使用Docker自动化部署MySQL集群-51CTO.COM

本文主要是简化mysql主从搭建，同时给需要学习mysql主从的同学一个现成的环境。

1. 编写master.sh脚本

此脚本主要是根据环境变量，创建出用于主从同步的用户。关于为什么要创建出master,slave脚本，是为了区分master和slave环境，因为只要把sql和sh文件放到master目录下面，master在第一次初始化的时候，会自动执行里面的文件内容

mkdir -p init init/master init/slave
cat > init/master/master.sh <<EOF
#!/bin/bash
#定义用于同步的用户名
MASTER_SYNC_USER=\${MASTER_SYNC_USER:-sync_admin}
#定义用于同步的用户密码
MASTER_SYNC_PASSWORD=\${MASTER_SYNC_PASSWORD:-123456}
#定义用于登录mysql的用户名
ADMIN_USER=\${ADMIN_USER:-root}
#定义用于登录mysql的用户密码
ADMIN_PASSWORD=\${ADMIN_PASSWORD:-123456}
#定义运行登录的host地址
ALLOW_HOST=\${ALLOW_HOST:-%}
#定义创建账号的sql语句
CREATE_USER_SQL="CREATE USER '\$MASTER_SYNC_USER'@'\$ALLOW_HOST' IDENTIFIED BY '\$MASTER_SYNC_PASSWORD';"
#定义赋予同步账号权限的sql,这里设置两个权限，REPLICATION SLAVE，属于从节点副本的权限，REPLICATION CLIENT是副本客户端的权限，可以执行show master status语句
GRANT_PRIVILEGES_SQL="GRANT SELECT,REPLICATION SLAVE,REPLICATION CLIENT ON *.* TO '\$MASTER_SYNC_USER'@'\$ALLOW_HOST';"
#定义刷新权限的sql
FLUSH_PRIVILEGES_SQL="FLUSH PRIVILEGES;"
#执行sql
mysql -u"\$ADMIN_USER" -p"\$ADMIN_PASSWORD" -e "\$CREATE_USER_SQL \$GRANT_PRIVILEGES_SQL \$FLUSH_PRIVILEGES_SQL"
EOF

2. 编写slave.sh脚本

slave主要是去master查询最新的pos位置和binlog文件名称，然后创建同步需要的相关信息，然后执行start slave;

cat >init/slave/slave.sh<<EOF
#定义连接master进行同步的账号
SLAVE_SYNC_USER="\${SLAVE_SYNC_USER:-sync_admin}"
#定义连接master进行同步的账号密码
SLAVE_SYNC_PASSWORD="\${SLAVE_SYNC_PASSWORD:-123456}"
#定义slave数据库账号
ADMIN_USER="\${ADMIN_USER:-root}"
#定义slave数据库密码
ADMIN_PASSWORD="\${ADMIN_PASSWORD:-123456}"
#定义连接master数据库host地址
MASTER_HOST="\${MASTER_HOST:-%}"
#连接master数据库，查询二进制数据，并解析出logfile和pos，这里同步用户要开启 REPLICATION CLIENT权限，才能使用SHOW MASTER STATUS;
RESULT=\`mysql -u"\$SLAVE_SYNC_USER" -h\$MASTER_HOST -p"\$SLAVE_SYNC_PASSWORD" -e "SHOW MASTER STATUS;" | grep -v grep |tail -n +2| awk '{print \$1,\$2}'\`
#解析出logfile
LOG_FILE_NAME=\`echo \$RESULT | grep -v grep | awk '{print \$1}'\`
#解析出pos
LOG_FILE_POS=\`echo \$RESULT | grep -v grep | awk '{print \$2}'\`
#设置连接master的同步相关信息
SYNC_SQL="change master to master_host='\$MASTER_HOST',master_user='\$SLAVE_SYNC_USER',master_password='\$SLAVE_SYNC_PASSWORD',master_log_file='\$LOG_FILE_NAME',master_log_pos=\$LOG_FILE_POS,get_master_public_key=1;"
#开启同步
START_SYNC_SQL="start slave;"
#查看同步状态
STATUS_SQL="show slave status\G;"
mysql -u"\$ADMIN_USER" -p"\$ADMIN_PASSWORD" -e "\$SYNC_SQL \$START_SYNC_SQL \$STATUS_SQL"
EOF

3. 编写docker-compose.yml脚本

version: '3'
services:
  master:
    image: mysql:8
    container_name: mysql-master
    ports:
    - '3306:3306'
    restart: always
    hostname: mysql-master
    environment:
      MYSQL_ROOT_PASSWORD: "123456"
      MASTER_SYNC_USER: "sync"
      MASTER_SYNC_PASSWORD: "123456"
      ADMIN_USER: "root"
      ADMIN_PASSWORD: "123456"
      TZ: "Asia/Shanghai"
    deploy:
      resources:
        limits:
          memory: 512M
          cpus: 50m
    healthcheck:
      test: ["CMD","mysqladmin","-uroot","-p$${MYSQL_ROOT_PASSWORD}","ping","-h","localhost"]
      timeout: 2s
      interval: 10s
      retries: 5
      start_period: 5s
    logging:
      options:
        max-file: '1'
        max-size: '128k'
    command:
    -  "--server-id=1"
    -  "--character-set-server=utf8mb4"
    -  "--collation-server=utf8mb4_unicode_ci"
    -  "--log-bin=mysql-bin"
    -  "--sync_binlog=1"
    -  "--binlog-ignore-db=mysql"
    -  "--binlog-ignore-db=sys"
    -  "--binlog-ignore-db=performance_schema"
    -  "--binlog-ignore-db=information_schema"
    -  "--sql_mode=NO_AUTO_VALUE_ON_ZERO,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION,PIPES_AS_CONCAT,ANSI_QUOTES"
    volumes:
    - ./init/master:/docker-entrypoint-initdb.d
    - ./data/master:/var/lib/mysql
  slave:
    image: mysql:8
    container_name: mysql-slave
    ports:
    - '3307:3306'
    restart: always
    hostname: mysql-slave
    environment:
      MYSQL_ROOT_PASSWORD: "123456"
      SLAVE_SYNC_USER: "sync"
      SLAVE_SYNC_PASSWORD: "123456"
      ADMIN_USER: "root"
      ADMIN_PASSWORD: "123456"
      MASTER_HOST: "mysql-master"
      TZ: "Asia/Shanghai"
    healthcheck: 
      test: ["CMD","mysqladmin","-uroot","-p$${MYSQL_ROOT_PASSWORD}","ping","-h","localhost"]
      timeout: 2s
      interval: 10s
      retries: 5
      start_period: 5s
    deploy:
      resources:
        limits:
          memory: 512M
          cpus: 50m
    logging:
      options:
        max-file: '1'
        max-size: '128k'
    depends_on:
      master:
        condition: service_healthy
    command:
    -  "--server-id=2"
    -  "--character-set-server=utf8mb4"
    -  "--collation-server=utf8mb4_unicode_ci"
    -  "--sql_mode=NO_AUTO_VALUE_ON_ZERO,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,NO_ZERO_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_ENGINE_SUBSTITUTION,PIPES_AS_CONCAT,ANSI_QUOTES"
    volumes:
    - ./init/slave:/docker-entrypoint-initdb.d
    - ./data/slave:/var/lib/mysql

启动脚本

docker-compose up -d

后期维护过程中可能用到的脚本

STOP SLAVE;
RESET SLAVE;
CHANGE MASTER TO master_log_pos=0;
START SLAVE;
SHOW SLAVE STATUS;

写在后面，基于docker-compose搭建mysql主从已经结束，这里主要演示了如何一键搭建mysql主从，同时测试在各种环境下，还能保证主从同步。但是单主从同步肯定也无法满足企业的需求，有兴趣的同学可以研究一些主主双写同步，然后通过nginx搭建主备模式，达到高可用。

这里提供一个判断的方法，编写#boot.sh脚本文件，这时候需要重写dockerfile文件，使用myslq debian的镜像文件。并通过apt添加netcat工具。具体不再详述了，以下为判断脚本。

#!/bin/bash
wait_for() {
    echo Waiting for $1 to listen on $2...
    sleep 1
    while ! nc -vz $1 $2
    do
       echo waiting...;
       sleep 1s;
    done
}
if [ -e "/slave" ]; then
   echo "slave..."
   cp /slave.cnf /etc/mysql/conf.d/
else
   echo "master..."
   cp /master.cnf /etc/mysql/conf.d/
fi


echo "master config ... "
/master.sh
echo "complete. "
if [ -n "$MASTER_HOST" ]; then
   echo "replicaof $MASTER_HOST"
   wait_for $MASTER_HOST $MASTER_PORT
   /slave.sh
        echo "Slave config complete."
fi