[omniORB] LocateRequest gets dropped on RHEL 6.1 system

[Duncan Grisby]

On Wed, 2011-09-07 at 12:06 -0500, Brad Fawcett wrote:

> I am seeing a problem with running my code using CORBA (omniORB -
> 4.1.4) on a RHEL 6.1 system. 
> 
> Problem scenario is this:
> 
> 1.) invoke the CORBA request  
> 2.) this causes a LOCATE_REQUEST to be sent out
> 3.) TCP ACK is sent back
> Note:  looking through the CORBA server traces.  this LOCATE_REQUEST
> is never received there.   However, a suspicious looking
>                                   one is received in its place.

Can you send complete client and server traces from traceLevel 40
traceTime 1 traceThreadId 1 ?  That will help show what omniORB thinks
is happening. Associated tcpdump output would be helpful too. It looks
as though this is a client and server on the same machine, is it?

Unpicking the message received on the server:

4749 4f50
GIOP

0102
version 1.2

01
little endian

00
Request message

0501 0000
261 octets in message

1200 0000
Request id 18

0000 0000
SyncScope NONE (i.e. oneway call)

0000
KeyAddr

0000
padding

0e00 0000

14 octet object key

feea 9166 4e00 0037 5300

10 octets of the object key



So that looks like the start of a valid oneway request message that
wasn't completed. It's definitely not the LocateRequest sent by the
client.

Cheers,

Duncan.

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