Re: Debugging mysterious freeze / crash
On Sun, 23 Sep 2018 11:27:19 +0500
"Alexander V. Makartsev" <avbetev@gmail.com> wrote:
> On 23.09.2018 07:51, Celejar wrote:
> > Hi,
> >
> > I've been experiencing a great deal of frustration recently with
> > intermittent freezes / crashes on my Debian Sid system (a Lenovo
> > W550s). The symptoms are that the screen totally freezes and the system
> > becomes completely unresponsive (even ssh attempts from another machine
> > fail), and the only thing that seems to have any effect is a hard
> > reboot (holding down the power button until the system restarts).
> >
> > Upon reboot, I can't find anything at all interesting in 'journalctl -b
> > -1', or /var/log/syslog - the former just shows everything looking
> > normal until the moment of the crash, at which point the log just ends,
> > and the latter also just shows everything seeming to be fine until the
> > moment of the crash, and then shows the boot messages from the reboot.
> >
> > Any ideas of what could be causing this, or how I could go about
> > debugging it? I've been using this machine for years without
> > experiencing anything like this, and I'm not sure for how long this has
> > been a problem. I did recently upgrade from stable to unstable, but I'm
> > not sure whether or not the problem's initial occurences coincide with
> > the upgrade.
> >
> > Celejar
> >
> With symptoms like this, I'd suspect hardware problems. If there is
> something bad happened with some software the kernel would know and warn
> about it.
>
> Best way to debug it, I think, is to test hardware under load with
> another fresh OS. Debian stable or Windows trial version would do.
> Severe over-heating can cause this. So monitor temperatures of CPU and
> ICH. Dangerous levels like 80C-90C-100C should be dealt with.
I suspected thermal issues, but the problem seems to occur when the
fans aren't even running (of course, maybe there's some fan
related malfunction ;)). I'll keep a closer eye on the sensors.
> You have to check if battery is ok, PSU performs alright and can deliver
> power under high-load.
> During tests you can apply small amount of vibration to the laptop (with
> constantly typing on the keyboard, or just tapping on the sides of the
> laptop) to check if there are some solder joints become crackled and
> could loose contact temporarily.
> Do all tests sequentially, to be able to find root cause for the
> freezes. If you got few lockups while tapping, you have faulty
> motherboard that should be serviced or replaced.
One of the reasons I buy Thinkpad T (or W) series machines is to not
have to worry about such things, but I understand that there are no
guarantees ;)
I'll try some stress / system testing (memtest, sysbench), with
tapping, and see what I find.
Thanks much,
Celejar
Reply to: