Hi Julien, this is actually expected behavior. With a shared file system, only one broker can be active and the second will have to be "passive" until the lock on the files has been removed. This can happen when the first broker crashes or normally stops. By the way, this is exactly the same behavior with replicated LevelDB.
I believe the topology you may want to look at is the network of brokers which can have multiple active brokers. However, this is not a pure HA solution since every persistent message that still sits in the message store of a broker that "crashed" will not get delivered until that broker gets started again. To improve such a setup, you may want to configure a network of brokers consisting of two HA cluster (two active and two passive). -- View this message in context: http://activemq.2283324.n4.nabble.com/ActiveMQ-shared-file-Shared-File-System-hangs-server-startups-tp4686579p4686584.html Sent from the ActiveMQ - User mailing list archive at Nabble.com.
