xxq250
/
homework-jianmu
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge branch 'dv' into 3

This commit is contained in:
yihaoDeng 2022-07-30 15:45:57 +08:00
parent d2f57b4ad7 02d97d97dc
commit d0d9526d44
265 changed files with 6884 additions and 6069 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
.gitmodules vendored
View File

@ -1,12 +0,0 @@
[submodule "src/connector/go"]
path = src/connector/go
url = git@github.com:taosdata/driver-go.git
[submodule "src/connector/hivemq-tdengine-extension"]
path = src/connector/hivemq-tdengine-extension
url = git@github.com:taosdata/hivemq-tdengine-extension.git
[submodule "deps/TSZ"]
path = deps/TSZ
url = https://github.com/taosdata/TSZ.git
[submodule "examples/rust"]
path = examples/rust
url = https://github.com/songtianyi/tdengine-rust-bindings.git

6
Jenkinsfile2
View File

@ -118,6 +118,7 @@ def pre_test(){
git rm --cached tools/taos-tools 2>/dev/null || :
git rm --cached tools/taosadapter 2>/dev/null || :
git rm --cached tools/taosws-rs 2>/dev/null || :
git rm --cached examples/rust 2>/dev/null || :
'''
sh '''
cd ${WKC}
@ -269,6 +270,7 @@ def pre_test_win(){
git rm --cached tools/taos-tools 2>nul
git rm --cached tools/taosadapter 2>nul
git rm --cached tools/taosws-rs 2>nul
git rm --cached examples/rust 2>nul
exit 0
'''
bat '''
@ -443,7 +445,7 @@ pipeline {
}
}
}
catchError(buildResult: 'FAILURE', stageResult: 'FAILURE') {
/*catchError(buildResult: 'FAILURE', stageResult: 'FAILURE') {
timeout(time: 15, unit: 'MINUTES'){
script {
sh '''
@ -455,7 +457,7 @@ pipeline {
'''
}
}
}
}*/
}
}
}

26
cmake/cmake.options
View File

@ -47,15 +47,21 @@ IF(${TD_WINDOWS})
)
option(
BUILD_TEST
"If build unit tests using googletest"
ON
)
BUILD_TEST
"If build unit tests using googletest"
ON
)
option(
TDENGINE_3
"TDengine 3.x"
ON
TDENGINE_3
"TDengine 3.x for taos-tools"
ON
)
option(
BUILD_CRASHDUMP
"If build crashdump on Windows"
ON
)
ELSEIF (TD_DARWIN_64)
@ -84,6 +90,12 @@ ELSE ()
ENDIF ()
ENDIF ()
option(
RUST_BINDINGS
"If build with rust-bindings"
ON
)
option(
JEMALLOC_ENABLED
"If build with jemalloc"

12
cmake/crashdump_CMakeLists.txt.in Normal file
View File

@ -0,0 +1,12 @@
# crashdump
ExternalProject_Add(crashdump
GIT_REPOSITORY https://github.com/Arnavion/crashdump.git
GIT_TAG master
SOURCE_DIR "${TD_CONTRIB_DIR}/crashdump"
BINARY_DIR "${TD_CONTRIB_DIR}/crashdump"
CONFIGURE_COMMAND ""
BUILD_COMMAND ""
INSTALL_COMMAND ""
TEST_COMMAND ""
)

12
cmake/rust-bindings_CMakeLists.txt.in Normal file
View File

@ -0,0 +1,12 @@
# rust-bindings
ExternalProject_Add(rust-bindings
GIT_REPOSITORY https://github.com/songtianyi/tdengine-rust-bindings.git
GIT_TAG 7ed7a97
SOURCE_DIR "${TD_SOURCE_DIR}/examples/rust"
BINARY_DIR "${TD_SOURCE_DIR}/examples/rust"
CONFIGURE_COMMAND ""
BUILD_COMMAND ""
INSTALL_COMMAND ""
TEST_COMMAND ""
)

2
cmake/taosadapter_CMakeLists.txt.in
View File

@ -1,5 +1,5 @@
# zlib
# taosadapter
ExternalProject_Add(taosadapter
GIT_REPOSITORY https://github.com/taosdata/taosadapter.git
GIT_TAG df8678f

4
cmake/taostools_CMakeLists.txt.in
View File

@ -1,8 +1,8 @@
# zlib
# taos-tools
ExternalProject_Add(taos-tools
GIT_REPOSITORY https://github.com/taosdata/taos-tools.git
GIT_TAG 2.1.1
GIT_TAG 9dc2fec
SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools"
BINARY_DIR ""
#BUILD_IN_SOURCE TRUE

2
cmake/taosws_CMakeLists.txt.in
View File

@ -1,5 +1,5 @@
# zlib
# taosws-rs
ExternalProject_Add(taosws-rs
GIT_REPOSITORY https://github.com/taosdata/taosws-rs.git
GIT_TAG 9de599d

54
contrib/CMakeLists.txt
View File

@ -27,10 +27,6 @@ else ()
cat("${TD_SUPPORT_DIR}/taosadapter_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
endif()
if(TD_LINUX_64 AND JEMALLOC_ENABLED)
cat("${TD_SUPPORT_DIR}/jemalloc_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
endif()
# pthread
if(${BUILD_PTHREAD})
cat("${TD_SUPPORT_DIR}/pthread_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
@ -109,6 +105,11 @@ if(${BUILD_WITH_SQLITE})
cat("${TD_SUPPORT_DIR}/sqlite_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
endif(${BUILD_WITH_SQLITE})
# rust-bindings
if(${RUST_BINDINGS})
cat("${TD_SUPPORT_DIR}/rust-bindings_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
endif(${RUST_BINDINGS})
# lucene
if(${BUILD_WITH_LUCENE})
cat("${TD_SUPPORT_DIR}/lucene_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
@ -120,6 +121,11 @@ if(${BUILD_WITH_NURAFT})
cat("${TD_SUPPORT_DIR}/nuraft_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
endif(${BUILD_WITH_NURAFT})
# crashdump
if(${BUILD_CRASHDUMP})
cat("${TD_SUPPORT_DIR}/crashdump_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
endif(${BUILD_CRASHDUMP})
# addr2line
if(${BUILD_ADDR2LINE})
if(NOT ${TD_WINDOWS})
@ -135,6 +141,24 @@ execute_process(COMMAND "${CMAKE_COMMAND}" -G "${CMAKE_GENERATOR}" .
execute_process(COMMAND "${CMAKE_COMMAND}" --build .
WORKING_DIRECTORY "${TD_CONTRIB_DIR}/deps-download")
# clear submodule
execute_process(COMMAND git submodule deinit -f tools/taos-tools
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git rm --cached tools/taos-tools
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git submodule deinit -f tools/taosadapter
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git rm --cached tools/taosadapter
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git submodule deinit -f tools/taosws-rs
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git rm --cached tools/taosws-rs
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git submodule deinit -f examples/rust
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
execute_process(COMMAND git rm --cached examples/rust
WORKING_DIRECTORY "${TD_SOURCE_DIR}")
# ================================================================================================
# Build
# ================================================================================================
@ -257,6 +281,16 @@ if(${BUILD_PTHREAD})
target_link_libraries(pthread INTERFACE libpthreadVC3)
endif()
# crashdump
if(${BUILD_CRASHDUMP})
add_executable(dumper "crashdump/dumper/dumper.c")
target_link_libraries(dumper User32.lib dbghelp.lib)
file(READ "crashdump/crasher/crasher.c" CRASHDUMP_CONTENT)
string(REPLACE "main(" "main_crashdump(" CRASHDUMP_CONTENT "${CRASHDUMP_CONTENT}")
file(WRITE "crashdump/crasher/crasher.c" "${CRASHDUMP_CONTENT}")
add_library(crashdump STATIC "crashdump/crasher/crasher.c")
endif()
# iconv
if(${BUILD_WITH_ICONV})
add_library(iconv STATIC iconv/win_iconv.c)
@ -403,18 +437,6 @@ if(${BUILD_ADDR2LINE})
endif(NOT ${TD_WINDOWS})
endif(${BUILD_ADDR2LINE})
# jemalloc
IF (TD_LINUX_64 AND JEMALLOC_ENABLED)
include(ExternalProject)
ExternalProject_Add(jemalloc
PREFIX "jemalloc"
SOURCE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/jemalloc
BUILD_IN_SOURCE 1
CONFIGURE_COMMAND ./autogen.sh COMMAND ./configure --prefix=${CMAKE_BINARY_DIR}/build/
BUILD_COMMAND ${MAKE}
)
INCLUDE_DIRECTORIES(${CMAKE_BINARY_DIR}/build/include)
ENDIF ()
# ================================================================================================
# Build test

4
docs/en/07-develop/03-insert-data/03-opentsdb-telnet.mdx
View File

@ -23,7 +23,7 @@ A single line of text is used in OpenTSDB line protocol to represent one row of
- `metric` will be used as the STable name.
- `timestamp` is the timestamp of current row of data. The time precision will be determined automatically based on the length of the timestamp. Second and millisecond time precision are supported.
- `value` is a metric which must be a numeric value, the corresponding column name is "value".
- `value` is a metric which must be a numeric value, the corresponding column name is "_value".
- The last part is the tag set separated by spaces, all tags will be converted to nchar type automatically.
For example:
@ -74,7 +74,7 @@ taos> show STables;
Query OK, 2 row(s) in set (0.002544s)
taos> select tbname, * from `meters.current`;
tbname | ts | value | groupid | location |
tbname | _ts | _value | groupid | location |
==================================================================================================================================
t_0e7bcfa21a02331c06764f275... | 2022-03-28 09:56:51.249 | 10.800000000 | 3 | California.LoSangeles |
t_0e7bcfa21a02331c06764f275... | 2022-03-28 09:56:51.250 | 11.300000000 | 3 | California.LoSangeles |

2
docs/en/07-develop/03-insert-data/04-opentsdb-json.mdx
View File

@ -91,7 +91,7 @@ taos> show STables;
Query OK, 2 row(s) in set (0.001954s)
taos> select * from `meters.current`;
ts | value | groupid | location |
_ts | _value | groupid | location |
===================================================================================================================
2022-03-28 09:56:51.249 | 10.300000000 | 2.000000000 | California.SanFrancisco |
2022-03-28 09:56:51.250 | 12.600000000 | 2.000000000 | California.SanFrancisco |

4
docs/zh/07-develop/03-insert-data/03-opentsdb-telnet.mdx
View File

@ -23,7 +23,7 @@ OpenTSDB 行协议同样采用一行字符串来表示一行数据。OpenTSDB
- metric 将作为超级表名。
- timestamp 本行数据对应的时间戳。根据时间戳的长度自动识别时间精度。支持秒和毫秒两种时间精度
- value 度量值，必须为一个数值。对应的列名也是 “value”。
- value 度量值，必须为一个数值。对应的列名是 “_value”。
- 最后一部分是标签集， 用空格分隔不同标签， 所有标签自动转化为 nchar 数据类型;
例如：
@ -74,7 +74,7 @@ taos> show stables;
Query OK, 2 row(s) in set (0.002544s)
taos> select tbname, * from `meters.current`;
tbname | ts | value | groupid | location |
tbname | _ts | _value | groupid | location |
==================================================================================================================================
t_0e7bcfa21a02331c06764f275... | 2022-03-28 09:56:51.249 | 10.800000000 | 3 | California.LosAngeles |
t_0e7bcfa21a02331c06764f275... | 2022-03-28 09:56:51.250 | 11.300000000 | 3 | California.LosAngeles |

2
docs/zh/07-develop/03-insert-data/04-opentsdb-json.mdx
View File

@ -91,7 +91,7 @@ taos> show stables;
Query OK, 2 row(s) in set (0.001954s)
taos> select * from `meters.current`;
ts | value | groupid | location |
_ts | _value | groupid | location |
===================================================================================================================
2022-03-28 09:56:51.249 | 10.300000000 | 2.000000000 | California.SanFrancisco |
2022-03-28 09:56:51.250 | 12.600000000 | 2.000000000 | California.SanFrancisco |

231
docs/zh/14-reference/12-config/index.md
View File

@ -1,6 +1,6 @@
---
title: 配置参数
description: 'TDengine 客户端和服务配置列表'
description: "TDengine 客户端和服务配置列表"
---
## 为服务端指定配置文件
@ -21,8 +21,6 @@ taosd -C
TDengine 系统的前台交互客户端应用程序为 taos，以及应用驱动，它可以与 taosd 共享同一个配置文件 taos.cfg，也可以使用单独指定配置文件。运行 taos 时，使用参数-c 指定配置文件目录，如 taos -c /home/cfg，表示使用/home/cfg/目录下的 taos.cfg 配置文件中的参数，缺省目录是/etc/taos。更多 taos 的使用方法请见帮助信息 `taos --help`。
**2.0.10.0 之后版本支持命令行以下参数显示当前客户端参数的配置**
```bash
taos -C
```
@ -47,19 +45,19 @@ taos --dump-config
### firstEp
| 属性 | 说明 |
| -------- | --------------------------------------------------------------- |
| 适用范围 | 服务端和客户端均适用 |
| 属性 | 说明 |
| -------- | -------------------------------------------------------------- |
| 适用范围 | 服务端和客户端均适用 |
| 含义 | taosd 或者 taos 启动时，主动连接的集群中首个 dnode 的 endpoint |
| 缺省值 | localhost:6030 |
| 缺省值 | localhost:6030 |
### secondEp
| 属性 | 说明 |
| -------- | -------------------------------------------------------------------------------------- |
| 适用范围 | 服务端和客户端均适用 |
| 属性 | 说明 |
| -------- | ------------------------------------------------------------------------------------- |
| 适用范围 | 服务端和客户端均适用 |
| 含义 | taosd 或者 taos 启动时，如果 firstEp 连接不上，尝试连接集群中第二个 dnode 的 endpoint |
| 缺省值 | 无 |
| 缺省值 | 无 |
### fqdn
@ -77,7 +75,6 @@ taos --dump-config
| 适用范围 | 仅服务端适用 |
| 含义 | taosd 启动后，对外服务的端口号 |
| 缺省值 | 6030 |
| 补充说明 | RESTful 服务在 2.4.0.0 之前（不含）由 taosd 提供，默认端口为 6041; 在 2.4.0.0 及后续版本由 taosAdapter，默认端口为 6041 |
:::note
确保集群中所有主机在端口 6030 上的 TCP 协议能够互通。（详细的端口情况请参见下表）
@ -87,8 +84,8 @@ taos --dump-config
| TCP | 6030 | 客户端与服务端之间通讯，多节点集群的节点间通讯。 | 由配置文件设置 serverPort 决定。 |
| TCP | 6041 | 客户端与服务端之间的 RESTful 通讯。 | 随 serverPort 端口变化。注意 taosAdapter 配置或有不同，请参考相应[文档](/reference/taosadapter/)。 |
| TCP | 6043 | TaosKeeper 监控服务端口。 | 随 TaosKeeper 启动参数设置变化。 |
| TCP | 6044 | 支持 StatsD 的数据接入端口。 | 随 taosAdapter 启动参数设置变化（2.3.0.1+以上版本）。 |
| UDP | 6045 | 支持 collectd 数据接入端口。 | 随 taosAdapter 启动参数设置变化（2.3.0.1+以上版本）。 |
| TCP | 6044 | 支持 StatsD 的数据接入端口。 | 随 taosAdapter 启动参数设置变化|
| UDP | 6045 | 支持 collectd 数据接入端口。 | 随 taosAdapter 启动参数设置变化 |
| TCP | 6060 | 企业版内 Monitor 服务的网络端口。 | |
### maxShellConns
@ -104,28 +101,28 @@ taos --dump-config
### monitor
| 属性 | 说明 |
| -------- | ---------------------------------------------------------------------------------------------------------------------------------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | 服务器内部的系统监控开关。监控主要负责收集物理节点的负载状况，包括 CPU、内存、硬盘、网络带宽的监控记录，监控信息将通过 HTTP 协议发送给由 `monitorFqdn` 和 `monitorProt` 指定的 TaosKeeper 监控服务 |
| 取值范围 | 0：关闭监控服务， 1：激活监控服务。 |
| 缺省值 | 1 |
| 取值范围 | 0：关闭监控服务， 1：激活监控服务。 |
| 缺省值 | 1 |
### monitorFqdn
| 属性 | 说明 |
| -------- | -------------------------------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | -------------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | TaosKeeper 监控服务的 FQDN |
| 缺省值 | 无 |
| 缺省值 | 无 |
### monitorPort
| 属性 | 说明 |
| -------- | -------------------------------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | --------------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | TaosKeeper 监控服务的端口号 |
| 缺省值 | 6043 |
| 缺省值 | 6043 |
### monitorInterval
@ -134,10 +131,9 @@ taos --dump-config
| 适用范围 | 仅服务端适用 |
| 含义 | 监控数据库记录系统参数（CPU/内存）的时间间隔 |
| 单位 | 秒 |
| 取值范围 | 1-200000 |
| 取值范围 | 1-200000 |
| 缺省值 | 30 |
### telemetryReporting
| 属性 | 说明 |
@ -149,25 +145,43 @@ taos --dump-config
## 查询相关
### queryBufferSize
### queryPolicy
| 属性 | 说明 |
| -------- | ----------------------------- |
| 适用范围 | 仅客户端适用 |
| 含义 | 查询语句的执行策略 |
| 单位 | 无 |
| 缺省值 | 1 |
| 补充说明 | 1: 只使用 vnode，不使用 qnode |
2: 没有扫描算子的子任务在 qnode 执行，带扫描算子的子任务在 vnode 执行
3: vnode 只运行扫描算子，其余算子均在 qnode 执行 |
### querySmaOptimize
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 仅客户端适用 |
| 含义 | sma index 的优化策略 |
| 单位 | 无 |
| 缺省值 | 0 |
| 补充说明 |
0: 表示不使用 sma index，永远从原始数据进行查询
1: 表示使用 sma index，对符合的语句，直接从预计算的结果进行查询 |
| 属性 | 说明 |
| -------- | ------------------------------------------------------------------------------------------------------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | 为所有并发查询占用保留的内存大小。 |
| 单位 | MB |
| 缺省值 | 无 |
| 补充说明 | 计算规则可以根据实际应用可能的最大并发数和表的数字相乘，再乘 170 。<br/>（2.0.15 以前的版本中，此参数的单位是字节） |
### maxNumOfDistinctRes
| 属性 | 说明 |
| -------- | -------------------------------- |
| -------- | -------------------------------- | --- |
| 适用范围 | 仅服务端适用 |
| 含义 | 允许返回的 distinct 结果最大行数 |
| 取值范围 | 默认值为 10 万，最大值 1 亿 |
| 缺省值 | 10 万 |
| 补充说明 | 2.3 版本新增。 | |
## 区域相关
@ -306,12 +320,12 @@ charset 的有效值是 UTF-8。
### supportVnodes
| 属性 | 说明 |
| -------- | ----------------------------------------------------------------------------------------------------------------------------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | dnode 支持的最大 vnode 数目 |
| 取值范围 | 0-4096 |
| 缺省值 | 256 |
| 属性 | 说明 |
| -------- | --------------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | dnode 支持的最大 vnode 数目 |
| 取值范围 | 0-4096 |
| 缺省值 | 256 |
## 时间相关
@ -366,7 +380,6 @@ charset 的有效值是 UTF-8。
| 单位 | bytes |
| 取值范围 | 0: 对所有查询结果均进行压缩 >0: 查询结果中任意列大小超过该值的消息才进行压缩 -1: 不压缩 |
| 缺省值 | -1 |
| 补充说明 | 2.3.0.0 版本新增。 |
## 日志相关
@ -464,7 +477,7 @@ charset 的有效值是 UTF-8。
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 服务端和客户端均适用 |
| 含义 | query 模块的日志开关 |
| 含义 | query 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
@ -481,7 +494,7 @@ charset 的有效值是 UTF-8。
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 仅服务端适用 |
| 适用范围 | 仅服务端适用 |
| 含义 | dnode 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | 135 |
@ -490,28 +503,28 @@ charset 的有效值是 UTF-8。
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 仅服务端适用 |
| 适用范围 | 仅服务端适用 |
| 含义 | vnode 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
### mDebugFlag
| 属性 | 说明 |
| -------- | ------------------ |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | mnode 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | 135 |
| 取值范围 | 同上 |
| 缺省值 | 135 |
### wDebugFlag
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | wal 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | 135 |
| 属性 | 说明 |
| -------- | ------------------ |
| 适用范围 | 仅服务端适用 |
| 含义 | wal 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | 135 |
### sDebugFlag
@ -533,57 +546,86 @@ charset 的有效值是 UTF-8。
### tqDebugFlag
| 属性 | 说明 |
| -------- | ------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | ----------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | tq 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
| 取值范围 | 同上 |
| 缺省值 | |
### fsDebugFlag
| 属性 | 说明 |
| -------- | ------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | ----------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | fs 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
| 取值范围 | 同上 |
| 缺省值 | |
### udfDebugFlag
| 属性 | 说明 |
| -------- | ---------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | ------------------ |
| 适用范围 | 仅服务端适用 |
| 含义 | UDF 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
| 取值范围 | 同上 |
| 缺省值 | |
### smaDebugFlag
| 属性 | 说明 |
| -------- | ---------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | ------------------ |
| 适用范围 | 仅服务端适用 |
| 含义 | sma 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
| 取值范围 | 同上 |
| 缺省值 | |
### idxDebugFlag
| 属性 | 说明 |
| -------- | ---------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | -------------------- |
| 适用范围 | 仅服务端适用 |
| 含义 | index 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
| 取值范围 | 同上 |
| 缺省值 | |
### tdbDebugFlag
| 属性 | 说明 |
| -------- | ---------------------- |
| 适用范围 | 仅服务端适用 |
| 属性 | 说明 |
| -------- | ------------------ |
| 适用范围 | 仅服务端适用 |
| 含义 | tdb 模块的日志开关 |
| 取值范围 | 同上 |
| 缺省值 | |
| 取值范围 | 同上 |
| 缺省值 | |
## Schemaless 相关
### smlChildTableName
| 属性 | 说明 |
| -------- | ------------------------- |
| 适用范围 | 仅客户端适用 |
| 含义 | schemaless 自定义的子表名 |
| 类型 | 字符串 |
| 缺省值 | 无 |
### smlTagName
| 属性 | 说明 |
| -------- | ------------------------------------ |
| 适用范围 | 仅客户端适用 |
| 含义 | schemaless tag 为空时默认的 tag 名字 |
| 类型 | 字符串 |
| 缺省值 | _tag_null |
### smlDataFormat
| 属性 | 说明 |
| -------- | ----------------------------- |
| 适用范围 | 仅客户端适用 |
| 含义 | schemaless 列数据是否顺序一致 |
| 值域 | 0：不一致；1: 一致 |
| 缺省值 | 1 |
## 其他
@ -596,3 +638,12 @@ charset 的有效值是 UTF-8。
| 取值范围 | 0：否，1：是 |
| 缺省值 | 1 |
| 补充说明 | 不同的启动方式，生成 core 文件的目录如下：1、systemctl start taosd 启动：生成的 core 在根目录下 <br/> 2、手动启动，就在 taosd 执行目录下。 |
### udf
| 属性 | 说明 |
| -------- | ------------------ |
| 适用范围 | 仅服务端适用 |
| 含义 | 是否启动 udf 服务 |
| 取值范围 | 0: 不启动；1：启动 |
| 缺省值 | 1 |

121
docs/zh/21-tdinternal/01-arch.md
View File

@ -5,7 +5,7 @@ title: 整体架构
## 集群与基本逻辑单元
TDengine 的设计是基于单个硬件、软件系统不可靠，基于任何单台计算机都无法提供足够计算能力和存储能力处理海量数据的假设进行设计的。因此 TDengine 从研发的第一天起，就按照分布式高可靠架构进行设计，是支持水平扩展的，这样任何单台或多台服务器发生硬件故障或软件错误都不影响系统的可用性和可靠性。同时，通过节点虚拟化并辅以自动化负载均衡技术，TDengine 能最高效率地利用异构集群中的计算和存储资源降低硬件投资。
TDengine 的设计是基于单个硬件、软件系统不可靠，基于任何单台计算机都无法提供足够计算能力和存储能力处理海量数据的假设进行设计的。因此 TDengine 从研发的第一天起，就按照分布式高可靠架构进行设计，是支持水平扩展的，这样任何单台或多台服务器发生硬件故障或软件错误都不影响系统的可用性和可靠性。同时，通过节点虚拟化并辅以负载均衡技术，TDengine 能最高效率地利用异构集群中的计算和存储资源降低硬件投资。
### 主要逻辑单元
@ -19,45 +19,43 @@ TDengine 分布式架构的逻辑结构图如下：
**物理节点（pnode）：** pnode 是一独立运行、拥有自己的计算、存储和网络能力的计算机，可以是安装有 OS 的物理机、虚拟机或 Docker 容器。物理节点由其配置的 FQDN（Fully Qualified Domain Name）来标识。TDengine 完全依赖 FQDN 来进行网络通讯，如果不了解 FQDN，请看博文[《一篇文章说清楚 TDengine 的 FQDN》](https://www.taosdata.com/blog/2020/09/11/1824.html)。
**数据节点（dnode）：** dnode 是 TDengine 服务器侧执行代码 taosd 在物理节点上的一个运行实例，一个工作的系统必须有至少一个数据节点。dnode 包含零到多个逻辑的虚拟节点（vnode），零或者至多一个逻辑的管理节点（mnode）。dnode 在系统中的唯一标识由实例的 End Point（EP）决定。EP 是 dnode 所在物理节点的 FQDN（Fully Qualified Domain Name）和系统所配置的网络端口号（Port）的组合。通过配置不同的端口，一个物理节点（一台物理机、虚拟机或容器）可以运行多个实例，或有多个数据节点。
**数据节点（dnode）：** dnode 是 TDengine 服务器侧执行代码 taosd 在物理节点上的一个运行实例，一个工作的系统必须有至少一个数据节点。dnode 包含零到多个逻辑的虚拟节点（vnode），零或者至多一个逻辑的管理节点（mnode），零或者至多一个逻辑的弹性计算节点（qnode），零或者至多一个逻辑的流计算节点（snode）。dnode 在系统中的唯一标识由实例的 End Point（EP）决定。EP 是 dnode 所在物理节点的 FQDN（Fully Qualified Domain Name）和系统所配置的网络端口号（Port）的组合。通过配置不同的端口，一个物理节点（一台物理机、虚拟机或容器）可以运行多个实例，或有多个数据节点。
**虚拟节点（vnode）：** 为更好的支持数据分片、负载均衡，防止数据过热或倾斜，数据节点被虚拟化成多个虚拟节点（vnode，图中 V2，V3，V4 等）。每个 vnode 都是一个相对独立的工作单元，是时序数据存储的基本单元，具有独立的运行线程、内存空间与持久化存储的路径。一个 vnode 包含一定数量的表（数据采集点）。当创建一张新表时，系统会检查是否需要创建新的 vnode。一个数据节点上能创建的 vnode 的数量取决于该数据节点所在物理节点的硬件资源。一个 vnode 只属于一个 DB，但一个 DB 可以有多个 vnode。一个 vnode 除存储的时序数据外，也保存有所包含的表的 schema、标签值等。一个虚拟节点由所属的数据节点的 EP，以及所属的 VGroup ID 在系统内唯一标识，由管理节点创建并管理。
**虚拟节点（vnode）：** 为更好的支持数据分片、负载均衡，防止数据过热或倾斜，数据节点被虚拟化成多个虚拟节点（vnode，图中 V2，V3，V4 等）。每个 vnode 都是一个相对独立的工作单元，是时序数据存储的基本单元，具有独立的运行线程、内存空间与持久化存储的路径。一个 vnode 包含一定数量的表（数据采集点）。当创建一个新 DB 时，系统会创建新的 vnode。一个数据节点上能创建的 vnode 的数量取决于该数据节点所在物理节点的硬件资源。一个 vnode 只属于一个 DB，但一个 DB 可以有多个 vnode。一个 vnode 除存储的时序数据外，也保存有所包含的表的 schema、标签值等。一个虚拟节点由所属的数据节点的 EP，以及所属的 VGroup ID 在系统内唯一标识，由管理节点创建并管理。
**管理节点（mnode）：** 一个虚拟的逻辑单元，负责所有数据节点运行状态的监控和维护，以及节点之间的负载均衡（图中 M）。同时，管理节点也负责元数据（包括用户、数据库、表、静态标签等）的存储和管理，因此也称为 Meta Node。TDengine 集群中可配置多个（最多不超过 3 个）mnode，它们自动构建成为一个虚拟管理节点组（图中 M0，M1，M2）。mnode 间采用 master/slave 的机制进行管理，而且采取强一致方式进行数据同步，任何数据更新操作只能在 Master 上进行。mnode 集群的创建由系统自动完成，无需人工干预。每个 dnode 上至多有一个 mnode，由所属的数据节点的 EP 来唯一标识。每个 dnode 通过内部消息交互自动获取整个集群中所有 mnode 所在的 dnode 的 EP。
**管理节点（mnode）：** 一个虚拟的逻辑单元，负责所有数据节点运行状态的监控和维护，以及节点之间的负载均衡（图中 M）。同时，管理节点也负责元数据（包括用户、数据库、超级表等）的存储和管理，因此也称为 Meta Node。TDengine 集群中可配置多个（最多不超过 3 个）mnode，它们自动构建成为一个虚拟管理节点组（图中 M1，M2，M3）。mnode 支持多副本，采用 RAFT 一致性协议，保证系统的高可用与高可靠，任何数据更新操作只能在 Leader 上进行。mnode 集群的第一个节点在集群部署时自动完成，其他节点的创建与删除由用户通过 SQL 命令完成。每个 dnode 上至多有一个 mnode，由所属的数据节点的 EP 来唯一标识。每个 dnode 通过内部消息交互自动获取整个集群中所有 mnode 所在的 dnode 的 EP。
**虚拟节点组（VGroup）：** 不同数据节点上的 vnode 可以组成一个虚拟节点组（vgroup）来保证系统的高可靠。虚拟节点组内采取 master/slave 的方式进行管理。写操作只能在 master vnode 上进行，系统采用异步复制的方式将数据同步到 slave vnode，这样确保了一份数据在多个物理节点上有拷贝。一个 vgroup 里虚拟节点个数就是数据的副本数。如果一个 DB 的副本数为 N，系统必须有至少 N 数据节点。副本数在创建 DB 时通过参数 replica 可以指定，缺省为 1。使用 TDengine 的多副本特性，可以不再需要昂贵的磁盘阵列等存储设备，就可以获得同样的数据高可靠性。虚拟节点组由管理节点创建、管理，并且由管理节点分配一个系统唯一的 ID，VGroup ID。如果两个虚拟节点的 VGroup ID 相同，说明他们属于同一个组，数据互为备份。虚拟节点组里虚拟节点的个数是可以动态改变的，容许只有一个，也就是没有数据复制。VGroup ID 是永远不变的，即使一个虚拟节点组被删除，它的 ID 也不会被收回重复利用。
**弹性计算节点（qnode）：** 一个虚拟的逻辑单元，运行查询计算任务，也包括基于系统表来实现的 show 命令（图中 Q）。集群中可配置多个 qnode，在整个集群内部共享使用（图中 Q1，Q2，Q3）。qnode 不与具体的 DB 绑定，即一个 qnode 可以同时执行多个 DB 的查询任务。每个 dnode 上至多有一个 qnode，由所属的数据节点的 EP 来唯一标识。客户端通过与 mnode 交互，获取可用的 qnode 列表，当没有可用的 qnode 时，计算任务在 vnode 中执行。
**Taosc** taosc 是 TDengine 给应用提供的驱动程序（driver），负责处理应用与集群的接口交互，提供 C/C++ 语言原生接口，内嵌于 JDBC、C#、Python、Go、Node.js 语言连接库里。应用都是通过 taosc 而不是直接连接集群中的数据节点与整个集群进行交互的。这个模块负责获取并缓存元数据；将插入、查询等请求转发到正确的数据节点；在把结果返回给应用时，还需要负责最后一级的聚合、排序、过滤等操作。对于 JDBC、C/C++、C#、Python、Go、Node.js 接口而言，这个模块是在应用所处的物理节点上运行。同时，为支持全分布式的 RESTful 接口，taosc 在 TDengine 集群的每个 dnode 上都有一运行实例。
**流计算节点（snode）：** 一个虚拟的逻辑单元，只运行流计算任务（图中 S）。集群中可配置多个 snode，在整个集群内部共享使用（图中 S1，S2，S3）。snode 不与具体的 stream 绑定，即一个 snode 可以同时执行多个 stream 的计算任务。每个 dnode 上至多有一个 snode，由所属的数据节点的 EP 来唯一标识。由 mnode 调度可用的 snode 完成流计算任务，当没有可用的 snode 时，流计算任务在 vnode 中执行。
**虚拟节点组（VGroup）：** 不同数据节点上的 vnode 可以组成一个虚拟节点组（vgroup），采用 RAFT 一致性协议，保证系统的高可用与高可靠。写操作只能在 leader vnode 上进行，系统采用异步复制的方式将数据同步到 follower vnode，这样确保了一份数据在多个物理节点上有拷贝。一个 vgroup 里虚拟节点个数就是数据的副本数。如果一个 DB 的副本数为 N，系统必须有至少 N 数据节点。副本数在创建 DB 时通过参数 replica 可以指定，缺省为 1。使用 TDengine 的多副本特性，可以不再需要昂贵的磁盘阵列等存储设备，就可以获得同样的数据高可靠性。虚拟节点组由管理节点创建、管理，并且由管理节点分配一个系统唯一的 ID，VGroup ID。如果两个虚拟节点的 VGroup ID 相同，说明他们属于同一个组，数据互为备份。虚拟节点组里虚拟节点的个数是可以动态改变的，容许只有一个，也就是没有数据复制。VGroup ID 是永远不变的，即使一个虚拟节点组被删除，它的 ID 也不会被收回重复利用。
**Taosc** taosc 是 TDengine 给应用提供的驱动程序（driver），负责处理应用与集群的接口交互，提供 C/C++ 语言原生接口，内嵌于 JDBC、C#、Python、Go、Node.js 语言连接库里。应用都是通过 taosc 而不是直接连接集群中的数据节点与整个集群进行交互的。这个模块负责获取并缓存元数据；将插入、查询等请求转发到正确的数据节点；在把结果返回给应用时，还需要负责最后一级的聚合、排序、过滤等操作。对于 JDBC、C/C++、C#、Python、Go、Node.js 接口而言，这个模块是在应用所处的物理节点上运行。同时，taosc 也可以与 taosAdapter 交互，支持全分布式的 RESTful 接口。
### 节点之间的通讯
**通讯方式：**TDengine 系统的各个数据节点之间，以及应用驱动与各数据节点之间的通讯是通过 TCP/UDP 进行的。因为考虑到物联网场景，数据写入的包一般不大，因此 TDengine 除采用 TCP 做传输之外，还采用 UDP 方式，因为 UDP 更加高效，而且不受连接数的限制。TDengine 实现了自己的超时、重传、确认等机制，以确保 UDP 的可靠传输。对于数据量不到 15K 的数据包，采取 UDP 的方式进行传输，超过 15K 的，或者是查询类的操作，自动采取 TCP 的方式进行传输。同时，TDengine 根据配置和数据包，会自动对数据进行压缩/解压缩，数字签名/认证等处理。对于数据节点之间的数据复制，只采用 TCP 方式进行数据传输。
**通讯方式：**TDengine 系统的各个数据节点之间，以及应用驱动与各数据节点之间的通讯是通过 TCP 进行的。TDengine 根据配置和数据包，会自动对数据进行压缩/解压缩，数字签名/认证等处理。
**FQDN 配置：**一个数据节点有一个或多个 FQDN，可以在系统配置文件 taos.cfg 通过参数“fqdn”进行指定，如果没有指定，系统将自动获取计算机的 hostname 作为其 FQDN。如果节点没有配置 FQDN，可以直接将该节点的配置参数 fqdn 设置为它的 IP 地址。但不建议使用 IP，因为 IP 地址可变，一旦变化，将让集群无法正常工作。一个数据节点的 EP（End Point）由 FQDN + Port 组成。采用 FQDN，需要保证 DNS 服务正常工作，或者在节点以及应用所在的节点配置好 hosts 文件。另外，这个参数值的长度需要控制在 96 个字符以内。
**端口配置：**一个数据节点对外的端口由 TDengine 的系统配置参数 serverPort 决定，对集群内部通讯的端口是 serverPort+5。为支持多线程高效的处理 UDP 数据，每个对内和对外的 UDP 连接，都需要占用 5 个连续的端口。
- 集群内数据节点之间的数据复制操作占用一个 TCP 端口，是 serverPort+10。
- 集群数据节点对外提供 RESTful 服务占用一个 TCP 端口，是 serverPort+11。
- 集群内数据节点与 Arbitrator 节点之间通讯占用一个 TCP 端口，是 serverPort+12。
因此一个数据节点总的端口范围为 serverPort 到 serverPort+12，总共 13 个 TCP/UDP 端口。确保集群中所有主机在端口 6030-6042 上的 TCP/UDP 协议能够互通。详细的端口情况请参见 [TDengine 2.0 端口说明](/train-faq/faq#port)
**端口配置：**一个数据节点对外的端口由 TDengine 的系统配置参数 serverPort 决定，默认为 6030。
**集群对外连接：**TDengine 集群可以容纳单个、多个甚至几千个数据节点。应用只需要向集群中任何一个数据节点发起连接即可，连接需要提供的网络参数是一数据节点的 End Point（FQDN 加配置的端口号）。通过命令行 CLI 启动应用 taos 时，可以通过选项-h 来指定数据节点的 FQDN，-P 来指定其配置的端口号，如果端口不配置，将采用 TDengine 的系统配置参数 serverPort。
**集群内部通讯：**各个数据节点之间通过 TCP/UDP 进行连接。一个数据节点启动时，将获取 mnode 所在的 dnode 的 EP 信息，然后与系统中的 mnode 建立起连接，交换信息。获取 mnode 的 EP 信息有三步：
**集群内部通讯：**各个数据节点之间通过 TCP 进行连接。一个数据节点启动时，将获取 mnode 所在的 dnode 的 EP 信息，然后与系统中的 mnode 建立起连接，交换信息。获取 mnode 的 EP 信息有三步：
1. 检查 mnodeEpSet.json 文件是否存在，如果不存在或不能正常打开获得 mnode EP 信息，进入第二步；
1. 检查 dnode.json 文件是否存在，如果不存在或不能正常打开获得 mnode EP 信息，进入第二步；
2. 检查系统配置文件 taos.cfg，获取节点配置参数 firstEp、secondEp（这两个参数指定的节点可以是不带 mnode 的普通节点，这样的话，节点被连接时会尝试重定向到 mnode 节点），如果不存在或者 taos.cfg 里没有这两个配置参数，或无效，进入第三步；
3. 将自己的 EP 设为 mnode EP，并独立运行起来。
获取 mnode EP 列表后，数据节点发起连接，如果连接成功，则成功加入进工作的集群，如果不成功，则尝试 mnode EP 列表中的下一个。如果都尝试了，但连接都仍然失败，则休眠几秒后，再进行尝试。
**Mnode 的选择：**TDengine 逻辑上有管理节点，但没有单独的执行代码，服务器侧只有一套执行代码 taosd。那么哪个数据节点会是管理节点呢？这是系统自动决定的，无需任何人工干预。原则如下：一个数据节点启动时，会检查自己的 End Point，并与获取的 mnode EP List 进行比对，如果在其中，该数据节点认为自己应该启动 mnode 模块，成为 mnode。如果自己的 EP 不在 mnode EP List 里，则不启动 mnode 模块。在系统的运行过程中，由于负载均衡、宕机等原因，mnode 有可能迁移至新的 dnode，但一切都是透明的，无需人工干预，配置参数的修改，是 mnode 自己根据资源做出的决定。
**Mnode 的选择：**TDengine 逻辑上有管理节点，但没有单独的执行代码，服务器侧只有一套执行代码 taosd。那么哪个数据节点会是管理节点呢？在集群部署时，第一个数据节点自动成为管理节点。集群中的其他管理节点的创建与删除，由用户通过 SQL 语句完成。
**新数据节点的加入：**系统有了一个数据节点后，就已经成为一个工作的系统。添加新的节点进集群时，有两个步骤，第一步：使用 TDengine CLI 连接到现有工作的数据节点，然后用命令“CREATE DNODE”将新的数据节点的 End Point 添加进去；第二步：在新的数据节点的系统配置参数文件 taos.cfg 里，将 firstEp，secondEp 参数设置为现有集群中任意两个数据节点的 EP 即可。具体添加的详细步骤请见详细的用户手册。这样就把集群一步一步的建立起来。
**重定向：**无论是 dnode 还是 taosc，最先都是要发起与 mnode 的连接，但 mnode 是系统自动创建并维护的，因此对于用户来说，并不知道哪个 dnode 在运行 mnode。TDengine 只要求向系统中任何一个工作的 dnode 发起连接即可。因为任何一个正在运行的 dnode，都维护有目前运行的 mnode EP List。当收到一个来自新启动的 dnode 或 taosc 的连接请求，如果自己不是 mnode，则将 mnode EP List 回复给对方，taosc 或新启动的 dnode 收到这个 list，就重新尝试建立连接。当 mnode EP List 发生改变，通过节点之间的消息交互，各个数据节点就很快获取最新列表，并通知 taosc。
**重定向：**无论是 dnode 还是 taosc，最先都是要发起与 mnode 的连接，由于 mnode 是可以动态调整的，所以对于用户来说，并不知道哪个 dnode 在运行 mnode。TDengine 只要求向系统中任何一个工作的 dnode 发起连接即可。因为任何一个正在运行的 dnode，都维护有目前运行的 mnode EP List。当收到一个来自新启动的 dnode 或 taosc 的连接请求，如果自己不是 mnode，则将 mnode EP List 回复给对方，taosc 或新启动的 dnode 收到这个 list，就重新尝试建立连接。当 mnode EP List 发生改变，通过节点之间的消息交互，各个数据节点就很快获取最新列表，并通知 taosc。
### 一个典型的消息流程
@ -68,15 +66,17 @@ TDengine 分布式架构的逻辑结构图如下：
<center> 图 2 TDengine 典型的操作流程 </center>
1. 应用通过 JDBC 或其他 API 接口发起插入数据的请求。
2. taosc 会检查缓存，看是否保存有该表的 meta data。如果有，直接到第 4 步。如果没有，taosc 将向 mnode 发出 get meta-data 请求。
3. mnode 将该表的 meta-data 返回给 taosc。Meta-data 包含有该表的 schema，而且还有该表所属的 vgroup 信息（vnode ID 以及所在的 dnode 的 End Point，如果副本数为 N，就有 N 组 End Point）。如果 taosc 迟迟得不到 mnode 回应，而且存在多个 mnode，taosc 将向下一个 mnode 发出请求。
4. taosc 向 master vnode 发起插入请求。
5. vnode 插入数据后，给 taosc 一个应答，表示插入成功。如果 taosc 迟迟得不到 vnode 的回应，taosc 会认为该节点已经离线。这种情况下，如果被插入的数据库有多个副本，taosc 将向 vgroup 里下一个 vnode 发出插入请求。
6. taosc 通知 APP，写入成功。
2. taosc 会检查缓存，看是否保存有该表所在数据库的 vgroup-info 信息。如果有，直接到第 4 步。如果没有，taosc 将向 mnode 发出 get vgroup-info 请求。
3. mnode 将该表所在数据库的 vgroup-info 返回给 taosc。Vgroup-info 包含数据库的 vgroup 分布信息（vnode ID 以及所在的 dnode 的 End Point，如果副本数为 N，就有 N 组 End Point），还包含每个 vgroup 中存储数据表的 hash 范围。如果 taosc 迟迟得不到 mnode 回应，而且存在多个 mnode，taosc 将向下一个 mnode 发出请求。
4. taosc 会继续检查缓存，看是否保存有该表的 meta-data。如果有，直接到第 6 步。如果没有，taosc 将向 vnode 发出 get meta-data 请求。
5. vnode 将该表的 meta-data 返回给 taosc。Meta-data 包含有该表的 schema。
6. taosc 向 leader vnode 发起插入请求。
7. vnode 插入数据后，给 taosc 一个应答，表示插入成功。如果 taosc 迟迟得不到 vnode 的回应，taosc 会认为该节点已经离线。这种情况下，如果被插入的数据库有多个副本，taosc 将向 vgroup 里下一个 vnode 发出插入请求。
8. taosc 通知 APP，写入成功。
对于第二和第三步，taosc 启动时，并不知道 mnode 的 End Point，因此会直接向配置的集群对外服务的 End Point 发起请求。如果接收到该请求的 dnode 并没有配置 mnode，该 dnode 会在回复的消息中告知 mnode EP 列表，这样 taosc 会重新向新的 mnode 的 EP 发出获取 meta-data 的请求。
对于第二步，taosc 启动时，并不知道 mnode 的 End Point，因此会直接向配置的集群对外服务的 End Point 发起请求。如果接收到该请求的 dnode 并没有配置 mnode，该 dnode 会在回复的消息中告知 mnode EP 列表，这样 taosc 会重新向新的 mnode 的 EP 发出获取 meta-data 的请求。
对于第四和第五步，没有缓存的情况下，taosc 无法知道虚拟节点组里谁是 master，就假设第一个 vnodeID 就是 master，向它发出请求。如果接收到请求的 vnode 并不是 master，它会在回复中告知谁是 master，这样 taosc 就向建议的 master vnode 发出请求。一旦得到插入成功的回复，taosc 会缓存 master 节点的信息。
对于第四和第六步，没有缓存的情况下，taosc 无法知道虚拟节点组里谁是 leader，就假设第一个 vnodeID 就是 leader，向它发出请求。如果接收到请求的 vnode 并不是 leader，它会在回复中告知谁是 leader，这样 taosc 就向建议的 leader vnode 发出请求。一旦得到插入成功的回复，taosc 会缓存 leader 节点的信息。
上述是插入数据的流程，查询、计算的流程也完全一致。taosc 把这些复杂的流程全部封装屏蔽了，对于应用来说无感知也无需任何特别处理。
@ -89,13 +89,13 @@ TDengine 分布式架构的逻辑结构图如下：
TDengine 存储的数据包括采集的时序数据以及库、表相关的元数据、标签数据等，这些数据具体分为三部分：
- 时序数据：存放于 vnode 里，由 data、head 和 last 三个文件组成，数据量大，查询量取决于应用场景。容许乱序写入，但暂时不支持删除操作，并且仅在 update 参数设置为 1 时允许更新操作。通过采用一个采集点一张表的模型，一个时间段的数据是连续存储，对单张表的写入是简单的追加操作，一次读，可以读到多条记录，这样保证对单个采集点的插入和查询操作，性能达到最优。
- 标签数据：存放于 vnode 里的 meta 文件，支持增删改查四个标准操作。数据量不大，有 N 张表，就有 N 条记录，因此可以全内存存储。如果标签过滤操作很多，查询将十分频繁，因此 TDengine 支持多核多线程并发查询。只要计算资源足够，即使有数千万张表，过滤结果能毫秒级返回。
- 元数据：存放于 mnode 里，包含系统节点、用户、DB、Table Schema 等信息，支持增删改查四个标准操作。这部分数据的量不大，可以全内存保存，而且由于客户端有缓存，查询量也不大。因此目前的设计虽是集中式存储管理，但不会构成性能瓶颈。
- 数据表元数据：包含标签信息和 Table Schema 信息，存放于 vnode 里的 meta 文件，支持增删改查四个标准操作。数据量很大，有 N 张表，就有 N 条记录，因此采用 LRU 存储，支持标签数据的索引。TDengine 支持多核多线程并发查询。只要计算内存足够，元数据全内存存储，千万级别规模的标签数据过滤结果能毫秒级返回。在内存资源不足的情况下，仍然可以支持数千万张表的快速查询。
- 数据库元数据：存放于 mnode 里，包含系统节点、用户、DB、STable Schema 等信息，支持增删改查四个标准操作。这部分数据的量不大，可以全内存保存，而且由于客户端有缓存，查询量也不大。因此目前的设计虽是集中式存储管理，但不会构成性能瓶颈。
与典型的 NoSQL 存储模型相比，TDengine 将标签数据与时序数据完全分离存储，它具有两大优势：
- 能够极大地降低标签数据存储的冗余度：一般的 NoSQL 数据库或时序数据库，采用的 K-V 存储，其中的 Key 包含时间戳、设备 ID、各种标签。每条记录都带有这些重复的内容，浪费存储空间。而且如果应用要在历史数据上增加、修改或删除标签，需要遍历数据，重写一遍，操作成本极其昂贵。
- 能够实现极为高效的多表之间的聚合查询：做多表之间聚合查询时，先把符合标签过滤条件的表查找出来，然后再查找这些表相应的数据块，这样大幅减少要扫描的数据集，从而大幅提高查询效率。而且标签数据采用全内存的结构进行管理和维护，千万级别规模的标签数据查询可以在毫秒级别返回。
- 能够实现极为高效的多表之间的聚合查询：做多表之间聚合查询时，先把符合标签过滤条件的表查找出来，然后再查找这些表相应的数据块，这样大幅减少要扫描的数据集，从而大幅提高查询效率。
### 数据分片
@ -103,11 +103,11 @@ TDengine 存储的数据包括采集的时序数据以及库、表相关的元
vnode（虚拟数据节点）负责为采集的时序数据提供写入、查询和计算功能。为便于负载均衡、数据恢复、支持异构环境，TDengine 将一个数据节点根据其计算和存储资源切分为多个 vnode。这些 vnode 的管理是 TDengine 自动完成的，对应用完全透明。
对于单独一个数据采集点，无论其数据量多大，一个 vnode（或 vgroup，如果副本数大于 1）有足够的计算资源和存储资源来处理（如果每秒生成一条 16 字节的记录，一年产生的原始数据不到 0.5G），因此 TDengine 将一张表（一个数据采集点）的所有数据都存放在一个 vnode 里，而不会让同一个采集点的数据分布到两个或多个 dnode 上。而且一个 vnode 可存储多个数据采集点（表）的数据，一个 vnode 可容纳的表的数目的上限为一百万。设计上，一个 vnode 里所有的表都属于同一个 DB。一个数据节点上，除非特殊配置，一个 DB 拥有的 vnode 数目不会超过系统核的数目。
对于单独一个数据采集点，无论其数据量多大，一个 vnode（或 vgroup，如果副本数大于 1）有足够的计算资源和存储资源来处理（如果每秒生成一条 16 字节的记录，一年产生的原始数据不到 0.5G），因此 TDengine 将一张表（一个数据采集点）的所有数据都存放在一个 vnode 里，而不会让同一个采集点的数据分布到两个或多个 dnode 上。而且一个 vnode 可存储多个数据采集点（表）的数据，一个 vnode 可容纳的表的数目的上限为一百万。设计上，一个 vnode 里所有的表都属于同一个 DB。
创建 DB 时，系统并不会马上分配资源。但当创建一张表时，系统将看是否有已经分配的 vnode，且该 vnode 是否有空余的表空间，如果有，立即在该有空位的 vnode 创建表。如果没有，系统将从集群中，根据当前的负载情况，在一个 dnode 上创建一新的 vnode，然后创建表。如果 DB 有多个副本，系统不是只创建一个 vnode，而是一个 vgroup（虚拟数据节点组）。系统对 vnode 的数目没有任何限制，仅仅受限于物理节点本身的计算和存储资源。
TDengine 3.0 采用 hash 一致性算法，确定每张数据表所在的 vnode。创建 DB 时，系统会立刻分配指定数目的 vnode，并确定每个 vnode 所负责的数据表范围。当创建一张表时，系统根据数据表名计算出所在的 vnodeID，立即在该 vnode 创建表。如果 DB 有多个副本，系统不是只创建一个 vnode，而是一个 vgroup（虚拟数据节点组）。系统对 vnode 的数目没有任何限制，仅仅受限于物理节点本身的计算和存储资源。
每张表的 meta data（包含 schema，标签等）也存放于 vnode 里，而不是集中存放于 mnode，实际上这是对 Meta 数据的分片，这样便于高效并行的进行标签过滤操作。
每张表的 meta data（包含 schema，标签等）也存放于 vnode 里，而不是集中存放于 mnode，实际上这是对 meta 数据的分片，这样便于高效并行的进行标签过滤操作。
### 数据分区
@ -117,77 +117,68 @@ TDengine 除 vnode 分片之外，还对时序数据按照时间段进行分区
### 负载均衡
每个 dnode 都定时向 mnode（虚拟管理节点）报告其状态（包括硬盘空间、内存大小、CPU、网络、虚拟节点个数等），因此 mnode 了解整个集群的状态。基于整体状态，当 mnode 发现某个 dnode 负载过重，它会将 dnode 上的一个或多个 vnode 挪到其他 dnode。在挪动过程中，对外服务继续进行，数据插入、查询和计算操作都不受影响。
每个 dnode 都定时向 mnode（虚拟管理节点）报告其状态（包括硬盘空间、内存大小、CPU、网络、虚拟节点个数等），因此 mnode 了解整个集群的状态。基于整体状态，当 mnode 发现某个 dnode 负载过重，它会将 dnode 上的一个或多个 vnode 挪到其他 dnode。在挪动过程中，对外服务继续进行，数据插入、查询和计算操作都不受影响。负载均衡的触发时间，由用户指定。
如果 mnode 一段时间没有收到 dnode 的状态报告，mnode 会认为这个 dnode 已经离线。如果离线时间超过一定时长（时长由配置参数 offlineThreshold 决定），该 dnode 将被 mnode 强制剔除出集群。该 dnode 上的 vnodes 如果副本数大于 1，系统将自动在其他 dnode 上创建新的副本，以保证数据的副本数。如果该 dnode 上还有 mnode，而且 mnode 的副本数大于 1，系统也将自动在其他 dnode 上创建新的 mnode，以保证 mnode 的副本数。
当新的数据节点被添加进集群，因为新的计算和存储被添加进来，系统也将自动启动负载均衡流程。
负载均衡过程无需任何人工干预，应用也无需重启，将自动连接新的节点，完全透明。
当新的数据节点被添加进集群，因为新的计算和存储被添加进来，用户需要手动触发负载均衡流程，使得系统在最优的情况下运行。
**提示：负载均衡由参数 balance 控制，决定开启/关闭自动负载均衡。**
## 数据写入与复制流程
如果一个数据库有 N 个副本，那一个虚拟节点组就有 N 个虚拟节点，但是只有一个是 master，其他都是 slave。当应用将新的记录写入系统时，只有 master vnode 能接受写的请求。如果 slave vnode 收到写的请求，系统将通知 taosc 需要重新定向。
如果一个数据库有 N 个副本，那一个虚拟节点组就有 N 个虚拟节点，但是只有一个是 leader，其他都是 follower。当应用将新的记录写入系统时，只有 leader vnode 能接受写的请求。如果 follower vnode 收到写的请求，系统将通知 taosc 需要重新定向。
### Master Vnode 写入流程
### Leader Vnode 写入流程
Master Vnode 遵循下面的写入流程：
Leader Vnode 遵循下面的写入流程：
![TDengine Database Master写入流程](./write_master.webp)
![TDengine Database Leader写入流程](./write_leader.webp)
<center> 图 3 TDengine Master 写入流程 </center>
<center> 图 3 TDengine Leader 写入流程 </center>
1. master vnode 收到应用的数据插入请求，验证 OK，进入下一步；
2. 如果系统配置参数 walLevel 大于 0，vnode 将把该请求的原始数据包写入数据库日志文件 WAL。如果 walLevel 设置为 2，而且 fsync 设置为 0，TDengine 还将 WAL 数据立即落盘，以保证即使宕机，也能从数据库日志文件中恢复数据，避免数据的丢失；
3. 如果有多个副本，vnode 将把数据包转发给同一虚拟节点组内的 slave vnodes，该转发包带有数据的版本号（version）；
4. 写入内存，并将记录加入到 skip list；
5. master vnode 返回确认信息给应用，表示写入成功；
6. 如果第 2、3、4 步中任何一步失败，将直接返回错误给应用。
1. leader vnode 收到应用的数据插入请求，验证 OK，进入下一步；
2. vnode 将该请求的原始数据包写入数据库日志文件 WAL。如果 walLevel 设置为 2，而且 fsync 设置为 0，TDengine 还将 WAL 数据立即落盘，以保证即使宕机，也能从数据库日志文件中恢复数据，避免数据的丢失；
3. 如果有多个副本，vnode 将把数据包转发给同一虚拟节点组内的 follower vnodes，该转发包带有数据的版本号（version）；
4. 写入内存，并将记录加入到 skip list。但如果未达成一致，会触发回滚操作；
5. leader vnode 返回确认信息给应用，表示写入成功；
6. 如果第 2、3、4 步中任何一步失败，将直接返回错误给应用；
### Slave Vnode 写入流程
### Follower Vnode 写入流程
对于 slave vnode，写入流程是：
对于 follower vnode，写入流程是：
![TDengine Database Slave 写入流程](./write_slave.webp)
![TDengine Database Follower 写入流程](./write_follower.webp)
<center> 图 4 TDengine Slave 写入流程 </center>
<center> 图 4 TDengine Follower 写入流程 </center>
1. slave vnode 收到 Master vnode 转发了的数据插入请求。检查 last version 是否与 master 一致，如果一致，进入下一步。如果不一致，需要进入同步状态。
2. 如果系统配置参数 walLevel 大于 0，vnode 将把该请求的原始数据包写入数据库日志文件 WAL。如果 walLevel 设置为 2，而且 fsync 设置为 0，TDengine 还将 WAL 数据立即落盘，以保证即使宕机，也能从数据库日志文件中恢复数据，避免数据的丢失。
1. follower vnode 收到 leader vnode 转发了的数据插入请求。
2. vnode 将把该请求的原始数据包写入数据库日志文件 WAL。如果 walLevel 设置为 2，而且 fsync 设置为 0，TDengine 还将 WAL 数据立即落盘，以保证即使宕机，也能从数据库日志文件中恢复数据，避免数据的丢失。
3. 写入内存，更新内存中的 skip list。
与 master vnode 相比，slave vnode 不存在转发环节，也不存在回复确认环节，少了两步。但写内存与 WAL 是完全一样的。
与 leader vnode 相比，follower vnode 不存在转发环节，也不存在回复确认环节，少了两步。但写内存与 WAL 是完全一样的。
### 主从选择
Vnode 会保持一个数据版本号（version），对内存数据进行持久化存储时，对该版本号也进行持久化存储。每个数据更新操作，无论是采集的时序数据还是元数据，这个版本号将增加 1。
一个 vnode 启动时，角色（master、slave）是不定的，数据是处于未同步状态，它需要与虚拟节点组内其他节点建立 TCP 连接，并互相交换 status，其中包括 version 和自己的角色。通过 status 的交换，系统进入选主流程，规则如下：
一个 vnode 启动时，角色（leader、follower）是不定的，数据是处于未同步状态，它需要与虚拟节点组内其他节点建立 TCP 连接，并互相交换 status，按照标准的 raft 一致性算法完成选主。
1. 如果只有一个副本，该副本永远就是 master
2. 所有副本都在线时，版本最高的被选为 master
3. 在线的虚拟节点数过半，而且有虚拟节点是 slave 的话，该虚拟节点自动成为 master
4. 对于 2 和 3，如果多个虚拟节点满足成为 master 的要求，那么虚拟节点组的节点列表里，最前面的选为 master
更多的关于数据复制的流程，请见[《TDengine 2.0 数据复制模块设计》](/tdinternal/replica/)。
更多的关于数据复制的流程，请见[《TDengine 3.0 数据复制模块设计》](/tdinternal/replica/)。
### 同步复制
对于数据一致性要求更高的场景，异步数据复制无法满足要求，因为有极小的概率丢失数据，因此 TDengine 提供同步复制的机制供用户选择。在创建数据库时，除指定副本数 replica 之外，用户还需要指定新的参数 quorum。如果 quorum 大于 1，它表示每次 master 转发给副本时，需要等待 quorum-1 个回复确认，才能通知应用，数据在 slave 已经写入成功。如果在一定的时间内，得不到 quorum-1 个回复确认，master vnode 将返回错误给应用。
对于数据一致性要求更高的场景，异步数据复制提供的最终一致性无法满足要求。因此 TDengine 提供同步复制的机制供用户选择。在创建数据库时，除指定副本数 replica 之外，用户还需要指定新的参数 strict。如果 strict 等于 1，它表示每次 leader 转发给副本时，需要等待半数以上副本达成一致后，才能通知应用，数据在 follower 已经写入成功。如果在一定的时间内，得不到半数以上副本的确认，leader vnode 将返回错误给应用。
采用同步复制，系统的性能会有所下降，而且 latency 会增加。因为元数据要强一致，mnode 之间的数据同步缺省就是采用的同步复制。
## 缓存与持久化
### 缓存
### 时序数据缓存
TDengine 采用时间驱动缓存管理策略（First-In-First-Out，FIFO），又称为写驱动的缓存管理机制。这种策略有别于读驱动的数据缓存模式（Least-Recent-Used，LRU），直接将最近写入的数据保存在系统的缓存中。当缓存达到临界值的时候，将最早的数据批量写入磁盘。一般意义上来说，对于物联网数据的使用，用户最为关心的是刚产生的数据，即当前状态。TDengine 充分利用这一特性，将最近到达的（当前状态）数据保存在缓存中。
TDengine 通过查询函数向用户提供毫秒级的数据获取能力。直接将最近到达的数据保存在缓存中，可以更加快速地响应用户针对最近一条或一批数据的查询分析，整体上提供更快的数据库查询响应能力。从这个意义上来说，**可通过设置合适的配置参数将 TDengine 作为数据缓存来使用，而不需要再部署 Redis 或其他额外的缓存系统**，可有效地简化系统架构，降低运维的成本。需要注意的是，TDengine 重启以后系统的缓存将被清空，之前缓存的数据均会被批量写入磁盘，缓存的数据将不会像专门的 key-value 缓存系统再将之前缓存的数据重新加载到缓存中。
每个 vnode 有自己独立的内存，而且由多个固定大小的内存块组成，不同 vnode 之间完全隔离。数据写入时，类似于日志的写法，数据被顺序追加写入内存，但每个 vnode 维护有自己的 skip list，便于迅速查找。当三分之一以上的内存块写满时，启动落盘操作，而且后续写的操作在新的内存块进行。这样，一个 vnode 里有三分之一内存块是保留有最近的数据的，以达到缓存、快速查找的目的。一个 vnode 的内存块的个数由配置参数 blocks 决定，内存块的大小由配置参数 cache 决定。
每个 vnode 有自己独立的内存，而且由多个固定大小的内存块组成，不同 vnode 之间完全隔离。数据写入时，类似于日志的写法，数据被顺序追加写入内存，但每个 vnode 维护有自己的 skip list，便于迅速查找。当三分之一以上的内存块写满时，启动落盘操作，而且后续写的操作在新的内存块进行。这样，一个 vnode 里有三分之一内存块是保留有最近的数据的，以达到缓存、快速查找的目的。一个 vnode 的内存大小由配置参数 buffer 决定。
### 持久化存储

BIN
docs/zh/21-tdinternal/message.webp
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 13 KiB

After

Width:  |  Height:  |  Size: 16 KiB

BIN
docs/zh/21-tdinternal/structure.webp
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 22 KiB

After

Width:  |  Height:  |  Size: 20 KiB

BIN
docs/zh/21-tdinternal/write_follower.webp Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 22 KiB

BIN
docs/zh/21-tdinternal/write_leader.webp Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 27 KiB

BIN
docs/zh/21-tdinternal/write_master.webp
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 23 KiB

BIN
docs/zh/21-tdinternal/write_slave.webp
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 19 KiB

1
examples/rust

@ -1 +0,0 @@
Subproject commit 7ed7a97715388fa144718764d6bf20f9bfc29a12

2
include/client/taos.h
View File

@ -271,8 +271,6 @@ DLL_EXPORT tmq_res_t tmq_get_res_type(TAOS_RES *res);
DLL_EXPORT int32_t tmq_get_raw(TAOS_RES *res, tmq_raw_data *raw);
DLL_EXPORT int32_t tmq_write_raw(TAOS *taos, tmq_raw_data raw);
DLL_EXPORT int taos_write_raw_block(TAOS *taos, int numOfRows, char *pData, const char* tbname);
DLL_EXPORT void tmq_free_raw(tmq_raw_data raw);
DLL_EXPORT char *tmq_get_json_meta(TAOS_RES *res); // Returning null means error. Returned result need to be freed by tmq_free_json_meta
DLL_EXPORT void tmq_free_json_meta(char* jsonMeta);

1
include/common/tglobal.h
View File

@ -89,7 +89,6 @@ extern uint16_t tsTelemPort;
// query buffer management
extern int32_t tsQueryBufferSize; // maximum allowed usage buffer size in MB for each data node during query processing
extern int64_t tsQueryBufferSizeBytes; // maximum allowed usage buffer size in byte for each data node
extern bool tsRetrieveBlockingModel; // retrieve threads will be blocked
// query client
extern int32_t tsQueryPolicy;

2
include/common/tmsg.h
View File

@ -1404,7 +1404,7 @@ typedef struct STableScanAnalyzeInfo {
uint32_t skipBlocks;
uint32_t filterOutBlocks;
double elapsedTime;
uint64_t filterTime;
double filterTime;
} STableScanAnalyzeInfo;
int32_t tSerializeSExplainRsp(void* buf, int32_t bufLen, SExplainRsp* pRsp);

2
include/libs/executor/executor.h
View File

@ -155,7 +155,7 @@ int32_t qGetQualifiedTableIdList(void* pTableList, const char* tagCond, int32_t
void qProcessRspMsg(void* parent, struct SRpcMsg* pMsg, struct SEpSet* pEpSet);
int32_t qGetExplainExecInfo(qTaskInfo_t tinfo, int32_t* resNum, SExplainExecInfo** pRes);
int32_t qGetExplainExecInfo(qTaskInfo_t tinfo, SArray* pExecInfoList/*,int32_t* resNum, SExplainExecInfo** pRes*/);
int32_t qSerializeTaskStatus(qTaskInfo_t tinfo, char** pOutput, int32_t* len);

10
include/libs/function/functionMgt.h
View File

@ -157,6 +157,13 @@ typedef enum EFunctionType {
FUNCTION_TYPE_UDF = 10000
} EFunctionType;
typedef enum EFuncReturnRows {
FUNC_RETURN_ROWS_NORMAL = 1,
FUNC_RETURN_ROWS_INDEFINITE,
FUNC_RETURN_ROWS_N,
FUNC_RETURN_ROWS_N_MINUS_1
} EFuncReturnRows;
struct SqlFunctionCtx;
struct SResultRowEntryInfo;
struct STimeWindow;
@ -167,6 +174,8 @@ void fmFuncMgtDestroy();
int32_t fmGetFuncInfo(SFunctionNode* pFunc, char* pMsg, int32_t msgLen);
EFuncReturnRows fmGetFuncReturnRows(SFunctionNode* pFunc);
bool fmIsBuiltinFunc(const char* pFunc);
bool fmIsAggFunc(int32_t funcId);
@ -198,6 +207,7 @@ bool fmIsImplicitTsFunc(int32_t funcId);
bool fmIsClientPseudoColumnFunc(int32_t funcId);
bool fmIsMultiRowsFunc(int32_t funcId);
bool fmIsKeepOrderFunc(int32_t funcId);
bool fmIsCumulativeFunc(int32_t funcId);
int32_t fmGetDistMethod(const SFunctionNode* pFunc, SFunctionNode** pPartialFunc, SFunctionNode** pMergeFunc);

6
include/libs/nodes/plannodes.h
View File

@ -96,6 +96,7 @@ typedef struct SScanLogicNode {
bool groupSort;
int8_t cacheLastMode;
bool hasNormalCols; // neither tag column nor primary key tag column
bool sortPrimaryKey;
} SScanLogicNode;
typedef struct SJoinLogicNode {
@ -204,6 +205,7 @@ typedef struct SWindowLogicNode {
int8_t igExpired;
EWindowAlgorithm windowAlgo;
EOrder inputTsOrder;
EOrder outputTsOrder;
} SWindowLogicNode;
typedef struct SFillLogicNode {
@ -212,6 +214,7 @@ typedef struct SFillLogicNode {
SNode* pWStartTs;
SNode* pValues; // SNodeListNode
STimeWindow timeRange;
EOrder inputTsOrder;
} SFillLogicNode;
typedef struct SSortLogicNode {
@ -410,6 +413,8 @@ typedef struct SWinodwPhysiNode {
int8_t triggerType;
int64_t watermark;
int8_t igExpired;
EOrder inputTsOrder;
EOrder outputTsOrder;
} SWinodwPhysiNode;
typedef struct SIntervalPhysiNode {
@ -434,6 +439,7 @@ typedef struct SFillPhysiNode {
SNode* pValues; // SNodeListNode
SNodeList* pTargets;
STimeWindow timeRange;
EOrder inputTsOrder;
} SFillPhysiNode;
typedef struct SMultiTableIntervalPhysiNode {

1
include/libs/nodes/querynodes.h
View File

@ -253,6 +253,7 @@ typedef struct SSelectStmt {
char stmtName[TSDB_TABLE_NAME_LEN];
uint8_t precision;
int32_t selectFuncNum;
int32_t returnRows; // EFuncReturnRows
bool isEmptyResult;
bool isTimeLineResult;
bool isSubquery;

3
packaging/tools/make_install.bat
View File

@ -1,6 +1,7 @@
@echo off
goto %1
:needAdmin
mshta vbscript:createobject("shell.application").shellexecute("%~s0",":hasAdmin","","runas",1)(window.close)&goto :eof
mshta vbscript:createobject("shell.application").shellexecute("%~s0",":hasAdmin","","runas",1)(window.close)&& echo To start/stop TDengine with administrator privileges: sc start/stop taosd &goto :eof
:hasAdmin
cp -f C:\\TDengine\\driver\\taos.dll C:\\Windows\\System32
sc query "taosd" >nul || sc create "taosd" binPath= "C:\\TDengine\\taosd.exe --win_service" start= DEMAND

5
source/client/CMakeLists.txt
View File

@ -20,11 +20,6 @@ target_link_libraries(
)
if(TD_WINDOWS)
set_target_properties(taos
PROPERTIES
LINK_FLAGS
/DEF:${CMAKE_CURRENT_SOURCE_DIR}/src/taos.def
)
INCLUDE_DIRECTORIES(jni/windows)
INCLUDE_DIRECTORIES(jni/windows/win32)
INCLUDE_DIRECTORIES(jni/windows/win32/bridge)

419
source/client/src/clientSml.c
View File

@ -84,32 +84,11 @@
typedef TSDB_SML_PROTOCOL_TYPE SMLProtocolType;
typedef enum {
SCHEMA_ACTION_CREATE_STABLE,
SCHEMA_ACTION_ADD_COLUMN,
SCHEMA_ACTION_ADD_TAG,
SCHEMA_ACTION_CHANGE_COLUMN_SIZE,
SCHEMA_ACTION_CHANGE_TAG_SIZE,
SCHEMA_ACTION_NULL,
SCHEMA_ACTION_COLUMN,
SCHEMA_ACTION_TAG
} ESchemaAction;
typedef struct {
char sTableName[TSDB_TABLE_NAME_LEN];
SArray *tags;
SArray *fields;
} SCreateSTableActionInfo;
typedef struct {
char sTableName[TSDB_TABLE_NAME_LEN];
SSmlKv *field;
} SAlterSTableActionInfo;
typedef struct {
ESchemaAction action;
union {
SCreateSTableActionInfo createSTable;
SAlterSTableActionInfo alterSTable;
};
} SSchemaAction;
typedef struct {
const char *measure;
const char *tags;
@ -226,18 +205,20 @@ static inline bool smlCheckDuplicateKey(const char *key, int32_t keyLen, SHashOb
}
static int32_t smlBuildInvalidDataMsg(SSmlMsgBuf *pBuf, const char *msg1, const char *msg2) {
memset(pBuf->buf, 0, pBuf->len);
if (msg1) strncat(pBuf->buf, msg1, pBuf->len);
int32_t left = pBuf->len - strlen(pBuf->buf);
if (left > 2 && msg2) {
strncat(pBuf->buf, ":", left - 1);
strncat(pBuf->buf, msg2, left - 2);
if(pBuf->buf){
memset(pBuf->buf, 0, pBuf->len);
if (msg1) strncat(pBuf->buf, msg1, pBuf->len);
int32_t left = pBuf->len - strlen(pBuf->buf);
if (left > 2 && msg2) {
strncat(pBuf->buf, ":", left - 1);
strncat(pBuf->buf, msg2, left - 2);
}
}
return TSDB_CODE_SML_INVALID_DATA;
}
static int32_t smlGenerateSchemaAction(SSchema *colField, SHashObj *colHash, SSmlKv *kv, bool isTag,
SSchemaAction *action, bool *actionNeeded, SSmlHandle *info) {
ESchemaAction *action, SSmlHandle *info) {
uint16_t *index = (uint16_t *)taosHashGet(colHash, kv->key, kv->keyLen);
if (index) {
if (colField[*index].type != kv->type) {
@ -251,25 +232,17 @@ static int32_t smlGenerateSchemaAction(SSchema *colField, SHashObj *colHash, SSm
(colField[*index].type == TSDB_DATA_TYPE_NCHAR &&
((colField[*index].bytes - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE < kv->length))) {
if (isTag) {
action->action = SCHEMA_ACTION_CHANGE_TAG_SIZE;
*action = SCHEMA_ACTION_TAG;
} else {
action->action = SCHEMA_ACTION_CHANGE_COLUMN_SIZE;
*action = SCHEMA_ACTION_COLUMN;
}
action->alterSTable.field = kv;
*actionNeeded = true;
}
} else {
if (isTag) {
action->action = SCHEMA_ACTION_ADD_TAG;
*action = SCHEMA_ACTION_TAG;
} else {
action->action = SCHEMA_ACTION_ADD_COLUMN;
*action = SCHEMA_ACTION_COLUMN;
}
action->alterSTable.field = kv;
*actionNeeded = true;
}
if (*actionNeeded) {
uDebug("SML:0x%" PRIx64 " generate schema action. kv->name: %s, action: %d", info->id, kv->key,
action->action);
}
return 0;
}
@ -284,171 +257,25 @@ static int32_t smlFindNearestPowerOf2(int32_t length, uint8_t type) {
} else if (type == TSDB_DATA_TYPE_NCHAR && result > (TSDB_MAX_BINARY_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE){
result = (TSDB_MAX_BINARY_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE;
}
if (type == TSDB_DATA_TYPE_NCHAR){
result = result * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE;
}else if (type == TSDB_DATA_TYPE_BINARY){
result = result + VARSTR_HEADER_SIZE;
}
return result;
}
static int32_t smlBuildColumnDescription(SSmlKv *field, char *buf, int32_t bufSize, int32_t *outBytes) {
uint8_t type = field->type;
char tname[TSDB_TABLE_NAME_LEN] = {0};
memcpy(tname, field->key, field->keyLen);
if (type == TSDB_DATA_TYPE_BINARY || type == TSDB_DATA_TYPE_NCHAR) {
int32_t bytes = smlFindNearestPowerOf2(field->length, type);
int out = snprintf(buf, bufSize, "`%s` %s(%d)", tname, tDataTypes[field->type].name, bytes);
*outBytes = out;
} else {
int out = snprintf(buf, bufSize, "`%s` %s", tname, tDataTypes[type].name);
*outBytes = out;
}
return 0;
}
static int32_t smlApplySchemaAction(SSmlHandle *info, SSchemaAction *action) {
int32_t code = 0;
int32_t outBytes = 0;
char *result = (char *)taosMemoryCalloc(1, TSDB_MAX_ALLOWED_SQL_LEN);
int32_t capacity = TSDB_MAX_ALLOWED_SQL_LEN;
uDebug("SML:0x%" PRIx64 " apply schema action. action: %d", info->id, action->action);
switch (action->action) {
case SCHEMA_ACTION_ADD_COLUMN: {
int n = sprintf(result, "alter stable `%s` add column ", action->alterSTable.sTableName);
smlBuildColumnDescription(action->alterSTable.field, result + n, capacity - n, &outBytes);
TAOS_RES *res = taos_query((TAOS*)&info->taos->id, result); // TODO async doAsyncQuery
code = taos_errno(res);
const char *errStr = taos_errstr(res);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " apply schema action. error: %s", info->id, errStr);
taosMsleep(100);
}
taos_free_result(res);
break;
}
case SCHEMA_ACTION_ADD_TAG: {
int n = sprintf(result, "alter stable `%s` add tag ", action->alterSTable.sTableName);
smlBuildColumnDescription(action->alterSTable.field, result + n, capacity - n, &outBytes);
TAOS_RES *res = taos_query((TAOS*)&info->taos->id, result); // TODO async doAsyncQuery
code = taos_errno(res);
const char *errStr = taos_errstr(res);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " apply schema action. error : %s", info->id, taos_errstr(res));
taosMsleep(100);
}
taos_free_result(res);
break;
}
case SCHEMA_ACTION_CHANGE_COLUMN_SIZE: {
int n = sprintf(result, "alter stable `%s` modify column ", action->alterSTable.sTableName);
smlBuildColumnDescription(action->alterSTable.field, result + n, capacity - n, &outBytes);
TAOS_RES *res = taos_query((TAOS*)&info->taos->id, result); // TODO async doAsyncQuery
code = taos_errno(res);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " apply schema action. error : %s", info->id, taos_errstr(res));
taosMsleep(100);
}
taos_free_result(res);
break;
}
case SCHEMA_ACTION_CHANGE_TAG_SIZE: {
int n = sprintf(result, "alter stable `%s` modify tag ", action->alterSTable.sTableName);
smlBuildColumnDescription(action->alterSTable.field, result + n, capacity - n, &outBytes);
TAOS_RES *res = taos_query((TAOS*)&info->taos->id, result); // TODO async doAsyncQuery
code = taos_errno(res);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " apply schema action. error : %s", info->id, taos_errstr(res));
taosMsleep(100);
}
taos_free_result(res);
break;
}
case SCHEMA_ACTION_CREATE_STABLE: {
int n = sprintf(result, "create stable `%s` (", action->createSTable.sTableName);
char *pos = result + n;
int freeBytes = capacity - n;
SArray *cols = action->createSTable.fields;
for (int i = 0; i < taosArrayGetSize(cols); i++) {
SSmlKv *kv = (SSmlKv *)taosArrayGetP(cols, i);
smlBuildColumnDescription(kv, pos, freeBytes, &outBytes);
pos += outBytes;
freeBytes -= outBytes;
*pos = ',';
++pos;
--freeBytes;
}
--pos;
++freeBytes;
outBytes = snprintf(pos, freeBytes, ") tags (");
pos += outBytes;
freeBytes -= outBytes;
cols = action->createSTable.tags;
for (int i = 0; i < taosArrayGetSize(cols); i++) {
SSmlKv *kv = (SSmlKv *)taosArrayGetP(cols, i);
smlBuildColumnDescription(kv, pos, freeBytes, &outBytes);
pos += outBytes;
freeBytes -= outBytes;
*pos = ',';
++pos;
--freeBytes;
}
if (taosArrayGetSize(cols) == 0) {
outBytes = snprintf(pos, freeBytes, "`%s` %s(%d)", tsSmlTagName, tDataTypes[TSDB_DATA_TYPE_NCHAR].name, 1);
pos += outBytes;
freeBytes -= outBytes;
*pos = ',';
++pos;
--freeBytes;
}
pos--;
++freeBytes;
outBytes = snprintf(pos, freeBytes, ")");
TAOS_RES *res = taos_query((TAOS*)&info->taos->id, result);
code = taos_errno(res);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " apply schema action. error : %s", info->id, taos_errstr(res));
taosMsleep(100);
}
taos_free_result(res);
break;
}
default:
break;
}
taosMemoryFreeClear(result);
if (code != 0) {
uError("SML:0x%" PRIx64 " apply schema action failure. %s", info->id, tstrerror(code));
}
return code;
}
static int32_t smlProcessSchemaAction(SSmlHandle *info, SSchema *schemaField, SHashObj *schemaHash, SArray *cols,
SSchemaAction *action, bool isTag) {
ESchemaAction *action, bool isTag) {
int32_t code = TSDB_CODE_SUCCESS;
for (int j = 0; j < taosArrayGetSize(cols); ++j) {
if(j == 0 && !isTag) continue;
SSmlKv *kv = (SSmlKv *)taosArrayGetP(cols, j);
bool actionNeeded = false;
code = smlGenerateSchemaAction(schemaField, schemaHash, kv, isTag, action, &actionNeeded, info);
code = smlGenerateSchemaAction(schemaField, schemaHash, kv, isTag, action, info);
if (code != TSDB_CODE_SUCCESS) {
return code;
}
if (actionNeeded) {
code = smlApplySchemaAction(info, action);
if (code != TSDB_CODE_SUCCESS) {
return code;
}
}
}
return TSDB_CODE_SUCCESS;
}
@ -475,6 +302,144 @@ static int32_t smlCheckMeta(SSchema *schema, int32_t length, SArray *cols, bool
return 0;
}
static int32_t getBytes(uint8_t type, int32_t length){
if (type == TSDB_DATA_TYPE_BINARY || type == TSDB_DATA_TYPE_NCHAR) {
return smlFindNearestPowerOf2(length, type);
} else {
return tDataTypes[type].bytes;
}
}
//static int32_t smlSendMetaMsg(SSmlHandle *info, SName *pName, SSmlSTableMeta *sTableData,
// int32_t colVer, int32_t tagVer, int8_t source, uint64_t suid){
static int32_t smlSendMetaMsg(SSmlHandle *info, SName *pName, SSmlSTableMeta *sTableData,
STableMeta *pTableMeta, ESchemaAction action){
SRequestObj* pRequest = NULL;
SMCreateStbReq pReq = {0};
int32_t code = TSDB_CODE_SUCCESS;
SCmdMsgInfo pCmdMsg = {0};
code = buildRequest(info->taos->id, "", 0, NULL, false, &pRequest);
if (code != TSDB_CODE_SUCCESS) {
goto end;
}
if (!pRequest->pDb) {
code = TSDB_CODE_PAR_DB_NOT_SPECIFIED;
goto end;
}
if (action == SCHEMA_ACTION_NULL){
pReq.colVer = 1;
pReq.tagVer = 1;
pReq.suid = 0;
pReq.source = TD_REQ_FROM_APP;
} else if (action == SCHEMA_ACTION_TAG){
pReq.colVer = pTableMeta->sversion;
pReq.tagVer = pTableMeta->tversion + 1;
pReq.suid = pTableMeta->uid;
pReq.source = TD_REQ_FROM_TAOX;
} else if (action == SCHEMA_ACTION_COLUMN){
pReq.colVer = pTableMeta->sversion + 1;
pReq.tagVer = pTableMeta->tversion;
pReq.suid = pTableMeta->uid;
pReq.source = TD_REQ_FROM_TAOX;
}
pReq.commentLen = -1;
pReq.igExists = true;
tNameExtractFullName(pName, pReq.name);
if(action == SCHEMA_ACTION_NULL || action == SCHEMA_ACTION_COLUMN){
pReq.numOfColumns = taosArrayGetSize(sTableData->cols);
pReq.pColumns = taosArrayInit(pReq.numOfColumns, sizeof(SField));
for (int i = 0; i < pReq.numOfColumns; i++) {
SSmlKv *kv = (SSmlKv *)taosArrayGetP(sTableData->cols, i);
SField field = {0};
field.type = kv->type;
field.bytes = getBytes(kv->type, kv->length);
memcpy(field.name, kv->key, kv->keyLen);
taosArrayPush(pReq.pColumns, &field);
}
}else if (action == SCHEMA_ACTION_TAG){
pReq.numOfColumns = pTableMeta->tableInfo.numOfColumns;
pReq.pColumns = taosArrayInit(pReq.numOfColumns, sizeof(SField));
for (int i = 0; i < pReq.numOfColumns; i++) {
SSchema *s = &pTableMeta->schema[i];
SField field = {0};
field.type = s->type;
field.bytes = s->bytes;
strcpy(field.name, s->name);
taosArrayPush(pReq.pColumns, &field);
}
}
if(action == SCHEMA_ACTION_NULL || action == SCHEMA_ACTION_TAG){
pReq.numOfTags = taosArrayGetSize(sTableData->tags);
if (pReq.numOfTags == 0){
pReq.numOfTags = 1;
pReq.pTags = taosArrayInit(pReq.numOfTags, sizeof(SField));
SField field = {0};
field.type = TSDB_DATA_TYPE_NCHAR;
field.bytes = 1;
strcpy(field.name, tsSmlTagName);
taosArrayPush(pReq.pTags, &field);
}else{
pReq.pTags = taosArrayInit(pReq.numOfTags, sizeof(SField));
for (int i = 0; i < pReq.numOfTags; i++) {
SSmlKv *kv = (SSmlKv *)taosArrayGetP(sTableData->tags, i);
SField field = {0};
field.type = kv->type;
field.bytes = getBytes(kv->type, kv->length);
memcpy(field.name, kv->key, kv->keyLen);
taosArrayPush(pReq.pTags, &field);
}
}
}else if (action == SCHEMA_ACTION_COLUMN){
pReq.numOfTags = pTableMeta->tableInfo.numOfTags;
pReq.pTags = taosArrayInit(pReq.numOfTags, sizeof(SField));
for (int i = 0; i < pReq.numOfTags; i++) {
SSchema *s = &pTableMeta->schema[i + pTableMeta->tableInfo.numOfColumns];
SField field = {0};
field.type = s->type;
field.bytes = s->bytes;
strcpy(field.name, s->name);
taosArrayPush(pReq.pTags, &field);
}
}
pCmdMsg.epSet = getEpSet_s(&info->taos->pAppInfo->mgmtEp);
pCmdMsg.msgType = TDMT_MND_CREATE_STB;
pCmdMsg.msgLen = tSerializeSMCreateStbReq(NULL, 0, &pReq);
pCmdMsg.pMsg = taosMemoryMalloc(pCmdMsg.msgLen);
if (NULL == pCmdMsg.pMsg) {
tFreeSMCreateStbReq(&pReq);
code = TSDB_CODE_OUT_OF_MEMORY;
goto end;
}
tSerializeSMCreateStbReq(pCmdMsg.pMsg, pCmdMsg.msgLen, &pReq);
SQuery pQuery;
pQuery.execMode = QUERY_EXEC_MODE_RPC;
pQuery.pCmdMsg = &pCmdMsg;
pQuery.msgType = pQuery.pCmdMsg->msgType;
pQuery.stableQuery = true;
launchQueryImpl(pRequest, &pQuery, true, NULL);
if(pRequest->code == TSDB_CODE_SUCCESS){
catalogRemoveTableMeta(info->pCatalog, pName);
}
code = pRequest->code;
taosMemoryFree(pCmdMsg.pMsg);
end:
destroyRequest(pRequest);
tFreeSMCreateStbReq(&pReq);
return code;
}
static int32_t smlModifyDBSchemas(SSmlHandle *info) {
int32_t code = 0;
SName pName = {TSDB_TABLE_NAME_T, info->taos->acctId, {0}, {0}};
@ -500,16 +465,9 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
code = catalogGetSTableMeta(info->pCatalog, &conn, &pName, &pTableMeta);
if (code == TSDB_CODE_PAR_TABLE_NOT_EXIST || code == TSDB_CODE_MND_STB_NOT_EXIST) {
SSchemaAction schemaAction;
schemaAction.action = SCHEMA_ACTION_CREATE_STABLE;
memset(&schemaAction.createSTable, 0, sizeof(SCreateSTableActionInfo));
memcpy(schemaAction.createSTable.sTableName, superTable, superTableLen);
schemaAction.createSTable.tags = sTableData->tags;
schemaAction.createSTable.fields = sTableData->cols;
code = smlApplySchemaAction(info, &schemaAction);
code = smlSendMetaMsg(info, &pName, sTableData, NULL, SCHEMA_ACTION_NULL);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlApplySchemaAction failed. can not create %s", info->id,
schemaAction.createSTable.sTableName);
uError("SML:0x%" PRIx64 " smlSendMetaMsg failed. can not create %s", info->id, superTable);
goto end;
}
info->cost.numOfCreateSTables++;
@ -521,24 +479,42 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
taosHashPut(hashTmp, pTableMeta->schema[i].name, strlen(pTableMeta->schema[i].name), &i, SHORT_BYTES);
}
SSchemaAction schemaAction;
memset(&schemaAction, 0, sizeof(SSchemaAction));
memcpy(schemaAction.createSTable.sTableName, superTable, superTableLen);
code = smlProcessSchemaAction(info, pTableMeta->schema, hashTmp, sTableData->tags, &schemaAction, true);
ESchemaAction action = SCHEMA_ACTION_NULL;
code = smlProcessSchemaAction(info, pTableMeta->schema, hashTmp, sTableData->tags, &action, true);
if (code != TSDB_CODE_SUCCESS) {
taosHashCleanup(hashTmp);
goto end;
}
if (action == SCHEMA_ACTION_TAG){
code = smlSendMetaMsg(info, &pName, sTableData, pTableMeta, action);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlSendMetaMsg failed. can not create %s", info->id, superTable);
goto end;
}
}
code = catalogRefreshTableMeta(info->pCatalog, &conn, &pName, -1);
if (code != TSDB_CODE_SUCCESS) {
goto end;
}
taosHashClear(hashTmp);
for (uint16_t i = 1; i < pTableMeta->tableInfo.numOfColumns; i++) {
taosHashPut(hashTmp, pTableMeta->schema[i].name, strlen(pTableMeta->schema[i].name), &i, SHORT_BYTES);
}
code = smlProcessSchemaAction(info, pTableMeta->schema, hashTmp, sTableData->cols, &schemaAction, false);
action = SCHEMA_ACTION_NULL;
code = smlProcessSchemaAction(info, pTableMeta->schema, hashTmp, sTableData->cols, &action, false);
taosHashCleanup(hashTmp);
if (code != TSDB_CODE_SUCCESS) {
goto end;
}
if (action == SCHEMA_ACTION_COLUMN){
code = smlSendMetaMsg(info, &pName, sTableData, pTableMeta, action);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlSendMetaMsg failed. can not create %s", info->id, superTable);
goto end;
}
}
code = catalogRefreshTableMeta(info->pCatalog, &conn, &pName, -1);
if (code != TSDB_CODE_SUCCESS) {
@ -1504,11 +1480,13 @@ static SSmlHandle* smlBuildSmlInfo(STscObj* pTscObj, SRequestObj* request, SMLPr
}
((SVnodeModifOpStmt *)(info->pQuery->pRoot))->payloadType = PAYLOAD_TYPE_KV;
info->taos = pTscObj;
code = catalogGetHandle(info->taos->pAppInfo->clusterId, &info->pCatalog);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " get catalog error %d", info->id, code);
goto cleanup;
if (pTscObj){
info->taos = pTscObj;
code = catalogGetHandle(info->taos->pAppInfo->clusterId, &info->pCatalog);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " get catalog error %d", info->id, code);
goto cleanup;
}
}
info->precision = precision;
@ -1518,9 +1496,12 @@ static SSmlHandle* smlBuildSmlInfo(STscObj* pTscObj, SRequestObj* request, SMLPr
} else {
info->dataFormat = true;
}
info->pRequest = request;
info->msgBuf.buf = info->pRequest->msgBuf;
info->msgBuf.len = ERROR_MSG_BUF_DEFAULT_SIZE;
if(request){
info->pRequest = request;
info->msgBuf.buf = info->pRequest->msgBuf;
info->msgBuf.len = ERROR_MSG_BUF_DEFAULT_SIZE;
}
info->exec = smlInitHandle(info->pQuery);
info->childTables = taosHashInit(32, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);

81
source/client/src/taos.def
View File

@ -1,81 +0,0 @@
taos_cleanup
taos_options
taos_set_config
taos_init
taos_connect
taos_connect_l
taos_connect_auth
taos_close
taos_data_type
taos_stmt_init
taos_stmt_prepare
taos_stmt_set_tbname_tags
taos_stmt_set_tbname
taos_stmt_set_sub_tbname
taos_stmt_is_insert
taos_stmt_num_params
taos_stmt_get_param
taos_stmt_bind_param
taos_stmt_bind_param_batch
taos_stmt_bind_single_param_batch
taos_stmt_add_batch
taos_stmt_execute
taos_stmt_use_result
taos_stmt_close
taos_stmt_errstr
taos_stmt_affected_rows
taos_stmt_affected_rows_once
taos_query
taos_query_l
taos_fetch_row
taos_result_precision
taos_free_result
taos_field_count
taos_num_fields
taos_affected_rows
taos_fetch_fields
taos_select_db
taos_print_row
taos_stop_query
taos_is_null
taos_is_update_query
taos_fetch_block
taos_fetch_block_s
taos_fetch_raw_block
taos_get_column_data_offset
taos_validate_sql
taos_reset_current_db
taos_fetch_lengths
taos_result_block
taos_get_server_info
taos_get_client_info
taos_errstr
taos_errno
taos_query_a
taos_fetch_rows_a
taos_subscribe
taos_consume
taos_unsubscribe
taos_load_table_info
taos_schemaless_insert
tmq_list_new
tmq_list_append
tmq_list_destroy
tmq_list_get_size
tmq_list_to_c_array
tmq_consumer_new
tmq_err2str
tmq_subscribe
tmq_unsubscribe
tmq_subscription
tmq_consumer_poll
tmq_consumer_close
tmq_commit
tmq_conf_new
tmq_conf_set
tmq_conf_destroy
tmq_conf_set_offset_commit_cb
tmq_get_topic_name
tmq_get_vgroup_id
tmq_create_stream
taos_check_server_status

26
source/client/test/clientTests.cpp
View File

@ -123,7 +123,7 @@ void createNewTable(TAOS* pConn, int32_t index) {
}
taos_free_result(pRes);
for(int32_t i = 0; i < 1000; i += 20) {
for(int32_t i = 0; i < 100000; i += 20) {
char sql[1024] = {0};
sprintf(sql,
"insert into tu%d values(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
@ -154,7 +154,7 @@ TEST(testCase, driverInit_Test) {
}
TEST(testCase, connect_Test) {
// taos_options(TSDB_OPTION_CONFIGDIR, "/home/ubuntu/first/cfg");
taos_options(TSDB_OPTION_CONFIGDIR, "/home/lisa/Documents/workspace/tdengine/sim/dnode1/cfg");
TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
if (pConn == NULL) {
@ -501,7 +501,6 @@ TEST(testCase, show_vgroup_Test) {
taos_close(pConn);
}
TEST(testCase, create_multiple_tables) {
TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
ASSERT_NE(pConn, nullptr);
@ -665,6 +664,7 @@ TEST(testCase, insert_test) {
taos_free_result(pRes);
taos_close(pConn);
}
#endif
TEST(testCase, projection_query_tables) {
TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
@ -697,7 +697,7 @@ TEST(testCase, projection_query_tables) {
}
taos_free_result(pRes);
for(int32_t i = 0; i < 100; ++i) {
for(int32_t i = 0; i < 1; ++i) {
printf("create table :%d\n", i);
createNewTable(pConn, i);
}
@ -723,6 +723,7 @@ TEST(testCase, projection_query_tables) {
taos_close(pConn);
}
#if 0
TEST(testCase, projection_query_stables) {
TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
ASSERT_NE(pConn, nullptr);
@ -820,21 +821,8 @@ TEST(testCase, async_api_test) {
getchar();
taos_close(pConn);
}
#endif
TEST(testCase, update_test) {
SInterval interval = {0};
interval.offset = 8000;
interval.interval = 10000;
interval.sliding = 4000;
interval.intervalUnit = 's';
interval.offsetUnit = 's';
interval.slidingUnit = 's';
// STimeWindow w = getAlignQueryTimeWindow(&interval, 0, 1630000000000);
STimeWindow w = getAlignQueryTimeWindow(&interval, 0, 1629999999999);
TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
ASSERT_NE(pConn, nullptr);
@ -869,4 +857,8 @@ TEST(testCase, update_test) {
taos_free_result(pRes);
}
}
#endif
#pragma GCC diagnostic pop

927
source/client/test/smlTest.cpp
View File

File diff suppressed because one or more lines are too long

61
source/common/src/tdatablock.c
View File

@ -16,65 +16,9 @@
#define _DEFAULT_SOURCE
#include "tdatablock.h"
#include "tcompare.h"
#include "tglobal.h"
#include "tlog.h"
#include "tname.h"
int32_t taosGetFqdnPortFromEp(const char* ep, SEp* pEp) {
pEp->port = 0;
strcpy(pEp->fqdn, ep);
char* temp = strchr(pEp->fqdn, ':');
if (temp) {
*temp = 0;
pEp->port = atoi(temp + 1);
}
if (pEp->port == 0) {
pEp->port = tsServerPort;
}
return 0;
}
void addEpIntoEpSet(SEpSet* pEpSet, const char* fqdn, uint16_t port) {
if (pEpSet == NULL || fqdn == NULL || strlen(fqdn) == 0) {
return;
}
int32_t index = pEpSet->numOfEps;
tstrncpy(pEpSet->eps[index].fqdn, fqdn, tListLen(pEpSet->eps[index].fqdn));
pEpSet->eps[index].port = port;
pEpSet->numOfEps += 1;
}
bool isEpsetEqual(const SEpSet* s1, const SEpSet* s2) {
if (s1->numOfEps != s2->numOfEps || s1->inUse != s2->inUse) {
return false;
}
for (int32_t i = 0; i < s1->numOfEps; i++) {
if (s1->eps[i].port != s2->eps[i].port || strncmp(s1->eps[i].fqdn, s2->eps[i].fqdn, TSDB_FQDN_LEN) != 0)
return false;
}
return true;
}
void updateEpSet_s(SCorEpSet* pEpSet, SEpSet* pNewEpSet) {
taosCorBeginWrite(&pEpSet->version);
pEpSet->epSet = *pNewEpSet;
taosCorEndWrite(&pEpSet->version);
}
SEpSet getEpSet_s(SCorEpSet* pEpSet) {
SEpSet ep = {0};
taosCorBeginRead(&pEpSet->version);
ep = pEpSet->epSet;
taosCorEndRead(&pEpSet->version);
return ep;
}
int32_t colDataGetLength(const SColumnInfoData* pColumnInfoData, int32_t numOfRows) {
ASSERT(pColumnInfoData != NULL);
if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
@ -1713,8 +1657,9 @@ void blockDebugShowDataBlocks(const SArray* dataBlocks, const char* flag) {
size_t numOfCols = taosArrayGetSize(pDataBlock->pDataBlock);
int32_t rows = pDataBlock->info.rows;
printf("%s |block type %d |child id %d|group id %" PRIu64 "\n", flag, (int32_t)pDataBlock->info.type,
pDataBlock->info.childId, pDataBlock->info.groupId);
printf("%s |block ver %" PRIi64 " |block type %d |child id %d|group id %" PRIu64 "\n", flag,
pDataBlock->info.version, (int32_t)pDataBlock->info.type, pDataBlock->info.childId,
pDataBlock->info.groupId);
for (int32_t j = 0; j < rows; j++) {
printf("%s |", flag);
for (int32_t k = 0; k < numOfCols; k++) {

16
source/common/src/tglobal.c
View File

@ -124,9 +124,6 @@ int32_t tsMinIntervalTime = 1;
int32_t tsQueryBufferSize = -1;
int64_t tsQueryBufferSizeBytes = -1;
// in retrieve blocking model, the retrieve threads will wait for the completion of the query processing.
bool tsRetrieveBlockingModel = false;
// tsdb config
// For backward compatibility
bool tsdbForceKeepFile = false;
@ -296,6 +293,7 @@ static int32_t taosAddServerLogCfg(SConfig *pCfg) {
if (cfgAddInt32(pCfg, "smaDebugFlag", smaDebugFlag, 0, 255, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "idxDebugFlag", idxDebugFlag, 0, 255, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "tdbDebugFlag", tdbDebugFlag, 0, 255, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "metaDebugFlag", metaDebugFlag, 0, 255, 0) != 0) return -1;
return 0;
}
@ -362,7 +360,6 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
if (cfgAddInt32(pCfg, "maxNumOfDistinctRes", tsMaxNumOfDistinctResults, 10 * 10000, 10000 * 10000, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "countAlwaysReturnValue", tsCountAlwaysReturnValue, 0, 1, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "queryBufferSize", tsQueryBufferSize, -1, 500000000000, 0) != 0) return -1;
if (cfgAddBool(pCfg, "retrieveBlockingModel", tsRetrieveBlockingModel, 0) != 0) return -1;
if (cfgAddBool(pCfg, "printAuth", tsPrintAuth, 0) != 0) return -1;
if (cfgAddInt32(pCfg, "multiProcess", tsMultiProcess, 0, 2, 0) != 0) return -1;
@ -476,6 +473,7 @@ static void taosSetServerLogCfg(SConfig *pCfg) {
smaDebugFlag = cfgGetItem(pCfg, "smaDebugFlag")->i32;
idxDebugFlag = cfgGetItem(pCfg, "idxDebugFlag")->i32;
tdbDebugFlag = cfgGetItem(pCfg, "tdbDebugFlag")->i32;
metaDebugFlag = cfgGetItem(pCfg, "metaDebugFlag")->i32;
}
static int32_t taosSetClientCfg(SConfig *pCfg) {
@ -547,7 +545,6 @@ static int32_t taosSetServerCfg(SConfig *pCfg) {
tsMaxNumOfDistinctResults = cfgGetItem(pCfg, "maxNumOfDistinctRes")->i32;
tsCountAlwaysReturnValue = cfgGetItem(pCfg, "countAlwaysReturnValue")->i32;
tsQueryBufferSize = cfgGetItem(pCfg, "queryBufferSize")->i32;
tsRetrieveBlockingModel = cfgGetItem(pCfg, "retrieveBlockingModel")->bval;
tsPrintAuth = cfgGetItem(pCfg, "printAuth")->bval;
tsMultiProcess = cfgGetItem(pCfg, "multiProcess")->bval;
@ -832,9 +829,7 @@ int32_t taosSetCfg(SConfig *pCfg, char *name) {
break;
}
case 'r': {
if (strcasecmp("retrieveBlockingModel", name) == 0) {
tsRetrieveBlockingModel = cfgGetItem(pCfg, "retrieveBlockingModel")->bval;
} else if (strcasecmp("rpcQueueMemoryAllowed", name) == 0) {
if (strcasecmp("rpcQueueMemoryAllowed", name) == 0) {
tsRpcQueueMemoryAllowed = cfgGetItem(pCfg, "rpcQueueMemoryAllowed")->i64;
} else if (strcasecmp("rpcDebugFlag", name) == 0) {
rpcDebugFlag = cfgGetItem(pCfg, "rpcDebugFlag")->i32;
@ -1100,12 +1095,12 @@ void taosCfgDynamicOptions(const char *option, const char *value) {
const char *options[] = {
"dDebugFlag", "vDebugFlag", "mDebugFlag", "wDebugFlag", "sDebugFlag", "tsdbDebugFlag",
"tqDebugFlag", "fsDebugFlag", "udfDebugFlag", "smaDebugFlag", "idxDebugFlag", "tdbDebugFlag",
"tmrDebugFlag", "uDebugFlag", "smaDebugFlag", "rpcDebugFlag", "qDebugFlag",
"tmrDebugFlag", "uDebugFlag", "smaDebugFlag", "rpcDebugFlag", "qDebugFlag", "metaDebugFlag",
};
int32_t *optionVars[] = {
&dDebugFlag, &vDebugFlag, &mDebugFlag, &wDebugFlag, &sDebugFlag, &tsdbDebugFlag,
&tqDebugFlag, &fsDebugFlag, &udfDebugFlag, &smaDebugFlag, &idxDebugFlag, &tdbDebugFlag,
&tmrDebugFlag, &uDebugFlag, &smaDebugFlag, &rpcDebugFlag, &qDebugFlag,
&tmrDebugFlag, &uDebugFlag, &smaDebugFlag, &rpcDebugFlag, &qDebugFlag, &metaDebugFlag,
};
int32_t optionSize = tListLen(options);
@ -1152,5 +1147,6 @@ void taosSetAllDebugFlag(int32_t flag) {
taosSetDebugFlag(&smaDebugFlag, "smaDebugFlag", flag);
taosSetDebugFlag(&idxDebugFlag, "idxDebugFlag", flag);
taosSetDebugFlag(&tdbDebugFlag, "tdbDebugFlag", flag);
taosSetDebugFlag(&metaDebugFlag, "metaDebugFlag", flag);
uInfo("all debug flag are set to %d", flag);
}

77
source/common/src/tmisce.c Normal file
View File

@ -0,0 +1,77 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#define _DEFAULT_SOURCE
#include "tdatablock.h"
#include "tglobal.h"
#include "tlog.h"
#include "tname.h"
int32_t taosGetFqdnPortFromEp(const char* ep, SEp* pEp) {
pEp->port = 0;
strcpy(pEp->fqdn, ep);
char* temp = strchr(pEp->fqdn, ':');
if (temp) {
*temp = 0;
pEp->port = atoi(temp + 1);
}
if (pEp->port == 0) {
pEp->port = tsServerPort;
}
return 0;
}
void addEpIntoEpSet(SEpSet* pEpSet, const char* fqdn, uint16_t port) {
if (pEpSet == NULL || fqdn == NULL || strlen(fqdn) == 0) {
return;
}
int32_t index = pEpSet->numOfEps;
tstrncpy(pEpSet->eps[index].fqdn, fqdn, tListLen(pEpSet->eps[index].fqdn));
pEpSet->eps[index].port = port;
pEpSet->numOfEps += 1;
}
bool isEpsetEqual(const SEpSet* s1, const SEpSet* s2) {
if (s1->numOfEps != s2->numOfEps || s1->inUse != s2->inUse) {
return false;
}
for (int32_t i = 0; i < s1->numOfEps; i++) {
if (s1->eps[i].port != s2->eps[i].port || strncmp(s1->eps[i].fqdn, s2->eps[i].fqdn, TSDB_FQDN_LEN) != 0)
return false;
}
return true;
}
void updateEpSet_s(SCorEpSet* pEpSet, SEpSet* pNewEpSet) {
taosCorBeginWrite(&pEpSet->version);
pEpSet->epSet = *pNewEpSet;
taosCorEndWrite(&pEpSet->version);
}
SEpSet getEpSet_s(SCorEpSet* pEpSet) {
SEpSet ep = {0};
taosCorBeginRead(&pEpSet->version);
ep = pEpSet->epSet;
taosCorEndRead(&pEpSet->version);
return ep;
}

28
source/common/src/tname.c
View File

@ -20,34 +20,6 @@
#define VALID_NAME_TYPE(x) ((x) == TSDB_DB_NAME_T || (x) == TSDB_TABLE_NAME_T)
bool tscValidateTableNameLength(size_t len) { return len < TSDB_TABLE_NAME_LEN; }
#if 0
// TODO refactor
SColumnFilterInfo* tFilterInfoDup(const SColumnFilterInfo* src, int32_t numOfFilters) {
if (numOfFilters == 0 || src == NULL) {
assert(src == NULL);
return NULL;
}
SColumnFilterInfo* pFilter = taosMemoryCalloc(1, numOfFilters * sizeof(SColumnFilterInfo));
memcpy(pFilter, src, sizeof(SColumnFilterInfo) * numOfFilters);
for (int32_t j = 0; j < numOfFilters; ++j) {
if (pFilter[j].filterstr) {
size_t len = (size_t) pFilter[j].len + 1 * TSDB_NCHAR_SIZE;
pFilter[j].pz = (int64_t) taosMemoryCalloc(1, len);
memcpy((char*)pFilter[j].pz, (char*)src[j].pz, (size_t) pFilter[j].len);
}
}
assert(src->filterstr == 0 || src->filterstr == 1);
assert(!(src->lowerRelOptr == 0 && src->upperRelOptr == 0));
return pFilter;
}
#endif
#if 0
int64_t taosGetIntervalStartTimestamp(int64_t startTime, int64_t slidingTime, int64_t intervalTime, char timeUnit, int16_t precision) {
if (slidingTime == 0) {

19
source/dnode/mgmt/exe/dmMain.c
View File

@ -27,6 +27,9 @@
#define DM_VERSION "Print program version."
#define DM_EMAIL "<support@taosdata.com>"
static struct {
#ifdef WINDOWS
bool winServiceMode;
#endif
bool dumpConfig;
bool generateGrant;
bool printAuth;
@ -93,6 +96,10 @@ static int32_t dmParseArgs(int32_t argc, char const *argv[]) {
global.dumpConfig = true;
} else if (strcmp(argv[i], "-V") == 0) {
global.printVersion = true;
#ifdef WINDOWS
} else if (strcmp(argv[i], "--win_service") == 0) {
global.winServiceMode = true;
#endif
} else if (strcmp(argv[i], "-e") == 0) {
global.envCmd[cmdEnvIndex] = argv[++i];
cmdEnvIndex++;
@ -169,6 +176,18 @@ int main(int argc, char const *argv[]) {
return -1;
}
#ifdef WINDOWS
int mainWindows(int argc,char** argv);
if (global.winServiceMode) {
stratWindowsService(mainWindows);
} else {
return mainWindows(argc, argv);
}
return 0;
}
int mainWindows(int argc,char** argv) {
#endif
if (global.generateGrant) {
dmGenerateGrant();
taosCleanupArgs();

6
source/dnode/mnode/impl/src/mndDnode.c
View File

@ -788,9 +788,9 @@ _OVER:
static int32_t mndProcessConfigDnodeReq(SRpcMsg *pReq) {
SMnode *pMnode = pReq->info.node;
const char *options[] = {
"debugFlag", "dDebugFlag", "vDebugFlag", "mDebugFlag", "wDebugFlag", "sDebugFlag",
"tsdbDebugFlag", "tqDebugFlag", "fsDebugFlag", "udfDebugFlag", "smaDebugFlag", "idxDebugFlag",
"tdbDebugFlag", "tmrDebugFlag", "uDebugFlag", "smaDebugFlag", "rpcDebugFlag", "qDebugFlag",
"debugFlag", "dDebugFlag", "vDebugFlag", "mDebugFlag", "wDebugFlag", "sDebugFlag", "tsdbDebugFlag",
"tqDebugFlag", "fsDebugFlag", "udfDebugFlag", "smaDebugFlag", "idxDebugFlag", "tdbDebugFlag", "tmrDebugFlag",
"uDebugFlag", "smaDebugFlag", "rpcDebugFlag", "qDebugFlag", "metaDebugFlag",
};
int32_t optionSize = tListLen(options);

1
source/dnode/vnode/CMakeLists.txt
View File

@ -31,6 +31,7 @@ target_sources(
"src/sma/smaOpen.c"
"src/sma/smaCommit.c"
"src/sma/smaRollup.c"
"src/sma/smaSnapshot.c"
"src/sma/smaTimeRange.c"
# tsdb

3
source/dnode/vnode/src/inc/sma.h
View File

@ -209,6 +209,9 @@ int32_t tdProcessTSmaGetDaysImpl(SVnodeCfg *pCfg, void *pCont, uint32_t contLen,
// smaFileUtil ================
typedef struct SQTaskFReader SQTaskFReader;
typedef struct SQTaskFWriter SQTaskFWriter;
#define TD_FILE_HEAD_SIZE 512
typedef struct STFInfo STFInfo;

21
source/dnode/vnode/src/inc/tsdb.h
View File

@ -97,7 +97,6 @@ int32_t tRowMergerGetRow(SRowMerger *pMerger, STSRow **ppRow);
// TABLEID
int32_t tTABLEIDCmprFn(const void *p1, const void *p2);
// TSDBKEY
int32_t tsdbKeyCmprFn(const void *p1, const void *p2);
#define MIN_TSDBKEY(KEY1, KEY2) ((tsdbKeyCmprFn(&(KEY1), &(KEY2)) < 0) ? (KEY1) : (KEY2))
#define MAX_TSDBKEY(KEY1, KEY2) ((tsdbKeyCmprFn(&(KEY1), &(KEY2)) > 0) ? (KEY1) : (KEY2))
// SBlockCol
@ -558,6 +557,26 @@ struct STsdbReadSnap {
STsdbFS fs;
};
// ========== inline functions ==========
static FORCE_INLINE int32_t tsdbKeyCmprFn(const void *p1, const void *p2) {
TSDBKEY *pKey1 = (TSDBKEY *)p1;
TSDBKEY *pKey2 = (TSDBKEY *)p2;
if (pKey1->ts < pKey2->ts) {
return -1;
} else if (pKey1->ts > pKey2->ts) {
return 1;
}
if (pKey1->version < pKey2->version) {
return -1;
} else if (pKey1->version > pKey2->version) {
return 1;
}
return 0;
}
#ifdef __cplusplus
}
#endif

21
source/dnode/vnode/src/inc/vnodeInt.h
View File

@ -62,6 +62,8 @@ typedef struct SMetaSnapReader SMetaSnapReader;
typedef struct SMetaSnapWriter SMetaSnapWriter;
typedef struct STsdbSnapReader STsdbSnapReader;
typedef struct STsdbSnapWriter STsdbSnapWriter;
typedef struct SRsmaSnapReader SRsmaSnapReader;
typedef struct SRsmaSnapWriter SRsmaSnapWriter;
typedef struct SSnapDataHdr SSnapDataHdr;
#define VNODE_META_DIR "meta"
@ -196,13 +198,21 @@ int32_t metaSnapWriterOpen(SMeta* pMeta, int64_t sver, int64_t ever, SMetaSnapWr
int32_t metaSnapWrite(SMetaSnapWriter* pWriter, uint8_t* pData, uint32_t nData);
int32_t metaSnapWriterClose(SMetaSnapWriter** ppWriter, int8_t rollback);
// STsdbSnapReader ========================================
int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapReader** ppReader);
int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, int8_t type, STsdbSnapReader** ppReader);
int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader);
int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData);
// STsdbSnapWriter ========================================
int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWriter** ppWriter);
int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData);
int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback);
// SRsmaSnapReader ========================================
int32_t rsmaSnapReaderOpen(SSma* pSma, int64_t sver, int64_t ever, SRsmaSnapReader** ppReader);
int32_t rsmaSnapReaderClose(SRsmaSnapReader** ppReader);
int32_t rsmaSnapRead(SRsmaSnapReader* pReader, uint8_t** ppData);
// SRsmaSnapWriter ========================================
int32_t rsmaSnapWriterOpen(SSma* pSma, int64_t sver, int64_t ever, SRsmaSnapWriter** ppWriter);
int32_t rsmaSnapWrite(SRsmaSnapWriter* pWriter, uint8_t* pData, uint32_t nData);
int32_t rsmaSnapWriterClose(SRsmaSnapWriter** ppWriter, int8_t rollback);
typedef struct {
int8_t streamType; // sma or other
@ -314,6 +324,15 @@ struct SSma {
// sma
void smaHandleRes(void* pVnode, int64_t smaId, const SArray* data);
enum {
SNAP_DATA_META = 0,
SNAP_DATA_TSDB = 1,
SNAP_DATA_DEL = 2,
SNAP_DATA_RSMA1 = 3,
SNAP_DATA_RSMA2 = 4,
SNAP_DATA_QTASK = 5,
};
struct SSnapDataHdr {
int8_t type;
int64_t index;

12
source/dnode/vnode/src/meta/metaOpen.c
View File

@ -183,11 +183,11 @@ int metaClose(SMeta *pMeta) {
int32_t metaRLock(SMeta *pMeta) {
int32_t ret = 0;
metaDebug("meta rlock %p B", &pMeta->lock);
metaTrace("meta rlock %p B", &pMeta->lock);
ret = taosThreadRwlockRdlock(&pMeta->lock);
metaDebug("meta rlock %p E", &pMeta->lock);
metaTrace("meta rlock %p E", &pMeta->lock);
return ret;
}
@ -195,11 +195,11 @@ int32_t metaRLock(SMeta *pMeta) {
int32_t metaWLock(SMeta *pMeta) {
int32_t ret = 0;
metaDebug("meta wlock %p B", &pMeta->lock);
metaTrace("meta wlock %p B", &pMeta->lock);
ret = taosThreadRwlockWrlock(&pMeta->lock);
metaDebug("meta wlock %p E", &pMeta->lock);
metaTrace("meta wlock %p E", &pMeta->lock);
return ret;
}
@ -207,11 +207,11 @@ int32_t metaWLock(SMeta *pMeta) {
int32_t metaULock(SMeta *pMeta) {
int32_t ret = 0;
metaDebug("meta ulock %p B", &pMeta->lock);
metaTrace("meta ulock %p B", &pMeta->lock);
ret = taosThreadRwlockUnlock(&pMeta->lock);
metaDebug("meta ulock %p E", &pMeta->lock);
metaTrace("meta ulock %p E", &pMeta->lock);
return ret;
}

4
source/dnode/vnode/src/meta/metaSnapshot.c
View File

@ -109,7 +109,7 @@ int32_t metaSnapRead(SMetaSnapReader* pReader, uint8_t** ppData) {
}
SSnapDataHdr* pHdr = (SSnapDataHdr*)(*ppData);
pHdr->type = 0; // TODO: use macro
pHdr->type = SNAP_DATA_META;
pHdr->size = nData;
memcpy(pHdr->data, pData, nData);
@ -145,6 +145,8 @@ int32_t metaSnapWriterOpen(SMeta* pMeta, int64_t sver, int64_t ever, SMetaSnapWr
pWriter->sver = sver;
pWriter->ever = ever;
metaBegin(pMeta);
*ppWriter = pWriter;
return code;

12
source/dnode/vnode/src/sma/smaSnapshot.c
View File

@ -49,7 +49,8 @@ int32_t rsmaSnapReaderOpen(SSma* pSma, int64_t sver, int64_t ever, SRsmaSnapRead
for (int32_t i = 0; i < TSDB_RETENTION_L2; ++i) {
if (pSma->pRSmaTsdb[i]) {
code = tsdbSnapReaderOpen(pSma->pRSmaTsdb[i], sver, ever, &pReader->pDataReader[i]);
code = tsdbSnapReaderOpen(pSma->pRSmaTsdb[i], sver, ever, i == 0 ? SNAP_DATA_RSMA1 : SNAP_DATA_RSMA2,
&pReader->pDataReader[i]);
if (code < 0) {
goto _err;
}
@ -221,10 +222,9 @@ int32_t rsmaSnapWriterClose(SRsmaSnapWriter** ppWriter, int8_t rollback) {
}
}
smaInfo("vgId:%d vnode snapshot rsma writer close succeed", SMA_VID(pWriter->pSma));
taosMemoryFree(pWriter);
*ppWriter = NULL;
smaInfo("vgId:%d vnode snapshot rsma writer close succeed", SMA_VID(pWriter->pSma));
return code;
_err:
@ -245,15 +245,17 @@ int32_t rsmaSnapWrite(SRsmaSnapWriter* pWriter, uint8_t* pData, uint32_t nData)
code = tsdbSnapWrite(pWriter->pDataWriter[1], pData, nData);
} else if (pHdr->type == SNAP_DATA_QTASK) {
code = rsmaSnapWriteQTaskInfo(pWriter, pData, nData);
} else {
ASSERT(0);
}
if (code < 0) goto _err;
_exit:
smaInfo("vgId:%d rsma snapshot write for data %" PRIi8 " succeed", SMA_VID(pWriter->pSma), pHdr->type);
smaInfo("vgId:%d rsma snapshot write for data type %" PRIi8 " succeed", SMA_VID(pWriter->pSma), pHdr->type);
return code;
_err:
smaError("vgId:%d rsma snapshot write for data %" PRIi8 " failed since %s", SMA_VID(pWriter->pSma), pHdr->type,
smaError("vgId:%d rsma snapshot write for data type %" PRIi8 " failed since %s", SMA_VID(pWriter->pSma), pHdr->type,
tstrerror(code));
return code;
}

2
source/dnode/vnode/src/tsdb/tsdbCache.c
View File

@ -175,7 +175,7 @@ int32_t tsdbCacheInsertLastrow(SLRUCache *pCache, STsdb *pTsdb, tb_uid_t uid, ST
cacheRow = (STSRow *)taosLRUCacheValue(pCache, h);
if (row->ts >= cacheRow->ts) {
if (row->ts == cacheRow->ts) {
STSRow *mergedRow;
STSRow *mergedRow = NULL;
SRowMerger merger = {0};
STSchema *pTSchema = metaGetTbTSchema(pTsdb->pVnode->pMeta, uid, -1);

6
source/dnode/vnode/src/tsdb/tsdbCommit.c
View File

@ -307,7 +307,11 @@ static int32_t tsdbCommitFileDataStart(SCommitter *pCommitter) {
fLast = (SLastFile){.commitID = pCommitter->commitID, .size = 0};
fSma = *pRSet->pSmaF;
} else {
wSet.diskId = (SDiskID){.level = 0, .id = 0};
SDiskID did = {0};
tfsAllocDisk(pTsdb->pVnode->pTfs, 0, &did);
wSet.diskId = did;
wSet.fid = pCommitter->commitFid;
fHead = (SHeadFile){.commitID = pCommitter->commitID, .offset = 0, .size = 0};
fData = (SDataFile){.commitID = pCommitter->commitID, .size = 0};

98
source/dnode/vnode/src/tsdb/tsdbRead.c
View File

@ -145,7 +145,8 @@ static int32_t doMergeRowsInFileBlocks(SBlockData* pBlockData, STableBlockScanI
SRowMerger* pMerger);
static int32_t doMergeRowsInBuf(SIterInfo* pIter, int64_t ts, SArray* pDelList, SRowMerger* pMerger,
STsdbReader* pReader);
static int32_t doAppendOneRow(SSDataBlock* pBlock, STsdbReader* pReader, STSRow* pTSRow);
static int32_t doAppendRowFromTSRow(SSDataBlock* pBlock, STsdbReader* pReader, STSRow* pTSRow);
static int32_t doAppendRowFromBlock(SSDataBlock* pResBlock, STsdbReader* pReader, SBlockData* pBlockData, int32_t rowIndex);
static void setComposedBlockFlag(STsdbReader* pReader, bool composed);
static void updateSchema(TSDBROW* pRow, uint64_t uid, STsdbReader* pReader);
static bool hasBeenDropped(const SArray* pDelList, int32_t* index, TSDBKEY* pKey, int32_t order);
@ -691,16 +692,13 @@ static int32_t copyBlockDataToSDataBlock(STsdbReader* pReader, STableBlockScanIn
SFileDataBlockInfo* pFBlock = getCurrentBlockInfo(pBlockIter);
SBlock* pBlock = getCurrentBlock(pBlockIter);
SSDataBlock* pResBlock = pReader->pResBlock;
int32_t numOfCols = blockDataGetNumOfCols(pResBlock);
int32_t numOfOutputCols = blockDataGetNumOfCols(pResBlock);
SBlockLoadSuppInfo* pSupInfo = &pReader->suppInfo;
SFileBlockDumpInfo* pDumpInfo = &pReader->status.fBlockDumpInfo;
int64_t st = taosGetTimestampUs();
SColVal cv = {0};
int32_t colIndex = 0;
int64_t st = taosGetTimestampUs();
bool asc = ASCENDING_TRAVERSE(pReader->order);
int32_t step = asc ? 1 : -1;
@ -724,7 +722,9 @@ static int32_t copyBlockDataToSDataBlock(STsdbReader* pReader, STableBlockScanIn
i += 1;
}
while (i < numOfCols && colIndex < taosArrayGetSize(pBlockData->aIdx)) {
int32_t colIndex = 0;
int32_t num = taosArrayGetSize(pBlockData->aIdx);
while (i < numOfOutputCols && colIndex < num) {
rowIndex = 0;
pColData = taosArrayGet(pResBlock->pDataBlock, i);
@ -744,7 +744,7 @@ static int32_t copyBlockDataToSDataBlock(STsdbReader* pReader, STableBlockScanIn
i += 1;
}
while (i < numOfCols) {
while (i < numOfOutputCols) {
pColData = taosArrayGet(pResBlock->pDataBlock, i);
colDataAppendNNULL(pColData, 0, remain);
i += 1;
@ -1256,7 +1256,7 @@ static int32_t doMergeBufAndFileRows(STsdbReader* pReader, STableBlockScanInfo*
}
tRowMergerClear(&merge);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
taosMemoryFree(pTSRow);
return TSDB_CODE_SUCCESS;
@ -1300,7 +1300,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
}
tRowMergerGetRow(&merge, &pTSRow);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
} else { // key > ik.ts || key > k.ts
ASSERT(key != ik.ts);
@ -1309,7 +1309,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
// [4] ik.ts < k.ts <= key
if (ik.ts < k.ts) {
doMergeMultiRows(piRow, uid, &pBlockScanInfo->iiter, pDelList, &pTSRow, pReader);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
}
@ -1317,7 +1317,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
// [6] k.ts < ik.ts <= key
if (k.ts < ik.ts) {
doMergeMultiRows(pRow, uid, &pBlockScanInfo->iter, pDelList, &pTSRow, pReader);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
}
@ -1326,7 +1326,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
ASSERT(key > ik.ts && key > k.ts);
doMergeMemIMemRows(pRow, piRow, pBlockScanInfo, pReader, &pTSRow);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
}
}
@ -1350,7 +1350,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
}
tRowMergerGetRow(&merge, &pTSRow);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
} else {
ASSERT(ik.ts != k.ts); // this case has been included in the previous if branch
@ -1359,7 +1359,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
// [4] ik.ts > key >= k.ts
if (ik.ts > key) {
doMergeMultiRows(piRow, uid, &pBlockScanInfo->iiter, pDelList, &pTSRow, pReader);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
}
@ -1371,7 +1371,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
doMergeRowsInFileBlocks(pBlockData, pBlockScanInfo, pReader, &merge);
tRowMergerGetRow(&merge, &pTSRow);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
}
@ -1383,7 +1383,7 @@ static int32_t doMergeThreeLevelRows(STsdbReader* pReader, STableBlockScanInfo*
tRowMerge(&merge, &fRow);
doMergeRowsInFileBlocks(pBlockData, pBlockScanInfo, pReader, &merge);
tRowMergerGetRow(&merge, &pTSRow);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
return TSDB_CODE_SUCCESS;
}
}
@ -1438,6 +1438,21 @@ static int32_t buildComposedDataBlockImpl(STsdbReader* pReader, STableBlockScanI
}
// imem & mem are all empty, only file exist
// opt version
// 1. it is not a border point
// 2. the direct next point is not an duplicated timestamp
if ((pDumpInfo->rowIndex < pDumpInfo->totalRows - 1 && pReader->order == TSDB_ORDER_ASC) ||
(pDumpInfo->rowIndex > 0 && pReader->order == TSDB_ORDER_DESC)) {
int32_t step = pReader->order == TSDB_ORDER_ASC? 1:-1;
int64_t nextKey = pBlockData->aTSKEY[pDumpInfo->rowIndex + step];
if (nextKey != key) { // merge is not needed
doAppendRowFromBlock(pReader->pResBlock, pReader, pBlockData, pDumpInfo->rowIndex);
pDumpInfo->rowIndex += step;
return TSDB_CODE_SUCCESS;
}
}
TSDBROW fRow = tsdbRowFromBlockData(pBlockData, pDumpInfo->rowIndex);
STSRow* pTSRow = NULL;
@ -1446,7 +1461,7 @@ static int32_t buildComposedDataBlockImpl(STsdbReader* pReader, STableBlockScanI
tRowMergerInit(&merge, &fRow, pReader->pSchema);
doMergeRowsInFileBlocks(pBlockData, pBlockScanInfo, pReader, &merge);
tRowMergerGetRow(&merge, &pTSRow);
doAppendOneRow(pReader->pResBlock, pReader, pTSRow);
doAppendRowFromTSRow(pReader->pResBlock, pReader, pTSRow);
taosMemoryFree(pTSRow);
tRowMergerClear(&merge);
@ -2201,7 +2216,7 @@ int32_t doMergeRowsInFileBlocks(SBlockData* pBlockData, STableBlockScanInfo* pSc
int32_t step = asc ? 1 : -1;
pDumpInfo->rowIndex += step;
if (pDumpInfo->rowIndex <= pBlockData->nRow - 1) {
if ((pDumpInfo->rowIndex <= pBlockData->nRow - 1 && asc) ||(pDumpInfo->rowIndex >= 0 && !asc)) {
pDumpInfo->rowIndex =
doMergeRowsInFileBlockImpl(pBlockData, pDumpInfo->rowIndex, key, pMerger, &pReader->verRange, step);
}
@ -2325,7 +2340,7 @@ int32_t tsdbGetNextRowInMem(STableBlockScanInfo* pBlockScanInfo, STsdbReader* pR
return TSDB_CODE_SUCCESS;
}
int32_t doAppendOneRow(SSDataBlock* pBlock, STsdbReader* pReader, STSRow* pTSRow) {
int32_t doAppendRowFromTSRow(SSDataBlock* pBlock, STsdbReader* pReader, STSRow* pTSRow) {
int32_t numOfRows = pBlock->info.rows;
int32_t numOfCols = (int32_t)taosArrayGetSize(pBlock->pDataBlock);
@ -2369,6 +2384,47 @@ int32_t doAppendOneRow(SSDataBlock* pBlock, STsdbReader* pReader, STSRow* pTSRow
return TSDB_CODE_SUCCESS;
}
int32_t doAppendRowFromBlock(SSDataBlock* pResBlock, STsdbReader* pReader, SBlockData* pBlockData, int32_t rowIndex) {
int32_t i = 0, j = 0;
int32_t outputRowIndex = pResBlock->info.rows;
SBlockLoadSuppInfo* pSupInfo = &pReader->suppInfo;
SColumnInfoData* pColData = taosArrayGet(pResBlock->pDataBlock, i);
if (pColData->info.colId == PRIMARYKEY_TIMESTAMP_COL_ID) {
colDataAppendInt64(pColData, outputRowIndex, &pBlockData->aTSKEY[rowIndex]);
i += 1;
}
SColVal cv = {0};
int32_t numOfInputCols = taosArrayGetSize(pBlockData->aIdx);
int32_t numOfOutputCols = blockDataGetNumOfCols(pResBlock);
while(i < numOfOutputCols && j < numOfInputCols) {
SColumnInfoData* pCol = taosArrayGet(pResBlock->pDataBlock, i);
SColData* pData = tBlockDataGetColDataByIdx(pBlockData, j);
if (pData->cid == pCol->info.colId) {
tColDataGetValue(pData, rowIndex, &cv);
doCopyColVal(pCol, outputRowIndex, i, &cv, pSupInfo);
j += 1;
} else { // the specified column does not exist in file block, fill with null data
colDataAppendNULL(pCol, outputRowIndex);
}
i += 1;
}
while (i < numOfOutputCols) {
SColumnInfoData* pCol = taosArrayGet(pResBlock->pDataBlock, i);
colDataAppendNULL(pCol, outputRowIndex);
i += 1;
}
pResBlock->info.rows += 1;
return TSDB_CODE_SUCCESS;
}
int32_t buildDataBlockFromBufImpl(STableBlockScanInfo* pBlockScanInfo, int64_t endKey, int32_t capacity,
STsdbReader* pReader) {
SSDataBlock* pBlock = pReader->pResBlock;
@ -2380,7 +2436,7 @@ int32_t buildDataBlockFromBufImpl(STableBlockScanInfo* pBlockScanInfo, int64_t e
break;
}
doAppendOneRow(pBlock, pReader, pTSRow);
doAppendRowFromTSRow(pBlock, pReader, pTSRow);
taosMemoryFree(pTSRow);
// no data in buffer, return immediately

95
source/dnode/vnode/src/tsdb/tsdbSnapshot.c
View File

@ -21,6 +21,7 @@ struct STsdbSnapReader {
int64_t sver;
int64_t ever;
STsdbFS fs;
int8_t type;
// for data file
int8_t dataDone;
int32_t fid;
@ -62,7 +63,8 @@ static int32_t tsdbSnapReadData(STsdbSnapReader* pReader, uint8_t** ppData) {
pReader->iBlockIdx = 0;
pReader->pBlockIdx = NULL;
tsdbInfo("vgId:%d vnode snapshot tsdb open data file to read, fid:%d", TD_VID(pTsdb->pVnode), pReader->fid);
tsdbInfo("vgId:%d vnode snapshot tsdb open data file to read for %s, fid:%d", TD_VID(pTsdb->pVnode), pTsdb->path,
pReader->fid);
}
while (true) {
@ -130,7 +132,7 @@ static int32_t tsdbSnapReadData(STsdbSnapReader* pReader, uint8_t** ppData) {
}
SSnapDataHdr* pHdr = (SSnapDataHdr*)(*ppData);
pHdr->type = 1;
pHdr->type = pReader->type;
pHdr->size = size;
TABLEID* pId = (TABLEID*)(&pHdr[1]);
@ -139,9 +141,9 @@ static int32_t tsdbSnapReadData(STsdbSnapReader* pReader, uint8_t** ppData) {
tPutBlockData((uint8_t*)(&pId[1]), &pReader->nBlockData);
tsdbInfo("vgId:%d vnode snapshot read data, fid:%d suid:%" PRId64 " uid:%" PRId64
tsdbInfo("vgId:%d vnode snapshot read data for %s, fid:%d suid:%" PRId64 " uid:%" PRId64
" iBlock:%d minVersion:%d maxVersion:%d nRow:%d out of %d size:%d",
TD_VID(pTsdb->pVnode), pReader->fid, pReader->pBlockIdx->suid, pReader->pBlockIdx->uid,
TD_VID(pTsdb->pVnode), pTsdb->path, pReader->fid, pReader->pBlockIdx->suid, pReader->pBlockIdx->uid,
pReader->iBlock - 1, pBlock->minVersion, pBlock->maxVersion, pReader->nBlockData.nRow, pBlock->nRow,
size);
@ -154,7 +156,8 @@ _exit:
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb read data failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb read data for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
return code;
}
@ -212,7 +215,7 @@ static int32_t tsdbSnapReadDel(STsdbSnapReader* pReader, uint8_t** ppData) {
}
SSnapDataHdr* pHdr = (SSnapDataHdr*)(*ppData);
pHdr->type = 2;
pHdr->type = SNAP_DATA_DEL;
pHdr->size = size;
TABLEID* pId = (TABLEID*)(&pHdr[1]);
@ -228,8 +231,8 @@ static int32_t tsdbSnapReadDel(STsdbSnapReader* pReader, uint8_t** ppData) {
n += tPutDelData((*ppData) + n, pDelData);
}
tsdbInfo("vgId:%d vnode snapshot tsdb read del data, suid:%" PRId64 " uid:%d" PRId64 " size:%d",
TD_VID(pTsdb->pVnode), pDelIdx->suid, pDelIdx->uid, size);
tsdbInfo("vgId:%d vnode snapshot tsdb read del data for %s, suid:%" PRId64 " uid:%d" PRId64 " size:%d",
TD_VID(pTsdb->pVnode), pTsdb->path, pDelIdx->suid, pDelIdx->uid, size);
break;
}
@ -238,11 +241,12 @@ _exit:
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb read del failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb read del for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->pVnode,
tstrerror(code));
return code;
}
int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapReader** ppReader) {
int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, int8_t type, STsdbSnapReader** ppReader) {
int32_t code = 0;
STsdbSnapReader* pReader = NULL;
@ -255,6 +259,7 @@ int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapRe
pReader->pTsdb = pTsdb;
pReader->sver = sver;
pReader->ever = ever;
pReader->type = type;
code = taosThreadRwlockRdlock(&pTsdb->rwLock);
if (code) {
@ -297,12 +302,13 @@ int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapRe
goto _err;
}
tsdbInfo("vgId:%d vnode snapshot tsdb reader opened", TD_VID(pTsdb->pVnode));
tsdbInfo("vgId:%d vnode snapshot tsdb reader opened for %s", TD_VID(pTsdb->pVnode), pTsdb->path);
*ppReader = pReader;
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb reader open failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb reader open for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
*ppReader = NULL;
return code;
}
@ -327,7 +333,7 @@ int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader) {
tsdbFSUnref(pReader->pTsdb, &pReader->fs);
tsdbInfo("vgId:%d vnode snapshot tsdb reader closed", TD_VID(pReader->pTsdb->pVnode));
tsdbInfo("vgId:%d vnode snapshot tsdb reader closed for %s", TD_VID(pReader->pTsdb->pVnode), pReader->pTsdb->path);
taosMemoryFree(pReader);
*ppReader = NULL;
@ -368,10 +374,12 @@ int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData) {
}
_exit:
tsdbDebug("vgId:%d vnode snapshot tsdb read for %s", TD_VID(pReader->pTsdb->pVnode), pReader->pTsdb->path);
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb read failed since %s", TD_VID(pReader->pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb read for %s failed since %s", TD_VID(pReader->pTsdb->pVnode),
pReader->pTsdb->path, tstrerror(code));
return code;
}
@ -436,7 +444,8 @@ static int32_t tsdbSnapWriteAppendData(STsdbSnapWriter* pWriter, uint8_t* pData,
return code;
_err:
tsdbError("vgId:%d tsdb snapshot write append data failed since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d tsdb snapshot write append data for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode),
pWriter->pTsdb->path, tstrerror(code));
return code;
}
@ -522,9 +531,12 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) {
}
_exit:
tsdbInfo("vgId:%d tsdb snapshot write table data end for %s", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path);
return code;
_err:
tsdbError("vgId:%d tsdb snapshot write table data end for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode),
pWriter->pTsdb->path, tstrerror(code));
return code;
}
@ -570,6 +582,8 @@ _exit:
return code;
_err:
tsdbError("vgId:%d tsdb snapshot move write table data for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode),
pWriter->pTsdb->path, tstrerror(code));
return code;
}
@ -708,8 +722,8 @@ static int32_t tsdbSnapWriteTableDataImpl(STsdbSnapWriter* pWriter) {
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb write table data impl failed since %s", TD_VID(pWriter->pTsdb->pVnode),
tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb write table data impl for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode),
pWriter->pTsdb->path, tstrerror(code));
return code;
}
@ -794,11 +808,12 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, TABLEID id) {
if (code) goto _err;
_exit:
tsdbDebug("vgId:%d vnode snapshot tsdb write data impl for %s", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path);
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb write data impl failed since %s", TD_VID(pWriter->pTsdb->pVnode),
tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb write data impl for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode),
pWriter->pTsdb->path, tstrerror(code));
return code;
}
@ -833,11 +848,12 @@ static int32_t tsdbSnapWriteDataEnd(STsdbSnapWriter* pWriter) {
}
_exit:
tsdbInfo("vgId:%d vnode snapshot tsdb writer data end", TD_VID(pTsdb->pVnode));
tsdbInfo("vgId:%d vnode snapshot tsdb writer data end for %s", TD_VID(pTsdb->pVnode), pTsdb->path);
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb writer data end failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb writer data end for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
return code;
}
@ -920,12 +936,13 @@ static int32_t tsdbSnapWriteData(STsdbSnapWriter* pWriter, uint8_t* pData, uint3
code = tsdbSnapWriteTableData(pWriter, id);
if (code) goto _err;
tsdbInfo("vgId:%d vnode snapshot tsdb write data, fid:%d suid:%" PRId64 " uid:%" PRId64 " nRow:%d",
TD_VID(pTsdb->pVnode), fid, id.suid, id.suid, pBlockData->nRow);
tsdbInfo("vgId:%d vnode snapshot tsdb write data for %s, fid:%d suid:%" PRId64 " uid:%" PRId64 " nRow:%d",
TD_VID(pTsdb->pVnode), pTsdb->path, fid, id.suid, id.suid, pBlockData->nRow);
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb write data failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb write data for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
return code;
}
@ -1015,7 +1032,8 @@ _exit:
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb write del failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb write del for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
return code;
}
@ -1056,11 +1074,12 @@ static int32_t tsdbSnapWriteDelEnd(STsdbSnapWriter* pWriter) {
}
_exit:
tsdbInfo("vgId:%d vnode snapshot tsdb write del end", TD_VID(pTsdb->pVnode));
tsdbInfo("vgId:%d vnode snapshot tsdb write del for %s end", TD_VID(pTsdb->pVnode), pTsdb->path);
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb write del end failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb write del end for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
return code;
}
@ -1127,10 +1146,12 @@ int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWr
}
*ppWriter = pWriter;
return code;
tsdbInfo("vgId:%d tsdb snapshot writer open for %s succeed", TD_VID(pTsdb->pVnode), pTsdb->path);
return code;
_err:
tsdbError("vgId:%d tsdb snapshot writer open failed since %s", TD_VID(pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d tsdb snapshot writer open for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path,
tstrerror(code));
*ppWriter = NULL;
return code;
}
@ -1157,14 +1178,16 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) {
if (code) goto _err;
}
tsdbInfo("vgId:%d vnode snapshot tsdb writer close for %s", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path);
taosMemoryFree(pWriter);
*ppWriter = NULL;
return code;
_err:
tsdbError("vgId:%d vnode snapshot tsdb writer close failed since %s", TD_VID(pWriter->pTsdb->pVnode),
tstrerror(code));
tsdbError("vgId:%d vnode snapshot tsdb writer close for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode),
pWriter->pTsdb->path, tstrerror(code));
taosMemoryFree(pWriter);
*ppWriter = NULL;
return code;
}
@ -1173,7 +1196,7 @@ int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData)
SSnapDataHdr* pHdr = (SSnapDataHdr*)pData;
// ts data
if (pHdr->type == 1) {
if (pHdr->type == SNAP_DATA_TSDB) {
code = tsdbSnapWriteData(pWriter, pData, nData);
if (code) goto _err;
@ -1186,15 +1209,17 @@ int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData)
}
// del data
if (pHdr->type == 2) {
if (pHdr->type == SNAP_DATA_DEL) {
code = tsdbSnapWriteDel(pWriter, pData, nData);
if (code) goto _err;
}
_exit:
tsdbDebug("vgId:%d tsdb snapshow write for %s succeed", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path);
return code;
_err:
tsdbError("vgId:%d tsdb snapshow write failed since %s", TD_VID(pWriter->pTsdb->pVnode), tstrerror(code));
tsdbError("vgId:%d tsdb snapshow write for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path,
tstrerror(code));
return code;
}

38
source/dnode/vnode/src/tsdb/tsdbUtil.c
View File

@ -151,26 +151,6 @@ int32_t tTABLEIDCmprFn(const void *p1, const void *p2) {
return 0;
}
// TSDBKEY =======================================================================
int32_t tsdbKeyCmprFn(const void *p1, const void *p2) {
TSDBKEY *pKey1 = (TSDBKEY *)p1;
TSDBKEY *pKey2 = (TSDBKEY *)p2;
if (pKey1->ts < pKey2->ts) {
return -1;
} else if (pKey1->ts > pKey2->ts) {
return 1;
}
if (pKey1->version < pKey2->version) {
return -1;
} else if (pKey1->version > pKey2->version) {
return 1;
}
return 0;
}
// TSDBKEY ======================================================
static FORCE_INLINE int32_t tPutTSDBKEY(uint8_t *p, TSDBKEY *pKey) {
int32_t n = 0;
@ -1401,7 +1381,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
break;
case TSDB_DATA_TYPE_BOOL:
break;
case TSDB_DATA_TYPE_TINYINT:{
case TSDB_DATA_TYPE_TINYINT: {
pColAgg->sum += colVal.value.i8;
if (pColAgg->min > colVal.value.i8) {
pColAgg->min = colVal.value.i8;
@ -1411,7 +1391,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
break;
}
case TSDB_DATA_TYPE_SMALLINT:{
case TSDB_DATA_TYPE_SMALLINT: {
pColAgg->sum += colVal.value.i16;
if (pColAgg->min > colVal.value.i16) {
pColAgg->min = colVal.value.i16;
@ -1441,7 +1421,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
break;
}
case TSDB_DATA_TYPE_FLOAT:{
case TSDB_DATA_TYPE_FLOAT: {
pColAgg->sum += colVal.value.f;
if (pColAgg->min > colVal.value.f) {
pColAgg->min = colVal.value.f;
@ -1451,7 +1431,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
break;
}
case TSDB_DATA_TYPE_DOUBLE:{
case TSDB_DATA_TYPE_DOUBLE: {
pColAgg->sum += colVal.value.d;
if (pColAgg->min > colVal.value.d) {
pColAgg->min = colVal.value.d;
@ -1463,7 +1443,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
case TSDB_DATA_TYPE_VARCHAR:
break;
case TSDB_DATA_TYPE_TIMESTAMP:{
case TSDB_DATA_TYPE_TIMESTAMP: {
if (pColAgg->min > colVal.value.i64) {
pColAgg->min = colVal.value.i64;
}
@ -1474,7 +1454,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
case TSDB_DATA_TYPE_NCHAR:
break;
case TSDB_DATA_TYPE_UTINYINT:{
case TSDB_DATA_TYPE_UTINYINT: {
pColAgg->sum += colVal.value.u8;
if (pColAgg->min > colVal.value.u8) {
pColAgg->min = colVal.value.u8;
@ -1484,7 +1464,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
break;
}
case TSDB_DATA_TYPE_USMALLINT:{
case TSDB_DATA_TYPE_USMALLINT: {
pColAgg->sum += colVal.value.u16;
if (pColAgg->min > colVal.value.u16) {
pColAgg->min = colVal.value.u16;
@ -1494,7 +1474,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
break;
}
case TSDB_DATA_TYPE_UINT:{
case TSDB_DATA_TYPE_UINT: {
pColAgg->sum += colVal.value.u32;
if (pColAgg->min > colVal.value.u32) {
pColAgg->min = colVal.value.u32;
@ -1504,7 +1484,7 @@ void tsdbCalcColDataSMA(SColData *pColData, SColumnDataAgg *pColAgg) {
}
break;
}
case TSDB_DATA_TYPE_UBIGINT:{
case TSDB_DATA_TYPE_UBIGINT: {
pColAgg->sum += colVal.value.u64;
if (pColAgg->min > colVal.value.u64) {
pColAgg->min = colVal.value.u64;

117
source/dnode/vnode/src/vnd/vnodeSnapshot.c
View File

@ -28,7 +28,8 @@ struct SVSnapReader {
int8_t tsdbDone;
STsdbSnapReader *pTsdbReader;
// rsma
int8_t rsmaDone[TSDB_RETENTION_L2];
int8_t rsmaDone;
SRsmaSnapReader *pRsmaReader;
};
int32_t vnodeSnapReaderOpen(SVnode *pVnode, int64_t sver, int64_t ever, SVSnapReader **ppReader) {
@ -57,6 +58,10 @@ _err:
int32_t vnodeSnapReaderClose(SVSnapReader *pReader) {
int32_t code = 0;
if (pReader->pRsmaReader) {
rsmaSnapReaderClose(&pReader->pRsmaReader);
}
if (pReader->pTsdbReader) {
tsdbSnapReaderClose(&pReader->pTsdbReader);
}
@ -99,7 +104,7 @@ int32_t vnodeSnapRead(SVSnapReader *pReader, uint8_t **ppData, uint32_t *nData)
if (!pReader->tsdbDone) {
// open if not
if (pReader->pTsdbReader == NULL) {
code = tsdbSnapReaderOpen(pReader->pVnode->pTsdb, pReader->sver, pReader->ever, &pReader->pTsdbReader);
code = tsdbSnapReaderOpen(pReader->pVnode->pTsdb, pReader->sver, pReader->ever, SNAP_DATA_TSDB, &pReader->pTsdbReader);
if (code) goto _err;
}
@ -118,40 +123,26 @@ int32_t vnodeSnapRead(SVSnapReader *pReader, uint8_t **ppData, uint32_t *nData)
}
// RSMA ==============
#if 0
if (VND_IS_RSMA(pReader->pVnode)) {
// RSMA1/RSMA2
for (int32_t i = 0; i < TSDB_RETENTION_L2; ++i) {
if (!pReader->rsmaDone[i]) {
if (!pReader->pVnode->pSma->pRSmaTsdb[i]) {
// no valid tsdb
pReader->rsmaDone[i] = 1;
continue;
}
if (pReader->pTsdbReader == NULL) {
code = tsdbSnapReaderOpen(pReader->pVnode->pSma->pRSmaTsdb[i], pReader->sver, pReader->ever,
&pReader->pTsdbReader);
if (code) goto _err;
}
if (VND_IS_RSMA(pReader->pVnode) && !pReader->rsmaDone) {
// open if not
if (pReader->pRsmaReader == NULL) {
code = rsmaSnapReaderOpen(pReader->pVnode->pSma, pReader->sver, pReader->ever, &pReader->pRsmaReader);
if (code) goto _err;
}
code = tsdbSnapRead(pReader->pTsdbReader, ppData);
if (code) {
goto _err;
} else {
if (*ppData) {
goto _exit;
} else {
pReader->tsdbDone = 1;
code = tsdbSnapReaderClose(&pReader->pTsdbReader);
if (code) goto _err;
}
}
code = rsmaSnapRead(pReader->pRsmaReader, ppData);
if (code) {
goto _err;
} else {
if (*ppData) {
goto _exit;
} else {
pReader->tsdbDone = 1;
code = rsmaSnapReaderClose(&pReader->pRsmaReader);
if (code) goto _err;
}
}
// QTaskInfoFile
// TODO ...
}
#endif
*ppData = NULL;
*nData = 0;
@ -186,6 +177,8 @@ struct SVSnapWriter {
SMetaSnapWriter *pMetaSnapWriter;
// tsdb
STsdbSnapWriter *pTsdbSnapWriter;
// rsma
SRsmaSnapWriter *pRsmaSnapWriter;
};
int32_t vnodeSnapWriterOpen(SVnode *pVnode, int64_t sver, int64_t ever, SVSnapWriter **ppWriter) {
@ -235,6 +228,11 @@ int32_t vnodeSnapWriterClose(SVSnapWriter *pWriter, int8_t rollback, SSnapshot *
if (code) goto _err;
}
if (pWriter->pRsmaSnapWriter) {
code = rsmaSnapWriterClose(&pWriter->pRsmaSnapWriter, rollback);
if (code) goto _err;
}
if (!rollback) {
SVnodeInfo info = {0};
char dir[TSDB_FILENAME_LEN];
@ -282,28 +280,51 @@ int32_t vnodeSnapWrite(SVSnapWriter *pWriter, uint8_t *pData, uint32_t nData) {
vInfo("vgId:%d vnode snapshot write data, index:%" PRId64 " type:%d nData:%d", TD_VID(pVnode), pHdr->index,
pHdr->type, nData);
if (pHdr->type == 0) {
// meta
switch (pHdr->type) {
case SNAP_DATA_META: {
// meta
if (pWriter->pMetaSnapWriter == NULL) {
code = metaSnapWriterOpen(pVnode->pMeta, pWriter->sver, pWriter->ever, &pWriter->pMetaSnapWriter);
if (code) goto _err;
}
if (pWriter->pMetaSnapWriter == NULL) {
code = metaSnapWriterOpen(pVnode->pMeta, pWriter->sver, pWriter->ever, &pWriter->pMetaSnapWriter);
code = metaSnapWrite(pWriter->pMetaSnapWriter, pData, nData);
if (code) goto _err;
}
} break;
case SNAP_DATA_TSDB: {
// tsdb
if (pWriter->pTsdbSnapWriter == NULL) {
code = tsdbSnapWriterOpen(pVnode->pTsdb, pWriter->sver, pWriter->ever, &pWriter->pTsdbSnapWriter);
if (code) goto _err;
}
code = metaSnapWrite(pWriter->pMetaSnapWriter, pData, nData);
if (code) goto _err;
} else {
// tsdb
if (pWriter->pTsdbSnapWriter == NULL) {
code = tsdbSnapWriterOpen(pVnode->pTsdb, pWriter->sver, pWriter->ever, &pWriter->pTsdbSnapWriter);
code = tsdbSnapWrite(pWriter->pTsdbSnapWriter, pData, nData);
if (code) goto _err;
}
} break;
case SNAP_DATA_RSMA1:
case SNAP_DATA_RSMA2: {
// rsma1/rsma2
if (pWriter->pRsmaSnapWriter == NULL) {
code = rsmaSnapWriterOpen(pVnode->pSma, pWriter->sver, pWriter->ever, &pWriter->pRsmaSnapWriter);
if (code) goto _err;
}
code = tsdbSnapWrite(pWriter->pTsdbSnapWriter, pData, nData);
if (code) goto _err;
code = rsmaSnapWrite(pWriter->pRsmaSnapWriter, pData, nData);
if (code) goto _err;
} break;
case SNAP_DATA_QTASK: {
// qtask for rsma
if (pWriter->pRsmaSnapWriter == NULL) {
code = rsmaSnapWriterOpen(pVnode->pSma, pWriter->sver, pWriter->ever, &pWriter->pRsmaSnapWriter);
if (code) goto _err;
}
code = rsmaSnapWrite(pWriter->pRsmaSnapWriter, pData, nData);
if (code) goto _err;
} break;
default:
break;
}
_exit:
return code;

2
source/dnode/vnode/src/vnd/vnodeSvr.c
View File

@ -447,6 +447,7 @@ _err:
static int32_t vnodeProcessCreateTbReq(SVnode *pVnode, int64_t version, void *pReq, int32_t len, SRpcMsg *pRsp) {
SDecoder decoder = {0};
SEncoder encoder = {0};
int32_t rcode = 0;
SVCreateTbBatchReq req = {0};
SVCreateTbReq *pCreateReq;
@ -515,7 +516,6 @@ static int32_t vnodeProcessCreateTbReq(SVnode *pVnode, int64_t version, void *pR
tdUidStoreFree(pStore);
// prepare rsp
SEncoder encoder = {0};
int32_t ret = 0;
tEncodeSize(tEncodeSVCreateTbBatchRsp, &rsp, pRsp->contLen, ret);
pRsp->pCont = rpcMallocCont(pRsp->contLen);

2
source/libs/catalog/inc/catalogInt.h
View File

@ -679,6 +679,8 @@ void ctgClearSubTaskRes(SCtgSubRes *pRes);
void ctgFreeQNode(SCtgQNode *node);
void ctgClearHandle(SCatalog* pCtg);
void ctgFreeTbCacheImpl(SCtgTbCache *pCache);
int32_t ctgRemoveTbMeta(SCatalog* pCtg, SName* pTableName);
int32_t ctgGetTbHashVgroup(SCatalog *pCtg, SRequestConnInfo *pConn, const SName *pTableName, SVgroupInfo *pVgroup);
extern SCatalogMgmt gCtgMgmt;

111
source/libs/catalog/src/catalog.c
View File

@ -92,7 +92,7 @@ int32_t ctgRefreshTbMeta(SCatalog* pCtg, SRequestConnInfo *pConn, SCtgTbMetaCtx*
int32_t code = 0;
if (!CTG_FLAG_IS_SYS_DB(ctx->flag)) {
CTG_ERR_RET(catalogGetTableHashVgroup(pCtg, pConn, ctx->pName, &vgroupInfo));
CTG_ERR_RET(ctgGetTbHashVgroup(pCtg, pConn, ctx->pName, &vgroupInfo));
}
STableMetaOutput moutput = {0};
@ -337,7 +337,10 @@ int32_t ctgGetTbType(SCatalog* pCtg, SRequestConnInfo *pConn, SName* pTableName,
}
STableMeta* pMeta = NULL;
CTG_ERR_RET(catalogGetTableMeta(pCtg, pConn, pTableName, &pMeta));
SCtgTbMetaCtx ctx = {0};
ctx.pName = (SName*)pTableName;
ctx.flag = CTG_FLAG_UNKNOWN_STB;
CTG_ERR_RET(ctgGetTbMeta(pCtg, pConn, &ctx, &pMeta));
*tbType = pMeta->tableType;
taosMemoryFree(pMeta);
@ -391,7 +394,7 @@ int32_t ctgGetTbCfg(SCatalog* pCtg, SRequestConnInfo *pConn, SName* pTableName,
CTG_ERR_RET(ctgGetTableCfgFromMnode(pCtg, pConn, pTableName, pCfg, NULL));
} else {
SVgroupInfo vgroupInfo = {0};
CTG_ERR_RET(catalogGetTableHashVgroup(pCtg, pConn, pTableName, &vgroupInfo));
CTG_ERR_RET(ctgGetTbHashVgroup(pCtg, pConn, pTableName, &vgroupInfo));
CTG_ERR_RET(ctgGetTableCfgFromVnode(pCtg, pConn, pTableName, &vgroupInfo, pCfg, NULL));
}
@ -477,6 +480,57 @@ _return:
CTG_RET(code);
}
int32_t ctgGetTbHashVgroup(SCatalog *pCtg, SRequestConnInfo *pConn, const SName *pTableName, SVgroupInfo *pVgroup) {
if (IS_SYS_DBNAME(pTableName->dbname)) {
ctgError("no valid vgInfo for db, dbname:%s", pTableName->dbname);
CTG_ERR_RET(TSDB_CODE_CTG_INVALID_INPUT);
}
SCtgDBCache* dbCache = NULL;
int32_t code = 0;
char db[TSDB_DB_FNAME_LEN] = {0};
tNameGetFullDbName(pTableName, db);
SDBVgInfo *vgInfo = NULL;
CTG_ERR_JRET(ctgGetDBVgInfo(pCtg, pConn, db, &dbCache, &vgInfo));
CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, vgInfo ? vgInfo : dbCache->vgCache.vgInfo, pTableName, pVgroup));
_return:
if (dbCache) {
ctgRUnlockVgInfo(dbCache);
ctgReleaseDBCache(pCtg, dbCache);
}
if (vgInfo) {
taosHashCleanup(vgInfo->vgHash);
taosMemoryFreeClear(vgInfo);
}
CTG_RET(code);
}
int32_t ctgRemoveTbMeta(SCatalog* pCtg, SName* pTableName) {
int32_t code = 0;
if (NULL == pCtg || NULL == pTableName) {
CTG_ERR_RET(TSDB_CODE_CTG_INVALID_INPUT);
}
if (NULL == pCtg->dbCache) {
return TSDB_CODE_SUCCESS;
}
CTG_ERR_JRET(ctgRemoveTbMetaFromCache(pCtg, pTableName, true));
_return:
CTG_RET(code);
}
int32_t catalogInit(SCatalogCfg* cfg) {
if (gCtgMgmt.pCluster) {
qError("catalog already initialized");
@ -772,21 +826,7 @@ _return:
int32_t catalogRemoveTableMeta(SCatalog* pCtg, SName* pTableName) {
CTG_API_ENTER();
int32_t code = 0;
if (NULL == pCtg || NULL == pTableName) {
CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
}
if (NULL == pCtg->dbCache) {
CTG_API_LEAVE(TSDB_CODE_SUCCESS);
}
CTG_ERR_JRET(ctgRemoveTbMetaFromCache(pCtg, pTableName, true));
_return:
CTG_API_LEAVE(code);
CTG_API_LEAVE(ctgRemoveTbMeta(pCtg, pTableName));
}
int32_t catalogRemoveStbMeta(SCatalog* pCtg, const char* dbFName, uint64_t dbId, const char* stbName, uint64_t suid) {
@ -878,12 +918,12 @@ int32_t catalogChkTbMetaVersion(SCatalog* pCtg, SRequestConnInfo *pConn, SArray*
case TSDB_CHILD_TABLE: {
SName stb = name;
strcpy(stb.tname, stbName);
catalogRemoveTableMeta(pCtg, &stb);
ctgRemoveTbMeta(pCtg, &stb);
break;
}
case TSDB_SUPER_TABLE:
case TSDB_NORMAL_TABLE:
catalogRemoveTableMeta(pCtg, &name);
ctgRemoveTbMeta(pCtg, &name);
break;
default:
ctgError("ignore table type %d", tbType);
@ -947,34 +987,7 @@ int32_t catalogGetTableDistVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const
int32_t catalogGetTableHashVgroup(SCatalog *pCtg, SRequestConnInfo *pConn, const SName *pTableName, SVgroupInfo *pVgroup) {
CTG_API_ENTER();
if (IS_SYS_DBNAME(pTableName->dbname)) {
ctgError("no valid vgInfo for db, dbname:%s", pTableName->dbname);
CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
}
SCtgDBCache* dbCache = NULL;
int32_t code = 0;
char db[TSDB_DB_FNAME_LEN] = {0};
tNameGetFullDbName(pTableName, db);
SDBVgInfo *vgInfo = NULL;
CTG_ERR_JRET(ctgGetDBVgInfo(pCtg, pConn, db, &dbCache, &vgInfo));
CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, vgInfo ? vgInfo : dbCache->vgCache.vgInfo, pTableName, pVgroup));
_return:
if (dbCache) {
ctgRUnlockVgInfo(dbCache);
ctgReleaseDBCache(pCtg, dbCache);
}
if (vgInfo) {
taosHashCleanup(vgInfo->vgHash);
taosMemoryFreeClear(vgInfo);
}
CTG_API_LEAVE(code);
CTG_API_LEAVE(ctgGetTbHashVgroup(pCtg, pConn, pTableName, pVgroup));
}
int32_t catalogGetAllMeta(SCatalog* pCtg, SRequestConnInfo *pConn, const SCatalogReq* pReq, SMetaData* pRsp) {
@ -1200,7 +1213,7 @@ int32_t catalogRefreshGetTableCfg(SCatalog* pCtg, SRequestConnInfo *pConn, const
}
int32_t code = 0;
CTG_ERR_JRET(catalogRemoveTableMeta(pCtg, (SName*)pTableName));
CTG_ERR_JRET(ctgRemoveTbMeta(pCtg, (SName*)pTableName));
CTG_ERR_JRET(ctgGetTbCfg(pCtg, pConn, (SName*)pTableName, pCfg));

2
source/libs/catalog/src/ctgAsync.c
View File

@ -398,7 +398,7 @@ int32_t ctgHandleForceUpdate(SCatalog* pCtg, int32_t taskNum, SCtgJob *pJob, con
SName* name = taosHashIterate(pTb, NULL);
while (name) {
catalogRemoveTableMeta(pCtg, name);
ctgRemoveTbMeta(pCtg, name);
name = taosHashIterate(pTb, name);
}

6
source/libs/command/inc/commandInt.h
View File

@ -19,6 +19,8 @@
#ifdef __cplusplus
extern "C" {
#endif
// clang-format off
#include "nodes.h"
#include "plannodes.h"
#include "ttime.h"
@ -77,6 +79,8 @@ extern "C" {
#define EXPLAIN_EXECINFO_FORMAT "cost=%.3f..%.3f rows=%" PRIu64
#define EXPLAIN_MODE_FORMAT "mode=%s"
#define EXPLAIN_STRING_TYPE_FORMAT "%s"
#define EXPLAIN_INPUT_ORDER_FORMAT "input_order=%s"
#define EXPLAIN_OUTPUT_ORDER_TYPE_FORMAT "output_order=%s"
#define COMMAND_RESET_LOG "resetLog"
#define COMMAND_SCHEDULE_POLICY "schedulePolicy"
@ -122,7 +126,7 @@ typedef struct SExplainCtx {
SHashObj *groupHash; // Hash<SExplainGroup>
} SExplainCtx;
#define EXPLAIN_ORDER_STRING(_order) ((TSDB_ORDER_ASC == _order) ? "Ascending" : "Descending")
#define EXPLAIN_ORDER_STRING(_order) ((ORDER_ASC == _order) ? "asc" : "desc")
#define EXPLAIN_JOIN_STRING(_type) ((JOIN_TYPE_INNER == _type) ? "Inner join" : "Join")
#define INVERAL_TIME_FROM_PRECISION_TO_UNIT(_t, _u, _p) (((_u) == 'n' || (_u) == 'y') ? (_t) : (convertTimeFromPrecisionToUnit(_t, _p, _u)))

9
source/libs/command/src/explain.c
View File

@ -13,6 +13,7 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
// clang-format off
#include "commandInt.h"
#include "plannodes.h"
#include "query.h"
@ -849,6 +850,10 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
EXPLAIN_ROW_APPEND(EXPLAIN_FUNCTIONS_FORMAT, pIntNode->window.pFuncs->length);
EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pIntNode->window.node.pOutputDataBlockDesc->totalRowSize);
EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
EXPLAIN_ROW_APPEND(EXPLAIN_INPUT_ORDER_FORMAT, EXPLAIN_ORDER_STRING(pIntNode->window.inputTsOrder));
EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
EXPLAIN_ROW_APPEND(EXPLAIN_OUTPUT_ORDER_TYPE_FORMAT, EXPLAIN_ORDER_STRING(pIntNode->window.outputTsOrder));
EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
EXPLAIN_ROW_END();
QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
@ -1154,7 +1159,9 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
EXPLAIN_ROW_NEW(level + 1, EXPLAIN_MERGE_KEYS_FORMAT);
for (int32_t i = 0; i < LIST_LENGTH(pMergeNode->pMergeKeys); ++i) {
SOrderByExprNode *ptn = (SOrderByExprNode *)nodesListGetNode(pMergeNode->pMergeKeys, i);
EXPLAIN_ROW_APPEND("%s ", nodesGetNameFromColumnNode(ptn->pExpr));
EXPLAIN_ROW_APPEND(EXPLAIN_STRING_TYPE_FORMAT, nodesGetNameFromColumnNode(ptn->pExpr));
EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
EXPLAIN_ROW_APPEND(EXPLAIN_STRING_TYPE_FORMAT, EXPLAIN_ORDER_STRING(ptn->order));
}
EXPLAIN_ROW_END();
QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));

2
source/libs/executor/inc/executil.h
View File

@ -103,7 +103,7 @@ void initGroupedResultInfo(SGroupResInfo* pGroupResInfo, SHashObj* pHashmap, int
void cleanupGroupResInfo(SGroupResInfo* pGroupResInfo);
void initMultiResInfoFromArrayList(SGroupResInfo* pGroupResInfo, SArray* pArrayList);
bool hasDataInGroupInfo(SGroupResInfo* pGroupResInfo);
bool hasRemainResults(SGroupResInfo* pGroupResInfo);
int32_t getNumOfTotalRes(SGroupResInfo* pGroupResInfo);

37
source/libs/executor/inc/executorimpl.h
View File

@ -297,6 +297,20 @@ enum {
TABLE_SCAN__BLOCK_ORDER = 2,
};
typedef struct SAggSupporter {
SHashObj* pResultRowHashTable; // quick locate the window object for each result
char* keyBuf; // window key buffer
SDiskbasedBuf* pResultBuf; // query result buffer based on blocked-wised disk file
int32_t resultRowSize; // the result buffer size for each result row, with the meta data size for each row
} SAggSupporter;
typedef struct {
// if the upstream is an interval operator, the interval info is also kept here to get the time window to check if current data block needs to be loaded.
SInterval interval;
SAggSupporter *pAggSup;
SExprSupp *pExprSup; // expr supporter of aggregate operator
} SAggOptrPushDownInfo;
typedef struct STableScanInfo {
STsdbReader* dataReader;
SReadHandle readHandle;
@ -312,12 +326,13 @@ typedef struct STableScanInfo {
SQueryTableDataCond cond;
int32_t scanFlag; // table scan flag to denote if it is a repeat/reverse/main scan
int32_t dataBlockLoadFlag;
SInterval interval; // if the upstream is an interval operator, the interval info is also kept here to get the time window to check if current data block needs to be loaded.
// SInterval interval; // if the upstream is an interval operator, the interval info is also kept here to get the time window to check if current data block needs to be loaded.
SSampleExecInfo sample; // sample execution info
int32_t currentGroupId;
int32_t currentTable;
int8_t scanMode;
int8_t noTable;
SAggOptrPushDownInfo pdInfo;
int8_t assignBlockUid;
} STableScanInfo;
@ -505,13 +520,6 @@ typedef struct SOptrBasicInfo {
SSDataBlock* pRes;
} SOptrBasicInfo;
typedef struct SAggSupporter {
SHashObj* pResultRowHashTable; // quick locate the window object for each result
char* keyBuf; // window key buffer
SDiskbasedBuf* pResultBuf; // query result buffer based on blocked-wised disk file
int32_t resultRowSize; // the result buffer size for each result row, with the meta data size for each row
} SAggSupporter;
typedef struct SIntervalAggOperatorInfo {
// SOptrBasicInfo should be first, SAggSupporter should be second for stream encode
SOptrBasicInfo binfo; // basic info
@ -523,7 +531,8 @@ typedef struct SIntervalAggOperatorInfo {
STimeWindow win; // query time range
bool timeWindowInterpo; // interpolation needed or not
SArray* pInterpCols; // interpolation columns
int32_t order; // current SSDataBlock scan order
int32_t resultTsOrder; // result timestamp order
int32_t inputOrder; // input data ts order
EOPTR_EXEC_MODEL execModel; // operator execution model [batch model|stream model]
STimeWindowAggSupp twAggSup;
bool invertible;
@ -533,8 +542,7 @@ typedef struct SIntervalAggOperatorInfo {
SArray* pDelWins; // SWinRes
int32_t delIndex;
SSDataBlock* pDelRes;
SNode *pCondition;
SNode* pCondition;
} SIntervalAggOperatorInfo;
typedef struct SMergeAlignedIntervalAggOperatorInfo {
@ -804,7 +812,7 @@ typedef struct STagFilterOperatorInfo {
typedef struct SJoinOperatorInfo {
SSDataBlock *pRes;
int32_t joinType;
int32_t inputTsOrder;
int32_t inputOrder;
SSDataBlock *pLeft;
int32_t leftPos;
@ -847,7 +855,6 @@ void doApplyFunctions(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, STimeWin
int32_t extractDataBlockFromFetchRsp(SSDataBlock* pRes, SLoadRemoteDataInfo* pLoadInfo, int32_t numOfRows, char* pData,
int32_t compLen, int32_t numOfOutput, int64_t startTs, uint64_t* total,
SArray* pColList);
STimeWindow getAlignQueryTimeWindow(SInterval* pInterval, int32_t precision, int64_t key);
STimeWindow getFirstQualifiedTimeWindow(int64_t ts, STimeWindow* pWindow, SInterval* pInterval, int32_t order);
int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t *order, int32_t* scanFlag);
@ -979,8 +986,7 @@ void setTaskStatus(SExecTaskInfo* pTaskInfo, int8_t status);
int32_t createExecTaskInfoImpl(SSubplan* pPlan, SExecTaskInfo** pTaskInfo, SReadHandle* pHandle, uint64_t taskId,
const char* sql, EOPTR_EXEC_MODEL model);
int32_t createDataSinkParam(SDataSinkNode *pNode, void **pParam, qTaskInfo_t* pTaskInfo, SReadHandle* readHandle);
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SExplainExecInfo** pRes, int32_t* capacity,
int32_t* resNum);
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SArray* pExecInfoList);
int32_t aggDecodeResultRow(SOperatorInfo* pOperator, char* result);
int32_t aggEncodeResultRow(SOperatorInfo* pOperator, char** result, int32_t* length);
@ -1019,6 +1025,7 @@ SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanN
void copyUpdateDataBlock(SSDataBlock* pDest, SSDataBlock* pSource, int32_t tsColIndex);
bool groupbyTbname(SNodeList* pGroupList);
int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle, SNodeList* groupKey);
SSDataBlock* createSpecialDataBlock(EStreamType type);
void* destroySqlFunctionCtx(SqlFunctionCtx* pCtx, int32_t numOfOutput);

4
source/libs/executor/inc/tfill.h
View File

@ -74,9 +74,9 @@ void taosFillSetInputDataBlock(struct SFillInfo* pFillInfo, const struct SSDataB
struct SFillColInfo* createFillColInfo(SExprInfo* pExpr, int32_t numOfOutput, const struct SNodeListNode* val);
bool taosFillHasMoreResults(struct SFillInfo* pFillInfo);
SFillInfo* taosCreateFillInfo(int32_t order, TSKEY skey, int32_t numOfTags, int32_t capacity, int32_t numOfCols,
SFillInfo* taosCreateFillInfo(TSKEY skey, int32_t numOfTags, int32_t capacity, int32_t numOfCols,
SInterval* pInterval, int32_t fillType, struct SFillColInfo* pCol, int32_t slotId,
const char* id);
int32_t order, const char* id);
void* taosDestroyFillInfo(struct SFillInfo *pFillInfo);
int64_t taosFillResultDataBlock(struct SFillInfo* pFillInfo, SSDataBlock* p, int32_t capacity);

2
source/libs/executor/src/executil.c
View File

@ -137,7 +137,7 @@ void initMultiResInfoFromArrayList(SGroupResInfo* pGroupResInfo, SArray* pArrayL
ASSERT(pGroupResInfo->index <= getNumOfTotalRes(pGroupResInfo));
}
bool hasDataInGroupInfo(SGroupResInfo* pGroupResInfo) {
bool hasRemainResults(SGroupResInfo* pGroupResInfo) {
if (pGroupResInfo->pRows == NULL) {
return false;
}

18
source/libs/executor/src/executor.c
View File

@ -242,25 +242,17 @@ int32_t qUpdateQualifiedTableId(qTaskInfo_t tinfo, const SArray* tableIdList, bo
}
// todo refactor STableList
bool assignUid = false;
size_t bufLen = (pScanInfo->pGroupTags != NULL) ? getTableTagsBufLen(pScanInfo->pGroupTags) : 0;
char* keyBuf = NULL;
if (bufLen > 0) {
assignUid = groupbyTbname(pScanInfo->pGroupTags);
keyBuf = taosMemoryMalloc(bufLen);
if (keyBuf == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
}
}
bool assignUid = false;
if (LIST_LENGTH(pScanInfo->pGroupTags) > 0) {
SNode* p = nodesListGetNode(pScanInfo->pGroupTags, 0);
if (p->type == QUERY_NODE_FUNCTION) {
// partition by tbname/group by tbname
assignUid = (strcmp(((struct SFunctionNode*)p)->functionName, "tbname") == 0);
}
}
for (int32_t i = 0; i < taosArrayGetSize(qa); ++i) {
uint64_t* uid = taosArrayGet(qa, i);
STableKeyInfo keyInfo = {.uid = *uid, .groupId = 0};
@ -504,11 +496,9 @@ void qDestroyTask(qTaskInfo_t qTaskHandle) {
doDestroyTask(pTaskInfo);
}
int32_t qGetExplainExecInfo(qTaskInfo_t tinfo, int32_t* resNum, SExplainExecInfo** pRes) {
int32_t qGetExplainExecInfo(qTaskInfo_t tinfo, SArray* pExecInfoList) {
SExecTaskInfo* pTaskInfo = (SExecTaskInfo*)tinfo;
int32_t capacity = 0;
return getOperatorExplainExecInfo(pTaskInfo->pRoot, pRes, &capacity, resNum);
return getOperatorExplainExecInfo(pTaskInfo->pRoot, pExecInfoList);
}
int32_t qSerializeTaskStatus(qTaskInfo_t tinfo, char** pOutput, int32_t* len) {

110
source/libs/executor/src/executorimpl.c
View File

@ -141,8 +141,7 @@ static int32_t doCopyToSDataBlock(SExecTaskInfo* taskInfo, SSDataBlock* pBlock,
SqlFunctionCtx* pCtx, int32_t numOfExprs);
static void initCtxOutputBuffer(SqlFunctionCtx* pCtx, int32_t size);
static void doSetTableGroupOutputBuf(SOperatorInfo* pOperator, SAggOperatorInfo* pAggInfo, int32_t numOfOutput,
uint64_t groupId);
static void doSetTableGroupOutputBuf(SOperatorInfo* pOperator, int32_t numOfOutput, uint64_t groupId);
// setup the output buffer for each operator
static bool hasNull(SColumn* pColumn, SColumnDataAgg* pStatis) {
@ -1393,10 +1392,11 @@ void extractQualifiedTupleByFilterResult(SSDataBlock* pBlock, const int8_t* rowR
}
}
void doSetTableGroupOutputBuf(SOperatorInfo* pOperator, SAggOperatorInfo* pAggInfo, int32_t numOfOutput,
uint64_t groupId) {
void doSetTableGroupOutputBuf(SOperatorInfo* pOperator, int32_t numOfOutput, uint64_t groupId) {
// for simple group by query without interval, all the tables belong to one group result.
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
SAggOperatorInfo* pAggInfo = pOperator->info;
SResultRowInfo* pResultRowInfo = &pAggInfo->binfo.resultRowInfo;
SqlFunctionCtx* pCtx = pOperator->exprSupp.pCtx;
int32_t* rowEntryInfoOffset = pOperator->exprSupp.rowEntryInfoOffset;
@ -1420,14 +1420,13 @@ void doSetTableGroupOutputBuf(SOperatorInfo* pOperator, SAggOperatorInfo* pAggIn
setResultRowInitCtx(pResultRow, pCtx, numOfOutput, rowEntryInfoOffset);
}
void setExecutionContext(SOperatorInfo* pOperator, int32_t numOfOutput, uint64_t groupId, SAggOperatorInfo* pAggInfo) {
if (pAggInfo->groupId != INT32_MIN && pAggInfo->groupId == groupId) {
static void setExecutionContext(SOperatorInfo* pOperator, int32_t numOfOutput, uint64_t groupId) {
SAggOperatorInfo* pAggInfo = pOperator->info;
if (pAggInfo->groupId != UINT64_MAX && pAggInfo->groupId == groupId) {
return;
}
#ifdef BUF_PAGE_DEBUG
qDebug("page_setbuf, groupId:%" PRIu64, groupId);
#endif
doSetTableGroupOutputBuf(pOperator, pAggInfo, numOfOutput, groupId);
doSetTableGroupOutputBuf(pOperator, numOfOutput, groupId);
// record the current active group id
pAggInfo->groupId = groupId;
@ -1594,7 +1593,7 @@ void doBuildResultDatablock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SG
pBlock->info.version = pTaskInfo->version;
blockDataCleanup(pBlock);
if (!hasDataInGroupInfo(pGroupResInfo)) {
if (!hasRemainResults(pGroupResInfo)) {
return;
}
@ -2931,7 +2930,7 @@ static int32_t doOpenAggregateOptr(SOperatorInfo* pOperator) {
}
// the pDataBlock are always the same one, no need to call this again
setExecutionContext(pOperator, pOperator->exprSupp.numOfExprs, pBlock->info.groupId, pAggInfo);
setExecutionContext(pOperator, pOperator->exprSupp.numOfExprs, pBlock->info.groupId);
setInputDataBlock(pOperator, pSup->pCtx, pBlock, order, scanFlag, true);
code = doAggregateImpl(pOperator, pSup->pCtx);
if (code != 0) {
@ -2966,7 +2965,7 @@ static SSDataBlock* getAggregateResult(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, pInfo, &pAggInfo->groupResInfo, pAggInfo->aggSup.pResultBuf);
doFilter(pAggInfo->pCondition, pInfo->pRes, NULL);
if (!hasDataInGroupInfo(&pAggInfo->groupResInfo)) {
if (!hasRemainResults(&pAggInfo->groupResInfo)) {
doSetOperatorCompleted(pOperator);
break;
}
@ -3501,7 +3500,7 @@ SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SExprInfo*
goto _error;
}
pInfo->groupId = INT32_MIN;
pInfo->groupId = UINT64_MAX;
pInfo->pCondition = pCondition;
pOperator->name = "TableAggregate";
pOperator->operatorType = QUERY_NODE_PHYSICAL_PLAN_HASH_AGG;
@ -3513,6 +3512,12 @@ SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SExprInfo*
pOperator->fpSet = createOperatorFpSet(doOpenAggregateOptr, getAggregateResult, NULL, NULL, destroyAggOperatorInfo,
aggEncodeResultRow, aggDecodeResultRow, NULL);
if (downstream->operatorType == QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
STableScanInfo* pTableScanInfo = downstream->info;
pTableScanInfo->pdInfo.pExprSup = &pOperator->exprSupp;
pTableScanInfo->pdInfo.pAggSup = &pInfo->aggSup;
}
code = appendDownstream(pOperator, &downstream, 1);
if (code != TSDB_CODE_SUCCESS) {
goto _error;
@ -3586,15 +3591,14 @@ void doDestroyExchangeOperatorInfo(void* param) {
}
static int32_t initFillInfo(SFillOperatorInfo* pInfo, SExprInfo* pExpr, int32_t numOfCols, SNodeListNode* pValNode,
STimeWindow win, int32_t capacity, const char* id, SInterval* pInterval, int32_t fillType) {
STimeWindow win, int32_t capacity, const char* id, SInterval* pInterval, int32_t fillType, int32_t order) {
SFillColInfo* pColInfo = createFillColInfo(pExpr, numOfCols, pValNode);
STimeWindow w = getAlignQueryTimeWindow(pInterval, pInterval->precision, win.skey);
w = getFirstQualifiedTimeWindow(win.skey, &w, pInterval, TSDB_ORDER_ASC);
int32_t order = TSDB_ORDER_ASC;
pInfo->pFillInfo =
taosCreateFillInfo(order, w.skey, 0, capacity, numOfCols, pInterval, fillType, pColInfo, pInfo->primaryTsCol, id);
taosCreateFillInfo(w.skey, 0, capacity, numOfCols, pInterval, fillType, pColInfo, pInfo->primaryTsCol, order, id);
pInfo->win = win;
pInfo->p = taosMemoryCalloc(numOfCols, POINTER_BYTES);
@ -3624,6 +3628,7 @@ SOperatorInfo* createFillOperatorInfo(SOperatorInfo* downstream, SFillPhysiNode*
? &((SMergeAlignedIntervalAggOperatorInfo*)downstream->info)->intervalAggOperatorInfo->interval
: &((SIntervalAggOperatorInfo*)downstream->info)->interval;
int32_t order = (pPhyFillNode->inputTsOrder == ORDER_ASC)? TSDB_ORDER_ASC:TSDB_ORDER_DESC;
int32_t type = convertFillType(pPhyFillNode->mode);
SResultInfo* pResultInfo = &pOperator->resultInfo;
@ -3635,7 +3640,7 @@ SOperatorInfo* createFillOperatorInfo(SOperatorInfo* downstream, SFillPhysiNode*
&numOfOutputCols, COL_MATCH_FROM_SLOT_ID);
int32_t code = initFillInfo(pInfo, pExprInfo, num, (SNodeListNode*)pPhyFillNode->pValues, pPhyFillNode->timeRange,
pResultInfo->capacity, pTaskInfo->id.str, pInterval, type);
pResultInfo->capacity, pTaskInfo->id.str, pInterval, type, order);
if (code != TSDB_CODE_SUCCESS) {
goto _error;
}
@ -3825,6 +3830,19 @@ static int32_t sortTableGroup(STableListInfo* pTableListInfo, int32_t groupNum)
return TDB_CODE_SUCCESS;
}
bool groupbyTbname(SNodeList* pGroupList) {
bool bytbname = false;
if (LIST_LENGTH(pGroupList) > 0) {
SNode* p = nodesListGetNode(pGroupList, 0);
if (p->type == QUERY_NODE_FUNCTION) {
// partition by tbname/group by tbname
bytbname = (strcmp(((struct SFunctionNode*)p)->functionName, "tbname") == 0);
}
}
return bytbname;
}
int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle, SNodeList* group) {
if (group == NULL) {
return TDB_CODE_SUCCESS;
@ -3851,12 +3869,21 @@ int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle,
return TSDB_CODE_OUT_OF_MEMORY;
}
bool assignUid = groupbyTbname(group);
int32_t groupNum = 0;
for (int32_t i = 0; i < taosArrayGetSize(pTableListInfo->pTableList); i++) {
size_t numOfTables = taosArrayGetSize(pTableListInfo->pTableList);
for (int32_t i = 0; i < numOfTables; i++) {
STableKeyInfo* info = taosArrayGet(pTableListInfo->pTableList, i);
int32_t code = getGroupIdFromTagsVal(pHandle->meta, info->uid, group, keyBuf, &info->groupId);
if (code != TSDB_CODE_SUCCESS) {
return code;
if (assignUid) {
info->groupId = info->uid;
} else {
int32_t code = getGroupIdFromTagsVal(pHandle->meta, info->uid, group, keyBuf, &info->groupId);
if (code != TSDB_CODE_SUCCESS) {
return code;
}
}
taosHashPut(pTableListInfo->map, &(info->uid), sizeof(uint64_t), &info->groupId, sizeof(uint64_t));
@ -4586,42 +4613,29 @@ void releaseQueryBuf(size_t numOfTables) {
atomic_add_fetch_64(&tsQueryBufferSizeBytes, t);
}
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SExplainExecInfo** pRes, int32_t* capacity,
int32_t* resNum) {
if (*resNum >= *capacity) {
*capacity += 10;
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SArray* pExecInfoList) {
SExplainExecInfo execInfo = {0};
SExplainExecInfo* pExplainInfo = taosArrayPush(pExecInfoList, &execInfo);
*pRes = taosMemoryRealloc(*pRes, (*capacity) * sizeof(SExplainExecInfo));
if (NULL == *pRes) {
qError("malloc %d failed", (*capacity) * (int32_t)sizeof(SExplainExecInfo));
return TSDB_CODE_QRY_OUT_OF_MEMORY;
}
}
SExplainExecInfo* pInfo = &(*pRes)[*resNum];
pInfo->numOfRows = operatorInfo->resultInfo.totalRows;
pInfo->startupCost = operatorInfo->cost.openCost;
pInfo->totalCost = operatorInfo->cost.totalCost;
pExplainInfo->numOfRows = operatorInfo->resultInfo.totalRows;
pExplainInfo->startupCost = operatorInfo->cost.openCost;
pExplainInfo->totalCost = operatorInfo->cost.totalCost;
pExplainInfo->verboseLen = 0;
pExplainInfo->verboseInfo = NULL;
if (operatorInfo->fpSet.getExplainFn) {
int32_t code = operatorInfo->fpSet.getExplainFn(operatorInfo, &pInfo->verboseInfo, &pInfo->verboseLen);
int32_t code = operatorInfo->fpSet.getExplainFn(operatorInfo, &pExplainInfo->verboseInfo, &pExplainInfo->verboseLen);
if (code) {
qError("%s operator getExplainFn failed, code:%s", GET_TASKID(operatorInfo->pTaskInfo), tstrerror(code));
return code;
}
} else {
pInfo->verboseLen = 0;
pInfo->verboseInfo = NULL;
}
++(*resNum);
int32_t code = 0;
for (int32_t i = 0; i < operatorInfo->numOfDownstream; ++i) {
code = getOperatorExplainExecInfo(operatorInfo->pDownstream[i], pRes, capacity, resNum);
if (code) {
taosMemoryFreeClear(*pRes);
code = getOperatorExplainExecInfo(operatorInfo->pDownstream[i], pExecInfoList);
if (code != TSDB_CODE_SUCCESS) {
// taosMemoryFreeClear(*pRes);
return TSDB_CODE_QRY_OUT_OF_MEMORY;
}
}

14
source/libs/executor/src/groupoperator.c
View File

@ -31,14 +31,21 @@ static int32_t* setupColumnOffset(const SSDataBlock* pBlock, int32_t rowCapacity
static int32_t setGroupResultOutputBuf(SOperatorInfo* pOperator, SOptrBasicInfo* binfo, int32_t numOfCols, char* pData, int16_t bytes,
uint64_t groupId, SDiskbasedBuf* pBuf, SAggSupporter* pAggSup);
static void freeGroupKey(void* param) {
SGroupKeys* pKey = (SGroupKeys*) param;
taosMemoryFree(pKey->pData);
}
static void destroyGroupOperatorInfo(void* param, int32_t numOfOutput) {
SGroupbyOperatorInfo* pInfo = (SGroupbyOperatorInfo*)param;
cleanupBasicInfo(&pInfo->binfo);
taosMemoryFreeClear(pInfo->keyBuf);
taosArrayDestroy(pInfo->pGroupCols);
taosArrayDestroy(pInfo->pGroupColVals);
taosArrayDestroyEx(pInfo->pGroupColVals, freeGroupKey);
cleanupExprSupp(&pInfo->scalarSup);
cleanupGroupResInfo(&pInfo->groupResInfo);
cleanupAggSup(&pInfo->aggSup);
taosMemoryFreeClear(param);
}
@ -301,8 +308,7 @@ static SSDataBlock* buildGroupResultDataBlock(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
doFilter(pInfo->pCondition, pRes, NULL);
bool hasRemain = hasDataInGroupInfo(&pInfo->groupResInfo);
if (!hasRemain) {
if (!hasRemainResults(&pInfo->groupResInfo)) {
doSetOperatorCompleted(pOperator);
break;
}
@ -415,8 +421,6 @@ SOperatorInfo* createGroupOperatorInfo(SOperatorInfo* downstream, SExprInfo* pEx
pOperator->blocking = true;
pOperator->status = OP_NOT_OPENED;
// pOperator->operatorType = OP_Groupby;
pOperator->exprSupp.pExprInfo = pExprInfo;
pOperator->exprSupp.numOfExprs = numOfCols;
pOperator->info = pInfo;
pOperator->pTaskInfo = pTaskInfo;

8
source/libs/executor/src/joinoperator.c
View File

@ -77,11 +77,11 @@ SOperatorInfo* createMergeJoinOperatorInfo(SOperatorInfo** pDownstream, int32_t
pInfo->pCondAfterMerge = NULL;
}
pInfo->inputTsOrder = TSDB_ORDER_ASC;
pInfo->inputOrder = TSDB_ORDER_ASC;
if (pJoinNode->inputTsOrder == ORDER_ASC) {
pInfo->inputTsOrder = TSDB_ORDER_ASC;
pInfo->inputOrder = TSDB_ORDER_ASC;
} else if (pJoinNode->inputTsOrder == ORDER_DESC) {
pInfo->inputTsOrder = TSDB_ORDER_DESC;
pInfo->inputOrder = TSDB_ORDER_DESC;
}
pOperator->fpSet =
@ -312,7 +312,7 @@ static void doMergeJoinImpl(struct SOperatorInfo* pOperator, SSDataBlock* pRes)
int32_t nrows = pRes->info.rows;
bool asc = (pJoinInfo->inputTsOrder == TSDB_ORDER_ASC) ? true : false;
bool asc = (pJoinInfo->inputOrder == TSDB_ORDER_ASC) ? true : false;
while (1) {
int64_t leftTs = 0;

126
source/libs/executor/src/scanoperator.c
View File

@ -166,6 +166,67 @@ static bool overlapWithTimeWindow(SInterval* pInterval, SDataBlockInfo* pBlockIn
return false;
}
// this function is for table scanner to extract temporary results of upstream aggregate results.
static SResultRow* getTableGroupOutputBuf(SOperatorInfo* pOperator, uint64_t groupId, SFilePage** pPage) {
if (pOperator->operatorType != QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
return NULL;
}
int64_t buf[2] = {0};
SET_RES_WINDOW_KEY((char*)buf, &groupId, sizeof(groupId), groupId);
STableScanInfo* pTableScanInfo = pOperator->info;
SResultRowPosition* p1 =
(SResultRowPosition*)taosHashGet(pTableScanInfo->pdInfo.pAggSup->pResultRowHashTable, buf, GET_RES_WINDOW_KEY_LEN(sizeof(groupId)));
if (p1 == NULL) {
return NULL;
}
*pPage = getBufPage(pTableScanInfo->pdInfo.pAggSup->pResultBuf, p1->pageId);
return (SResultRow*)((char*)(*pPage) + p1->offset);
}
static int32_t doDynamicPruneDataBlock(SOperatorInfo* pOperator, SDataBlockInfo* pBlockInfo, uint32_t* status) {
STableScanInfo* pTableScanInfo = pOperator->info;
if (pTableScanInfo->pdInfo.pExprSup == NULL) {
return TSDB_CODE_SUCCESS;
}
SExprSupp* pSup1 = pTableScanInfo->pdInfo.pExprSup;
SFilePage* pPage = NULL;
SResultRow* pRow = getTableGroupOutputBuf(pOperator, pBlockInfo->groupId, &pPage);
if (pRow == NULL) {
return TSDB_CODE_SUCCESS;
}
bool notLoadBlock = true;
for (int32_t i = 0; i < pSup1->numOfExprs; ++i) {
int32_t functionId = pSup1->pCtx[i].functionId;
SResultRowEntryInfo* pEntry = getResultEntryInfo(pRow, i, pTableScanInfo->pdInfo.pExprSup->rowEntryInfoOffset);
int32_t reqStatus = fmFuncDynDataRequired(functionId, pEntry, &pBlockInfo->window);
if (reqStatus != FUNC_DATA_REQUIRED_NOT_LOAD) {
notLoadBlock = false;
break;
}
}
// release buffer pages
releaseBufPage(pTableScanInfo->pdInfo.pAggSup->pResultBuf, pPage);
if (notLoadBlock) {
*status = FUNC_DATA_REQUIRED_NOT_LOAD;
}
return TSDB_CODE_SUCCESS;
}
static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanInfo* pTableScanInfo, SSDataBlock* pBlock,
uint32_t* status) {
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
@ -178,7 +239,7 @@ static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanInfo* pTableSca
*status = pInfo->dataBlockLoadFlag;
if (pTableScanInfo->pFilterNode != NULL ||
overlapWithTimeWindow(&pTableScanInfo->interval, &pBlock->info, pTableScanInfo->cond.order)) {
overlapWithTimeWindow(&pTableScanInfo->pdInfo.interval, &pBlock->info, pTableScanInfo->cond.order)) {
(*status) = FUNC_DATA_REQUIRED_DATA_LOAD;
}
@ -232,6 +293,16 @@ static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanInfo* pTableSca
ASSERT(*status == FUNC_DATA_REQUIRED_DATA_LOAD);
// todo filter data block according to the block sma data firstly
doDynamicPruneDataBlock(pOperator, pBlockInfo, status);
if (*status == FUNC_DATA_REQUIRED_NOT_LOAD) {
qDebug("%s data block skipped, brange:%" PRId64 "-%" PRId64 ", rows:%d", GET_TASKID(pTaskInfo),
pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows);
pCost->skipBlocks += 1;
return TSDB_CODE_SUCCESS;
}
#if 0
if (!doFilterByBlockStatistics(pBlock->pBlockStatis, pTableScanInfo->pCtx, pBlockInfo->rows)) {
pCost->filterOutBlocks += 1;
@ -263,18 +334,20 @@ static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanInfo* pTableSca
}
}
int64_t st = taosGetTimestampMs();
doFilter(pTableScanInfo->pFilterNode, pBlock, pTableScanInfo->pColMatchInfo);
if (pTableScanInfo->pFilterNode != NULL) {
int64_t st = taosGetTimestampUs();
doFilter(pTableScanInfo->pFilterNode, pBlock, pTableScanInfo->pColMatchInfo);
int64_t et = taosGetTimestampMs();
pTableScanInfo->readRecorder.filterTime += (et - st);
double el = (taosGetTimestampUs() - st) / 1000.0;
pTableScanInfo->readRecorder.filterTime += el;
if (pBlock->info.rows == 0) {
pCost->filterOutBlocks += 1;
qDebug("%s data block filter out, brange:%" PRId64 "-%" PRId64 ", rows:%d", GET_TASKID(pTaskInfo),
pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows);
} else {
qDebug("%s data block filter out, elapsed time:%" PRId64, GET_TASKID(pTaskInfo), (et - st));
if (pBlock->info.rows == 0) {
pCost->filterOutBlocks += 1;
qDebug("%s data block filter out, brange:%" PRId64 "-%" PRId64 ", rows:%d, elapsed time:%.2f ms",
GET_TASKID(pTaskInfo), pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows, el);
} else {
qDebug("%s data block filter applied, elapsed time:%.2f ms", GET_TASKID(pTaskInfo), el);
}
}
return TSDB_CODE_SUCCESS;
@ -607,10 +680,11 @@ SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode,
}
pInfo->scanInfo = (SScanInfo){.numOfAsc = pTableScanNode->scanSeq[0], .numOfDesc = pTableScanNode->scanSeq[1]};
// pInfo->scanInfo = (SScanInfo){.numOfAsc = 0, .numOfDesc = 1}; // for debug purpose
// pInfo->scanInfo = (SScanInfo){.numOfAsc = 0, .numOfDesc = 1}; // for debug purpose
// pInfo->cond.order = TSDB_ORDER_DESC;
pInfo->pdInfo.interval = extractIntervalInfo(pTableScanNode);
pInfo->readHandle = *readHandle;
pInfo->interval = extractIntervalInfo(pTableScanNode);
pInfo->sample.sampleRatio = pTableScanNode->ratio;
pInfo->sample.seed = taosGetTimestampSec();
@ -1489,14 +1563,14 @@ SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle, STableScanPhys
pInfo->tqReader = pHandle->tqReader;
}
if (pTSInfo->interval.interval > 0) {
pInfo->pUpdateInfo = updateInfoInitP(&pTSInfo->interval, pInfo->twAggSup.waterMark);
if (pTSInfo->pdInfo.interval.interval > 0) {
pInfo->pUpdateInfo = updateInfoInitP(&pTSInfo->pdInfo.interval, pInfo->twAggSup.waterMark);
} else {
pInfo->pUpdateInfo = NULL;
}
pInfo->pTableScanOp = pTableScanOp;
pInfo->interval = pTSInfo->interval;
pInfo->interval = pTSInfo->pdInfo.interval;
pInfo->readHandle = *pHandle;
pInfo->tableUid = pScanPhyNode->uid;
@ -2672,16 +2746,20 @@ static int32_t loadDataBlockFromOneTable(SOperatorInfo* pOperator, STableMergeSc
}
}
int64_t st = taosGetTimestampMs();
doFilter(pTableScanInfo->pFilterNode, pBlock, pTableScanInfo->pColMatchInfo);
if (pTableScanInfo->pFilterNode != NULL) {
int64_t st = taosGetTimestampMs();
doFilter(pTableScanInfo->pFilterNode, pBlock, pTableScanInfo->pColMatchInfo);
int64_t et = taosGetTimestampMs();
pTableScanInfo->readRecorder.filterTime += (et - st);
double el = (taosGetTimestampUs() - st) / 1000.0;
pTableScanInfo->readRecorder.filterTime += el;
if (pBlock->info.rows == 0) {
pCost->filterOutBlocks += 1;
qDebug("%s data block filter out, brange:%" PRId64 "-%" PRId64 ", rows:%d", GET_TASKID(pTaskInfo),
pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows);
if (pBlock->info.rows == 0) {
pCost->filterOutBlocks += 1;
qDebug("%s data block filter out, brange:%" PRId64 "-%" PRId64 ", rows:%d, elapsed time:%.2f ms",
GET_TASKID(pTaskInfo), pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows, el);
} else {
qDebug("%s data block filter applied, elapsed time:%.2f ms", GET_TASKID(pTaskInfo), el);
}
}
return TSDB_CODE_SUCCESS;

35
source/libs/executor/src/tfill.c
View File

@ -66,7 +66,7 @@ static void setNullRow(SSDataBlock* pBlock, int64_t ts, int32_t rowIndex) {
static void doSetVal(SColumnInfoData* pDstColInfoData, int32_t rowIndex, const SGroupKeys* pKey);
static void doSetUserSpecifiedValue(SColumnInfoData* pDst, SVariant* pVar, int32_t rowIndex, int64_t currentKey) {
static void doSetUserSpecifiedValue(SColumnInfoData* pDst, SVariant* pVar, int32_t rowIndex, int64_t currentKey) {
if (pDst->info.type == TSDB_DATA_TYPE_FLOAT) {
float v = 0;
GET_TYPED_DATA(v, float, pVar->nType, &pVar->i);
@ -184,7 +184,7 @@ static void doFillOneRow(SFillInfo* pFillInfo, SSDataBlock* pBlock, SSDataBlock*
continue;
}
SVariant* pVar = &pFillInfo->pFillCol[i].fillVal;
SVariant* pVar = &pFillInfo->pFillCol[i].fillVal;
SColumnInfoData* pDst = taosArrayGet(pBlock->pDataBlock, i);
doSetUserSpecifiedValue(pDst, pVar, index, pFillInfo->currentKey);
}
@ -298,7 +298,7 @@ static int32_t fillResultImpl(SFillInfo* pFillInfo, SSDataBlock* pBlock, int32_t
SColumnInfoData* pSrc = taosArrayGet(pFillInfo->pSrcBlock->pDataBlock, srcSlotId);
char* src = colDataGetData(pSrc, pFillInfo->index);
if (/*i == 0 || (*/!colDataIsNull_s(pSrc, pFillInfo->index)) {
if (/*i == 0 || (*/ !colDataIsNull_s(pSrc, pFillInfo->index)) {
bool isNull = colDataIsNull_s(pSrc, pFillInfo->index);
colDataAppend(pDst, index, src, isNull);
saveColData(pFillInfo->prev, i, src, isNull);
@ -313,7 +313,7 @@ static int32_t fillResultImpl(SFillInfo* pFillInfo, SSDataBlock* pBlock, int32_t
} else if (pFillInfo->type == TSDB_FILL_LINEAR) {
bool isNull = colDataIsNull_s(pSrc, pFillInfo->index);
colDataAppend(pDst, index, src, isNull);
saveColData(pFillInfo->prev, i, src, isNull); // todo:
saveColData(pFillInfo->prev, i, src, isNull); // todo:
} else if (pFillInfo->type == TSDB_FILL_NULL) {
colDataAppendNULL(pDst, index);
} else if (pFillInfo->type == TSDB_FILL_NEXT) {
@ -433,9 +433,9 @@ static int32_t taosNumOfRemainRows(SFillInfo* pFillInfo) {
return pFillInfo->numOfRows - pFillInfo->index;
}
struct SFillInfo* taosCreateFillInfo(int32_t order, TSKEY skey, int32_t numOfTags, int32_t capacity, int32_t numOfCols,
SInterval* pInterval, int32_t fillType, struct SFillColInfo* pCol, int32_t primaryTsSlotId,
const char* id) {
struct SFillInfo* taosCreateFillInfo(TSKEY skey, int32_t numOfTags, int32_t capacity, int32_t numOfCols,
SInterval* pInterval, int32_t fillType, struct SFillColInfo* pCol,
int32_t primaryTsSlotId, int32_t order, const char* id) {
if (fillType == TSDB_FILL_NONE) {
return NULL;
}
@ -446,10 +446,9 @@ struct SFillInfo* taosCreateFillInfo(int32_t order, TSKEY skey, int32_t numOfTag
return NULL;
}
pFillInfo->tsSlotId = primaryTsSlotId;
taosResetFillInfo(pFillInfo, skey);
pFillInfo->order = order;
pFillInfo->tsSlotId = primaryTsSlotId;
taosResetFillInfo(pFillInfo, skey);
switch (fillType) {
case FILL_MODE_NONE:
@ -535,6 +534,14 @@ void* taosDestroyFillInfo(SFillInfo* pFillInfo) {
return NULL;
}
void taosFillSetDataOrderInfo(SFillInfo* pFillInfo, int32_t order) {
if (pFillInfo == NULL || (order != TSDB_ORDER_ASC && order != TSDB_ORDER_DESC)) {
return;
}
pFillInfo->order = order;
}
void taosFillSetStartInfo(SFillInfo* pFillInfo, int32_t numOfRows, TSKEY endKey) {
if (pFillInfo->type == TSDB_FILL_NONE) {
return;
@ -581,7 +588,7 @@ int64_t getNumOfResultsAfterFillGap(SFillInfo* pFillInfo, TSKEY ekey, int32_t ma
int64_t numOfRes = -1;
if (numOfRows > 0) { // still fill gap within current data block, not generating data after the result set.
TSKEY lastKey = tsList[pFillInfo->numOfRows - 1];
TSKEY lastKey = (TSDB_ORDER_ASC == pFillInfo->order ? tsList[pFillInfo->numOfRows - 1] : tsList[0]);
numOfRes = taosTimeCountInterval(lastKey, pFillInfo->currentKey, pFillInfo->interval.sliding,
pFillInfo->interval.slidingUnit, pFillInfo->interval.precision);
numOfRes += 1;
@ -626,9 +633,9 @@ int64_t taosFillResultDataBlock(SFillInfo* pFillInfo, SSDataBlock* p, int32_t ca
}
qDebug("fill:%p, generated fill result, src block:%d, index:%d, brange:%" PRId64 "-%" PRId64 ", currentKey:%" PRId64
", current : % d, total : % d, %s", pFillInfo,
pFillInfo->numOfRows, pFillInfo->index, pFillInfo->start, pFillInfo->end, pFillInfo->currentKey,
pFillInfo->numOfCurrent, pFillInfo->numOfTotal, pFillInfo->id);
", current : % d, total : % d, %s",
pFillInfo, pFillInfo->numOfRows, pFillInfo->index, pFillInfo->start, pFillInfo->end, pFillInfo->currentKey,
pFillInfo->numOfCurrent, pFillInfo->numOfTotal, pFillInfo->id);
return numOfRes;
}

164
source/libs/executor/src/timewindowoperator.c
View File

@ -362,7 +362,7 @@ static void setNotInterpoWindowKey(SqlFunctionCtx* pCtx, int32_t numOfOutput, in
static bool setTimeWindowInterpolationStartTs(SIntervalAggOperatorInfo* pInfo, int32_t pos, SSDataBlock* pBlock,
const TSKEY* tsCols, STimeWindow* win, SExprSupp* pSup) {
bool ascQuery = (pInfo->order == TSDB_ORDER_ASC);
bool ascQuery = (pInfo->inputOrder == TSDB_ORDER_ASC);
TSKEY curTs = tsCols[pos];
@ -392,7 +392,7 @@ static bool setTimeWindowInterpolationStartTs(SIntervalAggOperatorInfo* pInfo, i
static bool setTimeWindowInterpolationEndTs(SIntervalAggOperatorInfo* pInfo, SExprSupp* pSup, int32_t endRowIndex,
SArray* pDataBlock, const TSKEY* tsCols, TSKEY blockEkey,
STimeWindow* win) {
int32_t order = pInfo->order;
int32_t order = pInfo->inputOrder;
TSKEY actualEndKey = tsCols[endRowIndex];
TSKEY key = (order == TSDB_ORDER_ASC) ? win->ekey : win->skey;
@ -550,7 +550,7 @@ static void doWindowBorderInterpolation(SIntervalAggOperatorInfo* pInfo, SSDataB
if (!done) {
int32_t endRowIndex = startPos + forwardRows - 1;
TSKEY endKey = (pInfo->order == TSDB_ORDER_ASC) ? pBlock->info.window.ekey : pBlock->info.window.skey;
TSKEY endKey = (pInfo->inputOrder == TSDB_ORDER_ASC) ? pBlock->info.window.ekey : pBlock->info.window.skey;
bool interp = setTimeWindowInterpolationEndTs(pInfo, pSup, endRowIndex, pBlock->pDataBlock, tsCols, endKey, win);
if (interp) {
setResultRowInterpo(pResult, RESULT_ROW_END_INTERP);
@ -639,7 +639,7 @@ static void doInterpUnclosedTimeWindow(SOperatorInfo* pOperatorInfo, int32_t num
setNotInterpoWindowKey(pSup->pCtx, numOfExprs, RESULT_ROW_START_INTERP);
doApplyFunctions(pTaskInfo, pSup->pCtx, &w, &pInfo->twAggSup.timeWindowData, startPos, 0, tsCols, pBlock->info.rows,
numOfExprs, pInfo->order);
numOfExprs, pInfo->inputOrder);
if (isResultRowInterpolated(pResult, RESULT_ROW_END_INTERP)) {
closeResultRow(pr);
@ -924,11 +924,11 @@ static void hashIntervalAgg(SOperatorInfo* pOperatorInfo, SResultRowInfo* pResul
int32_t numOfOutput = pSup->numOfExprs;
int64_t* tsCols = extractTsCol(pBlock, pInfo);
uint64_t tableGroupId = pBlock->info.groupId;
bool ascScan = (pInfo->order == TSDB_ORDER_ASC);
bool ascScan = (pInfo->inputOrder == TSDB_ORDER_ASC);
TSKEY ts = getStartTsKey(&pBlock->info.window, tsCols);
SResultRow* pResult = NULL;
STimeWindow win = getActiveTimeWindow(pInfo->aggSup.pResultBuf, pResultRowInfo, ts, &pInfo->interval, pInfo->order);
STimeWindow win = getActiveTimeWindow(pInfo->aggSup.pResultBuf, pResultRowInfo, ts, &pInfo->interval, pInfo->inputOrder);
int32_t ret = TSDB_CODE_SUCCESS;
if ((!pInfo->ignoreExpiredData || !isCloseWindow(&win, &pInfo->twAggSup)) &&
inSlidingWindow(&pInfo->interval, &win, &pBlock->info)) {
@ -946,7 +946,7 @@ static void hashIntervalAgg(SOperatorInfo* pOperatorInfo, SResultRowInfo* pResul
TSKEY ekey = ascScan ? win.ekey : win.skey;
int32_t forwardRows =
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, pInfo->order);
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, pInfo->inputOrder);
ASSERT(forwardRows > 0);
// prev time window not interpolation yet.
@ -969,7 +969,7 @@ static void hashIntervalAgg(SOperatorInfo* pOperatorInfo, SResultRowInfo* pResul
inSlidingWindow(&pInfo->interval, &win, &pBlock->info)) {
updateTimeWindowInfo(&pInfo->twAggSup.timeWindowData, &win, true);
doApplyFunctions(pTaskInfo, pSup->pCtx, &win, &pInfo->twAggSup.timeWindowData, startPos, forwardRows, tsCols,
pBlock->info.rows, numOfOutput, pInfo->order);
pBlock->info.rows, numOfOutput, pInfo->inputOrder);
}
doCloseWindow(pResultRowInfo, pInfo, pResult);
@ -977,14 +977,14 @@ static void hashIntervalAgg(SOperatorInfo* pOperatorInfo, SResultRowInfo* pResul
STimeWindow nextWin = win;
while (1) {
int32_t prevEndPos = forwardRows - 1 + startPos;
startPos = getNextQualifiedWindow(&pInfo->interval, &nextWin, &pBlock->info, tsCols, prevEndPos, pInfo->order);
startPos = getNextQualifiedWindow(&pInfo->interval, &nextWin, &pBlock->info, tsCols, prevEndPos, pInfo->inputOrder);
if (startPos < 0) {
break;
}
if (pInfo->ignoreExpiredData && isCloseWindow(&nextWin, &pInfo->twAggSup)) {
ekey = ascScan ? nextWin.ekey : nextWin.skey;
forwardRows =
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, pInfo->order);
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, pInfo->inputOrder);
continue;
}
@ -1002,14 +1002,14 @@ static void hashIntervalAgg(SOperatorInfo* pOperatorInfo, SResultRowInfo* pResul
ekey = ascScan ? nextWin.ekey : nextWin.skey;
forwardRows =
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, pInfo->order);
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, pInfo->inputOrder);
// window start(end) key interpolation
doWindowBorderInterpolation(pInfo, pBlock, pResult, &nextWin, startPos, forwardRows, pSup);
updateTimeWindowInfo(&pInfo->twAggSup.timeWindowData, &nextWin, true);
doApplyFunctions(pTaskInfo, pSup->pCtx, &nextWin, &pInfo->twAggSup.timeWindowData, startPos, forwardRows, tsCols,
pBlock->info.rows, numOfOutput, pInfo->order);
pBlock->info.rows, numOfOutput, pInfo->inputOrder);
doCloseWindow(pResultRowInfo, pInfo, pResult);
}
@ -1082,7 +1082,7 @@ static int32_t doOpenIntervalAgg(SOperatorInfo* pOperator) {
break;
}
getTableScanInfo(pOperator, &pInfo->order, &scanFlag);
getTableScanInfo(pOperator, &pInfo->inputOrder, &scanFlag);
if (pInfo->scalarSupp.pExprInfo != NULL) {
SExprSupp* pExprSup = &pInfo->scalarSupp;
@ -1090,13 +1090,13 @@ static int32_t doOpenIntervalAgg(SOperatorInfo* pOperator) {
}
// the pDataBlock are always the same one, no need to call this again
setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->order, scanFlag, true);
setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->inputOrder, scanFlag, true);
blockDataUpdateTsWindow(pBlock, pInfo->primaryTsIndex);
hashIntervalAgg(pOperator, &pInfo->binfo.resultRowInfo, pBlock, scanFlag, NULL);
}
initGroupedResultInfo(&pInfo->groupResInfo, pInfo->aggSup.pResultRowHashTable, pInfo->order);
initGroupedResultInfo(&pInfo->groupResInfo, pInfo->aggSup.pResultRowHashTable, pInfo->resultTsOrder);
OPTR_SET_OPENED(pOperator);
pOperator->cost.openCost = (taosGetTimestampUs() - st) / 1000.0;
@ -1218,7 +1218,7 @@ static SSDataBlock* doStateWindowAgg(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
doFilter(pInfo->pCondition, pBInfo->pRes, NULL);
bool hasRemain = hasDataInGroupInfo(&pInfo->groupResInfo);
bool hasRemain = hasRemainResults(&pInfo->groupResInfo);
if (!hasRemain) {
doSetOperatorCompleted(pOperator);
break;
@ -1249,7 +1249,6 @@ static SSDataBlock* doStateWindowAgg(SOperatorInfo* pOperator) {
}
pOperator->cost.openCost = (taosGetTimestampUs() - st) / 1000.0;
pOperator->status = OP_RES_TO_RETURN;
initGroupedResultInfo(&pInfo->groupResInfo, pInfo->aggSup.pResultRowHashTable, TSDB_ORDER_ASC);
@ -1258,7 +1257,7 @@ static SSDataBlock* doStateWindowAgg(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
doFilter(pInfo->pCondition, pBInfo->pRes, NULL);
bool hasRemain = hasDataInGroupInfo(&pInfo->groupResInfo);
bool hasRemain = hasRemainResults(&pInfo->groupResInfo);
if (!hasRemain) {
doSetOperatorCompleted(pOperator);
break;
@ -1295,7 +1294,7 @@ static SSDataBlock* doBuildIntervalResult(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
doFilter(pInfo->pCondition, pBlock, NULL);
bool hasRemain = hasDataInGroupInfo(&pInfo->groupResInfo);
bool hasRemain = hasRemainResults(&pInfo->groupResInfo);
if (!hasRemain) {
doSetOperatorCompleted(pOperator);
break;
@ -1550,7 +1549,7 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
SIntervalAggOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
pInfo->order = TSDB_ORDER_ASC;
pInfo->inputOrder = TSDB_ORDER_ASC;
SExprSupp* pSup = &pOperator->exprSupp;
if (pOperator->status == OP_EXEC_DONE) {
@ -1564,7 +1563,7 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
}
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
if (pInfo->binfo.pRes->info.rows == 0 || !hasDataInGroupInfo(&pInfo->groupResInfo)) {
if (pInfo->binfo.pRes->info.rows == 0 || !hasRemainResults(&pInfo->groupResInfo)) {
pOperator->status = OP_EXEC_DONE;
qDebug("===stream===single interval is done");
freeAllPages(pInfo->pRecycledPages, pInfo->aggSup.pResultBuf);
@ -1597,7 +1596,7 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
continue;
}
if (pBlock->info.type == STREAM_NORMAL) {
if (pBlock->info.type == STREAM_NORMAL && pBlock->info.version != 0) {
// set input version
pTaskInfo->version = pBlock->info.version;
}
@ -1610,7 +1609,7 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
// The timewindow that overlaps the timestamps of the input pBlock need to be recalculated and return to the
// caller. Note that all the time window are not close till now.
// the pDataBlock are always the same one, no need to call this again
setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->order, MAIN_SCAN, true);
setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->inputOrder, MAIN_SCAN, true);
if (pInfo->invertible) {
setInverFunction(pSup->pCtx, pOperator->exprSupp.numOfExprs, pBlock->info.type);
}
@ -1790,7 +1789,8 @@ SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo*
}
pInfo->win = pTaskInfo->window;
pInfo->order = TSDB_ORDER_ASC;
pInfo->inputOrder = (pPhyNode->window.inputTsOrder == ORDER_ASC)? TSDB_ORDER_ASC:TSDB_ORDER_DESC;
pInfo->resultTsOrder = (pPhyNode->window.outputTsOrder == ORDER_ASC)? TSDB_ORDER_ASC:TSDB_ORDER_DESC;
pInfo->interval = *pInterval;
pInfo->execModel = pTaskInfo->execModel;
pInfo->twAggSup = *pTwAggSupp;
@ -1807,7 +1807,6 @@ SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo*
}
pInfo->primaryTsIndex = primaryTsSlotId;
SExprSupp* pSup = &pOperator->exprSupp;
size_t keyBufSize = sizeof(int64_t) + sizeof(int64_t) + POINTER_BYTES;
@ -1879,7 +1878,7 @@ SOperatorInfo* createStreamIntervalOperatorInfo(SOperatorInfo* downstream, SExpr
goto _error;
}
pInfo->order = TSDB_ORDER_ASC;
pInfo->inputOrder = TSDB_ORDER_ASC;
pInfo->interval = *pInterval;
pInfo->execModel = OPTR_EXEC_MODEL_STREAM;
pInfo->win = pTaskInfo->window;
@ -2011,7 +2010,7 @@ static SSDataBlock* doSessionWindowAgg(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
doFilter(pInfo->pCondition, pBInfo->pRes, NULL);
bool hasRemain = hasDataInGroupInfo(&pInfo->groupResInfo);
bool hasRemain = hasRemainResults(&pInfo->groupResInfo);
if (!hasRemain) {
doSetOperatorCompleted(pOperator);
break;
@ -2054,7 +2053,7 @@ static SSDataBlock* doSessionWindowAgg(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
doFilter(pInfo->pCondition, pBInfo->pRes, NULL);
bool hasRemain = hasDataInGroupInfo(&pInfo->groupResInfo);
bool hasRemain = hasRemainResults(&pInfo->groupResInfo);
if (!hasRemain) {
doSetOperatorCompleted(pOperator);
break;
@ -2220,7 +2219,7 @@ static SSDataBlock* doTimeslice(SOperatorInfo* pOperator) {
// if (pOperator->status == OP_RES_TO_RETURN) {
// // doBuildResultDatablock(&pRuntimeEnv->groupResInfo, pRuntimeEnv, pIntervalInfo->pRes);
// if (pResBlock->info.rows == 0 || !hasDataInGroupInfo(&pSliceInfo->groupResInfo)) {
// if (pResBlock->info.rows == 0 || !hasRemainResults(&pSliceInfo->groupResInfo)) {
// doSetOperatorCompleted(pOperator);
// }
//
@ -3098,6 +3097,7 @@ SOperatorInfo* createStreamFinalIntervalOperatorInfo(SOperatorInfo* downstream,
pInfo->pDelRes = createSpecialDataBlock(STREAM_DELETE_RESULT);
pInfo->delIndex = 0;
pInfo->pDelWins = taosArrayInit(4, sizeof(SWinRes));
pInfo->pRecycledPages = taosArrayInit(4, sizeof(int32_t));
pOperator->operatorType = pPhyNode->type;
pOperator->blocking = true;
@ -3140,6 +3140,26 @@ void destroyStreamAggSupporter(SStreamAggSupporter* pSup) {
blockDataDestroy(pSup->pScanBlock);
}
void destroyStateWinInfo(void* ptr) {
if (ptr == NULL) {
return;
}
SStateWindowInfo* pWin = (SStateWindowInfo*) ptr;
taosMemoryFreeClear(pWin->stateKey.pData);
}
void destroyStateStreamAggSupporter(SStreamAggSupporter* pSup) {
taosMemoryFreeClear(pSup->pKeyBuf);
void** pIte = NULL;
while ((pIte = taosHashIterate(pSup->pResultRows, pIte)) != NULL) {
SArray* pWins = (SArray*)(*pIte);
taosArrayDestroyEx(pWins, (FDelete)destroyStateWinInfo);
}
taosHashCleanup(pSup->pResultRows);
destroyDiskbasedBuf(pSup->pResultBuf);
blockDataDestroy(pSup->pScanBlock);
}
void destroyStreamSessionAggOperatorInfo(void* param, int32_t numOfOutput) {
SStreamSessionAggOperatorInfo* pInfo = (SStreamSessionAggOperatorInfo*)param;
cleanupBasicInfo(&pInfo->binfo);
@ -3607,12 +3627,17 @@ static void doStreamSessionAggImpl(SOperatorInfo* pOperator, SSDataBlock* pSData
}
}
void deleteWindow(SArray* pWinInfos, int32_t index) {
void deleteWindow(SArray* pWinInfos, int32_t index, FDelete fp) {
ASSERT(index >= 0 && index < taosArrayGetSize(pWinInfos));
if (fp) {
void* ptr = taosArrayGet(pWinInfos, index);
fp(ptr);
}
taosArrayRemove(pWinInfos, index);
}
static void doDeleteTimeWindows(SStreamAggSupporter* pAggSup, SSDataBlock* pBlock, int64_t gap, SArray* result) {
static void doDeleteTimeWindows(SStreamAggSupporter* pAggSup, SSDataBlock* pBlock, int64_t gap,
SArray* result, FDelete fp) {
SColumnInfoData* pStartTsCol = taosArrayGet(pBlock->pDataBlock, START_TS_COLUMN_INDEX);
TSKEY* startDatas = (TSKEY*)pStartTsCol->pData;
SColumnInfoData* pEndTsCol = taosArrayGet(pBlock->pDataBlock, END_TS_COLUMN_INDEX);
@ -3626,7 +3651,7 @@ static void doDeleteTimeWindows(SStreamAggSupporter* pAggSup, SSDataBlock* pBloc
if (!pCurWin) {
break;
}
deleteWindow(pAggSup->pCurWins, winIndex);
deleteWindow(pAggSup->pCurWins, winIndex, fp);
if (result) {
taosArrayPush(result, pCurWin);
}
@ -3751,7 +3776,7 @@ SResultWindowInfo* getResWinForSession(void* pData) { return (SResultWindowInfo*
SResultWindowInfo* getResWinForState(void* pData) { return &((SStateWindowInfo*)pData)->winInfo; }
int32_t closeSessionWindow(SHashObj* pHashMap, STimeWindowAggSupp* pTwSup, SArray* pClosed, __get_win_info_ fn,
bool delete) {
bool delete, FDelete fp) {
// Todo(liuyao) save window to tdb
void** pIte = NULL;
size_t keyLen = 0;
@ -3773,7 +3798,7 @@ int32_t closeSessionWindow(SHashObj* pHashMap, STimeWindowAggSupp* pTwSup, SArra
pSeWin->isOutput = true;
}
if (delete) {
deleteWindow(pWins, i);
deleteWindow(pWins, i, fp);
i--;
size = taosArrayGetSize(pWins);
}
@ -3786,13 +3811,13 @@ int32_t closeSessionWindow(SHashObj* pHashMap, STimeWindowAggSupp* pTwSup, SArra
return TSDB_CODE_SUCCESS;
}
static void closeChildSessionWindow(SArray* pChildren, TSKEY maxTs, bool delete) {
static void closeChildSessionWindow(SArray* pChildren, TSKEY maxTs, bool delete, FDelete fp) {
int32_t size = taosArrayGetSize(pChildren);
for (int32_t i = 0; i < size; i++) {
SOperatorInfo* pChildOp = taosArrayGetP(pChildren, i);
SStreamSessionAggOperatorInfo* pChInfo = pChildOp->info;
pChInfo->twAggSup.maxTs = TMAX(pChInfo->twAggSup.maxTs, maxTs);
closeSessionWindow(pChInfo->streamAggSup.pResultRows, &pChInfo->twAggSup, NULL, getResWinForSession, delete);
closeSessionWindow(pChInfo->streamAggSup.pResultRows, &pChInfo->twAggSup, NULL, getResWinForSession, delete, fp);
}
}
@ -3835,7 +3860,7 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) {
return pInfo->pDelRes;
}
doBuildResultDatablock(pOperator, pBInfo, &pInfo->groupResInfo, pInfo->streamAggSup.pResultBuf);
if (pBInfo->pRes->info.rows == 0 || !hasDataInGroupInfo(&pInfo->groupResInfo)) {
if (pBInfo->pRes->info.rows == 0 || !hasRemainResults(&pInfo->groupResInfo)) {
doSetOperatorCompleted(pOperator);
}
printDataBlock(pBInfo->pRes, IS_FINAL_OP(pInfo) ? "final session" : "single session");
@ -3870,13 +3895,13 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) {
} else if (pBlock->info.type == STREAM_DELETE_DATA || pBlock->info.type == STREAM_DELETE_RESULT) {
SArray* pWins = taosArrayInit(16, sizeof(SResultWindowInfo));
// gap must be 0
doDeleteTimeWindows(&pInfo->streamAggSup, pBlock, 0, pWins);
doDeleteTimeWindows(&pInfo->streamAggSup, pBlock, 0, pWins, NULL);
if (IS_FINAL_OP(pInfo)) {
int32_t childIndex = getChildIndex(pBlock);
SOperatorInfo* pChildOp = taosArrayGetP(pInfo->pChildren, childIndex);
SStreamSessionAggOperatorInfo* pChildInfo = pChildOp->info;
// gap must be 0
doDeleteTimeWindows(&pChildInfo->streamAggSup, pBlock, 0, NULL);
doDeleteTimeWindows(&pChildInfo->streamAggSup, pBlock, 0, NULL, NULL);
rebuildTimeWindow(pInfo, pWins, pBlock->info.groupId, pOperator->exprSupp.numOfExprs, pOperator);
}
copyDeleteWindowInfo(pWins, pInfo->pStDeleted);
@ -3918,8 +3943,8 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) {
pOperator->status = OP_RES_TO_RETURN;
closeSessionWindow(pInfo->streamAggSup.pResultRows, &pInfo->twAggSup, pUpdated, getResWinForSession,
pInfo->ignoreExpiredData);
closeChildSessionWindow(pInfo->pChildren, pInfo->twAggSup.maxTs, pInfo->ignoreExpiredData);
pInfo->ignoreExpiredData, NULL);
closeChildSessionWindow(pInfo->pChildren, pInfo->twAggSup.maxTs, pInfo->ignoreExpiredData, NULL);
copyUpdateResult(pStUpdated, pUpdated);
taosHashCleanup(pStUpdated);
@ -4014,7 +4039,7 @@ static SSDataBlock* doStreamSessionSemiAgg(SOperatorInfo* pOperator) {
break;
} else if (pBlock->info.type == STREAM_DELETE_DATA || pBlock->info.type == STREAM_DELETE_RESULT) {
// gap must be 0
doDeleteTimeWindows(&pInfo->streamAggSup, pBlock, 0, NULL);
doDeleteTimeWindows(&pInfo->streamAggSup, pBlock, 0, NULL, NULL);
copyDataBlock(pInfo->pDelRes, pBlock);
pInfo->pDelRes->info.type = STREAM_DELETE_RESULT;
break;
@ -4120,7 +4145,7 @@ _error:
void destroyStreamStateOperatorInfo(void* param, int32_t numOfOutput) {
SStreamStateAggOperatorInfo* pInfo = (SStreamStateAggOperatorInfo*)param;
cleanupBasicInfo(&pInfo->binfo);
destroyStreamAggSupporter(&pInfo->streamAggSup);
destroyStateStreamAggSupporter(&pInfo->streamAggSup);
cleanupGroupResInfo(&pInfo->groupResInfo);
if (pInfo->pChildren != NULL) {
int32_t size = taosArrayGetSize(pInfo->pChildren);
@ -4132,6 +4157,10 @@ void destroyStreamStateOperatorInfo(void* param, int32_t numOfOutput) {
taosMemoryFreeClear(pChInfo);
}
}
colDataDestroy(&pInfo->twAggSup.timeWindowData);
blockDataDestroy(pInfo->pDelRes);
taosHashCleanup(pInfo->pSeDeleted);
destroySqlFunctionCtx(pInfo->pDummyCtx, 0);
taosMemoryFreeClear(param);
}
@ -4314,7 +4343,7 @@ static void doClearStateWindows(SStreamAggSupporter* pAggSup, SSDataBlock* pBloc
pSeDeleted);
ASSERT(isTsInWindow(pCurWin, tsCol[i]) || isEqualStateKey(pCurWin, pKeyData));
taosHashRemove(pSeUpdated, &pCurWin->winInfo.pos, sizeof(SResultRowPosition));
deleteWindow(pAggSup->pCurWins, winIndex);
deleteWindow(pAggSup->pCurWins, winIndex, destroyStateWinInfo);
}
}
@ -4357,7 +4386,7 @@ static void doStreamStateAggImpl(SOperatorInfo* pOperator, SSDataBlock* pSDataBl
appendOneRow(pAggSup->pScanBlock, &pCurWin->winInfo.win.skey, &pCurWin->winInfo.win.ekey,
&pSDataBlock->info.groupId);
taosHashRemove(pSeUpdated, &pCurWin->winInfo.pos, sizeof(SResultRowPosition));
deleteWindow(pAggSup->pCurWins, winIndex);
deleteWindow(pAggSup->pCurWins, winIndex, destroyStateWinInfo);
continue;
}
code = doOneStateWindowAgg(pInfo, pSDataBlock, &pCurWin->winInfo, &pResult, i, winRows, numOfOutput, pOperator);
@ -4391,7 +4420,7 @@ static SSDataBlock* doStreamStateAgg(SOperatorInfo* pOperator) {
return pInfo->pDelRes;
}
doBuildResultDatablock(pOperator, pBInfo, &pInfo->groupResInfo, pInfo->streamAggSup.pResultBuf);
if (pBInfo->pRes->info.rows == 0 || !hasDataInGroupInfo(&pInfo->groupResInfo)) {
if (pBInfo->pRes->info.rows == 0 || !hasRemainResults(&pInfo->groupResInfo)) {
doSetOperatorCompleted(pOperator);
}
printDataBlock(pBInfo->pRes, "single state");
@ -4415,7 +4444,7 @@ static SSDataBlock* doStreamStateAgg(SOperatorInfo* pOperator) {
continue;
} else if (pBlock->info.type == STREAM_DELETE_DATA) {
SArray* pWins = taosArrayInit(16, sizeof(SResultWindowInfo));
doDeleteTimeWindows(&pInfo->streamAggSup, pBlock, 0, pWins);
doDeleteTimeWindows(&pInfo->streamAggSup, pBlock, 0, pWins, destroyStateWinInfo);
copyDeleteWindowInfo(pWins, pInfo->pSeDeleted);
taosArrayDestroy(pWins);
continue;
@ -4437,8 +4466,8 @@ static SSDataBlock* doStreamStateAgg(SOperatorInfo* pOperator) {
pOperator->status = OP_RES_TO_RETURN;
closeSessionWindow(pInfo->streamAggSup.pResultRows, &pInfo->twAggSup, pUpdated, getResWinForState,
pInfo->ignoreExpiredData);
closeChildSessionWindow(pInfo->pChildren, pInfo->twAggSup.maxTs, pInfo->ignoreExpiredData);
pInfo->ignoreExpiredData, destroyStateWinInfo);
// closeChildSessionWindow(pInfo->pChildren, pInfo->twAggSup.maxTs, pInfo->ignoreExpiredData, destroyStateWinInfo);
copyUpdateResult(pSeUpdated, pUpdated);
taosHashCleanup(pSeUpdated);
@ -4564,7 +4593,7 @@ static int32_t outputMergeAlignedIntervalResult(SOperatorInfo* pOperatorInfo, ui
SExecTaskInfo* pTaskInfo = pOperatorInfo->pTaskInfo;
SExprSupp* pSup = &pOperatorInfo->exprSupp;
bool ascScan = (iaInfo->order == TSDB_ORDER_ASC);
bool ascScan = (iaInfo->inputOrder == TSDB_ORDER_ASC);
SET_RES_WINDOW_KEY(iaInfo->aggSup.keyBuf, &wstartTs, TSDB_KEYSIZE, tableGroupId);
SResultRowPosition* p1 = (SResultRowPosition*)taosHashGet(iaInfo->aggSup.pResultRowHashTable, iaInfo->aggSup.keyBuf,
@ -4618,7 +4647,7 @@ static void doMergeAlignedIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultR
} else {
updateTimeWindowInfo(&iaInfo->twAggSup.timeWindowData, &currWin, true);
doApplyFunctions(pTaskInfo, pSup->pCtx, &currWin, &iaInfo->twAggSup.timeWindowData, startPos, currPos - startPos,
tsCols, pBlock->info.rows, numOfOutput, iaInfo->order);
tsCols, pBlock->info.rows, numOfOutput, iaInfo->inputOrder);
outputMergeAlignedIntervalResult(pOperatorInfo, tableGroupId, pResultBlock, currTs);
@ -4637,7 +4666,7 @@ static void doMergeAlignedIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultR
}
updateTimeWindowInfo(&iaInfo->twAggSup.timeWindowData, &currWin, true);
doApplyFunctions(pTaskInfo, pSup->pCtx, &currWin, &iaInfo->twAggSup.timeWindowData, startPos, currPos - startPos,
tsCols, pBlock->info.rows, numOfOutput, iaInfo->order);
tsCols, pBlock->info.rows, numOfOutput, iaInfo->inputOrder);
outputMergeAlignedIntervalResult(pOperatorInfo, tableGroupId, pResultBlock, currTs);
}
@ -4682,8 +4711,8 @@ static SSDataBlock* doMergeAlignedIntervalAgg(SOperatorInfo* pOperator) {
break;
}
getTableScanInfo(pOperator, &iaInfo->order, &scanFlag);
setInputDataBlock(pOperator, pSup->pCtx, pBlock, iaInfo->order, scanFlag, true);
getTableScanInfo(pOperator, &iaInfo->inputOrder, &scanFlag);
setInputDataBlock(pOperator, pSup->pCtx, pBlock, iaInfo->inputOrder, scanFlag, true);
doMergeAlignedIntervalAggImpl(pOperator, &iaInfo->binfo.resultRowInfo, pBlock, scanFlag, pRes);
doFilter(miaInfo->pCondition, pRes, NULL);
if (pRes->info.rows >= pOperator->resultInfo.capacity) {
@ -4724,7 +4753,7 @@ SOperatorInfo* createMergeAlignedIntervalOperatorInfo(SOperatorInfo* downstream,
miaInfo->pCondition = pCondition;
iaInfo->win = pTaskInfo->window;
iaInfo->order = TSDB_ORDER_ASC;
iaInfo->inputOrder = TSDB_ORDER_ASC;
iaInfo->interval = *pInterval;
iaInfo->execModel = pTaskInfo->execModel;
iaInfo->primaryTsIndex = primaryTsSlotId;
@ -4806,7 +4835,7 @@ static int32_t finalizeWindowResult(SOperatorInfo* pOperatorInfo, uint64_t table
SMergeIntervalAggOperatorInfo* miaInfo = pOperatorInfo->info;
SIntervalAggOperatorInfo* iaInfo = &miaInfo->intervalAggOperatorInfo;
SExecTaskInfo* pTaskInfo = pOperatorInfo->pTaskInfo;
bool ascScan = (iaInfo->order == TSDB_ORDER_ASC);
bool ascScan = (iaInfo->inputOrder == TSDB_ORDER_ASC);
SExprSupp* pExprSup = &pOperatorInfo->exprSupp;
SET_RES_WINDOW_KEY(iaInfo->aggSup.keyBuf, &win->skey, TSDB_KEYSIZE, tableGroupId);
@ -4824,7 +4853,7 @@ static int32_t outputPrevIntervalResult(SOperatorInfo* pOperatorInfo, uint64_t t
SMergeIntervalAggOperatorInfo* miaInfo = pOperatorInfo->info;
SIntervalAggOperatorInfo* iaInfo = &miaInfo->intervalAggOperatorInfo;
SExecTaskInfo* pTaskInfo = pOperatorInfo->pTaskInfo;
bool ascScan = (iaInfo->order == TSDB_ORDER_ASC);
bool ascScan = (iaInfo->inputOrder == TSDB_ORDER_ASC);
SExprSupp* pExprSup = &pOperatorInfo->exprSupp;
SGroupTimeWindow groupTimeWindow = {.groupId = tableGroupId, .window = *newWin};
@ -4860,12 +4889,12 @@ static void doMergeIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultRowInfo*
int32_t numOfOutput = pExprSup->numOfExprs;
int64_t* tsCols = extractTsCol(pBlock, iaInfo);
uint64_t tableGroupId = pBlock->info.groupId;
bool ascScan = (iaInfo->order == TSDB_ORDER_ASC);
bool ascScan = (iaInfo->inputOrder == TSDB_ORDER_ASC);
TSKEY blockStartTs = getStartTsKey(&pBlock->info.window, tsCols);
SResultRow* pResult = NULL;
STimeWindow win =
getActiveTimeWindow(iaInfo->aggSup.pResultBuf, pResultRowInfo, blockStartTs, &iaInfo->interval, iaInfo->order);
getActiveTimeWindow(iaInfo->aggSup.pResultBuf, pResultRowInfo, blockStartTs, &iaInfo->interval, iaInfo->inputOrder);
int32_t ret =
setTimeWindowOutputBuf(pResultRowInfo, &win, (scanFlag == MAIN_SCAN), &pResult, tableGroupId, pExprSup->pCtx,
@ -4876,7 +4905,7 @@ static void doMergeIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultRowInfo*
TSKEY ekey = ascScan ? win.ekey : win.skey;
int32_t forwardRows =
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, iaInfo->order);
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, iaInfo->inputOrder);
ASSERT(forwardRows > 0);
// prev time window not interpolation yet.
@ -4897,7 +4926,7 @@ static void doMergeIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultRowInfo*
updateTimeWindowInfo(&iaInfo->twAggSup.timeWindowData, &win, true);
doApplyFunctions(pTaskInfo, pExprSup->pCtx, &win, &iaInfo->twAggSup.timeWindowData, startPos, forwardRows, tsCols,
pBlock->info.rows, numOfOutput, iaInfo->order);
pBlock->info.rows, numOfOutput, iaInfo->inputOrder);
doCloseWindow(pResultRowInfo, iaInfo, pResult);
// output previous interval results after this interval (&win) is closed
@ -4906,7 +4935,7 @@ static void doMergeIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultRowInfo*
STimeWindow nextWin = win;
while (1) {
int32_t prevEndPos = forwardRows - 1 + startPos;
startPos = getNextQualifiedWindow(&iaInfo->interval, &nextWin, &pBlock->info, tsCols, prevEndPos, iaInfo->order);
startPos = getNextQualifiedWindow(&iaInfo->interval, &nextWin, &pBlock->info, tsCols, prevEndPos, iaInfo->inputOrder);
if (startPos < 0) {
break;
}
@ -4921,14 +4950,14 @@ static void doMergeIntervalAggImpl(SOperatorInfo* pOperatorInfo, SResultRowInfo*
ekey = ascScan ? nextWin.ekey : nextWin.skey;
forwardRows =
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, iaInfo->order);
getNumOfRowsInTimeWindow(&pBlock->info, tsCols, startPos, ekey, binarySearchForKey, NULL, iaInfo->inputOrder);
// window start(end) key interpolation
doWindowBorderInterpolation(iaInfo, pBlock, pResult, &nextWin, startPos, forwardRows, pExprSup);
updateTimeWindowInfo(&iaInfo->twAggSup.timeWindowData, &nextWin, true);
doApplyFunctions(pTaskInfo, pExprSup->pCtx, &nextWin, &iaInfo->twAggSup.timeWindowData, startPos, forwardRows,
tsCols, pBlock->info.rows, numOfOutput, iaInfo->order);
tsCols, pBlock->info.rows, numOfOutput, iaInfo->inputOrder);
doCloseWindow(pResultRowInfo, iaInfo, pResult);
// output previous interval results after this interval (&nextWin) is closed
@ -4982,8 +5011,8 @@ static SSDataBlock* doMergeIntervalAgg(SOperatorInfo* pOperator) {
break;
}
getTableScanInfo(pOperator, &iaInfo->order, &scanFlag);
setInputDataBlock(pOperator, pExpSupp->pCtx, pBlock, iaInfo->order, scanFlag, true);
getTableScanInfo(pOperator, &iaInfo->inputOrder, &scanFlag);
setInputDataBlock(pOperator, pExpSupp->pCtx, pBlock, iaInfo->inputOrder, scanFlag, true);
doMergeIntervalAggImpl(pOperator, &iaInfo->binfo.resultRowInfo, pBlock, scanFlag, pRes);
if (pRes->info.rows >= pOperator->resultInfo.threshold) {
@ -5025,9 +5054,8 @@ SOperatorInfo* createMergeIntervalOperatorInfo(SOperatorInfo* downstream, SExprI
miaInfo->groupIntervals = tdListNew(sizeof(SGroupTimeWindow));
SIntervalAggOperatorInfo* iaInfo = &miaInfo->intervalAggOperatorInfo;
iaInfo->win = pTaskInfo->window;
iaInfo->order = TSDB_ORDER_ASC;
iaInfo->inputOrder = TSDB_ORDER_ASC;
iaInfo->interval = *pInterval;
iaInfo->execModel = pTaskInfo->execModel;

2
source/libs/function/inc/builtins.h
View File

@ -26,6 +26,7 @@ typedef int32_t (*FTranslateFunc)(SFunctionNode* pFunc, char* pErrBuf, int32_t l
typedef EFuncDataRequired (*FFuncDataRequired)(SFunctionNode* pFunc, STimeWindow* pTimeWindow);
typedef int32_t (*FCreateMergeFuncParameters)(SNodeList* pRawParameters, SNode* pPartialRes, SNodeList** pParameters);
typedef EFuncDataRequired (*FFuncDynDataRequired)(void* pRes, STimeWindow* pTimeWindow);
typedef EFuncReturnRows (*FEstimateReturnRows)(SFunctionNode* pFunc);
typedef struct SBuiltinFuncDefinition {
const char* name;
@ -44,6 +45,7 @@ typedef struct SBuiltinFuncDefinition {
const char* pPartialFunc;
const char* pMergeFunc;
FCreateMergeFuncParameters createMergeParaFuc;
FEstimateReturnRows estimateReturnRowsFunc;
} SBuiltinFuncDefinition;
extern const SBuiltinFuncDefinition funcMgtBuiltins[];

1
source/libs/function/inc/builtinsimpl.h
View File

@ -118,6 +118,7 @@ int32_t firstLastPartialFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
int32_t firstCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx);
int32_t lastCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx);
int32_t getFirstLastInfoSize(int32_t resBytes);
EFuncDataRequired lastDynDataReq(void* pRes, STimeWindow* pTimeWindow);
int32_t lastRowFunction(SqlFunctionCtx *pCtx);

1
source/libs/function/inc/functionMgtInt.h
View File

@ -48,6 +48,7 @@ extern "C" {
#define FUNC_MGT_CLIENT_PC_FUNC FUNC_MGT_FUNC_CLASSIFICATION_MASK(19)
#define FUNC_MGT_MULTI_ROWS_FUNC FUNC_MGT_FUNC_CLASSIFICATION_MASK(20)
#define FUNC_MGT_KEEP_ORDER_FUNC FUNC_MGT_FUNC_CLASSIFICATION_MASK(21)
#define FUNC_MGT_CUMULATIVE_FUNC FUNC_MGT_FUNC_CLASSIFICATION_MASK(22)
#define FUNC_MGT_TEST_MASK(val, mask) (((val) & (mask)) != 0)

49
source/libs/function/src/builtins.c
View File

@ -1277,6 +1277,8 @@ static int32_t translateCsum(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
return TSDB_CODE_SUCCESS;
}
static EFuncReturnRows csumEstReturnRows(SFunctionNode* pFunc) { return FUNC_RETURN_ROWS_N; }
static int32_t translateMavg(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
if (2 != LIST_LENGTH(pFunc->pParameterList)) {
return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
@ -1416,6 +1418,11 @@ static int32_t translateDerivative(SFunctionNode* pFunc, char* pErrBuf, int32_t
return TSDB_CODE_SUCCESS;
}
static EFuncReturnRows derivativeEstReturnRows(SFunctionNode* pFunc) {
return 1 == ((SValueNode*)nodesListGetNode(pFunc->pParameterList, 2))->datum.i ? FUNC_RETURN_ROWS_INDEFINITE
: FUNC_RETURN_ROWS_N_MINUS_1;
}
static int32_t translateIrate(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
if (1 != LIST_LENGTH(pFunc->pParameterList)) {
return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
@ -1551,6 +1558,14 @@ static int32_t translateDiff(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
return TSDB_CODE_SUCCESS;
}
static EFuncReturnRows diffEstReturnRows(SFunctionNode* pFunc) {
if (1 == LIST_LENGTH(pFunc->pParameterList)) {
return FUNC_RETURN_ROWS_N_MINUS_1;
}
return 1 == ((SValueNode*)nodesListGetNode(pFunc->pParameterList, 1))->datum.i ? FUNC_RETURN_ROWS_INDEFINITE
: FUNC_RETURN_ROWS_N_MINUS_1;
}
static int32_t translateLength(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
if (1 != LIST_LENGTH(pFunc->pParameterList)) {
return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
@ -2068,7 +2083,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "apercentile",
.type = FUNCTION_TYPE_APERCENTILE,
.classification = FUNC_MGT_AGG_FUNC,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC,
.translateFunc = translateApercentile,
.getEnvFunc = getApercentileFuncEnv,
.initFunc = apercentileFunctionSetup,
@ -2083,7 +2098,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
},
{
.name = "_apercentile_partial",
.type = FUNCTION_TYPE_APERCENTILE_PARTIAL,
.type = FUNCTION_TYPE_APERCENTILE_PARTIAL | FUNC_MGT_TIMELINE_FUNC,
.classification = FUNC_MGT_AGG_FUNC,
.translateFunc = translateApercentilePartial,
.getEnvFunc = getApercentileFuncEnv,
@ -2096,7 +2111,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "_apercentile_merge",
.type = FUNCTION_TYPE_APERCENTILE_MERGE,
.classification = FUNC_MGT_AGG_FUNC,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC,
.translateFunc = translateApercentileMerge,
.getEnvFunc = getApercentileFuncEnv,
.initFunc = apercentileFunctionSetup,
@ -2231,13 +2246,14 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "derivative",
.type = FUNCTION_TYPE_DERIVATIVE,
.classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_IMPLICIT_TS_FUNC,
.classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_IMPLICIT_TS_FUNC | FUNC_MGT_CUMULATIVE_FUNC,
.translateFunc = translateDerivative,
.getEnvFunc = getDerivativeFuncEnv,
.initFunc = derivativeFuncSetup,
.processFunc = derivativeFunction,
.sprocessFunc = derivativeScalarFunction,
.finalizeFunc = functionFinalize
.finalizeFunc = functionFinalize,
.estimateReturnRowsFunc = derivativeEstReturnRows
},
{
.name = "irate",
@ -2312,6 +2328,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
.type = FUNCTION_TYPE_LAST,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_MULTI_RES_FUNC | FUNC_MGT_IMPLICIT_TS_FUNC,
.translateFunc = translateFirstLast,
.dynDataRequiredFunc = lastDynDataReq,
.getEnvFunc = getFirstLastFuncEnv,
.initFunc = functionSetup,
.processFunc = lastFunction,
@ -2358,7 +2375,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "histogram",
.type = FUNCTION_TYPE_HISTOGRAM,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_MULTI_ROWS_FUNC | FUNC_MGT_FORBID_FILL_FUNC,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_MULTI_ROWS_FUNC | FUNC_MGT_FORBID_FILL_FUNC,
.translateFunc = translateHistogram,
.getEnvFunc = getHistogramFuncEnv,
.initFunc = histogramFunctionSetup,
@ -2373,7 +2390,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "_histogram_partial",
.type = FUNCTION_TYPE_HISTOGRAM_PARTIAL,
.classification = FUNC_MGT_AGG_FUNC,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_MULTI_ROWS_FUNC | FUNC_MGT_FORBID_FILL_FUNC,
.translateFunc = translateHistogramPartial,
.getEnvFunc = getHistogramFuncEnv,
.initFunc = histogramFunctionSetup,
@ -2385,7 +2402,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "_histogram_merge",
.type = FUNCTION_TYPE_HISTOGRAM_MERGE,
.classification = FUNC_MGT_AGG_FUNC,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_MULTI_ROWS_FUNC | FUNC_MGT_FORBID_FILL_FUNC,
.translateFunc = translateHistogramMerge,
.getEnvFunc = getHistogramFuncEnv,
.initFunc = functionSetup,
@ -2397,7 +2414,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "hyperloglog",
.type = FUNCTION_TYPE_HYPERLOGLOG,
.classification = FUNC_MGT_AGG_FUNC,
.classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC,
.translateFunc = translateHLL,
.getEnvFunc = getHLLFuncEnv,
.initFunc = functionSetup,
@ -2411,7 +2428,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
},
{
.name = "_hyperloglog_partial",
.type = FUNCTION_TYPE_HYPERLOGLOG_PARTIAL,
.type = FUNCTION_TYPE_HYPERLOGLOG_PARTIAL | FUNC_MGT_TIMELINE_FUNC,
.classification = FUNC_MGT_AGG_FUNC,
.translateFunc = translateHLLPartial,
.getEnvFunc = getHLLFuncEnv,
@ -2423,7 +2440,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
},
{
.name = "_hyperloglog_merge",
.type = FUNCTION_TYPE_HYPERLOGLOG_MERGE,
.type = FUNCTION_TYPE_HYPERLOGLOG_MERGE | FUNC_MGT_TIMELINE_FUNC,
.classification = FUNC_MGT_AGG_FUNC,
.translateFunc = translateHLLMerge,
.getEnvFunc = getHLLFuncEnv,
@ -2436,13 +2453,14 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "diff",
.type = FUNCTION_TYPE_DIFF,
.classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
.classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC | FUNC_MGT_CUMULATIVE_FUNC,
.translateFunc = translateDiff,
.getEnvFunc = getDiffFuncEnv,
.initFunc = diffFunctionSetup,
.processFunc = diffFunction,
.sprocessFunc = diffScalarFunction,
.finalizeFunc = functionFinalize
.finalizeFunc = functionFinalize,
.estimateReturnRowsFunc = diffEstReturnRows
},
{
.name = "statecount",
@ -2469,13 +2487,14 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
{
.name = "csum",
.type = FUNCTION_TYPE_CSUM,
.classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
.classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC | FUNC_MGT_CUMULATIVE_FUNC,
.translateFunc = translateCsum,
.getEnvFunc = getCsumFuncEnv,
.initFunc = functionSetup,
.processFunc = csumFunction,
.sprocessFunc = csumScalarFunction,
.finalizeFunc = NULL
.finalizeFunc = NULL,
.estimateReturnRowsFunc = csumEstReturnRows,
},
{
.name = "mavg",

16
source/libs/function/src/builtinsimpl.c
View File

@ -2700,6 +2700,22 @@ int32_t apercentileCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx)
return TSDB_CODE_SUCCESS;
}
EFuncDataRequired lastDynDataReq(void* pRes, STimeWindow* pTimeWindow) {
SResultRowEntryInfo* pEntry = (SResultRowEntryInfo*) pRes;
// not initialized yet, data is required
if (pEntry == NULL) {
return FUNC_DATA_REQUIRED_DATA_LOAD;
}
SFirstLastRes* pResult = GET_ROWCELL_INTERBUF(pEntry);
if (pResult->hasResult && pResult->ts >= pTimeWindow->ekey) {
return FUNC_DATA_REQUIRED_NOT_LOAD;
} else {
return FUNC_DATA_REQUIRED_DATA_LOAD;
}
}
int32_t getFirstLastInfoSize(int32_t resBytes) { return sizeof(SFirstLastRes) + resBytes; }
bool getFirstLastFuncEnv(SFunctionNode* pFunc, SFuncExecEnv* pEnv) {

17
source/libs/function/src/functionMgt.c
View File

@ -89,6 +89,14 @@ int32_t fmGetFuncInfo(SFunctionNode* pFunc, char* pMsg, int32_t msgLen) {
return TSDB_CODE_FUNC_NOT_BUILTIN_FUNTION;
}
EFuncReturnRows fmGetFuncReturnRows(SFunctionNode* pFunc) {
if (NULL != funcMgtBuiltins[pFunc->funcId].estimateReturnRowsFunc) {
return funcMgtBuiltins[pFunc->funcId].estimateReturnRowsFunc(pFunc);
}
return (fmIsIndefiniteRowsFunc(pFunc->funcId) || fmIsMultiRowsFunc(pFunc->funcId)) ? FUNC_RETURN_ROWS_INDEFINITE
: FUNC_RETURN_ROWS_NORMAL;
}
bool fmIsBuiltinFunc(const char* pFunc) {
return NULL != taosHashGet(gFunMgtService.pFuncNameHashTable, pFunc, strlen(pFunc));
}
@ -107,7 +115,12 @@ EFuncDataRequired fmFuncDynDataRequired(int32_t funcId, void* pRes, STimeWindow*
if (fmIsUserDefinedFunc(funcId) || funcId < 0 || funcId >= funcMgtBuiltinsNum) {
return TSDB_CODE_FAILED;
}
return funcMgtBuiltins[funcId].dynDataRequiredFunc(pRes, pTimeWindow);
if (funcMgtBuiltins[funcId].dynDataRequiredFunc == NULL) {
return FUNC_DATA_REQUIRED_DATA_LOAD;
} else {
return funcMgtBuiltins[funcId].dynDataRequiredFunc(pRes, pTimeWindow);
}
}
int32_t fmGetFuncExecFuncs(int32_t funcId, SFuncExecFuncs* pFpSet) {
@ -192,6 +205,8 @@ bool fmIsMultiRowsFunc(int32_t funcId) { return isSpecificClassifyFunc(funcId, F
bool fmIsKeepOrderFunc(int32_t funcId) { return isSpecificClassifyFunc(funcId, FUNC_MGT_KEEP_ORDER_FUNC); }
bool fmIsCumulativeFunc(int32_t funcId) { return isSpecificClassifyFunc(funcId, FUNC_MGT_CUMULATIVE_FUNC); }
bool fmIsInterpFunc(int32_t funcId) {
if (funcId < 0 || funcId >= funcMgtBuiltinsNum) {
return false;

2
source/libs/nodes/src/nodesCloneFuncs.c
View File

@ -443,6 +443,7 @@ static int32_t logicWindowCopy(const SWindowLogicNode* pSrc, SWindowLogicNode* p
COPY_SCALAR_FIELD(igExpired);
COPY_SCALAR_FIELD(windowAlgo);
COPY_SCALAR_FIELD(inputTsOrder);
COPY_SCALAR_FIELD(outputTsOrder);
return TSDB_CODE_SUCCESS;
}
@ -452,6 +453,7 @@ static int32_t logicFillCopy(const SFillLogicNode* pSrc, SFillLogicNode* pDst) {
CLONE_NODE_FIELD(pWStartTs);
CLONE_NODE_FIELD(pValues);
COPY_OBJECT_FIELD(timeRange, sizeof(STimeWindow));
COPY_SCALAR_FIELD(inputTsOrder);
return TSDB_CODE_SUCCESS;
}

21
source/libs/nodes/src/nodesCodeFuncs.c
View File

@ -1936,6 +1936,8 @@ static const char* jkWindowPhysiPlanTsEnd = "TsEnd";
static const char* jkWindowPhysiPlanTriggerType = "TriggerType";
static const char* jkWindowPhysiPlanWatermark = "Watermark";
static const char* jkWindowPhysiPlanIgnoreExpired = "IgnoreExpired";
static const char* jkWindowPhysiPlanInputTsOrder = "inputTsOrder";
static const char* jkWindowPhysiPlanOutputTsOrder = "outputTsOrder";
static int32_t physiWindowNodeToJson(const void* pObj, SJson* pJson) {
const SWinodwPhysiNode* pNode = (const SWinodwPhysiNode*)pObj;
@ -1962,6 +1964,12 @@ static int32_t physiWindowNodeToJson(const void* pObj, SJson* pJson) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddIntegerToObject(pJson, jkWindowPhysiPlanIgnoreExpired, pNode->igExpired);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddIntegerToObject(pJson, jkWindowPhysiPlanInputTsOrder, pNode->inputTsOrder);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddIntegerToObject(pJson, jkWindowPhysiPlanOutputTsOrder, pNode->outputTsOrder);
}
return code;
}
@ -1991,6 +1999,12 @@ static int32_t jsonToPhysiWindowNode(const SJson* pJson, void* pObj) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetTinyIntValue(pJson, jkWindowPhysiPlanIgnoreExpired, &pNode->igExpired);
}
if (TSDB_CODE_SUCCESS == code) {
tjsonGetNumberValue(pJson, jkWindowPhysiPlanInputTsOrder, pNode->inputTsOrder, code);
}
if (TSDB_CODE_SUCCESS == code) {
tjsonGetNumberValue(pJson, jkWindowPhysiPlanOutputTsOrder, pNode->outputTsOrder, code);
}
return code;
}
@ -2053,6 +2067,7 @@ static const char* jkFillPhysiPlanValues = "Values";
static const char* jkFillPhysiPlanTargets = "Targets";
static const char* jkFillPhysiPlanStartTime = "StartTime";
static const char* jkFillPhysiPlanEndTime = "EndTime";
static const char* jkFillPhysiPlanInputTsOrder = "inputTsOrder";
static int32_t physiFillNodeToJson(const void* pObj, SJson* pJson) {
const SFillPhysiNode* pNode = (const SFillPhysiNode*)pObj;
@ -2076,6 +2091,9 @@ static int32_t physiFillNodeToJson(const void* pObj, SJson* pJson) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddIntegerToObject(pJson, jkFillPhysiPlanEndTime, pNode->timeRange.ekey);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddIntegerToObject(pJson, jkFillPhysiPlanInputTsOrder, pNode->inputTsOrder);
}
return code;
}
@ -2103,6 +2121,9 @@ static int32_t jsonToPhysiFillNode(const SJson* pJson, void* pObj) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetBigIntValue(pJson, jkFillPhysiPlanEndTime, &pNode->timeRange.ekey);
}
if (TSDB_CODE_SUCCESS == code) {
tjsonGetNumberValue(pJson, jkFillPhysiPlanInputTsOrder, pNode->inputTsOrder, code);
}
return code;
}

35
source/libs/parser/src/parTranslater.c
View File

@ -1112,12 +1112,16 @@ static int32_t translateIndefiniteRowsFunc(STranslateContext* pCxt, SFunctionNod
if (!fmIsIndefiniteRowsFunc(pFunc->funcId)) {
return TSDB_CODE_SUCCESS;
}
if (!isSelectStmt(pCxt->pCurrStmt) || SQL_CLAUSE_SELECT != pCxt->currClause ||
((SSelectStmt*)pCxt->pCurrStmt)->hasIndefiniteRowsFunc || ((SSelectStmt*)pCxt->pCurrStmt)->hasAggFuncs ||
((SSelectStmt*)pCxt->pCurrStmt)->hasMultiRowsFunc) {
if (!isSelectStmt(pCxt->pCurrStmt) || SQL_CLAUSE_SELECT != pCxt->currClause) {
return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_NOT_ALLOWED_FUNC);
}
if (NULL != ((SSelectStmt*)pCxt->pCurrStmt)->pWindow || NULL != ((SSelectStmt*)pCxt->pCurrStmt)->pGroupByList) {
SSelectStmt* pSelect = (SSelectStmt*)pCxt->pCurrStmt;
if (pSelect->hasAggFuncs || pSelect->hasMultiRowsFunc ||
(pSelect->hasIndefiniteRowsFunc &&
(FUNC_RETURN_ROWS_INDEFINITE == pSelect->returnRows || pSelect->returnRows != fmGetFuncReturnRows(pFunc)))) {
return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_NOT_ALLOWED_FUNC);
}
if (NULL != pSelect->pWindow || NULL != pSelect->pGroupByList) {
return generateSyntaxErrMsgExt(&pCxt->msgBuf, TSDB_CODE_PAR_NOT_ALLOWED_FUNC,
"%s function is not supported in window query or group query", pFunc->functionName);
}
@ -1232,18 +1236,28 @@ static int32_t getMultiResFuncNum(SNodeList* pParameterList) {
return LIST_LENGTH(pParameterList);
}
static int32_t calcSelectFuncNum(SFunctionNode* pFunc, int32_t currSelectFuncNum) {
if (fmIsCumulativeFunc(pFunc->funcId)) {
return currSelectFuncNum > 0 ? currSelectFuncNum : 1;
}
return currSelectFuncNum + ((fmIsMultiResFunc(pFunc->funcId) && !fmIsLastRowFunc(pFunc->funcId))
? getMultiResFuncNum(pFunc->pParameterList)
: 1);
}
static void setFuncClassification(SNode* pCurrStmt, SFunctionNode* pFunc) {
if (NULL != pCurrStmt && QUERY_NODE_SELECT_STMT == nodeType(pCurrStmt)) {
SSelectStmt* pSelect = (SSelectStmt*)pCurrStmt;
pSelect->hasAggFuncs = pSelect->hasAggFuncs ? true : fmIsAggFunc(pFunc->funcId);
pSelect->hasRepeatScanFuncs = pSelect->hasRepeatScanFuncs ? true : fmIsRepeatScanFunc(pFunc->funcId);
pSelect->hasIndefiniteRowsFunc = pSelect->hasIndefiniteRowsFunc ? true : fmIsIndefiniteRowsFunc(pFunc->funcId);
if (fmIsIndefiniteRowsFunc(pFunc->funcId)) {
pSelect->hasIndefiniteRowsFunc = true;
pSelect->returnRows = fmGetFuncReturnRows(pFunc);
}
pSelect->hasMultiRowsFunc = pSelect->hasMultiRowsFunc ? true : fmIsMultiRowsFunc(pFunc->funcId);
if (fmIsSelectFunc(pFunc->funcId)) {
pSelect->hasSelectFunc = true;
pSelect->selectFuncNum += (fmIsMultiResFunc(pFunc->funcId) && !fmIsLastRowFunc(pFunc->funcId))
? getMultiResFuncNum(pFunc->pParameterList)
: 1;
pSelect->selectFuncNum = calcSelectFuncNum(pFunc, pSelect->selectFuncNum);
} else if (fmIsVectorFunc(pFunc->funcId)) {
pSelect->hasOtherVectorFunc = true;
}
@ -2483,6 +2497,9 @@ static int32_t translateInterp(STranslateContext* pCxt, SSelectStmt* pSelect) {
}
static int32_t translatePartitionBy(STranslateContext* pCxt, SNodeList* pPartitionByList) {
if (NULL == pPartitionByList) {
return TSDB_CODE_SUCCESS;
}
pCxt->currClause = SQL_CLAUSE_PARTITION_BY;
return translateExprList(pCxt, pPartitionByList);
}
@ -5571,7 +5588,7 @@ static int32_t rewriteCreateTable(STranslateContext* pCxt, SQuery* pQuery) {
int32_t code = checkCreateTable(pCxt, pStmt, false);
SVgroupInfo info = {0};
SName name;
SName name;
toName(pCxt->pParseCxt->acctId, pStmt->dbName, pStmt->tableName, &name);
if (TSDB_CODE_SUCCESS == code) {
code = getTableHashVgroupImpl(pCxt, &name, &info);

2
source/libs/planner/src/planLogicCreater.c
View File

@ -632,6 +632,7 @@ static int32_t createWindowLogicNodeFinalize(SLogicPlanContext* pCxt, SSelectStm
pWindow->igExpired = pCxt->pPlanCxt->igExpired;
}
pWindow->inputTsOrder = ORDER_ASC;
pWindow->outputTsOrder = ORDER_ASC;
int32_t code = nodesCollectFuncs(pSelect, SQL_CLAUSE_WINDOW, fmIsWindowClauseFunc, &pWindow->pFuncs);
if (TSDB_CODE_SUCCESS == code) {
@ -764,6 +765,7 @@ static int32_t createFillLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSelect
pFill->node.groupAction = GROUP_ACTION_KEEP;
pFill->node.requireDataOrder = DATA_ORDER_LEVEL_IN_GROUP;
pFill->node.resultDataOrder = DATA_ORDER_LEVEL_IN_GROUP;
pFill->inputTsOrder = ORDER_ASC;
int32_t code = nodesCollectColumns(pSelect, SQL_CLAUSE_WINDOW, NULL, COLLECT_COL_TYPE_ALL, &pFill->node.pTargets);
if (TSDB_CODE_SUCCESS == code && NULL == pFill->node.pTargets) {

185
source/libs/planner/src/planOptimizer.c
View File

@ -38,10 +38,13 @@ typedef struct SOptimizeRule {
FOptimize optimizeFunc;
} SOptimizeRule;
typedef enum EScanOrder { SCAN_ORDER_ASC = 1, SCAN_ORDER_DESC, SCAN_ORDER_BOTH } EScanOrder;
typedef struct SOsdInfo {
SScanLogicNode* pScan;
SNodeList* pSdrFuncs;
SNodeList* pDsoFuncs;
EScanOrder scanOrder;
} SOsdInfo;
typedef struct SCpdIsMultiTableCondCxt {
@ -97,6 +100,27 @@ static EDealRes optRebuildTbanme(SNode** pNode, void* pContext) {
return DEAL_RES_CONTINUE;
}
static void optSetParentOrder(SLogicNode* pNode, EOrder order) {
if (NULL == pNode) {
return;
}
switch (nodeType(pNode)) {
case QUERY_NODE_LOGIC_PLAN_WINDOW:
((SWindowLogicNode*)pNode)->inputTsOrder = order;
// window has a sorting function, and the operator behind it uses its output order
return;
case QUERY_NODE_LOGIC_PLAN_JOIN:
((SJoinLogicNode*)pNode)->inputTsOrder = order;
break;
case QUERY_NODE_LOGIC_PLAN_FILL:
((SFillLogicNode*)pNode)->inputTsOrder = order;
break;
default:
break;
}
optSetParentOrder(pNode->pParent, order);
}
EDealRes scanPathOptHaveNormalColImpl(SNode* pNode, void* pContext) {
if (QUERY_NODE_COLUMN == nodeType(pNode)) {
// *((bool*)pContext) = (COLUMN_TYPE_TAG != ((SColumnNode*)pNode)->colType);
@ -179,16 +203,18 @@ static int32_t scanPathOptGetRelatedFuncs(SScanLogicNode* pScan, SNodeList** pSd
SNodeList* pAllFuncs = scanPathOptGetAllFuncs(pScan->node.pParent);
SNodeList* pTmpSdrFuncs = NULL;
SNodeList* pTmpDsoFuncs = NULL;
SNode* pFunc = NULL;
SNode* pNode = NULL;
bool otherFunc = false;
FOREACH(pFunc, pAllFuncs) {
int32_t code = TSDB_CODE_SUCCESS;
if (scanPathOptNeedOptimizeDataRequire((SFunctionNode*)pFunc)) {
code = nodesListMakeStrictAppend(&pTmpSdrFuncs, nodesCloneNode(pFunc));
} else if (scanPathOptNeedDynOptimize((SFunctionNode*)pFunc)) {
code = nodesListMakeStrictAppend(&pTmpDsoFuncs, nodesCloneNode(pFunc));
FOREACH(pNode, pAllFuncs) {
SFunctionNode* pFunc = (SFunctionNode*)pNode;
int32_t code = TSDB_CODE_SUCCESS;
if (scanPathOptNeedOptimizeDataRequire(pFunc)) {
code = nodesListMakeStrictAppend(&pTmpSdrFuncs, nodesCloneNode(pNode));
} else if (scanPathOptNeedDynOptimize(pFunc)) {
code = nodesListMakeStrictAppend(&pTmpDsoFuncs, nodesCloneNode(pNode));
} else {
otherFunc = true;
break;
}
if (TSDB_CODE_SUCCESS != code) {
nodesDestroyList(pTmpSdrFuncs);
@ -206,12 +232,46 @@ static int32_t scanPathOptGetRelatedFuncs(SScanLogicNode* pScan, SNodeList** pSd
return TSDB_CODE_SUCCESS;
}
static int32_t scanPathOptGetScanOrder(SScanLogicNode* pScan, EScanOrder* pScanOrder) {
SNodeList* pAllFuncs = scanPathOptGetAllFuncs(pScan->node.pParent);
SNode* pNode = NULL;
bool hasFirst = false;
bool hasLast = false;
bool otherFunc = false;
FOREACH(pNode, pAllFuncs) {
SFunctionNode* pFunc = (SFunctionNode*)pNode;
if (FUNCTION_TYPE_FIRST == pFunc->funcType) {
hasFirst = true;
} else if (FUNCTION_TYPE_LAST == pFunc->funcType) {
hasLast = true;
} else if (FUNCTION_TYPE_SELECT_VALUE != pFunc->funcType) {
otherFunc = true;
}
}
if (hasFirst && hasLast && !otherFunc) {
*pScanOrder = SCAN_ORDER_BOTH;
} else if (hasLast) {
*pScanOrder = SCAN_ORDER_DESC;
} else {
*pScanOrder = SCAN_ORDER_ASC;
}
return TSDB_CODE_SUCCESS;
}
static int32_t scanPathOptSetOsdInfo(SOsdInfo* pInfo) {
int32_t code = scanPathOptGetRelatedFuncs(pInfo->pScan, &pInfo->pSdrFuncs, &pInfo->pDsoFuncs);
if (TSDB_CODE_SUCCESS == code) {
code = scanPathOptGetScanOrder(pInfo->pScan, &pInfo->scanOrder);
}
return code;
}
static int32_t scanPathOptMatch(SOptimizeContext* pCxt, SLogicNode* pLogicNode, SOsdInfo* pInfo) {
pInfo->pScan = (SScanLogicNode*)optFindPossibleNode(pLogicNode, scanPathOptMayBeOptimized);
if (NULL == pInfo->pScan) {
return TSDB_CODE_SUCCESS;
}
return scanPathOptGetRelatedFuncs(pInfo->pScan, &pInfo->pSdrFuncs, &pInfo->pDsoFuncs);
return scanPathOptSetOsdInfo(pInfo);
}
static EFuncDataRequired scanPathOptPromoteDataRequired(EFuncDataRequired l, EFuncDataRequired r) {
@ -258,15 +318,42 @@ static void scanPathOptSetScanWin(SScanLogicNode* pScan) {
}
}
static void scanPathOptSetScanOrder(EScanOrder scanOrder, SScanLogicNode* pScan) {
if (pScan->sortPrimaryKey || pScan->scanSeq[0] > 1 || pScan->scanSeq[1] > 1) {
return;
}
switch (scanOrder) {
case SCAN_ORDER_ASC:
pScan->scanSeq[0] = 1;
pScan->scanSeq[1] = 0;
optSetParentOrder(pScan->node.pParent, ORDER_ASC);
break;
case SCAN_ORDER_DESC:
pScan->scanSeq[0] = 0;
pScan->scanSeq[1] = 1;
optSetParentOrder(pScan->node.pParent, ORDER_DESC);
break;
case SCAN_ORDER_BOTH:
pScan->scanSeq[0] = 1;
pScan->scanSeq[1] = 1;
break;
default:
break;
}
}
static int32_t scanPathOptimize(SOptimizeContext* pCxt, SLogicSubplan* pLogicSubplan) {
SOsdInfo info = {0};
SOsdInfo info = {.scanOrder = SCAN_ORDER_ASC};
int32_t code = scanPathOptMatch(pCxt, pLogicSubplan->pNode, &info);
if (TSDB_CODE_SUCCESS == code && info.pScan) {
scanPathOptSetScanWin((SScanLogicNode*)info.pScan);
scanPathOptSetScanWin(info.pScan);
scanPathOptSetScanOrder(info.scanOrder, info.pScan);
}
if (TSDB_CODE_SUCCESS == code && (NULL != info.pDsoFuncs || NULL != info.pSdrFuncs)) {
info.pScan->dataRequired = scanPathOptGetDataRequired(info.pSdrFuncs);
info.pScan->pDynamicScanFuncs = info.pDsoFuncs;
}
if (TSDB_CODE_SUCCESS == code && info.pScan) {
OPTIMIZE_FLAG_SET_MASK(info.pScan->node.optimizedFlag, OPTIMIZE_FLAG_SCAN_PATH);
pCxt->optimized = true;
}
@ -987,12 +1074,13 @@ static bool sortPriKeyOptMayBeOptimized(SLogicNode* pNode) {
}
SSortLogicNode* pSort = (SSortLogicNode*)pNode;
if (pSort->groupSort || !sortPriKeyOptIsPriKeyOrderBy(pSort->pSortKeys) || 1 != LIST_LENGTH(pSort->node.pChildren)) {
return TSDB_CODE_SUCCESS;
return false;
}
return true;
}
static int32_t sortPriKeyOptGetScanNodesImpl(SLogicNode* pNode, bool* pNotOptimize, SNodeList** pScanNodes) {
static int32_t sortPriKeyOptGetSequencingNodesImpl(SLogicNode* pNode, bool* pNotOptimize,
SNodeList** pSequencingNodes) {
switch (nodeType(pNode)) {
case QUERY_NODE_LOGIC_PLAN_SCAN: {
SScanLogicNode* pScan = (SScanLogicNode*)pNode;
@ -1000,17 +1088,19 @@ static int32_t sortPriKeyOptGetScanNodesImpl(SLogicNode* pNode, bool* pNotOptimi
*pNotOptimize = true;
return TSDB_CODE_SUCCESS;
}
return nodesListMakeAppend(pScanNodes, (SNode*)pNode);
return nodesListMakeAppend(pSequencingNodes, (SNode*)pNode);
}
case QUERY_NODE_LOGIC_PLAN_JOIN: {
int32_t code =
sortPriKeyOptGetScanNodesImpl((SLogicNode*)nodesListGetNode(pNode->pChildren, 0), pNotOptimize, pScanNodes);
int32_t code = sortPriKeyOptGetSequencingNodesImpl((SLogicNode*)nodesListGetNode(pNode->pChildren, 0),
pNotOptimize, pSequencingNodes);
if (TSDB_CODE_SUCCESS == code) {
code =
sortPriKeyOptGetScanNodesImpl((SLogicNode*)nodesListGetNode(pNode->pChildren, 1), pNotOptimize, pScanNodes);
code = sortPriKeyOptGetSequencingNodesImpl((SLogicNode*)nodesListGetNode(pNode->pChildren, 1), pNotOptimize,
pSequencingNodes);
}
return code;
}
case QUERY_NODE_LOGIC_PLAN_WINDOW:
return nodesListMakeAppend(pSequencingNodes, (SNode*)pNode);
case QUERY_NODE_LOGIC_PLAN_AGG:
case QUERY_NODE_LOGIC_PLAN_PARTITION:
*pNotOptimize = true;
@ -1024,14 +1114,15 @@ static int32_t sortPriKeyOptGetScanNodesImpl(SLogicNode* pNode, bool* pNotOptimi
return TSDB_CODE_SUCCESS;
}
return sortPriKeyOptGetScanNodesImpl((SLogicNode*)nodesListGetNode(pNode->pChildren, 0), pNotOptimize, pScanNodes);
return sortPriKeyOptGetSequencingNodesImpl((SLogicNode*)nodesListGetNode(pNode->pChildren, 0), pNotOptimize,
pSequencingNodes);
}
static int32_t sortPriKeyOptGetScanNodes(SLogicNode* pNode, SNodeList** pScanNodes) {
static int32_t sortPriKeyOptGetSequencingNodes(SLogicNode* pNode, SNodeList** pSequencingNodes) {
bool notOptimize = false;
int32_t code = sortPriKeyOptGetScanNodesImpl(pNode, &notOptimize, pScanNodes);
int32_t code = sortPriKeyOptGetSequencingNodesImpl(pNode, &notOptimize, pSequencingNodes);
if (TSDB_CODE_SUCCESS != code || notOptimize) {
nodesClearList(*pScanNodes);
nodesClearList(*pSequencingNodes);
}
return code;
}
@ -1040,33 +1131,26 @@ static EOrder sortPriKeyOptGetPriKeyOrder(SSortLogicNode* pSort) {
return ((SOrderByExprNode*)nodesListGetNode(pSort->pSortKeys, 0))->order;
}
static void sortPriKeyOptSetParentOrder(SLogicNode* pNode, EOrder order) {
if (NULL == pNode) {
return;
}
if (QUERY_NODE_LOGIC_PLAN_WINDOW == nodeType(pNode)) {
((SWindowLogicNode*)pNode)->inputTsOrder = order;
} else if (QUERY_NODE_LOGIC_PLAN_JOIN == nodeType(pNode)) {
((SJoinLogicNode*)pNode)->inputTsOrder = order;
}
sortPriKeyOptSetParentOrder(pNode->pParent, order);
}
static int32_t sortPriKeyOptApply(SOptimizeContext* pCxt, SLogicSubplan* pLogicSubplan, SSortLogicNode* pSort,
SNodeList* pScanNodes) {
SNodeList* pSequencingNodes) {
EOrder order = sortPriKeyOptGetPriKeyOrder(pSort);
SNode* pScanNode = NULL;
FOREACH(pScanNode, pScanNodes) {
SScanLogicNode* pScan = (SScanLogicNode*)pScanNode;
if (ORDER_DESC == order && pScan->scanSeq[0] > 0) {
TSWAP(pScan->scanSeq[0], pScan->scanSeq[1]);
SNode* pSequencingNode = NULL;
FOREACH(pSequencingNode, pSequencingNodes) {
if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pSequencingNode)) {
SScanLogicNode* pScan = (SScanLogicNode*)pSequencingNode;
if ((ORDER_DESC == order && pScan->scanSeq[0] > 0) || (ORDER_ASC == order && pScan->scanSeq[1] > 0)) {
TSWAP(pScan->scanSeq[0], pScan->scanSeq[1]);
}
if (TSDB_SUPER_TABLE == pScan->tableType) {
pScan->scanType = SCAN_TYPE_TABLE_MERGE;
pScan->node.resultDataOrder = DATA_ORDER_LEVEL_GLOBAL;
pScan->node.requireDataOrder = DATA_ORDER_LEVEL_GLOBAL;
}
pScan->sortPrimaryKey = true;
} else if (QUERY_NODE_LOGIC_PLAN_WINDOW == nodeType(pSequencingNode)) {
((SWindowLogicNode*)pSequencingNode)->outputTsOrder = order;
}
if (TSDB_SUPER_TABLE == pScan->tableType) {
pScan->scanType = SCAN_TYPE_TABLE_MERGE;
pScan->node.resultDataOrder = DATA_ORDER_LEVEL_GLOBAL;
pScan->node.requireDataOrder = DATA_ORDER_LEVEL_GLOBAL;
}
sortPriKeyOptSetParentOrder(pScan->node.pParent, order);
optSetParentOrder(((SLogicNode*)pSequencingNode)->pParent, order);
}
SLogicNode* pChild = (SLogicNode*)nodesListGetNode(pSort->node.pChildren, 0);
@ -1083,12 +1167,13 @@ static int32_t sortPriKeyOptApply(SOptimizeContext* pCxt, SLogicSubplan* pLogicS
}
static int32_t sortPrimaryKeyOptimizeImpl(SOptimizeContext* pCxt, SLogicSubplan* pLogicSubplan, SSortLogicNode* pSort) {
SNodeList* pScanNodes = NULL;
int32_t code = sortPriKeyOptGetScanNodes((SLogicNode*)nodesListGetNode(pSort->node.pChildren, 0), &pScanNodes);
if (TSDB_CODE_SUCCESS == code && NULL != pScanNodes) {
code = sortPriKeyOptApply(pCxt, pLogicSubplan, pSort, pScanNodes);
SNodeList* pSequencingNodes = NULL;
int32_t code =
sortPriKeyOptGetSequencingNodes((SLogicNode*)nodesListGetNode(pSort->node.pChildren, 0), &pSequencingNodes);
if (TSDB_CODE_SUCCESS == code && NULL != pSequencingNodes) {
code = sortPriKeyOptApply(pCxt, pLogicSubplan, pSort, pSequencingNodes);
}
nodesClearList(pScanNodes);
nodesClearList(pSequencingNodes);
return code;
}

3
source/libs/planner/src/planPhysiCreater.c
View File

@ -1089,6 +1089,8 @@ static int32_t createWindowPhysiNodeFinalize(SPhysiPlanContext* pCxt, SNodeList*
pWindow->triggerType = pWindowLogicNode->triggerType;
pWindow->watermark = pWindowLogicNode->watermark;
pWindow->igExpired = pWindowLogicNode->igExpired;
pWindow->inputTsOrder = pWindowLogicNode->inputTsOrder;
pWindow->outputTsOrder = pWindowLogicNode->outputTsOrder;
SNodeList* pPrecalcExprs = NULL;
SNodeList* pFuncs = NULL;
@ -1363,6 +1365,7 @@ static int32_t createFillPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren
pFill->mode = pFillNode->mode;
pFill->timeRange = pFillNode->timeRange;
pFill->inputTsOrder = pFillNode->inputTsOrder;
SDataBlockDescNode* pChildTupe = (((SPhysiNode*)nodesListGetNode(pChildren, 0))->pOutputDataBlockDesc);
int32_t code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pFillNode->node.pTargets, &pFill->pTargets);

2
source/libs/planner/src/planSpliter.c
View File

@ -492,7 +492,7 @@ static int32_t stbSplSplitIntervalForBatch(SSplitContext* pCxt, SStableSplitInfo
((SWindowLogicNode*)pInfo->pSplitNode)->windowAlgo = INTERVAL_ALGO_MERGE;
SNodeList* pMergeKeys = NULL;
code = stbSplCreateMergeKeysByPrimaryKey(((SWindowLogicNode*)pInfo->pSplitNode)->pTspk,
((SWindowLogicNode*)pInfo->pSplitNode)->inputTsOrder, &pMergeKeys);
((SWindowLogicNode*)pInfo->pSplitNode)->outputTsOrder, &pMergeKeys);
if (TSDB_CODE_SUCCESS == code) {
code = stbSplCreateMergeNode(pCxt, NULL, pInfo->pSplitNode, pMergeKeys, pPartWindow, true);
}

10
source/libs/planner/test/planBasicTest.cpp
View File

@ -175,6 +175,16 @@ TEST_F(PlanBasicTest, pseudoColumn) {
"WHERE ts BETWEEN '2017-7-14 18:00:00' AND '2017-7-14 19:00:00' INTERVAL(10S)");
}
TEST_F(PlanBasicTest, indefiniteRowsFunc) {
useDb("root", "test");
run("SELECT DIFF(c1) FROM t1");
run("SELECT DIFF(c1), c2 FROM t1");
run("SELECT DIFF(c1), DIFF(c3), ts FROM t1");
}
TEST_F(PlanBasicTest, withoutFrom) {
useDb("root", "test");

13
source/libs/planner/test/planOptimizeTest.cpp
View File

@ -30,6 +30,11 @@ TEST_F(PlanOptimizeTest, scanPath) {
run("SELECT COUNT(CAST(c1 AS BIGINT)) FROM t1");
run("SELECT PERCENTILE(c1, 40), COUNT(*) FROM t1");
run("SELECT LAST(c1) FROM t1");
run("SELECT LAST(c1) FROM t1 WHERE ts BETWEEN '2022-7-29 11:10:10' AND '2022-7-30 11:10:10' INTERVAL(10S) "
"FILL(LINEAR)");
}
TEST_F(PlanOptimizeTest, pushDownCondition) {
@ -57,7 +62,15 @@ TEST_F(PlanOptimizeTest, sortPrimaryKey) {
run("SELECT c1 FROM t1 ORDER BY ts DESC");
run("SELECT c1 FROM st1 ORDER BY ts DESC");
run("SELECT COUNT(*) FROM t1 INTERVAL(10S) ORDER BY _WSTART DESC");
run("SELECT FIRST(c1) FROM t1 WHERE ts BETWEEN '2022-7-29 11:10:10' AND '2022-7-30 11:10:10' INTERVAL(10S) "
"FILL(LINEAR) ORDER BY _WSTART DESC");
run("SELECT LAST(c1) FROM t1 WHERE ts BETWEEN '2022-7-29 11:10:10' AND '2022-7-30 11:10:10' INTERVAL(10S) "
"FILL(LINEAR) ORDER BY _WSTART");
}
TEST_F(PlanOptimizeTest, PartitionTags) {

2
source/libs/qworker/inc/qwMsg.h
View File

@ -39,7 +39,7 @@ int32_t qwBuildAndSendFetchRsp(int32_t rspType, SRpcHandleInfo *pConn, SRetrieve
void qwBuildFetchRsp(void *msg, SOutputData *input, int32_t len, bool qComplete);
int32_t qwBuildAndSendCQueryMsg(QW_FPARAMS_DEF, SRpcHandleInfo *pConn);
int32_t qwBuildAndSendQueryRsp(int32_t rspType, SRpcHandleInfo *pConn, int32_t code, SQWTaskCtx *ctx);
int32_t qwBuildAndSendExplainRsp(SRpcHandleInfo *pConn, SExplainExecInfo *execInfo, int32_t num);
int32_t qwBuildAndSendExplainRsp(SRpcHandleInfo *pConn, SArray* pExecList);
int32_t qwBuildAndSendErrorRsp(int32_t rspType, SRpcHandleInfo *pConn, int32_t code);
void qwFreeFetchRsp(void *msg);
int32_t qwMallocFetchRsp(int32_t length, SRetrieveTableRsp **rsp);

8
source/libs/qworker/src/qwMsg.c
View File

@ -82,8 +82,9 @@ int32_t qwBuildAndSendQueryRsp(int32_t rspType, SRpcHandleInfo *pConn, int32_t c
return TSDB_CODE_SUCCESS;
}
int32_t qwBuildAndSendExplainRsp(SRpcHandleInfo *pConn, SExplainExecInfo *execInfo, int32_t num) {
SExplainRsp rsp = {.numOfPlans = num, .subplanInfo = execInfo};
int32_t qwBuildAndSendExplainRsp(SRpcHandleInfo *pConn, SArray* pExecList) {
SExplainExecInfo* pInfo = taosArrayGet(pExecList, 0);
SExplainRsp rsp = {.numOfPlans = taosArrayGetSize(pExecList), .subplanInfo = pInfo};
int32_t contLen = tSerializeSExplainRsp(NULL, 0, &rsp);
void * pRsp = rpcMallocCont(contLen);
@ -96,10 +97,9 @@ int32_t qwBuildAndSendExplainRsp(SRpcHandleInfo *pConn, SExplainExecInfo *execIn
.code = 0,
.info = *pConn,
};
rpcRsp.info.ahandle = NULL;
tmsgSendRsp(&rpcRsp);
return TSDB_CODE_SUCCESS;
}

14
source/libs/qworker/src/qworker.c
View File

@ -44,18 +44,24 @@ int32_t qwProcessHbLinkBroken(SQWorker *mgmt, SQWMsg *qwMsg, SSchedulerHbReq *re
QW_RET(TSDB_CODE_SUCCESS);
}
static void freeItem(void* param) {
SExplainExecInfo* pInfo = param;
taosMemoryFree(pInfo->verboseInfo);
}
int32_t qwHandleTaskComplete(QW_FPARAMS_DEF, SQWTaskCtx *ctx) {
qTaskInfo_t taskHandle = ctx->taskHandle;
if (TASK_TYPE_TEMP == ctx->taskType && taskHandle) {
if (ctx->explain) {
SExplainExecInfo *execInfo = NULL;
int32_t resNum = 0;
QW_ERR_RET(qGetExplainExecInfo(taskHandle, &resNum, &execInfo));
SArray* execInfoList = taosArrayInit(4, sizeof(SExplainExecInfo));
QW_ERR_RET(qGetExplainExecInfo(taskHandle, execInfoList));
SRpcHandleInfo connInfo = ctx->ctrlConnInfo;
connInfo.ahandle = NULL;
QW_ERR_RET(qwBuildAndSendExplainRsp(&connInfo, execInfo, resNum));
int32_t code = qwBuildAndSendExplainRsp(&connInfo, execInfoList);
taosArrayDestroyEx(execInfoList, freeItem);
QW_ERR_RET(code);
}
if (!ctx->needFetch) {

59
source/libs/sync/src/syncAppendEntries.c
View File

@ -790,65 +790,6 @@ int32_t syncNodeOnAppendEntriesSnapshotCb(SSyncNode* ths, SyncAppendEntries* pMs
}
} while (0);
#if 0
// fake match
//
// condition1:
// I have snapshot, no log, preIndex > myLastIndex
//
// condition2:
// I have snapshot, have log, log <= snapshot, preIndex > myLastIndex
//
// condition3:
// I have snapshot, preIndex < snapshot.lastApplyIndex
//
// condition4:
// I have snapshot, preIndex == snapshot.lastApplyIndex, no data
//
// operation:
// match snapshot.lastApplyIndex - 1;
// no operation on log
do {
SyncIndex myLastIndex = syncNodeGetLastIndex(ths);
SSnapshot snapshot;
ths->pFsm->FpGetSnapshotInfo(ths->pFsm, &snapshot);
bool condition0 = (pMsg->term == ths->pRaftStore->currentTerm) && (ths->state == TAOS_SYNC_STATE_FOLLOWER) &&
syncNodeHasSnapshot(ths);
bool condition1 =
condition0 && (ths->pLogStore->syncLogEntryCount(ths->pLogStore) == 0) && (pMsg->prevLogIndex > myLastIndex); // donot use syncLogEntryCount!!! use isEmpty
bool condition2 = condition0 && (ths->pLogStore->syncLogLastIndex(ths->pLogStore) <= snapshot.lastApplyIndex) &&
(pMsg->prevLogIndex > myLastIndex);
bool condition3 = condition0 && (pMsg->prevLogIndex < snapshot.lastApplyIndex);
bool condition4 = condition0 && (pMsg->prevLogIndex == snapshot.lastApplyIndex) && (pMsg->dataLen == 0);
bool condition = condition1 || condition2 || condition3 || condition4;
if (condition) {
char logBuf[128];
snprintf(logBuf, sizeof(logBuf), "recv sync-append-entries, fake match, pre-index:%" PRId64 ", pre-term:%" PRIu64,
pMsg->prevLogIndex, pMsg->prevLogTerm);
syncNodeEventLog(ths, logBuf);
// prepare response msg
SyncAppendEntriesReply* pReply = syncAppendEntriesReplyBuild(ths->vgId);
pReply->srcId = ths->myRaftId;
pReply->destId = pMsg->srcId;
pReply->term = ths->pRaftStore->currentTerm;
pReply->privateTerm = ths->pNewNodeReceiver->privateTerm;
pReply->success = true;
pReply->matchIndex = snapshot.lastApplyIndex;
// send response
SRpcMsg rpcMsg;
syncAppendEntriesReply2RpcMsg(pReply, &rpcMsg);
syncNodeSendMsgById(&pReply->destId, ths, &rpcMsg);
syncAppendEntriesReplyDestroy(pReply);
return ret;
}
} while (0);
#endif
// fake match
//
// condition1:

5
source/libs/sync/src/syncAppendEntriesReply.c
View File

@ -213,6 +213,11 @@ int32_t syncNodeOnAppendEntriesReplySnapshot2Cb(SSyncNode* ths, SyncAppendEntrie
if (nextIndex > SYNC_INDEX_BEGIN) {
--nextIndex;
// speed up
if (nextIndex > pMsg->matchIndex + 1) {
nextIndex = pMsg->matchIndex + 1;
}
bool needStartSnapshot = false;
if (nextIndex >= SYNC_INDEX_BEGIN && !ths->pLogStore->syncLogExist(ths->pLogStore, nextIndex)) {
needStartSnapshot = true;

10
source/libs/sync/src/syncMain.c
View File

@ -2222,13 +2222,18 @@ SyncTerm syncNodeGetPreTerm(SSyncNode* pSyncNode, SyncIndex index) {
SyncIndex preIndex = index - 1;
SSyncRaftEntry* pPreEntry = NULL;
int32_t code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, preIndex, &pPreEntry);
SSnapshot snapshot = {.data = NULL,
.lastApplyIndex = SYNC_INDEX_INVALID,
.lastApplyTerm = SYNC_TERM_INVALID,
.lastConfigIndex = SYNC_INDEX_INVALID};
if (code == 0) {
ASSERT(pPreEntry != NULL);
preTerm = pPreEntry->term;
taosMemoryFree(pPreEntry);
return preTerm;
} else {
SSnapshot snapshot = {.data = NULL, .lastApplyIndex = -1, .lastApplyTerm = 0, .lastConfigIndex = -1};
if (pSyncNode->pFsm->FpGetSnapshotInfo != NULL) {
pSyncNode->pFsm->FpGetSnapshotInfo(pSyncNode->pFsm, &snapshot);
if (snapshot.lastApplyIndex == preIndex) {
@ -2239,7 +2244,8 @@ SyncTerm syncNodeGetPreTerm(SSyncNode* pSyncNode, SyncIndex index) {
do {
char logBuf[128];
snprintf(logBuf, sizeof(logBuf), "sync node get pre term error, index:%" PRId64, index);
snprintf(logBuf, sizeof(logBuf), "sync node get pre term error, index:%ld, snap-index:%ld, snap-term:%lu", index,
snapshot.lastApplyIndex, snapshot.lastApplyTerm);
syncNodeErrorLog(pSyncNode, logBuf);
} while (0);

6
source/libs/sync/src/syncReplication.c
View File

@ -132,7 +132,8 @@ int32_t syncNodeAppendEntriesPeersSnapshot2(SSyncNode* pSyncNode) {
SyncIndex preLogIndex = syncNodeGetPreIndex(pSyncNode, nextIndex);
SyncTerm preLogTerm = syncNodeGetPreTerm(pSyncNode, nextIndex);
if (preLogTerm == SYNC_TERM_INVALID) {
SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
// SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
SyncIndex newNextIndex = nextIndex + 1;
syncIndexMgrSetIndex(pSyncNode->pNextIndex, pDestId, newNextIndex);
syncIndexMgrSetIndex(pSyncNode->pMatchIndex, pDestId, SYNC_INDEX_INVALID);
sError("vgId:%d sync get pre term error, nextIndex:%" PRId64 ", update next-index:%" PRId64
@ -222,7 +223,8 @@ int32_t syncNodeAppendEntriesPeersSnapshot(SSyncNode* pSyncNode) {
SyncIndex preLogIndex = syncNodeGetPreIndex(pSyncNode, nextIndex);
SyncTerm preLogTerm = syncNodeGetPreTerm(pSyncNode, nextIndex);
if (preLogTerm == SYNC_TERM_INVALID) {
SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
// SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
SyncIndex newNextIndex = nextIndex + 1;
syncIndexMgrSetIndex(pSyncNode->pNextIndex, pDestId, newNextIndex);
syncIndexMgrSetIndex(pSyncNode->pMatchIndex, pDestId, SYNC_INDEX_INVALID);
sError("vgId:%d sync get pre term error, nextIndex:%" PRId64 ", update next-index:%" PRId64

45
source/libs/sync/test/sh/auto_bench.sh Normal file
View File

@ -0,0 +1,45 @@
#!/bin/bash
if [ $# != 5 ] ; then
echo "Uasge: $0 instances vgroups replica ctables rows"
echo ""
exit 1
fi
instances=$1
vgroups=$2
replica=$3
ctables=$4
rows=$5
echo "params: instances:${instances}, vgroups:${vgroups}, replica:${replica}, ctables:${ctables}, rows:${rows}"
dt=`date "+%Y-%m-%d-%H-%M-%S"`
casedir=instances_${instances}_vgroups_${vgroups}_replica_${replica}_ctables_${ctables}_rows_${rows}_${dt}
mkdir ${casedir}
cp ./insert.tpl.json ${casedir}
cd ${casedir}
for i in `seq 1 ${instances}`;do
#echo ===$i===
cfg_file=bench_${i}.json
cp ./insert.tpl.json ${cfg_file}
rstfile=result_${i}
sed -i 's/tpl_vgroups_tpl/'${vgroups}'/g' ${cfg_file}
sed -i 's/tpl_replica_tpl/'${replica}'/g' ${cfg_file}
sed -i 's/tpl_ctables_tpl/'${ctables}'/g' ${cfg_file}
sed -i 's/tpl_stid_tpl/'${i}'/g' ${cfg_file}
sed -i 's/tpl_rows_tpl/'${rows}'/g' ${cfg_file}
sed -i 's/tpl_insert_result_tpl/'${rstfile}'/g' ${cfg_file}
done
for conf_file in `ls ./bench_*.json`;do
echo "nohup taosBenchmark -f ${conf_file} &"
nohup taosBenchmark -f ${conf_file} &
done
cd -
exit 0

77
source/libs/sync/test/sh/insert.tpl.json Normal file
View File

@ -0,0 +1,77 @@
{
"filetype": "insert",
"cfgdir": "/etc/taos/",
"host": "v3cluster-0001",
"port": 7100,
"user": "root",
"password": "taosdata",
"thread_count": 8,
"thread_count_create_tbl": 8,
"result_file": "./tpl_insert_result_tpl",
"confirm_parameter_prompt": "no",
"insert_interval": 0,
"interlace_rows": 0,
"num_of_records_per_req": 100000,
"databases": [
{
"dbinfo": {
"name": "db1",
"drop": "yes",
"vgroups": tpl_vgroups_tpl,
"replica": tpl_replica_tpl
},
"super_tables": [
{
"name": "stb_tpl_stid_tpl",
"child_table_exists": "no",
"childtable_count": tpl_ctables_tpl,
"childtable_prefix": "stb_tpl_stid_tpl_",
"auto_create_table": "no",
"batch_create_tbl_num": 50000,
"data_source": "rand",
"insert_mode": "taosc",
"insert_rows": tpl_rows_tpl,
"interlace_rows": 0,
"insert_interval": 0,
"max_sql_len": 10000000,
"disorder_ratio": 0,
"disorder_range": 1000,
"timestamp_step": 10,
"sample_format": "csv",
"use_sample_ts": "no",
"tags_file": "",
"columns": [
{
"type": "INT"
},
{
"type": "DOUBLE",
"count": 1
},
{
"type": "BINARY",
"len": 40,
"count": 1
},
{
"type": "nchar",
"len": 20,
"count": 1
}
],
"tags": [
{
"type": "TINYINT",
"count": 1
},
{
"type": "BINARY",
"len": 16,
"count": 1
}
]
}
]
}
]
}

2
source/os/CMakeLists.txt
View File

@ -41,7 +41,7 @@ target_link_libraries(
)
if(TD_WINDOWS)
target_link_libraries(
os PUBLIC ws2_32 iconv msvcregex wcwidth winmm
os PUBLIC ws2_32 iconv msvcregex wcwidth winmm crashdump
)
elseif(TD_DARWIN_64)
target_link_libraries(

4
source/os/src/osSysinfo.c
View File

@ -91,6 +91,7 @@ LONG WINAPI FlCrashDump(PEXCEPTION_POINTERS ep) {
return EXCEPTION_CONTINUE_SEARCH;
}
LONG WINAPI exceptionHandler(LPEXCEPTION_POINTERS exception);
#elif defined(_TD_DARWIN_64)
@ -841,7 +842,8 @@ char *taosGetCmdlineByPID(int pid) {
void taosSetCoreDump(bool enable) {
#ifdef WINDOWS
SetUnhandledExceptionFilter(&FlCrashDump);
// SetUnhandledExceptionFilter(exceptionHandler);
// SetUnhandledExceptionFilter(&FlCrashDump);
#elif defined(_TD_DARWIN_64)
#else
if (!enable) return;

57
source/os/src/osSystem.c
View File

@ -18,6 +18,63 @@
#include "os.h"
#if defined(WINDOWS)
typedef void (*MainWindows)(int argc,char** argv);
MainWindows mainWindowsFunc = NULL;
SERVICE_STATUS ServiceStatus;
SERVICE_STATUS_HANDLE hServiceStatusHandle;
void WINAPI windowsServiceCtrlHandle(DWORD request) {
switch (request) {
case SERVICE_CONTROL_STOP:
case SERVICE_CONTROL_SHUTDOWN:
raise(SIGINT);
ServiceStatus.dwCurrentState = SERVICE_STOP_PENDING;
if (!SetServiceStatus(hServiceStatusHandle, &ServiceStatus)) {
DWORD nError = GetLastError();
printf("failed to send stopped status to windows service: %d",nError);
}
break;
default:
return;
}
}
void WINAPI mainWindowsService(int argc,char** argv) {
int ret = 0;
ServiceStatus.dwServiceType = SERVICE_WIN32;
ServiceStatus.dwControlsAccepted = SERVICE_ACCEPT_PAUSE_CONTINUE | SERVICE_ACCEPT_STOP | SERVICE_ACCEPT_SHUTDOWN;
ServiceStatus.dwCurrentState = SERVICE_START_PENDING;
ServiceStatus.dwWin32ExitCode = 0;
ServiceStatus.dwCheckPoint = 0;
ServiceStatus.dwWaitHint = 0;
ServiceStatus.dwServiceSpecificExitCode = 0;
hServiceStatusHandle = RegisterServiceCtrlHandler("taosd", &windowsServiceCtrlHandle);
if (hServiceStatusHandle == 0) {
DWORD nError = GetLastError();
printf("failed to register windows service ctrl handler: %d",nError);
}
ServiceStatus.dwCurrentState = SERVICE_RUNNING;
if (SetServiceStatus(hServiceStatusHandle, &ServiceStatus)) {
DWORD nError = GetLastError();
printf("failed to send running status to windows service: %d",nError);
}
if (mainWindowsFunc != NULL) mainWindowsFunc(argc, argv);
ServiceStatus.dwCurrentState = SERVICE_STOPPED;
if (!SetServiceStatus(hServiceStatusHandle, &ServiceStatus)) {
DWORD nError = GetLastError();
printf("failed to send stopped status to windows service: %d",nError);
}
}
void stratWindowsService(MainWindows mainWindows) {
mainWindowsFunc = mainWindows;
SERVICE_TABLE_ENTRY ServiceTable[2];
ServiceTable[0].lpServiceName = "taosd";
ServiceTable[0].lpServiceProc = (LPSERVICE_MAIN_FUNCTION)mainWindowsService;
ServiceTable[1].lpServiceName = NULL;
ServiceTable[1].lpServiceProc = NULL;
StartServiceCtrlDispatcher(ServiceTable);
}
#elif defined(_TD_DARWIN_64)
#else
#include <dlfcn.h>

2
source/util/src/tlog.c
View File

@ -82,7 +82,7 @@ int64_t tsNumOfTraceLogs = 0;
// log
int32_t dDebugFlag = 135;
int32_t vDebugFlag = 135;
int32_t mDebugFlag = 131;
int32_t mDebugFlag = 135;
int32_t cDebugFlag = 131;
int32_t jniDebugFlag = 131;
int32_t tmrDebugFlag = 131;

35
tests/script/jenkins/basic.txt
View File

@ -89,13 +89,20 @@
./test.sh -f tsim/parser/alter_column.sim
./test.sh -f tsim/parser/alter_stable.sim
./test.sh -f tsim/parser/alter.sim
# TD-17661 ./test.sh -f tsim/parser/alter1.sim
# TD-17959 ./test.sh -f tsim/parser/alter1.sim
./test.sh -f tsim/parser/auto_create_tb_drop_tb.sim
./test.sh -f tsim/parser/auto_create_tb.sim
./test.sh -f tsim/parser/between_and.sim
./test.sh -f tsim/parser/binary_escapeCharacter.sim
# TD-17738 ./test.sh -f tsim/parser/col_arithmetic_operation.sim
# TD-17661 ./test.sh -f tsim/parser/columnValue.sim
./test.sh -f tsim/parser/col_arithmetic_operation.sim
./test.sh -f tsim/parser/columnValue_bigint.sim
./test.sh -f tsim/parser/columnValue_bool.sim
./test.sh -f tsim/parser/columnValue_double.sim
./test.sh -f tsim/parser/columnValue_float.sim
./test.sh -f tsim/parser/columnValue_int.sim
./test.sh -f tsim/parser/columnValue_smallint.sim
./test.sh -f tsim/parser/columnValue_tinyint.sim
./test.sh -f tsim/parser/columnValue_unsign.sim
./test.sh -f tsim/parser/commit.sim
./test.sh -f tsim/parser/condition.sim
./test.sh -f tsim/parser/constCol.sim
@ -145,7 +152,7 @@
./test.sh -f tsim/parser/select_across_vnodes.sim
./test.sh -f tsim/parser/select_distinct_tag.sim
./test.sh -f tsim/parser/select_from_cache_disk.sim
# TD-17832 ./test.sh -f tsim/parser/select_with_tags.sim
./test.sh -f tsim/parser/select_with_tags.sim
./test.sh -f tsim/parser/selectResNum.sim
./test.sh -f tsim/parser/set_tag_vals.sim
./test.sh -f tsim/parser/single_row_in_tb.sim
@ -154,15 +161,15 @@
./test.sh -f tsim/parser/slimit.sim
./test.sh -f tsim/parser/slimit1.sim
./test.sh -f tsim/parser/stableOp.sim
# TD-17661 ./test.sh -f tsim/parser/tags_dynamically_specifiy.sim
# TD-17661 ./test.sh -f tsim/parser/tags_filter.sim
./test.sh -f tsim/parser/tags_dynamically_specifiy.sim
./test.sh -f tsim/parser/tags_filter.sim
./test.sh -f tsim/parser/tbnameIn.sim
./test.sh -f tsim/parser/timestamp.sim
./test.sh -f tsim/parser/top_groupby.sim
./test.sh -f tsim/parser/topbot.sim
./test.sh -f tsim/parser/union.sim
# TD-17704 ./test.sh -f tsim/parser/union_sysinfo.sim
# TD-17661 ./test.sh -f tsim/parser/where.sim
./test.sh -f tsim/parser/union_sysinfo.sim
./test.sh -f tsim/parser/where.sim
# ---- query ----
./test.sh -f tsim/query/charScalarFunction.sim
@ -422,18 +429,18 @@
./test.sh -f tsim/tag/bool_binary.sim
./test.sh -f tsim/tag/bool_int.sim
./test.sh -f tsim/tag/bool.sim
# TD-17661 ./test.sh -f tsim/tag/change.sim
# TD-17407 ./test.sh -f tsim/tag/change.sim
./test.sh -f tsim/tag/column.sim
./test.sh -f tsim/tag/commit.sim
# TD-17661 ./test.sh -f tsim/tag/create.sim
# TD-17661 ./test.sh -f tsim/tag/delete.sim
# TD-17661 ./test.sh -f tsim/tag/double.sim
# TD-17661 ./test.sh -f tsim/tag/filter.sim
# TD-17407 ./test.sh -f tsim/tag/create.sim
# TD-17407 ./test.sh -f tsim/tag/delete.sim
# TD-17407 ./test.sh -f tsim/tag/double.sim
./test.sh -f tsim/tag/filter.sim
# TD-17407 ./test.sh -f tsim/tag/float.sim
./test.sh -f tsim/tag/int_binary.sim
./test.sh -f tsim/tag/int_float.sim
./test.sh -f tsim/tag/int.sim
# TD-17661 ./test.sh -f tsim/tag/set.sim
# TD-17959 ./test.sh -f tsim/tag/set.sim
./test.sh -f tsim/tag/smallint.sim
./test.sh -f tsim/tag/tinyint.sim

Some files were not shown because too many files have changed in this diff Show More