with 5.2 right? that sounds plausible as failover is the default transport and the default inactivity timeout is 30 seconds. I guess is is possible that the producerAck gets lost, I don't think there is a recovery cache on the broker sending side. Do you see the failover transport doing a reconnect? Does it take more than 30 seconds for the consumer to consume the messages and release the memory.
If you can produce a failing junit test case please raise a jira issue and attach it. 2008/12/9 Danielius Jurna <[EMAIL PROTECTED]>: > > I have one very slow consumer and activemq is configured for throttling. The > problem is that my producer deadlocks on producerWindow.waitForSpace(). All > my queues goes down to 0, but producer still waits for space. > My ideas is that if connection to broker is lost, usageMutex.norifyAll() is > never called. And it seems that failover connection closes connection on > inactivity. > Can any AMQ developer comment on how procuder throttling behaves on > reconnect to broker? > -- > View this message in context: > http://www.nabble.com/Deadlock-on-producerWindow.waitForSpace%28%29-tp20913298p20913298.html > Sent from the ActiveMQ - User mailing list archive at Nabble.com. > > -- http://blog.garytully.com Open Source SOA http://FUSESource.com
