Updating Proxmox VE (8 to 9, without a subscription)

Can I update Proxmox VE without a subscription?

Yes. Per the wiki, the enterprise repository needs a valid subscription key and the no-subscription repository does not: “you do not need a subscription key to access this repository”. The vendor names it for testing and non-production use and advises against it on production servers, because the packages are “not always as heavily tested and validated”.

To add the repository, see Is Proxmox VE free?. For what the tiers cost, see Proxmox subscription: tiers, cost, how to buy.

How do I install regular updates?

Use the web interface (Node → Updates) or the command line:

apt update
apt full-upgrade

Use apt full-upgrade, not apt upgrade. Proxmox VE is a rolling release on Debian, and per the admin guide apt upgrade may leave a partially upgraded or broken package state.

Do I have to move from version 8 to 9?

The support table in the FAQ gives August 2026 as the end of support for Proxmox VE 8 (Debian 12 “Bookworm”). That date has passed. Proxmox VE 9 is based on Debian 13 (“Trixie”).

The FAQ supports major upgrades, but “must be carefully planned and tested and should never be started without having a current backup ready”. The wiki offers two ways: an in-place upgrade via apt, or a new installation on new hardware.

For a new installation, you back up all VMs, containers and /etc, install Proxmox VE 9 from the ISO (this deletes all data on the host) and restore the VMs from the backups.

What has to be in place before an in-place upgrade?

The wiki lists these prerequisites:

  • All nodes run the latest Proxmox VE 8.4.
  • With hyper-converged Ceph, you upgrade to Ceph 19.2 Squid first.
  • Reliable access to the node, ideally through IPMI/IKVM or physically. With SSH only, test the upgrade first on an identical machine without production load.
  • A terminal multiplexer such as tmux or screen, so the upgrade carries on if the SSH connection drops.
  • A healthy cluster and a valid, tested backup of all VMs and containers.
  • At least 5 GB free on the root file system, ideally more than 10 GB.

How do I back up first?

According to the admin guide, Proxmox VE backups are always full backups, started in the web interface or with vzdump. A backup storage has to be defined first. The guide recommends a Proxmox Backup Server on a dedicated host, with an NFS server as a good alternative. More in Backing up and restoring Proxmox VMs and Installing and setting up Proxmox Backup Server.

How does the upgrade work, step by step?

The steps apply to each node individually. Per the wiki, work from the console or SSH, not from the web interface’s virtual console, which gets cut off during the upgrade.

What does pve8to9 check?

The script ships with the latest 8.4 packages:

pve8to9 --full

--full switches on all checks; run it at least once before the upgrade. The script only reports and fixes nothing, and it does not recognise every problem of a customised installation. Run it again after each fix.

Migrate guests that need to keep running away from the node first.

How do I bring the node up to date first?

Afterwards pveversion should report at least 8.4.1, and with Ceph ceph --version has to show Squid:

apt update
apt dist-upgrade
pveversion

How do I switch the repositories to Trixie?

The wiki first replaces bookworm with trixie, in sources.list and in the enterprise file:

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

Without a subscription, you create the Proxmox VE 9 repository in deb822 style, as the file proxmox.sources in /etc/apt/sources.list.d/:

Types: deb
URIs: http://download.proxmox.com/debian/pve
Suites: trixie
Components: pve-no-subscription
Signed-By: /usr/share/keyrings/proxmox-archive-keyring.gpg

With a subscription, pve-enterprise.sources has this instead:

Types: deb
URIs: https://enterprise.proxmox.com/debian/pve
Suites: trixie
Components: pve-enterprise
Signed-By: /usr/share/keyrings/proxmox-archive-keyring.gpg

How do I check the new repositories?

Check with apt update and apt policy that no errors appear, then remove the old Proxmox VE 8 entry from sources.list, pve-install-repo.list or any other .list file.

And the Ceph repository?

With hyper-converged Ceph, the wiki replaces ceph.com repositories with the proxmox.com ones. Without a subscription, in ceph.sources:

Types: deb
URIs: http://download.proxmox.com/debian/ceph-squid
Suites: trixie
Components: no-subscription
Signed-By: /usr/share/keyrings/proxmox-archive-keyring.gpg

Check with apt update and then apt policy: there must be no errors, and the new repository must show up in the output of apt policy. Then remove the old file ceph.list in /etc/apt/sources.list.d/.

Remove a backports line, because the upgrade has not been tested with it.

How do I start the upgrade, and what comes after?

apt update
apt dist-upgrade

According to the wiki this takes up to 60 minutes or more on a slow disk, under 5 minutes with SSD. For service restarts, take the default when in doubt, because the reboot afterwards restarts all services anyway. Configuration files are different: the wiki advises looking at the difference for each file, and for common files it recommends:

  • /etc/lvm/lvm.conf, /etc/chrony/chrony.conf: with no changes of your own and in doubt, the package maintainer’s version (“Yes”).
  • /etc/issue, /etc/default/grub: when in doubt, keep the installed version (“No”); for grub, look at the difference first.

If apt dist-upgrade finishes without errors, run pve8to9 once more and reboot the node. That applies even if you already run the 6.14 kernel on version 8.

What applies in a cluster and with Ceph?

In a cluster, check after each node that all nodes are up and on the latest packages. If not, continue with the next node and start over at the prerequisites.

What applies with Ceph?

Two pitfalls from the wiki:

  • “insecure key types” message: Ceph Squid from 19.2.6 sets the cluster status to HEALTH_ERR because of it, but the services keep running. If no other warning appears, continue the upgrade and migrate the keys once all nodes run version 9.
  • Ceph full-mesh network: Older guides contain a post-up line that restarts frr. It deadlocks on restart with Proxmox VE 9 and has to be changed beforehand.

More on Ceph is in Building a Proxmox cluster with Ceph.

Which known issues does the wiki list?

  • Containers with an old systemd: Proxmox VE 9 no longer supports cgroup v1. Containers running systemd 230 or older, such as CentOS 7 or Ubuntu 16.04, are not supported.
  • Network cards: The new kernel may rename interfaces. Proxmox VE 9 has pve-network-interface-pinning.
  • /etc/sysctl.conf: It is no longer read. Values belong in a file under /etc/sysctl.d/.

The full list is under “Known Upgrade Issues” in the wiki. If the upgrade stops after a network failure, try apt -f install.

Nexilon plans, installs and maintains Proxmox environments, is a certified Proxmox Silver Partner and supplies subscriptions.

Sources

Let's talk

Free initial consultation - no sales pitch, just an honest assessment.

Request a consultation →