I would like to see an OPTION where if a poller is down for X amount of time, the core will evenly distribute the devices on that poller to the other pollers.
Pollers could have an option to participate in poller failover, and nodes could also have an option to ignore.