IT32238: IN VERY RARE SCENARIOS CONCURRENT UPDATES OR INSERTS INTO INSERTTIME CLUSTERING (ITC) TABLES CAN LEAD TO DEADLATCH

Db2 Version 11.1 Mod 4 Fix Pack 6 for Linux, UNIX, and Windows
Db2 Version 11.1 Mod 4 Fix Pack 6 for Linux, UNIX, and Windows

APAR status

• Closed as fixed if next.

Error description

• The deadlatch is between a page latch
  (SQLO_LT_SQLB_BPD__bpdLatch_SX) and the DMS Page Cache Latch
  (SQLO_LT_SQLD_PAGE_CACHE__pageCacheLatch)
  
  The deadlatch is encountered during two or more concurrent
  update and/or insert statements on an ITC table. ITC tables are
  the only tables that are affected. Furthermore, for an ITC table
  to be susceptible it requires very specific conditions related
  to blocks being in the possibly-empty state while the concurrent
  transaction utilizes that block.
  
  The stack of one of the deadlatched threads will look like the
  following.  sqldbFindAppendOrEmptyBlock will be involved:
  
  SQLO_SLATCH_CAS64::getConflictComplex(const unsigned long)
  SQLO_SLATCH_CAS64::getConflict(const unsigned long)
  IPRA.$sqlbFindPageInBPOrSim(SQLB_FIX_CB*,int,unsigned
  long,SQLB_BufferPool*)
  sqlbfix(SQLB_FIX_CB*)
  sqldbFindAppendOrEmptyBlock(sqeAgent*,SQLD_DFM_WORK*,SQLD_TABLE_
  CACHE*,SQLD_TABLE_CACHE*,unsigned long*,SQLZ_RID*)
  IPRA.$sqldCreateAppendCachePagesForMDCTable(SQLD_DFM_WORK*,SQLD_
  PAGE_CACHE*,SQLB_FIX_CB*,long,unsigned
  int*,SQLD_TABLE_CACHE*,SQLD_TABLE_CACHE*,dmsBatchCreateCB*)
  sqldCreateNewAppendCache(SQLD_DFM_WORK*,SQLD_PAGE_CACHE*,unsigne
  d long,unsigned long,unsigned
  long*,SQLD_TABLE_CACHE*,SQLD_TABLE_CACHE*)
  sqldExtendAppendCache(SQLD_DFM_WORK*,unsigned
  long,SQLD_TABLE_CACHE*,SQLD_TABLE_CACHE*)
  sqldSearchPageCache(SQLD_DFM_WORK*,SQLD_PAGE_CACHE*,SQLD_TABLE_C
  ACHE*,SQLD_TABLE_CACHE*)
  sqldb_insertITCRow(SQLD_DFM_WORK*,SQLD_TABLE_CACHE*,SQLD_TABLE_C
  ACHE*)
  sqldMDCInsertRow(SQLD_DFM_WORK*,SQLD_PAGESLOT*,SQLD_PAGESLOT*,SQ
  LP_LOCK_INFO*)
  IPRA.$sqldCreateOverflow(SQLD_DFM_WORK*,SQLD_PAGESLOT,SQLD_PAGES
  LOT,SQLD_PAGESLOT,unsigned long,unsigned long,unsigned long)
  IPRA.$sqldUpdateRowComplex(SQLD_DFM_WORK*,SQLD_XOR_UPDATE*,unsig
  ned char*,int)
  sqldUpdateRow(SQLD_DFM_WORK*,SQLD_XOR_UPDATE*,unsigned
  char*,int)
  sqldRowUpdate(sqeAgent*,SQLD_CCB*,unsigned
  long,SQLD_VALUE**,unsigned long,void**)
  sqlriupd(sqlrr_cb*)
  sqlriSectInvoke(sqlrr_cb*,sqlri_opparm*)
  sqlrr_process_execute_request(sqlrr_cb*,int,bool)
  sqlrr_execute(db2UCinterface*,UCstpInfo*)
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * Users of INSERT TIME CLUSTERING (ITC) TABLES                 *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * See Error Description                                        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Upgrade to 11.1.4.6 (mod 4 fix pack 6)                       *
  ****************************************************************
  

Problem conclusion

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:

Fix information

Applicable component levels

• RB10 PSY

     UP