This episode guides you through integrating PBS into the Proxmox VE datacenter: adding PBS as storage via Datacenter → Storage with the server, datastore, and certificate fingerprint. You will also choose the backup mode (guest agent, snapshot, suspend, stop), schedule regular backups from PVE, and verify that snapshots appear in PBS.

Your PBS server is up from episode 3 — now we connect it to the real world: Proxmox VE. Episode 4 is the most satisfying moment of the series: your VMs and CTs start being protected by deduplicated, encrypted backups. It is also the first episode where two tools work together — PVE as the orchestrator, PBS as the storage engine.
Imagine PVE like an office that arranges all document deliveries, and PBS like a central archive warehouse. You simply give the office the warehouse address (server + datastore + fingerprint), and the office handles the schedule and delivery contents.
In the PVE web UI: Datacenter → Storage → Add → Proxmox Backup Server. Fill in:
pbs1).backup@pam).store1).proxmox-backup-manager cert info on PBS, or from the web UI login page).pvesm add pbs pbs1 \
--server 10.0.1.10 \
--datastore store1 \
--username backup@pam \
--password 'secret123' \
--fingerprint AA:BB:CC:DD:...pvesm add pbs registers a PBS-type storage. Verify with pvesm status:
pvesm statusStatus active means PVE successfully connected to PBS. Keep the fingerprint somewhere safe — it serves to verify identity (anti man-in-the-middle) on every connection.
Important
The fingerprint must be exactly right (format AA:BB:...). PVE refuses the connection if the fingerprint does not match. If it is wrong, PVE will show a "fingerprint mismatch" error — how to fix it is covered in episode 16.
Once the storage is active, you can select it as the target in two places:
The target storage is filled with the ID you created earlier (e.g. pbs1). PBS also offers extra options at the job level: compression (zstd default), mode, notification, and prune-backups (retention — episode 8).
The mode determines how the VM disk is read during backup:
qemu-guest-agent is installed in the VM, PBS uses fstrim and requests a filesystem freeze via the agent so the snapshot is consistent from inside the OS, and empty blocks can be thined. Enable it in PVE: VM → Options → QEMU Guest Agent → Enabled.From VM → Backup, or via the vzdump CLI:
vzdump 100 --storage pbs1 --mode snapshot --compress zstdvzdump is the PVE backup engine; when the target is PBS, it calls proxmox-backup-client behind the scenes. For CTs, the command is the same — LXC is supported natively.
Create a scheduled job in Datacenter → Backup → Add, set:
pbs1.02:30 (0 2 * * *) and weekly.keep-last 7, keep-daily 14 — details in episode 8.This job will run automatically and report status in the PVE backup column.
After the backup finishes, check from the PBS side that the snapshot really exists:
proxmox-backup-client snapshot list \
--repository backup@pbs@10.0.1.10:store1Or from the PBS web UI: navigate to Datastore → store1, and see the backup group per VM/CT with the latest snapshot timestamp. Status "OK" and the original vs dedup sizes will be shown here.
Key takeaways:
pvesm add pbs.vzdump runs the backup; a PBS target calls proxmox-backup-client internally.proxmox-backup-client snapshot list.In the next episode, episode 5, we will restore — full VM restore to the same or a different PVE, plus file-level recovery by mounting a snapshot to grab individual files without a full restore. This is the capability you will rely on most when disaster actually strikes!