Xcoms


  Key Value Timestamp Execution Date Task Id Dag Id
return_value j-1WX4JYGQ53NVH 2021-05-14 10:10:50.981124+00:00 2021-05-14 09:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-2K5TQX4BJNU5C 2021-05-14 10:10:51.174308+00:00 2021-05-14 09:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T10:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 11:00:16.861329+00:00 2021-05-14 10:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-23NKJAC3D61JC 2021-05-14 11:00:37.422590+00:00 2021-05-14 10:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T10:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 11:05:19.217032+00:00 2021-05-14 10:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-2KUOH4QUBY3NB 2021-05-14 11:05:39.554277+00:00 2021-05-14 10:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T10:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 11:09:12.485700+00:00 2021-05-14 10:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T10:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T10:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 11:09:16.161532+00:00 2021-05-14 10:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-QGYDDHL6BPVF 2021-05-14 11:09:32.309400+00:00 2021-05-14 10:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-375NEYEBEVW93 2021-05-14 11:09:36.419458+00:00 2021-05-14 10:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T11:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 12:00:16.119636+00:00 2021-05-14 11:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-297L41IIG69SN 2021-05-14 12:00:34.582651+00:00 2021-05-14 11:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T11:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 12:05:16.590346+00:00 2021-05-14 11:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-1VORE1DXS3YK8 2021-05-14 12:05:37.093270+00:00 2021-05-14 11:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T11:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T11:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 12:11:13.726745+00:00 2021-05-14 11:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T11:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 12:11:31.331495+00:00 2021-05-14 11:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-20JQ7AIFG43FA 2021-05-14 12:11:36.302363+00:00 2021-05-14 11:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-LYBYA4Q59ACT 2021-05-14 12:11:50.111071+00:00 2021-05-14 11:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T12:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 13:00:24.277807+00:00 2021-05-14 12:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-31WMAE735GYKS 2021-05-14 13:00:44.939247+00:00 2021-05-14 12:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T12:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 13:05:22.405263+00:00 2021-05-14 12:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-27N8V3E6OQCOZ 2021-05-14 13:05:42.812731+00:00 2021-05-14 12:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T12:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T12:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 13:12:35.363028+00:00 2021-05-14 12:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T12:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 13:12:49.272861+00:00 2021-05-14 12:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-36FVNIJE40UQL 2021-05-14 13:13:07.882683+00:00 2021-05-14 12:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-LGSL6TLGPSYX 2021-05-14 13:13:11.752777+00:00 2021-05-14 12:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T13:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 14:00:08.965643+00:00 2021-05-14 13:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-U1ALXV6ADCKT 2021-05-14 14:00:29.720638+00:00 2021-05-14 13:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T13:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 14:05:11.079202+00:00 2021-05-14 13:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-2YPUI9A92NXU 2021-05-14 14:05:31.675730+00:00 2021-05-14 13:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T13:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 14:10:31.791326+00:00 2021-05-14 13:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T13:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T13:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 14:10:34.254514+00:00 2021-05-14 13:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-3QFC10CY2MH8O 2021-05-14 14:10:52.452254+00:00 2021-05-14 13:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-2UW00SZWU051Z 2021-05-14 14:10:54.191586+00:00 2021-05-14 13:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T14:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 15:00:15.791652+00:00 2021-05-14 14:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-1AHFZ6PWXU0IO 2021-05-14 15:00:36.121448+00:00 2021-05-14 14:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T14:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 15:05:08.028216+00:00 2021-05-14 14:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-24YYHNFCER6SX 2021-05-14 15:05:28.552789+00:00 2021-05-14 14:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T14:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T14:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 15:11:37.435735+00:00 2021-05-14 14:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T14:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 15:11:39.345950+00:00 2021-05-14 14:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-3FZ15E8KOX3GB 2021-05-14 15:11:55.907732+00:00 2021-05-14 14:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-35GCTQRG3POJV 2021-05-14 15:11:58.092115+00:00 2021-05-14 14:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T15:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 16:00:21.720933+00:00 2021-05-14 15:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-11FZKB5Z4Z0XF 2021-05-14 16:00:41.911048+00:00 2021-05-14 15:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T15:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 16:05:07.662518+00:00 2021-05-14 15:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-13U91TTNVTVZU 2021-05-14 16:05:26.405533+00:00 2021-05-14 15:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T15:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T15:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 16:10:30.787882+00:00 2021-05-14 15:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T15:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 16:10:44.514736+00:00 2021-05-14 15:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-382VLY4ZNBYRI 2021-05-14 16:10:51.469857+00:00 2021-05-14 15:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-IHXKNL596NWS 2021-05-14 16:11:05.224899+00:00 2021-05-14 15:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T16:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 17:00:13.928792+00:00 2021-05-14 16:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-2C2KUZC5CFKOU 2021-05-14 17:00:34.295530+00:00 2021-05-14 16:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T16:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 17:05:12.186751+00:00 2021-05-14 16:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-5FWN92H02I8N 2021-05-14 17:05:32.983113+00:00 2021-05-14 16:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T16:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T16:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 17:10:32.934646+00:00 2021-05-14 16:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T16:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 17:10:33.016010+00:00 2021-05-14 16:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-O5BK2MJ6EL4N 2021-05-14 17:10:51.307165+00:00 2021-05-14 16:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-3OHDL6XIYY1UH 2021-05-14 17:10:51.512279+00:00 2021-05-14 16:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T17:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 18:01:53.250541+00:00 2021-05-14 17:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-2VJ0PKJRRL6EQ 2021-05-14 18:02:11.757037+00:00 2021-05-14 17:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T17:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 18:05:23.731587+00:00 2021-05-14 17:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-140T5NQER7FY2 2021-05-14 18:05:43.905675+00:00 2021-05-14 17:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T17:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T17:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 18:12:08.333080+00:00 2021-05-14 17:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T17:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 18:12:14.341950+00:00 2021-05-14 17:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-1L5XND6JHIJOU 2021-05-14 18:12:28.769325+00:00 2021-05-14 17:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-6JVC0OEUEEPF 2021-05-14 18:12:34.897861+00:00 2021-05-14 17:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T18:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 19:00:28.378399+00:00 2021-05-14 18:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-2XISSQX84DO9N 2021-05-14 19:00:47.058294+00:00 2021-05-14 18:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T18:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 19:05:19.442812+00:00 2021-05-14 18:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-29OLT0SPVMET7 2021-05-14 19:05:39.640720+00:00 2021-05-14 18:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T18:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 19:09:49.866568+00:00 2021-05-14 18:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T18:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T18:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 19:09:49.891158+00:00 2021-05-14 18:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-2MDQG4J9XIXNP 2021-05-14 19:10:08.645684+00:00 2021-05-14 18:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-19S8XOGH404WI 2021-05-14 19:10:10.354377+00:00 2021-05-14 18:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T19:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 20:00:14.564578+00:00 2021-05-14 19:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-1HKCFIT32MC4C 2021-05-14 20:00:35.124972+00:00 2021-05-14 19:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T19:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 20:05:15.068472+00:00 2021-05-14 19:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-39IHQP0SZW7FC 2021-05-14 20:05:33.307325+00:00 2021-05-14 19:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T19:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T19:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 20:10:33.496761+00:00 2021-05-14 19:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T19:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 20:10:45.181463+00:00 2021-05-14 19:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-16D686FZL46XU 2021-05-14 20:10:55.918308+00:00 2021-05-14 19:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-2I5JCY30RE8EZ 2021-05-14 20:11:14.234539+00:00 2021-05-14 19:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T20:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 21:00:08.405747+00:00 2021-05-14 20:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-D8JC1H0KMD0X 2021-05-14 21:00:27.109905+00:00 2021-05-14 20:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T20:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 21:05:24.922178+00:00 2021-05-14 20:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-1BB2MYE1Q18TD 2021-05-14 21:05:45.178049+00:00 2021-05-14 20:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T20:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 21:09:33.826561+00:00 2021-05-14 20:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T20:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T20:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 21:09:35.924190+00:00 2021-05-14 20:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-38O5PIGK058OC 2021-05-14 21:09:53.713199+00:00 2021-05-14 20:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-DY0MKGRERW6K 2021-05-14 21:09:53.925405+00:00 2021-05-14 20:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T21:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 22:00:27.901640+00:00 2021-05-14 21:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-3CR82OCW1OO3E 2021-05-14 22:00:48.535811+00:00 2021-05-14 21:00:00+00:00 create_cluster_and_add_emr_steps campaign-flow-job_v002
return_value {"Name": "vd-uat-viooh-vc-asset-import", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.xlarge"}, {"InstanceType": "m5.2xlarge"}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "PopulateVCAssetsToCassandra", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T21:05:00+00:00", "--class", "VioohCentralAssetsMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/daily-import/1.6.3/daily-import-1.6.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-vc-asset-import"}], "Configurations": []} 2021-05-14 22:05:26.306581+00:00 2021-05-14 21:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-261MZNT59VEO 2021-05-14 22:05:46.472560+00:00 2021-05-14 21:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002
return_value {"Name": "vd-uat-viooh-campaign-history-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignHistoryJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T21:00:00+00:00", "--class", "com.viooh.campaign.history.campaignHistoryMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-history/1.2.3/campaign-history-1.2.3-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-history-job"}], "Configurations": []} 2021-05-14 22:11:36.184393+00:00 2021-05-14 21:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value {"Name": "vd-uat-viooh-booking-processing-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 7, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}, {"InstanceType": "m5.4xlarge", "EbsConfiguration": {"EbsBlockDeviceConfigs": [{"VolumeSpecification": {"VolumeType": "gp2", "SizeInGB": 100}, "VolumesPerInstance": 1}], "EbsOptimized": true}}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 1, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-06c5b5a18c64a0316", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignProcessingJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T21:00:00+00:00", "--class", "com.viooh.CampaignProcessingMain", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaign-processing-job/1.3.6/campaign-processing-job-1.3.6.jar"], "Jar": "command-runner.jar"}}, {"Name": "CampaignDeltaJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T21:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-booking-processing-job"}], "Configurations": []} 2021-05-14 22:11:42.613405+00:00 2021-05-14 21:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-12R5EBWT3ENN2 2021-05-14 22:11:57.089183+00:00 2021-05-14 21:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-2PBMC742D677O 2021-05-14 22:12:01.217998+00:00 2021-05-14 21:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value {"Name": "vd-uat-viooh-campaign-delta-past-delete-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-09a80b42d0e6dedd8", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignDeltaPastDeleteJob", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-13T23:00:00+00:00", "--class", "com.viooh.booking.delta.CampaignDeltaPastDeleteMain", "--files", "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/1.0.0/job.yaml", "--conf", "spark.metrics.conf=job.conf", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/pandora-campaign-delta/1.4.9/pandora-campaign-delta-1.4.9-jar-with-dependencies.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-delta-past-delete-job"}], "Configurations": []} 2021-05-14 23:00:20.466628+00:00 2021-05-13 23:00:00+00:00 create_emr_steps campaign-delta-past-delete-job_v002
return_value {"Name": "vd-uat-viooh-campaign-flow-job", "LogUri": "s3n://viooh-emr-logs-lab-cn/bigdata/emr/", "ReleaseLabel": "emr-5.21.0", "Instances": {"InstanceFleets": [{"Name": "Masterfleet", "InstanceFleetType": "MASTER", "TargetOnDemandCapacity": 1, "InstanceTypeConfigs": [{"InstanceType": "m5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Corefleet", "InstanceFleetType": "CORE", "TargetOnDemandCapacity": 3, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge"}, {"InstanceType": "m5.4xlarge"}]}, {"Name": "Taskfleet", "InstanceFleetType": "TASK", "TargetSpotCapacity": 3, "LaunchSpecifications": {"SpotSpecification": {"TimeoutDurationMinutes": 120, "TimeoutAction": "SWITCH_TO_ON_DEMAND"}}, "InstanceTypeConfigs": [{"InstanceType": "r5.2xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}, {"InstanceType": "m5.4xlarge", "BidPriceAsPercentageOfOnDemandPrice": 90}]}], "Ec2KeyName": "data.engineer.lab", "KeepJobFlowAliveWhenNoSteps": false, "TerminationProtected": false, "Ec2SubnetId": "subnet-0e214a494e67e2266", "EmrManagedMasterSecurityGroup": "sg-0da002beeed87fb1a", "EmrManagedSlaveSecurityGroup": "sg-0f1a0b7517309fcf0", "ServiceAccessSecurityGroup": "sg-009528c399781cc69"}, "BootstrapActions": [], "Applications": [{"Name": "Spark"}, {"Name": "Hadoop"}, {"Name": "Ganglia"}, {"Name": "Zeppelin"}], "VisibleToAllUsers": true, "Steps": [{"Name": "Setup Hadoop Debugging", "ActionOnFailure": "TERMINATE_JOB_FLOW", "HadoopJarStep": {"Jar": "command-runner.jar", "Args": ["state-pusher-script"]}}, {"Name": "CampaignExtractor", "ActionOnFailure": "TERMINATE_CLUSTER", "HadoopJarStep": {"Args": ["spark-submit", "--deploy-mode", "cluster", "--master", "yarn", "--conf", "spark.yarn.submit.waitAppCompletion=true", "--conf", "spark.driver.memory=40g", "--conf", "spark.app.env=uat", "--conf", "spark.dag.execution.time=2021-05-14T22:00:00+00:00", "--class", "com.viooh.campaignextractor.CampaignExtractorMain", "--packages", "org.apache.spark:spark-avro_2.11:2.4.0", "s3://viooh-spark-artifacts-lab-cn/releases/com/viooh/campaignextractor/campaign-extractor/1.0.0/campaign-extractor-1.0.0.jar"], "Jar": "command-runner.jar"}}], "JobFlowRole": "DeEmrInstanceRole", "ServiceRole": "DeEmrServiceRole", "SecurityConfiguration": "bigdata_emr_cn_sec_conf", "Tags": [{"Key": "environment", "Value": "uat"}, {"Key": "role", "Value": "analysis"}, {"Key": "application", "Value": "emr"}, {"Key": "project", "Value": "campaign-compliance"}, {"Key": "TerminationPolicy", "Value": "OFFICE-HOURS"}, {"Key": "ContactEmail", "Value": "viooh.data@viooh.com"}, {"Key": "team", "Value": "bigdata"}, {"Key": "Name", "Value": "vd-uat-viooh-campaign-flow-job"}], "Configurations": []} 2021-05-14 23:00:24.162631+00:00 2021-05-14 22:00:00+00:00 create_emr_steps campaign-flow-job_v002