Authors:

Anjali Rajesh

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

This project aims to develop a comprehensive and interactive dashboard for the Anvil high performance computing cluster at Purdue University by providing detailed insights into user allocations, resource usage, and job performance metrics, ultimately focusing on enhancing job efficiency. This research project prioritizes making improvements on the current Anvil web-based dashboard built with the Open OnDemand platform. The current dashboard does not reveal any information to researchers pertinent to whether their jobs are efficiently utilizing their allocation. In the improved dashboard, information regarding service unit credits, disk usage, CPU efficiency, memory efficiency, and other metrics, is derived by parsing data from the open-source job scheduler Slurm, using scripts written in Ruby. Furthermore, the dashboard also integrates utilities such as sacct and reportseff. Sacct is used for displaying job accounting data from the Slurm database; reportseff is an open- source Python script from Princeton University which provides detailed job efficiency information for both CPU and memory utilization. The queried data is presented in a user-friendly manner using various Javascript graphing and table libraries, including Chart.js, Highcharts, and Datatables, using Ruby on Rails as the web framework to maintain the backbone of the dashboard. The objective of the dashboard is to provide a clean, well-structured, and extensible interface for researchers to visualize useful information about their utilization of Anvil without accessing the terminal. Individuals working on Anvil may view detailed statistics regarding job performance metrics via the dashboard, and can identify and debug any jobs that are noticeably inefficient. Keywords: Anvil; Open OnDemand; HPC; Dashboard; Slurm

Source:

Purdue University / 2024

Topics:

No topics listed

Co-authors:

Anjali Rajesh