def main_summary_subdag_factory(parent_dag, task_id, day):
ds = "{{{{ macros.ds_format(macros.ds_add(ds, {0}), '%Y-%m-%d', '%Y%m%d') }}}}".format(day)
subdag = DAG("{}.{}".format(parent_dag.dag_id, task_id),
schedule_interval=SCHEDULE_INTERVAL,
start_date=START_DATE,
default_args=default_args)
parent_job_flow_id = ("{{{{ task_instance.xcom_pull('setup_backfill_cluster', "
"key='return_value', dag_id={}) }}}}".format(parent_dag.dag_id))
# Try to alleviate throttling issues by introducing some slight jitter on each of the days
timedelta_task = TimeDeltaSensor(
task_id="day_start_jitter",
delta=timedelta(seconds=day),
dag=subdag
)
add_step_task = EmrAddStepsOperator(
task_id='submit_main_summary_day',
job_flow_id=parent_job_flow_id,
execution_timeout=timedelta(minutes=10),
aws_conn_id='aws_default',
steps=EmrAddStepsOperator.get_step_args(
job_name="main_summary {}".format(ds),
owner="*****@*****.**",
action_on_failure='CONTINUE',
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
env=tbv_envvar("com.mozilla.telemetry.views.MainSummaryView", {
"from": ds,
"to": ds,
"bucket": "telemetry-backfill"
}, {
"DO_ASSEMBLY": "False"
}),
),
dag=subdag
)
step_sensor_task = EmrStepSensor(
task_id="main_summary_step_sensor",
timeout=timedelta(hours=10).total_seconds(),
job_flow_id=parent_job_flow_id,
step_id="{{ task_instance.xcom_pull('submit_main_summary_day', key='return_value') }}",
poke_interval=timedelta(minutes=5).total_seconds(),
dag=subdag
)
step_sensor_task.set_upstream(add_step_task)
add_step_task.set_upstream(timedelta_task)
return subdag
task_id="main_summary_all_histograms",
job_name="Main Summary View - All Histograms",
execution_timeout=timedelta(hours=12),
instance_count=5,
max_instance_count=50,
enable_autoscale=True,
instance_type="c4.4xlarge",
spot_bid_price_percent=50,
ebs_volume_count=1,
ebs_volume_size=250,
env=tbv_envvar("com.mozilla.telemetry.views.MainSummaryView",
options={
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "telemetry-backfill",
"all_histograms": "",
"read-mode": "aligned",
"input-partition-multiplier": "400",
},
dev_options={
"channel": "nightly",
}),
dag=dag)
main_summary = MozDatabricksSubmitRunOperator(
task_id="main_summary",
job_name="Main Summary View",
execution_timeout=timedelta(hours=4),
email=[
"*****@*****.**", "*****@*****.**",
"*****@*****.**"
],
"date": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.public_output_bucket }}"
},
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/hardware_report.sh",
output_visibility="public",
dag=dag)
cross_sectional = EMRSparkOperator(
task_id="cross_sectional",
job_name="Cross Sectional View",
execution_timeout=timedelta(hours=10),
instance_count=30,
env=tbv_envvar(
"com.mozilla.telemetry.views.CrossSectionalView", {
"outName": "v" + DS_WEEKLY,
"outputBucket": "{{ task.__class__.private_output_bucket }}"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
distribution_viewer = EMRSparkOperator(
task_id="distribution_viewer",
job_name="Distribution Viewer",
owner="*****@*****.**",
email=["*****@*****.**", "*****@*****.**"],
execution_timeout=timedelta(hours=10),
instance_count=5,
env={"date": DS_WEEKLY},
uri=
}
# Make sure all the data for the given day has arrived before running.
# Running at 1am should suffice.
dag = DAG('main_summary',
default_args=default_args,
schedule_interval='0 1 * * *')
main_summary = EMRSparkOperator(
task_id="main_summary",
job_name="Main Summary View",
execution_timeout=timedelta(hours=14),
instance_count=40,
env=tbv_envvar(
"com.mozilla.telemetry.views.MainSummaryView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
experiments_error_aggregates = EMRSparkOperator(
task_id="experiments_error_aggregates",
job_name="Experiments Error Aggregates View",
execution_timeout=timedelta(hours=5),
instance_count=20,
owner="*****@*****.**",
email=["*****@*****.**", "*****@*****.**"],
env={
"date": "{{ ds_nodash }}",
execution_timeout=timedelta(hours=12),
instance_count=5,
max_instance_count=50,
enable_autoscale=True,
instance_type="c4.4xlarge",
spot_bid_price_percent=50,
ebs_volume_count=1,
ebs_volume_size=250,
env=tbv_envvar(
"com.mozilla.telemetry.views.MainSummaryView",
options={
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"schema-report-location":
"s3://{{ task.__class__.private_output_bucket }}/schema/main_summary/submission_date_s3={{ ds_nodash }}",
"bucket": "telemetry-backfill",
"all_histograms": "",
"read-mode": "aligned",
"input-partition-multiplier": "400",
},
dev_options={
"channel": "nightly",
}),
dag=dag)
main_summary = EMRSparkOperator(
task_id="main_summary",
job_name="Main Summary View",
execution_timeout=timedelta(hours=14),
instance_count=40,
env=tbv_envvar(
from operators.emr_spark_operator import EMRSparkOperator
from utils.constants import DS_WEEKLY
from utils.tbv import tbv_envvar
default_args = {
'owner': '*****@*****.**',
'depends_on_past': False,
'start_date': datetime(2017, 5, 26),
'email': ['*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('focus_event_longitudinal', default_args=default_args, schedule_interval='@weekly')
focus_event_longitudinal = EMRSparkOperator(
task_id="focus_event_longitudinal",
job_name="Focus Event Longitudinal View",
execution_timeout=timedelta(hours=12),
instance_count=10,
env = tbv_envvar("com.mozilla.telemetry.views.GenericLongitudinalView", {
"to": DS_WEEKLY,
"tablename": "telemetry_focus_event_parquet",
"output-path": "{{ task.__class__.private_output_bucket }}/focus_event_longitudinal",
"num-parquet-files": "30",
"ordering-columns": "seq,created"}),
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
timedelta(minutes=30),
}
dag = DAG('longitudinal',
default_args=default_args,
schedule_interval='@weekly')
longitudinal = MozDatabricksSubmitRunOperator(
task_id="longitudinal",
job_name="Longitudinal View",
execution_timeout=timedelta(hours=12),
instance_count=16,
instance_type="i3.8xlarge",
env=tbv_envvar("com.mozilla.telemetry.views.LongitudinalView", {
"bucket": "{{ task.__class__.private_output_bucket }}",
"to": DS_WEEKLY
},
metastore_location="s3://telemetry-parquet/longitudinal"),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
register_status(longitudinal, "Longitudinal",
"A 6-month longitudinal view of client history.")
addon_recommender = EMRSparkOperator(
task_id="addon_recommender",
job_name="Train the Addon Recommender",
execution_timeout=timedelta(hours=10),
instance_count=20,
owner="*****@*****.**",
devtools_release_events_to_amplitude = EMRSparkOperator(
task_id="devtools_release_events_to_amplitude",
job_name="DevTools Release Events to Amplitude",
execution_timeout=timedelta(hours=8),
instance_count=DEVTOOLS_INSTANCES,
dev_instance_count=DEVTOOLS_INSTANCES,
email=['*****@*****.**', '*****@*****.**'],
owner='*****@*****.**',
env=tbv_envvar(
"com.mozilla.telemetry.streaming.EventsToAmplitude", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"max_parallel_requests": str(
DEVTOOLS_INSTANCES * VCPUS_PER_INSTANCE),
"config_file_path": "devtools_release_schemas.json",
"url": "https://api.amplitude.com/httpapi",
"sample": "0.5",
"partition_multiplier": "5"
},
artifact_url=get_artifact_url(slug),
other={
"KEY_BUCKET": "telemetry-airflow",
"KEY_PATH": key_path("devtools"),
"DO_EVENTS_TO_AMPLITUDE_SETUP": "True"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
start_date=datetime(2018, 12, 4),
dag=dag)
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
# Make sure all the data for the given day has arrived before running.
# Running at 1am should suffice.
dag = DAG('main_summary', default_args=default_args, schedule_interval='0 1 * * *')
main_summary = EMRSparkOperator(
task_id="main_summary",
job_name="Main Summary View",
execution_timeout=timedelta(hours=14),
instance_count=40,
env=tbv_envvar("com.mozilla.telemetry.views.MainSummaryView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}"}),
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
experiments_error_aggregates = EMRSparkOperator(
task_id="experiments_error_aggregates",
job_name="Experiments Error Aggregates View",
execution_timeout=timedelta(hours=5),
instance_count=20,
release_label="emr-5.13.0",
owner="*****@*****.**",
email=["*****@*****.**", "*****@*****.**"],
env={"date": "{{ ds_nodash }}", "bucket": "{{ task.__class__.private_output_bucket }}"},
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/experiments_error_aggregates_view.sh",
dag=dag)
'email': ['*****@*****.**', '*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('sync_view', default_args=default_args, schedule_interval='@daily')
sync_view = EMRSparkOperator(
task_id="sync_view",
job_name="Sync Pings View",
execution_timeout=timedelta(hours=10),
instance_count=5,
env = tbv_envvar("com.mozilla.telemetry.views.SyncView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}"}),
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
sync_view_bigquery_load = SubDagOperator(
subdag=load_to_bigquery(
parent_dag_name=dag.dag_id,
dag_name="sync_view_bigquery_load",
default_args=default_args,
dataset_s3_bucket="telemetry-parquet",
aws_conn_id="aws_dev_iam_s3",
dataset="sync_summary",
dataset_version="v2",
gke_cluster_name="bq-load-gke-1",
),
job_flow_overrides=EmrCreateJobFlowSelectiveTemplateOperator.
get_jobflow_args(owner="*****@*****.**",
instance_count=20,
keep_alive=True,
job_name="Main Summary Backfill"),
templated_job_flow_overrides={
"Name":
"Main Summary Backfill {{ ds }}",
"Steps":
EmrCreateJobFlowSelectiveTemplateOperator.get_step_args(
job_name="compile_main_summary",
owner="*****@*****.**",
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
env=tbv_envvar(None,
options={},
branch="backfill",
other={"DO_SUBMIT": "False"})),
},
dag=dag)
cluster_start_sensor_task = MozEmrClusterStartSensor(
task_id="wait_for_cluster",
timeout=timedelta(hours=1).total_seconds(),
job_flow_id=job_flow_id_template,
dag=dag)
terminate_job_flow_task = EmrTerminateJobFlowOperator(
task_id="terminate_backfill_cluster",
aws_conn_id='aws_default',
execution_timeout=timedelta(minutes=10),
job_flow_id=job_flow_id_template,
'retry_delay': timedelta(minutes=30),
}
# Make sure all the data for the given day has arrived before running.
# Running at 1am should suffice.
dag = DAG('first_shutdown_summary', default_args=default_args, schedule_interval='0 1 * * *')
first_shutdown_summary = EMRSparkOperator(
task_id="first_shutdown_summary",
job_name="First Shutdown Summary View",
execution_timeout=timedelta(hours=1),
instance_count=1,
env=tbv_envvar("com.mozilla.telemetry.views.MainSummaryView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}",
"doc-type": "first_shutdown",
"read-mode": "aligned",
"input-partition-multiplier": "4"
}),
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
first_shutdown_summary_bigquery_load = SubDagOperator(
subdag=load_to_bigquery(
parent_dag_name=dag.dag_id,
dag_name="first_shutdown_summary_bigquery_load",
default_args=default_args,
dataset_s3_bucket="telemetry-parquet",
aws_conn_id="aws_dev_iam_s3",
dataset="first_shutdown_summary",
dataset_version="v4",
default_args = {
'owner': '*****@*****.**',
'depends_on_past': False,
'start_date': datetime(2018, 6, 27),
'email': ['*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('event_ping_events',
default_args=default_args,
schedule_interval='0 1 * * *')
event_ping_events = EMRSparkOperator(
task_id="event_ping_events",
job_name="Event Ping Events Dataset",
execution_timeout=timedelta(hours=8),
instance_count=5,
env=tbv_envvar(
"com.mozilla.telemetry.streaming.EventPingEvents", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"outputPath": "s3://{{ task.__class__.private_output_bucket }}/"
},
artifact_url=url),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
slug = "{{ task.__class__.telemetry_streaming_slug }}"
url = get_artifact_url(slug)
default_args = {
'owner': '*****@*****.**',
'depends_on_past': False,
'start_date': datetime(2018, 11, 26),
'email': ['*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('event_ping_events', default_args=default_args, schedule_interval='0 1 * * *')
event_ping_events = EMRSparkOperator(
task_id="event_ping_events",
job_name="Event Ping Events Dataset",
execution_timeout=timedelta(hours=8),
instance_count=5,
env=tbv_envvar("com.mozilla.telemetry.streaming.EventPingEvents", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"outputPath": "s3://{{ task.__class__.private_output_bucket }}/"
}, artifact_url=url),
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
# Make sure all the data for the given day has arrived before running.
# Running at 1am should suffice.
dag = DAG('main_summary',
default_args=default_args,
schedule_interval='0 1 * * *')
main_summary = EMRSparkOperator(
task_id="main_summary",
job_name="Main Summary View",
execution_timeout=timedelta(hours=14),
instance_count=40,
env=tbv_envvar(
"com.mozilla.telemetry.views.MainSummaryView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"schema-report-location":
"s3://{{ task.__class__.private_output_bucket }}/schema/main_summary/submission_date_s3={{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
main_summary_schema = EmailSchemaChangeOperator(
task_id="main_summary_schema",
email=["*****@*****.**", "*****@*****.**"],
to=["*****@*****.**"],
key_prefix='schema/main_summary/submission_date_s3=',
dag=dag)
experiments_error_aggregates = EMRSparkOperator(
from utils.constants import DS_WEEKLY
from utils.tbv import tbv_envvar
default_args = {
'owner': '*****@*****.**',
'depends_on_past': False,
'start_date': datetime(2018, 10, 21),
'email': ['*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('quantum_release_criteria_view',
default_args=default_args,
schedule_interval='@weekly')
quantum_release_criteria_view = EMRSparkOperator(
task_id="quantum_release_criteria_view",
job_name="Quantum Release Criteria View",
execution_timeout=timedelta(hours=2),
instance_count=10,
env=tbv_envvar("com.mozilla.telemetry.views.QuantumRCView", {
"to": DS_WEEKLY,
"bucket": "{{ task.__class__.private_output_bucket }}"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
task_id="main_summary_all_histograms",
job_name="Main Summary View - All Histograms",
execution_timeout=timedelta(hours=12),
instance_count=5,
max_instance_count=50,
enable_autoscale=True,
instance_type="c4.4xlarge",
spot_bid_price_percent=50,
ebs_volume_count=1,
ebs_volume_size=250,
env=tbv_envvar("com.mozilla.telemetry.views.MainSummaryView",
options={
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "telemetry-backfill",
"all_histograms": "",
"read-mode": "aligned",
"input-partition-multiplier": "400",
},
dev_options={
"channel": "nightly",
}),
dag=dag)
main_summary = MozDatabricksSubmitRunOperator(
task_id="main_summary",
job_name="Main Summary View",
execution_timeout=timedelta(hours=4),
email=["*****@*****.**", "*****@*****.**", "*****@*****.**"],
instance_count=5,
max_instance_count=40,
enable_autoscale=True,
'retries':
2,
'retry_delay':
timedelta(minutes=30),
}
dag = DAG('sync_view', default_args=default_args, schedule_interval='@daily')
sync_view = EMRSparkOperator(
task_id="sync_view",
job_name="Sync Pings View",
execution_timeout=timedelta(hours=10),
instance_count=5,
env=tbv_envvar(
"com.mozilla.telemetry.views.SyncView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
sync_view_bigquery_load = SubDagOperator(subdag=load_to_bigquery(
parent_dag_name=dag.dag_id,
dag_name="sync_view_bigquery_load",
default_args=default_args,
dataset_s3_bucket="telemetry-parquet",
aws_conn_id="aws_dev_iam_s3",
dataset="sync_summary",
dataset_version="v2",
gke_cluster_name="bq-load-gke-1",
'email': ['*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
# Make sure all the data for the given day has arrived before running.
# Running at 1am should suffice.
dag = DAG('first_shutdown_summary',
default_args=default_args,
schedule_interval='0 1 * * *')
first_shutdown_summary = EMRSparkOperator(
task_id="first_shutdown_summary",
job_name="First Shutdown Summary View",
execution_timeout=timedelta(hours=1),
instance_count=1,
env=tbv_envvar(
"com.mozilla.telemetry.views.MainSummaryView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "{{ task.__class__.private_output_bucket }}",
"doc-type": "first_shutdown",
"read-mode": "aligned",
"input-partition-multiplier": "4"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
'retries': 2,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('longitudinal', default_args=default_args, schedule_interval='@weekly')
longitudinal = MozDatabricksSubmitRunOperator(
task_id="longitudinal",
job_name="Longitudinal View",
execution_timeout=timedelta(hours=12),
instance_count=16,
instance_type="i3.8xlarge",
env=tbv_envvar(
"com.mozilla.telemetry.views.LongitudinalView",
{
"bucket": "{{ task.__class__.private_output_bucket }}",
"to": DS_WEEKLY
},
metastore_location="s3://telemetry-parquet/longitudinal"),
uri="https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
register_status(longitudinal, "Longitudinal", "A 6-month longitudinal view of client history.")
game_hw_survey = EMRSparkOperator(
task_id="game_hw_survey",
job_name="Firefox Hardware Report",
execution_timeout=timedelta(hours=5),
instance_count=15,
owner="*****@*****.**",
default_args = {
'owner': '*****@*****.**',
'depends_on_past': False,
'start_date': datetime(2018, 11, 20),
'email': ['*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 3,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('crash_summary',
default_args=default_args,
schedule_interval='@daily')
crash_summary_view = EMRSparkOperator(
task_id="crash_summary_view",
job_name="Crash Summary View",
instance_count=20,
execution_timeout=timedelta(hours=4),
env=tbv_envvar(
"com.mozilla.telemetry.views.CrashSummaryView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"outputBucket": "{{ task.__class__.private_output_bucket }}"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)
default_args = {
'owner': '*****@*****.**',
'depends_on_past': False,
'start_date': datetime(2016, 9, 20),
'email': ['*****@*****.**', '*****@*****.**'],
'email_on_failure': True,
'email_on_retry': True,
'retries': 3,
'retry_delay': timedelta(minutes=30),
}
dag = DAG('crash_aggregates_backfill',
default_args=default_args,
schedule_interval='@daily')
crash_aggregates_view_backfill = EMRSparkOperator(
task_id="crash_aggregates_view_backfill",
job_name="Crash Aggregates View Backfill",
instance_count=20,
execution_timeout=timedelta(hours=4),
env=tbv_envvar(
"com.mozilla.telemetry.views.CrashAggregateView", {
"from": "{{ ds_nodash }}",
"to": "{{ ds_nodash }}",
"bucket": "telemetry-test-bucket"
}),
uri=
"https://raw.githubusercontent.com/mozilla/telemetry-airflow/master/jobs/telemetry_batch_view.py",
dag=dag)