externalworkloads.yaml

name: External Workloads

# Any change in triggers needs to be reflected in the concurrency group.
on:
  ### FOR TESTING PURPOSES
  # This workflow runs in the context of `main`, and ignores changes to
  # workflow files in PRs. For testing changes to this workflow from a PR:
  # - Make sure the PR uses a branch from the base repository (requires write
  #   privileges). It will not work with a branch from a fork (missing secrets).
  # - Uncomment the `pull_request` event below, commit separately with a `DO
  #   NOT MERGE` message, and push to the PR. As long as the commit is present,
  #   any push to the PR will trigger this workflow.
  # - Don't forget to remove the `DO NOT MERGE` commit once satisfied. The run
  #   will disappear from the PR checks: please provide a direct link to the
  #   successful workflow run (can be found from Actions tab) in a comment.
  #
  # pull_request: {}
  ###
  pull_request_target: {}
  # Run every 6 hours
  schedule:
    - cron:  '0 4/6 * * *'

# By specifying the access of one of the scopes, all of those that are not
# specified are set to 'none'.
permissions:
  # To be able to access the repository with actions/checkout
  contents: read
  # To be able to request the JWT from GitHub's OIDC provider
  id-token: write

concurrency:
  group: ${{ github.workflow }}-${{ github.event.pull_request.number || 'scheduled' }}
  cancel-in-progress: true

env:
  zone: us-west2-a
  vmStartupScript: .github/gcp-vm-startup.sh
  # renovate: datasource=github-releases depName=cilium/cilium
  cilium_version: v1.16.4
  kubectl_version: v1.23.6
  USE_GKE_GCLOUD_AUTH_PLUGIN: True

jobs:
  installation-and-connectivity:
    name: External Workloads Installation and Connectivity Test
    if: ${{ github.repository == 'cilium/cilium-cli' }}
    runs-on: ubuntu-24.04
    timeout-minutes: 45
    steps:
      - name: Set cluster name
        run: |
          echo "clusterName=${{ github.event.repository.name }}-${{ github.run_id }}-vm${{ github.run_attempt }}" >> $GITHUB_ENV
          echo "vmName=${{ github.event.repository.name }}-${{ github.run_id }}-vm${{ github.run_attempt }}" >> $GITHUB_ENV

      - name: Checkout
        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2

      - name: Install kubectl
        run: |
          curl -sLO "https://dl.k8s.io/release/${{ env.kubectl_version }}/bin/linux/amd64/kubectl"
          curl -sLO "https://dl.k8s.io/${{ env.kubectl_version }}/bin/linux/amd64/kubectl.sha256"
          echo "$(cat kubectl.sha256)  kubectl" | sha256sum --check
          sudo install -o root -g root -m 0755 kubectl /usr/local/bin/kubectl
          kubectl version --client

      - name: Set up gcloud credentials
        id: 'auth'
        uses: google-github-actions/auth@6fc4af4b145ae7821d527454aa9bd537d1f2dc5f # v2.1.7
        with:
          workload_identity_provider: ${{ secrets.GCP_PR_WORKLOAD_IDENTITY_PROVIDER }}
          service_account: ${{ secrets.GCP_PR_SA_CLI }}
          create_credentials_file: true
          export_environment_variables: true

      - name: Set up gcloud CLI
        uses: google-github-actions/setup-gcloud@6189d56e4096ee891640bb02ac264be376592d6a # v2.1.2
        with:
          project_id: ${{ secrets.GCP_PR_PROJECT_ID }}
          version: "405.0.0"

      - name: Install gke-gcloud-auth-plugin
        run: |
          gcloud components install gke-gcloud-auth-plugin

      - name: Display gcloud CLI info
        run: |
          gcloud info

      - name: Set up job variables
        id: vars
        run: |
          if [ ${{ github.event.issue.pull_request || github.event.pull_request }} ]; then
            PR_API_JSON=$(curl \
              -H "Accept: application/vnd.github.v3+json" \
              -H "Authorization: Bearer ${{ secrets.GITHUB_TOKEN }}" \
              ${{ github.event.issue.pull_request.url || github.event.pull_request.url }})
            SHA=$(echo "$PR_API_JSON" | jq -r ".head.sha")
            OWNER=$(echo "$PR_API_JSON" | jq -r ".number")
          else
            SHA=${{ github.sha }}
            OWNER=${{ github.sha }}
          fi

          echo "sha=${SHA}" >> $GITHUB_OUTPUT
          echo "owner=${OWNER}" >> $GITHUB_OUTPUT

      - name: Create GCP VM
        uses: nick-invision/retry@7152eba30c6575329ac0576536151aca5a72780e # v3.0.0
        with:
          retry_on: error
          timeout_minutes: 1
          max_attempts: 10
          command: |
            gcloud compute instances create ${{ env.vmName }} \
              --labels "usage=${{ github.repository_owner }}-${{ github.event.repository.name }},owner=${{ steps.vars.outputs.owner }}" \
              --zone ${{ env.zone }} \
              --machine-type e2-custom-2-4096 \
              --boot-disk-type pd-standard \
              --boot-disk-size 15GB \
              --preemptible \
              --image-project ubuntu-os-cloud \
              --image-family ubuntu-2004-lts \
              --metadata hostname=${{ env.vmName }} \
              --metadata-from-file startup-script=${{ env.vmStartupScript}}

      - name: Create GKE cluster
        id: cluster
        run: |
          gcloud container clusters create ${{ env.clusterName }} \
            --labels "usage=${{ github.repository_owner }}-${{ github.event.repository.name }},owner=${{ steps.vars.outputs.owner }}" \
            --zone ${{ env.zone }} \
            --enable-ip-alias \
            --cluster-ipv4-cidr="/21" \
            --services-ipv4-cidr="/24" \
            --image-type COS_CONTAINERD \
            --num-nodes 2 \
            --machine-type e2-custom-2-4096 \
            --disk-type pd-standard \
            --disk-size 15GB \
            --node-taints node.cilium.io/agent-not-ready=true:NoExecute \
            --preemptible
          CLUSTER_CIDR=$(gcloud container clusters describe ${{ env.clusterName }} --zone ${{ env.zone }} --format="value(clusterIpv4Cidr)")
          echo "cluster_cidr=${CLUSTER_CIDR}" >> $GITHUB_OUTPUT

      - name: Get cluster credentials
        run: |
          gcloud container clusters get-credentials ${{ env.clusterName }} --zone ${{ env.zone }}

      - name: Install Cilium CLI
        uses: ./
        with:
          skip-build: 'true'
          image-tag: ${{ steps.vars.outputs.sha }}

      - name: Install Cilium
        timeout-minutes: 10
        run: |
          # Install Cilium in cluster
          cilium install \
            --version "${{ env.cilium_version }}" \
            --set cluster.id=1 \
            --set cluster.name="${{ env.clusterName }}" \
            --set bpf.monitorAggregation=none \
            --datapath-mode=tunnel \
            --set kubeProxyReplacement=true \
            --set loadBalancer.l7.backend=envoy \
            --set tls.secretsBackend=k8s \
            --set ipv4NativeRoutingCIDR="${{ steps.cluster.outputs.cluster_cidr }}"

          # Enable Relay
          cilium hubble enable

          # Wait for cilium and hubble relay to be ready
          # NB: necessary to work against occasional flakes due to https://github.com/cilium/cilium-cli/issues/918
          cilium status --wait

          # Enable cluster mesh
          # Explicitly specify LoadBalancer service type since the default type is NodePort in helm mode.
          # Ref: https://github.com/cilium/cilium-cli/pull/1527#discussion_r1177244379
          #
          # In Helm mode, externalWorkloads.enabled is set to false by default. You need to pass
          # --enable-external-workloads flag to enable it.
          # Ref: https://github.com/cilium/cilium/pull/25259
          cilium clustermesh enable --service-type LoadBalancer --enable-external-workloads

          # Wait for cluster mesh status to be ready
          cilium clustermesh status --wait

          # Add VM to cluster mesh
          cilium clustermesh vm create ${{ env.vmName }} -n default --ipv4-alloc-cidr 10.192.1.0/30
          cilium clustermesh vm status

          # Create install script for VMs
          cilium clustermesh vm install install-external-workload.sh --config debug
          sudo chmod +x install-external-workload.sh
          gcloud compute scp install-external-workload.sh ${{ env.vmName }}:~/ --zone ${{ env.zone }}
          gcloud compute ssh ${{ env.vmName }} --zone ${{ env.zone }} \
            --command "~/install-external-workload.sh"
          sleep 5s
          gcloud compute ssh ${{ env.vmName }} --zone ${{ env.zone }} \
            --command "sudo cilium status"

      - name: Verify cluster DNS on VM
        run: |
          gcloud compute ssh ${{ env.vmName }} --zone ${{ env.zone }} \
            --command "nslookup -d2 -retry=10 -timeout=5 -norecurse clustermesh-apiserver.kube-system.svc.cluster.local \$(systemd-resolve --status | grep -m 1 \"Current DNS Server:\" | cut -d':' -f2)"

      - name: Ping clustermesh-apiserver from VM
        run: |
          gcloud compute ssh ${{ env.vmName }} --zone ${{ env.zone }} \
            --command "ping -c 3 \$(sudo cilium service list get -o jsonpath='{[?(@.spec.flags.name==\"clustermesh-apiserver\")].spec.backend-addresses[0].ip}')"

      - name: Run Test
        timeout-minutes: 30
        run: |
          # Run connectivity test
          cilium connectivity test --test-concurrency=5 --all-flows --collect-sysdump-on-failure --external-target google.com.
          kubectl delete namespace -l "app.kubernetes.io/name=cilium-cli"

          # Run performance test
          cilium connectivity perf --duration 1s

      - name: Post-test information gathering
        if: ${{ !success() }}
        run: |
          echo "=== Retrieve VM state ==="
          gcloud compute ssh ${{ env.vmName }} --zone ${{ env.zone }} --command "sudo cilium status"
          gcloud compute ssh ${{ env.vmName }} --zone ${{ env.zone }} --command "sudo docker logs cilium --timestamps"

          echo "=== Retrieve cluster state ==="
          kubectl get pods --all-namespaces -o wide
          kubectl get cew --all-namespaces -o wide
          kubectl get cep --all-namespaces -o wide
          cilium status
          cilium clustermesh status
          cilium clustermesh vm status
          cilium sysdump --output-filename cilium-sysdump-out
        shell: bash {0} # Disable default fail-fast behaviour so that all commands run independently

      - name: Clean up GKE
        if: ${{ always() }}
        run: |
          while [ "$(gcloud container operations list --filter="status=RUNNING AND targetLink~${{ env.clusterName }}" --format="value(name)")" ];do
            echo "cluster has an ongoing operation, waiting for all operations to finish"; sleep 15
          done
          gcloud container clusters delete ${{ env.clusterName }} --zone ${{ env.zone }} --quiet --async
          gcloud compute instances delete ${{ env.vmName }} --zone ${{ env.zone }} --quiet
        shell: bash {0} # Disable default fail-fast behavior so that all commands run independently

      - name: Upload artifacts
        if: ${{ !success() }}
        uses: actions/upload-artifact@b4b15b8c7c6ac21ea08fcf65892d2ee8f75cf882 # v4.4.3
        with:
          name: cilium-sysdump-out.zip
          path: cilium-sysdump-out.zip
          retention-days: 5