IT16767: MQ client app running under control of XA TM calls MQDISC and MQ later rejects XA calls on the same thread

APAR status

• Closed as program error.

Error description

• An MQ client application is running under control of CICS or
  TUXEDO as XA transaction manager (TM) in a global transaction.
  
  The user code in the application disconnects by calling MQDISC.
   Shortly afterwards, the TM library code make calls to MQ XA
  functions on the same thread.
  
  MQ rejects these calls, with return codes such as -5
  (XAER_INVAL) and -6 (XAER_PROTO).
  
  But this is indeed a valid sequence, so no error should be
  returned by MQ.
  
  Separately, another problem has been seen if the XA application
  calls MQCONN and MQOPEN after calling MQDISC.  This is a valid
  sequence, but the MQ code did not create the 2nd connection
  properly, so the MQOPEN call failed with reason code 2009
  MQRC_CONNECTION_BROKEN.
  
  Separately, it is possible during shutdown of connections to see
  the queue manager end pre-emptively, after writing FDCs with
  these identifiers:
  AT077013/atxAssociationRemove    rc=arcE_XAER_PROTO
  AT080004/atxAssociationRemoveAll rc=arcE_XAER_PROTO
  ZL043105/zlaMainThread           rc=STOP_ALL
  
  These problems might also be seen when ending client-connected
  channels from applications hosted in JEE application servers
  such as WebSphere Application Server.
  
  These problems were introduced as a side effect in APAR IT08918.
  

Local fix

Problem summary

• ****************************************************************
  USERS AFFECTED:
  Users running a client application (connected to MQ over TCP/IP)
  under control of external XA TM software (for example, WebSphere
  Application Server, CICS or TUXEDO) in a global transaction.
  The fix applies to the server side, not the client side.  The
  problem only exists in server-side code levels that include
  IT08918 (beginning 7.5.0.6, 8.0.0.4, 9.0.0.0).
  
  
  Platforms affected:
  MultiPlatform
  
  ****************************************************************
  PROBLEM DESCRIPTION:
  When the SVRCONN channel program received the MQDISC request
  from the client, it called xa_close on the application's behalf.
   This caused later XA calls to be rejected because the XA
  context had been closed down.
  
  In certain states it was also possible for the queue manager's
  cleanup activity to attempt operations in an order that broke
  some internal rules, and caused the pre-emptive ending of the
  queue manager.
  

Problem conclusion

• The MQ product code has been corrected, so when the SVRCONN
  channel program receives the MQDISC request from the client, it
  does not call xa_close on the application's behalf.
  
  ---------------------------------------------------------------
  The fix is targeted for delivery in the following PTFs:
  
  Version    Maintenance Level
  v7.5       7.5.0.8
  v8.0       8.0.0.6
  v9.0 CD    9.0.2
  v9.0 LTS   9.0.0.1
  
  The latest available maintenance can be obtained from
  'WebSphere MQ Recommended Fixes'
  http://www-1.ibm.com/support/docview.wss?rs=171&uid=swg27006037
  
  If the maintenance level is not yet available information on
  its planned availability can be found in 'WebSphere MQ
  Planned Maintenance Release Dates'
  http://www-1.ibm.com/support/docview.wss?rs=171&uid=swg27006309
  ---------------------------------------------------------------
  

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:

  PI73983

Fix information

• Fixed component name

  WMQ BASE MULTIP

• Fixed component ID

  5724H7251

Applicable component levels

• R800 PSY

     UP