I want to set up a nextcloud server on an Ubunutu VM and use a zpool on the node for the nextcloud data storage. I have created the zpool on the node and added it as storage for the VM but how do I mount it from the VM?
Mount zfs pool on node from linux vm for data storage
- Thread starter Richard Peters
- Start date
-
- Tags
- zfs
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Sounds like you're trying to pass through the zpool filesystem to the vm and attach it from within the vm? Akin to something like a iscsi target/nfs share? Not quite sure it works like that in this implementation, though i could be mistaken.
When you create a zpool, add it to proxmox as storage, and you add vm + disk, proxmox will create a zfs volume within that zpool and directly attach that volume to the VM. So essentially, you are already "mounting" that zpools storage volume to your VM...just as a virtual disk/block device.
So if you create "zpool001/foo" and define it as storage, then create a vm., call it vmid 100.
You'll get a volume called "zpool001/foo/vm-100-disk-1" which attaches directly to the VM.
Goes the same for if you add another pool...
"zpool002/foo", define it as storage, go to your vm, create a new disk, and point it at "zpool002/foo", then you'll get "zpool002/foo/vm-100-disk-2".
So you don't mount it from within the VM like you would an iscsi target/nfs mount, you just mount it as a new disk from the hypervisor side and it simply shows up as /dev/{v,d}d{a-z}....ect ect ect...
Sorry if I completely missed the mark here, i might have mis-understood what you're trying to accomplish.
When you create a zpool, add it to proxmox as storage, and you add vm + disk, proxmox will create a zfs volume within that zpool and directly attach that volume to the VM. So essentially, you are already "mounting" that zpools storage volume to your VM...just as a virtual disk/block device.
So if you create "zpool001/foo" and define it as storage, then create a vm., call it vmid 100.
You'll get a volume called "zpool001/foo/vm-100-disk-1" which attaches directly to the VM.
Goes the same for if you add another pool...
"zpool002/foo", define it as storage, go to your vm, create a new disk, and point it at "zpool002/foo", then you'll get "zpool002/foo/vm-100-disk-2".
So you don't mount it from within the VM like you would an iscsi target/nfs mount, you just mount it as a new disk from the hypervisor side and it simply shows up as /dev/{v,d}d{a-z}....ect ect ect...
Sorry if I completely missed the mark here, i might have mis-understood what you're trying to accomplish.
Dmora, thanks for that. I am a newbie with linux. I see the dataset as a block device (sdb in my case pointing to proxmoxpoolr1/vm-102) but how do I create automatic mount point within the vm's file system. If I edit the fstab, I have to provide a type but (a) the version of ubuntu doesn't support zfs and (b) I can't imagine you would want to muck around with zfs from within the vm when the dataset was created on the node. I suspect the answer is obvious, just not to me.
The VM doesn't know that its a ZFS vol, it just knows its a block device, so just go about your normal business with mounting it....
I think you're over thinking it
So fstab is formatted like this:
If you say you want to mount /dev/sdb, you'll want to use its UUID instead of /dev/sdb as that naming convention isn't static anymore.
Play by play:
I think you're over thinking it
So fstab is formatted like this:
# <file system> <mount point> <type> <options> <dump> <pass>
If you say you want to mount /dev/sdb, you'll want to use its UUID instead of /dev/sdb as that naming convention isn't static anymore.
Play by play:
Code:
# partition the disk
fdisk /dev/sdb
# format it
mkfs.ext4 /dev/sdb1
# record the UUID, you'll get something like this...
blkid | grep sdb1
/dev/sdb1: UUID="6f86ee0b-0fa1-4467-93ca-c94a9fdc65c7" TYPE="ext4"
# create the directory you want to mount it
mkdir /foo_bar
# edit fstab, and place your settings, again the format is the following separated by spacing
# <file system> <mount point> <type> <options> <dump> <pass>
UUID="6f86ee0b-0fa1-4467-93ca-c94a9fdc65c7" /foo_bar ext4 default 0 0
# force it to mount
mount -a
# see if it mounted
mount