how to get max space from 3x2TB + 2x1TB ZFS drives w/Parity?

[TooTHPiCK]

Please forgive my noobishness with ZFS, raid and FreeNAS... I am trying to get max storage space out of 3x 2TB drives + 2x 1TB drives with single disk parity like Raid5. It's for backups and long term storage so write/read performance is not very important here. Space is #1, but I do need it to handle a SINGLE drive failure. Is it possible to configure them in a ZFS RAIDZ with the two 1TB drives striped in Raid0, providing a fourth 2TB drive. With that, could I then stack a Raid5 on top of the 4x 2TB drives? If possible, is that the best way? Would I have to configure it using the command line? The best solution I can get working in the GUI gives me just 4.4TB total with all disks in RAIDZ... I was hoping to get 6TB + 2TB for parity. I was getting 4TB before adding the 2 extra 1TB drives. Surely there is a way to use the 1TB drives without making a separate pool?

I just need a little expert advise on this one. TIA!

 

[Stephens]

I'm no expert, but did you use 2 vdev's? I ask because it sounds to me like what you ended up doing is creating one vdev, in which case the vdev was built using the size of the smallest drive. So what you have is a 5x1TB array (only using 1TB of the 2TB drives because you included 1TB drives in the vdev). You need to read noobsauce80's ZFS Presentation. There are definitely ways to achieve what you want, but I'll let the more knowledgeable answer that correctly. I believe you basically create a "RAID0" of the 2x1TB drives first, then create a ZPOOL with that plus the other 2TB drives to create (effectively) a 4x2TB RAID-Z1. I posted that to get you started in your thinking/reading, but please don't take my commentary as the final word.

On another note, a 2TB drive can be had for, what, $100 USD? I'd just get another 2TB drive and use the 1TB drives for something else (such as backup or off-line storage). The more complex you make your setup, the more complex it is to maintain it. When you're a self-professed noob, is that really the path you want to go down? Also, make sure you understand why RAIDZ2 exists. If you lose a drive with RAIDZ1, it's going to take ZFS a long time to rebuild the data using your remaining drives. If any of the remaining drives craps out because of all the constant I/O during the rebuild, you will lose your whole array.

 

[TooTHPiCK]

I made one pool in the GUI with 3x 2TB drives RAIDZ. Then I used the volume manager to "extend" the first pool with 2x 1GB drives set to stripe. It LOOKED like it did what I wanted and the size was close. The GUI then showed "Raid1+0" on the 2TB drives and "stripe" on the 1TBs. But when I removed a drive, to test the redundancy, it gave an error and could "..not recognize the pool size.." Not sure what that's about...

I am trying different configurations to get the most out of it. I partitioned all the drives into 16x 500GB chucks that I put in one RAIDZ3 pool. That gets me 6GB and it should be safe to recover 4 partitions (or more?) so the 2TB drives are safe. Am I understanding this right? I assume performance will suffer but I am not after speed. But, I could stripe the whole array and give it a boost... right?

It's FreeNAS 8.3 B1 running virtual on vmware ESXi 5 server. I had direct raw disk access working on the single partitions but, now I am using 16x 500GB virtual SCSI drives. That's three file systems layered on top of each other... And so far the speeds are as fast as my wireless will go so I'm happy! Still, I would love to hear from everyone who can actually understand this stuff!

 

[survive]

Hi TooTHPiCK,

There's really only one good way to do what you want to do if you want to do raidz with the drives you have right now, and that is to let FreeNAS treat all the drives as 1TB drives. All these gyrations you are attempting with the 500GB partitions will just add complexity and lead to heartbreak and disappointment later on. You can always replace the 1TB drives with 2TB drives later on and seamlessly grow the pool out to ~8TB.

Whether that's a good idea to do on drives that size is a whole different question.

-Will