Computer freezing how to troubleshoot?

dabd · 2012-01-02 00:53:13

My laptop running arch linux 64 bits is freezing randomly to the point that it becomes completely unresponsive. No keyboard input works the only way is to turn it off from the power button. I am using Gnome desktop and I switched temporarily to XFCE and the same problem occurs.
I looked at dmesg, everything.log, errors.log and I see nothing abnormal. How can I troubleshoot it?

I upgraded today to the latest kernel 3.1.6-1 in the hope that this upgrade can solve this issue. I will also try to run memtest to see if RAM is fine.
Any tips would be appreciated.
Thanks.

Last edited by dabd (2012-01-02 00:53:58)

ewaller · 2012-01-02 02:34:09

When it freezes, are any keyboard indicators flashing? That would indicate a kernel panic.

Do you have your box set up to allow you to ssh into it?
When it hangs, can you ssh into it?

You say the keyboard is non-responsive. Have you tried getting to a text console by pressing Ctrl-Alt-F1 ?

dabd · 2012-01-02 02:48:42

ewaller wrote:

When it freezes, are any keyboard indicators flashing? That would indicate a kernel panic.
Do you have your box set up to allow you to ssh into it?
When it hangs, can you ssh into it?
You say the keyboard is non-responsive. Have you tried getting to a text console by pressing Ctrl-Alt-F1 ?

When it freezes keyboard is not flashing.
I did not try to ssh into it if it happens again I will try to do that.
I can't get a text console with Ctrl-Alt-F1: no keys are functioning.
The freezing has not happened since the upgrade to 3.6.1-1 kernel a few hours ago.
I also ran memtest86 and RAM is fine.

skottish · 2012-01-02 03:35:27

It's always helpful when dealing with hard crashes to mention what your GPU is and what video drivers that you have installed.

lagagnon · 2012-01-02 03:40:11

You can determine if it is a hardware fault by running a Live Linux distro via your optical drive or USB port. Stress things a bit and if you still get freezes you know for certain it is not an operating system fault but a hardware fault. If this is the case test memory, hard drive, CPU, in that order.

dabd · 2012-01-02 04:03:40

skottish wrote:

It's always helpful when dealing with hard crashes to mention what your GPU is and what video drivers that you have installed.

graphics card: Mobility Radeon HD 4500/5100 Series
I have the following video driver: xf86-video-ati 6.14.3-1

dabd · 2012-01-02 04:04:17

lagagnon wrote:

You can determine if it is a hardware fault by running a Live Linux distro via your optical drive or USB port. Stress things a bit and if you still get freezes you know for certain it is not an operating system fault but a hardware fault. If this is the case test memory, hard drive, CPU, in that order.

Thanks for the idea!

gborzi · 2012-01-02 07:47:06

I've experienced similar freezes on my system, which has a "Radeon HD 5000M Series" graphics card (from lspci), but only when I use the xf86-video-ati driver. When I use the closed source catalyst driver there is no freezes.

dabd · 2012-01-02 16:55:48

gborzi wrote:

I've experienced similar freezes on my system, which has a "Radeon HD 5000M Series" graphics card (from lspci), but only when I use the xf86-video-ati driver. When I use the closed source catalyst driver there is no freezes.

A little off topic question: is your HDMI video output working with the catalyst driver? I cannot get any video output using the xf86-video-ati driver. thx

sitquietly · 2012-01-02 19:02:48

dabd wrote:

My laptop running arch linux 64 bits is freezing randomly to the point that it becomes completely unresponsive. ..... How can I troubleshoot it?

I had a similar lockup on two different computers running 64-bit Archlinux, but with kde 4.7.x or xmonad desktops on an Intel Sandy Bridge system with Intel HD 3000 graphics and a ION Nvidia system with nouveau graphics. I made progress by starting to watch memory use. I saw that free memory on each system kept decreasing until at times there was no free memory left. When there is no free memory Bad Things Happen.

In my case the culprit that leaked memory was Firefox 8 or 9. It could run me out of memory on my Stargate computer, which has 16 GB of ram and stays up at all times. I switched to Chromium and that has solved the problem. I still keep an eye on memory use

It might help you to keep a system monitor running, in a panel or separate window. Start watching how much memory is in use. If you see it creeping up and up and up then you can start the process of narrowing the problem down to some specific software. It took me days of watching the memory graph to realize my problem was caused by firefox: I ran each of several DE for 24-hour periods and realized that the problem occurred under each DE. Then I let just one or two applications sit at idle for up to a day, watching memory. I could see that firefox was the problem and then let just firefox sit at idle with ALL extensions turned off. The longer firefox sat there doing nothing the more memory was used, up and up and up. I hope you have a similar success, I know that lockups aren't supposed to happen (!) and it's very wrong and frustrating. It is probably not the fault of Archlinux ... there are a lot of upstream bugs eating away at us.

gborzi · 2012-01-04 11:40:30

@dabd
I've never tried HDMI. There is one problem with the current catalyst driver (11.12). When you try to use the XV extension for video it crashes Xwindow. So, you'll need to configure your video player to use a different output, like opengl or xshm.
@sitquietly
That's not the problem, at the least in my case. I do run a memory monitor on xfce, and I've never seen the memory use rising before the freeze.

EDIT: Maybe this is due to powersaving, the wiki page on ATI says that "in some configurations it even causes system freezes". But I don't know how to disable powersaving.

Last edited by gborzi (2012-01-04 11:44:19)

toad · 2012-01-04 11:45:01

keep a terminal with top and iotop open - that should give you an idea also

Lone_Wolf · 2012-01-06 13:51:46

Next time this happens, try the numlock key .

This one is processed on a lower level then the other keys.

If numlock led changes, it means the kernel/processor are still responding to input and the problem is likely in an application.

rvdli · 2012-01-06 14:30:54

I am also experiencing this issue, but it happens randomly, not too often, since the first upgrade to a 3.1 kernel.
The issue happens only on my laptop, a Sandy Bridge i7 with an onboard Intel HD graphics card.

dabd · 2012-01-06 14:42:28

I have not experienced this since the upgrade to 3.1.6-1 and beyond.

rvega · 2012-01-22 15:52:08

I have the same issue. Intel HD integrated graphics. I'll enable ssh and try to remote from another computer when it happens again...

If it's something related to the graphics driver, what would be the right log to look at?

Linux arch-rafa 3.1.6-1-ARCH #1 SMP PREEMPT Thu Dec 22 09:11:48 CET 2011 x86_64 Intel(R) Core(TM) i7-2620M CPU @ 2.70GHz GenuineIntel GNU/Linux

meph · 2012-01-22 20:28:29

I'm going to state the obvious, just because nobody else did so far Keep an eye on temperature. A lot of lockup problems are caused by overheating. If I understood your description ".... is freezing randomly to the point that it becomes completely unresponsive" correctly (as in it's getting less and less responsive before it finally freezes for good), then I'd even say it points to a temperature problem. It's a bit more complicated on a laptop, but you can still check whether the fans are working, air intake is not obstructed and full of dust. When it finally freezes, you can touch the casing on various places to check whether it's not unreasonably hot - unreasonably as in almost can't keep your hand on it, while it's not healthy for them most laptops can function properly with very high temperatures, and overheating problems probably wouldn't occur with less than say 70-80°C.

There are a lot of things that could cause overheating, but it's too early to elaborate on them now, until you confirm overheating itself. And if you don't... well, then there's another thing cleared out of your way

edit: one more thing to add, make sure you're not using your laptop while it's placed on anything soft, e.g. bed sheets. Anything soft fills the gap between the rubber pads on the bottom side of your laptop and the casing itself. But that gap is important to keep the air flowing and maintain the laptop's ability to cool itself down.

Last edited by meph (2012-01-22 20:32:18)

rvega · 2012-01-28 00:59:12

In my case it's not the temperature, I have setup a fan control daemon and the sensors report normal temperatures all the time. Also, I can't connect through ssh when the freeze happens so it looks like it's not Xorg crashing.

@toad, what should I look for in the iotop terminal? It shows which processes are reading/writing to disks, right?
@lone_wolf I don't have a numlock key, does the capslock key work like that as well?

rvega · 2012-01-28 01:23:47

One more thing: when the computer freezes, I can hear the fan changing speed (It was spinning fast and started to spin slower) ¿ which means that at least the fan control daemon is working ? Still confused on how to troubleshoot this issue :s

Lone_Wolf · 2012-01-31 17:23:25

rvega wrote:

In my case it's not the temperature, I have setup a fan control daemon and the sensors report normal temperatures all the time. Also, I can't connect through ssh when the freeze happens so it looks like it's not Xorg crashing.
@toad, what should I look for in the iotop terminal? It shows which processes are reading/writing to disks, right?
@lone_wolf I don't have a numlock key, does the capslock key work like that as well?

Capslock should be useful also, yes.

toad · 2012-01-31 17:32:04

@toad, what should I look for in the iotop terminal? It shows which processes are reading/writing to disks, right?

Yep. Hm, not being able to ssh into your box is bad...

Also note precise time of the lock up and trawl the logs.

liveeeee · 2012-02-02 16:08:08

I'm having basically the same problem. Some days ago the system froze every few minutes. Today it only happened once and hasn't happened in a couple hours. I don't know how to reproduce the problem.

System:

Linux 3.2.2-1-ARCH #1 SMP PREEMPT Thu Jan 26 08:40:20 CET 2012 x86_64 AMD E-300 APU with Radeon(tm) HD Graphics AuthenticAMD GNU/Linux
00:01.0 VGA compatible controller: ATI Technologies Inc Wrestler [Radeon HD 6310]

The problem isn't temperature, memory or the hdd. Also, MS Windows is working fine.

Given the similarities, I believe the problem is between xorg and xf86-video-ati.

subatomic · 2012-02-08 08:47:19

My system started crashing around the end Nov/begin Dec, around the time of the 3.1.x kernel. I still can't put my finger on it, but we can eliminate Intel, ATI and catalyst as I am running AMD64 and nvidia. Over Xmas I decided to reinstall Arch so now I have a fresh 64bit install and besides KDE, minimal packages until this is solved. I also installed Ubuntu which runs a 3.0 kernel - no problems - hey, even my bluetooth mouse works again!

I ran intensive memory and HDD tests before installation and now keeping my eye on memory and temperature, but really nothing unusual going on there.

Sometimes it'll run for a couple of hours before a random action click will trigger it's death. Last night while I was reading this very page I hit Kmenu to open a console to get some system info and that was it, no response, no logs. This morning while editing a conf file in nano, a very intensive CPU task, it froze 3 times within 5 mins of uptime, so I am eliminating powersave as well. Sunday I watched a film and luckily made it through to the end until it suddenly froze during the credits! No mouse or keyboard actions.

akey · 2012-02-09 08:51:15

+1
My system was installed around Janurary 21, 2012. X86_64 on core 2 duo 6600 dual core system with 4G memory. Graphics card is 7300GT with close source nvidia driver 29x. I have tested memory and hardware and no fault is found from memory or hardware. I'm running xfce4 and most time when it crashes it is running chromium but sometimes it is not. I only have a wireless keyboard without led so there is no way to tell if numlock is working from LED.

Hope this can help narrow down the problem if there is any. I noticed there is no 32bit system involved, is there anyone who installed 32bit system and still have the problem.

BTW, I transist from ubuntu 12.04 x86_64 and also have random crash. Before that I use an arch32 bit which works pretty well (already erased). So my best guess is something with x86_64, second guess is nvidia driver but seem it is not with nvidia from this thread.

meph · 2012-02-09 10:12:29

One question. After the crash, when your systems are locked up, are you able to use your capslock/numlock/scrolllock keys? Obviously you can't test their functionality in the system, but do their respective LEDs on your keyboard work? Can you repeatedly turn them on and off again? It tells us something about the nature of a crash. For example, your system might be in an infinite loop because a device driver used a spinlock but interrupts are turned off.

There's a kernel option that enables kernel tracing. That means a custom compiled kernel. I vaguely recall the option is somewhere under kernel hacking, and is called detect soft lockups or something similar. If you use a kernel with this option enabled during the lockup, it will generate a trace showing what exactly was going on when your system crashed, and therefore which device/driver/whatever is the likely culprit.

However, there are 2 problems with this. First is, all crashes aren't similar and this will help just in those cases, where your system didn't really crash, and just hangs. Hence the numlock led question. But still it's a start, and even if this doesn't work, I have a few more ideas so we can eventually go back to it.
Second problem is that I haven't used this feature in a while, so it might have been moved someplace else, renamed or completely removed (highly doubt that). If you can't find exactly the name I gave you, then just keep looking or try and google it up, I'm sure something with more precise information will come up.

Arch Linux

#1 2012-01-02 00:53:13

Computer freezing how to troubleshoot?

#2 2012-01-02 02:34:09

Re: Computer freezing how to troubleshoot?

#3 2012-01-02 02:48:42

Re: Computer freezing how to troubleshoot?

#4 2012-01-02 03:35:27

Re: Computer freezing how to troubleshoot?

#5 2012-01-02 03:40:11

Re: Computer freezing how to troubleshoot?

#6 2012-01-02 04:03:40

Re: Computer freezing how to troubleshoot?

#7 2012-01-02 04:04:17

Re: Computer freezing how to troubleshoot?

#8 2012-01-02 07:47:06

Re: Computer freezing how to troubleshoot?

#9 2012-01-02 16:55:48

Re: Computer freezing how to troubleshoot?

#10 2012-01-02 19:02:48

Re: Computer freezing how to troubleshoot?

#11 2012-01-04 11:40:30

Re: Computer freezing how to troubleshoot?

#12 2012-01-04 11:45:01

Re: Computer freezing how to troubleshoot?

#13 2012-01-06 13:51:46

Re: Computer freezing how to troubleshoot?

#14 2012-01-06 14:30:54

Re: Computer freezing how to troubleshoot?

#15 2012-01-06 14:42:28

Re: Computer freezing how to troubleshoot?

#16 2012-01-22 15:52:08

Re: Computer freezing how to troubleshoot?

#17 2012-01-22 20:28:29

Re: Computer freezing how to troubleshoot?

#18 2012-01-28 00:59:12

Re: Computer freezing how to troubleshoot?

#19 2012-01-28 01:23:47

Re: Computer freezing how to troubleshoot?

#20 2012-01-31 17:23:25

Re: Computer freezing how to troubleshoot?

#21 2012-01-31 17:32:04

Re: Computer freezing how to troubleshoot?

#22 2012-02-02 16:08:08

Re: Computer freezing how to troubleshoot?

#23 2012-02-08 08:47:19

Re: Computer freezing how to troubleshoot?

#24 2012-02-09 08:51:15

Re: Computer freezing how to troubleshoot?

#25 2012-02-09 10:12:29

Re: Computer freezing how to troubleshoot?

Board footer