PI84921: EXCESSIVE ECDSA STORAGE USAGE IN CICS

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• CICS regions show high ECDSA storage usage because of a memory
  leak in subpool RZGENRAL.
  
  CICS is acting as a web service provider.  A request arrives and
  starts a pipeline task.  A handler program within the PIPELINE
  causes a context switch to occur.  A second task gets attached
  in the same region and is connected to the first using request
  streams.
  
  The first task enters an RZCBNOTI suspend to wait for the second
  task to complete.  The DTIMOUT interval expires before the
  second task was able to send its response back.  The first task
  sends a SOAP fault to the client and terminates.
  
  Eventually the second task tries to send its response back to
  the first task.  The DFHRZIS SEND routine obtains a single
  RZISBUFF buffer in the RZGENRAL subpool to hold the entire
  response.  The routine checks and correctly determines that the
  first task is no longer present.  The DFHRZIS SEND routine then
  completes without sending any data, but also without freemaining
  the RZISBUFF buffer.  The address and length of the RZISBUFF are
  held in local variables so any reference to the buffer is lost.
  Eventually CICS would go SOS in ECDSA.
  
  Additional keywords: msgDFHSM0133 SM0133 msgDFHSM0142 SM0142
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICS users.                              *
  ****************************************************************
  * PROBLEM DESCRIPTION: Memory leak in ECDSA RZGENRAL subpool   *
  *                      when CICS is a Web Service provider.    *
  ****************************************************************
  * RECOMMENDATION: .                                            *
  ****************************************************************
  CICS Webservice Provider regions show high ECDSA storage usage
  because of a memory leak in subpool RZGENRAL.
  
  A URIMAP is initiating a background transaction. A context
  switch takes place to run the business application under a
  different transid, running in the same region. If the 'front
  end' transaction gets a DTIMOUT before the business application
  completes, there remains some storage in ECDSA (subpool
  RZGENRAL) that is not released.
  
  The problem is in the DFHRZIS SEND routine used to send the
  reply from the backend task. It correctly detects that the
  partner task is no longer present. This avoids any actual
  transfer of data. However, that check is only done after the
  RZISBUFF has been created and populated.
  
  KEYWORDS: RZ requeststream requeststreams SOS short DFHSM0133
            MSGDFHSM0133
  

Problem conclusion

• CICS Request Stream management has been changed to freemain
  the RZISBUFF storage before returning from the DFHRZIS SEND
  routine when the source task has already completed.
  

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:

  UI52360 UI52361

Modules/Macros

• DFHRZDM  DFHRZDUF DFHRZIX  DFHRZLN  DFHRZNR2 DFHRZRG2 DFHRZRM
  DFHRZRS1 DFHRZSO  DFHRZSO1 DFHRZTA  DFHRZTCX DFHRZTR1 DFHRZTRI
  DFHRZXM
  

Fix information

• Fixed component name

  CICS TS Z/OS V5

• Fixed component ID

  5655Y0400

Applicable component levels

• R000 PSY UI52361

     UP17/12/05 P F712

• R100 PSY UI52360

     UP17/12/05 P F712

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.