NetBackup 8.1.2 media servers in app_cluster configurations fail to update the active node after a failover

Problem

After upgrading NetBackup to 8.1.2 from previous versions or install fresh at NetBackup 8.1.2, the application cluster mechanism will not automatically determine the active node after the clustered application has failed over to a new host/node (media server).

The ability to associate media servers as application cluster (app_cluster) systems is important for High Availability when backing up the shared data.  If NetBackup does not recognise the active node automatically the backups will fail with STATUS 213.

To enable NetBackup to see the application cluster active node, restarting the NetBackup services on the active node is required as a workaround, but under normal circumstances this should not be required.

Error Message

STATUS 213

Cause

The NetBackup Volume Manage Daemon ( vmd ), was designed to update the master server every five minutes to ensure the ' app_cluster ' references were correct.  This was to ensure when a failover occurred of the clustered application the active node for the ' app_cluster ' server reference was associated to the correct host, but this feature was omitted at NetBackup version 8.1.2 (note, the same issues would have been seen at 8.1 and 8.1.1).

Solution

Veritas NetBackup engineering were notified of the issue under Etrack 3966527.  At the time this technical article was released, the Veritas NetBackup engineering team provided an Emergency Engineering Binary (EEB) to resolve the problems for NetBackup 8.1.2, which is available by contacting Veritas Support.  Note, if the NetBackup version is pre-8.1.2 no EEB is available and upgrading to 8.1.2 is required before applying the available fix.