Skip to main content

Asset jobs

Jobs are the main unit of execution and monitoring for asset definitions in Dagster. An asset job is a type of job that targets a selection of assets and can be launched:

  • Manually from the Dagster UI
  • At fixed intervals, by schedules
  • When external changes occur, using sensors

Creating asset jobs​

In this section, we'll demonstrate how to create a few asset jobs that target the following assets:

src/<project_name>/defs/assets.py
@dg.asset
def sugary_cereals() -> None:
    execute_query(
        "CREATE TABLE sugary_cereals AS SELECT * FROM cereals WHERE sugar_grams > 10"
    )


@dg.asset(deps=[sugary_cereals])
def shopping_list() -> None:
    execute_query("CREATE TABLE shopping_list AS SELECT * FROM sugary_cereals")

To create an asset job, use the define_asset_job method. An asset-based job is based on the assets the job targets and their dependencies.

You can target one or multiple assets, or create multiple jobs that target overlapping sets of assets. In the following example, we have two jobs:

  • all_assets_job targets all assets
  • sugary_cereals_job targets only the sugary_cereals asset
src/<project_name>/defs/jobs.py
all_assets_job = dg.define_asset_job(name="all_assets_job")

sugary_cereals_job = dg.define_asset_job(
    name="sugary_cereals_job", selection="sugary_cereals"
)

Making asset jobs available to Dagster tools​

Jobs are loaded automatically with dg and there is no need to explicity define a Definitions object for them. If you include schedules or sensors, the code location will automatically include jobs that those schedules or sensors target.

src/<project_name>/defs/assets.py
import dagster as dg


@dg.asset
def number_asset():
    yield dg.MaterializeResult(
        metadata={
            "number": 1,
        }
    )


number_asset_job = dg.define_asset_job(
    name="number_asset_job", selection="number_asset"
)

Testing asset jobs​

Dagster has built-in support for testing, including separating business logic from environments and setting explicit expectations on uncontrollable inputs. For more information, see the testing documentation.

Executing asset jobs​

You can run an asset job in a variety of ways:

  • In the Python process where it's defined
  • Via the command line
  • Via the GraphQL API
  • In the UI

Examples​

The Hacker News example builds an asset job that targets an asset group.