Health and Readiness Endpoints for Kubernetes

Without health check endpoints, Kubernetes can't distinguish a starting server from a crashed one, routing traffic to containers that aren't ready or restarts them prematurely.

Implement separate liveness and readiness endpoints that check different conditions.

// Liveness: is the process alive? (simple — should almost never fail)
app.get('/health/live', (req, res) => {
  res.status(200).json({ status: 'ok' });
});

// Readiness: can the process handle traffic? (check dependencies)
app.get('/health/ready', async (req, res) => {
  const checks = await Promise.allSettled([
    db.query('SELECT 1'),           // database
    cache.ping(),                   // redis
    queue.isConnected(),            // message queue
  ]);

  const failed = checks
    .map((c, i) => ({ name: ['db', 'cache', 'queue'][i], result: c }))
    .filter(c => c.result.status === 'rejected');

  if (failed.length > 0) {
    return res.status(503).json({
      status: 'not-ready',
      failures: failed.map(f => f.name),
    });
  }

  res.status(200).json({ status: 'ready' });
});

Liveness probes restart stuck processes. Readiness probes prevent traffic from reaching healthy-but-not-ready pods (e.g., during startup or DB connection loss). They serve different purposes.