[Mageia-dev] Strange system stalls - neep a tip

[Robert Fox]

On Mon, 2012-05-21 at 19:50 +0200, David Sjölin wrote:
> On Mon, May 21, 2012 at 7:33 PM, AL13N <alien at rmail.be> wrote:
> > perhaps check the logs (/var/log/messages) for information regarding those
> > downtimes.
> >
> > see if the ping also delays
> >
> > perhaps try to monitor something via ssh or in tty, to see if it's only
> > the graphic card.
> >
> > also, check ntpd settings, do they make big jumps?
> >
> > etc...
> >
> >
> > in short: logging is always the key.
> >
> >> Hello All -
> >>
> >> I am experiencing a strange problem with a specific machine running
> >> Cauldron.  I am beginning to believe it is NOT a Mageia problem because
> >> when I boot into Ubuntu 12.04 the problem exists as well.  I think it
> >> may be hardware related but I can not find what is causing this issue.
> >>
> >> The machine is an Acer Aspire x3900 with a quad core ‎AMD Phenom(tm) II
> >> X4 925 Processor - 6GB memory.  GeForce GT215 (GeForce 320) card. I am
> >> running the x86_64 version of Cauldron.
> >>
> >> The problem occurs under both KDE and Gnome
> >>
> >> Basically, the system seems to freeze up (mouse moves) but nothing else
> >> - gkrellem freezes and no errors occur - it just seems to hang - but
> >> magically wake up again after some time (many seconds at a time)
> >>
> >> I have checked various logs and can not find any problems - so I am
> >> stumped how to find the source of the problem.  I leave the machine
> >> running 24x7, but don't think it is a thermal issue.
> >>
> >> When the system runs under Windows 7 (64bit) it appears to run fine.
> >>
> >> Any hints what I could check?
> >>
> 
> Hello!
> 
> If you want to log cpuusage to see if some process has extra high cpu
> usage during this freeze you could do something like:
> 
> while true; do date >> testfilen.txt ; top -b -n1 >> testfilen.txt ;
> sleep 1; done
> 
> Then you will get top-output every second (sleep 1) or whatever time
> you specify for sleep. I think there is a way to do this with top
> only, without the delay but not sure which flags to use then (possibly
> top -d 1 -b)
> 
> The file testfilen.txt will of course grow quite large after a while,
> so you can't keep this running 24/7 maybe, but maybe you could clear
> out the file regularly until you get the error situation again and
> then based on the date output you should be able to find cpu usage
> around the time of the problem.
> 
> Regards,
> 
> David

Thanks for the quick responses.  I checked the cpu usage, but there
doesn't appear to be a spike - and the load average in htop is real low:

load average: 0.46, 0.67, 0.56

so if the X server or something else was maxing out the cpu, I'd see
that. 

Also, if I use a console in KDE (like yakuake) - the stalls happen in
there as well.  At first I though this was Firefox related (like flash
problem or something) because it was most noticeable when I surfed -
firefox freezes off and on and it is really frustrating.

This machine was working fine for at least a year - and I was suspecting
something with the latest kernels and the AMD processor - but have no
evidence.

The search continues . . .

Thanks again,
R.Fox