Add a new MultiNodeEnvironment CRD to setup the environment needed for running GPU workloads across multi-nodes

.gitignore

@@ -1,7 +1,8 @@
 .cache/
 .bash_history
-/nvidia-dra-controller
-/nvidia-dra-plugin
+/nvidia-dra-imex-controller
+/nvidia-dra-imex-plugin
+/nvidia-dra-gpu-plugin
 .idea
 [._]*.sw[a-p]
 coverage.out

Makefile

@@ -79,7 +79,7 @@ goimports:
 	find . -name \*.go \
 			-not -name "zz_generated.deepcopy.go" \
 			-not -path "./vendor/*" \
-			-not -path "./pkg/nvidia.com/resource/clientset/versioned/*" \
+			-not -path "./$(PKG_BASE)/clientset/versioned/*" \
 		-exec goimports -local $(MODULE) -w {} \;
 
 golangci-lint:
@@ -101,7 +101,22 @@ coverage: test
 	cat $(COVERAGE_FILE) | grep -v "_mock.go" > $(COVERAGE_FILE).no-mocks
 	go tool cover -func=$(COVERAGE_FILE).no-mocks
 
-generate: generate-deepcopy fmt
+generate: generate-crds generate-informers fmt
+
+generate-crds: generate-deepcopy .remove-crds
+	for dir in $(CLIENT_SOURCES); do \
+		controller-gen crd:crdVersions=v1 \
+			paths=$(CURDIR)/$${dir} \
+			output:crd:dir=$(CURDIR)/deployments/helm/tmp_crds; \
+	done
+	mkdir -p $(CURDIR)/deployments/helm/$(GPU_DRIVER_NAME)/crds
+	cp -R $(CURDIR)/deployments/helm/tmp_crds/* \
+		$(CURDIR)/deployments/helm/$(GPU_DRIVER_NAME)/crds
+	mkdir -p $(CURDIR)/deployments/helm/$(IMEX_DRIVER_NAME)/crds
+	cp -R $(CURDIR)/deployments/helm/tmp_crds/* \
+		$(CURDIR)/deployments/helm/$(IMEX_DRIVER_NAME)/crds
+	rm -rf $(CURDIR)/deployments/helm/tmp_crds
+
 
 generate-deepcopy: .remove-deepcopy
 	for dir in $(DEEPCOPY_SOURCES); do \
@@ -111,16 +126,69 @@ generate-deepcopy: .remove-deepcopy
 			output:object:dir=$(CURDIR)/$${dir}; \
 	done
 
+generate-informers: .remove-informers generate-listers
+	informer-gen \
+		--go-header-file=$(CURDIR)/hack/boilerplate.go.txt \
+		--output-package "$(MODULE)/$(PKG_BASE)/informers" \
+        --input-dirs "$(shell for api in $(CLIENT_APIS); do echo -n "$(MODULE)/$(API_BASE)/$$api,"; done | sed 's/,$$//')" \
+		--output-base "$(CURDIR)/pkg/tmp_informers" \
+		--versioned-clientset-package "$(MODULE)/$(PKG_BASE)/clientset/versioned" \
+		--listers-package "$(MODULE)/$(PKG_BASE)/listers"
+	mkdir -p $(CURDIR)/$(PKG_BASE)
+	mv $(CURDIR)/pkg/tmp_informers/$(MODULE)/$(PKG_BASE)/informers \
+	   $(CURDIR)/$(PKG_BASE)/informers
+	rm -rf $(CURDIR)/pkg/tmp_informers
+
+generate-listers: .remove-listers generate-clientset
+	lister-gen \
+		--go-header-file=$(CURDIR)/hack/boilerplate.go.txt \
+		--output-package "$(MODULE)/$(PKG_BASE)/listers" \
+        --input-dirs "$(shell for api in $(CLIENT_APIS); do echo -n "$(MODULE)/$(API_BASE)/$$api,"; done | sed 's/,$$//')" \
+		--output-base "$(CURDIR)/pkg/tmp_listers"
+	mkdir -p $(CURDIR)/$(PKG_BASE)
+	mv $(CURDIR)/pkg/tmp_listers/$(MODULE)/$(PKG_BASE)/listers \
+	   $(CURDIR)/$(PKG_BASE)/listers
+	rm -rf $(CURDIR)/pkg/tmp_listers
+
+generate-clientset: .remove-clientset
+	client-gen \
+		--go-header-file=$(CURDIR)/hack/boilerplate.go.txt \
+		--clientset-name "versioned" \
+		--build-tag "ignore_autogenerated" \
+		--output-package "$(MODULE)/$(PKG_BASE)/clientset" \
+		--input-base "$(MODULE)/$(API_BASE)" \
+		--output-base "$(CURDIR)/pkg/tmp_clientset" \
+		--input "$(shell echo $(CLIENT_APIS) | tr ' ' ',')" \
+		--plural-exceptions "$(shell echo $(PLURAL_EXCEPTIONS) | tr ' ' ',')"
+	mkdir -p $(CURDIR)/$(PKG_BASE)
+	mv $(CURDIR)/pkg/tmp_clientset/$(MODULE)/$(PKG_BASE)/clientset \
+       $(CURDIR)/$(PKG_BASE)/clientset
+	rm -rf $(CURDIR)/pkg/tmp_clientset
+
+.remove-crds:
+	rm -rf $(CURDIR)/deployments/helm/$(DRIVER_NAME)/crds
+
 .remove-deepcopy:
 	for dir in $(DEEPCOPY_SOURCES); do \
 		rm -f $(CURDIR)/$${dir}/zz_generated.deepcopy.go; \
 	done
 
+.remove-clientset:
+	rm -rf $(CURDIR)/$(PKG_BASE)/clientset
+
+.remove-listers:
+	rm -rf $(CURDIR)/$(PKG_BASE)/listers
+
+.remove-informers:
+	rm -rf $(CURDIR)/$(PKG_BASE)/informers
+
 build-image:
 	$(DOCKER) build \
 		--progress=plain \
 		--build-arg GOLANG_VERSION="$(GOLANG_VERSION)" \
 		--build-arg CLIENT_GEN_VERSION="$(CLIENT_GEN_VERSION)" \
+		--build-arg LISTER_GEN_VERSION="$(LISTER_GEN_VERSION)" \
+		--build-arg INFORMER_GEN_VERSION="$(INFORMER_GEN_VERSION)" \
 		--build-arg CONTROLLER_GEN_VERSION="$(CONTROLLER_GEN_VERSION)" \
 		--build-arg GOLANGCI_LINT_VERSION="$(GOLANGCI_LINT_VERSION)" \
 		--build-arg MOQ_VERSION="$(MOQ_VERSION)" \

README.md

@@ -84,9 +84,8 @@ This should show two pods running in the `nvidia` namespace:
 kubectl get pods -n nvidia
 ```
 ```
-NAME                                                          READY   STATUS    RESTARTS   AGE
-nvidia-dra-driver-k8s-dra-driver-controller-844fcb94b-ktbkc   1/1     Running   0          69s
-nvidia-dra-driver-k8s-dra-driver-kubelet-plugin-5vfp9         1/1     Running   0          69s
+NAME                                                              READY   STATUS    RESTARTS   AGE
+nvidia-dra-driver-k8s-dra-driver-gpu-kubelet-plugin-5vfp9         1/1     Running   0          69s
 ```
 
 ### Run the examples by following the steps in the demo script

api/nvidia.com/resource/gpu/v1alpha1/api.go → api/nvidia.com/resource/v1beta1/api.go

@@ -14,7 +14,7 @@
  * limitations under the License.
  */
 
-package v1alpha1
+package v1beta1
 
 import (
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
@@ -24,12 +24,14 @@ import (
 )
 
 const (
-	GroupName = "gpu.nvidia.com"
-	Version   = "v1alpha1"
+	GroupName = "resource.nvidia.com"
+	Version   = "v1beta1"
 
 	GpuConfigKind         = "GpuConfig"
 	MigDeviceConfigKind   = "MigDeviceConfig"
 	ImexChannelConfigKind = "ImexChannelConfig"
+	ImexDaemonConfigKind  = "ImexDaemonConfig"
+	ComputeDomainKind     = "ComputeDomain"
 )
 
 // Interface defines the set of common APIs for all configs
@@ -56,6 +58,8 @@ func init() {
 		&GpuConfig{},
 		&MigDeviceConfig{},
 		&ImexChannelConfig{},
+		&ImexDaemonConfig{},
+		&ComputeDomain{},
 	)
 	metav1.AddToGroupVersion(scheme, schemeGroupVersion)