Apache Airflow 2.10 with dbt-core, dbt-postgres, and astronomer-cosmos preinstalled for OrcheStack.
10K+
Apache Airflow 2.10 customised for OrcheStack — an open-source containerised data platform that integrates Airbyte, Apache Airflow, dbt Core, Great Expectations, Metabase, MinIO, OpenMetadata, and pgAdmin behind a single operator-facing interface.
This image extends the upstream apache/airflow image with the tooling an
OrcheStack operator needs to run dbt from Airflow with per-model task
granularity, without any runtime pip install.
| Package | Purpose |
|---|---|
dbt-core (1.8.0) | The dbt CLI for running transformations |
dbt-postgres (1.8.0) | dbt adapter for the OrcheStack PostgreSQL warehouse |
astronomer-cosmos (1.8.x) | Generates one Airflow task per dbt model + per dbt test, with per-model failure attribution and re-run-from-failed support |
Everything is baked in at build time so cold-start time is sub-15s on subsequent boots. The first start is slower because Airflow's metadata migrations run.
apache/airflow:2.10.5-python3.12
Apache 2.0 license inherited from upstream.
Custom DAGs, custom dbt projects, ingestion libraries (dlt, requests, pandas, etc.). The platform is intentionally unopinionated about those choices:
AIRFLOW_DAGS_REPO_URL. The
first-pipeline guide
documents three composition patterns operators can copy + adapt.DBT_REPO_URL. Mounted into both the dbt service container and this
Airflow container so Cosmos can parse the manifest at DAG-parse time.PythonVirtualenvOperator (works for dlt, requests, pandas, anything
on PyPI; isolated per task; cached after first run).The orchestrator's post-start hook creates the
orchestack_warehouse Airflow Connection on first start using the
warehouse credentials from .env. Cosmos's
PostgresUserPasswordProfileMapping reads it directly — operators
don't need to learn about Airflow Connections to run dbt.
Operators add other connections (Airbyte, Metabase, Fivetran, Tableau, etc.) through the Airflow UI per their stack.
This is part of OrcheStack and is not designed to run standalone. It
runs alongside the rest of the platform via the docker-compose.yml
shipped in the OrcheStack runtime bundle.
To deploy OrcheStack:
curl -fsSL https://orchestack.ng/install.sh | bash
Or download the latest runtime bundle
and follow its INSTALL.md. After setup, Airflow is reachable at
http://your-host/app/airflow/.
| Image | Purpose |
|---|---|
tripleaceme/orchestack-auth | Signup, login, setup wizard |
tripleaceme/orchestack-orchestrator | Service lifecycle daemon |
tripleaceme/orchestack-dashboard | Administrator dashboard |
tripleaceme/orchestack-ge | Great Expectations preinstalled |
This image extends and respects all licences of upstream Apache Airflow. See https://airflow.apache.org/ for the original project, and https://github.com/astronomer/astronomer-cosmos for Cosmos.
Content type
Image
Digest
sha256:58635013a…
Size
556.8 MB
Last updated
2 months ago
docker pull tripleaceme/orchestack-airflow