feat(glue): support Ray jobs

packages/@aws-cdk/aws-glue/README.md

@@ -90,6 +90,23 @@ new glue.Job(this, 'PythonShellJob', {
 });
 ```
 
+### Ray Jobs
+
+These jobs run in a Ray environment managed by AWS Glue.
+
+```ts
+new glue.Job(this, 'RayJob', {
+  executable: glue.JobExecutable.pythonRay({
+    glueVersion: glue.GlueVersion.V4_0,
+    pythonVersion: glue.PythonVersion.THREE_NINE,
+    script: glue.Code.fromAsset(path.join(__dirname, 'job-script/hello_world.py')),
+  }),
+  workerType: glue.WorkerType.Z_2X,
+  workerCount: 2,
+  description: 'an example Ray job'
+});
+```
+
 See [documentation](https://docs.aws.amazon.com/glue/latest/dg/add-job.html) for more information on adding jobs in Glue.
 
 ## Connection

packages/@aws-cdk/aws-glue/lib/job-executable.ts

@@ -95,12 +95,12 @@ export enum PythonVersion {
  */
 export class JobType {
   /**
-   * Command for running a Glue ETL job.
+   * Command for running a Glue Spark job.
    */
   public static readonly ETL = new JobType('glueetl');
 
   /**
-   * Command for running a Glue streaming job.
+   * Command for running a Glue Spark streaming job.
    */
   public static readonly STREAMING = new JobType('gluestreaming');
 
@@ -109,6 +109,11 @@ export class JobType {
    */
   public static readonly PYTHON_SHELL = new JobType('pythonshell');
 
+  /**
+   * Command for running a Glue Ray job.
+   */
+  public static readonly RAY = new JobType('glueray');
+
   /**
    * Custom type name
    * @param name type name
@@ -211,6 +216,11 @@ export interface PythonSparkJobExecutableProps extends SharedSparkJobExecutableP
  */
 export interface PythonShellExecutableProps extends SharedJobExecutableProps, PythonExecutableProps {}
 
+/**
+ * Props for creating a Python Ray job executable
+ */
+export interface PythonRayExecutableProps extends SharedJobExecutableProps, PythonExecutableProps {}
+
 /**
  * The executable properties related to the Glue job's GlueVersion, JobType and code
  */
@@ -281,6 +291,19 @@ export class JobExecutable {
     });
   }
 
+  /**
+   * Create Python executable props for Ray jobs.
+   *
+   * @param props Ray Job props.
+   */
+  public static pythonRay(props: PythonRayExecutableProps): JobExecutable {
+    return new JobExecutable({
+      ...props,
+      type: JobType.RAY,
+      language: JobLanguage.PYTHON,
+    });
+  }
+
   /**
    * Create a custom JobExecutable.
    *
@@ -297,10 +320,18 @@ export class JobExecutable {
       if (config.language !== JobLanguage.PYTHON) {
         throw new Error('Python shell requires the language to be set to Python');
       }
-      if ([GlueVersion.V0_9, GlueVersion.V2_0, GlueVersion.V3_0, GlueVersion.V4_0].includes(config.glueVersion)) {
+      if ([GlueVersion.V0_9, GlueVersion.V3_0, GlueVersion.V4_0].includes(config.glueVersion)) {
         throw new Error(`Specified GlueVersion ${config.glueVersion.name} does not support Python Shell`);
       }
     }
+    if (JobType.RAY === config.type) {
+      if (config.language !== JobLanguage.PYTHON) {
+        throw new Error('Ray requires the language to be set to Python');
+      }
+      if ([GlueVersion.V0_9, GlueVersion.V1_0, GlueVersion.V2_0, GlueVersion.V3_0].includes(config.glueVersion)) {
+        throw new Error(`Specified GlueVersion ${config.glueVersion.name} does not support Ray`);
+      }
+    }
     if (config.extraJarsFirst && [GlueVersion.V0_9, GlueVersion.V1_0].includes(config.glueVersion)) {
       throw new Error(`Specified GlueVersion ${config.glueVersion.name} does not support extraJarsFirst`);
     }
@@ -310,8 +341,11 @@ export class JobExecutable {
     if (JobLanguage.PYTHON !== config.language && config.extraPythonFiles) {
       throw new Error('extraPythonFiles is not supported for languages other than JobLanguage.PYTHON');
     }
-    if (config.pythonVersion === PythonVersion.THREE_NINE && config.type !== JobType.PYTHON_SHELL) {
-      throw new Error('Specified PythonVersion PythonVersion.THREE_NINE is only supported for JobType Python Shell');
+    if (config.pythonVersion === PythonVersion.THREE_NINE && config.type !== JobType.PYTHON_SHELL && config.type !== JobType.RAY) {
+      throw new Error('Specified PythonVersion PythonVersion.THREE_NINE is only supported for JobType Python Shell and Ray');
+    }
+    if (config.pythonVersion === PythonVersion.THREE && config.type === JobType.RAY) {
+      throw new Error('Specified PythonVersion PythonVersion.THREE is not supported for Ray');
     }
     this.config = config;
   }

packages/@aws-cdk/aws-glue/lib/job.ts

@@ -32,6 +32,16 @@ export class WorkerType {
    */
   public static readonly G_2X = new WorkerType('G.2X');
 
+  /**
+   * Each worker maps to 0.25 DPU (2 vCPU, 4 GB of memory, 64 GB disk), and provides 1 executor per worker. Suitable for low volume streaming jobs.
+   */
+  public static readonly G_025X = new WorkerType('G.025X');
+
+  /**
+   * Each worker maps to 2 high-memory DPU [M-DPU] (8 vCPU, 64 GB of memory, 128 GB disk). Supported in Ray jobs.
+   */
+  public static readonly Z_2X = new WorkerType('Z.2X');
+
   /**
    * Custom worker type
    * @param workerType custom worker type
@@ -726,6 +736,8 @@ export class Job extends JobBase {
   private setupSparkUI(executable: JobExecutableConfig, role: iam.IRole, props: SparkUIProps) {
     if (JobType.PYTHON_SHELL === executable.type) {
       throw new Error('Spark UI is not available for JobType.PYTHON_SHELL jobs');
+    } else if (JobType.RAY === executable.type) {
+      throw new Error('Spark UI is not available for JobType.RAY jobs');
     }
 
     const bucket = props.bucket ?? new s3.Bucket(this, 'SparkUIBucket');

packages/@aws-cdk/aws-glue/test/integ.job.js.snapshot/aws-glue-job.assets.json

@@ -14,15 +14,15 @@
         }
       }
     },
-    "977a2f07e22679bb04b03ce83cc1fac3e6cc269a794e38248ec67106ee39f0a2": {
+    "b553fef631f82898c826f3c20e1de0d155dbd3a35339ef92d0893052a5be69ce": {
       "source": {
         "path": "aws-glue-job.template.json",
         "packaging": "file"
       },
       "destinations": {
         "current_account-current_region": {
           "bucketName": "cdk-hnb659fds-assets-${AWS::AccountId}-${AWS::Region}",
-          "objectKey": "977a2f07e22679bb04b03ce83cc1fac3e6cc269a794e38248ec67106ee39f0a2.json",
+          "objectKey": "b553fef631f82898c826f3c20e1de0d155dbd3a35339ef92d0893052a5be69ce.json",
           "assumeRoleArn": "arn:${AWS::Partition}:iam::${AWS::AccountId}:role/cdk-hnb659fds-file-publishing-role-${AWS::AccountId}-${AWS::Region}"
         }
       }

packages/@aws-cdk/aws-glue/test/integ.job.js.snapshot/aws-glue-job.template.json

@@ -350,9 +350,11 @@
     },
     "GlueVersion": "2.0",
     "Name": "StreamingJob2.0",
+    "NumberOfWorkers": 10,
     "Tags": {
      "key": "value"
-    }
+    },
+    "WorkerType": "G.025X"
    }
   },
   "EtlJob30ServiceRole8E675579": {
@@ -705,9 +707,11 @@
     },
     "GlueVersion": "3.0",
     "Name": "StreamingJob3.0",
+    "NumberOfWorkers": 10,
     "Tags": {
      "key": "value"
-    }
+    },
+    "WorkerType": "G.025X"
    }
   },
   "EtlJob40ServiceRoleBDD9998A": {
@@ -1060,9 +1064,11 @@
     },
     "GlueVersion": "4.0",
     "Name": "StreamingJob4.0",
+    "NumberOfWorkers": 10,
     "Tags": {
      "key": "value"
-    }
+    },
+    "WorkerType": "G.025X"
    }
   },
   "ShellJobServiceRoleCF97BC4B": {
@@ -1314,6 +1320,133 @@
      "key": "value"
     }
    }
+  },
+  "RayJobServiceRole51433C3D": {
+   "Type": "AWS::IAM::Role",
+   "Properties": {
+    "AssumeRolePolicyDocument": {
+     "Statement": [
+      {
+       "Action": "sts:AssumeRole",
+       "Effect": "Allow",
+       "Principal": {
+        "Service": "glue.amazonaws.com"
+       }
+      }
+     ],
+     "Version": "2012-10-17"
+    },
+    "ManagedPolicyArns": [
+     {
+      "Fn::Join": [
+       "",
+       [
+        "arn:",
+        {
+         "Ref": "AWS::Partition"
+        },
+        ":iam::aws:policy/service-role/AWSGlueServiceRole"
+       ]
+      ]
+     }
+    ]
+   }
+  },
+  "RayJobServiceRoleDefaultPolicyA615640D": {
+   "Type": "AWS::IAM::Policy",
+   "Properties": {
+    "PolicyDocument": {
+     "Statement": [
+      {
+       "Action": [
+        "s3:GetBucket*",
+        "s3:GetObject*",
+        "s3:List*"
+       ],
+       "Effect": "Allow",
+       "Resource": [
+        {
+         "Fn::Join": [
+          "",
+          [
+           "arn:",
+           {
+            "Ref": "AWS::Partition"
+           },
+           ":s3:::",
+           {
+            "Fn::Sub": "cdk-hnb659fds-assets-${AWS::AccountId}-${AWS::Region}"
+           },
+           "/*"
+          ]
+         ]
+        },
+        {
+         "Fn::Join": [
+          "",
+          [
+           "arn:",
+           {
+            "Ref": "AWS::Partition"
+           },
+           ":s3:::",
+           {
+            "Fn::Sub": "cdk-hnb659fds-assets-${AWS::AccountId}-${AWS::Region}"
+           }
+          ]
+         ]
+        }
+       ]
+      }
+     ],
+     "Version": "2012-10-17"
+    },
+    "PolicyName": "RayJobServiceRoleDefaultPolicyA615640D",
+    "Roles": [
+     {
+      "Ref": "RayJobServiceRole51433C3D"
+     }
+    ]
+   }
+  },
+  "RayJob2F7864D9": {
+   "Type": "AWS::Glue::Job",
+   "Properties": {
+    "Command": {
+     "Name": "glueray",
+     "PythonVersion": "3.9",
+     "ScriptLocation": {
+      "Fn::Join": [
+       "",
+       [
+        "s3://",
+        {
+         "Fn::Sub": "cdk-hnb659fds-assets-${AWS::AccountId}-${AWS::Region}"
+        },
+        "/432033e3218068a915d2532fa9be7858a12b228a2ae6e5c10faccd9097b1e855.py"
+       ]
+      ]
+     }
+    },
+    "Role": {
+     "Fn::GetAtt": [
+      "RayJobServiceRole51433C3D",
+      "Arn"
+     ]
+    },
+    "DefaultArguments": {
+     "--job-language": "python",
+     "arg1": "value1",
+     "arg2": "value2"
+    },
+    "GlueVersion": "4.0",
+    "Name": "RayJob",
+    "NumberOfWorkers": 2,
+    "Tags": {
+     "key": "value"
+    },
+    "WorkerType": "Z.2X"
+   }
   }
  },
  "Parameters": {

packages/@aws-cdk/aws-glue/test/integ.job.js.snapshot/manifest.json

@@ -17,7 +17,7 @@
         "validateOnSynth": false,
         "assumeRoleArn": "arn:${AWS::Partition}:iam::${AWS::AccountId}:role/cdk-hnb659fds-deploy-role-${AWS::AccountId}-${AWS::Region}",
         "cloudFormationExecutionRoleArn": "arn:${AWS::Partition}:iam::${AWS::AccountId}:role/cdk-hnb659fds-cfn-exec-role-${AWS::AccountId}-${AWS::Region}",
-        "stackTemplateAssetObjectUrl": "s3://cdk-hnb659fds-assets-${AWS::AccountId}-${AWS::Region}/977a2f07e22679bb04b03ce83cc1fac3e6cc269a794e38248ec67106ee39f0a2.json",
+        "stackTemplateAssetObjectUrl": "s3://cdk-hnb659fds-assets-${AWS::AccountId}-${AWS::Region}/b553fef631f82898c826f3c20e1de0d155dbd3a35339ef92d0893052a5be69ce.json",
         "requiresBootstrapStackVersion": 6,
         "bootstrapStackVersionSsmParameter": "/cdk-bootstrap/hnb659fds/version",
         "additionalDependencies": [
@@ -213,6 +213,24 @@
             "data": "ShellJob390C141361"
           }
         ],
+        "/aws-glue-job/RayJob/ServiceRole/Resource": [
+          {
+            "type": "aws:cdk:logicalId",
+            "data": "RayJobServiceRole51433C3D"
+          }
+        ],
+        "/aws-glue-job/RayJob/ServiceRole/DefaultPolicy/Resource": [
+          {
+            "type": "aws:cdk:logicalId",
+            "data": "RayJobServiceRoleDefaultPolicyA615640D"
+          }
+        ],
+        "/aws-glue-job/RayJob/Resource": [
+          {
+            "type": "aws:cdk:logicalId",
+            "data": "RayJob2F7864D9"
+          }
+        ],
         "/aws-glue-job/BootstrapVersion": [
           {
             "type": "aws:cdk:logicalId",