Added: basic hpa

charts/clearml-serving/templates/_helpers.tpl

@@ -60,3 +60,33 @@ Create the name of the service account to use
 {{- default "default" .Values.serviceAccount.name }}
 {{- end }}
 {{- end }}
+
+{{/*
+Return the target Kubernetes version
+*/}}
+{{- define "common.capabilities.kubeVersion" -}}
+{{- if .Values.global }}
+    {{- if .Values.global.kubeVersion }}
+    {{- .Values.global.kubeVersion -}}
+    {{- else }}
+    {{- default .Capabilities.KubeVersion.Version .Values.kubeVersion -}}
+    {{- end -}}
+{{- else }}
+{{- default .Capabilities.KubeVersion.Version .Values.kubeVersion -}}
+{{- end -}}
+{{- end -}}
+
+{{/*
+Return the appropriate apiVersion for Horizontal Pod Autoscaler.
+*/}}
+{{- define "common.capabilities.hpa.apiVersion" -}}
+{{- if semverCompare "<1.23-0" (include "common.capabilities.kubeVersion" .context) -}}
+{{- if .beta2 -}}
+{{- print "autoscaling/v2beta2" -}}
+{{- else -}}
+{{- print "autoscaling/v2beta1" -}}
+{{- end -}}
+{{- else -}}
+{{- print "autoscaling/v2" -}}
+{{- end -}}
+{{- end -}}

charts/clearml-serving/templates/clearml-serving-inference-hpa.yaml

@@ -0,0 +1,42 @@
+{{- if .Values.clearml_serving_inference.autoscaling.enabled }}
+apiVersion: {{ include "common.capabilities.hpa.apiVersion" ( dict "context" $ ) }}
+kind: HorizontalPodAutoscaler
+metadata:
+  name: clearml-serving-inference-hpa
+  namespace: {{ .Release.Namespace | quote }}
+  annotations: {}
+  labels:
+    clearml.serving.service: clearml-serving-inference
+spec:
+  scaleTargetRef:
+    apiVersion: "apps/v1"
+    kind: Deployment
+    name: clearml-serving-inference
+  minReplicas: {{ .Values.clearml_serving_inference.autoscaling.minReplicas }}
+  maxReplicas: {{ .Values.clearml_serving_inference.autoscaling.maxReplicas }}
+  metrics:
+    {{- if .Values.clearml_serving_inference.autoscaling.targetCPU }}
+    - type: Resource
+      resource:
+        name: cpu
+        {{- if semverCompare "<1.23-0" (include "common.capabilities.kubeVersion" .) }}
+        targetAverageUtilization: {{ .Values.clearml_serving_inference.autoscaling.targetCPU }}
+        {{- else }}
+        target:
+          type: Utilization
+          averageUtilization: {{ .Values.clearml_serving_inference.autoscaling.targetCPU }}
+        {{- end }}
+    {{- end }}
+    {{- if .Values.clearml_serving_inference.autoscaling.targetMemory }}
+    - type: Resource
+      resource:
+        name: memory
+        {{- if semverCompare "<1.23-0" (include "common.capabilities.kubeVersion" .) }}
+        targetAverageUtilization: {{ .Values.clearml_serving_inference.autoscaling.targetMemory }}
+        {{- else }}
+        target:
+          type: Utilization
+          averageUtilization: {{ .Values.clearml_serving_inference.autoscaling.targetMemory }}
+        {{- end }}
+    {{- end }}
+{{- end }}

charts/clearml-serving/templates/clearml-serving-triton-hpa.yaml

@@ -0,0 +1,42 @@
+{{- if .Values.clearml_serving_triton.autoscaling.enabled }}
+apiVersion: {{ include "common.capabilities.hpa.apiVersion" ( dict "context" $ ) }}
+kind: HorizontalPodAutoscaler
+metadata:
+  name: clearml-serving-triton-hpa
+  namespace: {{ .Release.Namespace | quote }}
+  annotations: {}
+  labels:
+    clearml.serving.service: clearml-serving-triton
+spec:
+  scaleTargetRef:
+    apiVersion: "apps/v1"
+    kind: Deployment
+    name: clearml-serving-triton
+  minReplicas: {{ .Values.clearml_serving_triton.autoscaling.minReplicas }}
+  maxReplicas: {{ .Values.clearml_serving_triton.autoscaling.maxReplicas }}
+  metrics:
+    {{- if .Values.clearml_serving_triton.autoscaling.targetCPU }}
+    - type: Resource
+      resource:
+        name: cpu
+        {{- if semverCompare "<1.23-0" (include "common.capabilities.kubeVersion" .) }}
+        targetAverageUtilization: {{ .Values.clearml_serving_triton.autoscaling.targetCPU }}
+        {{- else }}
+        target:
+          type: Utilization
+          averageUtilization: {{ .Values.clearml_serving_triton.autoscaling.targetCPU }}
+        {{- end }}
+    {{- end }}
+    {{- if .Values.clearml_serving_triton.autoscaling.targetMemory }}
+    - type: Resource
+      resource:
+        name: memory
+        {{- if semverCompare "<1.23-0" (include "common.capabilities.kubeVersion" .) }}
+        targetAverageUtilization: {{ .Values.clearml_serving_triton.autoscaling.targetMemory }}
+        {{- else }}
+        target:
+          type: Utilization
+          averageUtilization: {{ .Values.clearml_serving_triton.autoscaling.targetMemory }}
+        {{- end }}
+    {{- end }}
+{{- end }}

charts/clearml-serving/values.yaml

@@ -65,6 +65,12 @@ clearml_serving_inference:
   extraPythonPackages: []
   #  - numpy==1.22.4
   #  - pandas==1.4.2
+  autoscaling:
+    enabled: false
+    minReplicas: 1
+    maxReplicas: 11
+    targetCPU: 50
+    targetMemory: 50
 
 clearml_serving_triton:
   enabled: true
@@ -77,3 +83,9 @@ clearml_serving_triton:
   extraPythonPackages: []
   #  - numpy==1.22.4
   #  - pandas==1.4.2
+  autoscaling:
+    enabled: false
+    minReplicas: 1
+    maxReplicas: 11
+    targetCPU: 50
+    targetMemory: 50