[BUG] Helm chart liveness and readiness probes hardcode port 8081, ignoring healthProbeBindAddress

What happened?

In charts/node-readiness-controller/values.yaml, healthProbeBindAddress: ":8081" is exposed as a configurable value, and templates/deployment.yaml passes --health-probe-bind-address={{ .Values.healthProbeBindAddress }} to /manager.

However, in templates/deployment.yaml, the container port and probe ports are hardcoded to 8081:

• ports[0].containerPort is hardcoded to 8081.
• livenessProbe.httpGet.port is hardcoded to 8081.
• readinessProbe.httpGet.port is hardcoded to 8081.

When a user overrides healthProbeBindAddress (such as healthProbeBindAddress: ":9090" or "0.0.0.0:8082" to avoid port collisions on shared networks), the controller manager listens on :9090, but kubelet probes port 8081. Because nothing is listening on 8081, the probes fail with connection refused. The controller is never marked ready, and after the failure threshold kubelet kills and restarts the container into a CrashLoopBackOff.

In addition, lines 77-84 in templates/deployment.yaml have extra indentation under ports:.

Steps to Reproduce

1. Render deployment.yaml with a custom probe bind address:

helm template test-nrc charts/node-readiness-controller \
  --set healthProbeBindAddress=":9090" \
  -s templates/deployment.yaml

2. Check the rendered arguments vs probe definitions:

          args:
            - --health-probe-bind-address=:9090
          ports:
                      - name: http
                        containerPort: 8081
          livenessProbe:
            httpGet:
              path: /healthz
              port: 8081
          readinessProbe:
            httpGet:
              path: /readyz
              port: 8081

Expected Behavior

The container port and probe ports should derive their port from healthProbeBindAddress (e.g. parsing the port via regex or using named port http with matching containerPort), so changing healthProbeBindAddress keeps probes pointed at the active probe port.

Controller Version / Image Tag

main (commit ca2acdc)

Kubernetes Version

Client Version: v1.32.0

Controller Logs

Liveness probe failed: Get "http://10.244.1.5:8081/healthz": dial tcp 10.244.1.5:8081: connect: connection refused
Readiness probe failed: Get "http://10.244.1.5:8081/readyz": dial tcp 10.244.1.5:8081: connect: connection refused

Additional Environment Details

Can submit a PR with the template adjustment and test coverage in charts/node-readiness-controller/tests/deployment_test.yaml.

AI Tools Usage

• AI tools were used (complete below)

How They Were Used

Used Antigravity IDE to inspect helm templates, verify rendered output with custom healthProbeBindAddress values, and prepare the issue description.

Start with charts/node-readiness-controller/templates/deployment.yaml and render it using the issue's helm template command with a custom healthProbeBindAddress. Check charts/node-readiness-controller/tests/deployment_test.yaml for existing coverage. Done means the rendered container port and both probe ports follow the configured health probe address, the indentation is corrected, and the deployment tests pass.