CPU overheats, Linux slows down, need some ideas.

Something happened on my server that made several processes do
intensive computations. (a bug in some of my perl scripts) That loaded
all cores of my two dual cored CPUs at 100% CPU. I got these messages
in /var/log/messages and a lot of "wall" messages:

Sep 5 13:17:09 ak74 kernel: CPU1: Running in modulated clock mode
Sep 5 13:17:18 ak74 kernel: CPU1: Temperature above threshold
Sep 5 13:17:18 ak74 kernel: CPU0: Temperature above threshold
Sep 5 13:17:18 ak74 kernel: CPU0: Running in modulated clock mode
Sep 5 13:17:18 ak74 kernel: CPU1: Running in modulated clock mode
Sep 5 13:17:27 ak74 kernel: CPU0: Temperature above threshold
Sep 5 13:17:27 ak74 kernel: CPU1: Temperature above threshold
Sep 5 13:17:27 ak74 kernel: CPU1: Running in modulated clock mode
Sep 5 13:17:27 ak74 kernel: CPU0: Running in modulated clock mode
Sep 5 13:17:38 ak74 kernel: CPU0: Temperature above threshold
Sep 5 13:17:38 ak74 kernel: CPU1: Temperature above threshold
Sep 5 13:17:38 ak74 kernel: CPU1: Running in modulated clock mode
Sep 5 13:17:38 ak74 kernel: CPU0: Running in modulated clock mode
Sep 5 13:17:44 ak74 kernel: CPU1: Temperature above threshold
Sep 5 13:17:44 ak74 kernel: CPU0: Temperature above threshold


So I think that Linux saved my ass and switched to modulated clock
mode instead of shutting down or burning my CPU. It simply slowed
down.

Thanks for that.

My question is, normally a computer can be very busy and should not
overheat. So something physical is wrong. Would you have any idea,
maybe I need to replace case fans or the CPU fan?

This is a 1U rackmount from Rackmounts Etc.

I called my ISP and they confirmed that they did not have any cooling
or heating issues on their racks.

thanks

i

For what it's worth, my new Dell Vostro 1500 runs very hot on Ubuntu, but okay on Windows XP Home.
I think fan management isn't perfect yet.
--
(E-Mail Removed)

On the internet, nobody knows you're a jerk.

On Wed, 05 Sep 2007 17:00:56 -0400, Ron Hardin <(E-Mail Removed)> wrote:
> For what it's worth, my new Dell Vostro 1500 runs very hot on Ubuntu, but okay on Windows XP Home.
> I think fan management isn't perfect yet.

It probably runs hotter because Linux is more efficient at making
computers work harder. Windows cannot get a CPU loaded like this due
to its inefficiency. I went through this 6 years ago with an AMD CPU.

This server sits on the ISP premises and I am paying a flat fee for
power. So I do not care for "fan management", I just want the baddest
fan to blow air as fast as it can. I hope that Linux would not
interfere with that.

i

Ignoramus22620 wrote:
>
> On Wed, 05 Sep 2007 17:00:56 -0400, Ron Hardin <(E-Mail Removed)> wrote:
> > For what it's worth, my new Dell Vostro 1500 runs very hot on Ubuntu, but okay on Windows XP Home.
> > I think fan management isn't perfect yet.
>
> It probably runs hotter because Linux is more efficient at making
> computers work harder. Windows cannot get a CPU loaded like this due
> to its inefficiency. I went through this 6 years ago with an AMD CPU.
>
> This server sits on the ISP premises and I am paying a flat fee for
> power. So I do not care for "fan management", I just want the baddest
> fan to blow air as fast as it can. I hope that Linux would not
> interfere with that.
>
> i

No, actually under Ubuntu, in the case at hand, it was running one processor at full 1.6 GHz
and the other idle at .8 GHz; Windows XP was running both at 1.6 GHz, presumably one at fast idle.

For the same computation (Cygwin under XP), taking the same amount of time (ten minutes).

Vast quantities of superhot air was blowing out under Ubuntu; nothing noticeable under XP.

(There's a laptop cooler under the thing, so air moves regardless what the system wants.)

So I think fan management is probably something to look at. It can differ over systems, and if the
hardware is newer than Ubuntu, it might be doing the wrong thing.
--
(E-Mail Removed)

On the internet, nobody knows you're a jerk.

At Wed, 05 Sep 2007 15:33:37 -0500 Ignoramus22620 <(E-Mail Removed)> wrote:

>
> Something happened on my server that made several processes do
> intensive computations. (a bug in some of my perl scripts) That loaded
> all cores of my two dual cored CPUs at 100% CPU. I got these messages
> in /var/log/messages and a lot of "wall" messages:
>
> Sep 5 13:17:09 ak74 kernel: CPU1: Running in modulated clock mode
> Sep 5 13:17:18 ak74 kernel: CPU1: Temperature above threshold
> Sep 5 13:17:18 ak74 kernel: CPU0: Temperature above threshold
> Sep 5 13:17:18 ak74 kernel: CPU0: Running in modulated clock mode
> Sep 5 13:17:18 ak74 kernel: CPU1: Running in modulated clock mode
> Sep 5 13:17:27 ak74 kernel: CPU0: Temperature above threshold
> Sep 5 13:17:27 ak74 kernel: CPU1: Temperature above threshold
> Sep 5 13:17:27 ak74 kernel: CPU1: Running in modulated clock mode
> Sep 5 13:17:27 ak74 kernel: CPU0: Running in modulated clock mode
> Sep 5 13:17:38 ak74 kernel: CPU0: Temperature above threshold
> Sep 5 13:17:38 ak74 kernel: CPU1: Temperature above threshold
> Sep 5 13:17:38 ak74 kernel: CPU1: Running in modulated clock mode
> Sep 5 13:17:38 ak74 kernel: CPU0: Running in modulated clock mode
> Sep 5 13:17:44 ak74 kernel: CPU1: Temperature above threshold
> Sep 5 13:17:44 ak74 kernel: CPU0: Temperature above threshold
>
>
> So I think that Linux saved my ass and switched to modulated clock
> mode instead of shutting down or burning my CPU. It simply slowed
> down.
>
> Thanks for that.
>
> My question is, normally a computer can be very busy and should not
> overheat. So something physical is wrong. Would you have any idea,
> maybe I need to replace case fans or the CPU fan?

Very much likely. Fans have bearings that after a time wearout, sieze
up, and burnout the fan motor. Good old 'entropy' at work here.


>
> This is a 1U rackmount from Rackmounts Etc.
>
> I called my ISP and they confirmed that they did not have any cooling
> or heating issues on their racks.
>
> thanks
>
> i
>

--
Robert Heller -- Get the Deepwoods Software FireFox Toolbar!
Deepwoods Software -- Linux Installation and Administration
http://www.deepsoft.com/ -- Web Hosting, with CGI and Database
(E-Mail Removed) -- Contract Programming: C/C++, Tcl/Tk

Robert Heller wrote:
> At Wed, 05 Sep 2007 15:33:37 -0500 Ignoramus22620
> <(E-Mail Removed)> wrote:
>
>> Something happened on my server that made several processes do
>> intensive computations. (a bug in some of my perl scripts) That loaded
>> all cores of my two dual cored CPUs at 100% CPU. I got these messages
>> in /var/log/messages and a lot of "wall" messages:
>>

>> So I think that Linux saved my ass and switched to modulated clock mode
>> instead of shutting down or burning my CPU. It simply slowed down.
>>
>> Thanks for that.
>>
>> My question is, normally a computer can be very busy and should not
>> overheat. So something physical is wrong. Would you have any idea,
>> maybe I need to replace case fans or the CPU fan?
>
> Very much likely. Fans have bearings that after a time wearout, sieze
> up, and burnout the fan motor. Good old 'entropy' at work here.
>
That's for sure. When I first put together this machine, it worked fine for
about 3 or 4 months, then the cpu fans (temperature controlled) speeded up
to slightly over 6000 rpm. It turned out the main case intake fan had
already seized up though not very old at all. Sleeve bearing type. I put in
a ball-bearing replacement one and it has worked over two years with no trouble.


--
.~. Jean-David Beyer Registered Linux User 85642.
/V\ PGP-Key: 9A2FC99A Registered Machine 241939.
/( )\ Shrewsbury, New Jersey http://counter.li.org
^^-^^ 19:25:01 up 27 days, 22:47, 4 users, load average: 5.29, 5.26, 5.19

"Ignoramus22620" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) ...
<snip>
> My question is, normally a computer can be very busy and should not
> overheat. So something physical is wrong. Would you have any idea,
> maybe I need to replace case fans or the CPU fan?

The only time I've had a problem with overheating was when dust clogged the
cpu fan (this was at home). Otherwise fans (as others have pointed out) are
the obvious thing to look at.

> This is a 1U rackmount from Rackmounts Etc.
>
> I called my ISP and they confirmed that they did not have any cooling
> or heating issues on their racks.

If the computer is owned by your ISP and you have logs showing that there is
a problem with overheating then surely its the ISP's responsibility? or have
I missed something? Tell them the problem and ask them to sort it. Just
becasue they are not aware of heating issues doesn't mean they don't have
heating issues - you've found the problem but I would have thought it was
their responsibility to fix it.
--
Brian Cryer
www.cryer.co.uk/brian

On Thu, 6 Sep 2007 10:29:18 +0100, Brian Cryer <(E-Mail Removed)> wrote:
> "Ignoramus22620" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed) ...
><snip>
>> My question is, normally a computer can be very busy and should not
>> overheat. So something physical is wrong. Would you have any idea,
>> maybe I need to replace case fans or the CPU fan?
>
> The only time I've had a problem with overheating was when dust clogged the
> cpu fan (this was at home). Otherwise fans (as others have pointed out) are
> the obvious thing to look at.

Yes, indeed I will replace the [CPU] fan.

>> This is a 1U rackmount from Rackmounts Etc.
>>
>> I called my ISP and they confirmed that they did not have any cooling
>> or heating issues on their racks.
>
> If the computer is owned by your ISP and you have logs showing that there is
> a problem with overheating then surely its the ISP's responsibility? or have
> I missed something? Tell them the problem and ask them to sort it. Just
> becasue they are not aware of heating issues doesn't mean they don't have
> heating issues - you've found the problem but I would have thought it was
> their responsibility to fix it.

The computer is owned by me, but it is co-located on their premises.

i

Ignoramus22620 wrote:
> On Wed, 05 Sep 2007 17:00:56 -0400, Ron Hardin <(E-Mail Removed)> wrote:
>> For what it's worth, my new Dell Vostro 1500 runs very hot on Ubuntu, but okay on Windows XP Home.
>> I think fan management isn't perfect yet.
>
> It probably runs hotter because Linux is more efficient at making
> computers work harder. Windows cannot get a CPU loaded like this due
> to its inefficiency. I went through this 6 years ago with an AMD CPU.
>
> This server sits on the ISP premises and I am paying a flat fee for
> power. So I do not care for "fan management", I just want the baddest
> fan to blow air as fast as it can. I hope that Linux would not
> interfere with that.
>
> i

I would have thought if something was more efficient, it would just get
the job done quicker, not run your machine hot all the time. Sounds to
me like the complete reverse is happening.

--

x theSpaceGirl (miranda)

http://www.northleithmill.com

-.-

Kammy has a new home: http://www.bitesizedjapan.com