Wednesday, 10 February 2021

Expanding Aggregates with New Full Disks as Partitioned Disks

Scenario:

You have a NetApp AFF with 36 disks (24 in shelf 0, 12 in shelf 1).
The system has been partitioned with ADPv2.
Currently each node has 18 disks assigned.
Each node has 18 P3 partitions for root and 18 P1 & P2 partitions for data (not a completely standard configuration, often you'll see node 1 with the P1's and node 2 with the P2's, but it's fine).

Existing aggregate layout:
Node 1: Root Aggr: rg0 of 16 * P3 partitions
Node 1: Data Aggr: rg0 of 17 * P1 partitions & rg1 of 17 * P2 partitions
Node 2: Root Aggr: rg0 of 16 * P3 partitions
Node 2: Data Aggr: rg0 of 17 * P1 partitions & rg1 of 17 * P2 partitions

You've bought 12 new disks (of the same size as the originals) and want to assign 6 disks to Node 1, and 6 disks to Node 2. And simply expand the existing raidgroups so your aggregate layout will be:

New aggregate layout:
Node 1: Root Aggr: rg0 of 16 * P3 partitions
Node 1: Data Aggr: rg0 of 23 * P1 partitions & rg1 of 23 * P2 partitions
Node 2: Root Aggr: rg0 of 16 * P3 partitions
Node 2: Data Aggr: rg0 of 23 * P1 partitions & rg1 of 23 * P2 partitions

Image: Existing Aggregate Partitions Layout and New Aggregate Layout

Note: 48 partitioned disks is the maximum you can go to as per hwu.netapp.com, and then you have to start using full disks.

How to do?

Note 1: We use the clustershell to achieve our objective.
Note 2: I'm using a vSIM here, so disk names will look different to what's seen in reality.

1) Disable disk auto assign:

disk option show
disk option modify -node * -autoassign off
disk option show

2) Physically insert the new disks.

3) Assign 6 disks to the Node 1:

disk show
disk assign -node CLU01-01 -disklist VMw-1.1,VMw-1.2,VMw-1.3,VMw-1.4,VMw-1.5,VMw-1.6
disk show

4) Make sure your data aggr maxraidsize is sufficient since it needs to be at least 23 here:

aggr show -aggregate aggr1 -fields maxraidsize
aggr modify -aggregate aggr1 -maxraidsize 24
aggr show -aggregate aggr1 -fields maxraidsize

5) Add the 6 newly assigned and spare but unpartitioned disks to your aggregate and see they get partitioned!:

Note: Key thing in the output below A) Disks are being added to existing raidgroups rg0 & rg1 B) It actually says "The following disks will be partitioned".

::> aggr add-disks -aggregate aggr1 -disklist VMw-1.1,VMw-1.2,VMw-1.3,VMw-1.4,VMw-1.5,VMw-1.6 -simulate

Disks would be added to aggregate "aggr1" on node "CLU01-01" in the following
manner:

First Plex

RAID Group rg0, 6 disks (block checksum, raid_dp)
Usable Physical
Position Disk Type Size Size
---------- ------------------------- ---------- -------- --------
shared VMw-1.1 SSD 12.39GB 12.42GB
shared VMw-1.2 SSD 12.39GB 12.42GB
shared VMw-1.3 SSD 12.39GB 12.42GB
shared VMw-1.4 SSD 12.39GB 12.42GB
shared VMw-1.5 SSD 12.39GB 12.42GB
shared VMw-1.6 SSD 12.39GB 12.42GB

RAID Group rg1, 6 disks (block checksum, raid_dp)
Usable Physical
Position Disk Type Size Size
---------- ------------------------- ---------- -------- --------
shared VMw-1.1 SSD 12.39GB 12.42GB
shared VMw-1.2 SSD 12.39GB 12.42GB
shared VMw-1.3 SSD 12.39GB 12.42GB
shared VMw-1.4 SSD 12.39GB 12.42GB
shared VMw-1.5 SSD 12.39GB 12.42GB
shared VMw-1.6 SSD 12.39GB 12.42GB

Aggregate capacity available for volume use would be increased by 133.8GB.

The following disks would be partitioned: VMw-1.1, VMw-1.2, VMw-1.3, VMw-1.4,
VMw-1.5, VMw-1.6.

::> aggr add-disks -aggregate aggr1 -disklist VMw-1.1,VMw-1.2,VMw-1.3,VMw-1.4,VMw-1.5,VMw-1.6

6) Repeat for 3 to 5 Node 2.

7) Finally, re-enable disk auto assign:

disk option modify -node * -autoassign on

THE END!

No comments:

Post a comment