This document is available on the Internet at: http://urbanmainframe.com/folders/blog/20040831/folders/blog/20040831/
I returned to the office after my week's sojourn in Paris only to find that one of my file-servers had died in my absence. It seems our building suffered a direct hit during an electrical storm and, in accordance with Murphy's Law, I had a single server that wasn't connected to our "clean" electrical circuit (that'll teach me).
The server in question is not business critical (so, thankfully, I am still employed). Nevertheless, it was a real pain having it out of action since it acts as my media server (music, video and imagery) - and I can't work without Pink Floyd!
The server was home to five hard disks: a single 20GB IDE unit upon which the OS (Windows 2000 Server) was installed and four SATA disks, in a RAID 10 configuration, for the data.
An examination of the server revealed that the IDE disk was damaged, seemingly beyond repair.
I rebooted the server with a Knoppix CD and was thus enable to ascertain that the RAID was still intact, with full data integrity. Of course, I wasn't too concerned about this, since we employ very strict backup procedures for our file-servers (experience is such a great teacher).
Restoring the server to a fully operational state should have been a no-brainer: substitute a new hard disk for the damaged one; install Windows 2000 server, reconfigure the shares and so forth.
It should have been so simple. Sadly Windows conspired against me in every possible way...
“I teased and cajoled Windows 2000 Server and for six interminable hours it responded by kicking me in the groin”
The OS install was uneventful in itself. It was only when I performed the post-install reboot that the problems began to manifest themselves.
The computer wouldn't boot! It got as far as the Windows 2000 Server splash screen then just hung. No disk activity, no response to the keyboard, nothing. I figured it was a driver issue. I rebooted and started the server in "safe" mode which got the desktop up. I dived into the device manager and found that various devices were flagged up for attention - both the sound and video cards were bereft of drivers and there were various problems with USB Root Hubs and PCI system devices.
I right-clicked and uninstalled everything that was flagged and rebooted. No dice! So I booted again in "safe" mode. The desktop came up and I was immediately assaulted with various "Found New Hardware" dialogues which I worked through ("Now where did I put the CD with that driver on?"). Then I fired up the device manager once more to check for warnings, there were none.
I tried to reboot. Splash screen, then... lock up!
Returning to the device manager in safe mode, I discovered that both the sound and video cards were suffering driver problems again and the NIC was not even listed! Huh?
I turned the machine off and removed the sound and video cards. I hooked the monitor up to the motherboard's VGA port. Rebooted, got the splash screen and the same damn failure. So I was starting to get a little bit pissed off. I powered down and opened up the computer, I pulled the RAID controller and the NIC - so there was nothing "non-standard" as far as the hardware was concerned. For good measure, I also pulled the RAM and dropped in a new 512MB module. It's worth reiterating at this juncture that the system worked perfectly with Knoppix, so I didn't really suspect a physical problem with the hardware (having already replaced the dead hard disk).
Another reboot and another crash with the splash screen. WTF?
Then I saw it... I had a scanner connected via USB - could that be the problem? I ditched the scanner and rebooted. Finally, success! Obviously I beat myself up a bit at that point for missing the fucking thing in the first place, a simple oversight, but I should have noticed it sooner.
I reinstalled the video, sound and RAID cards, and the NIC, then rebooted, with fingers crossed. After a few heart-stopping pauses the desktop reappeared. Perhaps my troubles were over? Oh no... Windows still wanted to play games.
My next step was the infamous "Windows Update". I installed everything available, rebooting frequently, until Windows Update had nothing left to offer me. I knew that, with everything downloaded, I should have restored full USB functionality. So, after the obligatory reboot, I hooked up the troublesome scanner (a HP Scanjet 4400C) and installed its software and drivers. I tested it, it worked. Then I rebooted and yes, you guessed it, the computer hung on that bastard splash screen. "Fuck it," says I, "the scanner can go on another PC."
The scanner sailed through the air in exactly the same manner that any violently thrown scanner would sail. Sadly, its all too brief experience of unpowered flight was interrupted by a rather solid wall. The scanner was shattered by this sudden barrier to its aerial progress. Oh well, scanners are cheap!
With the scanner issue resolved (ahem), my final task was to set up the shares to make my media library available to the LAN. Ah, writing the previous sentence makes it all sound so simple. Sadly, Windows hadn't yet delivered the kill shot.
The share configuration options weren't available to me. Of course, this shouldn't have been a problem. I simply had to install "File and Printer Sharing" in the network configuration dialogue.
"Not so fast buster," screamed Windows, "first you got to get me attached to your LAN!"
So that is what I tried to do. I tried and I tried and I cried. I could not get the standalone server onto my LAN. I couldn't get the computer to authorise itself with my domain controller. I uninstalled and reinstalled networking, I swapped out the NIC for a new one. I Googled for two fucking hours looking for answers.
This simple procedure, which I had successfully negotiated many times before, on LANs more complicated than my own, proved to be beyond me. I could ping the server from my workstation, I could see the LAN from the server. But I couldn't get the damn server to join the domain.
For six, long, hair-pulling hours I teased and cajoled Windows 2000 Server and for six interminable hours it responded by kicking me in the groin. It shouldn't be this bloody difficult to set up something as simple as a file server!
I don't like to quit, I don't like to be beaten. I could have persisted, but after six hours of futility I decided to bail out. I plucked Red Hat Linux 9 off a shelf and did what I should have done from the beginning. I had my media server up and running with Linux inside an hour. Less than one hour! Read 'em and weep Microsoft.
Despite the rant, this is not an anti-Microsoft article, it is not an anti-Windows article. I use Windows daily. I have computers running Windows XP, I have servers running Windows 2000 Server. I even have an old DEC Alpha server, still running Windows NT 4, that has over three years of continuous uptime under its belt. For the most part, I have found Windows NT, 2000 and XP to be stable and reliable. But this isn't always the case and I have written this article merely to document the pain of the end-user when faced with the sometimes inscrutable machinations of a Windows operating system.
This won't happen to me again. If another of my Windows 2000 servers fails, I will restore it on Linux. I will continue to do this until I have no Windows servers left. I will do this because my time is valuable and I can't afford to spend six hours holding the hands of a recalcitrant operating system.
- One angry Sys-Admin, signing off.