DSNT376I IDAEMK Timeouts seit DB2 11

Nur zur Info:

Seit dem Upgrade auf DB2 11 for z/OS kam es immer wieder zu überraschenden und zunächst unerklärlichen DSNT376I Timeout-Meldungen im Log des DB2 MSTR, wie z.B.:

13.45.35 STC21046  DSNT376I  -DB2T PLAN=IDAEMK WITH  261                         
   261           CORRELATION-ID=014.IDAEMK02                           
   261           CONNECTION-ID=DB2T                                   
   261           LUW-ID=DEBBSM0N.IPD20NBW.CEE394BE87CE                 
   261           THREAD-INFO=SYSOPR:*:*:*:*:*  :*:*                   
   261           IS TIMED OUT. ONE HOLDER OF THE RESOURCE IS PLAN=...
   261           CORRELATION-ID=MYPROGRM                               
   261           CONNECTION-ID=BATCH                                   
   261           LUW-ID=DEBBSM0N.IPD20NBW.CEE3C3ECF217=37071           
   261           THREAD-INFO=MYUSER1:BATCH:MYUSER1:MYPROGRM:STATIC:146 
   261           ON MEMBER DB2T                                       
13.45.35 STC21046  DSNT501I  -DB2T DSNILMCL RESOURCE UNAVAILABLE  262           
   262              CORRELATION-ID=014.IDAEMK02                       
   262              CONNECTION-ID=DB2T                                 
   262              LUW-ID=DEBBSM0N.IPD20NBW.CEE394BE87CE=0           
   262              REASON 00C9008E                                   
   262              TYPE 00000D01                                     
   262              NAME 00000211.00002012     

Verursacht werden die Meldungen durch den neuen Index Cleanup Daemon des DB2 11.

Mit IBM DB2 PTF PI31459/UI26612 vom 24.04.2015 wird folgende Problematik behohen:

An index pseudo delete cleanup thread needs to acquire DBD lock and table lock before updating the index to remove the pseudo deleted entries. If the required locks have already been held by other applications, index pseudo delete cleanup thread could timeout and issue many DSNT376I and DSNT501I messages. Because index pseudo delete cleanup threads tend to give other applications higher priority, normally they are the victims in the timeout situation. Index pseudo delete cleanup threads are designed to run in the background and to not block other applications, issuing a large amount of timeout messages. This can be alarming to the users while there is no real contention in the system.

Siehe auch http://www-01.ibm.com/support/docview.wss?uid=swg1PI31459