[PLUG] Mount can't make up its mind

[John Jason Jordan]

On Tue, 1 May 2007 17:58:53 -0700
"Rogan Creswick" <creswick at gmail.com> dijo:

> This sounds a *little* like an overheating thing, but that's really a
> long-shot.  I don't know of many things that will intentionally make a
> computer go black like that.  Generally this would be almost certainly
> be accompanied by some form of beeping, hopefully from a small
> on-board speaker (not the pc speaker, or not *just* the pc speaker),
> but if for some reason the speakers are borked, this could be the
> reason...
> 
> Anyway, just throwing that out as something to check.  Pop into your
> bios and see what the thermal alarms are set to, and maybe tweak them
> down a *lot* and see if the same thing happens under a controlled
> test.

Temperature is a good idea. However, checking the BIOS, I'm pretty sure
it is not the problem here. The BIOS page for PC Health says:

Shutdown Temperature	80C
Warning Temperature	75C

And then it displays the current stats, changing them while I watched
(I put the range next to the item)

CPU T Control			39-43C
System Temperature	40C
CPU fan Speed		1003 - 1054 RPM
System Fan Speed		0 RPM 
CPU Vcore			1.28V
VDIMM				1.87V

The system fan speed is 0 because it is mounted on the side of the case
and the side of the case has been off for days. In other words, small
animals could move in and nest.

However, the one thing of interest that is not listed is the temp of
the hard drives. I don't think there are any sensors on them. They are
warm to the touch, but not unpleasantly so. 

More to the point, the problem happens always and only about a minute
after starting to format the 250 GB partitions, and it occurs exactly
the same way whether the system has been running all day or has been
shut off and is dead cold.

I think we have to rule out heat.

I'm also leaning against a bug issue. It is certainly possible that
something in that messed up /dev/md3 could be so broken that it sends
e2fsck into a tizzy. But if there was a bug in e2fsk it should cause a
kernel panic and hang the computer, or puke up a screenful of error
messages and abort. I've never heard of a software bug that could
summarily power down the computer.

If it were the CPU I'd be having problems all over the place.

I think the only hardware left to consider are the hard disks. Is there
any software that will scan them to see if there are any physical
defects?