The Life (or Death) of a Kubernetes API Request, 2025 Edition
Talkat KubeCon + CloudNativeCon Europe 2025 · on Wednesday, 2025-04-02· at 15:15· Level 1 | Hall Entrance S10 | Room C·
summary
Chronicles the literal request path from kubectl to storage and back, showing how the HTTP server builds a handler chain, populates request context and audit events, and applies defaulting, conversion and admission before writing to etcd with protobuf and KMS encryption. Explains timeout enforcement with a 60s default deadline and how API Priority and Fairness uses APF 1/4 deadline = 15s to reject queued requests with 429/retry-after. Shows observable artifacts: panic manifests: EOF (HTTP/1) or stream reset (HTTP/2) and audit IDs tying logs, metrics, and audit annotations.
AI-written summary · about
chapters (11)
- 0:00Introduction and context
- 2:26Kind, resource, and mapping
- 3:32API request handler chain
- 6:54Panic recovery mechanism
- 9:36Request parsing and info
- 11:39Latency and logging
- 13:54Deadlines and timeouts
- 17:36Priority and fairness
- 20:40Resource handling and versions
- 23:44Storage and completion
- 28:58Client retries and conclusion
related talks
44mLife of a Kubernetes API Request
33mKubernetes API Server Performance Clinic: Auditing, and Priority...
25mAuditing in Kubernetes 101
05mKubernetes API Explained
22m2024: I Don't Know K8S and at This Point, I'm Too Afraid To Ask
08mKubernetes: The API of Everything
34mExtending the Kubernetes API: What the Docs Don't Tell You
32mInto the Deep Waters of API Machinery