instance_pool_ci.yml

name: Monitor the pool of WorkArena instances

on:
  workflow_dispatch:
  schedule:
    - cron: "0 3 * * *"       # daily at 03:00 UTC

jobs:

  test-l1-tasks:
    name: Test L1 tasks
    runs-on: ubuntu-22.04

    defaults:
      run:
        shell: bash -l {0}

    env:
      HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}

    steps:
      - name: Checkout Repository
        uses: actions/checkout@v4

      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: '3.12'
          cache: 'pip'

      - name: Install Python dependencies
        working-directory: ./dev
        run: |
          pip install -r requirements.txt
          pip install huggingface_hub

      - name: Pip list
        run: pip list

      - name: Install Playwright
        run: playwright install chromium --with-deps

      - name: Run L1 tests
        run: pytest -n 20 --durations=10 --slowmo 1000 -v tests/test_task_general.py


  test-snow-instance:
    name: Test snow instance
    runs-on: ubuntu-22.04
    needs: test-l1-tasks   # remove this line if you want both jobs to run in parallel

    defaults:
      run:
        shell: bash -l {0}

    env:
      HUGGING_FACE_HUB_TOKEN: ${{ secrets.HUGGING_FACE_HUB_TOKEN }}

    steps:
      - name: Checkout Repository
        uses: actions/checkout@v4

      - name: Set up Python
        uses: actions/setup-python@v5
        with:
          python-version: '3.12'
          cache: 'pip'

      - name: Install Python dependencies
        working-directory: ./dev
        run: |
          pip install -r requirements.txt
          pip install huggingface_hub

      - name: Pip list
        run: pip list

      - name: Install Playwright
        run: playwright install chromium --with-deps

      - name: Run snow instance tests
        run: pytest -n 20 --durations=10 --slowmo 1000 -v tests/test_snow_instance.py