NSX Load Balancer traffic steering fails with 404 errors after migration in VCF

Products

• VMware NSX

Issue/Introduction

In VCF environments, specifically when migrating from NGINX or legacy load balancing to NSX Load Balancer, traffic may fail to reach backend pool members. Symptoms include:

Users receive HTTP 404 Not Found errors.

Traffic continues to route to legacy NGINX configurations despite the service being disabled.

The NSX Virtual Server (VS) status may show as "Up," but backend communication fails.

Communication between the virtual server and server pool members requires the correct custom port number. 

 

Environment

VMware NSX 4.2.x

Cause

Custom port values are not being correctly applied to server pool members. If the port field is left unspecified, the configuration defaults to port 443 rather than configuring the expected port number in ther server pool.

Resolution

     1.    Correct Port Configuration:

1. • Navigate to Networking > Load Balancing > Server Pools.

   • Ensure the pool member port is set to the required port number

2. Separate Server Pool Configuration (NSX 4.2+ Requirement):

   • In NSX 4.2, you must configure the Server Pool separately before or during the Virtual Server setup.

   • Navigate to Networking > Load Balancing > Server Pools and create the pool.

   • Then, navigate to Virtual Servers and associate the existing pool. Unlike older versions, the "inline" creation may not apply all settings if not properly linked to the Tier-1 Gateway.

3. Edge Node Verification:

   • SSH into the active Edge Node and run: get load-balancer <LB-ID> virtual-server <VS-ID>

   • Verify that the Operational State is UP and the Backend Pool reflects the correct members on required port number.