How do i create a Jbod?

[Chris230291]

Hello can someone please explain how to create a volume that i can just add disks to as and when i please? I believe the term is called concatenated or JBOD?

I have 2 drives at the moment, a 1 and a 2 TB. I want them to appear as one drive on the Freenas server, but i would also like to be able to add more drives when they are needed without formatting all of the disks. If someone could help me with this it would be a huge help! I have looked for and followed guides, but i found very few and all of them have different interfaces with different buttons to the one im using (i have the latest i think).

Chris

 

[survive]

Hi Chris230291,

Do you want any sort of redundancy (protection against drive failure)?

-Will

 

[Chris230291]

WOW thanks for the quick reply Will!

Hmm well this is something im unsure of. I would be annoyed if i was to loose all of the data across all of the drives, but if one disk failing means that only the files on that disk are lost, im cool with that.

Thanks,
Chris.

 

[louisk]

If you concatenate disks together, a failure of 1 disk means an unrecoverable failure of the volume (read: all data lost).

I would suggest you start with something like RAIDZ, where you have a minimum of 3 spindles, and 1 is used for parity/checksum. This leaves you with the capacity of 2x. For example, if you have 3x2T in your ZFS pool as a RAIDZ, you would have ~4T of storage available. You would be able to have 1 failure (at a time) w/o losing any data (just be sure to replace the failed disk before another failure happens).

 

[Chris230291]

Thanks for the reply louisk.

So one disk failure means i cant recover data from the "good" disks? I thought one drive was filled, then it moved on to the next? This is ideally what i want, rather than spreading files across all of the drives. The data is only media (DVD's and Blu Rays) so a loss isn't a big deal as such, but a complete loss would be a massive kick in the balls as i would need to re rip everything. I think i might be in a little over my head here, but please try to bear with me.

Also i think i better explain what i want at the end of this incase that interferes with how i set this up. I Basically just want the Volume/s to show up in OS X, Linux and also be discoverable by DLNA and UPnP devices (xbox, blu ray players, smart TV's). I want everything to be readable by everyone, but only i can write.

Thank you,
Chris

 

[louisk]

If you spread a single filesystem across multiple disks with no parity, a failure of 1 disk causes the filesystem to be unusable. The solution for that is to either:
have multiple independant file systems (such as 1 per disk)
or have multiple disks with parity and a single filesystem

The cheapest way to do this would be with RAIDZ, and 3 spindles.

I would suggest that you setup a RAIDZ volume in ZFS (minimum of 3 spindles, max of 8 per vdev, you can have more than 1 vdev is you wish).
Share this volume with CIFS. OSX and Linux can both read CIFS just fine.

DLNA is not supported in a stable build of FreeNAS yet, so I would suggest you wait. Its currently in BETA, and I would expect in a couple months, it will become available.

 

[Chris230291]

Sorry, that went right over my head. Im going to look some words up.

 

[Chris230291]

Ok thanks i now understand all of that.

What you suggested sounds good but i only have 2 drives, a third drive has data on it that i want to save. Could i use a small USB drive to make up the numbers, then remove it once i have transfered the data over off of the third drive and added that drive to the volume?

I can just use a media sharing client on my computer for the DLNA until theres a release.

Thanks,
Chris

 

[louisk]

They would need to all be the same size (if you have a RAIDZ with 3 spindles, all 3 have to be the same size, but if you have 2x RAIDZ, you could have 3 of 1 size, and 3 of another size).

Once you add a volume, you can't remove it.

 

[survive]

Hi Chris23029,

Just to clarify, if you have drives of different sizes ZFS will use the smallest size drives e.g. 1 X 1TB, 1 X 2TB & 1 X3TB in raidz they would "act" like 1TB drives (and give you ~1.8TB of storage).

If you swap a 3TB drive for the 1TB drive in the above example the drives would suddenly all act like 2TB drives giving you ~4TB of space and when you swapped out the 2TB all the available space would become available.

-Will

 

[Chris230291]

Ok thanks.
This sucks, i just bought a new 2TB drive and my other two drives are 1TB. One of them is full of data too.
I dont know what to do.

When i create a concatenated array in OS X, i can use completely different size drives with no redundancy. How come i cant with this?

 

[Onichan]

Sounds like you want a non standard RAID type of system. There is a good quick overview of some popular ones here http://snapraid.sourceforge.net/compare.html. I haven't used any so not sure which to recommend.

 

[Chris230291]

Thanks Onichan.

I am going to try the windows server route. That SnapRAID sounds like it's worth a go.

 

[louisk]

Ok thanks.
This sucks, i just bought a new 2TB drive and my other two drives are 1TB. One of them is full of data too.
I dont know what to do.

When i create a concatenated array in OS X, i can use completely different size drives with no redundancy. How come i cant with this?

Concatenation doesn't require disks to be the same size. But be warned, concatenation in OSX is still just as fragile. If you lose one disk, the entire volume is gone.

 

[Chris230291]

Its ok .

An I right in saying if I create a spanned array in windows server, I can then add a thirt disk and create a raid 5 without loosening the data written from the spanned array?

Thanks.

 

[louisk]

From what I just read (I'm not a windows person and don't pretend to know detailed information about how it works), a spanned array is a concatenation (not to be confused with striping where it will write a chunk to each disk in the array sequentially).

Whether you can change a concatenation to a RAID in windows, I have no idea. Probably look for that answer in a different forum.