Environment Variables not passed through

[tstackhouse]

Describe the bug
Environment variables set on the workflow step (via the env keyword) are not passed through to the command when the workflow runs.

From my workflow:

- name: Destroy Infrastructure
  id: retry_on_error
  uses: nick-fields/retry@v2
  env:
    ENVIRONMENT: dev
    NODE_OPTIONS: '--max_old_space_size=7168'
  with:
    timeout_minutes: 60
    max_attempts: 3
    retry_wait_seconds: 60
    retry_on: error
    command: npx nx destroy ${{matrix.project}} --stage=${{ steps.git-branch.outputs.name }}

Expected behavior
Environment Variables set on the workflow step are used when invoking the command, which is the standard behaviour for the default run action, that this is meant to be a drop-in replacement for.

Screenshots
N/A

Logs
I can gather logs if necessary, but I feel like this is part feature request part bug report, so I'm not sure how useful logs would be.

[tstackhouse]

I'm continuing to research my issue, and I'm half wondering if it's connected with #76, as I've tried moving my environment variables inline with my command, as well as looking into the NodeJS docs, which indicates that child_process.exec should inherit from process.env if env is not provided as part of the options when called. I didn't think it was related to the SIGPIPE indicated in that issue because my command was editing with Code 1 and Signal null according to the debug log.

[nick-fields]

I'm continuing to research my issue, and I'm half wondering if it's connected with #76, as I've tried moving my environment variables inline with my command, as well as looking into the NodeJS docs, which indicates that child_process.exec should inherit from process.env if env is not provided as part of the options when called. I didn't think it was related to the SIGPIPE indicated in that issue because my command was editing with Code 1 and Signal null according to the debug log.

Thanks for the research on this, I'm looking into it now

[nick-fields]

Just published v2.8.0 that I think should resolve this. Thanks for your issue and research!

Can you confirm whether this resolves your issue?

[tstackhouse]

I had to explicitly make my reference nick-fields/retry@v2.8.0, but it looks like that did the trick! Thanks for looking into this so quickly, and I'm glad my research was helpful! ❤️

[Norfeldt]

I tried the following

jobs:
  jest:
    runs-on: ubuntu-latest
    env:
      NPM_AUTH_TOKEN: ${{ secrets.NPM_AUTH_TOKEN }}
      MAX_ATTEMPTS: 5
      CI: true
    steps:
      - name: checkout repository
        uses: actions/checkout@v3
      - name: get yarn cache directory path
        id: yarn-cache
        run: echo "::set-output name=dir::$(yarn cache dir)"
      - name: Setup cache key and directory for node_modules cache
        uses: actions/cache@v3
        with:
          path: ${{ steps.yarn-cache.outputs.dir }}
          key: ${{ runner.os }}-yarn-${{ hashFiles('**/yarn.lock') }}
      - name: yarn
        run: yarn --frozen-lockfile
      - name: jest
        uses: nick-fields/retry@v2.8.1
        with:
          command: yarn test --ci --bail --coverage=true
          max_attempts: $MAX_ATTEMPTS
          timeout_minutes: 10

That did not work. But using github secret ${{ secrets.MAX_ATTEMPTS }} worked - but it seems a bit silly to use secrets for this.