Xcoms


  Key Value Timestamp Execution Date Task Id Dag Id
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=2021-05-07T05:00:00+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": []} 2021-05-07 06:00:07.924804+00:00 2021-05-07 05:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-1FPGAVPOFHEDC 2021-05-07 06:00:26.879630+00:00 2021-05-07 05:00:00+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=2021-05-07T05:05:00+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": []} 2021-05-07 06:05:12.681326+00:00 2021-05-07 05:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-3296J2SPHY821 2021-05-07 06:05:33.379424+00:00 2021-05-07 05:05:00+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=2021-05-07T05:00:00+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": []} 2021-05-07 06:10:09.674919+00:00 2021-05-07 05:00:00+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=2021-05-07T05:00:00+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=2021-05-07T05:00:00+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": []} 2021-05-07 06:10:19.694226+00:00 2021-05-07 05:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-GEUHHAUONC93 2021-05-07 06:10:28.341250+00:00 2021-05-07 05:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-WH9GZTNYK7NJ 2021-05-07 06:10:38.148168+00:00 2021-05-07 05:00:00+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=2021-05-07T06:00:00+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": []} 2021-05-07 07:00:15.047214+00:00 2021-05-07 06:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-208EF4SJS9RYD 2021-05-07 07:00:35.573924+00:00 2021-05-07 06:00:00+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=2021-05-07T06:05:00+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": []} 2021-05-07 07:05:13.141947+00:00 2021-05-07 06:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-DHABKFZCBYW4 2021-05-07 07:05:31.899489+00:00 2021-05-07 06:05:00+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=2021-05-07T06:00:00+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": []} 2021-05-07 07:10:38.479896+00:00 2021-05-07 06:00:00+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=2021-05-07T06:00:00+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=2021-05-07T06:00:00+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": []} 2021-05-07 07:10:54.286836+00:00 2021-05-07 06:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-QF0BS74LOLCC 2021-05-07 07:11:02.903519+00:00 2021-05-07 06:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-3TMUGZDVHHWE5 2021-05-07 07:11:24.713263+00:00 2021-05-07 06:00:00+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=2021-05-07T07:00:00+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": []} 2021-05-07 08:00:07.541351+00:00 2021-05-07 07:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-2K12CH490ANO9 2021-05-07 08:00:28.498178+00:00 2021-05-07 07:00:00+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=2021-05-07T07:05:00+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": []} 2021-05-07 08:05:14.260057+00:00 2021-05-07 07:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-3AD5EOWXOWA9O 2021-05-07 08:05:37.065101+00:00 2021-05-07 07:05:00+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=2021-05-07T07:00:00+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=2021-05-07T07:00:00+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": []} 2021-05-07 08:11:25.837372+00:00 2021-05-07 07:00:00+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=2021-05-07T07:00:00+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": []} 2021-05-07 08:11:30.124315+00:00 2021-05-07 07:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-1F8ONIK3B821I 2021-05-07 08:11:43.894396+00:00 2021-05-07 07:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-1MWIGWY3J610J 2021-05-07 08:11:47.961694+00:00 2021-05-07 07:00:00+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=2021-05-07T08:00:00+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": []} 2021-05-07 09:00:37.366413+00:00 2021-05-07 08:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-FQOCLELKVTBR 2021-05-07 09:00:55.564377+00:00 2021-05-07 08:00:00+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=2021-05-07T08:05:00+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": []} 2021-05-07 09:05:19.393535+00:00 2021-05-07 08:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-MDKSZ0UWDRJW 2021-05-07 09:05:40.043258+00:00 2021-05-07 08:05:00+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=2021-05-07T08:00:00+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=2021-05-07T08:00:00+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": []} 2021-05-07 09:10:48.547324+00:00 2021-05-07 08:00:00+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=2021-05-07T08:00:00+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": []} 2021-05-07 09:10:56.425324+00:00 2021-05-07 08:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-1P5HDUWN89TCI 2021-05-07 09:11:08.652783+00:00 2021-05-07 08:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-1P2V53AILJJYX 2021-05-07 09:11:17.407387+00:00 2021-05-07 08:00:00+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=2021-05-07T09:00:00+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": []} 2021-05-07 10:00:22.364850+00:00 2021-05-07 09:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-1Q0GD9V7SMULS 2021-05-07 10:04:11.312511+00:00 2021-05-07 09:00:00+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=2021-05-07T09:05:00+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": []} 2021-05-07 10:05:07.043577+00:00 2021-05-07 09:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-1YF5I6F88U1QI 2021-05-07 10:05:27.657094+00:00 2021-05-07 09:05:00+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=2021-05-07T09:00:00+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": []} 2021-05-07 10:15:09.353330+00:00 2021-05-07 09:00:00+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=2021-05-07T09:00:00+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=2021-05-07T09:00:00+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": []} 2021-05-07 10:15:34.038902+00:00 2021-05-07 09:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-17YGXSG1TAOQ6 2021-05-07 10:15:38.353245+00:00 2021-05-07 09:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-28AUE2A3ULPA2 2021-05-07 10:15:54.380458+00:00 2021-05-07 09:00:00+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=2021-05-07T10:00:00+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": []} 2021-05-07 11:00:20.316751+00:00 2021-05-07 10:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-DQZBYPVRY0LS 2021-05-07 11:00:40.722098+00:00 2021-05-07 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-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=2021-05-07T10:05:00+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": []} 2021-05-07 11:05:32.698154+00:00 2021-05-07 10:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-17SQXDYV93CNK 2021-05-07 11:05:53.088717+00:00 2021-05-07 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=2021-05-07T10:00:00+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": []} 2021-05-07 11:10:41.550773+00:00 2021-05-07 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-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=2021-05-07T10:00:00+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=2021-05-07T10:00:00+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": []} 2021-05-07 11:10:45.569436+00:00 2021-05-07 10:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-2BNR8E00U467G 2021-05-07 11:11:00.351204+00:00 2021-05-07 10:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-FIO62RULMRQW 2021-05-07 11:11:06.686025+00:00 2021-05-07 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-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=2021-05-07T11:00:00+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": []} 2021-05-07 12:00:09.519815+00:00 2021-05-07 11:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-1O42PPADJ4CI 2021-05-07 12:00:28.397854+00:00 2021-05-07 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=2021-05-07T11:05:00+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": []} 2021-05-07 12:05:10.498880+00:00 2021-05-07 11:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-I8TU6W65AO 2021-05-07 12:10:05.267913+00:00 2021-05-07 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-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=2021-05-07T11:00:00+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=2021-05-07T11:00:00+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": []} 2021-05-07 12:10:24.733427+00:00 2021-05-07 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-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=2021-05-07T11:00:00+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": []} 2021-05-07 12:10:28.740667+00:00 2021-05-07 11:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-2O59BZ305U4V9 2021-05-07 12:10:42.753316+00:00 2021-05-07 11:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-76V8CMTNOI3O 2021-05-07 12:10:49.069120+00:00 2021-05-07 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-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=2021-05-07T12:00:00+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": []} 2021-05-07 13:00:28.323193+00:00 2021-05-07 12:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-1YLK5JW36JAUA 2021-05-07 13:00:46.903647+00:00 2021-05-07 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=2021-05-07T12:05:00+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": []} 2021-05-07 13:05:22.839441+00:00 2021-05-07 12:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-1WVYGQKIB4YLH 2021-05-07 13:05:53.305640+00:00 2021-05-07 12:05:00+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=2021-05-07T12:00:00+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": []} 2021-05-07 13:16:18.592087+00:00 2021-05-07 12:00:00+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=2021-05-07T12:00:00+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=2021-05-07T12:00:00+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": []} 2021-05-07 13:16:30.562856+00:00 2021-05-07 12:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-14G3ZLA98OQNH 2021-05-07 13:16:37.658231+00:00 2021-05-07 12:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-2I7BYNADI2LC 2021-05-07 13:16:51.238704+00:00 2021-05-07 12:00:00+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=2021-05-07T13:00:00+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": []} 2021-05-07 14:00:20.071466+00:00 2021-05-07 13:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-ABSJZD9P088E 2021-05-07 14:00:41.282088+00:00 2021-05-07 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-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=2021-05-07T13:05:00+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": []} 2021-05-07 14:05:08.781507+00:00 2021-05-07 13:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-2NIJEOUF0B4US 2021-05-07 14:05:29.873698+00:00 2021-05-07 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-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=2021-05-07T13:00:00+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": []} 2021-05-07 14:11:35.918700+00:00 2021-05-07 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-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=2021-05-07T13:00:00+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=2021-05-07T13:00:00+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": []} 2021-05-07 14:11:37.936910+00:00 2021-05-07 13:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-126PP849GHFS6 2021-05-07 14:11:56.830008+00:00 2021-05-07 13:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-18P94JQLH82K6 2021-05-07 14:11:58.795039+00:00 2021-05-07 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-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=2021-05-07T14:00:00+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": []} 2021-05-07 15:00:14.308505+00:00 2021-05-07 14:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-7QC46XC5YNGT 2021-05-07 15:00:34.979209+00:00 2021-05-07 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-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=2021-05-07T14:05:00+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": []} 2021-05-07 15:05:13.107699+00:00 2021-05-07 14:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-3RZQBVKGJRFHM 2021-05-07 15:05:33.986621+00:00 2021-05-07 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=2021-05-07T14:00:00+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=2021-05-07T14:00:00+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": []} 2021-05-07 15:10:32.097157+00:00 2021-05-07 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=2021-05-07T14:00:00+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": []} 2021-05-07 15:10:39.762662+00:00 2021-05-07 14:00:00+00:00 create_emr_steps campaign-history-job_v002
return_value j-38F11OTNNL9D4 2021-05-07 15:10:52.473409+00:00 2021-05-07 14:00:00+00:00 create_cluster_and_add_emr_steps booking-processing-job_v002
return_value j-3BO6SSETWT98D 2021-05-07 15:11:00.875962+00:00 2021-05-07 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-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=2021-05-07T15:00:00+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": []} 2021-05-07 16:00:24.492916+00:00 2021-05-07 15:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-2JNLFKIUGZ6MB 2021-05-07 16:00:45.082504+00:00 2021-05-07 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-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=2021-05-07T15:05:00+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": []} 2021-05-07 16:05:17.205919+00:00 2021-05-07 15:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-2MND2YH538N10 2021-05-07 16:05:40.062548+00:00 2021-05-07 15:05:00+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=2021-05-07T15:00:00+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": []} 2021-05-07 16:11:31.728214+00:00 2021-05-07 15:00:00+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=2021-05-07T15:00:00+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=2021-05-07T15:00:00+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": []} 2021-05-07 16:11:43.947565+00:00 2021-05-07 15:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-CE5HF3XZYV9H 2021-05-07 16:11:50.679458+00:00 2021-05-07 15:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-2S6K6QG0YTAIV 2021-05-07 16:12:03.014180+00:00 2021-05-07 15:00:00+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=2021-05-07T16:00:00+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": []} 2021-05-07 17:00:24.356642+00:00 2021-05-07 16:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-3EH90NN7UEJG8 2021-05-07 17:00:44.918873+00:00 2021-05-07 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-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=2021-05-07T16:05:00+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": []} 2021-05-07 17:05:06.809845+00:00 2021-05-07 16:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-1T2ZSSIB9037W 2021-05-07 17:05:27.268393+00:00 2021-05-07 16:05:00+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=2021-05-07T16:00:00+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": []} 2021-05-07 17:11:39.612298+00:00 2021-05-07 16:00:00+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=2021-05-07T16:00:00+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=2021-05-07T16:00:00+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": []} 2021-05-07 17:11:39.740724+00:00 2021-05-07 16:00:00+00:00 create_emr_steps booking-processing-job_v002
return_value j-3HU8KA21TE6IX 2021-05-07 17:11:58.343044+00:00 2021-05-07 16:00:00+00:00 create_cluster_and_add_emr_steps campaign-history-job_v002
return_value j-Y0BMKBV2D698 2021-05-07 17:11:58.641230+00:00 2021-05-07 16:00:00+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=2021-05-07T17:00:00+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": []} 2021-05-07 18:00:09.395019+00:00 2021-05-07 17:00:00+00:00 create_emr_steps campaign-flow-job_v002
return_value j-2C57811JRK7OB 2021-05-07 18:00:29.455902+00:00 2021-05-07 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-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=2021-05-07T17:05:00+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": []} 2021-05-07 18:05:15.462735+00:00 2021-05-07 17:05:00+00:00 create_emr_steps vc-asset-import_v002
return_value j-LXUZV9V95LLK 2021-05-07 18:05:36.094930+00:00 2021-05-07 17:05:00+00:00 create_cluster_and_add_emr_steps vc-asset-import_v002