vSphere Kubernetes Service Supervisor control plane VMs unreachable due to physical network misconfiguration

Products

• VMware vSphere ESXi

Issue/Introduction

• vCenter cannot reach the Supervisor VIP or Control Plane VMs.
• Management connectivity loss to Supervisor cluster.
• Supervisor VMs are deployed with two vNICs
• Ping fails to specific vNIC (eth1) Supervisor VMs on subset of ESXi hosts.

Environment

vSphere ESXi

Cause

vSphere with Tanzu Supervisor cluster VMs have two network cards (eth0 and eth1) to separate management traffic from workload traffic vSphere with Kubernetes on VCF. This design improves security and stops heavy app traffic from slowing down critical infrastructure.

eth0 (Management Network) Connects to the ESXi management network vSphere with Kubernetes on VCF. Used for infrastructure communication, including vCenter, DNS, NTP, and ESXi hosts.

eth1 (Workload/Overlay Network) Connects to the NSX overlay network or vSphere Distributed Switch Supervisor Networking. Used for internal communication with pods and deployed Tanzu Kubernetes clusters vSphere with Kubernetes on VCF.

Physical switchport configurations (VLAN tagging, trunking) do not match the Distributed Virtual Switch (DVS) properties for vNIC presenting eth1 on specific ESXi host uplinks.

 

Resolution

1. Identify impacted ESXi hosts and their physical uplinks.
2. Inspect physical switchport settings for VLAN mismatches.
3. Verify VLAN tagging and trunking settings match DVS properties.
4. Correct discrepancies in the physical network fabric to restore the data path.