Thanks for taking care, Zygo.

On Thu, 2021-12-02 13:11 -0500, Zygo Blaxell wrote as excerpted:

Does it always happen on the same block group?  If so, that points to
something lurking in your metadata.  If a reboot fixes it for one block
group and then it gets stuck on some other block group, it points to
an issue in kernel memory state.

Good point, I'll check. I can also do a memory test but the machine runs well
otherwise.

What do you get from 'btrfs check --readonly'?

Interestingly, the machine disappeared from the network shortly after I issued
the command. :) I'll drive to the machine today or tomorrow, see what is going
on and report back.

quoted

quoted

(The particular fs has been created 2016, I am otherwise happy with
btrfs and advocating; BTW I have backups and are ready to use them)

Another question I was asking myself: can btrfs be forced to forget
about a device (as in "delete from meta data) to then just run a
regular balance?

It can, but the way you do that is "mount in degraded mode (to force
forget the device), then run btrfs device delete," and you're getting
stuck on the "btrfs device delete" step.

"btrfs device delete" is itself "resize device to zero, then run balance"
and it's the balance step you're stuck on.

Yes, but btrfs still knows about the drive. But if it's really the balance
that hangs, it probably wouldn't make much sense to just "delete the device
from the metadata" if one would have to balance afterwards anyway.

Cheers

Lukas