How to properly use loadavg-target

[MNarath1]

How is loadavg-target properly used is it a factor of 1.0 for 100% cpu usage or a different factor i can't really find any documentation on it.

[kakra]

No, load average can be compared to the Windows run queue length. It measures the average amount of processes scheduled for immediate execution, it's the length of the scheduler run queue. In Linux it not only counts processes waiting on CPU, but also processes waiting on memory or IO, so it's probably different to Windows. I think original Unix OS'es also only count CPU waiting. So this is actually useful for bees, because it includes IO wait.

General rule of thumb is to keep your load at around the number of CPU cores. It means your system will be saturated with work it can handle. It doesn't mean that overall CPU usage is at 100%, usually it's below (because processes are also waiting on other resources).

Look at your system's loadavg while you're doing light work, like browsing the web or watching a youtube video, doing normal stuff. Put the bees loadavg limit slightly above it. If bees makes steady progress without impacting your usual work, you can still slightly increase it. I'm running a 20 thread CPU with 4 btrfs pool disks, and I've settled to bees loadavg target 5. That way, it can keep all disks busy on different threads before loadavg of my system raises. It should still have plenty of CPU left. But if your system becomes sluggish due to IO from bees, reduce it again.