[omniORB] Multiple Interfaces

[Sean Kessler]

I would first like to thank the supporters of this group
   for being so prompt, and precise, with regard to answering my   
previous questions.

   As I described in a previous mail, I have two interfaces
   on my Linux machine, ppp0 and eth0.  The ppp0 interface 
   is, of course, my connection to the net.  For arguments
   sake I'll say it's (209.139.139.80).  Behind the eth0
   interface is an NT machine, say 192.168.1.5.  I have been
   able to run both omniNames and my test servant on the 
   Linux machine and have an internet based client connect,
   obtain a reference to the servant, and invoke methods.

   I am trying to take this to the next step which is to relocate
   the servant to the NT machine, have it register itself with 
   omniNames (on the Linux box), and have an internet based client
   be able to obtain a reference to the servant, and invoke
   methods on the NT based servant.

   I run omniNames on the Linux box as follows...

./omniNames -ORBtraceLevel 25 -ORBInitialHost 209.139.139.80
-ORBInitialPort 900 -BOAiiop_name_port 209.139.139.80

  The servant on the NT machine gets initialized like this...

 fundsrv -ORBInitialHost 209.139.139.80 -BOAiiop_name_port
209.139.139.80

  The servant is able to bind it's name to the naming service
  without any problem.  The client is able to contact the 
  naming service and obtain a reference to the servant.
  However, when the client attempt to invoke a method, a
  COMM_FAILURE exception is generated (in the client), 
  scavenger on the servant doesn't event twitch.

  Am I going about this the wrong way?
  Any ideas or suggestions would be greatly appreciated.



  -Sean Kessler