We have setup an SBS 2008 setup. VPN protocols are passed-through from the internet to the local IP of the server (via NAT/port forwarding). We tested VPN and it works well, so we opened it for all employees.
Since then we experienced extreme network problems, usually during off-time when the employees are at home. The server becomes unreachable from the network, ping stops working, logging in at the local console hangs forever. If I'm already logged in while this problem occurs, I can use most system function but shutdown (as an example) won't work. From all I could see the system does not actually freeze, the shutdown/login animation still works.
We then disconnected internet, and everything worked. This brought my attention to the fact that it may be related to VPN logins. So I scanned the logs. While I cannot find any clues about successful VPN logins, I found a few other clues: Local services listening on the loopback network (not the loopback interface thou, read: not 127.0.0.1) can no longer be reached the very time when the remote access service tries to get an IP from the DHCP server.
Different services like DHCP server, DCOM, DNS and others a suddenly no longer reachable although these run locally on the SBS server.
I suspect the following problem: Something in the routing tables gets messed up, trying to route the local IP of the server (192.168.1.1) through the tunnel. Coincidently, 192.168.1.1 is the default IP of most popular DSL routers. I suppose this is the root cause of the problem, although I currently have no idea how exactly the problem starts during VPN login.
Any idea how to block such faulty configurations from establashing a tunnel? And why the heck does Windows messes up its own IP configuration just for a bloody VPN login?
