Proxmox VE is a Debian system with its own kernel and packages. Most broken upgrades come from one of four mistakes below.
1. Use dist-upgrade / full-upgrade, not apt upgrade
The Proxmox VE admin guide says to update with:
apt update
apt dist-upgrade # same as 'apt full-upgrade'Plain apt upgrade won’t add or remove packages, so it can leave Proxmox’s dependencies half-upgraded. The web UI’s Updates → Upgrade button runs the right command for you.
2. Have exactly one sensible repository
Without a subscription, the pve-enterprise repo returns 401 errors on apt update. Disable it and enable pve-no-subscription, as documented on Package Repositories. Never enable pvetest on a box you care about.
3. Back up guests first
Run a backup (or at least snapshots) of important VMs and containers before upgrading the host. See Backup and Restore. A host upgrade rarely touches guest disks, but a failed boot is much less scary when the backups sit on another machine.
4. Reboot for new kernels, and do clusters one node at a time
A new proxmox-kernel only takes effect after a reboot. In a cluster, migrate or shut down guests, upgrade one node, reboot, and check it rejoins quorum (pvecm status) before moving to the next.
Major versions are different
Major upgrades (e.g. 8 → 9) have their own wiki guide and a checker script (pve8to9). Read the whole guide and run the checker until it’s clean. Don’t treat a major upgrade like a routine dist-upgrade.
Checklist
- Correct repo enabled (no-subscription or enterprise, not both)
- Guests backed up
-
apt update && apt dist-upgrade - Reboot if a new kernel was installed
- Cluster: one node at a time, quorum verified