What's this?

A summary of freeCodeCamp - Docker Tutorial for Beginners - A Full DevOps Course on How to Run Applications in Containers, which you can watch here: https://youtu.be/fqMOX6JJhGo

The author has a very good approach, is a quick process explainer and has a course website https://kodekloud.com

Introduction

Why you need Docker?

No more long setup ENV or VM time
Different ENVs without a headache
No more issues about compatibility between OS versions, libraries, dependencies
Compared to a VM: quicker, less resources expensive

What are containers?

Completely isolated ENV with it your own network interfaces and mounts (volumes)
Shares same kernel

The methodology exists for a long time, docker came and turns it easier.

How's it done?

You have Docker Registry Hub (https://hub.docker.com) which has a large database of images ready to use like:

docker run ansible
docker run mongodb
docker run nodejs

Docker in DevOps

Developers can inject the infrastructure inside the codebase, linking the development and the ops. It's not required anymore write a manual on how to install/boot the application.

The main: docker run

It runs any image and creates a container. First try do find locally and if not exists, download it.

When you run some image as this, you will be in the foreground or attached mode and you will see the console output. You can't interact with terminal besides see the log output.

You will be on this console forever until you quit the container (CTRL+C) or the application which you run exits.

Running the container in background (dettached mode) you need to append the -d flag: docker run -d image. Container will run in background and the terminal prompt from your computer will be back to you.

If you ran the some container in the dettached mode and want to go to attached mode you can do this docker attach container_name.

Run in interactive mode

If you want to run the container in interactive mode, where you can see the console questions and can answer them you need to run as docker run -it some-image

Environment Variables

docker run -e APP_COLOR=green image/example

You can see the all the environment variables in use for the container using docker inspect

Notes

When you run docker run ubuntu. Ubuntu will boot up but it will exit immediately. If you run docker ps you will see that the container was exited.

Containers are made to boot up and run something. A task, service or something related. And not to boot up and wait for external commands as a VM is.

When the task finishes the container exits.

One example to run an Ubuntu image and just wait for a few seconds: docker run ubuntu sleep 5.

Port mapping

Every container receives an IP by default but it's only reachable from the docker host. It's locked from external access.

To allow access to some port inside the container, you need to map some port from docker host to the container.

docker run -p 80:5000 image/example

This tells to docker map the port 80 from docker host to the port 5000 inside the container.

Volume mapping

Everything wrote inside the container, which is outside a volume, will be lost when you delete the container.

You need to map a volume inside container to a place inside the docker host to persist the data.

This will auto-mount the volume inside the container: docker run -v /some/path/docker-host:/var/lib/mysql mysql

Networking

When docker installs, three networks are created: Bridge, none and host

To specify a network to use with the run command you need to: docker run ubuntu --network=bridge|none|host

Bridge

This is the default for every container
Every container can reach other containers inside this network
To access the services inside this network you map using Port Mapping

Host

Uses the same network of the host
Doesn't require do map ports as you can access everything inside the host

None

Doesn't use any network
There's no access between the other containers and from external network

Create your own network

docker network create \
  --driver bridge \
  --subnet 182.18.0.0/16 \
  custom-isolated-network

You can list all networks using docker network ls.

Connect to services container-to-container

Docker has a built-in DNS which helps you connect between the containers using the container_name.

Example: mysql.connect('container_mysql')

Storage

Layered architecture

Used in builds
Each layer stores only the difference from the before-layer. Even if you have more than one Dockerfile, Docker always will use the same layer for them.
These layers, used on builds, are read-only
After container boot up, a new layer is created read-write and will contain all the data from the container
When something changes any file inside the container which belongs to the image layer, a COPY-ON-WRITE is made

Volumes

Docker has a special storage for volumes which you can create inside your docker host (/var/lib/docker/volumes). This is called Volume Mounting

It's different from Bind Mounting, where you can points to any place on docker host

You can create these volumes as this: docker volume create data_volume

To use: docker run -v data_volume:/var/lib/mysql mysql

You can create the volume on the run command as well. If the volume doesn't exist, it will be created.

Example: docker run -v data_volume_which_doesnt_exist:/var/lib/mysql mysql

More verbose way to mount the volumes

docker run \
  --mount type=bind,source=/docker-host/folder,target=/var/lib/mysql \
  mysql

Creating you own image

You need to create a Dockerfile to create your own image, example:

FROM Ubuntu

RUN apt-get update
RUN apt-get install python

RUN pip install flask
RUN pip install flask-mysql

COPY . /opt/source-code

ENTRYPOINT FLASK_APP=/opt/source-code/app.py flask run

After you create the Dockerfile, you need build: docker build Dockerfile -t username/my-custom-app
To push to Docker Hub: docker push username/my-custom-app

Understanding the Dockerfile

Everything at the left and in uppercase is an instruction
It always needs to start with a FROM instruction (always should inherit some image already created)
Each line of instruction is a layer (helps in the build)
- In case of issue, the previous layers doesn't need to be repeated
- As in the example you are using a COPY instruction, only this layer will be updated when you are working with your code and building the image

CMD vs ENTRYPOINT

Every image by default has a CMD or ENTRYPOINT instruction inside Dockerfile. This instruction tell to Docker what to do when boot up.

For example, Nginx image does: CMD nginx, Ubuntu does CMD bash.

You can change the default instruction of any image just using a combination of FROM + CMD/ENTRYPOINT. Example:

FROM <image>
CMD ['my-new-command', 'my-new-param1']

You can override the image default CMD using docker run. Example: docker run image command-to-execute-instead-the-default

Accept a param from docker run

If you want to accept a param from docker run, you need to use ENTRYPONT instruction. Example:

FROM Ubuntu
ENTRYPOINT ["sleep"]

Run as: docker run ubuntu-sleeper 10

This will boot Ubuntu and run the sleep command waiting for 10 seconds.

Different from CMD, ENTRYPOINT accept a param from docker run and attaches to the default

You can define a default param when docker run doesn't define one. Example:

FROM Ubuntu
ENTRYPOINT ['sleep']
CMD ['5']

CMD needs to be use JSON format.

You can override the ENTRYPOINT too.

Example: docker run --entrypoint <new-command> image <param>

Docker Compose

Used to work with more than one service/image at the same time in the same project.

Overview

You always have to define a version
It creates by default the network to use inside the stack (to be used by all services)

Docker compose example file

version: 2
services:
  <container-name>:
    image: redis
  <another-container-name>:
    image: mysql
  <another-container-name-two>:
    image: nginx
    ports:
      - 5000:80
    depends_on:
      - db
    networks:
      - backend
networks:
  - backend:

To boot the entire stack you need to run docker-compose up

You can use the Dockerfile of some image and instruct docker-compose to use the correct file to build. Example:

services:
  <app-name>:
    build: ./app # path where the Dockerfile lives

Docker Engine

It's possible to run docker-cli accessing a remote docker
A container is in fact a process in docker host machine with a unique PID, but inside the container-process-tree it starts with PID 1
You can limit some resources inside the container
- docker run --cpus=.5 ubuntu uses max 50% of the CPU from host
- docker run --memory=100m ubuntu

Docker Registry

By default, any image you get is inside docker.io. Example: docker run nginx is in fact docker run docker.io/nginx/nginx

When you type only nginx, it's converted to nginx/nginx which is <user-account>/image-name.

You can work with private registries like AWS, Azure, GCP...:

You need to log in docker login private-registry.io
Install as docker run private-registry.io/apps/internal-app

Docker Orchestration

It helps to create/reboot containers when they stop in case of failure, or the traffic gets high.

Applications

Docker Swarm (less features)
Kubernetes (most popular)

Another useful commands

docker ps list running containers and some information about them
docker ps -a list all containers, even exited ones
docker stop container_name stops a container
docker rm container_name deletes the container permanently
docker images list all downloaded images
docker rmi nginx remove the image. Make sure you have no containers running which being using the image.
docker pull nginx only downloads the image without run the container
docker exec container_name <command> run a command inside the container
docker inspect container_name show s more information about the container (state, volumes, network, env vars)
docker logs container_name when you are in detached mode and want to see logs from container

In any place where you need to insert container name, you can use only a partial name, which should be different from other containers, and will be enough to docker knows which container you are trying to reach.

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
.gitignore		.gitignore
README.md		README.md

willjosefi/freecodecamp-docker-course-summary

Folders and files

Latest commit

History

Repository files navigation