| Author |
Message
|
| WillH |
 Posted: Thu Jun 30, 2005 11:21 am Post subject: Triggered application runs but intermittently sends to DLQ |
 |
|
Novice
Joined: 15 Jun 2005
Posts: 23
|
I am testing triggered processes with runmqtrm on AIX (MQ 5.3). I have a simple process defined to execute a perl script. The script prints the current time to STDOUT and a file.
I am using rfhutlc to write an empty message to the queue, and I am watching the trigger monitor output. I see the timestamp appear in the log, and I see the temp file with the timestamp in it. This works every time. But sometimes (actually a lot of the time), trigger monitor reports "Error starting triggered application.", and sends the trigger message to DLQ with the Reason "MQFB_APPL_CANNOT_BE_STARTED".
It doesn't seem to be related to the frequency I trigger the process. I can do it every second or every 5 seconds, and some will give that error.
The triggered app executes fine every time, and that's great. But I don't want my DLQ to fill up with these erroneous error messages. The Application Programmer's Reference does not go into detail about how trigger monitor decides the application could not be started.
I have tried exitting the script with 0, 1, -1, same results.
The Application Type is Unix (though actually I don't see this in the trigger header, that field appears to be empty).
The Application Identifier is /tmp/trigtest.pl. I also tried "exec /tmp/trigtest.pl".
There is no user data or env data.
Any ideas? |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Thu Jun 30, 2005 11:23 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Try setting the app identifier to '/tmp/trigtest.pl&'.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| EddieA |
| Posted: Thu Jun 30, 2005 11:28 am Post subject: |
|
|
Jedi
Joined: 28 Jun 2001
Posts: 2453
Location: Los Angeles
|
What's the Trigger type. Is there a concurrency issue, that a 2nd copy of your script can't run while the 1st is still active.
Cheers,
_________________
Eddie Atherton
IBM Certified Solution Developer - WebSphere Message Broker V6.1
IBM Certified Solution Developer - WebSphere Message Broker V7.0 |
|
| Back to top |
|
|
| WillH |
| Posted: Thu Jun 30, 2005 11:31 am Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 23
|
I got the same results with these variations of your suggestion:
These spit out an extra thing in trigger monitor log, ': not found'.
Setting appId to /tmp/trigtest.pl&
Setting appId to /tmp/trigtest.pl &
Setting appId to nohup /tmp/trigtest.pl &
These had the same results as my original post:
Setting appId to /tmp/trigtest.pl and userData to &
Settign appId to /tmp/trigtest.pl and envData to & |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Thu Jun 30, 2005 11:42 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
Can you change your runmqtrm execution to pipe it's output to a file?
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| WillH |
| Posted: Thu Jun 30, 2005 11:55 am Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 23
|
| jefflowrey wrote: |
| Can you change your runmqtrm execution to pipe it's output to a file? |
It does this already. The command I execute is
| Code: |
| nohup $MQBINPATH/runmqtrm -m QMWILL -q SYSTEM.DEFAULT.INITIATION.QUEUE >> $MQOUTPATH/SYSTEM.DEFAULT.INITIATION.QUEUE.log & |
Would you like to see some sample output? |
|
| Back to top |
|
|
| WillH |
| Posted: Thu Jun 30, 2005 12:12 pm Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 23
|
| EddieA wrote: |
| What's the Trigger type. Is there a concurrency issue, that a 2nd copy of your script can't run while the 1st is still active. |
The trigtype is Every. The script completes sub-second, but regardless, I am waiting 1-5 seconds with no difference. |
|
| Back to top |
|
|
| WillH |
| Posted: Thu Jun 30, 2005 1:57 pm Post subject: Solution |
|
|
Novice
Joined: 15 Jun 2005
Posts: 23
|
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Jun 30, 2005 2:12 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
You could just as well have found it using the search button on this site...
 |
|
| Back to top |
|
|
| WillH |
| Posted: Thu Jun 30, 2005 3:07 pm Post subject: |
|
|
Novice
Joined: 15 Jun 2005
Posts: 23
|
| Believe me, I tried. Hopefully this thread will show up in Search for the next person stumped by this problem. |
|
| Back to top |
|
|
| Nigelg |
| Posted: Fri Jul 01, 2005 12:54 am Post subject: |
|
|
Grand Master
Joined: 02 Aug 2004
Posts: 1046
|
This is not a bug in WMQ, but in AIX, if you read the fix report properly. Th WQ workround is to circumvent the AIX bug by using a different signal mechanism. Naturally AIX have denied all knowledge, but then they always do. Have you ever got a bug fix from AIX without threats?
Anyway, the WMQ fix referred to in the above link is for 5.2 CSD06, not 5.3 CSD07.
This saga continues with a defect introduced in 5.3 GA, which caused the env var not to be recognised. This was fixed in another APAR, in 5.3 CSD04.
However, the task of setting the env var in the right place and at the right time has proved far too complex for most system administrators, so a permanent fix has been done and will be issued in CSD11. |
|
| Back to top |
|
|
|
|
