Keepalived clustering deployment, nginx load balancing, high-availability clustering, single point of failure solution-biu~biu~@*·~!

Environmental preparation

C entos7 four, turn off the firewall, turn off selinux

1. No-1 Prevent single point of failure and nginx load balancing
   192.168.112.153
   192.168.112.154
2. No-2 httpd simulated web page
   192.168.112.160
   192.168.112.161

• Principle rendering
• 
  Description: The two highly available devices should be divided into master and backup, and VIP defined. After nginx load balancing is configured, keepalived configures the RS end of the local machine and the ip address of the other keepalived. When both are alive, regardless of access to VIP or Both master and backup can be accessed. When a single point of failure occurs (when one end of the master or backup fails), the VIP is elegant and continuous flow is realized!

No-1 installation service

yum -y install gcc gcc-c++ pcre-devel zlib-devel keepalived

• Upload nginx package

• 

• You can also download the latest nginx package from the official website

• nginx download address

Install and configure nginx to achieve load balancing

tar xf nginx-1.18.0.tar.gz  解压
cd nginx-1.18.0  进入目录
./configure  检测编译环境
make && make install  编译并安装

Modify nginx configuration file

vim /usr/local/nginx/conf/nginx.conf

• Lines 34~37 define the back-end server with load [outside the server]
  
• Add proxy [in server] on line 49
  

Arrangement No-2

• Install httpd
  yum -y install httpd
• Write test page

vim /var/www/html/index.html

Content web1, the other one is different, web2

• Start httpd, verify the port
  

No-1 start nginx, verify the load

/usr/local/nginx/sbin/nginx

No-1 placement keepalived

vim /etc/keepalived/keepalived.conf

• Delete the extra line
  
• 8 lines define master
• Line 9 modify the network card to ens33
• 11 Behavior weight [backup power is less important than master]
• The 18th line defines VIP. The ip master and backup that are not occupied in this network segment
  
  are required to have different configurations, and the others are the same!!!
  
• Fill in the VIP on line 22 and modify the port to 80
• 24 behavioral algorithm
• Comment out the persistence timeout on line 26 [symbol is! 】
• The ip and port of the RS end on line 29
• Modify line 31 to check for tcp
• 32 lines delete the redundant ones, add the connection port in the next line
  

Start the service and verify high availability

systemctl start keepalived

• View VIP

ip a show ens33

• Verification rendering, VIP of master
  
• All live, master, bakcup and VIP are all accessible to achieve load
  
  
  

Simulate a single point of failure

• Stop the master side, check VIP elegant
• systemctl stop keepalived
  
• Backup check
  
• Webpage test renderings
  
  
• Although the master side has been deactivated, the load can still be achieved, and the VIP is floating

The principle of keepalived please visit: keepalived principle and module

For detailed explanation of keepalived+LVS deployment and principles, please visit: keepalived+LVS clustered deployment