For the past 3-4 days, my system has been locking up about twice a day, at rather unpredictable intervals. No real hardware changes as of late, only added a fan about midweek last week.

CPU Temperatures stabilize at 51 deg C under full load (multiple parallel compiles).
RAM checked out good from memtest86+.

When system locks, no entries are written to any logfiles. Mouse does not move, pressing [num|caps|scroll] lock causes no changes in keyboard lights. System does not respond to ssh attempts or pings, and no hard disk activity is evident. Primary hard drive passes a long smartctl test. Secondary (older drive) does not support smart. Checked and reseated cables and RAM.

Only thing I can think is that it might be the power supply. Any way to reliably check that?

The problem most often occurs surfing the web or other light tasks. It has yet to occur during any sort of compile. It also only occurs when I am at the computer, never overnight or when I am in class or at work. It seems, but I'm not certain, that I am always moving the mouse when it occurs. Possible GPM or X.org bug?

I know there are a lot of threads about system lockups. I've read no fewer than 20, and all talk about temperature or bad RAM, which does not seem to be the case here. I have gkrellm running at all times, and the last time it locked, gkrellm was displaying 36.5 C for CPU, 36 C for hda, and 31C for MB.

CPU is P4-2.8 GHz, not overclocked. RAM is 2x512 Corsair DDR. Mobo is Asus P4C800-E Deluxe. Mouse is usb, but connected with PS/2 adaptor. Video card is ATI Radeon 9200SE, DRI enabled using FOSS drivers (not the ATI proprietary/binary only).

I hope someone has some ideas, because I'm out of them, and it seems the search on this site and google are as well.

The one big problem with asking Q's the right way, w/ complete relevant info, is that we can't bombard you w/ requests for more details.

Seriously, how about some general trouble shooting suggestions:

• The only recent change you can think of is the addition of a fan, have you tried disconnecting or removing it? I know that it's unlikely that your system is running too cool, but maybe it's an additional load on the PS -- the "straw that broke the camel's back" so to speak.
• The other outside possibility is that mouse you mention -- have tried connecting it to a USB port or substuting a PS/2 rodent?

That covers the items you mention, as for testing a PS, I don't know how, I should -- I've built simpler ones, but I don't. The only things I can think of are:[list=1][*]Unload it -- disconnect non-boot HD's &/or remove PCI cards.[*]Substitute a different PS, preferably of higher wattage.[/list=1]
You also might try re-seating RAM & other components (#1 & 2 above would have that side effect).

Sorry these are all so brute force & general, but maybe one is the key.

BTW, I can sympathize, my desktop has been doing the same thing less frequently & when unattended, but a similar PITA.

Well, for the moment, I'm trying running 2.6.11 kernel instead of 2.6.12. I was running this 2.6.11 kernel for about a month (20+ days uptime) before the upgrade to 2.6.12, so it should be good.

I'm beginning to think the PS is not at fault here because I put the system under some huge loads last night and the system handled it fine. I burned a DVD backup, did some (admittedly minor) compiling, and rsynced several gigs of data with no problem. All these seem like things that should stress the PSU.

So far (though I wasn't home much of the day) no lockups. We'll see if it was a kernel issue within a day or two, most likely.

As far as asking the question the right way: between having the link in my sig, and the number of times I've had to drag information from users, I do my best to ask good questions. I wonder if that's why I rarely get a response to my posts: just because most replies to posts are either the bleeding obvious or a "post more information" deal.

Ah, the loneliness at the top.

Again, but seriously, you have just given me a reminder in trouble shooting -- "What (else) has changed?"

Glad to hear you are on the track of a solution.

So my system has now been running 32 hours straight on 2.6.11. I wonder what 2.6.12 changed so substantially. I had to rebuild my fuse module... was wondering why it wouldn't work for like 20 minutes.