So, it depends on what you're monitoring. There is no one-size-fits-all tool.

I haven't needed to monitor remote systems much, especially recently. There's a whole menagerie of tools for aggregating that kind of stuff.

For the local system, I use a collection of things:

• In Sway, I use waybar, which can display current processor temperature and CPU load. I have it normally invisible so that the applications get the whole screen, but when I have the Super key down, it flashes up at the bottom of the screen.

• top is installed pretty much everywhere, and will show CPU and memory usage. I've played with atop, htop, and btop, but none have them really stuck. First, they aren't universally pre-installed on systems, and being everywhere is a nice advantage. Second, they all have some things that I don't like. htop defaults to blowing a line of the terminal per core on showing individual load on a core, which is a lot of space for something that probably isn't of general interest. btop looks prettiest, but uses a bunch of CPU time and refuses to run in a small terminal, besides wasting some (less) space on a per-core basis. I can't recall what it was about atop that didn't do it for me.

• bwm-ng is the tool I typically use to get an idea of how much data is moving in and out of a system.

• sensors (in the lm-sensors package) dumps most of the temperature data that your system's temperature sensors can see.

• mangohud is useful for showing an overlay in 3D games with a lot of performance data.

• I personally like sar (in the sysstat package) to log and show historical data, as it's been around for a long time. I've used collectd because it supports logging data from a temperature probe that I have.

• I think that the new hotness for logging data is maybe pcp, Performance Co-Pilot. I've been meaning to play with this more. Red Hat seems to very much recommend it. Red Hat has a list of PCP equivalents -- for example, pcp atopsar -A is comparable to sar. It also appears to support monitoring remote hosts. Just haven't had the occasion to really dig into it.

• radeontop for my AMD GPUs. This has been particularly useful recently, as it permits showing how much VRAM is in use

if you're playing around with AI stuff, it chews up a bunch of memory, and currently, Linux doesn't have a single all-encompassing system for automatically unloading and reloading stuff (though ollama will, by default, unload on a period of inactivity) so you need to keep an eye on it if you're simultaneously running games and other things that use VRAM.

• smartctl in the smartmontools package provides a bunch of data about hard drive performance and problems.