ZFS Proxmox Config

Proxmox ZFS Mount Pointsblog.kye.dev

1. Create ZFS pool in Proxmox UI : pve > disks > ZFS > Create ZFS

Create ZFS Datasets

Datasets in ZFS allow for organizing and structuring data within ZFS. You can create datasets on a Zpool to organize how and where specific data is stored. A dataset has no fixed size, it can expands to the pool size (or to the quota if set). For example you could have one datasets for your immich and another for your NFS share.

to create 'immich' dataset in 'tank' ZFS pool:

zfs create tank/immich  # zfs create POOL/DATASET

List ZFS pools/datasets

zfs list

Renaming datasets/snapshots

zfs rename pool1/data pool1/archive # Renaming dataset
zfs rename pool1/data@yesterday pool1/data@backup-2025-05-10  # Renaming snapshot

Migrating ZFS

Creating snapshot

zfs snapshot -r pool/dataset@migration

Sending snapshot

zfs send -R pool/dataset@migration | zfs receive -F newpool/dataset

You can also migrate whole ZFS pool by just not specifying any /dataset. E.g: zfs snapshot -r pool@migration ...

Access ZFS dataset in LXCs

Proxmox ZFS Mount Pointsblog.kye.dev

Proxmox setup (done once)

groupadd -g 110000 nas_shares
useradd nas -u 101000 -g 110000 -m -s /bin/bash
chown -R nas:nas_shares /tank/immich/
chown -R nas:nas_shares /tank/other_dataset/

Bind Mount Dataset to LXC

pct set 105 -mp0 /pool/dataset,mp=/mnt/media_root  # or whatever mount path

In LXC, adding nas_shares group

groupadd -g 10000 nas_shares
usermod -aG nas_shares $USER  # or whatever user needs to access data

Add disk to an existing pool

Use stable disk identifiers (e.g., /dev/disk/by-id/...) for all commands. You can see your disks ids by using ls -l /dev/disk/by-id/

Increase Redundancy: From 1 Disk to a Mirror (RAID 1)

Adds <new_disk> to duplicate the data from <existing_disk>.

zpool attach <POOL> <EXISTING_DISK_ID> <NEW_DISK_ID>

Example:

zpool attach rpool /dev/disk/by-id/wwn-0x5000c500c2f8832c /dev/disk/by-id/wwn-0x5000c500c2f8832d

Increase Capacity: Adding a New VDEV

Stripe (RAID 0)	Add a single disk to increase capacity without redundancy.	zpool add <pool_name> <disk_1>	Danger: Creates a RAID 0 stripe with the rest of the pool. Complete data loss upon any single drive failure.
New Mirror (RAID 10)	Add two mirrored disks to increase capacity with redundancy.	zpool add <pool_name> mirror <disk_1> <disk_2>	Increases the pool capacity by adding a new mirrored VDEV.
New RAIDZ-1	Add three disks with 1 parity disk to increase capacity.	zpool add <pool_name> raidz1 <d1> <d2> <d3>	Increases pool capacity with a new RAIDZ-1 VDEV.
New RAIDZ-2	Add four disks with 2 parity disks to increase capacity.	zpool add <pool_name> raidz2 <d1> <d2> <d3> <d4>	Increases pool capacity with a new RAIDZ-2 VDEV.

Expanding an Existing RAIDZ VDEV

zpool online -e <pool_name> <new_disk> (after physical addition) OR zpool expand <pool_name>

Verification after Adding

zpool status <pool_name>