Serving hpa (#113)

* Added: basic hpa

* Changed: version bump

charts/clearml-serving/Chart.yaml

@@ -2,8 +2,9 @@ apiVersion: v2
 name: clearml-serving
 description: ClearML Serving Helm Chart
 type: application
-version: 0.4.1
+version: 0.5.0
 appVersion: "0.9.0"
+kubeVersion: ">= 1.19.0-0 < 1.26.0-0"
 maintainers:
   - name: valeriano-manassero
     url: https://github.com/valeriano-manassero

charts/clearml-serving/README.md

@@ -1,6 +1,6 @@
 # clearml-serving
 
-![Version: 0.4.1](https://img.shields.io/badge/Version-0.4.1-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: 0.9.0](https://img.shields.io/badge/AppVersion-0.9.0-informational?style=flat-square)
+![Version: 0.5.0](https://img.shields.io/badge/Version-0.5.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: 0.9.0](https://img.shields.io/badge/AppVersion-0.9.0-informational?style=flat-square)
 
 ClearML Serving Helm Chart
 
@@ -11,6 +11,10 @@ ClearML Serving Helm Chart
 | valeriano-manassero |  | <https://github.com/valeriano-manassero> |
 | stefano-cherchi |  | <https://github.com/stefano-cherchi> |
 
+## Requirements
+
+Kubernetes: `>= 1.19.0-0 < 1.26.0-0`
+
 ## Values
 
 | Key | Type | Default | Description |
@@ -28,6 +32,11 @@ ClearML Serving Helm Chart
 | clearml.servingTaskId | string | `"ClearML Serving Task ID"` |  |
 | clearml.webHost | string | `"http://clearml-server-webserver:80"` |  |
 | clearml_serving_inference.affinity | object | `{}` |  |
+| clearml_serving_inference.autoscaling.enabled | bool | `false` |  |
+| clearml_serving_inference.autoscaling.maxReplicas | int | `11` |  |
+| clearml_serving_inference.autoscaling.minReplicas | int | `1` |  |
+| clearml_serving_inference.autoscaling.targetCPU | int | `50` |  |
+| clearml_serving_inference.autoscaling.targetMemory | int | `50` |  |
 | clearml_serving_inference.extraPythonPackages | list | `[]` | Extra Python Packages to be installed in running pods |
 | clearml_serving_inference.image | string | `"allegroai/clearml-serving-inference"` |  |
 | clearml_serving_inference.nodeSelector | object | `{}` |  |
@@ -40,6 +49,11 @@ ClearML Serving Helm Chart
 | clearml_serving_statistics.resources | object | `{}` |  |
 | clearml_serving_statistics.tolerations | list | `[]` |  |
 | clearml_serving_triton.affinity | object | `{}` |  |
+| clearml_serving_triton.autoscaling.enabled | bool | `false` |  |
+| clearml_serving_triton.autoscaling.maxReplicas | int | `11` |  |
+| clearml_serving_triton.autoscaling.minReplicas | int | `1` |  |
+| clearml_serving_triton.autoscaling.targetCPU | int | `50` |  |
+| clearml_serving_triton.autoscaling.targetMemory | int | `50` |  |
 | clearml_serving_triton.enabled | bool | `true` |  |
 | clearml_serving_triton.extraPythonPackages | list | `[]` | Extra Python Packages to be installed in running pods |
 | clearml_serving_triton.image | string | `"allegroai/clearml-serving-triton"` |  |

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