Beiträge von RudyBzh

Well, after searches, readings, searches, readings, I think I found the reason here : https://github.com/nfs-ganesha/nfs-ganesha/issues/331

It's seems that I'm facing the exact same issue with the exact same desing :

Proxmox (ZFS pool) => 9p => VM (KVM) with OMV. And I'm trying to activate NFS on OMV on top of the 9p FS.

Looks like it's not possible, so I'll do another way.

Proxmox (ZFS pool) => bind mount => Container (LXC) => NFS sharing (without OMV I think...)

Hopefully, it will work this way.

Thanks for your advises.

Unfortunately, even after these steps :

- Unmount (no need to force)

- Disabling NFS service

- Removing share

- Apply on OMV

- Enabling NFS

- Creating share

The issue is exactly the same.

Zitat von Soma

Thema

Remotemount lost a NFS shared connection

First, posting this on Plugins, since NFS is running normal.

This is busting my head the last days and could use some pointers.
After doing the latest updates on both Pis (Server && Client), one of the 3 NFS shares I have in use, just disappeared from the clients:
Wireguarded Windows Explorer says that I don't have permission (always worked normal since I regedited to mount as the UID:GID of the owner of the files/folders.
And the other Pi on the same LAN with remotemount. (Keeps timing out)

On pi…

Soma

12. Februar 2023

See if this makes any sense.

Thanks for the share, I will have a look at these long explanations and try to understand because, to be honest, I a total newbie on NFS.

I’m just trying it for the 1st time due to issues I have with SMB. And this kind of errors just at the beginning are not really motivating.

Hi, no ideas please ?

Hi,

I manage to create a NFS share with OMV and permissions seems OK.

I can mount it, create files (touch) and folders.

But, I got an issue with folder. The folder is created correctly but if I do, just after, a "ls dir", I get a "Stale file handle".

rudybzh@docker:/mnt$ sudo mount -t nfs 192.168.1.4:/NAS /mnt/temp/
rudybzh@docker:/mnt$ mkdir /mnt/temp/dir
rudybzh@docker:/mnt$ ll /mnt/temp | grep dir
drwxrwsr-x    2 rudybzh users    2 Mar 28 09:32 dir
rudybzh@docker:/mnt$ ll -n /mnt/temp | grep dir
drwxrwsr-x    2 1000 100    2 Mar 28 09:32 dir
rudybzh@docker:/mnt$ id
uid=1000(rudybzh) gid=1000(rudybzh) groups=1000(rudybzh),24(cdrom),25(floppy),27(sudo),29(audio),30(dip),44(video),46(plugdev),100(users),109(netdev)
rudybzh@docker:/mnt$ ll /mnt/temp/dir
ls: reading directory '/mnt/temp/dir': Stale file handle
total 0

On the server side, the NFS config is as follow :

# This file is auto-generated by openmediavault (https://www.openmediavault.org)
# WARNING: Do not edit this file, your changes will get lost.

# /etc/exports: the access control list for filesystems which may be exported
#               to NFS clients.  See exports(5).
/export/NAS 192.168.1.8/32(fsid=xxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx,rw,subtree_check,insecure,all_squash,anonuid=1000,anongid=100)

# NFSv4 - pseudo filesystem root
/export 192.168.1.8/32(ro,fsid=0,root_squash,no_subtree_check,hide)

Any ideas of what it could be please ?

It's 100% reproductible on my system... whatever directory I create, I got this error.

Thanks.

Zitat von ryecoaaron

I'm sure it could be added. I can look but not sure how much effort I will be putting toward OMV 5.x plugins with OMV 6.x on the horizon. You can always mount 9p from the command line and then use the sharerootfs plugin to make shared folders.

Thanks.

As mentionned in all this thread, I have this 9p mounted through fstab. From my point of view, it works perfectly from OMV in SSH.

Using sharerootfs is not working for me (see the issues I have in OMV GUI and attached services : SMB in my case).

Using mergerfs makes it work...

Just to share for those who are interested...

The missing subdirectories are in reality accessible by using "cd subdirectory" (with exact name) on an SMB Client. So, it's as if some directories have not "read access" but only "execute access".

After looking in ssh (from where I can see every folders/subfolders without any issue, so it's not a "9p" issue for me) the permissions and others ACL on all the directories/files (previously reset via OMV plugin... but did it applied everywhere ?! Not sure), I can't find a single difference...

So, I don't know why I have this strange behavior.

By the while, I manage to get it work using a dirty solution => mergerfs !

I put my 9p folder /mnt/data (the one mounted from the underlying Proxmox ZFS) in mergerfs. Yes, he's alone...

Next, I used the created mergerfs to set my shares and voila... It works "normaly' => I can see all my folders.

Any idea/explanation on why it's working with mergerfs an not with shareroot plugin ?!

I just have OMV throwing me some errors on some shares when I apply config (same error as in #11).

It looks like it's only for the shares that are not at the root.

If I share "/dir", I have no error (OMV doesn't try to put any ACL, I think ?!)

If I share "/dir/subdir", I have the error (OMV tries to put a "execute" only ACL on "dir" from group users... Why ?! Also, I don't know why it's not working because it should ; And, finally, look at my 2nd phrase in this post...)

Well, it's not really a wonderful way of doing things but, without help, I don't know what to look at to understand why I have these kind of issues and how to do it a better way.

As every reads I made about passing ZFS to VM, neither Proxmox, neither OMV (or others...) wants to make it easy to work and every one is staying at his position... "blabla Proxmox on a baremetal" & "blabla NAS on a baremetal"...

Adding 9p on remotemount plugin won't be possible ?

I'm going to look a way to activate SMB shares directly from Proxmox (ZFS Shares). Should be more "natural"...

Thank you.

I feel a little alone... 😢

I do not intend to use ACL.

By the while, the error displayed is related to ACL.. is it the default behaviour of OVM ?

How could it work for several folders and not for some ?

I looked at folder/files with getfacl, and they are the same ones (default as made by a chmod) between a working and not working folder/file in a share.

What can I look at to troubleshoot this ?

In smb.conf, all the folder should be shared... Why some files are not visible and even with differences between a windows and a linux client ?

More strange, Windows and Linux don't see the same number of folders...

Lets take a shared folder as an exemple...

# On my ZFS, it contains 951 subfolders (ls | wc -l)

# On OMV mount (9p), it contains 951 subfolders also

On OMV WebUI, I have a "shared folder" for this folder.

If I try to edit it, I don't see every subfolders (don't know if webUI is only limiting the number of result or so...)

I have an SMB on this sharefolder that I can use on windows or linux

# On Windows, I see 765 subfolders

# On Unix (2 different VM), I see 506 subfolders

Same user ; Looked at the permissions, same for all ; no ACL. (I even use the reset permission utility to remove ACL and restore every perms...)

Really strange.

Next, if I try to create a new sharefolder for on a missing subfolder (manually typing the path), when I apply, I have this error :

Failed to execute command 'export 
PATH=/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/usr/local/sbin; 
export LANG=C.UTF-8; setfacl --modify group:users:--x /mnt/data 
2>&1' with exit code '1': setfacl: /mnt/data: Operation not 
supported

Hope this will help do diagnose something and point me to the right way

Hi,

Follow of my use case, I think I successfully implemented virtFS.

On my OVM VM, connected in SSH, I can see all my folders/files through the 9p mount.

BUT, and I really don't understand why, my shares doesn't include every folders... some are missing, depending of the share.

Even on OMV WebUI, if I go to create a new "Share folder", in the path selection, all folders are not there ?!

Any idea please ?

Zitat von ryecoaaron

I wouldn't use fstab. I would make a systemd mount file and either set a Before= dependency in the mount file for the services that need it or add overrides to the services to wait for the mount.

Thanks, will look for this.

I'm not wondering about the systemd but about identifying the before list...

Well, we'll see.

Thanks a lot for your (very) fast and clear responses.

Zitat von ryecoaaron

OMV doesn't work in containers.

I do. My VMs aren't file servers though. So, I just create virtual disks. But, I am working on a kvm plugin and I will replace proxmox on the host with OMV and the plugin. New openmediavault-kvm plugin

OK, so the container way to do is not a good idea...

When you say your VM are not file servers, what are you using OMV for ?!

Proxmox is the host and OMV is the guest ? Right ? But OMV is not a file server for you ?

I saw the plugin to replace Promox by OMV...

Well, It would be a big (r)evolution for me... BTW, Not sure to go this way...

It's proxmox job to deal with VM/CT, not OMV (in my point of view).

I'm asking myself if it's not easier to make OMV as a container and use bind mount in place of virtFS...

Should probably be more straight forward...

Just have to use the same "sharerootfs" as suggested and no issues with fstab mount...

Well, except the privileged vs unprivileged container choice...

I remember have read that you also are using proxmox & OMV ?

What solution did you use to use both ?

Thanks.

Thanks aaron, that's what I was thinking of.

Any advise on the fstab way to mount this ? "_netshare" and so on to be sure service dependencies are in the right order ?

I guess I will have troubles with this

Hi all,

I know the question have been asked a lot of time but I never feel comfortable with all answers I saw.

I'm running OMV as a VM on Proxmox.

I've just created a ZFS pool ("tank") in proxmox and want to give that full FS to OMV.

I successfully did this using virtFS (9p) passthrough. On the OMV VM config file on the host (proxmox), I add this line :

args: -fsdev local,security_model=passthrough,id=fsdev0,path=/tank/data -device virtio-9p-pci,id=fs0,fsdev=fsdev0,mount_tag=dataZFS

and can successfully mount it on OMV with this :

sudo mount -t 9p -o trans=virtio,version=9p2000.L,msize=512000 dataZFS /mnt/data

Performances are quite as good from guest (OMV) as from host (Proxmox).

Unfortunately, I stuck now on OMV side to be able to use this directory (/mnt/data) as shared because it's not recognize as a drive ?!

I could mount it somewhere else if it's relevant (/media ? /srv ?) ?

I can use remotemount add on, who doesn't know 9p FS...

I wondering about using root share add on ?

What would you suggest to do here ?

(from fstab, with good options to mount in the right order, to ability to create shares in this virtFS remote folder)

Thanks a lot.

Regards

Sorry, but I don't know. Consider you're lucky because, for me, it's not working anymore, and I can't understand why...

I opened a bug in monit tracker because of this...

https://bitbucket.org/tildesla…ory-usage-reported-in-kvm

No response yet.

Hi,

Any ideas about this ? How do you manage your SMB shares ?

Thanks a lot.

Hi,

I'm having an issue following this guide : How to make a good Media share for Emby and PlexMediaServer

I use to make "/" SMB shares but, reading this guide, it's "stupid" to do so.

So, I'm trying to follow these best practices and I've just discovered that a big functionality is missing doing so : hardlinks !

Sharing the root folder permitted me to make hardlinks between 2 top folders - say "Folder A" & "Folder B" - from unix clients having only 1 mount (and native mount options like serverino and/or mapposix)

Now, if I make 1 share for "Folder A" and another one for "Folder B", it seems that I cannot make hardlinks between them...

Invalid cross-device link

Am I missing something ? A SMB option on the server side ? Or on the client side (having 2 different mounts for "Folder A" & "Folder B") ?

How do you manage this ?

Thanks a lot for you advices.

Solution found and explained here : RE: mem usage matches resource limit