programming

Profiling Go Apps

Radhi Fadlillah

2018-10-14 21:34

There are several ways for profiling a Go application, but the easiest one that I’ve found is by using github.com/pkg/profile.

To use it, first install it by running :

go get github.com/pkg/profile

Once installed, enable the profiler by starting it in the main function of your app :

package main

import "github.com/pkg/profile"

func main() {
    // Use this line for profiling CPU
    defer profile.Start().Stop()
    
    // Use this line for profiling memory
    defer profile.Start(profile.MemProfile).Stop()
}

Next, build and run your app like usual. Once finished, the app will generate a pprof file in a temporary folder, and it will tell you where it’s located, for example :

2018/10/1414:26:28 profile: cpu profiling enabled, /var/...../cpu.pprof

If you use memory profiler, the output will be mem.pprof instead of cpu.pprof.

To analyze the pprof file, we can convert it to PDF file by using standard Go tool :

go tool pprof --pdf ~/go/bin/yourbinary /path/to/cpu.pprof > file.pdf
# go # golang