I am trying to generate airflow dags using a template in a python code, and using globals() as defined here
To define dag object and saving it. Below is my code :
import datetime as dt
import sys
import airflow
from airflow.models import DAG
from airflow.operators.bash_operator import BashOperator
argumentList = sys.argv
owner = argumentList[1]
dag_name = argumentList[2]
taskID = argumentList[3]
bashCommand = argumentList[4]
default_args = {
'owner': owner,
'start_date': dt.datetime(2019, 6, 1),
'retries': 1,
'retry_delay': dt.timedelta(minutes=5),
}
def dagCreate():
with DAG(dag_name,
default_args=default_args,
schedule_interval=None,
) as dag:
print_hello = BashOperator(task_id=taskID, bash_command=bashCommand)
return dag
globals()[dag_name] = dagCreate()
I have kept this python code outside dag_folder, and executing it as follows :
python bash-dag-generator.py Airflow test_bash_generate auto_bash_task ls
But I don't see any DAG generated in the airflow webserver UI. I am not sure where I am going wrong.
As per the official documentation:
DAGs are defined in standard Python files that are placed in Airflow’s DAG_FOLDER. Airflow will execute the code in each file to dynamically build the DAG objects. You can have as many DAGs as you want, each describing an arbitrary number of tasks. In general, each one should correspond to a single logical workflow.
So unless your code is actually inside the DAG_FOLDER, it will not be registered as a DAG.
The way I have been able to implement Dynamic DAGs is by using Airflow Variable.
In the below example I have a csv file that has list of Bash command like ls, echo etc. As part of the read_file task I am updating the file location to the Airflow Variable. The part where we read the csv file and loop through the commands is where the dynamic DAGs get created.
from airflow import DAG
from airflow.operators.python_operator import PythonOperator
from airflow.operators.bash_operator import BashOperator
from airflow.models import Variable
from datetime import datetime, timedelta
import csv
'''
Orchestrate the Dynamic Tasks
'''
def read_file_task():
print('I am reading a File and setting variables ')
Variable.set('dynamic-dag-sample','/home/bashoperator.csv')
with DAG('dynamic-dag-sample',
start_date=datetime(2018, 11, 1)) as dag:
read_file_task = PythonOperator(task_id='read_file_task',
python_callable=read_file_task, provide_context=True,
dag=dag)
dynamic_dag_sample_file_path = Variable.get("dynamic-dag-sample")
if dynamic_dag_sample_file_path != None:
with open(dynamic_dag_sample_file_path) as csv_file:
reader = csv.DictReader(csv_file)
line_count = 0
for row in reader:
bash_task = BashOperator(task_id=row['Taskname'], bash_command=row['Command'])
read_file_task.set_downstream(bash_task)
Related
I am trying to trigger multiple external dag dataflow job via master dag.
I plan to use TriggerDagRunOperator and ExternalTaskSensor . I have around 10 dataflow jobs - some are to be executed in sequence and some in parallel .
For example: I want to execute Dag dataflow jobs A,B,C etc from master dag and before execution goes next task I want to ensure the previous dag run has completed. But I am having issues with importing ExternalTaskSensor module.
Is their any alternative path to achieve this ?
Note: Each Dag eg A/B/C has 6- 7 task .Can ExternalTaskSensor check if the last task of dag A has completed before DAG B or C can start.
I Used the below sample code to run dag’s which uses ExternalTaskSensor, I was able to successfully import the ExternalTaskSensor module.
import time
from datetime import datetime, timedelta
from pprint import pprint
from airflow import DAG
from airflow.operators.dagrun_operator import TriggerDagRunOperator
from airflow.operators.dummy_operator import DummyOperator
from airflow.operators.python_operator import PythonOperator
from airflow.sensors.external_task_sensor import ExternalTaskSensor
from airflow.utils.state import State
sensors_dag = DAG(
"test_launch_sensors",
schedule_interval=None,
start_date=datetime(2020, 2, 14, 0, 0, 0),
dagrun_timeout=timedelta(minutes=150),
tags=["DEMO"],
)
dummy_dag = DAG(
"test_dummy_dag",
schedule_interval=None,
start_date=datetime(2020, 2, 14, 0, 0, 0),
dagrun_timeout=timedelta(minutes=150),
tags=["DEMO"],
)
def print_context(ds, **context):
pprint(context['conf'])
with dummy_dag:
starts = DummyOperator(task_id="starts", dag=dummy_dag)
empty = PythonOperator(
task_id="empty",
provide_context=True,
python_callable=print_context,
dag=dummy_dag,
)
ends = DummyOperator(task_id="ends", dag=dummy_dag)
starts >> empty >> ends
with sensors_dag:
trigger = TriggerDagRunOperator(
task_id=f"trigger_{dummy_dag.dag_id}",
trigger_dag_id=dummy_dag.dag_id,
conf={"key": "value"},
execution_date="{{ execution_date }}",
)
sensor = ExternalTaskSensor(
task_id="wait_for_dag",
external_dag_id=dummy_dag.dag_id,
external_task_id="ends",
poke_interval=5,
timeout=120,
)
trigger >> sensor
In the above sample code, sensors_dag triggers tasks in dummy_dag using the TriggerDagRunOperator(). The sensors_dag will wait till the completion of the specified external_task in dummy_dag.
I have an Airflow DAG which I use to submit a spark job and for that I use the SparkSubmitOperator. In the DAG, I have to specify the application JAR that needs to be run. At the moment, it is hardcoded to spark-job-1.0.jar as following:
from airflow import DAG
from airflow.contrib.operators.spark_submit_operator import SparkSubmitOperator
from datetime import datetime, timedelta
args = {
'owner': 'joe',
'start_date': datetime(2020, 5, 23)
}
dag = DAG('spark_job', default_args=args)
operator = SparkSubmitOperator(
task_id='spark_sm_job',
conn_id='spark_submit',
java_class='com.mypackage',
application='/home/ubuntu/spark-job-1.0.jar', <---------
total_executor_cores='1',
executor_cores='1',
executor_memory='500M',
num_executors='1',
name='airflow-spark-job',
verbose=False,
driver_memory='500M',
application_args=["yarn", "10.11.21.12:9092"],
dag=dag,
)
The problem is, the release name will increment and I tried using a wildcard spark-job-*.jar but it didn't work! - Is it possible to use a wildcard, or is there any other way to get around this?
Assuming that there will always be only one file matching the glob /home/ubuntu/spark-job-*.jar, you can do the following:
import glob
...
application=glob.glob('/home/ubuntu/spark-job-*.jar')[0],
...
I want to make a parent DAG with a few child DAGs that get called via the SubDagOperator.
I can only find examples how to dynamically create Subdags in the SubDagOperator task.
However, in this case I want standalone child DAGs that are already defined in a DAG.py file and stitch those together in a parent dag
If I set the SubDAGOperator task with just the Dag Name of the child dag:
task_1 = SubDagOperator(
task_id="task_1",
subdag=child_dag_name,
dag=parent_dag
)
I get the following Error:
NameError: name 'child_dag_name' is not defined
This answer equally relies on knowledge of Python as much it does on having know-how of Airflow
Recall that
python: importing a module means that all top-level (indentation zero) stuff is immediately executed (during import process)
airflow: only those DAG objects are picked by scheduler / webserver that are occur on top-level (indentation zero) of dag-definition file
Keeping above 2 things in mind, here's what you can do
create a helper / utility function in your child_dag.py file to insantiate and return a DAG object for child-dag
use that helper function for instantiating the top-level child-DAG as well as for creating SubDagOperator task
dag_object_builder.py
from typing import Dict, Any
from airflow.models import DAG
def create_dag_object(dag_id: str, dag_params: Dict[str, Any]) -> DAG:
dag: DAG = DAG(dag_id=dag_id, **dag_params)
return dag
child_dag.py
from datetime import datetime
from typing import Dict, Any
from airflow.models import DAG
from src.main.subdag_example import dag_object_builder
default_args: Dict[str, Any] = {
"owner": "my_owner",
"email": ["my_username#my_domain.com"],
"weight_rule": "downstream",
"retries": 1
}
...
def create_child_dag_object(dag_id: str) -> DAG:
my_dag: DAG = dag_object_builder.create_dag_object(
dag_id=dag_id,
dag_params={
"start_date": datetime(year=2019, month=7, day=10, hour=21, minute=30),
"schedule_interval": None,
"max_active_runs": 1,
"default_view": "graph",
"catchup": False,
"default_args": default_args
}
)
return my_dag
my_child_dag: DAG = create_child_dag_object(dag_id="my_child_dag")
parent_dag.py
from datetime import datetime
from typing import Dict, Any
from airflow.models import DAG
from airflow.operators.subdag_operator import SubDagOperator
from src.main.subdag_example import child_dag
from src.main.subdag_example import dag_object_builder
default_args: Dict[str, Any] = {
"owner": "my_owner",
"email": ["my_username#my_domain.com"],
"weight_rule": "downstream",
"retries": 1
}
my_parent_dag: DAG = dag_object_builder.create_dag_object(
dag_id="my_parent_dag",
dag_params={
"start_date": datetime(year=2019, month=7, day=10, hour=21, minute=30),
"schedule_interval": None,
"max_active_runs": 1,
"default_view": "graph",
"catchup": False,
"default_args": default_args
}
)
...
my_subdag_task: SubDagOperator = SubDagOperator(
task_id="my_subdag_task",
dag=my_parent_dag,
subdag=child_dag.create_child_dag_object(dag_id="my_subdag")
)
If your intention is to link-up DAGs together and you don't have any particular requirement that necessitates using a SubDagOperator, then I would suggest using the TriggerDagRunOperator instead since SubDags have their share of nuisances.
Read more about it here: Wiring top-level DAGs together
I want to be able to publish and trigger a DAG object from my code which is not in control of scheduler (viz. $AIRFLOW_HOME/dags folder)
My last resort would be to programmatically create a py file containing the DAG definition that I want to publish and save this file to the $AIRFLOW_HOME/dags folder.
I'm sure it should be easier than that.
Below is what I've tried.
import airflow
from airflow import DAG
from datetime import timedelta
from airflow.models import DagPickle
from airflow.operators.dummy_operator import DummyOperator
from airflow.utils.db import provide_session
#provide_session
def submit_dag(session=None):
args = {
'owner': 'airflow',
'start_date': airflow.utils.dates.days_ago(2)
}
dag = DAG(
dag_id='sample', default_args=args,
schedule_interval=None, start_date=airflow.utils.dates.days_ago(2),
dagrun_timeout=timedelta(minutes=60))
task = DummyOperator(task_id='one', dag=dag)
dag_pickle = DagPickle(task)
session.add(dag_pickle)
session.commit()
submit_dag()
The above code does create entries in dag_pickle table but how do I publish and later trigger this dag?
I can do pickle.dump(dag,open(DAGS_FOLDER/pickled_dags,'wb')) and have a file in DAGS FOLDER that would pickle.load(DAGS_FOLDER/pickled_dags)
I mostly see Airflow being used for ETL/Bid data related jobs. I'm trying to use it for business workflows wherein a user action triggers a set of dependent tasks in future. Some of these tasks may need to be cleared (deleted) based on certain other user actions.
I thought the best way to handle this would be via dynamic task ids. I read that Airflow supports dynamic dag ids. So, I created a simple python script that takes DAG id and task id as command line parameters. However, I'm running into problems making it work. It gives dag_id not found error. Has anyone tried this? Here's the code for the script (call it tmp.py) which I execute on command line as python (python tmp.py 820 2016-08-24T22:50:00 ):
from __future__ import print_function
import os
import sys
import shutil
from datetime import date, datetime, timedelta
from airflow import DAG
from airflow.operators.bash_operator import BashOperator
execution = '2016-08-24T22:20:00'
if len(sys.argv) > 2 :
dagid = sys.argv[1]
taskid = 'Activate' + sys.argv[1]
execution = sys.argv[2]
else:
dagid = 'DAGObjectId'
taskid = 'Activate'
default_args = {'owner' : 'airflow', 'depends_on_past': False, 'start_date':date.today(), 'email': ['fake#fake.com'], 'email_on_failure': False, 'email_on_retry': False, 'retries': 1}
dag = DAG(dag_id = dagid,
default_args=default_args,
schedule_interval='#once',
)
globals()[dagid] = dag
task1 = BashOperator(
task_id = taskid,
bash_command='ls -l',
dag=dag)
fakeTask = BashOperator(
task_id = 'fakeTask',
bash_command='sleep 5',
retries = 3,
dag=dag)
task1.set_upstream(fakeTask)
airflowcmd = "airflow run " + dagid + " " + taskid + " " + execution
print("airflowcmd = " + airflowcmd)
os.system(airflowcmd)
After numerous trials and errors, I was able to figure this out. Hopefully, it will help someone. Here's how it works: You need to have an iterator or an external source (file/database table) to generate dags/task dynamically through a template. You can keep the dag and task names static, just assign them ids dynamically in order to differentiate one dag from the other. You put this python script in the dags folder. When you start the airflow scheduler, it runs through this script on every heartbeat and writes the DAGs to the dag table in the database. If a dag (unique dag id) has already been written, it will simply skip it. The scheduler also look at the schedule of individual DAGs to determine which one is ready for execution. If a DAG is ready for execution, it executes it and updates its status.
Here's a sample code:
from airflow.operators import PythonOperator
from airflow.operators import BashOperator
from airflow.models import DAG
from datetime import datetime, timedelta
import sys
import time
dagid = 'DA' + str(int(time.time()))
taskid = 'TA' + str(int(time.time()))
input_file = '/home/directory/airflow/textfile_for_dagids_and_schedule'
def my_sleeping_function(random_base):
'''This is a function that will run within the DAG execution'''
time.sleep(random_base)
def_args = {
'owner': 'airflow',
'depends_on_past': False,
'start_date': datetime.now(), 'email_on_failure': False,
'retries': 1, 'retry_delay': timedelta(minutes=2)
}
with open(input_file,'r') as f:
for line in f:
args = line.strip().split(',')
if len(args) < 6:
continue
dagid = 'DAA' + args[0]
taskid = 'TAA' + args[0]
yyyy = int(args[1])
mm = int(args[2])
dd = int(args[3])
hh = int(args[4])
mins = int(args[5])
ss = int(args[6])
dag = DAG(
dag_id=dagid, default_args=def_args,
schedule_interval='#once', start_date=datetime(yyyy,mm,dd,hh,mins,ss)
)
myBashTask = BashOperator(
task_id=taskid,
bash_command='python /home/directory/airflow/sendemail.py',
dag=dag)
task2id = taskid + '-X'
task_sleep = PythonOperator(
task_id=task2id,
python_callable=my_sleeping_function,
op_kwargs={'random_base': 10},
dag=dag)
task_sleep.set_upstream(myBashTask)
f.close()
From How can I create DAGs dynamically?:
Airflow looks in you [sic] DAGS_FOLDER for modules that contain DAG objects in their global namespace, and adds the objects it finds in the DagBag. Knowing this all we need is a way to dynamically assign variable in the global namespace, which is easily done in python using the globals() function for the standard library which behaves like a simple dictionary.
for i in range(10):
dag_id = 'foo_{}'.format(i)
globals()[dag_id] = DAG(dag_id)
# or better, call a function that returns a DAG object!
copying my answer from this question. Only for v2.3 and above:
This feature is achieved using Dynamic Task Mapping, only for Airflow versions 2.3 and higher
More documentation and example here:
Official Dynamic Task Mapping documentation
Tutorial from Astronomer
Example:
#task
def make_list():
# This can also be from an API call, checking a database, -- almost anything you like, as long as the
# resulting list/dictionary can be stored in the current XCom backend.
return [1, 2, {"a": "b"}, "str"]
#task
def consumer(arg):
print(list(arg))
with DAG(dag_id="dynamic-map", start_date=datetime(2022, 4, 2)) as dag:
consumer.expand(arg=make_list())
example 2:
from airflow import XComArg
task = MyOperator(task_id="source")
downstream = MyOperator2.partial(task_id="consumer").expand(input=XComArg(task))
The graph view and tree view are also updated:
Relevant issues here:
https://github.com/apache/airflow/projects/12