This article demonstrates how to configure a vSphere Kubernetes Service (VKS) Cluster to support 250 pods per worker node, discusses the prerequisites, and highlights the operational considerations that administrators should evaluate before increasing pod density.
Default VKS Behavior
In most Kubernetes deployments, kubelet uses a default pod limit to prevent node exhaustion.
You can verify the current limit using:
command : kubectl --kubeconfig <vks_cluster_kubeconfig_path> get nodes -o custom-columns=NAME:.metadata.name,CAPACITY:.status.capacity.pods,ALLOCATABLE:.status.allocatable.pods
Output Example :
NAME CAPACITY ALLOCATABLE vks-demo-c1-n9gpj-6rbph 110 110 vks-demo-c1-worker-86fwv 110 110
What You Need to check Before Adjusting the higher limit
- VKS Version : vSphere Kubernetes Service 3.5.0 (+v1.34) or higher.
- Resource Sizing: Ensure your worker node VMs have sufficient CPU and memory allocations. A node hosting 250 pods requires a robust underlying footprint to avoid memory pressure and eviction storms.
- IP Capacity : Every pod needs a unique IP address. For 250 pods , a larger Pod CIDR assignment per node is strongly recommended.
Configuring 250 Pods Per VKS Worker Node
- Add the following parameters to VKS cluster yaml:
topology:
class: builtin-generic-v3.5.0
version: <vkr-version>
variables:
- name: kubernetes
value:
kubeletConfiguration:
maxPods: <maximum-pods-per-node> # set 250 here- Deploy the VKS Cluster with above configuration and verify the pods capacity per Worker node
# kubectl --kubeconfig <vks_cluster_kubeconfig> get nodes -o custom-columns=NAME:.metadata.name,CAPACITY:.status.capacity.pods,ALLOCATABLE:.status.allocatable.pods NAME CAPACITY ALLOCATABLE vks-demo-ns1-d1-gk5fj-72stq 250 250 vks-demo-ns1-d1-vks-demo-ns1-d1-np-2vjr-kgwqf-2ntgz-2nv2h 250 250
Conclusion
Increasing pod density to 250 pods per worker node can significantly improve infrastructure utilization and reduce cluster footprint. However, successful deployment requires careful planning around node sizing, networking scalability, IP allocation, and workload characteristics.
Disclaimer
This blog is for informational and educational purposes only. The configurations, examples, and architectural guidance provided are based on general best practices and publicly available references.
Always validate configurations in a non-production environment before applying them to live systems. Features and integrations may vary depending on the versions of VKS, VKr, LCI, Supervisor and vCenter Server being used. The author is not responsible for any unintended impact caused by the use of this information in production environments.
Contents, and opinions are my own and do not reflect the opinions of my employer. .
