{
	"Description": "CloudFormation template for creating an EMR cluster",
	"Outputs": {
		"IPAddress": {
			"Description": "IP address of EMR cluster MasterNode",
			"Value": {
				"Fn::GetAtt": [
					"EMRSampleCluster",
					"MasterPublicDNS"
				]
			}
		}
	},
	"Parameters": {

		"CoreInstanceCount": {
			"Default": "3",
			"Description": "Number of core instances",
			"Type": "Number"
		},
		"CoreInstanceType": {
			"Default": "m4.xlarge",
			"Description": "Instance Type of the core node",
			"Type": "String"
		},
		"EMRClusterName": {
			"Default": "SecurityPOCCluster",
			"Description": "Cluster name for the EMR",
			"Type": "String"
		},
		"EMRLogDir": {
			"Description": "Log Dir for the EMR cluster",
			"Type": "String"
		},
		"KeyName": {
			"Description": "Name of an existing EC2 KeyPair to enable SSH to the instances",
			"Type": "AWS::EC2::KeyPair::KeyName"
		},

		"MasterInstanceType": {
			"Default": "m4.xlarge",
			"Description": "Instance Type of the master node",
			"Type": "String"
		},

		"Subnet": {
			"Description": "Subnet ID for creating the EMR cluster",
			"Type": "AWS::EC2::Subnet::Id"
		},
		"VPC": {
			"Description": "VPC ID for creating the EMR cluster",
			"Type": "AWS::EC2::VPC::Id"
		},
		"emrReleaseLabel": {
			"Default": "emr-5.9.0",
			"Description": "Release label for the EMR cluster",
			"Type": "String"
		}
	},
	"Resources": {
		"EMRSampleCluster": {
			"Properties": {
				"Applications": [{
						"Name": "Hive"
					},
					{
						"Name": "Spark"
					},
					{
						"Name": "Sqoop"
					},
					{
						"Name": "Hue"
					}
				],


				"Instances": {
					"CoreInstanceGroup": {

						"InstanceCount": {
							"Ref": "CoreInstanceCount"
						},
						"InstanceType": {
							"Ref": "CoreInstanceType"
						},
						"Market": "ON_DEMAND",
						"Name": "Core Instance"
					},
					"Ec2KeyName": {
						"Ref": "KeyName"
					},
					"Ec2SubnetId": {
						"Ref": "Subnet"
					},
					"MasterInstanceGroup": {
						"InstanceCount": "1",
						"InstanceType": {
							"Ref": "MasterInstanceType"
						},
						"Market": "ON_DEMAND",
						"Name": "Master Instance"
					},
					"TerminationProtected": "false"
				},
				"JobFlowRole": "EMR_EC2_DefaultRole",
				"LogUri": {
					"Ref": "EMRLogDir"
				},
				"Name": {
					"Ref": "EMRClusterName"
				},
				"ReleaseLabel": {
					"Ref": "emrReleaseLabel"
				},
				"ServiceRole": "EMR_DefaultRole",
				"Tags": [{
					"Key": "Name",
					"Value": "EMR Sample Cluster"
				}],
				"VisibleToAllUsers": "true"
			},
			"Type": "AWS::EMR::Cluster"
		},
		"LoadHDFSData": {
			"Properties": {
				"ActionOnFailure": "CONTINUE",
				"HadoopJarStep": {
					"Args": [
						"s3://mahis3bucket/moodys/loadloads3.sh"
					],
					"Jar": "s3://elasticmapreduce/libs/script-runner/script-runner.jar",
					"MainClass": ""
				},
				"JobFlowId": {
					"Ref": "EMRSampleCluster"
				},
				"Name": "LoadHDFSData"
			},
			"Type": "AWS::EMR::Step"
		},

		"SparkStep": {
			"Properties": {
				"ActionOnFailure": "CONTINUE",
				"HadoopJarStep": {
					"Args": [
						"spark-submit",
						"--deploy-mode",
						"cluster",
						"s3://mahis3bucket/moodys/py5.py",
						"10"
					],
					"Jar": "command-runner.jar",
					"MainClass": ""
				},
				"JobFlowId": {
					"Ref": "EMRSampleCluster"
				},
				"Name": "SparkStep"
			},
			"Type": "AWS::EMR::Step"
		}
	}
}