hawk-utils-scripts/monitoring/db
2024-02-21 15:55:05 +01:00
..
elklikeoutput First commit for getting data from PG-SQL 2023-08-09 17:39:31 +02:00
scripts Reimplement from_list in terms of pandas 2024-02-21 15:55:05 +01:00
README.md Update README 2024-02-15 11:18:11 +01:00

hawk-utils-scripts/monitoring/db

Use this directory for scripts, etc, that operate on or extract data from the the database.

scripts/get_detailed_power.py

Python script which querries the database for a list of jobids and produces a file for each jobid.

Those files contain time-resolved power consumption data and are meant to be consumed by the utility in monitoring/logs/scripts/plot_energy_logs.sh.

Requirements:

  • Python module sqlalchemy python -m pip install sqlalchemy psycopg2-binary

Usage:

ssh monitoring

./get_detailed_power.py --help

./get_detailed_power.py 2335001 7678069.hawk-pbs5
Created file detailed_power.2335001.hawk-pbs5.100000-1000004.csv
Created file detailed_power.7678069.hawk-pbs5.2432345-423423.csv

Querrying Hawk-AI nodes requires the option --hawk-ai.

The time resolution can be set with the option --interval.

Note, that small intervals will lead to data gaps. In particular, not all nodes will report power values for a given time stamp. Multiple values within an interval are averaged arithmetically. This migth make it difficult to use the output of this script to calculate total energy used by a job.