ability to filter processes in hostmetricsreceiver based on uptime

[davidmirza408]

Is your feature request related to a problem? Please describe.
HostMetricsReceiver collects telemetry data on all processes running in a system by default. It would be helpful to be able to filter out short lived processes that may be up for only a few seconds but result in a lot of noise bing reported.

Describe the solution you'd like
I would like to create a filter that allows the user to filter out processes that have been running for less than N seconds. I would also like to volunteer to work on this ticket if approved.

Describe alternatives you've considered
N/A

Additional context
N/A

cc @dmitryax

[davidmirza408]

This is what I had in mind: https://github.com/davidmirza408/opentelemetry-collector-contrib/pull/5

[jpkrohling]

cc @dmitryax

[davidmirza408]

Another possible solution is to include uptime in the metrics so that it can be filtered out by a processor if needed. Please advise as to what is the preferable solution to this.

[dmitryax]

Another possible solution is to include uptime in the metrics so that it can be filtered out by a processor if needed. Please advise as to what is the preferable solution to this.

I don't see a clean and safe way to include uptime in the metric. Adding it as an attribute will result in a new timeseries per each datapoint unless the attribute is removed by a processor

Let's add this as a configuration parameter for the receiver. Please suggest a configuration interface in this issue.

[dmitryax]

Maybe...

scrape_delay: 0s

or

scrape_process_delay: 0s

[davidmirza408]

I like scrape_process_delay: 0s. I can do the implementation for this issue