Failback occurs automatically on HACMP. When you start HACMP on the primary node, the stop_server event on the secondary node triggers the monitoring script to execute sp_companion 'prepare_failback'.
To failback to the primary node, make sure that the secondary companion is in secondary failover mode, and start HACMP services on the primary node. To make sure that sp_companion 'prepare_failback' was executed successfully, search for this string in /tmp/hacmp.out:
SYBASE HA MONITOR: Prepare_failback was successful.
Before you start the HACMP services on the primary node,
make sure the secondary node is running and the secondary companion
is running in secondary failover mode. If the secondary companion
or secondary node is not up and running, do not start the primary
companion. If both nodes are down, or the HACMP services has stopped
on both nodes, always restart the secondary node and its HACMP services
before restarting the primary node.
