One of the (Windows XP SP3) workstations of our Windows Domain could not access internet anymore, this problem suddenly happened overnight. The domain controllers (there are three of them) are all running Windows Server 2008.
First I compared the output of ipconfig /all on the faulty workstation with the output of a working workstation and it was just fine as it had always been. In particular the default gateway was correct and always remained pingable from the faulty workstation.
I guessed that something was wrong with the DHCP service and I restarted the DHCP server service on all of our three DCs as well as the DHCP client service on the faulty workstation. This didn't solve the issue.
I then proceeded to switch the TCP/IP configuration for that machine manually to another free valid IP address (192.168.0.41)... and then the internet access came back!
I then cleared any traces of the previous IP in the DHCP leases list and in the DNS tables of our DCs and, after setting back the TCP/IP configuration to 'automatic', finally, the new lease would be granted (192.168.0.41) alongside with the internet access.
My question: what went suddenly wrong with the original IP address?
It seems that the DHCP leases got messed up because of a very unfortunate manipulation from one of our junior staff.
We share some of our LAN's internet via a wireless router/firewall by plugging our LAN via Ethernet onto the WAN interface of the router. The staff in question had mistakenly plugged our LAN cable onto the LAN interface of the router turning it into a wireless switch. I then realized that our DNS tables where polluted with a lot more names than usual and - our DHCP range being quite small (192.168.0.25 to 99) - BAD_ADDRESS registration was given precisely to the faulty IP (the one under which internet access was lost.)
It is not yet clear 100% what exactly was going on but presumably some duplicate IPs were generated at some point.