Expand DRA Driver Integration for GPU Operator

api/nvidia/v1/clusterpolicy_types.go

@@ -53,6 +53,8 @@ type ClusterPolicySpec struct {
 	Toolkit ToolkitSpec `json:"toolkit"`
 	// DevicePlugin component spec
 	DevicePlugin DevicePluginSpec `json:"devicePlugin"`
+	// DRADriver component spec
+	DRADriver DRADriverSpec `json:"draDriver"`
 	// DCGMExporter spec
 	DCGMExporter DCGMExporterSpec `json:"dcgmExporter"`
 	// DCGM component spec
@@ -985,6 +987,104 @@ type SandboxDevicePluginSpec struct {
 	HostNetwork *bool `json:"hostNetwork,omitempty"`
 }
 
+// DRADriverSpec defines the properties for the NVIDIA DRA Driver deployment
+type DRADriverSpec struct {
+	// NVIDIA DRA Driver image repository
+	// +kubebuilder:validation:Optional
+	Repository string `json:"repository,omitempty"`
+
+	// NVIDIA DRA Driver image name
+	// +kubebuilder:validation:Pattern=[a-zA-Z0-9\-]+
+	Image string `json:"image,omitempty"`
+
+	// NVIDIA DRA Driver image tag
+	// +kubebuilder:validation:Optional
+	Version string `json:"version,omitempty"`
+
+	// Image pull policy
+	// +kubebuilder:validation:Optional
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Image Pull Policy"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:imagePullPolicy"
+	ImagePullPolicy string `json:"imagePullPolicy,omitempty"`
+
+	// Image pull secrets
+	// +kubebuilder:validation:Optional
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Image pull secrets"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:io.kubernetes:Secret"
+	ImagePullSecrets []string `json:"imagePullSecrets,omitempty"`
+
+	// GPUs defines configuration for GPUs in the NVIDIA DRA Driver
+	GPUs DRADriverGPUs `json:"gpus,omitempty"`
+
+	// ComputeDomains defines configuration for ComputeDomains in the NVIDIA DRA Driver
+	ComputeDomains DRADriverComputeDomains `json:"computeDomains,omitempty"`
+}
+
+// DRADriverGPUs defines configuration for GPUs in the NVIDIA DRA Driver
+type DRADriverGPUs struct {
+	// Enabled indicates if GPUs are enabled in the NVIDIA DRA Driver
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Enable GPUs in the NVIDIA DRA Driver"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:booleanSwitch"
+	Enabled *bool `json:"enabled,omitempty"`
+
+	// KubeletPlugin defines configuration for the NVIDIA DRA Driver kubelet plugin
+	KubeletPlugin DRADriverKubeletPlugin `json:"kubeletPlugin,omitempty"`
+}
+
+// DRADriverComputeDomains defines configuration for ComputeDomains in the NVIDIA DRA Driver
+type DRADriverComputeDomains struct {
+	// Enabled indicates if ComputeDomains are enabled in the NVIDIA DRA Driver
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Enable ComputeDomains in the NVIDIA DRA Driver"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:booleanSwitch"
+	Enabled *bool `json:"enabled,omitempty"`
+
+	// Controller defines configuration for the NVIDIA DRA Driver controller
+	Controller DRADriverController `json:"controller,omitempty"`
+
+	// KubeletPlugin defines configuration for the NVIDIA DRA Driver kubelet plugin
+	KubeletPlugin DRADriverKubeletPlugin `json:"kubeletPlugin,omitempty"`
+}
+
+// DRADriverController defines configuration for the NVIDIA DRA Driver controller
+type DRADriverController struct {
+	// Optional: List of environment variables
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Environment Variables"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:advanced,urn:alm:descriptor:com.tectonic.ui:text"
+	Env []EnvVar `json:"env,omitempty"`
+
+	// Optional: Define resources requests and limits
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Resource Requirements"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:advanced,urn:alm:descriptor:com.tectonic.ui:resourceRequirements"
+	Resources *ResourceRequirements `json:"resources,omitempty"`
+
+	// Optional: Set tolerations
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Tolerations"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:advanced,urn:alm:descriptor:io.kubernetes:Tolerations"
+	Tolerations []corev1.Toleration `json:"tolerations,omitempty"`
+}
+
+// DRADriverKubeletPlugin defines configuration for the NVIDIA DRA Driver kubelet plugin
+type DRADriverKubeletPlugin struct {
+	// Optional: List of environment variables
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Environment Variables"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:advanced,urn:alm:descriptor:com.tectonic.ui:text"
+	Env []EnvVar `json:"env,omitempty"`
+
+	// Optional: Define resources requests and limits
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors=true
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.displayName="Resource Requirements"
+	// +operator-sdk:gen-csv:customresourcedefinitions.specDescriptors.x-descriptors="urn:alm:descriptor:com.tectonic.ui:advanced,urn:alm:descriptor:com.tectonic.ui:resourceRequirements"
+	Resources *ResourceRequirements `json:"resources,omitempty"`
+}
+
 // DCGMExporterSpec defines the properties for NVIDIA DCGM Exporter deployment
 type DCGMExporterSpec struct {
 	// Enabled indicates if deployment of NVIDIA DCGM Exporter through operator is enabled
@@ -2079,6 +2179,9 @@ func ImagePath(spec interface{}) (string, error) {
 	case *SandboxDevicePluginSpec:
 		config := spec.(*SandboxDevicePluginSpec)
 		return imagePath(config.Repository, config.Image, config.Version, "SANDBOX_DEVICE_PLUGIN_IMAGE")
+	case *DRADriverSpec:
+		config := spec.(*DRADriverSpec)
+		return imagePath(config.Repository, config.Image, config.Version, "DRA_DRIVER_IMAGE")
 	case *DCGMExporterSpec:
 		config := spec.(*DCGMExporterSpec)
 		return imagePath(config.Repository, config.Image, config.Version, "DCGM_EXPORTER_IMAGE")
@@ -2194,6 +2297,21 @@ func (p *DevicePluginSpec) IsEnabled() bool {
 	return *p.Enabled
 }
 
+// IsEnabled returns true if the DRA Driver is enabled through gpu-operator
+func (d *DRADriverSpec) IsEnabled() bool {
+	return d.IsGPUsEnabled() || d.IsComputeDomainsEnabled()
+}
+
+// IsGPUsEnabled returns true if the GPUs resource is enabled in the DRA Driver
+func (d *DRADriverSpec) IsGPUsEnabled() bool {
+	return d.GPUs.Enabled != nil && *d.GPUs.Enabled
+}
+
+// IsComputeDomainsEnabled returns true if the ComputeDomains resource is enabled in the DRA Driver
+func (d *DRADriverSpec) IsComputeDomainsEnabled() bool {
+	return d.ComputeDomains.Enabled != nil && *d.ComputeDomains.Enabled
+}
+
 // IsEnabled returns true if dcgm-exporter is enabled(default) through gpu-operator
 func (e *DCGMExporterSpec) IsEnabled() bool {
 	if e.Enabled == nil {

assets/state-dra-driver/0100_service_account.yaml

@@ -0,0 +1,11 @@
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: nvidia-dra-driver-controller
+  namespace: gpu-operator
+  labels:
+    helm.sh/chart: nvidia-dra-driver-26.4.0-dev
+    app.kubernetes.io/version: 26.4.0-dev
+    app.kubernetes.io/managed-by: Helm
+    app.kubernetes.io/name: nvidia-dra-driver
+    app.kubernetes.io/instance: nvidia-dra-driver

assets/state-dra-driver/0110_compute_domain_daemon-service_account.yaml

@@ -0,0 +1,5 @@
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: compute-domain-daemon-service-account
+  namespace: gpu-operator

assets/state-dra-driver/0120_kubeletplugin-service_account.yaml

@@ -0,0 +1,11 @@
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: nvidia-dra-driver-kubeletplugin
+  namespace: gpu-operator
+  labels:
+    helm.sh/chart: nvidia-dra-driver-26.4.0-dev
+    app.kubernetes.io/version: 26.4.0-dev
+    app.kubernetes.io/managed-by: Helm
+    app.kubernetes.io/name: nvidia-dra-driver
+    app.kubernetes.io/instance: nvidia-dra-driver

assets/state-dra-driver/0200_clusterrole.yaml

@@ -0,0 +1,56 @@
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: nvidia-dra-driver-clusterrole-controller
+rules:
+- apiGroups:
+  - resource.nvidia.com
+  resources:
+  - computedomains
+  verbs:
+  - get
+  - list
+  - watch
+  - update
+- apiGroups:
+  - resource.nvidia.com
+  resources:
+  - computedomains/status
+  verbs:
+  - update
+- apiGroups:
+  - resource.k8s.io
+  resources:
+  - resourceclaimtemplates
+  verbs:
+  - get
+  - list
+  - watch
+  - create
+  - update
+  - delete
+- apiGroups:
+  - coordination.k8s.io
+  resources:
+  - leases
+  verbs:
+  - get
+  - create
+  - update
+- apiGroups:
+  - ''
+  resources:
+  - nodes
+  verbs:
+  - get
+  - list
+  - watch
+  - update
+- apiGroups:
+  - ''
+  resources:
+  - pods
+  verbs:
+  - get
+  - list
+  - watch

assets/state-dra-driver/0210_role.yaml

@@ -0,0 +1,27 @@
+apiVersion: rbac.authorization.k8s.io/v1
+kind: Role
+metadata:
+  name: nvidia-dra-driver-role-controller
+  namespace: gpu-operator
+rules:
+- apiGroups:
+  - apps
+  resources:
+  - daemonsets
+  verbs:
+  - get
+  - list
+  - watch
+  - create
+  - update
+  - patch
+  - delete
+- apiGroups:
+  - resource.nvidia.com
+  resources:
+  - computedomaincliques
+  verbs:
+  - get
+  - list
+  - watch
+  - update