MQSeries.net :: View topic - Broker fail to re-start after abnormal shutdown

lwl · Posted: Tue Feb 01, 2005 11:35 pm Post subject:

Hi

We use WBIMB 5.0 CSD04 with MQ5.3 CSD07 in solaris platform.

We are now trying the Broker failover senario. When the broker is shutdown abnormally (use kill -9 or machine hard-boot). The broker is fail to re-start and the following error in system error log file:

#########
Jan 11 18:21:54 sfv240 MQSIv500[19786]: [ID 702911 user.error] (DMBHKL.agent)[1]BIP2228E: Severe error: /build/S500_P/src/CommonServices/U
nix/ImbNamedMutex.cpp 1334 semctl Abend file: /var/mqsi/errors/DMBHKL.agent.13884.1.abend action: abort
Jan 11 18:21:54 sfv240 MQSIv500[19527]: [ID 702911 user.info] (DMBHKG)[1]BIP2001I: The WebSphere Business Integration Message Brokers serv
ice has started, process ID 19793. : DMBHKG.service: /build/S500_P/src/AdminAgent/ControlProcess/sparc_solaris_2/ImbControlService.cpp: 33
8: ImbControlService::StartNewAA: :
#########
Here is the content of the file "DMBHKL.agent.13884.1.abend "
------------------------------------------------------------------------
abend record for pid 13884 tid 1 time Wed Jan 5 16:53:00 2005
File: /build/S500_P/src/CommonServices/Unix/ImbNamedMutex.cpp
Line: 1334
Function: semctl
---- Inserts ----
-1
22
Relayed by ImbNamedMutex::ImbNamedMutex
@(#) 1.11.1.1 CommonServices/Unix/ImbNamedMutex.cpp, CommonServices, S500, S500-CSD03 03/10/18 11:26:27 [4/27/04 21:48:45]
-----------------
----------------------------- Stack dump for current thread ( 1)
(malloc allowed)
(0xfee5a954+0x01f4) __1cIImbAbendXdumpCurrentStacksToFile6Fpcpkv_v_ (0x1400, 0x0, 0x145c, 0xa, 0xfef75a64, 0xfef429a8) in /opt/mqsi/lib/li
bCommonServices.so
(0xfee59a2c+0x0194) __1cIImbAbendWlogSystemErrorInternal6Fpkcki2rkn0AGInsert_66666pkvkb_v_ (0xfef54616, 0x536, 0xfef5464e, 0xffbee828, 0xf
fbee820, 0xffbee818) in /opt/mqsi/lib/libCommonServices.so
(0xfee59ca0+0x0118) __1cIImbAbendbPlogSystemErrorAndTerminateProcessInternal6Fpkcki2rkn0AGInsert_66666pkvkb_v_ (0xfef54616, 0x536, 0xfef54
64e, 0xffbee828, 0xffbee820, 0xffbee818) in /opt/mqsi/lib/libCommonServices.so
(0xfeec3b54+0x072c) __1cNImbNamedMutex2t5B6MrknNImbStringBase4CHnDstdLchar_traits4CH__CcibL___v_ (0xffbeec28, 0xffbee870, 0xffbee890, 0xff
bee888, 0xffbee880, 0xfef53a26) in /opt/mqsi/lib/libCommonServices.so
(0x0004c438+0x020) __1cNImbAdminAgent2t6MpnbCImbAdminAgentStartParameters__v_ (0xffbeec28, 0xffbef9c4, 0x137d54, 0x0, 0xeb918, 0x0) in /ap
p/broker/opt/mqsi/bin/bipbroker
(0x0004b028+0x040c) main (0x0, 0xffbefbec, 0x109e94, 0x13b000, 0xffbef9c4, 0xffbef9a4) in /app/broker/opt/mqsi/bin/bipbroker
(0x0002d900+0x0108) _start (0x0, 0x0, 0x0, 0x0, 0x0, 0x0) in /app/broker/opt/mqsi/bin/bipbroker
----------------------------------------------------------------------
#########

After some investigations, it is found that in the directory "/var/mqsi/locks/", these 2 files are created when broker is started
1. mqsi5RetainedPubsTableLockSemaphore
2. mqsiV2ListenerConfigFileMutex
And the first file will be delete when the broker is shutdown normally.

But in abnormal shutdown scenario, the first file is still here. I can re-start the broker after I remove both file.

My questions are:
1. Are there any side effect of removing these files?
2. Are there any normal approach to do when the broker is fail to start after abnormal shutdown.
3. What are the purposes of these files?

Thank you
lwl