Memory leak in net/http

English is not my mother languange, help me to improve this if you’d like, thanks!

What happened?

Recently we’ve get stuck with a problem that we’re serving file download service in Go, but the service is killed by server because of OOM, so we decide to dig out the reason.

we are: @jiajunhuang, @lailin

Code in server is like this:

package main

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

	_ "net/http/pprof"
)

func main() {
	go func() { // here, this goroutine is for debug
		for {
			println("gonna gc")
			runtime.GC()
			time.Sleep(time.Second * 30)
		}
	}()

	http.Handle("/download", http.HandlerFunc(func(w http.ResponseWriter, req *http.Request) {
		data, err := ioutil.ReadFile("/Users/jiajun/Images/ubuntu-18.04.1-live-server-amd64.iso")
		if err != nil {
			log.Panicf("failed to read file: %s", err)
		}
		w.Header().Set("Connection", "close")
		w.Write(data)
	}))

	log.Fatal(http.ListenAndServe(":8080", nil))
}

and client code:

package main

import (
	"log"
	"net/http"
)

func main() {
	client := &http.Client{}

	req, err := http.NewRequest("GET", "http://127.0.0.1:8080/download", nil)
	if err != nil {
		log.Panicf("error: %s", err)
	}

	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
	req.Header.Set("Connection", "close")
	resp, err := client.Do(req)
	defer resp.Body.Close()

	println("file received")
	select {} // client will block here
}

run the server and client, we then use go tool pprof to inspect the memory usage:

$ go run server.go &
$ go run client.go &
$ go tool pprof http://localhost:8080/debug/pprof/heap

and we got this:

memory leak

although we call runtime.GC periodically, the memory is still hold by Golang runtime, until we close the client:

memory leak2

Reason

Although we’ve set Connection: close header, but as RFC2616 says:

HTTP/1.1 defines the “close” connection option for the sender to signal that the connection will be closed after completion of the response.

it does not speficy who is responsible for close the connection, so Golang wish the client to close the connection.

func (w *response) Write(data []byte) (n int, err error) {
	return w.write(len(data), data, "")
}

func (w *response) write(lenData int, dataB []byte, dataS string) (n int, err error) {
	if w.conn.hijacked() {
		if lenData > 0 {
			caller := relevantCaller()
			w.conn.server.logf("http: response.Write on hijacked connection from %s (%s:%d)", caller.Function, path.Base(caller.File), caller.Line)
		}
		return 0, ErrHijacked
	}
	if !w.wroteHeader {
		w.WriteHeader(StatusOK)
	}
	if lenData == 0 {
		return 0, nil
	}
	if !w.bodyAllowed() {
		return 0, ErrBodyNotAllowed
	}

	w.written += int64(lenData) // ignoring errors, for errorKludge
	if w.contentLength != -1 && w.written > w.contentLength {
		return 0, ErrContentLength
	}
	if dataB != nil {
		return w.w.Write(dataB)
	} else {
		return w.w.WriteString(dataS)
	}
}

and the writer use a byte slice to hold the data:

// Write writes the contents of p into the buffer.
// It returns the number of bytes written.
// If nn < len(p), it also returns an error explaining
// why the write is short.
func (b *Writer) Write(p []byte) (nn int, err error) {
	for len(p) > b.Available() && b.err == nil {
		var n int
		if b.Buffered() == 0 {
			// Large write, empty buffer.
			// Write directly from p to avoid copy.
			n, b.err = b.wr.Write(p)
		} else {
			n = copy(b.buf[b.n:], p)
			b.n += n
			b.Flush()
		}
		nn += n
		p = p[n:]
	}
	if b.err != nil {
		return nn, b.err
	}
	n := copy(b.buf[b.n:], p)
	b.n += n
	nn += n
	return nn, nil
}

How to solve it

  • Write files in chunk, instead of read large files into memory


更多文章
  • OAuth 2 详解(五):Device Authorization Flow
  • OAuth 2 详解(三):Resource Owner Password Credentials Grant
  • OAuth 2 详解(四):Client Credentials Flow
  • OAuth 2 详解(二):Implict Grant Flow
  • OAuth 2 详解(一):简介及 Authorization Code 模式
  • ElasticSearch 学习笔记
  • 三种git流程以及发版模型
  • 错误处理实践
  • 权限模型(RBAC/ABAC)
  • OIDC(OpenID Connect) 简介
  • 任务队列简介
  • 使用Drone CI构建CI/CD系统
  • PostgreSQL 操作笔记
  • Golang migrate 做数据库变更管理
  • 使用PostgreSQL做搜索引擎