Node.js => Desafíos de rendimiento

Procesando consultas de larga ejecución con Nodo

Dado que Node es de un solo hilo, hay una solución alternativa si se trata de cálculos de larga ejecución.

Nota: este es el ejemplo "listo para ejecutar". Simplemente, no olvide obtener jQuery e instalar los módulos necesarios.

Lógica principal de este ejemplo:

Cliente envía solicitud al servidor.
El servidor inicia la rutina en una instancia de nodo separada y envía una respuesta inmediata con el ID de tarea relacionado.
El cliente continuamente envía cheques a un servidor para actualizaciones de estado de la ID de tarea dada.

Estructura del proyecto:

    project
    │   package.json
    │   index.html    
    │
    ├───js
    │      main.js
    │      jquery-1.12.0.min.js
    │   
    └───srv
        │    app.js
        ├─── models
        │      task.js
        └─── tasks
               data-processor.js

app.js:

var express     = require('express');
var app         = express();
var http        = require('http').Server(app);
var mongoose    = require('mongoose');
var bodyParser  = require('body-parser');

var childProcess= require('child_process');

var Task        = require('./models/task');

app.use(bodyParser.urlencoded({ extended: true }));
app.use(bodyParser.json());

app.use(express.static(__dirname + '/../'));

app.get('/', function(request, response){
    response.render('index.html');
});

//route for the request itself
app.post('/long-running-request', function(request, response){
    //create new task item for status tracking
    var t = new Task({ status: 'Starting ...' });
    
    t.save(function(err, task){
        //create new instance of node for running separate task in another thread
        taskProcessor = childProcess.fork('./srv/tasks/data-processor.js');

        //process the messages comming from the task processor
        taskProcessor.on('message', function(msg){
            task.status = msg.status;
            task.save();
        }.bind(this));

        //remove previously openned node instance when we finished
        taskProcessor.on('close', function(msg){
            this.kill();
        });

        //send some params to our separate task
        var params = {
            message: 'Hello from main thread'
        };

        taskProcessor.send(params);
        response.status(200).json(task);
    });
});

//route to check is the request is finished the calculations
app.post('/is-ready', function(request, response){
    Task
        .findById(request.body.id)
        .exec(function(err, task){
            response.status(200).json(task);
        });
});

mongoose.connect('mongodb://localhost/test');
http.listen('1234');

task.js:

var mongoose    = require('mongoose');

var taskSchema = mongoose.Schema({
    status: {
        type: String
    }
});

mongoose.model('Task', taskSchema);

module.exports  = mongoose.model('Task');

data-processor.js:

process.on('message', function(msg){
    init = function(){
        processData(msg.message);
    }.bind(this)();
    
    function processData(message){
        //send status update to the main app
        process.send({ status: 'We have started processing your data.' });

        //long calculations ..
        setTimeout(function(){
            process.send({ status: 'Done!' });
            
            //notify node, that we are done with this task
            process.disconnect();
        }, 5000);
    }
});

process.on('uncaughtException',function(err){
    console.log("Error happened: " + err.message + "\n" + err.stack + ".\n");
    console.log("Gracefully finish the routine.");
});

index.html:

<!DOCTYPE html>
<html>
    <head>
        <script src="./js/jquery-1.12.0.min.js"></script>
        <script src="./js/main.js"></script>
    </head>
    <body>
        <p>Example of processing long-running node requests.</p>
        <button id="go" type="button">Run</button>
        
        <br />
        
        <p>Log:</p>
        <textarea id="log" rows="20" cols="50"></textarea>    
    </body>
</html>

main.js:

$(document).on('ready', function(){
    
    $('#go').on('click', function(e){
        //clear log
        $("#log").val('');
        
        $.post("/long-running-request", {some_params: 'params' })
            .done(function(task){
                $("#log").val( $("#log").val() + '\n' + task.status);
                
                //function for tracking the status of the task
                function updateStatus(){
                    $.post("/is-ready", {id: task._id })
                        .done(function(response){
                            $("#log").val( $("#log").val() + '\n' + response.status);
                            
                            if(response.status != 'Done!'){
                                checkTaskTimeout = setTimeout(updateStatus, 500);
                            }
                        });
                }
                
                //start checking the task
                var checkTaskTimeout = setTimeout(updateStatus, 100);
            });
    });
});

paquete.json:

{
  "name": "nodeProcessor",
  "dependencies": {
    "body-parser": "^1.15.2",
    "express": "^4.14.0",
    "html": "0.0.10",
    "mongoose": "^4.5.5"
  }
}

Descargo de responsabilidad: este ejemplo pretende darle una idea básica. Para usarlo en el entorno de producción, necesita mejoras.

Modified text is an extract of the original Stack Overflow Documentation

Licenciado bajo CC BY-SA 3.0

No afiliado a Stack Overflow

Node.js
Desafíos de rendimiento

Buscar..

Procesando consultas de larga ejecución con Nodo