PK69816: CICS DISPATCHER PROBLEMS WITH JVM TCBS

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• The root of the problem is within the TCB matching logic of
  the CICS dispatcher.  What can be seen is that a TCB was
  used for a JVM with profile DFHJVMPC.  At some later time
  all available JVMs were needed for transactions using
  profile DFHJVMCD.  This caused the JVM with profile DFHJVMPC
  to be destroyed.  However, two problems occured at
  this point.
  .
  1) The disptacher updated its DFHDSMATCH block for this TCB
     so it contained DFHJVMPC and DFHJVMCD.  This will cause
     any future JVM   requests specifying EITHER profile to be
     matched successfully to this TCB.
  .
  2) The JVM using profile DFHJVMPC that has just been destroyed
     wasnot removed from the JVMSET chain of the master JVM.
  .
  When a request comes in again for JVM profile DFHJVMPC it
  matches the TCB that was just being used for a DFHJVMCD JVM.
  As the match was    successful then the JVM is not destroyed
  and recreated.  So, the JVM  looks like it is running with
  profile DFHJVMPC (because the value in  the SJTCB block has
  been updated) but actually it is still running    with
  DFHJVMCD.  Also the SJTCB_WORKER and SJTCB_XNONRESETTABLE
  flags   have not been set.  Instead SJTCB_XRESETTABLE
  remains set on which is the value for a DFHJVMCD JVM.
  .
  When the TCB for this JVM is finally destroyed the master
  JVM will NOT be notified and so will remain in the system
  using up storage.
  

Local fix

• na
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: CICS short-on-storage condition occurs  *
  *                      with multiple master JVMs waiting for   *
  *                      non-existent worker JVMs to terminate.  *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  When a JVM starts up it may reuse the TCB from a previously
  terminated JVM.  However if it uses a different JVM profile,
  some of the attributes from the previous JVM are not being reset
  correctly, and the previous JVM is not being removed from the
  JVMSET of the master JVM.  When the master JVM is asked to
  terminate, it then waits indefinitely for non-existent worker
  JVMs to terminate.  This may lead to a CICS short-on-storage
  condition with multiple master JVMs waiting for termination,
  and may also prevent the CICS region shutting down.
  
  Additional keywords: java SOS MSGDFHSM0137 MSGDFHSM0139
  DFHSM0137 DFHSM0139 SM0137 SM0139
  

Problem conclusion

• DFHDSDS4 has been changed to allow only a single JVM profile
  match per TCB, and DFHSJIN has been changed to ensure that
  worker JVMs are always removed from the JVMSET.
  

Temporary fix

•             *********
              * HIPER *
              *********
  FIX AVAILABLE BY PTF ONLY
  

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  PK64827

• APAR is sysrouted TO one or more of the following:

  UK39362 UK39363

Modules/Macros

•    DESSJIN  DFHDSDS4 DFHSJIN
  

Fix information

• Fixed component name

  CICSTS V3 Z/OS

• Fixed component ID

  5655M1500

Applicable component levels

• R400 PSY UK39362

     UP08/09/03 P F809

• R500 PSY UK39363

     UP08/09/03 P F809

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.