Check your backups and restore on Dynamic.

Your window into the datastore

On Dynamic there's no PBS web interface — and you don't need one. Everything stored on your plan is visible in Proxmox VE: select the PBS storage on any node and open Backups. Every snapshot is listed per guest with its timestamp, size, and notes. If it's in that list, it's on our servers, offsite. This one view is where checking and restoring both start.

Checking: make failures loud

Three habits catch nearly every backup problem while it's still boring:

• Email on failure. Edit your backup job (Datacenter → Backup) and set a notification address. A failed backup that emails you is an annoyance; one that doesn't is a future incident.
• Scan the task log after changes. Task History on the node or datacenter shows every backup run. Green TASK OK is the goal; warnings about skipped guests deserve a look even when the job "succeeds."
• Look for missing guests, not just errors. Once a month, compare the Backups view against your guest list. The failure mode that hurts most is the VM that quietly dropped out of the job three months ago. A fresh snapshot for every guest that matters — that's the check.

Restoring a whole VM or container

From the storage's Backups view, select the snapshot you want and click Restore:

• Target storage — where the restored disks land on your side.
• VM ID — restoring over the existing ID replaces that guest. To restore alongside it (usually the safer first move), pick an unused ID.
• Start after restore — for VMs, PVE can boot the guest while the rest of the data streams in the background (live restore), which turns "wait for the full download" into "minutes to first boot."

Then validate: boot it, log in, check the application — not just the hypervisor's green checkmark. Note how long it took; that number is your real recovery time, and your boss will ask for it someday.

Restoring single files

Deleted one config file, not the whole server? Select the snapshot and click File Restore. Proxmox VE mounts the backup read-only and lets you browse the guest's filesystem in the browser and download exactly the files or directories you need — no full restore, nothing overwritten, works for VMs and containers alike.

Practice one restore per quarter

Pick one guest, restore it to a spare VM ID, boot it, check the app, delete it. Fifteen minutes, four times a year — and the day something real breaks, your restore is a routine instead of a first attempt. Server-side, we run verification on the datastore so silent corruption gets caught on our watch; the quarterly drill proves the part only you can prove — that your guests come back and run.