<feature>:merge from 3.0

.clang-format

@@ -5,6 +5,7 @@ AccessModifierOffset: -1
 AlignAfterOpenBracket: Align
 AlignConsecutiveAssignments: false
 AlignConsecutiveDeclarations: true
+AlignConsecutiveMacros: true
 AlignEscapedNewlinesLeft: true
 AlignOperands:   true
 AlignTrailingComments: true

cmake/cmake.define

@@ -36,11 +36,13 @@ IF (TD_WINDOWS)
     ENDIF ()
 
 ELSE ()
-    SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror -fPIC -gdwarf-2 -g3")
+    SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror -Werror=return-type -fPIC -gdwarf-2 -g3")
-    SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -fPIC -gdwarf-2 -g3")
+    SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Werror=return-type -fPIC -gdwarf-2 -g3")
+    #SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fno-sanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=null -fno-sanitize=alignment -static-libasan -g3")
+    #SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fno-sanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=null -fno-sanitize=alignment -static-libasan -g3")
 
 MESSAGE("System processor ID: ${CMAKE_SYSTEM_PROCESSOR}")
-IF (${CMAKE_SYSTEM_PROCESSOR} MATCHES "arm64")
+IF (${CMAKE_SYSTEM_PROCESSOR} MATCHES "arm64" OR ${CMAKE_SYSTEM_PROCESSOR} MATCHES "aarch64")
     ADD_DEFINITIONS("-D_TD_ARM_")
 ELSE ()
     ADD_DEFINITIONS("-msse4.2 -mfma")

cmake/cmake.options

@@ -55,17 +55,6 @@ option(
     OFF
 )
 
-IF(${TD_WINDOWS})
-    MESSAGE("Not build BDB on Windows")
-ELSE ()
-    option(
-            BUILD_WITH_BDB
-            "If build with BerkleyDB"
-            ON
-          )
-
-ENDIF ()
-
 option(
     BUILD_WITH_LUCENE 
     "If build with lucene" 

cmake/cmake.platform

@@ -42,7 +42,7 @@ IF (${CMAKE_SYSTEM_NAME} MATCHES "Linux" OR ${CMAKE_SYSTEM_NAME} MATCHES "Darwin
 
         SET(TD_DARWIN TRUE)
         SET(OSTYPE "macOS")
-        ADD_DEFINITIONS("-DDARWIN -Wno-tautological-pointer-compare -Wno-return-type")
+        ADD_DEFINITIONS("-DDARWIN -Wno-tautological-pointer-compare")
 
         MESSAGE("Current system processor is ${CMAKE_SYSTEM_PROCESSOR}.")
         IF (${CMAKE_SYSTEM_PROCESSOR} MATCHES "arm64" OR ${CMAKE_SYSTEM_PROCESSOR} MATCHES "x86_64")

contrib/CMakeLists.txt

@@ -63,9 +63,9 @@ if(${BUILD_WITH_UV})
 endif(${BUILD_WITH_UV})
 
 # bdb
-if(${BUILD_WITH_BDB})
+#if(${BUILD_WITH_BDB})
-    cat("${CMAKE_SUPPORT_DIR}/bdb_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    #cat("${CMAKE_SUPPORT_DIR}/bdb_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
-endif(${BUILD_WITH_BDB})
+#endif(${BUILD_WITH_BDB})
 
 # sqlite
 if(${BUILD_WITH_SQLITE})

contrib/test/CMakeLists.txt

@@ -7,9 +7,9 @@ if(${BUILD_WITH_LUCENE})
     add_subdirectory(lucene)
 endif(${BUILD_WITH_LUCENE})
 
-if(${BUILD_WITH_BDB})
+#if(${BUILD_WITH_BDB})
-    add_subdirectory(bdb)
+    #add_subdirectory(bdb)
-endif(${BUILD_WITH_BDB})
+#endif(${BUILD_WITH_BDB})
 
 if(${BUILD_WITH_SQLITE})
     add_subdirectory(sqlite)

example/src/tmq.c

@@ -28,7 +28,7 @@ int32_t init_env() {
     return -1;
   }
 
-  TAOS_RES* pRes = taos_query(pConn, "create database if not exists abc1 vgroups 1");
+  TAOS_RES* pRes = taos_query(pConn, "create database if not exists abc1 vgroups 2");
   if (taos_errno(pRes) != 0) {
     printf("error in create db, reason:%s\n", taos_errstr(pRes));
     return -1;
@@ -42,25 +42,33 @@ int32_t init_env() {
   }
   taos_free_result(pRes);
 
-  pRes = taos_query(pConn, "create stable if not exists st1 (ts timestamp, k int) tags(a int)");
+  pRes =
+      taos_query(pConn, "create stable if not exists st1 (ts timestamp, c1 int, c2 float, c3 binary(10)) tags(t1 int)");
   if (taos_errno(pRes) != 0) {
     printf("failed to create super table st1, reason:%s\n", taos_errstr(pRes));
     return -1;
   }
   taos_free_result(pRes);
 
-  pRes = taos_query(pConn, "create table if not exists tu1 using st1 tags(1)");
+  pRes = taos_query(pConn, "create table if not exists ct0 using st1 tags(1000)");
   if (taos_errno(pRes) != 0) {
     printf("failed to create child table tu1, reason:%s\n", taos_errstr(pRes));
     return -1;
   }
   taos_free_result(pRes);
 
-  pRes = taos_query(pConn, "create table if not exists tu2 using st1 tags(2)");
+  pRes = taos_query(pConn, "create table if not exists ct1 using st1 tags(2000)");
   if (taos_errno(pRes) != 0) {
     printf("failed to create child table tu2, reason:%s\n", taos_errstr(pRes));
     return -1;
   }
+
+  pRes = taos_query(pConn, "create table if not exists ct3 using st1 tags(3000)");
+  if (taos_errno(pRes) != 0) {
+    printf("failed to create child table tu3, reason:%s\n", taos_errstr(pRes));
+    return -1;
+  }
+
   taos_free_result(pRes);
   return 0;
 }
@@ -82,12 +90,40 @@ int32_t create_topic() {
 
   /*const char* sql = "select * from tu1";*/
   /*pRes = tmq_create_topic(pConn, "test_stb_topic_1", sql, strlen(sql));*/
-  pRes = taos_query(pConn, "create topic test_stb_topic_1 as select * from tu1");
+  pRes = taos_query(pConn, "create topic topic_ctb_column as select ts, c1 from ct1");
   if (taos_errno(pRes) != 0) {
-    printf("failed to create topic test_stb_topic_1, reason:%s\n", taos_errstr(pRes));
+    printf("failed to create topic topic_ctb_column, reason:%s\n", taos_errstr(pRes));
     return -1;
   }
   taos_free_result(pRes);
+
+#if 0
+  pRes = taos_query(pConn, "insert into tu1 values(now, 1, 1.0, 'bi1')");
+  if (taos_errno(pRes) != 0) {
+    printf("failed to insert, reason:%s\n", taos_errstr(pRes));
+    return -1;
+  }
+  taos_free_result(pRes);
+  pRes = taos_query(pConn, "insert into tu1 values(now+1d, 1, 1.0, 'bi1')");
+  if (taos_errno(pRes) != 0) {
+    printf("failed to insert, reason:%s\n", taos_errstr(pRes));
+    return -1;
+  }
+  taos_free_result(pRes);
+  pRes = taos_query(pConn, "insert into tu2 values(now, 2, 2.0, 'bi2')");
+  if (taos_errno(pRes) != 0) {
+    printf("failed to insert, reason:%s\n", taos_errstr(pRes));
+    return -1;
+  }
+  taos_free_result(pRes);
+  pRes = taos_query(pConn, "insert into tu2 values(now+1d, 2, 2.0, 'bi2')");
+  if (taos_errno(pRes) != 0) {
+    printf("failed to insert, reason:%s\n", taos_errstr(pRes));
+    return -1;
+  }
+  taos_free_result(pRes);
+#endif
+
   taos_close(pConn);
   return 0;
 }
@@ -115,7 +151,7 @@ tmq_t* build_consumer() {
 
 tmq_list_t* build_topic_list() {
   tmq_list_t* topic_list = tmq_list_new();
-  tmq_list_append(topic_list, "test_stb_topic_1");
+  tmq_list_append(topic_list, "topic_ctb_column");
   return topic_list;
 }
 
@@ -215,8 +251,8 @@ int main(int argc, char* argv[]) {
   if (argc > 1) {
     printf("env init\n");
     code = init_env();
-  }
     create_topic();
+  }
   tmq_t*      tmq = build_consumer();
   tmq_list_t* topic_list = build_topic_list();
   /*perf_loop(tmq, topic_list);*/

example/src/tstream.c

@@ -20,7 +20,7 @@
 #include "taos.h"
 
 int32_t init_env() {
-  TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 7010);
+  TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
   if (pConn == NULL) {
     return -1;
   }
@@ -65,7 +65,7 @@ int32_t init_env() {
 int32_t create_stream() {
   printf("create stream\n");
   TAOS_RES* pRes;
-  TAOS*     pConn = taos_connect("localhost", "root", "taosdata", NULL, 7010);
+  TAOS*     pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
   if (pConn == NULL) {
     return -1;
   }

include/client/taos.h

@@ -54,6 +54,7 @@ typedef void   TAOS_SUB;
 #define TSDB_DATA_TYPE_BLOB       18  // binary
 #define TSDB_DATA_TYPE_MEDIUMBLOB 19
 #define TSDB_DATA_TYPE_BINARY     TSDB_DATA_TYPE_VARCHAR  // string
+#define TSDB_DATA_TYPE_MAX        20
 
 typedef enum {
   TSDB_OPTION_LOCALE,
@@ -187,6 +188,9 @@ DLL_EXPORT void        taos_stop_query(TAOS_RES *res);
 DLL_EXPORT bool        taos_is_null(TAOS_RES *res, int32_t row, int32_t col);
 DLL_EXPORT bool        taos_is_update_query(TAOS_RES *res);
 DLL_EXPORT int         taos_fetch_block(TAOS_RES *res, TAOS_ROW *rows);
+DLL_EXPORT int         taos_fetch_block_s(TAOS_RES *res, int* numOfRows, TAOS_ROW *rows);
+DLL_EXPORT int         taos_fetch_raw_block(TAOS_RES *res, int* numOfRows, void** pData);
+DLL_EXPORT int        *taos_get_column_data_offset(TAOS_RES *res, int columnIndex);
 DLL_EXPORT int         taos_validate_sql(TAOS *taos, const char *sql);
 DLL_EXPORT void        taos_reset_current_db(TAOS *taos);
 

include/common/taosdef.h

@@ -76,6 +76,13 @@ typedef enum {
   TSDB_SMA_TYPE_ROLLUP = 2,      // Rollup SMA
 } ETsdbSmaType;
 
+typedef enum {
+  TSDB_BSMA_TYPE_NONE = 0,  // no block-wise SMA
+  TSDB_BSMA_TYPE_I = 1,     // sum/min/max(default)
+} ETsdbBSmaType;
+
+#define TSDB_BSMA_TYPE_LATEST TSDB_BSMA_TYPE_I
+
 extern char *qtypeStr[];
 
 #define TSDB_PORT_HTTP      11

include/common/tcommon.h

@@ -37,6 +37,14 @@ enum {
   TMQ_MSG_TYPE__EP_RSP,
 };
 
+enum {
+  STREAM_TRIGGER__AT_ONCE = 1,
+  STREAM_TRIGGER__WINDOW_CLOSE,
+  STREAM_TRIGGER__BY_COUNT,
+  STREAM_TRIGGER__BY_BATCH_COUNT,
+  STREAM_TRIGGER__BY_EVENT_TIME,
+};
+
 typedef struct {
   uint32_t  numOfTables;
   SArray*   pGroupList;
@@ -45,12 +53,12 @@ typedef struct {
 
 typedef struct SColumnDataAgg {
   int16_t colId;
-  int64_t sum;
-  int64_t max;
-  int64_t min;
   int16_t maxIndex;
   int16_t minIndex;
   int16_t numOfNull;
+  int64_t sum;
+  int64_t max;
+  int64_t min;
 } SColumnDataAgg;
 
 typedef struct SDataBlockInfo {
@@ -59,8 +67,11 @@ typedef struct SDataBlockInfo {
   int32_t     rowSize;
   int16_t     numOfCols;
   int16_t     hasVarCol;
-  union {int64_t uid; int64_t blockId;};
+  union {
-  int64_t        groupId;     // no need to serialize
+    int64_t uid;
+    int64_t blockId;
+  };
+  uint64_t    groupId;  // no need to serialize
 } SDataBlockInfo;
 
 typedef struct SSDataBlock {
@@ -182,24 +193,23 @@ typedef struct SColumn {
   uint8_t scale;
 } SColumn;
 
-typedef struct SLimit {
+typedef struct STableBlockDistInfo {
-  int64_t limit;
+  uint16_t  rowSize;
-  int64_t offset;
+  uint16_t  numOfFiles;
-} SLimit;
+  uint32_t  numOfTables;
-
+  uint64_t  totalSize;
-typedef struct SOrder {
+  uint64_t  totalRows;
-  uint32_t order;
+  int32_t   maxRows;
-  SColumn  col;
+  int32_t   minRows;
-} SOrder;
+  int32_t   firstSeekTimeUs;
-
+  uint32_t  numOfRowsInMemTable;
-typedef struct SGroupbyExpr {
+  uint32_t  numOfSmallBlocks;
-  SArray* columnInfo;  // SArray<SColIndex>, group by columns information
+  SArray   *dataBlockInfos;
-  bool    groupbyTag;  // group by tag or column
+} STableBlockDistInfo;
-} SGroupbyExpr;
 
 enum {
-  FUNC_PARAM_TYPE_VALUE = 0,
+  FUNC_PARAM_TYPE_VALUE = 0x1,
-  FUNC_PARAM_TYPE_COLUMN,
+  FUNC_PARAM_TYPE_COLUMN= 0x2,
 };
 
 typedef struct SFunctParam {
@@ -230,15 +240,6 @@ typedef struct SExprInfo {
   struct tExprNode*     pExpr;
 } SExprInfo;
 
-typedef struct SStateWindow {
-  SColumn col;
-} SStateWindow;
-
-typedef struct SSessionWindow {
-  int64_t gap;  // gap between two session window(in microseconds)
-  SColumn col;
-} SSessionWindow;
-
 #define QUERY_ASC_FORWARD_STEP 1
 #define QUERY_DESC_FORWARD_STEP -1
 

include/common/tdatablock.h

@@ -113,39 +113,53 @@ static FORCE_INLINE void colDataAppendNULL(SColumnInfoData* pColumnInfoData, uin
   pColumnInfoData->hasNull = true;
 }
 
-static FORCE_INLINE int32_t colDataAppendInt8(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int8_t* v) {
+static FORCE_INLINE void colDataAppendNNULL(SColumnInfoData* pColumnInfoData, uint32_t start, size_t nRows) {
+  if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
+    for(int32_t i = start; i < start + nRows; ++i) {
+      pColumnInfoData->varmeta.offset[i] = -1;  // it is a null value of VAR type.
+    }
+  } else {
+    for(int32_t i = start; i < start + nRows; ++i) {
+      colDataSetNull_f(pColumnInfoData->nullbitmap, i);
+    }
+  }
+
+  pColumnInfoData->hasNull = true;
+}
+
+static FORCE_INLINE void colDataAppendInt8(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int8_t* v) {
   ASSERT(pColumnInfoData->info.type == TSDB_DATA_TYPE_TINYINT ||
          pColumnInfoData->info.type == TSDB_DATA_TYPE_UTINYINT || pColumnInfoData->info.type == TSDB_DATA_TYPE_BOOL);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(int8_t*)p = *(int8_t*)v;
 }
 
-static FORCE_INLINE int32_t colDataAppendInt16(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int16_t* v) {
+static FORCE_INLINE void colDataAppendInt16(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int16_t* v) {
   ASSERT(pColumnInfoData->info.type == TSDB_DATA_TYPE_SMALLINT || pColumnInfoData->info.type == TSDB_DATA_TYPE_USMALLINT);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(int16_t*)p = *(int16_t*)v;
 }
 
-static FORCE_INLINE int32_t colDataAppendInt32(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int32_t* v) {
+static FORCE_INLINE void colDataAppendInt32(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int32_t* v) {
   ASSERT(pColumnInfoData->info.type == TSDB_DATA_TYPE_INT || pColumnInfoData->info.type == TSDB_DATA_TYPE_UINT);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(int32_t*)p = *(int32_t*)v;
 }
 
-static FORCE_INLINE int32_t colDataAppendInt64(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int64_t* v) {
+static FORCE_INLINE void colDataAppendInt64(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int64_t* v) {
   int32_t type = pColumnInfoData->info.type;
   ASSERT(type == TSDB_DATA_TYPE_BIGINT || type == TSDB_DATA_TYPE_UBIGINT || type == TSDB_DATA_TYPE_TIMESTAMP);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(int64_t*)p = *(int64_t*)v;
 }
 
-static FORCE_INLINE int32_t colDataAppendFloat(SColumnInfoData* pColumnInfoData, uint32_t currentRow, float* v) {
+static FORCE_INLINE void colDataAppendFloat(SColumnInfoData* pColumnInfoData, uint32_t currentRow, float* v) {
   ASSERT(pColumnInfoData->info.type == TSDB_DATA_TYPE_FLOAT);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(float*)p = *(float*)v;
 }
 
-static FORCE_INLINE int32_t colDataAppendDouble(SColumnInfoData* pColumnInfoData, uint32_t currentRow, double* v) {
+static FORCE_INLINE void colDataAppendDouble(SColumnInfoData* pColumnInfoData, uint32_t currentRow, double* v) {
   ASSERT(pColumnInfoData->info.type == TSDB_DATA_TYPE_DOUBLE);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(double*)p = *(double*)v;
@@ -163,11 +177,12 @@ void    colDataTrim(SColumnInfoData* pColumnInfoData);
 size_t blockDataGetNumOfCols(const SSDataBlock* pBlock);
 size_t blockDataGetNumOfRows(const SSDataBlock* pBlock);
 
-int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc);
+int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc, SArray* pIndexMap);
 int32_t blockDataSplitRows(SSDataBlock* pBlock, bool hasVarCol, int32_t startIndex, int32_t* stopIndex,
                            int32_t pageSize);
 int32_t blockDataToBuf(char* buf, const SSDataBlock* pBlock);
 int32_t blockDataFromBuf(SSDataBlock* pBlock, const char* buf);
+int32_t blockDataFromBuf1(SSDataBlock* pBlock, const char* buf, size_t capacity);
 
 SSDataBlock* blockDataExtractBlock(SSDataBlock* pBlock, int32_t startIndex, int32_t rowCount);
 
@@ -179,12 +194,14 @@ size_t blockDataGetSerialMetaSize(const SSDataBlock* pBlock);
 int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo);
 int32_t blockDataSort_rv(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullFirst);
 
-int32_t blockDataEnsureColumnCapacity(SColumnInfoData* pColumn, uint32_t numOfRows);
+int32_t colInfoDataEnsureCapacity(SColumnInfoData* pColumn, uint32_t numOfRows);
 int32_t blockDataEnsureCapacity(SSDataBlock* pDataBlock, uint32_t numOfRows);
 void    blockDataCleanup(SSDataBlock* pDataBlock);
 size_t  blockDataGetCapacityInRow(const SSDataBlock* pBlock, size_t pageSize);
 void*   blockDataDestroy(SSDataBlock* pBlock);
 
+int32_t blockDataTrimFirstNRows(SSDataBlock *pBlock, size_t n);
+
 SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock);
 
 void blockDebugShowData(const SArray* dataBlocks);

include/common/tdataformat.h

@@ -70,11 +70,13 @@ typedef struct {
 #pragma pack(pop)
 
 #define colType(col)   ((col)->type)
+#define colSma(col)    ((col)->sma)
 #define colColId(col)  ((col)->colId)
 #define colBytes(col)  ((col)->bytes)
 #define colOffset(col) ((col)->offset)
 
 #define colSetType(col, t)   (colType(col) = (t))
+#define colSetSma(col, s)    (colSma(col) = (s))
 #define colSetColId(col, id) (colColId(col) = (id))
 #define colSetBytes(col, b)  (colBytes(col) = (b))
 #define colSetOffset(col, o) (colOffset(col) = (o))
@@ -133,13 +135,13 @@ typedef struct {
 #define TD_VTYPE_PARTS 4  // 8 bits / TD_VTYPE_BITS = 4
 #define TD_VTYPE_OPTR  3  // TD_VTYPE_PARTS - 1, utilize to get remainder
 
-#define TD_BITMAP_BYTES(cnt) (ceil((double)cnt / TD_VTYPE_PARTS))
+#define TD_BITMAP_BYTES(cnt) (ceil((double)(cnt) / TD_VTYPE_PARTS))
-#define TD_BIT_TO_BYTES(cnt) (ceil((double)cnt / 8))
+#define TD_BIT_TO_BYTES(cnt) (ceil((double)(cnt) / 8))
 
 int32_t   tdInitTSchemaBuilder(STSchemaBuilder *pBuilder, schema_ver_t version);
 void      tdDestroyTSchemaBuilder(STSchemaBuilder *pBuilder);
 void      tdResetTSchemaBuilder(STSchemaBuilder *pBuilder, schema_ver_t version);
-int32_t   tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, col_id_t colId, col_bytes_t bytes);
+int32_t   tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t sma, col_id_t colId, col_bytes_t bytes);
 STSchema *tdGetSchemaFromBuilder(STSchemaBuilder *pBuilder);
 
 // ----------------- Semantic timestamp key definition
@@ -363,6 +365,7 @@ static FORCE_INLINE void tdCopyColOfRowBySchema(SDataRow dst, STSchema *pDstSche
 }
 #endif
 // ----------------- Data column structure
+// SDataCol arrangement: data => bitmap => dataOffset
 typedef struct SDataCol {
   int8_t          type;        // column type
   uint8_t         bitmap : 1;  // 0: has bitmap if has NULL/NORM rows, 1: no bitmap if all rows are NORM

include/common/tglobal.h

@@ -41,9 +41,6 @@ extern int32_t tsMaxConnections;
 extern int32_t tsMaxShellConns;
 extern int32_t tsShellActivityTimer;
 extern int32_t tsMaxTmrCtrl;
-extern float   tsNumOfThreadsPerCore;
-extern int32_t tsNumOfCommitThreads;
-extern float   tsRatioOfQueryCores;
 extern int32_t tsCompressMsgSize;
 extern int32_t tsCompressColData;
 extern int32_t tsMaxNumOfDistinctResults;
@@ -51,7 +48,30 @@ extern int32_t tsCompatibleModel;
 extern bool    tsEnableSlaveQuery;
 extern bool    tsPrintAuth;
 extern int64_t tsTickPerDay[3];
+
+// multi-process
 extern bool    tsMultiProcess;
+extern int32_t tsMnodeShmSize;
+extern int32_t tsVnodeShmSize;
+extern int32_t tsQnodeShmSize;
+extern int32_t tsSnodeShmSize;
+extern int32_t tsBnodeShmSize;
+
+// queue & threads
+extern int32_t tsNumOfRpcThreads;
+extern int32_t tsNumOfCommitThreads;
+extern int32_t tsNumOfTaskQueueThreads;
+extern int32_t tsNumOfMnodeQueryThreads;
+extern int32_t tsNumOfMnodeReadThreads;
+extern int32_t tsNumOfVnodeQueryThreads;
+extern int32_t tsNumOfVnodeFetchThreads;
+extern int32_t tsNumOfVnodeWriteThreads;
+extern int32_t tsNumOfVnodeSyncThreads;
+extern int32_t tsNumOfVnodeMergeThreads;
+extern int32_t tsNumOfQnodeQueryThreads;
+extern int32_t tsNumOfQnodeFetchThreads;
+extern int32_t tsNumOfSnodeSharedThreads;
+extern int32_t tsNumOfSnodeUniqueThreads;
 
 // monitor
 extern bool     tsEnableMonitor;

include/common/tmsg.h

@@ -265,6 +265,20 @@ typedef struct SSchema {
   char     name[TSDB_COL_NAME_LEN];
 } SSchema;
 
+typedef struct {
+  int8_t   type;
+  int8_t   sma;  // ETsdbBSmaType and default is TSDB_BSMA_TYPE_I
+  col_id_t colId;
+  int32_t  bytes;
+  char     name[TSDB_COL_NAME_LEN];
+} SSchemaEx;
+
+#define SSCHMEA_TYPE(s)  ((s)->type)
+#define SSCHMEA_SMA(s)   ((s)->sma)
+#define SSCHMEA_COLID(s) ((s)->colId)
+#define SSCHMEA_BYTES(s) ((s)->bytes)
+#define SSCHMEA_NAME(s)  ((s)->name)
+
 typedef struct {
   char    name[TSDB_TABLE_FNAME_LEN];
   int8_t  igExists;
@@ -469,7 +483,8 @@ typedef struct {
   int32_t tz;  // query client timezone
   char    intervalUnit;
   char    slidingUnit;
-  char    offsetUnit;  // TODO Remove it, the offset is the number of precision tickle, and it must be a immutable duration.
+  char
+      offsetUnit;  // TODO Remove it, the offset is the number of precision tickle, and it must be a immutable duration.
   int8_t  precision;
   int64_t interval;
   int64_t sliding;
@@ -497,6 +512,7 @@ typedef struct {
   int32_t maxRows;
   int32_t commitTime;
   int32_t fsyncPeriod;
+  int32_t ttl;
   int8_t  walLevel;
   int8_t  precision;  // time resolution
   int8_t  compression;
@@ -506,6 +522,7 @@ typedef struct {
   int8_t  cacheLastRow;
   int8_t  ignoreExist;
   int8_t  streamMode;
+  int8_t  singleSTable;
   int32_t numOfRetensions;
   SArray* pRetensions;  // SRetention
 } SCreateDbReq;
@@ -524,6 +541,7 @@ typedef struct {
   int8_t  walLevel;
   int8_t  quorum;
   int8_t  cacheLastRow;
+  int8_t  replications;
 } SAlterDbReq;
 
 int32_t tSerializeSAlterDbReq(void* buf, int32_t bufLen, SAlterDbReq* pReq);
@@ -570,6 +588,41 @@ int32_t tSerializeSUseDbRspImp(SCoder* pEncoder, const SUseDbRsp* pRsp);
 int32_t tDeserializeSUseDbRspImp(SCoder* pDecoder, SUseDbRsp* pRsp);
 void    tFreeSUsedbRsp(SUseDbRsp* pRsp);
 
+typedef struct {
+  char    db[TSDB_DB_FNAME_LEN];
+} SDbCfgReq;
+
+int32_t tSerializeSDbCfgReq(void* buf, int32_t bufLen, SDbCfgReq* pReq);
+int32_t tDeserializeSDbCfgReq(void* buf, int32_t bufLen, SDbCfgReq* pReq);
+
+typedef struct {
+  int32_t numOfVgroups;
+  int32_t cacheBlockSize;
+  int32_t totalBlocks;
+  int32_t daysPerFile;
+  int32_t daysToKeep0;
+  int32_t daysToKeep1;
+  int32_t daysToKeep2;
+  int32_t minRows;
+  int32_t maxRows;
+  int32_t commitTime;
+  int32_t fsyncPeriod;
+  int32_t ttl;
+  int8_t  walLevel;
+  int8_t  precision;
+  int8_t  compression;
+  int8_t  replications;
+  int8_t  quorum;
+  int8_t  update;
+  int8_t  cacheLastRow;
+  int8_t  streamMode;
+  int8_t  singleSTable;
+} SDbCfgRsp;
+
+int32_t tSerializeSDbCfgRsp(void* buf, int32_t bufLen, const SDbCfgRsp* pRsp);
+int32_t tDeserializeSDbCfgRsp(void* buf, int32_t bufLen, SDbCfgRsp* pRsp);
+
+
 typedef struct {
   int32_t rowNum;
 } SQnodeListReq;
@@ -664,6 +717,18 @@ typedef struct {
   char    charset[TD_LOCALE_LEN];     // tsCharset
 } SClusterCfg;
 
+typedef struct {
+  int32_t openVnodes;
+  int32_t totalVnodes;
+  int32_t masterNum;
+  int64_t numOfSelectReqs;
+  int64_t numOfInsertReqs;
+  int64_t numOfInsertSuccessReqs;
+  int64_t numOfBatchInsertReqs;
+  int64_t numOfBatchInsertSuccessReqs;
+  int64_t errors;
+} SVnodesStat;
+
 typedef struct {
   int32_t vgId;
   int8_t  role;
@@ -915,6 +980,21 @@ typedef struct {
   char    data[];
 } SRetrieveMetaTableRsp;
 
+typedef struct SExplainExecInfo {
+  uint64_t startupCost;
+  uint64_t totalCost;
+  uint64_t numOfRows;
+  void*    verboseInfo;
+} SExplainExecInfo;
+
+typedef struct {
+  int32_t           numOfPlans;
+  SExplainExecInfo* subplanInfo;
+} SExplainRsp;
+
+int32_t tSerializeSExplainRsp(void* buf, int32_t bufLen, SExplainRsp* pRsp);
+int32_t tDeserializeSExplainRsp(void* buf, int32_t bufLen, SExplainRsp* pRsp);
+
 typedef struct {
   char    fqdn[TSDB_FQDN_LEN];  // end point, hostname:port
   int32_t port;
@@ -1052,6 +1132,7 @@ typedef struct SSubQueryMsg {
   uint64_t taskId;
   int64_t  refId;
   int8_t   taskType;
+  int8_t   explain;
   uint32_t sqlLen;  // the query sql,
   uint32_t phyLen;
   char     msg[];
@@ -1381,10 +1462,9 @@ typedef struct {
 
 typedef struct {
   float      xFilesFactor;
-  int8_t   delayUnit;
+  int32_t    delay;
   int8_t     nFuncIds;
-  int32_t* pFuncIds;
+  func_id_t* pFuncIds;
-  int64_t  delay;
 } SRSmaParam;
 
 typedef struct SVCreateTbReq {
@@ -1403,12 +1483,11 @@ typedef struct SVCreateTbReq {
   union {
     struct {
       tb_uid_t    suid;
-      uint32_t    nCols;
+      col_id_t    nCols;
-      SSchema*    pSchema;
-      uint32_t    nTagCols;
-      SSchema*    pTagSchema;
       col_id_t    nBSmaCols;
-      col_id_t*   pBSmaCols;
+      SSchemaEx*  pSchema;
+      col_id_t    nTagCols;
+      SSchema*    pTagSchema;
       SRSmaParam* pRSmaParam;
     } stbCfg;
     struct {
@@ -1416,10 +1495,9 @@ typedef struct SVCreateTbReq {
       SKVRow   pTag;
     } ctbCfg;
     struct {
-      uint32_t    nCols;
+      col_id_t    nCols;
-      SSchema*    pSchema;
       col_id_t    nBSmaCols;
-      col_id_t*   pBSmaCols;
+      SSchemaEx*  pSchema;
       SRSmaParam* pRSmaParam;
     } ntbCfg;
   };
@@ -1799,6 +1877,7 @@ static FORCE_INLINE void* taosDecodeSMqMsg(void* buf, SMqHbMsg* pMsg) {
 typedef struct {
   int64_t leftForVer;
   int32_t vgId;
+  int32_t epoch;
   int64_t consumerId;
   char    topicName[TSDB_TOPIC_FNAME_LEN];
   char    cgroup[TSDB_CGROUP_LEN];
@@ -1812,6 +1891,7 @@ static FORCE_INLINE int32_t tEncodeSMqSetCVgReq(void** buf, const SMqSetCVgReq*
   int32_t tlen = 0;
   tlen += taosEncodeFixedI64(buf, pReq->leftForVer);
   tlen += taosEncodeFixedI32(buf, pReq->vgId);
+  tlen += taosEncodeFixedI32(buf, pReq->epoch);
   tlen += taosEncodeFixedI64(buf, pReq->consumerId);
   tlen += taosEncodeString(buf, pReq->topicName);
   tlen += taosEncodeString(buf, pReq->cgroup);
@@ -1825,6 +1905,7 @@ static FORCE_INLINE int32_t tEncodeSMqSetCVgReq(void** buf, const SMqSetCVgReq*
 static FORCE_INLINE void* tDecodeSMqSetCVgReq(void* buf, SMqSetCVgReq* pReq) {
   buf = taosDecodeFixedI64(buf, &pReq->leftForVer);
   buf = taosDecodeFixedI32(buf, &pReq->vgId);
+  buf = taosDecodeFixedI32(buf, &pReq->epoch);
   buf = taosDecodeFixedI64(buf, &pReq->consumerId);
   buf = taosDecodeStringTo(buf, pReq->topicName);
   buf = taosDecodeStringTo(buf, pReq->cgroup);
@@ -1835,11 +1916,43 @@ static FORCE_INLINE void* tDecodeSMqSetCVgReq(void* buf, SMqSetCVgReq* pReq) {
   return buf;
 }
 
+typedef struct {
+  int64_t leftForVer;
+  int32_t vgId;
+  int32_t epoch;
+  int64_t consumerId;
+  char    topicName[TSDB_TOPIC_FNAME_LEN];
+} SMqCancelConnReq;
+
+static FORCE_INLINE int32_t tEncodeSMqCancelConnReq(void** buf, const SMqCancelConnReq* pReq) {
+  int32_t tlen = 0;
+  tlen += taosEncodeFixedI64(buf, pReq->leftForVer);
+  tlen += taosEncodeFixedI32(buf, pReq->vgId);
+  tlen += taosEncodeFixedI32(buf, pReq->epoch);
+  tlen += taosEncodeFixedI64(buf, pReq->consumerId);
+  tlen += taosEncodeString(buf, pReq->topicName);
+  return tlen;
+}
+
+static FORCE_INLINE void* tDecodeSMqCancelConnReq(void* buf, SMqCancelConnReq* pReq) {
+  buf = taosDecodeFixedI64(buf, &pReq->leftForVer);
+  buf = taosDecodeFixedI32(buf, &pReq->vgId);
+  buf = taosDecodeFixedI32(buf, &pReq->epoch);
+  buf = taosDecodeFixedI64(buf, &pReq->consumerId);
+  buf = taosDecodeStringTo(buf, pReq->topicName);
+  return buf;
+}
+
+typedef struct {
+  int8_t reserved;
+} SMqCancelConnRsp;
+
 typedef struct {
   int64_t leftForVer;
   int32_t vgId;
   int64_t oldConsumerId;
   int64_t newConsumerId;
+  char*   topic;
 } SMqMVRebReq;
 
 static FORCE_INLINE int32_t tEncodeSMqMVRebReq(void** buf, const SMqMVRebReq* pReq) {
@@ -1848,6 +1961,7 @@ static FORCE_INLINE int32_t tEncodeSMqMVRebReq(void** buf, const SMqMVRebReq* pR
   tlen += taosEncodeFixedI32(buf, pReq->vgId);
   tlen += taosEncodeFixedI64(buf, pReq->oldConsumerId);
   tlen += taosEncodeFixedI64(buf, pReq->newConsumerId);
+  tlen += taosEncodeString(buf, pReq->topic);
   return tlen;
 }
 
@@ -1856,6 +1970,7 @@ static FORCE_INLINE void* tDecodeSMqMVRebReq(void* buf, SMqMVRebReq* pReq) {
   buf = taosDecodeFixedI32(buf, &pReq->vgId);
   buf = taosDecodeFixedI64(buf, &pReq->oldConsumerId);
   buf = taosDecodeFixedI64(buf, &pReq->newConsumerId);
+  buf = taosDecodeString(buf, &pReq->topic);
   return buf;
 }
 
@@ -1898,7 +2013,10 @@ int32_t tDecodeSMqCMCommitOffsetReq(SCoder* decoder, SMqCMCommitOffsetReq* pReq)
 
 typedef struct {
   uint32_t nCols;
+  union {
     SSchema*   pSchema;
+    SSchemaEx* pSchemaEx;
+  };
 } SSchemaWrapper;
 
 static FORCE_INLINE int32_t taosEncodeSSchema(void** buf, const SSchema* pSchema) {
@@ -2187,6 +2305,24 @@ static FORCE_INLINE void* tDecodeTSmaWrapper(void* buf, STSmaWrapper* pSW) {
   return buf;
 }
 
+typedef struct {
+  char indexFName[TSDB_INDEX_FNAME_LEN];
+} SUserIndexReq;
+
+int32_t tSerializeSUserIndexReq(void* buf, int32_t bufLen, SUserIndexReq* pReq);
+int32_t tDeserializeSUserIndexReq(void* buf, int32_t bufLen, SUserIndexReq* pReq);
+
+typedef struct {
+  char dbFName[TSDB_DB_FNAME_LEN];
+  char tblFName[TSDB_TABLE_FNAME_LEN];
+  char colName[TSDB_COL_NAME_LEN];
+  char indexType[TSDB_INDEX_TYPE_LEN];
+  char indexExts[TSDB_INDEX_EXTS_LEN];
+} SUserIndexRsp;
+
+int32_t tSerializeSUserIndexRsp(void* buf, int32_t bufLen, const SUserIndexRsp* pRsp);
+int32_t tDeserializeSUserIndexRsp(void* buf, int32_t bufLen, SUserIndexRsp* pRsp);
+
 typedef struct {
   int8_t  mqMsgType;
   int32_t code;
@@ -2204,6 +2340,7 @@ typedef struct {
   char    cgroup[TSDB_CGROUP_LEN];
 
   int64_t currentOffset;
+  uint64_t reqId;
   char    topic[TSDB_TOPIC_FNAME_LEN];
 } SMqPollReq;
 

include/common/tmsgcb.h

@@ -50,7 +50,6 @@ typedef struct {
   PutToQueueFp            queueFps[QUEUE_MAX];
   GetQueueSizeFp          qsizeFp;
   SendReqFp               sendReqFp;
-  SendMnodeReqFp          sendMnodeReqFp;
   SendRspFp               sendRspFp;
   RegisterBrokenLinkArgFp registerBrokenLinkArgFp;
   ReleaseHandleFp         releaseHandleFp;
@@ -60,7 +59,6 @@ void    tmsgSetDefaultMsgCb(const SMsgCb* pMsgCb);
 int32_t tmsgPutToQueue(const SMsgCb* pMsgCb, EQueueType qtype, SRpcMsg* pReq);
 int32_t tmsgGetQueueSize(const SMsgCb* pMsgCb, int32_t vgId, EQueueType qtype);
 int32_t tmsgSendReq(const SMsgCb* pMsgCb, const SEpSet* epSet, SRpcMsg* pReq);
-int32_t tmsgSendMnodeReq(const SMsgCb* pMsgCb, SRpcMsg* pReq);
 void    tmsgSendRsp(const SRpcMsg* pRsp);
 void    tmsgRegisterBrokenLinkArg(const SMsgCb* pMsgCb, SRpcMsg* pMsg);
 void    tmsgReleaseHandle(void* handle, int8_t type);

include/common/tmsgdef.h

@@ -155,6 +155,8 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_STREAM, "mnode-create-stream", SCMCreateStreamReq, SCMCreateStreamRsp)
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_STREAM, "mnode-alter-stream", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_STREAM, "mnode-drop-stream", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_GET_DB_CFG, "mnode-get-db-cfg", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_GET_INDEX, "mnode-get-index", NULL, NULL)
 
   // Requests handled by VNODE
   TD_NEW_MSG_SEG(TDMT_VND_MSG)
@@ -176,6 +178,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_DISCONNECT, "vnode-mq-disconnect", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_SET_CONN, "vnode-mq-set-conn", SMqSetCVgReq, SMqSetCVgRsp)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_REB, "vnode-mq-mv-rebalance", SMqMVRebReq, SMqMVRebRsp)
+  TD_DEF_MSG_TYPE(TDMT_VND_MQ_CANCEL_CONN, "vnode-mq-mv-cancel-conn", SMqCancelConnReq, SMqCancelConnRsp)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_SET_CUR, "vnode-mq-set-cur", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_RES_READY, "vnode-res-ready", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_TASKS_STATUS, "vnode-tasks-status", NULL, NULL)
@@ -188,6 +191,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_SHOW_TABLES_FETCH, "vnode-show-tables-fetch", SVShowTablesFetchReq, SVShowTablesFetchRsp)
   TD_DEF_MSG_TYPE(TDMT_VND_QUERY_CONTINUE, "vnode-query-continue", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_QUERY_HEARTBEAT, "vnode-query-heartbeat", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_VND_EXPLAIN, "vnode-explain", NULL, NULL)
 
   TD_DEF_MSG_TYPE(TDMT_VND_SUBSCRIBE, "vnode-subscribe", SMVSubscribeReq, SMVSubscribeRsp)
   TD_DEF_MSG_TYPE(TDMT_VND_CONSUME, "vnode-consume", SMqCVConsumeReq, SMqCVConsumeRsp)
@@ -215,6 +219,15 @@ enum {
   TD_NEW_MSG_SEG(TDMT_SCH_MSG)
   TD_DEF_MSG_TYPE(TDMT_SCH_LINK_BROKEN, "scheduler-link-broken", NULL, NULL)
   
+  // Monitor info exchange between processes
+  TD_NEW_MSG_SEG(TDMT_MON_MSG)
+  TD_DEF_MSG_TYPE(TDMT_MON_MM_INFO, "monitor-minfo", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MON_VM_INFO, "monitor-vinfo", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MON_QM_INFO, "monitor-qinfo", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MON_SM_INFO, "monitor-sinfo", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MON_BM_INFO, "monitor-binfo", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MON_VM_LOAD, "monitor-vload", NULL, NULL)
+
 #if defined(TD_MSG_NUMBER_)
   TDMT_MAX
 #endif

include/common/trow.h

@@ -148,15 +148,15 @@ typedef struct {
 typedef struct {
   // basic info
   int8_t       rowType;
-  int16_t sver;
+  schema_ver_t sver;
   STSRow      *pBuf;
 
   // extended info
   int32_t  flen;
-  int16_t nBoundCols;
+  col_id_t nBoundCols;
-  int16_t nCols;
+  col_id_t nCols;
-  int16_t nBitmaps;
+  col_id_t nBitmaps;
-  int16_t nBoundBitmaps;
+  col_id_t nBoundBitmaps;
   int32_t  offset;
   void    *pBitmap;
   void    *pOffset;
@@ -448,9 +448,9 @@ static FORCE_INLINE int32_t tdSRowSetExtendedInfo(SRowBuilder *pBuilder, int32_t
   }
 #ifdef TD_SUPPORT_BITMAP
   // the primary TS key is stored separatedly
-  pBuilder->nBitmaps = (int16_t)TD_BITMAP_BYTES(pBuilder->nCols - 1);
+  pBuilder->nBitmaps = (col_id_t)TD_BITMAP_BYTES(pBuilder->nCols - 1);
   if (nBoundCols > 0) {
-    pBuilder->nBoundBitmaps = (int16_t)TD_BITMAP_BYTES(pBuilder->nBoundCols - 1);
+    pBuilder->nBoundBitmaps = (col_id_t)TD_BITMAP_BYTES(pBuilder->nBoundCols - 1);
   } else {
     pBuilder->nBoundBitmaps = 0;
   }

include/common/ttime.h

@@ -60,8 +60,10 @@ int32_t parseNatualDuration(const char* token, int32_t tokenLen, int64_t* durati
 
 int32_t taosParseTime(const char* timestr, int64_t* time, int32_t len, int32_t timePrec, int8_t dayligth);
 void    deltaToUtcInitOnce();
+char getPrecisionUnit(int32_t precision);
 
 int64_t convertTimePrecision(int64_t time, int32_t fromPrecision, int32_t toPrecision);
+int64_t convertTimeFromPrecisionToUnit(int64_t time, int32_t fromPrecision, char toUnit);
 
 void taosFormatUtcTime(char *buf, int32_t bufLen, int64_t time, int32_t precision);
 

include/common/ttokendef.h

@@ -69,128 +69,154 @@
 #define TK_CACHELAST                       51
 #define TK_COMP                            52
 #define TK_DAYS                            53
-#define TK_FSYNC                           54
+#define TK_NK_VARIABLE                     54
-#define TK_MAXROWS                         55
+#define TK_FSYNC                           55
-#define TK_MINROWS                         56
+#define TK_MAXROWS                         56
-#define TK_KEEP                            57
+#define TK_MINROWS                         57
-#define TK_PRECISION                       58
+#define TK_KEEP                            58
-#define TK_QUORUM                          59
+#define TK_PRECISION                       59
-#define TK_REPLICA                         60
+#define TK_QUORUM                          60
-#define TK_TTL                             61
+#define TK_REPLICA                         61
-#define TK_WAL                             62
+#define TK_TTL                             62
-#define TK_VGROUPS                         63
+#define TK_WAL                             63
-#define TK_SINGLE_STABLE                   64
+#define TK_VGROUPS                         64
-#define TK_STREAM_MODE                     65
+#define TK_SINGLE_STABLE                   65
-#define TK_RETENTIONS                      66
+#define TK_STREAM_MODE                     66
-#define TK_TABLE                           67
+#define TK_RETENTIONS                      67
-#define TK_NK_LP                           68
+#define TK_NK_COMMA                        68
-#define TK_NK_RP                           69
+#define TK_NK_COLON                        69
-#define TK_STABLE                          70
+#define TK_TABLE                           70
-#define TK_ADD                             71
+#define TK_NK_LP                           71
-#define TK_COLUMN                          72
+#define TK_NK_RP                           72
-#define TK_MODIFY                          73
+#define TK_STABLE                          73
-#define TK_RENAME                          74
+#define TK_ADD                             74
-#define TK_TAG                             75
+#define TK_COLUMN                          75
-#define TK_SET                             76
+#define TK_MODIFY                          76
-#define TK_NK_EQ                           77
+#define TK_RENAME                          77
-#define TK_USING                           78
+#define TK_TAG                             78
-#define TK_TAGS                            79
+#define TK_SET                             79
-#define TK_NK_DOT                          80
+#define TK_NK_EQ                           80
-#define TK_NK_COMMA                        81
+#define TK_USING                           81
-#define TK_COMMENT                         82
+#define TK_TAGS                            82
-#define TK_BOOL                            83
+#define TK_NK_DOT                          83
-#define TK_TINYINT                         84
+#define TK_COMMENT                         84
-#define TK_SMALLINT                        85
+#define TK_BOOL                            85
-#define TK_INT                             86
+#define TK_TINYINT                         86
-#define TK_INTEGER                         87
+#define TK_SMALLINT                        87
-#define TK_BIGINT                          88
+#define TK_INT                             88
-#define TK_FLOAT                           89
+#define TK_INTEGER                         89
-#define TK_DOUBLE                          90
+#define TK_BIGINT                          90
-#define TK_BINARY                          91
+#define TK_FLOAT                           91
-#define TK_TIMESTAMP                       92
+#define TK_DOUBLE                          92
-#define TK_NCHAR                           93
+#define TK_BINARY                          93
-#define TK_UNSIGNED                        94
+#define TK_TIMESTAMP                       94
-#define TK_JSON                            95
+#define TK_NCHAR                           95
-#define TK_VARCHAR                         96
+#define TK_UNSIGNED                        96
-#define TK_MEDIUMBLOB                      97
+#define TK_JSON                            97
-#define TK_BLOB                            98
+#define TK_VARCHAR                         98
-#define TK_VARBINARY                       99
+#define TK_MEDIUMBLOB                      99
-#define TK_DECIMAL                        100
+#define TK_BLOB                           100
-#define TK_SMA                            101
+#define TK_VARBINARY                      101
-#define TK_ROLLUP                         102
+#define TK_DECIMAL                        102
-#define TK_FILE_FACTOR                    103
+#define TK_SMA                            103
-#define TK_NK_FLOAT                       104
+#define TK_ROLLUP                         104
-#define TK_DELAY                          105
+#define TK_FILE_FACTOR                    105
-#define TK_SHOW                           106
+#define TK_NK_FLOAT                       106
-#define TK_DATABASES                      107
+#define TK_DELAY                          107
-#define TK_TABLES                         108
+#define TK_SHOW                           108
-#define TK_STABLES                        109
+#define TK_DATABASES                      109
-#define TK_MNODES                         110
+#define TK_TABLES                         110
-#define TK_MODULES                        111
+#define TK_STABLES                        111
-#define TK_QNODES                         112
+#define TK_MNODES                         112
-#define TK_FUNCTIONS                      113
+#define TK_MODULES                        113
-#define TK_INDEXES                        114
+#define TK_QNODES                         114
-#define TK_FROM                           115
+#define TK_FUNCTIONS                      115
-#define TK_LIKE                           116
+#define TK_INDEXES                        116
-#define TK_INDEX                          117
+#define TK_FROM                           117
-#define TK_FULLTEXT                       118
+#define TK_ACCOUNTS                       118
-#define TK_FUNCTION                       119
+#define TK_APPS                           119
-#define TK_INTERVAL                       120
+#define TK_CONNECTIONS                    120
-#define TK_TOPIC                          121
+#define TK_LICENCE                        121
-#define TK_AS                             122
+#define TK_QUERIES                        122
-#define TK_EXPLAIN                        123
+#define TK_SCORES                         123
-#define TK_ANALYZE                        124
+#define TK_TOPICS                         124
-#define TK_VERBOSE                        125
+#define TK_VARIABLES                      125
-#define TK_NK_BOOL                        126
+#define TK_LIKE                           126
-#define TK_RATIO                          127
+#define TK_INDEX                          127
-#define TK_NULL                           128
+#define TK_FULLTEXT                       128
-#define TK_NK_VARIABLE                    129
+#define TK_FUNCTION                       129
-#define TK_NK_UNDERLINE                   130
+#define TK_INTERVAL                       130
-#define TK_ROWTS                          131
+#define TK_TOPIC                          131
-#define TK_TBNAME                         132
+#define TK_AS                             132
-#define TK_QSTARTTS                       133
+#define TK_DESC                           133
-#define TK_QENDTS                         134
+#define TK_DESCRIBE                       134
-#define TK_WSTARTTS                       135
+#define TK_RESET                          135
-#define TK_WENDTS                         136
+#define TK_QUERY                          136
-#define TK_WDURATION                      137
+#define TK_EXPLAIN                        137
-#define TK_BETWEEN                        138
+#define TK_ANALYZE                        138
-#define TK_IS                             139
+#define TK_VERBOSE                        139
-#define TK_NK_LT                          140
+#define TK_NK_BOOL                        140
-#define TK_NK_GT                          141
+#define TK_RATIO                          141
-#define TK_NK_LE                          142
+#define TK_COMPACT                        142
-#define TK_NK_GE                          143
+#define TK_VNODES                         143
-#define TK_NK_NE                          144
+#define TK_IN                             144
-#define TK_MATCH                          145
+#define TK_OUTPUTTYPE                     145
-#define TK_NMATCH                         146
+#define TK_AGGREGATE                      146
-#define TK_IN                             147
+#define TK_BUFSIZE                        147
-#define TK_JOIN                           148
+#define TK_STREAM                         148
-#define TK_INNER                          149
+#define TK_INTO                           149
-#define TK_SELECT                         150
+#define TK_KILL                           150
-#define TK_DISTINCT                       151
+#define TK_CONNECTION                     151
-#define TK_WHERE                          152
+#define TK_MERGE                          152
-#define TK_PARTITION                      153
+#define TK_VGROUP                         153
-#define TK_BY                             154
+#define TK_REDISTRIBUTE                   154
-#define TK_SESSION                        155
+#define TK_SPLIT                          155
-#define TK_STATE_WINDOW                   156
+#define TK_SYNCDB                         156
-#define TK_SLIDING                        157
+#define TK_NULL                           157
-#define TK_FILL                           158
+#define TK_FIRST                          158
-#define TK_VALUE                          159
+#define TK_LAST                           159
-#define TK_NONE                           160
+#define TK_NOW                            160
-#define TK_PREV                           161
+#define TK_ROWTS                          161
-#define TK_LINEAR                         162
+#define TK_TBNAME                         162
-#define TK_NEXT                           163
+#define TK_QSTARTTS                       163
-#define TK_GROUP                          164
+#define TK_QENDTS                         164
-#define TK_HAVING                         165
+#define TK_WSTARTTS                       165
-#define TK_ORDER                          166
+#define TK_WENDTS                         166
-#define TK_SLIMIT                         167
+#define TK_WDURATION                      167
-#define TK_SOFFSET                        168
+#define TK_BETWEEN                        168
-#define TK_LIMIT                          169
+#define TK_IS                             169
-#define TK_OFFSET                         170
+#define TK_NK_LT                          170
-#define TK_ASC                            171
+#define TK_NK_GT                          171
-#define TK_DESC                           172
+#define TK_NK_LE                          172
-#define TK_NULLS                          173
+#define TK_NK_GE                          173
-#define TK_FIRST                          174
+#define TK_NK_NE                          174
-#define TK_LAST                           175
+#define TK_MATCH                          175
+#define TK_NMATCH                         176
+#define TK_JOIN                           177
+#define TK_INNER                          178
+#define TK_SELECT                         179
+#define TK_DISTINCT                       180
+#define TK_WHERE                          181
+#define TK_PARTITION                      182
+#define TK_BY                             183
+#define TK_SESSION                        184
+#define TK_STATE_WINDOW                   185
+#define TK_SLIDING                        186
+#define TK_FILL                           187
+#define TK_VALUE                          188
+#define TK_NONE                           189
+#define TK_PREV                           190
+#define TK_LINEAR                         191
+#define TK_NEXT                           192
+#define TK_GROUP                          193
+#define TK_HAVING                         194
+#define TK_ORDER                          195
+#define TK_SLIMIT                         196
+#define TK_SOFFSET                        197
+#define TK_LIMIT                          198
+#define TK_OFFSET                         199
+#define TK_ASC                            200
+#define TK_NULLS                          201
 
 #define TK_NK_SPACE                       300
 #define TK_NK_COMMENT                     301
@@ -201,13 +227,10 @@
 #define TK_NK_FILE                        306
 #define TK_NK_QUESTION                    307   // denoting the placeholder of "?",when invoking statement bind query
 
-#define TK_NK_COLON                       500
 #define TK_NK_BITNOT                      501
 #define TK_INSERT                         502
-#define TK_INTO                           503
-#define TK_NOW                            504
 #define TK_VALUES                         507
-#define TK_IMPORT                         507
+#define TK_IMPORT                         509
 #define TK_NK_SEMI                        508
 
 #define TK_NK_NIL                         65535

include/common/ttypes.h

@@ -31,6 +31,7 @@ typedef int16_t  col_id_t;
 typedef int8_t   col_type_t;
 typedef int32_t  col_bytes_t;
 typedef uint16_t schema_ver_t;
+typedef int32_t  func_id_t;
 
 #pragma pack(push, 1)
 typedef struct {
@@ -145,6 +146,7 @@ typedef struct {
 #define IS_FLOAT_TYPE(_t)            ((_t) == TSDB_DATA_TYPE_FLOAT || (_t) == TSDB_DATA_TYPE_DOUBLE)
 
 #define IS_NUMERIC_TYPE(_t) ((IS_SIGNED_NUMERIC_TYPE(_t)) || (IS_UNSIGNED_NUMERIC_TYPE(_t)) || (IS_FLOAT_TYPE(_t)))
+#define IS_MATHABLE_TYPE(_t) (IS_NUMERIC_TYPE(_t) || (_t) == (TSDB_DATA_TYPE_BOOL) || (_t) == (TSDB_DATA_TYPE_TIMESTAMP))
 
 #define IS_VALID_TINYINT(_t)   ((_t) > INT8_MIN && (_t) <= INT8_MAX)
 #define IS_VALID_SMALLINT(_t)  ((_t) > INT16_MIN && (_t) <= INT16_MAX)

include/common/tvariant.h

@@ -59,6 +59,7 @@ int32_t taosVariantDumpEx(SVariant *pVariant, char *payload, int16_t type, bool
 #endif
 
 int32_t taosVariantTypeSetType(SVariant *pVariant, char type);
+char * taosVariantGet(SVariant *pVar, int32_t type);
 
 #ifdef __cplusplus
 }

include/dnode/mnode/mnode.h

@@ -29,8 +29,7 @@ extern "C" {
 typedef struct SMnode SMnode;
 
 typedef struct {
-  int32_t  dnodeId;
+  bool     deploy;
-  int64_t  clusterId;
   int8_t   replica;
   int8_t   selfIndex;
   SReplica replicas[TSDB_MAX_REPLICA];

include/libs/catalog/catalog.h

@@ -77,6 +77,9 @@ typedef struct SDbVgVersion {
   int32_t numOfTable; // unit is TSDB_TABLE_NUM_UNIT
 } SDbVgVersion;
 
+typedef SDbCfgRsp SDbCfgInfo;
+typedef SUserIndexRsp SIndexInfo;
+
 int32_t catalogInit(SCatalogCfg *cfg);
 
 /**
@@ -217,6 +220,10 @@ int32_t catalogGetExpiredSTables(SCatalog* pCatalog, SSTableMetaVersion **stable
 
 int32_t catalogGetExpiredDBs(SCatalog* pCatalog, SDbVgVersion **dbs, uint32_t *num);
 
+int32_t catalogGetDBCfg(SCatalog* pCtg, void *pRpc, const SEpSet* pMgmtEps, const char* dbFName, SDbCfgInfo* pDbCfg);
+
+int32_t catalogGetIndexInfo(SCatalog* pCtg, void *pRpc, const SEpSet* pMgmtEps, const char* indexName, SIndexInfo* pInfo);
+
 
 /**
  * Destroy catalog and relase all resources

include/libs/command/command.h

@@ -13,19 +13,18 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_DND_BNODE_H_
+#include "cmdnodes.h"
-#define _TD_DND_BNODE_H_
+#include "tmsg.h"
+#include "plannodes.h"
 
-#include "dnd.h"
+typedef struct SExplainCtx SExplainCtx;
 
-#ifdef __cplusplus
+int32_t qExecCommand(SNode* pStmt, SRetrieveTableRsp** pRsp);
-extern "C" {
-#endif
 
-void bmGetMgmtFp(SMgmtWrapper *pWrapper);
+int32_t qExecStaticExplain(SQueryPlan *pDag, SRetrieveTableRsp **pRsp);
+int32_t qExecExplainBegin(SQueryPlan *pDag, SExplainCtx **pCtx, int32_t startTs);
+int32_t qExecExplainEnd(SExplainCtx *pCtx, SRetrieveTableRsp **pRsp);
+int32_t qExplainUpdateExecInfo(SExplainCtx        *pCtx, SExplainRsp *pRspMsg, int32_t groupId, SRetrieveTableRsp **pRsp);
+void    qExplainFreeCtx(SExplainCtx *pCtx);
 
-#ifdef __cplusplus
-}
-#endif
 
-#endif /*_TD_DND_BNODE_H_*/

include/libs/executor/executor.h

@@ -174,6 +174,8 @@ void** qReleaseTask(void* pMgmt, void* pQInfo, bool freeHandle);
 
 void qProcessFetchRsp(void* parent, struct SRpcMsg* pMsg, struct SEpSet* pEpSet);
 
+int32_t qGetExplainExecInfo(qTaskInfo_t tinfo, int32_t *resNum, SExplainExecInfo **pRes);
+
 #ifdef __cplusplus
 }
 #endif

include/libs/function/function.h

@@ -36,7 +36,7 @@ typedef struct SFuncExecEnv {
 
 typedef bool (*FExecGetEnv)(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 typedef bool (*FExecInit)(struct SqlFunctionCtx *pCtx, struct SResultRowEntryInfo* pResultCellInfo);
-typedef void (*FExecProcess)(struct SqlFunctionCtx *pCtx);
+typedef int32_t (*FExecProcess)(struct SqlFunctionCtx *pCtx);
 typedef void (*FExecFinalize)(struct SqlFunctionCtx *pCtx);
 typedef int32_t (*FScalarExecProcess)(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 
@@ -52,6 +52,11 @@ typedef struct SFuncExecFuncs {
   FExecFinalize finalize;
 } SFuncExecFuncs;
 
+typedef struct SFileBlockInfo {
+  int32_t numBlocksOfStep;
+} SFileBlockInfo;
+
+#define TSDB_BLOCK_DIST_STEP_ROWS 8
 #define MAX_INTERVAL_TIME_WINDOW  1000000  // maximum allowed time windows in final results
 
 #define FUNCTION_TYPE_SCALAR       1
@@ -101,18 +106,15 @@ typedef struct SFuncExecFuncs {
 #define FUNCTION_DERIVATIVE   32
 #define FUNCTION_BLKINFO      33
 
-#define FUNCTION_HISTOGRAM    34
-#define FUNCTION_HLL          35
-#define FUNCTION_MODE         36
-#define FUNCTION_SAMPLE       37
 
 #define FUNCTION_COV          38
 
 typedef struct SResultRowEntryInfo {
-  int8_t   hasResult;       // result generated, not NULL value
+//  int8_t   hasResult:6;       // result generated, not NULL value
-  bool     initialized;     // output buffer has been initialized
+  bool     initialized:1;     // output buffer has been initialized
-  bool     complete;        // query has completed
+  bool     complete:1;        // query has completed
-  uint32_t numOfRes;        // num of output result in current buffer
+  uint8_t  isNullRes:6;       // the result is null
+  uint8_t  numOfRes;        // num of output result in current buffer
 } SResultRowEntryInfo;
 
 // determine the real data need to calculated the result
@@ -154,6 +156,7 @@ typedef struct SResultDataInfo {
 } SResultDataInfo;
 
 #define GET_RES_INFO(ctx)        ((ctx)->resultInfo)
+#define GET_ROWCELL_INTERBUF(_c) ((void*) ((char*)(_c) + sizeof(SResultRowEntryInfo)))
 
 typedef struct SInputColumnInfoData {
   int32_t           totalRows;      // total rows in current columnar data
@@ -183,15 +186,16 @@ typedef struct SqlFunctionCtx {
   int32_t          columnIndex;  // TODO remove it
   uint8_t          currentStage;  // record current running step, default: 0
   bool             isAggSet;
+  int64_t          startTs;       // timestamp range of current query when function is executed on a specific data block, TODO remove it
   /////////////////////////////////////////////////////////////////
   bool             stableQuery;
   int16_t          functionId;    // function id
   char *           pOutput;       // final result output buffer, point to sdata->data
-  int64_t          startTs;       // timestamp range of current query when function is executed on a specific data block
   int32_t          numOfParams;
   SVariant         param[4];      // input parameter, e.g., top(k, 20), the number of results for top query is kept in param
   int64_t         *ptsList;       // corresponding timestamp array list
-  void            *ptsOutputBuf;  // corresponding output buffer for timestamp of each result, e.g., top/bottom*/
+  SColumnInfoData *pTsOutput;     // corresponding output buffer for timestamp of each result, e.g., top/bottom*/
+  int32_t          offset;
   SVariant         tag;
   struct  SResultRowEntryInfo *resultInfo;
   SSubsidiaryResInfo     subsidiaryRes;

include/libs/function/functionMgt.h

@@ -72,10 +72,15 @@ typedef enum EFunctionType {
   FUNCTION_TYPE_ATAN,
 
   // string function
-  FUNCTION_TYPE_CHAR_LENGTH = 1500,
+  FUNCTION_TYPE_LENGTH = 1500,
+  FUNCTION_TYPE_CHAR_LENGTH,
   FUNCTION_TYPE_CONCAT,
   FUNCTION_TYPE_CONCAT_WS,
-  FUNCTION_TYPE_LENGTH,
+  FUNCTION_TYPE_LOWER,
+  FUNCTION_TYPE_UPPER,
+  FUNCTION_TYPE_LTRIM,
+  FUNCTION_TYPE_RTRIM,
+  FUNCTION_TYPE_SUBSTR,
 
   // conversion function
   FUNCTION_TYPE_CAST = 2000,
@@ -127,6 +132,7 @@ bool fmIsStringFunc(int32_t funcId);
 bool fmIsDatetimeFunc(int32_t funcId);
 bool fmIsTimelineFunc(int32_t funcId);
 bool fmIsTimeorderFunc(int32_t funcId);
+bool fmIsPseudoColumnFunc(int32_t funcId);
 bool fmIsWindowPseudoColumnFunc(int32_t funcId);
 bool fmIsWindowClauseFunc(int32_t funcId);
 

include/libs/monitor/monitor.h

@@ -19,6 +19,7 @@
 #include "tarray.h"
 #include "tdef.h"
 #include "tlog.h"
+#include "tmsg.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -29,6 +30,44 @@ extern "C" {
 #define MON_VER_LEN    12
 #define MON_LOG_LEN    1024
 
+typedef struct {
+  int64_t   ts;
+  ELogLevel level;
+  char      content[MON_LOG_LEN];
+} SMonLogItem;
+
+typedef struct {
+  SArray *logs;  // array of SMonLogItem
+  int32_t numOfErrorLogs;
+  int32_t numOfInfoLogs;
+  int32_t numOfDebugLogs;
+  int32_t numOfTraceLogs;
+} SMonLogs;
+
+typedef struct {
+  char      name[TSDB_FILENAME_LEN];
+  int8_t    level;
+  SDiskSize size;
+} SMonDiskDesc;
+
+typedef struct {
+  double  cpu_engine;
+  double  cpu_system;
+  float   cpu_cores;
+  int64_t mem_engine;     // KB
+  int64_t mem_system;     // KB
+  int64_t mem_total;      // KB
+  int64_t disk_engine;    // Byte
+  int64_t disk_used;      // Byte
+  int64_t disk_total;     // Byte
+  int64_t net_in;         // bytes
+  int64_t net_out;        // bytes
+  int64_t io_read;        // bytes
+  int64_t io_write;       // bytes
+  int64_t io_read_disk;   // bytes
+  int64_t io_write_disk;  // bytes
+} SMonSysInfo;
+
 typedef struct {
   int32_t dnode_id;
   char    dnode_ep[TSDB_EP_LEN];
@@ -36,6 +75,19 @@ typedef struct {
   int32_t protocol;
 } SMonBasicInfo;
 
+typedef struct {
+  float        uptime;  // day
+  int8_t       has_mnode;
+  SMonDiskDesc logdir;
+  SMonDiskDesc tempdir;
+} SMonDnodeInfo;
+
+typedef struct {
+  SMonBasicInfo basic;
+  SMonDnodeInfo dnode;
+  SMonSysInfo   sys;
+} SMonDmInfo;
+
 typedef struct {
   int32_t dnode_id;
   char    dnode_ep[TSDB_EP_LEN];
@@ -87,46 +139,65 @@ typedef struct {
 } SMonGrantInfo;
 
 typedef struct {
-  float   uptime;  // day
+  SMonClusterInfo cluster;
-  double  cpu_engine;
+  SMonVgroupInfo  vgroup;
-  double  cpu_system;
+  SMonGrantInfo   grant;
-  float   cpu_cores;
+  SMonSysInfo     sys;
-  int64_t mem_engine;   // KB
+  SMonLogs        log;
-  int64_t mem_system;   // KB
+} SMonMmInfo;
-  int64_t mem_total;    // KB
-  int64_t disk_engine;  // Byte
-  int64_t disk_used;    // Byte
-  int64_t disk_total;   // Byte
-  int64_t net_in;       // bytes
-  int64_t net_out;      // bytes
-  int64_t io_read;      // bytes
-  int64_t io_write;     // bytes
-  int64_t io_read_disk;   // bytes
-  int64_t io_write_disk;  // bytes
-  int64_t req_select;
-  int64_t req_insert;
-  int64_t req_insert_success;
-  int64_t req_insert_batch;
-  int64_t req_insert_batch_success;
-  int32_t errors;
-  int32_t vnodes_num;
-  int32_t masters;
-  int8_t  has_mnode;
-} SMonDnodeInfo;
 
-typedef struct {
+int32_t tSerializeSMonMmInfo(void *buf, int32_t bufLen, SMonMmInfo *pInfo);
-  char      name[TSDB_FILENAME_LEN];
+int32_t tDeserializeSMonMmInfo(void *buf, int32_t bufLen, SMonMmInfo *pInfo);
-  int8_t    level;
+void    tFreeSMonMmInfo(SMonMmInfo *pInfo);
-  SDiskSize size;
-} SMonDiskDesc;
 
 typedef struct {
   SArray *datadirs;  // array of SMonDiskDesc
-  SMonDiskDesc logdir;
-  SMonDiskDesc tempdir;
 } SMonDiskInfo;
 
-typedef struct SMonInfo SMonInfo;
+typedef struct {
+  SMonDiskInfo tfs;
+  SVnodesStat  vstat;
+  SMonSysInfo  sys;
+  SMonLogs     log;
+} SMonVmInfo;
+
+int32_t tSerializeSMonVmInfo(void *buf, int32_t bufLen, SMonVmInfo *pInfo);
+int32_t tDeserializeSMonVmInfo(void *buf, int32_t bufLen, SMonVmInfo *pInfo);
+void    tFreeSMonVmInfo(SMonVmInfo *pInfo);
+
+typedef struct {
+  SMonSysInfo sys;
+  SMonLogs    log;
+} SMonQmInfo;
+
+int32_t tSerializeSMonQmInfo(void *buf, int32_t bufLen, SMonQmInfo *pInfo);
+int32_t tDeserializeSMonQmInfo(void *buf, int32_t bufLen, SMonQmInfo *pInfo);
+void    tFreeSMonQmInfo(SMonQmInfo *pInfo);
+
+typedef struct {
+  SMonSysInfo sys;
+  SMonLogs    log;
+} SMonSmInfo;
+
+int32_t tSerializeSMonSmInfo(void *buf, int32_t bufLen, SMonSmInfo *pInfo);
+int32_t tDeserializeSMonSmInfo(void *buf, int32_t bufLen, SMonSmInfo *pInfo);
+void    tFreeSMonSmInfo(SMonSmInfo *pInfo);
+typedef struct {
+  SMonSysInfo sys;
+  SMonLogs    log;
+} SMonBmInfo;
+
+int32_t tSerializeSMonBmInfo(void *buf, int32_t bufLen, SMonBmInfo *pInfo);
+int32_t tDeserializeSMonBmInfo(void *buf, int32_t bufLen, SMonBmInfo *pInfo);
+void    tFreeSMonBmInfo(SMonBmInfo *pInfo);
+
+typedef struct {
+  SArray *pVloads;  // SVnodeLoad
+} SMonVloadInfo;
+
+int32_t tSerializeSMonVloadInfo(void *buf, int32_t bufLen, SMonVloadInfo *pInfo);
+int32_t tDeserializeSMonVloadInfo(void *buf, int32_t bufLen, SMonVloadInfo *pInfo);
+void    tFreeSMonVloadInfo(SMonVloadInfo *pInfo);
 
 typedef struct {
   const char *server;
@@ -138,16 +209,14 @@ typedef struct {
 int32_t monInit(const SMonCfg *pCfg);
 void    monCleanup();
 void    monRecordLog(int64_t ts, ELogLevel level, const char *content);
-
+int32_t monGetLogs(SMonLogs *logs);
-SMonInfo *monCreateMonitorInfo();
+void    monSetDmInfo(SMonDmInfo *pInfo);
-void      monSetBasicInfo(SMonInfo *pMonitor, SMonBasicInfo *pInfo);
+void    monSetMmInfo(SMonMmInfo *pInfo);
-void      monSetClusterInfo(SMonInfo *pMonitor, SMonClusterInfo *pInfo);
+void    monSetVmInfo(SMonVmInfo *pInfo);
-void      monSetVgroupInfo(SMonInfo *pMonitor, SMonVgroupInfo *pInfo);
+void    monSetQmInfo(SMonQmInfo *pInfo);
-void      monSetGrantInfo(SMonInfo *pMonitor, SMonGrantInfo *pInfo);
+void    monSetSmInfo(SMonSmInfo *pInfo);
-void      monSetDnodeInfo(SMonInfo *pMonitor, SMonDnodeInfo *pInfo);
+void    monSetBmInfo(SMonBmInfo *pInfo);
-void      monSetDiskInfo(SMonInfo *pMonitor, SMonDiskInfo *pInfo);
+void    monSendReport();
-void      monSendReport(SMonInfo *pMonitor);
-void      monCleanupMonitorInfo(SMonInfo *pMonitor);
 
 #ifdef __cplusplus
 }

include/libs/nodes/cmdnodes.h

@@ -20,27 +20,33 @@
 extern "C" {
 #endif
 
+#include "query.h"
 #include "querynodes.h"
 
+#define DESCRIBE_RESULT_COLS 4
+#define DESCRIBE_RESULT_FIELD_LEN (TSDB_COL_NAME_LEN - 1 + VARSTR_HEADER_SIZE)
+#define DESCRIBE_RESULT_TYPE_LEN (20 + VARSTR_HEADER_SIZE)
+#define DESCRIBE_RESULT_NOTE_LEN (8 + VARSTR_HEADER_SIZE)
+
 typedef struct SDatabaseOptions {
   ENodeType type;
-  int32_t numOfBlocks;
+  SValueNode* pNumOfBlocks;
-  int32_t cacheBlockSize;
+  SValueNode* pCacheBlockSize;
-  int8_t cachelast;
+  SValueNode* pCachelast;
-  int32_t compressionLevel;
+  SValueNode* pCompressionLevel;
-  int32_t daysPerFile;
+  SValueNode* pDaysPerFile;
-  int32_t fsyncPeriod;
+  SValueNode* pFsyncPeriod;
-  int32_t maxRowsPerBlock;
+  SValueNode* pMaxRowsPerBlock;
-  int32_t minRowsPerBlock;
+  SValueNode* pMinRowsPerBlock;
-  int32_t keep;
+  SNodeList* pKeep;
-  int32_t precision;
+  SValueNode* pPrecision;
-  int32_t quorum;
+  SValueNode* pQuorum;
-  int32_t replica;
+  SValueNode* pReplica;
-  int32_t ttl;
+  SValueNode* pTtl;
-  int32_t walLevel;
+  SValueNode* pWalLevel;
-  int32_t numOfVgroups;
+  SValueNode* pNumOfVgroups;
-  int8_t singleStable;
+  SValueNode* pSingleStable;
-  int8_t streamMode;
+  SValueNode* pStreamMode;
   SNodeList* pRetentions;
 } SDatabaseOptions;
 
@@ -70,13 +76,13 @@ typedef struct SAlterDatabaseStmt {
 
 typedef struct STableOptions {
   ENodeType type;
-  int32_t keep;
+  SNodeList* pKeep;
-  int32_t ttl;
+  SValueNode* pTtl;
-  char comments[TSDB_STB_COMMENT_LEN];
+  SValueNode* pComments;
   SNodeList* pSma;
   SNodeList* pFuncs;
-  float filesFactor;
+  SValueNode* pFilesFactor;
-  int32_t delay;
+  SValueNode* pDelay;
 } STableOptions;
 
 typedef struct SColumnDefNode {
@@ -84,6 +90,7 @@ typedef struct SColumnDefNode {
   char colName[TSDB_COL_NAME_LEN];
   SDataType dataType;
   char comments[TSDB_STB_COMMENT_LEN];
+  bool sma;
 } SColumnDefNode;
 
 typedef struct SCreateTableStmt {
@@ -187,6 +194,12 @@ typedef struct SShowStmt {
   SNode* pTbNamePattern; // SValueNode
 } SShowStmt;
 
+typedef struct SShowCreatStmt {
+  ENodeType type;
+  char dbName[TSDB_DB_NAME_LEN];
+  char tableName[TSDB_TABLE_NAME_LEN];
+} SShowCreatStmt;
+
 typedef enum EIndexType {
   INDEX_TYPE_SMA = 1,
   INDEX_TYPE_FULLTEXT
@@ -247,6 +260,13 @@ typedef struct SAlterLocalStmt {
   char value[TSDB_DNODE_VALUE_LEN];
 } SAlterLocalStmt;
 
+typedef struct SDescribeStmt {
+  ENodeType type;
+  char dbName[TSDB_DB_NAME_LEN];
+  char tableName[TSDB_TABLE_NAME_LEN];
+  STableMeta* pMeta;
+} SDescribeStmt;
+
 #ifdef __cplusplus
 }
 #endif

include/libs/nodes/nodes.h

@@ -30,11 +30,19 @@ extern "C" {
 #define FOREACH(node, list)	\
   for (SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); (NULL != cell ? (node = cell->pNode, true) : (node = NULL, false)); cell = cell->pNext)
 
-// only be use in FOREACH
-#define ERASE_NODE(list) cell = nodesListErase(list, cell);
-
 #define REPLACE_NODE(newNode) cell->pNode = (SNode*)(newNode)
 
+#define INSERT_LIST(target, src) nodesListInsertList((target), cell, src)
+
+#define WHERE_EACH(node, list) \
+  SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); \
+  while (NULL != cell ? (node = cell->pNode, true) : (node = NULL, false))
+
+#define WHERE_NEXT cell = cell->pNext
+
+// only be use in WHERE_EACH
+#define ERASE_NODE(list) cell = nodesListErase((list), cell)
+
 #define FORBOTH(node1, list1, node2, list2) \
   for (SListCell* cell1 = (NULL != (list1) ? (list1)->pHead : NULL), *cell2 = (NULL != (list2) ? (list2)->pHead : NULL); \
     (NULL == cell1 ? (node1 = NULL, false) : (node1 = cell1->pNode, true)), (NULL == cell2 ? (node2 = NULL, false) : (node2 = cell2->pNode, true)), (node1 != NULL && node2 != NULL); \
@@ -101,6 +109,17 @@ typedef enum ENodeType {
   QUERY_NODE_DROP_TOPIC_STMT,
   QUERY_NODE_ALTER_LOCAL_STMT,
   QUERY_NODE_EXPLAIN_STMT,
+  QUERY_NODE_DESCRIBE_STMT,
+  QUERY_NODE_RESET_QUERY_CACHE_STMT,
+  QUERY_NODE_COMPACT_STMT,
+  QUERY_NODE_CREATE_FUNCTION_STMT,
+  QUERY_NODE_DROP_FUNCTION_STMT,
+  QUERY_NODE_CREATE_STREAM_STMT,
+  QUERY_NODE_DROP_STREAM_STMT,
+  QUERY_NODE_MERGE_VGROUP_STMT,
+  QUERY_NODE_REDISTRIBUTE_VGROUP_STMT,
+  QUERY_NODE_SPLIT_VGROUP_STMT,
+  QUERY_NODE_SYNCDB_STMT,
   QUERY_NODE_SHOW_DATABASES_STMT,
   QUERY_NODE_SHOW_TABLES_STMT,
   QUERY_NODE_SHOW_STABLES_STMT,
@@ -113,6 +132,18 @@ typedef enum ENodeType {
   QUERY_NODE_SHOW_FUNCTIONS_STMT,
   QUERY_NODE_SHOW_INDEXES_STMT,
   QUERY_NODE_SHOW_STREAMS_STMT,
+  QUERY_NODE_SHOW_APPS_STMT,
+  QUERY_NODE_SHOW_CONNECTIONS_STMT,
+  QUERY_NODE_SHOW_LICENCE_STMT,
+  QUERY_NODE_SHOW_CREATE_DATABASE_STMT,
+  QUERY_NODE_SHOW_CREATE_TABLE_STMT,
+  QUERY_NODE_SHOW_CREATE_STABLE_STMT, 
+  QUERY_NODE_SHOW_QUERIES_STMT,
+  QUERY_NODE_SHOW_SCORES_STMT,
+  QUERY_NODE_SHOW_TOPICS_STMT,
+  QUERY_NODE_SHOW_VARIABLE_STMT,
+  QUERY_NODE_KILL_CONNECTION_STMT,
+  QUERY_NODE_KILL_QUERY_STMT,
 
   // logic plan node
   QUERY_NODE_LOGIC_PLAN_SCAN,
@@ -140,6 +171,8 @@ typedef enum ENodeType {
   QUERY_NODE_PHYSICAL_PLAN_SORT,
   QUERY_NODE_PHYSICAL_PLAN_INTERVAL,
   QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW,
+  QUERY_NODE_PHYSICAL_PLAN_STATE_WINDOW,
+  QUERY_NODE_PHYSICAL_PLAN_PARTITION,
   QUERY_NODE_PHYSICAL_PLAN_DISPATCH,
   QUERY_NODE_PHYSICAL_PLAN_INSERT,
   QUERY_NODE_PHYSICAL_SUBPLAN,
@@ -178,7 +211,9 @@ int32_t nodesListStrictAppend(SNodeList* pList, SNodeptr pNode);
 int32_t nodesListMakeAppend(SNodeList** pList, SNodeptr pNode);
 int32_t nodesListAppendList(SNodeList* pTarget, SNodeList* pSrc);
 int32_t nodesListStrictAppendList(SNodeList* pTarget, SNodeList* pSrc);
+int32_t nodesListPushFront(SNodeList* pList, SNodeptr pNode);
 SListCell* nodesListErase(SNodeList* pList, SListCell* pCell);
+void nodesListInsertList(SNodeList* pTarget, SListCell* pPos, SNodeList* pSrc);
 SNodeptr nodesListGetNode(SNodeList* pList, int32_t index);
 void nodesDestroyList(SNodeList* pList);
 // Only clear the linked list structure, without releasing the elements inside
@@ -191,16 +226,16 @@ typedef enum EDealRes {
 } EDealRes;
 
 typedef EDealRes (*FNodeWalker)(SNode* pNode, void* pContext);
-void nodesWalkNode(SNodeptr pNode, FNodeWalker walker, void* pContext);
+void nodesWalkExpr(SNodeptr pNode, FNodeWalker walker, void* pContext);
-void nodesWalkList(SNodeList* pList, FNodeWalker walker, void* pContext);
+void nodesWalkExprs(SNodeList* pList, FNodeWalker walker, void* pContext);
-void nodesWalkNodePostOrder(SNodeptr pNode, FNodeWalker walker, void* pContext);
+void nodesWalkExprPostOrder(SNodeptr pNode, FNodeWalker walker, void* pContext);
-void nodesWalkListPostOrder(SNodeList* pList, FNodeWalker walker, void* pContext);
+void nodesWalkExprsPostOrder(SNodeList* pList, FNodeWalker walker, void* pContext);
 
 typedef EDealRes (*FNodeRewriter)(SNode** pNode, void* pContext);
-void nodesRewriteNode(SNode** pNode, FNodeRewriter rewriter, void* pContext);
+void nodesRewriteExpr(SNode** pNode, FNodeRewriter rewriter, void* pContext);
-void nodesRewriteList(SNodeList* pList, FNodeRewriter rewriter, void* pContext);
+void nodesRewriteExprs(SNodeList* pList, FNodeRewriter rewriter, void* pContext);
-void nodesRewriteNodePostOrder(SNode** pNode, FNodeRewriter rewriter, void* pContext);
+void nodesRewriteExprPostOrder(SNode** pNode, FNodeRewriter rewriter, void* pContext);
-void nodesRewriteListPostOrder(SNodeList* pList, FNodeRewriter rewriter, void* pContext);
+void nodesRewriteExprsPostOrder(SNodeList* pList, FNodeRewriter rewriter, void* pContext);
 
 bool nodesEqualNode(const SNodeptr a, const SNodeptr b);
 
@@ -214,6 +249,10 @@ int32_t nodesStringToNode(const char* pStr, SNode** pNode);
 int32_t nodesListToString(const SNodeList* pList, bool format, char** pStr, int32_t* pLen);
 int32_t nodesStringToList(const char* pStr, SNodeList** pList);
 
+int32_t nodesNodeToSQL(SNode *pNode, char *buf, int32_t bufSize, int32_t *len);
+char *nodesGetNameFromColumnNode(SNode *pNode);
+int32_t nodesGetOutputNumFromSlotList(SNodeList* pSlots);
+
 #ifdef __cplusplus
 }
 #endif

include/libs/nodes/plannodes.h

@@ -104,6 +104,7 @@ typedef struct SWindowLogicNode {
   SFillNode* pFill;
   int64_t sessionGap;
   SNode* pTspk;
+  SNode* pStateExpr;
 } SWindowLogicNode;
 
 typedef struct SSortLogicNode {
@@ -194,11 +195,20 @@ typedef struct SSystemTableScanPhysiNode {
   int32_t accountId;
 } SSystemTableScanPhysiNode;
 
+typedef enum EScanRequired {
+  SCAN_REQUIRED_DATA_NO_NEEDED = 1,
+  SCAN_REQUIRED_DATA_STATIS_NEEDED,
+  SCAN_REQUIRED_DATA_ALL_NEEDED,
+  SCAN_REQUIRED_DATA_DISCARD,
+} EScanRequired;
+
 typedef struct STableScanPhysiNode {
   SScanPhysiNode scan;
   uint8_t scanFlag;         // denotes reversed scan of data or not
   STimeWindow scanRange;
   double ratio;
+  EScanRequired scanRequired;
+  SNodeList* pScanReferFuncs;
 } STableScanPhysiNode;
 
 typedef STableScanPhysiNode STableSeqScanPhysiNode;
@@ -257,17 +267,35 @@ typedef struct SIntervalPhysiNode {
   SFillNode* pFill;
 } SIntervalPhysiNode;
 
+typedef struct SMultiTableIntervalPhysiNode {
+  SIntervalPhysiNode interval;
+  SNodeList* pPartitionKeys;
+} SMultiTableIntervalPhysiNode;
+
 typedef struct SSessionWinodwPhysiNode {
   SWinodwPhysiNode window;
   int64_t    gap;
 } SSessionWinodwPhysiNode;
 
+typedef struct SStateWinodwPhysiNode {
+  SWinodwPhysiNode window;
+  SNode* pStateKey;
+} SStateWinodwPhysiNode;
+
 typedef struct SSortPhysiNode {
   SPhysiNode node;
   SNodeList* pExprs;   // these are expression list of order_by_clause and parameter expression of aggregate function
   SNodeList* pSortKeys; // element is SOrderByExprNode, and SOrderByExprNode::pExpr is SColumnNode
+  SNodeList* pTargets;
 } SSortPhysiNode;
 
+typedef struct SPartitionPhysiNode {
+  SPhysiNode node;
+  SNodeList* pExprs;   // these are expression list of partition_by_clause
+  SNodeList* pPartitionKeys;
+  SNodeList* pTargets;
+} SPartitionPhysiNode;
+
 typedef struct SDataSinkNode {
   ENodeType type;
   SDataBlockDescNode* pInputDataBlockDesc;
@@ -308,6 +336,7 @@ typedef enum EExplainMode {
 typedef struct SExplainInfo {
   EExplainMode mode;
   bool verbose;
+  double ratio;
 } SExplainInfo;
 
 typedef struct SQueryPlan {

include/libs/nodes/querynodes.h

@@ -22,6 +22,7 @@ extern "C" {
 
 #include "nodes.h"
 #include "tmsg.h"
+#include "tvariant.h"
 
 #define TABLE_TOTAL_COL_NUM(pMeta) ((pMeta)->tableInfo.numOfColumns + (pMeta)->tableInfo.numOfTags)
 #define TABLE_META_SIZE(pMeta) (NULL == (pMeta) ? 0 : (sizeof(STableMeta) + TABLE_TOTAL_COL_NUM((pMeta)) * sizeof(SSchema)))
@@ -79,6 +80,7 @@ typedef struct SValueNode {
   char* literal;
   bool isDuration;
   bool translate;
+  bool genByCalc;
   union {
     bool b;
     int64_t i;
@@ -122,6 +124,7 @@ typedef struct STableNode {
   char dbName[TSDB_DB_NAME_LEN];
   char tableName[TSDB_TABLE_NAME_LEN];
   char tableAlias[TSDB_TABLE_NAME_LEN];
+  uint8_t precision;
 } STableNode;
 
 struct STableMeta;
@@ -187,7 +190,7 @@ typedef struct SLimitNode {
 
 typedef struct SStateWindowNode {
   ENodeType type; // QUERY_NODE_STATE_WINDOW
-  SNode* pCol;
+  SNode* pExpr;
 } SStateWindowNode;
 
 typedef struct SSessionWindowNode {
@@ -234,6 +237,8 @@ typedef struct SSelectStmt {
   SNode* pLimit;
   SNode* pSlimit;
   char stmtName[TSDB_TABLE_NAME_LEN];
+  uint8_t precision;
+  bool isEmptyResult;
 } SSelectStmt;
 
 typedef enum ESetOperatorType {
@@ -256,6 +261,7 @@ typedef enum ESqlClause {
   SQL_CLAUSE_WINDOW,
   SQL_CLAUSE_GROUP_BY,
   SQL_CLAUSE_HAVING,
+  SQL_CLAUSE_DISTINCT,
   SQL_CLAUSE_SELECT,
   SQL_CLAUSE_ORDER_BY
 } ESqlClause;
@@ -277,7 +283,6 @@ typedef struct SVnodeModifOpStmt {
   ENodeType   nodeType;
   ENodeType   sqlNodeType;
   SArray*     pDataBlocks;         // data block for each vgroup, SArray<SVgDataBlocks*>.
-  int8_t      schemaAttache;       // denote if submit block is built with table schema or not
   uint8_t     payloadType;         // EPayloadType. 0: K-V payload for non-prepare insert, 1: rawPayload for prepare insert
   uint32_t    insertType;          // insert data from [file|sql statement| bound statement]
   const char* sql;                 // current sql statement position
@@ -306,6 +311,7 @@ int32_t nodesCollectFuncs(SSelectStmt* pSelect, FFuncClassifier classifier, SNod
 
 bool nodesIsExprNode(const SNode* pNode);
 
+bool nodesIsUnaryOp(const SOperatorNode* pOp);
 bool nodesIsArithmeticOp(const SOperatorNode* pOp);
 bool nodesIsComparisonOp(const SOperatorNode* pOp);
 bool nodesIsJsonOp(const SOperatorNode* pOp);
@@ -314,6 +320,9 @@ bool nodesIsTimeorderQuery(const SNode* pQuery);
 bool nodesIsTimelineQuery(const SNode* pQuery);
 
 void* nodesGetValueFromNode(SValueNode *pNode);
+char* nodesGetStrValueFromNode(SValueNode *pNode);
+char *getFillModeString(EFillMode mode);
+void valueNodeToVariant(const SValueNode* pNode, SVariant* pVal);
 
 #ifdef __cplusplus
 }

include/libs/parser/parser.h

@@ -44,8 +44,15 @@ typedef struct SCmdMsgInfo {
   void* pExtension;  // todo remove it soon
 } SCmdMsgInfo;
 
+typedef enum EQueryExecMode {
+  QUERY_EXEC_MODE_LOCAL = 1,
+  QUERY_EXEC_MODE_RPC,
+  QUERY_EXEC_MODE_SCHEDULE,
+  QUERY_EXEC_MODE_EMPTY_RESULT
+} EQueryExecMode;
+
 typedef struct SQuery {
-  bool directRpc;
+  EQueryExecMode execMode;
   bool haveResultSet;
   SNode* pRoot;
   int32_t numOfResCols;

include/libs/qcom/query.h

@@ -53,6 +53,7 @@ typedef struct SIndexMeta {
 
 } SIndexMeta;
 
+
 /*
  * ASSERT(sizeof(SCTableMeta) == 24)
  * ASSERT(tableType == TSDB_CHILD_TABLE)
@@ -235,6 +236,11 @@ extern int32_t (*queryProcessMsgRsp[TDMT_MAX])(void* output, char* msg, int32_t
     }                                                                    \
   } while (0)
 
+#define QRY_ERR_RET(c) do { int32_t _code = c; if (_code != TSDB_CODE_SUCCESS) { terrno = _code; return _code; } } while (0)
+#define QRY_RET(c) do { int32_t _code = c; if (_code != TSDB_CODE_SUCCESS) { terrno = _code; } return _code; } while (0)
+#define QRY_ERR_JRET(c) do { code = c; if (code != TSDB_CODE_SUCCESS) { terrno = code; goto _return; } } while (0)
+
+
 #ifdef __cplusplus
 }
 #endif

include/libs/scalar/filter.h

@@ -40,7 +40,7 @@ extern int32_t filterInitFromNode(SNode *pNode, SFilterInfo **pinfo, uint32_t op
 extern bool filterExecute(SFilterInfo *info, SSDataBlock *pSrc, int8_t** p, SColumnDataAgg *statis, int16_t numOfCols);
 extern int32_t filterSetDataFromSlotId(SFilterInfo *info, void *param);
 extern int32_t filterSetDataFromColId(SFilterInfo *info, void *param);
-extern int32_t filterGetTimeRange(SFilterInfo *info, STimeWindow *win);
+extern int32_t filterGetTimeRange(SNode *pNode, STimeWindow *win, bool *isStrict);
 extern int32_t filterConverNcharColumns(SFilterInfo* pFilterInfo, int32_t rows, bool *gotNchar);
 extern int32_t filterFreeNcharColumns(SFilterInfo* pFilterInfo);
 extern void filterFreeInfo(SFilterInfo *info);

include/libs/scalar/scalar.h

@@ -42,6 +42,7 @@ int32_t scalarGenerateSetFromList(void **data, void *pNode, uint32_t type);
 int32_t vectorGetConvertType(int32_t type1, int32_t type2);
 int32_t vectorConvertImpl(const SScalarParam* pIn, SScalarParam* pOut);
 
+/* Math functions */
 int32_t absFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t logFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t powFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
@@ -58,6 +59,17 @@ int32_t ceilFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
 int32_t floorFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t roundFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 
+/* String functions */
+int32_t lengthFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t charLengthFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t concatFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t concatWsFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t lowerFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t upperFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t ltrimFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t rtrimFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t substrFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+
 bool getTimePseudoFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 
 int32_t winStartTsFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);

include/libs/scheduler/scheduler.h

@@ -71,7 +71,7 @@ int32_t schedulerInit(SSchedulerCfg *cfg);
  * @param nodeList  Qnode/Vnode address list, element is SQueryNodeAddr
  * @return
  */
-int32_t schedulerExecJob(void *transport, SArray *nodeList, SQueryPlan* pDag, int64_t* pJob, const char* sql, SQueryResult *pRes);
+int32_t schedulerExecJob(void *transport, SArray *nodeList, SQueryPlan *pDag, int64_t *pJob, const char *sql, int64_t startTs, SQueryResult *pRes);
 
 /**
  * Process the query job, generated according to the query physical plan.

include/libs/transport/trpc.h

@@ -58,6 +58,9 @@ typedef struct {
   void    *pNode;
 } SNodeMsg;
 
+typedef void (*RpcCfp)(void *parent, SRpcMsg *, SEpSet *);
+typedef int (*RpcAfp)(void *parent, char *tableId, char *spi, char *encrypt, char *secret, char *ckey);
+
 typedef struct SRpcInit {
   uint16_t localPort;     // local port
   char *   label;         // for debug purpose
@@ -74,10 +77,10 @@ typedef struct SRpcInit {
   char *ckey;     // ciphering key
 
   // call back to process incoming msg, code shall be ignored by server app
-  void (*cfp)(void *parent, SRpcMsg *, SEpSet *);
+  RpcCfp cfp;
 
   // call back to retrieve the client auth info, for server app only
-  int (*afp)(void *parent, char *tableId, char *spi, char *encrypt, char *secret, char *ckey);
+  RpcAfp afp;;
 
   void *parent;
 } SRpcInit;

include/libs/wal/wal.h

@@ -157,6 +157,7 @@ typedef struct SWalReadHandle {
   int64_t       curVersion;
   int64_t       capacity;
   int64_t       status;  // if cursor valid
+  TdThreadMutex mutex;
   SWalHead     *pHead;
 } SWalReadHandle;
 #pragma pack(pop)
@@ -191,6 +192,7 @@ int32_t walEndSnapshot(SWal *);
 SWalReadHandle *walOpenReadHandle(SWal *);
 void            walCloseReadHandle(SWalReadHandle *);
 int32_t         walReadWithHandle(SWalReadHandle *pRead, int64_t ver);
+int32_t         walReadWithHandle_s(SWalReadHandle *pRead, int64_t ver, SWalReadHead **ppHead);
 
 // deprecated
 #if 0

include/os/osMath.h

@@ -56,6 +56,12 @@ extern "C" {
       __typeof(b) __b = (b);     \
       (__a < __b) ? __a : __b;   \
     })
+
+#define TRANGE(a, b, c) \
+  ({                    \
+    a = TMAX(a, b);     \
+    a = TMIN(a, c);     \
+  })
 #endif
 
 #ifdef __cplusplus

include/os/osMemory.h

@@ -32,13 +32,16 @@ extern "C" {
 void *taosMemoryMalloc(int32_t size);
 void *taosMemoryCalloc(int32_t num, int32_t size);
 void *taosMemoryRealloc(void *ptr, int32_t size);
+void *taosMemoryStrDup(void *ptr);
 void taosMemoryFree(const void *ptr);
 int32_t taosMemorySize(void *ptr);
 
 #define taosMemoryFreeClear(ptr) \
   do {                           \
+    if (ptr) {                   \
       taosMemoryFree(ptr);       \
       (ptr) = NULL;              \
+    }                            \
   } while (0)
 
 #ifdef __cplusplus

include/os/osProc.h

@@ -20,11 +20,12 @@
 extern "C" {
 #endif
 
-// start a copy of itself
+int32_t taosNewProc(char **args);
-int32_t taosNewProc(const char *args);
+void    taosWaitProc(int32_t pid);
-
+void    taosKillProc(int32_t pid);
-// the length of the new name must be less than the original name to take effect
+bool    taosProcExist(int32_t pid);
-void taosSetProcName(char **argv, const char *name);
+void    taosSetProcName(int32_t argc, char **argv, const char *name);
+void    taosSetProcPath(int32_t argc, char **argv);
 
 #ifdef __cplusplus
 }

include/os/osShm.h

@@ -26,7 +26,7 @@ typedef struct {
   void*   ptr;
 } SShm;
 
-int32_t taosCreateShm(SShm *pShm, int32_t shmsize) ;
+int32_t taosCreateShm(SShm *pShm, int32_t key, int32_t shmsize) ;
 void    taosDropShm(SShm *pShm);
 int32_t taosAttachShm(SShm *pShm);
 

include/os/osSignal.h

@@ -49,7 +49,7 @@ void taosSetSignal(int32_t signum, FSignalHandler sigfp);
 void taosIgnSignal(int32_t signum);
 void taosDflSignal(int32_t signum);
 
-void taosKillChildOnSelfStopped();
+void taosKillChildOnParentStopped();
 
 #ifdef __cplusplus
 }

include/os/osSysinfo.h

@@ -44,8 +44,8 @@ int32_t taosGetTotalMemory(int64_t *totalKB);
 int32_t taosGetProcMemory(int64_t *usedKB);
 int32_t taosGetSysMemory(int64_t *usedKB);
 int32_t taosGetDiskSize(char *dataDir, SDiskSize *diskSize);
-int32_t taosGetProcIO(int64_t *rchars, int64_t *wchars, int64_t *read_bytes, int64_t *write_bytes);
+void    taosGetProcIODelta(int64_t *rchars, int64_t *wchars, int64_t *read_bytes, int64_t *write_bytes);
-int32_t taosGetCardInfo(int64_t *receive_bytes, int64_t *transmit_bytes);
+void    taosGetCardInfoDelta(int64_t *receive_bytes, int64_t *transmit_bytes);
 
 void    taosKillSystem();
 int32_t taosGetSystemUUID(char *uid, int32_t uidlen);

include/os/osTime.h

@@ -46,6 +46,14 @@ extern "C" {
 #define MILLISECOND_PER_DAY    (MILLISECOND_PER_HOUR * 24)
 #define MILLISECOND_PER_WEEK   (MILLISECOND_PER_DAY * 7)
 
+#define NANOSECOND_PER_USEC   (1000L)
+#define NANOSECOND_PER_MSEC   (1000000L)
+#define NANOSECOND_PER_SEC    (1000000000L)
+#define NANOSECOND_PER_MINUTE (NANOSECOND_PER_SEC * 60)
+#define NANOSECOND_PER_HOUR   (NANOSECOND_PER_MINUTE * 60)
+#define NANOSECOND_PER_DAY    (NANOSECOND_PER_HOUR * 24)
+#define NANOSECOND_PER_WEEK   (NANOSECOND_PER_DAY * 7)
+
 int32_t taosGetTimeOfDay(struct timeval *tv);
 
 //@return timestamp in second

include/util/taoserror.h

@@ -217,6 +217,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_INVALID_DB_OPTION         TAOS_DEF_ERROR_CODE(0, 0x0385)
 #define TSDB_CODE_MND_INVALID_DB_ACCT           TAOS_DEF_ERROR_CODE(0, 0x0386)
 #define TSDB_CODE_MND_DB_OPTION_UNCHANGED       TAOS_DEF_ERROR_CODE(0, 0x0387)
+#define TSDB_CODE_MND_DB_INDEX_NOT_EXIST        TAOS_DEF_ERROR_CODE(0, 0x0388)
 
 // mnode-vgroup
 #define TSDB_CODE_MND_VGROUP_NOT_EXIST          TAOS_DEF_ERROR_CODE(0, 0x0390)
@@ -479,6 +480,24 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_INVALID_ENDPOINT          TAOS_DEF_ERROR_CODE(0, 0x2613)
 #define TSDB_CODE_PAR_EXPRIE_STATEMENT          TAOS_DEF_ERROR_CODE(0, 0x2614)
 #define TSDB_CODE_PAR_INTERVAL_VALUE_TOO_SMALL  TAOS_DEF_ERROR_CODE(0, 0x2615)
+#define TSDB_CODE_PAR_DB_NOT_SPECIFIED          TAOS_DEF_ERROR_CODE(0, 0x2616)
+#define TSDB_CODE_PAR_INVALID_IDENTIFIER_NAME   TAOS_DEF_ERROR_CODE(0, 0x2617)
+#define TSDB_CODE_PAR_CORRESPONDING_STABLE_ERR  TAOS_DEF_ERROR_CODE(0, 0x2618)
+#define TSDB_CODE_PAR_INVALID_RANGE_OPTION      TAOS_DEF_ERROR_CODE(0, 0x2619)
+#define TSDB_CODE_PAR_INVALID_STR_OPTION        TAOS_DEF_ERROR_CODE(0, 0x261A)
+#define TSDB_CODE_PAR_INVALID_ENUM_OPTION       TAOS_DEF_ERROR_CODE(0, 0x261B)
+#define TSDB_CODE_PAR_INVALID_TTL_OPTION        TAOS_DEF_ERROR_CODE(0, 0x261C)
+#define TSDB_CODE_PAR_INVALID_KEEP_NUM          TAOS_DEF_ERROR_CODE(0, 0x261D)
+#define TSDB_CODE_PAR_INVALID_KEEP_ORDER        TAOS_DEF_ERROR_CODE(0, 0x261E)
+#define TSDB_CODE_PAR_INVALID_KEEP_VALUE        TAOS_DEF_ERROR_CODE(0, 0x261F)
+#define TSDB_CODE_PAR_INVALID_COMMENT_OPTION    TAOS_DEF_ERROR_CODE(0, 0x2620)
+#define TSDB_CODE_PAR_INVALID_F_RANGE_OPTION    TAOS_DEF_ERROR_CODE(0, 0x2621)
+#define TSDB_CODE_PAR_INVALID_ROLLUP_OPTION     TAOS_DEF_ERROR_CODE(0, 0x2622)
+#define TSDB_CODE_PAR_INVALID_RETENTIONS_OPTION TAOS_DEF_ERROR_CODE(0, 0x2623)
+#define TSDB_CODE_PAR_GROUPBY_WINDOW_COEXIST    TAOS_DEF_ERROR_CODE(0, 0x2624)
+
+//planner
+#define TSDB_CODE_PLAN_INTERNAL_ERROR              TAOS_DEF_ERROR_CODE(0, 0x2700)
 
 #ifdef __cplusplus
 }

include/util/tcoding.h

@@ -57,6 +57,8 @@ static FORCE_INLINE void *taosDecodeFixedI8(const void *buf, int8_t *value) {
   return POINTER_SHIFT(buf, sizeof(*value));
 }
 
+static FORCE_INLINE void *taosSkipFixedLen(const void *buf, size_t len) { return POINTER_SHIFT(buf, len); }
+
 // ---- Fixed U16
 static FORCE_INLINE int32_t taosEncodeFixedU16(void **buf, uint16_t value) {
   if (buf != NULL) {

include/util/tdef.h

@@ -109,6 +109,9 @@ extern const int32_t TYPE_BYTES[15];
 #define TSDB_INS_TABLE_USER_USERS             "user_users"
 #define TSDB_INS_TABLE_VGROUPS                "vgroups"
 
+#define TSDB_INDEX_TYPE_SMA      "SMA"
+#define TSDB_INDEX_TYPE_FULLTEXT "FULLTEXT"
+
 #define TSDB_INS_USER_STABLES_DBNAME_COLID    2
 
 #define TSDB_TICK_PER_SECOND(precision)                      \
@@ -128,18 +131,20 @@ extern const int32_t TYPE_BYTES[15];
   } while (0)
 
 typedef enum EOperatorType {
-  // arithmetic operator
+  // binary arithmetic operator
   OP_TYPE_ADD = 1,
   OP_TYPE_SUB,
   OP_TYPE_MULTI,
   OP_TYPE_DIV,
   OP_TYPE_MOD,
+  // unary arithmetic operator
+  OP_TYPE_MINUS,
 
   // bit operator
   OP_TYPE_BIT_AND,
   OP_TYPE_BIT_OR,
 
-  // comparison operator
+  // binary comparison operator
   OP_TYPE_GREATER_THAN,
   OP_TYPE_GREATER_EQUAL,
   OP_TYPE_LOWER_THAN,
@@ -152,6 +157,7 @@ typedef enum EOperatorType {
   OP_TYPE_NOT_LIKE,
   OP_TYPE_MATCH,
   OP_TYPE_NMATCH,
+  // unary comparison operator
   OP_TYPE_IS_NULL,
   OP_TYPE_IS_NOT_NULL,
   OP_TYPE_IS_TRUE,
@@ -210,6 +216,9 @@ typedef enum ELogicConditionType {
 #define TSDB_FUNC_MAX_RETRIEVE   1024
 
 #define TSDB_INDEX_NAME_LEN      65  // 64 + 1 '\0'
+#define TSDB_INDEX_TYPE_LEN      10
+#define TSDB_INDEX_EXTS_LEN      256
+#define TSDB_INDEX_FNAME_LEN     (TSDB_DB_FNAME_LEN + TSDB_INDEX_NAME_LEN + TSDB_NAME_DELIMITER_LEN)
 #define TSDB_TYPE_STR_MAX_LEN    32
 #define TSDB_TABLE_FNAME_LEN     (TSDB_DB_FNAME_LEN + TSDB_TABLE_NAME_LEN + TSDB_NAME_DELIMITER_LEN)
 #define TSDB_TOPIC_FNAME_LEN     TSDB_TABLE_FNAME_LEN
@@ -224,6 +233,7 @@ typedef enum ELogicConditionType {
 
 #define TSDB_APP_NAME_LEN    TSDB_UNI_LEN
 #define TSDB_STB_COMMENT_LEN 1024
+
 /**
  *  In some scenarios uint16_t (0~65535) is used to store the row len.
  *  - Firstly, we use 65531(65535 - 4), as the SDataRow/SKVRow contains 4 bits header.
@@ -303,13 +313,13 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_TOTAL_BLOCKS     10000
 #define TSDB_DEFAULT_TOTAL_BLOCKS 6
 
-#define TSDB_MIN_DAYS_PER_FILE     1
+#define TSDB_MIN_DAYS_PER_FILE     60    // unit minute
-#define TSDB_MAX_DAYS_PER_FILE     3650
+#define TSDB_MAX_DAYS_PER_FILE     (3650 * 1440)
-#define TSDB_DEFAULT_DAYS_PER_FILE 10
+#define TSDB_DEFAULT_DAYS_PER_FILE (10 * 1440)
 
-#define TSDB_MIN_KEEP     1       // data in db to be reserved.
+#define TSDB_MIN_KEEP     (1 * 1440)       // data in db to be reserved. unit minute
-#define TSDB_MAX_KEEP     365000  // data in db to be reserved.
+#define TSDB_MAX_KEEP     (365000 * 1440) // data in db to be reserved.
-#define TSDB_DEFAULT_KEEP 3650    // ten years
+#define TSDB_DEFAULT_KEEP (3650 * 1440)    // ten years
 
 #define TSDB_MIN_MIN_ROW_FBLOCK     10
 #define TSDB_MAX_MIN_ROW_FBLOCK     1000
@@ -327,7 +337,7 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_FSYNC_PERIOD     180000  // millisecond
 #define TSDB_DEFAULT_FSYNC_PERIOD 3000    // three second
 
-#define TSDB_MIN_WAL_LEVEL     0
+#define TSDB_MIN_WAL_LEVEL     1
 #define TSDB_MAX_WAL_LEVEL     2
 #define TSDB_DEFAULT_WAL_LEVEL 1
 
@@ -388,6 +398,7 @@ typedef enum ELogicConditionType {
 #define TSDB_DEFAULT_EXPLAIN_RATIO   0.001
 
 #define TSDB_EXPLAIN_RESULT_ROW_SIZE 1024
+#define TSDB_EXPLAIN_RESULT_COLUMN_NAME "QUERY PLAN"
 
 #define TSDB_MAX_JOIN_TABLE_NUM 10
 #define TSDB_MAX_UNION_CLAUSE   5
@@ -467,17 +478,17 @@ typedef struct {
 
 #define TMQ_SEPARATOR ':'
 
-#define SND_UNIQUE_THREAD_NUM 2
-#define SND_SHARED_THREAD_NUM 2
-
 enum {
   SND_WORKER_TYPE__SHARED = 1,
   SND_WORKER_TYPE__UNIQUE,
 };
 
-#define MND_VGID -1
+#define MNODE_HANDLE -1
-#define QND_VGID 1
+#define QNODE_HANDLE 1
-#define VND_VGID 0
+#define DEFAULT_HANDLE 0
+
+#define MAX_NUM_STR_SIZE 40
+
 
 #ifdef __cplusplus
 }

include/util/tlockfree.h

@@ -76,6 +76,7 @@ void taosWLockLatch(SRWLatch *pLatch);
 void taosWUnLockLatch(SRWLatch *pLatch);
 void taosRLockLatch(SRWLatch *pLatch);
 void taosRUnLockLatch(SRWLatch *pLatch);
+int32_t taosWTryLockLatch(SRWLatch *pLatch);
 
 // copy on read
 #define taosCorBeginRead(x)                     \

include/util/tprocess.h

@@ -22,13 +22,12 @@
 extern "C" {
 #endif
 
-typedef enum { PROC_REQ, PROC_RSP, PROC_REG, PROC_RELEASE } ProcFuncType;
+typedef enum { PROC_REQ = 1, PROC_RSP, PROC_REGIST, PROC_RELEASE } ProcFuncType;
 
-typedef struct SProcQueue SProcQueue;
 typedef struct SProcObj SProcObj;
 typedef void *(*ProcMallocFp)(int32_t contLen);
 typedef void *(*ProcFreeFp)(void *pCont);
-typedef void *(*ProcConsumeFp)(void *pParent, void *pHead, int16_t headLen, void *pBody, int32_t bodyLen,
+typedef void (*ProcConsumeFp)(void *parent, void *pHead, int16_t headLen, void *pBody, int32_t bodyLen,
                                ProcFuncType ftype);
 
 typedef struct {
@@ -43,7 +42,7 @@ typedef struct {
   ProcMallocFp  parentMallocBodyFp;
   ProcFreeFp    parentFreeBodyFp;
   SShm          shm;
-  void         *pParent;
+  void         *parent;
   const char   *name;
   bool          isChild;
 } SProcCfg;
@@ -51,9 +50,13 @@ typedef struct {
 SProcObj *taosProcInit(const SProcCfg *pCfg);
 void      taosProcCleanup(SProcObj *pProc);
 int32_t   taosProcRun(SProcObj *pProc);
+void      taosProcStop(SProcObj *pProc);
+
 int32_t taosProcPutToChildQ(SProcObj *pProc, const void *pHead, int16_t headLen, const void *pBody, int32_t bodyLen,
-                              ProcFuncType ftype);
+                            void *handle, ProcFuncType ftype);
-int32_t   taosProcPutToParentQ(SProcObj *pProc, const void *pHead, int16_t headLen, const void *pBody, int32_t bodyLen,
+void    taosProcRemoveHandle(SProcObj *pProc, void *handle);
+void    taosProcCloseHandles(SProcObj *pProc, void (*HandleFp)(void *handle));
+void    taosProcPutToParentQ(SProcObj *pProc, const void *pHead, int16_t headLen, const void *pBody, int32_t bodyLen,
                              ProcFuncType ftype);
 
 #ifdef __cplusplus

packaging/install.sh

@@ -0,0 +1,536 @@
+#!/bin/bash
+#
+# This file is used to install database on linux systems. The operating system
+# is required to use systemd to manage services at boot
+
+set -e
+#set -x
+
+# -----------------------Variables definition---------------------
+script_dir=$(dirname $(readlink -f "$0"))
+# Dynamic directory
+data_dir="/var/lib/taos"
+log_dir="/var/log/taos"
+
+data_link_dir="/usr/local/taos/data"
+log_link_dir="/usr/local/taos/log"
+
+cfg_install_dir="/etc/taos"
+
+bin_link_dir="/usr/bin"
+lib_link_dir="/usr/lib"
+lib64_link_dir="/usr/lib64"
+inc_link_dir="/usr/include"
+
+#install main path
+install_main_dir="/usr/local/taos"
+
+# old bin dir
+bin_dir="/usr/local/taos/bin"
+
+service_config_dir="/etc/systemd/system"
+
+# Color setting
+RED='\033[0;31m'
+GREEN='\033[1;32m'
+GREEN_DARK='\033[0;32m'
+GREEN_UNDERLINE='\033[4;32m'
+NC='\033[0m'
+
+csudo=""
+if command -v sudo > /dev/null; then
+    csudo="sudo"
+fi
+
+update_flag=0
+prompt_force=0
+
+initd_mod=0
+service_mod=2
+if pidof systemd &> /dev/null; then
+    service_mod=0
+elif $(which service &> /dev/null); then
+    service_mod=1
+    service_config_dir="/etc/init.d"
+    if $(which chkconfig &> /dev/null); then
+         initd_mod=1
+    elif $(which insserv &> /dev/null); then
+        initd_mod=2
+    elif $(which update-rc.d &> /dev/null); then
+        initd_mod=3
+    else
+        service_mod=2
+    fi
+else
+    service_mod=2
+fi
+
+
+# get the operating system type for using the corresponding init file
+# ubuntu/debian(deb), centos/fedora(rpm), others: opensuse, redhat, ..., no verification
+#osinfo=$(awk -F= '/^NAME/{print $2}' /etc/os-release)
+if [[ -e /etc/os-release ]]; then
+  osinfo=$(cat /etc/os-release | grep "NAME" | cut -d '"' -f2)   ||:
+else
+  osinfo=""
+fi
+#echo "osinfo: ${osinfo}"
+os_type=0
+if echo $osinfo | grep -qwi "ubuntu" ; then
+#  echo "This is ubuntu system"
+  os_type=1
+elif echo $osinfo | grep -qwi "debian" ; then
+#  echo "This is debian system"
+  os_type=1
+elif echo $osinfo | grep -qwi "Kylin" ; then
+#  echo "This is Kylin system"
+  os_type=1
+elif  echo $osinfo | grep -qwi "centos" ; then
+#  echo "This is centos system"
+  os_type=2
+elif echo $osinfo | grep -qwi "fedora" ; then
+#  echo "This is fedora system"
+  os_type=2
+elif echo $osinfo | grep -qwi "Linx" ; then
+#  echo "This is Linx system"
+  os_type=1
+  service_mod=0
+  initd_mod=0
+  service_config_dir="/etc/systemd/system"
+else
+  echo " osinfo: ${osinfo}"
+  echo " This is an officially unverified linux system,"
+  echo " if there are any problems with the installation and operation, "
+  echo " please feel free to contact taosdata.com for support."
+  os_type=1
+fi
+
+
+# =============================  get input parameters =================================================
+
+# install.sh -v [server | client]  -e [yes | no] -i [systemd | service | ...]
+
+# set parameters by default value
+interactiveFqdn=yes   # [yes | no]
+verType=server        # [server | client]
+initType=systemd      # [systemd | service | ...]
+
+while getopts "hv:e:i:" arg
+do
+  case $arg in
+    e)
+      #echo "interactiveFqdn=$OPTARG"
+      interactiveFqdn=$( echo $OPTARG )
+      ;;
+    v)
+      #echo "verType=$OPTARG"
+      verType=$(echo $OPTARG)
+      ;;
+    i)
+      #echo "initType=$OPTARG"
+      initType=$(echo $OPTARG)
+      ;;
+    h)
+      echo "Usage: `basename $0` -v [server | client]  -e [yes | no]"
+      exit 0
+      ;;
+    ?) #unknow option
+      echo "unkonw argument"
+      exit 1
+      ;;
+  esac
+done
+
+#echo "verType=${verType} interactiveFqdn=${interactiveFqdn}"
+
+function kill_process() {
+  pid=$(ps -ef | grep "$1" | grep -v "grep" | awk '{print $2}')
+  if [ -n "$pid" ]; then
+    ${csudo} kill -9 $pid   || :
+  fi
+}
+
+function install_main_path() {
+    #create install main dir and all sub dir
+    ${csudo} rm -rf ${install_main_dir}    || :
+    ${csudo} mkdir -p ${install_main_dir}
+    ${csudo} mkdir -p ${install_main_dir}/cfg
+    ${csudo} mkdir -p ${install_main_dir}/bin
+    ${csudo} mkdir -p ${install_main_dir}/connector
+    ${csudo} mkdir -p ${install_main_dir}/driver
+    ${csudo} mkdir -p ${install_main_dir}/examples
+    ${csudo} mkdir -p ${install_main_dir}/include
+    ${csudo} mkdir -p ${install_main_dir}/init.d
+    if [ "$verMode" == "cluster" ]; then
+        ${csudo} mkdir -p ${nginx_dir}
+    fi
+
+    if [[ -e ${script_dir}/email ]]; then
+      ${csudo} cp ${script_dir}/email ${install_main_dir}/ ||:
+    fi
+}
+
+function install_bin() {
+    # Remove links
+    ${csudo} rm -f ${bin_link_dir}/taos     || :
+    ${csudo} rm -f ${bin_link_dir}/taosd    || :
+    ${csudo} rm -f ${bin_link_dir}/taosadapter     || :
+    ${csudo} rm -f ${bin_link_dir}/create_table || :
+    ${csudo} rm -f ${bin_link_dir}/tmq_sim   || :
+    ${csudo} rm -f ${bin_link_dir}/taosdump || :
+    ${csudo} rm -f ${bin_link_dir}/rmtaos   || :
+    #${csudo} rm -f ${bin_link_dir}/set_core   || :
+
+    ${csudo} cp -r ${script_dir}/bin/* ${install_main_dir}/bin && ${csudo} chmod 0555 ${install_main_dir}/bin/*
+
+    #Make link
+    [ -x ${install_main_dir}/bin/taos ] && ${csudo} ln -s ${install_main_dir}/bin/taos ${bin_link_dir}/taos                          || :
+    [ -x ${install_main_dir}/bin/taosd ] && ${csudo} ln -s ${install_main_dir}/bin/taosd ${bin_link_dir}/taosd                       || :
+    [ -x ${install_main_dir}/bin/create_table ] && ${csudo} ln -s ${install_main_dir}/bin/create_table ${bin_link_dir}/create_table   || :
+    [ -x ${install_main_dir}/bin/tmq_sim ] && ${csudo} ln -s ${install_main_dir}/bin/tmq_sim ${bin_link_dir}/tmq_sim     || :
+#    [ -x ${install_main_dir}/bin/taosdemo ] && ${csudo} ln -s ${install_main_dir}/bin/taosdemo ${bin_link_dir}/taosdemo              || :
+#    [ -x ${install_main_dir}/bin/taosdump ] && ${csudo} ln -s ${install_main_dir}/bin/taosdump ${bin_link_dir}/taosdump              || :
+    [ -x ${install_main_dir}/bin/remove.sh ] && ${csudo} ln -s ${install_main_dir}/bin/remove.sh ${bin_link_dir}/rmtaos              || :
+#    [ -x ${install_main_dir}/bin/set_core.sh ] && ${csudo} ln -s ${install_main_dir}/bin/set_core.sh ${bin_link_dir}/set_core        || :
+}
+
+function install_lib() {
+    # Remove links
+    ${csudo} rm -f ${lib_link_dir}/libtaos.*         || :
+    ${csudo} rm -f ${lib64_link_dir}/libtaos.*       || :
+
+    ${csudo} ln -s ${install_main_dir}/lib/libtaos.* ${lib_link_dir}/libtaos.so.1
+    ${csudo} ln -s ${lib_link_dir}/libtaos.so.1 ${lib_link_dir}/libtaos.so
+
+    ${csudo} ln -s ${install_main_dir}/lib/libtdb.* ${lib_link_dir}/libtdb.so.1
+    ${csudo} ln -s ${lib_link_dir}/libtdb.so.1 ${lib_link_dir}/libtdb.so
+
+    if [[ -d ${lib64_link_dir} && ! -e ${lib64_link_dir}/libtaos.so ]]; then
+      ${csudo} ln -s ${install_main_dir}/lib/libtaos.* ${lib64_link_dir}/libtaos.so.1     || :
+      ${csudo} ln -s ${lib64_link_dir}/libtaos.so.1 ${lib64_link_dir}/libtaos.so          || :
+      
+      ${csudo} ln -s ${install_main_dir}/lib/libtdb.* ${lib64_link_dir}/libtdb.so.1       || :
+      ${csudo} ln -s ${lib64_link_dir}/libtdb.so.1 ${lib64_link_dir}/libtdb.so            || :
+    fi
+
+    ${csudo} ldconfig
+}
+
+function install_header() {
+    ${csudo} rm -f ${inc_link_dir}/taos.h ${inc_link_dir}/taosdef.h ${inc_link_dir}/taoserror.h    || :
+    ${csudo} cp -f ${script_dir}/inc/* ${install_main_dir}/include && ${csudo} chmod 644 ${install_main_dir}/include/*
+    ${csudo} ln -s ${install_main_dir}/include/taos.h ${inc_link_dir}/taos.h
+#    ${csudo} ln -s ${install_main_dir}/include/taosdef.h ${inc_link_dir}/taosdef.h
+    ${csudo} ln -s ${install_main_dir}/include/taoserror.h ${inc_link_dir}/taoserror.h
+}
+
+function add_newHostname_to_hosts() {
+  localIp="127.0.0.1"
+  OLD_IFS="$IFS"
+  IFS=" "
+  iphost=$(cat /etc/hosts | grep $1 | awk '{print $1}')
+  arr=($iphost)
+  IFS="$OLD_IFS"
+  for s in "${arr[@]}"
+  do
+    if [[ "$s" == "$localIp" ]]; then
+      return
+    fi
+  done
+  ${csudo} echo "127.0.0.1  $1" >> /etc/hosts   ||:
+}
+
+function set_hostname() {
+  echo -e -n "${GREEN}Please enter one hostname(must not be 'localhost')${NC}:"
+    read newHostname
+  while true; do
+    if [[ ! -z "$newHostname" && "$newHostname" != "localhost" ]]; then
+      break
+    else
+      read -p "Please enter one hostname(must not be 'localhost'):" newHostname
+    fi
+  done
+
+  ${csudo} hostname $newHostname ||:
+  retval=`echo $?`
+  if [[ $retval != 0 ]]; then
+   echo
+   echo "set hostname fail!"
+   return
+  fi
+  #echo -e -n "$(hostnamectl status --static)"
+  #echo -e -n "$(hostnamectl status --transient)"
+  #echo -e -n "$(hostnamectl status --pretty)"
+
+  #ubuntu/centos /etc/hostname
+  if [[ -e /etc/hostname ]]; then
+    ${csudo} echo $newHostname > /etc/hostname   ||:
+  fi
+
+  #debian: #HOSTNAME=yourname
+  if [[ -e /etc/sysconfig/network ]]; then
+    ${csudo} sed -i -r "s/#*\s*(HOSTNAME=\s*).*/\1$newHostname/" /etc/sysconfig/network   ||:
+  fi
+
+  ${csudo} sed -i -r "s/#*\s*(fqdn\s*).*/\1$newHostname/" ${cfg_install_dir}/taos.cfg
+  serverFqdn=$newHostname
+
+  if [[ -e /etc/hosts ]]; then
+    add_newHostname_to_hosts $newHostname
+  fi
+}
+
+function is_correct_ipaddr() {
+  newIp=$1
+  OLD_IFS="$IFS"
+  IFS=" "
+  arr=($iplist)
+  IFS="$OLD_IFS"
+  for s in "${arr[@]}"
+  do
+   if [[ "$s" == "$newIp" ]]; then
+     return 0
+   fi
+  done
+
+  return 1
+}
+
+function set_ipAsFqdn() {
+  iplist=$(ip address |grep inet |grep -v inet6 |grep -v 127.0.0.1 |awk '{print $2}' |awk -F "/" '{print $1}') ||:
+  if [ -z "$iplist" ]; then
+    iplist=$(ifconfig |grep inet |grep -v inet6 |grep -v 127.0.0.1 |awk '{print $2}' |awk -F ":" '{print $2}') ||:
+  fi
+
+  if [ -z "$iplist" ]; then
+    echo
+    echo -e -n "${GREEN}Unable to get local ip, use 127.0.0.1${NC}"
+    localFqdn="127.0.0.1"
+    # Write the local FQDN to configuration file
+    ${csudo} sed -i -r "s/#*\s*(fqdn\s*).*/\1$localFqdn/" ${cfg_install_dir}/taos.cfg
+    serverFqdn=$localFqdn
+    echo
+    return
+  fi
+
+  echo -e -n "${GREEN}Please choose an IP from local IP list${NC}:"
+  echo
+  echo -e -n "${GREEN}$iplist${NC}"
+  echo
+  echo
+  echo -e -n "${GREEN}Notes: if IP is used as the node name, data can NOT be migrated to other machine directly${NC}:"
+  read localFqdn
+    while true; do
+      if [ ! -z "$localFqdn" ]; then
+        # Check if correct ip address
+        is_correct_ipaddr $localFqdn
+        retval=`echo $?`
+        if [[ $retval != 0 ]]; then
+          read -p "Please choose an IP from local IP list:" localFqdn
+        else
+          # Write the local FQDN to configuration file
+          ${csudo} sed -i -r "s/#*\s*(fqdn\s*).*/\1$localFqdn/" ${cfg_install_dir}/taos.cfg
+          serverFqdn=$localFqdn
+          break
+        fi
+      else
+        read -p "Please choose an IP from local IP list:" localFqdn
+      fi
+    done
+}
+
+function local_fqdn_check() {
+  #serverFqdn=$(hostname)
+  echo
+  echo -e -n "System hostname is: ${GREEN}$serverFqdn${NC}"
+  echo
+  if [[ "$serverFqdn" == "" ]] || [[ "$serverFqdn" == "localhost"  ]]; then
+    echo -e -n "${GREEN}It is strongly recommended to configure a hostname for this machine ${NC}"
+    echo
+
+    while true
+    do
+        read -r -p "Set hostname now? [Y/n] " input
+        if [ ! -n "$input" ]; then
+            set_hostname
+            break
+        else
+            case $input in
+                [yY][eE][sS]|[yY])
+                    set_hostname
+                    break
+                    ;;
+
+                [nN][oO]|[nN])
+                    set_ipAsFqdn
+                    break
+                    ;;
+
+                *)
+                    echo "Invalid input..."
+                    ;;
+            esac
+        fi
+    done
+  fi
+}
+
+function install_log() {
+    ${csudo} rm -rf ${log_dir}  || :
+    ${csudo} mkdir -p ${log_dir} && ${csudo} chmod 777 ${log_dir}
+
+    ${csudo} ln -s ${log_dir} ${install_main_dir}/log
+}
+
+function install_data() {
+    ${csudo} mkdir -p ${data_dir}
+
+    ${csudo} ln -s ${data_dir} ${install_main_dir}/data
+}
+
+function clean_service_on_systemd() {
+    taosd_service_config="${service_config_dir}/taosd.service"
+    if systemctl is-active --quiet taosd; then
+        echo "TDengine is running, stopping it..."
+        ${csudo} systemctl stop taosd &> /dev/null || echo &> /dev/null
+    fi
+    ${csudo} systemctl disable taosd &> /dev/null || echo &> /dev/null
+    ${csudo} rm -f ${taosd_service_config}
+
+    tarbitratord_service_config="${service_config_dir}/tarbitratord.service"
+    if systemctl is-active --quiet tarbitratord; then
+        echo "tarbitrator is running, stopping it..."
+        ${csudo} systemctl stop tarbitratord &> /dev/null || echo &> /dev/null
+    fi
+    ${csudo} systemctl disable tarbitratord &> /dev/null || echo &> /dev/null
+    ${csudo} rm -f ${tarbitratord_service_config}
+
+    if [ "$verMode" == "cluster" ]; then
+        nginx_service_config="${service_config_dir}/nginxd.service"
+        if systemctl is-active --quiet nginxd; then
+            echo "Nginx for TDengine is running, stopping it..."
+            ${csudo} systemctl stop nginxd &> /dev/null || echo &> /dev/null
+        fi
+        ${csudo} systemctl disable nginxd &> /dev/null || echo &> /dev/null
+        ${csudo} rm -f ${nginx_service_config}
+    fi
+}
+
+# taos:2345:respawn:/etc/init.d/taosd start
+
+function install_service_on_systemd() {
+    clean_service_on_systemd
+        
+    taosd_service_config="${service_config_dir}/taosd.service"
+    ${csudo} bash -c "echo '[Unit]'                             >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'Description=TDengine server service' >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'After=network-online.target taosadapter.service'        >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'Wants=network-online.target taosadapter.service'        >> ${taosd_service_config}"
+    ${csudo} bash -c "echo                                      >> ${taosd_service_config}"
+    ${csudo} bash -c "echo '[Service]'                          >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'Type=simple'                        >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'ExecStart=/usr/bin/taosd'           >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'ExecStartPre=/usr/local/taos/bin/startPre.sh'         >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'TimeoutStopSec=1000000s'            >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'LimitNOFILE=infinity'               >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'LimitNPROC=infinity'                >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'LimitCORE=infinity'                 >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'TimeoutStartSec=0'                  >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'StandardOutput=null'                >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'Restart=always'                     >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'StartLimitBurst=3'                  >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'StartLimitInterval=60s'             >> ${taosd_service_config}"
+    #${csudo} bash -c "echo 'StartLimitIntervalSec=60s'          >> ${taosd_service_config}"
+    ${csudo} bash -c "echo                                      >> ${taosd_service_config}"
+    ${csudo} bash -c "echo '[Install]'                          >> ${taosd_service_config}"
+    ${csudo} bash -c "echo 'WantedBy=multi-user.target'         >> ${taosd_service_config}"
+    ${csudo} systemctl enable taosd
+
+    ${csudo} systemctl daemon-reload 
+}
+
+function install_service() {
+    if ((${service_mod}==0)); then
+        install_service_on_systemd
+    elif ((${service_mod}==1)); then
+        install_service_on_sysvinit
+    else
+        # must manual stop taosd
+        kill_process taosd
+    fi
+}
+
+function install_TDengine() {
+    # Start to install
+    echo -e "${GREEN}Start to install TDengine...${NC}"
+
+    install_main_path
+    install_data
+    install_log
+    install_header
+    install_lib
+
+    if [ -z $1 ]; then # install service and client
+        # For installing new
+        install_bin
+        install_service
+        #install_config
+
+        # Ask if to start the service
+        #echo
+        #echo -e "\033[44;32;1mTDengine is installed successfully!${NC}"
+        echo
+        echo -e "${GREEN_DARK}To configure TDengine ${NC}: edit /etc/taos/taos.cfg"
+        if ((${service_mod}==0)); then
+          echo -e "${GREEN_DARK}To start TDengine     ${NC}: ${csudo} systemctl start taosd${NC}"
+        elif ((${service_mod}==1)); then
+          echo -e "${GREEN_DARK}To start TDengine     ${NC}: ${csudo} service taosd start${NC}"
+        else
+          echo -e "${GREEN_DARK}To start TDengine     ${NC}: taosd${NC}"
+        fi
+
+        if [ ! -z "$firstEp" ]; then
+            tmpFqdn=${firstEp%%:*}
+            substr=":"
+            if [[ $firstEp =~ $substr ]];then
+                tmpPort=${firstEp#*:}
+            else
+                tmpPort=""
+            fi
+            if [[ "$tmpPort" != "" ]];then
+                echo -e "${GREEN_DARK}To access TDengine    ${NC}: taos -h $tmpFqdn -P $tmpPort${GREEN_DARK} to login into cluster, then${NC}"
+            else
+                echo -e "${GREEN_DARK}To access TDengine    ${NC}: taos -h $tmpFqdn${GREEN_DARK} to login into cluster, then${NC}"
+            fi
+            echo -e "${GREEN_DARK}execute ${NC}: create dnode 'newDnodeFQDN:port'; ${GREEN_DARK}to add this new node${NC}"
+            echo
+        elif [ ! -z "$serverFqdn" ]; then
+            echo -e "${GREEN_DARK}To access TDengine    ${NC}: taos -h $serverFqdn${GREEN_DARK} to login into TDengine server${NC}"
+            echo
+        fi
+
+        echo -e "\033[44;32;1mTDengine is installed successfully!${NC}"
+        echo
+    else # Only install client
+        install_bin
+        #install_config
+        echo
+        echo -e "\033[44;32;1mTDengine client is installed successfully!${NC}"
+    fi
+
+    touch ~/.taos_history
+}
+
+
+## ==============================Main program starts from here============================
+serverFqdn=$(hostname)
+if [ "$verType" == "server" ]; then
+    # Install server and client
+    install_TDengine
+elif [ "$verType" == "client" ]; then
+    interactiveFqdn=no
+    # Only install client
+    install_TDengine client
+else
+    echo  "please input correct verType"
+fi

packaging/release.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 #
-# Generate the deb package for ubuntu, or rpm package for centos, or tar.gz package for other linux os
+# Generate the tar.gz package for linux os
 
 set -e
 #set -x
@@ -17,9 +17,12 @@ echo "=======================new version number: ${verNumber}===================
 
 build_time=$(date +"%F %R")
 
+echo "script_dir: ${script_dir}"
 echo "top_dir: ${top_dir}"
 
 cd ${top_dir}
+git checkout -- .
+git checkout 3.0
 git pull || :
 
 echo "curr_dir: ${curr_dir}"
@@ -49,28 +52,38 @@ cd ${release_dir}
 
 install_dir="${release_dir}/TDengine-server-${version}"
 mkdir -p ${install_dir}
+mkdir -p ${install_dir}/bin
 mkdir -p ${install_dir}/lib
+mkdir -p ${install_dir}/inc
 
-bin_files="${compile_dir}/source/dnode/mgmt/daemon/taosd ${compile_dir}/tools/shell/taos  ${compile_dir}/tests/test/c/create_table"
+install_files="${script_dir}/install.sh"
-cp ${bin_files} ${install_dir}/ && chmod a+x ${install_dir}/* || :
+chmod a+x ${script_dir}/install.sh || :
+cp ${install_files} ${install_dir} 
 
+header_files="${top_dir}/include/client/taos.h ${top_dir}/include/util/taoserror.h"
+cp ${header_files} ${install_dir}/inc
+ 
+bin_files="${compile_dir}/source/dnode/mgmt/taosd ${compile_dir}/tools/shell/taos  ${compile_dir}/tests/test/c/create_table ${compile_dir}/tests/test/c/tmq_sim ${script_dir}/remove.sh"
+cp ${bin_files} ${install_dir}/bin && chmod a+x ${install_dir}/bin/* || :
 
 cp ${compile_dir}/source/client/libtaos.so  ${install_dir}/lib/
-cp ${compile_dir}/source/dnode/mnode/impl/libmnode.so ${install_dir}/lib/
+cp ${compile_dir}/source/libs/tdb/libtdb.so ${install_dir}/lib/
-cp ${compile_dir}/source/dnode/qnode/libqnode.so ${install_dir}/lib/
+
-cp ${compile_dir}/source/dnode/snode/libsnode.so ${install_dir}/lib/
+#cp ${compile_dir}/source/dnode/mnode/impl/libmnode.so ${install_dir}/lib/
-cp ${compile_dir}/source/dnode/bnode/libbnode.so ${install_dir}/lib/
+#cp ${compile_dir}/source/dnode/qnode/libqnode.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/wal/libwal.so ${install_dir}/lib/
+#cp ${compile_dir}/source/dnode/snode/libsnode.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/scheduler/libscheduler.so ${install_dir}/lib/
+#cp ${compile_dir}/source/dnode/bnode/libbnode.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/planner/libplanner.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/wal/libwal.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/parser/libparser.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/scheduler/libscheduler.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/qcom/libqcom.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/planner/libplanner.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/transport/libtransport.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/parser/libparser.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/function/libfunction.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/qcom/libqcom.so ${install_dir}/lib/
-cp ${compile_dir}/source/common/libcommon.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/transport/libtransport.so ${install_dir}/lib/
-cp ${compile_dir}/source/os/libos.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/function/libfunction.so ${install_dir}/lib/
-cp ${compile_dir}/source/dnode/mnode/sdb/libsdb.so ${install_dir}/lib/
+#cp ${compile_dir}/source/common/libcommon.so ${install_dir}/lib/
-cp ${compile_dir}/source/libs/catalog/libcatalog.so ${install_dir}/lib/
+#cp ${compile_dir}/source/os/libos.so ${install_dir}/lib/
+#cp ${compile_dir}/source/dnode/mnode/sdb/libsdb.so ${install_dir}/lib/
+#cp ${compile_dir}/source/libs/catalog/libcatalog.so ${install_dir}/lib/
 
 pkg_name=${install_dir}-Linux-x64
 

packaging/remove.sh

@@ -0,0 +1,148 @@
+#!/bin/bash
+#
+# Script to stop the service and uninstall TDengine, but retain the config, data and log files.
+
+set -e
+#set -x
+
+verMode=edge
+
+RED='\033[0;31m'
+GREEN='\033[1;32m'
+NC='\033[0m'
+
+#install main path
+install_main_dir="/usr/local/taos"
+data_link_dir="/usr/local/taos/data"
+log_link_dir="/usr/local/taos/log"
+cfg_link_dir="/usr/local/taos/cfg"
+bin_link_dir="/usr/bin"
+lib_link_dir="/usr/lib"
+lib64_link_dir="/usr/lib64"
+inc_link_dir="/usr/include"
+
+service_config_dir="/etc/systemd/system"
+taos_service_name="taosd"
+
+csudo=""
+if command -v sudo > /dev/null; then
+    csudo="sudo"
+fi
+
+initd_mod=0
+service_mod=2
+if pidof systemd &> /dev/null; then
+    service_mod=0
+elif $(which service &> /dev/null); then
+    service_mod=1
+    service_config_dir="/etc/init.d"
+    if $(which chkconfig &> /dev/null); then
+         initd_mod=1
+    elif $(which insserv &> /dev/null); then
+        initd_mod=2
+    elif $(which update-rc.d &> /dev/null); then
+        initd_mod=3
+    else
+        service_mod=2
+    fi
+else
+    service_mod=2
+fi
+
+
+function kill_taosd() {
+  pid=$(ps -ef | grep "taosd" | grep -v "grep" | awk '{print $2}')
+  if [ -n "$pid" ]; then
+    ${csudo} kill -9 $pid   || :
+  fi
+}
+
+
+function clean_bin() {
+    # Remove link
+    ${csudo} rm -f ${bin_link_dir}/taos        || :
+    ${csudo} rm -f ${bin_link_dir}/taosd       || :
+    ${csudo} rm -f ${bin_link_dir}/create_table  || :
+    ${csudo} rm -f ${bin_link_dir}/tmq_sim || :
+    ${csudo} rm -f ${bin_link_dir}/taosdemo    || :
+    ${csudo} rm -f ${bin_link_dir}/taosdump    || :
+    ${csudo} rm -f ${bin_link_dir}/rmtaos      || :
+    #${csudo} rm -f ${bin_link_dir}/set_core    || :
+}
+
+function clean_lib() {
+    # Remove link
+    ${csudo} rm -f ${lib_link_dir}/libtaos.*      || :
+    ${csudo} rm -f ${lib64_link_dir}/libtaos.*    || :    
+    
+    ${csudo} rm -f ${lib_link_dir}/libtdb.*      || :
+    ${csudo} rm -f ${lib64_link_dir}/libtdb.*    || :
+}
+
+function clean_header() {
+    # Remove link
+    ${csudo} rm -f ${inc_link_dir}/taos.h       || :
+    ${csudo} rm -f ${inc_link_dir}/taosdef.h    || :
+    ${csudo} rm -f ${inc_link_dir}/taoserror.h  || :
+}
+
+function clean_config() {
+    # Remove link
+    ${csudo} rm -f ${cfg_link_dir}/*            || :
+}
+
+function clean_log() {
+    # Remove link
+    ${csudo} rm -rf ${log_link_dir}    || :
+}
+
+function clean_service_on_systemd() {
+    taosd_service_config="${service_config_dir}/${taos_service_name}.service"
+    if systemctl is-active --quiet ${taos_service_name}; then
+        echo "TDengine taosd is running, stopping it..."
+        ${csudo} systemctl stop ${taos_service_name} &> /dev/null || echo &> /dev/null
+    fi
+    ${csudo} systemctl disable ${taos_service_name} &> /dev/null || echo &> /dev/null
+    ${csudo} rm -f ${taosd_service_config}
+}
+
+function clean_service_on_sysvinit() {
+    echo " "
+}
+
+function clean_service() {
+    if ((${service_mod}==0)); then
+        clean_service_on_systemd
+    elif ((${service_mod}==1)); then
+        clean_service_on_sysvinit
+    else
+        # must manual stop taosd
+        kill_taosd
+    fi
+}
+
+# Stop service and disable booting start.
+clean_service
+# Remove binary file and links
+clean_bin
+# Remove header file.
+clean_header
+# Remove lib file
+clean_lib
+# Remove link log directory
+clean_log
+# Remove link configuration file
+clean_config
+# Remove data link directory
+${csudo} rm -rf ${data_link_dir}    || :
+
+${csudo} rm -rf ${install_main_dir}
+
+if [[ -e /etc/os-release ]]; then
+  osinfo=$(awk -F= '/^NAME/{print $2}' /etc/os-release)
+else
+  osinfo=""
+fi
+
+echo -e "${GREEN}TDengine is removed successfully!${NC}"
+echo

source/client/CMakeLists.txt

@@ -8,7 +8,7 @@ target_include_directories(
 target_link_libraries(
     taos
     INTERFACE api
-    PRIVATE os util common transport nodes parser planner catalog scheduler function qcom
+    PRIVATE os util common transport nodes parser command planner catalog scheduler function qcom
 )
 
 if(${BUILD_TEST})

source/client/inc/clientInt.h

@@ -152,7 +152,8 @@ typedef struct SResultColumn {
 typedef struct SReqResultInfo {
   const char*    pRspMsg;
   const char*    pData;
-  TAOS_FIELD*    fields;
+  TAOS_FIELD*    fields;     // todo, column names are not needed.
+  TAOS_FIELD*    userFields; // the fields info that return to user
   uint32_t       numOfCols;
   int32_t*       length;
   char**         convertBuf;
@@ -162,6 +163,7 @@ typedef struct SReqResultInfo {
   uint64_t       totalRows;
   uint32_t       current;
   bool           completed;
+  int32_t        precision;
   int32_t        payloadLen;
 } SReqResultInfo;
 
@@ -221,6 +223,7 @@ void  destroyRequest(SRequestObj* pRequest);
 
 char* getDbOfConnection(STscObj* pObj);
 void  setConnectionDB(STscObj* pTscObj, const char* db);
+void  resetConnectDB(STscObj* pTscObj);
 
 void taos_init_imp(void);
 int  taos_options_imp(TSDB_OPTION option, const char* str);

source/client/src/clientEnv.c

@@ -169,6 +169,7 @@ static void doFreeReqResultInfo(SReqResultInfo *pResInfo) {
   taosMemoryFreeClear(pResInfo->row);
   taosMemoryFreeClear(pResInfo->pCol);
   taosMemoryFreeClear(pResInfo->fields);
+  taosMemoryFreeClear(pResInfo->userFields);
 
   if (pResInfo->convertBuf != NULL) {
     for (int32_t i = 0; i < pResInfo->numOfCols; ++i) {

source/client/src/clientImpl.c

@@ -1,6 +1,7 @@
 
 #include "clientInt.h"
 #include "clientLog.h"
+#include "command.h"
 #include "scheduler.h"
 #include "tdatablock.h"
 #include "tdef.h"
@@ -55,7 +56,7 @@ TAOS* taos_connect_internal(const char* ip, const char* user, const char* pass,
   }
 
   char localDb[TSDB_DB_NAME_LEN] = {0};
-  if (db != NULL) {
+  if (db != NULL && strlen(db) > 0) {
     if (!validateDbName(db)) {
       terrno = TSDB_CODE_TSC_INVALID_DB_LENGTH;
       return NULL;
@@ -163,6 +164,7 @@ int32_t parseSql(SRequestObj* pRequest, bool topicQuery, SQuery** pQuery) {
     if ((*pQuery)->haveResultSet) {
       setResSchemaInfo(&pRequest->body.resInfo, (*pQuery)->pResSchema, (*pQuery)->numOfResCols);
     }
+
     TSWAP(pRequest->dbList, (*pQuery)->pDbList, SArray*);
     TSWAP(pRequest->tableList, (*pQuery)->pTableList, SArray*);
   }
@@ -170,7 +172,21 @@ int32_t parseSql(SRequestObj* pRequest, bool topicQuery, SQuery** pQuery) {
   return code;
 }
 
+int32_t execLocalCmd(SRequestObj* pRequest, SQuery* pQuery) {
+  SRetrieveTableRsp* pRsp = NULL;
+  int32_t code = qExecCommand(pQuery->pRoot, &pRsp);
+  if (TSDB_CODE_SUCCESS == code && NULL != pRsp) {
+    code = setQueryResultFromRsp(&pRequest->body.resInfo, pRsp);
+  }
+  return code;
+}
+
 int32_t execDdlQuery(SRequestObj* pRequest, SQuery* pQuery) {
+  // drop table if exists not_exists_table
+  if (NULL == pQuery->pCmdMsg) {
+    return TSDB_CODE_SUCCESS;
+  }
+
   SCmdMsgInfo* pMsgInfo = pQuery->pCmdMsg;
   pRequest->type = pMsgInfo->msgType;
   pRequest->body.requestMsg = (SDataBuf){.pData = pMsgInfo->pMsg, .len = pMsgInfo->msgLen, .handle = NULL};
@@ -213,12 +229,24 @@ void setResSchemaInfo(SReqResultInfo* pResInfo, const SSchema* pSchema, int32_t
   assert(pSchema != NULL && numOfCols > 0);
 
   pResInfo->numOfCols = numOfCols;
-  pResInfo->fields = taosMemoryCalloc(numOfCols, sizeof(pSchema[0]));
+  pResInfo->fields = taosMemoryCalloc(numOfCols, sizeof(TAOS_FIELD));
+  pResInfo->userFields = taosMemoryCalloc(numOfCols, sizeof(TAOS_FIELD));
 
   for (int32_t i = 0; i < pResInfo->numOfCols; ++i) {
     pResInfo->fields[i].bytes = pSchema[i].bytes;
     pResInfo->fields[i].type  = pSchema[i].type;
+
+    pResInfo->userFields[i].bytes = pSchema[i].bytes;
+    pResInfo->userFields[i].type  = pSchema[i].type;
+
+    if (pSchema[i].type == TSDB_DATA_TYPE_VARCHAR) {
+      pResInfo->userFields[i].bytes -= VARSTR_HEADER_SIZE;
+    } else if (pSchema[i].type == TSDB_DATA_TYPE_NCHAR) {
+      pResInfo->userFields[i].bytes = (pResInfo->userFields[i].bytes - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE;
+    }
+
     tstrncpy(pResInfo->fields[i].name, pSchema[i].name, tListLen(pResInfo->fields[i].name));
+    tstrncpy(pResInfo->userFields[i].name, pSchema[i].name, tListLen(pResInfo->userFields[i].name));
   }
 }
 
@@ -226,7 +254,7 @@ int32_t scheduleQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNodeList
   void* pTransporter = pRequest->pTscObj->pAppInfo->pTransporter;
 
   SQueryResult res = {.code = 0, .numOfRows = 0, .msgSize = ERROR_MSG_BUF_DEFAULT_SIZE, .msg = pRequest->msgBuf};
-  int32_t      code = schedulerExecJob(pTransporter, pNodeList, pDag, &pRequest->body.queryJob, pRequest->sqlstr, &res);
+  int32_t      code = schedulerExecJob(pTransporter, pNodeList, pDag, &pRequest->body.queryJob, pRequest->sqlstr, pRequest->metric.start, &res);
   if (code != TSDB_CODE_SUCCESS) {
     if (pRequest->body.queryJob != 0) {
       schedulerFreeJob(pRequest->body.queryJob);
@@ -253,20 +281,35 @@ int32_t scheduleQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNodeList
 SRequestObj* execQueryImpl(STscObj* pTscObj, const char* sql, int sqlLen) {
   SRequestObj* pRequest = NULL;
   SQuery* pQuery = NULL;
-  int32_t code = 0;
   SArray* pNodeList = taosArrayInit(4, sizeof(struct SQueryNodeAddr));
 
-  CHECK_CODE_GOTO(buildRequest(pTscObj, sql, sqlLen, &pRequest), _return);
+  int32_t code = buildRequest(pTscObj, sql, sqlLen, &pRequest);
-  CHECK_CODE_GOTO(parseSql(pRequest, false, &pQuery), _return);
+  if (TSDB_CODE_SUCCESS == code) {
-
+    code = parseSql(pRequest, false, &pQuery);
-  if (pQuery->directRpc) {
+  }
-    CHECK_CODE_GOTO(execDdlQuery(pRequest, pQuery), _return);
+
-  } else {
+  if (TSDB_CODE_SUCCESS == code) {
-    CHECK_CODE_GOTO(getPlan(pRequest, pQuery, &pRequest->body.pDag, pNodeList), _return);
+    switch (pQuery->execMode) {
-    CHECK_CODE_GOTO(scheduleQuery(pRequest, pRequest->body.pDag, pNodeList), _return);
+      case QUERY_EXEC_MODE_LOCAL:
+        code = execLocalCmd(pRequest, pQuery);
+        break;
+      case QUERY_EXEC_MODE_RPC:
+        code = execDdlQuery(pRequest, pQuery);
+        break;
+      case QUERY_EXEC_MODE_SCHEDULE:
+        code = getPlan(pRequest, pQuery, &pRequest->body.pDag, pNodeList);
+        if (TSDB_CODE_SUCCESS == code) {
+          code = scheduleQuery(pRequest, pRequest->body.pDag, pNodeList);
+        }
+        break;
+      case QUERY_EXEC_MODE_EMPTY_RESULT:
+        pRequest->type = TSDB_SQL_RETRIEVE_EMPTY_RESULT;
+        break;
+      default:
+        break;
+    }
   }
 
-_return:
   taosArrayDestroy(pNodeList);
   qDestroyQuery(pQuery);
   if (NULL != pRequest && TSDB_CODE_SUCCESS != code) {
@@ -464,9 +507,11 @@ static void destroySendMsgInfo(SMsgSendInfo* pMsgBody) {
   taosMemoryFreeClear(pMsgBody->msgInfo.pData);
   taosMemoryFreeClear(pMsgBody);
 }
+
 bool persistConnForSpecificMsg(void* parenct, tmsg_t msgType) {
   return msgType == TDMT_VND_QUERY_RSP || msgType == TDMT_VND_FETCH_RSP || msgType == TDMT_VND_RES_READY_RSP || msgType == TDMT_VND_QUERY_HEARTBEAT_RSP;
 }
+
 void processMsgFromServer(void* parent, SRpcMsg* pMsg, SEpSet* pEpSet) {
   SMsgSendInfo* pSendInfo = (SMsgSendInfo*)pMsg->ahandle;
   assert(pMsg->ahandle != NULL);
@@ -647,6 +692,11 @@ void* doFetchRow(SRequestObj* pRequest, bool setupOneRowPtr) {
       }
     }
 
+    if (pResultInfo->completed) {
+      pResultInfo->numOfRows = 0;
+      return NULL;
+    }
+
     SMsgSendInfo* body = buildMsgInfoImpl(pRequest);
 
     int64_t  transporterId = 0;
@@ -716,6 +766,7 @@ int32_t setResultDataPtr(SReqResultInfo* pResultInfo, TAOS_FIELD* pFields, int32
     pStart += colLength[i];
   }
 
+  // convert UCS4-LE encoded character to native multi-bytes character in current data block.
   for (int32_t i = 0; i < numOfCols; ++i) {
     int32_t type = pResultInfo->fields[i].type;
     int32_t bytes = pResultInfo->fields[i].bytes;
@@ -742,6 +793,7 @@ int32_t setResultDataPtr(SReqResultInfo* pResultInfo, TAOS_FIELD* pFields, int32
       }
 
       pResultInfo->pCol[i].pData = pResultInfo->convertBuf[i];
+      pResultInfo->row[i] = pResultInfo->pCol[i].pData;
     }
   }
 
@@ -767,6 +819,16 @@ void setConnectionDB(STscObj* pTscObj, const char* db) {
   taosThreadMutexUnlock(&pTscObj->mutex);
 }
 
+void resetConnectDB(STscObj* pTscObj) {
+  if (pTscObj == NULL) {
+    return;
+  }
+
+  taosThreadMutexLock(&pTscObj->mutex);
+  pTscObj->db[0] = 0;
+  taosThreadMutexUnlock(&pTscObj->mutex);
+}
+
 int32_t setQueryResultFromRsp(SReqResultInfo* pResultInfo, const SRetrieveTableRsp* pRsp) {
   assert(pResultInfo != NULL && pRsp != NULL);
 
@@ -776,6 +838,7 @@ int32_t setQueryResultFromRsp(SReqResultInfo* pResultInfo, const SRetrieveTableR
   pResultInfo->current    = 0;
   pResultInfo->completed  = (pRsp->completed == 1);
   pResultInfo->payloadLen = htonl(pRsp->compLen);
+  pResultInfo->precision  = pRsp->precision;
 
   // TODO handle the compressed case
   pResultInfo->totalRows += pResultInfo->numOfRows;

source/client/src/clientMain.c

@@ -146,7 +146,7 @@ TAOS_FIELD *taos_fetch_fields(TAOS_RES *res) {
   }
 
   SReqResultInfo *pResInfo = &(((SRequestObj *)res)->body.resInfo);
-  return pResInfo->fields;
+  return pResInfo->userFields;
 }
 
 TAOS_RES *taos_query(TAOS *taos, const char *sql) {
@@ -264,10 +264,19 @@ int *taos_fetch_lengths(TAOS_RES *res) {
 }
 
 TAOS_ROW *taos_result_block(TAOS_RES *res) {
-  // TODO
+  SRequestObj* pRequest = (SRequestObj*) res;
+  if (pRequest == NULL) {
+    terrno = TSDB_CODE_INVALID_PARA;
     return NULL;
   }
 
+  if (taos_is_update_query(res)) {
+    return NULL;
+  }
+
+  return &pRequest->body.resInfo.row;
+}
+
 // todo intergrate with tDataTypes
 const char *taos_data_type(int type) {
   switch (type) {
@@ -313,7 +322,14 @@ int taos_affected_rows(TAOS_RES *res) {
   return pResInfo->numOfRows;
 }
 
-int taos_result_precision(TAOS_RES *res) { return TSDB_TIME_PRECISION_MILLI; }
+int taos_result_precision(TAOS_RES *res) {
+  SRequestObj* pRequest = (SRequestObj*) res;
+  if (pRequest == NULL) {
+    return TSDB_TIME_PRECISION_MILLI;
+  }
+
+  return pRequest->body.resInfo.precision;
+}
 
 int taos_select_db(TAOS *taos, const char *db) {
   STscObj *pObj = (STscObj *)taos;
@@ -365,16 +381,24 @@ bool taos_is_null(TAOS_RES *res, int32_t row, int32_t col) {
 }
 
 bool taos_is_update_query(TAOS_RES *res) {
-  // TODO
+  return taos_num_fields(res) == 0;
-  return true;
 }
 
 int taos_fetch_block(TAOS_RES *res, TAOS_ROW *rows) {
-  if (res == NULL) {
+  int32_t numOfRows = 0;
+  /*int32_t code = */taos_fetch_block_s(res, &numOfRows, rows);
+  return numOfRows;
+}
+
+int taos_fetch_block_s(TAOS_RES *res, int* numOfRows, TAOS_ROW *rows) {
+  SRequestObj *pRequest = (SRequestObj *)res;
+  if (pRequest == NULL) {
     return 0;
   }
 
-  SRequestObj *pRequest = (SRequestObj *)res;
+  (*rows)      = NULL;
+  (*numOfRows) = 0;
+
   if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT || pRequest->type == TSDB_SQL_INSERT ||
       pRequest->code != TSDB_CODE_SUCCESS || taos_num_fields(res) == 0) {
     return 0;
@@ -385,18 +409,63 @@ int taos_fetch_block(TAOS_RES *res, TAOS_ROW *rows) {
   // TODO refactor
   SReqResultInfo *pResultInfo = &pRequest->body.resInfo;
   pResultInfo->current = pResultInfo->numOfRows;
-  *rows = pResultInfo->row;
 
-  return pResultInfo->numOfRows;
+  (*rows)      = pResultInfo->row;
+  (*numOfRows) = pResultInfo->numOfRows;
+  return pRequest->code;
+}
+
+int taos_fetch_raw_block(TAOS_RES *res, int* numOfRows, void** pData) {
+  SRequestObj *pRequest = (SRequestObj *)res;
+  if (pRequest == NULL) {
+    return 0;
+  }
+
+  if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT || pRequest->type == TSDB_SQL_INSERT ||
+      pRequest->code != TSDB_CODE_SUCCESS || taos_num_fields(res) == 0) {
+    return 0;
+  }
+
+  doFetchRow(pRequest, false);
+
+  SReqResultInfo *pResultInfo = &pRequest->body.resInfo;
+
+  pResultInfo->current = pResultInfo->numOfRows;
+  (*numOfRows) = pResultInfo->numOfRows;
+  (*pData) = (void*) pResultInfo->pData;
+
+  return 0;
+}
+
+int *taos_get_column_data_offset(TAOS_RES *res, int columnIndex) {
+  SRequestObj *pRequest = (SRequestObj *)res;
+  if (pRequest == NULL) {
+    return 0;
+  }
+
+  int32_t numOfFields = taos_num_fields(pRequest);
+  if (columnIndex < 0 || columnIndex >= numOfFields || numOfFields == 0) {
+    return 0;
+  }
+
+  TAOS_FIELD* pField = &pRequest->body.resInfo.userFields[columnIndex];
+  if (!IS_VAR_DATA_TYPE(pField->type)) {
+    return 0;
+  }
+
+  return pRequest->body.resInfo.pCol[columnIndex].offset;
 }
 
 int taos_validate_sql(TAOS *taos, const char *sql) { return true; }
 
 void taos_reset_current_db(TAOS *taos) {
-  // TODO
+  if (taos == NULL) {
     return;
   }
 
+  resetConnectDB(taos);
+}
+
 const char *taos_get_server_info(TAOS *taos) {
   if (taos == NULL) {
     return NULL;

source/client/src/tmq.c

@@ -78,6 +78,8 @@ struct tmq_t {
   STscObj*       pTscObj;
   tmq_commit_cb* commit_cb;
   int32_t        nextTopicIdx;
+  int8_t         epStatus;
+  int32_t        epSkipCnt;
   int32_t        waitingRequest;
   int32_t        readyRequest;
   SArray*        clientTopics;  // SArray<SMqClientTopic>
@@ -106,6 +108,7 @@ typedef struct {
   // connection info
   int32_t vgId;
   int32_t vgStatus;
+  int32_t vgSkipCnt;
   SEpSet  epSet;
 } SMqClientVg;
 
@@ -137,6 +140,7 @@ typedef struct {
   tmq_t*          tmq;
   SMqClientVg*    pVg;
   int32_t         epoch;
+  int32_t         vgId;
   tsem_t          rspSem;
   tmq_message_t** msg;
   int32_t         sync;
@@ -255,6 +259,7 @@ void tmqClearUnhandleMsg(tmq_t* tmq) {
       break;
   }
 
+  msg = NULL;
   taosReadAllQitems(tmq->mqueue, tmq->qall);
   while (1) {
     taosGetQitem(tmq->qall, (void**)&msg);
@@ -310,6 +315,8 @@ tmq_t* tmq_consumer_new(void* conn, tmq_conf_t* conf, char* errstr, int32_t errs
   pTmq->epoch = 0;
   pTmq->waitingRequest = 0;
   pTmq->readyRequest = 0;
+  pTmq->epStatus = 0;
+  pTmq->epSkipCnt = 0;
   // set conf
   strcpy(pTmq->clientId, conf->clientId);
   strcpy(pTmq->groupId, conf->groupId);
@@ -345,6 +352,8 @@ tmq_t* tmq_consumer_new1(tmq_conf_t* conf, char* errstr, int32_t errstrLen) {
   pTmq->epoch = 0;
   pTmq->waitingRequest = 0;
   pTmq->readyRequest = 0;
+  pTmq->epStatus = 0;
+  pTmq->epSkipCnt = 0;
   // set conf
   strcpy(pTmq->clientId, conf->clientId);
   strcpy(pTmq->groupId, conf->groupId);
@@ -787,7 +796,7 @@ void tmqShowMsg(tmq_message_t* tmq_message) {
   static bool noPrintSchema;
   char        pBuf[128];
   SMqPollRsp* pRsp = &tmq_message->msg;
-  int32_t     colNum = pRsp->schema->nCols;
+  int32_t     colNum = 2;
   if (!noPrintSchema) {
     printf("|");
     for (int32_t i = 0; i < colNum; i++) {
@@ -831,20 +840,21 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) {
   SMqClientVg*    pVg = pParam->pVg;
   tmq_t*          tmq = pParam->tmq;
   if (code != 0) {
-    tscWarn("msg discard, code:%x", code);
+    tscWarn("msg discard from vg %d, epoch %d, code:%x", pParam->vgId, pParam->epoch, code);
-    goto WRITE_QUEUE_FAIL;
+    goto CREATE_MSG_FAIL;
   }
 
   int32_t msgEpoch = ((SMqRspHead*)pMsg->pData)->epoch;
   int32_t tmqEpoch = atomic_load_32(&tmq->epoch);
   if (msgEpoch < tmqEpoch) {
-    tsem_post(&tmq->rspSem);
+    /*printf("discard rsp epoch %d, current epoch %d\n", msgEpoch, tmqEpoch);*/
-    tscWarn("discard rsp epoch %d, current epoch %d", msgEpoch, tmqEpoch);
+    /*tsem_post(&tmq->rspSem);*/
+    tscWarn("msg discard from vg %d since from earlier epoch, rsp epoch %d, current epoch %d", pParam->vgId, msgEpoch, tmqEpoch);
     return 0;
   }
 
   if (msgEpoch != tmqEpoch) {
-    tscWarn("mismatch rsp epoch %d, current epoch %d", msgEpoch, tmqEpoch);
+    tscWarn("mismatch rsp from vg %d, epoch %d, current epoch %d", pParam->vgId, msgEpoch, tmqEpoch);
   } else {
     atomic_sub_fetch_32(&tmq->waitingRequest, 1);
   }
@@ -871,7 +881,8 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) {
   /*SMqConsumeRsp* pRsp = taosMemoryCalloc(1, sizeof(SMqConsumeRsp));*/
   tmq_message_t* pRsp = taosAllocateQitem(sizeof(tmq_message_t));
   if (pRsp == NULL) {
-    goto WRITE_QUEUE_FAIL;
+    tscWarn("msg discard from vg %d, epoch %d since out of memory", pParam->vgId, pParam->epoch);
+    goto CREATE_MSG_FAIL;
   }
   memcpy(pRsp, pMsg->pData, sizeof(SMqRspHead));
   tDecodeSMqPollRsp(POINTER_SHIFT(pMsg->pData, sizeof(SMqRspHead)), &pRsp->msg);
@@ -880,23 +891,28 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) {
   // TODO: alloc mem
   /*pRsp->*/
   /*printf("rsp commit off:%ld rsp off:%ld has data:%d\n", pRsp->committedOffset, pRsp->rspOffset, pRsp->numOfTopics);*/
+#if 0
   if (pRsp->msg.numOfTopics == 0) {
     /*printf("no data\n");*/
     taosFreeQitem(pRsp);
-    goto WRITE_QUEUE_FAIL;
+    goto CREATE_MSG_FAIL;
   }
+#endif
+
+  tscDebug("consumer %ld recv poll: vg %d, req offset %ld, rsp offset %ld", tmq->consumerId, pParam->pVg->vgId, pRsp->msg.reqOffset,
+           pRsp->msg.rspOffset);
 
   pRsp->vg = pParam->pVg;
   taosWriteQitem(tmq->mqueue, pRsp);
   atomic_add_fetch_32(&tmq->readyRequest, 1);
-  tsem_post(&tmq->rspSem);
+  /*tsem_post(&tmq->rspSem);*/
   return 0;
 
-WRITE_QUEUE_FAIL:
+CREATE_MSG_FAIL:
   if (pParam->epoch == tmq->epoch) {
     atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
   }
-  tsem_post(&tmq->rspSem);
+  /*tsem_post(&tmq->rspSem);*/
   return code;
 }
 
@@ -905,6 +921,7 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
   bool    set = false;
   int32_t topicNumGet = taosArrayGetSize(pRsp->topics);
   char    vgKey[TSDB_TOPIC_FNAME_LEN + 22];
+  tscDebug("consumer %ld update ep epoch %d to epoch %d, topic num: %d", tmq->consumerId, tmq->epoch, epoch, topicNumGet);
   SArray* newTopics = taosArrayInit(topicNumGet, sizeof(SMqClientTopic));
   if (newTopics == NULL) {
     return false;
@@ -922,16 +939,19 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
     taosHashClear(pHash);
     topic.topicName = strdup(pTopicEp->topic);
 
+    tscDebug("consumer %ld update topic: %s", tmq->consumerId, topic.topicName);
     int32_t topicNumCur = taosArrayGetSize(tmq->clientTopics);
     for (int32_t j = 0; j < topicNumCur; j++) {
       // find old topic
       SMqClientTopic* pTopicCur = taosArrayGet(tmq->clientTopics, j);
       if (pTopicCur->vgs && strcmp(pTopicCur->topicName, pTopicEp->topic) == 0) {
         int32_t vgNumCur = taosArrayGetSize(pTopicCur->vgs);
+        tscDebug("consumer %ld new vg num: %d", tmq->consumerId, vgNumCur);
         if (vgNumCur == 0) break;
         for (int32_t k = 0; k < vgNumCur; k++) {
           SMqClientVg* pVgCur = taosArrayGet(pTopicCur->vgs, k);
           sprintf(vgKey, "%s:%d", topic.topicName, pVgCur->vgId);
+          tscDebug("consumer %ld epoch %d vg %d build %s", tmq->consumerId, epoch, pVgCur->vgId, vgKey);
           taosHashPut(pHash, vgKey, strlen(vgKey), &pVgCur->currentOffset, sizeof(int64_t));
         }
         break;
@@ -945,15 +965,19 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
       sprintf(vgKey, "%s:%d", topic.topicName, pVgEp->vgId);
       int64_t* pOffset = taosHashGet(pHash, vgKey, strlen(vgKey));
       int64_t  offset = pVgEp->offset;
+      tscDebug("consumer %ld epoch %d vg %d offset og to %ld", tmq->consumerId, epoch, pVgEp->vgId, offset);
       if (pOffset != NULL) {
         offset = *pOffset;
+        tscDebug("consumer %ld epoch %d vg %d found %s", tmq->consumerId, epoch, pVgEp->vgId, vgKey);
       }
+      tscDebug("consumer %ld epoch %d vg %d offset set to %ld", tmq->consumerId, epoch, pVgEp->vgId, offset);
       SMqClientVg clientVg = {
           .pollCnt = 0,
           .currentOffset = offset,
           .vgId = pVgEp->vgId,
           .epSet = pVgEp->epSet,
           .vgStatus = TMQ_VG_STATUS__IDLE,
+          .vgSkipCnt = 0,
       };
       taosArrayPush(topic.vgs, &clientVg);
       set = true;
@@ -971,7 +995,7 @@ int32_t tmqAskEpCb(void* param, const SDataBuf* pMsg, int32_t code) {
   SMqAskEpCbParam* pParam = (SMqAskEpCbParam*)param;
   tmq_t*           tmq = pParam->tmq;
   if (code != 0) {
-    tscError("get topic endpoint error, not ready, wait:%d\n", pParam->sync);
+    tscError("consumer %ld get topic endpoint error, not ready, wait:%d", tmq->consumerId, pParam->sync);
     goto END;
   }
 
@@ -980,6 +1004,7 @@ int32_t tmqAskEpCb(void* param, const SDataBuf* pMsg, int32_t code) {
   // Epoch will only increase when received newer epoch ep msg
   SMqRspHead* head = pMsg->pData;
   int32_t     epoch = atomic_load_32(&tmq->epoch);
+  tscDebug("consumer %ld recv ep, msg epoch %d, current epoch %d", tmq->consumerId, head->epoch, epoch);
   if (head->epoch <= epoch) {
     goto END;
   }
@@ -1004,10 +1029,11 @@ int32_t tmqAskEpCb(void* param, const SDataBuf* pMsg, int32_t code) {
     tDecodeSMqCMGetSubEpRsp(POINTER_SHIFT(pMsg->pData, sizeof(SMqRspHead)), pRsp);
 
     taosWriteQitem(tmq->mqueue, pRsp);
-    tsem_post(&tmq->rspSem);
+    /*tsem_post(&tmq->rspSem);*/
   }
 
 END:
+  atomic_store_8(&tmq->epStatus, 0);
   if (pParam->sync) {
     tsem_post(&pParam->rspSem);
   }
@@ -1015,10 +1041,18 @@ END:
 }
 
 int32_t tmqAskEp(tmq_t* tmq, bool sync) {
+  int8_t epStatus = atomic_val_compare_exchange_8(&tmq->epStatus, 0, 1);
+  if (epStatus == 1) {
+    int32_t epSkipCnt = atomic_add_fetch_32(&tmq->epSkipCnt, 1);
+    tscTrace("consumer %ld skip ask ep cnt %d", tmq->consumerId, epSkipCnt);
+    if (epSkipCnt < 5000) return 0;
+  }
+  atomic_store_32(&tmq->epSkipCnt, 0);
   int32_t           tlen = sizeof(SMqCMGetSubEpReq);
   SMqCMGetSubEpReq* req = taosMemoryMalloc(tlen);
   if (req == NULL) {
     tscError("failed to malloc get subscribe ep buf");
+    atomic_store_8(&tmq->epStatus, 0);
     return -1;
   }
   req->consumerId = htobe64(tmq->consumerId);
@@ -1029,6 +1063,7 @@ int32_t tmqAskEp(tmq_t* tmq, bool sync) {
   if (pParam == NULL) {
     tscError("failed to malloc subscribe param");
     taosMemoryFree(req);
+    atomic_store_8(&tmq->epStatus, 0);
     return -1;
   }
   pParam->tmq = tmq;
@@ -1040,6 +1075,7 @@ int32_t tmqAskEp(tmq_t* tmq, bool sync) {
     tsem_destroy(&pParam->rspSem);
     taosMemoryFree(pParam);
     taosMemoryFree(req);
+    atomic_store_8(&tmq->epStatus, 0);
     return -1;
   }
 
@@ -1057,6 +1093,8 @@ int32_t tmqAskEp(tmq_t* tmq, bool sync) {
 
   SEpSet epSet = getEpSet_s(&tmq->pTscObj->pAppInfo->mgmtEp);
 
+  tscDebug("consumer %ld ask ep", tmq->consumerId);
+
   int64_t transporterId = 0;
   asyncSendMsgToServer(tmq->pTscObj->pAppInfo->pTransporter, &epSet, &transporterId, sendInfo);
 
@@ -1111,6 +1149,7 @@ SMqPollReq* tmqBuildConsumeReqImpl(tmq_t* tmq, int64_t blockingTime, SMqClientTo
   pReq->consumerId = tmq->consumerId;
   pReq->epoch = tmq->epoch;
   pReq->currentOffset = reqOffset;
+  pReq->reqId = generateRequestId();
 
   pReq->head.vgId = htonl(pVg->vgId);
   pReq->head.contLen = htonl(sizeof(SMqPollReq));
@@ -1195,23 +1234,35 @@ int32_t tmqPollImpl(tmq_t* tmq, int64_t blockingTime) {
       SMqClientVg* pVg = taosArrayGet(pTopic->vgs, j);
       int32_t      vgStatus = atomic_val_compare_exchange_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE, TMQ_VG_STATUS__WAIT);
       if (vgStatus != TMQ_VG_STATUS__IDLE) {
+        int32_t vgSkipCnt = atomic_add_fetch_32(&pVg->vgSkipCnt, 1);
+        tscTrace("consumer %ld epoch %d skip vg %d skip cnt %d", tmq->consumerId, tmq->epoch, pVg->vgId, vgSkipCnt);
         continue;
+        /*if (vgSkipCnt < 10000) continue;*/
+#if 0
+        if (skipCnt < 30000) {
+          continue;
+        } else {
+        tscDebug("consumer %ld skip vg %d skip too much reset", tmq->consumerId, pVg->vgId);
         }
+#endif
+      }
+      atomic_store_32(&pVg->vgSkipCnt, 0);
       SMqPollReq* pReq = tmqBuildConsumeReqImpl(tmq, blockingTime, pTopic, pVg);
       if (pReq == NULL) {
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
-        tsem_post(&tmq->rspSem);
+        /*tsem_post(&tmq->rspSem);*/
         return -1;
       }
       SMqPollCbParam* pParam = taosMemoryMalloc(sizeof(SMqPollCbParam));
       if (pParam == NULL) {
         taosMemoryFree(pReq);
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
-        tsem_post(&tmq->rspSem);
+        /*tsem_post(&tmq->rspSem);*/
         return -1;
       }
       pParam->tmq = tmq;
       pParam->pVg = pVg;
+      pParam->vgId = pVg->vgId;
       pParam->epoch = tmq->epoch;
       pParam->sync = 0;
 
@@ -1220,7 +1271,7 @@ int32_t tmqPollImpl(tmq_t* tmq, int64_t blockingTime) {
         taosMemoryFree(pReq);
         taosMemoryFree(pParam);
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
-        tsem_post(&tmq->rspSem);
+        /*tsem_post(&tmq->rspSem);*/
         return -1;
       }
 
@@ -1229,7 +1280,7 @@ int32_t tmqPollImpl(tmq_t* tmq, int64_t blockingTime) {
           .len = sizeof(SMqPollReq),
           .handle = NULL,
       };
-      sendInfo->requestId = generateRequestId();
+      sendInfo->requestId = pReq->reqId;
       sendInfo->requestObjRefId = 0;
       sendInfo->param = pParam;
       sendInfo->fp = tmqPollCb;
@@ -1238,6 +1289,8 @@ int32_t tmqPollImpl(tmq_t* tmq, int64_t blockingTime) {
       int64_t transporterId = 0;
       /*printf("send poll\n");*/
       atomic_add_fetch_32(&tmq->waitingRequest, 1);
+      tscDebug("consumer %ld send poll to %s : vg %d, epoch %d, req offset %ld, reqId %lu", tmq->consumerId, pTopic->topicName, pVg->vgId, tmq->epoch, pVg->currentOffset, pReq->reqId);
+      /*printf("send vg %d %ld\n", pVg->vgId, pVg->currentOffset);*/
       asyncSendMsgToServer(tmq->pTscObj->pAppInfo->pTransporter, &pVg->epSet, &transporterId, sendInfo);
       pVg->pollCnt++;
       tmq->pollCnt++;
@@ -1247,13 +1300,13 @@ int32_t tmqPollImpl(tmq_t* tmq, int64_t blockingTime) {
 }
 
 // return
-int32_t tmqHandleRes(tmq_t* tmq, SMqRspHead* rspHead, bool* pReset) {
+int32_t tmqHandleNoPollRsp(tmq_t* tmq, SMqRspHead* rspHead, bool* pReset) {
   if (rspHead->mqMsgType == TMQ_MSG_TYPE__EP_RSP) {
     /*printf("ep %d %d\n", rspMsg->head.epoch, tmq->epoch);*/
     if (rspHead->epoch > atomic_load_32(&tmq->epoch)) {
       SMqCMGetSubEpRsp* rspMsg = (SMqCMGetSubEpRsp*)rspHead;
       tmqUpdateEp(tmq, rspHead->epoch, rspMsg);
-      tmqClearUnhandleMsg(tmq);
+      /*tmqClearUnhandleMsg(tmq);*/
       *pReset = true;
     } else {
       *pReset = false;
@@ -1284,6 +1337,11 @@ tmq_message_t* tmqHandleAllRsp(tmq_t* tmq, int64_t blockingTime, bool pollIfRese
         /*printf("vg %d offset %ld up to %ld\n", pVg->vgId, pVg->currentOffset, rspMsg->msg.rspOffset);*/
         pVg->currentOffset = rspMsg->msg.rspOffset;
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
+        if (rspMsg->msg.numOfTopics == 0) {
+          taosFreeQitem(rspMsg);
+          rspHead = NULL;
+          continue;
+        }
         return rspMsg;
       } else {
         /*printf("epoch mismatch\n");*/
@@ -1292,10 +1350,10 @@ tmq_message_t* tmqHandleAllRsp(tmq_t* tmq, int64_t blockingTime, bool pollIfRese
     } else {
       /*printf("handle ep rsp %d\n", rspMsg->head.mqMsgType);*/
       bool reset = false;
-      tmqHandleRes(tmq, rspHead, &reset);
+      tmqHandleNoPollRsp(tmq, rspHead, &reset);
       taosFreeQitem(rspHead);
       if (pollIfReset && reset) {
-        printf("reset and repoll\n");
+        tscDebug("consumer %ld reset and repoll", tmq->consumerId);
         tmqPollImpl(tmq, blockingTime);
       }
     }
@@ -1345,7 +1403,7 @@ tmq_message_t* tmq_consumer_poll(tmq_t* tmq, int64_t blocking_time) {
     tmqAskEp(tmq, false);
     tmqPollImpl(tmq, blocking_time);
 
-    tsem_wait(&tmq->rspSem);
+    /*tsem_wait(&tmq->rspSem);*/
 
     rspMsg = tmqHandleAllRsp(tmq, blocking_time, false);
     if (rspMsg) {
@@ -1354,6 +1412,7 @@ tmq_message_t* tmq_consumer_poll(tmq_t* tmq, int64_t blocking_time) {
     if (blocking_time != 0) {
       int64_t endTime = taosGetTimestampMs();
       if (endTime - startTime > blocking_time) {
+        tscDebug("consumer %ld (epoch %d) timeout, no rsp", tmq->consumerId, tmq->epoch);
         return NULL;
       }
     }
@@ -1534,24 +1593,3 @@ TAOS_ROW tmq_get_row(tmq_message_t* message) {
 }
 
 char* tmq_get_topic_name(tmq_message_t* message) { return "not implemented yet"; }
-
-#if 0
-tmq_t* tmqCreateConsumerImpl(TAOS* conn, tmq_conf_t* conf) {
-  tmq_t* pTmq = taosMemoryMalloc(sizeof(tmq_t));
-  if (pTmq == NULL) {
-    return NULL;
-  }
-  strcpy(pTmq->groupId, conf->groupId);
-  strcpy(pTmq->clientId, conf->clientId);
-  pTmq->pTscObj = (STscObj*)conn;
-  pTmq->pTscObj->connType = HEARTBEAT_TYPE_MQ;
-  return pTmq;
-}
-
-
-static void destroySendMsgInfo(SMsgSendInfo* pMsgBody) {
-  assert(pMsgBody != NULL);
-  taosMemoryFreeClear(pMsgBody->msgInfo.pData);
-  taosMemoryFreeClear(pMsgBody);
-}
-#endif

source/client/test/clientTests.cpp

@@ -400,6 +400,7 @@ TEST(testCase, show_vgroup_Test) {
   taos_free_result(pRes);
   taos_close(pConn);
 }
+#endif
 
 TEST(testCase, create_multiple_tables) {
   TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
@@ -458,7 +459,7 @@ TEST(testCase, create_multiple_tables) {
 
   taos_free_result(pRes);
 
-  for (int32_t i = 0; i < 20; ++i) {
+  for (int32_t i = 0; i < 25000; ++i) {
     char sql[512] = {0};
     snprintf(sql, tListLen(sql),
              "create table t_x_%d using st1 tags(2) t_x_%d using st1 tags(5) t_x_%d using st1 tags(911)", i,
@@ -652,7 +653,6 @@ TEST(testCase, projection_query_stables) {
   taos_free_result(pRes);
   taos_close(pConn);
 }
-#endif
 
 TEST(testCase, agg_query_tables) {
   TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);

source/common/src/tdatablock.c

@@ -17,6 +17,7 @@
 #include "tdatablock.h"
 #include "tcompare.h"
 #include "tglobal.h"
+#include "tlog.h"
 
 int32_t taosGetFqdnPortFromEp(const char* ep, SEp* pEp) {
   pEp->port = 0;
@@ -130,44 +131,7 @@ int32_t colDataAppend(SColumnInfoData* pColumnInfoData, uint32_t currentRow, con
     memcpy(pColumnInfoData->pData + len, pData, varDataTLen(pData));
     pColumnInfoData->varmeta.length += varDataTLen(pData);
   } else {
-    char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
+    memcpy(pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow, pData, pColumnInfoData->info.bytes);
-    switch (type) {
-      case TSDB_DATA_TYPE_BOOL: {
-        *(bool*)p = *(bool*)pData;
-        break;
-      }
-      case TSDB_DATA_TYPE_TINYINT:
-      case TSDB_DATA_TYPE_UTINYINT: {
-        *(int8_t*)p = *(int8_t*)pData;
-        break;
-      }
-      case TSDB_DATA_TYPE_SMALLINT:
-      case TSDB_DATA_TYPE_USMALLINT: {
-        *(int16_t*)p = *(int16_t*)pData;
-        break;
-      }
-      case TSDB_DATA_TYPE_INT:
-      case TSDB_DATA_TYPE_UINT: {
-        *(int32_t*)p = *(int32_t*)pData;
-        break;
-      }
-      case TSDB_DATA_TYPE_TIMESTAMP:
-      case TSDB_DATA_TYPE_BIGINT:
-      case TSDB_DATA_TYPE_UBIGINT: {
-        *(int64_t*)p = *(int64_t*)pData;
-        break;
-      }
-      case TSDB_DATA_TYPE_FLOAT: {
-        *(float*)p = *(float*)pData;
-        break;
-      }
-      case TSDB_DATA_TYPE_DOUBLE: {
-        *(double*)p = *(double*)pData;
-        break;
-      }
-      default:
-        assert(0);
-    }
   }
 
   return 0;
@@ -175,6 +139,8 @@ int32_t colDataAppend(SColumnInfoData* pColumnInfoData, uint32_t currentRow, con
 
 static void doBitmapMerge(SColumnInfoData* pColumnInfoData, int32_t numOfRow1, const SColumnInfoData* pSource,
                           int32_t numOfRow2) {
+  if (numOfRow2 <= 0) return;
+
   uint32_t total = numOfRow1 + numOfRow2;
 
   if (BitmapLen(numOfRow1) < BitmapLen(total)) {
@@ -189,22 +155,32 @@ static void doBitmapMerge(SColumnInfoData* pColumnInfoData, int32_t numOfRow1, c
 
   if (remindBits == 0) {  // no need to shift bits of bitmap
     memcpy(pColumnInfoData->nullbitmap + BitmapLen(numOfRow1), pSource->nullbitmap, BitmapLen(numOfRow2));
-  } else {
+    return;
+  }
+
+  uint8_t* p = (uint8_t*)pSource->nullbitmap;
+  pColumnInfoData->nullbitmap[BitmapLen(numOfRow1) - 1] |= (p[0] >> remindBits);  // copy remind bits
+
+  if (BitmapLen(numOfRow1) == BitmapLen(total)) {
+    return;
+  }
+
   int32_t len = BitmapLen(numOfRow2);
   int32_t i = 0;
 
-    uint8_t* p = (uint8_t*)pSource->nullbitmap;
-    pColumnInfoData->nullbitmap[BitmapLen(numOfRow1) - 1] |= (p[0] >> remindBits);
-
   uint8_t* start = (uint8_t*)&pColumnInfoData->nullbitmap[BitmapLen(numOfRow1)];
-    while (i < len) {
+  int32_t overCount = BitmapLen(total) - BitmapLen(numOfRow1);
-      start[i] |= (p[i] << shiftBits);
+  while (i < len) { // size limit of pSource->nullbitmap
-      i += 1;
+    if (i >= 1) {
+      start[i - 1] |= (p[i] >> remindBits); //copy remind bits
+    }
 
-      if (i > 1) {
+    if (i >= overCount) { // size limit of pColumnInfoData->nullbitmap
-        start[i - 1] |= (p[i] >> remindBits);
+      return;
-      }
     }
+
+    start[i] |= (p[i] << shiftBits);  //copy shift bits
+    i += 1;
   }
 }
 
@@ -216,6 +192,9 @@ int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, co
     return numOfRow1;
   }
 
+  if (pSource->hasNull) {
+    pColumnInfoData->hasNull = pSource->hasNull;
+  }
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
     // Handle the bitmap
     char* p = taosMemoryRealloc(pColumnInfoData->varmeta.offset, sizeof(int32_t) * (numOfRow1 + numOfRow2));
@@ -336,13 +315,18 @@ int32_t blockDataUpdateTsWindow(SSDataBlock* pDataBlock) {
   return 0;
 }
 
-int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc) {
+// if pIndexMap = NULL, merger one column by on column
-  assert(pSrc != NULL && pDest != NULL && pDest->info.numOfCols == pSrc->info.numOfCols);
+int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc, SArray* pIndexMap) {
+  assert(pSrc != NULL && pDest != NULL);
 
-  int32_t numOfCols = pSrc->info.numOfCols;
+  int32_t numOfCols = pDest->info.numOfCols;
   for (int32_t i = 0; i < numOfCols; ++i) {
+    int32_t mapIndex = i;
+    if(pIndexMap) {
+      mapIndex = *(int32_t*)taosArrayGet(pIndexMap, i);
+    }
     SColumnInfoData* pCol2 = taosArrayGet(pDest->pDataBlock, i);
-    SColumnInfoData* pCol1 = taosArrayGet(pSrc->pDataBlock, i);
+    SColumnInfoData* pCol1 = taosArrayGet(pSrc->pDataBlock, mapIndex);
 
     uint32_t oldLen = colDataGetLength(pCol2, pDest->info.rows);
     uint32_t newLen = colDataGetLength(pCol1, pSrc->info.rows);
@@ -399,26 +383,23 @@ int32_t blockDataSplitRows(SSDataBlock* pBlock, bool hasVarCol, int32_t startInd
   // TODO speedup by checking if the whole page can fit in firstly.
   if (!hasVarCol) {
     size_t  rowSize = blockDataGetRowSize(pBlock);
-    int32_t capacity = (payloadSize / (rowSize * 8 + bitmapChar * numOfCols)) * 8;
+    int32_t capacity = payloadSize / (rowSize + numOfCols * bitmapChar / 8.0);
+    ASSERT(capacity > 0);
 
-    *stopIndex = startIndex + capacity;
+    *stopIndex = startIndex + capacity - 1;
     if (*stopIndex >= numOfRows) {
       *stopIndex = numOfRows - 1;
     }
 
     return TSDB_CODE_SUCCESS;
-  } else {
+  }
   // iterate the rows that can be fit in this buffer page
   int32_t size = (headerSize + colHeaderSize);
-
   for (int32_t j = startIndex; j < numOfRows; ++j) {
     for (int32_t i = 0; i < numOfCols; ++i) {
       SColumnInfoData* pColInfoData = TARRAY_GET_ELEM(pBlock->pDataBlock, i);
       if (IS_VAR_DATA_TYPE(pColInfoData->info.type)) {
-          bool isNull = colDataIsNull(pColInfoData, numOfRows, j, NULL);
+        if (pColInfoData->varmeta.offset[j] != -1) {
-          if (isNull) {
-            // do nothing
-          } else {
           char* p = colDataGetData(pColInfoData, j);
           size += varDataTLen(p);
         }
@@ -433,9 +414,9 @@ int32_t blockDataSplitRows(SSDataBlock* pBlock, bool hasVarCol, int32_t startInd
       }
     }
 
-      if (size > pageSize) {
+    if (size > pageSize) {  // pageSize must be able to hold one row
       *stopIndex = j - 1;
-        ASSERT(*stopIndex > startIndex);
+      ASSERT(*stopIndex >= startIndex);
 
       return TSDB_CODE_SUCCESS;
     }
@@ -444,7 +425,7 @@ int32_t blockDataSplitRows(SSDataBlock* pBlock, bool hasVarCol, int32_t startInd
   // all fit in
   *stopIndex = numOfRows - 1;
   return TSDB_CODE_SUCCESS;
-  }
+
 }
 
 SSDataBlock* blockDataExtractBlock(SSDataBlock* pBlock, int32_t startIndex, int32_t rowCount) {
@@ -547,8 +528,13 @@ int32_t blockDataFromBuf(SSDataBlock* pBlock, const char* buf) {
   for (int32_t i = 0; i < numOfCols; ++i) {
     SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, i);
 
-    size_t metaSize = pBlock->info.rows * sizeof(int32_t);
     if (IS_VAR_DATA_TYPE(pCol->info.type)) {
+      size_t metaSize = pBlock->info.rows * sizeof(int32_t);
+      char* tmp = taosMemoryRealloc(pCol->varmeta.offset, metaSize);    // preview calloc is too small
+      if (tmp == NULL) {
+        return TSDB_CODE_OUT_OF_MEMORY;
+      }
+      pCol->varmeta.offset = (int32_t*)tmp;
       memcpy(pCol->varmeta.offset, pStart, metaSize);
       pStart += metaSize;
     } else {
@@ -581,6 +567,49 @@ int32_t blockDataFromBuf(SSDataBlock* pBlock, const char* buf) {
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t blockDataFromBuf1(SSDataBlock* pBlock, const char* buf, size_t capacity) {
+  pBlock->info.rows = *(int32_t*)buf;
+
+  int32_t     numOfCols = pBlock->info.numOfCols;
+  const char* pStart = buf + sizeof(uint32_t);
+
+  for (int32_t i = 0; i < numOfCols; ++i) {
+    SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, i);
+
+    if (IS_VAR_DATA_TYPE(pCol->info.type)) {
+      size_t metaSize = capacity * sizeof(int32_t);
+      memcpy(pCol->varmeta.offset, pStart, metaSize);
+      pStart += metaSize;
+    } else {
+      memcpy(pCol->nullbitmap, pStart, BitmapLen(capacity));
+      pStart += BitmapLen(capacity);
+    }
+
+    int32_t colLength = *(int32_t*)pStart;
+    pStart += sizeof(int32_t);
+
+    if (IS_VAR_DATA_TYPE(pCol->info.type)) {
+      if (pCol->varmeta.allocLen < colLength) {
+        char* tmp = taosMemoryRealloc(pCol->pData, colLength);
+        if (tmp == NULL) {
+          return TSDB_CODE_OUT_OF_MEMORY;
+        }
+
+        pCol->pData = tmp;
+        pCol->varmeta.allocLen = colLength;
+      }
+
+      pCol->varmeta.length = colLength;
+      ASSERT(pCol->varmeta.length <= pCol->varmeta.allocLen);
+    }
+
+    memcpy(pCol->pData, pStart, colLength);
+    pStart += pCol->info.bytes * capacity;
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
 size_t blockDataGetRowSize(SSDataBlock* pBlock) {
   ASSERT(pBlock != NULL);
   if (pBlock->info.rowSize == 0) {
@@ -627,6 +656,10 @@ double blockDataGetSerialRowSize(const SSDataBlock* pBlock) {
   return rowSize;
 }
 
+int32_t getAllowedRowsForPage(const SSDataBlock* pBlock, size_t pgSize) {
+  return (int32_t) ((pgSize - blockDataGetSerialMetaSize(pBlock))/ blockDataGetSerialRowSize(pBlock));
+}
+
 typedef struct SSDataBlockSortHelper {
   SArray*      orderInfo;  // SArray<SBlockOrderInfo>
   SSDataBlock* pDataBlock;
@@ -665,23 +698,13 @@ int32_t dataBlockCompar(const void* p1, const void* p2, const void* param) {
     void* left1 = colDataGetData(pColInfoData, left);
     void* right1 = colDataGetData(pColInfoData, right);
 
-    switch (pColInfoData->info.type) {
+    __compar_fn_t fn = getKeyComparFunc(pColInfoData->info.type, pOrder->order);
-      case TSDB_DATA_TYPE_INT: {
-        int32_t leftx = *(int32_t*)left1;
-        int32_t rightx = *(int32_t*)right1;
 
-        if (leftx == rightx) {
+    int ret = fn(left1, right1);
-          break;
+    if (ret == 0) {
+      continue;
     } else {
-          if (pOrder->order == TSDB_ORDER_ASC) {
+      return ret;
-            return (leftx < rightx) ? -1 : 1;
-          } else {
-            return (leftx < rightx) ? 1 : -1;
-          }
-        }
-      }
-      default:
-        assert(0);
     }
   }
 
@@ -735,61 +758,12 @@ static int32_t blockDataAssign(SColumnInfoData* pCols, const SSDataBlock* pDataB
         pDst->varmeta.offset[j] = pSrc->varmeta.offset[index[j]];
       }
     } else {
-      switch (pSrc->info.type) {
-        case TSDB_DATA_TYPE_UINT:
-        case TSDB_DATA_TYPE_INT: {
       for (int32_t j = 0; j < pDataBlock->info.rows; ++j) {
-            int32_t* p = (int32_t*)pDst->pData;
-            int32_t* srclist = (int32_t*)pSrc->pData;
-
-            p[j] = srclist[index[j]];
         if (colDataIsNull_f(pSrc->nullbitmap, index[j])) {
           colDataSetNull_f(pDst->nullbitmap, j);
+          continue;
         }
-          }
+        memcpy(pDst->pData + j * pDst->info.bytes, pSrc->pData + index[j] * pDst->info.bytes, pDst->info.bytes);
-          break;
-        }
-        case TSDB_DATA_TYPE_UTINYINT:
-        case TSDB_DATA_TYPE_TINYINT: {
-          for (int32_t j = 0; j < pDataBlock->info.rows; ++j) {
-            int32_t* p = (int32_t*)pDst->pData;
-            int32_t* srclist = (int32_t*)pSrc->pData;
-
-            p[j] = srclist[index[j]];
-            if (colDataIsNull_f(pSrc->nullbitmap, index[j])) {
-              colDataSetNull_f(pDst->nullbitmap, j);
-            }
-          }
-          break;
-        }
-        case TSDB_DATA_TYPE_USMALLINT:
-        case TSDB_DATA_TYPE_SMALLINT: {
-          for (int32_t j = 0; j < pDataBlock->info.rows; ++j) {
-            int32_t* p = (int32_t*)pDst->pData;
-            int32_t* srclist = (int32_t*)pSrc->pData;
-
-            p[j] = srclist[index[j]];
-            if (colDataIsNull_f(pSrc->nullbitmap, index[j])) {
-              colDataSetNull_f(pDst->nullbitmap, j);
-            }
-          }
-          break;
-        }
-        case TSDB_DATA_TYPE_UBIGINT:
-        case TSDB_DATA_TYPE_BIGINT: {
-          for (int32_t j = 0; j < pDataBlock->info.rows; ++j) {
-            int32_t* p = (int32_t*)pDst->pData;
-            int32_t* srclist = (int32_t*)pSrc->pData;
-
-            p[j] = srclist[index[j]];
-            if (colDataIsNull_f(pSrc->nullbitmap, index[j])) {
-              colDataSetNull_f(pDst->nullbitmap, j);
-            }
-          }
-          break;
-        }
-        default:
-          assert(0);
       }
     }
   }
@@ -862,33 +836,6 @@ static int32_t* createTupleIndex(size_t rows) {
 
 static void destroyTupleIndex(int32_t* index) { taosMemoryFreeClear(index); }
 
-static __compar_fn_t getComparFn(int32_t type, int32_t order) {
-  switch (type) {
-    case TSDB_DATA_TYPE_TINYINT:
-      return order == TSDB_ORDER_ASC ? compareInt8Val : compareInt8ValDesc;
-    case TSDB_DATA_TYPE_SMALLINT:
-      return order == TSDB_ORDER_ASC ? compareInt16Val : compareInt16ValDesc;
-    case TSDB_DATA_TYPE_INT:
-      return order == TSDB_ORDER_ASC ? compareInt32Val : compareInt32ValDesc;
-    case TSDB_DATA_TYPE_BIGINT:
-      return order == TSDB_ORDER_ASC ? compareInt64Val : compareInt64ValDesc;
-    case TSDB_DATA_TYPE_FLOAT:
-      return order == TSDB_ORDER_ASC ? compareFloatVal : compareFloatValDesc;
-    case TSDB_DATA_TYPE_DOUBLE:
-      return order == TSDB_ORDER_ASC ? compareDoubleVal : compareDoubleValDesc;
-    case TSDB_DATA_TYPE_UTINYINT:
-      return order == TSDB_ORDER_ASC ? compareUint8Val : compareUint8ValDesc;
-    case TSDB_DATA_TYPE_USMALLINT:
-      return order == TSDB_ORDER_ASC ? compareUint16Val : compareUint16ValDesc;
-    case TSDB_DATA_TYPE_UINT:
-      return order == TSDB_ORDER_ASC ? compareUint32Val : compareUint32ValDesc;
-    case TSDB_DATA_TYPE_UBIGINT:
-      return order == TSDB_ORDER_ASC ? compareUint64Val : compareUint64ValDesc;
-    default:
-      return order == TSDB_ORDER_ASC ? compareInt32Val : compareInt32ValDesc;
-  }
-}
-
 int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
   ASSERT(pDataBlock != NULL && pOrderInfo != NULL);
   if (pDataBlock->info.rows <= 1) {
@@ -922,11 +869,11 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
 
         int64_t p0 = taosGetTimestampUs();
 
-        __compar_fn_t fn = getComparFn(pColInfoData->info.type, pOrder->order);
+        __compar_fn_t fn = getKeyComparFunc(pColInfoData->info.type, pOrder->order);
         qsort(pColInfoData->pData, pDataBlock->info.rows, pColInfoData->info.bytes, fn);
 
         int64_t p1 = taosGetTimestampUs();
-        printf("sort:%" PRId64 ", rows:%d\n", p1 - p0, pDataBlock->info.rows);
+        uDebug("blockDataSort easy cost:%" PRId64 ", rows:%d\n", p1 - p0, pDataBlock->info.rows);
 
         return TSDB_CODE_SUCCESS;
       } else {  // var data type
@@ -955,24 +902,21 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
 
   SColumnInfoData* pCols = createHelpColInfoData(pDataBlock);
   if (pCols == NULL) {
+    destroyTupleIndex(index);
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return terrno;
   }
 
   int64_t p2 = taosGetTimestampUs();
 
-  int32_t code = blockDataAssign(pCols, pDataBlock, index);
+  blockDataAssign(pCols, pDataBlock, index);
-  if (code != TSDB_CODE_SUCCESS) {
-    terrno = code;
-    return code;
-  }
 
   int64_t p3 = taosGetTimestampUs();
 
   copyBackToBlock(pDataBlock, pCols);
   int64_t p4 = taosGetTimestampUs();
 
-  printf("sort:%" PRId64 ", create:%" PRId64 ", assign:%" PRId64 ", copyback:%" PRId64 ", rows:%d\n", p1 - p0, p2 - p1,
+  uDebug("blockDataSort complex sort:%" PRId64 ", create:%" PRId64 ", assign:%" PRId64 ", copyback:%" PRId64 ", rows:%d\n", p1 - p0, p2 - p1,
          p3 - p2, p4 - p3, rows);
   destroyTupleIndex(index);
 
@@ -1139,7 +1083,7 @@ void blockDataCleanup(SSDataBlock* pDataBlock) {
   }
 }
 
-int32_t blockDataEnsureColumnCapacity(SColumnInfoData* pColumn, uint32_t numOfRows) {
+int32_t colInfoDataEnsureCapacity(SColumnInfoData* pColumn, uint32_t numOfRows) {
   if (0 == numOfRows) {
     return TSDB_CODE_SUCCESS;
   }
@@ -1178,10 +1122,13 @@ int32_t blockDataEnsureColumnCapacity(SColumnInfoData* pColumn, uint32_t numOfRo
 
 int32_t blockDataEnsureCapacity(SSDataBlock* pDataBlock, uint32_t numOfRows) {
   int32_t code = 0;
+  if (numOfRows == 0) {
+    return TSDB_CODE_SUCCESS;
+  }
 
   for (int32_t i = 0; i < pDataBlock->info.numOfCols; ++i) {
     SColumnInfoData* p = taosArrayGet(pDataBlock->pDataBlock, i);
-    code = blockDataEnsureColumnCapacity(p, numOfRows);
+    code = colInfoDataEnsureCapacity(p, numOfRows);
     if (code) {
       return code;
     }
@@ -1201,6 +1148,9 @@ void* blockDataDestroy(SSDataBlock* pBlock) {
 }
 
 SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock) {
+  if(pDataBlock == NULL){
+    return NULL;
+  }
   int32_t numOfCols = pDataBlock->info.numOfCols;
 
   SSDataBlock* pBlock = taosMemoryCalloc(1, sizeof(SSDataBlock));
@@ -1220,7 +1170,7 @@ SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock) {
 }
 
 size_t blockDataGetCapacityInRow(const SSDataBlock* pBlock, size_t pageSize) {
-  return pageSize / (blockDataGetSerialRowSize(pBlock) + blockDataGetSerialMetaSize(pBlock));
+  return (int32_t) ((pageSize - blockDataGetSerialMetaSize(pBlock))/ blockDataGetSerialRowSize(pBlock));
 }
 
 void colDataDestroy(SColumnInfoData* pColData) {
@@ -1233,6 +1183,63 @@ void colDataDestroy(SColumnInfoData* pColData) {
   taosMemoryFree(pColData->pData);
 }
 
+
+static void doShiftBitmap(char* nullBitmap, size_t n, size_t total) {
+  int32_t len = BitmapLen(total);
+
+  int32_t newLen = BitmapLen(total - n);
+  if (n%8 == 0) {
+    memmove(nullBitmap, nullBitmap + n/8, newLen);
+  } else {
+    int32_t tail = n % 8;
+    int32_t i = 0;
+
+    uint8_t* p = (uint8_t*) nullBitmap;
+    while(i < len) {
+      uint8_t v = p[i];
+
+      p[i] = 0;
+      p[i] = (v << tail);
+
+      if (i < len - 1) {
+        uint8_t next = p[i + 1];
+        p[i] |= (next >> (8 - tail));
+      }
+
+      i += 1;
+    }
+  }
+}
+
+static void colDataTrimFirstNRows(SColumnInfoData* pColInfoData, size_t n, size_t total) {
+  if (IS_VAR_DATA_TYPE(pColInfoData->info.type)) {
+    memmove(pColInfoData->varmeta.offset, &pColInfoData->varmeta.offset[n], (total - n));
+    memset(&pColInfoData->varmeta.offset[total - n - 1], 0, n);
+  } else {
+    int32_t bytes = pColInfoData->info.bytes;
+    memmove(pColInfoData->pData, ((char*)pColInfoData->pData + n * bytes), (total - n) * bytes);
+    doShiftBitmap(pColInfoData->nullbitmap, n, total);
+  }
+}
+
+int32_t blockDataTrimFirstNRows(SSDataBlock *pBlock, size_t n) {
+  if (n == 0) {
+    return TSDB_CODE_SUCCESS;
+  }
+
+  if (pBlock->info.rows <= n) {
+    blockDataCleanup(pBlock);
+  } else {
+    for(int32_t i = 0; i < pBlock->info.numOfCols; ++i) {
+      SColumnInfoData* pColInfoData = taosArrayGet(pBlock->pDataBlock, i);
+      colDataTrimFirstNRows(pColInfoData, n, pBlock->info.rows);
+    }
+
+    pBlock->info.rows -= n;
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
 int32_t tEncodeDataBlock(void** buf, const SSDataBlock* pBlock) {
   int64_t tbUid = pBlock->info.uid;
   int16_t numOfCols = pBlock->info.numOfCols;
@@ -1372,6 +1379,7 @@ static char* formatTimestamp(char* buf, int64_t val, int precision) {
 
   return buf;
 }
+
 void blockDebugShowData(const SArray* dataBlocks) {
   char    pBuf[128];
   int32_t sz = taosArrayGetSize(dataBlocks);

source/common/src/tdataformat.c

@@ -33,7 +33,8 @@ int tdAllocMemForCol(SDataCol *pCol, int maxPoints) {
   spaceNeeded += (int)nBitmapBytes;
   // TODO: Currently, the compression of bitmap parts is affiliated to the column data parts, thus allocate 1 more
   // TYPE_BYTES as to comprise complete TYPE_BYTES. Otherwise, invalid read/write would be triggered.
-  spaceNeeded += TYPE_BYTES[pCol->type];
+  // spaceNeeded += TYPE_BYTES[pCol->type]; // the bitmap part is append as a single part since 2022.04.03, thus remove
+  // the additional space
 #endif
 
   if (pCol->spaceSize < spaceNeeded) {
@@ -47,6 +48,7 @@ int tdAllocMemForCol(SDataCol *pCol, int maxPoints) {
     }
   }
 #ifdef TD_SUPPORT_BITMAP
+
   if (IS_VAR_DATA_TYPE(pCol->type)) {
     pCol->pBitmap = POINTER_SHIFT(pCol->pData, pCol->bytes * maxPoints);
     pCol->dataOff = POINTER_SHIFT(pCol->pBitmap, nBitmapBytes);
@@ -85,6 +87,7 @@ int tdEncodeSchema(void **buf, STSchema *pSchema) {
   for (int i = 0; i < schemaNCols(pSchema); i++) {
     STColumn *pCol = schemaColAt(pSchema, i);
     tlen += taosEncodeFixedI8(buf, colType(pCol));
+    tlen += taosEncodeFixedI8(buf, colSma(pCol));
     tlen += taosEncodeFixedI16(buf, colColId(pCol));
     tlen += taosEncodeFixedI16(buf, colBytes(pCol));
   }
@@ -107,12 +110,14 @@ void *tdDecodeSchema(void *buf, STSchema **pRSchema) {
 
   for (int i = 0; i < numOfCols; i++) {
     col_type_t  type = 0;
+    int8_t      sma = TSDB_BSMA_TYPE_NONE;
     col_id_t    colId = 0;
     col_bytes_t bytes = 0;
     buf = taosDecodeFixedI8(buf, &type);
+    buf = taosDecodeFixedI8(buf, &sma);
     buf = taosDecodeFixedI16(buf, &colId);
     buf = taosDecodeFixedI32(buf, &bytes);
-    if (tdAddColToSchema(&schemaBuilder, type, colId, bytes) < 0) {
+    if (tdAddColToSchema(&schemaBuilder, type, sma, colId, bytes) < 0) {
       tdDestroyTSchemaBuilder(&schemaBuilder);
       return NULL;
     }
@@ -148,7 +153,7 @@ void tdResetTSchemaBuilder(STSchemaBuilder *pBuilder, schema_ver_t version) {
   pBuilder->version = version;
 }
 
-int tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, col_id_t colId, col_bytes_t bytes) {
+int32_t tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t sma, col_id_t colId, col_bytes_t bytes) {
   if (!isValidDataType(type)) return -1;
 
   if (pBuilder->nCols >= pBuilder->tCols) {
@@ -161,6 +166,7 @@ int tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, col_id_t colId, col
   STColumn *pCol = &(pBuilder->columns[pBuilder->nCols]);
   colSetType(pCol, type);
   colSetColId(pCol, colId);
+  colSetSma(pCol, sma);
   if (pBuilder->nCols == 0) {
     colSetOffset(pCol, 0);
   } else {
@@ -168,9 +174,6 @@ int tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, col_id_t colId, col
     colSetOffset(pCol, pTCol->offset + TYPE_BYTES[pTCol->type]);
   }
 
-  // TODO: set sma value by user input
-  pCol->sma = 1;
-
   if (IS_VAR_DATA_TYPE(type)) {
     colSetBytes(pCol, bytes);
     pBuilder->tlen += (TYPE_BYTES[type] + bytes);
@@ -305,7 +308,7 @@ static FORCE_INLINE const void *tdGetColDataOfRowUnsafe(SDataCol *pCol, int row)
 
 bool isNEleNull(SDataCol *pCol, int nEle) {
   if (isAllRowsNull(pCol)) return true;
-  for (int i = 0; i < nEle; i++) {
+  for (int i = 0; i < nEle; ++i) {
     if (!isNull(tdGetColDataOfRowUnsafe(pCol, i), pCol->type)) return false;
   }
   return true;
@@ -326,7 +329,7 @@ static FORCE_INLINE void dataColSetNullAt(SDataCol *pCol, int index) {
 static void dataColSetNEleNull(SDataCol *pCol, int nEle) {
   if (IS_VAR_DATA_TYPE(pCol->type)) {
     pCol->len = 0;
-    for (int i = 0; i < nEle; i++) {
+    for (int i = 0; i < nEle; ++i) {
       dataColSetNullAt(pCol, i);
     }
   } else {
@@ -342,7 +345,7 @@ void *dataColSetOffset(SDataCol *pCol, int nEle) {
   // char *tptr = (char *)(pCol->pData);
 
   VarDataOffsetT offset = 0;
-  for (int i = 0; i < nEle; i++) {
+  for (int i = 0; i < nEle; ++i) {
     pCol->dataOff[i] = offset;
     offset += varDataTLen(tptr);
     tptr = POINTER_SHIFT(tptr, varDataTLen(tptr));
@@ -370,6 +373,7 @@ SDataCols *tdNewDataCols(int maxCols, int maxRows) {
       tdFreeDataCols(pCols);
       return NULL;
     }
+#if 0  // no need as calloc used
     int i;
     for (i = 0; i < maxCols; i++) {
       pCols->cols[i].spaceSize = 0;
@@ -377,6 +381,7 @@ SDataCols *tdNewDataCols(int maxCols, int maxRows) {
       pCols->cols[i].pData = NULL;
       pCols->cols[i].dataOff = NULL;
     }
+#endif
   }
 
   return pCols;
@@ -390,17 +395,21 @@ int tdInitDataCols(SDataCols *pCols, STSchema *pSchema) {
     void *ptr = (SDataCol *)taosMemoryRealloc(pCols->cols, sizeof(SDataCol) * pCols->maxCols);
     if (ptr == NULL) return -1;
     pCols->cols = ptr;
-    for (i = oldMaxCols; i < pCols->maxCols; i++) {
+    for (i = oldMaxCols; i < pCols->maxCols; ++i) {
       pCols->cols[i].pData = NULL;
       pCols->cols[i].dataOff = NULL;
+      pCols->cols[i].pBitmap = NULL;
       pCols->cols[i].spaceSize = 0;
     }
   }
+#if 0
+  tdResetDataCols(pCols); // redundant loop to reset len/blen to 0, already reset in following dataColInit(...)
+#endif
 
-  tdResetDataCols(pCols);
+  pCols->numOfRows = 0;
   pCols->numOfCols = schemaNCols(pSchema);
 
-  for (i = 0; i < schemaNCols(pSchema); i++) {
+  for (i = 0; i < schemaNCols(pSchema); ++i) {
     dataColInit(pCols->cols + i, schemaColAt(pSchema, i), pCols->maxPoints);
   }
 
@@ -412,7 +421,7 @@ SDataCols *tdFreeDataCols(SDataCols *pCols) {
   if (pCols) {
     if (pCols->cols) {
       int maxCols = pCols->maxCols;
-      for (i = 0; i < maxCols; i++) {
+      for (i = 0; i < maxCols; ++i) {
         SDataCol *pCol = &pCols->cols[i];
         taosMemoryFreeClear(pCol->pData);
       }
@@ -463,7 +472,7 @@ SDataCols *tdDupDataCols(SDataCols *pDataCols, bool keepData) {
 void tdResetDataCols(SDataCols *pCols) {
   if (pCols != NULL) {
     pCols->numOfRows = 0;
-    for (int i = 0; i < pCols->maxCols; i++) {
+    for (int i = 0; i < pCols->maxCols; ++i) {
       dataColReset(pCols->cols + i);
     }
   }

source/common/src/tglobal.c

@@ -30,25 +30,45 @@ char     tsLocalEp[TSDB_EP_LEN] = {0};  // Local End Point, hostname:port
 uint16_t tsServerPort = 6030;
 int32_t  tsVersion = 30000000;
 int32_t  tsStatusInterval = 1;  // second
-bool     tsEnableTelemetryReporting = 0;
+bool     tsEnableTelemetryReporting = false;
 
 // common
 int32_t tsRpcTimer = 300;
 int32_t tsRpcMaxTime = 600;    // seconds;
-bool    tsRpcForceTcp = 1;   // disable this, means query, show command use udp protocol as default
+bool    tsRpcForceTcp = true;  // disable this, means query, show command use udp protocol as default
 int32_t tsMaxShellConns = 50000;
 int32_t tsMaxConnections = 50000;
 int32_t tsShellActivityTimer = 3;  // second
-float   tsNumOfThreadsPerCore = 1.0f;
-int32_t tsNumOfCommitThreads = 4;
-float   tsRatioOfQueryCores = 1.0f;
 int32_t tsMaxBinaryDisplayWidth = 30;
-bool    tsEnableSlaveQuery = 1;
+bool    tsEnableSlaveQuery = true;
-bool    tsPrintAuth = 0;
+bool    tsPrintAuth = false;
-bool    tsMultiProcess = 0;
+
+// multi process
+bool    tsMultiProcess = false;
+int32_t tsMnodeShmSize = TSDB_MAX_WAL_SIZE * 2;
+int32_t tsVnodeShmSize = TSDB_MAX_WAL_SIZE * 10;
+int32_t tsQnodeShmSize = TSDB_MAX_WAL_SIZE * 4;
+int32_t tsSnodeShmSize = TSDB_MAX_WAL_SIZE * 4;
+int32_t tsBnodeShmSize = TSDB_MAX_WAL_SIZE * 4;
+
+// queue & threads
+int32_t tsNumOfRpcThreads = 1;
+int32_t tsNumOfCommitThreads = 2;
+int32_t tsNumOfTaskQueueThreads = 1;
+int32_t tsNumOfMnodeQueryThreads = 1;
+int32_t tsNumOfMnodeReadThreads = 1;
+int32_t tsNumOfVnodeQueryThreads = 2;
+int32_t tsNumOfVnodeFetchThreads = 2;
+int32_t tsNumOfVnodeWriteThreads = 2;
+int32_t tsNumOfVnodeSyncThreads = 2;
+int32_t tsNumOfVnodeMergeThreads = 2;
+int32_t tsNumOfQnodeQueryThreads = 2;
+int32_t tsNumOfQnodeFetchThreads = 2;
+int32_t tsNumOfSnodeSharedThreads = 2;
+int32_t tsNumOfSnodeUniqueThreads = 2;
 
 // monitor
-bool     tsEnableMonitor = 1;
+bool     tsEnableMonitor = true;
 int32_t  tsMonitorInterval = 30;
 char     tsMonitorFqdn[TSDB_FQDN_LEN] = {0};
 uint16_t tsMonitorPort = 6043;
@@ -117,13 +137,13 @@ 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 = 0;
+bool tsRetrieveBlockingModel = false;
 
 // last_row(*), first(*), last_row(ts, col1, col2) query, the result fields will be the original column name
-bool tsKeepOriginalColumnName = 0;
+bool tsKeepOriginalColumnName = false;
 
 // kill long query
-bool tsDeadLockKillQuery = 0;
+bool tsDeadLockKillQuery = false;
 
 // tsdb config
 // For backward compatibility
@@ -283,7 +303,6 @@ static int32_t taosAddClientCfg(SConfig *pCfg) {
   if (cfgAddInt32(pCfg, "serverPort", defaultServerPort, 1, 65056, 1) != 0) return -1;
   if (cfgAddDir(pCfg, "tempDir", tsTempDir, 1) != 0) return -1;
   if (cfgAddFloat(pCfg, "minimalTempDirGB", 1.0f, 0.001f, 10000000, 1) != 0) return -1;
-  if (cfgAddFloat(pCfg, "numOfThreadsPerCore", tsNumOfThreadsPerCore, 0, 10, 1) != 0) return -1;
   if (cfgAddInt32(pCfg, "maxTmrCtrl", tsMaxTmrCtrl, 8, 2048, 1) != 0) return -1;
   if (cfgAddInt32(pCfg, "rpcTimer", tsRpcTimer, 100, 3000, 1) != 0) return -1;
   if (cfgAddInt32(pCfg, "rpcMaxTime", tsRpcMaxTime, 100, 7200, 1) != 0) return -1;
@@ -297,6 +316,11 @@ static int32_t taosAddClientCfg(SConfig *pCfg) {
     return -1;
   if (cfgAddBool(pCfg, "keepColumnName", tsKeepOriginalColumnName, 1) != 0) return -1;
   if (cfgAddInt32(pCfg, "maxBinaryDisplayWidth", tsMaxBinaryDisplayWidth, 1, 65536, 1) != 0) return -1;
+
+  tsNumOfTaskQueueThreads = tsNumOfCores / 4;
+  tsNumOfTaskQueueThreads = TRANGE(tsNumOfTaskQueueThreads, 1, 2);
+  if (cfgAddInt32(pCfg, "numOfTaskQueueThreads", tsNumOfTaskQueueThreads, 1, 1024, 0) != 0) return -1;
+
   return 0;
 }
 
@@ -329,8 +353,6 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
   if (cfgAddInt32(pCfg, "supportVnodes", 256, 0, 65536, 0) != 0) return -1;
   if (cfgAddDir(pCfg, "dataDir", tsDataDir, 0) != 0) return -1;
   if (cfgAddFloat(pCfg, "minimalDataDirGB", 2.0f, 0.001f, 10000000, 0) != 0) return -1;
-  if (cfgAddInt32(pCfg, "numOfCommitThreads", tsNumOfCommitThreads, 1, 100, 0) != 0) return -1;
-  if (cfgAddFloat(pCfg, "ratioOfQueryCores", tsRatioOfQueryCores, 0, 2, 0) != 0) return -1;
   if (cfgAddInt32(pCfg, "maxNumOfDistinctRes", tsMaxNumOfDistinctResults, 10 * 10000, 10000 * 10000, 0) != 0) return -1;
   if (cfgAddBool(pCfg, "telemetryReporting", tsEnableTelemetryReporting, 0) != 0) return -1;
   if (cfgAddInt32(pCfg, "maxConnections", tsMaxConnections, 1, 100000, 0) != 0) return -1;
@@ -347,7 +369,65 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
   if (cfgAddBool(pCfg, "printAuth", tsPrintAuth, 0) != 0) return -1;
   if (cfgAddBool(pCfg, "slaveQuery", tsEnableSlaveQuery, 0) != 0) return -1;
   if (cfgAddBool(pCfg, "deadLockKillQuery", tsDeadLockKillQuery, 0) != 0) return -1;
+
   if (cfgAddBool(pCfg, "multiProcess", tsMultiProcess, 0) != 0) return -1;
+  if (cfgAddInt32(pCfg, "mnodeShmSize", tsMnodeShmSize, 4096, INT32_MAX, 0) != 0) return -1;
+  if (cfgAddInt32(pCfg, "vnodeShmSize", tsVnodeShmSize, 4096, INT32_MAX, 0) != 0) return -1;
+  if (cfgAddInt32(pCfg, "qnodeShmSize", tsQnodeShmSize, 4096, INT32_MAX, 0) != 0) return -1;
+  if (cfgAddInt32(pCfg, "snodeShmSize", tsSnodeShmSize, 4096, INT32_MAX, 0) != 0) return -1;
+  if (cfgAddInt32(pCfg, "bnodeShmSize", tsBnodeShmSize, 4096, INT32_MAX, 0) != 0) return -1;
+
+  tsNumOfRpcThreads = tsNumOfCores / 2;
+  tsNumOfRpcThreads = TRANGE(tsNumOfRpcThreads, 1, 4);
+  if (cfgAddInt32(pCfg, "numOfRpcThreads", tsNumOfRpcThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfCommitThreads = tsNumOfCores / 2;
+  tsNumOfCommitThreads = TRANGE(tsNumOfCommitThreads, 2, 4);
+  if (cfgAddInt32(pCfg, "numOfCommitThreads", tsNumOfCommitThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfMnodeQueryThreads = tsNumOfCores / 8;
+  tsNumOfMnodeQueryThreads = TRANGE(tsNumOfMnodeQueryThreads, 1, 4);
+  if (cfgAddInt32(pCfg, "numOfMnodeQueryThreads", tsNumOfMnodeQueryThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfMnodeReadThreads = tsNumOfCores / 8;
+  tsNumOfMnodeReadThreads = TRANGE(tsNumOfMnodeReadThreads, 1, 4);
+  if (cfgAddInt32(pCfg, "numOfMnodeReadThreads", tsNumOfMnodeReadThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfVnodeQueryThreads = tsNumOfCores / 2;
+  tsNumOfVnodeQueryThreads = TMAX(tsNumOfVnodeQueryThreads, 1);
+  if (cfgAddInt32(pCfg, "numOfVnodeQueryThreads", tsNumOfVnodeQueryThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfVnodeFetchThreads = tsNumOfCores / 2;
+  tsNumOfVnodeFetchThreads = TRANGE(tsNumOfVnodeFetchThreads, 2, 4);
+  if (cfgAddInt32(pCfg, "numOfVnodeFetchThreads", tsNumOfVnodeFetchThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfVnodeWriteThreads = tsNumOfCores;
+  tsNumOfVnodeWriteThreads = TMAX(tsNumOfVnodeWriteThreads, 1);
+  if (cfgAddInt32(pCfg, "numOfVnodeWriteThreads", tsNumOfVnodeWriteThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfVnodeSyncThreads = tsNumOfCores / 2;
+  tsNumOfVnodeSyncThreads = TMAX(tsNumOfVnodeSyncThreads, 1);
+  if (cfgAddInt32(pCfg, "numOfVnodeSyncThreads", tsNumOfVnodeSyncThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfVnodeMergeThreads = tsNumOfCores / 8;
+  tsNumOfVnodeMergeThreads = TRANGE(tsNumOfVnodeMergeThreads, 1, 1);
+  if (cfgAddInt32(pCfg, "numOfVnodeMergeThreads", tsNumOfVnodeMergeThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfQnodeQueryThreads = tsNumOfCores / 2;
+  tsNumOfQnodeQueryThreads = TMAX(tsNumOfQnodeQueryThreads, 1);
+  if (cfgAddInt32(pCfg, "numOfQnodeQueryThreads", tsNumOfQnodeQueryThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfQnodeFetchThreads = tsNumOfCores / 2;
+  tsNumOfQnodeFetchThreads = TRANGE(tsNumOfQnodeFetchThreads, 2, 4);
+  if (cfgAddInt32(pCfg, "numOfQnodeFetchThreads", tsNumOfQnodeFetchThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfSnodeSharedThreads = tsNumOfCores / 4;
+  tsNumOfSnodeSharedThreads = TRANGE(tsNumOfSnodeSharedThreads, 2, 4);
+  if (cfgAddInt32(pCfg, "numOfSnodeSharedThreads", tsNumOfSnodeSharedThreads, 1, 1024, 0) != 0) return -1;
+
+  tsNumOfSnodeUniqueThreads = tsNumOfCores / 4;
+  tsNumOfSnodeUniqueThreads = TRANGE(tsNumOfSnodeUniqueThreads, 2, 4);
+  if (cfgAddInt32(pCfg, "numOfSnodeUniqueThreads", tsNumOfSnodeUniqueThreads, 1, 1024, 0) != 0) return -1;
 
   if (cfgAddBool(pCfg, "monitor", tsEnableMonitor, 0) != 0) return -1;
   if (cfgAddInt32(pCfg, "monitorInterval", tsMonitorInterval, 1, 360000, 0) != 0) return -1;
@@ -411,7 +491,6 @@ static int32_t taosSetClientCfg(SConfig *pCfg) {
     return -1;
   }
 
-  tsNumOfThreadsPerCore = cfgGetItem(pCfg, "numOfThreadsPerCore")->fval;
   tsMaxTmrCtrl = cfgGetItem(pCfg, "maxTmrCtrl")->i32;
   tsRpcTimer = cfgGetItem(pCfg, "rpcTimer")->i32;
   tsRpcMaxTime = cfgGetItem(pCfg, "rpcMaxTime")->i32;
@@ -424,7 +503,7 @@ static int32_t taosSetClientCfg(SConfig *pCfg) {
   tsMaxNumOfOrderedResults = cfgGetItem(pCfg, "maxNumOfOrderedRes")->i32;
   tsKeepOriginalColumnName = cfgGetItem(pCfg, "keepColumnName")->bval;
   tsMaxBinaryDisplayWidth = cfgGetItem(pCfg, "maxBinaryDisplayWidth")->i32;
-
+  tsNumOfTaskQueueThreads = cfgGetItem(pCfg, "numOfTaskQueueThreads")->i32;
   return 0;
 }
 
@@ -448,8 +527,6 @@ static void taosSetSystemCfg(SConfig *pCfg) {
 
 static int32_t taosSetServerCfg(SConfig *pCfg) {
   tsDataSpace.reserved = cfgGetItem(pCfg, "minimalDataDirGB")->fval;
-  tsNumOfCommitThreads = cfgGetItem(pCfg, "numOfCommitThreads")->i32;
-  tsRatioOfQueryCores = cfgGetItem(pCfg, "ratioOfQueryCores")->fval;
   tsMaxNumOfDistinctResults = cfgGetItem(pCfg, "maxNumOfDistinctRes")->i32;
   tsEnableTelemetryReporting = cfgGetItem(pCfg, "telemetryReporting")->bval;
   tsMaxConnections = cfgGetItem(pCfg, "maxConnections")->i32;
@@ -466,7 +543,27 @@ static int32_t taosSetServerCfg(SConfig *pCfg) {
   tsPrintAuth = cfgGetItem(pCfg, "printAuth")->bval;
   tsEnableSlaveQuery = cfgGetItem(pCfg, "slaveQuery")->bval;
   tsDeadLockKillQuery = cfgGetItem(pCfg, "deadLockKillQuery")->bval;
+
   tsMultiProcess = cfgGetItem(pCfg, "multiProcess")->bval;
+  tsMnodeShmSize = cfgGetItem(pCfg, "mnodeShmSize")->i32;
+  tsVnodeShmSize = cfgGetItem(pCfg, "vnodeShmSize")->i32;
+  tsQnodeShmSize = cfgGetItem(pCfg, "qnodeShmSize")->i32;
+  tsSnodeShmSize = cfgGetItem(pCfg, "snodeShmSize")->i32;
+  tsBnodeShmSize = cfgGetItem(pCfg, "bnodeShmSize")->i32;
+
+  tsNumOfRpcThreads = cfgGetItem(pCfg, "numOfRpcThreads")->i32;
+  tsNumOfCommitThreads = cfgGetItem(pCfg, "numOfCommitThreads")->i32;
+  tsNumOfMnodeQueryThreads = cfgGetItem(pCfg, "numOfMnodeQueryThreads")->i32;
+  tsNumOfMnodeReadThreads = cfgGetItem(pCfg, "numOfMnodeReadThreads")->i32;
+  tsNumOfVnodeQueryThreads = cfgGetItem(pCfg, "numOfVnodeQueryThreads")->i32;
+  tsNumOfVnodeFetchThreads = cfgGetItem(pCfg, "numOfVnodeFetchThreads")->i32;
+  tsNumOfVnodeWriteThreads = cfgGetItem(pCfg, "numOfVnodeWriteThreads")->i32;
+  tsNumOfVnodeSyncThreads = cfgGetItem(pCfg, "numOfVnodeSyncThreads")->i32;
+  tsNumOfVnodeMergeThreads = cfgGetItem(pCfg, "numOfVnodeMergeThreads")->i32;
+  tsNumOfQnodeQueryThreads = cfgGetItem(pCfg, "numOfQnodeQueryThreads")->i32;
+  tsNumOfQnodeFetchThreads = cfgGetItem(pCfg, "numOfQnodeFetchThreads")->i32;
+  tsNumOfSnodeSharedThreads = cfgGetItem(pCfg, "numOfSnodeSharedThreads")->i32;
+  tsNumOfSnodeUniqueThreads = cfgGetItem(pCfg, "numOfSnodeUniqueThreads")->i32;
 
   tsEnableMonitor = cfgGetItem(pCfg, "monitor")->bval;
   tsMonitorInterval = cfgGetItem(pCfg, "monitorInterval")->i32;

source/common/src/tmsg.c

@@ -296,33 +296,30 @@ int32_t tSerializeSVCreateTbReq(void **buf, SVCreateTbReq *pReq) {
   switch (pReq->type) {
     case TD_SUPER_TABLE:
       tlen += taosEncodeFixedI64(buf, pReq->stbCfg.suid);
-      tlen += taosEncodeFixedU32(buf, pReq->stbCfg.nCols);
+      tlen += taosEncodeFixedI16(buf, pReq->stbCfg.nCols);
-      for (uint32_t i = 0; i < pReq->stbCfg.nCols; i++) {
+      tlen += taosEncodeFixedI16(buf, pReq->stbCfg.nBSmaCols);
+      for (col_id_t i = 0; i < pReq->stbCfg.nCols; ++i) {
         tlen += taosEncodeFixedI8(buf, pReq->stbCfg.pSchema[i].type);
+        tlen += taosEncodeFixedI8(buf, pReq->stbCfg.pSchema[i].sma);
         tlen += taosEncodeFixedI16(buf, pReq->stbCfg.pSchema[i].colId);
         tlen += taosEncodeFixedI32(buf, pReq->stbCfg.pSchema[i].bytes);
         tlen += taosEncodeString(buf, pReq->stbCfg.pSchema[i].name);
       }
-      tlen += taosEncodeFixedU32(buf, pReq->stbCfg.nTagCols);
+      tlen += taosEncodeFixedI16(buf, pReq->stbCfg.nTagCols);
-      for (uint32_t i = 0; i < pReq->stbCfg.nTagCols; i++) {
+      for (col_id_t i = 0; i < pReq->stbCfg.nTagCols; ++i) {
         tlen += taosEncodeFixedI8(buf, pReq->stbCfg.pTagSchema[i].type);
         tlen += taosEncodeFixedI16(buf, pReq->stbCfg.pTagSchema[i].colId);
         tlen += taosEncodeFixedI32(buf, pReq->stbCfg.pTagSchema[i].bytes);
         tlen += taosEncodeString(buf, pReq->stbCfg.pTagSchema[i].name);
       }
-      tlen += taosEncodeFixedI16(buf, pReq->stbCfg.nBSmaCols);
-      for (col_id_t i = 0; i < pReq->stbCfg.nBSmaCols; ++i) {
-        tlen += taosEncodeFixedI16(buf, pReq->stbCfg.pBSmaCols[i]);
-      }
       if (pReq->rollup && pReq->stbCfg.pRSmaParam) {
         SRSmaParam *param = pReq->stbCfg.pRSmaParam;
-        tlen += taosEncodeFixedU32(buf, (uint32_t)param->xFilesFactor);
+        tlen += taosEncodeBinary(buf, (const void *)&param->xFilesFactor, sizeof(param->xFilesFactor));
-        tlen += taosEncodeFixedI8(buf, param->delayUnit);
+        tlen += taosEncodeFixedI32(buf, param->delay);
         tlen += taosEncodeFixedI8(buf, param->nFuncIds);
         for (int8_t i = 0; i < param->nFuncIds; ++i) {
           tlen += taosEncodeFixedI32(buf, param->pFuncIds[i]);
         }
-        tlen += taosEncodeFixedI64(buf, param->delay);
       }
       break;
     case TD_CHILD_TABLE:
@@ -330,26 +327,23 @@ int32_t tSerializeSVCreateTbReq(void **buf, SVCreateTbReq *pReq) {
       tlen += tdEncodeKVRow(buf, pReq->ctbCfg.pTag);
       break;
     case TD_NORMAL_TABLE:
-      tlen += taosEncodeFixedU32(buf, pReq->ntbCfg.nCols);
+      tlen += taosEncodeFixedI16(buf, pReq->ntbCfg.nCols);
-      for (uint32_t i = 0; i < pReq->ntbCfg.nCols; i++) {
+      tlen += taosEncodeFixedI16(buf, pReq->ntbCfg.nBSmaCols);
+      for (col_id_t i = 0; i < pReq->ntbCfg.nCols; ++i) {
         tlen += taosEncodeFixedI8(buf, pReq->ntbCfg.pSchema[i].type);
+        tlen += taosEncodeFixedI8(buf, pReq->ntbCfg.pSchema[i].sma);
         tlen += taosEncodeFixedI16(buf, pReq->ntbCfg.pSchema[i].colId);
         tlen += taosEncodeFixedI32(buf, pReq->ntbCfg.pSchema[i].bytes);
         tlen += taosEncodeString(buf, pReq->ntbCfg.pSchema[i].name);
       }
-      tlen += taosEncodeFixedI16(buf, pReq->ntbCfg.nBSmaCols);
-      for (col_id_t i = 0; i < pReq->ntbCfg.nBSmaCols; ++i) {
-        tlen += taosEncodeFixedI16(buf, pReq->ntbCfg.pBSmaCols[i]);
-      }
       if (pReq->rollup && pReq->ntbCfg.pRSmaParam) {
         SRSmaParam *param = pReq->ntbCfg.pRSmaParam;
-        tlen += taosEncodeFixedU32(buf, (uint32_t)param->xFilesFactor);
+        tlen += taosEncodeBinary(buf, (const void *)&param->xFilesFactor, sizeof(param->xFilesFactor));
-        tlen += taosEncodeFixedI8(buf, param->delayUnit);
+        tlen += taosEncodeFixedI32(buf, param->delay);
         tlen += taosEncodeFixedI8(buf, param->nFuncIds);
         for (int8_t i = 0; i < param->nFuncIds; ++i) {
           tlen += taosEncodeFixedI32(buf, param->pFuncIds[i]);
         }
-        tlen += taosEncodeFixedI64(buf, param->delay);
       }
       break;
     default:
@@ -370,45 +364,38 @@ void *tDeserializeSVCreateTbReq(void *buf, SVCreateTbReq *pReq) {
   switch (pReq->type) {
     case TD_SUPER_TABLE:
       buf = taosDecodeFixedI64(buf, &(pReq->stbCfg.suid));
-      buf = taosDecodeFixedU32(buf, &(pReq->stbCfg.nCols));
+      buf = taosDecodeFixedI16(buf, &(pReq->stbCfg.nCols));
-      pReq->stbCfg.pSchema = (SSchema *)taosMemoryMalloc(pReq->stbCfg.nCols * sizeof(SSchema));
+      buf = taosDecodeFixedI16(buf, &(pReq->stbCfg.nBSmaCols));
-      for (uint32_t i = 0; i < pReq->stbCfg.nCols; i++) {
+      pReq->stbCfg.pSchema = (SSchemaEx *)taosMemoryMalloc(pReq->stbCfg.nCols * sizeof(SSchemaEx));
+      for (col_id_t i = 0; i < pReq->stbCfg.nCols; ++i) {
         buf = taosDecodeFixedI8(buf, &(pReq->stbCfg.pSchema[i].type));
+        buf = taosDecodeFixedI8(buf, &(pReq->stbCfg.pSchema[i].sma));
         buf = taosDecodeFixedI16(buf, &(pReq->stbCfg.pSchema[i].colId));
         buf = taosDecodeFixedI32(buf, &(pReq->stbCfg.pSchema[i].bytes));
         buf = taosDecodeStringTo(buf, pReq->stbCfg.pSchema[i].name);
       }
-      buf = taosDecodeFixedU32(buf, &pReq->stbCfg.nTagCols);
+      buf = taosDecodeFixedI16(buf, &pReq->stbCfg.nTagCols);
       pReq->stbCfg.pTagSchema = (SSchema *)taosMemoryMalloc(pReq->stbCfg.nTagCols * sizeof(SSchema));
-      for (uint32_t i = 0; i < pReq->stbCfg.nTagCols; i++) {
+      for (col_id_t i = 0; i < pReq->stbCfg.nTagCols; ++i) {
         buf = taosDecodeFixedI8(buf, &(pReq->stbCfg.pTagSchema[i].type));
         buf = taosDecodeFixedI16(buf, &pReq->stbCfg.pTagSchema[i].colId);
         buf = taosDecodeFixedI32(buf, &pReq->stbCfg.pTagSchema[i].bytes);
         buf = taosDecodeStringTo(buf, pReq->stbCfg.pTagSchema[i].name);
       }
-      buf = taosDecodeFixedI16(buf, &(pReq->stbCfg.nBSmaCols));
-      if (pReq->stbCfg.nBSmaCols > 0) {
-        pReq->stbCfg.pBSmaCols = (col_id_t *)taosMemoryMalloc(pReq->stbCfg.nBSmaCols * sizeof(col_id_t));
-        for (col_id_t i = 0; i < pReq->stbCfg.nBSmaCols; ++i) {
-          buf = taosDecodeFixedI16(buf, pReq->stbCfg.pBSmaCols + i);
-        }
-      } else {
-        pReq->stbCfg.pBSmaCols = NULL;
-      }
       if (pReq->rollup) {
         pReq->stbCfg.pRSmaParam = (SRSmaParam *)taosMemoryMalloc(sizeof(SRSmaParam));
         SRSmaParam *param = pReq->stbCfg.pRSmaParam;
-        buf = taosDecodeFixedU32(buf, (uint32_t *)&param->xFilesFactor);
+        buf = taosDecodeBinaryTo(buf, (void*)&param->xFilesFactor, sizeof(param->xFilesFactor));
-        buf = taosDecodeFixedI8(buf, &param->delayUnit);
+        buf = taosDecodeFixedI32(buf, &param->delay);
         buf = taosDecodeFixedI8(buf, &param->nFuncIds);
         if (param->nFuncIds > 0) {
+          param->pFuncIds = (func_id_t *)taosMemoryMalloc(param->nFuncIds * sizeof(func_id_t));
           for (int8_t i = 0; i < param->nFuncIds; ++i) {
             buf = taosDecodeFixedI32(buf, param->pFuncIds + i);
           }
         } else {
           param->pFuncIds = NULL;
         }
-        buf = taosDecodeFixedI64(buf, &param->delay);
       } else {
         pReq->stbCfg.pRSmaParam = NULL;
       }
@@ -418,37 +405,30 @@ void *tDeserializeSVCreateTbReq(void *buf, SVCreateTbReq *pReq) {
       buf = tdDecodeKVRow(buf, &pReq->ctbCfg.pTag);
       break;
     case TD_NORMAL_TABLE:
-      buf = taosDecodeFixedU32(buf, &pReq->ntbCfg.nCols);
+      buf = taosDecodeFixedI16(buf, &pReq->ntbCfg.nCols);
-      pReq->ntbCfg.pSchema = (SSchema *)taosMemoryMalloc(pReq->ntbCfg.nCols * sizeof(SSchema));
+      buf = taosDecodeFixedI16(buf, &(pReq->ntbCfg.nBSmaCols));
-      for (uint32_t i = 0; i < pReq->ntbCfg.nCols; i++) {
+      pReq->ntbCfg.pSchema = (SSchemaEx *)taosMemoryMalloc(pReq->ntbCfg.nCols * sizeof(SSchemaEx));
+      for (col_id_t i = 0; i < pReq->ntbCfg.nCols; ++i) {
         buf = taosDecodeFixedI8(buf, &pReq->ntbCfg.pSchema[i].type);
+        buf = taosDecodeFixedI8(buf, &pReq->ntbCfg.pSchema[i].sma);
         buf = taosDecodeFixedI16(buf, &pReq->ntbCfg.pSchema[i].colId);
         buf = taosDecodeFixedI32(buf, &pReq->ntbCfg.pSchema[i].bytes);
         buf = taosDecodeStringTo(buf, pReq->ntbCfg.pSchema[i].name);
       }
-      buf = taosDecodeFixedI16(buf, &(pReq->ntbCfg.nBSmaCols));
-      if (pReq->ntbCfg.nBSmaCols > 0) {
-        pReq->ntbCfg.pBSmaCols = (col_id_t *)taosMemoryMalloc(pReq->ntbCfg.nBSmaCols * sizeof(col_id_t));
-        for (col_id_t i = 0; i < pReq->ntbCfg.nBSmaCols; ++i) {
-          buf = taosDecodeFixedI16(buf, pReq->ntbCfg.pBSmaCols + i);
-        }
-      } else {
-        pReq->ntbCfg.pBSmaCols = NULL;
-      }
       if (pReq->rollup) {
         pReq->ntbCfg.pRSmaParam = (SRSmaParam *)taosMemoryMalloc(sizeof(SRSmaParam));
         SRSmaParam *param = pReq->ntbCfg.pRSmaParam;
-        buf = taosDecodeFixedU32(buf, (uint32_t *)&param->xFilesFactor);
+        buf = taosDecodeBinaryTo(buf, (void*)&param->xFilesFactor, sizeof(param->xFilesFactor));
-        buf = taosDecodeFixedI8(buf, &param->delayUnit);
+        buf = taosDecodeFixedI32(buf, &param->delay);
         buf = taosDecodeFixedI8(buf, &param->nFuncIds);
         if (param->nFuncIds > 0) {
+          param->pFuncIds = (func_id_t *)taosMemoryMalloc(param->nFuncIds * sizeof(func_id_t));
           for (int8_t i = 0; i < param->nFuncIds; ++i) {
             buf = taosDecodeFixedI32(buf, param->pFuncIds + i);
           }
         } else {
           param->pFuncIds = NULL;
         }
-        buf = taosDecodeFixedI64(buf, &param->delay);
       } else {
         pReq->ntbCfg.pRSmaParam = NULL;
       }
@@ -1535,6 +1515,7 @@ int32_t tSerializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq) {
   if (tEncodeI32(&encoder, pReq->maxRows) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->commitTime) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->fsyncPeriod) < 0) return -1;
+  if (tEncodeI32(&encoder, pReq->ttl) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->walLevel) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->precision) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->compression) < 0) return -1;
@@ -1544,6 +1525,7 @@ int32_t tSerializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq) {
   if (tEncodeI8(&encoder, pReq->cacheLastRow) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->ignoreExist) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->streamMode) < 0) return -1;
+  if (tEncodeI8(&encoder, pReq->singleSTable) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->numOfRetensions) < 0) return -1;
   for (int32_t i = 0; i < pReq->numOfRetensions; ++i) {
     SRetention *pRetension = taosArrayGet(pReq->pRetensions, i);
@@ -1576,6 +1558,7 @@ int32_t tDeserializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq)
   if (tDecodeI32(&decoder, &pReq->maxRows) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->commitTime) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->fsyncPeriod) < 0) return -1;
+  if (tDecodeI32(&decoder, &pReq->ttl) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->walLevel) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->precision) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->compression) < 0) return -1;
@@ -1585,6 +1568,7 @@ int32_t tDeserializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq)
   if (tDecodeI8(&decoder, &pReq->cacheLastRow) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->ignoreExist) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->streamMode) < 0) return -1;
+  if (tDecodeI8(&decoder, &pReq->singleSTable) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->numOfRetensions) < 0) return -1;
   pReq->pRetensions = taosArrayInit(pReq->numOfRetensions, sizeof(SRetention));
   if (pReq->pRetensions == NULL) {
@@ -1629,6 +1613,7 @@ int32_t tSerializeSAlterDbReq(void *buf, int32_t bufLen, SAlterDbReq *pReq) {
   if (tEncodeI8(&encoder, pReq->walLevel) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->quorum) < 0) return -1;
   if (tEncodeI8(&encoder, pReq->cacheLastRow) < 0) return -1;
+  if (tEncodeI8(&encoder, pReq->replications) < 0) return -1;
   tEndEncode(&encoder);
 
   int32_t tlen = encoder.pos;
@@ -1650,6 +1635,7 @@ int32_t tDeserializeSAlterDbReq(void *buf, int32_t bufLen, SAlterDbReq *pReq) {
   if (tDecodeI8(&decoder, &pReq->walLevel) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->quorum) < 0) return -1;
   if (tDecodeI8(&decoder, &pReq->cacheLastRow) < 0) return -1;
+  if (tDecodeI8(&decoder, &pReq->replications) < 0) return -1;
   tEndDecode(&decoder);
 
   tCoderClear(&decoder);
@@ -1960,6 +1946,152 @@ void tFreeSUseDbBatchRsp(SUseDbBatchRsp *pRsp) {
   taosArrayDestroy(pRsp->pArray);
 }
 
+int32_t tSerializeSDbCfgReq(void* buf, int32_t bufLen, SDbCfgReq* pReq) {
+  SCoder encoder = {0};
+  tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeCStr(&encoder, pReq->db) < 0) return -1;
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tCoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSDbCfgReq(void* buf, int32_t bufLen, SDbCfgReq* pReq) {
+  SCoder decoder = {0};
+  tCoderInit(&decoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_DECODER);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pReq->db) < 0) return -1;
+  tEndDecode(&decoder);
+
+  tCoderClear(&decoder);
+  return 0;
+}
+
+int32_t tSerializeSDbCfgRsp(void* buf, int32_t bufLen, const SDbCfgRsp* pRsp) {
+  SCoder encoder = {0};
+  tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->numOfVgroups) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->cacheBlockSize) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->totalBlocks) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->daysPerFile) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->daysToKeep0) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->daysToKeep1) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->daysToKeep2) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->minRows) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->maxRows) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->commitTime) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->fsyncPeriod) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->walLevel) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->precision) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->compression) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->replications) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->quorum) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->update) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->cacheLastRow) < 0) return -1;
+  if (tEncodeI8(&encoder, pRsp->streamMode) < 0) return -1;
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tCoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSDbCfgRsp(void* buf, int32_t bufLen, SDbCfgRsp* pRsp) {
+  SCoder decoder = {0};
+  tCoderInit(&decoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_DECODER);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->numOfVgroups) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->cacheBlockSize) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->totalBlocks) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->daysPerFile) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->daysToKeep0) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->daysToKeep1) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->daysToKeep2) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->minRows) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->maxRows) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->commitTime) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->fsyncPeriod) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->walLevel) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->precision) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->compression) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->replications) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->quorum) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->update) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->cacheLastRow) < 0) return -1;
+  if (tDecodeI8(&decoder, &pRsp->streamMode) < 0) return -1;
+
+  tEndDecode(&decoder);
+
+  tCoderClear(&decoder);
+  return 0;
+}
+
+int32_t tSerializeSUserIndexReq(void* buf, int32_t bufLen, SUserIndexReq* pReq) {
+  SCoder encoder = {0};
+  tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeCStr(&encoder, pReq->indexFName) < 0) return -1;
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tCoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSUserIndexReq(void* buf, int32_t bufLen, SUserIndexReq* pReq) {
+  SCoder decoder = {0};
+  tCoderInit(&decoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_DECODER);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pReq->indexFName) < 0) return -1;
+  tEndDecode(&decoder);
+
+  tCoderClear(&decoder);
+  return 0;
+}
+
+int32_t tSerializeSUserIndexRsp(void* buf, int32_t bufLen, const SUserIndexRsp* pRsp) {
+  SCoder encoder = {0};
+  tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->dbFName) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->tblFName) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->colName) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->indexType) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->indexExts) < 0) return -1;
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tCoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSUserIndexRsp(void* buf, int32_t bufLen, SUserIndexRsp* pRsp) {
+  SCoder decoder = {0};
+  tCoderInit(&decoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_DECODER);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->dbFName) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->tblFName) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->colName) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->indexType) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->indexExts) < 0) return -1;
+  tEndDecode(&decoder);
+
+  tCoderClear(&decoder);
+  return 0;
+}
+
+
 int32_t tSerializeSShowReq(void *buf, int32_t bufLen, SShowReq *pReq) {
   SCoder encoder = {0};
   tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
@@ -2571,6 +2703,7 @@ int32_t tDeserializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *
 int32_t tFreeSCreateVnodeReq(SCreateVnodeReq *pReq) {
   taosArrayDestroy(pReq->pRetensions);
   pReq->pRetensions = NULL;
+  return 0;
 }
 
 int32_t tSerializeSDropVnodeReq(void *buf, int32_t bufLen, SDropVnodeReq *pReq) {
@@ -2787,6 +2920,48 @@ int32_t tDecodeSMqCMCommitOffsetReq(SCoder *decoder, SMqCMCommitOffsetReq *pReq)
   return 0;
 }
 
+int32_t tSerializeSExplainRsp(void* buf, int32_t bufLen, SExplainRsp* pRsp) {
+  SCoder encoder = {0};
+  tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->numOfPlans) < 0) return -1;
+  for (int32_t i = 0; i < pRsp->numOfPlans; ++i) {
+    SExplainExecInfo *info = &pRsp->subplanInfo[i];
+    if (tEncodeU64(&encoder, info->startupCost) < 0) return -1;
+    if (tEncodeU64(&encoder, info->totalCost) < 0) return -1;
+    if (tEncodeU64(&encoder, info->numOfRows) < 0) return -1;
+  }
+
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tCoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSExplainRsp(void* buf, int32_t bufLen, SExplainRsp* pRsp) {
+  SCoder decoder = {0};
+  tCoderInit(&decoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_DECODER);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->numOfPlans) < 0) return -1;
+  if (pRsp->numOfPlans > 0) {
+    pRsp->subplanInfo = taosMemoryMalloc(pRsp->numOfPlans * sizeof(SExplainExecInfo));
+    if (pRsp->subplanInfo == NULL) return -1;
+  }
+  for (int32_t i = 0; i < pRsp->numOfPlans; ++i) {
+    if (tDecodeU64(&decoder, &pRsp->subplanInfo[i].startupCost) < 0) return -1;
+    if (tDecodeU64(&decoder, &pRsp->subplanInfo[i].totalCost) < 0) return -1;
+    if (tDecodeU64(&decoder, &pRsp->subplanInfo[i].numOfRows) < 0) return -1;
+  }
+
+  tEndDecode(&decoder);
+
+  tCoderClear(&decoder);
+  return 0;
+}
+
 int32_t tSerializeSSchedulerHbReq(void *buf, int32_t bufLen, SSchedulerHbReq *pReq) {
   int32_t headLen = sizeof(SMsgHead);
   if (buf != NULL) {

source/common/src/tmsgcb.c

@@ -32,10 +32,6 @@ int32_t tmsgSendReq(const SMsgCb* pMsgCb, const SEpSet* epSet, SRpcMsg* pReq) {
   return (*pMsgCb->sendReqFp)(pMsgCb->pWrapper, epSet, pReq);
 }
 
-int32_t tmsgSendMnodeReq(const SMsgCb* pMsgCb, SRpcMsg* pReq) {
-  return (*pMsgCb->sendMnodeReqFp)(pMsgCb->pWrapper, pReq);
-}
-
 void tmsgSendRsp(const SRpcMsg* pRsp) { return (*tsDefaultMsgCb.sendRspFp)(tsDefaultMsgCb.pWrapper, pRsp); }
 
 void tmsgRegisterBrokenLinkArg(const SMsgCb* pMsgCb, SRpcMsg* pMsg) {

source/common/src/trow.c

@@ -503,7 +503,7 @@ SDataCols *tdDupDataCols(SDataCols *pDataCols, bool keepData) {
           memcpy(pRet->cols[i].dataOff, pDataCols->cols[i].dataOff, dataOffSize);
         }
         if (!TD_COL_ROWS_NORM(pRet->cols + i)) {
-          int32_t nBitmapBytes = (int32_t)TD_BITMAP_BYTES(pDataCols->maxPoints);
+          int32_t nBitmapBytes = (int32_t)TD_BITMAP_BYTES(pDataCols->numOfRows);
           memcpy(pRet->cols[i].pBitmap, pDataCols->cols[i].pBitmap, nBitmapBytes);
         }
       }

source/common/src/ttime.c

@@ -361,6 +361,18 @@ int32_t parseLocaltimeDst(char* timestr, int64_t* time, int32_t timePrec) {
   return 0;
 }
 
+char getPrecisionUnit(int32_t precision) {
+  static char units[3] = {TIME_UNIT_MILLISECOND, TIME_UNIT_MICROSECOND, TIME_UNIT_NANOSECOND};
+  switch (precision) {
+    case TSDB_TIME_PRECISION_MILLI:
+    case TSDB_TIME_PRECISION_MICRO:
+    case TSDB_TIME_PRECISION_NANO:
+      return units[precision];
+    default:
+      return 0;
+  }
+}
+
 int64_t convertTimePrecision(int64_t time, int32_t fromPrecision, int32_t toPrecision) {
   assert(fromPrecision == TSDB_TIME_PRECISION_MILLI || fromPrecision == TSDB_TIME_PRECISION_MICRO ||
          fromPrecision == TSDB_TIME_PRECISION_NANO);
@@ -370,6 +382,33 @@ int64_t convertTimePrecision(int64_t time, int32_t fromPrecision, int32_t toPrec
   return (int64_t)((double)time * factors[fromPrecision][toPrecision]);
 }
 
+int64_t convertTimeFromPrecisionToUnit(int64_t time, int32_t fromPrecision, char toUnit) {
+  assert(fromPrecision == TSDB_TIME_PRECISION_MILLI || fromPrecision == TSDB_TIME_PRECISION_MICRO ||
+         fromPrecision == TSDB_TIME_PRECISION_NANO);
+  static double factors[3] = {1000000., 1000., 1.};
+  switch (toUnit) {
+    case 's':
+      return time * factors[fromPrecision] / NANOSECOND_PER_SEC;
+    case 'm':
+      return time * factors[fromPrecision] / NANOSECOND_PER_MINUTE;
+    case 'h':
+      return time * factors[fromPrecision] / NANOSECOND_PER_HOUR;
+    case 'd':
+      return time * factors[fromPrecision] / NANOSECOND_PER_DAY;
+    case 'w':
+      return time * factors[fromPrecision] / NANOSECOND_PER_WEEK;
+    case 'a':
+      return time * factors[fromPrecision] / NANOSECOND_PER_MSEC;
+    case 'u':
+      return time * factors[fromPrecision] / NANOSECOND_PER_USEC;
+    case 'b':
+      return time * factors[fromPrecision];
+    default: {
+      return -1;
+    }
+  }  
+}
+
 static int32_t getDuration(int64_t val, char unit, int64_t* result, int32_t timePrecision) {
   switch (unit) {
     case 's':

source/common/src/ttypes.c

@@ -431,7 +431,7 @@ FORCE_INLINE void *getDataMax(int32_t type) {
   }
 }
 
-bool isValidDataType(int32_t type) { return type >= TSDB_DATA_TYPE_NULL && type <= TSDB_DATA_TYPE_UBIGINT; }
+bool isValidDataType(int32_t type) { return type >= TSDB_DATA_TYPE_NULL && type < TSDB_DATA_TYPE_MAX; }
 
 void setVardataNull(void *val, int32_t type) {
   if (type == TSDB_DATA_TYPE_BINARY) {

source/common/src/tvariant.c

@@ -512,6 +512,16 @@ static FORCE_INLINE int32_t convertToInteger(SVariant *pVariant, int64_t *result
     setNull((char *)result, type, tDataTypes[type].bytes);
     return 0;
   }
+
+  if (IS_SIGNED_NUMERIC_TYPE(pVariant->nType) || (pVariant->nType == TSDB_DATA_TYPE_BOOL)) {
+    *result = pVariant->i;
+  } else if (IS_UNSIGNED_NUMERIC_TYPE(pVariant->nType)) {
+    *result = pVariant->u;
+  } else if (IS_FLOAT_TYPE(pVariant->nType)) {
+    *result = (int64_t) pVariant->d;
+  } else {
+    //TODO: handling var types
+  }
 #if 0
   errno = 0;
   if (IS_SIGNED_NUMERIC_TYPE(pVariant->nType) || (pVariant->nType == TSDB_DATA_TYPE_BOOL)) {
@@ -1015,3 +1025,26 @@ int32_t taosVariantTypeSetType(SVariant *pVariant, char type) {
 
   return 0;
 }
+
+char * taosVariantGet(SVariant *pVar, int32_t type) {
+  switch (type) {
+    case TSDB_DATA_TYPE_BOOL: 
+    case TSDB_DATA_TYPE_TINYINT:
+    case TSDB_DATA_TYPE_SMALLINT:
+    case TSDB_DATA_TYPE_BIGINT:
+    case TSDB_DATA_TYPE_INT: 
+    case TSDB_DATA_TYPE_TIMESTAMP:
+      return (char *)&pVar->i;
+    case TSDB_DATA_TYPE_DOUBLE:
+    case TSDB_DATA_TYPE_FLOAT:
+      return (char *)&pVar->d;
+    case TSDB_DATA_TYPE_BINARY:
+      return (char *)pVar->pz;
+    case TSDB_DATA_TYPE_NCHAR:
+      return (char *)pVar->ucs4;
+    default: 
+      return NULL;
+  }
+
+  return NULL;
+}

source/dnode/mgmt/CMakeLists.txt

@@ -1,10 +1,10 @@
-aux_source_directory(dm/src DNODE_SRC)
+aux_source_directory(dm DNODE_SRC)
-aux_source_directory(qm/src DNODE_SRC)
+aux_source_directory(qm DNODE_SRC)
-aux_source_directory(bm/src DNODE_SRC)
+aux_source_directory(bm DNODE_SRC)
-aux_source_directory(sm/src DNODE_SRC)
+aux_source_directory(sm DNODE_SRC)
-aux_source_directory(vm/src DNODE_SRC)
+aux_source_directory(vm DNODE_SRC)
-aux_source_directory(mm/src DNODE_SRC)
+aux_source_directory(mm DNODE_SRC)
-aux_source_directory(main/src DNODE_SRC)
+aux_source_directory(main DNODE_SRC)
 add_library(dnode STATIC ${DNODE_SRC})
 target_link_libraries(
     dnode cjson mnode vnode qnode snode bnode wal sync taos tfs monitor
@@ -12,20 +12,14 @@ target_link_libraries(
 target_include_directories(
     dnode
     PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/mgmt"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/dm/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/qm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/bm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/sm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/vm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/mm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/main/inc"
 )
 
-aux_source_directory(main/exe EXEC_SRC)
+aux_source_directory(exe EXEC_SRC)
 add_executable(taosd ${EXEC_SRC})
 target_include_directories(
     taosd
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/main/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/inc"
 )
 target_link_libraries(taosd dnode)
 

source/dnode/mgmt/bm/bmHandle.c

@@ -16,6 +16,34 @@
 #define _DEFAULT_SOURCE
 #include "bmInt.h"
 
+void bmGetMonitorInfo(SMgmtWrapper *pWrapper, SMonBmInfo *bmInfo) {
+}
+
+int32_t bmProcessGetMonBmInfoReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq) {
+  SMonBmInfo bmInfo = {0};
+  bmGetMonitorInfo(pWrapper, &bmInfo);
+  dmGetMonitorSysInfo(&bmInfo.sys);
+  monGetLogs(&bmInfo.log);
+
+  int32_t rspLen = tSerializeSMonBmInfo(NULL, 0, &bmInfo);
+  if (rspLen < 0) {
+    terrno = TSDB_CODE_INVALID_MSG;
+    return -1;
+  }
+
+  void *pRsp = rpcMallocCont(rspLen);
+  if (pRsp == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  tSerializeSMonBmInfo(pRsp, rspLen, &bmInfo);
+  pReq->pRsp = pRsp;
+  pReq->rspLen = rspLen;
+  tFreeSMonBmInfo(&bmInfo);
+  return 0;
+}
+
 int32_t bmProcessCreateReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
   SDnode  *pDnode = pWrapper->pDnode;
   SRpcMsg *pReq = &pMsg->rpcMsg;
@@ -54,4 +82,6 @@ int32_t bmProcessDropReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
   }
 }
 
-void bmInitMsgHandles(SMgmtWrapper *pWrapper) {}
+void bmInitMsgHandle(SMgmtWrapper *pWrapper) {
+  dndSetMsgHandle(pWrapper, TDMT_MON_BM_INFO, bmProcessMonitorMsg, DEFAULT_HANDLE);
+}

source/dnode/mgmt/bm/bmInt.c

@@ -19,12 +19,7 @@
 static int32_t bmRequire(SMgmtWrapper *pWrapper, bool *required) { return dndReadFile(pWrapper, required); }
 
 static void bmInitOption(SBnodeMgmt *pMgmt, SBnodeOpt *pOption) {
-  SMsgCb msgCb = {0};
+  SMsgCb msgCb = dndCreateMsgcb(pMgmt->pWrapper);
-  msgCb.pWrapper = pMgmt->pWrapper;
-  msgCb.sendReqFp = dndSendReqToDnode;
-  msgCb.sendMnodeReqFp = dndSendReqToMnode;
-  msgCb.sendRspFp = dndSendRsp;
-  msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
   pOption->msgCb = msgCb;
 }
 
@@ -114,7 +109,7 @@ int32_t bmOpen(SMgmtWrapper *pWrapper) {
   return code;
 }
 
-void bmGetMgmtFp(SMgmtWrapper *pWrapper) {
+void bmSetMgmtFp(SMgmtWrapper *pWrapper) {
   SMgmtFp mgmtFp = {0};
   mgmtFp.openFp = bmOpen;
   mgmtFp.closeFp = bmClose;
@@ -122,7 +117,7 @@ void bmGetMgmtFp(SMgmtWrapper *pWrapper) {
   mgmtFp.dropMsgFp = bmProcessDropReq;
   mgmtFp.requiredFp = bmRequire;
 
-  bmInitMsgHandles(pWrapper);
+  bmInitMsgHandle(pWrapper);
   pWrapper->name = "bnode";
   pWrapper->fp = mgmtFp;
 }

source/dnode/mgmt/bm/bmWorker.c

@@ -18,7 +18,7 @@
 
 static void bmSendErrorRsp(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int32_t code) {
   SRpcMsg rpcRsp = {.handle = pMsg->rpcMsg.handle, .ahandle = pMsg->rpcMsg.ahandle, .code = code};
-  dndSendRsp(pWrapper, &rpcRsp);
+  tmsgSendRsp(&rpcRsp);
 
   dTrace("msg:%p, is freed", pMsg);
   rpcFreeCont(pMsg->rpcMsg.pCont);
@@ -33,7 +33,37 @@ static void bmSendErrorRsps(SMgmtWrapper *pWrapper, STaosQall *qall, int32_t num
   }
 }
 
-static void bmProcessQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
+static inline void bmSendRsp(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int32_t code) {
+  SRpcMsg rsp = {.handle = pMsg->rpcMsg.handle,
+                 .ahandle = pMsg->rpcMsg.ahandle,
+                 .code = code,
+                 .pCont = pMsg->pRsp,
+                 .contLen = pMsg->rspLen};
+  tmsgSendRsp(&rsp);
+}
+
+static void bmProcessMonQueue(SQueueInfo *pInfo, SNodeMsg *pMsg) {
+  SBnodeMgmt *pMgmt = pInfo->ahandle;
+
+  dTrace("msg:%p, get from bnode monitor queue", pMsg);
+  SRpcMsg *pRpc = &pMsg->rpcMsg;
+  int32_t  code = -1;
+
+  if (pMsg->rpcMsg.msgType == TDMT_MON_BM_INFO) {
+    code = bmProcessGetMonBmInfoReq(pMgmt->pWrapper, pMsg);
+  }
+
+  if (pRpc->msgType & 1U) {
+    if (code != 0 && terrno != 0) code = terrno;
+    bmSendRsp(pMgmt->pWrapper, pMsg, code);
+  }
+
+  dTrace("msg:%p, is freed, result:0x%04x:%s", pMsg, code & 0XFFFF, tstrerror(code));
+  rpcFreeCont(pRpc->pCont);
+  taosFreeQitem(pMsg);
+}
+
+static void bmProcessWriteQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
   SBnodeMgmt   *pMgmt = pInfo->ahandle;
   SMgmtWrapper *pWrapper = pMgmt->pWrapper;
 
@@ -72,18 +102,37 @@ int32_t bmProcessWriteMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
   return 0;
 }
 
+int32_t bmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
+  SBnodeMgmt    *pMgmt = pWrapper->pMgmt;
+  SSingleWorker *pWorker = &pMgmt->monitorWorker;
+
+  dTrace("msg:%p, put into worker:%s", pMsg, pWorker->name);
+  taosWriteQitem(pWorker->queue, pMsg);
+  return 0;
+}
+
 int32_t bmStartWorker(SBnodeMgmt *pMgmt) {
-  SMultiWorkerCfg cfg = {.max = 1, .name = "bnode-write", .fp = (FItems)bmProcessQueue, .param = pMgmt};
+  SMultiWorkerCfg cfg = {.max = 1, .name = "bnode-write", .fp = (FItems)bmProcessWriteQueue, .param = pMgmt};
   if (tMultiWorkerInit(&pMgmt->writeWorker, &cfg) != 0) {
-    dError("failed to start bnode write worker since %s", terrstr());
+    dError("failed to start bnode-write worker since %s", terrstr());
     return -1;
   }
 
+  if (tsMultiProcess) {
+    SSingleWorkerCfg mCfg = {
+        .min = 1, .max = 1, .name = "bnode-monitor", .fp = (FItem)bmProcessMonQueue, .param = pMgmt};
+    if (tSingleWorkerInit(&pMgmt->monitorWorker, &mCfg) != 0) {
+      dError("failed to start bnode-monitor worker since %s", terrstr());
+      return -1;
+    }
+   }
+
   dDebug("bnode workers are initialized");
   return 0;
 }
 
 void bmStopWorker(SBnodeMgmt *pMgmt) {
+  tSingleWorkerCleanup(&pMgmt->monitorWorker);
   tMultiWorkerCleanup(&pMgmt->writeWorker);
   dDebug("bnode workers are closed");
 }

source/dnode/mgmt/dm/dmFile.c

@@ -200,7 +200,7 @@ int32_t dmWriteFile(SDnodeMgmt *pMgmt) {
   taosMemoryFree(content);
 
   char realfile[PATH_MAX];
-  snprintf(realfile, sizeof(realfile), "%s%smnode.json", pMgmt->path, TD_DIRSEP);
+  snprintf(realfile, sizeof(realfile), "%s%sdnode.json", pMgmt->path, TD_DIRSEP);
 
   if (taosRenameFile(file, realfile) != 0) {
     terrno = TAOS_SYSTEM_ERROR(errno);

source/dnode/mgmt/dm/dmHandle.c

@@ -15,7 +15,6 @@
 
 #define _DEFAULT_SOURCE
 #include "dmInt.h"
-#include "vm.h"
 
 void dmSendStatusReq(SDnodeMgmt *pMgmt) {
   SDnode    *pDnode = pMgmt->pDnode;
@@ -43,8 +42,9 @@ void dmSendStatusReq(SDnodeMgmt *pMgmt) {
 
   SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, VNODES);
   if (pWrapper != NULL) {
-    req.pVloads = taosArrayInit(TSDB_MAX_VNODES, sizeof(SVnodeLoad));
+    SMonVloadInfo info = {0};
-    vmMonitorVnodeLoads(pWrapper, req.pVloads);
+    dmGetVnodeLoads(pWrapper, &info);
+    req.pVloads = info.pVloads;
     dndReleaseWrapper(pWrapper);
   }
 
@@ -53,11 +53,13 @@ void dmSendStatusReq(SDnodeMgmt *pMgmt) {
   tSerializeSStatusReq(pHead, contLen, &req);
   taosArrayDestroy(req.pVloads);
 
-  SRpcMsg rpcMsg = {.pCont = pHead, .contLen = contLen, .msgType = TDMT_MND_STATUS, .ahandle = (void *)9527};
+  SRpcMsg rpcMsg = {.pCont = pHead, .contLen = contLen, .msgType = TDMT_MND_STATUS, .ahandle = (void *)0x9527};
   pMgmt->statusSent = 1;
 
   dTrace("send req:%s to mnode, app:%p", TMSG_INFO(rpcMsg.msgType), rpcMsg.ahandle);
-  dndSendReqToMnode(pMgmt->pWrapper, &rpcMsg);
+  SEpSet epSet = {0};
+  dmGetMnodeEpSet(pMgmt, &epSet);
+  tmsgSendReq(&pMgmt->msgCb, &epSet, &rpcMsg);
 }
 
 static void dmUpdateDnodeCfg(SDnodeMgmt *pMgmt, SDnodeCfg *pCfg) {
@@ -95,6 +97,7 @@ int32_t dmProcessStatusRsp(SDnodeMgmt *pMgmt, SNodeMsg *pMsg) {
   }
 
   pMgmt->statusSent = 0;
+  return TSDB_CODE_SUCCESS;
 }
 
 int32_t dmProcessAuthRsp(SDnodeMgmt *pMgmt, SNodeMsg *pMsg) {
@@ -116,21 +119,98 @@ int32_t dmProcessConfigReq(SDnodeMgmt *pMgmt, SNodeMsg *pMsg) {
   return TSDB_CODE_OPS_NOT_SUPPORT;
 }
 
-void dmInitMsgHandles(SMgmtWrapper *pWrapper) {
+static int32_t dmProcessCreateNodeMsg(SDnode *pDnode, EDndType ntype, SNodeMsg *pMsg) {
+  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, ntype);
+  if (pWrapper != NULL) {
+    dndReleaseWrapper(pWrapper);
+    terrno = TSDB_CODE_NODE_ALREADY_DEPLOYED;
+    dError("failed to create node since %s", terrstr());
+    return -1;
+  }
+
+  pWrapper = &pDnode->wrappers[ntype];
+
+  if (taosMkDir(pWrapper->path) != 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to create dir:%s since %s", pWrapper->path, terrstr());
+    return -1;
+  }
+
+  int32_t code = (*pWrapper->fp.createMsgFp)(pWrapper, pMsg);
+  if (code != 0) {
+    dError("node:%s, failed to open since %s", pWrapper->name, terrstr());
+  } else {
+    dDebug("node:%s, has been opened", pWrapper->name);
+    pWrapper->deployed = true;
+  }
+
+  return code;
+}
+
+static int32_t dmProcessDropNodeMsg(SDnode *pDnode, EDndType ntype, SNodeMsg *pMsg) {
+  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, ntype);
+  if (pWrapper == NULL) {
+    terrno = TSDB_CODE_NODE_NOT_DEPLOYED;
+    dError("failed to drop node since %s", terrstr());
+    return -1;
+  }
+
+  taosWLockLatch(&pWrapper->latch);
+  pWrapper->deployed = false;
+
+  int32_t code = (*pWrapper->fp.dropMsgFp)(pWrapper, pMsg);
+  if (code != 0) {
+    pWrapper->deployed = true;
+    dError("node:%s, failed to drop since %s", pWrapper->name, terrstr());
+  } else {
+    pWrapper->deployed = false;
+    dDebug("node:%s, has been dropped", pWrapper->name);
+  }
+
+  taosWUnLockLatch(&pWrapper->latch);
+  dndReleaseWrapper(pWrapper);
+  return code;
+}
+
+int32_t dmProcessCDnodeReq(SDnode *pDnode, SNodeMsg *pMsg) {
+  switch (pMsg->rpcMsg.msgType) {
+    case TDMT_DND_CREATE_MNODE:
+      return dmProcessCreateNodeMsg(pDnode, MNODE, pMsg);
+    case TDMT_DND_DROP_MNODE:
+      return dmProcessDropNodeMsg(pDnode, MNODE, pMsg);
+    case TDMT_DND_CREATE_QNODE:
+      return dmProcessCreateNodeMsg(pDnode, QNODE, pMsg);
+    case TDMT_DND_DROP_QNODE:
+      return dmProcessDropNodeMsg(pDnode, QNODE, pMsg);
+    case TDMT_DND_CREATE_SNODE:
+      return dmProcessCreateNodeMsg(pDnode, SNODE, pMsg);
+    case TDMT_DND_DROP_SNODE:
+      return dmProcessDropNodeMsg(pDnode, SNODE, pMsg);
+    case TDMT_DND_CREATE_BNODE:
+      return dmProcessCreateNodeMsg(pDnode, BNODE, pMsg);
+    case TDMT_DND_DROP_BNODE:
+      return dmProcessDropNodeMsg(pDnode, BNODE, pMsg);
+    default:
+      terrno = TSDB_CODE_MSG_NOT_PROCESSED;
+      return -1;
+  }
+}
+
+void dmInitMsgHandle(SMgmtWrapper *pWrapper) {
   // Requests handled by DNODE
-  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_MNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_MNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_MNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_MNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_QNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_QNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_QNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_QNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_SNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_SNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_SNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_SNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_BNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_BNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_BNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_DROP_BNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_CONFIG_DNODE, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_CONFIG_DNODE, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_DND_NETWORK_TEST, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_DND_NETWORK_TEST, dmProcessMgmtMsg, DEFAULT_HANDLE);
 
   // Requests handled by MNODE
-  dndSetMsgHandle(pWrapper, TDMT_MND_STATUS_RSP, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_MND_STATUS_RSP, dmProcessMonitorMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_MND_GRANT_RSP, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_MND_GRANT_RSP, dmProcessMgmtMsg, DEFAULT_HANDLE);
-  dndSetMsgHandle(pWrapper, TDMT_MND_AUTH_RSP, dmProcessMgmtMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_MND_AUTH_RSP, dmProcessMgmtMsg, DEFAULT_HANDLE);
 }

source/dnode/mgmt/dm/dmInt.c

@@ -78,7 +78,7 @@ static int32_t dmStart(SMgmtWrapper *pWrapper) {
   return dmStartThread(pWrapper->pMgmt);
 }
 
-int32_t dmInit(SMgmtWrapper *pWrapper) {
+static int32_t dmInit(SMgmtWrapper *pWrapper) {
   SDnode     *pDnode = pWrapper->pDnode;
   SDnodeMgmt *pMgmt = taosMemoryCalloc(1, sizeof(SDnodeMgmt));
   dInfo("dnode-mgmt start to init");
@@ -112,22 +112,19 @@ int32_t dmInit(SMgmtWrapper *pWrapper) {
     return -1;
   }
 
-  if (dndInitServer(pDnode) != 0) {
+  if (dndInitTrans(pDnode) != 0) {
-    dError("failed to init trans server since %s", terrstr());
+    dError("failed to init transport since %s", terrstr());
-    return -1;
-  }
-
-  if (dndInitClient(pDnode) != 0) {
-    dError("failed to init trans client since %s", terrstr());
     return -1;
   }
 
   pWrapper->pMgmt = pMgmt;
+  pMgmt->msgCb = dndCreateMsgcb(pWrapper);
+
   dInfo("dnode-mgmt is initialized");
   return 0;
 }
 
-void dmCleanup(SMgmtWrapper *pWrapper) {
+static void dmCleanup(SMgmtWrapper *pWrapper) {
   SDnodeMgmt *pMgmt = pWrapper->pMgmt;
   if (pMgmt == NULL) return;
 
@@ -151,25 +148,24 @@ void dmCleanup(SMgmtWrapper *pWrapper) {
 
   taosMemoryFree(pMgmt);
   pWrapper->pMgmt = NULL;
-  dndCleanupServer(pDnode);
+  dndCleanupTrans(pDnode);
-  dndCleanupClient(pDnode);
 
   dInfo("dnode-mgmt is cleaned up");
 }
 
-int32_t dmRequire(SMgmtWrapper *pWrapper, bool *required) {
+static int32_t dmRequire(SMgmtWrapper *pWrapper, bool *required) {
   *required = true;
   return 0;
 }
 
-void dmGetMgmtFp(SMgmtWrapper *pWrapper) {
+void dmSetMgmtFp(SMgmtWrapper *pWrapper) {
   SMgmtFp mgmtFp = {0};
   mgmtFp.openFp = dmInit;
   mgmtFp.closeFp = dmCleanup;
   mgmtFp.startFp = dmStart;
   mgmtFp.requiredFp = dmRequire;
 
-  dmInitMsgHandles(pWrapper);
+  dmInitMsgHandle(pWrapper);
   pWrapper->name = "dnode";
   pWrapper->fp = mgmtFp;
 }

source/dnode/mgmt/dm/dmMonitor.c

@@ -0,0 +1,195 @@
+/*
+ * 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 "dmInt.h"
+
+static void dmGetMonitorBasicInfo(SDnode *pDnode, SMonBasicInfo *pInfo) {
+  pInfo->protocol = 1;
+  pInfo->dnode_id = pDnode->dnodeId;
+  pInfo->cluster_id = pDnode->clusterId;
+  tstrncpy(pInfo->dnode_ep, tsLocalEp, TSDB_EP_LEN);
+}
+
+static void dmGetMonitorDnodeInfo(SDnode *pDnode, SMonDnodeInfo *pInfo) {
+  pInfo->uptime = (taosGetTimestampMs() - pDnode->rebootTime) / (86400000.0f);
+  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, MNODE);
+  if (pWrapper != NULL) {
+    pInfo->has_mnode = pWrapper->required;
+    dndReleaseWrapper(pWrapper);
+  }
+  tstrncpy(pInfo->logdir.name, tsLogDir, sizeof(pInfo->logdir.name));
+  pInfo->logdir.size = tsLogSpace.size;
+  tstrncpy(pInfo->tempdir.name, tsTempDir, sizeof(pInfo->tempdir.name));
+  pInfo->tempdir.size = tsTempSpace.size;
+}
+
+static void dmGetMonitorInfo(SDnode *pDnode, SMonDmInfo *pInfo) {
+  dmGetMonitorBasicInfo(pDnode, &pInfo->basic);
+  dmGetMonitorSysInfo(&pInfo->sys);
+  dmGetMonitorDnodeInfo(pDnode, &pInfo->dnode);
+}
+
+void dmSendMonitorReport(SDnode *pDnode) {
+  if (!tsEnableMonitor || tsMonitorFqdn[0] == 0 || tsMonitorPort == 0) return;
+  dTrace("send monitor report to %s:%u", tsMonitorFqdn, tsMonitorPort);
+
+  SMonDmInfo dmInfo = {0};
+  SMonMmInfo mmInfo = {0};
+  SMonVmInfo vmInfo = {0};
+  SMonQmInfo qmInfo = {0};
+  SMonSmInfo smInfo = {0};
+  SMonBmInfo bmInfo = {0};
+
+  SRpcMsg req = {0};
+  SRpcMsg rsp;
+  SEpSet  epset = {.inUse = 0, .numOfEps = 1};
+  tstrncpy(epset.eps[0].fqdn, tsLocalFqdn, TSDB_FQDN_LEN);
+  epset.eps[0].port = tsServerPort;
+
+  SMgmtWrapper *pWrapper = NULL;
+  dmGetMonitorInfo(pDnode, &dmInfo);
+
+  bool getFromAPI = !tsMultiProcess;
+  pWrapper = &pDnode->wrappers[MNODE];
+  if (getFromAPI) {
+    if (dndMarkWrapper(pWrapper) != 0) {
+      mmGetMonitorInfo(pWrapper, &mmInfo);
+      dndReleaseWrapper(pWrapper);
+    }
+  } else {
+    if (pWrapper->required) {
+      req.msgType = TDMT_MON_MM_INFO;
+      dndSendRecv(pDnode, &epset, &req, &rsp);
+      if (rsp.code == 0 && rsp.contLen > 0) {
+        tDeserializeSMonMmInfo(rsp.pCont, rsp.contLen, &mmInfo);
+      }
+      rpcFreeCont(rsp.pCont);
+    }
+  }
+
+  pWrapper = &pDnode->wrappers[VNODES];
+  if (getFromAPI) {
+    if (dndMarkWrapper(pWrapper) != 0) {
+      vmGetMonitorInfo(pWrapper, &vmInfo);
+      dndReleaseWrapper(pWrapper);
+    }
+  } else {
+    if (pWrapper->required) {
+      req.msgType = TDMT_MON_VM_INFO;
+      dndSendRecv(pDnode, &epset, &req, &rsp);
+      if (rsp.code == 0 && rsp.contLen > 0) {
+        tDeserializeSMonVmInfo(rsp.pCont, rsp.contLen, &vmInfo);
+      }
+      rpcFreeCont(rsp.pCont);
+    }
+  }
+
+  pWrapper = &pDnode->wrappers[QNODE];
+  if (getFromAPI) {
+    if (dndMarkWrapper(pWrapper) != 0) {
+      qmGetMonitorInfo(pWrapper, &qmInfo);
+      dndReleaseWrapper(pWrapper);
+    }
+  } else {
+    if (pWrapper->required) {
+      req.msgType = TDMT_MON_QM_INFO;
+      dndSendRecv(pDnode, &epset, &req, &rsp);
+      if (rsp.code == 0 && rsp.contLen > 0) {
+        tDeserializeSMonQmInfo(rsp.pCont, rsp.contLen, &qmInfo);
+      }
+      rpcFreeCont(rsp.pCont);
+    }
+  }
+
+  pWrapper = &pDnode->wrappers[SNODE];
+  if (getFromAPI) {
+    if (dndMarkWrapper(pWrapper) != 0) {
+      smGetMonitorInfo(pWrapper, &smInfo);
+      dndReleaseWrapper(pWrapper);
+    }
+  } else {
+    if (pWrapper->required) {
+      req.msgType = TDMT_MON_SM_INFO;
+      dndSendRecv(pDnode, &epset, &req, &rsp);
+      if (rsp.code == 0 && rsp.contLen > 0) {
+        tDeserializeSMonSmInfo(rsp.pCont, rsp.contLen, &smInfo);
+      }
+      rpcFreeCont(rsp.pCont);
+    }
+  }
+
+  pWrapper = &pDnode->wrappers[BNODE];
+  if (getFromAPI) {
+    if (dndMarkWrapper(pWrapper) != 0) {
+      bmGetMonitorInfo(pWrapper, &bmInfo);
+      dndReleaseWrapper(pWrapper);
+    }
+  } else {
+    if (pWrapper->required) {
+      req.msgType = TDMT_MON_BM_INFO;
+      dndSendRecv(pDnode, &epset, &req, &rsp);
+      if (rsp.code == 0 && rsp.contLen > 0) {
+        tDeserializeSMonBmInfo(rsp.pCont, rsp.contLen, &bmInfo);
+      }
+      rpcFreeCont(rsp.pCont);
+    }
+  }
+
+  monSetDmInfo(&dmInfo);
+  monSetMmInfo(&mmInfo);
+  monSetVmInfo(&vmInfo);
+  monSetQmInfo(&qmInfo);
+  monSetSmInfo(&smInfo);
+  monSetBmInfo(&bmInfo);
+  tFreeSMonMmInfo(&mmInfo);
+  tFreeSMonVmInfo(&vmInfo);
+  tFreeSMonQmInfo(&qmInfo);
+  tFreeSMonSmInfo(&smInfo);
+  tFreeSMonBmInfo(&bmInfo);
+  monSendReport();
+}
+
+void dmGetVnodeLoads(SMgmtWrapper *pWrapper, SMonVloadInfo *pInfo) {
+  bool getFromAPI = !tsMultiProcess;
+  if (getFromAPI) {
+    vmGetVnodeLoads(pWrapper, pInfo);
+  } else {
+    SRpcMsg req = {.msgType = TDMT_MON_VM_LOAD};
+    SRpcMsg rsp = {0};
+    SEpSet  epset = {.inUse = 0, .numOfEps = 1};
+    tstrncpy(epset.eps[0].fqdn, tsLocalFqdn, TSDB_FQDN_LEN);
+    epset.eps[0].port = tsServerPort;
+
+    dndSendRecv(pWrapper->pDnode, &epset, &req, &rsp);
+    if (rsp.code == 0 && rsp.contLen > 0) {
+      tDeserializeSMonVloadInfo(rsp.pCont, rsp.contLen, pInfo);
+    }
+    rpcFreeCont(rsp.pCont);
+  }
+}
+
+void dmGetMonitorSysInfo(SMonSysInfo *pInfo) {
+  taosGetCpuUsage(&pInfo->cpu_engine, &pInfo->cpu_system);
+  taosGetCpuCores(&pInfo->cpu_cores);
+  taosGetProcMemory(&pInfo->mem_engine);
+  taosGetSysMemory(&pInfo->mem_system);
+  pInfo->mem_total = tsTotalMemoryKB;
+  pInfo->disk_engine = 0;
+  pInfo->disk_used = tsDataSpace.size.used;
+  pInfo->disk_total = tsDataSpace.size.total;
+  taosGetCardInfoDelta(&pInfo->net_in, &pInfo->net_out);
+  taosGetProcIODelta(&pInfo->io_read, &pInfo->io_write, &pInfo->io_read_disk, &pInfo->io_write_disk);
+}

source/dnode/mgmt/dm/dmWorker.c

@@ -14,12 +14,7 @@
  */
 
 #define _DEFAULT_SOURCE
-#include "bm.h"
 #include "dmInt.h"
-#include "mm.h"
-#include "qm.h"
-#include "sm.h"
-#include "vm.h"
 
 static void *dmThreadRoutine(void *param) {
   SDnodeMgmt *pMgmt = param;
@@ -37,7 +32,6 @@ static void *dmThreadRoutine(void *param) {
     }
 
     int64_t curTime = taosGetTimestampMs();
-
     float   statusInterval = (curTime - lastStatusTime) / 1000.0f;
     if (statusInterval >= tsStatusInterval && !pMgmt->statusSent) {
       dmSendStatusReq(pMgmt);
@@ -46,10 +40,22 @@ static void *dmThreadRoutine(void *param) {
 
     float monitorInterval = (curTime - lastMonitorTime) / 1000.0f;
     if (monitorInterval >= tsMonitorInterval) {
-      dndSendMonitorReport(pDnode);
+      dmSendMonitorReport(pDnode);
       lastMonitorTime = curTime;
     }
   }
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t dmStartThread(SDnodeMgmt *pMgmt) {
+  pMgmt->threadId = taosCreateThread(dmThreadRoutine, pMgmt);
+  if (pMgmt->threadId == NULL) {
+    dError("failed to init dnode thread");
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  return 0;
 }
 
 static void dmProcessQueue(SQueueInfo *pInfo, SNodeMsg *pMsg) {
@@ -61,16 +67,6 @@ static void dmProcessQueue(SQueueInfo *pInfo, SNodeMsg *pMsg) {
   dTrace("msg:%p, will be processed in dnode queue", pMsg);
 
   switch (pRpc->msgType) {
-    case TDMT_DND_CREATE_MNODE:
-    case TDMT_DND_CREATE_QNODE:
-    case TDMT_DND_CREATE_SNODE:
-    case TDMT_DND_CREATE_BNODE:
-    case TDMT_DND_DROP_MNODE:
-    case TDMT_DND_DROP_QNODE:
-    case TDMT_DND_DROP_SNODE:
-    case TDMT_DND_DROP_BNODE:
-      code = dndProcessNodeMsg(pMgmt->pDnode, pMsg);
-      break;
     case TDMT_DND_CONFIG_DNODE:
       code = dmProcessConfigReq(pMgmt, pMsg);
       break;
@@ -84,8 +80,8 @@ static void dmProcessQueue(SQueueInfo *pInfo, SNodeMsg *pMsg) {
       code = dmProcessGrantRsp(pMgmt, pMsg);
       break;
     default:
-      terrno = TSDB_CODE_MSG_NOT_PROCESSED;
+      code = dmProcessCDnodeReq(pMgmt->pDnode, pMsg);
-      dError("msg:%p, type:%s not processed in dnode queue", pRpc->handle, TMSG_INFO(pRpc->msgType));
+      break;
   }
 
   if (pRpc->msgType & 1u) {
@@ -100,17 +96,15 @@ static void dmProcessQueue(SQueueInfo *pInfo, SNodeMsg *pMsg) {
 }
 
 int32_t dmStartWorker(SDnodeMgmt *pMgmt) {
-  SSingleWorkerCfg mgmtCfg = {
+  SSingleWorkerCfg mcfg = {.min = 1, .max = 1, .name = "dnode-mgmt", .fp = (FItem)dmProcessQueue, .param = pMgmt};
-      .min = 1, .max = 1, .name = "dnode-mgmt", .fp = (FItem)dmProcessQueue, .param = pMgmt};
+  if (tSingleWorkerInit(&pMgmt->mgmtWorker, &mcfg) != 0) {
-  if (tSingleWorkerInit(&pMgmt->mgmtWorker, &mgmtCfg) != 0) {
     dError("failed to start dnode mgmt worker since %s", terrstr());
     return -1;
   }
 
-  SSingleWorkerCfg statusCfg = {
+  SSingleWorkerCfg scfg = {.min = 1, .max = 1, .name = "dnode-monitor", .fp = (FItem)dmProcessQueue, .param = pMgmt};
-      .min = 1, .max = 1, .name = "dnode-status", .fp = (FItem)dmProcessQueue, .param = pMgmt};
+  if (tSingleWorkerInit(&pMgmt->monitorWorker, &scfg) != 0) {
-  if (tSingleWorkerInit(&pMgmt->statusWorker, &statusCfg) != 0) {
+    dError("failed to start dnode monitor worker since %s", terrstr());
-    dError("failed to start dnode status worker since %s", terrstr());
     return -1;
   }
 
@@ -118,20 +112,9 @@ int32_t dmStartWorker(SDnodeMgmt *pMgmt) {
   return 0;
 }
 
-int32_t dmStartThread(SDnodeMgmt *pMgmt) {
-  pMgmt->threadId = taosCreateThread(dmThreadRoutine, pMgmt);
-  if (pMgmt->threadId == NULL) {
-    dError("failed to init dnode thread");
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return -1;
-  }
-
-  return 0;
-}
-
 void dmStopWorker(SDnodeMgmt *pMgmt) {
   tSingleWorkerCleanup(&pMgmt->mgmtWorker);
-  tSingleWorkerCleanup(&pMgmt->statusWorker);
+  tSingleWorkerCleanup(&pMgmt->monitorWorker);
 
   if (pMgmt->threadId != NULL) {
     taosDestoryThread(pMgmt->threadId);
@@ -143,9 +126,15 @@ void dmStopWorker(SDnodeMgmt *pMgmt) {
 int32_t dmProcessMgmtMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
   SDnodeMgmt    *pMgmt = pWrapper->pMgmt;
   SSingleWorker *pWorker = &pMgmt->mgmtWorker;
-  if (pMsg->rpcMsg.msgType == TDMT_MND_STATUS_RSP) {
+
-    pWorker = &pMgmt->statusWorker;
+  dTrace("msg:%p, put into worker %s", pMsg, pWorker->name);
-  }
+  taosWriteQitem(pWorker->queue, pMsg);
+  return 0;
+}
+
+int32_t dmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
+  SDnodeMgmt    *pMgmt = pWrapper->pMgmt;
+  SSingleWorker *pWorker = &pMgmt->monitorWorker;
 
   dTrace("msg:%p, put into worker %s", pMsg, pWorker->name);
   taosWriteQitem(pWorker->queue, pMsg);

source/dnode/mgmt/dm/inc/dm.h

@@ -1,38 +0,0 @@
-/*
- * 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_DND_DNODE_H_
-#define _TD_DND_DNODE_H_
-
-#include "dnd.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-typedef struct SDnodeMgmt SDnodeMgmt;
-
-void dmGetMgmtFp(SMgmtWrapper *pWrapper);
-void dmInitMsgHandles(SMgmtWrapper *pWrapper);
-
-void    dmGetMnodeEpSet(SDnodeMgmt *pMgmt, SEpSet *pEpSet);
-void    dmUpdateMnodeEpSet(SDnodeMgmt *pMgmt, SEpSet *pEpSet);
-void    dmSendRedirectRsp(SDnodeMgmt *pMgmt, const SRpcMsg *pMsg);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_DND_DNODE_H_*/

source/dnode/mgmt/exe/dndMain.c

@@ -14,7 +14,7 @@
  */
 
 #define _DEFAULT_SOURCE
-#include "dnd.h"
+#include "dndInt.h"
 #include "tconfig.h"
 
 static struct {
@@ -26,37 +26,30 @@ static struct {
   char     apolloUrl[PATH_MAX];
   SArray  *pArgs;  // SConfigPair
   SDnode  *pDnode;
-  ENodeType ntype;
+  EDndType ntype;
 } global = {0};
 
 static void dndStopDnode(int signum, void *info, void *ctx) {
-  dInfo("signal:%d is received", signum);
   SDnode *pDnode = atomic_val_compare_exchange_ptr(&global.pDnode, 0, global.pDnode);
   if (pDnode != NULL) {
     dndHandleEvent(pDnode, DND_EVENT_STOP);
   }
 }
 
-static void dndHandleChild(int signum, void *info, void *ctx) {
-  dInfo("signal:%d is received", signum);
-  dndHandleEvent(global.pDnode, DND_EVENT_CHILD);
-}
-
 static void dndSetSignalHandle() {
   taosSetSignal(SIGTERM, dndStopDnode);
   taosSetSignal(SIGHUP, dndStopDnode);
   taosSetSignal(SIGINT, dndStopDnode);
+  taosSetSignal(SIGTSTP, dndStopDnode);
   taosSetSignal(SIGABRT, dndStopDnode);
   taosSetSignal(SIGBREAK, dndStopDnode);
+  taosSetSignal(SIGQUIT, dndStopDnode);
 
   if (!tsMultiProcess) {
-    // Set the single process signal
+  } else if (global.ntype == DNODE || global.ntype == NODE_MAX) {
-  } else if (global.ntype == DNODE) {
+    taosIgnSignal(SIGCHLD);
-    // When the child process exits, the parent process receives a signal
-    taosSetSignal(SIGCHLD, dndHandleChild);
   } else {
-    // When the parent process exits, the child process will receive the SIGKILL signal
+    taosKillChildOnParentStopped();
-    taosKillChildOnSelfStopped();
   }
 }
 
@@ -77,14 +70,14 @@ static int32_t dndParseArgs(int32_t argc, char const *argv[]) {
       tstrncpy(global.apolloUrl, argv[++i], PATH_MAX);
     } else if (strcmp(argv[i], "-e") == 0) {
       tstrncpy(global.envFile, argv[++i], PATH_MAX);
-    } else if (strcmp(argv[i], "-k") == 0) {
-      global.generateGrant = true;
     } else if (strcmp(argv[i], "-n") == 0) {
       global.ntype = atoi(argv[++i]);
       if (global.ntype <= DNODE || global.ntype > NODE_MAX) {
-        printf("'-n' range is [1-5], default is 0\n");
+        printf("'-n' range is [1 - %d], default is 0\n", NODE_MAX - 1);
         return -1;
       }
+    } else if (strcmp(argv[i], "-k") == 0) {
+      global.generateGrant = true;
     } else if (strcmp(argv[i], "-C") == 0) {
       global.dumpConfig = true;
     } else if (strcmp(argv[i], "-V") == 0) {
@@ -140,23 +133,24 @@ static int32_t dndInitLog() {
   return taosCreateLog(logName, 1, configDir, global.envFile, global.apolloUrl, global.pArgs, 0);
 }
 
-static void dndSetProcName(char **argv) {
+static void dndSetProcInfo(int32_t argc, char **argv) {
-  if (global.ntype != DNODE) {
+  taosSetProcPath(argc, argv);
+  if (global.ntype != DNODE && global.ntype != NODE_MAX) {
     const char *name = dndNodeProcStr(global.ntype);
-    taosSetProcName(argv, name);
+    taosSetProcName(argc, argv, name);
   }
 }
 
 static int32_t dndRunDnode() {
   if (dndInit() != 0) {
-    dError("failed to initialize environment since %s", terrstr());
+    dError("failed to init environment since %s", terrstr());
     return -1;
   }
 
   SDnodeOpt option = dndGetOpt();
   SDnode   *pDnode = dndCreate(&option);
   if (pDnode == NULL) {
-    dError("failed to to create dnode object since %s", terrstr());
+    dError("failed to to create dnode since %s", terrstr());
     return -1;
   } else {
     global.pDnode = pDnode;
@@ -213,6 +207,6 @@ int main(int argc, char const *argv[]) {
     return 0;
   }
 
-  dndSetProcName((char **)argv);
+  dndSetProcInfo(argc, (char **)argv);
   return dndRunDnode();
 }

source/dnode/mgmt/inc/bmInt.h

@@ -16,7 +16,8 @@
 #ifndef _TD_DND_BNODE_INT_H_
 #define _TD_DND_BNODE_INT_H_
 
-#include "bm.h"
+#include "dndInt.h"
+
 #include "bnode.h"
 
 #ifdef __cplusplus
@@ -29,21 +30,24 @@ typedef struct SBnodeMgmt {
   SMgmtWrapper *pWrapper;
   const char   *path;
   SMultiWorker  writeWorker;
+  SSingleWorker monitorWorker;
 } SBnodeMgmt;
 
 // bmInt.c
 int32_t bmOpen(SMgmtWrapper *pWrapper);
 int32_t bmDrop(SMgmtWrapper *pWrapper);
 
-// bmMsg.c
+// bmHandle.c
-void    bmInitMsgHandles(SMgmtWrapper *pWrapper);
+void    bmInitMsgHandle(SMgmtWrapper *pWrapper);
 int32_t bmProcessCreateReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t bmProcessDropReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t bmProcessGetMonBmInfoReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq);
 
 // bmWorker.c
 int32_t bmStartWorker(SBnodeMgmt *pMgmt);
 void    bmStopWorker(SBnodeMgmt *pMgmt);
 int32_t bmProcessWriteMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t bmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/inc/dmInt.h

@@ -16,7 +16,7 @@
 #ifndef _TD_DND_DNODE_INT_H_
 #define _TD_DND_DNODE_INT_H_
 
-#include "dm.h"
+#include "dndInt.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -32,7 +32,8 @@ typedef struct SDnodeMgmt {
   TdThread     *threadId;
   SRWLatch      latch;
   SSingleWorker mgmtWorker;
-  SSingleWorker statusWorker;
+  SSingleWorker monitorWorker;
+  SMsgCb        msgCb;
   const char   *path;
   SDnode       *pDnode;
   SMgmtWrapper *pWrapper;
@@ -43,18 +44,25 @@ int32_t dmReadFile(SDnodeMgmt *pMgmt);
 int32_t dmWriteFile(SDnodeMgmt *pMgmt);
 void    dmUpdateDnodeEps(SDnodeMgmt *pMgmt, SArray *pDnodeEps);
 
-// dmMsg.c
+// dmHandle.c
+void    dmInitMsgHandle(SMgmtWrapper *pWrapper);
 void    dmSendStatusReq(SDnodeMgmt *pMgmt);
 int32_t dmProcessConfigReq(SDnodeMgmt *pMgmt, SNodeMsg *pMsg);
 int32_t dmProcessStatusRsp(SDnodeMgmt *pMgmt, SNodeMsg *pMsg);
 int32_t dmProcessAuthRsp(SDnodeMgmt *pMgmt, SNodeMsg *pMsg);
 int32_t dmProcessGrantRsp(SDnodeMgmt *pMgmt, SNodeMsg *pMsg);
+int32_t dmProcessCDnodeReq(SDnode *pDnode, SNodeMsg *pMsg);
+
+// dmMonitor.c
+void dmGetVnodeLoads(SMgmtWrapper *pWrapper, SMonVloadInfo *pInfo);
+void dmSendMonitorReport(SDnode *pDnode);
 
 // dmWorker.c
+int32_t dmStartThread(SDnodeMgmt *pMgmt);
 int32_t dmStartWorker(SDnodeMgmt *pMgmt);
 void    dmStopWorker(SDnodeMgmt *pMgmt);
-int32_t dmStartThread(SDnodeMgmt *pMgmt);
 int32_t dmProcessMgmtMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t dmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/inc/dndInt.h

@@ -13,8 +13,8 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_DND_H_
+#ifndef _TD_DND_INT_H_
-#define _TD_DND_H_
+#define _TD_DND_INT_H_
 
 #include "os.h"
 
@@ -27,13 +27,13 @@
 #include "tlockfree.h"
 #include "tlog.h"
 #include "tmsg.h"
+#include "tmsgcb.h"
 #include "tprocess.h"
 #include "tqueue.h"
 #include "trpc.h"
 #include "tthread.h"
 #include "ttime.h"
 #include "tworker.h"
-#include "tmsgcb.h"
 
 #include "dnode.h"
 #include "monitor.h"
@@ -49,7 +49,7 @@ extern "C" {
 #define dDebug(...) { if (dDebugFlag & DEBUG_DEBUG) { taosPrintLog("DND ", DEBUG_DEBUG, dDebugFlag, __VA_ARGS__); }}
 #define dTrace(...) { if (dDebugFlag & DEBUG_TRACE) { taosPrintLog("DND ", DEBUG_TRACE, dDebugFlag, __VA_ARGS__); }}
 
-typedef enum { DNODE, VNODES, QNODE, SNODE, MNODE, BNODE, NODE_MAX } ENodeType;
+typedef enum { DNODE, VNODES, QNODE, SNODE, MNODE, BNODE, NODE_MAX } EDndType;
 typedef enum { DND_STAT_INIT, DND_STAT_RUNNING, DND_STAT_STOPPED } EDndStatus;
 typedef enum { DND_ENV_INIT, DND_ENV_READY, DND_ENV_CLEANUP } EEnvStatus;
 typedef enum { PROC_SINGLE, PROC_CHILD, PROC_PARENT } EProcType;
@@ -92,6 +92,7 @@ typedef struct SMgmtWrapper {
   char       *path;
   int32_t     refCount;
   SRWLatch    latch;
+  EDndType    ntype;
   bool        deployed;
   bool        required;
   EProcType   procType;
@@ -125,7 +126,7 @@ typedef struct SDnode {
   int32_t      numOfDisks;
   uint16_t     serverPort;
   bool         dropped;
-  ENodeType    ntype;
+  EDndType     ntype;
   EDndStatus   status;
   EDndEvent    event;
   SStartupReq  startup;
@@ -134,34 +135,64 @@ typedef struct SDnode {
   SMgmtWrapper wrappers[NODE_MAX];
 } SDnode;
 
-const char *dndNodeLogStr(ENodeType ntype);
+// dndEnv.c
-const char *dndNodeProcStr(ENodeType ntype);
+const char *dndStatStr(EDndStatus stat);
+const char *dndNodeLogStr(EDndType ntype);
+const char *dndNodeProcStr(EDndType ntype);
+const char *dndEventStr(EDndEvent ev);
+
+// dndExec.c
+int32_t dndOpenNode(SMgmtWrapper *pWrapper);
+void    dndCloseNode(SMgmtWrapper *pWrapper);
+
+// dndFile.c
+int32_t   dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed);
+int32_t   dndWriteFile(SMgmtWrapper *pWrapper, bool deployed);
+TdFilePtr dndCheckRunning(const char *dataDir);
+int32_t   dndReadShmFile(SDnode *pDnode);
+int32_t   dndWriteShmFile(SDnode *pDnode);
+
+// dndInt.c
 EDndStatus    dndGetStatus(SDnode *pDnode);
 void          dndSetStatus(SDnode *pDnode, EDndStatus stat);
 void          dndSetMsgHandle(SMgmtWrapper *pWrapper, tmsg_t msgType, NodeMsgFp nodeMsgFp, int8_t vgId);
-void        dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc);
+SMgmtWrapper *dndAcquireWrapper(SDnode *pDnode, EDndType nType);
-void        dndSendMonitorReport(SDnode *pDnode);
-
-int32_t dndInitServer(SDnode *pDnode);
-void    dndCleanupServer(SDnode *pDnode);
-int32_t dndInitClient(SDnode *pDnode);
-void    dndCleanupClient(SDnode *pDnode);
-int32_t dndProcessNodeMsg(SDnode *pDnode, SNodeMsg *pMsg);
-int32_t dndSendReqToMnode(SMgmtWrapper *pWrapper, SRpcMsg *pMsg);
-int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, const SEpSet *pEpSet, SRpcMsg *pMsg);
-void    dndSendRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp);
-void    dndRegisterBrokenLinkArg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg);
-SMsgCb  dndCreateMsgcb(SMgmtWrapper *pWrapper);
-
-int32_t dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed);
-int32_t dndWriteFile(SMgmtWrapper *pWrapper, bool deployed);
-
-SMgmtWrapper *dndAcquireWrapper(SDnode *pDnode, ENodeType nodeType);
 int32_t       dndMarkWrapper(SMgmtWrapper *pWrapper);
 void          dndReleaseWrapper(SMgmtWrapper *pWrapper);
+void          dndHandleEvent(SDnode *pDnode, EDndEvent event);
+void          dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc);
+void          dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pMsg);
+
+// dndTransport.c
+int32_t  dndInitTrans(SDnode *pDnode);
+void     dndCleanupTrans(SDnode *pDnode);
+SMsgCb   dndCreateMsgcb(SMgmtWrapper *pWrapper);
+SProcCfg dndGenProcCfg(SMgmtWrapper *pWrapper);
+int32_t  dndInitMsgHandle(SDnode *pDnode);
+void     dndSendRecv(SDnode *pDnode, SEpSet *pEpSet, SRpcMsg *pReq, SRpcMsg *pRsp);
+
+// mgmt
+void dmSetMgmtFp(SMgmtWrapper *pWrapper);
+void bmSetMgmtFp(SMgmtWrapper *pWrapper);
+void qmSetMgmtFp(SMgmtWrapper *pMgmt);
+void smSetMgmtFp(SMgmtWrapper *pWrapper);
+void vmSetMgmtFp(SMgmtWrapper *pWrapper);
+void mmSetMgmtFp(SMgmtWrapper *pMgmt);
+
+void dmGetMnodeEpSet(SDnodeMgmt *pMgmt, SEpSet *pEpSet);
+void dmUpdateMnodeEpSet(SDnodeMgmt *pMgmt, SEpSet *pEpSet);
+void dmSendRedirectRsp(SDnodeMgmt *pMgmt, const SRpcMsg *pMsg);
+
+void dmGetMonitorSysInfo(SMonSysInfo *pInfo);
+void vmGetVnodeLoads(SMgmtWrapper *pWrapper, SMonVloadInfo *pInfo);
+void mmGetMonitorInfo(SMgmtWrapper *pWrapper, SMonMmInfo *mmInfo);
+void vmGetMonitorInfo(SMgmtWrapper *pWrapper, SMonVmInfo *vmInfo);
+void qmGetMonitorInfo(SMgmtWrapper *pWrapper, SMonQmInfo *qmInfo);
+void smGetMonitorInfo(SMgmtWrapper *pWrapper, SMonSmInfo *smInfo);
+void bmGetMonitorInfo(SMgmtWrapper *pWrapper, SMonBmInfo *bmInfo);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif /*_TD_DND_H_*/
+#endif /*_TD_DND_INT_H_*/

source/dnode/mgmt/inc/mmInt.h

@@ -16,7 +16,7 @@
 #ifndef _TD_DND_MNODE_INT_H_
 #define _TD_DND_MNODE_INT_H_
 
-#include "mm.h"
+#include "dndInt.h"
 #include "mnode.h"
 
 #ifdef __cplusplus
@@ -32,6 +32,7 @@ typedef struct SMnodeMgmt {
   SSingleWorker readWorker;
   SSingleWorker writeWorker;
   SSingleWorker syncWorker;
+  SSingleWorker monitorWorker;
   SReplica      replicas[TSDB_MAX_REPLICA];
   int8_t        replica;
   int8_t        selfIndex;
@@ -46,11 +47,12 @@ int32_t mmOpenFromMsg(SMgmtWrapper *pWrapper, SDCreateMnodeReq *pReq);
 int32_t mmDrop(SMgmtWrapper *pWrapper);
 int32_t mmAlter(SMnodeMgmt *pMgmt, SDAlterMnodeReq *pReq);
 
-// mmMsg.c
+// mmHandle.c
-void    mmInitMsgHandles(SMgmtWrapper *pWrapper);
+void    mmInitMsgHandle(SMgmtWrapper *pWrapper);
 int32_t mmProcessCreateReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t mmProcessDropReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t mmProcessAlterReq(SMnodeMgmt *pMgmt, SNodeMsg *pMsg);
+int32_t mmProcessGetMonMmInfoReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq);
 
 // mmWorker.c
 int32_t mmStartWorker(SMnodeMgmt *pMgmt);
@@ -59,6 +61,7 @@ int32_t mmProcessWriteMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t mmProcessSyncMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t mmProcessReadMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t mmProcessQueryMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t mmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t mmPutMsgToQueryQueue(SMgmtWrapper *pWrapper, SRpcMsg *pRpc);
 int32_t mmPutMsgToReadQueue(SMgmtWrapper *pWrapper, SRpcMsg *pRpc);
 int32_t mmPutMsgToWriteQueue(SMgmtWrapper *pWrapper, SRpcMsg *pRpc);

source/dnode/mgmt/inc/qmInt.h

@@ -16,7 +16,7 @@
 #ifndef _TD_DND_QNODE_INT_H_
 #define _TD_DND_QNODE_INT_H_
 
-#include "qm.h"
+#include "dndInt.h"
 #include "qnode.h"
 
 #ifdef __cplusplus
@@ -30,16 +30,18 @@ typedef struct SQnodeMgmt {
   const char   *path;
   SSingleWorker queryWorker;
   SSingleWorker fetchWorker;
+  SSingleWorker monitorWorker;
 } SQnodeMgmt;
 
 // qmInt.c
 int32_t qmOpen(SMgmtWrapper *pWrapper);
 int32_t qmDrop(SMgmtWrapper *pWrapper);
 
-// qmMsg.c
+// qmHandle.c
-void    qmInitMsgHandles(SMgmtWrapper *pWrapper);
+void    qmInitMsgHandle(SMgmtWrapper *pWrapper);
 int32_t qmProcessCreateReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t qmProcessDropReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t qmProcessGetMonQmInfoReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq);
 
 // qmWorker.c
 int32_t qmPutMsgToQueryQueue(SMgmtWrapper *pWrapper, SRpcMsg *pMsg);
@@ -50,6 +52,7 @@ int32_t qmStartWorker(SQnodeMgmt *pMgmt);
 void    qmStopWorker(SQnodeMgmt *pMgmt);
 int32_t qmProcessQueryMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t qmProcessFetchMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t qmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/inc/smInt.h

@@ -16,7 +16,7 @@
 #ifndef _TD_DND_SNODE_INT_H_
 #define _TD_DND_SNODE_INT_H_
 
-#include "sm.h"
+#include "dndInt.h"
 #include "snode.h"
 
 #ifdef __cplusplus
@@ -32,16 +32,18 @@ typedef struct SSnodeMgmt {
   int8_t        uniqueWorkerInUse;
   SArray       *uniqueWorkers;  // SArray<SMultiWorker*>
   SSingleWorker sharedWorker;
+  SSingleWorker monitorWorker;
 } SSnodeMgmt;
 
 // smInt.c
 int32_t smOpen(SMgmtWrapper *pWrapper);
 int32_t smDrop(SMgmtWrapper *pWrapper);
 
-// smMsg.c
+// smHandle.c
-void    smInitMsgHandles(SMgmtWrapper *pWrapper);
+void    smInitMsgHandle(SMgmtWrapper *pWrapper);
 int32_t smProcessCreateReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t smProcessDropReq(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t smProcessGetMonSmInfoReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq);
 
 // smWorker.c
 int32_t smStartWorker(SSnodeMgmt *pMgmt);
@@ -50,6 +52,7 @@ int32_t smProcessMgmtMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t smProcessUniqueMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t smProcessSharedMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t smProcessExecMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
+int32_t smProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/inc/vmInt.h

@@ -17,7 +17,7 @@
 #define _TD_DND_VNODES_INT_H_
 
 #include "sync.h"
-#include "vm.h"
+#include "dndInt.h"
 #include "vnode.h"
 
 #ifdef __cplusplus
@@ -28,6 +28,7 @@ typedef struct SVnodesMgmt {
   SHashObj     *hash;
   SRWLatch      latch;
   SVnodesStat   state;
+  SVnodesStat   lastState;
   STfs         *pTfs;
   SQWorkerPool  queryPool;
   SQWorkerPool  fetchPool;
@@ -38,6 +39,7 @@ typedef struct SVnodesMgmt {
   SDnode       *pDnode;
   SMgmtWrapper *pWrapper;
   SSingleWorker mgmtWorker;
+  SSingleWorker monitorWorker;
 } SVnodesMgmt;
 
 typedef struct {
@@ -84,13 +86,15 @@ void       vmReleaseVnode(SVnodesMgmt *pMgmt, SVnodeObj *pVnode);
 int32_t    vmOpenVnode(SVnodesMgmt *pMgmt, SWrapperCfg *pCfg, SVnode *pImpl);
 void       vmCloseVnode(SVnodesMgmt *pMgmt, SVnodeObj *pVnode);
 
-// vmMsg.c
+// vmHandle.c
-void    vmInitMsgHandles(SMgmtWrapper *pWrapper);
+void    vmInitMsgHandle(SMgmtWrapper *pWrapper);
 int32_t vmProcessCreateVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pReq);
 int32_t vmProcessAlterVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pReq);
 int32_t vmProcessDropVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pReq);
 int32_t vmProcessSyncVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pReq);
 int32_t vmProcessCompactVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pReq);
+int32_t vmProcessGetMonVmInfoReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq);
+int32_t vmProcessGetVnodeLoadsReq(SMgmtWrapper *pWrapper, SNodeMsg *pReq);
 
 // vmFile.c
 int32_t     vmGetVnodesFromFile(SVnodesMgmt *pMgmt, SWrapperCfg **ppCfgs, int32_t *numOfVnodes);
@@ -114,6 +118,7 @@ int32_t vmProcessQueryMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t vmProcessFetchMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t vmProcessMergeMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 int32_t vmProcessMgmtMsg(SMgmtWrapper *pWrappert, SNodeMsg *pMsg);
+int32_t vmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/main/dndEnv.c

@@ -58,39 +58,62 @@ void dndCleanup() {
   dInfo("dnode env is cleaned up");
 }
 
-void dndSetMsgHandle(SMgmtWrapper *pWrapper, tmsg_t msgType, NodeMsgFp nodeMsgFp, int8_t vgId) {
+const char *dndStatStr(EDndStatus status) {
-  pWrapper->msgFps[TMSG_INDEX(msgType)] = nodeMsgFp;
+  switch (status) {
-  pWrapper->msgVgIds[TMSG_INDEX(msgType)] = vgId;
+    case DND_STAT_INIT:
-}
+      return "init";
-
+    case DND_STAT_RUNNING:
-EDndStatus dndGetStatus(SDnode *pDnode) { return pDnode->status; }
+      return "running";
-
+    case DND_STAT_STOPPED:
-void dndSetStatus(SDnode *pDnode, EDndStatus status) {
+      return "stopped";
-  if (pDnode->status != status) {
+    default:
-    dDebug("dnode status set from %s to %s", dndStatStr(pDnode->status), dndStatStr(status));
+      return "UNKNOWN";
-    pDnode->status = status;
   }
 }
 
-void dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc) {
+const char *dndNodeLogStr(EDndType ntype) {
-  SStartupReq *pStartup = &pDnode->startup;
+  switch (ntype) {
-  tstrncpy(pStartup->name, pName, TSDB_STEP_NAME_LEN);
+    case VNODES:
-  tstrncpy(pStartup->desc, pDesc, TSDB_STEP_DESC_LEN);
+      return "vnode";
-  pStartup->finished = 0;
+    case QNODE:
+      return "qnode";
+    case SNODE:
+      return "snode";
+    case MNODE:
+      return "mnode";
+    case BNODE:
+      return "bnode";
+    default:
+      return "taosd";
+  }
 }
 
-void dndGetStartup(SDnode *pDnode, SStartupReq *pStartup) {
+const char *dndNodeProcStr(EDndType ntype) {
-  memcpy(pStartup, &pDnode->startup, sizeof(SStartupReq));
+  switch (ntype) {
-  pStartup->finished = (dndGetStatus(pDnode) == DND_STAT_RUNNING);
+    case VNODES:
+      return "taosv";
+    case QNODE:
+      return "taosq";
+    case SNODE:
+      return "taoss";
+    case MNODE:
+      return "taosm";
+    case BNODE:
+      return "taosb";
+    default:
+      return "taosd";
+  }
 }
 
-void dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pReq) {
+const char *dndEventStr(EDndEvent ev) {
-  dDebug("startup req is received");
+  switch (ev) {
-  SStartupReq *pStartup = rpcMallocCont(sizeof(SStartupReq));
+    case DND_EVENT_START:
-  dndGetStartup(pDnode, pStartup);
+      return "start";
-
+    case DND_EVENT_STOP:
-  dDebug("startup req is sent, step:%s desc:%s finished:%d", pStartup->name, pStartup->desc, pStartup->finished);
+      return "stop";
-  SRpcMsg rpcRsp = {
+    case DND_EVENT_CHILD:
-      .handle = pReq->handle, .pCont = pStartup, .contLen = sizeof(SStartupReq), .ahandle = pReq->ahandle};
+      return "child";
-  rpcSendResponse(&rpcRsp);
+    default:
+      return "UNKNOWN";
+  }
 }

source/dnode/mgmt/main/dndExec.c

@@ -20,9 +20,9 @@ static bool dndRequireNode(SMgmtWrapper *pWrapper) {
   bool    required = false;
   int32_t code = (*pWrapper->fp.requiredFp)(pWrapper, &required);
   if (!required) {
-    dDebug("node:%s, no need to start", pWrapper->name);
+    dDebug("node:%s, does not require startup", pWrapper->name);
   } else {
-    dDebug("node:%s, need to start", pWrapper->name);
+    dDebug("node:%s, needs to be started", pWrapper->name);
   }
   return required;
 }
@@ -45,7 +45,7 @@ int32_t dndOpenNode(SMgmtWrapper *pWrapper) {
 }
 
 void dndCloseNode(SMgmtWrapper *pWrapper) {
-  dDebug("node:%s, start to close", pWrapper->name);
+  dDebug("node:%s, mgmt start to close", pWrapper->name);
   pWrapper->required = false;
   taosWLockLatch(&pWrapper->latch);
   if (pWrapper->deployed) {
@@ -62,13 +62,42 @@ void dndCloseNode(SMgmtWrapper *pWrapper) {
     taosProcCleanup(pWrapper->pProc);
     pWrapper->pProc = NULL;
   }
-  dDebug("node:%s, has been closed", pWrapper->name);
+  dDebug("node:%s, mgmt has been closed", pWrapper->name);
+}
+
+
+static int32_t dndNewProc(SMgmtWrapper *pWrapper, EDndType n) {
+  char  tstr[8] = {0};
+  char *args[6] = {0};
+  snprintf(tstr, sizeof(tstr), "%d", n);
+  args[1] = "-c";
+  args[2] = configDir;
+  args[3] = "-n";
+  args[4] = tstr;
+  args[5] = NULL;
+
+  int32_t pid = taosNewProc(args);
+  if (pid <= 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("node:%s, failed to exec in new process since %s", pWrapper->name, terrstr());
+    return -1;
+  }
+
+  pWrapper->procId = pid;
+  dInfo("node:%s, continue running in new process:%d", pWrapper->name, pid);
+  return 0;
+}
+
+static void dndProcessProcHandle(void *handle) {
+  dWarn("handle:%p, the child process dies and send an offline rsp", handle);
+  SRpcMsg rpcMsg = {.handle = handle, .code = TSDB_CODE_DND_OFFLINE};
+  rpcSendResponse(&rpcMsg);
 }
 
 static int32_t dndRunInSingleProcess(SDnode *pDnode) {
-  dInfo("dnode start to run in single process");
+  dInfo("dnode run in single process");
 
-  for (ENodeType n = DNODE; n < NODE_MAX; ++n) {
+  for (EDndType n = DNODE; n < NODE_MAX; ++n) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     pWrapper->required = dndRequireNode(pWrapper);
     if (!pWrapper->required) continue;
@@ -81,7 +110,7 @@ static int32_t dndRunInSingleProcess(SDnode *pDnode) {
 
   dndSetStatus(pDnode, DND_STAT_RUNNING);
 
-  for (ENodeType n = 0; n < NODE_MAX; ++n) {
+  for (EDndType n = 0; n < NODE_MAX; ++n) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     if (!pWrapper->required) continue;
     if (pWrapper->fp.startFp == NULL) continue;
@@ -91,90 +120,56 @@ static int32_t dndRunInSingleProcess(SDnode *pDnode) {
     }
   }
 
-  dInfo("dnode running in single process");
+  dInfo("TDengine initialized successfully");
+  dndReportStartup(pDnode, "TDengine", "initialized successfully");
+  while (1) {
+    if (pDnode->event == DND_EVENT_STOP) {
+      dInfo("dnode is about to stop");
+      dndSetStatus(pDnode, DND_STAT_STOPPED);
+      break;
+    }
+    taosMsleep(100);
+  }
+
   return 0;
 }
 
-static void dndConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
-                                 ProcFuncType ftype) {
-  SRpcMsg *pRpc = &pMsg->rpcMsg;
-  pRpc->pCont = pCont;
-  dTrace("msg:%p, get from child queue, handle:%p app:%p", pMsg, pRpc->handle, pRpc->ahandle);
-
-  NodeMsgFp msgFp = pWrapper->msgFps[TMSG_INDEX(pRpc->msgType)];
-  int32_t   code = (*msgFp)(pWrapper, pMsg);
-
-  if (code != 0) {
-    dError("msg:%p, failed to process since code:0x%04x:%s", pMsg, code & 0XFFFF, tstrerror(code));
-    if (pRpc->msgType & 1U) {
-      SRpcMsg rsp = {.handle = pRpc->handle, .ahandle = pRpc->ahandle, .code = terrno};
-      dndSendRsp(pWrapper, &rsp);
-    }
-
-    dTrace("msg:%p, is freed", pMsg);
-    taosFreeQitem(pMsg);
-    rpcFreeCont(pCont);
-  }
-}
-
-static void dndConsumeParentQueue(SMgmtWrapper *pWrapper, SRpcMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
-                                  ProcFuncType ftype) {
-  pMsg->pCont = pCont;
-  dTrace("msg:%p, get from parent queue, handle:%p app:%p", pMsg, pMsg->handle, pMsg->ahandle);
-
-  switch (ftype) {
-    case PROC_REG:
-      rpcRegisterBrokenLinkArg(pMsg);
-      break;
-    case PROC_RELEASE:
-      rpcReleaseHandle(pMsg->handle, (int8_t)pMsg->code);
-      rpcFreeCont(pCont);
-      break;
-    case PROC_REQ:
-      // todo send to dnode
-      dndSendReqToMnode(pWrapper, pMsg);
-    default:
-      dndSendRpcRsp(pWrapper, pMsg);
-      break;
-  }
-  taosMemoryFree(pMsg);
-}
-
 static int32_t dndRunInParentProcess(SDnode *pDnode) {
-  dInfo("dnode start to run in parent process");
+  dInfo("dnode run in parent process");
   SMgmtWrapper *pDWrapper = &pDnode->wrappers[DNODE];
   if (dndOpenNode(pDWrapper) != 0) {
     dError("node:%s, failed to start since %s", pDWrapper->name, terrstr());
     return -1;
   }
 
-  for (ENodeType n = DNODE + 1; n < NODE_MAX; ++n) {
+  for (EDndType n = DNODE + 1; n < NODE_MAX; ++n) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     pWrapper->required = dndRequireNode(pWrapper);
     if (!pWrapper->required) continue;
 
-    int64_t shmsize = 1024 * 1024 * 2;  // size will be a configuration item
+    int32_t shmsize = tsMnodeShmSize;
-    if (taosCreateShm(&pWrapper->shm, shmsize) != 0) {
+    if (n == VNODES) {
-      terrno = TAOS_SYSTEM_ERROR(terrno);
+      shmsize = tsVnodeShmSize;
-      dError("node:%s, failed to create shm size:%" PRId64 " since %s", pWrapper->name, shmsize, terrstr());
+    } else if (n == QNODE) {
-      return -1;
+      shmsize = tsQnodeShmSize;
+    } else if (n == SNODE) {
+      shmsize = tsSnodeShmSize;
+    } else if (n == MNODE) {
+      shmsize = tsMnodeShmSize;
+    } else if (n == BNODE) {
+      shmsize = tsBnodeShmSize;
+    } else {
     }
 
-    SProcCfg cfg = {.childConsumeFp = (ProcConsumeFp)dndConsumeChildQueue,
+    if (taosCreateShm(&pWrapper->shm, n, shmsize) != 0) {
-                    .childMallocHeadFp = (ProcMallocFp)taosAllocateQitem,
+      terrno = TAOS_SYSTEM_ERROR(terrno);
-                    .childFreeHeadFp = (ProcFreeFp)taosFreeQitem,
+      dError("node:%s, failed to create shm size:%d since %s", pWrapper->name, shmsize, terrstr());
-                    .childMallocBodyFp = (ProcMallocFp)rpcMallocCont,
+      return -1;
-                    .childFreeBodyFp = (ProcFreeFp)rpcFreeCont,
+    }
-                    .parentConsumeFp = (ProcConsumeFp)dndConsumeParentQueue,
+    dInfo("node:%s, shm:%d is created, size:%d", pWrapper->name, pWrapper->shm.id, shmsize);
-                    .parentMallocHeadFp = (ProcMallocFp)taosMemoryMalloc,
-                    .parentFreeHeadFp = (ProcFreeFp)taosMemoryFree,
-                    .parentMallocBodyFp = (ProcMallocFp)rpcMallocCont,
-                    .parentFreeBodyFp = (ProcFreeFp)rpcFreeCont,
-                    .shm = pWrapper->shm,
-                    .pParent = pWrapper,
-                    .isChild = false,
-                    .name = pWrapper->name};
 
+    SProcCfg cfg = dndGenProcCfg(pWrapper);
+    cfg.isChild = false;
     pWrapper->procType = PROC_PARENT;
     pWrapper->pProc = taosProcInit(&cfg);
     if (pWrapper->pProc == NULL) {
@@ -188,22 +183,16 @@ static int32_t dndRunInParentProcess(SDnode *pDnode) {
     return -1;
   }
 
-  for (ENodeType n = DNODE + 1; n < NODE_MAX; ++n) {
+  for (EDndType n = DNODE + 1; n < NODE_MAX; ++n) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     if (!pWrapper->required) continue;
 
     if (pDnode->ntype == NODE_MAX) {
       dInfo("node:%s, should be started manually", pWrapper->name);
     } else {
-      char    args[PATH_MAX];
+      if (dndNewProc(pWrapper, n) != 0) {
-      int32_t pid = taosNewProc(args);
-      if (pid <= 0) {
-        terrno = TAOS_SYSTEM_ERROR(errno);
-        dError("node:%s, failed to exec in new process since %s", pWrapper->name, terrstr());
         return -1;
       }
-      pWrapper->procId = pid;
-      dInfo("node:%s, run in new process, pid:%d", pWrapper->name, pid);
     }
 
     if (taosProcRun(pWrapper->pProc) != 0) {
@@ -219,13 +208,57 @@ static int32_t dndRunInParentProcess(SDnode *pDnode) {
     return -1;
   }
 
-  dInfo("dnode running in parent process");
+  dInfo("TDengine initialized successfully");
+  dndReportStartup(pDnode, "TDengine", "initialized successfully");
+
+  while (1) {
+    if (pDnode->event == DND_EVENT_STOP) {
+      dInfo("dnode is about to stop");
+      dndSetStatus(pDnode, DND_STAT_STOPPED);
+
+      for (EDndType n = DNODE + 1; n < NODE_MAX; ++n) {
+        SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
+        if (!pWrapper->required) continue;
+        if (pDnode->ntype == NODE_MAX) continue;
+
+        if (pWrapper->procId > 0 && taosProcExist(pWrapper->procId)) {
+          dInfo("node:%s, send kill signal to the child process:%d", pWrapper->name, pWrapper->procId);
+          taosKillProc(pWrapper->procId);
+          dInfo("node:%s, wait for child process:%d to stop", pWrapper->name, pWrapper->procId);
+          taosWaitProc(pWrapper->procId);
+          dInfo("node:%s, child process:%d is stopped", pWrapper->name, pWrapper->procId);
+        }
+      }
+      break;
+    } else {
+      for (EDndType n = DNODE + 1; n < NODE_MAX; ++n) {
+        SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
+        if (!pWrapper->required) continue;
+        if (pDnode->ntype == NODE_MAX) continue;
+
+        if (pWrapper->procId <= 0 || !taosProcExist(pWrapper->procId)) {
+          dWarn("node:%s, process:%d is killed and needs to be restarted", pWrapper->name, pWrapper->procId);
+          taosProcCloseHandles(pWrapper->pProc, dndProcessProcHandle);
+          dndNewProc(pWrapper, n);
+        }
+      }
+    }
+
+    taosMsleep(100);
+  }
+
   return 0;
 }
 
 static int32_t dndRunInChildProcess(SDnode *pDnode) {
-  dInfo("dnode start to run in child process");
   SMgmtWrapper *pWrapper = &pDnode->wrappers[pDnode->ntype];
+  dInfo("%s run in child process", pWrapper->name);
+
+  pWrapper->required = dndRequireNode(pWrapper);
+  if (!pWrapper->required) {
+    dError("%s does not require startup", pWrapper->name);
+    return -1;
+  }
 
   SMsgCb msgCb = dndCreateMsgcb(pWrapper);
   tmsgSetDefaultMsgCb(&msgCb);
@@ -236,21 +269,8 @@ static int32_t dndRunInChildProcess(SDnode *pDnode) {
     return -1;
   }
 
-  SProcCfg cfg = {.childConsumeFp = (ProcConsumeFp)dndConsumeChildQueue,
+  SProcCfg cfg = dndGenProcCfg(pWrapper);
-                  .childMallocHeadFp = (ProcMallocFp)taosAllocateQitem,
+  cfg.isChild = true;
-                  .childFreeHeadFp = (ProcFreeFp)taosFreeQitem,
-                  .childMallocBodyFp = (ProcMallocFp)rpcMallocCont,
-                  .childFreeBodyFp = (ProcFreeFp)rpcFreeCont,
-                  .parentConsumeFp = (ProcConsumeFp)dndConsumeParentQueue,
-                  .parentMallocHeadFp = (ProcMallocFp)taosMemoryMalloc,
-                  .parentFreeHeadFp = (ProcFreeFp)taosMemoryFree,
-                  .parentMallocBodyFp = (ProcMallocFp)rpcMallocCont,
-                  .parentFreeBodyFp = (ProcFreeFp)rpcFreeCont,
-                  .shm = pWrapper->shm,
-                  .pParent = pWrapper,
-                  .isChild = true,
-                  .name = pWrapper->name};
-
   pWrapper->pProc = taosProcInit(&cfg);
   if (pWrapper->pProc == NULL) {
     dError("node:%s, failed to create proc since %s", pWrapper->name, terrstr());
@@ -264,39 +284,19 @@ static int32_t dndRunInChildProcess(SDnode *pDnode) {
     }
   }
 
+  dndSetStatus(pDnode, DND_STAT_RUNNING);
+
   if (taosProcRun(pWrapper->pProc) != 0) {
     dError("node:%s, failed to run proc since %s", pWrapper->name, terrstr());
     return -1;
   }
 
-  dInfo("dnode running in child process");
-  return 0;
-}
-
-int32_t dndRun(SDnode * pDnode) {
-  if (!tsMultiProcess) {
-    if (dndRunInSingleProcess(pDnode) != 0) {
-      dError("failed to run dnode since %s", terrstr());
-      return -1;
-    }
-  } else if (pDnode->ntype == DNODE || pDnode->ntype == NODE_MAX) {
-    if (dndRunInParentProcess(pDnode) != 0) {
-      dError("failed to run dnode in parent process since %s", terrstr());
-      return -1;
-    }
-  } else {
-    if (dndRunInChildProcess(pDnode) != 0) {
-      dError("failed to run dnode in child process since %s", terrstr());
-      return -1;
-    }
-  }
-
-  dndReportStartup(pDnode, "TDengine", "initialized successfully");
   dInfo("TDengine initialized successfully");
-
+  dndReportStartup(pDnode, "TDengine", "initialized successfully");
   while (1) {
     if (pDnode->event == DND_EVENT_STOP) {
-      dInfo("dnode is about to stop");
+      dInfo("%s is about to stop", pWrapper->name);
+      dndSetStatus(pDnode, DND_STAT_STOPPED);
       break;
     }
     taosMsleep(100);
@@ -304,3 +304,15 @@ int32_t dndRun(SDnode * pDnode) {
 
   return 0;
 }
+
+int32_t dndRun(SDnode *pDnode) {
+  if (!tsMultiProcess) {
+    return dndRunInSingleProcess(pDnode);
+  } else if (pDnode->ntype == DNODE || pDnode->ntype == NODE_MAX) {
+    return dndRunInParentProcess(pDnode);
+  } else {
+    return dndRunInChildProcess(pDnode);
+  }
+
+  return 0;
+}

source/dnode/mgmt/main/dndFile.c

@@ -164,7 +164,7 @@ int32_t dndReadShmFile(SDnode *pDnode) {
       goto _OVER;
     }
 
-    for (ENodeType ntype = DNODE + 1; ntype < NODE_MAX; ++ntype) {
+    for (EDndType ntype = DNODE + 1; ntype < NODE_MAX; ++ntype) {
       snprintf(itemName, sizeof(itemName), "%s_shmid", dndNodeProcStr(ntype));
       cJSON *shmid = cJSON_GetObjectItem(root, itemName);
       if (shmid && shmid->type == cJSON_Number) {
@@ -179,8 +179,8 @@ int32_t dndReadShmFile(SDnode *pDnode) {
     }
   }
 
-  if (!tsMultiProcess || pDnode->ntype == DNODE || pDnode->ntype == DNODE) {
+  if (!tsMultiProcess || pDnode->ntype == DNODE || pDnode->ntype == NODE_MAX) {
-    for (ENodeType ntype = DNODE; ntype < NODE_MAX; ++ntype) {
+    for (EDndType ntype = DNODE; ntype < NODE_MAX; ++ntype) {
       SMgmtWrapper *pWrapper = &pDnode->wrappers[ntype];
       if (pWrapper->shm.id >= 0) {
         dDebug("shmid:%d, is closed, size:%d", pWrapper->shm.id, pWrapper->shm.size);
@@ -194,10 +194,10 @@ int32_t dndReadShmFile(SDnode *pDnode) {
       dError("shmid:%d, failed to attach shm since %s", pWrapper->shm.id, terrstr());
       goto _OVER;
     }
-    dDebug("shmid:%d, is attached, size:%d", pWrapper->shm.id, pWrapper->shm.size);
+    dInfo("node:%s, shmid:%d is attached, size:%d", pWrapper->name, pWrapper->shm.id, pWrapper->shm.size);
   }
 
-  dDebug("successed to open %s", file);
+  dDebug("successed to load %s", file);
   code = 0;
 
 _OVER:
@@ -226,7 +226,7 @@ int32_t dndWriteShmFile(SDnode *pDnode) {
   }
 
   len += snprintf(content + len, MAXLEN - len, "{\n");
-  for (ENodeType ntype = DNODE + 1; ntype < NODE_MAX; ++ntype) {
+  for (EDndType ntype = DNODE + 1; ntype < NODE_MAX; ++ntype) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[ntype];
     len += snprintf(content + len, MAXLEN - len, "  \"%s_shmid\":%d,\n", dndNodeProcStr(ntype), pWrapper->shm.id);
     if (ntype == NODE_MAX - 1) {

source/dnode/mgmt/main/dndInt.c

@@ -46,7 +46,7 @@ static int32_t dndInitVars(SDnode *pDnode, const SDnodeOpt *pOption) {
 }
 
 static void dndClearVars(SDnode *pDnode) {
-  for (ENodeType n = 0; n < NODE_MAX; ++n) {
+  for (EDndType n = 0; n < NODE_MAX; ++n) {
     SMgmtWrapper *pMgmt = &pDnode->wrappers[n];
     taosMemoryFreeClear(pMgmt->path);
   }
@@ -61,7 +61,7 @@ static void dndClearVars(SDnode *pDnode) {
   taosMemoryFreeClear(pDnode->secondEp);
   taosMemoryFreeClear(pDnode->dataDir);
   taosMemoryFree(pDnode);
-  dDebug("dnode object memory is cleared, data:%p", pDnode);
+  dDebug("dnode memory is cleared, data:%p", pDnode);
 }
 
 SDnode *dndCreate(const SDnodeOpt *pOption) {
@@ -82,19 +82,20 @@ SDnode *dndCreate(const SDnodeOpt *pOption) {
   }
 
   dndSetStatus(pDnode, DND_STAT_INIT);
-  dmGetMgmtFp(&pDnode->wrappers[DNODE]);
+  dmSetMgmtFp(&pDnode->wrappers[DNODE]);
-  mmGetMgmtFp(&pDnode->wrappers[MNODE]);
+  mmSetMgmtFp(&pDnode->wrappers[MNODE]);
-  vmGetMgmtFp(&pDnode->wrappers[VNODES]);
+  vmSetMgmtFp(&pDnode->wrappers[VNODES]);
-  qmGetMgmtFp(&pDnode->wrappers[QNODE]);
+  qmSetMgmtFp(&pDnode->wrappers[QNODE]);
-  smGetMgmtFp(&pDnode->wrappers[SNODE]);
+  smSetMgmtFp(&pDnode->wrappers[SNODE]);
-  bmGetMgmtFp(&pDnode->wrappers[BNODE]);
+  bmSetMgmtFp(&pDnode->wrappers[BNODE]);
 
-  for (ENodeType n = 0; n < NODE_MAX; ++n) {
+  for (EDndType n = 0; n < NODE_MAX; ++n) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     snprintf(path, sizeof(path), "%s%s%s", pDnode->dataDir, TD_DIRSEP, pWrapper->name);
     pWrapper->path = strdup(path);
     pWrapper->shm.id = -1;
     pWrapper->pDnode = pDnode;
+    pWrapper->ntype = n;
     if (pWrapper->path == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       goto _OVER;
@@ -105,7 +106,7 @@ SDnode *dndCreate(const SDnodeOpt *pOption) {
   }
 
   if (dndInitMsgHandle(pDnode) != 0) {
-    dError("failed to msg handles since %s", terrstr());
+    dError("failed to init msg handles since %s", terrstr());
     goto _OVER;
   }
 
@@ -117,14 +118,14 @@ SDnode *dndCreate(const SDnodeOpt *pOption) {
   SMsgCb msgCb = dndCreateMsgcb(&pDnode->wrappers[0]);
   tmsgSetDefaultMsgCb(&msgCb);
 
-  dInfo("dnode object is created, data:%p", pDnode);
+  dInfo("dnode is created, data:%p", pDnode);
   code = 0;
 
 _OVER:
   if (code != 0 && pDnode) {
     dndClearVars(pDnode);
     pDnode = NULL;
-    dError("failed to create dnode object since %s", terrstr());
+    dError("failed to create dnode since %s", terrstr());
   }
 
   return pDnode;
@@ -133,29 +134,22 @@ _OVER:
 void dndClose(SDnode *pDnode) {
   if (pDnode == NULL) return;
 
-  if (dndGetStatus(pDnode) == DND_STAT_STOPPED) {
+  for (EDndType n = 0; n < NODE_MAX; ++n) {
-    dError("dnode is shutting down, data:%p", pDnode);
-    return;
-  }
-
-  dInfo("start to close dnode, data:%p", pDnode);
-  dndSetStatus(pDnode, DND_STAT_STOPPED);
-
-  for (ENodeType n = 0; n < NODE_MAX; ++n) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     dndCloseNode(pWrapper);
   }
 
   dndClearVars(pDnode);
-  dInfo("dnode object is closed, data:%p", pDnode);
+  dInfo("dnode is closed, data:%p", pDnode);
 }
 
 void dndHandleEvent(SDnode *pDnode, EDndEvent event) {
-  dInfo("dnode object receive event %d, data:%p", event, pDnode);
+  if (event == DND_EVENT_STOP) {
     pDnode->event = event;
   }
+}
 
-SMgmtWrapper *dndAcquireWrapper(SDnode *pDnode, ENodeType ntype) {
+SMgmtWrapper *dndAcquireWrapper(SDnode *pDnode, EDndType ntype) {
   SMgmtWrapper *pWrapper = &pDnode->wrappers[ntype];
   SMgmtWrapper *pRetWrapper = pWrapper;
 
@@ -196,3 +190,40 @@ void dndReleaseWrapper(SMgmtWrapper *pWrapper) {
   taosRUnLockLatch(&pWrapper->latch);
   dTrace("node:%s, is released, refCount:%d", pWrapper->name, refCount);
 }
+
+void dndSetMsgHandle(SMgmtWrapper *pWrapper, tmsg_t msgType, NodeMsgFp nodeMsgFp, int8_t vgId) {
+  pWrapper->msgFps[TMSG_INDEX(msgType)] = nodeMsgFp;
+  pWrapper->msgVgIds[TMSG_INDEX(msgType)] = vgId;
+}
+
+EDndStatus dndGetStatus(SDnode *pDnode) { return pDnode->status; }
+
+void dndSetStatus(SDnode *pDnode, EDndStatus status) {
+  if (pDnode->status != status) {
+    dDebug("dnode status set from %s to %s", dndStatStr(pDnode->status), dndStatStr(status));
+    pDnode->status = status;
+  }
+}
+
+void dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc) {
+  SStartupReq *pStartup = &pDnode->startup;
+  tstrncpy(pStartup->name, pName, TSDB_STEP_NAME_LEN);
+  tstrncpy(pStartup->desc, pDesc, TSDB_STEP_DESC_LEN);
+  pStartup->finished = 0;
+}
+
+static void dndGetStartup(SDnode *pDnode, SStartupReq *pStartup) {
+  memcpy(pStartup, &pDnode->startup, sizeof(SStartupReq));
+  pStartup->finished = (dndGetStatus(pDnode) == DND_STAT_RUNNING);
+}
+
+void dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pReq) {
+  dDebug("startup req is received");
+  SStartupReq *pStartup = rpcMallocCont(sizeof(SStartupReq));
+  dndGetStartup(pDnode, pStartup);
+
+  dDebug("startup req is sent, step:%s desc:%s finished:%d", pStartup->name, pStartup->desc, pStartup->finished);
+  SRpcMsg rpcRsp = {
+      .handle = pReq->handle, .pCont = pStartup, .contLen = sizeof(SStartupReq), .ahandle = pReq->ahandle};
+  rpcSendResponse(&rpcRsp);
+}

source/dnode/mgmt/main/dndTransport.c

@@ -0,0 +1,489 @@
+/*
+ * 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 "dndInt.h"
+
+#define INTERNAL_USER   "_dnd"
+#define INTERNAL_CKEY   "_key"
+#define INTERNAL_SECRET "_pwd"
+
+static void dndUpdateMnodeEpSet(SDnode *pDnode, SEpSet *pEpSet) {
+  SMgmtWrapper *pWrapper = &pDnode->wrappers[DNODE];
+  dmUpdateMnodeEpSet(pWrapper->pMgmt, pEpSet);
+}
+
+static inline NodeMsgFp dndGetMsgFp(SMgmtWrapper *pWrapper, SRpcMsg *pRpc) {
+  NodeMsgFp msgFp = pWrapper->msgFps[TMSG_INDEX(pRpc->msgType)];
+  if (msgFp == NULL) {
+    terrno = TSDB_CODE_MSG_NOT_PROCESSED;
+  }
+
+  return msgFp;
+}
+
+static inline int32_t dndBuildMsg(SNodeMsg *pMsg, SRpcMsg *pRpc) {
+  SRpcConnInfo connInfo = {0};
+  if ((pRpc->msgType & 1U) && rpcGetConnInfo(pRpc->handle, &connInfo) != 0) {
+    terrno = TSDB_CODE_MND_NO_USER_FROM_CONN;
+    dError("failed to build msg since %s, app:%p handle:%p", terrstr(), pRpc->ahandle, pRpc->handle);
+    return -1;
+  }
+
+  memcpy(pMsg->user, connInfo.user, TSDB_USER_LEN);
+  pMsg->clientIp = connInfo.clientIp;
+  pMsg->clientPort = connInfo.clientPort;
+  memcpy(&pMsg->rpcMsg, pRpc, sizeof(SRpcMsg));
+  return 0;
+}
+
+static void dndProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSet) {
+  int32_t   code = -1;
+  SNodeMsg *pMsg = NULL;
+  NodeMsgFp msgFp = NULL;
+
+  if (pEpSet && pEpSet->numOfEps > 0 && pRpc->msgType == TDMT_MND_STATUS_RSP) {
+    dndUpdateMnodeEpSet(pWrapper->pDnode, pEpSet);
+  }
+
+  if (dndMarkWrapper(pWrapper) != 0) goto _OVER;
+  if ((msgFp = dndGetMsgFp(pWrapper, pRpc)) == NULL) goto _OVER;
+  if ((pMsg = taosAllocateQitem(sizeof(SNodeMsg))) == NULL) goto _OVER;
+  if (dndBuildMsg(pMsg, pRpc) != 0) goto _OVER;
+
+  if (pWrapper->procType == PROC_SINGLE) {
+    dTrace("msg:%p, is created, handle:%p user:%s", pMsg, pRpc->handle, pMsg->user);
+    code = (*msgFp)(pWrapper, pMsg);
+  } else if (pWrapper->procType == PROC_PARENT) {
+    dTrace("msg:%p, is created and put into child queue, handle:%p user:%s", pMsg, pRpc->handle, pMsg->user);
+    code = taosProcPutToChildQ(pWrapper->pProc, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen, pRpc->handle,
+                               PROC_REQ);
+  } else {
+    dTrace("msg:%p, should not processed in child process, handle:%p user:%s", pMsg, pRpc->handle, pMsg->user);
+    ASSERT(1);
+  }
+
+_OVER:
+  if (code == 0) {
+    if (pWrapper->procType == PROC_PARENT) {
+      dTrace("msg:%p, is freed in parent process", pMsg);
+      taosFreeQitem(pMsg);
+      rpcFreeCont(pRpc->pCont);
+    }
+  } else {
+    dError("msg:%p, failed to process since 0x%04x:%s", pMsg, code & 0XFFFF, terrstr());
+    if (pRpc->msgType & 1U) {
+      SRpcMsg rsp = {.handle = pRpc->handle, .ahandle = pRpc->ahandle, .code = terrno};
+      tmsgSendRsp(&rsp);
+    }
+    dTrace("msg:%p, is freed", pMsg);
+    taosFreeQitem(pMsg);
+    rpcFreeCont(pRpc->pCont);
+  }
+
+  dndReleaseWrapper(pWrapper);
+}
+
+static void dndProcessMsg(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet) {
+  STransMgmt   *pMgmt = &pDnode->trans;
+  tmsg_t        msgType = pMsg->msgType;
+  bool          isReq = msgType & 1u;
+  SMsgHandle   *pHandle = &pMgmt->msgHandles[TMSG_INDEX(msgType)];
+  SMgmtWrapper *pWrapper = pHandle->pWrapper;
+
+  if (msgType == TDMT_DND_NETWORK_TEST) {
+    dTrace("network test req will be processed, handle:%p, app:%p", pMsg->handle, pMsg->ahandle);
+    dndProcessStartupReq(pDnode, pMsg);
+    return;
+  }
+
+  if (dndGetStatus(pDnode) != DND_STAT_RUNNING) {
+    dError("msg:%s ignored since dnode not running, handle:%p app:%p", TMSG_INFO(msgType), pMsg->handle, pMsg->ahandle);
+    if (isReq) {
+      SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_APP_NOT_READY, .ahandle = pMsg->ahandle};
+      rpcSendResponse(&rspMsg);
+    }
+    rpcFreeCont(pMsg->pCont);
+    return;
+  }
+
+  if (isReq && pMsg->pCont == NULL) {
+    dError("req:%s not processed since its empty, handle:%p app:%p", TMSG_INFO(msgType), pMsg->handle, pMsg->ahandle);
+    SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_DND_INVALID_MSG_LEN, .ahandle = pMsg->ahandle};
+    rpcSendResponse(&rspMsg);
+    return;
+  }
+
+  if (pWrapper == NULL) {
+    dError("msg:%s not processed since no handle, handle:%p app:%p", TMSG_INFO(msgType), pMsg->handle, pMsg->ahandle);
+    if (isReq) {
+      SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_MSG_NOT_PROCESSED, .ahandle = pMsg->ahandle};
+      rpcSendResponse(&rspMsg);
+    }
+    rpcFreeCont(pMsg->pCont);
+  }
+
+  if (pHandle->pMndWrapper != NULL || pHandle->pQndWrapper != NULL) {
+    SMsgHead *pHead = pMsg->pCont;
+    int32_t   vgId = ntohl(pHead->vgId);
+    if (vgId == QNODE_HANDLE) {
+      pWrapper = pHandle->pQndWrapper;
+    } else if (vgId == MNODE_HANDLE) {
+      pWrapper = pHandle->pMndWrapper;
+    } else {
+    }
+  }
+
+  dTrace("msg:%s will be processed by %s, app:%p", TMSG_INFO(msgType), pWrapper->name, pMsg->ahandle);
+  dndProcessRpcMsg(pWrapper, pMsg, pEpSet);
+}
+
+static int32_t dndInitClient(SDnode *pDnode) {
+  STransMgmt *pMgmt = &pDnode->trans;
+
+  SRpcInit rpcInit;
+  memset(&rpcInit, 0, sizeof(rpcInit));
+  rpcInit.label = "DND";
+  rpcInit.numOfThreads = 1;
+  rpcInit.cfp = (RpcCfp)dndProcessMsg;
+  rpcInit.sessions = 1024;
+  rpcInit.connType = TAOS_CONN_CLIENT;
+  rpcInit.idleTime = tsShellActivityTimer * 1000;
+  rpcInit.user = INTERNAL_USER;
+  rpcInit.ckey = INTERNAL_CKEY;
+  rpcInit.spi = 1;
+  rpcInit.parent = pDnode;
+
+  char pass[TSDB_PASSWORD_LEN + 1] = {0};
+  taosEncryptPass_c((uint8_t *)(INTERNAL_SECRET), strlen(INTERNAL_SECRET), pass);
+  rpcInit.secret = pass;
+
+  pMgmt->clientRpc = rpcOpen(&rpcInit);
+  if (pMgmt->clientRpc == NULL) {
+    dError("failed to init dnode rpc client");
+    return -1;
+  }
+
+  dDebug("dnode rpc client is initialized");
+  return 0;
+}
+
+static void dndCleanupClient(SDnode *pDnode) {
+  STransMgmt *pMgmt = &pDnode->trans;
+  if (pMgmt->clientRpc) {
+    rpcClose(pMgmt->clientRpc);
+    pMgmt->clientRpc = NULL;
+    dDebug("dnode rpc client is closed");
+  }
+}
+
+static inline void dndSendMsgToMnodeRecv(SDnode *pDnode, SRpcMsg *pReq, SRpcMsg *pRsp) {
+  SEpSet        epSet = {0};
+  SMgmtWrapper *pWrapper = &pDnode->wrappers[DNODE];
+  dmGetMnodeEpSet(pWrapper->pMgmt, &epSet);
+  rpcSendRecv(pDnode->trans.clientRpc, &epSet, pReq, pRsp);
+}
+
+static inline int32_t dndGetHideUserAuth(SDnode *pDnode, char *user, char *spi, char *encrypt, char *secret,
+                                         char *ckey) {
+  int32_t code = 0;
+  char    pass[TSDB_PASSWORD_LEN + 1] = {0};
+
+  if (strcmp(user, INTERNAL_USER) == 0) {
+    taosEncryptPass_c((uint8_t *)(INTERNAL_SECRET), strlen(INTERNAL_SECRET), pass);
+  } else if (strcmp(user, TSDB_NETTEST_USER) == 0) {
+    taosEncryptPass_c((uint8_t *)(TSDB_NETTEST_USER), strlen(TSDB_NETTEST_USER), pass);
+  } else {
+    code = -1;
+  }
+
+  if (code == 0) {
+    memcpy(secret, pass, TSDB_PASSWORD_LEN);
+    *spi = 1;
+    *encrypt = 0;
+    *ckey = 0;
+  }
+
+  return code;
+}
+
+static int32_t dndRetrieveUserAuthInfo(SDnode *pDnode, char *user, char *spi, char *encrypt, char *secret, char *ckey) {
+  if (dndGetHideUserAuth(pDnode, user, spi, encrypt, secret, ckey) == 0) {
+    dTrace("user:%s, get auth from mnode, spi:%d encrypt:%d", user, *spi, *encrypt);
+    return 0;
+  }
+
+  SAuthReq authReq = {0};
+  tstrncpy(authReq.user, user, TSDB_USER_LEN);
+  int32_t contLen = tSerializeSAuthReq(NULL, 0, &authReq);
+  void   *pReq = rpcMallocCont(contLen);
+  tSerializeSAuthReq(pReq, contLen, &authReq);
+
+  SRpcMsg rpcMsg = {.pCont = pReq, .contLen = contLen, .msgType = TDMT_MND_AUTH, .ahandle = (void *)9528};
+  SRpcMsg rpcRsp = {0};
+  dTrace("user:%s, send user auth req to other mnodes, spi:%d encrypt:%d", user, authReq.spi, authReq.encrypt);
+  dndSendMsgToMnodeRecv(pDnode, &rpcMsg, &rpcRsp);
+
+  if (rpcRsp.code != 0) {
+    terrno = rpcRsp.code;
+    dError("user:%s, failed to get user auth from other mnodes since %s", user, terrstr());
+  } else {
+    SAuthRsp authRsp = {0};
+    tDeserializeSAuthReq(rpcRsp.pCont, rpcRsp.contLen, &authRsp);
+    memcpy(secret, authRsp.secret, TSDB_PASSWORD_LEN);
+    memcpy(ckey, authRsp.ckey, TSDB_PASSWORD_LEN);
+    *spi = authRsp.spi;
+    *encrypt = authRsp.encrypt;
+    dTrace("user:%s, success to get user auth from other mnodes, spi:%d encrypt:%d", user, authRsp.spi,
+           authRsp.encrypt);
+  }
+
+  rpcFreeCont(rpcRsp.pCont);
+  return rpcRsp.code;
+}
+
+static int32_t dndInitServer(SDnode *pDnode) {
+  STransMgmt *pMgmt = &pDnode->trans;
+
+  SRpcInit rpcInit;
+  memset(&rpcInit, 0, sizeof(rpcInit));
+  rpcInit.localPort = pDnode->serverPort;
+  rpcInit.label = "DND";
+  rpcInit.numOfThreads = tsNumOfRpcThreads;
+  rpcInit.cfp = (RpcCfp)dndProcessMsg;
+  rpcInit.sessions = tsMaxShellConns;
+  rpcInit.connType = TAOS_CONN_SERVER;
+  rpcInit.idleTime = tsShellActivityTimer * 1000;
+  rpcInit.afp = (RpcAfp)dndRetrieveUserAuthInfo;
+  rpcInit.parent = pDnode;
+
+  pMgmt->serverRpc = rpcOpen(&rpcInit);
+  if (pMgmt->serverRpc == NULL) {
+    dError("failed to init dnode rpc server");
+    return -1;
+  }
+
+  dDebug("dnode rpc server is initialized");
+  return 0;
+}
+
+static void dndCleanupServer(SDnode *pDnode) {
+  STransMgmt *pMgmt = &pDnode->trans;
+  if (pMgmt->serverRpc) {
+    rpcClose(pMgmt->serverRpc);
+    pMgmt->serverRpc = NULL;
+    dDebug("dnode rpc server is closed");
+  }
+}
+
+int32_t dndInitTrans(SDnode *pDnode) {
+  if (dndInitServer(pDnode) != 0) return -1;
+  if (dndInitClient(pDnode) != 0) return -1;
+  return 0;
+}
+
+void dndCleanupTrans(SDnode *pDnode) {
+  dndCleanupServer(pDnode);
+  dndCleanupClient(pDnode);
+}
+
+int32_t dndInitMsgHandle(SDnode *pDnode) {
+  STransMgmt *pMgmt = &pDnode->trans;
+
+  for (EDndType n = 0; n < NODE_MAX; ++n) {
+    SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
+
+    for (int32_t msgIndex = 0; msgIndex < TDMT_MAX; ++msgIndex) {
+      NodeMsgFp msgFp = pWrapper->msgFps[msgIndex];
+      int8_t    vgId = pWrapper->msgVgIds[msgIndex];
+      if (msgFp == NULL) continue;
+
+      SMsgHandle *pHandle = &pMgmt->msgHandles[msgIndex];
+      if (vgId == QNODE_HANDLE) {
+        if (pHandle->pQndWrapper != NULL) {
+          dError("msg:%s has multiple process nodes", tMsgInfo[msgIndex]);
+          return -1;
+        }
+        pHandle->pQndWrapper = pWrapper;
+      } else if (vgId == MNODE_HANDLE) {
+        if (pHandle->pMndWrapper != NULL) {
+          dError("msg:%s has multiple process nodes", tMsgInfo[msgIndex]);
+          return -1;
+        }
+        pHandle->pMndWrapper = pWrapper;
+      } else {
+        if (pHandle->pWrapper != NULL) {
+          dError("msg:%s has multiple process nodes", tMsgInfo[msgIndex]);
+          return -1;
+        }
+        pHandle->pWrapper = pWrapper;
+      }
+    }
+  }
+
+  return 0;
+}
+
+static int32_t dndSendRpcReq(STransMgmt *pMgmt, const SEpSet *pEpSet, SRpcMsg *pReq) {
+  if (pMgmt->clientRpc == NULL) {
+    terrno = TSDB_CODE_DND_OFFLINE;
+    return -1;
+  }
+
+  rpcSendRequest(pMgmt->clientRpc, pEpSet, pReq, NULL);
+  return 0;
+}
+
+static void dndSendRpcRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
+  if (pRsp->code == TSDB_CODE_APP_NOT_READY) {
+    if (pWrapper->ntype == MNODE) {
+      dmSendRedirectRsp(pWrapper->pMgmt, pRsp);
+      return;
+    }
+  }
+
+  rpcSendResponse(pRsp);
+}
+
+static int32_t dndSendReq(SMgmtWrapper *pWrapper, const SEpSet *pEpSet, SRpcMsg *pReq) {
+  if (dndGetStatus(pWrapper->pDnode) != DND_STAT_RUNNING) {
+    terrno = TSDB_CODE_DND_OFFLINE;
+    dError("failed to send rpc msg since %s, handle:%p", terrstr(), pReq->handle);
+    return -1;
+  }
+
+  if (pWrapper->procType != PROC_CHILD) {
+    return dndSendRpcReq(&pWrapper->pDnode->trans, pEpSet, pReq);
+  } else {
+    char *pHead = taosMemoryMalloc(sizeof(SRpcMsg) + sizeof(SEpSet));
+    if (pHead == NULL) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      return -1;
+    }
+
+    memcpy(pHead, pReq, sizeof(SRpcMsg));
+    memcpy(pHead + sizeof(SRpcMsg), pEpSet, sizeof(SEpSet));
+    taosProcPutToParentQ(pWrapper->pProc, pHead, sizeof(SRpcMsg) + sizeof(SEpSet), pReq->pCont, pReq->contLen,
+                         PROC_REQ);
+    taosMemoryFree(pHead);
+    return 0;
+  }
+}
+
+static void dndSendRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
+  if (pWrapper->procType != PROC_CHILD) {
+    dndSendRpcRsp(pWrapper, pRsp);
+  } else {
+    taosProcPutToParentQ(pWrapper->pProc, pRsp, sizeof(SRpcMsg), pRsp->pCont, pRsp->contLen, PROC_RSP);
+  }
+}
+
+static void dndRegisterBrokenLinkArg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg) {
+  if (pWrapper->procType != PROC_CHILD) {
+    rpcRegisterBrokenLinkArg(pMsg);
+  } else {
+    taosProcPutToParentQ(pWrapper->pProc, pMsg, sizeof(SRpcMsg), pMsg->pCont, pMsg->contLen, PROC_REGIST);
+  }
+}
+
+static void dndReleaseHandle(SMgmtWrapper *pWrapper, void *handle, int8_t type) {
+  if (pWrapper->procType != PROC_CHILD) {
+    rpcReleaseHandle(handle, type);
+  } else {
+    SRpcMsg msg = {.handle = handle, .code = type};
+    taosProcPutToParentQ(pWrapper->pProc, &msg, sizeof(SRpcMsg), NULL, 0, PROC_RELEASE);
+  }
+}
+
+SMsgCb dndCreateMsgcb(SMgmtWrapper *pWrapper) {
+  SMsgCb msgCb = {
+      .pWrapper = pWrapper,
+      .sendReqFp = dndSendReq,
+      .sendRspFp = dndSendRsp,
+      .registerBrokenLinkArgFp = dndRegisterBrokenLinkArg,
+      .releaseHandleFp = dndReleaseHandle,
+  };
+  return msgCb;
+}
+
+static void dndConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
+                                 ProcFuncType ftype) {
+  SRpcMsg *pRpc = &pMsg->rpcMsg;
+  pRpc->pCont = pCont;
+  dTrace("msg:%p, get from child queue, handle:%p app:%p", pMsg, pRpc->handle, pRpc->ahandle);
+
+  NodeMsgFp msgFp = pWrapper->msgFps[TMSG_INDEX(pRpc->msgType)];
+  int32_t   code = (*msgFp)(pWrapper, pMsg);
+
+  if (code != 0) {
+    dError("msg:%p, failed to process since code:0x%04x:%s", pMsg, code & 0XFFFF, tstrerror(code));
+    if (pRpc->msgType & 1U) {
+      SRpcMsg rsp = {.handle = pRpc->handle, .ahandle = pRpc->ahandle, .code = terrno};
+      dndSendRsp(pWrapper, &rsp);
+    }
+
+    dTrace("msg:%p, is freed", pMsg);
+    taosFreeQitem(pMsg);
+    rpcFreeCont(pCont);
+  }
+}
+
+static void dndConsumeParentQueue(SMgmtWrapper *pWrapper, SRpcMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
+                                  ProcFuncType ftype) {
+  pMsg->pCont = pCont;
+  dTrace("msg:%p, get from parent queue, ftype:%d handle:%p, app:%p", pMsg, ftype, pMsg->handle, pMsg->ahandle);
+
+  switch (ftype) {
+    case PROC_REGIST:
+      rpcRegisterBrokenLinkArg(pMsg);
+      break;
+    case PROC_RELEASE:
+      taosProcRemoveHandle(pWrapper->pProc, pMsg->handle);
+      rpcReleaseHandle(pMsg->handle, (int8_t)pMsg->code);
+      rpcFreeCont(pCont);
+      break;
+    case PROC_REQ:
+      dndSendRpcReq(&pWrapper->pDnode->trans, (SEpSet *)((char *)pMsg + sizeof(SRpcMsg)), pMsg);
+      break;
+    case PROC_RSP:
+      taosProcRemoveHandle(pWrapper->pProc, pMsg->handle);
+      dndSendRpcRsp(pWrapper, pMsg);
+      break;
+    default:
+      break;
+  }
+  taosMemoryFree(pMsg);
+}
+
+SProcCfg dndGenProcCfg(SMgmtWrapper *pWrapper) {
+  SProcCfg cfg = {.childConsumeFp = (ProcConsumeFp)dndConsumeChildQueue,
+                  .childMallocHeadFp = (ProcMallocFp)taosAllocateQitem,
+                  .childFreeHeadFp = (ProcFreeFp)taosFreeQitem,
+                  .childMallocBodyFp = (ProcMallocFp)rpcMallocCont,
+                  .childFreeBodyFp = (ProcFreeFp)rpcFreeCont,
+                  .parentConsumeFp = (ProcConsumeFp)dndConsumeParentQueue,
+                  .parentMallocHeadFp = (ProcMallocFp)taosMemoryMalloc,
+                  .parentFreeHeadFp = (ProcFreeFp)taosMemoryFree,
+                  .parentMallocBodyFp = (ProcMallocFp)rpcMallocCont,
+                  .parentFreeBodyFp = (ProcFreeFp)rpcFreeCont,
+                  .shm = pWrapper->shm,
+                  .parent = pWrapper,
+                  .name = pWrapper->name};
+  return cfg;
+}
+
+void dndSendRecv(SDnode *pDnode, SEpSet *pEpSet, SRpcMsg *pReq, SRpcMsg *pRsp) {
+  rpcSendRecv(pDnode->trans.clientRpc, pEpSet, pReq, pRsp);
+}

source/dnode/mgmt/main/inc/dndInt.h

@@ -1,65 +0,0 @@
-/*
- * 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_DND_INT_H_
-#define _TD_DND_INT_H_
-
-#include "dnd.h"
-
-#include "bm.h"
-#include "dm.h"
-#include "mm.h"
-#include "qm.h"
-#include "sm.h"
-#include "vm.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-// dndInt.c
-int32_t     dndInit();
-void        dndCleanup();
-const char *dndStatStr(EDndStatus stat);
-void        dndGetStartup(SDnode *pDnode, SStartupReq *pStartup);
-void        dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pMsg);
-
-// dndMsg.c
-void dndProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg, SEpSet *pEpSet);
-
-// dndExec.c
-int32_t dndOpenNode(SMgmtWrapper *pWrapper);
-void    dndCloseNode(SMgmtWrapper *pWrapper);
-int32_t dndRun(SDnode *pDnode);
-
-// dndObj.c
-SDnode *dndCreate(const SDnodeOpt *pOption);
-void    dndClose(SDnode *pDnode);
-void    dndHandleEvent(SDnode *pDnode, EDndEvent event);
-
-// dndTransport.c
-int32_t dndInitMsgHandle(SDnode *pDnode);
-void    dndSendRpcRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp);
-
-// dndFile.c
-TdFilePtr dndCheckRunning(const char *dataDir);
-int32_t   dndReadShmFile(SDnode *pDnode);
-int32_t   dndWriteShmFile(SDnode *pDnode);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_DND_INT_H_*/

source/dnode/mgmt/main/src/dndMonitor.c

@@ -1,107 +0,0 @@
-/*
- * 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 "dndInt.h"
-
-static int32_t dndGetMonitorDiskInfo(SDnode *pDnode, SMonDiskInfo *pInfo) {
-  tstrncpy(pInfo->logdir.name, tsLogDir, sizeof(pInfo->logdir.name));
-  pInfo->logdir.size = tsLogSpace.size;
-  tstrncpy(pInfo->tempdir.name, tsTempDir, sizeof(pInfo->tempdir.name));
-  pInfo->tempdir.size = tsTempSpace.size;
-
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, VNODES);
-  if (pWrapper != NULL) {
-    vmMonitorTfsInfo(pWrapper, pInfo);
-    dndReleaseWrapper(pWrapper);
-  }
-  return 0;
-}
-
-static void dndGetMonitorBasicInfo(SDnode *pDnode, SMonBasicInfo *pInfo) {
-  pInfo->protocol = 1;
-  pInfo->dnode_id = pDnode->dnodeId;
-  pInfo->cluster_id = pDnode->clusterId;
-  tstrncpy(pInfo->dnode_ep, tsLocalEp, TSDB_EP_LEN);
-}
-
-static void dndGetMonitorDnodeInfo(SDnode *pDnode, SMonDnodeInfo *pInfo) {
-  pInfo->uptime = (taosGetTimestampMs() - pDnode->rebootTime) / (86400000.0f);
-  taosGetCpuUsage(&pInfo->cpu_engine, &pInfo->cpu_system);
-  taosGetCpuCores(&pInfo->cpu_cores);
-  taosGetProcMemory(&pInfo->mem_engine);
-  taosGetSysMemory(&pInfo->mem_system);
-  pInfo->mem_total = tsTotalMemoryKB;
-  pInfo->disk_engine = 0;
-  pInfo->disk_used = tsDataSpace.size.used;
-  pInfo->disk_total = tsDataSpace.size.total;
-  taosGetCardInfo(&pInfo->net_in, &pInfo->net_out);
-  taosGetProcIO(&pInfo->io_read, &pInfo->io_write, &pInfo->io_read_disk, &pInfo->io_write_disk);
-
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, VNODES);
-  if (pWrapper != NULL) {
-    vmMonitorVnodeReqs(pWrapper, pInfo);
-    dndReleaseWrapper(pWrapper);
-  }
-
-  pWrapper = dndAcquireWrapper(pDnode, MNODE);
-  if (pWrapper != NULL) {
-    pInfo->has_mnode = pWrapper->required;
-    dndReleaseWrapper(pWrapper);
-  }
-}
-
-void dndSendMonitorReport(SDnode *pDnode) {
-  if (!tsEnableMonitor || tsMonitorFqdn[0] == 0 || tsMonitorPort == 0) return;
-  dTrace("send monitor report to %s:%u", tsMonitorFqdn, tsMonitorPort);
-
-  SMonInfo *pMonitor = monCreateMonitorInfo();
-  if (pMonitor == NULL) return;
-
-  SMonBasicInfo basicInfo = {0};
-  dndGetMonitorBasicInfo(pDnode, &basicInfo);
-  monSetBasicInfo(pMonitor, &basicInfo);
-
-  SMonClusterInfo clusterInfo = {0};
-  SMonVgroupInfo  vgroupInfo = {0};
-  SMonGrantInfo   grantInfo = {0};
-
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, MNODE);
-  if (pWrapper != NULL) {
-    if (mmMonitorMnodeInfo(pWrapper, &clusterInfo, &vgroupInfo, &grantInfo) == 0) {
-      monSetClusterInfo(pMonitor, &clusterInfo);
-      monSetVgroupInfo(pMonitor, &vgroupInfo);
-      monSetGrantInfo(pMonitor, &grantInfo);
-    }
-    dndReleaseWrapper(pWrapper);
-  }
-
-  SMonDnodeInfo dnodeInfo = {0};
-  dndGetMonitorDnodeInfo(pDnode, &dnodeInfo);
-  monSetDnodeInfo(pMonitor, &dnodeInfo);
-
-  SMonDiskInfo diskInfo = {0};
-  if (dndGetMonitorDiskInfo(pDnode, &diskInfo) == 0) {
-    monSetDiskInfo(pMonitor, &diskInfo);
-  }
-
-  taosArrayDestroy(clusterInfo.dnodes);
-  taosArrayDestroy(clusterInfo.mnodes);
-  taosArrayDestroy(vgroupInfo.vgroups);
-  taosArrayDestroy(diskInfo.datadirs);
-
-  monSendReport(pMonitor);
-  monCleanupMonitorInfo(pMonitor);
-}

source/dnode/mgmt/main/src/dndMsg.c

@@ -1,174 +0,0 @@
-/*
- * 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 "dndInt.h"
-
-static void dndUpdateMnodeEpSet(SDnode *pDnode, SEpSet *pEpSet) {
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, DNODE);
-  if (pWrapper != NULL) {
-    dmUpdateMnodeEpSet(pWrapper->pMgmt, pEpSet);
-    dndReleaseWrapper(pWrapper);
-  }
-}
-
-static inline NodeMsgFp dndGetMsgFp(SMgmtWrapper *pWrapper, SRpcMsg *pRpc) {
-  NodeMsgFp msgFp = pWrapper->msgFps[TMSG_INDEX(pRpc->msgType)];
-  if (msgFp == NULL) {
-    terrno = TSDB_CODE_MSG_NOT_PROCESSED;
-  }
-
-  return msgFp;
-}
-
-static inline int32_t dndBuildMsg(SNodeMsg *pMsg, SRpcMsg *pRpc) {
-  SRpcConnInfo connInfo = {0};
-  if ((pRpc->msgType & 1U) && rpcGetConnInfo(pRpc->handle, &connInfo) != 0) {
-    terrno = TSDB_CODE_MND_NO_USER_FROM_CONN;
-    dError("failed to build msg since %s, app:%p RPC:%p", terrstr(), pRpc->ahandle, pRpc->handle);
-    return -1;
-  }
-
-  memcpy(pMsg->user, connInfo.user, TSDB_USER_LEN);
-  pMsg->clientIp = connInfo.clientIp;
-  pMsg->clientPort = connInfo.clientPort;
-  memcpy(&pMsg->rpcMsg, pRpc, sizeof(SRpcMsg));
-  return 0;
-}
-
-void dndProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSet) {
-  int32_t   code = -1;
-  SNodeMsg *pMsg = NULL;
-  NodeMsgFp msgFp = NULL;
-
-  if (pEpSet && pEpSet->numOfEps > 0 && pRpc->msgType == TDMT_MND_STATUS_RSP) {
-    dndUpdateMnodeEpSet(pWrapper->pDnode, pEpSet);
-  }
-
-  if (dndMarkWrapper(pWrapper) != 0) goto _OVER;
-  if ((msgFp = dndGetMsgFp(pWrapper, pRpc)) == NULL) goto _OVER;
-  if ((pMsg = taosAllocateQitem(sizeof(SNodeMsg))) == NULL) goto _OVER;
-  if (dndBuildMsg(pMsg, pRpc) != 0) goto _OVER;
-
-  if (pWrapper->procType == PROC_SINGLE) {
-    dTrace("msg:%p, is created, handle:%p app:%p user:%s", pMsg, pRpc->handle, pRpc->ahandle, pMsg->user);
-    code = (*msgFp)(pWrapper, pMsg);
-  } else if (pWrapper->procType == PROC_PARENT) {
-    dTrace("msg:%p, is created and will put into child queue, handle:%p app:%p user:%s", pMsg, pRpc->handle,
-           pRpc->ahandle, pMsg->user);
-    code = taosProcPutToChildQ(pWrapper->pProc, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen, PROC_REQ);
-  } else {
-    dTrace("msg:%p, should not processed in child process, handle:%p app:%p user:%s", pMsg, pRpc->handle, pRpc->ahandle,
-           pMsg->user);
-    ASSERT(1);
-  }
-
-_OVER:
-  if (code == 0) {
-    if (pWrapper->procType == PROC_PARENT) {
-      dTrace("msg:%p, is freed in parent process", pMsg);
-      taosFreeQitem(pMsg);
-      rpcFreeCont(pRpc->pCont);
-    }
-  } else {
-    dError("msg:%p, failed to process since 0x%04x:%s", pMsg, code & 0XFFFF, terrstr());
-    if (pRpc->msgType & 1U) {
-      SRpcMsg rsp = {.handle = pRpc->handle, .ahandle = pRpc->ahandle, .code = terrno};
-      dndSendRsp(pWrapper, &rsp);
-    }
-    dTrace("msg:%p, is freed", pMsg);
-    taosFreeQitem(pMsg);
-    rpcFreeCont(pRpc->pCont);
-  }
-
-  dndReleaseWrapper(pWrapper);
-}
-
-static int32_t dndProcessCreateNodeMsg(SDnode *pDnode, ENodeType ntype, SNodeMsg *pMsg) {
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, ntype);
-  if (pWrapper != NULL) {
-    dndReleaseWrapper(pWrapper);
-    terrno = TSDB_CODE_NODE_ALREADY_DEPLOYED;
-    dError("failed to create node since %s", terrstr());
-    return -1;
-  }
-
-  pWrapper = &pDnode->wrappers[ntype];
-
-  if (taosMkDir(pWrapper->path) != 0) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    dError("failed to create dir:%s since %s", pWrapper->path, terrstr());
-    return -1;
-  }
-
-  int32_t code = (*pWrapper->fp.createMsgFp)(pWrapper, pMsg);
-  if (code != 0) {
-    dError("node:%s, failed to open since %s", pWrapper->name, terrstr());
-  } else {
-    dDebug("node:%s, has been opened", pWrapper->name);
-    pWrapper->deployed = true;
-  }
-
-  return code;
-}
-
-static int32_t dndProcessDropNodeMsg(SDnode *pDnode, ENodeType ntype, SNodeMsg *pMsg) {
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, ntype);
-  if (pWrapper == NULL) {
-    terrno = TSDB_CODE_NODE_NOT_DEPLOYED;
-    dError("failed to drop node since %s", terrstr());
-    return -1;
-  }
-
-  taosWLockLatch(&pWrapper->latch);
-  pWrapper->deployed = false;
-
-  int32_t code = (*pWrapper->fp.dropMsgFp)(pWrapper, pMsg);
-  if (code != 0) {
-    pWrapper->deployed = true;
-    dError("node:%s, failed to drop since %s", pWrapper->name, terrstr());
-  } else {
-    pWrapper->deployed = false;
-    dDebug("node:%s, has been dropped", pWrapper->name);
-  }
-
-  taosWUnLockLatch(&pWrapper->latch);
-  dndReleaseWrapper(pWrapper);
-  return code;
-}
-
-int32_t dndProcessNodeMsg(SDnode *pDnode, SNodeMsg *pMsg) {
-  switch (pMsg->rpcMsg.msgType) {
-    case TDMT_DND_CREATE_MNODE:
-      return dndProcessCreateNodeMsg(pDnode, MNODE, pMsg);
-    case TDMT_DND_DROP_MNODE:
-      return dndProcessDropNodeMsg(pDnode, MNODE, pMsg);
-    case TDMT_DND_CREATE_QNODE:
-      return dndProcessCreateNodeMsg(pDnode, QNODE, pMsg);
-    case TDMT_DND_DROP_QNODE:
-      return dndProcessDropNodeMsg(pDnode, QNODE, pMsg);
-    case TDMT_DND_CREATE_SNODE:
-      return dndProcessCreateNodeMsg(pDnode, SNODE, pMsg);
-    case TDMT_DND_DROP_SNODE:
-      return dndProcessDropNodeMsg(pDnode, SNODE, pMsg);
-    case TDMT_DND_CREATE_BNODE:
-      return dndProcessCreateNodeMsg(pDnode, BNODE, pMsg);
-    case TDMT_DND_DROP_BNODE:
-      return dndProcessDropNodeMsg(pDnode, BNODE, pMsg);
-    default:
-      terrno = TSDB_CODE_MSG_NOT_PROCESSED;
-      return -1;
-  }
-}

source/dnode/mgmt/main/src/dndStr.c

@@ -1,64 +0,0 @@
-/*
- * 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 "dndInt.h"
-
-const char *dndStatStr(EDndStatus status) {
-  switch (status) {
-    case DND_STAT_INIT:
-      return "init";
-    case DND_STAT_RUNNING:
-      return "running";
-    case DND_STAT_STOPPED:
-      return "stopped";
-    default:
-      return "unknown";
-  }
-}
-
-const char *dndNodeLogStr(ENodeType ntype) {
-  switch (ntype) {
-    case VNODES:
-      return "vnode";
-    case QNODE:
-      return "qnode";
-    case SNODE:
-      return "snode";
-    case MNODE:
-      return "mnode";
-    case BNODE:
-      return "bnode";
-    default:
-      return "taosd";
-  }
-}
-
-const char *dndNodeProcStr(ENodeType ntype) {
-  switch (ntype) {
-    case VNODES:
-      return "taosv";
-    case QNODE:
-      return "taosq";
-    case SNODE:
-      return "taoss";
-    case MNODE:
-      return "taosm";
-    case BNODE:
-      return "taosb";
-    default:
-      return "taosd";
-  }
-}

source/dnode/mgmt/main/src/dndTransport.c

@@ -1,406 +0,0 @@
-/*
- * 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 "dndInt.h"
-
-#define INTERNAL_USER   "_dnd"
-#define INTERNAL_CKEY   "_key"
-#define INTERNAL_SECRET "_pwd"
-
-static inline void dndProcessQMVnodeRpcMsg(SMsgHandle *pHandle, SRpcMsg *pMsg, SEpSet *pEpSet) {
-  SMsgHead *pHead = pMsg->pCont;
-  int32_t   vgId = htonl(pHead->vgId);
-
-  SMgmtWrapper *pWrapper = pHandle->pWrapper;
-  if (vgId == QND_VGID) {
-    pWrapper = pHandle->pQndWrapper;
-  } else if (vgId == MND_VGID) {
-    pWrapper = pHandle->pMndWrapper;
-  }
-
-  dTrace("msg:%s will be processed by %s, handle:%p app:%p vgId:%d", TMSG_INFO(pMsg->msgType), pWrapper->name,
-         pMsg->handle, pMsg->ahandle, vgId);
-  dndProcessRpcMsg(pWrapper, pMsg, pEpSet);
-}
-
-static void dndProcessResponse(void *parent, SRpcMsg *pRsp, SEpSet *pEpSet) {
-  SDnode     *pDnode = parent;
-  STransMgmt *pMgmt = &pDnode->trans;
-  tmsg_t      msgType = pRsp->msgType;
-
-  if (dndGetStatus(pDnode) != DND_STAT_RUNNING) {
-    dTrace("rsp:%s ignored since dnode not running, handle:%p app:%p", TMSG_INFO(msgType), pRsp->handle, pRsp->ahandle);
-    rpcFreeCont(pRsp->pCont);
-    return;
-  }
-
-  SMsgHandle *pHandle = &pMgmt->msgHandles[TMSG_INDEX(msgType)];
-  if (pHandle->pWrapper != NULL) {
-    if (pHandle->pMndWrapper == NULL && pHandle->pQndWrapper == NULL) {
-      dTrace("rsp:%s will be processed by %s, handle:%p app:%p code:0x%04x:%s", TMSG_INFO(msgType),
-             pHandle->pWrapper->name, pRsp->handle, pRsp->ahandle, pRsp->code & 0XFFFF, tstrerror(pRsp->code));
-      dndProcessRpcMsg(pHandle->pWrapper, pRsp, pEpSet);
-    } else {
-      dndProcessQMVnodeRpcMsg(pHandle, pRsp, pEpSet);
-    }
-  } else {
-    dError("rsp:%s not processed since no handle, handle:%p app:%p", TMSG_INFO(msgType), pRsp->handle, pRsp->ahandle);
-    rpcFreeCont(pRsp->pCont);
-  }
-}
-
-int32_t dndInitClient(SDnode *pDnode) {
-  STransMgmt *pMgmt = &pDnode->trans;
-
-  SRpcInit rpcInit;
-  memset(&rpcInit, 0, sizeof(rpcInit));
-  rpcInit.label = "DND";
-  rpcInit.numOfThreads = 1;
-  rpcInit.cfp = dndProcessResponse;
-  rpcInit.sessions = 1024;
-  rpcInit.connType = TAOS_CONN_CLIENT;
-  rpcInit.idleTime = tsShellActivityTimer * 1000;
-  rpcInit.user = INTERNAL_USER;
-  rpcInit.ckey = INTERNAL_CKEY;
-  rpcInit.spi = 1;
-  rpcInit.parent = pDnode;
-
-  char pass[TSDB_PASSWORD_LEN + 1] = {0};
-  taosEncryptPass_c((uint8_t *)(INTERNAL_SECRET), strlen(INTERNAL_SECRET), pass);
-  rpcInit.secret = pass;
-
-  pMgmt->clientRpc = rpcOpen(&rpcInit);
-  if (pMgmt->clientRpc == NULL) {
-    dError("failed to init dnode rpc client");
-    return -1;
-  }
-
-  dDebug("dnode rpc client is initialized");
-  return 0;
-}
-
-void dndCleanupClient(SDnode *pDnode) {
-  STransMgmt *pMgmt = &pDnode->trans;
-  if (pMgmt->clientRpc) {
-    rpcClose(pMgmt->clientRpc);
-    pMgmt->clientRpc = NULL;
-    dDebug("dnode rpc client is closed");
-  }
-}
-
-static void dndProcessRequest(void *param, SRpcMsg *pReq, SEpSet *pEpSet) {
-  SDnode     *pDnode = param;
-  STransMgmt *pMgmt = &pDnode->trans;
-  tmsg_t      msgType = pReq->msgType;
-
-  if (msgType == TDMT_DND_NETWORK_TEST) {
-    dTrace("network test req will be processed, handle:%p, app:%p", pReq->handle, pReq->ahandle);
-    dndProcessStartupReq(pDnode, pReq);
-    return;
-  }
-
-  if (dndGetStatus(pDnode) != DND_STAT_RUNNING) {
-    dError("req:%s ignored since dnode not running, handle:%p app:%p", TMSG_INFO(msgType), pReq->handle, pReq->ahandle);
-    SRpcMsg rspMsg = {.handle = pReq->handle, .code = TSDB_CODE_APP_NOT_READY, .ahandle = pReq->ahandle};
-    rpcSendResponse(&rspMsg);
-    rpcFreeCont(pReq->pCont);
-    return;
-  }
-
-  if (pReq->pCont == NULL) {
-    dTrace("req:%s not processed since its empty, handle:%p app:%p", TMSG_INFO(msgType), pReq->handle, pReq->ahandle);
-    SRpcMsg rspMsg = {.handle = pReq->handle, .code = TSDB_CODE_DND_INVALID_MSG_LEN, .ahandle = pReq->ahandle};
-    rpcSendResponse(&rspMsg);
-    return;
-  }
-
-  SMsgHandle *pHandle = &pMgmt->msgHandles[TMSG_INDEX(msgType)];
-  if (pHandle->pWrapper != NULL) {
-    if (pHandle->pMndWrapper == NULL && pHandle->pQndWrapper == NULL) {
-      dTrace("req:%s will be processed by %s, handle:%p app:%p", TMSG_INFO(msgType), pHandle->pWrapper->name,
-             pReq->handle, pReq->ahandle);
-      dndProcessRpcMsg(pHandle->pWrapper, pReq, pEpSet);
-    } else {
-      dndProcessQMVnodeRpcMsg(pHandle, pReq, pEpSet);
-    }
-  } else {
-    dError("req:%s not processed since no handle, handle:%p app:%p", TMSG_INFO(msgType), pReq->handle, pReq->ahandle);
-    SRpcMsg rspMsg = {.handle = pReq->handle, .code = TSDB_CODE_MSG_NOT_PROCESSED, .ahandle = pReq->ahandle};
-    rpcSendResponse(&rspMsg);
-    rpcFreeCont(pReq->pCont);
-  }
-}
-
-static void dndSendMsgToMnodeRecv(SDnode *pDnode, SRpcMsg *pRpcMsg, SRpcMsg *pRpcRsp) {
-  STransMgmt *pMgmt = &pDnode->trans;
-  SEpSet      epSet = {0};
-
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, DNODE);
-  if (pWrapper != NULL) {
-    dmGetMnodeEpSet(pWrapper->pMgmt, &epSet);
-    dndReleaseWrapper(pWrapper);
-  }
-
-  rpcSendRecv(pMgmt->clientRpc, &epSet, pRpcMsg, pRpcRsp);
-}
-
-static int32_t dndGetHideUserAuth(SDnode *pDnode, char *user, char *spi, char *encrypt, char *secret, char *ckey) {
-  int32_t code = 0;
-  char    pass[TSDB_PASSWORD_LEN + 1] = {0};
-
-  if (strcmp(user, INTERNAL_USER) == 0) {
-    taosEncryptPass_c((uint8_t *)(INTERNAL_SECRET), strlen(INTERNAL_SECRET), pass);
-  } else if (strcmp(user, TSDB_NETTEST_USER) == 0) {
-    taosEncryptPass_c((uint8_t *)(TSDB_NETTEST_USER), strlen(TSDB_NETTEST_USER), pass);
-  } else {
-    code = -1;
-  }
-
-  if (code == 0) {
-    memcpy(secret, pass, TSDB_PASSWORD_LEN);
-    *spi = 1;
-    *encrypt = 0;
-    *ckey = 0;
-  }
-
-  return code;
-}
-
-static int32_t dndRetrieveUserAuthInfo(void *parent, char *user, char *spi, char *encrypt, char *secret, char *ckey) {
-  SDnode *pDnode = parent;
-
-  if (dndGetHideUserAuth(parent, user, spi, encrypt, secret, ckey) == 0) {
-    dTrace("user:%s, get auth from mnode, spi:%d encrypt:%d", user, *spi, *encrypt);
-    return 0;
-  }
-
-  SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, MNODE);
-  if (pWrapper != NULL) {
-    if (mmGetUserAuth(pWrapper, user, spi, encrypt, secret, ckey) == 0) {
-      dndReleaseWrapper(pWrapper);
-      dTrace("user:%s, get auth from mnode, spi:%d encrypt:%d", user, *spi, *encrypt);
-      return 0;
-    }
-    dndReleaseWrapper(pWrapper);
-  }
-
-  if (terrno != TSDB_CODE_APP_NOT_READY) {
-    dTrace("failed to get user auth from mnode since %s", terrstr());
-    return -1;
-  }
-
-  SAuthReq authReq = {0};
-  tstrncpy(authReq.user, user, TSDB_USER_LEN);
-  int32_t contLen = tSerializeSAuthReq(NULL, 0, &authReq);
-  void   *pReq = rpcMallocCont(contLen);
-  tSerializeSAuthReq(pReq, contLen, &authReq);
-
-  SRpcMsg rpcMsg = {.pCont = pReq, .contLen = contLen, .msgType = TDMT_MND_AUTH, .ahandle = (void *)9528};
-  SRpcMsg rpcRsp = {0};
-  dTrace("user:%s, send user auth req to other mnodes, spi:%d encrypt:%d", user, authReq.spi, authReq.encrypt);
-  dndSendMsgToMnodeRecv(pDnode, &rpcMsg, &rpcRsp);
-
-  if (rpcRsp.code != 0) {
-    terrno = rpcRsp.code;
-    dError("user:%s, failed to get user auth from other mnodes since %s", user, terrstr());
-  } else {
-    SAuthRsp authRsp = {0};
-    tDeserializeSAuthReq(rpcRsp.pCont, rpcRsp.contLen, &authRsp);
-    memcpy(secret, authRsp.secret, TSDB_PASSWORD_LEN);
-    memcpy(ckey, authRsp.ckey, TSDB_PASSWORD_LEN);
-    *spi = authRsp.spi;
-    *encrypt = authRsp.encrypt;
-    dTrace("user:%s, success to get user auth from other mnodes, spi:%d encrypt:%d", user, authRsp.spi,
-           authRsp.encrypt);
-  }
-
-  rpcFreeCont(rpcRsp.pCont);
-  return rpcRsp.code;
-}
-
-int32_t dndInitServer(SDnode *pDnode) {
-  STransMgmt *pMgmt = &pDnode->trans;
-
-  int32_t numOfThreads = (int32_t)((tsNumOfCores * tsNumOfThreadsPerCore) / 2.0);
-  if (numOfThreads < 1) {
-    numOfThreads = 1;
-  }
-
-  SRpcInit rpcInit;
-  memset(&rpcInit, 0, sizeof(rpcInit));
-  rpcInit.localPort = pDnode->serverPort;
-  rpcInit.label = "DND";
-  rpcInit.numOfThreads = numOfThreads;
-  rpcInit.cfp = dndProcessRequest;
-  rpcInit.sessions = tsMaxShellConns;
-  rpcInit.connType = TAOS_CONN_SERVER;
-  rpcInit.idleTime = tsShellActivityTimer * 1000;
-  rpcInit.afp = dndRetrieveUserAuthInfo;
-  rpcInit.parent = pDnode;
-
-  pMgmt->serverRpc = rpcOpen(&rpcInit);
-  if (pMgmt->serverRpc == NULL) {
-    dError("failed to init dnode rpc server");
-    return -1;
-  }
-
-  dDebug("dnode rpc server is initialized");
-  return 0;
-}
-
-void dndCleanupServer(SDnode *pDnode) {
-  STransMgmt *pMgmt = &pDnode->trans;
-  if (pMgmt->serverRpc) {
-    rpcClose(pMgmt->serverRpc);
-    pMgmt->serverRpc = NULL;
-    dDebug("dnode rpc server is closed");
-  }
-}
-
-int32_t dndInitMsgHandle(SDnode *pDnode) {
-  STransMgmt *pMgmt = &pDnode->trans;
-
-  for (ENodeType n = 0; n < NODE_MAX; ++n) {
-    SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
-
-    for (int32_t msgIndex = 0; msgIndex < TDMT_MAX; ++msgIndex) {
-      NodeMsgFp msgFp = pWrapper->msgFps[msgIndex];
-      int32_t   vgId = pWrapper->msgVgIds[msgIndex];
-      if (msgFp == NULL) continue;
-
-      // dTrace("msg:%s will be processed by %s, vgId:%d", tMsgInfo[msgIndex], pWrapper->name, vgId);
-
-      SMsgHandle *pHandle = &pMgmt->msgHandles[msgIndex];
-      if (vgId == QND_VGID) {
-        if (pHandle->pQndWrapper != NULL) {
-          dError("msg:%s has multiple process nodes", tMsgInfo[msgIndex]);
-          return -1;
-        }
-        pHandle->pQndWrapper = pWrapper;
-      } else if (vgId == MND_VGID) {
-        if (pHandle->pMndWrapper != NULL) {
-          dError("msg:%s has multiple process nodes", tMsgInfo[msgIndex]);
-          return -1;
-        }
-        pHandle->pMndWrapper = pWrapper;
-      } else {
-        if (pHandle->pWrapper != NULL) {
-          dError("msg:%s has multiple process nodes", tMsgInfo[msgIndex]);
-          return -1;
-        }
-        pHandle->pWrapper = pWrapper;
-      }
-    }
-  }
-
-  return 0;
-}
-
-static int32_t dndSendRpcReq(STransMgmt *pMgmt, const SEpSet *pEpSet, SRpcMsg *pReq) {
-  if (pMgmt->clientRpc == NULL) {
-    terrno = TSDB_CODE_DND_OFFLINE;
-    return -1;
-  }
-
-  rpcSendRequest(pMgmt->clientRpc, pEpSet, pReq, NULL);
-  return 0;
-}
-
-int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, const SEpSet *pEpSet, SRpcMsg *pReq) {
-  if (pWrapper->procType != PROC_CHILD) {
-    SDnode *pDnode = pWrapper->pDnode;
-    if (dndGetStatus(pDnode) != DND_STAT_RUNNING) {
-      terrno = TSDB_CODE_DND_OFFLINE;
-      dError("failed to send rpc msg since %s, handle:%p", terrstr(), pReq->handle);
-      return -1;
-    }
-    return dndSendRpcReq(&pDnode->trans, pEpSet, pReq);
-  } else {
-    while (taosProcPutToParentQ(pWrapper->pProc, pReq, sizeof(SRpcMsg), pReq->pCont, pReq->contLen, PROC_REQ) != 0) {
-      taosMsleep(1);
-    }
-  }
-}
-
-int32_t dndSendReqToMnode(SMgmtWrapper *pWrapper, SRpcMsg *pReq) {
-  SDnode     *pDnode = pWrapper->pDnode;
-  STransMgmt *pTrans = &pDnode->trans;
-  SEpSet      epSet = {0};
-
-  SMgmtWrapper *pWrapper2 = dndAcquireWrapper(pDnode, DNODE);
-  if (pWrapper2 != NULL) {
-    dmGetMnodeEpSet(pWrapper2->pMgmt, &epSet);
-    dndReleaseWrapper(pWrapper2);
-  }
-  return dndSendRpcReq(pTrans, &epSet, pReq);
-}
-
-void dndSendRpcRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
-  if (pRsp->code == TSDB_CODE_APP_NOT_READY) {
-    SMgmtWrapper *pDnodeWrapper = dndAcquireWrapper(pWrapper->pDnode, DNODE);
-    if (pDnodeWrapper != NULL) {
-      dmSendRedirectRsp(pDnodeWrapper->pMgmt, pRsp);
-      dndReleaseWrapper(pDnodeWrapper);
-    } else {
-      rpcSendResponse(pRsp);
-    }
-  } else {
-    rpcSendResponse(pRsp);
-  }
-}
-
-void dndSendRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
-  if (pWrapper->procType != PROC_CHILD) {
-    dndSendRpcRsp(pWrapper, pRsp);
-  } else {
-    while (taosProcPutToParentQ(pWrapper->pProc, pRsp, sizeof(SRpcMsg), pRsp->pCont, pRsp->contLen, PROC_RSP) != 0) {
-      taosMsleep(1);
-    }
-  }
-}
-
-void dndRegisterBrokenLinkArg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg) {
-  if (pWrapper->procType != PROC_CHILD) {
-    rpcRegisterBrokenLinkArg(pMsg);
-  } else {
-    while (taosProcPutToParentQ(pWrapper->pProc, pMsg, sizeof(SRpcMsg), pMsg->pCont, pMsg->contLen, PROC_REG) != 0) {
-      taosMsleep(1);
-    }
-  }
-}
-
-static void dndReleaseHandle(SMgmtWrapper *pWrapper, void *handle, int8_t type) {
-  if (pWrapper->procType != PROC_CHILD) {
-    rpcReleaseHandle(handle, type);
-  } else {
-    SRpcMsg msg = {.handle = handle, .code = type};
-    while (taosProcPutToParentQ(pWrapper->pProc, &msg, sizeof(SRpcMsg), NULL, 0, PROC_RELEASE) != 0) {
-      taosMsleep(1);
-    }
-  }
-}
-
-SMsgCb dndCreateMsgcb(SMgmtWrapper *pWrapper) {
-  SMsgCb msgCb = {
-      .pWrapper = pWrapper,
-      .registerBrokenLinkArgFp = dndRegisterBrokenLinkArg,
-      .releaseHandleFp = dndReleaseHandle,
-      .sendMnodeReqFp = dndSendReqToMnode,
-      .sendReqFp = dndSendReqToDnode,
-      .sendRspFp = dndSendRsp,
-  };
-  return msgCb;
-}