Failed DIMM Slot full build

Hardware
1

Hello,
I’m looking for additional troubleshooting.

When I first fired up my HL15, I did so " as configured" - full build ( 2 x 8Gb memory). Everything looked fine.

The next day I swapped out the original two DIMM’s for 6 x 32Gb sticks.

I got an error before boot that DIMM D1 was failing, so I hard pressed the power button and shut it down.
I’ve tried swapping all the other DIMMS into that slot and get the same error message. If I let the system boot, only 5 DIMMs are reported.
Today, I loaded the two original DIMMs into DIMM D1 and C1. I still got the same message that DIMM D1 is failing

HL15_Memory
HL15_Memory1591×174 54.8 KB

Any other ideas?

2

Troubleshooting ideas:

  • BIOS update, got latest?
  • Memtest 86
  • dust clean the DIMM slots with some compressed air, before loading them up; half the errors I ran into were just this issue…

I had a situation where I could memtest pass two dimms at a time, but not 4 together, until I updated the bios (on a DDR5 mobo)

YMMV

3

Hi @daemon1001, I’m sorry to hear you having issues with the drives. if you want to reach out to info@45homelab.com we can get a support member to reach out and troubleshoot your issues.

2 Likes
4

** Resolved **

Removed CPU cooler. Re-installed CPU cooler. Result: all 6 DIMMs recognized A1-F1.
My conclusion: CPU cooler middle bolt, closest to DIMM D1 not tight enough.

Also, while removing CPU cooler, I noticed motherboard screw missing between the 8-Pin power and 24-Pin power connectors. Standoff is under there. Easy fix. No big deal, but maybe let the assemblers at 45Drives know. I have not found a loose screw in the case.

Now onto fan replacements.

2 Likes
5

Hi @daemon1001,

Thank you for sharing this detail. I will forward this information to our team and make sure it does not happen again.

Please share your finds after your fan replacement.

Thank you,

6

I just want to share another data point on perhaps a similar issue. I bought a fully built HL15 when the product launched in 2023 with 32GB (2 x 16 GB) RAM. In Q3 of 2024, I upgraded the system to 96 GB (6 x 16 GB) with DIMM slots C1/B1/A1/D1/E/F1 populated and at first everything was behaving normally.
Then earlier this year, one day when I logged into TrueNAS, I noticed that the available memory was down to 80GB. In the IPMI, the server health page was showing DIMMA1 as not present even though a RAM stick was installed in the slot. I proceeded to swap RAM sticks in the hope that it was just bad RAM but the system would not recognize any RAM in DIMMA1 even when a known working stick was inserted.
Life got in the way , 80GB was still quite sufficient for my needs, my system probably was out of warranty, so left it at that… until last week when I decided to upgrade the CPU. I used the opportunity to remove all the RAM sticks, clear any dust, reseat the CPU but the issue still persisted. I then proceeded to try the RAM stick in the grey DIMM slots A2 / D2. The RAM was not detected in A2 but works in D2.

So now my system detects 96GB RAM but with DIMM slots A1/A2 vacant, which is fine for my homelab, but in theory, the performance of the system could be affected since the memory isn’t populated as per Supermicro’s memory population sequence.

So did the DIMMA1/A2 slots on by motherboard mysteriously fail? I don’t know.My last step would be to update the motherboard firmware and BIOS since I am still running what the system originally shipped with:
Firmware Revision: 01.73.14
Firmware Build Time: 11/09/2021
BIOS Version: 3.5
BIOS Build Time: 06/01/2021

But I am undecided whether I should go through with this risk since my system, despite the previously described issue, seems to be working fine and I have no intention to upgrade the system RAM past 96GB.

Update: Upgraded the BMC firmware and BIOS to the latest version from Supermicro’s website but to no avail… RAM is still not detected in DIMM slots A1/A2. Will revert to installing it in slot D2.