Advanced Helm Chart Configuration

Seldon Core Operator Chart Configuration

This page provides a detailed overview of the installation parameters available for the Seldon Core installation when using Helm 3.x. The high level workflows to install Seldon Core can be found in the Installation Page.

Below you can find the values.yaml file of the seldon-core-operator Helm chart, which contains basically all the values that you can configure in your installation by using the set flag in the format –set value.path=YOUR_VALUE.

The file has been written to be self documented, and has information on all the core parameters. Further information is referenced in the file to specific documentation pages.

  1# # Seldon Core Operator
  2# Below are the default values when installing Seldon Core
  3
  4# Defaults to .Release.Namespace
  5namespaceOverride: ""
  6
  7# ## Ingress Options
  8# You are able to choose between Istio and Ambassador
  9
 10# If you have ambassador installed you can just use the enabled flag
 11ambassador:
 12  enabled: false
 13  version: v2
 14  singleNamespace: false
 15# When activating Istio, respecive virtual services will be created
 16# You must make sure you create the seldon-gateway as well
 17istio:
 18  enabled: false
 19  gateway: istio-system/seldon-gateway
 20  tlsMode: ""
 21# If you have KEDA installed you can use it for autoscaling
 22keda:
 23  enabled: false
 24# ## Install with Cert Manager
 25# See installation page in documentation for more information
 26certManager:
 27  enabled: false
 28
 29# ## Install with limited namespace visibility
 30# If you want to ensure seldon-core-controller can only have visibility
 31#   to specify namespaces you can set the controllerId
 32controllerId: ""
 33
 34# Whether operator should create the webhooks and configmap on startup (false means created from chart)
 35managerCreateResources: false
 36
 37# Default user id to add to all Pod Security Context as the default
 38# Use this to ensure all container run as non-root by default
 39# For openshift leave blank as usually this will be injected automatically on an openshift cluster
 40# to all pods.
 41defaultUserID: "8888"
 42
 43# runtime user to use for manager
 44managerUserID: 8888
 45
 46# ## Service Orchestrator (Executor)
 47# The executor is the default service orchestrator which has superceded the "Java Engine"
 48executor:
 49  port: 8000
 50  metricsPortName: metrics
 51  image:
 52    pullPolicy: IfNotPresent
 53    registry: docker.io
 54    repository: seldonio/seldon-core-executor
 55    tag: 1.19.0-dev
 56  resources:
 57    cpuLimit: 500m
 58    cpuRequest: 500m
 59    memoryLimit: 512Mi
 60    memoryRequest: 512Mi
 61  prometheus:
 62    path: /prometheus
 63  serviceAccount:
 64    name: default
 65  user: 8888
 66# If you want to make available your own request logger for ELK integration you can set this
 67# For more information see the Production Integration for Payload Request Logging with ELK in the docs
 68  requestLogger:
 69    defaultEndpoint: 'http://default-broker'
 70    workQueueSize: 10000
 71    writeTimeoutMs: 2000
 72  # Whether to run full protocol API based health checks on models in deployment graph. False will just do TCP connects
 73  fullHealthChecks: false
 74
 75
 76# ## Seldon Core Controller Manager Options
 77image:
 78  pullPolicy: IfNotPresent
 79  registry: docker.io
 80  repository: seldonio/seldon-core-operator
 81  tag: 1.19.0-dev
 82manager:
 83  cpuLimit: 500m
 84  cpuRequest: 100m
 85  memoryLimit: 300Mi
 86  memoryRequest: 200Mi
 87  logLevel: INFO
 88  leaderElectionID: a33bd623.machinelearning.seldon.io
 89  leaderElectionResourceLock:
 90  leaderElectionLeaseDurationSecs: 15
 91  leaderElectionRenewDeadlineSecs: 10
 92  leaderElectionRetryPeriodSecs: 2
 93  annotations: {}
 94  containerSecurityContext: {}
 95  deploymentNameAsPrefix: false
 96  priorityClassName:
 97rbac:
 98  configmap:
 99    create: true
100  create: true
101serviceAccount:
102  create: true
103  name: seldon-manager
104singleNamespace: false
105storageInitializer:
106  cpuLimit: "1"
107  cpuRequest: 100m
108  image: seldonio/rclone-storage-initializer:1.19.0-dev
109  memoryLimit: 1Gi
110  memoryRequest: 100Mi
111usageMetrics:
112  enabled: false
113# In scenarios like EKS with non-standard CNI plugin like calico, the control plane cannot reach the webhook
114# hence it is needed to set hostNetwork: true
115hostNetwork: false
116webhook:
117  port: 4443 # If 'hostNetwork: true' you might need to change this port if it is already used by the node
118metrics: # these are the metrics exposed by the controller pod
119  port: 8080 # If 'hostNetwork: true' you might need to change this port if it is already used by the node
120
121# ## Predictive Unit Values
122predictiveUnit:
123  httpPort: 9000
124  grpcPort: 9500
125  metricsPortName: metrics
126  # If you would like to add extra environment variables to the init container to make available
127  #   secrets such as cloud credentials, you can provide a default secret name that will be loaded
128  #   to all the containers. You can then override this using the envSecretRefName in SeldonDeployments
129  defaultEnvSecretRefName: ""
130predictor_servers:
131  MLFLOW_SERVER:
132    protocols:
133      seldon:
134        defaultImageVersion: "1.19.0-dev"
135        image: seldonio/mlflowserver
136      v2:
137        defaultImageVersion: "1.3.5-mlflow"
138        image: seldonio/mlserver
139  SKLEARN_SERVER:
140    protocols:
141      seldon:
142        defaultImageVersion: "1.19.0-dev"
143        image: seldonio/sklearnserver
144      v2:
145        defaultImageVersion: "1.3.5-sklearn"
146        image: seldonio/mlserver
147  TENSORFLOW_SERVER:
148    protocols:
149      seldon:
150        defaultImageVersion: "1.19.0-dev"
151        image: seldonio/tfserving-proxy
152      tensorflow:
153        defaultImageVersion: 2.1.0
154        image:  tensorflow/serving
155  XGBOOST_SERVER:
156    protocols:
157      seldon:
158        defaultImageVersion: "1.19.0-dev"
159        image: seldonio/xgboostserver
160      v2:
161        defaultImageVersion: "1.3.5-xgboost"
162        image: seldonio/mlserver
163  TRITON_SERVER:
164    protocols:
165      v2:
166        defaultImageVersion: "21.08-py3"
167        image: nvcr.io/nvidia/tritonserver
168  HUGGINGFACE_SERVER:
169    protocols:
170      v2:
171        defaultImageVersion: "1.3.5-huggingface"
172        image: seldonio/mlserver
173  TEMPO_SERVER:
174    protocols:
175      v2:
176        defaultImageVersion: "1.3.5-slim"
177        image: seldonio/mlserver
178
179# ## Other
180# You can choose the crds to not be installed if you already installed them
181# This applies to just the yaml template. If you set managerCreateResources=true then
182# it will try to create the CRD but only if it does not exist
183crd:
184  create: true
185  # Whether to force the use of the v1beta1 or v1 CRD.
186  forceV1: false
187  forceV1beta1: false
188  annotations: {} # Annotations to add to the CRD if not created by the Manager
189
190# Warning: credentials will be depricated soon, please use defaultEnvSecretRefName above
191# For more info please check the documentation
192credentials:
193  gcs:
194    gcsCredentialFileName: gcloud-application-credentials.json
195  s3:
196    s3AccessKeyIDName: awsAccessKeyID
197    s3SecretAccessKeyName: awsSecretAccessKey
198
199kubeflow: false
200
201# Explainer image
202explainer:
203  image: seldonio/alibiexplainer:1.19.0-dev
204  image_v2: seldonio/mlserver:1.3.5-alibi-explain