node-worker-pool

node-worker-pool

node-worker-pool is a library for managing a pool of child workers in node.

It's primarily useful for scenarios where you have lots of highly parallelizable tasks you want to perform. It works exclusively via message-passing, so there is no need to share memory.

Specifically, node-worker-pool allows you to define your own worker executable that is capable of communicating over stdin/stdout (via a fairly simple protocol for which I have yet to propertly document :p).

Getting started

• Write a worker executable file
• Construct a WorkerPool object that points at the aforementioend worker executable
• Send messages to the WorkerPool object and wait for responses

Writing a worker executable file

You technically don't have to write this file in node, but for the time being there are only node helper libraries for abstracting away the communciation protocols. Here is an example worker:

worker.js

var workerUtils = require('node-worker-pool/nodeWorkerUtils');

/**
 * Executed once when the worker pool first starts
 * (before any messages are received)
 */
var initData;
function onInitialize(data) {
  initData = data;
}

/**
 * Executed each time a message is received from the worker pool.
 * Returns the response to the message (response must always be an object)
 */
function onMessage(data) {
  return {
    initData: initData,
    receivedData: data
  };
}

if (require.main === module) {
  try {
    workerUtils.startWorker(onInitialize, onMessage);
  } catch (e) {
    workerUtils.respondWithError(e);
  }
}

workerPool.js

if (require.main === module) {
  var workerPool = new WorkerPool(
    8, // number of workers
    process.execPath, // path to the node binary
    './worker.js' // path to the worker script
  );

  workerPool.sendMessage({message: 'hai!'}).then(function(response) {
    console.log(response); // Prints the response object from the worker
  });

  workerPool.shutDown().then(function() {
    console.log('All worker processes have now been killed');
  });
}