PM54811: IPIC CONNECTION ERROR BAD_CONVID FOR SYNCPOINT PROCESSING FOR LONG RUNNING MIRRORS.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• The problem requires a task in the TOR to be communicating over
  at least 2 IPIC sessions to remote systems.  When a syncpoint
  command is issued by the application all bar the last IPIC
  sessions gets sent a prepare message. There is an error in the
  prepare code for IPIC that causes ISSB_CCDS to get turned on.
  This bit being on causes the next synpoint to be issued to
  appear as an end of task syncpoint (whether it is or not).
  This will cause the remote mirror task to terminate.  In this
  case we had a user syncpoint so the mirror task should still
  exist.  When the next message is sent to the AOR it is marked
  as being part of an existing conversation but the mirror has
  terminated so that results in the BAD_CONVID error being
  returned to the TOR resulting in messages DFHIS1019 and
  DFHIS4000.
  .
  If there is only one IPIC session being used by the task in the
  TOR then the problem does not occur as the session is always
  last agent and so does not get sent a prepare message, it just
  gets sent a commit message.  If there is only a single user
  syncpoint in the task then the problem does not occur as the
  next syncpoint is at the end of the task and the mirror is
  expected to go away.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: Message "DFHIS1019 *** Conversation no  *
  *                      longer active on IPCONN." is issued     *
  *                      while SYNCPOINT command is issued to    *
  *                      coordinate among the distributed UOW    *
  *                      through IPIC.                           *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  The problem requires a task in the TOR to be communicating over
  at least 2 IPIC connections to remote AORs. Under these
  circumstances, one of the AORs acts as the last agent and
  becomes the co-ordinator for the distributed unit of work. The
  task in the TOR issues the user syncpoint and two phase commit
  processing takes place between the three regions. Flag ISSB_CCDS
  is turned on during this operation and that causes additional
  processing to take place during any subsequent user syncpoint by
  the same task. When any second syncpoint is issued in the TOR,
  then ps_spc_mod is set to '0000' which notifies the co-ordinator
  (the last agent AOR) to terminate its mirror task. When
  the next message is sent to this AOR, msgDFHIS1019 is issued on
  the TOR.
  Keywords: msgDFHIS4000 bad_convid
  

Problem conclusion

• DFHISCU is changed to not turn on the ISSB_CCDS flag after
  building the IS6 message in dfhiscu_issue_prepare.
  

Temporary fix

• FIX AVAILABLE BY PTF ONLY
  

Comments

APAR Information

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

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

  UK77797

Modules/Macros

•    DFHISCU
  

Fix information

• Fixed component name

  CICS TS Z/OS V4

• Fixed component ID

  5655S9700

Applicable component levels

• R700 PSY UK77797

     UP12/04/13 P F204

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.