How to migrate from Proxmox VE 8 to 9: step-by-step guide [2026]

Proxmox VE 9 is now available, bringing as its main new features the Debian 13 “Trixie” base, the Proxmox kernel 6.14 (based on Ubuntu 24.04 LTS), and ZFS 2.3 with draid support and compression improvements. If you are still on PVE 8, the migration is straightforward, but requires attention to a few details — especially repositories, cluster, and storage compatibility.

This guide covers the complete step-by-step process to migrate your Proxmox VE 8.x node to 9, including the exact commands, post-upgrade checks, and the most common pitfalls.

Before you start: requirements

PVE 9 maintains the same hardware requirements as PVE 8:

x86-64 CPU with VT-x/AMD-V and AES-NI (2+ cores, recommended 4+)
Minimum 4 GB RAM (recommended 8 GB+)
Minimum 32 GB disk for the system
Clusters: all nodes must be on PVE 8.x and healthy

Pre-migration checklist

Before any upgrade command, ensure that:

Full backup of configurations and VMs/CTs (vzdump, PBS, or snapshot)
Cluster in quorum: pvecm status — all nodes online
PVE 8 on the latest patch: apt update && apt dist-upgrade -y
Sufficient free disk space (at least 5 GB)
Backup restore test in an isolated environment (recommended)

If you have a cluster, the upgrade must be done node by node, removing HA from VMs before each node.

Step-by-step migration

1. Adjust the repositories

PVE 9 uses the same bookworm repository naming (don’t be fooled: PVE 9 packages are still in the repository called bookworm for internal compatibility, but they point to packages compiled for Debian 13).

For users with an enterprise subscription:

cat << EOF > /etc/apt/sources.list.d/pve-enterprise.list
deb https://enterprise.proxmox.com/debian/pve bookworm pve-enterprise
EOF

For users without a subscription (free repository):

cat << EOF > /etc/apt/sources.list.d/pve-no-subscription.list
deb http://download.proxmox.com/debian/pve bookworm pve-no-subscription
EOF

Remove old or incorrect repositories:

rm -f /etc/apt/sources.list.d/pve-enterprise.list
rm -f /etc/apt/sources.list.d/pve-no-subscription.list
# Then recreate with the content above

Note: If you used third-party repositories (Ceph, ZFS), check if they have a version compatible with Debian 13.

2. Update the Proxmox packages

apt update
apt dist-upgrade -y

This installs the latest PVE 9 package versions still in the bookworm repository. The system remains technically on Debian 12, but with updated Proxmox packages.

3. Update the Debian base (Bookworm → Trixie)

Now you switch the system base to Debian 13:

sed -i 's/bookworm/trixie/g' /etc/apt/sources.list

Check that no other files in /etc/apt/sources.list.d/ still point to bookworm:

grep -r "bookworm" /etc/apt/sources.list.d/

If found, evaluate whether they should also be updated (third-party repositories may or may not have a Trixie version).

Then perform the full upgrade:

apt update
apt upgrade -y
apt full-upgrade -y

4. Reboot

reboot

After reboot, verify the active kernel:

uname -r

The expected output is something like 6.14.x-1-pve (Proxmox kernel 6.14). If a 6.8 kernel still appears, check the bootloader:

dpkg -l | grep pve-kernel-6.14
proxmox-boot-tool status

5. Post-upgrade checks

# Cluster status (if applicable)
pvecm status

# Web service
systemctl status pveproxy

# ZFS storage
zpool status
zfs list

# VMs and containers
qm list
pct list

# Error logs
journalctl -p 3 -xb | grep -i error

# Proxmox version
pveversion

The pveversion command should return pve-manager/X.X.X with version 9.x.

What changed in PVE 9

Kernel 6.14

PVE 9 comes with the Proxmox VE Kernel 6.14, based on the Ubuntu 24.04 LTS kernel. Improvements include a more efficient scheduler, support for newer hardware (network cards, NVMe controllers, GPUs), and security fixes. The legacy kernel (5.15) has been removed — the pve-kernel-5.15 meta-package no longer exists.

Debian 13 “Trixie”

The new base brings updated glibc, systemd, and toolchain. Most user packages remain compatible, but check LXC containers with very old systems.

ZFS 2.3

ZFS has been updated to version 2.3, which brings:

Support for draid (distributed RAID) — more efficient allocation than traditional raidz
Compression improvements with Zstandard
Enhanced zfs_autoimport

Ceph Squid (19.2)

For those using Ceph, the version has been updated from Reef (18.2) to Squid (19.2). The Ceph migration must be done separately — refer to the official Ceph Squid guide.

Corosync 3.1

Corosync has been updated to 3.1.x. Configurations in 2.x format still work, but with warnings. PVE 9 recommends migrating to the new YAML configuration format.

Common pitfalls and solutions

Problem	Symptom	Solution
Repository 404	`apt update` fails	Check that sources.list use `bookworm` (not `trixie`) for PVE repositories
Kernel not updating	`uname -r` shows kernel 6.8	`apt install pve-kernel-6.14` and `proxmox-boot-tool refresh`
Corosync offline	`pvecm status` shows node offline	Edit `/etc/corosync/corosync.conf` to YAML format for corosync 3.x
ZFS not mounting	`zpool import` fails	`zpool import -a` then `zfs mount -a`
Network drops after reboot	No connectivity	Check `/etc/network/interfaces` — if using classic `ifupdown`, reinstall
VM fails to start	Kernel panic in guest	Convert IDE/SCSI disk controller to VirtIO SCSI single
PBS disconnects	Client error	`apt install proxmox-backup-client` for compatible version
HA resources stop	HA services shut down	`systemctl restart pve-ha-manager pve-ha-crm`

Cluster migration

For clusters with multiple nodes, the recommended procedure is:

Remove HA from all VMs/CTs on the node to be migrated
Migrate the VMs/CTs to other nodes
Upgrade the node (steps 1-4 above)
Verify the node has returned to the cluster (pvecm status)
Migrate the VMs/CTs back (optional)
Repeat for each node

Never try to upgrade all nodes simultaneously — you may lose cluster quorum.

Final considerations

If you encounter errors during the upgrade, the official Proxmox forum and the upgrade documentation are the best sources of reference.