周一（2014-11-17）有个项目进行变更，而且是重大变更，DB测操作从早上持续到下午17点，QA同事到晚上10点测试后发现，slave上的数据与master上不一致。
这个问题应该在情理之中但又在意料之外，其实DBA在下午DB变更时便遇到slave卡住：
Could not execute Delete_rows event on table codcpc_1004pc_0x3EC.mail; Can't find record in 'mail
', Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; the event's master log binlog.010464, end_log_pos 82113, Error_code: 1032
141117 7:54:31 [Warning] Slave: Can't find record in 'mail' Error_code: 1032
mysqlconn="mysql -uADMIN -p4word"
#while [ "1" = "1" ]
for i in `seq 1 30`
do
status=`$mysqlconn -e "show slave status\G;" | grep -i "Last_SQL_Error" | grep -i "mail"`
if [ -n "$status" ];then
$mysqlconn -vvve "stop slave;set global sql_slave_skip_counter=1;start slave;"
sleep 1
else
$mysqlconn -vvve "show slave status\G"
echo "error is OK!"
exit 0
fi
done
而binlog_format是row模式的时候，skip是以事务级别来的。