After upgrading to VCF 9.0.2, some ESXi hosts in a cluster are reporting an "Unknown" status in the NSX Manager UI: System > Fabric > Hosts

Products

• VMware NSX

Issue/Introduction

• In the NSX Manager UI, System > Fabric > Hosts, when clicking on the Unknown status, Status details indicate Controller Connectivity as Unknown, Tunnel Status as Unknown, and PNIC/Bond Status as Unknown
  
  
  
  
• Evaluation of the Unknown ESXi host via CLI shows it is working as expected with no issue found:
  
  
  • • 1 Controller is in a "connected" state: nsxcli -c get controllers
    • 3 Managers are in a "Connected" state: nsxcli -c get managers 
    • netcat to NSX Managers succeeds to ports 1234 and 1235: nc -zv  <NSX Manager IP/FQDN> 1234 and nc -zv  <NSX Manager IP/FQDN> 1235
    • TEP to TEP pings are successful: vmkping ++netstack=vxlan -I vmk10 <Destination vmk IP> -d -s <MTU Size> and vmkping ++netstack=vxlan -I vmk11 <Destination vmk IP> -d -s <MTU Size>

Environment

VCF 9.0.2
VCF NSX 9.0.2

Cause

This is a known issue in VCF 9.0.2 related to AggService status propagation.
Specifically, a race condition occurs in the call creation process during an opsagent restart. This prevents the Host Transport Node from correctly reporting its health telemetry to the NSX Manager UI.
There is no functional impact, the ESXi management and data plane remain fully operational.

Resolution

This is a known issue impacting VCF 9.0.2.

To restore the UI status of the ESX host to Up, services must be restarted in this specific order:

1. ssh to the ESX host as root user
2. Stop opsagent service
   /etc/init.d/nsx-opsagent stop
3. Stop exporter service
   /etc/init.d/nsx-exporter stop
4. Start opsagent service
   /etc/init.d/nsx-opsagent start
5. Start exporter service
   /etc/init.d/nsx-exporter start

Additional Information

Related Article:

ESX host status shows as "Unknown" in the NSX.