[clue] [hardware] Erratic boot behavior

[Matt Gushee]

Hi all--

Hope everybody is well!

I am having an issue with my desktop Linux box--though I think it is
really strictly a hardware or firmware problem. Anyway, before I go out
and spend a bunch of money based on my guesses, I thought I'd check with
you knowledgeable folks and see if anyone has encountered this before.

First of all, the computer in question is, hmm, less than
top-of-the-line. And it's quite old. The case is a Gateway mid-tower
that I bought used about 7 years ago, and I think it was made in about
1998. AFAIK the power supply is original equipment. The motherboard is
an ECS model that I got for cheap (closeout, I think) at MicroCenter.
Cheap, but the reviews I read said it was good for the money. AMD Athlon
1.8Gig CPU, SiS chipset (forget the model #). So maybe it's time to
upgrade anyway. But if there's reason to believe that the problem is
something else, then I'd like to take care of that something else first.

So anyway: the machine is now having serious trouble booting up after
being powered off. This started a long time ago, but it was just a mild
nuisance at first, gradually developing into a major problem. The
behavior has also been very inconsistent: sometimes the system will boot
on the first try. But when it doesn't, here's how it goes, more or less:

 * I hit the power button.

 * LEDs light up on the front of the case. The power-on LED will be
 steadily lit, and the CD-RW drive LED
   starts to flicker, as does the hard drive LED. Sometimes the DVD-RW
   drive LED will start to flash 
   and/or there will be disk-spinning noises. These are good signs,
   usually predicting a successful
   bootup.

 * Usually, the ViewSonic parrots appear at the upper left of the
 monitor, then disappear--followed by
   the words "No Signal" at the center of the monitor.

 * Sometimes the keyboard LEDs flash. That's also a good sign.

 * After a period of time (maybe 10 seconds, maybe 5 minutes), the POST
 screen appears. Then, if I'm
   lucky, the 2 hard drives are listed on screen. Often, the POST screen
   will just flash on and off,
   and it all starts again. The general trend is that it is taking
   longer to get to the POST screen,
   and it flashes on and off more times before a successful boot.
   Sometimes it won't POST at all. In
   those cases the CD-RW LED and the hard drive LED just keep steadily
   flickering, there are no sounds
   of moving parts, and pressing and holding the power switch fails to
   turn the machine off. So my only
   recourse is to cut off the power by switching off the power strip or
   pulling the plug (yikes!).

Now, a couple of facts that seems significant here are that (a) once
Linux is booted, the system runs very reliably, with no random failures;
and (b) as far as I can remember the system has never failed to boot
once the hard drives are detected. At least, I am assuming that when the
POST screen displays the hard drive info, that means that the drives
have just been detected. If that is the case, it seems likely that the
problem has nothing whatever to do with the OS.

BTW, In case it matters, I have upgraded the primary hard drive at least
once, probably twice, or maybe even thrice. As far as I can remember
those upgrades had no effect on this problem.

So, anybody have an idea about this? Any clues greatly appreciated.

-- 
Matt Gushee
matt at gushee.net