HPE Gen10 DL385 power issues

Computer Hardware & Server Infrastructure Builds
1

Hi,

I have a weird (possibly) power related issue with our HPE Server. Last 3-4 days it did it 2x. Randomly turns off the server and the iLO is unreachable. When I check it, the 2x PSU light is on, but I can’t turn it on.
After unplugging both cables (I tried with one only), the server restarts.

  • Proxmox has no information or log info about an OS triggered restart
  • today I updated everything with the latest SPP + BIOS to v4.02.

iLO has the following log, but nothing useful. When I try to run Hardware diagnostics, I got this error. (No matter if I run it from the BIOS or Intelligent Provisioning)

Screen Shot 2026-04-06 at 10.27.07 AM
Screen Shot 2026-04-06 at 10.27.07 AM732×451 68.8 KB

Screen Shot 2026-04-06 at 11.33.13 AM
Screen Shot 2026-04-06 at 11.33.13 AM1931×1167 386 KB

2

Do you have HP SmartMemory? Is the heat sink compond still good on the CPUs? Might be something there. Check iLO for the heat level, etc. Might be a dead end, though…

3

I have original, certified memory modules. They all report ok in iLO.

Regarding to the temperature, I don’t see any unusual high spots. I found a loos paper (just the small serial number sticker) and I changed the RTC battery just in case. (It was 3.11 V, the new has 3.2 V that should not be the problem.)

Screen Shot 2026-04-06 at 4.31.20 PM
Screen Shot 2026-04-06 at 4.31.20 PM1243×1089 238 KB

4

PCI Zone1 seems rather high. You have an add-in card in that slot? I wonder if you have an overheating problem for that card.

5

I was thinking about that. I did some tests, pushing the CPU to 50-60%, doing fio on the 10Gb card, but nothing happened.
I will do another attempt today, when I can observe it in person.

Pushing it for 20 minutes, only changed 67 to 73 C

6

Moving the network card down one PCI-E slot and slightly increasing the fan speed decreased the temp from 68 C to 54 C.

Usually the fans didn’t ramp up more than 17%, just occasionally. But it is in a temperature controlled room. Now the fans idle around 34%.

Only one thing left…convince HPE Support to talk with us on an hourly rate without purchasing a support contract.

Screen Shot 2026-04-07 at 11.26.23 AM
Screen Shot 2026-04-07 at 11.26.23 AM1256×1098 238 KB

7

Not sure if this is the fix? Worth a look..

8

Support said, I should start with minimal config (1 RAM stick / CPU) and remove everything. Gradually add back devices. (But I didn’t see the problem since Monday).

I removed all drives and I was able to start the built in diagnostic tool from the BIOS.

Which indicates that something is not OK with the Proxmox Install

Screen Shot 2026-04-07 at 2.57.52 PM
Screen Shot 2026-04-07 at 2.57.52 PM697×592 32.7 KB

9

Have you tried failing back to the previous firmware/BIOS to see if the issue persists on the previous version?

10

Yes, I tried. But still didn’t allow me to run the test.
However I was running it for a week and no issues. Probably the firmware update solved it. Even support was unable to tell me why. They just gave me the: “remove everything and gradually add back hardware” solution.

11

Thats actually good advice. Its a proper way to troubleshoot.

12

I agree, but I did all of that. For $310/hour I would expect that they check the logs in detail.
Also the only thing that I was able to remove is the 10 GbE card and RAM modules, but because the issue didn’t happen immediately waiting weeks between RAM sticks is not too efficient.

13

As a troubleshooting step, I changed the 2x500W PSU to 2x800W and updated every firmware again. In a week, I got the same result.
Now I will leave the server on, no load. Next steps are removing every RAM stick except one and the one addition (PCIe NIC).