btrfs for storage works

    • OMV 3.x
    • btrfs for storage works

      I hadn't seen too much talk about it, but since I was redoing my NAS I figured it would be a good time to try using btrfs as the filesystem for storage instead of mdadm/ext4.

      Instead of using the (very nice!) WebUI setup a RAID, used the CLI to run something to the effect of:

      Source Code

      1. mkfs.btrfs -d raid10 -m raid10 /dev/sdb /dev/sdc /dev/sdd /dev/sde

      Then I was able to see the newly created FS in the WebUI and it's been gravy since. It uses it just like I would expect a normal storage location to be used.

      Plus, now I get all the benefits of having a filesystem that does file level check sums and protects against bitrot or bit flips.
    • Master_Scythe wrote:

      I wouldn't raid10 instead of Raid1 with only 4 drives on BTRFS.... its just increasing your chance of failure.
      2 pairs, with 1 drive each, versus 4 drives, with 2 failures total.

      Just making sure btrf balance is done regularly would work.

      Why did you choose Raid10? Im happy to be corrected!
      Im not 100% sure, but I went searching around to try to confirm doing Raid 1 on 4 drives and it seems that you would have the capacity of 1 drive in the end.
      It makes sense if you think about it, Raid 1 is mirror, so if you have more than 2 drives, it will just keep mirroring the drives, having basically 4 copies of your data instead of 2.
      Not very effecient unlesss you want to be really safe.

      In case of Raid 10 you add performance with reliability. Just to clarify, Raid 10 is a Raid 0 of (Raid 1 + Raid 1).
      You can loose a drive in each bank and still be fine.
    • This is not how btrfs works.
      Raid means 2 copy of data regardless of number of drives.
      Btrfs uses chunks/blocks for raiding not devices. That is how it can use drives of different size in raid setup.
      But why not use raid10?

      Sent from my SM-N910T using Tapatalk
      omv 3.0.56 erasmus | 64 bit | 4.7 backport kernel
      SM-SC846(24 bay)| H8DME-2 |2x AMD Opteron Hex Core 2431 @ 2.4Ghz |49GB RAM
      PSU: Silencer 760 Watt ATX Power Supply
      IPMI |3xSAT2-MV8 PCI-X |4 NIC : 2x Realteck + 1 Intel Pro Dual port PCI-e card
      OS on 2×120 SSD in RAID-1 |
      DATA: 3x3T| 4x2T | 2x1T
    • I only have 4 drives, RAID10 will give better performance and give better protection than RAID5. Plus rebuilds are simple and fast. So *if* something were to go wrong, I'm not going to be stressing over RAID5/6 parity rebuilds. A RAID10 rebuild is super fast, it's just a straight copy.

      A RAID calculator like this one can be useful. raid-calculator.com/default.aspx

      For me, I like having the extra read and write performance boost from RAID10, the ~1.5 drive loss protection and the fast rebuild times. I could gain more space if I used RAID5, but then I'd be loosing performance and protection.

      I'm hoping that btrfs dev's will move beyond RAID levels and add in some N+X style data protection.
    • clickwir wrote:

      I only have 4 drives, RAID10 will give better performance and give better protection than RAID5. Plus rebuilds are simple and fast. So *if* something were to go wrong, I'm not going to be stressing over RAID5/6 parity rebuilds. A RAID10 rebuild is super fast, it's just a straight copy.

      A RAID calculator like this one can be useful. raid-calculator.com/default.aspx

      For me, I like having the extra read and write performance boost from RAID10, the ~1.5 drive loss protection and the fast rebuild times. I could gain more space if I used RAID5, but then I'd be loosing performance and protection.

      I'm hoping that btrfs dev's will move beyond RAID levels and add in some N+X style data protection.
      What do you mean with "N+X style data protection" ??