Recovering from BTRFS corruption

[timothyd]

Hi guys BTRFS scrub is reporting just a couple file / disk corruptions.

This was due to faulty a faulty memory stick so i am now trying to piece things back together.

I am getting errors such as:

Nov 17 11:19:23 Endeavour kernel: BTRFS error (device nvme0n1p7): bdev /dev/nvme0n1p7 errs: wr 0, rd 0, flush 0, corrupt 2255, gen 0
Nov 17 11:19:23 Endeavour kernel: BTRFS error (device nvme0n1p7): unable to fixup (regular) error at logical 265919631360 on dev /dev/nvme0n1p7
Nov 17 11:19:24 Endeavour kernel: BTRFS warning (device nvme0n1p7): checksum error at logical 391730319360 on dev /dev/nvme0n1p7, physical 195067793408, root 256, inode 677652, offset 28672, length 4096, links 1 (path: usr/lib/libglapi.so.0.0.0)

The problem is even if i re-install the corresponding package, and therefore replace the file i still get the error.

Two other file errors detected also seem to be for files which are no longer on the system (core dumps & removed software).

What is the best way to clean this up? Its an otherwise healthy fresh install so i dont want to bork it

Im guessing this is a metadata issue?

[timothyd]

booting off liveCD and running

“btrfs check --repair” fixed some stuff but not these file links

“btrfs rescue chunk-recover” didnt fix it.

[timothyd]

Givern that two of these files dont exist any more and one is replaceable via package rebuild (and already has been) ive made the assumption that this is a corrupt checksum tree.

i ran check with the DANGERIOUS option --init-csum-tree

That seem to have worked. No more errors and hopefully it didnt ranksack my otherwise good file-system.

I cba with the effort of the alternative of shrinking partition, adding, moving and then moving back the partitions and so rolled the dice.