GPU-Live/template/prd/gpu-operator/charts/node-feature-discovery/templates/topologyupdater.yaml

189 lines
7.2 KiB
YAML

{{- if .Values.topologyUpdater.enable -}}
apiVersion: apps/v1
kind: DaemonSet
metadata:
name: {{ include "node-feature-discovery.fullname" . }}-topology-updater
namespace: {{ include "node-feature-discovery.namespace" . }}
labels:
{{- include "node-feature-discovery.labels" . | nindent 4 }}
role: topology-updater
{{- with .Values.topologyUpdater.daemonsetAnnotations }}
annotations:
{{- toYaml . | nindent 4 }}
{{- end }}
spec:
revisionHistoryLimit: {{ .Values.topologyUpdater.revisionHistoryLimit }}
selector:
matchLabels:
{{- include "node-feature-discovery.selectorLabels" . | nindent 6 }}
role: topology-updater
template:
metadata:
labels:
{{- include "node-feature-discovery.selectorLabels" . | nindent 8 }}
role: topology-updater
annotations:
checksum/config: {{ include (print $.Template.BasePath "/nfd-topologyupdater-conf.yaml") . | sha256sum }}
{{- with .Values.topologyUpdater.annotations }}
{{- toYaml . | nindent 8 }}
{{- end }}
spec:
serviceAccountName: {{ include "node-feature-discovery.topologyUpdater.serviceAccountName" . }}
dnsPolicy: ClusterFirstWithHostNet
{{- with .Values.priorityClassName }}
priorityClassName: {{ . }}
{{- end }}
{{- with .Values.imagePullSecrets }}
imagePullSecrets:
{{- toYaml . | nindent 8 }}
{{- 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 }}"
imagePullPolicy: "{{ .Values.image.pullPolicy }}"
livenessProbe:
grpc:
port: {{ .Values.topologyUpdater.healthPort | default "8082" }}
{{- with .Values.topologyUpdater.livenessProbe.initialDelaySeconds }}
initialDelaySeconds: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.livenessProbe.failureThreshold }}
failureThreshold: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.livenessProbe.periodSeconds }}
periodSeconds: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.livenessProbe.timeoutSeconds }}
timeoutSeconds: {{ . }}
{{- end }}
readinessProbe:
grpc:
port: {{ .Values.topologyUpdater.healthPort | default "8082" }}
{{- with .Values.topologyUpdater.readinessProbe.initialDelaySeconds }}
initialDelaySeconds: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.readinessProbe.failureThreshold }}
failureThreshold: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.readinessProbe.periodSeconds }}
periodSeconds: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.readinessProbe.timeoutSeconds }}
timeoutSeconds: {{ . }}
{{- end }}
{{- with .Values.topologyUpdater.readinessProbe.successThreshold }}
successThreshold: {{ . }}
{{- end }}
env:
- name: NODE_NAME
valueFrom:
fieldRef:
fieldPath: spec.nodeName
- name: NODE_ADDRESS
valueFrom:
fieldRef:
fieldPath: status.hostIP
{{- with .Values.topologyUpdater.extraEnvs }}
{{- toYaml . | nindent 8 }}
{{- end}}
command:
- "nfd-topology-updater"
args:
- "-podresources-socket=/host-var/lib/kubelet-podresources/kubelet.sock"
{{- if .Values.topologyUpdater.updateInterval | empty | not }}
- "-sleep-interval={{ .Values.topologyUpdater.updateInterval }}"
{{- else }}
- "-sleep-interval=3s"
{{- end }}
{{- if .Values.topologyUpdater.watchNamespace | empty | not }}
- "-watch-namespace={{ .Values.topologyUpdater.watchNamespace }}"
{{- else }}
- "-watch-namespace=*"
{{- end }}
{{- if not .Values.topologyUpdater.podSetFingerprint }}
- "-pods-fingerprint=false"
{{- end }}
{{- if .Values.topologyUpdater.kubeletConfigPath | empty | not }}
- "-kubelet-config-uri=file:///host-var/kubelet-config"
{{- end }}
{{- if .Values.topologyUpdater.kubeletStateDir | empty }}
# Disable kubelet state tracking by giving an empty path
- "-kubelet-state-dir="
{{- end }}
- "-metrics={{ .Values.topologyUpdater.metricsPort | default "8081"}}"
- "-grpc-health={{ .Values.topologyUpdater.healthPort | default "8082" }}"
{{- with .Values.topologyUpdater.extraArgs }}
{{- toYaml . | nindent 10 }}
{{- end }}
ports:
- 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
mountPath: /host-var/kubelet-config
{{- end }}
- name: kubelet-podresources-sock
mountPath: /host-var/lib/kubelet-podresources/kubelet.sock
- name: host-sys
mountPath: /host-sys
{{- if .Values.topologyUpdater.kubeletStateDir | empty | not }}
- name: kubelet-state-files
mountPath: /host-var/lib/kubelet
readOnly: true
{{- end }}
- name: nfd-topology-updater-conf
mountPath: "/etc/kubernetes/node-feature-discovery"
readOnly: true
resources:
{{- toYaml .Values.topologyUpdater.resources | nindent 12 }}
securityContext:
{{- toYaml .Values.topologyUpdater.securityContext | nindent 12 }}
volumes:
- name: host-sys
hostPath:
path: "/sys"
{{- if .Values.topologyUpdater.kubeletConfigPath | empty | not }}
- name: kubelet-config
hostPath:
path: {{ .Values.topologyUpdater.kubeletConfigPath }}
{{- end }}
- name: kubelet-podresources-sock
hostPath:
{{- if .Values.topologyUpdater.kubeletPodResourcesSockPath | empty | not }}
path: {{ .Values.topologyUpdater.kubeletPodResourcesSockPath }}
{{- else }}
path: /var/lib/kubelet/pod-resources/kubelet.sock
{{- end }}
{{- if .Values.topologyUpdater.kubeletStateDir | empty | not }}
- name: kubelet-state-files
hostPath:
path: {{ .Values.topologyUpdater.kubeletStateDir }}
{{- end }}
- name: nfd-topology-updater-conf
configMap:
name: {{ include "node-feature-discovery.fullname" . }}-topology-updater-conf
items:
- key: nfd-topology-updater.conf
path: nfd-topology-updater.conf
{{- with .Values.topologyUpdater.nodeSelector }}
nodeSelector:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- with .Values.topologyUpdater.affinity }}
affinity:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- with .Values.topologyUpdater.tolerations }}
tolerations:
{{- toYaml . | nindent 8 }}
{{- end }}
{{- end }}