Appearance
Backups & Restore
Backups apply to PostgreSQL services — the only Infrly service kind that holds data you'd need to recover.
Automatic backups
Every PostgreSQL database gets a full backup when it's created and then once a day, with 7 days of retention by default. Change the retention window from the database's settings at any time — a longer window keeps more history but uses more backup storage.
There's nothing to turn on: the first backup starts as soon as the database is ready (it usually completes within a minute), and restores are available from then on.
Between full backups, changes are shipped to backup storage continuously, in small batches every few minutes. A restore can therefore reach any moment up to a few minutes ago; the last few minutes before you start a restore may not be included yet.
Restoring
Open a database's Backups tab to see available restore points, and choose either:
- Most recent — restores the latest recoverable state.
- A specific point in time — restores to any moment covered by your retention window, useful for recovering from "someone ran a bad migration/query an hour ago" rather than only being able to go back to last night's backup.
Restore creates a new database — it doesn't overwrite the original
Restoring never touches the source database. It provisions a brand-new PostgreSQL service, seeded from the backup, running alongside the original. This means a restore is always safe to try (the original is never at risk), but it also means you'll need to point your app's DATABASE_URL at the new database afterward — either by updating it directly or repointing a service reference at the new service's name.
The restored database is fully independent afterward — its own backup schedule, its own retention setting, its own lifecycle.
Project capacity
A restored database runs in the same project as the original. It starts as a single instance, so it only needs room for one copy of the plan's size; its standby is added right after the restore when the project has room. If there isn't room, the deployment log says so: once your app uses the restored database, delete the old one (or free other capacity) and redeploy the restored database to add its standby.