NSX Load Balancer Configuration Not Working for VCF Operations Cluster

Products

• VMware NSX
• VCF Operations

Issue/Introduction

• After deploying a new VCF 9.0.2 environment, the NSX Load Balancer configuration for a VCF Operations cluster (vRealize Operations) fails to function.
• The designated load-balanced FQDN is unreachable, and the web page fails to open.
• The Virtual IP (VIP) / FQDN does not return ICMP pings.
• VCF Operations nodes function normally when accessed individually via their respective IP addresses, but fail when accessed through the Load Balancer Virtual Server.
• Setting the "System Access URL" within VCF Operations does not resolve the connectivity issue.
• Backend server pool members may show as Down or Monitor Failed in the NSX Manager UI.

Environment

• VCF Version: 9.0.2
• NSX Version: 9.0.2
• ESX Version: 9.0.1
• Application: VCF Operations (Operations cluster deployment)

Cause

The Tier-1 Gateway hosting the Load Balancer Service Router (SR) lacks a valid routing path to communicate with the backend server pool or the external gateway. Without a default or specific static route, the Load Balancer cannot receive health check replies from the cluster nodes or route return traffic to the clients, resulting in an unreachable VIP.

Resolution

To restore connectivity, you must provide a valid return path for the Tier-1 Service Router by adding a static route.

Procedure

1. Log in to the NSX Manager UI.
2. Navigate to Networking > Tier-1 Gateways and select the gateway hosting the Load Balancer.
3. Under Static Routes, click Add Static Route.
4. Configure a default route using the physical gateway details:
   • Network: 0.0.0.0/0
   • Next Hops: Enter the next-hop IP and scope it to the appropriate interface (e.g., a Service Interface or Uplink like LB-INT).
5. Save the configuration.
6. Validation: Verify that the Server Pool members now report an Up status and that the FQDN is reachable via browser and ping.

Additional Information

• Static routes configured on a Tier-1 gateway are not reflected on the Edge node.
• After move T1 gateway from one T0 to another T0, load balancer virtual server went down