On 2020/2/27 下午1:59, 4e868df3 wrote:
> I updated kernels recently and now am getting a corrupt leaf error.
> The drives decrypt and mount, and I can touch a file briefly until the
> mount switches over to read-only mode. Extended SMART tests show all 6
> of my drives have a healthy status. I have a backup of the data. The
> array is configured as RAID10. As the BTRFS filesystem remains
> accessible / read-only, I am able to take an additional backup. What
> is the best way to recover from this error?
> 

> [  130.395696] BTRFS critical (device dm-0): corrupt leaf: root=7
block=2533706842112 slot=5, csum end range (68761223168) goes beyond the
start range (68761178112) of the next csum item
> [  130.395829] BTRFS error (device dm-0): block=2533706842112 read
time tree block corruption detected

This is not something caused by a powerloss, but more likely an older
kernel, or memory corruption.

Please provide the following command dump:
# btrfs ins dump-tree -b 2533706842112 /dev/dm-0

Furthermore, btrfs check output would be appreciated to ensure that's
the only corruption.

But from the generation mismatch, it looks like there are transid
mismatch error, which could be a bigger problem than your corrupted csum
tree.

Thanks,
Qu