PRIMECLUSTER Installation and Administration Guide 4.2 (Linux)

3.1.3 Setting Up Hardware Monitoring with ServerView

Overview

ServerView supports monitoring of the fan, temperature sensor, and power supply in addition to watchdog monitoring of the OS. In the ServerView, you can set up the behavior of each monitored target in the event of a failure. For example, if "Shut down the server immediately" is selected and an error is detected, the failed node will be shut down. The ongoing operations on the failed node are then quickly moved to the standby node.

Behavior setup of recovery function

When setting the behavior of the monitored targets in the event of a failure, you need to consider the operating environment in which a failover is generated. The recommended behaviors are as follows:

Monitored item			Recommended behavior
Fan
	Fan check time		00:00 AM/PM *
	Action after Fan Fail		Continues ongoing operations.
Temperature sensor
	When abnormally raised temperature is detected.		Shuts down the node immediately.
Restart settings
	Automatic Power On Delay (minutes)		2 **
	Number of Reboot Tries		3 **
	Action after exceeding reboot tries		Abandons attempts to restart and power off the node.
	Watchdog Settings		Not supported ***
	Monitoring time	Watchdog Timeout Delay (minutes)	3 **
		Action	Continues ongoing operations. **
	Startup monitoring		Not supported
	Monitoring time	Elapse time (minutes)	6 **
		Action	Restart. **

* The default value is 00:00 midnight. You need to change the value according to the operating environment.

** The default value is recommended. Note that the default value might vary depending on the machine model.

*** If communication with cluster interconnect is performed normally even when the operating system hangs up, no failover is generated with PRIMECLUSTER. This state can be avoided by enabling watchdog timer monitoring.

Since the hardware products that are not listed above only support monitoring, the behavior setup for the recovery function is not available.

For information about behavior setup, see the "ServerView User Guide."