PI61168: WHEN CANCELLING A JOB IN THE MIDDLE OF A CHECKPOINT , PRF MAY NOT HANDLE THE RESULTING IN-DOUBT CHECKPOINT PROPERLY

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• When cancelling a job in the middle of a checkpoint, PRF may not
  handle the resulting in-doubt checkpoint properly. There is a
  point in time where an in-process checkpoint creates a condition
  in the CTA/B file where the timestamps of the checkpoint
  records for the current checkpoint match the timestamps of the
  checkpoint records for the last checkpoint. When this happens,
  PRF discards the indoubt checkpoint, instead of notifying the
  user that an in-doubt situation occurs. In the case where the
  current checkpoint actually completed, this causes PRF to
  restart at the wrong checkpoint.
  

Local fix

• If possible, restore the databases in question to the point in
  time where the job started, then restart the job from the
  beginning.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All Users of Program Restart Facility V2R1.  *
  ****************************************************************
  * PROBLEM DESCRIPTION: If a PRF-controlled job abends in the   *
  *                      middle of processing a checkpoint, it   *
  *                      is possible for PRF to choose the prior *
  *                      checkpoint taken as the checkpoint from *
  *                      which to restart the job.               *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  PRF keeps information for the last two checkpoints taken by a
  program. To deal with the problem that can be caused by an abend
  in the midst of checkpoint processing, PRF first catalogs the
  checkpoint prior to sending it off to IMS for processing. At
  this point, PRF sets a bit in the CTDS file containing the
  checkpoint information that indicates this checkpoint has not
  finished processing yet, and is therefore "indoubt". PRF then
  sends the checkpoint off to IMS for processing. When IMS
  returns, PRF then sets a bit in the CTDS file containing the
  checkpoint information that indicates the checkpoint has been
  processed by IMS and is now available to be used for restart. If
  the job abends between the time IMS processes the checkpoint and
  the time PRF sets the bit indicating the checkpoint can be used
  for restart, PRF will choose the prior checkpoint for restart
  purposes. What PRF should do in this situation is to treat the
  current checkpoint as indoubt.
  
  NOTE: This situation can occur if you OS-cancel a batch DL/I or
  DBB job, or if you use the /STO REGION xxxx ABDUMP command to
  stop a BMP. If you think you may need to abnormally terminate a
  PRF-controlled job, you should specify FSTOP=YES in a GBL or
  job-related statement in your PRF inclusion options file, or in
  the job's IRT$CNTL file, and stop the job using the OS MODIFY
  <jobname>,STOP command. This will help avoid indoubt restart
  situations.
  

Problem conclusion

• The Program Restart Facility restart logic has been modified to
  correctly recognize this situation as indoubt.
  

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:

  UI38280

Modules/Macros

• IRTAOPT0 IRTFXSR0 IRTIXT09 IRTIXT10 IRTIXT11
  IRTIXT12 IRTIXT13 IRTPRE00 IRTXRI09 IRTXRI10 IRTXRI11 IRTXRI12
  IRTXRI13 IRTXSC09 IRTXSC10 IRTXSC11 IRTXSC12 IRTXSC13
  

Fix information

• Fixed component name

  IMS PGM RESTART

• Fixed component ID

  5655E1400

Applicable component levels

• R210 PSY UI38280

     UP16/06/03 P F606

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.