BM3.6 on NW5.1sp7
BM36c02 applied
Using Proxy.nlm from BM37sp3.
Using Craig's proxy.cfg
Ran tuneup.ncf from Craig's site.
Running HTTP and DNS proxies only.

Our BorderManager server to use 2 DNS servers at our ISP to resolve
hostnames. On the Proxy console, we see the status of those servers go
Up-->Down and back again throughout the day, and our users receive a lot of
"DNS server unreachable" errors in their browsers. We've configured the DNS
calls to go out over UDP with a timeout of 4 minutes. The DNS UDP replies
to timeouts ratio is about 1:3.

It's not a connectivity issue - during the period when both servers show as
"DOWN", we can still ping them from the Bordermanager server.

The problem seems to be getting worse and we're getting very frustrated.
Any help would be much appreciated.