We currently have the need to expand our aggr0 by adding new disks. Our current aggregate contains 11 disks in one raid group (rg0) utilizing RAID-DP (9 data; 2 parity). The disks are 1TB SATA disks. The current defined rg size for aggr0 is 20. The aggregate is currently a 32-bit aggregate, and the volumes within the aggregate are the source volumes in a Snapmirror relationship with another NetApp FAS2040, where the desetination volumes are in a 32-bit aggregate.
We have purchased 12 more disks, however, we were unable to get the 1TB disks and, instead, needed to purchase 2 TB SATA disks. We would like to add these disks to the existing aggregate, rather than create a new one, to avoid having to vMotion all of our VM's around, and create new volumes/datastores to present to ESXi.
This is my first time doing this type of expansion, and I want to confirm that I'm not missing any steps, or doing something that would cause irreperable harm.
The steps I've come up with the expand the aggregate are:
1) Change the rg size for rg's in aggr0 to 11 from 20. Since the current and only rg, rg0, is only 11 disks, this should have no impact on the existing rg0. It should also force any new disk additions to create their own rg1, which is what I want since adding 2TB disks to an existing rg with 1TB disks would effectively lop off 1TB of usable space per disk
2) Run the following command to confirm enough space for potential 64-bit upgrade (due to likely expanding aggregate beyond 16TB limit): aggr add aggr0 -64bit-upgrade check -g new -d [list of disks to add]
3) Once the check confirms there is enough space within the existing volumes to expand them to 64-bit, run the following command: aggr add aggr0 -64bit-upgrade normal -g new -d [list of disks to add]
4) Wait for disk add and expansion to complete
Questions I have:
1) The -64bit-upgrade switch has a couple options; growreserved and growall. I'm wondering if, in this particular case, i'm not better off using the 'growall' option rather than 'normal', so that Data OnTap will add the disks first and grow volumes as necessary.
2) Other than the difference in size, the disks are pretty much the same (both 7200rpm, same manufacturer, etc..).... if I add appropriately to their own rg1 and then to the aggr0, would there be any large performance impact of the mismatching disks?
3) Is there a major performance impact to the system during 32-bit to 64-bit expansion?
4) If there is an issue with performance, can I stop the expansion at any point and restart at the same point later? Can I stop it at all, even if I have to start all over again the next time we attempt the expansion? Or, once it starts, I have to let it complete?
5) NEW QUESTION: The aggr0 contains our root volume. Are there any concerns with expanding the aggregate containing the root volume of the filer?
Any help guidance that could be provided would be greatly appreciated.
Thank you for your reply. I had previously read both of these, and they are how I devised the steps. However, I'm asking for people who have potentially had experience with an add like this, or at least know that there will be no issues, to help answer the questions I posed. Also, the commands presented in the article are not the same as the OnTap version for my filer (e.g., I do not have a storage aggregate add-disks option; and for the storage command, it's show disk, not disk show, and -spare is not an option). Also, neither article describes the ability to pause/stop/revert an aggregate expansion.