IC83569: COMMIT CONTROL ERROR AND FDC GENERATED WHEN A PERSISTENT REFERENCE MESSAGE IS PUT TO A QUEUE IN MQ V5.3 NSK

WebSphere MQ for HP NonStop Server V5.3.1 fix pack 5.3.1.9

APAR status

• Closed as program error.

Error description

• When a persistent reference message is put to a queue the
  following
  fdc is generated and there is a commit control error:
  .
  Persistent Reference messages sent over a channel cause Commit
  Control Error Identification: -
  The receiver side generates the following FFST when trying to
  process the reference message:-
  +--------------------------------------------------------------
  |
  | WebSphere MQ First Failure Symptom Report
  | =========================================
  |
  | Date/Time         :- Wednesday April 25 14:12:41 EDT 2012
  | Host Name         :- xxxxxx (NONSTOP_KERNEL J06.12)
  | PIDS              :- 5724A3901
  | LVLS              :- 531.8  FP5318
  | Product Long Name :- WebSphere MQ for HP NonStop Server
  | Vendor            :- IBM
  | Probe Id          :- CS075003
  | Application Name  :- MQM
  | Component         :- craWriteSync
  | Build Date        :- May 13 2011
  | CMVC level        :- WMQ-5-3-1-8
  | Build Type        :- IKAP - (Production)
  | UserID            :- 44,3 (MQM.XXXXX)
  | Program Name      :- /home/xxxxx/wmqd/opt/mqm/bin/amqrmppa_r
  | Process           :- 1,1132 $X1120 OSS(538509550)
  | Thread            :- 00000004
  | QueueManager      :- FRED
  | Major Errorcode   :- rrcE_COMMIT_CONTROL_ERROR
  | Minor Errorcode   :- OK
  | Probe Type        :- MSGAMQ9538
  | Probe Severity    :- 2
  | Probe Description :- AMQ9538: Commit control error.
  | FDCSequenceNumber :- 0
  | Comment1          :- Error 2232 returned from
  lpiSPIHPNSSTxInfo
  |
  +---------------------------------------------------------------
  -
  .
  MQM Function Stack
  rriReceiveData
  rriConfirmAccept
  craWriteSync
  xcsFFST
  

Local fix

Problem summary

• Error is caused because the channel is attempting to commit a
  message batch,  but the reference message is not put to a queue,
  a TMF transaction is never started.
  Reference messages use DLQ handling logic to to discard the
  message, but because 'MQRO_DISCARD_MSG' is set, the message is
  not actually PUT to the DLQ, so no transaction is started.
  

Problem conclusion

• In message exit processing, if a batch commit is requsted and no
  TMF transaction has been started, then start one
  

Temporary fix

• Change application to use non-persisent messages. If message
  persistence is inherited from the initial queue, change the
  queue definition to DEFPSIST(yes)
  

Comments

APAR Information

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

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

Fix information

• Fixed component name

  WEBS MQ NSS MIP

• Fixed component ID

  5724A3901

Applicable component levels

• R531 PSY

     UP