Just to add to this, the above only applies to when the cache state is not STARTED or STARTING. For STARTING, if the call is remotely originating, I block for up to StateRetrievalTimeout millis, constantly trying again to see if the cache has entered the STARTED state. If even after StateRetrievalTimeout millis this still hasn't happened, I barf with an exception since just logging in this case can result in data inconsistency. I'd encourage you to check out the latest from SVN and try it out. View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4086240#... Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&a...