On 2016-01-05 07:25, Sylvain Joyeux wrote:
In the course of the few btrfs crashes I had on my USB backup drive
(NOT the drive from my other bug report, which is an internal SATA
drive) - in the last 6 months or so - I ended up having a 4 to 5 bad
checksums reported by scrub.
This drive is used to synchronize snapshots from my main machine, and
the corrupted files are system files that are still present on the
main machine.
I could obviously reset the filesystem on the USB drive, but since the
goal is to keep a backup history (which the main machine does not
keep), I would rather avoid that.
Would there be a way to replace the bad blocks using the good file on
the main filesystem ? Replacing it in each snapshot separately does
not look very appealing as the file is present on most of them.
Short of manually modifying the underlying block device directly, there
really isn't much you can do. However, if the file is in the same place
in every snapshot, it should be really easy to script from the command
line with a simple for loop. Assuming that the file is /bin/bash, the
following should work if run from the directory containing the snapshots
(assuming of course that the snapshots are read-only, if not you can
just remove both of the btrfs property set lines):
for snapshot in * ; do
btrfs property set ${snapshot} ro false
cp /bin/bash ${snapshot}/bin/bash
btrfs property set ${snapshot} ro true
done
You can of course replace /bin/bash in that with any file.
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
