Skip to content
Open
Changes from 11 commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
7 changes: 7 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -35,12 +35,15 @@ Monitor the release status by regions at [AKS-Release-Tracker](https://releases.
* [Windows Server 2025](https://aka.ms/aks/upgrade-windows-os-version) is now generally available. You no longer need to register a feature flag to create Windows Server 2025 node pools. Windows Server 2025 node pools can be created in Kubernetes version 1.32+ with a minimum GA CLI version of 2.87.0.
* [Azure Container Linux](https://learn.microsoft.com/azure/azure-linux/azure-container-linux-overview) is generally available (GA) as an OS option on AKS starting AKS v1.34. You can deploy ACL node pools in a new AKS cluster or add ACL node pools to your existing clusters. AKS also supports migrating existing node pools to ACL using in-place OS SKU migration or by creating new ACL node pools. For detailed migration steps, considerations, and rollback instructions, see [Migrate existing nodes to ACL](https://learn.microsoft.com/azure/azure-linux/tutorial-migrate-azure-container-linux-aks).
* [Azure Policy add-on](https://learn.microsoft.com/azure/governance/policy/concepts/policy-for-kubernetes) now generates ValidatingAdmissionPolicies (VAP) for all customers. This enforces CEL-based policies inside the API server process for minimal latency and enables fail-closed enforcement.
* Node Auto Provisioning (NAP) managed clusters can now enable custom OS settings for Linux nodes. See our [AKSNodeClass documentation](https://learn.microsoft.com/azure/aks/node-auto-provisioning-aksnodeclass#linux-custom-os-configuration-settings) for more information.
* Node Auto Provisioning (NAP) managed clusters can now support AMD GPU SKUs, NVIDIA GPU Operator, and custom driver installation via the AKSNodeClass. See our [AKSNodeClass documentation](https://learn.microsoft.com/azure/aks/node-auto-provisioning-aksnodeclass), and [GPU Operator on AKS documentation](https://learn.microsoft.com/azure/aks/nvidia-gpu-operator?tabs=nap-nodes) for more information.
* [AKS end of support notifications](https://learn.microsoft.com/azure/aks/aks-end-of-support-notifications) are now available. AKS automatically notifies you when your cluster's Kubernetes version is approaching or has passed its end of support date. Alerts are sent two weeks before end of support, one week before end of support, and then weekly after the support date passes—no monitoring add-on or maintenance window configuration required. Notifications are published to Azure Resource Graph and can be surfaced via email alerts or real-time Event Grid webhooks.

#### Preview features

* [Azure Linux 3.0 confidential VM (CVM)](https://learn.microsoft.com/azure/aks/use-azure-linux) is now available in preview in Fairfax (US Gov) regions. Register the `AzureLinuxCVMPreview` feature to enable it.
* [In-place node pool resize](https://learn.microsoft.com/azure/aks/resize-node-pool) is now available in preview. Resize the VM size of an existing VMSS-based node pool in place via `az aks nodepool update --node-vm-size`, without manually creating and migrating to a new node pool.
* [Mixed SKU autoscaling in virtual machine node pools](https://learn.microsoft.com/azure/aks/virtual-machines-node-pools#cluster-autoscaler-with-virtual-machines-node-pools-preview) is now available in preview. This allows users to select multiple VM sizes in the same node pool using cluster autoscaler with virtual machine node pools.
* [Automatic Pod Disruption Budget management](https://learn.microsoft.com/azure/aks/automatic-pod-disruption-budget-management) is now available in preview. The AKS extension automatically creates PDBs for deployments without PDBs and temporarily scales up replicas to unblock node drain when a PDB would prevent eviction, then scales back down—reducing the need for manual intervention during upgrade drain failures.

#### Behavioral changes
Expand All @@ -52,6 +55,10 @@ Monitor the release status by regions at [AKS-Release-Tracker](https://releases.
* AKS now allows migration from the `managedNATGatewayV2` outbound type to the `block` and `none` outbound types, supporting [network-isolated cluster](https://learn.microsoft.com/azure/aks/concepts-network-isolated) scenarios. Migration to other outbound types remains blocked.
* AKS now validates pod CIDR ranges during cluster create and update for kubenet and [Azure CNI Overlay](https://learn.microsoft.com/azure/aks/azure-cni-overlay) clusters. Clusters can no longer be created or updated with a pod CIDR that overlaps with reserved IP ranges (`172.30.0.0/16`, `172.31.0.0/16`), preventing potential in-cluster networking failures. Existing clusters with an overlapping pod CIDR are unaffected. See [CNI prerequisites](https://learn.microsoft.com/azure/aks/concepts-network-cni-overview#prerequisites).
* AKS now rejects [Calico NPM and Azure NPM](https://learn.microsoft.com/azure/aks/use-network-policies) install and uninstall operations on clusters running Kubernetes versions earlier than 1.30. Requests are rejected at the API level with a descriptive error directing customers to upgrade to a newer supported Kubernetes version before retrying. Existing clusters already using Calico NPM or Azure NPM are unaffected.
* Setting certain labels on AKS nodes is now restricted, including `agentpool`, `storageprofile`, `storagetier`, `accelerator`, and labels with a prefix of `kubernetes.azure.com/` that aren't well-known labels. New subscriptions using NAP can no longer set these labels, and existing subscriptions will be restricted on September 30, 2026. For more information, see [Azure Karpenter Provider Issue #1707](https://github.com/Azure/karpenter-provider-azure/issues/1707).
Comment thread
wdarko1 marked this conversation as resolved.
Outdated
* The cluster identity of Node Auto Provisioning (NAP) enabled clusters will now be granted the `AKS Nodepool Machine Manager` role.
* Custom subnets are no longer permitted on Node Auto Provisioning (NAP) managed clusters using system-assigned identities. For custom subnets with NAP managed clusters, use user-assigned identities. To set up a user-assigned identity, see our [documentation](https://learn.microsoft.com/azure/aks/user-assigned-managed-identity).
* The naming convention for Node Auto Provisioning (NAP) managed nodes will change from `nodepool-name-hash` to `aks-aksmanagedap-nodepool-name-########-vm`.
Comment thread
wdarko1 marked this conversation as resolved.
Outdated

#### Bug fixes

Expand Down
Loading