[borisha]

This looks very confusing. Most likely the problem is that it is not explained correctly in documentation. Maybe someone can point us to documentation that would make it clear how to use this and what was the reason of implementing it this way.

Now it looks like:

1) there is JGroups cluster between JBoss instances in JBoss cluster, so I guess every JBoss knows about all other instances inside same cluster.

2) Previous point is not enough for EJB clustering to work. On top of that every JBoss instance with EJB client trying to access remote EJB needs to define one or more outbound connections to few other JBoss instances. And to make things worse this is not done per JBoss instance but per application.

3) Again, previous point is not enough for EJB clustering to work. Also every application using stateless session beans MUST reference 'cluster' which has alias SFSB?! <cache-container name="ejb" aliases="sfsb sfsb-cache" default-cache="repl" module="org.jboss.as.clustering.ejb3.infinispan">. Why do I as SLSB must know anything about SFSB cluster? Even if it is used under the hood why do I have to reference it from my application when I do not use it?!

It would be very beneficial if all this was explained somewhere - so that it is clear what are benefits of constantly repeating very similar (or same) information.

thanks