| |
|
RSS Feed - WebSphere MQ Support
|
RSS Feed - Message Broker Support
|
  |
|
| problems with message exit |
« View previous topic :: View next topic » |
| Author |
Message
|
| PeteM |
 Posted: Mon Sep 25, 2006 5:24 am Post subject: problems with message exit |
 |
|
Newbie
Joined: 20 Apr 2006
Posts: 6
Location: UK
|
Hi all,
We have a message exit which does this;
message in format BBBBTRANCODE where "BBBB" is four blank bytes.
Exit establishes the length of the message data (buffer length minus MQXQH_LENGTH) and uses that to build LLZZ for IMS bridge which it overlays onto 'BBBB' - message length not altered.
Has run successfully for many months on 6 different queue managers processing millions of messages.
Last Monday on one system we had one message where the length put in the message was incorrect - right format just wrong value. On Tuesday 2 messages, wednesday 11 messages. Refreshed the CHIN task and problem went away (note there are hundres of thousands of messages through this exit each day). There were no changes.
If we take the messages and send them back in again the exit works fine so there is nothing wrong with the message.
This Monday one message on a different system.
PMR raised with IBM but requires tracing which has a big overhead given we don't know when/where it will hit also we are tight on ECSA on the system to increase the trace tables sufficiently and as TRAXTBL needs a refresh of the CHIN we may well fix the problem in the act of switching on the trace!
Basically what I am saying is.......Help!
_________________
Peter Moir,
Bank of America (UK) |
|
| Back to top |
|
 |
| rtsujimoto |
| Posted: Mon Sep 25, 2006 9:49 am Post subject: |
|
|
Centurion
Joined: 16 Jun 2004
Posts: 119
Location: Lake Success, NY
|
| Is it possible the message length exceeds LL? |
|
| Back to top |
|
|
| Nigelg |
| Posted: Tue Sep 26, 2006 12:07 am Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
Debug your code until you find the error.
_________________
MQSeries.net helps those who help themselves.. |
|
| Back to top |
|
|
| PeteM |
| Posted: Tue Sep 26, 2006 1:46 am Post subject: |
|
|
Newbie
Joined: 20 Apr 2006
Posts: 6
Location: UK
|
Hi,
thanks for teh replies...
not a length problem we check for that, also length of the rejected message is always fine.
As for debugging the code that is exactly what I have tried to do however when I take the message that failed and run it through the exit 10,000x and it works every time I'm at a bit of a loss as to how to debug it!
I believe this is a re-entrancy problem now and I'm reviewing the code for anything that would prevent it being properly reentrant.
thanks,
Pete
_________________
Peter Moir,
Bank of America (UK) |
|
| Back to top |
|
|
| tleichen |
| Posted: Tue Sep 26, 2006 8:04 am Post subject: Re: problems with message exit |
|
|
Yatiri
Joined: 11 Apr 2005
Posts: 663
Location: Center of the USA
|
| PeteM wrote: |
Hi all,
Last Monday on one system we had one message where the length put in the message was incorrect - right format just wrong value. On Tuesday 2 messages, wednesday 11 messages. Refreshed the CHIN task and problem went away (note there are hundres of thousands of messages through this exit each day). There were no changes.
.
...
not a length problem we check for that, also length of the rejected message is always fine.
|
Can you be more explicit when you say the length put in the message was the wrong value? Your statements seem to be in conflict. First you said the value was wrong, then you said it is not a length problem. 
Also, if you are dealing with reentrancy, this could be a task sequencing problem. 
_________________
IBM Certified MQSeries Specialist
IBM Certified MQSeries Developer |
|
| Back to top |
|
|
| PeteM |
| Posted: Tue Sep 26, 2006 8:20 am Post subject: |
|
|
Newbie
Joined: 20 Apr 2006
Posts: 6
Location: UK
|
Sorry, what I mean is that the value that is being put into LLZZ (1st 4 bytes of message) by the exit is wrong - that is to say it should be the length of the application message but isn't.
However there is nothing wrong with the actual message and its not related to the genuine length of actual the message i.e. its not too long to fit into 'LL' - which I think was what Rick was suggesting might be the cause.
For example - actual message length is 800 bytes, LLZZ edited by the exit says its 600 bytes or such (there's no pattern, its been over and under by varying amounts). IMS will reject the message when it reaches IMS.
If we put the exact same message back in to MQ the exit now fills out LLZZ correctly.
The error occurs very rarely and very randomly.
All that said I have now found a fundamental error in the exit which would prevent it being properly reentrant which showed up when I re-assembled it with the RENT checking option. Why we haven't had the problem until recently I don't know, perhaps just luck.
_________________
Peter Moir,
Bank of America (UK) |
|
| Back to top |
|
|
| tleichen |
| Posted: Tue Sep 26, 2006 6:55 pm Post subject: |
|
|
Yatiri
Joined: 11 Apr 2005
Posts: 663
Location: Center of the USA
|
Not sure how, but it really sounds like your exit, under certain strange conditions, is somehow using a length from one message and putting it in another.
_________________
IBM Certified MQSeries Specialist
IBM Certified MQSeries Developer |
|
| Back to top |
|
|
|
|
|
| |
|
Page 1 of 1 |
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
|
|
