This incident has been resolved.

load balancers appear happier after some manual pruning, we'll continue to monitor the siutation

Appears we are seeing some instability on our amazon edge nodes.
Reprioritizing and relaunching instances.