5.7 与 8.0 对相同文件的 LOAD DATA 语句结果不同

关注

原创最新推荐文章于 2026-09-24 12:45:12 发布 704 阅读 AI 写同款· GEO 优化 ›

使用 LOAD DATA LOCAL INFILE，sysbench 导数速度提升30% Python微信订餐小程序课程视频 https://edu.csdn.net/course/detail/36074 Python实战量化交易理财系统 https://edu.csdn.net/course/detail/35475 最近给 sysbench 提了一个 feature（https://github.com/akopytov/sysbench/pull/450），支持通过 LOAD DATA LOCAL INFILE 命令导入压测数据。下面我们来具体看看这个 feature 的使用方法和实现阅读详情

5.7 与 8.0 对相同文件的 LOAD DATA 语句结果不同

问题描述

某客户现场支持，由MySQL 5.7.21升级MySQL 8.0.25后，通过LOAD DATA导入文件，当同一会话连续导入不同的编码（UTF8/GB18030）文件时会出现乱码。数据库版本未升级之前，相同的导入操作在MySQL 5.7.21未出现乱码。

问题分析

1）查看简化后的 LOAD DATA语句

greatsql> LOAD DATA LOCAL INFILE '/home/greatdb/TEST_UTF8_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET UTF8MB4 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING UTF8MB4)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> LOAD DATA LOCAL INFILE '/home/greatdb/TEST_GB18030_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET GB18030 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING GB18030)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

2）查看表数据

+----------+------------------------------------------------------+
| AUTO_INC | D_NAME                                               |
+----------+------------------------------------------------------+
|        1 | xxx社会保险xxx                                        |
|        2 | xxx市路桥区xxx                                        |
|        4 | 鍙板窞甯傝矾妗ュ尯绀句細淇濋櫓浜嬩笟绠＄悊涓績             |
|        5 | 鍙板窞甯傝矾妗ュ尯绀句細淇濋櫓浜嬩笟绠＄悊涓績             |
+----------+------------------------------------------------------+
4 rows in set (0.00 sec)

3）检查业务表的字符集与校验集，发现字符集为 utf8mb4 、校验集为 utf8mb4_bin

4）检查数据库的字符集与校验集

greatsql> SHOW GLOBAL VARIABLES LIKE '%char%';
+--------------------------------------+--------------------------------+
| Variable_name                        | Value                          |
+--------------------------------------+--------------------------------+
| character_set_client                 | utf8mb4                        |
| character_set_connection             | utf8mb4                        |
| character_set_database               | utf8mb4                        |
| character_set_filesystem             | binary                         |
| character_set_results                | utf8mb4                        |
| character_set_server                 | utf8mb4                        |
| character_set_system                 | utf8mb3                        |
| character_sets_dir                   | /opt/mysql3301/share/charsets/ |
| validate_password_special_char_count | 1                              |
+--------------------------------------+--------------------------------+
9 rows in set (0.01 sec)

greatsql> SHOW GLOBAL VARIABLES LIKE '%coll%';
+-------------------------------+--------------------+
| Variable_name                 | Value              |
+-------------------------------+--------------------+
| collation_connection          | utf8mb4_bin        |
| collation_database            | utf8mb4_bin        |
| collation_server              | utf8mb4_bin        |
| default_collation_for_utf8mb4 | utf8mb4_general_ci |
+-------------------------------+--------------------+
4 rows in set (0.00 sec)

程序在MySQL 5.7.21跑了很长时间，一直没有问题，把数据库升级MySQL 8.0.25后，新导入的数据出现部分乱码，由此怀疑，MySQL 8.0定长数据导入LOAD DATA @row 出现BUG。

BUG场景：同一个会话 LOAD DATA多种字符集文件，使用@临时变量切割字段。将导致导入数据乱码，向MySQL官方提BUG,已证实为BUG（编号115824）

问题复现

MySQL: 8.0.25

greatsql> SELECT VERSION();
+-----------+
| version() |
+-----------+
| 8.0.25    |
+-----------+
1 row in set (0.00 sec)


table ddl:
CREATE TABLE `assp_sis_payres_imp_bak` (
  `AUTO_INC` bigint unsigned NOT NULL AUTO_INCREMENT COMMENT '自增列',
  `D_NAME` varchar(210) CHARACTER SET utf8mb4 COLLATE utf8mb4_bin DEFAULT NULL,
  PRIMARY KEY (`AUTO_INC`)
) ENGINE=InnoDB AUTO_INCREMENT=7 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin;


greatsql> SHOW GLOBAL VARIABLES LIKE '%char%';
+--------------------------------------+--------------------------------+
| Variable_name                        | Value                          |
+--------------------------------------+--------------------------------+
| character_set_client                 | utf8mb4                        |
| character_set_connection             | utf8mb4                        |
| character_set_database               | utf8mb4                        |
| character_set_filesystem             | binary                         |
| character_set_results                | utf8mb4                        |
| character_set_server                 | utf8mb4                        |
| character_set_system                 | utf8mb3                        |
| character_sets_dir                   | /opt/mysql3301/share/charsets/ |
| validate_password_special_char_count | 1                              |
+--------------------------------------+--------------------------------+
9 rows in set (0.01 sec)

greatsql> SHOW GLOBAL VARIABLES LIKE '%coll%';
+-------------------------------+--------------------+
| Variable_name                 | Value              |
+-------------------------------+--------------------+
| collation_connection          | utf8mb4_bin        |
| collation_database            | utf8mb4_bin        |
| collation_server              | utf8mb4_bin        |
| default_collation_for_utf8mb4 | utf8mb4_general_ci |
+-------------------------------+--------------------+
4 rows in set (0.00 sec)

greatsql> TRUNCATE TABLE assp_sis_payres_imp_bak;
Query OK, 0 rows affected (0.03 sec)

greatsql> SELECT charset(@row), @row;
+---------------+------------+
| charset(@row) | @row       |
+---------------+------------+
| binary        | NULL       |
+---------------+------------+
1 row in set (0.00 sec)

greatsql> LOAD DATA LOCAL INFILE '/root/dba_zc/load/TEST_UTF8_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET UTF8MB4 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING UTF8MB4)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> SELECT charset(@row), @row;  
+---------------+------------------------+
| charset(@row) | @row                   |
+---------------+------------------------+
| utf8mb4       | XXX路桥区社会保XXX       |
+---------------+------------------------+

greatsql> LOAD DATA LOCAL INFILE '/root/dba_zc/load/TEST_GB18030_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET GB18030 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING GB18030)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> SELECT charset(@row), @row;  
+---------------+-----------------------------------------+
| charset(@row) | @row                                    |
+---------------+-----------------------------------------+
| gb18030       | XXX路桥区社会保XXX       |
+---------------+-----------------------------------------+

greatsql>  SELECT * FROM  ASSP_SIS_PAYRES_IMP_BAK;
+----------+---------------------------------------------------------+
| AUTO_INC | D_NAME                                                  |
+----------+---------------------------------------------------------+
|        1 | XXX路桥区社会保XXX                                        |
|        2 | XXX路桥区社会保XXX                                        |
|        4 | 鍙板窞甯傝矾妗ュ尯绀句細淇濋櫓浜嬩笟绠＄悊涓績                 |
|        5 | 鍙板窞甯傝矾妗ュ尯绀句細淇濋櫓浜嬩笟绠＄悊涓績                 |
+----------+---------------------------------------------------------+
4 rows in set (0.00 sec)

MySQL 5.7.21

greatsql> SELECT VERSION();
+------------+
| version()  |
+------------+
| 5.7.21-log |
+------------+
1 row in set (0.01 sec)

table ddl:
CREATE TABLE `assp_sis_payres_imp_bak` (
  `AUTO_INC` bigint unsigned NOT NULL AUTO_INCREMENT COMMENT '自增列',
  `D_NAME` varchar(210) CHARACTER SET utf8mb4 COLLATE utf8mb4_bin DEFAULT NULL,
  PRIMARY KEY (`AUTO_INC`)
) ENGINE=InnoDB AUTO_INCREMENT=7 DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin;

greatsql>  SHOW GLOBAL VARIABLES LIKE '%char%';
+--------------------------------------+--------------------------------+
| Variable_name                        | Value                          |
+--------------------------------------+--------------------------------+
| character_set_client                 | utf8mb4                        |
| character_set_connection             | utf8mb4                        |
| character_set_database               | utf8mb4                        |
| character_set_filesystem             | binary                         |
| character_set_results                | utf8mb4                        |
| character_set_server                 | utf8mb4                        |
| character_set_system                 | utf8                           |
| character_sets_dir                   | /opt/mysql3305/share/charsets/ |
| validate_password_special_char_count | 1                              |
+--------------------------------------+--------------------------------+
9 rows in set (0.00 sec)

greatsql> SHOW GLOBAL VARIABLES LIKE '%coll%';
+----------------------+--------------------+
| Variable_name        | Value              |
+----------------------+--------------------+
| collation_connection | utf8mb4_general_ci |
| collation_database   | utf8mb4_general_ci |
| collation_server     | utf8mb4_general_ci |
+----------------------+--------------------+
3 rows in set (0.00 sec)

greatsql> SELECT charset(@row), @row;
+---------------+------------+
| charset(@row) | @row       |
+---------------+------------+
| binary        | NULL       |
+---------------+------------+
1 row in set (0.00 sec)

greatsql> LOAD DATA LOCAL INFILE '/root/dba_zc/load/TEST_UTF8_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET UTF8MB4 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING UTF8MB4)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> SELECT charset(@row), @row;  
+---------------+-----------------------+
| charset(@row) | @row                  |
+---------------+-----------------------+
| utf8mb4       | XXX路桥区社会保XXX      |
+---------------+-----------------------+

greatsql> LOAD DATA LOCAL INFILE '/root/dba_zc/load/TEST_GB18030_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET GB18030 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING GB18030)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> SELECT charset(@row), @row;  
+---------------+-----------------------+
| charset(@row) | @row                  |
+---------------+-----------------------+
| gb18030       | XXX路桥区社会保XXX      |
+---------------+-----------------------+

greatsql> SELECT * FROM  ASSP_SIS_PAYRES_IMP_BAK;                                                                                                                                                 
+---------------+-----------------------------+
| AUTO_INC      | D_NAME                      |
+---------------+-----------------------------+
|        1 | XXX路桥区社会保XXX                 |
|        2 | XXX路桥区社会保XXX                 |
|        4 | XXX路桥区社会保XXX                 |
|        5 | XXX路桥区社会保XXX                 |
+---------------+-----------------------------+
4 rows in set (0.00 sec)

BUG规避方案

通过SELECT``charset(@row), @row; 可以看到@row在执行LOAD DATA后在5.7.21和8.0.25是一样的，但最终的影响不一样。虽然MySQL官方确认此问题为BUG，但没有提供规避方案或者解决方案。通过万里工程师研究后，发现一种可行的规避方案。每次执行LOAD DATA命令前执行 [set @row=_binary'';] 进行规避。

greatsql> SELECT VERSION();
+-----------+
| version() |
+-----------+
| 8.0.25    |
+-----------+
1 row in set (0.00 sec)

greatsql> SET @row=_binary'';
Query OK, 0 rows affected (0.00 sec)

greatsql> LOAD DATA LOCAL INFILE '/home/greatdb/TEST_UTF8_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET UTF8MB4 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING UTF8MB4)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> SET @row=_binary'';
Query OK, 0 rows affected (0.00 sec)

greatsql> LOAD DATA LOCAL INFILE '/home/greatdb/TEST_GB18030_bak.txt' IGNORE INTO TABLE ASSP_SIS_PAYRES_IMP_BAK CHARACTER SET GB18030 IGNORE 0 LINES  (@row) SET `D_NAME` = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row), 1,240)) USING GB18030)),'');
Query OK, 2 rows affected (0.01 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

greatsql> SELECT * FROM assp_sis_payres_imp_bak;
+----------+--------------------------------------------------+
| AUTO_INC | D_NAME                                           |
+----------+--------------------------------------------------+
|        1 | XXX路桥区社会保XXX                 |
|        2 | XXX路桥区社会保XXX                 |
|        4 | XXX路桥区社会保XXX                 |
|        5 | XXX路桥区社会保XXX                 |
+----------+--------------------------------------------------+
4 rows in set (0.00 sec)

问题总结

1.BUG原因

MySQL8.0重构定长数据导入LOAD DATA @row 出现BUG.同一个数据库会话，多次执行LOAD DATA @row命令，则第n次执行LOAD DATA @row 的字符集使用的是n-1次的字符集，当文件的字符集存在不同，例如先后处理GB18030、UTF8字符集的文件就会数据乱码。此问题MySQL官方已证实为BUG（编号115824）

2.BUG触发条件

触发条件：需同时满足以下三个条件才会触发此bug。

1）LOAD DATA命令使用类似 @row临时变量进行数据处理，例如对定长记录按字节切割出多个字段：

LINES (@row) SET COLUMN_NAME = NULLIF(TRIM(CONVERT(UNHEX(SUBSTR(HEX(@row),1,20)) USING GB18030))。

2）在同一个连接中，多次执行LOAD DATA命令，且先后处理的文件字符集存在不同（例如GB18030和UTF8）。

3）使用MySQL 8.0。

3.BUG规避办法

由万里工程师提出，与MySQL官方社区沟通证实，涉及到满足上述BUG触发条件的场景，通过在每次执行LOAD DATA命令前执行 [set @row=_binary'';] 进行规避。

参考：https://bugs.mysql.com/bug.php?id=115824

Enjoy GreatSQL :)

关于 GreatSQL

GreatSQL是适用于金融级应用的国内自主开源数据库，具备高性能、高可靠、高易用性、高安全等多个核心特性，可以作为MySQL或Percona Server的可选替换，用于线上生产环境，且完全免费并兼容MySQL或Percona Server。

相关链接： GreatSQL社区 Gitee GitHub Bilibili

GreatSQL社区：

社区有奖建议反馈： https://greatsql.cn/thread-54-1-1.html

社区博客有奖征稿详情： https://greatsql.cn/thread-100-1-1.html

（对文章有疑问或者有独到见解都可以去社区官网提出或分享哦~）

技术交流群：

微信&QQ群：

QQ群：533341697

微信群：添加GreatSQL社区助手（微信号：wanlidbc ）好友，待社区助手拉您进群。

标签

#GreatSQL #MySQL

Gemma 4 12B QAT+MTP小显存部署实战指南大语言模型本地部署的核心瓶颈在于显存与推理效率的平衡。量化感知训练（QAT）通过在训练阶段模拟低精度计算，显著提升模型对INT4量化的鲁棒性；多令牌预测（MTP）则通过并行解码优化KV Cache复用，降低GPU kernel调用频次。二者协同可在8GB显存设备上稳定运行12B级模型，兼顾精度、速度与上下文长度。该方案适用于开发者本地AI助手、技术文档生成、SQL翻译等轻量但高交互需求场景，是当前Windows环境中小显存高效落地Gemma系列模型的工程最优解。阅读详情

5.7 与 8.0 对相同文件的 LOAD DATA 语句结果不同

5.7 与 8.0 对相同文件的 LOAD DATA 语句结果不同

问题描述

问题分析

问题复现

问题总结

参考：https://bugs.mysql.com/bug.php?id=115824

关于 GreatSQL

GreatSQL社区：

技术交流群：

相关推荐