from airflow import DAG
from template.utils import hours_ago
from template.dag_template import build_dag
dag_id = 'booking-processing-job_v002'
emr_steps = """[
{
"step-name": "CampaignExtractor",
"config-json": [
{"spark.driver.memory":"9g"}
],
"main-class": "com.viooh.campaignextractor.CampaignExtractorMain",
"group-id":"com/viooh/campaignextractor",
"artifact": "campaign-extractor",
"jars": "/usr/lib/spark/connector/lib/spark-avro.jar"
},
{
"step-name": "CampaignProcessingJob",
"config-json": [
{"spark.driver.memory":"9g"}
],
"main-class": "com.viooh.CampaignProcessingMain",
"artifact": "campaign-processing-job",
"files": "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/0.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/0.0.0/job.yaml"
},
{
"step-name": "CampaignDeltaJob",
"config-json": [
{"spark.driver.memory":"9g"}
],
"main-class": "com.viooh.booking.delta.CampaignDeltaMain",
"artifact": "pandora-campaign-delta",
"files": "s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/0.0.0/job.conf,s3://viooh-spark-artifacts-lab-cn/metrics/batch-job-metrics/0.0.0/job.yaml"
}
]"""
# cluster level parameters (optional)
cluster_args = {
"master-instance-types": "m5.xlarge,m5.2xlarge",
"core-instance-types": "m5.xlarge,m5.2xlarge",
"task-instance-types": "m5.xlarge,m5.2xlarge",
"core-instance-capacity": 3,
"task-instance-capacity": 0,
"ebs-volume-size": "50",
"job-type": "batch",
"emr-version": "emr-6.14.0"
}
# dag parameter
dag_args = {
'owner': 'data.engineers@viooh.com',
'start_date': hours_ago(3),
'depends_on_past': True,
'wait_for_downstream': True
}
dag = DAG(
dag_id,
schedule_interval="0 */1 * * *", # cron expression
default_args=dag_args)
build_dag(emr_steps=emr_steps, dag=dag, cluster_args=cluster_args)