Found a bug during performance benchmarking where there were duplicates of all scheduler airflow processes running on a scheduler container (so 1 container had 2 schedulers running).
This PR fixes this issue and also fixes issue where worker and webserver processes are not treated as essential (meaning failed conditions would not cause container to exit).
Tested deploying image locally and deploying to mwaa dev environment.
Description of changes:
Found a bug during performance benchmarking where there were duplicates of all scheduler airflow processes running on a scheduler container (so 1 container had 2 schedulers running).
This PR fixes this issue and also fixes issue where worker and webserver processes are not treated as essential (meaning failed conditions would not cause container to exit).
Tested deploying image locally and deploying to mwaa dev environment.