Backups That Restore: A Simple Monthly Drill

Updated: February 23, 2026

The uncomfortable truth

A backup you never restore is a hope, not a plan. Most backup failures are boring: a path changed, a volume moved, a permission broke, the disk filled up, or you backed up the wrong data. You only discover it when you need the data urgently.

Make it small and repeatable

You do not need to restore your entire server every month. You need a repeatable drill that proves the backup chain works end-to-end and that your notes are complete.

A good monthly restore drill

• Pick one service you care about (for example, a photo library or a notes app).
• Back up: its config + its data volume + any database dump it needs.
• Restore to a separate folder or a spare machine (not the live service).
• Verify one real workflow (login, open a file, search, etc.).
• Write down what you forgot and fix the checklist.

What to back up (typical home server)

• Reverse proxy config and certificates (or at least renewal automation notes).
• Docker Compose files / systemd unit files.
• App configuration directories.
• Database dumps (or raw volumes) with a documented restore command.
• A map of where secrets live (do not publish the secrets themselves).

Use the 3-2-1 idea without making it complicated

The classic rule is: 3 copies, 2 different media, 1 offsite. For a home server, “offsite” can be as simple as an encrypted backup to a different physical location or a cloud bucket you control. The details matter less than the habit of verifying restore.

Related: Guides: Backups