This section explains the conditions required for auto-recovery.
Auto-Recovery Using Server Switchover Settings
A server for which Auto-Recovery is enabled will be automatically switched over to its spare server if Resource Coordinator VE detects both a failure from the server hardware and determines that its physical OS (or VM host) has stopped.
Detecting hardware failures from servers
A hardware failure can be detected by an "Error" level SNMP trap failure notification sent to the admin server from either the ServerView Agent or the server management unit. Alternatively, Resource Coordinator VE can detect a failure by periodically polling the status of each managed server.
Detectable hardware failures
CPU faults
Memory errors
Temperature abnormalities
Fan failures
As a result of a FAN failure, it is detected as a temperature abnormality.
Detecting that a physical OS (or VM host) has stopped
A physical OS (or VM host) is seen to have stopped abnormally when the following conditions are met:
PRIMERGY BX series servers
An abnormal server status is obtained from a server management unit, and it is not possible to communicate with either the ServerView Agent or the Resource Coordinator VE agent
For rack mount, tower, and SPARC Enterprise servers
Communication using the ping command is unavailable
Auto-Recovery Using Monitoring Information Settings
When ping monitoring using monitoring information is enabled, server switchover is automatically performed when there is no response from physical OS on servers or VM hosts, and restoration by executing reboot fails.
The recovery process can be changed by configuring settings. For details on how to configure these settings, refer to "6.2.6 Configuring Monitoring Information".
No response detected by ping monitoring
When the period with no response in the ping command is over the time-out value, no response is detected.
Note
Notification of hardware failures on rack mount servers, tower servers, and SPARC Enterprise servers is only detected by SNMP traps.
Auto-Recovery is not triggered on servers that are in maintenance mode.
Even if a hardware failure is detected, Auto-Recovery will not be triggered if no response is received from the target server. In such cases, shutting down or restarting the server will temporarily stop the operating system, triggering an automatic switchover as the conditions for Auto-Recovery will be met. Under such conditions, automatic switchovers can be prevented by setting the server to maintenance mode before shutdown or restart.
