| Message ID | 20211116140206.291252-1-nborisov@suse.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Fix calculation of chunk size for RAID1/DUP profiles | expand |
On Tue, Nov 16, 2021 at 04:02:06PM +0200, Nikolay Borisov wrote: > Current formula calculates the stripe size, however that's not what we want > in the case of RAID1/DUP profiles. In those cases since chunkc are mirrored > across devices we want the full size of the chunk. Without this patch the > 'btrfs fi usage' output from an fs which is using RAID1 is: > > <snip> > > Data,RAID1: Size:2.00GiB, Used:1.00GiB (50.03%) > /dev/vdc 1.00GiB > /dev/vdf 1.00GiB > > Metadata,RAID1: Size:256.00MiB, Used:1.34MiB (0.52%) > /dev/vdc 128.00MiB > /dev/vdf 128.00MiB > > System,RAID1: Size:8.00MiB, Used:16.00KiB (0.20%) > /dev/vdc 4.00MiB > /dev/vdf 4.00MiB > > Unallocated: > /dev/vdc 8.87GiB > /dev/vdf 8.87GiB > > > So a 2 gigabyte RAID1 chunk actually will take up 4 gigabytes on the actual disks > 2 each. In this case this is being miscalculated as taking up 1gb on each device. > > This also leads to erroneously calculated unallocated space. The correct output > in this case is: > > <snip> > > Data,RAID1: Size:2.00GiB, Used:1.00GiB (50.03%) > /dev/vdc 2.00GiB > /dev/vdf 2.00GiB > > Metadata,RAID1: Size:256.00MiB, Used:1.34MiB (0.52%) > /dev/vdc 256.00MiB > /dev/vdf 256.00MiB > > System,RAID1: Size:8.00MiB, Used:16.00KiB (0.20%) > /dev/vdc 8.00MiB > /dev/vdf 8.00MiB > > Unallocated: > /dev/vdc 7.74GiB > /dev/vdf 7.74GiB > > > Fix it by only utilising the chunk formula for profiles which are not RAID1/DUP. > > Signed-off-by: Nikolay Borisov <nborisov@suse.com> Added to devel, thanks.
diff --git a/cmds/filesystem-usage.c b/cmds/filesystem-usage.c index 6195f633da44..5f2289a9b40d 100644 --- a/cmds/filesystem-usage.c +++ b/cmds/filesystem-usage.c @@ -805,11 +805,17 @@ int load_chunk_and_device_info(int fd, struct chunk_info **chunkinfo, */ static u64 calc_chunk_size(struct chunk_info *ci) { - u32 div; + u32 div = 1; - /* No parity + sub_stripes, so order of "-" and "/" does not matter */ - div = (ci->num_stripes - btrfs_bg_type_to_nparity(ci->type)) / - btrfs_bg_type_to_sub_stripes(ci->type); + /* + * The formula doesn't work for RAID1/DUP types, we should just return the + * chunk size + */ + if (!(ci->type & (BTRFS_BLOCK_GROUP_RAID1_MASK|BTRFS_BLOCK_GROUP_DUP))) { + /* No parity + sub_stripes, so order of "-" and "/" does not matter */ + div = (ci->num_stripes - btrfs_bg_type_to_nparity(ci->type)) / + btrfs_bg_type_to_sub_stripes(ci->type); + } return ci->size / div; }
Current formula calculates the stripe size, however that's not what we want in the case of RAID1/DUP profiles. In those cases since chunkc are mirrored across devices we want the full size of the chunk. Without this patch the 'btrfs fi usage' output from an fs which is using RAID1 is: <snip> Data,RAID1: Size:2.00GiB, Used:1.00GiB (50.03%) /dev/vdc 1.00GiB /dev/vdf 1.00GiB Metadata,RAID1: Size:256.00MiB, Used:1.34MiB (0.52%) /dev/vdc 128.00MiB /dev/vdf 128.00MiB System,RAID1: Size:8.00MiB, Used:16.00KiB (0.20%) /dev/vdc 4.00MiB /dev/vdf 4.00MiB Unallocated: /dev/vdc 8.87GiB /dev/vdf 8.87GiB So a 2 gigabyte RAID1 chunk actually will take up 4 gigabytes on the actual disks 2 each. In this case this is being miscalculated as taking up 1gb on each device. This also leads to erroneously calculated unallocated space. The correct output in this case is: <snip> Data,RAID1: Size:2.00GiB, Used:1.00GiB (50.03%) /dev/vdc 2.00GiB /dev/vdf 2.00GiB Metadata,RAID1: Size:256.00MiB, Used:1.34MiB (0.52%) /dev/vdc 256.00MiB /dev/vdf 256.00MiB System,RAID1: Size:8.00MiB, Used:16.00KiB (0.20%) /dev/vdc 8.00MiB /dev/vdf 8.00MiB Unallocated: /dev/vdc 7.74GiB /dev/vdf 7.74GiB Fix it by only utilising the chunk formula for profiles which are not RAID1/DUP. Signed-off-by: Nikolay Borisov <nborisov@suse.com> --- cmds/filesystem-usage.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-) -- 2.17.1