Self-describing proxies? about cloudstate HOT 2 CLOSED

Absolutely - there's multiple reasons why we might want to do this, one reason is so that the load balancer can more intelligently route requests. If the load balancer can know the entity key and compute the shard for each request, then we can do things like have the Akka sidecar include a hint in its response headers with which node to route the next request for that shard to, and the load balancer could use that. Or, alternatively, the load balancer could make its own decisions about where to route requests to shards to, and then Akka cluster sharding can use that information as a hint for its distribution.

from cloudstate.

Solved by implementing ServerReflection support and supporting FileDescriptorSets in current master.

from cloudstate.