因为之前发生502了，日志提示

[27-May-2015 13:54:34] WARNING: [pool www] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 32 children, there are 16 idle, and 602 total children

然后我就自己随意增加了pm.start_servers参数，一段时间后，php日志/usr/local/php/var/log/php-fpm.log部分信息如下：

[27-May-2015 15:13:48] NOTICE: [pool www] child 3998 started

[27-May-2015 15:13:59] NOTICE: [pool www] child 9242 exited with code 0 after 47575.960999 seconds from start

[27-May-2015 15:13:59] NOTICE: [pool www] child 4210 started

[27-May-2015 15:14:42] NOTICE: [pool www] child 9217 exited with code 0 after 47619.884065 seconds from start

[27-May-2015 15:14:42] NOTICE: [pool www] child 5621 started

[27-May-2015 15:14:43] NOTICE: [pool www] child 9561 exited with code 0 after 47620.024499 seconds from start

原因就是没有pm.start_servers这个参数没有按照下面的这个公式来:

pm.start_servers= min_spare_servers + (max_spare_servers - min_spare_servers) / 2

一般min_spare_servers设置为20

max_spare_servers是根据服务器本身的内存来计算的，标准算法就是内存大小除以30M

当然，有的php程序可能占用比较小，不到30M，这就看情况来计算了。

我设置的是这样的

过滤注释部分:

cat /usr/local/php/etc/php-fpm.conf | grep -v "^;" | grep -v "^$"

[global]

pid = run/php-fpm.pid

error_log = log/php-fpm.log

emergency_restart_threshold = 10

emergency_restart_interval = 1m

process_control_timeout = 10s

rlimit_files = 65535

events.mechanism = epoll

[www]

user = www

group = www

listen = 127.0.0.1:9000

pm = dynamic

pm.max_children = 1400

pm.start_servers = 710

pm.min_spare_servers = 20

pm.max_spare_servers = 1400

pm.max_requests = 10000

slowlog = log/$pool.log.slow

然后重新加载下
/usr/local/nginx/sbin/nginx -s reload

之后几天日志没有出现信息了。