We have a nodepool that initially consisted of 4x A200. We later purchased A300's and added a compability setting on the nodepool so that they could join the same pool.
We now want to remove the A200s from the pool, making the pool 100% A300. One way of doing this, according to a Dell PS consultant, is to simply remove the compability of the pool. According to him that would make the nodes "split out" and instantly form their own nodepool. It should have no impact to production according to him. Now I have no way of testing this, and our Dell sales rep has been unable to verify if it really works this way. So putting my hopes to this forum.
The pool looks like this:
isilon10-1# isi storagepool nodepools view --verbose 15 ID: 15 Name: a200_180tb_1.6tb-ssd_64gb Nodes: 5, 6, 7, 8, 9, 10, 11, 12, 17, 18, 19, 20 Node Type IDs: 2, 3 Protection Policy: +3d:1n1d Manual: No L3 Enabled: Yes L3 Migration Status: l3 Tier: tier2 Transfer Limit: 90% Transfer Limit State: default Usage Avail Bytes: 901.95T Avail SSD Bytes: 0.00 Avail HDD Bytes: 901.95T Balanced: No Free Bytes: 901.95T Free SSD Bytes: 0.00 Free HDD Bytes: 901.95T Total Bytes: 1.90P Total SSD Bytes: 0.00 Total HDD Bytes: 1.90P Used Bytes: 1.02P (54%) Used SSD Bytes: 0.00 (0%) Used HDD Bytes: 1.02P (54%) Virtual Hot Spare Bytes: 0.00
Happy to report the command above did exactly what we hoped.
"isi storagepool nodepools modify --remove-node-type-ids=3" instantly split the A300s into their own nodepool. No disruption noticed.
Now we've placed the new A300 nodepool into the tier and removed the A200 from it, to let the smartpools job move all data before we smartfail the A200s from the cluster.
mawi82
2 Intern
•
37 Posts
0
0
Posted June 19th, 2024 09:34
Happy to report the command above did exactly what we hoped.
"isi storagepool nodepools modify --remove-node-type-ids=3" instantly split the A300s into their own nodepool. No disruption noticed.
Now we've placed the new A300 nodepool into the tier and removed the A200 from it, to let the smartpools job move all data before we smartfail the A200s from the cluster.