Bump NFD to v0.16.1

Co-authored-by: Tariq Ibrahim <[email protected]> Signed-off-by: Christopher Desiniotis <[email protected]> bump NFD to v0.16.1 Signed-off-by: Tariq Ibrahim <[email protected]>
NVIDIA · Jul 11, 2024 · 27fb4b7 · 27fb4b7
1 parent f6c121f
commit 27fb4b7
Show file tree

Hide file tree

Showing 19 changed files with 645 additions and 169 deletions.
diff --git a/deployments/gpu-operator/Chart.lock b/deployments/gpu-operator/Chart.lock
@@ -1,6 +1,6 @@
 dependencies:
 - name: node-feature-discovery
   repository: https://kubernetes-sigs.github.io/node-feature-discovery/charts
-  version: 0.15.4
-digest: sha256:025d011a09907e4b274e550b36689c47ed12722543624debd1e7d709d70462d8
-generated: "2024-04-05T09:13:06.247476-07:00"
+  version: 0.16.1
+digest: sha256:9fe8c3cb9355cbeb51046a82541a332fbc2a545c465cc9d5425b056ba96ec036
+generated: "2024-07-03T09:21:33.977756-07:00"
diff --git a/deployments/gpu-operator/Chart.yaml b/deployments/gpu-operator/Chart.yaml
@@ -19,6 +19,6 @@ keywords:
 
 dependencies:
   - name: node-feature-discovery
-    version: v0.15.4
+    version: v0.16.1
     repository: https://kubernetes-sigs.github.io/node-feature-discovery/charts
     condition: nfd.enabled
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/Chart.yaml b/deployments/gpu-operator/charts/node-feature-discovery/Chart.yaml
@@ -1,5 +1,5 @@
 apiVersion: v2
-appVersion: v0.15.4
+appVersion: v0.16.1
 description: 'Detects hardware features available on each node in a Kubernetes cluster,
   and advertises those features using node labels. '
 home: https://github.com/kubernetes-sigs/node-feature-discovery
@@ -11,4 +11,4 @@ name: node-feature-discovery
 sources:
 - https://github.com/kubernetes-sigs/node-feature-discovery
 type: application
-version: 0.15.4
+version: 0.16.1
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/README.md b/deployments/gpu-operator/charts/node-feature-discovery/README.md
@@ -6,5 +6,5 @@ labels. NFD provides flexible configuration and extension points for a wide
 range of vendor and application specific node labeling needs.
 
 See
-[NFD documentation](https://kubernetes-sigs.github.io/node-feature-discovery/v0.15/deployment/helm.html)
+[NFD documentation](https://kubernetes-sigs.github.io/node-feature-discovery/v0.16/deployment/helm.html)
 for deployment instructions.
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/crds/nfd-api-crds.yaml b/deployments/gpu-operator/charts/node-feature-discovery/crds/nfd-api-crds.yaml
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/cert-manager-certs.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/cert-manager-certs.yaml
@@ -19,8 +19,12 @@ spec:
   - {{ include "node-feature-discovery.fullname" . }}-master.{{ include "node-feature-discovery.namespace" .  }}.svc
   - {{ include "node-feature-discovery.fullname" . }}-master.{{ include "node-feature-discovery.namespace" .  }}.svc.cluster.local
   issuerRef:
-    name: nfd-ca-issuer
+    name: {{ default "nfd-ca-issuer" .Values.tls.certManagerCertificate.issuerName }}
+    {{- if and .Values.tls.certManagerCertificate.issuerName .Values.tls.certManagerCertificate.issuerKind }}
+    kind: {{ .Values.tls.certManagerCertificate.issuerKind }}
+    {{- else }}
     kind: Issuer
+    {{- end }}
     group: cert-manager.io
 {{- end }}
 ---
@@ -39,8 +43,12 @@ spec:
   dnsNames:
   - {{ include "node-feature-discovery.fullname" . }}-worker.{{ include "node-feature-discovery.namespace" .  }}.svc.cluster.local
   issuerRef:
-    name: nfd-ca-issuer
+    name: {{ default "nfd-ca-issuer" .Values.tls.certManagerCertificate.issuerName }}
+    {{- if and .Values.tls.certManagerCertificate.issuerName .Values.tls.certManagerCertificate.issuerKind }}
+    kind: {{ .Values.tls.certManagerCertificate.issuerKind }}
+    {{- else }}
     kind: Issuer
+    {{- end }}
     group: cert-manager.io
 {{- end }}
 
@@ -60,8 +68,12 @@ spec:
   dnsNames:
   - {{ include "node-feature-discovery.fullname" . }}-topology-updater.{{ include "node-feature-discovery.namespace" .  }}.svc.cluster.local
   issuerRef:
-    name: nfd-ca-issuer
+    name: {{ default "nfd-ca-issuer" .Values.tls.certManagerCertificate.issuerName }}
+    {{- if and .Values.tls.certManagerCertificate.issuerName .Values.tls.certManagerCertificate.issuerKind }}
+    kind: {{ .Values.tls.certManagerCertificate.issuerKind }}
+    {{- else }}
     kind: Issuer
+    {{- end }}
     group: cert-manager.io
 {{- end }}
 

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/cert-manager-issuer.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/cert-manager-issuer.yaml
@@ -1,4 +1,4 @@
-{{- if .Values.tls.certManager }}
+{{- if and .Values.tls.certManager (not .Values.tls.certManagerCertificate.issuerName ) }}
 # See https://cert-manager.io/docs/configuration/selfsigned/#bootstrapping-ca-issuers
 # - Create a self signed issuer
 # - Use this to create a CA cert

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/clusterrole.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/clusterrole.yaml
@@ -21,10 +21,18 @@ rules:
   resources:
   - nodefeatures
   - nodefeaturerules
+  - nodefeaturegroups
   verbs:
   - get
   - list
   - watch
+- apiGroups:
+  - nfd.k8s-sigs.io
+  resources:
+  - nodefeaturegroups/status
+  verbs:
+  - patch
+  - update
 - apiGroups:
   - coordination.k8s.io
   resources:
@@ -58,6 +66,12 @@ rules:
   verbs:
   - get
   - list
+- apiGroups:
+  - ""
+  resources:
+  - namespaces
+  verbs:
+  - get
 - apiGroups:
     - ""
   resources:
@@ -80,7 +94,7 @@ rules:
   - update
 {{- end }}
 
-{{- if and .Values.gc.enable .Values.gc.rbac.create (or .Values.enableNodeFeatureApi .Values.topologyUpdater.enable) }}
+{{- if and .Values.gc.enable .Values.gc.rbac.create (or (and .Values.featureGates.NodeFeatureAPI .Values.enableNodeFeatureApi) .Values.topologyUpdater.enable) }}
 ---
 apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRole

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/clusterrolebinding.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/clusterrolebinding.yaml
@@ -33,7 +33,7 @@ subjects:
   namespace: {{ include "node-feature-discovery.namespace" .  }}
 {{- end }}
 
-{{- if and .Values.gc.enable .Values.gc.rbac.create (or .Values.enableNodeFeatureApi .Values.topologyUpdater.enable) }}
+{{- if and .Values.gc.enable .Values.gc.rbac.create (or (and .Values.featureGates.NodeFeatureAPI .Values.enableNodeFeatureApi) .Values.topologyUpdater.enable) }}
 ---
 apiVersion: rbac.authorization.k8s.io/v1
 kind: ClusterRoleBinding

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/master.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/master.yaml
@@ -27,6 +27,9 @@ spec:
         {{- toYaml . | nindent 8 }}
       {{- end }}
     spec:
+    {{- with .Values.priorityClassName }}
+      priorityClassName: {{ . }}
+    {{- end }}
     {{- with .Values.imagePullSecrets }}
       imagePullSecrets:
         {{- toYaml . | nindent 8 }}
@@ -43,12 +46,12 @@ spec:
           imagePullPolicy: {{ .Values.image.pullPolicy }}
           livenessProbe:
             grpc:
-              port: 8080
+              port: 8082
             initialDelaySeconds: 10
             periodSeconds: 10
           readinessProbe:
             grpc:
-              port: 8080
+              port: 8082
             initialDelaySeconds: 5
             periodSeconds: 10
             failureThreshold: 10
@@ -70,9 +73,8 @@ spec:
             {{- if .Values.master.instance | empty | not }}
             - "-instance={{ .Values.master.instance }}"
             {{- end }}
-            {{- if not .Values.enableNodeFeatureApi }}
+            {{- if not (and .Values.featureGates.NodeFeatureAPI .Values.enableNodeFeatureApi) }}
             - "-port={{ .Values.master.port | default "8080" }}"
-            - "-enable-nodefeature-api=false"
             {{- else if gt (int .Values.master.replicaCount) 1 }}
             - "-enable-leader-election"
             {{- end }}
@@ -108,6 +110,10 @@ spec:
             - "-key-file=/etc/kubernetes/node-feature-discovery/certs/tls.key"
             - "-cert-file=/etc/kubernetes/node-feature-discovery/certs/tls.crt"
             {{- end }}
+            # Go over featureGates and add the feature-gate flag
+            {{- range $key, $value := .Values.featureGates }}
+            - "-feature-gates={{ $key }}={{ $value }}"
+            {{- end }}
             - "-metrics={{ .Values.master.metricsPort  | default "8081" }}"
           volumeMounts:
             {{- if .Values.tls.enable }}

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/nfd-gc.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/nfd-gc.yaml
@@ -1,4 +1,4 @@
-{{- if and .Values.gc.enable (or .Values.enableNodeFeatureApi .Values.topologyUpdater.enable) -}}
+{{- if and .Values.gc.enable (or (and .Values.featureGates.NodeFeatureAPI .Values.enableNodeFeatureApi) .Values.topologyUpdater.enable) -}}
 apiVersion: apps/v1
 kind: Deployment
 metadata:
@@ -29,6 +29,9 @@ spec:
     spec:
       serviceAccountName: {{ include "node-feature-discovery.gc.serviceAccountName" . }}
       dnsPolicy: ClusterFirstWithHostNet
+    {{- with .Values.priorityClassName }}
+      priorityClassName: {{ . }}
+    {{- end }}
     {{- with .Values.imagePullSecrets }}
       imagePullSecrets:
         {{- toYaml . | nindent 8 }}

diff --git a/...yments/gpu-operator/charts/node-feature-discovery/templates/nfd-topologyupdater-conf.yaml b/...yments/gpu-operator/charts/node-feature-discovery/templates/nfd-topologyupdater-conf.yaml
@@ -1,3 +1,4 @@
+{{- if .Values.topologyUpdater.enable -}}
 apiVersion: v1
 kind: ConfigMap
 metadata:
@@ -8,3 +9,4 @@ metadata:
 data:
   nfd-topology-updater.conf: |-
     {{- .Values.topologyUpdater.config | toYaml | nindent 4 }}
+{{- end }}
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/post-delete-job.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/post-delete-job.yaml
@@ -0,0 +1,94 @@
+apiVersion: v1
+kind: ServiceAccount
+metadata:
+  name: {{ include "node-feature-discovery.fullname" . }}-prune
+  namespace: {{ include "node-feature-discovery.namespace" . }}
+  labels:
+    {{- include "node-feature-discovery.labels" . | nindent 4 }}
+  annotations:
+    "helm.sh/hook": post-delete
+    "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRole
+metadata:
+  name: {{ include "node-feature-discovery.fullname" . }}-prune
+  labels:
+    {{- include "node-feature-discovery.labels" . | nindent 4 }}
+  annotations:
+    "helm.sh/hook": post-delete
+    "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
+rules:
+- apiGroups:
+  - ""
+  resources:
+  - nodes
+  - nodes/status
+  verbs:
+  - get
+  - patch
+  - update
+  - list
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: ClusterRoleBinding
+metadata:
+  name: {{ include "node-feature-discovery.fullname" . }}-prune
+  labels:
+    {{- include "node-feature-discovery.labels" . | nindent 4 }}
+  annotations:
+    "helm.sh/hook": post-delete
+    "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: ClusterRole
+  name: {{ include "node-feature-discovery.fullname" . }}-prune
+subjects:
+- kind: ServiceAccount
+  name: {{ include "node-feature-discovery.fullname" . }}-prune
+  namespace: {{ include "node-feature-discovery.namespace" .  }}
+---
+apiVersion: batch/v1
+kind: Job
+metadata:
+  name:  {{ include "node-feature-discovery.fullname" . }}-prune
+  namespace: {{ include "node-feature-discovery.namespace" . }}
+  labels:
+    {{- include "node-feature-discovery.labels" . | nindent 4 }}
+  annotations:
+    "helm.sh/hook": post-delete
+    "helm.sh/hook-delete-policy": before-hook-creation,hook-succeeded
+spec:
+  template:
+    metadata:
+      labels:
+        {{- include "node-feature-discovery.labels" . | nindent 8 }}
+        role: prune
+    spec:
+      serviceAccountName: {{ include "node-feature-discovery.fullname" . }}-prune
+      containers:
+        - name: nfd-master
+          securityContext:
+            {{- toYaml .Values.master.securityContext | nindent 12 }}
+          image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
+          imagePullPolicy: {{ .Values.image.pullPolicy }}
+          command:
+            - "nfd-master"
+          args:
+            - "-prune"
+            {{- if .Values.master.instance | empty | not }}
+            - "-instance={{ .Values.master.instance }}"
+            {{- end }}
+      restartPolicy: Never
+      {{- with .Values.master.nodeSelector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.master.affinity }}
+      affinity:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- with .Values.master.tolerations }}
+      tolerations:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/prometheus.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/prometheus.yaml
@@ -12,7 +12,7 @@ metadata:
 spec:
   podMetricsEndpoints:
     - honorLabels: true
-      interval: 10s
+      interval: {{ .Values.prometheus.scrapeInterval }}
       path: /metrics
       port: metrics
       scheme: http

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/service.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/service.yaml
@@ -1,4 +1,4 @@
-{{- if and (not .Values.enableNodeFeatureApi) .Values.master.enable }}
+{{- if and (not (and .Values.featureGates.NodeFeatureAPI .Values.enableNodeFeatureApi)) .Values.master.enable }}
 apiVersion: v1
 kind: Service
 metadata:

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/serviceaccount.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/serviceaccount.yaml
@@ -27,7 +27,7 @@ metadata:
   {{- end }}
 {{- end }}
 
-{{- if and .Values.gc.enable .Values.gc.serviceAccount.create (or .Values.enableNodeFeatureApi .Values.topologyUpdater.enable) }}
+{{- if and .Values.gc.enable .Values.gc.serviceAccount.create (or (and .Values.featureGates.NodeFeatureAPI .Values.enableNodeFeatureApi) .Values.topologyUpdater.enable) }}
 ---
 apiVersion: v1
 kind: ServiceAccount

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/topologyupdater.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/topologyupdater.yaml
@@ -28,6 +28,9 @@ spec:
     spec:
       serviceAccountName: {{ include "node-feature-discovery.topologyUpdater.serviceAccountName" . }}
       dnsPolicy: ClusterFirstWithHostNet
+    {{- with .Values.priorityClassName }}
+      priorityClassName: {{ . }}
+    {{- end }}
     {{- with .Values.imagePullSecrets }}
       imagePullSecrets:
         {{- toYaml . | nindent 8 }}
@@ -38,6 +41,17 @@ spec:
       - name: topology-updater
         image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
         imagePullPolicy: "{{ .Values.image.pullPolicy }}"
+        livenessProbe:
+          grpc:
+            port: 8082
+          initialDelaySeconds: 10
+          periodSeconds: 10
+        readinessProbe:
+          grpc:
+            port: 8082
+          initialDelaySeconds: 5
+          periodSeconds: 10
+          failureThreshold: 10
         env:
         - name: NODE_NAME
           valueFrom:
@@ -66,8 +80,8 @@ spec:
           - "-key-file=/etc/kubernetes/node-feature-discovery/certs/tls.key"
           - "-cert-file=/etc/kubernetes/node-feature-discovery/certs/tls.crt"
           {{- end }}
-          {{- if .Values.topologyUpdater.podSetFingerprint }}
-          - "-pods-fingerprint"
+          {{- if not .Values.topologyUpdater.podSetFingerprint }}
+          - "-pods-fingerprint=false"
           {{- end }}
           {{- if .Values.topologyUpdater.kubeletConfigPath | empty | not }}
           - "-kubelet-config-uri=file:///host-var/kubelet-config"