Merge pull request #1044 from NVIDIA/bump-nfd

bump node-feature-discovery to v0.16.5
NVIDIA · Oct 15, 2024 · be0cc17 · be0cc17
2 parents 9e638b4 + e967505
commit be0cc17
Show file tree

Hide file tree

Showing 8 changed files with 46 additions and 10 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: 0.16.4
-digest: sha256:527fff3327048f6c749df93d105eddf3011ff1be4090e13c6455aedf472d259d
-generated: "2024-10-04T13:43:35.810538-07:00"
+  version: 0.16.5
+digest: sha256:7362fbdd7184c4855630a0bc84b9bda62742f7d59c0874ea069aedca8eb802e1
+generated: "2024-10-15T08:41:36.078658-07:00"
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: v0.16.4
+    version: v0.16.5
     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: v0.16.4
+appVersion: v0.16.5
 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: 0.16.4
+version: 0.16.5
diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/master.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/master.yaml
@@ -39,6 +39,7 @@ spec:
       enableServiceLinks: false
       securityContext:
         {{- toYaml .Values.master.podSecurityContext | nindent 8 }}
+      hostNetwork: {{ .Values.master.hostNetwork }}
       containers:
         - name: master
           securityContext:
@@ -54,11 +55,16 @@ spec:
             name: grpc
           - containerPort: {{ .Values.master.metricsPort | default "8081" }}
             name: metrics
+          - containerPort: {{ .Values.master.healthPort  | default "8082" }}
+            name: health
           env:
           - name: NODE_NAME
             valueFrom:
               fieldRef:
                 fieldPath: spec.nodeName
+        {{- with .Values.master.extraEnvs }}
+          {{- toYaml . | nindent 8 }}
+        {{- end}}
           command:
             - "nfd-master"
           resources:
@@ -109,6 +115,7 @@ spec:
             - "-feature-gates={{ $key }}={{ $value }}"
             {{- end }}
             - "-metrics={{ .Values.master.metricsPort  | default "8081" }}"
+            - "-grpc-health={{ .Values.master.healthPort  | default "8082" }}"
           volumeMounts:
             {{- if .Values.tls.enable }}
             - name: nfd-master-cert

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
@@ -39,6 +39,7 @@ spec:
     {{- end }}
       securityContext:
         {{- toYaml .Values.gc.podSecurityContext | nindent 8 }}
+      hostNetwork: {{ .Values.gc.hostNetwork }}
       containers:
       - name: gc
         image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
@@ -48,6 +49,9 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: spec.nodeName
+      {{- with .Values.gc.extraEnvs }}
+        {{- toYaml . | nindent 8 }}
+      {{- end}}
         command:
           - "nfd-gc"
         args:

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/topologyupdater.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/topologyupdater.yaml
@@ -38,6 +38,7 @@ spec:
     {{- end }}
       securityContext:
         {{- toYaml .Values.topologyUpdater.podSecurityContext | nindent 8 }}
+      hostNetwork: {{ .Values.topologyUpdater.hostNetwork }}
       containers:
       - name: topology-updater
         image: "{{ .Values.image.repository }}:{{ .Values.image.tag | default .Chart.AppVersion }}"
@@ -55,6 +56,9 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: status.hostIP
+      {{- with .Values.topologyUpdater.extraEnvs }}
+        {{- toYaml . | nindent 8 }}
+      {{- end}}
         command:
           - "nfd-topology-updater"
         args:
@@ -85,9 +89,12 @@ spec:
           - "-kubelet-state-dir="
           {{- end }}
           - -metrics={{ .Values.topologyUpdater.metricsPort | default "8081"}}
+          - "-grpc-health={{ .Values.topologyUpdater.healthPort  | default "8082" }}"
         ports:
-          - name: metrics
-            containerPort: {{ .Values.topologyUpdater.metricsPort | default "8081"}}
+          - containerPort: {{ .Values.topologyUpdater.metricsPort | default "8081"}}
+            name: metrics
+          - containerPort: {{ .Values.topologyUpdater.healthPort  | default "8082" }}
+            name: health
         volumeMounts:
         {{- if .Values.topologyUpdater.kubeletConfigPath | empty | not }}
         - name: kubelet-config

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/templates/worker.yaml b/deployments/gpu-operator/charts/node-feature-discovery/templates/worker.yaml
@@ -38,6 +38,7 @@ spec:
       serviceAccountName: {{ include "node-feature-discovery.worker.serviceAccountName" . }}
       securityContext:
         {{- toYaml .Values.worker.podSecurityContext | nindent 8 }}
+      hostNetwork: {{ .Values.worker.hostNetwork }}
       containers:
       - name: worker
         securityContext:
@@ -61,6 +62,9 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.uid
+      {{- with .Values.worker.extraEnvs }}
+        {{- toYaml . | nindent 8 }}
+      {{- end}}
         resources:
         {{- toYaml .Values.worker.resources | nindent 12 }}
         command:
@@ -79,9 +83,12 @@ spec:
         - "-feature-gates={{ $key }}={{ $value }}"
 {{- end }}
         - "-metrics={{ .Values.worker.metricsPort | default "8081"}}"
+        - "-grpc-health={{ .Values.worker.healthPort  | default "8082" }}"
         ports:
-          - name: metrics
-            containerPort: {{ .Values.worker.metricsPort | default "8081"}}
+          - containerPort: {{ .Values.worker.metricsPort | default "8081"}}
+            name: metrics
+          - containerPort: {{ .Values.worker.healthPort  | default "8082" }}
+            name: health
         volumeMounts:
         - name: host-boot
           mountPath: "/host-boot"

diff --git a/deployments/gpu-operator/charts/node-feature-discovery/values.yaml b/deployments/gpu-operator/charts/node-feature-discovery/values.yaml
@@ -20,6 +20,8 @@ priorityClassName: ""
 
 master:
   enable: true
+  extraEnvs: []
+  hostNetwork: false
   config: ### <NFD-MASTER-CONF-START-DO-NOT-REMOVE>
     # noPublish: false
     # autoDefaultNs: true
@@ -57,6 +59,7 @@ master:
   # be removed with it in a future release
   port: 8080
   metricsPort: 8081
+  healthPort: 8082
   instance:
   featureApi:
   resyncPeriod:
@@ -157,6 +160,8 @@ master:
 
 worker:
   enable: true
+  extraEnvs: []
+  hostNetwork: false
   config: ### <NFD-WORKER-CONF-START-DO-NOT-REMOVE>
     #core:
     #  labelWhiteList:
@@ -402,6 +407,7 @@ worker:
 ### <NFD-WORKER-CONF-END-DO-NOT-REMOVE>
 
   metricsPort: 8081
+  healthPort: 8082
   daemonsetAnnotations: {}
   podSecurityContext: {}
     # fsGroup: 2000
@@ -477,6 +483,8 @@ topologyUpdater:
 
   enable: false
   createCRDs: false
+  extraEnvs: []
+  hostNetwork: false
 
   serviceAccount:
     create: true
@@ -490,6 +498,7 @@ topologyUpdater:
     create: true
 
   metricsPort: 8081
+  healthPort: 8082
   kubeletConfigPath:
   kubeletPodResourcesSockPath:
   updateInterval: 60s
@@ -533,6 +542,8 @@ topologyUpdater:
 
 gc:
   enable: true
+  extraEnvs: []
+  hostNetwork: false
   replicaCount: 1
 
   serviceAccount: