add configurable EPP flags

Author: salonichf5

charts/nginx-gateway-fabric/README.md

@@ -245,7 +245,7 @@ The following table lists the configurable parameters of the NGINX Gateway Fabri
 | `nginx.usage.resolver` | The nameserver used to resolve the NGINX Plus usage reporting endpoint. Used with NGINX Instance Manager. | string | `""` |
 | `nginx.usage.secretName` | The name of the Secret containing the JWT for NGINX Plus usage reporting. Must exist in the same namespace that the NGINX Gateway Fabric control plane is running in (default namespace: nginx-gateway). | string | `"nplus-license"` |
 | `nginx.usage.skipVerify` | Disable client verification of the NGINX Plus usage reporting server certificate. | bool | `false` |
-| `nginxGateway` | The nginxGateway section contains configuration for the NGINX Gateway Fabric control plane deployment. | object | `{"affinity":{},"autoscaling":{"enable":false},"config":{"logging":{"level":"info"}},"configAnnotations":{},"extraVolumeMounts":[],"extraVolumes":[],"gatewayClassAnnotations":{},"gatewayClassName":"nginx","gatewayControllerName":"gateway.nginx.org/nginx-gateway-controller","gwAPIExperimentalFeatures":{"enable":false},"gwAPIInferenceExtension":{"enable":false},"image":{"pullPolicy":"Always","repository":"ghcr.io/nginx/nginx-gateway-fabric","tag":"edge"},"kind":"deployment","labels":{},"leaderElection":{"enable":true,"lockName":""},"lifecycle":{},"metrics":{"enable":true,"port":9113,"secure":false},"name":"","nodeSelector":{},"podAnnotations":{},"productTelemetry":{"enable":true},"readinessProbe":{"enable":true,"initialDelaySeconds":3,"port":8081},"replicas":1,"resources":{},"service":{"annotations":{},"labels":{}},"serviceAccount":{"annotations":{},"imagePullSecret":"","imagePullSecrets":[],"name":""},"snippetsFilters":{"enable":false},"terminationGracePeriodSeconds":30,"tolerations":[],"topologySpreadConstraints":[]}` |
+| `nginxGateway` | The nginxGateway section contains configuration for the NGINX Gateway Fabric control plane deployment. | object | `{"affinity":{},"autoscaling":{"enable":false},"config":{"logging":{"level":"info"}},"configAnnotations":{},"extraVolumeMounts":[],"extraVolumes":[],"gatewayClassAnnotations":{},"gatewayClassName":"nginx","gatewayControllerName":"gateway.nginx.org/nginx-gateway-controller","gwAPIExperimentalFeatures":{"enable":false},"gwAPIInferenceExtension":{"enable":false,"endpointPicker":{"disableTLS":false,"skipVerify":true}},"image":{"pullPolicy":"Always","repository":"ghcr.io/nginx/nginx-gateway-fabric","tag":"edge"},"kind":"deployment","labels":{},"leaderElection":{"enable":true,"lockName":""},"lifecycle":{},"metrics":{"enable":true,"port":9113,"secure":false},"name":"","nodeSelector":{},"podAnnotations":{},"productTelemetry":{"enable":true},"readinessProbe":{"enable":true,"initialDelaySeconds":3,"port":8081},"replicas":1,"resources":{},"service":{"annotations":{},"labels":{}},"serviceAccount":{"annotations":{},"imagePullSecret":"","imagePullSecrets":[],"name":""},"snippetsFilters":{"enable":false},"terminationGracePeriodSeconds":30,"tolerations":[],"topologySpreadConstraints":[]}` |
 | `nginxGateway.affinity` | The affinity of the NGINX Gateway Fabric control plane pod. | object | `{}` |
 | `nginxGateway.autoscaling` | Autoscaling configuration for the NGINX Gateway Fabric control plane. | object | `{"enable":false}` |
 | `nginxGateway.autoscaling.enable` | Enable or disable Horizontal Pod Autoscaler for the control plane. | bool | `false` |
@@ -258,6 +258,9 @@ The following table lists the configurable parameters of the NGINX Gateway Fabri
 | `nginxGateway.gatewayControllerName` | The name of the Gateway controller. The controller name must be of the form: DOMAIN/PATH. The controller's domain is gateway.nginx.org. | string | `"gateway.nginx.org/nginx-gateway-controller"` |
 | `nginxGateway.gwAPIExperimentalFeatures.enable` | Enable the experimental features of Gateway API which are supported by NGINX Gateway Fabric. Requires the Gateway APIs installed from the experimental channel. | bool | `false` |
 | `nginxGateway.gwAPIInferenceExtension.enable` | Enable Gateway API Inference Extension support. Allows for configuring InferencePools to route traffic to AI workloads. | bool | `false` |
+| `nginxGateway.gwAPIInferenceExtension.endpointPicker` | EndpointPicker TLS configuration. | object | `{"disableTLS":false,"skipVerify":true}` |
+| `nginxGateway.gwAPIInferenceExtension.endpointPicker.disableTLS` | Disable TLS for EndpointPicker communication. By default, TLS is enabled. Set to true only for development/testing or when using a service mesh for encryption. | bool | `false` |
+| `nginxGateway.gwAPIInferenceExtension.endpointPicker.skipVerify` | Disables TLS certificate verification when connecting to the EndpointPicker. By default, certificate verification is disabled. REQUIRED: Must be true until Gateway API Inference Extension EndpointPicker supports mounting certificates. See: https://github.com/kubernetes-sigs/gateway-api-inference-extension/issues/1556 | bool | `true` |
 | `nginxGateway.image` | The image configuration for the NGINX Gateway Fabric control plane. | object | `{"pullPolicy":"Always","repository":"ghcr.io/nginx/nginx-gateway-fabric","tag":"edge"}` |
 | `nginxGateway.image.repository` | The NGINX Gateway Fabric image to use | string | `"ghcr.io/nginx/nginx-gateway-fabric"` |
 | `nginxGateway.kind` | The kind of the NGINX Gateway Fabric installation - currently, only deployment is supported. | string | `"deployment"` |

charts/nginx-gateway-fabric/templates/deployment.yaml

@@ -103,6 +103,10 @@ spec:
         {{- if .Values.nginxGateway.gwAPIInferenceExtension.enable }}
         - --gateway-api-inference-extension
         {{- end }}
+        {{- if .Values.nginxGateway.gwAPIInferenceExtension.endpointPicker.disableTLS }}
+        - --endpoint-picker-disable-tls
+        {{- end }}
+        - --endpoint-picker-tls-skip-verify={{ .Values.nginxGateway.gwAPIInferenceExtension.endpointPicker.skipVerify }}
         {{- if .Values.nginxGateway.snippetsFilters.enable }}
         - --snippets-filters
         {{- end }}

charts/nginx-gateway-fabric/values.schema.json

@@ -846,6 +846,28 @@
               "required": [],
               "title": "enable",
               "type": "boolean"
+            },
+            "endpointPicker": {
+              "description": "EndpointPicker TLS configuration.",
+              "properties": {
+                "disableTLS": {
+                  "default": false,
+                  "description": "Disable TLS for EndpointPicker communication. By default, TLS is enabled.\nSet to true only for development/testing or when using a service mesh for encryption.",
+                  "required": [],
+                  "title": "disableTLS",
+                  "type": "boolean"
+                },
+                "skipVerify": {
+                  "default": true,
+                  "description": "Disables TLS certificate verification when connecting to the EndpointPicker.\nBy default, certificate verification is disabled.\nREQUIRED: Must be true until Gateway API Inference Extension EndpointPicker supports mounting certificates.\nSee: https://github.com/kubernetes-sigs/gateway-api-inference-extension/issues/1556",
+                  "required": [],
+                  "title": "skipVerify",
+                  "type": "boolean"
+                }
+              },
+              "required": [],
+              "title": "endpointPicker",
+              "type": "object"
             }
           },
           "required": [],

charts/nginx-gateway-fabric/values.yaml

@@ -214,6 +214,19 @@ nginxGateway:
     # -- Enable Gateway API Inference Extension support. Allows for configuring InferencePools to route traffic to AI workloads.
     enable: false
 
+    # -- EndpointPicker TLS configuration.
+    endpointPicker:
+      # -- Disable TLS for EndpointPicker communication. By default, TLS is enabled.
+      # Set to true only for development/testing or when using a service mesh for encryption.
+      disableTLS: false
+
+      # -- Disables TLS certificate verification when connecting to the EndpointPicker.
+      # By default, certificate verification is disabled.
+      # REQUIRED: Must be true until Gateway API Inference Extension EndpointPicker supports mounting certificates.
+      # See: https://github.com/kubernetes-sigs/gateway-api-inference-extension/issues/1556
+      skipVerify: true
+
+
   snippetsFilters:
     # -- Enable SnippetsFilters feature. SnippetsFilters allow inserting NGINX configuration into the generated NGINX
     # config for HTTPRoute and GRPCRoute resources.

cmd/gateway/commands.go

@@ -37,9 +37,11 @@ const (
 		`The controller name must be of the form: DOMAIN/PATH. The controller's domain is '%s'`
 	plusFlag = "nginx-plus"
 
-	serverTLSSecret               = "server-tls"
-	agentTLSSecret                = "agent-tls"
-	nginxOneTelemetryEndpointHost = "agent.connect.nginx.com"
+	serverTLSSecret                 = "server-tls"
+	agentTLSSecret                  = "agent-tls"
+	nginxOneTelemetryEndpointHost   = "agent.connect.nginx.com"
+	endpointPickerDisableTLSFlag    = "endpoint-picker-disable-tls"
+	endpointPickerTLSSkipVerifyFlag = "endpoint-picker-tls-skip-verify"
 )
 
 // usageReportParams holds the parameters for building the usage report configuration for PLUS.
@@ -162,6 +164,9 @@ func createControllerCommand() *cobra.Command {
 		nginxDockerSecrets = stringSliceValidatingValue{
 			validator: validateResourceName,
 		}
+
+		endpointPickerDisableTLS        bool
+		endpointPickerTLSSkipVerifyFlag = true
 	)
 
 	usageReportParams := usageReportParams{
@@ -288,6 +293,8 @@ func createControllerCommand() *cobra.Command {
 					EndpointPort:           nginxOneConsoleTelemetryEndpointPort.value,
 					EndpointTLSSkipVerify:  nginxOneConsoleTLSSkipVerify,
 				},
+				EndpointPickerDisableTLS:    endpointPickerDisableTLS,
+				EndpointPickerTLSSkipVerify: endpointPickerTLSSkipVerifyFlag,
 			}
 
 			if err := controller.StartManager(conf); err != nil {
@@ -441,6 +448,8 @@ func createControllerCommand() *cobra.Command {
 			"traffic to AI workloads.",
 	)
 
+	addEPPConnectionFlags(cmd, endpointPickerDisableTLS, endpointPickerTLSSkipVerifyFlag)
+
 	cmd.Flags().Var(
 		&nginxDockerSecrets,
 		nginxDockerSecretFlag,
@@ -758,19 +767,44 @@ func createSleepCommand() *cobra.Command {
 }
 
 func createEndpointPickerCommand() *cobra.Command {
+	var endpointPickerDisableTLS bool
+	endpointPickerTLSSkipVerifyFlag := true
 	cmd := &cobra.Command{
 		Use:   "endpoint-picker",
 		Short: "Shim server for communication between NGINX and the Gateway API Inference Extension Endpoint Picker",
 		RunE: func(_ *cobra.Command, _ []string) error {
 			logger := ctlrZap.New().WithName("endpoint-picker-shim")
-			handler := createEndpointPickerHandler(realExtProcClientFactory(), logger)
+			handler := createEndpointPickerHandler(
+				realExtProcClientFactory(endpointPickerDisableTLS, endpointPickerTLSSkipVerifyFlag),
+				logger,
+			)
 			return endpointPickerServer(handler)
 		},
 	}
 
+	addEPPConnectionFlags(cmd, endpointPickerDisableTLS, endpointPickerTLSSkipVerifyFlag)
+
 	return cmd
 }
 
+func addEPPConnectionFlags(cmd *cobra.Command, disableTLS, tlsSkipVerify bool) {
+	cmd.Flags().BoolVar(
+		&disableTLS,
+		endpointPickerDisableTLSFlag,
+		false,
+		"Disables TLS when connecting to the EndpointPicker. "+
+			"Set to true only for development/testing or when using a service mesh for encryption.",
+	)
+
+	cmd.Flags().BoolVar(
+		&tlsSkipVerify,
+		endpointPickerTLSSkipVerifyFlag,
+		true,
+		"Disables server certificate verification when connecting to the EndpointPicker, if TLS is enabled. "+
+			"REQUIRED: Must be true until Gateway API Inference Extension EndpointPicker supports mounting certificates.",
+	)
+}
+
 func parseFlags(flags *pflag.FlagSet) ([]string, []string) {
 	var flagKeys, flagValues []string
 

cmd/gateway/commands_test.go

@@ -161,6 +161,8 @@ func TestControllerCmdFlagValidation(t *testing.T) {
 				"--nginx-one-telemetry-endpoint-host=telemetry-endpoint-host",
 				"--nginx-one-telemetry-endpoint-port=443",
 				"--nginx-one-tls-skip-verify",
+				"--endpoint-picker-disable-tls",
+				"--endpoint-picker-tls-skip-verify",
 			},
 			wantErr: false,
 		},
@@ -924,3 +926,51 @@ func TestUsageReportConfig(t *testing.T) {
 		})
 	}
 }
+
+func TestEndpointPickerFlags(t *testing.T) {
+	t.Parallel()
+	tests := []flagTestCase{
+		{
+			name: "valid flags with default values",
+			args: []string{
+				"--endpoint-picker-disable-tls=false",
+				"--endpoint-picker-tls-skip-verify=true",
+			},
+			wantErr: false,
+		},
+		{
+			name: "valid flags with changed values",
+			args: []string{
+				"--endpoint-picker-disable-tls=true",
+				"--endpoint-picker-tls-skip-verify=false",
+			},
+			wantErr: false,
+		},
+		{
+			name: "endpoint-picker-disable-tls is not a bool",
+			args: []string{
+				"--endpoint-picker-disable-tls=not-a-bool",
+			},
+			wantErr: true,
+			expectedErrPrefix: `invalid argument "not-a-bool" for "--endpoint-picker-disable-tls" flag:` +
+				` strconv.ParseBool: parsing "not-a-bool": invalid syntax`,
+		},
+		{
+			name: "endpoint-picker-tls-skip-verify is not a bool",
+			args: []string{
+				"--endpoint-picker-tls-skip-verify=not-a-bool",
+			},
+			wantErr: true,
+			expectedErrPrefix: `invalid argument "not-a-bool" for "--endpoint-picker-tls-skip-verify" flag:` +
+				` strconv.ParseBool: parsing "not-a-bool": invalid syntax`,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			t.Parallel()
+			cmd := createEndpointPickerCommand()
+			testFlag(t, cmd, test)
+		})
+	}
+}

cmd/gateway/endpoint_picker.go

@@ -35,20 +35,19 @@ func endpointPickerServer(handler http.Handler) error {
 }
 
 // realExtProcClientFactory returns a factory that creates a new gRPC connection and client per request.
-func realExtProcClientFactory() extProcClientFactory {
+func realExtProcClientFactory(disableTLS, tlsSkipVerify bool) extProcClientFactory {
 	return func(target string) (extprocv3.ExternalProcessorClient, func() error, error) {
 		var opts []grpc.DialOption
-		enableTLS := true
-		insecureSkipVerify := true
 
-		if !enableTLS {
+		if disableTLS {
 			opts = append(opts, grpc.WithTransportCredentials(insecure.NewCredentials()))
 		} else {
 			creds := credentials.NewTLS(&tls.Config{
-				InsecureSkipVerify: insecureSkipVerify, //nolint:gosec
+				InsecureSkipVerify: tlsSkipVerify, //nolint:gosec
 			})
 			opts = append(opts, grpc.WithTransportCredentials(creds))
 		}
+
 		conn, err := grpc.NewClient(target, opts...)
 		if err != nil {
 			return nil, nil, err

deploy/azure/deploy.yaml

@@ -286,6 +286,7 @@ spec:
         - --metrics-port=9113
         - --health-port=8081
         - --leader-election-lock-name=nginx-gateway-leader-election
+        - --endpoint-picker-tls-skip-verify=true
         env:
         - name: POD_NAMESPACE
           valueFrom:

deploy/default/deploy.yaml

@@ -286,6 +286,7 @@ spec:
         - --metrics-port=9113
         - --health-port=8081
         - --leader-election-lock-name=nginx-gateway-leader-election
+        - --endpoint-picker-tls-skip-verify=true
         env:
         - name: POD_NAMESPACE
           valueFrom:

deploy/experimental-nginx-plus/deploy.yaml

@@ -295,6 +295,7 @@ spec:
         - --health-port=8081
         - --leader-election-lock-name=nginx-gateway-leader-election
         - --gateway-api-experimental-features
+        - --endpoint-picker-tls-skip-verify=true
         env:
         - name: POD_NAMESPACE
           valueFrom: