App Engine Go: How to kill a running goroutine - google-app-engine

Question

I want to run 2 goroutines parallel in App Engine, so that when the first goroutine finish its job, the handler doesn't need to wait the second goroutine - it stops the secend goroutine and returns the result to the client. Is this possible? I tried it with context.WithCancel(), but it didn't work (I use go1.6).
Here is my code:
package mytest
import (
"net/http"
"sync"
"time"
"golang.org/x/net/context"
"google.golang.org/appengine"
"google.golang.org/appengine/log"
"google.golang.org/appengine/urlfetch"
)
func init() {
http.HandleFunc("/test", handlerTest)
http.HandleFunc("/testwait10s", handlerTest10s)
http.HandleFunc("/testwait5s", handlerTest5s)
}
func handlerTest(w http.ResponseWriter, r *http.Request) {
c := appengine.NewContext(r)
ctx, _ := context.WithTimeout(c, 30*time.Second)
ctx1, ctx1Cancel := context.WithCancel(ctx)
ctx2, ctx2Cancel := context.WithCancel(ctx)
var wg sync.WaitGroup
wg.Add(2)
go func() {
defer wg.Done()
log.Infof(ctx1, "Go1 begin ...")
client1 := urlfetch.Client(ctx1)
_, err := client1.Get("http://APP_NAME.appspot.com/testwait5s")
if err != nil {
log.Errorf(ctx1, "Go1 failed: %v", err)
}
ctx2Cancel()
log.Infof(ctx1, "Go1 over ...")
}()
go func() {
defer wg.Done()
log.Infof(ctx2, "Go2 begin ...")
client2 := urlfetch.Client(ctx2)
_, err := client2.Get("http://APP_NAME.appspot.com/testwait10s")
if err != nil {
log.Errorf(ctx2, "Go2 failed %v", err)
}
ctx1Cancel()
log.Infof(ctx2, "Go2 over ...")
}()
wg.Wait()
log.Infof(ctx1, "Go1 and GO2 over")
}
func handlerTest10s(w http.ResponseWriter, r *http.Request) {
time.Sleep(10 * time.Second)
return
}
func handlerTest5s(w http.ResponseWriter, r *http.Request) {
time.Sleep(5 * time.Second)
return
}
Any ideas? Thanks!

Answer 1

Just create a notification channel and send there a signal that one of computations is over and you can proceed without waiting for the other.
func handlerTest(w http.ResponseWriter, r *http.Request) {
c := appengine.NewContext(r)
ctx, cancel := context.WithTimeout(c, 30*time.Second)
done := make(chan error, 2)
work := func(url, name string) {
log.Infof(ctx, "%s begin ...", name)
client := urlfetch.Client(ctx)
req, err := http.NewRequest(http.MethodGet, url, nil)
if err != nil {
log.Errorf(ctx, "%s failed: %v", name, err)
done <- err
return
}
req = req.WithContext(ctx)
_, err = client.Do(req)
done <- err
if err != nil {
log.Errorf(ctx, "%s failed: %v", name, err)
return
}
cancel()
log.Infof(ctx, "%s over ...", name)
}
go work("go1", "http://APP_NAME.appspot.com/testwait5s")
go work("go2", "http://APP_NAME.appspot.com/testwait10s")
for i := 0; i < cap(done); i++ {
if err := <-done; err == nil {
log.Infof(ctx, "executed without errors")
return
}
}
log.Error(ctx, "both computations have failed")
}

Answer 2

You can try reducing the value of wg.Add() to wg.Add(1) instead of wg.Add(2).
When one go-routine completes, wg.Done() will reduce the counter value by 1. So, In this case, the WaitGroup (wg) counter value will become ZERO. As a result, wg.Wait() on last line, will not wait for other go-routines to complete.
Note that, if the value of wg counter falls below zero, it will panic the remaining go-routines. So, the go-routines will be exited forcefully.