node.js - How to use nodeJS cluster with mySQL pool cluster?

Question

Quick question

If I make a node cluster application with 4 workers (4 instances of my application), should I use mySQL pool or mysql pool cluster? If I use pool it will create one pool for each application but If I use pool cluster it will create 4 pools for each application (16 total). Is this a good implementation or will it actually slow the performance?

Let's make a fake example to illustrate what I am asking. I am creating a nodeJS server app like this.

---

First, let's make the configuration file for the mysql database (important is the last section of this file where I make the db workers):

DBconfig.js

'use strict'
const mysql   = require('mysql'),
      workers = process.env.WORKERS || require('os').cpus().length,
      cluster = require('cluster');
      
//Local Database Settings
const local_settings = {
    user       : 'user',
    host       : '127.0.0.1',
    password   : 'pass',
    database   : 'dbname',
    debug      : false,
    dateStrings: true,
    connectionLimit     : 10,
    defaultSelector     : 'RR',
    multipleStatements  : true,
    removeNodeErrorCount: 1
};

let poolCluster = module.exports = mysql.createPoolCluster( local_settings );

//here I make one db worker for each app worker
for(let i = 0; i < workers; i++){
    poolCluster.add(`DBWORKER_${process.pid}_${i}`, local_settings);
}

Then I create a global Library to use in my application where I have my database connection function

globalLib.js

'use strict'
const path        = require('path'),
      poolCluster = require(path.join('path','to_the','DBconfig.js'));

global.db = obj => {
  return new Promise( (resolve, reject) => {
    poolCluster.getConnection(function(err, connection){
        if(err) {reject(err);return}
        
        connection.query(obj.query, obj.params, function(){
            connection.release();
            if(!err) resolve(rows)
            else reject(err);
        });
        
        connection.on('error'), function(err){
            reject(err);
        });
    }
  })
}

app.js

'use strict'
const express = require('express'),
      path    = require('path');
let   app     = module.exports = express();

//here I handle all with express.Router()
const index = require(path.join('path','to_my','index.js'));

app.use('/', index)

and in the end I have a cluster file from where I start the server (everything here is important):

cluster.js

'use strict'
const path    = require('path'),
      cluster = require('cluster'),
      sockets = require(path.join('path_to','sockets.js')),
      app     = require(path.join('path_to','app.js')),
      pclust  = require(path.join('path_to', 'DBconfig.js')),
      workers = process.env.WORKERS || require('os').cpus().length;

if (cluster.isMaster) {
  
    for (var i = 0; i < workers; ++i) {
        var worker = cluster.fork();
    }
   
    cluster.on('disconnect', function(worker) {
        pclust.remove(`DBWORKER_${worker.process.pid}_*`);//remove all connections relative to this process pid
        var worker = cluster.fork();
    });
}
else {

    //Start Application
    var server = app.listen(8080, '127.0.0.1', function(){
        console.log('yeeeeey');
    });

    // initialize socket
    sockets.initialize(server,app);
}

---

Let's assume that I have 4 cpus, so there will be 4 instances of my app.js. Therefore I create 4 database workers one for each cpu in each application so in the end I have:

cpuID    server instances    db workers

 1            1                 4

 2            1                 4

 3            1                 4

 4            1                 4

 4            4                 16 ( TOTAL )

So that leaves me with a lot of db workers ... The question is, is this a good implementation of nodeJS cluster and mySQL pool cluster? If not which is the right way?

Answer 1

I am also curious about this answer. I am assuming that with the higher work with the number of processors it would generally perform better with less requests, but with a lot of traffic it would be better to use.

In the app.js file I am using these packages within app.js:

const cluster = require('cluster');
const http = require('http');
let osu = require('node-os-utils'); //get's number of cpus
var cpu = osu.cpu;

I have also installed autocannon.

if(cluster.isMaster) {
  const availableCpus = cpu.count();

  console.log('clustering to', availableCpus, 'processes');
  for (let i = 0; i < availableCpus; i++) {
    cluster.fork();
  }

  cluster.on('online', function(worker) {
    console.log('worker:', worker.process.pid, 'is online');
  })

  cluster.on('exit', (worker, code, signal) => {
    if(code !== 0 && !worker.exitedAfterDisconnect) {
      console.log('Starting new worker:', worker.process.pid, + 'signal:', signal);
      cluster.fork();
    }
  })
}
else {

  app.get('/', function (req, res) {
    let i= 100000000; while (i>0) {i--}
    res.send();
  });

  app.listen(port, () => console.log("APP TEST: port %s", port));

}

Using autocannon without cluster:

Using autocannon with cluster and 8 workers:

Also, without node.js cluster: 44k requests. With node.js cluster and 8 workers: 40k requests.

I would find it helpful if anyone could give an explanation on this.