Merge pull request #127714 from dims/re-add-nvidia-gpu-device-plugin.yaml-in-test-suite-itself

Re-add nvidia-gpu-device-plugin.yaml in test suite itself

test/e2e/node/gpu.go

@@ -341,7 +341,7 @@ func SetupNVIDIAGPUNode(ctx context.Context, f *framework.Framework) {
 	} else {
 		// Using default local DaemonSet
 		framework.Logf("Using default local nvidia-driver-installer daemonset manifest.")
-		data, err := e2etestfiles.Read("test/e2e/testing-manifests/scheduling/nvidia-driver-installer.yaml")
+		data, err := e2etestfiles.Read("test/e2e/testing-manifests/gpu/gce/nvidia-driver-installer.yaml")
 		framework.ExpectNoError(err, "failed to read local manifest for nvidia-driver-installer daemonset")
 		ds, err = e2emanifest.DaemonSetFromData(data)
 		framework.ExpectNoError(err, "failed to parse local manifest for nvidia-driver-installer daemonset")
@@ -349,14 +349,27 @@ func SetupNVIDIAGPUNode(ctx context.Context, f *framework.Framework) {
 
 	prev, err := f.ClientSet.AppsV1().DaemonSets(f.Namespace.Name).Get(ctx, ds.Name, metav1.GetOptions{})
 	if err == nil && prev != nil {
-		framework.Logf("Daemonset already installed, skipping...")
+		framework.Logf("nvidia-driver-installer Daemonset already installed, skipping...")
-		return
+	} else {
+		ds.Namespace = f.Namespace.Name
+		_, err = f.ClientSet.AppsV1().DaemonSets(f.Namespace.Name).Create(ctx, ds, metav1.CreateOptions{})
+		framework.ExpectNoError(err, "failed to create nvidia-driver-installer daemonset")
+		framework.Logf("Successfully created daemonset to install Nvidia drivers.")
 	}
 
-	ds.Namespace = f.Namespace.Name
+	data, err := e2etestfiles.Read("test/e2e/testing-manifests/gpu/gce/nvidia-gpu-device-plugin.yaml")
-	_, err = f.ClientSet.AppsV1().DaemonSets(f.Namespace.Name).Create(ctx, ds, metav1.CreateOptions{})
+	framework.ExpectNoError(err, "failed to read local manifest for nvidia-gpu-device-plugin daemonset")
-	framework.ExpectNoError(err, "failed to create nvidia-driver-installer daemonset")
+	ds, err = e2emanifest.DaemonSetFromData(data)
-	framework.Logf("Successfully created daemonset to install Nvidia drivers.")
+	framework.ExpectNoError(err, "failed to parse local manifest for nvidia-gpu-device-plugin daemonset")
+
+	prev, err = f.ClientSet.AppsV1().DaemonSets(ds.Namespace).Get(ctx, ds.Name, metav1.GetOptions{})
+	if err == nil && prev != nil {
+		framework.Logf("nvidia-gpu-device-plugin Daemonset already installed, skipping...")
+	} else {
+		_, err = f.ClientSet.AppsV1().DaemonSets(ds.Namespace).Create(ctx, ds, metav1.CreateOptions{})
+		framework.ExpectNoError(err, "failed to create nvidia-gpu-device-plugin daemonset")
+		framework.Logf("Successfully created daemonset to install Nvidia device plugin.")
+	}
 
 	waitForGPUs(ctx, f, ds.Namespace, ds.Name)
 }

test/e2e/testing-manifests/embed.go

@@ -22,7 +22,7 @@ import (
 	e2etestfiles "k8s.io/kubernetes/test/e2e/framework/testfiles"
 )
 
-//go:embed dra flexvolume guestbook kubectl sample-device-plugin scheduling/nvidia-driver-installer.yaml statefulset storage-csi
+//go:embed dra flexvolume guestbook kubectl sample-device-plugin gpu statefulset storage-csi
 var e2eTestingManifestsFS embed.FS
 
 func GetE2ETestingManifestsFS() e2etestfiles.EmbeddedFileSource {

test/e2e/testing-manifests/gpu/gce/nvidia-gpu-device-plugin.yaml

@@ -0,0 +1,57 @@
+apiVersion: apps/v1
+kind: DaemonSet
+metadata:
+  name: nvidia-gpu-device-plugin
+  namespace: kube-system
+  labels:
+    k8s-app: nvidia-gpu-device-plugin
+    addonmanager.kubernetes.io/mode: EnsureExists
+spec:
+  selector:
+    matchLabels:
+      k8s-app: nvidia-gpu-device-plugin
+  template:
+    metadata:
+      labels:
+        k8s-app: nvidia-gpu-device-plugin
+    spec:
+      priorityClassName: system-node-critical
+      affinity:
+        nodeAffinity:
+          requiredDuringSchedulingIgnoredDuringExecution:
+            nodeSelectorTerms:
+            - matchExpressions:
+              - key: cloud.google.com/gke-accelerator
+                operator: Exists
+      tolerations:
+      - operator: "Exists"
+        effect: "NoExecute"
+      - operator: "Exists"
+        effect: "NoSchedule"
+      volumes:
+      - name: device-plugin
+        hostPath:
+          path: /var/lib/kubelet/device-plugins
+      - name: dev
+        hostPath:
+          path: /dev
+      containers:
+      - image: "registry.k8s.io/nvidia-gpu-device-plugin@sha256:4b036e8844920336fa48f36edeb7d4398f426d6a934ba022848deed2edbf09aa"
+        command: ["/usr/bin/nvidia-gpu-device-plugin", "-logtostderr"]
+        name: nvidia-gpu-device-plugin
+        resources:
+          requests:
+            cpu: 50m
+            memory: 10Mi
+          limits:
+            cpu: 50m
+            memory: 10Mi
+        securityContext:
+          privileged: true
+        volumeMounts:
+        - name: device-plugin
+          mountPath: /device-plugin
+        - name: dev
+          mountPath: /dev
+  updateStrategy:
+    type: RollingUpdate