Configuration for Veritas Global Clusters and Non-Veritas Clusters on UNIX

After the Commvault software is installed on all the physical nodes that host the cluster group, you must configure a cluster group client in the Command Center to protect cluster resources on the physical nodes.

Configure the cvclusternotify Script

A failure of any Agent software on the active node in a UNIX cluster will not cause a failover to be initiated.

The cvclusternotify script should be added as part of the normal cluster startup/shutdown procedure. The script is provided as a generic template, and it must be run at the beginning of node shutdown and at the end of new active node startup before any I/O or application starts on the cluster volumes. In both cases, data protection services must be up and running.

Run the following command to notify Commvault that the specified "Cluster Group" is going up or down because of a cluster failover:

Usage:

 
cvclusternotify -inst InstanceName -cn ClientName  -start | -shutdown

Where:

• cvclusternotify is the script that notifies the Commvault software about cluster failovers.

• -inst specifies the name of the Commvault instance on which you want to run the script. If you have a single instance, specify Instance001.

• -cn is the name of the cluster group client.

Example:

For a two-node cluster, if the cluster group client name is "ClusterGroup1" and the application instance is "Instance001", run the following command:

• To shutdown:

  cvclusternotify -inst Instance001 -cn "ClusterGroup1" -shutdown

• To start up:

  cvclusternotify -inst Instance001 -cn "ClusterGroup1" -start

Add Service Dependencies

Add the following service dependencies:

• Dependencies to IP Resource

• Dependencies to Disk Resource

Example:

When the node is going down:

• Run the command to shut down:

  cvclusternotify -inst Instance00X -cn cluster client name -shutdown

• Unmount any moved file system because of the failover

• Un-plumb virtual IP

When the node is starting up:

• Plumb virtual IP

• Mount any moved file system because of the failover

• Run the command to start up:

  cvclusternotify -inst Instance00X -cn cluster client name -start

Result

• A new application called GxClusterPlugin_service_group_name is created in the Veritas cluster, and dependency links will be set up from this newly created application to all of the resources in the group of type IP, mount point, and disk group. This ensures that the plug-in is started after all of the other resources have come online or is stopped before any other resource goes offline during failovers.

  The plug-in must be able to read/write the main.cmd file, which is typically located under /etc/VRTSvcs/conf/config. If the file is not present, run the following command:

   
  hacf -cftocmd directory_path

  After the command is run, the main.cmd file is created with the same ownership as that of Commvault processes.

• During the service group failover, the plug-in performs the following tasks:

  1. Turns off firewall on the service group, and enables it on the active node.

  2. Notifies the CommServe database where the service group is active for effective resource allocation.

  3. If archiving operations are configured, the plug-in switches the monitoring mount points to the active node to handle stub recalls.