Fan control on Xserve

[JF Mezei]

Does anyone know of an app that can control fans on an Xserve.

I tried smcfancontrol and it crashes/fails on the Xserve.

Ideally, I would like to test a lower default fan speed when idle,
especially in winter due to colder room.

My Xserve is only 1 CPU, so generates far less heat than what it has
been designed for and those tiny fans at 5000rpm are noisy.

Right now, I can see when fan speed starts to increase as CPU
temperature increases. So if lowering idle fan speed a bit causes no
increase in CPU temperature I know it is safe.

Any suggestions ?

[Jolly Roger]

I wouldn't advise it. I doubt they were designed with such modification
in mind. Safety should come first with expensive server-grade hardware,
which is why the fans run as fast and noisy as they do. Such hardware is
designed for data centers, where noise is the norm. I wouldn't advise
placing one in a home environment to begin with unless you are prepared
to deal with the cooling and noise associated with rack-mounted server
hardware like that. I admit I've been tempted in the past to install
rack-mounted systems in my home, but wouldn't dream of doing it without
specifically designing a closet/room for the additional cooling and
noise insulation required. For my needs, a Mac mini closet server and
Mac Pro workstation provide the server / data crunching ability quite
sufficiently for a home environment, without all the headaches of a rack
setup.

--
E-mail sent to this address may be devoured by my ravenous SPAM filter.
I often ignore posts from Google. Use a real news client instead.

JR

[JF Mezei]

On 14-02-18 17:10, Jolly Roger wrote:

> I wouldn't advise it. I doubt they were designed with such modification
> in mind. Safety should come first with expensive server-grade hardware,

But the Xserve ws designed to have 2 CPUs. When you have only 1, the
cooling requirements are much lower, especially if the system load is
very low.

For instance, if tests would show that reducing fan speed by 1000 rpm
resulted in no temperature increase for the CPU and other areas, then it
would be perfectly safe. The rpms woudl increase normally if temperature
rises.

The xserve doesn't produce much heat. You don't put it in a closet
because that would get hit. but in a room it is fine (about 150watts of
heat, an old lightbulb).

[nospam]

In article <5303d792$0$46364$c3e8da3$a909...@news.astraweb.com>, JF

Mezei <jfmezei...@vaxination.ca> wrote:

> Does anyone know of an app that can control fans on an Xserve.
>
> I tried smcfancontrol and it crashes/fails on the Xserve.

try fan control, a different product, despite the similar name. it
supports up to 10.6, but that's probably not an issue with the xserve.
it says macbooks, but it works on other macs too. it includes source so
you may be able to fix whatever doesn't work.
<http://www.lobotomo.com/products/FanControl/>

or try macs fan control, yet another tool with almost the same name.
apparently coming up something original is difficult.
<http://www.crystalidea.com/macs-fan-control>

[bi...@mix.com]

JF Mezei <jfmezei...@vaxination.ca> writes, quoting Jolly Roger:

> > I wouldn't advise it. I doubt they were designed with such modification
> > in mind. Safety should come first with expensive server-grade hardware
>

> But the Xserve ws designed to have 2 CPUs. When you have only 1, the
> cooling requirements are much lower, especially if the system load is
> very low.

I'd replace the fan, in this case with something less noisy. There is
no shortage of fans from which to choose.

In my case, I went with fans that can move the most air, because the
systems are in a separate room. Heat and dirt are the two enemies of
electronic equipment.

I also use http://www.bresink.com/osx/TemperatureMonitor.html and a
script to shutdown the systems if they overheat, as they are powered
up continuously.

Billy Y..
--
sub #'9+1 ,r0 ; convert ascii byte
add #9.+1 ,r0 ; to an integer
bcc 20$ ; not a number

[JF Mezei]

On 14-02-18 17:58, nospam wrote:

> or try macs fan control, yet another tool with almost the same name.
> apparently coming up something original is difficult.
> <http://www.crystalidea.com/macs-fan-control>

This one does run on the Xserve. (It can control up to 18 fans !)

While I am not able to lower the minimum speed, I was able to exercise
each fan individually.

What is interesting is that the above app shows a current RPM of 1500
for one fan, and Server Monitor shows 6.016 RPM. Roughly a factor of 4
in the values.

Running Handbrake or some Adobe encoders is about the only things that
normally cause the fans to speed up from idle, even in summer.

What I might try is to set different values for each fan to see if it
makes for quieter environment (fewer harmonics).

[Alan Browne]

On 2014.02.19, 00:05 , JF Mezei wrote:
> On 14-02-18 17:58, nospam wrote:
>
>> or try macs fan control, yet another tool with almost the same name.
>> apparently coming up something original is difficult.
>> <http://www.crystalidea.com/macs-fan-control>
>
>
> This one does run on the Xserve. (It can control up to 18 fans !)
>
> While I am not able to lower the minimum speed, I was able to exercise
> each fan individually.
>
> What is interesting is that the above app shows a current RPM of 1500
> for one fan, and Server Monitor shows 6.016 RPM. Roughly a factor of 4
> in the values.

Looks to me like some fans have different methods for sensing /
measuring speed and the factor is 4. This may be important to how you
control your fans. Perhaps by listening you can figure out how to get a
desired setting irrespective of the indicated setting.

eg: if it says "6000" and you ask for 6000 there should be no change in
fan noise. But it could be you need to ask for 1500 to get 6000
indicated (which still might be 1500 - but indicating 6000).

> Running Handbrake or some Adobe encoders is about the only things that
> normally cause the fans to speed up from idle, even in summer.
>
> What I might try is to set different values for each fan to see if it
> makes for quieter environment (fewer harmonics).

Do you have something like iStat Pro running that shows you temperatures?

I'd decide on max temperatures for various parts first - then set fan
speeds that keep those parts under that number by some margin. Note
that Google made a study some years ago showing that hotter hard disks
last longer than cooler hard disks (look it up for the details).

If you just have enclosure fans, then it's better to keep the
electronics as cool as possible. Heat kills electronics over time and
they're more expensive than disks. (Did you find a way to put ordinary
SATA disks in the Apple disk caddies?).

The real problem with your approach is that you're now serving the
server having to watch temperatures and set fan speeds and not the other
way around.

If noise is the issue, then putting the server in an appropriate
location with ample ventilation - but good sound deadening would be a
far better approach.

As to the fewer harmonics one could write a program to listen (audio
input) to the environment, do FFT's to determine beat frequencies and
then tune the fan RPM's to reduce the beat. Good science project.

--
... it may be that "in the cloud" really isn't the best term
for the services these companies offer. What they really
want is to have us "on the leash."
-David Pogue, Scientific American, 2014.02

[JF Mezei]

On 14-02-19 17:32, Alan Browne wrote:

> The real problem with your approach is that you're now serving the
> server having to watch temperatures and set fan speeds and not the other
> way around.

Mot fan control programs on Macs are able to set the lowest fan speed to
a value equal or higher than the hardware's specified limit, but it
leaves the SMC to raise fan speed on demand.

So you only change how low the fans can go, without removing the
automation to control/raise fan speeds.

[JF Mezei]

On 14-02-20 10:18, Lewis wrote:

> The total heat output isn't relevant, it's the heat of the CPU.

Half the fans are aimed at empty space where the second CPU would be.

[Alan Browne]

Could be, and that's a good compromise. But when I was using SMC Fan
Control it was not changing anything at any time. Of course I was using
it to increase speed to keep the temp of a failing HD down at the time -
it worked fine below a given temp, but malfunctioned above that temp.

--
Those who have reduced our privacy, whether they are state
or commercial actors, prefer that we do not reduce theirs.
- Jaron Lanier, Scientific American, 2013.11.