Dockerizing a dev environment

- 550 words -

⚠ This article is posted here so that I can find it back when I want to but it is just some rough note taking for myself. If you are not me, you probably have no interest in reading what's next.

Step 1 🔗

Expose python --version to the shell

FROM python:2

WORKDIR /usr/src/app

COPY pip_requirements.txt ./

RUN pip install --no-cache-dir -r pip_requirements.txt

CMD [ "python", "--version" ]

# Create the image
sdocker build -t dpython .

# Run the image and get the result of python --version
sdocker run -it --rm --name dpython dpython
# -> Python 2.7.18

# Create an alias to run the container
alias dpython='sdocker run -it --rm --name dpython dpython'

Step 2 🔗

The container stops running immediately so we can't use docker exec on it.
Solution: Never ending entry point.
Not sure this is the best solution, check https://devopscube.com/keep-docker-container-running/ for other options.

FROM python:2

WORKDIR /usr/src/app

COPY pip_requirements.txt ./

RUN pip install -r pip_requirements.txt

ENTRYPOINT ["tail", "-f", "/dev/null"]

# Create the image with the right name
docker build . -t provisioning/tools

# Create a container
docker run -d -t --name provitools provisioning/tools

# Run a test command in the container
docker exec -ti provitools sh -c "echo a"
# -> a

docker exec -ti provitools sh -c "python --version"
# -> Python 2.7.18

Step 3 🔗

Adding docker compose (maybe too early but at least it will be there)

docker-compose.yml

version: '3.7'

services:
    provitools:
        container_name: provitools
        image: provisioning/tools

build.sh

#!/usr/bin/env bash
sudo docker build . -t provisioning/tools
sudo docker-compose build

start.sh

#!/usr/bin/env bash
sudo docker-compose up -d

# Create alias
alias provitools='docker exec -ti provitools sh -c '

# Run command in container
provitools "python --version"
# -> Python 2.7.18

Step 4 🔗

Add access to provisioning directory

version: '3.7'

services:
    provitools:
        container_name: provitools
        image: provisioning/tools
        volumes:
          - /home/adrien/projects/provisioning:/provisioning

alias ,provitools='sdocker exec -ti provitools sh -c '

touch test.file

,provitools ls
# -> Shows the test.file

Step 5 🔗

Install jq dependency. Because the base image was python:2 the util apt-utils is not installed and that makes apt command fail. So we need to install it first

FROM python:2

WORKDIR /provisioning

# This is necessary because without apt-utils the apt command doesn't install packages
RUN apt-get update && apt-get install -y --no-install-recommends apt-utils

RUN apt update
RUN apt install -y jq

COPY pip_requirements.txt ./

RUN pip install -r pip_requirements.txt
# Yes this is the same command twice. Yes this is necessary. No do not delete it.
RUN pip install -r pip_requirements.txt

# CMD [ "bash", "--version" ]
ENTRYPOINT ["tail", "-f", "/dev/null"]

Step 6 🔗

We can run ansible playbook. And (somehow?) aws is installed on the container 🤷

But it fails because aws doen't work.

First try: Share the ~/.aws directory by adding this to docker-compose.yml

        volumes:
          - /home/adrien/projects/provisioning:/provisioning
          - /home/adrien/.aws:/root/.aws

But aws say it was not configured. So we need to share some environment variables like AWS_PROFILE and AWS_DEFAULT_PROFILE.

So we update the alias:

alias ,ansible-playbook='sdocker exec -e AWS_PROFILE=$AWS_PROFILE -ti provitools ansible-playbook '

Step 7 🔗

I am blocked because I need to forward my ssh agent.

I added -e SSH_AUTH_SOCK=$SSH_AUTH_SOCK to my alias but the directory is random one with each new invocation of the agent.

So I need to fix that.

This provides a solution but my SSH_AUTH_SOCK variable is empty

← Posts

No other posts in the category [docker]

The stuff I do