wjddn279 commented on PR #56044:
URL: https://github.com/apache/airflow/pull/56044#issuecomment-3426442031

   @potiuk 
   Sounds good. I'll think about how to actually verify this. In addition to 
ORM initialization, I'll also check for any unintended query operations in 
child processes.
   
   Before that, I have a question. Does the DB usage pattern we've classified 
for each component (whether parent/child processes use DB) reflect a design 
philosophy? For example, in the dag-processor, child processes don't connect to 
the DB but instead parse files and pass objects to the parent process. I'd like 
to know if this is based on a design principle of "dag processor child 
processes should never directly connect to the DB," or if it's just how it's 
currently implemented and could change in the future.
   
   The reason I'm asking is that our current verification might become invalid 
later and could potentially lead to new bugs. If this DB usage pattern is not 
absolute, then the uniform prevention approaches mentioned by @tirkarthi 
(NullPool, change driver, recreate engine) could be better long-term solutions.


-- 
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