Run electron scripts in managed workers
This module lets you run an electron script with scalability in mind, useful if you have to rely on electron to do heavy or long running tasks in parallel (web scraping, taking screenshots, generating PDFs, etc).
Works with electron@>=0.35.x including [email protected]
Note: This package is designed to work when used from a node.js process/app, since the main use case is to manage a pool of electron processes from a node.js process. if for some reason you need to use this package to manage a pool of electron processes from an electron process then PR are welcome 😃
- Install electron >= 0.35.x including electron@1, the easy way to install
electron in your app is
npm install electron --saveornpm install electron-prebuilt --save(or you can pass the path to yourelectronexecutable using thepathToElectronoption, see options)
There are two ways to communicate and distribute tasks between workers, each mode has its own way to use.
server-> Communication and task distribution will be doing using an embedded web server inside the electron process.ipc-> Communication and task distribution will be doing using an ipc channel.
The best mode to use will depend of how your electron app is implemented, however the recommended option is to use the ipc mode.
1.- First create an electron script wrapped in a webserver
script.js
var http = require('http'),
app = require('electron').app;
// every worker gets unique port, get it from a process environment variables
var port = process.env.ELECTRON_WORKER_PORT,
host = process.env.ELECTRON_WORKER_HOST,
workerId = process.env.ELECTRON_WORKER_ID; // worker id useful for logging
console.log('Hello from worker', workerId);
app.on('ready', function() {
// you can use any webserver library/framework you like (connect, express, hapi, etc)
var server = http.createServer(function(req, res) {
// You can respond with a status `500` if you want to indicate that something went wrong
res.writeHead(200, {'Content-Type': 'application/json'});
// data passed to `electronWorkers.execute` will be available in req body
req.pipe(res);
});
server.listen(port, host);
});2.- Start electron workers
var electronWorkers = require('electron-workers')({
connectionMode: 'server',
pathToScript: 'script.js',
timeout: 5000,
numberOfWorkers: 5
});
electronWorkers.start(function(startErr) {
if (startErr) {
return console.error(startErr);
}
// `electronWorkers` will send your data in a POST request to your electron script
electronWorkers.execute({ someData: 'someData' }, function(err, data) {
if (err) {
return console.error(err);
}
console.log(JSON.stringify(data)); // { someData: 'someData' }
electronWorkers.kill(); // kill all workers explicitly
});
});1.- First create an electron script
You will have an ipc channel available, what this means is that you can use process.send, and listen process.on('message', function() {}) inside your script
script.js
var app = require('electron').app;
var workerId = process.env.ELECTRON_WORKER_ID; // worker id useful for logging
console.log('Hello from worker', workerId);
app.on('ready', function() {
// first you will need to listen the `message` event in the process object
process.on('message', function(data) {
if (!data) {
return;
}
// `electron-workers` will try to verify is your worker is alive sending you a `ping` event
if (data.workerEvent === 'ping') {
// responding the ping call.. this will notify `electron-workers` that your process is alive
process.send({ workerEvent: 'pong' });
} else if (data.workerEvent === 'task') { // when a new task is executed, you will recive a `task` event
console.log(data); //data -> { workerEvent: 'task', taskId: '....', payload: <whatever you have passed to `.execute`> }
console.log(data.payload.someData); // -> someData
// you can do whatever you want here..
// when the task has been processed,
// respond with a `taskResponse` event, the `taskId` that you have received, and a custom `response`.
// You can specify an `error` field if you want to indicate that something went wrong
process.send({
workerEvent: 'taskResponse',
taskId: data.taskId,
response: {
value: data.payload.someData
}
});
}
});
});2.- Start electron workers
var electronWorkers = require('electron-workers')({
connectionMode: 'ipc',
pathToScript: 'script.js',
timeout: 5000,
numberOfWorkers: 5
});
electronWorkers.start(function(startErr) {
if (startErr) {
return console.error(startErr);
}
// `electronWorkers` will send your data in a POST request to your electron script
electronWorkers.execute({ someData: 'someData' }, function(err, data) {
if (err) {
return console.error(err);
}
console.log(JSON.stringify(data)); // { value: 'someData' }
electronWorkers.kill(); // kill all workers explicitly
});
});connectionMode - server, ipc mode, defaults to server mode if no specified.pathToScript (required) - path to the electron script.
pathToElectron - path to the electron executable, by default we will try to find the path using the value returned from the electron or electron-prebuilt packages (if any of them are found), otherwhise we will try to find it in your $PATH env var.
debug Number - pass debug port to electron process,see electron's debugging guide.
debugBrk Number - pass debug-brk port to electron process, see electron's debugging guide
electronArgs Array - pass custom arguments to the electron executable. ej: electronArgs: ['--some-value=2', '--enable-some-behaviour'].
env Object - pass custom env vars to workers. ej: env: { CUSTOM_ENV: 'foo' }.
stdio pass custom stdio option to worker's child process. see node.js documentation for details.
killSignal String - when calling electronWorkers.kill() this value will be used to kill the child process attached to the worker. see node.js docs for more info on signal events
pingTimeout Number - time in ms to wait for worker response in order to be considered alive, note that we retry the ping to a worker several times, this value is the interval between those pings. Default: 100
timeout - execution timeout in ms.
numberOfWorkers - number of electron instances, by default it will be the number of cores in the machine.
maxConcurrencyPerWorker - number of tasks a worker can handle at the same time, default Infinity
host - ip or hostname where to start listening electron web server, default localhost
portLeftBoundary - don't specify if you just want to take any random free port
portRightBoundary - don't specify if you just want to take any random free port
hostEnvVarName - customize the name of the environment variable passed to the electron script that specifies the worker host. defaults to ELECTRON_WORKER_HOST
portEnvVarName - customize the name of the environment variable passed to the electron script that specifies the worker port. defaults to ELECTRON_WORKER_PORT
If you are using node with nvm and you have installed electron with npm install -g electron-prebuilt you probably will see an error or log with env: node: No such file or directory, this is because the electron executable installed by electron-prebuilt is a node CLI spawning the real electron executable internally, since nvm don't install/symlink node to /usr/bin/env/node when the electron executable installed by electron-prebuilt tries to run, it will fail because node won't be found in that context.
1.- Install electron-prebuilt as a dependency in your app, this is the recommended option because you probably want to ensure your app will always run with the exact version you tested, and you probably don't want to install electron globally on your system.
2.- You can make a symlink to /usr/bin/env/node but this is not recommended by nvm authors, because you will lose all the power that nvm brings.
3.- Put the path to the real electron executable in your $PATH.
See license