Job Management Partner 1/Performance Management User's Guide

9.7 Failure recovery in a cluster system

When a failure occurs on the executing node, the cluster software executes a failover and the standby node takes over the processing. When a failover takes place, processing that was being executed by the executing node stops.

The system administrator identifies the cause of the failure that occurred on the executing node. After removing the cause of the failure, you need to switch the processing to the executing node to recover from the failure.

Collect and analyze the following log information to identify the cause of the failure:

Performance Management log information
This information is the same as that collected in a non-cluster system. Collected information includes:

System log

Common message log

Operation status log

Trace log

Cluster software and OS log information
Hitachi recommends that you collect the cluster software log information as well as logs output by the OS itself.

For details on the Performance Management log information, see 14.3 Log information.

Organization of this section

(1) Collecting the log information in a cluster system

(1) Collecting the log information in a cluster system

Pay attention to the following items when you collect the Performance Management log information in a cluster system:

The common message log and trace log are output to the shared disk when Performance Management is used on a logical host.
Log information before and after the failover is recorded in the same log file since the log file on the shared disk is inherited together with the system when a failover takes place.

If Performance Management is used on a logical host, it is necessary to refer to the information from around the time when a failure occurs. For this reason, you need to extract the log information on both the executing node that stopped the processing due to the failover and the failover destination standby node.

For details on how to extract the Performance Management log information, see 14.5 Data collection procedure.

[Contents][Back][Next]

[Trademarks]