ZFS on OMV6 - openmediavault (2024)

Zitat von geaves

either the proxmox kernel is not used or it's a must use, to enable zfs to function from a user perspective and yours

But the proxmox kernel is not required. It makes things easier though. It is just hard to get the dependencies in the plugin package to do the right thing depending on which kernel is installed. The plugin also works on non-amd64 platforms. So, the proxmox kernel cannot be a requirement there.

The following allows installing the zfs plugin when the proxmox kernel is installed because it forces the proxmox packages to be used.

apt-get install zfsutils-linux=2.1.1-pve3 zfs-zed=2.1.1-pve3

apt-get install openmediavault-zfs

Zitat von chente

To what extent is the proxmox kernel necessary?

It is never necessary. It just skips the compilation of the zfs module which helps installation time and reliability. The kernel is the Ubuntu LTS kernel which is tested with zfs extensively and in my opinion, a better choice for zfs. The problem right now happens whenever debian backports has a never version of zfs packages than the proxmox repos. Disabling backports prevents that.

Zitat von ryecoaaron

Also, on a fresh install with the proxmox kernel installed, disabling backports and clicking the apt clean button in omv-extras allowed the zfs plugin to be installed.

that worked by running apt clean as the backports were already disabled from your previous suggestion.

So if I redeploy, but run apt clean after disabling the backports this should be the way to go.

Zitat von chente

See Also

OpenMediaVault vs. TrueNAS (FreeNAS) in 2024 - WunderTech[Howto] Setup ZFS on OMV Step by Step for ZFS Fans. - openmediavaultOpenMediaVault and ZFS - It works?OpenMediaVault vs. TrueNAS (FreeNAS) in 2024 - WunderTech

Precisely for this reason I stopped using the proxmox kernel. I always try to keep the system as simple as possible to avoid complications.

That may "simplify" one thing but disabling backports simplifies another. I also think the proxmox kernel is better than the debian backports kernel. I have run it on all of my amd64 systems since I added the option to omv-extras.

Zitat von geaves

So if I redeploy, but run apt clean after disabling the backports this should be the way to go.

Yep. It is always safe to run apt clean (or omv-aptclean).

Zitat von ryecoaaron

I also think the proxmox kernel is better than the debian backports kernel

Because it is better? What advantages will I notice?

Ok redeploy;

1) Install OMV6

2) Login to GUI

3) Set time zone (first time around I saw some errors relating to this)

4) Install updates from Update Management

5) Install omv-extras

6) Omv-Extras -> Settings uncheck backports -> run apt clean

7) Install kernel plugin

-8) Install Proxmox kernel

9) Reboot

10) Remove non-Proxmox kernels

11) Install sharerootfs plugin

12) Install zfs plugin

Houston, we have lift off

Zitat von chente

Because it is better? What advantages will I notice?

The only one I think you will notice is that upgrades will not have to compile the zfs module every time the kernel is upgraded. Other than that, you will probably not notice anything. The kernel will just run and work. Take it from someone who uses this kernel (ubuntu lts) on hundreds of systems.

Hi guys, quick question about ZFS. Does RaidZ (the equivalent to Raid5) needs to make a parity sync the same way Raid5 needs to do?

Yesterday I installed OMV 6 on my home server and decide to use ZFS. Installed OMV Extras, Kernel plugin. Then installed Proxmox Kernel

Then I created the Zpool

sudo zpool create -m /srv/nas-data-zpool nas-data raidz /dev/sda /dev/sdc /dev/sdd----sudo zpool status pool: nas-data state: ONLINEconfig: NAME STATE READ WRITE CKSUM nas-data ONLINE 0 0 0 raidz1-0 ONLINE 0 0 0 sda ONLINE 0 0 0 sdc ONLINE 0 0 0 sdd ONLINE 0 0 0errors: No known data errors

On the UI, I've imported the pool and it seams ready to be used, but I thought I need to do a parity sync.

Other question, now about the way it shows on the file systems tab. It is supposed to show false on the device?

Zitat von ryecoaaron

The only one I think you will notice is that upgrades will not have to compile the zfs module every time the kernel is upgraded.

See Also

ZFS on OMV 7 - openmediavault

That is an advantage, yes. I guess this affects how long the update lasts.

Zitat von wultyc

On the UI, I've imported the pool and it seams ready to be used, but I thought I need to do a parity sync.

It is not necessary, this is one of the advantages of ZFS. The creation of a pool is instantaneous.

https://pthree.org/2012/04/17/…l-zfs-on-debian-gnulinux/

Zitat von wultyc

Other question, now about the way it shows on the file systems tab. It is supposed to show false on the device?

I don't know how to answer this, I have not yet migrated my real system to OMV6.

Zitat von wultyc

zpool create -m /srv/nas-data-zpool nas-data raidz /dev/sda /dev/sdc /dev/sdd

I think this is where it has gone wrong, installing the sharerootfs plugin allows the name of the pool to be placed in the root of OMV, so what I did (as a test on a VM) was

zpool create tank raidz1 /dev/sdb /dev/sdc /dev/sddzpool status pool: tank state: ONLINEconfig: NAME STATE READ WRITE CKSUM tank ONLINE 0 0 0 raidz1-0 ONLINE 0 0 0 sdb ONLINE 0 0 0 sdc ONLINE 0 0 0 sdd ONLINE 0 0 0errors: No known data errors

then for a filesystem zfs create tank/movies

After each cli command I imported on the zfs plugin

Zitat von geaves

I think this is where it has gone wrong, installing the sharerootfs plugin allows the name of the pool to be placed in the root of OMV, so what I did (as a test on a VM) was

Yes sure. Was my intention to mount my zpool on /srv/nas-data-zpool. My question was about the Filesystem tab showing false on the device. I understand maybe the filesystem tab is not ready for ZFS yet.

But on the ZFS tab everything looks fine

Zitat von wultyc

Other question, now about the way it shows on the file systems tab. It is supposed to show false on the device

No it should display like this;

Ok strange

Well I destroyed the pool and created a new one without the mounting point and it appears

sudo zpool create nas-data raidz /dev/sda /dev/sdc /dev/sddsudo zpool status pool: nas-data state: ONLINEconfig: NAME STATE READ WRITE CKSUM nas-data ONLINE 0 0 0 raidz1-0 ONLINE 0 0 0 sda ONLINE 0 0 0 sdc ONLINE 0 0 0 sdd ONLINE 0 0 0errors: No known data errors

But when I set the mounting point it shows as false

sudo zpool create -m /srv/nas-data-zpool nas-data raidz /dev/sda /dev/sdc /dev/sddsudo zpool status pool: nas-data state: ONLINEconfig: NAME STATE READ WRITE CKSUM nas-data ONLINE 0 0 0 raidz1-0 ONLINE 0 0 0 sda ONLINE 0 0 0 sdc ONLINE 0 0 0 sdd ONLINE 0 0 0errors: No known data errors

Is it the difference between nas-data-zpool and nas-data

so if you used

zpool create -m /srv/nas-data nas-data raidz /dev/sda /dev/sdc /dev/sdd or

zpool create -m /srv/nas-data-zpool nas-data-zpool raidz /dev/sda /dev/sdc /dev/sdd

I would say the problem is that you are trying to create the mount point after creating the pool.

When the pool is created, the mount point is automatically created. If you do not want the default mount point, you must define it when creating the pool. To do this later, you would have to disassemble and reassemble it.

https://docs.oracle.com/cd/E26921_01/html/E25823/gaynd.html

geaves I used this one and the result was the same zpool create -m /srv/nas-data nas-data raidz /dev/sda /dev/sdc /dev/sdd

chente using the set endpoint had the same result. started showing the pool and then shows false

# sudo zpool create nas-data raidz /dev/sda /dev/sdc /dev/sdd# sudo zfs get mountpoint nas-dataNAME PROPERTY VALUE SOURCEnas-data mountpoint /nas-data default

# sudo zfs set mountpoint=/srv/nas-data-zpool nas-data# sudo zfs get mountpoint nas-dataNAME PROPERTY VALUE SOURCEnas-data mountpoint /srv/nas-data-zpool local

But if I set to the original value is shows again

# sudo zfs set mountpoint=/nas-data nas-data

Isn't that of an issue for me tbh, I was simply trying to have all mount point in the same folder