dagster

TLDR

Start development server

$ dagster dev

Start with specific module

$ dagster dev -m [my_project]

Launch daemon for schedules/sensors

$ dagster-daemon run

Start webserver only

$ dagster-webserver

Run a job

$ dagster job execute -m [my_project] -j [job_name]

List jobs

$ dagster job list -m [my_project]

Scaffold a new project

$ dagster project scaffold --name [my-project]

Check definitions

$ dagster definitions validate -m [my_project]

SYNOPSIS

dagster command [options]
dagster-daemon run [options]
dagster-webserver [options]

dagster is the CLI for Dagster, a data orchestration platform built around software-defined assets. It manages the development environment, job execution, and infrastructure.
dagster dev starts both the webserver (UI) and daemon (schedules, sensors) for local development. In production, run dagster-webserver and dagster-daemon separately.
Software-defined assets are the core abstraction—functions that produce data assets with dependencies. Assets form a DAG that Dagster materializes. Jobs group assets for execution.
Schedules trigger jobs on cron patterns; sensors trigger based on external events. Both require the daemon process to run.
The webserver provides a UI showing asset lineage, run history, logs, and metrics. The asset graph visualizes data dependencies.

PARAMETERS

-m, --module name

Python module containing definitions.

-f, --file path

Python file containing definitions.

-j, --job name

Job name.

-p, --port port

Webserver port. Default: 3000.

-h, --host host

Webserver host. Default: localhost.

-w, --workspace file

Workspace YAML file.

-d, --working-directory path

Working directory for code.

COMMANDS

dev

Start development server (webserver + daemon).

job execute|list|print

Manage and run jobs.

asset materialize|list|wipe

Manage software-defined assets.

schedule list|start|stop|preview

Manage schedules.

sensor list|start|stop|preview

Manage sensors.

project scaffold|from-example

Create new projects.

definitions validate

Validate code definitions.

instance info|migrate

Manage Dagster instance.

run list|delete|terminate

Manage pipeline runs.

CAVEATS

Daemon is required for schedules and sensors. Asset materialization tracks state in instance storage. Production requires PostgreSQL for run storage. Some features require Dagster+ (cloud).

HISTORY

Dagster was created by Elementl, founded by Nick Schrock (co-creator of GraphQL) in 2018. The project introduced the concept of software-defined assets as an improvement over task-oriented workflows. Version 1.0 released in 2022. Dagster emphasizes developer experience with type checking, testing utilities, and local development. The company offers Dagster+ for managed cloud orchestration.