[GH-ISSUE #1335] frpc 在login登录失败时，连接关不掉 #1059

New issue

Closed

opened 2026-05-05 12:40:43 -06:00 by gitea-mirror · 8 comments

gitea-mirror commented

2026-05-05 12:40:43 -06:00

Owner

Originally created by @luyaotang on GitHub (Jul 23, 2019).
Original GitHub issue: https://github.com/fatedier/frp/issues/1335

问题描述：
frpc 在login登录失败时，连接关不掉，通过netstat -anp来查看还是维持着Established。这就会导致登录失败时，会创建了1000多个连接，状态还是Established。

版本：
目前用的是0.22.0,但对比了最新的0.27.2发现在相关的业务处理上代码并未变更

分析：
代码所有文件 server\service.go: Line 273。
conn.Close()这个动作并不能关掉连接，释放掉。
分析如下：在关闭前输出Remoaddr,LocalAddr,然后netsta -anp|grep portxxxx ，发现还是存在的，状态为Established。尝试使用frpConn.Close()就可以。
然后解决上面所说的连接没有关掉并释放的问题。

func (svr *Service) HandleListener(l frpNet.Listener) {
// Listen for incoming connections from client.
for {
c, err := l.Accept()
if err != nil {
log.Warn("Listener for incoming connections from client closed")
return
}

	// Start a new goroutine for dealing connections.
	go func(frpConn frpNet.Conn) {
		dealFn := func(conn frpNet.Conn) {
			var rawMsg msg.Message
			conn.SetReadDeadline(time.Now().Add(connReadTimeout))
			if rawMsg, err = msg.ReadMsg(conn); err != nil {
				log.Trace("Failed to read message: %v", err)
				conn.Close()
				return
			}
			conn.SetReadDeadline(time.Time{})

			switch m := rawMsg.(type) {
			case *msg.Login:
				err = svr.RegisterControl(conn, m)
				// If login failed, send error message there.
				// Otherwise send success message in control's work goroutine.
				if err != nil {
					conn.Warn("%v", err)
					msg.WriteMsg(conn, &msg.LoginResp{
						Version: version.Full(),
						Error:   err.Error(),
					})
					//log.Error("msg.Login RegisterControl fail,%s",conn.RemoteAddr().String())
					conn.Close()
					//不知道为什么，好像关不掉，netstat -anp还是维持着establised
					frpConn.Close()
				}
			case *msg.NewWorkConn:
				svr.RegisterWorkConn(conn, m)

Originally created by @luyaotang on GitHub (Jul 23, 2019). Original GitHub issue: https://github.com/fatedier/frp/issues/1335 问题描述： frpc 在login登录失败时，连接关不掉，通过netstat -anp来查看还是维持着Established。这就会导致登录失败时，会创建了1000多个连接，状态还是Established。版本：目前用的是0.22.0,但对比了最新的0.27.2发现在相关的业务处理上代码并未变更分析：代码所有文件 server\service.go: Line 273。 conn.Close()这个动作并不能关掉连接，释放掉。分析如下：在关闭前输出Remoaddr,LocalAddr,然后netsta -anp|grep portxxxx ，发现还是存在的，状态为Established。尝试使用frpConn.Close()就可以。然后解决上面所说的连接没有关掉并释放的问题。 func (svr *Service) HandleListener(l frpNet.Listener) { // Listen for incoming connections from client. for { c, err := l.Accept() if err != nil { log.Warn("Listener for incoming connections from client closed") return } // Start a new goroutine for dealing connections. go func(frpConn frpNet.Conn) { dealFn := func(conn frpNet.Conn) { var rawMsg msg.Message conn.SetReadDeadline(time.Now().Add(connReadTimeout)) if rawMsg, err = msg.ReadMsg(conn); err != nil { log.Trace("Failed to read message: %v", err) conn.Close() return } conn.SetReadDeadline(time.Time{}) switch m := rawMsg.(type) { case *msg.Login: err = svr.RegisterControl(conn, m) // If login failed, send error message there. // Otherwise send success message in control's work goroutine. if err != nil { conn.Warn("%v", err) msg.WriteMsg(conn, &msg.LoginResp{ Version: version.Full(), Error: err.Error(), }) //log.Error("msg.Login RegisterControl fail,%s",conn.RemoteAddr().String()) conn.Close() //不知道为什么，好像关不掉，netstat -anp还是维持着establised frpConn.Close() } case *msg.NewWorkConn: svr.RegisterWorkConn(conn, m)

gitea-mirror

2026-05-05 12:40:43 -06:00

closed this issue
added the
bug
label

gitea-mirror commented

2026-05-05 12:40:47 -06:00

Author

Owner

@fatedier commented on GitHub (Jul 23, 2019):

无法复现这个问题，建议更新版本在本地重新尝试验证一下。

@fatedier commented on GitHub (Jul 23, 2019): 无法复现这个问题，建议更新版本在本地重新尝试验证一下。

gitea-mirror commented

2026-05-05 12:40:47 -06:00

Author

Owner

@luyaotang commented on GitHub (Jul 23, 2019):

用了最新的release 0.27.1版本。复现流程如下，目标是让frpc和frps这两的的token不一样，引起登录失败。
1.frpc的配置：
[common]
server_addr = 192.168.22.254
server_port = 7000
token = abc123
login_fail_exit = false

[ssh]
type = tcp
local_ip = 192.168.22.254
local_port = 22
remote_port = 6000

2.frps的配置:
[common]
bind_addr = 0.0.0.0
bind_port = 7000
log_level = info
log_max_days = 3
allow_ports = 2000-3000,3001,3003,4000-50000
token = defa123

3.两边各自运行，加载配置
frpcp这边就开始报：
2019/07/23 15:24:58 [W] [service.go:82] login to server failed: authorization failed
2019/07/23 15:25:08 [E] [service.go:215] authorization failed

frps这边开始报：
2019/07/23 15:25:43 [I] [service.go:349] client login info: ip [192.168.22.30:25363] version [0.27.1] hostname [] os [windows] arch [amd64]
2019/07/23 15:25:43 [W] [service.go:291] authorization failed

4.这时候再看netstat -anp|grep frp就可以看到不断增长的连接数了。

补充go环境：
[root@zmdevserver frp]# go version
go version go1.11.5 linux/amd64

@luyaotang commented on GitHub (Jul 23, 2019): 用了最新的release 0.27.1版本。复现流程如下，目标是让frpc和frps这两的的token不一样，引起登录失败。 1.frpc的配置： [common] server_addr = 192.168.22.254 server_port = 7000 token = abc123 login_fail_exit = false [ssh] type = tcp local_ip = 192.168.22.254 local_port = 22 remote_port = 6000 2.frps的配置: [common] bind_addr = 0.0.0.0 bind_port = 7000 log_level = info log_max_days = 3 allow_ports = 2000-3000,3001,3003,4000-50000 token = defa123 3.两边各自运行，加载配置 frpcp这边就开始报： 2019/07/23 15:24:58 [W] [service.go:82] login to server failed: authorization failed 2019/07/23 15:25:08 [E] [service.go:215] authorization failed frps这边开始报： 2019/07/23 15:25:43 [I] [service.go:349] client login info: ip [192.168.22.30:25363] version [0.27.1] hostname [] os [windows] arch [amd64] 2019/07/23 15:25:43 [W] [service.go:291] authorization failed 4.这时候再看netstat -anp|grep frp就可以看到不断增长的连接数了。补充go环境： [root@zmdevserver frp]# go version go version go1.11.5 linux/amd64

gitea-mirror commented

2026-05-05 12:40:48 -06:00

Author

Owner

@fatedier commented on GitHub (Jul 23, 2019):

login_fail_exit = false 去掉这行

@fatedier commented on GitHub (Jul 23, 2019): `login_fail_exit = false` 去掉这行

gitea-mirror commented

2026-05-05 12:40:49 -06:00

Author

Owner

@fatedier commented on GitHub (Jul 23, 2019):