hot spare across active/active heads

[esquared1]

Getting back into NetApp after a 4 year hiatus and need a refresher on hot spares.

Hardware is a FAS2020 dual head active/active setup, single shelf, 12 x 450GB FC disks, OnTAP 7.3.2

If I do 1 aggregate, raid dp, will 1 hot spare disk be sufficient to fill in for ANY disk that would fail?  I have something stuck in the back of my head about hotspares needing to be owned by a head, and I'm not quote sure how that factors in with this hardware

[reinoud7]

Only owned disks can be a spare disk. So, it is very difficult to share a spare among two active-active nodes. (what maybe will work, is un-own the "disk" and create a script that set the correct ownership when a disk is failed. Un-owned disks doesn't receive disk firmware updates also). So, this is not the general practice.

Reinoud

[ogra]

Every disk has a owner it is as simple as that. Hence, they shouldn't be shared b/w controllers.

If there is a space crunch like always, you can ask  one controller to take 9 Disk  ( 8 in Aggregate and 1 spare ) ( which will serve your data as well). and 2nd controller only 3 disks. ( RAID-DP).

In this case even if in 2nd controller 2 disk fails ( very unlikely situation  it will be in panic and cause 1st controller to take over, But your data is still secure in 1st controller using RAID-DP and a spare.

However, its again not a best practice but will yield you a win-win situation, if at all there is a space crunch situation. Alternatively, you can add a shelf.

-Bakshana

[esquared1]

That is an interesting options Bakshana.  Why would that not be considered best practice?  Is it because of lack of actual hot spare on 1 of the aggregates, or because of general best practice being to balance disks as evenlly as possible across controllers?