This post is more than 5 years old
3 Posts
0
3002
db2syscs.exe crashes DB2 server during online backup: Bringing down all db2fmp processes as part of db2stop
Hi all,
We are facing unregular database crashes during online backups.
Facts:
IBM DB2 9.5 Fix Pack 8 64-bit & IBM DB2 9.5 Fix Pack 4 64-bit (AGENT_STACK_SZ=256)
2-4 databases per instance (1 instance per server)
Windows 2003 Server 64-bit SP2
Networker Module for DB2 4.0 (libnsrdb2.dll from 30.05.2009)
NetWorker Client 7.5.2 64-bit
db2diag.log:
2011-10-12-14.01.19.478000+120 E896654F508 LEVEL: Info
PID : 1840 TID : 2528 PROC : db2syscs.exe
INSTANCE: DB2_BXIP NODE : 000 DB : BXIP_ARP
APPHDL : 0-64948 APPID: *LOCAL.DB2_BXIP.111012120121
AUTHID : BXIP_BACKUP
EDUID : 2528 EDUNAME: db2agent (BXIP_ARP) 0
FUNCTION: DB2 UDB, database utilities, sqluxGetDegreeParallelism, probe:509
DATA #1 :
Autonomic backup/restore - using parallelism = 4.
2011-10-12-14.01.20.556000+120 E897164F532 LEVEL: Info
PID : 1840 TID : 2528 PROC : db2syscs.exe
INSTANCE: DB2_BXIP NODE : 000 DB : BXIP_ARP
APPHDL : 0-64948 APPID: *LOCAL.DB2_BXIP.111012120121
AUTHID : BXIP_BACKUP
EDUID : 2528 EDUNAME: db2agent (BXIP_ARP) 0
FUNCTION: DB2 UDB, database utilities, sqluxGetAvailableHeapPages, probe:620
DATA #1 :
Autonomic BAR - heap consumption.
Targetting (50%) - 2480 of 4960 pages.
2011-10-12-14.01.20.556000+120 E897698F521 LEVEL: Info
PID : 1840 TID : 2528 PROC : db2syscs.exe
INSTANCE: DB2_BXIP NODE : 000 DB : BXIP_ARP
APPHDL : 0-64948 APPID: *LOCAL.DB2_BXIP.111012120121
AUTHID : BXIP_BACKUP
EDUID : 2528 EDUNAME: db2agent (BXIP_ARP) 0
FUNCTION: DB2 UDB, database utilities, sqlubTuneBuffers, probe:871
DATA #1 :
Autonomic backup - tuning enabled.
Using buffer size = 577, number = 4.
2011-10-12-14.01.20.587000+120 E898221F469 LEVEL: Info
PID : 1840 TID : 2528 PROC : db2syscs.exe
INSTANCE: DB2_BXIP NODE : 000 DB : BXIP_ARP
APPHDL : 0-64948 APPID: *LOCAL.DB2_BXIP.111012120121
AUTHID : BXIP_BACKUP
EDUID : 2528 EDUNAME: db2agent (BXIP_ARP) 0
FUNCTION: DB2 UDB, database utilities, sqlubSetupJobControl, probe:1523
MESSAGE : Starting an online db backup.
2011-10-12-14.01.21.322000+120 I898692F415 LEVEL: Severe
PID : 1840 TID : 6460 PROC : db2syscs.exe
INSTANCE: DB2_BXIP NODE : 000
EDUID : 6460 EDUNAME: db2med.2528.0 (BXIP_ARP) 0
FUNCTION: DB2 UDB, base sys utilities, sqleCleanupResources, probe:5
DATA #1 : Hexdump, 4 bytes
0x0000000030B4E2B0 : 0000 0000
....
2011-10-12-14.01.21.353000+120 I899109F481 LEVEL: Warning
PID : 1840 TID : 6460 PROC : db2syscs.exe
INSTANCE: DB2_BXIP NODE : 000
EDUID : 6460 EDUNAME: db2med.2528.0 (BXIP_ARP) 0
FUNCTION: DB2 UDB, routine_infrastructure, sqlerKillAllFmps, probe:5
MESSAGE : Bringing down all db2fmp processes as part of db2stop
DATA #1 : Hexdump, 4 bytes
0x0000000030B4E120 : 0000 0000 ....
Resolution: ???
Does anybody has a solution for this problem? Might it be related to this 32-bit issue https://solutions.emc.com/emcsolutionview.asp?id=esg114206?
Thanks & regards,
Andreas
anza_55946c
3 Posts
0
January 16th, 2012 02:00
Hi all,
The DB2 crash was created by libnsrdb2.dll and is limited to Windows only. EMC provided a hotfix in the meanwhile.
The IBM technote http://www-01.ibm.com/support/docview.wss?uid=swg21578016 is referring to this problem.
Best regards,
Andreas
ble1
2 Intern
2 Intern
•
14.3K Posts
0
October 19th, 2011 07:00
Andreas, the part you quoted is normal part of message. Where is the part with crash?
anza_55946c
3 Posts
0
October 19th, 2011 09:00
Hi Hrvoje,
There is no other important information in the db2diag.log file. Just the message "Bringing down all db2fmp processes as part of db2stop" and a of corse a crash file is written and already analyzed by IBM last level support. Unfotunately it also doesn't contain any further information. So IBM came back wih the following recommondation:
...but the fact that the instance dies without producing a trap file, suggests that the libnsrdb2.dll file somehow replaces the trap handler for db2 so that db2 is unable to produce a trap file with diagnostics. So if the problem still happens, then chances are that we will still not see a trap file. Hence, I would suggest to make use of ADPlus to obtain a dump file when the instance dies.
Unfortunately it is very difficult to catch the next crash because we have several installations and the databases die very unregulary.
Best regards,
Andreas
ble1
2 Intern
2 Intern
•
14.3K Posts
0
October 20th, 2011 04:00
You can also try to use NMDA... I'm not sure how much different code is for DB2, but perhaps it does make some difference.
The error itself is something that used to exists with DB2 v8 in the past and was fixed by IBM.... not sure if this thingy is back now or this is something else.