PSProxmoxVE/.github/workflows/integration-tests.yml

name: Integration Tests

# Real integration tests against a live Proxmox VE instance.
#
# Architecture:
#   build           → GitHub-hosted, dotnet publish → upload artifact
#   container-image → GitHub-hosted, build+push Docker image to GHCR
#   provision       → self-hosted, provisions ALL nested PVE VMs via run-integration.sh
#   test            → self-hosted, runs Pester tests via run-integration.sh
#   cleanup         → self-hosted, tears down all VMs via run-integration.sh (always runs)
#   cleanup-images  → GitHub-hosted, prunes old GHCR container images
#
# The provision/test/cleanup logic lives in tests/infrastructure/scripts/run-integration.sh
# which is the single source of truth shared between CI and local dev containers.
#
# Required repository secrets (for provisioning):
#   PVE_ENDPOINT       - Parent PVE API URL (e.g. https://pve.example.com:8006)
#   PVE_API_TOKEN      - Parent PVE API token (for Terraform + uploads)
#   PVE_TEST_PASSWORD  - Root password for nested PVE instances (used for both provision and tests)
#
# Optional repository variables:
#   PVE_TARGET_NODE    - Parent PVE node name (variable, not secret, to avoid log
#                        masking). Unset or "auto": provisioning picks the online
#                        node with the most free memory and refuses to run when
#                        no node has enough headroom — the token's user needs
#                        PVEAuditor on /nodes for the memory stats.
#   STORAGE_VM_FQDN    - DNS name of the shared storage/answer-server VM; it
#                        boots via DHCP as hostname pvetest-storage and must
#                        resolve from the runner and the CI VLAN
#                        (default: pvetest-storage.test.local)
#
# Optional secrets (for skip_provision mode):
#   PVETEST_HOST       - Hostname or IP of a pre-existing nested PVE
#
# WARNING: Integration tests create and destroy real resources on the target
# node. Never run against production.

concurrency:
  group: integration-tests
  cancel-in-progress: false

on:
  push:
    branches: [ main ]
  # NOTE: pull_request trigger intentionally removed. Integration tests
  # run on a self-hosted runner with access to secrets and a live PVE
  # node. Fork PRs must not execute code on that runner. Integration
  # tests run automatically on push to main after merge.
  workflow_dispatch:
    inputs:
      skip_provision:
        description: 'Skip provisioning (use existing PVE from PVETEST_HOST secret)'
        required: false
        type: boolean
        default: false
      keep_vms:
        description: 'Skip cleanup so the nested PVE VMs survive for inspection'
        required: false
        type: boolean
        default: false

permissions:
  contents: read
  packages: write

env:
  SCRIPTS_DIR: tests/infrastructure/scripts
  TEST_IMAGE: ghcr.io/goodolclint/psproxmoxve-integration
  CACHE_DIR: /opt/pve-integration
  # PVE 8 reached EOL 2026-08-31; CI provisions and tests PVE 9 only.
  # run-integration.sh still accepts "8" for local runs against old targets.
  PVE_VERSIONS: '9'
  # Base ISO for the nested nodes. Set here rather than in run-integration.sh
  # so provision and cleanup agree on the name — force-cleanup deletes the
  # generated ISO by name and would miss it otherwise.
  PVE9_ISO: proxmox-ve_9.2-1.iso

jobs:
  # ── Build module artifact (GitHub-hosted) ────────────────────────
  build:
    if: github.actor != 'dependabot[bot]'
    runs-on: ubuntu-latest
    timeout-minutes: 10
    steps:
      - uses: actions/checkout@v7

      - name: Setup .NET
        uses: actions/setup-dotnet@v6
        with:
          dotnet-version: '10.0.x'

      - name: Build module
        run: dotnet publish src/PSProxmoxVE/PSProxmoxVE.csproj --configuration Release --framework netstandard2.0 --output ./publish/netstandard2.0

      - name: Clean publish output
        run: rm -f ./publish/netstandard2.0/*.deps.json

      - name: Upload module artifact
        uses: actions/upload-artifact@v7
        with:
          name: module-integration
          path: ./publish/netstandard2.0/

  # ── Build test container image (GitHub-hosted) ───────────────────────
  container-image:
    if: github.actor != 'dependabot[bot]'
    runs-on: ubuntu-latest
    timeout-minutes: 20
    steps:
      - uses: actions/checkout@v7

      - name: Log in to GHCR
        uses: docker/login-action@v4
        with:
          registry: ghcr.io
          username: ${{ github.actor }}
          password: ${{ secrets.GITHUB_TOKEN }}

      - name: Build and push
        uses: docker/build-push-action@v7
        with:
          context: .
          file: tests/Dockerfile.test
          target: dev-infra
          push: true
          tags: ${{ env.TEST_IMAGE }}:${{ github.sha }},${{ env.TEST_IMAGE }}:latest

  # ── Provision all nested PVE VMs + storage VM (self-hosted) ─────────
  provision:
    if: inputs.skip_provision != true
    needs: [build, container-image]
    runs-on: psproxmoxve
    timeout-minutes: 45
    container:
      image: ghcr.io/goodolclint/psproxmoxve-integration:${{ github.sha }}
      credentials:
        username: ${{ github.actor }}
        password: ${{ secrets.GITHUB_TOKEN }}
      volumes:
        - /opt/pve-integration:/opt/pve-integration

    steps:
      - uses: actions/checkout@v7

      - name: Provision PVE instances
        shell: bash
        env:
          PVE_ENDPOINT: ${{ secrets.PVE_ENDPOINT }}
          PVE_API_TOKEN: ${{ secrets.PVE_API_TOKEN }}
          PVE_TARGET_NODE: ${{ vars.PVE_TARGET_NODE }}
          PVE_PASSWORD: ${{ secrets.PVE_TEST_PASSWORD }}
          STORAGE_VM_FQDN: ${{ vars.STORAGE_VM_FQDN }}
          TF_VAR_disk_storage: ${{ vars.DISK_STORAGE }}
          TF_VAR_iso_storage: ${{ vars.ISO_STORAGE }}
          TF_VAR_network_bridge: ${{ vars.NETWORK_BRIDGE }}
          TF_VAR_pool_id: ${{ vars.POOL_ID }}
        run: bash ${SCRIPTS_DIR}/run-integration.sh provision

  # ── Integration tests (self-hosted, per PVE version) ────────────────
  test:
    needs: [build, provision]
    runs-on: psproxmoxve
    timeout-minutes: 20
    container:
      image: ghcr.io/goodolclint/psproxmoxve-integration:${{ github.sha }}
      credentials:
        username: ${{ github.actor }}
        password: ${{ secrets.GITHUB_TOKEN }}
      volumes:
        - /opt/pve-integration:/opt/pve-integration
    strategy:
      fail-fast: false
      matrix:
        pve_version: ['9']

    steps:
      - uses: actions/checkout@v7

      - name: Download module artifact
        uses: actions/download-artifact@v8
        with:
          name: module-integration
          path: ./publish/netstandard2.0/

      - name: Run integration tests (PVE ${{ matrix.pve_version }})
        shell: bash
        env:
          PVE_PASSWORD: ${{ secrets.PVE_TEST_PASSWORD }}
          MODULE_ARTIFACT: ./publish/netstandard2.0
          SKIP_PROVISION: ${{ inputs.skip_provision || 'false' }}
          PVETEST_HOST: ${{ secrets.PVETEST_HOST }}
          PVETEST_PASSWORD: ${{ secrets.PVE_TEST_PASSWORD }}
        run: bash ${SCRIPTS_DIR}/run-integration.sh test ${{ matrix.pve_version }}

      - name: Diagnose cluster state
        if: failure()
        shell: bash
        env:
          PVE_PASSWORD: ${{ secrets.PVE_TEST_PASSWORD }}
        run: bash ${SCRIPTS_DIR}/diagnose-cluster.sh ${{ matrix.pve_version }}

      - name: Upload test results
        if: always()
        uses: actions/upload-artifact@v7
        with:
          name: integration-test-results-pve${{ matrix.pve_version }}
          path: TestResults/

  # ── Cleanup: destroy all VMs (always runs) ──────────────────────
  cleanup:
    needs: [provision, test]
    if: always() && needs.provision.result != 'skipped' && github.actor != 'dependabot[bot]' && !inputs.keep_vms
    runs-on: psproxmoxve
    timeout-minutes: 15
    container:
      image: ghcr.io/goodolclint/psproxmoxve-integration:${{ github.sha }}
      credentials:
        username: ${{ github.actor }}
        password: ${{ secrets.GITHUB_TOKEN }}
      volumes:
        - /opt/pve-integration:/opt/pve-integration

    steps:
      - uses: actions/checkout@v7

      - name: Cleanup PVE instances
        shell: bash
        env:
          PVE_ENDPOINT: ${{ secrets.PVE_ENDPOINT }}
          PVE_API_TOKEN: ${{ secrets.PVE_API_TOKEN }}
          PVE_TARGET_NODE: ${{ vars.PVE_TARGET_NODE }}
          PVE_PASSWORD: ${{ secrets.PVE_TEST_PASSWORD }}
          STORAGE_VM_FQDN: ${{ vars.STORAGE_VM_FQDN }}
          TF_VAR_disk_storage: ${{ vars.DISK_STORAGE }}
          TF_VAR_iso_storage: ${{ vars.ISO_STORAGE }}
          TF_VAR_network_bridge: ${{ vars.NETWORK_BRIDGE }}
          TF_VAR_pool_id: ${{ vars.POOL_ID }}
        # Always use force-cleanup in CI — handles corrupted state from
        # cancelled runs, timeouts, or partial provisioning failures.
        run: bash ${SCRIPTS_DIR}/run-integration.sh force-cleanup

  # ── Clean up old container images from GHCR ──────────────────────
  cleanup-images:
    needs: test
    if: always() && github.actor != 'dependabot[bot]'
    runs-on: ubuntu-latest
    timeout-minutes: 5
    env:
      FORCE_JAVASCRIPT_ACTIONS_TO_NODE24: true
    steps:
      - name: Delete old container image versions
        uses: actions/delete-package-versions@v5
        with:
          package-name: psproxmoxve-integration
          package-type: container
          min-versions-to-keep: 3
          delete-only-untagged-versions: false
          token: ${{ secrets.GITHUB_TOKEN }}