一、问题
数据库mysql,在使用mybatis进行批量插入的时候,报错:### Cause: com.mysql.cj.jdbc.exceptions.PacketTooBigException: Packet for query is too large (4,850,051 > 4,194,304). You can change this value on the server by setting the 'max_allowed_packet' variable.
,如:
二、原因
在对mysql进行插入、更新或查询操作时,mysql server接收处理的数据包大小是有限制的,如果太大超过了设置的max_allowed_packet
参数的大小,会导致操作失败,我们可以通过命令:show VARIABLES like '%max_allowed_packet%';
查看参数值,如:
三、解决方案
数据库层面解决:调整
max_allowed_packet
参数,改大,额,多大合适呢?应用层面解决:批量插入时,将入参list进行分割,分批批量插入,控制每次批量插入的大小,避免超过最大限制
我们这里选择第2种解决方案
四、代码实现
1、v0.1版本(每个mapper对应的service方法中,都进行分割处理)
1.1、我们需要用到google提供的方法:Lists.partition()
对list进行分割,引入maven:
org.apache.commons
commons-collections4
4.1
1.2、UserMapper.xml
:
insert into user(username, password)
values
(#{item.username}, #{item.password})
update user
when id = #{item.id} then #{item.username}
when id = #{item.id} then #{item.password}
where id in
#{item.id}
1.3、UserServiceImpl
:
@Override
public void batchInsert(List list) {
if (CollectionUtils.isEmpty(list)) {
return;
}
List> partition = Lists.partition(list, MAX_SIZE_PER_TIME);
for (List batchList : partition) {
userMapper.batchInsert(batchList);
}
}
@Override
public void batchUpdate(List list) {
if (CollectionUtils.isEmpty(list)) {
return;
}
List> partition = Lists.partition(list, MAX_SIZE_PER_TIME);
for (List batchList : partition) {
userMapper.batchUpdate(batchList);
}
}
MAX_SIZE_PER_TIME
:即每次最大插入记录数,为方便测试,我这里取3,大家可以根据具体场景、数据库、业务需求指定,没有唯一
其他service以此类推,这样可以实现,但是会发现重复代码一堆,需改进
2、v0.2版本(代理模式实现,对方法进行增强,批量插入前,统一对集合进行分割)
2.1、首先定义一个mapper的批量处理接口:BatchProcessMapper
:
package com.wangzaiplus.test.service.batch;
import java.util.List;
public interface BatchProcessMapper {
void batchInsert(List list);
void batchUpdate(List list);
}
2.2、然后UserMapper
等mapper实现BatchProcessMapper
:
package com.wangzaiplus.test.mapper;
import com.wangzaiplus.test.service.batch.BatchProcessMapper;
public interface UserMapper extends BatchProcessMapper {
}
2.3、然后定义MapperProxy
也实现BatchProcessMapper
:
package com.wangzaiplus.test.service.batch.mapperproxy;
import com.google.common.collect.Lists;
import com.wangzaiplus.test.service.batch.BatchProcessMapper;
import org.apache.commons.collections4.CollectionUtils;
import java.util.List;
import static com.wangzaiplus.test.common.Constant.MAX_SIZE_PER_TIME;
public class MapperProxy implements BatchProcessMapper {
private BatchProcessMapper batchProcessMapper;
public MapperProxy(BatchProcessMapper batchProcessMapper) {
this.batchProcessMapper = batchProcessMapper;
}
@Override
public void batchInsert(List list) {
if (CollectionUtils.isEmpty(list)) {
return;
}
List> partition = Lists.partition(list, MAX_SIZE_PER_TIME);
for (List batchList : partition) {
batchProcessMapper.batchInsert(batchList);
}
}
@Override
public void batchUpdate(List list) {
if (CollectionUtils.isEmpty(list)) {
return;
}
List> partition = Lists.partition(list, MAX_SIZE_PER_TIME);
for (List batchList : partition) {
batchProcessMapper.batchUpdate(batchList);
}
}
}
2.4、使用:UserServiceImpl
中直接调用批量插入方法:
@Service
public class UserServiceImpl implements UserService {
@Autowired
private UserMapper userMapper;
@Override
public void batchInsert(List list) {
new MapperProxy(userMapper).batchInsert(list);
}
@Override
public void batchUpdate(List list) {
new MapperProxy(userMapper).batchUpdate(list);
}
}
其他service同上,一行代码搞定,省却重复代码
五、mybatis批量插入与更新
1、批量插入:
1.1、语法:
insert into user(username, password)
values
(#{item.username}, #{item.password})
1.2、sql实际执行语句:
2、批量更新:
2.1、语法:
update user
when id = #{item.id} then #{item.username}
when id = #{item.id} then #{item.password}
where id in
#{item.id}
2.2、sql实际执行语句:
格式化一下是这样子:
UPDATE USER
SET username = CASE
WHEN id = ? THEN
?
WHEN id = ? THEN
?
WHEN id = ? THEN
?
END,
PASSWORD = CASE
WHEN id = ? THEN
?
WHEN id = ? THEN
?
WHEN id = ? THEN
?
END
WHERE
id IN (?, ?, ?)
434616(Integer), batchUpdate_3366361bc15048129adfcf5dc851d7e5(String),
434617(Integer), batchUpdate_76370f7c845642f2be0ab1c35440f5bd(String),
434618(Integer), batchUpdate_6e06394ccbbd49c8b97c3e0600a7cada(String),
434616(Integer), 123456(String),
434617(Integer), 123456(String),
434618(Integer), 123456(String),
434616(Integer),
434617(Integer),
434618(Integer)
六、总结
在使用mybatis批量操作数据库时,需要注意集合大小,太大会导致sql执行异常,所以在批量操作前,应该对集合进行分割处理,然后分批操作,为了减少重复代码,可以通过代理模式对批量处理方法进行增强,客户端调用时只需将具体子类传给代理,由代理在执行批量语句前统一进行分割处理
代码已更新至Github,欢迎大家clone、交流,谢谢
Github
https://github.com/wangzaiplus/springboot/tree/wxw