How To: Change IPMI Sensor Thresholds using ipmiutil

[Stux]

I found that the BMC would stall the fans when set to Optimal. I posted my settings in my build thread.

https://forums.freenas.org/index.ph...24-supermicro-x10-sri-f-xeon-e5-1650v4.46262/

 

[ZeroBit]

I found that the BMC would stall the fans when set to Optimal. I posted my settings in my build thread.

https://forums.freenas.org/index.ph...24-supermicro-x10-sri-f-xeon-e5-1650v4.46262/

Thank you for the redirect to your build report. I've given the NF-F12 settings a try with mine, along with some similar settings for the NF-A14, since it has the same Min/Max RPMs. Also, for the NF-P12 I've tried the settings 000:lower non-recoverable, 000:lower critical, 200:lower non-critical, 1400:upper non-critical, 1500:upper critical, 1600:upper non-recoverable. Unfortunately, I still get revving.

So, I'm now going through, disconnecting two out of the three fans, and seeing if it can stabilize any of the fans. I'm theorizing that the mix of fans on SuperMicro's BMC may not work, due to the mixed min/max fan speeds. Unfortunately, the first test may have already blown that theory out of the water, since the NF-F12 alone still revved, but it was nowhere near as often, so it may be a settings issue for that fan. Alternatively, I'm hoping that if I do have a defective fan causing them all to rev, maybe I can isolate a bad fan. I'll have to update you over the next day or two, since all the changing is a bit time-consuming, and I only get at most a couple hours on weeknights to work on this.

 

[ZeroBit]

I found that the BMC would stall the fans when set to Optimal. I posted my settings in my build thread.

https://forums.freenas.org/index.ph...24-supermicro-x10-sri-f-xeon-e5-1650v4.46262/

One other quick thought I had, is that since FAN1, FAN2, FAN3 and FAN4 all seem to be CPU temperature controlled, and the CPU Fan is on FAN1, the BMC may be setting the power-settings (e.g. 10%, 20%, 30%, etc. power) based on the fan speeds of that fan, and expecting the others to adjust accordingly, but maybe the NF-F12's 10% or 20% power setting is just too low, so it's shutting off, which is where the critical and non-responsive alarms are set.

Unfortunately, I don't know the RPMs for the fan that comes with the Intel Boxed Processor E3-1245 V5. I can say it's from Foxconn.

Currently testing the NF-A14, plugged into FANA, and it seems to be rock-steady at 300 RPMs, with system at idle, CPU Temp 28 degrees C, PCH is 35 degrees C, and System Temp is 29 degrees C.

UPDATE: The NF-P12 (FAN4) when run exclusive of the other Noctua fans still revved and threw critical and non-responsive events with the lower threshold settings of 000:non-recoverable, 000:critical, 200:non-critical, but similar to the NF-F12 the events were at what appeared to be a much lower frequency. So, maybe there is some battling going on over the power-setting in the BMC, based on the sensor readings of all the attached fans?...

 

[Stux]

The Noctua fans stall below 30%. The BMC uses fan settings below 30%. This stalls the fan.
Install a good fan controller script (see my sig) and you should be fine.

 

[ZeroBit]

The Noctua fans stall below 30%. The BMC uses fan settings below 30%. This stalls the fan.
Install a good fan controller script (see my sig) and you should be fine.

I've been trying to get the script from your sig to play nicely in my system. So far it produces a similar revving with the script additionally resetting the BMC every couple minutes. On the plus side, it was able to read the drives w/o any script modifications.

I've set the BMC fan mode to "Full", so it shouldn't be fighting the script I don't believe.

The fans RPM ratings are:
Intel/Foxconn Boxed Processor Fan: 2000 RPM (based only on readings from IPMI, since I can't find any literature for this fan, assuming +/- 10%)
Noctua NF-F12: 1500 RPM +/- 10% (taken from the box)
Noctua NF-P12: 1300 RPM +/- 10% (taken from the box)
Noctua NF-A14: 1500 RPM +/- 10% (taken from the box)

I've set the max fan speeds based on what I know of the Max RPMs of the fastest fan in each zone, and their tolerances:

Code:

$cpu_max_fan_speed    = 2200;  #  2000 + (2000 * 0.1) = 2200
$hd_max_fan_speed    = 1700;     #  1500 + (1500 * 0.1) = 1650 rounded up to 1700

Here's a sample of the output:

Code:

2017-02-11 02:39:11: CPU Temp: 24.0 <= 35, CPU Fan going low.
2017-02-11 02:39:16: Maximum HD Temperature: 30
2017-02-11 02:39:16: Drives are cool enough, going to 30%
2017-02-11 02:39:28: CPU fan speed should be low, but 1900 > 1760.
2017-02-11 02:39:28: HD fan speed should be low, but 1400 > 1360.
2017-02-11 02:39:45: CPU fan speed should be low, but 1900 > 1760.
2017-02-11 02:39:45: HD fan speed should be low, but 1400 > 1360.
2017-02-11 02:39:50: Resetting BMC

I've read through the entire script, and don't see anything that looks like it should be kicking the fans back up to 100% duty-cycle. So, my guess is the BMC may still be interfering. Does that sound reasonable, or is there something I've missed?

 

[Kevin Horton]

Perhaps the BMC fan lower speed thresholds are too high. If the fan speed hits the BMC low speed thresholds, the BMC thinks there is a problem and it commands full speed. What is the output of ipmitool sensor | grep FAN ?

 

[ZeroBit]

Perhaps the BMC fan lower speed thresholds are too high. If the fan speed hits the BMC low speed thresholds, the BMC thinks there is a problem and it commands full speed. What is the output of ipmitool sensor | grep FAN ?

Good call! Somewhere along the line I must have set them back to the defaults of 300, 500 and 700. Thank you! :)

 

[Scampicfx]

Thank you for this guide! Helped a lot and save me from fans going up and down! :)

 

[Ericloewe]

Ericloewe submitted a new resource:

How To: Change IPMI Sensor Thresholds using ipmiutil - Introduction to IPMI sensor thresholds and ipmiutil

Many server motherboards, including the Supermicro X10 series, have their fans monitored by the BMC, which automatically attempts to recover stalling (of the motor, not in the aerodynamic sense) fans by spinning them up to maximum (PWM set to 100% speed). This behavior is determined by the sensor thresholds.

Supermicro X10 motherboard fan profiles are controlled via IPMI. Unfortunately, only very basic functions are exposed via its web interface and Supermicro's own IPMI...

Read more about this resource...

 

[Ericloewe]

Ericloewe updated How To: Change IPMI Sensor Thresholds using ipmiutil with a new update entry:

Initial migration

This is the original migration of the document to the Resources section. The original changelog follows:

Read the rest of this update entry...

 

[petesmc]

I'm using X11SSH-F motherboard and after the 2nd or 3rd reboot, i am now seeing critical error on FAN 2 saying that it is not reading and fan speed. The problem is, there is no fan connected to FAN 2. On FAN 4 which is also not connected, it correctly reports that there is no FAN connected.

No if i change FAN 3 or FAN 1, to be on FAN 2, the error just changes to one of those fan ports. Any idea if there is a command for me to reset the state of the sensors on these ports? With a critical warning, the fans will remain at 100%.

Peter

 

[Ericloewe]

Funny how life works, I was going through ebay looking for interesting deals on Supermicro motherboards and came across a listing that conspicuously pointed people towards post #33 of the original version of this thread about flashing the ME (albeit for a different problem). And this was written nearly eight years ago...