I am building a very basic web service that can either push text
messages onto a queue, or conversely consume messages from this same
queue.  ActiveMQ 5.0.0 is running standalone and is using the tcp://
transport and Kaha persistence.  (If it matters, I'm using JBoss as my
app server)

For my web service, I'm using a servlet, which using the doPost
method, builds up a very basic JMS producer client, sends the message,
and then tears the client down.  On the consumer side, it builds up a
very basic JMS consumer, consumes the message and then tears the
client down.

For testing purposes, I've written some HttpClient message producers
and consumers.  I'm producing/consuming 60kb messages as fast as they
will go.

It would seem that I can send messages to the queue all day long, but
after a random number of messages consumed, the consumers stop
consuming.

At this point, I have to stop ActiveMQ, restart, and then the
consumers will finish consuming the messages.  I've tried turning
debug on in ActiveMQ to try and get a hint as to what is going on, but
I don't see anything abnormal.

I'm thinking my design is probably pretty inefficient, and I was
curious if I was some how exhausting a resource somewhere by creating
and destroying so many JMS clients in such a short amount of time?

Is there anything in AMQ that I can switch on to get more debug info?
I've already tweaked the log4j settings.

thanks in advance

-- brian

Reply via email to