Merge branch '3.0' into feature/tq

CMakeLists.txt

@@ -45,6 +45,16 @@ if(${BUILD_WITH_ROCKSDB})
     add_definitions(-DUSE_ROCKSDB)
 endif(${BUILD_WITH_ROCKSDB})
 
+## bdb
+if(${BUILD_WITH_BDB})
+    cat("${CMAKE_SUPPORT_DIR}/bdb_CMakeLists.txt.in" ${DEPS_TMP_FILE})
+endif(${BUILD_WITH_DBD})
+
+## sqlite
+if(${BUILD_WITH_SQLITE})
+    cat("${CMAKE_SUPPORT_DIR}/sqlite_CMakeLists.txt.in" ${DEPS_TMP_FILE})
+endif(${BUILD_WITH_SQLITE})
+
 ## lucene
 if(${BUILD_WITH_LUCENE})
     cat("${CMAKE_SUPPORT_DIR}/lucene_CMakeLists.txt.in" ${DEPS_TMP_FILE})

cmake/bdb_CMakeLists.txt.in

@@ -0,0 +1,13 @@
+
+# bdb
+ExternalProject_Add(bdb
+        GIT_REPOSITORY https://github.com/berkeleydb/libdb.git
+        GIT_TAG v5.3.28
+        SOURCE_DIR "${CMAKE_SOURCE_DIR}/deps/bdb"
+        BINARY_DIR "${CMAKE_SOURCE_DIR}/deps/bdb"
+        #BUILD_IN_SOURCE TRUE
+        CONFIGURE_COMMAND "./dist/configure"
+        BUILD_COMMAND "$(MAKE)"
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+)

cmake/cmake.options

@@ -19,6 +19,18 @@ option(
     ON
 )
 
+option(
+    BUILD_WITH_SQLITE
+    "If build with sqlite" 
+    ON
+)
+
+option(
+    BUILD_WITH_BDB 
+    "If build with BerkleyDB" 
+    ON
+)
+
 option(
     BUILD_WITH_LUCENE 
     "If build with lucene" 
@@ -34,7 +46,7 @@ option(
 option(
     BUILD_DEPENDENCY_TESTS 
     "If build dependency tests" 
-    OFF
+    ON
 )
 
 option(

cmake/sqlite_CMakeLists.txt.in

@@ -0,0 +1,13 @@
+
+# sqlite
+ExternalProject_Add(sqlite
+        GIT_REPOSITORY https://github.com/sqlite/sqlite.git
+        GIT_TAG version-3.36.0
+        SOURCE_DIR "${CMAKE_SOURCE_DIR}/deps/sqlite"
+        BINARY_DIR "${CMAKE_SOURCE_DIR}/deps/sqlite"
+        #BUILD_IN_SOURCE TRUE
+        CONFIGURE_COMMAND "./configure"
+        BUILD_COMMAND "$(MAKE)"
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+)

deps/CMakeLists.txt

@@ -80,6 +80,33 @@ if(${BUILD_WITH_NURAFT})
     add_subdirectory(nuraft)
 endif(${BUILD_WITH_NURAFT})
 
+# BDB
+if(${BUILD_WITH_BDB})
+    add_library(bdb STATIC IMPORTED)
+    set_target_properties(bdb PROPERTIES
+        IMPORTED_LOCATION "${CMAKE_CURRENT_SOURCE_DIR}/bdb/libdb.a"
+        INTERFACE_INCLUDE_DIRECTORIES "${CMAKE_CURRENT_SOURCE_DIR}/bdb"
+    )
+    target_link_libraries(bdb 
+        INTERFACE pthread 
+    )
+endif(${BUILD_WITH_BDB})
+
+# SQLite
+if(${BUILD_WITH_SQLITE})
+    add_library(sqlite STATIC IMPORTED)
+    set_target_properties(sqlite PROPERTIES
+        IMPORTED_LOCATION "${CMAKE_CURRENT_SOURCE_DIR}/sqlite/.libs/libsqlite3.a"
+        INTERFACE_INCLUDE_DIRECTORIES "${CMAKE_CURRENT_SOURCE_DIR}/sqlite"
+    )
+    target_link_libraries(sqlite 
+        INTERFACE m 
+        INTERFACE pthread 
+        INTERFACE dl
+    )
+endif(${BUILD_WITH_SQLITE})
+
+
 
 # ================================================================================================
 # DEPENDENCY TEST

deps/test/CMakeLists.txt

@@ -6,3 +6,13 @@ endif(${BUILD_WITH_ROCKSDB})
 if(${BUILD_WITH_LUCENE})
     add_subdirectory(lucene)
 endif(${BUILD_WITH_LUCENE})
+
+if(${BUILD_WITH_BDB})
+    add_subdirectory(bdb)
+endif(${BUILD_WITH_BDB})
+
+if(${BUILD_WITH_SQLITE})
+    add_subdirectory(sqlite)
+endif(${BUILD_WITH_SQLITE})
+
+add_subdirectory(tdev)

deps/test/bdb/CMakeLists.txt

@@ -0,0 +1,7 @@
+add_executable(bdbTest "")
+target_sources(
+    bdbTest PRIVATE
+    "bdbTest.c"
+)
+
+target_link_libraries(bdbTest bdb)

deps/test/bdb/bdbTest.c

@@ -0,0 +1,28 @@
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "db.h"
+
+// refer: https://docs.oracle.com/cd/E17076_05/html/gsg/C/BerkeleyDB-Core-C-GSG.pdf
+
+int main(int argc, char const *argv[]) {
+  DB *     db;
+  int      ret;
+  uint32_t flags;
+
+  ret = db_create(&db, NULL, 0);
+  if (ret != 0) {
+    exit(1);
+  }
+
+  flags = DB_CREATE;
+
+  ret = db->open(db, NULL, "test.db", NULL, DB_BTREE, flags, 0);
+  if (ret != 0) {
+    exit(1);
+  }
+
+  db->close(db, 0);
+
+  return 0;
+}

deps/test/sqlite/CMakeLists.txt

@@ -0,0 +1,6 @@
+add_executable(sqliteTest "")
+target_sources(
+    sqliteTest PRIVATE
+    "sqliteTest.c"
+)
+target_link_libraries(sqliteTest sqlite)

deps/test/sqlite/sqliteTest.c

@@ -0,0 +1,84 @@
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "sqlite3.h"
+
+static void count_table(sqlite3 *db) {
+  int           rc;
+  char *        sql = "select * from t;";
+  sqlite3_stmt *stmt = NULL;
+  int           nrows = 0;
+
+  rc = sqlite3_prepare_v2(db, sql, -1, &stmt, NULL);
+  while (SQLITE_ROW == sqlite3_step(stmt)) {
+    nrows++;
+  }
+
+  printf("Number of rows: %d\n", nrows);
+}
+
+int main(int argc, char const *argv[]) {
+  sqlite3 *db;
+  char *   err_msg = 0;
+
+  int rc = sqlite3_open("test.db", &db);
+
+  if (rc != SQLITE_OK) {
+    fprintf(stderr, "Cannot open database: %s\n", sqlite3_errmsg(db));
+    sqlite3_close(db);
+
+    return 1;
+  }
+
+  char *sql =
+      "DROP TABLE IF EXISTS t;"
+      "CREATE TABLE t(id BIGINT);";
+
+  rc = sqlite3_exec(db, sql, 0, 0, &err_msg);
+
+  if (rc != SQLITE_OK) {
+    fprintf(stderr, "SQL error: %s\n", err_msg);
+
+    sqlite3_free(err_msg);
+    sqlite3_close(db);
+
+    return 1;
+  }
+
+  {
+    // Write a lot of data
+    int  nrows = 1000;
+    int  batch = 100;
+    char tsql[1024];
+    int  v = 0;
+
+    // sqlite3_exec(db, "PRAGMA journal_mode=WAL;", 0, 0, &err_msg);
+    sqlite3_exec(db, "PRAGMA read_uncommitted=true;", 0, 0, &err_msg);
+
+    for (int k = 0; k < nrows / batch; k++) {
+      sqlite3_exec(db, "begin;", 0, 0, &err_msg);
+
+      for (int i = 0; i < batch; i++) {
+        v++;
+        sprintf(tsql, "insert into t values (%d)", v);
+        rc = sqlite3_exec(db, tsql, 0, 0, &err_msg);
+
+        if (rc != SQLITE_OK) {
+          fprintf(stderr, "SQL error: %s\n", err_msg);
+
+          sqlite3_free(err_msg);
+          sqlite3_close(db);
+
+          return 1;
+        }
+      }
+
+      count_table(db);
+      sqlite3_exec(db, "commit;", 0, 0, &err_msg);
+    }
+  }
+
+  sqlite3_close(db);
+
+  return 0;
+}

deps/test/tdev/CMakeLists.txt

@@ -0,0 +1,4 @@
+aux_source_directory(src TDEV_SRC)
+add_executable(tdev ${TDEV_SRC})
+
+target_include_directories(tdev PUBLIC inc)

deps/test/tdev/src/main.c

@@ -0,0 +1,87 @@
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <sys/time.h>
+
+#define POINTER_SHIFT(ptr, s) ((void *)(((char *)ptr) + (s)))
+#define POINTER_DISTANCE(pa, pb) ((char *)(pb) - (char *)(pa))
+
+#define tPutA(buf, val)              \
+  ({                                 \
+    memcpy(buf, &val, sizeof(val));  \
+    POINTER_SHIFT(buf, sizeof(val)); \
+  })
+
+#define tPutB(buf, val)                         \
+  ({                                            \
+    ((uint8_t *)buf)[3] = ((val) >> 24) & 0xff; \
+    ((uint8_t *)buf)[2] = ((val) >> 16) & 0xff; \
+    ((uint8_t *)buf)[1] = ((val) >> 8) & 0xff;  \
+    ((uint8_t *)buf)[0] = (val)&0xff;           \
+    POINTER_SHIFT(buf, sizeof(val));            \
+  })
+
+#define tPutC(buf, val)              \
+  ({                                 \
+    ((uint64_t *)buf)[0] = (val);    \
+    POINTER_SHIFT(buf, sizeof(val)); \
+  })
+
+typedef enum { A, B, C } T;
+
+static void func(T t) {
+  uint64_t val = 198;
+  char     buf[1024];
+  void *   pBuf = buf;
+
+  switch (t) {
+    case A:
+      for (size_t i = 0; i < 10 * 1024l * 1024l * 1024l; i++) {
+        pBuf = tPutA(pBuf, val);
+        if (POINTER_DISTANCE(buf, pBuf) == 1024) {
+          pBuf = buf;
+        }
+      }
+      break;
+    case B:
+      for (size_t i = 0; i < 10 * 1024l * 1024l * 1024l; i++) {
+        pBuf = tPutB(pBuf, val);
+        if (POINTER_DISTANCE(buf, pBuf) == 1024) {
+          pBuf = buf;
+        }
+      }
+      break;
+    case C:
+      for (size_t i = 0; i < 10 * 1024l * 1024l * 1024l; i++) {
+        pBuf = tPutC(pBuf, val);
+        if (POINTER_DISTANCE(buf, pBuf) == 1024) {
+          pBuf = buf;
+        }
+      }
+      break;
+
+    default:
+      break;
+  }
+}
+
+static uint64_t now() {
+  struct timeval tv;
+  gettimeofday(&tv, NULL);
+
+  return tv.tv_sec * 1000000 + tv.tv_usec;
+}
+
+int main(int argc, char const *argv[]) {
+  uint64_t t1 = now();
+  func(A);
+  uint64_t t2 = now();
+  printf("A: %ld\n", t2 - t1);
+  func(B);
+  uint64_t t3 = now();
+  printf("B: %ld\n", t3 - t2);
+  func(C);
+  uint64_t t4 = now();
+  printf("C: %ld\n", t4 - t3);
+  return 0;
+}

include/common/taosmsg.h

@@ -63,6 +63,8 @@ TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_USER, "drop-user" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_DNODE, "create-dnode" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CONFIG_DNODE, "config-dnode" ) 
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_DNODE, "drop-dnode" )   
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_MNODE, "create-mnode" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_MNODE, "drop-mnode" ) 
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_DB, "create-db" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_DB, "drop-db" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_USE_DB, "use-db" )
@@ -631,7 +633,7 @@ typedef struct {
 
 typedef struct {
   int32_t statusInterval;
-  int8_t  reserved[4];
+  int32_t mnodeEqualVnodeNum;
   int64_t checkTime;                    // 1970-01-01 00:00:00.000
   char    timezone[TSDB_TIMEZONE_LEN];  // tsTimezone
   char    locale[TSDB_LOCALE_LEN];      // tsLocale
@@ -654,11 +656,14 @@ typedef struct {
 } SVnodeLoads;
 
 typedef struct SStatusMsg {
-  uint32_t    sversion;
+  int32_t     sver;
   int32_t     dnodeId;
-  int64_t     clusterId;
+  int32_t     clusterId;
   uint32_t    rebootTime;  // time stamp for last reboot
-  int32_t     numOfCores;
+  int16_t     numOfCores;
+  int16_t     numOfSupportMnodes;
+  int16_t     numOfSupportVnodes;
+  int16_t     numOfSupportQnodes;
   char        dnodeEp[TSDB_EP_LEN];
   SClusterCfg clusterCfg;
   SVnodeLoads vnodeLoads;
@@ -666,9 +671,9 @@ typedef struct SStatusMsg {
 
 typedef struct {
   int32_t dnodeId;
+  int32_t clusterId;
   int8_t  dropped;
-  char    reserved[3];
-  int64_t clusterId;
+  char    reserved[7];
 } SDnodeCfg;
 
 typedef struct {

include/common/trow.h

@@ -17,53 +17,108 @@
 #define _TD_COMMON_ROW_H_
 
 #include "os.h"
+#include "tbuffer.h"
+#include "tdataformat.h"
+#include "tdef.h"
+#include "tschema.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-// types
-typedef void *                  SRow;
-typedef struct SRowBatch        SRowBatch;
-typedef struct SRowBuilder      SRowBuilder;
-typedef struct SRowBatchIter    SRowBatchIter;
-typedef struct SRowBatchBuilder SRowBatchBuilder;
+#define TD_UNDECIDED_ROW 0
+#define TD_OR_ROW 1
+#define TD_KV_ROW 2
 
-// SRow
-#define ROW_HEADER_SIZE (sizeof(uint8_t) + 2 * sizeof(uint16_t) + sizeof(uint64_t))
-#define rowType(r) (*(uint8_t *)(r))                                // row type
-#define rowLen(r) (*(uint16_t *)POINTER_SHIFT(r, sizeof(uint8_t)))  // row length
-#define rowSVer(r) \
-  (*(uint16_t *)POINTER_SHIFT(r, sizeof(uint8_t) + sizeof(uint16_t)))  // row schema version, only for SDataRow
-#define rowNCols(r) rowSVer(r)                                         // only for SKVRow
-#define rowVer(r) (*(uint64_t)POINTER_SHIFT(r, sizeof(uint8_t) + 2 * sizeof(uint16_t)))  // row version
-#define rowCopy(dest, r) memcpy((dest), r, rowLen(r))
+typedef struct {
+  // TODO
+} SOrRow;
 
-static FORCE_INLINE SRow rowDup(SRow row) {
-  SRow r = malloc(rowLen(row));
-  if (r == NULL) {
-    return NULL;
-  }
+typedef struct {
+  col_id_t cid;
+  uint32_t offset;
+} SKvRowIdx;
 
-  rowCopy(r, row);
+typedef struct {
+  uint16_t  ncols;
+  SKvRowIdx cidx[];
+} SKvRow;
 
-  return r;
-}
+typedef struct {
+  union {
+    /// union field for encode and decode
+    uint32_t info;
+    struct {
+      /// row type
+      uint32_t type : 2;
+      /// row schema version
+      uint32_t sver : 16;
+      /// is delete row
+      uint32_t del : 1;
+      /// reserved for back compatibility
+      uint32_t reserve : 13;
+    };
+  };
+  /// row total length
+  uint32_t len;
+  /// row version
+  uint64_t ver;
+  /// timestamp
+  TSKEY ts;
+  /// the inline data, maybe a tuple or a k-v tuple
+  char data[];
+} STSRow;
 
-// SRowBatch
+typedef struct {
+  uint32_t nRows;
+  char     rows[];
+} STSRowBatch;
 
-// SRowBuilder
-SRowBuilder *rowBuilderCreate();
-void         rowBuilderDestroy(SRowBuilder *);
+typedef enum {
+  /// ordinary row builder
+  TD_OR_ROW_BUILDER = 0,
+  /// kv row builder
+  TD_KV_ROW_BUILDER,
+  /// self-determined row builder
+  TD_SD_ROW_BUILDER
+} ERowBbuilderT;
 
-// SRowBatchIter
-SRowBatchIter *rowBatchIterCreate(SRowBatch *);
-void           rowBatchIterDestroy(SRowBatchIter *);
-const SRow     rowBatchIterNext(SRowBatchIter *);
+typedef struct {
+  /// row builder type
+  ERowBbuilderT type;
+  /// buffer writer
+  SBufferWriter bw;
+  /// target row
+  STSRow *pRow;
+} STSRowBuilder;
 
-// SRowBatchBuilder
-SRowBatchBuilder *rowBatchBuilderCreate();
-void              rowBatchBuilderDestroy(SRowBatchBuilder *);
+typedef struct {
+  STSchema *pSchema;
+  STSRow *    pRow;
+} STSRowReader;
+
+typedef struct {
+  uint32_t   it;
+  STSRowBatch *pRowBatch;
+} STSRowBatchIter;
+
+// STSRowBuilder
+#define trbInit(rt, allocator, endian, target, size) \
+  { .type = (rt), .bw = tbufInitWriter(allocator, endian), .pRow = (target) }
+void trbSetRowInfo(STSRowBuilder *pRB, bool del, uint16_t sver);
+void trbSetRowVersion(STSRowBuilder *pRB, uint64_t ver);
+void trbSetRowTS(STSRowBuilder *pRB, TSKEY ts);
+int  trbWriteCol(STSRowBuilder *pRB, void *pData, col_id_t cid);
+
+// STSRowReader
+#define tRowReaderInit(schema, row) \
+  { .schema = (schema), .row = (row) }
+int tRowReaderRead(STSRowReader *pRowReader, col_id_t cid, void *target, uint64_t size);
+
+// STSRowBatchIter
+#define tRowBatchIterInit(pRB) \
+  { .it = 0, .pRowBatch = (pRB) }
+const STSRow *tRowBatchIterNext(STSRowBatchIter *pRowBatchIter);
 
 #ifdef __cplusplus
 }

include/common/tschema.h

@@ -16,10 +16,65 @@
 #ifndef _TD_COMMON_SCHEMA_H_
 #define _TD_COMMON_SCHEMA_H_
 
+#include "os.h"
+#include "tarray.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+typedef uint16_t col_id_t;
+
+#if 0
+typedef struct STColumn {
+  /// column name
+  char *cname;
+  union {
+    /// for encode purpose
+    uint64_t info;
+    struct {
+      uint64_t sma : 1;
+      /// column data type
+      uint64_t type : 7;
+      /// column id
+      uint64_t cid : 16;
+      /// max bytes of the column
+      uint64_t bytes : 32;
+      /// reserved
+      uint64_t reserve : 8;
+    };
+  };
+  /// comment about the column
+  char *comment;
+} STColumn;
+
+typedef struct STSchema {
+  /// schema version
+  uint16_t sver;
+  /// number of columns
+  uint16_t ncols;
+  /// sma attributes
+  struct {
+    bool    sma;
+    SArray *smaArray;
+  };
+  /// column info
+  STColumn cols[];
+} STSchema;
+
+typedef struct {
+  uint64_t  size;
+  STSchema *pSchema;
+} STShemaBuilder;
+
+#define tSchemaBuilderInit(target, capacity) \
+  { .size = (capacity), .pSchema = (target) }
+void tSchemaBuilderSetSver(STShemaBuilder *pSchemaBuilder, uint16_t sver);
+void tSchemaBuilderSetSMA(bool sma, SArray *smaArray);
+int  tSchemaBuilderPutColumn(char *cname, bool sma, uint8_t type, col_id_t cid, uint32_t bytes, char *comment);
+
+#endif
+
 #ifdef __cplusplus
 }
 #endif

include/common/type/type.h

@@ -13,18 +13,15 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_VNODE_FILE_SYSTEM_H_
-#define _TD_VNODE_FILE_SYSTEM_H_
+#ifndef _TD_TYPE_H_
+#define _TD_TYPE_H_
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef struct {
-} SVnodeFS;
-
 #ifdef __cplusplus
 }
 #endif
 
-#endif /*_TD_VNODE_FILE_SYSTEM_H_*/
+#endif /*_TD_TYPE_H_*/

include/dnode/mgmt/dnode.h

@@ -26,95 +26,25 @@ extern "C" {
 typedef struct SDnode SDnode;
 
 typedef struct {
-  /**
-   * @brief software version of the program.
-   *
-   */
-  int32_t sver;
-
-  /**
-   * @brief num of CPU cores.
-   *
-   */
-  int32_t numOfCores;
-
-  /**
-   * @brief number of threads per CPU core.
-   *
-   */
-  float numOfThreadsPerCore;
-
-  /**
-   * @brief the proportion of total CPU cores available for query processing.
-   *
-   */
-  float ratioOfQueryCores;
-
-  /**
-   * @brief max number of connections allowed in dnode.
-   *
-   */
-  int32_t maxShellConns;
-
-  /**
-   * @brief time interval of heart beat from shell to dnode, seconds.
-   *
-   */
-  int32_t shellActivityTimer;
-
-  /**
-   * @brief time interval of dnode status reporting to mnode, seconds, for cluster only.
-   *
-   */
-  int32_t statusInterval;
-
-  /**
-   * @brief first port number for the connection (12 continuous UDP/TCP port number are used).
-   *
-   */
+  int32_t  sver;
+  int16_t  numOfCores;
+  int16_t  numOfSupportMnodes;
+  int16_t  numOfSupportVnodes;
+  int16_t  numOfSupportQnodes;
+  int32_t  statusInterval;
+  int32_t  mnodeEqualVnodeNum;
+  float    numOfThreadsPerCore;
+  float    ratioOfQueryCores;
+  int32_t  maxShellConns;
+  int32_t  shellActivityTimer;
   uint16_t serverPort;
-
-  /**
-   * @brief data file's directory.
-   *
-   */
-  char dataDir[TSDB_FILENAME_LEN];
-
-  /**
-   * @brief local endpoint.
-   *
-   */
-  char localEp[TSDB_EP_LEN];
-
-  /**
-   * @brieflocal fully qualified domain name (FQDN).
-   *
-   */
-  char localFqdn[TSDB_FQDN_LEN];
-
-  /**
-   * @brief first fully qualified domain name (FQDN) for TDengine system.
-   *
-   */
-  char firstEp[TSDB_EP_LEN];
-
-  /**
-   * @brief system time zone.
-   *
-   */
-  char timezone[TSDB_TIMEZONE_LEN];
-
-  /**
-   * @brief system locale.
-   *
-   */
-  char locale[TSDB_LOCALE_LEN];
-
-  /**
-   * @briefdefault system charset.
-   *
-   */
-  char charset[TSDB_LOCALE_LEN];
+  char     dataDir[TSDB_FILENAME_LEN];
+  char     localEp[TSDB_EP_LEN];
+  char     localFqdn[TSDB_FQDN_LEN];
+  char     firstEp[TSDB_EP_LEN];
+  char     timezone[TSDB_TIMEZONE_LEN];
+  char     locale[TSDB_LOCALE_LEN];
+  char     charset[TSDB_LOCALE_LEN];
 } SDnodeOpt;
 
 /* ------------------------ SDnode ------------------------ */

include/dnode/mnode/mnode.h

@@ -30,133 +30,36 @@ typedef void (*SendRedirectMsgFp)(SDnode *pDnode, struct SRpcMsg *rpcMsg);
 typedef int32_t (*PutMsgToMnodeQFp)(SDnode *pDnode, SMnodeMsg *pMsg);
 
 typedef struct SMnodeLoad {
-  /**
-   * @brief the number of dnodes in cluster.
-   *
-   */
   int64_t numOfDnode;
-
-  /**
-   * @brief the number of mnodes in cluster.
-   *
-   */
   int64_t numOfMnode;
-
-  /**
-   * @brief the number of vgroups in cluster.
-   *
-   */
   int64_t numOfVgroup;
-
-  /**
-   * @brief the number of databases in cluster.
-   *
-   */
   int64_t numOfDatabase;
-
-  /**
-   * @brief the number of super tables in cluster.
-   *
-   */
   int64_t numOfSuperTable;
-
-  /**
-   * @brief the number of child tables in cluster.
-   *
-   */
   int64_t numOfChildTable;
-
-  /**
-   * @brief the number of normal tables in cluster.
-   *
-   */
   int64_t numOfNormalTable;
-
-  /**
-   * @brief the number of numOfTimeseries in cluster.
-   *
-   */
   int64_t numOfColumn;
-
-  /**
-   * @brief total points written in cluster.
-   *
-   */
   int64_t totalPoints;
-
-  /**
-   * @brief total storage in cluster.
-   *
-   */
   int64_t totalStorage;
-
-  /**
-   * @brief total compressed storage in cluster.
-   *
-   */
   int64_t compStorage;
 } SMnodeLoad;
 
 typedef struct {
-  /**
-   * @brief dnodeId of this mnode.
-   *
-   */
-  int32_t dnodeId;
-
-  /**
-   * @brief clusterId of this mnode.
-   *
-   */
-  int64_t clusterId;
-
-  /**
-   * @brief replica num of this mnode.
-   *
-   */
-  int8_t replica;
-
-  /**
-   * @brief self index in the array of replicas.
-   *
-   */
-  int8_t selfIndex;
-
-  /**
-   * @brief detail replica information of this mnode.
-   *
-   */
-  SReplica replicas[TSDB_MAX_REPLICA];
-
-  /**
-   * @brief the parent dnode of this mnode.
-   *
-   */
-  SDnode *pDnode;
-
-  /**
-   * @brief put apply msg to the write queue in dnode.
-   *
-   */
-  PutMsgToMnodeQFp putMsgToApplyMsgFp;
-
-  /**
-   * @brief the callback function while send msg to dnode.
-   *
-   */
-  SendMsgToDnodeFp sendMsgToDnodeFp;
-
-  /**
-   * @brief the callback function while send msg to mnode.
-   *
-   */
-  SendMsgToMnodeFp sendMsgToMnodeFp;
-
-  /**
-   * @brief the callback function while send redirect msg to clients or peers.
-   *
-   */
+  int32_t           dnodeId;
+  int32_t           clusterId;
+  int8_t            replica;
+  int8_t            selfIndex;
+  SReplica          replicas[TSDB_MAX_REPLICA];
+  SDnode           *pDnode;
+  PutMsgToMnodeQFp  putMsgToApplyMsgFp;
+  SendMsgToDnodeFp  sendMsgToDnodeFp;
+  SendMsgToMnodeFp  sendMsgToMnodeFp;
   SendRedirectMsgFp sendRedirectMsgFp;
+  int32_t           sver;
+  int32_t           statusInterval;
+  int32_t           mnodeEqualVnodeNum;
+  char             *timezone;
+  char             *locale;
+  char             *charset;
 } SMnodeOpt;
 
 /* ------------------------ SMnode ------------------------ */

include/dnode/mnode/sdb/sdb.h

@@ -38,6 +38,15 @@ extern "C" {
     dataPos += sizeof(int32_t);                    \
   }
 
+#define SDB_GET_INT16(pData, pRow, dataPos, val)   \
+  {                                                \
+    if (sdbGetRawInt16(pRaw, dataPos, val) != 0) { \
+      sdbFreeRow(pRow);                            \
+      return NULL;                                 \
+    }                                              \
+    dataPos += sizeof(int16_t);                    \
+  }
+
 #define SDB_GET_INT8(pData, pRow, dataPos, val)   \
   {                                               \
     if (sdbGetRawInt8(pRaw, dataPos, val) != 0) { \
@@ -74,6 +83,15 @@ extern "C" {
     dataPos += sizeof(int32_t);                    \
   }
 
+#define SDB_SET_INT16(pRaw, dataPos, val)          \
+  {                                                \
+    if (sdbSetRawInt16(pRaw, dataPos, val) != 0) { \
+      sdbFreeRaw(pRaw);                            \
+      return NULL;                                 \
+    }                                              \
+    dataPos += sizeof(int16_t);                     \
+  }
+
 #define SDB_SET_INT8(pRaw, dataPos, val)          \
   {                                               \
     if (sdbSetRawInt8(pRaw, dataPos, val) != 0) { \
@@ -100,6 +118,7 @@ extern "C" {
     }                                           \
   }
 
+typedef struct SMnode  SMnode;
 typedef struct SSdbRaw SSdbRaw;
 typedef struct SSdbRow SSdbRow;
 typedef enum { SDB_KEY_BINARY = 1, SDB_KEY_INT32 = 2, SDB_KEY_INT64 = 3 } EKeyType;
@@ -130,7 +149,7 @@ typedef struct SSdb SSdb;
 typedef int32_t (*SdbInsertFp)(SSdb *pSdb, void *pObj);
 typedef int32_t (*SdbUpdateFp)(SSdb *pSdb, void *pSrcObj, void *pDstObj);
 typedef int32_t (*SdbDeleteFp)(SSdb *pSdb, void *pObj);
-typedef int32_t (*SdbDeployFp)(SSdb *pSdb);
+typedef int32_t (*SdbDeployFp)(SMnode *pMnode);
 typedef SSdbRow *(*SdbDecodeFp)(SSdbRaw *pRaw);
 typedef SSdbRaw *(*SdbEncodeFp)(void *pObj);
 
@@ -190,6 +209,12 @@ typedef struct SSdbOpt {
    *
    */
   const char *path;
+
+  /**
+   * @brief The mnode object.
+   *
+   */
+  SMnode *pMnode;
 } SSdbOpt;
 
 /**
@@ -291,12 +316,14 @@ int32_t sdbGetSize(SSdb *pSdb, ESdbType type);
 SSdbRaw *sdbAllocRaw(ESdbType type, int8_t sver, int32_t dataLen);
 void     sdbFreeRaw(SSdbRaw *pRaw);
 int32_t  sdbSetRawInt8(SSdbRaw *pRaw, int32_t dataPos, int8_t val);
+int32_t  sdbSetRawInt16(SSdbRaw *pRaw, int32_t dataPos, int16_t val);
 int32_t  sdbSetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t val);
 int32_t  sdbSetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t val);
 int32_t  sdbSetRawBinary(SSdbRaw *pRaw, int32_t dataPos, const char *pVal, int32_t valLen);
 int32_t  sdbSetRawDataLen(SSdbRaw *pRaw, int32_t dataLen);
 int32_t  sdbSetRawStatus(SSdbRaw *pRaw, ESdbStatus status);
 int32_t  sdbGetRawInt8(SSdbRaw *pRaw, int32_t dataPos, int8_t *val);
+int32_t  sdbGetRawInt16(SSdbRaw *pRaw, int32_t dataPos, int16_t *val);
 int32_t  sdbGetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t *val);
 int32_t  sdbGetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t *val);
 int32_t  sdbGetRawBinary(SSdbRaw *pRaw, int32_t dataPos, char *pVal, int32_t valLen);

include/dnode/vnode/meta/impl/metaImpl.h

@@ -1,75 +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_META_IMPL_H_
-#define _TD_META_IMPL_H_
-
-#include "os.h"
-
-#include "taosmsg.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-typedef uint64_t tb_uid_t;
-
-/* ------------------------ SMetaOptions ------------------------ */
-struct SMetaOptions {
-  size_t lruCacheSize;  // LRU cache size
-};
-
-/* ------------------------ STbOptions ------------------------ */
-#define META_NORMAL_TABLE ((uint8_t)1)
-#define META_SUPER_TABLE ((uint8_t)2)
-#define META_CHILD_TABLE ((uint8_t)3)
-
-typedef struct {
-} SSMAOptions;
-
-// super table options
-typedef struct {
-  tb_uid_t  uid;
-  STSchema* pSchema;
-  STSchema* pTagSchema;
-} SSTbOptions;
-
-// child table options
-typedef struct {
-  tb_uid_t suid;
-  SKVRow   tags;
-} SCTbOptions;
-
-// normal table options
-typedef struct {
-  STSchema* pSchame;
-} SNTbOptions;
-
-struct STbOptions {
-  uint8_t     type;
-  char*       name;
-  uint32_t    ttl;   // time to live in (SECONDS)
-  SSMAOptions bsma;  // Block-wise sma
-  union {
-    SSTbOptions stbOptions;
-    SNTbOptions ntbOptions;
-    SCTbOptions ctbOptions;
-  };
-};
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_META_IMPL_H_*/

include/dnode/vnode/meta/meta.h

@@ -16,36 +16,96 @@
 #ifndef _TD_META_H_
 #define _TD_META_H_
 
-#include "impl/metaImpl.h"
+#include "os.h"
+#include "trow.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
 // Types exported
-typedef struct SMeta        SMeta;
-typedef struct SMetaOptions SMetaOptions;
-typedef struct STbOptions   STbOptions;
+typedef uint64_t     tb_uid_t;
+typedef struct SMeta SMeta;
+
+#define META_SUPER_TABLE 0
+#define META_CHILD_TABLE 1
+#define META_NORMAL_TABLE 2
+
+typedef struct SMetaCfg {
+  /// LRU cache size
+  uint64_t lruSize;
+} SMetaCfg;
+
+typedef struct STbCfg {
+  /// name of the table
+  char *name;
+  /// time to live of the table
+  uint32_t ttl;
+  /// keep time of this table
+  uint32_t keep;
+  /// type of table
+  uint8_t type;
+  union {
+    /// super table configurations
+    struct {
+      /// super table UID
+      tb_uid_t suid;
+      /// row schema
+      STSchema *pSchema;
+      /// tag schema
+      STSchema *pTagSchema;
+    } stbCfg;
+
+    /// normal table configuration
+    struct {
+      /// row schema
+      STSchema *pSchema;
+    } ntbCfg;
+    /// child table configuration
+    struct {
+      /// super table UID
+      tb_uid_t suid;
+      SKVRow   pTag;
+    } ctbCfg;
+  };
+} STbCfg;
 
 // SMeta operations
-SMeta *metaOpen(const char *path, const SMetaOptions *pOptions);
+SMeta *metaOpen(const char *path, const SMetaCfg *pOptions);
 void   metaClose(SMeta *pMeta);
 void   metaRemove(const char *path);
-int    metaCreateTable(SMeta *pMeta, const STbOptions *pTbOptions);
+int    metaCreateTable(SMeta *pMeta, STbCfg *pTbCfg);
 int    metaDropTable(SMeta *pMeta, tb_uid_t uid);
 int    metaCommit(SMeta *pMeta);
 
 // Options
-void metaOptionsInit(SMetaOptions *pOptions);
-void metaOptionsClear(SMetaOptions *pOptions);
+void metaOptionsInit(SMetaCfg *pOptions);
+void metaOptionsClear(SMetaCfg *pOptions);
 
-// STableOpts
-#define META_TABLE_OPTS_DECLARE(name) STableOpts name = {0}
-void metaNormalTableOptsInit(STbOptions *pTbOptions, const char *name, const STSchema *pSchema);
-void metaSuperTableOptsInit(STbOptions *pTbOptions, const char *name, tb_uid_t uid, const STSchema *pSchema,
-                            const STSchema *pTagSchema);
-void metaChildTableOptsInit(STbOptions *pTbOptions, const char *name, tb_uid_t suid, const SKVRow tags);
-void metaTableOptsClear(STbOptions *pTbOptions);
+// STbCfg
+#define META_INIT_STB_CFG(NAME, TTL, KEEP, SUID, PSCHEMA, PTAGSCHEMA)                   \
+  {                                                                                     \
+    .name = (NAME), .ttl = (TTL), .keep = (KEEP), .type = META_SUPER_TABLE, .stbCfg = { \
+      .suid = (SUID),                                                                   \
+      .pSchema = (PSCHEMA),                                                             \
+      .pTagSchema = (PTAGSCHEMA)                                                        \
+    }                                                                                   \
+  }
+
+#define META_INIT_CTB_CFG(NAME, TTL, KEEP, SUID, PTAG)                                                                \
+  {                                                                                                                   \
+    .name = (NAME), .ttl = (TTL), .keep = (KEEP), .type = META_CHILD_TABLE, .ctbCfg = {.suid = (SUID), .pTag = PTAG } \
+  }
+
+#define META_INIT_NTB_CFG(NAME, TTL, KEEP, SUID, PSCHEMA)                                                      \
+  {                                                                                                            \
+    .name = (NAME), .ttl = (TTL), .keep = (KEEP), .type = META_NORMAL_TABLE, .ntbCfg = {.pSchema = (PSCHEMA) } \
+  }
+
+#define META_CLEAR_TB_CFG(pTbCfg)
+
+int   metaEncodeTbCfg(void **pBuf, STbCfg *pTbCfg);
+void *metaDecodeTbCfg(void *pBuf, STbCfg **pTbCfg);
 
 #ifdef __cplusplus
 }

include/dnode/vnode/tq/tq.h

@@ -161,9 +161,9 @@ typedef struct TqLogReader {
   int64_t (*logGetLastVer)(void* logHandle);
 } TqLogReader;
 
-typedef struct TqConfig {
+typedef struct STqCfg {
   // TODO
-} TqConfig;
+} STqCfg;
 
 typedef struct TqMemRef {
   SMemAllocatorFactory *pAlloctorFactory;
@@ -256,14 +256,14 @@ typedef struct STQ {
   // the collection of group handle
   // the handle of kvstore
   char*  path;
-  TqConfig*    tqConfig;
+  STqCfg*    tqConfig;
   TqLogReader* tqLogReader; 
   TqMemRef     tqMemRef;
   TqMetaStore* tqMeta;
 } STQ;
 
 // open in each vnode
-STQ* tqOpen(const char* path, TqConfig* tqConfig, TqLogReader* tqLogReader, SMemAllocatorFactory *allocFac);
+STQ* tqOpen(const char* path, STqCfg* tqConfig, TqLogReader* tqLogReader, SMemAllocatorFactory *allocFac);
 void tqDestroy(STQ*);
 
 // void* will be replace by a msg type

include/dnode/vnode/tsdb/tsdb.h

@@ -22,21 +22,25 @@ extern "C" {
 
 // TYPES EXPOSED
 typedef struct STsdb             STsdb;
-typedef struct STsdbOptions      STsdbOptions;
+typedef struct STsdbCfg          STsdbCfg;
 typedef struct STsdbMemAllocator STsdbMemAllocator;
 
 // STsdb
-STsdb *tsdbOpen(const char *path, const STsdbOptions *);
+STsdb *tsdbOpen(const char *path, const STsdbCfg *);
 void   tsdbClose(STsdb *);
 void   tsdbRemove(const char *path);
+int    tsdbInsertData(STsdb *pTsdb, void *pData, int len);
 
-// STsdbOptions
-int  tsdbOptionsInit(STsdbOptions *);
-void tsdbOptionsClear(STsdbOptions *);
+// STsdbCfg
+int  tsdbOptionsInit(STsdbCfg *);
+void tsdbOptionsClear(STsdbCfg *);
 
 /* ------------------------ STRUCT DEFINITIONS ------------------------ */
-struct STsdbOptions {
+struct STsdbCfg {
   uint64_t lruCacheSize;
+  uint32_t keep0;
+  uint32_t keep1;
+  uint32_t keep2;
   /* TODO */
 };
 

include/dnode/vnode/vnode.h

@@ -23,24 +23,67 @@
 #include "tarray.h"
 #include "tq.h"
 #include "tsdb.h"
+#include "wal.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
 /* ------------------------ TYPES EXPOSED ------------------------ */
-typedef struct SVnode        SVnode;
-typedef struct SVnodeOptions SVnodeOptions;
+typedef struct SVnode SVnode;
+typedef struct SVnodeCfg {
+  /** vnode buffer pool options */
+  struct {
+    /** write buffer size */
+    uint64_t wsize;
+    /** use heap allocator or arena allocator */
+    bool isHeapAllocator;
+  };
+
+  /** time to live of tables in this vnode */
+  uint32_t ttl;
+
+  /** data to keep in this vnode */
+  uint32_t keep;
+
+  /** if TS data is eventually consistency */
+  bool isWeak;
+
+  /** TSDB config */
+  STsdbCfg tsdbCfg;
+
+  /** META config */
+  SMetaCfg metaCfg;
+
+  /** TQ config */
+  STqCfg tqCfg;
+
+  /** WAL config */
+  SWalCfg walCfg;
+} SVnodeCfg;
 
 /* ------------------------ SVnode ------------------------ */
+/**
+ * @brief Initialize the vnode module
+ *
+ * @return int 0 for success and -1 for failure
+ */
+int vnodeInit();
+
+/**
+ * @brief clear a vnode
+ *
+ */
+void vnodeClear();
+
 /**
  * @brief Open a VNODE.
  *
  * @param path path of the vnode
- * @param pVnodeOptions options of the vnode
+ * @param pVnodeCfg options of the vnode
  * @return SVnode* The vnode object
  */
-SVnode *vnodeOpen(const char *path, const SVnodeOptions *pVnodeOptions);
+SVnode *vnodeOpen(const char *path, const SVnodeCfg *pVnodeCfg);
 
 /**
  * @brief Close a VNODE
@@ -85,61 +128,55 @@ int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
  */
 int vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
 
-/* ------------------------ SVnodeOptions ------------------------ */
+/* ------------------------ SVnodeCfg ------------------------ */
 /**
  * @brief Initialize VNODE options.
  *
  * @param pOptions The options object to be initialized. It should not be NULL.
  */
-void vnodeOptionsInit(SVnodeOptions *pOptions);
+void vnodeOptionsInit(SVnodeCfg *pOptions);
 
 /**
  * @brief Clear VNODE options.
  *
  * @param pOptions Options to clear.
  */
-void vnodeOptionsClear(SVnodeOptions *pOptions);
+void vnodeOptionsClear(SVnodeCfg *pOptions);
 
-/* ------------------------ STRUCT DEFINITIONS ------------------------ */
-struct SVnodeOptions {
-  /**
-   * @brief write buffer size in BYTES
-   *
-   */
-  uint64_t wsize;
+/* ------------------------ REQUESTS ------------------------ */
+typedef STbCfg SVCreateTableReq;
+typedef struct {
+  tb_uid_t uid;
+} SVDropTableReq;
 
-  /**
-   * @brief time to live of tables in this vnode
-   * in SECONDS
-   *
-   */
-  uint32_t ttl;
+typedef struct {
+  // TODO
+} SVSubmitReq;
 
-  /**
-   * @brief if time-series requests eventual consistency
-   *
-   */
-  bool isWeak;
+typedef struct {
+  uint64_t ver;
+  union {
+    SVCreateTableReq ctReq;
+    SVDropTableReq   dtReq;
+  };
+} SVnodeReq;
 
-  /**
-   * @brief if the allocator is heap allcator or arena allocator
-   *
-   */
-  bool isHeapAllocator;
+typedef struct {
+  int  err;
+  char info[];
+} SVnodeRsp;
 
-  /**
-   * @brief TSDB options
-   *
-   */
-  STsdbOptions tsdbOptions;
+#define VNODE_INIT_CREATE_STB_REQ(NAME, TTL, KEEP, SUID, PSCHEMA, PTAGSCHEMA) \
+  { .ver = 0, .ctReq = META_INIT_STB_CFG(NAME, TTL, KEEP, SUID, PSCHEMA, PTAGSCHEMA) }
 
-  /**
-   * @brief META options
-   *
-   */
-  SMetaOptions metaOptions;
-  // STqOptions   tqOptions; // TODO
-};
+#define VNODE_INIT_CREATE_CTB_REQ(NAME, TTL, KEEP, SUID, PTAG) \
+  { .ver = 0, .ctReq = META_INIT_CTB_CFG(NAME, TTL, KEEP, SUID, PTAG) }
+
+#define VNODE_INIT_CREATE_NTB_REQ(NAME, TTL, KEEP, SUID, PSCHEMA) \
+  { .ver = 0, .ctReq = META_INIT_NTB_CFG(NAME, TTL, KEEP, SUID, PSCHEMA) }
+
+int   vnodeBuildReq(void **buf, const SVnodeReq *pReq, uint8_t type);
+void *vnodeParseReq(void *buf, SVnodeReq *pReq, uint8_t type);
 
 /* ------------------------ FOR COMPILE ------------------------ */
 
@@ -148,27 +185,27 @@ struct SVnodeOptions {
 #include "taosmsg.h"
 #include "trpc.h"
 
-typedef struct {
-  char     db[TSDB_FULL_DB_NAME_LEN];
-  int32_t  cacheBlockSize;  // MB
-  int32_t  totalBlocks;
-  int32_t  daysPerFile;
-  int32_t  daysToKeep0;
-  int32_t  daysToKeep1;
-  int32_t  daysToKeep2;
-  int32_t  minRowsPerFileBlock;
-  int32_t  maxRowsPerFileBlock;
-  int8_t   precision;  // time resolution
-  int8_t   compression;
-  int8_t   cacheLastRow;
-  int8_t   update;
-  int8_t   quorum;
-  int8_t   replica;
-  int8_t   selfIndex;
-  int8_t   walLevel;
-  int32_t  fsyncPeriod;  // millisecond
-  SReplica replicas[TSDB_MAX_REPLICA];
-} SVnodeCfg;
+// typedef struct {
+//   char     db[TSDB_FULL_DB_NAME_LEN];
+//   int32_t  cacheBlockSize;  // MB
+//   int32_t  totalBlocks;
+//   int32_t  daysPerFile;
+//   int32_t  daysToKeep0;
+//   int32_t  daysToKeep1;
+//   int32_t  daysToKeep2;
+//   int32_t  minRowsPerFileBlock;
+//   int32_t  maxRowsPerFileBlock;
+//   int8_t   precision;  // time resolution
+//   int8_t   compression;
+//   int8_t   cacheLastRow;
+//   int8_t   update;
+//   int8_t   quorum;
+//   int8_t   replica;
+//   int8_t   selfIndex;
+//   int8_t   walLevel;
+//   int32_t  fsyncPeriod;  // millisecond
+//   SReplica replicas[TSDB_MAX_REPLICA];
+// } SVnodeCfg;
 
 typedef enum {
   VN_MSG_TYPE_WRITE = 1,

include/os/osSysinfo.h

@@ -52,7 +52,7 @@ bool    taosGetSysMemory(float *memoryUsedMB);
 void    taosPrintOsInfo();
 int     taosSystem(const char *cmd);
 void    taosKillSystem();
-bool    taosGetSystemUid(char *uid, int32_t uidlen);
+int32_t taosGetSystemUid(char *uid, int32_t uidlen);
 char *  taosGetCmdlineByPID(int pid);
 void    taosSetCoreDump(bool enable);
 

include/util/mallocator.h

@@ -22,10 +22,10 @@
 extern "C" {
 #endif
 
-typedef struct SMemAllocator SMemAllocator;
+typedef struct SMemAllocator        SMemAllocator;
+typedef struct SMemAllocatorFactory SMemAllocatorFactory;
 
 struct SMemAllocator {
-  char  name[16];
   void *impl;
   void *(*malloc)(SMemAllocator *, uint64_t size);
   void *(*calloc)(SMemAllocator *, uint64_t nmemb, uint64_t size);
@@ -34,11 +34,11 @@ struct SMemAllocator {
   uint64_t (*usage)(SMemAllocator *);
 };
 
-typedef struct {
+struct SMemAllocatorFactory {
   void *impl;
-  SMemAllocator *(*create)();
-  void (*destroy)(SMemAllocator *);
-} SMemAllocatorFactory;
+  SMemAllocator *(*create)(SMemAllocatorFactory *);
+  void (*destroy)(SMemAllocatorFactory *, SMemAllocator *);
+};
 
 #ifdef __cplusplus
 }

include/util/taoserror.h

@@ -115,25 +115,25 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_TSC_VALUE_OUT_OF_RANGE        TAOS_DEF_ERROR_CODE(0, 0x0223)  //"Value out of range")
 
 // mnode
-#define TSDB_CODE_MND_MSG_NOT_PROCESSED         TAOS_DEF_ERROR_CODE(0, 0x0300)  //"Message not processed")
-#define TSDB_CODE_MND_ACTION_IN_PROGRESS        TAOS_DEF_ERROR_CODE(0, 0x0301)  //"Message is progressing")
-#define TSDB_CODE_MND_ACTION_NEED_REPROCESSED   TAOS_DEF_ERROR_CODE(0, 0x0302)  //"Messag need to be reprocessed")
-#define TSDB_CODE_MND_NO_RIGHTS                 TAOS_DEF_ERROR_CODE(0, 0x0303)  //"Insufficient privilege for operation")
-#define TSDB_CODE_MND_APP_ERROR                 TAOS_DEF_ERROR_CODE(0, 0x0304)  //"Unexpected generic error in mnode")
-#define TSDB_CODE_MND_INVALID_CONNECTION        TAOS_DEF_ERROR_CODE(0, 0x0305)  //"Invalid message connection")
-#define TSDB_CODE_MND_INVALID_MSG_VERSION       TAOS_DEF_ERROR_CODE(0, 0x0306)  //"Incompatible protocol version")
-#define TSDB_CODE_MND_INVALID_MSG_LEN           TAOS_DEF_ERROR_CODE(0, 0x0307)  //"Invalid message length")
-#define TSDB_CODE_MND_INVALID_MSG_TYPE          TAOS_DEF_ERROR_CODE(0, 0x0308)  //"Invalid message type")
-#define TSDB_CODE_MND_TOO_MANY_SHELL_CONNS      TAOS_DEF_ERROR_CODE(0, 0x0309)  //"Too many connections")
-#define TSDB_CODE_MND_INVALID_SHOWOBJ           TAOS_DEF_ERROR_CODE(0, 0x030B)  //"Data expired")
-#define TSDB_CODE_MND_INVALID_QUERY_ID          TAOS_DEF_ERROR_CODE(0, 0x030C)  //"Invalid query id")
-#define TSDB_CODE_MND_INVALID_STREAM_ID         TAOS_DEF_ERROR_CODE(0, 0x030D)  //"Invalid stream id")
-#define TSDB_CODE_MND_INVALID_CONN_ID           TAOS_DEF_ERROR_CODE(0, 0x030E)  //"Invalid connection id")
-#define TSDB_CODE_MND_MNODE_IS_RUNNING          TAOS_DEF_ERROR_CODE(0, 0x0310)  //"mnode is alreay running")
-#define TSDB_CODE_MND_FAILED_TO_CONFIG_SYNC     TAOS_DEF_ERROR_CODE(0, 0x0311)  //"failed to config sync")
-#define TSDB_CODE_MND_FAILED_TO_START_SYNC      TAOS_DEF_ERROR_CODE(0, 0x0312)  //"failed to start sync")
-#define TSDB_CODE_MND_FAILED_TO_CREATE_DIR      TAOS_DEF_ERROR_CODE(0, 0x0313)  //"failed to create mnode dir")
-#define TSDB_CODE_MND_FAILED_TO_INIT_STEP       TAOS_DEF_ERROR_CODE(0, 0x0314)  //"failed to init components")
+#define TSDB_CODE_MND_MSG_NOT_PROCESSED         TAOS_DEF_ERROR_CODE(0, 0x0300)
+#define TSDB_CODE_MND_ACTION_IN_PROGRESS        TAOS_DEF_ERROR_CODE(0, 0x0301)
+#define TSDB_CODE_MND_ACTION_NEED_REPROCESSED   TAOS_DEF_ERROR_CODE(0, 0x0302)
+#define TSDB_CODE_MND_NO_RIGHTS                 TAOS_DEF_ERROR_CODE(0, 0x0303)
+#define TSDB_CODE_MND_INVALID_OPTIONS           TAOS_DEF_ERROR_CODE(0, 0x0304)
+#define TSDB_CODE_MND_INVALID_CONNECTION        TAOS_DEF_ERROR_CODE(0, 0x0305)
+#define TSDB_CODE_MND_INVALID_MSG_VERSION       TAOS_DEF_ERROR_CODE(0, 0x0306)
+#define TSDB_CODE_MND_INVALID_MSG_LEN           TAOS_DEF_ERROR_CODE(0, 0x0307)
+#define TSDB_CODE_MND_INVALID_MSG_TYPE          TAOS_DEF_ERROR_CODE(0, 0x0308)
+#define TSDB_CODE_MND_TOO_MANY_SHELL_CONNS      TAOS_DEF_ERROR_CODE(0, 0x0309)
+#define TSDB_CODE_MND_INVALID_SHOWOBJ           TAOS_DEF_ERROR_CODE(0, 0x030B)
+#define TSDB_CODE_MND_INVALID_QUERY_ID          TAOS_DEF_ERROR_CODE(0, 0x030C)
+#define TSDB_CODE_MND_INVALID_STREAM_ID         TAOS_DEF_ERROR_CODE(0, 0x030D)
+#define TSDB_CODE_MND_INVALID_CONN_ID           TAOS_DEF_ERROR_CODE(0, 0x030E)
+#define TSDB_CODE_MND_MNODE_IS_RUNNING          TAOS_DEF_ERROR_CODE(0, 0x0310)
+#define TSDB_CODE_MND_FAILED_TO_CONFIG_SYNC     TAOS_DEF_ERROR_CODE(0, 0x0311)
+#define TSDB_CODE_MND_FAILED_TO_START_SYNC      TAOS_DEF_ERROR_CODE(0, 0x0312)
+#define TSDB_CODE_MND_FAILED_TO_CREATE_DIR      TAOS_DEF_ERROR_CODE(0, 0x0313)
+#define TSDB_CODE_MND_FAILED_TO_INIT_STEP       TAOS_DEF_ERROR_CODE(0, 0x0314)
 
 #define TSDB_CODE_SDB_APP_ERROR                 TAOS_DEF_ERROR_CODE(0, 0x0320)
 #define TSDB_CODE_SDB_OBJ_ALREADY_THERE         TAOS_DEF_ERROR_CODE(0, 0x0321)

source/common/src/trow.c

@@ -15,75 +15,21 @@
 
 #include "trow.h"
 
-/* ------------ Structures ---------- */
-struct SRowBatch {
-  int32_t compress : 1;  // if batch row is compressed
-  int32_t nrows : 31;    // number of rows
-  int32_t tlen;          // total length (including `nrows` and `tlen`)
-  char    rows[];
-};
-
-struct SRowBuilder {
+#if 0
+void trbSetRowInfo(SRowBuilder *pRB, bool del, uint16_t sver) {
   // TODO
-};
+}
 
-struct SRowBatchIter {
-  int32_t    counter;  // row counter
-  SRowBatch *rb;       // row batch to iter
-  SRow       nrow;     // next row
-};
-
-struct SRowBatchBuilder {
+void trbSetRowVersion(SRowBuilder *pRB, uint64_t ver) {
   // TODO
-};
+}
 
-/* ------------ Methods ---------- */
-
-// SRowBuilder
-SRowBuilder *rowBuilderCreate() {
-  SRowBuilder *pRowBuilder = NULL;
+void trbSetRowTS(SRowBuilder *pRB, TSKEY ts) {
   // TODO
-
-  return pRowBuilder;
 }
 
-void rowBuilderDestroy(SRowBuilder *pRowBuilder) {
-  if (pRowBuilder) {
-    free(pRowBuilder);
-  }
+int trbWriteCol(SRowBuilder *pRB, void *pData, col_id_t cid) {
+  // TODO
+  return 0;
 }
-
-// SRowBatchIter
-SRowBatchIter *rowBatchIterCreate(SRowBatch *pRowBatch) {
-  SRowBatchIter *pRowBatchIter = (SRowBatchIter *)malloc(sizeof(*pRowBatchIter));
-  if (pRowBatchIter == NULL) {
-    return NULL;
-  }
-
-  pRowBatchIter->counter = 0;
-  pRowBatchIter->rb = pRowBatch;
-  pRowBatchIter->nrow = pRowBatch->rows;
-
-  return pRowBatchIter;
-};
-
-void rowBatchIterDestroy(SRowBatchIter *pRowBatchIter) {
-  if (pRowBatchIter) {
-    free(pRowBatchIter);
-  }
-}
-
-const SRow rowBatchIterNext(SRowBatchIter *pRowBatchIter) {
-  SRow r = NULL;
-  if (pRowBatchIter->counter < pRowBatchIter->rb->nrows) {
-    r = pRowBatchIter->nrow;
-    pRowBatchIter->counter += 1;
-    pRowBatchIter->nrow = (SRow)POINTER_SHIFT(r, rowLen(r));
-  }
-
-  return r;
-}
-
-// SRowBatchBuilder
-SRowBatchBuilder *rowBatchBuilderCreate();
-void              rowBatchBuilderDestroy(SRowBatchBuilder *);
+#endif

source/common/test/trowTest.cpp

@@ -0,0 +1,23 @@
+#include <gtest/gtest.h>
+
+#include "trow.h"
+
+TEST(td_row_test, build_row_to_target) {
+#if 0
+  char        dst[1024];
+  SRow*       pRow = (SRow*)dst;
+  int         ncols = 10;
+  col_id_t    cid;
+  void*       pData;
+  SRowBuilder rb = trbInit(TD_OR_ROW_BUILDER, NULL, 0, pRow, 1024);
+
+  trbSetRowInfo(&rb, false, 0);
+  trbSetRowTS(&rb, 1637550210000);
+  for (int c = 0; c < ncols; c++) {
+    cid = c;
+    if (trbWriteCol(&rb, pData, cid) < 0) {
+      // TODO
+    }
+  }
+#endif
+}

source/common/test/tschemaTest.cpp

@@ -0,0 +1,6 @@
+#include <gtest/gtest.h>
+#include "tschema.h"
+
+TEST(td_schema_test, build_schema_test) {
+
+}

source/common/type/type.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeBigint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeBinary.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeBlob.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeBool.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeDecimal.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeDouble.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeFloat.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeInt.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeJson.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeLongblob.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeNchar.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeNull.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeSmallint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeTimestamp.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeTinyint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeUBigint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeUSmallint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeUTinyint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeUint.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/common/type/typeVarchar.c

@@ -0,0 +1,14 @@
+/*
+ * 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/>.
+ */

source/dnode/mgmt/daemon/src/daemon.c

@@ -138,11 +138,15 @@ void dmnWaitSignal() {
 void dmnInitOption(SDnodeOpt *pOption) {
   pOption->sver = tsVersion;
   pOption->numOfCores = tsNumOfCores;
+  pOption->numOfSupportMnodes = 1;
+  pOption->numOfSupportVnodes = 1;
+  pOption->numOfSupportQnodes = 1;
+  pOption->statusInterval = tsStatusInterval;
+  pOption->mnodeEqualVnodeNum = tsMnodeEqualVnodeNum;
   pOption->numOfThreadsPerCore = tsNumOfThreadsPerCore;
   pOption->ratioOfQueryCores = tsRatioOfQueryCores;
   pOption->maxShellConns = tsMaxShellConns;
   pOption->shellActivityTimer = tsShellActivityTimer;
-  pOption->statusInterval = tsStatusInterval;
   pOption->serverPort = tsServerPort;
   tstrncpy(pOption->dataDir, tsDataDir, TSDB_FILENAME_LEN);
   tstrncpy(pOption->localEp, tsLocalEp, TSDB_EP_LEN);

source/dnode/mgmt/impl/inc/dndDnode.h

@@ -27,7 +27,7 @@ void    dndProcessDnodeReq(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet);
 void    dndProcessDnodeRsp(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet);
 
 int32_t dndGetDnodeId(SDnode *pDnode);
-int64_t dndGetClusterId(SDnode *pDnode);
+int32_t dndGetClusterId(SDnode *pDnode);
 void    dndGetDnodeEp(SDnode *pDnode, int32_t dnodeId, char *pEp, char *pFqdn, uint16_t *pPort);
 void    dndGetMnodeEpSet(SDnode *pDnode, SEpSet *pEpSet);
 void    dndSendRedirectMsg(SDnode *pDnode, SRpcMsg *pMsg);

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

@@ -57,8 +57,8 @@ typedef struct {
 typedef struct {
   int32_t    dnodeId;
   int32_t    dropped;
+  int32_t    clusterId;
   uint32_t   rebootTime;
-  int64_t    clusterId;
   SEpSet     mnodeEpSet;
   char      *file;
   SHashObj  *dnodeHash;

source/dnode/mgmt/impl/src/dndDnode.c

@@ -26,10 +26,10 @@ int32_t dndGetDnodeId(SDnode *pDnode) {
   return dnodeId;
 }
 
-int64_t dndGetClusterId(SDnode *pDnode) {
+int32_t dndGetClusterId(SDnode *pDnode) {
   SDnodeMgmt *pMgmt = &pDnode->dmgmt;
   taosRLockLatch(&pMgmt->latch);
-  int64_t clusterId = pMgmt->clusterId;
+  int32_t clusterId = pMgmt->clusterId;
   taosRUnLockLatch(&pMgmt->latch);
   return clusterId;
 }
@@ -190,78 +190,78 @@ static int32_t dndReadDnodes(SDnode *pDnode) {
   }
 
   cJSON *dnodeId = cJSON_GetObjectItem(root, "dnodeId");
-  if (!dnodeId || dnodeId->type != cJSON_String) {
+  if (!dnodeId || dnodeId->type != cJSON_Number) {
     dError("failed to read %s since dnodeId not found", pMgmt->file);
     goto PRASE_DNODE_OVER;
   }
-  pMgmt->dnodeId = atoi(dnodeId->valuestring);
+  pMgmt->dnodeId = dnodeId->valueint;
 
   cJSON *clusterId = cJSON_GetObjectItem(root, "clusterId");
-  if (!clusterId || clusterId->type != cJSON_String) {
+  if (!clusterId || clusterId->type != cJSON_Number) {
     dError("failed to read %s since clusterId not found", pMgmt->file);
     goto PRASE_DNODE_OVER;
   }
-  pMgmt->clusterId = atoll(clusterId->valuestring);
+  pMgmt->clusterId = clusterId->valueint;
 
   cJSON *dropped = cJSON_GetObjectItem(root, "dropped");
-  if (!dropped || dropped->type != cJSON_String) {
+  if (!dropped || dropped->type != cJSON_Number) {
     dError("failed to read %s since dropped not found", pMgmt->file);
     goto PRASE_DNODE_OVER;
   }
-  pMgmt->dropped = atoi(dropped->valuestring);
+  pMgmt->dropped = dropped->valueint;
 
-  cJSON *dnodeInfos = cJSON_GetObjectItem(root, "dnodeInfos");
-  if (!dnodeInfos || dnodeInfos->type != cJSON_Array) {
-    dError("failed to read %s since dnodeInfos not found", pMgmt->file);
+  cJSON *dnodes = cJSON_GetObjectItem(root, "dnodes");
+  if (!dnodes || dnodes->type != cJSON_Array) {
+    dError("failed to read %s since dnodes not found", pMgmt->file);
     goto PRASE_DNODE_OVER;
   }
 
-  int32_t dnodeInfosSize = cJSON_GetArraySize(dnodeInfos);
-  if (dnodeInfosSize <= 0) {
-    dError("failed to read %s since dnodeInfos size:%d invalid", pMgmt->file, dnodeInfosSize);
+  int32_t numOfNodes = cJSON_GetArraySize(dnodes);
+  if (numOfNodes <= 0) {
+    dError("failed to read %s since numOfNodes:%d invalid", pMgmt->file, numOfNodes);
     goto PRASE_DNODE_OVER;
   }
 
-  pMgmt->dnodeEps = calloc(1, dnodeInfosSize * sizeof(SDnodeEp) + sizeof(SDnodeEps));
+  pMgmt->dnodeEps = calloc(1, numOfNodes * sizeof(SDnodeEp) + sizeof(SDnodeEps));
   if (pMgmt->dnodeEps == NULL) {
     dError("failed to calloc dnodeEpList since %s", strerror(errno));
     goto PRASE_DNODE_OVER;
   }
-  pMgmt->dnodeEps->num = dnodeInfosSize;
+  pMgmt->dnodeEps->num = numOfNodes;
 
-  for (int32_t i = 0; i < dnodeInfosSize; ++i) {
-    cJSON *dnodeInfo = cJSON_GetArrayItem(dnodeInfos, i);
-    if (dnodeInfo == NULL) break;
+  for (int32_t i = 0; i < numOfNodes; ++i) {
+    cJSON *node = cJSON_GetArrayItem(dnodes, i);
+    if (node == NULL) break;
 
     SDnodeEp *pDnodeEp = &pMgmt->dnodeEps->eps[i];
 
-    cJSON *dnodeId = cJSON_GetObjectItem(dnodeInfo, "dnodeId");
-    if (!dnodeId || dnodeId->type != cJSON_String) {
+    cJSON *dnodeId = cJSON_GetObjectItem(node, "id");
+    if (!dnodeId || dnodeId->type != cJSON_Number) {
       dError("failed to read %s, dnodeId not found", pMgmt->file);
       goto PRASE_DNODE_OVER;
     }
-    pDnodeEp->id = atoi(dnodeId->valuestring);
+    pDnodeEp->id = dnodeId->valueint;
 
-    cJSON *isMnode = cJSON_GetObjectItem(dnodeInfo, "isMnode");
-    if (!isMnode || isMnode->type != cJSON_String) {
-      dError("failed to read %s, isMnode not found", pMgmt->file);
-      goto PRASE_DNODE_OVER;
-    }
-    pDnodeEp->isMnode = atoi(isMnode->valuestring);
-
-    cJSON *dnodeFqdn = cJSON_GetObjectItem(dnodeInfo, "dnodeFqdn");
+    cJSON *dnodeFqdn = cJSON_GetObjectItem(node, "fqdn");
     if (!dnodeFqdn || dnodeFqdn->type != cJSON_String || dnodeFqdn->valuestring == NULL) {
       dError("failed to read %s, dnodeFqdn not found", pMgmt->file);
       goto PRASE_DNODE_OVER;
     }
     tstrncpy(pDnodeEp->fqdn, dnodeFqdn->valuestring, TSDB_FQDN_LEN);
 
-    cJSON *dnodePort = cJSON_GetObjectItem(dnodeInfo, "dnodePort");
-    if (!dnodePort || dnodePort->type != cJSON_String) {
+    cJSON *dnodePort = cJSON_GetObjectItem(node, "port");
+    if (!dnodePort || dnodePort->type != cJSON_Number) {
       dError("failed to read %s, dnodePort not found", pMgmt->file);
       goto PRASE_DNODE_OVER;
     }
-    pDnodeEp->port = atoi(dnodePort->valuestring);
+    pDnodeEp->port = dnodePort->valueint;
+
+    cJSON *isMnode = cJSON_GetObjectItem(node, "isMnode");
+    if (!isMnode || isMnode->type != cJSON_Number) {
+      dError("failed to read %s, isMnode not found", pMgmt->file);
+      goto PRASE_DNODE_OVER;
+    }
+    pDnodeEp->isMnode = isMnode->valueint;
   }
 
   code = 0;
@@ -307,16 +307,16 @@ static int32_t dndWriteDnodes(SDnode *pDnode) {
   char   *content = calloc(1, maxLen + 1);
 
   len += snprintf(content + len, maxLen - len, "{\n");
-  len += snprintf(content + len, maxLen - len, "  \"dnodeId\": \"%d\",\n", pMgmt->dnodeId);
-  len += snprintf(content + len, maxLen - len, "  \"clusterId\": \"%" PRId64 "\",\n", pMgmt->clusterId);
-  len += snprintf(content + len, maxLen - len, "  \"dropped\": \"%d\",\n", pMgmt->dropped);
-  len += snprintf(content + len, maxLen - len, "  \"dnodeInfos\": [{\n");
+  len += snprintf(content + len, maxLen - len, "  \"dnodeId\": %d,\n", pMgmt->dnodeId);
+  len += snprintf(content + len, maxLen - len, "  \"clusterId\": %d,\n", pMgmt->clusterId);
+  len += snprintf(content + len, maxLen - len, "  \"dropped\": %d,\n", pMgmt->dropped);
+  len += snprintf(content + len, maxLen - len, "  \"dnodes\": [{\n");
   for (int32_t i = 0; i < pMgmt->dnodeEps->num; ++i) {
     SDnodeEp *pDnodeEp = &pMgmt->dnodeEps->eps[i];
-    len += snprintf(content + len, maxLen - len, "    \"dnodeId\": \"%d\",\n", pDnodeEp->id);
-    len += snprintf(content + len, maxLen - len, "    \"isMnode\": \"%d\",\n", pDnodeEp->isMnode);
-    len += snprintf(content + len, maxLen - len, "    \"dnodeFqdn\": \"%s\",\n", pDnodeEp->fqdn);
-    len += snprintf(content + len, maxLen - len, "    \"dnodePort\": \"%u\"\n", pDnodeEp->port);
+    len += snprintf(content + len, maxLen - len, "    \"id\": %d,\n", pDnodeEp->id);
+    len += snprintf(content + len, maxLen - len, "    \"fqdn\": \"%s\",\n", pDnodeEp->fqdn);
+    len += snprintf(content + len, maxLen - len, "    \"port\": %u,\n", pDnodeEp->port);
+    len += snprintf(content + len, maxLen - len, "    \"isMnode\": %d\n", pDnodeEp->isMnode);
     if (i < pMgmt->dnodeEps->num - 1) {
       len += snprintf(content + len, maxLen - len, "  },{\n");
     } else {
@@ -344,23 +344,27 @@ static void dndSendStatusMsg(SDnode *pDnode) {
     return;
   }
 
-  bool changed = false;
-
   SDnodeMgmt *pMgmt = &pDnode->dmgmt;
   taosRLockLatch(&pMgmt->latch);
-  pStatus->sversion = htonl(pDnode->opt.sver);
+  pStatus->sver = htonl(pDnode->opt.sver);
   pStatus->dnodeId = htonl(pMgmt->dnodeId);
-  pStatus->clusterId = htobe64(pMgmt->clusterId);
+  pStatus->clusterId = htonl(pMgmt->clusterId);
   pStatus->rebootTime = htonl(pMgmt->rebootTime);
-  pStatus->numOfCores = htonl(pDnode->opt.numOfCores);
+  pStatus->numOfCores = htons(pDnode->opt.numOfCores);
+  pStatus->numOfSupportMnodes = htons(pDnode->opt.numOfCores);
+  pStatus->numOfSupportVnodes = htons(pDnode->opt.numOfCores);
+  pStatus->numOfSupportQnodes = htons(pDnode->opt.numOfCores);
   tstrncpy(pStatus->dnodeEp, pDnode->opt.localEp, TSDB_EP_LEN);
+
   pStatus->clusterCfg.statusInterval = htonl(pDnode->opt.statusInterval);
-  tstrncpy(pStatus->clusterCfg.timezone, pDnode->opt.timezone, TSDB_TIMEZONE_LEN);
-  tstrncpy(pStatus->clusterCfg.locale, pDnode->opt.locale, TSDB_LOCALE_LEN);
-  tstrncpy(pStatus->clusterCfg.charset, pDnode->opt.charset, TSDB_LOCALE_LEN);
+  pStatus->clusterCfg.mnodeEqualVnodeNum = htonl(pDnode->opt.mnodeEqualVnodeNum);
   pStatus->clusterCfg.checkTime = 0;
   char timestr[32] = "1970-01-01 00:00:00.00";
   (void)taosParseTime(timestr, &pStatus->clusterCfg.checkTime, (int32_t)strlen(timestr), TSDB_TIME_PRECISION_MILLI, 0);
+  pStatus->clusterCfg.checkTime = htonl(pStatus->clusterCfg.checkTime);
+  tstrncpy(pStatus->clusterCfg.timezone, pDnode->opt.timezone, TSDB_TIMEZONE_LEN);
+  tstrncpy(pStatus->clusterCfg.locale, pDnode->opt.locale, TSDB_LOCALE_LEN);
+  tstrncpy(pStatus->clusterCfg.charset, pDnode->opt.charset, TSDB_LOCALE_LEN);
   taosRUnLockLatch(&pMgmt->latch);
 
   dndGetVnodeLoads(pDnode, &pStatus->vnodeLoads);
@@ -373,7 +377,7 @@ static void dndSendStatusMsg(SDnode *pDnode) {
 static void dndUpdateDnodeCfg(SDnode *pDnode, SDnodeCfg *pCfg) {
   SDnodeMgmt *pMgmt = &pDnode->dmgmt;
   if (pMgmt->dnodeId == 0 || pMgmt->dropped != pCfg->dropped) {
-    dInfo("set dnodeId:%d clusterId:%" PRId64 " dropped:%d", pCfg->dnodeId, pCfg->clusterId, pCfg->dropped);
+    dInfo("set dnodeId:%d clusterId:%d dropped:%d", pCfg->dnodeId, pCfg->clusterId, pCfg->dropped);
 
     taosWLockLatch(&pMgmt->latch);
     pMgmt->dnodeId = pCfg->dnodeId;
@@ -414,7 +418,7 @@ static void dndProcessStatusRsp(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet) {
   SStatusRsp *pRsp = pMsg->pCont;
   SDnodeCfg  *pCfg = &pRsp->dnodeCfg;
   pCfg->dnodeId = htonl(pCfg->dnodeId);
-  pCfg->clusterId = htobe64(pCfg->clusterId);
+  pCfg->clusterId = htonl(pCfg->clusterId);
   dndUpdateDnodeCfg(pDnode, pCfg);
 
   if (pCfg->dropped) return;
@@ -434,6 +438,7 @@ static void dndProcessAuthRsp(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet) { a
 static void dndProcessGrantRsp(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet) { assert(1); }
 
 static void dndProcessConfigDnodeReq(SDnode *pDnode, SRpcMsg *pMsg) {
+  dDebug("config msg is received");
   SCfgDnodeMsg *pCfg = pMsg->pCont;
 
   int32_t code = TSDB_CODE_OPS_NOT_SUPPORT;
@@ -443,12 +448,12 @@ static void dndProcessConfigDnodeReq(SDnode *pDnode, SRpcMsg *pMsg) {
 }
 
 static void dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pMsg) {
-  dInfo("startup msg is received");
+  dDebug("startup msg is received");
 
   SStartupMsg *pStartup = rpcMallocCont(sizeof(SStartupMsg));
   dndGetStartup(pDnode, pStartup);
 
-  dInfo("startup msg is sent, step:%s desc:%s finished:%d", pStartup->name, pStartup->desc, pStartup->finished);
+  dDebug("startup msg is sent, step:%s desc:%s finished:%d", pStartup->name, pStartup->desc, pStartup->finished);
 
   SRpcMsg rpcRsp = {.handle = pMsg->handle, .pCont = pStartup, .contLen = sizeof(SStartupMsg)};
   rpcSendResponse(&rpcRsp);
@@ -464,7 +469,7 @@ static void *dnodeThreadRoutine(void *param) {
     pthread_testcancel();
 
     if (dndGetStat(pDnode) == DND_STAT_RUNNING) {
-      // dndSendStatusMsg(pDnode);
+      dndSendStatusMsg(pDnode);
     }
   }
 }

source/dnode/mgmt/impl/src/dndMnode.c

@@ -130,43 +130,43 @@ static int32_t dndReadMnodeFile(SDnode *pDnode) {
   }
 
   cJSON *deployed = cJSON_GetObjectItem(root, "deployed");
-  if (!deployed || deployed->type != cJSON_String) {
+  if (!deployed || deployed->type != cJSON_Number) {
     dError("failed to read %s since deployed not found", pMgmt->file);
     goto PRASE_MNODE_OVER;
   }
-  pMgmt->deployed = atoi(deployed->valuestring);
+  pMgmt->deployed = deployed->valueint;
 
   cJSON *dropped = cJSON_GetObjectItem(root, "dropped");
-  if (!dropped || dropped->type != cJSON_String) {
+  if (!dropped || dropped->type != cJSON_Number) {
     dError("failed to read %s since dropped not found", pMgmt->file);
     goto PRASE_MNODE_OVER;
   }
-  pMgmt->dropped = atoi(dropped->valuestring);
+  pMgmt->dropped = dropped->valueint;
 
-  cJSON *nodes = cJSON_GetObjectItem(root, "nodes");
-  if (!nodes || nodes->type != cJSON_Array) {
+  cJSON *mnodes = cJSON_GetObjectItem(root, "mnodes");
+  if (!mnodes || mnodes->type != cJSON_Array) {
     dError("failed to read %s since nodes not found", pMgmt->file);
     goto PRASE_MNODE_OVER;
   }
 
-  pMgmt->replica = cJSON_GetArraySize(nodes);
+  pMgmt->replica = cJSON_GetArraySize(mnodes);
   if (pMgmt->replica <= 0 || pMgmt->replica > TSDB_MAX_REPLICA) {
-    dError("failed to read %s since nodes size %d invalid", pMgmt->file, pMgmt->replica);
+    dError("failed to read %s since mnodes size %d invalid", pMgmt->file, pMgmt->replica);
     goto PRASE_MNODE_OVER;
   }
 
   for (int32_t i = 0; i < pMgmt->replica; ++i) {
-    cJSON *node = cJSON_GetArrayItem(nodes, i);
+    cJSON *node = cJSON_GetArrayItem(mnodes, i);
     if (node == NULL) break;
 
     SReplica *pReplica = &pMgmt->replicas[i];
 
     cJSON *id = cJSON_GetObjectItem(node, "id");
-    if (!id || id->type != cJSON_String || id->valuestring == NULL) {
+    if (!id || id->type != cJSON_Number) {
       dError("failed to read %s since id not found", pMgmt->file);
       goto PRASE_MNODE_OVER;
     }
-    pReplica->id = atoi(id->valuestring);
+    pReplica->id = id->valueint;
 
     cJSON *fqdn = cJSON_GetObjectItem(node, "fqdn");
     if (!fqdn || fqdn->type != cJSON_String || fqdn->valuestring == NULL) {
@@ -176,15 +176,15 @@ static int32_t dndReadMnodeFile(SDnode *pDnode) {
     tstrncpy(pReplica->fqdn, fqdn->valuestring, TSDB_FQDN_LEN);
 
     cJSON *port = cJSON_GetObjectItem(node, "port");
-    if (!port || port->type != cJSON_String || port->valuestring == NULL) {
+    if (!port || port->type != cJSON_Number) {
       dError("failed to read %s since port not found", pMgmt->file);
       goto PRASE_MNODE_OVER;
     }
-    pReplica->port = atoi(port->valuestring);
+    pReplica->port = port->valueint;
   }
 
   code = 0;
-  dInfo("succcessed to read file %s", pMgmt->file);
+  dDebug("succcessed to read file %s, deployed:%d dropped:%d", pMgmt->file, pMgmt->deployed, pMgmt->dropped);
 
 PRASE_MNODE_OVER:
   if (content != NULL) free(content);
@@ -213,15 +213,15 @@ static int32_t dndWriteMnodeFile(SDnode *pDnode) {
   char   *content = calloc(1, maxLen + 1);
 
   len += snprintf(content + len, maxLen - len, "{\n");
-  len += snprintf(content + len, maxLen - len, "  \"deployed\": \"%d\",\n", pMgmt->deployed);
+  len += snprintf(content + len, maxLen - len, "  \"deployed\": %d,\n", pMgmt->deployed);
 
-  len += snprintf(content + len, maxLen - len, "  \"dropped\": \"%d\",\n", pMgmt->dropped);
-  len += snprintf(content + len, maxLen - len, "  \"nodes\": [{\n");
+  len += snprintf(content + len, maxLen - len, "  \"dropped\": %d,\n", pMgmt->dropped);
+  len += snprintf(content + len, maxLen - len, "  \"mnodes\": [{\n");
   for (int32_t i = 0; i < pMgmt->replica; ++i) {
     SReplica *pReplica = &pMgmt->replicas[i];
-    len += snprintf(content + len, maxLen - len, "    \"id\": \"%d\",\n", pReplica->id);
+    len += snprintf(content + len, maxLen - len, "    \"id\": %d,\n", pReplica->id);
     len += snprintf(content + len, maxLen - len, "    \"fqdn\": \"%s\",\n", pReplica->fqdn);
-    len += snprintf(content + len, maxLen - len, "    \"port\": \"%u\"\n", pReplica->port);
+    len += snprintf(content + len, maxLen - len, "    \"port\": %u\n", pReplica->port);
     if (i < pMgmt->replica - 1) {
       len += snprintf(content + len, maxLen - len, "  },{\n");
     } else {
@@ -241,7 +241,7 @@ static int32_t dndWriteMnodeFile(SDnode *pDnode) {
     return -1;
   }
 
-  dInfo("successed to write %s", pMgmt->file);
+  dInfo("successed to write %s, deployed:%d dropped:%d", pMgmt->file, pMgmt->deployed, pMgmt->dropped);
   return 0;
 }
 
@@ -289,7 +289,6 @@ static void dndStopMnodeWorker(SDnode *pDnode) {
 
   taosWLockLatch(&pMgmt->latch);
   pMgmt->deployed = 0;
-  pMgmt->pMnode = NULL;
   taosWUnLockLatch(&pMgmt->latch);
 
   while (pMgmt->refCount > 1) taosMsleep(10);
@@ -332,6 +331,12 @@ static void dndInitMnodeOption(SDnode *pDnode, SMnodeOpt *pOption) {
   pOption->putMsgToApplyMsgFp = dndPutMsgIntoMnodeApplyQueue;
   pOption->dnodeId = dndGetDnodeId(pDnode);
   pOption->clusterId = dndGetClusterId(pDnode);
+  pOption->sver = pDnode->opt.sver;
+  pOption->statusInterval = pDnode->opt.statusInterval;
+  pOption->mnodeEqualVnodeNum = pDnode->opt.mnodeEqualVnodeNum;
+  pOption->timezone = pDnode->opt.timezone;
+  pOption->charset = pDnode->opt.charset;
+  pOption->locale = pDnode->opt.locale;
 }
 
 static void dndBuildMnodeDeployOption(SDnode *pDnode, SMnodeOpt *pOption) {
@@ -390,29 +395,34 @@ static int32_t dndBuildMnodeOptionFromMsg(SDnode *pDnode, SMnodeOpt *pOption, SC
 static int32_t dndOpenMnode(SDnode *pDnode, SMnodeOpt *pOption) {
   SMnodeMgmt *pMgmt = &pDnode->mmgmt;
 
-  int32_t code = dndStartMnodeWorker(pDnode);
-  if (code != 0) {
-    dError("failed to start mnode worker since %s", terrstr());
-    return code;
-  }
-
   SMnode *pMnode = mndOpen(pDnode->dir.mnode, pOption);
   if (pMnode == NULL) {
     dError("failed to open mnode since %s", terrstr());
-    code = terrno;
-    dndStopMnodeWorker(pDnode);
-    terrno = code;
-    return code;
+    return -1;
   }
+  pMgmt->deployed = 1;
 
-  if (dndWriteMnodeFile(pDnode) != 0) {
+  int32_t code = dndWriteMnodeFile(pDnode);
+  if (code != 0) {
     dError("failed to write mnode file since %s", terrstr());
     code = terrno;
+    pMgmt->deployed = 0;
+    mndClose(pMnode);
+    mndDestroy(pDnode->dir.mnode);
+    terrno = code;
+    return -1;
+  }
+
+  code = dndStartMnodeWorker(pDnode);
+  if (code != 0) {
+    dError("failed to start mnode worker since %s", terrstr());
+    code = terrno;
+    pMgmt->deployed = 0;
     dndStopMnodeWorker(pDnode);
     mndClose(pMnode);
     mndDestroy(pDnode->dir.mnode);
     terrno = code;
-    return code;
+    return -1;
   }
 
   taosWLockLatch(&pMgmt->latch);
@@ -420,6 +430,7 @@ static int32_t dndOpenMnode(SDnode *pDnode, SMnodeOpt *pOption) {
   pMgmt->deployed = 1;
   taosWUnLockLatch(&pMgmt->latch);
 
+  dInfo("mnode open successfully");
   return 0;
 }
 
@@ -908,6 +919,7 @@ void dndCleanupMnode(SDnode *pDnode) {
   dndStopMnodeWorker(pDnode);
   dndCleanupMnodeMgmtWorker(pDnode);
   tfree(pMgmt->file);
+  mndClose(pMgmt->pMnode);
   dInfo("dnode-mnode is cleaned up");
 }
 

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

@@ -58,6 +58,8 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TSDB_MSG_TYPE_CREATE_DNODE] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_CONFIG_DNODE] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_DROP_DNODE] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TSDB_MSG_TYPE_CREATE_MNODE] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TSDB_MSG_TYPE_DROP_MNODE] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_CREATE_DB] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_DROP_DB] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_USE_DB] = dndProcessMnodeWriteMsg;
@@ -115,12 +117,12 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TSDB_MSG_TYPE_CONFIG_DNODE_IN_RSP] = dndProcessMnodeWriteMsg;
 
   // message from dnode to mnode
-  pMgmt->msgFp[TSDB_MSG_TYPE_AUTH] = dndProcessMnodeReadMsg;
-  pMgmt->msgFp[TSDB_MSG_TYPE_AUTH_RSP] = dndProcessDnodeRsp;
   pMgmt->msgFp[TSDB_MSG_TYPE_GRANT] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_GRANT_RSP] = dndProcessDnodeRsp;
   pMgmt->msgFp[TSDB_MSG_TYPE_STATUS] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_STATUS_RSP] = dndProcessDnodeRsp;
+  pMgmt->msgFp[TSDB_MSG_TYPE_AUTH] = dndProcessMnodeReadMsg;
+  pMgmt->msgFp[TSDB_MSG_TYPE_AUTH_RSP] = dndProcessDnodeRsp;
 }
 
 static void dndProcessResponse(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet) {
@@ -131,19 +133,20 @@ static void dndProcessResponse(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet) {
 
   if (dndGetStat(pDnode) == DND_STAT_STOPPED) {
     if (pMsg == NULL || pMsg->pCont == NULL) return;
-    dTrace("RPC %p, rsp:%s is ignored since dnode is stopping", pMsg->handle, taosMsg[msgType]);
+    dTrace("RPC %p, rsp:%s app:%p is ignored since dnode is stopping", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
     rpcFreeCont(pMsg->pCont);
     return;
   }
 
   DndMsgFp fp = pMgmt->msgFp[msgType];
   if (fp != NULL) {
-    dTrace("RPC %p, rsp:%s will be processed, code:%s", pMsg->handle, taosMsg[msgType], tstrerror(pMsg->code));
     (*fp)(pDnode, pMsg, pEpSet);
+    dTrace("RPC %p, rsp:%s app:%p is processed, code:0x%0X", pMsg->handle, taosMsg[msgType], pMsg->ahandle,
+           pMsg->code & 0XFFFF);
   } else {
-    dError("RPC %p, rsp:%s not processed", pMsg->handle, taosMsg[msgType]);
-    rpcFreeCont(pMsg->pCont);
+    dError("RPC %p, rsp:%s app:%p not processed", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
   }
+  rpcFreeCont(pMsg->pCont);
 }
 
 static int32_t dndInitClient(SDnode *pDnode) {
@@ -187,19 +190,19 @@ static void dndProcessRequest(void *param, SRpcMsg *pMsg, SEpSet *pEpSet) {
 
   int32_t msgType = pMsg->msgType;
   if (msgType == TSDB_MSG_TYPE_NETWORK_TEST) {
-    dTrace("RPC %p, network test req will be processed", pMsg->handle);
+    dTrace("RPC %p, network test req, app:%p will be processed", pMsg->handle, pMsg->ahandle);
     dndProcessDnodeReq(pDnode, pMsg, pEpSet);
     return;
   }
 
   if (dndGetStat(pDnode) == DND_STAT_STOPPED) {
-    dError("RPC %p, req:%s is ignored since dnode exiting", pMsg->handle, taosMsg[msgType]);
+    dError("RPC %p, req:%s app:%p is ignored since dnode exiting", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
     SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_DND_EXITING};
     rpcSendResponse(&rspMsg);
     rpcFreeCont(pMsg->pCont);
     return;
   } else if (dndGetStat(pDnode) != DND_STAT_RUNNING) {
-    dError("RPC %p, req:%s is ignored since dnode not running", pMsg->handle, taosMsg[msgType]);
+    dError("RPC %p, req:%s app:%p is ignored since dnode not running", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
     SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_APP_NOT_READY};
     rpcSendResponse(&rspMsg);
     rpcFreeCont(pMsg->pCont);
@@ -207,7 +210,7 @@ static void dndProcessRequest(void *param, SRpcMsg *pMsg, SEpSet *pEpSet) {
   }
 
   if (pMsg->pCont == NULL) {
-    dTrace("RPC %p, req:%s not processed since content is null", pMsg->handle, taosMsg[msgType]);
+    dTrace("RPC %p, req:%s app:%p not processed since content is null", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
     SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_DND_INVALID_MSG_LEN};
     rpcSendResponse(&rspMsg);
     return;
@@ -215,10 +218,10 @@ static void dndProcessRequest(void *param, SRpcMsg *pMsg, SEpSet *pEpSet) {
 
   DndMsgFp fp = pMgmt->msgFp[msgType];
   if (fp != NULL) {
-    dTrace("RPC %p, req:%s will be processed", pMsg->handle, taosMsg[msgType]);
+    dTrace("RPC %p, req:%s app:%p will be processed", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
     (*fp)(pDnode, pMsg, pEpSet);
   } else {
-    dError("RPC %p, req:%s is not processed", pMsg->handle, taosMsg[msgType]);
+    dError("RPC %p, req:%s app:%p is not processed since no handle", pMsg->handle, taosMsg[msgType], pMsg->ahandle);
     SRpcMsg rspMsg = {.handle = pMsg->handle, .code = TSDB_CODE_MSG_NOT_PROCESSED};
     rpcSendResponse(&rspMsg);
     rpcFreeCont(pMsg->pCont);

source/dnode/mgmt/impl/src/dndVnodes.c

@@ -38,7 +38,7 @@ typedef struct {
   int32_t    threadIndex;
   pthread_t *pThreadId;
   SVnodeObj *pVnodes;
-  SDnode    *pDnode;
+  SDnode *   pDnode;
 } SVnodeThread;
 
 static int32_t dndInitVnodeReadWorker(SDnode *pDnode);
@@ -73,7 +73,7 @@ void           dndProcessVnodeSyncMsg(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEp
 void           dndProcessVnodeMgmtMsg(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet);
 static int32_t dndPutMsgIntoVnodeApplyQueue(SDnode *pDnode, int32_t vgId, SVnodeMsg *pMsg);
 
-static SVnodeObj  *dndAcquireVnode(SDnode *pDnode, int32_t vgId);
+static SVnodeObj * dndAcquireVnode(SDnode *pDnode, int32_t vgId);
 static void        dndReleaseVnode(SDnode *pDnode, SVnodeObj *pVnode);
 static int32_t     dndCreateVnodeWrapper(SDnode *pDnode, int32_t vgId, char *path, SVnode *pImpl);
 static void        dndDropVnodeWrapper(SDnode *pDnode, SVnodeObj *pVnode);
@@ -95,7 +95,7 @@ static int32_t dndProcessCompactVnodeReq(SDnode *pDnode, SRpcMsg *rpcMsg);
 
 static SVnodeObj *dndAcquireVnode(SDnode *pDnode, int32_t vgId) {
   SVnodesMgmt *pMgmt = &pDnode->vmgmt;
-  SVnodeObj   *pVnode = NULL;
+  SVnodeObj *  pVnode = NULL;
   int32_t      refCount = 0;
 
   taosRLockLatch(&pMgmt->latch);
@@ -128,7 +128,7 @@ static void dndReleaseVnode(SDnode *pDnode, SVnodeObj *pVnode) {
 
 static int32_t dndCreateVnodeWrapper(SDnode *pDnode, int32_t vgId, char *path, SVnode *pImpl) {
   SVnodesMgmt *pMgmt = &pDnode->vmgmt;
-  SVnodeObj   *pVnode = calloc(1, sizeof(SVnodeObj));
+  SVnodeObj *  pVnode = calloc(1, sizeof(SVnodeObj));
   if (pVnode == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
@@ -208,7 +208,7 @@ static SVnodeObj **dndGetVnodesFromHash(SDnode *pDnode, int32_t *numOfVnodes) {
   void *pIter = taosHashIterate(pMgmt->hash, NULL);
   while (pIter) {
     SVnodeObj **ppVnode = pIter;
-    SVnodeObj  *pVnode = *ppVnode;
+    SVnodeObj * pVnode = *ppVnode;
     if (pVnode) {
       num++;
       if (num < size) {
@@ -230,9 +230,9 @@ static int32_t dndGetVnodesFromFile(SDnode *pDnode, SVnodeObj **ppVnodes, int32_
   int32_t    code = TSDB_CODE_DND_VNODE_READ_FILE_ERROR;
   int32_t    len = 0;
   int32_t    maxLen = 30000;
-  char      *content = calloc(1, maxLen + 1);
-  cJSON     *root = NULL;
-  FILE      *fp = NULL;
+  char *     content = calloc(1, maxLen + 1);
+  cJSON *    root = NULL;
+  FILE *     fp = NULL;
   char       file[PATH_MAX + 20] = {0};
   SVnodeObj *pVnodes = NULL;
 
@@ -277,7 +277,7 @@ static int32_t dndGetVnodesFromFile(SDnode *pDnode, SVnodeObj **ppVnodes, int32_
   }
 
   for (int32_t i = 0; i < vnodesNum; ++i) {
-    cJSON     *vnode = cJSON_GetArrayItem(vnodes, i);
+    cJSON *    vnode = cJSON_GetArrayItem(vnodes, i);
     SVnodeObj *pVnode = &pVnodes[i];
 
     cJSON *vgId = cJSON_GetObjectItem(vnode, "vgId");
@@ -321,7 +321,7 @@ static int32_t dndWriteVnodesToFile(SDnode *pDnode) {
 
   int32_t     len = 0;
   int32_t     maxLen = 30000;
-  char       *content = calloc(1, maxLen + 1);
+  char *      content = calloc(1, maxLen + 1);
   int32_t     numOfVnodes = 0;
   SVnodeObj **pVnodes = dndGetVnodesFromHash(pDnode, &numOfVnodes);
 
@@ -403,8 +403,8 @@ static int32_t dndDropVnode(SDnode *pDnode, SVnodeObj *pVnode) {
 
 static void *dnodeOpenVnodeFunc(void *param) {
   SVnodeThread *pThread = param;
-  SDnode       *pDnode = pThread->pDnode;
-  SVnodesMgmt  *pMgmt = &pDnode->vmgmt;
+  SDnode *      pDnode = pThread->pDnode;
+  SVnodesMgmt * pMgmt = &pDnode->vmgmt;
 
   dDebug("thread:%d, start to open %d vnodes", pThread->threadIndex, pThread->vnodeNum);
   setThreadName("open-vnodes");
@@ -527,6 +527,7 @@ static int32_t dndParseCreateVnodeReq(SRpcMsg *rpcMsg, int32_t *vgId, SVnodeCfg
   SCreateVnodeMsg *pCreate = rpcMsg->pCont;
   *vgId = htonl(pCreate->vgId);
 
+#if 0
   tstrncpy(pCfg->db, pCreate->db, TSDB_FULL_DB_NAME_LEN);
   pCfg->cacheBlockSize = htonl(pCreate->cacheBlockSize);
   pCfg->totalBlocks = htonl(pCreate->totalBlocks);
@@ -549,6 +550,7 @@ static int32_t dndParseCreateVnodeReq(SRpcMsg *rpcMsg, int32_t *vgId, SVnodeCfg
     pCfg->replicas[i].port = htons(pCreate->replicas[i].port);
     tstrncpy(pCfg->replicas[i].fqdn, pCreate->replicas[i].fqdn, TSDB_FQDN_LEN);
   }
+#endif
 
   return 0;
 }
@@ -738,7 +740,7 @@ static void dndProcessVnodeFetchQueue(SVnodeObj *pVnode, SVnodeMsg *pMsg) {
 
 static void dndProcessVnodeWriteQueue(SVnodeObj *pVnode, taos_qall qall, int32_t numOfMsgs) {
   SVnodeMsg *pMsg = vnodeInitMsg(numOfMsgs);
-  SRpcMsg   *pRpcMsg = NULL;
+  SRpcMsg *  pRpcMsg = NULL;
 
   for (int32_t i = 0; i < numOfMsgs; ++i) {
     taosGetQitem(qall, (void **)&pRpcMsg);
@@ -895,6 +897,7 @@ static int32_t dndInitVnodeMgmtWorker(SDnode *pDnode) {
     return -1;
   }
 
+  dDebug("vnode mgmt worker is initialized");
   return 0;
 }
 
@@ -903,6 +906,7 @@ static void dndCleanupVnodeMgmtWorker(SDnode *pDnode) {
   tWorkerFreeQueue(&pMgmt->mgmtPool, pMgmt->pMgmtQ);
   tWorkerCleanup(&pMgmt->mgmtPool);
   pMgmt->pMgmtQ = NULL;
+  dDebug("vnode mgmt worker is closed");
 }
 
 static int32_t dndAllocVnodeQueryQueue(SDnode *pDnode, SVnodeObj *pVnode) {
@@ -963,6 +967,7 @@ static int32_t dndInitVnodeReadWorker(SDnode *pDnode) {
     return -1;
   }
 
+  dDebug("vnode read worker is initialized");
   return 0;
 }
 
@@ -970,6 +975,7 @@ static void dndCleanupVnodeReadWorker(SDnode *pDnode) {
   SVnodesMgmt *pMgmt = &pDnode->vmgmt;
   tWorkerCleanup(&pMgmt->fetchPool);
   tWorkerCleanup(&pMgmt->queryPool);
+  dDebug("vnode close worker is initialized");
 }
 
 static int32_t dndAllocVnodeWriteQueue(SDnode *pDnode, SVnodeObj *pVnode) {
@@ -1007,7 +1013,7 @@ static void dndFreeVnodeApplyQueue(SDnode *pDnode, SVnodeObj *pVnode) {
 }
 
 static int32_t dndInitVnodeWriteWorker(SDnode *pDnode) {
-  SVnodesMgmt  *pMgmt = &pDnode->vmgmt;
+  SVnodesMgmt * pMgmt = &pDnode->vmgmt;
   SMWorkerPool *pPool = &pMgmt->writePool;
   pPool->name = "vnode-write";
   pPool->max = tsNumOfCores;
@@ -1016,12 +1022,14 @@ static int32_t dndInitVnodeWriteWorker(SDnode *pDnode) {
     return -1;
   }
 
+  dDebug("vnode write worker is initialized");
   return 0;
 }
 
 static void dndCleanupVnodeWriteWorker(SDnode *pDnode) {
   SVnodesMgmt *pMgmt = &pDnode->vmgmt;
   tMWorkerCleanup(&pMgmt->writePool);
+  dDebug("vnode write worker is closed");
 }
 
 static int32_t dndAllocVnodeSyncQueue(SDnode *pDnode, SVnodeObj *pVnode) {
@@ -1046,7 +1054,7 @@ static int32_t dndInitVnodeSyncWorker(SDnode *pDnode) {
   if (maxThreads < 1) maxThreads = 1;
 
   SVnodesMgmt  *pMgmt = &pDnode->vmgmt;
-  SMWorkerPool *pPool = &pMgmt->writePool;
+  SMWorkerPool *pPool = &pMgmt->syncPool;
   pPool->name = "vnode-sync";
   pPool->max = maxThreads;
   if (tMWorkerInit(pPool) != 0) {
@@ -1054,12 +1062,14 @@ static int32_t dndInitVnodeSyncWorker(SDnode *pDnode) {
     return -1;
   }
 
+  dDebug("vnode sync worker is initialized");
   return 0;
 }
 
 static void dndCleanupVnodeSyncWorker(SDnode *pDnode) {
   SVnodesMgmt *pMgmt = &pDnode->vmgmt;
   tMWorkerCleanup(&pMgmt->syncPool);
+  dDebug("vnode sync worker is closed");
 }
 
 int32_t dndInitVnodes(SDnode *pDnode) {
@@ -1111,12 +1121,12 @@ void dndGetVnodeLoads(SDnode *pDnode, SVnodeLoads *pLoads) {
   pLoads->num = taosHashGetSize(pMgmt->hash);
 
   int32_t v = 0;
-  void   *pIter = taosHashIterate(pMgmt->hash, NULL);
+  void *  pIter = taosHashIterate(pMgmt->hash, NULL);
   while (pIter) {
     SVnodeObj **ppVnode = pIter;
     if (ppVnode == NULL || *ppVnode == NULL) continue;
 
-    SVnodeObj  *pVnode = *ppVnode;
+    SVnodeObj * pVnode = *ppVnode;
     SVnodeLoad *pLoad = &pLoads->data[v++];
 
     vnodeGetLoad(pVnode->pImpl, pLoad);

source/dnode/mnode/impl/CMakeLists.txt

@@ -10,4 +10,5 @@ target_link_libraries(
     PRIVATE sdb
     PRIVATE transport
     PRIVATE cjson
+    PRIVATE sync
 )

source/dnode/mnode/impl/inc/mndDef.h

@@ -24,6 +24,7 @@
 #include "thash.h"
 #include "cJSON.h"
 #include "mnode.h"
+#include "sync.h"
 
 #ifdef __cplusplus
 extern "C" {
@@ -78,6 +79,28 @@ typedef enum {
 
 typedef enum { TRN_POLICY_ROLLBACK = 1, TRN_POLICY_RETRY = 2 } ETrnPolicy;
 
+typedef enum {
+  DND_STATUS_OFFLINE = 0,
+  DND_STATUS_READY = 1,
+  DND_STATUS_CREATING = 2,
+  DND_STATUS_DROPPING = 3
+} EDndStatus;
+
+typedef enum {
+  DND_REASON_ONLINE = 0,
+  DND_REASON_STATUS_MSG_TIMEOUT,
+  DND_REASON_STATUS_NOT_RECEIVED,
+  DND_REASON_VERSION_NOT_MATCH,
+  DND_REASON_DNODE_ID_NOT_MATCH,
+  DND_REASON_CLUSTER_ID_NOT_MATCH,
+  DND_REASON_MN_EQUAL_VN_NOT_MATCH,
+  DND_REASON_STATUS_INTERVAL_NOT_MATCH,
+  DND_REASON_TIME_ZONE_NOT_MATCH,
+  DND_REASON_LOCALE_NOT_MATCH,
+  DND_REASON_CHARSET_NOT_MATCH,
+  DND_REASON_OTHERS
+} EDndReason;
+
 typedef struct STrans {
   int32_t    id;
   ETrnStage  stage;
@@ -92,36 +115,39 @@ typedef struct STrans {
 } STrans;
 
 typedef struct SClusterObj {
-  int64_t id;
+  int32_t id;
   char    uid[TSDB_CLUSTER_ID_LEN];
   int64_t createdTime;
   int64_t updateTime;
 } SClusterObj;
 
 typedef struct SDnodeObj {
-  int32_t  id;
-  int32_t  vnodes;
-  int64_t  createdTime;
-  int64_t  updateTime;
-  int64_t  lastAccess;
-  int64_t  rebootTime;  // time stamp for last reboot
-  char     fqdn[TSDB_FQDN_LEN];
-  char     ep[TSDB_EP_LEN];
-  uint16_t port;
-  int16_t  numOfCores;       // from dnode status msg
-  int8_t   alternativeRole;  // from dnode status msg, 0-any, 1-mgmt, 2-dnode
-  int8_t   status;           // set in balance function
-  int8_t   offlineReason;
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  int64_t    rebootTime;
+  int32_t    accessTimes;
+  int16_t    numOfMnodes;
+  int16_t    numOfVnodes;
+  int16_t    numOfQnodes;
+  int16_t    numOfSupportMnodes;
+  int16_t    numOfSupportVnodes;
+  int16_t    numOfSupportQnodes;
+  int16_t    numOfCores;
+  EDndStatus status;
+  EDndReason offlineReason;
+  uint16_t   port;
+  char       fqdn[TSDB_FQDN_LEN];
+  char       ep[TSDB_EP_LEN];
 } SDnodeObj;
 
 typedef struct SMnodeObj {
   int32_t    id;
-  int8_t     status;
-  int8_t     role;
-  int32_t    roleTerm;
-  int64_t    roleTime;
   int64_t    createdTime;
   int64_t    updateTime;
+  ESyncState role;
+  int32_t    roleTerm;
+  int64_t    roleTime;
   SDnodeObj *pDnode;
 } SMnodeObj;
 
@@ -273,25 +299,18 @@ typedef struct {
   char     payload[];
 } SShowObj;
 
-typedef struct {
-  int32_t len;
-  void   *rsp;
-} SMnodeRsp;
-
 typedef struct SMnodeMsg {
+  char    user[TSDB_USER_LEN];
   SMnode *pMnode;
-  void (*fp)(SMnodeMsg *pMsg, int32_t code);
-  SRpcConnInfo conn;
-  SUserObj    *pUser;
-  int16_t      received;
-  int16_t      successed;
-  int16_t      expected;
-  int16_t      retry;
-  int32_t      code;
-  int64_t      createdTime;
-  SMnodeRsp       rpcRsp;
-  SRpcMsg      rpcMsg;
-  char         pCont[];
+  int16_t received;
+  int16_t successed;
+  int16_t expected;
+  int16_t retry;
+  int32_t code;
+  int64_t createdTime;
+  SRpcMsg rpcMsg;
+  int32_t contLen;
+  void   *pCont;
 } SMnodeMsg;
 
 #ifdef __cplusplus

source/dnode/mnode/impl/inc/mndDnode.h

@@ -22,8 +22,10 @@
 extern "C" {
 #endif
 
-int32_t mndInitDnode(SMnode *pMnode);
-void    mndCleanupDnode(SMnode *pMnode);
+int32_t    mndInitDnode(SMnode *pMnode);
+void       mndCleanupDnode(SMnode *pMnode);
+SDnodeObj *mndAcquireDnode(SMnode *pMnode, int32_t dnodeId);
+void       mndReleaseDnode(SMnode *pMnode, SDnodeObj *pDnode);
 
 #ifdef __cplusplus
 }

source/dnode/mnode/impl/inc/mndInt.h

@@ -36,7 +36,7 @@ typedef struct {
 
 typedef struct SMnode {
   int32_t           dnodeId;
-  int64_t           clusterId;
+  int32_t           clusterId;
   int8_t            replica;
   int8_t            selfIndex;
   SReplica          replicas[TSDB_MAX_REPLICA];
@@ -50,12 +50,14 @@ typedef struct SMnode {
   SendMsgToMnodeFp  sendMsgToMnodeFp;
   SendRedirectMsgFp sendRedirectMsgFp;
   PutMsgToMnodeQFp  putMsgToApplyMsgFp;
+  int32_t           sver;
+  int32_t           statusInterval;
+  int32_t           mnodeEqualVnodeNum;
+  char             *timezone;
+  char             *locale;
+  char             *charset;
 } SMnode;
 
-tmr_h   mndGetTimer(SMnode *pMnode);
-int32_t mndGetDnodeId(SMnode *pMnode);
-int64_t mndGetClusterId(SMnode *pMnode);
-
 void mndSendMsgToDnode(SMnode *pMnode, SEpSet *pEpSet, SRpcMsg *rpcMsg);
 void mndSendMsgToMnode(SMnode *pMnode, SRpcMsg *pMsg);
 void mndSendRedirectMsg(SMnode *pMnode, SRpcMsg *pMsg);

source/dnode/mnode/impl/inc/mndMnode.h

@@ -24,8 +24,7 @@ extern "C" {
 
 int32_t mndInitMnode(SMnode *pMnode);
 void    mndCleanupMnode(SMnode *pMnode);
-void    mndGetMnodeEpSetForPeer(SEpSet *epSet, bool redirect);
-void    mndGetMnodeEpSetForShell(SEpSet *epSet, bool redirect);
+bool    mndIsMnode(SMnode *pMnode, int32_t dnodeId);
 
 #ifdef __cplusplus
 }

source/dnode/mnode/impl/inc/mndTrans.h

@@ -13,8 +13,8 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_TRANSACTION_INT_H_
-#define _TD_TRANSACTION_INT_H_
+#ifndef _TD_MND_TRANS_H_
+#define _TD_MND_TRANS_H_
 
 #include "mndInt.h"
 
@@ -44,4 +44,4 @@ SSdbRow *mndTransActionDecode(SSdbRaw *pRaw);
 }
 #endif
 
-#endif /*_TD_TRANSACTION_INT_H_*/
+#endif /*_TD_MND_TRANS_H_*/

source/dnode/mnode/impl/inc/mndUser.h

@@ -22,8 +22,10 @@
 extern "C" {
 #endif
 
-int32_t mndInitUser(SMnode *pMnode);
-void    mndCleanupUser(SMnode *pMnode);
+int32_t   mndInitUser(SMnode *pMnode);
+void      mndCleanupUser(SMnode *pMnode);
+SUserObj *mndAcquireUser(SMnode *pMnode, const char *userName);
+void      mndReleaseUser(SMnode *pMnode, SUserObj *pUser);
 
 #ifdef __cplusplus
 }

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

@@ -44,6 +44,7 @@ static SSdbRow *mnodeAcctActionDecode(SSdbRaw *pRaw) {
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
   if (sver != SDB_ACCT_VER) {
+    mError("failed to decode acct since %s", terrstr());
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     return NULL;
   }
@@ -68,18 +69,30 @@ static SSdbRow *mnodeAcctActionDecode(SSdbRaw *pRaw) {
   return pRow;
 }
 
-static int32_t mnodeAcctActionInsert(SSdb *pSdb, SAcctObj *pAcct) { return 0; }
-
-static int32_t mnodeAcctActionDelete(SSdb *pSdb, SAcctObj *pAcct) { return 0; }
-
-static int32_t mnodeAcctActionUpdate(SSdb *pSdb, SAcctObj *pSrcAcct, SAcctObj *pDstAcct) {
-  SAcctObj tObj;
-  int32_t  len = (int32_t)((int8_t *)&tObj.info - (int8_t *)&tObj);
-  memcpy(pDstAcct, pSrcAcct, len);
+static int32_t mnodeAcctActionInsert(SSdb *pSdb, SAcctObj *pAcct) {
+  mTrace("acct:%s, perform insert action", pAcct->acct);
+  memset(&pAcct->info, 0, sizeof(SAcctInfo));
   return 0;
 }
 
-static int32_t mnodeCreateDefaultAcct(SSdb *pSdb) {
+static int32_t mnodeAcctActionDelete(SSdb *pSdb, SAcctObj *pAcct) {
+  mTrace("acct:%s, perform delete action", pAcct->acct);
+  return 0;
+}
+
+static int32_t mnodeAcctActionUpdate(SSdb *pSdb, SAcctObj *pSrcAcct, SAcctObj *pDstAcct) {
+  mTrace("acct:%s, perform update action", pSrcAcct->acct);
+
+  memcpy(pSrcAcct->acct, pDstAcct->acct, TSDB_USER_LEN);
+  pSrcAcct->createdTime = pDstAcct->createdTime;
+  pSrcAcct->updateTime = pDstAcct->updateTime;
+  pSrcAcct->acctId = pDstAcct->acctId;
+  pSrcAcct->status = pDstAcct->status;
+  pSrcAcct->cfg = pDstAcct->cfg;
+  return 0;
+}
+
+static int32_t mnodeCreateDefaultAcct(SMnode *pMnode) {
   int32_t code = 0;
 
   SAcctObj acctObj = {0};
@@ -98,7 +111,8 @@ static int32_t mnodeCreateDefaultAcct(SSdb *pSdb) {
   if (pRaw == NULL) return -1;
   sdbSetRawStatus(pRaw, SDB_STATUS_READY);
 
-  return sdbWrite(pSdb, pRaw);
+  mTrace("acct:%s, will be created while deploy sdb", acctObj.acct);
+  return sdbWrite(pMnode->pSdb, pRaw);
 }
 
 int32_t mndInitAcct(SMnode *pMnode) {

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

@@ -14,8 +14,97 @@
  */
 
 #define _DEFAULT_SOURCE
-#include "os.h"
-#include "mndInt.h"
+#include "mndCluster.h"
+#include "mndTrans.h"
 
-int32_t mndInitCluster(SMnode *pMnode) { return 0; }
-void    mndCleanupCluster(SMnode *pMnode) {}
+#define SDB_CLUSTER_VER 1
+
+static SSdbRaw *mndClusterActionEncode(SClusterObj *pCluster) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_CLUSTER, SDB_CLUSTER_VER, sizeof(SClusterObj));
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pCluster->id);
+  SDB_SET_INT64(pRaw, dataPos, pCluster->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pCluster->updateTime)
+  SDB_SET_BINARY(pRaw, dataPos, pCluster->uid, TSDB_CLUSTER_ID_LEN)
+
+  return pRaw;
+}
+
+static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != SDB_CLUSTER_VER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode cluster since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow     *pRow = sdbAllocRow(sizeof(SClusterObj));
+  SClusterObj *pCluster = sdbGetRowObj(pRow);
+  if (pCluster == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pCluster->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pCluster->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pCluster->updateTime)
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pCluster->uid, TSDB_CLUSTER_ID_LEN)
+
+  return pRow;
+}
+
+static int32_t mndClusterActionInsert(SSdb *pSdb, SClusterObj *pCluster) {
+  mTrace("cluster:%d, perform insert action", pCluster->id);
+  return 0;
+}
+
+static int32_t mndClusterActionDelete(SSdb *pSdb, SClusterObj *pCluster) {
+  mTrace("cluster:%d, perform delete action", pCluster->id);
+  return 0;
+}
+
+static int32_t mndClusterActionUpdate(SSdb *pSdb, SClusterObj *pSrcCluster, SClusterObj *pDstCluster) {
+  mTrace("cluster:%d, perform update action", pSrcCluster->id);
+  return 0;
+}
+
+static int32_t mndCreateDefaultCluster(SMnode *pMnode) {
+  SClusterObj clusterObj = {0};
+  clusterObj.createdTime = taosGetTimestampMs();
+  clusterObj.updateTime = clusterObj.createdTime;
+
+  int32_t code = taosGetSystemUid(clusterObj.uid, TSDB_CLUSTER_ID_LEN);
+  if (code != 0) {
+    strcpy(clusterObj.uid, "tdengine2.0");
+    mError("failed to get uid from system, set to default val %s", clusterObj.uid);
+  } else {
+    mDebug("cluster:%d, uid is %s", clusterObj.id, clusterObj.uid);
+  }
+  clusterObj.id = MurmurHash3_32(clusterObj.uid, TSDB_CLUSTER_ID_LEN);
+  clusterObj.id = abs(clusterObj.id);
+  pMnode->clusterId = clusterObj.id;
+
+  SSdbRaw *pRaw = mndClusterActionEncode(&clusterObj);
+  if (pRaw == NULL) return -1;
+  sdbSetRawStatus(pRaw, SDB_STATUS_READY);
+
+  mTrace("cluster:%d, will be created while deploy sdb", clusterObj.id);
+  return sdbWrite(pMnode->pSdb, pRaw);
+}
+
+int32_t mndInitCluster(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_CLUSTER,
+                     .keyType = SDB_KEY_INT32,
+                     .deployFp = (SdbDeployFp)mndCreateDefaultCluster,
+                     .encodeFp = (SdbEncodeFp)mndClusterActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndClusterActionDecode,
+                     .insertFp = (SdbInsertFp)mndClusterActionInsert,
+                     .updateFp = (SdbUpdateFp)mndClusterActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndClusterActionDelete};
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupCluster(SMnode *pMnode) {}

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

@@ -14,8 +14,339 @@
  */
 
 #define _DEFAULT_SOURCE
-#include "os.h"
-#include "mndInt.h"
+#include "mndDnode.h"
+#include "mndMnode.h"
+#include "mndTrans.h"
+#include "ttime.h"
 
-int32_t mndInitDnode(SMnode *pMnode) { return 0; }
-void    mndCleanupDnode(SMnode *pMnode) {}
+#define SDB_DNODE_VER 1
+
+static char *offlineReason[] = {
+    "",
+    "status msg timeout",
+    "status not received",
+    "version not match",
+    "dnodeId not match",
+    "clusterId not match",
+    "mnEqualVn not match",
+    "interval not match",
+    "timezone not match",
+    "locale not match",
+    "charset not match",
+    "unknown",
+};
+
+static SSdbRaw *mndDnodeActionEncode(SDnodeObj *pDnode) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_DNODE, SDB_DNODE_VER, sizeof(SDnodeObj));
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pDnode->id);
+  SDB_SET_INT64(pRaw, dataPos, pDnode->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pDnode->updateTime)
+  SDB_SET_INT16(pRaw, dataPos, pDnode->port)
+  SDB_SET_BINARY(pRaw, dataPos, pDnode->fqdn, TSDB_FQDN_LEN)
+  SDB_SET_DATALEN(pRaw, dataPos);
+
+  return pRaw;
+}
+
+static SSdbRow *mndDnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != SDB_DNODE_VER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode dnode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SDnodeObj));
+  SDnodeObj *pDnode = sdbGetRowObj(pRow);
+  if (pDnode == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pDnode->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pDnode->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pDnode->updateTime)
+  SDB_GET_INT16(pRaw, pRow, dataPos, &pDnode->port)
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pDnode->fqdn, TSDB_FQDN_LEN)
+
+  return pRow;
+}
+
+static void mnodeResetDnode(SDnodeObj *pDnode) {
+  pDnode->rebootTime = 0;
+  pDnode->accessTimes = 0;
+  pDnode->numOfCores = 0;
+  pDnode->numOfMnodes = 0;
+  pDnode->numOfVnodes = 0;
+  pDnode->numOfQnodes = 0;
+  pDnode->numOfSupportMnodes = 0;
+  pDnode->numOfSupportVnodes = 0;
+  pDnode->numOfSupportQnodes = 0;
+  pDnode->status = DND_STATUS_OFFLINE;
+  pDnode->offlineReason = DND_REASON_STATUS_NOT_RECEIVED;
+  snprintf(pDnode->ep, TSDB_EP_LEN, "%s:%u", pDnode->fqdn, pDnode->port);
+}
+
+static int32_t mndDnodeActionInsert(SSdb *pSdb, SDnodeObj *pDnode) {
+  mTrace("dnode:%d, perform insert action", pDnode->id);
+  mnodeResetDnode(pDnode);
+  return 0;
+}
+
+static int32_t mndDnodeActionDelete(SSdb *pSdb, SDnodeObj *pDnode) {
+  mTrace("dnode:%d, perform delete action", pDnode->id);
+  return 0;
+}
+
+static int32_t mndDnodeActionUpdate(SSdb *pSdb, SDnodeObj *pSrcDnode, SDnodeObj *pDstDnode) {
+  mTrace("dnode:%d, perform update action", pSrcDnode->id);
+  pSrcDnode->id = pDstDnode->id;
+  pSrcDnode->createdTime = pDstDnode->createdTime;
+  pSrcDnode->updateTime = pDstDnode->updateTime;
+  pSrcDnode->port = pDstDnode->port;
+  memcpy(pSrcDnode->fqdn, pDstDnode->fqdn, TSDB_FQDN_LEN); 
+  return 0;
+}
+
+static int32_t mndCreateDefaultDnode(SMnode *pMnode) {
+  SDnodeObj dnodeObj = {0};
+  dnodeObj.id = 1;
+  dnodeObj.createdTime = taosGetTimestampMs();
+  dnodeObj.updateTime = dnodeObj.createdTime;
+  dnodeObj.port = pMnode->replicas[0].port;
+  memcpy(&dnodeObj.fqdn, pMnode->replicas[0].fqdn, TSDB_FQDN_LEN);
+
+  SSdbRaw *pRaw = mndDnodeActionEncode(&dnodeObj);
+  if (pRaw == NULL) return -1;
+  sdbSetRawStatus(pRaw, SDB_STATUS_READY);
+
+  mTrace("dnode:%d, will be created while deploy sdb", dnodeObj.id);
+  return sdbWrite(pMnode->pSdb, pRaw);
+}
+
+static SDnodeObj *mndAcquireDnodeByEp(SMnode *pMnode, char *pEpStr) {
+  SSdb *pSdb = pMnode->pSdb;
+
+  void *pIter = NULL;
+  while (1) {
+    SDnodeObj *pDnode = NULL;
+    pIter = sdbFetch(pSdb, SDB_DNODE, pIter, (void **)&pDnode);
+    if (pIter == NULL) break;
+
+    if (strncasecmp(pEpStr, pDnode->ep, TSDB_EP_LEN) == 0) {
+      sdbCancelFetch(pSdb, pIter);
+      return pDnode;
+    }
+  }
+
+  return NULL;
+}
+
+static int32_t mndGetDnodeSize(SMnode *pMnode) {
+  SSdb *pSdb = pMnode->pSdb;
+  return sdbGetSize(pSdb, SDB_DNODE);
+}
+
+static void mndGetDnodeData(SMnode *pMnode, SDnodeEps *pEps, int32_t numOfEps) {
+  SSdb *pSdb = pMnode->pSdb;
+
+  int32_t i = 0;
+  void   *pIter = NULL;
+  while (1) {
+    SDnodeObj *pDnode = NULL;
+    pIter = sdbFetch(pSdb, SDB_DNODE, pIter, (void **)&pDnode);
+    if (pIter == NULL) break;
+    if (i >= numOfEps) {
+      sdbCancelFetch(pSdb, pIter);
+      break;
+    }
+
+    SDnodeEp *pEp = &pEps->eps[i];
+    pEp->id = htonl(pDnode->id);
+    pEp->port = htons(pDnode->port);
+    memcpy(pEp->fqdn, pDnode->fqdn, TSDB_FQDN_LEN);
+    pEp->isMnode = 0;
+    if (mndIsMnode(pMnode, pDnode->id)) {
+      pEp->isMnode = 1;
+    }
+    i++;
+  }
+
+  pEps->num = htonl(i);
+}
+
+static int32_t mndCheckClusterCfgPara(SMnode *pMnode, const SClusterCfg *pCfg) {
+  if (pCfg->mnodeEqualVnodeNum != pMnode->mnodeEqualVnodeNum) {
+    mError("\"mnodeEqualVnodeNum\"[%d - %d] cfg inconsistent", pCfg->mnodeEqualVnodeNum, pMnode->mnodeEqualVnodeNum);
+    return DND_REASON_MN_EQUAL_VN_NOT_MATCH;
+  }
+
+  if (pCfg->statusInterval != pMnode->statusInterval) {
+    mError("\"statusInterval\"[%d - %d] cfg inconsistent", pCfg->statusInterval, pMnode->statusInterval);
+    return DND_REASON_STATUS_INTERVAL_NOT_MATCH;
+  }
+
+  int64_t checkTime = 0;
+  char    timestr[32] = "1970-01-01 00:00:00.00";
+  (void)taosParseTime(timestr, &checkTime, (int32_t)strlen(timestr), TSDB_TIME_PRECISION_MILLI, 0);
+  if ((0 != strcasecmp(pCfg->timezone, pMnode->timezone)) && (checkTime != pCfg->checkTime)) {
+    mError("\"timezone\"[%s - %s] [%" PRId64 " - %" PRId64 "] cfg inconsistent", pCfg->timezone, tsTimezone,
+           pCfg->checkTime, checkTime);
+    return DND_REASON_TIME_ZONE_NOT_MATCH;
+  }
+
+  if (0 != strcasecmp(pCfg->locale, pMnode->locale)) {
+    mError("\"locale\"[%s - %s]  cfg parameters inconsistent", pCfg->locale, pMnode->locale);
+    return DND_REASON_LOCALE_NOT_MATCH;
+  }
+
+  if (0 != strcasecmp(pCfg->charset, pMnode->charset)) {
+    mError("\"charset\"[%s - %s] cfg parameters inconsistent.", pCfg->charset, pMnode->charset);
+    return DND_REASON_CHARSET_NOT_MATCH;
+  }
+
+  return 0;
+}
+
+static void mndParseStatusMsg(SStatusMsg *pStatus) {
+  pStatus->sver = htonl(pStatus->sver);
+  pStatus->dnodeId = htonl(pStatus->dnodeId);
+  pStatus->clusterId = htonl(pStatus->clusterId);
+  pStatus->rebootTime = htonl(pStatus->rebootTime);
+  pStatus->numOfCores = htons(pStatus->numOfCores);
+  pStatus->numOfSupportMnodes = htons(pStatus->numOfSupportMnodes);
+  pStatus->numOfSupportVnodes = htons(pStatus->numOfSupportVnodes);
+  pStatus->numOfSupportQnodes = htons(pStatus->numOfSupportQnodes);
+
+  pStatus->clusterCfg.statusInterval = htonl(pStatus->clusterCfg.statusInterval);
+  pStatus->clusterCfg.mnodeEqualVnodeNum = htonl(pStatus->clusterCfg.mnodeEqualVnodeNum);
+  pStatus->clusterCfg.checkTime = htobe64(pStatus->clusterCfg.checkTime);
+}
+
+static int32_t mndProcessStatusMsg(SMnode *pMnode, SMnodeMsg *pMsg) {
+  SStatusMsg *pStatus = pMsg->rpcMsg.pCont;
+  mndParseStatusMsg(pStatus);
+
+  SDnodeObj *pDnode = NULL;
+  if (pStatus->dnodeId == 0) {
+    pDnode = mndAcquireDnodeByEp(pMnode, pStatus->dnodeEp);
+    if (pDnode == NULL) {
+      mDebug("dnode:%s, not created yet", pStatus->dnodeEp);
+      return TSDB_CODE_MND_DNODE_NOT_EXIST;
+    }
+  } else {
+    pDnode = mndAcquireDnode(pMnode, pStatus->dnodeId);
+    if (pDnode == NULL) {
+      pDnode = mndAcquireDnodeByEp(pMnode, pStatus->dnodeEp);
+      if (pDnode != NULL && pDnode->status != DND_STATUS_READY) {
+        pDnode->offlineReason = DND_REASON_DNODE_ID_NOT_MATCH;
+      }
+      mError("dnode:%d, %s not exist", pStatus->dnodeId, pStatus->dnodeEp);
+      mndReleaseDnode(pMnode, pDnode);
+      return TSDB_CODE_MND_DNODE_NOT_EXIST;
+    }
+  }
+
+  if (pStatus->sver != pMnode->sver) {
+    if (pDnode != NULL && pDnode->status != DND_STATUS_READY) {
+      pDnode->offlineReason = DND_REASON_VERSION_NOT_MATCH;
+    }
+    mndReleaseDnode(pMnode, pDnode);
+    mError("dnode:%d, status msg version:%d not match cluster:%d", pStatus->dnodeId, pStatus->sver, pMnode->sver);
+    return TSDB_CODE_MND_INVALID_MSG_VERSION;
+  }
+
+  if (pStatus->dnodeId == 0) {
+    mDebug("dnode:%d %s, first access, set clusterId %d", pDnode->id, pDnode->ep, pMnode->clusterId);
+  } else {
+    if (pStatus->clusterId != pMnode->clusterId) {
+      if (pDnode != NULL && pDnode->status != DND_STATUS_READY) {
+        pDnode->offlineReason = DND_REASON_CLUSTER_ID_NOT_MATCH;
+      }
+      mError("dnode:%d, clusterId %d not match exist %d", pDnode->id, pStatus->clusterId, pMnode->clusterId);
+      mndReleaseDnode(pMnode, pDnode);
+      return TSDB_CODE_MND_INVALID_CLUSTER_ID;
+    } else {
+      pDnode->accessTimes++;
+      mTrace("dnode:%d, status received, access times %d", pDnode->id, pDnode->accessTimes);
+    }
+  }
+
+  if (pDnode->status == DND_STATUS_OFFLINE) {
+    // Verify whether the cluster parameters are consistent when status change from offline to ready
+    int32_t ret = mndCheckClusterCfgPara(pMnode, &pStatus->clusterCfg);
+    if (0 != ret) {
+      pDnode->offlineReason = ret;
+      mError("dnode:%d, cluster cfg inconsistent since:%s", pDnode->id, offlineReason[ret]);
+      mndReleaseDnode(pMnode, pDnode);
+      return TSDB_CODE_MND_CLUSTER_CFG_INCONSISTENT;
+    }
+
+    mInfo("dnode:%d, from offline to online", pDnode->id);
+  }
+
+  pDnode->rebootTime = pStatus->rebootTime;
+  pDnode->numOfCores = pStatus->numOfCores;
+  pDnode->numOfSupportMnodes = pStatus->numOfSupportMnodes;
+  pDnode->numOfSupportVnodes = pStatus->numOfSupportVnodes;
+  pDnode->numOfSupportQnodes = pStatus->numOfSupportQnodes;
+  pDnode->status = DND_STATUS_READY;
+
+  int32_t     numOfEps = mndGetDnodeSize(pMnode);
+  int32_t     contLen = sizeof(SStatusRsp) + numOfEps * sizeof(SDnodeEp);
+  SStatusRsp *pRsp = rpcMallocCont(contLen);
+  if (pRsp == NULL) {
+    mndReleaseDnode(pMnode, pDnode);
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  pRsp->dnodeCfg.dnodeId = htonl(pDnode->id);
+  pRsp->dnodeCfg.dropped = 0;
+  pRsp->dnodeCfg.clusterId = htonl(pMnode->clusterId);
+  mndGetDnodeData(pMnode, &pRsp->dnodeEps, numOfEps);
+
+  pMsg->contLen = contLen;
+  pMsg->pCont = pRsp;
+  mndReleaseDnode(pMnode, pDnode);
+
+  return 0;
+}
+
+static int32_t mndProcessCreateDnodeMsg(SMnode *pMnode, SMnodeMsg *pMsg) { return 0; }
+
+static int32_t mndProcessDropDnodeMsg(SMnode *pMnode, SMnodeMsg *pMsg) { return 0; }
+
+static int32_t mndProcessConfigDnodeMsg(SMnode *pMnode, SMnodeMsg *pMsg) { return 0; }
+
+int32_t mndInitDnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_DNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .deployFp = (SdbDeployFp)mndCreateDefaultDnode,
+                     .encodeFp = (SdbEncodeFp)mndDnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndDnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndDnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndDnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndDnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TSDB_MSG_TYPE_CREATE_DNODE, mndProcessCreateDnodeMsg);
+  mndSetMsgHandle(pMnode, TSDB_MSG_TYPE_DROP_DNODE, mndProcessDropDnodeMsg);
+  mndSetMsgHandle(pMnode, TSDB_MSG_TYPE_CONFIG_DNODE, mndProcessConfigDnodeMsg);
+  mndSetMsgHandle(pMnode, TSDB_MSG_TYPE_STATUS, mndProcessStatusMsg);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupDnode(SMnode *pMnode) {}
+
+SDnodeObj *mndAcquireDnode(SMnode *pMnode, int32_t dnodeId) {
+  SSdb *pSdb = pMnode->pSdb;
+  return sdbAcquire(pSdb, SDB_DNODE, &dnodeId);
+}
+
+void mndReleaseDnode(SMnode *pMnode, SDnodeObj *pDnode) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pDnode);
+}

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

@@ -14,11 +14,125 @@
  */
 
 #define _DEFAULT_SOURCE
-#include "os.h"
-#include "mndInt.h"
+#include "mndTrans.h"
 
-int32_t mndInitMnode(SMnode *pMnode) { return 0; }
-void    mndCleanupMnode(SMnode *pMnode) {}
+#define SDB_MNODE_VER 1
 
-void mndGetMnodeEpSetForPeer(SEpSet *epSet, bool redirect) {}
-void mndGetMnodeEpSetForShell(SEpSet *epSet, bool redirect) {}
+static SSdbRaw *mndMnodeActionEncode(SMnodeObj *pMnodeObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_MNODE, SDB_MNODE_VER, sizeof(SMnodeObj));
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pMnodeObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pMnodeObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pMnodeObj->updateTime)
+
+  return pRaw;
+}
+
+static SSdbRow *mndMnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != SDB_MNODE_VER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode mnode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SMnodeObj));
+  SMnodeObj *pMnodeObj = sdbGetRowObj(pRow);
+  if (pMnodeObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pMnodeObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pMnodeObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pMnodeObj->updateTime)
+
+  return pRow;
+}
+
+static void mnodeResetMnode(SMnodeObj *pMnodeObj) {
+  pMnodeObj->role = TAOS_SYNC_STATE_FOLLOWER;
+  pMnodeObj->roleTerm = 0;
+  pMnodeObj->roleTime = 0;
+}
+
+static int32_t mndMnodeActionInsert(SSdb *pSdb, SMnodeObj *pMnodeObj) {
+  mTrace("mnode:%d, perform insert action", pMnodeObj->id);
+  pMnodeObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pMnodeObj->id);
+  if (pMnodeObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("mnode:%d, failed to perform insert action since %s", pMnodeObj->id, terrstr());
+    return -1;
+  }
+
+  mnodeResetMnode(pMnodeObj);
+  return 0;
+}
+
+static int32_t mndMnodeActionDelete(SSdb *pSdb, SMnodeObj *pMnodeObj) {
+  mTrace("mnode:%d, perform delete action", pMnodeObj->id);
+  if (pMnodeObj->pDnode != NULL) {
+    sdbRelease(pSdb, pMnodeObj->pDnode);
+    pMnodeObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndMnodeActionUpdate(SSdb *pSdb, SMnodeObj *pSrcMnode, SMnodeObj *pDstMnode) {
+  mTrace("mnode:%d, perform update action", pSrcMnode->id);
+  pSrcMnode->id = pDstMnode->id;
+  pSrcMnode->createdTime = pDstMnode->createdTime;
+  pSrcMnode->updateTime = pDstMnode->updateTime;
+  return 0;
+}
+
+static int32_t mndCreateDefaultMnode(SMnode *pMnode) {
+  SMnodeObj mnodeObj = {0};
+  mnodeObj.id = 1;
+  mnodeObj.createdTime = taosGetTimestampMs();
+  mnodeObj.updateTime = mnodeObj.createdTime;
+
+  SSdbRaw *pRaw = mndMnodeActionEncode(&mnodeObj);
+  if (pRaw == NULL) return -1;
+  sdbSetRawStatus(pRaw, SDB_STATUS_READY);
+
+  mTrace("mnode:%d, will be created while deploy sdb", mnodeObj.id);
+  return sdbWrite(pMnode->pSdb, pRaw);
+}
+
+static int32_t mndProcessCreateMnodeMsg(SMnode *pMnode, SMnodeMsg *pMsg) { return 0; }
+
+static int32_t mndProcessDropMnodeMsg(SMnode *pMnode, SMnodeMsg *pMsg) { return 0; }
+
+int32_t mndInitMnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_MNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .deployFp = (SdbDeployFp)mndCreateDefaultMnode,
+                     .encodeFp = (SdbEncodeFp)mndMnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndMnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndMnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndMnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndMnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TSDB_MSG_TYPE_CREATE_MNODE, mndProcessCreateMnodeMsg);
+  mndSetMsgHandle(pMnode, TSDB_MSG_TYPE_DROP_MNODE, mndProcessDropMnodeMsg);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupMnode(SMnode *pMnode) {}
+
+bool mndIsMnode(SMnode *pMnode, int32_t dnodeId) {
+  SSdb *pSdb = pMnode->pSdb;
+
+  SMnodeObj *pMnodeObj = sdbAcquire(pSdb, SDB_MNODE, &dnodeId);
+  if (pMnodeObj == NULL) {
+    return false;
+  }
+
+  sdbRelease(pSdb, pMnodeObj);
+  return true;
+}

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

@@ -203,9 +203,9 @@ static void mndSendTelemetryReport() {
     return;
   }
 
-  int64_t clusterId = mndGetClusterId(NULL);
+  int32_t clusterId = 0;
   char    clusterIdStr[20] = {0};
-  snprintf(clusterIdStr, sizeof(clusterIdStr), "%" PRId64, clusterId);
+  snprintf(clusterIdStr, sizeof(clusterIdStr), "%d", clusterId);
 
   SBufferWriter bw = tbufInitWriter(NULL, false);
   mndBeginObject(&bw);

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

@@ -15,13 +15,13 @@
 
 #define _DEFAULT_SOURCE
 #include "mndSync.h"
-#include "tkey.h"
 #include "mndTrans.h"
+#include "tkey.h"
 
 #define SDB_USER_VER 1
 
 static SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, SDB_USER_VER, sizeof(SAcctObj));
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, SDB_USER_VER, sizeof(SUserObj));
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -41,6 +41,7 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
   if (sver != SDB_USER_VER) {
+    mError("failed to decode user since %s", terrstr());
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     return NULL;
   }
@@ -61,15 +62,18 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
 }
 
 static int32_t mndUserActionInsert(SSdb *pSdb, SUserObj *pUser) {
+  mTrace("user:%s, perform insert action", pUser->user);
   pUser->prohibitDbHash = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   if (pUser->prohibitDbHash == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
+    mError("user:%s, failed to perform insert action since %s", pUser->user, terrstr());
     return -1;
   }
 
   pUser->pAcct = sdbAcquire(pSdb, SDB_ACCT, pUser->acct);
   if (pUser->pAcct == NULL) {
     terrno = TSDB_CODE_MND_ACCT_NOT_EXIST;
+    mError("user:%s, failed to perform insert action since %s", pUser->user, terrstr());
     return -1;
   }
 
@@ -77,12 +81,13 @@ static int32_t mndUserActionInsert(SSdb *pSdb, SUserObj *pUser) {
 }
 
 static int32_t mndUserActionDelete(SSdb *pSdb, SUserObj *pUser) {
+  mTrace("user:%s, perform delete action", pUser->user);
   if (pUser->prohibitDbHash) {
     taosHashCleanup(pUser->prohibitDbHash);
     pUser->prohibitDbHash = NULL;
   }
 
-  if (pUser->acct != NULL) {
+  if (pUser->pAcct != NULL) {
     sdbRelease(pSdb, pUser->pAcct);
     pUser->pAcct = NULL;
   }
@@ -91,13 +96,17 @@ static int32_t mndUserActionDelete(SSdb *pSdb, SUserObj *pUser) {
 }
 
 static int32_t mndUserActionUpdate(SSdb *pSdb, SUserObj *pSrcUser, SUserObj *pDstUser) {
-  SUserObj tObj;
-  int32_t  len = (int32_t)((int8_t *)tObj.prohibitDbHash - (int8_t *)&tObj);
-  memcpy(pDstUser, pSrcUser, len);
+  mTrace("user:%s, perform update action", pSrcUser->user);
+  memcpy(pSrcUser->user, pDstUser->user, TSDB_USER_LEN);
+  memcpy(pSrcUser->pass, pDstUser->pass, TSDB_KEY_LEN);
+  memcpy(pSrcUser->acct, pDstUser->acct, TSDB_USER_LEN);
+  pSrcUser->createdTime = pDstUser->createdTime;
+  pSrcUser->updateTime = pDstUser->updateTime;
+  pSrcUser->rootAuth = pDstUser->rootAuth;
   return 0;
 }
 
-static int32_t mndCreateDefaultUser(SSdb *pSdb, char *acct, char *user, char *pass) {
+static int32_t mndCreateDefaultUser(SMnode *pMnode, char *acct, char *user, char *pass) {
   SUserObj userObj = {0};
   tstrncpy(userObj.user, user, TSDB_USER_LEN);
   tstrncpy(userObj.acct, acct, TSDB_USER_LEN);
@@ -113,15 +122,16 @@ static int32_t mndCreateDefaultUser(SSdb *pSdb, char *acct, char *user, char *pa
   if (pRaw == NULL) return -1;
   sdbSetRawStatus(pRaw, SDB_STATUS_READY);
 
-  return sdbWrite(pSdb, pRaw);
+  mTrace("user:%s, will be created while deploy sdb", userObj.user);
+  return sdbWrite(pMnode->pSdb, pRaw);
 }
 
-static int32_t mndCreateDefaultUsers(SSdb *pSdb) {
-  if (mndCreateDefaultUser(pSdb, TSDB_DEFAULT_USER, TSDB_DEFAULT_USER, TSDB_DEFAULT_PASS) != 0) {
+static int32_t mndCreateDefaultUsers(SMnode *pMnode) {
+  if (mndCreateDefaultUser(pMnode, TSDB_DEFAULT_USER, TSDB_DEFAULT_USER, TSDB_DEFAULT_PASS) != 0) {
     return -1;
   }
 
-  if (mndCreateDefaultUser(pSdb, TSDB_DEFAULT_USER, "_" TSDB_DEFAULT_USER, TSDB_DEFAULT_PASS) != 0) {
+  if (mndCreateDefaultUser(pMnode, TSDB_DEFAULT_USER, "_" TSDB_DEFAULT_USER, TSDB_DEFAULT_PASS) != 0) {
     return -1;
   }
 
@@ -196,7 +206,7 @@ static int32_t mndProcessCreateUserMsg(SMnode *pMnode, SMnodeMsg *pMsg) {
     return -1;
   }
 
-  SUserObj *pOperUser = sdbAcquire(pMnode->pSdb, SDB_USER, pMsg->conn.user);
+  SUserObj *pOperUser = sdbAcquire(pMnode->pSdb, SDB_USER, pMsg->user);
   if (pOperUser == NULL) {
     terrno = TSDB_CODE_MND_NO_USER_FROM_CONN;
     mError("user:%s, failed to create since %s", pCreate->user, terrstr());
@@ -229,4 +239,15 @@ int32_t mndInitUser(SMnode *pMnode) {
   return sdbSetTable(pMnode->pSdb, table);
 }
 
-void mndCleanupUser(SMnode *pMnode) {}
+void mndCleanupUser(SMnode *pMnode) {}
+
+SUserObj *mndAcquireUser(SMnode *pMnode, const char *userName) {
+  SSdb *pSdb = pMnode->pSdb;
+  return sdbAcquire(pSdb, SDB_USER, &userName);
+}
+
+void mndReleaseUser(SMnode *pMnode, SUserObj *pUser) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pUser);
+}
+

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

@@ -32,30 +32,6 @@
 #include "mndUser.h"
 #include "mndVgroup.h"
 
-int32_t mndGetDnodeId(SMnode *pMnode) {
-  if (pMnode != NULL) {
-    return pMnode->dnodeId;
-  }
-
-  return -1;
-}
-
-int64_t mndGetClusterId(SMnode *pMnode) {
-  if (pMnode != NULL) {
-    return pMnode->clusterId;
-  }
-
-  return -1;
-}
-
-tmr_h mndGetTimer(SMnode *pMnode) {
-  if (pMnode != NULL) {
-    return pMnode->timer;
-  }
-
-  return NULL;
-}
-
 void mndSendMsgToDnode(SMnode *pMnode, SEpSet *pEpSet, SRpcMsg *pMsg) {
   if (pMnode != NULL && pMnode->sendMsgToDnodeFp != NULL) {
     (*pMnode->sendMsgToDnodeFp)(pMnode->pDnode, pEpSet, pMsg);
@@ -112,6 +88,7 @@ static int32_t mnodeCreateDir(SMnode *pMnode, const char *path) {
 static int32_t mndInitSdb(SMnode *pMnode) {
   SSdbOpt opt = {0};
   opt.path = pMnode->path;
+  opt.pMnode = pMnode;
 
   pMnode->pSdb = sdbInit(&opt);
   if (pMnode->pSdb == NULL) {
@@ -177,11 +154,11 @@ static void mndCleanupSteps(SMnode *pMnode, int32_t pos) {
   if (pMnode->pSteps == NULL) return;
 
   if (pos == -1) {
-    pos = taosArrayGetSize(pMnode->pSteps);
+    pos = taosArrayGetSize(pMnode->pSteps) - 1;
   }
 
   for (int32_t s = pos; s >= 0; s--) {
-    SMnodeStep *pStep = taosArrayGet(pMnode->pSteps, pos);
+    SMnodeStep *pStep = taosArrayGet(pMnode->pSteps, s);
     mDebug("step:%s will cleanup", pStep->name);
     if (pStep->cleanupFp != NULL) {
       (*pStep->cleanupFp)(pMnode);
@@ -189,6 +166,7 @@ static void mndCleanupSteps(SMnode *pMnode, int32_t pos) {
   }
 
   taosArrayClear(pMnode->pSteps);
+  taosArrayDestroy(pMnode->pSteps);
   pMnode->pSteps = NULL;
 }
 
@@ -225,15 +203,22 @@ static int32_t mndSetOptions(SMnode *pMnode, const SMnodeOpt *pOption) {
   pMnode->sendMsgToDnodeFp = pOption->sendMsgToDnodeFp;
   pMnode->sendMsgToMnodeFp = pOption->sendMsgToMnodeFp;
   pMnode->sendRedirectMsgFp = pOption->sendRedirectMsgFp;
+  pMnode->sver = pOption->sver;
+  pMnode->statusInterval = pOption->statusInterval;
+  pMnode->mnodeEqualVnodeNum = pOption->mnodeEqualVnodeNum;
+  pMnode->timezone = strdup(pOption->timezone);
+  pMnode->locale = strdup(pOption->locale);
+  pMnode->charset = strdup(pOption->charset);
 
   if (pMnode->sendMsgToDnodeFp == NULL || pMnode->sendMsgToMnodeFp == NULL || pMnode->sendRedirectMsgFp == NULL ||
-      pMnode->putMsgToApplyMsgFp == NULL) {
-    terrno = TSDB_CODE_MND_APP_ERROR;
+      pMnode->putMsgToApplyMsgFp == NULL || pMnode->dnodeId < 0 || pMnode->clusterId < 0 ||
+      pMnode->statusInterval < 1 || pOption->mnodeEqualVnodeNum < 0) {
+    terrno = TSDB_CODE_MND_INVALID_OPTIONS;
     return -1;
   }
 
-  if (pMnode->dnodeId < 0 || pMnode->clusterId < 0) {
-    terrno = TSDB_CODE_MND_APP_ERROR;
+  if (pMnode->timezone == NULL || pMnode->locale == NULL || pMnode->charset == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
@@ -259,8 +244,9 @@ SMnode *mndOpen(const char *path, const SMnodeOpt *pOption) {
   }
 
   int32_t code = mnodeCreateDir(pMnode, path);
-  if (mnodeCreateDir(pMnode, path) != 0) {
-    mError("failed to open mnode since %s", tstrerror(code));
+  if (code != 0) {
+    code = terrno;
+    mError("failed to open mnode since %s", terrstr());
     mndClose(pMnode);
     terrno = code;
     return NULL;
@@ -268,7 +254,8 @@ SMnode *mndOpen(const char *path, const SMnodeOpt *pOption) {
 
   code = mndSetOptions(pMnode, pOption);
   if (code != 0) {
-    mError("failed to open mnode since %s", tstrerror(code));
+    code = terrno;
+    mError("failed to open mnode since %s", terrstr());
     mndClose(pMnode);
     terrno = code;
     return NULL;
@@ -276,7 +263,8 @@ SMnode *mndOpen(const char *path, const SMnodeOpt *pOption) {
 
   code = mndInitSteps(pMnode);
   if (code != 0) {
-    mError("failed to open mnode since %s", tstrerror(code));
+    code = terrno;
+    mError("failed to open mnode since %s", terrstr());
     mndClose(pMnode);
     terrno = code;
     return NULL;
@@ -284,7 +272,8 @@ SMnode *mndOpen(const char *path, const SMnodeOpt *pOption) {
 
   code = mndExecSteps(pMnode);
   if (code != 0) {
-    mError("failed to open mnode since %s", tstrerror(code));
+    code = terrno;
+    mError("failed to open mnode since %s", terrstr());
     mndClose(pMnode);
     terrno = code;
     return NULL;
@@ -299,6 +288,9 @@ void mndClose(SMnode *pMnode) {
     mDebug("start to close mnode");
     mndCleanupSteps(pMnode, -1);
     tfree(pMnode->path);
+    tfree(pMnode->charset);
+    tfree(pMnode->locale);
+    tfree(pMnode->timezone);
     tfree(pMnode);
     mDebug("mnode is closed");
   }
@@ -335,31 +327,36 @@ SMnodeMsg *mndInitMsg(SMnode *pMnode, SRpcMsg *pRpcMsg) {
   SMnodeMsg *pMsg = taosAllocateQitem(sizeof(SMnodeMsg));
   if (pMsg == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
+    mError("failed to create msg since %s", terrstr());
     return NULL;
   }
 
-  if (rpcGetConnInfo(pRpcMsg->handle, &pMsg->conn) != 0) {
+  SRpcConnInfo connInfo = {0};
+  if (rpcGetConnInfo(pRpcMsg->handle, &connInfo) != 0) {
     mndCleanupMsg(pMsg);
-    mError("can not get user from conn:%p", pMsg->rpcMsg.handle);
     terrno = TSDB_CODE_MND_NO_USER_FROM_CONN;
+    mError("failed to create msg since %s", terrstr());
     return NULL;
   }
+  memcpy(pMsg->user, connInfo.user, TSDB_USER_LEN);
 
+  pMsg->pMnode = pMnode;
   pMsg->rpcMsg = *pRpcMsg;
   pMsg->createdTime = taosGetTimestampSec();
 
+  mTrace("msg:%p, is created", pMsg);
   return pMsg;
 }
 
 void mndCleanupMsg(SMnodeMsg *pMsg) {
-  if (pMsg->pUser != NULL) {
-    sdbRelease(pMsg->pMnode->pSdb, pMsg->pUser);
-  }
-
   taosFreeQitem(pMsg);
+  mTrace("msg:%p, is destroyed", pMsg);
 }
 
-void mndSendRsp(SMnodeMsg *pMsg, int32_t code) {}
+void mndSendRsp(SMnodeMsg *pMsg, int32_t code) {
+  SRpcMsg rpcRsp = {.handle = pMsg->rpcMsg.handle, .code = code};
+  rpcSendResponse(&rpcRsp);
+}
 
 static void mndProcessRpcMsg(SMnodeMsg *pMsg) {
   SMnode *pMnode = pMsg->pMnode;
@@ -368,29 +365,34 @@ static void mndProcessRpcMsg(SMnodeMsg *pMsg) {
   void   *ahandle = pMsg->rpcMsg.ahandle;
   bool    isReq = (msgType % 2 == 1);
 
+  mTrace("msg:%p, app:%p will be processed", pMsg, ahandle);
+
   if (isReq && !mndIsMaster(pMnode)) {
     code = TSDB_CODE_APP_NOT_READY;
+    mDebug("msg:%p, app:%p failed to process since %s", pMsg, ahandle, terrstr());
     goto PROCESS_RPC_END;
   }
 
   if (isReq && pMsg->rpcMsg.pCont == NULL) {
-    mError("msg:%p, app:%p type:%s content is null", pMsg, ahandle, taosMsg[msgType]);
     code = TSDB_CODE_MND_INVALID_MSG_LEN;
+    mError("msg:%p, app:%p failed to process since %s", pMsg, ahandle, terrstr());
     goto PROCESS_RPC_END;
   }
 
   MndMsgFp fp = pMnode->msgFp[msgType];
   if (fp == NULL) {
-    mError("msg:%p, app:%p type:%s not processed", pMsg, ahandle, taosMsg[msgType]);
     code = TSDB_CODE_MSG_NOT_PROCESSED;
+    mError("msg:%p, app:%p failed to process since not handle", pMsg, ahandle);
     goto PROCESS_RPC_END;
   }
 
   code = (*fp)(pMnode, pMsg);
   if (code != 0) {
     code = terrno;
-    mError("msg:%p, app:%p type:%s failed to process since %s", pMsg, ahandle, taosMsg[msgType], terrstr());
+    mError("msg:%p, app:%p failed to process since %s", pMsg, ahandle, terrstr());
     goto PROCESS_RPC_END;
+  } else {
+    mTrace("msg:%p, app:%p is processed", pMsg, ahandle);
   }
 
 PROCESS_RPC_END:
@@ -398,13 +400,13 @@ PROCESS_RPC_END:
     if (code == TSDB_CODE_APP_NOT_READY) {
       mndSendRedirectMsg(pMnode, &pMsg->rpcMsg);
     } else if (code != 0) {
-      SRpcMsg rspMsg = {.handle = pMsg->rpcMsg.handle, .code = code};
-      rpcSendResponse(&rspMsg);
+      SRpcMsg rpcRsp = {.handle = pMsg->rpcMsg.handle, .code = code};
+      rpcSendResponse(&rpcRsp);
     } else {
+      SRpcMsg rpcRsp = {.handle = pMsg->rpcMsg.handle, .contLen = pMsg->contLen, .pCont = pMsg->pCont};
+      rpcSendResponse(&rpcRsp);
     }
   }
-
-  mndCleanupMsg(pMsg);
 }
 
 void mndSetMsgHandle(SMnode *pMnode, int32_t msgType, MndMsgFp fp) {

source/dnode/mnode/sdb/inc/sdbInt.h

@@ -38,8 +38,8 @@ extern "C" {
 
 typedef struct SSdbRaw {
   int8_t  type;
-  int8_t  sver;
   int8_t  status;
+  int8_t  sver;
   int8_t  reserved;
   int32_t dataLen;
   char    pData[];
@@ -53,6 +53,7 @@ typedef struct SSdbRow {
 } SSdbRow;
 
 typedef struct SSdb {
+  SMnode     *pMnode;
   char       *currDir;
   char       *syncDir;
   char       *tmpDir;

source/dnode/mnode/sdb/src/sdb.c

@@ -27,7 +27,7 @@ SSdb *sdbInit(SSdbOpt *pOption) {
   }
 
   char path[PATH_MAX + 100];
-  snprintf(path, PATH_MAX + 100, "%s", pOption->path);
+  snprintf(path, PATH_MAX + 100, "%s%sdata", pOption->path, TD_DIRSEP);
   pSdb->currDir = strdup(path);
   snprintf(path, PATH_MAX + 100, "%s%ssync", pOption->path, TD_DIRSEP);
   pSdb->syncDir = strdup(path);
@@ -40,10 +40,11 @@ SSdb *sdbInit(SSdbOpt *pOption) {
     return NULL;
   }
 
-  for (int32_t i = 0; i < SDB_MAX; ++i) {
+  for (ESdbType i = 0; i < SDB_MAX; ++i) {
     taosInitRWLatch(&pSdb->locks[i]);
   }
 
+  pSdb->pMnode = pOption->pMnode;
   mDebug("sdb init successfully");
   return pSdb;
 }
@@ -68,47 +69,67 @@ void sdbCleanup(SSdb *pSdb) {
     tfree(pSdb->tmpDir);
   }
 
-  for (int32_t i = 0; i < SDB_MAX; ++i) {
+  for (ESdbType i = 0; i < SDB_MAX; ++i) {
     SHashObj *hash = pSdb->hashObjs[i];
-    if (hash != NULL) {
-      taosHashClear(hash);
-      taosHashCleanup(hash);
+    if (hash == NULL) continue;
+
+    SdbDeleteFp deleteFp = pSdb->deleteFps[i];
+    SSdbRow   **ppRow = taosHashIterate(hash, NULL);
+    while (ppRow != NULL) {
+      SSdbRow *pRow = *ppRow;
+      if (pRow == NULL) continue;
+
+      if (deleteFp != NULL) {
+        (*deleteFp)(pSdb, pRow->pObj);
+      }
+      sdbFreeRow(pRow);
+      ppRow = taosHashIterate(hash, ppRow);
     }
-    pSdb->hashObjs[i] = NULL;
   }
 
+  for (ESdbType i = 0; i < SDB_MAX; ++i) {
+    SHashObj *hash = pSdb->hashObjs[i];
+    if (hash == NULL) continue;
+
+    taosHashClear(hash);
+    taosHashCleanup(hash);
+    pSdb->hashObjs[i] = NULL;
+    mTrace("sdb table:%d is cleaned up", i);
+  }
+
+  free(pSdb);
   mDebug("sdb is cleaned up");
 }
 
 int32_t sdbSetTable(SSdb *pSdb, SSdbTable table) {
-  ESdbType sdb = table.sdbType;
-  pSdb->keyTypes[sdb] = table.keyType;
-  pSdb->insertFps[sdb] = table.insertFp;
-  pSdb->updateFps[sdb] = table.updateFp;
-  pSdb->deleteFps[sdb] = table.deleteFp;
-  pSdb->deployFps[sdb] = table.deployFp;
-  pSdb->encodeFps[sdb] = table.encodeFp;
-  pSdb->decodeFps[sdb] = table.decodeFp;
+  ESdbType sdbType = table.sdbType;
+  EKeyType keyType = table.keyType;
+  pSdb->keyTypes[sdbType] = table.keyType;
+  pSdb->insertFps[sdbType] = table.insertFp;
+  pSdb->updateFps[sdbType] = table.updateFp;
+  pSdb->deleteFps[sdbType] = table.deleteFp;
+  pSdb->deployFps[sdbType] = table.deployFp;
+  pSdb->encodeFps[sdbType] = table.encodeFp;
+  pSdb->decodeFps[sdbType] = table.decodeFp;
 
-  for (int32_t i = 0; i < SDB_MAX; ++i) {
-    int32_t type;
-    if (pSdb->keyTypes[i] == SDB_KEY_INT32) {
-      type = TSDB_DATA_TYPE_INT;
-    } else if (pSdb->keyTypes[i] == SDB_KEY_INT64) {
-      type = TSDB_DATA_TYPE_BIGINT;
-    } else {
-      type = TSDB_DATA_TYPE_BINARY;
-    }
-
-    SHashObj *hash = taosHashInit(64, taosGetDefaultHashFunction(type), true, HASH_NO_LOCK);
-    if (hash == NULL) {
-      terrno = TSDB_CODE_OUT_OF_MEMORY;
-      return -1;
-    }
-
-    pSdb->hashObjs[i] = hash;
-    taosInitRWLatch(&pSdb->locks[i]);
+  int32_t hashType = 0;
+  if (keyType == SDB_KEY_INT32) {
+    hashType = TSDB_DATA_TYPE_INT;
+  } else if (keyType == SDB_KEY_INT64) {
+    hashType = TSDB_DATA_TYPE_BIGINT;
+  } else {
+    hashType = TSDB_DATA_TYPE_BINARY;
   }
 
+  SHashObj *hash = taosHashInit(64, taosGetDefaultHashFunction(hashType), true, HASH_NO_LOCK);
+  if (hash == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  pSdb->hashObjs[sdbType] = hash;
+  taosInitRWLatch(&pSdb->locks[sdbType]);
+  mTrace("sdb table:%d is initialized", sdbType);
+
   return 0;
 }

source/dnode/mnode/sdb/src/sdbFile.c

@@ -42,11 +42,11 @@ static int32_t sdbCreateDir(SSdb *pSdb) {
 static int32_t sdbRunDeployFp(SSdb *pSdb) {
   mDebug("start to deploy sdb");
 
-  for (int32_t i = SDB_MAX - 1; i > SDB_START; --i) {
+  for (ESdbType i = SDB_MAX - 1; i > SDB_START; --i) {
     SdbDeployFp fp = pSdb->deployFps[i];
     if (fp == NULL) continue;
 
-    if ((*fp)(pSdb) != 0) {
+    if ((*fp)(pSdb->pMnode) != 0) {
       mError("failed to deploy sdb:%d since %s", i, terrstr());
       return -1;
     }
@@ -150,7 +150,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
     return -1;
   }
 
-  for (int32_t i = SDB_MAX - 1; i > SDB_START; --i) {
+  for (ESdbType i = SDB_MAX - 1; i > SDB_START; --i) {
     SdbEncodeFp encodeFp = pSdb->encodeFps[i];
     if (encodeFp == NULL) continue;
 
@@ -173,6 +173,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
         if (taosWriteFile(fd, pRaw, writeLen) != writeLen) {
           code = TAOS_SYSTEM_ERROR(terrno);
           taosHashCancelIterate(hash, ppRow);
+          free(pRaw);
           break;
         }
 
@@ -180,6 +181,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
         if (taosWriteFile(fd, &cksum, sizeof(int32_t)) != sizeof(int32_t)) {
           code = TAOS_SYSTEM_ERROR(terrno);
           taosHashCancelIterate(hash, ppRow);
+          free(pRaw);
           break;
         }
       } else {
@@ -188,6 +190,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
         break;
       }
 
+      free(pRaw);
       ppRow = taosHashIterate(hash, ppRow);
     }
     taosWUnLockLatch(pLock);

source/dnode/mnode/sdb/src/sdbHash.c

@@ -237,7 +237,15 @@ void *sdbFetch(SSdb *pSdb, ESdbType type, void *pIter, void **ppObj) {
   SRWLatch *pLock = &pSdb->locks[type];
   taosRLockLatch(pLock);
 
-  SSdbRow **ppRow = taosHashIterate(hash, ppRow);
+  if (pIter != NULL) {
+    SSdbRow *pLastRow = *(SSdbRow **)pIter;
+    int32_t  ref = atomic_sub_fetch_32(&pLastRow->refCount, 1);
+    if (ref <= 0 && pLastRow->status == SDB_STATUS_DROPPED) {
+      sdbFreeRow(pLastRow);
+    }
+  }
+
+  SSdbRow **ppRow = taosHashIterate(hash, pIter);
   while (ppRow != NULL) {
     SSdbRow *pRow = *ppRow;
     if (pRow == NULL || pRow->status != SDB_STATUS_READY) {

source/dnode/mnode/sdb/src/sdbRaw.c

@@ -61,6 +61,21 @@ int32_t sdbSetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t val) {
   return 0;
 }
 
+int32_t sdbSetRawInt16(SSdbRaw *pRaw, int32_t dataPos, int16_t val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int16_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *(int16_t *)(pRaw->pData + dataPos) = val;
+  return 0;
+}
+
 int32_t sdbSetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t val) {
   if (pRaw == NULL) {
     terrno = TSDB_CODE_INVALID_PTR;
@@ -146,6 +161,21 @@ int32_t sdbGetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t *val) {
   return 0;
 }
 
+int32_t sdbGetRawInt16(SSdbRaw *pRaw, int32_t dataPos, int16_t *val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int16_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *val = *(int16_t *)(pRaw->pData + dataPos);
+  return 0;
+}
+
 int32_t sdbGetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t *val) {
   if (pRaw == NULL) {
     terrno = TSDB_CODE_INVALID_PTR;

source/dnode/vnode/impl/CMakeLists.txt

@@ -19,5 +19,5 @@ target_link_libraries(
 
 # test
 if(${BUILD_TEST})
-    add_subdirectory(test)
+    # add_subdirectory(test)
 endif(${BUILD_TEST})

source/dnode/vnode/impl/inc/vnodeBufferPool.h

@@ -25,8 +25,9 @@ extern "C" {
 
 typedef struct SVBufPool SVBufPool;
 
-int  vnodeOpenBufPool(SVnode *pVnode);
-void vnodeCloseBufPool(SVnode *pVnode);
+int   vnodeOpenBufPool(SVnode *pVnode);
+void  vnodeCloseBufPool(SVnode *pVnode);
+void *vnodeMalloc(SVnode *pVnode, uint64_t size);
 
 #ifdef __cplusplus
 }

source/dnode/vnode/impl/inc/vnodeCfg.h

@@ -13,8 +13,8 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_VNODE_OPTIONS_H_
-#define _TD_VNODE_OPTIONS_H_
+#ifndef _TD_VNODE_CFG_H_
+#define _TD_VNODE_CFG_H_
 
 #include "vnode.h"
 
@@ -22,13 +22,13 @@
 extern "C" {
 #endif
 
-extern const SVnodeOptions defaultVnodeOptions;
+extern const SVnodeCfg defaultVnodeOptions;
 
-int  vnodeValidateOptions(const SVnodeOptions *);
-void vnodeOptionsCopy(SVnodeOptions *pDest, const SVnodeOptions *pSrc);
+int  vnodeValidateOptions(const SVnodeCfg *);
+void vnodeOptionsCopy(SVnodeCfg *pDest, const SVnodeCfg *pSrc);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif /*_TD_VNODE_OPTIONS_H_*/
+#endif /*_TD_VNODE_CFG_H_*/

source/dnode/vnode/impl/inc/vnodeDef.h

@@ -20,12 +20,14 @@
 #include "sync.h"
 #include "tlockfree.h"
 #include "wal.h"
+#include "tcoding.h"
 
 #include "vnode.h"
 #include "vnodeBufferPool.h"
+#include "vnodeCfg.h"
 #include "vnodeCommit.h"
-#include "vnodeFileSystem.h"
-#include "vnodeOptions.h"
+#include "vnodeFS.h"
+#include "vnodeRequest.h"
 #include "vnodeStateMgr.h"
 #include "vnodeSync.h"
 
@@ -34,16 +36,16 @@ extern "C" {
 #endif
 
 struct SVnode {
-  char*         path;
-  SVnodeOptions options;
-  SVState       state;
-  SVBufPool*    pBufPool;
-  SMeta*        pMeta;
-  STsdb*        pTsdb;
-  STQ*          pTq;
-  SWal*         pWal;
-  SVnodeSync*   pSync;
-  SVnodeFS*     pFs;
+  char*       path;
+  SVnodeCfg   config;
+  SVState     state;
+  SVBufPool*  pBufPool;
+  SMeta*      pMeta;
+  STsdb*      pTsdb;
+  STQ*        pTq;
+  SWal*       pWal;
+  SVnodeSync* pSync;
+  SVnodeFS*   pFs;
 };
 
 #ifdef __cplusplus

source/dnode/vnode/impl/inc/vnodeFS.h

@@ -0,0 +1,47 @@
+/*
+ * 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_VNODE_FS_H_
+#define _TD_VNODE_FS_H_
+
+#include "vnode.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct {
+} SDir;
+
+typedef struct {
+} SFile;
+
+typedef struct SFS {
+  void *pImpl;
+  int (*startEdit)(struct SFS *);
+  int (*endEdit)(struct SFS *);
+} SFS;
+
+typedef struct {
+} SVnodeFS;
+
+int  vnodeOpenFS(SVnode *pVnode);
+void vnodeCloseFS(SVnode *pVnode);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_VNODE_FS_H_*/

source/dnode/vnode/impl/inc/vnodeRequest.h

@@ -16,23 +16,15 @@
 #ifndef _TD_VNODE_REQUEST_H_
 #define _TD_VNODE_REQUEST_H_
 
+#include "vnode.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef struct SVnodeReq SVnodeReq;
-typedef struct SVnodeRsp SVnodeRsp;
-
-typedef enum {
-} EVReqT;
-
-struct SVnodeReq {
-  /* TODO */
-};
-
-struct SVnodeRsp {
-  /* TODO */
-};
+// SVDropTableReq
+int   vnodeBuildDropTableReq(void **buf, const SVDropTableReq *pReq);
+void *vnodeParseDropTableReq(void *buf, SVDropTableReq *pReq);
 
 #ifdef __cplusplus
 }

source/dnode/vnode/impl/inc/vnodeStateMgr.h

@@ -21,6 +21,9 @@ extern "C" {
 #endif
 
 typedef struct {
+  uint64_t processed;
+  uint64_t committed;
+  uint64_t applied;
 } SVState;
 
 #ifdef __cplusplus

source/dnode/vnode/impl/src/vnodeBufferPool.c

@@ -19,9 +19,12 @@
 #define VNODE_BUF_POOL_SHARDS 3
 
 struct SVBufPool {
+  // buffer pool impl
   SList      free;
   SList      incycle;
   SListNode *inuse;
+  // MAF for submodules
+  SMemAllocatorFactory maf;
 };
 
 typedef enum {
@@ -49,6 +52,11 @@ typedef struct {
   SVArenaNode  node;
 } SVArenaAllocator;
 
+typedef struct {
+  SVnode *   pVnode;
+  SListNode *pNode;
+} SVMAWrapper;
+
 typedef struct {
   T_REF_DECLARE()
   uint64_t        capacity;
@@ -59,8 +67,11 @@ typedef struct {
   };
 } SVMemAllocator;
 
-static SListNode *vBufPoolNewNode(uint64_t capacity, EVMemAllocatorT type);
-static void       vBufPoolFreeNode(SListNode *pNode);
+static SListNode *    vBufPoolNewNode(uint64_t capacity, EVMemAllocatorT type);
+static void           vBufPoolFreeNode(SListNode *pNode);
+static SMemAllocator *vBufPoolCreateMA(SMemAllocatorFactory *pmaf);
+static void           vBufPoolDestroyMA(SMemAllocatorFactory *pmaf, SMemAllocator *pma);
+static void *         vBufPoolMalloc(SVMemAllocator *pvma, uint64_t size);
 
 int vnodeOpenBufPool(SVnode *pVnode) {
   uint64_t        capacity;
@@ -74,8 +85,8 @@ int vnodeOpenBufPool(SVnode *pVnode) {
   tdListInit(&(pVnode->pBufPool->free), 0);
   tdListInit(&(pVnode->pBufPool->incycle), 0);
 
-  capacity = pVnode->options.wsize / VNODE_BUF_POOL_SHARDS;
-  if (pVnode->options.isHeapAllocator) {
+  capacity = pVnode->config.wsize / VNODE_BUF_POOL_SHARDS;
+  if (pVnode->config.isHeapAllocator) {
     type = E_V_HEAP_ALLOCATOR;
   }
 
@@ -89,6 +100,10 @@ int vnodeOpenBufPool(SVnode *pVnode) {
     tdListAppendNode(&(pVnode->pBufPool->free), pNode);
   }
 
+  pVnode->pBufPool->maf.impl = pVnode;
+  pVnode->pBufPool->maf.create = vBufPoolCreateMA;
+  pVnode->pBufPool->maf.destroy = vBufPoolDestroyMA;
+
   return 0;
 }
 
@@ -115,6 +130,24 @@ void vnodeCloseBufPool(SVnode *pVnode) {
   }
 }
 
+void *vnodeMalloc(SVnode *pVnode, uint64_t size) {
+  void *ptr;
+
+  if (pVnode->pBufPool->inuse == NULL) {
+    SListNode *pNode;
+    while ((pNode = tdListPopHead(&(pVnode->pBufPool->free))) == NULL) {
+      // todo
+      // tsem_wait();
+      ASSERT(0);
+    }
+
+    pVnode->pBufPool->inuse = pNode;
+  }
+
+  SVMemAllocator *pvma = (SVMemAllocator *)(pVnode->pBufPool->inuse->data);
+  return vBufPoolMalloc(pvma, size);
+}
+
 /* ------------------------ STATIC METHODS ------------------------ */
 static void vArenaAllocatorInit(SVArenaAllocator *pvaa, uint64_t capacity, uint64_t ssize, uint64_t lsize) { /* TODO */
   pvaa->ssize = ssize;
@@ -185,4 +218,103 @@ static void vBufPoolFreeNode(SListNode *pNode) {
   }
 
   free(pNode);
+}
+
+static void *vBufPoolMalloc(SVMemAllocator *pvma, uint64_t size) {
+  void *ptr = NULL;
+
+  if (pvma->type == E_V_ARENA_ALLOCATOR) {
+    SVArenaAllocator *pvaa = &(pvma->vaa);
+
+    if (POINTER_DISTANCE(pvaa->inuse->ptr, pvaa->inuse->data) + size > pvaa->inuse->size) {
+      SVArenaNode *pNode = (SVArenaNode *)malloc(sizeof(*pNode) + MAX(size, pvaa->ssize));
+      if (pNode == NULL) {
+        // TODO: handle error
+        return NULL;
+      }
+
+      pNode->prev = pvaa->inuse;
+      pNode->size = MAX(size, pvaa->ssize);
+      pNode->ptr = pNode->data;
+
+      pvaa->inuse = pNode;
+    }
+
+    ptr = pvaa->inuse->ptr;
+    pvaa->inuse->ptr = POINTER_SHIFT(ptr, size);
+  } else if (pvma->type == E_V_HEAP_ALLOCATOR) {
+    /* TODO */
+  }
+
+  return ptr;
+}
+
+static SMemAllocator *vBufPoolCreateMA(SMemAllocatorFactory *pmaf) {
+  SVnode *        pVnode;
+  SMemAllocator * pma;
+  SVMemAllocator *pvma;
+  SVMAWrapper *   pvmaw;
+
+  pVnode = (SVnode *)(pmaf->impl);
+  pma = (SMemAllocator *)calloc(1, sizeof(*pma) + sizeof(SVMAWrapper));
+  if (pma == NULL) {
+    // TODO: handle error
+    return NULL;
+  }
+  pvmaw = (SVMAWrapper *)POINTER_SHIFT(pma, sizeof(*pma));
+
+  // No allocator used currently
+  if (pVnode->pBufPool->inuse == NULL) {
+    while (listNEles(&(pVnode->pBufPool->free)) == 0) {
+      // TODO: wait until all released ro kill query
+      // tsem_wait();
+      ASSERT(0);
+    }
+
+    pVnode->pBufPool->inuse = tdListPopHead(&(pVnode->pBufPool->free));
+    pvma = (SVMemAllocator *)(pVnode->pBufPool->inuse->data);
+    T_REF_INIT_VAL(pvma, 1);
+  } else {
+    pvma = (SVMemAllocator *)(pVnode->pBufPool->inuse->data);
+  }
+
+  T_REF_INC(pvma);
+
+  pvmaw->pVnode = pVnode;
+  pvmaw->pNode = pVnode->pBufPool->inuse;
+
+  pma->impl = pvmaw;
+  pma->malloc = NULL;
+  pma->calloc = NULL;  /* TODO */
+  pma->realloc = NULL; /* TODO */
+  pma->free = NULL;    /* TODO */
+  pma->usage = NULL;   /* TODO */
+
+  return pma;
+}
+
+static void vBufPoolDestroyMA(SMemAllocatorFactory *pmaf, SMemAllocator *pma) { /* TODO */
+  SVnode *        pVnode = (SVnode *)(pmaf->impl);
+  SListNode *     pNode = ((SVMAWrapper *)(pma->impl))->pNode;
+  SVMemAllocator *pvma = (SVMemAllocator *)(pNode->data);
+
+  if (T_REF_DEC(pvma) == 0) {
+    if (pvma->type == E_V_ARENA_ALLOCATOR) {
+      SVArenaAllocator *pvaa = &(pvma->vaa);
+      while (pvaa->inuse != &(pvaa->node)) {
+        SVArenaNode *pNode = pvaa->inuse;
+        pvaa->inuse = pNode->prev;
+        /* code */
+      }
+
+      pvaa->inuse->ptr = pvaa->inuse->data;
+    } else if (pvma->type == E_V_HEAP_ALLOCATOR) {
+    } else {
+      ASSERT(0);
+    }
+
+    // Move node from incycle to free
+    tdListAppendNode(&(pVnode->pBufPool->free), tdListPopNode(&(pVnode->pBufPool->incycle), pNode));
+    // tsem_post(); todo: sem_post
+  }
 }

source/dnode/vnode/impl/src/vnodeCfg.c

@@ -15,20 +15,20 @@
 
 #include "vnodeDef.h"
 
-const SVnodeOptions defaultVnodeOptions = {0}; /* TODO */
+const SVnodeCfg defaultVnodeOptions = {0}; /* TODO */
 
-void vnodeOptionsInit(SVnodeOptions *pVnodeOptions) { /* TODO */
+void vnodeOptionsInit(SVnodeCfg *pVnodeOptions) { /* TODO */
   vnodeOptionsCopy(pVnodeOptions, &defaultVnodeOptions);
 }
 
-void vnodeOptionsClear(SVnodeOptions *pVnodeOptions) { /* TODO */
+void vnodeOptionsClear(SVnodeCfg *pVnodeOptions) { /* TODO */
 }
 
-int vnodeValidateOptions(const SVnodeOptions *pVnodeOptions) {
+int vnodeValidateOptions(const SVnodeCfg *pVnodeOptions) {
   // TODO
   return 0;
 }
 
-void vnodeOptionsCopy(SVnodeOptions *pDest, const SVnodeOptions *pSrc) {
-  memcpy((void *)pDest, (void *)pSrc, sizeof(SVnodeOptions));
+void vnodeOptionsCopy(SVnodeCfg *pDest, const SVnodeCfg *pSrc) {
+  memcpy((void *)pDest, (void *)pSrc, sizeof(SVnodeCfg));
 }

source/dnode/vnode/impl/src/vnodeMain.c

@@ -15,27 +15,40 @@
 
 #include "vnodeDef.h"
 
-static SVnode *vnodeNew(const char *path, const SVnodeOptions *pVnodeOptions);
+static SVnode *vnodeNew(const char *path, const SVnodeCfg *pVnodeCfg);
 static void    vnodeFree(SVnode *pVnode);
 static int     vnodeOpenImpl(SVnode *pVnode);
 static void    vnodeCloseImpl(SVnode *pVnode);
 
-SVnode *vnodeOpen(const char *path, const SVnodeOptions *pVnodeOptions) {
+int vnodeInit() {
+  // TODO
+  if (walInit() < 0) {
+    return -1;
+  }
+
+  return 0;
+}
+
+void vnodeClear() {
+  walCleanUp();
+}
+
+SVnode *vnodeOpen(const char *path, const SVnodeCfg *pVnodeCfg) {
   SVnode *pVnode = NULL;
 
   // Set default options
-  if (pVnodeOptions == NULL) {
-    pVnodeOptions = &defaultVnodeOptions;
+  if (pVnodeCfg == NULL) {
+    pVnodeCfg = &defaultVnodeOptions;
   }
 
   // Validate options
-  if (vnodeValidateOptions(pVnodeOptions) < 0) {
+  if (vnodeValidateOptions(pVnodeCfg) < 0) {
     // TODO
     return NULL;
   }
 
   // Create the handle
-  pVnode = vnodeNew(path, pVnodeOptions);
+  pVnode = vnodeNew(path, pVnodeCfg);
   if (pVnode == NULL) {
     // TODO: handle error
     return NULL;
@@ -62,7 +75,7 @@ void vnodeClose(SVnode *pVnode) {
 void vnodeDestroy(const char *path) { taosRemoveDir(path); }
 
 /* ------------------------ STATIC METHODS ------------------------ */
-static SVnode *vnodeNew(const char *path, const SVnodeOptions *pVnodeOptions) {
+static SVnode *vnodeNew(const char *path, const SVnodeCfg *pVnodeCfg) {
   SVnode *pVnode = NULL;
 
   pVnode = (SVnode *)calloc(1, sizeof(*pVnode));
@@ -72,7 +85,7 @@ static SVnode *vnodeNew(const char *path, const SVnodeOptions *pVnodeOptions) {
   }
 
   pVnode->path = strdup(path);
-  vnodeOptionsCopy(&(pVnode->options), pVnodeOptions);
+  vnodeOptionsCopy(&(pVnode->config), pVnodeCfg);
 
   return pVnode;
 }
@@ -94,7 +107,7 @@ static int vnodeOpenImpl(SVnode *pVnode) {
 
   // Open meta
   sprintf(dir, "%s/meta", pVnode->path);
-  pVnode->pMeta = metaOpen(dir, &(pVnode->options.metaOptions));
+  pVnode->pMeta = metaOpen(dir, &(pVnode->config.metaCfg));
   if (pVnode->pMeta == NULL) {
     // TODO: handle error
     return -1;
@@ -102,23 +115,23 @@ static int vnodeOpenImpl(SVnode *pVnode) {
 
   // Open tsdb
   sprintf(dir, "%s/tsdb", pVnode->path);
-  pVnode->pTsdb = tsdbOpen(dir, &(pVnode->options.tsdbOptions));
+  pVnode->pTsdb = tsdbOpen(dir, &(pVnode->config.tsdbCfg));
   if (pVnode->pTsdb == NULL) {
     // TODO: handle error
     return -1;
   }
 
   // TODO: Open TQ
-  sprintf(dir, "%s/wal", pVnode->path);
-  // pVnode->pTq = tqOpen(dir, NULL /* TODO */);
-  // if (pVnode->pTq == NULL) {
-  //   // TODO: handle error
-  //   return -1;
-  // }
+  sprintf(dir, "%s/tq", pVnode->path);
+  pVnode->pTq = tqOpen(dir, &(pVnode->config.tqCfg), NULL, NULL);
+  if (pVnode->pTq == NULL) {
+    // TODO: handle error
+    return -1;
+  }
 
   // Open WAL
   sprintf(dir, "%s/wal", pVnode->path);
-  pVnode->pWal = walOpen(dir, NULL /* TODO */);
+  pVnode->pWal = walOpen(dir, &(pVnode->config.walCfg));
   if (pVnode->pWal == NULL) {
     // TODO: handle error
     return -1;

source/dnode/vnode/impl/src/vnodeRequest.c

@@ -11,4 +11,105 @@
  *
  * 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/>.
- */
+ */
+
+#include "vnodeDef.h"
+
+static int   vnodeBuildCreateTableReq(void **buf, const SVCreateTableReq *pReq);
+static void *vnodeParseCreateTableReq(void *buf, SVCreateTableReq *pReq);
+
+int vnodeBuildReq(void **buf, const SVnodeReq *pReq, uint8_t type) {
+  int tsize = 0;
+
+  tsize += taosEncodeFixedU64(buf, pReq->ver);
+  switch (type) {
+    case TSDB_MSG_TYPE_CREATE_TABLE:
+      tsize += vnodeBuildCreateTableReq(buf, &(pReq->ctReq));
+      /* code */
+      break;
+
+    default:
+      break;
+  }
+  /* TODO */
+  return tsize;
+}
+
+void *vnodeParseReq(void *buf, SVnodeReq *pReq, uint8_t type) {
+  buf = taosDecodeFixedU64(buf, &(pReq->ver));
+
+  switch (type) {
+    case TSDB_MSG_TYPE_CREATE_TABLE:
+      buf = vnodeParseCreateTableReq(buf, &(pReq->ctReq));
+      break;
+
+    default:
+      break;
+  }
+
+  // TODO
+  return buf;
+}
+
+static int vnodeBuildCreateTableReq(void **buf, const SVCreateTableReq *pReq) {
+  int tsize = 0;
+
+  tsize += taosEncodeString(buf, pReq->name);
+  tsize += taosEncodeFixedU32(buf, pReq->ttl);
+  tsize += taosEncodeFixedU32(buf, pReq->keep);
+  tsize += taosEncodeFixedU8(buf, pReq->type);
+
+  switch (pReq->type) {
+    case META_SUPER_TABLE:
+      tsize += taosEncodeFixedU64(buf, pReq->stbCfg.suid);
+      tsize += tdEncodeSchema(buf, pReq->stbCfg.pSchema);
+      tsize += tdEncodeSchema(buf, pReq->stbCfg.pTagSchema);
+      break;
+    case META_CHILD_TABLE:
+      tsize += taosEncodeFixedU64(buf, pReq->ctbCfg.suid);
+      tsize += tdEncodeKVRow(buf, pReq->ctbCfg.pTag);
+      break;
+    case META_NORMAL_TABLE:
+      tsize += tdEncodeSchema(buf, pReq->ntbCfg.pSchema);
+      break;
+    default:
+      break;
+  }
+
+  return tsize;
+}
+
+static void *vnodeParseCreateTableReq(void *buf, SVCreateTableReq *pReq) {
+  buf = taosDecodeString(buf, &(pReq->name));
+  buf = taosDecodeFixedU32(buf, &(pReq->ttl));
+  buf = taosDecodeFixedU32(buf, &(pReq->keep));
+  buf = taosDecodeFixedU8(buf, &(pReq->type));
+
+  switch (pReq->type) {
+    case META_SUPER_TABLE:
+      buf = taosDecodeFixedU64(buf, &(pReq->stbCfg.suid));
+      buf = tdDecodeSchema(buf, &(pReq->stbCfg.pSchema));
+      buf = tdDecodeSchema(buf, &(pReq->stbCfg.pTagSchema));
+      break;
+    case META_CHILD_TABLE:
+      buf = taosDecodeFixedU64(buf, &(pReq->ctbCfg.suid));
+      buf = tdDecodeKVRow(buf, &(pReq->ctbCfg.pTag));
+      break;
+    case META_NORMAL_TABLE:
+      buf = tdDecodeSchema(buf, &(pReq->ntbCfg.pSchema));
+      break;
+    default:
+      break;
+  }
+
+  return buf;
+}
+
+int vnodeBuildDropTableReq(void **buf, const SVDropTableReq *pReq) {
+  // TODO
+  return 0;
+}
+
+void *vnodeParseDropTableReq(void *buf, SVDropTableReq *pReq) {
+  // TODO
+}

source/dnode/vnode/impl/src/vnodeWrite.c

@@ -16,51 +16,84 @@
 #include "vnodeDef.h"
 
 int vnodeProcessWMsgs(SVnode *pVnode, SArray *pMsgs) {
-  /* TODO */
-  return 0;
-}
+  SRpcMsg *  pMsg;
+  SVnodeReq *pVnodeReq;
 
-int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
-#if 0
-  int      reqType;        /* TODO */
-  size_t   reqSize;        /* TODO */
-  uint64_t reqVersion = 0; /* TODO */
-  int      code = 0;
+  for (int i = 0; i < taosArrayGetSize(pMsgs); i++) {
+    pMsg = *(SRpcMsg **)taosArrayGet(pMsgs, i);
 
-  // Copy the request to vnode buffer
-  void *pReq = mMalloc(pVnode->inuse, reqSize);
-  if (pReq == NULL) {
-    // TODO: handle error
-  }
+    // ser request version
+    void *   pBuf = pMsg->pCont;
+    uint64_t ver = pVnode->state.processed++;
+    taosEncodeFixedU64(&pBuf, ver);
 
-  memcpy(pReq, pMsg, reqSize);
-
-  // Push the request to TQ so consumers can consume
-  tqPushMsg(pVnode->pTq, pReq, 0);
-
-  // Process the request
-  switch (reqType) {
-    case TSDB_MSG_TYPE_CREATE_TABLE:
-      code = metaCreateTable(pVnode->pMeta, NULL /* TODO */);
-      break;
-    case TSDB_MSG_TYPE_DROP_TABLE:
-      code = metaDropTable(pVnode->pMeta, 0 /* TODO */);
-      break;
-    case TSDB_MSG_TYPE_SUBMIT:
-      /* TODO */
-      break;
-    default:
-      break;
-  }
-
-  if (vnodeShouldCommit(pVnode)) {
-    if (vnodeAsyncCommit(pVnode) < 0) {
+    if (walWrite(pVnode->pWal, ver, pMsg->msgType, pMsg->pCont, pMsg->contLen) < 0) {
       // TODO: handle error
     }
   }
 
-  return code;
-#endif
+  walFsync(pVnode->pWal, false);
+
+  // Apply each request now
+  for (int i = 0; i < taosArrayGetSize(pMsgs); i++) {
+    pMsg = *(SRpcMsg **)taosArrayGet(pMsgs, i);
+    SVnodeReq vReq;
+
+    // Apply the request
+    {
+      void *ptr = vnodeMalloc(pVnode, pMsg->contLen);
+      if (ptr == NULL) {
+        // TODO: handle error
+      }
+
+      // TODO: copy here need to be extended
+      memcpy(ptr, pMsg->pCont, pMsg->contLen);
+
+      // todo: change the interface here
+      uint64_t ver;
+      taosDecodeFixedU64(pMsg->pCont, &ver);
+      if (tqPushMsg(pVnode->pTq, ptr, ver) < 0) {
+        // TODO: handle error
+      }
+
+      vnodeParseReq(pMsg->pCont, &vReq, pMsg->msgType);
+
+      switch (pMsg->msgType) {
+        case TSDB_MSG_TYPE_CREATE_TABLE:
+          if (metaCreateTable(pVnode->pMeta, &(vReq.ctReq)) < 0) {
+            // TODO: handle error
+          }
+
+          // TODO: maybe need to clear the requst struct
+          break;
+        case TSDB_MSG_TYPE_DROP_TABLE:
+          if (metaDropTable(pVnode->pMeta, vReq.dtReq.uid) < 0) {
+            // TODO: handle error
+          }
+          break;
+        case TSDB_MSG_TYPE_SUBMIT:
+          /* code */
+          break;
+        default:
+          break;
+      }
+
+      pVnode->state.applied = ver;
+    }
+
+    // Check if it needs to commit
+    if (vnodeShouldCommit(pVnode)) {
+      if (vnodeAsyncCommit(pVnode) < 0) {
+        // TODO: handle error
+      }
+    }
+  }
+
+  return 0;
+}
+
+int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+  // TODO
   return 0;
 }
 

source/dnode/vnode/impl/test/vnodeApiTests.cpp

@@ -3,11 +3,173 @@
 
 #include "vnode.h"
 
+static STSchema *createBasicSchema() {
+  STSchemaBuilder sb;
+  STSchema *      pSchema = NULL;
+
+  tdInitTSchemaBuilder(&sb, 0);
+
+  tdAddColToSchema(&sb, TSDB_DATA_TYPE_TIMESTAMP, 0, 0);
+  for (int i = 1; i < 10; i++) {
+    tdAddColToSchema(&sb, TSDB_DATA_TYPE_INT, i, 0);
+  }
+
+  pSchema = tdGetSchemaFromBuilder(&sb);
+
+  tdDestroyTSchemaBuilder(&sb);
+
+  return pSchema;
+}
+
+static STSchema *createBasicTagSchema() {
+  STSchemaBuilder sb;
+  STSchema *      pSchema = NULL;
+
+  tdInitTSchemaBuilder(&sb, 0);
+
+  tdAddColToSchema(&sb, TSDB_DATA_TYPE_TIMESTAMP, 0, 0);
+  for (int i = 10; i < 12; i++) {
+    tdAddColToSchema(&sb, TSDB_DATA_TYPE_BINARY, i, 20);
+  }
+
+  pSchema = tdGetSchemaFromBuilder(&sb);
+
+  tdDestroyTSchemaBuilder(&sb);
+
+  return pSchema;
+}
+
+static SKVRow createBasicTag() {
+  SKVRowBuilder rb;
+  SKVRow        pTag;
+
+  tdInitKVRowBuilder(&rb);
+
+  for (int i = 10; i < 12; i++) {
+    void *pVal = malloc(sizeof(VarDataLenT) + strlen("foo"));
+    varDataLen(pVal) = strlen("foo");
+    memcpy(varDataVal(pVal), "foo", strlen("foo"));
+
+    tdAddColToKVRow(&rb, i, TSDB_DATA_TYPE_BINARY, pVal);
+    free(pVal);
+  }
+
+  pTag = tdGetKVRowFromBuilder(&rb);
+  tdDestroyKVRowBuilder(&rb);
+
+  return pTag;
+}
+
+#if 0
+TEST(vnodeApiTest, test_create_table_encode_and_decode_function) {
+  tb_uid_t  suid = 1638166374163;
+  STSchema *pSchema = createBasicSchema();
+  STSchema *pTagSchema = createBasicTagSchema();
+  char      tbname[128] = "st";
+  char *    buffer = new char[1024];
+  void *    pBuf = (void *)buffer;
+  SVnodeReq vCreateSTbReq = VNODE_INIT_CREATE_STB_REQ(tbname, UINT32_MAX, UINT32_MAX, suid, pSchema, pTagSchema);
+
+  vnodeBuildReq(&pBuf, &vCreateSTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
+
+  SVnodeReq decoded_req;
+
+  vnodeParseReq(buffer, &decoded_req, TSDB_MSG_TYPE_CREATE_TABLE);
+
+  int k = 10;
+}
+#endif
+
 TEST(vnodeApiTest, vnodeOpen_vnodeClose_test) {
+  GTEST_ASSERT_GE(vnodeInit(), 0);
+
   // Create and open a vnode
   SVnode *pVnode = vnodeOpen("vnode1", NULL);
   ASSERT_NE(pVnode, nullptr);
 
+  tb_uid_t suid = 1638166374163;
+  {
+    // Create a super table
+    STSchema *pSchema = createBasicSchema();
+    STSchema *pTagSchema = createBasicTagSchema();
+    char      tbname[128] = "st";
+
+    SArray *  pMsgs = (SArray *)taosArrayInit(1, sizeof(SRpcMsg *));
+    SVnodeReq vCreateSTbReq = VNODE_INIT_CREATE_STB_REQ(tbname, UINT32_MAX, UINT32_MAX, suid, pSchema, pTagSchema);
+
+    int      zs = vnodeBuildReq(NULL, &vCreateSTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
+    SRpcMsg *pMsg = (SRpcMsg *)malloc(sizeof(SRpcMsg) + zs);
+    pMsg->msgType = TSDB_MSG_TYPE_CREATE_TABLE;
+    pMsg->contLen = zs;
+    pMsg->pCont = POINTER_SHIFT(pMsg, sizeof(SRpcMsg));
+
+    void *pBuf = pMsg->pCont;
+
+    vnodeBuildReq(&pBuf, &vCreateSTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
+    META_CLEAR_TB_CFG(&vCreateSTbReq);
+
+    taosArrayPush(pMsgs, &(pMsg));
+
+    vnodeProcessWMsgs(pVnode, pMsgs);
+
+    free(pMsg);
+    taosArrayDestroy(pMsgs);
+    tdFreeSchema(pSchema);
+    tdFreeSchema(pTagSchema);
+  }
+
+  {
+    // Create some child tables
+    int ntables = 100000;
+    int batch = 10;
+    for (int i = 0; i < ntables / batch; i++) {
+      SArray *pMsgs = (SArray *)taosArrayInit(batch, sizeof(SRpcMsg *));
+      for (int j = 0; j < batch; j++) {
+        SKVRow pTag = createBasicTag();
+        char   tbname[128];
+        sprintf(tbname, "tb%d", i * batch + j);
+        SVnodeReq vCreateCTbReq = VNODE_INIT_CREATE_CTB_REQ(tbname, UINT32_MAX, UINT32_MAX, suid, pTag);
+
+        int      tz = vnodeBuildReq(NULL, &vCreateCTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
+        SRpcMsg *pMsg = (SRpcMsg *)malloc(sizeof(SRpcMsg) + tz);
+        pMsg->msgType = TSDB_MSG_TYPE_CREATE_TABLE;
+        pMsg->contLen = tz;
+        pMsg->pCont = POINTER_SHIFT(pMsg, sizeof(*pMsg));
+        void *pBuf = pMsg->pCont;
+
+        vnodeBuildReq(&pBuf, &vCreateCTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
+        META_CLEAR_TB_CFG(&vCreateCTbReq);
+        free(pTag);
+
+        taosArrayPush(pMsgs, &(pMsg));
+      }
+
+      vnodeProcessWMsgs(pVnode, pMsgs);
+
+      for (int j = 0; j < batch; j++) {
+        SRpcMsg *pMsg = *(SRpcMsg **)taosArrayPop(pMsgs);
+        free(pMsg);
+      }
+
+      taosArrayDestroy(pMsgs);
+
+      // std::cout << "the " << i << "th batch is created" << std::endl;
+    }
+  }
+
   // Close the vnode
   vnodeClose(pVnode);
+
+  vnodeClear();
+}
+
+TEST(vnodeApiTest, DISABLED_vnode_process_create_table) {
+  STSchema *       pSchema = NULL;
+  STSchema *       pTagSchema = NULL;
+  char             stname[15];
+  SVCreateTableReq pReq = META_INIT_STB_CFG(stname, UINT32_MAX, UINT32_MAX, 0, pSchema, pTagSchema);
+
+  int k = 10;
+
+  META_CLEAR_TB_CFG(pReq);
 }

source/dnode/vnode/meta/inc/metaCfg.h

@@ -13,8 +13,8 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_META_OPTIONS_H_
-#define _TD_META_OPTIONS_H_
+#ifndef _TD_META_CFG_H_
+#define _TD_META_CFG_H_
 
 #include "meta.h"
 
@@ -22,13 +22,13 @@
 extern "C" {
 #endif
 
-extern const SMetaOptions defaultMetaOptions;
+extern const SMetaCfg defaultMetaOptions;
 
-int  metaValidateOptions(const SMetaOptions *);
-void metaOptionsCopy(SMetaOptions *pDest, const SMetaOptions *pSrc);
+int  metaValidateOptions(const SMetaCfg *);
+void metaOptionsCopy(SMetaCfg *pDest, const SMetaCfg *pSrc);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif /*_TD_META_OPTIONS_H_*/
+#endif /*_TD_META_CFG_H_*/

source/dnode/vnode/meta/inc/metaDB.h

@@ -34,7 +34,7 @@ typedef struct {
 
 int  metaOpenDB(SMeta *pMeta);
 void metaCloseDB(SMeta *pMeta);
-int  metaSaveTableToDB(SMeta *pMeta, const STbOptions *pTbOptions);
+int  metaSaveTableToDB(SMeta *pMeta, const STbCfg *pTbOptions);
 int  metaRemoveTableFromDb(SMeta *pMeta, tb_uid_t uid);
 
 #ifdef __cplusplus

source/dnode/vnode/meta/inc/metaDef.h

@@ -20,10 +20,10 @@
 
 #include "meta.h"
 #include "metaCache.h"
+#include "metaCfg.h"
 #include "metaDB.h"
 #include "metaIdx.h"
-#include "metaOptions.h"
-#include "metaTbOptions.h"
+#include "metaTbCfg.h"
 #include "metaTbTag.h"
 #include "metaTbUid.h"
 
@@ -33,7 +33,7 @@ extern "C" {
 
 struct SMeta {
   char*                 path;
-  SMetaOptions          options;
+  SMetaCfg              options;
   meta_db_t*            pDB;
   meta_index_t*         pIdx;
   meta_cache_t*         pCache;

source/dnode/vnode/meta/inc/metaIdx.h

@@ -28,7 +28,7 @@ typedef rocksdb_t meta_index_t;
 
 int  metaOpenIdx(SMeta *pMeta);
 void metaCloseIdx(SMeta *pMeta);
-int  metaSaveTableToIdx(SMeta *pMeta, const STbOptions *pTbOptions);
+int  metaSaveTableToIdx(SMeta *pMeta, const STbCfg *pTbOptions);
 int  metaRemoveTableFromIdx(SMeta *pMeta, tb_uid_t uid);
 
 #ifdef __cplusplus

source/dnode/vnode/meta/inc/metaTbCfg.h

@@ -13,8 +13,8 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_META_TABLE_OPTIONS_H_
-#define _TD_META_TABLE_OPTIONS_H_
+#ifndef _TD_META_TABLE_CFG_H_
+#define _TD_META_TABLE_CFG_H_
 
 #include "meta.h"
 
@@ -22,11 +22,15 @@
 extern "C" {
 #endif
 
-int    metaValidateTbOptions(SMeta *pMeta, const STbOptions *);
-size_t metaEncodeTbObjFromTbOptions(const STbOptions *, void *pBuf, size_t bsize);
+#define META_SUPER_TABLE 0
+#define META_CHILD_TABLE 1
+#define META_NORMAL_TABLE 2
+
+int    metaValidateTbOptions(SMeta *pMeta, const STbCfg *);
+size_t metaEncodeTbObjFromTbOptions(const STbCfg *, void *pBuf, size_t bsize);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif /*_TD_META_TABLE_OPTIONS_H_*/
+#endif /*_TD_META_TABLE_CFG_H_*/

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

@@ -18,8 +18,8 @@
 
 int metaOpenCache(SMeta *pMeta) {
   // TODO
-  if (pMeta->options.lruCacheSize) {
-    pMeta->pCache = rocksdb_cache_create_lru(pMeta->options.lruCacheSize);
+  if (pMeta->options.lruSize) {
+    pMeta->pCache = rocksdb_cache_create_lru(pMeta->options.lruSize);
     if (pMeta->pCache == NULL) {
       // TODO: handle error
       return -1;

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

@@ -15,20 +15,20 @@
 
 #include "metaDef.h"
 
-const SMetaOptions defaultMetaOptions = {.lruCacheSize = 0};
+const SMetaCfg defaultMetaOptions = {.lruSize = 0};
 
 /* ------------------------ EXPOSED METHODS ------------------------ */
-void metaOptionsInit(SMetaOptions *pMetaOptions) { metaOptionsCopy(pMetaOptions, &defaultMetaOptions); }
+void metaOptionsInit(SMetaCfg *pMetaOptions) { metaOptionsCopy(pMetaOptions, &defaultMetaOptions); }
 
-void metaOptionsClear(SMetaOptions *pMetaOptions) {
+void metaOptionsClear(SMetaCfg *pMetaOptions) {
   // TODO
 }
 
-int metaValidateOptions(const SMetaOptions *pMetaOptions) {
+int metaValidateOptions(const SMetaCfg *pMetaOptions) {
   // TODO
   return 0;
 }
 
-void metaOptionsCopy(SMetaOptions *pDest, const SMetaOptions *pSrc) { memcpy(pDest, pSrc, sizeof(*pSrc)); }
+void metaOptionsCopy(SMetaCfg *pDest, const SMetaCfg *pSrc) { memcpy(pDest, pSrc, sizeof(*pSrc)); }
 
 /* ------------------------ STATIC METHODS ------------------------ */

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

@@ -92,7 +92,7 @@ void metaCloseDB(SMeta *pMeta) {
   }
 }
 
-int metaSaveTableToDB(SMeta *pMeta, const STbOptions *pTbOptions) {
+int metaSaveTableToDB(SMeta *pMeta, const STbCfg *pTbOptions) {
   tb_uid_t uid;
   char *   err = NULL;
   size_t   size;
@@ -102,13 +102,14 @@ int metaSaveTableToDB(SMeta *pMeta, const STbOptions *pTbOptions) {
 
   // Generate a uid for child and normal table
   if (pTbOptions->type == META_SUPER_TABLE) {
-    uid = pTbOptions->stbOptions.uid;
+    uid = pTbOptions->stbCfg.suid;
   } else {
     uid = metaGenerateUid(pMeta);
   }
 
   // Save tbname -> uid to tbnameDB
   rocksdb_put(pMeta->pDB->nameDb, wopt, pTbOptions->name, strlen(pTbOptions->name), (char *)(&uid), sizeof(uid), &err);
+  rocksdb_writeoptions_disable_WAL(wopt, 1);
 
   // Save uid -> tb_obj to tbDB
   size = metaEncodeTbObjFromTbOptions(pTbOptions, pBuf, 1024);
@@ -117,22 +118,22 @@ int metaSaveTableToDB(SMeta *pMeta, const STbOptions *pTbOptions) {
   switch (pTbOptions->type) {
     case META_NORMAL_TABLE:
       // save schemaDB
-      metaSaveSchemaDB(pMeta, uid, pTbOptions->ntbOptions.pSchame);
+      metaSaveSchemaDB(pMeta, uid, pTbOptions->ntbCfg.pSchema);
       break;
     case META_SUPER_TABLE:
       // save schemaDB
-      metaSaveSchemaDB(pMeta, uid, pTbOptions->stbOptions.pSchema);
+      metaSaveSchemaDB(pMeta, uid, pTbOptions->stbCfg.pSchema);
 
       // save mapDB (really need?)
       rocksdb_put(pMeta->pDB->mapDb, wopt, (char *)(&uid), sizeof(uid), "", 0, &err);
       break;
     case META_CHILD_TABLE:
       // save tagDB
-      rocksdb_put(pMeta->pDB->tagDb, wopt, (char *)(&uid), sizeof(uid), pTbOptions->ctbOptions.tags,
-                  kvRowLen(pTbOptions->ctbOptions.tags), &err);
+      rocksdb_put(pMeta->pDB->tagDb, wopt, (char *)(&uid), sizeof(uid), pTbOptions->ctbCfg.pTag,
+                  kvRowLen(pTbOptions->ctbCfg.pTag), &err);
 
       // save mapDB
-      metaSaveMapDB(pMeta, pTbOptions->ctbOptions.suid, uid);
+      metaSaveMapDB(pMeta, pTbOptions->ctbCfg.suid, uid);
       break;
     default:
       ASSERT(0);
@@ -157,6 +158,7 @@ static void metaSaveSchemaDB(SMeta *pMeta, tb_uid_t uid, STSchema *pSchema) {
   char * err = NULL;
 
   rocksdb_writeoptions_t *wopt = rocksdb_writeoptions_create();
+  rocksdb_writeoptions_disable_WAL(wopt, 1);
 
   metaGetSchemaDBKey(key, uid, schemaVersion(pSchema));
   vsize = tdEncodeSchema((void **)(&ppBuf), pSchema);
@@ -190,10 +192,12 @@ static int metaSaveMapDB(SMeta *pMeta, tb_uid_t suid, tb_uid_t uid) {
   memcpy(POINTER_SHIFT(nval, vlen), (void *)(&uid), sizeof(uid));
 
   rocksdb_writeoptions_t *wopt = rocksdb_writeoptions_create();
+  rocksdb_writeoptions_disable_WAL(wopt, 1);
 
   rocksdb_put(pMeta->pDB->mapDb, wopt, (char *)(&suid), sizeof(suid), nval, vlen + sizeof(uid), &err);
 
   rocksdb_writeoptions_destroy(wopt);
+  free(nval);
 
   return 0;
 }

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

@@ -47,7 +47,12 @@ void metaCloseIdx(SMeta *pMeta) { /* TODO */
   }
 }
 
-int metaSaveTableToIdx(SMeta *pMeta, const STbOptions *pTbOptions) {
+int metaSaveTableToIdx(SMeta *pMeta, const STbCfg *pTbOptions) {
+  // TODO
+  return 0;
+}
+
+int metaRemoveTableFromIdx(SMeta *pMeta, tb_uid_t uid) {
   // TODO
   return 0;
 }