As I see this it works when there is behind the scenes backup schedule of database (because once a day isn’t enough, is it?) and then inbuild warningsystem is lagging behind, max 24 hours. But it would work as early warning system if Discourse is down because of database failing, but because of caching users don’t see it right away.
Am I totally wrong? Not even close?