We were seeing this today as well with endpoints we know to be valid. Azure appears to be having (or had) some sort of downtime they haven't owned up to.