Random DNS Client Issue with BIND9/Windows Server 2003 DNS

Question

Within our office, we have a local server running DNS, for internal related "domains", (e.g. .internal, .office, .lan, .vpn, etc.). Randomly, only the hosts configured with those extensions will stop resolving on the Windows-based workstations. Sometimes it'll work for a couple weeks without issue on one machine, then suddenly stop working, or it'll happen on another 15 times per day. It's completely random for all workstations.

When troubleshooting, I have opened up a command prompt, and issued various nslookup commands for some of these hosts, and they resolve, however I've been told that nslookup uses different "libraries" for name resolution than other applications such as web browsers, email clients, etc.

The only solution thus far, is manually restarting the Windows DNS Client on each workstation when this happens. Issuing the ipconfig /flushdns command multiple times helps every now and then, but is not successful enough to even attempt before restarting the DNS Client.

I have tried two different DNS servers; BIND9, and Windows Server 2003 R2 DNS, and the behavior is the same.

We have a single Netgear JGS524 switch all workstations and servers are connected to within the office, and a Linksys SR224G switch in another department with workstations attached.

Answer 1

In this particular situation, it appears that Windows will randomly start using a secondary name server instead of the primary, even if the primary is available.

My solution: remove the secondary. This is not a great solution as it obviously will kill the whole name resolution if this single name server goes down, but given this network is small and name resolution isn't mission critical (read: it can go down for an hour), this solution is acceptable.