Zfs

⚠️ Bluefin Delivery Options

Option A — ZFS inside Distrobox (Recommended for safety)

Run ZFS tools inside a Fedora Distrobox with DKMS. Keeps the host clean and doesn't risk breaking on bootc upgrade rebase.

Option B — ZFS via DKMS on the host (Advanced)

Install the ZFS kernel module on the host via COPR. Works, but carries risk: the DKMS module may not survive a bootc upgrade rebase if the kernel version changes.

Option C — External NAS (Recommended for homelab)

Run TrueNAS or an OpenZFS server externally. Mount shares on Bluefin via NFS or SMB. The host stays completely clean.

Option A — Distrobox

# Create a Fedora container
distrobox create --name zfs-box --image registry.fedoraproject.org/fedora:latest

# Enter and install ZFS
distrobox enter zfs-box
sudo dnf install -y https://zfsonlinux.org/fedora/zfs-release-2-5$(rpm --eval "%{dist}").noarch.rpm
sudo dnf install -y zfs
sudo modprobe zfs

# Verify
zpool list

If you need to create or import pools that require direct disk access, recreate the container with elevated privileges:

distrobox create --name zfs-box \
  --image registry.fedoraproject.org/fedora:latest \
  --additional-flags "--privileged --volume /dev:/dev"

Note: Privileged mode gives the container full access to host devices. Only use this when you need to manage actual disks from inside the container.

Option B — Host DKMS (Advanced, use with caution)

# Enable the ZFS COPR repository
sudo rpm-ostree install \
  https://zfsonlinux.org/fedora/zfs-release-2-5$(rpm --eval "%{dist}").noarch.rpm

# Install ZFS and the DKMS build tooling
sudo rpm-ostree install zfs zfs-dkms

# Reboot to activate the layered packages
sudo systemctl reboot

# After reboot, load the kernel module
sudo modprobe zfs

# Verify
zpool list

⚠️ Risk: After bootc upgrade, the DKMS module may not rebuild automatically if the kernel version changes. You may need to reinstall the DKMS module after a rebase. See Troubleshooting below.

Option C — External NAS (Recommended for homelab)

Mount a TrueNAS or OpenZFS share without touching the host kernel:

# NFS mount (one-time)
sudo mkdir -p /mnt/nas
sudo mount -t nfs 192.168.1.100:/mnt/pool/share /mnt/nas

# Persistent NFS mount — add to /etc/fstab
192.168.1.100:/mnt/pool/share  /mnt/nas  nfs  defaults,_netdev  0  0

# SMB/CIFS mount (one-time)
sudo mount -t cifs //192.168.1.100/share /mnt/nas \
  -o username=user,password=pass,uid=$(id -u),gid=$(id -g)

This is the lowest-risk option: the host never loads a ZFS module, and upgrades/rebases have no impact on your storage access.

Core ZFS Commands

Pool Management

# Create a single-disk pool
sudo zpool create mypool /dev/sdX

# Create a mirrored pool (RAID-1 equivalent)
sudo zpool create mypool mirror /dev/sdX /dev/sdY

# Create a RAIDZ pool (RAID-5 equivalent)
sudo zpool create mypool raidz /dev/sdX /dev/sdY /dev/sdZ

# Check pool health
sudo zpool status
sudo zpool list

# Import an existing pool (e.g., after moving disks)
sudo zpool import              # list all importable pools
sudo zpool import mypool       # import by name

# Export a pool (safe unmount before moving disks)
sudo zpool export mypool

# Scrub (integrity check)
sudo zpool scrub mypool

Dataset Management

# Create datasets
sudo zfs create mypool/data
sudo zfs create mypool/backups
sudo zfs create mypool/vms

# List datasets
sudo zfs list

# Set common properties
sudo zfs set compression=lz4 mypool/data
sudo zfs set atime=off mypool/data
sudo zfs set quota=100G mypool/data
sudo zfs set recordsize=1M mypool/vms    # better for large sequential I/O

# Get a property
sudo zfs get compression mypool/data

# Set mountpoint
sudo zfs set mountpoint=/mnt/data mypool/data

Snapshots

# Take a snapshot
sudo zfs snapshot mypool/data@backup-2024-01-01

# List snapshots
sudo zfs list -t snapshot

# Roll back to a snapshot (destroys data written after the snapshot)
sudo zfs rollback mypool/data@backup-2024-01-01

# Destroy a snapshot
sudo zfs destroy mypool/data@backup-2024-01-01

# Clone a snapshot into a new dataset
sudo zfs clone mypool/data@backup-2024-01-01 mypool/data-clone

Send / Receive (Backup and Replication)

# Full send to another pool
sudo zfs send mypool/data@snap1 | sudo zfs receive backuppool/data

# Incremental send (only changes since snap1)
sudo zfs send -i mypool/data@snap1 mypool/data@snap2 | \
  sudo zfs receive backuppool/data

# Send over SSH to a remote host
sudo zfs send mypool/data@snap1 | \
  ssh backup-host sudo zfs receive backuppool/data

Automated Snapshots with sanoid

sanoid manages automatic snapshots and retention policies via a config file. syncoid handles replication.

Install sanoid

# Inside Distrobox (Option A) — recommended
distrobox enter zfs-box
sudo dnf install -y sanoid

# On host (Option B)
sudo rpm-ostree install sanoid

Configure sanoid

# /etc/sanoid/sanoid.conf

[mypool/data]
use_template = production

[mypool/backups]
use_template = backup

[template_production]
frequently = 0
hourly     = 24
daily      = 30
monthly    = 3
yearly     = 0
autosnap   = yes
autoprune  = yes

[template_backup]
frequently = 0
hourly     = 0
daily      = 90
monthly    = 12
yearly     = 1
autosnap   = no
autoprune  = yes

Run sanoid

# Manual run
sudo sanoid --take-snapshots --verbose

# Prune old snapshots
sudo sanoid --prune-snapshots --verbose

# Automate with a systemd timer or cron
# cron example — every 15 minutes
*/15 * * * * root /usr/sbin/sanoid --take-snapshots --quiet

Mounting ZFS Datasets on the Bluefin Host (Option B only)

# Enable ZFS services so pools auto-import and datasets auto-mount at boot
sudo systemctl enable --now zfs-import-cache zfs-mount zfs-share

# Manual mount all datasets in a pool
sudo zfs mount -a

Troubleshooting

Module not loaded after reboot

sudo modprobe zfs

# If the module is missing (kernel update broke DKMS build):
sudo dkms install zfs/$(modinfo -F version zfs 2>/dev/null || echo "2.2") \
  -k $(uname -r)

Pool not found after rebase (Option B)

sudo zpool import              # list all importable pools
sudo zpool import -f mypool    # force import by name

Cannot import pool — wrong hostid

sudo zgenhostid                # generate a new /etc/hostid
sudo zpool import -f mypool

Container can't access /dev (Option A)

Recreate the Distrobox with device access:

distrobox create --name zfs-box \
  --image registry.fedoraproject.org/fedora:latest \
  --additional-flags "--privileged --volume /dev:/dev"

Pool health degraded

sudo zpool status -v mypool    # verbose — shows which vdev is faulted
sudo zpool scrub mypool        # start integrity scrub
sudo zpool clear mypool        # clear error counters after replacing a disk

Replacing a failed disk in a mirror

sudo zpool replace mypool /dev/sdX /dev/sdY   # replace old with new disk
sudo zpool status mypool                       # watch resilver progress

Decision Guide

References

• OpenZFS Documentation
• ZFS on Linux (zfsonlinux.org)
• sanoid on GitHub
• Distrobox documentation