PK49017: MULTIPLE SYNCONRETURNS OVER IPIC CONNECTION FROM SINGLE TASK FAILS TERMERR, RESP2=17

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Any of the following errors may occur when DPL requests are sent
  over IPIC connections using the SYNCONRETURN option:-
  A transaction hang may occur if a large channel (>60K) is
  returned in response to a DPL request with the SYNCONRETURN
  option.
  Second and subsequent DPL requests with the SYNCONRETURN option
  from a single task fail with TERMERR, RESP2=17. or, when I tried
  to reproduce the problem, INVREQ, RESP2=14.
  An incorrect abend, followed by subsequent transaction hangs, on
  the client side, may occur after a server side time-out.  Should
  this occur then message DFHIS0123 will be issued on the server
  side.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS Users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: When using an IPIC connection, a        *
  *                      transaction hang may occur if a         *
  *                      large channel (>60K) is returned in     *
  *                      response to a DPL request with the      *
  *                      SYNCONRETURN option.                    *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  1. A transaction hang may occur if a large channel (>60K) is
  returned in response to a DPL request with the SYNCONRETURN
  option. This occurs because the end bracket indicator is
  being set before all of the container data is returned to
  the caller.
  2. Second and subsequent DPL requests with the SYNCONRETURN
  option from a single task fail with TERMERR, RESP2=17 or
  sometimes INVREQ, RESP2=14. This occurs because the logic
  that controls the freeing of sessions for SYNCONRETURN
  requests is incorrect.
  3. An abend AITJ, followed by a subsequent transaction
  hang on the client side, may occur after a server
  side time-out. The abendAITJ occurs because the server
  incorrectly sends a purge request back to the client when
  it times out.
  Additional keywords: IPCONN hung LINK
  

Problem conclusion

• 1. DFHISSR has been amended to only set EB on the last element
  of a chain and to not set issb_ebs until EB has been sent.
  DFHISRR has been amended to not set issb_cebr until the last
  element in a chain has been received (to accommodate messages
  received from code that does not have this fix applied).
  2. DFHISIS has been amended to delete the RM link and free the
  ISSB on receipt of EB from the partner, for both good and
  bad responses.
  3. DFHISSR has been amended to only send purge when the client
  (send session) task times out. The transaction hang occurred
  because the server side was not freeing its ISSB either
  (a) at end of task, because EB had not been sent or received
  DFHISZA has been amended to allow server side (receive session)
  ISSBs to be freed even if EB has not been sent or received; it
  is important that the client ISSB is not freed before the
  server, not the other way round.
  (b) on receipt of an incoming request when the
  tid_isrr_task_not_active condition is detected  DFHISRR has
  been amended to set the ISSB state correctly so that IS7 is
  sent for error message DFHIS1023 (although this  condition
  will not occur after this fix, the ISSB will not be found,
  rather than being found but inactive, So DFHIS1019 will be
  issued instead). MSGDFHIS1019 has been amended to better
  describe the problem and DFHISEM amended to add the
  conversation id to the message.
  
  The CICS Transaction Server for z/OS Version 3 Release 2
  Messages and Codes manual, GC34-6827-00, has been updated
  to alter messages DFHIS1019 and DFHIS1023 as follows:
  
  DFHIS1019 DATE TIME APPLID Conversation
            convid no longer active on IPCONN
            ipconn.
  
  EXPLANATION:  The IS domain request/response receiver
  module, DFHISRR, has received an input message for IPCONN
  ipconn with an IPIC HTTP header that contains a
  conversation id for which there is no active ISSB.  This is
  probably due to the late arrival of a request or response
  for a task which has timed out or been purged and
  relinquished its ISSB.
  
  SYSTEM ACTION:  An exception trace is written by IS
  domain, the IPCONN session is added to the IS domain error
  queue for processing by the long running error and message
  task, CISE.
  
  USER RESPONSE:  Inspect the trace, dump and IS domain
  message log to determine the cause of the error.
  
  DESTINATION:  CISO
  
  MODULE:  DFHISEM
  
  XMEOUT PARAMETERS: date, time, applid,
  convid, ipconn
  
  
  DFHIS1023 DATE TIME APPLID Conversation
            error (code X'errorcode') occurred on IPIC
            session name in IPCONN ipconn.
  
  EXPLANATION:  The IS domain request/response receiver
  module, DFHISRR, has received data that it is unable to
  process for the conversation on session block (ISSB)
  name on IPCONN ipconn.  The errorcode
  X'errorcode' is the exception trace point ID which
  uniquely identifies what the error is and where the error
  was detected, for example:
  
  o   0614 The ISSB task is not waiting for data.
  
  o   061C The mirror or user task resume failed unexpectedly.
  
  o   061D No ISSB available for new task attach request.
  
  o   061E The ISSB is not in RECEIVE state; data is not
      expected.
  
  o   0629 Chain state in header does not match ISSB chain
      state.
  
  o   062A Chain element received before pacing message sent.
  
  o   062D Invalid message sequence number.
  
  o   062E Invalid chain sequence number.
  
  o   062F IS header contains unrecognised IBM CCSID.
  
  o   0630 Data received for a task which is no longer active.
  
  SYSTEM ACTION:  An exception entry (code
  X'errorcode' in the message) is made in the trace
  table. The IPCONN session is added to the IS domain error
  queue for processing by the long running error and message
  task, CISE.  The user or mirror task using the ISSB in error
  is abended, either immediately or when it next tries to
  communicate with the partner.  An FMH7 error response is
  sent for sessions in SEND state.
  
  USER RESPONSE:  Inspect the trace, dump and CICS message
  log to determine the cause of the error.
  
  DESTINATION:  CISO
  
  MODULE:  DFHISEM
  
  XMEOUT PARAMETERS: date, time, applid,
  X'errorcode', name, ipconn
  

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:

  UK28918

Modules/Macros

•    DESISEM  DESISIS  DESISRR  DESISSR  DESISZA
  DFHISEM  DFHISEMA DFHISEMM DFHISEMP DFHISEMT DFHISIS  DFHISISA
  DFHISISM DFHISIST DFHISRR  DFHISRRA DFHISRRM DFHISRRP DFHISRRT
  DFHISSR  DFHISSRA DFHISSRM DFHISSRT DFHISZA  DFHISZAA DFHISZAM
  DFHISZAT DFHMEISC DFHMEISE DFHMEISK DFH49017
  

Fix information

• Fixed component name

  CICSTS V3 Z/OS

• Fixed component ID

  5655M1500

Applicable component levels

• R500 PSY UK28918

     UP07/09/14 P F709

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.