I have a system that has two brokers that use a shared file system for persistance.
For purposes of my explaining my issue, consider that Broker1 came up as the master and Broker2 is the slave. If Broker1 goes down, the applications (producer and consumer) failover to Broker2 just fine. Broker2 now asks as the master. However, I lose some messages. If I later bring Broker1 back online (as a slave) and kill Broker2 (so that Broker1 becomes master again), the missing messages are suddenly processed. I don't understand why this happens. Since Broker1 and Broker2 share a persistance location, shouldn't broker2 have all the message when Broker1 crashes? Liz -- View this message in context: http://www.nabble.com/Failover-config-tf4577069s2354.html#a13065448 Sent from the ActiveMQ - User mailing list archive at Nabble.com.
