Hi,
We have a problem on a 2.2.17: sometimes it crashs
without any reason (no high load), there is no kernel panic,
the screan is black. We setup watchdog software and
we realized watchdog can not reboot this box whe it crashs
(on the others servers it works fine).
my question is:
what kind of problem can have this serveur:
hardware or software ?
Thanks for help
Octave
> We have a problem on a 2.2.17: sometimes it crashs
> without any reason (no high load), there is no kernel panic,
> the screan is black. We setup watchdog software and
> we realized watchdog can not reboot this box whe it crashs
> (on the others servers it works fine).
>
> my question is:
> what kind of problem can have this serveur:
> hardware or software ?
What sort of watchdog are you using ?
Alan Cox wrote:
>
> > We have a problem on a 2.2.17: sometimes it crashs
> > without any reason (no high load), there is no kernel panic,
> > the screan is black. We setup watchdog software and
> > we realized watchdog can not reboot this box whe it crashs
> > (on the others servers it works fine).
> >
> > my question is:
> > what kind of problem can have this serveur:
> > hardware or software ?
>
> What sort of watchdog are you using ?
software. no hardware solution.
http://www.ibiblio.org/pub/Linux/system/daemons/watchdog/watchdog-5.1.tar.gz
Octave
> > > my question is:
> > > what kind of problem can have this serveur:
> > > hardware or software ?
> >
> > What sort of watchdog are you using ?
>
> software. no hardware solution.
> http://www.ibiblio.org/pub/Linux/system/daemons/watchdog/watchdog-5.1.tar.gz
The software watchdog will fail if the kernel is badly mashed or interrupts
are disabled. That means it doesn't help tell me if the problem was hardware
or software (nor in general do hardware watchdogs). Is this one box running
different loads to the others or different in any notable way ?
Alan Cox wrote:
>
> > > > my question is:
> > > > what kind of problem can have this serveur:
> > > > hardware or software ?
> > >
> > > What sort of watchdog are you using ?
> >
> > software. no hardware solution.
> > http://www.ibiblio.org/pub/Linux/system/daemons/watchdog/watchdog-5.1.tar.gz
>
> The software watchdog will fail if the kernel is badly mashed or interrupts
> are disabled.
yes.
> That means it doesn't help tell me if the problem was hardware
> or software (nor in general do hardware watchdogs). Is this one box running
> different loads to the others or different in any notable way ?
nothing really different. we have 5-6 servers like this one and
this one have been crashed for 2 days only (once per day).
never before !? when the server crashed the load was 1.2. nothing
really bad.
the only difference is: this server has the serial cart. I was
in contact with Ted for the serial problem on Linux, but I have
no news from about the crashs.
I do not know if the serial cart can crash the server that
interrupts are disabled ?
Octave