HP Application Recovery Manager software A.06.10 - Product announcements, software notes, and references (March 2008)

Cluster related issues
Common issues
• If a backup session stops responding during a cluster failover, and all session
agents fail, a timeout will be reported but the session itself will not abort. The
default session timeout occurs after 7200 seconds (two hours). As long as the
session is not responding, another session using the same backup specification
cannot be started.
Workaround: Manually abort the backup session at any time and restart the
session.
• If a cluster failover occurs during an Application Recovery Manager backup
session in which an application database that resides on the cluster is being
backed up with the appropriate integration agent, particular problem may occur
after the failover which prevents the session from succeeding.
Under such circumstances, in Monitoring context of the Application Recovery
Manager GUI, two backup sessions are displayed: the backup session that was
restarted after the failover, and another, unknown session. Output of the unknown
session contains messages similar to the following:
[Critical] From: BSM@ClusterNode01Name
"BackupSpecificationName" Time: Date Time
[12:1243] Device not found.
[Critical] From: OB2BAR_VSSBAR@ClusterNode02Name "MSVSSW"
Time: Date Time
Failed VSSBAR agent.
[Major] From: OB2BAR_VSSBAR@ClusterNode02Name "MSVSSW"
Time: Date Time
Aborting connection to BSM. Abort code -1.
[Critical] From: BSM@ClusterNode01Name
"BackupSpecificationName" Time: Date Time
None of the Disk Agents completed successfully.
Session has failed.
The root cause of the problem is unsuccessful identification of the restarted backup
session after a cluster failover. The involved integration agent is not notified about
the backup session restart. Depending on the particular situation, the integration
agent either starts a new backup session or connects to the restarted backup
Recognized issues and workarounds28