Skip to main content
Infrastructure

Proxmox VE 8 Is Past End of Life: Upgrading a Small Office to Proxmox VE 9

What the end of support means for every guest on an 8.x host, the checks to pass before you touch the repositories, and how to plan the window, the cluster order and the way back.

By William Bradshaw | November 16, 2026 | 10 min read

The version table in the Proxmox FAQ lists end of life for Proxmox VE 8 in August 2026, a month after Debian 12 "Bookworm", the release it is built on. A host still running 8.x does not stop working. It just stops changing: no more fixes for the kernel, QEMU or the management stack underneath every virtual machine and container it runs, while new vulnerabilities keep being published against them.

This article is for IT leads at 10 to 50 seat offices that moved off VMware onto Proxmox VE 8 and have applied minor updates since, and for Ohio townships and fire districts whose cybersecurity program has to show that systems run on supported software. It covers what end of life means in practice, the choice between upgrading in place and reinstalling, the checks to pass first, the repository change, the order for a cluster, what behaves differently afterward, and what to write down.

It follows the official Proxmox upgrade guide and is not a replacement for it: read the current version of that page before you start, because its known-issues list changes. If you do not yet have an offsite copy of your guests, fix that first with offsite backups for Proxmox. Every route below assumes a backup you have restored at least once.

What End of Life Means for the Guests

The hypervisor is the one layer every guest trusts. A Windows server can be patched to the day and still run on a host kernel and a QEMU build that will never be fixed again. That is the practical meaning of end of life: the gap between what is known about the platform and what is patched on it only grows.

  • Debian LTS does not cover it. Debian's volunteer long-term support effort extends part of Debian 12, but Proxmox builds its own kernel, QEMU and management packages, and those come only from the Proxmox repositories.
  • Frameworks ask for supported software. CIS Controls v8 safeguard 2.2 asks that authorized software be currently supported, and that is the kind of baseline an ORC 9.64 cybersecurity program is measured against.
  • Insurers ask too. Cyber insurance questionnaires ask whether any end-of-life or unsupported systems remain. The questionnaire walkthrough explains how to disclose one with a dated migration plan if the upgrade cannot happen before your renewal.

Check where you stand with pveversion on each host. Anything that starts with 8 is the subject of this article. Anything that starts with 7 is further behind: Proxmox publishes one upgrade guide per major version, so a 7.x host goes to the latest 8.4 first.

Two Routes: Upgrade in Place, or Reinstall and Restore

The Proxmox upgrade guide describes two ways to get from 8.x to 9.x: a new installation on new hardware with the guests restored from backup, or an in-place upgrade through the package manager. Both are legitimate. The right one depends on the age of the hardware and how much you trust your restores.

Question Upgrade in place Reinstall and restore
Host configurationKept: network, storage, users, jobsRebuilt by hand, so document it first
Guest downtimeThe package run and a reboot per hostEach guest's restore time
What it provesThat the checks were rightThat every backup restores
Best fitClusters, and hardware with years leftSingle hosts on aging hardware due for replacement
The way backRestore from backup onto a reinstalled hostThe old host, untouched, until you retire it

The upgrade guide also notes that hardware more than ten years old is less tested. If the host is already on your replacement list, a new server installed with Proxmox VE 9 and the guests restored from Proxmox Backup Server turns an upgrade project into a hardware refresh and a full restore test in one window. Size the new host with the Proxmox host sizing guide.

Before You Touch the Repositories

These come from the prerequisites in the Proxmox guide. Most failed upgrades trace back to one of them being skipped.

  • Latest 8.4 on every node. Run apt update and apt dist-upgrade on the 8.x repositories first, then confirm pveversion reports at least 8.4.1.
  • A valid, tested backup of every VM and container. Tested means restored, not just listed as successful.
  • Disk space. At least 5 GB available on the root file system, ideally more than 10 GB. Check with df -h /.
  • A console that does not depend on the host's own network. IPMI or iKVM, or a monitor and keyboard. If SSH is all you have, the guide says to test on an identical non-production machine first and to run the upgrade inside tmux or screen so a dropped session does not kill it.
  • A healthy cluster, if you run one: every node online and quorate in pvecm status.
  • Ceph first. A hyperconverged Ceph cluster must be on Ceph 19.2 Squid before the Proxmox upgrade starts.
  • Proxmox Backup Server on the same host needs its own 3 to 4 upgrade, following the PBS guide alongside this one.
  • Old containers. Proxmox VE 9 drops cgroup v1, and containers running systemd 230 or older are not supported. CentOS 7 (systemd 219) is the usual case. Migrate or rebuild them first.
  • Third-party tools. The guide says to wait for third-party storage plugins to support 9.0. At the time of writing it also lists Veeam backups as broken for VMs on QEMU machine version 10.0 or later, with pinning the machine version as the workaround. Check your backup vendor's notes before the window, not during it.

Run pve8to9, Fix, Run It Again

Proxmox ships a checklist script for this upgrade. It reports problems and changes nothing:

pve8to9 --full

Run the full version at least once, fix what it flags, and run it again until it is clean apart from warnings you understand. Two findings worth expecting on older installs: the systemd-boot meta-package, which the guide says to remove when the checker suggests it unless you use systemd-boot deliberately, and UEFI systems with the root file system on LVM, which need grub-efi-amd64 installed.

Save the output. The first clean run is the opening page of your change record, and you will run the script once more after the upgrade.

Switch the Repositories From Bookworm to Trixie

Proxmox VE 9 is built on Debian 13 "Trixie". The Debian entries change by name:

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

The Proxmox repositories also move to the deb822 .sources format, which the guide recommends. For a host on the no-subscription repository, the guide's /etc/apt/sources.list.d/proxmox.sources reads:

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

A subscribed host uses a pve-enterprise.sources file instead, with the enterprise URI and the pve-enterprise component. Ceph gets a ceph.sources file for Squid. Then remove the old Proxmox and Ceph .list entries, and any backports line, since the guide says the upgrade is untested with backports.

Trixie's apt can convert existing lists for you with apt modernize-sources. Answer "n" at its prompt to preview the change, then run it again to apply; it keeps the old files with a .bak suffix.

Finish with apt update and apt policy. Every source should name trixie, and nothing should still point at bookworm. A 401 error on the enterprise repository usually means the subscription needs refreshing (the guide gives pvesubscription update --force).

The Upgrade Itself

apt dist-upgrade

The guide puts the run at under five minutes on a server with SSDs and an hour or more on slow spinning disks. Stay at the console: apt will stop to ask about configuration files you have changed. The guide's advice, in short:

  • /etc/issue: keep your current version.
  • /etc/lvm/lvm.conf and /etc/chrony/chrony.conf: take the maintainer's version if you made no custom changes.
  • /etc/ssh/sshd_config: the maintainer's version is recommended if you have no other changes. If you hardened SSH, compare before you accept anything (the Proxmox hardening checklist lists what to re-check).
  • /etc/default/grub: keep your current version if unsure.

When apt finishes, run pve8to9 once more, then reboot into the new kernel. Confirm with pveversion, force-reload the web interface so the browser drops cached scripts, and start the guests in the order their owners need them.

Clusters: One Node at a Time

A cluster lets you upgrade without a full outage, as long as the order respects one rule from the guide: migrating a guest from an older Proxmox VE to a newer one always works, while the other direction may work but is not supported.

  1. Move every guest off the first node to its partner, which is still on 8.4. For HA-managed guests, ha-manager crm-command node-maintenance enable NODENAME drains them for you.
  2. Upgrade the empty node, reboot it, and confirm it rejoins: pvecm status, and the QDevice still counted if you run one.
  3. Move the guests from the next node onto the upgraded one (older to newer, the supported direction), then upgrade that node. Repeat until every node runs 9.

While versions are mixed, the guide warns against managing 8.x nodes from a web session on an upgraded node; if something errors, retry from the 8.x node's own interface. It also warns that live migration between hosts with different CPUs can leave a VM unresponsive, so test with a non-production VM first. On Ceph clusters, a health warning about insecure key types can appear after the first upgraded monitor node reboots; the guide says to carry on if nothing else is wrong and to migrate the keys once every node is upgraded.

HA groups convert to HA rules automatically once every node runs 9. The Proxmox high availability guide explains the rules model and the two-node QDevice design. The QDevice machine itself is not part of this upgrade; it is a separate Debian host and keeps its own patch schedule.

What Behaves Differently Afterward

  • /etc/sysctl.conf is no longer read. Any kernel setting you put there (forwarding, rp_filter, connection tracking limits) needs to move to a file such as /etc/sysctl.d/90-local.conf. Check what you have before the upgrade, while it still applies.
  • Network interface names can change with the new kernel, which leaves a bridge such as vmbr0 pointing at a port that no longer exists, and a host you can only reach from the console. The guide names a pve-network-interface-pinning tool that pins interfaces to stable nicX names. The bridge and bond layout those names feed is in the Proxmox networking guide.
  • VM memory looks higher. Host-side accounting now includes each VM's overhead. It is a reporting change, not a leak, but it will move your monitoring baselines.
  • HA groups are now HA rules, as above. Tutorials written for version 8 describe the old interface.

Proxmox VE 9.2 is the current release. The Proxmox VE 9.2 release article covers what arrived after 9.0, including the dynamic load balancer for HA clusters.

Plan the Window and the Way Back

There is no supported downgrade from Debian 13 to Debian 12. The way back from a failed in-place upgrade is your backups: reinstall the host and restore the guests. That is why the tested backup sits at the top of the prerequisites and not in a footnote.

  • Single host: every guest is down for the package run, the reboot and the checks. Book an evening or weekend window, tell users in advance, and name who confirms each system works afterward.
  • Cluster: guests move instead of stopping, but each migration is a short interruption for that guest, and the cluster carries the full workload on fewer nodes for the length of the window.
  • Either: keep the console session, the backup job IDs and the restore steps in front of you, not in an email you cannot reach when the host is down.

The change record

One page is enough, and it is the evidence an auditor, an insurer or the next administrator will ask for:

  1. Date, window, and who did the work.
  2. pveversion -v before and after, for each node.
  3. The pve8to9 --full output before, and the run after.
  4. The backup for each guest that was verified, and when it was last restored.
  5. Each guest started and checked, and by whom.
  6. What went wrong, what you changed, and anything left to follow up.

The FAQ lists the end-of-life date for Proxmox VE 9 as still to be announced. Put a yearly reminder in the calendar to check that table, so the next major upgrade is a planned project rather than a discovery.

Frequently Asked Questions

When did Proxmox VE 8 reach end of life?

The version table in the Proxmox FAQ lists end of life for Proxmox VE 8 in August 2026, one month after Debian 12, the release it is built on, in July 2026. A host still on 8.x keeps running, but the 8.x series no longer receives updates.

Can I upgrade straight from Proxmox VE 7 to Proxmox VE 9?

No. Proxmox publishes one upgrade guide per major version, and the 8 to 9 guide starts from the latest Proxmox VE 8.4. A 7.x host follows the 7 to 8 guide first, then the 8 to 9 guide.

How long does the Proxmox VE 8 to 9 upgrade take?

The package upgrade itself can finish in under five minutes on a server with SSDs and take an hour or more on slow spinning disks, according to the Proxmox upgrade guide. Add the reboot, the checks afterward and the time to move guests between cluster nodes, and plan a maintenance window rather than a lunch break.

Will my old Linux containers still start on Proxmox VE 9?

Not if they are very old. Proxmox VE 9 drops cgroup v1, and the upgrade guide says containers running systemd 230 or older are not supported. CentOS 7, with systemd 219, is the common example. Migrate or rebuild those containers before you upgrade the host.

Should I upgrade in place or reinstall Proxmox VE 9 and restore?

Proxmox documents both. An in-place upgrade keeps the host configuration and is usually faster. A fresh install on new hardware with guests restored from backup suits older hardware, and every restore doubles as a backup test. Single-host offices with aging servers often do best with the second route.

Still Running Proxmox VE 8?

Bullium plans and carries out Proxmox upgrades for small offices, townships and fire districts: the pre-upgrade checks, a verified restore, the maintenance window, the cluster order, and a change record you can file. We start with a walkthrough of what you run today and which route fits it. No commitment to engage further.