Hello, I’m trying to run the Getting Started wine classification example on my Flyte setup, but have faced several errors. My setup is consisted of: Jupyter Hub + Flyte on the same K8S cluster. The codes are

from flytekit.remote import FlyteRemote
from flytekit.configuration import Config, PlatformConfig, ImageConfig, SerializationSettings

remote = FlyteRemote(
    config=Config(
        platform=PlatformConfig(
            endpoint="dns:///flyte-backend-flyte-binary-grpc.namespace.svc.cluster.local:8089",
            insecure=True,
            insecure_skip_verify=True,
        )
    ),
    default_project="flytesnacks",
    default_domain="development",
    data_upload_location="<s3a://flyte-storage/>"
)

HYPERPARAMS = {
    "hyperparameters": {"C": 0.1}
}

flyte_entity = training_workflow

flyte_workflow = remote.register_workflow(
    entity=flyte_entity,
    serialization_settings=SerializationSettings(image_config=ImageConfig.auto_default_image()),
    version="v1"
)

launch_plan = LaunchPlan.get_or_create(
    workflow=flyte_workflow,
    name="wine_classification_lp",
    default_inputs=HYPERPARAMS,
)
remote.register_launch_plan(launch_plan, "v1", project="flytesnacks", domain="development")

And the workflow failed right at the first task

get_data

, here is the error I’ve got

[1/1] currentAttempt done. Last Error: USER::rapped(*args, **kwargs)                        │
│                                                                              │
│ /usr/local/lib/python3.8/site-packages/flytekit/bin/entrypoint.py:347 in     │
│ _execute_task                                                                │
│                                                                              │
│ ❱ 347 │   │   _task_def = resolver_obj.load_task(loader_args=resolver_args)  │
│                                                                              │
│ /usr/local/lib/python3.8/site-packages/flytekit/core/utils.py:295 in wrapper │
│                                                                              │
│ ❱ 295 │   │   │   │   return func(*args, **kwargs)                           │
│                                                                              │
│ /usr/local/lib/python3.8/site-packages/flytekit/core/python_auto_container.p │
│ y:235 in load_task                                                           │
│                                                                              │
│ ❱ 235 │   │   task_module = importlib.import_module(name=task_module)  # typ │
│                                                                              │
│ /usr/local/lib/python3.8/importlib/__init__.py:127 in import_module          │
│                                                                              │
│ ❱ 127 │   return _bootstrap._gcd_import(name[level:], package, level)        │
│ in _gcd_import:1011                                                          │
│ in _sanity_check:950                                                         │
╰──────────────────────────────────────────────────────────────────────────────╯
ValueError: Empty module name

Any suggestion on which might have caused this error and how to fix it? I’m suspecting that I haven’t registered the workflow or the launch plan properly.

Hi Erik! You need to specify the module name and source path. Here's an example: https://docs.flyte.org/projects/cookbook/en/latest/auto/case_studies/feature_engineering/feast_integration/Feast_Flyte_Demo.html#register-the-code.

Thank you very much for your response. That works, and I no longer face the

ValueError

issue. However, I’m dealing with an S3 issue. Basically I have MinIO installed on the same k8s cluster as Flyte. I have supply

data_config

in FlyteRemote.

remote = FlyteRemote(
    config=Config(
        platform=PlatformConfig(
            endpoint="dns:///flyte-backend-flyte-binary-grpc.namespace.svc.cluster.local:8089",
            insecure=True,
            insecure_skip_verify=True,
        ),
        data_config=DataConfig(s3=S3Config(
            endpoint="<https://my-minio-endpoint.company.com>",
            access_key_id="access-key",
            secret_access_key="secret-key"
        ))
    ),
    default_project="flytesnacks",
    default_domain="development",
    data_upload_location="<s3a://flyte-storage/>",
)

However, I can see that the

get_data

is failing with error

FlyteAssertion: Failed to get data from 
<s3://flyte-storage/flytesnacks/development/Z6G7IM5HUD77U7SVV2UQRHP5KQ===>
===/script_mode.tar.gz to ./ (recursive=False).

Original exception: Unable to locate credentials

I guess I’m still not configuring the S3/data config properly. Any idea how to fix this issue?

Why does the

data_upload_location

have s3a?

@Samhita Alla Because it’s stated in the documentation that “the default location - s3://my-s3-bucket/data works for sandbox/demo environment. Please override this for non-sandbox cases.” So I thought I need to supply an s3a URI. But perhaps I only need s3:// then

Yeah, s3:// should suffice.

I’ve changed the config to `s3://´, but the error is still there.

I think this isn't a problem with FlyteRemote. You may need to edit the propeller configmap (

kubectl -n flyte edit cm flyte-sandbox-config

): https://discuss.flyte.org/t/9671922/running-the-greeting-example-i-am-getting-the-following-erro#ae1029d6-077a-4081-bee7-4fe85f57f3ae.

@Samhita Alla Thank you very much. It worked for me. Even though now I’m facing with the SIGKILL problem, I think I no longer have the S3 problem. The SIGKILL problem might be resolved by increasing the resources for the task and workflow.