activemq replicated leveldb cluster hangs upon exception in client's message handler
This post was updated on .
Following issue looks like a potential bug of critical nature. I am able to reproduce it easily.
I have the following setup:
Versions: activemq-5.9.1, zookeeper-3.3.6
3 nodes of activemq in cluster (using replicated leveldb)
I'm using the following sample application i wrote using camel: https://gist.github.com/imdhmd/86fb59190706f82d5fe4 This sample application has both the consumer and the producer. When the application is run, producer randomly adds a number (1-9) into a particular Q. Consumer reads from this Q and adds it to the value in a accumulator. The application shows the current value of the accumulator.
Consumer also throws a RuntimeException whenever the current value of accumulator is divisible by 3.
Issue: Whenever an exception occurs in Consumer's processor, activemq broker hangs indefinitely. It is able to accept connections, but it is not able receive messages on any Qs. If you send a message to it the broker in this state, it will not respond.
This occurs only when activemq is in replicated leveldb cluster. In standalone setup, one retry of the failed message happens and then the message goes into DLQ, and broker resumes functioning normally.
Is this a known issue, am i missing some configuration or have i discovered a potentially critical bug??