You can specify the number of replicas for clouddriver. But what does that do? Am I just going to end up doing the caching runs N times? Do multiple clouddrivers coordinate somehow?
Clouddriver caching agents are sharded by location and resource type. Caching agents that handle a resource type in a location each attempt to grab a lock in Redis before caching (even across multiple Clouddriver replicas), and the first agent that does acquire the lock actually handles the resource caching and writes to Redis.