client and server odd behaviour

client.go

package main

import (
	"bytes"
	"io"
	"log"
	"net"
	"time"
)

/*
usage:
  go run -race client.go
*/

func main() {
	start := time.Now()
	Backend := "localhost:3000"

	b := `POST /post HTTP/1.1
Host: localhost:3000
User-Agent: curl/7.54.0
accept: application/json
Content-Type: application/json
Content-Length: 15

{"name":"komu"}`

	requestBytes := []byte(b)
	backendConn, err := net.Dial("tcp", Backend)
	if err != nil {
		log.Fatalf("dial failed %+v", err)
	}
	defer backendConn.Close()
	err = backendConn.SetDeadline(time.Now().Add(15 * time.Second))
	if err != nil {
		log.Fatalf("unable to set backendConn deadline %+v", err)
	}
	log.Printf("frontend connected to backend %v", Backend)

	io.Copy(backendConn, bytes.NewReader(requestBytes))
	log.Printf("send request to backend %v took %v seconds", Backend, time.Since(start).Seconds())
	responseBuf := new(bytes.Buffer)
	io.Copy(responseBuf, backendConn)

	log.Printf("read response from backend %v took %v seconds", Backend, time.Since(start).Seconds())
	log.Println(string(responseBuf.Bytes()))
}

server.go

package main

import (
	"io/ioutil"
	"log"
	"net/http"
	"time"
)

/*
usage:
  go run -race server.go
*/

func main() {
	http.HandleFunc("/post", echoHandler)
	log.Println("listening at localhost:3000")
	err := http.ListenAndServe(":3000", nil)
	if err != nil {
		log.Fatal("unable to listen and serve on port 3000", err)
	}
}

func echoHandler(w http.ResponseWriter, r *http.Request) {
	start := time.Now()
	reqBytes, err := ioutil.ReadAll(r.Body)
	if err != nil {
		log.Printf("unable to read request body %+v", err)

		http.Error(w, err.Error(), http.StatusInternalServerError)
		return
	}

	time.Sleep(1 * time.Second)
	w.Header().Set("Content-Type", "application/json")
	w.Write(reqBytes)
	log.Printf("served response in %v seconds", time.Since(start).Seconds())
}

Run the server in one terminal;
go run -race server.go
and the client in another;
go run -race client.go

The odd thing is;
Even though the server only takes 1second to reply,
the client always waits 15seconds(ie the time specified in backendConn.SetDeadline) before exiting.

2019/01/04 17:11:48 read response from backend localhost:3000 took 15.006758846 seconds

However, if we changed the backend to something else, eg;

Backend := "google.com:80"

the client does not wait the full 15seconds before exiting

2019/01/04 17:05:52 read response from backend google.com:80 took 0.582892931 seconds

I would expect the behaviour of the client in the second case to be the right one always