iriscast-pipeline

This is a set of scripts and config files I used during the IRISCAST project. A project to collect energy usage metrics from multiple HPC facilities

Essentially I set up numerous data pipelines to collect energy usage information and ingest it into OpenSearch

The scripts include

  • Tests for PowerTOP to collect energy usage information at the process-level
  • Tests for Nvidia SMI to collect GPU energy usage information
  • Filebeats and Logstash configuration files for reading csvs of power metrics from different sites/facilities (csvs not included)

Run Opensearch

cd ~/<repo>/opensearch
docker-compose up

change the docker-compose script to point to directories with correct csv files

Run Logstash and Filebeats

cd ~/<repo>
docker-compose up

Scripts to parse external info

see iriscast-read-data folder for scripts used to parse information into csv files so that they could be uploaded to OpenSearch