The CA Dispatch started task abends with a U3970 preceded by program check messages.
The IDMS r10.2 messages and codes manual documents possible causes for a U3970 abend as:
* Error during physical read/write operation.
* Error condition during recovery.
* System internal error or internal storage overwritten.
* Program check or abend in a user defined abend exit.
* A timeout occurred during abend processing.
* A deadlock occurred during abend processing
Anytime the the U3970 abend is preceded with program check messages in either, module IDMSDBMS at offset BA6A or module IDMSDBIO at offset 1F1C, the abend occurred because of a journal sequence number field (DMCJTSEQ) overflow problem that exists in the version if IDMS (r10.2) that CA Dispatch utilizes.
In release 10.2 of IDMS, this DMCJTSEQ field is only four bytes in length and can overflow. The field has been expanded in more current releases of IDMS.
* A message warning you of this impending overflow condition may be observed in the output produced by the DSEXJNL1 job that gets submitted by the CA Dispatch STC each time a journal file fills up and needs to be offloaded. The warning message appears in the DSEXJNL1 job output as:
When the U3970 abend occurs as the result of a journal sequence overflow problem, the appropriate resolution is as follows:
1. Restart the CA Dispatch STC using the special command S DISPATCH,START=DSSTAR00 so that no subtasks get restarted.
2. Confirm a WARMSTART SUCCESSFUL message is received during startup and as soon as CA Dispatch is all the way up, issue the 99STOPCADS command to initiate the normal CA Dispatch shutdown routine.
NOTE: During normal CLEAN shutdown processing, you may see messages containing the text of "ABORTING TASKS" and/or "ABENDED WITH CODE SHUT". Messages containing this specific text are not representative of a problem. These are normal shutdown messages and are to be expected.
3. Submit the CADSJCL library job DSEXJFIX with the MIGRATE symbolic set as MIGRATE='LE' to initialize all four of the existing journal files.
4. Review the output from the DSEXJFIX jobs execution and ensure that the job ended with RC=00.
Initialization of the journal files as outlined above will reset the DMCJTSEQ field counter and will also clean up any other journal file corruption or fragmentation that may have occurred over time.
CA actually encourages our CA Dispatch customers to periodically execute the DSEXJFIX job as part of a regular journal health maintenance. For additional information and details on this subject please refer to knowledge article: Article Id: 43183 - Having problems with your Dispatch journal files? Consider running periodic journal file maintenance.