potiuk commented on PR #60033:
URL: https://github.com/apache/airflow/pull/60033#issuecomment-3756252508

   > Why do we need to add fallbacks to many configs? Did they not work without 
the fallbacks?
   
   @uranusjr  -> This is manifestation of doing too much and in a wrong 
sequence during imports. The problem is that currently the sdk config does not 
have the defaults "hard-coded" fallbacks for celery and kubernetes providers 
(@amoghrajesh deliberately skipped them when moving conf to task.sdk) - because 
initialization of executors (that already need those fallbacks) is happening 
during importing airflow and before provider's manager initializes 
configuration retrieved from providers. This has been discussed and I proposed 
to bring it back as a temporary solution before we finish task isolation and 
will be able to do explicit initialization rather than import-based 
initialization: 
https://github.com/apache/airflow/pull/60074#issuecomment-3710183909
   
   But maybe there are other proposals?
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to