High Fan Speed Issue on Lenovo ThinkSystem Servers

gduperrey

You can find information about kernel-alt here: https://docs.xcp-ng.org/installation/hardware/#-alternate-kernel

RIX_IT

bleader

Just did that on 8.2 (kernel-alt.x86_64 0:4.19.265-1.xcpng8.2), not the testing one yet.

Got a few errors on startup, not related afaik, but still fyi:

That didn't make a difference as it seems:

I'm on UEFI 1.41, could try the most recent somewhat later once people leave their offices.

LennertvdBerg

gduperrey said in High Fan Speed Issue on Lenovo ThinkSystem Servers:

https://docs.xcp-ng.org/installation/hardware/#-alternate-kernel

I tried installing the kernel-alt using

yum install kernel-alt

However I receive the following error;

python: can't open file '/opt/xensource/bin/updategrub.py': [Errno 2] No such file or directory
warning: %postun(kernel-alt-4.19.227-5.xcpng8.3.x86_64) scriptlet failed, exit status 2

kernel-alt failure.png

rmaclachlan

RIX_IT I tried kernel-alt before rolling back the uefi and it didn't help. Also tried the new Xen beta stormi posted (4.19 was it?) in case something was added but that didn't fix it either.

Thanks for trying the new Lenovo fw - I saw it come out and was going to test when I had time but now I don't need to!

Gheppy

Did it come fully equipped?
Did you install Lenovo signed HDD in it? Or you have compatibles HDD.
If you have compatibility, try to put a HDD with (original) Lenovo firmware.
I had the same problem with an Lenovo ThinkSystem SR630, I had installed a Lenovo non-firmware HDD, all of them. I bought one with Lenovo firmware, I installed and the problem disappeared.

rmaclachlan

Gheppy Our SR635v3 are running factory installed Lenovo hardware, although I did test swapping out the Broadcom 57504 OCP NIC with Intel X710 but the fan issue persisted. Thanks for the suggestion!

LennertvdBerg

Gheppy I just installed Ubuntu 22.044 LTS with kernel 5.15.0-102-generic just to test if there could be anything like a 'vendor lock'. Using Ubuntu I just see my memory temperatures and all my fan speeds are around 6000 rpm. So it really seems to be something with XCP and Lenovo.

Gheppy

Try to install lm-sensors and see what it detects.
It seems to me that the kernel it does not have drivers for IPMI BMC KCS.
My fans have 8000 - 8100 rpm

yum -y install lm_sensors
sensors -v
sensors-detect
sensors

LennertvdBerg

Gheppy I've just reinstalled xcp-ng-8.3.0-beta2 after my Ubuntu experiment and installed lm_sensors. The output is indeed:

Driver `to-be-written':
  * ISA bus, address 0xcc0
    Chip `IPMI BMC KCS' (confidence: 8)

Note: there is no driver for IPMI BMC KCS yet.
Check http://www.lm-sensors.org/wiki/Devices for updates.

No modules to load, skipping modules configuration.

Unloading i2c-dev... OK
Unloading cpuid... OK

The complete output is:
Screenshot 2024-04-09 at 11.57.06.png

What will be the solution for this?

Gheppy

We need to talk to the XCP-ng team and see if they can provide the module for the xen kernel.

LennertvdBerg

olivierlambert can you help us with providing the module for the xen kernel, which Gheppy is talking about?

rmaclachlan

LennertvdBerg The above message is indicating there is no driver written for IPMI BMC KCS for the LM_Sensors application, not that a kernel module is missing in XCPNG.

You can read more about this on the lm_sensors github issue https://github.com/lm-sensors/lm-sensors/issues/69

If you wish to view sensor data in XPCNG you can do so through IPMI still using ipmitool

ipmitool sensor

This will list the sensors in the server.

DIMM 1           | 0x0        | discrete   | 0x0080| na        | na        | na        | na        | na        | na
DIMM 1 Temp      | na         | degrees C  | na    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 2           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 2 Temp      | 23.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 3           | 0x0        | discrete   | 0x0080| na        | na        | na        | na        | na        | na
DIMM 3 Temp      | na         | degrees C  | na    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 4           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 4 Temp      | 25.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 5           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 5 Temp      | 26.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 6           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 6 Temp      | 26.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 7           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 7 Temp      | 26.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 8           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 8 Temp      | 26.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 9           | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 9 Temp      | 25.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 10          | 0x0        | discrete   | 0x0080| na        | na        | na        | na        | na        | na
DIMM 10 Temp     | na         | degrees C  | na    | na        | na        | na        | 85.000    | 87.000    | 91.000
DIMM 11          | 0x0        | discrete   | 0x4080| na        | na        | na        | na        | na        | na
DIMM 11 Temp     | 24.000     | degrees C  | ok    | na        | na        | na        | 85.000    | 87.000    | 91.000

The issue is when upgrading the UEFI we start seeing the sensor data read NA for the RAM modules which spins up the fans on the server, I don't know how to determine what on the OS is causing that but it sounds like something is trying to read that information and is locking up the sensor.

deric created this issue in lm-sensors/lm-sensors

closed Driver for IPMI BMC KCS #69

LennertvdBerg

rmaclachlan Thanks. I'm also unsure how we can determine what in the OS is causing this issue. Are there other installations or modifications we could try to help isolate the problem, such as another Linux distribution with the same kernel, to see if it's a kernel-related issue? gduperrey or olivierlambert any suggestions how we can help the team with identifying this?

olivierlambert

I have no idea, sorry. Hopefully someone could tell us what we need to add in XCP-ng to fix it.

ThierryEscande

LennertvdBerg To verify that the IPMI drivers are not messing up with the firmware, you could try to blacklist the IPMI modules.
Create the file /etc/modprobe.d/blacklist-ipmi.conf containing the following and reboot

blacklist ipmi_si
blacklist ipmi_devintf
blacklist ipmi_msghandler

Actually add a line for whatever modules the command lsmod | grep ipmi gives you

LennertvdBerg

ThierryEscande . When I do lsmod | grep ipmiI get the following results

ipmi_si                65536  0 
ipmi_devintf           20480  0 
ipmi_msghandler        61440  2 ipmi_devintf,ipmi_si

So, I created the file with vi /etc/modprobe.d/blacklist-ipmi.conf and added the following:

blacklist ipmi_si
blacklist ipmi_devintf
blacklist ipmi_msghandler

I saved the file and rebooted the system using shutdown -r now. However, I still don't see the memory temperatures in Xclarity, and the server's fans are still running at over 13,000 RPM. The system is running XCP-NG 8.3 beta 2 with kernel 4.19.0+1.

ThierryEscande

We have a new kernel-alt release available for testing purpose. This is only for XCP-ng 8.2.1 for now. The package for XCP-ng 8.3 should land soon.

THIS IS A TEMPORARY REPO FOR TESTING PURPOSE ONLY

To install this kernel-alt package:

Edit the xcp-ng.repo file

# vi /etc/yum.repos.d/xcp-ng.repo

Add the following lines at the end of the file:

[xcp-ng-tescande]
name=XCP-ng 8.2 tescande User Repository
baseurl=https://koji.xcp-ng.org/repos/user/8/8.2/tescande1/x86_64/
enabled=0
gpgcheck=0
priority=1

Install the kernel-alt package

# yum --enablerepo=xcp-ng-tescande install kernel-alt

Reboot and select XCP-ng kernel-alt 4.19.309 at the grub screen.

RIX_IT

ThierryEscande
I've upgraded to the kernel-alt 4.19.309 while still using the old UEFI (kae110k 1.41). Currently the IPMI modules are not blacklisted in the modprobe config.

The fan speeds remain stuck at 9k RPMs, consistent with both the stable and previous alt kernel versions.

I also attempted to upgrade again to the newest UEFI (kae118m 4.11) but didn't notice any discernible difference in fan behavior - here it's still around 13k RPMs.
The ipmitool output also didn't change from the one rmaclachlan provided.

If there are any other suggestions for testing or specific logs you'd like me to provide, please let me know.

ThierryEscande

RIX_IT Thanks for your feedback.

Can you please share the output of dmesg and xl dmesg ?

ThierryEscande

Even if this will probably be same as with XCP-ng 8.2, the kernel-alt package is available for 8.3.

Same steps as in my previous post, except for the repo URL in xcp-ng.repo:

[xcp-ng-tescande]
name=XCP-ng 8.3 tescande User Repository
baseurl=https://koji.xcp-ng.org/repos/user/8/8.3/tescande2/x86_64/
enabled=0
gpgcheck=0
priority=1