In HADR environments with DBHEAP constrained at fairly low
level,
DB2 fast writer EDUs (required for event monitor infrastructure)
might fail to initialize properly during the takeover. As the
result following messages will be logged to db2diag.log:
2016-07-14-15.12.34.936413+120 E230488573E619 LEVEL:
Error
PID : 6210 TID : 140684346517248 PROC :
db2sysc 0
INSTANCE: db2inst1 NODE : 000 DB :
SAMPLE
HOSTNAME: hostname
EDUID : 100 EDUNAME: db2hadrs.0.0 (SAMPLE) 0
FUNCTION: DB2 UDB, database monitor,
sqmFastWriterQueueMgr::initFastWriterQueueMgr, probe:20
MESSAGE : ZRC=0x8B0F0002=-1961951230=SQLO_NOMEM_DBH
"No memory available in 'Database Heap'"
DIA8302C No memory available in the database heap.
DATA #1 : String, 42 bytes
Error occured initializing fast writer mgr
2016-07-14-15.12.34.936726+120 I230489193E548 LEVEL:
Severe
PID : 6210 TID : 140684346517248 PROC :
db2sysc 0
INSTANCE: db2inst1 NODE : 000 DB :
SAMPLE
HOSTNAME: hostname
EDUID : 100 EDUNAME: db2hadrs.0.0 (SAMPLE) 0
FUNCTION: DB2 UDB, database monitor,
sqmFastWriterManager::initFastWriterManager, probe:20
MESSAGE : ZRC=0x8B0F0002=-1961951230=SQLO_NOMEM_DBH
"No memory available in 'Database Heap'"
DIA8302C No memory available in the database heap.
and one of the fast writer EDUs might trap with SEGV:
2016-07-14-15.12.39.058618+120 I230500996E613 LEVEL:
Error
PID : 6210 TID : 140684287796992 PROC :
db2sysc 0
INSTANCE: db2inst1 NODE : 000 DB :
ICMPLSDB
APPHDL : 0-25 APPID:
10.237.113.40.42823.160714131238
AUTHID : ICMCONCT HOSTNAME: ulxxpcmls2
EDUID : 103 EDUNAME: db2agent (ICMPLSDB) 0
FUNCTION: DB2 UDB, base sys utilities, sqleagnt_sigsegvh,
probe:1
MESSAGE : Error in agent servicing application with coor_node:
DATA #1 : Hexdump, 2 bytes
0x00007FF39CC0F6B2 : 0000
with following functions on the top of the stack:
sqmFastWriter::initFastWriter
sqmFastWriterManager::startupWriters
sqeLocalDatabase::StartBackgroundAgents
sqeApplication::InitEngineComponents
sqeApplication::AppStartUsing
sqeApplication::AppLocalStart
sqlelostWrp
The trap will be sustained, that is, will not bring down DB2
engine, but to recover one will have to follow the steps from
"Recovering from sustained traps" Knowledge Center chapter. Most
importantly, after such a trap regular TAKEOVER will fail
because PRIMARY will not be able to force off trapped EDU, so
TAKEOVER BY FORCE must be used. |
