I've noticed that when I scale down the pods of the operator to zero also the check for uptime will stop, the graph will show a blank period and the uptime percentage will ignore this period and calculate the uptime using as the base only the period for which it has data: