Catch Scrapy exception when crawling from Airflow - scrapy

Question

I'm trying to catch the exception that occurs on my spider in a manner that I can mark the task instance as failed. Currently the task finishes and is marked as succeeded. I'm calling the crawl() from PythonOperator in Airflow, as follow:
with DAG(
'MySpider',
default_args=default_args,
schedule_interval=None) as dag:
t1 = python_task = PythonOperator(
task_id="crawler_task",
python_callable=run_crawler,
op_kwargs=dag_kwargs
)
Here is my run_crawler() method:
def run_crawler(**kwargs):
project_settings = set_project_settings({
'FEEDS': {
f'{kwargs["bucket"]}%(time)s.{kwargs["format"]}': {
'format': kwargs["format"],
'encoding': 'utf8',
'store_empty': kwargs["store_empty"]
}
}
})
print("Project settings: ")
pprint(project_settings.attributes.items())
set_connection("airflow", kwargs["gcs_connection_id"])
process = CrawlerProcess(project_settings)
process.crawl(spider.MySpider)
print("Starting crawler...")
process.start()
When running, I'm having problems with GCS credentials, which leads me to an Exception, as follow:
google.auth.exceptions.DefaultCredentialsError: The file /tmp/file_my_credentials.json does not have a valid type. Type is None, expected one of ('authorized_user', 'service_account', 'external_account', 'external_account_authorized_user', 'impersonated_service_account', 'gdch_service_account').
{logging_mixin.py:115} WARNING - [scrapy.statscollectors] INFO: Dumping Scrapy stats:
{'downloader/request_bytes': 21087,
'downloader/request_count': 68,
'downloader/request_method_count/GET': 68,
'downloader/response_bytes': 1863876,
'downloader/response_count': 68,
'downloader/response_status_count/200': 68,
'elapsed_time_seconds': 25.647386,
'feedexport/failed_count/GCSFeedStorage': 1,
'httpcompression/response_bytes': 9212776,
'httpcompression/response_count': 68,
'item_scraped_count': 66,
'log_count/DEBUG': 136,
'log_count/ERROR': 1,
'log_count/INFO': 10,
'log_count/WARNING': 3,
'memusage/max': 264441856,
'memusage/startup': 264441856,
'request_depth_max': 1,
'response_received_count': 68,
'scheduler/dequeued': 68,
'scheduler/dequeued/memory': 68,
'scheduler/enqueued': 68,
'scheduler/enqueued/memory': 68,
[2032-13-13, 09:04:28 UTC] {engine.py:389} INFO - Spider closed (finished)
[2032-13-13, 09:04:28 UTC] {logging_mixin.py:115} WARNING -
[scrapy.core.engine] INFO: Spider closed (finished)
[2032-13-13, 09:04:28 UTC] {python.py:173} INFO - Done. Returned value was: None
[2032-13-13, 09:04:28 UTC] {taskinstance.py:1408} INFO - Marking task as SUCCESS. dag_id=MySpider, task_id=crawler_task, execution_date=2032-13-13, start_date=2032-13-13, end_date=2032-13-13
[2032-13-13, 09:04:28 UTC] {local_task_job.py:156} INFO - Task exited with return code 0
[2032-13-13, 09:04:28 UTC] {local_task_job.py:279} INFO - 0 downstream tasks scheduled from follow-on schedule check
As you can see, even having this exception, the task itself is marked as "SUCCESS". Is it possible to catch it in order to mark as FAILED, then we can follow it on airflow (Composer) interface?
Thank you

Answer 1

I don't understand why in this case the exception doesn't break the task.
You can add a try except in the run_crawler method and then raise you own exception in the except bloc :
import logging
def run_crawler(**kwargs):
class CustomException(Exception):
pass
try:
project_settings = set_project_settings({
'FEEDS': {
f'{kwargs["bucket"]}%(time)s.{kwargs["format"]}': {
'format': kwargs["format"],
'encoding': 'utf8',
'store_empty': kwargs["store_empty"]
}
}
})
print("Project settings: ")
pprint(project_settings.attributes.items())
set_connection("airflow", kwargs["gcs_connection_id"])
process = CrawlerProcess(project_settings)
process.crawl(spider.MySpider)
print("Starting crawler...")
process.start()
except Exception as err:
logging.error("Error in the Airflow task !!!!!", err)
raise CustomException("Error in the Airflow task !!!!!", err)
In this case when your custom exception will be raised, it will break the Airflow and mark it as failed.