1、首先进行数据训的XltraBackup备份，有备无患，切记切记！
2、
mysql -uroot -pD********
-- 导出csv文件
use dsideal_db;
MariaDB [dsideal_db]> SELECT * from t_resource_info INTO OUTFILE "/usr/local/mysql/t_resource_info.txt" FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"' LINES TERMINATED BY '\n';
Query OK,1582463 rows affected (29.97sec)
3、切分csv文件，这样分批导入更快，更方便，参考这里：http://www.nowamagic.net/librarys/veda/detail/2495，但是不能使用按容量啊，一定要按行。说日志太大拿不回来的，罚面壁一小时！
mkdir/usr/local/huanghai -p
split-a 2 -d -l 50000 /usr/local/mysql/t_resource_info.txt /usr/local/huanghai/prefix2-3秒吧
3、清空原表，修改字段，反正有备份，不怕的
truncate t_resource_info;alter table t_resource_info add huanghai_test int;
4、优化环境配置，准备开始导入
SET autocommit=0;
SET unique_checks=0;
SET foreign_key_checks=0;set sql_log_bin=0;
SET @innodb_additional_mem_pool_size=26214400;set @innodb_buffer_pool_size=1073741824;set @innodb_log_buffer_size=8388608;set @innodb_log_file_size=268435456;
load data infile'/usr/local/huanghai/prefix00' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix01' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix02' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix03' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix04' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix05' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix06' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix07' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix08' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix09' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix10' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix11' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix12' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix13' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix14' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix15' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix16' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix17' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix18' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
load data infile'/usr/local/huanghai/prefix19' IGNORE into table dsideal_db.t_resource_info_huanghai fields terminated by ',' enclosed by '"';
commit;
5、恢复现场
SET autocommit=1;
SET unique_checks=1;
SET foreign_key_checks=1;set sql_log_bin=1;
6、建议使用python3进行开发一个程序，这样方便串连起来，这是我目前能想到的最好办法，基本无风险，速度上基本能利用磁盘的最大IO，不建议采用修改frm等暴力办法，那个对于最后面追加字段的可能还行，对于字段在中间的
可能就是灾难，而且没有办法程序化，这个办法是用python3开发起来，基本无困难。
补充一下测试结果，但这台机器实在是太NB了，可能一般的客户没有这样的条件，供参考吧：
测试表：
t_resource_info
记录个数：1582937
一、生成
[[email protected] TestLoadFile]# python3 ExportData.py
2017-11-05 17:03:57 成功创建工作目录！
2017-11-05 17:03:59 开始导出数据...
2017-11-05 17:04:29 成功导出数据！
2017-11-05 17:04:29 正在进行分割...
2017-11-05 17:04:32 成功进行分割！
导出需要35秒
二、重新导入
[[email protected] TestLoadFile]# python3 ImportData.py
2017-11-05 16:58:08,378 INFO : 开始生成SQL脚本...
2017-11-05 16:58:08,380 INFO : 成功生成SQL脚本！
2017-11-05 16:58:08,380 INFO : 开始执行SQL脚本...
2017-11-05 16:59:27,223 INFO : SQL脚本执行成功！
导入需要79秒
合计需要114秒。
===================================================================================================
测试用机：
物理机，4颗CPU
cat /proc/cpuinfo| grep "physical id"| sort| uniq| wc -l
4
CPU核数：
cat /proc/cpuinfo| grep "cpu cores"
逻辑处理器个数：
cat /proc/cpuinfo| grep "processor"| wc -l
64
cpu 型号:
[[email protected] TestLoadFile]# cat /proc/cpuinfo | grep name | cut -f2 -d: | uniq -c
64 Intel(R) Xeon(R) CPU E7-4809 v4 @ 2.10GHz
内存：
cat /proc/meminfo
MemTotal: 65845352 kB
===================================================================================================
附10.10.14.224测试用例
[[email protected] TestLoadFile]# python3 ExportData.py
2017-11-06 07:51:14 成功创建工作目录！
2017-11-06 07:51:14 开始导出数据...
2017-11-06 07:53:12 成功导出数据！
2017-11-06 07:53:12 正在进行分割...
2017-11-06 07:53:27 成功进行分割！
You have new mail in /var/spool/mail/root
[[email protected] TestLoadFile]# python3 ImportData.py
2017-11-06 07:55:37,622 INFO : 开始生成SQL脚本...
2017-11-06 07:55:37,629 INFO : 成功生成SQL脚本！
2017-11-06 07:55:37,630 INFO : 开始执行SQL脚本...
2017-11-06 08:07:40,093 INFO : SQL脚本执行成功！
===================================================================================================
附：测试用例 链接：http://pan.baidu.com/s/1dFbCEIl 密码：75j5