PH20899: SPIN LOOP IN THE MQ RECOVERY PROCESSING

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• An allied thread (in the customer's case, CICS) was cancelled
  in a must complete state due to an unexpected abend 878.
  To prevent an integrity issue, the queue manager must
  terminate, and would require to be started once termination is
  complete. However, due to a timing window as a result of the
  manner the allied threads were cancelled, MQ was unable to
  gracefully come down, resulting in delays in the termination.
  
  During the termination process further abends occurred when
  applications were attempting new connections to the queue
  manager. Due to the small timing window hit in this rare
  circumstance, recovery processing entered a spin loop.
  
  The spin loop may eventually result in SCOM being overlaid with
  10 bytes.
  
  If such overlay happens, abends 5C6-00E80031 can occur in
  applications connecting to MQ due to validation failing early
  in the connect. They happen prior to the MQ termination
  completion.
  

Local fix

• Avoid unexpected allied address spaces termination, such as
  circumstances that lead to abends 878, 80A, 40D.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All users of IBM MQ for z/OS Version 9       *
  *                 Release 1 Modification 0.                    *
  ****************************************************************
  * PROBLEM DESCRIPTION: Various symptoms occur if CSQVSRX is    *
  *                      called while the CPU lock is held.      *
  *                      Symptoms include:                       *
  *                      - Abend 5C6-00E80031 in CSQYALLI        *
  *                      - overlay of the MQ SCOM control block  *
  *                        with 10 bytes ending 'CSQVSTRM'       *
  *                      - disabled loop in recovery processing  *
  *                      - repeated 073 abends reported in       *
  *                        logrec                                *
  ****************************************************************
  An MQ task was dispatched on a TCB with the CPU lock
  unexpectedly held, and the task then called CSQVSRX to suspend
  or resume a ROB.
  CSQVSRX detected that the lock was unexpectedly held, and
  abended 5C6-00E50050, causing CSQVSRRX to be called as part of
  recovery processing.
  CSQVSRRX determined that the queue manager should be terminated
  by CSQVATRM, however this abended S073 due to the CPU lock being
  held. An error in MQ recovery processing caused the 073 abend
  to be retried incorrectly, leading to the SCOM being overlaid,
  and another call to CSQVATRM.
  MQ recovery processing continued to loop until the disabled loop
  was detected by the system and the queue manager was terminated.
  

Problem conclusion

• MQ abnormal termination processing (CSQVATRM) is changed to
  better tolerate being called with the CPU lock held, so that the
  intended abnormal termination of the queue manager can complete.
  

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:

  UI67983

Modules/Macros

• CSQ3EXT2 CSQVATRM CSQVATRR CSQWDSDM
  

Fix information

• Fixed component name

  IBM MQ Z/OS V9

• Fixed component ID

  5655MQ900

Applicable component levels

• R100 PSY UI67983

     UP20/03/11 P F003 ¢

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.