Here, we describe how to use Jobflow with FireWorks as the execution backend instead of Jobflow-Remote. If you aren't familiar with Jobflow, refer to 👷Jobflow + Jobflow-Remote. While FireWorks can also be used to define the workflows themselves, we do not encourage this approach.
FireWorks has no server, so everything can be run from the login node. It has more pieces to set up, but as long as you are using Jobflow to define the workflows, the usage is pretty similar and takes a similar MongoDB approach.
Setup
Follow the Fireworks setup instructions in the Workflow Engine Setup section of the quacc documentation. Note that for the ~/.jobflow.yaml file, you can construct the JOB_STORE section in an analogous way as the jobstore section described in the 👷Jobflow + Jobflow-Remote document.
If you are running FireWorks on a NERSC machine, you will not use the same database credentials as we have locally on the Princeton Research Computing machines. Instead, follow the instructions in 🌌Perlmutter first.
Usage
To use FireWorks in practice, refer to the Deploying Calculations section of the quacc documentation.