[zfs-discuss] ZFS on top of iSCSI on top of ZFS

Schweiss, Chip chip at innovates.com
Fri Jan 8 17:50:46 EST 2016


4K blocks cause a lot of extra space for metadata in ZFS especially under
raidz.   Go all the way to 128K and metadata is going to kill your space
consumption.

ZFS block size has nothing to do with disk block size.   Set your disk
block size to the smallest your disk physically supports.   On SSDs don't
worry if they have larger internal blocks, set it to what the SSD
advertises to the SAS/SATA and let the internal controller take care of
it.

If you are using compression the ZFS block size is variable ranging from
the disk block size to the size set.   If you're using compression and your
ZFS is new enough use 1M for best results.

-Chip

On Fri, Jan 8, 2016 at 6:01 AM, kaazoo via zfs-discuss <
zfs-discuss at list.zfsonlinux.org> wrote:

> Am 08.01.16 um 12:45 schrieb Gordan Bobic via zfs-discuss:
> >
> > On 8 Jan 2016 11:59 am, kaazoo via zfs-discuss <
> zfs-discuss at list.zfsonlinux.org> wrote:
> >>
> >> Am 07.01.16 um 16:25 schrieb Markus Koeberl:
> >>> On Thursday 07 January 2016 09:51:41 kaazoo via zfs-discuss wrote:
> >
> >>>> The iSCSI volumes have 128 KB blocksize, so ashift=17 is used.
> >>>> ZFS features like compression and snapshots are used here.
> >>>
> >>> If I understand it right how compression is working you may not
> benefit with this settings.
> >>> Because of ashift=17 you cannot use smaller blocks than 128 KB and
> because of the max blocksize of 128 KB you never get bigger blocks.
> >>> So the only way to save space is if you get 128 KB of zeros I guess.
> >
> > I'm pretty sure there is no such think as ashift=17 in ZFS. The maximum
> value is 13 (8KB).
>
>
> Hello Gordan,
>
> I was wondering as well.
>
> The 'Linux IO target' iSCSI device based on a zvol with blocksize=128K
> will report a physical blockisze of 128K. When creating a zpool without
> specifiing a value for ashift 'ZFS on Linux' seems to do autodetection
> and try to match the physical blocksize which is reported by the device.
> 'ashift=17' was shown via 'zdb' command.
>
> For some devices which only report 512 Byte but use 4 KB internally,
> other hardcoded ashift values are used.
>
>
> Regards,
> kaazoo
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at list.zfsonlinux.org
> http://list.zfsonlinux.org/cgi-bin/mailman/listinfo/zfs-discuss
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.zfsonlinux.org/pipermail/zfs-discuss/attachments/20160108/3b5002a3/attachment-0001.html>


More information about the zfs-discuss mailing list