Time and Location: Dec. 16thth, 2020, Wed., 10 a.m., Online seminar
 
 Speaker: Mr. Misha Ahmadian
 
Title: Collecting Job Info Parameters from Slurm Job Scheduler for HPC Provenance
 
Abstract: In December 2020, the High-Performance Computing Center of Texas Tech University introduced a brand-new cluster, named RedRaider, consisting of 240 AMD CPU nodes and 20 Intel/Nvidia V100 GPU nodes. RedRaider cluster, along with the current Quanah and Ivy clusters, is expected to significantly increase data generation and processing volume in this HPC center. Extensive growth in data volume and file operations will lead to more demand for Provenance systems for HPC clusters. Provenance refers to a set of metadata that describes the history of data, including how data are generated, used, and modified. These metadata annotate the datasets with plenty of details to track, identify, and cite all the generated data. Moreover, Provenance systems describe the relationships between all the elements in workflows that contribute to utilize or generate the data and support many advanced data management functionalities such as: Recognizing the source of data, finding assumptions behind the given results, auditing users’ file operations, and understanding how particular inputs transforms into desired outputs. In this talk, we will introduce the Texas Tech RedRaider cluster and the current equipment and resources. We will then spend the rest of the time discussing the Slurm resource manager components and how our proposed HPC Provenance system can leverage the Slurm REST API to collect the relevant job info parameters.