# vim /data/scripts/delete_mysql_binlog.sh
#!/bin/bash
#=======================================================================================
#   用于删除 MySQL Master 端已经同步完的 binlog【需在 Master 端运行】，以减少磁盘空间
#   每天凌晨 5:30 分运行一次
#
#   注：需在 Slave 端添加允许 Master 端访问的帐号【帐号：check_binlog，密码：binlog_2356】
#          运行于 MySQL Master 端【目前只用于一主一从的同步模式，对于多从的情况暂时未考虑】
#=======================================================================================
PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/sbin:/usr/local/bin

## Slave端连接信息
SLAVE_ADDR="XXX.XXX.XXX.XXX"
SLAVE_USER="check_binlog"
SLAVE_PWD="binlog_2356"

LOGFILE="/data/logs/db_sync_info.log"
PINGFILE="/tmp/mysqlping.log"

## MySQL状态信息查看命令
SQLCMD="show slave status"

#=======================================================================================
## 检查MySQL是否已经运行
retval=`ps aux | grep mysql[d] | wc -l`
if [ ${retval} -eq 0 ]; then
       echo The MySQL is not running at: `date +%F" "%H-%M-%S` >> ${LOGFILE}
       exit 1
fi

## 测试Slave端的连通性
nohup mysqladmin -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} ping > ${PINGFILE}
retval=`grep "^error" ${PINGFILE}`
rm -f ${PINGFILE}
if [ "${retval}X" != "X" ]; then
       echo The MySQL Slave can not be connected at: `date +%F" "%H-%M-%S` >> ${LOGFILE}
       exit 1
fi

## 检查是否合法的Slave
MASTER_ADDR=`mysql -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} -e "${SQLCMD}\G;" | awk '$1=="Master_Host:" {print $2}'`
LOCAL_ADDR=`/sbin/ifconfig eth1 | awk -F ':' '/inet addr/{print $2}' | sed 's/[a-zA-Z ]//g'`
if [ "${MASTER_ADDR}" != "${LOCAL_ADDR}" ]; then
       echo The MySQL Slave is not lawful at: `date +%F" "%H-%M-%S` >> ${LOGFILE}
       exit 1
fi

## 获得Slave端信息，以此来确定是否处于正常同步的情况
IO_STATUS=`mysql -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} -e "${SQLCMD}\G;" | awk '$1=="Slave_IO_Running:" {print $2}'`
SQL_STATUS=`mysql -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} -e "${SQLCMD}\G;" | awk '$1=="Slave_SQL_Running:" {print $2}'`
if [[ "${IO_STATUS}" != "Yes" || "${SQL_STATUS}" != "Yes" ]]; then
       echo The MySQL Replication is not synchronous at: `date +%F" "%H-%M-%S` >> ${LOGFILE}
       exit 1
fi

## 再做一次判断，以保证数据同步绝对正常【创建测试数据】
mysql -uroot -e "create database if not exists mytestdb;"
sleep 3

retval=`mysql -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} -e "show databases;" | grep mytestdb`
mysql -uroot -e "drop database if exists mytestdb;"
if [ "${retval}X" = "X" ]; then
       echo The MySQL Replication is not synchronous at: `date +%F" "%H-%M-%S` >> ${LOGFILE}
       exit 1
fi

## 在已经同步的情况，还需要判断当前同步的binlog，以此来确定哪些已经是过期的binlog
SLAVE_BINLOG1=`mysql -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} -e "${SQLCMD}\G;" | awk '$1=="Master_Log_File:" {print $2}'`
SLAVE_BINLOG2=`mysql -h${SLAVE_ADDR} -u${SLAVE_USER} -p${SLAVE_PWD} -e "${SQLCMD}\G;" | awk '$1=="Relay_Master_Log_File:" {print $2}'`

## 获得Master端，当前的binlog文件以及binlog路径
MASTER_BINLOG=`mysql -uroot -e "show master status;" | grep -v '^+' | tail -1 | awk '{print $1}'`

## 主从端已经同步到相同的binlog
if [[ "${SLAVE_BINLOG1}" = "${SLAVE_BINLOG2}" && "${SLAVE_BINLOG1}" = "${MASTER_BINLOG}" ]]; then
       CURR_BINLOG="${MASTER_BINLOG}"
## 主从端已经同步，但从端的binlog还没有追赶到主端最新的binlog
elif [[ "${SLAVE_BINLOG1}" = "${SLAVE_BINLOG2}" && "${SLAVE_BINLOG1}" != "${MASTER_BINLOG}" ]]; then
       CURR_BINLOG="${SLAVE_BINLOG1}"
## 主从端已经同步，主从端的binlog一致，但relaylog还不一致
elif [[ "${SLAVE_BINLOG1}" != "${SLAVE_BINLOG2}" && "${SLAVE_BINLOG1}" = "${MASTER_BINLOG}" ]]; then
       CURR_BINLOG="${SLAVE_BINLOG2}"
else
       echo Has noknown error at：`date +%F" "%H-%M-%S` >> ${LOGFILE}
       exit 1
fi
mysql -uroot -e "purge binary logs to '${CURR_BINLOG}';"
if [ $? -eq 0 ]; then
       echo Clear MySQL binlog is ok at: `date +%F" "%H-%M-%S` >> ${LOGFILE}
fi

# crontab -e
30 05 * * * /data/scripts/delete_mysql_binlog.sh >/dev/null 2>&1
转自：http://sofar.blog.51cto.com/353572/1260086