Which file system/RAID-ish solution to choose?

  • I'm a long time Windows Home Server user, and I'm finally about to EOL my server "Columbia". I'm currently in the process of acquiring the hardware to build the new server "Leviathan", and have been testing various implementations of server software in VirtualBox. After much fiddling, I've decided on OMV. With that being said, I've also identified various plugins that will help me replicate my WHS experience for my user (my wife!) so she doesn't have to learn too much new stuff.


    That being said, I'm having trouble figuring out file system/pooling/parity solution to choose. So I figured I'd ask.


    I plan on starting my server with 5 x 4TB drives and the design of my server has space for future total of 15 drives. I'm planning on reserving 2 drives for parity. I want to be able to add drives of various sizes (whichever is on sale/reuse some of the drives from my WHS - they're 2TB), and I want to be able to incrementally grow my storage. i.e. - I have the original 5 x 4TB drives, add in 4 x 2TB from my WHS, if I start to run out of space and 8 TB drives are on sale? etc...


    Because of this, I'm moving away from a traditional RAID solution, since I don't want to be limited to the smallest drive as I grow (nor do I want to fiddle with multiple partitions on one drive).


    Fundamentally, I'd like some sort of JBOD solution that allows expansion of the array, and expanding with different size drives. I'd also like some level of data replication/parity, so that with up to 2 drive failures - I lose minimal to zero data.


    I've looked at Greyhole, unionfs, mergerfs, etc and can't seem to really tell pluses/minuses and what is the best solution. SnapRAID looks like it might be the only solution for the parity portion, but if there's a better idea - I'm all ears. Thanks in advance!

    Is this old airplane safe to fly? How in the world do you think it got to be this old?

  • I use ext4 formatted drives with SnapRaid with two parity drives of 12TB each and six data drives of mixed sizes of 3TB, 4TB, 8TB and mergerfs. The current UnionFS solution within OMV is mergerfs, so there is nothing to compare between those. Greyhole on current OMV is AFAIK deprecated.

    --
    Google is your friend and Bob's your uncle!


    OMV AMD64 7.x on headless Chenbro NR12000 1U 1x 8m Quad Core E3-1220 3.1GHz 32GB ECC RAM.

  • I use ext4 formatted drives with SnapRaid with two parity drives of 12TB each and six data drives of mixed sizes of 3TB, 4TB, 8TB and mergerfs. The current UnionFS solution within OMV is mergerfs, so there is nothing to compare between those. Greyhole on current OMV is AFAIK deprecated.

    Thanks! I couldn't really tell if there was a difference between the two, and saw on a webpage a reference to them as two different pooling solutions. If they're essentially the same, and mergerfs is built in - then I think going with that might be easiest.


    As for SnapRAID - I like your idea of two big parity drives, but it's not in the budget right now (I've already blown through the budget, and my wife keeps raising her eyebrow as I order more stuff) - can I swap in larger parity drives later on down the line?

    Is this old airplane safe to fly? How in the world do you think it got to be this old?

    • Offizieller Beitrag

    As you've described your requirements, I'd say, take a closer look at UnionFS+SNAPRAID. I've been using the combination on one server, as a long term test, and I've found that both are living up to their advertised capabilities.


    UnionFS will pool drives of dissimilar size, even with dissimilar formats, and a new drive can be added as needed.
    SNAPRAID will protect all selected drives, as long as the parity drive(s) are larger than those to be protected. SNAPRAID parity drives can be removed at any time, replaced, or SNAPRAID can be abandoned altogether, without affecting the UnionFS pool. SNAPRAID can restore a protected drive, or even files and folders, to the state they were in as of the last SYNC.
    (Full disclosure - I'm not running 2 parity drives so I don't know if there are any idiosyncrasies involved in that setup. I don't believe there are.)


    There's a few setting selections to be aware of, going in, but I'm real happy with the combo so far.
    ___________________________________________


    **Edit: It seems there is a reply posted, along the same lines.**

  • As for SnapRAID - I like your idea of two big parity drives, but it's not in the budget right now (I've already blown through the budget, and my wife keeps raising her eyebrow as I order more stuff) - can I swap in larger parity drives later on down the line?

    SnapRaid requires that your parity drives be either the largest in the array or you don't quite fill up your data drives 100% if they are the same size as the parity drives. I have grown my array over the years by buying new drives that are larger than any existing drives and making these new larger drives my parity drives. The former parity drives, being smaller than the new ones are reused as data drives. So yes, you can swap in larger parity drives at any time.

    --
    Google is your friend and Bob's your uncle!


    OMV AMD64 7.x on headless Chenbro NR12000 1U 1x 8m Quad Core E3-1220 3.1GHz 32GB ECC RAM.

    • Offizieller Beitrag

    I have grown my array over the years by buying new drives that are larger than any existing drives and making these new larger drives my parity drives

    That's another aspect that I consider to be a good feature. It's best that parity drives be relatively new and clean. In an hardware refresh, previously used parity drives, with a bit of age on them, can safely be reused and protected by newer parity drives.

  • Thanks for all the suggestions. After setting up unionfs on the virtualbox, and watching some videos on SnapRAID - I think you guys are right, that is exactly the solution I'm looking for.


    The only weird thing I noticed was trying to link it with Plex, had to identify each individual HD's Media folder with Plex vice the pool that I created with unionfs. Not a big deal, but I could see as HDs get more and more (just two in the test environment) it could turn into a PITA.

    Is this old airplane safe to fly? How in the world do you think it got to be this old?

  • Thanks for all the suggestions. After setting up unionfs on the virtualbox, and watching some videos on SnapRAID - I think you guys are right, that is exactly the solution I'm looking for.


    The only weird thing I noticed was trying to link it with Plex, had to identify each individual HD's Media folder with Plex vice the pool that I created with unionfs. Not a big deal, but I could see as HDs get more and more (just two in the test environment) it could turn into a PITA.

    I had no problems referencing the pool in Plex. Plex is extremely touchy about permissions, so verify that your pool directory has the correct permissions/ownership by comparing them to those of one of the individual media folders.

    --
    Google is your friend and Bob's your uncle!


    OMV AMD64 7.x on headless Chenbro NR12000 1U 1x 8m Quad Core E3-1220 3.1GHz 32GB ECC RAM.

  • I had no problems referencing the pool in Plex. Plex is extremely touchy about permissions, so verify that your pool directory has the correct permissions/ownership by comparing them to those of one of the individual media folders.

    I've gone through all the permissions/ownership with the pool. I realized what was going on - I didn't catch that in Plex the unionfs pool is identified by the device label. I was looking at the SMB shares folders vice the device folder. Once I removed all the other library folders, browsed to the correct unionfs folder, there it was.


    Rookie mistake, I know. :)

    Is this old airplane safe to fly? How in the world do you think it got to be this old?

Jetzt mitmachen!

Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!