跨云服务器厂商--低成本--实现--伪·灾备

1.背景说明

（网上找了点资料，下面是自己的一些浅见，有不当之处，恳请大家指出。技术交流，只为进步。）

主要场景假设ip如下

1台阿里云服务器 10.0.0.2

1台华为云服务器 10.0.0.3

要求：部署服务，实现灾备双活。（重点不是云服务器做灾备。重点是客户想要在这个环境上模拟灾备双活实现）

正常解决方式：

数据库双主同步+keepalived
前后端连接数据库连接的是 keepalived提供的ip。

这总方式在内网很容易实现，多准备1个ip地址即可。

但是在云环境上，要么买云厂商的服务，要么自己做一个伪·灾备。

思路如下：

前端nginx权重轮询反向代理两个后端。
两个后端分别连接相应的数据库。
数据库之间做双向同步。
文件存储类文件通过NFS进行同步。

特殊时期特殊环境的特殊手段。如果有其他选择，建议大家不要选择这个方式。

2.遇到的问题

（1）数据库双向同步（云服务器安全组 ——配置信任ip，两个ip允许所有端口互相访问）

10.0.0.2操作如下

依赖
yum install -y perl perl-devel
yum -y install autoconf libaio
解压安装
tar xf mysql-5.6.37-linux-glibc2.12-x86_64.tar.gz
mv mysql-5.6.37-linux-glibc2.12-x86_64 /usr/local/mysql-5.6
cd /usr/local/mysql-5.6/
groupadd mysql
useradd -r -g mysql mysql
chown -R mysql.mysql /usr/local/mysql-5.6/
初始化
/usr/local/mysql-5.6/scripts/mysql_install_db --user=mysql --basedir=/usr/local/mysql-5.6/ --datadir=/usr/local/mysql-5.6/data

配置文件
vim /etc/my.cnf

[mysqld]
    basedir=/usr/local/mysql-5.6
    datadir=/usr/local/mysql-5.6/data
    port=3306
    server-id=1
    log-bin=/usr/local/mysql-5.6/mysql-bin
    sync_binlog = 1
    binlog_checksum = none
    binlog_format = mixed
    auto-increment-increment = 2
    auto-increment-offset = 1
    slave-skip-errors = all    
    innodb_flush_log_at_trx_commit=2
    sync_binlog=1000
    transaction_isolation=READ-COMMITTED
    expire_logs_days=7
    user=mysql
    socket = /tmp/mysql.sock
    character-set-server=utf8
    skip-name-resolve
    lower_case_table_names=1
    max_connections=5000  
    max_connect_errors=1000
    max_allowed_packet=500M
    wait_timeout=1814400   
    slow_query_log=on   
    long_query_time=2   
    slow_query_log_file=/usr/local/mysql-5.6/slow.log
    symbolic-links=0
[mysqld_safe]
    log-err=/usr/local/mysql-5.6/data/error.log
    pid-file=/usr/local/mysql-5.6/data/mysql.pid
    sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES

开机启动
cp /usr/local/mysql-5.6/support-files/mysql.server /etc/init.d/mysql
chkconfig mysql on
环境变量
[ `egrep mysql /etc/profile|wc -l` -eq 0 ] && echo 'export PATH=$PATH:/usr/local/mysql-5.6/bin' >> /etc/profile ;source /etc/profile
启动
service mysql start
同步账户，锁表只读，查看log-bin.
mysql
grant replication slave,replication client,select on *.* to slave@'10.0.0.3' identified by 'Slave@123';
flush privileges;
flush tables with read lock;
flush logs;
show master status;

两台机器按照上述步骤完成操作。注意以下配置区别。

第二个 10.0.0.3 配置

[mysqld]
    basedir=/usr/local/mysql-5.6
    datadir=/usr/local/mysql-5.6/data
    port=3306
    server-id=2
    log-bin=/usr/local/mysql-5.6/mysql-bin
    sync_binlog = 1
    binlog_checksum = none
    binlog_format = mixed
    auto-increment-increment = 2
    auto-increment-offset = 2
    slave-skip-errors = all    
    innodb_flush_log_at_trx_commit=2
    sync_binlog=1000
    transaction_isolation=READ-COMMITTED
    expire_logs_days=7
    user=mysql
    socket = /tmp/mysql.sock
    character-set-server=utf8
    skip-name-resolve
    lower_case_table_names=1
    max_connections=5000  
    max_connect_errors=1000
    max_allowed_packet=500M
    wait_timeout=1814400   
    slow_query_log=on   
    long_query_time=2   
    slow_query_log_file=/usr/local/mysql-5.6/slow.log
    symbolic-links=0
[mysqld_safe]
    log-err=/usr/local/mysql-5.6/data/error.log
    pid-file=/usr/local/mysql-5.6/data/mysql.pid
    sql_mode=NO_ENGINE_SUBSTITUTION,STRICT_TRANS_TABLES

grant replication slave,replication client,select on *.* to slave@'10.0.0.2' identified by 'Slave@123';
flush privileges;
flush tables with read lock;
flush logs;
show master status;

单向同步

10.0.0.3同步10.0.0.2的数据

先查看10.0.0.2的logbin

再10.0.0.3 进入mysql中操作

unlock tables;
stop slave;
change master to master_host='10.0.0.2',master_user='slave',master_password='Slave@123',master_log_file='mysql-bin.000004',master_log_pos=120;
start slave;
show slave status\G;

10.0.0.2同步10.0.0.3的数据

再10.0.0.3上刷新log-bin

然会回到 10.0.0.2上如下操作

unlock tables;
stop slave;
change master to master_host='10.0.0.3',master_user='slave',master_password='Slave@123',master_log_file='mysql-bin.000004',master_log_pos=120;
start slave;
show slave status\G;

最后去验证下。其中一台创建表，第二台查看表.建库建表语句如下，可以去尝试测试。

create database if not exists ywjiasql default character set utf8;
grant all privileges on ywjiasql.* to "ywjiasql"@"%" identified by "ywjiasql";
flush privileges;

CREATE TABLE IF NOT EXISTS `runoob_tbl`(
   `runoob_id` INT UNSIGNED AUTO_INCREMENT,
   `runoob_title` VARCHAR(100) NOT NULL,
   `runoob_author` VARCHAR(40) NOT NULL,
   `submission_date` DATE,
   PRIMARY KEY ( `runoob_id` ))ENGINE=InnoDB DEFAULT CHARSET=utf8;
desc runoob_tbl;

（2）nginx权重轮询配置

具体大家可以去看一下。这里选择一种展示

upstream soft_up{
    ip_hash; 
    server 10.0.0.2:19998 weight=1;
    server 10.0.0.3:19998  weight=1;
}

（3）文件同步NFS

server端部署如下:

yum -y install nfs-utils rpcbind
vim /etc/sysconfig/nfs
cat /etc/sysconfig/nfs |grep -v ^# |grep -v ^$

修改了一些端口，添加信任，端口该不该问题都不大

LOCKD_TCPPORT=32803
LOCKD_UDPPORT=32769
RPCNFSDARGS=""
RPCMOUNTDOPTS=""
MOUNTD_PORT=892
STATDARG=""
STATD_PORT=1001
STATD_OUTGOING_PORT=2001
SMNOTIFYARGS=""
RPCIDMAPDARGS=""
RPCGSSDARGS=""
GSS_USE_PROXY="yes"
BLKMAPDARGS=""

vim /etc/exports

/usr/local/eli_file 10.0.0.3/32(insecure,rw,sync,no_subtree_check,no_root_squash)

exportfs -rv
systemctl start rpcbind
systemctl start nfs
systemctl enable rpcbind
systemctl enable nfs

客户端挂载操作如下：
yum -y install nfs-utils rpcbind
showmount -e  10.0.0.2
mount -t nfs -o nolock,nfsvers=3,vers=3,soft,intr,bg,rw,rsize=32768,wsize=32768 10.0.0.2:/usr/local/soft_file /usr/local/soft_file

（4）服务启动。数据库停止，服务停止。数据库恢复，服务恢复。

写一个死循环脚本持续运行。简简单单实现一下。稍后会更新一版通用化点的这个脚本。

#!/bin/bash
function print_info_log(){
    echo -e "\e[1;35m INFO:$(date "+%Y-%m-%d %H:%M:%S") \e[0m $1." 
}
function jar_pid_cmd(){
    jar_pid=`ps -ef|grep eli-main-1.0|grep -v grep|awk '{print $2}'`
    kill -9 $jar_pid
}

function jar_start_cmd(){
start_jar="nohup java -jar /usr/local/springboot_eli/eli-main-1.0-SNAPSHOT.jar -Xms1024m -Xms3072m -XX:PermSize=512m -XX:MaxPermSize=1024m -XX:MaxNewSize=1024m  --spring.profiles.active=eli  --spring.config.additional-location=/usr/local/springboot_eli/  >>/usr/local/springboot_eli/eli.out 2 >&1 & "
echo "$start_jar" | bash  &>/dev/null
}


#死循环
aaaaa=1
while [  $aaaaa -eq 1 ]
do
    mysql_status=`netstat -tnlp|grep 3306|wc -l`
    jar_status=`ps -ef|grep eli-main-1.0|grep -v grep  |wc -l`
    
#检测数据库关闭，获取jar包运行状态，运行的进行关闭
if [ $mysql_status -eq 0 ];then
    [ $jar_status -ne 0 ] && jar_pid_cmd || print_info_log "已关闭" 
fi

#检测数据库开启，获取jar包运行状态，未运行的进行启动
if [ $mysql_status -eq 1 ];then
    [ $jar_status -eq 0 ] &&  jar_start_cmd  ||print_info_log "已启动" 
fi

    sleep 5
    #结束死循环
    #aaaaa=2
done

3.常规实现--数据库双主同步+keepalived

服务器ip：10.0.0.2
yum -y install gcc gcc-c++ autoconf automake make
yum -y install zlib zlib-devel openssl openssl-devel pcre pcre-devel
yum -y install keepalived
配置安装文件 文件需要适配修改
vim  /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
   router_id MASTER-1
}
vrrp_script check_mysql_alive {
    script "/etc/keepalived/check_mysql_alive.sh"
    interval 2
    weight -5
}
vrrp_instance VI_1 {
    state BACKUP
    interface eth0
    virtual_router_id 51
    mcast_src_ip 10.0.0.2
    priority 100
    nopreempt
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 123456
    }
    track_script {
    check_mysql_alive
    }
    virtual_ipaddress {
        10.0.0.3
        }
}
 
检测脚本
cd /etc/keepalived/
vim check_mysql_alive.sh
内容如下：
#!/bin/bash
counter=$(ss -tnlp|grep 3306 |wc -l)
if [ $counter -eq 0 ];then
    systemctl stop keepalived
fi

chmod +x check_mysql_alive.sh 
启动（保持mysql服务运行下，否则脚本检测到服务未运行会进行关闭keepalived）
systemctl enable keepalived.service
systemctl start keepalived.service
systemctl status keepalived.service

10.0.0.3 操作
yum -y install gcc gcc-c++ autoconf automake make
yum -y install zlib zlib-devel openssl openssl-devel pcre pcre-devel
yum -y install keepalived

配置安装文件 文件需要适配修改，参照图解说明
vim  /etc/keepalived/keepalived.conf
! Configuration File for keepalived
global_defs {
   router_id MASTER-2
}
vrrp_script check_mysql_alive {
    script "/etc/keepalived/check_mysql_alive.sh"
    interval 2
    weight -5
}
vrrp_instance VI_1 {
    state BACKUP
    interface eth0
    virtual_router_id 51
    mcast_src_ip 10.0.0.3
    priority 90
    nopreempt
    advert_int 1
    authentication {
        auth_type PASS
        auth_pass 123456
    }
    track_script {
    check_mysql_alive
    }
    virtual_ipaddress {
        10.0.0.4
        }
}

检测脚本
cd /etc/keepalived/
vim check_mysql_alive.sh
内容如下：
#!/bin/bash
counter=$(ss -tnlp|grep 3306 |wc -l)
if [ $counter -eq 0 ];then
    systemctl stop keepalived
fi
执行权限
chmod +x check_mysql_alive.sh 

启动（保持mysql服务运行下，否则脚本检测到服务未运行会进行关闭keepalived）
systemctl enable keepalived.service
systemctl start keepalived.service
systemctl status keepalived.service

协助理解下

posted on 2022-01-14 11:50 执手仗剑天涯阅读(60) 评论(0) 编辑收藏举报

刷新页面返回顶部

导航

跨云服务器厂商--低成本--实现--伪·灾备

跨云服务器厂商--低成本--实现--伪·灾备

1.背景说明

2.遇到的问题

（1）数据库双向同步（云服务器安全组 ——配置信任ip，两个ip允许所有端口互相访问）

（2）nginx权重轮询配置

（3）文件同步NFS

（4）服务启动。数据库停止，服务停止。数据库恢复，服务恢复。

3.常规实现--数据库双主同步+keepalived