Kubernetes Conformance Deep Dive
summary
Collects API server audit logs (currently via SSH to the master because managed control planes like GKE block it) and loads raw JSON into SQLite to produce CSVs and usage charts, using API snoop to prioritize untested endpoints by real-world calls rather than topic lists. Finds only ~18% stable API coverage in v1.10, sets goal to raise coverage +40% by Q3 and push top three APIs to 80%. Integrates sonobuoy and a test results upload for TestGrid, and experiments with darker and darker for hermetic runs.
AI-written summary · about
chapters (21)
- 0:00Start
- 1:00Kubernetes API Consumers
- 1:33Audit Webhook
- 2:18API Usage Logging
- 5:12Conformance for KBs Implementations
- 12:34Scaffold Untested Stable API Usage
- 15:14Data Driven Conformance
- 16:29OSS Challenges
- 18:45Charter
- 18:59Defining the Process
- 19:48Improving Conformance Test Coverage
- 20:37Communication
- 20:49How to increase test coverage?
- 21:20Test Authoring Guidelines
- 22:58Where to focus?
- 23:41Hermetic Conformance testing
- 24:16DIND Testing Limitation
- 27:22Conformance Dashboard on TestGrid
- 27:36Openstack Conformance Test Results
- 30:43Revisiting the framework
- 31:15Roadmap for 2018
related talks
20mFilling the Gaps in Kubernetes Test Coverage - Hippie Hacker, ii.coop
23mContributing to Kubernetes Conformance Coverage
21mFilling the Gaps in Kubernetes Test Coverage
20mConformance Testing the Kubernetes API: Tooling that Makes the Difference
01h 27mIntro + Deep Dive: Kubernetes Conformance WG
32mIntro + Deep Dive: Kubernetes Conformance WG
35mK8s Conformance & SIG Architecture Conformance Subgroup
27mAdventures in Conformance