First steps to create a ZFS mirror

[andersostling56]

So I have a PVE setup on an /dev/nvme1n0 (OS only) and two blank 1TB disks that I want to use for storage of VM's, templates etc.

Should I create an md (software raid) before adding the mounted mirror as a Directory?

Or would it be better to let PVE create a ZFS mirror from the two blank disks? If so, what are the steps to achieve this?

The third step that I see (for now) is to manually create and mount a dataset, and then use the GUI to add the mounted directory as new Storage

mkdir /mnt/zfs-vm
zpool create zfspool1 mirror /dev/sda /dev/sdb
zfs set mountpoint=/mnt/zfs-vm zfspool1

Any advice for a beginner?

 

[mariol]

Yes you can use Softwareraid (mdadm) too, but it is not official support. We recommend using a hardware raid controller or ZFS Raid [1].
For ZFS you can also use the WebUI, which will take care of all the steps for you. Here is an example with a mirror:



If you want to create an extra dataset for the VMs, uncheck the “Add Storage” checkbox (create the pool only). Then create a new dataset on the CMD with “zfs create <pool>/<dataset>”.
After you are able ton add this as ZFS VM storage [2] under “Datacenter -> Storage”.

The third step that I see (for now) is to manually create and mount a dataset, and then use the GUI to add the mounted directory as new Storage

You can do this, but do not use the directory store (ZFS) for VMs or containers. You would be using cow on cow. For ISO images, backups... perfectly ok. You don't have to specify a mountpoint either, it is created automatically when you create a dataset. Unless you want to change it.


[1] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_zfs
[2] https://pve.proxmox.com/pve-docs/pve-admin-guide.html#storage_zfspool