Metrics as a First-Class Citizen in the E2E Testing Landscape
summary
Claims that instrumented apps can drive E2E tests by asserting on exposed Prometheus metrics rather than external exit codes, implemented with a Go e2e framework, Docker and Prometheus. The demo spins up an instrumented example app and Prometheus, waits until head_samples > 50 as a precondition, opens a browser and requires 5 HTTP requests observed by http_request_total before completing. Applied to Thanos and Observatorium repositories, the metric-driven interactive tests run successfully and enable fine-grained control, retries, and fail‑fast behavior.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 0:44About Matej
- 1:09Agenda
- 2:57Why do you need metrics
- 3:41Resources
- 3:58Instrumentation
- 4:44Metrics in Testing
- 6:31What is even better
- 6:56Making more detailed test assertions
- 7:50Better control over test scenarios
- 8:54More complex scenarios
- 9:22Extra data points
- 9:53Test types
- 11:51Unit test integration test entry test
- 14:21The Framework
- 18:47Demonstration
- 24:22Demo
- 27:38Running the test
related talks
45mMicroProfile Metrics: Practical Use Cases
22mPrometheus instrumentation: the Practical Way
27mGrand Treatise of Modern Instrumentation and Orchestration
32mSynthetic Monitoring and E2E Testing: 2 Sides of the Same Coin
21mImproving Monitoring Systems Interoperability with Prometheus & OpenMetrics
38mPerformance Testing of Prometheus Based Metric Platforms
30mSynthetic Monitoring and E2E Testing: 2 Sides of the Same Coin
52mIs your JVM app flying blind? Unmask hidden issues with Observability superpowers!