I am using ActiveMQ 5.8.0. I am having mesh of brokers with 4 broker instances (broker configuration is attached) with JDBC persistence. All 4 brokers are configured in same way. I am attaching 3 listeners to each queue and each listener takes around 400ms to process the message I am pushing message after every 20ms. First I bring up all 4 brokers and attaché listeners to queues on all the nodes. Once setup is up, I start producer threads. I can see messages are being consumed properly and are shared between nodes. No I do failover as follows: First I bring down broker 1 (First I shut down producers and consumers and then broker). Then I start it again (First I start broker, then consumers and producers) I repeat same steps for all brokers Now when all brokers are up after failover, I stop all the producers and wait for pending messages to be consumed. In the end I am left with few messages stuck (randomly on any node, 4 to 5 messages) in database table which are not delivered. These messages are not delivered even after waiting for long time. I tried to access queues using JMX and few message browsing tools, but no success. Once I restart broker on which these messages was stuck, they got delivered Is there something wrong with my setup or I am getting same problem as described in https://issues.jboss.org/browse/ENTMQ-444 If Yes, Is similar issue raised with ActiveMQ? Is there any workaround?
-- View this message in context: http://activemq.2283324.n4.nabble.com/Small-number-of-messages-stuck-in-full-mesh-failover-tp4674635.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
