PH40903: CICS SUFFERS SOS PROBLEMS IN THE ECDSA AFTER MIGRATING TO CICS 5.6

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• When a program does multiple DPL requests to a program in a
  remote CICS region without specifying SYNCONRETURN the ECDSA
  storage in the remote region climbs by the size of the commarea
  until the ECDSA finally goes short on storage (SOS).
  
  
  The affected CICS region runs with MROLRM and MRO connectivity
  is used.
  
  This problem started to appear after the CICS regions were
  migrated from CICS 5.4 to 5.6.
  
  The dump showed that there are thousands of storage segments
  attached to a long running CSMI task. Those segements got a size
  of 24K each and have an "DFHXFPC-COMM FREEMAIN" eyecatcher.
  
  The investigations showed that the commarea is not being free'd
  once the data has been copied into a container to send back to
  the originating region.
  
  The commareas will be freed as part of the clean up performed
  when the mirror task completes, but until then they will hang
  around as the current code does not free these once it has
  finished with them.
  

Local fix

• Not available
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS Users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: A CICS region that has SIT MROLRM=YES   *
  *                      specified (to use a long running mirror *
  *                      ), receiving DPL requests from a remote *
  *                      region can see an increase in ECDSA.    *
  ****************************************************************
  A CICS region can see an SOS condition when SIT MROLRM=YES is
  specified and a large number of DPL requests are received from
  a remote region.
  ECDSA will show storage segments attached to the long running
  mirror transaction with the DFHXFPC-COMM FREEMAIN eyecatcher.
  These segments are used to hold the commarea data passed to the
  remote region.
  The problem here is that these commareas are not being freed
  until end of task when either:
  - The DPL request has sent a channel and a commarea
  - The DPL request has sent a commarea >24K. In these cases CICS
    uses a channel to pass the data.
  
  Currently when a channel is passed on a DPL request it is
  assumed that there is no commarea to free.
  These commareas do get freed as part of task clean up. However
  in the case where a long running mirror is being used, this can
  be some time in the future and can lead to a build up of
  commareas that are no longer needed.
  

Problem conclusion

• CICS has been changed so that a commarea is freed when it is no
  longer needed as opposed to being freed at end of task.
  

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:

  UI78786

Modules/Macros

• DFHXFP   DFHXFQ
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R300 PSY UI78786

     UP22/01/05 P F201

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.