cluster.js
const cluster = require('cluster');
const http = require('http');
const numCPUs = require('os').cpus().length;
if (cluster.isMaster) {
// Fork workers.
for (let i = 0; i < numCPUs; i++) {
cluster.fork();
}
cluster.on('exit', (worker, code, signal) => {
console.log(`worker ${worker.process.pid} died`);
});
} else {
// Workers can share any TCP connection
// In this case it is an HTTP server
require('./server.js')();
}
server.js
const http = require('http');
function startServer() {
const server = http.createServer((req, res) => {
res.writeHead(200);
res.end('Hello Http');
});
server.listen(3000);
}
if(!module.parent) {
// Start server if file is run directly
startServer();
} else {
// Export server, if file is referenced via cluster
module.exports = startServer;
}
In this example, we host a basic web server, however, we spin up workers (child
processes) using the built-in cluster module. The number of processes forker depend on
the number of CPU cores available. This enables a Node.js application to take
advantage of multi-core CPUs, since a single instance of Node.js runs in a single
thread. The application will now share the port 3000
across all the processes. Loads
will automatically be distributed between workers using the Round-Robin method by
default.