release.yaml

on:
  push:
    branches:
      - main
    paths-ignore:
      - README.md
      - withdrawn-images.txt
      - withdrawn-repos.txt
      - reinstated-images.txt
  schedule:
    - cron: "0 0 * * *"
  workflow_dispatch:
    inputs:
      only:
        description: "Specific image name to build"
        type: string
        required: false
        default: ""

concurrency:
  group: ${{ inputs.only || 'release' }}
  cancel-in-progress: false

env:
  TF_VAR_target_repository: cgr.dev/chainguard

permissions: {}

jobs:
  shard:
    runs-on: ubuntu-latest

    permissions:
      contents: read

    steps:
      - uses: step-security/harden-runner@20cf305ff2072d973412fa9b1e3a4f227bda3c76 # v2.14.0
        with:
          egress-policy: audit

      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2

      - id: shard
        name: Shard
        shell: bash # bash math foo required
        run: |
          images=($(find ./images -maxdepth 1 -type d -not -path "./images/TEMPLATE" | awk -F'/' '{print $3}' | sort -u | shuf))

          total=${#images[@]}

          # put each image module into its own shard
          declare -a bins
          for ((i = 0; i < total; i++)); do
            bins[$i]="${images[$i]}"
          done

          matrix=$(printf "%s\n" "${bins[@]}" | jq -cRnjr '[inputs] | [ range(0; length) as $i | { "index": $i | tostring, "images": .[$i] } ]')
          echo "matrix=${matrix}" >> $GITHUB_OUTPUT

          # Overwrite the output above if workflow_dispatch'd with `only`
          if [ -n "${{ inputs.only }}" ]; then
            shard='[{"index": 0, "images": "${{ inputs.only }}"}]'
            echo "matrix=${shard}" >> $GITHUB_OUTPUT
          fi

      - name: Shard Results
        run: echo '${{ steps.shard.outputs.matrix }}'

    outputs:
      # This is of the format [{"index": 0, "images": "a b c"}, {"index": 1, "images": "d e f"}, ...]
      matrix: "${{steps.shard.outputs.matrix}}"

  build:
    runs-on: ubuntu-latest
    needs: shard

    strategy:
      fail-fast: false
      matrix:
        shard: ${{ fromJson(needs.shard.outputs.matrix) }}

    permissions:
      id-token: write
      packages: write
      contents: read
      actions: read

    steps:
      - uses: step-security/harden-runner@20cf305ff2072d973412fa9b1e3a4f227bda3c76 # v2.14.0
        with:
          egress-policy: audit

      - uses: hashicorp/setup-terraform@b9cd54a3c349d3f38e8881555d616ced269862dd # v3.1.2
        with:
          terraform_version: "1.8.*"
          terraform_wrapper: false

      - uses: chainguard-dev/setup-chainctl@c125f765e82b09a42af3185f3214465314d75c5d # v0.5.0
        with:
          # This allows chainguard-images/images-private to publish images to cgr.dev/chainguard-private
          # We maintain this identity here:
          # https://github.com/chainguard-dev/mono/blob/main/env/chainguard-images/iac/images-pusher.tf
          identity: 720909c9f5279097d847ad02a2f24ba8f59de36a/b6461e99e132298f

      # Make cosign/crane CLI available to the tests
      - uses: sigstore/cosign-installer@faadad0cce49287aee09b3a48701e75088a2c6ad # v4.0.0
      - uses: imjasonh/setup-crane@31b88efe9de28ae0ffa220711af4b60be9435f6e # v0.4

      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
        with:
          persist-credentials: false

      # note this step takes a long time, thus only do it when we
      # really need lots of disk space, ie. for pytorch
      - name: Disk cleanup
        if: ${{ contains(matrix.shard.images, 'pytorch') }}
        run: |
          ## All disk reclaim actions look sus and are not actively
          ## maintained, doing it by hand as per https://dev.to/mathio/squeezing-disk-space-from-github-actions-runners-an-engineers-guide-3pjg#6-how-to-automate-cleanup-in-your-ci

          # Remove Java (JDKs)
          sudo rm -rf /usr/lib/jvm

          # Remove .NET SDKs
          sudo rm -rf /usr/share/dotnet

          # Remove Swift toolchain
          sudo rm -rf /usr/share/swift

          # Remove Haskell (GHC)
          sudo rm -rf /usr/local/.ghcup

          # Remove Julia
          sudo rm -rf /usr/local/julia*

          # Remove Android SDKs
          sudo rm -rf /usr/local/lib/android

          # Remove Chromium (optional if not using for browser tests)
          sudo rm -rf /usr/local/share/chromium

          # Remove Microsoft/Edge and Google Chrome builds
          sudo rm -rf /opt/microsoft /opt/google

          # Remove Azure CLI
          sudo rm -rf /opt/az

          # Remove PowerShell
          sudo rm -rf /usr/local/share/powershell

          # Remove CodeQL and other toolcaches
          sudo rm -rf /opt/hostedtoolcache

          docker system prune -af || true
          docker builder prune -af || true
          df -h

      - name: Terraform apply
        timeout-minutes: 60
        run: |
          set -exo pipefail
          env | grep '^TF_VAR_'

          make init-upgrade

          # Enable active tag updates
          make enable-active-tag-update

          targets=""
          for image in ${{ matrix.shard.images }}; do
            targets+=' -target='module."${image}"''
          done

          terraform plan ${targets} -out mega-module.tfplan
          terraform apply ${targets} -auto-approve --parallelism=$(nproc) -json mega-module.tfplan | tee /tmp/mega-module.tf.json | jq -r '.["@message"]'

      - name: Collect TF diagnostics
        if: ${{ always() }}
        id: tf-diag
        uses: chainguard-dev/actions/terraform-diag@f77a7c3431f961556be5a41e2602c450a590f3e5 # v1.5.12
        with:
          json-file: /tmp/mega-module.tf.json

      - name: Collect K8s diagnostics and upload
        if: ${{ failure() }}
        run: |
          echo 'This step is deprecated. Please use the imagetest provider and reference the imagetest logs'

      - name: Upload terraform logs
        if: always()
        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6.0.0
        with:
          name: "mega-module-${{ matrix.shard.index }}.tf.json"
          path: /tmp/mega-module.tf.json

      - name: Upload imagetest logs
        if: always()
        uses: actions/upload-artifact@b7c566a772e6b6bfb58ed0dc250532a479d7789f # v6.0.0
        with:
          name: "mega-module-${{ matrix.shard.index }}-imagetest-logs"
          path: imagetest-logs

      - uses: rtCamp/action-slack-notify@e31e87e03dd19038e411e38ae27cbad084a90661 # v2.3.3
        if: ${{ failure() && github.event_name == 'schedule' }}
        env:
          SLACK_ICON: http://github.com/chainguard-dev.png?size=48
          SLACK_USERNAME: guardian
          SLACK_WEBHOOK: ${{ secrets.DISTROLESS_SLACK_WEBHOOK }}
          SLACK_MSG_AUTHOR: chainguardian
          SLACK_CHANNEL: chainguard-images-alerts
          SLACK_COLOR: "#8E1600"
          MSG_MINIMAL: "true"
          SLACK_TITLE: "[images] release failed (shard ${{ matrix.shard.index }} of ${{ env.TOTAL_SHARDS }})"
          SLACK_MESSAGE: |
            https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}

            ${{ steps.tf-diag.outputs.errors }}

  summary:
    name: "Build Summary"
    runs-on: ubuntu-latest
    if: ${{ always() }}
    needs: build

    permissions:
      contents: read

    steps:
      - uses: step-security/harden-runner@20cf305ff2072d973412fa9b1e3a4f227bda3c76 # v2.14.0
        with:
          egress-policy: audit

      - name: "Download shard logs"
        uses: actions/download-artifact@37930b1c2abaa49bbe596cd826c3c89aef350131 # v7.0.0
        with:
          path: /tmp/shard-logs
          pattern: mega-module-*

      # Cat all the files into one, while maintaining their shard
      - run: |
          find '/tmp/shard-logs' -name 'mega-module.tf.json' | while read file; do
            shard_index=$(echo "$file" | sed -E 's/.*mega-module-([0-9]+)\.tf\.json.*/\1/')
            echo $shard_index
            jq -cr --arg shard_index "$shard_index" '. + {"shard_index":$shard_index}' $file >> logs.tf.json
          done

      # process the logs
      - run: |
          # Create a file just for errors
          jq -r 'select(.["@level"]=="error")' logs.tf.json > errors.tf.json

      # Build the markdown table
      - run: |
          echo "| Status | Shard | Image | Summary | Address |" >> $GITHUB_STEP_SUMMARY
          echo "| :-:    | ----- | ----- | ------- | ------- |" >> $GITHUB_STEP_SUMMARY

          # append the rows to the table
          export rows="$(jq -r '"| ❌ | " + .shard_index + " | " + (.diagnostic.address | split(".")[1]) + " | ```" + .diagnostic.summary + "``` | ```" + .diagnostic.address + "``` |"' errors.tf.json)"
          echo "${rows}"

          cat >> $GITHUB_STEP_SUMMARY <<EOR
          ${rows}
          EOR

      - name: Error Details
        run: |
          # Print the errors as expandable groups
          jq -r '"::group:: shard: " + .shard_index + " | " + (.diagnostic.address | split(".")[1]) + "\nresource: " + .diagnostic.address + "\n\nsummary: " + .diagnostic.summary + "\n\ndetails:\n\n" + .diagnostic.detail + "\n::endgroup::"' errors.tf.json || true