Week Ending October 12, 2025

Developer News

The ballots for the Steering Committee Elections are due on October 24th. If you haven’t already, submit your Steering votes. If you have contributed to Kubernetes in the last year but haven’t met the eligibility requirements, you will need to submit an exception request to vote in the steering election, the deadline for which is October 22nd.

The CFP for Maintainer Summit: KubeCon + CloudNativeCon Europe 2026 is open. Please send in your submissions before 14th December 2025.

SIG-Testing is continuing to improve alpha/beta feature coverage, including moving kind-beta-features to release blocking and several other beta jobs to release-informing.

Release Schedule

Next Deadline: Docs Deadline for placeholder PRs, October 23

We are in PRR freeze. Enhancements Freeze will begin this week (16th October). If you are going to miss the deadline, please file an Exception.

Patch releases have been delayed until 22nd October.

Featured PRs

134433 : kubeadm print errors during control-plane-wait retries

This PR improves troubleshooting during control plane startup by ensuring that errors encountered while waiting for control plane components are printed during each retry at log verbosity level 5. Previously, these errors were not shown, which made it harder to identify issues when components failed to become ready. With this change, administrators can now see the actual errors without additional steps, making failure causes more visible and debugging faster.

KEP of the Week

KEP-4622: New TopologyManager Policy which configure the value of maxAllowableNUMANodes

This KEP introduces a new TopologyManager policy option called max-allowable-numa-nodes, allowing users to configure the maximum number of NUMA nodes supported by the TopologyManager. Previously, this value was hardcoded to 8 as a temporary measure to prevent state explosion. By making it configurable, the KEP enables better support for high-end CPUs with more than 8 NUMA nodes, without changing existing TopologyManager policies or addressing broader resource management aspects.

This KEP is tracked as stable in v1.35

Other Merges

• Enforce valid label-key format in device tolerations
• Add declarative validation and path normalization for ResourceClaim fields
• Remove runtime gogo protobuf dependencies from Kubernetes API types
• Fix IPv6 allocator for /64 CIDRs
• Add -n shorthand flag for kubectl config set-context
• Add k8s.update flag to enable validation rules just for updates
• Prevent panic when creating an invalid CronJob schedule
• Stop calling --chunk-size beta, it’s been around since 2017
• Make sure that the eviction controller knows about NoExecute device tolerations
• APIApprovalController can run with contextual logging
• kubeadm: show control plane retry errors
• ResourceClaim: ensure that fields don’t exceed list limits, that shareID is validated, and that it supports the immutable tag and long name format
• Add test for endpoint/endpointslice headless label propagation
• Maybe don’t let folks create ResourceQuotas with request > limit
• kubectl gets -n shorthand for --namespace
• Set FeatureGates simultaneously during tests to avoid dependency problems
• DeviceRequests exactly and firstAvailable shortcut some logic
• Refactor away most of the dependencies on the unmaintained gogo protobuf library
• Allocate within IPv6 subnets correctly
• resource.k8s.io v1 API is now the default
• Prometheus client can handle deprecated/missing metrics
• APIserver will abort startup due to invalid CA configuration

Subprojects and Dependency Updates

• headlamp v0.36.0 adds EndpointSlice support, label-based search, and clipboard copy for resource names.
• cloud-provider-openstack v1.34.1 updates test dependencies and fixes build-script issues across OCCM and CSI plugins. Multiple Helm charts were also updated.
• csi-driver-nfs v4.12.1 updates CSI release tools and documentation for NFS volumes.
• csi-driver-smb v1.19.1 updates CSI release tooling and improves maintenance scripts.
• kubespray v2.29.0 adds new configuration options, supports Kubernetes v1.33.1 and Debian 13 Trixie, and upgrades major components
• prometheus v3.7.0 adds experimental anchored and smoothed rate functions, introduces NHCB, improves rule evaluation and TSDB logging, and deprecates several remote-write metrics.