The default max open conns (50) was causing issues with small DB instances (`db.t4g.medium`) in loadtesting (we run 10 fleet instances, and 500 was over the limit number of connections for that instance type). Also upgrading Redis version to a more recent version that users are most likely using.