Proxmox 9.0 Ryzen 7 1700 crashes

[WolfAURman_IKrell]

Hi everyone! I am getting certain problems when proxmox is running on my device.

Configuration:
Lenovo m715s, ryzen 7 1700, 16gb ddr4, bios version from 2023 latest

Device freezes, or sometimes restarts. I see such errors in the logs:

root@virt-hideyoshi-01:~# journalctl -k -g "mce"
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: Machine check events logged
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: CPU 14: Machine Check: 0 Bank 5: bea0000000000108
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: TSC 0 ADDR 1ffffb1cc0336 MISC d012000100000000 SYND 4d000000 IPID 500b000000000
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1757958617 SOCKET 0 APIC e microcode 8001138
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: Machine check events logged
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: CPU 1: Machine Check: 0 Bank 5: bea0000000000108
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: TSC 0 ADDR 1ffffb2c3cab2 MISC d012000100000000 SYND 4d000000 IPID 500b000000000
Sep 15 23:50:20 virt-hideyoshi-01 kernel: mce: [Hardware Error]: PROCESSOR 2:800f11 TIME 1757958617 SOCKET 0 APIC 1 microcode 8001138
Sep 15 23:50:21 virt-hideyoshi-01 kernel: MCE: In-kernel MCE decoding enabled.

it is impossible to turn off power saving in the bios, I was able to set only these parameters in grub:

GRUB_CMDLINE_LINUX_DEFAULT="quiet pcie_aspm=off usbcore.autosuspend=-1 processor.max_cstate=5 intel_idle.max_cstate=5 idle=poll"

Memtest revealed no problems.

Does anyone know how to fix this? Or will it be fixed only by replacing the processor?

 

[dcsapak]

coincidentally i recently had also a ryzen 1700 processor that seems to be defective. it ran without problems for years, but recently i changed the power supply (that is good & new) and the processor would hang or reset. i replaced it with a different processor and the issues are gone.

there are multiple possible other solutions mentioned e.g. on the arch wiki: https://wiki.archlinux.org/title/Ryzen#Troubleshooting

so it seems these first generation ryzen chips often had a problem with halts/reboots/etc.

 

[WolfAURman_IKrell]

coincidentally i recently had also a ryzen 1700 processor that seems to be defective. it ran without problems for years, but recently i changed the power supply (that is good & new) and the processor would hang or reset. i replaced it with a different processor and the issues are gone.

there are multiple possible other solutions mentioned e.g. on the arch wiki: https://wiki.archlinux.org/title/Ryzen#Troubleshooting

so it seems these first generation ryzen chips often had a problem with halts/reboots/etc.

Oh, that's sad. I want to replace it with 7 2700 or 5 1600af. Does zen+ have such problems? Or should I not engage in necrophilia and replace the device?
I'll try to use this https://github.com/r4m0n/ZenStates-Linux I'll see if it helps...

 

[leesteken]

I still run a 2700X with PVE 9 (and PBS 4) and it still runs fine (but I don't run it daily). Ryzen CPUs draw less power on idle than older CPUs and there is a motherboard BIOS setting to increase this so older PSUs don't shut off (due to lack of power requested). Have you tried changing it? Or disabling C-States in the motherboard BIOS (which plagued the first generation of Ryzen CPUs)? Or updating the motherboard BIOS (which has fixes compatibilities with kernel 6.14 for some people)?

 

[WolfAURman_IKrell]

I still run a 2700X with PVE 9 (and PBS 4) and it still runs fine (but I don't run it daily). Ryzen CPUs draw less power on idle than older CPUs and there is a motherboard BIOS setting to increase this so older PSUs don't shut off (due to lack of power requested). Have you tried changing it? Or disabling C-States in the motherboard BIOS (which plagued the first generation of Ryzen CPUs)? Or updating the motherboard BIOS (which has fixes compatibilities with kernel 6.14 for some people)?

I have the latest BIOS from 2023. I can't disable C6 in the BIOS, there are no such settings, lenovo's bios has limited configuration options. I turned off python using the zenstates script with the command python3 zenstates.py --c6-disable It seems to be disabled, see screenshot. I'll be watching. But I think I'm leaning towards replacing the processor, eh