How to Set Up Zero-Downtime Deploys for Go

Achieve zero-downtime Go deploys by running new instances alongside old ones and using Server.Shutdown to gracefully drain active connections before stopping the old process.

Zero-downtime deploys in Go are achieved by running a new instance alongside the old one, draining connections, and swapping traffic. This approach relies on the net/http package's Server.Shutdown method to gracefully stop accepting new requests while finishing in-flight ones.

// 1. Start new server on a different port or via load balancer
// 2. Signal old server to drain
ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
err := oldServer.Shutdown(ctx)
cancel()
if err != nil {
    log.Fatalf("Server forced to shutdown: %v", err)
}
// 3. Update load balancer to point to new server
// 4. Stop old server process

The Shutdown method closes the listener and waits for all active connections to close or timeout, ensuring no requests are dropped during the transition.

Zero-downtime deployment means updating your software without ever taking it offline for users. Think of it like changing the tires on a moving car: you build the new version while the old one keeps running, then you smoothly switch traffic to the new version before shutting down the old one. This ensures your users never see an error page or experience a delay.