A long-running database migration casing a timeout. Kubernetes was retrying the to run the migration, which (unexpectedly) made the Dashboard offline.

We rolled-back the production deployment. We are still investigating why the table is locked and how to solve it. For the moment, we used a different strategy that does not require database migration so we can proceed with the implementation without being affected by table locks.

Scarf Web Application went offline after production deployment. Scarf Gateway was not affected.