Losing Connection To PDC

Our backup resides on a machine running 2003 Server. The PDC for the domain is running NT 4.0. The issue I am running into is that they will see each other for maybe a minute then the connection drops for maybe 10 or 15 minutes, then it reestablishes itself for another minute or so. It isn't the physical connection because each machine sees every other machine on the network regardless of operating system. Has anyone ever seen this? I'm worried because if I can't resolve this then I can't back up the information on the PDC. Any help would be much appreciated.

Thanks!

 

Other than using your WINS server addresses, are you set like the picture on your 2K3 box?

 

Yes, it's set up exactly like that. It's really odd. It's almost as if it sees it every 10 or 15 minutes then something stops them from seeing one another. It wouldn't be so frusterating, except both machines can see and ping any other machine on the network.

 

OK. Thanks.

The timing makes me strongly suspect an issue caused somehow by browser update traffic. The time duration between Domain Master Browser and a Master Browser update is 12 minutes by default. Network announcements to keep this list current start immediately after boot followed by announcements at intervals of 2, 4, 8, and finally every 12 minutes.

I'm not exactly sure what is happening but 2003 is not really designed to rely on NetBIOS traffic for anything and more or less must be forced to play nice on a classic NT4 domain.

Try turning off browser service on the 2K3 server and see if that helps. Your PDC functions best as the Domain Master Browser and if you have any BDCs, they do well as backup browsers.

 

The browser service is turned off on the 2K3 machine. That is what is so perplexing about it.

 

Found out what the problem was. The PDC was not the Master Browser. Nothing was. I changed that setting to make it the master and rebooted. It's working great now. Thanks for setting me down the right path!

 

That would do it all right.

Glad you found the problem and got it corrected.

 

Just curious, Were there any clues in the event log?