jscheffl commented on PR #65943:
URL: https://github.com/apache/airflow/pull/65943#issuecomment-4444389148

   @wjddn279 @diogosilva30 Thanks for all valid discussions. Really cool. So 
improvements in general welcome.
   
   @wjddn279 The optimization that was added to LocalExecutor is still on my 
(long) bucket list. I even left a comment in 
https://github.com/apache/airflow/blob/main/providers/edge3/src/airflow/providers/edge3/cli/worker.py#L460
 to remind me whenever I see the code again. Feel free to raise a PR (to be 
faster than me)!
   
   In general I am mostly wondering why at @diogosilva30 this error is happeing 
often while in our environment we do not see this at all. Must be some 
environmental side effect. So for the moment I'd accept it as an option but as 
of memory overhead and time only optional. Main line should be kept as is until 
more errors reported.
   
   I am not sure if for many environments a pre-fork makes sense. It might be 
another (tuning) option. I assume the typical pattern for celery is different 
than for Edge... at least I think. I#d in general favor simplicity over 
complexity. And a pre-fork pool might add a level of complexity making it 
harder to maintain or adding bugs.
   
   My actual "dream" would be that we could make supervisor bing mostly async 
such that we do not need to fork() or spawn another process at all just to run 
another process... but just have a set of in-process supervisor instances which 
themself just fork the execution for the task. Because today we fork worker -> 
supervisor -> task execution.


-- 
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