/network

a great network project

Primary LanguageCMIT LicenseMIT

Network

purpose for this repositories

  1. Basic code used for learning network with POSIX functions
  2. Test for the speed of each implementation

protocol

  1. TCP

  • Linux TCP server and client demo

  • linux查看socket连接

netstat -antp
  • mac查看socket连接
lsof -p PID | grep TCP

  1. UDP

    Linux UDP server and client demo

  2. HTTP

    HTTP client and server demo

TCP State

  • LISTEN represents waiting for a connection request from any remote TCP and port.
  • SYN-SENT represents waiting for a matching connection request after having sent a connection request.
  • SYN-RECEIVED represents waiting for a confirming connection request acknowledgment after having both received and sent a connection request.
  • ESTABLISHED represents an open connection, data received can be delivered to the user. The normal state for the data transfer phase of the connection.
  • FIN-WAIT-1 represents waiting for a connection termination request from the remote TCP, or an acknowledgment of the connection termination request previously sent.
  • FIN-WAIT-2 represents waiting for a connection termination request from the remote TCP.
  • CLOSE-WAIT represents waiting for a connection termination request from the local user.
  • CLOSING represents waiting for a connection termination request acknowledgment from the remote TCP.
  • LAST-ACK represents waiting for an acknowledgment of the connection termination request previously sent to the remote TCP (which includes an acknowledgment of its connection termination request).
  • TIME-WAIT represents waiting for enough time to pass to be sure the remote TCP received the acknowledgment of its connection termination request.
  • CLOSED represents no connection state at all.

establish

/*
state			local			remote			state
CLOSED			|					|			CLOSED
				|					|
				|	  SYN	seq=n	|			listen
----------------|==================>|----------------------
				|					|
				|	ACK	ack_seq=n+1	|
				|	SYN seq=m		|
SYN-SENT		|<==================|			SYN-RECEIVED
				|					|
				|	ACK	ack_seq=m+1	|
----------------|==================>|----------------------
				|					|
ESTABLISHED		|					|			ESTABLISHED
*/

close

/*
state			local			remote			state
ESTABLISHED		|					|			ESTABLISHED
				|					|
				|	  FIN	seq=n	|
----------------|==================>|----------------------
FIN-WAIT-1		|					|
				|	ACK	ack_seq=n+1	|
----------------|<==================|			CLOSE-WAIT
				|	  FIN   seq=n	|
				|<==================|----------------------
FIN-WAIT-2		|					|
				|					|			LAST-ACK
				|	ACK	ack_seq=n+1	|
----------------|==================>|----------------------
				|					|
TIME-WAIT		|					|
				|					|
----------------|					|			CLOSED
				|					|
CLOSED			|					|
*/

  1. TIME-WAIT

    当最后一个ack丢失时，remote端启动定时重传fin,如果没有TIME-WAIT,会一直发出无效包

    默认值120s，在linux下设置方式

/etc/sysctl.conf
# 表示开启SYN Cookies。当出现SYN等待队列溢出时，启用Cookie来处理，可防范少量的SYN攻击。该参数默认为0，表示关闭
net.ipv4.tcp_syncookies＝1
# 表示开启重用，即允许将TIME-WAIT套接字重新用于新的TCP连接。该参数默认为0，表示关闭
net.ipv4.tcp_tw_reuse＝1
# 表示开启TCP连接中TIME-WAIT套接字的快速回收，该参数默认为0，表示关闭
net.ipv4.tcp_tw_recycle＝1
# 表示如果套接字由本端要求关闭，那么这个参数将决定它保持在FIN-WAIT-2状态的时间
net.ipv4.tcp_fin_timeout＝30
# 表示当Keepalived启用时，TCP发送Keepalived消息的频度改为20分钟，默认值是2小时
net.ipv4.tcp_keepalive_time＝1200
# 表示系统同时保持TIME_WAIT套接字的最大数量，如果超过这个数字，TIME_WAIT套接字将立刻被清除并打印警告信息，默认值为180 000，此处改为5000。对于Apache、Nginx等服务器，前面介绍的几个参数已经可以很好地减少TIME_WAIT套接字的数量，但是对于Squid来说，效果却不大，有了此参数就可以控制TIME_WAIT套接字的最大数量，避免Squid服务器被大量的TIME_WAIT套接字拖死
net.ipv4.tcp_max_tw_buckets＝5000

sysctl -p /etc/sysctl.conf

  1. FIN-WAIT-1

    client先异常断开没有发送FIN,server主动关闭进入FIN-WAIT-1
    无法收到FIN,会一直处于该状态,该状态系统有time_out,尚未查证具体时间

# record what tcp_max_orphans's current value
original_value=$(cat /proc/sys/net/ipv4/tcp_max_orphans)

#set the tcp_max_orphans to 0 temporarily
echo 0 > /proc/sys/net/ipv4/tcp_max_orphans

# watch /var/log/messages
# it will split out "kernel: TCP: too many of orphaned sockets"
# it won't take long for the connections to be killed

# restore the value of tcp_max_orphans whatever it was before.
echo $original_value > /proc/sys/net/ipv4/tcp_max_orphans

常见名词

  • 慢启动

    门限值以下是慢启动，以上是拥塞控制

  • 拥塞控制

    rtt(round trip time) rtt = 0.9*pre_rtt + cur_rtt rtt增大window减小，rtt减小window增大

  • 滑动窗口 [ack,ack+window]

  • 超时重传

  • 延迟确认

  • 半连接/全连接(syn队列/accept队列)

  • 全双工/半双工

  • 三次握手

  • 四次挥手

TCP vs UDP

  1. UDP实时性高于TCP
  2. UDP传输效率高于TCP

epoll

reactor & proactor

两种高效的IO事件处理模型

单台服务器百万并发

  1. server系统限制
  • 单个进程最多打开的文件数
ulimit -a # 查看全部
ulimit -n # 关注open files

vim /etc/security/limits.conf
修改fs.file_max
  • 单个进程的maxfd
sysctl -a | grep file_max
# 修改文件
/etc/sysctl.conf
net.ipv4.tcp_men = 262144 524288 786432
net.ipv4.tcp_wmen = 1024 1024 2048
net.ipv4.tcp_rmen = 1024 1024 2048
fs.file_max =1048576 #max fd

sysctl -p #生效

  1. 单个客户端最多连接数限制

    server端开发不用关注，单台客户端机器测试用

# 修改文件
/etc/sysctl.conf
net.nf_conntrack_max = 1048576
net.netfilter.nf_conntrack_tcp_timeout_established = 1200

sysctl -p #生效

单机服务器IO模型

  1. 单线程同步 NTP
  2. 多线程同步 Natty
  3. 纯异步 redis,HAProxy
  4. 半同步半异步 Natty
  5. 多进程同步 fastcgi
  6. 多线程异步 memcached
  7. 多进程异步 nginx
  8. 每请求每进程(线程) Apache/CGI
  9. 微进程架构 erlang/go/lua
  10. 协程架构 libco

opensource network framework

boost.asio libevent https://github.com/Tencent/mars