DOCUMENTATION: Checklist and general information for upgrading a clustered environment from NetBackup 5.x to NetBackup 6.x

Article:TECH49483  |  Created: 2008-01-17  |  Updated: 2011-05-10  |  Article URL http://www.symantec.com/docs/TECH49483
Article Type
Technical Solution

Product(s)

Environment

Issue



DOCUMENTATION: Checklist and general information for upgrading a clustered environment from NetBackup 5.x to NetBackup 6.x


Solution



Manual: Veritas NetBackup (tm) 6.0 Installation Guide for UNIX

Modification Type: Supplement

Modification:
The following is a collection of information to be aware of when attempting to upgrade a cluster aware environment from NetBackup 5.x to the NetBackup 6.x release.  This TechNote includes a checklist that should be performed before the upgrade to ensure the cluster is configured properly prior to the upgrade.  It also includes a means to verify the new Enterprise Media Manager (EMM) database is created properly during an upgrade.  The last section includes notes from the NetBackup 6.0 Maintenance Pack (MP) README.

Checklist for upgrading a cluster aware environment running NetBackup 5.x to 6.x
 
It will be necessary to review the expected NetBackup 5.x Master Server configuration when working in a cluster environment with Veritas Cluster Server (tm).
 

 
The following steps should be taken prior to running the NetBackup 6.x upgrade installation. Please note:  The following steps only cover some of the possible configuration issues that could exist in a clustered environment.  It is still possible that other configurations could exist that would impact the upgrade.  If any errors occur during or after the upgrade please contact Symantec Enterprise Technical Support for assistance.
 

 
When NetBackup 5.x is running in cluster the binaries are installed locally and the catalog is installed in the shared folder.  The online script creates links to the shared folder before starting NetBackup.  The offline script will then remove those links after stopping NetBackup. This represents a change in behavior from NetBackup 4.5, where both binaries and catalog were in the shared folder.  For more information on upgrading a cluster from NetBackup 4.5 to NetBackup 5.x refer to TECH31396, found below, in the Related Documents section.
 

 
1. The first thing to check is that a response file already exists.  The default location for a response file would be:
 
# cd /usr/openv/netbackup/bin/cluster/vcs
 
# cat VCS_NBU_RSP
 

 
2. Look at the contents of this file to determine the following:
 
- Determine if the entry for SHARED_DISK exists and is set to any location other than /usr/openv on the local system.  (If it is not set to /usr/openv, confirm the validity of any LINK entries by examining their paths relative to the location specified by SHARED_DISK.)
 
- Determine if the entry for VNAME matches the CLUSTER_NAME defined in the /usr/openv/netbackup/bp.conf file.  The CLUSTER_NAME entry is mandatory for NetBackup 5.1 systems.
 
Note: Make sure that the file is identical in all nodes to which NetBackup service group should failover. If they are not identical, choose the most appropriate one and copy it to other nodes.  Symantec does not recommend hand editing this file. Ensure sure there are no customized comment-lines (i.e. lines with a leading #, other than the very first comment line generated by cluster_config script).  The cluster_upgrade script will not recognize additional comment lines and attempt to process them during the upgrade.  If these entries exist it will lead to unpredictable results during the installation.
 

 
3. Verify the contents of the /usr/openv/netbackup/bp.conf file:
 
- The CLUSTER_NAME entry should match the first SERVER entry in the bp.conf file. This entry should reflect the virtual name and be different from the hostname of the local system.
 
- There should be SERVER entries for each host name in the cluster.
 

 
4. Verify the contents of the /usr/openv/volmgr/vm.conf file:
 
- If the REQUIRED_INTERFACE option exists then this entry should be removed.  For more information, refer to TECH48026, found below, in the Related Documents section.
 

 
5. Verify that the NetBackup agent already installed:
 
# hatype -list | egrep NetBackup
 

 
6. Verify the /opt/VRTSvcs/bin/NetBackup directory is a link to /usr/openv/netbackup/bin/cluster/vcs
 

 
If any of these requirements is not met then correct the configuration before attempting to upgrade to NetBackup 6.x.  These values are used by the NetBackup installer to determine the settings for the cluster.  If any of these options are incorrect or missing it can cause errors during the upgrade.
 

Verification that the EMM database is dropped and recreated during upgrade process in cluster environments
 
When upgrading to NetBackup 6.x in a cluster, the active node should be the last node to be upgraded.  The install_trace log file can be checked to verify the Enterprise Media Manager (EMM) database is created in the local folder on the active node.  By default this will be created in the /usr/openv/db/data directory.  
 

 
The following example is from the /usr/openv/tmp/install_trace.<pid>
 
Database server is VERITAS_NB_nbuvcs
 
Creating the NetBackup database
 
Creating NetBackup data files in: /usr/openv/db/data
 
Creating /usr/openv/db/data
 
Creating NetBackup index files in: /usr/openv/db/data
 
Creating NetBackup transaction log file in: /usr/openv/db/data
 
Database "/usr/openv/db/data/NBDB.db" created successfully
 
Starting the NetBackup database
 

 
When upgrading a NetBackup master server in a cluster, the EMM database will be dropped and recreated on the shared folder.  It is important to verify that this was done properly by reviewing the install_trace file for the following entries.  These show that the database is being dropped and then be created in the shared folder.  By default the shared folder is created in the /opt/VRTSnbu/db/data directory.  If the shared folder is in a different location verify that directory name is reflected in the install_trace logs.
 

 
1. Check the /usr/openv/tmp/install_trace.<pid> for the following entries.
 
Database server is VERITAS_NB_nbuvcs
 
Dropping the NetBackup database: /usr/openv/db/data/NBDB.db
 
Creating the NetBackup database
 
Creating NetBackup data files in: /opt/VRTSnbu/db/data
 
Creating /opt/VRTSnbu/db/data
 
Creating NetBackup index files in: /opt/VRTSnbu/db/data
 
Creating NetBackup transaction log file in: /opt/VRTSnbu/db/data
 
Database "/opt/VRTSnbu/db/data/NBDB.db" created successfully
 
Starting the NetBackup database
 

 
2. After verifying the database was created in the shared folder, it is necessary to verify the bp.conf file.  Check the following file /usr/openv/netbackup/bp.conf and verify the VXDBMS_NB_DATA points to the shared folder. This entry should reflect the correct name for the shared directory.
 
VXDBMS_NB_DATA = /opt/VRTSnbu/db/data
 

 
3. Finally, check the /opt/VRTSnbu/db/data/vxdbms.conf file to make sure it refers to the shared folder.  These entries should reflect the correct name for the shared directory.
 
VXDBMS_NB_DATA = /opt/VRTSnbu/db/data
 
VXDBMS_NB_INDEX = /opt/VRTSnbu/db/data
 
VXDBMS_NB_TLOG = /opt/VRTSnbu/db/data
 

 
4. The NetBackup 6.x upgrade contains a new step, which is to run the nbpushdata command to update the contents of each master and media server into the new EMM database.  After the nbpushdata commands have been run ensure each host is listed correctly.  This can be done by running the nbemmcmd command.  The following example shows a correctly configured hosts list for the EMM database after a successful upgrade.  In the following example nbuvcs is the master server virtual name, and sun02 and sun04 the node names:
 
# nbemmcmd -listhosts -verbose
 
NBEMMCMD, Version:6.0(20050906)
 
The following hosts were found:
 
nbuvcs
 
       MachineName = "nbuvcs"
 
       MachineDescription = ""
 
       MachineNbuType = server (6)
 
nbuvcs
 
       MachineName = "nbuvcs"
 
       MachineDescription = ""
 
       MachineNbuType = cluster (5)
 
       Active Node Name = "sun02"
 
sun02
 
       ClusterName = "nbuvcs"
 
       MachineName = "sun02"
 
       GlobalDriveSeed = "VEND:#.:PROD:#.:IDX"
 
       LocalDriveSeed = ""
 
       MachineDescription = ""
 
       MachineFlags = 7
 
       MachineNbuType = master (3)
 
       MachineState = active for disk jobs (12)
 
       NetBackupVersion = 6.00 (600000)
 
       OperatingSystem = solaris (2)
 
       ScanAbility = 5
 
sun04
 
       ClusterName = "nbuvcs"
 
       MachineName = "sun04"
 
       GlobalDriveSeed = "VEND:#.:PROD:#.:IDX"
 
       LocalDriveSeed = ""
 
       MachineDescription = ""
 
       MachineFlags = 3
 
       MachineNbuType = master (3)
 
       MachineState = active for disk jobs (12)
 
       NetBackupVersion = 6.00 (600000)
 
       OperatingSystem = solaris (2)
 
       ScanAbility = 5
 

Caveats for Installing a NetBackup 6.0 maintenance pack on a cluster
 
When upgrading to a NetBackup 6.0 Maintenance Pack (MP) in a clustered environment it is important to freeze the NetBackup group prior to the install.  This will prevent the failover of the cluster during the patch installation and the associated problems it will cause when updating the Enterprise Media Manager (EMM) database.
 

 
The NetBackup 6.0 maintenance pack readme lists the following about updating in a UNIX clustered environment.
 
For Maintenance Pack installation on a UNIX Cluster Environment:
 
----------------------------------------------------------------
 
1. Before you install this Maintenance Pack, make sure that NetBackup is at
 
  release level 6.0 and configured to run in a cluster.
 

 
2. Freeze the NetBackup group. (By freezing the group, a
 
  'failover' will be prevented from occurring during a patch installation).
 

 
If the steps mentioned herein are not taken before an upgrade, it is possible the EMM database will not get updated.   An error was reported when upgrading a Solaris master server to NetBackup 6.0 Maintenance Pack 4 (MP4).  During the installation the NetBackup group in the cluster was not put in failover mode before the upgrade.  After the upgrade the error "Permission denied: you do not have permission to select from EMM_DriveType" appeared in the nbrb logs.  This error was preventing a successful connection to the nbemm daemon resulting in the daemons not starting and a failure to run any subsequent backup or restore jobs.
 

 
For information on how to take the NetBackup group offline during an upgrade refer to the Veritas NetBackup (tm) 6.0 High Availability Guide.  This contains a "Upgrading a NetBackup Failover Server" chapter for each of the supported Cluster types.  Once the NetBackup group is in a failover or inactive state then the NetBackup maintenance packs can be installed in the cluster.
 

Symantec recommends that the above information be reviewed prior to beginning any NetBackup 6.x upgrade in a clustered environment.  For additional information review the NetBackup (tm) Installation Guides for UNIX, TECH43924 found below, in the Related Documents section.  Also review the NetBackup (tm) High Availability Guides, TECH43985 found below, in the Related Documents section.
 



Legacy ID



285699


Article URL http://www.symantec.com/docs/TECH49483


Terms of use for this information are found in Legal Notices