PM72512: REALLOCATE OF A DUPLEX GBP MAY GET ABEND026 RSN08118001 AND MESSAGE IXL040E IF THE GBP IS IN SHORTAGE.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• IXL040E CONNECTOR NAME: DB2_DAA1, JOBNAME: DAA1DBM1, ASID: 0093
  HAS NOT RESPONDED TO THE REBUILD QUIESCE EVENT.
           REBUILD PROCESSING
           FOR STRUCTURE DAA0_GBP6 CANNOT CONTINUE.
           MONITORING FOR RESPONSE STARTED: 08/12/2012 09:28:03.
           DIAG: 0000 0000 00000000
  .
  Dump Title: ABEND S026 REASON 08118001 CONNECTOR HANG
              CONNAME=DB2_DAA1 JOBNAME=DAA1DBM1
  REALLOCATE of a duplex GBP may issue above message and ABEND
  after message MSGDSNB743I if the GBP is in shortage status.
  .
  DSNB743I  -DAA1 DSNB1GBR DUPLEXING IS BEING STOPPED
             FOR GROUP BUFFER POOL GBP6
             FALLING BACK TO PRIMARY
             REASON = OPERATOR
             DB2 REASON CODE = 00000000
  .
  This may cause DB2 crash by ABENDS026 RSN08110102 if
  CFSTRHANGTIME interval is defined in the SFM policy.
  

Local fix

• Don't issue REALLOCATE to duplex GBP while it is in shortage
  status ( message DSNB319A and DSNB325A ).
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: DB2 data sharing users.                      *
  ****************************************************************
  * PROBLEM DESCRIPTION: Various types of delays following a     *
  *                      rebuild of a group buffer pool.  This   *
  *                      can result in application hangs, lock   *
  *                      timeouts, or IXL040E messages stating   *
  *                      that DB2 is not responding to a rebuild *
  *                      event.                                  *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  At the end of a group buffer pool rebuild, DB2 goes through a
  "resume castout" process attempting to reschedule castouts that
  were interrupted by the rebuild.  There are timing windows that
  can cause this process to take longer than it should (waiting
  for castouts that started after the rebuild finished), or even
  to take forever (waiting for castout completion when no castout
  engine is actually running).  These delays can cause application
  hangs, usually waiting on a drain lock held by the pseudo-close
  process.  They can also cause subsequent GBP rebuild events to
  be ignored, resulting in IXL040E messages complaining that DB2
  is not responding.
  

Problem conclusion

• The pageset castout logic has been modified to no longer handle
  a GBP rebuild by requeueing its request and terminating the
  engine, requiring a reschedule at the end of the rebuild.  The
  castout engine will now remain active but wait on the GBP fence
  for the rebuild to complete.  Additionally, the "resume castout"
  step at the end of the GBP rebuild has been eliminated.
  

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:

  UK83514 UK83515

Modules/Macros

• DSNB1GBR DSNB1PCD DSNB5PCO
  

Fix information

• Fixed component name

  DB2 OS/390 & Z/

• Fixed component ID

  5740XYR00

Applicable component levels

• RA10 PSY UK83514

     UP12/11/29 P F211

• R910 PSY UK83515

     UP12/11/29 P F211

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.