VMware ESXi 8.x host, Cisco Discovery Protocol (CDP) information for a physical network adapter (VMNIC) intermittently changes or 'flips' between different upstream switch ports.

Products

• VMware vCenter Server

Issue/Introduction

Specific symptoms include:

• The `Device ID` remains the same (identifying the logical switch/stack), but the `Port ID` (e.g., Ethernet1/15 vs. Ethernet1/16) toggles every few seconds.
• Fluctuation is visible in the vSphere Client under **Physical Adapters** discovery info.
• Fluctuating output is observed when running the following command repeatedly:  `vim-cmd hostsvc/net/query_networkhint --pnic-name=vmnic[X]`
  
  
  
    

Environment

• VMware vSphere ESXi 8.x
• Cisco Nexus or Catalyst Upstream Switches (typically in a stack or VPC configuration)
  
  

Cause

This behavior occurs when an ESXi uplink receives CDP advertisements from multiple physical switch ports. This is typically caused by one of the following:

1.  **Physical Cabling Loop:** A physical loop exists in the network underlay, allowing discovery packets to reach the same NIC from different directions.
2.  **LACP / Port-Channel Misconfiguration:** The physical switch ports are bundled into a Port-Channel, but the load-balancing or trunking configuration does not match the ESXi virtual switch policy (e.g., LACP is enabled on the switch but the vSwitch is set to 'Route based on originating virtual port').
3.  **Native VLAN Mismatch:** CDP packets may be 'leaking' across ports due to inconsistent Native VLAN tagging between the ESXi host and the upstream switch.

Resolution

• Step 1: Analyze Packet Captures
  Perform a packet capture on the affected uplink to identify the source of the conflicting CDP packets.

`pktcap-uw --uplink vmnic[X] --ethtype 0x2000 -o /vmfs/volumes/datastore/cdp_capture.pcap`

Review the capture in Wireshark to confirm if multiple Source MAC addresses or different Port IDs are advertising the same Device ID to that specific VMNIC.

• Step 2: Verify Physical Topology
  Coordinate with the network team to ensure:
  - There are no physical cable loops bridged between ports.
  - The specific VMNIC is connected to exactly one intended physical port.
  
  
• Step 3: Align Virtual and Physical Switch Policies
  If the uplinks are intended to be part of a team (LAG/Port-Channel):
  - **Standard Switch (VSS):** Ensure the switch side is a Static Port-Channel and ESXi is set to **Route based on IP hash**.
  - **Distributed Switch (VDS):** Ensure LACP is correctly negotiated if using an LACP LAG.
  
  
• Step 4: Firmware and Driver Alignment
  Ensure the ESXi host and server hardware (e.g., Dell PowerEdge MX750c) are updated to the latest supported BIOS and Driver/Firmware versions as identified by the VMware Compatibility Guide.

Additional Information

- CDP data shows multiple ESXi host uplinks (vmnics) connected to same neighbor
- CDP information unavailable on ESXi physical network adapters