OK, this may be a little lenghty but it's a strange problem so I want to be
as complete as possible. I have multiple schools connected to a central
office via T1s. All schools have local DHCP servers and point to two DNS
servers at the central office for name resolution. At one school, name
resolution seems to stop working randomly throughout the day. When it stops
the users at the site can't access any resources at the central office via
NCP, RDP connections drop, websites become inaccessible, etc. The server
starts getting "Unable to communicate with server" errors for the servers
in its replica ring and SLP gets deactivated. It sounds like IP has stopped
routing or the WAN link has dropped, however, the workstations and the
server are still able to ping servers at the central office and on the
Internet by IP - just not by name. The WAN link also consistently tests
error free.

There doesn't seem to be any pattern to this problem - the console.log
indicates that the connections are even lost in the middle of the night
when nobody is onsite. Sometimes it reconnects in a few minutes and other
times it takes a half hour.

Any ideas?