mlflow

A Helm chart for Mlflow open source platform for the machine learning lifecycle

Get Helm Repository Info

helm repo add community-charts https://community-charts.github.io/helm-charts
helm repo update

See helm repo for command documentation.

Installing the Chart

helm install [RELEASE_NAME] community-charts/mlflow

See configuration below.

See helm install for command documentation.

Tip: Search all available chart versions using helm search repo community-charts -l. Please don't forget to run helm repo update before the command.

Supported Databases

Currently, we support the following two databases as a backend repository for Mlflow.

• PostgreSQL
• MySQL

Supported Cloud Providers

We currently support the following three cloud providers for BLOB storage integration.

• AWS (S3)
• Google Cloud Platform (Cloud Storage)
• Azure Cloud (Azure Blob Storage)

Values Files Examples

Postgres Database Migration Values Files Example

backendStore:
  databaseMigration: true
  postgres:
    enabled: true
    host: "postgresql-instance1.cg034hpkmmjt.eu-central-1.rds.amazonaws.com"
    port: 5432
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

MySQL Database Migration Values Files Example

backendStore:
  databaseMigration: true
  mysql:
    enabled: true
    host: "mysql-instance1.cg034hpkmmjt.eu-central-1.rds.amazonaws.com"
    port: 3306
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

Postgres Database Connection Check Values Files Example

backendStore:
  databaseConnectionCheck: true
  postgres:
    enabled: true
    host: "postgresql-instance1.cg034hpkmmjt.eu-central-1.rds.amazonaws.com"
    port: 5432
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

MySQL Database Connection Check Values Files Example

backendStore:
  databaseConnectionCheck: true
  mysql:
    enabled: true
    host: "mysql-instance1.cg034hpkmmjt.eu-central-1.rds.amazonaws.com"
    port: 3306
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

AWS Installation Examples

You can use 2 different way to connect your S3 backend.

• First way, you can access to your S3 with IAM user's awsAccessKeyId and awsSecretAccessKey.
• Second way, you can create an aws role for your service account. And you can assign your role ARN from serviceAccount annotation. You don't need to create or manage IAM user anymore. Please find more information from here.

Tip: Please follow this tutorial to create your own RDS postgres cluster.

S3 (Minio) and PostgreSQL DB Configuration on Helm Upgrade Command Example

helm upgrade --install mlflow community-charts/mlflow \
  --set backendStore.databaseMigration=true \
  --set backendStore.postgres.enabled=true \
  --set backendStore.postgres.host=postgres-service \
  --set backendStore.postgres.port=5432 \
  --set backendStore.postgres.database=postgres \
  --set backendStore.postgres.user=postgres \
  --set backendStore.postgres.password=postgres \
  --set artifactRoot.s3.enabled=true \
  --set artifactRoot.s3.bucket=mlflow \
  --set artifactRoot.s3.awsAccessKeyId=minioadmin \
  --set artifactRoot.s3.awsSecretAccessKey=minioadmin \
  --set extraEnvVars.MLFLOW_S3_ENDPOINT_URL=http://minio-service:9000 \
  --set serviceMonitor.enabled=true

S3 (Minio) and MySQL DB Configuration on Helm Upgrade Command Example

helm upgrade --install mlflow community-charts/mlflow \
  --set backendStore.databaseMigration=true \
  --set backendStore.mysql.enabled=true \
  --set backendStore.mysql.host=mysql-service \
  --set backendStore.mysql.port=3306 \
  --set backendStore.mysql.database=mlflow \
  --set backendStore.mysql.user=mlflow \
  --set backendStore.mysql.password=mlflow \
  --set artifactRoot.s3.enabled=true \
  --set artifactRoot.s3.bucket=mlflow \
  --set artifactRoot.s3.awsAccessKeyId=minioadmin \
  --set artifactRoot.s3.awsSecretAccessKey=minioadmin \
  --set extraEnvVars.MLFLOW_S3_ENDPOINT_URL=http://minio-service:9000 \
  --set serviceMonitor.enabled=true

S3 Access with awsAccessKeyId and awsSecretAccessKey Values Files Example

backendStore:
  postgres:
    enabled: true
    host: "postgresql-instance1.cg034hpkmmjt.eu-central-1.rds.amazonaws.com"
    port: 5432
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

artifactRoot:
  s3:
    enabled: true
    bucket: "my-mlflow-artifact-root-backend"
    awsAccessKeyId: "a1b2c3d4"
    awsSecretAccessKey: "a1b2c3d4"

S3 Access with AWS EKS Role ARN Values Files Example

Tip: Associate an IAM role to a service account

serviceAccount:
  create: true
  annotations:
    eks.amazonaws.com/role-arn: "arn:aws:iam::account-id:role/iam-role-name"
  name: "mlflow"

backendStore:
  postgres:
    enabled: true
    host: "postgresql-instance1.cg034hpkmmjt.eu-central-1.rds.amazonaws.com"
    port: 5432
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

artifactRoot:
  s3:
    enabled: true
    bucket: "my-mlflow-artifact-root-backend"

Azure Cloud Installation Example

Tip: Please follow this tutorial to create your own postgres database. Tip: Please follow this tutorial to create your azure blob storage and container.

backendStore:
  postgres:
    enabled: true
    host: "mydemoserver.postgres.database.azure.com"
    port: 5432
    database: "mlflow"
    user: "mlflowuser"
    password: "Pa33w0rd!"

artifactRoot:
  azureBlob:
    enabled: true
    container: "mlflow"
    storageAccount: "mystorageaccount"
    accessKey: "Eby8vdM02xNOcqFlqUwJPLlmEtlCDXJ1OUzFT50uSRZ6IFsuFq2UVErCz4I6tq/K1SZFPTOtr/KBHBeksoGMGw=="

Requirements

Kubernetes: >=1.16.0-0

Uninstall Helm Chart

helm uninstall [RELEASE_NAME]

This removes all the Kubernetes components associated with the chart and deletes the release.

See helm uninstall for command documentation.

Upgrading Chart

helm upgrade [RELEASE_NAME] community-charts/mlflow

Values

Key	Type	Default	Description
affinity	object	{}	Set the affinity for the pod.
artifactRoot.azureBlob.accessKey	string	""	Azure Cloud Storage Account Access Key for the container
artifactRoot.azureBlob.connectionString	string	""	Azure Cloud Connection String for the container. Only onnectionString or accessKey required
artifactRoot.azureBlob.container	string	""	Azure blob container name
artifactRoot.azureBlob.enabled	bool	false	Specifies if you want to use Azure Blob Storage Mlflow Artifact Root
artifactRoot.azureBlob.path	string	""	Azure blobk container folder. If you want to use root level, please don't set anything.
artifactRoot.azureBlob.storageAccount	string	""	Azure storage account name
artifactRoot.gcs.bucket	string	""	Google Cloud Storage bucket name
artifactRoot.gcs.enabled	bool	false	Specifies if you want to use Google Cloud Storage Mlflow Artifact Root
artifactRoot.gcs.path	string	""	Google Cloud Storage bucket folder. If you want to use root level, please don't set anything.
artifactRoot.proxiedArtifactStorage	bool	false	Specifies if you want to enable proxied artifact storage access
artifactRoot.s3.awsAccessKeyId	string	""	AWS IAM user AWS_ACCESS_KEY_ID which has attached policy for access to the S3 bucket
artifactRoot.s3.awsSecretAccessKey	string	""	AWS IAM user AWS_SECRET_ACCESS_KEY which has attached policy for access to the S3 bucket
artifactRoot.s3.bucket	string	""	S3 bucket name
artifactRoot.s3.enabled	bool	false	Specifies if you want to use AWS S3 Mlflow Artifact Root
artifactRoot.s3.path	string	""	S3 bucket folder. If you want to use root level, please don't set anything.
backendStore.databaseConnectionCheck	bool	false	Add an additional init container, which checks for database availability
backendStore.databaseMigration	bool	false	Specifies if you want to run database migration
backendStore.mysql.database	string	""	mlflow database name created before in the mysql instance
backendStore.mysql.driver	string	"pymysql"	mysql database connection driver. e.g.: "pymysql"
backendStore.mysql.enabled	bool	false	Specifies if you want to use mysql backend storage
backendStore.mysql.host	string	""	MySQL host address. e.g. your Amazon RDS for MySQL
backendStore.mysql.password	string	""	mysql database user password which can access to mlflow database
backendStore.mysql.port	int	3306	MySQL service port
backendStore.mysql.user	string	""	mysql database user name which can access to mlflow database
backendStore.postgres.database	string	""	mlflow database name created before in the postgres instance
backendStore.postgres.driver	string	""	postgres database connection driver. e.g.: "psycopg2"
backendStore.postgres.enabled	bool	false	Specifies if you want to use postgres backend storage
backendStore.postgres.host	string	""	Postgres host address. e.g. your RDS or Azure Postgres Service endpoint
backendStore.postgres.password	string	""	postgres database user password which can access to mlflow database
backendStore.postgres.port	int	5432	Postgres service port
backendStore.postgres.user	string	""	postgres database user name which can access to mlflow database
extraArgs	object	{}	A map of arguments and values to pass to the mlflow server command Keys must be camelcase. Helm will turn them to kebabcase style.
extraContainers	list	[]	Extra containers for the mlflow pod
extraEnvVars	object	{}	Extra environment variables
extraFlags	list	[]	A list of flags to pass to mlflow server command Items must be camelcase. Helm will turn them to kebabcase style.
extraSecretNamesForEnvFrom	list	[]	Extra secrets for environment variables
extraVolumeMounts	list	[]	Extra Volume Mounts for the mlflow container
extraVolumes	list	[]	Extra Volumes for the pod
fullnameOverride	string	""	String to override the default generated fullname
image.pullPolicy	string	"IfNotPresent"	The docker image pull policy
image.repository	string	"burakince/mlflow"	The docker image repository to use
image.tag	string	""	The docker image tag to use. Default app version
imagePullSecrets	list	[]	Image pull secrets for private docker registry usages
ingress.annotations	object	{}	Additional ingress annotations
ingress.className	string	""	New style ingress class name. Only possible if you use K8s 1.18.0 or later version
ingress.enabled	bool	false	Specifies if you want to create an ingress access
ingress.hosts[0].host	string	"chart-example.local"
ingress.hosts[0].paths[0].path	string	"/"
ingress.hosts[0].paths[0].pathType	string	"ImplementationSpecific"	Ingress path type
ingress.tls	list	[]	Ingress tls configuration for https access
initContainers	list	[]	Init Containers for Mlflow Pod
livenessProbe	object	{}	Liveness probe configurations. Please look to here.
nameOverride	string	""	String to override the default generated name
nodeSelector	object	{}	Set the node selector for the pod.
podAnnotations	object	{}	Annotations for the pod
podSecurityContext	object	{}	Security context for all pod
readinessProbe	object	{}	Readiness probe configurations. Please look to here.
replicaCount	int	1	Numbers of replicas
resources	object	{}	Set the resources requests and limits
securityContext	object	{}	Security context for the mlflow container
service.annotations	object	{}	Additional service annotations
service.name	string	"http"	Default Service name
service.port	int	5000	Default Service port
service.type	string	"ClusterIP"	Specifies what type of Service should be created
serviceAccount.annotations	object	{}	Annotations to add to the service account. AWS EKS users can assign role arn from here. Please find more information from here: https://docs.aws.amazon.com/eks/latest/userguide/specify-service-account-role.html
serviceAccount.create	bool	true	Specifies whether a ServiceAccount should be created
serviceAccount.name	string	""	The name of the ServiceAccount to use. If not set and create is true, a name is generated using the fullname template
serviceMonitor.enabled	bool	false	When set true then use a ServiceMonitor to configure scraping
serviceMonitor.interval	string	"30s"	Set how frequently Prometheus should scrape
serviceMonitor.labels	object	{"release":"prometheus"}	Set labels for the ServiceMonitor, use this to define your scrape label for Prometheus Operator
serviceMonitor.labels.release	string	"prometheus"	default kube prometheus stack helm chart serviceMonitor selector label Mostly it's your prometheus helm release name. Please find more information from here: https://github.com/prometheus-operator/prometheus-operator/blob/main/Documentation/troubleshooting.md#troubleshooting-servicemonitor-changes
serviceMonitor.metricRelabelings	list	[]	Set of rules to relabel your exist metric labels
serviceMonitor.namespace	string	"monitoring"	Set the namespace the ServiceMonitor should be deployed
serviceMonitor.targetLabels	list	[]	Set of labels to transfer on the Kubernetes Service onto the target.
serviceMonitor.telemetryPath	string	"/metrics"	Set path to mlflow telemtery-path
serviceMonitor.timeout	string	"10s"	Set timeout for scrape
serviceMonitor.useServicePort	bool	false	When set true then use a service port. On default use a pod port.
tolerations	list	[]	Set the tolerations for the pod.

Homepage: https://mlflow.org

Source Code

• https://github.com/community-charts/helm-charts
• https://github.com/burakince/mlflow
• https://github.com/mlflow/mlflow

Chart Development

Please install unittest helm plugin with helm plugin install https://github.com/quintush/helm-unittest command and use following command to run helm unit tests.

helm unittest --helm3 --strict --file unittests/*.yaml --file unittests/**/*.yaml charts/mlflow

Maintainers

Name	Email	Url
burakince	burak.ince@linux.org.tr	https://www.burakince.net

Updated

Name	Email	Url
serdarcanb	serdarcanbuyukdereli@gmail.com	https://serdarcanb.github.io