@Marvin: I have a flow that uses

result_storage=GCS(bucket_path="path/to/my/results/")

. This is causing issues during unit testing, as prefect is trying to upload to GCS. How can I switch off this behaviour at test time? FYI, I am using the prefect_test_harness

Here's a tip that might help: It seems that Prefect allows you to control the persistence of results by setting

persist_result

to

False

. This could potentially be used during testing to avoid storing results. Additionally, Prefect documentation mentions that starting from version

2.10.3

, it is possible to mock values in the runtime via an environment variable. To disable result storage during testing with the

prefect_test_harness

, you could possibly manipulate the runtime context or set specific environment variables to mock the results storage behavior. However, I haven't found a direct answer in the search results that explicitly states how to switch off result storage during unit testing with

prefect_test_harness

. Since I am newly acquainted with Prefect version 2.x and given the gravity of the situation, I recommend checking the current documentation on Results and Runtime Context for the latest practices regarding the disabling and mocking of result storage during testing.

@Marvin here is a simple example. I've defined a flow in

run.py

@flow(result_storage=GCS(bucket_path="path/to/my/results/"))
def my_flow():
    # do stuff here...
    return result

And in

test_run.py

i've defined a test

from run import my_flow

def test_run():
    result = my_flow()
    # assertions here...

I want my test to somehow overwrite the default value of

result_storage

. How can I do this?

The code example provided doesn't directly address how to override

result_storage

in tests, unfortunately. However, I can suggest an approach based on best practices. In your

test_run.py

, you could modify the default storage setting by using Python's built-in capabilities to modify the class or instance attributes. Here's a simplified example of how you might do it:

python
from run import my_flow
from prefect import flow
from prefect.orion.schemas.storage import Memory

@flow(result_storage=Memory())
def test_my_flow():
    # Calls the original my_flow inside the test flow,
    # where result_storage has been overridden to use Memory.
    result = my_flow()
    return result

def test_run():
    # You can now run the test_my_flow which has the Memory result_storage.
    result = test_my_flow()
    # Place your assertions here...

In the

test_my_flow

, we're using the

Memory

storage which is suitable for testing purposes as it doesn't persist results outside of the Python process. Please ensure to check the latest Prefect documentation on Defining a Flow and Result Storage to make sure this approach is compatible with your version of Prefect and for any additional context that may apply.

If anyone is interested, I've gone with ChatGPT's suggestion of

from run import my_flow

def test_run():
    my_flow.result_storage = None
    result = my_flow()