ONTAP Discussions

Remove a Disk Shelf from an Aggregate

jku
3,696 Views

I have a FAS3040 running Data Ontap 7.3.2P4. I have just connected another Disk Shelf and just added it to the Aggregate. I see the disks being zero'd out right now. Is there some way to cancel this process? I don't want to add this shelf to this Aggregate. I want to use this shelf somewhere else. Is it too late now? Once the Shelf is added to the AGGR and the AGGR grows in size, is there some way to shrink the AGGR or remove the Entire disk shelf?

1 ACCEPTED SOLUTION

spence
3,696 Views

The only way to stop a disk zero is to reboot.  There is no way to remove disks from an aggregate once they are added and there is no way to shrink an aggregate.

Now given your scenario it might still be possible to reboot and have the aggr add command cancelled out as a result of the reboot stoping the disk zero, but that is something I have NEVER tried and would be VERY cautious about. You could potentially loose data and the existing aggr.

This is a reason why aggr add commands should be very carefully executed.

If you choose to reboot, I would only do so if you have aggr snapshots enabled and present:

snap list -A <aggr name>

This would definately be an unsupport method and if something goes wrong wafliron may be required AND data loss potential is still present.

View solution in original post

1 REPLY 1

spence
3,697 Views

The only way to stop a disk zero is to reboot.  There is no way to remove disks from an aggregate once they are added and there is no way to shrink an aggregate.

Now given your scenario it might still be possible to reboot and have the aggr add command cancelled out as a result of the reboot stoping the disk zero, but that is something I have NEVER tried and would be VERY cautious about. You could potentially loose data and the existing aggr.

This is a reason why aggr add commands should be very carefully executed.

If you choose to reboot, I would only do so if you have aggr snapshots enabled and present:

snap list -A <aggr name>

This would definately be an unsupport method and if something goes wrong wafliron may be required AND data loss potential is still present.

Public