| Author |
Message
|
| vinodsasidharan |
 Posted: Fri May 09, 2003 10:23 pm Post subject: Deadlocks fmcdb : But keep it and reduce side effects |
 |
|
Apprentice
Joined: 25 Apr 2003
Posts: 47
Location: Norwich
|
Dead locks:
=======
1)
get snapshot for database on fmcdb
Deadlocks detected = 11
This is a part of fmcsys.err :
==================
05/09/03 15:52:39 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 15:52:39 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 16:41:11 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 16:41:11 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 16:49:43 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 16:49:43 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 16:49:45 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 16:49:45 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:15:12 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:15:12 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:15:39 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:15:39 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:19:46 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:19:46 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:39:03 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 17:39:03 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
-----------------------------
05/09/03 17:55:41 FMC10100I Administration server starting.
05/09/03 17:55:42 FMC10200I Execution server for system FMCSYS started.
05/09/03 17:55:42 FMC10500I Execution server instance started.
05/09/03 17:55:46 FMC10000I System startup complete. System FMCSYS in system gro
up FMCGRP is now running.
05/09/03 17:56:32 FMC10500I Execution server instance started.
05/09/03 18:20:05 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 18:20:05 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
05/09/03 18:42:07 FMC31100W The message ChgWorkItemDesc could not be processed b
ecause of a database deadlock or timeout. The message will be retried.
05/09/03 18:42:07 FMC31100W The message ChgWorkItemDesc could not be processed b
ecause of a database deadlock or timeout. The message will be retried.
05/09/03 19:09:04 FMC31100W The message CheckoutWorkItem could not be processed
because of a database deadlock or timeout. The message will be retried.
This is a part of the live production system and these dead locks are happening from nearly 1 year . we did not want to make any changes as it was not giving any major problems . but now some times the admin server hangs .so need a resolution .
Also note that the DB2_RR_TO_RS=YES fmcdb . is this the culprit leading to dead locks
so we have started the reorg and re bind of the database .?
1)So what is the best approach .Also
2)i also notices that hold quee has reched some 88 may be it is maximum .is there any problem if i purge these . most of the messages in the hold quee are related to dead locks . so i doubt if this is affecting the server .
plss reply
_________________
Vinod sasidharan
Ibm Certfied MQ Admin 5.3
Ibm Certfied MQ Admin 6.0
Ibm Certfied WAS Admin 6.0
Ibm Certfied WMB Admin 5.0
Ibm Certfied Db2 Specialist.
Sun certified Java Programmer.
"Ai carte, ai parte ....................." |
|
| Back to top |
|
 |
| jmac |
| Posted: Sat May 10, 2003 5:24 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
You need to have a look at
"MQSeries Workflow - Considerations for production roll out" Version 1.0 written by Wolfgang Kulhanek
I downloaded it from
http://www-3.ibm.com/software/ts/mqseries/txppacs/wd02.html
This will cover the deadlock issue.
I have never seen anything on the hold queue. I have been told that one reason that items end up there is because of log files being full (i.e file system out of space).
It sounds to me like you need to do some maintenance.
When was the last time you:
Purged records from the System Log Table?
Purged records from the Error Log Table?
Deleted fmcerr.log?
Deleted fmcsys.log?
Ran fmcsdclad to clean the Audit table?
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| Ratan |
| Posted: Sat May 10, 2003 6:01 pm Post subject: |
|
|
Grand Master
Joined: 18 Jul 2002
Posts: 1245
|
In addition to what John said, If you have not applied the latest support pac, do it.
-Ratan |
|
| Back to top |
|
|
| kevinf2349 |
| Posted: Mon May 12, 2003 6:11 am Post subject: |
|
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
Now this is interesting.
Ratan....on what basis are you advising the latest support pack? Are there some definate 'fixes' for this problem in there? Also what platforms are you referring too?
We sometimes end up with "Expiry Notification" messages on the Hold queue. This is on OS/390 (or to be more accurate z/OS) and we were told by the company that implemented the initial system (and "yes" it was SYSCOM) that this was caused by other processes locking out the database. We adjusted the the number of retry attempts and it reduced the number of these that we got but I am not truly satified that this is the best solution for this site so I am extremely interested in reading more about the support packs.  |
|
| Back to top |
|
|
| jmac |
| Posted: Mon May 12, 2003 6:24 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
Kevin:
There is definitely some "deadlock" improvement noted in SP1 for V3.4. There was also some in several of the 332 service packs.
What I don't know is if they applied to 390 or not.
GOOD LUCK
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
| kevinf2349 |
| Posted: Mon May 12, 2003 7:03 am Post subject: |
|
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
Thanks John
I will do some research to see if it applies to z/OS and post my finding here. |
|
| Back to top |
|
|
|
|
