lightning/.github/workflows/tpu-tests.yml

name: Test PyTorch - TPU

on:
  push:
    branches: [master, "release/*"]
  pull_request_target:
    branches: [master, "release/*"]
    types: [opened, reopened, ready_for_review, synchronize]  # add `ready_for_review` since draft is skipped
    paths:
      - ".github/workflows/tpu-tests.yml"
      - "requirements/pytorch/**"
      - "src/pytorch_lightning/**"
      - "tests/tests_pytorch/**"
      - "setup.cfg"  # includes pytest config
      - ".actions/**"

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref }}
  cancel-in-progress: ${{ ! (github.ref == 'refs/heads/master' || startsWith(github.ref, 'refs/heads/release/')) }}

env:
  PROJECT_ID: ${{ secrets.GKE_PROJECT }}
  GKE_CLUSTER: lightning-cluster
  GKE_ZONE: us-central1-a

jobs:
  # TODO: package parametrization
  test-on-tpus:
    runs-on: ubuntu-22.04
    if: github.event.pull_request.draft == false
    env:
      PYTHON_VER: 3.7
    timeout-minutes: 100  # should match the timeout in `tpu_test_cases.jsonnet`

    steps:
    - uses: actions/checkout@v3
      with:
        ref: ${{ github.event.pull_request.head.sha }}

    - uses: actions/setup-python@v4
      with:
        python-version: ${{ env.PYTHON_VER }}

    - name: Checkout ml-testing-accelerators
      run: |
       git clone https://github.com/GoogleCloudPlatform/ml-testing-accelerators.git
       cd ml-testing-accelerators
       git fetch origin 5e88ac24f631c27045e62f0e8d5dfcf34e425e25:stable
       git checkout stable

    - uses: actions/setup-go@v3
      with:
        go-version: '1.19'

    - name: Install jsonnet
      run: go install github.com/google/go-jsonnet/cmd/jsonnet@latest

    - name: Update jsonnet
      env:
        XLA_VER: 1.12
        PR_NUMBER: ${{ github.event.pull_request.number }}
        SHA: ${{ github.event.pull_request.head.sha }}
      run: |
       python -c "fname = 'dockers/tpu-tests/tpu_test_cases.jsonnet' ; data = open(fname).read().replace('{PYTORCH_VERSION}', '$XLA_VER')
       data = data.replace('{PYTHON_VERSION}', '$PYTHON_VER').replace('{PR_NUMBER}', '$PR_NUMBER').replace('{SHA}', '$SHA') ; open(fname, 'w').write(data)"
       cat dockers/tpu-tests/tpu_test_cases.jsonnet
      shell: bash

    - uses: google-github-actions/auth@v0
      with:
        credentials_json: ${{ secrets.GKE_SA_KEY_BASE64 }}

    # https://docs.github.com/en/actions/deployment/deploying-to-your-cloud-provider/deploying-to-google-kubernetes-engine
    - uses: google-github-actions/get-gke-credentials@v0
      with:
        cluster_name: ${{ env.GKE_CLUSTER }}
        location: ${{ env.GKE_ZONE }}

    - name: Deploy cluster
      run: |
        export PATH=$PATH:$HOME/go/bin
        job_name=$(jsonnet -J ml-testing-accelerators/ dockers/tpu-tests/tpu_test_cases.jsonnet | kubectl create -f -)
        job_name=${job_name#job.batch/}
        job_name=${job_name% created}
        pod_name=$(kubectl get po -l controller-uid=`kubectl get job $job_name -o "jsonpath={.metadata.labels.controller-uid}"` | awk 'match($0,!/NAME/) {print $1}')
        echo "GKE pod name: $pod_name"
        echo "Waiting on kubernetes job: $job_name"
        status_code=2 &&
        # Check on the job periodically. Set the status code depending on what happened to the job in Kubernetes.
        printf "Waiting for job to finish: "
        while true; do
          if kubectl get jobs $job_name -o jsonpath='Failed:{.status.failed}' | grep "Failed:1"; then
            status_code=1 && break;
          elif kubectl get jobs $job_name -o jsonpath='Succeeded:{.status.succeeded}' | grep "Succeeded:1"; then
            status_code=0 && break;
          else
            printf ".";
          fi;
          sleep 5;
        done
        echo "Done waiting. Job status code: $status_code"
        kubectl logs -f $pod_name --container=train > /tmp/full_output.txt
        grep '<?xml version="1.0" ?>' /tmp/full_output.txt  # sanity check
        csplit /tmp/full_output.txt '/<?xml version="1.0" ?>/'
        mv xx01 coverage.xml
        exit $status_code
      shell: bash

    - name: Upload coverage to Codecov
      uses: codecov/codecov-action@v3
      # see: https://github.com/actions/toolkit/issues/399
      continue-on-error: true
      with:
        token: ${{ secrets.CODECOV_TOKEN }}
        file: coverage.xml
        flags: tpu,pytest,python${{ env.PYTHON_VER }}
        name: TPU-coverage
        fail_ci_if_error: false