This morning someone had a server with a strange problem. Its a Windows NT 4 terminal server.
I could launch a term session, logon and ping the server’s netbios name, but we could not ping any IP address from within the server. Its own IP could be pinged, but I could not ping its gateway nor any other IP addresses out there. Eventlogs seem to be in order with nothing usual.
So I asked networks to check that gateway or if any ACL was applied to the networks and they said it was okay, which is correct as I could TS to the server. Its puzzles me for a while as I try to remember if its possible to have any IP filtering policies in NT that could prevent this… not that I am aware of.
But something the network guys said peaked my interest. The server was configured with 2 IP addresses in the same subnet, one on each NIC. They told me that one of the IP addresses was not pingable. I quickly double checked the eventlogs and sure enough there was a NIC error earlier on. I fired up the HP network dialog and it tells me that NIC was disconnected.
Apparently, the server as using the faulty NIC as the gateway in its route table and is routing all TCP/IP traffic out from it. As a result, network connections out of the box does not work, but connections into the box is okay because the other NIC is still working.
Disabled the faulty NIC and rebooting the server resolves my problem.