Hi
it happened (at least) twice, with two different sensors, that for some reasons the Bias-T connector was suddenly detached. This caused the coverage area to shrink drastically. It took some days to realize the problem, and during that period the sensors were marked as "online" but were sending very little data.
It would be probably easy to install some automatic sanity check, that e.g. compares the coverage area at the end of each day with the maximum coverage area recorded in the past, and if the former drops below say 50% of the latter, then an automatic email alert could be send.
It takes very little to reconnect the Base-T cable, but it may take several days until somebody go and check manually the status of own sensor. That automatic alert would help us to react more promptly (and perhaps it's not too difficult to implement, since coverage area is already part of the computed stats).
Along the same line, you could consider implementing simple automatic alerts that are triggered upon anomalous drop of message rate, number of aircrafts etc. (again, based on daily threshold learned from the past). That will improve the reactivity of opensky contributors to hardware glitches and other minor problem (minor to fix, but not easy to detect).
Just a suggestion ...