[Solved] System freezes, how can I troubleshoot it?

Gumper · 2009-12-02 01:06:59

I've been having troubles with my system randomly locking up. When it does this the mouse doesn't work and neither does the keyboard. If I have music playing at the time the music starts studdering the same thing over and over again. The only way to fix is to reboot with the hard reboot button. I'm not even able to power it down by holding the power button for several seconds.

What steps can I take to figure out what is causing my issue? I know that there have been issues with Xorg crashing, but how can I determine if this is the same problem that I'm having or if it's a kernel panic or something else?

I'm running Arch x64 with Kde 4.3.3, xorg-server 1.7.2-1 and an ATI video card.

Thanks for any help.

Last edited by Gumper (2009-12-22 23:27:02)

jmss · 2009-12-02 01:13:01

One simple test can be:

pacman -S memtest86+

and then you get a boot entry from which you can test your RAM.

Gumper · 2009-12-02 01:50:17

jmss wrote:

One simple test can be:
pacman -S memtest86+
and then you get a boot entry from which you can test your RAM.

I guess it could be the memory but I don't have any problems when running in windows.

alphe · 2009-12-02 06:11:48

I have the same issue with the same configuration as yours.

My video card is:

lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc RV350 AP [Radeon 9600]

I have already run memtest without any error.

Nothing in logs when it is happened.

The only thing that I have noticed until now is that it happens only when the amarok is running no matters if it plays music or no.

Gumper · 2009-12-02 11:41:02

alphe wrote:

I have the same issue with the same configuration as yours.
My video card is:
lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc RV350 AP [Radeon 9600]
I have already run memtest without any error.
Nothing in logs when it is happened.
The only thing that I have noticed until now is that it happens only when the amarok is running no matters if it plays music or no.

Mine freezes at different times. I first thought that it only did it when I was using Firefox but last evening it did it while using Kontact. I'm thinking that it has something to do with the Xorg server seeing there has been other postings about problems with it lately. I'd like to try downgrading it but I'm not sure what all I have to downgrade.

drelyn86 · 2009-12-02 14:50:44

alphe wrote:

I have the same issue with the same configuration as yours.
My video card is:
lspci | grep VGA
01:00.0 VGA compatible controller: ATI Technologies Inc RV350 AP [Radeon 9600]
I have already run memtest without any error.
Nothing in logs when it is happened.
The only thing that I have noticed until now is that it happens only when the amarok is running no matters if it plays music or no.

We are the Arch community and we DO NOT SUPPORT THE FGLRX DRIVER!!!

lol, just kidding! I just hated my HD3850 a lot.

Actually, what video driver do you have running? I do remember having similar crashes with my 3850 while playing games in Wine with the FGLRX driver.

One thing you might try - I never got to do this simply because I didn't think of it until I upgraded to the nvidia driver ... do you have an OpenSSH server running on this machine? You could try to remote in and diagnose the issue from there. I'm really not sure what you should be looking for if your log files are clean... do a "ps -A" or maybe even top... I'm out of ideas, but if you can SSH into it while the crash has taken place, that would be a good start.

Last edited by drelyn86 (2009-12-02 14:51:42)

alphe · 2009-12-02 15:44:17

Gumper wrote:

Mine freezes at different times

I had two freezes without to do anything (screensaver, DPMS). So has nothing to do with the app in use.

Gumper wrote:

I'm thinking that it has something to do with the Xorg server seeing there has been other postings about problems

I also suspect Xorg and KMS but I have not find any similar symptoms with our configuration and that's why I reply to your post

Gumper wrote:

I'd like to try downgrading it but I'm not sure what all I have to downgrade

I have read a recent post that gives a guide to downgrade Xorg but I cannot find it right now. I will try to debug it so I will not downgrade for the moment.

drelyn86 wrote:

We are the Arch community and we DO NOT SUPPORT THE FGLRX DRIVER!!!

Who said anything about fglrx? I am using the open-source radeon driver
I thought that this was the driver that drelyn86 uses also. If it is not I will stop to post in this thread because we do not have the same configuration.

drelyn86 wrote:

but if you can SSH into it while the crash has taken place, that would be a good start.

I have think about it but I have to setup SSH first which is in my to-do list but for later on. I have try to access it using samba but it was not accessible.

tavianator · 2009-12-02 16:27:35

A couple tests to see if it's a kernel crash:
- The easiest test is pressing numlock. If your indicator light toggles, then the kernel is working.
- You should enable the magic sysrq key in /etc/sysctl.conf. Then the key combination Alt+SysRq+{R,E,I,S,U,B} will reboot your machine after flushing and unmounting your disks, provided the kernel hasn't crashed.

alphe · 2009-12-02 16:39:35

tavianator wrote:

A couple tests to see if it's a kernel crash:
- The easiest test is pressing numlock. If your indicator light toggles, then the kernel is working.
- You should enable the magic sysrq key in /etc/sysctl.conf. Then the key combination Alt+SysRq+{R,E,I,S,U,B} will reboot your machine after flushing and unmounting your disks, provided the kernel hasn't crashed.

Both tested none works.
Thanks

tavianator · 2009-12-02 18:39:32

Try toggling whether KMS is enabled ("options radeon modeset=<0 or 1>" in /etc/modprobe.d/modprobe.conf).

james_earl · 2009-12-02 19:39:58

tavianator wrote:

A couple tests to see if it's a kernel crash:
- The easiest test is pressing numlock. If your indicator light toggles, then the kernel is working.
- You should enable the magic sysrq key in /etc/sysctl.conf. Then the key combination Alt+SysRq+{R,E,I,S,U,B} will reboot your machine after flushing and unmounting your disks, provided the kernel hasn't crashed.

Good info on numlock there, didn't know that.

I've been dealing with a similar situation, but I can cause it by loading up a game with Wine. I found that disabling the Composite extension in my xorg.conf lowered the frequency of the crashes.

Card is a HD4850 X2, using X 1.7.1, E17, 32bit Arch, 2.6.32rc8. I'm using the open drivers for my card, built 2 days ago from git.

So I guess chances are good that it's either a kernel or video driver issue. I was going to update X today to latest, will check back in with any progress.

tavianator · 2009-12-02 20:15:27

james_earl: Updating xorg-server is a pain and usually not worth it. I'd update the DRM module first (by compiling kernel26-git from the AUR), then mesa, libgl, and ati-dri (there's a split mesa-git PKBUILD in the comments for mesa-git in the AUR).

Yeah my grade 4 school tech guy taught me the numlock trick, and it's been pretty useful ever since.

jithin1987 · 2009-12-03 01:07:58

When you system freezes and its an X freeze. You can ssh to your box and look and xorg logs situated in /var/log/Xorg.0.log

stryder · 2009-12-03 02:44:55

tavianator wrote:

A couple tests to see if it's a kernel crash:
- The easiest test is pressing numlock. If your indicator light toggles, then the kernel is working.
- You should enable the magic sysrq key in /etc/sysctl.conf. Then the key combination Alt+SysRq+{R,E,I,S,U,B} will reboot your machine after flushing and unmounting your disks, provided the kernel hasn't crashed.

If the keyboard already froze surely these tests will be meaningless?

tavianator · 2009-12-03 03:14:25

stryder: Just because X isn't responding to the keyboard doesn't mean the kernel isn't. Especially regarding the magic sysrq sequences. These are indispensable for saving otherwise frozen boxes.

Gumper · 2009-12-03 03:38:08

tavianator wrote:

A couple tests to see if it's a kernel crash:
- The easiest test is pressing numlock. If your indicator light toggles, then the kernel is working.
- You should enable the magic sysrq key in /etc/sysctl.conf. Then the key combination Alt+SysRq+{R,E,I,S,U,B} will reboot your machine after flushing and unmounting your disks, provided the kernel hasn't crashed.

Thanks for the magic sysrq key, I wasn't aware of that. I'm sure it will be much better to use this than to do a hard reboot like I've been doing it. File systems don't seem to like doing it the way that I was doing it.

My box has been on for a couple of hours tonight without any problems. I'm sure it will be back though. I think that I'll try the new xorg that is available and if that doesn't work I'll try to disable "KMS" to see if that helps.

jithin1987 · 2009-12-03 04:08:19

If it happens again its always good to ssh to ur box from some other machine and collect logs. May be you can file a bug report with it.
But you will need one more machine connected to your network.

Gumper · 2009-12-03 11:21:40

jithin1987 wrote:

If it happens again its always good to ssh to ur box from some other machine and collect logs. May be you can file a bug report with it.
But you will need one more machine connected to your network.

Unfortunately this is my only machine.

stryder · 2009-12-04 14:29:03

tavianator wrote:

stryder: Just because X isn't responding to the keyboard doesn't mean the kernel isn't. Especially regarding the magic sysrq sequences. These are indispensable for saving otherwise frozen boxes.

Thanks. Have always done a hard reset when I can't even switch to console. Hopefully will remember the sysrq sequence.

leeyee · 2009-12-05 09:42:25

alphe wrote:

Gumper wrote:
Mine freezes at different times
I had two freezes without to do anything (screensaver, DPMS). So has nothing to do with the app in use.

I had a same issue before with Intel driver. However, it seems that add

xset -dpms

into your x starting up configuration file(such as .xinitrc, .xprofile or whatever it is on your computer) will avoid this problem. Just to disable DPMS. I'm not sure if it is still working, since I haven't met with this problem for a long time.

Gumper · 2009-12-09 01:51:46

Here I thought my problem was fixed but last night my box locked up again. It was fine for about 5 days. Arrgghh...:mad:

When this happened I wasn't even able to use the magic sysreq keys, they had no effect at all. If I recall correctly, my numb lock did respond, at least initially and then that stopped working as well. I'm using a wireless mouse / keyboard, not sure if that matters or not.

I see that xorg-server 1.7.3-1 is out, perhaps that will help.

jithin1987 · 2009-12-09 04:03:53

I think magic ssyreq keys are disabled by default.

To enable it back. Edit file /etc/sysctl.conf and change kernel.sysrq to 1

Gumper · 2009-12-10 11:39:47

jithin1987 wrote:

I think magic ssyreq keys are disabled by default.
To enable it back. Edit file /etc/sysctl.conf and change kernel.sysrq to 1

I did have this enabled and I tested it earlier so I know it works. Unfortunately it didn't work when my system locked up.

jithin1987 · 2009-12-10 13:19:13

Hmm that sounds like a kernel panic. Could you check /var/log/kernel.log or /var/log/messages.log and see of there is any error logged during he time your system froze.

graysky · 2009-12-13 20:27:18

I just had a series a lock ups as well. IN my case my system would fail memtest in about 6 sec. After reseating the memory, all is well. Odd. Thanks for the magic key tip. Learn something new every day.

Last edited by graysky (2009-12-13 20:37:01)

Arch Linux

#1 2009-12-02 01:06:59

[Solved] System freezes, how can I troubleshoot it?

#2 2009-12-02 01:13:01

Re: [Solved] System freezes, how can I troubleshoot it?

#3 2009-12-02 01:50:17

Re: [Solved] System freezes, how can I troubleshoot it?

#4 2009-12-02 06:11:48

Re: [Solved] System freezes, how can I troubleshoot it?

#5 2009-12-02 11:41:02

Re: [Solved] System freezes, how can I troubleshoot it?

#6 2009-12-02 14:50:44

Re: [Solved] System freezes, how can I troubleshoot it?

#7 2009-12-02 15:44:17

Re: [Solved] System freezes, how can I troubleshoot it?

#8 2009-12-02 16:27:35

Re: [Solved] System freezes, how can I troubleshoot it?

#9 2009-12-02 16:39:35

Re: [Solved] System freezes, how can I troubleshoot it?

#10 2009-12-02 18:39:32

Re: [Solved] System freezes, how can I troubleshoot it?

#11 2009-12-02 19:39:58

Re: [Solved] System freezes, how can I troubleshoot it?

#12 2009-12-02 20:15:27

Re: [Solved] System freezes, how can I troubleshoot it?

#13 2009-12-03 01:07:58

Re: [Solved] System freezes, how can I troubleshoot it?

#14 2009-12-03 02:44:55

Re: [Solved] System freezes, how can I troubleshoot it?

#15 2009-12-03 03:14:25

Re: [Solved] System freezes, how can I troubleshoot it?

#16 2009-12-03 03:38:08

Re: [Solved] System freezes, how can I troubleshoot it?

#17 2009-12-03 04:08:19

Re: [Solved] System freezes, how can I troubleshoot it?

#18 2009-12-03 11:21:40

Re: [Solved] System freezes, how can I troubleshoot it?

#19 2009-12-04 14:29:03

Re: [Solved] System freezes, how can I troubleshoot it?

#20 2009-12-05 09:42:25

Re: [Solved] System freezes, how can I troubleshoot it?

#21 2009-12-09 01:51:46

Re: [Solved] System freezes, how can I troubleshoot it?

#22 2009-12-09 04:03:53

Re: [Solved] System freezes, how can I troubleshoot it?

#23 2009-12-10 11:39:47

Re: [Solved] System freezes, how can I troubleshoot it?

#24 2009-12-10 13:19:13

Re: [Solved] System freezes, how can I troubleshoot it?

#25 2009-12-13 20:27:18

Re: [Solved] System freezes, how can I troubleshoot it?

Board footer