In this article discuss about VMware vSphere Kubernetes Service (VKS) 3.5 and vSphere Kubernetes Release (VKr) 1.34 With the introduction of VKS 3.5, support has been extended to include VKr version 1.34. Under this update, each vSphere Kubernetes minor release will now receive a 24-month support window from the date it becomes available. This longer lifecycle significantly reduces the pressure to upgrade frequently, helps maintain stable environments, and allows teams to channel their time toward business-driven priorities rather than constant upgrade planning. Organizations that prefer to adopt the latest Kubernetes capabilities can continue tracking new versions, while those that value long-term stability can operate on a consistent release with greater confidence. VKS 3.5 accommodates both approaches.
In Kubernetes 1.34, Dynamic Resource Allocation (DRA) has reached stable status. This feature enables centralized classification of specialized hardware—such as GPUs—through DeviceClass, ensuring pods are scheduled consistently onto nodes that match their required hardware category. By supporting device sharing and request-based resource claims similar to dynamic storage provisioning, DRA enhances utilization and mitigates issues like uneven distribution or idle hardware. Users can employ ResourceClaims and Resource Claim Templates to precisely request the device characteristics needed for their workloads.
DRA also introduces the ability to share GPU devices across multiple containers or pods. Using Common Expression Language (CEL) filters and ResourceSlice-based requests, workloads can select specific GPU devices, enabling more efficient allocation compared to traditional count-based GPU requests.
Cluster API v1beta2:
Improved Kubernetes Lifecycle Management in VKS 3.5
VKS 3.5 now includes support for Cluster API v1beta2 (CAPI v1.11.1), bringing an updated set of resource APIs and several architectural enhancements over the earlier v1beta1 release. For a complete list of upgrades and adjustments, refer to the official CAPI release notes.
As part of the upgrade VMware vSphere Kubernetes Service (VKS) 3.5 , clusters that were previously created using the v1beta1 API will be seamlessly migrated to the v1beta2 format by the enhanced Cluster API controllers. For all future lifecycle management activities, the use of the v1beta2 API is strongly recommended.
The introduction of v1beta2 marks a significant milestone toward the eventual graduation of these APIs to v1. Resource status reporting has also been strengthened across the board, improving overall transparency and addressing long-standing feedback from field teams and customers.
Preventing Cluster Upgrade Failures Caused by Misconfigured PodDisruptionBudgets
A PodDisruptionBudget (PDB) defines the minimum number of pods that must remain available during voluntary disruptions such as maintenance, scaling actions, or node upgrades. If a PDB is configured incorrectly, it can block pod evictions—resulting in stalled or failed cluster upgrades.
To mitigate this, VKS 3.5 introduces validation guardrails designed to detect PDB misconfigurations that may hinder cluster operations. When such issues are found—for example, when no disruptions are allowed—the system will halt the upgrade to avoid failures.
To further enhance upgrade reliability, a new SystemChecksSucceeded condition has been added to the Cluster object. This condition gives operators a clear view of whether the cluster is ready for an upgrade, with a specific focus on PodDisruptionBudget health.
VMware vSphere Kubernetes Service (VKS) 3.5
Key advantages of this enhancement include:
Pre-upgrade Protection: Automatically identifies PDB settings that would prevent pod evictions, blocking the upgrade before it starts.
Improved Issue Visibility: Marks SystemChecksSucceeded as false when any PDB shows allowedDisruptions <= 0, signaling that an upgrade may fail without corrective action.
Better Upgrade Planning: Provides administrators with the insights needed to correct PDB issues ahead of time, ensuring smooth and predictable upgrade workflows.
Optimized Operations with Enhanced CLI, Add-On Lifecycle Management, and Support Utilities
VKS 3.5 introduces major improvements aimed at simplifying day-to-day operational tasks for platform teams, particularly around add-on lifecycle management and support bundle creation.
Unified VKS Add-On Management
The updated Add-On Management capability consolidates how standard packages are installed, upgraded, and configured. It provides a consistent workflow for all package-related operations, supported by a declarative API framework and built-in validation to avoid compatibility or dependency issues during upgrades. Although the previous workflow is still available, the new system offers stronger reliability and operational efficiency.
Key enhancements include:
Centralized management of VKS Standard Packages through the Supervisor Namespace, enabling installation, updates, and configuration from a single interface. These features can be accessed via VKS-M 9.0.1 or the new VCF CLI addon plugin.
Simplified Cluster Autoscaler installation with automated upgrades aligned with Kubernetes cluster version updates.
Declarative Add-On APIs (AddonInstall, AddonConfig) that integrate seamlessly with GitOps tools such as Argo CD, enabling version-controlled, repeatable add-on deployments across multiple environments with improved traceability.
Cluster-wide Addon Repository support using AddonRepository and AddonRepositoryInstall APIs. This streamlines configuration of VKS Standard Packages, including scenarios where repositories are mirrored to a private registry. Starting with the 2025.10.22 package release for VKS 3.5.0, a default AddonRepository for embedded standard packages will be automatically installed.
Automated compatibility validation during Kubernetes upgrades, ensuring that all add-ons are updated to the most recent compatible releases. If incompatibilities are detected, VKS halts the upgrade to prevent issues.
Pre-checks for add-on installation and upgrades, including detection of duplicate PackageInstalls and reporting of non-blocking dependency warnings.
Improved visibility into available add-on versions and compatibility information, helping operators quickly identify suitable releases.
Conclusion: –
VKS 3.5 significantly elevates platform reliability by unifying and automating add-on lifecycle operations. Its declarative APIs and integrated pre-checks ensure streamlined upgrades with minimal risk. By enabling consistent, GitOps-driven deployments, it brings stronger control and auditability across clusters. Overall, VKS 3.5 empowers teams with a more efficient, predictable, and scalable operational experience. In my previous article discussed about VMware Cloud Foundation.