At which point all the traffic causes another outage so they can't reconnect, so they timeout for a fixed amount of time and then try back again all at the same time...