LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Software (https://www.linuxquestions.org/questions/linux-software-2/)
-   -   examine my lm_sensors output (https://www.linuxquestions.org/questions/linux-software-2/examine-my-lm_sensors-output-550173/)

JustinHoMi 04-30-2007 02:11 PM

examine my lm_sensors output
 
Hey. I just installed lm_sensors on a remote server, and the output looks a bit suspicious. I wonder if it's a problem with the configuration, or if there is a legitimate hardware problem.

As you can see, most of the voltages are within range. However, the -5V is way off, as is the -12V.

Additionally, in looking at my graphs, Temp3 is really sporadic. It has a huge range, so I wonder if it's monitoring something else, or if the sensor is bad. How can I figure out exactly what Temp3 is?

Code:

eeprom-i2c-0-50
Adapter: SMBus Via Pro adapter at 5000
Memory type:            DDR SDRAM DIMM
Memory size (MB):      512
Note that eeprom support will be dropped from libsensors soon.
Please use the decode-dimms.pl script instead.

it8712-isa-0290
Adapter: ISA adapter
VCore 1:  +1.65 V  (min =  +4.08 V, max =  +4.08 V)  ALARM
VCore 2:  +2.45 V  (min =  +4.08 V, max =  +4.08 V)  ALARM
+3.3V:    +3.34 V  (min =  +4.08 V, max =  +4.08 V)  ALARM
+5V:      +4.89 V  (min =  +6.85 V, max =  +6.85 V)  ALARM
+12V:    +12.28 V  (min = +16.32 V, max = +16.32 V)  ALARM
-12V:      -2.20 V  (min =  +3.93 V, max =  +3.93 V)  ALARM
-5V:      -9.27 V  (min =  +3.94 V, max =  +4.03 V)  ALARM
Stdby:    +5.01 V  (min =  +6.85 V, max =  +6.85 V)  ALARM
VBat:      +2.99 V
fan1:    4017 RPM  (min =  672 RPM, div = 8)         
fan2:        0 RPM  (min =    0 RPM, div = 8)         
fan3:        0 RPM  (min =    0 RPM, div = 8)         
M/B Temp:    +43?C  (low  =    -1?C, high =    -1?C)  sensor = thermistor 
CPU Temp:    +29?C  (low  =    -1?C, high =    -1?C)  sensor = thermistor 
Temp3:      +33?C  (low  =    -1?C, high =  -65?C)  sensor = thermistor

Justin

lazlow 04-30-2007 02:27 PM

It is a configuration problem. The way I figure it out goes something like this. Look at all the temps with nothing running on the machine (write them down). Put the cpu under high strain (mprime or whatever). The actual cpu temperature will rise rapidly. Regardless of which label is marked, the one that shoots up is the cpu. You can then go to /etc/sensors.conf and alter the label that goes to the cpu appropriately. You can do the same thing with the northbridge (?) temp(heavy files transfers, etc). Once you have the first two figured out, the third one is by default the MB temp.

Look at man sensors for more help.

You can also search the web using your motherboard and sensors.conf, often you will find a premade one.

Good Luck
Lazlow

GrapefruiTgirl 04-30-2007 02:30 PM

There's enough odd readings there to strongly suspect the configuration, either the config file itself, and/or the kernel support for your motherboard's hardware monitoring devices.
When I set up LMsensors on my machine, I spent a LOT of time configuring the values and stuff for the various sensor readings, and also, researching my board's chipsed on the net. (It is a w63287hf or something like that) the default 'descriptors' for what was supposed to be my onboard monitoring device were mostly wrong: Fans were labeled as voltages, voltages were labeles as temperatures, some of the defaults returned zero-values.. Some of the fans were configured with names like 'Sensor3' rather than 'Fan3' etc..

And to expand on both what I and what lazlow said, I realize you said the machine is 'remote', but if you could get access to or reports of the BIOS power/setup screen, you would have the most accurate low-baseline numbers you will get on the machine, to start with.

weibullguy 04-30-2007 03:33 PM

Quote:

Originally Posted by JustinHoMi
Hey. I just installed lm_sensors on a remote server, and the output looks a bit suspicious. I wonder if it's a problem with the configuration, or if there is a legitimate hardware problem.

As you can see, most of the voltages are within range. However, the -5V is way off, as is the -12V.

It's the configuration. There are lots of tell-tale signs. Not the least of which is that the server works, right? If you have the sensors configured correctly and something alarms, nine times out of ten it's already too late. Just shut things down to prevent further damage.

In any event, look at your temperatures. According to your output, the motherboard is 25F hotter than the CPU. Not very likely. Your processor voltages are also too high. For example, the absolute maximum rating for an Opteron is 1.65V. If it really was 2.45V, the processor would be toast. Intel processors will have similar ratings. I'm guessing you don't have a battery in your server, but the sensors are reporting a battery voltage (VBat).

I wouldn't trust a single output, even the ones that look good, until you configure everything. The it8712 chip is what is referred to as a Super I/O chip. It controls all sorts of I/O from floppy drives to joysticks to keyboards to fans. Some of the I/O are dedicated (e.g., the fans) some are general purpose I/O (firmware configured by the mobo designer), and some are both.

The maximum recommended input voltage for an it8712 pin is 4.096V (IIRC). That means just about every voltage is going to require a voltage divider; probably all the voltages use a divider to prevent loading the power supply. You have no way of knowing how the designers of your motherboard divided each voltage. Is the +12V divided so the it8712 input is 2V, 3V, 0.5V, etc. All would work, but each would require a different multiplier.

If you're lucky you'll find a configuration file on the internet somewhere. But, of course, how was that all figured out? You could contact the motherboard vendor and ask. That's how I figured out what's what on my PC.

Quote:

Originally Posted by JustinHoMi
Additionally, in looking at my graphs, Temp3 is really sporadic. It has a huge range, so I wonder if it's monitoring something else, or if the sensor is bad. How can I figure out exactly what Temp3 is?

The only way to really know is to look at the schematic for the motherboard and see what is connected to pin 87 on the it8712. I'd bet it's floating and not monitoring anything at all.

JustinHoMi 05-01-2007 06:35 PM

Thanks for all the info. Really helpful. Any ideas on how to determine the model of the motherboard without contacting the support guys?

HappyTux 05-01-2007 09:24 PM

Quote:

Originally Posted by JustinHoMi
Thanks for all the info. Really helpful. Any ideas on how to determine the model of the motherboard without contacting the support guys?

Try the lshw or dmidecode programs or open the box and look at the motherboard.


All times are GMT -5. The time now is 10:38 PM.