If
the automatic failback fails, examine the logs to make sure that
the high availability system performed the following steps. If it
did not, perform them manually. You must perform them in the sequence
described below.
Stop the HACMP subsystem with the takeover mode on the primary node (See your IBM documentation for information). This shuts down the primary companion and fails over its resources to the secondary companion.
Shut down and restart the secondary companion. RUNHA_<servername>.sh restarts the companion automatically after you shut it down if RETRY is set to a value greater than 0.
Log in as HA_LOGIN, as specified in RUNHA_<servername>.sh, to the secondary companion through isql and verify that it is running in secondary failover mode.
Issue sp_companion 'prepare_failback'. For example, to fail back from the secondary companion PERSONNEL1:
sp_companion MONEY1, 'prepare_failback'
Restart HACMP on the primary node.
Log in to the primary companion using isql and make sure it is running in primary failback mode.
Issue sp_companion 'resume'. For example, to resume companion mode for primary companion MONEY1:
sp_companion PERSONNEL1, 'resume'
You cannot connect clients with the failover property
until you issue sp_companion resume.
If you do try to reconnect them after issuing sp_companion
prepare_failback, the client stops responding
until you issue sp_companion resume.
