[omniORB] No connection to Naming Service from omniORB client

[Duncan Grisby]

On Tuesday 11 March, Christoph Thomalla wrote:

[...]
> Is there a way to see what actually happens?

This is showing us what actually happens. The problem can be seen
in this (slightly edited for clarity) fragment...

> omniORB: LocateRequest to remote: 
> key<0xafabcaff0000000262f589f100000008000000000000000001>
> omniORB: Client opened connection to giop:tcp:10.1.1.55:3302
> omniORB: sendChunk: to giop:tcp:10.1.1.55:3302 45 bytes
> 4749 4f50 0101 0103 2100 0000 0200 0000 GIOP....!.......
> 1900 0000 afab caff 0000 0002 62f5 89f1 ............b...
> 0000 0008 0000 0000 0000 0000 01        .............
> omniORB: inputMessage: from giop:tcp:10.1.1.55:3302 20 bytes
> 4749 4f50 0100 0004 0000 0008 0000 0002 GIOP............
> 0000 0001                               ....
> omniORB: From endpoint: giop:tcp:10.1.1.55:3302. Detected GIOP 1.1 
> protocol error in input message. Connection is closed.

omniORB sends a GIOP 1.1 LocateRequest message. It did that because
the IOR received from the naming service bootstrap agent claimed to
support GIOP 1.1. The server responds with a GIOP 1.0 LocateReply.
That is not permitted, so omniORB was right to reject it.

I don't think the JDK 1.3 ORB supports anything above GIOP 1.0, so it
was wrong to issue an IOR with a GIOP 1.1 profile in it. The
work-around it to limit omniORB to GIOP 1.0, in which case it will use
GIOP 1.0 to talk to the object, regardless of the version it claims to
support. Do that with -ORBmaxGIOPVersion 1.0.

Cheers,

Duncan.

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