Hello all,

Small update on this issue, it happened again to the same disk:

Nov 22 13:32:15 TV1 emhttpd: shcmd (126): mount -t btrfs -o
noatime,space_cache=v2 /dev/md20 /mnt/disk20
Nov 22 13:32:15 TV1 kernel: BTRFS info (device md20): enabling free space tree
Nov 22 13:32:15 TV1 kernel: BTRFS info (device md20): using free space tree
Nov 22 13:32:15 TV1 kernel: BTRFS info (device md20): has skinny extents
Nov 22 13:32:16 TV1 kernel: BTRFS error (device md20): bad tree block
start, want 1589201485824 have 620757024
Nov 22 13:32:16 TV1 kernel: BTRFS error (device md20): bad tree block
start, want 1589201485824 have 620757024
Nov 22 13:32:16 TV1 kernel: BTRFS error (device md20): failed to read
block groups: -5

This is a new filesystem, previous one was unrecoverable, again it
happened on boot after a clean shutdown, I have over a 100 similar
btrfs filesystems, more than 10 using this same disk model with no
other issues for years, so the same thing happening to the same disk
in a space of a few months suggests to me it's not just a firmware
issue, something else must be going one, maybe something in the disk
is going bad, controller is an LSI 9207 HBA with 17 more disks
connected, for now I'm going to restore the data to a different disk
and see if it doesn't happen again, might also use this disk in a
small zfs pool I have to see if it also gives issues there.

Regards,
Jorge Bastos

On Wed, Jul 21, 2021 at 7:14 PM Jorge Bastos [off-list ref] wrote:

On Wed, Jul 21, 2021 at 6:47 PM David Sterba [off-list ref] wrote:

quoted

For the record summing up the discussion from IRC with Zygo, this
particular firmware 80.00A80 on WD Green is known to have problematic
firmware and would explain the observed errors.

Recommendation is not to use WD Green or periodically disable the write
cache by 'hdparm -W0'.

Thank you for the reply, yes, from now on I intend to disable write
cache on those disks, since I still have a lot of them in use.

Jorge