VMs lose connectivity and MAC addresses aren't learned after a vmnic link failure when using "Route based on Physical NIC Load" loadbalancing with trunk VLANs.
Article ID: 404730
Updated On:
Products
Issue/Introduction
1. In the presence of multiple uplinks (vmnic1, vmnic2), an uplink switchover (Vmnic1 goes down) will result in workloads moving to vmnic2.
2. Workloads connecting to trunk DVPGs/segments will face a downtime.
Environment
VMware vSphere ESXi
VMware vCenter Server
VMware NSX
Cause
RARP Broadcast Scope Limitation during uplink switchover, Reverse ARP (RARP) broadcasts are not flooded to all trunk VLANs where the affected workloads reside. This limitation prevents physical switches from promptly updating their MAC address tables, causing them to retain the VM's MAC address pointing to the old, failed uplink.
Resolution
Configure Link Aggregation Control Protocol (LACP) on both your physical network switch and the ESXi host. This ensures consistent MAC-to-uplink mappings and proper MAC address table updates.
Feedback
