Dagster Jobs and Schedules¶

Truthound integrates cleanly into Dagster jobs when quality checks are modeled as explicit ops or resource-backed asset steps. The goal is to keep scheduling and automation Dagster-native while preserving the shared Truthound runtime contract underneath.

Who This Is For¶

Dagster operators designing scheduled validation jobs
teams combining asset materialization and quality checks in one graph
platform engineers standardizing reusable quality jobs

When To Use It¶

Use this page when:

a resource-backed check should run on a schedule
you need a dedicated validation job separate from asset materialization
you want to reuse data_quality_check_op across datasets

Prerequisites¶

truthound-orchestration[dagster] installed
Dagster Definitions-based project layout
familiarity with DataQualityResource and the prebuilt ops

Minimal Quickstart¶

Create a small quality job with the exported op surface:

from dagster import Definitions, job
from truthound_dagster import DataQualityResource, data_quality_check_op

@job(resource_defs={"data_quality": DataQualityResource()})
def users_quality_job():
    data_quality_check_op.alias("check_users")()

defs = Definitions(jobs=[users_quality_job])

For mixed jobs, keep materialization and validation as separate nodes:

from dagster import job, op
from truthound_dagster import create_check_op

@op
def load_users():
    return read_users_dataframe()

validate_users = create_check_op(
    name="validate_users",
    rules=[{"column": "id", "check": "not_null"}],
)

Production Pattern¶

Use one of these shapes:

Pattern	Best Fit
asset-only schedule	quality logic naturally belongs to assets
dedicated validation job	operators want a separate operational surface
mixed pipeline job	materialization and validation must stay in one Dagster graph

Recommended schedule policy:

keep validation schedules aligned with materialization freshness windows
separate “fast gate” checks from expensive profile or learn jobs
name jobs after the dataset or domain, not the engine

Failure Modes and Troubleshooting¶

Symptom	Likely Cause	What To Do
schedule is green but asset quality is stale	validation job is not aligned with materialization cadence	reschedule to run after the producing asset
one validation graph becomes too large	too many unrelated checks live in one job	split by domain or ownership boundary
jobs are hard to debug	materialization and validation share the same op name patterns	use explicit aliases for quality steps