19

In my action.yml I defined an input:

name: 'test action' author: Param Thakkar description: 'test' inputs: test_var: description: 'A test variable' required: true runs: using: 'docker' image: 'Dockerfile'

And in my workflow I passed the test_var:

name: CI on: [push] jobs: build: runs-on: ubuntu-latest steps: - name: Test the GH action uses: paramt/github-actions-playground@master with: test_var: "this is just a test"

So there should be an environment variable that's created when the workflow runs, right? But when I run this short python script:

import os print(os.getenv('TEST_VAR')) print("It works!") exit(0)

It prints:

None It works!

I think that I have to pass the ENV variable through my Dockerfile... Right now my Dockerfile looks like this: 

FROM python:latest # Add files to the image ADD entrypoint.py /entrypoint.py ADD requirements.txt /requirements.txt # Save ENV var in a temp file RUN $TEST_VAR > /temp_var # Install dependencies and make script executable RUN pip install -r requirements.txt RUN chmod +x entrypoint.py RUN echo "temp var: " RUN cat /temp_var # Run script with the ENV var ENTRYPOINT export TEST_VAR="$TEST_VAR"; /entrypoint.py

But the variable isn't echoed and isn't passed to the pythons script either.. am I missing something? When I tried to set my $TEMP_VAR to a random piece of string, it is sent through to the Python script. Is this a mistake on my behalf or is the GitHub action not working as intended?

Here's the link to the test repo

Improve this question
1
  • How were you able to access your input arg defined in action.yml inside your Dockerfile? I'm trying to do the same and I'm having some trouble with that. Commented Jan 19, 2021 at 1:24

4 Answers 4

Reset to default
13

I think you are trying to read the wrong environment variable name. GitHub Actions adds INPUT_ to the name of the input variable. So try the following:

print(os.getenv('INPUT_TEST_VAR'))

From the documentation:

When you specify an input to an action in a workflow file or use a default input value, GitHub creates an environment variable for the input with the name INPUT_. The environment variable created converts input names to uppercase letters and replaces spaces with _ characters.

For example, if a workflow defined the numOctocats and octocatEyeColor inputs, the action code could read the values of the inputs using the INPUT_NUMOCTOCATS and INPUT_OCTOCATEYECOLOR environment variables.

https://help.github.com/en/articles/metadata-syntax-for-github-actions#inputs

Improve this answer
Sign up to request clarification or add additional context in comments.

Comments

9

A bit late but for the next one, you can also use the env field :

name: CI on: [push] jobs: build: runs-on: ubuntu-latest steps: - name: Test the GH action uses: paramt/github-actions-playground@master env: test_var: "this is just a test"

which will be included during the creation of your docker and pass without the prefix INPUT_

Improve this answer

1 Comment

does that make it available inside the container or does that require another step?
5

Keep env vars secret by specifying them in Settings -> Secrets in the repo, and then calling them in the workflow:

For example, consider a workflow that runs an R script followed by a Python script. First, in .github/workflows/my_job.yml notice the MY_VAR variable, which points to a stored secret with ${{ secrets.MY_VAR}}. The rest is standard code (run on cron, specify Ubuntu OS and Docker image, define workflow steps).

on: schedule: - cron: '0 17 * * *' jobs: my_job: name: my job env: MY_VAR: ${{ secrets.MY_VAR }} runs-on: ubuntu-18.04 container: image: docker.io/my_username/my_image:my_tag steps: - name: checkout_repo uses: actions/checkout@v2 - name: run some code run: bash ./src/run.sh

Next, in the scripts that compose your workflow, you can access the env var specified in the workflow file above as you would locally.

For example, in the repo, let's assume src/run.sh calls an R script followed by a Python script.

In R access the env var and store as an object:

my_var <- Sys.getenv("MY_VAR") . . .

In Python access the env var and store as an object:

import os my_var = os.getenv("MY_VAR") . . .

See the docs here.

Improve this answer

Comments

3

Edit: This approach is not secure. I was using it on a dummy-dev project. See explanation from Sun (in the reply) and check Peter Evan's answer

However, in my case, none of the answers worked. Here's how I fixed it.

--- name: Build and Push Docker Image to AWS ECR on: push: branches: [ master ] env: FOO: '${{ secrets.FOO }}' jobs: build-and-push: name: Build Project and Push to AWS ECR runs-on: ubuntu-latest steps: - name: Checkout uses: actions/checkout@v2 ... - name: Build and Push to AWS ECR id: build-image env: ECR_REGISTRY: ${{ steps.login-ecr.outputs.registry }} run: | docker build --build-arg FOO=$FOO -t $ECR_REGISTRY/crew-charge-app:latest . docker push $ECR_REGISTRY/crew-charge-app:latest

I first had to get the FOO variable from github secrets using ${{secrets.FOO}} then pass it onto the docker file using docker build --build-arg FOO=$FOO --build-arg BAR=$BAR -t .

Then inside the docker file I had to declare both as an ARG and ENV to be available at all times.

FROM node:14 ARG FOO=${FOO} ENV FOO=${FOO} RUN mkdir -p /usr/src/app WORKDIR /usr/src/app COPY package.json /usr/src/app RUN yarn install COPY . /usr/src/app RUN FOO=$FOO yarn build EXPOSE 80 CMD ["yarn", "start" ]

Important part was to RUN FOO=$FOO yarn build because setting the ENV alone doesn't pass it onto the container.

Improve this answer

2 Comments

Never do this. this is insecure in the case where FOO supposed to be secret value --build-arg FOO=$FOO means you will be able to see the cached secret when any user who has access to image can run the command docker history <imagename>
I presume this is perfectly safe in cases where FOO is not an important value?

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.