me-sajib/docker-playground

Docker Playground

• Docker Playground
  • Installing Docker
  • Architecture and Theory
    • Kernel primitives
    • Docker Engine
    • DE for windows
  • Working with Images
    • Docker images
      • Manifest file
    • Registries
      • Best practices
  • Containerizing an App
    • Discussion
    • Containerize an App
    • Dig Deeper
    • Multi-stage Builds
  • Working with Containers
    • Container & logging
      • Engine/Daemon logs
      • Container/App logs
  • Building Secure Swarm
    • The big picture
    • Swarm clustering deep dive
    • Building a secure swarm
    • Orchestration
  • Container Networking
    • The three pillars of Docker Networking
    • Use Cases & Drivers
      • Single Host Networking (with the bridge driver)
      • Multi-host networking
      • Working with Existing networks
  • Working with volumes and persistent data
  • Working with secrets

Installing Docker

wget -qO- https://get.docker.com/ | sh

adding your user to the "docker" group:

sudo usermod -aG docker <your-user-name>

Architecture and Theory

Operating System, Container, Isolation, Control Groups, Namespaces

Kernel primitives

• Control Groups (C Groups / in Windows Job Objects): grouping objects and setting limit
• Namespaces: isolation.
• Container
  • Process ID (pid)
  • Network (net)
  • Filesystem/mount (mnt)
  • Inter-proc comms (ipc)
  • UTS (uts)
  • User (user)

Docker Engine

CLI -> API -> daemon -> containerd -> OCI

DE for windows

there are two types of windows containers

Attempt	Native windows containers	Hyper-V Containers
os tech	Namespace isolation	VM, but less performance than a full blown VM
kernel/os	Uses hosts kurnel	not using the host's karnet, have own OS
linux	can't have linux here	as it can have it's own karnel and OS, so we can have linux here
run	docker container run ..	docker container run --isolation=hyperv

Working with Images

Docker images

Image	Container
read only template to create application containers	a running image
build time	run time

Image: a bunch of independent layer that are very loosely connected through manifest file (or config file).

Manifest file

to play with the manifest command, you need to enable docker experimental feature on your machine. You can get help from this link.

1. create a file in .docker/config.json
2. paste the following code into the file:

{
  "experimental": "enabled",
  "debug": true
}

3. do want to now more about a manifest file? run the following command

docker manifest inspect <imagename>

4. what does docker pull <imagename> command do?

   1. get the manifest file
      1. Fat Manifest: list architecture supported and and a manifest for all of those. first it get the fat manifest to match with my system architecture config match.
      2. Image Manifest: the actual manifest for my machine. Then it get the manifest that machaes my machine architecture.
   2. Pull layers

5. Content addressable storage

   docker image ls --digests

6. To see the docker system info:

   docker system info

7. The file system inside a container will be the base layer file system, not the host machine's file system.

8. do you want to see the layer history of an docker image?

   docker history <imagename>

9. do you want to know the details of a docker image?

   docker image inspect <imagename>

10. and finally you may want to free up space by deleting any unused image?

    docker image rm <imagename>

Registries

Images line in regitries. Within the registries we have repos. Within repos, we have images or tags. registry/repo/image(tag) Along with many public registries other than docker hub, we can have our own private docker registry.

repo:latest

latest tagged repo: does nto alway means that it is the later version of the repo. To make a repo latest, we have to explicitely mention while creating the image.

mehdihasan/pokemon

you may need to add name of the user or the organization while dealing with a repo because they are not official version of any image.

distribution hash / cotent hash

Best practices

1. using official images
2. keep the images small
3. explicitely reference of the image version, not just only latest.

Containerizing an App

Discussion

Dockerfile: where developers describe their apps and how they work, and for ops to read the file and understand.

Containerize an App

1. Instructions for building images

   • CAPITALIZE instructions
   • <'INSTRUCTION'> <'value'>
   • FROM = base image
   • Good practice to list maintainer
   • RUN = execute command and create layer
   • COPY = copy code into image as new layer
   • Some instructions add metadada instead of layers
   • ENTRYPOINT = default app for image/container

2. create a Dockerfile

3. Build your image

docker image build -t `myAppNameTag` .

4. running the image

docker image ls

docker container run -d --name `nameYourContainer` -p `host-port`:`container-port` `myAppNameTag`

Dig Deeper

Build Context: Location of my code in my machine.

It is possible to build from git repo as well! Run the following command to run from git repo:

docker image build -t thesaurus https://github.com/mehdihasan/springboot-thesaurus-app.git

Multi-stage Builds

Very interesting topics! The concept is to build my expected image step by step. In the last stage, may be the production build, it will going to take only the output layers from the each stage, and build the final image out from those! Do you know why do they put the Dockerfile into the app directory other than the root directory?

example

Working with Containers

the most atomic unit - container. in k8s - pod.

1. here -it stands for interactive terminal

docker container run -it alpine sh

2. Here what is -d and sleep 1d doing?

docker container run -d alpine sleep 1d

3. You might want to run a single command inside a container. And you are lazy enough to avoid get inside the container using the -it command.

docker container exec <CONTAINER-ID> ls -al
docker container exec <CONTAINER-ID> cat newFile

4. do you want to remove all your containers?

docker container rm $(docker container ls -aq) -f

Container & logging

Engine/Daemon logs

Linux

• systemd
  • journalctl -u docker.service
• non systemd
  • try /var/log/messages

Windows ~/AppData/Local/Docker

Container/App logs

• STDOUT, STDERR
• It is possible to collect the docker logs and forward those to existing logging solution system like Syslog, Gelf, Splunk.
• we can set default logging driver in daemon.json
• it is possible to override per container with --log-driver & || --log-opts
• inspect logs with docker logs <container>

Building Secure Swarm

He just showed how to create a swarm and then create some nodes. I was not sure what did me ment by nodes. Probably different machines into same network.

The big picture

Swarm clustering deep dive

Building a secure swarm

• check if docker swarm is enabled

  docker system info

• initialize docker swarm with the following command

  docker swarm init

• check the command to add manager to the docker swarm

  docker swarm join-token manager

• list out all docker nodes

  docker node ls

Orchestration

Container Networking

The three pillars of Docker Networking

Must read item for docker networking: Docker networking Grand Design DNA

1. CNM (Container Network Model)
2. Libnetwork
3. Drivers

You should know about CNI as well. It is about K8s networking.

Use Cases & Drivers

AGENDA

Single Host Networking (with the bridge driver)

we are goin to create a user defined bridge network in docker host.

Let's create a new network

docker network create -d bridge --subnet 10.0.1.1/24 ps-bridge

Now, I want to test the bridge. Let's install the testing kit first

sudo apt install bridge-utils

after the package installed, run the following command:

brctl show
ip link show

Now, let's run 2 containers on the bridge:

docker run -dt --name c1 --network ps-bridge alpine sleep 1d
docker run -dt --name c2 --network ps-bridge alpine sleep 1d

Now, inspect the container:

docker network inspect ps-bridge

Now, if you run the following command you will see that there are two interfaces connected to the bridge

brctl show

Now, let's get inside into one of our docker container and ping the other one:

docker exec -it c1 sh

Now, you can check own IP and ping the other instance

# ip a
# ping c2

So, we have created a network bridge in which 2 instances are connected.

Now, how can other docker instances can able to access any of the instance outsside of this network?

we need to publish a container service in a host port

To, show a demo, let's make another container running in port 8080 in the container and map to the 5000 port in the host machine

docker run -d --name web1 \
--network ps-bridge \
-p 5000:8080 \
nigelpoulton/pluralsight-docker-ci

DONE! Now you can access the web1 container by going http://localhost:5000

Multi-host networking

Working with Existing networks

Working with volumes and persistent data

using docker volume, we can take leverage of persistent data.

to know mode about the commond run the following

docker volume

to create a container with persistant data, run the following command

docker container run -dit --name voltest \
--mount source=ubervol,target=/vol alpine:latest

you can see the created volume in the host machine in the following directory

sudo ls -al /var/lib/docker/volumes/

do you want to test if file really writes to the host machine's location? TRY IT BY YOURSELF.

Working with secrets