Sign inSign up

tripleaceme/orchestack-airflow

By tripleaceme

Updated 2 months ago

Apache Airflow 2.10 with dbt-core, dbt-postgres, and astronomer-cosmos preinstalled for OrcheStack.

Image
Web servers
Databases & storage
Monitoring & observability
0

10K+

tripleaceme/orchestack-airflow repository overview

orchestack-airflow

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.

What this image adds to upstream Airflow

PackagePurpose
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.

Base image

apache/airflow:2.10.5-python3.12

Apache 2.0 license inherited from upstream.

What this image does NOT add

Custom DAGs, custom dbt projects, ingestion libraries (dlt, requests, pandas, etc.). The platform is intentionally unopinionated about those choices:

  • DAGs — operators bring their own from a Git repository via AIRFLOW_DAGS_REPO_URL. The first-pipeline guide documents three composition patterns operators can copy + adapt.
  • dbt project — operators bring their own from a Git repository via DBT_REPO_URL. Mounted into both the dbt service container and this Airflow container so Cosmos can parse the manifest at DAG-parse time.
  • Ingestion libraries — install per-DAG via Airflow's PythonVirtualenvOperator (works for dlt, requests, pandas, anything on PyPI; isolated per task; cached after first run).

Connections

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.

How this image is used

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/.

ImagePurpose
tripleaceme/orchestack-authSignup, login, setup wizard
tripleaceme/orchestack-orchestratorService lifecycle daemon
tripleaceme/orchestack-dashboardAdministrator dashboard
tripleaceme/orchestack-geGreat Expectations preinstalled

Upstream

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.

Tag summary

Content type

Image

Digest

sha256:58635013a

Size

556.8 MB

Last updated

2 months ago

docker pull tripleaceme/orchestack-airflow