[mythtvnz] Pointers on how to track down machine lockup

[Solor Vox]

On Mon, October 11, 2010 11:16, Wade Maxfield wrote:
>
> I went through that list and didn't find anything that stood out as an
> issue. Can't find any evidence of bad RAM after running memtest for 20+
> hours, CPU and other temps are all under 40°C, no mention of segfaults in
> any log, and I reset the BIOS to the failsafe defaults.
>
> So I clean installed Ubuntu 10.04 Server, and then added enough packages
> to get MythTV & gdm running and a few other things like lm-sensors,
> smartmon & epgsnoop, etc. I have't added any Nvidia packages. Machine
> still locked up every few days for the following 2-3 weeks.
>
> I then found a BIOS update for the motherboard, installed that Oct 2 and
> since then the machine hasn't locked up.  8)
>
> But it has spontaneously restarted itself twice.  Oct 9 16:32, and Oct 11
> 9:05.
>
> I can't find anything in the logs to indicate why it's doing that though.
> Going through syslog and kern.log the machine is just operating normally,
> then all of a sudden it's booting. It's connected to a UPS so it shouldn't
> be losing power or restarting because of some sort of power spike. Nobody
> was sitting in front of the machine at the time, but a remote frontend was
> playing back a recording during the Oct 9 restart.
>
> Any thoughts on where to look or other things to try?
>
>
>  - Wade

If you have another PSU on hand, that would be the next best thing to
check.  You said you check everything on the list... how did you verify
the PSU was good?  You might also try a CPU burn-in (cpuburn) along with
some disk IO (iozone3) to simulate both system and power loads.  I
recommend using one cpuburn mmx process for each core.  These should cause
any weak components to fail rather quickly.

Good luck,
sV