[gentoo-user] [OT overheat] How to cause shutdown on overheat

[gentoo-user] [OT overheat] How to cause shutdown on overheat

[reader]

Group, I recently built a ventilated stucture around my 4 desktops to
try to quiet things down and get rid of the heat.

I made no provision for forced shutdown in case of overheat, which is
quite likely to happen if, for example the main ventilation fan went
down for some reason.

Well, that happened due to stupidity on my part with getting used to
the new setup.  I fired up a computer and neglected to turn the fan
on.  Then left it running overnight.

Well, given the confined space and very little/no ventilation (of my
homemade structure) the computer got hot...

Sometime this morning I see syslog messages written to tty that say:

  Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
  reader kernel: CPU0: Temperature above threshold

  Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
  reader kernel: CPU0: Running in modulated clock mode

[...]

Some kind of attempt by kernel to cool things down.  But will it
actually shutdown if it gets dangerously hot?

Further, how can I discover what temperatures were involved when this
happened?

Or can I set something to make a shutdown happen at a specific
temperature? 

A nicer solution would be somekind of added stand alone temperature
monitor in the enclosure that causes a controlled shutdown like one
gets with `shutdown -h now'.

Anyone here with some experience in this kind of thing that can steer me
to some good information?


-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] [OT overheat] How to cause shutdown on overheat

[Daniel da Veiga]

On 9/30/06, reader@newsguy.com <reader@newsguy.com> wrote:
> Group, I recently built a ventilated stucture around my 4 desktops to
> try to quiet things down and get rid of the heat.
>
> I made no provision for forced shutdown in case of overheat, which is
> quite likely to happen if, for example the main ventilation fan went
> down for some reason.
>
> Well, that happened due to stupidity on my part with getting used to
> the new setup.  I fired up a computer and neglected to turn the fan
> on.  Then left it running overnight.
>
> Well, given the confined space and very little/no ventilation (of my
> homemade structure) the computer got hot...
>
> Sometime this morning I see syslog messages written to tty that say:
>
>   Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
>   reader kernel: CPU0: Temperature above threshold
>
>   Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
>   reader kernel: CPU0: Running in modulated clock mode
>
> [...]
>
> Some kind of attempt by kernel to cool things down.  But will it
> actually shutdown if it gets dangerously hot?
>
> Further, how can I discover what temperatures were involved when this
> happened?
>
> Or can I set something to make a shutdown happen at a specific
> temperature?
>
> A nicer solution would be somekind of added stand alone temperature
> monitor in the enclosure that causes a controlled shutdown like one
> gets with `shutdown -h now'.
>
> Anyone here with some experience in this kind of thing that can steer me
> to some good information?
>

If you have built your kernel with ACPI options for THERMAL or some
kind of frequency changer, you can use a daemon like cpufreqd to
monitor the temperature and take actions like reduce the clock and
voltage to avoid damage to the processor.

http://www.gentoo.org/doc/en/power-management-guide.xml

When my computer reaches the limit (wich in my case is 73 degrees C)
it automatically shutdown, I didn't have to configure anything, it is
builtin with the thermal ACPI module. I'm talking about a Pentium IV
Northwood here, and they tend to get really hot. It used to happen to
my Athlon XP also... To monitor the temperature, you can read the
/proc entries created by ACPI, for example
/proc/acpi/thermal/TZ0/info.

-- 
Daniel da Veiga
Computer Operator - RS - Brazil
-----BEGIN GEEK CODE BLOCK-----
Version: 3.1
GCM/IT/P/O d-? s:- a? C++$ UBLA++ P+ L++ E--- W+++$ N o+ K- w O M- V-
PS PE Y PGP- t+ 5 X+++ R+* tv b+ DI+++ D+ G+ e h+ r+ y++
------END GEEK CODE BLOCK------
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] [OT overheat] How to cause shutdown on overheat

[Richard Fish]

On 9/30/06, reader@newsguy.com <reader@newsguy.com> wrote:
>   Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
>   reader kernel: CPU0: Temperature above threshold
>
>   Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
>   reader kernel: CPU0: Running in modulated clock mode
>
> [...]
>
> Some kind of attempt by kernel to cool things down.  But will it
> actually shutdown if it gets dangerously hot?

Nope, not the kernel.  Modern CPUs have built-in thermal
throttling...if they get too hot, they reduce their clocks until
things cool down.  They generate an exception when this occurs, which
the kernel sees  and logs if you have the right ACPI options enabled.

They will also simply shutdown if the temperature gets too warm.
However, this would generally be bad, as no clean shutdown or even
disk sync would occur.

> Further, how can I discover what temperatures were involved when this
> happened?

Download the processor specs from the manufacturer's web site....that
should list the throttle and shutdown temperatures.

> Or can I set something to make a shutdown happen at a specific
> temperature?

Install and run acpid, and you should get ACPI temperature events
passed through to /etc/acpi/default.sh, which you can edit to handle
in whatever way you see fit.

-Richard
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] [OT overheat] How to cause shutdown on overheat

[Michael Gisbers]

[-- Attachment #1: Type: text/plain, Size: 2010 bytes --]

Am Samstag 30 September 2006 14:16 schrieb reader@newsguy.com:
> Group, I recently built a ventilated stucture around my 4 desktops to
> try to quiet things down and get rid of the heat.
>
> I made no provision for forced shutdown in case of overheat, which is
> quite likely to happen if, for example the main ventilation fan went
> down for some reason.
>
> Well, that happened due to stupidity on my part with getting used to
> the new setup.  I fired up a computer and neglected to turn the fan
> on.  Then left it running overnight.
>
> Well, given the confined space and very little/no ventilation (of my
> homemade structure) the computer got hot...
>
> Sometime this morning I see syslog messages written to tty that say:
>
>   Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
>   reader kernel: CPU0: Temperature above threshold
>
>   Message from syslogd@reader at Sat Sep 30 04:41:32 2006 ...
>   reader kernel: CPU0: Running in modulated clock mode
>
> [...]
>
> Some kind of attempt by kernel to cool things down.  But will it
> actually shutdown if it gets dangerously hot?
>
> Further, how can I discover what temperatures were involved when this
> happened?
>
> Or can I set something to make a shutdown happen at a specific
> temperature?
>
> A nicer solution would be somekind of added stand alone temperature
> monitor in the enclosure that causes a controlled shutdown like one
> gets with `shutdown -h now'.
>
> Anyone here with some experience in this kind of thing that can steer me
> to some good information?

Just have a look to your trip_point:

mgisbers@mars ~ $ cat /proc/acpi/thermal_zone/THRM/trip_points
critical (S5):           95 C
passive:                 87 C: tc1=1 tc2=5 tsp=10 devices=0xdff72dc4

When reaching 95° my computer starts to shutdown fast.

I found more Information on 
http://acpi.sourceforge.net/documentation/thermal.html

Maybe it helps you too.
-- 
 Michael Gisbers
 http://www.lugor.de

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] [OT overheat] How to cause shutdown on overheat

[Joost Roeleveld]

On Sunday 01 October 2006 17:31, Michael Gisbers wrote:
>
> Just have a look to your trip_point:
>
> mgisbers@mars ~ $ cat /proc/acpi/thermal_zone/THRM/trip_points
> critical (S5):           95 C
> passive:                 87 C: tc1=1 tc2=5 tsp=10 devices=0xdff72dc4
>
> When reaching 95° my computer starts to shutdown fast.
>

My trip_point reads:
cat /proc/acpi/thermal_zone/THRM/trip_points
critical (S5):           80 C
passive:                 78 C: tc1=4 tc2=3 tsp=60 devices=0xffff8800007fce00
active[0]:               78 C: devices=0xffff8800004f2200

The value for 'critical' has been entered into the BIOS, at which point the
computer will perform a hard shutdown (equivalent to pulling the plug)

On my old laptop (died recently) I had it set up to initiate a proper
shutdown
(shutdown -h now) when it would get near this value.

How can I configure acpid to perform a proper shutdown when it reaches 78 C,
which ACPI-event would I get?
I'd rather not experiment for obvious reasons as this machine is used to
store
my emails and other important files.

Thanks,

Joost Roeleveld

-- 
gentoo-user@gentoo.org mailing list