SATA kernel panics after some time on Supermicro AMD Epyc boards

[bytemine]

Hi,
we recently got 2 Supermicro servers running AMD EPYC 7702P on H12SSW-NT boards. Both have 2 SAMSUNG MZ7LH240HAHQ SATA SSDs connected to the onboard SATA ports via ASMedia ASM1062. After some time, both systems panic with WRITE FPDMA QUEUED issues, finally with I/O errors on both drives (the whole panic can be seeen in the attachment). The issues appear with 5.18-2/3 and also 5.4.24-1. In contrast, there are no issues with the NVMe devices.

We are in contact with our vendor, but we are also wondering if both boards have real hardware issues or if it's some kind of configuration issue like the random reboots mentioned in https://www.thomas-krenn.com/de/wiki/Random_Reboots_AMD_EPYC_Server

 

[aaron]

Since you did not mention it I need to ask if you did try to install the latest BIOS and firmwares (motherboard components, SSDs)?

 

[bytemine]

Hi,
thanks for your reply. Sorry for not mentioning it, we checked that already and the BIOS etc. are at the newest version and there are no firmware updates for the 883 DCT series yet.

 

[velocity08]

Hi,
thanks for your reply. Sorry for not mentioning it, we checked that already and the BIOS etc. are at the newest version and there are no firmware updates for the 883 DCT series yet.

Hi @bytemine

just wondering if this issue was resolved?

Where you running ZFS on the onboard ASMEDIA SATA?

""Cheers
G

 

[bytemine]

Hi @bytemine

just wondering if this issue was resolved?

Where you running ZFS on the onboard ASMEDIA SATA?

""Cheers
G

We were not able to resolve this issue, so we moved the disks from the onboard SATA to an also installed Trimode controller and never experienced these problems again.