Author |
Message
|
wraymore |
Posted: Tue Dec 01, 2009 5:48 am Post subject: BIP5424S - No Message Id Recieved |
|
|
Centurion
Joined: 16 Aug 2005 Posts: 114 Location: Burlington, NC USA
|
Currently running WMB 6.008 on AIX 5.3.x.
Input to message flow is TDS and I have the Input Message Parsing identified in the MQInput Node.
We did not encounter this error in Development or QA testing. When we moved to production for User Acceptance Testing with 3000 to 5000 messages, we have encountered this error on three occasions. All three times the error has occurred at 2:37 AM EST, which corresponds with approximately half way through the testing. When we move the message from the failure queue to the input queue the message processes correctly.
I have looked for environmental reasons for this occurring but have not been able to identify any processes that might affect the broker executions. |
|
Back to top |
|
 |
sirsi |
Posted: Tue Dec 01, 2009 9:37 am Post subject: |
|
|
Disciple
Joined: 11 Mar 2005 Posts: 177
|
|
Back to top |
|
 |
wraymore |
Posted: Tue Dec 01, 2009 12:27 pm Post subject: |
|
|
Centurion
Joined: 16 Aug 2005 Posts: 114 Location: Burlington, NC USA
|
I read the posting at developerworks, but that reads like the parser fails to load the first time. In my case there are several hundred processing correctly then one fails and then the following several hundred process correctly. And I can replay the message that failed and it processes. I would think that the flow would fail all the time if the value of message type was not correct. |
|
Back to top |
|
 |
kimbert |
Posted: Wed Dec 02, 2009 4:58 am Post subject: |
|
|
 Jedi Council
Joined: 29 Jul 2003 Posts: 5542 Location: Southampton
|
Quote: |
I can replay the message that failed and it processes. I would think that the flow would fail all the time if the value of message type was not correct. |
I agree. This sounds like an intermittent defect - always the worst kind to diagnose.
Quote: |
All three times the error has occurred at 2:37 AM EST, which corresponds with approximately half way through the testing. |
That might be an important clue to finding the problem.
- Does *every* test run fail in this way?
- If not, what is the ratio of successful/unsuccessful runs
- Has it ever failed at any other time, or is it *always* 2:37AM?
- Is the failing message always the same test message? ( I know it always has the same message type - but I am asking whether the problem is related to the content of the message).
- Is the production system identical to the test systems ( same fix packs applied etc )? |
|
Back to top |
|
 |
wraymore |
Posted: Wed Dec 02, 2009 6:16 am Post subject: |
|
|
Centurion
Joined: 16 Aug 2005 Posts: 114 Location: Burlington, NC USA
|
Quote: |
Does *every* test run fail in this way? |
There were 4 test runs of 2500 to 3000 messages this past month and each run had one failure with this error
Quote: |
Has it ever failed at any other time, or is it *always* 2:37AM? |
Not exactly the same time
The dates and time are:
10/21 02:23:24
11/05 02:49:56
11/18 02:37:10
11/19 02:47:17
Quote: |
Is the failing message always the same test message? ( I know it always has the same message type - but I am asking whether the problem is related to the content of the message). |
Messages all have different content
Quote: |
Is the production system identical to the test systems ( same fix packs applied etc )? |
Yes; the Development, QA, and Production systems are all at 6.0.0.8.
A little background:
The data originates as a tagged delimited file that is read into a program that builds a MQ Message and puts to a queue. This queue is the input to the flow that transforms the message from TDS to XML.
The one difference between the dev/QA testing and the User Acceptance Testing is the way the order entry is occurring. In dev and QA testing we used auto-generated order entry, whereas in the User Acceptance Testing order entry is following the normal production process. |
|
Back to top |
|
 |
kimbert |
Posted: Wed Dec 02, 2009 7:56 am Post subject: |
|
|
 Jedi Council
Joined: 29 Jul 2003 Posts: 5542 Location: Southampton
|
Quote: |
In dev and QA testing we used auto-generated order entry, whereas in the User Acceptance Testing order entry is following the normal production process. |
I'm not sure exactly what that means - I don't know what 'normal' is in your businsess. I presume that the dev/QA system is randomizing the message content, whereas the User Acceptance tests are a fixed set of real-world 'order' records? |
|
Back to top |
|
 |
sirsi |
Posted: Wed Dec 02, 2009 8:40 am Post subject: |
|
|
Disciple
Joined: 11 Mar 2005 Posts: 177
|
when the program builds an mq message does it add any header details w.r.t message set ? |
|
Back to top |
|
 |
wraymore |
Posted: Wed Dec 02, 2009 2:30 pm Post subject: |
|
|
Centurion
Joined: 16 Aug 2005 Posts: 114 Location: Burlington, NC USA
|
Quote: |
I'm not sure exactly what that means - I don't know what 'normal' is in your businsess. I presume that the dev/QA system is randomizing the message content, whereas the User Acceptance tests are a fixed set of real-world 'order' records? |
That is correct since they are generated automatically. They also process through the systems at a higher performance rate than production. The auto generation mimics the Order Entry process which releases the orders to the process that generates the tag delimited files (COBOL). When the tag delimited files are written to disc storage, the MQ protocol transformation process (COBOL) has a listener that picks up the files.
This part of the data flow is on a Unisys mainframe.
The transformation from TDS to XML in the MB is on a P-Server running AIX.
Quote: |
when the program builds an mq message does it add any header details w.r.t message set ? |
No there is not a reference to the message set in the message. |
|
Back to top |
|
 |
sirsi |
Posted: Wed Dec 02, 2009 2:45 pm Post subject: |
|
|
Disciple
Joined: 11 Mar 2005 Posts: 177
|
may be a trace, which will capture only header details related to msg set, might help... ofcourse its pain to check for thosands of message's...
otherwise raise an pmr? |
|
Back to top |
|
 |
|