Ray provider

Orchestrate your Ray jobs using Apache Airflow® combining Airflow's workflow management with Ray's distributed computing capabilities.

Benefits of using this provider include:

• Integration: Incorporate Ray jobs into Airflow DAGs for unified workflow management.
• Distributed computing: Use Ray's distributed capabilities within Airflow pipelines for scalable ETL, LLM fine-tuning etc.
• Monitoring: Track Ray job progress through Airflow's user interface.
• Dependency management: Define and manage dependencies between Ray jobs and other tasks in DAGs.
• Resource allocation: Run Ray jobs alongside other task types within a single pipeline.

Table of Contents

• Quickstart
• Sample DAGs
• Getting Involved
• Changelog
• Contributing Guide

Quickstart

Check out the Getting Started guide in our docs. Sample DAGs are available at example_dags/.

Sample DAGs

Example 1: Using @ray.task for job life cycle

The below example showcases how to use the @ray.task decorator to manage the full lifecycle of a Ray cluster: setup, job execution, and teardown. The configuration for the decorator can provided statically or at runtime.

This approach is ideal for jobs that require a dedicated, short-lived cluster, optimizing resource usage by cleaning up after task completion

https://github.com/astronomer/astro-provider-ray/blob/bd6d847818be08fae78bc1e4c9bf3334adb1d2ee/example_dags/ray_taskflow_example.py#L1-L57

Example 2: Using SetupRayCluster, SubmitRayJob & DeleteRayCluster

This example shows how to use separate operators for cluster setup, job submission, and teardown, providing more granular control over the process.

This approach allows for more complex workflows involving Ray clusters.

Key Points:

• Uses SetupRayCluster, SubmitRayJob, and DeleteRayCluster operators separately.
• Allows for multiple jobs to be submitted to the same cluster before deletion.
• Demonstrates how to pass cluster information between tasks using XCom.

This method is ideal for scenarios where you need fine-grained control over the cluster lifecycle, such as running multiple jobs on the same cluster or keeping the cluster alive for a certain period.

https://github.com/astronomer/astro-provider-ray/blob/bd6d847818be08fae78bc1e4c9bf3334adb1d2ee/example_dags/setup-teardown.py#L1-L44

Getting Involved

Changelog

We follow Semantic Versioning for releases. Check CHANGELOG.rst for the latest changes.

Contributing Guide

All contributions, bug reports, bug fixes, documentation improvements, enhancements are welcome.

A detailed overview on how to contribute can be found in the Contributing Guide.

License

Apache 2.0 License