MQSeries.net :: View topic - Reason code 2033 even though message seems to exist on queue

westi · Posted: Tue Jun 12, 2007 8:17 am Post subject:

Hi all,

We have a test app that uses JMS API to send a message on a queue and then straight after receives the very same message and logs it.
This is done to reproduce the following problem in our server: From time to time we wait for ever and never receives the message.

We are using:
MQ Java Client 6.0.2.0
MQ Manager V.6

The test application does the following:

1) Create ConnectionFactory using TCP client connection type

2) Create QueueConnection
3) Create QueueSession (not transacted with auto-acknowledge)
4) Create QueueSender
5) Create TextMessage
6) Send message
7) Save msgId for later retrieval

Close QueueSender
9) Close QueueSession
10) Close QueueConnection

11) Create QueueConnection from factory in 1)
12) Create QueueSession (not transacted with auto-acknowledge)
13) Create QueueReceiver
14) Receive message using a filter like "JMSMessageID = '<msgId>'" with 2000 ms. timeout
If timeout expires with no meesage received we loop back to the receive method again
15) Close QueueSender
16) Close QueueSession
17) Close QueueConnection

This should always work - and is quite simple as well: we have one app, one thread, one queue, and the app sequentially sends and then receives one simple text message.

Still, the test fails randomly after having processed X messages.
The one message getting stucked is in fact placed on the queue as it can be browsed by the "MQ Visual Edit" tool (and HermesJMS for that matter)
So it looks like the Receiver never realizes that a message that fits the filter is actually available on the queue.
In the JMS app the receive method returns with null.
Enabling API logging by calling com.ibm.mq.jms.services.ConfigEnvironment.start() reveals that a reason code 2033 is returned from the Queue Manager.

How is this possible?

Thanks in advance.

John