Fedora and the heat issue - please help if you can!

[GMorris]

Hi all,

It's been a little while since I had to post here, but this is starting to bug the crap out of me. I got a Dell laptop that had Windows XP installed on it, and it got beat to death by trojans and viruses despite "Windows Defender" and McAffee. (McAfee?). I got fed up after a few months and installed Linux. I looked and looked to find something kind of "in-between" super user system and almost complete idiot/total newbie, remembered how well Mandrake suited but it was gone or "morphed" by then. I saw that Fedora was part of RedHat and figured I'd give it a try. I've used Slackware, RedHat, Mandrake and from time to time other distros a little, and found Fedora to be a decent fit. Of all the things I like about it, YUM has to be the one I like best. As long as there's disk space, I don't have a problem with it. Of all the things I DON'T really like about it, the heat issue has to be the worst. The laptop RARELY, if ever got hot enough to shut down with Windoze, but it seems to do it all the time with Fedora. I'll be in the graphical interface (gnome 99% of the time), and all of a sudden it just throws up a message about the temperature and is going to shut down.

No warning, like "critical temperature approaching, save your work and shut down ASAP" or anything, it just goes to the splash screen like when it starts. No warning AT ALL. I put it on the empty box it came in, put a towel between it and the box, and at the edges between the towel and laptop I put an icepack on each side. That seemed to stretch out the time before crash a little, but it still does it. I don't know if working at the command line would warn me, but Gnome does not at all. It just stops and shuts down.

NOW, if there's a way for the computer to know the temperature, it has to be in the Linux kernel code or something as Windoze never even DID that. If so, I'd like to have a script or even program that could check it every so often (like every minute or two) so I could set up an alarm or something that lets me know well ahead of time that shutdown is imminent and unstoppable. Is there a temp file somewhere that has this temp, or is it closer to the system? All I want to do is monitor it and have a window pop up and let me know it is reaching critical point and will shut down SOON. That way, I don't have to save work every 30 seconds, or just lose it completely. There is no time or choice, it's going to shut down then, and I don't get any say in it!

Anybody else familiar with this, or just knows how to monitor that temp? I'd be SO grateful, and I could handle the script/program (whatever is necessary) to do it. This is VERY annoying at the very least, catastrophic at worst!

Thanks,

G.

[BoDiddley]

What I found is that Linux uses processors governors, whether it be a generic one - or the one that could accompany your motherboard. I am AMD. Linux was using the processor governor on my system, and it ran like crap. Since I am ReiserFS, after careful research - I found that CPUDYN was a more useful governor. Now my variable speed system fan usually runs on low.

I have two processor speeds, one processor. The governor - based on system demand alternates between the two speeds. Seems to me that you might be stuck on "high", and simply need to find a more suitable governor.

I would also not be surprised to find that Windows is encouraging this glitch you are experiencing to battle Linux Operating Systems.

You could also, simply - have a bad system.

[GMorris]

I would also not be surprised to find that Windows is encouraging this glitch you are experiencing to battle Linux Operating Systems.

You could also, simply - have a bad system.

Not quite sure what you mean by that, but there IS no Windows on this machine anywhere. I have an external drive which I copied important files to right before I COMPLETELY wiped Win, reformatted and partitioned, so there shouldn't even be any remnants of Windows on the local drive.

I know I've seen system files that change depending on the state of the machine at the time. I've seen the processor speed, network performance and a plethora of others, I was just hoping there would be some place I could check it. I'd like to write a little program with an icon I can use to check the current system temperature or be warned when it's getting too hot, or even find one somebody else did already. The only thing is, I don't know where to find that temperature so I can keep an eye on it. I'd consider just about anything that would work!

[GMorris]

Hello all,

I know it's been a while since this thread started, but I have found something that helps some. Somebody posted on another board about ACPI! I found that 'acpi -t' gives me A temperature, I'm just not sure WHAT temperature. Does anybody know enough about this to say what that temp is? I also read that there is more than one temperature, so now I'm really confused! I can't figure out why there are more than one, or if I'm getting the right one from 'acpi -t'.

If it turns out that this IS the one I'm after, anybody have any suggestions as to how to go about this? I mean, should I write a script, a compiled program, or what? Should I try and use cron to make this run every minute or two, or is there a better way? I don't even know if there are many options. All I want to do is have the script/program, whatever, check it every so often, then when it gets up over a certain temp, pop up a warning that the temperature is too high and then I'd have time to save everything and go ahead and shut down naturally rather than forced. Right now I get around 55 - 60 degrees C, but when it shuts down like that sometimes I'll see a message REAL quick that goes away almost instantly, and it usually has above 95C. Any suggestions appreciated.

[BoDiddley]

That is good info, thanks. acpi -V shows everything, and tells you your trip points. Any command can become part of a script. I am presently learning Bash,myself. Thanks again.

[BoDiddley]

You said you got a Dell laptop, was it new? Long story short.... My notebook was purchased in 2004. About a year ago I started having the same problem. I brought a new fan assembly, that - as it turns out I never needed.

What I did was unclog the wad of dust in the fan (internally) - and re-seat the heat sink (on top of the processor), applying fresh thermal grease - and it has been fine ever since. However, this is risky - static discharge can blow your processor.

You might try something simple like a can of compressed gas designed to eat dust, and use it in your ventilators.

The really odd thing is this does not happen, to you - on Windows.

[GMorris]

Yeah, I thought it was a pretty handy command to have around. I've played around with it at the command line some, setting a loop and then "sleep 2" before the acpi -t command to give it a pause before the next reading, just for fun. I could likely write a small script in Bash that would do a longer "sleep", with an unending loop, thought not sure if that itself would contribute to the heat issue. The only thing that I'm stuck on now is, 1) is that the temp I want to monitor? and 2) how do I send this message to gnome when the temp is getting close (trip point on this one said 98C, I'd like to catch it before that anyway)?

I suppose I might use something like TCL, but maybe somebody knows an easier way? Thank, y'all are great!

[atreyu]

Just a thought: you could install the lm_sensors package (should be in your yum repo). Provided that your machine comes with sensor chips (most current ones have some kinds of sensor chips), this package will read and report what these sensors have to say (CPU and ambient temperature, fan speed, voltage, etc.). You have to install the package, then run:

 sensors-detect

This script will run thru an interactive probe of your system, load appropriate kernel drivers, and try to identify your chip(s). When it is done, it will hopefully have generated an lm_sensors config file (probably /etc/sensors3.conf), which you may have to modify/calibrate.

Once your done with that, you can run:

sensors

and a lot of good info will be sent to the terminal.

The program can be run as a daemon, too, though you could just cron calls to 'sensors' and parse the output if you wanted.

There is also the hddtemp tool, part of a package by the same name, which is handy for reporting hard drive temperatures, e.g.:

hddtemp /dev/sda