Domain Issues and W2K Professional...

We got a big Thunderstorm Monday that knocked out power. As far as I can tell, nothing of ours was hit, but the resultant spike and sag from a lightning strike next door caused some problem (fried a Wireless bridge among other things) for us. Most of them are solved, but one I have run into that I can't fix is really frustrating me:

My Win2K professional machine lost power, and now won't join or log in to the domain. When I attempt to log in, it tells me the domain is unavailable. Look at the Event Viewer on the server (NT 4.0 SP6a) reveals an Event ID 8003, stating that my computer thinks it's a Master Browser.

So I removed my computer account from the server using Server Manager and resynched the domain. I waited about 4 hours, then added the computer back into the domain, and still was unable to log into the domain.

So... I removed my computer from the domain in the Network Configuration and put it back into the Workgroup (same name as the domain, as it should be). Then deleted the computer account on the server using Server Manager again. I logged into my computer locally with Administrative rights, looked into the "Computers Near Me" that represents the workgroup and can see everyone fine, including the PDC.

Now, while trying to join the domain again using the wizard, I get to the part where I create a computer in the domain, and it returns an error saying "the domain is invalid or unavailable ".

Huh?

I don't want to have to change my computer name just to get back into the domain as I have a number of computers using my computer as where they get their antivirus definitions from and I don't want to have to reconfigure all of them.

Any ideas, anyone? Oh, and btw, I did a loopback ping on my machine and it works fine. I can also ping the PDC from my computer just fine, so I'm loathe to believe the nic is gone, but I could be wrong....

 

If you can ping out, you have eliminated lots of possibilities including NIC function.

Removing the computer from the domain and re-adding it will create a new SID (security ID) so even if the name is the same, your Domain will see it as a new machine which should have cleared up the problems.

One thing to do is to stop and then disable browser service on the workstation. No reason whatever it should be running the service as long as you have a PDC and at least one BDC.

You didn't say how IP addresses are being assigned but while you are troubleshooting, I'd suggest putting in fixed values rather than letting DHCP assign them.

Try this:

• PC back to a workgroup account - and use something other than the domain name.
• Set the NIC's IP, netmask, gateway, etc. manually. Reboot.
• Run ipconfig /flushdns.
• Unplug the NIC cable.
• Rename your hosts and lmhosts files to something else like hosts.old or similar.
• Power it down.
• Get into WINS from another PC and sunset all the existing records for your PC. Same with DNS if you are running it.
• Power back up and try to rejoin the domain.

 

Ah-ha!!!

Newt,

Thanks for the suggestions, though they weren't necessary. I discovered the culprit!

I checked the hub and the port my computer hooks into didn't have a link light. Strange that it was allowing pinging at all (though it got successively worse as time went on, until just now, it wasn't able to ping the server. I literally watched as my pings went from 100% success, to 50% loss, to 75% loss, to no response in a span of a half hour). I've never actually seen a hub port fails, but it looks like that's what occurred. On the interesting level, it was neat to see one failing in action, on the headache side of things, it's sure hell to diagnose!

So... switching my patch cable to a new port and wohola! I am up and working again!

Again, Newt, thanks for the suggestions!

Oh, and btw, the IP's are all assigned manually anyway (we aren't using DHCP)...

Adios!

 

Galdor, a word of warning...

One of my clients had a direct strike to the building around 3 months ago, and though the whole computer room and all desktops are surge protected by a big-ass UPS, a number of ports on switches & hubs have progressively died since then (possibly fried by induction through CAT5, though unsure) and have needed replacement.

If financially acceptable (most insurance policies cover you for strike damage), replace the hub, and expect possible random NIC failures... This one strike may drag on for a while

 

Redzed,

Thanks. I'll keep an eye on it.

It's not cost effective for us to replace the hubs just out of sorts (even though I'm one of 2 who get to make that decision.) even though the potential for other problems exists. I'll just have to keep an eye on them. We have 4 cascading 3com hubs that would need to be replaced if we just decided that they should be replaced because of potential surge damage. The budget won't eat that without some finangling and I'm not even sure they're insured? I'll have to look into that.

Thanks for the suggestion though. I'll keep watch on them....

 

I've also seen modules in routes go bad. Thank goodness they tend to be hot swappable (meaning you pull the bad module out WITH the power on) and replace with a new one. Other strange things I've seen in almost 6 yrs as an NT Admin is word documents getting scrambled for no apparent reason. We had one division on a certain floor that was really bad for this. We changed hubs, patch cables, reinstalled OS, eliminated auto save in word, etc. Nothing seemed to work Very frustrating.

 

pweegar,

I've seen that word scramble problem too. The way we solved it is that it was faulty wiring in the outlets causing problems with the power supplies. At first we thought it was the power supplies themselves, but it ended up being the wiring itself...

 

Galdor - having a hub munch the occasional port isn't that unusual. Happens even without a power hit like the one you took. I've had some that just munched about 1 a month until they stopped at 3 or 4 dead ports. No idea why either.

For some reason switches don't seem to be as bad as hubs and some brands are worse than others.