S067 abend with Cleanup for Top Secret

Products

• Top Secret
• Cleanup

Issue/Introduction

On one system, the Cleanup for Top Secret address space came down with a S067 abend. Cleanup is still running on the other systems. The messages from the log are:

 00.27.25 xxxxxxxx  ETCL135I Security interface deactivated
 00.27.25 xxxxxxxx  ETCL910I Dataspace ending, standby 30 seconds
 00.27.55 xxxxxxxx  ETCL113I DBASE file Built: 2025/08/27 06:00  SYSID:ssss
 00.27.56 xxxxxxxx  ETCL109I Dataspace being allocated 44704K
 00.27.57 xxxxxxxx  ETCL130I Security interface activated
 00.27.57 xxxxxxxx  ETCL330I Database reload complete

Resolution

This is a timing issue. The Cleanup address space started the nightly database refresh process around 12:27:25am. During this refresh process, Cleanup drops its dataspace, creates a new one, and updates it with the latest information from the Cleanup database/tracking file. This refresh process completed by 12:27:57am.

A batch job was running during this Cleanup refresh activity. Due to low dispatch priority, this batch job was suspended within the Cleanup processing by the initiator some time before the Cleanup refresh process. The batch job resumed at 12:31:20am. By this time, the Cleanup dataspace had been dropped and its address was no longer valid. When that invalid address was referenced in the batch process, it resulted in the Cleanup failure with the system code S067. Cleanup is designed to terminate during these failures to maintain the integrity of the database.

Below are the criteria that should be satisfied for this issue to happen:

1) A batch/online process should be active within the Cleanup processing

2) This process is suspended just before the Cleanup refresh processing and thus referencing currently active Cleanup dataspace

3) Cleanup waits for 30 seconds before it drops the old dataspace. So the suspended batch process should not resume processing for at least 30 seconds

The batch job satisfied all these criteria for this failure to happen. The batch job's dispatch priority was very low and it did not resume the processing for a long time (4 minutes).