Mergerfs + snapraid project

    • OMV 4.x

    This site uses cookies. By continuing to browse this site, you are agreeing to our Cookie Policy.

    • Mergerfs + snapraid project

      Hi all

      plan to change my NAS


      Currently I have a raid 5 on OMV 4.
      The lack of space in my current nas pushed me to invest in a supermicro SC836.
      I want to use mergerfs + snapraid to make evolution easier.


      I have some questions about this:


      how much space did you leave in the mergerfs settings (4G by default)?


      I see that the cache on the disks is not enabled by default in OMV, should it but like it or should it activate?


      I also saw that directories disappear in the sharefolders, what is the best thing to do to avoid this.


      All your advice is good to take!

      ps : I understand how work snapraid.

      Thank you
    • Disclaimer: New to OMV so I'm happy to be corrected about any mistake

      Revers62 wrote:

      how much space did you leave in the mergerfs settings (4G by default)?
      50G. It all depends on your block size (I use 512KiB instead of default 256KiB) and the number of files you have.
      Since I'm using 10TB disks, also for the parity, I increased the storage for the parity by 70GB by using the following format command:

      mkfs.ext4 -m 0 -T largefile4 -L LabelXYZ /dev/sdX1

      According to the SnapRAID documentation one can assume roughly half of the block size is wasted with 256KiB. I assume the higher the block size the higher the waste so let's assume 75% for 512KiB. Thus with 70GB headroom in the parity + 50GB free space on the drives (120GB) I can have a maximum of 312500 files. If I understand the parity calculation correctly this is not the maximum file limit for the whole pool but the maximum one drive can have.

      Revers62 wrote:

      I see that the cache on the disks is not enabled by default in OMV, should it but like it or should it activate?
      The cache is enabled by default since the hdparm settings shown in the GUI are not applied if you don't press "Save" for each disk at least one time.
      Personally I enabled disk cache since my 10TB drives have plenty (256MB) and I noticed significant SnapRAID sync speed increases with cache enabled (although it was enough to only enable cache for the parity drive).

      Revers62 wrote:

      I also saw that directories disappear in the sharefolders, what is the best thing to do to avoid this.

      This happened on every reboot for me. The problem is that OMV tries to initialize the shared folder before MergerFS mounts its pool thus resulting in an empty shared folder.
      Easy fix was to do the following:
      Comment out line 8 of the file:
      /etc/systemd/system/sharedfolders-<name>.mount
      Credit:
      forum.openmediavault.org/index…?postID=196466#post196466

      The post was edited 1 time, last by Paradido ().

    • Thanks Paradido!

      But I don't understand any rules for this free space if you have a link to check ?

      I will use my NAS with big files. +/- 50G per files

      Badblocks is running on 2x 10 Tb and a 8Tb come tomorow. long process !
      I will put
      1x 10 Tb for parity
      1x 10 Tb for data
      1x 8 Tb for data
      1x 4 Tb from my old nas but disk from last year
      1x 4 Tb from my old nas but disk from last year


      I have 5 x 4Tb on my old nas but I see max data disk for 1 parity disk is 4 data disk.
      I must buy another 10Tb for double parity and go for more data disk.

      Next week I will can play :)
    • Revers62 wrote:

      But I don't understand any rules for this free space if you have a link to check ?

      I will use my NAS with big files. +/- 50G per files
      snapraid.it/manual#7.8

      The default 4G of MergerFS will be enough for you if you format your parity drive with the command I mentioned above since you don't have a lot of small files.

      Revers62 wrote:

      I have 5 x 4Tb on my old nas but I see max data disk for 1 parity disk is 4 data disk.
      I must buy another 10Tb for double parity and go for more data disk.
      You don't have to. I'm running 7 disks with one parity disk without an issue. If I understand correctly the recommended drive count is there because the more disks you have the higher the chance of a bad block on one of the drives, thus you are losing data in the restore process. The more disks the higher the probability.

      The post was edited 1 time, last by Paradido ().