By default, groups are set not to fall back to the original (failed) machine. Unless you manually configure a DB2 group to fall back after failing over, it continues to run on the alternative MSCS node after the cause of the failover has been resolved.
If you configure a DB2 group to automatically fall back to the original machine, all the resources in the DB2 group including the DB2 resource will fall back as soon as the original machine is available. If, during the fall back, a database connection exists, the DB2 resource cannot be brought offline, and the fallback processing will fail.
If you want to force all database connections off the database during the fallback processing, set the DB2_FALLBACK registry variable to ON. This variable must be set as follows:
You do not have to reboot or restart the cluster service after setting this registry variable.
[ DB2 List of Books | Search the DB2 Books ]