Restarting IMS

You can restart IMS after you have completed any necessary recovery. You can perform a normal, emergency, or emergency with OVERRIDE restart.

• Normal restart initializes the IMS subsystem without reference to any previous execution of IMS (a cold start), or restarts a system that was terminated using a checkpoint command (a warm start).
• Emergency restart initializes an IMS subsystem after a system failure. During an emergency restart, IMS does the following:
  • Closes the OLDS from the WADS
  • Resets each active transaction to its last sync point
  • Resets each active region (BMPs, and CCTL or ODBA threads) to its last sync point

    You do not need to restart connected ODBA application programs or CCTLs. An IMS shutdown or failure simply disconnects from the ODBA application or CCTL and generally does not affect it.

  • Restores the databases to the last sync point
  • Restores local message queues to their condition at the time of failure

    An IMS failure does not generally affect shared message queues.

  • If dynamic resource definition is enabled in the IMS system, the runtime resource definitions are restored by processing the checkpoint and X'22' logs

  You must manually restart regions and BMPs. However, when restarting a batch message processing region, IMS automatically determines the last BMP sync point if you specified LAST as an EXEC parameter. Otherwise, IMS indicates what the last BMP sync point was so that you can specify it during restart.

  If you have an IMS DBCTL standby environment, you can send the /ERESTART command to the pre-initialized DBCTL subsystem. Doing this is faster than starting the IMS job and waiting for the DBCTL-ready message before sending the /ERESTART command.

• Emergency restart with OVERRIDE is necessary after a failure in which DBRC is unable to mark the SSYS record in the RECON data set as abnormally terminated, such as a power, CPC, z/OS®, or DBRC failure.

Automatic restart reduces MTO intervention and makes restart faster. With automatic restart, IMS automatically chooses the appropriate restart command (either /NRESTART or /ERESTART). If restart processing abnormally terminates before initial checkpoint, the appropriate automatic restart command is the same type (either /NRE or /ERE) as the aborted restart. The operator does not enter a restart command. Specify automatic restart by including AUTO=Y in the JCL.

When AUTO=Y is specified in the startup procedure, you cannot cold start IMS or modify any restart options before IMS restarts.

You can implement normal restart procedures easily, and generally do not need to worry about it. However, an operator error could adversely affect system integrity in the following situations:

• Specifying cold start after a /CHECKPOINT FREEZE | DUMPQ | PURGE command when some messages are not completely processed
• Failing to supply the proper security options