Add additional disk to volume

[Wolfeman0101]

Yes :)

Doesn't sound like you're confused ;)

So with this one, I'm running FreeNAS in a VMWare enviroment and I can't seem to figure out how to expand the volume. Any tips?

 

[paleoN]

So with this one, I'm running FreeNAS in a VMWare enviroment and I can't seem to figure out how to expand the volume. Any tips?

How about the FreeNAS details and what you want to expand to?

I wouldn't run 4 2TB drives in raidz1, but if that's the pool add another 4 2TB drives. You could also add 6 2TB drives in raidz2, move the data to it and delete the old pool.

You don't say what your space or performance requirements are.

 

[Wolfeman0101]

How about the FreeNAS details and what you want to expand to?

I wouldn't run 4 2TB drives in raidz1, but if that's the pool add another 4 2TB drives. You could also add 6 2TB drives in raidz2, move the data to it and delete the old pool.

You don't say what your space or performance requirements are.

Why wouldn't you run 4x2TB in RAIDZ1? I want to have ~10TB eventually. My case & motherboard give me the expandability to 6 drives. This is for backups of a few windows boxes, storage of movies, TV, & running Plex, Sick Beard, Couch Potato, etc.

Right now I have 2x1.5TB and 2x2TB drives in RAIDZ1. I could add 2 more 2TB drives, make it 6 and recreate the array. I really don't think I need RAIDZ2. That's 1/3 of my drives just for redundancy.

 

[survive]

Hi Wolfeman0101,

I agree with paleoN....I wouldn't run 4 2TB drives in raidz either. The risk of a second disk failing on rebuild is simply more risk than I feel is worth it.

Look at it this way....you are valuing your data at ~$125.00 because you don't want to pay for the extra drive.

-Will

 

[oxyris]

Hello!

I'm new to FreeNAS and in the process of building my own NAS around it. Before I finalize my configuration I'd like to check here if my understanding is correct.

Currently I have 2x2TB drives in my main rig and 2x2TB drives in my HTPC. I'm currently mirroring my data from the main rig to the HTPC, so I have 4TB of space available. Initially I wanted to go for a 6x2TB in RAIDZ1 mode. However, I have two problems with that

1) As far as I understood, when I build the 6x2TB configuration all my data would be deleted. So I'd require another 4TB to temporarily store my data, build the 6x2TB RAID and then transfer my 4TB data to the RAID. But after that I basically have 2x2TB disks that I don't need anymore.
2) At the moment, I don't want to purchase 2 (or even 4) additional disks.

So my idea was then to build a 3x2TB RAIDZ1 (buy 1 additional disk) and have the possibility to expand it by another 3x2TB RAID. Concerning that practice I have some questions

1) Is it correct that "3x2TB RAIDZ1 + 3x2TB RAIDZ1" is equivalent to 6x2TB RAIDZ2? So I will have 8TB available and 4TB for parity?
2) If 1) is correct, is there an advantage of building the 6x2TB RAIDZ2 from the start rather than expanding the 3x2TB RAIDZ1?
3) If 1) is correct, what happens if I expand the 3x2TB RAIDZ1 with 3x3TB RAIDZ1? Is this even possible? Will I have 8TB available store or 10TB?

Many thanks in advance.

 

[Milhouse]

1) No, it's not equivalent - with 6x2TB RAIDZ2 you can lose any two disks of the 6... with 2x 3x2TB RAIDZ1 you can stand to lose two disks, as long as it's one disk from each RAIDZ1 - if you were to lose two disks from the same RAIDZ1 vdev you'd lose the entire volume (both vdevs). In short, 2x3xTB RAIDZ1 is more redundant than 6x2TB RAIDZ1 but it's not as resilient (while arguably equally redundant) as 6x2TB RAIDZ2. In terms of storage available, you're about right.

 

[oxyris]

But why is it only possible to expand a RAIDZ1 by a RAIDZ1 of the exact same configuration if the two are independent anyways? Shouldn't it then be possible to extend a 3x2TB RAIDZ1 with e.g. a 5x2TB RAIDZ1?

 

[Milhouse]

But why is it only possible to expand a RAIDZ1 by a RAIDZ1 of the exact same configuration if the two are independent anyways? Shouldn't it then be possible to extend a 3x2TB RAIDZ1 with e.g. a 5x2TB RAIDZ1?

You can "expand" a volume that consists of a RAIDZ1 vdev by adding a second RAIDZ2 vdev if you want, the additional vdevs could even be mirrored - and if you're crazy - striped, new vdevs certainly don't have to be the same configuration as any existing vdevs. The thing to remember is that ZFS will stripe new data across all of the available vdevs in the volume, which is why if you subsequently lose a single vdev (eg. two disks failing in a RAIDZ1 vdev, or just one disk failing in a striped vdev) you will lose the whole ZFS volume, ie. all of your data will be lost.

The configuration of each vdev in a volume is entirely independent of the other vdevs, so yes you could add a second vdev of 5x2TB RAIDZ1 to an existing 3x2TB RAIDZ1 vdev (this would be close what is sometimes known as RAID 50 - aka RAID5 distributed parity with RAID0 striping).

 

[paleoN]

1) Is it correct that "3x2TB RAIDZ1 + 3x2TB RAIDZ1" is equivalent to 6x2TB RAIDZ2? So I will have 8TB available and 4TB for parity?
2) If 1) is correct, is there an advantage of building the 6x2TB RAIDZ2 from the start rather than expanding the 3x2TB RAIDZ1?
3) If 1) is correct, what happens if I expand the 3x2TB RAIDZ1 with 3x3TB RAIDZ1? Is this even possible? Will I have 8TB available store or 10TB?

1. What Milhouse said
2. Yes there is an advantage to starting with raidz2
3. Yes it's possible, though not entirely recommended for multiple reasons. You would have 4TB raw with the first vdev and 6TB raw with the second vdev.

If you start out with raidz2 you not only get the additional redunancy, but all data is striped across all the drives equally.

If you start out with 3x2TB raidz1 and mostly filled it up you would get uneven access patterns when you added a second 3x2TB raidz1 vdev. The original vdev, being mostly full & having all the current data, would service most of the reads while the second new vdev, having almost all the free space, would service most of the writes. All of which has a negative impact on performance and resilvering. It's not the end of the world by any means, but you want to have your fully built configuration as early as possible.

But why is it only possible to expand a RAIDZ1 by a RAIDZ1 of the exact same configuration if the two are independent anyways? Shouldn't it then be possible to extend a 3x2TB RAIDZ1 with e.g. a 5x2TB RAIDZ1?

As Milhouse said it's possible, but not recommended. You would again run into uneven access patterns. It's also just a plain bad idea to mix different redundancy levels, which you did not in your example.

Everyone has their own threshold as to when to move to double-parity, raidz2, vs single-parity, raidz1. I'm actually OK with running 2TB drives in a 3 disk raidz1 configuration. I am completely uncomfortable running with a 5x2TB disk raidz1 vdev. Some of the quite knowledgeable storage regulars on here draw the line with 1TB drives, discounting mirrors. You could not make me run any 3TB drive with single parity besides in a mirror configuration. The rebuild/resilver times are simply too long and the chance of additional drive failure too great.

 

[oxyris]

Many thanks for your clarifications, the issue wasn't clear to me previously from what I read in the documentation.