other: merge main.

cmake/apr-util_CMakeLists.txt.in

+# apr-util
+ExternalProject_Add(aprutil-1
+        URL https://dlcdn.apache.org//apr/apr-util-1.6.3.tar.gz
+        URL_HASH SHA256=2b74d8932703826862ca305b094eef2983c27b39d5c9414442e9976a9acf1983
+        DOWNLOAD_NO_PROGRESS 1
+        DOWNLOAD_DIR "${TD_CONTRIB_DIR}/deps-download"
+        #GIT_REPOSITORY https://github.com/apache/apr-util.git
+	#GIT_TAG 1.5.4
+        SOURCE_DIR "${TD_CONTRIB_DIR}/apr-util"
+        #BINARY_DIR ""
+        BUILD_IN_SOURCE TRUE
+        BUILD_ALWAYS 1
+        #UPDATE_COMMAND ""
+        CONFIGURE_COMMAND ./configure --prefix=$ENV{HOME}/.cos-local.1/ --with-apr=$ENV{HOME}/.cos-local.1
+        #CONFIGURE_COMMAND ./configure --with-apr=/usr/local/apr
+        BUILD_COMMAND make
+        INSTALL_COMMAND make install
+        TEST_COMMAND ""
+)

cmake/apr_CMakeLists.txt.in

+# apr
+ExternalProject_Add(apr-1
+        URL https://dlcdn.apache.org//apr/apr-1.7.4.tar.gz
+        URL_HASH SHA256=a4137dd82a185076fa50ba54232d920a17c6469c30b0876569e1c2a05ff311d9
+        DOWNLOAD_NO_PROGRESS 1
+        DOWNLOAD_DIR "${TD_CONTRIB_DIR}/deps-download"
+        #GIT_REPOSITORY https://github.com/apache/apr.git
+	#GIT_TAG 1.5.2
+        SOURCE_DIR "${TD_CONTRIB_DIR}/apr"
+        BUILD_IN_SOURCE TRUE
+        UPDATE_DISCONNECTED TRUE
+        BUILD_ALWAYS 1
+        #UPDATE_COMMAND ""
+        CONFIGURE_COMMAND ./configure --prefix=$ENV{HOME}/.cos-local.1/ --enable-shared=no
+        #CONFIGURE_COMMAND ./configure
+        BUILD_COMMAND make
+        INSTALL_COMMAND make install
+        TEST_COMMAND ""
+)

cmake/cmake.define

@@ -77,6 +77,12 @@ ELSE ()
     SET(TD_TAOS_TOOLS TRUE)
 ENDIF ()
 
+IF (${TD_WINDOWS})
+    SET(TAOS_LIB taos_static)
+ELSE ()
+    SET(TAOS_LIB taos)
+ENDIF ()
+
 IF (TD_WINDOWS)
     MESSAGE("${Yellow} set compiler flag for Windows! ${ColourReset}")
     SET(COMMON_FLAGS "/w /D_WIN32 /DWIN32 /Zi /MTd")

cmake/cmake.options

@@ -125,6 +125,16 @@ option(
     ON
 )
 
+IF(${TD_LINUX})
+
+option(
+    BUILD_WITH_COS
+    "If build with cos"
+    ON
+)
+
+ENDIF ()
+
 option(
     BUILD_WITH_SQLITE
     "If build with sqlite" 

cmake/cos_CMakeLists.txt.in

+# cos
+ExternalProject_Add(cos
+        GIT_REPOSITORY https://github.com/tencentyun/cos-c-sdk-v5.git
+	GIT_TAG v5.0.16
+        SOURCE_DIR "${TD_CONTRIB_DIR}/cos-c-sdk-v5"
+        BINARY_DIR ""
+        #BUILD_IN_SOURCE TRUE
+        CONFIGURE_COMMAND ""
+        BUILD_COMMAND ""
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+)

cmake/curl_CMakeLists.txt.in

+# curl
+ExternalProject_Add(curl
+        URL  https://curl.se/download/curl-8.2.1.tar.gz
+        DOWNLOAD_NO_PROGRESS 1
+        DOWNLOAD_DIR "${TD_CONTRIB_DIR}/deps-download"
+        #GIT_REPOSITORY https://github.com/curl/curl.git
+        #GIT_TAG curl-7_88_1
+        SOURCE_DIR "${TD_CONTRIB_DIR}/curl"
+        BUILD_IN_SOURCE TRUE
+        BUILD_ALWAYS 1
+        #UPDATE_COMMAND ""
+        CONFIGURE_COMMAND ./configure --prefix=$ENV{HOME}/.cos-local.1 --without-ssl --enable-shared=no --disable-ldap --disable-ldaps --without-brotli
+        #CONFIGURE_COMMAND ./configure --without-ssl
+        BUILD_COMMAND make
+        INSTALL_COMMAND make install
+        TEST_COMMAND ""
+)

cmake/mxml_CMakeLists.txt.in

+# cos
+ExternalProject_Add(mxml
+        GIT_REPOSITORY https://github.com/michaelrsweet/mxml.git
+	GIT_TAG release-2.10
+        SOURCE_DIR "${TD_CONTRIB_DIR}/mxml"
+        #BINARY_DIR ""
+        BUILD_IN_SOURCE TRUE
+        #UPDATE_COMMAND ""
+        CONFIGURE_COMMAND ./configure --prefix=$ENV{HOME}/.cos-local.1 --enable-shared=no
+        #CONFIGURE_COMMAND ./configure
+        BUILD_COMMAND make
+        INSTALL_COMMAND make install
+        TEST_COMMAND ""
+)

contrib/CMakeLists.txt

@@ -6,6 +6,39 @@ function(cat IN_FILE OUT_FILE)
     file(APPEND ${OUT_FILE} "${CONTENTS}")
 endfunction(cat IN_FILE OUT_FILE)
 
+if(${TD_LINUX})
+
+set(CONTRIB_TMP_FILE3 "${CMAKE_BINARY_DIR}/deps_tmp_CMakeLists.txt.in3")
+configure_file("${TD_SUPPORT_DIR}/deps_CMakeLists.txt.in" ${CONTRIB_TMP_FILE3})
+
+if(${BUILD_WITH_COS})
+    file(MAKE_DIRECTORY $ENV{HOME}/.cos-local.1/)
+    cat("${TD_SUPPORT_DIR}/mxml_CMakeLists.txt.in" ${CONTRIB_TMP_FILE3})
+    cat("${TD_SUPPORT_DIR}/apr_CMakeLists.txt.in" ${CONTRIB_TMP_FILE3})
+    cat("${TD_SUPPORT_DIR}/curl_CMakeLists.txt.in" ${CONTRIB_TMP_FILE3})
+endif(${BUILD_WITH_COS})
+
+configure_file(${CONTRIB_TMP_FILE3} "${TD_CONTRIB_DIR}/deps-download/CMakeLists.txt")
+execute_process(COMMAND "${CMAKE_COMMAND}" -G "${CMAKE_GENERATOR}" .
+        WORKING_DIRECTORY "${TD_CONTRIB_DIR}/deps-download")
+execute_process(COMMAND "${CMAKE_COMMAND}" --build .
+        WORKING_DIRECTORY "${TD_CONTRIB_DIR}/deps-download")
+
+set(CONTRIB_TMP_FILE2 "${CMAKE_BINARY_DIR}/deps_tmp_CMakeLists.txt.in2")
+configure_file("${TD_SUPPORT_DIR}/deps_CMakeLists.txt.in" ${CONTRIB_TMP_FILE2})
+
+if(${BUILD_WITH_COS})
+    cat("${TD_SUPPORT_DIR}/apr-util_CMakeLists.txt.in" ${CONTRIB_TMP_FILE2})
+endif(${BUILD_WITH_COS})
+
+configure_file(${CONTRIB_TMP_FILE2} "${TD_CONTRIB_DIR}/deps-download/CMakeLists.txt")
+execute_process(COMMAND "${CMAKE_COMMAND}" -G "${CMAKE_GENERATOR}" .
+        WORKING_DIRECTORY "${TD_CONTRIB_DIR}/deps-download")
+execute_process(COMMAND "${CMAKE_COMMAND}" --build .
+        WORKING_DIRECTORY "${TD_CONTRIB_DIR}/deps-download")
+
+endif(${TD_LINUX})
+
 set(CONTRIB_TMP_FILE "${CMAKE_BINARY_DIR}/deps_tmp_CMakeLists.txt.in")
 configure_file("${TD_SUPPORT_DIR}/deps_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
 
@@ -122,6 +155,16 @@ if(${BUILD_WITH_SQLITE})
     cat("${TD_SUPPORT_DIR}/sqlite_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
 endif(${BUILD_WITH_SQLITE})
 
+# cos
+if(${BUILD_WITH_COS})
+    #cat("${TD_SUPPORT_DIR}/mxml_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    #cat("${TD_SUPPORT_DIR}/apr_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    #cat("${TD_SUPPORT_DIR}/apr-util_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    #cat("${TD_SUPPORT_DIR}/curl_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    cat("${TD_SUPPORT_DIR}/cos_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    add_definitions(-DUSE_COS)
+endif(${BUILD_WITH_COS})
+
 # lucene
 if(${BUILD_WITH_LUCENE})
     cat("${TD_SUPPORT_DIR}/lucene_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
@@ -347,6 +390,31 @@ if (${BUILD_WITH_ROCKSDB})
     endif()
 endif()
 
+# cos
+if(${BUILD_WITH_COS})
+  if(${TD_LINUX})
+    set(CMAKE_PREFIX_PATH $ENV{HOME}/.cos-local.1)
+    #ADD_DEFINITIONS(-DMINIXML_LIBRARY=${CMAKE_BINARY_DIR}/build/lib/libxml.a)
+    option(ENABLE_TEST "Enable the tests" OFF)
+    INCLUDE_DIRECTORIES($ENV{HOME}/.cos-local.1/include)
+    MESSAGE("$ENV{HOME}/.cos-local.1/include")
+
+    set(CMAKE_BUILD_TYPE debug)
+    set(ORIG_CMAKE_PROJECT_NAME ${CMAKE_PROJECT_NAME})
+    set(CMAKE_PROJECT_NAME cos_c_sdk)
+
+    add_subdirectory(cos-c-sdk-v5 EXCLUDE_FROM_ALL)
+    target_include_directories(
+      cos_c_sdk
+      PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/cos-c-sdk-v5/cos_c_sdk>
+      )
+
+    set(CMAKE_PROJECT_NAME ${ORIG_CMAKE_PROJECT_NAME})
+  else()
+
+  endif(${TD_LINUX})
+endif(${BUILD_WITH_COS})
+
 # lucene
 # To support build on ubuntu: sudo apt-get install libboost-all-dev
 if(${BUILD_WITH_LUCENE})

contrib/test/CMakeLists.txt

@@ -3,6 +3,11 @@ if(${BUILD_WITH_ROCKSDB})
     add_subdirectory(rocksdb)
 endif(${BUILD_WITH_ROCKSDB})
 
+# cos
+if(${BUILD_WITH_COS})
+    add_subdirectory(cos)
+endif(${BUILD_WITH_COS})
+
 if(${BUILD_WITH_LUCENE})
     add_subdirectory(lucene)
 endif(${BUILD_WITH_LUCENE})

contrib/test/cos/CMakeLists.txt

+add_executable(cosTest "")
+target_sources(cosTest
+  PRIVATE
+  "${CMAKE_CURRENT_SOURCE_DIR}/main.c"
+  )
+
+#find_path(APR_INCLUDE_DIR apr-1/apr_time.h)
+#find_path(APR_UTIL_INCLUDE_DIR apr/include/apr-1/apr_md5.h)
+#find_path(MINIXML_INCLUDE_DIR mxml.h)
+#find_path(CURL_INCLUDE_DIR curl/curl.h)
+
+#include_directories (${MINIXML_INCLUDE_DIR})
+#include_directories (${CURL_INCLUDE_DIR})
+FIND_PROGRAM(APR_CONFIG_BIN NAMES apr-config apr-1-config PATHS /usr/bin /usr/local/bin /usr/local/apr/bin/)
+#FIND_PROGRAM(APU_CONFIG_BIN NAMES apu-config apu-1-config PATHS /usr/bin /usr/local/bin /usr/local/apr/bin/)
+
+IF (APR_CONFIG_BIN)
+  EXECUTE_PROCESS(
+    COMMAND ${APR_CONFIG_BIN} --includedir
+    OUTPUT_VARIABLE APR_INCLUDE_DIR
+    OUTPUT_STRIP_TRAILING_WHITESPACE
+    )
+ENDIF()
+#IF (APU_CONFIG_BIN)
+#  EXECUTE_PROCESS(
+#    COMMAND ${APU_CONFIG_BIN} --includedir
+#    OUTPUT_VARIABLE APR_UTIL_INCLUDE_DIR
+#    OUTPUT_STRIP_TRAILING_WHITESPACE
+#    )
+#ENDIF()
+
+include_directories (${APR_INCLUDE_DIR})
+#include_directories (${APR_UTIL_INCLUDE_DIR})
+
+target_include_directories(
+  cosTest
+  PUBLIC "${TD_SOURCE_DIR}/contrib/cos-c-sdk-v5/cos_c_sdk"
+  )
+
+#find_library(APR_LIBRARY apr-1 PATHS /usr/local/apr/lib/)
+#find_library(APR_UTIL_LIBRARY aprutil-1 PATHS /usr/local/apr/lib/)
+#find_library(MINIXML_LIBRARY mxml)
+#find_library(CURL_LIBRARY curl)
+
+target_link_libraries(cosTest cos_c_sdk)
+target_link_libraries(cosTest apr-1})
+target_link_libraries(cosTest aprutil-1})
+target_link_libraries(cosTest mxml)
+target_link_libraries(cosTest curl)

docs/en/07-develop/07-tmq.mdx

@@ -24,24 +24,19 @@ By subscribing to a topic, a consumer can obtain the latest data in that topic i
 To implement these features, TDengine indexes its write-ahead log (WAL) file for fast random access and provides configurable methods for replacing and retaining this file. You can define a retention period and size for this file. For information, see the CREATE DATABASE statement. In this way, the WAL file is transformed into a persistent storage engine that remembers the order in which events occur. However, note that configuring an overly long retention period for your WAL files makes database compression inefficient. TDengine then uses the WAL file instead of the time-series database as its storage engine for queries in the form of topics. TDengine reads the data from the WAL file; uses a unified query engine instance to perform filtering, transformations, and other operations; and finally pushes the data to consumers.
 
 Tips:(c interface for example)
-- A consumption group consumes all data under the same topic, and different consumption groups are independent of each other;
-- A consumption group consumes all vgroups of the same topic, which can be composed of multiple consumers, but a vgroup is only consumed by one consumer. If the number of consumers exceeds the number of vgroups, the excess consumers do not consume data;
-- On the server side, only one offset is saved for each vgroup, and the offsets for each vgroup are monotonically increasing, but not necessarily continuous. There is no correlation between the offsets of various vgroups;
-- Each poll server will return a result block, which belongs to a vgroup and may contain data from multiple versions of wal. This block can be accessed through tmq_get_vgroup_offset. The offset interface obtains the offset of the first record in the block;
-- If a consumer group has never committed an offset, when its member consumers restart and pull data again, they start consuming from the set value of the parameter auto.offset.reset; In a consumer lifecycle, the client locally records the offset of the most recent pull data and will not pull duplicate data;
-- If a consumer terminates abnormally (without calling tmq_close), they need to wait for about 12 seconds to trigger their consumer group rebalance. The consumer's status on the server will change to LOST, and after about 1 day, the consumer will be automatically deleted; Exit normally, and after exiting, the consumer will be deleted; Add a new consumer, wait for about 2 seconds to trigger Rebalance, and the consumer's status on the server will change to ready;
-- The consumer group Rebalance will reassign Vgroups to all consumer members in the ready state of the group, and consumers can only assign/see/commit/poll operations to the Vgroups they are responsible for;
-- Consumers can tmq_position to obtain the offset of the current consumption, seek to the specified offset, and consume again;
-- Seek points the position to the specified offset without executing the commit operation. Once the seek is successful, it can poll the specified offset and subsequent data;
-- Position is to obtain the current consumption position, which is the position to be taken next time, not the current consumption position
-- Commit is the submission of the consumption location. Without parameters, it is the submission of the current consumption location (the location to be taken next time, not the current consumption location). With parameters, it is the location in the submission parameters (i.e. the location to be taken after the next exit and restart)
-- Seek is to set the consumer's consumption position. Wherever the seek goes, the position will be returned, all of which are the positions to be taken next time
-- Seek does not affect commit, commit does not affect seek, independent of each other, the two are different concepts
-- The begin interface is the offset of the first data in wal, and the end interface is the offset+1 of the last data in wal10.
-- Before the seek operation, tmq must be call tmq_get_topic_assignment, The assignment interface obtains the vgroup ID and offset range of the consumer. The seek operation will detect whether the vgroup ID and offset are legal, and if they are illegal, an error will be reported;
-- Due to the existence of a WAL expiration deletion mechanism, even if the seek operation is successful, it is possible that the offset has expired when polling data. If the offset of poll is less than the WAL minimum version number, it will be consumed from the WAL minimum version number;
-- The tmq_get_vgroup_offset interface obtains the offset of the first data in the result block where the record is located. When seeking to this offset, it will consume all the data in this block. Refer to point four;
-- Data subscription is to consume data from the wal. If some wal files are deleted according to WAL retention policy, the deleted data can't be consumed any more. So you need to set a reasonable value for parameter `WAL_RETENTION_PERIOD` or `WAL_RETENTION_SIZE` when creating the database and make sure your application consume the data in a timely way to make sure there is no data loss. This behavior is similar to Kafka and other widely used message queue products.
+1. A consumption group consumes all data under the same topic, and different consumption groups are independent of each other;
+2. A consumption group consumes all vgroups of the same topic, which can be composed of multiple consumers, but a vgroup is only consumed by one consumer. If the number of consumers exceeds the number of vgroups, the excess consumers do not consume data;
+3. On the server side, only one offset is saved for each vgroup, and the offsets for each vgroup are monotonically increasing, but not necessarily continuous. There is no correlation between the offsets of various vgroups;
+4. Each poll server will return a result block, which belongs to a vgroup and may contain data from multiple versions of wal. This block can be accessed through tmq_get_vgroup_offset. The offset interface obtains the offset of the first record in the block;
+5. If a consumer group has never committed an offset, when its member consumers restart and pull data again, they start consuming from the set value of the parameter auto.offset.reset; In a consumer lifecycle, the client locally records the offset of the most recent pull data and will not pull duplicate data;
+6. If a consumer terminates abnormally (without calling tmq_close), they need to wait for about 12 seconds to trigger their consumer group rebalance. The consumer's status on the server will change to LOST, and after about 1 day, the consumer will be automatically deleted; Exit normally, and after exiting, the consumer will be deleted; Add a new consumer, wait for about 2 seconds to trigger Rebalance, and the consumer's status on the server will change to ready;
+7. The consumer group Rebalance will reassign Vgroups to all consumer members in the ready state of the group, and consumers can only assign/see/commit/poll operations to the Vgroups they are responsible for;
+8. Consumers can tmq_position to obtain the offset of the current consumption, seek to the specified offset, and consume again;
+9. Seek points the position to the specified offset without executing the commit operation. Once the seek is successful, it can poll the specified offset and subsequent data;
+10. Before the seek operation, tmq must be call tmq_get_topic_assignment, The assignment interface obtains the vgroup ID and offset range of the consumer. The seek operation will detect whether the vgroup ID and offset are legal, and if they are illegal, an error will be reported;
+11. Due to the existence of a WAL expiration deletion mechanism, even if the seek operation is successful, it is possible that the offset has expired when polling data. If the offset of poll is less than the WAL minimum version number, it will be consumed from the WAL minimum version number;
+12. The tmq_get_vgroup_offset interface obtains the offset of the first data in the result block where the record is located. When seeking to this offset, it will consume all the data in this block. Refer to point four;
+13. Data subscription is to consume data from the wal. If some wal files are deleted according to WAL retention policy, the deleted data can't be consumed any more. So you need to set a reasonable value for parameter `WAL_RETENTION_PERIOD` or `WAL_RETENTION_SIZE` when creating the database and make sure your application consume the data in a timely way to make sure there is no data loss. This behavior is similar to Kafka and other widely used message queue products.
 
 ## Data Schema and API
 
@@ -104,8 +99,6 @@ The related schemas and APIs in various languages are described as follows:
     DLL_EXPORT const char *tmq_err2str(int32_t code);
 ```
 
-For more information, see [C/C++ Connector](/reference/connector/cpp).
-
 The following example is based on the smart meter table described in Data Models. For complete sample code, see the C language section below.
 
 </TabItem>

docs/en/10-deployment/01-deploy.md

@@ -62,12 +62,13 @@ serverPort            6030
 
 For all the dnodes in a TDengine cluster, the below parameters must be configured exactly the same, any node whose configuration is different from dnodes already in the cluster can't join the cluster.
 
-| **#** | **Parameter**      | **Definition**                                                                    |
-| ----- | ------------------ | ------------------------------------------- |
-| 1     | statusInterval     | The interval by which dnode reports its status to mnode                           |
-| 2     | timezone           | Timezone                                                                          |
-| 3     | locale             | System region and encoding                       |
-| 4     | charset            | Character set |
+| **#** | **Parameter**    | **Definition**                                                                |
+| ----- | ---------------- | ----------------------------------------------------------------------------- |
+| 1     | statusInterval   | The interval by which dnode reports its status to mnode                       |
+| 2     | timezone         | Timezone                                                                      |
+| 3     | locale           | System region and encoding                                                    |
+| 4     | charset          | Character set                                                                 |
+| 5     | ttlChangeOnWrite | Whether the ttl expiration time changes with the table modification operation |
 
 ## Start Cluster
 
@@ -97,7 +98,7 @@ Then, on the first dnode i.e. h1.tdengine.com in our example, use TDengine CLI `
 CREATE DNODE "h2.taos.com:6030";
 ````
 
-This adds the end point of the new dnode (from Step 4) into the end point list of the cluster. In the command "fqdn:port" should be quoted using double quotes. Change `"h2.taos.com:6030"` to the end point of your new dnode. 
+This adds the end point of the new dnode (from Step 4) into the end point list of the cluster. In the command "fqdn:port" should be quoted using double quotes. Change `"h2.taos.com:6030"` to the end point of your new dnode.
 
 Then on the first dnode h1.tdengine.com, execute `show dnodes` in `taos`
 

docs/en/12-taos-sql/10-function.md

@@ -698,7 +698,7 @@ ELAPSED(ts_primary_key [, time_unit])
 LEASTSQUARES(expr, start_val, step_val)
 ```
 
-**Description**: The linear regression function of the specified column and the timestamp column (primary key), `start_val` is the initial value and `step_val` is the step value.
+**Description**: The linear regression function of a specified column, `start_val` is the initial value and `step_val` is the step value.
 
 **Return value type**: A string in the format of "(slope, intercept)"
 

docs/en/14-reference/03-connector/04-java.mdx

@@ -1174,7 +1174,6 @@ public abstract class ConsumerLoop {
         config.setProperty("client.id", "1");
         config.setProperty("value.deserializer", "com.taosdata.jdbc.tmq.ConsumerTest.ConsumerLoop$ResultDeserializer");
         config.setProperty("value.deserializer.encoding", "UTF-8");
-        config.setProperty("experimental.snapshot.enable", "true");
 
         this.consumer = new TaosConsumer<>(config);
         this.topics = Collections.singletonList("topic_speed");

docs/zh/07-develop/07-tmq.mdx

@@ -26,24 +26,19 @@ import CDemo from "./_sub_c.mdx";
 本文档不对消息队列本身的基础知识做介绍，如果需要了解，请自行搜索。
 
 说明（以c接口为例）：
-- 一个消费组消费同一个topic下的所有数据，不同消费组之间相互独立；
-- 一个消费组消费同一个topic所有的vgroup，消费组可由多个消费者组成，但一个vgroup仅被一个消费者消费，如果消费者数量超过了vgroup数量，多余的消费者不消费数据；
-- 在服务端每个vgroup仅保存一个offset，每个vgroup的offset是单调递增的，但不一定连续。各个vgroup的offset之间没有关联；
-- 每次poll服务端会返回一个结果block，该block属于一个vgroup，可能包含多个wal版本的数据，可以通过 tmq_get_vgroup_offset 接口获得是该block第一条记录的offset；
-- 一个消费组如果从未commit过offset，当其成员消费者重启重新拉取数据时，均从参数auto.offset.reset设定值开始消费；在一个消费者生命周期中，客户端本地记录了最近一次拉取数据的offset，不会拉取重复数据；
-- 消费者如果异常终止（没有调用tmq_close），需等约12秒后触发其所属消费组rebalance，该消费者在服务端状态变为LOST，约1天后该消费者自动被删除；正常退出，退出后就会删除消费者；新增消费者，需等约2秒触发rebalance，该消费者在服务端状态变为ready；
-- 消费组rebalance会对该组所有ready状态的消费者成员重新进行vgroup分配，消费者仅能对自己负责的vgroup进行assignment/seek/commit/poll操作；
-- 消费者可利用 tmq_position 获得当前消费的offset，并seek到指定offset，重新消费；
-- seek将position指向指定offset，不执行commit操作，一旦seek成功，可poll拉取指定offset及以后的数据；
-- seek 操作之前须调用 tmq_get_topic_assignment 接口获取该consumer的vgroup ID和offset范围。seek 操作会检测vgroup ID 和 offset是否合法，如非法将报错；
-- position是获取当前的消费位置，是下次要取的位置，不是当前消费到的位置
-- commit是提交消费位置，不带参数的话，是提交当前消费位置（下次要取的位置，不是当前消费到的位置），带参数的话，是提交参数里的位置（也即下次退出重启后要取的位置）
-- seek是设置consumer消费位置，seek到哪，position就返回哪，都是下次要取的位置
-- seek不会影响commit，commit不影响seek，相互独立，两个是不同的概念
-- begin接口为wal 第一条数据的offset，end 接口为wal 最后一条数据的offset + 1
-- tmq_get_vgroup_offset接口获取的是记录所在结果block块里的第一条数据的offset，当seek至该offset时，将消费到这个block里的全部数据。参见第四点；
-- 由于存在 WAL 过期删除机制，即使seek 操作成功，poll数据时有可能offset已失效。如果poll 的offset 小于 WAL 最小版本号，将会从WAL最小版本号消费；
-- 数据订阅是从 WAL 消费数据，如果一些 WAL 文件被基于 WAL 保留策略删除，则已经删除的 WAL 文件中的数据就无法再消费到。需要根据业务需要在创建数据库时合理设置 `WAL_RETENTION_PERIOD` 或 `WAL_RETENTION_SIZE` ，并确保应用及时消费数据，这样才不会产生数据丢失的现象。数据订阅的行为与 Kafka 等广泛使用的消息队列类产品的行为相似；
+1. 一个消费组消费同一个topic下的所有数据，不同消费组之间相互独立；
+2. 一个消费组消费同一个topic所有的vgroup，消费组可由多个消费者组成，但一个vgroup仅被一个消费者消费，如果消费者数量超过了vgroup数量，多余的消费者不消费数据；
+3. 在服务端每个vgroup仅保存一个offset，每个vgroup的offset是单调递增的，但不一定连续。各个vgroup的offset之间没有关联；
+4. 每次poll服务端会返回一个结果block，该block属于一个vgroup，可能包含多个wal版本的数据，可以通过 tmq_get_vgroup_offset 接口获得是该block第一条记录的offset；
+5. 一个消费组如果从未commit过offset，当其成员消费者重启重新拉取数据时，均从参数auto.offset.reset设定值开始消费；在一个消费者生命周期中，客户端本地记录了最近一次拉取数据的offset，不会拉取重复数据；
+6. 消费者如果异常终止（没有调用tmq_close），需等约12秒后触发其所属消费组rebalance，该消费者在服务端状态变为LOST，约1天后该消费者自动被删除；正常退出，退出后就会删除消费者；新增消费者，需等约2秒触发rebalance，该消费者在服务端状态变为ready；
+7. 消费组rebalance会对该组所有ready状态的消费者成员重新进行vgroup分配，消费者仅能对自己负责的vgroup进行assignment/seek/commit/poll操作；
+8. 消费者可利用 tmq_position 获得当前消费的offset，并seek到指定offset，重新消费；
+9. seek将position指向指定offset，不执行commit操作，一旦seek成功，可poll拉取指定offset及以后的数据；
+10. seek 操作之前须调用 tmq_get_topic_assignment 接口获取该consumer的vgroup ID和offset范围。seek 操作会检测vgroup ID 和 offset是否合法，如非法将报错；
+11. tmq_get_vgroup_offset接口获取的是记录所在结果block块里的第一条数据的offset，当seek至该offset时，将消费到这个block里的全部数据。参见第四点；
+12. 由于存在 WAL 过期删除机制，即使seek 操作成功，poll数据时有可能offset已失效。如果poll 的offset 小于 WAL 最小版本号，将会从WAL最小版本号消费；
+13. 数据订阅是从 WAL 消费数据，如果一些 WAL 文件被基于 WAL 保留策略删除，则已经删除的 WAL 文件中的数据就无法再消费到。需要根据业务需要在创建数据库时合理设置 `WAL_RETENTION_PERIOD` 或 `WAL_RETENTION_SIZE` ，并确保应用及时消费数据，这样才不会产生数据丢失的现象。数据订阅的行为与 Kafka 等广泛使用的消息队列类产品的行为相似；
 
 ## 主要数据结构和 API
 
@@ -60,17 +55,17 @@ import CDemo from "./_sub_c.mdx";
     typedef void(tmq_commit_cb(tmq_t *tmq, int32_t code, void *param));
 
     typedef enum tmq_conf_res_t {
-        TMQ_CONF_UNKNOWN = -2,
-        TMQ_CONF_INVALID = -1,
-        TMQ_CONF_OK = 0,
-    } tmq_conf_res_t;
+    TMQ_CONF_UNKNOWN = -2,
+    TMQ_CONF_INVALID = -1,
+    TMQ_CONF_OK = 0,
+} tmq_conf_res_t;
 
     typedef struct tmq_topic_assignment {
-        int32_t vgId;
-        int64_t currentOffset;
-        int64_t begin;
-        int64_t end;  // The last version of wal + 1
-    } tmq_topic_assignment;
+    int32_t vgId;
+    int64_t currentOffset;
+    int64_t begin;
+    int64_t end;
+} tmq_topic_assignment;
 
     DLL_EXPORT tmq_conf_t    *tmq_conf_new();
     DLL_EXPORT tmq_conf_res_t tmq_conf_set(tmq_conf_t *conf, const char *key, const char *value);
@@ -89,24 +84,24 @@ import CDemo from "./_sub_c.mdx";
     DLL_EXPORT int32_t   tmq_subscription(tmq_t *tmq, tmq_list_t **topics);
     DLL_EXPORT TAOS_RES *tmq_consumer_poll(tmq_t *tmq, int64_t timeout);
     DLL_EXPORT int32_t   tmq_consumer_close(tmq_t *tmq);
-    DLL_EXPORT int32_t   tmq_commit_sync(tmq_t *tmq, const TAOS_RES *msg); //Commit the msg’s offset + 1
+    DLL_EXPORT int32_t   tmq_commit_sync(tmq_t *tmq, const TAOS_RES *msg);
     DLL_EXPORT void      tmq_commit_async(tmq_t *tmq, const TAOS_RES *msg, tmq_commit_cb *cb, void *param);
     DLL_EXPORT int32_t   tmq_commit_offset_sync(tmq_t *tmq, const char *pTopicName, int32_t vgId, int64_t offset);
     DLL_EXPORT void      tmq_commit_offset_async(tmq_t *tmq, const char *pTopicName, int32_t vgId, int64_t offset, tmq_commit_cb *cb, void *param);
     DLL_EXPORT int32_t   tmq_get_topic_assignment(tmq_t *tmq, const char *pTopicName, tmq_topic_assignment **assignment,int32_t *numOfAssignment);
     DLL_EXPORT void      tmq_free_assignment(tmq_topic_assignment* pAssignment);
     DLL_EXPORT int32_t   tmq_offset_seek(tmq_t *tmq, const char *pTopicName, int32_t vgId, int64_t offset);
-    DLL_EXPORT int64_t   tmq_position(tmq_t *tmq, const char *pTopicName, int32_t vgId);  // The current offset is the offset of the last consumed message + 1
+    DLL_EXPORT int64_t   tmq_position(tmq_t *tmq, const char *pTopicName, int32_t vgId);
     DLL_EXPORT int64_t   tmq_committed(tmq_t *tmq, const char *pTopicName, int32_t vgId);
 
     DLL_EXPORT const char *tmq_get_topic_name(TAOS_RES *res);
     DLL_EXPORT const char *tmq_get_db_name(TAOS_RES *res);
     DLL_EXPORT int32_t     tmq_get_vgroup_id(TAOS_RES *res);
-    DLL_EXPORT int64_t     tmq_get_vgroup_offset(TAOS_RES* res);  // Get current offset of the result
-    DLL_EXPORT const char *tmq_err2str(int32_t code);
+    DLL_EXPORT int64_t     tmq_get_vgroup_offset(TAOS_RES* res);
+    DLL_EXPORT const char *tmq_err2str(int32_t code);DLL_EXPORT void           tmq_conf_set_auto_commit_cb(tmq_conf_t *conf, tmq_commit_cb *cb, void *param);
 ```
 
-这些 API 的文档请见 [C/C++ Connector](../../connector/cpp)，下面介绍一下它们的具体用法（超级表和子表结构请参考“数据建模”一节），完整的示例代码请见下面 C 语言的示例代码。
+下面介绍一下它们的具体用法（超级表和子表结构请参考“数据建模”一节），完整的示例代码请见下面 C 语言的示例代码。
 
 </TabItem>
 <TabItem value="java" label="Java">

docs/zh/10-deployment/01-deploy.md

@@ -62,12 +62,13 @@ serverPort            6030
 
 加入到集群中的数据节点 dnode，下表中涉及集群相关的参数必须完全相同，否则不能成功加入到集群中。
 
-| **#** | **配置参数名称**   | **含义**                                    |
-| ----- | ------------------ | ------------------------------------------- |
-| 1     | statusInterval     | dnode 向 mnode 报告状态时长                 |
-| 2     | timezone           | 时区                                        |
-| 3     | locale             | 系统区位信息及编码格式                       |
-| 4     | charset            | 字符集编码                                 |
+| **#** | **配置参数名称** | **含义**                             |
+| ----- | ---------------- | ------------------------------------ |
+| 1     | statusInterval   | dnode 向 mnode 报告状态时长          |
+| 2     | timezone         | 时区                                 |
+| 3     | locale           | 系统区位信息及编码格式               |
+| 4     | charset          | 字符集编码                           |
+| 5     | ttlChangeOnWrite | ttl 到期时间是否伴随表的修改操作改变 |
 
 ## 启动集群
 
@@ -196,10 +197,10 @@ dnodeID 是集群自动分配的，不得人工指定。它在生成时是递增
 1、建立集群时使用 CREATE DNODE 增加新节点后，新节点始终显示 offline 状态？
 ```sql
   1）首先要检查增加的新节点上的 taosd 服务是否已经正常启动
-  
+
   2）如果已经启动，再检查到新节点的网络是否通畅，可以使用 ping fqdn 验证下
-  
+
   3）如果前面两步都没有问题，这一步要检查新节点做为独立集群在运行了，可以使用 taos -h fqdn 连接上后，show dnodes; 命令查看.
-    如果显示的列表与你主节点上显示的不一致，说明此节点自己单独成立了一个集群，解决的方法是停止新节点上的服务，然后清空新节点上 
+    如果显示的列表与你主节点上显示的不一致，说明此节点自己单独成立了一个集群，解决的方法是停止新节点上的服务，然后清空新节点上
     taos.cfg 中配置的 dataDir 目录下的所有文件，重新启动新节点服务即可解决。
-```  
+```

docs/zh/12-taos-sql/10-function.md

@@ -700,7 +700,7 @@ ELAPSED(ts_primary_key [, time_unit])
 LEASTSQUARES(expr, start_val, step_val)
 ```
 
-**功能说明**：统计表中某列的值是主键（时间戳）的拟合直线方程。start_val 是自变量初始值，step_val 是自变量的步长值。
+**功能说明**：统计表中某列的值的拟合直线方程。start_val 是自变量初始值，step_val 是自变量的步长值。
 
 **返回数据类型**：字符串表达式（斜率, 截距）。
 

include/client/taos.h

@@ -272,7 +272,7 @@ typedef struct tmq_topic_assignment {
   int32_t vgId;
   int64_t currentOffset;
   int64_t begin;
-  int64_t end;  // The last invalidate version of wal + 1
+  int64_t end;
 } tmq_topic_assignment;
 
 DLL_EXPORT tmq_conf_t    *tmq_conf_new();
@@ -305,7 +305,7 @@ DLL_EXPORT int64_t   tmq_committed(tmq_t *tmq, const char *pTopicName, int32_t v
 DLL_EXPORT const char *tmq_get_topic_name(TAOS_RES *res);
 DLL_EXPORT const char *tmq_get_db_name(TAOS_RES *res);
 DLL_EXPORT int32_t     tmq_get_vgroup_id(TAOS_RES *res);
-DLL_EXPORT int64_t     tmq_get_vgroup_offset(TAOS_RES* res);  // Get current offset of the result
+DLL_EXPORT int64_t     tmq_get_vgroup_offset(TAOS_RES* res);
 DLL_EXPORT const char *tmq_err2str(int32_t code);
 
 /* ------------------------------ TAOSX -----------------------------------*/

include/libs/executor/storageapi.h

@@ -98,6 +98,16 @@ typedef struct SMTbCursor {
   int8_t      paused;
 } SMTbCursor;
 
+typedef struct SMCtbCursor {
+  SMeta   *pMeta;
+  void    *pCur;
+  tb_uid_t suid;
+  void    *pKey;
+  void    *pVal;
+  int      kLen;
+  int      vLen;
+} SMCtbCursor;
+
 typedef struct SRowBuffPos {
   void* pRowBuff;
   void* pKey;
@@ -278,13 +288,15 @@ typedef struct SStoreMeta {
   void (*getBasicInfo)(void* pVnode, const char** dbname, int32_t* vgId, int64_t* numOfTables,
                        int64_t* numOfNormalTables);  // vnodeGetInfo(void *pVnode, const char **dbname, int32_t *vgId) &
                                                      // metaGetTbNum(SMeta *pMeta) & metaGetNtbNum(SMeta *pMeta);
-
   int64_t (*getNumOfRowsInMem)(void* pVnode);
   /**
 int32_t vnodeGetCtbIdList(void *pVnode, int64_t suid, SArray *list);
 int32_t vnodeGetCtbIdListByFilter(void *pVnode, int64_t suid, SArray *list, bool (*filter)(void *arg), void *arg);
 int32_t vnodeGetStbIdList(void *pVnode, int64_t suid, SArray *list);
  */
+  SMCtbCursor* (*openCtbCursor)(void *pVnode, tb_uid_t uid, int lock);
+  void         (*closeCtbCursor)(SMCtbCursor *pCtbCur, int lock);
+  tb_uid_t     (*ctbCursorNext)(SMCtbCursor* pCur);
 } SStoreMeta;
 
 typedef struct SStoreMetaReader {

include/libs/nodes/nodes.h

@@ -169,7 +169,7 @@ typedef enum ENodeType {
   QUERY_NODE_REVOKE_STMT,
   QUERY_NODE_SHOW_DNODES_STMT,
   QUERY_NODE_SHOW_MNODES_STMT,
-  QUERY_NODE_SHOW_MODULES_STMT,
+//  QUERY_NODE_SHOW_MODULES_STMT,
   QUERY_NODE_SHOW_QNODES_STMT,
   QUERY_NODE_SHOW_SNODES_STMT,
   QUERY_NODE_SHOW_BNODES_STMT,

include/libs/nodes/plannodes.h

@@ -107,6 +107,7 @@ typedef struct SScanLogicNode {
   bool          sortPrimaryKey;
   bool          igLastNull;
   bool          groupOrderScan;
+  bool          onlyMetaCtbIdx; // for tag scan with no tbname
 } SScanLogicNode;
 
 typedef struct SJoinLogicNode {
@@ -334,7 +335,11 @@ typedef struct SScanPhysiNode {
   bool       groupOrderScan;
 } SScanPhysiNode;
 
-typedef SScanPhysiNode STagScanPhysiNode;
+typedef struct STagScanPhysiNode {
+  SScanPhysiNode scan;
+  bool       onlyMetaCtbIdx; //no tbname, tag index not used.
+} STagScanPhysiNode;
+
 typedef SScanPhysiNode SBlockDistScanPhysiNode;
 
 typedef struct SLastRowScanPhysiNode {

include/libs/tfs/tfs.h

@@ -300,6 +300,25 @@ void tfsClosedir(STfsDir *pDir);
  */
 int32_t tfsGetMonitorInfo(STfs *pTfs, SMonDiskInfo *pInfo);
 
+/**
+ * @brief Check if disk space available at level
+ *
+ * @param pTfs The fs object.
+ * #param level the level
+ * @return bool
+ */
+bool tfsDiskSpaceAvailable(STfs *pTfs, int32_t level);
+
+/**
+ * @brief Check if disk space sufficient at disk of level
+ *
+ * @param pTfs The fs object.
+ * @param level the level
+ * @param disk the disk
+ * @return bool
+ */
+bool tfsDiskSpaceSufficient(STfs *pTfs, int32_t level, int32_t disk);
+
 #ifdef __cplusplus
 }
 #endif

include/os/osFile.h

@@ -76,7 +76,7 @@ int32_t taosUnLockFile(TdFilePtr pFile);
 
 int32_t taosUmaskFile(int32_t maskVal);
 
-int32_t taosStatFile(const char *path, int64_t *size, int32_t *mtime);
+int32_t taosStatFile(const char *path, int64_t *size, int32_t *mtime, int32_t *atime);
 int32_t taosDevInoFile(TdFilePtr pFile, int64_t *stDev, int64_t *stIno);
 int32_t taosFStatFile(TdFilePtr pFile, int64_t *size, int32_t *mtime);
 bool    taosCheckExistFile(const char *pathname);

include/os/osTime.h

@@ -95,6 +95,8 @@ struct tm *taosLocalTime(const time_t *timep, struct tm *result, char *buf);
 struct tm *taosLocalTimeNolock(struct tm *result, const time_t *timep, int dst);
 time_t     taosTime(time_t *t);
 time_t     taosMktime(struct tm *timep);
+int64_t    user_mktime64(const uint32_t year, const uint32_t mon, const uint32_t day, const uint32_t hour,
+                         const uint32_t min, const uint32_t sec, int64_t time_zone);
 
 #ifdef __cplusplus
 }

source/client/CMakeLists.txt

@@ -16,7 +16,7 @@ target_include_directories(
 target_link_libraries(
     taos
     INTERFACE api
-    PRIVATE os util common transport nodes parser command planner catalog scheduler function qcom
+    PRIVATE os util common transport nodes parser command planner catalog scheduler function qcom geometry
 )
 
 if(TD_DARWIN_ARM64)
@@ -57,7 +57,7 @@ target_include_directories(
 target_link_libraries(
     taos_static
     INTERFACE api
-    PRIVATE os util common transport nodes parser command planner catalog scheduler function qcom
+    PRIVATE os util common transport nodes parser command planner catalog scheduler function qcom geometry
 )
 
 if(${BUILD_TEST})

source/client/inc/clientSml.h

@@ -33,6 +33,7 @@ extern "C" {
 #include "ttime.h"
 #include "ttypes.h"
 #include "cJSON.h"
+#include "geosWrapper.h"
 
 #if (defined(__GNUC__) && (__GNUC__ >= 3)) || (defined(__INTEL_COMPILER) && (__INTEL_COMPILER >= 800)) || defined(__clang__)
 #  define expect(expr,value)    (__builtin_expect ((expr),(value)) )
@@ -192,7 +193,7 @@ typedef struct {
   //
   SArray      *preLineTagKV;
   SArray      *maxTagKVs;
-  SArray      *masColKVs;
+  SArray      *maxColKVs;
 
   SSmlLineInfo preLine;
   STableMeta  *currSTableMeta;

source/client/src/clientSml.c

@@ -1073,6 +1073,7 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
   return 0;
 
 end:
+  taosHashCancelIterate(info->superTables, tmp);
   taosHashCleanup(hashTmp);
   taosMemoryFreeClear(pTableMeta);
   catalogRefreshTableMeta(info->pCatalog, &conn, &pName, 1);
@@ -1191,6 +1192,7 @@ void freeSSmlKv(void *data) {
   SSmlKv *kv = (SSmlKv *)data;
   if (kv->keyEscaped) taosMemoryFree((void *)(kv->key));
   if (kv->valueEscaped) taosMemoryFree((void *)(kv->value));
+  if (kv->type == TSDB_DATA_TYPE_GEOMETRY) geosFreeBuffer((void *)(kv->value));
 }
 
 void smlDestroyInfo(SSmlHandle *info) {
@@ -1433,6 +1435,7 @@ static int32_t smlInsertData(SSmlHandle *info) {
     code = smlCheckAuth(info, &conn, pName.tname, AUTH_TYPE_WRITE);
     if(code != TSDB_CODE_SUCCESS){
       taosMemoryFree(measure);
+      taosHashCancelIterate(info->childTables, oneTable);
       return code;
     }
 
@@ -1441,6 +1444,7 @@ static int32_t smlInsertData(SSmlHandle *info) {
     if (code != TSDB_CODE_SUCCESS) {
       uError("SML:0x%" PRIx64 " catalogGetTableHashVgroup failed. table name: %s", info->id, tableData->childTableName);
       taosMemoryFree(measure);
+      taosHashCancelIterate(info->childTables, oneTable);
       return code;
     }
     taosHashPut(info->pVgHash, (const char *)&vg.vgId, sizeof(vg.vgId), (char *)&vg, sizeof(vg));
@@ -1450,6 +1454,7 @@ static int32_t smlInsertData(SSmlHandle *info) {
     if (unlikely(NULL == pMeta || NULL == (*pMeta)->tableMeta)) {
       uError("SML:0x%" PRIx64 " NULL == pMeta. table name: %s", info->id, tableData->childTableName);
       taosMemoryFree(measure);
+      taosHashCancelIterate(info->childTables, oneTable);
       return TSDB_CODE_SML_INTERNAL_ERROR;
     }
 
@@ -1465,6 +1470,7 @@ static int32_t smlInsertData(SSmlHandle *info) {
     taosMemoryFree(measure);
     if (code != TSDB_CODE_SUCCESS) {
       uError("SML:0x%" PRIx64 " smlBindData failed", info->id);
+      taosHashCancelIterate(info->childTables, oneTable);
       return code;
     }
     oneTable = (SSmlTableInfo **)taosHashIterate(info->childTables, oneTable);

source/client/src/clientSmlLine.c

@@ -102,6 +102,30 @@ int32_t smlParseValue(SSmlKv *pVal, SSmlMsgBuf *msg) {
     return TSDB_CODE_TSC_INVALID_VALUE;
   }
 
+  if (pVal->value[0] == 'g' || pVal->value[0] == 'G') {  // geometry
+    if (pVal->value[1] == '"' && pVal->value[pVal->length - 1] == '"' && pVal->length >= sizeof("POINT")+3) {
+      int32_t code = initCtxGeomFromText();
+      if (code != TSDB_CODE_SUCCESS) {
+        return code;
+      }
+      char* tmp = taosMemoryCalloc(pVal->length, 1);
+      memcpy(tmp, pVal->value + 2, pVal->length - 3);
+      code = doGeomFromText(tmp, (unsigned char **)&pVal->value, &pVal->length);
+      taosMemoryFree(tmp);
+      if (code != TSDB_CODE_SUCCESS) {
+        return code;
+      }
+
+      pVal->type = TSDB_DATA_TYPE_GEOMETRY;
+      if (pVal->length > TSDB_MAX_BINARY_LEN - VARSTR_HEADER_SIZE) {
+        geosFreeBuffer((void*)(pVal->value));
+        return TSDB_CODE_PAR_INVALID_VAR_COLUMN_LEN;
+      }
+      return TSDB_CODE_SUCCESS;
+    }
+    return TSDB_CODE_TSC_INVALID_VALUE;
+  }
+
   if (pVal->value[0] == 't' || pVal->value[0] == 'T') {
     if (pVal->length == 1 ||
         (pVal->length == 4 && (pVal->value[1] == 'r' || pVal->value[1] == 'R') &&
@@ -390,14 +414,14 @@ static int32_t smlParseColKv(SSmlHandle *info, char **sql, char *sqlEnd, SSmlLin
           SSmlKv   kv = {.key = tag->name, .keyLen = strlen(tag->name), .type = tag->type};
           if (tag->type == TSDB_DATA_TYPE_NCHAR) {
             kv.length = (tag->bytes - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE;
-          } else if (tag->type == TSDB_DATA_TYPE_BINARY) {
+          } else if (tag->type == TSDB_DATA_TYPE_BINARY || tag->type == TSDB_DATA_TYPE_GEOMETRY) {
             kv.length = tag->bytes - VARSTR_HEADER_SIZE;
           }
           taosArrayPush((*tmp)->cols, &kv);
         }
       }
       info->currSTableMeta = (*tmp)->tableMeta;
-      info->masColKVs = (*tmp)->cols;
+      info->maxColKVs = (*tmp)->cols;
     }
   }
 
@@ -512,13 +536,13 @@ static int32_t smlParseColKv(SSmlHandle *info, char **sql, char *sqlEnd, SSmlLin
         freeSSmlKv(&kv);
         return TSDB_CODE_SUCCESS;
       }
-      if (cnt >= taosArrayGetSize(info->masColKVs)) {
+      if (cnt >= taosArrayGetSize(info->maxColKVs)) {
         info->dataFormat = false;
         info->reRun = true;
         freeSSmlKv(&kv);
         return TSDB_CODE_SUCCESS;
       }
-      SSmlKv *maxKV = (SSmlKv *)taosArrayGet(info->masColKVs, cnt);
+      SSmlKv *maxKV = (SSmlKv *)taosArrayGet(info->maxColKVs, cnt);
       if (kv.type != maxKV->type) {
         info->dataFormat = false;
         info->reRun = true;
@@ -663,14 +687,15 @@ int32_t smlParseInfluxString(SSmlHandle *info, char *sql, char *sqlEnd, SSmlLine
   if (info->dataFormat) {
     uDebug("SML:0x%" PRIx64 " smlParseInfluxString format true, ts:%" PRId64, info->id, ts);
     ret = smlBuildCol(info->currTableDataCtx, info->currSTableMeta->schema, &kv, 0);
-    if (ret != TSDB_CODE_SUCCESS) {
-      return ret;
+    if (ret == TSDB_CODE_SUCCESS) {
+      ret = smlBuildRow(info->currTableDataCtx);
     }
-    ret = smlBuildRow(info->currTableDataCtx);
-    if (ret != TSDB_CODE_SUCCESS) {
+
+    clearColValArray(info->currTableDataCtx->pValues);
+    if (unlikely(ret != TSDB_CODE_SUCCESS)) {
+      smlBuildInvalidDataMsg(&info->msgBuf, "smlBuildCol error", NULL);
       return ret;
     }
-    clearColValArray(info->currTableDataCtx->pValues);
   } else {
     uDebug("SML:0x%" PRIx64 " smlParseInfluxString format false, ts:%" PRId64, info->id, ts);
     taosArraySet(elements->colArray, 0, &kv);

source/client/test/CMakeLists.txt

@@ -20,7 +20,7 @@ TARGET_LINK_LIBRARIES(
 ADD_EXECUTABLE(smlTest smlTest.cpp)
 TARGET_LINK_LIBRARIES(
         smlTest
-        PUBLIC os util common transport parser catalog scheduler function gtest taos_static qcom
+        PUBLIC os util common transport parser catalog scheduler function gtest taos_static qcom geometry
 )
 
 TARGET_INCLUDE_DIRECTORIES(

source/common/CMakeLists.txt

@@ -16,6 +16,14 @@ ENDIF ()
 IF (TD_STORAGE)
   ADD_DEFINITIONS(-D_STORAGE)
   TARGET_LINK_LIBRARIES(common PRIVATE storage)
+
+  IF(${TD_LINUX})
+    IF(${BUILD_WITH_COS})
+      add_definitions(-DUSE_COS)
+    ENDIF(${BUILD_WITH_COS})
+
+  ENDIF(${TD_LINUX})
+
 ENDIF ()
 
 target_include_directories(

source/common/src/systable.c

@@ -314,7 +314,7 @@ static const SSysDbTableSchema userUserPrivilegesSchema[] = {
 static const SSysTableMeta infosMeta[] = {
     {TSDB_INS_TABLE_DNODES, dnodesSchema, tListLen(dnodesSchema), true},
     {TSDB_INS_TABLE_MNODES, mnodesSchema, tListLen(mnodesSchema), true},
-    {TSDB_INS_TABLE_MODULES, modulesSchema, tListLen(modulesSchema), true},
+    // {TSDB_INS_TABLE_MODULES, modulesSchema, tListLen(modulesSchema), true},
     {TSDB_INS_TABLE_QNODES, qnodesSchema, tListLen(qnodesSchema), true},
     {TSDB_INS_TABLE_SNODES, snodesSchema, tListLen(snodesSchema), true},
     {TSDB_INS_TABLE_CLUSTER, clusterSchema, tListLen(clusterSchema), true},

source/common/src/tglobal.c

@@ -238,6 +238,14 @@ bool    tsFilterScalarMode = false;
 int32_t tsKeepTimeOffset = 0;  // latency of data migration
 int     tsResolveFQDNRetryTime = 100; //seconds
 
+char   tsS3Endpoint[TSDB_FQDN_LEN] = "<endpoint>";
+char   tsS3AccessKey[TSDB_FQDN_LEN] = "<accesskey>";
+char   tsS3AccessKeyId[TSDB_FQDN_LEN] = "<accesskeyid>";
+char   tsS3AccessKeySecret[TSDB_FQDN_LEN] = "<accesskeysecrect>";
+char   tsS3BucketName[TSDB_FQDN_LEN] = "<bucketname>";
+char   tsS3AppId[TSDB_FQDN_LEN] = "<appid>";
+int8_t tsS3Enabled = false;
+
 #ifndef _STORAGE
 int32_t taosSetTfsCfg(SConfig *pCfg) {
   SConfigItem *pItem = cfgGetItem(pCfg, "dataDir");
@@ -259,7 +267,43 @@ int32_t taosSetTfsCfg(SConfig *pCfg) {
 int32_t taosSetTfsCfg(SConfig *pCfg);
 #endif
 
-struct SConfig *taosGetCfg() { return tsCfg; }
+int32_t taosSetS3Cfg(SConfig *pCfg) {
+  tstrncpy(tsS3AccessKey, cfgGetItem(pCfg, "s3Accesskey")->str, TSDB_FQDN_LEN);
+  if (tsS3AccessKey[0] == '<') {
+    return 0;
+  }
+  char *colon = strchr(tsS3AccessKey, ':');
+  if (!colon) {
+    uError("invalid access key:%s", tsS3AccessKey);
+    return -1;
+  }
+  *colon = '\0';
+  tstrncpy(tsS3AccessKeyId, tsS3AccessKey, TSDB_FQDN_LEN);
+  tstrncpy(tsS3AccessKeySecret, colon + 1, TSDB_FQDN_LEN);
+  tstrncpy(tsS3Endpoint, cfgGetItem(pCfg, "s3Endpoint")->str, TSDB_FQDN_LEN);
+  tstrncpy(tsS3BucketName, cfgGetItem(pCfg, "s3BucketName")->str, TSDB_FQDN_LEN);
+  char *cos = strstr(tsS3Endpoint, "cos.");
+  if (cos) {
+    char *appid = strrchr(tsS3BucketName, '-');
+    if (!appid) {
+      uError("failed to locate appid in bucket:%s", tsS3BucketName);
+      return -1;
+    } else {
+      tstrncpy(tsS3AppId, appid + 1, TSDB_FQDN_LEN);
+    }
+  }
+  if (tsS3BucketName[0] != '<' && tsDiskCfgNum > 1) {
+#ifdef USE_COS
+    tsS3Enabled = true;
+#endif
+  }
+
+  return 0;
+}
+
+struct SConfig *taosGetCfg() {
+  return tsCfg;
+}
 
 static int32_t taosLoadCfg(SConfig *pCfg, const char **envCmd, const char *inputCfgDir, const char *envFile,
                            char *apolloUrl) {
@@ -583,6 +627,10 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
   if (cfgAddInt32(pCfg, "pqSortMemThreshold", tsPQSortMemThreshold, 1, 10240, CFG_SCOPE_SERVER) != 0) return -1;
   if (cfgAddInt32(pCfg, "resolveFQDNRetryTime", tsResolveFQDNRetryTime, 1, 10240, 0) != 0) return -1;
 
+  if (cfgAddString(pCfg, "s3Accesskey", tsS3AccessKey, CFG_SCOPE_SERVER) != 0) return -1;
+  if (cfgAddString(pCfg, "s3Endpoint", tsS3Endpoint, CFG_SCOPE_SERVER) != 0) return -1;
+  if (cfgAddString(pCfg, "s3BucketName", tsS3BucketName, CFG_SCOPE_SERVER) != 0) return -1;
+
   GRANT_CFG_ADD;
   return 0;
 }
@@ -1505,6 +1553,7 @@ int32_t taosInitCfg(const char *cfgDir, const char **envCmd, const char *envFile
     if (taosSetServerCfg(tsCfg)) return -1;
     if (taosSetReleaseCfg(tsCfg)) return -1;
     if (taosSetTfsCfg(tsCfg) != 0) return -1;
+    if (taosSetS3Cfg(tsCfg) != 0) return -1;
   }
   taosSetSystemCfg(tsCfg);
 

source/common/src/ttime.c

@@ -25,46 +25,6 @@
 
 #include "tlog.h"
 
-/*
- * mktime64 - Converts date to seconds.
- * Converts Gregorian date to seconds since 1970-01-01 00:00:00.
- * Assumes input in normal date format, i.e. 1980-12-31 23:59:59
- * => year=1980, mon=12, day=31, hour=23, min=59, sec=59.
- *
- * [For the Julian calendar (which was used in Russia before 1917,
- * Britain & colonies before 1752, anywhere else before 1582,
- * and is still in use by some communities) leave out the
- * -year/100+year/400 terms, and add 10.]
- *
- * This algorithm was first published by Gauss (I think).
- *
- * A leap second can be indicated by calling this function with sec as
- * 60 (allowable under ISO 8601).  The leap second is treated the same
- * as the following second since they don't exist in UNIX time.
- *
- * An encoding of midnight at the end of the day as 24:00:00 - ie. midnight
- * tomorrow - (allowable under ISO 8601) is supported.
- */
-static int64_t user_mktime64(const uint32_t year0, const uint32_t mon0, const uint32_t day, const uint32_t hour,
-                             const uint32_t min, const uint32_t sec, int64_t time_zone) {
-  uint32_t mon = mon0, year = year0;
-
-  /* 1..12 -> 11,12,1..10 */
-  if (0 >= (int32_t)(mon -= 2)) {
-    mon += 12; /* Puts Feb last since it has leap day */
-    year -= 1;
-  }
-
-  // int64_t res = (((((int64_t) (year/4 - year/100 + year/400 + 367*mon/12 + day) +
-  //                year*365 - 719499)*24 + hour)*60 + min)*60 + sec);
-  int64_t res;
-  res = 367 * ((int64_t)mon) / 12;
-  res += year / 4 - year / 100 + year / 400 + day + ((int64_t)year) * 365 - 719499;
-  res = res * 24;
-  res = ((res + hour) * 60 + min) * 60 + sec;
-
-  return (res + time_zone);
-}
 
 // ==== mktime() kernel code =================//
 static int64_t m_deltaUtc = 0;

source/dnode/mgmt/mgmt_mnode/src/mmFile.c

@@ -46,7 +46,7 @@ static int32_t mmDecodeOption(SJson *pJson, SMnodeOpt *pOption) {
     if (code < 0) return -1;
     tjsonGetInt32ValueFromDouble(replica, "role", pOption->nodeRoles[i], code);
     if (code < 0) return -1;
-    if(pOption->nodeRoles[i] == TAOS_SYNC_ROLE_VOTER){
+    if (pOption->nodeRoles[i] == TAOS_SYNC_ROLE_VOTER) {
       pOption->numOfReplicas++;
     }
   }
@@ -65,7 +65,7 @@ int32_t mmReadFile(const char *path, SMnodeOpt *pOption) {
   char      file[PATH_MAX] = {0};
   snprintf(file, sizeof(file), "%s%smnode.json", path, TD_DIRSEP);
 
-  if (taosStatFile(file, NULL, NULL) < 0) {
+  if (taosStatFile(file, NULL, NULL, NULL) < 0) {
     dInfo("mnode file:%s not exist", file);
     return 0;
   }

source/dnode/mgmt/mgmt_vnode/src/vmFile.c

@@ -97,7 +97,7 @@ int32_t vmGetVnodeListFromFile(SVnodeMgmt *pMgmt, SWrapperCfg **ppCfgs, int32_t
   SWrapperCfg *pCfgs = NULL;
   snprintf(file, sizeof(file), "%s%svnodes.json", pMgmt->path, TD_DIRSEP);
 
-  if (taosStatFile(file, NULL, NULL) < 0) {
+  if (taosStatFile(file, NULL, NULL, NULL) < 0) {
     dInfo("vnode file:%s not exist", file);
     return 0;
   }

source/dnode/mgmt/mgmt_vnode/src/vmInt.c

@@ -460,7 +460,6 @@ static void vmCleanup(SVnodeMgmt *pMgmt) {
   vmCloseVnodes(pMgmt);
   vmStopWorker(pMgmt);
   vnodeCleanup();
-  tfsClose(pMgmt->pTfs);
   taosThreadRwlockDestroy(&pMgmt->lock);
   taosMemoryFree(pMgmt);
 }
@@ -535,20 +534,9 @@ static int32_t vmInit(SMgmtInputOpt *pInput, SMgmtOutputOpt *pOutput) {
   pMgmt->msgCb.mgmt = pMgmt;
   taosThreadRwlockInit(&pMgmt->lock, NULL);
 
-  SDiskCfg dCfg = {0};
-  tstrncpy(dCfg.dir, tsDataDir, TSDB_FILENAME_LEN);
-  dCfg.level = 0;
-  dCfg.primary = 1;
-  SDiskCfg *pDisks = tsDiskCfg;
-  int32_t   numOfDisks = tsDiskCfgNum;
-  if (numOfDisks <= 0 || pDisks == NULL) {
-    pDisks = &dCfg;
-    numOfDisks = 1;
-  }
-
-  pMgmt->pTfs = tfsOpen(pDisks, numOfDisks);
+  pMgmt->pTfs = pInput->pTfs;
   if (pMgmt->pTfs == NULL) {
-    dError("failed to init tfs since %s", terrstr());
+    dError("tfs is null.");
     goto _OVER;
   }
   tmsgReportStartup("vnode-tfs", "initialized");

source/dnode/mgmt/mgmt_vnode/src/vmWorker.c

@@ -15,6 +15,7 @@
 
 #define _DEFAULT_SOURCE
 #include "vmInt.h"
+#include "vnodeInt.h"
 
 static inline void vmSendRsp(SRpcMsg *pMsg, int32_t code) {
   if (pMsg->info.handle == NULL) return;
@@ -114,9 +115,10 @@ static void vmProcessFetchQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
     const STraceId *trace = &pMsg->info.traceId;
     dGTrace("vgId:%d, msg:%p get from vnode-fetch queue", pVnode->vgId, pMsg);
 
+    terrno = 0;
     int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, pMsg, pInfo);
     if (code != 0) {
-      if (terrno != 0) {
+      if (code == -1 && terrno != 0) {
         code = terrno;
       }
 
@@ -158,6 +160,15 @@ static void vmSendResponse(SRpcMsg *pMsg) {
   }
 }
 
+static bool vmDataSpaceSufficient(SVnodeObj *pVnode) {
+  STfs *pTfs = pVnode->pImpl->pTfs;
+  if (pTfs) {
+    return tfsDiskSpaceSufficient(pTfs, 0, pVnode->diskPrimary);
+  } else {
+    return osDataSpaceSufficient();
+  }
+}
+
 static int32_t vmPutMsgToQueue(SVnodeMgmt *pMgmt, SRpcMsg *pMsg, EQueueType qtype) {
   const STraceId *trace = &pMsg->info.traceId;
   if (pMsg->contLen < sizeof(SMsgHead)) {
@@ -203,7 +214,7 @@ static int32_t vmPutMsgToQueue(SVnodeMgmt *pMgmt, SRpcMsg *pMsg, EQueueType qtyp
       taosWriteQitem(pVnode->pFetchQ, pMsg);
       break;
     case WRITE_QUEUE:
-      if (!osDataSpaceSufficient()) {
+      if (!vmDataSpaceSufficient(pVnode)) {
         terrno = TSDB_CODE_NO_ENOUGH_DISKSPACE;
         code = terrno;
         dError("vgId:%d, msg:%p put into vnode-write queue failed since %s", pVnode->vgId, pMsg, terrstr(code));

source/dnode/mgmt/node_mgmt/inc/dmMgmt.h

@@ -20,6 +20,7 @@
 #include "uv.h"
 
 #include "dmInt.h"
+#include "tfs.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -79,6 +80,7 @@ typedef struct SDnode {
   TdThreadMutex mutex;
   TdFilePtr     lockfile;
   SDnodeData    data;
+  STfs         *pTfs;
   SMgmtWrapper  wrappers[NODE_END];
 } SDnode;
 
@@ -124,4 +126,4 @@ void dmGetQnodeLoads(SQnodeLoad *pInfo);
 }
 #endif
 
-#endif /*_TD_DND_MGMT_H_*/
+#endif /*_TD_DND_MGMT_H_*/
\ No newline at end of file

source/dnode/mgmt/node_mgmt/src/dmEnv.c

@@ -96,28 +96,23 @@ _exit:
   return code;
 }
 
-static bool dmCheckDiskSpace() {
-  osUpdate();
-  // sufficiency
-  if (!osDataSpaceSufficient()) {
-    dWarn("free data disk size: %f GB, not sufficient, expected %f GB at least",
-          (double)tsDataSpace.size.avail / 1024.0 / 1024.0 / 1024.0,
-          (double)tsDataSpace.reserved / 1024.0 / 1024.0 / 1024.0);
-  }
-  if (!osLogSpaceSufficient()) {
-    dWarn("free log disk size: %f GB, not sufficient, expected %f GB at least",
-          (double)tsLogSpace.size.avail / 1024.0 / 1024.0 / 1024.0,
-          (double)tsLogSpace.reserved / 1024.0 / 1024.0 / 1024.0);
+static bool dmDataSpaceAvailable() {
+  SDnode *pDnode = dmInstance();
+  if (pDnode->pTfs) {
+    return tfsDiskSpaceAvailable(pDnode->pTfs, 0);
   }
-  if (!osTempSpaceSufficient()) {
-    dWarn("free temp disk size: %f GB, not sufficient, expected %f GB at least",
-          (double)tsTempSpace.size.avail / 1024.0 / 1024.0 / 1024.0,
-          (double)tsTempSpace.reserved / 1024.0 / 1024.0 / 1024.0);
+  if (!osDataSpaceAvailable()) {
+    dError("data disk space unavailable, i.e. %s", tsDataDir);
+    return false;
   }
+  return true;
+}
+
+static bool dmCheckDiskSpace() {
+  osUpdate();
   // availability
   bool ret = true;
-  if (!osDataSpaceAvailable()) {
-    dError("data disk space unavailable, i.e. %s", tsDataDir);
+  if (!dmDataSpaceAvailable()) {
     terrno = TSDB_CODE_NO_DISKSPACE;
     ret = false;
   }
@@ -134,6 +129,34 @@ static bool dmCheckDiskSpace() {
   return ret;
 }
 
+int32_t dmDiskInit() {
+  SDnode  *pDnode = dmInstance();
+  SDiskCfg dCfg = {0};
+  tstrncpy(dCfg.dir, tsDataDir, TSDB_FILENAME_LEN);
+  dCfg.level = 0;
+  dCfg.primary = 1;
+  SDiskCfg *pDisks = tsDiskCfg;
+  int32_t   numOfDisks = tsDiskCfgNum;
+  if (numOfDisks <= 0 || pDisks == NULL) {
+    pDisks = &dCfg;
+    numOfDisks = 1;
+  }
+
+  pDnode->pTfs = tfsOpen(pDisks, numOfDisks);
+  if (pDnode->pTfs == NULL) {
+    dError("failed to init tfs since %s", terrstr());
+    return -1;
+  }
+  return 0;
+}
+
+int32_t dmDiskClose() {
+  SDnode *pDnode = dmInstance();
+  tfsClose(pDnode->pTfs);
+  pDnode->pTfs = NULL;
+  return 0;
+}
+
 static bool dmCheckDataDirVersion() {
   char checkDataDirJsonFileName[PATH_MAX] = {0};
   snprintf(checkDataDirJsonFileName, PATH_MAX, "%s/dnode/dnodeCfg.json", tsDataDir);
@@ -147,6 +170,7 @@ static bool dmCheckDataDirVersion() {
 
 int32_t dmInit() {
   dInfo("start to init dnode env");
+  if (dmDiskInit() != 0) return -1;
   if (!dmCheckDataDirVersion()) return -1;
   if (!dmCheckDiskSpace()) return -1;
   if (dmCheckRepeatInit(dmInstance()) != 0) return -1;
@@ -177,6 +201,7 @@ void dmCleanup() {
   udfcClose();
   udfStopUdfd();
   taosStopCacheRefreshWorker();
+  dmDiskClose();
   dInfo("dnode env is cleaned up");
 
   taosCleanupCfg();
@@ -367,6 +392,7 @@ SMgmtInputOpt dmBuildMgmtInputOpt(SMgmtWrapper *pWrapper) {
   SMgmtInputOpt opt = {
       .path = pWrapper->path,
       .name = pWrapper->name,
+      .pTfs = pWrapper->pDnode->pTfs,
       .pData = &pWrapper->pDnode->data,
       .processCreateNodeFp = dmProcessCreateNodeReq,
       .processAlterNodeTypeFp = dmProcessAlterNodeTypeReq,

source/dnode/mgmt/node_util/inc/dmUtil.h

@@ -37,6 +37,7 @@
 #include "monitor.h"
 #include "qnode.h"
 #include "sync.h"
+#include "tfs.h"
 #include "wal.h"
 
 #include "libs/function/tudf.h"
@@ -111,6 +112,7 @@ typedef struct {
 typedef struct {
   const char         *path;
   const char         *name;
+  STfs               *pTfs;
   SDnodeData         *pData;
   SMsgCb              msgCb;
   ProcessCreateNodeFp processCreateNodeFp;

source/dnode/mgmt/node_util/src/dmEps.c

@@ -100,7 +100,7 @@ int32_t dmReadEps(SDnodeData *pData) {
     goto _OVER;
   }
 
-  if (taosStatFile(file, NULL, NULL) < 0) {
+  if (taosStatFile(file, NULL, NULL, NULL) < 0) {
     dInfo("dnode file:%s not exist", file);
     code = 0;
     goto _OVER;
@@ -350,7 +350,7 @@ void dmRotateMnodeEpSet(SDnodeData *pData) {
 }
 
 void dmGetMnodeEpSetForRedirect(SDnodeData *pData, SRpcMsg *pMsg, SEpSet *pEpSet) {
-  if(!pData->validMnodeEps) return;
+  if (!pData->validMnodeEps) return;
   dmGetMnodeEpSet(pData, pEpSet);
   dTrace("msg is redirected, handle:%p num:%d use:%d", pMsg->info.handle, pEpSet->numOfEps, pEpSet->inUse);
   for (int32_t i = 0; i < pEpSet->numOfEps; ++i) {
@@ -469,7 +469,7 @@ static int32_t dmReadDnodePairs(SDnodeData *pData) {
   char      file[PATH_MAX] = {0};
   snprintf(file, sizeof(file), "%s%sdnode%sep.json", tsDataDir, TD_DIRSEP, TD_DIRSEP);
 
-  if (taosStatFile(file, NULL, NULL) < 0) {
+  if (taosStatFile(file, NULL, NULL, NULL) < 0) {
     dDebug("dnode file:%s not exist", file);
     code = 0;
     goto _OVER;

source/dnode/mgmt/node_util/src/dmFile.c

@@ -38,7 +38,7 @@ int32_t dmReadFile(const char *path, const char *name, bool *pDeployed) {
   char      file[PATH_MAX] = {0};
   snprintf(file, sizeof(file), "%s%s%s.json", path, TD_DIRSEP, name);
 
-  if (taosStatFile(file, NULL, NULL) < 0) {
+  if (taosStatFile(file, NULL, NULL, NULL) < 0) {
     dInfo("file:%s not exist", file);
     code = 0;
     goto _OVER;

source/dnode/mnode/impl/src/mndDnode.c

@@ -41,7 +41,7 @@ static const char *offlineReason[] = {
     "timezone not match",
     "locale not match",
     "charset not match",
-    "ttl change on write not match"
+    "ttlChangeOnWrite not match",
     "unknown",
 };
 

source/dnode/mnode/impl/src/mndShow.c

@@ -58,8 +58,10 @@ static int32_t convertToRetrieveType(char *name, int32_t len) {
     type = TSDB_MGMT_TABLE_DNODE;
   } else if (strncasecmp(name, TSDB_INS_TABLE_MNODES, len) == 0) {
     type = TSDB_MGMT_TABLE_MNODE;
+/*
   } else if (strncasecmp(name, TSDB_INS_TABLE_MODULES, len) == 0) {
     type = TSDB_MGMT_TABLE_MODULE;
+*/
   } else if (strncasecmp(name, TSDB_INS_TABLE_QNODES, len) == 0) {
     type = TSDB_MGMT_TABLE_QNODE;
   } else if (strncasecmp(name, TSDB_INS_TABLE_SNODES, len) == 0) {

source/dnode/mnode/impl/src/mndSubscribe.c

@@ -489,7 +489,7 @@ static int32_t mndDoRebalance(SMnode *pMnode, const SMqRebInputObj *pInput, SMqR
             SMqVgEp *pVgEp = taosArrayGetP(pConsumerEpNew->vgs, i);
             if(pVgEp->vgId == d1->vgId){
               jump = true;
-              mInfo("pSub->offsetRows jump, because consumer id:%"PRIx64 " and vgId:%d not change", pConsumerEp->consumerId, pVgEp->vgId);
+              mInfo("pSub->offsetRows jump, because consumer id:0x%"PRIx64 " and vgId:%d not change", pConsumerEp->consumerId, pVgEp->vgId);
               break;
             }
           }

source/dnode/mnode/impl/src/mndUser.c

@@ -863,6 +863,7 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
+      mndReleaseDb(pMnode, pDb);
     } else {
       while (1) {
         SDbObj *pDb = NULL;
@@ -887,6 +888,7 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
+      mndReleaseDb(pMnode, pDb);
     } else {
       while (1) {
         SDbObj *pDb = NULL;
@@ -908,6 +910,7 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
         goto _OVER;
       }
       taosHashRemove(newUser.readDbs, alterReq.objname, len);
+      mndReleaseDb(pMnode, pDb);
     } else {
       taosHashClear(newUser.readDbs);
     }
@@ -922,6 +925,7 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
         goto _OVER;
       }
       taosHashRemove(newUser.writeDbs, alterReq.objname, len);
+      mndReleaseDb(pMnode, pDb);
     } else {
       taosHashClear(newUser.writeDbs);
     }

source/dnode/vnode/CMakeLists.txt

@@ -8,6 +8,7 @@ set(
   "src/vnd/vnodeCommit.c"
   "src/vnd/vnodeQuery.c"
   "src/vnd/vnodeModule.c"
+  "src/vnd/vnodeCos.c"
   "src/vnd/vnodeSvr.c"
   "src/vnd/vnodeSync.c"
   "src/vnd/vnodeSnapshot.c"
@@ -155,6 +156,45 @@ target_link_libraries(
     PUBLIC index
 )
 
+if(${TD_LINUX})
+set(CMAKE_FIND_LIBRARY_SUFFIXES ".a")
+find_library(APR_LIBRARY apr-1 PATHS /usr/local/apr/lib/)
+find_library(APR_UTIL_LIBRARY aprutil-1 PATHS /usr/local/apr/lib/)
+find_library(MINIXML_LIBRARY mxml)
+find_library(CURL_LIBRARY curl)
+target_link_libraries(
+    vnode
+
+    # s3
+    PUBLIC cos_c_sdk_static
+    PUBLIC ${APR_UTIL_LIBRARY}
+    PUBLIC ${APR_LIBRARY}
+    PUBLIC ${MINIXML_LIBRARY}
+    PUBLIC ${CURL_LIBRARY}
+)
+
+# s3
+FIND_PROGRAM(APR_CONFIG_BIN NAMES apr-config apr-1-config PATHS /usr/bin /usr/local/bin /usr/local/apr/bin/)
+IF (APR_CONFIG_BIN)
+  EXECUTE_PROCESS(
+    COMMAND ${APR_CONFIG_BIN} --includedir
+    OUTPUT_VARIABLE APR_INCLUDE_DIR
+    OUTPUT_STRIP_TRAILING_WHITESPACE
+    )
+ENDIF()
+include_directories (${APR_INCLUDE_DIR})
+target_include_directories(
+  vnode
+  PUBLIC "${TD_SOURCE_DIR}/contrib/cos-c-sdk-v5/cos_c_sdk"
+  PUBLIC "$ENV{HOME}/.cos-local.1/include"
+  )
+
+if(${BUILD_WITH_COS})
+  add_definitions(-DUSE_COS)
+endif(${BUILD_WITH_COS})
+
+endif(${TD_LINUX})
+
 IF (TD_GRANT)
   TARGET_LINK_LIBRARIES(vnode PUBLIC grant)
 ENDIF ()
@@ -169,8 +209,6 @@ if(${BUILD_WITH_ROCKSDB})
     add_definitions(-DUSE_ROCKSDB)
 endif(${BUILD_WITH_ROCKSDB})
 
-
-
 if(${BUILD_TEST})
     add_subdirectory(test)
 endif(${BUILD_TEST})

source/dnode/vnode/src/inc/vndCos.h

+/*
+ * 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/>.
+ */
+
+#ifndef _TD_VND_COS_H_
+#define _TD_VND_COS_H_
+
+#include "vnd.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+extern int8_t tsS3Enabled;
+
+int32_t s3Init();
+void    s3CleanUp();
+int32_t s3PutObjectFromFile(const char *file, const char *object);
+void    s3DeleteObjects(const char *object_name[], int nobject);
+bool    s3Exists(const char *object_name);
+bool    s3Get(const char *object_name, const char *path);
+void    s3EvictCache(const char *path, long object_size);
+long    s3Size(const char *object_name);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_VND_COS_H_*/

source/dnode/vnode/src/inc/vnodeInt.h

@@ -167,7 +167,7 @@ int metaAddIndexToSTable(SMeta* pMeta, int64_t version, SVCreateStbReq* pReq);
 int metaDropIndexFromSTable(SMeta* pMeta, int64_t version, SDropIndexReq* pReq);
 
 int64_t       metaGetTimeSeriesNum(SMeta* pMeta);
-SMCtbCursor*  metaOpenCtbCursor(SMeta* pMeta, tb_uid_t uid, int lock);
+SMCtbCursor*  metaOpenCtbCursor(void* pVnode, tb_uid_t uid, int lock);
 void          metaCloseCtbCursor(SMCtbCursor* pCtbCur, int lock);
 tb_uid_t      metaCtbCursorNext(SMCtbCursor* pCtbCur);
 SMStbCursor*  metaOpenStbCursor(SMeta* pMeta, tb_uid_t uid);

source/dnode/vnode/src/meta/metaQuery.c

@@ -408,17 +408,9 @@ _err:
   return NULL;
 }
 
-struct SMCtbCursor {
-  SMeta   *pMeta;
-  TBC     *pCur;
-  tb_uid_t suid;
-  void    *pKey;
-  void    *pVal;
-  int      kLen;
-  int      vLen;
-};
 
-SMCtbCursor *metaOpenCtbCursor(SMeta *pMeta, tb_uid_t uid, int lock) {
+SMCtbCursor *metaOpenCtbCursor(void* pVnode, tb_uid_t uid, int lock) {
+  SMeta* pMeta = ((SVnode*)pVnode)->pMeta;
   SMCtbCursor *pCtbCur = NULL;
   SCtbIdxKey   ctbIdxKey;
   int          ret = 0;
@@ -435,7 +427,7 @@ SMCtbCursor *metaOpenCtbCursor(SMeta *pMeta, tb_uid_t uid, int lock) {
     metaRLock(pMeta);
   }
 
-  ret = tdbTbcOpen(pMeta->pCtbIdx, &pCtbCur->pCur, NULL);
+  ret = tdbTbcOpen(pMeta->pCtbIdx, (TBC**)&pCtbCur->pCur, NULL);
   if (ret < 0) {
     metaULock(pMeta);
     taosMemoryFree(pCtbCur);
@@ -1373,7 +1365,7 @@ int32_t metaGetTableTagsByUids(void *pVnode, int64_t suid, SArray *uidList) {
 }
 
 int32_t metaGetTableTags(void *pVnode, uint64_t suid, SArray *pUidTagInfo) {
-  SMCtbCursor *pCur = metaOpenCtbCursor(((SVnode *)pVnode)->pMeta, suid, 1);
+  SMCtbCursor *pCur = metaOpenCtbCursor(pVnode, suid, 1);
 
   // If len > 0 means there already have uids, and we only want the
   // tags of the specified tables, of which uid in the uid list. Otherwise, all table tags are retrieved and kept
@@ -1434,35 +1426,36 @@ int32_t metaGetInfo(SMeta *pMeta, int64_t uid, SMetaInfo *pInfo, SMetaReader *pR
   int     nData = 0;
   int     lock = 0;
 
-  metaRLock(pMeta);
+  if (pReader && !(pReader->flags & META_READER_NOLOCK)) {
+    lock = 1;
+  }
+
+  if(!lock) metaRLock(pMeta);
 
   // search cache
   if (metaCacheGet(pMeta, uid, pInfo) == 0) {
-    metaULock(pMeta);
+    if(!lock) metaULock(pMeta);
     goto _exit;
   }
 
   // search TDB
   if (tdbTbGet(pMeta->pUidIdx, &uid, sizeof(uid), &pData, &nData) < 0) {
     // not found
-    metaULock(pMeta);
+    if(!lock) metaULock(pMeta);
     code = TSDB_CODE_NOT_FOUND;
     goto _exit;
   }
 
-  metaULock(pMeta);
+  if(!lock) metaULock(pMeta);
 
   pInfo->uid = uid;
   pInfo->suid = ((SUidIdxVal *)pData)->suid;
   pInfo->version = ((SUidIdxVal *)pData)->version;
   pInfo->skmVer = ((SUidIdxVal *)pData)->skmVer;
 
-  if (pReader != NULL) {
-    lock = !(pReader->flags & META_READER_NOLOCK);
-    if (lock) {
-      metaULock(pReader->pMeta);
-      // metaReaderReleaseLock(pReader);
-    }
+  if (lock) {
+    metaULock(pReader->pMeta);
+    // metaReaderReleaseLock(pReader);
   }
   // upsert the cache
   metaWLock(pMeta);

source/dnode/vnode/src/sma/smaRollup.c

@@ -905,6 +905,7 @@ int32_t tdProcessRSmaSubmit(SSma *pSma, int64_t version, void *pReq, void *pMsg,
         tb_uid_t *pTbSuid = (tb_uid_t *)taosHashGetKey(pIter, NULL);
         if (tdExecuteRSmaAsync(pSma, version, pMsg, len, inputType, *pTbSuid) < 0) {
           smaError("vgId:%d, failed to process rsma submit exec 2 since: %s", SMA_VID(pSma), terrstr());
+          taosHashCancelIterate(uidStore.uidHash, pIter);
           goto _err;
         }
       }

source/dnode/vnode/src/tq/tq.c

@@ -644,7 +644,8 @@ int32_t tqProcessVgCommittedInfoReq(STQ* pTq, SRpcMsg* pMsg) {
   SDecoder decoder;
   tDecoderInit(&decoder, (uint8_t*)data, len);
   if (tDecodeMqVgOffset(&decoder, &vgOffset) < 0) {
-    return TSDB_CODE_OUT_OF_MEMORY;
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return terrno;
   }
 
   tDecoderClear(&decoder);
@@ -652,19 +653,22 @@ int32_t tqProcessVgCommittedInfoReq(STQ* pTq, SRpcMsg* pMsg) {
   STqOffset* pOffset = &vgOffset.offset;
   STqOffset* pSavedOffset = tqOffsetRead(pTq->pOffsetStore, pOffset->subKey);
   if (pSavedOffset == NULL) {
-    return TSDB_CODE_TMQ_NO_COMMITTED;
+    terrno = TSDB_CODE_TMQ_NO_COMMITTED;
+    return terrno;
   }
   vgOffset.offset = *pSavedOffset;
 
   int32_t code = 0;
   tEncodeSize(tEncodeMqVgOffset, &vgOffset, len, code);
   if (code < 0) {
-    return TSDB_CODE_INVALID_PARA;
+    terrno = TSDB_CODE_INVALID_PARA;
+    return terrno;
   }
 
   void* buf = rpcMallocCont(len);
   if (buf == NULL) {
-    return TSDB_CODE_OUT_OF_MEMORY;
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return terrno;
   }
   SEncoder encoder;
   tEncoderInit(&encoder, buf, len);

source/dnode/vnode/src/tq/tqOffsetSnapshot.c

@@ -60,7 +60,7 @@ int32_t tqOffsetSnapRead(STqOffsetReader* pReader, uint8_t** ppData) {
   }
 
   int64_t sz = 0;
-  if (taosStatFile(fname, &sz, NULL) < 0) {
+  if (taosStatFile(fname, &sz, NULL, NULL) < 0) {
     taosCloseFile(&pFile);
     taosMemoryFree(fname);
     return -1;

source/dnode/vnode/src/tq/tqRead.c

@@ -1092,6 +1092,7 @@ int32_t tqUpdateTbUidList(STQ* pTq, const SArray* tbUidList, bool isAdd) {
         if(ret != TDB_CODE_SUCCESS) {
           tqError("qGetTableList in tqUpdateTbUidList error:%d handle %s consumer:0x%" PRIx64, ret, pTqHandle->subKey, pTqHandle->consumerId);
           taosArrayDestroy(list);
+          taosHashCancelIterate(pTq->pHandle, pIter);
           return ret;
         }
         tqReaderSetTbUidList(pTqHandle->execHandle.pTqReader, list, NULL);

source/dnode/vnode/src/tsdb/tsdbFS.c

@@ -176,7 +176,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) {
   // SDelFile
   if (pTsdb->fs.pDelFile) {
     tsdbDelFileName(pTsdb, pTsdb->fs.pDelFile, fname);
-    if (taosStatFile(fname, &size, NULL)) {
+    if (taosStatFile(fname, &size, NULL, NULL)) {
       code = TAOS_SYSTEM_ERROR(errno);
       TSDB_CHECK_CODE(code, lino, _exit);
     }
@@ -195,7 +195,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) {
 
     // head =========
     tsdbHeadFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pHeadF, fname);
-    if (taosStatFile(fname, &size, NULL)) {
+    if (taosStatFile(fname, &size, NULL, NULL)) {
       code = TAOS_SYSTEM_ERROR(errno);
       TSDB_CHECK_CODE(code, lino, _exit);
     }
@@ -206,7 +206,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) {
 
     // data =========
     tsdbDataFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pDataF, fname);
-    if (taosStatFile(fname, &size, NULL)) {
+    if (taosStatFile(fname, &size, NULL, NULL)) {
       code = TAOS_SYSTEM_ERROR(errno);
       TSDB_CHECK_CODE(code, lino, _exit);
     }
@@ -221,7 +221,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) {
 
     // sma =============
     tsdbSmaFileName(pTsdb, pSet->diskId, pSet->fid, pSet->pSmaF, fname);
-    if (taosStatFile(fname, &size, NULL)) {
+    if (taosStatFile(fname, &size, NULL, NULL)) {
       code = TAOS_SYSTEM_ERROR(errno);
       TSDB_CHECK_CODE(code, lino, _exit);
     }
@@ -237,7 +237,7 @@ static int32_t tsdbScanAndTryFixFS(STsdb *pTsdb) {
     // stt ===========
     for (int32_t iStt = 0; iStt < pSet->nSttF; iStt++) {
       tsdbSttFileName(pTsdb, pSet->diskId, pSet->fid, pSet->aSttF[iStt], fname);
-      if (taosStatFile(fname, &size, NULL)) {
+      if (taosStatFile(fname, &size, NULL, NULL)) {
         code = TAOS_SYSTEM_ERROR(errno);
         TSDB_CHECK_CODE(code, lino, _exit);
       }

source/dnode/vnode/src/tsdb/tsdbReaderWriter.c

@@ -14,6 +14,7 @@
  */
 
 #include "tsdb.h"
+#include "vndCos.h"
 
 // =============== PAGE-WISE FILE ===============
 int32_t tsdbOpenFile(const char *path, int32_t szPage, int32_t flag, STsdbFD **ppFD) {
@@ -34,9 +35,24 @@ int32_t tsdbOpenFile(const char *path, int32_t szPage, int32_t flag, STsdbFD **p
   pFD->flag = flag;
   pFD->pFD = taosOpenFile(path, flag);
   if (pFD->pFD == NULL) {
-    code = TAOS_SYSTEM_ERROR(errno);
-    taosMemoryFree(pFD);
-    goto _exit;
+    const char *object_name = taosDirEntryBaseName((char *)path);
+    long        s3_size = s3Size(object_name);
+    if (!strncmp(path + strlen(path) - 5, ".data", 5) && s3_size > 0) {
+      s3EvictCache(path, s3_size);
+      s3Get(object_name, path);
+
+      pFD->pFD = taosOpenFile(path, flag);
+
+      if (pFD->pFD == NULL) {
+        code = TAOS_SYSTEM_ERROR(errno);
+        taosMemoryFree(pFD);
+        goto _exit;
+      }
+    } else {
+      code = TAOS_SYSTEM_ERROR(errno);
+      taosMemoryFree(pFD);
+      goto _exit;
+    }
   }
   pFD->szPage = szPage;
   pFD->pgno = 0;
@@ -50,7 +66,7 @@ int32_t tsdbOpenFile(const char *path, int32_t szPage, int32_t flag, STsdbFD **p
 
   // not check file size when reading data files.
   if (flag != TD_FILE_READ) {
-    if (taosStatFile(path, &pFD->szFile, NULL) < 0) {
+    if (taosStatFile(path, &pFD->szFile, NULL, NULL) < 0) {
       code = TAOS_SYSTEM_ERROR(errno);
       taosMemoryFree(pFD->pBuf);
       taosCloseFile(&pFD->pFD);

source/dnode/vnode/src/tsdb/tsdbRetention.c

@@ -15,6 +15,7 @@
 
 #include "tsdb.h"
 #include "tsdbFS2.h"
+#include "vndCos.h"
 
 typedef struct {
   STsdb  *tsdb;
@@ -41,6 +42,28 @@ static int32_t tsdbDoRemoveFileObject(SRTNer *rtner, const STFileObj *fobj) {
   return TARRAY2_APPEND(rtner->fopArr, op);
 }
 
+static int32_t tsdbRemoveFileObjectS3(SRTNer *rtner, const STFileObj *fobj) {
+  int32_t code = 0, lino = 0;
+
+  STFileOp op = {
+      .optype = TSDB_FOP_REMOVE,
+      .fid = fobj->f->fid,
+      .of = fobj->f[0],
+  };
+
+  code = TARRAY2_APPEND(rtner->fopArr, op);
+  TSDB_CHECK_CODE(code, lino, _exit);
+
+  const char *object_name = taosDirEntryBaseName((char *)fobj->fname);
+  s3DeleteObjects(&object_name, 1);
+
+_exit:
+  if (code) {
+    TSDB_ERROR_LOG(TD_VID(rtner->tsdb->pVnode), lino, code);
+  }
+  return code;
+}
+
 static int32_t tsdbDoCopyFile(SRTNer *rtner, const STFileObj *from, const STFile *to) {
   int32_t code = 0;
   int32_t lino = 0;
@@ -76,6 +99,34 @@ _exit:
   return code;
 }
 
+static int32_t tsdbCopyFileS3(SRTNer *rtner, const STFileObj *from, const STFile *to) {
+  int32_t code = 0;
+  int32_t lino = 0;
+
+  char      fname[TSDB_FILENAME_LEN];
+  TdFilePtr fdFrom = NULL;
+  TdFilePtr fdTo = NULL;
+
+  tsdbTFileName(rtner->tsdb, to, fname);
+
+  fdFrom = taosOpenFile(from->fname, TD_FILE_READ);
+  if (fdFrom == NULL) code = terrno;
+  TSDB_CHECK_CODE(code, lino, _exit);
+
+  char *object_name = taosDirEntryBaseName(fname);
+  code = s3PutObjectFromFile(from->fname, object_name);
+  TSDB_CHECK_CODE(code, lino, _exit);
+
+  taosCloseFile(&fdFrom);
+
+_exit:
+  if (code) {
+    TSDB_ERROR_LOG(TD_VID(rtner->tsdb->pVnode), lino, code);
+    taosCloseFile(&fdFrom);
+  }
+  return code;
+}
+
 static int32_t tsdbDoMigrateFileObj(SRTNer *rtner, const STFileObj *fobj, const SDiskID *did) {
   int32_t  code = 0;
   int32_t  lino = 0;
@@ -123,6 +174,53 @@ _exit:
   return code;
 }
 
+static int32_t tsdbMigrateDataFileS3(SRTNer *rtner, const STFileObj *fobj, const SDiskID *did) {
+  int32_t  code = 0;
+  int32_t  lino = 0;
+  STFileOp op = {0};
+
+  // remove old
+  op = (STFileOp){
+      .optype = TSDB_FOP_REMOVE,
+      .fid = fobj->f->fid,
+      .of = fobj->f[0],
+  };
+
+  code = TARRAY2_APPEND(rtner->fopArr, op);
+  TSDB_CHECK_CODE(code, lino, _exit);
+
+  // create new
+  op = (STFileOp){
+      .optype = TSDB_FOP_CREATE,
+      .fid = fobj->f->fid,
+      .nf =
+          {
+              .type = fobj->f->type,
+              .did = did[0],
+              .fid = fobj->f->fid,
+              .cid = fobj->f->cid,
+              .size = fobj->f->size,
+              .stt[0] =
+                  {
+                      .level = fobj->f->stt[0].level,
+                  },
+          },
+  };
+
+  code = TARRAY2_APPEND(rtner->fopArr, op);
+  TSDB_CHECK_CODE(code, lino, _exit);
+
+  // do copy the file
+  code = tsdbCopyFileS3(rtner, fobj, &op.nf);
+  TSDB_CHECK_CODE(code, lino, _exit);
+
+_exit:
+  if (code) {
+    TSDB_ERROR_LOG(TD_VID(rtner->tsdb->pVnode), lino, code);
+  }
+  return code;
+}
+
 typedef struct {
   STsdb  *tsdb;
   int32_t sync;
@@ -201,8 +299,14 @@ static int32_t tsdbDoRetention2(void *arg) {
       for (int32_t ftype = 0; (ftype < TSDB_FTYPE_MAX) && (fobj = rtner->ctx->fset->farr[ftype], 1); ++ftype) {
         if (fobj == NULL) continue;
 
-        code = tsdbDoRemoveFileObject(rtner, fobj);
-        TSDB_CHECK_CODE(code, lino, _exit);
+        int32_t nlevel = tfsGetLevel(rtner->tsdb->pVnode->pTfs);
+        if (tsS3Enabled && nlevel > 1 && TSDB_FTYPE_DATA == ftype && fobj->f->did.level == nlevel - 1) {
+          code = tsdbRemoveFileObjectS3(rtner, fobj);
+          TSDB_CHECK_CODE(code, lino, _exit);
+        } else {
+          code = tsdbDoRemoveFileObject(rtner, fobj);
+          TSDB_CHECK_CODE(code, lino, _exit);
+        }
       }
 
       SSttLvl *lvl;
@@ -228,8 +332,15 @@ static int32_t tsdbDoRetention2(void *arg) {
         if (fobj == NULL) continue;
 
         if (fobj->f->did.level == did.level) continue;
-        code = tsdbDoMigrateFileObj(rtner, fobj, &did);
-        TSDB_CHECK_CODE(code, lino, _exit);
+
+        int32_t nlevel = tfsGetLevel(rtner->tsdb->pVnode->pTfs);
+        if (tsS3Enabled && nlevel > 1 && TSDB_FTYPE_DATA == ftype && did.level == nlevel - 1) {
+          code = tsdbMigrateDataFileS3(rtner, fobj, &did);
+          TSDB_CHECK_CODE(code, lino, _exit);
+        } else {
+          code = tsdbDoMigrateFileObj(rtner, fobj, &did);
+          TSDB_CHECK_CODE(code, lino, _exit);
+        }
       }
 
       // stt
@@ -281,4 +392,4 @@ int32_t tsdbRetention(STsdb *tsdb, int64_t now, int32_t sync) {
     tsdbFreeRtnArg(arg);
   }
   return code;
-}
+}
\ No newline at end of file

source/dnode/vnode/src/tsdb/tsdbWrite.c

@@ -76,7 +76,7 @@ int tsdbScanAndConvertSubmitMsg(STsdb *pTsdb, SSubmitReq2 *pMsg) {
   int32_t       code = 0;
   STsdbKeepCfg *pCfg = &pTsdb->keepCfg;
   TSKEY         now = taosGetTimestamp(pCfg->precision);
-  TSKEY         minKey = now - tsTickPerMin[pCfg->precision] * pCfg->keep2;
+  TSKEY         minKey = now - tsTickPerMin[pCfg->precision] * pCfg->keep1;
   TSKEY         maxKey = tsMaxKeyByPrecision[pCfg->precision];
   int32_t       size = taosArrayGetSize(pMsg->aSubmitTbData);
 
@@ -107,4 +107,4 @@ int tsdbScanAndConvertSubmitMsg(STsdb *pTsdb, SSubmitReq2 *pMsg) {
 
 _exit:
   return code;
-}
+}
\ No newline at end of file

source/dnode/vnode/src/vnd/vnodeInitApi.c

@@ -96,6 +96,10 @@ void initMetadataAPI(SStoreMeta* pMeta) {
 
   pMeta->metaGetCachedTbGroup = metaGetCachedTbGroup;
   pMeta->metaPutTbGroupToCache = metaPutTbGroupToCache;
+
+  pMeta->openCtbCursor = metaOpenCtbCursor;
+  pMeta->closeCtbCursor = metaCloseCtbCursor;
+  pMeta->ctbCursorNext = metaCtbCursorNext;
 }
 
 void initTqAPI(SStoreTqReader* pTq) {

source/dnode/vnode/src/vnd/vnodeModule.c

@@ -14,6 +14,7 @@
  */
 
 #include "vnd.h"
+#include "vndCos.h"
 
 typedef struct SVnodeTask SVnodeTask;
 struct SVnodeTask {
@@ -81,6 +82,9 @@ int vnodeInit(int nthreads) {
   if (tqInit() < 0) {
     return -1;
   }
+  if (s3Init() < 0) {
+    return -1;
+  }
 
   return 0;
 }
@@ -112,6 +116,7 @@ void vnodeCleanup() {
   walCleanUp();
   tqCleanUp();
   smaCleanUp();
+  s3CleanUp();
 }
 
 int vnodeScheduleTaskEx(int tpid, int (*execute)(void*), void* arg) {

source/dnode/vnode/src/vnd/vnodeQuery.c

@@ -440,7 +440,7 @@ int32_t vnodeGetTableList(void* pVnode, int8_t type, SArray* pList) {
 }
 
 int32_t vnodeGetAllTableList(SVnode *pVnode, uint64_t uid, SArray *list) {
-  SMCtbCursor *pCur = metaOpenCtbCursor(pVnode->pMeta, uid, 1);
+  SMCtbCursor *pCur = metaOpenCtbCursor(pVnode, uid, 1);
 
   while (1) {
     tb_uid_t id = metaCtbCursorNext(pCur);
@@ -462,7 +462,7 @@ int32_t vnodeGetCtbIdListByFilter(SVnode *pVnode, int64_t suid, SArray *list, bo
 
 int32_t vnodeGetCtbIdList(void *pVnode, int64_t suid, SArray *list) {
   SVnode      *pVnodeObj = pVnode;
-  SMCtbCursor *pCur = metaOpenCtbCursor(pVnodeObj->pMeta, suid, 1);
+  SMCtbCursor *pCur = metaOpenCtbCursor(pVnodeObj, suid, 1);
 
   while (1) {
     tb_uid_t id = metaCtbCursorNext(pCur);
@@ -521,7 +521,7 @@ int32_t vnodeGetStbIdListByFilter(SVnode *pVnode, int64_t suid, SArray *list, bo
 }
 
 int32_t vnodeGetCtbNum(SVnode *pVnode, int64_t suid, int64_t *num) {
-  SMCtbCursor *pCur = metaOpenCtbCursor(pVnode->pMeta, suid, 0);
+  SMCtbCursor *pCur = metaOpenCtbCursor(pVnode, suid, 0);
   if (!pCur) {
     return TSDB_CODE_FAILED;
   }

source/libs/command/src/explain.c

@@ -291,17 +291,17 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
   switch (pNode->type) {
     case QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN: {
       STagScanPhysiNode *pTagScanNode = (STagScanPhysiNode *)pNode;
-      EXPLAIN_ROW_NEW(level, EXPLAIN_TAG_SCAN_FORMAT, pTagScanNode->tableName.tname);
+      EXPLAIN_ROW_NEW(level, EXPLAIN_TAG_SCAN_FORMAT, pTagScanNode->scan.tableName.tname);
       EXPLAIN_ROW_APPEND(EXPLAIN_LEFT_PARENTHESIS_FORMAT);
       if (pResNode->pExecInfo) {
         QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
         EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       }
-      if (pTagScanNode->pScanPseudoCols) {
-        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pTagScanNode->pScanPseudoCols->length);
+      if (pTagScanNode->scan.pScanPseudoCols) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pTagScanNode->scan.pScanPseudoCols->length);
         EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       }
-      EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pTagScanNode->node.pOutputDataBlockDesc->totalRowSize);
+      EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pTagScanNode->scan.node.pOutputDataBlockDesc->totalRowSize);
       EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
       EXPLAIN_ROW_END();
       QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
@@ -309,11 +309,11 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
       if (verbose) {
         EXPLAIN_ROW_NEW(level + 1, EXPLAIN_OUTPUT_FORMAT);
         EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT,
-                           nodesGetOutputNumFromSlotList(pTagScanNode->node.pOutputDataBlockDesc->pSlots));
+                           nodesGetOutputNumFromSlotList(pTagScanNode->scan.node.pOutputDataBlockDesc->pSlots));
         EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
-        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pTagScanNode->node.pOutputDataBlockDesc->outputRowSize);
-        EXPLAIN_ROW_APPEND_LIMIT(pTagScanNode->node.pLimit);
-        EXPLAIN_ROW_APPEND_SLIMIT(pTagScanNode->node.pSlimit);
+        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pTagScanNode->scan.node.pOutputDataBlockDesc->outputRowSize);
+        EXPLAIN_ROW_APPEND_LIMIT(pTagScanNode->scan.node.pLimit);
+        EXPLAIN_ROW_APPEND_SLIMIT(pTagScanNode->scan.node.pSlimit);
         EXPLAIN_ROW_END();
         QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
 

source/libs/executor/inc/executil.h

@@ -190,4 +190,6 @@ void printDataBlock(SSDataBlock* pBlock, const char* flag);
 void getNextTimeWindow(const SInterval* pInterval, STimeWindow* tw, int32_t order);
 void getInitialStartTimeWindow(SInterval* pInterval, TSKEY ts, STimeWindow* w, bool ascQuery);
 
+SSDataBlock* createTagValBlockForFilter(SArray* pColList, int32_t numOfTables, SArray* pUidTagList, void* pVnode,
+                                        SStorageAPI* pStorageAPI);
 #endif  // TDENGINE_EXECUTIL_H

source/libs/executor/inc/executorInt.h

@@ -25,6 +25,7 @@ extern "C" {
 #include "tsort.h"
 #include "ttszip.h"
 #include "tvariant.h"
+#include "theap.h"
 
 #include "dataSinkMgt.h"
 #include "executil.h"
@@ -250,6 +251,12 @@ typedef struct STableMergeScanInfo {
   SSortExecInfo   sortExecInfo;
 } STableMergeScanInfo;
 
+typedef struct STagScanFilterContext {
+  SHashObj* colHash;
+  int32_t   index;
+  SArray*   cInfoList;
+} STagScanFilterContext;
+
 typedef struct STagScanInfo {
   SColumnInfo*    pCols;
   SSDataBlock*    pRes;
@@ -258,6 +265,14 @@ typedef struct STagScanInfo {
   SLimitNode*     pSlimit;
   SReadHandle     readHandle;
   STableListInfo* pTableListInfo;
+  uint64_t        suid;
+  void*           pCtbCursor;
+  SNode*          pTagCond;
+  SNode*          pTagIndexCond;
+  STagScanFilterContext filterCtx;
+  SArray*         aUidTags; // SArray<STUidTagInfo>
+  SArray*         aFilterIdxs; // SArray<int32_t>
+  SStorageAPI*    pStorageAPI;
 } STagScanInfo;
 
 typedef enum EStreamScanMode {
@@ -417,6 +432,14 @@ typedef struct SIntervalAggOperatorInfo {
   EOPTR_EXEC_MODEL   execModel;          // operator execution model [batch model|stream model]
   STimeWindowAggSupp twAggSup;
   SArray*            pPrevValues;  //  SArray<SGroupKeys> used to keep the previous not null value for interpolation.
+  // for limit optimization
+  bool          limited;
+  int64_t       limit;
+  bool          slimited;
+  int64_t       slimit;
+  uint64_t      curGroupId; // initialize to UINT64_MAX
+  uint64_t      handledGroupNum;
+  BoundedQueue* pBQ;
 } SIntervalAggOperatorInfo;
 
 typedef struct SMergeAlignedIntervalAggOperatorInfo {

source/libs/executor/inc/operator.h

@@ -81,7 +81,7 @@ SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode,
 
 SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanNode, SReadHandle* readHandle, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
 
-SOperatorInfo* createTagScanOperatorInfo(SReadHandle* pReadHandle, STagScanPhysiNode* pPhyNode, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
+SOperatorInfo* createTagScanOperatorInfo(SReadHandle* pReadHandle, STagScanPhysiNode* pPhyNode, STableListInfo* pTableListInfo, SNode* pTagCond, SNode*pTagIndexCond, SExecTaskInfo* pTaskInfo);
 
 SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSystemTableScanPhysiNode* pScanPhyNode, const char* pUser, SExecTaskInfo* pTaskInfo);
 

source/libs/executor/inc/querytask.h

@@ -99,7 +99,7 @@ struct SExecTaskInfo {
 void           buildTaskId(uint64_t taskId, uint64_t queryId, char* dst);
 SExecTaskInfo* doCreateTask(uint64_t queryId, uint64_t taskId, int32_t vgId, EOPTR_EXEC_MODEL model, SStorageAPI* pAPI);
 void           doDestroyTask(SExecTaskInfo* pTaskInfo);
-bool           isTaskKilled(SExecTaskInfo* pTaskInfo);
+bool           isTaskKilled(void* pTaskInfo);
 void           setTaskKilled(SExecTaskInfo* pTaskInfo, int32_t rspCode);
 void           setTaskStatus(SExecTaskInfo* pTaskInfo, int8_t status);
 int32_t        createExecTaskInfo(SSubplan* pPlan, SExecTaskInfo** pTaskInfo, SReadHandle* pHandle, uint64_t taskId,

source/libs/executor/inc/tsort.h

@@ -191,6 +191,9 @@ int32_t getProperSortPageSize(size_t rowSize, uint32_t numOfCols);
 bool tsortIsClosed(SSortHandle* pHandle);
 void tsortSetClosed(SSortHandle* pHandle);
 
+void tsortSetSingleTableMerge(SSortHandle* pHandle);
+void tsortSetAbortCheckFn(SSortHandle* pHandle, bool (*checkFn)(void* param), void* param);
+
 #ifdef __cplusplus
 }
 #endif

source/libs/executor/src/executil.c

@@ -47,8 +47,6 @@ static int32_t optimizeTbnameInCondImpl(void* metaHandle, SArray* list, SNode* p
 
 static int32_t      getTableList(void* pVnode, SScanPhysiNode* pScanNode, SNode* pTagCond, SNode* pTagIndexCond,
                                  STableListInfo* pListInfo, uint8_t* digest, const char* idstr, SStorageAPI* pStorageAPI);
-static SSDataBlock* createTagValBlockForFilter(SArray* pColList, int32_t numOfTables, SArray* pUidTagList, void* pVnode,
-                                               SStorageAPI* pStorageAPI);
 
 static int64_t getLimit(const SNode* pLimit) { return NULL == pLimit ? -1 : ((SLimitNode*)pLimit)->limit; }
 static int64_t getOffset(const SNode* pLimit) { return NULL == pLimit ? -1 : ((SLimitNode*)pLimit)->offset; }
@@ -846,7 +844,7 @@ static int32_t optimizeTbnameInCondImpl(void* pVnode, SArray* pExistedUidList, S
   return -1;
 }
 
-static SSDataBlock* createTagValBlockForFilter(SArray* pColList, int32_t numOfTables, SArray* pUidTagList, void* pVnode,
+SSDataBlock* createTagValBlockForFilter(SArray* pColList, int32_t numOfTables, SArray* pUidTagList, void* pVnode,
                                                SStorageAPI* pStorageAPI) {
   SSDataBlock* pResBlock = createDataBlock();
   if (pResBlock == NULL) {
@@ -2118,8 +2116,9 @@ int32_t buildGroupIdMapForAllTables(STableListInfo* pTableListInfo, SReadHandle*
     if (code != TSDB_CODE_SUCCESS) {
       return code;
     }
+    if (pScanNode->groupOrderScan) pTableListInfo->numOfOuputGroups = taosArrayGetSize(pTableListInfo->pTableList);
 
-    if (groupSort) {
+    if (groupSort || pScanNode->groupOrderScan) {
       code = sortTableGroup(pTableListInfo);
     }
   }