Hi Andy, Thanks for the pointers. Based on your suggestions I have been able to send and receive 1000 messages on my Windows XP box without any problems! I tried to push this number upwards but it breaks at about 3000 messages - same "Connection failure has been detected" message. Anyway, I am not worried about that right now since I am not expecting those kinds of loads on my system. Moreover, it may just be that I am running out of steam on my development laptop. Bottom line is that my MDB configuration has always been correct. As you indicated, since I am running in my local server, I can just use the default INVM settings and there is no need for netty. So I have reverted to the simple MDB configuration I previously had: | @MessageDriven(activationConfig={ | @ActivationConfigProperty(propertyName="destinationType", propertyValue="javax.jms.Queue"), | @ActivationConfigProperty(propertyName="destination", propertyValue="queue/jmstest") | }) | public class TestMessageListener implements MessageListener { | ... | } | The problem really was on the sender side. As Tim pointed out, because I was using "/ConnectionFactory", I was really not going through the JCA adapter. As soon as I changed that to "java:/JmsXA", messages started flowing smoothly. So I am a happy camper now :-). Couple of quick followup questions: 1) When I monitor my queue from JBoss jmx-console I see very low message counts, typically less that 50. I know that all my 1000 messages have been pumped into the queue because the call to send them has returned. I also know that only a handful have been received by watching the logs. So why does the console not show a big number in the queue? 2) In the stand-alone server use case, where is the connection factory located - on the JBM server or on client (i.e. a standalone application or servlet/EJB running on an application server)? I am assuming that it is on the client and you might even provide an implementation that manages a pool of connections just like the JCA adapter on the application server. Is this assumption correct? Finally, I have couple of suggestions that would make the documentation a lot clearer - at least for me :-). I like the two diagrams in sections 3.3 and 3.4 - they give a nice high level overview of the two use cases - standalone and embedded. I suggest showing an external client in the standalone case to make it crystal clear from where the services are accessed. Next, I would use these diagrams as a base for explaining deeper concepts. Overlay them with connection factories to show where they are located and how many of them are there. Especially useful would be to show what we get "out-of-the-box". For example, I had no idea that we get the JmsXA factory out-of-the-box and it is a JCA factory that I can start using right away. The documentation leads me to believe that I got to do a #~it-load of configuration before I can get JCA working - which is absolutely not true. Finally, it would be very helpful to use the diagrams to show which configuration file controls which piece. There are so many configuration files flying out there that it is confusing for a beginner. You guys have set up very good defaults so that a new user could get started in just 10 minutes, but it has taken so much energy for so many people to get a simple use case going. I guess if you create a 10 minute getting started guide for two use cases (embedded and stand-alone), that would go a long way! (The existing Quick Start Guide is very mechanical, it gets the examples running but leaves the reader without any context around what they did. Also, I was having problems getting the jca-config example to run because of couple of issues - I could provide details separately.) I hope this is feedback constructive. Thanks again for all your help. Naresh View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4242971#... Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&a...