Maintenance:
Enable the point-in-time recovery option on the main Postgres database. This should be a zero-downtime operation but we noticed that connections were dropping while testing in the Staging Environment. Our backend systems do reconnect but there could be an impact on transactions that are in-flight. We estimate that the operation will take 20min to 30min, but the connection dropping only occurred during a smaller time frame of a couple of minutes in the beginning.
Why does it matter:
- Reduces data loss from hours to seconds in case of a database failure
- Enables us to conduct database recovery exercises without downtime as required per SOC II and ISO 27001.
