The admin interface went down due to a misconfiguration -- we didn't get notified of this problem but rather noticed it.
While CI / CD won't be able to detect this a priori, we should somehow have a health check (either on the DO side of things, or via uptime robot)