PH24659: SHUTDOWN IMMED WITH CICS 5.5 USING COMM SERVER CICS SOCKETS WITH THE OTE=YES WAITS AND WILL NOT TERMINATE.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• The TCPIP sockets code which earlier ran under L8007 as part of
  TASK-00066 has created a USS process for TCB L8007. During
  shutdown, several tasks get purged including TASK-00066. This
  task was running OPENAPI under L8007 when the purge got issued
  which resulted in an ABTERM of the L8 TCB. From this point,
  nothing can execute on L8007 and any CHANGE_MODEs to this TCB
  will fail. At task termination, new code added to DFHDSKE in
  CICS TS 5.5 attempts to release any USS process associated with
  an open TCB. However, this code fails to take account of the
  fact that the L8 is no longer be available. It ends making a
  BPX1MPC (process cleanup call) while running on the QR TCB
  (after the CHANGE_MODE to L8 fails). The code should never do
  this. This appears to hang for a period. Eventually, this call
  triggers a series of severe errors elsewhere in CICS.
  
  This is what the task's kernel stacks look like:
  
  KE_NUM @STACK LEN TYPE ADDRESS LINK REG OFFSET ERR NAME
  
  00F5 28D61040 0200 Bot 27102F00 A71033FE 0004FE DFHKETA
  00F5 28D61240 03D0 Dom 2711F298 A711FC3E 0009A6 DFHDSKE
                     Int +000946 A711F726 00048E FREE_USS_PROCESS
  
  Trace shows that the CHANGE_MODE to the L8 TCB fails which is
  what leaves processing on the QR TCB:
  
  0002 DSAT ENTRY - FUNCTION(CHANGE_MODE) MODENAME(L8)
        TASK-XM KE_NUM-00F5 TCB-C/QR
  
  0003 DSAT EXIT - FUNCTION(CHANGE_MODE) RESPONSE(EXCEPTION)
       REASON(TCB_FAILED) OLD_TCB_TOKEN(27315F00 , 00000001)
        TASK-XM KE_NUM-00F5 TCB-C/QR
  
  Additional Symptoms:
  DFHKE0002 severe error code 0506 in DFHKEDS
  DFHSO0002 severe error code 0C48 in DFHSOLS
  BPXP023I THREAD 2B6AC00000000000, IN PROCESS 207, WAS
   TERMINATED BY SIGNAL SIGKILL, SENT FROM THREAD
   2B6A780000000000, IN PROCESS 198, UID 1251, IN JOB jjjjjjjj.
  Additional Symptoms:
  CICS region unresponsive hang
  In Dump, there are many EC6 abends in the kernel error table
  CICS Trace you find the following exception:
  .
  DS 0213 DSKE  *EXC* - FREE_USS_PROCESS_FAILED -
   TASK-XM    KE_NUM-xxxx TCB-C/QR
       1-0000  FFFFFFFF
       2-0000  0000009D
       3-0000  0B7000B9
  Data area 3 is the reason code- 0B7000B9 which means
  JRActiveProcess.  Backing up in the trace, a task was
  running on an open TCB, when CICS Recovery was driven for
  that open TCB, on our main QR TCB:
  ERM *EXC* RECOVERY  PROGRAM(EZACIC01)
   TASK-nnnnn KE_NUM-xxxx TCB-C/QR
  Trace only had DS level 1 tracing, so we don't
  see the change mode to that open TCB that failed.
  This caused the bpx_process_cleanup call to
  incorrectly be issued on the QR TCB, which failed
  with reason code 0B7000B9.
  LMQUEUE waits
  

Local fix

• n/a
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: CICS hangs, other tasks (including      *
  *                      CSOL) abend  and JVM servers disable    *
  *                      after a task is purged.                 *
  ****************************************************************
  CICS is processing an IP CICS Socket workload with OTE=YES.
  This processing causes a USS process to be associated with each
  L8/L9 TCB.  One such task is purged whilst in a wait. This cases
  the TCB to be queued for termination when the task ends. Task
  termination is run on the QR TCB and an attempt is made to
  switch to the L8/L9 in order to make a BPX1MPC call to clean up
  the USS process. Switching TCBs fails as the TCB is queued for
  termination so the BPX1MPC call is made on the QR TCB.
  The BPX1MPC call may hang the QR TCB and will cause a SIGKILL to
  be sent to all USS processes associated with any other TCBs in
  the region. This will cause any other IP CICS Sockets tasks to
  abend and JVM servers to disable.
  
  As a result of this, the following may also be seen:
  - msgDFHKE0002 with serve error code X'0506' in module
  DFHKEDS
  - msgDFHSO0002 with serve error code X'0C48' in module
  DFHSOLS
  -Thread terminated with msgBPXP023I for a thread
  terminated with signal SIGKILL in the CICS region job
  

Problem conclusion

• USS Clean up processing in DFHDSKE, for these TCBs to handle the
  case where the associated L8/L9 TCB has been abterm'd, to not
  continue the processing, and not call BPX1MPC if the change_mode
  response shows the TCB has been terminated.
  

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:

  UI70777 UI70778

Modules/Macros

• DFHDSKE
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R200 PSY UI70777

     UP20/07/29 P F007 ¢

• R300 PSY UI70778

     UP20/07/29 P F007 ¢

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.