Re: [gentoo-user] spontaneous reboots.. what to look for

[Mark Knecht <markknecht@gmail.com>]

On Sun, Feb 15, 2009 at 3:42 PM, Harry Putnam <reader@newsguy.com> wrote:
> I've been experiencing spontaneous reboots on one gentoo machine
> lately.  Looking thru /var/log/messages... I see the restarts but
> looking above that... I'm not seeing anything I recognize as being a
> culprit.
>
> Its been happening for a few weeks... but I've been busy and only now
> digging into it ( The machine is no kind of server ).
>
> It appears to only happen in X (I'm using xfce4) and I've only noticed
> it since I started running 2.6.28 kernels.  Although I couldn't say
> that it seemed to be directly related.
>
> I mean I didn't boot into 2.6.28 and suddenly notice spontaneous
> rebooting.
>
> It does not appear to be heat realated... but I am only now using
> lm_sensors to keep an accurate record and see if there appears to be a
> relationship.
>
> I've had two today so either its happening more often or I'm just
> spending more time on that machine.
>
> It may also be on the first or second time its happened while I as
> actually right at the keyboard.
>
> I'm sorry to be so vague about it, but in truth, I've been pretty lazy
> about it... since no real harm comes of an unexpected reboot on that
> machine (so far anyway).  But clearly something that has to be figured
> out.
>
> The only things I've checked so far...
> 1) browsing thru /var/log/messages (Having trouble recognizing any
>   thing that looks suspicious.
>
>   I have noticed what appears to be a time/date anomaly where the
>   progression of time is suddenly irregular.  That is, an earlier
>   time shows up amongst some later times.
>
>   It appears to have been me sudoing to visudo.  And apparently
>   having /etc/sudoers open long enough for the closing of it to be
>   earlier than other events taking place.
>
>   Again ... I'm not real sure exactly what happened there but it
>   does not appear to coincide with a reboot anyway.
>
> 2) checking how hot the cpu is getting (Doesn't appear to be a
>   problem) But now running a cron job recording temperatures every 10
>   minutes. So that may turn up something.
>
> 3) checking for overfilled disks.  (none show in df -h)
>

Reseat memory and PCI cards, etc. Consider removing for a period of
time any hardware not absolutely necessary to debug the problem. (I.e.
- second video card, extra disk drives, extra network adapters, etc.)
Run memtest86 for a few days if you can spare the machine. Run
spinrite, etc., to look for drive problems. Open the box up and place
a fan blowing extra air for additional cooling.

good luck,
Mark