This seems to be really tricky. My latest tests indicates that the
problem is not in the "sending" end, but in the receiving end. I can for
example duplicate the error when client connect, sends a ping and the
whole connection process end up beeing locked while the serverside
callback server tries to do the initial ping back.
I have trace instructions on all the Corba parts, and there really is no
answer (no notify runaway conditions), the client callback server simple
locks the reply.
However, on the client, I can not find any locks that seems to be the
result of the ping callback.
So either there is a lock problem in some other part of Jacorb, either
at the sending end or the replying end, or there is some strange lockup
in XmlBlaster which translates to these lockups.
Simply put. I do not think the problem is in ReplyPlaceholder or
ReplyReceiver, but I can not understand where else to look. I have a
testclass which can be used to triggers these lockups with the
standalone XmlBlaster, and which dumps (kill -3) when it detects a lock.