Merge branch 'main' into docs/dclow-restruct-2

This commit is contained in:
danielclow 2024-12-03 10:34:32 +08:00 committed by GitHub
commit 8e256cc9c9
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
8 changed files with 151 additions and 68 deletions

View File

@ -2,7 +2,7 @@
IF (DEFINED VERNUMBER)
SET(TD_VER_NUMBER ${VERNUMBER})
ELSE ()
SET(TD_VER_NUMBER "3.3.4.3.alpha")
SET(TD_VER_NUMBER "3.3.4.8.alpha")
ENDIF ()
IF (DEFINED VERCOMPATIBLE)

View File

@ -25,6 +25,10 @@ Download links for TDengine 3.x version installation packages are as follows:
import Release from "/components/ReleaseV3";
## 3.3.4.8
<Release type="tdengine" version="3.3.4.8" />
## 3.3.4.3
<Release type="tdengine" version="3.3.4.3" />

View File

@ -4,69 +4,61 @@ sidebar_label: 3.3.4.3
slug: /release-history/release-notes/3-3-4-3
---
### Behavioral Changes and Compatibility
## New Features
1. feat: the twa function in stream computing support time-driven result push mode.
1. feat: the interp function in stream computing support time-driver result push mode.
1. feat: support Azure Blob
1. snode must be used in multi-replica stream computing
1. Added a compatibility assurance mechanism for stream computing to avoid compatibility issues with subsequent function changes. However, stream computing in previous versions must be rebuilt.
1. Adjusted the method for determining the result type of the case when statement
## Optimizations
1. enh: improve the efficiency of pulling data between dnodes when performing large concurrent query.
1. enh: support use AVX2 and AVX512 to perform decoding optimization for double, timestamp and bigint types.
1. enh: adjust the result type of the case when statement.
1. enh: optimized the error msg when compact + split vgroup operation is performed in sequence
1. enh: improve the data scanning speed of the query "select... from... where ts in (...)".
1. enh: add a compatiability guarantee mechanism for stream, to avoid new compatiblity issued caused by subsequent functions, but the previously created stream must be deleted and recreated.
1. enh: improve the data synchronization performance of taosX in interlace scenarios.
1. enh: support turn off encoding for integer and float data types.
1. enh: in a multi-replica database, snode must be used in stream computing.
1. enh: generate unique ID to identify each query task to avoid memory corruption.
1. enh: speed up the creation time of the database.
1. enh: modify the default value of s3MigrateEnabled to 0.
1. enh: support recording delete operations in the audit database.
1. enh: support create database in the specified dnode list [enterprise].
1. enh: adjust the error message when deleting data columns of the super table.
### New Features
1. New feature: TWA function in stream computing supports time-driven result push mode
1. New feature: Interp function in stream computing supports time-driven result push mode
1. New feature: Support for Microsoft object storage
### Optimizations
1. Optimization: Improved efficiency of data pulling between nodes during concurrent large queries
1. Optimization: Support for using AVX2 and AVX512 to optimize decoding for double, timestamp, and bigint types
1. Optimization: Adjusted the method for determining the result type of the case when statement
1. Optimization: Improved error logging during sequential execution of compact and split vgroup operations
1. Optimization: Increased data scan speed for queries like “select ... from ... where ts in (...)”
1. Optimization: Added a compatibility assurance mechanism for stream computing to avoid compatibility issues with subsequent function changes. Stream computing in previous versions must be rebuilt
1. Optimization: Improved data synchronization performance of taosX in cross-writing scenarios
1. Optimization: Support for disabling encoding of integer/floating-point types
1. Optimization: snode must be used in multi-replica stream computing
1. Optimization: Client generates a unique ID for each query task to avoid memory corruption caused by duplicate IDs
1. Optimization: Accelerated database creation time
1. Optimization: Changed the default value of s3MigrateEnabled to 0
1. Optimization: Support for recording delete operations in the audit database
1. Optimization: Support for creating databases in specified dnodes [Enterprise Edition]
1. Optimization: Adjusted error messages when deleting columns from supertables
### Fixes
1. Fix: Significant performance degradation in last_row queries in version 3.3.3.0
1. Fix: taosd unable to start when WAL entries are incomplete
1. Fix: Incorrect query results when partitioning by constants
1. Fix: Incorrect calculation results when scalar functions include _wstart and the fill method is prev
1. Fix: Time zone setting issues on Windows platforms
1. Fix: Transaction cannot end when compact operation is performed on an empty database [Enterprise Edition]
1. Fix: Logical errors in transaction conflicts
1. Fix: Certain errors in management nodes causing transactions to not stop
1. Fix: Retry errors in taosc after clearing dnode data
1. Fix: Intermediate files not cleaned up after Data Compact is abnormally terminated
1. Fix: Kafka connector's earliest mode not consuming new column data after adding columns
1. Fix: Incorrect behavior of interp function when using fill(prev)
1. Fix: TSMA stops unexpectedly during high-frequency metadata operations
1. Fix: Incorrect label display in the result of the show create stable statement
1. Fix: Percentile function crashes during large data volume queries
1. Fix: Syntax error when using partition by and having together
1. Fix: tbname is empty when interp is used with partition by tbname,c1
1. Fix: taosd may crash when writing illegal boolean values via stmt
1. Fix: Library symbol version conflicts with libraries using the same symbol
1. Fix: Continuous increase in handle count for JDBC drivers on Windows platforms
1. Fix: Occasional startup failures when upgrading from 3.3.3.1 to 3.3.4.0
1. Fix: Memory leaks on Windows platform when repeatedly adding and deleting tables
1. Fix: Unable to limit the number of concurrent checkpoints, causing excessive resource consumption in stream computing
1. Fix: too many session issues during concurrent queries
1. Fix: taos shell crashes in slow query scenarios on Windows platforms
1. Fix: Encrypted databases cannot recover when dnode logs are enabled
1. Fix: taosd unable to start due to mnode synchronization timeout
1. Fix: Vnode (virtual node) unable to recover due to slow data organization during snapshot synchronization
1. Fix: taosd crashes when writing data with escape characters to string-type fields via row protocol
1. Fix: Metadata file corruption caused by logical errors in Error Code processing
1. Fix: Query errors caused by not setting scalar mode when the query statement contains multiple nested "not" conditions
1. Fix: dnode offline issues caused by vnode statistics reporting timeout
1. Fix: taosd startup failures on servers that do not support the avx instruction set
1. Fix: Error code 0x09xx during taosX data migration fault tolerance handling
[Fixed issues]
1. fix: query performance of last_row has significantly decreased in version 3.3.3.0.
1. fix: taosd cannot start when the WAL entry is incomplete.
1. fix: incorrect query results when partitioning by a constant.
1. fix: when the scalar function include _wstart and the fill method is prev, the calculation result is wrong.
1. fix: the time zone setting problem on Windows.
1. fix: the translation can't finish if compacting on empty database [enterprise]
1. fix: logic error with transaction conflicts
1. fix: some errors of the mnode can cause transactions to be unable to stop.
1. fix: some errors of the mnode can cause transactions to be unable to stop.
1. fix: the problem of taosc driver retrying errors after clearing the dnode data.
1. fix: after Data Compact is abnormally terminated, intermediate files are not cleaned up.
1. fix: after adding a new column, the earliest mode of the Kafka connector cannot consume the data of the new column.
1. fix: the behavior of the interp function is incorrect when use fill(prev).
1. fix: the problem that TSMA abnormally stops during high-frequency metadata operations.
1. fix: the execution result of the "show create stable" statement is incorrect.
1. fix: the percentile function will crash when querying a large amount of data.
1. fix: syntax error when using partition by and having together.
1. fix: tbname is empty when interp function is used with partition by tbname, c1.
1. fix: taosd may crash while writing illegal boolean values through stmt.
1. fix: library symbol 'version' confict with other lib which use the same symbol.
1. fix: the number of JAVA process handles in JDBC driver keeps increasing on Windows environment.
1. fix: the occasional startup failure when upgrading from version 3.3.3.1 to 3.3.4.0.
1. fix: memory leak caused by repeated addition and deletion of tables on the Windows platform.
1. fix(stream): check the right return code for concurrent checkpoint trans.
1. fix: the "too many session" problem while perform large concurrent queries.
1. fix: the problem of taos shell crashing in slow query scenarios on the Windows platform.
1. fix: the encrypted database cannot be recovered when opening the dnode log.
1. fix: the problem that taosd cannot be started due to mnode synchronization timeout.
1. fix: the slow sorting of file group data during snapshot synchronization leads to the inability of Vnode to recover.
1. fix: when writing data with escape characters to a varchar field throug line protocol, taosd will crash.
1. fix: metadata file damage caused by incorrect logic processing of error code
1. fix: when a query statement contains multiple nested "not" conditional statements, not setting the scalar mode will lead to query errors.
1. fix: the problem of dnode going offline due to timeout of vnode stat report.
1. fix: taosd failed to start on servers that not support AVX instructions.
1. fix(taosX): handle 0x09xx error codes in migration

View File

@ -0,0 +1,40 @@
---
title: TDengine 3.3.4.8 Release Notes
sidebar_label: 3.3.4.8
description: Version 3.3.4.8 Notes
slug: /release-history/release-notes/3.3.4.8
---
## New Features
1. feat: Interp function supports the fill(near) mode, supports returning the timestamp where the original data is located, and supports finding interpolations within a specified range.
## Optimizations
1. enh: improve the query performance of TSBS.
2. enh: support alter multiple tag values of a single child table at the same time.
3. enh: optimize the error code of the status window.
4. enh: reduce resource consumption in the force window close mode of stream computing.
5. enh: remove the restriction that window query must include aggregate functions.
## Fixed issues
1. fix: use websocket in node.js example of explorer
2. fix: the problem that WAL is not distributed to multiple level-0 mount points.
3. fix: abnormal exit of stream computing may lead to a deadlock.
4. fix(taosX): fix drop table unexpected error when point set enable=0 in opc csv
5. fix: failed to read WAL file when alter WAL_level from 0 to 2.
6. fix: deadlock when create snode.
7. fix: The race condition during replica snapshot synchronization.
8. fix: the performance degradation of the percentile function after upgrading to 3.3.4.3.
9. fix: the info of dataDir in ins_dnode_variables is incomplete.
10. fix: WHERE condition not work in statements with INTERP function
11. fix: taosd restart failed while use force window close mode in stream computing.
12. Fix: SQL statements containing “GROUP BY constants” will report “Executor internal error”.
13. fix: fix OPC-DA connection incorrect NOCONFIG error
14. fix: in force_window_close mode, the stream computing fails to generate the sub-table names according to the specified rules.
15. fix: 'schema is old' occurring in schemaless write mode under specific scenarios. https://github.com/taosdata/TDengine/issues/28846
16. fix: taosX sync with subscription may lost messages in websocket
17. fix: incorrect sub-table count when filtering by tags (which may cause taosdump crash).
18. fix: restart a single dnode when RPC compression is enabled may lead to an abnormal cluster state.
19. fix: unexpected Grafana alerting with CASE WHEN
20. fix: taosd may crash when performing UNION queries frequently.
21. fix: subscribe with enable.auto.commit=false not work as expected

View File

@ -3,6 +3,10 @@ title: Release Notes
slug: /release-history/release-notes
---
[3.3.4.3](./3-3-4-3)
[3.3.3.0](./3-3-3-0)
[3.3.2.0](./3-3-2-0)
[3.3.4.8](./3-3-4-8/)
[3.3.4.3](./3-3-4-3/)
[3.3.3.0](./3-3-3-0/)
[3.3.2.0](./3-3-2-0/)

View File

@ -24,6 +24,10 @@ TDengine 3.x 各版本安装包下载链接如下:
import Release from "/components/ReleaseV3";
## 3.3.4.8
<Release type="tdengine" version="3.3.4.8" />
## 3.3.4.3
<Release type="tdengine" version="3.3.4.3" />

View File

@ -0,0 +1,38 @@
---
title: 3.3.4.8 版本说明
sidebar_label: 3.3.4.8
description: 3.3.4.8 版本说明
---
## 新特性
1. 新功能Interp 函数支持 fill(near) 填充模式,支持返回原始数据所在的 timestamp支持在指定的范围内查找插值
## 优化
1. 优化:提升 TSBS 测试的查询性能
2. 优化:支持同时修改单个子表的多个标签值
3. 优化:优化状态窗口的错误码
4. 优化:降低流计算 force windows close 模式下的资源消耗
5. 优化:取消窗口查询必须使用聚合函数的限制
## 修复
1. 修复Explorer Node.js 示例代码修复
2. 修复WAL 未分布到多个 0 级挂载点的问题
3. 修复:流计算异常退出可能导致死锁
4. 修复: OPC CSV 点位配置 enabled=0 时需要转义的table drop 失败
5. 修复:修改 WAL_level 时 WAL 文件读取失败的问题
6. 修复:创建流计算节点时的死锁问题
7. 修复:副本快照同步时的 race condition.
8. 修复:升级到 3.3.4.3 后 Percentile 函数性能下降的问题
9. 修复:系统表 ins_dnode_variables 中的 dataDir 信息不完整问题
10. 修复WHERE 条件在 INTERP 函数中不生效的问题
11. 修复:流计算 force window close 模式下重启失败的问题
12. 修复:包含 ”GROUP BY“ 常量的 SQL 语句会报语法解析错误
13. 修复: OPC-DA 连接检查报 NOCONFIG 错误
14. 修复:流计算在 force_window_close 模式下未按指定规则生成子表名
15. 修复在特定场景下schemaless 写入模式中出现的“schema is old”问题 https://github.com/taosdata/TDengine/issues/28846
16. 修复:在弱网下可能出现订阅数据部分缺失
17. 修复:通过标签过滤子表时数目不正确的问题(可能导致 taosdump crash
18. 修复:打开 RPC 压缩时单个 dnode 重启可能导致集群状态异常
19. 修复Grafana 使用 CASE WHEN 配置告警结果不符合预期
20. 修复:频繁执行 union 查询时可能出现的 crash
21. 修复:订阅参数 enable.auto.commit=false 不生效

View File

@ -4,6 +4,7 @@ sidebar_label: 版本说明
description: 各版本版本说明
---
[3.3.4.8](./3.3.4.8)
[3.3.4.3](./3.3.4.3)
[3.3.3.0](./3.3.3.0)
[3.3.2.0](./3.3.2.0)