mysql命令导入超大csv文件常见问题

• 远程新建表

CREATE TABLE `UserBehavior` (
  `id` int NOT NULL AUTO_INCREMENT,
  `user_id` varchar(45) CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci NULL DEFAULT NULL,
  `item_id` varchar(45) CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci NULL DEFAULT NULL,
  `category_id` varchar(45) CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci NULL DEFAULT NULL,
  `behavior_id` varchar(45) CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci NULL DEFAULT NULL,
  `timestamp` varchar(45) CHARACTER SET utf8mb4 COLLATE utf8mb4_0900_ai_ci NULL DEFAULT NULL,
  PRIMARY KEY (`id`) USING BTREE
) ENGINE = InnoDB AUTO_INCREMENT = 10660001 CHARACTER SET = utf8mb4 COLLATE = utf8mb4_0900_ai_ci ROW_FORMAT = DYNAMIC;

• 准备csv文件
  

• 将csv文件导出指定行数，文件会小点，如果直接导入超大文件，加载时间过长，会导入失败

import os
import pandas as pd
# 每份csv文件截取100万行，保存到当前路径
chunksize = 1000000
# csv路径
filename = r"C:\work\mysql\UserBehavior.csv"
data1 = pd.read_csv(filename, chunksize=chunksize, header=None, sep=',')
num = 0
for chunk in data1:
    num += len(chunk)
    head, tail = os.path.split(filename)
    data2 = pd.read_csv(filename, chunksize=chunksize, header=None, sep=',')
    # 保存了多少份csv文件
    i = 0
    for chunk in data2:
        chunk.to_csv('{0}\{1}{2}'.format(head, i, tail), header=True, index=False)
        print('保存第{0}个数据'.format(i))
        i += 1

• 测试1

# 本地进入UserBehavior.csv所在文件夹，右键打开Git Bash

# 执行如下代码，导入成功，导入速度极快，100w条数据(36MB)导入花费6秒
ychen@DESKTOP-49HGBQ1 MINGW64 /c/work/mysql
$ mysqlimport -h192.168.128.80 -P3306 -uroot -p123456 db_test --fields-terminated-by=',' UserBehavior.csv --columns='user_id,item_id,category_id,behavior_id,timestamp' --local
mysqlimport: [Warning] Using a password on the command line interface can be insecure.
db_test.UserBehavior: Records: 1000001  Deleted: 0  Skipped: 0  Warnings: 0

• 测试2

# 打开cmd，登录mysql
C:\Users\ychen>mysql -h 192.168.128.80 -P 3306 -u root -p --local-infile
Enter password: **********

# 切换库
mysql> use db_test
Database changed

# 执行如下代码，只导入成功9739条数据
mysql> LOAD DATA LOCAL INFILE 'C:\\work\\mysql\\UserBehavior.csv'
    -> INTO TABLE UserBehavior
    -> FIELDS TERMINATED BY ','
    -> ENCLOSED BY '"'
    -> LINES TERMINATED BY '\n'
    -> IGNORE 1 ROWS;
Query OK, 9739 rows affected, 65535 warnings (3.76 sec)
Records: 1000000  Deleted: 0  Skipped: 990261  Warnings: 1990261

# 复制如下代码
LOAD DATA LOCAL INFILE 'C:\\work\\mysql\\UserBehavior.csv'
INTO TABLE UserBehavior
FIELDS TERMINATED BY ','
ENCLOSED BY '"'
LINES TERMINATED BY '\n'
IGNORE 1 ROWS;

• 连接失败，防火墙开放端口

firewall-cmd --zone=public --add-port=3306/tcp --permanent
firewall-cmd --reload

• 报错1114

[root@hadoop01 ~]# whereis my.cnf 
my: /etc/my.cnf
[root@hadoop01 ~]# cd /etc
[root@hadoop01 etc]# vim my.cnf
# 添加或修改如下
tmp_table_size=256M
max_heap_table_size=256M
# 重启
[root@hadoop01 etc]# systemctl restart mysqld.service
# 查看
show variables like '%max_heap_table_size%';
show variables like '%tmp_table_size%';

• 报错1130，修改为允许任意ip连接

mysql -u root -p
use mysql;
update user set host = '%' where user ='root';
flush privileges;
select host, user from user

• 报错

> 2 - File 'C:xxx' not found (Errcode: 2 "No such file or directory")

# 将单斜杠改为双斜杠

• 报错

ychen@DESKTOP-49HGBQ1 MINGW64 /c/work/mysql
$ mysqlimport -h192.168.128.80 -P3306 -uroot -pvmc79m8035 db_test --fields-terminated-by=',' UserBehavior.csv --columns='user_id,item_id,category_id,behavior_id,timestamp' --local
mysqlimport: [Warning] Using a password on the command line interface can be insecure.
mysqlimport: Error: 3948, Loading local data is disabled; this must be enabled on both the client and server sides, when using table: UserBehavior

# 修改权限
show variables like 'local_infile';
set global local_infile = 1;