3.3. Server Utilization¶

What is the Server Utilization?¶

The Server Utilization cron job connects to the d3VIEW sever, as configured in the Lucy config. Once connected it will check with the scheduler or d3VIEW server to determine the utilization at the time run.

There are two main utilization modes: Weekly and Hourly. The Weekly mode is intended to be run at a longer interval, as it will give the history of the last 10000 jobs, whereas the hourly should be at a short interval, as it is just a simple snapshot.

Most of the time the hourly can be run once or twice an hour, but it depends on the metrics that should be taken.

Configure Server Utilization¶

To configure the Server Utilization utility, configure the d3VIEW server url and port correctly in the d3view.json file.

Each of the hpc servers in d3VIEW should have a correct scheduler path and scheduler set up.

Also a directory to hold utilization files should be created.

How to run¶

The Server Utilization cron job can be used by invoking Lucy manually or in a crontab. The most basic command is:

$ lucy server utilization output_file_name

lucy server utilization output_file_name

Arguments¶

All arguments are shown directly below, with links to their sections.

Parameter name	When required

`output_file_name`	ALWAYS
`-period`
`-log-file`
`-verbose`

output_file_name¶: The output file name should be the full path to the output file. Expect the output file to be prefixed with the HPC Server’s id. For example if the value specified is /tmp/utilization.csv expect the file to be named /tmp/336_utilization.csv

-period¶

There are 2 modes for utilization period:

hourly (default) : takes a snapshot of the current utilization
weekly: takes a longer period of utilization history

-verbose¶: This adds extra debug information while running.