feat(client-sagemaker): This release adds support for rolling deploym…

…ent in SageMaker Inference.
aws · Jun 30, 2023 · 83d2ff0 · 83d2ff0
1 parent 73ae32e
commit 83d2ff0
Show file tree

Hide file tree

Showing 14 changed files with 276 additions and 140 deletions.
diff --git a/clients/client-sagemaker/src/commands/CreateEndpointCommand.ts b/clients/client-sagemaker/src/commands/CreateEndpointCommand.ts
@@ -147,6 +147,18 @@ export interface CreateEndpointCommandOutput extends CreateEndpointOutput, __Met
  *         },
  *       ],
  *     },
+ *     RollingUpdatePolicy: { // RollingUpdatePolicy
+ *       MaximumBatchSize: {
+ *         Type: "INSTANCE_COUNT" || "CAPACITY_PERCENT", // required
+ *         Value: Number("int"), // required
+ *       },
+ *       WaitIntervalInSeconds: Number("int"), // required
+ *       MaximumExecutionTimeoutInSeconds: Number("int"),
+ *       RollbackMaximumBatchSize: {
+ *         Type: "INSTANCE_COUNT" || "CAPACITY_PERCENT", // required
+ *         Value: Number("int"), // required
+ *       },
+ *     },
  *   },
  *   Tags: [ // TagList
  *     { // Tag

diff --git a/clients/client-sagemaker/src/commands/DeleteImageCommand.ts b/clients/client-sagemaker/src/commands/DeleteImageCommand.ts
@@ -13,8 +13,7 @@ import {
   SerdeContext as __SerdeContext,
 } from "@smithy/types";
 
-import { DeleteImageRequest } from "../models/models_1";
-import { DeleteImageResponse } from "../models/models_2";
+import { DeleteImageRequest, DeleteImageResponse } from "../models/models_2";
 import { de_DeleteImageCommand, se_DeleteImageCommand } from "../protocols/Aws_json1_1";
 import { SageMakerClientResolvedConfig, ServiceInputTypes, ServiceOutputTypes } from "../SageMakerClient";
 

diff --git a/clients/client-sagemaker/src/commands/DescribeEndpointCommand.ts b/clients/client-sagemaker/src/commands/DescribeEndpointCommand.ts
@@ -92,7 +92,7 @@ export interface DescribeEndpointCommandOutput extends DescribeEndpointOutput, _
  * //     DestinationS3Uri: "STRING_VALUE", // required
  * //     KmsKeyId: "STRING_VALUE", // required
  * //   },
- * //   EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed", // required
+ * //   EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed" || "UpdateRollbackFailed", // required
  * //   FailureReason: "STRING_VALUE",
  * //   CreationTime: new Date("TIMESTAMP"), // required
  * //   LastModifiedTime: new Date("TIMESTAMP"), // required
@@ -120,6 +120,18 @@ export interface DescribeEndpointCommandOutput extends DescribeEndpointOutput, _
  * //         },
  * //       ],
  * //     },
+ * //     RollingUpdatePolicy: { // RollingUpdatePolicy
+ * //       MaximumBatchSize: {
+ * //         Type: "INSTANCE_COUNT" || "CAPACITY_PERCENT", // required
+ * //         Value: Number("int"), // required
+ * //       },
+ * //       WaitIntervalInSeconds: Number("int"), // required
+ * //       MaximumExecutionTimeoutInSeconds: Number("int"),
+ * //       RollbackMaximumBatchSize: {
+ * //         Type: "INSTANCE_COUNT" || "CAPACITY_PERCENT", // required
+ * //         Value: Number("int"), // required
+ * //       },
+ * //     },
  * //   },
  * //   AsyncInferenceConfig: { // AsyncInferenceConfig
  * //     ClientConfig: { // AsyncInferenceClientConfig

diff --git a/clients/client-sagemaker/src/commands/DescribeInferenceExperimentCommand.ts b/clients/client-sagemaker/src/commands/DescribeInferenceExperimentCommand.ts
@@ -68,7 +68,7 @@ export interface DescribeInferenceExperimentCommandOutput
  * //   EndpointMetadata: { // EndpointMetadata
  * //     EndpointName: "STRING_VALUE", // required
  * //     EndpointConfigName: "STRING_VALUE",
- * //     EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed",
+ * //     EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed" || "UpdateRollbackFailed",
  * //     FailureReason: "STRING_VALUE",
  * //   },
  * //   ModelVariants: [ // ModelVariantConfigSummaryList // required

diff --git a/clients/client-sagemaker/src/commands/ListEndpointsCommand.ts b/clients/client-sagemaker/src/commands/ListEndpointsCommand.ts
@@ -53,7 +53,7 @@ export interface ListEndpointsCommandOutput extends ListEndpointsOutput, __Metad
  *   CreationTimeAfter: new Date("TIMESTAMP"),
  *   LastModifiedTimeBefore: new Date("TIMESTAMP"),
  *   LastModifiedTimeAfter: new Date("TIMESTAMP"),
- *   StatusEquals: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed",
+ *   StatusEquals: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed" || "UpdateRollbackFailed",
  * };
  * const command = new ListEndpointsCommand(input);
  * const response = await client.send(command);
@@ -64,7 +64,7 @@ export interface ListEndpointsCommandOutput extends ListEndpointsOutput, __Metad
  * //       EndpointArn: "STRING_VALUE", // required
  * //       CreationTime: new Date("TIMESTAMP"), // required
  * //       LastModifiedTime: new Date("TIMESTAMP"), // required
- * //       EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed", // required
+ * //       EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed" || "UpdateRollbackFailed", // required
  * //     },
  * //   ],
  * //   NextToken: "STRING_VALUE",

diff --git a/clients/client-sagemaker/src/commands/SearchCommand.ts b/clients/client-sagemaker/src/commands/SearchCommand.ts
@@ -870,7 +870,7 @@ export interface SearchCommandOutput extends SearchResponse, __MetadataBearer {}
  * //           DestinationS3Uri: "STRING_VALUE", // required
  * //           KmsKeyId: "STRING_VALUE", // required
  * //         },
- * //         EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed", // required
+ * //         EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed" || "UpdateRollbackFailed", // required
  * //         FailureReason: "STRING_VALUE",
  * //         CreationTime: new Date("TIMESTAMP"), // required
  * //         LastModifiedTime: new Date("TIMESTAMP"), // required
@@ -1772,7 +1772,7 @@ export interface SearchCommandOutput extends SearchResponse, __MetadataBearer {}
  * //             EndpointArn: "STRING_VALUE", // required
  * //             CreationTime: new Date("TIMESTAMP"), // required
  * //             LastModifiedTime: new Date("TIMESTAMP"), // required
- * //             EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed", // required
+ * //             EndpointStatus: "OutOfService" || "Creating" || "Updating" || "SystemUpdating" || "RollingBack" || "InService" || "Deleting" || "Failed" || "UpdateRollbackFailed", // required
  * //           },
  * //         ],
  * //         LastBatchTransformJob: {

diff --git a/clients/client-sagemaker/src/commands/UpdateEndpointCommand.ts b/clients/client-sagemaker/src/commands/UpdateEndpointCommand.ts
@@ -92,6 +92,18 @@ export interface UpdateEndpointCommandOutput extends UpdateEndpointOutput, __Met
  *         },
  *       ],
  *     },
+ *     RollingUpdatePolicy: { // RollingUpdatePolicy
+ *       MaximumBatchSize: {
+ *         Type: "INSTANCE_COUNT" || "CAPACITY_PERCENT", // required
+ *         Value: Number("int"), // required
+ *       },
+ *       WaitIntervalInSeconds: Number("int"), // required
+ *       MaximumExecutionTimeoutInSeconds: Number("int"),
+ *       RollbackMaximumBatchSize: {
+ *         Type: "INSTANCE_COUNT" || "CAPACITY_PERCENT", // required
+ *         Value: Number("int"), // required
+ *       },
+ *     },
  *   },
  *   RetainDeploymentConfig: true || false,
  * };

diff --git a/clients/client-sagemaker/src/models/models_0.ts b/clients/client-sagemaker/src/models/models_0.ts
@@ -6585,7 +6585,10 @@ export type CapacitySizeType = (typeof CapacitySizeType)[keyof typeof CapacitySi
 
 /**
  * @public
- * <p>Specifies the endpoint capacity to activate for production.</p>
+ * <p>Specifies the type and size of the endpoint capacity to activate for a blue/green deployment, a rolling deployment, or a rollback strategy.
+ *         You can specify your batches as either instance count or the overall percentage or your fleet.</p>
+ *          <p>For a rollback strategy, if you don't specify the fields in this object, or if you set the <code>Value</code> to 100%, then SageMaker
+ *         uses a blue/green rollback strategy and rolls all traffic back to the blue fleet.</p>
  */
 export interface CapacitySize {
   /**

diff --git a/clients/client-sagemaker/src/models/models_1.ts b/clients/client-sagemaker/src/models/models_1.ts
@@ -24,6 +24,7 @@ import {
   Bias,
   BlueGreenUpdatePolicy,
   CanvasAppSettings,
+  CapacitySize,
   CaptureContentTypeHeader,
   CaptureOption,
   CaptureStatus,
@@ -598,6 +599,37 @@ export interface CreateEdgePackagingJobRequest {
   Tags?: Tag[];
 }
 
+/**
+ * @public
+ * <p>Specifies a rolling deployment strategy for updating a SageMaker endpoint.</p>
+ */
+export interface RollingUpdatePolicy {
+  /**
+   * <p>Batch size for each rolling step to provision capacity and turn on traffic on the new
+   *             endpoint fleet, and terminate capacity on the old endpoint fleet. Value must be between
+   *             5% to 50% of the variant's total instance count.</p>
+   */
+  MaximumBatchSize: CapacitySize | undefined;
+
+  /**
+   * <p>The length of the baking period, during which SageMaker monitors alarms for each batch on the new fleet.</p>
+   */
+  WaitIntervalInSeconds: number | undefined;
+
+  /**
+   * <p>The time limit for the total deployment. Exceeding this limit causes a timeout.</p>
+   */
+  MaximumExecutionTimeoutInSeconds?: number;
+
+  /**
+   * <p>Batch size for rollback to the old endpoint fleet. Each rolling step to provision
+   *             capacity and turn on traffic on the old endpoint fleet, and terminate capacity on the new
+   *             endpoint fleet. If this field is absent, the default value will be set to 100% of total
+   *             capacity which means to bring up the whole capacity of the old fleet at once during rollback.</p>
+   */
+  RollbackMaximumBatchSize?: CapacitySize;
+}
+
 /**
  * @public
  * <p>The deployment configuration for an endpoint, which contains the desired deployment
@@ -612,13 +644,18 @@ export interface DeploymentConfig {
    *             specified, SageMaker uses a blue/green deployment strategy with all at once traffic shifting
    *             by default.</p>
    */
-  BlueGreenUpdatePolicy: BlueGreenUpdatePolicy | undefined;
+  BlueGreenUpdatePolicy?: BlueGreenUpdatePolicy;
 
   /**
    * <p>Automatic rollback configuration for handling endpoint deployment failures and
    *             recovery.</p>
    */
   AutoRollbackConfiguration?: AutoRollbackConfig;
+
+  /**
+   * <p>Specifies a rolling deployment strategy for updating a SageMaker endpoint.</p>
+   */
+  RollingUpdatePolicy?: RollingUpdatePolicy;
 }
 
 /**
@@ -10754,16 +10791,6 @@ export interface DeleteHumanTaskUiRequest {
  */
 export interface DeleteHumanTaskUiResponse {}
 
-/**
- * @public
- */
-export interface DeleteImageRequest {
-  /**
-   * <p>The name of the image to delete.</p>
-   */
-  ImageName: string | undefined;
-}
-
 /**
  * @internal
  */

diff --git a/clients/client-sagemaker/src/models/models_2.ts b/clients/client-sagemaker/src/models/models_2.ts
@@ -177,6 +177,16 @@ import {
   VendorGuidance,
 } from "./models_1";
 
+/**
+ * @public
+ */
+export interface DeleteImageRequest {
+  /**
+   * <p>The name of the image to delete.</p>
+   */
+  ImageName: string | undefined;
+}
+
 /**
  * @public
  */
@@ -2256,6 +2266,7 @@ export const EndpointStatus = {
   OUT_OF_SERVICE: "OutOfService",
   ROLLING_BACK: "RollingBack",
   SYSTEM_UPDATING: "SystemUpdating",
+  UPDATE_ROLLBACK_FAILED: "UpdateRollbackFailed",
   UPDATING: "Updating",
 } as const;
 
@@ -10094,19 +10105,6 @@ export interface PropertyNameQuery {
   PropertyNameHint: string | undefined;
 }
 
-/**
- * @public
- * <p>Specified in the <a href="https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_GetSearchSuggestions.html">GetSearchSuggestions</a> request.
- *       Limits the property names that are included in the response.</p>
- */
-export interface SuggestionQuery {
-  /**
-   * <p>Defines a property name hint. Only property
-   *       names that begin with the specified hint are included in the response.</p>
-   */
-  PropertyNameQuery?: PropertyNameQuery;
-}
-
 /**
  * @internal
  */