Hello, we have on have a little bit more abstraction between cluster setup and our python code and plan on hiding the cluster behind load balancing proxy (nginx), thus just single endpoint from client point of view. I have checked various load balancing strategies and noticed, that some of these stragies forward traffic to different elastic node for every single request from the same client - while other e.g. ip_hash just try to use same node for given client as long as possible
Does the forwarding to different node for each call leads to repeated authentication of the client to the node (I assume creating some unnecessary overhead), or is the previous connection somehow shared across the elasticsearch cluster?
Thanks!