Testing Basic

This section is meant to be read in order: first learn which test catches which bug, then learn why tests lose their signal and how to fix them, then load testing and testing against real dependencies.

It suits backend developers new to writing tests, or tech leads cleaning up a CI that keeps going red for no reason. The section does not yet cover mutation testing or frontend testing.

Reading path

#	Article	You will be able to answer	Level	Read
1	Unit, integration, e2e	Which layer should catch this bug? When do you need a real DB or a contract test? Pyramid or trophy?	Basic	~12 min
2	Flaky tests	Why does a test flip between green and red? How do you fix clock, ordering, shared state and async issues? When do you quarantine?	Basic	~12 min
3	Load testing with k6	How many RPS can the API take at the target p95? Why does a closed model hide a slow server? How do thresholds fail CI?	Intermediate	~15 min
4	Testcontainers	How do you run tests against real Postgres, Redis and Kafka without slow, shared state? Rollback, truncate or template database?	Intermediate	~15 min

Article 2 assumes you already understand the test-layer boundaries from article 1.

If you run into…

Symptom	Read
Unit tests are green but production still breaks on a response/field	Which bug, which test layer
A mocked repository passes but production hits a unique constraint error	Real database and Testcontainers
Two services drift apart on the API schema after a deploy	Contract tests
CI is red, a rerun is green	Classifying flaky causes
A TTL/expiry test uses `sleep` and fails often	Sleep and an injected clock
Fails only when run in parallel or as a full suite	Classifying flaky causes
The team is used to retrying until green	Quarantine and flake policy

Test boundaries are easier to draw when code boundaries are clear: the Strategy pattern separates what varies into an independently testable policy, and the modular monolith discusses boundaries at the module level. For persistence risks see ACID and transactions.