Hi, I'm kind of new to prefect, so it's probably a dumb question, but I wanted to know if anyone has tried to do something similar to me. I created a python class to extend the prefect.Flow. Among other things, it overrides the run function (mainly to enforce a few company rules for logging and to setup the context on runtime for each flow), but now I'm having dificulty running my code on a local agent, since the agent does not call flow.run. Is there a way for me to configure those things on my flow class without needing to create my own FlowRunner?

Hey @Leonardo Rocha, would you be able to give me an idea of how you subclassed the Flow? Normally people would add configuration and then call the

super().run()

. Not sure everything will hold well when you move to

flow.register()

though. Maybe you can just add your filter to the Prefect logger?

class My_Flow(prefect.Flow):

    

def __init__(self, name, schedule=None, **kwargs):

        

...

        

super(My_Flow, self).__init__(name, schedule, **kwargs)

    

def run(self, **kwargs):

        

...

        

with prefect.context(**context_kwargs):

            

return super(My_Flow, self).run(**kwargs)

    

def add_task(self, task: prefect.core.task.Task) -> prefect.core.task.Task:

        

task = super(My_Flow, self).add_task(task)

        

...

        

return task

What are you setting in the context? Is it environment variables?

Kind of, I have an internal API that I call on My_Flow.run that provides me with the variables needed on the context for each flow.

Is this something that can’t be done with the RunConfigs , you can set environment variables on the Flow through this. You can also set them on the agent and the environment variables will be carried over to the Flow.

I'll look into it, thank you very much