"dimitris(a)jboss.org" wrote : Can't find the discussion thread, sorry. | | Again, my point is why not printing a warning and choosing a temporary id? | | Well if it was that simple, we'd have probably done it already? The problem is that the id is used to identify the nodes permanent storage - so it cannot be transient, it has to persist between restarts. Also it's an integer, and we'd have to make sure it's unique across all nodes in the cluster. Doing that automatically is non trivial - you need some kind of HA singleton counter that dishes out IDs (yuck). Alternatives would be to make it a GUID, but it's used as a key in the message data tables so this would have performance implications. Also making it a GUID would probably p*off some sys admins since an integer id makes it easy to identify a particular server (server 1, server 2, server 3 etc), whereas a GUID does not. As Aaron pointed out, passing in a sys prop is a good idea. Brian Stansberry also suggested we should have some kind of concept of server id at the AS level - then we could just use that. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4125477#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...