Feature: Serverless Endpoint (#193)

Issue #, if available:

Description of changes:

This pull request contains code/unit tests for the Serverless Endpoint.
Changes made:
`generator.yaml` - Uncommenting shape and adding hook.
`pkg/resource/endpoint_config/custom_delta.go` - Code that handles server side default. Late initialization isnt possible because this field is an array.
`testdata/` - Unit tests

The Server returns a value for VolumeSizeInGB despite it being a parameter the user cannot even specify while using a serverless endpoint. The same problem is not there for an instance based endpoint config.

Manual testing (Will delete this before merge):

Endpoint Config:

[Serverless] Async Conifg - Validation Exception
[Serverless] Multi Variant - Validation Exception
[Serverless] Data Capture - Validation Exception
[Serverless] Including VolumeSizeInGB in spec: Validation exception
[Serverless] No VolumeSizeInGB: Normal reconciliation, no errors after restart.
[Normal] No VolumeSizeInGB - Normal reconciliation, no errors after restart.
[Normal] VolumeSizeInGB - Normal reconciliation, no errors after restart.

Endpoint:
Launching Serverless: Works, user sees information in PendingDeploymentSummary
Updating Instance→ Serverless: Validation Exception
Updating Serverless → Instance: Production variant must have a different name, works.


By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

Author: ananth102

apis/v1alpha1/ack-generate-metadata.yaml

@@ -1,13 +1,13 @@
 ack_generate_info:
-  build_date: "2023-01-04T22:46:39Z"
-  build_hash: 12246c7da82841b351ec7a9e1f139f9338f2784b
+  build_date: "2023-01-19T03:37:00Z"
+  build_hash: 1b20baf45a0b73a11b296050322a384c705fa897
   go_version: go1.17.13
-  version: v0.21.0
-api_directory_checksum: 19b63f2ee6b7e14b2f6d254a6caff0b9867539e5
+  version: v0.22.0
+api_directory_checksum: c38df824c3c6e41cd4eaa559dccc4c05c7e8d075
 api_version: v1alpha1
 aws_sdk_go_version: v1.44.117
 generator_config_info:
-  file_checksum: fd845048bf7dc84ecd3ea1ed1784b831d8f42645
+  file_checksum: e3688a5487a10dccd7a5e488de025d9489043f7d
   original_file_name: generator.yaml
 last_modification:
   reason: API generation

apis/v1alpha1/generator.yaml

@@ -57,6 +57,9 @@ resources:
       Tags:
         compare:
           is_ignored: true
+    hooks:
+      delta_pre_compare:
+        code: customSetDefaults(a, b)
   Endpoint:
     reconcile: 
       requeue_on_success_seconds: 30
@@ -1017,7 +1020,6 @@ ignore:
     shape_names:
       # RSessionAppSettings is an empty struct that causes generation errors
       - RSessionAppSettings
-      - ProductionVariantServerlessConfig
       - ExecutionRoleIdentityConfig
       - HyperParameterTuningResourceConfig
       - InstanceMetadataServiceConfiguration

apis/v1alpha1/types.go

@@ -167,6 +167,17 @@ spec:
                       type: integer
                     modelName:
                       type: string
+                    serverlessConfig:
+                      description: Specifies the serverless configuration for an endpoint
+                        variant.
+                      properties:
+                        maxConcurrency:
+                          format: int64
+                          type: integer
+                        memorySizeInMB:
+                          format: int64
+                          type: integer
+                      type: object
                     variantName:
                       type: string
                     volumeSizeInGB:

apis/v1alpha1/zz_generated.deepcopy.go

@@ -268,6 +268,17 @@ spec:
                         currentInstanceCount:
                           format: int64
                           type: integer
+                        currentServerlessConfig:
+                          description: Specifies the serverless configuration for
+                            an endpoint variant.
+                          properties:
+                            maxConcurrency:
+                              format: int64
+                              type: integer
+                            memorySizeInMB:
+                              format: int64
+                              type: integer
+                          type: object
                         currentWeight:
                           type: number
                         deployedImages:
@@ -295,6 +306,17 @@ spec:
                         desiredInstanceCount:
                           format: int64
                           type: integer
+                        desiredServerlessConfig:
+                          description: Specifies the serverless configuration for
+                            an endpoint variant.
+                          properties:
+                            maxConcurrency:
+                              format: int64
+                              type: integer
+                            memorySizeInMB:
+                              format: int64
+                              type: integer
+                          type: object
                         desiredWeight:
                           type: number
                         instanceType:
@@ -332,6 +354,17 @@ spec:
                     currentInstanceCount:
                       format: int64
                       type: integer
+                    currentServerlessConfig:
+                      description: Specifies the serverless configuration for an endpoint
+                        variant.
+                      properties:
+                        maxConcurrency:
+                          format: int64
+                          type: integer
+                        memorySizeInMB:
+                          format: int64
+                          type: integer
+                      type: object
                     currentWeight:
                       type: number
                     deployedImages:
@@ -359,6 +392,17 @@ spec:
                     desiredInstanceCount:
                       format: int64
                       type: integer
+                    desiredServerlessConfig:
+                      description: Specifies the serverless configuration for an endpoint
+                        variant.
+                      properties:
+                        maxConcurrency:
+                          format: int64
+                          type: integer
+                        memorySizeInMB:
+                          format: int64
+                          type: integer
+                      type: object
                     desiredWeight:
                       type: number
                     variantName:

config/crd/bases/sagemaker.services.k8s.aws_endpointconfigs.yaml

@@ -57,6 +57,9 @@ resources:
       Tags:
         compare:
           is_ignored: true
+    hooks:
+      delta_pre_compare:
+        code: customSetDefaults(a, b)
   Endpoint:
     reconcile: 
       requeue_on_success_seconds: 30
@@ -1017,7 +1020,6 @@ ignore:
     shape_names:
       # RSessionAppSettings is an empty struct that causes generation errors
       - RSessionAppSettings
-      - ProductionVariantServerlessConfig
       - ExecutionRoleIdentityConfig
       - HyperParameterTuningResourceConfig
       - InstanceMetadataServiceConfiguration

config/crd/bases/sagemaker.services.k8s.aws_endpoints.yaml

@@ -167,6 +167,17 @@ spec:
                       type: integer
                     modelName:
                       type: string
+                    serverlessConfig:
+                      description: Specifies the serverless configuration for an endpoint
+                        variant.
+                      properties:
+                        maxConcurrency:
+                          format: int64
+                          type: integer
+                        memorySizeInMB:
+                          format: int64
+                          type: integer
+                      type: object
                     variantName:
                       type: string
                     volumeSizeInGB: