Controller Service selector/port mismatch prevents agent connection

[piwi3910]

Description

The novaedge-controller ClusterIP service has:

• Selector: app.kubernetes.io/instance=novaedge, app.kubernetes.io/name=novaedge
• Ports: 8082/TCP, 8080/TCP

But controller pods have:

• Labels: app.kubernetes.io/name=novaedge-controller (not novaedge)
• Missing: app.kubernetes.io/instance=novaedge

The agent connects to <service-ip>:9090 for gRPC config stream, but port 9090 is not exposed by the service.

Result: Service endpoints are <none>, causing agent connection timeouts on some nodes.

Impact

• Agent on master-11 unable to receive config snapshots
• 7/8 agents work (likely connecting via other mechanism)

Fix

Either update the service selector/ports to match the actual deployment labels, or fix the deployment template labels to match the service selector.