ClusterInstanceGroupSpecification

Gets and sets the property CapacityRequirements.

Specifies the capacity requirements for the instance group.

Gets and sets the property ExecutionRole.

Specifies an IAM execution role to be assumed by the instance group.

Gets and sets the property ImageId.

When configuring your HyperPod cluster, you can specify an image ID using one of the following options:

• HyperPodPublicAmiId: Use a HyperPod public AMI

• CustomAmiId: Use your custom AMI

• default: Use the default latest system image

If you choose to use a custom AMI (CustomAmiId), ensure it meets the following requirements:

• Encryption: The custom AMI must be unencrypted.

• Ownership: The custom AMI must be owned by the same Amazon Web Services account that is creating the HyperPod cluster.

• Volume support: Only the primary AMI snapshot volume is supported; additional AMI volumes are not supported.

When updating the instance group's AMI through the UpdateClusterSoftware operation, if an instance group uses a custom AMI, you must provide an ImageId or use the default as input. Note that if you don't specify an instance group in your UpdateClusterSoftware request, then all of the instance groups are patched with the specified image.

Gets and sets the property InstanceCount.

Specifies the number of instances to add to the instance group of a SageMaker HyperPod cluster.

Gets and sets the property InstanceGroupName.

Specifies the name of the instance group.

Gets and sets the property InstanceStorageConfigs.

Specifies the additional storage configurations for the instances in the SageMaker HyperPod cluster instance group.

Gets and sets the property InstanceType.

Specifies the instance type of the instance group.

Gets and sets the property KubernetesConfig.

Specifies the Kubernetes configuration for the instance group. You describe what you want the labels and taints to look like, and the cluster works to reconcile the actual state with the declared state for nodes in this instance group.

Gets and sets the property LifeCycleConfig.

Specifies the LifeCycle configuration for the instance group.

Gets and sets the property MinInstanceCount.

Defines the minimum number of instances required for an instance group to become InService. If this threshold isn't met within 3 hours, the instance group rolls back to its previous state - zero instances for new instance groups, or previous settings for existing instance groups. MinInstanceCount only affects the initial transition to InService and does not guarantee maintaining this minimum afterward.

Gets and sets the property OnStartDeepHealthChecks.

A flag indicating whether deep health checks should be performed when the cluster instance group is created or updated.

Gets and sets the property OverrideVpcConfig.

To configure multi-AZ deployments, customize the Amazon VPC configuration at the instance group level. You can specify different subnets and security groups across different AZs in the instance group specification to override a SageMaker HyperPod cluster's default Amazon VPC configuration. For more information about deploying a cluster in multiple AZs, see Setting up SageMaker HyperPod clusters across multiple AZs.

Gets and sets the property ScheduledUpdateConfig.

The configuration object of the schedule that SageMaker uses to update the AMI.

Gets and sets the property ThreadsPerCore.

Specifies the value for Threads per core. For instance types that support multithreading, you can specify 1 for disabling multithreading and 2 for enabling multithreading. For instance types that doesn't support multithreading, specify 1. For more information, see the reference table of CPU cores and threads per CPU core per instance type in the Amazon Elastic Compute Cloud User Guide.

Gets and sets the property TrainingPlanArn.

The Amazon Resource Name (ARN); of the training plan to use for this cluster instance group.

For more information about how to reserve GPU capacity for your SageMaker HyperPod clusters using Amazon SageMaker Training Plan, see CreateTrainingPlan .