Hi,
Yes I found a workaround.
Initially I had problems because I was using SQLite and I had lock problems that led to ugly SQLAlchemy errors.
After changing to Postgres SQL I got rid of lock errors but I had same issue.
So I tried to create a simple script (my workflow has 14 processes)
from common import submit
submit(...
submit(...
submit(...
...
x8 times
and same issue, after the 4th or 5th submit call, the HTTP response of of the call to the api takes ages (minutes) to return (or never returns)
However, if I run separate threads each calling submit, everything works fine :S...
I know it does not make much sense but before implementing threads I did first experiments in serial and it was failing