smoke.yml

name: Smoke Test
permissions:
  id-token: write
on:
  pull_request_target:
    paths: 'task/**'
  schedule:
  - cron: '0 1 * * *' # Everyday @ 0100 UTC
  workflow_dispatch:
    inputs:
      test_target:
        description: Type of instances to be provisioned during test.
        required: true
        default: 'gpu+spot'
        type: choice
        options:
        - quick
        - quick+spot
        - gpu
        - gpu+spot
concurrency:
  group: ${{ github.workflow }}-${{ github.ref_name }}
  cancel-in-progress: true
env:
 SMOKE_TEST_TARGET: ${{ github.event_name == 'pull_request_target' && 'quick' || inputs.test_target || 'gpu+spot' }}
 SMOKE_TEST_IDENTIFIER: smoke-test-${{ github.run_id }}
jobs:
  authorize:
    environment: ${{ (github.event_name == 'pull_request_target' &&
      github.event.pull_request.head.repo.full_name != github.repository) &&
      'manual' || 'automatic' }}
    runs-on: ubuntu-latest
    steps:
    - run: echo ✓
  test:
    needs: authorize
    environment: automatic
    runs-on: ubuntu-latest
    timeout-minutes: 60
    strategy:
      fail-fast: false
      matrix:
        provider: [AWS, AZ, GCP]
    concurrency: ${{ github.workflow }}-${{ matrix.provider }}
    env:
      AZURE_CLIENT_ID: ${{ vars.AZURE_CLIENT_ID }}
      AZURE_CLIENT_SECRET: ${{ secrets.AZURE_CLIENT_SECRET }}
      AZURE_TENANT_ID: ${{ vars.AZURE_TENANT_ID }}
      AZURE_SUBSCRIPTION_ID: ${{ vars.AZURE_SUBSCRIPTION_ID }}
      GOOGLE_APPLICATION_CREDENTIALS_DATA: ${{ secrets.GOOGLE_APPLICATION_CREDENTIALS_DATA }}
      SMOKE_TEST_ENABLE_${{ matrix.provider }}: true
    steps:
    - uses: actions/checkout@v3
      with:
        ref: ${{ github.event.pull_request.head.sha || github.ref }}
    - uses: actions/setup-go@v3
      with:
        go-version-file: go.mod
    - if: matrix.provider == 'AWS'
      uses: aws-actions/configure-aws-credentials@v1
      with:
        aws-region: us-west-1
        role-to-assume: arn:aws:iam::342840881361:role/SandboxUser
    - run: go test ./task -v -timeout=30m -count=1 -tags=smoke
    - if: always()
      uses: actions/checkout@v3
      with:
        ref: main
    - if: always()
      run: go test ./task -v -timeout=30m -count=1 -tags=smoke
      env:
        SMOKE_TEST_SWEEP: true
  test-k8s:
    name: test (K8S)
    needs: authorize
    environment: automatic
    runs-on: ubuntu-latest
    timeout-minutes: 60
    concurrency: ${{ github.workflow }}-K8S
    steps:
    - uses: actions/checkout@v3
      with:
        ref: ${{ github.event.pull_request.head.sha || github.ref }}
    - uses: actions/setup-go@v3
      with:
        go-version-file: go.mod
    - uses: azure/login@v1
      with:
        client-id: ${{ secrets.AZURE_CLIENT_ID }}
        tenant-id: ${{ secrets.AZURE_TENANT_ID }}
        subscription-id: ${{ secrets.AZURE_SUBSCRIPTION_ID }}
    - if: contains(env.SMOKE_TEST_TARGET, 'gpu')
      run: |
        az extension add \
          --name=aks-preview
        az provider register \
          --namespace=Microsoft.ContainerService
        az feature register \
          --namespace=Microsoft.ContainerService \
          --name=GPUDedicatedVHDPreview
    - run: |
        az group create \
          --name="tpiSmokeTestCluster$GITHUB_RUN_ID" \
          --location=westus2
        az aks create \
          --resource-group="tpiSmokeTestCluster$GITHUB_RUN_ID" \
          --name="tpiSmokeTestCluster$GITHUB_RUN_ID" \
          --node-vm-size=${{ contains(env.SMOKE_TEST_TARGET, 'gpu') && 'Standard_NC4as_T4_v3' || 'Standard_A2_v2' }} \
          --node-count=1 \
          --aks-custom-headers=${{ contains(env.SMOKE_TEST_TARGET, 'gpu') && 'UseGPUDedicatedVHD=true' || '' }} \
          --generate-ssh-keys
    - id: credentials
      run: |
        az aks get-credentials \
          --resource-group="tpiSmokeTestCluster$GITHUB_RUN_ID" \
          --name="tpiSmokeTestCluster$GITHUB_RUN_ID" \
          --file - |
        perl -0777p \
          -e 's/%/%25/g;' \
          -e 's/\n/%0A/g;' \
          -e 's/\r/%0D/g;' \
          -e 's/(.+)/::add-mask::\1\n::set-output name=kubeconfig::\1\n/g'
    - run: go test ./task -v -timeout=60m -count=1 -tags=smoke
      env:
        KUBECONFIG_DATA: ${{ steps.credentials.outputs.kubeconfig }}
        SMOKE_TEST_ENABLE_K8S: true
    - if: always()
      run: |
        az group delete \
          --name="tpiSmokeTestCluster$GITHUB_RUN_ID" \
          --yes