feedback loop to orchestrator

[willscott]

currently the only way that caboose reports success/failures of backends in a way to trigger dynamic re-balancing is through logs

related to #16 , there should be some explicit feedback loop to the orchestrator - e.g. saying when a member of the pool has been excluded for misbehaving, that can allow for re-balancing / smarter pool refreshes in realtime

[aarshkshah1992]

@willscott So Caboose informs the Orchestrator of misbehaving L1s and the Orchestrator does some internal downvoting of them too so as to not return those L1s subsequently ?

[willscott]

right

• this is important for helping the orchestrator coordinate exclusion across multiple clients
• this is useful as orchestrator's input is used in returning to DNS resoluion