scenes to be used

At the beginning of this article, talk about the usage scenarios encountered. After the nodejs program is written, it must be packaged into a Docker image. When compiling the Dockerfile, the system will download the used modules from the npm module repository. This will encounter two problems:

First, in most cases, we update the program, but only the source code, not the module configuration in package.json. It is a waste of time to download modules from the network every time the image is compiled.

Second, the production environment is isolated from the external network; or the network environment is unstable.

I gave solutions around these two problems and wrote an example in this article.

Solution

First we need to create a nodejs-base image to store the npm modules we need to put in this docker image.

Create a node-js project, because this project is mainly for downloading npm modules, so there is no program and the structure is simple, as shown below:

nodejs-base
  │
  ├─Dockerfile
  ├─main.js
  └─package.json

Dockerfile

FROM node:8-onbuild

# set timezone as china/shanghai
RUN cp /usr/share/zoneinfo/PRC /etc/localtime

# Create app directory
RUN mkdir -p /usr/src/app
WORKDIR /usr/src/app

# Install app dependencies
#COPY package.json /usr/src/app/

# copy app source
COPY . /usr/src/app

RUN npm install

CMD ["node","/usr/src/app/main.js"]

main.js

console.log("main")

package.json writes all modules used in the project into this file. In order to avoid errors caused by the automatic module update, a certain version is used.

{
  "name": "train-outline",
  "version": "0.0.1",
  "description": "Learn node.js and express.",
  "main": "main.js",
  "author": "Zhang Chao",
  "license": "MIT",
  "dependencies": {
    "body-parser": "1.17.2",
    "cookie-parser": "1.4.3",
    "express": "4.15.3",
    "jade": "1.11.0",
    "multer": "1.3.0",
    "superagent": "3.5.2"
  }
}

After the local compilation runs normally, upload the entire node-base folder (except node_modules) to the linux server, enter the node-base folder in linux through the cd command, and enter the following command to compile the image:

docker build -t zhangchao/nodejs-base:v1 .

Then I write an application example to try it out.

nodejs-base
  │
  ├─.npmrc
  ├─Dockerfile
  ├─package.json
  ├─server.js
  ├─test.jade
  └─yarn.lock

.npmrc

sass_binary_site=https://npm.taobao.org/mirrors/node-sass/
phantomjs_cdnurl=https://npm.taobao.org/mirrors/phantomjs/
electron_mirror=https://npm.taobao.org/mirrors/electron/
registry=https://registry.npm.taobao.org

Dockerfile

FROM zhangchao/nodejs-base:v1

WORKDIR /usr/src/app

# copy app source
COPY . /usr/src/app

EXPOSE 18091
CMD ["node","/usr/src/app/server.js"]

package.json

{
  "name": "train-outline",
  "version": "0.0.1",
  "description": "Learn node.js and express.",
  "main": "main.js",
  "author": "Zhang Chao",
  "license": "MIT",
  "dependencies": {
    "body-parser": "1.17.2",
    "cookie-parser": "1.4.3",
    "express": "4.15.3",
    "jade": "1.11.0",
    "multer": "1.3.0",
    "superagent": "3.5.2"
  }
}

server.js

// var express = require('express');
// var http = require('http');
// var app = express();
// app.set('view engine','jade'); // set template engine
// app.set('views', __dirname); // Set template relative path (relative to current directory)

// app.get('/', function(req, res) {
// res.render('test'); // call the test.jade template under the current path
// })

// var server = http.createServer(app);
// server.listen(3002);


//----------------------------------
var super_request = require('superagent');
var express = require('express');
var app = express();

app.set('view engine', 'jade'); // Set the template engine
app.set('views', __dirname);  // Set the template relative path (relative to the current directory)

app.use(express.static('public'));// static file


// begin 
app.get('/', function (req, res) {
    var md = "";

    md = "Hello world !!!"
    res.render('test',{
        s:'hello world!',
        md:md
    });

})



var server = app.listen(18091, function () {

  var host = server.address().address
  var port = server.address().port

  console.log("Application example, access address is http://%s:%s", host, port)

})

test.jade

doctype html
html
    head
        meta(charset="utf-8")
        meta(http-equiv="X-UA-Compatible", content="IE=edge")
        meta(name="viewport", content="width=device-width, initial-scale=1, maximum-scale=1, user-scalable=0")
        title #{s}
    body
        div #{md}

After writing, remove the node_modules folder of the project, upload it to the Linux server, use the cd command to enter the project folder, and execute the following command:

docker build -t zhangchao/app:v1 .

docker run --name app -p 18091:18091 --restart=always -d zhangchao/app:v1

In this way, when only changing the source code without changing package.json, just recompile the app image. The nodejs-base image needs to be recompiled only when package.json needs to be updated.
If the production environment requires isolation from the external network, first find a machine that can connect to the external network to compile the nodejs-base image, and then install it in the production environment.

This deployment method is more suitable for products or projects that require frequent updates.