Error in worker logs: KeyError: 'task_name'

[nicolamarangoni]

Apache Airflow version

3.0.0

If "Other Airflow 2 version" selected, which one?

No response

What happened?

I'm testing a fresh 3.0.0 installation and started the worker with the command airflow celery worker, after the usual logs, I see the wollowing lines in the logs of the worker service:

[tasks]
. execute_workload
--- Logging error ---
Traceback (most recent call last):
File "/usr/local/lib/python3.12/logging/__init__.py", line 464, in format
return self._format(record)
File "/usr/local/lib/python3.12/logging/__init__.py", line 460, in _format
return self._fmt % values
KeyError: 'task_name'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/usr/local/lib/python3.12/logging/__init__.py", line 1160, in emit
msg = self.format(record)
....

The service looks healthy however, but the installation has no DAGs at the moment therefore I didn't testet some DAG executions yet.

What you think should happen instead?

The KeyError: 'task_name' should not happen.

How to reproduce

Install a fresh 3.0.0 from the greenfield and launch the worker with the command:
airflow celery worker

Operating System

Official docker image,

Versions of Apache Airflow Providers

No response

Deployment

Other

Deployment details

Official docker image running on AWS ECS.

Anything else?

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[amoghrajesh]

@nicolamarangoni can you help with your installation steps?

[nailo2c]

I can't reproduce it either, could you share your celery version?

[nicolamarangoni]

My installation steps consist in a huge terraform deployment.
However, I think that the error happen here:
celery/apps/worker.py, line 176: https://github.com/celery/celery/blob/main/celery/apps/worker.py
This is the code:

    def on_consumer_ready(self, consumer):
        signals.worker_ready.send(sender=consumer)
        logger.info('%s ready.', safe_str(self.hostname))

It's related to this issue: #42136 (comment)

[nicolamarangoni]

After setting AIRFLOW__CORE__EXECUTION_API_SERVER_URL I still have the same erro but in another position:
https://github.com/celery/celery/blob/main/celery/worker/consumer/connection.py line 22:

    def start(self, c):
        c.connection = c.connect()
        info('Connected to %s', c.connection.as_uri())

[nicolamarangoni]

Thiese are the logs:

extract-2025-07-21T15_23_57.822Z.csv

[amoghrajesh]

@nicolamarangoni I merged a recent fix related to this: Handle invalid execution API urls in supervisor #53082

Could you pull in that commit in your deployment and try again?

[nicolamarangoni]

@amoghrajesh Thank you for the hint. I downloaded and substituted the supervisor.py file into my image and restarted the ECS services. I still see the error (see logs in attachment). However I think it's only a false positive because the ECS service of the worker is running.
Currently my issue is this one: #42136 (comment)
I don't know if the 2 issues are related.

[amoghrajesh]

@nicolamarangoni could you check what's set as config for base_url?

[nicolamarangoni]

@amoghrajesh I have this in my config:

airflow config list
....
base_url = https://dev.myinternal.company.domain.tech
....

It's the URL of the web UI

[amoghrajesh]

@nicolamarangoni looking more carefully, this could be a logging error

Your logging formatter is trying to substitute a field %(task_name)s, but the log record being emitted doesn’t always have that attribute.

Check your task code for potential usage of task_name or if you could share your dag code

[nicolamarangoni]

@amoghrajesh I did a fresh install of Airflow. There are currently no DAGs at all in the instance.
I still see the same error at startup in the worker logs.
This error is not related to any DAG/task run. It appears at startup.