Reallocation after adding new disks

[johnwood73]

Hi all

Please bear with me new user and my knowledge of NetApp is good but not expert 🙂

So in essence we are adding a new disk to one aggregate on our FAS2020, it contains about 20 flexvols a mix of thin and thick provisioned.

I understand that after I add the disk its best to do a reallocation -p on each flexvol, however a third party support has also mentioned that we should check to make sure that the target volume's space guarantee is enabled so that the reallocation scan does not cause an overcommitment of the volume's storage space.

Now my question is if you set the guarantee to 'volume' wouldnt this in essence make a Thin provisioned volume Thick provisioned? I am basing this on the following NetApp article

https://library.netapp.com/ecmdocs/ECMP1196986/html/GUID-C9F36C22-8A46-4D44-B74D-54B56F834D29.html

And is it necessary ? it seems a tad ambigious to me

Thanks in advance

[aborzenkov]

Physical reallocation does not inflate volume snapshots (which is likely what your support meant). It does inflate aggregate snapshots so you may need to delete them.

As for your other question - thin provisioning may be on different levels so no blanket answer is possible, but yes, if you thin provision on volume level it means making volumes thick.

[johnwood73]

Thanks for your reply Aborzenkov

Just to confirm if i do the command 'vol options volname guarantee' as requested by our 3rd party support and the felxvol is Thin provisioned it will make it Thick provisioned ?

If this is the case why would you want to do this or should you do this before running reallocation -p pn a flex vol?

I appreciate your help

[J_curl]

I don;t think you need to worry about setting space guarantee (removing thin provisioning)

-p with volume reallocate makes the process skip snapshots.

[J_curl]

I think that by adding 1 disk, you will want to do aggr reallocate.  This will level the free space among all disks, so you don't end up with a hot spot on the newly added disk. (otherwise all new writes could target only the new disk since it will be a large area of free space)

You can also reallocate the volumes, which will evenly spread blocks between all disks (and the new one). But I think in this case it's more important to do aggr reallocate.