Currently kas keeps spinning, trying to contact another kas that is unreachable (perhaps it crashed or was shut down but somehow didn't clean up tunnel records from Redis). This generates a lot of logs (1 record every 50ms!) which is a waste of resources.