cloud-ci.yml

name: tests

on:
  # Runs for pull requests
  pull_request:
    branches:
      - master

permissions:
  id-token: write

jobs:
  cloud-tests:
    strategy:
      fail-fast: true
      matrix:
        include:
          - arch: cuda
            exclude: "no-cuda"
            run_on: azure__a100
          # - arch: rocm
          #   exclude : "no-rocm"

    runs-on: ubuntu-latest
    environment: cloud-ci

    # Cancel previous jobs if a new version was pushed
    concurrency:
      group: "${{ github.ref }}-${{ matrix.arch }}-${{ matrix.run_on }}"
      cancel-in-progress: true

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

    env:
      MILABENCH_CONFIG: "config/standard.yaml"
      MILABENCH_SYSTEM: "config/cloud-system.yaml"
      MILABENCH_BASE: "output"
      MILABENCH_ARGS: ""
      MILABENCH_GPU_ARCH: "${{ matrix.arch }}"
      MILABENCH_DASH: "no"
      ARM_TENANT_ID: "${{ secrets.ARM_TENANT_ID }}"
      ARM_SUBSCRIPTION_ID: "${{ secrets.ARM_SUBSCRIPTION_ID }}"
      AZURE_CORE_OUTPUT: none

    steps:
      - uses: actions/checkout@v3
        with:
          token: ${{ github.token }}

      - uses: actions/setup-python@v2
        with:
          python-version: 3.9

      # Follow
      # https://registry.terraform.io/providers/hashicorp/azurerm/latest/docs/guides/service_principal_client_secret
      # to generate a clientId as well as a clientSecret
      - name: Azure login
        uses: azure/login@v2
        environment: cloud-ci
        with:
          creds: |
            {
              "clientId": "${{ secrets.ARM_CLIENT_ID }}",
              "clientSecret": "${{ secrets.ARM_CLIENT_SECRET }}",
              "subscriptionId": "${{ secrets.ARM_SUBSCRIPTION_ID }}",
              "tenantId": "${{ secrets.ARM_TENANT_ID }}"
            }

      - name: dependencies
        run: |
          python -m pip install -U pip
          python -m pip install -U poetry
          poetry lock --no-update
          poetry install

      - name: setup cloud credentials
        run: |
          mkdir -p ~/.aws
          mkdir -p ~/.ssh/covalent
          echo "${{ secrets.COVALENT_EC2_EXECUTOR_KEYPAIR }}" >~/.ssh/covalent/covalent-ec2-executor-keypair.pem
          echo "[default]" >~/.aws/credentials
          echo "aws_access_key_id=${{ secrets.AWS_ACCESS_KEY_ID }}" >>~/.aws/credentials
          echo "aws_secret_access_key=${{ secrets.AWS_SECRET_ACCESS_KEY }}" >>~/.aws/credentials
          chmod -R a-rwx,u+rwX ~/.aws ~/.ssh

      - name: setup cloud
        run: |
          _system=$(
          poetry run milabench cloud \
            --setup \
            --run-on ${{ matrix.run_on }}
          )
          { read _hash ; }< <(
          echo -n "$_system" | while read l
          do
              if [[ "$l" == "# hash::>"* ]]
              then
                  echo -n "${l#*::>}"
              fi
          done
          echo
          )
          if [[ -z "${_hash}" ]]
          then
            >&2 echo "Failed to fetch system config hash"
            exit 1
          fi
          echo -n "$_system" >$MILABENCH_SYSTEM.$_hash
          echo "MILABENCH_SYSTEM=$MILABENCH_SYSTEM.$_hash" >>$GITHUB_ENV

      - name: install benchmarks
        run: |
          poetry run milabench install --variant ${{ matrix.arch }}

      - name: prepare benchmarks
        run: |
          poetry run milabench prepare

      - name: run benchmarks
        run: |
          poetry run milabench run

      - name: Summary
        run: |
          # git remote set-url origin "https://${{ vars.REPORTS_USERNAME }}:${{ secrets.REPORTS_PAT }}@$(git remote get-url origin | cut -d'/' -f3-)"
          git config --global user.email "github-ci@example.com"
          git config --global user.name "GitHub CI"
          poetry run milabench report --push

      - name: teardown cloud
        if: always()
        run: |
          if [[ -f "${MILABENCH_SYSTEM%.*}" ]]
          then
            export MILABENCH_SYSTEM=${MILABENCH_SYSTEM%.*}
          fi
          poetry run milabench cloud \
            --teardown \
            --run-on ${{ matrix.run_on }} \
            --all