Hi! Any advice in running debug / troubleshooting runs of a flow locally without it being registered on Prefect Cloud? Flow is being run like so

python flows/flow.py

and parameters are filled in via

if __name__ == "__main__":
    my_flow(
        parameter_1=value_1,
        parameter_2=value_2,
        parameter_3=value_3,
    )

I do not want

my_flow

run to be registered on prefect cloud due to many different reasons 😞 How can I achieve this?

You can always spin a local server with

prefect server start

You could use a separate profile, i.e.

prefect profile create local

and then

prefect profile use local

. Or, something like this should work as well:

# your code here 

if __name__ == "__main__":
    from prefect.settings import temporary_settings, PREFECT_API_URL

    with temporary_settings(restore_defaults={PREFECT_API_URL}):
        my_flow(
            parameter_1=value_1,
            parameter_2=value_2,
            parameter_3=value_3,
        )

Thank you for suggestions, but this will not work because the same flow run I do not want to be registered on the server, fetches a few secret blocks from it. The way I see it, is to 1. Spin up my local prefect server for development 2. Sync all required secret blocks and other to it 3. Do my tests 4. Nuke the dev server --- Repeat for every dev session.. facepalm

I have one other idea that might work; I'm just going to give it a quick test

@jpuris I personally do not use any prefect blocks. Any block you need can be setup using the python prefect client in memory without having to store it in the server.

It is the secrets we use across many different flows, hence it is out of convenience and to avoid stray .env files

Yeah in that case. I just put all the secrets into a cloud key vault.

Indeed. We sort of are leaning towards doing something similar, but no plans currently 👍

Just a question, are the flows in multiple repo?

Flows are currently on local fs the prefect agent is running on

you can use the prefect test harness to easily use a local temporary server rather than the prefect cloud:

from prefect.testing.utilities import prefect_test_harness

with prefect_test_harness():
    my_flow()

I've never heard of this one! Thank you @Austin Weisgrau!