omv-extras plugins - porting progress to OMV 6.x (done)

I have now also given OMV6 with mergerfs (on RasPi OS 64) a try. The problem I have is that while setting up a mergerfs pool with wildcards works fine and all files are there, after a reboot only the contents of one disk appear in the mergerfs mount.

I've checked the systemd mount file, and it includes the line

RequiresMountsFor=findmnt: options --target and --source can't be used together with command line element that is not an option

This looks like an error to me. I'm going to try to put the required mount point manually in there.

Zitat von ryecoaaron

It is. If you don't have a *, it won't work right. What was your wildcard?

/srv/dev-disk-by-uuid-*/mergerfs*

Indeed the RequiresMountsFor line was the issue. After putting the required mount points there manually, all my files showed up after reboot.

findmnt doesn't seem to like wildcards. On the shell this gave me the desired output, but I don't know if that is feasible for your implementation:

echo $(for i in /srv/dev-disk-by-uuid-*/mergerfs*; do findmnt --first-only --noheadings --output TARGET --target $i; done)

Zitat von ryecoaaron

True but nothing I can control. Since systemd converted fstab entries to mounts, I don't understand why fstab entries don't have the limit

This is true, however I think they still recommend using fstab as they are auto converted to mount entries as needed.: https://www.freedesktop.org/so…/systemd.mount.html#fstab

Zitat von ryecoaaron

I might just add a checkbox to the pool creation that will give the user the option to create a systemd mount file or /etc/fstab entry to get around the 256 character limit. I need to think about that.

I do like the idea of using an fstab or systemd mount option for creating a mergerfs volume (albeit double work) and could keep things working while a solution or deeper detail is found.

Regardless of mergerfs, is systemd just completing ignoring filesystems or fuse types that could have long What= fields? That does seem weird.

Even if short name symlinks are used, at some point, it too will surpass the 255 character limit.

As long as the limit is there unsure how much a systemd mount unit will be extensible

Zitat von KM0201

I've gotten so fond of symlinks I've ditched UUID's completely (unless I absolutely have to use it). They set up in seconds.. you can put them in a single centralized location (I have all mine under /NAS).. Obviously that doesn't matter much with nfs/smb, etc.. but when needing access to my files at the command line level, or writing docker-compose files.. it's WAY easier.

I guess this could work for now as I do like a more explicit way to set this up which makes it more idempotent.

There is a new 5.15 proxmox kernel,and it works great with Epyc, and also with older ryzen and intel.

hello

im using ver 6 and see KVM plugin was reported as done, i have installed and when trying to create a pool i see an error, the same error is no matter which "Type" i select

 500 - Internal Server Error
Failed to execute XPath query '//system/fstab/mntent[uuid='79684322-3eac-11ea-a974-63a080abab18']'. 

Zitat von ryecoaaron

I missed the sharerootfs plugin as a dependency

Seems to be working after installing that thanks

Hi Guys

Strange issue here

I've mounted my drives and created a folder in each of it.

Then I'm trying to merge them with merger Fs on OMV 6 (.0.34)

And

I get the error you see in the picture

(it works merging only two drives)

Zitat von zipporobotics

Hi Guys

Strange issue here

I've mounted my drives and created a folder in each of it.

Then I'm trying to merge them with merger Fs on OMV 6 (.0.34)

And

I get the error you see in the picture

(it works merging only two drives)

Alles anzeigen

2 posts before your's has the reason why, you only had to look UP.

RE: omv-extras plugins - porting progress to OMV 6.x

Zitat von ryecoaaron

mergerfs change (not in repo yet) - uses symlink for each branch (if the branch is not using wildcard). Should be able to get up to 15 branches (depending if subfolder is chosen) in the 256 limit. The symlink will be in /srv/mfs and be an 8 digit crc32 value created from the long path (Volker's idea). Still testing this. Working on fstab option too.

No, i don't usually use ntfs on linux machines, either extfat or something native.