[omniORB] omniORB/omniNames with multiple network interfaces

[Duncan Grisby]

On Wednesday 22 August, Chris Sandu wrote:

> The omniNames name service is started by the company's product and I
> am not necessarily allowed to play with it on a remote server.
> 
> This server has two network interfaces 10.0.0.0/255.255.0.0, which is
> inaccessible to my machine and 10.255.11.0/255.255.255.0 which is the
> network through which I access this server. Now, I am using omniORB
> 4.0.7(I can't use the newest version) and I run my server on the
> remote machine and the client on my machine using the name service on
> the remote machine. Now, for some reason the objects published on the
> server ahve the 10.0.0.0 network address which is inaccessible to my
> computer and therefore I can't invoke any methods on the object. I've
> tried the -ORBendPoint giop:tcp:10.205.11.112: option on the server;
> although the server seems to publish them correctly(I traced this via
> -ORBtraceLevel 40), my client still gets an object with 10.0.0.1
> address. What am I doing wrong? Keep in mind that I can't really touch
> the name service. It seems like a simple issue, and yet I can't make
> it work. Also, I am not at liberty to add a route to the 10.0.0.0
> network; it's just not a solution for a client, for example. Help?

omniNames doesn't do anything to modify the object references you store
in it. It's a very simple thing that just stores object references as it
receives them, and passes them on to clients. Configuring omniNames
itself to publish the correct IP address has no bearing on the
addresses in the object references stored by omniNames. Have you
configured all the servers that register themselves in omniNames to have
the appropriate address?

Cheers,

Duncan.

-- 
 -- Duncan Grisby         --
  -- duncan at grisby.org     --
   -- http://www.grisby.org --