Re: file system failure after hard reboot support request

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2024/6/2 11:30, William Rotach 写道:
>
>    I have reached out to a couple different locations for help with
> BTRFS, and I have reached a dead end at my google-foo, and my
> distributions discussion boards.  If there is a better place to ask for
> help with respect to a hard file system crash, please let me know.
>    I have a corrupted BTRFS file system, with a quick summary in bullet
> points:
> -    single new NVME, 6 months old, 2Tb size, 1Tb used, 200Mb of really
> needed stuff
> -    one backup exists, 6 months old, still need the new copies of above
> needed files
> -    Fedora 39
> -    chromium, high tab count, apparent culprit
> -    hard system lock (no ssh response, no console switching)
> -    repeated hard system restarts, last one crashed btrfs file system
> -    btrfs hard failures:
> -        parent transid;
> -        checksum;
> -        bytenr;
> -        search for next directory;
> -        can't read tree root;
> -        bad tree block start;
> -        212 found roots, none seem to have a "good match", as noted above

Full "btrfs check --readonly" output please, not some truncated output.

Meanwhile you can try "-o rescue=all,ro mount" option to see if it can
mount, and the dmesg for such attempt.

But according to the "can't read tree root" line, rescue mount may not help.

Thanks,
Qu
>
>    I started looking into the kernel driver source code to help my
> understanding of the physical layout, and may build a more forensic
> method of retrieving the binary/text data that I need.  I hope there is
> something that is already included in the btrfs-progs packaging,
> available through the source code, or by the developer team itself.
>