Alex
Sieni
Anvil REU Automating HPC Software Deployment Mathematical/Computation Sciences
Abstract profile. Full document pending author claim.
Authors:
Alex Sieni
Date Created:
Not specified
Course Title:
Professor:
Not specified
About Paper:
On HPC systems, many researchers require the installation of various large, complex scientific applications. This process is generally cumbersome due to the large number of dependencies and the steep learning curve for building software from source. The proposed solution aims to automate that process through a user-friendly interface, providing real-time build information while achieving enough modularity for use at various institutions. Resources utilized encompass Go, Charm.sh library implementations (TUI: terminal user interface), Spack, Jenkins, PostgreSQL, and Kubernetes. The user interface is accessible via SSH into a frontend node or via log-in on a web-based cluster portal. After inputting the necessary parameters, a Jenkins CI/CD pipeline builds the Spack applications, configures a Spack environment, and confirms functionality through testing. A full Spack spec is returned, and environment modules are placed within the user's directory to be loaded for later use. Currently, a proof-of-concept for this workflow is complete. The next steps include hosting the TUI, pipeline, and database on Kubernetes and providing the option to pass a Spack YAML file for quicker Spack and application configuration. Through the proposed solution, automating software deployment will increase accessibility to powerful scientific software, automate testing and error reporting, and allow for greater research output. Keywords: HPC Applications; CICD; Spack; Build Automation
Source:
Purdue University / 2024
Topics:
No topics listed
Co-authors:
Alex Sieni