FireWorks workflow software
FireWorks allows you to define calculation workflows and execute them on remote computers, usually through a queueing system. Workflows are stored in a centralized database, and jobs are pulled from the database by registered computers.
Unique features of FireWorks include:
- Dynamic workflows that react to results programmatically. A job can be automatically restarted, modified, or cancelled in case of error or other condition. Entire workflows can be changed automatically based on calculation results.
- Distribute calculations over multiple computing resources simultaneously.
- Automated duplicate workflow detection
- Plug-and-play on several large supercomputing clusters and queueing systems (future)
- Web-based monitoring of workflows (future)
FireWorks is intended for applications where realtime performance of the workflow software is not a big issue. For example, if you require steps in a workflow to execute within a few seconds of one another, FireWorks might not be for you. In addition, FireWorks is a centralized workflow system.
TODO: add description
- Use pip-install
- run python setup.py nosetests
Setup on clusters / Tutorial (Future)
TODO: link to another page…
TODO: add proper docs
- Create a subclass of QueueAdapter that handles queue issues - an example is PBSAdapterNersc
- Create an appropriate JobParameters file for your cluster - an example is provided.
- Try running rocket_launcher.py on your cluster with a test job config. See if it prints ‘howdy, you won’ or whatever.
- Try changing the executable to be the Rocket. See if it grabs a job properly…
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.