prefect-saturn

prefect-saturn is a Python package that makes it easy to run Prefect Cloud flows on a Dask cluster with Saturn Cloud. For a detailed tutorial, see "Fault-Tolerant Data Pipelines with Prefect Cloud ".

Installation

prefect-saturn is available on PyPi.

pip install prefect-saturn

prefect-saturn can be installed directly from GitHub

pip install git+https://github.com/saturncloud/prefect-saturn.git@main

Getting Started

prefect-saturn is intended for use inside a Saturn Cloud environment, such as a Jupyter notebook.

import prefect
from prefect import Flow, task
from prefect_saturn import PrefectCloudIntegration


@task
def hello_task():
    logger = prefect.context.get("logger")
    logger.info("hello prefect-saturn")


flow = Flow("sample-flow", tasks=[hello_task])

project_name = "sample-project"
integration = PrefectCloudIntegration(
    prefect_cloud_project_name=project_name
)
flow = integration.register_flow_with_saturn(flow)

flow.register(
    project_name=project_name,
    labels=["saturn-cloud"]
)

Customize Dask

You can customize the size and behavior of the Dask cluster used to run prefect flows. prefect_saturn.PrefectCloudIntegration.register_flow_with_saturn() accepts to arguments to accomplish this:

For example, the code below tells Saturn that this flow should run on a Dask cluster with 3 xlarge workers, and that prefect should shut down the cluster once the flow run has finished.

flow = integration.register_flow_with_saturn(
    flow=flow,
    dask_cluster_kwargs={
        "n_workers": 3,
        "worker_size": "xlarge",
        "autoclose": True
    }
)

flow.register(
    project_name=project_name,
    labels=["saturn-cloud"]
)

GitHub

https://github.com/saturncloud/prefect-saturn