These are machine level metrics, it's the responsibility of the node exporter to export those - not individual applications.

@brian-brazil I agree with you about "committedVirtualMemorySize, freePhysicalMemorySize, freeSwapSpaceSize, totalPhysicalMemorySize, totalSwapSpaceSize and systemCpuLoad are machine level metrics"(processCpuLoad is JVM level metric.). But, if one can get these metrics from JMX why would he need node exporter to get these metrics?

Maybe it's no appropriate to include these metrics in 'StandardExports', but I think it will be helpful to provide these metrics in another Exports, for example OperatingSystemExports, and let the develop decide which one to use: OperatingSystemExports or node export.

https://prometheus.io/docs/instrumenting/writing_exporters/#machine-and-process-metrics

well I modified the PR just add processCpuLoad metric, that's process-level metric.

process_cpu_seconds_total should already cover that, and in a better way for Prometheus.

Well, PR is not a good place for discussion, I created an issue.