| Author |
Message
|
| kevinf2349 |
 Posted: Thu May 13, 2004 12:59 pm Post subject: |
 |
|
Grand Master
Joined: 28 Feb 2003
Posts: 1311
Location: USA
|
CHF,
All I can do is point you at the manual for z?OS and quote our workflow system programmer (I deal more with MQ Series)
| Quote: |
I guess I'm against it, because on z/OS the start and stop parameters are
ignored. To me, this indicates that it is to run 24 hours a day. The
mainframe should be able to handle this load. To change the start and stop
times of the server would require manual intervention and I don't think IBM
would have it set up like that. |
We ended up auotmating the start and stop of the cleanup server through our automation package and have never had a repeat. |
|
| Back to top |
|
 |
| manoj |
| Posted: Thu May 13, 2004 1:18 pm Post subject: |
|
|
Master
Joined: 30 Jan 2002
Posts: 237
Location: Virgina
|
You can specify when you want to run the clean up server in the FDL....
MQ Workflow will automatically start and stop the clean up server
_________________
-manoj |
|
| Back to top |
|
|
| keithng |
| Posted: Thu May 13, 2004 6:09 pm Post subject: |
|
|
Novice
Joined: 27 Apr 2004
Posts: 12
|
CHF,
It is pity to hear that you have the same problem.
At this current moment, after 1 week struggling, we finally drill down to only 2 problems,
1. when there is deadlock happen, all other mqwf messages will have error and throw message retry limit exceeds.
2. when there is retry limit exceeds, it mention there should be message in hold queue of execution server, but i find nothing.
Now my deadlock problem becomes 1 per 2-3 days. Since i have done 2 things in additional to rebind and reorg.
1. increase locklist
2. make the recyclethreshold of execution server to be 1000 transaction.
I am currently discussing with IBM, will give u guys update.
Regards,
Keith |
|
| Back to top |
|
|
| manoj |
| Posted: Thu May 13, 2004 7:16 pm Post subject: |
|
|
Master
Joined: 30 Jan 2002
Posts: 237
Location: Virgina
|
1000 may be too small. when i put 1000 i saw too many server stop and start messages in a time frame of 5 minutes. i had to change it to 30000...again it depends on your workflow application
_________________
-manoj |
|
| Back to top |
|
|
| keithng |
| Posted: Thu May 13, 2004 7:18 pm Post subject: |
|
|
Novice
Joined: 27 Apr 2004
Posts: 12
|
| yep, 1000 is very small...but u know, when u find the server become stable...u will not try to modify any configuration..... |
|
| Back to top |
|
|
| CHF |
| Posted: Fri May 14, 2004 5:06 am Post subject: |
|
|
Master
Joined: 16 Dec 2003
Posts: 297
|
Thanks for sharing your ideas guys. We contacted IBM regarding this. I will let you know if I got any fixes for this problem.
CHF  |
|
| Back to top |
|
|
| CHF |
| Posted: Mon May 17, 2004 4:43 am Post subject: |
|
|
Master
Joined: 16 Dec 2003
Posts: 297
|
Keith,
Where do I change these parameters?
1. increase locklist
2. make the recyclethreshold of execution server to be 1000 transaction.
Can you elaborate on the general purpose of these parameters?
Thanks
CHF |
|
| Back to top |
|
|
| hos |
| Posted: Mon May 24, 2004 12:23 am Post subject: |
|
|
Chevalier
Joined: 03 Feb 2002
Posts: 470
|
Hi,
just to clarify one - minor - side aspect of this problem:
Athough the error message indicates that the failing message will be put onto the hold queue this is not always the case. If the retry limit of a client request message exceeds is will be just thrown away. It does not make sense to store this request and replay it later asynchronously from the client. The client is anyway aware that a problem occurred as he enfaces a timeout.
So only server messages are put onto the hold queue. |
|
| Back to top |
|
|
| CHF |
| Posted: Mon Jun 07, 2004 6:27 am Post subject: Update |
|
|
Master
Joined: 16 Dec 2003
Posts: 297
|
Hello all,
Applying Workflow SP5 solved our problem. There are no more database deadlocks.
FYI:
My environment is: WF 3.4 SP5 on OS/390
Thanks
CHF  |
|
| Back to top |
|
|
| blane99 |
| Posted: Fri Jun 11, 2004 5:24 am Post subject: make RTMessageRetryLimit and Message Retry Limit the same |
|
|
Apprentice
Joined: 12 Jun 2002
Posts: 41
|
Here's a problem we had with SDDS messages and messages going to holdq re: MQwf 3.4 SP5
Problem description
When RTMessageRetryLimit is increased to a value above the default of 5, then in error situations (more than 5 retries), multiple XML error messages are sent to the REPLY2Q, i.e. a value of 6 for the parameter yields 2 error responses, a value of 10 yields 6, and so on (RTMessageRetryLimit - MESSAGE_RETRY_LIMIT + 1).
Cause of the problem
There is a second parameter for controlling message retry and error handling: the FDL parameter MESSAGE_RETRY_LIMIT, controlled via Buildtime option "Domain / Messages / Message Retry Limit".
Workaround / Solution of the problem
There is not, and will not be a fix for this problem for Workflow 3.4 or lower. For these releases, both parameters have to be kept in Synch manually by following the description below.
Attention: Never advise a customer to modify RTMessageRetryLimit without also setting MESSAGE_RETRY_LIMIT to the same value also! |
|
| Back to top |
|
|
| praveenchhangani |
| Posted: Fri Jun 11, 2004 5:46 am Post subject: |
|
|
Disciple
Joined: 20 Sep 2002
Posts: 192
Location: Chicago, IL
|
Good to know Blane. Thank you.
_________________
Praveen K. Chhangani,
IBM Certified Solutions Designer -
MQ Workflow 3.4. |
|
| Back to top |
|
|
| CHF |
| Posted: Sun Mar 06, 2005 7:54 am Post subject: |
|
|
Master
Joined: 16 Dec 2003
Posts: 297
|
Regarding runstats......
just wondering if the workflow should be brought down while we do runstats/rebinds/reorg on the workflow DB.
I am of the opinion that WF needs to be down only while rebinding the executables from fmcztuil. It is also mentioned in the WD02 supportpac about this.
IBM recommends running the runstats/rebinds on a nightly basis. Not sure if we need to bounce WF everynight.
_________________
CHF |
|
| Back to top |
|
|
| jmac |
| Posted: Sun Mar 06, 2005 11:37 am Post subject: |
|
|
Jedi Knight
Joined: 27 Jun 2001
Posts: 3081
Location: EmeriCon, LLC
|
CHF:
I hold the same opinion as you. WF must be down for the rebind, but not for the runstats. This means if you rebind every day, you need to bring down WF everyday.
_________________
John McDonald
RETIRED |
|
| Back to top |
|
|
|
|
