网站被大量爬虫访问，经常出现访问奇慢的情况，经过不断的探索，终于把php-fpm多开、nginx限制并发、定时重启php-fpm三者结合起来，解决了困扰已久的难题。
第一步、php-fpm多开
要修改三个地方，nginx配置，php配置，/etc/init.d/中增加php-fpm2
nginx里面添加
upstream backend
{
server 127.0.0.1:9000;
server 127.0.0.1:9002;
}转发用tcp协议，不用sock，主要的原因是tcp更稳定，速度差别感觉不到
server中修改
location ~ [^/]\.php(/|$)
{
try_files $uri =404;
fastcgi_pass backend;
fastcgi_index index.php;
include fastcgi.conf;
}
php配置文件复制一份，分别是php-fpm.conf和php-fpm2.conf
位置在php/etc里面
php-fpm.conf
[global]
pid = run/php-fpm.pid
error_log = /root/php.log
log_level = warning
[www]
listen = 127.0.0.1:9000
listen.backlog = 4096
listen.allowed_clients = 127.0.0.1
listen.owner = www
listen.group = www
listen.mode = 0666
user = www
group = www
pm = dynamic
pm.max_children = 60
pm.start_servers = 10
pm.min_spare_servers = 10
pm.max_spare_servers = 30
pm.max_requests = 10240
request_terminate_timeout = 100
request_slowlog_timeout = 20
slowlog = /root/slow.log
php-fpm2.conf
[global]
pid = run/php-fpm2.pid
error_log = /root/php.log
log_level = warning
[www]
listen = 127.0.0.1:9002
listen.backlog = 4096
listen.allowed_clients = 127.0.0.1
listen.owner = www
listen.group = www
listen.mode = 0666
user = www
group = www
pm = dynamic
pm.max_children = 60
pm.start_servers = 10
pm.min_spare_servers = 10
pm.max_spare_servers = 30
pm.max_requests = 10240
request_terminate_timeout = 100
request_slowlog_timeout = 20
slowlog = /root/slow.log
重点讲讲
pm = dynamic
pm.max_children = 60
pm.start_servers = 10
pm.min_spare_servers = 10
pm.max_spare_servers = 30
pm.max_requests = 10240
由于是开了两个php-fpm，因此开始的时候是10+10=20个进程，最高的时候限制为120个，正常在60个以内是够用的。
max_spare_servers要低于max_children，好处是高并发后会自动降低进程，减少系统开销。
cd /etc/init.d
复制一份php-fpm，名称? php-fpm2? ? ?
php-fpm2 全部内容
#! /bin/sh
### BEGIN INIT INFO
# Provides: php-fpm
# Required-Start: $remote_fs $network
# Required-Stop: $remote_fs $network
# Default-Start: 2 3 4 5
# Default-Stop: 0 1 6
# Short-Description: starts php-fpm
# Description: starts the PHP FastCGI Process Manager daemon
### END INIT INFO
prefix=/www/server/php/56
exec_prefix=${prefix}
php_fpm_BIN=${exec_prefix}/sbin/php-fpm
php_fpm_CONF=${prefix}/etc/php-fpm2.conf
php_fpm_PID=${prefix}/var/run/php-fpm2.pid
php_opts="--fpm-config $php_fpm_CONF --pid $php_fpm_PID"
wait_for_pid () {
try=0
while test $try -lt 35 ; do
case "$1" in
'created')
if [ -f "$2" ] ; then
try=''
break
fi
;;
'removed')
if [ ! -f "$2" ] ; then
try=''
break
fi
;;
esac
echo -n .
try=`expr $try + 1`
sleep 1
done
}
case "$1" in
start)
echo -n "Starting php-fpm "
$php_fpm_BIN --daemonize $php_opts
if [ "$?" != 0 ] ; then
echo " failed"
exit 1
fi
wait_for_pid created $php_fpm_PID
if [ -n "$try" ] ; then
echo " failed"
exit 1
else
echo " done"
fi
;;
stop)
echo -n "Gracefully shutting down php-fpm "
if [ ! -r $php_fpm_PID ] ; then
echo "warning, no pid file found - php-fpm is not running ?"
exit 1
fi
kill -QUIT `cat $php_fpm_PID`
wait_for_pid removed $php_fpm_PID
if [ -n "$try" ] ; then
echo " failed. Use force-quit"
exit 1
else
echo " done"
fi
;;
status)
if [ ! -r $php_fpm_PID ] ; then
echo "php-fpm is stopped"
exit 0
fi
PID=`cat $php_fpm_PID`
if ps -p $PID | grep -q $PID; then
echo "php-fpm (pid $PID) is running..."
else
echo "php-fpm dead but pid file exists"
fi
;;
force-quit)
echo -n "Terminating php-fpm "
if [ ! -r $php_fpm_PID ] ; then
echo "warning, no pid file found - php-fpm is not running ?"
exit 1
fi
kill -TERM `cat $php_fpm_PID`
wait_for_pid removed $php_fpm_PID
if [ -n "$try" ] ; then
echo " failed"
exit 1
else
echo " done"
fi
;;
restart)
$0 stop
$0 start
;;
reload)
echo -n "Reload service php-fpm "
if [ ! -r $php_fpm_PID ] ; then
echo "warning, no pid file found - php-fpm is not running ?"
exit 1
fi
kill -USR2 `cat $php_fpm_PID`
echo " done"
;;
configtest)
$php_fpm_BIN -t
;;
*)
echo "Usage: $0 {start|stop|force-quit|restart|reload|status|configtest}"
exit 1
;;
esac
核心是修改
prefix=/www/server/php/56
exec_prefix=${prefix}
php_fpm_BIN=${exec_prefix}/sbin/php-fpm
php_fpm_CONF=${prefix}/etc/php-fpm2.conf
php_fpm_PID=${prefix}/var/run/php-fpm2.pid
然后执行
chkconfig --add php-fpm2
设置开机启动
至此，php-fpm双开已经完成。
第二步、限制nginx并发
在nginx的http中添加
limit_conn_zone $binary_remote_addr zone=one:10m;
limit_req_zone $binary_remote_addr zone=two:100m rate=10r/s;server中添加
limit_conn one 10;
limit_req zone=two burst=10;限制单个ip连接数低于10个，并发数10（burst=10），可根据情况修改为20或者30，经测试不带nodelay更稳定。
第三步、定时重启
重启服务
vi /sbin/cq
#/bin/bash
/etc/init.d/nginx restart
/etc/init.d/php-fpm restart
/etc/init.d/php-fpm2 restart
/etc/init.d/mysql restart平滑重启服务
vi /sbin/cq2
#! /bin/sh
getDateTime(){
date "+%m-%d %H:%M:%S"
}
date=$(getDateTime)
count=$(ps aux | grep -c php-fpm)
echo "$date -- $count" >> /www/jt160/cq.txt
/etc/init.d/php-fpm reload
sleep 1
/etc/init.d/php-fpm2 reload
/etc/init.d/mysql reload添加执行权限
chmod +x /sbin/*
定时每20秒平滑重启php-fpm，解决php-fpm卡死的问题，经测试，非常好用。
crontab -e
* * * * * /sbin/cq2
* * * * * sleep 20;/sbin/cq2
* * * * * sleep 40;/sbin/cq2重启的记录保存在 /www/jt160/cq.txt? ?可以是网站目录，可以随时查看，通常100以内是正常的，php-fpm超过100就是负载太高的时候，一目了然。
重启所有服务输入命令? cq 即可
通过php-fpm多开、nginx限制并发、定时重启php-fpm进程三者结合，网站稳定性上一个大台阶，分享给大家。
如有疑问，欢迎交流。?
定时重启记录如下
