Add optional NFD dependencies for Helm deployment

If nfd.enabled is set to true, the daemonsets are only deployed to nodes
with AMD devices (as identified by label
feature.node.kubernetes.io/pci-0300_1002.present)

Bumping Helm chart version to 0.2.0

Author: y2kenny

helm/amd-gpu/Chart.yaml

@@ -15,5 +15,11 @@ keywords:
   - gpu
 
 kubeVersion: ">= 1.18.0"
-version: 0.1.0
+version: 0.2.0
 appVersion: "1.18.0"
+
+dependencies:
+- name: node-feature-discovery
+  version: "0.8.1"
+  repository: "https://kubernetes-sigs.github.io/node-feature-discovery/charts"
+  condition: nfd.enabled

helm/amd-gpu/README.md

@@ -1,13 +1,19 @@
 # AMD GPU Helm Chart
 
-![Version: 0.1.0](https://img.shields.io/badge/Version-0.1.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: 1.18.0](https://img.shields.io/badge/AppVersion-1.18.0-informational?style=flat-square)
+![Version: 0.2.0](https://img.shields.io/badge/Version-0.2.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: 1.18.0](https://img.shields.io/badge/AppVersion-1.18.0-informational?style=flat-square)
 
 A Helm chart for deploying Kubernetes AMD GPU device plugin
 
 ## Requirements
 
 Kubernetes: `>= 1.18.0`
 
+## Optional Dependencies
+
+| Repository | Name | Version |
+|------------|------|---------|
+| https://kubernetes-sigs.github.io/node-feature-discovery/charts | node-feature-discovery | 0.8.1 |
+
 ## Values
 
 | Key | Type | Default | Description |
@@ -19,6 +25,8 @@ Kubernetes: `>= 1.18.0`
 | lbl.image.repository | string | `"docker.io/rocm/k8s-device-plugin"` |  |
 | lbl.image.tag | string | `"labeller-latest"` |  |
 | namespace | string | `"kube-system"` |  |
+| nfd.enabled | bool | `false` |  |
+| node_selector."feature.node.kubernetes.io/pci-0300_1002.present" | string | `"true"` |  |
 | securityContext.allowPrivilegeEscalation | bool | `false` |  |
 | securityContext.capabilities.drop[0] | string | `"ALL"` |  |
 | tolerations[0].key | string | `"CriticalAddonsOnly"` |  |

helm/amd-gpu/templates/deviceplugin-daemonset.yaml

@@ -18,6 +18,12 @@ spec:
       imagePullSecrets:
         {{- toYaml . | nindent 8 }}
       {{- end }}
+      {{- if .Values.nfd.enabled }}
+      {{- with .Values.node_selector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- end }}
       {{- with .Values.tolerations }}
       tolerations:
         {{- toYaml . | nindent 8 }}

helm/amd-gpu/templates/labeller.yaml

@@ -41,9 +41,16 @@ spec:
       labels:
         name: amdgpu-lr-ds
     spec:
+      {{- if .Values.nfd.enabled }}
+      {{- with .Values.node_selector }}
+      nodeSelector:
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
+      {{- end }}
+      {{- with .Values.tolerations }}
       tolerations:
-      - key: CriticalAddonsOnly
-        operator: Exists
+        {{- toYaml . | nindent 8 }}
+      {{- end }}
       containers:
       - image: {{ .Values.lbl.image.repository }}:{{ .Values.lbl.image.tag }}
         name: {{ .Chart.Name }}-lr-cntr

helm/amd-gpu/values.yaml

@@ -1,3 +1,6 @@
+nfd:
+  enabled: false
+
 labeller:
   enabled: false
 
@@ -25,3 +28,6 @@ securityContext:
 tolerations:
   - key: CriticalAddonsOnly
     operator: Exists
+
+node_selector:
+  feature.node.kubernetes.io/pci-0300_1002.present: "true"