feat(client-sagemaker): Amazon Sagemaker supports orchestrating SageMaker HyperPod clusters with Amazon EKS

Author: awstools

clients/client-sagemaker/src/commands/CreateAppImageConfigCommand.ts

@@ -5,7 +5,7 @@ import { Command as $Command } from "@smithy/smithy-client";
 import { MetadataBearer as __MetadataBearer } from "@smithy/types";
 
 import { commonParams } from "../endpoint/EndpointParameters";
-import { CreateAppImageConfigRequest, CreateAppImageConfigResponse } from "../models/models_0";
+import { CreateAppImageConfigRequest, CreateAppImageConfigResponse } from "../models/models_1";
 import { de_CreateAppImageConfigCommand, se_CreateAppImageConfigCommand } from "../protocols/Aws_json1_1";
 import { SageMakerClientResolvedConfig, ServiceInputTypes, ServiceOutputTypes } from "../SageMakerClient";
 

clients/client-sagemaker/src/commands/CreateArtifactCommand.ts

@@ -5,8 +5,7 @@ import { Command as $Command } from "@smithy/smithy-client";
 import { MetadataBearer as __MetadataBearer } from "@smithy/types";
 
 import { commonParams } from "../endpoint/EndpointParameters";
-import { CreateArtifactRequest } from "../models/models_0";
-import { CreateArtifactResponse } from "../models/models_1";
+import { CreateArtifactRequest, CreateArtifactResponse } from "../models/models_1";
 import { de_CreateArtifactCommand, se_CreateArtifactCommand } from "../protocols/Aws_json1_1";
 import { SageMakerClientResolvedConfig, ServiceInputTypes, ServiceOutputTypes } from "../SageMakerClient";
 

clients/client-sagemaker/src/commands/CreateClusterCommand.ts

@@ -58,6 +58,9 @@ export interface CreateClusterCommandOutput extends CreateClusterResponse, __Met
  *           },
  *         },
  *       ],
+ *       OnStartDeepHealthChecks: [ // OnStartDeepHealthChecks
+ *         "InstanceStress" || "InstanceConnectivity",
+ *       ],
  *     },
  *   ],
  *   VpcConfig: { // VpcConfig
@@ -74,6 +77,12 @@ export interface CreateClusterCommandOutput extends CreateClusterResponse, __Met
  *       Value: "STRING_VALUE", // required
  *     },
  *   ],
+ *   Orchestrator: { // ClusterOrchestrator
+ *     Eks: { // ClusterOrchestratorEksConfig
+ *       ClusterArn: "STRING_VALUE", // required
+ *     },
+ *   },
+ *   NodeRecovery: "Automatic" || "None",
  * };
  * const command = new CreateClusterCommand(input);
  * const response = await client.send(command);

clients/client-sagemaker/src/commands/CreatePresignedMlflowTrackingServerUrlCommand.ts

@@ -5,10 +5,8 @@ import { Command as $Command } from "@smithy/smithy-client";
 import { MetadataBearer as __MetadataBearer } from "@smithy/types";
 
 import { commonParams } from "../endpoint/EndpointParameters";
-import {
-  CreatePresignedMlflowTrackingServerUrlRequest,
-  CreatePresignedMlflowTrackingServerUrlResponse,
-} from "../models/models_1";
+import { CreatePresignedMlflowTrackingServerUrlRequest } from "../models/models_1";
+import { CreatePresignedMlflowTrackingServerUrlResponse } from "../models/models_2";
 import {
   de_CreatePresignedMlflowTrackingServerUrlCommand,
   se_CreatePresignedMlflowTrackingServerUrlCommand,

clients/client-sagemaker/src/commands/CreatePresignedNotebookInstanceUrlCommand.ts

@@ -5,7 +5,7 @@ import { Command as $Command } from "@smithy/smithy-client";
 import { MetadataBearer as __MetadataBearer } from "@smithy/types";
 
 import { commonParams } from "../endpoint/EndpointParameters";
-import { CreatePresignedNotebookInstanceUrlInput, CreatePresignedNotebookInstanceUrlOutput } from "../models/models_1";
+import { CreatePresignedNotebookInstanceUrlInput, CreatePresignedNotebookInstanceUrlOutput } from "../models/models_2";
 import {
   de_CreatePresignedNotebookInstanceUrlCommand,
   se_CreatePresignedNotebookInstanceUrlCommand,

clients/client-sagemaker/src/commands/DescribeClusterCommand.ts

@@ -65,6 +65,9 @@ export interface DescribeClusterCommandOutput extends DescribeClusterResponse, _
  * //           },
  * //         },
  * //       ],
+ * //       OnStartDeepHealthChecks: [ // OnStartDeepHealthChecks
+ * //         "InstanceStress" || "InstanceConnectivity",
+ * //       ],
  * //     },
  * //   ],
  * //   VpcConfig: { // VpcConfig
@@ -75,6 +78,12 @@ export interface DescribeClusterCommandOutput extends DescribeClusterResponse, _
  * //       "STRING_VALUE",
  * //     ],
  * //   },
+ * //   Orchestrator: { // ClusterOrchestrator
+ * //     Eks: { // ClusterOrchestratorEksConfig
+ * //       ClusterArn: "STRING_VALUE", // required
+ * //     },
+ * //   },
+ * //   NodeRecovery: "Automatic" || "None",
  * // };
  *
  * ```

clients/client-sagemaker/src/commands/DescribeClusterNodeCommand.ts

@@ -47,7 +47,7 @@ export interface DescribeClusterNodeCommandOutput extends DescribeClusterNodeRes
  * //     InstanceGroupName: "STRING_VALUE",
  * //     InstanceId: "STRING_VALUE",
  * //     InstanceStatus: { // ClusterInstanceStatusDetails
- * //       Status: "Running" || "Failure" || "Pending" || "ShuttingDown" || "SystemUpdating", // required
+ * //       Status: "Running" || "Failure" || "Pending" || "ShuttingDown" || "SystemUpdating" || "DeepHealthCheckInProgress", // required
  * //       Message: "STRING_VALUE",
  * //     },
  * //     InstanceType: "ml.p4d.24xlarge" || "ml.p4de.24xlarge" || "ml.p5.48xlarge" || "ml.trn1.32xlarge" || "ml.trn1n.32xlarge" || "ml.g5.xlarge" || "ml.g5.2xlarge" || "ml.g5.4xlarge" || "ml.g5.8xlarge" || "ml.g5.12xlarge" || "ml.g5.16xlarge" || "ml.g5.24xlarge" || "ml.g5.48xlarge" || "ml.c5.large" || "ml.c5.xlarge" || "ml.c5.2xlarge" || "ml.c5.4xlarge" || "ml.c5.9xlarge" || "ml.c5.12xlarge" || "ml.c5.18xlarge" || "ml.c5.24xlarge" || "ml.c5n.large" || "ml.c5n.2xlarge" || "ml.c5n.4xlarge" || "ml.c5n.9xlarge" || "ml.c5n.18xlarge" || "ml.m5.large" || "ml.m5.xlarge" || "ml.m5.2xlarge" || "ml.m5.4xlarge" || "ml.m5.8xlarge" || "ml.m5.12xlarge" || "ml.m5.16xlarge" || "ml.m5.24xlarge" || "ml.t3.medium" || "ml.t3.large" || "ml.t3.xlarge" || "ml.t3.2xlarge",

clients/client-sagemaker/src/commands/DescribeMlflowTrackingServerCommand.ts

@@ -5,7 +5,7 @@ import { Command as $Command } from "@smithy/smithy-client";
 import { MetadataBearer as __MetadataBearer } from "@smithy/types";
 
 import { commonParams } from "../endpoint/EndpointParameters";
-import { DescribeMlflowTrackingServerRequest, DescribeMlflowTrackingServerResponse } from "../models/models_2";
+import { DescribeMlflowTrackingServerRequest, DescribeMlflowTrackingServerResponse } from "../models/models_3";
 import {
   de_DescribeMlflowTrackingServerCommand,
   se_DescribeMlflowTrackingServerCommand,

clients/client-sagemaker/src/commands/ListClusterNodesCommand.ts

@@ -57,7 +57,7 @@ export interface ListClusterNodesCommandOutput extends ListClusterNodesResponse,
  * //       InstanceType: "ml.p4d.24xlarge" || "ml.p4de.24xlarge" || "ml.p5.48xlarge" || "ml.trn1.32xlarge" || "ml.trn1n.32xlarge" || "ml.g5.xlarge" || "ml.g5.2xlarge" || "ml.g5.4xlarge" || "ml.g5.8xlarge" || "ml.g5.12xlarge" || "ml.g5.16xlarge" || "ml.g5.24xlarge" || "ml.g5.48xlarge" || "ml.c5.large" || "ml.c5.xlarge" || "ml.c5.2xlarge" || "ml.c5.4xlarge" || "ml.c5.9xlarge" || "ml.c5.12xlarge" || "ml.c5.18xlarge" || "ml.c5.24xlarge" || "ml.c5n.large" || "ml.c5n.2xlarge" || "ml.c5n.4xlarge" || "ml.c5n.9xlarge" || "ml.c5n.18xlarge" || "ml.m5.large" || "ml.m5.xlarge" || "ml.m5.2xlarge" || "ml.m5.4xlarge" || "ml.m5.8xlarge" || "ml.m5.12xlarge" || "ml.m5.16xlarge" || "ml.m5.24xlarge" || "ml.t3.medium" || "ml.t3.large" || "ml.t3.xlarge" || "ml.t3.2xlarge", // required
  * //       LaunchTime: new Date("TIMESTAMP"), // required
  * //       InstanceStatus: { // ClusterInstanceStatusDetails
- * //         Status: "Running" || "Failure" || "Pending" || "ShuttingDown" || "SystemUpdating", // required
+ * //         Status: "Running" || "Failure" || "Pending" || "ShuttingDown" || "SystemUpdating" || "DeepHealthCheckInProgress", // required
  * //         Message: "STRING_VALUE",
  * //       },
  * //     },

clients/client-sagemaker/src/commands/ListModelCardExportJobsCommand.ts

@@ -5,7 +5,7 @@ import { Command as $Command } from "@smithy/smithy-client";
 import { MetadataBearer as __MetadataBearer } from "@smithy/types";
 
 import { commonParams } from "../endpoint/EndpointParameters";
-import { ListModelCardExportJobsRequest, ListModelCardExportJobsResponse } from "../models/models_3";
+import { ListModelCardExportJobsRequest, ListModelCardExportJobsResponse } from "../models/models_4";
 import { de_ListModelCardExportJobsCommand, se_ListModelCardExportJobsCommand } from "../protocols/Aws_json1_1";
 import { SageMakerClientResolvedConfig, ServiceInputTypes, ServiceOutputTypes } from "../SageMakerClient";
 

clients/client-sagemaker/src/commands/UpdateClusterCommand.ts

@@ -55,8 +55,12 @@ export interface UpdateClusterCommandOutput extends UpdateClusterResponse, __Met
  *           },
  *         },
  *       ],
+ *       OnStartDeepHealthChecks: [ // OnStartDeepHealthChecks
+ *         "InstanceStress" || "InstanceConnectivity",
+ *       ],
  *     },
  *   ],
+ *   NodeRecovery: "Automatic" || "None",
  * };
  * const command = new UpdateClusterCommand(input);
  * const response = await client.send(command);

clients/client-sagemaker/src/models/models_0.ts

@@ -9280,6 +9280,20 @@ export interface ClusterLifeCycleConfig {
   OnCreate: string | undefined;
 }
 
+/**
+ * @public
+ * @enum
+ */
+export const DeepHealthCheckType = {
+  INSTANCE_CONNECTIVITY: "InstanceConnectivity",
+  INSTANCE_STRESS: "InstanceStress",
+} as const;
+
+/**
+ * @public
+ */
+export type DeepHealthCheckType = (typeof DeepHealthCheckType)[keyof typeof DeepHealthCheckType];
+
 /**
  * <p>Details of an instance group in a SageMaker HyperPod cluster.</p>
  * @public
@@ -9340,6 +9354,12 @@ export interface ClusterInstanceGroupDetails {
    * @public
    */
   InstanceStorageConfigs?: ClusterInstanceStorageConfig[];
+
+  /**
+   * <p>A flag indicating whether deep health checks should be performed when the cluster instance group is created or updated.</p>
+   * @public
+   */
+  OnStartDeepHealthChecks?: DeepHealthCheckType[];
 }
 
 /**
@@ -9396,6 +9416,12 @@ export interface ClusterInstanceGroupSpecification {
    * @public
    */
   InstanceStorageConfigs?: ClusterInstanceStorageConfig[];
+
+  /**
+   * <p>A flag indicating whether deep health checks should be performed when the cluster instance group is created or updated.</p>
+   * @public
+   */
+  OnStartDeepHealthChecks?: DeepHealthCheckType[];
 }
 
 /**
@@ -9423,6 +9449,7 @@ export interface ClusterInstancePlacement {
  * @enum
  */
 export const ClusterInstanceStatus = {
+  DEEP_HEALTH_CHECK_IN_PROGRESS: "DeepHealthCheckInProgress",
   FAILURE: "Failure",
   PENDING: "Pending",
   RUNNING: "Running",
@@ -9528,6 +9555,20 @@ export interface ClusterNodeDetails {
   Placement?: ClusterInstancePlacement;
 }
 
+/**
+ * @public
+ * @enum
+ */
+export const ClusterNodeRecovery = {
+  AUTOMATIC: "Automatic",
+  NONE: "None",
+} as const;
+
+/**
+ * @public
+ */
+export type ClusterNodeRecovery = (typeof ClusterNodeRecovery)[keyof typeof ClusterNodeRecovery];
+
 /**
  * <p>Lists a summary of the properties of an instance (also called a
  *             <i>node</i> interchangeably) of a SageMaker HyperPod cluster.</p>
@@ -9565,6 +9606,30 @@ export interface ClusterNodeSummary {
   InstanceStatus: ClusterInstanceStatusDetails | undefined;
 }
 
+/**
+ * <p>The configuration settings for the Amazon EKS cluster used as the orchestrator for the SageMaker HyperPod cluster.</p>
+ * @public
+ */
+export interface ClusterOrchestratorEksConfig {
+  /**
+   * <p>The Amazon Resource Name (ARN) of the Amazon EKS cluster associated with the SageMaker HyperPod cluster.</p>
+   * @public
+   */
+  ClusterArn: string | undefined;
+}
+
+/**
+ * <p>The type of orchestrator used for the SageMaker HyperPod cluster.</p>
+ * @public
+ */
+export interface ClusterOrchestrator {
+  /**
+   * <p>The Amazon EKS cluster used as the orchestrator for the SageMaker HyperPod cluster.</p>
+   * @public
+   */
+  Eks: ClusterOrchestratorEksConfig | undefined;
+}
+
 /**
  * @public
  * @enum
@@ -9682,7 +9747,6 @@ export interface CodeEditorAppSettings {
 
   /**
    * <p>Settings that are used to configure and manage the lifecycle of CodeEditor applications.</p>
-   *          <p></p>
    * @public
    */
   AppLifecycleManagement?: AppLifecycleManagement;
@@ -11122,116 +11186,3 @@ export interface CreateAppResponse {
    */
   AppArn?: string;
 }
-
-/**
- * <p>Resource being accessed is in use.</p>
- * @public
- */
-export class ResourceInUse extends __BaseException {
-  readonly name: "ResourceInUse" = "ResourceInUse";
-  readonly $fault: "client" = "client";
-  Message?: string;
-  /**
-   * @internal
-   */
-  constructor(opts: __ExceptionOptionType<ResourceInUse, __BaseException>) {
-    super({
-      name: "ResourceInUse",
-      $fault: "client",
-      ...opts,
-    });
-    Object.setPrototypeOf(this, ResourceInUse.prototype);
-    this.Message = opts.Message;
-  }
-}
-
-/**
- * @public
- */
-export interface CreateAppImageConfigRequest {
-  /**
-   * <p>The name of the AppImageConfig. Must be unique to your account.</p>
-   * @public
-   */
-  AppImageConfigName: string | undefined;
-
-  /**
-   * <p>A list of tags to apply to the AppImageConfig.</p>
-   * @public
-   */
-  Tags?: Tag[];
-
-  /**
-   * <p>The KernelGatewayImageConfig. You can only specify one image kernel in the
-   *          AppImageConfig API. This kernel will be shown to users before the
-   *          image starts. Once the image runs, all kernels are visible in JupyterLab.</p>
-   * @public
-   */
-  KernelGatewayImageConfig?: KernelGatewayImageConfig;
-
-  /**
-   * <p>The <code>JupyterLabAppImageConfig</code>. You can only specify one image kernel in the <code>AppImageConfig</code> API. This kernel is shown to users before the image starts. After the image runs, all kernels are visible in JupyterLab.</p>
-   * @public
-   */
-  JupyterLabAppImageConfig?: JupyterLabAppImageConfig;
-
-  /**
-   * <p>The <code>CodeEditorAppImageConfig</code>. You can only specify one image kernel
-   *       in the AppImageConfig API. This kernel is shown to users before the image starts.
-   *       After the image runs, all kernels are visible in Code Editor.</p>
-   * @public
-   */
-  CodeEditorAppImageConfig?: CodeEditorAppImageConfig;
-}
-
-/**
- * @public
- */
-export interface CreateAppImageConfigResponse {
-  /**
-   * <p>The ARN of the AppImageConfig.</p>
-   * @public
-   */
-  AppImageConfigArn?: string;
-}
-
-/**
- * @public
- */
-export interface CreateArtifactRequest {
-  /**
-   * <p>The name of the artifact. Must be unique to your account in an Amazon Web Services Region.</p>
-   * @public
-   */
-  ArtifactName?: string;
-
-  /**
-   * <p>The ID, ID type, and URI of the source.</p>
-   * @public
-   */
-  Source: ArtifactSource | undefined;
-
-  /**
-   * <p>The artifact type.</p>
-   * @public
-   */
-  ArtifactType: string | undefined;
-
-  /**
-   * <p>A list of properties to add to the artifact.</p>
-   * @public
-   */
-  Properties?: Record<string, string>;
-
-  /**
-   * <p>Metadata properties of the tracking entity, trial, or trial component.</p>
-   * @public
-   */
-  MetadataProperties?: MetadataProperties;
-
-  /**
-   * <p>A list of tags to apply to the artifact.</p>
-   * @public
-   */
-  Tags?: Tag[];
-}