On Thu, Aug 5, 2010 at 2:23 PM, Simon Vallet svallet@genoscope.cns.fr wrote:

Hi list,

I currently have to address a quite interesting problem : long story short, we need to remove disks from the root (and only) aggregate of a 3020 cluster.

Since this is not possible [1], and as we don't have other aggregates on this filer, do you think it would be feasible to :

- copy the root volume on another filer ;        - netboot a head ;        - destroy the root aggregate ;        - create another one, with the correct number of disk drives ;        - copy the root volume back ?

Yes, I did contact NetApp about the subject.

I guess you'll get there eventually, although I think you'll lose your igroups and LUN mappings in the process.

Why not get 2 spare disks (eg by converting from RAID-DP to RAID-4 on an existing aggregate) somehow, create a 2-disk RAID-4 aggregate, migrate the root volume onto that and save you the trouble of having to go through a netboot ?

Best regards, Filip