Avoid Spikes: Unexpected Kubernetes Behaviors
summary
Kubernetes releases three times a year and that cadence causes surprising breakage: pod security policy deprecated in 1.21 removed in 1.25 forces migration off a long‑used beta feature; CoreDNS default ndot=5 causes extra DNS lookups that bite at scale; service scaling moves to EndpointSlice introduced in 1.19 changing how endpoints are represented; rolling updates create temporary extra pods so tune maxSurge to avoid unexpected capacity and cost spikes.
AI-written summary · about
chapters (4)
related talks
35mKubernetes, Lessons Learnt
18mKubernetes Project Update
34mLatest Kubernetes Scalability Improvements
36mKubernetes Confessions: Real Mistakes We’ve All Made (and What They Teach Us)
36m10 More Weird Ways to Blow Up Your Kubernetes
37m10 Ways to Shoot Yourself in the Foot with Kubernetes, #9 Will Surprise You
19mUpgrading Your K8s Clusters: Key Considerations for a Smooth Transition
28mHow to Break your Kubernetes Cluster with Networking