Image source: by Wei Chu

Table:

• Step 00 | Prerequisite
• Step 01 | Creating an express app with node.js for testing the container function
• Step 02 | Setting up Docker container
• Step 03 | Container and Network Traffic Management
• Step 04 | Starting a exited container
• Step 05 | .dockerignore to enhance security
• Step 06 | Docker Bind Mount for Directory Files Synchronisation to the Container Node
• Step 07 | Automatically Restarting the Node Process with Nodemon
• Step 08 | Filtering Cached Item
• Step 09 | Restricting directory management privilege of the container node to avoid the main work directory being corrupted
• Step 10 | Environment variables
• Step 11 | Cleaning up redundant volumes built-up
• Comments

Step 00 | Prerequisite

1. Installing npm

   bash

2. Installing the node.js module

   bash

Step 01 | Creating an express app with node.js for testing the container function

1. Using the npm initto create a new package.json file for a Node.js project.

   bash

2. We will see a package.json file is now created in the directory.

   package.json

3. creating the express app file index.js for testing.

   index.js

Step 02 | Setting Docker container

1. Installing docker on the local machine .
2. Go to hub.docker.com , search “node” .
3. As the default image of node in the hub won't have every dependencies we will need for our app, we are going to write our own customised image but based on the image that is shown on the hub.
4. Creating a Dockerfile, . ./ will cache all the docker build result, which will allow speed up next time when run the image build again.

   Dockerfile

5. Run the command to build the docker image

   bash

   
6. Check the now existed docker image to see if it has successfully created.

   bash

   
7. (Optional) you can also remove the image by specifying the image ID.

   bash

   
8. (Optional) you can rebuild the image again but specifying a name for the docker image this time by adding a flag -t in the command.

   bash

   Now go ahead and run it.

   bash

   Double check.

   bash

   Go to the browser and check it, the website should be not working just yet . Go to the next major step about the port and how network can talk to the container.

Step 03 | Container and Network Traffic Management

bash

Step 04 | Starting a exited container

1. The following command will only show the container is currently running.

   bash

   
2. The following command will show all the containers including the one that is not active.

   bash

   
3. Run the following command start running the exited container again.

   bash

   

   bash

   

   💡 If you want to start the container and attach to it, you can use:

   bash

   💡 If you need to run a new container from an image, you can use the docker run command:

   bash

Step 05 | .dockerignore to enhance security

Run the following commands, and we will find that the dockerfile is also included in the deployed container, which is not the safest practice:

1. Run the follow command to enter the shell of the container node.

   bash

2. Run the command ls in the instance shell to check the files that has been synced from the working directory to the container node. We will find the Dockerfile and node.js modules are all synced, which is not necessary for those files existed in there and potentially contain security risk.
3. This is when .dockerignore file comes in handy.

   .dockerignore

4. After creating the .dockerignore file and specifying the directory item to be ignored, rebuilding the image and then re-deploy the container.
   1. Stop the container by running command sudo docker stop node-app
   2. Remove the container by running command sudo docker rm node-app
   
   3. Run the command sudo docker image ls -a to get the image ID and run the command sudo docker image rm cec14ac641d6 to remove the image.
   
   4. Run the command sudo docker build -t node-app-image . to rebuild the image again
   
   5. Run the command sudo docker run -p 3000:3000 -d --name node-app node-app-image to redeploy the container again
   6. Run the command sudo docker exec -it node-app bash to enter the shell of the container, and we will find the items in the .dockerignore file are now ignored while deploying the container.
   

   💡 We might still be able to find the node_modules in the container's directory even though we asked the modules to be ignored. It is because we specify npm to install the dependencies into the container in the dockerfile (see the screenshot below).

Step 06 | Docker Bind Mount for Directory Files Synchronisation to the Container Node

1. Edit the frontend content in the express app definition file.
2. We will find the content still looks the same and updated. That means the frontend app file in the work directory is not synced to the frontend app file in the container node.
3. Question is how to not constantly rebuild the container image when something is changed ? - This is when “Bind Mount” comes in handy.

   💡 What is Bind Mount

   A bind mount in Docker is a way to mount a directory or file from the host machine into a container. This allows the container to access and modify files on the host system. Bind mounts are useful for development, as changes made on the host are immediately reflected in the container.

   The following is the command syntax to set up the bind mount mechanism between the working directory and the container node.

   bash

   According to my case, it would be :

   bash

   To avoid the syntax become too long, it is better to specify the path in the dynamic way as normally docker only accept the full path address and doesn't like relative path.

   bash

   Stop and remove the existing container node and redeploy a container with the command syntax above. The new container now should be in sync with the host work directory. Remember to restart the container node process when frontend content is updated.

Step 07 | Automatically Restarting the Node Process with Nodemon

Using together with bind mount, when frontend content has been updated and synced to the container node, restarting the node process will be required. In order to prevent from manually running commands to achieve the task, Nodemon will be setup together with Bind Mount.

1. We need to install nodemon package with npm. Run the command sudo npm install nodemon --save-dev. The flag --save-dev in the command indicates that the package should be added to the devDependencies section of your package.jsonfile. Development dependencies are only needed during development and not in production.
2. We will find the devDependencies section is created in the package.json file
3. We need to adding few scripts in the package.json file to make nodemon works properly.
4. Update the Dockerfile
5. Repeat the steps in the previous sections to rebuild the image.(Stop&Remove the running node >> Remove the image with ID >> Rebuild the image )
6. As package.json is updated, therefore the Bind Mount will need to be conducted again when deploying the container node. The command: sudo docker run -v $(pwd):/app -p 3000:3000 -d --name node-app node-app-image
7. Now the frontend content should be in sync between the work directory and container node without manually starting the node processes.

Step 08 | Filtering Cached Item

Context:

1. Deleting the node_modules package hosted in the work directory as those node.js package are no longer needed after we containerise the dependency of the app.
2. We will then find out the web app is crashed if we will deploy the container node again in the future. This is due to the caching specified in the Dockerfile. Which means that even though npm will install packages in the new container node, the caching means the work directory structure will still override the directory cached in the container node.

Fix: Excluding a specific part of the volume from being overwritten by using the following command:

bash

-v /app/node_modules creates an anonymous volume for the /app/node_modules directory to avoid overwriting it with the host's node_modules.

Step 09 | Restricting directory management privilege of the container node to avoid the main work directory being corrupted

Context:

1. After the container node has being deployed, you will find that it is possible to create, update or delete files from the container node side because of the bind mount synchronisation
2. For example, if we enter the shell of the containder node, and use the command touch test.txt to create a new file. We will find the new file immediately is synced to the root work directory. This will make the root work directory risks from being edited or corrupted.

Fix: Re-deploy the image by specifying read-only privilege with the following command:

bash

Step 10 | Environment variables

1. Setting default ENV value
   for example: setting port number at 3000 in the Dockerfile, and then reference the ENV variable PORT by EXPOSE for documentation purpose.
2. Now killing the existing container and re-build the image. Re-deploy the container after the image is rebuild the with the environment variable referenced in the command line.
   The command:

   bash

   
3. To check if the environment variables are parsed into the container, enter the container shell and run the command printenv:
4. To set multiple environment variables with their own values, we need set up a .env file and reference it in the container deployment command.
   1. Create an .env file
   2. Delete the existing container and redeploy it with the following command with .env file specified to reference the desired environment variable values.

      bash

   3. Run the command printenv again in the container bash shell to check if the environment variables are parsed into the container.

Step 11 | Cleaning up redundant volumes built-up

When a container is removed, some of the volumes attached to it will not be removed with it. In our case, we specify volume /app/node_modules when deploying a container, which is a anonymous volume that won't be deleted when a container is removed.

1. Check the exiting redundant volumes
   Using the command docker volume ls to check what are the volumes existing in our system:
2. There are 2 methods to handle them:>
   1. Using the command docker volume prune to trim the redundant volumes.
   2. Using -fv flag when removing a container, which will have the attached volume deleted together with the container to prevent volumes from building up.