First i will explain my use case so that you can get a general sense of what i'm trying to do: 1/ Use case I have a clustered JBoss 5.1.0 ear application (with default JBM so 1.4?). A Swing application is connected to both nodes and receives JMS notifications via a ClusteredXAConnectionFactory. When one of of the nodes fails, the other detects the change in topology and posts a "LostNode" message on a topic which the client listens to. 2/ Problem The failover sometimes works, sometimes not (maybe depending on which node the connection was last pointing to?) 3/ Things i'm confused about: * The FailoverOnNodeLeave attribute in oracle-persistence-service.xml. From what i read it forces same failover on Ctrl-C as on brutal kill-9 or crash? * Do i have to do anything in the client (interceptor, exception listener,...)? I'm getting mixed messages, so is the failover transparent or not? If it's not, what do i need to do (the connection is injected via Spring if it makes any difference) * Does specifying ClusteredXAConnectionFactory/ClusteredConnectionFactory/others in the client make any difference? I don't see what XA or not would change for a remote client but i could be horribly wrong? Any examples of a transparently reconnecting client? I see only xml files in the JBoss messaging examples directory? View the original post : http://www.jboss.org/index.html?module=bb&op=viewtopic&p=4237704#... Reply to the post : http://www.jboss.org/index.html?module=bb&op=posting&mode=reply&a...