PyJobRunner 1.7

Job runner for scientific computing workloads

Jobrunner is a command line tool to manage and deploy computing jobs, organize complex workloads, and enforce a directory based hierarchy to enable reuse of files and bash scripts within a project. Organization details of a directory tree is encoded in Jobfiles which serve as an index of files/scripts, and indicate their purpose when deploying or setting up a job.

Functionality of Jobrunner is best understood through example projects which can be found in following repositories:

• akashdhruv/boiling-simulations: A collection of high-fidelity flow/pool boiling simulations

• akashdhruv/paramesh-bittree-tests: A lab notebook for performance tests of multiphysics scientific software instrument, Flash-X

• akashdhruv/channel-flow: Example simulations of the channel flow problem to showcase applicability of containerization tools for scientific computing problems

Writing a Jobfile

A Jobfile provides details on functionality of each file in a directory tree along with schedular configuration

[schedular]

   # schedular command `bsub`, `qsub`, `slurm` or `bash`
   command = "slurm"
   options = [
              "#SBATCH --ntasks=5",
              "#SBATCH -t 0-30:00",
              "#SBATCH --job-name=myjob",
              ]

[job]
   setup = []
   input = []
   target = "target"
   submit = []
   archive = []

Jobrunner commands

• Setup: Executes

• Submit: Executes

• Archive: Executes

• Clean: Executes

Installation

Using Python Package Index (PyPI)

pip3 install PyJobrunner

Development mode

pip3 install click && ./setup develop