LCI : Local Consumption Interface 9.1.1
What's New in 9.1.1
vSphere Kubernetes Service (VKS)
LCI 9.1.1 introduces support for VKS 3.7 with the following features:
- ClusterClass 3.7 Support: Includes schema changes and consumes variable metadata.
- Cluster Scaling Enhancements: Increased the worker node ceiling to 250 and added Pod CIDR limit adjustments.
- Multi-Control-Plane Nodes: Added support for up to 5 control plane nodes per cluster.
- Node Pool Management: Added day-2 delete operations for node pools.
- Secondary NIC & Driver Exposure: Exposed the NIC VFIO-PCI driver field and second-level ClusterClass variables schema handling for advanced networking.
- CNI Selection: You can now select a CNI during cluster creation.
- TKC Support Deprecation: Removed TKC support from LCI.
Fixed Issues
- In VKS IaaS plugin: After retiring a TKC, you can now update the cluster to a newer ClusterClass without manually removing the
kubernetes.vmware.com/skip-auto-cc-rebase annotation from the cluster YAML (previously a known issue in 9.1.0).
Known Issues and Limitations
-
For 8.0U3 installations: Occasionally, the plug-in may fail to load on the initial attempt. To check if the plug-in has loaded correctly, click the vSphere Client menu icon, then to Administration -> Client -> Plug-ins. Check the Status column of the VMware Local Consumption Interface plug-in, and in case you see a Plug-in Configuration with Reverse Proxy failed. message, reinstall the plug-in.
-
You must uninstall version 1.0.x before using the 9.x.x version of LCI. Failure to do so will result in the interface not starting correctly when looking at the Resources tab for a namespace.
-
In VKS IaaS plugin:
- Retiring a 1.32.x VMware Kubernetes Service cluster and then upgrading it to 1.33.x fails with an admission webhook error and requires manually editing the cluster object.
- Windows OS images are visible in Add Nodepool on a cluster when using a non-Antrea CNI.
- If a nodepool is using a Windows OS image, the OS image field is blank on the cluster's Summary tab for that nodepool.
- Cluster variables that require CIDR values under Create Cluster - Advanced Settings reject valid CIDR values.
- Provisioning a cluster using the Default Configuration may trigger an admission webhook error in the event Pod CIDR blocks overlap with Supervisor-managed IP ranges. To ensure successful cluster deployment, select the Custom configuration and specify the Pod CIDR to prevent IP range overlap.
-
Previous release notes