If the system behavior is unstable, for example there are insufficient temporary system resources, the Mirroring Controller automatic switch may fail.
Perform the switch manually using one of the following methods:
Refer to the procedures in "3.4 Manually Switching the Primary Server".
In the standby server, execute the mc_ctl command in switch mode with the -force option specified to forcibly perform the switch.
Example)
$ mc_ctl switch -M /mcdir/inst1 --force
Point
Even if connection cannot be established between database servers, it is possible to fence the primary server and forcibly switch by executing the mc_ctl command in switch mode with the --force option specified.
The primary server is not fenced in the cases below, so stop Mirroring Controller and instances of the primary server database in advance:
The --no-fencing option is specified when performing forced switch.
The heartbeat_error_action parameter in serverIdentifier.conf is set to "message" and the fencing command is not configured to be used (the fencing_command parameter is omitted in serverIdentifier.conf).
The heartbeat_error_action parameter in serverIdentifier.conf is set to "fallback".
See
Recovery to database multiplexing mode
Refer to "4.1.1.2 Rebuild the Standby Server" and "4.1.1.3 Failback of the Primary Server" for information on recovery to database multiplexing mode.
