How should we use the remaining disks in the shelf partially configured for root aggr?

[netappmagic]

We have a shelf with 24x890GB SSD disks. Currently, 2x3 of them have been assigned to 2 root aggrs respectively to two nodes in HA. then left total of 2x9x890G = 16TB physical space.

How should we use the 16TB space? creating separate aggregate seems too small. I

If I  add 8TB into the root aggregate, and create a separated volume for data, would that cause any potential issues?

Thanks you!

[AlexDawson]

Hi there!

With older ONTAP (before version 9) you could store user data on root aggregates, but you shouldn't. With Clustered ONTAP 8.x, and later, you cannot. 

Is it a new system with no user data? And running ONTAP 8.3 or later? My question would be - why isn't the system using partitioned disks to enable using all 24 disks for user data?

[netappmagic]

It is running cluster mode 9.6p7, and the shelf was added due to swap heads to A700's for increase root volumes required by the A700's. 

In the root aggr, rg0 is still using inherited R-D partition from old heads, but rg1 is using 3 new and entire 890GB disk's.  

If we should not store user data on root aggr, why not? we will then left total of 16TB unused. 

Thanks!

[paul_stejskal]

No it is an ONTAP limitation. Use the disks to create a new aggregate on a single node. 

[AlexDawson]

Hmm. I think the headswap to the A700 could have been done differently - by adding the shelf to the existing system and partitioning the disks to add to the existing root aggr+volume before doing it, but I'm not sure exactly what led to the current situation, and I think moving away from it is probably too difficult to consider at this time.

At this point, I think the best action would be to consider adding the SSDs as a seperate RAID group into an existing SSD data aggregate (even if they don't match in size, it would normally be ok) or create a new aggregate with just them in.

You can send me a message with your system serial number and company name if you have autosupport on and would like me to take a more detailed look.

[netappmagic]

@AlexDawson , 

In our case, we added the new 24x890GB SSD for increasing the required root volume size, but didn't partition them as you suggested we should.

As it is now, I am interested in two options that you pointed out, and would like you to weigh in which one I should go for. 

1. With total of 18x890GB, creating a new aggr on a node would cause two nodes in HA out of balance, personally, I prabably would not prefer. 

2. Adding 9x890GB to each node into the existing aggr as a new rg, AND deducting two disk as parities, that would probably add 6TB more in each aggr, assuming the size mismatching is alright,  is it worth doing it?

I am going to send you 2 serial numbers for these two new heads. Due to issues here with sending out on-demand Autosupport, you might not be able to see the updates, but there is a manually trigged one there. 

Thanks a lot for your time.