clearml-helm-charts/charts/clearml

ClearML Ecosystem for Kubernetes

MLOps platform

Homepage: https://clear.ml

Maintainers

Name	Email	Url
valeriano-manassero		https://github.com/valeriano-manassero

Introduction

The clearml-server is the backend service infrastructure for ClearML. It allows multiple users to collaborate and manage their experiments.

clearml-server contains the following components:

• The ClearML Web-App, a single-page UI for experiment management and browsing
• RESTful API for:
  • Documenting and logging experiment information, statistics and results
  • Querying experiments history, logs and results
• Locally-hosted file server for storing images and models making them easily accessible using the Web-App

Local environment

For development/evaluation it's possible to use kind. After installation, following commands will create a complete ClearML insatllation:

mkdir -pm 777 /tmp/clearml-kind

cat <<EOF > /tmp/clearml-kind.yaml
kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
nodes:
- role: control-plane
  extraPortMappings:
  - containerPort: 30008
    hostPort: 30008
    listenAddress: "127.0.0.1"
    protocol: TCP
  - containerPort: 30080
    hostPort: 30080
    listenAddress: "127.0.0.1"
    protocol: TCP
  - containerPort: 30081
    hostPort: 30081
    listenAddress: "127.0.0.1"
    protocol: TCP
  extraMounts:
  - hostPath: /tmp/clearml-kind/
    containerPath: /var/local-path-provisioner
EOF

kind create cluster --config /tmp/clearml-kind.yaml

helm install clearml allegroai/clearml

After deployment, the services will be exposed on localhost on the following ports:

• API server on 30008
• Web server on 30080
• File server on 30081

Data persisted in every Kubernetes volume by ClearML will be accessible in /tmp/clearml-kind folder on the host.

Production cluster environment

In a production environment it's suggested to install an ingress controller and verify that is working correctly. During ClearML deployment enable ingress section of chart values. This will create 3 ingress rules:

• app.<your domain name>
• files.<your domain name>
• api.<your domain name>

(for example, app.clearml.mydomainname.com, files.clearml.mydomainname.com and api.clearml.mydomainname.com)

Just pointing the domain records to the IP where ingress controller is responding will complete the deployment process.

Additional Configuration for ClearML Server

You can also configure the clearml-server for:

• fixed users (users with credentials)
• non-responsive experiment watchdog settings

For detailed instructions, see the Optional Configuration section in the clearml-server repository README file.

Source Code

• https://github.com/allegroai/clearml-helm-charts
• https://github.com/allegroai/clearml

Requirements

Repository	Name	Version
https://charts.bitnami.com/bitnami	mongodb	~10.3.2
https://charts.bitnami.com/bitnami	redis	~10.9.0
https://helm.elastic.co	elasticsearch	~7.10.1

Values

Key	Type	Default	Description
agentGroups.agent-group-cpu.affinity	object	{}
agentGroups.agent-group-cpu.agentVersion	string	""
agentGroups.agent-group-cpu.awsAccessKeyId	string	nil
agentGroups.agent-group-cpu.awsDefaultRegion	string	nil
agentGroups.agent-group-cpu.awsSecretAccessKey	string	nil
agentGroups.agent-group-cpu.azureStorageAccount	string	nil
agentGroups.agent-group-cpu.azureStorageKey	string	nil
agentGroups.agent-group-cpu.clearmlAccessKey	string	nil
agentGroups.agent-group-cpu.clearmlConfig	string	"sdk {\n}"
agentGroups.agent-group-cpu.clearmlGitPassword	string	nil
agentGroups.agent-group-cpu.clearmlGitUser	string	nil
agentGroups.agent-group-cpu.clearmlSecretKey	string	nil
agentGroups.agent-group-cpu.image.pullPolicy	string	"IfNotPresent"
agentGroups.agent-group-cpu.image.repository	string	"ubuntu"
agentGroups.agent-group-cpu.image.tag	string	"18.04"
agentGroups.agent-group-cpu.name	string	"agent-group-cpu"
agentGroups.agent-group-cpu.nodeSelector	object	{}
agentGroups.agent-group-cpu.nvidiaGpusPerAgent	int	0
agentGroups.agent-group-cpu.podAnnotations	object	{}
agentGroups.agent-group-cpu.queues	string	"default"
agentGroups.agent-group-cpu.replicaCount	int	1
agentGroups.agent-group-cpu.tolerations	list	[]
agentGroups.agent-group-cpu.updateStrategy	string	"Recreate"
agentGroups.agent-group-gpu.affinity	object	{}
agentGroups.agent-group-gpu.agentVersion	string	""
agentGroups.agent-group-gpu.awsAccessKeyId	string	nil
agentGroups.agent-group-gpu.awsDefaultRegion	string	nil
agentGroups.agent-group-gpu.awsSecretAccessKey	string	nil
agentGroups.agent-group-gpu.azureStorageAccount	string	nil
agentGroups.agent-group-gpu.azureStorageKey	string	nil
agentGroups.agent-group-gpu.clearmlAccessKey	string	nil
agentGroups.agent-group-gpu.clearmlConfig	string	"sdk {\n}"
agentGroups.agent-group-gpu.clearmlGitPassword	string	nil
agentGroups.agent-group-gpu.clearmlGitUser	string	nil
agentGroups.agent-group-gpu.clearmlSecretKey	string	nil
agentGroups.agent-group-gpu.image.pullPolicy	string	"IfNotPresent"
agentGroups.agent-group-gpu.image.repository	string	"nvidia/cuda"
agentGroups.agent-group-gpu.image.tag	string	"11.0-base-ubuntu18.04"
agentGroups.agent-group-gpu.name	string	"agent-group-gpu"
agentGroups.agent-group-gpu.nodeSelector	object	{}
agentGroups.agent-group-gpu.nvidiaGpusPerAgent	int	1
agentGroups.agent-group-gpu.podAnnotations	object	{}
agentGroups.agent-group-gpu.queues	string	"default"
agentGroups.agent-group-gpu.replicaCount	int	0
agentGroups.agent-group-gpu.tolerations	list	[]
agentGroups.agent-group-gpu.updateStrategy	string	"Recreate"
agentservices.affinity	object	{}
agentservices.agentVersion	string	""
agentservices.awsAccessKeyId	string	nil
agentservices.awsDefaultRegion	string	nil
agentservices.awsSecretAccessKey	string	nil
agentservices.azureStorageAccount	string	nil
agentservices.azureStorageKey	string	nil
agentservices.clearmlFilesHost	string	nil
agentservices.clearmlGitPassword	string	nil
agentservices.clearmlGitUser	string	nil
agentservices.clearmlHostIp	string	nil
agentservices.clearmlWebHost	string	nil
agentservices.clearmlWorkerId	string	"clearml-services"
agentservices.extraEnvs	list	[]
agentservices.googleCredentials	string	nil
agentservices.image.pullPolicy	string	"IfNotPresent"
agentservices.image.repository	string	"allegroai/clearml-agent-services"
agentservices.image.tag	string	"latest"
agentservices.nodeSelector	object	{}
agentservices.podAnnotations	object	{}
agentservices.replicaCount	int	1
agentservices.resources	object	{}
agentservices.storage.data.class	string	"standard"
agentservices.storage.data.size	string	"50Gi"
agentservices.tolerations	list	[]
apiserver.affinity	object	{}
apiserver.configDir	string	"/opt/clearml/config"
apiserver.extraEnvs	list	[]
apiserver.image.pullPolicy	string	"IfNotPresent"
apiserver.image.repository	string	"allegroai/clearml"
apiserver.image.tag	string	"1.1.1"
apiserver.livenessDelay	int	60
apiserver.nodeSelector	object	{}
apiserver.podAnnotations	object	{}
apiserver.prepopulateArtifactsPath	string	"/mnt/fileserver"
apiserver.prepopulateEnabled	string	"true"
apiserver.prepopulateZipFiles	string	"/opt/clearml/db-pre-populate"
apiserver.readinessDelay	int	60
apiserver.replicaCount	int	1
apiserver.resources	object	{}
apiserver.service.port	int	8008
apiserver.service.type	string	"NodePort"
apiserver.storage.config.class	string	"standard"
apiserver.storage.config.size	string	"1Gi"
apiserver.storage.enableConfigVolume	bool	false
apiserver.tolerations	list	[]
clearml.defaultCompany	string	"d1bd92a3b039400cbafc60a7a5b1e52b"
elasticsearch.clusterHealthCheckParams	string	"wait_for_status=yellow&timeout=1s"
elasticsearch.clusterName	string	"clearml-elastic"
elasticsearch.enabled	bool	true
elasticsearch.esConfig."elasticsearch.yml"	string	"xpack.security.enabled: false\n"
elasticsearch.esJavaOpts	string	"-Xmx2g -Xms2g"
elasticsearch.extraEnvs[0].name	string	"bootstrap.memory_lock"
elasticsearch.extraEnvs[0].value	string	"false"
elasticsearch.extraEnvs[1].name	string	"cluster.routing.allocation.node_initial_primaries_recoveries"
elasticsearch.extraEnvs[1].value	string	"500"
elasticsearch.extraEnvs[2].name	string	"cluster.routing.allocation.disk.watermark.low"
elasticsearch.extraEnvs[2].value	string	"500mb"
elasticsearch.extraEnvs[3].name	string	"cluster.routing.allocation.disk.watermark.high"
elasticsearch.extraEnvs[3].value	string	"500mb"
elasticsearch.extraEnvs[4].name	string	"cluster.routing.allocation.disk.watermark.flood_stage"
elasticsearch.extraEnvs[4].value	string	"500mb"
elasticsearch.extraEnvs[5].name	string	"http.compression_level"
elasticsearch.extraEnvs[5].value	string	"7"
elasticsearch.extraEnvs[6].name	string	"reindex.remote.whitelist"
elasticsearch.extraEnvs[6].value	string	"*.*"
elasticsearch.extraEnvs[7].name	string	"xpack.monitoring.enabled"
elasticsearch.extraEnvs[7].value	string	"false"
elasticsearch.extraEnvs[8].name	string	"xpack.security.enabled"
elasticsearch.extraEnvs[8].value	string	"false"
elasticsearch.httpPort	int	9200
elasticsearch.minimumMasterNodes	int	1
elasticsearch.persistence.enabled	bool	true
elasticsearch.replicas	int	1
elasticsearch.resources.limits.memory	string	"4Gi"
elasticsearch.resources.requests.memory	string	"4Gi"
elasticsearch.roles.data	string	"true"
elasticsearch.roles.ingest	string	"true"
elasticsearch.roles.master	string	"true"
elasticsearch.roles.remote_cluster_client	string	"true"
elasticsearch.volumeClaimTemplate.accessModes[0]	string	"ReadWriteOnce"
elasticsearch.volumeClaimTemplate.resources.requests.storage	string	"50Gi"
fileserver.affinity	object	{}
fileserver.extraEnvs	list	[]
fileserver.image.pullPolicy	string	"IfNotPresent"
fileserver.image.repository	string	"allegroai/clearml"
fileserver.image.tag	string	"1.1.1"
fileserver.nodeSelector	object	{}
fileserver.podAnnotations	object	{}
fileserver.replicaCount	int	1
fileserver.resources	object	{}
fileserver.service.port	int	8081
fileserver.service.type	string	"NodePort"
fileserver.storage.data.class	string	"standard"
fileserver.storage.data.size	string	"50Gi"
fileserver.tolerations	list	[]
ingress.annotations	object	{}
ingress.enabled	bool	false
ingress.host	string	""
ingress.hostPrefixApi	string	"api."
ingress.hostPrefixApp	string	"app."
ingress.hostPrefixFiles	string	"files."
ingress.name	string	"clearml-server-ingress"
ingress.tls.secretName	string	""
mongodb.architecture	string	"standalone"
mongodb.auth.enabled	bool	false
mongodb.enabled	bool	true
mongodb.persistence.accessModes[0]	string	"ReadWriteOnce"
mongodb.persistence.enabled	bool	true
mongodb.persistence.size	string	"50Gi"
mongodb.replicaCount	int	1
mongodb.service.name	string	"{{ .Release.Name }}-mongodb"
mongodb.service.port	int	27017
mongodb.service.portName	string	"mongo-service"
mongodb.service.type	string	"ClusterIP"
redis.cluster.enabled	bool	false
redis.databaseNumber	int	0
redis.enabled	bool	true
redis.master.name	string	"{{ .Release.Name }}-redis-master"
redis.master.persistence.accessModes[0]	string	"ReadWriteOnce"
redis.master.persistence.enabled	bool	true
redis.master.persistence.size	string	"5Gi"
redis.master.port	int	6379
redis.usePassword	bool	false
webserver.affinity	object	{}
webserver.extraEnvs	list	[]
webserver.image.pullPolicy	string	"IfNotPresent"
webserver.image.repository	string	"allegroai/clearml"
webserver.image.tag	string	"1.1.1"
webserver.nodeSelector	object	{}
webserver.podAnnotations	object	{}
webserver.replicaCount	int	1
webserver.resources	object	{}
webserver.service.port	int	80
webserver.service.type	string	"NodePort"
webserver.tolerations	list	[]