Skip to content

How to Upgrade Proxmox VE 8 to 9 Safely (8 Is Now EOL)

By Bikram Bhujel

If your homelab or office server still runs Proxmox VE 8, it's now out of support. Proxmox's own FAQ lists the end of life for version 8 as August 2026, which means no more security fixes. The good news is that the in-place upgrade to Proxmox VE 9 is well documented and, done carefully, boring.

This is the careful version for a single node or a small cluster: the checks to run first, the repository changes (Proxmox 9 switches to a new repository file format), the upgrade itself, and the gotchas from the official notes that are easy to miss. I've linked the official page throughout, because for an upgrade like this you want the source, not my memory.

Proxmox VE shell running the pve8to9 checklist before upgrading Proxmox VE 8 to Proxmox VE 9

Quick answer

Proxmox VE 8 reached end of life in August 2026. To upgrade in place to Proxmox VE 9 (Debian 13 Trixie): back up every VM and container, update to the latest 8.4, run pve8to9 --full and fix what it reports, change your apt repositories from bookworm to trixie (Proxmox 9 uses the new .sources format), then run apt update and apt dist-upgrade and reboot. Hyper-converged Ceph must be on 19.2 Squid first. Proxmox 9 drops cgroup v1, so very old containers with systemd 230 or older won't run.

Why upgrade now

VersionBased onReleasedEnd of life
Proxmox VE 8Debian 12 BookwormJune 2023August 2026
Proxmox VE 9Debian 13 TrixieAugust 2025Not announced yet

The dates come from the Proxmox VE FAQ. The current release is 9.2, announced on 21 May 2026 with Linux kernel 7.0 as the default. A hypervisor is the one machine in a homelab you really don't want running unpatched, because everything else sits on top of it.

Step 1: Back up, properly

The official upgrade guide is blunt about this: a valid and tested backup is always required. That means:

  • Back up every VM and container to storage that is not on this host (a Proxmox Backup Server, a NAS share, a USB disk).
  • Restore one of them somewhere as a test. A backup you've never restored is a guess.
  • Copy /etc, especially /etc/pve, /etc/network/interfaces, /etc/passwd and /etc/resolv.conf.
tar czf /root/pve-etc-$(hostname)-$(date +%F).tar.gz /etc
scp /root/pve-etc-*.tar.gz user@192.168.0.50:/backups/

Also check you have at least 5 GB free on the root filesystem, ideally more than 10 GB:

df -h /

Step 2: Get to the latest 8.4 and run the checker

apt update
apt dist-upgrade
pveversion

pveversion needs to show at least 8.4.1. Then run the checklist script. It only reports problems, it changes nothing:

pve8to9 --full

Read every WARN and FAIL. Fix them, then run it again until it's clean. The guide says to run it before, during and after the upgrade, and it's genuinely useful each time.

If you run Ceph on the same nodes, check ceph --version. It must be 19.2 Squid before you start. Older Ceph versions (Reef, Quincy) need upgrading first, on Proxmox 8.

Step 3: Things to sort out before the upgrade

These come from the known issues on the official page. Most homelabs hit one or two:

If you have...Do this
Old containers (systemd 230 or older, e.g. CentOS 7)Proxmox 9 has no cgroup v1 support, so they won't start. Upgrade or replace them first
Settings in /etc/sysctl.confMove them to a file in /etc/sysctl.d/. The old file is no longer read
Proxmox installed on top of plain DebianRemove linux-image-amd64 before the upgrade
Veeam backups of your VMsBackups fail on QEMU machine version 10.0+. Pin VMs to 9.2+pve1 or wait
NVIDIA vGPUUse GRID 18.3 (driver 570.158.02) or newer
UEFI boot with GRUB on LVMMake sure grub-efi-amd64 is installed

The guide also suggests turning off the kernel audit log socket first to avoid log noise:

systemctl disable --now systemd-journald-audit.socket

Step 4: Change the repositories to Trixie

Do this from a tmux or screen session, or better, from the physical console or IPMI. You don't want a dropped SSH connection halfway through a hypervisor upgrade.

First, switch the Debian sources from Bookworm to Trixie:

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

Proxmox 9 moves to the deb822 .sources format. For the free no-subscription repository, which is what most homelabs use, create /etc/apt/sources.list.d/proxmox.sources:

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

If you have an enterprise subscription, use pve-enterprise.sources with https://enterprise.proxmox.com/debian/pve and the pve-enterprise component instead. Ceph users need a matching ceph.sources pointing at ceph-squid for Trixie. The exact blocks for each are on the official page.

Then remove the old .list files for Proxmox and Ceph so you don't have both, and check what apt now sees:

ls /etc/apt/sources.list.d/
apt update
apt policy

Every Proxmox line in apt policy should say trixie. If anything still says bookworm, fix it before going further.

Step 5: Upgrade and reboot

apt dist-upgrade

This takes a while. You'll get questions about config files. If you changed a file yourself, keep your version. If you didn't, accepting the maintainer's version is usually right. When it finishes:

pve8to9
pveversion
reboot

Reboot even if you were already using the newer kernel on Proxmox 8. The guide is explicit about that. After the reboot, log in to the web UI (clear the browser cache or use a private window if it looks odd), check that every VM and container starts, and run pve8to9 one last time.

What can go wrong

  • The upgrade stopped halfway (network drop, power cut). Run apt -f install, then apt dist-upgrade again. This is why you use tmux or the console.
  • A container won't start. Check if it's an old distribution. The cgroup v1 removal is the usual cause.
  • Clusters with mixed CPUs. The guide warns that live migration between hosts with different CPU vendors or models can cause problems. Upgrade one node at a time, migrate guests off it first, and keep the cluster healthy between nodes.
  • Shared LVM storage. New logical volumes are created with autoactivation turned off. If pve8to9 suggests it, run /usr/share/pve-manager/migrations/pve-lvm-disable-autoactivation.

How to roll back

There's no supported downgrade from 9 to 8. Your rollback is a fresh Proxmox 8 install plus your /etc backup and guest backups, or a reinstall of Proxmox 9 with the guests restored. That's why Step 1 matters more than any other step here. If the host boots from ZFS, a snapshot of the root dataset before the upgrade gives you a much quicker way back.

A few things I learned

  • Run the checker until it's boring. pve8to9 --full catches most of the problems before they happen.
  • Use the console, not SSH over Wi-Fi. Of all the machines to lose halfway through an upgrade, the hypervisor is the worst.
  • Old containers are the real work. The Proxmox upgrade itself is quick. Replacing a CentOS 7 container that nobody remembers setting up is what takes the afternoon.

If you're installing fresh instead, see my guide to installing Proxmox VE 9.2 on ARM64. My TrueNAS home server post covers the storage side, and the full picture of my setup is in Homelab Intelligence. If you'd rather have someone upgrade a production cluster for you, that's what my IT services are for.

Frequently asked questions

When does Proxmox VE 8 reach end of life?

Proxmox's FAQ lists end of life for Proxmox VE 8 as August 2026. After that it gets no more security updates, so nodes still on version 8 should be upgraded to Proxmox VE 9, which is based on Debian 13 Trixie.

How do I upgrade Proxmox VE 8 to 9?

Back up all guests, update to the latest 8.4, run pve8to9 --full and fix its warnings, change the apt repositories from bookworm to trixie, run apt update and apt dist-upgrade, then reboot. Ceph must be on 19.2 Squid first.

What does pve8to9 do?

pve8to9 is Proxmox's upgrade checklist script. It checks the node for known problems before, during and after the upgrade from version 8 to 9 and reports them. It does not change anything, so it is safe to run as often as you like.

Why won't my LXC container start after upgrading to Proxmox 9?

Proxmox VE 9 removed support for the legacy cgroup v1 environment, so containers running systemd 230 or older, such as CentOS 7, no longer start. Upgrade the distribution inside the container or replace it with a newer template.

Can I downgrade Proxmox VE 9 back to 8?

No, there is no supported downgrade. To go back you would reinstall Proxmox VE 8 and restore your configuration and guest backups. That is why a tested backup of every VM and container is required before upgrading.

Comments

Questions and corrections are welcome. I read every comment before it appears, and remove spam and promotional links.