Weekly GitHub Report for Kubernetes: November 03, 2025 - November 10, 2025 (12:05:45)

Weekly GitHub Report for Kubernetes

Thank you for subscribing to our weekly newsletter! Each week, we deliver a comprehensive summary of your GitHub project's latest activity right to your inbox, including an overview of your project's issues, pull requests, contributors, and commit activity.

---

Table of Contents

• I. News
  • 1.1. Recent Version Releases
  • 1.2. Other Noteworthy Updates
• II. Issues
  • 2.1. Top 5 Active Issues
  • 2.2. Top 5 Stale Issues
  • 2.3. Open Issues
  • 2.4. Closed Issues
  • 2.5. Issue Discussion Insights

• III. Pull Requests
  • 3.1. Open Pull Requests
  • 3.2. Closed Pull Requests
  • 3.3. Pull Request Discussion Insights

• IV. Contributors
  • 4.1. Contributors

I. News

1.1 Recent Version Releases:

The current version of this repository is v1.32.3

1.2 Version Information:

The Kubernetes version released on March 11, 2025, introduces key updates detailed in the official CHANGELOG, with additional binary downloads available. For comprehensive information on new features and changes, users are encouraged to refer to the Kubernetes announce forum and the linked CHANGELOG.

II. Issues

2.1 Top 5 Active Issues:

We consider active issues to be issues that that have been commented on most frequently within the last week. Bot comments are omitted.

1. [InPlacePodVerticalScaling] e2e tests fail when upgrading containerd v2.1.5 on kOps: This issue reports that end-to-end tests for the InPlacePodVerticalScaling feature fail when upgrading containerd to version 2.1.5 on kOps, due to unexpected CPU cgroup weight values caused by changes in runc v1.3.3. The discussion centers on whether to adjust test expectations to accommodate the new cgroup behavior, the need for backporting fixes to older Kubernetes releases, and confirming that recent patches have addressed the problem in newer test runs.

   • The comments reveal that the change in CPU weight handling was intentional in runc, leading to test failures with the new containerd version; maintainers consider adding new expected values rather than version-specific tests. They confirm a recent Kubernetes PR addresses the issue for newer versions, discuss the implications for backporting to older releases, and seek clarification on testing policies for containerd and runc versions across supported branches.
   • Number of comments this week: 16

2. failing device-plugin health checks causing kubelet SIGABRT by systemd watchdog during initialization.: This issue describes a problem where the kubelet process running under systemd crashes due to failing device-plugin health checks triggered by the systemd watchdog during kubelet initialization. The root cause is that the updateRuntimeUp function can be blocked, preventing timely watchdog updates and causing kubelet to crash-loop, with the device-plugin health check failing because it expects initialization to be complete before passing.

   • The discussion focused on confirming the severity and priority of the bug, exploring whether it was a regression, and proposing a solution to extend the HealthChecker interface with an IsInitialized() method to distinguish between components that are still initializing versus truly unhealthy. Commenters also debated the impact of slow dependencies like CNI plugins on startup and watchdog behavior, emphasizing the need for better handling of initialization states and timeouts to prevent false health check failures and unnecessary kubelet restarts.
   • Number of comments this week: 15

3. The pod has been successfully removed, but the volume’s project quota mapping still remains on the node: This issue describes a problem where, after deleting a pod with the LocalStorageCapacityIsolationFSQuotaMonitoring feature enabled and hostUsers set to false, the project quota mappings for the pod’s volume are not properly cleaned up on the node, leaving stale entries in /etc/project and /etc/projid. The root cause appears to be that during volume teardown, only the pod UUID is available and the pod spec (including the hostUsers field) is nil, causing the quota cleanup logic to be bypassed and resulting in leftover quota mappings.

   • The discussion confirmed the issue is specific to the LocalStorageCapacityIsolationFSQuotaMonitoring feature and does not block user namespace GA. It was explained that the pod spec is not fully populated during volume teardown, which prevents proper quota cleanup. Contributors agreed the cleanup should always run regardless of user namespace settings, and a detailed reproduction setup was provided involving a single-node cluster with a specially prepared ext4 filesystem supporting project quotas. The issue was triaged as important-longterm and referred to SIG node and SIG storage for further input on handling cleanup failures.
   • Number of comments this week: 7

4. apiserver: Watch cache consistency check failure constantly panics apiserver instances: This issue describes a problem encountered when upgrading to Kubernetes 1.33 with the environment variable KUBE_WATCHCACHE_CONSISTENCY_CHECKER enabled, which causes the apiserver instances to constantly panic due to a "Cache inconsistency check failed" error, leading to repeated termination and restarts. The problem was observed in two clusters, with one recovering after a few hours and the other remaining unstable for seven days until the consistency checker was disabled, and no stable reproduction steps have been identified.

   • The comments discuss a possible root cause related to the watch cache processing events individually rather than transactionally, though this theory is considered unlikely due to Kubernetes’ single-key transaction model; contributors note that the consistency checker is intended for testing and not production, and a pull request was created to add logging for better debugging if the issue recurs.
   • Number of comments this week: 6

5. Live container migration from one node to another to support long-running AI/ML workloads: This issue proposes adding the capability to live migrate containers from one node to another within Kubernetes, specifically to support long-running AI/ML workloads that are costly to restart and prone to resource exhaustion. The feature aims to address scenarios such as moving containers to nodes with more resources or migrating from pre-empted spot nodes, thereby improving the efficiency and reliability of running large-scale batch jobs and model training on Kubernetes.

   • The comments reveal that a working group has been formed to tackle this feature, with the main challenge now being how to properly integrate it into Kubernetes rather than technical feasibility. The original poster expressed interest in joining or observing the working group discussions, and was encouraged to follow the group's progress once the necessary communication infrastructure is established.
   • Number of comments this week: 4

2.2 Top 5 Stale Issues:

We consider stale issues to be issues that has had no activity within the last 30 days. The team should work together to get these issues resolved and closed as soon as possible.

1. Zone-aware down scaling behavior: This issue describes a problem with the horizontal pod autoscaler's (HPA) scale-in behavior in a Kubernetes deployment that uses topology spread constraints to evenly distribute pods across zones. Specifically, during scale-in events, the pods become unevenly distributed with one zone ending up with significantly fewer pods than allowed by the maxSkew: 1 setting, causing high CPU usage on the lone pod in that zone and violating the expected balanced pod distribution.
2. apimachinery's unstructured converter panics if the destination struct contains private fields: This issue describes a panic occurring in the apimachinery's DefaultUnstructuredConverter when it attempts to convert an unstructured object into a destination struct that contains private (non-exported) fields. The reporter highlights that the converter should ideally ignore these private fields instead of panicking, especially since protobuf-generated gRPC structs often include private fields that cause this failure even when only public fields are present in the unstructured data.
3. Integration tests for kubelet image credential provider: This issue proposes adding integration tests for the kubelet image credential provider, similar to the existing tests for client-go credential plugins. It suggests that since there are already integration tests for pod certificate functionality, implementing tests for the kubelet credential plugins would be a logical and beneficial extension.
4. conversion-gen generates code that leads to panics when fields are accessed after conversion: This issue describes a bug in the conversion-gen tool where it generates incorrect conversion code for structs that have changed field types between API versions, specifically causing unsafe pointer conversions instead of properly calling the conversion functions. As a result, accessing certain fields like ExclusiveMaximum after conversion leads to runtime panics, highlighting the need for conversion-gen to produce safe and correct code to prevent such crashes.
5. Failure cluster [ff7a6495...] TestProgressNotify fails when etcd in k/k upgraded to 3.6.2: This issue describes a failure in the TestProgressNotify test that occurs when the etcd component in the Kubernetes project is upgraded to version 3.6.2. The test times out after 30 seconds waiting on a result channel, with multiple errors indicating that the embedded etcd server fails to set up serving due to closed network connections and server shutdowns.

2.3 Open Issues

This section lists, groups, and then summarizes issues that were created within the last week in the repository.

Issues Opened This Week: 31

Summarized Issues:

• Performance Testing and Optimization: This topic covers issues related to ensuring Kubernetes scheduler performance tests include experimental allocators and setting thresholds to prevent regressions. It also includes proposals for performance improvements such as using node hints when nominated node names fail and tracking flaky tests affecting resource allocation.
• [issues/135058, issues/135163, issues/135185]

• Resource Management and Allocation Issues: Several issues address problems with resource allocation, including deferred pod resizes causing overcommitment, inconsistencies in pod resource handling affecting QoS classification, and race conditions delaying PVC to PV binding. These issues highlight challenges in resource tracking and scheduling efficiency.
• [issues/135082, issues/135107, issues/135127]

• Dynamic Resource Allocation (DRA) Enhancements and Metrics: This topic includes migration of declarative validation for DRA resource groups, smarter allocation result formatting for device classes, and discussions on unifying metric naming for dynamic resource allocation subsystems to improve API tooling and observability.
• [issues/135073, issues/135074, issues/135075]

• Kubelet and Container Manager Improvements: Issues here focus on refactoring kubelet and containermanager code to reduce technical debt, adding declarative on-demand debugging features, and fixing goroutine leaks in dynamic plugin probing to improve maintainability and reliability.
• [issues/135076, issues/135110, issues/135162]

• Pod and Controller Status Handling: Problems include leftover volume quota mappings after pod deletion, kube-controller-manager failing to update pod NotReady status after restart, and inability to update multiple pod conditions atomically in the scheduler plugin, all affecting pod lifecycle and status accuracy.
• [issues/135063, issues/135159, issues/135205]

• Cluster Stability and Upgrade Failures: This covers critical pod crashes after upgrading to Kubernetes 1.33.5 causing cluster unusability, apiserver panics due to watch cache consistency checks during upgrades, and node authorization race conditions causing 403 errors on startup or scale-up.
• [issues/135113, issues/135115, issues/135169, issues/135175]

• API and CLI Usability Issues: Issues include unsupported field selectors on ReplicaSet resources causing errors, removal of deprecated HPA API versions to reduce complexity, and a request for a ShareID test to ensure no interference with existing enhancements.
• [issues/135136, issues/135141, issues/135190]

• Logging and Debugging Challenges: This topic addresses ambiguity in klog's file path stripping that hinders fine-grained logging control and causes confusion in verbosity settings, impacting debugging and log analysis.
• [issues/135198]

• Deployment and Scaling Controller Bugs: Problems with duplicated deployment controller methods causing rollout test failures and improper handling of replica set annotations highlight the need for careful testing and cleanup in scaling logic.
• [issues/135222]

• Node and Runtime Performance Issues: Reports include increased CPU usage after kube-proxy upgrade linked to garbage collection and sync operations, and container runtime blackbox test failures due to unreachable private registries, affecting node and runtime stability.
• [issues/135230, issues/135233]

• Container and Workload Migration Features: A proposal to add live container migration capabilities aims to support resource-intensive AI/ML workloads by enabling seamless transfer between nodes, improving workload reliability and resource utilization.
• [issues/135178]

• Test Failures Due to External Dependencies: Failures in Pod InPlacePodVerticalScaling tests caused by containerd and runc version changes demonstrate how external runtime updates can break Kubernetes test expectations.
• [issues/135214]

• Build and Tooling Warnings: A build warning on Ubuntu 25.10 with uutils coreutils due to an unbound DATE variable in a version script indicates tooling compatibility issues affecting developer workflows.
• [issues/135210]

2.4 Closed Issues

This section lists, groups, and then summarizes issues that were closed within the last week in the repository. This section also links the associated pull requests if applicable.

Issues Closed This Week: 16

Summarized Issues:

• End-to-end test failures in kinder workflows: Multiple end-to-end tests in the sig-cluster-lifecycle kinder workflows have been consistently timing out or exiting with status 1 since October 18, 2025, causing failures in related Kubernetes CI jobs. These persistent failures affected jobs such as sig-release-master-informing and kubeadm-kinder-latest until they were eventually resolved.
• issues/135039, issues/135040

• Container runtime test failures due to client rate limiting: Container Runtime blackbox tests and container runtime conformance tests have been failing due to client rate limiter errors causing context deadline exceeded timeouts. These issues appeared after specific pull requests and affect jobs like pull-kubernetes-node-kubelet-serial-containerd and private registry image pulls with credential providers.
• issues/135041, issues/135132

• CRI interface container filtering behavior: The CRI interface's ContainerFilter returns nil instead of an error or a list when multiple containers match a partial container ID, leading to unexpected empty results in commands like crictl ps --id <partial-id>. This behavior causes confusion and unexpected outcomes when querying containers by partial IDs.
• issues/135049

• Scheduling throughput degradation in benchmarks: The SteadyStateClusterResourceClaimTemplate benchmark in the ci-benchmark-scheduler-perf-master job has shown consistently low scheduling throughput since October 31st, causing test failures. This issue has prompted investigation and fixes within the Kubernetes scheduling SIG to restore acceptable performance levels.
• issues/135061

• Security vulnerabilities in kubectl release: The kubectl v1.33.5 release was flagged with multiple security vulnerabilities in the Go standard library, prompting requests to rebuild the binary using Go version 1.24.8 or later. This action is necessary to address the identified security concerns and ensure a secure release.
• issues/135083

• Kubeadm kinder dry-run upgrade failures due to CRI socket issues: The kubeadm kinder dry-run upgrade workflow fails preflight checks because it cannot connect to the container runtime due to invalid or missing CRI socket addresses. This prevents validation of container runtime version compatibility and causes upgrade failures.
• issues/135086

• Kubelet panic from concurrent map writes with emptyDir volumes: A fatal concurrent map write error occurs in the kubelet when creating and deleting pods with emptyDir volumes simultaneously while the LocalStorageCapacityIsolationFSQuotaMonitoring feature gate is enabled. This leads to a panic caused by improper locking in quota support maps during volume teardown.
• issues/135089

• Duplicate kubeadmConfigPatches causing kind presubmit job failures: DRA kind presubmit jobs fail due to duplicate top-level kubeadmConfigPatches entries in the generated /tmp/kind.yaml file. The presubmit script appends a kubeadmConfigPatches block to a file that already contains one, resulting in YAML parsing errors after rebasing onto the latest master.
• issues/135099

• RemoteEndpoints not deleted on Deployment scale down in AKS Windows NodePools: RemoteEndpoints persist stale when a Deployment referenced by multiple Services is scaled down or terminated in large-scale AKS environments with Windows NodePools. This is likely caused by race conditions or delayed cleanup in the controller logic, leading to resource inconsistencies.
• issues/135144

• ValidatingAdmissionPolicy crashes with CRD matchConstraints: A ValidatingAdmissionPolicy referencing a CRD with a property defined as type: object and additionalProperties: true in its matchConstraints causes the kube-controller-manager to panic repeatedly. This prevents stable reconciliation of these policies due to type-checking failures.
• issues/135145

• Deployment iterative rollouts test failures: The Kubernetes deployment iterative rollouts test experiences persistent failures where deployment status does not progress as expected. Although a recent revert temporarily mitigated the issue, further investigation is required to fully resolve the increased failure rates.
• issues/135150

• DRA extended resource quota test timeouts due to resource accounting discrepancies: The DRA extended resource quota test was timing out because of discrepancies in resource usage accounting after a fix to initcontainer resource accounting. It was clarified that the failure was not a flake, and a corrective fix was merged to update the test accordingly.
• issues/135177

• Flaky scheduler dynamic resources plugin test due to CEL runtime error: A flaky test failure in the Kubernetes scheduler's dynamic resources plugin occurs sporadically due to a CEL runtime error caused by a missing "healthy" key in CEL evaluation. This is attributed to a race condition from a missing HasSynced check on the resource slice informer.
• issues/135184

• MirrorPod EnvFiles failing on CRI-O runtime: The MirrorPod feature with EnvFiles fails to consume environment variables from a file when running on CRI-O container runtime lanes. This results in container creation errors due to unparseable environment files.
• issues/135223

2.5 Issue Discussion Insights

This section will analyze the tone and sentiment of discussions within this project's open and closed issues that occurred within the past week. It aims to identify potentially heated exchanges and to maintain a constructive project environment.

Based on our analysis, there are no instances of toxic discussions in the project's open or closed issues from the past week.

---

III. Pull Requests

3.1 Open Pull Requests

This section provides a summary of pull requests that were opened in the repository over the past week. The top three pull requests with the highest number of commits are highlighted as 'key' pull requests. Other pull requests are grouped based on similar characteristics for easier analysis. Up to 25 pull requests are displayed in this section, while any remaining pull requests beyond this limit are omitted for brevity.

Pull Requests Opened This Week: 58

Key Open Pull Requests

1. KEP 5598: Opportunistic Batching: This pull request introduces opportunistic batching as described in KEP 5598 to improve scheduling performance for pods with similar configurations by grouping them together, incorporating pod signatures, fixing batch size limitations, and integrating these changes with existing performance tests.

• URL: pull/135231

• Merged: No

• Associated Commits: 0e6f3, 5716e, 818a2, 964d0, 63c75, 95719, 1e5bb, 466f3

2. [WIP] KEP-4671: Add Workload API. Copy of PR #134564: This pull request proposes adding a new Workload API based on KEP-4671, including the introduction of a WorkloadReference to the Pod specification, along with related feature gates, kubectl enhancements, and end-to-end tests, while excluding the gang scheduling implementation covered in a separate PR.

• URL: pull/135143

• Merged: No

• Associated Commits: 55554, 8a560, 24a10, e5de6, 6c84e, 310be, af49e

3. KEP-4671: Add Declarative Validation to Workload API: This pull request adds declarative validation to the Workload API in Kubernetes by introducing validation tags and immutability tests for various workload-related fields, enhancing the robustness and correctness of workload scheduling configurations.

• URL: pull/135164

• Merged: No

• Associated Commits: 89244, de8f5, 8fc53, 8de74, 559c9, 86e87, 7375b

Other Open Pull Requests

• Device Resource Allocator (DRA) Enhancements: Multiple pull requests improve the Kubernetes Device Resource Allocator by implementing a tombstone mechanism to retain pod resource claims after termination, adding support for human-readable device health messages, refactoring scheduler plugin code, adding unit tests for extended resources, promoting the feature to beta, and fixing a scheduler crash related to nil pointer dereference. These changes collectively enhance resource tracking, health reporting, code maintainability, and stability of the DRA subsystem.
• [pull/135202, pull/135196, pull/135199, pull/135200, pull/135048, pull/135051]

• Declarative Validation and API Improvements: Several pull requests enable declarative validation for the node.k8s.io API group and the RBAC ClusterRoleBinding resource, add missing field label conversion for ReplicaSet status, and enable commentstart checks on the admissionregistration API group. These updates improve validation coverage, API consistency, and user experience without introducing user-facing changes.
• [pull/135046, pull/135050, pull/135139, pull/135106]

• Scheduler Metrics and Feature Gate Updates: Pull requests introduce a new alpha counter metric to track pods scheduled after flush from the unschedulable queue and mark the KubeletEnsureSecretPulledImages feature gate as beta and enabled by default with related tests and fixes. These changes help identify scheduling issues and stabilize feature usage.
• [pull/135126, pull/135228]

• Kubelet and CPU Manager Refactoring: A pull request refactors the kubelet CPU manager by moving the fake CPU manager into a dedicated subpackage and changing the fake manager constructor to return a concrete type, improving code organization and testability.
• [pull/135220]

• Etcd Client Enhancements: One pull request adds DNS-based service discovery, gRPC health checking, round-robin load balancing, and gRPC trace logging to the etcd client, improving high availability and failure detection while maintaining backward compatibility.
• [pull/135047]

• CSI Driver Error Handling Fixes: Multiple automated cherry picks mark API server errors as transient in the CSI raw block driver to prevent volumes from being incorrectly marked as unmounted due to temporary failures, aligning error handling with the file mode plugin.
• [pull/135064, pull/135065, pull/135066]

• Test Stability and Flakiness Fixes: A pull request addresses flakiness in the CSI Mock volume expansion quota validation test by replacing the cached client with an uncached one, adding retry logic, and ensuring deterministic recovery verification to eliminate timing-based failures.
• [pull/135131]

• Leadership Callbacks Improvement: One pull request adds support for invoking the OnStoppedLeading callback only after the OnStartedLeading callback has fully completed, enabling graceful cleanup during leadership transitions.
• [pull/135062]

• Filesystem Watcher Bug Fix: A pull request implements a graceful shutdown mechanism for the fsnotify watcher’s Run() function to prevent indefinite blocking and goroutine leaks in the Kubernetes filesystem watcher.
• [pull/135078]

• Statusz Registry Refactor: One pull request moves the componentName into the statusz registry to make the registry the single source of truth for all statusz metadata, replacing the previous approach of passing componentName separately.
• [pull/135091]

• FIFO Queue Metrics Feature: A pull request introduces a new feature adding a FIFO queued size metric to monitor the length of FIFO queues for various Kubernetes resource informers, identified by item type and unique name with override options.
• [pull/135112]

• Miscellaneous Preliminary Test Submission: One pull request is a preliminary test submission with minimal descriptive information and no specified changes or linked issues, likely intended for initial validation or experimentation.
• [pull/135125]

• Workload API Cleanup: A pull request removes the Basic pod group policy from the Workload API as a follow-up to a previous change, reflecting that the basic policy is no longer needed.
• [pull/135179]

• In-place Pod-level Resource Resizing Feature: One pull request adds a node-declared feature for in-place pod-level resource resizing (IPPR) support, enabling valid update resize during admission as introduced in a related Kubernetes enhancement.
• [pull/135203]

3.2 Closed Pull Requests

This section provides a summary of pull requests that were closed in the repository over the past week. The top three pull requests with the highest number of commits are highlighted as 'key' pull requests. Other pull requests are grouped based on similar characteristics for easier analysis. Up to 25 pull requests are displayed in this section, while any remaining pull requests beyond this limit are omitted for brevity.

Pull Requests Closed This Week: 92

Key Closed Pull Requests

1. KEP-4671: Implement Gang scheduling in kube-scheduler. Copy of PR #134722: This pull request implements gang scheduling in the kube-scheduler by adding the necessary API changes, feature gates, workload references, and integration tests to enable coordinated scheduling of related pods as a single unit, addressing multiple related issues and aligning with KEP-4671.

• URL: pull/135138

• Merged: No

• Associated Commits: 8a809, 99e72, a9358, aa244, 2e184, a57b0, 6e7a7, 5b177, cd8fc, 8694c, 3fba1

2. Nnn integration tests: This pull request aims to add and refine integration tests related to the NominatedNodeNameForExpectation and ClearingNominatedNodeNameAfterBinding feature gates, including promoting these features to beta and verifying preemption behavior involving NominatedNodeName settings.

• URL: pull/135096

• Merged: No

• Associated Commits: 7a605, 55ea4, 95161, 2991a, d6100, 42a6d, b64c6, c1c6d, 581bc, 0c8bc

3. KEP-4671: Add Workload API. Copy of PR #134564: This pull request proposes adding a new Workload API along with a WorkloadReference field to the Pod specification, based on Kubernetes Enhancement Proposal 4671, to enhance workload management capabilities without including the gang scheduling implementation.

• URL: pull/135140

• Merged: No

• Associated Commits: 86da2, a94b2, 98791, cfaa8, a8674, dcb0a, 1480c

Other Closed Pull Requests

• SupplementalGroupsPolicy GA promotion: This pull request promotes the SupplementalGroupsPolicy feature from beta to GA by updating the feature gate and removing redundant end-to-end tests. It also finalizes unit tests and documentation as specified in KEP-3619.
  pull/135088

• CSI manifests update for e2e tests: This pull request synchronizes CSI manifests used in end-to-end tests with the latest csi-driver-hostpath master branch changes. It includes significant updates to the update-hostpath.sh script to maintain compatibility and test accuracy.
  pull/135135

• Deterministic device class selection for extended resources: This pull request implements a method to deterministically select one device class when multiple are available, choosing either the most recently created or alphabetically earliest by name. This ensures consistent resource allocation behavior.
  pull/135037

• Volume limits scheduling tests: This pull request adds verification tests for scheduling behavior related to volume limits without changing the CSIDriver. Although it addresses review comments, it was not merged.
  pull/135077

• DRA device taint eviction improvements: This pull request enhances the DRA device taint eviction feature by adding a separate feature gate for DeviceTaintRules to limit bugs without disabling device taints. It also optimizes eviction simulation with a NOP queue and tracks evicting rules to reduce unnecessary rule listing.
  pull/135068

• VolumeGroupSnapshots e2e test updates: This pull request updates VolumeGroupSnapshots end-to-end tests to use the v1beta2 API by bumping CRDs from the external-snapshotter and disables the v1beta1 API due to webhook limitations. It ensures group snapshot tests are enabled and run in CI.
  pull/135069

• DRA validation logic enhancements: This pull request improves DRA validation by short-circuiting on maxSize checks to enhance denial-of-service protection. It adds a declarative validation test exposing a mismatch with the "+k8s:maxLength" tag and refactors related validation functions.
  pull/135079

• StatefulSet rollout regression fix: This automated cherry pick fixes a regression in kube-controller-manager causing spurious StatefulSet rollouts during control plane upgrades. It introduces a feature gate enabled by default to prevent unnecessary rollouts.
  pull/135087

• kubeadm preflight check skip on dry-run: This pull request proposes skipping the ContainerRuntimeVersion preflight check during kubeadm upgrades when the --dry-run flag is used. This prevents unnecessary warnings in dry-run scenarios.
  pull/135090

• Node e2e image pull test fixes: This pull request fixes failures in node end-to-end image pull tests by implementing finite waits for pod status, logging fake registry credentials on failure, and running registry tests as pods even in-cluster. These changes ensure reliable test execution.
  pull/135094

• Validation ratcheting logic improvement: This pull request introduces a boolean flag to distinguish explicitly nil fields from absent fields in old objects. This prevents incorrect skipping of validation during update operations.
  pull/135123

• Node conformance test connectivity fix: This pull request fixes node conformance test failures by improving connectivity to the e2e test registry through restarting the kubelet after loading credential configurations. It addresses issues without requiring longer credential cache expiration.
  pull/135142

• Storage version selection revert: This pull request reverts a previous change that prevented selecting versions with replacements as the storage version. It fixes issues related to the removal of the original v1alpha3 API candidate and corrects the replacement tag behavior.
  pull/135197

• kube-controller-manager nodelifecyclecontroller bug fix: This pull request fixes a race condition causing pods to be incorrectly marked ready by ensuring pods are processed only when nodeHealth data is available or pods are deleted. This prevents premature or missed NotReady status updates during node transitions.
  pull/135212

• kubectl setup code cleanup: This pull request refactors kubectl setup code by moving PluginHandler to a separate file and improving the readability of NewDefaultKubectlCommandWithArgs. These changes enhance code clarity and maintainability.
  pull/135053

• DRA allocator performance revert and fix: This pull request reverts a previous change causing a performance regression in the DRA allocator and implements a simpler fix that performs extra work only when the resource pool is incomplete. It also restores and expands relevant test cases.
  pull/135056

• InPlacePodVerticalScaling conformance tests promotion: This pull request promotes InPlacePodVerticalScaling end-to-end tests to conformance status following GA. It demonstrates minimal test flakiness, ensuring compliance with Kubernetes conformance criteria.
  pull/135067

• Typed workqueue shutdown improvement: This pull request improves the typed workqueue by ensuring graceful shutdown waits for the updateUnfinishedWorkLoop() goroutine to terminate. It introduces a stop channel and stop-once mechanism similar to the delaying queue to prevent blocking.
  pull/135072

• kubeadm preflight container runtime version check revert: This pull request reverts a previous commit adding a container runtime version check to kubeadm’s preflight process and removes a socket from a fake node.
  pull/135097

• Test submission for dry-run failure fix: This pull request contains test commits aimed at fixing a potential dry-run failure related to NodeLocalCRISocket GA and includes an additional unspecified update labeled "option 2." No changes were merged.
  pull/135120

• DRA device-specific health check timeouts: This pull request implements configurable device-specific health check timeouts in DRA health monitoring. It allows drivers to specify custom timeouts via the gRPC health API while maintaining a default timeout for backward compatibility.
  pull/135147

• MinimumKubeletVersion test tag cleanup: This pull request removes MinimumKubeletVersion test tags for kubelet versions 1.20, 1.21, 1.22, 1.23, and 1.27, which are no longer supported or tested. This streamlines the testing framework for supported Kubernetes versions.
  pull/135157

• Scheduler unit test race condition fix: This pull request fixes a race condition in the scheduler unit test by ensuring the ResourceSlice informer is synced before running. It also corrects deferred logging in the allocator to capture parameter values at function exit.
  pull/135186

• IngressClassParametersReference validation refactor: This pull request proposes extracting kind and name validations into a dedicated function and removes copyright information in a declarative test. The changes were not merged.
  pull/135224

• CONNECT proxy response header size limit: This pull request addresses a bug by limiting the size of CONNECT proxy response headers to prevent memory exhaustion from misbehaving proxies. It extends a previous fix applied only to http.Transport users to cover additional cases.
  pull/135038

3.3 Pull Request Discussion Insights

This section will analyze the tone and sentiment of discussions within this project's open and closed pull requests that occurred within the past week. It aims to identify potentially heated exchanges and to maintain a constructive project environment.

Based on our analysis, there are no instances of toxic discussions in the project's open or closed pull requests from the past week.

---

IV. Contributors

4.1 Contributors

Active Contributors:

We consider an active contributor in this project to be any contributor who has made at least 1 commit, opened at least 1 issue, created at least 1 pull request, or made more than 2 comments in the last month.

If there are more than 10 active contributors, the list is truncated to the top 10 based on contribution metrics for better clarity.

Contributor	Commits	Pull Requests	Issues	Comments
pohly	53	16	14	85
liggitt	54	4	0	76
BenTheElder	55	6	1	61
macsko	44	7	3	61
neolit123	18	6	1	88
yongruilin	67	3	1	25
aaron-prindle	40	6	1	20
HirazawaUi	21	8	3	33
michaelasp	12	5	0	48
kannon92	13	3	4	44