GeeCON Prague 2017: Anton Arhipov - TestContainers - integration testing without the hassle

### Overview In this GeeCON Prague 2017 presentation, Anton Arhipov from ZeroTurnaround introduces the Testcontainers Java library and explains how it simplifies integration testing. Integration testing is critical because unit tests alone—even with 100% code coverage—fail to detect bugs arising from real external interactions, such as network protocols, driver behavior, and database engine quirks. Arhipov illustrates this with an incident where moving from Java 1.4 to 1.5 modified the `BigDecimal.toString()` implementation, causing an Oracle database driver to insert millions of euros instead of cents because the test suite relied on an in-memory mock database rather than a real database.

### Key Topics - **Challenges of Integration Testing**: Setting up testing environments is traditionally difficult due to complex manual configurations, long onboarding times, resource contention, and slow feedback cycles. Running concurrent tests often causes collisions over static ports or local resources. - **Requirements for Effective Integration Tests**: Modern integration test suites require reproducible environments across developer machines (Mac, Linux, Windows) and continuous integration (CI) systems, automatic resource isolation, ease of maintenance, and testing against real dependencies rather than mocks. - **Docker and Testcontainers**: Testcontainers acts as a wrapper around the `docker-java` library, providing JUnit rules and fluent Java APIs to manage container lifecycles directly within JUnit tests. It automates container discovery, pulling images, port mapping, and teardown.

### Technical Details - **Dynamic Port Mapping**: Instead of hardcoding port bindings that cause collisions during parallel execution, Testcontainers dynamically binds container ports (e.g., Redis port 6379 or Jetty port 8080) to random, sequential local host ports, accessible programmatically via helper methods. - **GenericContainer and MockServer**: For dependencies without dedicated modules (such as Redis or PostgreSQL), users can instantiate `GenericContainer`. The talk demonstrates mocking third-party microservices over the network using MockServer inside a container, avoiding in-process JVM mocking. - **Testing JVM Agents**: Arhipov demonstrates how ZeroTurnaround tests monitoring agents that instrument bytecode. Testing involves packaging a lightweight Groovy test script using the Spark framework and embedded Jetty, mounting the local agent JAR and Groovy scripts into a Groovy Docker container via classpath resource mapping, configuring `JAVA_OPTS` to pass `-javaagent`, waiting for readiness via a wait strategy, and asserting against HTTP response headers with a Feign client.

### Takeaways Testcontainers eliminates the need for proprietary setup scripts and manual wiki-driven environment setups. By running disposable, containerized dependencies on demand directly from JUnit, teams achieve isolated, reliable, and cross-platform integration tests that closely mirror production environments.