add improvements to k8s agent

charts/clearml/Chart.yaml

@@ -2,7 +2,7 @@ apiVersion: v2
 name: clearml
 description: MLOps platform
 type: application
-version: "3.5.1"
+version: "3.6.0"
 appVersion: "1.2.0"
 home: https://clear.ml
 icon: https://raw.githubusercontent.com/allegroai/clearml/master/docs/clearml-logo.svg

charts/clearml/README.md

@@ -163,16 +163,19 @@ For detailed instructions, see the [Optional Configuration](https://github.com/a
 | agentGroups.agent-group-gpu.replicaCount | int | `0` |  |
 | agentGroups.agent-group-gpu.tolerations | list | `[]` |  |
 | agentGroups.agent-group-gpu.updateStrategy | string | `"Recreate"` |  |
-| agentk8sglue.defaultDockerImage | string | `"nvidia/cuda:11.3.1-cudnn8-runtime-ubuntu18.04"` |  |
+| agentk8sglue.defaultDockerImage | string | `"nvidia/cuda:11.3.1-cudnn8-runtime-ubuntu20.04"` |  |
 | agentk8sglue.enabled | bool | `false` |  |
 | agentk8sglue.id | string | `"k8s-agent"` |  |
 | agentk8sglue.image.repository | string | `"allegroai/clearml-agent-k8s"` |  |
 | agentk8sglue.image.tag | string | `"aws-latest-1.21"` |  |
 | agentk8sglue.maxPods | int | `10` |  |
+| agentk8sglue.podTemplate.env | object | `{}` |  |
 | agentk8sglue.podTemplate.nodeSelector | object | `{}` |  |
 | agentk8sglue.podTemplate.resources | object | `{}` |  |
 | agentk8sglue.podTemplate.tolerations | object | `{}` |  |
+| agentk8sglue.podTemplate.volumes | list | `[]` |  |
 | agentk8sglue.queue | string | `"aws-instances"` |  |
+| agentk8sglue.serviceAccountName | string | `"default"` |  |
 | agentservices.affinity | object | `{}` |  |
 | agentservices.agentVersion | string | `""` |  |
 | agentservices.awsAccessKeyId | string | `nil` |  |

charts/clearml/templates/configmap-agentk8s-template.yaml

@@ -9,9 +9,23 @@ data:
     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: "http://{{ include "clearml.fullname" . }}-apiserver:{{ .Values.apiserver.service.port }}"
@@ -29,6 +43,7 @@ data:
             secretKeyRef:
               name: clearml-conf
               key: apiserver_secret
+        {{- toYaml .Values.agentk8sglue.podTemplate.env | nindent 8 }}
       tolerations:
         {{- toYaml .Values.agentk8sglue.podTemplate.tolerations | nindent 8 }}
       nodeSelector:

charts/clearml/templates/rbac-agentk8s.yaml

@@ -8,7 +8,7 @@ rules:
       - ""
     resources:
       - pods
-    verbs: ["get", "list", "watch", "create", "patch"]
+    verbs: ["get", "list", "watch", "create", "patch", "delete"]
 ---
 apiVersion: rbac.authorization.k8s.io/v1
 kind: RoleBinding

charts/clearml/values.yaml

@@ -302,11 +302,24 @@ agentk8sglue:
   image:
     repository: "allegroai/clearml-agent-k8s"
     tag: "aws-latest-1.21"
+  serviceAccountName: default
   maxPods: 10
-  defaultDockerImage: nvidia/cuda:11.3.1-cudnn8-runtime-ubuntu18.04  # default docker image that is spawned as new pod
+  defaultDockerImage: nvidia/cuda:11.3.1-cudnn8-runtime-ubuntu20.04 # default docker image that is spawned as new pod
   queue: aws-instances  # create this queue manually in the UI first for it to work
   id: k8s-agent
   podTemplate:
+    volumes: []
+      # - name: "yourvolume"
+      #   path: "/yourpath"
+    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: {}
       # limits:
       #   nvidia.com/gpu: 1