I have a cache pool with 2 ssds on brtfs and today i woke up with all my docker containers and vms stopped. After having a look, i saw a notification with unrecoverable errors on my cache disks, so i moved all appdata and other data away from the cache and onto the array.

 

As the mover seemed not to move everything, i did a scrub on the pool, which fixed most of the errors (but left around 64 unrecoverable errors). I managed to move everything back to the array, all but a 70GB vm img of my blueiris vm.

 

This last file (and most important) i am unable to move to the array. The mover stops with io errors. I tried to cp it manually and after copying around 35GB i got a "cp: error reading 'vdisk1.img': Input/output error"

 

I tried to rsync, but rsync after copying all the data, gave that the verification of the file failed and retries.

 

The thing is that if i start the vm, it works fine. So the corruption must be minimal, yet i am unable to copy the img to the array in order to replace the cache drives.

 

Is there a way to copy the image (even with the small corruption, since it seems to work fine) or repair it somehow ?

 

It's normal to get an i/o error is btrfs detects the file is corrupt, so the user knows there's problem, you can recover the file with btrfs restore, it will not check data integrity.

Thank you for replying

 

I tried what you suggested, umounted the cache, and used btrfs restore, and i still got an error and got half of the img file, as i did with the cp command (although i did use the -i option to ignore errors):

 

root:~# btrfs restore -iv /dev/mapper/sdb1 /mnt/disk1/restore_cache/
Restoring /mnt/disk1/restore_cache/domains
Restoring /mnt/disk1/restore_cache/domains/Windows 10
Restoring /mnt/disk1/restore_cache/domains/Windows 10/vdisk1.img
Short read for 102430212096, read 8192, read_len 65536
trying another mirror
Short read for 102430212096, read 8192, read_len 65536
ERROR: exhausted mirrors trying to read (3 > 2)
ERROR: copying data for /mnt/disk1/restore_cache/domains/Windows 10/vdisk1.img failed
Done searching /domains/Windows 10
ERROR: searching directory /mnt/disk1/restore_cache/domains/Windows 10/vdisk1.img failed: -1
Done searching /domains
ERROR: searching directory /mnt/disk1/restore_cache/domains/Windows 10/vdisk1.img failed: -1

 

hmm, maybe it's not a checksum error, try to copy the file with cp and post the diags after you get an error.