ZFS vs OSX Time Machine

[Malcolm Waltz]

ZFS volumes (zvol s) can definitely be resized using the volsize property:
# zfs get volsize mypool/myvol
NAME        PROPERTY  VALUE    SOURCE
mypool/myvol  volsize   2G       -
# zfs set volsize=4g mypool/myvol

Mac OS 10.5 and later allows you to resize Journaled HFS+ volumes (using diskutil or Disk Utility.app).  Doing a quick google search, I see plenty of references to decreasing the size of a TimeMachine volume, so it's probably possible to increase it as well.  I'm sure you can find more with a little googleing.

"man diskutil" (look for resizeVolume) indicates that you can increase and decrease the size and doesn't mention anything special about Time Machine.


On Apr 28, 2011, at 5:02 PM, Daniel O'Connor wrote:

> 
> On 29/04/2011, at 2:16, Malcolm Waltz wrote:
>> I doubt the issues you are encountering have much to do with ZFS.
>> 
>> It sounds like you are using TimeMachine over NFS.  Obviously, Apple does not support that configuration:
>> http://www.google.com/search?q=time+machine+nfs+site:apple.com
>> 
>> In my opinion, TimeMachine should only be used with block storage.  If you use any kind of file-sharing protocol (AFP, SMB/CIFS or NFS), TimeMachine is implemented using a sparse disk image broken into hundreds or thousands of separate files.  This is a hack at best.
>> 
>> Time machine works very well with locally attached storage, but if you need to use network storage, you might want to try iSCSI:
>> http://thegreyblog.blogspot.com/2010/02/using-zfs-with-apple-time-machine.html
>> http://people.freebsd.org/~rse/iscsi/iscsi.txt
> 
> Hmm, I _am_ using AFPD, not NFS for this.. I will see about using an ISCSI disk image instead (although that would make it impossible to resize once it's created right?)
> 
> I see that the sparse disk image does use ~80000 files in a single directory which does take.. a while.. to stat..
> 
> --
> Daniel O'Connor software and network engineer
> for Genesis Software - http://www.gsoft.com.au
> "The nice thing about standards is that there
> are so many of them to choose from."
>  -- Andrew Tanenbaum
> GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C
> 
> 
> 
> 
> 
>