[omniORB] Tracing options - Rel. 3.0.2

[Ivanov, Roumen]

	Hi,
	omniORB release 3.0.2 on HP 11.0

	I'm experiencing problems with a heavily multithreaded server with
many (multithreaded) clients. Every client has heartbeat thread and possibly
sends some messages using another thread.
	The entire structure is a kind of notification service with clients.

	When the number of clients reaches a particular number (about 123)
no more clients could connect.
	Even method _non_existent() invoked on the server's IOR returns
true.
	Although, the server and all, that are already connected keep
working happily.

	Sounds like resource problem, but our administrators already
increased many thresholds and we are running out of ideas.

	Could you suggest what ORBtraceLevel should we use in order to trace
the maximum of internal invocations?
	The exchanged data is not interesting.

	Thanks in anticipation,
	Roumen Ivanov

	P.S. Any suggestions about kernel configuration parameters that
could positively affect the process would be very appreciated.