On Thu, Jul 19, 2012 at 02:29:38PM +0200, Hans J. Albertsson wrote:
> I think the problem is with disks that are 4k organised, but report
> their blocksize as 512.
> If the disk reports it's blocksize correctly as 4096, then ZFS should
> not have a problem.
> At least my 2TB Seagate Barracuda disks seemed to report their
> blocksizes as 4096, and my zpools on those machines have ashift set to
> 12, which is correct, since 2¹² = 4096
> You cannot mix 512 and 4096 byte blocksize disks in one pool, at least
> not in a mirror. All disks in a single pool should have the same blocksize.
Ashift is a vdev property. As far as I know you can have several vdevs
with different ashifts in one pool.
Besides, it is possible to mix 512 and 4096 byte blocksizes in a mirror
vdev (and likely also in other vdev configurations), if and only if you
start with the 4096 byte blocksize disk. That is, you can attach 512
byte blocksize disks to ashift=12 vdevs, but you can't attach a 4096
byte blocksize disk to an ashift=9 vdev.
And even that is not entirely true: some implementation of ZFS might
choose to just warn about it or not care at all, while others will
refuse to create or import such a pool.
%SYSTEM-F-ANARCHISM, The operating system has been overthrown
zfs-discuss mailing list