man btrfs.5 FILESYSTEM WITH MULTIPLE PROFILES section

[CorrosiveTruths]

Hi, it would be nice if this section reflected the current possible profiles (add in raid1c3/raid1c4) for the order of selection.

Currently reads: "Having just one profile is desired as this also clearly defines the profile of newly allocated block groups, otherwise this depends on internal allocation policy. When there are multiple profiles present, the order of selection is RAID56, RAID10, RAID1, RAID0 as long as the device number constraints are satisfied."

Also the section above that says: "In the example above, the filesystem started as a single device and single block group profile. Then another device was added, followed by balance with convert=raid1 but for some reason hasn’t finished. Restarting the balance with convert=raid1 will continue and end up with filesystem with all block group profiles RAID1."

Should probably be convert=raid1,soft. Without the soft, it will also re-convert already raid1 blocks rather than continue converting them? Maybe re-word the note as well.

[Zygo]

it would be nice if this section reflected the current possible profiles (add in raid1c3/raid1c4) for the order of selection.

Currently, the kernel doesn't define the order of selection for raid1c3 and raid1c4 either. From btrfs_reduce_alloc_profile:

        if (allowed & BTRFS_BLOCK_GROUP_RAID6)
                allowed = BTRFS_BLOCK_GROUP_RAID6;
        else if (allowed & BTRFS_BLOCK_GROUP_RAID5)
                allowed = BTRFS_BLOCK_GROUP_RAID5;
        else if (allowed & BTRFS_BLOCK_GROUP_RAID10)
                allowed = BTRFS_BLOCK_GROUP_RAID10;
        else if (allowed & BTRFS_BLOCK_GROUP_RAID1)
                allowed = BTRFS_BLOCK_GROUP_RAID1;
        else if (allowed & BTRFS_BLOCK_GROUP_RAID0)
                allowed = BTRFS_BLOCK_GROUP_RAID0;

The first matching profile of those listed above will be chosen instead of dup and single. Presumably these would also be chosen over raid1c3 and raid1c4.

If a conversion is in progress, the conversion target profile is chosen instead (which allows converting up or down).

Should probably be convert=raid1,soft.

Yes; otherwise, it will explicitly redo everything.

[CorrosiveTruths]

Hi @Zygo.

I'm seeing (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/fs/btrfs/block-group.c):

	/* Select the highest-redundancy RAID level. */
	if (allowed & BTRFS_BLOCK_GROUP_RAID1C4)
		allowed = BTRFS_BLOCK_GROUP_RAID1C4;
	else if (allowed & BTRFS_BLOCK_GROUP_RAID6)
		allowed = BTRFS_BLOCK_GROUP_RAID6;
	else if (allowed & BTRFS_BLOCK_GROUP_RAID1C3)
		allowed = BTRFS_BLOCK_GROUP_RAID1C3;
	else if (allowed & BTRFS_BLOCK_GROUP_RAID5)
		allowed = BTRFS_BLOCK_GROUP_RAID5;
	else if (allowed & BTRFS_BLOCK_GROUP_RAID10)
		allowed = BTRFS_BLOCK_GROUP_RAID10;
	else if (allowed & BTRFS_BLOCK_GROUP_RAID1)
		allowed = BTRFS_BLOCK_GROUP_RAID1;
	else if (allowed & BTRFS_BLOCK_GROUP_DUP)
		allowed = BTRFS_BLOCK_GROUP_DUP;
	else if (allowed & BTRFS_BLOCK_GROUP_RAID0)
		allowed = BTRFS_BLOCK_GROUP_RAID0;

[Zygo]

OK, two corrections:

1. Apparently I haven't done a git fetch since 6.4, and that code is newer (and not part of a release yet).

   v6.5-rc1: 160fe8f6fdb1 btrfs: add handling for RAID1C23/DUP to btrfs_reduce_alloc_profile

2. The old behavior wasn't "use dup or single with equal priority," it was "force read-only". Yikes!

It has been backported to the -stable kernels (5.10.188, 5.15.121, 6.4.4, and 6.1.39, earlier LTS kernels don't have raid1c3 at all), so anyone who has upgraded after June should have the new profiles handled properly.