Skip to content
xaweho

Knowledge base · basic

Running your first backup to the PBS — step by step

A manual test backup of a single VM to check that the datastore is connected correctly and everything works.

basic ·

Before you set up an automatic backup job, it’s worth running a manual test backup of a single VM. That way you see the connection works, you can measure the transfer speed, and you check that the backup actually arrives in the PBS.

Test backup via the PVE GUI

  1. In the PVE tree, select the VM or container you want to back up. Ideally a small VM for testing (e.g. an LXC with 5–10 GB).
  2. Click Backup in the VM’s/container’s left submenu.
  3. Click Backup now.

In the dialog:

FieldValue
StorageYour PBS storage (e.g. xaweho-pbs)
Modesnapshot (default, runs in the background) — possibly suspend for databases
Compressionzstd (default, good balance of speed and compression)
Notification modeauto

Click Backup. A task window opens.

What happens during the backup

The task window shows you the output live. These lines matter:

  • INFO: starting new backup job — the backup has started
  • INFO: Starting Backup of VM 100 (...) — VM ID and name
  • INFO: creating Proxmox Backup Server archive 'vm/100/...' — the backup snapshot ID
  • INFO: backup is sparse: ... bytes (XX%) total zero data — the share of data that doesn’t need to be transferred (zeroes)
  • INFO: read X% (... of ...) — progress of the current device
  • INFO: backup was done incrementally, ... GB (XX%) of ... GB total — at the end: how much was actually transferred
  • INFO: Finished Backup of VM 100 (... ) — done

The first backup transfers everything — it’s a full backup. For a 100 GB VM, that can take 30 minutes to several hours, depending on your internet connection. With a symmetric 1 Gbit line, about 15 minutes.

Subsequent backups are incremental: PBS recognizes unchanged blocks and only transfers the difference. After the first backup, a 100 GB VM typically produces 1–2 GB of transfer per day.

Test backup via the CLI

vzdump 100 --storage xaweho-pbs --mode snapshot --compress zstd

100 is the VM ID. The command produces the same output as the GUI.

Checking success

After it finishes:

  1. In PVE: VM → Backup → list, the new backup is at the top with date/time.
  2. In PBS: Datastore → Content → under the VM ID you see the snapshot with date/time. The snapshot is clickable and expandable.

If you click the snapshot, you see the disk files with size and index status. Here you can also extract individual files directly — see single-file restore.

What you should do next

Frequently asked questions

How long does the first backup take? Depends on VM size and connection. A 100 GB VM on a symmetric 1 Gbit line takes about 15–30 minutes. With 50 Mbit home internet, more like 4–6 hours. Subsequent backups are incremental and often only take a few minutes.

My backup aborted — what to do? Check the task logs (Tasks → last failed task). Common causes: connection dropout (internet), full disk on the PVE host (for temporary data), fingerprint mismatch. You have to restart an aborted backup — PBS doesn’t resume mid-run, and an aborted run leaves no snapshot. Only once the first full backup has completed once are the follow-up runs incremental and correspondingly fast.

What if the VM is running during the backup? With mode: snapshot (the default), PVE takes a live snapshot — the VM keeps running, and the backup is consistent as of the snapshot moment. Works well for most workloads. For active databases, we still recommend regularly exporting the DB into a dump as well — snapshots are crash-consistent, not necessarily transactionally consistent.

How do I see how much storage the backup really uses? In PBS under Datastore → Content. Per snapshot, the “Size” is the uncompressed data volume. The storage actually used is usually 30–60% of that — thanks to deduplication and compression. More in Understanding storage usage.

Can I back up multiple VMs in parallel? Yes. If you select multiple VMs (via Datacenter → Backup as a job, not via individual VMs), a backup job runs through all VMs sequentially. PBS handles up to 8 parallel write operations — that matters with multiple PVE hosts; with a single host everything runs serially through the bandwidth anyway.

What’s next

Related products
Tags
pbs backup erste-schritte

Did this article help?

If not, open a ticket. If it did, we're happy about a referral — both sides get €25 credit on their customer account.