2017-03-21 10:04 AM
We recently purchased a number of DS224C shelves that are now mixed in with older DS2246 shelves. Both have 900GB drives populated in them but they're on two different stacks.
We've created seperate aggregates to keep the disks behind the IOM6 and IOM12 apart. Seems it's not a problem to have the disks mixed, but is this recommended? What happens when a drive fails? Is there a way to set some affinity to the correct drive based on the shelf it's a member of?
2 spares from each stack of shelves for each node (8 total)
7 DS224C shelves (IOM12)
5 DS2246 shelves (IOM6)
Solved! SEE THE SOLUTION
2017-03-21 01:26 PM - edited 2017-03-22 06:22 AM
in Ontap 8.2.3 7 mode
options disk.auto_assign on
options disk.auto_assign_shelf on
storage disk option modify ‑autoassign on ‑autoassign-shelf on -node *
or in cdot 8.3 and higher
storage disk option modify ‑autoassign on ‑autoassign-policy shelf -node *
or - much better still in 8.3 and higher;
storage disk option modify ‑autoassign on ‑autoassign-policy stack -node *
2017-03-22 05:12 AM
Thank you. This is exactly what I was looking for. Seems the default setting on our system, a FAS8200, would be stack. Guess it did indeed already have the intelligence built in to assign the appropriate disk.
Thanks for the quick reply.
2017-03-22 05:54 AM
Disk ownership rules are not relevant for spare selection (if I understood your question correctly) unless all 6G shelves are owned by one node and all 12G shelves are owned by another node.
2017-03-22 06:21 AM
@aborzenkovis right. i was not thinking on the whole process.
the assignment for replacement in the raid group is controlled by:
storage raid-options show