diff --git a/kubernetes/helm/templates/ollama-statefulset.yaml b/kubernetes/helm/templates/ollama-statefulset.yaml index 19204347a..393787163 100644 --- a/kubernetes/helm/templates/ollama-statefulset.yaml +++ b/kubernetes/helm/templates/ollama-statefulset.yaml @@ -17,7 +17,10 @@ spec: spec: containers: - name: {{ include "ollama.name" . }} - image: {{ .Values.ollama.image }} + {{- with .Values.ollama.image }} + image: {{ .repository }}:{{ .tag }} + imagePullPolicy: {{ .pullPolicy }} + {{- end }} ports: - name: http containerPort: {{ .Values.ollama.service.containerPort }} @@ -29,7 +32,7 @@ spec: value: /usr/local/nvidia/lib:/usr/local/nvidia/lib64 - name: NVIDIA_DRIVER_CAPABILITIES value: compute,utility - {{- end}} + {{- end }} {{- with .Values.ollama.resources }} resources: {{- toYaml . | nindent 10 }} {{- end }} @@ -41,12 +44,10 @@ spec: nodeSelector: {{- toYaml . | nindent 8 }} {{- end }} + {{- with .Values.ollama.tolerations }} tolerations: - {{- if .Values.ollama.gpu.enabled }} - - key: nvidia.com/gpu - operator: Exists - effect: NoSchedule - {{- end }} + {{- toYaml . | nindent 8 }} + {{- end }} volumeClaimTemplates: - metadata: name: ollama-volume @@ -54,4 +55,4 @@ spec: accessModes: [ "ReadWriteOnce" ] resources: requests: - storage: {{ .Values.ollama.volumeSize }} + storage: {{ .Values.ollama.persistence.size }} diff --git a/kubernetes/helm/templates/webui-deployment.yaml b/kubernetes/helm/templates/webui-deployment.yaml index 3e62ac9f2..9250d315c 100644 --- a/kubernetes/helm/templates/webui-deployment.yaml +++ b/kubernetes/helm/templates/webui-deployment.yaml @@ -16,7 +16,10 @@ spec: spec: containers: - name: {{ .Chart.Name }} - image: {{ .Values.webui.image }} + {{- with .Values.webui.image }} + image: {{ .repository }}:{{ .tag }} + imagePullPolicy: {{ .pullPolicy }} + {{- end }} ports: - name: http containerPort: {{ .Values.webui.service.containerPort }} diff --git a/kubernetes/helm/values.yaml b/kubernetes/helm/values.yaml index 3e9f6a958..dde2db17d 100644 --- a/kubernetes/helm/values.yaml +++ b/kubernetes/helm/values.yaml @@ -1,6 +1,9 @@ ollama: replicaCount: 1 - image: ollama/ollama:latest + image: + repository: ollama/ollama + tag: latest + pullPolicy: Always resources: requests: cpu: "2000m" @@ -9,9 +12,14 @@ ollama: cpu: "4000m" memory: "4Gi" nvidia.com/gpu: "0" - volumeSize: 30Gi + persistence: + enabled: true + size: 30Gi nodeSelector: {} - tolerations: [] + tolerations: + - key: nvidia.com/gpu + operator: Exists + effect: NoSchedule service: type: ClusterIP port: 80 @@ -21,7 +29,10 @@ ollama: webui: replicaCount: 1 - image: ghcr.io/open-webui/open-webui:main + image: + repository: ghcr.io/open-webui/open-webui + tag: main + pullPolicy: Always resources: requests: cpu: "500m"