So I have a windows 2000 server and it's hanging randomly at least once a day. I'm unsure what is wrong with it other than the problem is not hardware.
I have quite literally transplanted the hard drives and raid controller from the server's original configuration (1GHZ PIII) to a new machine (2x550mhz PIII Xeon). No change to my lockups. I also ran memtest 86, nothing wrong.
My problem seems to be that at higher CPU utilization (on either of the boxes tested) the computer will drop all network connections (remote desktop, local FTP, it won't even let me ping it from another computer on the network). The strange thing is that if I'm physically at the server itself, it will still go to the internet, but nothing else can reach it (making it useless).
I have SP4 and the latest updates, and I read that remote desktop can sometimes cause this problem so I removed terminal services and installed tightVNC instead, still have the same problem. The only thing I think is left is my Active directory or DNS settings may be way off.
DHCP is handled by my router (pfsense distro), and I have set the router's dns relay to forward all requests for home.local to the IP address of my server. My server is set up to control the domain home.local and I set up DNS on the server to forward requests it cannot resolve to my router (so the server can still reach the internet). All DNS not related to home.local is passed from computers on the network directly to the router (not through the server).
The only other mentionable configuration is that I'm running Active directory for the domain in question with a couple of users, and that I am running IIS to allow for a FTP server on the network (the computer kept dropping connections before IIS and the FTP were installed.
I'm not sure what to do, but this is getting really frustrating. Is it possible that I have a problem with my DNS configuration? (after I set it up with the wizard, I'm lost with all of these settings).
Any ideas would be greatly appreciated and thanks for your time reading this long winded post

Maybe?...

http://support.microsoft.com/defaul...

Sounds more like a failing switch. What is everything plugged into?

Imagine the power of knowing how to internet search
http://www.lib.berkeley.edu/Teachin...

Now that you mention it 813130 was one of the KB articles I read, but the hotfix wasn't a quick and easy download, so I just removed terminal services completely from the system for the time being until I found out if it was the problem (figured it can't hurt me if it's not there). Even with terminal services out of the picture it still does this dance it does...

As for failing switch I hadn't thought to look into it, but my first thought was maybe the NIC card was acting up, so I put a different one in, no change.

I'm pretty confident it isn't a switch problem though. Switching is handled by a Dell powerconnect smart switch (gigabit, don't remember the model number exactly). Not the fanciest switch on the planet, but pretty reliable and it has never failed yet to pass traffic except to the machine in question.

The only other switch on the network is a D-Link 5 port gigabit switch on my desk (downstream from the Dell), but it too allows me to ping all but the machine in question, and I can even ping that machine and get to shared files/etc before it decides to stop cooperating. As soon as that happens it's a restart and it works again... for as little as 30 mins to an hour all the way up to as long as a day, but never more (I think).

It was a wild guess, but the only thing I wasn't sure if I got right was the DNS (up two posts). The only other thing I could think of is if there was some sort of bug in 2k server that hasn't been addressed by SP4 or the update rollup.

DNS has nothing to do with pings not working. If you can't ping the server by ip you can't get to DNS anyway.

Your description doesn't fit any bugs I am aware of.

What does event viewer logs say during these periods of outages?

What are the server specs? [mem, cpu..]

If a smart [managed] switch you should be able to go in and look a the port stats for the port the server connects to. Any errors?

When you are at the server "if I'm physically at the server itself, it will still go to the internet, but nothing else can reach it " can you ping a workstation?

Imagine the power of knowing how to internet search
http://www.lib.berkeley.edu/Teachin...

Actually I don't use the switch in managed mode as it's pretty much worthless, but the port is okay.

The logs were interesting though. I had never set the server to sync with an external time server (apparently I needed to do that when it is the PDC for a forest?), and the windows time service was shutting down periodically.

I set it to sync with us.pool.ntp.org. Time will tell if this is the only problem I had... I'm not 100 percent sure how NTP works, but if I'm not mistaken, both machines need to be closely in tune with each other there (even if they are in different time zones) or you get nothing (just like I was getting).

Things seem to be working again without any restarts.

I'll post again if this didn't fix the problem...

Well.. that tears that..

Less than minute after I post and it's not working again.

Okay it's definately a time problem, the problem now is.. how do i fix it??

I ran net time /setsntp:"0.time.windows.com" and then w32tm /s

(It's eaisier than setting all the other boxes to go to ntp.org)

It seems to hold for about 5-10 minutes, then goes to crap... with lots more of the same two errors, first is a warning that your server is a PDC.... etc etc sync with an external time server, and then an error that says the time service has stopped

As of about 30 minutes later now, the errors and warnings have stopped, but connections still get dropped.

Running the command w32tm /s fixes the problem for a while, then it goes haywire again....