大型网站架构之百万PV

• 简介

PV即点击量，通常是衡量一个网络新闻频道或网站甚至一条网络新闻的重要指标。PV从某种程度上已经成为投资者衡量商业网站表现的最重要的尺度。

PV是一个访问者在24小时到底看了网站的几个页面。

• 案例描述

本案例设计采用四层模式实现，主要分为前端反向代理，web层，数据库缓存层和数据库层。前端反向代理层采用主备模式，web层采用集群模式，数据库缓存采用主备模式，数据库层采用主从模式。

• 案例环境

主：192.168.177.145 centos7-1

从：192.168.177.135 centos7-2

节点1：192.168.177.132 centos7-3

节点2：192.168.177.133 centos7-4

主服务器 从服务器

• 安装带有nginx rpm软件包的源

  # rpm -ivh http://nginx.org/packages/centos/7/noarch/RPMS/\
  nginx-release-centos-7-0.el7.ngx.noarch.rpm

• 使用centos 默认仓库完成下面的安装 主

  #  yum install -y keepalived nginx
  #  vim /etc/keepalived/keepalived.conf  //从上修改三个参数

  ! Configuration File for keepalived

vrrp_script nginx {
script "/opt/shell/nginx.sh"
interval 2
} //添加
global_defs {
route_id NGINX_HA //修改
}
vrrp_instance VI_1 {
state MASTER
interface ens33
virtual_router_id 51
priority 100
advert_int 1
authentication {
auth_type PASS
auth_pass 1111
}
track_script {
nginx
}
virtual_ipaddress {
192.168.177.188 //虚拟IP
192.168.200.188
}
}

#  mkdir /opt/shell  
#  cd /opt/shell
#  vim /opt/shell/nginx.sh

#!/bin/bash
k=`ps -ef | grep keepalived | grep -v grep | wc -l`
if [ $k -gt 0 ];then
    /bin/systemctl start nginx.service
else
/bin/systemctl stop nginx.service
fi

# chmod +x /opt/shell/nginx.sh   //变成可执行的脚本

从服务器与主服务器配置一样（内容要修改）

# yum install -y keepalived nginx
# vim /etc/keepalived/keepalived.conf  //从上修改三个参数

! Configuration File for keepalived

vrrp_script nginx {
    script "/opt/shell/nginx.sh"
    interval 2
}    //添加
global_defs {
    route_id NGINX_HB   //修改
}
vrrp_instance VI_1 {
    state MASTER
    interface ens33
    virtual_router_id 52
    priority 90
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 1111
}
track_script {
    nginx
}
virtual_ipaddress {
    192.168.177.188      //虚拟IP
    192.168.200.188
    }
}
# mkdir /opt/shell
# cd /opt/shell
# vim /opt/shell/nginx.sh

#!/bin/bash
k=`ps -ef | grep keepalived | grep -v grep | wc -l`
if [ $k -gt 0 ];then
    /bin/systemctl start nginx.service
else
/bin/systemctl stop nginx.service
fi

# chmod +x /opt/shell/nginx.sh   //变成可执行的脚本

配置nginx前端调度功能(主，从)

# vim /etc/nginx/nginx.conf //在#gzip on;下面添加：

 upstream tomcat_pool {
                server 192.168.177.132:8080;
                server 192.168.177.133:8080;
                ip_hash;           #会话稳固功能，否则无法通过vip地址登陆
        }
        server {
                listen 80;
                server_name 192.168.177.188; #虚拟出的IP  
                location / {
                        proxy_pass http://tomcat_pool;
                        proxy_set_header X-Real-IP $remote_addr;
                }
    }

# nginx -t -c /etc/nginx/nginx.conf  //测试配置文件语法
# systemctl start keepalived.service    //nginx启动会等待一会

两台节点都要做tomcat

# tar xf apache-tomcat-8.5.23.tar.gz
# tar xf jdk-8u144-linux-x64.tar.gz

# mv apache-tomcat-8.5.23/ /usr/local/tomcat8
# mv jdk1.8.0_144/ /usr/local/java

# vim /etc/profile

export JAVA_HOME=/usr/local/java
export JRE_HOME=/usr/local/java/jre
export PATH=$PATH:/usr/local/java/bin
export CLASSPATH=./:/usr/local/java/lib:/usr/local/java/jre/lib

# source /etc/profile

# java -version

# ln -s /usr/local/tomcat8/bin/startup.sh /usr/bin/tomcatup
# ln -s /usr/local/tomcat8/bin/shutdown.sh /usr/bin/tomcatdown

# tomcatup     //开启服务
# netstat -anpt | grep 8080

# vim /usr/local/tomcat8/webapps/ROOT/index.jsp //修改默认网页内容

<h1>Server 132!!!</h1>
http://192.168.177.132:8080/         //测试默认测试页是否正常显示（节点）
http://192.168.177.133:8080/
http://192.168.177.188/  输入调度器地址，也就是虚拟地址，测试两台节点的调度情况。

将192.168.177.132宕机（tomcatdown)，看看http://192.168.177.188/会不会显示<h1>Server 133!!!</h1>

# vim server.xml //跳到行尾，在Host name下新增 148行

<Context path="" docBase="SLSaleSystem" reloadable="true" debug="0"></Context>
日志调试信息debug为0表示信息越少，docBase指定访问目录

Mysql安装（主，从）

# yum install -y mariadb-server mariadb
# systemctl start mariadb.service
# systemctl enable mariadb.service    //开机自启动
# netstat -anpt | grep 3306
# mysql_secure_installation    //常规安全设置
# mysql -u root -p

导入数据库

先挂载
# mysql -u root -p  < slsaledb-2014-4-10.sql
# mysql -u root -p
  show databases;
  GRANT all ON slsaledb.* TO 'root'@'%' IDENTIFIED BY 'abc123';
  flush privileges;

以下在两台tomcat节点做

# tar xf SLSaleSystem.tar.gz -C /usr/local/tomcat8/webapps/
# cd /usr/local/tomcat8/webapps/SLSaleSystem/WEB-INF/classes
# vim jdbc.properties //修改数据库IP地址是VRRP的虚拟IP，以及授权的用户名root和密码abc123。

网站测试

http://192.168.177.132:8080/   //默认的用户名admin 密码：123456
http://192.168.177.133:8080/

http://192.168.177.188  //输入虚拟地址测试登录，并且关闭主再测试登录

redis集群（主，从）

# yum install -y epel-release
# yum install redis -y
# vim /etc/redis.conf
  bind 0.0.0.0
# systemctl start redis.service
# netstat -anpt | grep 6379
# redis-cli -h 192.168.177.145 -p 6379 //测试连接
  192.168.177.145:6379> set name test  //设置name 值是test
  192.168.177.145:6379> get name //获取name值

从服务器上多如下一行配置

 266行 slaveof 192.168.177.145  6379 //主服务器的IP不是虚拟IP

# redis-cli -h 192.168.177.135 -p 6379 //登录从，获取值，成功说明主从同步成功
  192.168.177.135:6379> get name
"test"

配置商城项目中连接redis的参数

38         <!--redis 配置 开始-->

  47                 <constructor-arg value="192.168.177.188"/>
  48                 <constructor-arg value="6379"/>

测试缓存效果

# redis-cli -h 192.168.177.188 -p 6379
  192.168.177.188:6379> info
 keyspace_hits:1  或者 keyspace_misses:2//关注这个值，命中数和未命中数

登录商城，然后反复点击需要数据库参与的操作页面，再回来检查keyspace_hits或者keyspace_misses: 值变化。

配置redis集群主从切换，只在主服务器上操作

# redis-cli -h  192.168.177.145 info Replication       //获取当前服务器的角色
# vim /etc/redis-sentinel.conf
17行 protected-mode no17G
68行 sentinel monitor mymaster 192.168.177.145 6379 1 //1表示1台从 注意：修改
98行 sentinel down-after-milliseconds mymaster 3000 //故障切换时间单位是毫秒
# service redis-sentinel start //启动集群
# netstat -anpt | grep 26379
# redis-cli -h 192.168.177.145 -p 26379 info Sentinel //查看集群信息

验证主从切换

在主上
# systemctl stop redis.service
# redis-cli -h 192.168.177.145 -p 26379 info Sentinel //发现主变成了135

验证数据同步情况

# redis-cli -h 192.168.177.135 -p 6379 
  192.168.177.135:6379> set name2 test2
OK
  192.168.177.135:6379> get name2
"test2"
# service redis start //把主启动
# redis-cli -h 192.168.177.145 -p 6379
  192.168.175.128:6379> get name2
  "test2"

Mysql主从

mysql主服务器配置

# vim /etc/my.cnf //[mysqld]下

binlog-ignore-db=mysql,information_schema
character_set_server=utf8
log_bin=mysql_bin
server_id=1
log_slave_updates=true
sync_binlog=1

# systemctl restart mariadb
# netstat -anpt | grep 3306
# mysql -u root -p
  show master status; //记录日志文件名称和 位置值
  grant replication slave on *.* to 'rep'@'192.168.177.%' identified by '123456';
  flush privileges;

mysql从服务器配置

# vim /etc/my.cnf //[mysqld]下

binlog-ignore-db=mysql,information_schema
character_set_server=utf8
log_bin=mysql_bin
server_id=2
log_slave_updates=true
sync_binlog=1

# systemctl restart mariadb
# netstat -anpt | grep 3306
# mysql -u root -p

  change master to master_host='192.168.177.145',master_user='rep',master_password='123456',master_log_file='mysql_bin.000001',master_log_pos=245;

  start slave;

  show slave status\G;

    Slave_IO_Running: Yes
    Slave_SQL_Running: Yes