If a delegate service broker is restarted, the aggregate client doesnt attempt to recreate a listener to a temp reply queue. This may cause a hang or a timeout in the aggregate client. In this scenario there are two brokers: one managing aggregate service queue and a second that manages delegate service temp reply queue. When the latter is restarted, the dispatch code does not attempt to create a new listener. The aggregate code should detect a failure of the broker, mark a delegate as FAILED and kill the listener. On subsequent CAS request, the aggregate should attempt to create a new listener and a new JMS Connection to a broker. Failure to re-establish the connection is subject to error handling as defined in deployment descriptor Process error configuration for the delegate.