Application can not be forced off because there is deadlatch
between its subagent processes which are performing hash join.
The deadlatch can be like the following:
process id: 7024680
Waiting on latch type: (SQLO_LT_sqlriSectInit_C__dynamic_latch)
- Address: (7800011455a0e80) , Line: 11023, File:
sqlri_hsjn_util.C
Holding Latch type: (SQLO_LT_sqlri_hsjnLatch__ioLatch)
- Address: (780000cb6ae1348) , Line: 9817, File:
sqlri_hsjn_util.C
process id: 7594048
Waiting on latch type: (SQLO_LT_sqlri_hsjnLatch__ioLatch)
- Address: (780000cb6ae10c8) , Line: 9817, File:
sqlri_hsjn_util.C
Holding Latch type: (SQLO_LT_sqlriSectInit_C__dynamic_latch)
- Address: (7800011455a0e00) , Line: 9603, File:
sqlri_hsjn_util.C
Holding Latch type: (SQLO_LT_sqlri_hsjnLatch__partLatch)
- Address: (780000cb6ae10c0) , Line: 9804, File:
sqlri_hsjn_util.C
process id: 2601118
Waiting on latch type: (SQLO_LT_sqlriSectInit_C__dynamic_latch)
- Address: (7800011455a0e80) , Line: 11023, File:
sqlri_hsjn_util.C
Holding Latch type: (SQLO_LT_sqlri_hsjnLatch__ioLatch)
- Address: (780000cb6ae10c8) , Line: 9817, File:
sqlri_hsjn_util.C
Process id 7594048 is holding the
SQLO_LT_sqlriSectInit_C__dynamic_latch, that process 7024680 is
waiting for, but this info is not properly dumped in trap file.
The stack of these processes are like:
Stack for process id: 7594048:
sqloXlatchConflict
sqloXlatchAIX
sqloXlatchConflict
sqloXlatchAIX
sqlri_hsjnGetTupleBlock
sqlri_hsjnPartitionFull
sqlri_hsjnNewTuple
sqlrihsjn
sqlriSectInvoke
Stack for process id: 7024680:
sqloXlatchConflict
sqloXlatchAIX
sqlri_hsjnFreeTupleBlockList
sqlri_hsjnFlushBlocks
sqlri_hsjnPartitionFull
sqlri_hsjnNewTuple
sqlrihsjn
sqlriSectInvoke |
