Graceful shutdown

Problem

What becomes confusing, fragile, or impossible without understanding graceful shutdown? This lesson answers that through explanation, a worked example, two runnable exercises, and a reference solution. No teacher-supplied worksheet is required.

Node applied JavaScript’s event-driven model to servers and command-line programs, pairing a single process with nonblocking I/O.

Concept explanation

Graceful shutdown belongs to “One process under load”. graceful shutdown must be connected to layer below by tracing representation, ownership, control, and failure. Capacity ends where CPU, memory, descriptors, event-loop delay, or dependencies saturate.

For graceful shutdown, trace concrete input, state transition, output, and failure through a server-side runtime boundary involving process lifecycle, filesystem or network I/O, streams, and capacity.

One process owns an event loop, module graph, heap, handles, streams, signals, and explicit boundaries to the OS and network. Apply that model to supplied normal, boundary, and failure cases; each case below names its input and expected evidence.

Example bank

Compare normal, boundary, failure, and cross-layer cases. Predict each observation before revealing the explanation.

Worked example

Start from supplied server.mjs. Focus: Trace graceful shutdown end-to-end while running: Load-test one-process server and record latency percentiles, throughput, handles, CPU, memory.

1. Run: node server.mjs > server.log 2>&1 & server_pid=$!; trap 'kill $server_pid' EXIT; sleep 1; curl -i http://127.0.0.1:3000/
2. Save baseline evidence. Use Node inspector, active handles, process signals, curl, CPU profiles, heap snapshots, and OS process tools.
3. Boundary case: At every graceful shutdown boundary, label owner and representation during: Increase concurrency until first measured knee.
4. Failure case: Locate first layer where evidence diverges during: Send termination under load and verify graceful drain timeout.

RESULT
Trace explains Report names first bottleneck from evidence and distinguishes worker/child-process trade-offs. without skipping any conversion, queue, process, protocol, or storage boundary. Starter-level baseline: Server prints its URL; curl receives 200, JSON content type, correct byte length, and a body containing topic and GET method.

Starter material

Create server.mjs, paste this exact content, then run the command below.

import { createServer } from "node:http";

const server = createServer((request, response) => {
  const body = JSON.stringify({ topic: "graceful shutdown", method: request.method });
  response.writeHead(200, { "content-type": "application/json", "content-length": Buffer.byteLength(body) });
  response.end(body);
});

server.listen(3000, "127.0.0.1", () => console.log("http://127.0.0.1:3000"));

STOP / CLEANUPSupplied command traps shell exit and stops server. If running server alone, press Ctrl+C.

Independent exercise

Remove one abstraction: graceful shutdown

1. Create second case from blank file: At every graceful shutdown boundary, label owner and representation during: Increase concurrency until first measured knee.
2. Then create controlled failure: Locate first layer where evidence diverges during: Send termination under load and verify graceful drain timeout.
3. Use Node inspector · curl · SQLite CLI · process tools to prove behavior, then repair controlled failure.
4. Compare result against supplied acceptance checks and reference approach before marking complete.

Expected result

• Trace explains Report names first bottleneck from evidence and distinguishes worker/child-process trade-offs. without skipping any conversion, queue, process, protocol, or storage boundary.
• Server prints its URL; curl receives 200, JSON content type, correct byte length, and a body containing topic and GET method.
• Controlled graceful shutdown failure produces captured evidence; repair restores stated invariant without hiding error.

Hints

Solution

Attempt both exercises before opening reference approach.