LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Hardware (http://www.linuxquestions.org/questions/linux-hardware-18/)
-   -   Computer freezes randomly (http://www.linuxquestions.org/questions/linux-hardware-18/computer-freezes-randomly-932735/)

odiseo77 03-04-2012 09:50 PM

Computer freezes randomly
 
Hello folks,

Specs:
OS: Debian Sid with kernel 3.2.0-1-686-pae
CPU: Core2Duo 7500 @ 2.93GHz
Mother Board: Asrock G31M-S R2.0
Memory: 2 sticks of 2 Gb each. (I think they work at 800 Mhz).
Graphics card: nVidia GeForce 7300 GT.

Problem:
Since about 2 or 3 weeks ago the computer freezes randomly and all of a sudden with no apparent reason (it has happened about 6 times during this period). I have checked /var/log/kern.log, /var/log/messages and /var/log/dmesg and they never show anything by the time the freezes have occurred. At first I suspected it could be caused by a defective memory stick, so I grabbed a LinuxMint CD and let memtest86+ run during 7 hours, but when I came back, it had not found any memory error. I haven't made any hardware or BIOS change recently and this is a computer I put together about 2 years ago (this didn't use to happen before; it was completely stable).

BTW, I'm not sure if this is a hardware problem or a software problem, but for the type of freeze it seems to be hardware-related (everything just freezes completely and suddenly, as I said before; if, for example, I'm listening music, the sound just keeps repeating itself, the pointer gets stuck at one point of the screen, etc). So far it has only happened on Debian Sid, but it's almost the only OS I use on this machine; though I have Windows and Ubuntu installed as well, I never used them.

How can I know what is causing these freezes? Should I let memtest86+ run longer (if so, how many hours)?

Thanks in advance for any help you can provide!

afreitascs 03-04-2012 11:04 PM

Use onboard video (could be the video card!)

see processor temperature
see psu
psu voltages ok?

business_kid 03-05-2012 02:34 PM

Are the voltage supplies right?

I like to see:
5V & VSB - 5V < supply < 5.3V
3.3 - Over 3.3 < supply < 3.5
The rest aren't too fussy - just in the ball park

odiseo77 03-05-2012 06:38 PM

Hi, thanks for your answers. I have a question: How do I check the PSU voltages? Directly in the PSU inscription or is there a way check them directly from Linux?

onebuck 03-05-2012 07:03 PM

Member response
 
Hi,

For PSU voltages it would be best if you use a DVM(Digital Volt Meter) with the PSU loaded.

I would first do a memtest86+ test on the RAM at least over night.

odiseo77 03-05-2012 07:24 PM

Hi,

Ok, will let memtest86+ run again (I'll try for 10 or 12 hours this time), and then I'll try to find a DVM to see if everything's fine.

BTW, I had completely forgotten: I did change the PSU about 3 months ago, but the problem only started happening like 3 weeks ago. The brand of this PSU is Delux (600 W).

I'll keep you posted of any news I find. Thanks!

kuser:) 03-05-2012 08:15 PM

Not sure if you can watch youtube videos, but here's a pretty good tutorial on how to test the PSU here:
http://youtu.be/rpzeTiqqjtg

ATX 2.2 specification
There should be two tests: one when the computer is idle, and one when it's working (PSU under load).

Don't forget you will be working with live electricity. Best to have someone else near you just in case.

odiseo77 03-05-2012 09:47 PM

Hi kuser. Thanks for the links (and the warning). I can watch youtube videos, I'll take a look later. I booted a CD with memtest86+ and will let it run for 12 hours to see if it finds some error, then I will try to find a DVM to measure the voltages.

Regards, and thanks again.

onebuck 03-05-2012 10:08 PM

Member response
 
Hi,

Be sure to use caution when making any measurements. One hand rule! Place your reference probe(power ground) via clip then be sure to use one hand to hold the other probe to measure. Do not ground yourself.

Be sure to have the pinout for the measurement points. ATX pinouts. Use caution! Always be safe.

odiseo77 03-05-2012 10:17 PM

Thanks onebuck, I'll keep your advices in mind when making the measurements. (I'll probably buy the DVM tomorrow or the day after, when I have some time).

Regards.

cascade9 03-06-2012 08:42 AM

If you've got away with running memtest for a 7 hours, I doubt its your RAM.

I'm suprised that nobody has suggested cleaning the CPU heatsink out. Overheating CPUs can cause random lockups, etc..

Quote:

Originally Posted by odiseo77 (Post 4619317)
Hi, thanks for your answers. I have a question: How do I check the PSU voltages? Directly in the PSU inscription or is there a way check them directly from Linux?

You can normally check the voltages from inside the BIOS. Its not as accurate as using a voltmeter, but it at least gives you a good idea of where the voltages are.

The only problem with using the BIOS voltages is that you cant check when you put the machine under load, which is where power problems are mostly likely to happen.

Quote:

Originally Posted by odiseo77 (Post 4619341)
BTW, I had completely forgotten: I did change the PSU about 3 months ago, but the problem only started happening like 3 weeks ago. The brand of this PSU is Delux (600 W).

Why did you change the PSU?

I'd be very careful with brands like 'delux'. Its better to have a good quality power supply than some cheap 'yum-cha' junk with a 600watt sticker on the side......

odiseo77 03-06-2012 05:38 PM

Quote:

Originally Posted by cascade9 (Post 4619858)
If you've got away with running memtest for a 7 hours, I doubt its your RAM.

I'm suprised that nobody has suggested cleaning the CPU heatsink out. Overheating CPUs can cause random lockups, etc..

You can normally check the voltages from inside the BIOS. Its not as accurate as using a voltmeter, but it at least gives you a good idea of where the voltages are.

The only problem with using the BIOS voltages is that you cant check when you put the machine under load, which is where power problems are mostly likely to happen.

Why did you change the PSU?

I'd be very careful with brands like 'delux'. Its better to have a good quality power supply than some cheap 'yum-cha' junk with a 600watt sticker on the side......

Hi,

I checked the heatsink and though it's a little bit dusty, I don't think it's as dusty as to cause any malfunction. I also used cpuburn (a cpu stress program) for about 5 or 8 minutes, checked the CPU temperature and they didn't rise too much during this time; from 38 C idle to around 52 C under load, although it probably would have rised more if I had let cpuburn run longer; (the machine didn't freeze during the stress test).

I also let memtest86+ run for 12 hours last night and it didn't find any error, so I guess we can discard the memory sticks from the list of suspects.

Now I'm starting to be suspcicious about the PSU. I replaced the one I had before because the machine just stopped working one day (it didn't even beeped, no leds, the PSU fan didn't spin, etc.); so I took this one I had bought like 2 years ago but had not used, connected it, and the PC worked again. Will have to do the voltage test these days to see what I find. BTW; in case I decide -or have- to buy a new PSU, which brands would you recommend?

Thanks for the help guys, I'll keep you posted.

sycamorex 03-06-2012 05:42 PM

Are you using NVidia's binary driver? You might want to look at this thread and H_texmex_h's comments.

http://www.linuxquestions.org/questi...ponent-922215/

odiseo77 03-06-2012 06:48 PM

Quote:

Originally Posted by sycamorex (Post 4620264)
Are you using NVidia's binary driver? You might want to look at this thread and H_texmex_h's comments.

http://www.linuxquestions.org/questi...ponent-922215/

Hi,

Yes, I'm using the nvidia binary driver available on the Debian repositories. I'm not on my PC right now to check, but according to this, the nvidia-glx package on Debian Sid is at version 295.20-1 at the moment, so I probably have this version installed. In fact, if memory doesn't fail me, the nvidia-glx package was upgraded to this version about 2 or 3 weeks ago, around the time the problem started happening, so it might very well be the driver causing issues. Now, in that thread you say you could type "reboot" to reboot the machine; can you switch to a virtual console when it happens? In my case I can't because the system is completely frozen and unresponsive when the problem occurs, so I can't switch to a virtual console in order to reboot.

sycamorex 03-07-2012 06:21 AM

Fortunately, it hasn't happened to me for over 3 weeks now so I can't verify it. As far as I remember that time it froze while I was in X and a terminal was open, music was playing, etc., so I typed 'reboot' in sakura and it worked (rebooted). In my case it was clear that it's not the system itself that froze, just the display. I've been using nouveau drivers for some time now and it works fine.


All times are GMT -5. The time now is 02:11 AM.