We recently gave setFCV a dry run argument, allowing administrators to detect possible errors in downgrade before committing to executing the command fully. 

This is extremely useful when evaluating FCV downgrade for one cluster

In a larger fleet, evaluating the riskiness of a fleetwide downgrade is a little harder. An orchestrator must mass issue setFCV with dry run and examine the success or failure in aggregate. It also requires work on every cluster.

This feature requests that the use of features which are not compatible with earlier FCVs result in some kind of counter or signal changing in serverStatus. Perhaps, a minSupportedFCV setting which is evaluated for increment each time a write requires a particular FCV. 

This would allow serverStatus to be evaluated in aggregate, without inducing work on the fleet