We need to collect statistics on user work and server load, linux. How can this be done?

Question

linux

It is necessary to collect statistics on the work of users on the server with scientific linux 6.1, not a virtual machine. I would like to understand who loads it the most, with what tasks.
How and with what can this be done?
Thank you.

Answer 1

man atop
https://habr.com/en/post/140010/

Answer 2

I recommend to roll monitoring system. This requires a database, a data collection agent, and some service to display this data.
I advise InfluxDB - it is installed in one click, it requires almost no maintenance on your volumes.
Put telegraf to it - a statistics collection agent, collects everything that can be collected, processes, users, network, disks, temperature, etc. Also put in one click.
And to view all this through the browser - put grafana. It is put in two clicks from packages.
In grafana, you need to create graphs for viewing data, but you can take them ready-made decanters from the site itself and import them to yourself.
In general, we monitor a whole cluster of machines in this way, but I also monitor a bunch of individual servers, where everything is installed on them.