Continuous Integration Testing
summary
Explains concrete CI integration options and tradeoffs: use docker‑compose with Cloud Build for fast ephemeral builds but lower fidelity; use a shared GKE namespaces staging cluster for high fidelity, quick runs and namespace isolation yet limited CRD/version testing; create self‑destructing k3s VM instances (via Terraform + startup timers) for perfect isolation and inspectable failures; or provision a full GKE cluster per test for maximum fidelity at high cost and slower provisioning.
AI-written summary · about
chapters (28)
- 0:00Intro
- 0:18Continuous Integration Testing
- 0:40Shift Left
- 1:24Shift all the things. Shift them left.
- 3:17Shifting left is good for your business
- 4:00Ideal test pyramid
- 4:58Actual test pyramid
- 5:24Integration testing is hard
- 7:26Why bother?
- 8:43Don't bother?
- 10:03How bad could it be?
- 11:27We can still do integration tests
- 11:52What are we optimizing for?
- 12:03Fidelity
- 12:33Isolation
- 12:43Scalability
- 13:13Speed
- 13:23Ephemerality
- 14:05An example
- 14:58Integration test flow
- 15:53Docker Compose
- 19:07Technique 2: K8s Staging Cluster
- 20:04Self-Destructing VM?
- 21:18Technique 3: Self-destructing VM!
- 22:58K8s cluster per test
- 23:37Integration testing techniques
- 25:02But Dave, what about...
- 27:26Yes we can (still) do integration testing
related talks
26mContinuous Integration Testing
28mContinuous Integration Testing: Fully Test your Microservices Application, Early and Often
24mOrchestrating automation tests for Continuous testing
35mTesting in Infrastructure as Code
41mContinuous Integration Tips & Tricks
30mUsing Docker & Compose to Simplify Integration Testing
45mRunning Tests in Continuous Integration
26mWho broke the build? Using Kuttl to release fast