Scheduled maintenance is currently in progress. We will provide updates as necessary.
Posted Jan 26, 2026 - 10:30 UTC
Scheduled
We are making some improvements to the way that we take and manage backups of projects across the fleet. We introduced this new method some time ago, and it's already in use for larger projects, projects with PITR enabled, and projects with Read Replicas. We also made the new method the default for all new projects last week.
We are now pushing this change out to all remaining projects across the fleet in stages through the week of January 26, 2026. We will start with Free plan projects, and move through paid subscribers after.
What's not changing?
You'll still get daily backups as you do today — this is an infrastructure improvement on our end that benefits you automatically.
What this means for you:
Faster restores — If you ever need to restore your database, it'll be back online significantly faster, especially for larger databases.
Less impact on your database — Backups will run with lower overhead, so you won't notice any performance impact during the backup window.
Is there any downtime?
Transitioning to the new backup method does involve a Postgres restart. The downtime is extremely minimal, just a few seconds in most circumstances. Any direct or pooler connections will disconnect on restart. As a best practice for using any hosted database, we recommend including reconnect logic in our application to re-establish connections if they're severed for any reason.