btrfs-check for filesystems with many snapshots (with many files) gets slow

btrfs-check for filesystems with many snapshots (with many files) gets slow

[Christoph Anton Mitterer]

Hey.

That's not really a priority issue, I just figured that no one might
have thought about it before:

When having a filesystem with many files (like a million or so) and on
the same fs, snapshots are continuously made of the "main" subvol with
those files... then eventually btrfs-check get's super slow, even
though there weren't really that many changes between the snapshots
(mostly added files).

"Super slow" in the sense of 6 hours and more.


I just wondered whether there's any room for speed up (without loosing
any safety in checking) in btrfs-check to make it much faster?

Very naively thought something like: it doesn't have to check the stuff
it already checked on the other snapshot.
But of course I'm aware that with refs between the snapshots and
extents, things are much more complex in reality.


Cheers,
Chris.