[REOPENED] Auto-Rebooting System

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

Recently my system started to reboot instantly without any kind of warning. I don't know what it can be related, 'cause I don't know if there is any kind of log where kernel may register this kind of thing, so the only thing I could do is to check every piece I could find of my system.

I noticed that it crashes only when it is performing a hard task, like compiling, or a hard math calculation (I use matlab frequently). For any other meanings, its running fine.

First I thought it has something to do with kernel. I performed tests with 4 diferent kernel versions: zen-2.6.29, zen-2.6.30-rc5, gentoo-2.6.26-rc4, gentoo-2.6.28.5. All of them presented the same behaviour.

Then I considered any kind of weird bug with composition or graphical related stuffs. I tried to run compilations into pure console, and got the same behaviour.

The crashes are kinda random... it does not have a approximate time to happen, it just happens sometime if I let my computer doing hardwor, sometimes in 4min, sometimes in 4s, sometimes in 4h. So as it does not have a pattern I considered it a memory fail... but I don't know how to test the memory.

Appreciate any help.
_________________
"Nolite arbitrari quia venerim mittere pacem in terram non veni pacem mittere sed gladium" (Yeshua Ha Mashiach)

eccerr0r · Posted: Fri May 22, 2009 9:31 pm Post subject:

Some motherboards are designed to do something when the processor overheats. Might want to check if this is happening. If you can underclock your cpu, that's another thing to try.

Usually RAM errors don't behave like this. Normally it's motherboard or possibly CPU issues.

You can download and test memtest86+ to check your RAM. It's a good idea to check it anyway.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

AaronPPC · Guru Joined: 29 May 2005 Posts: 522 Location: Tucson, AZ

I have had 3 computers that had CPU problems and they exibited symptoms very similar to yours.
_________________
--Aaron

ronmon · Posted: Fri May 22, 2009 11:37 pm Post subject:

I'll bet that it's heat. Mine started doing this a few weeks ago when the weather started getting warmer here in the sub-tropics. Watching in gkrellm, I saw my CPU temp go up to almost 70C under heavy loads.

Going into BIOS and dropping the CPU voltage from 1.35 to 1.30 did the trick. Just .05 volt difference, and now it never goes over 40C.
_________________
Ask Questions the Smart Way - by ESR

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

I didn't think it was the heat 'cause I've never overclocked my CPU. But when you guys mentioned the BIOS I remembered that I've enabled some "fancy" options in there, like CPU Fan warnings that are disabled by default. "Load Optimal Defaults" fixed. Maybe this kind of "Warning" signal was confusing the kernel?

Anyway, the temperature sensors are not found here even though I enabled it in kernel... where am I supposed to find them?
_________________
"Nolite arbitrari quia venerim mittere pacem in terram non veni pacem mittere sed gladium" (Yeshua Ha Mashiach)

ronmon · Posted: Sat May 23, 2009 11:08 am Post subject:

Mine wasn't overclocked either.

You need to emerge lm_sensors, start it and set it to start automatically on boot.

entrophie · n00b Joined: 24 Aug 2006 Posts: 13

Holysword: you don't need to overclock your CPU to get overheating. Some silver contacts beetwen cpu and radiator tends to leak. So after you check the temperature with sensors, you can try to improve the contact.

szczerb · Posted: Tue May 26, 2009 6:59 am Post subject:

First of all check if your radiator is not full of thick dust that very well stops the airflow.

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

Thank you guys for answering, but I still don't think its overheat. I've checked the temperature with sensors and it seems okay.
_________________
"Nolite arbitrari quia venerim mittere pacem in terram non veni pacem mittere sed gladium" (Yeshua Ha Mashiach)

eccerr0r · Posted: Sun May 31, 2009 1:28 pm Post subject:

I guess if you rule out all the normal reasons that cause reboot then the only things that remain are the non-normal...

and that are

1. Hackers. But very unlikely.

2. Your hardware is broken. You need to buy a new power supply most likely, or possibly motherboard.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

I still don't think its any of those. As I stated before, I disabled some non-default bios options and the problem has gone.
_________________
"Nolite arbitrari quia venerim mittere pacem in terram non veni pacem mittere sed gladium" (Yeshua Ha Mashiach)

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

Turns out that the problem came back, but more frequently.

Again, its not about the temperature (I still check the temperature sensors, nothing odd. Sometimes it crashes at 47°C). The difference is that now I don't need to be doing something aggressive, which makes this problem even more annoying.
I'll try to install the memtest86+ into that grub fancy way and check the memory, and I'll post here the results.
_________________
"Nolite arbitrari quia venerim mittere pacem in terram non veni pacem mittere sed gladium" (Yeshua Ha Mashiach)

rjw8703 · Posted: Wed Jun 10, 2009 2:34 pm Post subject:

I had this problem a while back. It turned out to be the voltage regulators on the m/b were dying intermittently. Replacing the m/b fixed the problem.

Gusar · Posted: Wed Jun 10, 2009 3:13 pm Post subject:

Faulty power supply would be my guess. I had a problem kinda like that. After an hour or so, the machine would reboot. And then it would reboot every few minutes. Leaving it off for a while, it would work for an hour again. Changing the power supply made the problem go away.

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

eccerr0r · Posted: Sat Jun 13, 2009 7:51 am Post subject:

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

Well, while I don't find someone to test my motherboard, I was checking dmesg and I suddenly realized that from time to time it complains about the voltage of in5 being 0. Follows the relevant part of "sensors":

eccerr0r · Posted: Thu Jun 25, 2009 11:32 pm Post subject:

Well that proves to show that lm-sensors is unreliable, and nothing else. There are no transistors that will work at 0V so obviously something's wrong with the detection there, or that input is simply unused. Or perhaps your PSU doesn't supply -5V as most mb's don't use it nowadays, and it's perfectly fine for it to be at 0V.

Not only that, likely the multiplier constants in your lm-sensors.conf do not match your motherboard and is producing really unreliable results - where's your 12V line? Where's your -12V line? Which line is which?

"ALARM" just means the chip detected a number from the voltage line that was outside its bounds. But what are the bounds? The bounds are set up by software which once again may or may not match your motherboard to produce valid results.

I'm sorry, the cheapest way is to buy a new MB or PSU to test. You might be able to get away with a multimeter but again, multimeters also tend to be slow and won't detect fast glitches in your power. There's no other way you can really tell for sure. The on board sensor chips are not only inaccurate, but also slow - at most a few samples every second versus millions of samples for a DSO.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

Holysword · l33t Joined: 19 Nov 2006 Posts: 946 Location: Greece

eccerr0r · Posted: Mon Jun 29, 2009 10:57 pm Post subject: