default-config.yaml

# Common tags applied to all resources
tags:
  stack-name: covid19-glue-workflow

awsAccount:
  awsRegion: us-east-1

iam:
  glueRoleName: 'Covid19-glue-Role'
  lambdaRoleName : 'Covid19-lambda-Role'
  #boundaryPolicyArn: 'arn:aws:iam::<ACCOUNT_ID>:policy/permissions-boundaries-policy'

glueJobConfig:
  jobTypeSpark : 'glueetl'
  jobTypePythonShell : 'pythonshell'
  databaseName: covid19db
  glueKSMKeyName : 'glue-kms-key'
  s3BucketNameScript : 'covid19-glue-config'
  scriptKeyPrefix : 'etl-script'

  # Crawlers Name
  neherlabCaseCntCrawlerName : 'c_neherlab_case_counts'
  neherlabCntryCdCrawlerName : 'c_neherlab_country_codes'
  neherlabICUCapCrawlerName : 'c_neherlab_icu_capacity'
  neherlabPopulationCrawlerName : 'c_neherlab_population'

  securityConfigurationName : 'covid19-glue-sc'

  # Glue ETL Config
  neherlabDenormGlueJobName :  'j_neherlab_denorm'
  neherlabDenormGlueJobCapacity : 10
  neherlabDenormGlueELTScriptName : 'j_neherlab_denorm_etl.py'

  neherlabCntryCdTableName : 'neherlab_country_codes'
  neherlabCaseCntTableName : 'neherlab_case_counts'
  neherlabICUCapTableName : 'neherlab_icu_capacity'
  neherlabPopulationTableName : 'neherlab_population'
  neherlabDenormTableName : 'neherlab_denormalized'

  emitStartGlueJobName : 'j_emit_start_event'
  emitStartGlueJobCapacity : 0.0625
  emitStartScriptName : 'j_emit_start_event.py'

  emitEndGlueJobName : 'j_emit_ended_event'
  emitEndGlueJobCapacity : 0.0625
  emitEndScriptName : 'j_emit_ended_event.py'

  logManagerScriptName : 'j_log_manager.py'

  # Wrokflow Configuration
  glueWorkflowName : 'Covid_19'

  # Triggers Configuration
  startJobTriggerName : 't_start'
  endJobTriggerName : 't_end'
  neherlabCaseCntCrawlerTriggerName: 't_neherlab_case_counts'
  neherlabCntryCdCrawlerTriggerName: 't_neherlab_country_codes'
  neherlabICUCapCrawlerTriggerName: 't_neherlab_icu_capacity'
  neherlabPopulationCrawlerTriggerName: 't_neherlab_population'
  neherlabDenormGlueJobTriggerName : 't_neherlab_denorm'

dataSourcePath:
  # S3 source bucket for Crawlers
  neherlabSourceS3Bucket : 'covid19-harmonized-dataset'
  crawlerSourceS3PathNeherlabCaseCnt : 'covid19-harmonized-dataset/covid19tos3/neherlab_case_counts/'
  crawlerSourceS3PathNeherlabCntryCd : 'covid19-harmonized-dataset/covid19tos3/neherlab_country_codes/'
  crawlerSourceS3PathNeherlabICUCap : 'covid19-harmonized-dataset/covid19tos3/neherlab_icu_capacity/'
  crawlerSourceS3PathNeherlabPopulation : 'covid19-harmonized-dataset/covid19tos3/neherlab_population/'
  
  # S3 bucket for Glue ETL
  s3BucketNameNeherlabDenorm : 'neherlab-denormalized-dataset'

  # S3 bucket for Glue Temp
  s3PathGlueTemp : 'temp-glue/'

codepipeline:
  applicationName: 'cdk-covid19-glue'
  pipelineStackName: 'cdk-covid19-glue-stack-pipeline'
  glueStackName: 'cdk-covid19-glue-stack'
  repoName: 'cdk-covid19-glue-blog'
  repoBranch: 'master'
  pipelineNotification: 'username@mail.com'