PH70988: JVMSERVER SUDDENLY DISABLED AND RECYCLED DUE TO AKEA ABEND IN DFHSJIN

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• From the job log, we can see that the JVMSERVER recycled was
  due to a AKEA abend from DFHSJIN.  The messages received:
  
  +DFHSJ0001 An abend (code ---/AKEA) has occurred at offset
  X'123A' in module DFHSJIN
  (Module:DFHMEME) CICS symptom string for message DFHSJ0001 is
  PIDS/5655YA100 LVLS/750 MS/DFHSJ0001
  RIDS/DFHSJIN PTFS/UO05978
  AB/UAKEA RIDS/DFHSJIN ADRS/0000123A
  
  +DFHDU0201 ABOUT TO TAKE SDUMP.
  DUMPCODE: SJ0001  , DUMPID: 1/0002
  
  IEA794I SVC DUMP HAS CAPTURED: DUMPID=173 REQUESTED BY JOB
  DUMP TITLE=CICS DUMP: SYSTEM=XXXXXXXX CODE=SJ0001   ID=1/0002
  
  DUMP CAPTURED USING OPTIMIZE=YES
  +DFHDU0202 SDUMPX COMPLETE. SDUMPX RETURN CODE X'00'
  
  +DFHSJ1007 W JVMSERVER is being disabled and restarted
  by CICS because it is in an inconsistent
  
  From the CICS trace we can see that task on TCB T808F is just
  finishing and during the termination processing it gets an error
  within DFHAPLJ.
  SJ    T808F SJ 0B02 SJJS  EXIT  TERMINATE_TRANSACTION/OK
  
  SJ    T808F AP 196C APLJP *EXC* Recovery
  SJ    T808F AP 196D APLJP *EXC* Percolate
  SJ    T808F AP 1962 APLJ  *EXC* Percolate
  DISPATCH_PTHREAD,4A2F2CF0
  SJ    T808F SJ 0203 SJIN  *EXC* RECOVERY_ENTERED
  INVOKE_PTHREAD,49A700D4,INVOKE_SJ_PTHREAD
  SJ    T808F LM 0003 LMLM  ENTRY TEST_LOCK_OWNER       5FA6DE00
  
  Looking at the thread within the JVM we can see that its
  current Java stack is
      bp: 0x0000005b5cf75340  method: void com/ibm/cics/delegate/
  impl/DelegateWrapperImpl.SetCommArea(String,String, byte[])  (Na
       objects: <no objects in this frame>
      bp: 0x0000000000000000  method: void com/ibm/cics/delegate/
  impl/DelegateWrapperImpl.setCommArea(String,String, byte[])  sou
       objects: <no objects in this frame>
      bp: 0x0000005b5cf75d68  method: void
  com/ibm/cics/server/Wrapper.setCommAreaFromDelegate(String,
  String, byte[])  source: Wrapper.java:529
       objects: <no objects in this frame>
      bp: 0x0000005b5cf75e70  method: void com/ibm/cics/server/Wr
  apper.callOSGiClass(com.ibm.cics.server.invocation.InvocationPr
  oxy,com.ibm.cics.server.Task, long)  source: Wrapper.java:2846
       objects: 0x594be53bc8 0x594be58f18 0x594be56bd0
      bp: 0x0000005b5cf75f50  method: void com/ibm/cics/server/Wr
  apper.invokeJvmServerOSGiClass(com.ibm.cics.server.invocation.I
  nvocationProxy,com.ibm.cics.server.Task, long)  source: Wrapper.
       objects: 0x594be53bc8 0x594be55770 0x58024412e0
  0x594be557e8 0x594be557e8 0x5801d45b38 0x5801d45af0 0x5801d45a28
  0x594be55788 0x594be557e8 0x5801d45af0
      bp: 0x0000005b5cf76030  method: void com/ibm/cics/server/Wr
  apper.jvmServerOSGiEntry(com.ibm.cics.server.invocation.Invocat
  ionProxy,long)  source: Wrapper.java:2630
       objects: 0x594be53bc8
  
  This shows that we are within the SetCommArea DTC call (native c
  call) and we can see the same from the java native stack.
  Looking at the kernel domain we can see that we are within the
  DFHTRC64.ASM module (from the location of the BEAR within the
  dll file).
  
  We believe the problem is similar to PH58620 which was in base
  at CICS TS 6.2 but it looks like there might be another route
  through the code at CICS TS 6.2 that needs to be covered.
  
  Additional Symptoms and Keywords:
  ABENDAKEA  KIXREVDAM
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: CICS JVM server is recycled due to a    *
  *                      crash resulting from residual DTC data  *
  *                      associated with the pthread.            *
  ****************************************************************
  If a JVM server incorrectly has residual DTC data associated
  with the pthread and no active task running on that pthread,
  then it will crash in DTC module DFHTRC64, causing the JVM
  server to recycle.
  

Problem conclusion

• CICS has been updated so that this crash will no longer occur
  if there is residual pthread DTC data.
  

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:

  UO07912 UO07913

Modules/Macros

• DFJ@H356
  

Fix information

• Fixed component name

  CICS TS Z/OS V6

• Fixed component ID

  5655YA100

Applicable component levels

• R500 PSY UO07913

     UP26/05/21 P F605

• R600 PSY UO07912

     UP26/05/21 P F605

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.