Skip to content

Latest commit

 

History

History
109 lines (77 loc) · 3 KB

03-docker.md

File metadata and controls

109 lines (77 loc) · 3 KB
Raw

The Docker container engine: Hands on

  • First clone the cna-code repo from github
git clone http://github.com/intracom-telecom-sdn/cna-code.git
cd cna-code

The first component of our cloud native application is a basic HTTP server implemented in server/main.go.

This server exposes 2 REST endpoints

  • /doWork: The server handles an incoming request to do a unit of work
  • /stop: The server exits.

(Part 1/3): Deploy server instance on baremetal

  1. Run server application

    go run server/main.go

  2. Perform a GET request using CURL and observe the result

    curl -X GET http://localhost:4444/doWork

  3. Try to run a second server on baremetal. You should get an error because the port the server is trying to bind is already in use. You can bypass this by making the user specify the port, but what about other resources? (config files, log files, system side effects)

  4. Now stop the server

    curl -X GET http://localhost:4444/stop

(Part 2/3): Deploy server instance with docker

  1. Open server/Dockerfile. This file is a template for a docker image to deploy and run our server application.

  2. Build the server image with

    docker build -t cna-server:latest server/

  3. Verify cna-server image exists with

    docker images

  4. Deploy a server instance container using cna-server:latest image. The -p argument maps the internal port of the container 4444 where the server is listening to port 8000 in the host

    docker run -ti -p 8000:4444 --name server1 cna-server:latest

  5. Verify container is up

    docker ps

  6. Inspect server1 container. Search for IPAddress field to get the server IP

    docker inspect server1

  7. Make server1 do some work with CURL using the IPAddress from step 6 and port 4444 (don't /stop the server)

  8. Make server1 do some work with CURL using localhost and port 8000

  9. Repeat steps 4-8 to create and test a new server instance named server2. Map to host port 8001.

  10. Delete server2 instance

docker rm -f server2

(Part 3/3): Call Docker API programmatically

Here we'll learn how to interact programmatically with Docker. This is the first step towards building an orchestrator that will provide some elasticity to our application.

We provide the boilerplate in Python.

We'll give you ~20 minutes for the coding part.

  1. Implement the TODOs in orchestrator/python/docker_utils.py. The functions in this module wrap the calls to Docker programmatic API and allow us to manage cna-server containers in an elastic and dynamic way

  2. Test the implementation inside Python REPL (bpython).

Conclusions

  • We deployed and tested an application on baremetal. Does the trick, but somewhat inflexible.
  • We deployed and tested the same application using Docker. We saw that Docker streamlines the deployment and destruction of application and makes it 100% reproduciple.