DGX Cloud Admission Controller Helm Chart#

Values for Amazon Web Services (AWS)#

The following table shows the values for the DGX Cloud Admission Controller Helm Chart for Amazon Web Services (AWS).

Key

Type

Default

Description

kubernetesAffinity.enabled

boolean

false

Whether to enable Kubernetes affinity rules.

systemTolerationKeyName

string

"dedicated"

Key name for system workload toleration.

systemTolerationValueName

string

"system-workload"

Value for system workload toleration.

imagePullSecret

string

"nvcrimagepullsecret"

Name of the image pull secret for accessing container images.

dgxcController.image

string

""

Container image for DGX Admission Controller.

dgxcController.controllerPort

integer

8443

Port number for the controller service.

dgxcController.replicas

integer

1

Number of controller replicas to deploy.

tolerations.customer.key

string

"dedicated"

Toleration key for customer workloads.

tolerations.customer.value

string

"user-workload"

Toleration value for customer workloads.

tolerations.system.key

string

"dedicated"

Toleration key for system workloads.

tolerations.system.value

string

"system-workload"

Toleration value for system workloads.

cloudProviderSpec.type

string

"aws"

Cloud provider type.

cloudProviderSpec.aws.efa

boolean

true

Whether to enable EFA.

cloudProviderSpec.aws.region

string

"us-east-1"

AWS region.

cloudProviderSpec.aws.accountName

string

"111122223333"

AWS account name or ID.

cloudProviderSpec.aws.clusterName

string

""

Name of the cluster.

cloudProviderSpec.aws.statefulRuleGroupName

string

""

Name of the stateful rule group.

cloudProviderSpec.aws.gpuNode.gpuAllocatable

integer

8

Specifies the number of GPUs allocatable.

cloudProviderSpec.aws.gpuNode.efaAllocatable

integer

32

Specifies the number of EFA allocatable.

cloudProviderSpec.aws.gpuNode.hugePagesAllocatable

string

"42242Mi"

Specifies the amount of huge pages allocatable.

envVars.user.efa

list

See values file

Environment variables for EFA configuration.

webhooks.registeredWebhooks

list

See values file

Registered webhooks. Do not modify.

webhooks.podMutatingWebhook.customSettings

list

See values file

Pod mutating webhook configuration. Do not modify.

enableIntegrationsSupport

boolean

false

Whether to enable integrations support. Do not modify.

enableLabeler

boolean

false

Whether to enable labeler. Do not modify.

enableRunAIRbac

boolean

false

Whether to enable RunAI RBAC. Do not modify.

enableNetworkPolicies

boolean

false

Whether to enable network policies. Do not modify.

enableDGXCRbac

boolean

false

Whether to enable DGX CRBAC. Do not modify.

enableDGXCAPIBootstrap

boolean

false

Whether to enable DGX CA bootstrap. Do not modify.

enableServiceMonitor

boolean

false

Whether to enable service monitor. Do not modify.

enabledComponents.dgxcAdmissionController

boolean

true

Whether to enable the DGX Admission Controller.

enabledComponents.dgxcRunaiController

boolean

false

Whether to enable the DGX RunAI Controller.

enabledComponents.dgxcExporter

boolean

false

Whether to enable the DGX Exporter.

Values for Azure#

The following table shows the values for the DGX Cloud Admission Controller Helm Chart for Azure.

Key

Type

Default

Description

kubernetesAffinity.enabled

boolean

false

Whether to enable Kubernetes affinity rules.

systemTolerationKeyName

string

"dedicated"

Key name for system workload toleration.

systemTolerationValueName

string

"system-workload"

Value for system workload toleration.

imagePullSecret

string

"nvcrimagepullsecret"

Name of the image pull secret for accessing container images.

dgxcController.replicas

integer

1

Number of the DGX Admission Controller replicas to deploy.

dgxcController.image

string

""

Container image for the DGX Admission Controller.

webhooks.registeredWebhooks

list

See values file

Registered webhooks. Do not modify.

webhooks.podMutatingWebhook.customSettings

list

See values file

Pod mutating webhook configuration. Do not modify.

enableIntegrationsSupport

boolean

false

Whether to enable integrations support. Do not modify.

enableLabeler

boolean

false

Whether to enable labeler. Do not modify.

enableRunAIRbac

boolean

false

Whether to enable RunAI RBAC. Do not modify.

enableNetworkPolicies

boolean

false

Whether to enable network policies. Do not modify.

enableDGXCRbac

boolean

false

Whether to enable DGX CRBAC. Do not modify.

enableDGXCAPIBootstrap

boolean

false

Whether to enable DGX CA bootstrap. Do not modify.

enableServiceMonitor

boolean

false

Whether to enable service monitor. Do not modify.

tolerations.customer.key

string

"dedicated"

Toleration key for customer workloads.

tolerations.customer.value

string

"user-workload"

Toleration value for customer workloads.

tolerations.system.key

string

"dedicated"

Toleration key for system workloads.

tolerations.system.value

string

"system-workload"

Toleration value for system workloads.

cloudProviderSpec.type

string

"azure"

Specifies the cloud provider type.

cloudProviderSpec.azure.instanceType

string

"Standard_ND96amsr_A100_v4"

Specifies the Azure instance type.

enabledComponents.dgxcAdmissionController

boolean

true

Whether to enable the DGX Admission Controller.

enabledComponents.dgxcRunaiController

boolean

false

Whether to enable the DGX RunAI Controller.

enabledComponents.dgxcExporter

boolean

false

Whether to enable the DGX Exporter.

Values for Oracle Cloud Infrastructure (OCI)#

The following table shows the values for the DGX Cloud Admission Controller Helm Chart for Oracle Cloud Infrastructure (OCI).

Key

Type

Default

Description

kubernetesAffinity.enabled

boolean

false

Whether to enable Kubernetes affinity rules.

systemTolerationKeyName

string

"CriticalAddonsOnly"

System toleration key name.

systemTolerationValueName

string

"true"

System toleration value name.

imagePullSecret

string

"nvcrimagepullsecret"

Image pull secret name.

dgxcController.replicas

integer

1

Number of the DGX Cloud Admission Controller replicas to deploy.

dgxcController.image

string

""

Container image for the DGX Cloud Admission Controller.

webhooks.registeredWebhooks

list

See values file

Registered webhooks. Do not modify.

webhooks.podMutatingWebhook.customSettings

list

See values file

Pod mutating webhook configuration. Do not modify.

enableIntegrationsSupport

boolean

false

Whether to enable integrations support. Do not modify.

enableLabeler

boolean

false

Whether to enable labeler. Do not modify.

enableRunAIRbac

boolean

false

Whether to enable Run:AI RBAC. Do not modify.

enableNetworkPolicies

boolean

false

Whether to enable network policies. Do not modify.

enableDGXCRbac

boolean

false

Whether to enable DGX CRBAC. Do not modify.

enableDGXCAPIBootstrap

boolean

false

Whether to enable DGX CA bootstrap. Do not modify.

enableServiceMonitor

boolean

false

Whether to enable service monitor. Do not modify.

tolerations.customer.key

string

"dedicated"

Toleration key for customer workloads.

tolerations.customer.value

string

"user-workload"

Toleration value for customer workloads.

tolerations.system.key

string

"CriticalAddonsOnly"

Toleration key for system workloads.

tolerations.system.value

string

"true"

Toleration value for system workloads.

cloudProviderSpec.type

string

"oci"

Specifies the cloud provider type.

cloudProviderSpec.oci.resourcePrincipalVersion

string

"2.2"

Specifies the resource principal version.

storageControllerSpecOCI.enabled

boolean

false

Whether to enable the storage controller.

enabledComponents.dgxcAdmissionController

boolean

true

Whether to enable the DGX Cloud Admission Controller.

enabledComponents.dgxcRunaiController

boolean

false

Whether to enable the DGX Cloud Run:AI Controller.

enabledComponents.dgxcExporter

boolean

false

Whether to enable the DGX Cloud Exporter.

Values for Google Cloud Platform (GCP)#

The following table shows the values for the DGX Cloud Admission Controller Helm Chart for Google Cloud Platform (GCP).

Key

Type

Default

Description

kubernetesAffinity.enabled

boolean

true

Whether to enable Kubernetes affinity rules to ensure pods are running on different nodes.

kubernetesAffinity.antiAffinityKey

string

"cloud.google.com/gke-gpu"

Anti-affinity key.

kubernetesAffinity.antiAffinityValue

boolean

true

Anti-affinity value.

kubernetesAffinity.topologyKey

string

"kubernetes.io/hostname"

Topology key.

kubernetesAffinity.runOnMasters

boolean

false

Whether to run on masters. Run on masters should remain false unless there is a specific need, such as if you have been given a minimal dev cluster, with only masters.

kubernetesAffinity.systemTolerationKeyName

string

"components.gke.io/gke-managed-components"

System toleration key name.

kubernetesAffinity.systemTolerationValueName

string

"true"

System toleration value name.

imagePullSecret

string

"nvcrimagepullsecret"

Name of the image pull secret for accessing container images.

dgxcController.replicas

integer

1

Number of the DGX Admission Controller replicas to deploy.

dgxcController.image

string

""

Container image for the DGX Admission Controller.

webhooks.registeredWebhooks

list

See values file

Registered webhooks. Do not modify.

webhooks.podMutatingWebhook.customSettings

list

See values file

Pod mutating webhook configuration. Do not modify.

tolerations.customer.key

string

"dedicated"

Toleration key for customer workloads.

tolerations.customer.value

string

"user-workload"

Toleration value for customer workloads.

tolerations.system.key

string

"components.gke.io/gke-managed-components"

Toleration key for system workloads.

tolerations.system.value

string

"true"

Toleration value for system workloads.

enableIntegrationsSupport

boolean

false

Whether to enable integrations support. Do not modify.

enableLabeler

boolean

false

Whether to enable labeler. Do not modify.

enableRunAIRbac

boolean

false

Whether to enable Run:AI RBAC. Do not modify.

enableNetworkPolicies

boolean

false

Whether to enable network policies. Do not modify.

enableDGXCRbac

boolean

false

Whether to enable DGX CRBAC. Do not modify.

enableDGXCAPIBootstrap

boolean

false

Whether to enable DGX CA bootstrap. Do not modify.

enableServiceMonitor

boolean

false

Whether to enable service monitor. Do not modify.

cloudProviderSpec.type

string

"gcp"

Specifies the cloud provider type.

enabledComponents.dgxcAdmissionController

boolean

true

Whether to enable the DGX Admission Controller.

enabledComponents.dgxcRunaiController

boolean

false

Whether to enable the DGX Run:AI Controller.

enabledComponents.dgxcExporter

boolean

false

Whether to enable the DGX Exporter.