IMS log dataset space problem

When IMS log dataset space problem, remember that IMS batch and online share databases through DBRC, which serializes access — contention or a missed authorization stops work fast.

What it means

• The message queue is the shock absorber: when it grows, the bottleneck is downstream, not in the queue itself.
• Checkpoint restart replays from the last checkpoint — anything not checkpointed gets reprocessed, so restart logic must be idempotent.
• A PSB change means nothing until ACBGEN runs and the region picks up the new ACB.

Common causes

• PSB or PCB in the program does not match the ACB in the IMS system (PSBGEN/ACBGEN out of sync).
• A batch job holds the database through DBRC while an online region needs it.
• A dependent region (MPP/BMP) fails to start because its PSB is not defined to the region.
• IMS log data sets (OLDS) fill up, stalling the whole IMS system.
• A slow or stopped transaction lets the message queue backlog grow.

How to diagnose

• Use /DIS DB to see database status and which job or region holds it.
• Check the IMS master terminal output and the failing region JES log for the first error.
• Verify the PSB was generated into the ACB library that the failing region actually uses.
• For queue backlog, watch the queue depth trend with /DIS Q — steady growth means the consumer is the problem.

How to fix it

• Run ACBGEN after any PSB or DBD change, then restart the region so it picks up the new ACB.
• Coordinate batch windows with DBRC so online regions are not starved of the database.
• Switch or add OLDS data sets when the log fills; never let IMS stall on logging.
• Restart the transaction (/STA TRAN) only after fixing the underlying cause.

Example: IMS operator checks

Always /DBR a database before batch update work, then /STA it when done.

Interview tip

• Q: An IMS transaction works in test but fails in production with a PCB status error, like IMS log dataset space problem. Where do you look?
• A: At the ACB library the production region uses — the PSB or DBD was likely changed and ACBGEN ran for test but not production. I compare the ACB members, regenerate for production, and restart the region.

Prevention checklist

• Make ACBGEN part of every IMS change deployment, for every region.
• Monitor OLDS usage and queue depths with alerts before they fill.
• Keep a batch-versus-online database schedule so DBRC contention is planned, not discovered.