Thank you for bringing up this important point!

You're absolutely right—when deploying a WebSocket service with FastAPI across multiple workers or pods, managing connection states becomes more complex. WebSocket connections are typically stateful, and when you scale horizontally with multiple workers or pods, each instance may not be aware of the connections handled by others.

To address this, additional infrastructure like Redis, a message broker, or a shared memory solution is often necessary to synchronize connection states across all workers. This ensures that messages can be routed correctly, and the overall connection management remains consistent.

It's an essential consideration for anyone scaling WebSocket services, and I'm glad you highlighted it!

Thanks again for your input. If you have more insights or questions, feel free to share!