After performing a manual rolling reboot of NSX manager nodes, customer observed transient "configuration failure" status for some hosts.

Products

• VMware NSX

Issue/Introduction

• Multiple ESXi Transport Nodes report a Failed or Disconnected state in the NSX Manager UI.
• The issue occurs during or immediately after a manually initiated rolling reboot of NSX Manager nodes.
• No VM connectivity issues are observed.
• Under /var/log/proton/nsxapi.log on the NSX Manager, you see entries similar to: 
  Corfu StreamListener entry UfoTable ClientHeartbeat, (####) : payload status_change_reason: "messaging_heartbeat_client_unresponsive" Heartbeat status is down. Returning HOST_DISCONNECTED status for node.
• Manual reboot of managers can be traced under /var/log/nsxcli.log.
  2026-06-12T15:46:13.392Z 1760648 cli.server.cli_command_service INFO {10000} CMD: reboot
2026-06-12T15:54:18.971Z 2730636 cli.server.cli_command_service INFO {10000} CMD: reboot
2026-06-12T16:02:35.542Z 2510792 cli.server.cli_command_service INFO {10000} CMD: reboot

Environment

VMware NSX 

Cause

This issue arises from an edge case encountered during the manual restart of manager nodes.

In an UC workflow, hosts would be evacuated (re-sharded) from a manager node before it rebooted. But this is not done during manual reboot.

When a manager node suddenly "disappears" during reboot, hosts assigned to that manager fail over to some other manager node. But this failover results in the appearance of a host heartbeat being lost, which translates into the host reporting "configuration failure".

Once the manager nodes reboot, the host status is cleared again in the course of restoring normal MP connectivity.

Resolution

No active remediation is required.

The system automatically resolves the false disconnection status once the NSX Manager node's Control plane services are fully restored and the connection statuses are reset.