When the primary companion or machine is prepared to resume operation, the user with the ha_role performs fail back to return the servers to normal companion mode. Because fail back temporarily shuts down the databases of the failed-over companion, you should choose a time for fail back when the application load is light. If you choose a time when the Adaptive Server is very busy, failback succeeds, but is very slow, and the performance of the secondary companion is sacrificed. Choosing the appropriate time for fail back can dramatically reduce the amount of time clients must wait to reconnect.
