Server Problem

I've got a real problem with my company's main server. Please bear with me.

This AM, the server was found dead in the water. When we tried to restart it, we received the message "HAL.DLL is missing or corrupt ". No problem, thought I, as I popped in the Windows 2K server CD, and went to the recovery console. Ran "chkdsk /r ". Started promisingly, but hit 75%, backedup to 50% and then progressed agonizingly slowly. Reached 72% and backed to 52%. But started to run more quickly. Finally completed and said "unrecoverable errors ". BTW, this process took 3 hours. I was concerned about stopping... feeling that I might leave something in mid-fix. Convinced that the server had crashed overnight and that I had some disk damage, I tried the "easy way" and (using another computer) did a drive image of the disk to a brand new drive. I figured there might be some problems, but that the recovery console could take care of them.

Put the new disk in the box. Good new, system started .... almost. Got through logon when BSOD. Tried to bring it up in Safe Mode. This worked. System said that it had installed new hardware and rebooted. BSOD. Ran Automatic Recovery from the CD. All is well. New BSOD message. 5 hours have passed.

Out of desperation, put the original damaged disk back in the box. System booted and ran OK. Made backup of company database and let the troops try to work with the system ... at least we can do some work. This went fine, there wer no problems during the remainder of the day. BUT, I noticed that drive E was "missing ". I brought up the disk manager console and found the disk was "offline ". Tried to re-activate the drive and succeeded ... but drive F disappeared. This box has 4 disk controllers for the 4 drives and 1 CD.

At 5 PM, I kicked everyone off, backed up the database, shut the system down. Replaced the damaged old drive with the BSOD new one. I also chedked on the "missing drive" by placing it as the master drive in the secondary controller of the XP box. Drive was fine. Ran chkdsk. Drive is fine. Put it back into the win2k server and put the BSOD generating new drive. Botted, all is well. No BSOD. But, also No D or E drives.

While this is going on, I put the "damaged OS drive" into the XP box as a secondary device (drive d). Ran chkdsk /r. No problems.

I had thought that I had a simple disk crash with some damage to the media. Now, I don't know what to think. I've ruled out bad disk controllers, because the disappearing disks are on different controllers. My current candidates are: power supply(s), motherboard, or some kind of virus/trojan/malware.

Anyone have a similar experience, or any ideas as to how to check this out?

Thanks for any suggestions.

 

What hardware are you running?

If your "server" is a server class machine you should have some diagnostics for the box that will do lower level snooping than the OS can. Beats guessing any day.

When you mentioned 'power supply' as a possible problem it sounds like there is only one so I'm sorta betting on a beefed up PC rather than a regular server with redundant everything.

Failing in diagnostics for the whole machine, maybe some diagnostics for the motherboard.

IDE or SCSI?

Where do you stand with tech support for the box?

 

Sure sounds like a bad IDE controler to me.

 

The server is a doit yourself box. But, it has dual hot swap power supplies, dual AMD 2GHZ processors on a gigabyte board, and 4GB of registered ram.

There are two IDE controllers and two promise raid controllers. Started out to use the raid controllers for the OS and data with the IDE for the cd. We're using the Navision MRP package and their techie recommended the current configuration. There is no tech support for the box. I don't think there is anything in the way of diagnostics. But, this box has been working almost flawlessly for a year and a half. It is on a big, big ups and has never had a real power outage.

 

Problem Resolved

It turns out that it was the dynamic disk problem. Finally got MS tech support expert who know exactly what the problem was and how to fix it. He reset the disks from dyamic to basic by changing one byte in the header of each of the three disks. Turns out the disks had been basic and somehow got changed to dynamic. I didn't do it, but it must have happened during all of my flailing trying to get the system to work correctly.

 

Excellent news.

And on a more personal note, really glad to find that you are using real server hardware for your server. Too many horror tales of companies trying to get by with the much less expensive PC hardware.