Mergefs folders

    • Offizieller Beitrag

    Yes, gparted before OMV. Is this a mistake

    Yes

    Waiting for an official explanation about how to use it correctly

    Did you read my post above, I have tried this in a VM and cannot get it to work, the other option maybe the cli.

  • Managed to mess permissions on /etc and had to reinstall OMV. Made backup using fsarchiver, but no option to boot recovery. Raspberry 4 here.

    Will take some time to reconfigure. Will be using whole drive, no partitions.

  • Last time I did it, it was like you said: "you can type the path to each shared folder in that box and it will work." The exception was that those three folders were not shared separately. The parent folder was shared and typed three paths. The plugin worked fine, it joined the folders into one.


    This time, after reinstalling OMV, shared each folder and typed the paths. It works fine this time, too.

    fstab looks the same.


    First time:

    Code
    /srv/dev-disk-by-label-hdd1part1/1/:/srv/dev-disk-by-label-hdd1part1/2/:/srv/dev-disk-by-label-hdd1part1/3/        /srv/26b47f43-0bba-4a25-bd4a-1b32ad55e8ed    fuse.mergerfs    defaults,allow_other,cache.files=off,use_ino,category.create=epmfs,minfreespace=4G,fsname=Utilizator:26b47f43-0bba-4a25-bd4a-1b32ad55e8ed,x-systemd.requires=/srv/dev-disk-by-label-hdd1part1,x-systemd.requires=/srv/dev-disk-by-label-hdd1part1,x-systemd.requires=/srv/dev-disk-by-label-hdd1part1    0 0


    Second time:

    Code
    /srv/dev-disk-by-label-hdd1part1/1:/srv/dev-disk-by-label-hdd1part1/2:/srv/dev-disk-by-label-hdd1part1/3        /srv/2c851eb2-47e4-405c-8b52-5942898adb93    fuse.mergerfs    defaults,allow_other,cache.files=off,use_ino,category.create=epmfs,minfreespace=4G,fsname=Utilizator:2c851eb2-47e4-405c-8b52-5942898adb93,x-systemd.requires=/srv/dev-disk-by-label-hdd1part1,x-systemd.requires=/srv/dev-disk-by-label-hdd1part1,x-systemd.requires=/srv/dev-disk-by-label-hdd1part1    0 0


    PROBLEM: OMV is not working when using this plugin if HDD is missing. It enters in safe mode.

    Do we have a solution for this situation?

  • You could try manually removing all those x-systemd.requires= statements in fstab. If that solves it for your use case, then don't use the plugin - code it yourself in fstab without those requires.

    --
    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

    You could try manually removing all those x-systemd.requires= statements in fstab

    That's interesting, the plugin creates those for each disk, I've been testing in a VM with 4 virtual disks, if you set it up with the plugin it's accessible, if you shutdown then remove one of the virtual disks the VM doesn't boot and goes into emergency mode, adding the drive back works.


    So if those were removed from fstab, to do that I would assume locating the reference within config.xml then run the salt deploy for fstab might be worth trying as I take that statement is not created/relevant if mergerfs is created manually.

    OMV is not working when using this plugin if HDD is missing

    Why would a drive go missing, if it's a USB possibly, if a drive fails that's a different story.

  • That's interesting, the plugin creates those for each disk, I've been testing in a VM with 4 virtual disks, if you set it up with the plugin it's accessible, if you shutdown then remove one of the virtual disks the VM doesn't boot and goes into emergency mode, adding the drive back works.

    I don't use the plugins. I code my mergers by hand in fstab outside of the openmediavault stanza. And there are no x-systemd.requires= statements present.


    If a disk is not present, there is a one minute thirty second timeout (per missing disk) before the boot continues normally. Of course, once booted any references to that merger are broken.

    --
    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 don't use the plugins. I code my mergers by hand in fstab outside of the openmediavault stanza. And there are no x-systemd.requires= statements present.

    I finally found your reply where you referenced that, I have been trying to understand how to do that by going back over mergerfs in github and just generally searching, and your own fstab entries.

    If a disk is not present, there is a one minute thirty second timeout (per missing disk) before the boot continues normally. Of course, once booted any references to that merger are broken.

    With the plugin you get the timeout as a disk is missing but it doesn't continue to boot normally due to the x-systemd.requires


    EDIT: You have stated that you don't use the plugin, I assume then that you install mergerfs from the cli

  • You could try manually removing all those x-systemd.requires= statements in fstab. If that solves it for your use case, then don't use the plugin - code it yourself in fstab without those requires.

    Does not solve it.

    Tryed within # >>> [openmediavault] and outside of if.

    • Offizieller Beitrag

    Does not solve it.

    Tryed within # >>> [openmediavault] and outside of if

    You cannot edit that, do a search for x-systemd.requires (from debian manpages) and look at this thread I'm trying to understand this myself, basically if a drive/disk/folder is missing the meregrfs mount point will not mount.

    • Offizieller Beitrag

    If added after created, by editing from gui, it does not show on /etc/fstab

    I have just tested this using the unionfs plugin, added the nofail to the options save, apply, reboot nofail gets added to the mergerfs mount point in fstab (the plugin actually tells you to remount or reboot for the changes to apply)

    The above actually works and the system will boot, is it accessible, not tested that.

    Screenshot of WebUI

    you can clearly see the drive as missing

    • Offizieller Beitrag

    The talk here is about mergefs folders.

    I know that!! but thanks for reminding me, I was in the process of testing using mergerfs folders but perhaps I won't bother and FYI they are one of the same as they use the same backend.

    • Offizieller Beitrag

    How?

    Add nofail to the options, click save, you do get the message to remount or reboot for the changes to take effect (in the unionfs plugin the apply option would appear and add the nofail to fstab) you therefore have to add the nofail to fstab manually, save the changes and reboot. I can the shutdown the server again remove a virtual disk and start the VM, but there is no indication that a drive is missing because the mergerfs folder plugin looks OK, the drive does get removed, tagged as missing under file systems.

    By adding this option you get no warning that a drive is missing.

Jetzt mitmachen!

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