TL;DR: Proxmox VE 8 (Debian 12 Bookworm-based) EOL is August 2026. PVE 9 (Debian 13 Trixie-based) can be upgraded in-place via the pve8to9 tool, but plan for: ZFS pool compatibility, QEMU machine version, kernel versions, cluster member order, network config (ifupdown2), and guest VM compatibility. This guide covers pre-check, upgrade steps, and rollback scenarios. Important: Start at least 1 month early; in cluster environments, do node-by-node rolling upgrade, not all at once.

EOL timeline and why plan now

Version	Debian Base	EOL	Status (2026-06)
Proxmox VE 7	Debian 11 Bullseye	2024-07	EOL — unsupported
Proxmox VE 8	Debian 12 Bookworm	2026-08	~2 months left
Proxmox VE 9	Debian 13 Trixie	2028+ (estimated)	Recommended version

After EOL: no security patches, missing new hardware drivers, vendor support ends. Production environments must transition before August.

Step 1 — Pre-flight check

# Version check
pveversion
# Required: pve-manager/8.4.x or higher

# Pre-flight check (shows all warnings and errors)
pve8to9 --full

# Sample output:
# PASSED: 42
# WARNINGS: 2
# FAILURES: 0

If there are FAILURES, do not start the upgrade. Read each WARNING carefully; many require manual action.

Step 2 — Update APT sources (Bookworm → Trixie)

# Backup current sources
cp /etc/apt/sources.list /etc/apt/sources.list.bak

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

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

# Ceph if present:
sed -i 's/bookworm/trixie/g' /etc/apt/sources.list.d/ceph.list 2>/dev/null

Step 3 — Run dist-upgrade

apt update
apt --simulate dist-upgrade | tail -20
DEBIAN_FRONTEND=noninteractive apt -y dist-upgrade

Interactive prompts may appear (/etc/issue, /etc/lvm/lvm.conf). Default “keep current local version” is usually safe.

Step 4 — Reboot and verify

shutdown -r now

# After reboot
pveversion -v   # Expected: proxmox-ve/9.0.x
uname -r        # Expected: 6.8.x or 6.11.x
pvecm status    # All nodes quorate
pvesm status    # All datastores active
ip addr show    # All bridges up

Cluster rolling upgrade (important!)

1. Upgrade first node (typically non-master, backup node)
2. After first node is up, check cluster health: pvecm status
3. If HA enabled, wait for HA Manager rebalance (~5 min)
4. Move to second node; same process
5. Upgrade master/quorum master node last
6. After all nodes on PVE 9: pvesh get /cluster/status

Guest VM compatibility after upgrade

• Existing VMs stay on old QEMU machine version (e.g. i440fx 8.1) by default. To use new features, set VM Hardware → Machine → “Latest”; caution with Windows VMs (boot sector sensitive)
• Linux guests: update virtio-net driver (especially CentOS 7 / Ubuntu 18)
• Windows guests: update VirtIO drivers to virtio-win-0.1.248 or higher
• LXC containers: Debian 11 containers may be incompatible with new kernel; upgrade guest OS inside with apt dist-upgrade

Rollback scenario

• ZFS root: choose old boot environment via zfsbootmenu (snapshot required beforehand)
• LVM-thin: no rollback, fresh install required
• Cluster node down: move VMs to other nodes (qm move-disk), reinstall affected node
• VM-level rollback: PBS restore (best practice)

Common issues

• “pve8to9 ‘no-subscription warning’.” Just paid vs free repo notice; doesn’t block upgrade.
• “/etc/network/interfaces not working after boot.” ifupdown2 may be missing. apt install ifupdown2 + systemctl restart networking.
• “VM won’t start, ‘machine type not supported’.” Old machine version removed in PVE 9. Choose newer machine type (i440fx-8.x or q35-8.x) in VM hardware.
• “Ceph monitor down after upgrade.” Ceph has its own upgrade procedure; ceph status, systemctl restart ceph-mon@<id>.
• “GUI web console (8006) not opening.” systemctl status pveproxy. SSL may be corrupt: pvecm updatecerts --force.

Post-upgrade best practices

• First week: scan journalctl -p 3 -xb on each node for errors
• Monitor VM CPU/memory metrics for anomalies (Grafana/Prometheus)
• Verify backup jobs run: cat /var/log/vzdump/log
• Daily cluster health: pvecm status

Sources: Proxmox VE Roadmap (pve.proxmox.com/wiki), XDA Developers analysis. Tested environments: PVE 8.4.1 → 9.0 RC1 in-place upgrade, 3-node cluster + PBS. Last updated: 2026-06-15.