During a fail-over, if a local database has multiple alias
(and NUMDB is set to > 1), and there is an attempt to
restart/recover that same database multiple time in parallel, it
may cause the race condition in restarting that same database
and this can cause the panic in the target host.
The following is the snippet of db2diag.log:
2011-11-30-17.52.12.867102+540 I336592832A891 LEVEL: Info
PID : 4653228 TID : 16963 PROC :
db2sysc 3
INSTANCE: pmninst NODE : 003 DB : TPCC
APPHDL : 3-53 APPID: *N3.DB2.111130085206
HOSTNAME: c1mem01
EDUID : 16963 EDUNAME: db2agnti (TPCC ) 3
FUNCTION: DB2 UDB, base sys utilities,
sqeLocalDatabase::SubsequentConnect, probe:9005
DATA #1 : String, 20 bytes
DB is not yet active
DATA #2 : SQLCA, PD_DB2_TYPE_SQLCA, 136 bytes
sqlcaid : SQLCA sqlcabc: 136 sqlcode: -952 sqlerrml: 0
sqlerrmc:
sqlerrp : SQLEDOSD
sqlerrd : (1) 0x00000000 (2) 0x00000000 (3)
0x00000000
(4) 0x00000000 (5) 0x00000000 (6)
0x00000003
sqlwarn : (1) (2) (3) (4) (5) (6)
(7) (8) (9) (10) (11)
sqlstate:
2011-11-30-17.52.12.873432+540 I336593724A947 LEVEL:
Warning
PID : 4653228 TID : 16963 PROC :
db2sysc 3
INSTANCE: pmninst NODE : 003 DB : TPCC
APPHDL : 3-53 APPID: *N3.DB2.111130085206
HOSTNAME: c1mem01
EDUID : 16963 EDUNAME: db2agnti (TPCC ) 3
FUNCTION: DB2 UDB, base sys utilities,
sqeDBMgr::StartUsingLocalDatabase, probe:2582
DATA #1 : String, 75 bytes
Non-fatal error during automatic restart connect. Remapping
below to -1015
DATA #2 : SQLCA, PD_DB2_TYPE_SQLCA, 136 bytes
sqlcaid : SQLCA sqlcabc: 136 sqlcode: -952 sqlerrml: 0
sqlerrmc:
sqlerrp : SQLESBCN
sqlerrd : (1) 0x00000000 (2) 0x00000000 (3)
0x00000000
(4) 0x00000000 (5) 0x00000000 (6)
0x00000003
sqlwarn : (1) (2) (3) (4) (5) (6)
(7) (8) (9) (10) (11)
sqlstate:
....
2011-11-30-17.52.14.352551+540 I336606945A948 LEVEL:
Severe
PID : 4653228 TID : 37780 PROC :
db2sysc 3
INSTANCE: pmninst NODE : 003
HOSTNAME: c1mem01
EDUID : 37780 EDUNAME: db2ard 3
FUNCTION: DB2 UDB, base sys utilities, sqle_panic_light,
probe:10
DATA #1 : <preformatted>
sqle_panic_light: Panic/sleep = 0x0
CALLSTCK:
[0] 0x09000000071FBFF0 pdLogVPrintf + 0xD0
[1] 0x09000000071FBD44 pdLogPrintf + 0x58
[2] 0x0900000006C7A200 sqle_panic_light__Fv + 0x1B0
[3] 0x0900000006C79FFC
@78@sqleARHandleFailedRecoveryOperation__Fi + 0x224
[4] 0x0900000006C79D10
sqleARHandleFailedRecoveryOperation__Fi@glue7B2 + 0x7C
[5] 0x090000000863BC38 processChain__17sqeARChainBuilderFv +
0xD8
[6] 0x0900000006E55E1C sqleARThreadEntry__FPUcUi + 0x3C8
[7] 0x0900000006F868A4 sqloEDUEntry + 0x414
[8] 0x0900000000C24E10 _pthread_body + 0xF0
[9] 0xFFFFFFFFFFFFFFFC ?unknown + 0xFFFFFFFF |
****************************************************************
* USERS AFFECTED: *
* All pureScale users *
****************************************************************
* PROBLEM DESCRIPTION: *
* During a fail-over, if a local database has multiple alias *
* (and NUMDB is set to > 1), and there is an attempt to *
* restart/recover that same database multiple time in *
* parallel, and this can cause the panic in the target host. *
****************************************************************
* RECOMMENDATION: *
* Upgrade to db2 Version 9.8 FixPak 5 *
**************************************************************** |
