Skip to content
uploy docs

Download a backup

Goal: Pull a nightly .sql.gz backup of your database to your local machine, for example to restore into a local Postgres for inspection or to archive off-platform.

Prerequisites:

  • A Hobby-tier project with a managed Postgres database.
  • At least one nightly backup in the retention window (backups run daily; retention is 7 days on Hobby).

Download is Hobby-only. Dev / Pro / Business tiers use continuous physical backups, which aren’t a single-file artifact — use PITR for those.

Steps

Section titled “Steps”
  1. Open the project that owns the database and go to the Database tab.
  2. Scroll to Backups. Each row is one nightly backup with its date.
  3. Click Download on the row you want.
  4. Your browser receives a presigned S3 URL and starts the download automatically. The file name is the backup date, e.g. 2026-04-14.sql.gz.

The presigned URL is valid for 5 minutes. If your download fails or times out, click Download again to mint a new URL.

Verify

Section titled “Verify”

The downloaded file is a gzipped SQL dump produced by pg_dump. Restore it into a local Postgres to inspect:

Terminal window
gunzip -c 2026-04-14.sql.gz | psql -h localhost -U postgres -d scratch

A clean restore with no errors confirms the backup is valid.

Notes

Section titled “Notes”
  • Every download request is logged server-side with the user, backup, and database IDs.
  • Backups older than the retention window are garbage-collected and can no longer be downloaded.
  • The shared Hobby cluster is not exposed for direct client connections; the downloaded .sql.gz is the only way to obtain a point-in-time copy of Hobby data.

Next steps

Section titled “Next steps”