IT33464: IBM MQ FOR HPE NONSTOP V8 IF A GLOBAL TMF TRANSACTION CANCELED BY AUTOABORT CAN LEAD TO VARIOUS FDCS IN ZLATMFTHREAD FUNCTION.

IBM MQ for HP NonStop Server V8.1 Fix Pack 8.1.0.6

APAR status

• Closed as program error.

Error description

• With IBM MQ for HPE NonStop Version 8, for instance working
  with 'IBM MQ for HPE NonStop I (ia64) 8.1.0.5'.  If a global
  TMF transaction is canceled in the system by AUTOABORT can lead
  to various FDC entries in zlaTMFThread function.  Other
  negative effects can be hanging MQGET calls.
  .
  Following is an example of such a FDC entry (an excerpt of it):
  -----
  Probe Id          :- XC307100
  Application Name  :- MQM
  Component         :- xlsRequestMutex
  ...
  Major Errorcode   :- xecL_W_LONG_LOCK_WAIT
  Minor Errorcode   :- OK
  Probe Type        :- MSGAMQ6150
  Probe Severity    :- 3
  Probe Description :- AMQ6150: IBM MQ semaphore is busy.
  .
  MQM Function Stack
  zlaTMFThread
  atxStart
  xlsRequestMutex
  xcsFFST
  -----
  

Local fix

Problem summary

• Access to a queue can be blocked by an outstanding
  syncpointed MQGET call, in case a global transaction
  it is involved with is aborted.
  Should the wait interval be longer than 5 minutes
  it may happen that FFST entries are generated.
  The queue will be accessible again after the MQGET
  has returned. This is caused by a problem in the
  TMF transaction facility in MQ with
  the MQ wide mutex handling.
  

Problem conclusion

• The TMF and mutex handling has
  been enhanced. In the situation above the MQGET will
  return after 30 seconds, so the queue is not blocked.
  

Temporary fix

• Cancel the application hanging in the MQGET.
  

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

  MQ FOR HPE NS O

• Fixed component ID

  5724A3903

Applicable component levels

• R810 PSY

     UP