Getting the following error when attempting to run my Flow

Failed to load and execute Flow's environment: KeyError('my-flow-name')

The Flow is uses DockerStorage, and I’ve got an image containing only this Flow; are there some

prefect

cli invocations I can do inside the container to debug this KeyError?

Hi @Joseph Haaga - can you provide some more context to this and perhaps some minimum amount of code?

yeah it’s a little hard since we’ve actually broken out the

build

and

register

step into separate scripts (we’re trying to write a gold-standard template repo for devs to fork when writing Flows), but this is basically what’s going on

# src/flow.py
from prefect import Flow, task


@task
def task_one():
    print("hi")
    return 1


@task
def task_two(x):
    return x + 1


with Flow("my-flow-name") as flow:
    a = task_one()
    b = task_two(a)



# build_flow.py
from src.flow import flow
from prefect.environments.storage import Docker

storage = Docker(
    dockerfile='Dockerfile',
    registry_url="<http://myregistry.com|myregistry.com>",
    image_name="my-flow-name",
)
storage.add_flow(flow)
storage.build()



# register_flow.py
flow.storage = Docker(
    registry_url=args.repo,
    image_name="my-flow-name",
    image_tag=args.tag,
)

flow.environment = DaskKubernetesEnvironment(
    min_workers=1,
    max_workers=3,
    worker_spec_file="worker_spec.yaml"
)
flow.register(project_name=args.project, build=False, labels=labels)

@Joseph Haaga it looks like you've got 2 storage objects between those scripts, 1 in

build_flow.py

and another in

register_flow.py

- and it's a little tough to tell but it looks like the actual flow you're registering doesn't ever have its storage built

yeah I share your suspicion - we’ve consolidated everything into

register_flow.py

and are now encountering another (albeit more encouraging) error:

Failed to load and execute Flow's environment: ModuleNotFoundError("No module named 'src'")

We usually just write a

setup.py

and

RUN pip install -e .

to circumvent this issue, but curious if anyone knows a better way to resolve this!

Hm, are you explicitly creating the

src

directory in your dockerfile? if not, it won't be present when you try to run your flow in the container

oh yeah we’re COPYing src in to the container

So is this failing at the build step or when running?

when running - everything builds and registers nicely, and these error messages are pulled from the UI on page for the FLow Run

gotcha, that's a little odd - do you have an

__init__.py

file in that directory?