The runtime package

Learn the basic functions of the runtime package.

The runtime package

The runtime package lets you talk to the Go runtime underneath your program: scheduler, memory and stack information.

fmt.Println(runtime.NumCPU())        // number of cores
fmt.Println(runtime.NumGoroutine())  // live goroutines
fmt.Println(runtime.GOOS, runtime.GOARCH)

These functions are for diagnosis and monitoring; tying application logic to them is usually not a good idea.

runtime/metrics

runtime/metrics, added in Go 1.16, offers cheaper measurements with stable names than ReadMemStats:

s := []metrics.Sample{{Name: "/sched/goroutines:goroutines"}}
metrics.Read(s)
fmt.Println(s[0].Value.Uint64())

All names are listed by metrics.All(); Prometheus clients use this package too.

Exercise Goroutine metric

Read the "/sched/goroutines:goroutines" sample with runtime/metrics and print whether the value is greater than 0.

Show example solution

package main

import (
    "fmt"
    "runtime/metrics"
)

func main() {
    s := []metrics.Sample{{Name: "/sched/goroutines:goroutines"}}
    metrics.Read(s)
    fmt.Println(s[0].Value.Uint64() > 0)
}

Test yourself

When is <code>runtime.GOOS</code> determined?
- The user sets it
- At compile time (the target OS)
- It is re-measured on each run
- During a GC cycle
Show answer

At compile time (the target OS) — GOOS and GOARCH are the target of the built binary; when cross-compiling they can differ from the build machine.

Core count

n := runtime.____()

Show answer

n := runtime.NumCPU()

runtime function

NumCPU

Logical CPUs
NumGoroutine

Live goroutines
Version

Go version
metrics.Read

Named measurements

Next lesson: GOMAXPROCS

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies