Normally, if HADR primary and standby are not consistent,
reintegration will fail. However, if "db2 activate db dbname" is
issued after the previous command "db2 start hadr on db dbname
as standby" which fails to start up HADR, HADR reintegration
could succeed, and primary and standby get into PEER state.
.
Now, although "db2pd -hadr" shows that HADR is in PEER state,
but data on primary and standby are actually out of sync.
Afterwards, transactions on primary can be committed as normal.
But standby will fail when replaying certain log record.
Following is one possible error from db2diag.log on standby:
2011-10-12-04.46.00.497327-420 I211419E613 LEVEL: Severe
PID : 16594 TID : 47278437951808PROC :
db2sysc
INSTANCE: db2inst1 NODE : 000 DB : SAMPLE
APPHDL : 0-28 APPID: *LOCAL.DB2.111012113926
EDUID : 83 EDUNAME: db2redow (SAMPLE)
FUNCTION: DB2 UDB, data management, sqldRedoUpsert, probe:1864
MESSAGE : Free Space does not match during redo!
DATA #1 : Hexdump, 26 bytes
0x00002B004DA16A78 : 01A2 0200 0400 0000 0C01 A00D 0100 9200
................
0x00002B004DA16A88 : 0000 A40D 0000 0C01 0100
..........
2011-10-12-04.46.00.497534-420 I212033E498 LEVEL: Severe
PID : 16594 TID : 47278437951808PROC :
db2sysc
INSTANCE: db2inst1 NODE : 000 DB : SAMPLE
APPHDL : 0-28 APPID: *LOCAL.DB2.111012113926
EDUID : 83 EDUNAME: db2redow (SAMPLE)
FUNCTION: DB2 UDB, data management, sqldRedoUpsert, probe:1867
MESSAGE : Space used:
DATA #1 : Hexdump, 4 bytes
0x00002AFFDE7FBA68 : 0C01 0000
.... |
