e2e_tests.yaml

name: e2e

on:
  pull_request:
    branches:
      - main
      - 'release-*'
    paths-ignore:
      - 'docs/**'
      - '**.adoc'
      - '**.md'
      - 'LICENSE'
  push:
    branches:
      - main
      - 'release-*'
    paths-ignore:
      - 'docs/**'
      - '**.adoc'
      - '**.md'
      - 'LICENSE'

concurrency:
  group: ${{ github.head_ref }}-${{ github.workflow }}
  cancel-in-progress: true

jobs:
  kubernetes:

    runs-on: ubuntu-20.04

    steps:
      - name: Cleanup
        run: |
          ls -lart
          echo "Initial status:"
          df -h

          echo "Cleaning up resources:"
          sudo swapoff -a
          sudo rm -f /swapfile
          sudo apt clean
          sudo rm -rf /usr/share/dotnet
          sudo rm -rf /opt/ghc
          sudo rm -rf "/usr/local/share/boost"
          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
          docker rmi $(docker image ls -aq)

          echo "Final status:"
          df -h

      - name: Checkout code
        uses: actions/checkout@v3
        with:
          submodules: recursive

      - name: Init directories
        run: |
          TEMP_DIR="$(pwd)/tmp"
          mkdir -p "${TEMP_DIR}"
          echo "TEMP_DIR=${TEMP_DIR}" >> $GITHUB_ENV

          mkdir -p "$(pwd)/bin"
          echo "$(pwd)/bin" >> $GITHUB_PATH

      - name: Set Go
        uses: actions/setup-go@v3
        with:
          go-version: v1.18

      - name: Set up gotestfmt
        uses: gotesttools/gotestfmt-action@v2
        with:
          token: ${{ secrets.GITHUB_TOKEN }}

      - name: Container image registry
        run: |
          podman run -d -p 5000:5000 --name registry registry:2.8.1

          export REGISTRY_ADDRESS=$(hostname -i):5000
          echo "REGISTRY_ADDRESS=${REGISTRY_ADDRESS}" >> $GITHUB_ENV
          echo "Container image registry started at ${REGISTRY_ADDRESS}"

          KIND_CONFIG_FILE=${{ env.TEMP_DIR }}/kind.yaml
          echo "KIND_CONFIG_FILE=${KIND_CONFIG_FILE}" >> $GITHUB_ENV
          envsubst < ./test/e2e/kind.yaml > ${KIND_CONFIG_FILE}

          sudo --preserve-env=REGISTRY_ADDRESS sh -c 'cat > /etc/containers/registries.conf.d/local.conf <<EOF
          [[registry]]
          prefix = "$REGISTRY_ADDRESS"
          insecure = true
          location = "$REGISTRY_ADDRESS"
          EOF'

      - name: Setup KinD cluster
        uses: helm/kind-action@v1.5.0
        with:
          cluster_name: cluster
          version: v0.17.0
          config: ${{ env.KIND_CONFIG_FILE }}

      - name: Print cluster info
        run: |
          echo "KinD cluster:"
          kubectl cluster-info
          kubectl describe nodes

      - name: Deploy CodeFlare stack
        id: deploy
        run: |
          echo Deploying CodeFlare operator
          IMG="${REGISTRY_ADDRESS}"/codeflare-operator
          make image-push -e IMG="${IMG}"
          make deploy -e IMG="${IMG}"
          kubectl wait --timeout=120s --for=condition=Available=true deployment -n openshift-operators codeflare-operator-manager

          echo Setting up CodeFlare stack
          make setup-e2e

      - name: Run e2e tests
        run: |
          export CODEFLARE_TEST_TIMEOUT_SHORT=1m
          export CODEFLARE_TEST_TIMEOUT_MEDIUM=3m
          export CODEFLARE_TEST_TIMEOUT_LONG=8m

          export CODEFLARE_TEST_OUTPUT_DIR=${{ env.TEMP_DIR }}
          echo "CODEFLARE_TEST_OUTPUT_DIR=${CODEFLARE_TEST_OUTPUT_DIR}" >> $GITHUB_ENV

          set -euo pipefail
          go test -timeout 30m -v ./test/e2e -json 2>&1 | tee ${CODEFLARE_TEST_OUTPUT_DIR}/gotest.log | gotestfmt

      - name: Print CodeFlare operator logs
        if: always() && steps.deploy.outcome == 'success'
        run: |
          echo "Printing CodeFlare operator logs"
          kubectl logs -n openshift-operators --tail -1 -l app.kubernetes.io/name=codeflare-operator | tee ${CODEFLARE_TEST_OUTPUT_DIR}/codeflare-operator.log

      - name: Print MCAD controller logs
        if: always() && steps.deploy.outcome == 'success'
        run: |
          echo "Printing MCAD controller logs"
          kubectl logs -n codeflare-system --tail -1 -l component=multi-cluster-application-dispatcher | tee ${CODEFLARE_TEST_OUTPUT_DIR}/mcad.log

      - name: Print KubeRay operator logs
        if: always() && steps.deploy.outcome == 'success'
        run: |
          echo "Printing KubeRay operator logs"
          kubectl logs -n ray-system --tail -1 -l app.kubernetes.io/name=kuberay | tee ${CODEFLARE_TEST_OUTPUT_DIR}/kuberay.log

      - name: Upload logs
        uses: actions/upload-artifact@v3
        if: always() && steps.deploy.outcome == 'success'
        with:
          name: logs
          retention-days: 10
          path: |
            ${{ env.CODEFLARE_TEST_OUTPUT_DIR }}/**/*.log