PH45794: CHINIT HUNG DURING STOP PROCESS

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• The dump shows that the CHINIT is hung because the supervisor
  is waiting for all of the adapters to end. The supervisor is
  looping round each of the adapters and releasing any which
  haven't ended with a code telling them to end.
  
  This hasn't worked in this case, because both adapters are
  waiting in the QMGR for a post to arrive. The wait is occurring
  to serialize the handle post/wait logic between a successful
  get and subsequent gets. This processing is not awoken by STOP
  QMGR MODE(FORCE) processing, in fact, the expectation is that
  the wait is being issued to be woken by a post in progress.
  .
  Additional symptoms and keywords:
  Systrace shows STIMERM calls that are issued from CSQXSRVT
  in routine WAIT_A_SEC.  The TASK=1 keyword for the CSQXDPRD
  formatter shows that at least one adapter TCB is still active
  and has a "Term" (termination) ECB that is not posted.
  

Local fix

• z/OS Cancel CHINIT address apace
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All users of IBM MQ for z/OS Version 9       *
  *                 Release 2 Modification 0 and                 *
  *                 Release 3 Modification 0.                    *
  ****************************************************************
  * PROBLEM DESCRIPTION: Channel initiator adapter tasks hang    *
  *                      in CSQMGET processing when using        *
  *                      multiplexed svrconn channels            *
  *                      (i.e. SHARECNV > 0).                    *
  *                                                              *
  *                      When the chinit and/or queue manager    *
  *                      are stopped, the hung adapter prevents  *
  *                      normal shutdown completion.             *
  ****************************************************************
  A timing window exists where a handle is posted while a get
  request is posted at the same time as an earlier post is being
  processed by CSQXACNO. This can leave the handle in a state
  where it is being marked as posted, but the post flag is no
  longer on, leading to CSQMGET hanging in ?WAIT processing.
  
  When the chinit is stopped (either explicitly or due to queue
  manager termination), shutdown processing suspends waiting for
  the hung adapter(s) to end, preventing shutdown processing
  completing.
  

Problem conclusion

• CSQXACNO is changed to prevent the post flag being prematurely
  reset.
  

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:

  UI82767 UI82768

Modules/Macros

• CSQXACNO
  

Fix information

• Fixed component name

  IBM MQ Z/OS V9

• Fixed component ID

  5655MQ900

Applicable component levels

• R200 PSY UI82768

     UP22/10/22 P F210 ¢

• R300 PSY UI82767

     UP22/10/22 P F210 ¢

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.