in_mem_metrics: initial documentation.

[pwhelan]

This adds documentation for the new in_mem_metrics plugin: fluent/fluent-bit#7615.

[pwhelan]

@patrick-stephens should I add a specific example for using docker to log metrics for kubernetes nodes?

[patrick-stephens]

@patrick-stephens should I add a specific example for using docker to log metrics for kubernetes nodes?

I don't really understand so probably :)

Although why would you use it over node exporter, kubelet metrics, etc. is probably the main question I'd have? Be good to have an example of what it provides in addition.

[pwhelan]

Although why would you use it over node exporter, kubelet metrics, etc. is probably the main question I'd have? Be good to have an example of what it provides in addition.

The main reason to use it over those options is granularity. This plugin grabs the memory usage of each individual process and various different types, ie: shared, etc... If you want to be able to correctly diagnose which process is actually leaking this would be what you need to be able to discern that correctly.