cockpit-project / cockpit

Cockpit is a web-based graphical interface for servers.
http://www.cockpit-project.org/
GNU Lesser General Public License v2.1
10.95k stars 1.1k forks source link

No CPU temperature showing #18799

Open wadrie opened 1 year ago

wadrie commented 1 year ago

Explain what happens

I'm using Cockpit on an Orange Pi 5 running Armbian. I installed the backport versions of cockpit and cockpit-pcp. On the metrics page, there is no CPU temperature showing: image

From other Github tickets about this issue I learned that this is useful information: grep -r . /sys/class/hwmon/hwmon*/name /sys/class/hwmon/hwmon0/name:tcpm_source_psy_6_0022 /sys/class/hwmon/hwmon1/name:soc_thermal /sys/class/hwmon/hwmon2/name:bigcore0_thermal /sys/class/hwmon/hwmon3/name:bigcore1_thermal /sys/class/hwmon/hwmon4/name:littlecore_thermal /sys/class/hwmon/hwmon5/name:center_thermal /sys/class/hwmon/hwmon6/name:gpu_thermal /sys/class/hwmon/hwmon7/name:npu_thermal

grep -r . /sys/class/hwmon/hwmon*/temp* /sys/class/hwmon/hwmon1/temp1_crit:115000 /sys/class/hwmon/hwmon1/temp1_input:32384 /sys/class/hwmon/hwmon2/temp1_input:32384 /sys/class/hwmon/hwmon3/temp1_input:31461 /sys/class/hwmon/hwmon4/temp1_input:32384 /sys/class/hwmon/hwmon5/temp1_input:31461 /sys/class/hwmon/hwmon6/temp1_input:31461 /sys/class/hwmon/hwmon7/temp1_input:31461

There is nothing showing in the syslogs about this issue

Version of Cockpit

287-1~bpo22.04.1

Where is the problem in Cockpit?

Metrics

Server operating system

Ubuntu

Server operating system version

Armbian 23.02.2 Jammy with Linux 5.10.110-rockchip-rk3588

What browsers are you using?

Firefox, Chrome

System log

No response

jelly commented 1 year ago

On ARM we only checkfor cpu_thermal which does not seem like something you have. I am not sure what we should even use here for determing the cpu thermals as it's a Big.Little CPU. I don't see this board being supported yet upstream's u-boot or the Linux kernel so that might explain why the patched kernel uses a different thermal zone.

darmach commented 1 year ago

@jelly Rocky Linux 9 here, running cockpit 276.1 on Intel J1900 (x86-64) I can't see temps aswell:

image

root@nas ~ 0 $ grep -r . /sys/class/hwmon/hwmon*/name
/sys/class/hwmon/hwmon0/name:acpitz
/sys/class/hwmon/hwmon1/name:coretemp
/sys/class/hwmon/hwmon2/name:hidpp_battery_0
/sys/class/hwmon/hwmon3/name:nct6776
root@nas ~ 0 $ grep -r . /sys/class/hwmon/hwmon*/temp*
/sys/class/hwmon/hwmon0/temp1_crit:90000
/sys/class/hwmon/hwmon0/temp1_input:26800
/sys/class/hwmon/hwmon1/temp2_crit:105000
/sys/class/hwmon/hwmon1/temp2_crit_alarm:0
/sys/class/hwmon/hwmon1/temp2_input:44000
/sys/class/hwmon/hwmon1/temp2_label:Core 0
/sys/class/hwmon/hwmon1/temp2_max:105000
/sys/class/hwmon/hwmon1/temp3_crit:105000
/sys/class/hwmon/hwmon1/temp3_crit_alarm:0
/sys/class/hwmon/hwmon1/temp3_input:45000
/sys/class/hwmon/hwmon1/temp3_label:Core 1
/sys/class/hwmon/hwmon1/temp3_max:105000
/sys/class/hwmon/hwmon1/temp4_crit:105000
/sys/class/hwmon/hwmon1/temp4_crit_alarm:0
/sys/class/hwmon/hwmon1/temp4_input:47000
/sys/class/hwmon/hwmon1/temp4_label:Core 2
/sys/class/hwmon/hwmon1/temp4_max:105000
/sys/class/hwmon/hwmon1/temp5_crit:105000
/sys/class/hwmon/hwmon1/temp5_crit_alarm:0
/sys/class/hwmon/hwmon1/temp5_input:47000
/sys/class/hwmon/hwmon1/temp5_label:Core 3
/sys/class/hwmon/hwmon1/temp5_max:105000
/sys/class/hwmon/hwmon3/temp1_alarm:1
/sys/class/hwmon/hwmon3/temp1_beep:0
/sys/class/hwmon/hwmon3/temp1_input:40000
/sys/class/hwmon/hwmon3/temp1_label:SYSTIN
/sys/class/hwmon/hwmon3/temp1_max:0
/sys/class/hwmon/hwmon3/temp1_max_hyst:0
/sys/class/hwmon/hwmon3/temp1_offset:0
/sys/class/hwmon/hwmon3/temp1_type:4
/sys/class/hwmon/hwmon3/temp2_alarm:0
/sys/class/hwmon/hwmon3/temp2_beep:0
/sys/class/hwmon/hwmon3/temp2_input:43000
/sys/class/hwmon/hwmon3/temp2_label:CPUTIN
/sys/class/hwmon/hwmon3/temp2_max:80000
/sys/class/hwmon/hwmon3/temp2_max_hyst:75000
/sys/class/hwmon/hwmon3/temp2_offset:0
/sys/class/hwmon/hwmon3/temp2_type:4
/sys/class/hwmon/hwmon3/temp3_alarm:0
/sys/class/hwmon/hwmon3/temp3_beep:0
/sys/class/hwmon/hwmon3/temp3_input:30000
/sys/class/hwmon/hwmon3/temp3_label:AUXTIN
/sys/class/hwmon/hwmon3/temp3_max:80000
/sys/class/hwmon/hwmon3/temp3_max_hyst:75000
/sys/class/hwmon/hwmon3/temp3_offset:0
/sys/class/hwmon/hwmon3/temp3_type:4
/sys/class/hwmon/hwmon3/temp7_input:0
/sys/class/hwmon/hwmon3/temp7_label:PCH_CHIP_TEMP
/sys/class/hwmon/hwmon3/temp8_input:0
/sys/class/hwmon/hwmon3/temp8_label:PCH_CPU_TEMP
/sys/class/hwmon/hwmon3/temp9_input:0
/sys/class/hwmon/hwmon3/temp9_label:PCH_MCH_TEMP
wadrie commented 1 year ago

On ARM we only checkfor cpu_thermal which does not seem like something you have. I am not sure what we should even use here for determing the cpu thermals as it's a Big.Little CPU. I don't see this board being supported yet upstream's u-boot or the Linux kernel so that might explain why the patched kernel uses a different thermal zone.

Am I correct in assuming you mean in other systems something like "/sys/class/hwmon/hwmon1/name:cpu_thermal" exists, and you only check for that value?

If so, maybe an idea would be to make that configurable, so that in my case I could say: use "/sys/class/hwmon/hwmon1/name:soc_thermal" for displaying the temperature. This may be useful for everybody who wants to monitor more than just the CPU temp.

jelly commented 1 year ago

On ARM we only checkfor cpu_thermal which does not seem like something you have. I am not sure what we should even use here for determing the cpu thermals as it's a Big.Little CPU. I don't see this board being supported yet upstream's u-boot or the Linux kernel so that might explain why the patched kernel uses a different thermal zone.

Am I correct in assuming you mean in other systems something like "/sys/class/hwmon/hwmon1/name:cpu_thermal" exists, and you only check for that value?

Yes, for arm we check cpu_thermal.

If so, maybe an idea would be to make that configurable, so that in my case I could say: use "/sys/class/hwmon/hwmon1/name:soc_thermal" for displaying the temperature. This may be useful for everybody who wants to monitor more than just the CPU temp.

That is misleading the soc thermal isn't the CPU temperature

wadrie commented 1 year ago

OK, bad example, I might use bigcore0_thermal or something like that. My point is, if it's configurable, users can choose themselves if they want just the CPU temp, and what would be the source of that. You could even have multiple so you can monitor the CPU, GPU, SOC, whatever is available in /sys/class/hwmon/

darmach commented 1 year ago

@jelly I have the same issue (no temps displayed) although on different platform (x86-64). Should I open a separate issue? (details of my sensors are posted in previous post)

jelly commented 1 year ago

@darmach I am afraid that you might need to update to 286 or higher as that contains a fix for cpu sensor detection.

g6avk commented 1 year ago

@jelly sadly that fix/update never helped me, as I mentioned in my original BZ https://github.com/cockpit-project/cockpit/issues/18098 Running 292 here, must be something odd with AMD/Ryzen?

hovancik commented 1 month ago

Hi, I am on Debian GNU/Linux 11 (bullseye) aarch64 and I also do not see the temperature. Raspberry Pi 4 B.

grep -r . /sys/class/hwmon/hwmon*/name
/sys/class/hwmon/hwmon0/name:cpu_thermal
/sys/class/hwmon/hwmon1/name:rpi_volt
grep -r . /sys/class/hwmon/hwmon*/temp*
/sys/class/hwmon/hwmon0/temp1_crit:110000
/sys/class/hwmon/hwmon0/temp1_input:53556
jelly commented 1 month ago

@hovancik and which Cockpit version?

hovancik commented 1 month ago

@jelly ah, nevermind my question, I see I have some super-old version 239 https://packages.debian.org/bullseye/cockpit. sorry for bother