Hardware errors detected, likely a RAM stick, but which?

[Earan]

Hi Everyone, for a while now, my unraid server throws hardware errors, every now and then, which seem to be RAM related. I recently saw this on the screen it's attached to:

Here's the parts that I'm using:

• Supermicro MBD-H11DSi-NT-B
• 2x AMD Epyc 7301
• 8x16 GB of Kingston Server Premier KSM26RD8/16HAI DDR4-2666 regECC

 

One RAM stick seems to have issues, since the server reports as 112GB of Memory sometimes, and not 128GB after a reboot.

How do I find out which RAM stick it is, since those errors come up infrequently?

Are there other issues in the logs on the screen?

[JorgeB]

Looks at the system event log in the BIOS, or IPMI event viewer, there should be more info there.

[Earan]

Neither on the CLI with IPMITool from the Nerdpack, nor with the IPMI support plugin can I see any RAM Related issues. Downloading the full syslog I see quite a few events like the one on the screen, but all of them say,

[Hardware Error]: Corrected error, no action required.

also not really stating which RAMslot it is, or atleast, I cannot make it out.

 

this is one full event:

Dec 16 21:46:50 itXsvr kernel: mce: [Hardware Error]: Machine check events logged
Dec 16 21:46:50 itXsvr kernel: [Hardware Error]: Corrected error, no action required.
Dec 16 21:46:50 itXsvr kernel: [Hardware Error]: CPU:8 (17:1:2) MC15_STATUS[Over|CE|MiscV|AddrV|-|-|SyndV|CECC|-|-|-]: 0xdc2040000000011b
Dec 16 21:46:50 itXsvr kernel: [Hardware Error]: Error Addr: 0x0000000143092400
Dec 16 21:46:50 itXsvr kernel: [Hardware Error]: IPID: 0x0000009600050f00, Syndrome: 0x000067100a400401
Dec 16 21:46:50 itXsvr kernel: [Hardware Error]: Unified Memory Controller Ext. Error Code: 0, DRAM ECC error.
Dec 16 21:46:50 itXsvr kernel: EDAC MC2: 1 CE on mc#2csrow#1channel#0 (csrow:1 channel:0 page:0x973092 offset:0x400 grain:64 syndrome:0x6710)
Dec 16 21:46:50 itXsvr kernel: [Hardware Error]: cache level: L3/GEN, tx: GEN, mem-tx: RD

 

[JorgeB]

Take a look at the SEL in the BIOS, this is how it appears for one of my SM boards: