Message from syslogd@justice at Fri Dec 8 15:29:07 2006 ... justice kernel: CPU0: Temperature above threshold
when I was fiddling with the BIOS of my fancy new 1U server, I set "fan always on" to "turn fan on when exceeding temperature" but apparently the temperature there is not low enough to blow the heat out when I do something CPU-intensive (such as run my .sig for twenty minutes)
Who knows how to find out what the current temperature is? Is it in /proc/misc or somewhere?
This new-fangled magic is beyond me; in my days when the computers overheated, they simply caught fire and halted.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Message from syslogd@justice at Fri Dec 8 15:29:07 2006 ... justice kernel: CPU0: Temperature above threshold
when I was fiddling with the BIOS of my fancy new 1U server, I set "fan always on" to "turn fan on when exceeding temperature" but apparently the temperature there is not low enough to blow the heat out when I do something CPU-intensive (such as run my .sig for twenty minutes)
Who knows how to find out what the current temperature is? Is it in /proc/misc or somewhere?
This new-fangled magic is beyond me; in my days when the computers overheated, they simply caught fire and halted.
-- perl -le'1while(1x++$_)=~/^(11+)\1+$/||print'
lm-sensors is pretty useful. There are several x/kde/gnome clients that will read data based on this client and libs.
smartmon-tools is also helpful for monitoring the temp and status of harddrives.
sensors it8712-isa-0290 Adapter: ISA adapter VCore 1: +1.26 V (min = +1.42 V, max = +1.57 V) ALARM VCore 2: +1.49 V (min = +2.40 V, max = +2.61 V) ALARM +3.3V: +6.62 V (min = +3.14 V, max = +3.46 V) ALARM +5V: +5.11 V (min = +4.76 V, max = +5.24 V) ALARM +12V: +11.65 V (min = +11.39 V, max = +12.61 V) ALARM -12V: -20.12 V (min = -12.63 V, max = -11.41 V) ALARM -5V: -8.58 V (min = -5.26 V, max = -4.77 V) ALARM Stdby: +5.08 V (min = +4.76 V, max = +5.24 V) ALARM VBat: +3.18 V fan1: 5818 RPM (min = 0 RPM, div = 8) ALARM fan2: 0 RPM (min = 3013 RPM, div = 8) ALARM fan3: 3835 RPM (min = 3013 RPM, div = 8) M/B Temp: +48°C (low = +15°C, high = +40°C) sensor = thermistor ALARM CPU Temp: +45°C (low = +15°C, high = +45°C) sensor = thermistor ALARM Temp3: +53°C (low = +15°C, high = +45°C) sensor = diode ALARM
are these voltage readings real? With the power supply providing power so far out of spec I'm surprised that the box works at all.
On 12/8/06, Christopher A. Bier chris.bier@cymor.com wrote:
lm-sensors is pretty useful. There are several x/kde/gnome clients that will read data based on this client and libs.
On 12/8/06, David Nicol davidnicol@gmail.com wrote:
sensors it8712-isa-0290 Adapter: ISA adapter VCore 1: +1.26 V (min = +1.42 V, max = +1.57 V) ALARM VCore 2: +1.49 V (min = +2.40 V, max = +2.61 V) ALARM +3.3V: +6.62 V (min = +3.14 V, max = +3.46 V) ALARM +5V: +5.11 V (min = +4.76 V, max = +5.24 V) ALARM +12V: +11.65 V (min = +11.39 V, max = +12.61 V) ALARM -12V: -20.12 V (min = -12.63 V, max = -11.41 V) ALARM -5V: -8.58 V (min = -5.26 V, max = -4.77 V) ALARM Stdby: +5.08 V (min = +4.76 V, max = +5.24 V) ALARM VBat: +3.18 V fan1: 5818 RPM (min = 0 RPM, div = 8) ALARM fan2: 0 RPM (min = 3013 RPM, div = 8) ALARM fan3: 3835 RPM (min = 3013 RPM, div = 8) M/B Temp: +48°C (low = +15°C, high = +40°C) sensor = thermistor ALARM CPU Temp: +45°C (low = +15°C, high = +45°C) sensor = thermistor ALARM Temp3: +53°C (low = +15°C, high = +45°C) sensor = diode ALARM
are these voltage readings real? With the power supply providing power so far out of spec I'm surprised that the box works at all.
Sensor and voltage readings are suspect unless verified. That said,do also consider that thermistors are inherently tied to Vref . The "on board" Vregs also can compensate for incredible out of ranges HIGH but not so much LOW being largely "linear" as opposed to Switchers. Then also consider Vcore being clock and duty cycle subjective in REAL value as to it's impact. All this may tie into the rest of your situation.
On 12/8/06, Oren Beck orenbeck@gmail.com wrote:
Sensor and voltage readings are suspect unless verified.
I've heard the same from someone who spent 5 years working high up in the Apple stress testing lab. Basically he said to monitor the temps and note the _change_ in temp, not so much the actual reading. Make a note of the "normal" temp during idle at normal operating air temp. Watch temp changes over stress loading to determine the "normal" changes you get. If the machine goes beyond that temp, whatever the reading, it's on the way to overheating.
Jon.
-
Christopher A. Bier -
David Nicol -
Jon Pruente -
Oren Beck