Riding The Whale

@jochen_christ

IT Consultant && Software Developer

https://www.docker.com/what-container

DEV = TEST = PROD

Container vs VM

https://www.docker.com/what-container

Container = OS Virtualization

VM = Hardware Virtualization

Linux & Windows

Images and Containers

docker pull ubuntu
docker images

docker run -it --rm ubuntu bash

# Now you can do things in your container
root@ada3108b4d6b:/# ps -auxwww
root@ada3108b4d6b:/# apt-get update && apt-get -y install curl

# CTRL-PQ -> Back to host
docker ps
docker stats
docker stop 552e864fbe70

Quit Terminal: Crtl-PQ

Stop Container: Stop Process (exit) or docker stop

docker run -d -p 27017:27017 --name mymongo --rm mongo

docker ps

# Run commands in running container
docker exec -d mymongo touch hello
docker exec -it mymongo bash
docker exec -it mymongo mongo

docker run -d -p 8081:8081 --link mymongo:mongo --rm mongo-express

open http://localhost:8081

https://docs.docker.com/engine/docker-overview/#docker-architecture

Let's Build a Full-Stack Application

Web-App

EmberJS frontend served by nginx

Service

Spring Boot Java Application

Database

MongoDB

FROM openjdk:8-jdk-alpine

# Add files to image
ADD build/libs/simple-todo-service-0.0.1-SNAPSHOT.jar app.jar

# expose a port (internally!)
EXPOSE 8080

# Specify where to find the mongo host
ENV SPRING_DATA_MONGODB_HOST mongo

# Command when container runs
# Use exec forms (array-style) to run as PID 1
ENTRYPOINT ["java", "-jar", "/app.jar"]

Dockerfile (for Java Application)

Build Command 

git clone https://github.com/jochenchrist/simple-todo-service.git
docker build -t simple-todo-service .
docker run -d -p 8080:8080 --link mymongo:mongo --rm simple-todo-service

# Base image
FROM nginx:alpine

# Add the web application
ADD dist/ /usr/share/nginx/html

# A custom nginx configuration is required for single page applications
ADD default.conf /etc/nginx/conf.d/default.conf

Dockerfile (for Web Application)

Build Command 

git clone https://github.com/jochenchrist/simple-todo-web.git
docker build -t simple-todo-web .
docker run -d -p 80:80 --rm simple-todo-web

Good Practices for Dockerfile

• Run only one process in a container
• Keep images as small as possible
• Minimize the number of layers
  • Combine statements
  • apt-get update && apt-get install -y curl git
• Use the cache (reuse layers!)
• Process (with PID 1) should stop gracefully with SIGTERM

Web-App

EmberJS frontend served by nginx

Service

Spring Boot Java Application

Database

MongoDB

Now we have a Full-Stack Application

Docker Compose

# docker-compose.yml
version: '3'
services:
  web:
    image: "simple-todo-web"
    ports: 
     - "80:80"
  service:
    image: "simple-todo-service"
    ports: 
     - "8080:8080"
    networks:
     - database
    depends_on: 
     - mongo
  mongo:
    image: "mongo"
    networks:
      - database
  monog-express:
    image: "mongo-express"
    networks:
      - database

networks:
  database:

docker-compose up -d

Docker Hub

Docker Hub

• Hosted Docker Registry
• Default
• Public and Private (paid) Repos
• Tight Integration with Github

docker login

docker tag todo-web jochenchrist/simple-todo-web
docker push jochenchrist/simple-todo-web

docker tag todo-service jochenchrist/simple-todo-service
docker push jochenchrist/simple-todo-service

Docker Swarm

for i in 1 2 3; do \
docker-machine create --driver digitalocean \
--digitalocean-image ubuntu-16-04-x64 \
--digitalocean-access-token $DOTOKEN \
--engine-opt experimental=true \
devnight-node-$i \
; done

for i in 1 2 3; do \
docker-machine ssh devnight-node-$i ufw disable \
; done

docker-machine ssh devnight-node-1 \
docker swarm init --advertise-addr $(docker-machine ip devnight-node-1)

# follow the join instructions for node-2 and node-3

eval $(docker-machine env devnight-node-1)
docker deploy --compose-file docker-compose-swarm.yml simple-todo-app

Facts

• Apache 2.0 Licence
• OCI compliant
• Written in Go
• libcontainer as core technology
  • namespaces (isolation)
  • control groups (resource limits)
  • UnionFS (layered file system)

Wait!

Don't use it in production!

Think about...