Outcome
You will create a guest backup schedule with a storage destination, retention policy, and owner. You will decide whether the built-in Proxmox VE backup feature is enough or whether a separate Proxmox Backup Server (PBS) is worth operating. A scheduled job is not proof of recoverability; guide 16 tests that.
Before you start
List each VM and container, its importance, how often its data changes, and how much data you can lose. Inventory a backup destination independent of the Proxmox host and its guest-disk pool. Confirm capacity, access, and who will notice job failures. Safeguard encryption keys and backup credentials outside the host; losing them may make backups unusable. Choose one small, disposable VM for the initial test.
Steps
- Choose a recovery target. Write a recovery point target (acceptable data loss) and a recovery time target (acceptable outage) for each service. A nightly backup cannot promise less than a day's possible data loss. The target helps set job frequency and offsite copy needs; do not copy a generic schedule without considering your data.
- Select a destination. Built-in Proxmox VE backups can write guest archives to configured backup-capable storage. PBS is a separate product and datastore that supports deduplicated backups, verification, pruning, and other backup operations. PBS adds another server and its own storage, updates, access control, and recovery work. Start with a separate backup-capable destination if that covers your small lab; consider PBS when its features justify the added operations.
- Configure the store. In Datacenter → Storage, add the supported destination using its documented backend. Verify it accepts VZDump backup file or, for PBS, that the PBS store is reachable and its credentials are scoped. Protect the storage from routine guest access. Do not put the only backup on the same physical disk as the guests.
- Create a job. In Datacenter → Backup, add a job selecting the test guest, destination, schedule, and backup mode appropriate to the workload. Proxmox's backup modes have different consistency and interruption characteristics; review them before backing up a live database. Set retention so old archives do not silently fill the destination.
- Run one backup now. Rather than waiting for the first schedule, start a manual job for the disposable VM. Read the task log through completion, inspect the resulting archive or snapshot entry, and note duration and size. Then confirm the recurring job is enabled and covers the intended guests, not an accidental empty set.
- Monitor and maintain. Put failed jobs, space usage, and backup age on a recurring review checklist. For PBS, configure and review datastore maintenance such as verification and garbage collection using PBS documentation. Save the restore procedure and required credentials separately from the host.
Check it worked
The backup task completed without errors, the independent destination shows the artifact, a future schedule is visible, retention is explicit, and someone is responsible for reviewing failures. Recovery remains unverified until guide 16.
If it fails or rollback
Read the task log for network, permission, free-space, or guest errors. Correct the specific issue and rerun a small backup. If a destination fills, do not delete the newest or only known-good copy without confirming another restore path. Disable a bad recurring job while repairing it; record the resulting protection gap.
Safety and data notes
Backup archives can contain credentials and personal data. Restrict access and document retention. Never claim a backup works because an archive exists; restoration is the meaningful test.
Sources
Official Proxmox documentation checked 2026-09-29: VE backup jobs, modes, retention, restore, storage backends and PBS integration, PBS documentation.
Next guide
Continue with guide 16 and restore the test guest to a separate VM ID.