10.6.6 Performing realtime operation monitoring by alarms in a cluster system
Alarms need to be set in order to notify users when an error occurs in a monitoring target system. Note that the method of setting alarms is different from that for a non-cluster system when a logical host is used in a cluster system.
- Notes about the nodes that executes actions:
-
-
If LOCAL is set for Command Execution Action Handler, actions are executed on the node on which PFM - Agent or PFM - RM that performs alarm monitoring operates. For example, when an alarm occurs with PFM - Agent or PFM - RM used on a logical host, actions are executed on the executing node on which PFM - Agent or PFM - RM runs.
-
When Performance Management runs on a logical host, if you specify the logical host name or LOCAL for Command Execution Action Handler, commands are executed on the node on which Performance Management operates. For that reason, you need to configure the environment in such a way that commands can be executed in the same way on both the executing and standby nodes.
-
Also, if the Action Handler service runs on a logical host, the current directory is as described below. The environment directory means the environment directory name specified with the jpcconf ha setup command.
environment-directory\jp1pc\bin\action\
-
When Performance Management is running on a cluster system and a JP1 event is issued as an alarm action, the JP1 event is registered in the event server of JP1/Base on a physical host as standard behavior.
When Performance Management and JP1/Base are operated on the same logical host, use the -r logical-host-name option to additionally specify the logical host name as the event server name in the message text (JP1 event attribute to be passed to the jpcimevt command) field where the JP1 event is registered.
You cannot specify the event server name of JP1/Base that runs on a different logical host.
For details on how to set alarms, see 6.4 Setting alarms using the Web browser (Alarms tree).
-