WilliamLam.com

Quick Tip: How to Identify Which Kubernetes Cluster Owns a vSphere Container Volume (PV)

by // Leave a Comment

I was recently cleaning up some orphaned vSphere Container Volumes and while I was able to compare the Persistent Volumes (PV) IDs against the ones currently in use by a couple of Kubernetes clusters, I wanted to be 100% sure before removing them, especially since I planned bulk delete using PowerCLI’s Remove-CnsVolume cmdlet.

You can view all provisioned vSphere Container Volumes in the vSphere UI by navigating to vSphere Cluster->Monitor->Cloud Native Storage->Container Volumes. In addition to the volume details, you also get access to all of the associated metadata and labels, very similar to what you would see using kubectl, which is quite nice!


While reviewing the PV IDs in the UI, I noticed an attribute named Kubernetes Cluster, which appeared to identify the Kubernetes (K8s) cluster that owned the volume. How do I find this Kubernetes Cluster ID?

[Read more...]

VCF 9.1 - Auditing VCF Management Services (VCFMS) IP Pool Usage 

by // Leave a Comment

As part of deploying the new VCF Management Services (VCFMS) component, a pool of IP addresses is provided to support the automatic deployment and scale-out of additional worker nodes for Day-N services such as Log Management and Real-time Metrics. In addition, these IP addresses are also reserved for lifecycle operations such as node recovery, rolling upgrades and scaling out additional service replicas, all of which may require additional IP capacity.


While you can view the current IP allocations by logging into VCF Operations and navigating to Build->Lifecycle->VCF Management and selecting the VCF Services Runtime component (VCFMS), there is currently no high-level view showing how many IP addresses have been consumed relative to the configured IP address pool(s). This can make capacity planning difficult and leave users unsure when additional IP pools may need to be added.

With the help of the VCF Fleet LCM APIs, we can easily audit the IP usage for a given VCFMS deployment and generate a more friendly report

[Read more...]

VCF 9.1 - OCuLink External Graphics (eGPU) Passthrough with vSphere Kubernetes Service (VKS)

by // Leave a Comment

The Minisforum MS-A2 continues to be a popular and versatile platform for running VMware Cloud Foundation (VCF) 9.1 in a lab environment. While the MS-A2 may be small in size, it is surprisingly capable and can even be configured to support an NVIDIA RTX 4000 Ada (20GB VRAM) GPU, which I had used to deploy VMware Private AI Services (PAIS).

Although there are several half-height GPUs that can fit in an MS-A2, the overall selection remains relatively limited. For users looking for additional GPU flexibility, the MS-A2 supports an OCuLink half-height PCIe card, enabling the use of a much broader range of discrete GPUs.


Thanks to a fellow colleague who recently shared their success with me using this setup on VMware Cloud Foundation (VCF) 9.1, including GPU passthrough to vSphere Kubernetes Service (VKS) cluster for AI experimentation, below are the specific BOM details for those interested.

[Read more...]