Error "/dev/sda not large enough to join array" when trying to "Grow" Raid Mirror

  • Hello there


    First time forum-poster here :). I recently migrated from FreeNAS to OMV and while it's had it's challenges I think I'm largely there. I've created a RAID Mirror with 2 x 4TB disks and it's populated with data. I would now like to expand this mirror by adding another 2 x 3TB disks, however I now get the above error "/dev/sda not large enough to join array" when I use the "Grow" button in the Storage->Raid Management menu option. I can't imagine I need another set of 2 drive identical to the existing ones in the mirror.


    Has anyone perhaps come across this before and able to advise on how to resolve?


    Thanks very much
    Phil

    • Offizieller Beitrag

    I've created a RAID Mirror with 2 x 4TB disks and it's populated with data. I would now like to expand this mirror by adding another 2 x 3TB disks

    mdadm Raid always use amount of space on each disk equal to the size of the smallest disk. Because you created it with 4TB disks, that is the smallest disk allowed. If you would have started with the 3TB disks, you could add 4TB disks but it would only use 3TB worth of space.


    Either way, ask yourself why are you using raid anyway? You could setup two merger pools (unionfilesystems plugin) with one 3 tb disk and one 4tb disk each. Instead of using a mirror, setup a scheduled rsync job to clone one pool to the other.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | cputemp 7.0 | mergerfs 7.0.3


    omv-extras.org plugins source code and issue tracker - github


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

  • Thank you for your reply


    I guess my intention was to have 7TB of disk space that was mirrored across the four drives so that I could have some real-time mirroring. I'm not too hot on RAID setups so mirror seemed simple yet reliable enough for me. Does the way you suggest mean that I would have to rely on a delayed scheduled job to run before the data is synced?


    Thank you

    • Offizieller Beitrag

    Does the way you suggest mean that I would have to rely on a delayed scheduled job to run before the data is synced?

    Yes. Do you really need real time sync? realtime sync can bad sometimes especially when you accidentally delete something If you sync with rsnapshot instead of rsync, then you fairly well protected against ransomware as well.

    omv 7.0-32 sandworm | 64 bit | 6.5 proxmox kernel

    plugins :: omvextrasorg 7.0 | kvm 7.0.9 | compose 7.0.9 | cputemp 7.0 | mergerfs 7.0.3


    omv-extras.org plugins source code and issue tracker - github


    Please try ctrl-shift-R and read this before posting a question.

    Please put your OMV system details in your signature.
    Please don't PM for support... Too many PMs!

  • Thank you for your reply


    I guess my intention was to have 7TB of disk space that was mirrored across the four drives so that I could have some real-time mirroring. I'm not too hot on RAID setups so mirror seemed simple yet reliable enough for me. Does the way you suggest mean that I would have to rely on a delayed scheduled job to run before the data is synced?


    Thank you

    Hi


    Do you have LVM2 on top of the raid in use?


    If so the easiest way is to create a second (new) raid 1 on the two 3 TB disks and create a physical volume within the same volumegroup on it. After that you can grow the logical volume by 3TB and then grow the filesystem finally.


    That would end up like this:


    Raid 1: 2x4TB - MD1 : pv1
    Raid 2: 2x3TB - MD2 : pv2
    DG 1: consist out of PV1 and PV2
    LVOL1: running accross PV1 and PV2 (concatenated).
    FS in LVOL1


    If you have no LVM at the moment but you have enough space (or can temporarily move stuff) you can do the following:


    Create a second MD with 2x3TB drives (MD2).
    Create a new volume group
    Create a new physical disk on MD2 for the new volumegroup
    Create a new logical device on the Volume group (full extend)
    Create a new filesystem on the new device ...
    Then copy everything (cp-a) from old disk to new disk. Ensure that the drives are not actievly served anymore.


    After that has finished - create a new phyiscal Drive on MD1 (your old 2 TB).
    Add the PD to the volume group
    extend the volume and the filesystem
    Done


    You need to unmap and unmount everything from OMV guy before you start, as this operation conflicts with /etc/fstab and mountpoints and OMV might now mount anything anymore after you have manually done all this.

    Everything is possible, sometimes it requires Google to find out how.

Jetzt mitmachen!

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