Stopping a data server

Stopping a data server ends the address space that the data server is running in after stopping all of the services that are running in that data server.

You can stop a data server in two ways:

Controlled shutdown

With a controlled shutdown, the data server stops after all activity in the data server completes normally and the services complete an orderly shutdown. Services stop when:

  • Replication ends for all actively replicating subscriptions. This is equivalent to requesting a controlled end of replication.
  • Replication stops after the target completes applying all changes that it received. The capture service waits indefinitely for the target to report that replication is complete for a subscription.
  • For a persistent subscription, replication automatically restarts when the data server is restarted.

You initiate a controlled shutdown by issuing one of the following commands (the commands are not case sensitive):

z/OS modify command
F name,STOP,ALL
z/OS STOP command
STOP name

Where name is the started task or job name of the data server.

Immediate shutdown

You initiate an immediate data server shutdown by using the following z/OS modify command:

F name,STOP,ALL,IMMEDIATE
Note: You can request an immediate shutdown after a controlled shutdown is requested, but after an immediate shutdown is accepted further stop commands are ignored.

An immediate shutdown is equivalent to requesting an immediate end of replication for any actively replicating subscriptions, including those that might be in the process of performing a controlled end of replication.

Replication automatically ends for a subscription based on the capture service SHUTDOWNTIMEOUT setting. The default is 30 seconds.

Replication does not automatically restart when the data server is restarted for any persistent subscriptions that were replicating at the time an immediate shutdown is performed. You must manually restart replication for these subscriptions.

After replication stops for any active subscriptions, the capture service stops, followed by the other services that run in the data server. If a service does not stop within five minutes, the service is forced to stop to ensure that the data server shuts down as rapidly as possible. A forced stop might result in a system abend.

The total time that is required to shut down the data server might be longer than five minutes because an internal service hierarchy exists among the various services. Services that are in one level must all stop before the next level can stop. The logger service is the last to stop, immediately before the data server stops.

If you tried the controlled stop, elevated to the immediate stop, and waited a sufficient period (five minutes), take a dump, cancel the address space, and open a case with IBM Support. You should also have backups for all of the DDs that are in the address space.