I just moved my unraid config to a new machine -

amd ryzen 4650g + B550 motherboard.

my array is 4 8TB HDDs (3 + 1 parity).

the issue I'm seeing is that the computer freezes when the disks are spinning down, either when I manually spin down or when they do it themselves after the timeout.

it freezes/crashes in a way that the monitor is no longer receiving a signal and the keyboard won't respond.

it won't even respond to a hard shutdown (long press on the power button). I need to physically cut the power at the power supply to shutdown and reboot.

there is no message in the syslog. I've mirrored it to the flash drive and watched i using tail -f /var/log/syslog.

there are some drives I can spin down once and it's ok, others that crash the machine first time I try to spin down.

if I spin down 2 disks that would spin down fine separately, it will crash.

for example (sdc is the parity, sdb is a drive):

spin down parity, spin back up, all fine

spin down drive, spin back up, all fine

spin down drive, down parity - crash.

I should note I can hear "disk reading" noise before the crash.

and this is all that's saved in the log (last line is after cutting power & rebooting):

 

Jan 17 19:44:42 Media  emhttpd: spinning down /dev/sdc
Jan 17 19:44:47 Media  emhttpd: spinning up /dev/sdc
Jan 17 19:44:53 Media  emhttpd: read SMART /dev/sdc
Jan 17 19:45:04 Media  emhttpd: spinning down /dev/sdb
Jan 17 19:45:10 Media  emhttpd: spinning up /dev/sdb
Jan 17 19:45:16 Media  emhttpd: read SMART /dev/sdb
Jan 17 19:45:20 Media  emhttpd: spinning down /dev/sdb
Jan 17 19:45:25 Media  emhttpd: spinning down /dev/sdc

Jan 17 18:45:54 Media kernel: Linux version 5.19.17-Unraid (root@Develop) (gcc (GCC) 12.2.0, GNU ld version 2.39-slack151) #2 SMP PREEMPT_DYNAMIC Wed Nov 2 11:54:15 PDT 2022

 

the bios is updated to the latest version. MB is gigabyte auros pro.

 

I tried setting the power supply idle control to "typical current idle" - didn't help.

I disabled c-states globally - didn't help.

I removed one RAM stick (had 2x16 ECC 3200) and downclocked it to 2133- didn't help.

it seems the issue was the PSU.

while it works fine with my old core 2 duo setup, it doesn't work with the new ryzen one.

oddly enough, it refused to work when "underloaded" as opposed to being overloaded - it would not POST unless at least 1 hdd was connected and would crash unless more than 2 were.

not sure what exactly is wrong there, perhaps the second 12v cpu rail that isn't used in the core 2 mobo is faulty, but connecting a different more moden PSU seems to have fixed the issue.