airflow-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sophie Herrmann <>
Subject Re: Taskes not running with celery gevent/eventlet worker
Date Tue, 07 Apr 2020 14:26:37 GMT

Thanks for the quick answer! Sadly that is not an option in our setup.

It seems to me Airflow 1.10.9 does not properly create the gevent celery worker. When I check
the workers in Flower's Web UI (localhost:5555) I see the worker but when clicking on the
worker's name  I just get "Unknown worker 'celery@dfa2e29c0b96'".

Did you find any other workaround not requiring celery CLI? Do you know if there is an active
issue on this? Do you know if it will be fixed in Airflow 2.0?


From: Aaron Grubb <>
Sent: Tuesday, April 7, 2020 3:05 PM
To: <>
Subject: RE: Taskes not running with celery gevent/eventlet worker

I think this is the same issue I encountered. Try running the celery workers manually (i.e.
using the celery command to launch them).

From: Sophie Herrmann <>
Sent: Tuesday, April 7, 2020 8:21 AM
Subject: Taskes not running with celery gevent/eventlet worker

Hi everyone,

I am using Airflow 1.10.9 with the CeleryExecutor (docker image based on puckel/docker-airflow)
and I am trying to switch from prefork to gevent/eventlet pool for one worker.
With prefork everything is working as expected but when I use a gevent/eventlet worker instead,
it seems tasks are not picked up.

I checked the logs and the scheduler queues the task and sends them to the worker. The worker
then receives the task but after that nothing else happens (execution is not started).

Below the last logs I get from the worker:

[2020-04-07 12:13:13,376: INFO/MainProcess] Received task: airflow.executors.celery_executor.execute_command[32ba869c-80b8-47b5-8eca-aadd64c7fd7c]

[2020-04-07 12:13:13,614: INFO/MainProcess] Scaling up 1 processes.

[2020-04-07 12:13:13,630: DEBUG/MainProcess] TaskPool: Apply <function _fast_trace_task
at 0x7fb8af20fdd0> (args:('airflow.executors.celery_executor.execute_command', '32ba869c-80b8-47b5-8eca-aadd64c7fd7c',
{'lang': 'py', 'task': 'airflow.executors.celery_executor.execute_command', 'id': '32ba869c-80b8-47b5-8eca-aadd64c7fd7c',
'shadow': None, 'eta': None, 'expires': None, 'group': None, 'retries': 0, 'timelimit': [None,
None], 'root_id': '32ba869c-80b8-47b5-8eca-aadd64c7fd7c', 'parent_id': None, 'argsrepr': "[['airflow',
'run', 'jb-10baee25-8b57-46d1-bb9b-75380504df33', 'cancel_sensor', '2020-04-07T12:13:09.378481+00:00',
'--local', '--pool', 'default_pool', '-sd', '/usr/local/airflow/dags/']]",
'kwargsrepr': '{}', 'origin': 'gen76@f2822480e988', 'reply_to': 'e07ad7c8-e252-3533-a528-11b3eb712a22',
'correlation_id': '32ba869c-80b8-47b5-8eca-aadd64c7fd7c', 'hostname': 'celery@9ede4f9fb1d1',
'delivery_info': {'exchange': '', 'routing_key': 'sensor', 'priority': 0, 'redelivered': None},
'args': [['airflow', 'run', 'jb-10baee25-8b57-46d1-bb9b-75380504df33', 'cancel_sensor',...

In the Airflow UI the task is set to queued and does not change.

Any idea what could be the problem?



View raw message