awoooi/.github/workflows/runner-healthcheck.yml

# =============================================================================
# AWOOOI - Self-hosted Runner 健康檢查 Workflow
# =============================================================================
# 設計原則:
# - 每 10 分鐘檢查一次 Runner 狀態
# - 只使用 self-hosted runner (禁止 ubuntu-latest)
#
# 🔴 HARD RULE: 禁止 ubuntu-latest (GitHub Billing 限制)
# =============================================================================

name: Runner Health Check

on:
  schedule:
    # 每 10 分鐘執行一次
    - cron: '*/10 * * * *'
  workflow_dispatch:
    inputs:
      notify_telegram:
        description: '發送 Telegram 通知'
        required: false
        default: true
        type: boolean

env:
  OPENCLAW_URL: http://192.168.0.188:8088
  RUNNER_HOST: 192.168.0.110

jobs:
  # ===========================================
  # 內部健康檢查 - 在 Self-hosted Runner 執行
  # ===========================================
  health-check:
    name: "Runner Health Check"
    runs-on: [self-hosted, harbor, k8s]
    timeout-minutes: 5

    steps:
      # =================================================================
      # 2026-03-29 Claude Code: 修復 _diag/pages 檔案衝突
      # 根本原因: GitHub Actions Runner 診斷檔案沒有自動清理
      # =================================================================
      - name: "Clean Runner Diagnostics"
        run: |
          RUNNER_ROOT=$(dirname "$(dirname "$RUNNER_TEMP")")
          rm -rf "$RUNNER_ROOT/_diag/pages" 2>/dev/null || true
          mkdir -p "$RUNNER_ROOT/_diag/pages" 2>/dev/null || true
          echo "✅ Cleaned _diag/pages"

      - name: "Check Docker Service"
        id: docker
        run: |
          echo "=== Docker 版本 ==="
          docker version --format '{{.Server.Version}}' || echo "DOCKER_FAILED"

          if docker info > /dev/null 2>&1; then
            echo "docker_status=healthy" >> $GITHUB_OUTPUT
          else
            echo "docker_status=unhealthy" >> $GITHUB_OUTPUT
          fi

      - name: "Check Disk Space"
        id: disk
        run: |
          echo "=== 磁碟使用量 ==="
          df -h /

          USAGE=$(df / | tail -1 | awk '{print $5}' | tr -d '%')
          echo "disk_usage=$USAGE" >> $GITHUB_OUTPUT

          if [ "$USAGE" -gt 95 ]; then
            echo "disk_status=critical" >> $GITHUB_OUTPUT
          elif [ "$USAGE" -gt 85 ]; then
            echo "disk_status=warning" >> $GITHUB_OUTPUT
          else
            echo "disk_status=healthy" >> $GITHUB_OUTPUT
          fi

      - name: "Check Harbor Registry"
        id: harbor
        run: |
          HTTP_CODE=$(curl -s -o /dev/null -w '%{http_code}' --max-time 10 \
            "http://192.168.0.110:5000/v2/" 2>/dev/null || echo "000")

          if [ "$HTTP_CODE" = "200" ] || [ "$HTTP_CODE" = "401" ]; then
            echo "harbor_status=healthy" >> $GITHUB_OUTPUT
            echo "Harbor Registry: OK (HTTP $HTTP_CODE)"
          else
            echo "harbor_status=unhealthy" >> $GITHUB_OUTPUT
            echo "::error::Harbor Registry 無法連接"
          fi

      - name: "Check K8s Connectivity"
        id: k8s
        run: |
          if timeout 10 ssh -o StrictHostKeyChecking=no -o ConnectTimeout=5 \
            wooo@192.168.0.125 "kubectl get nodes" 2>/dev/null; then
            echo "k8s_status=healthy" >> $GITHUB_OUTPUT
          else
            echo "k8s_status=unhealthy" >> $GITHUB_OUTPUT
            echo "::warning::K8s Master 連接失敗"
          fi

      - name: "Report to OpenClaw"
        if: always()
        env:
          DOCKER_STATUS: ${{ steps.docker.outputs.docker_status }}
          DISK_STATUS: ${{ steps.disk.outputs.disk_status }}
          DISK_USAGE: ${{ steps.disk.outputs.disk_usage }}
          HARBOR_STATUS: ${{ steps.harbor.outputs.harbor_status }}
          K8S_STATUS: ${{ steps.k8s.outputs.k8s_status }}
        run: |
          OVERALL="healthy"
          if [ "$DOCKER_STATUS" != "healthy" ] || [ "$HARBOR_STATUS" != "healthy" ]; then
            OVERALL="critical"
          elif [ "$DISK_STATUS" = "warning" ] || [ "$K8S_STATUS" != "healthy" ]; then
            OVERALL="warning"
          fi

          curl -sf -X POST "${{ env.OPENCLAW_URL }}/api/v1/webhook/runner-health" \
            -H "Content-Type: application/json" \
            -d "{
              \"runner\": \"${{ runner.name }}\",
              \"host\": \"${{ env.RUNNER_HOST }}\",
              \"overall_status\": \"${OVERALL}\",
              \"checks\": {
                \"docker\": \"${DOCKER_STATUS}\",
                \"disk\": {\"status\": \"${DISK_STATUS}\", \"usage\": ${DISK_USAGE:-0}},
                \"harbor\": \"${HARBOR_STATUS}\",
                \"k8s\": \"${K8S_STATUS}\"
              },
              \"timestamp\": \"$(date -Iseconds)\"
            }" || echo "::warning::OpenClaw 通知失敗"

  # ===========================================
  # 自動修復 (如果有問題)
  # ===========================================
  auto-repair:
    name: "Auto Repair (if needed)"
    needs: [health-check]
    if: failure()
    runs-on: [self-hosted, harbor, k8s]

    steps:
      - name: "Clean Runner Diagnostics"
        run: |
          RUNNER_ROOT=$(dirname "$(dirname "$RUNNER_TEMP")")
          rm -rf "$RUNNER_ROOT/_diag/pages" 2>/dev/null || true
          mkdir -p "$RUNNER_ROOT/_diag/pages" 2>/dev/null || true

      - name: "Attempt Docker Cleanup"
        run: |
          echo "=== 嘗試清理 Docker ==="
          docker system prune -f --volumes || true
          docker image prune -a -f --filter "until=168h" || true
          df -h /