Posts by luxflow

    hmm there is miscommunication about dataset


    zfs terminology
    Dataset is the generic term for a ZFS file system, volume, snapshot or clone.
    https://www.freebsd.org/doc/handbook/zfs-term.html
    you should understand term in above link


    so roughly,
    physical devices is used to build vdev(s)
    on vdev(s), pool is created
    pool can have several dataset (that means filesystem, volume, clone, snapshot) and have nested structure
    each dataset have different properties
    zfs filesystem behave like other filesystem
    zfs volume is block device, it means, it act as hard disk, so you can format with another filesystem on it(such as ext4, swap)
    use case for volume is providing disk for windows VM
    zfs snapshot is taken from filesystem or volume to save specific point


    generally,
    it is more wise to use zfs filesystem rather than creating another filesystem on volume in your case
    but you need encryption, so I suggest workaround.


    so my final suggestion is like..
    -tank (pool)
    --Document (volume)
    --Media (volume)


    go filesystem -> create -> choose device -> ext4 -> mount -> go share

    - BigPool 7.02Tb
    -vol01 documents 1Tb (snapshot)
    -vol02 Media 3Tb (volume)
    -vol03 dataset 1To (volume) (don't know what it is for exactly and how much space it needs)

    snapshot can be applied to any volume, any zfs filesystem (don't be confused filesystem(such as ext4) and zfs filesystem, they are different)



    I will be left with 2Tb not used. So if In a few months I see that documents needs more space, I'll be able to expend it, right?

    yes, using cli
    http://alittlestupid.com/2010/…how-to-grow-a-zfs-volume/
    keep in mind thing is ..
    shirinking filesystem process usaully more complex than expand
    so assign volume size conservatively


    Also, in order to see all my data from my windows PC, I guess that I'll need to create at least one shared folder per volume right?

    right, at least one shared folder per volume,
    you can also make share for subdirectories in volume

    one per your need
    one guide line for dividing dataset(whether it is zfs filesystem or volume) is attribute for that dataset
    zfs provide several attribute such as compression,disk size quota ..
    also zfs provide snapshot
    it depends on your need whether divide or not
    for example documents need compression on, but media don't want to compression on since media is already compressed
    and for documents, you need snapshot scheduling per hour (that means your changes is kept per hour)
    but you don't need snapshot per hour for media, then divide it
    (additional note: automatic snapshot is not supported in gui, use cli, I have plan for supporting gui)


    and if one driver fail(because raid z1 only support 1 drive fail), you need `zfs replace` command in shell (no gui support yet)

    zfs native encryption support is in progress
    https://github.com/zfsonlinux/zfs/pull/4329
    no one know when it is released, (my wild guess for release is within 6 month)


    and yes, you need to set encryption before creating dataset (according to above link)

    • Encryption must be specified at dataset creation time.

    so you need to move all data, and recreate dataset with encryption enabled


    I didn't test but I think you can use luks encryption currently with zfs volume
    create volume -> luks encryption -> format with ext4 -> mount and use

    no need to use OMV in this case
    just use official crashplan client..
    as I said that docker plugin is for uploading(upload backup), not for downloading
    that means it is for backuping OMV system



    I can think following 2 options
    1. use crashplan to backup and forget OMV, don't need
    or
    2. drop idea using crash plan, use samba & time machine to backup from windows and mac to OMV

    so do you want to backup from all other computer (including mac,windows) using crashplan
    and want to download those backup from crashplan to omv?
    that docker plugin is for uploading, not for downloading

    currently no native encryption for zfs
    that feature will be included next zfs release


    1. no you dont need use raid although it is recommended.. (parity hdd for failure)
    use omv-zfs plugin
    2. omv 3
    3. use omv install cd

    recommended plugin
    omv-flashmemory for usb boot
    usb 2.0 would affect long boot delay


    but I think ssd is better
    why? because you can install mysql for owncloud on ssd


    I recommend ZFS rather than linux raid
    because of bitrot protection, more flexible configuration, snapshot


    ZFS vs snapraid + union filesystem
    hmm it depends on..
    if you can buy multiple disks at once
    (2 for mirror, 3 for raid 5, 6 for raid 6)
    then zfs
    or
    buy one by one
    then snapraid + union


    raid configuration
    3 disk raid 5
    or
    6 disk raid 6
    or
    all miror
    http://louwrentius.com/the-hid…fs-for-your-home-nas.html


    yeah many people says no raid 5
    but it is not bad acually
    there is debate for this topic reddit/datahoarder check about it


    no need to hba unless need more sata port since sata 2 alreday saturate hdd full speed

    My choice is to install all of them to OMV3


    Those service works properly when your storage for those works properly
    and If installed separately, add overhead for connection between those services
    No benefit for seperation.

    Thumb up to skylake pentium, ECC, low power, ZFS
    Look cool, competive with other commercial NAS solution for me
    Considering power consumption, expandability, stability, I think it is more wise to choose Intel platform rather than ARM based board except price


    Do you have any specific reason to choose Fujitsu D3417-B?


    C232 chipset might be good choice since omv doesn't need gpu
    or turning off intel intergated gpu saves power AFAIK

    There are two main problem in omv-zfs plugin


    First is NFS shares problem
    NFS share prevent from mounting at boot


    Second is Fstab problem
    something like a ...
    mount: can't find /openmediavault in /etc/fstab'


    Both are solved, but it is not released yet, since OMV 3.0.60 is not released which omv-zfs depends on
    and I think OMV 3.0.60 release term will be longer than others due to many changes


    if you want to test these developer version
    install deb I attached like omv-extras and
    Developer->Setting->Install OMV from github
    Developer->Plugins->select omv-zfs->Package->Install


    but remember due to that 3.0.60 is not officially released
    others plugins may not works properly


    apt-get install --reinstall openmediavault openmediavault-zfs
    to rollback packages previous version