@Marvin does the virtual environment you call

flow.from_source.deploy()

need to be the same python version and have the same dependencies installed as the source location? How would you make that happen if it's not the case?

From the documentation, it looks like you can skip having a

flow.from_source

call when building an image with

flow.deploy

. Prefect manages the flow's source code location within the image and loads it from there during execution. However, it is recommended that the Python environments match due to dependency compatibility. Here's the relevant guide. If the environments diverge, you can use

prefect.yaml

to define the deployment configurations including dependencies. Running

prefect deploy

from the root of your repo will set this up. To make sure the environments align, you should: 1. Ensure the same Python version is used in both environments. 2. Synchronize the dependencies using a

requirements.txt

or

environment.yml

file shared across environments. You can read more about deploying flows and managing environments in the Prefect deployment guide.