Can i use different drives with same capacity in one pool?

Hi, can i create pool in raidz1 with different disks but the same capacity? And can i extend this pool in the future?

1 Like

Yes, the maker of the disk does not matter, you could have a WD, a Seagate, an HGST. They could even be different sizes, but ideally you want them they same to be able to use the max size of the largest disk.

As for expanding in future, you can not.

1 Like

Do I understand correctly, that is still not possible to extend pool at all?

1 Like

You can extend the pool by adding further vdevs.
You can increase vdev capacity by replacing the current drives with larger drives.

You can NOT (yet) add further drives to a raidz1 vdev (or any level of raidz) to make it wider.

You can NOT change the level of a raidz vdev (raidz1 → raidz2).

4 Likes

…and the “yet” here is important–this feature is slated to appear in SCALE by the end of this year. But there are still traps and quirks.

1 Like

I cant extend raidz pool, but i can extend another pools like stripe or mirror, right?

1 Like

You can extend a raidz pool… by adding vdevs.

With mirrors (including single drives) you can do whatever you want: Add vdevs, widen vdevs (single → 2-way → 3-way mirror), shrink vdevs, and even (free space permitting) remove vdevs.

2 Likes

Ie, say you had a 24 bay chassis.

You might’ve decided to use 8-way RaidZ2… So you buy 8 drives which might be 4TB.

Then you buy 8 more drives… and add another 8-way RaidZ2… to the same pool as another vdev.

Your data will then begin to stripe across the original vdev… and the new one… with a bias to the new one.

Eventually, you add another vdev.

And then you decide you want even more space… so you replace your original 4TB drives in the first vdev with 20TB drives… or even larger…

ONE DAY, you might be able to add/remove raidz vdevs… and you might be able to increase/decrease the width of raidz vdevs… and you might be able to increase/decrease the level of raidz vdev

But that day is not here yet.

1 Like

I don’t think two of the three things you mentioned are under serious consideration to ever happen. Increasing (not decreasing) the width of RAIDZn vdevs should be in SCALE before the end of this year. I don’t think anyone’s talking seriously about RAIDZ vdev removal, changing the redundancy level, or decreasing the width of RAIDZ vdevs.

2 Likes

These are things taken for granted in other systems

And I didn’t even talk about dissimilar drive topologies, but that’s actually a trick when you can change increase/decrease raid widths etc

1 Like

Which is why it’s best to put forward that these are not available at all with ZFS.

Raidz vdev expansion is being expected “soon” for years. Although iX is adamant that it is coming this year, I’ll believe it when I’ll see it.

As for the other features… To make it practical, vdev expansion is implemented without reflowing data: So there’s not much hope for vdev shrinking or changing raidz level, which would require reflowing data.
And as for raidz vdev removal and Block Pointer Rewrite…

1 Like

I’ll likewise believe it when I see it, but the code has landed in OpenZFS as I understand, so that’s a pretty strong indication that it really is Coming Soon™. But yeah, it’s been a long time coming. This thread is almost seven years old:

2 Likes

Ganz groĂźes Kino: The Seven Year Itch, iXsystems remake?

Granted, hopefully the discussion helps with that :slight_smile: