DL385 Gen11 AMD 9174F Fan Noise

[sdrake48]

Anyone else have a DL385 Gen11 with AMD 9174F (4.1Ghz) processor with extremely loud fan noise?

[TVVJ]

Hello,

You may refer to section "Fan and thermal issues" of the "Troubleshooting Guide for HPE ProLiant Gen11 servers" and perform them to check if it is of help.

Regards,

[sdrake48]

@TVVJ 

Cooling set to Optimal Cooling.  All other settings are for more cooling, more noise.  Any time the 6 fans in this server are over 35%, the noise is louder than any other HP server we have - we have every generation since Compaq Proliant 2500 and 1600 in 1997. 

[Bunsol]

Check if there is any component is degraded as degraded or failed components can cause fan to spin fast. Make sure firmware are updated to latest. This being fairly new generation we would recommend you to log a support ticket if issue persists.

[sdrake48]

@Bunsol 

Nothing degraded, nothing failed. Both DL385 and ML350 Gen11 servers do this...especially during POST at about the time F9 appears.  When this happens I can feel the blowing air 15 feet away.   But the DL385 small fans are the nosiest.  I have a support ticket for this issue.  Sent many AHS logs with and without anything in pcie slot3.  ILO shows no temperatures for any pci slot.  While working this issue via my case, there is no visibility to the community. I'm looking to see if everyone with this generation server has the same complaint.

[jdacircle]

Hi sdrake48

Yes, we have the same annoying problem here. A lot of noise and nothing happens on the server. Did you manage to get an answer from HP? I was thinking too of opening a case.

[Vinky_99]

@sdrake48 @jdacircle 

I can provide you with some general suggestions to address excessive fan noise in servers:

>> Ensure that you have the latest firmware and BIOS versions installed for your server. Sometimes, updates can include fixes or improvements related to fan control and noise.

>> Check the fan speed settings in the BIOS or server management software. Adjust the fan speed profiles to balance cooling performance and noise levels. You may need to experiment with different settings to find the right balance for your needs.

>> Dust accumulation can impede proper airflow and cause fans to work harder, resulting in increased noise. Regularly clean the server's air vents, heat sinks, and fans to remove any dust or debris that may have built up.

>> Ensure that all components are properly installed and securely connected. Loose or incorrectly installed components can cause increased fan noise due to poor thermal conductivity or airflow.

>> Keep an eye on the server's temperature readings using appropriate monitoring tools. Excessively high temperatures can cause fans to run at higher speeds. Consider optimizing the server's cooling system if temperatures consistently exceed recommended levels.

>> If the fan noise persists and none of the above steps resolve the issue, as you have already open the support case, please co-ordinate with the team.

Remember that server noise can vary depending on the workload and configuration. However, if you believe the noise level is significantly louder than expected or compared to other similar servers, it's worth investigating further to ensure there are no underlying issues impacting the fan performance.

[sdrake48]

|Vinky_99  Have you been near a DL385 Gen11 9174F AMD processor?  Heard it boot?  Listen to it at idle...and then when simply opening a terminal window?  Try it.  Its not like any HP Proliant server Gen1-G10.

[jdacircle]

It is a complete new server. However, an update to iLO6 v1.41 seems to have resolved the issue. Currently running fan speed below 20% which should be stated as normal I assume. Unfortunately i wasn't able to apply the latest bios, cause the SPP download is giving me some errors. Will examine later, need to get these server up and running. I will keep you posted.

[sdrake48]

At one point I was told this problem had been escalated to the Power and Thermal Group at HPE and would be addressed by a future firmware update.  But then they shipped me a new PCIe slot 3 board for replacement (something about no pcie slot 3 temp reading in AHS).  The replacement board had no effect.   I have not heard back since I provided another AHS log after replacing that part.   Another open case wouldn't hurt.  

[sdrake48]

They replaced the main board (system board) but it won't recognize MR408 in ocp slot, so won't boot. Fans still noisy. Server is now powered off awaiting 3rd system board.

[Itzikbenabu]

I have the same problem with ML350 gen 11 After I shut down the OS the fan started to make a very loud noise I needed to physically disconnect the power cable so it would stop.

also when I connect a third-party NVME PCIE card the noise starts from the ILO POST stage and does not stop.

I have installed the third-party NVME PCIE card in ML 350 Gen 10 and I don't have the fan noise at all.

[zzif]

You can adjust gen11 fans with REST API:

curl --request PATCH --url 'https://<ILO IP>/redfish/v1/Chassis/1/Thermal/' --user 'Administrator:<PASS>' --header 'content-type: application/json' --insecure --data '{"Oem": {"Hpe": {"FanPercentAdjust": 50}}}'

[Itzikbenabu]

@zzif can you also give me the syntax for how to readjust it to default?

Thanks,

Itzik

[Itzikbenabu]

I don't think it work on ILO 6.
II did GET to the API endpoint you post I get all the settings but the setting you post don't show in the list.



נשלח מה-Galaxy שלי

[zzif]

It works with the latest ILO6 version. I have tried with DL325 gen11. Removing power cords will reset the value back to defaults.

[Itzikbenabu]

I am running Firmware Version 1.52 and when I try your command I get this message:

{"error":{"code":"iLO.0.10.ExtendedInfo","message":"See @Message.ExtendedInfo for more information.","@Message.ExtendedInfo":[{"MessageId":"Base.1.17.Success"}]}

[Itzikbenabu]

You're right it works  

I looked for the definition in the GET response but it does not appear there, probably it is a definition that is not saved there.

Thanks

[Itzikbenabu]

Do you know how can I set this OEM API curl --request PATCH --url 'https://<ILO IP>/redfish/v1/Chassis/1/Thermal/' --user 'Administrator:<PASS>' --header 'content-type: application/json' --insecure --data '{"Oem": {"Hpe": {"FanPercentAdjust": 50}}}' to be permanent?

If I do ILO reset or the server loses power I need to rerun the API.

Thanks,

Itzik

[sdrake48]

Not interested in ilorest cli solution.  There is no temperature anywhere near critical threshhold anywhere in the server.  Firmware should be modified to slow the fans down.  Maybe investigate why they spin even faster when any pcie card is plugged in.

[Itzikbenabu]

HPE don't want us to connect 3 party cards to the system so they increase the fans when the system recognized non supported cards. I open case in hpe support this is there answer regarding the ilo redfish api command this is the only way to lower the fans but the problem is that you need to reapply it every time you do cold reboot or unplug the power cable I also open support case about this and there answer was that the ilo api can't be save after reboot this is by design.

HPE is not like other corporation like cisco for example they allow to connect 3 party sfp hardware by running cli command and we need to confirm that when there is problem we don't have support until we remove the unsupported hardware.

My problem was with PCIE to NVME M.2 adapter when I connect it the system fans was on 60% this make a lot of noise.



נשלח מה-Galaxy שלי

[Nikolai1]

Getting same error message:

{"error":{"code":"iLO.0.10.ExtendedInfo","message":"See @Message.ExtendedInfo for more information.","@Message.ExtendedInfo":[{"MessageId":"Base.1.17.Success"}]}}

How did you get it to work?

[Itzikbenabu]

It work you got successful.



נשלח מה-Galaxy שלי

[sdrake48]

In my case, fans running at 58% with server idle and very loud. Noise reduced by physically removing MR408i-o OCP RAID controller. For other reasons we switched to SR932i-p RAID controller, but left MR408i-o in the OCP slot (disabled in bios).  Physically removing the OCP card results in fans running at 18% during idle.  Still not 36dBA as listed in the quickspecs (more like 60 dBA), but the complaining stopped. Consider the issue resolved.  Now if I could only find a way to reduce the 58% fan speed on a ML350 Gen11 in the same basic configuration (removing the MR408 from this server had no effect). Since the fans are larger than DL385 fans, they are not as noisy, but noise could be reduced.  

[sdrake48]

Perhaps HPE Power and Thermal considers MR408i-o a 3rd party card?