For database recovery, we rely on the automated backups set up in AWS RDS that takes snapshots daily. This process also maintains a rolling window of 35 days of previous snapshots. We rely on version control of our Hasura metadata and migrations.
Taking a snapshot of the RDS instance is also part of the Release Process in case recovery to a pre-release state is needed.
Discontinued process
We previously implemented cron jobs on our on-prem servers to take backups of the database and Hasura-related artifacts. These files were stored in a S3 bucket called atd-moped-db-backups but this process was stopped after 5/16/2024.
The epic issue that tracked this work can be used to find out more.
