On 6/25/23 12:38 PM, Fernando CMK wrote:
Ok, I'll try it on a new copy. BTW, are you sure it's "swidth"?
I erroneously replied to this off-list. (The field name is "width")
Fernando let me know that fixing the stripe units made things recoverable.
This is still an unexplainable situation though, to me. But glad it worked!
-Eric
It's complaining as follows:
Metadata corruption detected at 0x5627b63fec08, xfs_sb block 0x98580000/0x1000
field swidth not found
Thanks for your help.
On Sat, Jun 24, 2023 at 10:48 PM Eric Sandeen <sandeen@xxxxxxxxxxx> wrote:
On 6/24/23 1:25 PM, Fernando CMK wrote:
It seems that the only problem w/ the filesystem detected by repair is a
ridiculously large stripe width, and that's found on every superblock.
If that's the issue, is there a way to set the correct stripe width?
Also... the md array involved has 3 disks, if that's of any help.
Yes, you can rewrite each superblock (all 42) with xfs_db in -x mode.
I would suggest trying it on a fresh copy of the image file in case
something goes wrong.
for S in `seq 0 41`; do
xfs_db -x -c "sb $S" -c "write swidth 256" <image_file>
done
or something similar
I'm really baffled about how your filesystem possibly got into this
shape, though.
