> Hi, just a sidenote: deploying two MDB ataching to the same destination
> has unspecifyed behaviour if a remember the EJB 2.0 spec correct - so we
> are spec compliant ;-)
The specs (EJB + JMS) are broken, in more ways than one.
Lack of specification in this area is good and bad. Good because it does
not limit how vendors can architect there system, bad because each vendor
will behave differently. The only safe usage of JMS/MDB in this respect is
one queue, one consumer, n+1 clients (which might be satisfactory in some
cases, but lets not stop there).
> But it would be really good if this could be fixed.
I hope so, otherwise using MDB in a highly distributed environment is almost
worthless. If there are n+1 machines each with a MDB attached to a queue
(on a remote router) and one of them sucks up all the messages, then a super
system would have to be put in place to manage queues for each host,
copying from master queue to host queue, managing the resources on both ends
to make intelligent decisions about when to move messages and so on...
which would be much easier to implement in the provider itself.
Perhaps JMS is not the best solution for robust distributed message passing
in Java? I am already frustrated with the highly redudant API (I think
there are about 5 methods that differ between queues and topics). Perhaps a
JINI/JMS based system would provide a work around to some of these issues...
--jason
_______________________________________________
Jboss-development mailing list
[EMAIL PROTECTED]
http://lists.sourceforge.net/lists/listinfo/jboss-development
