I've been running arch linux for quite a few months and it has been the best distribution i've used. However, sometimes, it just decides to freeze. I may be using the computer, the bottom bar shows the CPU load is at 0.30, perfectly normal, then all of sudden it starts going up, 1, 3, 5, 10 and goes beyond 20 if I let it. The whole system becomes unresponsive and I can't even kill X. I'm pretty sure it's my browser the one which is doing that, since when I realize the load is going up, I can kill it and everything goes back to normal. However this happens both with firefox and chromium, which leaves me thinking it might be a flash bug, however, everything is up to date and I don't see anyone else complaining about it. Is there some way I can profile it to make sure what is the process taking over the CPU?

  • 1
    Be sure to do a memory test. Memory and disk are the two components that fail the most often, and this doesn't look like a disk failure. – Gilles Mar 12 '14 at 23:39
  • @Gilles o_O? In fact it looks exactly like a disk failure. They can take a long time on their way out. A memtest is always worthwhile in any case. – goldilocks Mar 13 '14 at 2:46

Sometimes the CPU cooler is not properly mounted, so you need to check the CPU temperature.

If you have some issues and you don't know what it is, then the best approach is the black-box testing. To do that, remove some components, sound-cards, drives, GPU and so on. If you still have issues try another OS.


Sounds like hardware I/O failure; the reason the load average goes up is because multiple process are trapped in uninterruptible sleep. It also tends to max out the CPU1 because the kernel is busy looping trying to access a hardware device that is malfunctioning.

In short, the kernel does not deal with the situation very well because there not much point in trying to compensate for failing hardware on a low level. You will probably find lots and lots of evidence of the problem if you look in the system logs afterwards. Chances are you have a hard drive with some bad blocks on it; this may be repairable, although you have not described the problem in sufficient detail2 for me to offer an opinion.

If the problem is bad blocks, you can check decisively by running e2fsck -c on all the (ext) partitions you normally have mounted; they'll have to be unmounted so you may need to use a live CD for this. An extra -c (i.e., -cc) and -y could be used too, although the former will require quite a bit of time. Make sure you look at man e2fsck so you understand what this is about.

1. Note that CPU usage and load average are not the same thing, although the way this is reported by some tools can be confusing.

2. How long has this been recurring, how often does it happen, how long does it last.

  • Well, this has been happening some months by now. It usually happen once or twice a day(the computer is used about 12~14 hours a day), and the only way to recover is rebooting. Which logs should I check? – Leonardo Arroyo Mar 12 '14 at 22:54
  • The system logs, i.e. the ones produced by the system logger, syslog (or rsyslog). The kernel logs to there, init (systemd) processes log to there, and most everything else logs to there too. They're almost always in /var/log, but what gets placed in which file depends on your (r)syslog configuration -- now is the time to start learning about that, evidently. Usually there's one that catches everything, possibly /var/log/system. – goldilocks Mar 13 '14 at 0:10
  • @Sonar I've added a third paragraph WRT testing this premise. – goldilocks Mar 13 '14 at 3:22

What graphics chipset do you have? Newer kernels have a huge problem with some Intel graphics chipsets, like Intel 82845G "Brookdale". Do this:

lspci -v | grep -i -A8 VGA

Then google for "Arch Linux" and whatever chipset you find.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.