Unoid

What version of rocm does radeon top plugin provide? Anyone have success? I ahve a mi50 on the way and would like unraid to have the necessary rocm and or amdgpu vulkan.

The lack of GPU support that isn't ngreedia is getting tiring.

Thanks for always helping JorgeB, It's scary having to re-wipe the GPT of each disk in a zpool from a normal reboot.

I ran sgdisk on 3n1, and rebooted, the pool now mounts. No clue what happened...

I'm going to just nuke the zpool. I was able to get the zpool status in a available state, disks [0-2] were unavail while 3n1 was available (parity disk), I ran sgdisk -o -a 8 -n 1:1M:0 /dev/nvme0n1 also on 1 and 2. Then zpool import would work. However mounting in unraid would still fail. I think the ugprade somehow broke the GPT info on some disks which is very concerning if true. At this point I'm going to re-create the zpool and restore data. jorge, if you see what happened from the diagnostic, that's great to know. otherwise disregard. btw I hate zfs on nvme in unraid.

I don't see anything specific in the syslogs that would destroy the zpool 'speedteam'. All I can see if working, then today upon upgrade the zpool doesn't exist system wide anymore.

I had a working zpool with 4 nvme's in z1. If I remember correctly, I did create the zpool via CLI in order to set a custom ashift. It imported into the GUI just fine and has been working. Did the upgrade to 6.12.14. Now the GUI still shows the pool configuration but states that the disks are unmountable. via bash: zpool list and zfs list show nothing. it's as if my pool never existed. Any advice on recovery?

that was it! thank you again for the zfs help!

I may be doing steps wrong, here's what I'm doing: sudo zpool create -o ashift=13 -O sync=disabled -O recordsize=1M mypool raidz /dev/nvme0n1 /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1 Then in GUI I create 4 disk pool with fs=auto. upon array startup, it keeps saying: unmountable: unsupported or no filesystem. Should I be formatting the zpool or anything else manually before I drop into gui? I also tried the following after manual pool creation before setting in gui: sudo zfs set aclmode=passthrough speedteam sudo zfs set aclinherit=passthrough speedteam sudo chmod -R 777 /speedteam sudo zfs set mountpoint=/speedteam speedteam

Major slowdowns over time. a fresh z1 of 4x4tb nvme, fio speeds a good, except 4k sizes. Load up the disks to 50% and fio tests show substancial speed regressions down to 200MB/s moving a 200GB.img file. scrubs get progressively slowers every month as well, started out at 45minutes with 50% used, down to 8 hours 9 months later. George helped me try to tune some things with recordsize etc but some options can't be changed after the fact like ashift or sync mode. in summary, I wish I could just add a 2nd array of 4x4tb nvme with 1 parity on btrfs.

I run a zpool of 4x4TB nvme. I'm not happy with sync=auto and ashift=12, I can use the cli to manually create a pool with the options I want. HOW do I 'import' it into the GUI? If I create a pool there, it'll reformat with defaults of ashift=12 etc. Is it possible?

Multiple arrays would be godsend. I want one array of all NVME storage. ZFS zpool has been a bad experience for nvme

Rclone is what you're thinking of. essentially rsync for cloud/other dc backup

zpool scrub of 4x4TB nvme went from 2 hours (50% used) to 6 hours (62% used) over the last 8 months. transfers reflect that as well. I so want to have multiple pools One for HDD's and one for NVME's

I can't wait for multiple data arrays. I have 3x12TB HDD's as slow storage array. For a while I've had 4x4TB nvme's in ZFS z1, but their performance slow down over time quite bad. everything I read of nvme + zfs isn't good. Normal 3 disk nvme btrfs with 1 parity would be great. nvme speeds for single disk is more than plenty on my home network.