How to schedule local scripts without keeping flow.serve() running or using Docker?

[TomMonkeyMan]

Hi Prefect team and community,

I’m new here and I was trying to run local Python flows (i.e., scripts on my filesystem, no Git/Docker) using Prefect 3’s (version 3.6.1+2.gd96416c96b) deployment system, but I’m hitting a usability gap:

✅ flow.serve() works perfectly for development — it registers the deployment and starts an embedded worker.

❌ But as soon as I Ctrl+C the python deploy.py process, new flow runs become "Late" because the embedded worker is gone ( Although I can do actions on UI, but the flow will not be running, another discussion also reporting this: #16150 ). I don't know whether this behavior is as expected.

🚫 The new prefect.deploy(...) / flow.deploy() API requires either a Docker image or remote storage (e.g., GitHub), which is overkill for simple local automation.

I’d like a way to:
Register a deployment from a local .py file (once),
Maybe start a long-running worker (e.g., via nohup prefect worker start --pool local-pool),
Trigger runs via UI/CLI without keeping the original deploy script alive.

I checked this was possible in earlier versions with Deployment.build_from_flow(...).apply() + process work pool. but looks like it's not an available approach in Prefect 3.6+?

May I ask what is the best practice for local script scheduling & deployment?

Thanks!

[zzstoatzz]

hi @TomMonkeyMan - have you checked this?

if you use from_source and refer to a local filepath, you don't need remote storage or an image

e.g.

from pathlib import Path
from prefect import flow


@flow(log_prints=True)
def my_flow(name: str = "world"):
    print(f"Hello {name}! I'm a flow from a Python script!")


if __name__ == "__main__":
    my_flow.from_source(
        source=str(Path(__file__).parent),
        entrypoint="example.py:my_flow",
    ).deploy(
        name="my-deployment",
        parameters=dict(name="Marvin"),
        work_pool_name="local",
    )

[zzstoatzz]

this works when the worker has filesystem access to the source/entrypoint, but otherwise you need docker/remote fs

[TomMonkeyMan]

Thanks for the guidance! I initially tried using prefect.deployments.Deployment from the older documentation, but it seems this has been removed in recent versions (I got an import error). Because of that, I didn’t explore further options like from_source. I'll go for a quick test then!

[TomMonkeyMan]

wonderful! it works well. thanks so much for the quick reply! @zzstoatzz

one quick extra question: Is flow.serve() designed for development only?

[zzstoatzz]

Is flow.serve() designed for development only?

no! it just happens to be the easiest way to make a deployment

check this doc on running serve in a docker container

which you could easily extend to the entrypoint of a docker compose or a kubernetes setup

[TomMonkeyMan]

Got it! That makes sense to me. great work btw! I'm planning to migrate some of my flows from airflow to prefect, it's much more convenient!