nginx+php-fpm性能参数优化原则

1.worker_processes越大越好（一定数量后性能增加不明显）

2.worker_cpu_affinity所有cpu平分worker_processes 要比每个worker_processes都跨cpu分配性能要好;不考虑php的执行，测试结果worker_processes数量是cpu核数的2倍性能最优

3.unix domainsocket（共享内存的方式）要比tcp网络端口配置性能要好

不考虑backlog，请求速度有量级的飞跃，但错误率超过50%

加上backlog，性能有10%左右提升

4.调整nginx、php-fpm和内核的backlog（积压），connect()to unix:/tmp/php-fpm.socket failed (11: Resource temporarilyunavailable) while connecting to upstream错误的返回会减少

nginx：

配置文件的server块

listen 80 defaultbacklog=1024;

php-fpm:

配置文件的

listen.backlog =2048

kernel参数：

/etc/sysctl.conf，不能低于上面的配置

net.ipv4.tcp_max_syn_backlog =4096

net.core.netdev_max_backlog =4096

5.增加单台服务器上的php-fpm的master实例，会增加fpm的处理能力，也能减少报错返回的几率

多实例启动方法，使用多个配置文件：

/usr/local/php/sbin/php-fpm -y/usr/local/php/etc/php-fpm.conf &

/usr/local/php/sbin/php-fpm -y/usr/local/php/etc/php-fpm1.conf &

nginx的fastcgi配置

   upstream phpbackend {

#     server  127.0.0.1:9000 weight=100 max_fails=10fail_timeout=30;

#     server  127.0.0.1:9001 weight=100 max_fails=10fail_timeout=30;

#     server  127.0.0.1:9002 weight=100 max_fails=10fail_timeout=30;

#     server  127.0.0.1:9003 weight=100 max_fails=10fail_timeout=30;

     server  unix:/var/www/php-fpm.sock weight=100 max_fails=10fail_timeout=30;

     server  unix:/var/www/php-fpm1.sock weight=100 max_fails=10fail_timeout=30;

     server  unix:/var/www/php-fpm2.sock weight=100 max_fails=10fail_timeout=30;

     server  unix:/var/www/php-fpm3.sock weight=100 max_fails=10fail_timeout=30;

#     server  unix:/var/www/php-fpm4.sock weight=100 max_fails=10fail_timeout=30;

#     server  unix:/var/www/php-fpm5.sock weight=100 max_fails=10fail_timeout=30;

#     server  unix:/var/www/php-fpm6.sock weight=100 max_fails=10fail_timeout=30;

#     server  unix:/var/www/php-fpm7.sock weight=100 max_fails=10fail_timeout=30;

   }

       location ~\.php* {

          fastcgi_pass  phpbackend;

#         fastcgi_pass  unix:/var/www/php-fpm.sock;

          fastcgi_indexindex.php;

      ..........

      }

6.测试环境和结果

内存2G

swap2G

cpu 2核 Intel(R) Xeon(R) CPUE5405  @ 2.00GHz

采用ab远程访问测试，测试程序为php的字符串处理程序

1）在开4个php-fpm实例，nginx8个worker_processes 每个cpu4个worker_processes，backlog为1024，php的backlog为2048，内核backlog为4096，采用unix domainsocket连接的情况下，其他保持参数不变

性能和错误率较为平衡，可接受，超过4个fpm实例，性能开始下降，错误率并没有明显下降

结论是fpm实例数，worker_processes数和cpu保持倍数关系，性能较高

影响性能和报错的参数为

php-fpm实例， nginxworker_processes数量，fpm的max_request， php的backlog， unix domain socket

10W请求，500并发无报错，1000并发报错率为0.9%

500并发：

Timetaken for tests:   25 secondsavg.

Complete requests:    100000

Failed requests:      0

Write errors:         0

Requests per second:   4000 [#/sec] (mean) avg.

Timeper request:      122.313 [ms] (mean)

Timeper request:      0.245 [ms] (mean, across all concurrentrequests)

Transfer rate:         800[Kbytes/sec] received avg.

1000并发：

Time taken for tests:  25 seconds avg.

Complete requests:    100000

Failed requests:      524

  (Connect: 0, Length: 524, Exceptions: 0)

Write errors:         0

Non-2xx responses:    524

Requests per second:   3903.25 [#/sec](mean)

Time per request:      256.197[ms] (mean)

Time per request:      0.256 [ms](mean, across all concurrent requests)

Transfer rate:        772.37 [Kbytes/sec]received

2）在其他参数不变，unix domainsocket换为tcp网络端口连接，结果如下

500并发：

Concurrency Level:    500

Timetaken for tests:   26.934431seconds

Complete requests:    100000

Failed requests:      0

Write errors:         0

Requests per second:   3712.72 [#/sec](mean)

Time per request:      134.672[ms] (mean)

Time per request:      0.269 [ms](mean, across all concurrent requests)

Transfer rate:        732.37 [Kbytes/sec]received

1000并发：

Concurrency Level:    1000

Time taken for tests:  28.385349 seconds

Complete requests:    100000

Failed requests:      0

Write errors:         0

Requests per second:   3522.94 [#/sec](mean)

Time per request:      283.853[ms] (mean)

Time per request:      0.284 [ms](mean, across all concurrent requests)

Transfer rate:        694.94 [Kbytes/sec]received

与1）比较，有大约10%的性能下降

7.5.16调整fpm的max_request参数为1000，并发1000报错返回降到200个以下，

Transfer rate在800左右