If your database management software
or cluster management software detects that a database server has
failed, your database solution must respond to that failure as quickly
and as smoothly as possible.
Your database solution must
attempt to shield user applications from the failure by rerouting
workload, if possible, and failover to a secondary or standby database,
if one is available.
Procedure
If your database or cluster management software detects
that a database server has failed, you or your database or cluster
management software must do the following:
- Identify, bring online, and initialize a secondary database
server to take over operations for the failed database server.
If you are using Db2® High Availability
Disaster Recover (HADR) to manage primary and standby database servers, HADR will manage keeping the
standby database synchronized with the primary database; and HADR will manage the takeover of the
primary database by the standby database.
- Reroute user application workload to the secondary database
server.
Db2 client
reroute can automatically reroute client application away from a failed database server to a
secondary database server previously identified and configured for this purpose.
- Remove the failed database server from the system to repair
it.
Once the user applications have been rerouted to a secondary or standby database server, the
failed database server can not handle any client application requests until it has been restarted or
otherwise repaired. For example, if the cause of the failure on the primary database was that a
database instance terminated unexpectedly, the Db2 fault monitor
facility will automatically restart it.