Hi,

I've encountered an error in my environment that I want to reproduce in a
activemq test environment.  I am running 5.4.2 with a pure master/slave,
persistence=true setup.  In my environment I have multiple consumers on a
queue each with the exclusive.consumer=true flag set (consumerA and
consumerB).  These consumers are created from a transacted session.  What I
notice in my tests of killing the master:

* test starts, consumerA gets the "lock on the queue", receiving fine in a
transacted session
* producer will send ~100 messages, the last message will cause a consumer
to commit session
* consumerB is in a receiveNoWait loop (it checks some other things sleeps
and comes back)
* master killed around message 30
* The exclusive consumer is switched from from one to another (say from
consumerA to consumerB)
* consumerB will start getting messages from message 1, but some window of
messages around the kill are not replayed

I assume this is not expected behavior and would like to reproduce in the
ActiveMQ test environment.  Can someone point me to a few tests that I
should draw from to produce this test?  I've found some master/slave tests
and some exclusive consumer tests, but not any exclusive consumer transacted
tests or other combinations.

Thanks,
Martin

--
View this message in context: 
http://activemq.2283324.n4.nabble.com/best-place-to-start-to-write-a-master-slave-exclusive-consumer-failover-test-tp3628964p3628964.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.

Reply via email to