john-jac commented on issue #14583:
URL: https://github.com/apache/airflow/issues/14583#issuecomment-823574467


   > I'm not sure if there is one (or even range) universal values to be 
considered healthy that match all.
   > For example `dag_processing.total_parse_time |  Seconds taken to scan and 
import all DAG files once`
   > 
   > What is the appropriate value that can be recommended? Consider an Airflow 
instance with 3 dags and instance with 3000 Dags.
   
   What would be helpful is recommended guidelines, regardless of the number of 
DAGs (as the number of DAGs don't linearly equate to parse time given other 
variables like DAG complexity).  For example, if Total Parse Time > x minutes, 
how does that effect task execution?  Does that mean that a user should add 
more resources to the scheduler?  If so, should they add memory/cpu/both? 
   
   Another option would be mapping values like "if you have metric x, then you 
should have a configuration value of x".  Going back to the parse time example, 
if I have a parse time of 2 minutes then I should set processor_poll_interval 
to X and min_file_process_interval to Y


-- 
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.

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


Reply via email to