Diagnose random freezes (Linux)

"Please leave a message at the beep, we will get back to you when your support contract expires."

Moderators: phlip, Moderators General, Prelates

J. Curwen
Posts: 15
Joined: Wed May 02, 2012 11:45 am UTC
Location: Germany
Contact:

Diagnose random freezes (Linux)

Postby J. Curwen » Thu Mar 27, 2014 6:16 pm UTC

I have a Linux box that crashes quite often without giving any clues what might be wrong.

This behaviour ist consistent over different OSs: First Ubuntu (upgraded since 9.04 to 12.10) and now Debian 7 + KDE.

The machine boots up normaly, login via display manager also works but after starting an application (say LibreOffice) the whole system freezes. This is not application-specific and also does not occur every time. The solution is a hard reboot, after that it usually works.

What I found out so far:

  • No messages in /var/log/syslog from before the crash
  • CapsLock and NumLock keys do not work
  • Mouse pointer not movable
  • The "magic sysreq" command does not work (AltGr+Print+{r,e,i,s,u,b}), but is enabled; I checked when not crashed.

Now I think it could be some faulty hardware, but I do not know where to look.

How do I proceed?
Toast, being an inanimate object, obviously lacks both the ability and the desire to right itself.
-- Wikipedia on the "Buttered cat paradox"

User avatar
Infernalis
Posts: 68
Joined: Mon Nov 10, 2008 10:13 pm UTC
Location: Cube farm

Re: Diagnose random freezes (Linux)

Postby Infernalis » Thu Mar 27, 2014 7:33 pm UTC

I've dabbled in linux, but not nearly enough to know everything so my advice is a little skewed towards Windows troubleshooting.

Anyways, I think you're correct about it being a hardware issue. I've never seen software crash to the point of not being able to toggle numlock without generating at least some kind of log.

If this were a windows pc I would suggest you run a chkdsk and see if it reports any errors (is there a linux version of chkdsk?). I would also take a look at running a RAM test, checking for overheating, and checking for power issues in that order. Unfortunately when it's hardware and you have no logs the best you can do is look for discrepancies in diagnostic tests that might point towards the fault.

User avatar
forward4
Posts: 30
Joined: Wed Mar 05, 2014 2:40 pm UTC

Re: Diagnose random freezes (Linux)

Postby forward4 » Fri Mar 28, 2014 1:11 pm UTC

Try to run diagnostics from your BIOS. This is a hardware issue, as it is present across multiple platforms and programs and also leaves no logs.
Minecraft and Mathematics; dragons and dinosaurs; raptors and rotational forces

Carnildo
Posts: 2023
Joined: Fri Jul 18, 2008 8:43 am UTC

Re: Diagnose random freezes (Linux)

Postby Carnildo » Sat Mar 29, 2014 8:12 am UTC

Based on the trigger you describe, the most likely candidates are faulty RAM or a faulty video card (bad video drivers are also a possibility). Memory's easy to test: boot into Memtest86+ and leave it running overnight. Video is harder to test: you can rule out driver issues by switching between the open-source and closed-source drivers for your card, but the only good way to test a video card is to move it to another machine and see if the crashes still occur.

J. Curwen
Posts: 15
Joined: Wed May 02, 2012 11:45 am UTC
Location: Germany
Contact:

Re: Diagnose random freezes (Linux)

Postby J. Curwen » Sun Apr 06, 2014 8:39 am UTC

forward4 wrote:Try to run diagnostics from your BIOS. This is a hardware issue, as it is present across multiple platforms and programs and also leaves no logs.


Browsed through the BIOS, but found nothing that looked like diagnostics tools. Is there anything that can be run from linux or some live CD?

Carnildo wrote:Memory's easy to test: boot into Memtest86+ and leave it running overnight. Video is harder to test: you can rule out driver issues by switching between the open-source and closed-source drivers for your card, but the only good way to test a video card is to move it to another machine and see if the crashes still occur.


Memtest ran over night (12 hours, 12 passes) and reported no errors.

Concerning the video card: I will try switching drivers if it is not too complicated. By "move it to another machine and see if the crashes still occur" you mean "see if the first machine now does not crash" or "see if the second machine now crashes"?
Toast, being an inanimate object, obviously lacks both the ability and the desire to right itself.
-- Wikipedia on the "Buttered cat paradox"

User avatar
LucasBrown
Posts: 298
Joined: Thu Apr 15, 2010 2:57 am UTC
Location: Poway, CA

Re: Diagnose random freezes (Linux)

Postby LucasBrown » Sun Apr 13, 2014 1:49 pm UTC

J. Curwen wrote:Concerning the video card: I will try switching drivers if it is not too complicated. By "move it to another machine and see if the crashes still occur" you mean "see if the first machine now does not crash" or "see if the second machine now crashes"?
I'd find a problem-free computer and swap the video cards, then turn them both on, and then see which of them crashes. If the faulty machine still crashes, then it has a non-video problem; if the second machine crashes, the video card has a problem.

J. Curwen
Posts: 15
Joined: Wed May 02, 2012 11:45 am UTC
Location: Germany
Contact:

Re: Diagnose random freezes (Linux)

Postby J. Curwen » Thu May 22, 2014 10:47 am UTC

So, quite late, but an update.

I removed the nvidia graphics card and purged all of its drivers. This seemed to fix it, but now I am getting the same freezes as before. There are still no errors in the logs.

What could I check next? Harddrive?

I know it has "SMART"-capabilities, but I don't know how to interpret its output.

Any SMART-wizards here who could help me out?
Toast, being an inanimate object, obviously lacks both the ability and the desire to right itself.
-- Wikipedia on the "Buttered cat paradox"

beojan
Posts: 165
Joined: Mon May 23, 2011 12:11 pm UTC
Location: Oxford / London, United Kingdom, Europe

Re: Diagnose random freezes (Linux)

Postby beojan » Thu May 22, 2014 1:25 pm UTC

Install smartmontools, then run smartctl -H /dev/sda as root. It should give a simple PASSED / FAILED message.

J. Curwen
Posts: 15
Joined: Wed May 02, 2012 11:45 am UTC
Location: Germany
Contact:

Re: Diagnose random freezes (Linux)

Postby J. Curwen » Sat May 24, 2014 6:33 pm UTC

beojan wrote:Install smartmontools, then run smartctl -H /dev/sda as root. It should give a simple PASSED / FAILED message.


This gives a PASSED message.

I also did a full smartctl -a /dev/sda on the crashy machine (~4 years old) as well as my laptop (~2 years) and compared these two.

The current error number in the first one is at 16127 whereas on my laptop no errors have been logged.

If anyone is interested in the full SMART error-log of the crashy machine I can post ist.
Toast, being an inanimate object, obviously lacks both the ability and the desire to right itself.
-- Wikipedia on the "Buttered cat paradox"

beojan
Posts: 165
Joined: Mon May 23, 2011 12:11 pm UTC
Location: Oxford / London, United Kingdom, Europe

Re: Diagnose random freezes (Linux)

Postby beojan » Sun May 25, 2014 7:32 am UTC

Perhaps run a short, then an extended test first, and let those complete, before posting the full output.


Return to “The Help Desk”

Who is online

Users browsing this forum: No registered users and 5 guests