MicroServer Gen10 CPU temperature

[ThomasHall]

I purchased a Microserver Gen10 (EDIT: the 4-core version) a while ago and I'm running FreeNAS on it with 4 disks. I recently noticed the fan speeding up and making a lot of noise, and the temperature seems abnormally high on the CPU.

Peak utilization during a period I was looking at was 33%, while the temperature went slightly above 85C. The CPU during the same period was generally around 10-15% utilization, while the temperature was around 70-75C. During idle, utilization is below 5% and the temperature is around 30C. Ambient temperature is around 21C and the airflow to the server is not restricted in any way.

This doesn't seem neither normal nor reasonable under the circumstances, as I can't imagine the CPU getting anywhere close to 100% utilization without melting if this kind of temperature is expected behavior. I'm tempted to open it up and see if someone forgot to apply thermal paste to the cooler or something, but I don't want to void my warranty either, especially not if something actually is wrong with the server. Any advice on what I should do?

[AmRa]

Greetings,

Please refer HPE ProLiant MicroServer Gen10 - Troubleshooting Hardware, section : Fans are running at a higher than expected speed

https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-a00036944en_us&docLocale=en_US#N10235 

Symptom

The fans are running at a higher speed than expected.

Cause

The system temperature sensor reading is high and within approximately 10°C of the maximum operating temperature.

An air baffle or blank is missing or not installed properly, causing a disruption in the airflow.

The processor heatsink is not installed properly.

Action

1. Update the server to the latest firmware version, such as system BIOS, option firmware, etc.

2. Check the temperature. If a temperature sensor is high and within approximately 10°C of the maximum operating temperature, fan speeds can be higher than normal.

3. Verify that all air baffles and required blanks, such as processor heatsink blanks, power supply blanks, etc., are installed.

4. If unable to isolate the issue, please log a support ticket with HPE support team for further troubleshooting.

[ThomasHall]

I have read that article and it is of little to no help.

1) I can't update BIOS as it, for reasons I cannot fathom, requires me to provide warranty information or whatever it is that I don't have just to download. I have the server itself and 3 A4s of text which consists mostly of linking me to documents and downloads that I can't access. With the exception of the packaging material, that's all that was delivered.

2) I checked the temperature. 80+ degrees C when the CPU is barely at a quarter of the max is beyond anything I've ever seen. Even 30-40 degrees when it's basically idling is quite horrendous given the specs. When the CPU actually starts doing something, the fan spins up and the server is louder than my desktop which currently contains a total of 6 fans yet still manages to keep its CPU cooler.

3) I have no idea what "air baffles and required blanks" refers to. Short of removing components or the use of either a welder or bolt cutter, I don't see how I could do anything to alter the air flow in the case. Everything is factory default as far as I'm aware, and I got no spare parts to install.

4) I've already removed the heatsink and replaced the thermal paste (the factory application was shoddy, I might add). I'm perhaps 5C lower at most now, so not much of an improvement, although the fan doesn't seem to spin up quite as often so perhaps it keeps the PWM down instead of the temperature.

I'm at a loss here. Either the temperature sensor is faulty and reports a way higher temperature than it actually is, the CPU is faulty and somehow produces way more heat than it should, or the gen10 is simply a garbage product. I've even started looking at the possibility of installing some kind of fan for the CPU heatsink, but it's a tight fit which leaves few options, would require several power adapters and some custom fan mounting job, and should in no way be required of me when I've purchased a product that promised me it could do far more than I'm asking of it.

[Torsten.]

Consider to check if you can get an older BIOS, it is maybe still newer than yours.

https://support.hpe.com/hpsc/swd/public/detail?swItemId=MTX_fbd76116099d444fa16a7263bf

[ThomasHall]

That particular BIOS is not. According to the startup information, the BIOS is version ZA10A360 with build date 06/26/2018. I'm running FreeNAS on it anyway (FreeBSD-based), so the Windows version would probably not be the one to use. I'm not sure if the Linux one would work since there's at least some binary compatibility between them, or if it's "other", whatever that includes.

[AmRa]

The BIOS version ZA10A360 (06/26/2018) is the latest BIOS version for MicroServer Gen10.
However in the older version of MicroServer Gen10 BIOS version (ZA10A340 (15 Mar 2018)) there was major fix available with reference to HW thermal alert and Change thermal alert's temp low limit to default value 0x81(-127).

Hence check by downgrading BIOS version to ZA10A340. Download OS Independent BIOS version and flash BIOS firmware through offline method.

Type: BIOS (Entitlement Required) - System ROM
Version: ZA10A340(15 Mar 2018)
https://support.hpe.com/hpsc/swd/public/detail?swItemId=MTX_6b7e60c1fd1d4b49a657c39ea6#tab1

Steps to BIOS firmware through offline method:

1. Download the zip file to a directory on the target server.
2. Unzip the file to a temporary directory or USB key.
3. Switch to the "System_BIOS_v_ZA10A340_for_MicroServer_Gen10_UEFI" folder and verify that the following files are extracted:
EFI folder
Instruction.txt
releasenote.txt
ZA10A340.rom
4. Reboot the system and boot to the embedded UEFI shell.
5. Select the target device containing the downloaded firmware by issuing "fsx:", and then press Enter ("x" is the number of the target storage device number in the UEFI shell).
6. Switch to the "EFI" folder.
7. run "flashbios.nsh" and press Enter to start the firmware update.
8. After the firmware update process is completed, power cycle the system to ensure that the update is applied and to maintain hardware stability.
9. Check the BIOS settings and if needed, configure these settings to restore the user customized values.

If still issue continued capture system temperature in the Aptio Setup Utility.

1. Press F2 or Del (during POST), to access Aptio.

2 Select Advanced > Hardware Health Configuration, and then press the Enter key.

3 Capture hardware monitor temperature screen and share with us.

[ThomasHall]

Again, I can't download BIOS updates or anything else marked as requiring entitlement. I don't know what it expects me to do to get it and all manners of support have similar requirements; information that I must enter that I don't have as far as I'm aware. I purchased the server from an online store, received the delivery, kept everything that was in the package (what little there was), and that's it. How this could not implicitly include such a basic service as being able to download firmware updates is frankly beyond me.

[Torsten.]

If still under warranty, you need to registrr the serialno to your passport profile.

[AmRa]

Please refer HPE article “how to Link Contracts and Warranties to your HPE Support Center User Profile to Access HPE's Protected Knowledge and Downloads.”

https://support.hpe.com/hpsc/doc/public/display?docLocale=en_US&docId=emr_na-c04055475

Also refer HPE ProLiant Servers Firmware Access Update article.

https://support.hpe.com/hpsc/doc/public/display?docId=c04044353

[britcons]

Was there any further feedback on this problem? I have exactly the same thing going on. Temp goes up to 90 degrees C.