Internal NT 4.0 DNS resolution fails with W2K clients

[TxJas]

We have a NT 4.0 DNS server used for internal resolutions only. ISP DNS servers are setup as secondary for resolving public domains. For some reason W2K clients will stop resolving internal host names at what seems like random times. NT 4.0 clients do not have this problem. Refreshing W2K client DNS parameters re-enables internal resolutions, but eventually the problem returns.

The internal domain we use is also one of our public domains. For instance, our registered domain of company.com with our ISP is also setup on our internal DNS server as company.com. I realize this make no sense, but I inherited the setup. I've thought this might be the problem, but why wouldn't the NT 4.0 clients behave the same if it were?

We also have a WINS setup on another NT 4.0 server. DNS is setup to resolve host names through WINS and DNS submits domain as NETBIOS scope.

Can anyone tell me what happening?

Environment:

Server: NT 4.0 SP6
Client: Windows 2000 SP3

DNS Parameters used on clientonly used options are listed)

-DNS servers in order of use:
--Private address (NT 4.0 server)
--Public address (ISP)
--Public address (ISP)

-For resolution of unqualified names:
--Append primary and connection specific DNS suffixes
--DNS suffix for this connection: company.com

 

[djhawthorn]

Have you tried debugging on the client using nslookup? MCSE NT4/W2K

 

[TxJas]

To a degree, yes. I'm sorry to say that I'm not very familiar with nslookup. I didn't think to use this tool when the problem occurred, but I did check to see if DNS responses were as expected.

However, I think I may have solved the riddle on my own. It appears this is due to a bad server config and a difference in the way W2K behaves verses NT 4.0 clients.

After doing research on DNS order of operations in MS, I found that a failed query would force the default DNS address to the bottom of the name server list. It would also remain there until subsequent DNS queries failed with name server addresses on the top of the list, and so on. So, I created a secondary internal DNS server and setup both the primary and secondary to use forwarders (ISP name servers). Then I changed the default and alternate addresses on my workstation to query only the internal servers. So far this has worked. However I still don't understand why W2K would handle this differently than NT 4.0?

I will post to this again after a few more days if the new config still appears to have solved the problem. Any insight to this problem would be appreciated.

 

[ykwong]

TxJas, in my company I had a similar problem with what you described. I must say that your postings helped me a lot to find the light of a dark tunnel. But I am also interested in the part you mentioned about how the DNS order change when a query failed. May I know where you found that? I have searched many books, I still cannot find any solid description of that mechanism.

Thanks in advance.

 

[TxJas]

I found the information in Microsoft's Technet. I'm sorry to say that I do not remember the article number. I would try to sear on, "DNS search order" or something similar at MS' Technet site or Knowledge Base.

I had previously promised to repost if my changes fixed the problem permanently.

If anyone was wondering, DNS queries still work perfectly with the posted changes.

 

[ykwong]

Yes, my LAN is also behaving well since I applied the fix.