We are having some issues with our 1

[joner]

We are having some issues with our set-up now it is deployed over a firewall
and I was wondering if anybody could help. The constituent parts are
several objects running under MTS on an app server, a DB on a DB server,
web pages on a web server. These are all in a contained within a domain,
effectively the DMZ. Between this domain and the back office domain there
is a firewall using NAT. Within the back office domain are Win32 clients,
using the objects running under MTS on the app server in the DMZ. The
firewall is set-up to allow traffic through which instantiated through the
back office domain, there is no NAT between the back office and the DMZ but
if one of the DMZ machines attempts to talk to the back office machines, NAT
is employed. Our problem is that after six minutes of inactivity between a
Win32 client and the MTS objects we start getting RPC server unavailable
errors. I have read about DCOM servers pinging their clients to check if
they have not been torn down but from the documents I have found it is not
clear if the client pings the server or vice versa. If the client is
pinging the server I cannot see a problem with our set up but if the true is
opposite, maybe the documents I have found stating that NAt cannot work with
DCOM may come into play. The way I read these documents is that the IP
address of the server is wrappered up with the call from the client to the
server and so any NAT would cause issues.
Could anybody shed any light on this matter?
Thanks.
Paul ...

 

[jmarler]

From what I've heard, DCOM will not work with firewalls using NAT. You may want to look at SOAP and XML to send your data/calls through Port 80 on the firewall. - Jeff Marler

http://www.JeffMarler.com

(please note, that the page is under construction)

 

[joner]

My own take on the whole 'DCOM does not work with NAT topic' was that it works if you are going from one side of the firewall to the other where you are not using NAT i.e. from private IP adrsses to public ones. I thought (maybe erroneously) that DCOM and NAT did not work if you were going from public to private and therefore using NAT.
Do you have any information as to how 'DCOM does not work with NAT' manifiests itself, I mean does it stop any client connecting to a server or is it more subtle than that?
Paul ...

 

[jmarler]

joner,
From MSDN . . .


When a COM client on a Windows NT computer runs under an identity that cannot be authenticated on the remote computer, a COM server started by the client shuts down in approximately six minutes.

http://support.microsoft.com/support/kb/articles/Q175/0/20.ASP

If your client and server computers are connected through the Internet with firewalls and proxies between them, DCOM does not work if there is any type of Address Translation (NAT) in between them. If there is no address translation, you need to configure these proxies and firewalls to enable DCOM to communicate. You can find several white papers related to this subject on the Microsoft Developers Network (MSDN) or at the following Microsoft Web site:

http://www.microsoft.com/com/tech/dcom.asp

- Jeff Marler

http://www.JeffMarler.com

(please note, that the page is under construction)