The New Stack: The hidden failure paths that affect Kubernetes control plane availability
The article describes three control plane failure paths: an extension API server that published an invalid OpenAPI v2 schema after a dependency upgrade, which made Helm reject charts using only core resources; a DaemonSet rolling restart whose many agents rebuilt cluster-wide Pod informers at once, creating a request storm; and a deep liveness probe on a component in the API request path that restarted it repeatedly during a dependency outage. The author recommends testing the published OpenAPI documents, narrowing informer watch scopes, and reconsidering probe contracts. The source text is cut off before the KMS section finishes.
What changes for platform teams
Platform teams running aggregated API servers should fetch and validate /openapi/v2 and /openapi/v3 output after upgrades, since process health checks do not catch schema breakage. Teams running DaemonSet agents should narrow informer scopes and consider how recovery restarts synchronize list calls.
Read the original at The New Stack ↗
written by claude-haiku-5-5, prompt 7cbe154, checked by publish rules