Testing Patterns: Serial vs Parallel

This guide explains when and how to use serial vs parallel test execution patterns.

Overview

Pattern	Execution	Browser Context	Use Case
Parallel (default)	Tests run concurrently	Each test gets fresh context	Independent tests
Serial	Tests run sequentially	Shared browser context	Workflow tests

Parallel Tests (Default)

By default, Playwright runs tests in parallel with isolated browser contexts.

Characteristics

• Isolation: Each test gets a fresh browser context
• Speed: Tests run concurrently across workers
• Independence: Tests cannot affect each other
• Login overhead: Each test must login separately

When to Use

• Tests that don't depend on each other
• Tests that modify shared state (to avoid conflicts)
• General functional tests
• Smoke tests

Example

import { test, expect } from "rhdh-e2e-test-utils/test";

// Each test runs independently with its own browser context
test.describe("Catalog Tests", () => {
  test.beforeEach(async ({ loginHelper }) => {
    // Each test logs in fresh
    await loginHelper.loginAsGuest();
  });

  test("can view catalog", async ({ uiHelper }) => {
    await uiHelper.openSidebar("Catalog");
    await uiHelper.verifyHeading("My Org Catalog");
  });

  test("can search catalog", async ({ uiHelper }) => {
    await uiHelper.openSidebar("Catalog");
    await uiHelper.searchInputPlaceholder("Filter", "component");
  });

  test("can filter by kind", async ({ uiHelper }) => {
    await uiHelper.openSidebar("Catalog");
    await uiHelper.selectMuiBox("Kind", "API");
  });
});

Configuration

// playwright.config.ts
import { defineConfig } from "@playwright/test";
import { baseConfig } from "rhdh-e2e-test-utils/playwright-config";

export default defineConfig({
  ...baseConfig,
  workers: 4, // Run 4 tests in parallel
  fullyParallel: true, // All tests in parallel
});

Serial Tests (Shared Session)

Serial tests share a browser context and run in sequence.

Characteristics

• Shared state: Tests share the same browser session
• Order dependency: Tests run in defined order
• Efficiency: Single login for all tests
• Workflow testing: Perfect for multi-step flows

When to Use

• Testing multi-step workflows
• Tests that build on each other
• Resource-intensive operations (single login)
• State-dependent test sequences

Example

import { test } from "@playwright/test";
import { setupBrowser, UIhelper, LoginHelper } from "rhdh-e2e-test-utils/helpers";

// Configure as serial
test.describe.configure({ mode: "serial" });

test.describe("Entity Creation Workflow", () => {
  let uiHelper: UIhelper;
  let loginHelper: LoginHelper;

  test.beforeAll(async ({ browser }) => {
    // Setup shared browser context
    const context = await setupBrowser(browser);
    const page = await context.newPage();
    uiHelper = new UIhelper(page);
    loginHelper = new LoginHelper(page, uiHelper);

    // Single login for all tests
    await loginHelper.loginAsGuest();
  });

  test("step 1: navigate to create", async () => {
    await uiHelper.openSidebar("Create...");
    await uiHelper.verifyHeading("Create");
  });

  test("step 2: fill template form", async () => {
    // Uses same session from step 1
    await uiHelper.clickButton("Choose");
    await uiHelper.fillTextInputByLabel("Name", "my-new-component");
    await uiHelper.fillTextInputByLabel("Description", "A test component");
  });

  test("step 3: submit and verify", async () => {
    // Uses same session from steps 1-2
    await uiHelper.clickButton("Create");
    await uiHelper.verifyHeading("my-new-component");
  });

  test("step 4: verify in catalog", async () => {
    // Uses same session from previous steps
    await uiHelper.openSidebar("Catalog");
    await uiHelper.verifyRowsInTable(["my-new-component"]);
  });
});

Key Differences from Parallel

Aspect	Parallel	Serial
test.describe	Normal	test.describe.configure({ mode: "serial" })
Fixtures	Use { uiHelper }	Create manually with new UIhelper(page)
Browser setup	Automatic	Use setupBrowser(browser)
Login	Each beforeEach	Once in beforeAll
State between tests	Isolated	Shared

Hybrid Approach

You can mix parallel and serial within the same test file:

import { test, expect } from "rhdh-e2e-test-utils/test";
import { setupBrowser, UIhelper, LoginHelper } from "rhdh-e2e-test-utils/helpers";

// Parallel tests (use fixtures)
test.describe("Parallel Catalog Tests", () => {
  test.beforeEach(async ({ loginHelper }) => {
    await loginHelper.loginAsGuest();
  });

  test("view catalog", async ({ uiHelper }) => {
    await uiHelper.openSidebar("Catalog");
  });

  test("view APIs", async ({ uiHelper }) => {
    await uiHelper.openSidebar("APIs");
  });
});

// Serial tests (shared session)
test.describe("Serial Workflow Tests", () => {
  test.describe.configure({ mode: "serial" });

  let uiHelper: UIhelper;
  let loginHelper: LoginHelper;

  test.beforeAll(async ({ browser }) => {
    const context = await setupBrowser(browser);
    const page = await context.newPage();
    uiHelper = new UIhelper(page);
    loginHelper = new LoginHelper(page, uiHelper);
    await loginHelper.loginAsGuest();
  });

  test("step 1", async () => { /* ... */ });
  test("step 2", async () => { /* ... */ });
});

Best Practices

For Parallel Tests

1. Keep tests independent - Each test should work in isolation
2. Use unique data - Generate unique names to avoid conflicts
3. Clean up after - Remove any created resources
4. Don't rely on order - Tests may run in any sequence

test("create entity with unique name", async ({ apiHelper }) => {
  const uniqueName = `test-${Date.now()}-${Math.random().toString(36).slice(2)}`;

  await apiHelper.importEntity(`https://example.com/${uniqueName}/catalog-info.yaml`);

  // Clean up
  test.afterEach(async () => {
    await apiHelper.deleteEntity(uniqueName);
  });
});

For Serial Tests

1. Use meaningful step names - Makes failures easier to diagnose
2. Handle cleanup properly - Clean up in afterAll, not afterEach
3. Consider failure impact - Later tests may fail if early ones fail
4. Minimize shared state - Only share what's necessary

test.describe("Workflow with proper cleanup", () => {
  test.describe.configure({ mode: "serial" });

  let createdEntity: string;

  test.afterAll(async ({ browser }) => {
    // Clean up after all tests complete (or fail)
    if (createdEntity) {
      // Cleanup logic here
    }
  });

  test("create entity", async () => {
    createdEntity = "my-entity";
    // ... creation logic
  });
});

Performance Comparison

Scenario	Parallel (4 workers)	Serial
10 independent tests	~30 seconds	~2 minutes
5-step workflow	N/A (can't parallelize)	~1 minute
10 tests with shared login	~30 sec + 10 logins	~1 min + 1 login

Decision Flowchart

┌─────────────────────────────────────────┐
│ Do tests depend on each other's state?  │
└─────────────────────┬───────────────────┘
                      │
         ┌────────────┴────────────┐
         │                         │
        Yes                        No
         │                         │
         ▼                         ▼
┌─────────────────┐    ┌─────────────────────┐
│  Use Serial     │    │ Is login expensive? │
│  mode: "serial" │    └──────────┬──────────┘
└─────────────────┘               │
                      ┌───────────┴───────────┐
                      │                       │
                     Yes                      No
                      │                       │
                      ▼                       ▼
             ┌────────────────┐    ┌─────────────────┐
             │Consider Serial │    │  Use Parallel   │
             │for same user   │    │  (default)      │
             └────────────────┘    └─────────────────┘

Related Pages

• Error Handling - Handle failures gracefully
• Playwright Fixtures - Understanding fixtures
• Serial Tests Example - Complete example