We currently have a 4 node Netapp Cluster with a FAS8040 (being replaced with a FAS2720) & AFF300 with 1 DS2246 IOM6 Shelf with 24x1.6TB SSD drives with ADP enabled. The cluster is running 9.3P19. Currently there are two root aggregates and two data aggregates. All the SSD's are shared between both nodes in the HA pair and the existing disk shelf is ID 00.
The new shelf is a DS224C with IOM12. My plan was to add the shelf on a separate SAS loop of 10 because its IOM 12 and different sizes. After I add the shelf's will Ontap be smart enough to create a shared aggregate with 11 of the 3.8TB SSDs, one spare, and split the storage between the nodes? Has anyone done this lately and have the best process? I will open a case too.
My concerns are as follows:
1. I see the 1.6TB SSD drives being retired in the future and our root aggregates currently live on there.
2. What is the best way to setup the new aggregates and maybe move the root aggregates to the new drives. I have done this before with mechanical drives and it required a controller restart.
Depending which 1.6TB SSDs you have, EOS dates have already been released:
X365 & X366 - January 31, 2023
X439 & X576 - October 31, 2021
I'm willing to wager a fair bit of money you have the X36x drives though so you have some time before you need to retire the 1.6s, unless you're retiring them for other, internal reasons. As an aside, the A300 goes EOS on November 30, 2026.
The root volume relocation is pretty easy actually, and the procedure can be found here. I would be more concerned about ensuring that you're still able to take advantage of ADP on the new half shelf. ONTAP has been getting "smarter" with respect to aggregate creation, but even if it does get created in a way you are not happy with, tearing it down and recreating them isn't so bad as zeroing SSDs takes seconds, not minutes/hours...I'm looking at you 8TB+ drives; also, depending which version of ONTAP you deployed the A300 with, it may also have the Rapid Zeroing feature which came with ONTAP 9.4.
I don't believe the new drives will partitions automatically since you won't be adding them to the same RAID group or aggregate so if you went with the automatic approach, this would mean dedicating 3 SSD per node for root and 1 spare per node, losing 8 of those precious SSDs. If this was my system, I'd be looking into partitioning the drives manually to the same sizes as an A300 with 12x3.8GB SSDs would come from the factory as, and only then migrating root. If any of this makes you uncomfortable, it's probably time to talk to your local SE or the support centre on how to proceed.
Hi that is exactly what I am thinking I should do. This was originally a AFF8040 that was replaced with a AFFA300 with the free head upgrade at year 3. The drives the AFF8040 came with were the X439_S16351T6ATD. They are end of support on 10/31/2021 however our support on the array is good until 6/30/2022. That means Netapp will support the old drives.
So if I do not root partition the new 12x3.84tb disks this time around, I would have to move data off the array temporarily in the future to move the root volumes. So the question is do I take care of the issue now instead of waiting for it to byte me later. Another option is when we buy another 12x3.84tb disks, I assume I could root partition those new disks to hold the root volume. Is there any documentation for how to setup root data partitioning with newly added disks? I setup the AFFA300 all by myself and it was option 9a on bootup if I remember.
Sure looks right to me, and while the intent of the KB is to get you from whole drives to ADP I don't see why it wouldn't work going from ADP to ADP.
Your first response has one grave error in it however:
The drives the AFF8040 came with were the X439_S16351T6ATD. They are end of support on 10/31/2021 however our support on the array is good until 6/30/2022. That means Netapp will support the old drives.
This is not so, the support for your X439s will end on October 31, 2021 regardless of the support contract that exists on the controller. I highly recommend you get clarification here before it becomes a problem. I'm going through this with a bunch of my clients running 3TB drives right now.
Actually they will be covered. The array was sold with those drives originally with 6 years of support. We are still on those 6 years of support. I understand your argument but I have had this discussion with 450gb sas 15k drives which were end of support but they replaced them because the array has warranty. It depends, is the right answer. However from what I have seen is if the array was originally purchased with those drives, they are covered for the original duration of the warranty. Now when 2022 comes along. I will not be able to renew support on those drives but NetApp will not say oh well yeah we screwed up and sold you 6 years of support but are only going to cover the drives for 5.
The only way your assertion is true is if you have explicitly requested and obtained approval for a support extension product variation request (PVR). If you do not have a support extension PVR in place, do not expect to receive replacements for the drive when it passes its HW end-of-support (EOS) date. I would check with your NetApp sales representative/partner.