In simulated cases of heavy load (writes occurring every 1-10ms) it is possible to prevent the rescheduling of server monitoring checks. In this scenario, the logic to debounce calls to monitoring within minHeartbeatFrequencyMS is tying up the event loop and preventing future monitoring attempts from being scheduled.