PI41442: ABEND026 REASON=08118001 CONNECTOR HANG

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• An abends026 rsn08118001 occurred with the following title:
  .
     ABEND=S026,REASON=08118001,CONNECTOR HANG:
     CONNAME=DB2_DSN1,JOBNAME=DSN1DBM1
                                                                 .
  Messages were given out, such as:
  *IXL040E CONNECTOR NAME: DB2_DSN1, JOBNAME: DSN1DBM1, ASID:00xx
   HAS NOT RESPONDED TO THE REBUILD CONNECT EVENT, IXLREBLD
   REQUEST= COMPLETE EXPECTED.
   DUPLEXING REBUILD PROCESSING FOR STRUCTURE DSNxxxx_GBP2 CANNOT
   CONTINUE.
  ..
  *IXL049E HANG RESOLUTION ACTION FOR CONNECTOR NAME: DB2_DSN1
   TO STRUCTURE DSNxxxx_GBP2, JOBNAME: DSN1DBM1, ASID: 00xx:
   SFM POLICY REQUIRES MANUAL INTERVENTION.
                                                                 .
  The event exit (thread) for GBP2 'Corrid: 010.GC1REE2' was
  suspended in dsnb1pcd waiting for copy rebuild castout to
  finish.
  The castout engine being waited for has 'corrid 010.PCOPCO02'
  (castout engine for rebuild), and was waiting for I/O to finish.
                                                                 .
  DB2 went down this castout path due to an arithmetic error
  during its determination of whether to do copy or castout to get
  changed pages from old to new structure. The error caused DB2 to
  think the pages would not fit, and went down the castout path.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: DB2 data sharing users.                      *
  ****************************************************************
  * PROBLEM DESCRIPTION: ABEND026 with REASON=08118001,          *
  *                      CONNECTOR HANG, during a DB2 group      *
  *                      buffer pool rebuild.                    *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  When performing a GBP rebuild, DB2 will copy all of the changed
  pages from the old structure to the new one if the number of
  pages in the new structure is more than the number of changed
  pages in the old structure.  Otherwise, the pages will be cast
  out from the old structure to DASD.
  
  For GBP structures larger than around 2048 megabytes, the check
  for available pages in the new structure may come up with an
  incorrect result due to an arithmetic overflow, leading DB2 to
  do the castout process even though the new structure had plenty
  of room for all of the changed pages in the old structure.
  
  Since the castout process is significantly slower than the copy
  process, this may result in a "connector hang" abend if there is
  a very large number of changed pages.
  

Problem conclusion

• The GBP rebuild logic has been modified to avoid the arithmetic
  overflow when calculating the available page count for the new
  structure.
  

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:

  UI28804 UI28805

Modules/Macros

• DSNB1GBR DSNB1REE
  

Fix information

• Fixed component name

  DB2 OS/390 & Z/

• Fixed component ID

  5740XYR00

Applicable component levels

• RA10 PSY UI28804

     UP15/07/10 P F507

• RB10 PSY UI28805

     UP15/07/10 P F507

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.