Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
guide
HP OpenView Storage
Data Protector I:
Fundamentals
u1610s b.00
training
2003 Hewlett-Packard Development Company, L.P.
Netscape, Netscape Commerce Server, Netscape Communications, Netscape Communications
Server, "N" logo, Netscape Navigator, Netscape Navigator Included logo, and Netscape Proxy
Server are U.S. trademarks of Netscape Communications Corporation
All other product names mentioned herein may be trademarks of their respective companies.
Hewlett-Packard Company shall not be liable for technical or editorial errors or omissions
contained herein. The information is provided “as is” without warranty of any kind and is subject
to change without notice. The warranties for HP products are set forth in the express limited
warranty statements accompanying such products. Nothing herein should be construed as
constituting an additional warranty.
HP OpenView Storage Data Protector I: Fundamentals
Student Guide
October 2003
Contents
Module 1 — Introduction
1-1. SLIDE: Welcome...................................................................................................................... 1-2
1-2. SLIDE: Agenda (1).................................................................................................................. 1-3
1-3. SLIDE: Agenda (2).................................................................................................................. 1-4
1-4. SLIDE: Additional Resources................................................................................................ 1-5
Module 8 — Backup
8-1. SLIDE: Performing Backups...................................................................................................8-2
8-2. SLIDE: Backup Specification Types ......................................................................................8-5
8-3. SLIDE: The Backup Specification (datalist).........................................................................8-7
8-4. SLIDE: Backup Specification Contents.................................................................................8-9
8-5. SLIDE: Backup Specification Sequence..............................................................................8-11
8-6. SLIDE: Creating Backup Specifications..............................................................................8-12
8-7. SLIDE: Load Balancing..........................................................................................................8-14
8-8. SLIDE: Static Device Allocation...........................................................................................8-16
8-9. SLIDE: Load Balancing — Object Allocation .....................................................................8-17
8-10. SLIDE: Interactive Backup Specifications........................................................................8-19
8-11. SLIDE: Source ......................................................................................................................8-20
8-12. SLIDE: Destination ..............................................................................................................8-22
8-13. SLIDE: Backup Specification Options..............................................................................8-24
8-14. SLIDE: Pre- and Post-Execution .......................................................................................8-27
8-15. SLIDE: Pre- and Post-Exec Script Failures......................................................................8-29
8-16. SLIDE: Reconnect Broken Sessions .................................................................................8-33
8-17. SLIDE: File System Options ..............................................................................................8-35
8-18. SLIDE: Object Summary ....................................................................................................8-41
8-19. SLIDE: Object Properties...................................................................................................8-42
8-20. SLIDE: Parallel Data Streams from Object ......................................................................8-44
8-21. SLIDE: Configure Parallel Data Streams..........................................................................8-45
8-22. SLIDE: The Backup Process Flow ....................................................................................8-46
8-23. SLIDE: Templates and Groups ..........................................................................................8-48
8-24. SLIDE: Preview ...................................................................................................................8-50
8-25. Review Questions ...............................................................................................................8-61
Module 9 — Restore
9-1. SLIDE: Performing Restores .................................................................................................. 9-2
9-2. SLIDE: Restore Objects .......................................................................................................... 9-4
9-3. SLIDE: Restore from a Session.............................................................................................. 9-6
9-4. SLIDE: Parallel Restore .......................................................................................................... 9-7
9-5. SLIDE: Restore Sequence....................................................................................................... 9-9
9-6. SLIDE: Restore Source ......................................................................................................... 9-10
9-7. SLIDE: Restore Object Properties....................................................................................... 9-13
9-8. SLIDE: Destination................................................................................................................ 9-15
9-9. SLIDE: Restore Options........................................................................................................ 9-17
9-10. SLIDE: Restore Devices...................................................................................................... 9-19
9-11. SLIDE: Restore Media......................................................................................................... 9-20
9-12. SLIDE: Restore Summary................................................................................................... 9-21
9-13. SLIDE: Parallel or Single Restore...................................................................................... 9-22
9-14. SLIDE: Point in Time Restore ............................................................................................ 9-23
9-15. Review Questions ................................................................................................................ 9-25
Module 16 — Troubleshooting
16-1. SLIDE: Log Files .................................................................................................................. 16-2
16-2. SLIDE: Execution Tracing.................................................................................................. 16-4
16-3. SLIDE: Message Details .................................................................................................... 16-12
16-4. SLIDE: Network Connectivity ......................................................................................... 16-14
16-5. SLIDE: Services ................................................................................................................. 16-18
16-6. TEXT PAGE: User Interface Startup Problems ............................................................ 16-23
16-7. SLIDE: Backup Devices .................................................................................................... 16-26
16-8. SLIDE: Backup and Restore............................................................................................. 16-33
16-9. SLIDE: omnihealthcheck.................................................................................................. 16-39
Module 17 — Customizing
17-1. SLIDE: Customizing............................................................................................................17-2
17-2. TEXT PAGE: Contents of the globals File .......................................................................17-4
17-3. TEXT PAGE: Contents of the omnirc.TMPL File..........................................................17-21
Course Description
This course is designed for system administrators and consultants who will be implementing,
planning or administering the HP OpenView Storage Data Protector product on HP-UX,
Windows NT/2000 and Solaris systems.
Throughout this course, we will refer to HP OpenView Storage Data Protector simply as Data
Protector for simplicity.
Course Goals
• This course is targeted at system administrators who are responsible for managing the
system backup and recovery in a heterogeneous networked environment with HP
OpenView Storage Data Protector software.
• This course teaches system administrators and network administrators how to install,
configure, and customize the HP OpenView Storage Data Protector product.
HP-UX System and Network Administration II (H3065S)(for students who will be working
in the UNIX environment)
• For Windows NT/2000, this level equates to Microsoft Windows Server administration or
equivalent experience.
• For Solaris, system and network administration training or equivalent experience.
• Networking knowledge and backup device knowledge is also recommended.
Conventions
For convenience, we will refer to specific product directory names by their logical names
rather than the fully qualified paths.
Unix
Logical Name Directory Path Usage
$OMNIHOME or /opt/omni Binaries, man pages, etc.
<OMNIHOME>
$OMNICONFIG or /etc/opt/omni Configuration directory
<OMNICONFIG>
$OMNIVAR or /var/opt/omni Database and log files
<OMNIVAR>
Windows NT/2000
Logical Name Default Directory Path Usage
$OMNIHOME or C:\program files\Omniback The product root directory
<OMNIHOME>
$OMNICONFIG or C:\program files\Omniback\config Configuration directory
<OMNICONFIG>
$OMNIVAR or C:\program files\Omniback The product root directory
<OMNIVAR>
Agenda
Day 1
Module 1 — Introduction
Module 2 — Data Protector Overview and Architecture
Module 3 — Data Protector Installation
Module 4 — Data Protector Basics
Day 2
Module 5 — Tape Library Configuration and Implementation
Module 6 — Media Management
Module 7 — Logical Devices
Module 8 — Backup
Day 3
Module 9 — Restore
Module 10 — Internal Database
Module 11 — Monitoring and Reporting
Module 12 — Event Notification
Module 13 — Access Control and Security
Day 4
Module 14 — Disaster Recovery
Module 15 — Manager of Managers
Module 16 — Troubleshooting
Module 17 — Customizing
Welcome
Student Notes
Welcome to HP Education, and the HP OpenView Storage Data Protector 1: Fundamentals
course (U1610S). This course is designed for system administrators who will be responsible
for the installation, configuration, and management of the Data Protector storage
management software.
This course covers the HP OpenView Storage Data Protector product functionality for
version 5.1, released June 2003. Throughout this course, the product name “HP OpenView
Storage Data Protector” will be shortened to just Data Protector or DP for simplicity.
Agenda (1)
• Architecture
• Installation
• DP Basics
• Library Implementation
• Media Management
• Logical Devices
• Backup
• Restore
Student Notes
The main topics in this course are listed on the slide.
Agenda (2)
• Internal Database
• Monitoring and Reporting
• Event Notification
• Cell Security
• Disaster Recovery
• Manager of Managers
• Troubleshooting
Student Notes
Additional Resources
• Product documentation
• Web sites
• Support services
• Consulting services
• Users’ group
Student Notes
Hewlett Packard provides several additional resources designed to make you successful with
our products. These include:
• Product documentation
− Soft copy (Acrobat format) is included with the software distribution as well on the
on the web.
− Suggested reading:
− HP OpenView Storage Data Protector Administrator’s Guide
− HP OpenView Storage Data Protector Concepts Guide
− HP OpenView Storage Data Protector Installation and Licensing Guide
• Web Sites
− http://education.hp.com
− http://openview.hp.com
− http://itresourcecenter.hp.com
• Support services
− HP Response Center
− Account Support Organization
• Consulting services
− HP Consulting & Integration
In addition to the support and services available from HP, there is an HP sponsored user
group called OpenView Forum. They typically have yearly conferences and have several
other benefits available to members. Their information is available via the web at:
http://ovforum.org.
What is it?
• Software that provides automated data protection for businesses
with 24x7 availability needs.
What does it do?
• Data Protection: copies data onto a storage device, so that in
case of a disaster, data can be easily recovered and made
accessible.
• Media management: easily manages the library catalogues to
keep track of all media and copies of data for fast recovery.
Most important features:
• Automated backups that scale from small workgroups to multi-
site, heterogeneous SAN & NAS environments with thousands of
servers.
• Fully-integrated Zero-Downtime backup with Instant-Recovery.
Student Notes
HP OpenView Storage Data Protector is a new generation of HP OpenView software that
manages data protection as an integral component of an overall IT service.
By managing data protection as a set of services rather than a set of data objects and IT
resources, Data Protector helps you meet your service level objectives (SLO) with increased
staff efficiency. This in particular addresses the SLM requirements of service providers.
Data Protector builds upon the capabilities of its predecessor, HP OpenView Omniback II, for
tape management, backup, and disaster recovery, and establishes a new focus on recovery
and service-centric management.
Managed Environment
Student Notes
The typical IT environment today consists of many systems distributed across the enterprise.
The traditional data center has experienced tremendous change and become a server and
storage farm. The systems that operate today’s’ corporations are very numerous and contain
huge quantities of data.
The picture above is representative of the IT environment today. Many systems from the
desktop to the data center, connected via high-speed local area networks (LANs).
Behind these systems are increasingly large and complex data storage systems. As the need
to access data from multiple systems and the quantity of data increases, companies are
turning to large storage systems, such as the HP StorageWorks disk arrays for on-line storage
and automated tape systems for near-line storage. Many storage devices are either directly
connected to a host or connected via a Storage Area Network (SAN) to meet data storage
accessibility needs.
Managing the complexities of the IT infrastructure today requires an even more capable
solution to meet the ever changing IT Service Management environment.
Backup Models
Backup Server
tape
media
host P
disk
host S S S
tape
Backup Server
Student Notes
To protect data from all risks of loss, Data Protector offers a variety of ways to back it up and
recover it including Zero Downtime Backup (ZDB) and Instant Recovery (IR). Data Protector
offers several models for data security and backup including:
• Direct attached storage
• Zero Downtime Backup with Split-Mirror (StoragWorks XP)
• Zero Downtime Backup with Snapshot (StorageWorks EVA, VA, MSA)
• Heterogeneous network backup
• Storage Area Network (SAN) attached online and nearline storage
• Network Attached Storage (NAS, using Network Data Management Protocol (NDMP)
• Direct backup using X-Copy (extended copy)
Data Protector’s Instant Recovery (IR) is capable of recovering terabytes of data in minutes
rather than hours. Unlike traditional tools that focus exclusively on backup to tape, Data
Protector enables a variety of techniques to create recovery images using disk resources as
well as tape. These techniques can maximize information availability and minimize
application impact, by incorporating near zero-impact, zero-downtime backup or Direct
Backup (server-less backup from disk to tape), depending on your business needs and
available hardware.
P – primary LDEV
M – mirror copy (MU0-2)
Student Notes
The general idea behind split mirror backups is to stream the backup from the mirror instead
of the production disk. The mirror is typically connected to a separate host (called the
backup host) with a tape device attached. Usually, hardware mirror technologies such as
Business Copy XP or Continuous Access XP are used to create the mirror.
Before a backup of a mirror can be started, a valid point in time disk image needs to be
created. The disk image needs to be consistent so that it can be fully restored. The mirror is
not created at backup time but needs to be established ahead of time. To create the backup
image, the mirror will simply be split off the production disk at backup time.
As the application host and backup host are different, it is very important that all cached
information (database cache, filesystem cache) on the host is flushed to the disk before the
mirror is split off. One of the following options achieves this (depending upon the type of
data to backup):
The above must occur prior to the split of the mirror to guarantee that the backup image will
be consistent.
In case of a plain filesystem backup, it won’t be required to unmount the filesystem first. The
split-mirror backup will complete successfully also with the filesystem mounted. However, a
successful restore of all files and directories cannot be guaranteed since cached data won’t
be written to disk prior to the split. It’s therefore recommended to unmount a filesystem
before performing a spit-mirror backup.
In case a database is running on a filesystem, there will be no need to unmount the filesystem
as the database controls the write to the disk and ensures that data is really written to the
disk and not to the filesystem cache.
For the online database backup, the backup image alone cannot be restored. The archive log
files from the application host are also needed. The archive log backup can be started when
the database is taken out of backup mode. This will happen right after the mirrors were
successfully split off their productive disks. The backup duration (from the perspective of the
application) is only the time required to perform the split, during which the consistent
backup copy is created. The backup and the resynchronization of the mirrors do not affect
the production database’s I/O performance as they happen inside of the XP Disk Array.
The HP Education course, U1611S, covers the concept of Zero Downtime Backup and Instant
Recovery within a hands-on SAN environment.
Mirror Rotation
Mirror rotation relies on Business Copy’s capability to maintain up to three independent
secondary volumes (S-Vols) of one primary volume (P-Vol). The different S-Vols are labeled
as Mirror Units (MU#0, MU#1 and MU#2).
Data Protector can perform split mirror backups from each of the split mirrors.
Administrators can either supply one dedicated S-Vol or multiple S-Vols for backup. If two or
more mirrors are available, Data Protector will automatically use them in a cyclic fashion. At
the end of the backup, the S-Vol used will be left split off the P-Vol thus keeping the backup
versions on the S-Vol available for Instant Recovery. For the next backup, another S-Vol will
be used. This provides a high level of data protection.
S S
Backup host
P – primary LUN
S – snapshot / child
Student Notes
The snapshot backup concepts are similar to those of the split-mirror backup. The snapshot
backup currently is supported with the HP StorageWorks Virtual Arrays, VA71xx and
VA74xx, the HP StorageWorks Enterprise Virtual Array, EVA3000 and EVA5000 as well as the
HP StorageWorks Modular Storage Array, MSA 1000 (MSA available later).
Snapshots may be created on the fly within the array, or they may be designated for re-use for
backup utilizing a rotation strategy.
Snapshots may be designated for use with the Instant Recovery capabilities of Data
Protector. (At the time of this printing, Instant Recovery is only supported for the HP VA
products. It is expected that the EVA and MSA will be supported via a patch release due later
in 2003).
The HP Education course, U1611S, covers the snapshot integration in detail along with
Instant Recovery within both a file system and RDBMS environment.
Network
Desktop Network Access Servers Databases Applications
Managing Distributed UNIX and Windows Environments, End-to-End
Student Notes
Illustrated above is the current OpenView building block architecture.
What sets OpenView apart from other solutions is the flexible architecture that allows you to
build an IT management environment according to needs and requirements.
Our different product offerings can be used as standalone products or in an integrated
fashion. Network Node Manager (NNM) and OpenView Operations (OVO) are the most
common integration points for HP and third-party management products. The flexible OVO
and Service Navigator consoles also function as one of the OpenView Enterprise Consoles.
The Service Desk and Service Information Portal products form the service management
umbrella and add a service management process layer and functionality on top of the
integrated OpenView solution to complete the service management product offering.
There are over 400 OpenView products. This course obviously will not cover all of the
products, but it will focus on the Data Protector storage management product.
Storage Data Protector holds the backup performance record of 3.6TB/hour. It supports
business and IT alignment, and offers turnkey control to create one complete, integrated
backup solution for heterogeneous environments.
IT managers are equipped with the key data to enable proactive monitoring and planning of
backup and data recovery operations.
Deep integration from Data Protector along with the OpenView Operations centric
environment provides unmatched service level management capabilities.
Integration with other HP OpenView service management solutions through the Application
Response Measurement (ARM) API and utilization of Data Source Integration (DSI) allows
data to be leveraged into service availability and recovery planning activities that are critical
to maintaining service level agreements in a heterogeneous environment.
With HP OpenView Data Protector 5.0 (OV DP) there are four new Service Management
Integrations introduced which aggregate data and reduces complexity in a large scale, global
data center.
Enterprise IT departments are increasingly using service management tools, techniques, and
methods to set service level expectations, measure service delivery against those
expectations, and to justify future service expansion. In short, the IT department now is run
like a business.
Part of IT’s business is managing the risk of data loss. Threats ranging from user error, to
viruses or other unauthorized data access and modification, or to the occasional failure of the
storage device itself put data at risk twenty four hours a day. Business critical data loss can
cost the enterprise thousands, even millions of dollars per hour of downtime. While all data is
at risk, not all data justify equal recoverability. IT department must protect the business
critical data to a higher level of protection than the less valuable data, and do so cost
effectively.
Service providers use Service Level Agreements (SLAs) to document the provider-customer
contractual expectations. SLAs typically establish availability and performance objectives.
Using this model, a provider can offer multiple service levels each at its own cost structure.
By identifying the relative value of data placed within its care, IT department can set service
expectations on backup and recovery consistent with the protected data’s business value.
Backup and recovery now is managed like the enterprise itself: that is, like a business.
Demonstrating SLA compliance requires constant monitoring and periodic reporting to show
whether SLA expectations have been met. Data Protector out of the box has monitoring,
notification, and reporting tools to document backup and recovery operations. Data
Protector integration with other OpenView service management products consolidates
service views, service performance data and other capabilities into one console, giving a
service provider better information and insight into the overall IT service delivery.
• OVO DP SPI (OpenView Operations Data Protector SMART Plug In) is a package
containing components of Data Protector that are fully integrated into OVO. The
integration includes users, message groups, node groups, applications, reports, service
definitions and command executables. Installation of the Data Protector cell manager
onto the OVO management server is required for the enterprise console functionality and
scalability that this integration provides.
• OVR (OpenView Reporter) is a reporting service that further analyzes, inspects, and
collects data gathered by OVO and formats them into a human readable and usable web-
based presentation.
• DP – OVR integration integrates DP 5.0 with HP OVO, OVSN, OVP Agent and OV
Reporter
The integration of DP 5.0 with HP OVO is extended by adding HP OpenView Reporter
(OVR 3.0 English version). With OVR service providers can generate reports from data
obtained from the OVO management server. An IT Service Provider can use these reports
to demonstrate to a customer its SLA compliance. For example, “DP Transaction
Performance” Report consists of the service performance metrics (one of the IT SLA
parameters). In addition to SLA compliance reports, an IT Service Provider can generate
monthly operational reports for DP5.0 environment. For example, “DP5.0 Operational
Error Status” report aggregates the “problem” data and can be used by an IT service
provider for operational planning.
• DP – OVSIP integration integrates DP 5.0 with HP OpenView Service Information
Portal (OV SIP).
OV SIP gives an IT service provider customer visibility into the services that they are
outsourcing. OV SIP instead of giving the customer a generalized view of the service
provider’s infrastructure, personalizes that information for each customer and shows
status and business information specific to customer’s outsourced environment. OV SIP
contains a portal foundation and a range of management information modules. The Data
Protector module on OV SIP extracts status information from DP 5.0. With this module,
an IT service provider can give its customers a view into the status of their outsourced
data protection operations.
• DP – OVSD integration integrates DP5.0 with HP OpenView Service Desk (OV SD).
OV SD is a help desk solution. It enables the IT support organization to implement
configuration, help desk, incident resolution, problem resolution, and change
management processes into a single workflow. OV SD automates and regulates IT
troubleshooting processes. It stores SLAs and monitors support service compliance to
them. When integrated with DP5.0, OV SD (without a human involvement) monitors the
time taken to resolve backup-related problems, such as adding media or restarting a
failed backup, increasing DP’s monitoring and measuring capabilities. OV SD manages
service help desk workflow, measures service quality levels, and generates reports
demonstrating SLA compliance. DP 5.0’s integration with OV SD gives support personnel
access to DP5.0 data for a timely response and resolution of operational problems before
they affect vital data protection service.
cell
cell
manager
manager
cell
cell
clients
clients
Student Notes
The basic HP Data Protector implementation utilizes only two architecture layers, the Cell
Manager, and the Cell Client layers.
The architecture is highly scalable and lends itself to the simplest single system
configuration, right up to the most complex multi-system, multi-site enterprise-wide solution.
Data Protector is available as a Single Server Edition, designed for smaller environments.
The Data Protector client/server architecture provides multiple manager layers, which offers
tremendous flexibility and adjusts easily to organizational needs and changes.
Enterprise Console
The Data Protector integration with HP OpenView Operations provides the concept of the
Enterprise Console. HP OpenView Operations allows remote administration and monitoring
of one or more Data Protector cells from a single Enterprise Console.
Manager of Managers—MoM
An existing Data Protector Cell Manager can be configured as the Manager of Managers
(M.o.M.) which allows remote administration and monitoring of many cells from a single
consolidated GUI. A centralized media management database (CMMDB), cross-cell device
sharing as well as central license management may also be configured with MoM.
Firewall Support
Data Protector has support for backups to be managed through a firewall. This gives
administrators more control for remote managed environments.
SAN Support
Data Protector is used today in several different SAN implementations. As this technology is
evolving, consult the OpenView web site for the latest information about the supported
environments.
Scalability
Data Protector is used in environments from one system (which could be a data server) to
environments with thousands of systems. Through its architecture, it is highly scalable and
suitable for nearly any kind of environment.
Easy-to-Use
Data Protector comes with an easy-to-use cross-platform consistent Windows style GUI and
allows easy administration of a complex environment.
Disaster Recovery
Data Protector allows easy disaster recovery of a complete Windows system.
NDMP Support
Data Protector allows the backup of data stored on an NDMP server such as NetApp filers.
NetApp Filers have their own operating system, called ONTAP, and contains a NDMP server
implementation, which is used by Data Protector to perform a backup and restore on such a
system.
Tape-Library-Support
Data Protector supports multiple tape libraries, which allow for fast unattended backup
times.
Flexible
Because of multiple backup and restore options, Data Protector is very flexible. It fits all
kinds of end-users and administrator requirements.
Multi-Vendor Support
The various Data Protector agents (Disk Agent, Media Agent, and Online Application
Integration Agents) are supported on various platforms, making Data Protector truly a
backup solution for multivendor environments.
Integrations
In addition to the online backup integrations, Data Protector offers integrations with
OpenView Operations, OpenView Media Operations, MC/ServiceGuard, MS Clusters, and
OmniStorage. Data Protector also integrates into the Microsoft Management Console for
more convenient access.
Cell Concept
• Backup domain
• Logical organization of systems
• Can match your organization
• Heterogeneous system support
• Independent but can be centrally
managed
Student Notes
The Data Protector architecture breaks down the size and complexity of the enterprise
network by defining Data Protector Cells.
A Data Protector Cell consists of a Cell Manager system and all of the systems that are to
have backup managed by it. A cell can be all the systems within a department, or all systems
within one room or building. It is also possible to have a cell that contains only one system
(called a single-system cell).
The Data Protector Cell configuration can reflect the organization within a company, with
each department having its own administrators. However, there is no reason that two
machines, thousands of miles apart, cannot be in the same cell.
There is no enforced limit to the number of systems per Data Protector Cell, but the cell size
may be limited by a number of factors:
• The maximum supported number of systems is 1000, although 100 is recommended
• The size of the Data Protector internal database
• The quantity of backups that can be effectively managed
The Data Protector internal database can grow to be many GB. A good rule of thumb is that
you should allocate enough disk space to allow the internal database be approximately 2% of
the quantity of data that is backed up. You may find that if you are backing up many large
files (50 MB–100 MB each), then the percentage size of the database compared to data can be
as little as 0.25%; this is especially true when backing up large database files. Backing up
many small files means more records in the database, which means more space is required
for the database.
Later in this module you will see how to estimate more accurately the size of your Data
Protector internal database. (The module on Database Management will give more specifics
about how to plan for and manage database growth.)
Cells are generally independent parts of the enterprise network. They are administered and
operate independently of each other. Data Protector has the capability to monitor and
administer all the cells from a central administration point utilizing the Cell Console or
Enterprise Console or the Manager of Managers console.
NOTE: If systems in the same cell are in different time zones, some of the Data
Protector messages can be confusing. In addition, all backups are configured
according to the Cell Manager’s clock.
Client-Server Modules
cell console
cell manager (CC)
(CM)
media agent
(MA)
disk agent
(DA)
Student Notes
Data Protector is composed of separate modules, each of which performs a specialized task.
The major component is the Cell Manager; it is responsible for the control of the entire Data
Protector Cell and the invocation of the specialized agent processes.
Client/Server Architecture
The basis of the client/server model is that the Data Protector software consists of client
modules and a server module. These modules can all be installed on a single node (a single
node cell) or be distributed across many nodes.
Platform Support
Cell Manager:
• HP-UX Application (Integration) Agents
• Windows • Oracle
• Solaris Catalog/Media Database: • Informix
• files, versions, hosts • IBM DB2 UDB
Disk Agent: • media, drives, libraries • Sybase
• HP-UX 11.X • SAP R/3
• HP OpenVMS • Baan IV on Oracle, Informix
• HP Tru64 UNIX • Lotus Domino
• HP MPE/iX • MS SQL Server
• Win NT, XP • MS Exchange Server
• Win 2000, 2003 • MS VSS
• Sun Solaris 7,8,9 • MS Cluster Server
• SunOS • HP MC ServiceGuard
• IBM AIX • HP OpenView Operations
• Linux Redhat/SuSE/Caldera • HP OpenView OmniStorage
• Novell NetWare Media Agent:
• SGI IRIX • Windows
• Windows NT-Alpha • HP-UX
• SCO OpenServer • Sun Solaris
• SCO Unixware • IBM AIX
• SNI SINIX • Linux Redhat/SuSE
• NCR MP-RAS • Novell NetWare
• Sequent DYNIX • HP MPE/iX
• Additional platforms via NFS / shared disk (CIFS) • SCO OpenServer
• SNI Sinix
Student Notes
The Data Protector product consists of several product components: the Manager of
Manager, the Cell Manager, Backup Device Manager (with the Media Agent), Backup Agent
(with the Disk Agent) and various Application Agents.
Included in the product documentation you will find several release specific documents
describing the supported platforms and integrations.
Cell Manager
z Internal database
Session Managers
z Session managers
z Scheduler
Disk, Media and Integration
z Cell Console and agents Agents
Student Notes
The Cell Manager is the key component of a Data Protector Cell. It contains the Data
Protector database, and is responsible for the starting of backup, restore, and media
management sessions.
The UNIX Cell Manager system always has three daemon processes running to provide Data
Protector services:
crs Cell Request Server
mmd Media Management Daemon
rds Raima Database Server
The Windows Cell Manager system always has three service processes running to provide
Data Protector services:
Data Protector CRS Cell Request Server
Data Protector Inet Remote Connection Server
Data Protector RDS Raima Database Server
In the Windows environment, the MMD runs as an application process, (mmd.exe) not as a
service.
Daemon/Service Control
The manager programs will reside in /opt/omni/lbin directory on Unix, and the
C:\Program Files\Omniback\bin directory on Windows. The three services/daemons
will normally be started when the system boots up. A program has been provided to stop,
start, and check on the status of these services, the program name is omnisv. There are
three options available for the omnisv program, they are: -stop, -start, -status.
Session Managers
The Cell Manager listens for session requests and starts the appropriate Session Manager,
which in turn starts the required clients. A dedicated Session Manager controls the clients for
each operation. If a new session is started, an additional Session Manager is generated.
These session manager programs will reside in the /opt/omni/lbin directory on UNIX and
C:\Program Files\Omniback\bin (default) on Windows, once they are installed with the
Cell Manager.
Student Notes
The Data Protector Internal Database (IDB) is comprised of several structures that store
data. The three main structures are shown above, they are:
The IDB has several defined, supported limits. These limits should not be exceeded under
any circumstances. The limits illustrated on the slide are also available from the product
Release Notes document that ships with the product.
The file names database file is initialized with a 2GB (2047MB) maximum size by default, but
may be extended in up to 2 GB (2047MB) increments to a maximum of 32 GB. The minimum
size per extension is 1MB.
The file versions stored in the DCBF is initially configured as one directory capable of storing
up to 4 GB, but may be extended in up to 4 GB increments to a default maximum of 10
directories. To reach the 50 directory limit changes to the global options file (covered later)
must be made. Each extension directory may contain up to 10,000 files; the limit for file
versions is set to allow approximately 10 times the number of filenames. This represents
approximately 80% of all the data stored by Data Protector.
Student Notes
The capacity planning worksheet (spreadsheet) shown above is included in the Data
Protector product distribution. The spreadsheet contains macros, which will help in planning
future database growth potential. Simply plug-in the appropriate data and the macros will
calculate the amount of disk space that is needed.
The spreadsheet is installed in the UX: doc or Windows: Docs directory on the cell manager
and is called IDB_capacity_planning.xls.
Note! The spreadsheet must be copied to an appropriate system to view and use the tool.
An alternate approach to using the spreadsheet is to use the documented formulas for
estimating the disk space needed. The Data Protector Concepts Guide documents the
formula for each part of the database. The spreadsheet is the preferred method.
z Provides:
z Graphical user interface
z Command-line interface
z Web reporting java interface
Student Notes
Data Protector provides user interfaces for the UNIX and Windows platforms. The user
interface is commonly referred to as the Cell Console. Both UNIX and Windows platforms
include the following components:
• Graphical user interface
• Command line interface
• Java based reporting interface
The user interface is installed as a Data Protector software component onto the Cell Manager
system, but it may also be installed on any number of clients within the cell. A system
administrator or a backup operator will use the cell console to control the cell. Therefore, it
should run on the platform that will simplify the configuration and administration of Data
Protector. It is common practice to install the Cell Console user interface on both UNIX and
Windows clients within the same cell.
Once you have installed the user interface on a system in the cell, you can access the Cell
Manager remotely from the local machine. You do not have to use the Cell Manager as the
central graphical user interface system, although the user interface is installed there by
default. The Data Protector graphical user interface for Windows can be installed on any
Windows NT/2000/XP/2003 system, and the Data Protector graphical user interface for UNIX
(Motif) can be installed on any HP-UX or Solaris system in the cell. You can have an HP-UX
Cell Manager, with the user interface installed on a Windows system.
Data Protector provides a rich and powerful command line interface. The commands can be
used in situations where a GUI is not available, for example, when dialing in to a system for
remote support, or when writing shell scripts or batch files. Most of the Data Protector
commands will reside in the bin directory. Some additional platforms support a subset of the
cell console in order to control some of the local integrations with Data Protector. In many
cases the support is for parts of the command line interface only.
NOTE: The distributed Cell Console must be authorized from the User Manager
interface running on the Cell Manager. Details are covered in the User
Configuration module
Disk Agent
z Media
Student Notes
The Disk Agent module is responsible for all read and write actions to disk storage
performed by the Data Protector backup and restore managers. Therefore, in order to back
up or restore a client node, you must have a Disk Agent module installed on the client
system. The Disk Agent module consists of specialized processes that are started on demand
by the respective Backup or Restore Manager process. These programs are installed in the
/opt/omni/lbin directory on HP-UX and
C:\Program Files\Omniback\bin on Windows.
Media Agent
Student Notes
The Media Agent module is responsible for all read and write actions performed to tape by
the Data Protector backup, restore and media managers. Therefore, in order to utilize such
devices for backup or restore, a Media Agent module must be installed on the client system
to which the backup device is physically attached.
The Media Agent module consists of specialized processes that are started on demand by the
respective Backup, Restore or Media Manager process. . These programs are installed in the
<omnihome>/lbin directory on Unix and <omnihome>\bin on Windows:
Integration Agent
Student Notes
Data Protector provides a set of integration modules that enable data to be exchanged
between the most popular databases and Data Protector. Data Protector hooks into the
vendors API in order to perform online backups and restores. The ability to perform online
backups is a highly desirable feature in mission-critical, high-availability environments.
Data Protector also provides integrations with many other applications that assist in areas
such as high availability, system control, and monitoring.
Database Integrations
• SAP R/3
• Oracle
• Informix
• IBM DB2 UDB
• Sybase
• MS SQL
• MS Exchange
• Lotus Notes/Lotus Domino
Application/Device Integrations
• HP OpenView Operations
• HP OpenView Manage/X
• HP OpenView OmniStorage
• HP MC/ServiceGuard
• HP StorageWorks Disk Array (Zero Downtime backup)
• MS Cluster
• MS Volume Shadow Copy Service (VSS; Windows 2003 only)
• EMC Symmetrix (Fastrax)
• GRAU DAS
• StorageTek ACSLS
HP StorageWorks and EMC Symmetrix integrations provide special capabilities to allow data
on their disks to be backed up without downtime. These integrations require special licenses
in order to operate.
Installation Server
z Manually installed
z Repository for agent software
z Must be registered with cell
manager
z HP-UX, Windows, and Solaris
platforms
HP-UX
z Used separately by UNIX and
Windows clients
z Distributes the installation load Solaris
z May be used by multiple cells
Windows
Student Notes
The Installation Server acts as a repository for the agent software modules. The Installation
Server does not need to be a client/agent of the Data Protector cell for which it provides
installation services. The Installation Server must be registered as such with a Cell Manager,
and may provide installation services for more than one cell.
When the Cell Manager system pushes agent software to a client system, the particular
Installation Server from which the software is to be obtained is specified. Unix and Windows
Cell Managers must maintain two separate Installation Servers, one for each platform.
Data Protector software patches must be applied to the Installation Servers(s) and then
distributed to clients during an update/push from the Cell Manager.
cell server
request read/write
crs rds
start
cell console
connect
session IDB
bsm catalog
control/report control/report
Student Notes
There are several processes that execute while backup or restore jobs are executing. The
slide above illustrates the location of the processes that execute on the various systems, as
well as their roles.
NOTE: Data from the backup flows directly between the agents, and does not flow
through the manager.
Acronyms:
cell console
network local
backup/restore backup/restore
disk agent
disk agent tcp/ip
tcp/ip
cell manager
shared
scheduler
tcp/ip memory
session session
manager manager
Student Notes
Data Protector is a distributed application and relies heavily on multiple cooperating local
and remote processes. Its IPC mechanisms are designed and implemented with great care to
maximize system response time and data throughput. Data Protector concentrates on simple
bi-directional messaging for both data and message transfer.
As both network capacity and backup device speed are expected to increase significantly
during the lifetime of the Data Protector product, all IPC channels are carefully designed to
avoid communication bottlenecks. Data Protector uses the following fast and reliable IPC
mechanisms, available on all major platforms today:
Within a Data Protector Cell, all systems must have the same port number configured, but it
may vary from cell to cell. The default port number used is 5555. If this port is already in
use, Data Protector can use another port number. This number must be identified in the
global (in addition to the Windows Registry) file before installing the clients.
The Data Protector session manager invokes specific agent processes, depending on the
request it has received, and uses the following mechanism to achieve this:
1. The session manager connects to the system on which it wants to start a media or disk
agent process via the predefined port number, 5555.
2. At the Unix Agent host, the inetd daemon process is listening on port 5555 and starts
the HP Data Protector inet process, as defined in the /etc/inetd.conf. (This
assumes the system security). On the Windows platforms, the Data Protector inet service
is already running on port 5555 to handle incoming requests.
3. The session manager sends a control block that informs the remote system exactly what
agents to start and what ports are to use for communication, etc.
4. The Data Protector inet process then starts the desired agents.
HKEY_LOCAL_MACHINE
-SOFTWARE
-Hewlett-Packard
-OpenView
-OmniBackII
-Common
-Parameters
InetPort 5555
rptschedules amoschedules
barlists
barschedules
sap cc win ma opc sybase
snmp
acs oracle oracle8 da das stk informix
mom Note: these directories contain the installation server components
Student Notes
The following table outlines the directories used by Data Protector on the Unix Cell Manager
system.
<product_home>
utilns
Student Notes
No files are outside the <OMNIHOME> tree. The database and all log files are kept under the
<OMNIHOME> tree.
<OMNIHOME> (default install dir) Data Protector home directory (default, may be
(C:\Program Files\Omniback relocated during installation process)
cell log
bin lbin databases
install tmp
newconfig lib lib sbin
customize
vendor
nls
Student Notes
The directories used on the Unix clients are a subset of the directories used by Data
Protector on the Unix Cell Manager system.
<product_home>
utilns
setupdir sap cc win ma opc sybase
Student Notes
The directories used on the Windows clients are a subset of the directories used by the Data
Protector on the Windows Cell Manager system.
Global Options
• Centrally managed
• Product defaults (documented)
• Customizable
<config_dir>
options
global
Student Notes
In most situations, the Data Protector default configuration and options are adequate for
everyone. However, many options can be changed to affect the behavior of the product for
large and more complex environments.
On Unix Systems
<OMNICONFIG>/options/global
On Windows NT Systems
<OMNICONFIG>\options\global
Option settings from this file are available to all user interface programs (Windows, Motif,
and the command line interface) and all Cell Manager programs. These options are not
directly distributed to disk or media agents.
This file may be modified whenever the need to affect the options in the file is necessary. The
options file contains many of the Data Protector defaults, but is only used if the items are
uncommented. Each option currently in the file has a hash mark, or pound sign (#), which
comments out the option. This means that it does not affect Data Protector.
Localized Options
• Locally managed
• Agent parameters
• Customizable
<product_home>
omnirc/.omnirc omnirc.tmpl
copy/modify
Student Notes
Using omnirc Options
The omnirc variables are most useful for troubleshooting or overriding other settings,
affecting the behavior of the Data Protector client only.
Even advanced users should not use them unless specifically required by documentation or
an HP support representative. The Disk and Media Agents use the values of these options ad
environment variables. These variables are found in the following locations:
When changing omnirc file, the Data Protector services/daemons on the affected system
must be restarted. This is mandatory for the crs daemon on UNIX and recommended for
Data Protector CRS and Data Protector Inet services on NT. Specifically on NT, restarting is
not required when adding or changing entries, only when removing entries (or renaming the
file).
OB2VXDIRECT: Enables direct (without cache) reading for Advanced VxFS file
systems, as well as improving performance.
OB2OEXECOFF: Allows a user to restrict or disable any object pre- and post-exec
scripts defined in backup specifications for a specific client.
OB2REXECOFF: Allows a user to disable any remote session pre- and post-exec
scripts for a specific client.
OB2DEVSLEEP: Changes the sleep time between each retry while loading a
device.
OB2RECONNECT_RETRY: Defines how long Data Protector should wait before trying to
reconnect after a socket connection has been broken (the default
is 1200 seconds). In other words, the WAN line between the
Backup Session Manager and agents cannot be down more than
OB2RECONNECT_RETRY seconds.
OB2RECONNECT_ACK: Defines how long Data Protector should wait for
the message of acknowledgement (default 600
seconds). In other words, if the agent does not
get an acknowledgement in OB2RECONNECT_ACK
seconds, then it will assume that the socket
connection is no longer valid.
1. What are the names of the five main components of the Data Protector architecture?
__________________________________________
__________________________________________
__________________________________________
__________________________________________
4. What, if any, is the limit to how many systems may be in the Data Protector cell?
5. How many Data Protector cells may an individual system be configured into?
8. What are the main directories for the Data Protector programs:
Unix:
Windows:
Installation Sequence
Student Notes
Planning
Before you start to install the Data Protector software, it is helpful to understand how your
Data Protector Cell should be assembled.
One of the systems within your Data Protector Cell must be the Cell Manager. If you are
running Data Protector in a mixed environment, you need at least two installation servers,
one for Windows and one for UNIX. The Cell Manager is typically used as an installation
server; this is an option available during the Cell Manager installation.
Install Clients
After you have installed the Cell Manager and Installation Servers, you may install the agents
on the client systems using the Data Protector GUI, or manually from the local CD-ROM.
Licensing
An Instant-On license is automatically created when the product is first installed. This gives
you usage for 60 days, during which time you must apply for and install a permanent license.
NOTE: Three symbolic names may be used throughout the rest of this manual for
paths to various files and directories.
<OMNIHOME> represents:
Unix: /opt/omni
Windows: C:\program files\Omniback
<OMNICONFIG> represents:
Unix: /etc/opt/omni
Windows: C:\program files\Omniback\config
<OMNIVAR> represents:
Unix: /var/opt/omni
Windows: C:\program files\Omniback
Installation Methods
Student Notes
Planning
Installation of the Data Protector A.05.10 version uses native installation tools on major
platforms: HP-UX, Solaris, and Windows (MSI 2.0).
The above table summarizes installation methods and Install Server availability.
Unix Installation server can be hosted on HP-UX 11.x or Solaris 7, 8, 9 platforms. They all are
capable of supporting every supported Unix client with the following exceptions:
• Dynix client
No installation server is capable of supporting Novell or MPE/iX clients in any way. These
clients must be installed and maintained locally.
The following types of local manual installations are supported with omnisetup.sh:
• New installation of Data Protector 5.1
• Detects and upgrades previously installed components including the Cell Manager and
Internal Database
Omnisetup will attempt to install the subset of the following components. The exact list of
the components is subject to availability on the particular platform.
cc Cell Console
da Disk Agent
das DAS Media Agent
acs ACS Media Agent
ma Media Agent
informix Informix Integration
lotus Lotus Notes Integration
oracle8 Oracle8 Integration (also used for Oracle 9)
oracle Oracle7 Integration
sap SAP R/3 Integration
Default answer is YES for da, generic ma, and cc components, and NO for any other
component – all subject to availability of the component on the host.
If DP 5.1 exists on the system and a specified component is already installed, the omnisetup
script provides a prompt, e.g.
(da) Disk Agent is already installed.
Reinstall (da) Disk Agent (YES, no, quit)?
If -install option is specified, or if a component is not found on the system, then the
behavior is as it would be a new installation.
Manual
If manual removal is selected, the script ends with brief instructions on how to remove
previous product and where to obtain additional information.
To remove a previous version of data protector:
Automatic
If automatic removal is selected, the script saves the omnirc file, then it compiles a list of
already installed components (in -install parameter format). Finally, it removes /usr/omni
(or /opt/omni) directory. If not successful, it aborts with the message:
Removal of previous installation FAILED. This is most likely because
some process is running and is blocking deletion of files. Please
remove the product manually.
If an automated removal of previous version is successful, omnisetup continues as it were a
new installation. However, if -install parameter was not specified, a list compiled before
product removal is used (and no further prompt is issued).
The first time any component is selected for installation or reinstallation, CORE component
is automatically installed (reinstalled if already there). In other words, the only way that the
script does not install CORE component is that user selects no component for installation or
reinstallation (always answers NO). Subsequent components do not trigger installation of
CORE component.
Similarly, the first time any integration component is selected for installation or
reinstallation, CORE-INTEG component is automatically installed (reinstalled if already
there).
If user confirms, the component is unpacked and installed. After the component is installed, a
message appears:
(da) Disk Agent installed.
Once a media agent is selected, subsequent media agents are not prompted for.
If the cell server host name was specified, installed client is automatically imported into the
cell.
Installation completed. Client was imported into a cell.
If host name was not specified, but cell_server file is available, host name will be taken
from there.
If the host specified cannot be contacted, or if no host was specified and cell_server file is
not available, user is reminded to import client into cell:
Installation completed, but client was not imported into a cell.
Please import a client manually.
Alternatively, user will be able to list components that are to be installed as a parameter.
./omnisetup.sh –server testbox –install da,ma,cc,informix
Parameters are NOT checked. For each component that is installed a message will appear
(as stated above). Misspelled components and components that do not apply to the system
are skipped with no message. If several media agents are specified, only the first is installed
(in the order stated in the above table - if all are specified, “das agent” is selected). In exactly
the same manner Oracle8 Integration overrides Oracle Integration.
1. Cell Manager or
2. Client or
3. Installation Server
From the command line (or batch file) using the following sytax:
Applies to
Value Description Cell Install.
Client
Manager Server
CM........... ...Local installation of Cell Manager √
Type CLIENT….Local installation of client √
IS…….Local installation of Installation server √
List of components. Each component is
preceded by a hyphen (dash) and followed by
Components √ √
a component version. Enclose list in double
quotes.
home- Folder where Data Protector A.05.00 is
√ √ √
directory to be installed.
crs-name Name of the account under which CRS service
√
runs.
crs-password Password for the account under which CRS
√
runs.
cell-server Name of the host that acts as a cell server. √ √
log-file Name of the log file √ √ √
Only basic error checking is performed. In case of an error, the installation is aborted.
<log-file> contains further information on this.
msiexec is part of Microsoft Installer.
However, when installing a Cell Manager, a password for CRS account must be
specified in clear text.
Supported Upgrades
Student Notes
Data Protector 5.1 is supported as an upgrade from previous versions of OmniBack as well as
Data Protector as shown above. Versions of OmniBack prior to 3.5 are not supported for
upgrade directly to 5.1.
Windows Cell Managers will automatically detect and upgrade Omniback or Data Protector
while running the setup.exe from the CD-ROM. Similar to the UX Cell Manager, the older
product will be removed and the 5.1 product will be installed.
When the Cell Manager is upgraded the software is replaced by the new version and the
Internal Database is also migrated. In the case of the 3.5 to 5.1 upgrade the database is
converted in two steps; first the core part of the database is converted to the new structure,
then the administrator may upgrade the detail part by using Data Protector utilities described
in the following sections.
Much of the 3.5 database will remain in the <OMNIVAR>/db directory, while the new
database is initialized in the <OMNIVAR>/db40 directory. The name db40 is used to represent
the Internal Database architecture which was developed for OmniBack 4.0. The 3.5 directory
<OMNIVAR>/db/catalog has to be changed since the IDB (Velocis) has been upgraded to its
version 3.5, and is not compatible with the old catalog contents. Before the changes are made
to the <OMNIVAR>/db/catalog it is copied to <OMNIVAR>/db/catalog.old.
Note: The old 3.5 database may be removed at any time after the core and detail
parts of the database are upgraded.
The core part of the database upgrade from 3.5 to 5.1 will transfer all vital data from the old
to the new database; this is started unconditionally as part of the upgrade. The entire MMDB
as well as the session information is transferred. Session messages, filenames and file
versions are not transferred during the core upgrade.
After the database core upgrade is completed, all Data Protector functionality is available
except browsing single files and directories as well as session message reporting.
Data Protector offers the administrator both a GUI as well as command line interface for the
upgrade of the detail parts of the IDB (these should be executed after the core part has
completed):
UX GUI xomnidbupg
Windows GUI idbupgwiz.exe
CLI omnidbupgrade -udp
These utilities shown above process all of the details from the CDB and import the data in the
5.1 IDB. The session messages are also imported. The sessions having media that was
overwritten or exported from the MMDB are removed. Catalogs for unprotected objects are
skipped. The number of objects skipped is reported in the <OMNIVAR>log/upgrade.log file.
On Unix there are two files, upgrade.log and Upgrade.log. The upgrade.log is generated by
the binary files and the Upgrade.log is produced by the scripts used for the upgrade.
When the Cell Manager is upgraded from 3.5, 4.x, or 5.0 to version 5.1, the contents of the
existing global file is merged into the new 5.1 global file as listed below, the old file is
renamed to global.#, where # is the next available integer starting with one:
• Uncommented parameters in the old file (active) are copied into the new file and
annotated with the string “This value was automatically copied from previous version.”
• Obsolete parameters are merged, but converted to comments and annotated with the
string “This variable is no longer in use.”
• Comments from the old global file are not transferred to the new file.
Manual
Installation/
Windows Update HP- UX/Solaris
Cell Manager Cell Manager
Installation Server Novell Windows 98 Installation Server
Agents Disk Agent CM
CM
IS
IS
CC CC
MPE/iX
push push
Agents
HP-UX Solaris
Agents Agents
Windows Application
Agents Agents
Unix Application
Agents Agents
Student Notes
Planning the Cell
The Cell Manager must be one of the following: HP-UX 11.x, Windows NT/2000/XP Pro/2003
or a Solaris (7,8,9) system. The Cell Manager system should be reliable and ideally configured
with high availability characteristics (RAID, Disk Mirroring, etc). The Cell Manager and
Internal Database must be available for backup operations to be performed.
In the previous slide, the Cell Manager systems are installed from local media. Depending
upon the platform, this may be accomplished by way of a network depot or shared drive
accessible to the native installation utilities for the respective operating systems.
Any supported system can be the Installation Server, as it is not confined to a single cell
usage. In a mixed environment, allocate one system to be the Installation Server for UNIX
platforms and one for Windows platforms. Creating more than one installation server for a
large environment can help distribute the installation load during updates or patches to the
Data Protector software.
After the Cell Manager and Installation Servers have been installed manually, most of the
agents can be installed on the systems via the Data Protector GUI. In most cases, the agents
are pushed from the Installation Server under the direction of the User Interface.
The OpenVMS, Novell, MPE/IX, and Windows 98 agents must all be installed manually from
the product media. This is because they do not support receiving software from an
installation server.
The installation procedure of the OpenVMS client has to be performed on the OpenVMS
system. You can install the Data Protector Disk Agent, Media Agent, and CLI Interface on
systems running OpenVMS/Alpha 7.3-1 or above. The product is called Data Protector (DP),
but the files and directories placed on the disk are named OMNI*. This is historical due to
the name change from OmniBack to Data Protector.
Prerequisites
Before you install Data Protector on the OpenVMS platform, check the
following:
For additional details, see the release notes document (DPVMSKIT) on the Windows CD-
ROM in the OpenVMS folder.
There are many pre-requisites that must be satisfied before starting the installation; see the
HP OpenView Storage Data Protector Installation and Licensing Guide for more details
and configuration steps to backup the NDS database.
The installation procedure can be performed from the Data Protector Windows CD-ROM.
Note that the Novell NetWare installation is not a part of the Installation Server functionality.
To install Data Protector on the Novell NetWare server, proceed as follows:
1. Run a command prompt on your Windows system and change the current path to the
CD-ROM root directory, then to the NetWare sub-directory.
2. Run the installation script:
Student Notes
General Requirements
• Networking software (TCP/IP) is installed and running.
• Port 5555 is available for the Data Protector services.
• Hostname resolution mechanism is implemented (consistent across all systems).
• FTP service is enabled.
• Kernel parameter: maxdsiz set to a minimum of 128MB.
The DP 5.1 Installation Server must meet the following minimum requirements:
• 64 MB RAM (minimum)
Student Notes
General Requirements
• 190 MB of disk space + approximately 2% of planned data to be backed up (for use by the
IDB)
Student Notes
All Data Protector configuration files and directories, as well as the Data Protector internal
databases reside on this Cell Manager system. If the Installation Server will also be installed
on this system (the default option), allow approximately 750 MB of disk space in
/opt/omni.
The software is installed from CD with swinstall. Within the swinstall utility you can
select the Data Protector bundle or manually select the required products, sub-products or
filesets. There are sub-products for all integrations. You need to install only the components
required in your environment. You can skip the components for the integrations that you do
not need.
The Data Protector installation procedure configures an automatic start and shutdown of all
Data Protector processes whenever a system is rebooted. The following files are
automatically configured:
Installation Steps
3. Specify the source as Local Directory and enter the mount point of the CD-ROM
drive followed by DP_DEPOT/DP_A0510_UX11x.sd_depot; Click OK to open the
Install - Software Selection window.
4. Select the B6960MA bundle, then click Actions -> Mark for Install. (This will
install the cell manager, installation server and all of the on-line documentation)
1. After you’ve installed the software, you can check that the Data Protector daemon
processes are running:
/opt/omni/sbin/omnisv –status
2. Check the swinstall log file /var/adm/sw/swagent.log for errors.
Then start the Data Protector GUI.
• The /etc/opt/omni/cell/cell_server file consists of only one line with the cell
manager name. This file exists on all systems within the cell.
Student Notes
If you want to install a Windows system as Data Protector Cell Manager or installation server,
you must install the software from CD-ROM.
To install Data Protector Windows client systems, you can either install the software from
CD or you can use the Data Protector GUI and define from which Windows installation
server you want to install Data Protector client systems. On the CD-ROM, execute (run) the
Windows Installer Package (setup.exe, located in the i386 folder or the IA64 folder as
appropriate) and you will get three options for the installation:
• Cell Manager (includes agents and option for installation server)
Installation Steps
The installation procedure consists of the following steps:
1. Run the Windows Installer Package and install the Cell Manager. If you deselect the
default option to install the Installation Server on the Cell Manager, then you must install
the Installation Server to another system before you are able to push agent software to
client systems.
You must also install the Installation Server on an HP-UX or Solaris system if you have a
mixed (Unix/Windows) environment.
2. Use the Data Protector user interface to distribute the agent software to the client
systems.
During the installation, a number of Data Protector registry entries are added to the
Windows registry, and three Data Protector services are configured and started:
NOTE: You can check the status of these services with the Control Panel. They
should be set to start automatically. You may also use the
<OMNIHOME>/bin/omnisv –status command to verify their status
On Windows systems, Data Protector runs all the services under a default system account
or the one specified during the installation.
A special Data Protector service user must be created to back up shared Windows disks
or integrations with databases and applications, such as MS SQL, MS Exchange, Oracle,
etc.
NOTE: See the Installation and Licensing Guide for more details.
After a successful installation, start the Data Protector GUI with the start button:
Student Notes
Follow the procedure below to install the Cell Manager on a Solaris
system:
1. Insert the Solaris installation CD-ROM.
2. Change to the main <package_source> directory, i.e. the directory that contains the
installation depot file (in this case <Mount_point>/DP-DEPOT).
The following sub-product packages related to Cell Manager installation are included in the
product:
OB2-CORE Data Protector Core software.
OB2-C-IS Installation Server Core software.
OB2-CC Cell Console software. This contains the graphical userinterface and the
command-line interface.
OB2-CS Cell Manager software.
OB2-DA Disk Agent software. This is required, otherwise it is not possible to back up
the IDB.
OB2-MA Media Agent. This is required if you want to attach a backup device to the Cell
Manager.
OB2-DOCS Data Protector online manuals.
Use the pkgadd facility to install the above packages in the order in which they are listed,
using the following command in each case:
If you want to install an Installation Server for UNIX on your Cell Manager, you can do it at
this point. Refer to “Installing an Installation Server for UNIX” later in this module for the
additional steps required.
IMPORTANT If you want to install Data Protector to linked directories, for instance:
/opt/omni/ -> /<prefix>/opt/omni/
/var/opt/omni/ -> /<prefix>/var/opt/omni/
/etc/opt/omni/ -> /<prefix>/etc/opt/omni/
you must create the links before the installation and ensure that the
destination directories exist.
Installation Servers
Student Notes
Installation Servers allow for distributed client software installations. Because the Cell
Manager is not responsible for the installation, remote installations can complete faster in
complex network environments. Thus, the Cell Manager is free for tasks that are more
important. In mixed UNIX and NT environments, an installation server of each type should be
installed to avoid manual client installations. However, this does not apply to Novell and
MPE/IX; these require manual client installations.
The choice for which platforms to use for the installation server depends largely upon the
cell clients. If you are using UX clients, then you will need the HP-UX installation server; this
may also be the Cell Manager system. If you are installing Windows clients, you will need at
least one Windows installation server; otherwise, you will have to install all of the clients
from the distribution media, or set up a network share.
After the Data Protector Installation Server is installed, it may no longer possible to remotely
install only the Data Protector Agent software onto the same system. (this is a older Windows
limitation that does not affect other newer versions of Windows).
In all cases, install the Data Protector client software before installing the Data Protector
Installation Server depot.
Troubleshooting DNS
The Data Protector 5.1 version introduces a new tool to help troubleshoot DNS problems
associated with clients within the cell.
root@r848c61 [/opt/omni]
# omnicheck -dns -host r848c61 -verbose
DNS check: checking connection between r848c61.dow.edunet.hp.com and dlthost.atl.edunet.hp.com
DNS check: checking connection between r848c61.dow.edunet.hp.com and r848c76.dow.edunet.hp.com
DNS check: checking connection between r848c61.dow.edunet.hp.com and r848c77.dow.edunet.hp.com
DNS check: checking connection between dlthost.atl.edunet.hp.com and r848c61.dow.edunet.hp.com
DNS check: checking connection between r848c76.dow.edunet.hp.com and r848c61.dow.edunet.hp.com
DNS check: checking connection between r848c77.dow.edunet.hp.com and r848c61.dow.edunet.hp.com
DNS check: all checks completed successfully.
Student Notes
The Data Protector software products are shipped on three separate CD-ROMs for the
supported Cell Manager platforms. They are:
• HP-UX
• Windows
• Solaris
This graphic above illustrates the HP-UX CD-ROM contents. Included are:
ADOBE contains instructions for obtaining the Acrobat Reader for HP-UX
DOCS contains the complete set of Data Protector Manuals
DP_DEPOT contains the software depot used for swinstall
LOCAL_DP_AGENT_INSTALL contains the omnisetup.sh script for local agent install
and cell manager upgrade
MISC some unsupported tools
OV_INTEGRATIONS contains the software for the OV Integrations
ReadMe.UX contains an overview of the CD-ROM
Student Notes
Shown above are the contents of the Data Protector for Windows CD-ROM. Included are:
Adobe contains an installable version of the Acrobat reader
Alpha contains the agent installation components for the Alpha platform
Docs contains the complete Data Protector manual set as PDF files
DP_Demo contains product demonstration material
i386 contains the setup.exe and all of the binaries for the Windws-Intel 32-bit
platform
ia64 contains the setup.exe and all of the binaries for the Windows-Intel 64-bit
platform
License Checker contains tools to help with licensing
MPE contains the components to install the agents on the MPE/IX platform
NetWare contains the components to install the agents on the Netware platform
OFM_8.1 contains the installation tools to load the Open File Manager version 8.1
OPENVMS contains the components to install the agents on the HP OpenVMS platform
OV_Integrations Contains the software to install the Openview integrations
Product_Information
autorun the executable invoked when the CD-ROM is inserted into a Windows system
Student Notes
This graphic above illustrates the HP-UX CD-ROM contents. Included are:
ADOBE contains instructions for obtaining the Acrobat Reader for HP-UX
DOCS contains the complete set of Data Protector Manuals
DP_DEPOT contains the software depot used for swinstall
LOCAL_DP_AGENT_INSTALL contains the omnisetup.sh script for local agent install
and cell manager upgrade
MISC some unsupported tools
OV_INTEGRATIONS contains the software for the OV Integrations
ReadMe.Solaris contains an overview of the CD-ROM
# /opt/omni/bin/xomni &
Student Notes
The Cell Console (user interface) on HP-UX and Solaris is called xomni, and it is located in
the /opt/omni/bin (<OMNIHOME>/bin) directory.
The Data Protector GUI may be started with the /opt/omni/bin/xomni command.
While the GUI is the recommended tool for configuring Data Protector, it is possible to make
configuration changes by using the command line interface or by editing the configuration
files in the /etc/opt/omni/cell and /etc/opt/omni/users directories.
The users that have access to the Cell Manager are registered in the
<OMNICONFIG>/uses/UserList file. This file will need to be modified to allow all
distributed GUI’s to access the cell manager. If there is no local GUI running on the Cell
Manager, then this file should be edited before any remote GUI will be able to connect to the
Cell Manager. To get started, add a new entry (on a single line) containing four asterisk
characters separated by spaces, followed by the string admin.
Example:
* * * * admin
This will allow any GUI client to connect to the Cell Manager, and is not recommended as a
long term solution. This entry should be removed once a remote cell console is able to
connect to the cell manager.
When Data Protector is installed, the /etc/PATH file is updated to contain the
<OMNIHOME>/bin directory. This will allow all of the Data Protector commands in
<OMNIHOME>/bin to be available after a new login session is started. In addition, the
/etc/MANPATH file is updated to allow for easy access to the man-pages.
The Data Protector administrator may want to add the <OMNIHOME>/sbin directory to the
PATH on the cell manager system. This will allow simpler access to the binaries for some cell
management tasks.
Start -> Programs -> HP OpenView Storage Data Protector -> Data Protector Manager
Student Notes
The user interface on Windows systems is accessed via the Start button. The actual
program running is the <OMNIHOME>/bin/manager.exe. The <OMNIHOME>/bin
directory is not added to the system PATH by default. The Administrator may want to add the
“bin” directory to the system Path for simpler access to the Data Protector executables.
Additional commands exist in this directory for command-line execution; simply make the
<OMNIHOME>/bin your working directory, and execute the programs by name if the
directory is not added to the Path.
All of the programs that make up the command line interface are documented in the
<OMNIHOME>/Docs/MAN directory in a single file named CLIReference.pdf.
Client
context
Student Notes
Adding an Installation Server
When the Cell Manager is installed, the Installation Server software is also installed. This
default Installation Server does not need to be registered with the cell manager in order to be
used. Additional Installation Servers needed by the Cell Manager must each be registered
after they are installed. This is true for Windows Installation Servers and HP-UX Installation
Servers in the cell.
2. Select: Edit -> Add -> Import Installation Server from the Menu Bar.
3. Add the name of the Windows server in the Name Field or select it by Browsing the
Microsoft Windows Network, and click Finish.
NOTE: You cannot remotely install a Data Protector client on the Windows
installation server system. To use the same system as both an installation
server and a client, install the client components first and check the feature to
include the installation server.
1. As with the cell manager installation for HP-UX, run swinstall. This time, select only
the sub-product of Data Protector, called OB2-IS, and complete the installation.
4. Add the FQDN (Fully Qualified Domain Name) of the HP-UX server in the Name Field,
and select Finish .
Clients
context
Cell
Actions
Student Notes
The Data Protector user interface supports cross platform client installation. Software is
distributed to clients using the Data Protector user interface (GUI)
In the Data Protector Client context, select Clients, and Add Clients from the pop-up menu.
Select an installation server to use, then select which components to install on which system.
Possible components are the disk agent, media agent, the cell console (GUI and command
line interface), as well as the various integration components.
The GUI shows which agents and versions are installed on the client systems. This
information is also stored in the <OMNICONFIG>/cell/cell_info file, which can be used
if the GUI is not available.
The next set of slides illustrates the various installation and deinstallation options that are
available for these clients through the supported GUIs.
If the system is remote, rlogin is attempted as the root user. If this is a first time
installation of Data Protector, a password is requested. If Data Protector has previously
been installed (version 3.1 or greater), it does not require a password.
1. Checks to see if the client already has the agent software installed. If so, what version it
is.
2. Determines if available disk space on the client system meets the needs requested in the
require.dat file.
3. Copies the package containing the software onto the client system. If client system is in
the .rhosts file, a normal rcp command is used. An ftp script is generated if the client
system is not in the .rhosts list, and ftp is executed to copy the files.
4. Installs software. Next, the installation procedure found in the utils package is used to
install the software on the remote system (omni_rinst.sh).
5. Updates information about currently installed software packets and configuration on the
client system. From this information, Data Protector can determine which version of the
software is currently running on the client system and from which cell manager it was
installed.
Files:
<OMNICONFIG>/cell/cell_server
<OMNICONFIG>/cell/omni_info
6. The script performs some cleanups and deletes all temporary scripts, utilities, and files
created on the client system.
7. Prints a summary for every packet requested to be installed on the client system.
From within the Windows GUI, right clicking on the Clients icon accesses the new client
window. Alternately, the Add New Client icon can be selected from the Tool Bar in the
top right-hand corner of the same screen.
3. Starts setup with the option -components component1 component2 -client host1 host2
host3 ...
4. Setup checks for access permission to the remote registry. If it fails, it proceeds with the
next host.
5. Setup checks if Data Protector has already been installed (update). If not, it asks in
which directory the software should be installed.
6. Setup copies all the files to the destination directory. If the destination directory is not
available as the default shared disk (\\hostname\C$ for C: and \\hostname\D$
for D:), the user is prompted to give the correct shared disk.
7. Using the remote service manager, it starts the inet service on the remote host.
8. When started, inet contacts the Cell Manager and instructs it to import the host into the
cell. (In some cases, a manual import may still be necessary to complete the registration
process.)
Select
client
Pop-up menu
using mouse
right-button
Student Notes
The Add Components option is used when a client already exists within the cell, but more
agents or integrations are required. An example might be if an HP-UX server has been
installed with disk and media agents (i.e., it has data to be backed up and contains a backup
media of some sort). If it has an Oracle database added to it, then a further integration is
required for online backups to occur. This is achieved via the Add Components screen.
Importing Clients
Select
clients, use
pop-up menu
Student Notes
Importing an HP-UX Client
To move a system from one cell to another cell (or to remove a system from your local
network), use the export/import actions. These actions do not install or delete the Data
Protector client software; they simply amend the configuration on the cell manager and client
machines. An option is available to remove the Data Protector software if this is necessary.
The relevant files for export and import are:
Importing a Cluster
Data Protector supports the HP MC/ServiceGuard clusters running on HP-UX. The Data
Protector clustered systems must be installed locally from the CD-ROM on every system
within the cluster, and then manually imported to the Data Protector cell using the graphical
user interface and specifying the virtual names for floating IP addresses. Use the Import
Cluster feature in the GUI for this function.
On Windows the cell_server information is also kept in the registry, if entered incorrectly
during the installation process, this entry may be altered manually using regedit:
HKEY_LOCAL_MACHINE\
SOFTWARE\
Hewlett-Packard\
OpenView\
Omniback II\
Site\ [fully qualified name of the cell manager]
Note: During the Windows client installation, specifying the name of the Cell
Manager is optional, simply leave the field empty and continue the install
process. In this way the import must be used to register the client with the
Cell Manager.
Importing a Cluster
Data Protector supports the Microsoft Cluster Server (MSCS) for Windows. The Data
Protector cluster-aware clients must be installed locally from the CD-ROM on every system
within the cluster, and then manually imported to the Data Protector cell using the graphical
user interface; select “MS Clusters” in client context, and use the Import Cluster tool or pop-
up menu in the GUI.
Option
to delete
software
Student Notes
Exporting (Deleting) Clients
Exporting a client system from the Data Protector cell removes the client references from the
Data Protector database and configuration files on the Cell Manager and client system
without uninstalling the software on the client computer (unless selected). This can be done
using the GUI or the omnicc -export_host command. A client export may be required in
the following situations:
• The client needs to be moved to another cell.
• The administrator wants to remove any systems from the Data Protector cell
configuration that are no longer part of the network.
The software components may be deleted when a client is exported from the cell when using
the Data Protector GUI. If the client is to be imported into another cell, do not remove the
software.
Licensing
Evaluation
120 Days
Permanent
Instant-On
60 Days
Emergency
14 Days
Default installed
with product
Installed using :
• GUI
• Command Line Interface
• Editing lic.dat
Student Notes
When you install Data Protector for the first time, it runs with an instant-on license, which is
valid for 60 days. Furthermore, in special cases you may be provided with a temporary
license which is valid for three months. This means that you can use Data Protector for up to
three months without any permanent license. During this time, you should set up and
configure your Data Protector environment, and request your permanent license string.
In the event of a loss of the Cell Manager and subsequent recovery using a new system, an
emergency license valid for 14 days may be obtained from HP Customer Support.
After you receive the permanent license string, you can install it with the Data Protector
Installation GUI, or using the command-line interface. You can also use an editor to type the
string into the license file, <OMNICONFIG>/cell/lic.dat. You must then issue the
omnicc command to activate and verify the changes.
Use the command omnicc -query to display the current license information, and the
omnicc -password_info for a more extensive report.
Student Notes
Shown above are the licenses available for Data Protector 5.1. The Starter Packs are for the
Cell Manager including a single drive license with unlimited clients.
NOTE: All Omniback 3.5, 4.0, 4.1 and Data Protector 5.0 licenses will work with Data
Protector 5.1. To take advantage of new product features, additional licenses
must be purchased.
The Cell Manager installation includes a license.txt file which may be printed and then faxed
to the HP Password Delivery Center to obtain a permanent license keys.
NOTE: For further information on licensing refer to the HP OpenView Storage Data
Protector Installation and Licensing Guide. B6960-90079
• Schedule a backup.
Purpose
Gain familiarity with the GUI
Authorize remote administration
Understand the backup concept
Add a simple tape device into the cell
Format media for backup
Perform a simple backup
Schedule a backup
Perform a restore
Respond to a mount request
Execute a report
Student Notes
The purpose of this module is to gain familiarity with the Data Protector product and its
associated GUI. This module will serve as a tutorial to the basic concepts behind Data
Protector backup.
At this point you have installed the product and created a cell, this module will guide you to a
point at which a simple backup can be initiated. Several steps are required; some are simple
checks; others are configuration tasks. The end-to-end process will introduce many of the
initial features within the main Data Protector GUI. Each of the features and functions will be
discussed in much greater detail in later topics in this training. The purpose of this module is
to have you explore the functionality of Data Protector by configuring the cell to perform a
simple backup and restore.
• Using the new logical device to initialize a tape into a media pool
Menu bar
Tool bar
Context list
Results
area
Scoping
pane
Navigation
tab Results
tabs Status bar
Student Notes
HP OpenView Data Protector Main GUI
The slide above depicts the main Data Protector GUI provided when an administrator
executes the xomni(UNIX) or manager(windows) command. The GUI contains several
contexts, each designed to allow for control of a specific functional area.
UNIX: /opt/omni/bin/xomni
Windows: Start -> Programs -> HP OpenView Storage Data Protector -> Data Protector Manager
or c:\program files\Ominback\bin\manager
The Scoping Pane: Provides a tree-like structure of items that may be selected to allow for
configuration or properties for the selected item.
The Results Area: Provides properties for selected items as well as configuration
procedures for the selected items in the Scoping Pane.
The Navigation Tabs: Objects and Tasks that appear at the bottom of the Scoping Pane.
They are used to switch between the Object list, and configuration Tasks.
The Results Tabs: Allows several tasks to be executing in parallel from the single GUI.
Each tab shows the results of a particular task.
• xomni -server <Host> to start the user interface for the complete Data Protector
functionality and connect to a specific Host which is a cell manager
• xomnibackup to start the interface for the backup context (xomni –backup)
• xomnirestore to start the interface for the restore context (xomni –restore)
• xomnimonitor to start the interface for monitoring a single cell (xomni –monitor)
• xomnimm to start the interface for devices and media (xomni –admin)
• xomniadmin to start the interface for users, clients, database and reporting
(xomni –users –clients –db –report)
• manager -server <Host> to start the user interface for the complete Data Protector
functionality and connect to a specific host, which is a cell manager
• manager -backup to start the user interface for backup
• manager -restore to start the user interface for restore
• manager -monitor to start the user interface for monitoring a single cell
• manager -admin to start the user interface for devices and media
• manager -db to start the user interface for the Data Protector Database
• manager -users to start the user interface for users
• manager -clients to start the user interface for clients
• manager -report to start the reporting interface
• manager -? to see a list of options
NOTE: Data Protector also provides a “snap-in” for the Microsoft Management
Console (MMC). If you have the MMC loaded, you may add the OB2_Snap as a
standalone Snap-in.
Most of these components of the GUI and the command interface will be discussed in further
detail throughout the rest of this training course.
You may want to view the man-page or WordPad files for omniintro and omnigui for an
overview of all the commands available to Data Protector.
export MANPATH=$MANPATH:/opt/omni/lib/man
man omniintro
man omnigui
<OMNIHOME>/config/users/UserList (Windows)
<OMNICONFIG>users/UserList (UNIX)
Student Notes
In many Data Protector installations, more than one Cell Console is distributed during the
installation process. In order to be able to access the Cell Manager remotely via the
distributed Cell Console it must be authorized.
By default, only the Cell Console installed on the Cell Manager system can access the cell
server process. Any attempts to use a remote cell console will be blocked; a permission-
denied message will appear on the system where the console is running.
To authorize another Cell Console, you must add a remote user to the User configuration. If
the user is to be a remote administrator, add them to the Data Protector “admin” user group.
The screen above illustrates the necessary steps:
Select the Users context, select the admin group, then use the right-mouse button to access
the context-sensitive pop-up menu, select “Add Users.”
2
Fields in the
UserList file 3
1
4
Student Notes
You will need the following information to add an additional user of the Cell Console:
• The Platform type (where the console is installed UNIX or Windows)
• The User Name (operating system user ID of the person authorized to use the cell
console)
• The System Name (where the cell console will connect from)
• All of the above entries may contain <Any> instead of a specific name. This is essentially
a wildcard. Use with caution!
NOTE: The Cell Console system does not need to be a client of the cell; this allows an
authorized cell console to connect to many different cell managers.
“Any” * * * admin
The above entry allows any user from any cell console (user interface) to access the cell
manager; this is not a secure long term solution, but does allow for quick remote access to
the cell manager. Once access is gained remotely, modify the user configuration as necessary
to tighten up the security.
System-A
Backup Specification
Objects Disk1 DA System-C
SystemA - Disk1 Logical Device
Student Notes
• The Data Protector product performs either local or network backups using the Disk
Agents and Media Agents that are installed onto the various systems in the Data Protector
Cell.
• The Backup Specification is essentially a configuration file that contains a list of the
objects to be backed up along with the devices to use for the backup. Disk Agents are
used to access the object data, and Media Agents are used to write to the backup devices.
• The physical devices attached to a system are configured into Data Protector as a Logical
Device, allowing additional features to be used for backup and restore. One of the
features of the Logical Device is a Media Pool.
• The Media Pool groups tapes together into a logical set, and has policies for how the
tapes in the set may be used and accessed. Media Pools may be assigned to a Logical
Device when the device is configured, and become the default set of tapes to be used
during backup when a logical device is used.
• Data Protector backups are configured such that tape selection is done automatically
during the running of the backup job. You assign an Object to a Device, which is already
assigned to a Media Pool. Data Protector then chooses an appropriate tape for the
backup. If no tapes are available, Data Protector will issue a Mount Request for the
desired medium.
NOTE: We will explore all of these topics in much detail in the modules following this
introduction.
Backup Specification
Student Notes
Before going through the individual steps of configuring a backup, it is helpful to know how a
backup is defined and processed with Data Protector.
Data Protector requires the following fundamental components for all backups:
• A list of what is to be backed up. Data Protector refers to the data source as an Object.
Data Protector supports specific object types, which will be discussed later in this
module.
• A list of what devices Logical Devices to be used. (Details covered in Module 6)
• A set of media for the backup to be written to. This is in the form of a media pool.
(Details covered in Module 5)
• The options that are to be used for the backup. Data Protector provides many flexible
options that can be used to completely define all characteristics of the backup and the
information relating to it. These options will be discussed in throughout this module.
These components can be grouped together and saved as a backup specification. Backup
specifications are used to run repeated backups of the same source data, either manually
invoked or scheduled.
Backup Checklist
Student Notes
The slide illustrates some of the main tasks that are necessary to complete an Data Protector
backup. The remainder of the module will guide you through these steps, with a lab at the
end for you to work through them on your own.
Backup Wizard
Data Protector provides a “wizard” to guide you through the main steps in configuring a new
backup specification. The wizard is not covered here, but rather the steps used within the
wizard. To use the Next Step Wizard, select it from the View menu in the Menu Bar:
Verify Agents
Client
context
Select
clients
Property list
shows installed
agents
• Initialize Medium
Student Notes
The initial installation of the cell manager installed three key components:
• User Interface — Allows the GUI to be started, also called the Cell Console.
• Disk Agent — Allows data to be backed up from the cell manager
• Media Agent — Allows a backup device to be configured on the cell manager
The Clients context on the Data Protector GUI provides the list of the client hosts, along
with the software that has been successfully configured on each.
TIP! You may see a report from the command line as follows:
omnicellinfo –cell brief
Before configuring devices or backups, be sure that each system in the Cell has the correct
software components (agents) installed.
Device/media
context
Select
media
Property list
shows configured
pools
• Install Disk and Media Agents
• Initialize Medium
Student Notes
The Devices & Media context in the GUI is used to configure media into the relevant pool (a
media pool is a logical collection of media). It is also used to configure logical devices for
backup and restore (discussed in the next slide).
During the initial configuration of Data Protector, a default media pool was created for each
type of supported medium, you may use or remove these pools as needed.
In this module, you will perform a backup to a tape that will be part of the standard media
pool called Default DDS. To check that this pool is available, click the Devices & Media
context in the GUI, select Media in the Scoping Pane. The Default DDS pool should be listed
in the Results Area. It should have a 0 in the column #Media indicating that there are no
tapes in this pool yet.
Configure a Device
Select
devices
Properties list
Pop-up menu shows existing
with mouse devices
right-click
• Initialize Medium
Student Notes
As mentioned in the previous slide, the Devices & Media context can be used to configure
logical devices. The logical devices are the entities that represent physical devices on a client
system. They are used to initialize tapes as well as perform backups and restores. You must
create a logical device on the cell manager that can be used to perform the backup at the end
of this module.
Device Specification
Logical device
name and
description
Student Notes
11. Select the Next button (at the bottom of the Results Area)
12. Verify that the Media Type is DDS, and Default Media Pool is Default DDS, select the
Finish button.
Highlight
desired pool
Select
media
Pop-up - context
sensitive menu -
mouse right-click
• Initialize Medium
Student Notes
Before a backup can take place to the logical device that has been created, some media
should be added to the pool.
TIP! Data Protector can use blank media, so you do not have to initialize it. At
backup time, the blank media is simply added to the pool name specified by
the logical device into which the media was inserted. See previous slide for
specifying the pool name for a device.
In this case, we will use the new logical device to initialize a tape and add it to the default
DDS pool.
Double-click Media in the Scoping Pane; this opens the Media Pool list.
1. Select Default DDS, select Format from the pop-up menu (use the right-mouse button)
2. Select the DDS-Practice device from the pull-down list next to the Device field.
3. Select the Next button.
4. Select the Automatically generate option (selected by default)
5. Select the Location field, and enter “Device Repository”
6. Select the Next button
7. Select the Force Operation button (we assume to have a previously used tape)
8. Select the “Default” for Medium Size.
9. Insert a tape into the tape drive (be sure it’s not write protected).
10. Wait for the device to be ready.
11. Select the Finish button.
Messages should appear in the Results Area window, and the tape should be initialized within
a short period of time, this may take several minutes depending upon previous usage of the
medium.
Format Medium
Media
checked
before format
proceeds
Unique
medium-id
assigned
Session
summary
Session
summary
pop-up window
Student Notes
The graphic above illustrates the messages displayed during a media initialization session.
Note the following:
• Data Protector reads the tape before writing to it.
• Data Protector will not initialize (without the force option) a used tape.
• Data Protector adds a unique MediumID to each tape initialized, and stores this in its
internal media management database.
• The default label for a tape (when auto-label is used) is the pool name and a sequence
number within the pool.
• All tapes must be labeled (formatted) before use. This consists of simply writing a header
to the tape and registering it in the media database.
Pop-up
File system menu
selected
• Initialize Medium
Student Notes
Now that we have a logical device and media in a media pool, we are ready for backup. A
definition must now be created to tell Data Protector which objects to place on the tape. This
definition is called a Backup Specification or Datalist. Datalist is synonymous with
Backup Specification and is more commonly used to refer to the file containing the backup
specification. The directory that contains the backup specification file is called datalists,
and is in the <OMNICONFIG> directory.
The Backup context in the GUI opens the Backup tools. There are many options for creating
the datalist file. In this module, we will address only the ones required for a simple backup.
The Filesystem folder under Backup Specification in the Scoping Pane is the place to
start, select it, and use the menu as shown on the graphic above to add a new Backup
Specification.
Blank
template
Direct
attached
or remote
Student Notes
With the Create New Backup window open, select the Blank Filesystem Backup,
then select the OK button.
Select
objects
Select Next>
for
more options
Student Notes
For this first backup, a relatively small object is desirable; locate a directory or filesystem
that is suitable.
1. Select the “plus” in front of your system to open the object list, repeat if necessary to see
the directory to backup. (Double-click to open the list as well.)
2. Select the small check box in front of the desired object or directory, a small blue check
mark will appear in the box, this indicates that it is selected for backup.
3. After you have selected (checked) the items for backup, select the Next button.
4. Select (check) your logical device (it may be selected by default), select Next. Leave
most of the options as their default values, except as indicated below.
5. Change the file system options: Protection to 4 days.
6. Select Next. (You will see the scheduler.)
7. Select Next. (You will see the job summary.)
8. Select Next. (You will see Save, Start, and Preview buttons, go to the next page)
Save backup
specification to Select finish to
a file complete the
definition
• Initialize Medium
Student Notes
Data Protector backup specifications are stored as files in the <OMNICONFIG>/datalists
directory. From the Backup Results area, you must perform a save, in order to keep the
specification.
1. Select Save as.
2. Enter a name for the file (for example, exercise-1); leave the group name, default.
3. Click OK.
The new backup (datalist) will appear in the Scoping Pane.
Schedule tab
for recurring
backup
Select the
saved backup
specification Start an
interactive
backup
• Initialize Medium
Student Notes
When Data Protector invokes a backup, it does so using a datalist file. The datalist definition
ensures that Data Protector knows which objects to backup, which to ignore, and which
device(s) to use. To start the backup, perform the following:
1. In the backup context, select a backup specification then:
Start Backup … from the popup menu
2. You will be prompted for the Backup Type (Full) and Network Load (High).
4. The Backup Results Area window will become a Backup Monitor window, and
will show a session running.
5. The file system object will switch from Pending to Running to Completed
6. The logical device part of the window switches from Inactive/Waiting to Running,
and shows that amount of KB of data backed up.
7. The messages part of the window updates the progress as various file system objects are
backed up.
The Scheduler
Student Notes
Once a backup specification has been created, the scheduler can be used to execute the
backup at a predefined date or time. The scheduler can also run the backup on a regular
basis, defined by the administrator.
Select Add…
for recurring
backup
Student Notes
The Data Protector Scheduler offers many possibilities for producing re-occurring backups.
Each backup specification may have a single schedule file, but be executed at many different
times, each time with a different scope (full, incr #, incr).
Notice on the picture above, that you may not schedule backups in the past! Data Protector’s
scheduler only allows for forward scheduling.
Select Add… to create a new schedule, or simply double-click on a particular day to schedule
a backup. You may override the protection for individual backups.
• Holidays
The scheduler can be configured to skip backups on days that are defined as holidays.
(blackout days) Data Protector can use the standard hp-ux style calendar holiday file as
input. Holidays are seen in the schedule window and are color-coded black.The Holidays
file resides in <OMNICONFIG>/Holidays.
Selection
determines
options
Student Notes
Backups may be scheduled to be recurring; they may execute daily, weekly, or monthly. Each
schedule (file) may contain multiple time parameters. Backup specifications may also be
setup to start executing at some future date.
• Backup Type
Backup Types
Incremental
Incremental
plus level
Student Notes
As is common with many other backup mechanisms, Data Protector provides a method of
performing full, and various levels of incremental backups:
• Full
Everything is backed up.
• Incremental 1 - 9
Data Protector tries to find the latest protected backup session with a lower backup level
on which the incremental backup will be based. With any incremental backup the entire
directory tree is backed up; so even if no files changed, the directory tree is recorded.
For example, if the user starts backup with backup type incr5, Data Protector will search
through the database to find the latest protected version of the object with a lower level
(full or incr 1-4) backup level, and use it as the reference point version on which
incremental will be based. A full backup is equivalent to level 0.
NOTE If Data Protector cannot find a valid reference point version of the object
(hostname, mount point and description identifies an object) on which the
incremental backup is to be based, it will start a full backup. This can then
be seen in the monitor screen (backup type will be Full instead of
incremental). This backup promotion option may be disabled by changing
the UpgradeIncrToFull global option (default is 1, on).
• The object is not completed (aborted and failed versions of object) during the backup
session.
• Object versions without all media (one or more of the media on which the object was
stored has been overwritten/exported or is not protected any more).
• Full restore chain for object version is broken (full restore chain is broken if the chain
from full to this object version has some missing versions or if one or more object
versions from the chain are invalid)
NOTE Use the UNIX command touch -m with options to change the modification
time on a file.
Last Full
Incr Incr Incr Incr
Incr n
Incr n
Incr n
Incr n
Incr Incr Incr Incr Incr
22:00 12:00 22:00 12:00 22:00 12:00 22:00 12:00 22:00 12:00
Day 1 Day 2 Day 3 Day 4 Day 5
Student Notes
There are three different backup models illustrated above, each has certain characteristics
that may be desirable for certain environments.
Ä Nightly leveled incremental increases in size and duration and includes changes since
the previous nightly leveled incremental backup
Context Tabs to
change data
view
Highlight
session
Object
properties
Student Notes
After your backup finishes, you can review the operation from the database context within
the GUI. Open the Session folder of the Object folder to view the data.
Perform a Restore
Context
Tabs to change
restore options
Highlight
object
Pop-up dialog
Select
items to allows for media
preview
restore Start the
restore
Student Notes
In the final part of this module, a restore will be started to retrieve a file from the tape
following its removal from the Cell Manager.
NOTE: This restore should be performed without the tape in the tape drive to force
Data Protector to generate a Mount Request for the tape.
Start a Restore
As the root/Administrator user on the system where the backup was performed remove
a file or sub-directory (folder) that was included in the previous backup job.
1. Remove the backup tape from the drive. If you have another tape, insert that one instead
(it makes the mount request appear sooner).
5. Select the “List restored files” and “Display Statistical Information” options.
6. Start the restore by selecting the Restore <object_name> button on the bottom of the
Results area.
7. Select Finish within the “Start Restore Session” pop-up window to proceed with the
restore operation.
8. If you ejected your tape from the device, Data Protector will prompt you (after a short
delay, which could be several minutes) with a mount request for the specific tape to
complete the restore operation.
Device needs a
specific tape to
continue
Student Notes
Data Protector uses a Mount Event to notify you that it needs a particular tape loaded into a
logical device. The Mount Event is used for both backup as well as restore. Having started a
restore and created a situation whereby there is a mount request outstanding, we must satisfy
that request in order for the files to be restored and for the restore session to complete.
Mount requests will remain pending until they are either confirmed or cancelled.
Student Notes
Data Protector will present the above window in the form of a pop-up when you are
performing interactive operations, such as backup and restore. You may use it to confirm the
mount request once you have placed the appropriate tape within the device. If you close this
pop-up window, you may still respond to the mount request, see the next page.
Monitor
context
Highlight
device
Student Notes
Mount requests may be handled within the monitor context. Select the session that is in the
Mount Request state, then select the device waiting for the tape, and confirm the request
from the Actions menu. You may also use the pop-up menu while selecting the mount request
with the right mouse button.
NOTE: Be sure to place the tape in the required drive before confirming the request,
or you will be back to the same mount request state.
Introduction to Reporting
Reporting
context
Wizard to
configure report
parameters
Report
to execute
Tasks
scope
Student Notes
Data Protector includes very extensive reporting and notification capabilities. These will be
discussed in much more detail later. You may want to experiment with the reporting tasks
after you have run backup to see the kind of information that is available from the internal
database.
There are several ways to have reports generated. Use the report tasks wizard to execute
reports interactively.
Reporting
Select report
parameters
Finish to
view
report
Student Notes
Data Protector comes with several categories of reports that may be executed against the
internal database. Each report selected within the reporting wizard will require different sets
of optional parameters in order to execute.
The example above demonstrates the selection of a session for a single session report. Data
Protector provides pull down lists within the reporting wizard to make report parameter
selections very simple. You may Cancel or Finish to abort or complete the report execution.
Objectives
Student Notes
This module focuses on the basic knowledge and skills necessary to successfully implement
tape libraries for use with Data Protector. Today HP manufactures and sells a variety of tape
devices ranging from single drive DDS units to multi-drive ESL libraries.
This section will provide general tape library implementation techniques, yet focus on the HP
StorageWorks MSL product line of midrange tape libraries. Student without this particular
tape library brand will benefit from the implementation and troubleshooting concepts
presented here.
Students pursuing HP Certification in the HPCP program for tape libraries are encouraged to
pursue further studies; including the HP course for library installation as well as the MSL
Library WBT available on the HP ITRC training site (also available on the classroom
systems).
Library Terminology
• Tape drive(s)
– Data transfer element
• Repository slots
– Cartridge/magazine slots
– Storage element
• Media exchanger
– Transport element
– Robotics
• Barcode scanner
• Mail-slot(s)
– Import/export slot
– Eject element
• Management interface card
– SCSI (scsi2, scsi3)
– NSR (fibre channel/SCSI bridge)
Student Notes
The tape library is a complex system used for near-line or off-line storage of data. Data is
typically written onto high capacity tape cartridges by utilizing multiple tape drives
simultaneously. The tape library system differs from a standalone tape drive in many ways,
not the least of which is the automated handling (load/unload) of media to/from the
embedded tape drives.
• tape drive(s)
• repository slots
• media transport
• mail slot(s)
• barcode scanner
• management interface
The SCSI interface within the library presents the various components to all attached host
systems as objects. There are usually four objects, commonly referred to as elements that are
presented:
• transport
o the robotic that moves tape cartridges between slots and drives
o in SCSI terms, “Medium Transport Element”
o the MSL 5000/6000 Series Libraries have a single SCSI transport
• slot
o repository of tape cartridges, stored in magazines in the MSL
o in SCSI terms, “Storage Element”
• drive
o tape drives such as DLT, SDLT, LTO
o in SCSI terms, “Data Transfer Element”
• ports
o commonly referred to as import/export slot or mail slot
o MSL libraries may have one (5U models) or two mail slots (10U models)
o in SCSI terms, “Import/Export Element”
o other tape libraries (non-MSL) may have external access ports or other
numbers of mail-slots (typically 0, 1, or 5)
The management interface card varies depending upon the library. HP MSL libraries may be
equipped with either Ultra-2 SCSI (MSL 5000) or Ultra-3 SCSI (MSL 6000) controllers in
addition to a Fibre Channel interface called the Network Storage Router (NSR E-1200). The
NSR provides Serial, LAN, FC and SCSI ports in a single cPCI card that is installed into the
card cage at the rear of the MSL library.
• Background
– Designed for backup operations with high-end networks.
– Features high availability and maximum storage density.
– Combines high-end tape drive technology with advanced
robotics.
• HP Midrange Libraries
– MSL#xyz – medium storage library
– MSL5000 series library GUI
touch-screen
– MSL6000 series library
– (xyz indicates slot count)
Student Notes
Background
• Designed for backup operations with high-end networks and high-performance servers.
• Features high availability, maximum storage density, and easy serviceability.
• Combines Digital Linear Tape (DLT), Super DLT (SDLT) or Linear Tape-Open (LTO) drive
technology with advanced robotics.
• Features a GUI-Touch Screen for configuration and management of the library.
History
• The StorageWorks MSL5000 series library models are the first generation.
• Introduced under Compaq brand. Some still sold under this brand.
• The HP brand was introduced with LTO Ultrium 230 (Gen 1) tape drives and
supports high-end SDLT tape drives as well.
• The MSL6000 series library models, announced in April 2003, are the next generation
of MSL libraries that include features such as:
• Auto-power-on
• Ultra3 SCSI interface
• LTO Ultrium 460 (Gen 2) drives
Native capacity for available MSL drives based upon drive type:
External Access
enter/eject tapes
Mail Slot(s)
Configured in Repository
0, 1, or 5 slots
3D views available at http://www.smb.compaq.com
Student Notes
One of the most common ways of entering and ejecting individual tapes from a tape library is
via the mail slot. There are various names for this Import/Export Element and different ways
for libraries to provide it.
On the MSL libraries the first slot in the left magazine tray on each of the upper and lower
drawers is the 10U models are the mail slots. (one in the 5U models). Other legacy libraries
have implemented this as an external access port, or as 0, 1, or 5 reserved slots within the
tape magazines.
Once implemented, the contents of the mail-slot may be managed by Data Protector via the
library manager functions of enter and eject. See the Logical Devices module for more
details.
Other slots within the library are numbered beginning with slot 0 (default is zero; may be
changed to start with slot 1 as the first position) which is adjacent to the mail slot. For the
10U models this slot is in the upper left magazine. The remaining slots in this drawer are
number 1-13. Slots 14-28 are in the right (upper) magazine. If so equipped, the lower left
magazine slots are numbered 29 to 42, and 43 to 57 in the right.
In the MSL6060 there are 58 fixed cartridge slots and two mail slots for a total of 60 slots. The
MSL 6030 has 30 slots, one of which is used for the mail slot.
LAN port
Serial (RJ11)
port
[NSR E1200]
Student Notes
The graphic above illustrates the rear panel components of the MSL libraries. (10U shown; U
is the rack-unit measurement, each U is approximately 1.75 inches; the 5U model is half the
size of the 10U)
Library Controller
The library controller board contains a single microprocessor and associated logic
devices to control robotic operations and manage overall library functions. The
microprocessor also manages the SCSI interface between the library and the host system.
Student Notes
The graphic above shows an LTO Ultrium tape drive and shoe mechanism removed from the
library.
Tape drives are mounted at the rear of the library in a hot-pluggable shoe that permits a tape
drive to be removed and replaced while other tape drives and the library robotics remain
active. The hot-pluggable capability of the tape drives result in uninterrupted SCSI bus
activities during removal or installation.
The Ultrium tape drive shoe assembly has a slightly smaller base than the DLT/SDLT tape
drives. The Ultrium 230 and the Ultrium 460 may be differentiated by the Ultrium logo at the
rear of the drive.
Note: The LTO-2 media cannot be used in the Ultrium 230 tape drive. The Ultrium
460 will accept both LTO-2 as well as LTO-1 media.
The DLT tape drive load handle extends from the rear of the drive shoe.
As of April 2003, media partitioning is supported only on the HP MSL 5026 and MSL 5052
models with DLT 40/80, SDLT 110/220 and SDLT 160/320.
SCSI Interface
Student Notes
With the recommended standard Ultra/Wide LVD/SE SCSI HBA interface, the data transfer
rate of the module (shown above) is achievable. Actual rates depend upon the type of drive,
number of drives, and the number of drives connected to the SCSI bus.
Each drive in the MSL 5000/6000 series library has a maximum sustained rate of twice the
data transfer rate with 2:1 compression on the data.
Each of the tape drives and the library controller constitute an independent SCSI target.
When any two or more devices are connected to the same SCSI bus, each must be assigned a
unique SCSI ID. HBA’s are typically assigned SCSI ID 7 by default.
The library is equipped with a Low Voltage Differential and Single-Ended (LVD/SE) SCSI
interface. The standard configuration is a SCSI LVD/SE using two VHDCI-series (ultra high
desity) 68-pin, Micro-D SCSI connectors.
SCSI LVD differs from SCSI-SE in that it overcomes the distance limits imposed upon the
single-ended standard through an enhanced signaling scheme. This signaling scheme has
made the two signaling mediums incompatible on the same bus. The LVD bus must operate
with only LVD devices connected; otherwise the bus will default to single-ended if any SE
devices are connected (including the use of a SE terminator instead of the normal LVD
terminator.
Note: LVD and HVD (High Voltage Differential) will not operate together on the
same bus. The bus itself won’t function with both types of devices connected.
To connect the library to the host system, the host must have at least one Wide SCSI
controller and the appropriate driver software. The controller must support LVD/SE.
SCSI Evolution
SCSI-1
The maximum data transfer speed for this implementation of SCSI is 2 to 4MB/s (actual
average is around 2.5MB/s), using a limited instruction set. Under SCSI-1, all devices use
different commands.
SCSI-2
SCSI-2 (referred to as plain SCSI) is the second-generation SCSI standard. It consists of the
basic SCSI-1 standard with many additions and some deletions.
Two alternative signaling systems are available when implementing SCSI-2:
Single-ended interface — This is “regular” SCSI and uses the type of conventional
signaling that is used on other buses.
Differential interface — The differential SCSI bus minimized the
potentialbottleneck created by bus length limitations experienced with single-ended
SCSI.
These two alternatives are incompatible, resulting in two main groups of SCSI devices and
controllers that cannot be mixed on the same bus. It is possible to use special converter
hardware to transform a single-ended bus into a differential one (and vice versa). Single-
ended implementations are the most common. They are suitable for internal cabling.
Differential interfaces are used externally.
SCSI-3
SCSI-3 defines new physical-level transports, IEEE 1394 and Fibre Channel, as a means of
transporting SCSI data packets. SCSI-3 defines a new low-voltage differential (LVD) SCSI
specification.
LVD SCSI is a technology that combines the advantages of both its predecessors. LVD uses
differential signaling techniques instead of single-ended, making the bus more stable. It will
support up to 15 devices on one cable and enables the use of external SCSI cabling up to 12m
long.
Library Performance
Student Notes
Native performance of the MSL5000/6000 series library indicated above is based upon:
the native sustained transfer rate
the number of drives
Note: the HP Ultrium drives support Adaptive Tape Speed (ATS), also referred to as Matching
Data Rate (MDR). ATS reduces performance degradation due to slower data rates as well as
minimizes drive wear due to frequent repositioning (DLT) as a result of lack of streaming.
Multi-module Configuration
The MSL series libraries support a range of two to eight rack-mounted modules configured
into a multi-module mode. This multi-module configuration provides:
more capacity by adding additional cartridges (52-240 including mail slots)
more throughput by adding more drives (2-16)
¾ For example, if the maximum of four MSL 6060 Library modules (each with
four LTO Ultrium 460 drives) are placed in a multi-module configuration, the
native capacity would be up to 48 TB and native backup performance would
be up to 1,728 GB/hour.
Storage hardware/software
Support for specific hardware and software depends upon the library model and
environment.
Types of Connection
The type of connection between the servers and clients to be backed up and the secondary
storage system affects the backup performance. This connection is typically one of the
following:
Directly connected SCSI tape device: Devices connected directly to the server through a
SCSI connection are very fast at backing up that server.
Network connection between client and backup server: The LAN bandwidth affects the
speed at which data can be transmitted between the client devices and the backup server.
Fibre Channel connection between backup server and tape device: Data transmitted
over a Fibre Channel connection to the tape device is very fast, 1Gbs(gigabit) or 2Gbs
(approximately 100MB/s or 200MB/s both support full duplex operation).
E1200-160
Student Notes
The MSL 5000/6000 Fibre Channel option kit includes:
Network Storage Router (NSR N1200 or NSR E1200-160)
Serial Cable
Two SCSI cables (.5m VHDCI-VHDCI)
Documentation
The NSR E1200/E1200-160 is installed in the PCI slot adjacent to (the left of) the library
controller board. The N1200/E1200 support Ultra2 LVD SCSI interfaces. The N1200 is an
external storage router, while the E1200 is an embedded cPCI card.
Caution! The library controller board must always be housed in the correct option slot
(right most position when facing the rear of the library). Insertion of an option
card into the library controller board slot damages the PCI backplane and
renders the library inoperable.
When installing the NSR into the Opal colored (Compaq brand) MSL 5026, there is a
requirement to install a cooling kit. See the MSL user guide for more details. There is a
section in the WBT as well, look for “Installing the Thermal Upgrade Kit” topic.
The NSR E2400 solution is designed for embedded use in the HP StorageWorks ESL tape
libraries.
The NSR E1200-160 uses a null device driver (hp_cpq_router.inf) on Windows 2000.
The NSR supports indexed maps containing a Fibre Channel controller LUN
not default
not necessary for Windows OS as long as there are less than 8 devices attached
use the Port 0 or Port 1 mapping
When controller LUNs are enabled (SCSI array controller device at FC LUN0) the Windows
2000 Device Manager will discover the controller device and prompt for the installation of a
device driver. The hp_cp_router.inf installs a null device driver and creates a device entry
under System Devices in the Device Manager. A true device driver is not required for the
proper operation of the NSR.
Library Operations/Configuration
Student Notes
Menu
o allow for viewing, configuring and operating the library, see appendix for
more details.
Online
o allows the library to be placed online or offline
o default setting is online after power-up initialization
Status
o view tape drive types
o physical tape drive status including cleaning status
o cleaning cartridge information
Security Levels
The MSL5000/MSL6000 Series Library features GUI touch screen security to prevent
unauthorized access to the library operation.
The GUI touch screen offers four levels of security. Only the first three are supported
in the field. (level1, level2, and Service). the fourth security level is Factory and is
reserved for HP.
Password
Each password is represented by four decimal digits that are stored in NVRAM in a range of
0001 to 9999.
Enabling a password at a lower level re-enables disabled higher levels to that value. As a
result, prior to accessing any higher level operation, you are prompted first to enter the new
higher level password. You can also use a higher level password to gain access to a lower
level operation.
For example, use the Service password to access the Move Media option.
Using the Service password to access the Menu option also gives full access (without
validating) to the Service operations. To restore passwords if forgotten, use Set User Defaults
using the MSL5000 Utility and diagnostic (serial) cable.
Menu Screen
• Use this screen to view,
configure, and operate the
library.
• Screen has three areas:
– View System Data
– Utilities
– Edit Options
• Each option button in these
areas brings you to other
options that further define
the desired task.
• See the WBT or user guide for
detailed information.
• GUI Simulator
(NeoSimHp_414.exe installed
on classroom systems)
Student Notes
Edit Options - lets you set library, SCSI, and network options.
The options available are:
Library
SCSI
Network
Passwords
Use the GUI simulator, installed on the student systems to practice navigating the menu
option. Choose Lightening for 2-drive 5U models, Thunder for four-drive 10U models.
Student Notes
The MSL supports remote management via the network interface once it is configured from
the GUI touch screen. The default network configuration should be changed to allow for this
remote access. There are two screens to configure, both shown above. First configure the
network identity, and then enable the Level 1 and Level 2 web logins. Each web login
provides access to a portion of the library functionality. Level 1 is primarily for operations,
and Level 2 for administration. Library maintenance is not available from the Remote
Management Interface (RMI, web enabled).
Note: the GUI touch screen defaults to a level 1 password of “1” and level 2
password of “2”. The service password defaults to “5566”. Service level access
is not available from the RMI.
The RMI allows for a different password (login) than the GUI touch screen. Enable an alpha-
numeric string for both level-1 and level-2 as desired.
Student Notes
The library is designed with many configuration options, each offering multiple settings to
support a variety of applications and platforms.
The setting of each option is stored in NVRAM in the module. For most applications, many of
the factory defaults will be sufficient. The exceptions are host and network specific settings.
Library Options:
Reserved slots (used for cleaning cartridges)
Configuring the master module
Configuring the slave module
SCSI Options:
Setting SCSI IDs
Setting element bases
Network Options:
Setting IP address (must be altered for site network connectivity)
For example set the ID for Drive 0 to be SCSI 0; Drive 1, SCSI 1, etc.
1. Navigate the GUI touch screen as follows: Status Screen Æ Menu Æ SCSI (in the Edit
Options area) Æ <the desired SCSI ID>
2. Touch the box next to the SCSI ID that is to be changed; a numeric keypad appears.
3. Select the desired ID number using the keypad.
4. Touch the Save button to confirm the change
5. Touch one of the following: OK, or Cancel.
6. Repeat the steps for each SCSI ID that is to be modified
7. Return to the main menu
Web Level 1 or
Level 2 login
Student Notes
Shown above is the “Login to the Remote Management Interface” for the HP StorageWorks
MSL. The password must be configured via the GUI touch screen. There are two different
access levels (level 1 or level 2) each with a different password. See the previous page
“Configuring Network Access” for web login details.
To access the Login screen, enter the IP Address of the tape library into a web browser (the
IP name will not work!). For example:
http://156.152.82.114
Student Notes
After a successful login, the tape library automation management console appears, allowing
for selection and access to the tape library.
Along the top of the web interface are the function buttons that allow for remote
management and monitoring.
Default Login:
User name: root
Password: password
Student Notes
HP recommends that all of the factory default values for the Ethernet configuration be
changed to site specific values.
IP address: http://1.1.1.1/
Subnet mask: 255.255.255.0
Gateway address: 0.0.0.0
User name: root
Password: password
These settings may be backed up to a configuration file and restored back to the router in
case the settings need to be recovered. Use the FTP method for this backup to a connected
host.
Configuration settings
To provide connectivity between hosts and devices, the router must establish an address on
each connected Fibre Channel network and SCSI bus.
The following slides identify configuration settings that are commonly modified and are
available in the Visual Manager UI and the Serial/Telnet UI.
• For procedural information on accessing and changing these settings, see the NSR
e1200-160 user guide:
• Chapter 3, Visual Manager User Interface.
• Chapter 4, Serial/Telnet User Interface.
Note: By default, the Fibre Channel port speed is set to 2 GB/s. Changes to the Fibre
Channel port speed must be manually set, such as for 1 GB/s. If set incorrectly
and the router is plugged into a Loop or Fabric, the unit may receive framing
errors, which can be found in the trace logs, and the fiber link light will be off
because of the incorrect Fibre Channel link speed.
Discovery mode
This feature makes it easy to discover attached SCSI target devices and automatically map
them on the INDEXED MAP for the bus/port in question.
• These devices get automatically mapped to the INDEXED map.
• HP recommendation: To map devices to the host, use Port 0 and 1, depending on the
port in question.
Note: This is a type of “mapping protection” so only known hosts will have access to
the maps.
Host device
A host system using a Fibre Channel Host Bus Adapter (HBA) will typically map devices into
the existing device-mapping scheme used by that operating system. Refer to the HBA manual
for the mapping table.
Switch WWN
Host WWN’s
Host assignment
Map Edit
Student Notes
The Mapping Menu is used to view and modify the host-to-map information for a Fibre
Channel port. Maps and hosts may be added, edited or deleted.
Fibre-Channel to host mapping provides a form of LUN security, similar to that used by most
SAN connected disk arrays. A map defines the devices accessible through a particular Fibre
Channel port (on the NSR). The Administrator then assigns each host a map for each Fibre
Channel port. Multiple maps may be created for a single Fibre Channel port.
The NSR discovers the WWN of the connected FC switch as well as the WWN of the hosts
connected via the FC switch. Shown above are the default configurations obtained without
specific mappings defined. In the default configuration, all hosts have access to all devices.
This default may not be optimum depending upon the desired level of access to the tape
library controller and drives.
Student Notes
With high-speed tape devices, consider the following:
The HBA must be able to transfer data at maximum tape speeds. The Ultrium 460 has a native
data rate of 30MB/s, double that if 2:1 compression is achieved.
For optimum performance, the total PCI bandwidth needs to be more than double the backup
rate. For a single Ultrium 460 tape drive the bandwidth must be more than 200MB/s to be able
to achieve maximum drive backup performance.
o Virus scans
o Disk defragmenters
Server Consideration
Student Notes
For optimum performance, the total PCI bandwidth needs to be more than double the backup
rate. For a single Ultrium 460 tape drive the bandwidth must be more than 200MB/s to be able
to achieve maximum drive backup performance.
Student Notes
Prior to use as an Enterprise Backup Solution (EBS) client or server, a number of preliminary
steps must be completed on the host running Windows operating systems (NT and above).
The EBS platform support matrix is available at: http://www.hp.com/go/connect under the
“Automated Backup” link.
Power Up Sequence
After verifying the configuration of the HBA, NSR and MSL, the power up sequence is very
important for host access to the devices connected to the NSR.
Power up sequence:
FC switch (usually takes several minutes to complete the boot up (5-6 minutes for
B-series switches))
MSL Library (allow boot up to complete, green LED on library front glows steady)
NSR (must be powered after MSL, or reboot and discovery will have to be run)
Hosts (may be already powered on depending upon the operating system)
The server must be installed for Fibre Channel operation with Tru64 UNIX, patched, and
configured so that it can become part of an overall EBS environment
AlphaServers utilize a robust hardware console known as the System Reference Manual
(SRM). Updates to the SRM firmware are regularly released. Firmware version 5.7 or higher
is required to function with EBS.
HP-UX includes many tools for managing devices files, the most common are:
insf – automatically creates device files (runs at system startup, or manually invoked)
lssf – display properties for specific device files
ioscan – scan and display all system devices and associated device files
HP-UX drivers for tape library controllers:
sctl – requires the manual (mknod) process for creating device files
schgr – automatically binds to library controller devices at boot time; supports insf
device file creation
Student Notes
When tape libraries are to be used with Data Protector; the Windows 2000 Removable
Storage Manager must be disabled for the tape library or Data Protector will not be able to
access the library successfully.
Shown above is the result of having the RSM enabled while trying to use Data Protector to
configure the device. The Data Protector device agent (devbra –devices) cannot properly
access the library drive(s). The device agent will not be able to access the library, and may
hang if the RSM is enabled.
Student Notes
Shown above is the simple procedure for disabling the Windows 2000 RSM for a selected
library. Once disabled, Data Protector will be able to function correctly when accessing the
library. See the next slide for details.
Student Notes
Shown above is the result of having the RSM disabled for the library to be configured with
Data Protector (covered in more detail in the next module on logical devices). Proper device
detection by the Data Protector device agent is necessary for successful for all Data
Protector media accesses (backup, restore, initialization, etc.).
Changer0:0:5:0
Tape1:0:6:0C (hw compression)
Student Notes
Once installed onto the host operating system a tape library may be tested to determine if it is
operating correctly. This is a recommended step prior to use with Data Protector. The first
step in testing is to determine the device file associated with the tape library.
Shown above are the Windows Device Manger screens showing the SCSI paths for the library
changer and tape device. Windows uses the device name: Bus: Target: Lun as the path to the
devices. For example, the path to the tape device shown is Tape1:0:6:0C, where C indicates
the use of hardware compression. The changer is identified as Changer0:0:5:0, where 0:5:0 is
the Bus: Target: LUN for the device.
The Data Protector Media Agent provides two programs for tape devices (installed in the
C:\Program_Files\Omniback\bin directory by default:
Shown above are the results of navigating the Windows (2000) Device Manager to determine
the device paths for the tape library controller as well as the tape drive(s). Once known these
devices may be tested with the Data Protector Utility Media Agent (UMA) to determine
operational status.
1. scan the system for available devices (shows the Windows device paths)
C:\Program Files\Omniback\bin\devbra –devices
2. locate the device path for the tape library robotics (media changer)
3. invoke the utility media agent to interact with the tape library (load/unload tapes,
status inquiry, etc.)
C:\Program Files\Omniback\bin\uma –ioctl <dev> -barcode
d. move <from> <to> move media from element to element (return to original
position when testing is completed; use integer value or element name)
– Fibre Channel/SCSI
ext_bus 13 1/2/0/0.8.0.255.7 fcpdev CLAIMED INTERFACE FCP Device Interface
target 17 1/2/0/0.8.0.255.7.11 tgt CLAIMED DEVICE
autoch 1 1/2/0/0.8.0.255.7.11.0 schgr CLAIMED DEVICE HP C7200-8000
– SCSI
ext_bus 2 56/52 scsi1 CLAIMED INTERFACE HP 28655A - SE SCSI ID=7
target 4 56/52.4 target CLAIMED DEVICE
spt 0 56/52.4.0 spt CLAIMED DEVICE HP C5177-7000
Student Notes
Mastery of devices and associated files on UNIX requires a bit more knowledge than for
Windows systems. Administrators must be able to build a kernel with the appropriate drivers
and in some cases even create (manually) the necessary devices files.
The HP-UX operating system typically creates device files at system startup for all known
hardware devices that are powered on prior to the system startup. After the system starts,
some additional (manual) steps will be needed to enable missing devices. HP-UX does not
continually scan for new devices; it is up to the system administrator to perform this task
when it is convenient. It is common for administrators to execute these steps when LUNs are
added to disk arrays, as well as when new devices are connected to the SAN while the system
is running. The needed device files will automatically be created after this procedure as long
as the needed device drivers are already loaded into the kernel.
Host Connections
The tape library systems typically connect to the Host Bus Adapter of the host system or
systems. The most common interface is SCSI, although Fibre Channel connections are
becoming increasingly common. In some cases, the library system will connect to the host via
Fibre Channel and then to the internal drives and library controller with SCSI. When this is
the case, the library includes a protocol converter interface card (NSR in MSL libraries).
Another possibility is the connection of the SCSI library to a SCSI/Fibre bridge device. In this
case the library is SCSI, but is connected to the host through the bridge device.
In the case of the SCSI connected device, the controller is simply the system path from the
system bus to the SCSI device, and it will be displayed as a class of device called ext_bus by
the ioscan command.
HW Path: 0/4/0/0.9.23.198.0.3.1
In the case of Fibre Channel, the actual hardware path for the controller can represent
several different parameters including the back-plane slot of the interface card and all of the
Fibre Channel values (varies by switch type). The ioscan command, however, still displays
the ext_bus as the class of the device controller. It does not matter from a local perspective,
if the device is on a public or private loop, connected via switch or switched fabric, ioscan
will still show the hardware path to the library controller and SCSI devices as if they were
“local” devices.
The key to the configuration of the library controller device is the location and identification
of the ext_bus interface that connects to the library device. Once this is known we can easily
locate the SCSI target and LU of the controller that needs to be configured.
The three components will be used to construct the minor number (once converted to hex)
used to create the device file for the library controller.
Sample MSL 5030 connected to a B-series switch and HP 9000 server running HP-UX 11i:
Class I H/W Path Driver S/W State H/W Type Description
=============================================================================
ext_bus 7 0/2/0/0.1.3.255.0 fcpdev CLAIMED INTERFACE FCP Device Interface
target 15 0/2/0/0.1.3.255.0.0 tgt CLAIMED DEVICE
autoch 0 0/2/0/0.1.3.255.0.0.0 schgr CLAIMED DEVICE HP MSL5000 Series
/dev/rac/c7t0d0
tape 1 0/2/0/0.1.3.255.0.0.1 stape CLAIMED DEVICE HP Ultrium 1-SCSI
/dev/rmt/1m /dev/rmt/c7t0d1BEST
/dev/rmt/1mb /dev/rmt/c7t0d1BESTb
/dev/rmt/1mn /dev/rmt/c7t0d1BESTn
/dev/rmt/1mnb /dev/rmt/c7t0d1BESTnb
tape 2 0/2/0/0.1.3.255.0.0.2 stape CLAIMED DEVICE HP Ultrium 1-SCSI
/dev/rmt/2m /dev/rmt/c7t0d2BEST
/dev/rmt/2mb /dev/rmt/c7t0d2BESTb
/dev/rmt/2mn /dev/rmt/c7t0d2BESTn
/dev/rmt/2mnb /dev/rmt/c7t0d2BESTnb
ctl 5 0/2/0/0.1.3.255.0.0.3 sctl CLAIMED DEVICE COMPAQ SWMODULAR ROUTER
/dev/rscsi/c7t0d3
Notice: The device instance for the ext_bus class device is 7 (I column), the target device is 0
(last digit of the HW path for the target class device) and the Lun number is 0 (last digit of the
HW path for the autoch class device) producing a device file: /dev/rac/c7t0d0 for the
autochanger device (library controller/robotic). The ext_bus path includes the hardware path
of the HBA as well as the Domain.Area.Port (1.3.255) of the FC device.
The example above shows the LUN mode for the Network Storage Router, where Lun 0 is the
tape contoller, Lun 1 is the first drive and Lun 2 is the second drive. This Lun addressing
mode is used when the NSR is enabled for Active Fabric mode to support the use of the X-
copy serverless backup mode.
In order to create a device file for use in controlling the tape library, you will need to gather
all three components. These may be collected by using the ioscan and lsdev commands.
Most HP-UX systems prior to HP-UX 11.11 (11i) do not have the schgr driver configured by
default; it is available and supported but not as widely used as the sctl driver.
The HP-UX command kmsystem may be used to check if the drivers are available on the
system. This command will also display the configured state for the driver. If the drivers are
configured into the system, the lsdev command will display them along with their character
major number. The typical output for lsdev shows the major number 203 for the sctl driver,
231 for schgr.
Converting a hardware path to a (hex) minor number (output from ioscan –f):
Fibre Channel/SCSI (private loop or point to point, domain id=8)
ext_bus 13 1/2/0/0.8.0.255.7 fcpdev CLAIMED INTERFACE
FCP Device Interface
target 17 1/2/0/0.8.0.255.7.11 tgt CLAIMED DEVICE
autoch 1 1/2/0/0.8.0.255.7.11.0 schgr CLAIMED DEVICE
HP C7200-8000
SCSI
ext_bus 2 56/52 scsi1 CLAIMED INTERFACE HP 28655A
- SE SCSI ID=7
target 4 56/52.4 target CLAIMED DEVICE
spt 0 56/52.4.0 spt CLAIMED DEVICE HP
C5177-7000
Example creating a device file using mknod with output from ioscan –fn and the sctl driver:
Class I H/W Path Driver S/W State H/W Type Description
=======================================================================
ext_bus 1 8/4 c720 CLAIMED INTERFACE GSC add-on Fast/Wide
SCSI Interface
target 3 8/4.0 tgt CLAIMED DEVICE
unknown -1 8/4.0.0 UNCLAIMED UNKNOWN HP C7200-8000
target 4 8/4.1 tgt CLAIMED DEVICE
tape 1 8/4.1.0 stape CLAIMED DEVICE QUANTUM DLT8000
/dev/rmt/1m /dev/rmt/c1t1d0BEST
/dev/rmt/1mb /dev/rmt/c1t1d0BESTb
/dev/rmt/1mn /dev/rmt/c1t1d0BESTn
/dev/rmt/1mnb /dev/rmt/c1t1d0BESTnb
The Data Protector Media Agent provides two utilities that may be used for device testing
prior to configuration as a Logical Device. The utilities are located in the /opt/omni/lbin
directory as devbra and uma. Once the devices are known, they may be tested with the Data
Protector Utility Media Agent (UMA) to determine operational status.
2. locate the device path for the tape library robotics (media changer)
3. invoke the utility media agent to interact with the tape library (load/unload tapes,
status inquiry, etc.)
/opt/omni/lbin/uma –ioctl <dev> -barcode
h. move <from> <to> move media from element to element (return to original
position when testing is completed; use integer value or element name)
5. HP-UX additionally offers the “mc” utility which operates similarly to the Data
Protector “uma”.
Example:
/usr/sbin/mc –p /dev/rac/c0t7d0 –q (SCSI inquiry)
(SCSI report element status, Drive, Import/Export, Media transport, Storage slots)
– Library Exercise
– Firmware Management
Student Notes
In addition to the utilities provided by Data Protector, HP offers the Library and Tape Tools
for tape library and drive management. Library and Tape Tools is available for no charge
from the HP web site shown above. As of the printing of this manual, L&TT version 3.3 is
available.
Software Features
L&TT offers the following features:
• Installation Check — L&TT guides you through a basic installation check of your
product. The software assists the user in choosing an appropriate HBA and SCSI ID(s),
ensuring that the device is detected by the system, and verifying key device functionality.
The installation check feature is essentially HTML documentation that helps with the
most common generic installation issues while also describing how to use L&TT to verify
the device installation.
• Device Identification — L&TT clearly identifies the storage products connected to the
system, along with key information on product configuration and status.
Student Notes
Shown above is the Windows GUI, illustrating the devices connected to the host. From the
Test area of the GUI, the device to host access may be tested. The host to tape buffer is used
to verify physical connectivity and device availability. Note; no data is written to the tape.
Note: the L&TT indicates whether the OBDR capable firmware is loaded onto a device;
shown above the DDS drive does not currently have the OBDR firmware loaded. This would
be necessary to support Data Protector Disaster Recovery (OBDR) functionality.
By Product—This option shows a list of all the products connected to the system.
The list is grouped into the following four categories:
o Libraries and autoloaders
o Drives
o Enclosures and processors
o Other devices
The three number fields listed after the device represent the device address. Each
field in the address is separated by a period: the first field represents the HBA
channel, the second field represents the SCSI ID, and the third field represents the
LUN.
3. Device Information screen—All the main functions of the program are displayed in
this window. The content of this window depends on the device and tool function
selected.
Student Notes
Selecting Device Analysis in the Test Group analyzes data in the internal logs on the device
and finds problems if they exist. Advice is given on how to solve the problems.
When test is
in progress,
time
remaining will
display here.
Student Notes
Selecting Library Exercise in the Test Group will perform robotic exercise as well as drive
load and unload exercises. This requires a “scratch” medium to load and unload. Library and
Tape Tools will prompt for a tape to be loaded into the mail-slot for this exercise.
Consult with your instructor about availability of the MSL simulator (GUI touch-screen)
and/or remote access to a Network Storage Router and Remote Management Interface for the
MSL library. Due to the nature of this course, students are not expected to configure the tape
library nor the NSR; access may be provided to allow for demonstration purposes. Later in
this course the MSL libraries will be configured as Logical Devices for use with Data
Protector.
Additionally, the MSL Web Based Training may be available on the classroom training
systems or may be accessed via the HP IT Resource Center (http://itrc.hp.com).
Media Management
Library Features
• Logical organization of media
• Online catalog
• Location tracking
Protection Features
• Media labeling
• Media duplication
• Media condition monitoring
Student Notes
Today, modern backup utilities must offer more than just a mechanism for backing up
computers data. As the term storage management implies, the management of the data
once it has been backed up is just as important as the act of backing up the data. Data
Protector has powerful features to organize and protect your backups.
Library Features
Logical Organization of Media
Data Protector organizes your media into Media Pools; a Media Pool is simply a logical
collection in which media that belong together are kept.
Online Catalog
Data Protector maintains a record of all the data that been backed up and what media was
used to perform the backup. When it is necessary to restore data, the on-line catalog can be
browsed to locate the file to be restored and to find the candidate backups that could be
used. This catalog is part of the Data Protector Internal Database, more details are in the
Database module.
Location Tracking
Once a backup has been performed, the media usually is moved physically from one location
to another, for example to offsite storage or a fire-safe.
Data Protector can keep track of the physical location of the media by use of predefined
Vaulting Locations.
In addition to tracking external changes to the media locations, Data Protector stores the
current physical location of media. When a tape is inserted into a Logical Device, and then
accessed by Data Protector, the device repository is stored in the database. This media
tracking provides for quick access to known tapes. This device repository feature is available
for tape library as well as standalone devices.
Protection Features
Media Labeling
When backups or restores are performed, we need to be able to verify that the correct
medium has been selected for the desired action. Without this capability, it would be possible
to restore from the wrong media or erase a media that we want to keep.
Data Protector media contains header information that enables the use of the media to be
tracked and controlled.
Media Duplication
For extra security, it may be necessary to have multiple copies of a particular backup. For
example, if the data were being changed in some way, or removed after the backup has taken
place, the only place that the original data would reside is on the backup media.
In this situation, it is desirable to have multiple copies of the backup available in case there is
a fault with the original copy or it is somehow lost.
The backup could be performed twice. One disadvantage of this is that the data would be
unavailable to the users for twice the length of time required for a single backup.
Additionally, there is a possibility that some data may change from one backup to another,
thereby not creating exact copies of the original backup.
Data Protector provides a mechanism for copying media. This has the advantage of being
able to be performed while the data is back online following the original backup. Copied
media is also tracked in the Media Management Database.
Data Protector (5.1) provides automated media operations. This has the ability to either
schedule automated media copy or execute automatic media copy after a backup job is
completed. This method combines the flexibility of the manual copy plus the automation
associated with lights out operations.
Purpose
Purpose
•• Logical
Logical grouping
grouping of
of media
media
Database
Backups
•• Media
Media usage
usage policies
policies
•• Device
Deviceproperty
property(default)
(default) Full Daily
Incremental
Backups
•• Assigned
Assigned to
todevice
device by
bybackup
backup Backups
DB Month End
Archive Backups
Logs Backups
Student Notes
Logical Organization
Data Protector organizes media into Media Pools. A Media Pool is simply a logical collection
in which media that belong together are organized into a single structure within the internal
media management database.
The only restriction is that all the media in the same media pool must be of the same physical
type, for example, DDS or DLT.
Media Pools should be used to organize media in a logical fashion; for example, they should
contain only media that is related in some way and has the same usage policy.
As you can see from these examples, each backup, while similar, is slightly different, either in
the number of tapes required to complete the backup, or the cycle in which the media will be
used.
It is also possible to have all these backups share the same single media pool. This approach
has certain disadvantages. The quantity of media in the pool may be too large, and managing
the pool may be difficult. It would be difficult to verify that you have sufficient media in the
pool to complete all the backups that utilize it (as each backup has a different requirement).
TIP A simplified way to think about media pools: View them as a destination for
your backup, while you look at the devices as a transfer mechanism between
the data and the media pools.
Grouping media used for a similar kind of backup into a Media Pool allows you to apply
common media handling policies on a group level. In this case, you will not have to bother
with each medium individually. All media in a pool are tracked as one set and have the same
media allocation and usage policies.
Using the
command line
Student Notes
The Data Protector administrator can create a media pool via the GUI or by using the
omnimm command. Data Protector provides a set of default media pools, one for each media
type.
Configuration via the GUI is the easier method of creating a new pool, however, the
command line offers several possibilities for automation.
TIP For more information on these commands, refer to the online man pages.
Example: Create a new media pool using the omnimm command and verify
its attributes:
Pool Description :
Media type : DDS
Policy : App+Loose
Blocks used[MB] : 0
Blocks total [MB] : 0
Altogether media : 0
Poor media : 0
Fair media : 0
Medium age limit : 36 months
Maximum overwrites : 100
Magazine support : No
Free pool support : Uses free pool + Move free media to free pool
Automatic
32-character name allocation
64-character
description automatic
de-allocation
Age and
overwrites
Student Notes
The Media Pool properties may be set when a media pool is created or modified at a later
time. The name for the pool may contain up to 32 characters, spaces are allowed but not
suggested (complicates scripting, etc).
The description field (64 characters maximum) is optional, and may be used to convey a
purpose or usage characteristics for the pool.
The media type is selected when the pool is created and is not modifiable. To change the
media type of a pool you must first delete the pool and then re-create it.
The Allocation Policy as well as the Usage Policies may be altered for new or existing pools.
The life expectancy and number of overwrites should be set according the media
manufacturer's recommendations. Data Protector simply provides a default value based upon
the media type.
Properties
Properties Loose
Loose
•• Media
Media allocation
allocationpolicy
policy or
or
Strict
Strict
•• Media usage policy
Media usage policy
•• Media
Mediacondition
conditionfactors
factors
•• Free
Free Pool
Pooluse
use Appendable
Appendable
•• Magazine
Magazine support
support
or
or
Non-appendable
Non-appendable
Age
Age
and
and
Overwrites
Overwrites
Student Notes
In addition to being a logical container for your media, a Media Pool is configured so that the
media within the pool exhibit particular characteristics. These characteristics depend on the
properties and policies that you have set for the Media Pool.
Properties
TIP While it is acceptable to use spaces in the name, the recommendation is not to
do so. Use the underscore instead. While using the command line for Data
Protector, you will need to use double quotes around any names that contain
spaces. Notice that the default pools use spaces.
The description field can be used to give a more detailed explanation of the usage of the pool.
Limit the description to 64 characters.
Media Type
This defines what type of media the pool contains. Remember that a pool can contain only
one type of media. The currently supported media types are:
• DDS
• DLT
• LTO-Ultrium
• SuperDLT
• DTF
• ExaByte
• AIT
• QIC
• T3480/T4890/T9490
• T9840
• T3590
• T9940
• SD-3
• Tape
• Optical
• File
Appendable
This enables Data Protector to append multiple backups to the same piece of media. This can
be very useful when backing up small amounts of data throughout the day, for example
Database Redo Logs.
When using this policy, Data Protector will always request a media that has the most data on
it but is not full (See Media Allocation Policy).
When a backup is performed, it is directed to a specific media pool, via the Logical Device
definition. Data Protector will choose the particular media to be used from the pool, based on
certain factors. If the media pool allocation policy is appendable, the media that is the most
full, but still has spare capacity is used. Ideally, Data Protector wants to fill up existing media
before going on to use empty media. This policy will save generally be less expensive in
terms of media cost, but will not allow for easy tape rotations. Data Protector will continue to
request the medium until it is filled.
Non-Appendable
This specifies that Data Protector will write to a media from the beginning. Data Protector
will request a media that has been used the least amount of times (See Media Allocation
Policy).
If the media pool is non-appendable, Data Protector chooses a tape that has been used the
least number of times. In this way, Data Protector ensures even wear across all media, rather
than the same tape being used each time. This may make media more reliable due to less
wear as a result of fewer loadings.
The choice of which tape to use is based on the allocation number associated with the
medium. The allocation number is viewable from the Media Management GUI; select the
Device & Media context, right click on a media pool, select properties, then the allocation
tab.
CAUTION Media that are marked as Poor should not be reinitialized and registered as a
new medium unless the poor condition was as a result of a tape drive failure,
and the tape is new.
NOTE Media that are marked as Poor may be the result of a failure to write due to a
hardware (drive) failure. In this case the tape quality may be verified by
scanning and/or verifying the tape. (omnimver)
Maximum # of Overwrites
In addition to the number of months that a media is to be considered valid, the number of
overwrites can also be configured. Again, when this threshold is reached the media is marked
Poor. Tapes reaching 80% of this threshold are marked as Fair.
NOTE Both the age and overwrite thresholds may be altered via the MMFairLimit
parameter in the <OMNICONFIG>/options/global file. Eighty percent is
the default for the use of fair quality marking.
Loose or Strict
The loose policy defines that even while Data Protector will request a particular medium, it
still accepts an alternative that is available for use.
The strict policy determines that the medium Data Protector requests must be used.
Allocation order is “strictly enforced.”
The most commonly used setting is loose because it is more forgiving. (Loose is also
required when you want the ability to use a new, unformatted medium).
NOTE For more details on the allocation policies, see the next slide.
The setting of this flag tells Data Protector to initialize and use blank media that may be
loaded in preference to media that is already initialized. If the device being used is a library, it
must be scanned (barcode) prior to using this feature, or Data Protector will not know where
the uninitialized media are located.
Magazine Support
Certain SCSI II Library devices, such as small auto-changers are equipped to manage media
loaded in magazines. Individual media is never removed from a magazine; rather, the whole
magazine is replaced. In addition, the order of the tapes in the magazines should not be
changed. In effect, Data Protector treats the magazine as one large piece of media.
Magazine Pools are very useful when backups consistently require multiple tapes. The
handling of these tapes as one unit eases the process of media loading, unloading and
storage, but may be more expensive if the magazine is not filled before getting removed from
the device.
Data Protector can use media within small auto-changers with a standard media pool, or the
media pool can be configured specifically for magazine support.
When magazine support is enabled for a media pool, the media pool view can be changed so
that magazine’s are shown rather than individual media. When using this view, the following
commands operate on the entire magazine:
• format magazine
• modify
• verify
• move
• recycle
• ungroup media
• import magazine
• export
When a magazine is formatted, Data Protector assigns a name to the whole group of media
and gives every single media in the magazine the same name, suffixed with a sequential
number.
Whenever Data Protector scans a magazine, it reads the label of the first tape to identify the
magazine. It assumes the magazine is completely loaded and does not scan the remaining
slots. Therefore, you should never change the order of the tapes in the magazine or remove
individual media.
Loose Strict
•• Even
Evenmedia
mediausage
usage •• Even
Evenmedia
mediausage
usage
•• Auto-initialization
Auto-initialization(optional)
(optional) •• Manual
Manualinitialization
initialization
•• Allocation
Allocation order notenforced
order not enforced •• Allocation
Allocationorder
orderenforced
enforced
•• Works
Works best with smalldevices
best with small devices •• Works
Works best with librarydevices
best with library devices
–– standalone
standalonedrives
drives •• More mount requests
More mount requests with with
–– small autochangers
small autochangers small
smalldevices
devices
•• Fewer
Fewermount
mountrequests
requests •• Override
Overrideduring
duringbackup
backup
•• Auto-allocation/de-allocation
Auto-allocation/de-allocation •• Auto-allocation/de-allocation
Auto-allocation/de-allocation
Student Notes
One of the most important decisions for creating Media Pools is to choose whether you
would like to use loose or strict media allocation. The following will summarize each policy.
Loose Allocation
The loose allocation policy usually works best with standalone devices and small
autochangers. With these smaller devices, it is common to want the ability to use any
unprotected tape to perform a backup (provided that the tape belongs to the pool assigned to
the needed device, or is new). Data Protector will try to use media that exists according to
the allocation order, but this will not be enforced. The operator will be presented with mount
requests for any tape "unprotected or new" if Data Protector finds an invalid tape in the drive.
Blank Media
Data Protector may “format on the fly”, or auto-initialize (format) when it finds blank
medium in either a standalone or a library device. Data Protector may auto-format the
medium when the backup starts. It will give the media a default name and put it in the media
pool associated with the Logical Device used for the backup. The media policy must be set to
loose to take advantage of this feature. Additionally, a global option, InitOnLoosePolicy must
have the appropriate value to allow for automatic initialization. See the note below.
Note: The auto-initialize feature may be controlled by modifying the global option
InitOnLoosePolicy. The default value of this parameter varies with the
product version. The default value at version 5.1 is 0, which is disabled. To
enable, set the parameter to 1.
With automatic formatting, you will not have initial control over the labels (called description
in the GUI) for the media prior to the backup. After the backup completes, however, you will
be able to modify the media label (description) in addition to producing reports on the media
used for a particular backup session.
Strict Allocation
The strict allocation policy usually works best with library devices. With these larger devices,
it is desirable to have an even usage of media within the library. In the case of the strict
policy, the even usage would be "strictly enforced" by Data Protector. In order for Data
Protector to assign an allocation number to each tape, they must all be manually formatted
prior to the start of a backup session. The order of the initialization (formatting) will
determine the initial allocation sequence.
If a mount request is given to the operator for a tape from a strict allocation media pool, Data
Protector will request a specific medium. At that time only the requested medium will be
acceptable to complete the backup.
The use of strict allocation for standalone devices will require proactive media management,
to be sure that you always have the correct tape in the drive prior to backups starting. The
combination of strict and standalone devices (or small autoloaders) is not usually the best
combination for a lights out operation.
To verify the media allocation order, open the GUI and select a particular media pool in the
“Device and Media” context. In the results area you will see “Order” as one of the column
headings, this is the current Allocation Order for the tape. The tape order has the potential to
change with each backup as a result of tape usage.
The sequence of media allocation is in the order of the following Data Protector media sets:
• Pre-allocated Media
• Appendable Media
• Uninitialized Media
• Free Media
• Overflow Media
Each of these media sets has its own definition and rules about the sequence of media. These
are explained below:
Preallocated Media
Media named in the datalist device options pre-allocation list. Pre-allocated media in 'Poor'
condition will not be used. The pool policy can be Strict or Loose. This media set is not
sorted.
Order of use:
as specified in the datalist, provided that this won't break any other rules such as those
relating to protection and appendable media.
Appendable Media
Media in 'Good' condition, with some currently protected data objects, but the media is not
full. The pool must be 'appendable'. This media set is sorted according to the time of the last
write. The most recently written medium is listed first.
Order of use:
when one or more media have protected objects, the most recently written media is reused
first.
Uninitialized Media
Uninitialized Media is media without a recognizable header; Data Protector assumes that it
can be auto-initialized as required, during backups. The pool policy must be Loose to allow
auto-initialization and the global file needs InitOnLoosePolicy=1. This media set is only
available in exchanger devices. This media set is sorted with 'Blank' media ahead of media
with an 'Unknown' header.
Order of use:
a. 'Blank' media is used first.
b. 'Unknown' media is only used when there is no 'Blank' media.
Free Media
Free Media is that which is in 'Good' condition with no currently protected objects contained
on it. This media set is sorted according to the time of the last write. The least recently
written medium is listed first.
Order of use:
• least recently medium is used first.
Overflow Media
Overflow Media is that which is in 'Fair' condition with no currently protected objects. This
media will only be used if no 'Good' condition media are available. This media set is sorted
according to the total number of overwrites. The medium with the least number of overwrites
is listed first.
Order of use:
• least recently medium is used first.
Unclassified Media
Media in the following categories are not classified into any of the sets by Data Protector. As
a result, they are not allocated for use by Data Protector.
• Media in 'Fair' condition which is protected.
• Media in 'Poor' condition.
• Media which is recognized by Data Protector as having a header for another backup
utility such as 'tar' or 'fbackup'.
Strict Policy
The Strict allocation policy is not directly related to the use of a preallocation list. A
preallocation list can be used by both the Loose or Strict policy. The order of media use is
generally the same for Loose and Strict policy. The difference is in Data Protector's response
when the medium in the device is not the one dictated by the allocation rules. Strict policy
means that Data Protector will not use any other medium than the one its own rules indicate
should be used. If the policy is Loose, any unprotected medium can be used, if found in the
device.
The order that media are selected for use depends in part on whether or not the data on the
media is protected. In general, if the data is protected, a medium will be used for appending
more data. If the data is not protected, it will not be used for appending, and will not be
overwritten until no other medium is available.
Example 1:
A pool is configured as 'Appendable' and 'Loose'. Five newly initialized media are loaded into
an autoloader.
If several backups are made, with protection on the data, they will all be appended to the first
media.
Then several more backups are made, now with NO protection on the data. The unprotected
backups will also be appended to the first medium and they will not overwrite each other.
Example 2:
Again, five newly initialized media are loaded into an autoloader. If a backup is made with no
protection, it will be written to the first medium. A second unprotected backup will be
written to the second medium. Each subsequent unprotected backup will be written to a new
medium, until all five media are used. The sixth unprotected backup will overwrite the first
medium, which is now the least recently used medium.
Because the data on each medium is unprotected, the media are not considered appendable,
even if the pool configuration is 'Appendable.'
Example 3:
Again, five newly initialized media are loaded into an autoloader. A backup is made with no
protection, and will be written to the first medium. The second backup in this sequence is
protected and it is written to the second medium. The subsequent backups are all
unprotected and they will be appended to the second medium. The presence of a protected
object on the second medium makes it 'appendable.'
While any backup object on a medium is still protected, the whole medium is prevented from
being reused. Sometimes a medium that is expected to have no active protection is rejected
for use by OBII.
It may be that one object on the medium is actually still protected. A typical scenario is that
an ad hoc backup was added to a medium outside the normal schedules. As the default
protection is 'Permanent' this can be what is preventing the medium being used. Check the
medium in question with this command:
The expression 'first session medium' is found in the Data Protector documentation. It refers
to the medium that is used as the first in a sequence of media, if the backup session requires
more than one medium. If the pool policy is appendable, only the medium that is used as first
in the session can be appended.
Subsequent media for that session must begin at the start of a medium so they need to be
empty or unprotected.
If the first session medium becomes full, and there are no empty or unprotected media
available, a mount prompt will occur. This will be true even if some space remains on the
other media in an exchanger. Different parts of backup cannot be appended to the ends of
several different media.
The cleaning status is NOT checked while a medium is in the drive and being written to. The
status of the cleaning request is checked only when a medium is being loaded or changed. So,
if the drive sets the cleaning bit, part way through a backup session, the medium is not
immediately unloaded. It will be used until it is full, assuming that there is enough data to fill
it, in the current backup session.
When medium #1 is unloaded, the cleaning request will be checked; the cleaning tape will be
loaded and used. After that, medium #2 will be loaded to continue the backup. It is possible
that medium #1 is in bad shape and could be the cause of the cleaning request. However,
Data Protector assumes the medium is still 'Good' and does not mark medium #1 as 'Poor'.
The medium will be used again, when it is the 'least recently used'.
If medium #1 is really faulty, this will show up next time it is used, when the write operations
fail. Then it will be marked 'Poor' and not used again.
DDS cleaning tapes are a fixed length, with just enough for '25 times 30 seconds' of cleaning.
When 30 seconds of cleaning has completed successfully, the clean bit on the drive is reset.
The cleaning tape moves forward, over one section of tape, for each 30 seconds of cleaning
and it NEVER rewinds! It just moves along until it gets to the end and then it stops.
Once it is at the end, it is an 'expired' cleaning tape. If you load an 'expired' cleaning tape, the
tape is active for less than 20 seconds, it does not actually do any cleaning and the cleaning
bit is NOT reset. Data Protector can initiate a cleaning operation when the drive sets the
cleaning bit. After the cleaning is done, Data Protector will recheck the cleaning bit. If it is
still set, Data Protector will terminate the session and will report that the cleaning tape was
requested twice for the device.
Initialized Size
The size that a tape media is initialized to will not ultimately affect the amount of data that
can be written to it. Data Protector writes to the tape until the device reports early end of
tape (EOT) warning.
If a tape is formatted to a smaller size than the physical size of the tape, Data Protector will
write to the end of tape, and then update the Media Management (MM) database. The
recorded tape size will be reset to the value of the physical tape size.
The same thing applies to tapes that are initialized to a very large size. Once the tape has
been filled with data, the size will be reset in the MM database.
Statistical Information
The correct settings for the 'Full' flag and for 'Data Protection' are the only tape details
necessary for Data Protector operation.
Data Protector does not care about the order of slots inside a random access tape library. It
assumes that the slots it controls are assigned exclusively to Data Protector and performs
media allocation based on the media allocation rules.
The rules do not relate to the order of media in the slots, so Data Protector does not
necessarily start with the lowest slot number and progress towards higher slot numbers. The
media allocation rules are the same for small exchangers and large tape libraries.
DDS Pool-1
Protected
Expired
Free-DDS Pool
Protected Expired
De-allocate Expired
Protected
De-allocate
Protected
Expired
Student Notes
Data Protector supports the use of a Free Pool of unprotected media. These “free” tapes may
be newly formatted or have expired backups on them. See the next page for more details.
Each media type supported by Data Protector may have an associated Free Pool; so one for
DDS, DLT, LTO, etc.
To implement the Free Pool for an individual media type, create a new media pool or modify
the properties of an existing pool and select the “Use free pool” feature on the Allocation
properties tab of the GUI. Data Protector will automatically create an additional pool called
“Free <Media_type>”, such as “Free DDS.” There will be only one free pool for each media
type and it may be shared with all of the other pools of the same media type.
off de-allocation (by de-selecting the “Move free media to free pool) you may still move
media manually to the free pool as long as the media is not protected.
De-allocation Times
The de-allocation process occurs periodically during the day. The frequency of the de-
allocation is controlled by the “FreePoolDeallocFreq” parameter in the global file. The
default frequency is once per day at 00:00 (midnight). The parameter “FreePoolDeallocFreq”
is set to one by default, but may be set as high as 96 to produce a 15-minute de-allocation
frequency. The first de-allocation occurs at 00:00, and then the day is divided according to the
frequency that you specify. As an example, a frequency of 3 causes de-allocation at 00:00,
08:00 and 16:00.
You have the option of forcing a manual de-allocation at any time by using the command:
omnidbutil –free_pool_update
NOTE The omnidbutil command is available only on the Cell Manager as it is not
part of the command line part of the Cell Console. On the Unix Cell Manager
the command is in the OMNIHOME/sbin directory; on the Windows Cell
Manager the command is in the OMNIHOME/bin directory.
The media pool allocation and usage policies will be established by properties of the regular
media pools, as free pools do not have such policies available. Tapes that exist in the free
pools are not used for backup until allocated and moved to a regular media pool.
Media Life
Media Pool
properties establish
expected media life
Related commands:
omnimm -list_media ... Medium
omnimm -media_info ... properties track
age and usage
omnimm -catalog ...
Student Notes
Once the media is formatted, it should not be formatted again. Data Protector keeps usage
and quality information regarding the tape in its database. Formatting a medium more than
once resets the quality information stored within the media management database.
If the session information is not required (and it is still protected), use the recycle feature.
NOTE Media can only be exported if the protection of the sessions has expired, or a
recycle has been performed to remove the protection.
Media Operations
Medium
Operations
Pool
Operations
Student Notes
Data Protector provides the following Media Management operations for pools:
Format Initialize a medium. Prepare it for Data Protector use by writing a header
to the tape, and register it in the media management database
Import Read the header and detail catalog information from a tape. The tape may
be from a different cell or may have been exported from the current cell.
Delete Removes an empty media pool. Delete media in pool first. This is useful for
removing the Default pools that are not needed.
Select Media Search a Media Pool for specific media.
Useful when a pool contains a large number of media.
Media Operations
Data Protector provides the following Media Management operations for media within pools:
Export Delete an unprotected medium from the media management database. The
contents of the tape are unaffected.
Change Alter the vaulting location string associated with a tape. The tape does not need to
Location be in a device for this operation.
Recycle Remove all of the protection from the data that is backed up on the selected tape.
The tape does not need to in a device for this operation.
Move Change the pool that a particular tape(s) is assigned to. The tape does not need to
be in a device for this operation.
Copy Replicate a tape. Two devices of the same type and a blank tape are required. This
uses the omnimcopy functionality for duplicating a single tape.
Verify Read the tape header and verify that it is written in Data Protector format. The
data may also be verified if the tape contains crc blocks.
Import Recover the detail catalogs from a tape that is still in the database but has had its
Catalog detail catalog expire. The detail catalog is automatically purged from the database
when it expires.
Formatting Media
Format each
tape only
one time
omniminit [-options]
Student Notes
Before any media can be used with Data Protector, it must undergo an initialization process
called Formatting. The Data Protector GUI now presents the format.. option, where
initialize was previously used.
The Data Protector media management system requires a unique medium ID for each tape. A
unique ID is generated when the media is initialized. This ID is written to the media header
and to the Media Management Database (MMDB). Data Protector uses this header to identify
one media from another. Each time a medium is accessed, the header information is read to
ensure that the correct medium is being used. It is also possible to manually read media
header information by using the Scan operation of a Logical Device or with the “omnimlist –
device <logical device> -header” command.
The media format process is performed within the media pool where the formatted medium
is to be added.
Media can be formatted from within the GUI in the Devices and Media context, or from the
command line using the omniminit command.
Specify allows the user to input a specific capacity in megabytes that the medium is
expected to hold. The capacity is used only for statistical purposes and does not set a
hard limit on the amount of data that any media can hold.
Each time Data Protector writes to a newly formatted media, the media capacity figure is
updated. It reflects the largest amount of data that has ever been written to the tape.
NOTE When using media type File, the specified size will limit the size of the file
medium; the Data Protector default is 100 MB. This may be altered by
modifying the global option: FileMediumCapacity.
• Force
During initialization, Data Protector checks the media to see if it already contains data
that is in a recognizable format. If the format is recognized, then, by default, Data
Protector does not initialize the media. The reason is that this media may contain
valuable data.
If the format is recognized but initialization is still required, the force option must be
specified.
Media Duplication
omnimcopy –options
omniamo -options
Student Notes
For extra security, it may be necessary to have two copies of a particular backup. For
example, if the data were being changed in some way or removed after the backup has taken
place, the only place that the original data resides is on the backup media. If an individual
medium is lost or damaged the ability to recover data will be lost. In addition, it may be
desirable or required to retain media copies both on-site as well as off-site.
The manual, single copy operation can be initiated through the GUI or the command line
interface with the omnimcopy command. The source and destination devices are logical
devices. The logical devices may be located anywhere in the Data Protector cell, but must be
of the same media class. During the copy, the target media is initialized before all data from
the source media is copied.
After the copy, both media are tracked in the media management database so that the
original media and its copies can be easily identified.
If a mount request is issued during a restore session, all tapes that contain the requested
data are listed; this includes both originals and copies.
If the original media is overwritten or is exported from the MMDB, the first copy becomes an
original.
NOTE: When the copy process is completed, both the original and the copy are
marked as non-appendable. The copy may also be permanently protected. A
copy of a copy is not permitted.
Tape Variation
There are slight variations in the overall capacity of individual tapes. This can pose a
significant challenge when attempting to make an exact copy from a tape that is slightly
larger than the destination tape. Planning for this eventual issue must be done before media
is initialized.
There is a local parameter that may be specified per device called OB2BLKPADDING. This
parameter is placed in the omnirc file on each system with connected devices and indicates
the number of blocks to add after the tape header. This additional padding should allow tapes
of the same type to be duplicated, even if they vary slightly in capacity. See the
<OMNIHOME>omnirc.tmpl located on the cell manager for more information.
AMO types:
• Post-backup
– Enables automatic media copy
at the end of a backup session
– Copy all media used in that
particular session
• Scheduled-backup
– Pre-determined start time
– Automatic copy of media used
in time window
Student Notes
Automated Media Operations (AMO) is a new feature in Data Protector 5.1 that facilitates
automated copying of media containing backups.
• post-backup: enables automatic media copy at the end of a backup session, which
can copy media used in that particular session
• operation type: Media Copy is the only possible value, because currently automated
Media Copy is the only member of Automated Media Operations suite.
• number of copies: N = 1 to 10, where default is 1; each medium will be copied N times
• source drives
• destination drives
• eject = none/copy/original/both
• new location for target media for vaulting purposes
• target media protection: default = same as original, permanent; none, days & weeks
All media management operations, in the realm of AMO classify themselves as sessions, like a
backup or restore session, as opposed to being mere utility tools such as omnimcopy. These
sessions will be tracked in the database and therefore can be monitored and reviewed later.
The source medium defines the destination pool of the target medium. This effectively means
that the copied media will belong to the same pool as the original media.
Each source medium is mapped to a pair of devices from among the devices that were
specified in the AMO specification. Once this device pair is established, a copy session will
copy the data from the source to the target medium.
The AMO functionality provides for its own load balancing. It optimizes the usage of the
available devices by utilizing as many devices as possible and even selecting local devices, if
they are available.
Device locking takes place at the outset of an AMO session. Since the devices that are not
available at the beginning cannot be utilized for the session, device locking after the
beginning of the session is not possible. An available ‘pair of devices’ of a certain device type
is a minimum prerequisite for successful completion of any automated copy session.
The data protection for the copy defaults to the original’s protection. However, you can alter
the protection period either during the creation or modification of the AMC specification.
Automated Media Copy does not handle mount or cleanme requests. Incase a mount request
pops up, the media pair aborts, while the session continues to reach its logical end.
Student Notes
Post-backup AMO is automatically initiated after a backup session finishes, and results in the
copying of the media used in that particular session. Session records for AMO post-backup
sessions are stored in the IDB and are also able to be monitored.
The BSM reports the end of a backup session, and thereby supplies information about
session, sessionID, datalist, success etc to omnitrig utility. After a backup finishes, omnitrig
matches the backup (backup specification name) to a corresponding post-backup AMO
configuration file. On a successful match, AMO initiates the copy process and copies all
media used in that particular backup session. There may be only one AMO configuration per
backup specification.
2. In the Scoping pane, right-click Automated Operations and click Add Post-Backup
Media Operation to start the configuration wizard.
3. In the Backup Specification drop-down list, select the desired backup specification.
5. Specify:
Media Properties
Tracking of the duplicated media is done within the MMDB. While displaying the properties
of the copied media, there appears a button in the upper right corner in the GUI called
“Original..”. Select “Original” and a pop-up window displays the following information about
the source medium:
Additionally, when displaying the properties of the original medium, a new tab labeled
“Copies” is available to show a list of all valid copies of the medium as well as a summary of
their properties.
• poor
Select to • fair
change • good
month • any condition
Select to
change
year
Choose
timeframe
Specify
options
Student Notes
The scheduled Automated Media Operation (Media Copy) is the process of duplicating media
used in one or more backup sessions at a scheduled time. Scheduled Media Copy seeks
backup sessions that started and have completed, within a user-defined timeframe. Once the
sessions are known, AMO copies all of the media that belong to the backup sessions using a
single AMO session.
The media will be copied simultaneously, if enough devices are available. Otherwise, they
will be copied sequentially. Load balancing in AMO strives to simultaneously use the
maximum number of media during the copy process.
The omnitrig module polls every fifteen minutes to see if there are any scheduled tasks
(including backups or reports) to be processed. Under the auspices of AMO, omnitrig will
check for scheduled AMO sessions and triggers omniamo and provides to it the parameters
from the saved AMO configuration file(s).
Schedule Media Copy offers three timeframe choices when defining the schedule, they are
Relative, Absolute, and No Time Limit.
Relative
The relative time option apportions a timeframe based on the two input values, namely
Started Within (hours) and Duration (hours). Started Within establishes the beginning
of the timeframe, while Duration sets the actual duration of the time frame. This defines a
so-called window of opportunity, starting some number of hours before the actual AMO start
time.
For example, an AMO is scheduled at 1200; specifying relative time option, we may choose
Started Within = 14 hours and Duration = 8 hours. Now AMO seeks all media associated
with backup sessions that started between 2200 the night before and 0600 the next morning,
and attempts to copy them.
A conflict can be anticipated in case one or more backup sessions that were started within
the AMO time frame were still running beyond this time frame, and simultaneously AMO was
attempting to copy the media that this particular backup specification would produce.
In such situations, AMO will not be able to copy media that are related to that particular
backup specification because they are still locked by the BSM. The AMO session displays the
following error message:
Source medium <medium ID> could not be locked and will not be copied
in this session.
The different times used in this hypothetical case conform to a typical business enterprises
backup and copy time window.
Absolute
You set the scope in terms of absolute days to search for backup sessions. The drops down
arrows serve to open a calendar. This option would probably be used for one-time vaulting
purposes, or to vault media from a certain time to another!
No Time Limit
This option selected will include all backup sessions, irrespective of when they were
performed. It is expected that this option would be used rarely.
2. In the Scoping pane, right-click Automated Operations and select “Add Scheduled
Media Operation” to start the configuration wizard.
3. In the Media Operation Name field, type a user-defined name. This user-defined
name acts as a prefix for that particular configuration file name, and may include or
be the same as the name of a backup specification.
4. In the Media Operation Type drop-down list, select Media Copy; (the only choice in
DP51) and click Next to select the devices for the copy process.
5. Select the source drive(s) and map them to their corresponding destination(s) to
produce a ‘device pair’. On closer observation of the figure above, notice that in
Library 2, lib2_drive1 is disallowed from the Destination list; this is due to the
selection in the Source list. Click Next.
6. Specify the time frame within which you want to search for completed backup
sessions, for scheduling to take place. There are three timeframe choices, namely
Relative, Absolute, and No time limits. Click Next.
7. Select the backup specification(s) of the backups you want to copy. Click Next.
Media Condition:
Media Protection:
9. Specify the number of copies to be made, eject mode stipulating whether either of the
media will be ejected after copying, location for the target media (if ejected) and
protection for target media. Click Next.
10. Right click on a date and select Schedule from the pop-up to open the Schedule
Media Operation dialog box (shown below). Specify the various options
accordingly, and click OK. Click Finish to exit the wizard.
Select Use Starting to delay the first performance of the copy operation and specify the
starting date.
duration 8h
Student Notes
The sequence on the slide demonstrates the sequence of events for a scheduled media copy.
They are as follows:
1. Omnitrig reads the AMO schedule file and starts the session manager (MSM).
2. The MSM selects media from the database that match the selection criteria
(timeframe).
3. The MSM starts the necessary agents (CMA, BMA) to create the tape copies.
4. The MSM ejects the tapes as configured.(optional)
5. The session ends.
Student Notes
Data Protector 5.1 introduces a new CLI utility called omniamo.
omniamo will initiate Automatic Media Operations (AMO), it currently accepts media copy
parameters. The syntax is as follows:
For example:
In case of post-backup AMO, omniamo requires the session ID of the backup session; the
media of which you want to make a copy of. The session id may be exported as follows:
NOTE: If the session ID is unknown, use the omnidb –session command to list all
previous session stored in the internal database.
Configuration Directories
AMOs configuration files are stored in two directories namely amo and amoschedules. The
directories will contain AMO associated configuration files and schedule files respectively.
Windows
Configuration: <OMNIHOME>\Config\amo
Schedules: <OMNIHOME>\Config\amoschedules
UNIX
Configuration: /etc/opt/omni/amo
Schedules: /etc/opt/omni/amoschedules
The directory amo contains both the post-backup as well as the scheduled AMO
configuration files. On the other hand, amoschedules contains only scheduled AMO
configuration files corresponding to scheduled AMO specifications. A typical CONFIG file in
amoschedules stores detailed AMO schedule information, whereas its corresponding file in
amo stores only AMO specification information.
For example, see below the different contents of the same AMO configuration file
ScheduleTape.amcs, in amo and amoschedules directories:
Configuration File:
Schedule File:
-start
-starting 1 4 2003 -every
-day Sat -month Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
-at 15:45
The name of a typical AMO configuration file can be broken down into its two constituent
parts, namely a prefix and a suffix.
The prefix may not necessarily be suggestive of the backup specification name, however the
suffix defines whether the AMO operation is post or scheduled. Post-backup AMC
configuration files suffix in .amc, while scheduled AMO suffix in. amcs.
Regarding the prefix part of the configuration files, Data Protector clearly distinguishes the
manner in which the configuration files of the two forms of AMC are labeled. While, it allows
a user-defined name as prefix for scheduled AMCs, it requires that the post-backup AMO
prefix is the same as the backup specification name..
Limitations
• Only entire media can be copied as opposed to copying selected objects or sessions
contained within the media. It is expected that Object Copy (in addition to Media
Copy) shall enable such functionality in future Data Protector versions.
• Media Copy marks both source as well as target media as non appendable.
• During media copy, the media being copied will be unavailable for restore.
• Data Protector treats the various sub-types of the same device type (e.g. DLT or LTO
etc.) as homogeneous, for the purpose of copying; however, it forbids forward
compatibility, while allowing backward compatibility within one device type (e.g.
DLT7000 can use DLT4000 media, but DLT4000 cannot use DLT7000 media). i.e., it is
possible to copy a DLT4000 to a DLT7000 media, but not vice-versa Therefore, you are
advised to choose appropriate source and destination devices.
• The time frame options under scheduled AMO can only be specified in hours. (started
within and duration options can only be specified in hours)
• Mount request handling is not implemented. If, indeed a mount request is received
either from BMA or CMA, the device pair is aborted, while the session continues to its
end.
• Device locking in AMO takes place at the outset of a session. It is imperative for AMO
to lock at least the pair of devices (corresponding to source and target media for each
media type) to complete the session successfully. The AMO session will fail
prematurely if the minimum number of devices necessary for the session cannot be
secured at the beginning of the session.
Vaulting:
• Offsite protection
• Stored in safe facility
Student Notes
The process of “vaulting" media is essentially a form of protection. Tapes are typically packed
up and sent to an off-site safe storage facility. Tape rotations typically involve moving the
tapes off site, and then back onsite after some defined period of time.
Data Protector supports the following features to facilitate tape rotations and vaulting:
• Media Protection - inherited from Backup Operations
• Media Pools with Strict Allocation - Media Pool Feature
• Multiple media pools of the same type
• Media Location tracking - Individual Media Feature
• Media Labeling - Individual Media Feature
• Multiple Media Pools - Each with a specific purpose
• Media Duplication (scheduled, automatic, or manual)
• Vaulting Locations - Pre-configured locations
Vaulting Locations
Device and
allocation order
media
context
List of media
vault [barcode label] description - [Physical Vault
locations label if no location] location
barcode Library slot
<config>/vault_locations
Student Notes
The vaulting locations may be pre-configured into Data Protector using the Device and Media
Management GUI. Vaulting locations are stored in the OMNICONFIG directory as an ASCII
file named vault_locations. You may edit this file using an editor instead of using the GUI.
The vaulting locations are used as media location strings, assigned to individual tapes, and
stored in the Data Protector Media Management Database. When media are moving from one
physical location to another it is a good idea to update the Data Protector Database with the
correct physical location of the media, as this will be displayed whenever a tape is requested
by a running session (backup or restore).
You may change the media location from within the Media Manager, simply select a medium
and then Edit -> Modify…. You may also change the media label at the same time.
NOTE The tape does not have to be in a tape drive to be modified. The label and
location are stored in the Data Protector database along with the Medium-ID
that Data Protector generated for the tape when it was initialized.
Label may
be modified
Medium
selected
Location
Vaulting
may be
locations
modified
Media pool
may change
omnimm -modify_medium
omnimm -move_medium
Student Notes
Data Protector provides another possibility for vaulting operations. Multiple media pools may
serve as media repositories when media are to be taken offsite, or just removed from a device
repository. You may want to create a media pool for each physical location that a tape may
be stored, such as:
Active_pool This is the set of media available within a device repository (library)
the active pool could also be considered a "scratch" pool
On-site_vault Tapes here are out of the device, but not yet offsite
Off-site_vault Tapes are physically at a remote location.
Free Pool May be used as a holding area for expired media, prior to moving to
the active pool.
Data Protector provides both the GUI and command line to allow you to move media from
one pool to another of the same type. The command line could be used in conjunction with
an automation script to make the media management simpler. Consider the following worked
example for providing automated vaulting operations and media management.
Command Example
The following example demonstrates the command line method of modifying an existing
medium that is moving to a different location (you may change the label at the same time):
Scenario: Tapes are to be taken out of the library each morning and sent offsite for vault
storage. The post-exec feature for the backup specification is used to create a list of media
used during the backup session and store it in a file for further processing. Two scripts will
be used.
The collection.sh script is used as a session post-exec script. It simply starts the
manage_media.sh as a background job, passing to it the session information, then exits.
There may be two ways to pass data from the running session to the manage_media.sh
script; one way would be to create temporary files, the other way would be to pass variables
directly as positional parameters. The sample provided includes both methods, with one
method commented out.
The manage_media.sh processes the session information and collects the media data from
the Data Protector database. The number of media used by the particular session is counted
and stored in a temporary file. The file is read and each medium listed is moved to a different
media pool, and the label and location are updated. A record of the changes is also logged
into another file for verification. (this could be printed or emailed if desired).
Collection.sh (sample only, not provided with the product; create the script in
/opt/omni/lbin on the management server)
#!/usr/bin/sh
# collection.sh
# execute as a session post-exec (from /opt/omni/lbin)
# optionally store the session and key in a file
#print $SESSIONKEY >/tmp/KEY
#print $SESSIONID > /tmp/ID
Manage_media.sh (sample only, not provided with the product; create the script on
/opt/omni/lbin on the Unix Cell Manager system)
#!/usr/bin/sh
# manage_media.sh
# executed from collection.sh
# Env Vars for session Post_Exec script
# DATALIST name of the datalist
# MODE full or incr backup
# OWNER session owner
# PPID parent process of the session
# PREVIEW 0 or 1 if in the preview mode
# PWD current directory
# RESTARTED 0 or 1 if restarted due to prior failure
# SESSIONID omni session id of the running session
# SESSIONKEY session key of the running session
# SHELL type of unix shell in use
# SMEXIT status of the backup session
#######
# optionally read data from KEY and ID files if created by collection.sh
#SESSIONKEY="$(cat /tmp/KEY)"
#SESSIONID="$(cat /tmp/ID)"
#######
SESSIONKEY="$1"
SESSIONID="$2"
#
VAULT=file_vault # destination pool (considered to be the vault)
MED_PRE=VAULT # prefix added to each medium moved to the vault
LOC="onsite vault" # location for each medium moved to the vault
# Collect the media list from the database
/opt/omni/bin/omnidb -session $SESSIONID -media -detail|\
awk -F ":" '$1 ~ /Medium Label/ {print $2}' > /tmp/media_$SESSIONKEY
sleep 30
#sleep to allow the database record locks to be freed
num_media=$(wc -l /tmp/media_$SESSIONKEY)
print number of media used = $num_media
print
print contents of /tmp/media_$SESSIONKEY
cat /tmp/media_$SESSIONKEY
sleep 5
exec 4< /tmp/media_$SESSIONKEY
COUNT=1
# perform the media operations move and change label/location
while read -u4 CURRENT_TAPE
do
print moving media ${COUNT}, $CURRENT_TAPE to $VAULT
/opt/omni/bin/omnimm -move_medium "${CURRENT_TAPE}" ${VAULT} && \
/opt/omni/bin/omnimm -modify_medium "${CURRENT_TAPE}" \
"${MED_PRE}_${CURRENT_TAPE}" "${LOC}"
((COUNT+=1))
done
exec 4<&-
exit 0
2. What protection features does Data Protector provide to safeguard the integrity of your
backups?
•
3. Data Protector provides two media allocation policies, strict and loose. Briefly,
describe the purpose of these policies.
• strict
• loose
6. How can you change the label on an existing tape? What about the location?
Does the tape need to be loaded into a tape drive in order to change the label or location?
8. What happens during a media import? Explain why this would be necessary.
9. To remove bad media from a pool, you export it. TRUE or FALSE?
10. What must be done before Data Protector can use media?
12. When initializing media, the size in megabytes set by specify or determine specifies
the hard limit as to how much data it will hold. TRUE or FALSE?
13. By default, Data Protector will automatically initialize blank media. TRUE or FALSE?
Explain.
Usage:
Usage:
•• Backup
Backup
•• Restore
Restore Logical Device Definition
Physical
Device
•• Format
Format Device Options
•• Copy
Copy Physical
•• Scan
Scan Properties
•• Verify
Verify
Media
Student Notes
Data Protector does not reference physical devices directly; rather, it uses a logical
representation of the device known simply as a Logical Device.
The Logical Device concept is used because it allows for easy configuration of device options
and greater flexibility in changing devices after backups have been configured.
A Logical Device consists of a physical part (such as a device file name, SCSI path, or drive
index), and a logical part (parameters that control Data Protector's usage of the device).
The Logical Device definition is stored in the Data Protector media management database,
commonly referred to as the MMDB.
Logical Devices are used for all Data Protector operations that require access to a physical
device, for example backup and restore, media initialization, media scanning, media
verification, media duplication and media listing.
Student Notes
Data Protector allows you to configure the following predefined Logical Device types.
• Standalone
• Stacker
• SCSI-II Library (used for SCSI and Fiber Channel)
• Jukebox
• External Control
• GRAU DAS Library
• StorageTek ACS Library
The configuration of a file device is very similar to that of a tape drive, except that the
filenames of the Physical Devices are the actual files that the backups will be written to. The
files would be treated just as any other medium, they will be initialized with an Data
Protector tape header. The default size for the Data Protector file medium is 100 MB, this
may be increased by manually initializing the file. (see the module for Media Management for
details)
Standalone
This is the most common type of device used with Data Protector.
Cascades
A cascade is a series of standalone type devices (of the same type) connected to one host
that the user wants to use in sequence. Cascades are useful for backing up more data than
would be able to fit on a single tape, and no operator intervention is desired to manually
change tapes during the backup. A cascade is essentially a standalone device with more than
one physical device.
To configure a device cascade within Data Protector you have to select the device policy
“standalone.” In your standalone device configuration, it is possible to define multiple device
files. For example:
/dev/rmt/c1t0d0BEST
/dev/rmt/c1t1d0BEST
/dev/rmt/c1t2d0BEST
Data Protector will use the physical devices in the same order as defined in the standalone
device configuration.
Only one backup device license is required per cascade during backup operations.
When you add a new SCSI-II Library, you first define the robotic. This includes the
definition of the control device file, the system to which the robotic is connected, the
number of slots used, and the use of such items as cleaning tapes and barcode readers.
The configuration of the data drives is similar to the configuration of standalone devices.
You define the data device file, the system to which the drive is connected, the drive
index number within the library, and the type of drive and medium pool used by this
device.
The SCSI-II library policy cannot be used to run GRAU devices or large StorageTek silos.
These are managed with StorageTek ACSLS software.
Jukebox Configuration
The Jukebox is very similar to the SCSI-II Library, but only works for Magneto Optical
devices. The major difference to SCSI-II library is that a pair of special MO drivers, which are
available only on HP-UX, operates the Jukebox. (schgr, ssrfc)
• As an MO Jukebox (preferred)
The difference with a Jukebox is that the user accesses a side of a platter and the operating
system driver automatically mounts the platter into an available drive. The driver for the
Jukebox understands the concept of media rotation, so as to allow both sides of the MO
medium to be accessed.
The user cannot define into which drive the platter will be loaded.
This is implemented with the use of a special magneto optical driver set, available on HP-UX.
NOTE There are separate device files for side A and side B of the Magneto Optical
platters.
The configuration of an MO Jukebox is similar to an SCSI-II library configuration. Again, it is
split into two steps:
1. Define a library name and the device files for all MO platters.
2. Define the drives.
To use all the drives within a magneto optical jukebox simultaneously, you have to create as
many “logical” drives as there are physical drives available.
In comparison to a SCSII library, you are not required to specify a device file, as the HP-UX
Jukebox device drivers handle the assignment of the drive to the device file. See the
procedure later in this module for configuration of the Jukebox on HP-UX.
NOTE If you configure a jukebox within Data Protector and use it for backups and
restores, this jukebox should not be used by any other application
Stacker
Stacker devices have a cartridge with multiple media slots. The difference between a Library
Device and a Stacker is that a Stacker has no control over media selection, simply load and
unload. Stackers can only load media sequentially from the cartridge, while libraries can
randomly access the media loaded in their repository slots
Device Configurations
Logical device
Drives
Hardware compression
Library controller
Robotic and barcode reader
Repository slots
Mail slot (import/export)
Cleaning slot
Student Notes
While most of the configuration within Data Protector for a Logical Device contains
parameters added by Data Protector, you must first understand how the operating system
makes the physical device available to you before you can configure it as a Logical Device.
Data Protector supports devices connected to many different Media Agent platforms, and
each one represents devices in a slightly different manner. The two platforms for the Cell
Manager, Unix and Windows usually detect and add their devices automatically.
There are numerous ways to configure Data Protector to make use of the backup devices that
are available. Within Data Protector, from a Logical Device perspective, it is possible to:
• Configure a single physical device multiple times, each with a different name and set of
properties
• Configure a physical device to have multiple device files, and then configure each one a
separate Logical Device (most common on Unix systems)
• Configure multiple single physical devices as a single Logical Device
• Configure a tape library more than once, each with a sub-set of all the available drives
and slots (this is necessary when a library contains drives of different types and where
the repository contains more than one type of media)
NOTE: There is a global option to control the queuing time (in minutes),
SmWaitForDevice, 60 is the default value.
Windows Host:
Unix Host:
# /opt/omni/lbin/devbra –devices
Configuration Methods
Automatic Manual
• Standalone devices • All device types
• Libraries (SAN or SCSI) • All parameters user selected
• Two step approach • Adjust options as needed
• select system/device
• select autoconfigure
• Automatic Lock Names
• Adjust options as needed
Student Notes
Data Protector offers two methods of configuring devices, namely Automatic or Manual.
With Automatic device configuration, Data Protector executes the device agent (devbra) to
scan the Media Agent hosts and assemble a device list. From the list, hosts or devices are
selected by the Administrator and the Logical Device configuration process is then started.
Individual device options are set to the product defaults, but may be altered as needed once
the auto-configuration process completes.
With Manual device configuration, the Administrator specifies all parameters needed to
configure the Logical Device. Using this method, a high degree of control over all of the
parameters is available, but this procedure requires more experience and a thorough
understanding of the hardware and the operating system device to be configured.
Using the
command line
omnidownload, omniupload
Student Notes
The Data Protector Administrator can define a Logical Device via the GUI or by editing a
template. The templates in <OMNICONFIG>/devices may be uploaded into the Data
Protector Media Management Internal Database with the omniupload command.
Logical Device configuration via the GUI is easier than the using the command line, and
therefore the recommended method.
To modify the configuration of an existing device or to just extract the device configuration
from the database use the omnidownload command; this creates a file which then may be
edited and uploaded with omniupload command.
NDMP Server Server name required when NDMP is the interface type, this also
requires the NDMP Media Agent to be installed
Library Options
In addition to the general options available to the standalone device, the library device adds
some configuration parameters for the library components.
NDMP Server Server name required when NDMP is the interface type, this also
requires the NDMP Media Agent to be installed
TIP! Using the omnidownload command to save device configurations in text files
may be beneficial when performing disaster recovery, as device configurations
would need to be created before restore operations would take place.
omniupload/omnidownload
omnidownload may be used to extract device information from Data Protector
database into a file.
omniupload may be used to upload a new device configuration into the Data
Protector database from a file.
Examples:
The following table provides some samples using the omniupload/download commands
Standalone device
Library robotic
device
Student Notes
Robotic Device File The system device file that is used by the operating system to
(SCSI-pass-through communicate with robotic controller.
device file) Example: /dev/rac/c0t5d1 (HP-UX)
Drive Index This number indicates the position of the drive within the library.
This is used to identify to the robot which drive to load tapes
into.
Barcode Reader Informs Data Protector if the library includes a barcode reader
Support mechanism.
Busy Drive Handling This tells Data Protector how to react if it unexpectedly finds that
a drive already has media in it. Choices will vary depending upon
the capability of the library.
Repository slots The repository slots available in this logical library (This may be
different from the physical settings.)
Drive(s) A library can contain one or more drives. Each drive must be
defined within Data Protector in a similar way as any other
standalone Logical Device. As of version 4.0 of Data Protector,
each drive within the library may be a different type of device,
such as DLT, LTO, etc.
Libraries on HP-UX
Before you configure a SCSI-II Library (Tape or Optical), a SCSI Pass through driver or auto-
changer driver (sctl or schgr) must be installed and configured properly on your system.
When you have a choice of drivers, the schgr driver makes the Logical Device configuration
simpler, as Data Protector is able to detect the device name. If the sctl device file is manually
configured with the mknod command, Data Protector is able to use the device but may be
unable to detect it when performing the system scan. The library controller in this case would
need to be specified by manually entering its name in the “SCSI address or filename of the
library robotic” field when configuring the tape library.
Slot range
Used with
Detect dirty
drive
Student Notes
The configuration of the Library Repository allows the Administrator to select all or some of
the available slots within a particular library. The slots may be specified in a range, as shown,
or as individual slot numbers. The slots need not be sequential, although this is most
common. It is possible that a tape library is configured more than once as a logical device,
each time with a different set of physical slots associated with it. This type of configuration is
very useful when the tape library contains more than one type of tape drive (DLT, LTO).
The Cleaning Slot option specifies which (if any) of the library repository slots contains a
cleaning tape(s). Data Protector will use this slot with the logical devices that have enabled
the “dirty drive detection” option. Dirty drive detection is performed only once, and before
the backup process starts. If a drive reports that it is in the clean-me mode, Data Protector
will load the cleaning tape before the backup of data begins.
Uses of the cleaning tapes (loaded automatically by Data Protector) are logged in the
cleaning.log file stored on the cell manager in the /var/opt/omni/log/cleaning.log on Unix and
C:\Program Files\Omniback\log\cleaining.log on Windows.
Student Notes
The Data Protector Library configuration controls access to the library only; you must
configure a logical device type "drive" for each tape drive within the library. After the
completion of the Library configuration you will be prompted to create a drive for the library,
select yes to configure the drive now. Later you will be able to add a new drive to the library
configuration by selecting the library from the "Device and Media Management" context and
then Edit -> Add -> Drive… . Be prepared to provide the device file that corresponds with the
library drive index number.
The example shown on the slide is for a HP SureStore Library; note the data device path may
be something like: c0t1d0BEST, where t1 represents the SCSI target ID; this is used to
match with the drive 1 (index 1) for the physical drive in the library.
NOTE On some devices, the drives may be numbered starting with zero (0); this
would correspond to the first drive and the Data Protector index number of
one (1). Data Protector starts index numbers at one.
NOTE If you get the Exchanger Data Device and the Drive Index numbers matched
incorrectly, Data Protector may load a tape into the wrong drive. For example
Drive 2 gets a tape loaded and then the data is sent to Drive 1. This will result
in a backup failure!
Library Operations
Data Protector uses its various agents to access the library devices during backup, restore
and media management sessions. Which agents are used depends upon the session type.
Data Protector uses the following agents to access the data (tape) devices in the library:
Data Protector uses the following agent to manage the library activities, such as load and
unload:
The UMA is the only agent that is interactive, and available as a command. The command
executable is in: <OMNIHOME>/lbin on Unix and in the <OMNIHOME>\bin directory on
Windows. There is also a man-page for it. The UMA is useful for troubleshooting and testing
the tape library operations.
NOTE When backup, restore and media operations are in progress, no interaction
with UMA is recommended.
Student Notes
Data protector offers the following choices for device media type and Default Media Pool:
Advanced Options
Student Notes
Library drives as well as standalone devices support additional Advanced options to control
how the devices are to be used during backup. The table below presents these options. All of
the advanced options have default values, and do not require changes in order to use a
particular device. The advanced options allow for more granular control over performance
characteristics of the device.
Concurrency This defines the maximum number of concurrent data streams (from
disk agents) that the device will receive. Setting this to an optimum
value for a particular device type allows the device to stream. This can
have various effects on backup performance. Generally, faster devices
such as DLT 7000 should be configured with higher concurrency values
than slower devices such as DDS.
Eject Specifies whether the tape should be ejected after the operation that
has accessed it completes. The default is not to eject. This is only
useful for standalone tape drives.
CRC Check Use this button to write CRC checks to the media used with this logical
device. The CRC check allows you to verify the accuracy of the data
written to the media with the verify operation in the GUI or by the
omnimver command.
Rescan This library option instructs Data Protector to rescan the device
repository before a backup starts. This is useful if manual media
changes were performed since the last media scan. This rescan
synchronizes the Data Protector media database with the media that is
currently present within the library repository. For devices that
support barcode readers, this is a barcodes scan; otherwise the scan
requires each tape to be loaded into a drive to scan the header.
Dirty Drive This tells Data Protector to detect when a drive is in need of cleaning.
Detection It senses this via the SCSI status bytes received back from the drive. If
this option is enabled, Data Protector will either automatically clean
the drive itself or issue a mount request for a cleaning tape to be
loaded. The cleaning check is only performed once, and before the
backup begins. Drives that set the SCSI status during a backup
execution may cause the backup to fail with IO errors.
Block Size The device hardware processes data it receives using a device type
specific block size. Data Protector allows the adjustment of the size of
blocks it sends to the device. The default is 64 KB on devices
connected to UNIX systems and 56 KB on devices connected to some
Windows systems. For Data Protector to use tapes for backup in
different devices, the block size must be set the same for all devices.
The maximum block size is currently 1024 KB. Your device/interface
adapter may not allow for large block sizes, consult with the vendor for
supported block size for your devices. Larger block sizes (greater than
56K) on some Windows systems require modifications to the registry,
and may not be supported. Disaster recovery requires the default block
size to be used on Windows systems.
See the Administrators Reference Manual for details.
Segment Size Use this field to enter the size of the data segments on the media. The
segment size affects the speed of a restore. A smaller segment size
requires additional space on the media, because each segment has a
fast search mark. The additional fast search marks result in faster
restores, because the Media Agent can more quickly locate the segment
containing the restore data. Optimal segment size depends on the
media type used in the device and the kind of data backed up. By
default, the segment size is in the range of 100 to 2000 MB, depending
on the medium type.
Disk Agent Data Protector Media Agents and Disk Agents use memory buffers to
hold data waiting to be transferred. This memory is divided into a
Buffers number of buffer areas (one for each Disk Agent, depending on device
concurrency).
Each buffer area consists of 8 Disk Agent buffers (of the same size as
the block size configured for the device). You can change this value to
be anything between 1 and 32, although this is rarely necessary. There
are two basic reasons to change this setting:
• Shortage of memory
The shared memory required for a Media Agent can be calculated
as follows:
DAConcurrency*NumberOfBuffers*BlockSize
Reducing the number of buffers from 8 to 4, for instance, results in
a 50% reduction in memory consumption, with performance
implications.
• Streaming
If the available network bandwidth varies significantly during
backup, then it becomes more important that a Media Agent has
enough data ready for writing to keep the device in the streaming
mode. In this case, increase the number of buffers.
Mount Request The script to be executed after a mount prompt request has been
Script outstanding for the number of minutes configured as the Mount
Prompt Delay. The default script simply sends an email notification to
the backup owner containing the relevant details.
Default template: /opt/omni/lbin/Mount.sh (Unix)
C:\Program Files\Omniback\bin\Mounts.sh (Windows)
Mount Request The time in minutes that must of elapsed since a mount prompt was
Delay issued before the Mount Prompt Notification script is executed.
Default: 30 Minutes.
Lock Name Used when a physical device is defined more than once for Data
Protector. The use of the same lock name for each use of the physical
device prevents Data Protector from trying to use (and failing) the
same physical device more than once concurrently. It is common to
configure a tape drive as more than one Logical Device when you
would like to apply more than one set of option to be used for different
operations such as backup. In some cases you may want a different
combination of block size, segment size and concurrency. The lock is
name is usually optional, and just a string of text that you choose;
required for SAN configured libraries.
Device Concurrency
Concurrency = 1
Disk Agent
Concurrency = 3
Student Notes
Data Protector supports a Logical Device feature called concurrency; this allows for the
simultaneous (concurrent) backup of multiple objects to a single Logical Device at the same
time. Concurrency is one of the most commonly altered device options. This option controls
the DA to MA ratio for the device. The maximum concurrency per device is 32; however the
device defaults are usually much lower, in the range of 2-5.
The concurrency feature is primarily designed to keep the Media Agent streaming data to the
device to achieve the best device performance. Data Protector will attempt to start the
required number of Disk Agents sending data to the Media Agent simultaneously to satisfy
the concurrency of the logical device during backup operations. Fewer Disk Agents may be
started for backup due to the number of objects included in the backup specification.
Data Protector supports parallel backup as well as concurrency. Parallel backup allows for
multiple tape drives to be used (in parallel) within both a single backup job execution as well
as with multiple backups running within the Data Protector Cell at the same time. The
example at the top of the slide (above) shows parallel backup, both logical devices set to
concurrency one. Any number of logical devices with varying concurrency values may be
used within the cell according to cell’s license limits.
There are two global variables that have an impact on the concurrency level as well as the
level of parallelism. They are :
MaxMAperSM (32 default, range 1-32) controls the maximum number of load balanced media
agents per session manager.
MaxDAperMA (default 32, range 1-32) controls the maximum concurrency value (DA to MA)
for logical devices.
Concurrency Implications
• Backup Performance
The performance of some linear tape devices, such as DLT, may be negatively impacted
by a lack of streaming. This should be a consideration when configuring the logical device
for backup. Some newer linear tape devices such as HP’s Ultrium allow for speed
variation (adaptive write), and can better accommodate the data flow provided by the
Media Agent to stream the drive.
Higher values for concurrency, however, don't necessarily mean higher backup
performance. Consider performing a full backup on a server with a partitioned disk. In
most cases multiple objects will reside on the same physical disk. Using concurrency to
backup the partitioned disk may lead to disk head contention, and the backup may
actually take longer and not allow the tape device to stream due to the disk bottleneck.
The bottom line is that concurrency values are designed to improve backup performance
in most situations. The organization of the objects within the backup specification will
have an impact as well. You will need to test various solutions in your environment to
find the best possible combination of objects and concurrency to achieve the best
performance.
• Restore Performance
The performance of Data Protector restore may be negatively affected when a device
configured for concurrency was used for backup and you are not using parallel restore to
recover data. Data Protector will have to read more of the tape to restore a single object
that was interleaved onto the tape with other objects during the backup process.
Generally speaking, backups are performed much more than restores, so concurrency
values should be configured to achieve the best backup performance possible.
Tape Format
Tape
Image
Data Segments
Header minimum 10 MB
150-1000 MB
defaults
EOD
End of
Segment Data
Student Notes
The Data Protector tape format supports the following features:
Tape Sections
The Data Protector tape is comprised of the following sections:
• Tape Header — Data Protector writes a tape header and tape label. There are two Data
Protector labels: one is the user-defined label; the other is the Data Protector medium ID.
This ID is unique and used by Data Protector internally. The tape header uses only one
block on the tape.
• Segments — Data to be backed up is written to a segment. The size of the segment can
be configured in the Logical Device configuration window with advanced options. If
larger segments are used, more memory is required on the system on which the media
agent is running. This memory is used to store catalog information. If device concurrency
is used, the data within the segment will be from more than one disk agent. Larger tape
segments can improve the performance of the backup in many cases; this should be
tested within your environment.
• Dynamic Segment Size --- Segment size is no longer a fixed size (as of 4.0). The
parameter above is used to specify the maximum size of the segment on tape. The
segment size used will now be determined by the segment size parameter, or a system
specific parameter named OMNIMAXCATALOG_<device_name>. By specifying a
catalog size per device on a particular system, you can limit how large the catalog
segment will be on the tape. The default segment size is 12 MB, and can range from 1 to
60 MB. Data Protector may adjust the size of the segment if the catalog reaches the
defined limit. The catalog size takes precedence over the specified segment size. The
parameter to define the catalog limit must be in the omnirc file on the system where the
device is connected.
• Data Blocks ---- Data stored within the segments are written in blocks. The block size for
most Data Protector devices is 64 KB by default (file devices and reel tapes use 16 KB, D3
uses 256 KB.) This default is now used for both Unix as well as Windows NT. In prior
versions of Data Protector the block size for Windows NT was at 56K. You should set the
block size to equal values if you want to exchange tapes between different devices. In
many cases, when backing up a large data set, a larger block size may improve
performance. The required block size for most Disaster Recovery backups is 64KB.
• Catalog Information — Catalog information is stored after each segment is written and
records what data (file names, etc. ...) was backed up in that segment. When the data is
written to tape, the catalog information is kept in memory and then written to the tape at
the end of each segment. The larger the segment, the more memory is required to keep
the backup information. The catalog information is also stored within the Data Protector
database. This information is later used during the restore process. Catalog information
may be read from the tape into the database by performing a media import. (Media
Import is covered in the next module). The size of the catalog per segment by default is
12 MB, but can range from 1 to 60 MB. See the previous description for Dynamic Segment
Size.
Block Size
Although it is possible to change the block size for a device, it is advisable to be mindful of
the following when doing so:
Mount Notification
Defaults
Delay: Programs: _____________
30min /opt/omni/lbin/Mount.sh
<product>/bin/mount.bat
Parameters Description
USER User name
GROUP User group name
HOSTNAME Name of host (short)
STARTPID PID of the process that caused the mount prompt
DEVNAME Logical Device Name
DEVHOST Short name of host where device is connected to
DEVFILE File representing the device
DEVCLASS DAT standalone, DAT exchanger, DAT stacker, MO standalone ...
DEVCLASSNAME Currently same as devclass
MEDID Medium ID
MEDLABEL Medium user label
MEDLOC Location of the medium
POOLNAME Name of the pool
POLICY Strict, Loose, App+Loose, App+Strict
MEDCLASS Medium class number
MEDCLASSNAME DAT/DDS, Optical, Double Sided Optical, Exabyte, 3480, Reel Tape, File
SESSIONKEY Session key to be used by omnimnt <config>/options/global:
MountDelay=<DelayInMinutes>
MountScript=<Full Pathname>
Student Notes
Data Protector provides a script template called Mount (Mount.sh on Unix, Mount.bat on
Windows) that may be executed whenever a device needs a tape. Every device may use a
different script, or they all may share a single Mount script. By default, the Mount script will
send an e-mail, or Windows broadcast to owner of the backup at the cell server after 30
minutes of waiting for a tape to be loaded.
The Mount Prompt Script executes in response to an event. The event in this case is a Mount
Request. So in essence, Data Protector is providing an event driven notification mechanism.
The type of notification or notification method is up to you. For example, you may want to
execute some paging software in response to the event.
NOTE The mount script may be customized to perform functions other than
standard email. For example, it can interface with paging software to alert the
operator, or to issue a message to a management application such as
OpenView Operations by issuing the opcmsg command.
A mount notification is issued when a tape that Data Protector requires is not loaded in the
drive or available within the repository of the library. For example, when a backup is writing
to a standalone device and the tape has been filled, Data Protector will request a further tape
to be loaded to continue the backup. The mount request may also occur when the tape in the
drive is from a different media pool than the one assigned for the backup. When this occurs,
the following events can take place:
• Data Protector puts the session into the Mount Request state.
• Data Protector issues a mount prompt (when interacting with the session) and then waits
until the mount request is satisfied.
• Data Protector waits for the mount delay time (30 minutes) and then executes only once
the notification script.
The Data Protector Operator should satisfy the mount request from the Monitor GUI or
with the omnimnt command.
The Mount Notification Script does not confirm the mount; it simply issues the
notification according to the script instructions.
NOTE In a later section of this course we will discuss in more detail some additional
notification procedures that may be used for Data Protector events including
mount events.
When Data Protector executes the mount notification script, it passes the following
positional parameters to it:
Parameter Value
THIS=${0} or %0 The name of this script
USER=${1} or %1 The UNIX username
GROUP=${2} or %2 The UNIX group name
HOSTNAME=${3} or %3 The expanded hostname
STARTPID=${4} or %4 The UNIX PID of the client process that started backup
session.
DEVNAME=${5} or %5 The name of the logical device
DEVHOST=${6} or %6 The hostname where the logical device is located
DEVFILE=${7} or %7 The physical device file
DEVCLASS=${8} or %8 The device class number
DEVCLASSNAME=${9} or %9 The device class name
MEDID=${10} The medium ID
Library Sharing
System A
Data Only MA
inet
MA
TCP/IP SCSI or
Fiber Channel
Control
and
Session Data MA
Manager
inet
UMA
Robotic
System B
Student Notes
Data Protector allows the sharing of library devices between multiple systems. This is can be
achieved because Data Protector separates the definition of a library into two parts, control
and data.
Data Protector controls the robotic of a library via a separate device than that of the drives.
The robotic control will either have a dedicated SCSI interface that is attached to the
controlling system, or will share the same interface as the drives.
The Data Protector Session Manager instructs the UMA (Universal Media Agent) running on
the system, which has the library robotic control attached, to perform all the library functions
(load, unload, eject, scan) etc. This allows other systems that have library drives attached but
no direct robotic control to issue library control requests (load, unload, eject) to the session
manager. The session manager then gets the Universal media agent to carry out the requests
on behalf of the requesting system.
With direct library access, all hosts may send library control commands directly to the
library. The library control host which coordinates access to the library robotics. If this
controlling host if unavailable then the host sends controls directly to the library. This is
typically used within a SAN environment, where many hosts reside within the same SAN, or
SAN zone.
To enable the direct access mode, a configuration file must be created on all hosts called the
“libtab” file. After the libtab file is created, the “direct access” option may be used for the
library configuration.
The purpose of the libtab file is to map the library robotic access from local host (media
agent client). Create the libtab file on each Windows or Unix system requiring direct access
in the event of a library control host failure. The libtab file is a plain text file with the
following format:
<FQDN of the local client> <Device File or SCSI path> <Library Name>
On HP-UX systems, the Device File must be the same on all hosts that are to access the
library. If the device files are not the same, a symbolic link may be used on the secondary
hosts to create a device name that matches the original name on the controlling host.
Note: The libtab file may be copied to each host requiring direct library access or a
client specific file may be created.
Tape Library
smocl4
HP:C7200-800
(DLT 8000)
SAN
smocl3
smohpu04
Student Notes
This example assumes a configuration consisting of three systems and one library consisting
of 4 drives. The library is connected via SAN to three systems, i.e. all three systems can “see”
all four drives and also the robotic of the library.
The goal of the configuration is that one system (smohpu04) gets access to the robotic of the
library, and all systems access all drives. After configuration, the same lock name for all
logical drives should be created, representing the same physical drive.
Select
Autoconfigure
Select hosts
for library
configuration
Student Notes
Data Protector 5.1 includes the functionality from the “sanconf” tool (from DP 5.0) now as
part of the device GUI. This new capability makes SAN based library configuration very
simple and mostly automatic (with the exception of the previously mention libtab).
Right click Devices and select Autoconfigure Devices to start the auto configuration wizard.
The wizard asks you to select all the systems for which libraries or standalone devices are to
be configured. Click Next. Data Protector scans all selected systems for attached devices.
This scan is done by the tool devbra, and may take several minutes.
The DP 5.1 devbra tool now produces much more user-friendly output, as compared to
previous versions.
Switch between
Device or Host
grouping
Select system
to control the
robotic
Select systems
which should get
access to the
devices
Student Notes
This slide contains two separate screenshots, which illustrates the step two of the auto
configuration wizard. Two different presentations help to determine the assignments
between the systems and the library. With the grouping by devices (left screenshot) the
presentation is based on libraries. The right screenshot shows a grouping based on systems.
This shows all assignments from the system point of view.
NOTE: Only one system can access the library robotic, indicated by the radio button
(circle). For the drives it is different. Several systems can access the same
device (square button).
where n is only used in case the drive is already setup inside Data Protector with the same
name. Since the name must be unique within DP a number is appended. Right-click the drive
name to rename it before the configuration starts.
drive1: QUANTUM:DLT8000:CKA32P3224
drive2: QUANTUM:DLT8000:JF90908606
drive3: QUANTUM:DLT8000:JF90413627
drive4: QUANTUM:DLT8000:JF90909085
Library Scanning
Student Notes
Data Protector provides a Library Management interface that may be useful verifying library
operations and checking library status.
The slide above shows how to access Library Management to perform scans of the library. If
the library is configured for barcode capability, the barcode scan will report the media label
without having to load the any tapes into the drive. Scan will actually load a tape and read the
tape header. The scan is considered to be a "hard scan" of the tape and is usually avoided
when using libraries, especially if multiple slots are selected due to the amount of time that
the operation may take.
Example: The output of the barcode scan is also available with the uma utility.
NOTE The uma utility may read from the shell standard output on Unix: Also notice
the cleaning tape in slot 15. The CLN prefix for barcode labels is recognized by
Data Protector to indicate a cleaning cartridge.
/dev/dltrobot> stat
Element Status (T=Transport, X=Im/Export, D=Drive, S=Storage):
0 T1 Empty "" ""
Student Notes
Data Protector is able to perform many functions related to media management by selecting
the Logical Device within the device and media context. Selecting the device (library, slot or
drive) will allow for the following operations:
Operation Description
Scan Identify the tape label and format. Data Protector will recognize
several common Unix tape formats. (this is a hard-scan)
Enter Load a tape from the Mail-Slot (supported on libraries that have
such a feature enabled)
Eject Eject a tape to the Mail-Slot (supported on libraries that have such
a feature). Can be automated for bulk-eject, see the reporting/
notification or media management sections of this course manual.
Import Read the header and catalog information from a tape to register it
into the media management database.
Recycle Remove all protection from a tape. Use with caution, as this
allows a tape to be overwritten!
List Cartridge Allows access to the data stored in the LTO (Ultrium)
Memory cartridge memory.
Bulk Operations
Data Protector allows for “bulk” media processing tasks to be performed in accordance with
the general capabilities of the individual tape library. The use of the term “bulk” in this
context means multiple slots/tapes simultaneously or sequentially from a single select
operation. Within a single session Data Protector allows for the selection and then
initialization of all (if desired) of the tapes within the library repository. The Media Session
Manager will load and unload tapes as needed to perform the requested task.
Data Protector is also able to eject multiple tapes for library devices that support the mail-
slot feature. Both enter and eject operations are available relative to the available mail-slot.
External Control
Student Notes
This device type is designed to allow for simple and efficient support of non-standard
autochanger devices that do not work with the standard SCSI pass-through driver.
To keep this mechanism as simple as possible, but also powerful and flexible enough to deal
with a large variety of autochanger devices, the Data Protector media agent uses only two
operations: medium load and medium unload.
Both commands are invoked through the same external script, which must be capable of
parsing predefined options and parameters.
At runtime, the Data Protector media agent will call the external script, parsing it for all
necessary information to complete the required action. The script should perform the action
and return an exit code of zero if the operation was successful, or a positive integer in case of
an unsuccessful operation.
During the execution of the script, any messages read from its stdout will be picked up by
the media agent and passed to the controlling session manager as error messages at level
minor.
The script should be able to handle the commands:
Drive <Drive #>
flip
load
unload
The media agent assumes that the autochanger is online and reset to an operational state
before it is used. If the autochanger has been used before and left in an inconsistent state, the
load/unload operations will probably fail and abort the media agent.
The external script should catch this situation and reinitialize the autochanger to a default
operational state. The media agent does not issue a special reset or initialize command
for this purpose.
Some autochangers have removable cartridge magazines that can be loaded and unloaded
under software control. The media agent assumes that the magazine is loaded and does not
attempt to preload it at startup or unload it at shutdown. If the specific autochanger supports
magazine loading, the external scripts should detect an unloaded magazine and load it
transparently to the media agent.
The media agent launches the external script; therefore, it runs on the system on which the
Logical Device is configured.
Caution! The external script runs with root permissions, therefore the security of
this script is very important.
let num=0
let flip=0
for arg in $*
do
TAB[${num}]=${arg}
let num=${num}+1
done
let i=0
while [ ${i} -lt ${num} ]
do
if [ ${TAB[${i}]} = "-slot" ]
then
let slot=TAB[${i}+1]+1
fi
if [ ${TAB[${i}]} = "-drive" ]
then
let drive=TAB[${i}+1]
fi
if [ ${TAB[${i}]} = "-load" ]
then
let load=1
fi
if [ ${TAB[${i}]} = "-unload" ]
then
let load=0
fi
if [ ${TAB[${i}]} = "-flip" ]
then
let flip=1
fi
let i=${i}+1
done
if [ ${load} -eq 1 ]
then
/usr/omni/bin/HTC4 insert $slot
exit $?
else
/usr/omni/bin/HTC4 eject
exit $?
fi
• Integration modules
installed on client
systems
Student Notes
These large libraries are typically seen in mainframe environments and are controlled by a
dedicated system running either DAS or ACS software.
In order to use these devices with Data Protector, the additional Unlimited Slot Library
License must be purchased. The special integration modules also must be installed on the
client systems that are connected to the library.
1. What Data Protector device type would you use to configure a single DLT drive?
2. What Data Protector utility can be used to check communication with a library device
before you configure it as a Logical Device?
3. What HP-UX driver does Data Protector use to control the robotics of library devices?
HP-UX?
Windows?
8. What command can be used to get a listing of all the logical devices in your cell?
10. What command can be used to perform a “scan” of the media in a device?
11. The “Drive Index” is related to the SCSI address of the drive. True/False?
12. Is it possible to have more than one logical device that maps onto the same physical
device?
13. If you were able to, why would you create more than one logical device for a single
physical device?
14. What is the purpose of the Lock Name advanced option? Is it ever required by Data
Protector? If so, when?
• Adjust all of the options that control the execution of a backup job.
Performing Backups
Student Notes
Data Protector provides several methods of performing backups. The method to use depends
on the flexibility required.
To execute this type of backup, the user simply selects the backup specification required,
and executes it via a menu item. No other information is required, as all the information is
contained in the backup specification.
Example: using omnib to backup a filesystem, notice the interrupt takes you into
command mode for the session where interaction is possible, disconnect to keep the session
running without further interaction:
# omnib -filesystem na168w2:/tmp "practice-command" -device dlt_drive1
[Normal] From: BMA@na168w2 "dlt_drive1" Time: 04/20/02 14:43:07
STARTING Medium Agent "dlt_drive1"
Datalist:
• DP - General host and file system
Barlist:
• SAP - SAP database and logs
• Oracle 7 - Oracle 7 databases and logs
• Oracle 8 - Oracle 8 databases and logs
• Informix - Informix databases and logs
• Sybase - Sybase databases and logs
• MS SQL - Microsoft SQL databases and logs
• MS Exchange - Microsoft Exchange server
Student Notes
Special types of backup specifications must be used to suit the particular type of source data
being backed up (objects).
Datalist
This is the most commonly used type of backup specification. It is used to backup:
• General file systems (drives) of UNIX, Windows, and Novell systems
• Windows Registry (configuration)
• Data Protector Internal Database
• Rawdisk sections
Barlist
• SAP/R3
An SAP database instance, online or offline, including log files
• Oracle 8/9
An Oracle database instance, online or offline, including log files
• Informix
An Informix database instance, online or offline, including log files
• Sybase
A Sybase database instance, online or offline, including log files
• MS SQL
An MS SQL database instance, online or offline, including log files
• MS Exchange
A Microsoft Exchange mail server.
NOTE The MS backup specifications cannot be configured via the HP-UX GUI. These
must be configured via the NT GUI.
• Filesystem (UNIX/Windows/Netware)
• Hosts (expanded into file systems).
• rawdisk
• the DP internal database
Student Notes
The source data specified within a backup specification is referred to as Objects. When
backed up, the source data is classified according to the specific object type required.
Backup specifications may contain different object types, such as a host backup mixed with
filesystem backups from different systems.
Data Protector provides the following object types:
• file system
This type of object is used for data residing in the operating system file systems. A
backup using this object type backs up the source data, file by file.
Examples: /usr (Unix file system) C: (windows file system)
Vxfs, HFS, NFS, UFS, FAT16, FAT32, VFAT, NTFS, NTFS 5.0, others.
NOTE See the Data Protector Administrator’s Guide for details about each of the file
systems and some specific details or special considerations for each type.
• host
Also called Disk Discovery; this type of object is used for data residing in file systems. A
backup using this object type backs up the source data, file by file. It is different from the
file system object in that the specific objects to be backed up are determined dynamically
at run time. File systems such as NFS cannot be backed up using this object type, they
must be included as file system objects.
• rawdisk
This type of object is used for data backed up from rawdisk sections (image copy). For
example raw logical volumes on HP-UX systems; /dev/vg01/rlvol1 or
/dev/rdsk/c0t6d0.
• omnidb
Backups of the Data Protector internal database have this special object type. This
ensures that the database is backed up in a consistent state. This backup will be
discussed in more detail in the Database chapter of this manual.
• vbfs
Data originating from HP OpenView OmniStorage MFS (Migrating File Systems).
OmniStorage is HP’s hierarchical storage management solution. Data Protector uses a
special integration to backup OmniStorage data.
• winfs
Data originating from Windows NT/2000, ME, 98 and XP drives. Registry data is also
stored using this object type and is called CONFIGURATION. There are many
considerations for backing up and restoring the CONFIGURATION, see the Data
Protector Administrator’s Guide for more specific details.
• netware
Data originating from Novell NetWare server drives.
Template Options
Defaults
Object Options
Objects
Device Options
Devices
Student Notes
Data Protector provides a rich set of options that can be used to define all the characteristics
that the administrator wants the backup job and stored data to have.
• Template Level
Options associated with a template will set the initial options for the backup. These
“defaults” may be altered for each of the following three additional levels.
• Backup Specification Level
Options available at this level are common to the whole backup session, including the
objects and device specifications that are defined within the backup specification.
• Object Level
Options at object level are specific to that particular object. A backup specification can
contain one or many separate objects. Each object can have different options. Typically,
objects within a backup specification tend to have the same or similar options. (Defaults
are inherited from templates.)
• Device Level
Options here are specific to the particular logical device. A backup specification can
contain many device definitions. Each one can have different options.
Choose:
Choose:
•• Load/non-load
Load/non-loadbalance
balance
•• Source
Source
•• Destination
Destination
•• Backup
Backupspecification
specificationadvanced
advancedoptions
options
•• Filesystem advanced options
Filesystem advanced options
•• Schedule
Schedule
•• Backup
Backupobject
object summary
summary(properties/options)
(properties/options)
Student Notes
The list shown above illustrates the typical sequence (as guided by the GUI) used when
defining a backup specification.
The next several pages will first explain and then illustrate each of the choices shown above.
Load balance was previously explained, so next will be the explanation for Source.
Choose a
template for a set
of default options
Static or
dynamic
devices
Use the
command
to create a
file
••omnicreatedl
omnicreatedl… …
••Edit
EditASCII
ASCIIfile
fileininOMNICONFIG/datalists
OMNICONFIG/datalists
Student Notes
Data Protector is able to generate backup specifications within the GUI, first by selecting an
appropriate template, and then which template options to apply to the new specification.
Once the template is chosen, objects and their options, backup specification options and
devices and their options may all be modified.
Command Line Example-1: Create a datalist containing all of the file systems for a single
host, the file produced will be host168w2, for the host na168w2, and the logical device
dlt_drive1:
Command Line Example-2: Create a datalist containing all of the file systems from two
hosts, the file produced will be <OMNICONFIG>/datalists/gencombo.
NOTE See the man page called omnidatalist for details on the structure and
syntax used within the file.
The datalist file produced by the command line compared to the one produced
by the GUI may be structured differently when more than one host is included.
Load Balancing
Objects
Devices
Student Notes
One of the first choices you will have to make when creating a backup specification is the use
of load balancing. If you refer back to the previous slide, you will notice the callout indicating
the on/off switch for its use. Load balancing avoids the pitfalls of static device allocation by
enabling dynamic device allocation. With dynamic device allocation, objects are not targeted
at a specific device, rather at a pool of devices. This feature is designed to be used with multi-
drive tape libraries, but will also work with stand-alone devices.
Using this method, the creator of the backup specification does not need to worry about what
objects are sent to each device. Data Protector allocates the objects to a device when the
device finishes backing up the previous object.
Load balancing can also make a backup more robust, as it is possible to define more backup
devices in the backup than are actually required. If a device fails to start, Data Protector will
mark the drive as failed and another device will automatically be used in its place. In some
earlier versions of Data Protector, this was known as Auxiliary Devices.
The creator of the backup specification defines the minimum number of devices that must be
available to successfully complete the backup. In addition, the maximum number of devices
is also defined.
For example: A load balancing setting of MIN=1, MAX=3 defines that one device is required
to complete the backup successfully, however if three or more devices defined in the backup
specification are available and needed, then three (maximum) will be used.
Student Notes
Static device allocation occurs load balance is disabled, then each object defined in the
backup specification is targeted at a specific logical device. All objects can be targeted to the
same device, or to different ones.
The creator of the backup specification must make a decision as to which objects are sent to
each device. This feature allows for the most control during backups.
The following factors need to be taken into account to get the best performance:
• The number of objects sent to the device
• The size of the objects
• The speed of the device and its concurrency
• How well the object data compresses
• The order in which the objects are backed up
If one device in the list has more or larger objects directed to it, it could still be in use and
have objects pending to it while other devices have finished their objects and are sitting idle.
This is not usually ideal for the best overall backup performance, but may improve
performance by giving more control to someone that is very familiar with the data source,
such as the administrator.
System B Device
Pending Unused
(Max=2)
Student Notes
On the left, we can see a backup specification; the backup specification contains four objects
and three logical devices. The options for the backup specification include: “Load Balanced,
Min 1, Max 2.” Also, note that the backup objects are in a specific order, that is to say they
have an order within the object list.
When a backup specification is configured as load balanced, you will notice that the device
field for each object that normally shows the name of the logical device the object is targeted
at, now shows “Load Balanced.”
At run time, media agents are started for each logical device specified in the backup
specification. However, Data Protector will only start the number of media agents defined in
the MAX parameter, in this case “2.”
The media agents that are started depend on the order defined in the backup specification. In
this example, it is System D and System A Device. The third device, System B Device, is not
used. The backup specification defines “2” as the maximum number of devices that can be
used.
If one of the devices as not available, for example it is in use by another backup or restore
session, or it failed to start, then Data Protector can use the third device instead.
Once the media agents have started, the disk agents are started. The number of disk agents
started is the combined concurrency values for the running devices; in this case, the total is
three.
Data Protector first sends an object to its local tape drive rather than send it over the
network to a drive on another system.
In our example, the first System-A object has been assigned to the System-A device. The
device concurrency is set to one, so no additional objects are started for the device.
The other System-A object has been allocated to System D device, along with the System B
object, as this device has a concurrency of two.
The reason these two devices are used is that they are the first two devices in the list.
The System C object is in the pending state, as it is waiting for one of the devices to have a
free concurrency slot. The device that finishes backing up one of the objects first will be the
device that receives the System C object.
In this way, devices are constantly fed objects until the backup is complete.
NOTE As the user does not know in advance what objects will be written to each
device, it makes sense to use a common media pool for each device that is to
be a part of a load balanced backup.
Load
balanced
Non-load
balanced
Student Notes
When defining objects interactively to be added to a backup specification, you may use a
“task wizard” to create the specification. Within the backup context of the GUI, select
“Tasks” form the bottom of the Scoping Pane. There you will find the wizards. Select either
the load balanced, or non-load balanced wizard. You will not be able to change the load
balance selection later unless you edit the datalist file with a text editor.
As you select objects to be backed up, you may select the check box in front of a host to
include the “host” object, or you may expand the host object and select file systems
individually. The coloring used for the check marks in front of the objects indicate whether
the items were selected directly (blue) or indirectly (black)) because of another selection.
The lightened colors (green and gray) are used to indicate partial selections.
Proceed through the rest of the choices by using the Next Æ button at the bottom of the
Results Area.
Source
primary
Host object
secondary
secondary
Filesystem primary
objects
Student Notes
What to backup?
Multiple object types/selections may be combined together within a single specification for
backup. The choice of host object and file system object may also be mixed with different
agent platforms. The backup is executed as a single session (a single job) comprised of many
objects.
While the flexibility of combining objects together is quite high, you may want to consider
how you will restore data before you define your backups.
How to restore?
Data Protector provides essentially two methods for data restore, object-based, and session-
based.
With session-based restore, Data Protector is able to restore all at once the objects from a
single backup session. The backup session is stored in the Data Protector database and may
be selected for restore. This makes restoration of a complete system very simple, but may
change the way that you will define your specifications for backup. With object restore, you
may select to restore an entire object version, or any subset of it, down to the file level.
NOTE See the Data Protector Administrator’s Guide for a list of files to exclude.
Destination
Select library to
use all drives
Use drive
default
media pool
Student Notes
The destination for a backup may be a library or individually selected logical devices. Each
device selected may have modifications made to its default properties. You may override the
device default properties for each backup without affecting the device defaults saved within
the Media Management Database.
Device Properties
Properties are modified on a per device basis. Therefore, each device in the backup
specification can have different options. The following options are available:
• CRC Check (default is per device definition, and is usually off)
Set this option to ON to have Data Protector calculate the Cyclical Redundancy Check
(CRC) when a backup is run. The CRC check is an enhanced checksum function that lets
you confirm whether or not data has been written correctly to the medium. This causes
additional processing and tape writes to occur for each backup.
NOTE CRC data can be rechecked using the Data Protector verify function in the
Media Manager, or with the Data Protector command: omnimver
Concurrency allows more than one Disk Agent (up to five) to write to one backup device
concurrently. This helps Data Protector keep the device streaming when it can accept
data faster than a Disk Agent can send it.
If this option is ON, Data Protector will updates library repository information before
starting your backup. This is useful when you manually change media order, or enter and
eject media from a library and you want to avoid mount requests during backup.
NOTE If the library has been defined to support barcode reading, then a barcode
scan takes place; otherwise, a physical scan will take place.
• Media Pool (default is as per the logical device definition)
This option selects the media pool from which the device should source media. The
sample above indicates “no media pool selected,” which is interpreted to mean use the
device default pool, no specific pool requested for this backup.
NOTE Pre-allocation should be used only with the strict media allocation policy and
on limited basis. Pre-allocation lists can be inflexible and confusing. If a
backup does not take place, the order of media in the list may not be what is
required for the next backup. Use of this feature will require daily changes to
the datalists. It is possible to request in-appropriate media (full/protected).
Data Protector only checks this list at run-time.
Non-modifiable
unless disk
image added
manually
Student Notes
Backup specification options encompass the whole backup session, including the objects and
devices contained within it.
Example: shutdown_application.sh
• Post-exec
The partner to pre-exec, here you can specify a script file that will be executed after the
backup has completed. Such a script can do the necessary application startups that must
be performed following a backup, so that they are available to the users. This post-exec
may be executed on any system within the Data Protector cell, but must reside within the
same directory tree as the pre-exec scripts.
Example: startup_application.sh
We will discuss exec scripts in more detail later in this module.
• Reconnect Broken Sessions (default is off)
This option can be used to increase the robustness of a backup when it is susceptible to
communication breakdowns due to unstable networks.
Ownership is not truly a backup specification option. However, it affects the backup
specification and the objects that it contains.
Clustering Options
The options for the backup specification fall into two categories, Data Protector Options and
Clustering.
Features that are designed to work within the MC/ServiceGuard and Microsoft Cluster
environments.
• Automatic Session Restart
− Abort if less than n minutes: Prevents newly started backups from being restarted.
− Abort if more than n minutes: Prevents long running jobs from being restarted.
NOTE See the man page for the omniclus command for more details and examples.
System B Object
Backup Specification Level can be BSM
Object Post-exec
executed on any system in the cell.
Object Pre-exec
Object Level is always executed on System C Object
the system where the object resides. DA
Object Post-exec
Student Notes
Typically, before a system can be backed up, the administrator must secure the system by
shutting down the various applications and databases that the system is running. After the
backup is complete, the administrator restarts the applications and databases, making them
available again to the users.
Data Protector through its Pre- and Post-exec facilities can perform the shutdown and
startup from within the backup itself.
Data Protector provides two levels of pre/post-exec, the first is at the backup specification
level, and the other is at the object level.
Pre/post execution scripts must complete (or send output) within 15 minutes or Data
Protector will abort the backup. This is used to avoid execution hangs. This timeout value is
adjustable by modify the ScriptOutputTimeout parameter in the
<OMNICONFIG>/options/global file.
Object Level
The level is available for every object within the backup specification. The pre-exec will run
before the actual object is backed up, and the post-exec will run after the object is backed up.
The scripts are executed on the system where the disk agent retrieves the object data from.
CAUTION When using Host objects, the pre/post execs are run before any objects are
backed up, and once for each object to which the host is expanded.
CAUTION When using combined concurrency levels greater than one (1), you cannot
predict the order in which objects will complete, therefore, do not put commands at the
object level, which rely on particular object completion orders.
NOTE Tasks that never terminate, such as startup of applications or daemons, must be
detached from the scripts in order to avoid time-outs. To achieve this the scripts that contain
such daemon startups may use the UNIX at command to detach the script from the pre-exec
script:
/usr/bin/at -f script_file.sh now
or with some delay
/usr/bin/at -f script_file.sh now + 3 minutes
To check the error output of your pre- and post-exec command and make it
visible in your message window (monitor), always redirect stderr to stdout. To achieve this
on a UNIX system, use output redirection:
unix_command 2>&1 &
Any non-zero return value from the pre-execution command will result in the
backup or backup object being aborted. We suggest that the pre- and post-exec commands be
scripts terminate with exit(0) when executed successfully.
Fail
1 2
Fail 3 Fail 4 Fail 5
Fail
BDACC SMEXIT
Set Set
Student Notes
The pre/post-exec feature of Data Protector is extremely useful. However, it is very important
that you understand what effect errors and failures of the scripts have on the backup.
1. The backup specification level, pre-exec script is run. If the script completes successfully
(exit code 0), it moves on to step 2. If the script fails, then the backup does not take
place, and it jumps straight to step 5.
2. At this stage, the pre-exec script for the first object is run; if it works successfully, it
jumps to step 3.
3. At this stage, the object is backed up. If the backup of the object succeeds or fails, it
jumps to step 4. However, before it goes to step 4, the BDACC environment variable is set
to reflect the status of the object.
4. At this stage, the object post-exec script is run; if the script succeeds or fails, it jumps to
step 5. However, before it goes to step 5, the SMEXIT environment variable is set to
reflect the status of the entire backup session.
5. At this stage, the backup specification, post-exec script is run; then the backup session
completes.
The following table shows the resulting backup session status that is seen in the monitor
window, following various failures (pre/post with non-zero return values):
BDACC Status
Value
0 Normal, successful termination
1 Program failed, user error
2 Program failed, environmental
malfunction
3 Program failed, internal malfunction
4 Program failed, reason unknown.
SMEXIT Status
Value
0 Completed.
10 Completed with file errors.
11 One or more disk agents failed.
12 All disk agents failed.
13 Session was aborted.
For example, your post-exec script may read the SESSIONID variable and use it as a
parameter to the omnidb command, to find out what media was used by the session. The list
could then be emailed to the operator:
TIP A Pre or Post- exec script may hang because it did not close all file descriptors
before forking the new process. This is the case if the new process runs in the
background and does not exit (for example, database server process
(dbstart), some daemon processes, etc.).
In this case, a user can use the detach command. The source of the
detach command is provided in the detach.c file. This command is
officially unsupported.
For example:
/opt/omni/bin/utilns/detach pre_script
[arguments...]
Instructs
InstructsData
DataProtector
Protectorto
touse
useaamore
morerobust
robustprotocol.
protocol.
IfIfcommunication
communication between BSM and anagent
between BSM and an agentfails,
fails,the
the
BSM and the agent attempt to re-establish communication.
BSM and the agent attempt to re-establish communication.
Network Backup
Cell Manager
Scheduler
TCP/IP
Session Session
Manager Manager
Student Notes
The “reconnect broken session” backup specification option can be used to increase the
robustness of the backup. It can be used when backups are taking place over networks that
are susceptible to interruptions, such as WANs. When this option is enabled, a more
advanced protocol is used for agent communication and data transfer. This protocol has a
performance overhead, and therefore, should be used only if the link reliability is a problem.
If the BSM loses communication with the disk or media agents, the BSM and agents will both
try to re-establish communication.
The agent will try for OB2RECONNECT_RETRY seconds to re-establish the TCP/IP connection
and will then wait for another OB2RECONNECT_ACK seconds to get the acknowledgement
from the server. If either one of these time parameters times-out, the object will abort. The
default for the OB2RECONNECT_ACK variable is 10 minutes, and for the
OB2RECONNECT_RETRY the default is 20 minutes. These settings can be changed by placing
the variables in an OMNIHOME/.omnirc file on all systems involved.
If the BSM is unable to contact the host to start the agents for the first time, the object is
rescheduled to the end of the backup specification, where one further attempt is made to
communicate with the host.
Student Notes
Each object to be backed up within Data Protector may have a unique set of options to
control how the data is backed up and protected within the Data Protector database.
The only general option is the session protection. The default is permanent and may be
changed to:
None
Days <number>
Weeks <number>
Until <yyyy> <mm> <dd> (year, month, day)
Data Protector will protect the session data on tape from overwrite until the protection
expires.
Options at the object level are specific to that particular object. A backup specification may
contain one or many separate objects. Each object can have different options. Typically,
objects within a backup specification tend to have the same or similar options. Rather than
setting these options individually for each object, Data Protector provides a template that can
be modified in advance of adding objects to the specification. Newly added objects then
inherit the new default settings provided by the template.
Object options are divided into two categories, options and other:
Advanced Options
NOTE Object level pre/post-exec scripts will be run on the system that the object
resides on. They must be located within the OMNIHOME/lbin directory
tree if a relative path is used; they may be located anywhere if an absolute
path is used.
• Catalog Protection (default same as data)
This option enables you to set periods of protection for information about backed up
objects in the Data Protector database. You may want to expire the catalog but keep the
tape protected if the tape is stored off-site for extended periods of time, and you want to
keep your Data Protector database smaller. This concept will be explored further in the
database chapter of this course manual. The default value is the same as for data
protection.
− Until
Information in the Data Protector database is retained until a specified date. You
enter the year, month, and day. Protection for the information will cease at noon of
the entered day.
− Days
Information in the Data Protector database is retained for the number of days
specified.
− Weeks
Information in the Data Protector database is retained for the number of weeks
specified.
Other Options
TIP Software compression can be useful when backing up data over a WAN.
The data is compressed on the source system before it is sent over the
network, thus reducing network traffic.
If set to ON, Data Protector will back up the entire file contents for each hard link. This
will prevent hard links from being recreated during the restore process; every filename
that was previously a link is restored as a file. Data Protector traverses the file system
tree only once during the backup, thus speeding up the backup process. When this option
is set to ON, Data Protector cannot estimate the size of the backup or display the
percentage of the backup finished.
NOTE Use this option when there are no hard links in your file system.
• Logging (default is log all)
Data Protector provides four levels of detail on files and directories stored in the Data
Protector database. Data may be restored regardless of the level chosen. The logging
option potentially reduces the amount of data stored in the database. The logging level
also affects the amount of detail available to the restore browser as well as the backup
performance.
− Log All
This is the default option. All detail about backed up files and directories are logged
to the database. This complete information allows you to search for backed up files,
and allows Data Protector to fast position on the tape when restoring a specific file.
However, this information may take a lot of space if there are many files.
− Log File
Details on directories and only file name information are stored in the database. No
file version attributes such as modification time, protection, owner, size, etc are
stored in the database (they are on tape only). This represents a savings of about 70%
over the log all option.
− Log Directories
Details on directories only are stored in the database. This disables the search feature
during restore, and you will be able only, to browse directories. However, Data
Protector still performs fast positioning because a file is located on the tape near the
directory where it actually resides. This option is suitable for file systems with many
auto-generated files, such as news and mail systems. The data stored with this option
represents a savings of about 90% as compared to log all.
− No Log
No details on files or directories are logged in the database. You will not be able to
search and browse files and directories while restoring. The restore will take longer
because Data Protector cannot use fast positioning on the tape but will read from the
start of the backup. The primary information stored here is at the object level. You
would use this if you expect to restore an entire file system object and not select
individual files and directories by name.
Winfs Options
Netware Options
Object Summary
List sorted
based upon
order
Student Notes
The final step in creating a backup specification is to review, and possibly change the objects
and options selected for the backup. Here you may also change the order of the objects in the
list. The order will affect the execution sequence and pairings for concurrency. The object list
order along with the algorithm for load balancing will determine the backup sequence.
From this object summary screen you may also modify any object options individually.
Notice that you may select the column headings in the summary within the Results Area to
change the sorting preference for the list.
This object summary list is the last configuration screen before you save the specification
and/or start the backup.
Object Properties
Only…
Skip …
Student Notes
The slide illustrates where to find some of the object properties where you may fine-tune the
scope of the backup. From here, you may select the parts of a files tem object to backup,
instead of the entire tree, which is the default. The Trees list is essentially an Include-only list
for the backup. The exclude list allows you to specify the absolute path of the files or
directories to exclude from the backup. When the lists are empty, the entire object (file
system) will be backed up.
Use the Filter … button to specify a wildcard list of names to include or exclude. The “Only”
list is used for include, and the “Skip” list is used for exclude. In both cases, the list
represents a filter for the entire file system. Whenever a match for the filter is found, the item
is either skipped or included in the backup.
Object Qualifiers
The data that is to be backed up requires qualification that is more detailed for Data
Protector. Configuring Data Protector to backup a file system is not sufficient information. A
complete description of the object, such as where it resides, which parts of it are to be
backed up, etc., must be specified.
Specialized backup specifications, such as Oracle, Informix, MS Exchange, etc., have other
qualifiers, such as the instance or SID name. However, we will not be detailing these options,
but focusing on the backup specification datalist and options instead.
TIP Data Protector uses three key qualifiers to identify file system objects in the
database, they are Hostname, Mountpoint and Description. These are used for
restore and reporting.
The following list details the most commonly used qualifiers used with the backup
specifications:
• Hostname
Specifies the particular system in the cell that the object resides on.
Example: vindaloo.uksr.hp.com is a fully qualified hostname.
• Mountpoint
Specifies the file system mount point on a UNIX type system, or the drive letter on a
Windows or Novell system
Example: /opt - a UNIX file system mount point
/ - a UNIX file system mount point for the root file system
C: - a Windows drive letter
CONFIGURATION - The windows registry and recovery files
The more meaningful your object description, the easier to locate the object for restore!
Disk Agent
5
Filesystem Disk Agent Don’t forget to
Disk
consider how
you will use
restore!
Student Notes
Data Protector allows for multiple data streams from one file system to be used for a backup.
This is very much different than device concurrency, with is multiple objects being sent to
one device. The purpose for creating the multiple streams is for improved backup
performance.
Great care must be taken in order that you do not overload the disk, (and reduce
performance), and overlap the data streams. This configuration is manual; be aware, it is
possible to be backing up the same data more than once.
Device concurrency is still available, as Data Protector will start a disk agent for each data
stream that you configure.
Student Notes
To configure parallel data streams, use the Manual Add… feature and select the Trees option
for an object. One key to success here is that you will add a file system multiple times; this is
only possible if you change the description for each one. Recall that an object is defined by
three parts, “Hostname, Mountpoint and description.”
As long as the object description is different, you may add another thread (directory) to be
backed up for each file system. Add the data streams (directory to backup) one at a time.
Parse Backup MA
DA
Specification
Student Notes
1. A backup is started either interactively by the user or by the Data Protector scheduler.
The backup request is received by the cell manager, which in turn starts a Backup
Session Manager process (BSM).
2. The BSM parses the backup specification and checks it for errors.
3. If the backup specification contains host objects, these objects are expanded into a list of
the host's currently mounted file systems. The file systems inherit the options defined for
the host object.
5. The BSM checks that sufficient licenses are available for the logical devices defined in the
backup specification, and whether they can be locked. It will then start the media agents.
For each media agent, it starts the number of disk agents that correspond to the device
concurrency. If load balancing is used, then license checking and locking takes place, just
before the devices start, not at parsing time.
6. The Disk Agent (DA) does a tree-walk (mount point and tree information are checked).
On the first traversal, it computes the file system statistics and builds a catalog of hard
links. The DA then connects to the MA.
7. If the backup is being run in Preview mode (a dummy run), then stop.
8. The Disk Agent connects to the Media Agent and another tree-walk is performed, during
which it reads the files and sends the information to the Media Agent. When the second
tree walk is finished, the Disk Agent disconnects from the Media Agent, and terminates.
On Windows NT file systems, only one tree walk is performed unless you choose to use
the NTFS hard link option.
The Session Manager (SM) starts a new Disk Agent (if required) for the same Media
Agent. If not, the SM shuts down the Media Agent and terminates. The session is
complete. The restore session runs in a very similar way.
Templates:
• Often used backup specification and schedule
characteristics can be saved in a template.
Groups:
• In large environments where many backup
specifications and templates are required, groups can
be created to organize them more effectively.
Student Notes
Data Protector provides the ability to create templates and groups to aid the generation and
organization of backup specifications.
Templates
Many environments can require many separate backup specifications; each backup
specification may have very similar characteristics to previous ones, for example, the same
schedule, the same media pool, etc. Rather than specifying the same options each time a
specification is generated, they can be applied en masse, via a template.
Backup options applied to objects. Backup options for an object are not applied to the
objects in the backup specification unless the Override object options is selected. You
cannot undo this action.
• Object Qualifiers (Tree Options)
Options skip, exclude and only.
• Schedule
How the backup is to be scheduled.
Groups
Large numbers of backup specifications and templates can be difficult to administer. The
sheer size of a list can make it confusing and difficult to find what you are looking for. Data
Protector allows the creation of Groups that allow you to place related backup specifications
and templates together.
For example, you can create groups based on system usage (production, development, etc).
Preview
Student Notes
The Preview function allows the user to perform a “preliminary run” of a backup
specification. The preview will run through all of the backup steps, with the exception of
actually writing any data to the backup device. Therefore, it is a very good test of backup
specification correctness. It is highly recommend that a preview be run on all new or
modified backup specifications, especially when they are to be scheduled.
When you preview a backup, the backup monitor shows exactly the same kind of information
you would see if the backup were actually running. The main difference noticed is the speed
with which the objects are completed. This is because only a tree walk and space calculation
are made, rather than any transfer of data.
CAUTION The pre/post execute scripts may run during the preview mode. This may
cause some interruptions within a production environment. See the options in
the <OMNICONFIG>i/options/global file to turn off pre-exec during
preview. (ExecScriptOnPreview)
1. What command is used to perform a backup from the Data Protector command line
interface?
2. What are the four fundamental components of all Data Protector backups?
•
•
•
•
4. What is an object?
11. What are the differences between object level and backup specification level pre and
post execution?
Performing Restores
Student Notes
Data Protector offers two methods of restoring data, interactively through the GUI or via the
command line interface omnir command. All restores are guided interactive sessions, as
opposed to scheduled datalist backups.
In general, restores are occasional events that are performed only once in the same manner.
As such, there is no need to have the equivalent of a backup specification for restores.
Data Protector does not provide a method for predefining a restore session’s requirements,
each restore session must be defined when required. This is not a limitation with Data
Protector, rather a design consideration as restoring data is a destructive process and
incorrect or out of date predefined restores are a potential disaster, waiting to happen. The
Data Protector Administrator may create a script to automate the restore process using the
omnir command, if desired.
Data Protector restore definitions are done on backup session or object level. Within one
restore session, one or more backup objects can be selected. For each object, files and
versions can be selected. Options can be set on the restore session level, as well as on object
level.
When the restore is started, the restore session manager is executed and a restore session ID
is assigned to the restore session. The restore session is stored in the database in a similar
manner to backup sessions. These sessions may be removed, up to the discretion of the
administrator. While backup data in the Data Protector database is necessary for restore,
restore data is only necessary for auditing and reporting purposes.
Restore Objects
Student Notes
Depending on the types of backups performed, many different object types may be available
for restore. Each object has restore options that are specific to the individual object type.
There are also restore options that are common to all object types.
Object Types
• omnidb
From the omnidb object, the Data Protector internal database can be recovered,
including the <OMNICONFIG> directories. This topic will be addressed in much more
detail later in this course.
• sap, oracle, oracle8, sybase, informix, mssql, msexchange
From these database objects it is typically possible to restore an entire database, a
portion of the database (dataset) or point in time via the redo logs. Integrated third-party
databases are restored using the databases own tools, for example rman for Oracle or
sapdba for SAP, onbar for Informix.
Data
Data Protector
Protector Session
Session Restore
Restore
•• Restores
Restores an
an entire
entire client
client
•• Can
Can restore
restore all
all objects
objects from
from aa backup
backup
together
together
•• Operates
Operates like
like aa datalist
datalist for
for restore
restore
•• Can
Can exclude
exclude individual
individual objects
objects
•• Provides
Provides aa high
high degree
degree of of control
control for
for each
each
object
object
•• Makes
Makes extensive
extensive useuse of
of the
the database
database
Student Notes
In some cases, the restore of an entire system is necessary. Normally this would be after a
disaster recovery. Usually, the disaster recovery includes some out of date files or data. Data
Protector in conjunction with your disaster recovery tools allows for easy recovery of your
data from the most current backup session. Disaster recovery within Data Protector is
discussed in more detail later in this course.
The session restore capability within Data Protector is based upon how you perform your
backups. Data backed up within a single session, usually from a backup specification
(datalist), can be restored in parallel.
While selecting a session to restore, Data Protector provides individual object selection, so
you are not limited to an all or nothing restore. Each object that is recorded in the database
may be restored in parallel with any other object. By selecting a backup session for restore
purposes you are able to restore all of the data that was a part of the backup.
The Data Protector internal database plays a key role in making the session and object data
available for restore. Within each session, you will be able to browse the data trees and select
down to the file level if a partial rather than a full restore is necessary.
Parallel Restore
Client system
disk agents
DA Disk 1
MA DA Disk 2
Client system
media agent
DA Disk 1
Client system
disk agent
Student Notes
When a backup is performed to a logical device that has a concurrency value greater than one
and/or multiple logical devices, backing up multiple objects in parallel maximizes
performance.
Conversely, when it comes to restore time, the same performance benefits can be realized by
choosing to restore objects in parallel. Parallel restore can be used to restore all of the
objects to a client system from one restore session as well as objects from unrelated
sessions.
A parallel restore requires only one pass of a media in order to extract all the selected objects
from it. A sequential restore only allows the selection of a single object at a time; thus,
multiple passes of the media are required if more than one object from the backup is selected
for the restore.
• Objects that were backed up to separate devices using different media are capable of
being restored in parallel.
• Objects that exist on the same medium in different tape segments are restored
sequentially, even if configured for parallel restore.
NOTE A parallel restore may execute multiple DA processes for a single MA, just the
reverse compared to concurrent backup. In addition, Data Protector may start
multiple DA processes for a single object if the data was backed up in that
manner using the trees options.
Restore Sequence
Choose:
• Object(s) or Session for restore
• Source
• Destination
• Options
• Devices
• Media
• Backup Object Summary (properties/options)
Student Notes
The general sequence for restoring data from Data Protector tapes is listed above. The
process is conceptually similar to backup, but you may start the restore at any point. You do
not have to work through all of the option screens if you want to accept all of the restore
defaults, such as:
The following pages will illustrate the typical restore sequence; remember you can start a
restore anywhere within the sequence, once you have changed all of the necessary defaults.
Restore Source
Select objects
or a session from
the database
Choose items to
restore and their
properties
Select tasks to
search for files
Student Notes
When a restore is performed, the user is able to browse the catalog database to select
directories, files etc to restore. The graphics shown above illustrate the selection of an object
to be used for restore, and the browse features allowing for file level selection.
NOTE The three parameters that identify an object in the database: Hostname,
Mount Point, Description if these change over the life of an object, you will
end up with multiple object names in the database. Notice how /tmp appears
more than once.
The level to which the user will be able to browse the detail catalog of the database depends
on the following criteria.
• Log File: If “Log File” was used, then you may browse the tree as in “Log All”, but you
will not be able to see all of the file attributes, as they are not stored in the database, but
only on the backup tape. The choice would be made to save database space of
approximately 70% over Log All.
• Log Directories: If “Log Directories” was used, the user will only be able to browse to
directory level and must know the name of the file to be restored, as it will not be
possible to browse it. This is a tradeoff between flexibility and database space. This
choice would be made to save database space of approximately 90% over LogAll.
• No Log: If “No Log” was used, no browsing at all is possible, and the user must know
the complete path, filename, etc. This uses the least database space, but makes the
restore more challenging. This option would be used if restores are only performed at the
object level, and browsing the file/directory tree is unnecessary. Minimal disk space is
needed to store the object data as compared to Log All.
Catalog Protection
Catalog protection refers to how long the online catalog information is retained. When the
backup is performed, you can set the catalog information to keep it as long as the media is
valid, or until a particular date. If this date has passed, browsing will not be possible.
2. Fill in as much information as you have about the file name and object that contains the
file. (Case sensitive checking is optional, and wildcards are available)
3. Choose a time frame for which to search, or a range of dates for which the backup was
taken.
5. Data Protector will present a list of all matches based upon your selection criteria. Select
from the list and configure the options for the items that you want to have restored.
Partial
Version and restores
destination
also available
in pop-up
menu
Default
version is
the latest
Student Notes
Data Protector allows each object selected for restore to be “fine tuned” to meet your
requirements. You can select the version of the object, its destination, and desired, only parts
of the object restored.
Restore only allows you to specify wildcard matches to the object contents. For example
maybe you would like all of the Adobe Acrobat documents restored; you would enter *.pdf
in the “Restore Only” option screen. Perhaps you would like to exclude certain files from
the restore, such as anything containing the name core, such as *core, or core.*; enter
these in the “Skip” option screen.
Within the Properties GUI shown above, the Destination tab contains options that allow the
destination of the object to be changed. You may alter the name of the object, or place it into
a new directory. When the Into option is chosen Data Protector will append the original
object path to the selected new location.
Restore As / Into
• As
Restore the file or directory as the given path, any path specified as non-absolute will be
created relative to the root (/) filesystem.
• Into
Restore the file or directory into a different directory. Directories will be created as
needed and are appended onto the original path.
PC Restore Options
Netware
By default, Data Protector will backup NetWare sparse files in their compressed sparse
format. This choice will speed up the backup process.
Windows 2000
For Windows 2000 there is restore option for restore of Windows 2000 active directories.
• Authoritative:
This is a Windows 2000 specific option dealing with active directory restores. The Active
Directory database is not updated after the restore and the restored data overwrites the
existing data in the target destination.
• Non-authoritative:
The Active Directory database is updated after the restore using standard replication
techniques. The Nonauthoritative replication mode is the default option.
• Primary:
The Primary replication mode allows you to keep the NT directory Service online and is
used when you restore FileReplicationService along with the Active Directory service.
This option has to be used when all replication partners for a replicated share have been
lost.
Destination
Select or enter
any host with a
disk agent
Student Notes
Default Destination
The default destination for any restored object is the objects original location. You may
restore data onto any client with a Disk Agent installed (even non-cell clients) as well as
restore the data into any new location within the system. Data Protector will create the
necessary directory trees to accommodate your requested location.
Security Concern
The default behavior with Data Protector, as well as its predecessor Omniback, is to allow
any restore disk agent to respond to any session manager to start the restore process. This
feature is designed to allow for simpler restores in case of a disaster. In this way any cell
manager could be used to restore data to any client, regardless of the cell affiliation.
The client and therefore the cell may be secured to limit which cell managers have access to
the various clients. The specific details of the security configuration is covered in the security
module, later in this course.
Overwrite: replaces any files or directories already on the disk with those from the
restore tapes.
No Overwrite: determines that files and directories are not overwritten, Data Protector
will not restore files from tape even if the version on disk is older. Restored files are only
those that are not currently on the disk, such as files that were deleted.
Directory_2
File_B
If File_A and File_B are deleted, and a restore is performed selecting the contents of
Directory_1 for restore, File_A will be restored but File_B will not be restored,
because Directory_2 exists, and the "no overwrite" option will not overwrite an existing
directory to restore its contents. The Warning: “Cannot restore – name conflict. Object
Exists!” will be presented during the restore session to indicate that there are items missing
from a directory that were not restored. To restore the items, choose one of the other two file
conflict handlers.
Restore Options
Student Notes
NOTE Selecting this can greatly reduce performance because Data Protector must
examine the data to determine if it contains null data blocks; conversely, some
files may not be useable by their applications if they are not restored as
sparse.
Restore Devices
Student Notes
When restoring data, Data Protector will choose the same Logical Device that was used
during the backup for the object. In most cases, this is desirable, especially if the needed tape
is still within the repository of the library.
If the device no longer exists, a permanent change to an alternate device would be in order.
Use the omnidbutil command with the -change_bdev option to permanently change a
device to another within the Data Protector database. The omnidbutil command will be
discussed in more detail within the database module later in this course.
Restore Media
Student Notes
Shows the media that will be required in order to perform the restore.
Restore Summary
Student Notes
Last minute changes to the object list. Here Data Protector allows the addition or removal of
objects for the restore session. The properties for each object may be changed by
highlighting the object and using the pop-up menu (right-mouse-button) to select its
properties. The properties include Version, Destination, Restore Only, and Skip choices.
From the pop-up menu, an additional choice of version selection by time, allows a file version
to be chosen from “best available.” You can specify an acceptable time range for an alternate
version, if your preferred version is not available. Your selection may be from a date and time
range from seconds to hours.
Start restore
from the toolbar
or Actions
menu
Student Notes
Single Restore
When single restore is chosen, you will be prompted for the object to restore. After that
object completes, chose the “start restore” icon from the Tool Bar, and select another of the
configured objects to restore. Repeat this process until all of your objects are restored.
When
Whenrestoring
restoringto toaa
point
pointinintime,
time,the
thelast
last
full
full backup of thedata
backup of the data
must be restored first,
must be restored first,
followed
followedby bythe
the note
incremental
incremental backupsinin
backups
the
thecorrect
correctorder.
order.
Data
DataProtector
Protectortakes
takes
care
care ofmedia
of mediaselection
selection
and
andrestore
restoreorder,
order,all
allinin
the same restore
the same restore
session.
session.
Student Notes
When using a full backup plus incremental backup scheme, restores are more complex. This
is because to restore to a particular point in time (a date/time of a particular backup) multiple
restores must be performed. For example if a weekly full backup is performed followed by
daily multi-level incrementals (Mon-Fri), the user wants to recover a directory to the state it
was on the Tuesday, then the following restores must be performed:
1. Restore directory from last full backup.
The slide shows a screen shot from such a restore session, you can see the objects that Data
Protector has decided are necessary to restore and the order.
With this type of restore, it is also possible to omit files that were deleted between backups.
See file system restore options for more information.
2. Is it possible to restore a single file from a rawdisk backup? If yes, describe the
limitations.
3. The level of detail available in the restore file browser depends on which factors?
8. What is the difference between the Restore As and Restore Into options?
9. Is it possible to restore a file backed up on one system to a different system? Must the
system be a member of the same Data Protector Cell?
10. Give an example of a use for the pre/post-exec facility for restores:
11. In order to perform a valid Rawdisk backup of a file system, the file system must first be
unmounted. TRUE/FALSE?
Features
• Scalable
• Flexible
Topics
• High performance
• Reliable • Architecture
• Low maintenance • Command summary
• Automated recovery • Maintenance
• Backup
• Recovery
Student Notes
In this module, the following Internal Database topics will be addressed:
Architecture Composition of the database and general overview
Command Summary Details on using the database related commands
Maintenance Required routine maintenance
Backup Backup procedures and recommendations
Restore Restore procedures and recommendations
Recovery Recovery from corruption
Scalability
One of the major features is the overall scalability, which allows for:
• Single point control for many more systems
• Storage of catalog data for hundreds of millions of files / many backup session versions
for convenient restore browsing for a long time
Higher Performance
In the Data Protector 5.0 Database, there is:
• Less CPU load, higher insertion rate, less disk IO
• Many more systems / disks backup in parallel while tracking the catalog data for
convenient restore browsing
• Using the IDB archive transaction logs the IDB can be restored to the point in time of
disaster
• The administrator can decide when to upgrade the IDB detail part
• The detail part upgrade is performed while the IDB is online, thus backups can run
(upgrade will be suspended during that time). However, there are several limitations:
− Browsing of objects, residing on tapes not yet upgraded, will not work or will work
partially.
− Tapes with the detail catalogs not yet upgraded will not be allocated for appended
backup.
− Filename purge cannot be run.
− The upgrade will be suspended while backup is running.
The detail upgrade procedure is recoverable. If the system fails in any stage or if Data
Protector shuts down, the upgrade is resumed automatically when services are restarted.
• Growth planning
• Disk space allocation
• Transaction logs
• Recovery data
• Event configuration
• Reporting
Student Notes
The Internal Database is configured by default when the cell manager is installed. The default
configuration may be acceptable for smaller sites, but larger sites should consider changing
the default database due to higher expected quantities of data.
Configuring the Internal Database (or changing its configuration) consists of several facets:
• Growth planning
• Disk space allocation
• Transaction log management (archival)
• Recovery data management
• Event configuration
• Reporting
Internal Database
• User configuration
• Backup specifications
• Schedules
• Cell environment
• Report groups Flat File
• Report schedules Storage
• Notifications
UX: /etc/opt/omni
Windows: /<product>/config
Student Notes
Data Protector stores its information in two main locations, the first being the Data Protector
internal database, /var/opt/omni/db40 on Unix, <OMNIHOME>/db40 on Windows, and
the other being flat ASCII files located under the /etc/opt/omni on Unix,
<OMNIHOME>/config in the windows directory tree.
Data Protector IDB utilizes an embedded database technology provided by the Raima
Company called Velocis. The following information is stored in this Data Protector IDB:
• Media Management
Records describing media labels, media pools, backup objects, media locations, media
utilization, current device repository, etc.
• Device Management
Records describing logical device configuration, including logical device names, physical
device files, descriptions, lock names, etc.
• Backup and Restore Session Data
The session progress messages and object status as seen through the monitor screen and
also when viewing previous sessions.
IDB Tablespaces
Catalog
Database Catalog DB:
• Objects
• Object versions
Filenames • Sessions
(up to 32 GB) • Media positions
• Directory names
Student Notes
The Internal Database resides in the /var/opt/omni/db40 on Unix. and in the
<OMNIHOME>\db40 directory on Windows. It is highly recommended that the database
always have its own filesystem(s) or partition(s), because it can grow to be very large. The
filesystem may be accessed via a mount to the db40 directory, or on Unix, a symbolic link
from the directory /var/opt/omni to the mount point. On Windows 2000, the empty directory,
<OMNIHOME> may be used as a drive letter path to a partition prior to the install of the
product, or the <OMNIHOME>\db40 may be used for a partition containing only the
database. The downside to the partition is that it must be suitable for disaster recovery,
discussed later in this course.
The embedded Data Protector (Velocis) database is composed of two separate tablespaces
managed by the RDS server:
• The Media Management Database (MMDB) — stores logical device definitions, media
and media pool information.
Location:
− Unix: /var/opt/omni/db40/datafiles/mmdb
− Windows: <OMNIHOME>/db40/datafiles/mmdb
• The Catalog Database (CDB) — stores information about data backed up, such as
files, directories, versions, and so on.
Location:
− Unix /var/opt/omni/db40/datafiles/cdb
− Windows: <OMNIHOME>/db40/datafiles/cdb
NOTE All changes made to the MMDB and CDB are updated using transaction logs.
This is discussed in more detail later in this section.
The CDB (objects and positions) and MMDB parts represent the core part of the IDB.
The CDB tends to be much larger than the MMDB. The largest file in the CDB is:
The initial limit for the fnames.dat is 2 GB, but may be extended in up to 2 GB increments
to 32 GB. The growth of the database depends on the number of backup sessions and the
growth and dynamics of the client environment (number of new files). By frequently
maintaining the database, the size should be kept to a minimum (the minimum amount of
space required to store the information).
The filenames table will initially grow rapidly, but reach somewhat of a plateau after the
catalog retention time expires; at that point the growth will slow dramatically and remain
fairly constant. The size of the database will ultimately be determined by one complete
backup cycle of all of the data.
In an ordinary single-cell environment, both parts of the database are located on the same
cell server.
In a multi-cell environment, with the Manager of Manager licenses, you may configure a
central MMDB database for many cells. In such a configuration, the MMDB is stored on the
Manager of Manager (MoM) system. Multiple Data Protector cells can share it, and therefore,
share devices. (see the module "Manager of Managers" for more details)
Session SMBF
messages
One file per SIBF
session
Student Notes
Data Protector (as of Omniback version 4.0) stores a great deal of data associated with
backup session in data-files that are external to the Velocis database (IDB). These binary files
are updated directly by the session managers; transaction logs are not created for them.
There are three data-store directories used by default:
The DCBF will contain one file for every backup medium (tape). The names of the files in the
DCBF directory are derived from the medium-id that Data Protector assigned to the tape
when it was initialized. Data Protector will automatically remove the file associated with a
medium that becomes obsolete (exported or overwritten.)
The CDB will contain about 20% of backup file data, and the DCBF will contain about 80%,
when the backup option of “Log All” is used.
By default, there is only one DCBF directory enabled; Data Protector supports up to 10 DCBF
directories per cell.
Location:
• Unix: /var/opt/omni/db40/dcbf
• Windows: <OMNIHOME>/db40/dcbf
NOTE All changes made to the DCBF are done directly, without the use of
transaction logs.
Location:
• Unix: /var/opt/omni/db40/msg
• Windows: <OMNIHOME>/db40/msg
NOTE All changes made to the SMBF are done directly, without the use of
transaction logs.
Location:
• Unix: /var/opt/omni/db40/meta
• Windows: <OMNIHOME>/db40/meta
NOTE All changes made to the SIBF are done directly, without the use of transaction
logs.
Location:
• HP-UX: /var/opt/omni/db40/vadb
• HP-UX: /var/opt/omni/db40/xpdb
• Windows: <OMNIHOME>/db40/vadb
• Windows: <OMNIHOME>/db40/xpdb
NOTE All changes made to the vadb and xpdb are done directly, without the use of
transaction logs.
Location:
• HP-UX: /var/opt/omni/db40/sysdb
• Windows: <OMNIHOME>/db40/sysdb
Directory Structure
db40
<file
cdb mmdb catalog rlog syslog <year>
versions>
<session id>
Student Notes
The directory structure for the default Internal Database is shown above. Data Protector
allows many parts of this database to be relocated for optimum performance and
recoverability. Later in this section we will discuss the optimum disk layouts.
The previous pages discussed several of the objects listed above, some not yet mentioned:
Data Protector stores the recovery data in a file called obrindex.dat. This file is needed
for the automated off-line recovery. There can be two copies of the obrindex.dat created
during the IDB backup. The second copy is created by altering the “RecoveryIndexDir”
parameter in the global options file, and performed by the IDB backup. The copies should be
on different disk locations.
The off line recovery will also replay the transaction logs to bring the database to the state of
the last backup. Transactions that affected the DCBF are not logged, and cannot be
recovered with the transaction log replay. The DCBF data can be imported from the last used
media if it is needed.
It is recommended that the transaction logs are located on the system disk where the IDB
was installed, and the other parts of the IDB are relocated to other disks. In the case of
disaster caused by a disk failure, the logfiles and recovery information would be stored
online, separate from the database.
Unix: /var/opt/omni/db40/logfiles/syslog
Windows: <OMNIHOME>/db40/logfiles/syslog
The names of the transaction logs will be similar to the following (from an HP-UX host):
/var/opt/omni/db40/logfiles/syslog/rAAAAAAI.chg
NOTE On Windows NT, it is not possible to change the locations of the IDB
directories.
Omnis -stop
….
Archiving=1
…
Omnis –start
To minimize the disk space needed for the archived transaction logs, wait until at least one
cycle of full backups has been completed within the cell before enabling Archiving. If the
names of files are already in the CDB, then the transaction logs will be fairly small. If
filenames are not in the CDB, then each new filename added will add approximately 200
bytes to the logs. In a large environment, the size of the transaction logs may be substantial.
The suggested approach is to enable the Archiving after most of the filenames are already
recorded in the CDB.
Student Notes
The Data Protector Internal Database limits were introduced in the Architecture module.
They are mentioned here as a review for planning purposes.
The IDB has several defined (supported) limits. These limits should not be exceeded under
any circumstances. The limits illustrated on the slide are also available from the product
Release Notes document that ships with the product.
The file-names (fnames.dat) database file is initially limited to 2 GB, but may be extended
in up to 2-GB increments to a maximum of 32 GB. The minimum extension is 1 MB per
extension.
The file-versions stored in the DCBF is initially limited to one directory of 4 GB, but may be
extended in up to 4 GB increments to a default of 10 directories. The maximum number of
DCBF directories is 50, but this requires the modification of the global option “MaxDCDirs”
from the default value of 10. Each extension directory may contain up to 10,000 files; the limit
for the file versions is set to approximately 10 times the number of file-names. The file
versions represents approximately 80% of all the data stored by Data Protector within the
IDB.
TIP See the Release Notes for last minute changes and further details.
Recommended Distribution
Tablespaces
MMDB
and Binary File
CDB Directories
Logs
Transaction
and
Recovery
Student Notes
For the best performance using Data Protector’s database and for the easier recovery, it is
best to separate the most active parts of the database onto separate disk volumes. Separating
the tablespaces from the external binary files will increase the performance of Data
Protector. It is also wise to move the transaction logs onto a different disk than the
tablespaces, and make a copy of the recovery information.
Depending upon the operating system of the cell manager, it may be possible to allocation
another partition and then simply move the Internal Database onto the new disk space while
the Data Protector servers are not running. (on UNIX systems, this could be done with
symbolic links as well)
It is best to relocation these directories during the installation process, when the database
contains very little data. (Volume mounts can be built before the installation of the database.)
DCBF Manage the locations with the database maintenance commands (later
in this section)
Transaction logs Keep in the default location if CDB, MMDB, and DCBF are relocated
Student Notes
The Internal Database will continue to grow, as more sessions are executed within the cell.
Data Protector stores all the details of successful as well as failed sessions for later reporting.
The growth and size of the Internal Database are determined by the following factors:
• Catalog Detail Level
The number of files and directories backed up and the level of detail held in the database
to describe them (Log None, Log All, Log File, Log Directory)
• Catalog Protection Time
How long detail information is to be kept in the database (should be less than the media
protection in most cases)
Clients, and further objects (file systems) that reside on the clients that have very high
dynamics can negatively impact the overall performance and growth of the IDB. These high
dynamic clients are candidates for reduced catalog data logging during backup.
The selection of “Log File” or “Log Directory” will prevent the unnecessary storage of file
version information for dynamic files. The files will be recoverable from tape, but their
details will not need to be stored in the database, as they are unlikely to be requested
individually. Typically restore by object, or restore by directory is used to put back the files
onto the system.
Recall (from the backup chapter) that the “Log File” option reduce about 70%, and the “Log
Directory” option will reduce about 90% of the file information that is stored.
Setting the catalog retention time to a period lower than the physical protection time can be
useful. For example, if media is required to be kept for a long time span, but realistically, will
not be required for restore (archives, etc), the catalog can be kept for only one month, while
the data on tape is protected for 3 years.
If the catalog protection is set equal to that of the media protection, then the IDB will
continue to grow rapidly. The protection of the catalog for particular sessions may be altered
in the database by using the GUI or the command line.
One source of messages is output from pre/post exec scripts. These messages, by default,
appear on stdout, and thus will appear in the session log. If a pre/post-exec script is
generating large numbers of messages for each backup, it can take up a large amount of
space in the database. Such information can be reduced by either reducing the verboseness
of messages or by redirecting them to a logfile.
Messages generated by Data Protector itself during the backup operation are another source.
Setting the message filter level (Normal, Warning, Minor, Major, and Critical) can control
the number of such messages.
The global option file contains the parameter “DailyMaintenanceTime” to control when the
ASM executes. The default time is 12:00 (noon). A typical purge session will last for several
seconds while obsolete DCBF and SMBF files are removed for the expired media.
The manual invocation of a purge may still be needed. The omnidbutil command provides
several options to control how this type of data purge will execute.
Data Protector has two features that will help determine if an additional purge is needed.
They include a purge report as well as an automatic notification. The chapter on Reporting
and Notifications will cover both in more detail.
Student Notes
Keeping tabs on the disk space consumed by the IDB is very important. Database corruption
could occur if the disk fills up before a transaction is completed. Data Protector provides
sufficient tools to allow advanced monitoring of the space consumption for its database.
Frequent monitoring is highly recommended.
Data Protector has several built in features to aid in the monitoring of the database:
• Built-in size graphs
• Event logging and notification (database full event)
• Scheduled monitoring via reports
• Web Based reporting tools
Shown above is the size-graph available within the database context of the GUI. Each of the
icons within the Scoping Pane within the MMDB and CDB components support the graph
tools.
Student Notes
In addition to the graphs, Data Protector provides pre-configured reports to show the
physical size and usage of the internal database components. The Reporting context in the
Scoping Pane has several reports available under the Tasks menu tab.
The report shown above should be frequently executed to monitor the growth of the
database. The amount of maintenance needed for the database will largely depend upon the
rate of growth and length of time that the data needs to be kept.
Database Maintenance
Reasons:
• Low disk space
• File version purge
• Filenames purge
• DB size management
• DB corruption
Student Notes
There are several maintenance tasks required to keep the IDB running smoothly. They
include:
• Monitoring for devices with low disk space (where the database is stored)
• Ensuring that regular purging of obsolete data is occurring
• Periodically purging the file names tablespace (CDB)
• Monitoring for high client dynamics, and making adjustments to the backup
specifications
• Monitoring and managing the regular growth of the IDB
• Preventing corruption, or detecting it early before major problems occur
The rest of this section will deal with the commands that are provided with the cell manager
to manage, monitor and maintain the IDB. These maintenance commands will allow for
relocation and allocation of additional space for the DCBF.
Database Commands
Data Protector provides several commands for access to and control of the database.
There is a section at the end of this module containing an overview of the database related
commands and examples for their usage.
Database Cleanup
omnidb -purge …
omnidb -strip …
Student Notes
The GUI provides many database maintenance features and actions. Removal of session,
session messages, and session versions are provided. The functionality here behaves the
same as the omnidb command. See the omnidb man-page or the command reference section
lager in this module for details of the selected choices and some examples.
NOTE The GUI allows for the selection of multiple sessions as targets for the
operation.
Up to 2047 MB
per extension
Student Notes
When the size of the fnames.dat table is insufficient for the needs of the cell, the
administrator must extend it. This extension is typically done via the GUI as shown above,
but may also be performed from the command line. The omnidbutil command introduced in
the slide, is covered by a reference section near the end of this module.
Each extension to the fnames.dat may range from 1-2047 MB in size. It is generally
recommended to keep all of the extensions within the same directory structure, that is,
within the <OMNIVAR>/db40/datafiles/cdb directory.
Student Notes
It is desirable in larger cell environments to have more than a single directory used for the
storage of the DCBF. Data Protector allows for a maximum of 50 directories, but has an
initial limit set to 10 by a global option (covered earlier). The new DCBF may be created and
initialized via the Internal Database GUI, or by using the omnidbutil command. When using
the omnidbutil command, there are additional options for relocating the DCBF, as well as
removing the DCBF. Refer to the omnidutil reference near the end of this module for
examples.
Considerations:
Considerations:
•• Distribution
Distribution
•• Transaction
Transaction logging
logging
•• Backup
Backup
Student Notes
Internal Database recovery may be necessary if omnidbcheck reports critical or major
corruption to some parts of the database. Preparation is necessary if you are to recover the
previous non-corrupted database.
• Restoring the database without the server processes running (crs, rds)
The choice of which type of recovery is needed depends upon preparation and the report
output from the omnidbcheck command. Corruption reported as Major or Critical will
require some form of recovery. Corruption that is Minor allows for removal of the corrupted
data, and then continued operations; recovery of the data is optional and in some cases
unnecessary.
•• The
TheOMNIDB
OMNIDBobject
objectmay
maybe
beincluded
includedininaabackup
backup
specification
specification
•• The
Theconfiguration
configurationand
anddatabase
databaseare
areboth
bothincluded
included
ininthe OMNIDB object
the OMNIDB object
•• AA“hot
“hotbackup”
backup”isisperformed
performed
•• Consistency
Consistencycheck
checkperformed
performedbefore
beforebackup
backup
(default=on)
(default=on)
Student Notes
The database is an extremely important part of Data Protector, and must be backed up
regularly. A special object type of OMNIDB is provided, and must be used in order to obtain a
consistent backup. A normal file system backup is not sufficient!
Only one database backup can run at a time. During the Internal Database backup, Data
Protector also performs the following:
• Checks the integrity of the database before backup, thus preventing back up and
restoration of a corrupted database. (-quick enabled by default, this takes about 1.5 hours
for a 10 GB database)
• Online backup while the database is being used. Therefore, other backup or restore
sessions can run while the database is being backed up.
• Backup all Data Protector configuration data that is stored in the database and flat files,
including data on devices, backup specifications, and schedules.
1. Create a separate backup specification for the database. This simplifies scheduling and
restoring in case of a disk crash.
2. Schedule a database backup every night. This ensures that you always have an up-to-date
backup of the database. You can set the data and catalog protections to only a few days.
3. Make the database backup using a separate media pool on a separate media, on a specific
device. Make sure you know which media you use for a database backup. This greatly
simplifies eventual restore, since you know precisely on which medium your database is
backed up. (use of standalone devices is preferred in case of a disaster; they are easier to
configure for the restore process)
Data Protector by default checks the integrity of the database before the database is backed
up.
Modifying the object properties can disable the automatic database check that is performed
prior to backup of an OMNIDB object. It may be necessary to disable the automatic check if
the database is very large and the check takes too much time. In this case, a manual check
scheduled with the system scheduler may be a better option. The scheduled check should not
be performed while there is any other Data Protector activity.
Data
base Internal Database
MMDB
Student Notes
The restore of the Internal Database must be accomplished in a series of steps, because it is
not possible to do a "hot-restore" of the data with this method. All restore operations in Data
Protector require that Data Protector is operational; that is to say, Data Protector is the only
product that will read the Data Protector tapes.
It is possible that the restore of the Cell Database may be onto a different system than the
original Cell Server (Manager) system; in this case additional steps are needed in order to
proceed with the restore.
• Corruption
• Recovery due to loss of data
• Restore of failed Cell Server
Database Restore
The process below will assume a recovery onto a newly installed Cell Manager System or
after a disaster recovery of the Operating System. This procedure may also be used to restore
the database to a previously non-corrupted state.
Requirements
Overview:
Depending upon the condition of the Cell Server and Database, the Database may need to be
re-initialized so that it is operational prior to the restore. Use the omnidbinit command to
initialize the database if necessary; this requires that the Database Server (rds) is running.
If the rds is unable to be started due to database corruption, see the next topic,
"Reconfiguring a Corrupted DB."
Procedure
1. Import the tape into the existing (new) IDB, into a Media Pool using a Logical Device.
(This is not needed if the database is still operational and contains the session
information from the desired backup session.) This may require the configuration of a
new Logical Device if a new database was created. Consult the media log for the medium
ID if it is not known.
2. Restore the desired backup session data onto the system in an alternate location using
the "into" feature of Restore. (you may be able to restore into the partition or directory
where you have located the “db40”, since you will likely have available disk space there,
just don't overwrite the existing active database, "db40" directory)
3. After the "restore - into" has completed, stop the Data Protector servers. Be sure to stop
all GUI's and sessions before proceeding, the database will be moved! On the slide, this is
indicated as the crs, mmd, rds not running. Do not relocate the database with the servers
running!
omnisv -stop
4. Move/rename the current database to a temporary name, then move the restored
database into place.
mv /var/opt/omni/db40 /var/opt/omni/db40.bkup
mv <location>/db40 /var/opt/omni/db40
(where <location> is the full path to the restored database directory (eg. HP-UX))
5. The restore process also restored the configuration files into the same location as the
database files. You may want to move them into place as well if they need to be recovered
(this step may be optional, if the files are intact).
mv /etc/opt/omni /etc/opt/omni.bkup
mv <location>/omni /etc/opt/omni
6. Start the Data Protector Servers using the newly recovered database.
omnisv -start
7. Verify that the database and all of the configurations are operational.
omnidbcheck …
Student Notes
This essentially the same as the previous manual procedure, but only takes into account the
actual restore process, and not the activation of the restored database. After using this task,
follow the manual steps outlined previously.
Automated Restore/Recovery
Process Overview
• Stop daemons/services
• Read recovery information
• Restore database session
• Log replay (roll forward)
• Start daemons/services
omnidbrestore
omnidbrestore–autorecover
–autorecover[[ …
… options]
options]
Student Notes
The previous recovery sequence was required if the entire database is replaced, and the cell
server processes are operational.
In the case of the cell server being inoperable, it is still possible to restore the database in
place by using the command omnidbrestore. This is the preferred choice for database
recovery, but some key preparation is required.
Preparations must be made in advance if this type of restore is to be successful. Follow the
preparations mentioned earlier in this section; especially the configuration of a second copy
for the obrindex.dat file.
In this mode the obrindex.dat file is scanned for the media, RMA, and VRDA options
needed, as well as the name for the transaction log from the last IDB backup which is used
for the restore. When the options are retrieved, the database is automatically retrieved from
the last backup tape using the same physical device that the backup was executed with.
The read mode reads from a file created with the –autorecover –save <file> options.
The obrindex.dat file must be available, but requires changes. This is mostly used when
the original device for the backup is not available for the restore, or attached to a different
system. In this case the <file> may be manually updated with the appropriate restore device
information, and then used instead of the obrindex.dat file.
Manual mode:
This manual mode is used when the obrindex.dat file is not available. All of the options
needed for a restore must be specified manually. See the man page for omnidbrestore for
all the details. Good preparation should help to avoid this more difficult type of restore.
newconfig!
Internal Database
CDB MMDB
The
Thedatabase
databasemust
mustbebeoperational
operationalininorder
order
to
to manually restore the database fromtape
manually restore the database from tape
Student Notes
The corruption of the Internal Database is rare, but it is comprised of files and directories,
and is stored within a filesystem. There may be some circumstances where the Database
Server (RDS) is unable to start due to a corrupted database. When this occurs you are likely
to need to restore the Internal Database using the procedure previously discussed.
Each of the three mentioned above is recoverable, with the proper preparations.
Critical
To recover from critical level corruption, the IDB will need to be recovered from tape using
either of the previous procedures for recovery.
Major
To recover from major level corruption, the IDB may be recovered from tape (preferred) or
restored by exporting and importing the current database without the details. The
omnidbutil –writedb –no_detail, followed by an omnidbutil –read_db will
recreate the database without the file detail catalogs. The database will appear as to have
been created with backups using the “No Log” option. Once the omnidbutil –read_db
completes, the contents of the DCBF may be removed, as they are no longer referenced. This
recovery operation will take approximately 5-20 minutes.
All new backups may use the “Log All” option to create new DCBF entries.
Minor
To recover from minor corruption, removal of the files in the DCBF may be performed. Then
recreate the DCBF files by importing the media that was deleted. If these files are missing,
then restore will report errors when browsing.
During the installation of Data Protector, a copy of the database directory structure was
placed within the <OMNIHOME> directory called “newconfig.” The newconfig/db40
directory is a new, uninitialized Internal Database. To use it, copy it into the correct location,
and use the omnidbinit command to initialize it. Then start the manual recovery process as
indicated earlier.
1. Stop the Data Protector Servers (be sure to exit all Data Protector GUI's and sessions):
/opt/omni/sbin/omnisv -stop
2. Check the state of the Data Protector product filesets, they should be configured:
swconfig -u DATA-PROTECTOR
rm -r /var/opt/omni/db
swconfig DATA-PROTECTOR
/opt/omni/sbin/omnisv.sh -start
9. Follow the restore procedure discussed earlier to restore the Internal Database from the
latest backup.
NOTE This procedure may be used to create a new, empty database that may quickly
be enabled to allow for a restore of the database from tape.
This section will provide some brief examples so that the basic format of data stored in the
database will be understood. This module is primarily about maintenance, which may be
performed on individual items within the database.
omnidb -session
Access : Private
Number of warnings : 0
Number of errors : 0
Device name : dlt_drive1
-change_catprotection Protection
Changes the current protection of the catalog retention time. Protection can be
none, same_as_data_protection, until a specific date, or for a time interval.
same_as_data_protection means that the catalog will stay until data is
overwritten or exported. When the protection is until a specified date or for a time
interval, you must specify the value. The date form is [YY]YY/MM/DD. In the first
case, the value is the date until which the data is protected. In the second case, the
time interval is the number of days (after today) during which the data cannot be
overwritten.
You can mark detailed data for removing using the omnidb -strip command, with various
options; there are four possibilities:
• strip the detail catalogs for an object older than a specific number of days
Purging Sessions
The omnidb command may also be used to purge an entire session from the database.
All objects within a session will be marked as unprotected. It will still be possible to restore
from this session until the media used is overwritten; at that time the session data will
become obsolete.
Example:
Purging removes unneeded information from the database and frees space for new
information. Purging the CDB does not actually shrink the size of the files; it merely makes
space for new information.
The data previously selected (marked) for removal by omnidb will be permanently removed
from the database by this purging process.
The following command can purge information relating to restore sessions, as well as
obsolete (overwritten) backup sessions, and sessions without any media, such as failed or
aborted sessions.
Data Protector records detailed data, such as the names of each file being backed up during
each session, in the CDB. You can create space for new data by removing this data from the
CDB. You can remove detailed data for a backed-up object in one specific backup session.
The restore of single files is slower when the data about the file is not in the database; this is
because Data Protector has to search for the file from the beginning of the media Data
Protector normally tracks file positions on tape in its database.
Purge Schedule
The default installation of Data Protector includes automatic database purging. A purge
session manager (ASM) will execute based upon a schedule. The default schedule calls for a
standard purge every day at 12:00 (noon). You may notice the ASM running if the monitor
context is available while it is running. The purge (maintenance schedule) is controlled by
the global options file parameters.
Reporting on Usage
Monitoring the size of the Internal Database if very important. The -info as well as the
-extendinfo options for omnidbutil report on table and record usage for the database.
In addition to table size, you should also verify that the disk space where the database is
located is not full; this would be fatal to the database server.
Extending fnames.dat
The omnidbutil command can create new extended fnames.dat files where new data is to
be stored. The maximum size of each new file extension is 2 GB, and the maximum size of
the data table is 32 GB. It is important to plan the location of database extension files
carefully.
Once you define the extended fnames.dat files, their size cannot be reduced.
Where pathname is the full path to the directory where the new database files will reside,
(typically <OMNIVAR>/db40/datafiles/cdb) and size is the size of the database
extension file in MB (range 1-2047 MB per extension). The directory for the extension must
already exist.
For example:
omnidbutil -extendfnames /var/opt/omni/db40/datafiles/cdb -maxsize 500
This command creates an additional database file in the same directory as the original
database, and extends the size by 500 MB, thus giving a full database size of 2.5 GB.
Data Protector creates a new, extended fnames.datN file in the specified directory, each
time you run the command. The database extension files are backed up as part of the
database backup and are restored with database recovery.
omnidbutil –extendinfo
omnidbutil –list_dcdirs
Additional options:
-maxsize <MB> The maximu size of the DCBF directory (limit
4096)
-maxfiles <n> The limit of files in the DCBF directory (limit
10000)
-spacelow <MB> The minimum amount of disk space needed to
use the DCBF directory (default 100)
-seq <n> Assign the usage sequence to the DCBF directory
Data Protector first determines which DCBF directories are still useable (those not already
exceeding any limits). The DCBF directory is considered full when any of the following is
true:
Based upon the algorithm selected, Data Protector chooses an appropriate directory to use.
1) omnisv –stop
2) <move the files to another DCBF directory>
3) omnisv -start
4) omnidbutil –remap_dcdir
5) omnidbutil –remove_dcdir <dir_name> (will fail if not empty)
To ensure the integrity of the database, before copying the database, be sure that backups,
restores, or GUIs are not running.
Use the following command to copy the database to a set of ASCII files. You must specify
which part of the database you want to copy, MMDB, CDB, or both. You will be prompted to
archive (copy) the DCBF and SMBF before omnidbutil removes exclusive access to the IDB.
mkdir /tmp/mmdb
mkdir /tmp/cdb
omnidbutil -writedb -mmdb /tmp/mmdb -cdb /tmp/cdb
NOTE Writing the database out in ASCII format requires approximately 20% more
disk space than the current database size.
Use the following command to read the database from an ASCII file.
omnidbutil -readdb [-mmdb <dir>] [-cdb <dir>] [-no_details]
By default, the device is set to the device that was used for the backup. While most
restore operations allow the device to be changed, database restores such as Oracle/SAP
provide no such mechanism. Therefore, this command must be used.
-change_cell_name [old_host]
This option changes the owner of the catalog database to the current cell server. It also
changes all references in the (central) media management database from old_host to
current cell server. If the old_host option is not specified, omnidbutil tries to get the
previous owner of the catalog database (old host) from the database itself. This command
should be used after moving databases from one cell server to another and after
-readdb of files that were created on another cell server. The old_host option must
be specified after -readdb of files that were created on another cell server.
-show_cell_name
Reports the owner of the catalog database.
-clear
This option is used to bring the Internal Database back to a consistent state. If something
happens in Data Protector that causes some processes to hang, like the backup session
manager (BSM), this process may block the Internal Database. In this case, the clear
option is used. This option kills any running Data Protector process, and frees up every
database resource the process is using. This is also useful after a restore of the database
from tape; a running session backing up the database may be cleared (failed) and then
removed. (see the section on database restore)
-free_pool_update
Transfer unprotected media from the backup pools to the free pool. This is the manual
version of the automatic de-allocation feature of media management. This is normally
executed during the day automatically, and is controlled by the global option:
“FreePoolDeallocFreq”.
NOTE For more details on the omnidbutil command, see the man page.
The directory structure for the database must exist before execution of omnidbinit.
omnidbinit {-force}
This command is used to initialize the database and erase all data in it. omnidbinit is
automatically executed during the first Data Protector installation. It can be used to reset
Data Protector to its default state. Be very careful!
-force
Data Protector normally prompts you to verify that you really want to delete all data in
the database. The force option suppresses this prompt. This command is useful if you
want to initialize the database from a script.
omnidbcheck
The omnidbcheck command checks the consistency of all or parts of the Internal Database.
For this, the command requires exclusive access to the database. If errors are detected, an
error report is sent to the standard output. Several different options exist to allow for full or
partial checking of certain sections of the database. Errors found during the check process
may require a repair or rebuild of the database. The omnidbcheck command should be run
frequently to ensure continued operations of the database. Early detection of problems is
very important.
The choice of which type of recovery is needed depends upon preparation and the report
output from the omnidbcheck command. Corruption reported as Major or Critical will
require some form of recovery. Corruption that is Minor allows for removal of the corrupted
data, and then continued operations; recovery of the data is optional and in some cases
unnecessary.
The command can be also used to display the contents of the DCBF and SMBF sections of
the database.
Even if the database is corrupt, it may still be possible to use it, but it is not wise to do so, as
more corruption may result.
The omnidbcheck creates a log file for each part of the database that is checked. The log
files will be written to <OMNIVAR>/log on the cell manager system. The file names created
will be the database section name with a “.txt” suffix. Each log file will contain details of
the check along with a timestamp of when the check was performed.
• Object versions
• Media positions
The database may be verified in parts, or as a whole. Some options for the omnidbcheck
command are: (see the man page for more options and details)
-quick
Check the core, CDB filenames, and the presence and size of DCBF
-extended
Full check of database excluding the SMBF (full check of the database consistency
including detail file information; this may take several hours on a large database)
2. The Data Protector internal database is comprised of two tablespaces. What are their
names?
•
•
3. List three types of data that are stored in the Data Protector internal database.
•
•
•
4. Name 5 parts of the Internal Database that are external to the tablespaces:
•
•
•
•
•
5. What does the term invalid mean when describing records in the database?
7. What would you need to add to your database to increase its capacity beyond the 2-GB
limit?
8. Which database files are likely to be the biggest in the database, and what information do
they hold?
9. By default, an automatic database purge takes place every day at midday. TRUE/FALSE?
•
•
12. Which option can be set in a backup to reduce the amount of detail information stored in
the database?
13. What command would you use to perform a thorough consistency check of the database?
•
•
•
15. What command can be used to display information about database size and utilization?
22. What happens to the archive logs when the database is backed up?
• Backup Specifications:
• Backup Schedules:
• Logical Devices:
• Media Pools:
Monitor
Monitor GUI
GUI
•• Monitor
Monitorcontext
context
•• Database
Databasecontext
context
•• OpenView
OpenView Cell
Database
Reporting
ReportingCommands
Commands
Java
Java Reporting
ReportingGUI
GUI
ARM/DSI
ARM/DSIIntegration
Integration
Report
Report Schedules
Schedules Events
Events
Student Notes
Data Protector provides a rich set of tools and features to enable the administrator to manage
the backup function effectively.
Monitoring and Reporting capabilities are available from the following Data Protector tools:
• Data Protector GUI (UNIX, Windows NT, OpenView)
Reporting capability allows generation of formatted reports against all aspects of cell
operation, for example, media used for the previous night’s backups, media available for use,
etc. These reports may be scheduled as report groups.
Notification capability enables the administrator to be alerted of a predefined event such, like
Mount Request or Device Error. Notifications are sent in various forms, such as email and
SNMP.
Service level reporting can be achieved by taking various metrics from Data Protector DSI
(Data Source Integration) and feeding this into MeasureWare. In addition, Data Protector
response times can be measured using Application Response Measurement (ARM).
Connect to
another cell
Student Notes
Monitoring the status of current sessions is accomplished by using the Data Protector
Monitor context. To start the Monitor only context use xomni –monitor (HP-UX) or
manager –monitor (Windows).
• Respond to mount requests for media for an active session (backup, restore)
Upon invocation, the monitor will display current active sessions.
To check the backups of last night, start the monitor to see if any sessions are still active,
perhaps in a mount_request state.
omnidb -rpt -wo 3600 3600 Show a summary of all of the sessions
that ran in the last hour (3600 seconds)
omnidb -session <ID> -media List media used by a session <ID>
omnistat Show currently running sessions
omnistat -previous -last 1 Show the sessions that ran in the last 1
day
omnistat -previous -until 2000/05/12 Show all previous sessions through May
12, 2000
omnistat –session <ID> -monitor "Attach" to a running session and
monitor it interactively
omnidb -session $(omnidb -rpt -latest |awk 'NR==1 {print }' ) –media
Optionally you may start the xomni -monitor or Manager.exe -monitor with the
-server <cell_manager> option to monitor a remote cell.
NOTE By default, all cell consoles will have read-only permission to other cells
through a remote cell monitor. You may use the User Manager to allow for
more remote control if desired.
Student Notes
To view the previous sessions, switch to the internal database context, then select sessions
to view all of the previous sessions.
In the case of a failed session, you may select the failed backup session, check the details on
it, find out what the problem was, and perhaps restart the failed backup objects.
Session Details
By double clicking on one particular session, you get into the session details monitor. Here
you can see an overview of all backed up objects and their size. You can check the used
media, and you can check how long the backup took.
All messages that occurred during a session are shown here as well. You can search for
messages or delete them from the Data Protector internal database to save space. Session
detail information should be deleted only when it is no longer required.
For each backup object, the full object description is provided, such as client system, mount
prompt, description, and its backup status. Furthermore, the size of each object is displayed
along with the number of errors or warnings that occurred during its backup. The object
backup status can be Completed, Failed or Running. The session status is a summary of
the status of all objects, plus the completion status of the pre/post exec command.
Reporting Possibilities
Student Notes
Shown above are some of the possibilities that Data Protector provides for collecting data
from the embedded database. Some of these capabilities require some configuration before
they may be used.
The interactive reporting command line and GUI are available for use right out of the box.
Data Protector also comes pre-configured with several events that trigger logging. You may
want to configure some additional triggers for reporting. The event driven reporting functions
are covered in the next module.
The configuration of a Report Group would allow for collections of reports to be executed in
a single action. This grouping is necessary for scheduling and event based execution.
Report Categories
• Backup specifications
• Configuration
• Internal database
• Pools and media
• Sessions in timeframe
• Single session
Student Notes
Data Protector provides a rich set of predefined detailed reports to provide all the typical
information that the Cell Administrator may need to assist with the day to day Data Protector
tasks: (These are all available via the Web Interface and Reporting context of the GUI)
Backup Specifications
• Trees in backup specifications
• Objects without backup
• Objects latest backup
• Objects average sizes
• Not configured file systems
• Backup specification information
• Backup specification schedule
Configuration
• Cell information
• Configured clients not used
• Configured devices not used
• Lookup schedule
• Clients not configured
• Licensing report
• Client backup report
Sessions in Timeframe
• List of backup specifications
• Session flow report
• Device flow report
• Report on used media
• Client statistics
• Backup statistics
• Backup errors
Single Session
• Single session report
• Session objects report
• Session per client report
• Session devices report
• Session media report
NOTE In an MOM environment, reports can include information from multiple cells.
Reporting Overview
Interactive Report or
Add Report Group and
Add Report to Group
Define Report Group
Schedule
Choose Report Content
Student Notes
Data Protector provides a large number of predefined reports. Reports can be generated
interactively via GUI or command line, in a selected format, such as ASCII, HTML, etc.
Reports can also be used within notifications, such as email, broadcast, etc., and can be
scheduled to provide regular information.
Reporting Tools
Data Protector provides the following mechanisms for defining and running reports:
• Data Protector GUI (reporting context)
• omnirpt command
• Web Interface
• Data Protector commands (omnidb, omnistat, omnicellinfo, omnimm, omnidbutil)
Report Groups
Reports can be run interactively or can be placed in Report Groups to allow multiple reports
to be collated together to provide more useful information. A report must also reside in a
report group if it is to be scheduled. Data Protector allows schedules for the group of reports,
not the individual reports.
Report Formats
To generate a report, a database session manager (DBSM) is started. It provides the
requested information. The report can be created in four different formats:
• ASCII
A report is generated in plain text format.
• Short
A report is generated in plain text format, but in a summary form, showing the most
important information. This is a suggested format for broadcast messages.
• HTML
A report is generated in HTML format, which is useful when using a Web browser. For
example, users could check to see if their systems were backed up by following a link on
the Intranet site. (reports would have to be scheduled, and their data saved as HTML files
(documents) as content for a web server.)
• Tab
A report is generated with fields separated by tabs. This format is useful to import the
reports into some other applications for further analysis, such as a spreadsheet program.
Delivery Methods
Reports may be delivered using the following methods:
Broadcast Allows for pop-up window within the Microsoft Windows environment.
Email Sends the report as Email, requires a mail sending capability to be available on
the Cell Manager.
External Executes a program external to the Data Protector product. The report data is
sent to this executable as command line parameters.
SNMP Sends the report data to an SNMP manager, such as OpenView NNM or OVO.
Reporting GUI
Student Notes
The Data Protector GUI provides an easy to use method of generating and viewing reports
online. The GUI provides mini-Wizards to guide you through report generation and
scheduling.
The Data Protector GUI can be used to define, generate, and schedule reports on both
Windows and HP-UX cell servers.
Data Protector reports can be viewed individually, or grouped into report groups.
Some reports can be used only in a report group and are not available as interactive reports,
for example, a mount request. Documentation for all of the supported reports may be found
on the man-page for omnirpt. (see the next page)
2. Click the Tasks tab at the bottom of the Scoping pane to switch to the tasks wizards.
3. Browse the provided reports and select the one you want to view.
Student Notes
Data Protector provides a Java applet Web-based online reporting capability that lets you
configure, run, and print all the Data Protector built-in reports of the omnirpt command
interactively. During reporting operations, Data Protector's Java applet directly accesses the
Cell Manager to retrieve current data.
The Java reporting interface is installed as a component of the cell console, which means that
it is available on any client system that supports the cell console user interface.
The Java applet requires a web browser (Netscape Navigator 4.x, Microsoft Internet Explorer
4.x, with JDK 1.1 or later support).
Not only can you use the Java reporting facility to get online access to your reports, you can
also configure your reporting structure through it, such as adding new reports to a report
group and changing a report's parameters.
The Java interface is started from a web browser by opening the Java GUI file
webreporting.html located in /opt/omni/java/bin on Unix, or
<OMNIHOME>\java\bin on Windows.
URL: file:/opt/omni/java/bin/webreporting.html
Address: C:\Program Files\Data Protector\java\bin\webreporting.html
To generate reports using the Data Protector Web reporting interface, you must access the
Data Protector Web reporting interface. Once you are logged onto the Cell Manager via the
web-reporting tool, you can generate various types of reports in a fashion similar to the Data
Protector GUI. (the reports are the same, the presentation is HTML)
To view a report, click the report and provide the needed information. When the report is
displayed, you can print it or save it. When the report is saved, the definition to collect the
data is stored, not the report data. When you save the report, you must add this report to an
existing or a new report group. The Web reporting interface does allow for the creation of
report groups, a new group is created when saving a report to a new group. The report
groups are saved on the cell manager in the <OMNICONFIG> directory.
Reporting Command
Student Notes
The omnirpt command is used to generate reports. For a detailed description of the
command, see the omnirpt man page.
or
Generate a report on media used in the last 24 hours in tabulated format, saved in a file:
omnirpt -report used_media –timeframe 24 24 –tab –log data.txt
NOTE If you have an HP-UX cell manager, you may want to take advantage of the
Netscape Fasttrack Server that is available from the applications media.
Report Groups
Student Notes
The report group defines a collection of reports that will be executed together. The report
group, unlike individual reports, may be scheduled. Report groups may also be triggered by
the event notification system.
To create a report group, switch the GUI to the reporting context, highlight reports, and using
the pop-up menu (right mouse button) select Add Report Group… .You will be prompted for
a report group name; names without spaces and special characters are acceptable.
After the report group is created, you may assign (save) report definitions to the group, and
then schedule the report group.
Student Notes
Once reports have been defined and grouped into report groups, they can be scheduled to
generate reports on a regular basis.
The omnitrig process initiates scheduled reports in the same way as backups.
The easiest way to schedule a report is to use the Data Protector GUI as shown in the slide
above. Alternately, the schedule file can be created manually. A few examples of schedule
files are shown below.
Example 1: every Saturday at 8:00 PM, beginning May 20, 2000:
-start -starting 20 5 2000 -every
-day Sat -month May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr
-at 20:00
Student Notes
The report group is used to create a report collection that may be scheduled and executed
together. The report group has the concept of a folder, or container for the report definitions.
After the report group is created, report definitions may be added to it to form the collection.
Data Protector allows for several reports to be added to a single report group, as well as
several report groups to be created. Once defined, the properties of the report group and
reports within the group may be modified.
Report groups are saved as ASCII files in the <OMNICONFIG>\rptgroups directory on the
cell manager.
Report Groups
A report group can contain multiple individual reports that will be executed together via the
execution of the report group.
Normally, a user will require information that is not available from any single report type. In
some cases the requirements must be satisfied by a combination of reports. For example, a
report of the previous night's sessions and details of the media used. A report group would
satisfy the multi-report requirements. Report groups must also be defined if reports are to be
launched regularly via the scheduler.
The whole report group can also be scheduled and delivered by the notification service.
Report groups can be defined by all three reporting mechanisms (Data Protector GUI,
command line, and the Web interface).
<OMNICONFIG>/rptgroups/Manage_Media
NAME "Manage_Media"
{ comments
REPORT "show-all-media" report name
{
ID "media_list" type of report
POOL "Default File" pool name
STATUS "Good" media status
LOG report method, log file
{
TYPE ASCII file type
TO file location
"/var/opt/omni/log/media_rpt.log"
}
}
}
TIP To create a report group from the Web interface, you must choose to save a
report; one of the choices for saving is to a report group. If the report group
does not yet exist, it will be created. After the report group is created it will
appear at the bottom of the reports list, under notification; from there it may
be executed.
See the omnirpt man page for more details on report ID and options
for each type of report.
operational level
presentation
Data
Protector OV-R
16
16
crystal
crystal
Omnirpt for skipped files
reports
reports
service level
DP-OVO SPI DP-SIP
presentation
Cfg
Cfgfiles
files
SIP scripts
scripts
11
Events
snmp traps crystal
crystal
event collection report
report
and consolidation
Cfg problem
Cfgfiles
files management
OVO SLA
OV-SN Service Mapping Cfg
Cfgfiles
files
SD scripts
scripts
Student Notes
Data Protector, out of the box has monitoring, notification, and reporting tools to document
backup and recovery operations. Data Protector 5.0 contains several integration possibilities
to enable Service Level Management (SLM). Data Protector extends its service centric
approach to SLM through leverage of several OpenView products. The SLM integrations with
other OpenView service management products consolidates service views, service
performance data and other capabilities into one console, giving a service provider better
information and insight into the overall IT service delivery.
With HP OpenView Data Protector 5.0 there are four new Service Management Integrations
introduced which aggregate data and reduces complexity in a large scale, global data center.
• DP 5.0 - DP-SPI - OVO - OVSD (Service Desk 4.0) for problem management
• DP 5.0 - DP-SPI - OVO - OVR (Reporter 3.0) for operational level information presentation
• DP 5.0 - DP-SPI - OVO - OVSN - OVSIP (Service Information Portal 3.0) for service level
information presentation
• DP 5.0 - OVSIP (Service Information Portal 3.0) for service level information presentation
Enterprise IT departments are increasingly using service management tools, techniques, and
methods to set service level expectations, measure service delivery against those
expectations, and to justify future service expansion. In short, the IT department now is run
like a business.
Part of IT’s business is managing the risk of data loss. Threats ranging from user error, to
viruses or other unauthorized data access and modification, or to the occasional failure of the
storage device itself put data at risk twenty four hours a day. Business critical data loss can
cost the enterprise thousands, even millions of dollars per hour of downtime. While all data is
at risk, not all data justify equal recoverability. IT department must protect the business
critical data to a higher level of protection than the less valuable data, and do so cost
effectively.
Service providers use Service Level Agreements (SLAs) to document the provider-customer
contractual expectations. SLAs typically establish availability and performance objectives.
Using this model, a provider can offer multiple service levels each at its own cost structure.
By identifying the relative value of data placed within its care, IT department can set service
expectations on backup and recovery consistent with the protected data’s business value.
Backup and recovery now is managed like the enterprise itself: that is, like a business.
Demonstrating SLA compliance requires constant monitoring and periodic reporting to show
whether SLA expectations have been met.
• HP OVO – a software solution designed to help service providers and their system
administrators detect, solve, and prevent problems occurring in networks, systems, and
applications in any enterprise. It is a central management point for various remote
OpenView applications. Collects and analyzes data, automates critical response, as well
as message forwarding to other services.
• OVR (OpenView Reporter) - a reporting service that further analyzes, inspects, and
collects data gathered by OVO and formats them into a human readable and usable web-
based presentation.
• DP – OVR integration: integration of DP 5.0 with HP OVO, OVSN, OVP Agent and OV
Reporter
The integration of DP 5.0 with HP OVO is extended by adding HP OpenView Reporter
(OVR 3.0 English version). With OVR service providers can generate reports from data
obtained from the OVO management server.
An IT Service Provider can use these reports to demonstrate to a customer its SLA
compliance. For example, “DP Transaction Performance” Report consists of the service
performance metrics (one of the IT SLA parameters).
Java
Notifications Tabular Reporting
Events/Scheduled
Reports Reporting
ASCII Reporting
Web Reporting
Student Notes
What Is ARM?
The ARM API is an emerging standard for measuring end-to-end response times of
transactions in distributed environments. Application programs that use the ARM API act as
sources of response time information (and also user supplied information that may be
relevant to a particular transaction) for ARM compliant system management and monitoring
tools, like HP MeasureWare.
HP MeasureWare will log ARM transaction information in its repository for subsequent
analysis and reporting. It can also raise real time alerts (or alarms) when the elapsed time of
a specific transaction, such as a backup operation, exceeds a predefined threshold.
When a real time alert is raised, a number of actions are possible, including but not limited to,
informing a central operations console, such as HP OpenView IT/Operations, paging a
system operator, or taking automated remedial action to resolve the problem.
As Data Protector is already ARM equipped, it is simple to integrate Data Protector with an
application like HP OpenView Performance Agent, formerly called MeasureWare, which
supports the ARM API. On the Windows platform, this is completely automatic. If Data
Protector is installed on a system where MeasureWare is already present or vice versa, the
transaction data will immediately show up in MeasureWare and PerfView. On HP-UX, the
only required task is to create a link from a MeasureWare library to an Data Protector
directory. See the HP OpenView Data Protector Administrator’s Guide for more
information.
The ARM integration is provided only on Data Protector cell managers. In order to use it,
install the ARM 2.0-compliant library to the specific system.
Measuring
The following information can be measured with the ARM integration:
• Overall session duration
• Disk agent read times
• Disk agent network write times
• Media agent network read times
• Media agent data write times
• Session manager write to database time.
• Database purge duration
What Is DSI?
The Data Source Integration (DSI) allows you to use the HP OpenView Performance Agent
(MeasureWare) to log data, define alarms, and access metrics from sources of data other
than the metrics logged by the MeasureWare Agents scopeux collector. Data Protector
provides a sample script and configuration file that shows you how to use the Data Protector
reporting command line interface with Data Source Integration to log data about the Data
Protector environment and backup and restore sessions.
Data Protector provides a sample Korn shell (ksh) script and class specification file that, by
default, log two metrics: the number of clients in cell and the size of Data Protector internal
database size. The script and class specification file can be easily modified for collecting
other information from Data Protector. The scripts are supported on UNIX systems.
Select what data you want to log. Data Protector provides a reporting command,
omnirpt, located in the /opt/omni/bin/ directory. This command can be used to
gather information about the Data Protector environment. See the omnirpt man page
for more information on the command.
2. Second, write a script that loops querying the selected data and writes it to standard
output.
5. Configure perflbd.rc.
Before you start modifying the perflbd.rc file, stop the mwa services, with the
following command:
/opt/perf/bin/mwa stop
Now you can edit the file /var/opt/perf/perflbd.rc. If you are configuring Data
Protector sample metrics, add the following line to the file. It must be added as a single
line:
Here are some examples of what you can do with the data that Data Protector provides:
• Real time alerting of backup or restore sessions that exceed the specified time window
(MeasureWare).
• Forecasting of the Data Protector database growth to be able to spot points in time when
certain limits will be reached (PerfView Planner).
• Regular email reports to backup operators, end users, and the management (Data
Protector built-in reporting with the capability to send emails).
• Backup reports written to a web server to make them available on an on-demand basis
(built-in Data Protector reporting with the capability to write HTML).
• Send major and critical Data Protector events to your network management solution,
such as HP OpenView Network Node Manager (Data Protector built-in notification engine
sending SNMP traps).
1. Using the Data Protector GUI, how do you determine if last night's backups succeeded?
2. Using the command line interface, how do you determine if last nights backups
succeeded?
3. Using the command line interface, how do you determine if there are sessions currently
running?
6. Using the command line interface, how do you respond to a mount request?
•
2. What is a report group?
•
•
Monitor
Monitor GUI
GUI
•• Monitor
Monitorcontext
context
•• Database
Databasecontext
context
•• OpenView
OpenView Cell
Database
Reporting
ReportingCommands
Commands
Java
JavaReporting
ReportingGUI
GUI
Report
Report Schedules
Schedules Events
Events
Student Notes
As discussed in the Monitoring and Reporting module, Data Protector provides several
possibilities for retrieving stored data from the Internal Database.
Events occur frequently within the Data Protector cell, and are logged into file and presented
to the administrator within the GUI. While this capability is very necessary, the event driven
notifications can take automation to another level. Data Protector is able to have automated
handling of events that cause reporting or notification to occur, in addition to logging events
to a file.
Notifications Concept
Student Notes
Data Protector provides an event driven notification system. There are two main types of
notifications:
• Event triggered (a result of a run-time event occurrence)
Maintenance Time
Maintenance time is normally schedule for 00:00 (midnight). The global option
“DailyMaintenanceTime” is configurable to cause these checks to be done at another time.
Example 1
Every morning at 7:00, a report about all backup sessions in the last 24 hours is created and
sent by email in ASCII format to the backup administrator's mailbox. Additionally, the same
report is written to a file on your web server in HTML format, so that others can also access
this information.
Example 2
In case of a device failure or a mount request, a Windows NT broadcast message is
immediately sent to the backup administrator's Windows NT workstation and an external
command is triggered, which activates the backup administrator's pager.
Example 3:
Data Protector executes a media report that gets fed to an external script. The media report
lists all of the tapes used for the evening backups. The external script send eject commands
to the tape library to eject the tapes so they may be taken off-site.
(See the Data Protector Administrators Guide Appendix-A, for a worked example of
automated media eject from a tape library.)
The following events can be trapped and notifications generated which alert interested
parties so that remedial action can be taken.
Notification Events Description
Alarm Critical messages as a result of internal Data Protector
conditions, or errors.
Backup Error The termination status of a backup session is completed with
errors.
Database Corrupted The result of Data Protector detecting a problem while trying
to add or retrieve data.
Device Error A failure of Data Protector to be able to write to a device.
End of Session Occurs when every backup session completes.
Mail Slots Full Occurs when ejecting media from a tape library, some libraries
have more than one repository position to be used as the mail
slot(s).
Mount Request A backup or restore session is running and requesting a tape in
order to continue.
Database Purge Needed Evaluates the configured parameters (days since last purge,
number of filenames, estimated purge time) to determine if a
data purge is needed.
Database Space Low Evaluates the configured threshold parameters (maximum size
of CDB, free disk space, maximum size of DCBF) to avert a
disk full condition.
Health Check Failed Check if Data Protector services are running, the media
database is consistent, and if one backup of the OBDB exists.
License Will Expire Evaluates the configured parameter (number of days) to warn
of license expiration.
Not Enough Free Media The free pool media count has dipped below the configured
threshold.
Unexpected Events The number of events in the event log has exceed the
configured threshold.
User Check Failed Execution of a customized (admin defined) script to perform
some checking has failed.
Notification Send Methods
• Broadcast
Broadcast message notifications allow you to send a broadcast message with desired
information to specified systems when a specified event occurs. Broadcast messages can
be sent to Windows systems only, and are received as pop-up windows on the system
display even if no one is logged in. (This is not supported on UNIX systems)
You must specify the system to which the broadcast message should be sent. Broadcast
messages are limited in length, so the short format is preferred. The reports are limited to
1000 characters.
• Email
Email notifications allow you to receive email with desired information when a specified
event occurs.
If you are sending email notifications from a Windows system with Microsoft Exchange,
you must create an Data Protector Exchange profile, called Data Protector, on a
system that will be sending email (usually the Data Protector Cell Manager). On UNIX
systems, no additional configuration is needed.
• External
External script notification allows you to process the output of the report in your own
script. The script receives the report output as standard input (STDIN). The
recommended format for script processing is the tab format.
You must choose, and specify, the full path name of the script that is located on the cell
manager system which is to process the report data.
• Logfile
Log to file notifications allow you to post a log file with desired information when a
specified event occurs.
The log file is posted on the cell manager system. You specify the name of the file to
which you want to post the report.
• Report Group
Report group notifications allow you to execute all of the reports from a report group
when a specified event occurs.
• SNMP
SNMP trap notifications allow you to send an SNMP trap with desired information when a
specified event occurs. The SNMP trap can be further processed by applications using
SNMP traps, such as the HP OpenView Operations. An OV_APPL_ALLERT SNMP trap is
generated, which includes the report in variable $6.
On a UNIX cell manager, SNMP traps are sent to the systems configured in the
notification. On a Windows NT cell manager, SNMP traps are sent to the systems
configured in the Windows SNMP traps configuration.
On a Windows NT Cell Manager, SNMP traps are sent to the systems configured in the
Windows SNMP traps configuration.
NOTE The OVdests file does not exist by default in all versions of Data Protector, it
may need to be created.
The OVdests will contain the name of the trap recipient, such as:
The OVfilter identifies the minimum message levels as Normal, Warning, Minor, Major,
Critical. Enter the list of messages levels that would be sent as SNMP traps.
SNMP trap configuration allows for all session messages to be sent to a trap destination.
While this may be useful, the trap recipient may easily become overwhelmed with the volume
of messages received through this mechanism. It is strongly suggested that some filtering be
put in place at the receiver to control the volume of messages that Data Protector may
potentially send, or configure the minimum message level and allow Data Protector to send
only Major or Critical messages.
Student Notes
Data Protector tracks several different cell events and stores them in an event log. The event
log is viewable as a text file, or by using the Reporting context of the GUI and selecting Event
Log.
The event log is stored in readable form in the <OMNIVAR>log directory as Ob2EventLog.txt.
The events stored here are the result of the Data Protector event notification system. Events
that occur as a result of Data Protector monitoring, or executing jobs may trigger event
processing.
Data Protector provides several events types that may be configured. By default most of the
event types simply log to the event log files.
Periodic purging of this event log is suggested, as this file will grow without limits. To remove
entries from the GUI, select Event Log, and use the pop-up menu item “Empty Event Log.”
The items in the log will not be removed, but the GUI will no longer display the “old” entries.
To view the old events, simply display the text file “Ob2EventLog.txt.
In addition to the Data Protector Event Log, the events that occur for the cell server process,
crs may also be logged separately. The global option that controls this behavior is called
“LogCrsEvents” and is set to 0 (zero) by default, which means off. If the administrator
enables this additional logging, then manual purging of the log file over time is necessary. The
log file where these events will be stored is called <OMNIVAR>/log/crsevents.log.
On Windows systems additional logging may be sent to the system event log. To enable this
use the global parameter “EventLogMessages”. The default value is set to 0 (zero) which
means off. The events that may be logged as a result of this feature are:
These events are only logged on the Cell Manager system (where the CRS service is running).
Default Notifications
Event names
Student Notes
Each of the default notification send alerts to the Data Protector Event Log. Many of the
notifications send their alerts based upon pre-configured thresholds. The thresholds and
parameters may be viewed using the GUI shown above.
You may create additional notifications that use all of the pre-configured events; your
customized notification may use any of the methods shown previously, such as E-mail,
broadcast, etc.
Notifications are stored in an easy to edit file. The event configurations may be altered by
using a text editor as well as via the GUI. The format for the configurations is shown on the
next page.
Student Notes
The Data Protector Java based Web GUI may be used to add, delete or view the notifications
that are configured into the cell. The native GUI must be used to edit notifications, this is not
performed within the Java GUI.
Notification Format
Student Notes
Defining Notifications
Notifications can be defined using the following methods:
• Data Protector GUI
• Java Web Reporting
• Manually by modifying the Notifications file
The main structure for the Notifications file is outlined on the slide above.
The following examples may be referenced if you would like to manually add to the
Notifications file on the cell manager:
NOTIFICATION "Notify_1"
{
-event "EndOfSession"
-object "*" (could be a datalist name instead of *)
-log "/var/opt/omni/log/end_of_session.log"
}
NOTIFICATION "Notify_2"
{
-event "DeviceError"
-object "*" (could be a logical device name)
-external "/opt/paging/page_admin.sh"
}
NOTIFICATION "Failed_Device:DLT-1"
{
-event "DeviceError"
-object "DLT-1" (send notice for logical device DLT-1)
-email "root@r207w100"
}
NOTIFICATION "Notify_3"
{
-event "DBSpaceLow"
-object "*"
-email "root@na168w2.nap.edunet.hp.com"
}
Example 4: Notify root@cell_server via email the database space is low by executing a
report group that performs that action.
NOTIFICATION "Notify_4"
{
-event "DBSpaceLow"
-object "*"
-report "Database"
NOTIFICATION "Session_End"
{
-event "EndOfSession"
-object "w100_gen"
-external "/opt/omni/lbin/manage_media.sh"
}
The following is the configuration for the report group called Database that will be executed
according to Notify_4 above. The report group was created using the Windows NT GUI.
NAME "Database"
{
REPORT "Size_Report"
{
ID "db_size" execute the database size report
MAIL action for the report is mail
{
TYPE ASCII format is ascii
TO
"root@na168w2" recipient is root at <host>
}
}
}
NOTIFICATION "Notify_5"
{
-event "MountRequest"
-object "*" (could be a logical device name)
-external "/opt/omni/lbin/Mount_notify.sh"
}
NOTE The notification above will execute in addition to any configure Mount Script
that is associated with the device. In this manner you have multiple
notification methods for a single mount request event.
Scheduled Event
omnitrig
• Broadcast
dbsm invokes omnirpt to • Email
Delivery
• External
generate report Method
• Logfile
• SNMP
Notification Event
Session Manager
Student Notes
This is done by first defining a report group and reports, then defining a notification. The
send method for the notification can be set to Report Group. When the particular event
occurs, the DBSM invokes omnirpt to generate the report, and deliver it by the defined
delivery mechanism, such as E-mail.
This form of report generation can be very useful. For example, when the end of a backup
session condition occurs, generate a report of the media used and email it to the operator, so
he or she can remove the tapes from the library. This concept could be extended to automate
the eject of media from a library (to the mail-slots) based upon a daily report connected to an
event.
6. More than one notification can be configured for the same event type. TRUE/FALSE?
•• Local
Local user
useraccess
accessto tothe
thecell
cell console
console
•• Remote
Remotecell
cell console
console access
access toto the
the cell
cellmanager
manager
•• Access
Access to
to the
the web
web reporting
reportinginterface
interface
•• Remote
Remote control
control of
of non-cell
non-cell clients
clientsforfor restore
restore
•• Remote
Remoteaccess
accessthrough
through aafirewall
firewall
Student Notes
Out of the box, Data Protector is both secure and un-secure from different perspectives. By
default, the only user that is able to operate the cell console and use the command line
interface is the root/Administrator user logged into the cell manager system. On the other
hand, the disk agent is configured to respond to “any” session manager requesting to restore
data. What appears to be a huge security hole is easily closed, and a design feature of the
product.
This section of the course will address all of the topics listed above, including the support for
access through a firewall.
Access Control
Student Notes
Access to Data Protector’s functional areas, such as Client Installation, Device
Configuration, Backup, and Restore, is strictly controlled by the allocation of specific
permissions to Data Protector User Groups.
Specific operating system users, such as root, Administrator, Oracle, etc. may be
configured as members of a Data Protector User Group.
The Data Protector operations that the users are able to perform depend on the capabilities
assigned to the User Group to which they belong. Users may be assigned to more than one
group, although this is not very common.
User Groups
Data
Data Protector
Protector provides
provides three
three default
defaultuser
usergroups.
groups.
New groups can be added by the administrator.
New groups can be added by the administrator.
Student Notes
A Data Protector user group is a set of access rights that permit execution of certain portions
of Data Protector functionality.
Data Protector provides three default user groups that provide the typical level of delegation
and control required by most customers:
• Admin
• Operator
• User
Student Notes
The Admin Group is all-powerful. Members of this group have complete control of all Data
Protector operations. Accordingly, the Admin group cannot be modified in any way, as it
must always have full-control.
When Data Protector is installed, the root/Administrator user of the Cell Manager
system is automatically added to the Admin group.
If you require other users to have full control of the Data Protector Cell, they must be added
to the Admin group.
•• Complete
Completecontrol
controlof
ofall
allData
Data Protector
Protectorfunctions,
functions,excluding
excluding
installation,configuration and reporting/notification.
installation,configuration and reporting/notification.
•• By
Bydefault,
default, database
database DBA
DBAusers,
users, such
such as
as Oracle,
Oracle, are
are
added
added toto this
this group
group when
when the
the integration
integration isis configured.
configured.
Student Notes
The Operator group has fewer capabilities than the Admin group. The members of the
Operator group are prevented from executing the following operations:
Although the Operator group does not have these permissions, it still has many other
powerful rights; therefore, care should be taken when assigning users to this group.
NOTE Operators have super-user like privileges through Backup and Restore!
The main purpose of the Operator group is to provide operators the ability to perform the
day-to-day operation of the Data Protector Cell. This is why the Operator group does not have
any Configuration type permissions, as these are functions typically performed by the
system administrator.
•• Permission
Permission to
to initiate
initiate restores
restoresonly
only
•• Can
Can only access their own data (private)
only access their own data (private)
•• Can
Can be
beused
usedby
byresponsible
responsible users
users who
who have
haveaccess
access
to their own tape drives or libraries
to their own tape drives or libraries
Student Notes
The User Group has permission only to initiate a restore of the user's own data. Those
responsible for backup must assign ownership of the backup job to allow a member of the
user group permission to see the data available for restore within the restore GUI.
Any media requests that accompany the restore session must be satisfied by the members of
the Operator or Admin groups.
Giving users the ability to restore their own data may be desirable in environments like
developer workgroups, where they have access to their own tape drives or libraries.
No intervention on the part of the Admin or Operator group members is required to satisfy
mount requests, if the correct media is loaded in the device specified by the restoring user.
By default, there are no users configured into the Data Protector User Group.
Custom Groups
•• Customer
Customerdefined
defined groups
groups can
can be
be created
created to
to suit
suit the
the local
local
environment.
environment.
•• Default
Default groups,
groups,operator,
operator, and
and user
usercan
canalso
also bebe modified.
modified.
Student Notes
In addition to the predefined default groups, Data Protector allows you to create your own
groups.
You may choose to create custom groups that match the structure and requirements of your
IT department.
Example
The User Group operator has all access rights, except client configuration, User
configuration, Device configuration, and Reporting and Notification. The user group user has
only the Start restore access right.
The IT organization may require some sort of hybrid where more senior users can format
tapes (media configuration), monitor, start backups, start backup specifications, mount
prompt, abort, and restore. In this case, a custom group can be created to satisfy this
requirement. The relevant users are then added to (or modified in) this group.
Group Permissions
Student Notes
The above slide displays the complete set of Data Protector permissions, and how these
permissions are assigned to the default user groups.
NOTE Many of the permissions will allow super-user capability indirectly, and are
considered very powerful rights. (e.g. Restore as root)
Permissions Explained
Select Users
to add a new
group
Select a group
to add new users
or modify access
Manually
add a user
Student Notes
Users and Groups can be created, modified, and deleted from the Data Protector GUI.
Alternately, if no GUI is available, modifications can be made directly to the configuration
files. Note, the “*” is displayed in the GUI as <Any>; this is a wildcard that may be used in any
of the first 4 fields.
Example:
<OMNICONFIG>/users/UserList file:
Additionally, the integrations with third party databases, such as Oracle, typically require that
a special user be added to the admin group or operator group to allow the backups to be
performed by the database administrator's user id. This will require that the backup
specifications are "owned" by that user as well.
No old password
by default
Password saved in
<OMNICONFIG/users/WebAccess
Student Notes
When Data Protector is installed on the Cell Manager, there is a web user (java) inserted into
the Admin group. There is no password required to access the Web Reporting applet by
default.
To provide more security, you may want to password protect the web functionality. The
protection requirement is largely due to the fact that through the web interface, notifications
and report groups may be modified, as well as the cell data is available.
Client Security
Secure
clients Secure
cell
<OMNICONFIG>cell/allow_hosts
Student Notes
Data Protector normally provides a secure environment from a user perspective. However,
the default installation allows for any cell manager or restore session manager to access any
disk and media agent, even those that are not members of the same cell.
This designed-in feature allows for remote recovery of data from one cell to another.
Although this may be a very valuable feature for some environments, it is also a security risk.
Cell administrators may want to prevent outside access to systems in the cell, or at least
restrict access to only known, trusted cell managers. This may be accomplished by
configuring the access limit to the cell by using the Secure feature shown above.
Using the Data Protector Install (cell administration) GUI, choose the cell manager or client
systems to restrict access to only certain systems. The resulting list of configured hosts is
stored in the <OMNICONFIG>/cell/allow_hosts file. This file is created on the cell
manager, and distributed to each cell client automatically when the secure cell choice is
made. Otherwise, each client may be independently secured. By default the cell manager is
always able to access the client, as the cell manager is registered in the
<OMNICONFIG>/cell/cell_server file.
The limited access restriction must include all systems that are to be remote managers, such
as OpenView Operations management stations.
To remove the access limit, either modify the allow_hosts file on each system in the cell, or
use the GUI to unsecure the cell or individual clients.
<OMNIVAR>/tmp/inet.log /sbin/init.d/inetd
ns
ectio 1
nn
rts
co 6
sta
s
lo g
/etc/services:
omni 5555/tcp
$OMNIHOME/bin/inet
s ta
5 rts
d Checked at startup
he cke inetd
ly c
n uous 4
i
c on t 2
/var/adm/inetd.sec
/var/adm/inetd.sec 3 /etc/inetd.conf:
omni stream tcp nowait root
$OMNIHOME/bin/inet/inet -log
$OMNIVAR/tmp/inet.log
DP - port 5555 request
Student Notes
The Omni-Inet process is the server started on each client to initiate the agents. Data
Protector sends service request on the well-known port 5555 to the Omni-Inet server on client
systems.
NOTE The service name is omni, and the daemon started is inet, hence the name
omni-inet.
On UNIX systems, this service request is intercepted by the inetd (super daemon). The
inetd.sec allows administrators to control access to the system “pre-service.” That is, the
service daemon (in this case inet) would not be started if the requestor does not pass the
initial security check. This level of security further enhances the security checking already
performed by Data Protector.
2. The inetd matches the port number of the requested service to the server by consulting
the /etc/services file; here the port identifies the desired service name. When service
requests come in the service configured in the /etc/inetd.conf file that matches the
service request port is identified.
3. After the service name is known, the inetd checks the security limitations for the
requested service by consulting the /var/adm/inetd.sec file.
4. If the remote system is authorized to start the local server then the inetd consults the
/etc/inetd.conf file for server startup instructions. Inetd starts the inet service daemon to
initiate the agents.
Example:
The /var/adm/inetd.sec file may contain an entry similar to the following entry to limit the
Omni-Inet daemon:
Firewall Support
VPN tunnel
Internet
Internet GUI
Internet
VPN tunnel
Firewall
DA MA
DA MA
DMZ
CM
Firewall
5555 (outbound)
5555 + OB2PORTRANGE(outbound)
Intranet
DA
CM
MA GUI
GUI
Student Notes
The goal of Data Protector’s support of systems outside a firewall is to provide a solution for
the backup of systems in the Demilitarized Zone (DMZ) while keeping either the cell console
(GUI) or some other clients within the Intranet.
Shown above are the two supported firewall configurations for Data Protector. Technically,
there are other possibilities but they represent security risks and should be avoided.
Supported Configurations
Below are the two supported and recommended configurations. In both cases, only ports
need to be opened for outbound connections:
• The disk agent and media agent in the DMZ (this is the recommended configuration)
• The cell manager, disk agent, and media agent in the DMZ.
Notice that the GUI connecting through the Internet using a point-to-point or VPN connection
requires no additional configuration on the cell manager, except the configuration for user
that will be logging in. Other Internet connections are not recommended due to the number
of ports that would need to be opened on the inbound connection from the GUI to the cell
manager.
In the examples shown on the slide, only the cell manager that is in the DMZ would need this
configuration.
The OB2PORTRANGE parameter controls the dynamic listening port range used by Data
Protector on an individual system. This range will not affect the inet port, by default set to
5555.
The range of ports sufficient for most Data Protector cells is around 100.
OB2PORTRANGE=5000-5099
The above formula represents the “worst case scenario” of concurrent processes.
Process connections
Process Connects to Port Numbers Used
Cell Manager Disk Agent/Media Agent 5555
Disk Agent Media Agent OB2PORTRANGE on Media
Agent system
User Interface (GUI/CLI) Cell Manager 5555 + OB2PORTRANGE on
the Cell Manager
NOTE Remote installation of client systems across the firewall is not supported.
Client systems need to be manually installed.
With Data Protector 5.0, a new implementation allows for specifying a port range for every
binary. The selection is based on the so-called ‘progname’ which is a platform independent,
functionality related identification of a Data Protector 5.0 executable. The same name
appears in version output, debug trace name as well as in some protocol version checking.
With this new method, administrators have the ability to target a group of agents that is
relevant for the cross-firewall operation. The size of the range for this group will be much
smaller than the range as described above. This will allow for fewer ports to be open in the
firewall.
The IPC library will check for a new omnirc variable named OB2PORTRANGESPEC. The
value of this new variable will be parsed and checked against the progname for the process
(thread) obtained from the programming libraries. If no match is found, the
OB2PORTRANGE variable will be parsed next to maintain compatibility with previous
versions.
Example:
OB2PORTRANGESPEC=CRS:7000-7009;xSM:7050-7099;DBSM:7200-7499
When a specification begins with a lower case x, as shown above in “xSM”, this means that
the range applies to the processes that have the rest of the string in the progname as a sub-
string. In other words, “xSM” would be used for all session managers, except for the DBSM
which is specifically identified with its own clause. If a specification is identified more than
once, the last one in the sequence applies.
Generally, the port ranges should not overlap, to prevent one group of processes from
exhausting the port range from another group. It is possible for two or more process groups
to share a process range, but then the range should be exactly the same for all of them.
3. Which two files store the Data Protector group and user assignments?
4. UNIX and Windows users can be added to the same Data Protector group. TRUE or
FALSE?
5. Any user in the Admin group may run backup and restore?
6. The /var/adm/inetd.sec must be modified for normal Data Protector operations, TRUE or
FALSE?
In this module, we will introduce you to the different recovery procedures that are available
with Data Protector. This module contains selected extracts from the Data Protector
Administrators Guide. It is essential that you do not use the contents in this module alone as
a basis for your own disaster recovery procedures. Refer to the Data Protector
Administrators Guide and Data Protector Concepts guide for more information on this
subject, as well as specific operating system documentation.
Disaster Recovery
Student Notes
A disaster is any situation in which a system does not function properly, whether due to
human error, hardware failure, or natural disaster. In these cases, the root (boot) partition of
the system is not available, and the environment needs to be recovered before the normal
restore operation can begin. This includes re-partitioning and re-formatting the boot partition
and recovery of the operating system with all the configuration information that defines the
environment.
This step must be complete in order to recover other user data.
Disaster is always serious; however, the following factors can exacerbate the situation:
• The system must be returned to online status as quickly and efficiently as possible.
• Disaster recovery is not usually necessary, and therefore, administrators are not familiar
with the required steps.
• The available personnel to perform the recovery may have only fundamental system
knowledge.
Corruption/loss of the
Data Protector database
Student Notes
There are three components of the Data Protector architecture that may require recovery:
• Client System
Recovery of a client system may be necessary because of hardware failure, or corruption
or loss of critical system software or configuration.
• The Data Protector Database
It may be necessary to recover the Data Protector database if it becomes corrupted and
beyond repair with normal database maintenance tools. The database must also be
recovered as a part of the cell manager recovery procedures if the cell manager fails.
(The IDB recovery is covered in the Internal Database module of this course)
• Cell Manager System
The cell manager system recovery is more complicated that a client system, as it holds
the Data Protector database and software. This database is not available during disaster
recovery of the Cell Manager, so offline recovery is necessary.
DR OS
P1S
SRD
OBDR
EADR
ASR
AMDR
Student Notes
There are several terms used in this discussion that are likely to be new, and need to be
introduced for clarity.
OBDR This One Button Disaster Recovery capability exists as a feature of most HP
tape drives. Data Protector prepares the image needed for the OBDR
automated system recovery feature.
EADR The Data Protector Enhanced Automated Disaster Recovery feature allows for
automated or semi-automated recovery of failed system. This requires
significant preparation to be used.
ASR The Data Protector integration with Microsoft Automated System Recovery
feature for Windows XP and 2003; provides for automated operating system
recovery and Data Protector DR processes.
AMDR The Data Protector Assisted Manual Disaster Recovery feature allows for
post-operating system installation recovery. Data Protector provides a “mini-
server/agent” installation with automated tools.
Phase
Phase0:
0:Preparation
Preparation
Phase
Phase1:
1:Configuration
Configuration
Phase
Phase 2:
2: Re-activation
Re-activation
Phase
Phase3:
3:Restore
Restore
Student Notes
The disaster recovery process consists of 4 phases:
Phase 0 Preparation
• Perform full client backups
• Update the DR OS Image after hardware/software changes
Phase 1 Boot the DR OS
• Replace any faulty hardware
• Boot the system from the DR OS CD-ROM
• Select the scope of the recovery
Phase 2 OS configured and Data Protector installed
• Critical volumes are automatically restored
• (including the boot partition, OS, and the partition containing Data Protector)
Phase 3 Restore missing data
• Restore any data not restored from the Phase 1 and 2 using Data Protector
Regardless of the disaster recovery method chosen, there are several steps that need to be
addressed before a successful disaster recovery can be executed.
Planning
Developing a detailed disaster recovery plan is crucial for successful disaster recovery. The
plan must be prepared by IT administration and should include consideration of the
following:
1. What systems must be recovered, within what specified time, and what other systems
need to be recovered, but with more relaxed conditions (like time frame, point in time to
which to recover, application data only)?
This will yield several categories of systems: High priority HP-UX, high priority Windows
NT, and standard priority Windows NT systems, for example.
On Windows, the most important configuration data is centralized in the registry, which
can be backed up by backing up the configuration object.
To confirm the validity of the planned recovery procedures, corresponding tests are required.
At what time intervals and using which systems (or copies of them) do you perform these
tests?
Some applications are not completely inactive, even if they have been shut down. Some have
daemons or processes active as soon as the system finishes booting, for various reasons (HP-
UX example: License server at run level-2). Such an early process may even read the data into
memory and write a “ dirty flag” into some file while it runs. A backup taken at the standard
operating stage (the standard run level-4) cannot be expected to yield a problem-free restart
of such an application. To follow the example, the license server, if started after such a
pseudo-recovery, will realize that the data read from the file is inconsistent, and will refuse to
run the service as expected.
Depending on what is active on the system when the backup runs, data consistency for an
application can be violated and result in restart and execution issues after recovery.
The best concept would be to perform the backup with the relevant partition(s) set offline.
However, this cannot be accomplished in most cases.
Student Notes
Data Protector 5.0 supports all of the disaster recovery options listed on the slide. The rest of
this module will outline the concepts for each type.
Student Notes
Data Protector 5.1 supports all of the disaster recovery options listed on the slide. The rest of
this module will outline the concepts for each type.
Student Notes
Assisted Manual Disaster Recovery (AMDR) is available for Windows NT, 2000, XP and 2003
operating systems. Any time after a backup for a client or Cell Manager is completed, a few
manual steps are necessary to be able to use the AMDR tools. Additionally, a one time
preparation (per platform) of the AMDR diskettes is required.
Student Notes
The graphic above highlights the location of the stored “dr” data used for AMDR as well as
other procedures discussed later in this module. The contents of these directories are used
by most of the methods used by Data Protector for disaster recovery. In most cases these
files are automatically copied via the graphical user interface (DR wizards) or the
omnisrdupdate command.
default
location: local
floppy drive
(a:)
Student Notes
Disaster recovery media for the Cell Manager must be prepared in advance to allow the
system to be fully recovered. Shown above is the Cell Manger AMDR preparation wizard. This
is useful for systems that don’t support the other more automated methods. The
omnisrdupdate command is used to update recovery.srd file on disk1 of the two disk AMDR
disk set.
The location for the stored recovery.srd file may be a local floppy drive (A:) or a network
share if available. When a share is used, the recovery.srd may be copied to the diskette when
needed for recovery of the Cell Manager.
omnisrdupdate.exe
–location may be
used with client
backup also
Student Notes
The graphic above illustrates the use of the backup post-exec to update the recovery.srd file
in the desired location for the cell manager. This procedure should be used whenever a new
configuration backup for the Cell Manager is performed. If the recovery.srd is outdated, then
additional manual recovery/restore steps will be necessary to recover the current Data
Protector Internal Database as well as other data to bring the system to the most recent
backup state.
Student Notes
The Data Protector disaster recovery wizard may be used to update disk1 of the recovery
disk set for a client system just when it is needed (or ahead of time if preferred). The GUI
shown above allows for a selection of any disk location and defaults to the local A: drive.
This is most likely run from the Cell Manager when a client system is unavailable and in need
of recovery. The AMDR disk set may also be created at this time and then updated as shown
above.
omnisrdupdate
Student Notes
The contents for the AMDR disk set for a 32-bit system is shown above. The 64-bit systems
require a third disk due to the size of the zipped executables on disk2.
The AMDR disk1 includes the drstart.exe program which is used after the installation of the
operating system (DR OS). The disk1 should contain the latest recovery.srd for the specific
client that is in need of recovery. The recovery.srd file is available on the cell manager by
utilizing the data in the <OMNICONFIG>\dr\srd directory and omnisrdupdate command.
Student Notes
The procedure for recovering a failed client is as follows:
Recover other partitions using normal DP restore procedures as well as any other vendor
specific partitions. (see the Administrators guide for more details)
Perform full
OBDR backups
Restore other
data
System Restored
DP Platforms:
Windows NT
Windows 2000 Initiate OBDR
Student Notes
HP’s one button disaster recovery (OBDR) support within Data Protector was introduced
with version Omniback version 3.5 and is currently supported for Windows NT/2000.
The principle behind OBDR is simple. The tape drive is used to perform a full Data Protector
backup in addition to an ISO image of the Operating System. Using a patented process, the
tape drive is able to emulate a SCSI CD-ROM drive so that during a Disaster Recovery the
computer system may be booted from this device. Once the boot process has completed, the
tape drive switches back into “normal” operating mode and proceeds to restore the system
automatically.
The entire recovery process is automatic and removes the need to locate drivers, manuals
and the software applications and licenses during a disaster recovery. The creation and
maintenance of cumbersome boot disks becomes obsolete and everything you need to
perform the recovery can be stored on tape.
The HP OBDR is supported on a limited number of hardware (PC) platforms with certain HP
tape devices. Consult the documentation on the HP URL:
http://www.hp.com/products1/storage/products/tapebackup/obdr.html for the current
platform support.
If OBDR is supported for your platform, use the Data Protector Windows GUI to create an
OBDR backup. You will need to have the Windows installation CD-ROM during this backup
process for Windows NT. The Data Protector OBDR wizard will guide you through the
backup process. Go to the Backup context within the GUI, then select the tasks tab, select
the OBDR wizard. (see the next slide)
OBDR requirements:
• Automatic Disaster Recovery component must be installed.
• It is essential to have an OBDR capable computer configuration: the system’s BIOS must
support bootable CD extensions as defined in the El-Torito standard and read/write
access to hard disk drive using LBA addressing via INT13h function XXh. The OBDR
device must be conform to the same standard when emulating the CD-ROM.
• The hardware configuration of the target system must be the same as that of the original
system. This includes SCSI BIOS settings (sector remapping).
• Replacement disks have to be attached to the same host bus adapter on the same bus.
• An additional 200 MB of free disk space is required on the boot partition. If this disk
space is not available, the disaster recovery fails.
• All drivers, required for boot must be installed under the <%SystemRoot%> folder.
• Network must be available when you boot the system in Safe Mode with Networking
(Windows 2000) or in Directory Services Restore Mode (Windows 2000 Domain
Controller).
• A media pool with a Non-appendable media usage policy and Loose media allocation
policy has to be created for the OBDR capable device. Only the media from such a pool
can be used for disaster recovery.
• The logical device must have a 64Kb block size as its default, and be assigned the non-
appendable/loose media pool.
OBDR limitations:
• Multiboot systems that do not use Microsoft's boot loader are not supported.
(e.g. LILO boot loader from Linux)
• Internet Information Server (IIS) Database, Terminal Services Database, Certificate
Server Database and MS Cluster Service Database are not restored automatically
during Phase 2. They can be restored on the target system using the standard Data
Protector restore procedure (except for Microsoft Cluster Service database on
Windows NT).
When backing up the client, the default 64K block size should be used to write to the device
if you plan to perform offline restore. This is the only default block size available on
Windows when performing disaster recovery.
OBDR Preparation
Requirements:
• 64Kb block size (logical device)
• Non-appendable media pool
• Logical device default media pool
non-appendable
• Local GUI
• Local OBDR capable tape device
Student Notes
The Data Protector requirements for OBDR are as follows:
• 64KB block size set for the logical device
• media pool allocation set to non-appendable
• logical device must be assigned to the non-appendable pool in advance, no device
properties may be changed during the OBDR wizard
• local GUI may be required depending upon the platform of the Cell Manager; the
OBDR wizard only runs on supported platforms.
• the local device must have HP’s OBDR capable firmware installed. Use HP
StorageWorks Library and Tape Tools to verify the OBDR capability, or to upgrade
the drive firmware.
Student Notes
The Data Protector OBDR Wizard may be used to create the recovery tape for the Cell
Manager; this must include the Internal Database. Often the Internal Database is backed up to
media separate from the filesystem of the Cell Manager, however, in the case of OBDR they
must be included in one backup. The hot-backup mode for the Internal Database is used as
normal, and the database is also checked for corruption during the backup.
device properties
not modifiable
via wizard
Student Notes
While using the OBDR wizard, device properties may not be modified. Because of this, it may
be beneficial to have a separate logical device defined as the OBDR device, with all of the
specific settings already prepared.
OBDR Session
Student Notes
Data Protector uses the omniiso.exe to create the boot image used to restore the operating
system. This will require additional storage space in the <OMNIHOME>\tmp directory while
the backup is executing. The “recovery.iso” file for Windows 2000 is approximately 80 MB
and will be removed from the temporary location after it is moved to the tape.
A log file called autodr.log is stored in the same directory as the recovery.iso file, and may be
used to help troubleshoot failed OBDR sessions. Additionally the autodr file contains OBDR
processing details.
Student Notes
Use the DR OS CD to boot the failed system. Data Protector automatically installs and
configures the DR OS, formats and partitions the disks, and finally recovers the original
system as it was at the time of backup.
The general steps using the Enhanced Automated Disaster Recovery method for a Windows
NT/2000 client are:
EADR requirements
• Automatic Disaster Recovery component must be installed.
• The hardware configuration of the target system must be the same as of the
original system. This includes SCSI BIOS settings (sector re-mapping).
• Replacement disks have to be attached to the same host bus adapter on the same
bus.
• Boot partition has to be larger than 100 MB or disaster recovery will fail.
• An additional 200 MB of free disk space is required on the boot partition. If this
disk space is not available, the disaster recovery fails. This additional free space is
needed during collecting information for EADR (e.g. scanning registry, WINNT and
system32 directory etc…)
• All drivers required for boot must be installed under <%SystemRoot%> folder.
• Network must be available when you boot the system in Safe Mode with
Networking (Windows 2000) or in Directory Services Restore Mode (Windows 2000
Domain Controller).
• The systems BIOS must support bootable CD extensions as defined in the El-Torito
standard and read/write access to hard disk drive using LBA addressing via INT13h
function XXh.
NOTE When backing up the client, the default 64K block size must be used to write
to the device if you plan to perform offline restore. This is the only default
block size available on Windows NT/2000 when performing disaster recovery.
EADR steps
• Full host backup with configuration (if this is also Cell manager include backup of Data
Protector database)
• Verify all Warnings in backup session log check if there are any system/application
critical files locked during backup.
• Inspect AUTODR.log file and search for any Error or Critical warning. In case if there are
Errors reported; check if reported files are needed during installation, logon and starting
restore session.
• Use Restore Tasks start wizard for preparing .iso image which will be used to burn
bootable CD.
• After you have iso image You need to use one of CD burning software to create bootable
CD for crashed system. CD burning software must support burn CD from iso image.
Created bootable CD contains usually about 90 Mb size. Depends which operating system
was installed on crashed host.
• Verify settings in BIOS of crashed system: boot-up sequence, security and CD-ROM
section in BIOS must be set the way it is possible to boot from CD drive.
• Verify if backup device and backup server host are available on the network. Check if
Inet service is running and if needed media for restore session is available.
• Switch on / restart crashed system and follow instructions. You must press F12 to
proceed with EADR process otherwise system will try to boot from hard disk.
• Select recovery scope from menu. Small amount of data is written to disk followed by
reboot. Now you can remove bootable CD from drive.
• Follow instructions on screen. Several reboots might be needed during phase one. First
when small amount of data is written to disk and partition information is placed on hard
disk. After reboot partition information is checked. At first time partition is FAT and
needs to be converted to NTFS if there were NTFS partition used before crash. (Microsoft
recommends NTFS type of partition for bootable data)
• Next step is to start mini operating system and updating information of original system.
After that in command window full session restore is started.
Optional you can monitor restore session from DP GUI is Cell manager is available.
Hint: If you suspect that restore session has stopped, verify the activity on backup device and
in monitor session (amount of restored data).
Student Notes
Student Notes
The disaster recovery image must be prepared ahead of time for the Cell Manager, but just in
time for the cell clients (or ahead of time if desired). Use the EADR wizard to create the
recovery file that is to be “burned” to CD-R.
Image may be
created from disk
file or recovered
from latest backup
tape
Student Notes
The image to use for the creation of the EADR CD is selectable at the time of the creation. If
the full DR image was copied to the Cell Manager during the backup, then it should be
available in the <OMNICONFIG>\dr\p1s directory; otherwise it may be recovered from the
last backup tape. The Cell Manager disk copy will be much faster, but requires some
additional disk space to store the image for each client. (approximately 30Mb per client)
Student Notes
When multiple backup sets exist for a client, the wizard presents a list of possible objects to
use for the recovery set. Only one version for each object may be selected. After the objects
are selected, Data Protector will prompt for the location of the recovery image.
Student Notes
The recovery image (recovery.iso) may be stored in any location prior to recording it to CD.
As soon as the destination is selected the ISO image will be created. The ISO image file
creation usually takes 5 to 10 minutes. The resulting file may be 80-100MB depending upon
the system. The recovery image creates a bootable DR OS CD for system recovery.
Student Notes
Once the recovery.iso file is created, Data Protector is essentially done with the EADR
backup process. Use an ISO compatible CD-burning software to copy the recovery.iso
contents to CD and then use this EADR CD to boot the system.
Student Notes
The boot process for OBDR and EADR is very similar. The main difference is getting to the
point where the Disaster Recovery medium is used to start the system. In the case of OBDR,
the tape drive must be switched into the CD-emulation mode at power-up and then chosen
for the boot device (manually; the boot menu is available by using F8 on some systems). The
EADR ISO CD must also be selected as the boot medium, and the F12 key must be depressed
to initiate the DR process (as shown on the slide). If F12 is not selected within 10 seconds,
the default boot (from the hard drive) is attempted.
Phase 2 aspects:
Offline recovery is performed if the Cell Manager is not accessible (e.g. due to network
problems, Cell Manager has experienced a disaster, online recovery has failed, etc.). Only
standalone and SCSI-II Library devices can be used for offline recovery. Note that recovery of
Cell Manager is always offline.
Remote restore is the most common way of restoring data on DP client because of central
media repository and managing. Local restore has an advantage in that it may be faster and
no need for network access or Cell Manager access is required.(in case if network has down
etc…). This allows for clients to be restored in advance of DR on the Cell Manager in case of
a site failure.
The Data Protector command (installed by the temporary DP installation) begins with a call
to drstart.exe, which looks for the latest system recovery data (SRD) and then executes
omnidr.
GeneralOptions
-target <hostname> Target system hostname.
-local Forces restore to a local device.
This command is used to restore any type of backup objects in the absence of a working
database, which may have been caused by a disaster or lost connection with it. It is used as a
standalone CLI utility or --better yet-- the higher level utility omnidr.exe, which assumes
default invocation of omniofflr.exe based on the SRD file contents.
Offline restore is based upon the omnidbrestore program code (used to restore the IDB). This
command is used when the Cell Manager is unavailable.
The omniofflr.exe requires exhaustive details about the restore device and backup media,
including the position of backup objects on the medium. Information regarding the media can
be obtained from the SRD file or alternatively by querying the DP database using omnidb.exe
command in phase 0 i.e., when the system is still intact. The user can also write a script,
which queries the database and prepares another script where a bunch of omniofflr.exe
commands are executed with appropriate options.
NOTE UMA for media management is not working if there is no Cell server up
and running. The required tape must be manually loaded using library
functionality (manually).
Syntax:
Options: Device
-name <LogicalDeviceName>
-dev {<PhysicalDevice>}
-mahost <DeviceHostname>
-policy <Logical Device Policy No.>
-type <Logical Device Type No.>
-description <DeviceDescription>
-blksize <BlkSize>
-name <LogicalDeviceName>
parameter that specifies the logical device name
-dev <PhysicalDevice>
specifies the physical device
(c:\temp\dev1, scsi1:0:0:0, /dev/tape0…)
-mahost <DeviceHostname>
specifies the name of the host, where the restore device is located
Media Agent started
-description <DeviceDescription>]
optional parameter that specifies the logical device description.
-blksize <BlockSize>]
optional parameter that specifies the block size the device is going to use when
accessing media
Student Notes
Windows XP and .Net Windows 2003 offer a process called automated system recovery
(ASR), which can perform a system recovery after a system crash occurred (for example
hard disk is defective). ASR has two parts: ASR backup and ASR restore.
You can access the backup portion through the Automated System Recovery Preparation
Wizard located in the Windows Backup functionality. The Automated System Recovery
Preparation Wizard backs up the System State data, system services, and all disks associated
with the operating system components. It also creates a floppy disk, which contains
information about the backup, the disk configurations (including basic and dynamic
volumes), and how to accomplish a restore.
The system state includes a collection of system-specific data maintained by the operating
system that must be backed up as a unit. It is not a backup of the entire system. The System
State data includes the registry, COM+ Class Registration database, system files, boot files,
and files under Windows File Protection. For servers, the System State data also includes the
Certificate Services database (if the server is a certificate server). If the server is a domain
controller, the System State data also includes the Active Directory database and the SYSVOL
directory. If the server is a node in a cluster, it includes the Cluster database information. The
IIS Metabase is included if Internet Information Services (IIS) is installed.
You can access the restore part of ASR by pressing F2 when prompted in the text mode
portion of setup. ASR reads the disk configurations from the floppy disk and restores all of
the disk signatures, volumes, and partitions on the disks required to start your computer. (It
will attempt to restore all of the disk configurations, but under some circumstances, it may
not be able to do so). ASR then installs a simple version of Windows and automatically
restores data from the backup created by the Automated System Recovery Preparation
Wizard.
For more information on how to use ASR directly with Windows XP or 2003, see Microsoft
documentation.
Data Protector 5.1 offers a complete integration with ASR that provides all the benefits of
ASR, without the need to deal with Windows XP/2003 backup and restore tool directly. The
following slides describe the integration in detail.
Disaster occurs
Phase 3 • All user and application data are restored using the
restore normal restore procedure
Student Notes
This slide gives a brief overview about all necessary steps, required to perform a successful
disaster recovery integration of ASR with Data Protector.
There are four distinct phases:
phase 1: This is the phase right after the disaster happened, i.e. the system has
configuration crashed and can’t be started. In this phase, a temporarily existing OS and
some Data Protector binaries are installed, which automatically launch
the restore process.
phase 2: In this phase, the original storage structure and its contents are
re-activation automatically re-established. All data belonging to the root and system
partition, as well as the partition where Data Protector previously resided,
is restored.
phase 3: Within this phase the user restores user related data, which were not
restore automatically restored during phase 2.
ASR focuses only on three partitions (also called volume or disk): boot, system and data
protector. These three partitions are also called the critical partitions.
A boot partition (also called disk or volume) contains the files required for the initial step of
the boot process, whereas the system partition contains operating system files. The Data
Protector partition hosts Data Protector executables, and in case of a Cell Manager also the
IDB.
Student Notes
Phase 0 comprises all of the necessary steps which have to be done prior to a disaster. The
first prerequisite is to perform a full system backup, containing all partitions as well as the
CONFIGURATION part.
During such a backup, the so-called system recovery data (SRD) information and the data
required for ASR is collected. This is warranted only if the Automatic Disaster Recovery
module is installed on the client system. To check whether these data are collected properly,
the session should contain the following messages:
The SRD and the ASR data are stored in separate files. For a CM running on HP-UX, the data
are stored under /etc/opt/omni/dr/srd/ and /etc/opt/omni/dr/asr/ respectively. For windows
CM, the corresponding files are located under
<Omni Home>\config\dr\srd\ and <Omni Home>\config\dr\asr\.
Both files inherit the corresponding system name. The ASR file contains data used for the
ASR procedure. The SRD file is used for other disaster recovery procedures as well.
After a backup, the SRD file doesn’t reflect any information about session and media Ids.
However, this information is essential to successfully execute a disaster recovery. During the
creation of ASR diskette sets, (see next slide) the SRD file is updated; this effectively means
that the information about session and media ID, important for the restore, is extracted from
the internal database of Data Protector and added to the SRD file.
This update process can also be done manually, via the GUI or the command omnisrdupdate.
Since the updating of the SRD file is accomplished as an integral part of the creation of the
ARS file set, it is not necessary to do it manually.
Student Notes
For all Data Protector client systems, the creation of the floppy set or the so-called ASR set
can be done after disaster strikes (just in time). This only requires another Window system,
where the Data Protector GUI component is installed. For the Cell Manager, however, the
floppy disk set must be prepared before the disaster occurs as the Cell Manager is the source
of the data for the ASR set.
The ASR set can be created after a full client backup was performed. This is done via the
Data Protector GUI wizard, under the Restore context, Tasks, and Disaster Recovery.
In the wizard, when the version of the object is selected, corresponding information about
the media and session ID is extracted, and subsequently added to the SRD file (This is
otherwise, also, known as update the SRD file).
When the ASR disks are created for the first time it is necessary to select the option Copy DR
installation, in order to make sure that all necessary data are written to the diskettes. Once
the ASR set is created, only the first diskette (which contains ASR information) has to be
updated, i.e. the option Copy DR installation doesn’t have to be selected. Such an update is
necessary after each hardware or software configuration change. This is also true for any
network changes, such as a change of IP address or DNS server.
Having the option Copy DR installation enabled, files which are needed in order to start the
restore, like vrda_2l.exe, inet_2k.exe, vrda_nt.exe, inet_nt.exe, omnienu.dll, rma.exe,
omnidr.exe, omnir.exe, devbra.exe, omnioflr.exe, etc. are grouped together as a zip file and
copied to the floppy disk, disk 2 (and/or disk 3 in the case of the 64-bit systems).
Note: When performing the creation with the option Copy DR installation, and using
the floppy drive as the destination, the following windows pop up after the
creation of the configuration files:
This means that the diskette, which is already in the floppy drive should be used for this step
too, i.e. the diskette remains in the floppy drive while the dialog box is confirmed. Only, in
case the space is not enough to accommodate the DR installation bits, an additional disk has
to be utilized.
This precautionary measure is taken, because the ASR configuration data files (like SRD)
vary in size and if they reach a certain size, the maximum size of the floppy disk is exceeded,
and therefore an additional disk has to be used.
NOTE: The option Copy full DR image to disk can be selected but has no impact. This option
is only valid for Enhanced Automated Disaster Recovery (EARD).
Student Notes
To create the ASR set, Data Protector requires the selection of the versions of the critical
volume backups that are desired for disaster recovery. The selection of these volumes causes
an automatic update of the SRD (recovery.srd) to be created on the destination disk(s).
creates disk2
Student Notes
The destination for the ASR set us usually the local floppy drive, but may also be folder on
the local system. For successful use of the ASR set, the disk folders contents must be copied
to floppy disks and be available during the ASR recovery procedure.
The selection of the Copy DR Installation causes the creation of Disk2, and potentially Disk3.
These disks will hold the DR installation for Data Protector (agents, etc).
Student Notes
In the first phase of the ASR procedure, a temporary OS and some Data Protector (DR)
installation binaries are installed. This is required to initiate a restore of the complete system
(boot, system and DP partition) afterwards (phase 2 see next slide).
2. Press F2 during the start of the OS setup to enter the ASR mode.
3. Insert the first (updated) diskette from the ASR set. Change the diskette(s) when
prompted. After inserting the first diskette ASR re-creates the layout of the boot and
system disk(s).
NOTE: If third party SCSI or RAID driver are to be installed or configured then this
must be done before step 2 (invoked with F6).
After the first diskette is inserted, ASR will reformat the disk layout and copy files from the
CD-ROM. After that the system is automatically rebooted. When the system comes up again
the Data Protector disaster recovery wizard (drstart.exe) is started and prompts for the
second floppy disk.
After the restore the system is rebooted. The Windows CD and the floppy disk should be
withdrawn.
NOTE: Only in phase 1 it is necessary to boot from the CD-ROM. The hard disk is
used for subsequent reboots.
Student Notes
If the Cell Manager can be reached then, a so-called online restore is performed. In this case a
session manager is started on the Cell Manager system and the session can be monitored in
the DP 5.1 GUI.
Offline recovery is performed if the Cell Manager is not accessible (for example, due to
network problems, Cell Manager has experienced a disaster, online recovery has failed, etc.).
Only standalone and SCSI-II Library devices can be used for offline recovery. Note that the
recovery of a Cell Manager is always offline.
Offline recovery invokes the omniofflr.exe which was installed during the DR installation.
• After phase 2, the following there sessions should have been performed
• session 1: restore windows file protection catalog
• session 2: file system restore
• session 3: CONFIGURATION and IDB (in case of CM)
Student Notes
In phase 2, the system, boot, and data protector partitions are restored. This phase is
performed automatically and doesn’t require any manual intervention,.
At the end of phase 2, the system is rebooted. Once the system is up again, all the data on the
boot, disk and Data Protector partition are available again.
Data Protector will install a command script into the directory \All Users\Start
Menu\Programs\Startup, which is executed at the first login after the recovery. It performs
some cleanups and is then removed automatically.
After phase 2, the following three separate restore sessions will have been executed and are
visible inside the GUI:
Student Notes
In phase three all partitions, not yet restored, must be restored manually. There are also some
limitations of the MS ASR procedure: Special OS services aren’t restored within the disaster
recovery scope, because their backup API requires the services be online, which is not the
case during ASR, where the services are not running.
• When backing up the client, the default 64 Kb block size should be used to write
to the device if you plan to perform an offline restore. This is the only default
block size available on Windows when performing disaster recovery.
• Multiboot systems that do not use Microsoft's boot loader are not supported.
• Internet Information Server (IIS) Database, Terminal Services Database, and
Certificate Server Database are not restored automatically during Phase 2. They
can be restored to the target system using the standard Data Protector restore
procedure.
• Data stored on vendor specific partitions is not automatically restored during
ASR. The partitions will be recreated during the ASR, but you will have to restore
the data manually using the vendor specific procedure for restoring data.
However, you can restore data on EISA utility partitions using the standard Data
Protector restore procedure.
• Only those local backup devices are supported that can be installed by Windows
during OS installation (no additional drivers are required).
drstart.exe (interaction)
enable
debug
logs
Student Notes
The first Data Protector component, which is started by ASR is the Disaster Recovery Wizard
(drstart.exe). This wizard normally runs automatically without any user interaction.
1. When the Disaster Recovery Wizard appears, any button can be clicked in
order to stop the automated procedure.
2. Click on Cmd button to open a command line window.
3. The option Use Debugs must be selected.
4. Click Finish to continue processing
When the error appears, the debugs can be copied from the directory
%System32%\ob2dr\tmp either to a floppy disk or to a different partition. This can be done in
the command line window opened in step 2.
In case the backup, required for the disaster recovery, was done to a library connected to the
system, which needs to be recovered (local attached), there are certain aspects, which might
require additional considerations.
Since the unattended setup of Windows uses dhcp, the system name is set to a dhcp name
(like, dhcp-15-139-46-5.bbn.hp.com). Because of this system name mismatch, arising because
the original name stored on the Cell Manager is different from the name used on the client
(dhcp system name), the communication between the Restore Session Manager and the
Media Agent doesn’t work, and consequently the online restore will fail with an appropriate
message.
If the online restore fails, the ASR procedure automatically attempts an offline restore. In this
case, the above mentioned issue becomes invalid.
If the released scsitab file was updated with the library used for the disaster recovery
process, it must be copied onto the first ASR disk. To find out whether this copy is necessary,
follow the steps below:
Removable Storage Manager is a Windows service used for managing removable media (such
as tapes and disks) and storage devices (libraries). Removable Storage allows applications to
access and share the same media resources. Before, any robotic libraries are configured on
Windows system this service must be disabled. Because during an offline restore, DP auto
configures the library, this general requirement is also true for ASR.
To disable the Removable Storage Manager during ASR, follow these steps:
1. When the disaster recovery wizard window pops up, stop the procedure by
pressing any key.
2. Open a command prompt window by clicking Cmd button.
3. Enter the command “net stop NtmsSvc”.
4. Click Finish to continue.
Note: The autoconfiguration is performed in case the restore MA host is not responding. This
is the case when a locally attached drive is used, because the unattended installation of ASR
sets the hostname to “machinename”, which, of course, does not match the name as
described in the recovery.srd file. In order to skip such an autoconfiguration, change the
system name specified with –mahost inside the recovery.srd file to “machinename”
(unsupported feature).
In case, DP can’t fully auto configure the library, and detects the drives only as standalone
then, that medium, containing the backup, has to be manually loaded into the drive.
When the clients within the DP cell are secure, they allow only certain Cell Managers to
connect to them. This causes problems in case of an offline restore. In this case, the system
which is going to be recovered acts as a Cell Manager, and tries to connect to the client to
which the library is connected. If this client does not allow this, then the offline restore will
fail. In this case the secure option for this system has to be switched off.
DP 5.1 supports: HP-UX, Solaris, Aix, Tru64, Window NT 4.0, Windows 2000, Windows XP
Student Notes
As with manual disaster recovery, the administrator must ensure that before the disaster,
enough data to be able to correctly format and partition the disk has been collected. For
Windows NT/2000, however, Data Protector automatically stores the relevant information as
part of the configuration backup if the Automated Disaster Recovery component is installed
on the client and Cell Manager systems. (Use the Data Protector Client context to add
components)
An auxiliary disk is a disk that has been prepared in advance, specifically to recover
failed client systems. The disk must contain a bootable operating system, networking, and
the Data Protector disk and media agents. A separate auxiliary disk is required for each
platform type that may be recovered, (i.e. HP-UX, NT, etc).
1. Connect the auxiliary disk to the faulty system, replace the faulty disk with a new
disk, and reboot the system to the minimal installation installed on the auxiliary disk.
This establishes the network connection to the Data Protector Cell.
3. Use the Restore function provided by Data Protector to restore the boot disk of the
faulty client onto the replacement disk.
4. Remove the auxiliary disk, and reboot from the new disk.
The first method, using a working Data Protector client system, is described in this section.
Preparation
It is imperative that you complete a few steps in order to prepare for disaster recovery. In
addition to completing the steps listed in this section, read and follow the section, “ Preparing
for a Disaster Recovery,” in the Data Protector Administrators Guide.
Remember that once the disaster occurs, it will be too late to perform disaster recovery
successfully, if you have not prepared in advance.
In order to recover from a disaster quickly, efficiently, and effectively, you will need the
following:
• The most recent, successful full backup of the client that you want to recover.
• In case you want to use the hosting Data Protector client method for recovery, you need a
client of the same platform type as the crashed client, (i.e. Windows NT, and the
hardware I/O path to connect the new disk).
• If you choose the auxiliary method for recovery, you need a bootable disk, compatible
with the hardware of the crashed system. The disk should contain:
Recovery
This section provides the procedure for recovering your Windows NT client system using the
disk delivery method.
Using the disk delivery method on NT, you use a working Data Protector client to restore the
latest full backup of your crashed disk onto a new hard disk connected to the hosting client.
You then replace your crashed disk on the faulty system with this new hard disk. The actual
disk delivery disaster recovery procedure consists of the following steps:
1. Connect the new disk to a working Data Protector host client system.
2. Reboot the host client system so that the new disk is recognized.
3. From an Data Protector client that has the GUI installed, run the
Manager>Restore>Tasks tab>Disaster Recovery wizard. From there, select
the hosting client.
4. If partitioning has not already been done using one of the commercial partitioning
packages, when prompted in the disaster recovery Wizard, partition the new disk using
the disk administrator and the original partition size information provided by Data
Protector. When partitioning the system, it is recommended that you assign partitions in
the same order as they were at the time the full backup was performed. This is an
especially important procedure in the case of system partition. It simplifies drive letter
reassignment after restore and prevents a possibility of failure at system restart, because
of an inappropriate path to the system partition in the boot.ini file.
5. Perform any necessary drive letter mappings. Drive letter assignments — available by
right clicking on the original drive letter — serve as anchor points for the Restore
Into option, when performing a restore of data.
6. Restore the latest full backup of your crashed disk onto the new hard disk using the
Restore Into option.
NOTE Do not close down the GUI when performing a disk delivery restore because
the restore will cease. This is not the same as a normal restore session.
7. Remove the new disk from the client, and then connect it to the crashed system.
8. Reboot the faulty system. This completes the recovery of the client system.
The first method is to use a working Data Protector client system and create the new disk
while connected to this host.
Connecting a bootable disk that contains a minimal OS installation and Data Protector Disk
Agent to the crashed system will also work is the second method, and is described in this
section.
The administrator must ensure that enough data has been collected before the disaster to be
able to correctly format and partition the disk.
This is the preferred disaster recovery method for an HP-UX client.
Limitations
• HP-UX 10.x and 11.x
This description does not cover the recovery of a cluster environment. Depending on the
configuration of the MC/Service Guard environment, additional steps and modification to
the environment are necessary.
Preparation
Preparation for this disaster recovery is provided on several levels:
• Gathering the information for your backup specification
• Preparing the disk
• Preparing your backup specification (pre-exec)
• Executing the backup.
All of these preparatory steps are necessary before executing disaster recovery on the client
system.
One-Time Preparation
This section provides a list of items that need to be executed for each target system at
backup time, in order to perform successful disaster recovery. If the information is collected
as part of a pre-exec command, it is important to document the location of these files in the
disaster recovery plan, so that the information can be found when disaster strikes. Also,
version administration (there is a collection of the “ auxiliary information” per backup) must
be considered. The details given apply to HP-UX 10.x.
In case the system to be backed up has application processes active at low run levels, you
must establish a state of “ minimal activity” (modified “init 1 run level"):
2. Ensure that rpcd is configured on the system (configure the variable RPCD=1 within
the file /etc/rc.config.d/dce). This prepares the system so that it can enter the
state of minimal activity. The state can be characterized as follows:
a. Collect all the necessary information about the environment and put it in a
place where it is available if disaster recovery is needed. It is suggested that
you put it onto a different system that can be accessed easily. The information
should cover:
• Physical and logical storage structure of the storage
• Current logical volume structure (vgcfgbackup and vgdisplay -v)
• ServiceGuard configuration data, disk-mirroring, striping
• File systems and mount points overview (bdf or copy of /etc/fstab)
• The output of the swapinfo command
• I/O-structure overview (ioscan -fun and ioscan -fkn)
• Client network settings
b. An emergency copy of the data can also be put into the backup itself.
However, if done so, the information must then be extracted prior to the
actual recovery.
d. Shut down all applications, unless the application data gets backed up
separately, for example, using online database backup.
e. You may want to restrict network access to the system, so that nobody can log
on to the system while the backup is running (overwrite inetd.sec and use
inetd -c).
f. If needed, enter the state of minimal system activity (sbin/init 1; wait 60;
check if run_level 1 is reached). Note that this is a modified “ init 1" state.
g. Provide a post-exec script that elevates the system to the standard run-level,
restarts applications, and so on.
h. Set up a backup specification for the client on the Data Protector cell
manager. It should include all the disks (with disk discovery) and include the
pre-exec and post-exec scripts.
2. Attach the auxiliary disk (which contains the HP-UX operating system and the
Data Protector client) to the system, and make it the boot device.
4. Reconstruct the LVM structure, if applicable. Use the saved data for the non-root
volume groups (with vgcfgrestore or SAM).
5. Additionally, the root volume group (to be restored) must be created on the
repaired disk (using vgimport). It will not look like a root volume group during
the restore process, because we are currently running on the OS from the
auxiliary disk.
7. Reconstruct any other storage structure, like mirror, striping, ServiceGuard, and
so on, from the data saved on a secondary storage device during backup.
8. Create the file systems and mount them as required by the data from the backup.
Use similar, but not the original, mountpoint names (like /etc_restore for
/etc, and so on).
10. Start the Data Protector user interface and open a connection to the Data
Protector cell manager. Import the system with the auxiliary disk into the cell.
11. Select the version from which you want to restore. First, list all the required
media for the restore and make sure they are available. Restore all the required
mountpoints, including the (future) root-volume to the system, using the option
"Restore As <new_mountpoint>". The root-volume from the backup is
restored to the root-volume on the "repaired disk." Nothing is restored to the
currently running auxiliary operating system on the auxiliary disk.
14. Make the system reboot from the new (or repaired) disk.
NOTE Instead of using an auxiliary disk, the new disk can also be temporarily
connected to a hosting system that requires an Data Protector disk agent.
After being restored, it can be connected to the faulty system and booted.
HP-UX Clients
Student Notes
The recovery process that must be followed for a HP-UX client is made up of three basic
steps:
• Install Minimal OS
With a worst case disaster, the system disk or all disks have been lost or corrupted. When
the faulty or destroyed hardware has been replaced the first step is to install the
operating system. Usually, a minimal installation is all that is required to get the system
up and running. A minimal installation leaves out all the non-core software (which will be
recovered from backup later).
• Configure Networking
TIP Ignite/UX can be used to prepare recovery tapes that make the recovery much
simpler and faster. See the Ignite/UX documentation for more information.
TIP When performing a full restore, Data Protector's restore option “move busy
files” can be used to restore program files and libraries that may already be in
use.
Student Notes
The recovery process that must be followed for an HP-UX cell manager is composed of three
basic steps:
Step 1 — Recover the Operating System
• Install Minimal OS
The cell manager software must be installed again from the CD media. We covered this
same procedure earlier in the Installation module
Following installation of the Data Protector software, the Data Protector configuration
will only contain the default media pools. The recovery process requires that the Data
Protector database be recovered. Before the database can be recovered, a Logical Device
must first be defined, so that the import procedure can be used. A temporary logical
device definition is created for the import.
A tape containing the most recent copy of the Data Protector database must be
re-imported into one of the default media pools. You must do this, because the database
is now empty and has no knowledge of this or any other previous backups that have been
performed. The import procedure reads the tape and populates the database with
information regarding the objects present on the tape. The import procedure can take
quite some time, depending on the number of objects on the tape.
TIP It is highly recommended that you have Data Protector database backups
write to their own media pool and tapes; also, that you have a manual method
of identifying the tapes. In this way, the import procedure can be faster (no
other objects on tape) and the tapes can be identified more easily.
CAUTION When performing a full recovery of a cell manager system, take special care
not to overwrite the active Data Protector database that you have just
recovered, with one on the backup tape that may be part of a file system or
host backup. For this reason, it is highly recommended that you always
exclude the IDB from all filesystem and host level backups!
TIP When performing a full restore, Data Protector's restore option, “move busy
files,” can be used to restore program files and libraries that may already
be in use.
The
Theenterprise
enterprisemay
mayconsist
consistof
ofmultiple
multiplecells
cellsdue
dueto:
to:
•• Geographical
Geographicalreasons
reasons
•• Database limitations
Database limitations
•• Implementation
Implementationof
ofseparate
separatecell
cellmanager
managerplatforms
platforms(UX,Win)
(UX,Win)
The
Thepitfalls
pitfallsof
ofmaintaining
maintainingmultiple
multiplecells
cellsinclude:
include:
•• More
Moredifficult
difficultto
toadminister
administer(than
(thanaasingle
singlecell)
cell)
•• Unable
Unabletotoshare
sharebackup
backupLogical
LogicalDevices
Devices(libraries)
(libraries)among
amongcells
cells
•• Lack
Lackof
ofenterprise-wide
enterprise-widereporting
reportingcapability
capability
Student Notes
• Geographical Reasons
Separate cells may exist in each geographical location that the company has, for example,
different cities, states, or countries.
• Communications
The communications between the separate locations may be unreliable (WANs). Some
local control and management of backups is required.
The Pitfalls
While the implementation of multiple cells may be justifiable for any of the reasons
described above, it can cause many headaches:
• Difficult to Administer
Administration of multiple cells is more complicated. Each cell is autonomous and
designed to be managed by the cell manager of the same cell. For example, if a backup of
a particular system is to be performed:
− The administrator must first log on to the cell manager to which the system belongs.
− Redirect the display variable to his local system. (unless the cell console is registered)
− Then, start the GUI to perform the backup.
If a backup is to be performed on another system belonging to another cell, the process
must be repeated. In essence, it may be tedious to control all of the backups from a single
point. The same is true with backup specifications, schedules, templates, and monitoring.
When multiple cells are implemented, it is not possible to share a library between systems
in different cells, because each cell has its own media management database.
• Lack of Enterprise-Wide Reporting Capability
Reports can be generated only by the cell manager system. They can contain data relating
only to the systems that reside in the cell.
Features
M.O.M.
M.O.M. Features
Features
•• Centralized
Centralizedmanagement
managementof ofall
alltasks
tasks
•• Enterprise reporting
Enterprise reporting
•• Centralized
Centralizedlicensing
licensing
•• Distributed
Distributedcatalog
catalogdatabases
databases(CDB)
(CDB)
•• Centralized
Centralized media managementdatabase
media management database(CMMDB)
(CMMDB)
•• Sharing of library backup devices
Sharing of library backup devices
Student Notes
• Centralized Management of All Tasks
Centralized management of all tasks allows configuration, management, and control over
the enterprise backup environment of up to 50 Data Protector cells from a single point.
This includes backup configuration, media management, monitoring, and reporting of the
status of the whole environment.
• Centralized Media Management Database (CMMDB)
Optionally, all the cells in the environment can share a common, central database to
manage devices and media within the enterprise. This means that any device in a cell
using the CMMDB is available to all cells using the CMMDB.
• Shared Libraries
When using the centralized media management database, you can share high-end devices
among cells in a multi-cell environment. Allowing systems in different cells to share
expensive devices may better utilize them.
• Centralized Licensing
Centralized licensing allows one central location to administer the licenses for all backup
environments (cells).
• Distributed Data Protector Database
The database containing all the information on data backed up is split amongst each of
the cell managers in the environment. This lets you have multiple databases in your
enterprise backup environment, resulting in up to 50 times the single cell database space.
• Enterprise Reporting
The Data Protector Manager-of -Managers can generate reports on a single cell, as well as
for the entire enterprise environment. When the MoM environment is configured,
connecting to the MoM cell server from the Web interface will give access to multi-cell
reporting.
Concepts
Large Enterprise
Backup Environment
Student Notes
• One Data Protector cell manager acts as the Manager of Managers server.
With the Data Protector Manager-of-Manager concept, one cell acts as the manager cell and
the other cells run as client cells. The manager cell can be thought of as having a special cell
manager running. However, there are no additional processes installed on this system. The
Data Protector cell manager acting as the MoM cell manager can handle additional tasks,
share its media management database, and provide a new graphical user interface
(xomnimom/mom.exe) to show the view of multiple cells.
Configuration Steps
•• Choose
ChooseaaMoM MoMserver.
server.
•• Install
Install “Managerof
“Manager ofManager
ManagerExtension”
Extension”licenses.
licenses.
•• Configure MoM server.
Configure MoM server.
•• Create
Createaacommon
commonAdministrator
Administratoraccount
accounton onall
allcells.
cells.
•• Import cells to the MoM environment.
Import cells to the MoM environment.
•• Restart
RestartData
DataProtector
Protectorservices/daemons
services/daemonson onall
allcells.
cells.
Student Notes
Configuring the Data Protector Manager-of-Mangers (MoM) consists of the following steps:
2. Install the proper licenses on the Manager-of-Managers, as well as on every cell in the
MoM environment.
4. Create an Data Protector Administrator account that is common for all cells.
There are additional configuration tasks that you may want to perform to use some specific
MoM functionality:
• The system must already be an Data Protector cell manager, with the software installed.
See the HP OpenView Data Protector Installation and Licensing Guide or the
Installation module in this course for more information on how to configure the Data
Protector cell manager system.
Install Licenses
In order to activate the Data Protector MoM features, the Manager-of-Manager extension
license must be installed on each Data Protector cell manager that takes part in the MoM
environment
Configuring a MoM Manager
This section describes how to configure an HP-UX Data Protector cell manager as a MoM
Manager, manually and via the GUI.
touch /etc/opt/omni/cell/mom_info
3. Run the omnisv start command to start the Data Protector services.
4. Make sure the file has permissions as shown. It will contain the list of managed cell
managers. This file cannot be edited manually.
Note: Creating the empty mom_info file may be accomplished via the GUI:
6. Verify that the mom_server file has been created in the <OMNICONFIG>/cell
directory, and that it contains the FQDN of the MoM cell server.
MoM GUI
In
In addition
additionto
tothe
thestandard
standard cell
cell manager
managertasks,
tasks,
the
the MoM
MoMGUI GUI has
hasadditional
additionalfeatures:
features:
•• Import
Importandandexport
exportcells
cells
•• Move
Move clients betweencells
clients between cells
•• Distribute
Distributecommon
commonMoM MoMconfiguration
configurationfiles:
files:
–– User class specifications
User class specifications
–– Holidays
Holidaysfilefile
–– Global
Globaloptions
optionsfile
file
–– Vaulting
Vaulting locationsfile
locations file
Student Notes
In order to run the Data Protector MoM GUI, use the following procedure:
Add a MoM Administrator to Cells
Before you can use the MoM GUI to administer remote cells, you must have a user, belonging
to the Admin user group, in every cell in the environment. This user will be the MoM
administrator.
For example, you have a user called Mom_Admin in every Admin group on each cell in the
enterprise. This user will be the MoM administrator.
2. Add the MoM Administrator (Mom_Admin) to the Data Protector admin User Class.
(or another username of your choosing)
2. (Optional) You do not need to perform this step if the system is the MoM Server system.
It is for other systems from which you want to be able to run the GUI:
Create the following file in the <OMNICONFIG>/cell directory, containing the system
name of the MoM Manager:
Make sure the file has permissions as shown. If the client is to connect to the correct
MoM server (and not the local cell manager), you must manually create this file. This
would be necessary in those cases when the cell console is distributed amongst the
clients in the cell.
NOTE The mom_server file is created automatically on the cell manager system
when it is imported into the MoM environment.
2. If you are not the root/Administrator user of the imported cell, you will not have
permission to the remote cell, although you are registered as the mom_server. Add the
root/Administrator of the mom_server to the Admin group on the imported cell to gain
full permission to control the cell.
When you have configured the MoM environment, you will be notified to restart the Data
Protector daemons on each imported cell that uses CMMDB on the MoM Server.
The user must login in as root/Administrator and do the following:
Communication
Cell Manager/
Manager of Cell #1
Manager
Client Client
System System
Mom GUI
Cell Manager Cell #3
Cell Manager
Client Client
System System
Client Client
System System
Cell #2
Student Notes
The MoM GUI communicates directly with the cell manager systems that are part of the MoM
environment.
The MoM GUI is part of the standard, cell console package; therefore, it is possible that the
GUI can be run from any of the cell manager systems as long as configuration allows. In other
words, the MoM GUI is not restricted to the MoM cell manager system alone.
If the Data Protector device and media management databases are being shared, it is possible
for a client system to directly communicate with another cell manager system for non-MoM
GUI activity.
• If the MoM GUI is running from the MoM cell manager system. This is covered in the
client system and cell manager communication above since in this case the MoM cell
manager system is also acting as a client system.
• When central licensing is distributed from the Manager of Managers cell manager.
Cell #1
MoM Server
CDB
IDB
MMDB WAN
LAN
Cell #2
Cell #3
Cell Manager
Cell Manager
MMDB
CDB CDB
IDB IDB
MMDB MMDB
MMDB
MMDB
Student Notes
In a normal configuration, each Data Protector cell manager system maintains its own
internal database. The database is actually composed of two embedded parts:
• The Media Management database (MMDB), which stores information about backup
devices, their configuration, and media used for backup.
• The Catalog database (CDB) stores information about data backed up, such as files,
directories, and versions.
In a typical, cell-oriented environment, both parts are located on the cell manager system and
keep information on devices, media, and backup information for that cell. For security
reasons, it is impossible to access and use this data from another Data Protector cell.
Therefore, media and devices used in a cell cannot be accessed and used in some other cell,
without moving them to the other cell.
In larger multi-cell environments with high-end backup devices, you may want to share these
devices and media among several cells. Device sharing can be achieved by having one
centralized MMDB for all the cells and keeping an individual CDB for each cell. This allows
media and device sharing while preserving the security capabilities of the multi-cell structure.
Each cell is autonomous, in that only the cell manager of that cell maintains information
about the backups, devices, media, and catalog information.
When using Manager-of-Manager, this distributed model can be left in place (default), or the
media management databases can be merged and centralized.
Cell #1
MoM Server
CDB
IDB
WAN
CMMDB
LAN
Cell #2
Cell #3
Cell Manager
Cell Manager
CDB CDB
CMMDB IDB
IDB
Student Notes
The CMMDB on the MoM cell server allows you to share devices and media across several
cells in a MoM environment. This makes devices of one cell (using the CMMDB) accessible to
other cells that use the CMMDB.
Once the media management databases have been centralized, each cell is no longer
autonomous, as it must obtain information on devices and media from the central MoM
server's centralized media management database.
A reliable network connection is essential between the MoM cell and the other Data
Protector cells. Therefore, it is recommended only when using local LAN connections, but
not WAN.
Note that it is optional to centralize the Media Management database.
It may be highly desirable to centralize the media management databases if multiple cells are
being maintained on the same local site (connected via LAN). Typically, this is when multiple
cells have been implemented to overcome database capacity issues.
The implementation of the CMMDB is optional. If it is implemented, you can also choose on
which systems MMDBs will be centralized, which is useful when certain client cell managers
are connected to the MoM server over unreliable data links.
NOTE Each cell manager ALWAYS has its own Catalog database (CDB), where the
bulk of the Data Protector database resides.
NOTE The centralized MMDB has a big effect on licensing. Immediately after the
MMDB is changed from local to remote, all the licenses associated with
libraries and devices are validated from the MoM server. They can be removed
from client.
When CMMDB is used, Data Protector goes to the cell manager specified in the file
mmdb_server, located in following directory for all media requests:
On HP-UX: /etc/opt/omni/cell
Data Protector operation and the user interface see the MMDB in the same way.
Each medium that contains protected data has information showing which cell currently
owns the data. Once this protection has expired, any cell can reuse the medium. If a tape has
been initialized by one cell, any other cell can use it, as it does not have any protected data on
it. If a tape is loaded in a library and not yet initialized, any cell can initialize it, assuming
there is a loose policy, and no other tapes are available.
The media allocation rules apply in exactly the same way to shared tapes, except that
appendable media can be appended, only by the cell that owns it.
NOTE If you are configuring a new cell (and you do not yet have devices and media
configured), there is no need to merge the database. You want to merge cells
only with the CMMDB that already has devices and media.
NOTE Once you have configured the CMMDB and start using it, it is not possible to
split it back into local MMDBs. It may be possible that you recover the old
state of an MMDB, and import all the tapes since the merge occurred. It is
recommended that you start from scratch with a new MMDB, however.
Procedure
On the MoM, you must add one cell at a time to the CMMDB.
Data Protector cell managers in all cells must have the new version of Data Protector
installed and running.
1. Check that there are no backup, restore, or media management sessions running on
any of the cells to be added to the enterprise (multi-cell) environment.
a. /opt/omni/sbin/omnisv –stop
b. cd /var/opt/omni/db40/datafiles
c. cp –rp mmdb mmdb.before_mom
d. /opt/omni/sbin/omnisv –start
3. Before proceeding, you must have exclusive access to both the MoM database, as well
as the client cell server’s database. On the MoM manager, run the following command:
4. On the MoM server, edit the duplicated names of media pools and devices (in the user
interface). The duplicated names have a “ _N” appended to their name, where N
represents a number. This always happens to the default pools, if they exist in both
cells. In this case, you must manually change the backup specifications that use these
devices to use the new device names. It is a good idea to add a line to the media pool’s
description, to indicate from which cell the pool has come.
Repeat steps 3 and 4 for all the client cells you want to add to the CMMDB.
NOTE The mmdb_server file may be a copied from the mom_server file.
omnisv stop
omnisv start
omnicc -update_mom_server
Central Licensing
Student Notes
Data Protector allows you to configure centralized licensing for the whole MoM environment.
All licenses are installed and kept on the single MoM server system.
NOTE This means that the licenses must be generated by the HP password center
using the IP address of the MoM server system. Therefore, if you are
converting an existing environment from separate cells, and licensing to MoM
and centralized licensing, you must complete “License Move” requests and
submit them to the licensing center. See the “Installation and Licensing
Guide” for more information.
You can then allocate licenses to specific cells to suit your needs. Licenses are allocated for a
specific cell and are not floating licenses. Once a license has been allocated to a cell, it can
only be used in that cell. However, you can change the license at any time.
For example, suppose you have two cell managers in your enterprise environment. One is
named Aztec, and the other is named Mayan. The Aztec cell manager is a UNIX system with
four multi-drive servers in its cell. Aztec has four “ Multi-drive Server for UNIX” licenses.
The Mayan cell manager is another UNIX system with one multi-drive server in its cell. Thus,
Mayan has one “Multi-drive Server for UNIX” license.
The Aztec cell is reorganized, with most of the clients and three of the multi-drive servers
being transferred to the Mayan cell.
The Mayan cell now needs three more “ Multi-drive Server for UNIX” licenses, for a total of
four. Because these two cell managers are in an enterprise environment, configured with
centralized licensing, removing the licenses from one cell and adding them to another is a
relatively simple task that can be achieved via the MoM GUI.
NOTE The instant-on evaluation license does not permit MoM license configuration.
1. In the Manager-of –Managers GUI, switch to the Clients context, select the
MoM cell server, from the pop-up menu choose Configure Licensing to open
licensing configuration in the Results Area. In this window, you can see the current
licensing configuration for all cell managers in the multi-cell environment.
2. Select the server that has the licensing information you want to change. The types and
numbers of licenses available to your selected cell manager appear in the Results Area:
The USED column shows the number of licenses assigned to that particular server.
Increasing the number in this column will correspondingly decrease the number of
available licenses, and vice-versa.
The AVAILABLE column shows the number of licenses available to the entire
enterprise. This is the number of licenses not taken by any cell within the enterprise
environment.
The TOTAL column shows the total number of licenses, both used and available, in
the entire enterprise.
3. Select the desired license “Used” column, select to increase or decrease the value.
Giving up Licenses
To give up a license type, thus increasing the number available, reduce its
corresponding number in the USED column.
Getting Licenses
To get a license type, increase its corresponding number in the USED column.
Systems that are selected to use the remote, MoM managed licenses will receive a
configuration file called <OMNICONFIG>/cell/lic_server. The lic_server file will
contain the name of the MoM manager acting as the license server. This hostname in the
lic_server file must be the fully qualified name for the MoM server (this should occur by
default) if DNS is used for host name lookup.
When the Data Protector services start and find this lic_server, Data Protector will check
licenses with the MoM server every hour. In case of a communication problem, the last
licensing status is kept for 72 hours. If Data Protector does not find this file, licensing
information from the local <OMNIHOME/cell/ lic.dat file is used if available. The
lic.dat file is used to hold the license information when local licensing is used.
• It checks to see if the system is configured in any backup specifications, then leads you
through the steps to reconfigure backup of this system in the new cell.
• Checks if there are any devices configured on the system then leads you through the
steps to move devices to another system.
• Checks if there are media used in the devices on this system, then leads you through the
steps to move the media.
2. Select the client you want to move to another the cell manager.
3. Select Move Client System to Another cell from the pop-up menu in the
Scoping Pane.
2. In the Scoping Pane, select “Enterprise Clients,” from the pop-up menu select:
Distribute Configuration.
4. What configuration information can be distributed from the MoM GUI when using the
Distribute Configuration function?
5. When using a centralized media management database, the MoM system should be
highly available. Why?
8. When using MoM, the catalog databases can also be merged. TRUE/FALSE?
• Process tracing
• A description of common problems, such as network, user interface, devices, and starting
backup/restore sessions
Log Files
UNIX: /var/opt/omni/log
Windows: <OMNIHOME>/log
cleaning.log
debug.log
inet.log
IS_install.log
media.log
Valuable troubleshooting information omnisv.log
can be obtained by examining the RDS.log
Data Protector log files. sm.log
Upgrade.log
upgrade.log
trace.log
Student Notes
The directory in which Data Protector log files are kept depends on which operating system
you are using. The following table shows the directories where the log files can be found:
Unix
/var/opt/omni/log
Windows NT
All logs except the RDS.log file are stored in this directory:
<Data_Protectory_home>\log
Netware
SYS:\USR\OMNI\LOG
The following list shows the Data Protector log files and describes their contents:
cleaning.log This file records all Data Protector initiated tape-cleaning events. This
file is updated when Logical Devices have the Cleaning Tape option
enabled and they are in the clean-me state.
debug.log Unexpected conditions are logged into this file. While some can be
meaningful to the user, it is used mainly by the HP support
organization.
IS_install.log This file contains the trace of the remote installation and is located on
the installation server.
media.log This is a very important file. Each time a medium is used for backup,
initialized, or imported, a new entry is made to this log. Media that
contains the Data Protector database backup is also marked. For this
reason, media.log can be used after disaster recovery to find the
tape where that database was backed up and what media were used
after the database’s last backup.
omnisv.log This file is updated when the Data Protector services are stopped and
started.
RDS.log This is the log file of the Data Protector internal database.
sm.log This log file contains errors that occur in backup and restore sessions,
such as errors in parsing the backup specifications.
Upgrade.log This log is created during the upgrade and contains traces of the
(UNIX only) upgrade processing.
Online Help
Data Protector provides extensive help via an integrated help system. To start the help
system from within the GUI, select Help Æ Help Topics from the toolbar. On Windows, this
starts the integrated help, on HP-UX the Netscape browser will start.
Execution Tracing
Example:
Student Notes
Data Protector processes may be started in a special mode called the "debug" mode to allow
for extensive tracing of their execution. This execution tracing produces voluminous data
sets which may consume a significant amount of disk space; use with caution.
This is useful for the HP Support Center to assist you in troubleshooting situations with your
cell. The tracing produces a set of “debugs” that may be needed by HP Support.
To set the options for debugging using the Data Protector GUI, in the File menu, click
Preferences, and then click the Debug tab. Specify the debug options and restart the
GUI. The GUI will be restarted in the debug mode.
Debugging parameters for Data Protector integrations can be set using the OB2OPTS
environment variable. For more details about the OB2OPTS variable refer to the HP
OpenView Storage Data Protector Integration Guide.
Most all Data Protector commands can be started with an additional -debug
parameter:
Where:
1-99 is the debug range.
C:n is the size limit (KB) for the debug file to enable circular logging
T:s is the timestamp resolution, where 0 is off, 1 is seconds, 1000 is milliseconds.
POST is the debug postfix added to each file produced by the tracing
Host is the quoted list of hostnames where the debug is turned on.
The list of hostnames limits the hosts where the debug is turned on during execution
of the Data Protector command. If more than one host is on the list, they should be
delimited by spaces. The entire list must be put in quotes. For example:
"host1.domain host2.domain".
The debug postfix option is used for creating the trace file name.
Trace logging
Data Protector creates a log file called trace.log on the Cell Manager whenever tracing
is enabled. This trace log contains information about when and where debug traces
where generated within the cell.
And named:
OB2DBG_<DID>_[<SID>]_<Program_Name>_<Host>_<PID>_<Postfix
Where:
DID is the debug ID; this is the PID of the first process that accepts
the debug parameter; all debugs are “children” of this process
SID is the session id added by backup and restore agents (MA, DA)
Program Name is the program name of the Data Protector program writing the
trace
Host is the system name where the trace file is created
PID is the process ID.
POST is the postfix as specified in the -debug parameter.
SYS:\USR\OMNI\TMP\ProgTime.POST
Where:
to
NOTE If you enable INET debugs, all integrations will generate trace log files.
NOTE Execution traces can become very large. For this reason, use the -debug
option with care.
Discuss with the support organization what detail level should be applied and what
environmental conditions to use for the debug run.
• Prepare the environment for the debug run.
• Perform the debug run.
• Collect the debug information.
• Return your environment to normal operation.
The following steps show how to collect debug information for a typical issue, which
involves one client and the cell manager (example from the Admin Guide):
1. Discuss with the support organization the technical issue and request the following
information:
− Debug level (this is a command option needed later, such as "1-99")
− Debug scope (client only, cell manager only, every system)
− Collect also the standard debug logs
2. Try to reduce the error environment as much as possible, such as:
Create a backup specification that contains just one file or a few objects.
Have only ONE failing client involved in the debug run
On UNIX /tmp
On Windows \<OMNIHOME>\tmp
You should see only crs.pid and the mmd.ctx files written by Data Protector.
Hardware identification of the systems (cell manager, media agent and disk agent
system); examples: HP N-4000 Series; HP Netserver
Operating System information; examples: HP-UX 11.11; Windows 2000 Pro SP2
On the system where the (tape) devices are connected, execute the command:
devbra –DEV
Put the output in the info file. (Careful, output redirection might not work.)
5. Exit all OB2 GUIs and stop all other backup activities in the cell.
6. In case you need to collect CRS debugs as well (per support organization instruction) you
need to:
Stop the Data Protector services on the cell manager. (omnisv –stop)
Restart the services with debug.
This brings up the manager GUI and starts the debug mode. You can define the postfix
of the trace file names created by substituting the error_run text with your
preference.
10. Exit all OB2 GUIs (close the OB2 application). This cancels the debug mode for the Data
Protector manager.
11. Look at the /tmp directories (on the cell manager and on the clients). You will see
several ..._error_run.txt files.
12. Compress and pack ALL the ..._error_run.txt files and the info file using
compress, WINZIP, or TAR.
13. If the error includes the usage of a client system, you also need to collect the
..._error_run.txt files from the client and include them in the package.
In case the support organization asked you to collect the standard debug log files, collect
them as well.
14. Gzip, Shar/uuencode or otherwise mail the files to the support organization.
15. Give information about how you packed or compressed the file in the mail.
17. In case you need to collect CRS debugs as well (per support organization instruction),
you need to:
Process Overview
Which, When, and Where Processes Run
The following table shows which processes run where, and when, when Data Protector is
idle, during a backup, restore or media sessions.
Message Details
Student Notes
In case of difficulties during the operation, Data Protector provides additional information
with an interactive troubleshooting dialog. You can get a detailed explanation of messages
that occur within a running session by selecting the message ID number.
An example of the error message ID number format is: [x:y]. When displayed during a
session, the message number may be selected to reveal the troubleshooting utility dialog
window. The dialog window consists of four text fields:
Message Text You will see the message as displayed in the session.
The troubleshooting file is available only in the directory where the cell manager is installed.
It can be found at the following locations:
/opt/omni/gui/help/Trouble.txt on Unix
MESSAGE:
DESCRIPTION:
An internal error occurred. The process was not able to recover and aborted
ungracefully immediately after reporting this condition.
ACTION:
Before contacting your post-sales Data Protector Support
Representative, please gather as much information as possible:
* Write down product version and build number.
* Make a note of the circumstances that cause this error.
* Save session output to a file (e.g. session.txt).
* Collect all log files (*.log) in DATA PROTECTORHOME/log directories on all
hosts involved in the situation when this error occurred (i.e. host running
VBDA, host running BMA and host running BSM).
Network Connectivity
Cell Console
Disk Agent
Disk Agent TCP/IP TCP/IP TCP/IP
Cell Manager
Student Notes
A very common problem in a Data Protector environment is host name resolution. This
means that host A is unable to communicate with host B.
The table below shows Data Protector components and how they should communicate
among one another within the Data Protector environment.
Communication among hosts means that host A in the table should resolve host B by its fully
qualified domain name (FQDN). Resolving a host means that host A can interpret the FQDN
and determine its IP address.
Example One
For example, we have host A (nevizex1.domain.com) and host B
(nevizex2.domain.com). Host A is a client with a disk agent, and host B is the cell
manager with a media agent.
If host A properly communicates with host B, you should get a response from host B.
This response consists of host B’s IP address and confirmation on messages being
sent to it.
C:\users>ping active.domain.com
Pinging active.domain.com [10.10.10.10] with 32 bytes of data:
Request timed out.
Request timed out.
Request timed out.
Request timed out.
C:\users>
To resolve this problem, check for and resolve possible network problems on the
remote host.
/usr/sbin/ping active.domain.com
PING active.domain.com: 64 byte packets
Name Resolution
DNS Resolution Problems
Windows NT If you encounter resolution problems when using DNS on
Windows NT, enter the DNS Server in the Microsoft TCP/IP
Properties window
(Control Panel,--> Network-->Protocols--> TCP/IP Protocol-->
Properties-->DNS)
nslookup active.domain.com
or
nsquery hosts active.domain.com
(nsquery is in /usr/contrib/bin)
If the target system was not defined on the DNS, then the response would be similar
to the one below:
On HP-UX, verify the lookup service that you are using by viewing the contents of the
/etc/nsswitch.conf file. (dns, files, nis, nis+)
Test the lookup service using the nslookup or nsquery commands, for example:
nslookup <client_name.domain>
or
nsquery hosts <client_name.domain>
If you are using DNS, and the nameserver is not reachable, you may still be able to
use /etc/hosts as a fallback service; this is configured with the /etc/nsswitch.conf.
TIP You may also be able to configure the cell manager so that it may operate even
in those cases when the DNS server is unavailable. The following files will
need to be modified to allow the cell to operate when the nameserver is
unavailable.
/etc/opt/omni/cell/cell_manager contains the cell manager host name
Services
Windows
Student Notes
Services are critical components on Data Protector systems. They are used for
communication of Data Protector components and unattended systems tasks.
To test if the Data Protector inet service is running on a remote system, by using the telnet
command and the Data Protector inet service port number; use the following command:
If you get a response similar to the one below in the telnet window, the Data Protector inet
service is installed and running/responding on the remote system:
If you get a response, connection refused, you know there are some problems with the
current installation. Possible problems include:
For more details on service problems, see “ Problems Starting Data Protector Services”
below.
The Data Protector Inet program (/opt/omni/lbin/inet) is started by the system inet
daemon when an application tries to connect to the Data Protector port, which, by default, is
port 5555.
Normally, these daemons are started automatically during the system’s startup.
To manually stop, start, and get the status of Data Protector daemons, log on to the cell
manager system as superuser.
Stopping Daemons
To stop the Data Protector daemons, enter the following command in the /opt/omni/sbin
directory:
omnisv -stop
Starting Daemons
To start the Data Protector daemons, enter the following command in the /opt/omni/sbin
directory on Unix, <OMNIHOME>\bin on Windows:
omnisv -start
omnisv -status
Possible reasons why a Data Protector daemon may fail to start include:
/opt/omni/sbin/omnisv start
Could not start Raima Velocis server daemon. See
/var//opt/omni/log//RDS.log for details.
Check that you have all Data Protector database files in the <OMNIVAR>/db40 directory.
Compare the list of files in the <OMNIHOME>/newconfig/var/opt/omni/db40 on Unix
and <OMNIHOME/newconfig/db40 on Windows to the list of files in the
<OMNIVAR>/db40 directory.
One possible reason for this message is that the volume where the Data Protector internal
database resides is not mounted.
You would most probably get this message if the global options file were missing. Verify that
the <OMNICONFIG>/options/global file exists.
1. Check that the database server is really not running using the following command:
omnisv -status
If the database server is not running, start it with the following command:
omnisv -start
Otherwise, continue with the next step. If the server cannot be started, see: "Raima
Velocis server daemon could not be started" above.
3. For recovery, follow the steps in “ Recovering the Complete Database” in the
Administrators Guide.
There are many possible reasons why Data Protector services may fail to start:
• "You do not have permissions to start the services."
If you get this error message when starting services, you do not have permission to start
and stop services. Ask your system administrator to grant you the permission to start,
stop, and modify services on the system that you are administering. In this case, log off
and log on again for the changes to take effect. You can also log off the system and log on
as system administrator, then start/stop or modify the services. If the service still fails to
start, try one of the other options in this list.
• The service user does not have permissions to start the service
If you get the following error message “Service Start-up failed due to logon
failure,” it means that the user who is configured to run the services does not have
permission to start/stop services. Ask your system administrator to grant this user
permission to start, stop, and modify services on the system you are administering. If the
service still fails to start, try one of the other options in this list.
First, uninstall the current installation, either on the client system or on the server
system. If this is a server system, copy the Data Protector\db40 and Data
Protector\config directories to a safe location before uninstalling Data Protector;
then reinstall the software.
This will give you a clean installation and all the binaries of the services and other
binaries will be in place. If the service still fails to start, try one of the other options in
this list.
• Database log files are corrupted and the MMD crashes when starting the CRS
service.
Because of improper database shutdowns, some database log files can be corrupted, and
the Data Protector CRS service can fail to start. Also, the MBD.exe file invokes a Dr.
Watson diagnosis. This points to a corruption in the database log files.
Delete the mmd.ctx file on the \Data Protector\tmp directory to resolve the
problems. Then restart the services.
Port Usage
In general, the communication between Data Protector components, which are distributed
over several systems on a LAN, uses a dynamic port assignment. This means that you cannot
predict which ports are used.
However, the communication between the graphical user interface and the Cell Manager can
be restricted with respect to the ports being used.
INET port This is the port used to initiate any communication between Data Protector
components. It is, by default, port 5555, which can be modified to a different
number. See the HP OpenView Storage Data Protector Installation and
Licensing Guide for information on how to modify this number.
CRS port This port is the only port used by the GUI to communicate to the cell
manager, as well as the INET port. No other system should be using this port
if the Data Protector graphical user interface must use it.
NOTE The two ports cannot be mapped to the same number. Follow the procedure
below to configure the CRS port.
1. Change the CrsPort entry in the GLOBAL file, and specify the port number to use.
To stop Data Protector Inet service on a Windows 95/98 Client system, use the <Data
Protector_home>\bin\omnii95 -kill command.
• If communication between the systems is not the problem, check the installation, using
the telnet command.
• If the installation is okay, check that all services are running properly on the cell
manager.
• If you are able to run the GUI, but have no access, your DNS server may be down, and
hostname resolution is falling back to local files. This fallback mechanism is a common
problem if the local host lookup file does not contain fully qualified domain names.
To remedy this problem, you may want to add an additional user to the UserList
configuration file. Be sure to have an entry for the cell manager that uses the fully
qualified name, as well as the short name for the hostname. This should be considered a
short-term fix to be able to access the GUI.
• Access Denied message when starting the GUI on a local or remote system: This event
may be similar to the above DNS problem, or simply your user is not authorized to access
the Cell Manager. To ensure that you are able to start the GUI from any system (for
testing) add the following entry to the <OMNICONFIG>\users\UserList file on the Call
Manager:
* * * * Admin
The above entry is considered to be a security risk, and thus should not be left in place
after testing is completed. If you are able to connect from the remote GUI, then go to the
users context and refine the access limits for the cell.
in the background. Settings that are modified in the GUI are saved in the home directory in a
file named “.windu.<hostname>.” This file is not modifiable, but may be removed if the GUI is
not running to reset to the Data Protector installation defaults.
If the server processes (crs, mmd, rds) are stopped while the GUI is running, and then the
GUI is stopped, you will see “Wind/U Error (251); these errors occurring as a result of the
loss of communication between the windu_registryd and the crs processes. This should not
cause any lasting problems, but the messages will look very ominous. The messages may be
safely disregarded, and the GUI may be restarted after all the servers are running again.
Backup Devices
Student Notes
General Troubleshooting
When you have a device problem, you may jump to the conclusion that it is a Data Protector
configuration problem. The best practice is to eliminate Data Protector as the source by
accessing the device with another native backup utility. For example on HP-UX, try tar,
fbackup, cpio, etc; on Windows try NTbackup. If they do not work, the problem is not
with Data Protector. However, if they do, the problem may be with Data Protector, and
further investigation is required.
Can the system see the device? On HP-UX, use the ioscan –fnCtape command to verify
connectivity and device files. On Windows use the Data Protector command: devbra –
devices.
Supported Devices (SCSITAB)
Data protector provides support for devices of many types from HP and other vendors. On
the Cell Manager is a file named “scsitab” which provides a list of all supported models.
Periodically new devices are added to this list by HP. Download the most recent scsitab from
HP to add support for the newest devices. Do not modify this file manually.
Media Problems
Is the media bad, does the operation work with other media? Use the Data Protector verify
function to verify existing backups.
Library Devices
The most common error message for an improper configuration is, “Cannot access
exchanger control device.” This implies a problem with the robotic control device
file. Refer to the Appendix for detailed information on how to properly configure a library
device.
1. Check the drive indexes as they appear on the library’s control panel.
2. Verify that they match the drive index number assigned to them in Data Protector.
4. After configuring each drive, verify that it is configured correctly: Select the drive, and
click Scan from the Actions menu.
UMA is packaged and installed as part of OB2's media agent fileset. If you have received uma
as a standalone program, or if you run it on a system on which Data Protector has not been
installed, uma will be fully functional and behave as documented, but it will probably not be
able to locate and use Data Protector's NLS message catalog, omni.cat.
On HP-UX systems, uma is located in /opt/omni/lbin/uma, and the Data Protector NLS
message catalog is in /opt/omni/lib/nls/C/omni.cat.
Uma can be started both interactively or in batch mode. The only option that needs to be
specified is the pathname of the device file that controls the robotics of the target
autochanger, this is specified as –ioctl <device name>.
For your convenience, uma allows you to specify symbolic instead of physical element
addresses (slot IDs). When you need to refer to the 1st drive of the autochanger, you can
specify the physical address, "128," or the more convenient, symbolic "D1." The output of the
addr command reflects the duality of this addressing convention.
Syntax
uma version
uma -ioctl deviceFile –[barcode] -tty
Commands:
help
inq
res
rel
init
addr
offl drive
sense
pos slot
stat
Options
-version Displays the version of the uma command.
-ioctl deviceFile Specifies the pathname of the device file that controls the robotics
of the target autochanger.
Examples:
echo stat | C:\Program Files\Omniback\bin\uma –ioctl <dev> –tty
print stat | /opt/omni/lbin/uma –ioctl <dev> -tty
Commands
help Displays the usage synopsis for the uma command
inq The inq command performs an SCSI-II inquiry operation on the
UNIX device file specified with the -ioctl option. It returns
the device's type, vendor ID, product ID, and firmware revision
number.
res Reserve control device file for exclusive access. This command
puts an exclusive write lock on the device file, which prevents
other programs from accessing it until it is released (or the
current uma process terminates). It requires no additional
arguments.
rel Release control device file. Reverses the effect caused by a res
command. It requires no additional arguments.
init The init command performs an SCSI-II "initialize element
status" operation, which (if applied to an autochanger robotic
device) forces the autochanger to reset its internal state and
perform an inventory of its repository. This command should
not be used if another process is accessing the autochanger at
the same time. The effects will be unpredictable.
addr Query and print the autochanger's element assignment page.
Each addressable item inside the autochanger mechanism
(drive, repository slot, robotic arm, import/export slot) has a
unique integer number (slotID), which can be used to address
this specific item. As the element assignment differs between
different autochangers, the software that controls the
movement of media inside the autochanger must find out and
use these numbers to perform move, pos, and stat
operations (see below).
The following output is obtained by executing the addr command on the UNIX device file,
referring to an ACL 4/52 DLT autochanger:
/dev/spt/sctl0> addr
Element Addresses:
The numbers returned by the addr command are the physical element addresses of different
elements within the autochanger ( i.e., element address "256" corresponds to the 1st
repository slot, element address "65" corresponds to the location of the 2nd data drive, etc.
offl This command can be used only if at least one drive was specified at uma -dev
option. If a medium is loaded in the specified drive, it will eject the medium, the
same as if an mt offl command was specified. The mandatory argument is a
symbolic drive ID (i.e., D3 for the 3rd drive == the 3rd device file specified with the
-dev option).
sense Read the device's sense data and dump them in hex-dump format. This command is
not directly useful to the user, but can be used by developers and support engineers
to diagnose potential device problems. No additional arguments are required.
pos The pos command positions the autochanger transport mechanism in front of the
specified slot ID. This operation is meaningful only if the specified slot ID refers to
an import/export, data drive, or repository element. The actual meaning of this
operation may differ among different autochanger models. This command is
generally not required, but is provided for testing purposes and convenience. Both
physical and symbolic slot addressing may be used.
move Move a medium from a source slot into a destination slot. This command has two
mandatory arguments, the source and destination slot IDs (address numbers, as
reported by the addr command described above) and an optional numeric Boolean
argument that can be used to instruct the robotics to flip the medium before
inserting it into the destination slot. By default (if no flipping argument is specified),
flipping is disabled.
NOTE Flipping is supported only for double-sided optical media. For tapes, the effect
of a flip command is not defined.
NOTE Most autochangers do not allow you to move a tape from a drive to a
repository location, if the tape has not been dismounted and ejected by the
drive. Use the offl command on the drive device file to put the drive off-line
before executing the move command.
stat This command queries the device for information about the state of each of its
addressable elements. The output of this command is a table of physical and
symbolic element IDs and their states, indicating which elements are free (Empty)
and occupied (Full). Additionally, if barcode support is available and enabled, the
barcode for each medium is printed. Uma recognizes one specific environment
variable, which can be used to enable barcode support for autochangers that are
equipped with barcode reading hardware. By default, uma OB2BARCODE=1 into
uma.
The stat command can be used to query the status of a specific slot (i.e., stat 290
or stat S35) or a related group of slots (i.e., stat D will query all drives, stat S will
query all repository slots, etc.).
If no additional arguments are specified, the stat command will query and print the
status information for all slot IDs it can address.
Examples
Uma can be started both interactively or in batch mode. The only option that must be
specified is the pathname of the device file that controls the robotics of the target
autochanger:
uma -ioctl /dev/spt/sctl0
/dev/spt/sctl0> inq
SCSI Inquiry:
Type: 8
To let uma execute a batch script of its own commands, simply redirect its stdin to a file
containing a list of uma commands, separated with new lines:
cat >/tmp/cmdFile
inq
addr
stat
<ctrl-D>
To list all physical devices that are configured on the system, run the
<OMNIHOME>\bin\devbra -dev command from the Data Protector command line
interface on the system on which the devices are located, . If any of the SCSI addresses have
a CLAIMED status, they are in use by another device driver.
To resolve the problem, click on Control Panel, then Tape Devices to open the Tape Devices
dialog box. Remove configured tape devices. Reboot the system.
Other common problems are hardware-related. They are most commonly resolved by
checking the SCSI communication between the system and the device (SCSI cables, adapters,
length of cables, and so on).
Student Notes
To resolve this issue, you can configure a session owner for the backup in the Advanced
Backup Specification options. The backup owner should be a user from the Admin
class. This will make this user owner of all backups, regardless of who actually starts the
backup session.
On Standalone Devices
If a mount request is issued for a standalone device and media is available in the device, the
following is possible:
• The media in the device is in a media pool that has a non- appendable policy.
In this case, although there is still available space on the media, the media will not be
used because of the non-appendable policy of the pool. To append backups to media,
change the media pool policy to Appendable to append backups until the media is
FULL..
• The media in the device is not formatted, and the media pool to be used has a strict
policy.
If your pool is defined to use a strict media allocation policy, media that is not formatted
will not be used for backup. If no formatted media is available, Data Protector issues a
mount request. If you would like Data Protector to automatically format unformatted
media, set the media pool policy to Loose for media available in the device.
• The media in the device is formatted, but is different than the one in the Prealoc List of
the Backup Specification, and the Pool specified has a strict policy.
If you are using a Prealoc list of media in combination with the strict media policy, the
exact media specified in the Prealoc list must be available in the device when a backup is
started. If the exact media is not available, a mount request is issued.
To use another media (if it is available in the device), in combination with the Prealoc
list, change the media pool allocation policy to Loose. To use any available media in the
device without the Prealoc list in the Backup Specification, remove the Prealoc list
in the Properties of the Backup Specification.
On Library Devices
If a mount request is issued for a library device, and media is available in the library, the
following is possible:
• The media in the library is not formatted and the media pool that the device is in has a
strict policy.
If your pool is defined to use a strict media allocation policy, media that is not formatted
will not be used for backup. If no formatted media is available in the library, Data
Protector will issue a mount request. If you want Data Protector to automatically format
unformatted media, (if it is available in the Library), set the media pool policy to Loose.
This can be modified in the media pool properties.
• The media in the library is formatted, but is different than the one in the Prealloc list of
the backup specification, and the pool specified has a strict policy.
If you are using a Prealloc list of media in combination with the strict media policy, the
exact media specified in the Prealloc list must be available in the device when backup is
started. If the exact media is not available, a mount request is issued. To use another
media (if available in the device) in combination with the Prealloc list, change the media
pool allocation policy to Loose. To use any available media in the device without the
Prealloc list in the backup specification, remove the Prealloc list in the Properties of
the Backup Specification
If the following line does not appear, stop and start the Data Protector daemons by
entering the omnisv.sh stop and the omnisv.sh start commands in the
/opt/omni/sbin directory.
To resolve this issue, you will need to request new licenses and apply them to the Data
Protector system. See the HP OpenView Data Protector Licensing and Installation
Guide for licensing details.
The solution is to configure the protection for your full backups so that they are protected for
longer than your incremental backups. The time difference between the protection for the
FULL backup and the incremental backup should be the amount of time between the FULL
backup and the last incremental backup before the next full backup. For example, if you run
incremental backups Monday through Friday and FULL backups on Saturday, you should set
the protection of the FULL backup to six days more than the incremental backups. This will
keep your FULL backup protected and available until your last incremental backup expires.
Windows/Unix NT Solution
In Windows NT, set the protection as follows:
2. Select the Backup Specification used to back up the data you want protected.
When backing up the database on a system, the agent that starts on the system logs the
system’s name to the database as system.domain.com.
When a restore is performed, the restore session manager wants to restore to the
system_name.domain.com, but it cannot, because it does not know this system as
system_name.domain.com, only as system_name. It cannot expand the system to the
long name, because its DNS is improperly configured. This situation can also be the other
way around, where DNS is configured on the cell manager and not on the application client.
The Data Protector Inet service on the remote computer is running under a user account that
does not have access to the Data Protector share on the installation server computer, which
is most probably a local user. You should change the username for the Data Protector Inet
service with access to the Data Protector share.
Reason
This problem can occur if the ActiveX control hhctrl.ocx used by Data Protector is
replaced with a different version during installation of another software application.
Solution
Rename or delete the file %SystemRoot%\system32\hhctrl.ocx and restart Data
Protector Manager or Data Protector MoM Manager. The application reports that the Data
Protector Help system is not installed, and will install it. This will reinstall the original
version of the hhctrl.ocx control.
omnihealthcheck
UX: /opt/omni/sbin
Windows: <Data Protector_home>\bin>
Student Notes
The omnihealthcheck command reads a configuration file and automatically and
periodically checks the status of Data Protector services.
Each line in the file is treated as a command line and is executed. Commands must
be listed with full pathnames except if they are Data Protector commands located in
bin (Windows) or sbin (UNIX) directories.
HealthCheckConfig File
UX: /etc/opt/omni/HealthCheckConfig
Windows: <Data Protector_home>/Config/HealthCheckConfig
Student Notes
The omnihealthcheck config file may be modified to include additional checks to the
defaults provided; operating system commands may be used.
This is the text file describing which commands must run when omnihealthcheck
command is used. omnihealthcheck can be used for monitoring reliability and conditions
of Data Protector cell.
If you want to use other non-Data Protector commands, then full path must be used to run a
command. Commands used in HealthCheckConfig file run under administrator / root account
during omnihealthcheck command execution.
On Windows, commands are executed under account, which NT Data Protector CRS service
run., with may be different than the administrator.
Running omnihealthcheck with default HealthCheckConfig does not have any impact on
the performance of Cell manager. Also, there is almost no impact on running backup / restore
sessions.
omnihealthcheck Log
UX: /var/opt/omni/log/HealthCheck.log
Windows: <Data Protector_Home>/Log/HealthCheck.log
Student Notes
This text file HealthCheck.log is appended during omnihealthcheck command, together with
information about each separated command used in omnihealthcheck configuration file.
When the omnihealthcheck command is first started, the HealthCheck.log file is created.
If all commands are executed without any error then the Healthcheck exit code is 0 (no
errors).
EXIT CODES:
0 All commands listed were executed and their exit codes
1 At least one command could not be executed or completed with non
zero exit code
2 Config file could not be read
3 Command timed out (depending on Timeout variable)
omnitrig -run_checks
UX: /opt/omni/sbin
Windows: <Data Protector_home>\bin>
Event Configured
================================
UserCheckFailed -
DbSpaceLow Yes/1
NotEnoughFreeMedia Yes/1
UnexpectedEvents Yes/1
HealthCheckFailed Yes/1
DbPurgeNeeded Yes/1
LicenseWillExpire Yes/1
Student Notes
The omnitrig -run_checks command may be used to start checks for the following Data
Protector notifications:
By default omnitrig runs once per day at midnight (can be changed in global options
file, variable: DailyMaintenanceTime)
16–13. TEXT PAGE: Debugging UNIX Pre- and Post- exec Scripts
First a little understanding of how Data Protector starts an exec script and the limitations.
The exec scripts are stated using popen() for object exec scripts and fork() and
execvp() for session exec scripts. See the man page for more details on these system and
function calls.
When an object exec script is started, it will show up in the ps output as "sh -c
script_name.” When a session exec script is started it will show up in the ps output as "sh
script_name". The owner specified in the backup specification will own the process for the
script. Data Protector sets the SESSIONKEY, PREVIEW, SESSIONID, SMEXIT and OWNER
environment variables for a session exec script and sets PREVIEW and BDACC for an object
exec script. See the Administrator's Guide under the heading “Backing up Data - Pre- and
Post- Exec Commands “ for more details on these variables.
After the script is started, Data Protector monitors the output of a pipe attached to the
stdout of the script. stdout of ALL processes started by the script are also attached to this
pipe unless otherwise explicitly closed by the process or script. The function feof()
monitors the pipe looking for non-zero, feof() will not return non-zero until ALL processes
that have stdout attached to the pipe Data Protector is reading, close their file descriptor. If
the script forks new processes that have stdout attached to the pipe and they are still running
when the script is completed, feof() will not return non-zero until these processes
complete or close the file descriptor.
This is the case of processes launched in the background. To demonstrate this launch a
"sleep 300 &" in an exec script. You will notice the script completes but Data Protector will
not complete until the sleep is finished. The reason is that there are still file descriptors
attached to the pipe. The best way to prevent this is before the fork, the process does a set
process group and so becomes a process leader (detaches). This is important because Data
Protector monitors the data coming from the pipe. If no data is passed through the pipe over
a global file defined timeout, Data Protector will report this timeout and consider this an
error. See the global file for the ScriptOutputTimeout variable for session exec scripts
and SmDaIdleTimeout, which can affect object exec scripts.
See below for a sample program, which shows how Data Protector starts exec scripts. Other
ways to work around this are:
1. In the script, use at(1) to start commands which normally keep the pipe open.
2. Redirect the output and stdout to a file when starting a command, which normally
keeps the pipe open. You can then cat this file in the end of the script to report anything
that has been output to the file.
An issue with the pipe monitoring is, Data Protector only captures the output of stdout.
stdout is not captured or reported. It is up to the designer of the script to allow for
capturing of data sent to stdout.
Data Protector checks the return of waitpid() for session exec scripts and pclose() for
object exec scripts that monitor the return of the script. If the script returns a non-zero value
upon completion, this will be considered an error. This is documented in the Administrator's
Guide in the section for "Advanced Tasks and Concepts."
Data Protector's function is to start the script with the correct owner, set the proper
environment variables and capture the return. How the script functions is totally the
responsibility of the designer of the script. It is completely the responsibility of the script
designer to set any environment necessary for the script Data Protector does not set, to
handle error conditions within the script and exit with the desired exit code. If the script is
started with the correct ownership and Data Protector environment variables but does not
function as expected this is NOT a Data Protector supported issue.
If problems arise with the script, the main tool available is to add the following lines after the
first line in the script:
set -x
exec 2>&1
These lines put the shell into debug mode. The commands are echoed to stdout. The exec
line redirects stdout to the same place as stdout. Since Data Protector is capturing
stdout, you will be able to see the commands as they are executed. Anything the commands
send to stdout unless your script redirects it somewhere else will also be captured in the
session output.
Other ideas are to execute the ID command to confirm the ownership is correct. Execute the
env command to ensure your environment is correct. Confirm nothing is hung on an
attempted read from stdin. Another idea is to confirm the script works when run from cron
as the owner of the backup specification or barlist. Except for Data Protector setting up the
environment for some Data Protector specific variables, the environment is very similar. The
script can be modified as a test to explicitly set the Data Protector environment variables
required by the script. If the script does not work from cron, it is unlikely to work from Data
Protector. If the script does run from cron, this does NOT mean it this is a Data Protector
problem.
Data Protector does not support the use of “su” from an exec script. Also, su may alter the
environment Data Protector sets up to run the script. HP will not debug a customer exec
script. We can help in assuring the script is started and with the correct ownership and Data
Protector environment variables set. We can help if Data Protector is not detecting the return
correctly. Anything else is consulting and beyond the scope of what HP is able to provide.
Read and understand the section for “Advanced Tasks and Concepts” in the “Administrator's
Guide,” when working with exec scripts. If the following test scripts echo the correct
information, you have confirmed that Data Protector is not the problem. You can also change
the exit values to confirm Data Protector is checking the return code properly.
Customizing
Student Notes
In most situations, the Data Protector default configuration and options are adequate for
everyone. However, many options can be changed that affect the appearance and behavior of
the product.
/etc/opt/omni/options/global
This file may be modified whenever the need to affect the options in the file is necessary. The
options file contains many of the Data Protector defaults, but is only used if the items are
uncommented. Each option currently in the file has a hash mark, or pound sign (#), which
comments out the option. This means that it does not affect Data Protector.
How To Use Global Options
To use a global option, uncomment the line that has the option name and set appropriate
value. To uncomment a line, simply remove the ‘#’ mark. Average users should be able to
operate the product without changing them.
Commonly Used Variables
The following list includes the most often used global variables. See the Global Options file
for a complete description.
• MediaView: Change the fields and their order in the Media Management context.
• MaxBSessions: Increase the default limit of five concurrent backups
• InitOnLoosePolicy: Prevents Data Protector from automatically initializing blank or
unknown tapes on under a loose media policy.
• MaxMAperSM: Increases the default limit of five concurrent devices per backup session.
• SmWaitForDevice: The amount of time spent waiting for a device; this is the backup
queuing time. (default is 60 minutes)
• ExecScriptOnPreview: Determines if pre/post execs are executed during the preview
mode. (default is 0, off)
• ScriptOutputTimeout: The amount of time that the SM will wait for a pre/post exec
script to complete. (default is 15 minutes)
# default: 0
# If set to 1 and MIN value of load balancing is greater than
# number of devices used in backup session the
# session aborts.
# Port=InetPortNumber
# default: 5555
# Data Protector inet process listen port. If the Data Protector
# default port is already used by some other product you should
# change this. Enter the new port number in the
# /tmp/omni_tmp/socket.dat file before installing the cell
# server. Cell server installation scripts will also update
# your global file.
# CrsPort=CrsPortNumber
# LogCrsEvents=0 or 1
# default: 0
# If set to 1, CRS creates <Data_Protector_home>/log/crsevents.log
file
# where it logs all CRS events along with the date, time,
# hostname and user information. NOTE: it is up to the Data Protector
# administrator to delete the file if it grows too large and
# occupies too much disk space.
# EventLogMessages=0 or 1
# default: 0
# If set to 1, the following Data Protector events are logged to a Windows
# NT EventLog:
# -start/stop Cell Request Server
# -start/stop Session Manager
# -device mount request
# -device error
# -all MAJOR/CRITICAL session messages
# Note that this is Data Protector NT specific option and it logs
# events only on cell manager computer.
# DBMaxFilesFound=NumberOfFiles
# default: 500
# This limit is used by the Data Protector file pattern search
# function. Data Protector will stop searching for files when more
# than <NumberOfFiles> files matches the specified search
# pattern.
# DBMaxFilesInDir=NumberOfFilesInDirectory
# default: 2000000
# This limit is used by the Data Protector database. It prevents
# endless loops if the database is corrupted. Increase the
# number if a directory with more than <NumberOfFilesInDirectory>
# files exists, otherwise Data Protector reports the database is
# corrupted.
# DBPurgeSuspension=0 or 1
# default: 1
# If this option is set to 1 (default), a purge session will be
# suspended whenever a backup/restore session is running.
# If this option is set to 0, a purge session will continue in
# parallel with backup/restore sessions. This influences
# backup performance as if there was another backup/restore
# session running. Additionally, if a database backup is
# started, the purge session creates a number of transactions
# stored in the online backup transactions file. This affects
# the size of the file and a time needed to restore the
# transactions from the online backup transactions file to the
# database when the backup is finished.
# DBPurgeSuspensionDuringDBCheck=0 or 1
# default: 1
# If this option is set (=1), purge session manager
# ("PSM") will suspend its execution while database
# check is in progress.
# DBFreeDiskSpace=MinSpaceInMBytes
# default: 50
# This option is used for checking free disk space on disks
# where Data Protector internal database is located. If free
# disk space is lower than this number, special event will be
# triggered (DbSpaceLow). This is a standard notification
# event and can be used to trigger predefined report or action.
# DBFreeExtFileSpace=MinSpaceInMBytes
# default: 250
# This option is used for checking free extension file space.
# If fvers.dat and fnames.dat (base and extension files) have
# less then specified number free space, Data Protector will
# trigger special event (DbSpaceLow).
# BrowseHistoryStart=NumberOfDays
# default: 0 (disabled)
# If "Search interval" in GUI is set to "None", then global option
# BrowseHistoryStart, if set, overrides it to whatever value
# specified in BrowseHistoryStart. Option is meant to prevent
# CU with large history from selecting "None" by accident and
# thus slowing down browsing. Even if this global option is set,
# CU can still browse history before BrowseHistoryStart by
# specifying in GUI search interval larger than BrowseHistoryStart.
# BrowseMPosCache=NumberOfMegabytes
# default: 40
# This option specifies upper limit of memory used by DBSM when
# browsing Detail Catalog. Specifying 0 disables cache all
# together.
# DCDirAllocation=0, 1, 2
# default: 0
# This global option controls which algorithm will be used to
# select the directory for the creation of the new DCBF file.
# 0 - Fill in sequence
# 1 - Balance size
# 2 - Balance number
# MaxDCDirs=NumberOfDirectories
# default: 10
# minimum: 1
# maximum: 50
# This option specifies maximum number of configured DCBF
# Directories.
# KeepObsoleteSessions=NumberOfDays
# default: 30
# This global option controls how many days obsolete
# sessions (restore sessions, backup sessions without any
# valid media, and media management sessions) are kept in the
# database.
# KeepMessages=NumberOfDays
# default: 0
# This global option controls how many days the session messages
# are kept in the database (if the session is not obsoleted and
# removed sooner). If the value is 0 then the session message
# are not removed before session.
# KeepDiskonlySessions=NumberOfDays
# default: 365
# This global option controls how many days the diskonly
# backup sessions are kept in the database.
# RepositionWithinRestoredObject=0 or 1
# default: 1
# This option specifies if restore should use repositioning and
# multiple disk agents to restore trees and files within one
# object.
# DailyMaintenanceTime=HH:MM
# default: 12:00
# This option is used for starting daily maintenance tasks at
# first omnitrig run after the specified time each day. Valid
# values are hour:minute, using the twenty-four hour clock notation.
# DailyCheckTime=HH:MM
# default: 12:30
# This option is used for starting daily checks at first
# omnitrig run after the specified time each day. Valid values
# are hour:minute, using the twenty-four hour clock notation.
# Specifying 'None' disables starting daily checks.
# RecoveryIndexDir=FullPathToTheBackupDir
# default: none
# This option sets backup directory for the obrindex.dat file.
# If this directory pathname is writable, the recovery index
# will be created/appended into this directory in addition to
# the default recovery index file.
# SessionMessagesDir=FullPathToTheMessageDir
# BackupPreexecBeforeDevLock=0 or 1
# default: 0
# If this option is set (=1), Backup Session Manager (BSM)
# executes session pre-exec before successfully locking needed
# devices. Otherwise, session pre-exec is started after device
# lock.
# PrePostExecOnEveryVolumeForHostBackup=0 or 1
# default: 0
# If this option is set (=1), Backup Disk Agent (BDA) for every
# volume of host object will execute pre/post-exec specified.
# If this option is not set (=0), Backup Session Manager (BSM)
# will execute pre-exec just prior to starting first volume BDA
# and post-exec after last volume BDA finished. Note that for
# the duration of pre/post-exec scripts, BSM will not be
# respondable.
# BackupCatalogProtectionEqualToObject=0 or 1
# default: 0
# If this option is set (=1), Backup Session Manager (BSM) sets
# catalog retention to object protection if catalog retention is
# larger than object protection.
# ExternalScriptMode=0 or 1 or 2
# default: 0
# This option is used to check if omnirpt is allowed to
# execute script on CM as external method. If set to 0, no
# checking is performed, if set to 1, script can be executed
# only in bin directory, and if set to 2 executing of external
# scripts is not allowed.
# BackupDeviceIdle=seconds or -1
# default: 600
# During "BAR" backups, devices can be running without assigned
# objects. This option specifies maximum idle time for devices.
# If set to -1, BSM doesn't close and unlock device until
# session has finished.
# FileMediumCapacity=MaxSizeInMBytes
# default: 100
# If you are using "file" devices, you should change this to
# specify the maximum size of a "file" medium. When Data Protector
# writes specified size of data in one file, you will get a
# mount prompt. Or if you are using "file" exchanger, the next
# "file medium" will be loaded. If the global option is
# changed, the logical medium must be exported and the file
# ClnTapeBCPrefix=Prefix
# default: CLN
# If this option is set to a non-empty value, Data Protector will
# check barcodes (during barcode rescan) and if the barcode
# starts with specified Prefix, the medium will be treated as a
# cleaning tape. This is useful if you have devices with barcode
# support, but it cannot detect a cleaning tape.
# MountDelay=DelayInMinutes
# default: 30
# This is the default mount prompt delay (session manager will
# wait for specified time for someone to confirm/cancel a mount
# request. If there is no answer on mount prompt, a mount
# script will be executed). You can override this setting for
# each configured device with advanced options of the device.
# MountScript=FullPathname
# default: "/opt/omni/lbin/Mount.sh"
# Default mount script. You can override this setting for
# each configured device with advanced options of the device.
# ScriptUser=username
# default: "omniback"
# The backup session manager tries to execute scripts with
# permission from the user who started the session or from the
# owner of the session if the owner of the session has switched.
# If Data Protector cannot find the user in the /etc/passwd file, it
# tries to execute the script as the ScriptUser. If even the
# ScriptUser is not found in the /etc/passwd file, the script is
# not executed.
# MMFairLimit=PercentageOfPoor
# default: 80
# This limit is used for detecting "Fair" (almost "Poor") media.
# When a medium exceeds a specified percentage of limit
# specified for "Poor" media (this limit can be set for each
# pool), it is marked as "Fair". Data Protector Media Management
# uses such ("Fair") media only if there is no "Good" media
# available.
# NonResidentLocation=NonResidentLocationString
# default: ""
# This is the string displayed in location field, when the media
# is ejected from the slot.
# InitOnLoosePolicy=0 or 1
# default: 0
# This option is used by backup session manager. When using
# loose policy media checking, this option is checked if the
# session manager should automatically initialize new media.
# FreePoolDeallocFreq=TimesPerDay
# default: 1
# limit: 1 <= FreePoolDeallocFreq <= 96
# This period is used to run free pool deallocation process
# on Data Protector Cell Manager (Media Mgmt 2).
# If set to 1, deallocation is performed once per day (00:00),
# set to 2 two times per day (00:00,12:00), set to 3 three times
# per day (00:00, 08:00, 16:00), set to 4 four times per day
# (00:00, 06:00, 12:00, 18:00). If maximum (96) is specified,
# free pool deallocation will be started every 15 minutes.
# EnableSCSIReserveRelease=0 or 1
# default: 0
# If set to 1, this option enables access to the SCSI reserve/release
# functionality used with library robotics. Set this variable only
# after consulting the Hewlett-Packard support. General
# use of this variable is not supported, unless prior agreement
# from Hewlett-Packard support is obtained.
# MinUID=MinimumUID
# default: 100
# This limit is used by the User Configuration GUI. The GUI
# will show in list of users to configure only users with higher
# UID than MinimumUID. This is useful for excluding non-human
# users like root, bin, daemon, uucp...
# ShowAllMsgs=0 or 1
# default: 1
# If this option is set, the Monitor GUI will show all messages
# generated by a session when the user connects to a running
# session. If this is not set, the Monitor GUI will show only
# new messages (messages which were generated after connect).
# MaxGUIMsg=SizeOfMsgWindowInBytes
# default: 1 MB
# This limit is used to limit the output to Backup, Restore and
# Monitor GUI message windows. In some cases, there can be a
# lot of messages to put in the message part of the window.
# This may cause the GUI to abort because it runs out of memory.
# So if the total sum of all messages (length) is greater than
# the specified limit, the GUIs will stop writing messages to
# the message window (The user will get a warning message). The
# Backup, Restore or Monitor operation will continue.
# WebBrowser=WebBrowserPath
# default: "/opt/NSCPcom:/usr/dt/bin"
# The search path for finding the Netscape browser used in
# displaying Online Help, Topic Help and "Data Protector on the Web".
# AcroRead=AcrobatReaderCommand
# default: "/opt/Acrobat4/bin/acroread"
# The command to start the Acrobat reader for displaying
# documentation files. Several values can be divided by ":".
# HideMethodsPerClient=0 or 1
# default: 1
# This option is used to hide the End User Notification capability by
# default.
# BindTimeout=TimeoutInSeconds
# default: 15
# BindTimeout is used by the MOM GUI. BindTimeout is expected
# maximum response time of each Cell Server to MOM GUI.
# TM_LABEL_xx=MenuLabel
# OnlyValidObj=0 or 1
# default: 1
# This option changes the list mode for versions of the object
# (in GUI and CLI). If this option is set to "1" (default),
# only valid (at least one medium of object exists) object
# versions are listed and if set to "0", all object versions are
# listed (even invalid object versions, unusable for restore).
# AlignedDefault=0 or 1
# default: 0
# If this option is set, VBDA performs aligned backup.
# Otherwise, aligned backup in VBDA is disabled (default).
# IncrOnProtected=0 or 1
# default: 1
# If this option is set (default), incremental backup is done
# only on a protected chain (all backups from full to last
# incremental must be protected). Otherwise, protection of a
# related chain is not checked at all.
# UpgradeIncrToFull=0 or 1
# default: 1
# If this option is set (default), incremental backup is
# automatically upgraded to a full backup if there is no valid
# related chain to be used for incremental backup. Otherwise,
# the backup session manager will abort incremental backup.
# ForceFull=0 or 1
# default: 0
# If this option is set (=1), Data Protector will force full backup
# for every object that failed or was aborted in last full
# backup. This allows very strict handling of backup
# generations and does not allow incremental backup to be made
# on full/incremental backup from previous generation. In such
# case, Data Protector will make new Full backup of the object
# instead.
# StrictPrivacyCheck=0 or 1
# default: 1
# If this option is set (=1), Data Protector strictly checks
# private objects (backups). Username, groupname and hostname
# must match for an object to be accessible by a user. If an
# option is not set (=0), only the username is checked. Admin
# users have full access (no checking).
# HidePrivateObj=0 or 1
# default: 1
# If this option is set (=1), Data Protector will never show
# objects that are private to other users (except admin users).
# If this option is not set (=0), objects are shown in list of
# objects of session / on medium, but restore is still denied.
# Admin users have full access (no checking).
# MaxSessions=MaxNumberOfSessions
# default: 200
# limit: 25 <= MaxNumberOfSessions <= 200
# Maximum number of concurrently running sessions (session
# managers).
# MaxBSessions=MaxNumberOfBackupSess
# default: 5
# limit: 1 <= MaxNumberOfBackupSess <= MaxSessions
# The maximum number of concurrently running backup sessions
# (backup session managers). If the number provided is out
# of specified range, default (5) will be used.
# SmSendReceiveTimeout=timeout in seconds
# default: 1800
# This timeout is used when sending or receiving data from
# Session Manager (BSM, RSM, etc.). If operation can not be
# completed within timeout value, error is issued and
# connection closed.
# MaxMAperSM=MaxMediaAgentsPerSM
# default: 32
# limit: 1 <= MaxMediaAgentsPerSM <= 32
# The maximum number of concurrently running media agents per
# session manager. It is used for load balancing and does not
# limit the number of media agents with statically assigned
# objects.
# MaxDAperMA=MaxDiskAgentsPerMA
# default: 32
# limit: 1 <= MaxDiskAgentsPerMA <= 32
# Maximum number of concurrently running disk agents per media
# agent. This is actually the maximum concurrency of any device
# in the cell.
# UseMaxConc=0 or 1
# default: 0
# If this option is set (=1), BSM will always start the MA with
# maximum concurrency. Check the global option 'MaxDAperMA' for
# to find out the maximum concurrency.
# SegmentsPerDA=SegmentsPerDA
# default: 30
# limit: 0 <= SegmentsPerDA <= 2147483647
# This number is used by the Restore SM to determine how often
# to start disk agents for the restore of OmniStorage (VBFS)
# volume. If 0 is specified, only one disk agent will be
# started and therefore all media will be read (slow).
# SmMaxAgentMessages=NumberOfAgentMessages
# default: 3000
# This option specifies the maximum number of messages received
# from a single Media/Disk agent that Backup/Restore Session
# Manager will store into the database. When this value is
# exceeded, a warning message is displayed indicating that any
# additional messages from this particular agent will not be
# stored into to the database. However, messages are still
# visible in the session monitor while session is alive.
# SmMaxMessagesPerSession=NumberOfMessagesMessagesPerSession
# default: 30000
# This option specifies the maximum number of messages per one
# session that Backup/Restore Session Manager will store into
# the database. When this value is exceeded, a warning message
# is displayed indicating that any additional messages from this
# session will not be stored into to the database. However,
# messages are still visible in the session monitor while
# session is alive.
# SmMaxAgentStartupRetries=NumberOfRetriesAtAgentStartup
# default: 1
# limit: 1 <= SmMaxAgentStartupRetries <= 50
# This option specifies the maximum number of retries that a
# session manager will use to start an agent before it fails.
# MinDelayForConnectionRetry=TimeInSeconds
# default: 30
# limit: 0 <= MinDelayForConnectionRetry <= 120
# Specifies the minimum time in seconds between connection
# retries. Used when starting an agent. If connection fails next
# one will not be attempted until this time has elapsed.
# SmWaitForNewClient=WaitForInMinutes
# SmWaitForNewClientSec=WaitForInSeconds
# default: 30 seconds
# This timeout is used by backup and restore session managers
# (BSM and RSM). BSM: After the last BAR client disconnects, BSM
# will wait for specified timeout for new client connections.
# If there are no new connections in the specified timeout, BSM
# will complete the session and go down. RSM: After receiving
# request for restore for specified device, RSM will wait for
# specified timeout before actually starting RMA. Changing this
# timeout may influence restore performance.
# SmWaitForFirstClient=WaitForInMinutes
# SmWaitForFirstClientSec=WaitForInSeconds
# default: 10 minutes
# This timeout is used by the backup session manager. After the
# backup script on the BAR system starts, the BSM will wait for
# specified timeout for the first client connections. If there
# are no connections in the specified timeout, BSM will go down.
# SmLogStartStop=0 or 1
# default: 0
# If this option is set (=1), session managers will put an entry
# into the /var/opt/omni/log/sm.log file to log the start and
# end time.
# SmWaitForDevice=WaitForInMinutes
# default: 60 minutes
# This timeout is used by the Backup Session Manager. The
# Backup Session Manager first tries to get a lock for all
# devices that will be used for backup. If a lock can not be
# obtained for the desired devices, the BSM will wait for the
# specified time-out. If it can not get a lock for the devices,
# the backup session will fail. If dynamic is used then the
# Backup Session Manager tries to get a lock for the minimum
# number of devices used for backup. In general,
# SmWaitForDevice is used as a time-out for different kinds of
# queuing (queuing for lock, license etc.).
# SmWaitForDB=WaitForInMinutes
# default: 60 minutes
# This timeout is used by the backup session manager. BSM tries
# to open the database at startup. If the database cannot be
# opened (omnidbcheck or omnidbutil are running), BSM will wait
# for specified timeout. If it can not open the database within
# the timeout it will go down and report it in the global debug
# SmDisableScript=0 or 1
# default: 0
# This value is used to disable any execution of pre/post-exec
# scripts: session, remote session and object pre/post-exec.
# ExecScriptOnPreview=0 or 1
# default: 0
# This option is used by the Backup SM to check if it should
# execute the session pre-exec script or post-exec script when
# backup preview is running. The alternative is to check for
# environment variable PREVIEW (1 or 0), which is exported to
# the environment of the pre/post-exec script.
# ScriptOutputTimeout=TimeoutInMinutes
# default: 15 minutes
# This timeout is used by the Backup SM. The session pre-exec
# or post-exec script should send some output at least every
# ScriptOutputTimeout minutes, or the session gets aborted by
# the session manager.
# MaxWaitForSm=WaitForInSeconds
# default: 25 seconds
# This timeout is used by the Cell Request Server (CRS). When
# CRS starts the session manager, it will wait for specified
# timeout for an initial identification from the session
# manager. If the session manager does not identify itself, CRS
# will kill the session manager.
# MC_x=Visible
# DP_x=Visible
# DBIdleTimeout=TimeoutInMinutes
# default: 30 minutes
# This timeout is only used by the Database Session Manager
# (DBSM). When the DBSM is inactive (GUI/command does not
# request any data from it) for a specified time, it will close
# the database. When the GUI/command request a data again, the
# DBSM will try to open the database again and reply to request.
# MMDLockTimeout=TimeoutInMinutes
# default: 60 minutes
# This timeout is used in case of communication problems between
# "MMD" and Session Managers. If after a specified timeout, the
# Session Manager does not reconnect to "MMD", the Session
# Manager's device, media and slot locks are cleared.
# DBLockTries=NumberOfTries
# default: 10
# limit: 1 <= NumberOfTries <= 100
# This value specifies how many times the "DB" will retry on a
# failed lock request.
# DBLockTimeout=TimeoutInSeconds
# default: 1800
# limit: 120 <= TimeoutInSeconds <= 7200
# This timeout is used by Velocis. A Velocis call will wait for
# a specified timeout for a table lock, before the lock request
# will fail. See also: DBLockTries.
# SmPeerID=WaitForInMinutes
# default: 5 minutes
# This timeout is used by all session managers
# (BSM/RSM/MSM/DBSM). If the new connection was established to
# the session manager (for example: the started
# Disk Agent/Media Agent has connected to SM, new monitor
# (xomnimonitor/omnistat) has connected to SM...) the connected
# client should identify (send its name and version) itself to
# the session manager in the specified time. If the client does
# not identify itself in the specified time, SM closes
# connection to the client.
# SmFirstConn=WaitForInSeconds
# default: 30 seconds
# This timeout is used by all session managers. The front-end
# (GUI or command) which started the session manager, should
# connect to the SM in the specified time. Otherwise the SM
# will go down.
# SmWaitToOpenDevice=WaitForInSeconds
# default: 30 seconds
# If the BMA cannot open the device for SmWaitToOpenDevice time
# period (for example: in the case when OB2SLEEP and OB2RETRY
# are big), the BSM will start a new media agent (if any). The
# running BMA will NOT be aborted. It will be used when the
# device becomes available.
# SmMaIdleTimeout=TimeoutInMinutes
# SmDaIdleTimeout=TimeoutInMinutes
# DaStartDelay=DelayInSeconds
# default: 5 seconds
# limit: 0 <= TimeoutInSeconds <= 30
# This delay is used by backup & restore session managers.
# After the disk agent disconnects from the session manager, SM
# will wait for a specified delay period before it starts a new
# disk agent. This delay solves timing problems in TCP/IP on
# some platforms.
# UsePanScripts=0 or 1
# default: 0
# If this variable is set (=1), Data Protector will always try to
# start the session pre-, post-exec scripts and mount request
# scripts from the /opt/omni/lbin/scripts directory. No scripts
# in other directories can be started for security reasons.
# MultiHomed=0 or 1
# default: 1
# If set, the user can add additional alias hostnames for
# ServiceGuard installations.
# Ob2TapeStatistics=0 or 1
# default: 0
# If enabled, this option allows tape statistics logging into
# media.log file.
# Ob2HeaderCheck=0 or 1
# default: 1
# If enabled, this option allows checking of medium header
# before medium is removed (ejected).
# HostBackupExcludesApplyToSingleVolume=0 or 1
# default: 0
# If enabled, host backup exclude options apply only to a single volume.
# SessSuccessfulWhenNoObjectsBackedUp=0 or 1
# default: 0
# This option is used by Data Protector. By setting
# this option to 1 user can change the Data Protector
# behavior in order to report session as successful when no objects
# were backed up.
#
#---------------------------------------------------------------------------
#
# Filename: <Data_Protector_home>\omnirc.TMPL
#
# $Revision: /main/dp51/42 $
#
# This file is a template for Data Protector agent environment variables.
#
# To make it active, copy it into file <Data_Protector_home>\omnirc.
#
# To make a variable active, uncomment the line containing the variable,
# remove all leading spaces and set the desired value for the variable
# as follows:
# <VariableName>=<VariableValue>
#
# An omnirc variable must be set on the Data Protector client that performs
# the process affected by the variable.
# The omnirc variables' default values are set by Data Protector binaries
# and can not be set in this file.
# When changing the omnirc file, you have to restart the Data Protector
# services/daemons on the Data Protector client where you modified the
# omnirc file. This is mandatory for the crs daemon on UNIX and recommended
# for Data Protector CRS and Data Protector Inet services on Windows.
# Specifically on Windows, restarting is not required when adding or
# changing entries, but only when removing entries (or renaming the file).
#
# NOTE: Changing these variables can cause misbehavior of Data Protector
# software components. Please contact Data Protector customer support
# before changing any variables!
#
#===========================================================================
# Variables for a Media Agent client
#===========================================================================
# OB2READ0_FSM=0|1
# Default: 0
# Some backup devices do not report the FILEMARK when 0 bytes is
# read. If the variable is set to 1, Data Protector treats such a situation
# as a filemark. If the variable is set to 0, Data Protector does not treat
# such a situation as a filemark.
#
# OB2NOLOCKDRIVE=0|1
# Default: 0
# A StorageTek related variable, relevant only if the Library Control System
# and not Data Protector controls the library robotics. When this variable
# is set to 1, the drive is not locked during the Data Protector activity
# on a drive. If the variable is set to 0, the drive is locked during the
# Data Protector activity on a drive.
#
# OB2BLKPADDING_<DeviceType>=<number_of_empty_blocks>
# Default: 0
# During copying tapes it may happen that the data on the source tape does
# not fit on the target tape. This can be due to a perfect streaming of
# the source tape during the backup, which may not be the case for the
# target tape during the copy operation. It is also possible that the
# source and the target tape are of a slightly different length.
# In order to avoid this, the below set of variables can be used to
# specify the number of empty blocks written after the tape header at
# the initialization time for various device types. The effect is that
# the size of data on the medium is smaller, and tape copy should run
# without problems. When the medium is copied, the empty
# blocks are not copied.
#
# For each device type, there is a special variable used, specifying the
# number of empty blocks to be written after the medium header (device
# default block size is used):
#
# OB2BLKPADDING_1 DAT/DDS
# OB2BLKPADDING_2 QIC Quarter Inch Cartridge
# OB2BLKPADDING_3 8mm - ExaByte
# OB2BLKPADDING_4 AIT - Advanced Information Technology
# OB2BLKPADDING_5 3480 Cartridge
# OB2BLKPADDING_6 Raw Magnetic Disk
# OB2BLKPADDING_7 Regular Disk File
# OB2BLKPADDING_8 STK 9840
# OB2BLKPADDING_9 Generic Magnetic Tape Device
# OB2BLKPADDING_10 DLT - Digital Linear Tape
# OB2BLKPADDING_11 StorageTek SD-3 - Redwood
# OB2BLKPADDING_12 3590 Cartridge (Magstar)
# OB2BLKPADDING_13 LTO Ultrium
# OB2BLKPADDING_14 Quantum SuperDLT
#
# For example:
# OB2BLKPADDING_3=5
# This will put 5 empty blocks on the beginning of every 8mm - Exabyte
# media written.
#
# The optimum value of the variable differs from one medium to another.
# The number of blocks should be calculated according to the set block size.
# Normally, the empty blocks should take up approximately 1 percent of the
# length of the entire tape. If this is still not enough for successful
# copying, increase the value by 0.5 percent until you find the most suitable
# value.
#
# OMNIMAXCATALOG_<LogicalDevice>=1-60
# Default: 12
# Sets the maximum size (in MB) of the Data Protector catalog that is
# sent by the MA to the BSM at the end of each segment. Note that when the
# specified size is reached, the MA will finish the current data segment
# on the tape and start a new one.
#
# You must specify a specific value for each configured logical device
# separately, by changing the <LogicalDevice> with a specific logical device
# name. You can get the logical device names in the Data Protector GUI
# or by using the omnidownload -list_devices command (the names are listed
# under the "Device name" column in the output of the command).
#
# For example, if you have configured a logical device with the name
# DLT_STD, you need to set the OMNIMAXCATALOG_DLT_STD variable.
#
# OB2XS2RETRY=n
# Default: 15
# Number of retries for the open call for the
# external SCSI II robotics control.
#
# OB2XS2SLEEP=s
# Default: 10
# Timeout (in seconds) between the open calls the
# external SCSI II robotics control.
#
# OB2DEVRETRY=n
# Default: 30
# Number of retries for the open call for the drives. On Tru64 system the retries
# may end earlier, if time limit for open retries (OB2DEVTIMEOUT) is exceeded.
#
# OB2DEVSLEEP=s
# Default: 30
# Timeout (in seconds) between the open calls for the drives.
#
# OB2DEVTIMEOUT=s
# Default: 6600
# This is a Tru64 specific variable. It specifies the amount of time (in seconds) in
# which Data Protector will re-trying to open a device special file of a drive.
#
# OB2CHECKCLNSLOT=0|1
# Default: 1
# If a cleaning slot is specified on non-barcode device, and this variable is set to
1,
# Data Protector will try to check if cleaning tape is in the slot. If
# variable is set to 0, it is assumed that cleaning tape is in slot and no check
will
# be performed.
#
# OB2STKRETRY=n
# Default: 40
# A StorageTek related variable, relevant only if the Library Control System
# and not Data Protector controls the library robotics. Sets the number of
# retries for the open call for stacker devices.
#
# OB2STKSLEEP=s
# Default: 5
# A StorageTek related variable, relevant only if the Library Control System
# and not Data Protector controls the library robotics. Sets the timeout
# (in seconds) between the open the calls for stacker devices.
#
# OB2CLNRETRY=n
# Default: 60
# Number of retries when accessing a slot that has the cleaning medium loaded.
#
# OB2CLNSLEEP=s
# Default: 5
# Timeout (in seconds) between retries when accessing a slot that has the
# cleaning medium loaded.
#
# OB2SCTLMOVETIMEOUT=s
# Default: 240000
# Data Protector timeout period (in milliseconds) after the SCSI II
# MoveMedium command has been issued. If the move is not completed after
# this timeout, the ioctl call will fail.
#
# OB2TAPEALERT=0|1
# Default: 1
# By issuing tape-alert command to devices Data Protector can retrieve and
# report statistical information maintained by the device about the device
# or its logical units. When OB2TAPEALERT variable is set to 1, which is the default
# setting, Data Protector issues tape-alert command to devices and reports
# the retrieved tape-alert information. If the OB2TAPEALERT is set to 0
# the tape-alert reporting is disabled.
#
# DAS_CLIENT=<GRAU_client_name>
# Default: If DAS_CLIENT variable is not set the client name OMNIBACK is used.
# This variable, together with a DAS_SERVER variable, is a ADIC/GRAU DAS
# Library backup device related. When a ADIC/GRAU DAS library is configured a
# list of all DAS clients (GRAU_client_names) is defined in file
# \DAS\ETC\CONFIG on the DAS server computer. Each Data Protector client with
# DAS agent installed that wants to access ADIC/GRAU DAS library robotics
# individually should have the DAS_CLIENT variable set to unique
# GRAU_client_name defined on DAS server computer. In this way different Data
Protector
# clients can access ADIC/GRAU DAS library, each one with its unique
# GRAU_clinet_name. If DAS_CLIENT is not specially declared the DAS_CLIENT
# client name OMNIBACK is used.
#
# DAS_SERVER=<GRAU_server_name>
# Default: if DAS_SERVER variable is not set the autochanger ioctl host is
# used (parameter: -ioctl <server_name>).
# This variable, together with a DAS_CLIENT variable, is a ADIC/GRAU DAS
# Library backup device related. When a ADIC/GRAU DAS library is configured a
# list of all DAS clients (GRAU client names) is defined in file
# \DAS\ETC\CONFIG on the DAS server computer. Each Data Protector client with
# DAS agent installed that wants to access ADIC/GRAU DAS library robotics
# individually should have the DAS_SERVER variable set to the host name of
# the DAS server computer.
#
# OB2_DAS_ENTER_RETRY=<number of retries>
# Default: 10
# OB2_DAS_ENTER_SLEEP=<sleep for n seconds>
# Default: 6
# These variables are ADIC/GRAU DAS Library backup device related. When you wish
# to insert new catridge into library this is done through I/O unit(mail slot).
# DP tries to insert media but it takes some time for user to insert the media
# and close the library door. Therefore DP retries this operation for
# OB2_DAS_ENTER_RETRY times using timeout of OB2_DAS_ENTER_SLEEP seconds.
#
# OB2LIB_STACKIMP=<scsii_address>
# OB2LIB_STACKEXP=<scsii_address>
# Default: -1 (no default value)
# If a library supports stack mode for entering and ejecting the media,
# with a special SCSI II address for the eject Mailslot, and one for the
# enter Mailslot (although physically the same slot), the above two
# variables are used to tell the Data Protector MA which
# addresses to use for entering (OB2LIB_STACKIMP) and ejecting the media
# (OB2LIB_STACKEXP).
# Example:
# OB2LIB_STACKEXP=289
# OB2LIB_STACKIMP=288
#
# OB2DASDRIVESTATUS2=0|1
# Default: 0
# By default only up to 15 DAS drives can be used in a GRAU library.
# By setting this variable to 1, you can use up to 250 DAS drives in a
# GRAU library.
#
# OB2NDMPFH=Y|N
# Default: Y
# NDMP MA specific environment variable.
# If this variable is set to Y, the NDMP server will create the file
# history information.
# If it is switched on, Data Protector can restore single files.
# If this variable is set to N, the NDMP server will not create the file
# history information, which is the recommended setting.
#
# OB2NDMPDIRECT=Y|N
# Default: Y
# NDMP MA specific environment variable.
# This variable defines the use of the direct restore functionality. It
# is used only if the variable OB2NDMPFH is switched on at backup time.
# If this variable is set to Y, Data Protector uses the direct access
# restore functionality.
# If this variable is set to N, the direct access restore functionality
# is disabled.
#
# OB2SANCONFSCSITIMEOUT=s
# Default: if not set, the OB2SCSITIMEOUT variable is used.
# A sanconf command specific variable.
# This variable sets the timeout value for the sanconf command triggered
# operations. If used, it must be set on all clients affected by the
# sanconf command, before running the sanconf command. The recommended
# value is 15 seconds.
#
#===========================================================================
# Variables for a Disk Agent client
#===========================================================================
#
# OB2ENCODE=0|1
# Default: 0 (encoding is off)
# Setting the variable overrides the backup option -encode for the
# backup disk agent.
#
# OB2VXDIRECT=0|1
# Default: 0
# A VxFS related option. If set to 1, then VxFS VX_DIRECT advisory is set,
# meaning that direct data transfer between the disk and the user-supplied
# buffer for reads and writes is used. This improves backup performance.
# Install the latest VxFS patch on your system before setting this variable.
#
# OB2ALIGN=n
# Default: 0
# Used to define the alignment factor for backup read calls. If greater
# than zero, OB2ALIGN variable sets the alignment to
# (media_blocksize * align factor), and all subsequent file backup reads
# are performed according to this size. It is used for performance
# improvement on HP-UX VxFS only.
#
# OB2CHECKCHANGETIME=0|1|2
# Default: 1
# During the incremental backup, the Disk Agent uses the "last modified"
# and "last accessed" time attributes to detect the files changed
# since the last backup. Use this variable to control which files should
# be backed up when an incremental backup is performed.
# There are three valid values:
# 0 - Only the "last accessed" time attribute is checked. Backs up modified
# files, but does not back up moved files.
# 1 - Same as 0, except when the option "do not preserve access time attributes"
# is selected and file locking is disabled. In this case it acts the same
# as when the variable is set to 2.
# 2 - Always checks both time attributes. Backs up modified and moved files.
#
# OB2INCRDIFFTIME=s
# Default: 0
# Specifies an "incremental latency" period (in minutes) that is enforced
# when checking the inode "last accessed" time with incremental backups. The
# referential time received from the Session Manager (time of the previous
# backup) is the first incremented by the specified period and then compared
# to the inode "last accessed" to qualify for backup. This variable takes
# effect only when specified together with the "OB2CHECKCHANGETIME" variable.
#
# OB2SKIPDIRECTORY=0|1
# Default: 0
# Setting the variable allows the backup option -skip for directory not
# only files.
#
# OB2LOGENABLED=0|1
# Default: 0 (logging is off)
# Setting the variable will enable logging of names of all objects backed
# up during a backup session.
#
# OB2CLUSTERDISKTYPES=<resource_type_name>
# Default: none
# This option enables the Data Protector Inet service to
# differentiate between local and cluster disk resources which use their
# own resource driver and not the Microsoft resource driver. To specify a
# disk resource, define the disk resource name as the value of the
# OB2CLUSTERDISKTYPES options.
# For example, if you are using NetRAID4M disk, specify:
# OB2CLUSTERDISKTYPES=NETRaid4M Diskset
# If you are also using Veritas Volume Manager, specify:
# OB2CLUSTERDISKTYPES=NETRaid4M Diskset;Volume Manager Disk Group
#
# OB2NOREMOTEWARNINGS=0 | 1
# Default: 0
# If enabled, no warnings will be displayed when trying to back up filesystem
# containing mount point to local or NFS mounted filesystem.
# It is used only on Unix-like systems.
#
#===========================================================================
# Common variables for split mirror, snapshot and direct backup (HP StorageWorks
# XP Agent (SSEA) client, HP StorageWorks VA Agent (SNAPA) client,
# EMC Symmetrix Agent (SYMA) client, HP StorageWorks EVA Agent(EVAA) client
# and HP StorageWorks SA Agent (SAA) client)
#===========================================================================
#
# SMB_DISK_RESCAN=s
# Default: 30
# The SMB_DISK_RESCAN variable sets the time interval (in seconds) needed
# for successful disk rescan. The default value is 30 seconds.
#
#===========================================================================
# Variables for an HP StorageWorks XP Agent (SSEA) client
#===========================================================================
#
# SSEA_SYNC_SLEEP_TIME=s
# Default: 5
# During the resynchronization of mirrored disks, the HP StorageWorks
# XP Agent checks the status of mirrored disks for so many times as specified
# by the SSEA_SYNC_RETRY variable. The SSEA_SYNC_SLEEP_TIME variable sets
# the time interval (in seconds) between these status checks. The default
# value is 5.
#
# SSEA_SYNC_REPORT_RATE=n
# Default: 2
# During the resynchronization of mirrored disks, the HP StorageWorks
# XP Agent checks the status of mirrored disks within a check interval
# specified by the SSEA_SYNC_SLEEP_TIME variable for so many times as
# specified by the SSEA_SYNC_RETRY variable. The SSEA_SYNC_REPORT_RATE
# variable specifies the rate of displaying the status of the mirrored disks
# to the Data Protector Monitor. For example, if the SSEA_SYNC_SLEEP_TIME option
# is set to 5 seconds and the SSEA_SPLIT_REPORT_RATE is set to 2, the status
# will be displayed to monitor every 2nd check, that is every 10 seconds. The
# default value is 2.
#
# SSEA_SYNC_RETRY=n
# Default: 10
# During the resynchronization of mirrored disks, the HP StorageWorks
#
# In the case of an HP-UX or Solaris Data Protector HP StorageWorks Disk Array XP
client:
# /var/opt/omni/tmp, if the SSEA_MOUNT_PATH is not specified or
# <SSEA_MOUNT_PATH>, if the SSEA_MOUNT_PATH is set to <SSEA_MOUNT_PATH>.
#
# In the case of a Windows 2000 Data Protector HP StorageWorks Disk Array XP
client:
# <Data_Protector_home>\tmp, if the SSEA_MOUNT_PATH is not specified or
# <SSEA_MOUNT_PATH>, if the SSEA_MOUNT_PATH is set to <SSEA_MOUNT_PATH>.
#
# If this variable is set to 1:
#
# The mountpoint for the backed up filesystem is created in the:
# /<mountpoint_name_on_application_system> (HP-UX or Solaris Data Protector HP
StorageWorks Disk Array XP client)
# \<mountpoint_name_on_application_system> (Windows 2000 Data Protector HP
StorageWorks Disk Array XP client)
# <Drive_letter_on_the_app_system>: (Windows NT Data Protector HP StorageWorks Disk
Array XP client)
# directory or drive letter on the backup system.
#
# SSEA_MULTI_MOUNT=0|1
# Default: 0
# This variable specifies, together with the SSEA_PRESERVE_MOUNTPOINTS and
# with the SSEA_MOUNT_PATH variables, how the mountpoints are created on
# the backup system. If this variable is set to 1, the application system
# LDEV MU# is appended at the end of mount point path on the backup system,
# thus enabling every first level mirror to be mounted to its own mountpoint.
# If this variable is set to 0, the integration always mounts the selected
# first level mirror to the same mountpoint on the backup system.
# This variable is ignored if the SSEA_PRESERVE_MOUNTPOINTS is set to 1.
#
# SSEA_MOUNT_PATH=<first_part_of_the_mountpoint_path>
# Default: none
# This variable specifies, together with the SSEA_PRESERVE_MOUNTPOINTS and
# with the SSEA_MOUNT_PATH variables, how the mountpoints are created on the
# backup system. By default, this variable is not set. If this variable is not
# set, the first part of the mountpoint path is set as:
# /var/opt/omni/tmp (HP-UX or Solaris Data Protector HP StorageWorks Disk Array
XP client) or
# <Data_Protector_home>\tmp (Windows 2000 Data Protector HP StorageWorks Disk
Array XP client).
# Specify the first part of the mountpoint path to set this variable.
# This variable is ignored if the SSEA_PRESERVE_MOUNTPOINTS is set to 1.
#
# SSEA_IR_VGCHANGE_A=vgchange -a e|vgchange -a s|vgchange -q n -a y
# Default: vgchange -q n -a y
# This an HP-UX instant recovery related variable. It controls in which mode
# the mirrored volume groups on the application system are activated after
# the SVOLs are synchronized to PVOLs during the instant recovery process.
#
# The variable is must be set on the application system only.
#
# By setting this variable, the mirrored volume groups can be activated in
# the following modes:
# exclusive; the variable has to be set as follows: SSEA_IR_VGCHANGE_A=vgchange -
a e
# shared; the variable has to be set as follows: SSEA_IR_VGCHANGE_A=vgchange -a s
# normal (vgchange -q n -a y); the variable is not set (default)
#
# Use the exclusive mode to enable the instant recovery if an application/filesystem
# is running in the MC/ServiceGuard cluster on the application system.
#
# SSEA_ALWAYS_POST_SCRIPT: 0|1
# Default: 0
# By default, the command/script set by the Restart application
# option is not executed if the command/script set by the Stop/quiesce application
# option fails.
# If this variable is set to 1, the command/script set by the Restart application
# option is always executed if set.
#
#=================================================================================
# Variable for an HP StorageWorks XP Agent (SSEA) and VA Agent (SNAPA) client
#=================================================================================
#
# OB2AUTOPATH_BALANCING_POLICY=0|1|2
# Default: 1
# HP Auto Path for HP-UX provides enhanced data availability for host systems
configured
# with multiple host adapters and connections to a disk array. When several
alternate
# paths are available, HP Auto Path will dinamically balance data load between the
# alternate paths to achieve optimum performance. The load balance policies
supported are:
#
# 0 [none] - No Load Balance policy.
# 1 [RR] - Round Robin policy (Default).
# 2 [SQL] - Shortest Queue Length policy.
# 3 [SST] - Shortest Service Time policy.
#
# During a Data Protector session, when the HP StorageWorks AutoPath
# Shortest Queue Length load balance policy is set, and if the failover to
# an alternate path occurs, Data Protector will fail the session.
#
#===========================================================================
# Variables for an HP StorageWorks VA Agent (SNAPA) client
#===========================================================================
#
# The varaibles described in this section, with the exception of the
# SNAPA_IR_VGCHANGE_A variable are used for the purpose of backward compatibility.
# It is recommended to use the common snapshot agents variables. Refer to
# "Variables for any ZDB Agent client".
#
# SNAPA_PRESERVE_MOUNTPOINTS=0|1
# Default: 0
# Data Protector creates unique mount points and mounts them automatically
# on the backup system. This variable controls how the mountpoints on the
# backup system are created.
#
# In the case stated below, this variable is set to 1 and its override is ignored;
# Oracle8
#
# If this variable is set to 0:
#
#
<BU_MOUNT_PATH>/<application_system_name>/<mountpoint_name_on_application_system>_<Sess
ionID>,
# directory on the backup system, where <BU_MOUNT_PATH> is:
#
# In the case of an HP-UX Data Protector HP StorageWorks Disk Array VA client:
# /var/opt/omni/tmp, if the SNAPA_MOUNT_PATH is not specified or
# <SNAPA_MOUNT_PATH>, if the SNAPA_MOUNT_PATH is set to <SNAPA_MOUNT_PATH>.
#
# In the case of a Windows 2000 Data Protector HP StorageWorks Disk Array VA
client:
# In the cases stated below, this variable is set to 1 and its override is
# ignored; the ZDB_MULTI_MOUNT and the ZDB_MOUNT_PATH variables are ignored
# if this variable is set to 1:
# Windows NT
# disk images
# Oracle8/9
# SAP R/3
# MS SQL Server 2000
# MS Exchange 5.x
#
# If this variable is set to 0:
#
#
<BU_MOUNT_PATH>/<application_system_name>/<mountpoint_name_on_application_system>_<Sess
ionID>,
# if the ZDB_MULTI_MOUNT is set to 1, or in the
#
# <BU_MOUNT_PATH>/<application_system_name>/<mountpoint_name_on_application_system>,
# if the ZDB_MULTI_MOUNT is set to 0;
#
# directory on the backup system, where <BU_MOUNT_PATH> is:
#
# In the case of a UNIX client:
# /var/opt/omni/tmp, if the ZDB_MOUNT_PATH is not specified or
# <ZDB_MOUNT_PATH>, if the ZDB_MOUNT_PATH is set to <ZDB_MOUNT_PATH>.
#
# In the case of a Windows 2000 client:
# <Data_Protector_home>\tmp, if the ZDB_MOUNT_PATH is not specified or
# <ZDB_MOUNT_PATH>, if the ZDB_MOUNT_PATH is set to <ZDB_MOUNT_PATH>.
#
# If this variable is set to 1:
#
# The mountpoint for the backed up filesystem is created in the:
# /<mountpoint_name_on_application_system> (UNIX client)
# \<mountpoint_name_on_application_system> (Windows 2000 client)
# directory or drive letter on the backup system.
#
# ZDB_MULTI_MOUNT=0|1
# Default: 0
# This variable specifies, together with the ZDB_PRESERVE_MOUNTPOINTS and
# with the ZDB_MOUNT_PATH variables, how the mountpoints are created on
# the backup system. If this variable is set to 1, the SessionID
# is appended at the end of the mount point path on the backup system,
# thus enabling every group of mount points for one replica storage version in
# the storage pool to be mounted to their own mount points.
# If this variable is set to 0, the integration always mounts the selected
# group of mount points for one replica storage version in the storage pool to the
# same mountpoint on the backup system.
# This variable is ignored if the ZDB_PRESERVE_MOUNTPOINTS is set to 1.
# This variable is ignored and set to 1 on HP StorageWorks Virtual Array.
#
# ZDB_MOUNT_PATH=<first_part_of_the_mountpoint_path>
# Default: none
# This variable specifies, together with the ZDB_PRESERVE_MOUNTPOINTS variable,
# how the mountpoints are created on the backup system. By default, this variable
# is not set. If this variable is not set, the first part of the mountpoint path is
# set as:
# /var/opt/omni/tmp (UNIX client) or
# <Data_Protector_home>\tmp (Windows 2000 client).
# Specify the first part of the mountpoint path to set this variable.
# This variable is ignored if the ZDB_PRESERVE_MOUNTPOINTS is set to 1.
#
# ZDB_ALWAYS_POST_SCRIPT=0|1
# Default: 0
# By default, the command/script set by the Restart application option is not
# executed if the command/script set by the Stop/quiesce application option fails.
# If this variable is set to 1, the command/script set by the Restart application
# option is always executed if set.
#===========================================================================
# Variables for an HP StorageWorks EVA Agent (EVAA) client
#===========================================================================
#
# EVA_HOSTNAMEALIASES=<host_object_name_1>,<host_object_name 2>,...
# Default: no host object names specified
# As a part of ZDB backup, the HP StorageWorks EVA client
# searches the HSV Element Manager for host objects that match the backup host. The
# search is done by the host object name. By default the HP StorageWorks EVA
# client will only search by two names:
# - full backup host hostname (as seen on the IP network)
# - short backup host hostname (as seen on the IP network)
#
# If you whish to add more hostnames to the search, you can do it using this
variable.
# Example:
# Supposing your backup host is represented within the Element Manager by the
following
# two host objects:
# - /Hosts/Backup hosts/MyHost_Port1
# - /Hosts/Backup hosts/MyHost_Port2
#
# You can force the the HP StorageWorks EVA client to find
# these two host objects by setting:
# EVA_HOSTNAMEALIASES=MyHost_Port1,MyHost_Port2
#
# EVA_EMAPI_MAX_RETRY=30
# Default: 10
# Some HSV Element Manager commands that the HP StorageWorks EVA client
# executes tend to fail when the target HP StorageWorks EVA system is under
# heavy load. In this case Data Protector will wait for the period set
# by the EVA_EMAPI_RETRY_DELAY variable and try to execute the command again. With
this
# variable, you can define how many times Data Protector should retry before
# concluding that a command has failed. You might need to increase the default value
when
# your HP StorageWorks EVA system is under extreamly heavy load.
#
# EVA_EMAPI_RETRY_DELAY=20
# Default: 10
# Some HSV Element Manager commands that the HP StorageWorks EVA client
# executes tend to fail when the target HP StorageWorks EVA system is under
# heavy load. In this case Data Protector will wait for the period set
# by the EVA_EMAPI_RETRY_DELAY variable (in seconds) and try to execute the command
again.
# See also the variable EVA_EMAPI_MAX_RETRY which sets the maximum number of retires.
# Since each HSV Element Manager command increases the load on your HP StorageWorks
EVA
# system, it is usually better to increase the delay than the number of retries.
#
# EVA_GETOBJID_MAX_RETRY=n
# Default: 10
# When creating new HSV Element Manager objects (virtual disks, presentations, etc)
the
# HP StorageWorks EVA client sends the creation requests to
# HSV Element Manager and then collects data about the created objects. When the
target HP StorageWorks EVA
# system is under heavy load, it may happen that object creation takes some time,
therefore
# Data Protector has to retry to collect the created object until all of them are
# found. With this variable you can define how many times Data Protector should retry
# before concluding that a command has failed. You might need to increase the default
# value when your HP StorageWorks EVA system is under extreamly heavy load.
# See also the variable EVA_GETOBJID_RETRY_DELAY which sets the delay between the
retries.
#
# EVA_GETOBJID_RETRY_DELAY=n
# Default: 10
# When creating new HSV Element Manager objects (virtual disks, presentations, etc)
the
# HP StorageWorks EVA client sends the creation requests to HSV Element Manager and
then
# collects data about the created objects. When the target HP StorageWorks EVA
# system is under heavy load, it may happen that object creation takes some time,
therefore
# Data Protector has to retry to collect the created object until all of them
# are found. With this variable you can define how long (in seconds) should Data
Protector
# wait between the retires. See also the variable EVA_GETOBJID_MAX_RETRY which sets
maximum
# number of retries.
# Since each HSV Element Manager command increases the load on your HP StorageWorks
EVA
# system, it is ususally better to increase the delay than the number of retries.
#
#===========================================================================
# Variables for an EMC Symmetrix Integration client
#===========================================================================
#
# SYMA_LOCK_RETRY=n
# Default: 15
# When syma agent interacts with Symmetrix disk array, it has to
# exclusively lock the SYMAPI database. If it is already locked by
# another process, syma agent retries to lock the database for as many
# times as defined by this variable.
#
# SYMA_MOUNT_PATH=<first_part_of_the_mountpoint_path>
# Default: none
# This variable specifies, together with the SYMA_PRESERVE_MOUNTPOINTS
# variable, how the mountpoints are created on the Backup (R2) System.
#
# By default, this variable is not set.
#
# If this variable is not set, the first part of the mountpoint path is set as:
# /var/opt/omni/tmp (HP-UX Data Protector EMC Symmetrix client).
#
# Specify the first part of the mountpoint path to set this variable.
#
# This variable is ignored if the SYMA_PRESERVE_MOUNTPOINTS is set to 1.
#
# SYMA_MOUNT_R2_READWRITE=0|1
# Default: 0
# This is an HP-UX related variable. It is not to be used on Windows systems.
# If this variable is set to 0, the volume groups and filesystems on the
# Backup (R2) System are activated and mounted in read-only mode. If this
# variable is set to 1, the volume groups and filesystem on the
# Backup (R2) System will be activated in read/write mode. For backup, it
# is sufficient to activate the Backup (R2) System volume groups and
# filesystem in read-only mode. Should the mirror be used for DSS or other
# tasks after the backup, this might not be sufficient.
#
# SYMA_PRESERVE_MOUNTPOINTS=0|1
# Default: 0
# Data Protector
# creates unique mount points and mounts them automatically on the Backup (R2)
System.
# This variable controls how the mountpoints on the Backup (R2) System are created.
#
# The creation of mount points is also influenced by the SYMA_MOUNT_PATH variable.
#
# In the cases stated below, this variable is set to 1 and its override is
# ignored; the SYMA_MOUNT_PATH variable is ignored if this variable is set to 1:
# Windows NT
# disk image
# Oracle8
# SAP R/3
#
# If this variable is set to 0, the mountpoint for the backed up filesystem is
# created in the:
# <BU_MOUNT_PATH>/<application_system_name>/<mountpoint_name_on_application_system>
# directory on the Backup (R2) System, where <BU_MOUNT_PATH> is:
#
# In the case of an HP-UX Data Protector EMC Symmetrix client:
# /var/opt/omni/tmp, if the SYMA_MOUNT_PATH is not specified or
# <SYMA_MOUNT_PATH>, if the SYMA_MOUNT_PATH is set to <SYMA_MOUNT_PATH>.
#
# If this variable is set to 1, the mountpoint for the backed up filesystem is
created in the:
# /<mountpoint_name_on_application_system> (HP-UX Data Protector EMC Symmetrix
client)
# <Drive_letter_on_the_app_system>: (Windows NT Data Protector EMC Symmetrix
client)
# directory or drive letter on the Backup (R2) System.
#
# SYMA_REC_FILE_LIMIT=n
# Default: 102400
# Maximum size (in bytes) of syma recovery file:
# /var/opt/omni/emc/symmR1.rec or /var/opt/omni/emc/symmR2.rec (UNIX client)
# <Data_Protector_home>\Config\Emc\symmR1.rec or
<Data_Protector_home>\Config\Emc\symmR2.rec (Windows client)
# All critical operations, which are done on the R1 and R2 systems during the
# backup (EMC), are logged in these file (e.g. de-activating volume groups)
# for recovery purposes. If the size of this file exceeds the maximum size,
# syma agent at next successful backup, starts cleaning process,
# which deletes all obsolete records in these files.
#
# SYMA_SLEEP_FOR_LOCK=s
# Default: 30 seconds
# Timeout (in seconds) between two subsequent syma agent attempts to lock
# the SYMAPI database.
#
# SYMA_SYNC_RETRY=n
# Default: 15
# A successful split and restore operation of the links
# (SRDF,TimeFinder) requires a valid status of the links
# (e.g. devices not fully synchronized, existing write pending tracks,...).
# Syma agent is sequentially checking for the invalid link statuses
# (before split or restore operation) until they are in the valid
# condition but not for more then SYMA_SYNC_RETRY times.
#
# SYMA_SLEEP_FOR_SYNC=s
# Default: 30
# Timeout (in seconds) between the last unsuccessful link status check
Lab 3: Installation
Objective: The objective of this lab is to install and configure the cell manager and clients.
Directions
Objective: To become familiar with installing a Data Protector cell and configuring clients.
Small teams allow you to configure a multi-system cell. Teaming is optional, but if chosen,
teams are best with two members.
The ideal cell for classroom use would be to have one Windows 2000 system per student and
an HP-UX system to be shared.
The objective of this lab is to create a Data Protector Cell, with one system being the Cell
Manager and Installation Server and the others being client systems and, optionally,
installation servers.
Pre-installation Activities
Plan your cell.
a. Decide which system will be the Cell Manager.
b. Decide which systems will be disk/media agent clients.
c. Decide which systems will be installation servers.
d. List the systems to be included in the cell, and check the agents to be installed there.
Document the systems that will become part of your cell (ask your instructor):
Cell Manager:
Unix Client(s):
Windows Client(s):
Check to see that all of your systems meet the prerequisites for disk space, operating system
version, RAM, etc. It is extremely unlikely that your systems do not meet them, but it is good
practice to check anyway.
1. For the HP-UX and Windows 2000 Cell Managers it is highly recommended that before
starting the installation, you create a separate file system for <OMNIVAR>/db40. For
the classroom environment, 500 MB should be sufficient.
On HP-UX, create a filesystem and mount it to /var/opt/omni/db40.
4. Verify that the name server configuration uses the HP Education name server. (Ask your
instructor for the specific IP address and search entries to use for this domain)
cat /etc/resolv.conf
5. For Windows, verify that the same DNS configuration as the Unix systems is in use. (Use
the Education site name server)
6. Proceed to the Cell Manager Installation labs which follow; choose either HP-UX or
Windows or both depending upon systems availability.
You will install from the following software depot (provided by your instructor):
1. Create a “snapshot” of the current disk space to determine how much space is consumed
by the installation.
Change the source host and depot to the one supplied by the instructor.
You are required to install the full product. However, now is a good opportunity for you to
drill down into the product to see all the available Data Protector file sets.
2. Verify the amount of disk space consumed by the installation. (notice the changes to /,
/opt, /var)
cat /tmp/before_dp.txt
bdf
/ change:
/var change:
/opt change:
or
/opt/omni/sbin/omnisv -status
8. Log off and then log on again as root. This will refresh your profile, which has been
modified by the Data Protector installation. (PATH and MANPATH variables are usually
set during login using the /etc/PATH and /etc/MANPATH, both are modified by Data
Protector).
Directions
Hardware permitting, you can complete this lab which will demonstrate the installation
process of a Cell Server and Client on the Windows platform. If you do not have a Windows
system available, the instructor can demonstrate this to the entire class instead.
1. Obtain from the instructor the location of the Data Protector software. This may be a
network share or it may be copied to a partition on the local system.
5. Select Install Data Protector. (this may require an update to the Microsoft Installer, which
will require a reboot to be able to continue)
6. Follow the installation wizard to install Data Protector. Be sure to select the Cell manager
as the choice to install. Other options will be the client, or installation server. Verify the
default location for the installation, C:\program files\Omniback (notice that
Omniback is still used by Data Protector for support of previous versions)
Installation Prompt screens:
7. When complete, you will be asked if you want to start using Data Protector. Click “NO.”
10. Verify that the Data Protector services have been added:
Use the Windows 2000 Computer Management,right click on My Computer, then
select manage, Open the Services and Applications tab, select services. Look for the three
Data Protector services; ensure they are started and configured for automatic startup
using the system account.
11. Using regedit (Start -> Run “regedit”), look at the following registry hives (do
not make any changes):
HKEY_LOCAL_MACHINE\SOFTWARE\Hewlett-
Packard\OpenView\OmnibackII\Site
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\omniback_
crs
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\OmniInet
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Velocis
Take some time to look around the GUI. See if you can find some of the controls that you
used in the Motif GUI earlier. When finished, exit the GUI.
If the system is not to be the primary cell manager, then uninstall Data Protector and re-
install just the client. While installing the Cell manager and client, you may optionally add
or remove the installation server at the same time. The selection to add/remove the
installation server will be available in the Custom Setup screen during the installation
process.
13. Uninstall Data Protector from your windows system. (Start -> Settings ->
Control Panel (Add/Remove Programs) Select Data Protector, and remove
it.
14. Run setup.exe (or Autorun)again, but this time, install the system as a client and
add the Installation Server.
15. When prompted for the Cell Server name, you have two choices:
a. Leave the field empty, this will allow the client to be imported from the cell manager
(preferred).
b. Enter the fully qualified name of the cell server (requires the use of DNS as the
lookup service). This should instruct the Cell Server to “import” your system as a
client. (Note, if you enter the cell manager name incorrectly, you will may have to
modify the registry to change it, or uninstall Data Protector and start over.
16. From the Cell Manager, start the GUI and go into the Clients list to verify that the
Windows system has been imported. If not, import it manually.
3. This will explode to a listing of the file sets contained within the full product. In the list of
file sets, select the Installation Server file set.
grep omni /etc/inetd.conf (You should see an entry for the service, omni.)
Before completing this lab, you must have already installed the cell manager and an
installation server to a second system.
At this stage the Cell Manager system does not know that the Installation Server(s) exist. In
order for the Cell Manager to see the Installation server, you must register it on the Cell
Manager.
1. Before you configure the Installation Server, look at the contents of the following file on
the Cell Manager:
<OMNICONFIG>/cell/installation_servers
At the top of the GUI, select Clients from the pull down list to switch to the clients
context.
Highlight “Installation Server” (single click with the right mouse button) to display the
pop-up menu. Choose Add Installation Server.
Enter the hostname of the system containing the installation server software. (It may be
another cell manager containing an installation server.)
The Cell Manager will now connect to the system to verify that it is indeed an Installation
Server, and if it’s okay will configure it.
Import all of the Installation Servers that you configured (Windows and Unix)
installation_servers
If you have both systems listed, congratulations! You have just configured your
Installation Server.
2. Decide which systems have backup devices to be used by Data Protector (Media Agent
clients). Review your list from the beginning of this exercise.
c. Select Clients from the pull down list to switch to the client context.
4. Verify that the appropriate agents are installed. Using the client’s context of the Data
Protector interface can perform this.
Select a host from the list of configured hosts.
The properties will automatically be shown.
What other information can you get about each host (client)?
5. On the Cell Manager (Installation Server) check the contents of the install log text files;
check for errors. (Depending upon the system, not every log will exist.)
HP-UX logs:
<OMNIVAR>/log/IS_install.log
Windows Logs:
<OMNIVAR>\log\IS_install
Before completing this lab; verify with your instructor that access to the HP Openview
Support Web is available from the classroom. The following procedure is demonstrated for a
Windows Cell Manager/Installation Server.
Objectives:
• practice the update of the Cell Manager and or Installation Server within the cell
• practice using the patch check functions of Data Protector
Procedure:
1. Create a new temporary directory to hold the downloaded patch files (some may
consume a large quantity of disk space):
Windows: C:\temp\DP51_Patches
UNIX: //tmp/DP51_Patches
2. Open the web browser and access the HP OpenView Support Web site:
http://support.openview.hp.com
3. Select the Data Protector product from the list of available OpenView products: (see
below)
5. Browse the available patch selections for the necessary/desired patches; review the “view
path full text” to determine applicability to your environment as well as patch
dependencies; then download the patch(s).
6. Select the download patch now link; then save the patch to the local disk.
7. Specify the save location as the directory created in step 1 of this procedure.
8. Continue downloading all of the necessary patches, then execute them one at a time to
install them.
9. Executing the patch file, in this case DPWIN_00038.exe, presents the following dialogue:
10. Scroll through the dialogue to note any special requirements prior to completing the
patch installation; look for the section entitled “Special Installation Instructions”, shown
below:
11. Selecting Next (from the screen above) starts the Install Shield Wizard; select Next to
being the installation process (shown below):
12. The installation process verifies that no sessions are currently running; as shown below,
at least one GUI session was connected to the Cell Manager (dbsm is running). Be sure to
stop all sessions before continuing:
13. You will be prompted to save the original files, to allow the patch to be un-installed later.
The default directory for the saved files is: C:\Program Files\Omniback\Patched_files.
This directory will contain the executable named PatchUninstall.exe as well as the
original files that were replaced by the patch (see below).
14. After the patching process completes, start the DP GUI and verify that the patches were
installed on the Cell Manager as well as Installation Server. (shown below are the
properties for the Cell Manager; the properties for the Installation Server is essentially the
same screen)
15. Data Protector also provides a command line interface to verify the installation of
patches; execute the “omnicheck –patches” to see the listing. (see below)
16. Finally; after verifying that the Installation Server and Cell Manager are successfully
patched; upgrade the clients to distribute the patches to other systems in the cell.
2. The programs that begin with omni… make up the Command Line Interface to Data
Protector.
Try the following commands from the command line (DOS prompt):
omnicellinfo –cell
omnicc –query
NOTE: On windows, you may want to add the <OMNIHOME> directory to the
environment variables, so they are more convenient to use
3. Read the command reference (man page) for the omnicc command. How could you use
this command to perform the import/export functions you have just performed via the
GUI ?
Objective: To create a standalone logical device, initialize a tape and perform a backup of
some objects on the Cell Manager, then restore a single file.
1. Use the procedures presented in this module to perform a backup of some data on
your systems within your cell. If you have both Windows and UNIX clients in the cell,
include all of the following into a single backup specification.
/tmp
/home
c:\winnt\system32\drivers\etc
c:\temp
c:\winnt\temp
c:\Inetpub
c:\Perl.
2. You should:
• Configure a standalone logical device.
• Initialize at least one tape into the “Default DDS” media pool.
• Create a backup specification and save it as “backup 1”
3. Perform the required steps for backup (including all the checks listed in the slides).
Execute the backup in “Full” mode.
4. Remove your tape from the tape drive after the backup job has completed.
5. Prepare to perform a restore: If you want to restore a file, you will have to delete
something (be sure not to delete any critical files yet). Remove the following files
prior to the restore:
a. HP-UX:
rm –r /home/user1
rm –r /tmp/*
delete c:\Perl
delete c:\Inetpub
6. Perform the required steps for the restore of a file from your previous backup making
sure to create and respond to the mount request (as a result of the tape removal)
7. Schedule the same Backup Specification (from steps1-2) to execute tonight at 11:00
PM as a full backup. Verify that the backup is correctly scheduled by selecting the
Filesystem tab under Backup Specifications in the Backup context. Then examine the
Backup results area columns named “Scheduled” and “Backup Type.”
8. Create another backup specification for any part of your cell. Schedule this backup to
run at least two hours after your previous backup.
9. When you come in tomorrow, be sure to verify that your backup session completed
successfully. Proceed as follows:
a. From the “Internal Database” context, open the sessions folder, select each
session for details.
b. From the “Device & Media” context open the Default DDS pool, select each
tape listed, and verify the Objects (use the Objects tab).
c. From the “Restore” context open the Sessions and Objects folders, examine
the contents.
d. From the “Reporting” context, “Tasks” tab, execute the “List of Backup
Sessions” report from the “Sessions in Timeframe” folder; accept all of the
default parameters.
e. From the “Reporting” context, select “Web Reporting” from the Actions menu;
execute the same report as in the previous step.
Objective: The objective of this lab is to learn how to use the device and media
management features of Data Protector.
Objective: The objective of this lab is to learn how to use the media management features
of Data Protector, including:
It is assumed that you have configured a standalone device for your use from the “Getting
Started” module. This standalone device will be used during this lab.
1. Start the Data Protector GUI, and switch to the Devices and Media context.
2. When Data Protector is first installed, it provides you with a set of default media pools.
There is one default pool for each media type. We want to keep only the default pools for
the media types that we will be using within our cell. Otherwise, the list will be too
cluttered. Start by deleting all the pools that you will not be using. Typically, you will
keep only default DLT , DDS, and File, pools, proceed as follows:
7. Open the list of media pools in the Scoping Pane.
8. Right-click on the pool you want to remove. Select Delete from the pop-up menu.
9. You will be prompted, “Do you want to delete media pool xyz?” Click
Yes.
10. Unfortunately, you cannot delete pools en masse, so you will need to repeat this
process for each of the unwanted pools.
Edit-> Locations
b. Using the Location field, add a few locations, one at a time. (for example, “Offsite
Storage” , “Device Repository”, “Fire Safe”).
c. Examine the contents of the <OMNICONFIG>/vault_locations file. Is this file
modifiable?
b. You are now presented with the Media Pool Editor window. Complete the options as
follows:
Pool Name The name you want to call your Media Pool.
Use Yourname_test.
Description "A loose and appendable test pool"
Media type The media type for this pool. (The default is
DDS.)
Media Usage Policy Appendable (the default).
Media Allocation Policy Loose (default).
Media Condition Factors Leave these fields with the defaults.
Your pool will now be added to the list of configured media pools.
5. Create two further media pools, one for full backups and the other for incremental
backups. Call these media pools yourname_full, yourname_inc.
a. For the full backup pool, use <yourname_full> for the pool name and set the media
allocation policy to Non-appendable.
b. For the incremental pool use <yourname_inc> for the pool namd, and set the media
allocation policy to use Append Incrementals Only.
6. Use the Data Protector command line interface to get information about your new media
pools:
• omnicellinfo -mm
• omnimm -list_pool or omnimm -show_pools
• omnimm –list_pool <poolname> -detail
7. Initialize some media (use the Standalone device created in Module 4) and add them to
your media pools as follows:
a. Using the GUI, Select the Folder labeled “Media”
b. Open the test media pool by double clicking on it. You will now be presented with
the Media list in the Results area. The list will be empty, because you have no media
in the pool.
c. Load a tape into your standalone device.
d. From the menu bar choose: Actions -> Format... (Or, use the pop-up menu by
right clicking on the media pool name and select Format...
In the message section in the Results Area, you will see the media agent start to perform
the initialization of the media.
If the media is a recognized format, this procedure will fail, and you will need to repeat
the process to specify the Force option.
Upon completion of the initialization, you will see a dialog box telling you “1 out of 1
media was successfully initialized.” Click OK.
Your newly initialized media will now appear in the Media List.
Repeat this process to initialize the remainder of your media, but try to initialize at least
one of your media using the command line interface omniminit command.
10. When you have initialized all your media, look at what information is available for the
media and media pool.
(media properties)
A list of media pools of the same media type will be selectable in a pull down list.
Select the destination media pool by clicking on it; then press Finish.
The list of all media pools will be displayed in the Results Area.
Check that the media has been moved to the pool as you requested, then move it back.
You will be asked to verify that you want to export the media, click OK.
14. Make sure that the media you just exported is loaded in one of your devices. Then scan
the media using a logical device to check its status:
Open the Devices list in the Scoping pane
Select the logical device you want to use, from the pop-up menu select: Scan
(Do not eject after scan.)
In the Results Area screen you will see the scan take place. The media type will show
Data Protector Foreign.
Select the device where your media is loaded; leave the other options unchanged, and
click Next. (leave all of the other options to their default settings)
Select Finish.
Standalone
1. What Data Protector device type would you use to configure a single DDS/DAT drive?
3. What are the default block and segment sizes? What is the default concurrency?
4. Unless you want to re-initialize all of your existing tapes, do not change the block size for
the devices after the media is initialized.
6. Modify the device options so it will use the new DDS script after 15 minutes of waiting for
a tape to be loaded.
File Jukebox
Objective: create two multi-drive file jukeboxes, for performing more sophisticated backups
and automated media copy functions. The first jukebox may contain more slots than the
second and will be used primarily for load balanced backup. The second jukebox will only be
used to demonstrate AMO functions.
Note: this device type is a bit unusual, as it is not common to be used in a production
environment, but you will be able to exercise Data Protector functionality that you may
otherwise not be able to test with standalone devices, such as load balancing.
In order to perform this exercise you will need a substantial amount of free disk space; at
least 1GB is recommended. Ask your instructor about available disk space. This space could
be on either a Unix or Windows system with the Data Protector media agent installed.
Note: In some cases the /jukebox filesystem may already exist; if so, verify its
capacity and use it in place of <path> below.
If the /jukebox filesystem does not exist complete the following, or skip to step 1 if the
/jukebox already exists:
Using the Operating system tools (SAM on HP-UX, or Device Manager on Windows) create a
new partition and put a filesystem on it. For Windows 2000 this may be a drive-letter-path, or
just a drive letter; if using a driver letter path, first create a new empty directory such as
c:\jukebox. On Unix, use the volume manager and create a new mount point called /jukebox.
1. Create a “file” media pool using any unused name; accept all of the default options.
2. Create a directory under /jukebox (c:\jukebox) for each jukebox. Name them library_1
and library_2
3. Configure the logical device: (substitute <path> with the mount point for your partition).
4. Create at least two “virtual” drives for the Library/Jukebox. (more if you would like)
Drive-1
Drive Name: file_drive1
Description: virtual drive
ClientName: <your_host>
Pool Name: <your choice>
Concurrency: 2
Drive-2
Drive Name: file_drive2
Description: virtual drive
Host Name: <your_host>
Pool Name: <your choice>
Concurrency: 2
5. Expand contents of the newly created file_jukebox device in the Scroping Pane.
Select the "Slots" component, the properties will be displayed in the Results Area. Select
all of the slots in the Results Area (use Shift+Mouse, or Control+Mouse).
7. What Format does Data Protector recognize for the media that is in the file slots?
8. Initialize all of the file media into the an appropriate “file” pool.
Select all of the slots, the Actions -> Format…
Medium Size:
Note: be sure to allow for space in the partition for the second set of files to
be used by jukebox-2.
Objective: create additional devices to be used for AMO to be configured with backup; use
two file jukeboxes (or multiple drives in the same jukebox). Create a second file jukebox.
Requirements: two library devices of the same media type and media capacity.
1. Follow the procedure used previously for configuration of the file jukebox, with the
following exceptions (unless you have sufficient disk space for more):
a. Use <path>/library_2/file01-nn as the path/names for the slots; only 3-4 media
slots are needed to demonstrate AMO.
b. Create only a single drive jukebox
c. Format the media to the same size as the original file jukebox media (optional;
will be done by AMO if necessary)
d. Assign the drives in the new jukebox to the same pool as the original jukebox.
1. Using the backup specification created for module 4 that was named “backup_1”; change
the destination to use one of the drives in Jukebox_1. Apply the change to the
specification.
d. Select the source and target drives (choose only one drive from each of the file
jukeboxes)
e. Select Relative Time frame (media used within the last 1 hour)
f. Started within 1 hour
g. Duration 1 hour
h. Select the “All backup specifications” option (default)
i. Media Condition: Good
j. Media Protection: Any
k. Number of copies: 1
l. Eject media: “Do Not Eject Media”
m. Protection for target media: “Same as original”
n. Add a schedule so the Media Copy executes in the next 15-minutes.
o. Finish
2. Examine the configuration file for the scheduled media copy; use notepad or another text
editor to view the file from the <OMNICONFIG>\amo directory. The file should be named
“AMO_lab1.amcs” and is an ASCII file. Do not make any changes to this file.
3. Examine the schedule for the scheduled media copy; use notepad or another text editor
to view the file from the <OMNICONFIG>\amoschedules directory. The file should be
named “AMO_lab1.amcs” and is an ASCII file. Do not make any changes to this file.
4. In the Device & Media context of the GUI, open the Automated Operations container;
select the AMO_lab1 specification. Examine all of the tabs in the results area to view the
possible values for the media operation. Do not make any changes to this scheduled
operation.
5. Open the Monitor context of the DP GUI; watch for the AMO session to start. Continue
monitoring the progress to completion.
6. After the AMO completes, switch to the Internal Database context and review the session
that was created. Hint: Open the “Sessions” container to see the session ID’s.
7. Switch the context to Devices & Media; open the media pool containing the copied media.
Select the original (source) medium used for the replication process and view the
properties. Select the Copies tab to determine the name of the copy.
8. Select (within the media pool) the destination medium used for the replication. Notice in
the “General” tab of the properties results, there is a new button labeled “Original”. Select
it to view the name of the original medium. Select Ok to the pop-up.
9. end of lab.
NOTE Use the "cookbooks" presented in the appendix to accomplish the robotics
device configuration first if necessary, or ask your instructor for assistance.
1. Verify that the Media Agents are installed on the library host system.
2. Execute the Device Agent to determine if the SCSI devices are properly configured
(drivers and device paths are needed); note the paths to the respective devices.
/opt/omni/lbin/devbra –devices
Changer:
Tape Drive(s):
3. Start the Data Protector GUI; Select the Devices & Media Context.
4. Create a Logical Device of type SCSI-II Library and set the attributes shown in the table.
Edit -> Add -> Device
Client Name Select the host where the device is attached, and then
click OK.
Library’s robotic SCSI Type in the name of the control device file eg:
Address
HP-UX /dev/rac/c#t#d#
Windows: scsi2:0:5:1
Cleaning Slot If the instructor has supplied you with a cleaning tape,
you can use this feature. Normally, cleaning tapes are
loaded in the last slot. However, you can choose any
slot. Load the cleaning tape in slot 6 of the magazine,
select slot 6 from the list.
Media type The media type for this kind of device (Default is DDS).
5. When prompted “Do you want to create a Drive for library?” click Yes.
Set the following attributes:
Data Drive select the SCSI address or Device File for the exchanger
data (tape) device
Drive Index If this is the first and only drive in the library, enter 1.,
otherwise change the index as necessary.
Default Media Pool Choose an appropriate media pool, or use the Default
<Type> for now, and then create a new media pool and
change this field later
7. Verify that OmniBack can communicate with the library. Have it check the status of at
least one of the tape slots by scanning it:
NOTE Before performing the next step, make sure that no tape is currently loaded in
the drive. All tapes must be stored in the repository or magazine; otherwise,
the scan will abort. The Logical Device option “Busy Drive handling” can be
set to change this behavior, but the default is abort. You may want to interact
with the device using the Utility Media Agent (UMA) first to verify that it is
operational.
9. Select the library entry in the list of configured devices in the Scoping Pane, open the
contents, and select the slots component.Select one of the slots in the Results area and
then:
Actions -> Scan (barcode scan if available)
You will be asked to select the drive to be used for performing the scan operation. Select
the drive you have configured in this library, then click OK.
In the messages window you will see the media agent start and attempt to load the first
tape into the drive. When loaded, it will read the header of the tape, then unload it back to
its slot. It will repeat this process for each slot that you have selected.
When the scan operation has completed, you will see a dialog box, indicating how many
slots were scanned.
The messages window should contain a summary of the status of the slots, for example,
“empty, tar, fbackup, blank, etc. This should also be shown in the media status list.
NOTE The scan operation is very important. Without it, Data Protector has no
knowledge of what tapes are in which slots. Therefore a scan should be
performed after every manual movement of media to/from the library.
10. Using the command line interface, list the devices configured on the system and the
detailed information on each device using the following commands:
omnicellinfo –dev (-detail is also available)
omnidownload –list_devices
NOTE Refer to the online man pages or the appendix at the back for syntax.
11. Using the command line interface, perform a scan of a standalone drive and a library (if
you have a library available) using the following command:
12. Using the command line interface, modify the description field of one of your devices.
Use the following commands:
omnidownload
omniupload
Notes: Many HP Education sites have access to additional hardware either locally or
remotely. Verify with your instructor what systems/devices are available for this MSL
configuration before proceeding. In many cases these systems and devices are in remote
locations and are shared with other training sites.
This lab will verify SAN connectivity and then proceed to library configuration.
Library type:
Systems that have library access; use FQDN (media agent hosts):
1. Login: Password:
2. Login: Password:
3. Login: Password:
4. Login: Password:
Login: Password:
Login: Password:
1. Start a web browser on a local PC and access the Visual Manager Interface for the NSR.
http://<IP_address_of_NSR>
2. Login when prompted, using the login name and password collected earlier.
3. Select the “Report” link in the Main Menu. Scroll down the report to locate the “Discovery
Information” and “Mapping Information” sections.
4. Verify that the CHGR and TAPE devices are listed in the “Discovery Information” section
of the report.
5. Note the host maps in the “Mapping Information” section; there should be “Auto
Assigned” maps for each WWN of the HBA for the SAN connected hosts as well as an
Indexed map for the WWN of the FC switch. This indicates that all hosts in the SAN
should have access to the devices. If this is not the case, ask your instructor how to
proceed; you may have to manually build the maps.
6. Using the same web browser that was used for the connection to the Visual Manager for
the NSR, connect to the IP address of the MSL tape library (you may have to turn off use
of the proxy server for the web browser)
http://<IP_Address_of_the_ MSL_Library>
7. Login to the Remote Management Interface using the Password collected earlier.
Select Here
9. Check the status and configuration of the MSL Library. Note the number of drives,
number of slots, the status of the mail slot and number and locations of tapes.
10. Using the same web browser, connect to the Fabric Manager (B-series switches only) for
the FC switch used in the SAN that connects the library to the servers.
11. Select the “Admin View” folder; login using the information collected earlier.
12. Select the “Report” tab, then select the “View Report” button.
13. Scroll down the report until you see the “Name Server” section; verify that the NSR is
visible (may have different names depending upon the model used)
14. The hosts connected to the SAN must have Data Protector Media Agents installed in
order to be successful. Ask your instructor how to proceed if this has not been done
already. You may be able to add/import the clients into your existing cell.
15. (UNIX Clients only) Telnet to each client system that is connected to the library to verify
that the Data Protector Agents are installed.
telnet <unix_client> (login using root id, provide password collected earlier)
16. Verify that the client system contains device files for the MSL Changer as well as the tape
devices.
ioscan –fn (note the device name for the schgr device)
17. While connected to the prospective client system; verify that the Data Protector Agents
are installed.
18. If the DATA-PROTECTOR.OMNI-MA is installed, verify that the client doesn’t already
belong to an existing cell.
Notify your instructor if the client is already in use in another cell. Otherwise import or add
the client as appropriate to the Data Protector cell.
19. Execute the Data Protector Device Agent to test access to the MSL Changer and Tape
devices:
/opt/omni/lbin/devbra –devices
20. Repeat the “Client Verification Steps” for each system connected to the tape library.
21. Add the client systems into the Data Protector cell by importing them (preferred) or by
adding media agents to them. (Again, consult with your instructor before proceeding if
you are unsure)
22. Decide which of the connected clients will be the controller for the robotics and then use
the “Auto-configure” method to create the logical device configuration within Data
Protector: (proceed as follows)
a. Select the Device & Media context in the Data Protector GUI.
23. For the client that was not chosen for the Robotics Path create a configuration that
allows for direct library access in the event that the controlling host is unavailable;
proceed as follows:
a. Verify that the robotics device file has the same name on each client. If the names are
different, create a symbolic link or rename the device file so the names match exactly.
b. Execute the following to verify that the device file and library are available:
Libtab contents: (according to the example above and data collected from ioscan)
24. In the Device & Media context of the DP GUI, modify the library drives that are
associated with the system containing the Libtab file that was created in the previous
step; Enable direct access for each drive as shown below. (Troubleshooting tip: errors in
the Libtab file will prevent the enabling of direct access)
25. Select the MSL Logical Library and perform a “Barcode Scan”. Verify that the tapes shown
in the Remote Management Interface match those shown by Data Protector.
26. Verify the Lock name for each drive within the library.
28. Initialize all of the tapes in the MSL into the MSL_Library media pool; use the autolabel
feature and accept the defaults for the media capacity.
Notes: Many HP Education sites have access to additional hardware either locally or
remotely. Verify with your instructor what systems/devices are available for this additional
library configuration before proceeding. In many cases these systems and devices are in
remote locations and are shared with other training sites. Your instructor may choose to
demonstrate this if hardware access is limited.
Library type:
Systems that have direct attached library access; use FQDN (media agent hosts):
1. Login: Password:
2. Login: Password:
Changer:
Drive 1:
Drive 2:
Drive 3:
Drive 4:
1. Verify DNS configuration for the system before adding it to the cell by connecting to the
system and displaying the /etc/resolv.conf and /etc/nsswitch.conf files; make the
necessary adjustments to enable name resolution using DNS. Some sites have the DNS
configuration documented in the beginning of the /etc/hosts file, look there for the
needed parameters for the next step:
2. Open a terminal window; and telnet to the client system. Display the /etc/resolv.conf.
cat /etc/resolv.conf
search
nameserver
nameserver
3. Import the client system connected to the tape library into the cell. This does not verify
successful DNS configuration. Verify that the imported client is registered with the FQDN
of the client.
omnicheck -dns
5. Look for “cannot resolve” messages in the output of the omnicheck command. These
usually indicate an incomplete name resolution issue on one or more of the client
systems. Verify the /etc/resolv.conf contents are correct on all clients and then run the
check again
omnicheck -dns
6. Verify that the Media and Disk Agents are installed on the client and are the same version
as the cell manager. If the agents are older upgrade them from the Clients context of the
DP GUI.
7. Open a terminal window for the client system where the SCSI library is connected and
execute the Data Protector Device Agent to discover the available devices (see sample):
8. Start the Data Protector GUI, switch to the Devices and Media Context.
9. Select the Devices Folder, and from the Actions menu choose “Autoconfigure Devices…”
10. Select the client system with the Library connected; Select Next ->.
12. Use the Barcode Scan to determine the number of available media.
13. Create a new media pool that matches the drive types in the library.
14. Format 2-3 tapes from the available library repository into the new pool created in the
previous step.
15. Modify the drive properties for each drive in the library so that they use the Media Pool
created in step 11 as the default pool.
Objective: create additional devices to be used for AMO to be configured with backup.
Requirements: two library devices of the same type (we will assume file jukebox is the only
classroom solution)
1. Create a new filesystem/partition to be used for a file jukebox. This should be sufficiently
large to accommodate 3-4 media from your original file jukebox to allow for practice with
Automated Media Copy.
2. Follow the procedure used previously for configuration of the file jukebox, with the
following exceptions (unless you have sufficient disk space for more):
• Create only a single drive jukebox
• Allocate a minimum number of file (slots) such as three to four.
• Format the media to the same size as the original file jukebox media.
• Assign the drives in the new jukebox to the same pool as the original jukebox.
Lab 8: Backup
Objective: During the following labs, you will try out several methods of backup and the
options available. It is unlikely that all of the labs will be completed, you may pick and
choose ones that look interesting. For the backup tasks, use any available library or
standalone devices that you configured into the cell. This may be a DDS-Autochanger, MSL
tape library, DLT tape library, or file jukebox, or any other available device.
From the bottom of the Scoping Pane, Select the Tasks tab.
From the Scoping Pane select the Interactive Backup Wizard (no load balance)
d. Select the Next > button at the bottom to make the device selection.
Next>
Choose one of the devices that you created from the previous module.
Click the Advanced… button; Select the tab named “Other”; set the following option:
g. Select your object in the Summaries and then select the “Change device” button to
assign your object to a device. Notice the device name in the Device column.
h. At the summaries screen, just select Next>, or review the properties for the backup.
i. Save the backup as a new specification, choose any unused name for the file. Choose
to save it to a group with your initials as the group name.
In the Results Area, you will see the media and disk agents start. The object will go
into the running state, and the percentage complete indicator will go from 0 to 100 %.
A window will appear informing you when the preview is complete. Acknowledge the
message by clicking OK.
If the preview has worked, you have a good indication that the backup will work also.
1. Now we will show you how to get information regarding the backup you just performed
using the Command Line Interface:
a. Let's start with the ASCII source of the backup specification you just created.
Bring up a terminal/command window and use the following commands on the Cell
Manager system:
cd <OMNICONFIG>/datalists
ll or dir
Your backup specification should be listed.
b. Now take a look at its source with the command: (note, don’t make any changes)
more BackupSpecificationName (hp-ux)
e. Each session listed has a session-id, which is based on the data and a run number.
Identify one of your backup sessions, then get detailed information on it with the
command:
omnidb –session sessionid –detail
f. You should see the session report; it will be the same information you saw in the
Messages window when you performed the backup. You can also get object level
detail from the session:
omnidb –session sessionid –report
h. Discover the names of the objects that were backed up as part of the session:
Unix:
2. Now we will show you how you can get information regarding the backup you have just
performed, via the GUI:
c. Switch to the Internal Database context to see all previous sessions; open the
sessions tree to see the data.
You will now be shown a list of the previous sessions. It is the same list generated by
the omnidb –session command.
d. Identify a backup session and view the properties for it by double clicking on it. You
will see the objects backed up in that session. Double click one of the objects to see
its messages and other properties.
e. Find out what media was used for this session. From the tabs at the top of the Results
Area, select: the Media tab.
3. Now that you have identified the media that was used for the session, look at the media in
the media pool.
You will be presented with a list of the media that is in the pool, locate the specific
media that was used.
4. Now, take a quick look at what is available under the Restore window. We will not go into
much detail here, as restore is covered in the next module. The objective here is to
show you what new data appears after each backup you perform.
d. Open one of the systems to see the object backups for it by double-clicking on it.
e. Select one of the objects to see the filesystem browser appear in the Results Area of
the GUI.
f. Explore/Expand parts of the filesystem directory tree by clicking on the plus “+” sign.
Repeat this process to expand the subdirectories also.
When finished viewing the files, click Cancel if necessary then (optionally)
5. Perform another backup, but this time use the command line interface to initiate the
session. (Choose a short backup so as to not occupy the command line for too long)
6. Verify that the backup has completed successfully by using either the GUI or the
command line interface.
Select one of your backup specifications that will execute very quickly.
8. Select the Options tab from the top of the Results Area; Select the Advanced option
in the Filesystems Options section.
9. In the Options tab, fill in the full path to the pre/post exec scripts that you created.
Set the following:
Pre-exec /tmp/yourname_pre_exec.sh
Post-exec /tmp/yourname_post_exec.sh
Select High Network Load. (This step is irrelevant if you are not backing up over
a network).
Click Preview Backup.
TIP If you get critical errors, such as " exitcode Pre-exec failed with 127 =>
backup aborted!" or "Post-exec failed with exitcode 127 => backup
aborted!” make sure that the script contains the correct PATH to all the
commands that are contained in the script.
Error Code 126 is also very common. Check the permissions of the script.
It must have at least "rx. (read/execute)"
NOTE By default, Pre and Post-exec scripts at backup specification level are not
executed during previews. Set the ExecScriptOnPreview option in the
global file to change this.
Click Ok.
10. Check the output files that your pre and post-exec scripts generated.
You should be able to see the environment variables and their values discussed in this
module, i.e. SMEXIT, SESSIONID, etc.
Incremental Backup
11. Now perform an incremental backup of any object that you choose. This incremental
backup will be based on the full backup of the object performed earlier. (You will
need to do a full backup first, if you choose an incremental for a new object).
Again, decide which logical device to use (perhaps use a library if you have not done
so yet), decide what media pool to use and the pool properties (perhaps try append
incrementals only).
Start by making some modifications to the contents of the directory that you intend to
backup by copying or renaming some files within the directory. (choice of a temp
directory is wise, as this will not usually corrupt the operating system in any way)
c:\program files\omniback\bin\pre_exec.bat
Contents:
echo “Pre exec environment” > c:\temp\backup_pre.txt
set >> c:\temp\backup_pre.txt
echo “end of pre exec environment” >> c:\temp\backup_pre.txt
c:\program files\omniback\bin\post_exec.bat
Contents:
echo “Post exec environment” > c:\temp\backup_post.txt
set >> c:\temp\backup_post.txt
echo “end of post exec environment” >> c:\temp\backup_post.txt
Select one of your windows backup specifications that will execute very quickly (such as
the one used in the Data Protector Basics module.
7. Select the Options tab from the top of the Results Area; Select the Advanced option in
the Filesystems Options section.
8. In the Options tab, fill in the full path to the pre/post exec scripts that you created.
Pre-exec pre_exec.bat
Post-exec post_exec.bat
Select High Network Load. (This step is irrelevant if you are not backing up over a
network).
Click Preview Backup.
NOTE: By default, Pre and Post-exec scripts at backup specification level are not
executed during previews. Set the ExecScriptOnPreview option in the
global file to change this.
9. Check the output files that your pre and post-exec scripts generated.
You should be able to see the environment variables and their values discussed in this
module, i.e. SMEXIT, SESSIONID, etc.
Incremental Backup
10. Now perform an incremental backup of any object that you choose. This incremental
backup will be based on the full backup of the object performed earlier. (You will need to
do a full backup first, if you choose an incremental for a new object).
Again, decide which logical device to use (perhaps use a library if you have not done so
yet), decide what media pool to use and the pool properties (perhaps try append
incrementals only).
Start by making some modifications to the contents of the directory that you intend to
backup by copying or renaming some files within the directory. (choice of a temp
directory is wise, as this will not usually corrupt the operating system in any way)
The remaining lab is designed to allow you to experiment with many of the features not
yet covered by the previous labs. Since by now you should have a little experience with
the GUI, the individual steps are left out.
Your instructor will guide you if necessary, but this lab is designed to allow you to
experiment on your own.
1. Create and preview a backup specification that does a full backup of all the hosts in
your cell using load balancing. You will require at least two logical devices. You may
want to use the file jukebox that we created earlier in the course. If you have very
limited disk space, you may want to add only a few file systems, not the entire host to
the backup specification.
2. Schedule a Full backup of the cell manager to be executed at midnight. The backup
should run each night starting tonight. Exclude the Data Protector database from the
file system tree where it resides.
3. Schedule an incremental backup to run for the cell manager. The backup should
execute at least 3-4 hours after the full backup from step 2. Use any incremental level
that you would like. You may re-use the backup specification from the previous step,
or create a new one.
4. Modify a file before the backup starts. Modify the hosts file, either /etc/hosts, or
c:\winnt\system32\drivers\etc\hosts, add an additional comment to the top of the
file.
When you arrive the next day, check the status of your backups.
5. To finish this section we will take a quick look at Templates and Groups.
6. Create your own template, by opening up the Templates collection in the Scoping
Pane (this is in the Backup Context). The options that you select are just a sub-set of
those used for the Backup Specification. Experiment with the View menu in the Menu
Bar, try changing the views and note the results in the Scoping Pane (open all the
component trees within the Scoping Pane)
Choose from:
7. Organize your backup specifications using the grouping feature so that you can easily
identify your backup specifications. Create two to three groups to exercise the
capabilities.
1. Use the omnicreatedl command to create a data list that contains the file systems on
your cell server. Give the data list an appropriate name.
b. Which directory did you look in to view your new data list?
c. What must be done with the data list produced by omnicreatedl before it may be
used for backup?
b) Use the GUI to view the /tmp file system object that now resides on the tape?
c) Looking at the Start Time and End Time, how long did the backup take to run?
Objective: The objective of this lab is to practice with some additional Data Protector
backup features.
1. If you have an HP-UX server, perform a rawdisk backup of any logical volume. This
backup should use a pre-exec script to umount the filesystem and a post-exec to mount
again the filesystem after the backup completes.
2. If you have an HP-UX server, perform a full system backup to the autochanger. What are
the causes of all of the warnings? Is this full backup useful for a system recovery?
3. If you have a Windows 2000 system with an internal tape drive create a full system
backup using the local tape drive. Explain what to do with the “Minor” error messages
that are produced. Could this full backup be useful for disaster recovery, even with these
error messages?
4. What is in the CONFIGURATION object that is part of a full Windows backup? Try using
the Restore context to browse its contents after the previous full backup completes.
5. Experiment with deleting and restoring some files that were previously part of a full
backup from either Windows or Unix. (This may take some time to complete, if you have
to schedule the backups from steps 2 and 3 to run during the evening.)
Lab 9: Restore
Objective: The objective of this lab is to learn how to Data Protector restore features.
Objective: During this lab you will perform the Data Protector restore function and use
some of available options, including filesystem restore.
Note! You may have performed various different backups to get to this point of the course.
By now, you should be aware of the Restore GUI, and its general options from the previous
discussion. This lab will intentionally be very unstructured to allow you time to experiment.
Restore is usually quite simple; in most cases, it is a matter of deciding what to restore, to
where and from which backup. This lab assumes that you have taken at least one full backup
of your cell manager.
1. Let’s start with a simple single-file restore. On your local host file. Restore the file
/etc/hosts or C:\winnt\system32\drivers\etc\hosts.
First, look at the contents of the hosts file. Add a line to the top of the file by using one
of the system editors (vi, or notepad) if you have not done the previous lab that included
this step.
After you examine the file, make a copy of it in the same directory; choose an appropriate
new name, such as host_copy.
Restore the file from the full backup that included the hosts file.
Restore1: Restore the file into a new directory, use the option to list restored files.
Restore2: Restore the file to the original location, use the default options.
Verify that the files were restored in each case, or explain why they were not.
2. You have performed three restores. Each one was performed by selecting a single
file-system component. Now you should try a parallel restore that will restore files from
three separate objects in the same restore session. This will not be possible on Windows
unless you have multiple partitions; HP-UX has multiple mount points by default.
You may want to delete some non-critical files before the restore starts, then verify that
they were restored. (non-critical files on HP-UX may be found within the /home
filesystem.
3. You will restore a single file, but you will not be given the specific location. Use the Data
Protector Restore GUI to restore: rgb.txt to the latest version on HP-UX and trace to
Windows. Restore the file to the original location and overwrite the one on the system, as
if it is corrupted.
Hint: use the Restore by Query from the Restore Context of the GUI.
4. All the restores you have performed up to now have used the GUI. Now you will learn
how to use the command line interface, omnir command, to perform the same point in
time restore that you have just performed. The examples below will give some guidance
on how to construct the command line for the restore. Your data will be different, so use
the following as a reference only.
# omnidb -filesystem
Object Name Object type
===================================================================
scorpio:/ 'Nigels /etc backup' FileSystem
scorpio:/ ' [/]' FileSystem
scorpio:/home ' [/home]' FileSystem
scorpio:/Data Protector ' [/Data Protector]'
FileSystem
scorpio:/opt ' [/opt]' FileSystem
scorpio:/stand ' [/stand]' FileSystem
scorpio:/tmp ' [/tmp]' FileSystem
scorpio:/usr ' [/usr]' FileSystem
scorpio:/var ' [/var]' FileSystem
The preceding command will produce a listing of all the backup sessions that contain
this object. Insert your hostname:mount point and description into the command and
review your output. You will see something like this:
From this listing, you can see the two previous backup sessions. The larger one (in
KB size) is the full, and the smaller is the incremental. You want to restore to the
point in time of the incremental, so make a note of the SessionID. In this example, it
is 1999/12/12-2.
b. Now with the above information you can initiate a command line restore:
# omnir -filesystem scorpio:/ 'Nigels /etc backup' -session 1999/12/12-2 -tree /etc –catalog
[d:0] /etc
[f:0] /etc/hosts.bkup
The preceding command will produce a listing of all the backup sessions that contain
this object. Insert your hostname:mount point and description into the command and
review your output. You will see something like this:
From this listing, you can see the two previous backup sessions. The larger one (in
KB size) is the full, and the smaller is the incremental. You want to restore to the
point in time of the incremental, so make a note of the SessionID. In this example, it
is 2003/09/17-2.
c. Now with the above information you can initiate a command line restore:
# omnir –winfs pc99.dow.edunet.hp.com:/C “C:” –session 2003/09/17-2 –tree /temp –catalog
• Restore a file to a different system (the client system does not need to be in your cell)
• Try restoring a busy file
• Restore an entire system, all objects in parallel
Objective: The following lab demonstrates the management tasks for the Internal Database.
As discussed earlier, the Internal Database is critical to the operation of the product. It
should be protected by being backed up.
It is essential that a backup of the database is performed on a regular basis, before any
manipulation of the database, such as write_db/read_db or adding database extension files.
Backup using a standalone device (this may be a remote device, such as one connected to
Windows system, as long as it is a device you will have physical access to)
Backup the database daily, after all of the other backups have completed.
Procedure:
1. Create a new Media Pool to be used exclusively for the Internal Database backups. Add at
least one tape to the pool.
2. (optional) Create a new Logical Device that is assigned to use the pool created in step 1.
Be sure to use this device for the backup. (This must be a device that you will have
physical access to for this lab; a local Windows device for the remote Cell Manager is
Ok).
3. Create a backup specification to perform a full backup of your Internal Database. Protect
the backup for one month.
4. Open a terminal window on the Cell Manager and execute the following command to
display the list of directories that are to be included in the Internal Database backup:
omnidbutil –list
5. Perform an interactive backup of the database using your newly defined backup
specification.
6. Monitor the backup to verify that it completes successfully. Remove the tape from the
tape drive and physically label the tape as Data Protector IDB Backup.
Now that you have a backup of the database, you can continue with the remainder of
the lab, secure in the knowledge that if anything goes wrong, you can recover from this
backup!
Extension files allow the database to be extended beyond its default capacity. The file that
occupies the most database tablespace is:
fnames.dat This file contains information on the names of the files backed up. Typically,
this file occupies about 20% of the Internal Database. The default limit of the
file is 2 GB; maximum size is 32 GB, in up to 2047MB increments.
omnidbutil -info
Save a copy of the command's output. You may need this later.
From the command's output, record the disk space used for Filenames:
omnidbutil -extendinfo
fnames.dat
3. Add a 500 MB extension to the base file fnames.dat . (replace OMNIVAR with the
correct path to the CDB on your Cell Manager.
or
omnidbutil -extendinfo
Objective: Add two additional DCBF directories to the cell. Add one directory using the
GUI, and add another using the command line.
Using the Internal Database context, add one additional DCBF directory as follows:
a. Open the Usage folder, and select Detail Catalog Binary Files, right click to get the
pop-up menu and select Add Detail Catalog Directory
c. Enter the Path to the directory where you will place the DCBF extension. (this should
be in a filesystem that has at least 100 MB available) Choose the name for the
directory to be DCBF_2.
g. Select Finish.
2. Change the allocation algorithm in the global file; DCDirAllocation=2. What is the
meaning of this change?
3. Perform a few (at least 3) quick backups, perhaps to your file jukebox. Check the
contents of the newly created DCBF directories. What are your observations?
4. Move a file from DCBF_3 to DCBF_2. Is there any way to notify Data Protector that this
has been done? Check the man-page (word pad document) for the options to omnidbutil.
Inform Data Protector of the change, and then use the correct command to remove the
DCBF_3 directory from Data Protector.
1. Using the “Data Protector Database” context of the Scoping Pane, purge any failed or
aborted backup sessions:
Select some or all sessions of status Aborted/Failed. (Use the [Ctrl] key to
select multiple sessions.) From the pop-up menu:
Remove Session
Click Yes to the prompt, "Are you sure you want to remove these
<num> sessions?"
When the purge has finished, check the database utilization again, using the
omnidbutil –info command. Compare the output to the previous time you
ran the command.
What information has changed?
omnidb –session
3. Using the GUI, remove only the session messages for a selected session.
4. Using the command line, purge restore sessions (including detail catalog information)
older than 1 days:
1. Make a backup copy of the original file before proceeding. Use the operating system copy
command. (just in case…)
3. Note the use of the following parameters, as well as their default values:
KeepObsoleteSessions:
KeepMessages:
DailyMaintenanceTime:
DailyCheckTime:
Objective: This recovery will focus on the automated recovery capabilities. You can also
use the manual recovery, if this is desirable, use the procedure outlined within the student
notes of this module.
6. Confirm the media prompt by typing OK, unless the medium indicated does not match the
tape you have. See note below.
NOTE If the media you have available is different than the one that you’ve been
prompted to confirm, enter cancel at the prompt; then create a new database
backup. Once you have a recent backup of the database, start this process
again.
7. Once the database has been successfully restored, check the status of the Data Protector
services.
8. Start the GUI and verify that the Devices, Media and Database content is as it was at the
time of your backup.
9. Open the “global” file and alter the value of the RecoveryIndexDir parameter to contain
the path to the login directory of the system administrator user. For Unix this may be
/root, and for Windows 2000 it may be C:\Documents and Settings\Administrator.
12. Verify that the new obrindex.dat file was created; compare the date/time of the new file to
the obrindex.dat file that exists in the <OMNIVAR>/db40/logfiles/rlog directory.
14. Perform another restore using the omnidbrestore –autorecover command. Which
obrindex.dat file is used for the session?
15. If time allows, investigate some of the other possibilities for omnidbrestore; such as how
to manually replay the transaction logs, and what to do if you don’t have the obrindex.dat
file. See the man-page for omnidbrestore.
Objective: This lab (while no longer necessary) offers some possibilities to export and then
re-import data within the Internal Database. This may be useful to execute at periodic
intervals to maintain peak performance after long periods of record add/record delete cycles.
These cycles are completely normal, but may fragment the database over time.
1. Create a temporary directory (this can be any directory that has sufficient space to
contain the Data Protector DB):
2. Make sure that no Data Protector activity is taking place; this includes exiting from all
GUI’s since these use a DBSM to connect to the database.
omnidbutil -info
Expand the Usage tree. Try selecting the various subentities. You should see similar
information to that seen with the omnidbutil command, plus information in graphic pie
chart form in the Disk Usage tab.
5. Write the Internal Database information into ASCII format into the temporary directory.
(Purged data will not be written to the ASCII format, only valid data).
omnidbinit
When the database has been reinitialized, it will be empty, and will contain only the
default media pool definitions. Check the size again using omnidbutil.
7. Read the ASCII formatted data back into the empty Internal Database:
8. Stop the Data Protector services, and copy the msg and dcbf contents back into their
original locations.
10. Verify that your database is now occupying less disk space:
Compare the disk space usage in this output with the previous output.
omnidbutil -info
11. Using the GUI, review database size and record utilization:
From the Objects tab, expand the Usage tree. Try selecting the various subentities.
You should see similar information to that seen with the omnidbutil command.
View the information in graphics pie chart form in the Disk Usage tab.
Objective: To relocate the Internal Database to overcome disk space limitations that are
encountered over time. This lab will be performed on HP-UX where the LVM
capability exists. (A similar procedure may be executed on Windows 2000)
1. Create a new logical volume (partition) that is at least 100 MB (for this lab).
lvcreate –L 100 –n dpidb /dev/vg00
newfs –F vxfs /dev/vg00/rdpidb
7. Copy the contents of the “old” database to the new mount point
cp –rp /var/opt/omni/db40_temp/* /var/opt/omni/db40
9. Verify (using the GUI) that all of the session and media data are present and accessible.
10. Reclaim the disk space consumed by the old database directory.
rm –r /var/opt/omni/db40_save
Objective: The objective of this lab is to learn how to use the command line and GUI to
monitor and interact with activities within the cell.
2. Start the Data Protector monitor GUI, switch to the Monitor context.
What information is provided within the Monitor while sessions are running?
•
•
•
•
•
•
At the detail level, you are monitoring the individual session. You should see the status of
all the backup objects and devices, as well as progress messages.
5. Now, use the command line interface to get the same information you have just obtained
through the GUI.
6. Bring up a terminal window or command prompt and issue the following command:
omnistat
7. Display the session information on the running backup session with the following
command, where sessionid is the one reported from the omnistat command (your
running test backup).
omnidb –session <sessionid> -report
You should see very familiar output to the GUI from this command.
The command returns to the prompt and gives you a snapshot of the running session and
its state at that point in time. This is not the same as the GUI monitor; it shows the
progress of the session as long as the monitor GUI window is open.
Monitoring the session through to completion using the command line is also possible.
Issue the following command:
You will notice that the command does not return immediately to the prompt as it is
waiting for output from the session. Only new session messages are displayed, not the
entire session log. You may have to wait a while before you see a session message.
You will see a list of the previous sessions that have been run in your cell.
11. Double-click on one of the old sessions to see the details of the session.
Find out what tapes were used in this session.
Select Properties
15. Open a terminal window on your workstation, and execute the following command:
omnidb -session
This gives a listing of previous Data Protector sessions. The listing will contain only
“valid” sessions, as sessions are removed during the database maintenance process.
18. Obtain a listing of the backup sessions run in the last day:
omnidb –session –type backup –last 1
20. Produce a list of all of the file systems that have backups in the cell:
omnidb –filesystem <or> omnidb -winfs
21. Produce a detailed report describing what was backed up in the latest session.
Record the session ID and one of the object names from the report. The object name for a
file system object is in the form:
22. Produce a report showing what files were backed up by this session.
(You will need the session ID and the object name you have recorded.)
omnidb -filesystem <object_name> -session <session_id> -catalog
Note: the description part of the object name for Windows objects must
be enclosed in double-quotes, even though the session report shows
single quotes!
Objective: The objective of this lab is to learn how to use the Reporting and Notification
features of Data Protector.
c. Fill in the name for the report group, do not schedule the report group yet.
Protection: Any
Libraries: Choose any library or jukebox device that you have (dds or file)
Duration: 12hours
Format: ASCII
4. Select (right mouse button) the newly created, select “Preview” to execute the report.
Protection: <Any>
c. Using a terminal window view the two text reports that were created
8. Schedule the report group to be run daily. You may want to schedule it to run at the next
closest 15 or 30 minute mark. (In other words, check the current time, and schedule it to
run soon).
a. Select (right mouse button) the Event_Lab1 report group
b. Select “Properties”
c. Select the schedule tab in the Results Area.
d. Choose the Add… button near the bottom of the Results Area.
e. Select Daily, and fill in a time that is within the next 15 minutes
9. Add the report group as a notification event that will be executed with every backup job.
Notification definitions are stored in the file <OMNICONFIG>/Notifications. Look
at this file to see how your current (default) notifications are configured.
a. In the Scoping Pane, select Notifications with the right mouse button.
11. After the backup completes examine the directory to see that the reports were generated
by the notification event.
12. Examine all of the Notifications that already exist, as well as the parameters that control
their execution.
13. Look at the default notification for the Data Protector Database Space; notice the default
limits and thresholds.
15. Modify the default event for the Database Space Low that goes into the Data Protector
Event Log. Set the thresholds sufficiently low so that the event is triggered. Execute the
omnitrig command to trigger the event.
16. Examine the Data Protector Event Log for the events that have occurred. What has
caused the events to be written to the log?
17. Clear the events out of the Data Protector Event Log by selecting the Event Log within
the Scoping Pane using the right mouse button, select Empty Event Log. Can the entries
still be read? Which file contains a record of the past events? Open the file to examine its
contents.
Windows: Create an additional directory alias on the Windows Peer IIS web server:
1. As Administrator select:
Start -> Programs -> Administrative Tools -> Internet Services Manager
2. Open the Container representing your system.
7. Browse and Select the C:\Program Files\Omniback\java\bin directory, select Ok, then
Next.
8. Enable Execute permission in addition to the already selected Read and Run scripts
permissions.
HP-UX: Create an additional alias for the default Apache web server running on port 80.
1. Edit the system file that controls the automatic startup of the Apache Web Server:
vi /etc/rc.config.d/hpapache2
APACHE2_START=1
3. Save the file and exit the editor. ( you will have to use :wq! to save the file since it is read-
only)
4. Stop the Apache Web Server (if it is running; this may not be necessary if the
Apache_Start parameter was 0)
/sbin/init.d/hpapache2 stop
5. Edit (vi) the /opt/hpapache2/conf/httpd.conf and add the following in the Alias definition
section of the configuration file (be careful where these lines are added).
Look for the section in the file that starts with: # Aliases: Add here
There are some existing Alias definitions, be certain not to mix this new entry in the
middle of an existing definition.
/sbin/init.d/hpapache2 start
Start Internet Explorer (access either the Windows web server or UNIX web server)
Address: http://<FQDN>/DP_Report/WebReporting.html
1. Using the Web interface, configure a notification that sends a Broadcast message to a
Windows system (If no NT system is available, use email to an HP-UX system):
a. Log in to the cell using the Web interface. Select Notifications, then Add
Notification. Fill in the boxes as follows:
Object <ALL>
The outcome is that when any session finishes, notification of this fact is
broadcast or emailed to the system specified.
Test your notification by performing a simple interactive backup of a single file.
Note that broadcasts can be sent only to NT systems!
2. Configure a notification to send a report of the media used when the session completes,
in the form of a broadcast or email message:
a. Using the Web interface, expand the Single Session report category and
select the Session Media Report. You will then be asked to select the
specific session for which you want to generate the report. Unfortunately, you
cannot select a blank session, so accept the one that is displayed by default,
and click Show.
b. The report is displayed. Click Save Report, and then enter a Report Name,
such as mediareport. click the Save to New Report Group button, and
enter a name for your new report group, such as mediareportgroup, then
click Define and Save Report.
c. Select Broadcast or Email. Set the report format to ASCII. In the Send
Report To field, enter either the hostname for a Broadcast or the email
recipient, and press Add, then Save Report.
d. You need to make a modification to the report definition file so that it is not
reporting on a specific session (this way it will report on the session it is run
for).
e. Using an editor, edit the report group file you just created. The file will be
named as you specified and located in the <OMNICONFIG>/rptgroups
directory. Locate the line that contains the session specification, (i.e.
SESSION “1999/12/09-6”), and delete it. Save the file. The file should
look something like this when edited:
NAME "mediareportgroup"
{
REPORT "mediareport"
{
ID "session_media"
BROADCAST
{
TYPE ASCII
TO
"saturn"
}
}
}
When using the Data Protector GUI to define the same report, this
modification is not required, as it is possible to leave the session name blank.
4. Try some notifications of your own using different formats and delivery methods.
Remember that report groups can be used to send a combination of reports upon a
notification event.
Objective: The objective of this lab is learning how to assign specific privileges to users by
using the Data Protector Groups and Users features.
User Configuration
Consult your instructor about the possibility of using an existing operating system login in
order to bypass step 1.
1. Add a new user to the operating system, preferably not an administrator, but an end user
or operator level. (Using the HP-UX System Administration Manager (SAM) utility, create
four new users called <yourname>1 – <yourname>4.) Be sure to test the login
capability of the user before proceeding to the next step.
2. As the new user, try executing the Data Protector GUI. By default, you should not have
any permission; exit Data Protector and then logoff, and back on as the system
administrator.
Make sure you are logged on as root; then start the Data Protector GUI.
Edit->Add->User Group
The Add Group dialog box will appear. Fill the boxes as follows:
Add your first user yourname1 to the Admin group. Select the Admin group, then from
the menu bar select:
Edit->Add->Users
You will then be presented with the user add screen. Fill the boxes as follows:
Click (…) and select the user you want to add from the
User Name list, the list is simply from the /etc/passwd file.
Group / Enter the hp-ux group or Windows domain that the
Domain user belongs to, (i.e. users). You can find out the
Name group name by logging on as the user and typing id.
Real Name Test user <yourname>X.
Host Name Click (…) and select the system on which the user
resides. The list shows all systems in the cell.
User Group Click (…) and select the Data Protector group to
which you want to add the user, that is, Admin.
Unix or Click on UNIX User.
Windows
User
Now repeat this process for users yourname2-4, assigning the users to the following:
User Group
<yourname>2 Operator
<yourname>3 User
<yourname>4 Student
Start the Data Protector GUI as this user: (the PATH to the command may need to be
entered)
eg: /opt/omni/bin/xomni
Does the Data Protector GUI appear as it did when you ran it from the root user?
Do you have access to all the functional areas via the GUI as before?
Why is this?
Does the xomni GUI appear as it did when you ran it from the root user?
Do you have access to all the functional areas via the GUI as before?
Why is this?
Does the xomni GUI appear as it did when you ran it from the root user?
Do you have access to all the functional areas via the GUI as before?
Why is this?
Does the xomni GUI appear as it did when you ran it from the root user?
Do you have access to all the functional areas via the GUI as before?
Why is this?
Objective: Work with another team using a different Data Protector cell; limit access to
prevent restores between cells.
1. Try to restore some of your files into a temporary directory to a cell client that belongs to
the other cell team that you’re working with. Were you successful? Why?
2. Try to perform a backup of any filesystem from the system you restored to. Were you
successful? Why?
3. Try importing that other client into your cell. Are you successful? Why?
4. Secure each cell to prevent outside access to your clients. Be sure to specify your cell
manager as a trusted server. What if you forget to add yourself?
5. Is there another way to secure your cell from remote restores if the system is HP-UX?
Objective: In this lab, working with a partner, you will configure one of your systems as the
MoM server and the other as a MoM client.
Next, you will centralize the media management databases so that the MoM client gets its
entire media and device information from the MoM server. This will enable you to share
devices and media between cells.
Decide whose system will be the MoM cell server. The other system will be the cell server for
the client cell:
1. Log on to the cell manager you want to configure as the MoM manager as user root.
2. Create an empty file in the /etc/opt/omni/cell directory:
cd /etc/opt/omni/cell
touch mom_info
chmod 600 mom_info
ll mom_info
Make sure the file has permissions as shown. It will eventually contain the list of
managed cell mangers. This file cannot be manually edited.
/opt/omni/sbin/omnisv stop
/opt/omni/sbin/omnisv start
5. Add the root user of the MoM cell to the admin group so that root can use the
xomnimom GUI to manage the remote cell.
Group sys
Now, import your client cell manager into the MoM environment:
10. In the Scoping Pane, highlight Enterprise clients, and from the pop-up menu select:
Import Cell Manager
11. Enter the fully qualified hostname of the client cell server and click OK.
This results in:
The imported cell manager should appear in the list of cell servers on the left of the
screen. Verify that the imported client cell has been added to the list.
Familiarize yourself with information available through the GUI by selecting the various
contexts from the scoping pane.
Before sharing can occur, you must merge the media management databases of the client
cells with the MOM server:
As root on the MoM server, merge the client MMDB with the MoM’s MMDB:
This will extract information from the client's MMDB and merge it with the existing
MMDB on the MoM server.
Example:
# /opt/omni/sbin/omnidbutil -mergemmdb blake7
When complete, run the GUI on the MoM server, and edit the duplicated names of media
pools and devices (in the user interface). The duplicated names have a “ _N” appended to
their name, where N represents a number.
This always happens to the default pools if they exist in both cells. In this case, you must
manually change the backup specifications that use these devices to use the new device
names. It is a good idea to add a line to the media pool’s description to indicate from
which cell the pool has come.
Log on to the cell manager of the client cell as user root or administrator.
omnisv stop
omnisv start
omnicc -update_mom_server
Now that the database has been merged, the MMDB on the client is no longer required,
however, save it by moving to another location:
Notice how the client cell is now "indented" under the MoM server.
Take a look at the devices and media views for both systems. You should see that they
are now the same. Also, try running the Data Protector GUI on the client system and look
at the devices and media section (this information is now being supplied from the MoM.
Try performing a backup of some files from the client system to a device that is located
on the manager system (device sharing).
From the MoM GUI menu bar, select the Clients context, highlight “Enterprise Clients”
and select Distribute Configuration from the pop-up menu.
In the list of cell managers, select the client cell manager to which to distribute.
Click Distribute.
The configuration file is sent to the client cell manager.
15. Check on the licenses configured for the MOM server. If you are using the instant-on
license that ships with the product, you will not be able to distribute the licenses as
described below; stop here. If you have a “real” license proceed as follows:
Note that some steps must be performed at the client cell server, while others are
performed on the MoM server.
Configure the client cell manager to use centralized licensing by performing the following
steps:
Check the current status of licensing on the client, using the following command on the
client system:
omnicc -query
Note that the Licensing Mode is set to Local.
Example:
# omnicc -query
Using the MoM GUI, configure the client to use Remote (centralized) licensing:
From the Scoping Pane, select the MOM server system and from the pop-up menu, select:
Configure Licensing
From the servers list, select the client system by clicking on it.
Select the Remote mode button. (It should currently be local.)
At this stage, you could run the omnicc –query command again on the client system.
You should now see that zero licenses are available, and that the license mode is now
Remote.
In the MoM GUI, the Type of Licensing screen changes views and shows two
columns for each license, which indicate how many licenses are available, and how may
have been allocated to the client system.
In the Allocated column, set the number of Single Drive for UNIX licenses to
be half of that available, (i.e. 250) by either typing this in (recommended!), or using the
up/down arrows. Also, select an Online Extension license. (If you are using NT, use
the appropriate licenses.)
When ready, click Redistribute.
Again, on the client system, try the omnicc –query command. You should now see that
the licenses have been configured.
1. Configuring the DAT Autoloader or DLT Library using the SCTL driver
The following procedure can be used to install the connection to a tape library
on an HP-UX 11.x system. Knowledge of system administration is required.
This procedure will make modifications to your kernel.
Procedural Outline
1. Make sure you have the required driver(s).
2. Determine the hardware path for the device.
3. Shutdown the system.
4. Connect the new device.
5. Create device file(s) for the autochanger/library unit.
6. Test the operation of the device.
Overview: You should not need to build a new kernel if you have any other
device on your system that requires the SCSI PASS-THRU driver sctl.
Sctl is used along with the SCSI interfaces and is in the HP-UX kernel by default.
There is also man-page that may help (man 7 scsi_ctl)
Check if your kernel definition already has the sctl driver installed:
# ioscan -f -k -C ext_bus
Class I H/W Path Driver S/W State H/W Type Description
===============================================================
ext_bus 0 8/0 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
ext_bus 1 8/4 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
ext_bus 2 8/8 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
ext_bus 4 8/16/0 CentIf CLAIMED INTERFACE Built-in Parallel Interface
ext_bus 3 8/16/5 c720 CLAIMED INTERFACE Built-in SCSI
Note: the "I" column (instance number) of the 8/16/5 interface. For our system the
instance number of this interface is a 3. We will need this for the device file later.
B. Select one of your SCSI busses, scan the bus for available addresses:
The H/W Path for this example is: 8/16/5 (from a D-class system)
# ioscan -f -k -H 8/16/5
C. On the 8/16/5 bus, addresses 0, and 2 are used; we will use 4 for the
autochanger. The hardware address including the SCSI target of the autoloader
will be: 8/16/5.4. Note; if the device is a DDS Autoloader, this will be the
SCSI target for the tape drive as well as the autoloader since SCSI lu’s are used
for the internal devices within the unit.
A. Shutdown the system to allow connection of the new device, and execution of
the new kernel.
# cd /
# shutdown -h (power off the system when instructed to do so)
4 7
Scsi Address
Autoloader Options
A. Connect the new device to the system, then power all peripherals on and power up
the system last.
B. When the system reboots, scan for the new device in the ioconfig:
# ioscan -f
# lsdev -d sctl
Character Block Driver Class
203 -1 sctl ctl
note the “Character (major) number for the sctl driver, 203)
Overview: The tape drive in the autoloader will autoconfigure at reboot time. We will
only need to build a device file for the autochanger unit using the scsi
pass-thru device driver. We will use the mknod command for this purpose.
A. Gather the information needed to create the device file for for the robotics:
Device Name: /dev/scsi_library (any name will work)
Device Type: c (for character type devices)
Driver Major#: 203 (output from the “lsdev -d spt” command)
Device Minor#: 0x034100 (03 from the scsi card, target id 4, lu 1, zero fill)
B. Manually bind the sctl driver to the device using the ioscan command (this is
optional, but makes the ioscan output look nicer):
# ioscan -f -H 8/16/5.4.1
Overview: HP OmniBack II provides uma (universal media agent), a utility that will
get installed with the media agent module. The utility is used for
interacting with scsi changers. We can test the new device file we built in
the previous section to be sure that it allows us to pass commands to the
autochanger. Once the test is verified, configure the Logical Device.
1. Start the uma utility using the -ioctl option for the new device file:
/dev/scsi_library> inq
SCSI Inquiry:
Type: 8
Vendor ID: "HP "
Product ID: "C1557A "
F/W Revision: "U709"
/dev/scsi_library> stat
Element Status (T=transport, X=Im/Export, D=Drive, S=Storage):
1 D1 Empty ""
2 S1 Full ""
3 S2 Full ""
4 S3 Full ""
5 S4 Full ""
6 S5 Full ""
7 S6 Full ""
/dev/scsi_library> exit
Configuring the DAT Autoloader or DLT Library using the SPT driver
The following procedure can be used to install the connection to a tape library
on an HP-UX 11.x system. Knowledge of system administration is required.
This procedure will make modifications to your kernel.
Procedural Outline
1. Make sure you have the required driver(s).
2. Determine the hardware path for the device.
3. Build a new kernel (if necessary).
4. Shutdown the system.
5. Connect the new device.
6. Create device file(s) for the autochanger/library unit.
7. Test the operation of the device.
Overview: You will most likely need to build a new kernel if you have no other
device on your system that requires the SCSI PASS-THRU driver or SCSI
There is also man-page that may help (man 7 scsi_pt)
Installation for a server (E,G,H,I,T,some K’s) systems using HP-PB bus slot for the
autochanger/library robotics SCSI connection.
# ls -l /usr/conf/master.d/spt
-r--r--r-- 1 bin bin 4270 Nov 6 1997 /usr/conf/master.d/spt
B. Check if your kernel definition already has the spt driver installed:
# ioscan -f -k -C ext_bus
Class I H/W Path Driver S/W State H/W Type Description
===============================================================
ext_bus 0 56/40 scsi3 CLAIMED INTERFACE HP 28696A - Wide SCSI ID=7
ext_bus 1 56/48 scsi3 CLAIMED INTERFACE HP 28696A - Wide SCSI ID=7
ext_bus 2 56/52 scsi1 CLAIMED INTERFACE HP 28655A - SE SCSI ID=7
ext_bus 3 56/53 lpr2 CLAIMED INTERFACE HP 28655A - Parallel Interface
Note: the "I" column (instance number) of the 56/52 interface. For our system the
instance number of this interface is a 2. We will need this for the device file later.
B. Select one of your SCSI busses, scan the bus for available addresses:
The H/W Path for this example is: 56/52 (from a E-class system)
# ioscan -f -k -H 56/52
C. On the 56/52 bus, addresses 0, 2, 5, and 6 are used; we will use 4 for the
autochanger. The hardware address including the SCSI target of the autoloader
will be: 56/52.4. Note; if the device is a DDS Autoloader, this will be the
SCSI target for the tape drive as well as the autoloader since SCSI lu’s are used
for the internal devices within the unit.
Overview: If your kernel does have the driver in already, skip to section 4.
If you have determined from step 1 that your kernel does not contain the
scsi pass-thru driver, spt, proceed as follows.
Note: this procedure will modify the kernel and system files.
# cp -i /stand/system /stand/system.prev
B. Edit the system file to include (configure) the spt driver into the kernel.
# kmsystem -c y spt
# mk_kernel
D. Schedule the kernel to be moved into place during the next shutdown cycle.
# kmupdate
4. Shutdown the system.
A. Shutdown the system to allow connection of the new device, and execution of
the new kernel.
# cd /
# shutdown -h (power off the system when instructed to do so)
4 7
Scsi Address
Autoloader Options
A. Connect the new device to the system, then power all peripherals on and power up
the system last.
B. When the system reboots, scan for the new device in the ioconfig:
# ioscan -f
note the I (instance number) for the ext_bus (SCSI card) for the device, in this case “2”
note the Target id for the unit, in this case 4, and the lu of the robotic 1, from the unknown device
at 56/52.4.1, which means SCSI instance 2, target 4, device lu 1
# lsdev -d spt
Character Block Driver Class
172 -1 spt spt
Overview: The tape drive in the autoloader will autoconfigure at reboot time. We will
only need to build a device file for the autochanger unit using the scsi
pass-thru device driver. We will use the mknod command for this purpose.
A. Gather the information needed to create the device file for for the robotics:
Device Name: /dev/scsi_library (any name will work)
Device Type: c (for character type devices)
Driver Major#: 172 (output from the “lsdev -d spt” command)
Device Minor#: 0x024100 (02 from the scsi card, target id 4, lu 1)
B. Manually bind the spt driver to the device using the ioscan command:
# ioscan -f -H 56/52.4.1
Overview: HP OmniBack II provides uma (universal media agent), a utility that will get
installed with the media agent module. The utility is used for interacting with scsi
changers. We can test the new device file we built in the previous section to be sure that it
allows us to pass commands to the autochanger. Once the test is verified, configure the
Logical Device.
1. Start the uma utility using the -ioctl option for the new device file:
/dev/scsi_library> inq
SCSI Inquiry:
Type: 8
Vendor ID: "HP "
Product ID: "C1557A "
F/W Revision: "U709"
/dev/scsi_library> stat
Element Status (T=transport, X=Im/Export, D=Drive, S=Storage):
1 D1 Empty ""
2 S1 Full ""
3 S2 Full ""
4 S3 Full ""
5 S4 Full ""
6 S5 Full ""
7 S6 Full ""
/dev/scsi_library> exit
Configuring the DAT Autoloader or DLT Library using the SCHGR driver.
The following procedure can be used to install the connection to a tape library
on a HP-UX 11.x system. Knowledge of system administration is required. This
procedure will make modifications to your kernel. (schgr is in 11.11 by default)
Procedural Outline
1. Make sure you have the required driver(s).
2. Determine the hardware path for the device.
3. Build a new kernel (if necessary).
4. Shutdown the system.
5. Connect the new device.
6. Verify the device file creation.
7. Test the operation of the device.
Overview: You will most likely need to build a new kernel if you have no other
device on your system that requires the SCSI Autochanger driver schgr.
There is also man-page that may help (man 7 autochanger)
Installation for a server systems using HP-HSC, HP-GSC or PCI bus slot for the
autochanger/library robotics SCSI connection.
A. Check if your kernel definition already has the schgr driver installed:
# ioscan -f -k -C ext_bus
Class I H/W Path Driver S/W State H/W Type Description
===============================================================
ext_bus 0 56/40 scsi3 CLAIMED INTERFACE HP 28696A - Wide SCSI ID=7
ext_bus 1 56/48 scsi3 CLAIMED INTERFACE HP 28696A - Wide SCSI ID=7
ext_bus 2 56/52 scsi1 CLAIMED INTERFACE HP 28655A - SE SCSI ID=7
ext_bus 3 56/53 lpr2 CLAIMED INTERFACE HP 28655A - Parallel Interface
Note: the "I" column (instance number) of the 56/52 interface. For our system the instance
number of this interface is a 2. We will need this for the device file later.
B. Select one of your SCSI busses, scan the bus for available addresses:
The H/W Path for this example is: 56/52 (from a E-class system)
# ioscan -f -k -H 56/52
C. On the 56/52 bus, addresses 0, 2, 5, and 6 are used; we will use 4 for the
autochanger. The hardware address including the SCSI target of the autoloader
will be: 56/52.4. Note; if the device is a DDS Autoloader, this will be the
SCSI target for the tape drive as well as the autoloader since SCSI lu’s are used
for the internal devices within the unit.
Overview: If your kernel does have the driver in already, skip to section 4.
If you have determined from step 1 that your kernel does not contain the
scsi pass-thru driver, spt, proceed as follows.
Note: this procedure will modify the kernel and system files.
# cp -i /stand/system /stand/system.prev
B. Edit the system file to include (configure) the spt driver into the kernel.
# kmsystem -c y schgr
# mk_kernel
D. Schedule the kernel to be moved into place during the next shutdown cycle.
# kmupdate
4. Shutdown the system.
A. Shutdown the system to allow connection of the new device, and execution of
the new kernel.
# cd /
# shutdown -h (power off the system when instructed to do so)
4 7
Scsi Address
Autoloader Options
A. Connect the new device to the system, then power all peripherals on and power up
the system last.
B. When the system reboots, scan for the new device in the ioconfig:
# ioscan -f
# lsdev -d schgr
Character Block Driver Class
231 29 schgr autoch
Overview: The tape drive in the autoloader will autoconfigure at reboot time along with
the autochanger robotics if the schgr driver is loaded in the kernel.
# ls -l /dev/rac
total 0
crw------- 1 bin sys 231 0x024100 Aug 10 13:17 c2t4d1
Overview: HP OmniBack II provides uma (universal media agent), a utility that will get
installed with the media agent module. The utility is used for interacting with scsi changers.
We can test the new device file we built in the previous section to be sure that it allows us to
pass commands to the autochanger. Once the test is verified, configure the Logical Device.
1. Start the uma utility using the -ioctl option for the new device file:
/dev/rac/c2t4d1> inq
SCSI Inquiry:
Type: 8
Vendor ID: "HP "
Product ID: "C1557A "
F/W Revision: "U709"
/dev/scsi_library> stat
Element Status (T=transport, X=Im/Export, D=Drive, S=Storage):
1 D1 Empty ""
2 S1 Full ""
3 S2 Full ""
4 S3 Full ""
5 S4 Full ""
6 S5 Full ""
7 S6 Full ""
/dev/rac/c2t4d1> exit
Procedural Outline
1. Make sure you have the required driver(s).
2. Determine the hardware path for the device.
3. Build a new kernel (if necessary).
4. Shutdown the system.
5. Connect the new device.
6. Verify the device file creation.
7. Test the operation of the device.
1. Make sure that you have the required drivers: schgr and ssrfc
Overview: You will most likely need to build a new kernel if you have no other
device on your system that requires the SCSI Autochanger Hardware driver schgr
and the MO Autochanger surface driver ssrfc. Neither of the previous drivers are in the
HP-UX 11.0 kernel by default.
Installation for a server systems using HP-HSC, HP-GSC or PCI bus slot for the
autochanger/library robotics SCSI connection.
A. Check if your kernel definition already has the schgr driver installed:
OR
# kmsystem -q ssrfc (use kmsystem to list existing kernel drivers)
# ioscan -f -k -C ext_bus
Class I H/W Path Driver S/W State H/W Type Description
===================================================================
ext_bus 0 8/0 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
ext_bus 1 8/4 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
ext_bus 2 8/8 c720 CLAIMED INTERFACE GSC add-on Fast/Wide SCSI Interface
ext_bus 4 8/16/0 CentIf CLAIMED INTERFACE Built-in Parallel Interface
ext_bus 3 8/16/5 c720 CLAIMED INTERFACE Built-in SCSI
Note: the "I" column (instance number) of the 8/16/5 interface. For our system the
instance number of this interface is a 3. We will need this for the device file later.
B. Select one of your SCSI busses, scan the bus for available addresses:
The H/W Path for this example is: 8/16/5 (from a D-class server)
# ioscan -f -k -H 8/16/5
Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
ext_bus 3 8/16/5 c720 CLAIMED INTERFACE Built-in SCSI
target 5 8/16/5.0 tgt CLAIMED DEVICE
tape 0 8/16/5.0.0 stape CLAIMED DEVICE HP C1533A
target 7 8/16/5.2 tgt CLAIMED DEVICE
disk 2 8/16/5.2.0 sdisk CLAIMED DEVICE TOSHIBA CD-ROM XM-5701TA
target 9 8/16/5.7 tgt CLAIMED DEVICE
ctl 3 8/16/5.7.0 sctl CLAIMED DEVICE Initiator
C. On the 8/16/5 bus, addresses 0 and 2 are used; we will use 3 for the
autochanger. The hardware address including the SCSI target of the autoloader
will be: 8/16/5.3.0.
Overview: If your kernel does have the driver in already, skip to section 4.
If you have determined from step 1 that your kernel does not contain the
scsi pass-thru driver, spt, proceed as follows.
Note: this procedure will modify the kernel and system files.
# cp -i /stand/system /stand/system.prev
B. Edit the system file to include (configure) the spt driver into the kernel.
# kmsystem -c y schgr
# kmsystem -c y ssrfc
# mk_kernel
D. Schedule the kernel to be moved into place during the next shutdown cycle.
# kmupdate
4. Shutdown the system.
A. Shutdown the system to allow connection of the new device, and execution of
the new kernel.
# cd /
# shutdown -h (power off the system when instructed to do so)
A. Connect the new device to the system, then power all peripherals on and power up
the system last.
B. When the system reboots, scan for the new device in the ioconfig:
# ioscan -f
Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
ext_bus 3 8/16/5 c720 CLAIMED INTERFACE Built-in SCSI
target 5 8/16/5.0 tgt CLAIMED DEVICE
tape 0 8/16/5.0.0 stape CLAIMED DEVICE HP C1533A
target 6 8/16/5.1 tgt CLAIMED DEVICE
disk 3 8/16/5.1.0 sdisk CLAIMED DEVICE HP C1113J
target 7 8/16/5.2 tgt CLAIMED DEVICE
disk 2 8/16/5.2.0 sdisk CLAIMED DEVICE TOSHIBA CD-ROM XM-5701TA
target 8 8/16/5.3 tgt CLAIMED DEVICE
autoch 0 8/16/5.3.0 schgr CLAIMED DEVICE HP C1100J
target 9 8/16/5.7 tgt CLAIMED DEVICE
ctl 3 8/16/5.7.0 sctl CLAIMED DEVICE Initiator
# lsdev -d schgr
Character Block Driver Class
231 29 schgr autoch
Overview: The tape drive in the autoloader will autoconfigure at reboot time along
with the autochanger robotics if the schgr driver is loaded in the kernel.
# ls -l /dev/rac
crw------- 1 bin sys 231 0x033000 Jan 26 14:56 c3t3d0
crw-r----- 1 bin sys 231 0x033013 Jan 26 14:56 c3t3d0_10a
crw-r----- 1 bin sys 231 0x033014 Jan 26 14:56 c3t3d0_10b
crw-r----- 1 bin sys 231 0x033015 Jan 26 14:56 c3t3d0_11a
crw-r----- 1 bin sys 231 0x033016 Jan 26 14:56 c3t3d0_11b
crw-r----- 1 bin sys 231 0x033017 Jan 26 14:56 c3t3d0_12a
crw-r----- 1 bin sys 231 0x033018 Jan 26 14:56 c3t3d0_12b
crw-r----- 1 bin sys 231 0x033019 Jan 26 14:56 c3t3d0_13a
crw-r----- 1 bin sys 231 0x03301a Jan 26 14:56 c3t3d0_13b
crw-r----- 1 bin sys 231 0x03301b Jan 26 14:56 c3t3d0_14a
crw-r----- 1 bin sys 231 0x03301c Jan 26 14:56 c3t3d0_14b
crw-r----- 1 bin sys 231 0x03301d Jan 26 14:56 c3t3d0_15a
crw-r----- 1 bin sys 231 0x03301e Jan 26 14:56 c3t3d0_15b
crw-r----- 1 bin sys 231 0x03301f Jan 26 14:56 c3t3d0_16a
crw-r----- 1 bin sys 231 0x033020 Jan 26 14:56 c3t3d0_16b
crw-r----- 1 bin sys 231 0x033021 Jan 26 14:56 c3t3d0_17a
crw-r----- 1 bin sys 231 0x033022 Jan 26 14:56 c3t3d0_17b
crw-r----- 1 bin sys 231 0x033023 Jan 26 14:56 c3t3d0_18a
crw-r----- 1 bin sys 231 0x033024 Jan 26 14:56 c3t3d0_18b
crw-r----- 1 bin sys 231 0x033025 Jan 26 14:56 c3t3d0_19a
crw-r----- 1 bin sys 231 0x033026 Jan 26 14:56 c3t3d0_19b
crw-r----- 1 bin sys 231 0x033001 Jan 26 14:56 c3t3d0_1a
crw-r----- 1 bin sys 231 0x033002 Jan 26 16:48 c3t3d0_1b
crw-r----- 1 bin sys 231 0x033027 Jan 26 14:56 c3t3d0_20a
...
Overview: HP OmniBack II provides uma (universal media agent), a utility that will
get installed with the media agent module. The utility is used for
interacting with scsi changers. We can test the new device file we built in
the previous section to be sure that it allows us to pass commands to the
autochanger. Once the test is verified, configure the Logical Device.
1. Start the uma utility using the -ioctl option for the new device file:
/dev/rac/c3t3d0> inq
SCSI Inquiry:
Type: 8
Vendor ID: "HP "
Product ID: "C1100J "
F/W Revision: "4.06"
/dev/scsi_library> stat
Element Status (T=transport, X=Im/Export, D=Drive, S=Storage):
0 T1 Empty "" ""
/dev/rac/c3t3d0> exit
• Cell Manager: Maintain the configuration, the database, and control all sessions
• Cell Console: Provides a graphical and command-line interface
• Disk Agent: Read/write data to/from the disk, coordinates data transfer with
media agent.
• Media Agent: Read/write data to/from the tape, coordinates data transfer with
disk agent
The Data Protector cell is a logical collection of systems that have their backups
centrally managed from a single cell manager. There must be only one cell
manager, and any number of disk and media agents. The cell console is a
distributed client interface that may be a member of multiple cells.
4. What if any, is the limit to how many systems may be in an Data Protector cell?
The (supported) limit to the Data Protector cell is 1000 systems, however HP
recommends keeping the number of systems to 100 or less. Ultimately the limit has
to do with the quantity of data that would be stored in the Data Protector embedded
database.
5555
On HP-UX, the inet server is started by the inetd; on Windows, the omni-inet is
running as a service (daemon). The omni-inet process starts the agents.
8. What are the main directories for the Data Protector programs:
The Media Pool is a logical collection of media within the Data Protector Database. The
media pool provides a mechanism to organize tapes that have a common set of
attributes or sets of data. The media pool may be used to represent data collections or
physical locations of tapes.
2. What protection features does Data Protector provide to safeguard the integrity of your
backups?
• Private (ownership) of the data in the database, owned by the owner of the backup.
• CRC (Cyclic Redundancy Check) written to the backup by the logical device
3. Data Protector provides two media allocation policies, strict and loose. Briefly,
describe the purpose of these policies.
• Strict: forces even usage of media according to the allocation sequence number
assigned to the tapes within the media pool. The strict policy works best with tape
libraries, which have a relatively large number of tapes available for backup.
• Loose: allows for automatic labeling of unused tapes, as well as any unprotected
tape to be used for backup. Tapes are requested in the allocation order, but order is
not enforced.
Magazine support is intended for small tape autochanges. The media pool that uses
magazine support will manage the tapes within the magazine as a single unit.
Managing media in magazines is more convenient if you routinely change the
entire “set” of tapes as a set. Data Protector will expect the tapes within the
magazine to be always in the same positions (slots) within the magazine.
5. What Media Condition Factors does Data Protector implement?
6. How can you change the label on an existing tape? What about the location?
Use the Device and Media GUI, open a media pool, select the tape and then
properties; apply changes to label as well as location. The omnimm command may
also be used with the –modify_medium option.
Does the tape need to be loaded into a tape drive in order to change the label or
location?
No!
7. What does recycle do to media, and does the media have to be loaded?
Removes all of the protection for all sessions on an individual tape. The medium does
not need to be in the drive for this operation
8. What happens during a media import? Explain why this would be necessary.
The media import process reads the header and catalog sections from the tape to
restore data into the Data Protector catalog database. This would be necessary if the
catalog retention time expired, and then you wanted to browse the (file level) contents
of a particular tape, and then restore the data.
9. To remove bad media from a pool, you export it. TRUE or FALSE?
False, you can create another media pool and simply move the media that is marked as
poor to the new pool. It may be possible to change the status of a tape by using the
media verification process for Data Protector. If a tape was marked as poor because of
a device failure, or dirty tape drive, then the media verify may restore the tape to a
good status.
10. What must be done before Data Protector can use media?
The tape must have an Data Protector header written to the tape; this is called
initialization, or formatting. The entire tape is not formatted, just the header.
11. When is force required?
Force initialization is needed if the tape was previously used for another purpose. If
Data Protector detects data written in one of the formats that it recognizes it will fail to
initialize the tape without the use of force. Force will not overwrite a tape that is still
under protection within the current cell.
12. When initializing media, the size in megabytes set by specify or determine specifies
the hard limit as to how much data it will hold. TRUE or FALSE?
False. Data Protector uses this only for media reporting. The only exception to this is
the use of file media, which are limited by the size specification.
13. By default, Data Protector will automatically initialize blank media. TRUE or FALSE?
Explain.
True; using the location and multiple media pools you have a lot of flexibility in
locating tapes in off-site locations.
1. What Data Protector device type would you use to configure a single DLT drive?
Standalone
2. What Data Protector utility can be used to check communication with a library device
before you configure it as a Logical Device?
3. What HP-UX driver does Data Protector use to control the robotics of library devices?
Windows? SCSI
Concurrency allows multiple disk agents to write to a single media agent in parallel;
this attempts to keep the tape drive streaming for maximum performance.
Scanning reads the tape header to identify the tape within the drive.
Scan reads the tape header; barcode scan reads the barcode cache from tape library.
8. What command can be used to get a listing of all the logical devices in your cell?
omnicellinfo -dev
10. What command can be used to perform a “scan” of the media in a device?
11. The “Drive Index” is related to the SCSI address of the drive. True/False?
False. The drive index is the physical drive number within the library. The library
controller understands the drive sequence, not the host device file name.
12. Is it possible to have more than one logical device that maps onto the same physical
device?
True. This is useful if you want to configure a single device with more than one set of
Data Protector properties, or use a different physical device configuration, such as
without hardware compression.
13. If you were able to, why would you create more than one logical device for a single
physical device?
14. What is the purpose of the Lock Name advanced option? Does Data Protector ever
require it? If so, when?
The lock name is needed if a physical device is configured with more than one logical
device name. Each time a physical device is used again, the logical device should use
the same lock name. Data Protector will lock logical devices using the lock name to
prevent multiple opens of the same physical device. The lock name is never “required,”
but device failures may result if it is needed and not used.
omnib
2. What are the four fundamental components of all Data Protector backups?
• Defaults
• Objects
• Devices
• Options
A text file containing the 4 components listed above. The backup specification is stored
in the <OMNICONFIG>/datalists directory, and may be edited with a text editor.
4. What is an object?
Something that can be accessed by the disk agent for backup. A File system, directory,
or file may be an object.
To control the access from the restore users, and to allow 3rd party backup products to
have access to the appropriate data, such as Oracle.
Load balancing allows Data Protector to dynamically select devices for backup. Objects
are balanced across the number of running devices to keep the drives running at their
requested concurrency. Load balancing allows Data Protector to work around failed
devices by simply allocating the next available device.
Software compression is useful when network bottlenecks are affecting the performance
of backup, or backup is affecting network performance.
Pre/post execution allows for the automation of tasks before and after backup jobs.
11. What are the differences between object level and backup specification level pre and
post execution?
Backup specification pre/post execution allows for a pre/post backup job control. The
object level scripts execute before and after individual objects. The backup specification
pre-exec runs first, then all of the object level executions, then finally the backup
specification pos-exec executes last.
The template allows for a set of defaults to be saved and used to define a backup job.
False, Data Protector allows for session level restore or object level restore based upon
the objects that are previously backed up.
2. Is it possible to restore a single file from a Rawdisk backup? If yes, describe the
limitations.
Yes, only on HP-UX HFS file system. (not commonly used anymore)
3. The level of detail available in the restore file browser depends on what factors?
The detail level designated in the backup specification: log all, log directory, log file or
no log are possible.
Selecting an individual object from any available version from the database, Data
Protector will put together the restore chain of all of the needed media to perform the
requested restore.
Choose the report level Minor, Major, or Critical when the restore session is started.
• Merge: restore missing files and newer files from tape onto the disk
• Overwrite: restore everything selected from the tape onto the system
Parallel restore allows for the reverse of the backup concurrency to be used to restore
objects by making a single pass through a tape. Normal restore may make several
passes through a tape for the same data.
8. What is the difference between the Restore As and Restore Into options?
9. Is it possible to restore a file backed up on one system to a different system? Must the
system be a member of the same Data Protector Cell?
The restore of data is only limited by cell access. By default, any cell manager may
restore data onto any system that contains a disk agent, even disk agents in remote
cells.
10. Give an example of a use for the pre/post-exec facility for restores:
The pre-exec may be used to stop applications processes prior to restore of data, and
then restarted after the data is restored.
11. In order to perform a valid Rawdisk backup of a file system, the file system must first be
unmounted. TRUE/FALSE?
Exclude requires a full path, and is a specific object. Skip allow for wildcards and may
be anywhere within a file system.
2. The Data Protector internal database is comprised of two tablespaces. What are their
names?
• mmdb
• cdb
3. List three types of data that are stored in the Data Protector internal database.
• Media pools
• File names
• Devices
4. Name 5 parts of the Data Protector Database that are external to the tablespaces:
5. What does the term invalid mean when describing records in the database?
Obsolete data ready to be purged, such as data from a tape that was overwritten.
7. What would you need to add to your database to increase its capacity beyond the 2-GB
limit?
8. Which database files are likely to be the biggest in the database, and what information do
they hold?
• fvers.dat (3.5version only) all of the file versions now kept in the DCBF.
9. By default, an automatic database purge takes place every day at midday. TRUE/FALSE?
• DailyMaintenanceTime
• KeepObsoleteSessions
• KeepMessages
12. Which option can be set in a backup to reduce the amount of detail information stored in
the database?
13. What command would you use to perform a thorough consistency check of the database?
• omnidbcheck -extended
• omnidbutil –writedb …
• omnidbinit
• omnidbutil -readdb
15. What command can be used to display information about database size and utilization?
• omnidbutil –info
• omnidbutil -extendinfo
False! Backing up the database as a file system will not be able to be restored as the
database because it is open at the time of the backup.
<OMNIVAR>/db40/logfiles/syslog
After one full backup cycle, to minimize the amount of archiving necessary when
updating the filenames catalog.
22. What happens to the archive logs when the database is backed up?
• Database
• Internal Events
• Device Events
• Backup Events
• Log to file
• External program execution
This would be useful as notification that a session has finished. Upon completion of a
backup job, several reports may be executed.
TRUE
<OMNICONFIG>Notifications
6. More than one notification can be configured for the same event type. TRUE/FALSE?
TRUE, you can have several different notifications configured for each event type.
Controlling access to functional areas within the Data Protector GUI, and delegating
capabilities to non-administrator users.
3. Which two files store the Data Protector group and user assignments?
• <OMNICONFIG>users/Userlist
• <OMNICONFIG>users/ClassSpec
4. UNIX and NT users can be added to the same Data Protector group. TRUE or FALSE?
TRUE
5. Any user in the Admin group may run backup and restore?
TRUE
FALSE, this file on HP-UX systems may be used to control access to applications that
use the inetd to start their services
To control (allow or prevent) remote cell managers from accessing the disk agent for
restore.
• Central Administration
• Central Licensing
• Central MMDB
• Shared Devices between cells
TRUE, the license is required for each manager that will participate in the MOM
configuration.
To obtain a consolidated password from all of the cell managers to put in a single
license file on the cell manager.
4. What configuration information can be distributed from the MoM GUI when using the
Distribute Configuration function?
• User configuration
• Holidays
• Global options
• Vaulting locations
5. When using a centralized media management database, the MoM system should be highly
available. Why?
The remote cell managers must be able to access the Central MMDB to be able to
perform backup.
opmnidbutil -mergemmdb
8. When using MoM, the catalog databases can also be merged. TRUE/FALSE?
False, MoM only allows the media database to be merged. Each cell manager
maintains its own catalog database.