Copying a 1TB disk image on a btrfs partition with cp takes less than a second on my Ubuntu 14.10 amd64 with Linux 3.16.0. Why does it take at least 5 minutes and requires > 1 GB to be read on a ZFS pool with deduplication turned on and compression turned off?

I just want to improve my understanding of ZFS and figure out whether ZFS is an appropriate alternative as deduplicating filesystem for me (by understanding e.g. whether copy action described for btrfs are possible on ZFS at all (on linux 3.16)).

EDIT: After Stéphane Chazelas' comment I run cp --reflink=always file file.bk which fails with

$ LANG=C cp --reflink=always file file.bk
cp: failed to clone 'image_lvm.img.bk2' from 'image_lvm.img': Inappropriate ioctl for device
  • That's probably down to your cp implementation that does a reflink copy, which is not supported by ZFS yet. – Stéphane Chazelas Nov 1 '14 at 10:03
  • 1
    That is very much like the info I was looking for. Do you know where plans are mentioned or documented to support that in the future (assuming you added "yet" based on such rather than your own expectation and wishes) and how I could follow the development? Thank you! – Karl Richter Nov 1 '14 at 10:26

The feature is in development which can be followed at https://github.com/zfsonlinux/zfs/issues/405. The last action is to remove it from 0.8.0 milestone which is quite ahead of our time (current version is 0.6.3 AFAIK), so I'll just sit and wait...

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.