- Take different actions for different database errors while connecting. * Network error should keep retry on the next worker run. * Authentication error should not (put the server in a permanent condition until restart). - Create **runtime** error/information aggregation counters for each database.