Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations IamaSherpa on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Microsoft DNS / Unix Clients 1

Status
Not open for further replies.

spamly

MIS
Apr 1, 2004
447
US
I've got an interesting issue that has now bit me twice.

My organization has three Microsoft DNS servers (Windows Server 2003 SP1). Normally, they work just as advertised and we have no issues from clients.

Unfortunately, we've experienced a scenario that is impacting our UNIX clients (~200). One DNS server will be in a limbo state. By this I mean that the service is up, it is accepting connections, but not responding with any information. Our Microsoft servers and clients accessing this particular server don't miss a beat. They automtically skip this one and move on to the next DNS server. If they manually lookup a DNS entry from that specific server, they simply timeout after 2 seconds.

The UNIX servers (AIX, Solaris, RedHat, probably others) all hang for about 10 seconds before moving on to the next DNS server.

As this problem is only affecting UNIX clients, our Microsoft DNS admins don't really see it as an issue. Further, they believe that it is a "UNIX" problem and I should deal with it.

The particular DNS server that has the issues requires a reboot to bring the DNS service back online.

You can see my problem. I've got a variety of UNIX flavors that are all experiencing the same symptom. Microsoft clients (all versions?) seem to be working fine.

Has anyone out there seen this before? I'd love to identify a Microsoft patch for the DNS server, but the Microsoft support guys tell me that there isn't one available.

Alas, google hasn't been able to show me the way either.

I could set some timeouts for all of our UNIX clients, but this would be time consuming and I don't believe it addresses the root cause of the problem.
 
Seen it, been there.

Not a bug on the Microsoft side, I believe that their client implementation probably selects the "best" one, and invalidates the "bad" one, so after the first timeout it never asks that DNS server.

I am sure you have read the resolv.conf man page, one can only twiddle the retry count, timeout and set the round robin.

What I usually do is setup a Unix/bind cache server that is forwarding to the Windows DNS.

eugene

 
Thanks for the response. We're bringing in an appliance that uses bind to resolve this (They were already coming in, but several months down the line). A caching DNS server is a failback plan if the appliances don't work.
 

As far as your Windows Admin thinking a DNS server should only be up optionally, let them know that resolv.conf, bind, etc. have been around longer than the D in DNS.

I've had bind servers (with the correct log trimming and memory tuning) that have run for years, until we updated them.
eugene
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top