Network Drive Timeouts

I am in a windows 2000 server enviorment with windows 2000 workstations. When users try to connect to their mapped drives after a while of inactivity they get an error message saying drive is not accessible, but if they continue to click the drive eventually it will open up. Does anyone know how to fix this?

 

Are you running at least SP3 on all servers & workstations?

Are all your servers running the same SP?

 

No. Both the servers and workstations are at least at SP2 but some are SP3 and SP4.

 

Your Domain Controllers absolutely need to be all at the same SP level.

I'd strongly suggest all servers be up to at least SP3 and especially any that have users mapped to them.

Also take a look Here and Here.

 

I will try it. Thanks.

 

Make sure speed and duplex settings are hard code to your switch settings. i.e 100MB / FD.
You can use netmon to capture a network trace of the problem, and if you see duplicate packets (based on sequence number), then you have a duplexing / speed problem.

 

Interesting Bursley. Would you give the 'why' on dups and if it happens with NIC & switch port both on auto (yuck) or just the PC NIC on auto (yuck again).

I've seen speed issues with either of the 'yuck' settings and always hard code these days but was never really sure what the problem was. Sorta figured maybe the two ends just enjoyed negotiating when they shoulda been working.

 

I am looking at the network monitor, but the problem only happens once in a blue moon. I guess I will keep the network monitor running and then have the person tell me when they get the error. Then I can back track the log and look for dup frames. Is that how you would do it?

 

Have you already matched up the speed/duplex settings?

 

From a basic understanding of this problem, certain nics don't auto negotiate very well with certain switches / hubs. In theory as we all know, speed should be negotiated properly. The problem here is the human factor of trusting computers to always be right when we humans invented them. So human error in programming of the chipset or drivers / OS cause nics to renegotiate a speed that has already been determined. During this time period network packets are sent.
In the TCP if that packet is not received in full and checked for consistency, it will ask for a retransmission, and the sequence number of the packet will be the same.
Using UDP if a packet is not received in full, a higher level protocol in the application layer of the OSI will need to determine the problem and ask for the information again. This will not be seen as a retransmission.

I have also seen problems with switches, where the switch fabric is causing a problem between 2 switch ports talking together. For example. My switch at home, ports 1 and 3 can ping and talk UDP w/o to much trouble, but if I start using TCP for a file tranfer for example, I get network timeouts. However ports 1 and 3 talk with port 2 just fine and can tranfer files w/o a problem. Hard setting my switch ports and nic don't matter. In my case, I just need a better switch.

 

Appreciate the information.

A part of the initial setup of any server here is to set switch port and NIC to the same settings. Always. But we've just been doing it because it seems to work better and never had a good idea of the cause of problems if either or both ends were allowed to float.