testing-strategies

A robust Kailash testing strategy organizes tests into three tiers: unit, integration, and end-to-end. It uses real infrastructure for tiers 2-3 and Docker-based environments with proper fixtures to validate workflows across databases, APIs, and LLMs.

Avoiding mocks in integration and end-to-end testing ensures you validate real infrastructure behavior across databases, APIs, and LLMs. The testing strategy recommends using Docker-based test environments to provide reproducible real infrastructure parity for tiers 2 and 3.

Docker-based test environments provide runtime parity for integration and e2e tests. The testing strategy defines fixture patterns and Docker configurations that ensure reproducible tests across LocalRuntime and AsyncLocalRuntime without relying on mocked external services.

The best way to test Kailash workflows with databases and LLMs is applying a 3-tier testing strategy. Tiers 2 and 3 use real infrastructure via Docker-based test environments with established fixtures, ensuring reliable end-to-end validation across all connected services.

Yes, the testing strategy supports runtime parity testing across both LocalRuntime and AsyncLocalRuntime. It provides fixture patterns and Docker-based test environments that ensure reproducible test execution for Kailash workflows regardless of the runtime used.

Real infrastructure testing is required for integration and end-to-end tiers (2 and 3) when validating Kailash workflows across databases, APIs, and LLMs. Unit tests in tier 1 can use mocks, but tiers 2-3 must avoid mocks to catch edge cases and prevent flaky deployments.