Attempts to register a Supervisor Cluster with Tanzu Mission Control Self-Managed (TMC SM) fail. Communication from the Supervisor nodes is unable to reach the TMC SM API, preventing the bootstrap process from initiating.
While the TMC SM API remains accessible from other network segments, the Supervisor nodes specifically time out when attempting to reach the Load Balancer VIP. The presumed traffic path is from the Control Plane VM (CPVM) to the Load Balancer; however, the Supervisor remains unable to establish a handshake even when the Load Balancer appears healthy.
This failure is caused by the Supervisor's networking behavior where traffic destined for a LoadBalancer-type service is intercepted locally via iptables and redirected internally.
Instead of exiting the node to the Load Balancer VIP (##.##.4.46), the Supervisor’s kube-proxy rules perform a Destination Network Address Translation (DNAT). This redirects the traffic directly to the internal NodePort IP and Port of the TMC Service (e.g., ##.##.12.238:30833).
The network path shifts as follows:
The issue occurs when a firewall rule blocks traffic originating from the Supervisor Node's management subnet to the TMC Service Node network (##.##.12.0/24) on the specific NodePort range. Because the traffic never actually reaches the Load Balancer, rules applied only to the VIP are insufficient.
The following iptables chains on the Supervisor confirm this internal redirection:
To resolve this issue, update the underlying network firewall policies to allow the traffic path created by the Supervisor's internal networking.