How do you implement graceful shutdown in a Node.js service?

A deployment sends SIGTERM before killing a process. If the process exits immediately, in-flight requests fail and queue messages are abandoned. Graceful shutdown gives the instance time to finish work while the load balancer stops sending new traffic.

const server = app.listen(3000);

process.on("SIGTERM", () => {
  server.close(async () => {
    await pool.end();
    process.exit(0);
  });

  setTimeout(() => process.exit(1), 10000).unref();
});

The sequence

• Flip the readiness probe to false so the load balancer removes the instance
• Stop accepting new connections with server.close
• Let in-flight requests finish, with a bounded timeout
• Stop consuming from queues and finish or requeue current messages
• Close database pools, Redis clients and telemetry exporters
• Exit with a success code, or force exit after the deadline

The platform's termination grace period must be longer than your drain timeout, otherwise the process is killed before it finishes. In Kubernetes, readiness removal and SIGTERM are not instantaneous, so allow a short pre-stop delay if traffic still arrives. Background jobs need their own drain logic.