@josef put up a few patches to add prominent warnings to btrfs-progs: https://lore.kernel.org/linux-btrfs/bf9594ea55ce40af80548888070427ad97daf78a.1598374255.git.josef@toxicpanda.com/ https://lore.kernel.org/linux-btrfs/164e102b2a6179f9af35ded962c11d780ccf8400.1598375602.git.josef@toxicpanda.com/
We should also mention this prominently in the docs.
Metadata Update from @dcavalca: - Issue set to the milestone: Fedora 33
This is a good explanation and how to use Btrfs raid56 safely. https://lore.kernel.org/linux-btrfs/20200627032414.GX10769@hungrycats.org/
I also saw patches for btrfs-progs to add a warning when mkfs or converting to raid56.
If it's necessary to recommend against using btrfs raid56 then it's necessary to recommend against anything using md raid56 without enabling integrity, which LVM now has an option for. There are enough caveats of software raid that I think it's better to state the facts rather than make recommendations. Zygo's write up states the facts and work arounds or things to be aware of.
Also related to this multiple disk problem is timeout mismatches between the drive firmware (SCT ERC) and the kernel's SCSI command timer. This guide should be updated to include a udev example, but is otherwise applicable to mdadm, lvm, and btrfs raid. I have no idea if ZFS on Linux is also affected, I'm not sure why it wouldn't be but I don't know if they implement their own block layer. https://raid.wiki.kernel.org/index.php/Timeout_Mismatch
Yeah I'd mostly filed this to make sure we didn't forget to pull the warning patches into Fedora. I like the idea of signal boosting Zygo's writeup.
Metadata Update from @ngompa: - Issue set to the milestone: Future Release (was: Fedora 33)
Login to comment on this ticket.