MQSeries.net :: View topic

MikeClark · Posted: Thu Feb 24, 2005 11:35 am Post subject:

I did as suggested and checked on the Status of the queue (the triggered, not the initiation queue), and it showed my application that I use to send test messages to the queue as an input source. This seems to suggest it isn't closing the connection after doing the put, but I've followed the app in my debugger and it is definitely performing a close.

I was able to get a trigger to run anyway, by setting triggering to EVERY on that queue, then sending (putting) a test in. Alarmingly, the trigger fired, the application ran successfully, and did what it was supposed to. One message remained on the queue -- wouldn't trigger, and I can't "clear" the queue because of the Open Input count. The app that put the message in there is long closed and gone. I ran a getter to pull the message off the queue, but the Open Input count remains at 1.

I cannot get the Open Input count to 0.

This has been driving me nuts over the past few months. From time to time I've been getting the "Not a valid trigger message" error, especially (although not always) after having made a change to the application that is set to be started up, with no change to the underlying local queue and process definition. Other times this has happened when the mainframe programmer (the other side to at least part of this mess, because that's usually where test data is coming from) has made some change. I had hoped to find some consistent cause to it all. The crazy thing is, at some point the problem goes away -- the mainframer, who is a contractor with extensive MQ experience, claims not to have done anything.

For JT, I did check the name of the Destination Queue in the DLQ header, and it was the name of the queue as shown defined in my first post. No other trigger monitors were running, and I'm not sure what message from the DLQ I should post. The "data"? Well, here are the first few bytes: