system: aoostar wtr pro, 5825U 4 bay
ever since several updates ago, around 25.10.3-ish and later after all the kernel security updates, the system would hang if I did not disable C6.
on update, 25.10.6, the system would boot and freeze/hang. Tried 25.10.7, same issue
I saw the disk size stuff bugging out
then I saw ethernet loop flapping in the middle of my jellyfin stream/transcoding testing
Oct 09 08:36:39 snuwtrpro kernel: br-3e17519bc528: port 6(vethbdba881) entered blocking state
Oct 09 08:36:39 snuwtrpro kernel: br-3e17519bc528: port 6(vethbdba881) entered disabled state
Oct 09 08:36:39 snuwtrpro kernel: vethbdba881: entered allmulticast mode
Oct 09 08:36:39 snuwtrpro kernel: vethbdba881: entered promiscuous mode
Oct 09 08:36:39 snuwtrpro kernel: eth1: renamed from veth3c9cfa0
Oct 09 08:36:39 snuwtrpro kernel: br-3e17519bc528: port 6(vethbdba881) entered blocking state
Oct 09 08:36:39 snuwtrpro kernel: br-3e17519bc528: port 6(vethbdba881) entered forwarding state
this but repeated dozens of times, I think my router eventually kicked it off the LAN until I reset the router itself
then now it boots and immediately crashes/freezes, reverting to older kernel doesn’t work as it can’t find the debug kernel from prior releases for some strange reason
I did try live booting a ubuntu liveiso (latest) and it also freezes so I believe it’s a newer kernel issue.
I did update the BIOS to the latest available one on the Aoostar website and the machine no longer hangs but it did reset a few times on its own which should not happen (but at least it stopped freezing/hanging)
I tried ubuntu 26.04 liveiso old bios, hangs kernel 7.0
ubuntu 24.04 liveiso old+new bios, does not hang. kernel 6.8
I don’t know where to report this issue and I mainly use truenas, so please excuse the sloppy writing and tell me what logs/details to provide so this can be sent upstream. I’m also (of course) in the middle of pulling my hair out to see if I can make sure it’s stable before freaking out some more.
https://www.reddit.com/r/Kubuntu/comments/1vfjxi9/kernel_70028_29_shutdown_hangup_amdgpu_driver/
https://askubuntu.com/questions/1566809/ubuntu-26-04-lts-probably-unstable-driver-amdgpu
testing w/ blacklisting amdgpu module from kernel now, looks like that + current kernel version is the culprit.
truenas 25.10.7 kernel version 6.12.105-production+truenas need to blacklist amdgpu in kernel or else it’ll keep resetting
midclt call system.advanced.update ‘{“kernel_extra_options”:“amdgpu.dcdebugmask=0x10”}’ still crashes, not the problem
downgrading attempts:
25.10.7 - crash | kernel 6.12.105
25.10.6 - crash | kernel 6.12.99
25.10.4 - crash | kernel 6.12.91
25.10.3.1 - crash | kernel 6.12.33
25.10.3 - crash | kernel 6.12.33
25.10.2 - crash | kernel 6.12.33
I’m beginning to suspect something in the jellyfin container caused the crash. The machine hung when I was starting and stopping a transcode. Trying to clear the transcode dir by destroying the container and recreating it, also by stopping the jellyfin container so it stops running by default when the system starts up.
jellyfin container recreated (tmpfs + transcode folder deleted/destroyed), host not down
looks like it’s a bug in amdgpu driver for all versions/upstream, has nothing to do with truenas kernel versions it seems. jellyfin using ffmpeg using amdgpu should never crash the host like that.
starting jellyfin → host crashed again
starting truenas → no /dev/dri passed to jellyfin, amdgpu enabled for all other containers from before → no crash o_O
there might be 2 issues going on, hard stall/frozen, and a reboot that resets the machine
I suspect the one that resets the machine is the c6 disable that’s required and the hard stall is amdgpu, something going on with jellyfin+immich
might’ve been a combination of 4k media import/thumbnail generation/trickplay + transcoding, not sure. it also just crashed when I opened my ssh client to remote in for some reason.
but after that reboot (and the 12 hours in between where /dev/dri was removed from jellyfin and immich, I’m guessing allowing the background processing to be done on cpu) the machine is now stable again on latest 25.10.7.
machine still crashing, trying 27.0.0-RC.1 kernel 6.18.52
