We've been running a Remote Access Server for VPN clients at our medium sized office. It was running on a small workstation sized server running Win2K server. Yesterday the power supply literally exploded and killed the motherboard. At the time, it also spiked the ethernet connection and our router and other servers lost ethernet connectivity for a split second.
Instead of rebuild the motherboard, I took another server also running Win2K server, which wasn't in production and enabled the RAS services. I enabled the second NIC interface on the machine, set it up for the same IP as the original machine (to get it through the firewall) and put in a range of IP addresses for it to assign manually (not DHCP).
Easy setup, seemed to work great at first; but as our users started dialing back in; about 1 in 5 connections seems to be bogus. They get a connection, can resolve IP addresses, but cannot ping any of the servers they need to access. Doing a tracert fails at the VPN's internal interface. These clients can dial back in, and usually get a 'good' connection, but its luck of the draw at that point.
'External' IP ends in .199 (the IP clients connect to, and open on the firewall)
'Internal' IP ends in .190
External IP got assigned .150 (I don't see anyway to configure this really)
When a tracert is done from either side, it ends at the .150 address.