Curious to know whether Proteus broker preserves the reactive chain while bridging two clients. I mean, is the back-pressure escalated end-to-end? If so, what is the behaviour when a service (one of the clients) is shut down in an active communication? Does the other end gets notified or broker assigns new service instance transparently. The question is more in the perspective of service HA.
I did setup a Nginx based HA where client and service is communicating over a channel. In this traditional HA set up, client is getting ChannelClosed exception because of websocket’s connection oriented nature. So when that exact server is terminated where client is connecting to, HA setup is not worthy.
How does Proteus broker address this scenario?