[dcgm-exporter] Add option hostNetwork for dcgmExporter

api/nvidia/v1/clusterpolicy_types.go

@@ -884,6 +884,13 @@ type DCGMExporterSpec struct {
 	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:io.kubernetes:Secret"
 	ImagePullSecrets []string `json:"imagePullSecrets,omitempty"`
 
+	// HostNetwork
+	// +kubebuilder:validation:Optional
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="HostNetwork"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:booleanSwitch"
+	HostNetwork *bool `json:"hostNetwork,omitempty"`
+
 	// Optional: Define resources requests and limits for each pod
 	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
 	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Resource Requirements"

bundle/manifests/nvidia.com_clusterpolicies.yaml

@@ -337,6 +337,9 @@ spec:
                       - name
                       type: object
                     type: array
+                  hostNetwork:
+                    description: Run the DCGM Exporter pod on the host’s network, sharing the node’s network interfaces and IP address
+                    type: boolean
                   image:
                     description: NVIDIA DCGM Exporter image name
                     pattern: '[a-zA-Z0-9\-]+'

config/crd/bases/nvidia.com_clusterpolicies.yaml

@@ -337,6 +337,9 @@ spec:
                       - name
                       type: object
                     type: array
+                  hostNetwork:
+                    description: Run the DCGM Exporter pod on the host’s network, sharing the node’s network interfaces and IP address
+                    type: boolean
                   image:
                     description: NVIDIA DCGM Exporter image name
                     pattern: '[a-zA-Z0-9\-]+'

controllers/object_controls.go

@@ -1608,6 +1608,10 @@ func TransformDCGMExporter(obj *appsv1.DaemonSet, config *gpuv1.ClusterPolicySpe
 	if len(config.DCGMExporter.ImagePullSecrets) > 0 {
 		addPullSecrets(&obj.Spec.Template.Spec, config.DCGMExporter.ImagePullSecrets)
 	}
+	// set hostNetwork
+	if config.DCGMExporter.HostNetwork != nil {
+		obj.Spec.Template.Spec.HostNetwork = *config.DCGMExporter.HostNetwork
+	}
 	// set resource limits
 	if config.DCGMExporter.Resources != nil {
 		// apply resource limits to all containers

deployments/gpu-operator/crds/nvidia.com_clusterpolicies.yaml

@@ -337,6 +337,9 @@ spec:
                       - name
                       type: object
                     type: array
+                  hostNetwork:
+                    description: Run the DCGM Exporter pod on the host’s network, sharing the node’s network interfaces and IP address
+                    type: boolean
                   image:
                     description: NVIDIA DCGM Exporter image name
                     pattern: '[a-zA-Z0-9\-]+'

deployments/gpu-operator/templates/clusterpolicy.yaml

@@ -499,6 +499,9 @@ spec:
     {{- if .Values.dcgmExporter.imagePullSecrets }}
     imagePullSecrets: {{ toYaml .Values.dcgmExporter.imagePullSecrets | nindent 6 }}
     {{- end }}
+    {{- if .Values.dcgmExporter.hostNetwork }}
+    hostNetwork: {{ .Values.dcgmExporter.hostNetwork }}
+    {{- end }}
     {{- if .Values.dcgmExporter.resources }}
     resources: {{ toYaml .Values.dcgmExporter.resources | nindent 6 }}
     {{- end }}

deployments/gpu-operator/values.yaml

@@ -322,6 +322,7 @@ dcgmExporter:
       value: "true"
     - name: DCGM_EXPORTER_COLLECTORS
       value: "/etc/dcgm-exporter/dcp-metrics-included.csv"
+  hostNetwork: false
   resources: {}
   service:
     internalTrafficPolicy: Cluster