This utility is useful to analyze the usage of a SLURM computing cluster and to create reports allowing to track the cluster usage. It uses the accounting data provided by 'sacct' to create the reports. It's not useful for billing but is used for capacity planning and to tailor an adequate Scheduling Policy.
It's a Python program. It's been developed using Python 2.7.6 and uses Python's Standard libraries plus:
- numpy (for histogram creation)
This project uses the standard Python packaging system. Therefore, just run:
python setup.py install
It will also automatically install any dependencies needed.
This setup does not create an appropriate config file. If you need one to customize the tool's behaviour, read the following section. By default config files are looked for in the current directory.
The configuration file uses INI file's syntax. The options available are:
[general]
report_title=Report for my cluster
avail_cpu_number=20
sacct_path=/bin/sacct
- sacct_path: Path to the sacct command. This is the option most likely to need configuration (default: /bin/sacct)
- report_title: Provides the title for the Report generated (default: Report)
- avail_cpu_number: Number of CPUs available to compute the usage on the period (default: None)
Show help:
./report_colas.py --help
Show standard report (using defaults):
./report_colas.py --start 2016-05-30 --end 2016-06-01