Autosubmit Workflow Manager#

Autosubmit

Autosubmit is an open source Python experiment and workflow manager used to manage complex workflows on Cloud and HPC platforms.

$ pip install autosubmit
Illustration of a person and workflows running on a platform.

Autosubmit is a lightweight workflow manager designed to meet climate research necessities. Unlike other workflow solutions in the domain, it integrates the capabilities of an experiment manager, workflow orchestrator and monitor in a self-contained application.

It is a Python package available at PyPI. The source code in Git contains a Dockerfile used in cloud environments with Kubernetes, and there are examples of how to install Autosubmit with Conda.

What Autosubmit Does#

Automation

Management of job submission and dependencies without user intervention.

Data Provenance

Experiments with unique PIDs, use of open standards for data provenance in the experiments and workflows.

Fault Tolerance

Automatic retries and ability to re-run specific parts of the experiment in case of failure.

Resource Management

Individual platform configuration, allowing users to run their experiments without having to modify job scripts.

Multiplatform

Widely used to run experiments on different platforms simultaneously, using batch schedulers such as Slurm. It is deployed and used on various HPC and cloud systems.

Open Source

Autosubmit code is hosted on GitHub, licensed under the GPLv3 License, and under active development.

Where Autosubmit Is Used#

Autosubmit runs the workflows behind these research centres, projects and operational services.

The Autosubmit Ecosystem#

Two companion tools extend Autosubmit with monitoring and a graphical interface. Both are open source and documented separately.

Autosubmit API

A Python web application that monitors, analyses and controls workflows created and managed with Autosubmit. It exposes experiment state over HTTP and is what the GUI reads from.

https://autosubmit-api.readthedocs.io/
Autosubmit GUI

A graphical interface that presents workflow execution in the browser, with job trees, run history and log inspection for experiments managed by Autosubmit.

https://autosubmit-gui.readthedocs.io/

Citing Autosubmit#

If you use Autosubmit, please cite it using these metadata.

Seamless management of ensemble climate prediction experiments on HPC platforms
D. Manubens-Gil, J. Vegas-Regidor, C. Prodhomme, O. Mula-Valls, F. J. Doblas-Reyes (2016)
2016 International Conference on High Performance Computing & Simulation (HPCS), IEEE 10.1109/HPCSim.2016.7568429

https://doi.org/10.1109/HPCSim.2016.7568429

Contact Us#

Source code

Browse the code, read the changelog, or open a pull request.

https://github.com/BSC-ES/autosubmit
Report a problem

Found a bug or want to request a feature? Open an issue on the tracker.

https://github.com/BSC-ES/autosubmit/issues
Discussions

Ask questions, share workflows, and see how other centres use Autosubmit.

https://github.com/BSC-ES/autosubmit/discussions
Ask the team

Questions about running Autosubmit at your centre, or anything the docs do not answer.

mailto:support-autosubmit@bsc.es

Media and Press Kit#

Logos in SVG and PNG, and slide decks — everything you need to include Autosubmit in your own materials.

Open the media kit