helm-charts

Aizen package

A Helm chart for Aizen machine learning platform

Prerequisites

Get Helm Repository Info

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

Install Aizen infra and core components using chart properties

helm install [RELEASE_NAME] aizen/aizen

Install Aizen INFRA components (pre-requistes)

Ray operator

Key Type Default Description
infra.enabled string true Set this to “true” to install infrastructure components
infra.rayoperator.enabled string true  
infra.rayoperator.kuberay-operator.enabled string true  

Spark operator

Key Type Default Description
infra.enabled string true Set this to “true” to install infrastructure components
infra.sparkoperator.enabled string true  
infra.sparkoperator.spark-operator.enabled string true  
infra.sparkoperator.spark-webhook.enabled string true  

Filecopy server

Key Type Default Description
infra.enabled string true Set this to “true” to install infrastructure components
infra.filecopyserver.enabled string true  
global.image_registry string aizen.repo.com The docker image registry to use
global.image_tag string ”” common tag for all container images
global.image_secret string ”” Secret to access docker image registry
global.image_pullpolicy string IfNotPresent Image pull policy

Install Aizen CORE components

Key Type Default Description
core.enabled string false Set this to “true” to install aizen core components
global.image_registry string aizen.repo.com The docker image registry to use
global.image_tag string ”” common tag for all container images
global.image_secret string ”” Secret to access docker image registry
global.image_pullpolicy string IfNotPresent Image pull policy
global.storage_class string standard Backend storage
global.clustername string mycluster Name used to create unique buckets for backend object store
global.cloud_provider_type string ”” if using metadata in cloud set cloud provider type
global.cloud_provider_region string ”” if using metadata in cloud set cloud provider region name
global.customer_bucket_name string ”” if using metadata in cloud set customer bucket name
global.storage_type string local Default storage type. Options are ‘local’ or ‘cloud’
global.s3.endpoint_url string ”” S3 endpoint url for cloud provider
global.s3.secrets.enabled boolean false Enable s3 access
global.s3.secrets.values.s3_access_key string ”” Define s3 access key
global.s3.secrets.values.s3_secret_key string ”” Define s3 secret key
global.ingress.enabled boolean false Enable ingress
global.ingress.host string ”” Specify ingress ip address
global.log.enabled boolean false Enable logging
global.log.volume_size string 50Gi Volume size for storing log files
global.mlflow.mysql.primary.persistence.storageClass string standard Default storage class
global.mlflow.mysql.primary.persistence.size string 8Gi Default size for mlflow mysql pvc
global.mlflow.artifact.secrets.values.mlflow_artifacts_destination string s3://aizen-mlflow Default s3 destination
global.mlflow.artifact.secrets.values.mlflow_access_key_id string ”” Define s3 access key
global.mlflow.artifact.secrets.values.mlflow_access_secret_key string ”” Define s3 secret key
global.mlflow.artifact.secrets.values.mlflow_endpoint_url string ”” Define s3 end point url for mlflow
global.mlflow.artifact.region string ”” Define s3 region name
core.storage.blkcache_size_mb int 2048 Block cache size
core.storage.memtable_size_mb int 1024 Mem table size
core.storage.resources.limits.cpu string 8 cpu limit
core.storage.resources.limits.memory string 16Gi memory limit
core.storage.resources.requests.cpu string 2 requested cpu from the cpu limit
core.storage.resources.requests.memory string 2Gi requested memory from the memory limit
This site is open source. Improve this page.