To briefly go through the cases:
1) A client running outside of JBoss AS. You need to code the retry mechanism that's
laid out in detail on the wiki page.
2) A non-MDB client running inside JBoss AS. Use java:JmsXA.
3) An MDB deployed normally, i.e. not deploy-hasingleton. When the HA-JMS server fails
over, there is no deploy/undeploy going on with the MDB. The existing MDBs will
automatically keep trying to reacquire the HA-JMS server, and when it starts on the new
master, they'll find it. This should all be transparent to other JMS clients, other
than there may be a temporary delay in processing messages.
4) An MDB deployed in deploy-hasingleton. Here there will be a deployment of the MDB on
the new master. Against, this should be transparent to other JMS clients. Note that
there is no requirement whatsoever to deploy MDBs in deploy-hasingleton in order to have
HA-JMS work properly. You'd only deploy them in deploy-hasingleton if for some reason
you wanted them only running on one server in the cluster.
View the original post :
http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4092271#...
Reply to the post :
http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...