40Reporting & Observability

40.1 Built-in reporters

// playwright.config.ts
reporter: [
  ['list'],
  ['html', { open: 'never', outputFolder: 'playwright-report' }],
  ['junit', { outputFile: 'results/junit.xml' }],
  ['github'],
],

40.2 HTML report — what's inside

• Pass/fail summary, per project, per file
• Per-test page with steps, errors, attached screenshots / videos / traces
• Filtering by project / status / file
• Link directly to a trace
• Shareable — copy folder, drop into S3 / GitHub Pages

npx playwright test
npx playwright show-report   # opens locally

40.3 Sharding + blob merge

Each shard writes a binary blob report. Merge into one HTML at the end.

# In each shard (CI matrix job)
npx playwright test --shard=1/4 --reporter=blob,line

# In the final merge job
npx playwright merge-reports --reporter=html ./all-blob-reports

40.4 Allure (stakeholder reports)

npm i -D allure-playwright

// playwright.config.ts
reporter: [['line'], ['allure-playwright', { detail: true, suiteTitle: false }]],

npx allure generate ./allure-results --clean -o ./allure-report
npx allure open ./allure-report

Allure gives a richer "feature → story → step" hierarchy, history trends, links to JIRA / Confluence. Common in enterprise where the report is consumed by non-engineers.

40.5 Custom reporter

// reporters/my-reporter.ts
import type { Reporter, TestCase, TestResult } from '@playwright/test/reporter';

export default class MyReporter implements Reporter {
  onTestEnd(test: TestCase, result: TestResult) {
    if (result.status === 'failed') {
      const payload = {
        name: test.title, file: test.location.file, error: result.error?.message,
        retry: result.retry, duration: result.duration,
      };
      /* fire-and-forget to your DB / dashboard */
      fetch('https://internal/qa-events', { method: 'POST', body: JSON.stringify(payload) });
    }
  }
}

reporter: [['list'], ['./reporters/my-reporter.ts']]

40.6 KPIs to track

40.7 Where to put the data

• Postgres for per-test records (test_run table: id, name, status, duration, retry, build, branch)
• ClickHouse / BigQuery if volume is huge (millions of runs)
• Grafana / Looker / Metabase for dashboards
• PagerDuty / Slack for alerts on KPI breaches

40.8 Alerts that earn their keep

• "Pass rate dropped below 95% on main" — page on-call
• "Retry rate > 5% over the last 24h" — open a JIRA
• "PR feedback p95 > 15 min" — investigate infra
• "A new test is failing > 50% of the time" — auto-quarantine after 24h

40.9 Observability of the platform itself

• Worker pod CPU / RAM / disk usage
• Queue depth — backlog warning
• Browser launch time trend
• S3 artifact upload latency
• Cost per worker-hour