[lm-sensors] fan control questions

[Mark Nienberg]

Jean Delvare wrote:
> Hi Mark,
> 
> On Tue, 27 Jan 2009 10:02:28 -0800, Mark Nienberg wrote:
>> I was hoping to use the fan control program on a server with an ASUS KFN4-DRE 
>> motherboard.  The BIOS has three settings for control of fans (disabled, smart fan, 
>> smart fan II). The two smart fan options work very poorly, cycling through full speed 
>> and low speed on about a 24 hour cycle with no apparent regard to temps.  That is why 
>> I would like to take control and force something more rational.

> It might be better to tweak the settings of the automatic modes than to
> switch to software control. Sometimes the BIOS allows this. If not,
> some of our drivers have sysfs files to tweak the settings.

There is nothing in the BIOS that allows me to tweak the way the BIOS fan control 
works.  I would love to let the BIOS take care of this, but it does a poor job, often 
running the CPU fan noisily at full speed for hours, even when temps are low.

>> I assumed that if I am controlling fans from the OS then I don't want the BIOS to 
>> also try to control fans, so I set it to "disabled" and then ran the pwmconfig 
>> program. The program found three pwm devices and tried each one in turn, but there 
>> was no effect on the two fans connected to the motherboard.  There are a total of 10 
>> fan connectors on this board, six of which show up in lm_sensors.  The manual says 
>> that the "smart fan" setting applies to all fan connectors.
>>
>> Should I have tested with the BIOS settings other than "disabled"?
> 
> This shouldn't change anything. If pwmconfig finds a PWM output in
> automatic mode, it'll warn you that you probably don't want to change
> it to manual mode, but that's about it (you can override that if you
> really want to.)
> 
>> Should I test the other 4 fan connectors that lm_sensors recognizes?
> 
> Yes, it can't hurt.

I'll try some more fans then.

> It would help if you would provide more information about your board.
> In particular, which hardware monitoring chip(s) are found by
> sensors-detect? What does the output of "sensors" look like?
> 
> If there is more than one chip, are the PWM outputs found on the same
> chip that reports the fan speeds?

I'll run pwmconfig again and see if I can answer that.  Maybe I should try a newer 
lm_sensors, but so far I have been sticking with rpms rather than source.

> And do you see fan speeds change when in automatic control mode?

Yes, the BIOS does change the fan speed, though not rationally.

>> Or is this just one of the boards that fan control won't be able to handle?
> 
> If the BIOS has options to control the fans then the wiring on the board
> must be correct. You're probably not too far away.

Thanks.  I won't give up then.  This is the office file server, so I am limited in 
the times I can work on it.  Maybe I'll have another go this weekend.

-- 
Mark Nienberg
Sent from an invalid address. Please reply to the group.