Chapter 6. Operations 111
6.2.2 z/OS system failure
If the underlying z/OS system fails, the DB2 subsystem in the z/OS image should
be restarted on another z/OS image in the Parallel Sysplex as soon as possible
so that retained locks are released. You can automate this cross-system restart if
your z/OS systems team has implemented automatic restart manager (ARM)
policy.
In non-data sharing, some users wait for an IPL of the failed z/OS image. For
data sharing users, this is too long to wait. The focus needs to be changed from
the failed image (and its DB2) to the DB2 members that are active on another
image. The active workload can incur timeouts and failures due to the retained
locks still held by the member that remains down.
Once the DB2 ...