Nginx合集-性能优化（nginx性能调优）
1、工具
ab：Apache Benchmark(简称ab) 是Apache安装包中的压力测试工具 ，本案使用ab进行性能测试。
2、ab安装
需要安装依赖apr-util包
yum install apr-util yumdownload命令安装：
yum install yum-utils创建目录
mkdir -p /appdata/apache/ab && cd /appdata/apache/ab下载安装包
yum install yum-utils.noarch
yumdownloader httpd-tools* 解压安装包
rpm2cpio httpd-tools*.rpm | cpio -idmv将./user/bin/ab复制到系统bin下
cp /appdata/apache/ab/usr/bin/ab /usr/bin3、ab压测
服务器配置
操作系统：CentOS Linux release 7.9.2009 (Core)
cpu：8核
内存：16G
磁盘：90G压测：20万，无静态资源
ab -n 200000 -c 5000 http://xxx.xxx.xxx.xxx/index.html # -n 请求次数 -c 并发数
#报告
Server Software: nginx
Server Hostname: 10.0.53.84
Server Port: 80
Document Path: /index.html
Document Length: 612 bytes
Concurrency Level: 5000 #并发数
Time taken for tests: 12.075 seconds #总请求时长
Complete requests: 200000 #总请求数
Failed requests: 0
Write errors: 0
Total transferred: 204200000 bytes
HTML transferred: 122400000 bytes
Requests per second: 16562.75 [#/sec] (mean) #QPS 吞吐量
Time per request: 301.882 [ms] (mean) #用户请求平均等待时间
Time per request: 0.060 [ms] (mean, across all concurrent requests) #服务器处理每个请求花费的时间
Transfer rate: 16514.23 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 131 35.5 130 1139
Processing: 56 168 29.5 161 276
Waiting: 0 125 31.4 110 204
Total: 162 299 36.2 294 1301
Percentage of the requests served within a certain time (ms)
50% 294
66% 309
75% 312
80% 313
90% 317
95% 323
98% 332
99% 337
100% 1301 (longest request)压测：100万，无静态资源
ab -n 1000000 -c 5000 http://10.0.53.84/index.html
#报告
Server Software: nginx
Server Hostname: 10.0.53.84
Server Port: 80
Document Path: /index.html
Document Length: 612 bytes
Concurrency Level: 5000
Time taken for tests: 59.294 seconds
Complete requests: 1000000
Failed requests: 0
Write errors: 0
Total transferred: 1021000000 bytes
HTML transferred: 612000000 bytes
Requests per second: 16865.11 [#/sec] (mean)
Time per request: 296.470 [ms] (mean)
Time per request: 0.059 [ms] (mean, across all concurrent requests)
Transfer rate: 16815.70 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 131 73.2 129 3136
Processing: 52 165 25.7 156 227
Waiting: 1 123 29.2 106 195
Total: 171 296 74.4 288 3305
Percentage of the requests served within a certain time (ms)
50% 288
66% 308
75% 310
80% 311
90% 313
95% 315
98% 317
99% 320
100% 3305 (longest request)压测：20万，1M内的静态资源（一般前端包不大于1Mb）
ab -n 200000 -c 5000 http://10.0.53.84/1.jpg
#报告
Server Software: nginx
Server Hostname: 10.0.53.84
Server Port: 80
Document Path: /1.jpg
Document Length: 999398 bytes #静态资源文件大小
Concurrency Level: 5000 #并发数
Time taken for tests: 85.020 seconds #总请求时长
Complete requests: 200000 #总请求数
Failed requests: 0
Write errors: 0
Total transferred: 199962600000 bytes
HTML transferred: 199879600000 bytes
Requests per second: 2352.39 [#/sec] (mean) #QPS 吞吐量
Time per request: 2125.498 [ms] (mean) #用户请求平均等待时间
Time per request: 0.425 [ms] (mean, across all concurrent requests) #服务器处理每个请求花费的时间
Transfer rate: 2296826.54 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 0 36 16.2 34 174
Processing: 937 2081 132.4 2094 2711
Waiting: 0 44 57.0 37 676
Total: 1086 2117 128.5 2129 2791
Percentage of the requests served within a certain time (ms)
50% 2129
66% 2139
75% 2146
80% 2150
90% 2166
95% 2174
98% 2186
99% 2398
100% 2791 (longest request)
服务器性能截图
编辑
结论：根据连云港nginx请求统计在高峰时刻1分钟内达到6万次请求时候系统保持稳定，超过8万次请求时候接口会返回系统异常情况。优化配置系统内核和nginx配置文件之后，进行nginx入口并发压测，读取1Mb的文件，每秒可请求2352.39次，预估1分钟可达到10万请求，能够满足前端的高峰性能要求。
4、优化记录
优化nginx的配置文件中http模块
sendfile on;
tcp_nopush on;
client_max_body_size 1024m;
client_body_buffer_size 10m;
client_header_buffer_size 10m;
proxy_buffers 4 128k;
proxy_busy_buffers_size 128k;
open_file_cache max=102400 inactive=20s;
#这个将为打开文件指定缓存，默认是没有启用的，max指定缓存数量，建议和打开文件数一致，inactive是指经过多长
时间文件没被请求后删除缓存。
open_file_cache_valid 30s;
keepalive_timeout 60;
优化linux内核参数
#增加tcp支持的队列数
net.ipv4.tcp_max_syn_backlog = 262144
#减少断开连接时 ，资源回收
net.ipv4.tcp_max_tw_buckets = 8000
# 表示开启重用。
# 允许将TIME-WAIT sockets重新用于新的TCP连接，默认为0，表示关闭；
net.ipv4.tcp_tw_reuse = 1
net.ipv4.tcp_tw_recycle = 1
# 表示如果套接字由本端要求关闭，这个参数决定了它保持在FIN-WAIT-2状态的时间。针对nginx
net.ipv4.tcp_fin_timeout = 10
#改变本地的端口范围
net.ipv4.ip_local_port_range = 1024 65535
#对于只在本地使用的数据库服务器
net.ipv4.tcp_fin_timeout = 1
#端口监听队列
net.core.somaxconn=65535
#接受数据的速率
net.core.netdev_max_backlog=262144
net.core.wmem_default=8388608
net.core.wmem_max=16777216
net.core.rmem_default=8388608
net.core.rmem_max=16777216
# 开启SYN Cookies。
# 当出现SYN等待队列溢出时，启用cookies来处理，可防范少量SYN攻击，
# 默认为0，表示关闭；
net.ipv4.tcp_syncookies = 1
net.ipv4.tcp_max_orphans = 3276800
# 系统默认值为”65536”，系统会主动丢掉新连接包，直到连接小于此设置值才会恢复。
net.nf_conntrack_max = 655360
# 单位s，系统默认值为”432000”，代表nf_conntrack的TCP连接记录时间默认是5天，致使nf_conntrack的值减不下来，丢包持续时间长。
net.netfilter.nf_conntrack_tcp_timeout_established = 1200
