Failover of the ESXi uplinks does not work as expected when a physical switch went down

Products

• VMware NSX

Issue/Introduction

• A physical switch has gone down and ESX experienced a total network outage
• vDS portgroup has 2 or more active uplinks e.g. uplink1 and uplink2
• Failover to the other physical switch has not occurred as expected

Environment

VMware NSX

Cause

The ESX host has multiple vmnics, some vmnics connect to one physical switch and others connect to the second physical switch.
This design should provide network redundancy.
uplink1 and uplink2 on the vDS are incorrectly mapped to vmnics which connect to the same physical switch.

Resolution

The mapping of the vDS uplinks must be changed so that they map to vmnics connected to different physical switches

1. On vSphere client, click on a Hosts > Configure > Virtual switches, select Manage Physical Adaptors.
2. Change the dropdowns to remap the vmnics.
   In this example vmnic5 will be changed to uplink4 and vmnic4 will be changed to uplink2
   
   
   
   
3. Click OK and repeat for each host that is impacted