I hope I have all my terminology correct. I knew what I was talking about when I set this up, a year later and I’m a little rusty
I have a raidz2 vdev in a zfs pool made up of 5 16tb exos drives. I am in need of additional storage space in the same vdev. I only have the means to add a few drives, and one at a time most likely.
I know this feature has been right around the corner for a while:
It seems to be in beta now! My data is not so mission critical that I need to await full integration, I can try the beta.
Is it as simple as running zpool attach POOL raidzP-N NEW_DEVICE ?
Am I missing anything?
Any warnings I should heed?
The code is available for over a year now…if you want, you can compile it yourself. Devs always ask for people to test stuff.
But don’t expect the feature in mainline before Summer '24 and end of '24 for most distros. And even then I wouldn’t touch it for at least half a year. Because #blockcloning
Ad a new vdev with 5 drives. That’s how this works. If you would have gone with mirrors, you could just plug in 2 new drives and be done. 60% storage efficiency isn’t that much different from 50%.
Otherwise just backup everything and recreate the pool with the additional drive. That’s how everyone does their RAIDZ expansion at the moment. So there are options. And you write the data with new parity and storage efficiency, not have a permanent memory footprint and a more safe method in general.