Overview
A lib for monitoring runtime goroutine stack. Such as wait for goroutines to exit, leak detection, etc.
Features
context.Contextfirst design- Concurrent leak detection
- No dependencies and 100% test coverage
- Provides handy low-level APIs to extend the lib
Guides
Check the examples and godoc for detailed usage.
Let's get started with a typical async producer-consumer model, can you find the leaking issue of the code below? If we visit the port 3000 again and again the service will leak memory.
package main
import (
"fmt"
"net/http"
)
func main() {
_ = http.ListenAndServe(":3000", http.HandlerFunc(handle))
}
func handle(_ http.ResponseWriter, _ *http.Request) {
c := make(chan int)
go produce(c)
go consume(c)
}
func produce(c chan int) {
for i := range "..." {
c <- i
}
}
func consume(c chan int) {
for i := range c {
fmt.Println(i)
}
}
We can use gotrace to locate the leak point, let's create a main_test.go file:
package main
import (
"testing"
"github.com/ysmood/gotrace"
)
func TestHandle(t *testing.T) {
// Just add one line before the standard test case
gotrace.CheckLeak(t, 0)
handle(nil, nil)
}
Run GODEBUG="tracebackancestors=10" go test, wait for 3 seconds you will see the below output:
--- FAIL: TestHandle (3.00s)
main_test.go:10: leaking goroutines: goroutine 6 [chan receive]:
play.consume(0x0)
/Users/ys/repos/play/default/main.go:25 +0x74
The range on line 25 get stuck, the reason is chan receive, it keeps waiting for new messages from c, but the produce function has already exited, it will not send messages to c anymore.
We have several ways to fix it, such as change the produce to:
func produce(c chan int) {
for i := range "..." {
c <- i
}
close(c)
}
Run go test again, the test should exit immediately without errors.
