a month ago
Description: Verified pg_dump/mysqldump backups into your own S3 bucket. No state.
Category: Storage
1 Reply
a month ago
I published a template for scheduled database backups: https://railway.com/deploy/dumpling-backup
It runs pg_dump or mysqldump on a cron inside your project and talks to the database over the private network, so you don't need the TCP proxy. It's also quicker than dumping a big database from your laptop over the public connection. The dump streams straight to any S3-compatible bucket (S3, R2, B2, Wasabi, and so on).
Railway's own backups are good for rolling back, but they stay on Railway. This is for having a copy somewhere else.
Each run gets checked. Postgres dumps go through pg_restore --list while uploading, and MySQL dumps are checked for mysqldump's completion line. If the file can't be read back, the run fails and calls your webhook (Slack, Discord, anything). It keeps the last N backups and only deletes files it created.
Set DATABASE_URL=${{Postgres.DATABASE_URL}}, add your bucket details, and that's it.
Source is MIT: https://github.com/siddhujz/dumpling-backup. Issues and PRs welcome.