JR30253: AUTO-PURGE (AFTER N RUNS / N DAYS) SHOULD NOT PURGE LOGS FROM ABORTED JOB RUNS

APAR status

• Closed as program error.

Error description

• The customer runs many many instances of the same job
  concurrently.  The auto-purge setting is configured (a single
  digit value for example) to keep the log size manageable and
  avoid issues (startup peformance, timeouts, gui performance,
  etc.) The problem occurs when one of the job instances aborts,
  but the rest complete. The completions will trigger the purge,
  which wipes out the log items from the failed instance, long
  before the user has the opportunity to review the errors. The
  auto-purge should not purge aborted job runs; the user should
  have to do this with manual purge after reviewing the errors and
  taking appropriate action.
  

Local fix

Problem summary

• ****************************************************************
  USERS AFFECTED:
  Customers using Auto-Purging that require failed jobs to not be
  purged so that the status can be investigated.
  ****************************************************************
  PROBLEM DESCRIPTION:
  When Auto-pugring is in use, all instances of a job will be
  purged, without checking for the success or fail status.
  ****************************************************************
  RECOMMENDATION:
  This change is included in 8.0.1 Fix Pack 2.
  ****************************************************************
  

Problem conclusion

• The Auto-Purge functionality has been enhanced so that jobs
  which have failed do not have their Status or Log records
  removed, so that subsequent evaluation of the failure can
  occur. This is enabled by adding an environment variable for the
  relevant project in the DataStage Administrator.
  DS_LOG_AUTOPURGE_IGNORE_STATUS should contain a comma-separated
  list of the status codes to ignore. A failed job has a status of
  3, so the default value of this environment should be 3. By
  default this environment variable will not exist and will have
  no impact on existing Auto-Purge logic. The variable should be
  added in the User-defined section and is likely to slow the
  operation of the Auto-purge code when a job completes. Job Logs
  & Status can still be completely cleared from the DataStage
  Director.
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Modules/Macros

• SERVER
  

Fix information

• Fixed component name

  WIS DATASTAGE

• Fixed component ID

  5724Q36DS

Applicable component levels

• R801 PSY

     UP

• R810 PSY

     UP