Release 3.23.08.0

Release Notes:

We have released a new version 3.23.08.0.

Features:

• [#3831] Support OLAP/OLTP mixed load. More Detail, see doc.
• [#3832] New ManagerFrontWorker thread pool for management side. More Detail, see doc.
• [#3833] Supports monitoring of some timing threads and interrupt operations of some threads. More Detail, see doc.
• [#3834] Optimize and unify dble related internal statistics. More Detail, see doc.
• [#3838] Strengthen the handling of captured thread ‘error’ exceptions.

Bug-fixes:

• [#3835] When dble starts, some asynchronous initialization tasks take a long time
• [#3836] When the lower_case_table_name of the backend is inconsistent, some operations of dble do not meet expectations.
• Fix some DDL issues
• Fix clickhouse synchronization variable error when reloading
• Supports specifyCharset attribute configuration for ChildTable

Breaking backward compatibility:

• [#3834] Optimize and unify dble related internal statistics. More Detail, see doc.

特性：

• [#3831] 支持 OLAP/OLTP 混合负载. 更多详细信息，请参阅 doc.
• [#3832] ManagerFrontWorker 独立管理端线程池. 更多详细信息，请参阅 doc.
• [#3833] 支持监控部分定时线程以及部分线程的中断操作. 更多详细信息，请参阅 doc.
• [#3834] 优化并统一dble相关内部统计信息. 更多详细信息，请参阅 doc.
• [#3838] 加强对捕获的线程“错误”异常的处理.

缺陷修复:

• [#3835] dble启动时，某些异步初始化任务需要很长时间
• [#3836] 后端的lower_case_table_name不一致时，dble部分操作不符合预期
• 修复一些 DDL 问题
• 修复重新加载时clickhouse同步变量错误
• ChildTable支持specifyCharset属性配置

和旧版本的不兼容项：

• [#3834] 优化并统一dble相关内部统计信息. 更多详细信息，请参阅 doc.

Release 3.21.10.15

Release Notes:

We have released a bug-fixed version 3.21.10.15.

Bug-fixes:

• When the max function processes datetime/timestamp/time columns, the precision is discarded
• Modifying the white_ips column in dble_rw_split_entry to null does not take effect.
• Strengthen sequence configuration change scenario optimization
• Some npe processing

发行说明：

我们发布了3.21.10.15,这是个bug修复版本

问题修复:

• max函数处理datetime/timestamp/time列时，精度丢弃了
• 修改dble_rw_split_entry中white_ips列为null，不生效
• 加强sequence配置改动场景优化
• 一些NPE的处理

Release 3.22.11.1

Release Notes:

We have released a bug-fixed version 3.22.11.1.

Bug-fixes:

• Improve the logic of getting connections from connection pool
• In some scenarios, when executing a large package statement, hang, Lost connection...
• Support update/delete shardingTable with join globalTable
• Handle the semi-connection scenario when the backend connection is initialized
• After the master-slave switch, the 8066 side starts a transaction and rolls it back, some data may not be rolled back.
• The expansion and contraction thread connection-pool-evictor-thread of the connection pool exits unexpectedly
• Too many slow logs blocked, resulting in heartbeat failure and connection creation failure

发行说明：

我们发布了3.22.11.1,这是个bug修复版本

问题修复:

• 改善从连接池中获取连接的逻辑
• 某些场景下执行大包语句出现hang、Lost connection...
• 支持update/delete多表针对分片表join全局表
• 后端连接初始化时出现半连接场景的处理
• 主从切换后，业务端开启事务并进行回滚时，可能存在部分数据没有被回滚的现象
• 连接池的扩缩容线程connection-pool-evictor-thread意外退出
• 慢日志过多阻塞，导致心跳失败、创建连接失败

Release 3.22.01.6

Release Notes:

We have released a bug-fixed version 3.22.01.6.

Bug-fixes:

• After the master-slave switch, the 8066 side starts a transaction and rolls it back, some data may not be rolled back.
• Enable sql statistics, concurrent threads do not grab the lock and fall into an infinite loop, causing the cpu usage to be too high
• The expansion and contraction thread connection-pool-evictor-thread of the connection pool exits unexpectedly
• Too many slow logs blocked, resulting in heartbeat failure and connection creation failure
• Slow log bug causes queue consumption thread to exit abnormally
• Strengthen sequence configuration change scenario optimization
• Lost connection appears when issuing multiple statements...

发行说明：

我们发布了3.22.01.6,这是个bug修复版本

问题修复:

• 主从切换后，业务端开启事务并进行回滚时，可能存在部分数据没有被回滚的现象
• 开启sql统计，并发线程没有抢到锁陷入死循环引起cpu使用率过高
• 连接池的扩缩容线程connection-pool-evictor-thread意外退出
• 慢日志过多阻塞，导致心跳失败、创建连接失败
• 慢日志bug导致队列消费线程异常退出
• 多语句下发出现lost connnection...
• 加强sequence配置改动场景优化

Release 3.23.04.0

Release Notes:

We have released a new version 3.23.04.0.

Features:

• [#3738] Support update/delete shardingTable with join globalTable. More Detail, see doc1, doc2
• [#3743] Adjusted the -DfakeMysqlVersion setting rule in RW-split mode. More Detail, see doc
• [#3742] Adjust the default value of minimum stack size (-Xss).
• [#3739] Support for adjusting tcp related parameters（alpha version）. More Detail, see doc.
• [#3727] Optimize the global sequence loading logic.
• [#3745] Enhance/optimize the internal logic code that supports SSL, and handle scenarios that may cause memory leaks.

Bug-fixes:

• [#3741] In RW-split mode, when the auto_increment_increment of the mysql is not equal to 1, insert wrong data using jdbc connection dble
• [#3744] In RW-split mode, executing large package sql should not be limited by the maxPacketSize set by dble
• [#3746] The scaling thread(connection-pool-evictor-thread) exited unexpectedly
• [#3747] After the master-slave switch, the 8066 side opens the transaction for rollback and some data is not rolled back
• Support column aliases can be set when querying tables on the management side(9066)
• Adjust some log printing
• Fix some NPE issues

Breaking backward compatibility:

• [#3743] Adjusted the -DfakeMysqlVersion setting rule in rwsplit mode. More Detail, see doc

---

发行说明：

我们发布了新版本 3.23.04.0。

特性：

• [#3738] 针对分片表join全局表，支持update/delete多表. 更多详细信息，请参见 doc1, doc2
• [#3743] 对读写分离模式的-DfakeMysqlVersion设置规则进行了调整. 更多详细信息，请参见 doc
• [#3742] 调整最小栈值(-Xss)的默认值.
• [#3739] 支持调整tcp相关参数（实验功能）. 更多详细信息，请参见 doc
• [#3727] 优化全局序列加载内部逻辑
• [#3745] 加强/优化支持SSL的内部逻辑代码，以及可能发成内存泄漏的场景处理

缺陷修复:

• [#3741] 读写分离中，当数据库的自增步长不等于1时，使用jdbc连接dble插入错误数据
• [#3744] 读写分离中，执行大包语句不受dble设置的maxPacketSize的限制
• [#3746] 扩缩容connection-pool-evictor-thread线程偶现退出
• [#3747] 主从切换后，业务端开启事务进行rollback时，部分数据没有被回滚
• 支持在管理端查询表时可以设置列别名
• 调整一些日志打印
• 解决一些NPE问题

和旧版本的不兼容项：

• [#3743] 对读写分离模式的-DfakeMysqlVersion设置规则进行了调整。更多详细信息，请参见 doc

Release 3.21.10.14

Release Notes:

We have released a bug-fixed version 3.21.10.14.

Bug-fixes:

• Enable sql statistics, concurrent threads do not grab the lock and fall into an infinite loop, causing the cpu usage to be too high
• The expansion and contraction thread connection-pool-evictor-thread of the connection pool exits unexpectedly
• Improve the logic of fetching connections in the connection pool
• Handle the semi-connection scenario when the backend connection is initialized
• In some scenarios, when executing a large package statement, hang, Lost connection...
• Support update/delete shardingTable with join globalTable
• After the master-slave switch, the 8066 side opens the transaction for rollback and some data is not rolled back
• Too many slow logs blocked, resulting in heartbeat failure and connection creation failure

发行说明：

我们发布了3.21.10.14,这是个基于3.21.10.12的bug修复版本

问题修复:

• 开启sql统计，并发线程没有抢到锁陷入死循环引起cpu使用率过高
• 连接池的扩缩容线程connection-pool-evictor-thread意外退出
• 改善连接池中取连接的逻辑
• 后端连接初始化时出现半连接场景的处理
• 某些场景下执行大包语句出现hang、Lost connection...
• 支持update/delete多表针对分片表join全局表
• 主从切换后，业务端开启事务进行rollback时，部分数据没有被回滚
• 慢日志过多阻塞，导致心跳失败、创建连接失败

Release 3.22.01.5

Release Notes:

We have released a bug-fixed version 3.22.01.5.

Bug-fixes:

• Improve the logic of getting connections from connection pool
• Handle the semi-connection scenario when the backend connection is initialized
• In some scenarios, when executing a large package statement, hang, Lost connection...
• Support update/delete shardingTable with join globalTable
• After the master-slave switch, the 8066 side opens the transaction for rollback and some data is not rolled back
• When executing load data, Lost connection... and other errors occur
• Fix some NPE issues

发行说明：

我们发布了3.22.01.5,这是个基于3.22.01.4的bug修复版本

问题修复:

• 改善从连接池中获取连接的逻辑
• 后端连接初始化时出现半连接场景的处理
• 某些场景下执行大包语句出现hang、Lost connection...
• 支持update/delete多表针对分片表join全局表
• 主从切换后，业务端开启事务进行rollback时，部分数据没有被回滚
• 执行load data出现Lost connection...和其他报错
• 解决一些NPE问题

Release 3.22.07.3

Release Notes:

We have released a bug-fixed version 3.22.07.3.

Bug-fixes:

• In some scenarios, when executing a large package statement, hang, Lost connection...
• Two tables are queried through a non-shard key association, resulting in a memory leak
• Load data local execution error
• When the main instance network fails, the execution of complex queries causes the Business thread to hang
• Some heartbeat issues

发行说明：

我们发布了3.22.07.3,这是个基于3.22.07.01的bug修复版本

问题修复:

• 某些场景下执行大包语句出现hang、Lost connection...
• 两张表通过非分片键关联查询，出现内存泄露
• Load data local执行报错
• 主实例网络故障时，执行复杂查询导致Business线程hang住
• 一些心跳问题

Release 3.22.11.0

Release Notes:

We have released a new version 3.22.11.0.
Firstly, we must thanks to the following people for their contributions to our community：@xhailiang, @huashusu, @13611582598, @2779382063, @zeroxxmmbm, @24pawa.

Features:

• [#3529] Add monitor for off-heap memory leak. More Detail, see doc.

• [#3549] Support delay detection under master-slave structure. More Detail, see doc.

• [#3431] Support sql_dump log. More Detail,see doc.

• [#3429] Support local read for Read-write separation when using sharding-table. More Detail, see doc.Thanks to @2779382063 for reporting the issue.

• [#3586] Support some scenes of multi-table update. More Detail, see doc

• [#3587] Pseudo support for dropping dble database. More Detail, see doc

• [#3585] Support native json decode function. More Detail, see doc

Bug-fixes:

• [#3420] When set useCursorFetch=true in mysql-connector-java-8.0.27 , dble may report ArrayIndexOutOfBoundsException, thanks to @xhailiang for reporting the bug.
• [#3450] Some parameters of the JDK like The java.specification.maintenance.version is not supported, thanks to @huashusu for reporting the bug.
• [#3444] MultiNodeSelectHandler may cause dble to hang , thanks to @13611582598 for reporting the bug.
• [#3495] CPU usage up to 100% and report lots of exceptions, thanks to @zeroxxmmbm for reporting the bug.
• [#3427] When load data local infile into dble, the used memory will not be recycled, thanks to @24pawa for reporting the bug.
• Optimize some log problems.
• Optimize some performance problems

Breaking backward compatibility:

• [#3546] Change the default unit of parameter delayThreshold, from second to millssecond. More Detail,seedoc

---

发行说明：

我们发布了新版本 3.22.11.0。
首先，我们必须感谢以下人士对我们社区的贡献：@xhailiang, @huashusu, @13611582598, @2779382063, @zeroxxmmbm, @24pawa .

特性：

• [#3529] 添加堆外内存泄漏监视器. 更多详细信息，请参见doc.

• [#3549] 支持主从结构下的延时检测. 更多详细信息，请参见doc.

• [#3431] 支持审计日志. 更多详细信息，请参见doc.

• [#3429] 支持分库分表时候的读写分离本地读功能. 更多详细信息，请参见doc.感谢 @2779382063 贡献此需求.

• [#3586] 支持部分场景的多表update. 更多详细信息，请参见doc

• [#3587] 伪支持删除 dble 数据库. 更多详细信息，请参见doc

• [#3585] 支持本地json函数的计算.更多详细信息，请参见doc

缺陷修复:

• [#3420] 当我们在 mysql-connector-java-8.0.27里设置了useCursorFetch=true, dble 可能会报错ArrayIndexOutOfBoundsException, 感谢 @xhailiang 报告此bug.
• [#3450] 不支持 JDK 的一些参数，例如 The java.specification.maintenance.version, 感谢 @huashusu 报告此bug.
• [#3444] MultiNodeSelectHandler 可能导致dble 卡死 , 感谢 @13611582598 报告此bug.
• [#3495] CPU使用率高达100% 并且报出大量异常, 感谢 @zeroxxmmbm 报告此bug.
• [#3427] 当使用load data local infile导入dble, 使用过的内存不会被回收, 感谢 @24pawa 报告此bug.
• 优化一些日志问题。
• 优化一些性能问题。

和旧版本的不兼容项：

• [#3546] 将参数“delayThreshold”的默认单位从秒更改为毫秒。更多详细信息，请参见doc

Release 3.21.10.12

Release Notes:

We have released a bug-fixed version 3.21.10.12.

Bug-fixes:

• Some parameters of the JDK like The java.specification.maintenance.version is not supported.
• MultiNodeSelectHandler may cause dble to hang.
• CPU usage up to 300%
• Some problems about maxPacketSize
• When load data local infile into dble, the used memory will not be recycled
• Optimize some problems about complex query.
• Optimize some heartbeat problems
• Memory leak caused by some complex queries

发行说明：

我们发布了3.21.10.12,这是个基于3.21.10.10的bug修复版本

问题修复:

• 不支持 JDK 的一些参数，例如 The java.specification.maintenance.version
• MultiNodeSelectHandler 可能导致dble 卡死
• CPU使用率高达300%问题
• maxPacketSize相关问题。
• 当使用load data local infile导入dble, 使用过的内存不会被回收
• 复杂查询的一些问题
• 心跳的优化
• 某些复杂查询可能引起的内存泄漏