Currently there is no way to know if the configured amount of pollers is enough to handle all requests coming from the server. Some kind of '% of time busing serving requests by pollers' would be very handy here.