Split up end2end frontend tests to speed them up

.github/workflows/end2end_test.yml

@@ -20,7 +20,16 @@ env:
 
 jobs:
   end-to-end-test:
-    name: End2End Test
+    name: End2End Test (shard ${{ matrix.shard.index }}/${{ matrix.shard.total }})
+    strategy:
+      fail-fast: false
+      matrix:
+        # Run each Playwright shard on its own runner to split the suite by file.
+        shard:
+          - index: 1
+            total: 2
+          - index: 2
+            total: 2
     runs-on: ubuntu-latest
     steps:
       - name: Checkout Code
@@ -52,29 +61,13 @@ jobs:
           path: lightly_studio/datasets/coco-128
           key: coco-example-dataset
 
-      - name: Cache lightly_studio app build
-        id: cache-python-build
-        uses: actions/cache@v4
-        with:
-          path: lightly_studio/dist
-          key: python-build-${{ hashFiles('lightly_studio/src/**', 'lightly_studio/pyproject.toml', 'lightly_studio_view/src/**', 'lightly_studio_view/public/**', 'lightly_studio_view/package.json', 'lightly_studio_view/package-lock.json') }}
-
-      - name: Cache e2e-tests
-        id: cache-e2e-tests
-        uses: actions/cache@v4
-        with:
-          path: lightly_studio_view/playwright-report
-          key: python-build-${{ hashFiles('lightly_studio/e2e-tests/index_dataset_for_end2end_ui_tests.py', 'lightly_studio/src/**', 'lightly_studio/pyproject.toml', 'lightly_studio_view/e2e/**', 'lightly_studio_view/src/**', 'lightly_studio_view/public/**', 'lightly_studio_view/package.json', 'lightly_studio_view/package-lock.json') }}
-
       ### Setup steps for the workflow ###
       - name: Set Up Python
-        if: steps.cache-python-build.outputs.cache-hit != 'true' || steps.cache-e2e-tests.outputs.cache-hit != 'true'
         uses: actions/setup-python@v5
         with:
           python-version: "3.8"
 
       - name: Set Up uv
-        if: steps.cache-python-build.outputs.cache-hit != 'true' || steps.cache-e2e-tests.outputs.cache-hit != 'true'
         uses: astral-sh/setup-uv@v6
         with:
           version: "0.8.17"
@@ -87,8 +80,9 @@ jobs:
           cache: 'npm'
           cache-dependency-path: lightly_studio_view/package-lock.json
 
-      - name: Install dependencies
-        if: steps.cache-e2e-tests.outputs.cache-hit != 'true' || steps.cache-python-build.outputs.cache-hit != 'true'
+      # Always install npm dependencies (needed for npx playwright commands)
+      # The setup-node npm cache makes this fast even on repeated runs
+      - name: Install npm dependencies
         working-directory: lightly_studio_view
         run: npm ci
 
@@ -102,13 +96,12 @@ jobs:
 
       # Always ensure system packages for browsers are present on Ubuntu
       - name: Install Playwright system dependencies
-        if: steps.cache-e2e-tests.outputs.cache-hit != 'true'
         working-directory: lightly_studio_view
         run: npx playwright install-deps chromium
 
       # Only download browser binaries when cache misses; otherwise reuse cache
       - name: Install Playwright browsers
-        if: steps.cache-e2e-tests.outputs.cache-hit != 'true' && steps.cache-playwright-browsers.outputs.cache-hit != 'true'
+        if: steps.cache-playwright-browsers.outputs.cache-hit != 'true'
         working-directory: lightly_studio_view
         run: npx playwright install chromium
 
@@ -119,33 +112,26 @@ jobs:
           git clone --depth=1 https://github.com/lightly-ai/dataset_examples_studio.git tmp_dataset_repo
           cp -r tmp_dataset_repo/coco_subset_128_images/* lightly_studio/datasets/coco-128/
           rm -rf tmp_dataset_repo
-      
+
       - name: Install lightly_studio python dependencies
-        if: steps.cache-python-build.outputs.cache-hit != 'true'
         working-directory: lightly_studio
         run: make install-optional-deps
 
-      - name: Export schema from backend
-        if: steps.cache-python-build.outputs.cache-hit != 'true'
-        working-directory: lightly_studio
-        run: make export-schema
-
-      ### Buildging and steps for the workflow ###
+      ### Building steps for the workflow ###
       - name: Build the ui application
-        if: steps.cache-ui-build.outputs.cache-hit != 'true' || steps.cache-python-build.outputs.cache-hit != 'true'
+        if: steps.cache-ui-build.outputs.cache-hit != 'true'
         working-directory: lightly_studio_view
         run: |
           rm -rf build
           npm run build
 
       - name: Copy the built ui app to lightly_studio source folder
-        if: steps.cache-ui-build.outputs.cache-hit != 'true' || steps.cache-python-build.outputs.cache-hit != 'true'
+        if: steps.cache-ui-build.outputs.cache-hit != 'true'
         run: |
           rm -rf lightly_studio/src/lightly_studio/dist_lightly_studio_view_app
           cp -r lightly_studio_view/build/ lightly_studio/src/lightly_studio/dist_lightly_studio_view_app
 
-      - name: Index dataset with using lightly_studio pypi package
-        if: steps.cache-e2e-tests.outputs.cache-hit != 'true'
+      - name: Start lightly_studio server with test dataset
         working-directory: lightly_studio
         # TODO(Michal, 10/2025): Starting up the server redownloads the embedding model. Consider caching it.
         run: |
@@ -154,14 +140,13 @@ jobs:
         id: server
 
       - name: Wait for the server to start
-        if: steps.cache-e2e-tests.outputs.cache-hit != 'true'
         timeout-minutes: 5
         run: |
           echo "Waiting for the server to become available..."
           timeout=360
           elapsed=0
           interval=5
-          
+
           while [ $elapsed -lt $timeout ]; do
             if curl -s -f ${E2E_BASE_URL}/healthz > /dev/null 2>&1; then
               echo "✅ Server is up and running after ${elapsed} seconds!"
@@ -171,22 +156,26 @@ jobs:
             sleep $interval
             elapsed=$((elapsed + interval))
           done
-          
+
           if [ $elapsed -ge $timeout ]; then
             echo "❌ Timed out waiting for server to become available"
             exit 1
           fi
 
       ### Running end-to-end tests ###
+      # Shard Playwright specs across runners for parallel execution.
       - name: Run end-to-end tests for indexed dataset
-        if: steps.cache-e2e-tests.outputs.cache-hit != 'true'
         working-directory: lightly_studio_view
-        run: npx playwright test --reporter=html --trace=on
+        run: |
+          npx playwright test \
+            --reporter=html \
+            --trace=on \
+            --shard=${{ matrix.shard.index }}/${{ matrix.shard.total }}
         timeout-minutes: 10
 
       ### Cleanup steps for the workflow ###
       - name: Show server logs after failure
-        if: failure() && steps.cache-e2e-tests.outputs.cache-hit != 'true'
+        if: failure()
         working-directory: lightly_studio
         run: |
           echo "Server logs:"
@@ -196,11 +185,27 @@ jobs:
         uses: actions/upload-artifact@v4
         if: failure()
         with:
-          name: Upload Playwright Report
+          # Each shard uploads its own Playwright HTML report.
+          name: playwright-report-shard-${{ matrix.shard.index }}-of-${{ matrix.shard.total }}
           path: lightly_studio_view/playwright-report
           retention-days: 5
 
       - name: Stop lightly_studio application running in background
-        if: always() && steps.cache-e2e-tests.outputs.cache-hit != 'true'
+        if: always()
         run: |
           kill ${{ env.SERVER_PID }} || true
+
+  # Summary job that reports overall test status
+  end-to-end-test-summary:
+    name: End2End Test
+    needs: end-to-end-test
+    runs-on: ubuntu-latest
+    if: always()
+    steps:
+      - name: Check test results
+        run: |
+          if [[ "${{ needs.end-to-end-test.result }}" != "success" ]]; then
+            echo "❌ One or more test shards failed"
+            exit 1
+          fi
+          echo "✅ All end-to-end test shards passed successfully"