Integration of MINCER with LDMS for GPU Workflows
"Explore the integration of MINCER with LDMS in a project focused on GPU workflows and cluster-wide performance data collection. Learn about creating lightweight containers for monitoring, easy deployment across nodes, and goals for future enhancements. Discover the implementation approach and the seamless deployment of LDMS containers in cluster environments."
Download Presentation
Please find below an Image/Link to download the presentation.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author. If you encounter any issues during the download, it is possible that the publisher has removed the file from their server.
You are allowed to download the files provided on this website for personal or commercial use, subject to the condition that they are used lawfully. All files are the property of their respective owners.
The content on the website is provided AS IS for your information and personal use only. It may not be sold, licensed, or shared on other websites without obtaining consent from the author.
E N D
Presentation Transcript
Integration of MINCER with Irvin Lopez-Audetat, Ashkan Arabi, Shirley V. Moore, Deepak K. Tosh University of Texas at El Paso SLIDESMANIA.COM
Table of contents 01 - Project Overview 02 - Goals 03- Implementation Approach 06 - Demo 04 - DCGM & LDMS in MINCER Containers 05 - Docker Swarm for Deployment SLIDESMANIA.COM
Project Overview Mincer Project Provides containerized tools for in-depth insight and analysis for distributed heterogeneous computing environments. To work in conjunction with OSG by providing MINCER containers for integration equipped with all the necessary tools and metrics, and following best practices to ensure portability and reusability in accordance to OSG standards. Focuses on GPU workflows and cluster-wide performance data collection. SLIDESMANIA.COM
Goals Primary & Future Goals Create lightweight containers for LDMS-based monitoring. Enable easy deployment and cleanup of monitoring containers across compute nodes. Aggregate monitoring data centrally to avoid manually checking every node. Future Goals: Incorporate DCGM for GPU-specific metrics. Contribute final containers and usage examples to OSG. SLIDESMANIA.COM
LDMS in MINCER Containers Overview and Integration LDMS Overview: A low-overhead metric collection tool with plugin-based architecture. Collects system-level metrics like CPU, memory, and network usage. Integration in Our Work: We incorporated LDMS into a container image. Configured it for seamless deployment in cluster environments. SLIDESMANIA.COM No DCGM Integration Yet: DCGM integration is planned, but not yet part of this deployment.
Implementation Approach Work done & work in progress What We Did: Built a container image with LDMS and essential dependencies. Wrote a script that uses a Swarm stack file to deploy LDMS containers to all nodes. Ensured the manager node aggregates data for easier cluster-wide insights. Modules used: vmstat, meminfo. Source Code: https://github.com/AshkanArabim/osg-cluster-monitoring/ What We Did Not Do: SLIDESMANIA.COM Integrate DCGM yet will be a future step. Export the LDMS and DCGM data into OSG Dashboard: https://ce-dashboard.ospool.osg- htc.org/index.html
Demo by Ashkan SLIDESMANIA.COM
Thank you! Do you have any questions? ilopezaude@miners.utep.edu aarabimian@miners.utep.edu svmoore@utep.edu This work is supported by the National Science Foundation under award #2346423. SLIDESMANIA.COM
