Network adapter reset on Exchange server

[PainOfDeath]

I have a very weird problem on a server running Exchange 2003 that just cropped up a day ago. I set the server up a couple weeks ago and it had been working fine until then.

The server has two network interfaces on two different subnets, 192.168.136.x and 192.168.130.x., each configured with different gateway routers. The .136 subnet is where the DC resides and it has a Contivity router with no port forwarding/NAT configured. The .130 subnet has an IPCop firewall which handles forwarding for SMTP, POP3, and SSL (no HTTP) to the Exchange server. The IPCop also has a VPN tunnel to a remote office on a different subnet, where there is a second DC for the domain (the Contivity is also tunneled to the same office). The NIC on the .130 subnet is configured to use each DC as its primary and secondary DNS servers (local and remote respectively).

What happens is a few times a day the server will stop responding on the .130 NIC - for example using OWA. However if I ping that NIC internally it responds. If I disable and re-enable that NIC within Windows, it starts working again. One time when it stopped working I tried to telnet port 25 on the .130 NIC and it worked, so I thought the IPCop may not be forwarding traffic properly, but then it forwards traffic to another internal machine just fine. Also, why would it start working again just by disabling and re-enabling the NIC? I am very puzzled. I already tried re-booting EVERYTHING with no luck (router, server, and switch).

Any help would be greatly appreciated - its driving me and my users crazy!

 

[Zelandakh]

This isn't an Exchange problem, it is a routing problem I think caused by the multiple routes. I can't get my head around it immediately but I'd say you should move to Exchange on a single IP and use a router for each route.