การจัดการการตัดการเชื่อมต่อและการเชื่อมต่อใหม่

In the real world, network connections aren't always perfect. WebSockets, despite being persistent, can break due to many reasons.

Ignoring these disconnections can lead to unresponsive applications and a poor user experience. This lesson explores how to gracefully handle these interruptions.

What makes a WebSocket connection drop?

• Network Issues: Wi-Fi drops, internet outages, proxy problems.
• Server Restarts: The server application might restart for updates or maintenance.
• Client Offline: The user closes their browser, loses power, or puts their device to sleep.
• Idle Timeouts: Some proxies or firewalls might close idle connections.

On the client side, the browser's native WebSocket API provides events to notify you about connection status changes.

The most important event for detecting a disconnection is the onclose event. It fires when the connection is either gracefully closed by the server or abruptly lost.

When onclose fires, you can immediately try to reconnect. This simple approach works for temporary glitches but can be problematic.

Try running this basic client-side example:

Continuously trying to reconnect immediately (e.g., every 1 second) can overwhelm both the client and the server.

• Client Resource Drain: Constant connection attempts consume CPU and network resources.
• Server Overload: If many clients disconnect and then flood the server with reconnect requests, it can lead to a Denial of Service (DoS) situation.
• No Backoff: It doesn't account for prolonged outages.

To prevent overloading and provide a better experience, we use reconnection backoff strategies.

A backoff strategy introduces delays between reconnection attempts, and these delays typically increase with each failed attempt. This gives the server time to recover and prevents a "thundering herd" problem.

Exponential backoff is a common and effective strategy. It means the delay between retries increases exponentially.

• First retry: 1 second
• Second retry: 2 seconds
• Third retry: 4 seconds
• Fourth retry: 8 seconds

You usually cap the maximum delay to prevent excessively long waits and add some random "jitter" to avoid synchronized retries.

Let's enhance our client-side code to use exponential backoff with a maximum delay and some random jitter. Try running it!

The server also needs to know when a client disconnects. This is crucial for cleaning up resources, updating connected user lists, or stopping data streams for that client.

Using a library like ws in Node.js, the WebSocket server emits a 'close' event on the individual client connection object when it disconnects.

Here's a basic Node.js WebSocket server example showing how to handle client disconnections and remove them from an active connections list.

We've explored how crucial it is to handle disconnections in realtime applications. Both client and server play a role in maintaining connection resilience.

• Clients use onclose to detect disconnections.
• Exponential backoff with jitter and a max delay prevents server overload during reconnections.
• Servers use on('close') to clean up resources when clients leave.

These strategies ensure your realtime apps remain responsive and reliable, even in unstable network conditions.