openshift-docling-e2e.yaml

# ============================================================================
# OpenShift Docling E2E Test Workflow (EC2 Runner)
# ============================================================================
#
# This workflow tests the Spark operator using Kustomize manifests and
# shell-based E2E tests.
#
# Tests performed:
#   1. Installs Spark operator via Kustomize (config/default/)
#   2. Verifies fsGroup is NOT 185 (OpenShift security requirement)
#   3. Runs the docling-spark-app workload from examples/openshift/k8s/
#
# Infrastructure:
#   - Uses self-hosted EC2 runners for larger disk space and resources
#   - EC2 instances are started on-demand and terminated after tests
#
# Triggers:
#   - Manual (workflow_dispatch)
#   - Push to main branch when OpenShift-related files change
#
# ============================================================================

name: OpenShift Docling E2E Test

on:
  # Manual trigger - allows running from GitHub Actions UI or CLI
  workflow_dispatch:
    inputs:
      k8s_version:
        description: "Kubernetes version to test"
        required: false
        default: "v1.30.8"

  # Automatic trigger on push to release branches
  push:
    branches:
      - main
    paths:
      - 'examples/openshift/**'
      - 'config/**'
      - '.github/workflows/openshift-docling-e2e.yaml'

env:
  INSTANCE_TYPE: "m5.xlarge"

# We don't need anything other than Bash for our shell
defaults:
  run:
    shell: bash

# Cancel in-progress runs for the same branch
concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

jobs:
  # ============================================================================
  # Job 1: Launch EC2 Runner
  # ============================================================================
  launch-ec2-runner:
    runs-on: ubuntu-latest
    permissions:
      id-token: write # Required for OIDC (AWS auth)
      contents: read
    outputs:
      label: ${{ steps.start-ec2-runner.outputs.label }}
      ec2-instance-id: ${{ steps.start-ec2-runner.outputs.ec2-instance-id }}

    steps:
      - name: Checkout repository
        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6

      - name: Configure AWS Credentials
        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708 # v5.1.1
        with:
          role-to-assume: "arn:aws:iam::${{ secrets.AWS_ACCOUNT_ID }}:role/${{ vars.SPARK_OPERATOR_IAM_ROLE }}"
          aws-region: us-east-2
          role-session-name: odh-spark-operator  # For tracking in CloudTrail

      - name: Start EC2 runner
        id: start-ec2-runner
        uses: machulav/ec2-github-runner@a6dbcefcf8a31a861f5e078bb153ed332130c512 # v2.4.3
        with:
          mode: start
          github-token: "${{ secrets.SPARK_OPERATOR_GH_PERSONAL_ACCESS_TOKEN }}"
          ec2-instance-type: "${{ env.INSTANCE_TYPE }}"
          availability-zones-config: >
            [
              {"imageId": "ami-03ae00a3c1d00fd0f", "subnetId": "${{ vars.US_EAST_2A_SUBNET_ID }}", "securityGroupId": "${{ vars.US_EAST_2_SG_ID }}"},
              {"imageId": "ami-03ae00a3c1d00fd0f", "subnetId": "${{ vars.US_EAST_2B_SUBNET_ID }}", "securityGroupId": "${{ vars.US_EAST_2_SG_ID }}"},
              {"imageId": "ami-03ae00a3c1d00fd0f", "subnetId": "${{ vars.US_EAST_2C_SUBNET_ID }}", "securityGroupId": "${{ vars.US_EAST_2_SG_ID }}"}
            ]
          block-device-mappings: >
            [
              {"DeviceName": "/dev/sda1", "Ebs": {"VolumeSize": 150}}
            ]
          aws-resource-tags: >
            [
              {"Key": "Name", "Value": "spark-operator-gh-runner"},
              {"Key": "GitHubRepository", "Value": "${{ github.repository }}"},
              {"Key": "GitHubRef", "Value": "${{ github.ref }}"},
              {"Key": "GitHubPR", "Value": "${{ github.event.number || 'N/A' }}"}
            ]

  # ============================================================================
  # Job 2: OpenShift E2E Tests (runs on EC2)
  # ============================================================================
  openshift-e2e:
    name: OpenShift E2E Tests
    needs:
      - launch-ec2-runner
    runs-on: ${{ needs.launch-ec2-runner.outputs.label }}
    env:
      K8S_VERSION: ${{ inputs.k8s_version || 'v1.30.8' }}
      # Home and Kubernetes config (EC2 runner may not have HOME set)
      HOME: ${{ github.workspace }}/home
      KUBECONFIG: ${{ github.workspace }}/home/.kube/config

    steps:
      - name: Setup Environment
        run: |
          echo "Running on EC2 ${{ needs.launch-ec2-runner.outputs.ec2-instance-id }}"
          echo "Instance type: ${{ env.INSTANCE_TYPE }}"
          cat /etc/os-release
          df -h
          
          # Create home and kubeconfig directories
          mkdir -p $HOME/.kube
          echo "HOME=$HOME"
          echo "KUBECONFIG=$KUBECONFIG"

      # ======================================================================
      # Step 1: Checkout the repository
      # ======================================================================
      - name: Checkout source code
        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6
        with:
          fetch-depth: 0

      # ======================================================================
      # Step 2: Install dependencies (docker, make, kubectl, kind)
      # The EC2 runner AMI doesn't have these pre-installed
      # ======================================================================
      - name: Install dependencies
        run: |
          set -eou pipefail
          echo "Installing Docker CE..."
          # Install required packages
          sudo yum install -y yum-utils
          
          # Add Docker's official repository
          sudo yum-config-manager --add-repo https://download.docker.com/linux/rhel/docker-ce.repo
          
          # Install Docker CE
          sudo yum install -y docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
          
          # Start and enable Docker
          sudo systemctl start docker
          sudo systemctl enable docker
          
          # Allow docker commands without sudo for this session
          sudo chmod 666 /var/run/docker.sock
          docker --version

          echo "Installing build tools..."
          sudo yum install -y make git

          echo "Installing kubectl..."
          curl -LO "https://dl.k8s.io/release/$(curl -L -s https://dl.k8s.io/release/stable.txt)/bin/linux/amd64/kubectl"
          sudo install -o root -g root -m 0755 kubectl /usr/local/bin/kubectl
          kubectl version --client

          echo "Installing Kind..."
          curl -Lo ./kind https://kind.sigs.k8s.io/dl/v0.31.0/kind-linux-amd64
          sudo install -o root -g root -m 0755 kind /usr/local/bin/kind
          kind version

          echo "All dependencies installed successfully!"

      # ======================================================================
      # Step 3: Free up disk space
      # The docling-spark image is very large (~9.5GB), need space
      # ======================================================================
      - name: Free up disk space
        run: |
          echo "=== Disk space BEFORE cleanup ==="
          df -h
          
          echo ""
          echo "Removing unused Docker resources..."
          docker system prune -af --volumes || true
          
          echo ""
          echo "=== Disk space AFTER cleanup ==="
          df -h

      # ======================================================================
      # Step 4: Create Kind cluster, load docling image, upload test PDFs
      # Uses the standardized make target for full setup
      # ======================================================================
      - name: Setup Kind cluster with docling image and test data
        env:
          K8S_VERSION: ${{ inputs.k8s_version || 'v1.30.8' }}
          KIND_KUBE_CONFIG: ${{ env.KUBECONFIG }}
          DOCLING_IMAGE: quay.io/rishasin/docling-spark@sha256:7e8431fc89dbc6c10aec1f0401aadd9c9cd66b9728fbcb98f6bf40ba3e3b4cdb
        run: make -C examples/openshift kind-setup-full

      # ======================================================================
      # Step 5: Verify cluster is ready
      # ======================================================================
      - name: Verify cluster
        run: |
          echo "Verifying Kind cluster is ready..."
          kubectl cluster-info
          kubectl get nodes
          kubectl get pvc -n spark-operator
          echo "Cluster is ready!"

      # ======================================================================
      # Step 6: Install Spark operator via Kustomize
      # ======================================================================
      - name: Install Spark operator
        run: CLEANUP=false make -C examples/openshift operator-install

      # ======================================================================
      # Step 7: Run Docling Spark E2E test
      # ======================================================================
      - name: Run Docling Spark E2E test
        run: CLEANUP=false make -C examples/openshift test-docling-spark

      # ======================================================================
      # Step 8: Download and show results
      # Multi-output format produces per-PDF files:
      #   - {name}.md          (markdown content)
      #   - {name}.json        (docling JSON format)
      #   - {name}_metadata.json (processing metadata)
      #   - {name}_error.json  (for failed PDFs)
      #   - summary.jsonl      (combined summary)
      # ======================================================================
      - name: Download and show results
        if: success()
        run: |
          echo "Downloading results from output PVC..."
          mkdir -p ./output
          
          cd examples/openshift
          chmod +x k8s/deploy.sh
          ./k8s/deploy.sh download ../../output/
          
          echo ""
          echo "=== Output Directory Contents ==="
          ls -la ../../output/
          
          # Count different file types
          MD_COUNT=$(find ../../output/ -maxdepth 1 -name "*.md" 2>/dev/null | wc -l | tr -d ' ')
          JSON_COUNT=$(find ../../output/ -maxdepth 1 -name "*.json" ! -name "*_error.json" ! -name "*_metadata.json" 2>/dev/null | wc -l | tr -d ' ')
          METADATA_COUNT=$(find ../../output/ -maxdepth 1 -name "*_metadata.json" 2>/dev/null | wc -l | tr -d ' ')
          ERROR_COUNT=$(find ../../output/ -maxdepth 1 -name "*_error.json" 2>/dev/null | wc -l | tr -d ' ')
          
          echo ""
          echo "=== File Summary ==="
          echo "Markdown files (.md):       $MD_COUNT"
          echo "JSON files (.json):         $JSON_COUNT"
          echo "Metadata files:             $METADATA_COUNT"
          echo "Error files (*_error.json): $ERROR_COUNT"
          
          # Check summary.jsonl
          if [ -s ../../output/summary.jsonl ]; then
            LINE_COUNT=$(wc -l < ../../output/summary.jsonl | tr -d ' ')
            echo ""
            echo "=== Processing Summary (summary.jsonl: $LINE_COUNT documents) ==="
            python3 -c "
          import sys, json
          success_count = 0
          fail_count = 0
          for line in open('../../output/summary.jsonl'):
              d = json.loads(line.strip())
              doc_path = d.get('document_path', 'unknown')
              # Extract just filename
              doc_name = doc_path.split('/')[-1] if '/' in doc_path else doc_path
              if d.get('success', False):
                  print(f'  ✅ {doc_name}')
                  success_count += 1
              else:
                  error_msg = d.get('error_message', 'Unknown error')
                  # Truncate long error messages
                  if error_msg and len(error_msg) > 80:
                      error_msg = error_msg[:77] + '...'
                  print(f'  ❌ {doc_name}: {error_msg}')
                  fail_count += 1
          print(f'\nTotal: {success_count} succeeded, {fail_count} failed')
          "
          fi
          
          # Show sample markdown if any exist
          if [ "$MD_COUNT" -gt 0 ]; then
            echo ""
            echo "=== Sample Markdown Output ==="
            FIRST_MD=$(find ../../output/ -maxdepth 1 -name "*.md" | head -1)
            echo "File: $(basename $FIRST_MD)"
            echo "Size: $(wc -c < "$FIRST_MD" | tr -d ' ') bytes"
            echo "Preview (first 500 chars):"
            echo "---"
            head -c 500 "$FIRST_MD"
            echo ""
            echo "---"
          fi
          
          # Show error details if any failures
          if [ "$ERROR_COUNT" -gt 0 ]; then
            echo ""
            echo "=== Error Details ==="
            for err_file in ../../output/*_error.json; do
              if [ -f "$err_file" ]; then
                echo "File: $(basename $err_file)"
                python3 -c "
          import json
          with open('$err_file') as f:
              d = json.load(f)
              print(f'  Source: {d.get(\"source_file\", \"unknown\")}')
              err = d.get('error_message', 'No error message')
              if len(err) > 200:
                  err = err[:197] + '...'
              print(f'  Error: {err}')
          "
              fi
            done
          fi
          
          # Determine success/failure
          echo ""
          if [ -s ../../output/summary.jsonl ]; then
            echo "✅ Download complete! Found $MD_COUNT markdown, $JSON_COUNT JSON, $ERROR_COUNT error files."
          else
            echo "❌ ERROR: summary.jsonl is empty or missing"
            exit 1
          fi

      # ======================================================================
      # Step 10: Collect logs on failure
      # ======================================================================
      - name: Collect logs on failure
        if: failure()
        run: |
          echo "=== Pod Status ==="
          kubectl get pods -A

          echo ""
          echo "=== Spark Operator Logs ==="
          kubectl logs -n spark-operator -l app.kubernetes.io/name=spark-operator --tail=100 || true

          echo ""
          echo "=== Spark Operator Namespace Pods ==="
          kubectl get pods -n spark-operator -o wide || true

          echo ""
          echo "=== Driver Pod Details ==="
          kubectl describe pod -n spark-operator -l spark-role=driver || true

          echo ""
          echo "=== Driver Pod Logs ==="
          kubectl logs -n spark-operator -l spark-role=driver --tail=100 || true

          echo ""
          echo "=== SparkApplication Status ==="
          kubectl get sparkapplication -A -o yaml || true

          echo ""
          echo "=== Events in spark-operator namespace ==="
          kubectl get events -n spark-operator --sort-by='.lastTimestamp' || true

          echo ""
          echo "=== All Recent Events ==="
          kubectl get events -A --sort-by='.lastTimestamp' | tail -50

      # ======================================================================
      # Step 11: Cleanup Kind cluster
      # ======================================================================
      - name: Cleanup Kind cluster
        if: always()
        run: make -C examples/openshift kind-cleanup || true

  # ============================================================================
  # Job 3: Stop EC2 Runner
  # ============================================================================
  stop-ec2-runner:
    permissions:
      id-token: write # Required for OIDC (AWS auth)
      contents: read
    needs:
      - launch-ec2-runner
      - openshift-e2e
    runs-on: ubuntu-latest
    if: ${{ always() && needs.launch-ec2-runner.outputs.ec2-instance-id != '' }}

    steps:
      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@61815dcd50bd041e203e49132bacad1fd04d2708 # v5.1.1
        with:
          role-to-assume: "arn:aws:iam::${{ secrets.AWS_ACCOUNT_ID }}:role/${{ vars.SPARK_OPERATOR_IAM_ROLE }}"
          aws-region: us-east-2
          role-session-name: odh-spark-operator  # For tracking in CloudTrail

      - name: Stop EC2 runner
        uses: machulav/ec2-github-runner@a6dbcefcf8a31a861f5e078bb153ed332130c512 # v2.4.3
        with:
          mode: stop
          github-token: "${{ secrets.SPARK_OPERATOR_GH_PERSONAL_ACCESS_TOKEN }}"
          label: ${{ needs.launch-ec2-runner.outputs.label }}
          ec2-instance-id: ${{ needs.launch-ec2-runner.outputs.ec2-instance-id }}