I have figured out that if I always give an aggregator the same name eg. "server", then I can do a Network discovery (ND) command to find the "server", or a Discover Node (DN) command, I can identify the aggregator. I can then load the DL, DH values with the aggregator MAC address.
For the second scenario, I can implement a watchdog timer on the secondary aggregator that pings the primary one. If the watchdog times out, the secondary can issue a AG command to have all members of the network send through the new primary.