ClearML standalone agent chart (#79)

* Added: agent chart * Changed: base image tag * Changed: updated helm-docs * Fixed: maintainers * Changed: updated radme * Fixed: http code to check * Fixed: default values to check * Changed: updated helm-docs * Added: default values to be substituted by GH action * Added: sed on the fly for testing * Changed: updated CI images for Kind
2025-04-17 01:31:13 +00:00 · 2022-06-08 10:01:33 +02:00 · 2022-06-08 10:01:33 +02:00 · 562815e97a
commit 562815e97a
parent e317610397
12 changed files with 467 additions and 5 deletions
--- a/.github/workflows/ci.yaml
+++ b/.github/workflows/ci.yaml
@ -21,19 +21,23 @@ jobs:
    strategy:
      matrix:
        k8s:
-          - v1.21.2
-          - v1.22.1
-          - v1.23.0
+          - v1.22.7
+          - v1.23.6
+          - v1.24.0
    steps:
      - name: Checkout
        uses: actions/checkout@v1
      - name: Create kind ${{ matrix.k8s }} cluster
-        uses: helm/kind-action@v1.1.0
+        uses: helm/kind-action@v1.2.0
        with:
-          version: v0.11.1
+          version: v0.13.0
          node_image: kindest/node:${{ matrix.k8s }}
      - name: Set up chart-testing
        uses: helm/chart-testing-action@v2.2.1
+      - name: Inject secrets
+        run: |
+          find ./charts/*/ci/*.yaml -type f -exec sed -i "s/AGENTK8SGLUEKEY/${{ secrets.agentk8sglueKey }}/g" {} \;
+          find ./charts/*/ci/*.yaml -type f -exec sed -i "s/AGENTK8SGLUESECRET/${{ secrets.agentk8sglueSecret }}/g" {} \;
      - name: Run chart-testing (list-changed)
        id: list-changed
        run: |
--- a/charts/clearml-agent/.helmignore
+++ b/charts/clearml-agent/.helmignore
@ -0,0 +1,23 @@
+# Patterns to ignore when building packages.
+# This supports shell glob matching, relative path matching, and
+# negation (prefixed with !). Only one pattern per line.
+.DS_Store
+# Common VCS dirs
+.git/
+.gitignore
+.bzr/
+.bzrignore
+.hg/
+.hgignore
+.svn/
+# Common backup files
+*.swp
+*.bak
+*.tmp
+*.orig
+*~
+# Various IDEs
+.project
+.idea/
+*.tmproj
+.vscode/
--- a/charts/clearml-agent/Chart.yaml
+++ b/charts/clearml-agent/Chart.yaml
@ -0,0 +1,19 @@
+apiVersion: v2
+name: clearml-agent
+description: MLOps platform
+type: application
+version: "1.0.0"
+appVersion: "1.21"
+kubeVersion: ">= 1.19.0-0 < 1.25.0-0"
+home: https://clear.ml
+icon: https://raw.githubusercontent.com/allegroai/clearml/master/docs/clearml-logo.svg
+sources:
+  - https://github.com/allegroai/clearml-helm-charts
+  - https://github.com/allegroai/clearml
+maintainers:
+  - name: valeriano-manassero
+    url: https://github.com/valeriano-manassero
+keywords:
+  - clearml
+  - "machine learning"
+  - mlops
--- a/charts/clearml-agent/README.md
+++ b/charts/clearml-agent/README.md
@ -0,0 +1,57 @@
+# clearml-agent
+
+![Version: 1.0.0](https://img.shields.io/badge/Version-1.0.0-informational?style=flat-square) ![Type: application](https://img.shields.io/badge/Type-application-informational?style=flat-square) ![AppVersion: 1.21](https://img.shields.io/badge/AppVersion-1.21-informational?style=flat-square)
+
+MLOps platform
+
+**Homepage:** <https://clear.ml>
+
+## Maintainers
+
+| Name | Email | Url |
+| ---- | ------ | --- |
+| valeriano-manassero |  | <https://github.com/valeriano-manassero> |
+
+## Source Code
+
+* <https://github.com/allegroai/clearml-helm-charts>
+* <https://github.com/allegroai/clearml>
+
+## Requirements
+
+Kubernetes: `>= 1.19.0-0 < 1.25.0-0`
+
+## Values
+
+| Key | Type | Default | Description |
+|-----|------|---------|-------------|
+| agentk8sglue | object | `{"apiServerUrlReference":"https://api.clear.ml","defaultContainerImage":"ubuntu:18.04","fileServerUrlReference":"https://files.clear.ml","id":"k8s-agent","image":{"repository":"allegroai/clearml-agent-k8s","tag":"base-1.21"},"maxPods":10,"podTemplate":{"env":[],"nodeSelector":{},"resources":{},"tolerations":[],"volumes":[]},"queue":"default","replicaCount":1,"serviceAccountName":"default","webServerUrlReference":"https://app.clear.ml"}` | This agent will spawn queued experiments in new pods, a good use case is to combine this with GPU autoscaling nodes. https://github.com/allegroai/clearml-agent/tree/master/docker/k8s-glue |
+| agentk8sglue.apiServerUrlReference | string | `"https://api.clear.ml"` | Reference to Api server url |
+| agentk8sglue.defaultContainerImage | string | `"ubuntu:18.04"` | default container image for ClearML Task pod |
+| agentk8sglue.fileServerUrlReference | string | `"https://files.clear.ml"` | Reference to File server url |
+| agentk8sglue.id | string | `"k8s-agent"` | ClearML worker ID (must be unique across the entire ClearMLenvironment) |
+| agentk8sglue.image | object | `{"repository":"allegroai/clearml-agent-k8s","tag":"base-1.21"}` | Glue Agent image configuration |
+| agentk8sglue.maxPods | int | `10` | maximum concurrent consume ClearML Task pod |
+| agentk8sglue.podTemplate | object | `{"env":[],"nodeSelector":{},"resources":{},"tolerations":[],"volumes":[]}` | template for pods spawned to consume ClearML Task |
+| agentk8sglue.podTemplate.env | list | `[]` | environment variables for pods spawned to consume ClearML Task (example in values.yaml comments) |
+| agentk8sglue.podTemplate.nodeSelector | object | `{}` | nodeSelector setup for pods spawned to consume ClearML Task (example in values.yaml comments) |
+| agentk8sglue.podTemplate.resources | object | `{}` | resources declaration for pods spawned to consume ClearML Task (example in values.yaml comments) |
+| agentk8sglue.podTemplate.tolerations | list | `[]` | tolerations setup for pods spawned to consume ClearML Task (example in values.yaml comments) |
+| agentk8sglue.podTemplate.volumes | list | `[]` | volumes definition for pods spawned to consume ClearML Task (example in values.yaml comments) |
+| agentk8sglue.queue | string | `"default"` | ClearML queue this agent will consume |
+| agentk8sglue.replicaCount | int | `1` | Glue Agent number of pods |
+| agentk8sglue.serviceAccountName | string | `"default"` | serviceAccountName for pods spawned to consume ClearML Task |
+| agentk8sglue.webServerUrlReference | string | `"https://app.clear.ml"` | Reference to Web server url |
+| clearml | object | `{"agentk8sglueKey":"ACCESSKEY","agentk8sglueSecret":"SECRETKEY"}` | ClearMl generic configurations |
+| clearml.agentk8sglueKey | string | `"ACCESSKEY"` | Agent k8s Glue basic auth key |
+| clearml.agentk8sglueSecret | string | `"SECRETKEY"` | Agent k8s Glue basic auth secret |
+| imageCredentials | object | `{"email":"someone@host.com","enabled":false,"existingSecret":"","password":"pwd","registry":"docker.io","username":"someone"}` | Private image registry configuration |
+| imageCredentials.email | string | `"someone@host.com"` | Email |
+| imageCredentials.enabled | bool | `false` | Use private authentication mode |
+| imageCredentials.existingSecret | string | `""` | If this is set, chart will not generate a secret but will use what is defined here |
+| imageCredentials.password | string | `"pwd"` | Registry password |
+| imageCredentials.registry | string | `"docker.io"` | Registry name |
+| imageCredentials.username | string | `"someone"` | Registry username |
+
+----------------------------------------------
+Autogenerated from chart metadata using [helm-docs v1.10.0](https://github.com/norwoodj/helm-docs/releases/v1.10.0)
--- a/charts/clearml-agent/ci/default-values.yaml
+++ b/charts/clearml-agent/ci/default-values.yaml
@ -0,0 +1,3 @@
+clearml:
+  agentk8sglueKey: "AGENTK8SGLUEKEY"
+  agentk8sglueSecret: "AGENTK8SGLUESECRET"
--- a/charts/clearml-agent/templates/NOTES.txt
+++ b/charts/clearml-agent/templates/NOTES.txt
@ -0,0 +1 @@
+Glue Agent deployed.
--- a/charts/clearml-agent/templates/_helpers.tpl
+++ b/charts/clearml-agent/templates/_helpers.tpl
@ -0,0 +1,86 @@
+{{/*
+Expand the name of the chart.
+*/}}
+{{- define "clearml.name" -}}
+{{- default .Chart.Name .Values.nameOverride | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Create a default fully qualified app name.
+We truncate at 63 chars because some Kubernetes name fields are limited to this (by the DNS naming spec).
+If release name contains chart name it will be used as a full name.
+*/}}
+{{- define "clearml.fullname" -}}
+{{- if .Values.fullnameOverride }}
+{{- .Values.fullnameOverride | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- $name := default .Chart.Name .Values.nameOverride }}
+{{- if contains $name .Release.Name }}
+{{- .Release.Name | trunc 63 | trimSuffix "-" }}
+{{- else }}
+{{- printf "%s-%s" .Release.Name $name | trunc 63 | trimSuffix "-" }}
+{{- end }}
+{{- end }}
+{{- end }}
+
+{{/*
+Create chart name and version as used by the chart label.
+*/}}
+{{- define "clearml.chart" -}}
+{{- printf "%s-%s" .Chart.Name .Chart.Version | replace "+" "_" | trunc 63 | trimSuffix "-" }}
+{{- end }}
+
+{{/*
+Common labels
+*/}}
+{{- define "clearml.labels" -}}
+helm.sh/chart: {{ include "clearml.chart" . }}
+{{ include "clearml.selectorLabels" . }}
+{{- if .Chart.AppVersion }}
+app.kubernetes.io/version: {{ .Chart.AppVersion | quote }}
+{{- end }}
+app.kubernetes.io/managed-by: {{ .Release.Service }}
+{{- end }}
+
+{{/*
+Selector labels
+*/}}
+{{- define "clearml.selectorLabels" -}}
+app.kubernetes.io/name: {{ include "clearml.name" . }}
+app.kubernetes.io/instance: {{ .Release.Name }}
+{{- end }}
+
+{{/*
+Reference Name (agentk8sglue)
+*/}}
+{{- define "agentk8sglue.referenceName" -}}
+{{- include "clearml.name" . }}-agentk8sglue
+{{- end }}
+
+{{/*
+Selector labels (agentk8sglue)
+*/}}
+{{- define "agentk8sglue.selectorLabels" -}}
+app.kubernetes.io/name: {{ include "clearml.name" . }}
+app.kubernetes.io/instance: {{ include "agentk8sglue.referenceName" . }}
+{{- end }}
+
+{{/*
+Create the name of the service account to use
+*/}}
+{{- define "clearml.serviceAccountName" -}}
+{{- if .Values.serviceAccount.create }}
+{{- default (include "clearml.fullname" .) .Values.serviceAccount.name }}
+{{- else }}
+{{- default "default" .Values.serviceAccount.name }}
+{{- end }}
+{{- end }}
+
+{{/*
+Create secret to access docker registry
+*/}}
+{{- define "imagePullSecret" }}
+{{- with .Values.imageCredentials }}
+{{- printf "{\"auths\":{\"%s\":{\"username\":\"%s\",\"password\":\"%s\",\"email\":\"%s\",\"auth\":\"%s\"}}}" .registry .username .password .email (printf "%s:%s" .username .password | b64enc) | b64enc }}
+{{- end }}
+{{- end }}
--- a/charts/clearml-agent/templates/agentk8sglue-configmap.yaml
+++ b/charts/clearml-agent/templates/agentk8sglue-configmap.yaml
@ -0,0 +1,55 @@
+apiVersion: v1
+kind: ConfigMap
+metadata:
+  name: k8sagent-pod-template
+data:
+  template.yaml: |
+    apiVersion: v1
+    metadata:
+      namespace: {{ .Release.Namespace }}
+    spec:
+      serviceAccountName: {{ .Values.agentk8sglue.serviceAccountName }}
+      volumes:
+      {{- range .Values.agentk8sglue.podTemplate.volumes }}
+      - name: {{ .name }}
+        persistentVolumeClaim:
+          claimName: {{ .name }}
+      {{- end }}
+      containers:
+      - resources:
+          {{- toYaml .Values.agentk8sglue.podTemplate.resources | nindent 10 }}
+        ports:
+        - containerPort: 10022
+        volumeMounts:
+        {{- range .Values.agentk8sglue.podTemplate.volumes }}
+        - mountPath: {{ .path }}
+          name: {{ .name }}
+        {{- end }}
+        env:
+        - name: CLEARML_API_HOST
+          value: {{.Values.agentk8sglue.apiServerUrlReference}}
+        - name: CLEARML_WEB_HOST
+          value: {{.Values.agentk8sglue.webServerUrlReference}}
+        - name: CLEARML_FILES_HOST
+          value: {{.Values.agentk8sglue.fileServerUrlReference}}
+        - name: CLEARML_API_ACCESS_KEY
+          valueFrom:
+            secretKeyRef:
+              name: clearml-agent-conf
+              key: agentk8sglue_key
+        - name: CLEARML_API_SECRET_KEY
+          valueFrom:
+            secretKeyRef:
+              name: clearml-agent-conf
+              key: agentk8sglue_secret
+        {{- if .Values.agentk8sglue.podTemplate.env }}
+        {{ toYaml .Values.agentk8sglue.podTemplate.env | nindent 8 }}
+        {{- end }}
+        {{- with .Values.agentk8sglue.podTemplate.nodeSelector}}
+      nodeSelector:
+          {{- toYaml . | nindent 8 }}
+        {{- end }}
+        {{- with .Values.agentk8sglue.podTemplate.tolerations }}
+      tolerations:
+          {{- toYaml . | nindent 8 }}
+        {{- end }}
--- a/charts/clearml-agent/templates/agentk8sglue-deployment.yaml
+++ b/charts/clearml-agent/templates/agentk8sglue-deployment.yaml
@ -0,0 +1,91 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: {{ include "agentk8sglue.referenceName" . }}
+  labels:
+    {{- include "clearml.labels" . | nindent 4 }}
+spec:
+  replicas: {{ .Values.agentk8sglue.replicaCount }}
+  selector:
+    matchLabels:
+      {{- include "agentk8sglue.selectorLabels" . | nindent 6 }}
+  template:
+    metadata:
+      annotations:
+        checksum/config: {{ printf "%s" .Values.clearml | sha256sum }}
+      labels:
+        {{- include "agentk8sglue.selectorLabels" . | nindent 8 }}
+    spec:
+      {{- if .Values.imageCredentials.enabled }}
+      imagePullSecrets:
+      {{- if .Values.imageCredentials.existingSecret }}
+      - name: .Values.imageCredentials.existingSecret
+      {{- else }}
+      - name: clearml-agent-registry-key
+      {{- end }}
+      {{- end }}
+      initContainers:
+          - name: init-k8s-glue
+            image: "{{ .Values.agentk8sglue.image.repository }}:{{ .Values.agentk8sglue.image.tag }}"
+            command:
+              - /bin/sh
+              - -c
+              - >
+                set -x;
+                while [ $(curl -sw '%{http_code}' "{{.Values.agentk8sglue.apiServerUrlReference}}/debug.ping" -o /dev/null) -ne 200 ] ; do
+                  echo "waiting for apiserver" ;
+                  sleep 5 ;
+                done;
+                while [ $(curl -sw '%{http_code}' "{{.Values.agentk8sglue.fileServerUrlReference}}/" -o /dev/null) -ne 403 ] ; do
+                  echo "waiting for fileserver" ;
+                  sleep 5 ;
+                done;
+                while [ $(curl -sw '%{http_code}' "{{.Values.agentk8sglue.webServerUrlReference}}/" -o /dev/null) -ne 200 ] ; do
+                  echo "waiting for webserver" ;
+                  sleep 5 ;
+                done
+      containers:
+      - name: k8s-glue
+        image: "{{ .Values.agentk8sglue.image.repository }}:{{ .Values.agentk8sglue.image.tag }}"
+        imagePullPolicy: Always
+        command: ["/bin/bash", "-c", "export PATH=$PATH:$HOME/bin; source /root/.bashrc && /root/entrypoint.sh"]
+        volumeMounts:
+          - name: k8sagent-pod-template
+            mountPath: /root/template
+        env:
+          - name: CLEARML_API_HOST
+            value: "{{.Values.agentk8sglue.apiServerUrlReference}}"
+          - name: CLEARML_WEB_HOST
+            value: "{{.Values.agentk8sglue.webServerUrlReference}}"
+          - name: CLEARML_FILES_HOST
+            value: "{{.Values.agentk8sglue.fileServerUrlReference}}"
+          - name: K8S_GLUE_MAX_PODS
+            value: "{{.Values.agentk8sglue.maxPods}}"
+          - name: K8S_GLUE_QUEUE
+            value: "{{.Values.agentk8sglue.queue}}"
+          - name: K8S_GLUE_EXTRA_ARGS
+            value: "--namespace {{ .Release.Namespace }} --template-yaml /root/template/template.yaml"
+          - name: K8S_DEFAULT_NAMESPACE
+            value: "{{ .Release.Namespace }}"
+          - name: CLEARML_API_ACCESS_KEY
+            valueFrom:
+              secretKeyRef:
+                name: clearml-agent-conf
+                key: agentk8sglue_key
+          - name: CLEARML_API_SECRET_KEY
+            valueFrom:
+              secretKeyRef:
+                name: clearml-agent-conf
+                key: agentk8sglue_secret
+          - name: CLEARML_WORKER_ID
+            value: "{{.Values.agentk8sglue.id}}"
+          - name: CLEARML_AGENT_UPDATE_REPO
+            value: ""
+          - name: FORCE_CLEARML_AGENT_REPO
+            value: ""
+          - name: CLEARML_DOCKER_IMAGE
+            value: "{{.Values.agentk8sglue.defaultContainerImage}}"
+      volumes:
+        - name: k8sagent-pod-template
+          configMap: 
+            name: k8sagent-pod-template
--- a/charts/clearml-agent/templates/agentk8sglue-rbac.yaml
+++ b/charts/clearml-agent/templates/agentk8sglue-rbac.yaml
@ -0,0 +1,23 @@
+apiVersion: rbac.authorization.k8s.io/v1
+kind: Role
+metadata:
+  name: k8sagent-pods-access
+rules:
+  - apiGroups:
+      - ""
+    resources:
+      - pods
+    verbs: ["get", "list", "watch", "create", "patch", "delete"]
+---
+apiVersion: rbac.authorization.k8s.io/v1
+kind: RoleBinding
+metadata:
+  name: k8sagent-pods-access
+subjects:
+  - kind: ServiceAccount
+    name: default
+    namespace: {{ .Release.Namespace }}
+roleRef:
+  apiGroup: rbac.authorization.k8s.io
+  kind: Role
+  name: k8sagent-pods-access
--- a/charts/clearml-agent/templates/clearml-secrets.yaml
+++ b/charts/clearml-agent/templates/clearml-secrets.yaml
@ -0,0 +1,19 @@
+apiVersion: v1
+kind: Secret
+metadata:
+  name: clearml-agent-conf
+data:
+  agentk8sglue_key: {{ .Values.clearml.agentk8sglueKey | b64enc }}
+  agentk8sglue_secret: {{ .Values.clearml.agentk8sglueSecret | b64enc }}
+---
+{{- if .Values.imageCredentials.enabled }}
+{{- if not .Values.imageCredentials.existingSecret }}
+apiVersion: v1
+kind: Secret
+metadata:
+  name: clearml-agent-registry-key
+type: kubernetes.io/dockerconfigjson
+data:
+  .dockerconfigjson: {{ template "imagePullSecret" . }}
+{{- end }}
+{{- end }}
--- a/charts/clearml-agent/values.yaml
+++ b/charts/clearml-agent/values.yaml
@ -0,0 +1,81 @@
+# -- Private image registry configuration
+imageCredentials:
+  # -- Use private authentication mode
+  enabled: false
+  # -- If this is set, chart will not generate a secret but will use what is defined here
+  existingSecret: ""
+  # -- Registry name
+  registry: docker.io
+  # -- Registry username
+  username: someone
+  # -- Registry password
+  password: pwd
+  # -- Email
+  email: someone@host.com
+
+# -- ClearMl generic configurations
+clearml:
+  # -- Agent k8s Glue basic auth key
+  agentk8sglueKey: "ACCESSKEY"
+  # -- Agent k8s Glue basic auth secret
+  agentk8sglueSecret: "SECRETKEY"
+
+# -- This agent will spawn queued experiments in new pods, a good use case is to combine this with
+# GPU autoscaling nodes.
+# https://github.com/allegroai/clearml-agent/tree/master/docker/k8s-glue
+agentk8sglue:
+  # -- Glue Agent image configuration
+  image:
+    repository: "allegroai/clearml-agent-k8s"
+    tag: "base-1.21"
+
+  # -- Glue Agent number of pods
+  replicaCount: 1
+
+  # -- Reference to Api server url
+  apiServerUrlReference: "https://api.clear.ml"
+  # -- Reference to File server url
+  fileServerUrlReference: "https://files.clear.ml"
+  # -- Reference to Web server url
+  webServerUrlReference: "https://app.clear.ml"
+
+  # -- serviceAccountName for pods spawned to consume ClearML Task
+  serviceAccountName: default
+  # -- maximum concurrent consume ClearML Task pod
+  maxPods: 10
+  # -- default container image for ClearML Task pod
+  defaultContainerImage: ubuntu:18.04
+  # -- ClearML queue this agent will consume
+  queue: default
+
+  # -- ClearML worker ID (must be unique across the entire ClearMLenvironment)
+  id: k8s-agent
+
+  # -- template for pods spawned to consume ClearML Task
+  podTemplate:
+    # -- volumes definition for pods spawned to consume ClearML Task (example in values.yaml comments)
+    volumes: []
+      # - name: "yourvolume"
+      #   path: "/yourpath"
+    # -- environment variables for pods spawned to consume ClearML Task (example in values.yaml comments)
+    env: []
+      # # to setup access to private repo, setup secret with git credentials:
+      # - name: CLEARML_AGENT_GIT_USER
+      #   value: mygitusername
+      # - name: CLEARML_AGENT_GIT_PASS
+      #   valueFrom:
+      #     secretKeyRef:
+      #       name: git-password
+      #       key: git-password
+    # -- resources declaration for pods spawned to consume ClearML Task (example in values.yaml comments)
+    resources: {}
+      # limits:
+      #   nvidia.com/gpu: 1
+    # -- tolerations setup for pods spawned to consume ClearML Task (example in values.yaml comments)
+    tolerations: []
+      # - key: "nvidia.com/gpu"
+      #   operator: Exists
+      #   effect: "NoSchedule"
+    # -- nodeSelector setup for pods spawned to consume ClearML Task (example in values.yaml comments)
+    nodeSelector: {}
+      # fleet: gpu-nodes