[CentOS] guest vms crash host systems

Negative negativebinomial at gmail.com
Tue Oct 4 19:51:29 UTC 2011


On Tue, Oct 4, 2011 at 2:24 PM, <m.roth at 5-cent.us> wrote:

> Negative wrote:
> > On Tue, Oct 4, 2011 at 1:18 PM, <m.roth at 5-cent.us> wrote:
> >> Negative wrote:
> >> > On Tue, Oct 4, 2011 at 10:41 AM, <m.roth at 5-cent.us> wrote:
> >> >> Brian Mathis wrote:
> >> >> > On Mon, Oct 3, 2011 at 10:16 PM, Negative
> >> <negativebinomial at gmail.com
> >> >
> >> >> > wrote:
> >> >>>> I built guest vm's one for Windows 7 and one for Windows XP using
> >> >>>> the
> >> >>>> virtual machine manager on a  just updated to centos 5.7, and they
> >> >>>> are both crashing the host machine. They run only  for a few
> >> minutes,
> >> <snip>
> >> >> > Is this new hardware?  Have you run any hardware burn testing (CPU,
> >> >> > RAM, etc...) and/or memtest86+ on the RAM?  This sounds like a
> >> >> > hardware issue to me.
> >> >
> <snip>
> >> More and more it sounds like a hardware issue. Hmm, every time X is
> >> running, and you say you had one video card fried - how did it fry?
> >> Also,
> >> is this machine on a good quality surge protector? Have you had a
> >> thunderstorm, or power outages recently?
> >
> > The vendor told me that the particular video card model (I forget which)
> > had some flaw. In any case the fan stopped running, and it heated up. I
> > usually use the machine remotely but I was at the console at that moment.
> > The monitor started flickering and then went gray.
> >
> > The vendor sent a replacement, but I had thrown an old ATI in before it
> > arrived. When the crashes occurred now, I finally put in the replacement.
> > Same behavior.
>
> Have you examined the m/b and cards *around* where the card fried? Its
> heat death may have affected things around it.
>

It looks good to the eye. The capacitors look good.


> >
> > The surge protector is good. I live in NYC and the biggest environmental
> > hazard is the cleaning lady, who has in the past tripped the surge
> > protector switch.
>
> <g> Do you know the story about the mainframe shop, the racks of tapes,
> and the cleaning staff?
>

Don't know it but I can imagine.


> >
> > I fear you're right about the hardware. But as far as I can tell
> > everything else works fine. I went overboard in buying two quad
> > processors -- so I could live with one if that's the problem.
>
> A replacement m/b?
>
>
That's a tough one!

Since the crashes can be duplicated and are only caused by this one
combination of events, I don't know.

On another machine, I had a case where none of the kvm guests would boot. It
turned out to be a conflict between libvirt and the nvidia proprietary
driver. I used an old version of the video driver until Nvidia caught up.
(It only affected amd processors.)



More information about the CentOS mailing list