Add additional metrics to the internal plugin

[zoonage]

Feature Request

Whilst troubleshooting a Telegraf issue earlier I realised it would have been nice to know how many goroutines were running at a given point in time. This can be found by runtime.NumGoroutine(). I wanted to kickstart a discussion about other metrics that people might want to add to the internal plugin.

Proposal:

Add a metric from runtime.NumGoroutine() to the internal plugin.

Current behavior:

Telegraf does not currently support this metric

Desired behavior:

To be able to see the number of goroutines that Telegraf is using

Use case:

Helps troubleshoot issues with Telegraf

[danielnelson]

Seems like a good idea, can you open a pull request?

[russorat]

Which measurement would you put numgoroutine in? it doesn't seem to fit well into any of the existing ones. maybe adding an internal_runtime measurement?

Other things that might be useful to have (from: https://github.com/influxdata/influxdb/blob/master/prometheus/influx.go#L27):

• "version": build.Version,
• "commit": build.Commit,
• "build_date": build.Date,
• "os": runtime.GOOS,
• "arch": runtime.GOARCH,
• "cpus": strconv.Itoa(runtime.NumCPU()),

connect: #5958

[zoonage]

The number of CPUs is already reported via the n_cpus field on the system measurement

[danielnelson]

We can add num_goroutine to internal_agent. Let's skip build date and commit for now, they impose requirements on the build. Things like number os and arch would make more sense to me to add them to the system plugin potentially with additional platform information.