On Sun, Apr 09, 2023 at 07:07:39 -0400, Genes Lists wrote: > On 4/8/23 19:55, Luna Celeste wrote: > > Sounds very annoying You have no idea. > Couple of thoughts: > > - does the system journal reveal anything? If not it seems less like a > kernel crash and more like graphics type hangs - good to avoid any nvidia > drivers. Not just kernel but 'other processes' - journal --since -1h or > whatever. There's nothing in any of the logs. Nothing in dmesg, nothing in journalctl. The system works fine, hangs, I reboot it, and the next messages I see are the system booting. This is an AMD machine with no NVIDIA hardware / software whatsoever. Unlike in Jonas's case, my capslock does not flash when the machine hangs--unless this is something that only happens a few times immediately after a crash. > - Can you keep a remote ssh from another computer - and when hang occurs > is the remote ssh also dead or just the console/monitor keyboard? I almost always have a mosh connection from my laptop to the Linux machine, and it too hangs. I've gotten disappointingly familiar with mosh's blue/white "can't reach remote machine" message. > - Definitely worth checking memtest86+ as well as smarctl test your > disk(s). While bad mem is possible your symptoms dont seem quite consistent > with that - i'd check anyway. This was one of the first things I tried! I've run both the proprietary and the open source versions and both show no errors. Beyond this, the system is using ECC memory, so I would expect to see checksum errors or the like in the logs, but again, nothing. I've also checked smartctl, and again nothing. Also, as I said, I'm running ZFS (and only ZFS), and I would expect it to tell me that there are checksum errors *long* before smartctl shows anything. > - is your CPU AMD? Can you turn off sleep and see if it makes a > difference - there were some issues with s2idle and AMD, though I thought > the were worked around/fixed back in 6.1. Also is your bios fully up to > date? As said above, yes, the CPU is AMD, but I wasn't clear in my original message, I'm sorry. *Only* the display is sleeping. I do not have any other power saving measures configured. And, even if that was the case, my research shows that the AMD s2idle bug was fixed in 5.15. This machine is running 6.2.10. I'm not sure about the BIOS, will have to check that. Thank you everyone for helping me dig into this. It has been quite frustrating! I am thoroughly out of ideas, so anything helps. -- Cheers, Luna Celeste
