Back to updates

K8s - 2

2026 (12)

September (2)

k8s: Kubernetes v1.37: Scale Workloads to Zero with HorizontalPodAutoscaler

Kubernetes Official Blog

2026-09-02

Kubernetes v1.37 makes HorizontalPodAutoscaler scale workloads down to zero replicas by default when using object or external metrics. The feature is Beta; it requires a metric source that remains available at zero replicas, and the controller marks zero-state ownership with a ScaledToZero condition.

k8s: Kubernetes v1.37: etcd RangeStream Cuts Memory Use on Large List Reads

Kubernetes Official Blog

2026-09-01

Kubernetes v1.37 graduates the EtcdRangeStream feature to beta and enables it by default. With etcd v3.7+, the API server can stream large collection reads instead of assembling whole pages in memory, reducing peak memory use and making it more predictable.

August (6)

k8s: Kubernetes v1.37: Storage Version Migration Enabled by Default

Kubernetes Official Blog

2026-08-31

Kubernetes v1.37 makes built-in StorageVersionMigration (storagemigration.k8s.io/v1) GA and enabled by default in all clusters. It lets administrators and CRD authors declaratively rewrite stored objects to the current storage version, replacing manual re-write scripts or the out-of-tree migrator.

k8s: Kubernetes v1.37: Pod Certificates and Cluster Trust Bundles

Kubernetes Official Blog

2026-08-28

Kubernetes 1.37 graduates Pod Certificates and Cluster Trust Bundles to GA, adding built-in support for workload X.509 identity and trust bundle distribution. The feature uses Kubelet, PodCertificateRequest objects, and signer controllers to issue and rotate certificates for TLS/mTLS while keeping key material on the node filesystem.

k8s: Kubernetes v1.37: Metrics API graduates to stable

Kubernetes Official Blog

2026-08-27

Kubernetes v1.37 graduates the Metrics API from v1beta1 to stable v1. The API surface is unchanged, and v1beta1 remains available for compatibility.

k8s: Kubernetes v1.37: Garhwal

Kubernetes Official Blog

2026-08-26

Kubernetes v1.37 ships 67 enhancements: 16 Stable, 23 Beta, 27 Alpha, and 1 deprecation/removal. Key changes include resilient watchcache initialization, HPA scale-to-zero, manifest-based admission config, Pod checkpoint/restore, broader Dynamic Resource Allocation support, and multiple scheduler, storage, node, and observability updates.

k8s: How to Pretty-Print Your Kubernetes YAML as KYAML and Why You'd Want To

Kubernetes Official Blog

2026-08-11

Kubernetes introduced KYAML, a strict YAML subset that uses explicit braces, brackets, quotes, and trailing commas to avoid common YAML pitfalls like indentation errors and silent type coercion. The post explains how to output or format manifests as KYAML with kubectl and yamlfmt, and notes that KYAML remains valid YAML with no new parser required.

k8s: Gateway API v1.6: TCPRoute and UDPRoute Graduate to Standard

Kubernetes Official Blog

2026-08-03

Gateway API v1.6.0 promotes TCPRoute and UDPRoute to Standard/v1, adding GA-level raw TCP and UDP routing. It also moves experimental APIs into a separate gateway.networking.x-k8s.io group with X-prefixed kinds to make the experimental boundary explicit.

July (4)

k8s: Kubernetes v1.37 Sneak Peek

Kubernetes Official Blog

2026-07-31

Kubernetes v1.37 is planning several deprecations and a few feature graduations. Notable changes include deprecating kubectl run -f, forbidding Static Pods from referencing Secrets/ConfigMaps, starting deprecation of kube-proxy ipvs mode, and moving SELinuxMount to GA with possible workload impact.

k8s: How the controller-runtime Cache Actually Works, and Why Your Controller Does Not Crash the API Server

Kubernetes Official Blog

2026-07-29

The article explains that controller-runtime controllers read from a local informer cache, not directly from the API server, and that writes go to the API server while the cache catches up asynchronously via watch events. It also details how shared informers, DeltaFIFO, indexing, cache scoping, APIReader, and cache disabling affect consistency, memory use, and query behavior.

k8s: Building a Custom Metrics Exporter for Kubernetes

Kubernetes Official Blog

2026-07-14

The post explains how to build a custom Prometheus metrics exporter for Kubernetes in Go, package it as a minimal container image, and expose it for scraping by Prometheus. It also shows how to verify the target in Prometheus and notes that a metrics adapter such as Prometheus Adapter is needed if you want to use the exported metrics for HorizontalPodAutoscaler scaling.

k8s: Operating AI/ML Workloads on Kubernetes: A Headlamp Plugin for Kubeflow

Kubernetes Official Blog

2026-07-13

Kubernetes announced a Headlamp plugin for Kubeflow that surfaces Kubeflow CRDs and related Pod-level state directly in the Headlamp Kubernetes UI. The plugin is aimed at operators and SREs troubleshooting ML workloads and works with modular Kubeflow installs by discovering available API groups at runtime.