test(nightly): stabilize the exhaustive nightly suite (#345)

Triaged the nightly failures (all pre-existing, unrelated to the analytics
work) and fixed the ones with clear root causes:

- video-speed: a 1s tiny.mp4 sped up 2x rounds to ~0.75s, flaking the +/-25%
  duration assertion under heavy CI load. Use the 8s hero.mp4 (still 44.1kHz)
  so rounding is negligible. Verified locally.
- Extended Matrix + Coverage timeouts: full-matrix / coverage-instrumented runs
  starve the heavy media tests under 4 forks at the 30s default. Make maxForks
  env-overridable (VITEST_MAX_FORKS) and run those jobs with 2 forks + a 300s
  timeout so format-matrix conversions and qr-generate stop timing out.
- Device Matrix visual baselines: the update-visual-baselines workflow could
  not start the app ('failed to create database') because it never provisioned
  Postgres/Redis. Add the same services block the e2e jobs use.
- Docker E2E: a container pnpm install network blip exits 254. Add fetch
  retries + a longer network timeout (frozen-lockfile already passes locally).
- Cross-browser: the home page is the tool catalog now (no dropzone), and the
  tool routes moved to /<section>/<toolId>. Point the upload test at a real
  tool page and fix the stale single-segment routes (/resize -> /image/resize,
  etc.).

The flaky/timeout and cross-browser fixes can only be confirmed by the nightly
(they are load- and browser-specific); a fresh nightly run will verify.

.github/workflows/nightly.yml

@@ -273,6 +273,12 @@ jobs:
           FULL_MATRIX: "1"
           FUZZ: "1"
           FUZZ_RUNS: "50"
+          # Full-matrix tests iterate every format x tool; with 4 forks the heavy
+          # media conversions starve and hit the 30s default. Fewer forks (more
+          # CPU each) plus a generous timeout keeps them from flaking.
+          VITEST_MAX_FORKS: "2"
+          VITEST_TEST_TIMEOUT: "300000"
+          VITEST_HOOK_TIMEOUT: "120000"
 
   api-fuzz:
     name: Schemathesis API Fuzz
@@ -361,6 +367,12 @@ jobs:
       - uses: ./.github/actions/setup
       - name: Run tests with coverage
         run: pnpm vitest run --coverage tests/unit/ tests/integration/
+        # Coverage instrumentation slows every test; fewer forks + a longer
+        # timeout keep slow media tests (e.g. qr-generate) from timing out.
+        env:
+          VITEST_MAX_FORKS: "2"
+          VITEST_TEST_TIMEOUT: "300000"
+          VITEST_HOOK_TIMEOUT: "120000"
       - name: Upload coverage artifact
         uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
         with:

.github/workflows/update-visual-baselines.yml

@@ -12,6 +12,32 @@ jobs:
     name: Regenerate linux visual baselines
     runs-on: ubuntu-latest
     timeout-minutes: 60
+    # The e2e webServer (playwright.config.ts) creates an isolated database via
+    # tests/e2e-pg-create-db.cjs, so it needs Postgres + Redis on localhost.
+    # Without these services the webServer fails with "failed to create database".
+    services:
+      postgres:
+        image: postgres:17-alpine
+        env:
+          POSTGRES_USER: snapotter
+          POSTGRES_PASSWORD: snapotter
+          POSTGRES_DB: snapotter
+        ports:
+          - 5432:5432
+        options: >-
+          --health-cmd "pg_isready -U snapotter"
+          --health-interval 5s
+          --health-timeout 3s
+          --health-retries 10
+      redis:
+        image: redis:8-alpine
+        ports:
+          - 6379:6379
+        options: >-
+          --health-cmd "redis-cli ping"
+          --health-interval 5s
+          --health-timeout 3s
+          --health-retries 10
     steps:
       - uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
       - uses: ./.github/actions/setup

docker/Dockerfile.test

@@ -81,7 +81,13 @@ COPY packages/ai/package.json packages/ai/tsconfig.json ./packages/ai/
 # imports at module load; without it that suite fails to collect.
 COPY packages/enterprise/package.json packages/enterprise/tsconfig.json ./packages/enterprise/
 
-# Install ALL dependencies (including devDependencies for testing)
+# Install ALL dependencies (including devDependencies for testing).
+# Registry hiccups during the container build surface as a hard exit 254
+# (interrupted fetch / integrity check). Harden the install with more retries
+# and a longer network timeout so a transient blip doesn't fail the whole job.
+RUN pnpm config set fetch-retries 5 \
+ && pnpm config set fetch-retry-maxtimeout 120000 \
+ && pnpm config set network-timeout 600000
 RUN pnpm install --frozen-lockfile
 
 # Copy source code

tests/e2e/gui-cross-browser.spec.ts

@@ -37,13 +37,18 @@ test.describe("Cross-browser smoke tests", () => {
     expect(errors).toHaveLength(0);
   });
 
-  test("home page file upload: upload image, verify preview", async ({ loggedInPage: page }) => {
+  test("tool page file upload: upload image, verify it is accepted", async ({
+    loggedInPage: page,
+  }) => {
     const errors = collectConsoleErrors(page);
+    // The home page is the tool catalog now; uploads happen on a tool page.
+    await page.goto("/image/resize");
     await page.waitForLoadState("networkidle");
 
     await uploadImage(page);
 
-    await expect(page.locator("[class*='text-green']").first()).toBeVisible();
+    // After a successful upload the tool's settings panel renders.
+    await expect(page.getByText("Settings").first()).toBeVisible();
 
     expect(errors).toHaveLength(0);
   });
@@ -51,7 +56,7 @@ test.describe("Cross-browser smoke tests", () => {
   test("resize E2E: upload, set dimensions, process, download", async ({ loggedInPage: page }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/resize");
+    await page.goto("/image/resize");
     await page.waitForLoadState("networkidle");
 
     await uploadImage(page);
@@ -81,7 +86,7 @@ test.describe("Cross-browser smoke tests", () => {
   }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/compress");
+    await page.goto("/image/compress");
     await page.waitForLoadState("networkidle");
 
     // Upload and wait for processing to produce the before-after view
@@ -278,7 +283,7 @@ test.describe("Cross-browser smoke tests", () => {
     expect(bgColor).not.toBe("");
 
     // Navigate to a tool page and verify CSS layout
-    await page.goto("/resize");
+    await page.goto("/image/resize");
     await page.waitForLoadState("networkidle");
     await page.waitForTimeout(500);
 
@@ -297,7 +302,7 @@ test.describe("Cross-browser smoke tests", () => {
   }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/resize");
+    await page.goto("/image/resize");
     await page.waitForLoadState("networkidle");
 
     await uploadImage(page);
@@ -338,7 +343,7 @@ test.describe("Cross-browser smoke tests", () => {
   test("drag-and-drop: drop image file onto dropzone", async ({ loggedInPage: page }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/resize");
+    await page.goto("/image/resize");
     await page.waitForLoadState("networkidle");
     await page.waitForTimeout(500);
 
@@ -363,7 +368,7 @@ test.describe("Cross-browser smoke tests", () => {
   test("canvas interactions: crop tool draw and adjust region", async ({ loggedInPage: page }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/crop");
+    await page.goto("/image/crop");
     await page.waitForLoadState("networkidle");
 
     await uploadImage(page);
@@ -458,7 +463,7 @@ test.describe("Cross-browser smoke tests", () => {
   test("convert E2E: upload, select format, process", async ({ loggedInPage: page }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/convert");
+    await page.goto("/image/convert");
     await page.waitForLoadState("networkidle");
 
     await uploadImage(page);
@@ -557,7 +562,7 @@ test.describe("Cross-browser smoke tests", () => {
   test("qr-generate: enter text and verify preview renders", async ({ loggedInPage: page }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/qr-generate");
+    await page.goto("/image/qr-generate");
     await page.waitForLoadState("networkidle");
     await page.waitForTimeout(500);
 
@@ -593,7 +598,7 @@ test.describe("Cross-browser smoke tests", () => {
   test("collage tool renders templates across browsers", async ({ loggedInPage: page }) => {
     const errors = collectConsoleErrors(page);
 
-    await page.goto("/collage");
+    await page.goto("/image/collage");
     await page.waitForLoadState("networkidle");
     await page.waitForTimeout(500);
 
@@ -645,12 +650,12 @@ test.describe("Cross-browser smoke tests", () => {
     const errors = collectConsoleErrors(page);
 
     // Navigate to resize
-    await page.goto("/resize");
+    await page.goto("/image/resize");
     await page.waitForLoadState("networkidle");
     await page.waitForTimeout(300);
 
     // Navigate to compress
-    await page.goto("/compress");
+    await page.goto("/image/compress");
     await page.waitForLoadState("networkidle");
     await page.waitForTimeout(300);
 

tests/integration/tools/video/video-speed.test.ts

@@ -11,7 +11,10 @@ import {
   type TestApp,
 } from "../../test-server.js";
 
-const MP4 = readFixture(fixtures.video.tiny("mp4"));
+// An 8s clip (44.1kHz audio): long enough that the 2x-speed duration check is
+// robust to ffmpeg's frame/packet rounding. The 1s tiny.mp4 rounded to ~0.75s
+// for a 2x speed-up, which flaked the +/-25% assertion under heavy CI load.
+const MP4 = readFixture(fixtures.video.hero.mp4);
 
 let testApp: TestApp;
 let adminToken: string;

vitest.config.ts

@@ -41,8 +41,13 @@ export default defineConfig({
     poolOptions: {
       forks: {
         // Parallel forks; each fork gets an isolated DB + workspace via
-        // tests/setup/per-fork-env.ts. CI runners have 4 vCPUs.
-        maxForks: process.env.CI ? 4 : Math.max(2, Math.floor(os.availableParallelism() / 2)),
+        // tests/setup/per-fork-env.ts. CI runners have 4 vCPUs. Env-overridable
+        // (VITEST_MAX_FORKS) so the heavy nightly jobs (FULL_MATRIX, coverage)
+        // can drop to fewer forks and give each slow media test more CPU, which
+        // keeps format-matrix conversions from starving and timing out.
+        maxForks:
+          Number(process.env.VITEST_MAX_FORKS) ||
+          (process.env.CI ? 4 : Math.max(2, Math.floor(os.availableParallelism() / 2))),
       },
     },
     globalSetup: ["tests/global-setup.ts"],