Friday, May 27, 2005

Death Watch

Problem: I'm writing this on a machine which I though was completely dead. In fact it might die before I get done with the posting. It has had the most bizarre set of problems I've ever seen up until about 4 hours ago. I was up working on it until 3:00 am this morning, slept a few hours and then got right back to working on it. I was alternating between installing XP for about the 6th time, running diagnostics and removing components. I really thought it was ready for the trash; it might still be. The really frustrating thing is that the machine would seem to run fine from the initial OS install through SP2 and all the current virus definition updates. Then I would install the latest Windows updates and the machine would go haywire. The reboot after the updates wouldn't finish; it would get a BSOD which flashed by before I could read it and then the system would recursively reboot. I also got various messages like:
  • Dell E771a hardware error
  • Recursive WMI failures (while downloading Windows fixes)
  • Total system hangs; no cursor movement
  • Odd screen settings, like classic Windows display settings when I didn't even choose them. After a reboot the settings would change back to Standard XP

I was thinking, this has got to be hardware, right? Wrong! I ran extended diagnostics this morning on all system components. It took 7 hours and I didn't find a single error! Damn .........

Solution: I'm trying not to get my hopes up but this is the longest the machine has run in 2 months. What I've done so far on this rebuild is:

  • Install XP
  • Install McAfee and get the current definitions
  • Install SP2
  • Run sfc to check for corrupt files (sfc.exe /scannow). While this was running I got several WMI failure notices.
  • Run CHKDSK /F (which says it fixed some files)

At this point there are some Windows updates ready to install. As soon as I complete this posting I'm going to run the updates. If the machine comes back up, I'll update the blog to let you know it worked. If it didn't work you'll know by the sound of me kicking the damn tower all the way to my trash can and cussing all the way back.....

. . . . . . . . . . . . . . . . . . . . . . (^%#$^%$&%)

4 days later - You probably guessed that it didn't work!

Since I find it very difficult to throw away hardware no matter how flawed it might be, I haven't completely given up yet. Stay tuned .......

Many days later - After a few days of going blind while trying to read the BSOD, a split second at a time, I was finally able to see that the Stop Code was C000021a (fatal something or other....). I did some internet searches on the Stop Code and found some interesting hits. Thanks to the hard work of some diligent people on the CNet XP Forum, I finally got to the bottom of this. It turns out that the problem was caused by applying MS hotfix #890859, Article MS05-018. The CNet thread suggested turning off automatic updates, which I had already done. Then they suggested going to the Windows update site and choose Custom Install instead of Express Install. In the Custom Install dialog, uncheck fix #890859 and mark the fix as hidden. Then install the rest of the fixes. This did the trick! Thanks guys!!!

Note: To-date the MS article still doesn't warn us about this problem ......