It would be helpful to record all of the current operations during shutdown. This would be especially helpful in the case of a fatal assertion, or when diagnosing slow operations if the server is shutdown manually.

This would also be useful during transitions within a replica member (like primary->secondary, of secondary->recovery).