On Jan 12, 2011, at 7:43 AM, Jeff Rose wrote: > On Wed, Jan 12, 2011 at 12:29 AM, Alex Dean <a...@crackpot.org> wrote: > >> >> On Jan 11, 2011, at 8:14 AM, Jeff Rose wrote: >> >>> On Mon, Jan 10, 2011 at 4:57 PM, Alex Dean <a...@crackpot.org> wrote: >>> >>>> >>>> On Jan 10, 2011, at 5:02 AM, Jeff Rose wrote: >>>> >>> >>> The last issue I do see is that fairly often the consumer stops listening >> to >>> the topic even though the ajax poll requests are still coming. I can see >> in >>> jconsole that the consumer count on my topic drops to zero, even though >> at >>> the same time I'm watching empty ajax responses coming back periodically. >> >> Hm. I've never seen that happen. >> >> Can you enable debug logging for activemq-web? I've been adding >> log4j.logger.org.apache.activemq.web=DEBUG >> to config/log4j.properties. >> >> When a new ajax client connects, you should see a line like: >> DEBUG | creating new AjaxWebClient in >> p8v92prp07trmhghrpltw0rh-1294788329515 >> >> If you close the browser window and wait a minute, you should then see >> something like: >> DEBUG | Cleaning up expired web clients. >> DEBUG | AjaxWebClient p8v92prp07trmhghrpltw0rh-1294788329515 last accessed >> 95 seconds ago. >> DEBUG | Removing expired AjaxWebClient >> p8v92prp07trmhghrpltw0rh-1294788329515 >> >> If you see a message like that 'Removing expired AjaxWebClient' while your >> browser is still actively polling, that's wrong. Not sure what else might >> cause the consumer count to drop back to 0. >> >> alex > > > Below is a dump of the relevant log messages. Starting from the initial > subscription when the first AJAX request is made. Then showing the 10 > messages sent by the test php script, which were correctly sent up to the > browser. But after this first message batch arrived the consumer count in > jconsole dropped to zero and no more messages made it. You can see that the > GET long-poll requests keep coming, and it isn't expiring the client because > it seems to be active. I'm not sure why the consumer is dropped though. I > wonder if there could be an exception firing somewhere in the message > processing that could result in this behavior? > > -Jeff >
Hi Jeff. I haven't been able to reproduce that behavior. Does this happen every time, or intermittently? What do you see with a stock 5.4.2 build? This is a shot in the dark, but I notice some log messages like > DEBUG | [2] N-Messages: 1 ... which I don't think are coming from ActiveMQ's trunk code or from my patch. Is it possible you've got other changes in the build you're testing? alex
