Kubernetes API Governance: The Gatekeeper Devs Can't Ignore
Your Kubernetes clusters stay up because someone sweats the API details. Jordan Liggitt's SIG Architecture work ensures changes don't wreck production.
Your Kubernetes clusters stay up because someone sweats the API details. Jordan Liggitt's SIG Architecture work ensures changes don't wreck production.
Picture this: your zonal disk just got migrated to regional storage, but Kubernetes won't let Pods schedule anywhere new. v1.35's mutable PV node affinity aims to fix that – if you dare enable alpha.
Kubernetes pods get preempted 40% of the time in busy clusters, torching hours of compute. The new Checkpoint/Restore WG promises to freeze and thaw them smoothly — but I've seen this movie before.
Prod debugging in Kubernetes shouldn't mean handing out god-mode keys. Here's the enthusiastic blueprint for secure, speedy access that scales to AI workloads.
Imagine rebuilding a rocket's engine mid-flight. That's exactly what the Kubernetes team did with kpromo, the invisible hero behind every container image pull—and nobody noticed. Here's why it matters.
Staring down kubectl's endless options? Clientcmd lets you borrow Kubernetes' CLI smarts without the headache. Here's how — and why it's not just another library.
Hit 'go install' and watch your messy Ingress-NGINX manifests spit out clean Gateway API YAML. With retirement looming in March 2026, this tool isn't optional—it's survival gear for Kubernetes ops.
Your Kubernetes node flips to 'Ready' too soon, pods crash spectacularly. Node Readiness Controller steps in with smart taints to wait for the full infrastructure handshake.
Kubernetes is killing off Ingress NGINX, the ingress controller half your clusters rely on. No more security fixes after March 2026—time to panic or plan?
Forget endless machine recreations—Cluster API v1.12 introduces in-place updates that tweak live nodes without the chaos. And chained upgrades? They skip the manual minor-version hopscotch.
Kubernetes isn't waiting for AI chaos to hit critical mass. A fresh Working Group targets gateways built for inference traffic—rate limits, payload checks, the works.
AI agents are ditching quickie inference calls for long-haul marathons. Enter Kubernetes' Agent Sandbox: a CRD promising to wrangle them without turning your cluster into a madhouse.