Tagged VLAN traffic stops on VDS uplinks due to IPFIX sampling rate 0

Products

• VMware vSphere ESXi

Issue/Introduction

On a vSphere Distributed Switch (VDS), tagged VLAN traffic may suddenly stop passing through physical uplinks (vmnics) even though the interfaces appear 'Up'. This issue occurs when IPFIX (NetFlow) is configured with an invalid sampling rate.

Symptoms:

• All tagged VLAN traffic fails on vmnics associated with a specific VDS.
• Cisco Discovery Protocol (CDP) or LLDP information is no longer visible on the physical uplinks.
• Virtual Machines lose network connectivity on the affected vmnics.
• Checking the VDS configuration shows sampling rate = 0

Environment

• VMware vSphere ESXi 8.0.x
• VMware Cloud Foundation (VCF) 5.x / 9.x

Cause

When NetFlow/IPFIX is enabled on a VDS with a sampling rate of 0, the ESXi host attempts to capture and process every single packet instead of a subset. Under heavy network loads, this causes the ESXi kernel's network stack helper threads to choke. The data path locks up, preventing tagged VLAN traffic from traversing the physical interfaces.

Resolution

Disable the monitoring protocol and reset the physical interfaces to restore traffic flow.

1. Log in to the vSphere Client.
2. Navigate to the Distributed Switch where the issue is occurring.
3. Go to Configure > Settings > NetFlow.
4. Click Edit and disable NetFlow/IPFIX or ensure the Sampling Rate is set to a non-zero value (e.g., 100 or 1000).
5. Access the ESXi host via SSH or DCUI.
6. Bounce the affected physical uplinks to clear the locked state: esxcli network nic down -n vmnic#### esxcli network nic up -n vmnic#### (Note: Replace #### with your specific vmnic number, such as vmnic2).
7. Verify that CDP/LLDP information and VLAN traffic are restored.

For additional guidance, Contact Broadcom Support

Additional Information

Configure the NetFlow Settings of a vSphere Distributed Switch