This article covers a version of Ubuntu that is no longer supported. If you are currently operate a server running Ubuntu 12.04, we highly recommend upgrading or migrating to a supported version of Ubuntu:
Reason: Ubuntu 12.04 reached end of life (EOL) on April 28, 2017 and no longer receives security patches or updates. This guide is no longer maintained.
See Instead: This guide might still be useful as a reference, but may not work on other Ubuntu releases. If available, we strongly recommend using a guide written for the version of Ubuntu you are using. You can use the search functionality at the top of the page to find a more recent version.
A daemon is a program that runs in background and has no controlling terminal. They are often used to provide background services. For example, a web-server or a database server can run as a daemon.
This tutorial will show you how to write a daemon with Node.js and deploy it on your VPS with Upstart.
I will focus on implementing a standard daemon. I’m using upstart just for simplicity, however you can write a System-V init script or use anything you like to start your daemon.
For this tutorial, you will need a Linux VPS (Preferably Ubuntu or CentOS), Node.js, and Upstart.
There are several ways to install Node.js. The easiest way in my opinion is to use nvm. It also allows you to manage different versions of Node.
Alternatively, you can follow one of these guides:
Upstart comes pre-installed on many Linux distros. If it’s not installed on the distro of your choice, you should be able to install it from official repositories. You can also compile it from the source-code. Refer to Upstart Getting Started page for more info.
Basically a daemon starts like a normal process. the following will occur afterwards:
Instead of closing the standard file descriptors, the parent process can open the null device and attach it to the child’s standard file descriptors.
For the sake of this tutorial, we’re going to create a simple HTTP daemon.
Our daemon will be able to:
The following is the initial project structure:
node-simple-http-daemon/
|
|-- README.md
|-- bin/
| `-- node-simple-http-daemon
`-- lib/
`-- app.js
Let’s start by creating a package.json file:
$ npm init
Install daemon module:
$ npm --save install daemon
And here is how the package.json should look like:
{
"name": "node-simple-http-daemon",
"version": "0.0.0",
"description": "Simple HTTP Daemon written in Node.js",
"main": "lib/app.js",
"scripts": {
"test": "echo \"Error: no test specified\" && exit 1",
"start": "node lib/app.js"
},
"author": "Fardjad Davari",
"license": "MIT",
"dependencies": {
"daemon": "~1.1.0"
}
}
Note that we added a start script, So we can start the app with npm start
command later.
For now, we’re going to create a simple HTTP server that starts listening on port 80 and responds with “Hello world” for every request.
Put the following in lib/app.js file:
/**
* lib/app.js
*/
const PORT = 80;
const ADDRESS = '0.0.0.0';
var http = require('http');
var server = http.createServer(function (req, res) {
res.writeHead(200, {'Content-Type': 'text/plain'});
res.end('Hello World\n');
});
server.listen(PORT, ADDRESS, function () {
console.log('Server running at http://%s:%d/', ADDRESS, PORT);
console.log('Press CTRL+C to exit');
});
You can start the VPS by running sudo npm start
.
The following is the daemon executable code.
The following should be placed in bin/node-simple-http-daemon:
#!/usr/bin/env node
/**
* bin/node-simple-http-daemon
*/
// Everything above this line will be executed twice
require('daemon')();
var cluster = require('cluster');
// Number of CPUs
var numCPUs = require('os').cpus().length;
/**
* Creates a new worker when running as cluster master.
* Runs the HTTP server otherwise.
*/
function createWorker() {
if (cluster.isMaster) {
// Fork a worker if running as cluster master
var child = cluster.fork();
// Respawn the child process after exit
// (ex. in case of an uncaught exception)
child.on('exit', function (code, signal) {
createWorker();
});
} else {
// Run the HTTP server if running as worker
require('../lib/app');
}
}
/**
* Creates the specified number of workers.
* @param {Number} n Number of workers to create.
*/
function createWorkers(n) {
while (n-- > 0) {
createWorker();
}
}
/**
* Kills all workers with the given signal.
* Also removes all event listeners from workers before sending the signal
* to prevent respawning.
* @param {Number} signal
*/
function killAllWorkers(signal) {
var uniqueID,
worker;
for (uniqueID in cluster.workers) {
if (cluster.workers.hasOwnProperty(uniqueID)) {
worker = cluster.workers[uniqueID];
worker.removeAllListeners();
worker.process.kill(signal);
}
}
}
/**
* Restarts the workers.
*/
process.on('SIGHUP', function () {
killAllWorkers('SIGTERM');
createWorkers(numCPUs * 2);
});
/**
* Gracefully Shuts down the workers.
*/
process.on('SIGTERM', function () {
killAllWorkers('SIGTERM');
});
// Create two children for each CPU
createWorkers(numCPUs * 2);
Time to start our daemon! But before that, we should modify our app.js to handle SIGTERM.
Add the following to the end of the app.js file:
process.on('SIGTERM', function () {
if (server === undefined) return;
server.close(function () {
// Disconnect from cluster master
process.disconnect && process.disconnect();
});
});
Make the daemon script executable:
$ chmod +x bin/node-simple-http-daemon
And run it (Make sure nothing else is running on port 80):
$ sudo bin/node-simple-http-daemon
Now, your daemon and it’s workers should be running in background. You can confirm that by sending an HTTP GET request via cURL:
$ curl 127.0.0.1
Before moving to the next step, let’s take a look at the process IDs:
$ ps -axf | grep [n]ode-simple-http-daemon | \
awk '{ print "pid:"$2", parent-pid:"$3 }'
Sample output:
pid:33811, parent-pid:1
pid:33853, parent-pid:33811
pid:33856, parent-pid:33811
pid:33857, parent-pid:33811
pid:33858, parent-pid:33811
pid:33859, parent-pid:33811
pid:33860, parent-pid:33811
pid:33861, parent-pid:33811
pid:33862, parent-pid:33811
Note that the daemon is the process with parent-pid of 1.
Try restarting the workers by sending SIGHUP to the daemon:
$ kill -HUP 33811 # (replace 33811 with the daemon PID)
Now if you list the PIDs again, you should be able to see new worker processes with new PIDs. Amazing, isn’t it?
To stop the daemon, just run:
$ kill -TERM 33811 # (replace 33811 with the daemon PID)
We’re almost done. We only need to make workers drop their privileges after the VPS gets started.
Modify server.listen()
callback in app.js so it reads:
server.listen(PORT, ADDRESS, function () {
console.log('Server running at http://%s:%d/', ADDRESS, PORT);
console.log('Press CTRL+C to exit');
// Check if we are running as root
if (process.getgid() === 0) {
process.setgid('nobody');
process.setuid('nobody');
}
});
And that’s it for the daemon part.
Now you can install it system-wide:
$ sudo npm link
Making an Upstart job is very easy. Create a file in /etc/init/node-simple-http-daemon.conf
with the following contents:
# /etc/init/node-simple-http-daemon.conf
start on started network
stop on stopping network
respawn
expect daemon
exec https-proxy-daemon
Now you can:
$ sudo start node-simple-http-daemon # Start the job
$ initctl --system node-simple-http-daemon # Check the job status
$ sudo reload node-simple-http-daemon # Send SIGHUP (restart the workers)
$ sudo stop node-simple-http-daemon # Stop the job
You may want to modify the daemon to give up spawning workers when they’re getting killed too frequently (spinning).
I can’t emphasize enough how important it is to have adequate logging in your application. Make sure to check out node-bunyan.
<div class=“author”>Submitted by: <a href=“http://www.fardjad.com/”>Fardjad Davari</a></div>
Thanks for learning with the DigitalOcean Community. Check out our offerings for compute, storage, networking, and managed databases.
This textbox defaults to using Markdown to format your answer.
You can type !ref in this text area to quickly search our full set of tutorials, documentation & marketplace offerings and insert the link!
Running a VM on Azure. Went through your setup, started the UpStart job for node-simple-http-daemon, and verified that it was running. Then, I restarted my Azure VM from Azure Portal, and ssh’d back into VM and did a
initctl status node-simple-http-daemon
, and I got anode-simple-http-daemon stop/waiting
. My assumption would be that it would be running. Am I correct, or am I missing something? Thank you!Ahh, I see. My apologies to you @fardjad! That’s an interesting behavior of node.js that I didn’t realize. My conclusion came from putting a print statement at the location of the require and finding lots of duplicate printing. If I put the print statement in the required file I see what you intended. Thanks for the clarification! Maybe you can link to that modules behavior in the example directly?
This is wrong!!!
You have the following function:
This is being called in your createWorkers() function. For my system that means it will get called 8 times…or so we think. Take a look at:
That will fork a child process and run through all of the code again. That means createWorkers() will get called again by the child process, and that means createWorker() will get called another 8 times by that child process and will run require(‘…/lib/app’); 8 times over. In total createWorker() will get called 64 times!!!
Instead you should have something like this:
Great introductory article. Thanks for the suggestion @n664352, it worked for me using the absolute address
exec /var/www/mynodeApp/bin/node-simple-http-daemon
for those getting “Job failed” change “exec https-proxy-daemon” to use the full path to your daemon script. e…g “exec /var/www/bin/node-simple-http-daemon”
same problem as otamealsnap. I can run the daemon just fine and my app is up and running fine. However, when I try via the upstart script with ‘sudo start…’ I get: start: Job failed to start
help :)
@oatmealsnap: If you run “node-simple-http-daemon” directly, are you seeing any errors?
I got to the end with everything working alright, but when I try to run:
I get
Has anyone else run into this?
Actually, here is an excellent guide on deploying NodeJS daemons using systemd: http://savanne.be/articles/deploying-node-js-with-systemd/
excellent post, thanks!
two recommended alternatives (based on my experience) that folks can check out also:
systemd instead of upstart. winston instead of bunyan.