Challenge 2: Trace the Cross-Server Flow — Possible Solution
====================================================================

1. Client A sends a chat message over its WebSocket connection — but
   that connection only reaches SERVER A, since a WebSocket connection
   is tied to one specific server process.

2. Server A does two things: it relays the message to any of its OWN
   locally-connected WebSocket clients (if any are in the same room),
   and it PUBLISHes the message to a shared Redis channel that every
   server instance subscribes to.

3. Redis delivers that published message to every current subscriber
   of that channel — which includes SERVER B, because Server B
   subscribed to the same channel when it started up. Server A and
   Server B never communicate with each other directly at all; Redis
   is the only thing connecting them.

4. Server B receives the message FROM REDIS (not from Server A
   directly) and relays it over its own local WebSocket connections —
   reaching Client B, who was never connected to Server A and has no
   direct relationship with it whatsoever.

THE KEY INSIGHT: Server A and Server B don't need to know about each
other's existence, their network addresses, or how many other servers
exist at all — they only need to agree on which Redis channel to use.
Redis pub/sub is what makes an arbitrary number of otherwise-isolated
server instances behave as one logical system to connected clients.