I submitted https://issues.apache.org/jira/browse/AMQ-3380 this bug with a unit test that demonstrates what I've observed. Unfortunately I cannot use a shared master/slave as the system I'm making this for does not have access to shared disk.
It is really clear that I'm using the system in an unexpected way since the "failure" is consistent. Some questions: Why does the chosen consumer in an exclusive consumer setup get switched when the master fails? Shouldn't an exclusive consumer's transaction get rolled back if it is no longer the consumer? If not, is there a way that a consumer can get notified that it is no longer the exclusive choice? Thanks -- View this message in context: http://activemq.2283324.n4.nabble.com/best-place-to-start-to-write-a-master-slave-exclusive-consumer-failover-test-tp3628964p3630698.html Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.
