0
I just cloned a btrfs file system from one disk to another using partclone.btrfs --dev-to-dev
. Afterwards, I mounted the cloned file system and started a scrub to see if everything is in order. After a while I checked the status and was shocked to see hundreds of thousands of correctable csum errors. Should I be worried? Since they are correctable errors, I guess it's OK, but still. Is this an expected result of the cloning? I'm using Oracle Linux 6, i.e. kernel 3.8.13.
Thanks for any insight in advance.
Have you performed a scrub on the source disk too? I'd definitely be concerned. You should probably build the latest kernel and btrfs-progs and try again... – Hygrinet – 2015-06-29T08:18:31.873
@HoJi No, I haven't checked the source disk. Good idea. I'll do that and report back. – user2323470 – 2015-06-29T08:34:41.417
@HoJi I scrubbed the source disk. It checks out OK. 0 errors. I also compared some of the cloned files with sha256sum, they are all identical. – user2323470 – 2015-06-30T19:03:33.203
1Hmm... It could well be a bug in btrfs in that case. The best course of action is to compile yourself the latest kernel version and btrfs-progs - if the problem still exists then post the details to the btrfs mailing list. The 3.8 kernel is pretty old now from a btrfs point of view, as I'm assuming the userspace tools are the same age too. – Hygrinet – 2015-07-01T11:39:36.977