github.com/travelaudience/pprofetheus

pprofetheus

pprofetheus is a collector for Prometheus that collects CPU profiling data for the current process and exports them as metrics. It can be used to monitor, visualize, and alert on profiling data from any Go process that imports pprofetheus and exports metrics via Prometheus.

Quick start

In order to use pprofetheus in your Prometheus-enabled Go application, you just need to

go get github.com/travelaudience/pprofetheus

and then import the same package, and set up the collector with Prometheus in your code, e.g. like this:

cpuProfileCollector, err := pprofetheus.NewCPUProfileCollector()
if err != nil {
	/* handle error */
}
prometheus.MustRegister(cpuProfileCollector)
cpuProfileCollector.Start()

After these changes, your application will export the Prometheus metrics pprof_cpu_time_used_ms, pprof_cpu_time_used_cum_ms, pprof_cpu_started and pprof_cpu_stopped.

pprof_cpu_time_used_ms contains the amount of milliseconds the program spent in the function provided in the label function.

pprof_cpu_time_used_cum_ms contains the cumulated amount of milliseconds the program spent in the function provided in the label function. This means that the amount of time spent in a function is accounted both for the function that spent the time and all functions up the call stack. The cumulated time thus includes the total time that a function spent, including all other functions that were called by that function.

pprof_cpu_started counts how often the Start method has been called on the collector, while pprof_cpu_stopped counts how often the Stop method has been called on the collector.

License

Please see the file LICENSE for licensing information.

The files in the subdirectory internal are subject to a separate license. See internal/LICENSE for licensing information.