3

I have a problem on my Ubuntu (11.10 still) which causes my machine to do a critical shutdown. I identified the problem as an overheating issue, seeing the following line in /var/log/syslog:

kernel: [ 1815.505465] Critical temperature reached (128 C), shutting down.

Now I'd like to know what's causing this overheating problem. It has happened several times already, but the circumstances are not yet identifiable for me. Sometimes I'm watching a movie, somtimes just browsing the web... It could have something to do with a segfault thrown by Opera, but I'm not sure. But this was the last syslog line before the nose dive:

operapluginwrap[2613]: segfault at 506 ip 00007fab1814bf40 sp 00007fff6ff89258 error 4 in libgobject-2.0.so.0.3000.0[7fab18119000+4e000

Anyway, my question:
Is there a way to make the system alert me in case of an overheat, maybe some 5-10 degrees under the actual overheat limit?

This would probably give me the chance to identify the process which is causing the massive load/heat... Mind, I don't want to stop the machine from going down if it has to - I rather suffer some crashes than having my CPU go up in smoke...

Nota Bene: I am trying to identifiy a rogue process/thread. I'm not (yet) interested in preventing the overheat itself.

Improve this question
7
  • 3
    Your CPU is overheating so perform the normal steps that you would to deal with an overheating CPU. Start by cleaning the computer, then if its a desktop, apply new thermal paste.
    – Ramhound
    Nov 19, 2012 at 19:57
  • The installation is as clean as possible without a full reinstall. The machine is a Lenovo ThinkPad T400 - it served me well over years bearing high loads. Suggesting thermal paste is really not helpful. Also there is no reason on earth why a browser, a video client or an office application should overheat a 2 GHz dual core machine.
    – fgysin
    Nov 19, 2012 at 20:37
  • If a laptop CPU is REALLY reaching 128C, there is an obvious issue somewhere, and an alert would just be a workaround. Check your power settings, to make sure you're not on some "always 100%" setting. Also make sure your fans are working and are dust-free. And why dismiss the thermal paste comment? These are just standard maintenance steps.
    – trpt4him
    Nov 19, 2012 at 20:53
  • The point is, that I can run my CPU at seemingly 100% for several minutes with no overheating whatsoever (it stabilises at a high, but sustainable 97-100C); in idle mode the temp is around 40C with basically no CPU load. The reason for me dismissing the thermal paste solution is simply that it has nothing to do with my question. I asked for the possibility of an alert/message in case of high temperature. I did NOT ask how to prevent my CPU from overheating.
    – fgysin
    Nov 19, 2012 at 21:04
  • Off the top of my head, you could probably use lm_sensors, poll the appropriate value, then compare it with the value you're comfortable with. If no one else can answer, I'll try hacking at it over the weekend.
    – Journeyman Geek ♦
    Jan 23, 2013 at 12:11

1 Answer 1

Reset to default
2

Try installing the cpu frequency indicator:

sudo apt-get install indicator-cpufreq

This will install an icon in the appmenu, here you can click and toggle your CPU to more conservative settings; see if this helps.

Or you can try Psensor from Ubuntu Software Centre.

If there is no improvement, maybe check out this resource:

https://wiki.ubuntu.com/Kernel/PowerManagement

For example "Indicator-Weather" is problematic because it constantly writes to a log file, stopping the drive from becoming fully idle in power saving mode, causing heat, triggering fans, etc. This is mentioned in the research. Try to find apps that do this and uninstall them.

Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .