Resource icon

How To: Change IPMI Sensor Thresholds using ipmitool

DarkLogix

Cadet
Joined
Feb 3, 2018
Messages
4
You may have to reset the BMC (either reboot it from the IPMI webGUI or completely cut power to the server for a few seconds).
Guessing that would reset it to default?

After that how would I get to where it's not oscillating or going full speed?
 

DarkLogix

Cadet
Joined
Feb 3, 2018
Messages
4
So I've done the following
1. shut down the server
2. flipped the power switch on the PSU to off for awhile
3. moved the AIO from Fan2 (previously had it on fan1) to Fan4 (I thought it was fanA (just realized that mistake)
4. switched the AIO from PWM control to Auto.
5. powered on

And now it's properly quiet and the idle CPU temp is about 24c (or 35c, the sensor reading keeps bouncing.)
So this should be good.

Thanks.
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
These BMCs can be very weird in their behavior and a reset is occasionally needed when manipulating sensor thresholds.

3. moved the AIO from Fan2 (previously had it on fan1) to Fan4 (I thought it was fanA (just realized that mistake)
4. switched the AIO from PWM control to Auto.
You mean the pump?
 

DarkLogix

Cadet
Joined
Feb 3, 2018
Messages
4
These BMCs can be very weird in their behavior and a reset is occasionally needed when manipulating sensor thresholds.


You mean the pump?
The AIO I'm using is the Fractal Design C360, which has a cable from the pump to the a PWM splitter on the radiator where the 3 fan's it comes with are then connected. So by AIO I mean the pump and 3x120mm fans, though I think the RPM sense wire is only reading the pump. https://www.newegg.com/Product/Product.aspx?Item=N82E16835352030

The reason I picked this AIO is since it's a 360 I figure it would be able to run very quiet more so since the CPU has a TDP of only 72W (Xeon E3-1275 v6) also the cabling for it is fairly clean by design.
 

guermantes

Patron
Joined
Sep 27, 2017
Messages
213
The failure rate rise when temp is below roughly 30 °C, it rises even faster if the temp is over 40 °C.

But I always wondered if it's because of the temp, or because of the temp changes, or both.

Or can it not simply be because being over 40 °C generally indicates that the drive is reading/writing and not just idly spinning, and thus is more subjected to "wear and tear" and thus breaks down faster?
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
Or can it not simply be because being over 40 °C generally indicates that the drive is reading/writing and not just idly spinning, and thus is more subjected to "wear and tear" and thus breaks down faster?
No. A single point in time is meaningless, the data is from a longer period. SMART conveniently logs the lifetime extremes the disk is subjected to, as well.
 

HELLiON

Cadet
Joined
Jan 6, 2019
Messages
4
To lower fan sensor limits using ipmitool

ipmitool sensor list all (gives sensor name and current settings)

ipmitool sensor thresh "FAN 1" lower 100.000 150.000 200.000
ipmitool sensor thresh "FAN 2" lower 100.000 150.000 200.000
ipmitool sensor thresh "FAN 3" lower 100.000 150.000 200.000
ipmitool sensor thresh "FAN 4" lower 100.000 150.000 200.000
ipmitool sensor thresh "FAN A" lower 100.000 150.000 200.000
 

Christopher Ward

Contributor
Joined
May 14, 2015
Messages
104
Just came across this thread after noticing my new ML120 PRO fans from corsair kept ramping up and down every few seconds when connected to my motherboard and it was extremely loud. My fans are rated from 400-2400RPM so would i be correct in assuming this is the correct way to set them or have i done it in the wrong order:

ipmitool sensor thresh "FAN2" lower 200 300 400
ipmitool sensor thresh "FAN2" upper 2400 2500 2600

ipmitool sensor thresh "FAN3" lower 200 300 400
ipmitool sensor thresh "FAN3" upper 2400 2500 2600

https://www.corsair.com/uk/en/Categ.../ml-pro-config/p/CO-9050040-WW#tab-tech-specs
https://youtu.be/zEeTr0Aaqf8
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
You might want to widen the range if you get log spam.
 

Christopher Ward

Contributor
Joined
May 14, 2015
Messages
104
Still can't get this to work.

I currently have the intel stock cooler attached to FAN1 and 2 Corsair ML 120's connected via a fan splitter to FANA

Fan mode is set to optimal in the IPMI

Code:
ipmitool sensor thresh "FANA" lower 400 500 600
ipmitool sensor thresh "FANA" upper 2400 2500 2600


showing this:
Code:
FAN1             | 2000.000   | RPM        | nr    | 500.000   | 600.000   | 700.000   | 1000.000  | 1100.000  | 1200.000
FAN2             | na         |            | na    | na        | na        | na        | na        | na        | na
FAN3             | na         |            | na    | na        | na        | na        | na        | na        | na
FAN4             | na         |            | na    | na        | na        | na        | na        | na        | na
FANA             | 2500.000   | RPM        | cr    | 200.000   | 300.000   | 400.000   | 2400.000  | 2500.000  | 2600.000


The max speed the corsair fans can do is 2400 RPM but that's showing them maxing out at 2500? and i can't lower the speed


Motherboard: supermicro x10sl7-f
 
Last edited:

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
The max speed the corsair fans can do is 2400 RPM but that's showing them maxing out at 2500? and i can't lower the speed
You're assuming the spec is infallible - it isn't. It could also be that the motherboard isn't measuring the fan speed correctly. Either way, the solution is to not set the limits right at what the datasheet says, but to give a bit of a buffer.
 
Joined
Dec 2, 2015
Messages
730
Still can't get this to work.

I currently have the intel stock cooler attached to FAN1 and 2 Corsair ML 120's connected via a fan splitter to FANA

Fan mode is set to optimal in the IPMI

Code:
ipmitool sensor thresh "FANA" lower 400 500 600
ipmitool sensor thresh "FANA" upper 2400 2500 2600


showing this:
Code:
FAN1             | 2000.000   | RPM        | nr    | 500.000   | 600.000   | 700.000   | 1000.000  | 1100.000  | 1200.000
FAN2             | na         |            | na    | na        | na        | na        | na        | na        | na
FAN3             | na         |            | na    | na        | na        | na        | na        | na        | na
FAN4             | na         |            | na    | na        | na        | na        | na        | na        | na
FANA             | 2500.000   | RPM        | cr    | 200.000   | 300.000   | 400.000   | 2400.000  | 2500.000  | 2600.000
The Intel stock cooler, connected to FAN1 was running at 2000 rpm at the moment you checked with ipmitool, but the upper fan thresholds are set to much lower speeds than that. This will cause the BMC to think there is a problem, and the fan speed will be keep going up and down. Change the upper speed thresholds on FAN1 to faster than the max speed of that fan.
 

Christopher Ward

Contributor
Joined
May 14, 2015
Messages
104
The Intel stock cooler, connected to FAN1 was running at 2000 rpm at the moment you checked with ipmitool, but the upper fan thresholds are set to much lower speeds than that. This will cause the BMC to think there is a problem, and the fan speed will be keep going up and down. Change the upper speed thresholds on FAN1 to faster than the max speed of that fan.

Thanks that seems to have fixed it


Code:
FAN1             | 900.000    | RPM        | ok    | 600.000   | 700.000   | 800.000   | 2500.000  | 2600.000  | 2700.000
FAN2             | na         |            | na    | na        | na        | na        | na        | na        | na
FAN3             | na         |            | na    | na        | na        | na        | na        | na        | na
FAN4             | na         |            | na    | na        | na        | na        | na        | na        | na
FANA             | 600.000    | RPM        | ok    | 200.000   | 300.000   | 400.000   | 2400.000  | 2500.000  | 2600.000


I can barely hear any fans spinning now!

1553589154804.png
 

Octopuss

Patron
Joined
Jan 4, 2019
Messages
461
How would you set the thresholds for a fan with these specs? 300 ~ 1600 RPM ±10 %
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
100, 200, 300, 1700, 1800, 1900
 

walks

Dabbler
Joined
May 18, 2019
Messages
13
Hey @Ericloewe, what about the thresholds for my fan, the Supermicro FAN-0126L4? Have not been able to find detailed specs on this fan. At most it states the Rated fan speed as 7000RPM. And have not seen anyone in this thread with a fan similar to this.

I figure I can use the following for the upper: 7100 7200 7300

but not sure what to use for the lower. Would 100 200 300 be to low for a fan rated at 7000RPM?

The FAN mode is set to Optimal in IPMI and the 3 fans are averaging between 2600-2800 rpms right now.
 

Ericloewe

Server Wrangler
Moderator
Joined
Feb 15, 2014
Messages
20,175
It's probably a Delta, SanAce or Nidec fan. The OEM datasheet would have detailed specs.
Would 100 200 300 be to low for a fan rated at
In the sense that its rated minimum is probably much higher, yes.
 

francisaugusto

Contributor
Joined
Nov 16, 2018
Messages
153
The command for 50% duty cycle is:
Code:
ipmitool raw 0x30 0x70 0x66 0x01 0x00 0x32
.

Yes, the duty cycle can be specified anywhere between 0 and 100% (I'm not sure how many discrete values can be usefully specified - on my machine I seem to be able to get roughly 8 different fan speeds).

The last value in that command sets the duty cycle, with 64 = 100% and 0 =minimum speed. i.e., "0x32" is 50%. "0x64" is 100% duty cycle. "0x48" is 75% duty cycle.

Hi Kevin,

Let me ask you something - I have a X11SDV-4C-TP4F board on a Fractal Design node 804. The case's fans are connected, well, to the case, and not to the motherboard, since they ahave three pins, and my MB have only 4-pin fans.

For the CPU, I have the original cooler from Supermicro, which is as loud as a Supermicro cooler can be. My readings, however, were showing the CPU fan was at 3800 rpm, which was weird, as my CPU was around 39 degrees.

I've set up the lower thresholds to 200 300 and 400, and the upper to 2600, 2700 and 2800. Still, the fan would keep spinning at 3800.

I then found your post, quoted above. By setting the duty cycle to 0, the fan became considerably less noisy, and was spinning around 1800, which is still high, compared to what people say here, but still much better than the 3800 I was having. The temperature got a bit higher, rarely touching the 50. But when it got to 50, the fan went back to 3800 rpm! And even when the temps got low, it wouldn't go back to the 1800 rpm I had.

I changed the duty cycle to 0x2, and it's just marginally louder, the fan spins at 2200 rpm, and the temps are stable at 44ºC. However, once or twice the fans went back to 3800 rpm, and the temps were below 50.

What am I doing wrong here? Is it my thresholds that need to be different? Is it because my case fans are not connected to the motherboard? Should I then replace them with 4-pin ones and connect them to the motherboard instead?
 
Joined
Dec 2, 2015
Messages
730
Hi Kevin,

Let me ask you something - I have a X11SDV-4C-TP4F board on a Fractal Design node 804. The case's fans are connected, well, to the case, and not to the motherboard, since they ahave three pins, and my MB have only 4-pin fans.

For the CPU, I have the original cooler from Supermicro, which is as loud as a Supermicro cooler can be. My readings, however, were showing the CPU fan was at 3800 rpm, which was weird, as my CPU was around 39 degrees.

I've set up the lower thresholds to 200 300 and 400, and the upper to 2600, 2700 and 2800. Still, the fan would keep spinning at 3800.

I then found your post, quoted above. By setting the duty cycle to 0, the fan became considerably less noisy, and was spinning around 1800, which is still high, compared to what people say here, but still much better than the 3800 I was having. The temperature got a bit higher, rarely touching the 50. But when it got to 50, the fan went back to 3800 rpm! And even when the temps got low, it wouldn't go back to the 1800 rpm I had.

I changed the duty cycle to 0x2, and it's just marginally louder, the fan spins at 2200 rpm, and the temps are stable at 44ºC. However, once or twice the fans went back to 3800 rpm, and the temps were below 50.

What am I doing wrong here? Is it my thresholds that need to be different? Is it because my case fans are not connected to the motherboard? Should I then replace them with 4-pin ones and connect them to the motherboard instead?
I suspect that the BMC is also thinking it should control the CPU fan speed and it jumps in once the CPU temperature hits some threshold. If you want to have full control over the CPU fan, you would need to set the BMC Fan Mode to FULL and it will then allow you full control over the fan duty cycle. But beware, you could cook your CPU.

Set fan mode to FULL:

Code:
ipmitool raw 0x30 0x45 0x01 0x01


The best answer might be to replace the current CPU fan with something quieter and let the BMC control the fan speed as the CPU temp varies.
 

francisaugusto

Contributor
Joined
Nov 16, 2018
Messages
153
I suspect that the BMC is also thinking it should control the CPU fan speed and it jumps in once the CPU temperature hits some threshold. If you want to have full control over the CPU fan, you would need to set the BMC Fan Mode to FULL and it will then allow you full control over the fan duty cycle. But beware, you could cook your CPU.

Set fan mode to FULL:

Code:
ipmitool raw 0x30 0x45 0x01 0x01


The best answer might be to replace the current CPU fan with something quieter and let the BMC control the fan speed as the CPU temp varies.

Thanks Kevin. I ordered a Noctua A6x25 and will see if I man up to replace the fan that's on the cooler.

And I think you're right - because I am on the optimal mode, it might be that it kicks in based on power, for example. I suspect that the full mode will also work, so I might try that at some point. I don't think the CPU is cookable, as others reported that using it without a cooler actually led to a shutdown when it got too hot, but one never knows, so I think that those thresholds might help keeping it into a zone.

But so weird it was giving me such high rpm's before, even when the temp seemed to be reasonable...
 
Top