[omniORB] giopstream::CommFailure

[Duncan Grisby]

On Thursday 28 June, "=?UTF-8?Q?Jo=C3=A3o_Lopes?=" wrote:

> AFAIK you have to restart the client ORB every time the server dies.
> The reason for this is that the client make a TCP connection to the
> server and when the server comes up again it cannot recover the old
> TCP connection from the client.

That's not true. If the server comes up with the same endpoint
information, the client can reconnect with a new TCP connection. In some
cases of unclean server shutdown, the client is unable to tell that the
connection has broken, so the first call it tries to do fails with a
COMM_FAILURE, but subsequence calls will open a new connection and
succeed.

You can use a COMM_FAILURE exception handler to automatically retry the
call once, rather than having to handle it specifically on all calls.
See section 4.7 of the omniORB manual.

Cheers,

Duncan.

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