[KCM-6031] FinOps Agent release automation

.github/workflows/release.yaml

@@ -0,0 +1,277 @@
+name: Automated FinOps Agent Release
+
+on:
+  workflow_dispatch:
+    inputs:
+      version:
+        description: "Version number (e.g., 1.0.22)"
+        required: true
+        type: string
+      # Every irreversible action here is a `git push`, so a dry run passes
+      # --dry-run to each one: git still contacts the remote and validates auth
+      # and the ref update, but nothing lands. Steps that only observe a push's
+      # side effects are skipped with `if: !inputs.dry_run`. Not exercised: the
+      # image build, the chart publish, and Chart.lock (which needs the new
+      # chart version to actually be published).
+      dry_run:
+        description: "Dry run (validate and edit locally, push nothing)"
+        required: false
+        default: false
+        type: boolean
+
+permissions:
+  contents: read
+
+concurrency:
+  group: finops-agent-release
+  cancel-in-progress: false
+
+env:
+  TAG: v${{ inputs.version }}
+
+  # The only difference between a real release and a dry run.
+  PUSH_ARGS: ${{ inputs.dry_run && '--dry-run' || '' }}
+
+  # The cross-repo contract. The two workflow filenames name files in repos
+  # this one cannot see, so a rename there surfaces here only as a "did not
+  # start" timeout — keep them visible in one place.
+  BUILD_REPO: kubecost/integration-ci-cd
+  BUILD_WORKFLOW: finops-agent-build-release.yaml
+  CHART_REPO: kubecost/finops-agent-chart
+  CHART_WORKFLOW: publish-chart.yaml
+  KUBECOST_REPO: kubecost/kubecost
+  KUBECOST_BRANCH: develop
+
+  # Applies to every commit and annotated tag this workflow creates.
+  GIT_AUTHOR_NAME: kc-actions-bot[bot]
+  GIT_AUTHOR_EMAIL: kc-actions-bot[bot]@users.noreply.github.com
+  GIT_COMMITTER_NAME: kc-actions-bot[bot]
+  GIT_COMMITTER_EMAIL: kc-actions-bot[bot]@users.noreply.github.com
+
+jobs:
+  tag-agent:
+    name: Validate and tag ibm-finops-agent
+    runs-on: ubuntu-latest
+    steps:
+      - name: Validate version format
+        run: |
+          if ! [[ "${{ inputs.version }}" =~ ^[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
+            echo "❌ Invalid version format. Expected: X.Y.Z (e.g., 1.0.22)"
+            exit 1
+          fi
+          echo "✅ Version format valid: ${{ inputs.version }}"
+
+      - name: Generate app token to tag ibm-finops-agent and watch the build
+        id: generate-app-token
+        uses: actions/create-github-app-token@v3.2.0
+        with:
+          client-id: ${{ vars.KC_ACTIONS_BOT_APP_CLIENT_ID }}
+          private-key: ${{ secrets.KC_ACTIONS_BOT_APP_PRIVATE_KEY }}
+          repositories: |
+            ibm-finops-agent
+            integration-ci-cd
+
+      - name: Checkout ibm-finops-agent
+        uses: actions/checkout@v6
+        with:
+          ref: develop
+          token: ${{ steps.generate-app-token.outputs.token }}
+
+      - name: Check tag doesn't exist
+        run: |
+          if git ls-remote --exit-code --tags origin "$TAG" >/dev/null 2>&1; then
+            echo "❌ Tag $TAG already exists"
+            exit 1
+          fi
+          echo "✅ Tag $TAG is available"
+
+      - name: Create and push tag
+        run: |
+          git tag -a "$TAG" -m "Release $TAG"
+          git push $PUSH_ARGS origin "$TAG"
+
+      # The tag push triggers build-release.yaml, which dispatches to
+      # $BUILD_REPO. Poll for the run that appears there, then watch it.
+      - name: Wait for image build
+        if: ${{ !inputs.dry_run }}
+        env:
+          GH_TOKEN: ${{ steps.generate-app-token.outputs.token }}
+        timeout-minutes: 30
+        run: |
+          for i in {1..24}; do
+            RUN_ID=$(gh run list --repo "$BUILD_REPO" --workflow "$BUILD_WORKFLOW" \
+              --event repository_dispatch --limit 1 \
+              --json databaseId --jq '.[0].databaseId // empty')
+            if [ -n "$RUN_ID" ]; then break; fi
+            echo "⏳ Waiting for build to start (attempt $i/24)..."
+            sleep 5
+          done
+          : "${RUN_ID:?build workflow did not start within 2 minutes}"
+
+          gh run watch "$RUN_ID" --repo "$BUILD_REPO" --exit-status
+
+  update-chart:
+    name: Update finops-agent-chart
+    needs: tag-agent
+    runs-on: ubuntu-latest
+    steps:
+      - name: Generate app token to update finops-agent-chart
+        id: generate-app-token
+        uses: actions/create-github-app-token@v3.2.0
+        with:
+          client-id: ${{ vars.KC_ACTIONS_BOT_APP_CLIENT_ID }}
+          private-key: ${{ secrets.KC_ACTIONS_BOT_APP_PRIVATE_KEY }}
+          repositories: finops-agent-chart
+
+      - name: Checkout finops-agent-chart
+        uses: actions/checkout@v6
+        with:
+          repository: ${{ env.CHART_REPO }}
+          ref: main
+          token: ${{ steps.generate-app-token.outputs.token }}
+
+      - name: Bump image tag and chart version
+        run: |
+          sed -i -E "/^image:/,/^[^ ]/ s|^  tag: .*|  tag: $TAG|" charts/finops-agent/values.yaml
+          sed -i -E "s|^version: .*|version: ${{ inputs.version }}|;
+                     s|^appVersion: .*|appVersion: $TAG|" charts/finops-agent/Chart.yaml
+
+          git --no-pager diff
+
+          verify() {
+            actual=$(yq "$2" "$1")
+            [ "$actual" = "$3" ] || { echo "❌ $1: $2 is '$actual', expected '$3'"; exit 1; }
+          }
+          verify charts/finops-agent/values.yaml .image.tag "$TAG"
+          verify charts/finops-agent/Chart.yaml .version "${{ inputs.version }}"
+          verify charts/finops-agent/Chart.yaml .appVersion "$TAG"
+
+      - name: Commit, tag, and push
+        run: |
+          git add charts/finops-agent/values.yaml charts/finops-agent/Chart.yaml
+          git commit -m "Release $TAG"
+          git tag -a "$TAG" -m "Release $TAG"
+          git push $PUSH_ARGS origin main "$TAG"
+
+      # --branch matches the tag ref, so this correlates to the tag just
+      # pushed rather than to whatever happened to run most recently.
+      - name: Wait for chart publish
+        if: ${{ !inputs.dry_run }}
+        env:
+          GH_TOKEN: ${{ steps.generate-app-token.outputs.token }}
+        timeout-minutes: 20
+        run: |
+          for i in {1..24}; do
+            RUN_ID=$(gh run list --repo "$CHART_REPO" --workflow "$CHART_WORKFLOW" \
+              --event push --branch "$TAG" --limit 1 \
+              --json databaseId --jq '.[0].databaseId // empty')
+            if [ -n "$RUN_ID" ]; then break; fi
+            echo "⏳ Waiting for chart publish to start (attempt $i/24)..."
+            sleep 5
+          done
+          : "${RUN_ID:?chart publish workflow did not start within 2 minutes}"
+
+          gh run watch "$RUN_ID" --repo "$CHART_REPO" --exit-status
+
+  update-kubecost:
+    name: Update kubecost chart dependency
+    needs: update-chart
+    runs-on: ubuntu-latest
+    steps:
+      - name: Generate app token to update kubecost
+        id: generate-app-token
+        uses: actions/create-github-app-token@v3.2.0
+        with:
+          client-id: ${{ vars.KC_ACTIONS_BOT_APP_CLIENT_ID }}
+          private-key: ${{ secrets.KC_ACTIONS_BOT_APP_PRIVATE_KEY }}
+          repositories: kubecost
+
+      - name: Checkout kubecost
+        uses: actions/checkout@v6
+        with:
+          repository: ${{ env.KUBECOST_REPO }}
+          ref: ${{ env.KUBECOST_BRANCH }}
+          token: ${{ steps.generate-app-token.outputs.token }}
+
+      - name: Install Helm
+        uses: azure/setup-helm@v3
+        with:
+          version: v3.12.0
+
+      - name: Bump finops-agent dependency
+        run: |
+          sed -i -E "/^  - name: finops-agent$/,/^  - name:/ s|^    version: .*|    version: \"$TAG\"|" kubecost/Chart.yaml
+
+          git --no-pager diff
+
+          actual=$(yq '.dependencies[] | select(.name == "finops-agent") | .version' kubecost/Chart.yaml)
+          [ "$actual" = "$TAG" ] || { echo "❌ dependency is '$actual', expected '$TAG'"; exit 1; }
+
+      # Needs the new chart version to actually be published, so it can only
+      # run for real. Chart.lock is left untouched in a dry run.
+      - name: Update Chart.lock
+        if: ${{ !inputs.dry_run }}
+        run: |
+          helm repo add finops-agent https://kubecost.github.io/finops-agent-chart/ || true
+          helm repo update finops-agent
+          helm dependency update ./kubecost
+
+      - name: Commit and push
+        run: |
+          git add kubecost/Chart.yaml kubecost/Chart.lock
+          git commit -m "Update finops-agent to $TAG"
+          git push $PUSH_ARGS origin "$KUBECOST_BRANCH"
+
+  summary:
+    name: Release Summary
+    needs: [tag-agent, update-chart, update-kubecost]
+    runs-on: ubuntu-latest
+    if: always()
+    steps:
+      - name: Report status
+        env:
+          TAG_RESULT: ${{ needs.tag-agent.result }}
+          CHART_RESULT: ${{ needs.update-chart.result }}
+          KUBECOST_RESULT: ${{ needs.update-kubecost.result }}
+        run: |
+          mark() {
+            case "$1" in
+              success) echo "✅" ;;
+              skipped) echo "⏭️" ;;
+              *)       echo "❌" ;;
+            esac
+          }
+
+          if [ "${{ inputs.dry_run }}" = "true" ]; then
+            TITLE="🔍 Dry Run — FinOps Agent Release $TAG (nothing was pushed)"
+          else
+            TITLE="FinOps Agent Release $TAG"
+          fi
+
+          cat >> "$GITHUB_STEP_SUMMARY" <<EOF
+          # $TITLE
+
+          - $(mark "$TAG_RESULT") ibm-finops-agent tagged and image built
+          - $(mark "$CHART_RESULT") finops-agent-chart updated and published
+          - $(mark "$KUBECOST_RESULT") kubecost chart dependency updated
+
+          [agent tag](https://github.com/${{ github.repository }}/releases/tag/$TAG) &middot;
+          [chart release](https://github.com/$CHART_REPO/releases/tag/$TAG) &middot;
+          [kubecost commits](https://github.com/$KUBECOST_REPO/commits/$KUBECOST_BRANCH)
+          EOF
+
+      # A dry run pushes no tags, so there is never anything to clean up.
+      - name: Add cleanup instructions
+        if: ${{ !inputs.dry_run && contains(needs.*.result, 'failure') }}
+        run: |
+          cat >> "$GITHUB_STEP_SUMMARY" <<EOF
+
+          ## Manual cleanup
+
+          The release stopped partway through. Delete whichever tags were pushed:
+
+          \`\`\`bash
+          gh api -X DELETE repos/${{ github.repository }}/git/refs/tags/$TAG
+          gh api -X DELETE repos/$CHART_REPO/git/refs/tags/$TAG
+          \`\`\`
+          EOF