When there is too much load on a pod it will currently not work harder until CPU Utilization is exhausted. Instead jobs for the fetcher will queue up higher and higher so no results will be returned to the user.