VisiBroker CORBA using 99% CPU and increasing infinite memory usage

Question

I've read most of the CORBA posts on this newsgroup and found that no one else has had this problem, but I tried the other solutions just in case. None of them worked.
In my swing app running in JWS, I have a thread that initializes a VisiBroker ORB. Somewhere along the initialization line, my CPU usage for the JVM controlling my app jumps up to 99%, memory increases infinitely. When i finally killed the app, memory allocated was way over 160 megs. This does not happen with this application outside of JWS. I then used a memory profiler and found that the method call that was causing all of this was in some a class of VisiBroker's, the com.inprise.vbroker.ds.DSUser$DatagramThread in particular its run method.
If i decompile this class gasp, I actually find that there is a possibility of an infinite loop. Here is the code for the method:
        public void run()
            do
                try
                    do
                        byte abyte0[] = new byte[8192];
                        DatagramPacket datagrampacket = new DatagramPacket(abyte0, abyte0.length);
                        _socket.receive(datagrampacket);
                        IStream istream = new IStream(datagrampacket);
                        DSResource dsresource = istream.read_Resource();
                        InetAddress inetaddress = datagrampacket.getAddress();
                        int i = datagrampacket.getPort();
                        synchronized(_handlers)
                            Enumeration enumeration = _handlers.elements();
                            if(debug)
                                System.out.println("Received message: " + dsresource);
                            for(; enumeration.hasMoreElements(); ((MessageHandler)enumeration.nextElement()).handle(dsresource, inetaddress, i));
                    } while(true);
                catch(Exception _ex) { }
            while(!_shutdownInProgress);
        }Now using the information i get from the memory profiler, the actual place where the problem is is _socket.receive(datagrampacket); . This method performs the java.net.PlainDatagramSocketImpl.peek method as well, but basically you get the point. Using the memory profiler I also find that this causes a SocketException, which leads to an IOException.
With all this being said, I tried a few things from the newsgroup. I tried making sure my JNLP file specified to use 1.3 because of the classloader issues. This didn't work. I also tried putting code into the thread that does this and perform the setContextClassLoader(getClass().getClassLoader()); so that if it were the classloader issues I've been reading aobut it would take care of that. I have my ORB properties set up in the code and as I said before, this application works perfectly fine outside of JWS. I have ALL needed jar files referenced in my JNLP, it just comes to when it gets to that one line of code. Does anyone out there have any suggestions?

Answer 1

No one responded to this post but I came across the solution that I will share in case anyone else needs this in the future. Although I think it's a bug in the VisiBroker code that shouldn't allow this, the problem seemed to be with OSAgent. Upon the recommendation from someone else, I started OSAgent before running JWS and everything worked appropriately. Not familiar enough with VisiBroker to know why this settled the memory leak and infinite loop issue, but it did.