Changing number of disks in the root aggregate

[SkyMover]

We have a system that is configured as follows:

FAS8020, DataONTAP : 8.3.2 P2 HDD size 2.0TB SATA

Name                           Node                 Type          used%            Available               Used                   Total                  Volume Count                   Disk Count

System_A_aggr01   System_A           Std            2                      51.38TB                 1.01TB                52.39TB               2                                          42

System_A_root         System_A           Std            32                   2.83TB                 1.32TB                4.15TB                  1                                         5

System_B_aggr01   System_B          Std            2                      51.38TB                 1.01TB                52.39TB                2                                          42

System_B_root         System_B          Std           32                    2.83TB                 1.32TB                4.15TB                   1                                          5

Only 1 spare drive per system. All HDD used.

Client has decided they want to increase the number of spares by reducing the root aggregate from 5 to 3 HDD. There is one SVM defined and although there is only a small amount of data on it they would prefer to preserve that if possible.

What is the best way to achieve this? Worse case is trashing the system and rebuild it, but we'd prefer not to if we can avoid it

Is it  possible to drop disks out of an aggregate providing there is sufficent remaining capacity to hold the data on the reduced number of disks?

My experience of DataONTAP is very limited so any guidance will be very much appreciated.

Many thanks

Chris

[aborzenkov]

There is no way to remove data disks from aggregate. It is possible to convert RAID_DP into RAID4 by removing one parity disk (but RAID4 is frowned upon today). Alternatively you can build new aggregate (it can use minimal number of disks initially, even 2, so you can utilize existing spares) and relocate root volume to this aggregate, then destroy original root aggergate. Relocating root volume is possible non-disruptively in your case (but see http://community.netapp.com/t5/Data-ONTAP-Discussions/Epsilon-and-root-move-from-an-aggregate-to-anoter/m-p/138789#M30619 for caveats).

[SkyMover]

Thanks for the swift reply. I'll review the procedure. If I used the two spares to make a new aggregate I guess you can simply add a 3rd disk later after the original root aggregate has been deleted and more drives become available?

Otherwise I would need to delete the data aggregate to release sufficient disks to make a new one. That isn't a major problem as the small amout of data that exists (<1GB) can easily be copied elsewhere.

Chris

[aborzenkov]

---

@SkyMover wrote:

I guess you can simply add a 3rd disk later after the original root aggregate has been deleted and more drives become available?

---

Correct. You can both go from RAID4 to RAID_DP as well as add more data disks.

[GidonMarcus]

one more thing.

if you do willing to re-initialize your system (both nodes) and upgrade to 9.2+, you can use ADP. and gain around 4.5TB more then the raid DP>RAID4 conversation (so around 8.5 TB saving in total from what you have now)..

https://kb.netapp.com/app/answers/answer_view/a_id/1030450

Gidi