I am attempting to build a highly available Artemis cluster with zero message loss. Currently I am experimenting with a share nothing architecture, so each broker in the cluster has its own independent store with replication in between them. As far as I can tell when you are running Artemis cluster with replication the broker will acknowledge a message before the message is successfully replicated. The problem that this generates is that if the broker dies before it is replicated, but after the acknowledgement, the message is essentially lost. Is there any plan to add this functionality in order to ensure zero message loss during failover without using a shared store?
-- View this message in context: http://activemq.2283324.n4.nabble.com/Artemis-Replication-before-acknowledgement-tp4712215.html Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.
