Merge remote-tracking branch 'origin/3.0' into feature/qnode

cmake/cmake.define

@@ -44,7 +44,6 @@ ENDIF ()
 
 IF (TD_WINDOWS)
     MESSAGE("${Yellow} set compiler flag for Windows! ${ColourReset}")
-    SET(CMAKE_GENERATOR "NMake Makefiles" CACHE INTERNAL "" FORCE)
     SET(COMMON_FLAGS "/W3 /D_WIN32")
 
     # IF (MSVC AND (MSVC_VERSION GREATER_EQUAL 1900))

example/src/tmq.c

@@ -24,7 +24,7 @@ static void msg_process(TAOS_RES* msg) {
   char buf[1024];
   memset(buf, 0, 1024);
   printf("topic: %s\n", tmq_get_topic_name(msg));
-  printf("vg:%d\n", tmq_get_vgroup_id(msg));
+  printf("vg: %d\n", tmq_get_vgroup_id(msg));
   while (1) {
     TAOS_ROW row = taos_fetch_row(msg);
     if (row == NULL) break;
@@ -140,8 +140,8 @@ int32_t create_topic() {
   return 0;
 }
 
-void tmq_commit_cb_print(tmq_t* tmq, tmq_resp_err_t resp, tmq_topic_vgroup_list_t* offsets) {
-  printf("commit %d\n", resp);
+void tmq_commit_cb_print(tmq_t* tmq, tmq_resp_err_t resp, tmq_topic_vgroup_list_t* offsets, void* param) {
+  printf("commit %d tmq %p offsets %p param %p\n", resp, tmq, offsets, param);
 }
 
 tmq_t* build_consumer() {
@@ -161,7 +161,7 @@ tmq_t* build_consumer() {
   tmq_conf_set(conf, "td.connect.user", "root");
   tmq_conf_set(conf, "td.connect.pass", "taosdata");
   /*tmq_conf_set(conf, "td.connect.db", "abc1");*/
-  tmq_conf_set_offset_commit_cb(conf, tmq_commit_cb_print);
+  tmq_conf_set_offset_commit_cb(conf, tmq_commit_cb_print, NULL);
   tmq_t* tmq = tmq_consumer_new(conf, NULL, 0);
   assert(tmq);
   return tmq;
@@ -215,12 +215,24 @@ void sync_consume_loop(tmq_t* tmq, tmq_list_t* topics) {
     return;
   }
 
+  tmq_list_t* subList = NULL;
+  tmq_subscription(tmq, &subList);
+  char**  subTopics = tmq_list_to_c_array(subList);
+  int32_t sz = tmq_list_get_size(subList);
+  printf("subscribed topics: ");
+  for (int32_t i = 0; i < sz; i++) {
+    printf("%s, ", subTopics[i]);
+  }
+  printf("\n");
+  tmq_list_destroy(subList);
+
   while (running) {
     TAOS_RES* tmqmessage = tmq_consumer_poll(tmq, 1000);
     if (tmqmessage) {
       msg_process(tmqmessage);
       taos_free_result(tmqmessage);
 
+      tmq_commit(tmq, NULL, 1);
       /*if ((++msg_count % MIN_COMMIT_COUNT) == 0) tmq_commit(tmq, NULL, 0);*/
     }
   }

include/client/taos.h

@@ -213,7 +213,7 @@ typedef struct tmq_topic_vgroup_list_t tmq_topic_vgroup_list_t;
 typedef struct tmq_conf_t tmq_conf_t;
 typedef struct tmq_list_t tmq_list_t;
 
-typedef void(tmq_commit_cb(tmq_t *, tmq_resp_err_t, tmq_topic_vgroup_list_t *));
+typedef void(tmq_commit_cb(tmq_t *, tmq_resp_err_t, tmq_topic_vgroup_list_t *, void *param));
 
 DLL_EXPORT tmq_list_t *tmq_list_new();
 DLL_EXPORT int32_t     tmq_list_append(tmq_list_t *, const char *);
@@ -253,12 +253,12 @@ typedef enum tmq_conf_res_t tmq_conf_res_t;
 DLL_EXPORT tmq_conf_t    *tmq_conf_new();
 DLL_EXPORT tmq_conf_res_t tmq_conf_set(tmq_conf_t *conf, const char *key, const char *value);
 DLL_EXPORT void           tmq_conf_destroy(tmq_conf_t *conf);
-DLL_EXPORT void           tmq_conf_set_offset_commit_cb(tmq_conf_t *conf, tmq_commit_cb *cb);
+DLL_EXPORT void           tmq_conf_set_offset_commit_cb(tmq_conf_t *conf, tmq_commit_cb *cb, void *param);
 
 /* -------------------------TMQ MSG HANDLE INTERFACE---------------------- */
 
-DLL_EXPORT char   *tmq_get_topic_name(TAOS_RES *res);
-DLL_EXPORT int32_t tmq_get_vgroup_id(TAOS_RES *res);
+DLL_EXPORT const char *tmq_get_topic_name(TAOS_RES *res);
+DLL_EXPORT int32_t     tmq_get_vgroup_id(TAOS_RES *res);
 // TODO
 #if 0
 DLL_EXPORT char   *tmq_get_table_name(TAOS_RES *res);

include/common/taosdef.h

@@ -28,15 +28,16 @@ typedef int64_t tb_uid_t;
 #define TSWINDOW_INITIALIZER       ((STimeWindow){INT64_MIN, INT64_MAX})
 #define TSWINDOW_DESC_INITIALIZER  ((STimeWindow){INT64_MAX, INT64_MIN})
 #define IS_TSWINDOW_SPECIFIED(win) (((win).skey != INT64_MIN) || ((win).ekey != INT64_MAX))
+#define TSWINDOW_IS_EQUAL(t1, t2)  (((t1).skey == (t2).skey) && ((t1).ekey == (t2).ekey))
 
 typedef enum {
-  TSDB_SUPER_TABLE  = 1,    // super table
-  TSDB_CHILD_TABLE  = 2,    // table created from super table
-  TSDB_NORMAL_TABLE = 3,    // ordinary table
-  TSDB_STREAM_TABLE = 4,    // table created from stream computing
-  TSDB_TEMP_TABLE   = 5,    // temp table created by nest query
+  TSDB_SUPER_TABLE = 1,   // super table
+  TSDB_CHILD_TABLE = 2,   // table created from super table
+  TSDB_NORMAL_TABLE = 3,  // ordinary table
+  TSDB_STREAM_TABLE = 4,  // table created from stream computing
+  TSDB_TEMP_TABLE = 5,    // temp table created by nest query
   TSDB_SYSTEM_TABLE = 6,
-  TSDB_TABLE_MAX    = 7
+  TSDB_TABLE_MAX = 7
 } ETableType;
 
 typedef enum {
@@ -71,16 +72,22 @@ typedef enum {
   TSDB_SMA_STAT_DROPPED = 2,   // sma dropped
 } ETsdbSmaStat;                // bit operation
 
-
 typedef enum {
   TSDB_SMA_TYPE_BLOCK = 0,       // Block-wise SMA
   TSDB_SMA_TYPE_TIME_RANGE = 1,  // Time-range-wise SMA
   TSDB_SMA_TYPE_ROLLUP = 2,      // Rollup SMA
 } ETsdbSmaType;
 
+typedef enum {
+  TSDB_RETENTION_L0 = 0,
+  TSDB_RETENTION_L1 = 1,
+  TSDB_RETENTION_L2 = 2,
+  TSDB_RETENTION_MAX = 3
+} ERetentionLevel;
+
 extern char *qtypeStr[];
 
-#define TSDB_PORT_HTTP      11
+#define TSDB_PORT_HTTP 11
 
 #undef TD_DEBUG_PRINT_ROW
 

include/common/tcommon.h

@@ -65,20 +65,18 @@ typedef struct SDataBlockInfo {
   STimeWindow window;
   int32_t     rows;
   int32_t     rowSize;
-  union {
-    int64_t uid;  // from which table of uid, comes from this data block
-    int64_t blockId;
-  };
-  uint64_t groupId;  // no need to serialize
-  int16_t  numOfCols;
-  int16_t  hasVarCol;
-  int16_t  capacity;
+  int64_t     uid;      // the uid of table, from which current data block comes
+  int64_t     blockId;  // block id, generated by physical planner
+  uint64_t    groupId;  // no need to serialize
+  int16_t     numOfCols;
+  int16_t     hasVarCol;
+  int32_t     capacity;
 } SDataBlockInfo;
 
 typedef struct SSDataBlock {
-  SColumnDataAgg* pBlockAgg;
-  SArray*         pDataBlock;  // SArray<SColumnInfoData>
-  SDataBlockInfo  info;
+  SColumnDataAgg** pBlockAgg;
+  SArray*          pDataBlock;  // SArray<SColumnInfoData>
+  SDataBlockInfo   info;
 } SSDataBlock;
 
 typedef struct SVarColAttr {

include/common/tdatablock.h

@@ -183,7 +183,7 @@ static FORCE_INLINE void colDataAppendDouble(SColumnInfoData* pColumnInfoData, u
 }
 
 int32_t colDataAppend(SColumnInfoData* pColumnInfoData, uint32_t currentRow, const char* pData, bool isNull);
-int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, const SColumnInfoData* pSource,
+int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, int32_t* capacity, const SColumnInfoData* pSource,
                         uint32_t numOfRow2);
 int32_t colDataAssign(SColumnInfoData* pColumnInfoData, const SColumnInfoData* pSource, int32_t numOfRows);
 int32_t blockDataUpdateTsWindow(SSDataBlock* pDataBlock);
@@ -225,6 +225,9 @@ SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData);
 
 void blockDebugShowData(const SArray* dataBlocks);
 
+int32_t buildSubmitReqFromDataBlock(SSubmitReq** pReq, const SArray* pDataBlocks, STSchema* pTSchema, int32_t vgId,
+                                 tb_uid_t uid, tb_uid_t suid);
+
 static FORCE_INLINE int32_t blockGetEncodeSize(const SSDataBlock* pBlock) {
   return blockDataGetSerialMetaSize(pBlock) + blockDataGetSize(pBlock);
 }

include/common/tdataformat.h

@@ -64,19 +64,19 @@ typedef struct {
   col_id_t colId;        // column ID(start from PRIMARYKEY_TIMESTAMP_COL_ID(1))
   int32_t  type : 8;     // column type
   int32_t  bytes : 24;   // column bytes (0~16M)
-  int32_t  sma : 8;      // block SMA: 0, no SMA, 1, sum/min/max, 2, ...
+  int32_t  flags : 8;    // flags: 0 no index, 1 SCHEMA_SMA_ON, 2 SCHEMA_IDX_ON
   int32_t  offset : 24;  // point offset in STpRow after the header part.
 } STColumn;
 #pragma pack(pop)
 
 #define colType(col)   ((col)->type)
-#define colSma(col)    ((col)->sma)
+#define colFlags(col)  ((col)->flags)
 #define colColId(col)  ((col)->colId)
 #define colBytes(col)  ((col)->bytes)
 #define colOffset(col) ((col)->offset)
 
 #define colSetType(col, t)   (colType(col) = (t))
-#define colSetSma(col, s)    (colSma(col) = (s))
+#define colSetFlags(col, f)  (colFlags(col) = (f))
 #define colSetColId(col, id) (colColId(col) = (id))
 #define colSetBytes(col, b)  (colBytes(col) = (b))
 #define colSetOffset(col, o) (colOffset(col) = (o))
@@ -146,7 +146,7 @@ typedef struct {
 int32_t   tdInitTSchemaBuilder(STSchemaBuilder *pBuilder, schema_ver_t version);
 void      tdDestroyTSchemaBuilder(STSchemaBuilder *pBuilder);
 void      tdResetTSchemaBuilder(STSchemaBuilder *pBuilder, schema_ver_t version);
-int32_t   tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t sma, col_id_t colId, col_bytes_t bytes);
+int32_t   tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t flags, col_id_t colId, col_bytes_t bytes);
 STSchema *tdGetSchemaFromBuilder(STSchemaBuilder *pBuilder);
 
 // ----------------- Semantic timestamp key definition

include/common/tglobal.h

@@ -121,6 +121,10 @@ extern char     tsCompressor[];
 extern int32_t  tsDiskCfgNum;
 extern SDiskCfg tsDiskCfg[];
 
+// internal 
+extern int32_t tsTransPullupMs;
+extern int32_t tsMaRebalanceMs;
+
 #define NEEDTO_COMPRESSS_MSG(size) (tsCompressMsgSize != -1 && (size) > tsCompressMsgSize)
 
 int32_t taosCreateLog(const char *logname, int32_t logFileNum, const char *cfgDir, const char **envCmd, const char *envFile,

include/common/tmsg.h

@@ -181,8 +181,8 @@ typedef struct SField {
 } SField;
 
 typedef struct SRetention {
-  int32_t freq;
-  int32_t keep;
+  int64_t freq;
+  int64_t keep;
   int8_t  freqUnit;
   int8_t  keepUnit;
 } SRetention;
@@ -243,18 +243,9 @@ typedef struct {
 
 int32_t tInitSubmitMsgIter(const SSubmitReq* pMsg, SSubmitMsgIter* pIter);
 int32_t tGetSubmitMsgNext(SSubmitMsgIter* pIter, SSubmitBlk** pPBlock);
-int32_t tInitSubmitBlkIter(SSubmitBlk* pBlock, SSubmitBlkIter* pIter);
+int32_t tInitSubmitBlkIter(SSubmitMsgIter* pMsgIter, SSubmitBlk* pBlock, SSubmitBlkIter* pIter);
 STSRow* tGetSubmitBlkNext(SSubmitBlkIter* pIter);
 
-// TODO: KEEP one suite of iterator API finally.
-// 1) use tInitSubmitMsgIterEx firstly as not decrease the merge conflicts
-// 2) replace tInitSubmitMsgIterEx with tInitSubmitMsgIter later
-// 3) finally, rename tInitSubmitMsgIterEx to tInitSubmitMsgIter
-int32_t tInitSubmitMsgIterEx(const SSubmitReq* pMsg, SSubmitMsgIter* pIter);
-int32_t tGetSubmitMsgNextEx(SSubmitMsgIter* pIter, SSubmitBlk** pPBlock);
-int32_t tInitSubmitBlkIterEx(SSubmitMsgIter* pMsgIter, SSubmitBlk* pBlock, SSubmitBlkIter* pIter);
-STSRow* tGetSubmitBlkNextEx(SSubmitBlkIter* pIter);
-
 typedef struct {
   int32_t index;  // index of failed block in submit blocks
   int32_t vnode;  // vnode index of failed block
@@ -281,8 +272,10 @@ typedef struct SSchema {
   char     name[TSDB_COL_NAME_LEN];
 } SSchema;
 
+#define IS_BSMA_ON(s) (((s)->flags & 0x01) == SCHEMA_SMA_ON)
+
 #define SSCHMEA_TYPE(s)  ((s)->type)
-#define SSCHMEA_SMA(s)   ((s)->sma)
+#define SSCHMEA_FLAGS(s) ((s)->flags)
 #define SSCHMEA_COLID(s) ((s)->colId)
 #define SSCHMEA_BYTES(s) ((s)->bytes)
 #define SSCHMEA_NAME(s)  ((s)->name)
@@ -842,7 +835,7 @@ typedef struct {
   char     db[TSDB_DB_FNAME_LEN];
   int64_t  dbUid;
   int32_t  vgVersion;
-   int32_t numOfStables;
+  int32_t  numOfStables;
   int32_t  buffer;
   int32_t  pageSize;
   int32_t  pages;
@@ -1442,32 +1435,32 @@ typedef struct {
   SArray* lostConsumers;     // SArray<int64_t>
   SArray* removedConsumers;  // SArray<int64_t>
   SArray* newConsumers;      // SArray<int64_t>
-} SMqRebSubscribe;
+} SMqRebInfo;
 
-static FORCE_INLINE SMqRebSubscribe* tNewSMqRebSubscribe(const char* key) {
-  SMqRebSubscribe* pRebSub = (SMqRebSubscribe*)taosMemoryCalloc(1, sizeof(SMqRebSubscribe));
-  if (pRebSub == NULL) {
+static FORCE_INLINE SMqRebInfo* tNewSMqRebSubscribe(const char* key) {
+  SMqRebInfo* pRebInfo = (SMqRebInfo*)taosMemoryCalloc(1, sizeof(SMqRebInfo));
+  if (pRebInfo == NULL) {
     goto _err;
   }
-  strcpy(pRebSub->key, key);
-  pRebSub->lostConsumers = taosArrayInit(0, sizeof(int64_t));
-  if (pRebSub->lostConsumers == NULL) {
+  strcpy(pRebInfo->key, key);
+  pRebInfo->lostConsumers = taosArrayInit(0, sizeof(int64_t));
+  if (pRebInfo->lostConsumers == NULL) {
     goto _err;
   }
-  pRebSub->removedConsumers = taosArrayInit(0, sizeof(int64_t));
-  if (pRebSub->removedConsumers == NULL) {
+  pRebInfo->removedConsumers = taosArrayInit(0, sizeof(int64_t));
+  if (pRebInfo->removedConsumers == NULL) {
     goto _err;
   }
-  pRebSub->newConsumers = taosArrayInit(0, sizeof(int64_t));
-  if (pRebSub->newConsumers == NULL) {
+  pRebInfo->newConsumers = taosArrayInit(0, sizeof(int64_t));
+  if (pRebInfo->newConsumers == NULL) {
     goto _err;
   }
-  return pRebSub;
+  return pRebInfo;
 _err:
-  taosArrayDestroy(pRebSub->lostConsumers);
-  taosArrayDestroy(pRebSub->removedConsumers);
-  taosArrayDestroy(pRebSub->newConsumers);
-  taosMemoryFreeClear(pRebSub);
+  taosArrayDestroy(pRebInfo->lostConsumers);
+  taosArrayDestroy(pRebInfo->removedConsumers);
+  taosArrayDestroy(pRebInfo->newConsumers);
+  taosMemoryFreeClear(pRebInfo);
   return NULL;
 }
 
@@ -1524,6 +1517,7 @@ typedef struct {
 int32_t tEncodeSRSmaParam(SCoder* pCoder, const SRSmaParam* pRSmaParam);
 int32_t tDecodeSRSmaParam(SCoder* pCoder, SRSmaParam* pRSmaParam);
 
+// TDMT_VND_CREATE_STB ==============
 typedef struct SVCreateStbReq {
   const char*    name;
   tb_uid_t       suid;
@@ -1536,17 +1530,14 @@ typedef struct SVCreateStbReq {
 int tEncodeSVCreateStbReq(SCoder* pCoder, const SVCreateStbReq* pReq);
 int tDecodeSVCreateStbReq(SCoder* pCoder, SVCreateStbReq* pReq);
 
+// TDMT_VND_DROP_STB ==============
 typedef struct SVDropStbReq {
-  // data
-#ifdef WINDOWS
-  size_t avoidCompilationErrors;
-#endif
-
+  const char* name;
+  tb_uid_t    suid;
 } SVDropStbReq;
 
-typedef struct SVCreateStbRsp {
-  int code;
-} SVCreateStbRsp;
+int32_t tEncodeSVDropStbReq(SCoder* pCoder, const SVDropStbReq* pReq);
+int32_t tDecodeSVDropStbReq(SCoder* pCoder, SVDropStbReq* pReq);
 
 typedef struct SVCreateTbReq {
   tb_uid_t    uid;
@@ -1603,19 +1594,37 @@ int tDecodeSVCreateTbBatchRsp(SCoder* pCoder, SVCreateTbBatchRsp* pRsp);
 int32_t tSerializeSVCreateTbBatchRsp(void* buf, int32_t bufLen, SVCreateTbBatchRsp* pRsp);
 int32_t tDeserializeSVCreateTbBatchRsp(void* buf, int32_t bufLen, SVCreateTbBatchRsp* pRsp);
 
+// TDMT_VND_DROP_TABLE =================
 typedef struct {
-  int64_t  ver;
-  char*    name;
-  uint8_t  type;
-  tb_uid_t suid;
+  const char* name;
+  int8_t      igNotExists;
 } SVDropTbReq;
 
 typedef struct {
-  int tmp;  // TODO: to avoid compile error
+  int32_t code;
 } SVDropTbRsp;
 
-int32_t tSerializeSVDropTbReq(void** buf, SVDropTbReq* pReq);
-void*   tDeserializeSVDropTbReq(void* buf, SVDropTbReq* pReq);
+typedef struct {
+  int32_t nReqs;
+  union {
+    SVDropTbReq* pReqs;
+    SArray*      pArray;
+  };
+} SVDropTbBatchReq;
+
+int32_t tEncodeSVDropTbBatchReq(SCoder* pCoder, const SVDropTbBatchReq* pReq);
+int32_t tDecodeSVDropTbBatchReq(SCoder* pCoder, SVDropTbBatchReq* pReq);
+
+typedef struct {
+  int32_t nRsps;
+  union {
+    SVDropTbRsp* pRsps;
+    SArray*      pArray;
+  };
+} SVDropTbBatchRsp;
+
+int32_t tEncodeSVDropTbBatchRsp(SCoder* pCoder, const SVDropTbBatchRsp* pRsp);
+int32_t tDecodeSVDropTbBatchRsp(SCoder* pCoder, SVDropTbBatchRsp* pRsp);
 
 typedef struct {
   SMsgHead head;

include/common/tmsgcb.h

@@ -42,6 +42,7 @@ typedef int32_t (*GetQueueSizeFp)(SMgmtWrapper* pWrapper, int32_t vgId, EQueueTy
 typedef int32_t (*SendReqFp)(SMgmtWrapper* pWrapper, const SEpSet* epSet, SRpcMsg* pReq);
 typedef int32_t (*SendMnodeReqFp)(SMgmtWrapper* pWrapper, SRpcMsg* pReq);
 typedef void (*SendRspFp)(SMgmtWrapper* pWrapper, const SRpcMsg* pRsp);
+typedef void (*SendRedirectRspFp)(SMgmtWrapper* pWrapper, const SRpcMsg* pRsp, const SEpSet* pNewEpSet);
 typedef void (*RegisterBrokenLinkArgFp)(SMgmtWrapper* pWrapper, SRpcMsg* pMsg);
 typedef void (*ReleaseHandleFp)(SMgmtWrapper* pWrapper, void* handle, int8_t type);
 typedef void (*ReportStartup)(SMgmtWrapper* pWrapper, const char* name, const char* desc);
@@ -52,6 +53,7 @@ typedef struct {
   GetQueueSizeFp          qsizeFp;
   SendReqFp               sendReqFp;
   SendRspFp               sendRspFp;
+  SendRedirectRspFp       sendRedirectRspFp;
   RegisterBrokenLinkArgFp registerBrokenLinkArgFp;
   ReleaseHandleFp         releaseHandleFp;
   ReportStartup           reportStartupFp;
@@ -62,6 +64,7 @@ int32_t tmsgPutToQueue(const SMsgCb* pMsgCb, EQueueType qtype, SRpcMsg* pReq);
 int32_t tmsgGetQueueSize(const SMsgCb* pMsgCb, int32_t vgId, EQueueType qtype);
 int32_t tmsgSendReq(const SMsgCb* pMsgCb, const SEpSet* epSet, SRpcMsg* pReq);
 void    tmsgSendRsp(const SRpcMsg* pRsp);
+void    tmsgSendRedirectRsp(const SRpcMsg* pRsp, const SEpSet* pNewEpSet);
 void    tmsgRegisterBrokenLinkArg(const SMsgCb* pMsgCb, SRpcMsg* pMsg);
 void    tmsgReleaseHandle(void* handle, int8_t type);
 void    tmsgReportStartup(const char* name, const char* desc);

include/common/tmsgdef.h

@@ -170,9 +170,9 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_UPDATE_TAG_VAL, "vnode-update-tag-val", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_TABLE_META, "vnode-table-meta", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_TABLES_META, "vnode-tables-meta", NULL, NULL)
-  TD_DEF_MSG_TYPE(TDMT_VND_CREATE_STB, "vnode-create-stb", SVCreateTbReq, SVCreateTbRsp)
+  TD_DEF_MSG_TYPE(TDMT_VND_CREATE_STB, "vnode-create-stb", SVCreateStbReq, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_ALTER_STB, "vnode-alter-stb", NULL, NULL)
-  TD_DEF_MSG_TYPE(TDMT_VND_DROP_STB, "vnode-drop-stb", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_VND_DROP_STB, "vnode-drop-stb", SVDropStbReq, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_CONSUME, "vnode-mq-consume", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_QUERY, "vnode-mq-query", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_CONNECT, "vnode-mq-connect", NULL, NULL)
@@ -202,6 +202,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_CREATE_SMA, "vnode-create-sma", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_CANCEL_SMA, "vnode-cancel-sma", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_DROP_SMA, "vnode-drop-sma", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_VND_SUBMIT_RSMA, "vnode-submit-rsma", SSubmitReq, SSubmitRsp)
 
   // sync integration
   TD_DEF_MSG_TYPE(TDMT_VND_SYNC_TIMEOUT, "vnode-sync-timeout", NULL, NULL)

include/dnode/mnode/sdb/sdb.h

@@ -107,10 +107,9 @@ typedef enum {
 typedef enum {
   SDB_STATUS_INIT = 0,
   SDB_STATUS_CREATING = 1,
-  SDB_STATUS_UPDATING = 2,
-  SDB_STATUS_DROPPING = 3,
+  SDB_STATUS_DROPPING = 2,
+  SDB_STATUS_DROPPED = 3,
   SDB_STATUS_READY = 4,
-  SDB_STATUS_DROPPED = 5
 } ESdbStatus;
 
 typedef enum {

include/libs/function/function.h

@@ -165,7 +165,7 @@ typedef struct SInputColumnInfoData {
   SColumnInfoData  *pPTS;           // primary timestamp column
   SColumnInfoData **pData;
   SColumnDataAgg  **pColumnDataAgg;
-  uint64_t          uid;            // table uid
+  uint64_t          uid;            // table uid, used to set the tag value when building the final query result for selectivity functions.
 } SInputColumnInfoData;
 
 // sql function runtime context

include/libs/function/functionMgt.h

@@ -40,6 +40,7 @@ typedef enum EFunctionType {
   FUNCTION_TYPE_STDDEV,
   FUNCTION_TYPE_SUM,
   FUNCTION_TYPE_TWA,
+  FUNCTION_TYPE_HISTOGRAM,
 
   // nonstandard SQL function
   FUNCTION_TYPE_BOTTOM = 500,
@@ -112,6 +113,9 @@ typedef enum EFunctionType {
   FUNCTION_TYPE_WENDTS,
   FUNCTION_TYPE_WDURATION,
 
+  // internal function
+  FUNCTION_TYPE_SELECT_VALUE,
+
   // user defined funcion
   FUNCTION_TYPE_UDF = 10000
 } EFunctionType;
@@ -140,6 +144,7 @@ bool fmIsScalarFunc(int32_t funcId);
 bool fmIsNonstandardSQLFunc(int32_t funcId);
 bool fmIsStringFunc(int32_t funcId);
 bool fmIsDatetimeFunc(int32_t funcId);
+bool fmIsSelectFunc(int32_t funcId);
 bool fmIsTimelineFunc(int32_t funcId);
 bool fmIsTimeorderFunc(int32_t funcId);
 bool fmIsPseudoColumnFunc(int32_t funcId);

include/libs/index/index.h

@@ -17,6 +17,7 @@
 #define _TD_INDEX_H_
 
 #include "os.h"
+#include "taoserror.h"
 #include "tarray.h"
 
 #ifdef __cplusplus
@@ -41,11 +42,12 @@ typedef enum {
   UPDATE_VALUE,  // update index column value
   ADD_INDEX,     // add index on specify column
   DROP_INDEX,    // drop existed index
-  DROP_SATBLE    // drop stable
+  DROP_SATBLE,   // drop stable
+  DEFAULT        // query
 } SIndexOperOnColumn;
 
-typedef enum { MUST = 0, SHOULD = 1, NOT = 2 } EIndexOperatorType;
-typedef enum { QUERY_TERM = 0, QUERY_PREFIX = 1, QUERY_SUFFIX = 2, QUERY_REGEX = 3, QUERY_RANGE = 4 } EIndexQueryType;
+typedef enum { MUST = 0, SHOULD, NOT } EIndexOperatorType;
+typedef enum { QUERY_TERM = 0, QUERY_PREFIX, QUERY_SUFFIX, QUERY_REGEX, QUERY_RANGE } EIndexQueryType;
 
 /*
  * create multi query
@@ -166,8 +168,8 @@ void        indexOptsDestroy(SIndexOpts* opts);
  * @param:
  */
 
-SIndexTerm* indexTermCreate(int64_t suid, SIndexOperOnColumn operType, uint8_t colType, const char* colName,
-                            int32_t nColName, const char* colVal, int32_t nColVal);
+SIndexTerm* indexTermCreate(int64_t suid, SIndexOperOnColumn operType, int8_t qType, uint8_t colType,
+                            const char* colName, int32_t nColName, const char* colVal, int32_t nColVal);
 void        indexTermDestroy(SIndexTerm* p);
 
 /*

include/libs/nodes/nodes.h

@@ -22,19 +22,20 @@ extern "C" {
 
 #include "tdef.h"
 
-#define nodeType(nodeptr) (((const SNode*)(nodeptr))->type)
+#define nodeType(nodeptr)          (((const SNode*)(nodeptr))->type)
 #define setNodeType(nodeptr, type) (((SNode*)(nodeptr))->type = (type))
 
 #define LIST_LENGTH(l) (NULL != (l) ? (l)->length : 0)
 
-#define FOREACH(node, list)	\
-  for (SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); (NULL != cell ? (node = cell->pNode, true) : (node = NULL, false)); cell = cell->pNext)
+#define FOREACH(node, list)                                       \
+  for (SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); \
+       (NULL != cell ? (node = cell->pNode, true) : (node = NULL, false)); cell = cell->pNext)
 
 #define REPLACE_NODE(newNode) cell->pNode = (SNode*)(newNode)
 
 #define INSERT_LIST(target, src) nodesListInsertList((target), cell, src)
 
-#define WHERE_EACH(node, list) \
+#define WHERE_EACH(node, list)                               \
   SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); \
   while (NULL != cell ? (node = cell->pNode, true) : (node = NULL, false))
 
@@ -43,16 +44,26 @@ extern "C" {
 // only be use in WHERE_EACH
 #define ERASE_NODE(list) cell = nodesListErase((list), cell)
 
-#define FORBOTH(node1, list1, node2, list2) \
-  for (SListCell* cell1 = (NULL != (list1) ? (list1)->pHead : NULL), *cell2 = (NULL != (list2) ? (list2)->pHead : NULL); \
-    (NULL == cell1 ? (node1 = NULL, false) : (node1 = cell1->pNode, true)), (NULL == cell2 ? (node2 = NULL, false) : (node2 = cell2->pNode, true)), (node1 != NULL && node2 != NULL); \
-    cell1 = cell1->pNext, cell2 = cell2->pNext)
+#define FORBOTH(node1, list1, node2, list2)                                               \
+  for (SListCell* cell1 = (NULL != (list1) ? (list1)->pHead : NULL),                      \
+                  *cell2 = (NULL != (list2) ? (list2)->pHead : NULL);                     \
+       (NULL == cell1 ? (node1 = NULL, false) : (node1 = cell1->pNode, true)),            \
+                  (NULL == cell2 ? (node2 = NULL, false) : (node2 = cell2->pNode, true)), \
+                  (node1 != NULL && node2 != NULL);                                       \
+       cell1 = cell1->pNext, cell2 = cell2->pNext)
 
 #define REPLACE_LIST1_NODE(newNode) cell1->pNode = (SNode*)(newNode)
 #define REPLACE_LIST2_NODE(newNode) cell2->pNode = (SNode*)(newNode)
 
-#define FOREACH_FOR_REWRITE(node, list)	\
-  for (SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); (NULL != cell ? (node = &(cell->pNode), true) : (node = NULL, false)); cell = cell->pNext)
+#define FOREACH_FOR_REWRITE(node, list)                           \
+  for (SListCell* cell = (NULL != (list) ? (list)->pHead : NULL); \
+       (NULL != cell ? (node = &(cell->pNode), true) : (node = NULL, false)); cell = cell->pNext)
+
+#define DESTORY_LIST(list)  \
+  do {                      \
+    nodesDestroyList(list); \
+    list = NULL;            \
+  } while (0)
 
 typedef enum ENodeType {
   // Syntax nodes are used in parser and planner module, and some are also used in executor module, such as COLUMN,
@@ -174,6 +185,7 @@ typedef enum ENodeType {
   QUERY_NODE_LOGIC_PLAN_VNODE_MODIF,
   QUERY_NODE_LOGIC_PLAN_EXCHANGE,
   QUERY_NODE_LOGIC_PLAN_WINDOW,
+  QUERY_NODE_LOGIC_PLAN_FILL,
   QUERY_NODE_LOGIC_PLAN_SORT,
   QUERY_NODE_LOGIC_PLAN_PARTITION,
   QUERY_NODE_LOGIC_SUBPLAN,
@@ -191,6 +203,7 @@ typedef enum ENodeType {
   QUERY_NODE_PHYSICAL_PLAN_EXCHANGE,
   QUERY_NODE_PHYSICAL_PLAN_SORT,
   QUERY_NODE_PHYSICAL_PLAN_INTERVAL,
+  QUERY_NODE_PHYSICAL_PLAN_FILL,
   QUERY_NODE_PHYSICAL_PLAN_SESSION_WINDOW,
   QUERY_NODE_PHYSICAL_PLAN_STATE_WINDOW,
   QUERY_NODE_PHYSICAL_PLAN_PARTITION,
@@ -202,7 +215,7 @@ typedef enum ENodeType {
 
 /**
  * The first field of a node of any type is guaranteed to be the ENodeType.
- * Hence the type of any node can be gotten by casting it to SNode. 
+ * Hence the type of any node can be gotten by casting it to SNode.
  */
 typedef struct SNode {
   ENodeType type;
@@ -211,41 +224,36 @@ typedef struct SNode {
 typedef struct SListCell {
   struct SListCell* pPrev;
   struct SListCell* pNext;
-  SNode* pNode;
+  SNode*            pNode;
 } SListCell;
 
 typedef struct SNodeList {
-  int32_t length;
+  int32_t    length;
   SListCell* pHead;
   SListCell* pTail;
 } SNodeList;
 
-#define SNodeptr void*  
+#define SNodeptr void*
 
 SNodeptr nodesMakeNode(ENodeType type);
-void nodesDestroyNode(SNodeptr pNode);
+void     nodesDestroyNode(SNodeptr pNode);
 
 SNodeList* nodesMakeList();
-int32_t nodesListAppend(SNodeList* pList, SNodeptr pNode);
-int32_t nodesListStrictAppend(SNodeList* pList, SNodeptr pNode);
-int32_t nodesListMakeAppend(SNodeList** pList, SNodeptr pNode);
-int32_t nodesListMakeStrictAppend(SNodeList** pList, SNodeptr pNode);
-int32_t nodesListAppendList(SNodeList* pTarget, SNodeList* pSrc);
-int32_t nodesListStrictAppendList(SNodeList* pTarget, SNodeList* pSrc);
-int32_t nodesListPushFront(SNodeList* pList, SNodeptr pNode);
+int32_t    nodesListAppend(SNodeList* pList, SNodeptr pNode);
+int32_t    nodesListStrictAppend(SNodeList* pList, SNodeptr pNode);
+int32_t    nodesListMakeAppend(SNodeList** pList, SNodeptr pNode);
+int32_t    nodesListMakeStrictAppend(SNodeList** pList, SNodeptr pNode);
+int32_t    nodesListAppendList(SNodeList* pTarget, SNodeList* pSrc);
+int32_t    nodesListStrictAppendList(SNodeList* pTarget, SNodeList* pSrc);
+int32_t    nodesListPushFront(SNodeList* pList, SNodeptr pNode);
 SListCell* nodesListErase(SNodeList* pList, SListCell* pCell);
-void nodesListInsertList(SNodeList* pTarget, SListCell* pPos, SNodeList* pSrc);
-SNodeptr nodesListGetNode(SNodeList* pList, int32_t index);
-void nodesDestroyList(SNodeList* pList);
+void       nodesListInsertList(SNodeList* pTarget, SListCell* pPos, SNodeList* pSrc);
+SNodeptr   nodesListGetNode(SNodeList* pList, int32_t index);
+void       nodesDestroyList(SNodeList* pList);
 // Only clear the linked list structure, without releasing the elements inside
 void nodesClearList(SNodeList* pList);
 
-typedef enum EDealRes {
-  DEAL_RES_CONTINUE = 1,
-  DEAL_RES_IGNORE_CHILD,
-  DEAL_RES_ERROR,
-  DEAL_RES_END
-} EDealRes;
+typedef enum EDealRes { DEAL_RES_CONTINUE = 1, DEAL_RES_IGNORE_CHILD, DEAL_RES_ERROR, DEAL_RES_END } EDealRes;
 
 typedef EDealRes (*FNodeWalker)(SNode* pNode, void* pContext);
 void nodesWalkExpr(SNodeptr pNode, FNodeWalker walker, void* pContext);
@@ -261,18 +269,18 @@ void nodesRewriteExprsPostOrder(SNodeList* pList, FNodeRewriter rewriter, void*
 
 bool nodesEqualNode(const SNodeptr a, const SNodeptr b);
 
-SNodeptr nodesCloneNode(const SNodeptr pNode);
+SNodeptr   nodesCloneNode(const SNodeptr pNode);
 SNodeList* nodesCloneList(const SNodeList* pList);
 
 const char* nodesNodeName(ENodeType type);
-int32_t nodesNodeToString(const SNodeptr pNode, bool format, char** pStr, int32_t* pLen);
-int32_t nodesStringToNode(const char* pStr, SNode** pNode);
+int32_t     nodesNodeToString(const SNodeptr pNode, bool format, char** pStr, int32_t* pLen);
+int32_t     nodesStringToNode(const char* pStr, SNode** pNode);
 
 int32_t nodesListToString(const SNodeList* pList, bool format, char** pStr, int32_t* pLen);
 int32_t nodesStringToList(const char* pStr, SNodeList** pList);
 
-int32_t nodesNodeToSQL(SNode *pNode, char *buf, int32_t bufSize, int32_t *len);
-char *nodesGetNameFromColumnNode(SNode *pNode);
+int32_t nodesNodeToSQL(SNode* pNode, char* buf, int32_t bufSize, int32_t* len);
+char*   nodesGetNameFromColumnNode(SNode* pNode);
 int32_t nodesGetOutputNumFromSlotList(SNodeList* pSlots);
 
 #ifdef __cplusplus

include/libs/nodes/plannodes.h

@@ -102,7 +102,6 @@ typedef struct SWindowLogicNode {
   int64_t     sliding;
   int8_t      intervalUnit;
   int8_t      slidingUnit;
-  SFillNode*  pFill;
   int64_t     sessionGap;
   SNode*      pTspk;
   SNode*      pStateExpr;
@@ -110,6 +109,14 @@ typedef struct SWindowLogicNode {
   int64_t     watermark;
 } SWindowLogicNode;
 
+typedef struct SFillLogicNode {
+  SLogicNode  node;
+  EFillMode   mode;
+  SNode*      pWStartTs;
+  SNode*      pValues;  // SNodeListNode
+  STimeWindow timeRange;
+} SFillLogicNode;
+
 typedef struct SSortLogicNode {
   SLogicNode node;
   SNodeList* pSortKeys;
@@ -223,10 +230,12 @@ typedef struct SProjectPhysiNode {
 typedef struct SJoinPhysiNode {
   SPhysiNode node;
   EJoinType  joinType;
-  SNode*     pOnConditions;  // in or out tuple ?
+  SNode*     pOnConditions;
   SNodeList* pTargets;
 } SJoinPhysiNode;
 
+typedef SJoinPhysiNode SSortMergeJoinPhysiNode;
+
 typedef struct SAggPhysiNode {
   SPhysiNode node;
   SNodeList* pExprs;  // these are expression list of group_by_clause and parameter expression of aggregate function
@@ -263,9 +272,17 @@ typedef struct SIntervalPhysiNode {
   int64_t          sliding;
   int8_t           intervalUnit;
   int8_t           slidingUnit;
-  SFillNode*       pFill;
 } SIntervalPhysiNode;
 
+typedef struct SFillPhysiNode {
+  SPhysiNode  node;
+  EFillMode   mode;
+  SNode*      pWStartTs;  // SColumnNode
+  SNode*      pValues;    // SNodeListNode
+  SNodeList*  pTargets;
+  STimeWindow timeRange;
+} SFillPhysiNode;
+
 typedef struct SMultiTableIntervalPhysiNode {
   SIntervalPhysiNode interval;
   SNodeList*         pPartitionKeys;
@@ -340,7 +357,7 @@ typedef struct SQueryPlan {
   int32_t      numOfSubplans;
   SNodeList*   pSubplans;  // Element is SNodeListNode. The execution level of subplan, starting from 0.
   SExplainInfo explainInfo;
-  SArray* pPlaceholderValues;
+  SArray*      pPlaceholderValues;
 } SQueryPlan;
 
 void nodesWalkPhysiPlan(SNode* pNode, FNodeWalker walker, void* pContext);

include/libs/nodes/querynodes.h

@@ -89,7 +89,7 @@ typedef struct SValueNode {
     char*    p;
   } datum;
   int64_t typeData;
-  char unit;
+  char    unit;
 } SValueNode;
 
 typedef struct SOperatorNode {
@@ -209,9 +209,11 @@ typedef enum EFillMode {
 } EFillMode;
 
 typedef struct SFillNode {
-  ENodeType type;  // QUERY_NODE_FILL
-  EFillMode mode;
-  SNode*    pValues;  // SNodeListNode
+  ENodeType   type;  // QUERY_NODE_FILL
+  EFillMode   mode;
+  SNode*      pValues;    // SNodeListNode
+  SNode*      pWStartTs;  // _wstartts pseudo column
+  STimeWindow timeRange;
 } SFillNode;
 
 typedef struct SSelectStmt {
@@ -231,6 +233,7 @@ typedef struct SSelectStmt {
   uint8_t     precision;
   bool        isEmptyResult;
   bool        hasAggFuncs;
+  bool        isTimeOrderQuery;
 } SSelectStmt;
 
 typedef enum ESetOperatorType { SET_OP_TYPE_UNION_ALL = 1, SET_OP_TYPE_UNION } ESetOperatorType;
@@ -295,12 +298,13 @@ void nodesWalkSelectStmt(SSelectStmt* pSelect, ESqlClause clause, FNodeWalker wa
 void nodesRewriteSelectStmt(SSelectStmt* pSelect, ESqlClause clause, FNodeRewriter rewriter, void* pContext);
 
 typedef enum ECollectColType { COLLECT_COL_TYPE_COL = 1, COLLECT_COL_TYPE_TAG, COLLECT_COL_TYPE_ALL } ECollectColType;
-
 int32_t nodesCollectColumns(SSelectStmt* pSelect, ESqlClause clause, const char* pTableAlias, ECollectColType type,
                             SNodeList** pCols);
 
 typedef bool (*FFuncClassifier)(int32_t funcId);
-int32_t nodesCollectFuncs(SSelectStmt* pSelect, FFuncClassifier classifier, SNodeList** pFuncs);
+int32_t nodesCollectFuncs(SSelectStmt* pSelect, ESqlClause clause, FFuncClassifier classifier, SNodeList** pFuncs);
+
+int32_t nodesCollectSpecialNodes(SSelectStmt* pSelect, ESqlClause clause, ENodeType type, SNodeList** pNodes);
 
 bool nodesIsExprNode(const SNode* pNode);
 
@@ -312,11 +316,11 @@ bool nodesIsJsonOp(const SOperatorNode* pOp);
 bool nodesIsTimeorderQuery(const SNode* pQuery);
 bool nodesIsTimelineQuery(const SNode* pQuery);
 
-void* nodesGetValueFromNode(SValueNode* pNode);
-int32_t nodesSetValueNodeValue(SValueNode* pNode, void *value);
-char* nodesGetStrValueFromNode(SValueNode* pNode);
-char* getFillModeString(EFillMode mode);
-void  valueNodeToVariant(const SValueNode* pNode, SVariant* pVal);
+void*   nodesGetValueFromNode(SValueNode* pNode);
+int32_t nodesSetValueNodeValue(SValueNode* pNode, void* value);
+char*   nodesGetStrValueFromNode(SValueNode* pNode);
+char*   getFillModeString(EFillMode mode);
+void    valueNodeToVariant(const SValueNode* pNode, SVariant* pVal);
 
 #ifdef __cplusplus
 }

include/libs/sync/sync.h

@@ -89,7 +89,7 @@ typedef struct SSyncFSM {
 struct SSyncRaftEntry;
 typedef struct SSyncRaftEntry SSyncRaftEntry;
 
-#define SYNC_INDEX_BEGIN   0
+#define SYNC_INDEX_BEGIN 0
 #define SYNC_INDEX_INVALID -1
 
 // abstract definition of log store in raft
@@ -149,6 +149,8 @@ int32_t     syncReconfig(int64_t rid, const SSyncCfg* pSyncCfg);
 ESyncState  syncGetMyRole(int64_t rid);
 const char* syncGetMyRoleStr(int64_t rid);
 SyncTerm    syncGetMyTerm(int64_t rid);
+void        syncGetEpSet(int64_t rid, SEpSet* pEpSet);
+int32_t     syncGetVgId(int64_t rid);
 
 typedef enum {
   TAOS_SYNC_PROPOSE_SUCCESS = 0,

include/os/osDef.h

@@ -62,7 +62,7 @@ extern "C" {
   #define strncasecmp _strnicmp
   #define wcsncasecmp _wcsnicmp
   #define strtok_r strtok_s
-  #define snprintf _snprintf
+  // #define snprintf _snprintf
   #define in_addr_t unsigned long
 //  #define socklen_t int
 

include/util/taoserror.h

@@ -87,6 +87,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MSG_NOT_PROCESSED             TAOS_DEF_ERROR_CODE(0, 0x0113)
 #define TSDB_CODE_CFG_NOT_FOUND                 TAOS_DEF_ERROR_CODE(0, 0x0114)
 #define TSDB_CODE_REPEAT_INIT                   TAOS_DEF_ERROR_CODE(0, 0x0115)
+#define TSDB_CODE_DUP_KEY                       TAOS_DEF_ERROR_CODE(0, 0x0116)
 
 #define TSDB_CODE_REF_NO_MEMORY                 TAOS_DEF_ERROR_CODE(0, 0x0140)
 #define TSDB_CODE_REF_FULL                      TAOS_DEF_ERROR_CODE(0, 0x0141)
@@ -263,7 +264,8 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_TRANS_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x03D0)
 #define TSDB_CODE_MND_TRANS_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x03D1)
 #define TSDB_CODE_MND_TRANS_INVALID_STAGE       TAOS_DEF_ERROR_CODE(0, 0x03D2)
-#define TSDB_CODE_MND_TRANS_CANT_PARALLEL       TAOS_DEF_ERROR_CODE(0, 0x03D4)
+#define TSDB_CODE_MND_TRANS_CONFLICT            TAOS_DEF_ERROR_CODE(0, 0x03D3)
+#define TSDB_CODE_MND_TRANS_UNKNOW_ERROR        TAOS_DEF_ERROR_CODE(0, 0x03D4)
 
 // mnode-mq
 #define TSDB_CODE_MND_TOPIC_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x03E0)
@@ -318,6 +320,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_VND_TB_NOT_EXIST              TAOS_DEF_ERROR_CODE(0, 0x0515)
 #define TSDB_CODE_VND_SMA_NOT_EXIST             TAOS_DEF_ERROR_CODE(0, 0x0516)
 #define TSDB_CODE_VND_HASH_MISMATCH             TAOS_DEF_ERROR_CODE(0, 0x0517)
+#define TSDB_CODE_VND_TABLE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0518)
 
 // tsdb
 #define TSDB_CODE_TDB_INVALID_TABLE_ID          TAOS_DEF_ERROR_CODE(0, 0x0600)
@@ -616,9 +619,13 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_OFFSET_LESS_ZERO          TAOS_DEF_ERROR_CODE(0, 0x2637)
 #define TSDB_CODE_PAR_SLIMIT_LEAK_PARTITION_BY  TAOS_DEF_ERROR_CODE(0, 0x2638)
 #define TSDB_CODE_PAR_INVALID_TOPIC_QUERY       TAOS_DEF_ERROR_CODE(0, 0x2639)
+#define TSDB_CODE_PAR_INVALID_DROP_STABLE       TAOS_DEF_ERROR_CODE(0, 0x263A)
+#define TSDB_CODE_PAR_INVALID_FILL_TIME_RANGE   TAOS_DEF_ERROR_CODE(0, 0x263B)
 
 //planner
 #define TSDB_CODE_PLAN_INTERNAL_ERROR           TAOS_DEF_ERROR_CODE(0, 0x2700)
+#define TSDB_CODE_PLAN_EXPECTED_TS_EQUAL        TAOS_DEF_ERROR_CODE(0, 0x2701)
+#define TSDB_CODE_PLAN_NOT_SUPPORT_CROSS_JOIN   TAOS_DEF_ERROR_CODE(0, 0x2702)
 
 //function
 #define TSDB_CODE_FUNC_FUNTION_ERROR            TAOS_DEF_ERROR_CODE(0, 0x2800)

include/util/tencode.h

@@ -79,17 +79,6 @@ typedef struct {
 #define TD_CODER_CURRENT(CODER)                        ((CODER)->data + (CODER)->pos)
 #define TD_CODER_MOVE_POS(CODER, MOVE)                 ((CODER)->pos += (MOVE))
 #define TD_CODER_CHECK_CAPACITY_FAILED(CODER, EXPSIZE) (((CODER)->size - (CODER)->pos) < (EXPSIZE))
-// #define TCODER_MALLOC(PCODER, SIZE)                                         \
-//   ({                                                                        \
-//     void*      ptr = NULL;                                                  \
-//     SCoderMem* pMem = (SCoderMem*)taosMemoryMalloc(sizeof(*pMem) + (SIZE)); \
-//     if (pMem) {                                                             \
-//       pMem->next = (PCODER)->mList;                                         \
-//       (PCODER)->mList = pMem;                                               \
-//       ptr = (void*)&pMem[1];                                                \
-//     }                                                                       \
-//     ptr;                                                                    \
-//   })
 static FORCE_INLINE void* tCoderMalloc(SCoder* pCoder, int32_t size) {
   void*      ptr = NULL;
   SCoderMem* pMem = (SCoderMem*)taosMemoryMalloc(sizeof(SCoderMem*) + size);
@@ -102,8 +91,9 @@ static FORCE_INLINE void* tCoderMalloc(SCoder* pCoder, int32_t size) {
 }
 
 #define tEncodeSize(E, S, SIZE, RET)                           \
-  do{                                                          \
+  do {                                                         \
     SCoder coder = {0};                                        \
+    RET = 0;                                                   \
     tCoderInit(&coder, TD_LITTLE_ENDIAN, NULL, 0, TD_ENCODER); \
     if ((E)(&coder, S) == 0) {                                 \
       SIZE = coder.pos;                                        \
@@ -111,7 +101,7 @@ static FORCE_INLINE void* tCoderMalloc(SCoder* pCoder, int32_t size) {
       RET = -1;                                                \
     }                                                          \
     tCoderClear(&coder);                                       \
-  }while(0)
+  } while (0)
 // #define tEncodeSize(E, S, SIZE)                                \
 //   ({                                                           \
 //     SCoder coder = {0};                                        \

include/util/tskiplist.h

@@ -52,72 +52,21 @@ typedef struct SSkipListNode {
 #define SL_NODE_GET_FORWARD_POINTER(n, l)  (n)->forwards[(l)]
 #define SL_NODE_GET_BACKWARD_POINTER(n, l) (n)->forwards[(n)->level + (l)]
 
-/*
- * @version 0.3
- * @date   2017/11/12
- * the simple version of skip list.
- *
- * for multi-thread safe purpose, we employ TdThreadRwlock to guarantee to generate
- * deterministic result. Later, we will remove the lock in SkipList to further enhance the performance.
- * In this case, one should use the concurrent skip list (by using michael-scott algorithm) instead of
- * this simple version in a multi-thread environment, to achieve higher performance of read/write operations.
- *
- * Note: Duplicated primary key situation.
- * In case of duplicated primary key, two ways can be employed to handle this situation:
- * 1. add as normal insertion without special process.
- * 2. add an overflow pointer at each list node, all nodes with the same key will be added in the overflow pointer.
- *    In this case, the total steps of each search will be reduced significantly.
- *    Currently, we implement the skip list in a line with the first means, maybe refactor it soon.
- *
- *    Memory consumption: the memory alignment causes many memory wasted. So, employ a memory
- *    pool will significantly reduce the total memory consumption, as well as the calloc/malloc operation costs.
- *
- */
-
-// state struct, record following information:
-// number of links in each level.
-// avg search steps, for latest 1000 queries
-// avg search rsp time, for latest 1000 queries
-// total memory size
-typedef struct tSkipListState {
-  // in bytes, sizeof(SSkipList)+sizeof(SSkipListNode)*SSkipList->nSize
-  uint64_t nTotalMemSize;
-  uint64_t nLevelNodeCnt[MAX_SKIP_LIST_LEVEL];
-  uint64_t queryCount;  // total query count
-
-  /*
-   * only record latest 1000 queries
-   * when the value==1000, = 0,
-   * nTotalStepsForQueries = 0,
-   * nTotalElapsedTimeForQueries = 0
-   */
-  uint64_t nRecQueries;
-  uint16_t nTotalStepsForQueries;
-  uint64_t nTotalElapsedTimeForQueries;
-
-  uint16_t nInsertObjs;
-  uint16_t nTotalStepsForInsert;
-  uint64_t nTotalElapsedTimeForInsert;
-} tSkipListState;
-
 typedef enum { SSkipListPutSuccess = 0, SSkipListPutEarlyStop = 1, SSkipListPutSkipOne = 2 } SSkipListPutStatus;
 
 typedef struct SSkipList {
-  uint32_t          seed;
-  __compar_fn_t     comparFn;
-  __sl_key_fn_t     keyFn;
-  TdThreadRwlock *lock;
-  uint16_t          len;
-  uint8_t           maxLevel;
-  uint8_t           flags;
-  uint8_t           type;  // static info above
-  uint8_t           level;
-  uint32_t          size;
-  SSkipListNode    *pHead;  // point to the first element
-  SSkipListNode    *pTail;  // point to the last element
-#if SKIP_LIST_RECORD_PERFORMANCE
-  tSkipListState state;  // skiplist state
-#endif
+  uint32_t           seed;
+  __compar_fn_t      comparFn;
+  __sl_key_fn_t      keyFn;
+  TdThreadRwlock    *lock;
+  uint16_t           len;
+  uint8_t            maxLevel;
+  uint8_t            flags;
+  uint8_t            type;  // static info above
+  uint8_t            level;
+  uint32_t           size;
+  SSkipListNode     *pHead;  // point to the first element
+  SSkipListNode     *pTail;  // point to the last element
   tGenericSavedFunc *insertHandleFn;
 } SSkipList;
 

source/client/src/tmq.c

@@ -68,6 +68,7 @@ struct tmq_conf_t {
   char*          pass;
   char*          db;
   tmq_commit_cb* commitCb;
+  void*          commitCbUserParam;
 };
 
 struct tmq_t {
@@ -78,7 +79,8 @@ struct tmq_t {
   int32_t        autoCommitInterval;
   int32_t        resetOffsetCfg;
   int64_t        consumerId;
-  tmq_commit_cb* commit_cb;
+  tmq_commit_cb* commitCb;
+  void*          commitCbUserParam;
 
   // status
   int8_t  status;
@@ -372,10 +374,18 @@ int32_t tmqSubscribeCb(void* param, const SDataBuf* pMsg, int32_t code) {
 int32_t tmqCommitCb(void* param, const SDataBuf* pMsg, int32_t code) {
   SMqCommitCbParam* pParam = (SMqCommitCbParam*)param;
   pParam->rspErr = code == 0 ? TMQ_RESP_ERR__SUCCESS : TMQ_RESP_ERR__FAIL;
-  if (pParam->tmq->commit_cb) {
-    pParam->tmq->commit_cb(pParam->tmq, pParam->rspErr, NULL);
+  if (pParam->tmq->commitCb) {
+    pParam->tmq->commitCb(pParam->tmq, pParam->rspErr, NULL, pParam->tmq->commitCbUserParam);
+  }
+  if (!pParam->async)
+    tsem_post(&pParam->rspSem);
+  else {
+    tsem_destroy(&pParam->rspSem);
+    /*if (pParam->pArray) {*/
+    /*taosArrayDestroy(pParam->pArray);*/
+    /*}*/
+    taosMemoryFree(pParam);
   }
-  if (!pParam->async) tsem_post(&pParam->rspSem);
   return 0;
 }
 
@@ -384,7 +394,7 @@ tmq_resp_err_t tmq_subscription(tmq_t* tmq, tmq_list_t** topics) {
     *topics = tmq_list_new();
   }
   for (int i = 0; i < taosArrayGetSize(tmq->clientTopics); i++) {
-    SMqClientTopic* topic = taosArrayGetP(tmq->clientTopics, i);
+    SMqClientTopic* topic = taosArrayGet(tmq->clientTopics, i);
     tmq_list_append(*topics, topic->topicName);
   }
   return TMQ_RESP_ERR__SUCCESS;
@@ -477,7 +487,8 @@ tmq_t* tmq_consumer_new(tmq_conf_t* conf, char* errstr, int32_t errstrLen) {
   strcpy(pTmq->groupId, conf->groupId);
   pTmq->autoCommit = conf->autoCommit;
   pTmq->autoCommitInterval = conf->autoCommitInterval;
-  pTmq->commit_cb = conf->commitCb;
+  pTmq->commitCb = conf->commitCb;
+  pTmq->commitCbUserParam = conf->commitCbUserParam;
   pTmq->resetOffsetCfg = conf->resetOffset;
 
   // assign consumerId
@@ -557,7 +568,7 @@ tmq_resp_err_t tmq_commit(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, in
     tscError("failed to malloc request");
   }
 
-  SMqCommitCbParam* pParam = taosMemoryMalloc(sizeof(SMqCommitCbParam));
+  SMqCommitCbParam* pParam = taosMemoryCalloc(1, sizeof(SMqCommitCbParam));
   if (pParam == NULL) {
     return -1;
   }
@@ -572,6 +583,7 @@ tmq_resp_err_t tmq_commit(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, in
   };
 
   SMsgSendInfo* sendInfo = buildMsgInfoImpl(pRequest);
+  sendInfo->requestObjRefId = 0;
   sendInfo->param = pParam;
   sendInfo->fp = tmqCommitCb;
   SEpSet epSet = getEpSet_s(&tmq->pTscObj->pAppInfo->mgmtEp);
@@ -582,13 +594,12 @@ tmq_resp_err_t tmq_commit(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, in
   if (!async) {
     tsem_wait(&pParam->rspSem);
     resp = pParam->rspErr;
-  }
-
-  tsem_destroy(&pParam->rspSem);
-  taosMemoryFree(pParam);
+    tsem_destroy(&pParam->rspSem);
+    taosMemoryFree(pParam);
 
-  if (pArray) {
-    taosArrayDestroy(pArray);
+    if (pArray) {
+      taosArrayDestroy(pArray);
+    }
   }
 
   return resp;
@@ -688,9 +699,10 @@ FAIL:
   return code;
 }
 
-void tmq_conf_set_offset_commit_cb(tmq_conf_t* conf, tmq_commit_cb* cb) {
+void tmq_conf_set_offset_commit_cb(tmq_conf_t* conf, tmq_commit_cb* cb, void* param) {
   //
   conf->commitCb = cb;
+  conf->commitCbUserParam = param;
 }
 
 #if 0
@@ -1306,10 +1318,10 @@ const char* tmq_err2str(tmq_resp_err_t err) {
   return "fail";
 }
 
-char* tmq_get_topic_name(TAOS_RES* res) {
+const char* tmq_get_topic_name(TAOS_RES* res) {
   if (TD_RES_TMQ(res)) {
     SMqRspObj* pRspObj = (SMqRspObj*)res;
-    return pRspObj->topic;
+    return strchr(pRspObj->topic, '.') + 1;
   } else {
     return NULL;
   }

source/common/src/tdatablock.c

@@ -168,13 +168,6 @@ static void doBitmapMerge(SColumnInfoData* pColumnInfoData, int32_t numOfRow1, c
 
   uint32_t total = numOfRow1 + numOfRow2;
 
-  if (BitmapLen(numOfRow1) < BitmapLen(total)) {
-    char*    tmp = taosMemoryRealloc(pColumnInfoData->nullbitmap, BitmapLen(total));
-    uint32_t extend = BitmapLen(total) - BitmapLen(numOfRow1);
-    memset(tmp + BitmapLen(numOfRow1), 0, extend);
-    pColumnInfoData->nullbitmap = tmp;
-  }
-
   uint32_t remindBits = BitPos(numOfRow1);
   uint32_t shiftBits = 8 - remindBits;
 
@@ -194,25 +187,24 @@ static void doBitmapMerge(SColumnInfoData* pColumnInfoData, int32_t numOfRow1, c
   int32_t i = 0;
 
   uint8_t* start = (uint8_t*)&pColumnInfoData->nullbitmap[BitmapLen(numOfRow1)];
-  int32_t overCount = BitmapLen(total) - BitmapLen(numOfRow1);
-  while (i < len) { // size limit of pSource->nullbitmap
+  int32_t  overCount = BitmapLen(total) - BitmapLen(numOfRow1);
+  while (i < len) {  // size limit of pSource->nullbitmap
     if (i >= 1) {
-      start[i - 1] |= (p[i] >> remindBits); //copy remind bits
+      start[i - 1] |= (p[i] >> remindBits);  // copy remind bits
     }
 
-    if (i >= overCount) { // size limit of pColumnInfoData->nullbitmap
+    if (i >= overCount) {  // size limit of pColumnInfoData->nullbitmap
       return;
     }
 
-    start[i] |= (p[i] << shiftBits);  //copy shift bits
+    start[i] |= (p[i] << shiftBits);  // copy shift bits
     i += 1;
   }
 }
 
-int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, const SColumnInfoData* pSource,
-                        uint32_t numOfRow2) {
+int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, int32_t* capacity,
+                        const SColumnInfoData* pSource, uint32_t numOfRow2) {
   ASSERT(pColumnInfoData != NULL && pSource != NULL && pColumnInfoData->info.type == pSource->info.type);
-
   if (numOfRow2 == 0) {
     return numOfRow1;
   }
@@ -221,14 +213,19 @@ int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, co
     pColumnInfoData->hasNull = pSource->hasNull;
   }
 
+  uint32_t finalNumOfRows = numOfRow1 + numOfRow2;
   if (IS_VAR_DATA_TYPE(pColumnInfoData->info.type)) {
     // Handle the bitmap
-    char* p = taosMemoryRealloc(pColumnInfoData->varmeta.offset, sizeof(int32_t) * (numOfRow1 + numOfRow2));
-    if (p == NULL) {
-      return TSDB_CODE_OUT_OF_MEMORY;
+    if (finalNumOfRows > *capacity) {
+      char* p = taosMemoryRealloc(pColumnInfoData->varmeta.offset, sizeof(int32_t) * (numOfRow1 + numOfRow2));
+      if (p == NULL) {
+        return TSDB_CODE_OUT_OF_MEMORY;
+      }
+
+      *capacity = finalNumOfRows;
+      pColumnInfoData->varmeta.offset = (int32_t*)p;
     }
 
-    pColumnInfoData->varmeta.offset = (int32_t*)p;
     for (int32_t i = 0; i < numOfRow2; ++i) {
       if (pSource->varmeta.offset[i] == -1) {
         pColumnInfoData->varmeta.offset[i + numOfRow1] = -1;
@@ -253,15 +250,27 @@ int32_t colDataMergeCol(SColumnInfoData* pColumnInfoData, uint32_t numOfRow1, co
     memcpy(pColumnInfoData->pData + oldLen, pSource->pData, len);
     pColumnInfoData->varmeta.length = len + oldLen;
   } else {
-    doBitmapMerge(pColumnInfoData, numOfRow1, pSource, numOfRow2);
+    if (finalNumOfRows > *capacity) {
+      char* tmp = taosMemoryRealloc(pColumnInfoData->pData, finalNumOfRows * pColumnInfoData->info.bytes);
+      if (tmp == NULL) {
+        return TSDB_CODE_VND_OUT_OF_MEMORY;
+      }
 
-    int32_t newSize = (numOfRow1 + numOfRow2) * pColumnInfoData->info.bytes;
-    char*   tmp = taosMemoryRealloc(pColumnInfoData->pData, newSize);
-    if (tmp == NULL) {
-      return TSDB_CODE_VND_OUT_OF_MEMORY;
+      pColumnInfoData->pData = tmp;
+
+      if (BitmapLen(numOfRow1) < BitmapLen(finalNumOfRows)) {
+        char*    btmp = taosMemoryRealloc(pColumnInfoData->nullbitmap, BitmapLen(finalNumOfRows));
+        uint32_t extend = BitmapLen(finalNumOfRows) - BitmapLen(numOfRow1);
+        memset(btmp + BitmapLen(numOfRow1), 0, extend);
+
+        pColumnInfoData->nullbitmap = btmp;
+      }
+
+      *capacity = finalNumOfRows;
     }
 
-    pColumnInfoData->pData = tmp;
+    doBitmapMerge(pColumnInfoData, numOfRow1, pSource, numOfRow2);
+
     int32_t offset = pColumnInfoData->info.bytes * numOfRow1;
     memcpy(pColumnInfoData->pData + offset, pSource->pData, pSource->info.bytes * numOfRow2);
   }
@@ -350,29 +359,22 @@ int32_t blockDataUpdateTsWindow(SSDataBlock* pDataBlock) {
 // if pIndexMap = NULL, merger one column by on column
 int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc, SArray* pIndexMap) {
   assert(pSrc != NULL && pDest != NULL);
+  int32_t capacity = pDest->info.capacity;
 
-  int32_t numOfCols = pDest->info.numOfCols;
-  for (int32_t i = 0; i < numOfCols; ++i) {
+  for (int32_t i = 0; i < pDest->info.numOfCols; ++i) {
     int32_t mapIndex = i;
-    if(pIndexMap) {
+    if (pIndexMap) {
       mapIndex = *(int32_t*)taosArrayGet(pIndexMap, i);
     }
+
     SColumnInfoData* pCol2 = taosArrayGet(pDest->pDataBlock, i);
     SColumnInfoData* pCol1 = taosArrayGet(pSrc->pDataBlock, mapIndex);
 
-    uint32_t oldLen = colDataGetLength(pCol2, pDest->info.rows);
-    uint32_t newLen = colDataGetLength(pCol1, pSrc->info.rows);
-
-    int32_t newSize = oldLen + newLen;
-    char*   tmp = taosMemoryRealloc(pCol2->pData, newSize);
-    if (tmp != NULL) {
-      pCol2->pData = tmp;
-      colDataMergeCol(pCol2, pDest->info.rows, pCol1, pSrc->info.rows);
-    } else {
-      return TSDB_CODE_VND_OUT_OF_MEMORY;
-    }
+    capacity = pDest->info.capacity;
+    colDataMergeCol(pCol2, pDest->info.rows, &capacity, pCol1, pSrc->info.rows);
   }
 
+  pDest->info.capacity = capacity;
   pDest->info.rows += pSrc->info.rows;
   return TSDB_CODE_SUCCESS;
 }
@@ -451,7 +453,6 @@ int32_t blockDataSplitRows(SSDataBlock* pBlock, bool hasVarCol, int32_t startInd
   // all fit in
   *stopIndex = numOfRows - 1;
   return TSDB_CODE_SUCCESS;
-
 }
 
 SSDataBlock* blockDataExtractBlock(SSDataBlock* pBlock, int32_t startIndex, int32_t rowCount) {
@@ -490,7 +491,7 @@ SSDataBlock* blockDataExtractBlock(SSDataBlock* pBlock, int32_t startIndex, int3
     SColumnInfoData* pDstCol = taosArrayGet(pDst->pDataBlock, i);
 
     for (int32_t j = startIndex; j < (startIndex + rowCount); ++j) {
-      bool  isNull = colDataIsNull(pColData, pBlock->info.rows, j, pBlock->pBlockAgg);
+      bool  isNull = colDataIsNull(pColData, pBlock->info.rows, j, pBlock->pBlockAgg[i]);
       char* p = colDataGetData(pColData, j);
 
       colDataAppend(pDstCol, j - startIndex, p, isNull);
@@ -556,7 +557,7 @@ int32_t blockDataFromBuf(SSDataBlock* pBlock, const char* buf) {
 
     if (IS_VAR_DATA_TYPE(pCol->info.type)) {
       size_t metaSize = pBlock->info.rows * sizeof(int32_t);
-      char* tmp = taosMemoryRealloc(pCol->varmeta.offset, metaSize);    // preview calloc is too small
+      char*  tmp = taosMemoryRealloc(pCol->varmeta.offset, metaSize);  // preview calloc is too small
       if (tmp == NULL) {
         return TSDB_CODE_OUT_OF_MEMORY;
       }
@@ -702,8 +703,8 @@ int32_t dataBlockCompar(const void* p1, const void* p2, const void* param) {
     SColumnInfoData* pColInfoData = pOrder->pColData;  // TARRAY_GET_ELEM(pDataBlock->pDataBlock, pOrder->colIndex);
 
     if (pColInfoData->hasNull) {
-      bool leftNull = colDataIsNull(pColInfoData, pDataBlock->info.rows, left, pDataBlock->pBlockAgg);
-      bool rightNull = colDataIsNull(pColInfoData, pDataBlock->info.rows, right, pDataBlock->pBlockAgg);
+      bool leftNull = colDataIsNull(pColInfoData, pDataBlock->info.rows, left, NULL);
+      bool rightNull = colDataIsNull(pColInfoData, pDataBlock->info.rows, right, NULL);
       if (leftNull && rightNull) {
         continue;  // continue to next slot
       }
@@ -742,7 +743,7 @@ static int32_t doAssignOneTuple(SColumnInfoData* pDstCols, int32_t numOfRows, co
     SColumnInfoData* pDst = &pDstCols[i];
     SColumnInfoData* pSrc = taosArrayGet(pSrcBlock->pDataBlock, i);
 
-    if (pSrc->hasNull && colDataIsNull(pSrc, pSrcBlock->info.rows, tupleIndex, pSrcBlock->pBlockAgg)) {
+    if (pSrc->hasNull && colDataIsNull(pSrc, pSrcBlock->info.rows, tupleIndex, pSrcBlock->pBlockAgg[i])) {
       code = colDataAppend(pDst, numOfRows, NULL, true);
       if (code != TSDB_CODE_SUCCESS) {
         return code;
@@ -938,8 +939,9 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
   copyBackToBlock(pDataBlock, pCols);
   int64_t p4 = taosGetTimestampUs();
 
-  uDebug("blockDataSort complex sort:%" PRId64 ", create:%" PRId64 ", assign:%" PRId64 ", copyback:%" PRId64 ", rows:%d\n", p1 - p0, p2 - p1,
-         p3 - p2, p4 - p3, rows);
+  uDebug("blockDataSort complex sort:%" PRId64 ", create:%" PRId64 ", assign:%" PRId64 ", copyback:%" PRId64
+         ", rows:%d\n",
+         p1 - p0, p2 - p1, p3 - p2, p4 - p3, rows);
   destroyTupleIndex(index);
 
   return TSDB_CODE_SUCCESS;
@@ -1176,7 +1178,7 @@ void* blockDataDestroy(SSDataBlock* pBlock) {
 }
 
 SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData) {
-  if(pDataBlock == NULL){
+  if (pDataBlock == NULL) {
     return NULL;
   }
 
@@ -1187,7 +1189,7 @@ SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData) {
 
   pBlock->info.numOfCols = numOfCols;
   pBlock->info.hasVarCol = pDataBlock->info.hasVarCol;
-  pBlock->info.rowSize   = pDataBlock->info.rows;
+  pBlock->info.rowSize = pDataBlock->info.rows;
 
   for (int32_t i = 0; i < numOfCols; ++i) {
     SColumnInfoData  colInfo = {0};
@@ -1217,7 +1219,7 @@ SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData) {
 }
 
 size_t blockDataGetCapacityInRow(const SSDataBlock* pBlock, size_t pageSize) {
-  return (int32_t) ((pageSize - blockDataGetSerialMetaSize(pBlock))/ blockDataGetSerialRowSize(pBlock));
+  return (int32_t)((pageSize - blockDataGetSerialMetaSize(pBlock)) / blockDataGetSerialRowSize(pBlock));
 }
 
 void colDataDestroy(SColumnInfoData* pColData) {
@@ -1234,14 +1236,14 @@ static void doShiftBitmap(char* nullBitmap, size_t n, size_t total) {
   int32_t len = BitmapLen(total);
 
   int32_t newLen = BitmapLen(total - n);
-  if (n%8 == 0) {
-    memmove(nullBitmap, nullBitmap + n/8, newLen);
+  if (n % 8 == 0) {
+    memmove(nullBitmap, nullBitmap + n / 8, newLen);
   } else {
     int32_t tail = n % 8;
     int32_t i = 0;
 
-    uint8_t* p = (uint8_t*) nullBitmap;
-    while(i < len) {
+    uint8_t* p = (uint8_t*)nullBitmap;
+    while (i < len) {
       uint8_t v = p[i];
 
       p[i] = 0;
@@ -1268,7 +1270,7 @@ static void colDataTrimFirstNRows(SColumnInfoData* pColInfoData, size_t n, size_
   }
 }
 
-int32_t blockDataTrimFirstNRows(SSDataBlock *pBlock, size_t n) {
+int32_t blockDataTrimFirstNRows(SSDataBlock* pBlock, size_t n) {
   if (n == 0) {
     return TSDB_CODE_SUCCESS;
   }
@@ -1276,7 +1278,7 @@ int32_t blockDataTrimFirstNRows(SSDataBlock *pBlock, size_t n) {
   if (pBlock->info.rows <= n) {
     blockDataCleanup(pBlock);
   } else {
-    for(int32_t i = 0; i < pBlock->info.numOfCols; ++i) {
+    for (int32_t i = 0; i < pBlock->info.numOfCols; ++i) {
       SColumnInfoData* pColInfoData = taosArrayGet(pBlock->pDataBlock, i);
       colDataTrimFirstNRows(pColInfoData, n, pBlock->info.rows);
     }
@@ -1462,3 +1464,128 @@ void blockDebugShowData(const SArray* dataBlocks) {
   }
 }
 
+/**
+ * @brief TODO: Assume that the final generated result it less than 3M
+ *
+ * @param pReq
+ * @param pDataBlocks
+ * @param vgId
+ * @param uid set as parameter temporarily // TODO: remove this parameter, and the executor should set uid in
+ * SDataBlock->info.uid
+ * @param suid  // TODO: check with Liao whether suid response is reasonable
+ * 
+ * TODO: colId should be set
+ */
+int32_t buildSubmitReqFromDataBlock(SSubmitReq** pReq, const SArray* pDataBlocks, STSchema *pTSchema, int32_t vgId, tb_uid_t uid,
+                                        tb_uid_t suid) {
+  int32_t sz = taosArrayGetSize(pDataBlocks);
+  int32_t bufSize = sizeof(SSubmitReq);
+  for (int32_t i = 0; i < sz; ++i) {
+    SDataBlockInfo* pBlkInfo = &((SSDataBlock*)taosArrayGet(pDataBlocks, i))->info;
+    bufSize += pBlkInfo->rows * (TD_ROW_HEAD_LEN + pBlkInfo->rowSize + BitmapLen(pBlkInfo->numOfCols));
+    bufSize += sizeof(SSubmitBlk);
+  }
+
+  ASSERT(bufSize < 3 * 1024 * 1024);
+
+  *pReq = taosMemoryCalloc(1, bufSize);
+  if(!(*pReq)) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return TSDB_CODE_FAILED;
+  }
+  void* pDataBuf = *pReq;
+
+  int32_t msgLen = sizeof(SSubmitReq);
+  int32_t     numOfBlks = 0;
+  SRowBuilder rb = {0};
+  tdSRowInit(&rb, 0); // TODO: use the latest version
+
+  for (int32_t i = 0; i < sz; ++i) {
+    SSDataBlock* pDataBlock = taosArrayGet(pDataBlocks, i);
+    int32_t      colNum = pDataBlock->info.numOfCols;
+    int32_t      rows = pDataBlock->info.rows;
+    int32_t      rowSize = pDataBlock->info.rowSize;
+    int64_t      groupId = pDataBlock->info.groupId;
+    
+    if(rb.nCols != colNum) {
+      tdSRowSetTpInfo(&rb, colNum, pTSchema->flen);
+    }
+
+    SSubmitBlk* pSubmitBlk = POINTER_SHIFT(pDataBuf, msgLen);
+    pSubmitBlk->suid = suid;
+    pSubmitBlk->uid = uid;
+    pSubmitBlk->numOfRows = rows;
+
+    ++numOfBlks;
+
+    msgLen += sizeof(SSubmitBlk);
+    int32_t dataLen = 0;
+    for (int32_t j = 0; j < rows; ++j) {     // iterate by row
+      tdSRowResetBuf(&rb, POINTER_SHIFT(pDataBuf, msgLen)); // set row buf
+      printf("|");
+      bool  isStartKey = false;
+      for (int32_t k = 0; k < colNum; ++k) {  // iterate by column
+        SColumnInfoData* pColInfoData = taosArrayGet(pDataBlock->pDataBlock, k);
+        void*            var = POINTER_SHIFT(pColInfoData->pData, j * pColInfoData->info.bytes);
+        switch (pColInfoData->info.type) {
+          case TSDB_DATA_TYPE_TIMESTAMP:
+            if (!isStartKey) {
+              isStartKey = true;
+              tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID, TSDB_DATA_TYPE_TIMESTAMP, TD_VTYPE_NORM, var, true, 0, 0);
+            } else {
+              tdAppendColValToRow(&rb, 2, TSDB_DATA_TYPE_TIMESTAMP, TD_VTYPE_NORM, var, true, 8, k);
+              break;
+            }
+            break;
+          case TSDB_DATA_TYPE_NCHAR: {
+            tdAppendColValToRow(&rb, 2, TSDB_DATA_TYPE_NCHAR, TD_VTYPE_NORM, var, true, 8, k);
+            break;
+          }
+          case TSDB_DATA_TYPE_VARCHAR: {  // TSDB_DATA_TYPE_BINARY
+            tdAppendColValToRow(&rb, 2, TSDB_DATA_TYPE_VARCHAR, TD_VTYPE_NORM, var, true, 8, k);
+            break;
+          }
+          case TSDB_DATA_TYPE_VARBINARY:
+          case TSDB_DATA_TYPE_DECIMAL:
+          case TSDB_DATA_TYPE_BLOB:
+          case TSDB_DATA_TYPE_MEDIUMBLOB:
+            printf("the column type %" PRIi16 " is defined but not implemented yet\n", pColInfoData->info.type);
+            TASSERT(0);
+            break;
+          default:
+            if (pColInfoData->info.type < TSDB_DATA_TYPE_MAX && pColInfoData->info.type > TSDB_DATA_TYPE_NULL) {
+              tdAppendColValToRow(&rb, 2, pColInfoData->info.type, TD_VTYPE_NORM, var, true, 8, k);
+            } else {
+              printf("the column type %" PRIi16 " is undefined\n", pColInfoData->info.type);
+              TASSERT(0);
+            }
+            break;
+        }
+      }
+      dataLen += TD_ROW_LEN(rb.pBuf);
+    }
+    pSubmitBlk->dataLen = dataLen;
+    msgLen += pSubmitBlk->dataLen;
+  }
+
+  (*pReq)->length = msgLen;
+
+  (*pReq)->header.vgId = htonl(vgId);
+  (*pReq)->header.contLen = htonl(msgLen);
+  (*pReq)->length = (*pReq)->header.contLen;
+  (*pReq)->numOfBlocks = htonl(numOfBlks);
+  SSubmitBlk* blk = (SSubmitBlk*)((*pReq) + 1);
+  while (numOfBlks--) {
+    int32_t dataLen = blk->dataLen;
+    blk->uid = htobe64(blk->uid);
+    blk->suid = htobe64(blk->suid);
+    blk->padding = htonl(blk->padding);
+    blk->sversion = htonl(blk->sversion);
+    blk->dataLen = htonl(blk->dataLen);
+    blk->schemaLen = htonl(blk->schemaLen);
+    blk->numOfRows = htons(blk->numOfRows);
+    blk = (SSubmitBlk*)(blk->data + dataLen);
+  }
+
+  return TSDB_CODE_SUCCESS;
+}

source/common/src/tdataformat.c

@@ -87,7 +87,7 @@ int tdEncodeSchema(void **buf, STSchema *pSchema) {
   for (int i = 0; i < schemaNCols(pSchema); i++) {
     STColumn *pCol = schemaColAt(pSchema, i);
     tlen += taosEncodeFixedI8(buf, colType(pCol));
-    tlen += taosEncodeFixedI8(buf, colSma(pCol));
+    tlen += taosEncodeFixedI8(buf, colFlags(pCol));
     tlen += taosEncodeFixedI16(buf, colColId(pCol));
     tlen += taosEncodeFixedI16(buf, colBytes(pCol));
   }
@@ -110,14 +110,14 @@ void *tdDecodeSchema(void *buf, STSchema **pRSchema) {
 
   for (int i = 0; i < numOfCols; i++) {
     col_type_t  type = 0;
-    int8_t      sma = 0;
+    int8_t      flags = 0;
     col_id_t    colId = 0;
     col_bytes_t bytes = 0;
     buf = taosDecodeFixedI8(buf, &type);
-    buf = taosDecodeFixedI8(buf, &sma);
+    buf = taosDecodeFixedI8(buf, &flags);
     buf = taosDecodeFixedI16(buf, &colId);
     buf = taosDecodeFixedI32(buf, &bytes);
-    if (tdAddColToSchema(&schemaBuilder, type, sma, colId, bytes) < 0) {
+    if (tdAddColToSchema(&schemaBuilder, type, flags, colId, bytes) < 0) {
       tdDestroyTSchemaBuilder(&schemaBuilder);
       return NULL;
     }
@@ -153,7 +153,7 @@ void tdResetTSchemaBuilder(STSchemaBuilder *pBuilder, schema_ver_t version) {
   pBuilder->version = version;
 }
 
-int32_t tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t sma, col_id_t colId, col_bytes_t bytes) {
+int32_t tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t flags, col_id_t colId, col_bytes_t bytes) {
   if (!isValidDataType(type)) return -1;
 
   if (pBuilder->nCols >= pBuilder->tCols) {
@@ -166,7 +166,7 @@ int32_t tdAddColToSchema(STSchemaBuilder *pBuilder, int8_t type, int8_t sma, col
   STColumn *pCol = &(pBuilder->columns[pBuilder->nCols]);
   colSetType(pCol, type);
   colSetColId(pCol, colId);
-  colSetSma(pCol, sma);
+  colSetFlags(pCol, flags);
   if (pBuilder->nCols == 0) {
     colSetOffset(pCol, 0);
   } else {

source/common/src/tglobal.c

@@ -169,6 +169,10 @@ uint32_t tsMaxRange = 500;                      // max range
 uint32_t tsCurRange = 100;                      // range
 char     tsCompressor[32] = "ZSTD_COMPRESSOR";  // ZSTD_COMPRESSOR or GZIP_COMPRESSOR
 
+// internal
+int32_t tsTransPullupMs = 6000;
+int32_t tsMaRebalanceMs = 2000;
+
 void taosAddDataDir(int32_t index, char *v1, int32_t level, int32_t primary) {
   tstrncpy(tsDiskCfg[index].dir, v1, TSDB_FILENAME_LEN);
   tsDiskCfg[index].level = level;

source/common/src/tmsg.c

@@ -34,77 +34,6 @@ int32_t tInitSubmitMsgIter(const SSubmitReq *pMsg, SSubmitMsgIter *pIter) {
     return -1;
   }
 
-  pIter->totalLen = pMsg->length;
-  ASSERT(pIter->totalLen > 0);
-  pIter->len = 0;
-  pIter->pMsg = pMsg;
-  if (pMsg->length <= sizeof(SSubmitReq)) {
-    terrno = TSDB_CODE_TDB_SUBMIT_MSG_MSSED_UP;
-    return -1;
-  }
-
-  return 0;
-}
-
-int32_t tGetSubmitMsgNext(SSubmitMsgIter *pIter, SSubmitBlk **pPBlock) {
-  ASSERT(pIter->len >= 0);
-
-  if (pIter->len == 0) {
-    pIter->len += sizeof(SSubmitReq);
-  } else {
-    if (pIter->len >= pIter->totalLen) {
-      ASSERT(0);
-    }
-
-    SSubmitBlk *pSubmitBlk = (SSubmitBlk *)POINTER_SHIFT(pIter->pMsg, pIter->len);
-    pIter->len += (sizeof(SSubmitBlk) + pSubmitBlk->dataLen + pSubmitBlk->schemaLen);
-    ASSERT(pIter->len > 0);
-  }
-
-  if (pIter->len > pIter->totalLen) {
-    terrno = TSDB_CODE_TDB_SUBMIT_MSG_MSSED_UP;
-    *pPBlock = NULL;
-    return -1;
-  }
-
-  *pPBlock = (pIter->len == pIter->totalLen) ? NULL : (SSubmitBlk *)POINTER_SHIFT(pIter->pMsg, pIter->len);
-
-  return 0;
-}
-
-int32_t tInitSubmitBlkIter(SSubmitBlk *pBlock, SSubmitBlkIter *pIter) {
-  if (pBlock->dataLen <= 0) return -1;
-  pIter->totalLen = pBlock->dataLen;
-  pIter->len = 0;
-  pIter->row = (STSRow *)(pBlock->data + pBlock->schemaLen);
-  return 0;
-}
-
-STSRow *tGetSubmitBlkNext(SSubmitBlkIter *pIter) {
-  STSRow *row = pIter->row;
-
-  if (pIter->len >= pIter->totalLen) {
-    return NULL;
-  } else {
-    pIter->len += TD_ROW_LEN(row);
-    if (pIter->len < pIter->totalLen) {
-      pIter->row = POINTER_SHIFT(row, TD_ROW_LEN(row));
-    }
-    return row;
-  }
-}
-
-// TODO: KEEP one suite of iterator API finally.
-// 1) use tInitSubmitMsgIterEx firstly as not decrease the merge conflicts
-// 2) replace tInitSubmitMsgIterEx with tInitSubmitMsgIter later
-// 3) finally, rename tInitSubmitMsgIterEx to tInitSubmitMsgIter
-
-int32_t tInitSubmitMsgIterEx(const SSubmitReq *pMsg, SSubmitMsgIter *pIter) {
-  if (pMsg == NULL) {
-    terrno = TSDB_CODE_TDB_SUBMIT_MSG_MSSED_UP;
-    return -1;
-  }
-
   pIter->totalLen = htonl(pMsg->length);
   ASSERT(pIter->totalLen > 0);
   pIter->len = 0;
@@ -117,7 +46,7 @@ int32_t tInitSubmitMsgIterEx(const SSubmitReq *pMsg, SSubmitMsgIter *pIter) {
   return 0;
 }
 
-int32_t tGetSubmitMsgNextEx(SSubmitMsgIter *pIter, SSubmitBlk **pPBlock) {
+int32_t tGetSubmitMsgNext(SSubmitMsgIter *pIter, SSubmitBlk **pPBlock) {
   ASSERT(pIter->len >= 0);
 
   if (pIter->len == 0) {
@@ -152,7 +81,7 @@ int32_t tGetSubmitMsgNextEx(SSubmitMsgIter *pIter, SSubmitBlk **pPBlock) {
   return 0;
 }
 
-int32_t tInitSubmitBlkIterEx(SSubmitMsgIter *pMsgIter, SSubmitBlk *pBlock, SSubmitBlkIter *pIter) {
+int32_t tInitSubmitBlkIter(SSubmitMsgIter *pMsgIter, SSubmitBlk *pBlock, SSubmitBlkIter *pIter) {
   if (pMsgIter->dataLen <= 0) return -1;
   pIter->totalLen = pMsgIter->dataLen;
   pIter->len = 0;
@@ -160,7 +89,7 @@ int32_t tInitSubmitBlkIterEx(SSubmitMsgIter *pMsgIter, SSubmitBlk *pBlock, SSubm
   return 0;
 }
 
-STSRow *tGetSubmitBlkNextEx(SSubmitBlkIter *pIter) {
+STSRow *tGetSubmitBlkNext(SSubmitBlkIter *pIter) {
   STSRow *row = pIter->row;
 
   if (pIter->len >= pIter->totalLen) {
@@ -490,21 +419,6 @@ int32_t tDeserializeSClientHbBatchRsp(void *buf, int32_t bufLen, SClientHbBatchR
   return 0;
 }
 
-int32_t tSerializeSVDropTbReq(void **buf, SVDropTbReq *pReq) {
-  int32_t tlen = 0;
-  tlen += taosEncodeFixedI64(buf, pReq->ver);
-  tlen += taosEncodeString(buf, pReq->name);
-  tlen += taosEncodeFixedU8(buf, pReq->type);
-  return tlen;
-}
-
-void *tDeserializeSVDropTbReq(void *buf, SVDropTbReq *pReq) {
-  buf = taosDecodeFixedI64(buf, &pReq->ver);
-  buf = taosDecodeString(buf, &pReq->name);
-  buf = taosDecodeFixedU8(buf, &pReq->type);
-  return buf;
-}
-
 int32_t tSerializeSMCreateStbReq(void *buf, int32_t bufLen, SMCreateStbReq *pReq) {
   SCoder encoder = {0};
   tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);
@@ -1697,8 +1611,8 @@ int32_t tSerializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq) {
   if (tEncodeI32(&encoder, pReq->numOfRetensions) < 0) return -1;
   for (int32_t i = 0; i < pReq->numOfRetensions; ++i) {
     SRetention *pRetension = taosArrayGet(pReq->pRetensions, i);
-    if (tEncodeI32(&encoder, pRetension->freq) < 0) return -1;
-    if (tEncodeI32(&encoder, pRetension->keep) < 0) return -1;
+    if (tEncodeI64(&encoder, pRetension->freq) < 0) return -1;
+    if (tEncodeI64(&encoder, pRetension->keep) < 0) return -1;
     if (tEncodeI8(&encoder, pRetension->freqUnit) < 0) return -1;
     if (tEncodeI8(&encoder, pRetension->keepUnit) < 0) return -1;
   }
@@ -1743,8 +1657,8 @@ int32_t tDeserializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq)
 
   for (int32_t i = 0; i < pReq->numOfRetensions; ++i) {
     SRetention rentension = {0};
-    if (tDecodeI32(&decoder, &rentension.freq) < 0) return -1;
-    if (tDecodeI32(&decoder, &rentension.keep) < 0) return -1;
+    if (tDecodeI64(&decoder, &rentension.freq) < 0) return -1;
+    if (tDecodeI64(&decoder, &rentension.keep) < 0) return -1;
     if (tDecodeI8(&decoder, &rentension.freqUnit) < 0) return -1;
     if (tDecodeI8(&decoder, &rentension.keepUnit) < 0) return -1;
     if (taosArrayPush(pReq->pRetensions, &rentension) == NULL) {
@@ -2173,8 +2087,8 @@ int32_t tSerializeSDbCfgRsp(void *buf, int32_t bufLen, const SDbCfgRsp *pRsp) {
   if (tEncodeI32(&encoder, pRsp->numOfRetensions) < 0) return -1;
   for (int32_t i = 0; i < pRsp->numOfRetensions; ++i) {
     SRetention *pRetension = taosArrayGet(pRsp->pRetensions, i);
-    if (tEncodeI32(&encoder, pRetension->freq) < 0) return -1;
-    if (tEncodeI32(&encoder, pRetension->keep) < 0) return -1;
+    if (tEncodeI64(&encoder, pRetension->freq) < 0) return -1;
+    if (tEncodeI64(&encoder, pRetension->keep) < 0) return -1;
     if (tEncodeI8(&encoder, pRetension->freqUnit) < 0) return -1;
     if (tEncodeI8(&encoder, pRetension->keepUnit) < 0) return -1;
   }
@@ -2217,8 +2131,8 @@ int32_t tDeserializeSDbCfgRsp(void *buf, int32_t bufLen, SDbCfgRsp *pRsp) {
 
   for (int32_t i = 0; i < pRsp->numOfRetensions; ++i) {
     SRetention rentension = {0};
-    if (tDecodeI32(&decoder, &rentension.freq) < 0) return -1;
-    if (tDecodeI32(&decoder, &rentension.keep) < 0) return -1;
+    if (tDecodeI64(&decoder, &rentension.freq) < 0) return -1;
+    if (tDecodeI64(&decoder, &rentension.keep) < 0) return -1;
     if (tDecodeI8(&decoder, &rentension.freqUnit) < 0) return -1;
     if (tDecodeI8(&decoder, &rentension.keepUnit) < 0) return -1;
     if (taosArrayPush(pRsp->pRetensions, &rentension) == NULL) {
@@ -2835,8 +2749,8 @@ int32_t tSerializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *pR
   if (tEncodeI32(&encoder, pReq->numOfRetensions) < 0) return -1;
   for (int32_t i = 0; i < pReq->numOfRetensions; ++i) {
     SRetention *pRetension = taosArrayGet(pReq->pRetensions, i);
-    if (tEncodeI32(&encoder, pRetension->freq) < 0) return -1;
-    if (tEncodeI32(&encoder, pRetension->keep) < 0) return -1;
+    if (tEncodeI64(&encoder, pRetension->freq) < 0) return -1;
+    if (tEncodeI64(&encoder, pRetension->keep) < 0) return -1;
     if (tEncodeI8(&encoder, pRetension->freqUnit) < 0) return -1;
     if (tEncodeI8(&encoder, pRetension->keepUnit) < 0) return -1;
   }
@@ -2892,8 +2806,8 @@ int32_t tDeserializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *
 
   for (int32_t i = 0; i < pReq->numOfRetensions; ++i) {
     SRetention rentension = {0};
-    if (tDecodeI32(&decoder, &rentension.freq) < 0) return -1;
-    if (tDecodeI32(&decoder, &rentension.keep) < 0) return -1;
+    if (tDecodeI64(&decoder, &rentension.freq) < 0) return -1;
+    if (tDecodeI64(&decoder, &rentension.keep) < 0) return -1;
     if (tDecodeI8(&decoder, &rentension.freqUnit) < 0) return -1;
     if (tDecodeI8(&decoder, &rentension.keepUnit) < 0) return -1;
     if (taosArrayPush(pReq->pRetensions, &rentension) == NULL) {
@@ -3811,3 +3725,119 @@ int tDecodeSVCreateTbRsp(SCoder *pCoder, SVCreateTbRsp *pRsp) {
   tEndDecode(pCoder);
   return 0;
 }
+
+// TDMT_VND_DROP_TABLE =================
+static int32_t tEncodeSVDropTbReq(SCoder *pCoder, const SVDropTbReq *pReq) {
+  if (tStartEncode(pCoder) < 0) return -1;
+
+  if (tEncodeCStr(pCoder, pReq->name) < 0) return -1;
+  if (tEncodeI8(pCoder, pReq->igNotExists) < 0) return -1;
+
+  tEndEncode(pCoder);
+  return 0;
+}
+
+static int32_t tDecodeSVDropTbReq(SCoder *pCoder, SVDropTbReq *pReq) {
+  if (tStartDecode(pCoder) < 0) return -1;
+
+  if (tDecodeCStr(pCoder, &pReq->name) < 0) return -1;
+  if (tDecodeI8(pCoder, &pReq->igNotExists) < 0) return -1;
+
+  tEndDecode(pCoder);
+  return 0;
+}
+
+static int32_t tEncodeSVDropTbRsp(SCoder *pCoder, const SVDropTbRsp *pReq) {
+  if (tStartEncode(pCoder) < 0) return -1;
+
+  if (tEncodeI32(pCoder, pReq->code) < 0) return -1;
+
+  tEndEncode(pCoder);
+  return 0;
+}
+
+static int32_t tDecodeSVDropTbRsp(SCoder *pCoder, SVDropTbRsp *pReq) {
+  if (tStartDecode(pCoder) < 0) return -1;
+
+  if (tDecodeI32(pCoder, &pReq->code) < 0) return -1;
+
+  tEndDecode(pCoder);
+  return 0;
+}
+
+int32_t tEncodeSVDropTbBatchReq(SCoder *pCoder, const SVDropTbBatchReq *pReq) {
+  int32_t      nReqs = taosArrayGetSize(pReq->pArray);
+  SVDropTbReq *pDropTbReq;
+
+  if (tStartEncode(pCoder) < 0) return -1;
+
+  if (tEncodeI32v(pCoder, nReqs) < 0) return -1;
+  for (int iReq = 0; iReq < nReqs; iReq++) {
+    pDropTbReq = (SVDropTbReq *)taosArrayGet(pReq->pArray, iReq);
+    if (tEncodeSVDropTbReq(pCoder, pDropTbReq) < 0) return -1;
+  }
+
+  tEndEncode(pCoder);
+  return 0;
+}
+
+int32_t tDecodeSVDropTbBatchReq(SCoder *pCoder, SVDropTbBatchReq *pReq) {
+  if (tStartDecode(pCoder) < 0) return -1;
+
+  if (tDecodeI32v(pCoder, &pReq->nReqs) < 0) return -1;
+  pReq->pReqs = (SVDropTbReq *)tCoderMalloc(pCoder, sizeof(SVDropTbReq) * pReq->nReqs);
+  if (pReq->pReqs == NULL) return -1;
+  for (int iReq = 0; iReq < pReq->nReqs; iReq++) {
+    if (tDecodeSVDropTbReq(pCoder, pReq->pReqs + iReq) < 0) return -1;
+  }
+
+  tEndDecode(pCoder);
+  return 0;
+}
+
+int32_t tEncodeSVDropTbBatchRsp(SCoder *pCoder, const SVDropTbBatchRsp *pRsp) {
+  int32_t nRsps = taosArrayGetSize(pRsp->pArray);
+  if (tStartEncode(pCoder) < 0) return -1;
+
+  if (tEncodeI32v(pCoder, nRsps) < 0) return -1;
+  for (int iRsp = 0; iRsp < nRsps; iRsp++) {
+    if (tEncodeSVDropTbRsp(pCoder, (SVDropTbRsp *)taosArrayGet(pRsp->pArray, iRsp)) < 0) return -1;
+  }
+
+  tEndEncode(pCoder);
+  return 0;
+}
+
+int32_t tDecodeSVDropTbBatchRsp(SCoder *pCoder, SVDropTbBatchRsp *pRsp) {
+  if (tStartDecode(pCoder) < 0) return -1;
+
+  if (tDecodeI32v(pCoder, &pRsp->nRsps) < 0) return -1;
+  pRsp->pRsps = (SVDropTbRsp *)tCoderMalloc(pCoder, sizeof(SVDropTbRsp) * pRsp->nRsps);
+  if (pRsp->pRsps == NULL) return -1;
+  for (int iRsp = 0; iRsp < pRsp->nRsps; iRsp++) {
+    if (tDecodeSVDropTbRsp(pCoder, pRsp->pRsps + iRsp) < 0) return -1;
+  }
+
+  tEndDecode(pCoder);
+  return 0;
+}
+
+int32_t tEncodeSVDropStbReq(SCoder *pCoder, const SVDropStbReq *pReq) {
+  if (tStartEncode(pCoder) < 0) return -1;
+
+  if (tEncodeCStr(pCoder, pReq->name) < 0) return -1;
+  if (tEncodeI64(pCoder, pReq->suid) < 0) return -1;
+
+  tEndEncode(pCoder);
+  return 0;
+}
+
+int32_t tDecodeSVDropStbReq(SCoder *pCoder, SVDropStbReq *pReq) {
+  if (tStartDecode(pCoder) < 0) return -1;
+
+  if (tDecodeCStr(pCoder, &pReq->name) < 0) return -1;
+  if (tDecodeI64(pCoder, &pReq->suid) < 0) return -1;
+
+  tEndDecode(pCoder);
+  return 0;
+}

source/common/src/tmsgcb.c

@@ -15,33 +15,83 @@
 
 #define _DEFAULT_SOURCE
 #include "tmsgcb.h"
+#include "taoserror.h"
 
 static SMsgCb tsDefaultMsgCb;
 
 void tmsgSetDefaultMsgCb(const SMsgCb* pMsgCb) { tsDefaultMsgCb = *pMsgCb; }
 
 int32_t tmsgPutToQueue(const SMsgCb* pMsgCb, EQueueType qtype, SRpcMsg* pReq) {
-  return (*pMsgCb->queueFps[qtype])(pMsgCb->pWrapper, pReq);
+  PutToQueueFp fp = pMsgCb->queueFps[qtype];
+  if (fp != NULL) {
+    return (*fp)(pMsgCb->pWrapper, pReq);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
 }
 
 int32_t tmsgGetQueueSize(const SMsgCb* pMsgCb, int32_t vgId, EQueueType qtype) {
-  return (*pMsgCb->qsizeFp)(pMsgCb->pWrapper, vgId, qtype);
+  GetQueueSizeFp fp = pMsgCb->qsizeFp;
+  if (fp != NULL) {
+    return (*fp)(pMsgCb->pWrapper, vgId, qtype);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
 }
 
 int32_t tmsgSendReq(const SMsgCb* pMsgCb, const SEpSet* epSet, SRpcMsg* pReq) {
-  return (*pMsgCb->sendReqFp)(pMsgCb->pWrapper, epSet, pReq);
+  SendReqFp fp = pMsgCb->sendReqFp;
+  if (fp != NULL) {
+    return (*fp)(pMsgCb->pWrapper, epSet, pReq);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
 }
 
-void tmsgSendRsp(const SRpcMsg* pRsp) { return (*tsDefaultMsgCb.sendRspFp)(tsDefaultMsgCb.pWrapper, pRsp); }
+void tmsgSendRsp(const SRpcMsg* pRsp) {
+  SendRspFp fp = tsDefaultMsgCb.sendRspFp;
+  if (fp != NULL) {
+    return (*fp)(tsDefaultMsgCb.pWrapper, pRsp);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+  }
+}
+
+void tmsgSendRedirectRsp(const SRpcMsg* pRsp, const SEpSet* pNewEpSet) {
+  SendRedirectRspFp fp = tsDefaultMsgCb.sendRedirectRspFp;
+  if (fp != NULL) {
+    (*fp)(tsDefaultMsgCb.pWrapper, pRsp, pNewEpSet);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+  }
+}
 
 void tmsgRegisterBrokenLinkArg(const SMsgCb* pMsgCb, SRpcMsg* pMsg) {
-  (*pMsgCb->registerBrokenLinkArgFp)(pMsgCb->pWrapper, pMsg);
+  RegisterBrokenLinkArgFp fp = pMsgCb->registerBrokenLinkArgFp;
+  if (fp != NULL) {
+    (*fp)(pMsgCb->pWrapper, pMsg);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+  }
 }
 
 void tmsgReleaseHandle(void* handle, int8_t type) {
-  (*tsDefaultMsgCb.releaseHandleFp)(tsDefaultMsgCb.pWrapper, handle, type);
+  ReleaseHandleFp fp = tsDefaultMsgCb.releaseHandleFp;
+  if (fp != NULL) {
+    (*fp)(tsDefaultMsgCb.pWrapper, handle, type);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+  }
 }
 
 void tmsgReportStartup(const char* name, const char* desc) {
-  (*tsDefaultMsgCb.reportStartupFp)(tsDefaultMsgCb.pWrapper, name, desc);
+  ReportStartup fp = tsDefaultMsgCb.reportStartupFp;
+  if (fp != NULL && tsDefaultMsgCb.pWrapper != NULL) {
+    (*fp)(tsDefaultMsgCb.pWrapper, name, desc);
+  } else {
+    terrno = TSDB_CODE_INVALID_PTR;
+  }
 }
\ No newline at end of file

source/dnode/mgmt/implement/src/dmTransport.c

@@ -16,8 +16,8 @@
 #define _DEFAULT_SOURCE
 #include "dmImp.h"
 
-#define INTERNAL_USER   "_dnd"
-#define INTERNAL_CKEY   "_key"
+#define INTERNAL_USER "_dnd"
+#define INTERNAL_CKEY "_key"
 #define INTERNAL_SECRET "_pwd"
 
 static void dmGetMnodeEpSet(SDnode *pDnode, SEpSet *pEpSet) {
@@ -72,6 +72,7 @@ static void dmProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSe
   NodeMsgFp msgFp = NULL;
   uint16_t  msgType = pRpc->msgType;
   bool      needRelease = false;
+  bool      isReq = msgType & 1U;
 
   if (pEpSet && pEpSet->numOfEps > 0 && msgType == TDMT_MND_STATUS_RSP) {
     dmSetMnodeEpSet(pWrapper->pDnode, pEpSet);
@@ -85,13 +86,13 @@ static void dmProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSe
   if (dmBuildMsg(pMsg, pRpc) != 0) goto _OVER;
 
   if (pWrapper->procType == DND_PROC_SINGLE) {
-    dTrace("msg:%p, is created, type:%s handle:%p user:%s", pMsg, TMSG_INFO(msgType), pRpc->handle, pMsg->user);
+    dTrace("msg:%p, created, type:%s handle:%p user:%s", pMsg, TMSG_INFO(msgType), pRpc->handle, pMsg->user);
     code = (*msgFp)(pWrapper, pMsg);
   } else if (pWrapper->procType == DND_PROC_PARENT) {
-    dTrace("msg:%p, is created and put into child queue, type:%s handle:%p user:%s", pMsg, TMSG_INFO(msgType),
-           pRpc->handle, pMsg->user);
-    code = taosProcPutToChildQ(pWrapper->procObj, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen, pRpc->handle,
-                               pRpc->refId, PROC_FUNC_REQ);
+    dTrace("msg:%p, created and put into child queue, type:%s handle:%p code:0x%04x user:%s contLen:%d", pMsg,
+           TMSG_INFO(msgType), pRpc->handle, pMsg->rpcMsg.code & 0XFFFF, pMsg->user, pRpc->contLen);
+    code = taosProcPutToChildQ(pWrapper->procObj, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen,
+                               (isReq && (pMsg->rpcMsg.code == 0)) ? pRpc->handle : NULL, pRpc->refId, PROC_FUNC_REQ);
   } else {
     dTrace("msg:%p, should not processed in child process, handle:%p user:%s", pMsg, pRpc->handle, pMsg->user);
     ASSERT(1);
@@ -100,12 +101,13 @@ static void dmProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSe
 _OVER:
   if (code == 0) {
     if (pWrapper->procType == DND_PROC_PARENT) {
-      dTrace("msg:%p, is freed in parent process", pMsg);
+      dTrace("msg:%p, freed in parent process", pMsg);
       taosFreeQitem(pMsg);
       rpcFreeCont(pRpc->pCont);
     }
   } else {
-    dError("msg:%p, type:%s failed to process since 0x%04x:%s", pMsg, TMSG_INFO(msgType), code & 0XFFFF, terrstr());
+    dError("msg:%p, type:%s handle:%p failed to process since 0x%04x:%s", pMsg, TMSG_INFO(msgType), pRpc->handle,
+           code & 0XFFFF, terrstr());
     if (msgType & 1U) {
       if (terrno != 0) code = terrno;
       if (code == TSDB_CODE_NODE_NOT_DEPLOYED || code == TSDB_CODE_NODE_OFFLINE) {
@@ -128,10 +130,10 @@ _OVER:
 }
 
 static void dmProcessMsg(SDnode *pDnode, SRpcMsg *pMsg, SEpSet *pEpSet) {
-  SDnodeTrans * pTrans = &pDnode->trans;
+  SDnodeTrans  *pTrans = &pDnode->trans;
   tmsg_t        msgType = pMsg->msgType;
   bool          isReq = msgType & 1u;
-  SMsgHandle *  pHandle = &pTrans->msgHandles[TMSG_INDEX(msgType)];
+  SMsgHandle   *pHandle = &pTrans->msgHandles[TMSG_INDEX(msgType)];
   SMgmtWrapper *pWrapper = pHandle->pNdWrapper;
 
   if (msgType == TDMT_DND_SERVER_STATUS) {
@@ -254,8 +256,17 @@ static void dmSendRpcRedirectRsp(SDnode *pDnode, const SRpcMsg *pReq) {
 
     epSet.eps[i].port = htons(epSet.eps[i].port);
   }
-
-  rpcSendRedirectRsp(pReq->handle, &epSet);
+  SRpcMsg resp;
+  SMEpSet msg = {.epSet = epSet};
+  int32_t len = tSerializeSMEpSet(NULL, 0, &msg);
+  resp.pCont = rpcMallocCont(len);
+  resp.contLen = len;
+  tSerializeSMEpSet(resp.pCont, len, &msg);
+
+  resp.code = TSDB_CODE_RPC_REDIRECT;
+  resp.handle = pReq->handle;
+  resp.refId = pReq->refId;
+  rpcSendResponse(&resp);
 }
 
 static inline void dmSendRpcRsp(SDnode *pDnode, const SRpcMsg *pRsp) {
@@ -309,6 +320,37 @@ static inline void dmSendRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
   }
 }
 
+static inline void dmSendRedirectRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp, const SEpSet *pNewEpSet) {
+  ASSERT(pRsp->code == TSDB_CODE_RPC_REDIRECT);
+  ASSERT(pRsp->pCont == NULL);
+  if (pWrapper->procType != DND_PROC_CHILD) {
+    SRpcMsg resp = {0};
+    SMEpSet msg = {.epSet = *pNewEpSet};
+    int32_t len = tSerializeSMEpSet(NULL, 0, &msg);
+    resp.pCont = rpcMallocCont(len);
+    resp.contLen = len;
+    tSerializeSMEpSet(resp.pCont, len, &msg);
+
+    resp.code = TSDB_CODE_RPC_REDIRECT;
+    resp.handle = pRsp->handle;
+    resp.refId = pRsp->refId;
+    rpcSendResponse(&resp);
+  } else {
+    taosProcPutToParentQ(pWrapper->procObj, pRsp, sizeof(SRpcMsg), pRsp->pCont, pRsp->contLen, PROC_FUNC_RSP);
+  }
+}
+
+#if 0
+static inline void dmSendRedirectRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp, const SEpSet *pNewEpSet) {
+  ASSERT(pRsp->code == TSDB_CODE_RPC_REDIRECT);
+  if (pWrapper->procType != DND_PROC_CHILD) {
+    rpcSendRedirectRsp(pRsp->handle, pNewEpSet);
+  } else {
+    taosProcPutToParentQ(pWrapper->procObj, pRsp, sizeof(SRpcMsg), pRsp->pCont, pRsp->contLen, PROC_FUNC_RSP);
+  }
+}
+#endif
+
 static inline void dmRegisterBrokenLinkArg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg) {
   if (pWrapper->procType != DND_PROC_CHILD) {
     rpcRegisterBrokenLinkArg(pMsg);
@@ -350,29 +392,31 @@ static void dmConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int16_t
 
 static void dmConsumeParentQueue(SMgmtWrapper *pWrapper, SRpcMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
                                  EProcFuncType ftype) {
+  int32_t code = pMsg->code & 0xFFFF;
   pMsg->pCont = pCont;
-  dTrace("msg:%p, get from parent queue, ftype:%d handle:%p code:0x%04x mtype:%d, app:%p", pMsg, ftype, pMsg->handle,
-         pMsg->code & 0xFFFF, pMsg->msgType, pMsg->ahandle);
-
-  switch (ftype) {
-    case PROC_FUNC_REGIST:
-      rpcRegisterBrokenLinkArg(pMsg);
-      break;
-    case PROC_FUNC_RELEASE:
-      taosProcRemoveHandle(pWrapper->procObj, pMsg->handle);
-      rpcReleaseHandle(pMsg->handle, (int8_t)pMsg->code);
-      rpcFreeCont(pCont);
-      break;
-    case PROC_FUNC_REQ:
-      dmSendRpcReq(pWrapper->pDnode, (SEpSet *)((char *)pMsg + sizeof(SRpcMsg)), pMsg);
-      break;
-    case PROC_FUNC_RSP:
-      pMsg->refId = taosProcRemoveHandle(pWrapper->procObj, pMsg->handle);
-      dmSendRpcRsp(pWrapper->pDnode, pMsg);
-      break;
-    default:
-      break;
+
+  if (ftype == PROC_FUNC_REQ) {
+    dTrace("msg:%p, get from parent queue, send req:%s handle:%p code:0x%04x, app:%p", pMsg, TMSG_INFO(pMsg->msgType),
+           pMsg->handle, code, pMsg->ahandle);
+    dmSendRpcReq(pWrapper->pDnode, (SEpSet *)((char *)pMsg + sizeof(SRpcMsg)), pMsg);
+  } else if (ftype == PROC_FUNC_RSP) {
+    dTrace("msg:%p, get from parent queue, rsp handle:%p code:0x%04x, app:%p", pMsg, pMsg->handle, code, pMsg->ahandle);
+    pMsg->refId = taosProcRemoveHandle(pWrapper->procObj, pMsg->handle);
+    dmSendRpcRsp(pWrapper->pDnode, pMsg);
+  } else if (ftype == PROC_FUNC_REGIST) {
+    dTrace("msg:%p, get from parent queue, regist handle:%p code:0x%04x, app:%p", pMsg, pMsg->handle, code,
+           pMsg->ahandle);
+    rpcRegisterBrokenLinkArg(pMsg);
+  } else if (ftype == PROC_FUNC_RELEASE) {
+    dTrace("msg:%p, get from parent queue, release handle:%p code:0x%04x, app:%p", pMsg, pMsg->handle, code,
+           pMsg->ahandle);
+    taosProcRemoveHandle(pWrapper->procObj, pMsg->handle);
+    rpcReleaseHandle(pMsg->handle, (int8_t)pMsg->code);
+    rpcFreeCont(pCont);
+  } else {
+    dError("msg:%p, invalid ftype:%d while get from parent queue, handle:%p", pMsg, ftype, pMsg->handle);
   }
+
   taosMemoryFree(pMsg);
 }
 
@@ -393,6 +437,14 @@ SProcCfg dmGenProcCfg(SMgmtWrapper *pWrapper) {
   return cfg;
 }
 
+bool rpcRfp(int32_t code) {
+  if (code == TSDB_CODE_RPC_REDIRECT) {
+    return true;
+  } else {
+    return false;
+  }
+}
+
 static int32_t dmInitClient(SDnode *pDnode) {
   SDnodeTrans *pTrans = &pDnode->trans;
 
@@ -407,6 +459,7 @@ static int32_t dmInitClient(SDnode *pDnode) {
   rpcInit.ckey = INTERNAL_CKEY;
   rpcInit.spi = 1;
   rpcInit.parent = pDnode;
+  rpcInit.rfp = rpcRfp;
 
   char pass[TSDB_PASSWORD_LEN + 1] = {0};
   taosEncryptPass_c((uint8_t *)(INTERNAL_SECRET), strlen(INTERNAL_SECRET), pass);
@@ -464,7 +517,7 @@ static inline int32_t dmRetrieveUserAuthInfo(SDnode *pDnode, char *user, char *s
   SAuthReq authReq = {0};
   tstrncpy(authReq.user, user, TSDB_USER_LEN);
   int32_t contLen = tSerializeSAuthReq(NULL, 0, &authReq);
-  void *  pReq = rpcMallocCont(contLen);
+  void   *pReq = rpcMallocCont(contLen);
   tSerializeSAuthReq(pReq, contLen, &authReq);
 
   SRpcMsg rpcMsg = {.pCont = pReq, .contLen = contLen, .msgType = TDMT_MND_AUTH, .ahandle = (void *)9528};
@@ -538,6 +591,7 @@ SMsgCb dmGetMsgcb(SMgmtWrapper *pWrapper) {
   SMsgCb msgCb = {
       .sendReqFp = dmSendReq,
       .sendRspFp = dmSendRsp,
+      .sendRedirectRspFp = dmSendRedirectRsp,
       .registerBrokenLinkArgFp = dmRegisterBrokenLinkArg,
       .releaseHandleFp = dmReleaseHandle,
       .reportStartupFp = dmReportStartupByWrapper,

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

@@ -112,7 +112,9 @@ static void vmGenerateVnodeCfg(SCreateVnodeReq *pCreate, SVnodeCfg *pCfg) {
   pCfg->tsdbCfg.keep2 = 3650;
   pCfg->tsdbCfg.keep0 = 3650;
   pCfg->tsdbCfg.keep1 = 3650;
-  pCfg->tsdbCfg.retentions = pCreate->pRetensions;
+  for (size_t i = 0; i < taosArrayGetSize(pCreate->pRetensions); ++i) {
+    memcpy(&pCfg->tsdbCfg.retentions[i], taosArrayGet(pCreate->pRetensions, i), sizeof(SRetention));
+  }
   pCfg->walCfg.vgId = pCreate->vgId;
   pCfg->hashBegin = pCreate->hashBegin;
   pCfg->hashEnd = pCreate->hashEnd;
@@ -282,6 +284,7 @@ void vmInitMsgHandle(SMgmtWrapper *pWrapper) {
   dmSetMsgHandle(pWrapper, TDMT_VND_CREATE_SMA, vmProcessWriteMsg, DEFAULT_HANDLE);
   dmSetMsgHandle(pWrapper, TDMT_VND_CANCEL_SMA, vmProcessWriteMsg, DEFAULT_HANDLE);
   dmSetMsgHandle(pWrapper, TDMT_VND_DROP_SMA, vmProcessWriteMsg, DEFAULT_HANDLE);
+  dmSetMsgHandle(pWrapper, TDMT_VND_SUBMIT_RSMA, vmProcessWriteMsg, DEFAULT_HANDLE);
   dmSetMsgHandle(pWrapper, TDMT_VND_MQ_VG_CHANGE, (NodeMsgFp)vmProcessWriteMsg, DEFAULT_HANDLE);
   dmSetMsgHandle(pWrapper, TDMT_VND_CONSUME, vmProcessFetchMsg, DEFAULT_HANDLE);
   dmSetMsgHandle(pWrapper, TDMT_VND_TASK_DEPLOY, vmProcessWriteMsg, DEFAULT_HANDLE);

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

@@ -136,7 +136,7 @@ static void vmProcessWriteQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
   // sync integration response
   for (int i = 0; i < taosArrayGetSize(pArray); i++) {
     SNodeMsg *pMsg;
-    SRpcMsg * pRpc;
+    SRpcMsg  *pRpc;
 
     pMsg = *(SNodeMsg **)taosArrayGet(pArray, i);
     pRpc = &pMsg->rpcMsg;
@@ -149,8 +149,15 @@ static void vmProcessWriteQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
 
     int32_t ret = syncPropose(vnodeGetSyncHandle(pVnode->pImpl), pRpc, false);
     if (ret == TAOS_SYNC_PROPOSE_NOT_LEADER) {
-      rsp.code = TSDB_CODE_SYN_NOT_LEADER;
-      tmsgSendRsp(&rsp);
+      // rsp.code = TSDB_CODE_SYN_NOT_LEADER;
+      // tmsgSendRsp(&rsp);
+      dTrace("syncPropose not leader redirect, vgId:%d ", syncGetVgId(vnodeGetSyncHandle(pVnode->pImpl)));
+      rsp.code = TSDB_CODE_RPC_REDIRECT;
+      SEpSet newEpSet;
+      syncGetEpSet(vnodeGetSyncHandle(pVnode->pImpl), &newEpSet);
+      newEpSet.inUse = (newEpSet.inUse + 1) % newEpSet.numOfEps;
+      tmsgSendRedirectRsp(&rsp, &newEpSet);
+
     } else if (ret == TAOS_SYNC_PROPOSE_OTHER_ERROR) {
       rsp.code = TSDB_CODE_SYN_INTERNAL_ERROR;
       tmsgSendRsp(&rsp);
@@ -175,7 +182,7 @@ static void vmProcessWriteQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
 
 static void vmProcessApplyQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
   SVnodeObj *pVnode = pInfo->ahandle;
-  SNodeMsg * pMsg = NULL;
+  SNodeMsg  *pMsg = NULL;
   SRpcMsg    rsp;
 
   for (int32_t i = 0; i < numOfMsgs; ++i) {
@@ -218,7 +225,7 @@ static void vmProcessApplyQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
 
 static void vmProcessSyncQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
   SVnodeObj *pVnode = pInfo->ahandle;
-  SNodeMsg * pMsg = NULL;
+  SNodeMsg  *pMsg = NULL;
 
   for (int32_t i = 0; i < numOfMsgs; ++i) {
     taosGetQitem(qall, (void **)&pMsg);
@@ -231,7 +238,7 @@ static void vmProcessSyncQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOf
 
 static void vmProcessMergeQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
   SVnodeObj *pVnode = pInfo->ahandle;
-  SNodeMsg * pMsg = NULL;
+  SNodeMsg  *pMsg = NULL;
 
   for (int32_t i = 0; i < numOfMsgs; ++i) {
     taosGetQitem(qall, (void **)&pMsg);
@@ -248,7 +255,7 @@ static void vmProcessMergeQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
 }
 
 static int32_t vmPutNodeMsgToQueue(SVnodesMgmt *pMgmt, SNodeMsg *pMsg, EQueueType qtype) {
-  SRpcMsg * pRpc = &pMsg->rpcMsg;
+  SRpcMsg  *pRpc = &pMsg->rpcMsg;
   SMsgHead *pHead = pRpc->pCont;
   pHead->contLen = ntohl(pHead->contLen);
   pHead->vgId = ntohl(pHead->vgId);
@@ -262,23 +269,23 @@ static int32_t vmPutNodeMsgToQueue(SVnodesMgmt *pMgmt, SNodeMsg *pMsg, EQueueTyp
   int32_t code = 0;
   switch (qtype) {
     case QUERY_QUEUE:
-      dTrace("msg:%p, will be written into vnode-query queue", pMsg);
+      dTrace("msg:%p, type:%s will be written into vnode-query queue", pMsg, TMSG_INFO(pRpc->msgType));
       taosWriteQitem(pVnode->pQueryQ, pMsg);
       break;
     case FETCH_QUEUE:
-      dTrace("msg:%p, will be written into vnode-fetch queue", pMsg);
+      dTrace("msg:%p, type:%s will be written into vnode-fetch queue", pMsg, TMSG_INFO(pRpc->msgType));
       taosWriteQitem(pVnode->pFetchQ, pMsg);
       break;
     case WRITE_QUEUE:
-      dTrace("msg:%p, will be written into vnode-write queue", pMsg);
+      dTrace("msg:%p, type:%s will be written into vnode-write queue", pMsg, TMSG_INFO(pRpc->msgType));
       taosWriteQitem(pVnode->pWriteQ, pMsg);
       break;
     case SYNC_QUEUE:
-      dTrace("msg:%p, will be written into vnode-sync queue", pMsg);
+      dTrace("msg:%p, type:%s will be written into vnode-sync queue", pMsg, TMSG_INFO(pRpc->msgType));
       taosWriteQitem(pVnode->pSyncQ, pMsg);
       break;
     case MERGE_QUEUE:
-      dTrace("msg:%p, will be written into vnode-merge queue", pMsg);
+      dTrace("msg:%p, type:%s will be written into vnode-merge queue", pMsg, TMSG_INFO(pRpc->msgType));
       taosWriteQitem(pVnode->pMergeQ, pMsg);
       break;
     default:
@@ -317,7 +324,7 @@ int32_t vmProcessMergeMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
 }
 
 int32_t vmProcessMgmtMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
-  SVnodesMgmt *  pMgmt = pWrapper->pMgmt;
+  SVnodesMgmt   *pMgmt = pWrapper->pMgmt;
   SSingleWorker *pWorker = &pMgmt->mgmtWorker;
   dTrace("msg:%p, will be written to vnode-mgmt queue, worker:%s", pMsg, pWorker->name);
   taosWriteQitem(pWorker->queue, pMsg);
@@ -325,7 +332,7 @@ int32_t vmProcessMgmtMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
 }
 
 int32_t vmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
-  SVnodesMgmt *  pMgmt = pWrapper->pMgmt;
+  SVnodesMgmt   *pMgmt = pWrapper->pMgmt;
   SSingleWorker *pWorker = &pMgmt->monitorWorker;
 
   dTrace("msg:%p, put into worker:%s", pMsg, pWorker->name);
@@ -335,7 +342,7 @@ int32_t vmProcessMonitorMsg(SMgmtWrapper *pWrapper, SNodeMsg *pMsg) {
 
 static int32_t vmPutRpcMsgToQueue(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, EQueueType qtype) {
   SVnodesMgmt *pMgmt = pWrapper->pMgmt;
-  SMsgHead *   pHead = pRpc->pCont;
+  SMsgHead    *pHead = pRpc->pCont;
 
   SVnodeObj *pVnode = vmAcquireVnode(pMgmt, pHead->vgId);
   if (pVnode == NULL) return -1;

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

@@ -57,11 +57,11 @@ typedef enum {
   TRN_STAGE_PREPARE = 0,
   TRN_STAGE_REDO_LOG = 1,
   TRN_STAGE_REDO_ACTION = 2,
-  TRN_STAGE_COMMIT = 3,
-  TRN_STAGE_COMMIT_LOG = 4,
-  TRN_STAGE_UNDO_ACTION = 5,
-  TRN_STAGE_UNDO_LOG = 6,
-  TRN_STAGE_ROLLBACK = 7,
+  TRN_STAGE_ROLLBACK = 3,
+  TRN_STAGE_UNDO_ACTION = 4,
+  TRN_STAGE_UNDO_LOG = 5,
+  TRN_STAGE_COMMIT = 6,
+  TRN_STAGE_COMMIT_LOG = 7,
   TRN_STAGE_FINISHED = 8
 } ETrnStage;
 
@@ -72,6 +72,7 @@ typedef enum {
   TRN_TYPE_DROP_USER = 1003,
   TRN_TYPE_CREATE_FUNC = 1004,
   TRN_TYPE_DROP_FUNC = 1005,
+  
   TRN_TYPE_CREATE_SNODE = 1006,
   TRN_TYPE_DROP_SNODE = 1007,
   TRN_TYPE_CREATE_QNODE = 1008,
@@ -91,10 +92,12 @@ typedef enum {
   TRN_TYPE_CONSUMER_LOST = 1022,
   TRN_TYPE_CONSUMER_RECOVER = 1023,
   TRN_TYPE_BASIC_SCOPE_END,
+
   TRN_TYPE_GLOBAL_SCOPE = 2000,
   TRN_TYPE_CREATE_DNODE = 2001,
   TRN_TYPE_DROP_DNODE = 2002,
   TRN_TYPE_GLOBAL_SCOPE_END,
+
   TRN_TYPE_DB_SCOPE = 3000,
   TRN_TYPE_CREATE_DB = 3001,
   TRN_TYPE_ALTER_DB = 3002,
@@ -102,6 +105,7 @@ typedef enum {
   TRN_TYPE_SPLIT_VGROUP = 3004,
   TRN_TYPE_MERGE_VGROUP = 3015,
   TRN_TYPE_DB_SCOPE_END,
+
   TRN_TYPE_STB_SCOPE = 4000,
   TRN_TYPE_CREATE_STB = 4001,
   TRN_TYPE_ALTER_STB = 4002,
@@ -131,7 +135,7 @@ typedef struct {
   int32_t    id;
   ETrnStage  stage;
   ETrnPolicy policy;
-  ETrnType   transType;
+  ETrnType   type;
   int32_t    code;
   int32_t    failedTimes;
   void*      rpcHandle;
@@ -388,15 +392,12 @@ typedef struct {
   int8_t         type;
   int8_t         replica;
   int16_t        numOfColumns;
-  int32_t        rowSize;
   int32_t        numOfRows;
   void*          pIter;
   SMnode*        pMnode;
   STableMetaRsp* pMeta;
   bool           sysDbRsp;
   char           db[TSDB_DB_FNAME_LEN];
-  int16_t        offset[TSDB_MAX_COLUMNS];
-  int32_t        bytes[TSDB_MAX_COLUMNS];
 } SShowObj;
 
 typedef struct {
@@ -520,6 +521,7 @@ void*          tDecodeSMqConsumerEp(const void* buf, SMqConsumerEp* pEp);
 typedef struct {
   char      key[TSDB_SUBSCRIBE_KEY_LEN];
   SRWLatch  lock;
+  int64_t   dbUid;
   int32_t   vgNum;
   int8_t    subType;
   int8_t    withTbName;
@@ -556,9 +558,8 @@ int32_t             tEncodeSMqSubActionLogObj(void** buf, const SMqSubActionLogO
 void*               tDecodeSMqSubActionLogObj(const void* buf, SMqSubActionLogObj* pLog);
 
 typedef struct {
-  const SMqSubscribeObj* pOldSub;
-  const SMqTopicObj*     pTopic;
-  const SMqRebSubscribe* pRebInfo;
+  int32_t           oldConsumerNum;
+  const SMqRebInfo* pRebInfo;
 } SMqRebInputObj;
 
 typedef struct {

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

@@ -31,6 +31,8 @@ void         mndReleaseTopic(SMnode *pMnode, SMqTopicObj *pTopic);
 SSdbRaw *mndTopicActionEncode(SMqTopicObj *pTopic);
 SSdbRow *mndTopicActionDecode(SSdbRaw *pRaw);
 
+int32_t mndDropTopicByDB(SMnode *pMnode, STrans *pTrans, SDbObj *pDb);
+
 #ifdef __cplusplus
 }
 #endif

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

@@ -24,9 +24,12 @@ extern "C" {
 
 int32_t   mndInitUser(SMnode *pMnode);
 void      mndCleanupUser(SMnode *pMnode);
-SUserObj *mndAcquireUser(SMnode *pMnode, char *userName);
+SUserObj *mndAcquireUser(SMnode *pMnode, const char *userName);
 void      mndReleaseUser(SMnode *pMnode, SUserObj *pUser);
 
+// for trans test
+SSdbRaw *mndUserActionEncode(SUserObj *pUser);
+
 #ifdef __cplusplus
 }
 #endif

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

@@ -453,7 +453,7 @@ static int32_t mndRetrieveBnodes(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *p
     colDataAppend(pColInfo, numOfRows, (const char *)&pObj->id, false);
 
     char buf[TSDB_EP_LEN + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(buf, pObj->pDnode->ep, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(buf, pObj->pDnode->ep, pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, buf, false);

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

@@ -194,7 +194,7 @@ static int32_t mndRetrieveClusters(SNodeMsg *pMsg, SShowObj *pShow, SSDataBlock
     colDataAppend(pColInfo, numOfRows, (const char*) &pCluster->id, false);
 
     char buf[tListLen(pCluster->name) + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(buf, pCluster->name, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(buf, pCluster->name, pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, buf, false);

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

@@ -134,15 +134,15 @@ FAIL:
   return -1;
 }
 
-static SMqRebSubscribe *mndGetOrCreateRebSub(SHashObj *pHash, const char *key) {
-  SMqRebSubscribe *pRebSub = taosHashGet(pHash, key, strlen(key) + 1);
+static SMqRebInfo *mndGetOrCreateRebSub(SHashObj *pHash, const char *key) {
+  SMqRebInfo *pRebSub = taosHashGet(pHash, key, strlen(key) + 1);
   if (pRebSub == NULL) {
     pRebSub = tNewSMqRebSubscribe(key);
     if (pRebSub == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return NULL;
     }
-    taosHashPut(pHash, key, strlen(key) + 1, pRebSub, sizeof(SMqRebSubscribe));
+    taosHashPut(pHash, key, strlen(key) + 1, pRebSub, sizeof(SMqRebInfo));
   }
   return pRebSub;
 }
@@ -189,7 +189,7 @@ static int32_t mndProcessMqTimerMsg(SNodeMsg *pMsg) {
         char  key[TSDB_SUBSCRIBE_KEY_LEN];
         char *removedTopic = taosArrayGetP(pConsumer->currentTopics, i);
         mndMakeSubscribeKey(key, pConsumer->cgroup, removedTopic);
-        SMqRebSubscribe *pRebSub = mndGetOrCreateRebSub(pRebMsg->rebSubHash, key);
+        SMqRebInfo *pRebSub = mndGetOrCreateRebSub(pRebMsg->rebSubHash, key);
         taosArrayPush(pRebSub->removedConsumers, &pConsumer->consumerId);
       }
       taosRUnLockLatch(&pConsumer->lock);
@@ -200,7 +200,7 @@ static int32_t mndProcessMqTimerMsg(SNodeMsg *pMsg) {
         char  key[TSDB_SUBSCRIBE_KEY_LEN];
         char *newTopic = taosArrayGetP(pConsumer->rebNewTopics, i);
         mndMakeSubscribeKey(key, pConsumer->cgroup, newTopic);
-        SMqRebSubscribe *pRebSub = mndGetOrCreateRebSub(pRebMsg->rebSubHash, key);
+        SMqRebInfo *pRebSub = mndGetOrCreateRebSub(pRebMsg->rebSubHash, key);
         taosArrayPush(pRebSub->newConsumers, &pConsumer->consumerId);
       }
 
@@ -209,7 +209,7 @@ static int32_t mndProcessMqTimerMsg(SNodeMsg *pMsg) {
         char  key[TSDB_SUBSCRIBE_KEY_LEN];
         char *removedTopic = taosArrayGetP(pConsumer->rebRemovedTopics, i);
         mndMakeSubscribeKey(key, pConsumer->cgroup, removedTopic);
-        SMqRebSubscribe *pRebSub = mndGetOrCreateRebSub(pRebMsg->rebSubHash, key);
+        SMqRebInfo *pRebSub = mndGetOrCreateRebSub(pRebMsg->rebSubHash, key);
         taosArrayPush(pRebSub->removedConsumers, &pConsumer->consumerId);
       }
       taosRUnLockLatch(&pConsumer->lock);

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

@@ -107,8 +107,8 @@ static SSdbRaw *mndDbActionEncode(SDbObj *pDb) {
   for (int32_t i = 0; i < pDb->cfg.numOfRetensions; ++i) {
     TASSERT(taosArrayGetSize(pDb->cfg.pRetensions) == pDb->cfg.numOfRetensions);
     SRetention *pRetension = taosArrayGet(pDb->cfg.pRetensions, i);
-    SDB_SET_INT32(pRaw, dataPos, pRetension->freq, _OVER)
-    SDB_SET_INT32(pRaw, dataPos, pRetension->keep, _OVER)
+    SDB_SET_INT64(pRaw, dataPos, pRetension->freq, _OVER)
+    SDB_SET_INT64(pRaw, dataPos, pRetension->keep, _OVER)
     SDB_SET_INT8(pRaw, dataPos, pRetension->freqUnit, _OVER)
     SDB_SET_INT8(pRaw, dataPos, pRetension->keepUnit, _OVER)
   }
@@ -180,8 +180,8 @@ static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
     if (pDb->cfg.pRetensions == NULL) goto _OVER;
     for (int32_t i = 0; i < pDb->cfg.numOfRetensions; ++i) {
       SRetention retension = {0};
-      SDB_GET_INT32(pRaw, dataPos, &retension.freq, _OVER)
-      SDB_GET_INT32(pRaw, dataPos, &retension.keep, _OVER)
+      SDB_GET_INT64(pRaw, dataPos, &retension.freq, _OVER)
+      SDB_GET_INT64(pRaw, dataPos, &retension.keep, _OVER)
       SDB_GET_INT8(pRaw, dataPos, &retension.freqUnit, _OVER)
       SDB_GET_INT8(pRaw, dataPos, &retension.keepUnit, _OVER)
       if (taosArrayPush(pDb->cfg.pRetensions, &retension) == NULL) {
@@ -1387,12 +1387,13 @@ static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, in
                            bool sysDb) {
   int32_t cols = 0;
 
-  char       *buf = taosMemoryMalloc(pShow->bytes[cols]);
+  int32_t bytes = pShow->pMeta->pSchemas[cols].bytes;
+  char       *buf = taosMemoryMalloc(bytes);
   const char *name = mndGetDbStr(pDb->name);
   if (name != NULL) {
-    STR_WITH_MAXSIZE_TO_VARSTR(buf, name, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(buf, name, bytes);
   } else {
-    STR_WITH_MAXSIZE_TO_VARSTR(buf, "NULL", pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(buf, "NULL", bytes);
   }
 
   char *status = "ready";

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

@@ -241,7 +241,7 @@ int32_t tEncodeSMqConsumerEp(void **buf, const SMqConsumerEp *pConsumerEp) {
 
 void *tDecodeSMqConsumerEp(const void *buf, SMqConsumerEp *pConsumerEp) {
   buf = taosDecodeFixedI64(buf, &pConsumerEp->consumerId);
-  buf = taosDecodeArray(buf, &pConsumerEp->vgs, (FDecode)tDecodeSMqVgEp, sizeof(SMqSubVgEp));
+  buf = taosDecodeArray(buf, &pConsumerEp->vgs, (FDecode)tDecodeSMqVgEp, sizeof(SMqVgEp));
 #if 0
   int32_t sz;
   buf = taosDecodeFixedI32(buf, &sz);
@@ -277,6 +277,7 @@ SMqSubscribeObj *tCloneSubscribeObj(const SMqSubscribeObj *pSub) {
   memcpy(pSubNew->key, pSub->key, TSDB_SUBSCRIBE_KEY_LEN);
   taosInitRWLatch(&pSubNew->lock);
 
+  pSubNew->dbUid = pSub->dbUid;
   pSubNew->subType = pSub->subType;
   pSubNew->withTbName = pSub->withTbName;
   pSubNew->withSchema = pSub->withSchema;
@@ -310,6 +311,7 @@ void tDeleteSubscribeObj(SMqSubscribeObj *pSub) {
 int32_t tEncodeSubscribeObj(void **buf, const SMqSubscribeObj *pSub) {
   int32_t tlen = 0;
   tlen += taosEncodeString(buf, pSub->key);
+  tlen += taosEncodeFixedI64(buf, pSub->dbUid);
   tlen += taosEncodeFixedI32(buf, pSub->vgNum);
   tlen += taosEncodeFixedI8(buf, pSub->subType);
   tlen += taosEncodeFixedI8(buf, pSub->withTbName);
@@ -336,6 +338,7 @@ int32_t tEncodeSubscribeObj(void **buf, const SMqSubscribeObj *pSub) {
 void *tDecodeSubscribeObj(const void *buf, SMqSubscribeObj *pSub) {
   //
   buf = taosDecodeStringTo(buf, pSub->key);
+  buf = taosDecodeFixedI64(buf, &pSub->dbUid);
   buf = taosDecodeFixedI32(buf, &pSub->vgNum);
   buf = taosDecodeFixedI8(buf, &pSub->subType);
   buf = taosDecodeFixedI8(buf, &pSub->withTbName);

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

@@ -705,7 +705,7 @@ static int32_t mndRetrieveDnodes(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *p
     colDataAppend(pColInfo, numOfRows, (const char *)&pDnode->id, false);
 
     char buf[tListLen(pDnode->ep) + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(buf, pDnode->ep, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(buf, pDnode->ep,   pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, buf, false);

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

@@ -517,14 +517,14 @@ static int32_t mndRetrieveFuncs(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pB
     cols = 0;
 
     char b1[tListLen(pFunc->name) + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(b1, pFunc->name, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(b1, pFunc->name, pShow->pMeta->pSchemas[cols].bytes);
 
     SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)b1, false);
 
     if (pFunc->pComment) {
-      char *b2 = taosMemoryCalloc(1, pShow->bytes[cols]);
-      STR_WITH_MAXSIZE_TO_VARSTR(b2, pFunc->pComment, pShow->bytes[cols]);
+      char *b2 = taosMemoryCalloc(1, pShow->pMeta->pSchemas[cols].bytes);
+      STR_WITH_MAXSIZE_TO_VARSTR(b2, pFunc->pComment, pShow->pMeta->pSchemas[cols].bytes);
 
       pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
       colDataAppend(pColInfo, numOfRows, (const char *)b2, false);
@@ -540,7 +540,7 @@ static int32_t mndRetrieveFuncs(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pB
 
     char b3[TSDB_TYPE_STR_MAX_LEN] = {0};
     STR_WITH_MAXSIZE_TO_VARSTR(b3, mnodeGenTypeStr(buf, TSDB_TYPE_STR_MAX_LEN, pFunc->outputType, pFunc->outputLen),
-                               pShow->bytes[cols]);
+                               pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)b3, false);

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

@@ -619,14 +619,14 @@ static int32_t mndRetrieveMnodes(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *p
     colDataAppend(pColInfo, numOfRows, (const char *)&pObj->id, false);
 
     char b1[TSDB_EP_LEN + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(b1, pObj->pDnode->ep, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(b1, pObj->pDnode->ep, pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, b1, false);
 
     const char *roles = syncStr(pObj->role);
-    char       *b2 = taosMemoryCalloc(1, strlen(roles) + VARSTR_HEADER_SIZE);
-    STR_WITH_MAXSIZE_TO_VARSTR(b2, roles, pShow->bytes[cols]);
+    char       *b2 = taosMemoryCalloc(1, 12 + VARSTR_HEADER_SIZE);
+    STR_WITH_MAXSIZE_TO_VARSTR(b2, roles, pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)b2, false);

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

@@ -570,38 +570,39 @@ static int32_t mndRetrieveConns(SNodeMsg *pReq, SShowObj *pShow, char *data, int
     if (pConn == NULL) break;
 
     cols = 0;
-
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+#if 0
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
     *(uint32_t *)pWrite = pConn->id;
     cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConn->user, pShow->bytes[cols]);
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConn->user, pShow->pMeta->pSchemas[cols].bytes);
     cols++;
 
     // app name
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConn->app, pShow->bytes[cols]);
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConn->app, pShow->pMeta->pSchemas[cols].bytes);
     cols++;
 
     // app pid
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
     *(int32_t *)pWrite = pConn->pid;
     cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
     taosIpPort2String(pConn->ip, pConn->port, ipStr);
-    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, pShow->pMeta->pSchemas[cols].bytes);
     cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
     *(int64_t *)pWrite = pConn->loginTimeMs;
     cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
     if (pConn->lastAccessTimeMs < pConn->loginTimeMs) pConn->lastAccessTimeMs = pConn->loginTimeMs;
     *(int64_t *)pWrite = pConn->lastAccessTimeMs;
     cols++;
+#endif
 
     numOfRows++;
   }
@@ -643,67 +644,67 @@ static int32_t mndRetrieveQueries(SNodeMsg *pReq, SShowObj *pShow, char *data, i
       SQueryDesc *pDesc = pConn->pQueries + i;
       cols = 0;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(int64_t *)pWrite = htobe64(pDesc->queryId);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(int64_t *)pWrite = htobe64(pConn->id);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConn->user, pShow->bytes[cols]);
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConn->user, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       snprintf(str, tListLen(str), "%s:%u", taosIpStr(pConn->ip), pConn->port);
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, str, pShow->bytes[cols]);
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, str, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
       char handleBuf[24] = {0};
       snprintf(handleBuf, tListLen(handleBuf), "%" PRIu64, htobe64(pDesc->qId));
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
 
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, handleBuf, pShow->bytes[cols]);
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, handleBuf, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(int64_t *)pWrite = htobe64(pDesc->stime);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(int64_t *)pWrite = htobe64(pDesc->useconds);
       cols++;
 
       snprintf(str, tListLen(str), "0x%" PRIx64, htobe64(pDesc->sqlObjId));
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, str, pShow->bytes[cols]);
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, str, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(int32_t *)pWrite = htonl(pDesc->pid);
       cols++;
 
       char epBuf[TSDB_EP_LEN + 1] = {0};
       snprintf(epBuf, tListLen(epBuf), "%s:%u", pDesc->fqdn, pConn->port);
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, epBuf, pShow->bytes[cols]);
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, epBuf, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(bool *)pWrite = pDesc->stableQuery;
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
       *(int32_t *)pWrite = htonl(pDesc->numOfSub);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDesc->subSqlInfo, pShow->bytes[cols]);
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDesc->subSqlInfo, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
-      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDesc->sql, pShow->bytes[cols]);
+      pWrite = data + pShow->offset[cols] * rows + pShow->pMeta->pSchemas[cols].bytes * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDesc->sql, pShow->pMeta->pSchemas[cols].bytes);
       cols++;
 
       numOfRows++;

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

@@ -517,7 +517,7 @@ static int32_t mndRetrieveQnodes(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *p
     colDataAppend(pColInfo, numOfRows, (const char *)&pObj->id, false);
 
     char ep[TSDB_EP_LEN + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(ep, pObj->pDnode->ep, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(ep, pObj->pDnode->ep, pShow->pMeta->pSchemas[cols].bytes);
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)ep, false);
 

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

@@ -209,16 +209,6 @@ static int32_t mndProcessRetrieveSysTableReq(SNodeMsg *pReq) {
 
     pShow->pMeta = pMeta;
     pShow->numOfColumns = pShow->pMeta->numOfColumns;
-    int32_t offset = 0;
-
-    for (int32_t i = 0; i < pShow->pMeta->numOfColumns; ++i) {
-      pShow->offset[i] = offset;
-
-      int32_t bytes = pShow->pMeta->pSchemas[i].bytes;
-      pShow->rowSize += bytes;
-      pShow->bytes[i] = bytes;
-      offset += bytes;
-    }
   } else {
     pShow = mndAcquireShowObj(pMnode, retrieveReq.showId);
     if (pShow == NULL) {

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

@@ -463,7 +463,7 @@ static int32_t mndRetrieveSnodes(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *p
     colDataAppend(pColInfo, numOfRows, (const char *)&pObj->id, false);
 
     char ep[TSDB_EP_LEN + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(ep, pObj->pDnode->ep, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(ep, pObj->pDnode->ep, pShow->pMeta->pSchemas[cols].bytes);
 
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)ep, false);

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

@@ -428,17 +428,23 @@ static void *mndBuildVCreateStbReq(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pSt
 }
 
 static void *mndBuildVDropStbReq(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb, int32_t *pContLen) {
-  SName name = {0};
+  SName        name = {0};
+  SVDropStbReq req = {0};
+  int32_t      contLen = 0;
+  int32_t      ret = 0;
+  SMsgHead    *pHead = NULL;
+  SCoder       coder = {0};
+
   tNameFromString(&name, pStb->name, T_NAME_ACCT | T_NAME_DB | T_NAME_TABLE);
 
-  SVDropTbReq req = {0};
-  req.ver = 0;
   req.name = (char *)tNameGetTableName(&name);
-  req.type = TD_SUPER_TABLE;
   req.suid = pStb->uid;
 
-  int32_t   contLen = tSerializeSVDropTbReq(NULL, &req) + sizeof(SMsgHead);
-  SMsgHead *pHead = taosMemoryMalloc(contLen);
+  tEncodeSize(tEncodeSVDropStbReq, &req, contLen, ret);
+  if (ret < 0) return NULL;
+
+  contLen += sizeof(SMsgHead);
+  pHead = taosMemoryMalloc(contLen);
   if (pHead == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
@@ -448,7 +454,10 @@ static void *mndBuildVDropStbReq(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb,
   pHead->vgId = htonl(pVgroup->vgId);
 
   void *pBuf = POINTER_SHIFT(pHead, sizeof(SMsgHead));
-  tSerializeSVDropTbReq(&pBuf, &req);
+
+  tCoderInit(&coder, TD_LITTLE_ENDIAN, pBuf, contLen - sizeof(SMsgHead), TD_ENCODER);
+  tEncodeSVDropStbReq(&coder, &req);
+  tCoderClear(&coder);
 
   *pContLen = contLen;
   return pHead;
@@ -670,8 +679,8 @@ static int32_t mndCreateStb(SMnode *pMnode, SNodeMsg *pReq, SMCreateStbReq *pCre
     memcpy(stbObj.pAst2, pCreate->pAst2, stbObj.ast2Len);
   }
 
-  stbObj.pColumns = taosMemoryMalloc(stbObj.numOfColumns * sizeof(SSchema));
-  stbObj.pTags = taosMemoryMalloc(stbObj.numOfTags * sizeof(SSchema));
+  stbObj.pColumns = taosMemoryCalloc(1, stbObj.numOfColumns * sizeof(SSchema));
+  stbObj.pTags = taosMemoryCalloc(1, stbObj.numOfTags * sizeof(SSchema));
   if (stbObj.pColumns == NULL || stbObj.pTags == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
@@ -722,7 +731,6 @@ _OVER:
 static int32_t mndProcessMCreateStbReq(SNodeMsg *pReq) {
   SMnode        *pMnode = pReq->pNode;
   int32_t        code = -1;
-  SStbObj       *pTopicStb = NULL;
   SStbObj       *pStb = NULL;
   SDbObj        *pDb = NULL;
   SUserObj      *pUser = NULL;
@@ -753,12 +761,6 @@ static int32_t mndProcessMCreateStbReq(SNodeMsg *pReq) {
     goto _OVER;
   }
 
-  pTopicStb = mndAcquireStb(pMnode, createReq.name);
-  if (pTopicStb != NULL) {
-    terrno = TSDB_CODE_MND_NAME_CONFLICT_WITH_TOPIC;
-    goto _OVER;
-  }
-
   pDb = mndAcquireDbByStb(pMnode, createReq.name);
   if (pDb == NULL) {
     terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
@@ -776,7 +778,7 @@ static int32_t mndProcessMCreateStbReq(SNodeMsg *pReq) {
 
   int32_t numOfStbs = -1;
   mndGetNumOfStbs(pMnode, pDb->name, &numOfStbs);
-  if (pDb->cfg.numOfStables == 1 && numOfStbs != 0 ) {
+  if (pDb->cfg.numOfStables == 1 && numOfStbs != 0) {
     terrno = TSDB_CODE_MND_SINGLE_STB_MODE_DB;
     goto _OVER;
   }
@@ -790,7 +792,6 @@ _OVER:
   }
 
   mndReleaseStb(pMnode, pStb);
-  mndReleaseStb(pMnode, pTopicStb);
   mndReleaseDb(pMnode, pDb);
   mndReleaseUser(pMnode, pUser);
   tFreeSMCreateStbReq(&createReq);
@@ -1102,7 +1103,7 @@ static int32_t mndSetAlterStbRedoLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pD
   SSdbRaw *pRedoRaw = mndStbActionEncode(pStb);
   if (pRedoRaw == NULL) return -1;
   if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
-  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_UPDATING) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_READY) != 0) return -1;
 
   return 0;
 }

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

@@ -80,6 +80,7 @@ static SMqSubscribeObj *mndCreateSub(SMnode *pMnode, const SMqTopicObj *pTopic,
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
   }
+  pSub->dbUid = pTopic->dbUid;
   pSub->subType = pTopic->subType;
   pSub->withTbName = pTopic->withTbName;
   pSub->withSchema = pTopic->withSchema;
@@ -144,6 +145,10 @@ static int32_t mndPersistSubChangeVgReq(SMnode *pMnode, STrans *pTrans, const SM
 
   int32_t vgId = pRebVg->pVgEp->vgId;
   SVgObj *pVgObj = mndAcquireVgroup(pMnode, vgId);
+  if (pVgObj == NULL) {
+    taosMemoryFree(buf);
+    return -1;
+  }
 
   STransAction action = {0};
   action.epSet = mndGetVgroupEpset(pMnode, pVgObj);
@@ -170,27 +175,20 @@ static int32_t mndSplitSubscribeKey(const char *key, char *topic, char *cgroup)
   return 0;
 }
 
-static SMqRebSubscribe *mndGetOrCreateRebSub(SHashObj *pHash, const char *key) {
-  SMqRebSubscribe *pRebSub = taosHashGet(pHash, key, strlen(key) + 1);
+static SMqRebInfo *mndGetOrCreateRebSub(SHashObj *pHash, const char *key) {
+  SMqRebInfo *pRebSub = taosHashGet(pHash, key, strlen(key) + 1);
   if (pRebSub == NULL) {
     pRebSub = tNewSMqRebSubscribe(key);
     if (pRebSub == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return NULL;
     }
-    taosHashPut(pHash, key, strlen(key) + 1, pRebSub, sizeof(SMqRebSubscribe));
+    taosHashPut(pHash, key, strlen(key) + 1, pRebSub, sizeof(SMqRebInfo));
   }
   return pRebSub;
 }
 
 static int32_t mndDoRebalance(SMnode *pMnode, const SMqRebInputObj *pInput, SMqRebOutputObj *pOutput) {
-  if (pInput->pTopic != NULL) {
-    // create subscribe
-    pOutput->pSub = mndCreateSub(pMnode, pInput->pTopic, pInput->pRebInfo->key);
-    ASSERT(taosHashGetSize(pOutput->pSub->consumerHash) == 0);
-  } else {
-    pOutput->pSub = tCloneSubscribeObj(pInput->pOldSub);
-  }
   int32_t totalVgNum = pOutput->pSub->vgNum;
 
   mInfo("mq rebalance subscription: %s, vgNum: %d", pOutput->pSub->key, pOutput->pSub->vgNum);
@@ -241,12 +239,8 @@ static int32_t mndDoRebalance(SMnode *pMnode, const SMqRebInputObj *pInput, SMqR
   }
 
   // 3. calc vg number of each consumer
-  int32_t oldSz = 0;
-  if (pInput->pOldSub) {
-    oldSz = taosHashGetSize(pInput->pOldSub->consumerHash);
-  }
-  int32_t afterRebConsumerNum =
-      oldSz + taosArrayGetSize(pInput->pRebInfo->newConsumers) - taosArrayGetSize(pInput->pRebInfo->removedConsumers);
+  int32_t afterRebConsumerNum = pInput->oldConsumerNum + taosArrayGetSize(pInput->pRebInfo->newConsumers) -
+                                taosArrayGetSize(pInput->pRebInfo->removedConsumers);
   int32_t minVgCnt = 0;
   int32_t imbConsumerNum = 0;
   // calc num
@@ -484,22 +478,34 @@ static int32_t mndProcessRebalanceReq(SNodeMsg *pMsg) {
     rebOutput.touchedConsumers = taosArrayInit(0, sizeof(void *));
     rebOutput.rebVgs = taosArrayInit(0, sizeof(SMqRebOutputVg));
 
-    SMqRebSubscribe *pRebSub = (SMqRebSubscribe *)pIter;
-    SMqSubscribeObj *pSub = mndAcquireSubscribeByKey(pMnode, pRebSub->key);
+    SMqRebInfo      *pRebInfo = (SMqRebInfo *)pIter;
+    SMqSubscribeObj *pSub = mndAcquireSubscribeByKey(pMnode, pRebInfo->key);
+
+    rebInput.pRebInfo = pRebInfo;
 
     if (pSub == NULL) {
       // split sub key and extract topic
       char topic[TSDB_TOPIC_FNAME_LEN];
       char cgroup[TSDB_CGROUP_LEN];
-      mndSplitSubscribeKey(pRebSub->key, topic, cgroup);
+      mndSplitSubscribeKey(pRebInfo->key, topic, cgroup);
       SMqTopicObj *pTopic = mndAcquireTopic(pMnode, topic);
       ASSERT(pTopic);
       taosRLockLatch(&pTopic->lock);
-      rebInput.pTopic = pTopic;
-    }
 
-    rebInput.pRebInfo = pRebSub;
-    rebInput.pOldSub = pSub;
+      rebOutput.pSub = mndCreateSub(pMnode, pTopic, pRebInfo->key);
+      ASSERT(taosHashGetSize(rebOutput.pSub->consumerHash) == 0);
+
+      taosRUnLockLatch(&pTopic->lock);
+      mndReleaseTopic(pMnode, pTopic);
+
+      rebInput.oldConsumerNum = 0;
+    } else {
+      taosRLockLatch(&pSub->lock);
+      rebInput.oldConsumerNum = taosHashGetSize(pSub->consumerHash);
+      rebOutput.pSub = tCloneSubscribeObj(pSub);
+      taosRUnLockLatch(&pSub->lock);
+      mndReleaseSubscribe(pMnode, pSub);
+    }
 
     // TODO replace assert with error check
     ASSERT(mndDoRebalance(pMnode, &rebInput, &rebOutput) == 0);
@@ -509,14 +515,8 @@ static int32_t mndProcessRebalanceReq(SNodeMsg *pMsg) {
     /*ASSERT(taosArrayGetSize(rebOutput.rebVgs) != 0);*/
 
     // TODO replace assert with error check
-    ASSERT(mndPersistRebResult(pMnode, pMsg, &rebOutput) == 0);
-
-    if (rebInput.pTopic) {
-      SMqTopicObj *pTopic = (SMqTopicObj *)rebInput.pTopic;
-      taosRUnLockLatch(&pTopic->lock);
-      mndReleaseTopic(pMnode, pTopic);
-    } else {
-      mndReleaseSubscribe(pMnode, pSub);
+    if (mndPersistRebResult(pMnode, pMsg, &rebOutput) < 0) {
+      mError("persist rebalance output error, possibly vnode splitted or dropped");
     }
   }
 
@@ -587,7 +587,7 @@ static SSdbRow *mndSubActionDecode(SSdbRaw *pRaw) {
   int32_t dataPos = 0;
   int32_t tlen;
   SDB_GET_INT32(pRaw, dataPos, &tlen, SUB_DECODE_OVER);
-  buf = taosMemoryMalloc(tlen + 1);
+  buf = taosMemoryMalloc(tlen);
   if (buf == NULL) goto SUB_DECODE_OVER;
   SDB_GET_BINARY(pRaw, dataPos, buf, tlen, SUB_DECODE_OVER);
   SDB_GET_RESERVE(pRaw, dataPos, MND_SUBSCRIBE_RESERVE_SIZE, SUB_DECODE_OVER);
@@ -673,3 +673,36 @@ static int32_t mndProcessSubscribeInternalRsp(SNodeMsg *pRsp) {
   mndTransProcessRsp(pRsp);
   return 0;
 }
+
+static int32_t mndSetDropSubCommitLogs(SMnode *pMnode, STrans *pTrans, SMqSubscribeObj *pSub) {
+  SSdbRaw *pCommitRaw = mndSubActionEncode(pSub);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+int32_t mndDropSubByDB(SMnode *pMnode, STrans *pTrans, SDbObj *pDb) {
+  int32_t code = -1;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  void            *pIter = NULL;
+  SMqSubscribeObj *pSub = NULL;
+  while (1) {
+    pIter = sdbFetch(pSdb, SDB_SUBSCRIBE, pIter, (void **)&pSub);
+    if (pIter == NULL) break;
+
+    if (pSub->dbUid != pDb->uid) {
+      sdbRelease(pSdb, pSub);
+      continue;
+    }
+
+    if (mndSetDropSubCommitLogs(pMnode, pTrans, pSub) < 0) {
+      goto END;
+    }
+  }
+
+  code = 0;
+END:
+  return code;
+}

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

@@ -38,6 +38,8 @@ static int32_t mndProcessDropTopicInRsp(SNodeMsg *pRsp);
 static int32_t mndRetrieveTopic(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
 static void    mndCancelGetNextTopic(SMnode *pMnode, void *pIter);
 
+static int32_t mndSetDropTopicCommitLogs(SMnode *pMnode, STrans *pTrans, SMqTopicObj *pTopic);
+
 int32_t mndInitTopic(SMnode *pMnode) {
   SSdbTable table = {.sdbType = SDB_TOPIC,
                      .keyType = SDB_KEY_BINARY,
@@ -553,7 +555,41 @@ static int32_t mndRetrieveTopic(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pB
   return numOfRows;
 }
 
+static int32_t mndSetDropTopicCommitLogs(SMnode *pMnode, STrans *pTrans, SMqTopicObj *pTopic) {
+  SSdbRaw *pCommitRaw = mndTopicActionEncode(pTopic);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+
+  return 0;
+}
+
 static void mndCancelGetNextTopic(SMnode *pMnode, void *pIter) {
   SSdb *pSdb = pMnode->pSdb;
   sdbCancelFetch(pSdb, pIter);
 }
+
+int32_t mndDropTopicByDB(SMnode *pMnode, STrans *pTrans, SDbObj *pDb) {
+  int32_t code = -1;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  void        *pIter = NULL;
+  SMqTopicObj *pTopic = NULL;
+  while (1) {
+    pIter = sdbFetch(pSdb, SDB_TOPIC, pIter, (void **)&pTopic);
+    if (pIter == NULL) break;
+
+    if (pTopic->dbUid != pDb->uid) {
+      sdbRelease(pSdb, pTopic);
+      continue;
+    }
+
+    if (mndSetDropTopicCommitLogs(pMnode, pTrans, pTopic) < 0) {
+      goto END;
+    }
+  }
+
+  code = 0;
+END:
+  return code;
+}

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

@@ -25,7 +25,6 @@
 #define USER_RESERVE_SIZE 64
 
 static int32_t  mndCreateDefaultUsers(SMnode *pMnode);
-static SSdbRaw *mndUserActionEncode(SUserObj *pUser);
 static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw);
 static int32_t  mndUserActionInsert(SSdb *pSdb, SUserObj *pUser);
 static int32_t  mndUserActionDelete(SSdb *pSdb, SUserObj *pUser);
@@ -90,7 +89,7 @@ static int32_t mndCreateDefaultUsers(SMnode *pMnode) {
   return 0;
 }
 
-static SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
+SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
 
   int32_t numOfReadDbs = taosHashGetSize(pUser->readDbs);
@@ -238,7 +237,7 @@ static int32_t mndUserActionUpdate(SSdb *pSdb, SUserObj *pOld, SUserObj *pNew) {
   return 0;
 }
 
-SUserObj *mndAcquireUser(SMnode *pMnode, char *userName) {
+SUserObj *mndAcquireUser(SMnode *pMnode, const char *userName) {
   SSdb     *pSdb = pMnode->pSdb;
   SUserObj *pUser = sdbAcquire(pSdb, SDB_USER, userName);
   if (pUser == NULL) {
@@ -276,9 +275,6 @@ static int32_t mndCreateUser(SMnode *pMnode, char *acct, SCreateUserReq *pCreate
   }
   sdbSetRawStatus(pRedoRaw, SDB_STATUS_READY);
 
-  char *param = strdup("====> test code to be deleted later <=====");
-  mndTransSetCb(pTrans, TEST_TRANS_START_FUNC, TEST_TRANS_STOP_FUNC, param, strlen(param) + 1);
-
   if (mndTransPrepare(pMnode, pTrans) != 0) {
     mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
     mndTransDrop(pTrans);
@@ -657,7 +653,7 @@ static int32_t mndRetrieveUsers(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pB
     SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols);
 
     char name[TSDB_USER_LEN + VARSTR_HEADER_SIZE] = {0};
-    STR_WITH_MAXSIZE_TO_VARSTR(name, pUser->user, pShow->bytes[cols]);
+    STR_WITH_MAXSIZE_TO_VARSTR(name, pUser->user, pShow->pMeta->pSchemas[cols].bytes);
 
     colDataAppend(pColInfo, numOfRows, (const char *)name, false);
 

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

@@ -540,7 +540,7 @@ static int32_t mndRetrieveVgroups(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock*
 
         char        buf1[20] = {0};
         const char *role = syncStr(pVgroup->vnodeGid[i].role);
-        STR_WITH_MAXSIZE_TO_VARSTR(buf1, role, pShow->bytes[cols]);
+        STR_WITH_MAXSIZE_TO_VARSTR(buf1, role,   pShow->pMeta->pSchemas[cols].bytes);
 
         pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
         colDataAppend(pColInfo, numOfRows, (const char *)buf1, false);

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

@@ -43,9 +43,6 @@
 #include "mndUser.h"
 #include "mndVgroup.h"
 
-#define MQ_TIMER_MS    2000
-#define TRNAS_TIMER_MS 6000
-
 static void *mndBuildTimerMsg(int32_t *pContLen) {
   SMTimerReq timerReq = {0};
 
@@ -68,7 +65,7 @@ static void mndPullupTrans(void *param, void *tmrId) {
     tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
   }
 
-  taosTmrReset(mndPullupTrans, TRNAS_TIMER_MS, pMnode, pMnode->timer, &pMnode->transTimer);
+  taosTmrReset(mndPullupTrans, tsTransPullupMs, pMnode, pMnode->timer, &pMnode->transTimer);
 }
 
 static void mndCalMqRebalance(void *param, void *tmrId) {
@@ -84,7 +81,7 @@ static void mndCalMqRebalance(void *param, void *tmrId) {
     tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
   }
 
-  taosTmrReset(mndCalMqRebalance, MQ_TIMER_MS, pMnode, pMnode->timer, &pMnode->mqTimer);
+  taosTmrReset(mndCalMqRebalance, tsMaRebalanceMs, pMnode, pMnode->timer, &pMnode->mqTimer);
 }
 
 static void mndPullupTelem(void *param, void *tmrId) {
@@ -106,12 +103,12 @@ static int32_t mndInitTimer(SMnode *pMnode) {
     return -1;
   }
 
-  if (taosTmrReset(mndPullupTrans, TRNAS_TIMER_MS, pMnode, pMnode->timer, &pMnode->transTimer)) {
+  if (taosTmrReset(mndPullupTrans, tsTransPullupMs, pMnode, pMnode->timer, &pMnode->transTimer)) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
-  if (taosTmrReset(mndCalMqRebalance, MQ_TIMER_MS, pMnode, pMnode->timer, &pMnode->mqTimer)) {
+  if (taosTmrReset(mndCalMqRebalance, tsMaRebalanceMs, pMnode, pMnode->timer, &pMnode->mqTimer)) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }

source/dnode/mnode/impl/test/trans/CMakeLists.txt

-aux_source_directory(. MNODE_TRANS_TEST_SRC)
-add_executable(transTest ${MNODE_TRANS_TEST_SRC})
+add_executable(transTest1 "")
+target_sources(transTest1
+    PRIVATE
+    "${CMAKE_CURRENT_SOURCE_DIR}/trans1.cpp"
+)
 target_link_libraries(
-    transTest
+    transTest1
     PUBLIC sut
 )
+target_include_directories(
+    transTest1
+    PUBLIC "${TD_SOURCE_DIR}/include/dnode/mnode"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/../../inc"
+)
+add_test(
+    NAME transTest1
+    COMMAND transTest1
+)
 
+add_executable(transTest2 "")
+target_sources(transTest2
+    PRIVATE
+    "${CMAKE_CURRENT_SOURCE_DIR}/trans2.cpp"
+)
+target_link_libraries(
+    transTest2
+    PUBLIC dnode mnode gtest_main
+)
+target_include_directories(
+    transTest2
+    PUBLIC "${TD_SOURCE_DIR}/include/dnode/mnode"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/../../inc"
+)
 add_test(
-    NAME transTest
-    COMMAND transTest
+    NAME transTest2
+    COMMAND transTest2
 )

source/dnode/mnode/impl/test/trans/trans.cpp → source/dnode/mnode/impl/test/trans/trans1.cpp

@@ -11,10 +11,10 @@
 
 #include "sut.h"
 
-class MndTestTrans : public ::testing::Test {
+class MndTestTrans1 : public ::testing::Test {
  protected:
   static void SetUpTestSuite() {
-    test.Init("/tmp/mnode_test_trans", 9013);
+    test.Init("/tmp/mnode_test_trans1", 9013);
     const char* fqdn = "localhost";
     const char* firstEp = "localhost:9013";
     server2.Start("/tmp/mnode_test_trans2", fqdn, 9020, firstEp);
@@ -26,7 +26,7 @@ class MndTestTrans : public ::testing::Test {
   }
 
   static void KillThenRestartServer() {
-    char      file[PATH_MAX] = "/tmp/mnode_test_trans/mnode/data/sdb.data";
+    char      file[PATH_MAX] = "/tmp/mnode_test_trans1/mnode/data/sdb.data";
     TdFilePtr pFile = taosOpenFile(file, TD_FILE_READ);
     int32_t   size = 3 * 1024 * 1024;
     void*     buffer = taosMemoryMalloc(size);
@@ -60,10 +60,10 @@ class MndTestTrans : public ::testing::Test {
   void TearDown() override {}
 };
 
-Testbase   MndTestTrans::test;
-TestServer MndTestTrans::server2;
+Testbase   MndTestTrans1::test;
+TestServer MndTestTrans1::server2;
 
-TEST_F(MndTestTrans, 00_Create_User_Crash) {
+TEST_F(MndTestTrans1, 00_Create_User_Crash) {
   {
     test.SendShowReq(TSDB_MGMT_TABLE_TRANS, "trans", "");
     EXPECT_EQ(test.GetShowRows(), 0);
@@ -83,7 +83,7 @@ TEST_F(MndTestTrans, 00_Create_User_Crash) {
   }
 }
 
-TEST_F(MndTestTrans, 01_Create_User_Crash) {
+TEST_F(MndTestTrans1, 01_Create_User_Crash) {
   {
     SCreateUserReq createReq = {0};
     strcpy(createReq.user, "u1");
@@ -107,7 +107,7 @@ TEST_F(MndTestTrans, 01_Create_User_Crash) {
   EXPECT_EQ(test.GetShowRows(), 2);
 }
 
-TEST_F(MndTestTrans, 02_Create_Qnode1_Crash) {
+TEST_F(MndTestTrans1, 02_Create_Qnode1_Crash) {
   {
     SMCreateQnodeReq createReq = {0};
     createReq.dnodeId = 1;
@@ -142,7 +142,7 @@ TEST_F(MndTestTrans, 02_Create_Qnode1_Crash) {
   }
 }
 
-TEST_F(MndTestTrans, 03_Create_Qnode2_Crash) {
+TEST_F(MndTestTrans1, 03_Create_Qnode2_Crash) {
   {
     SCreateDnodeReq createReq = {0};
     strcpy(createReq.fqdn, "localhost");

source/dnode/mnode/impl/test/trans/trans2.cpp

+/**
+ * @file trans.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief MNODE module trans tests
+ * @version 1.0
+ * @date 2022-05-02
+ *
+ * @copyright Copyright (c) 2022
+ *
+ */
+
+#include <gtest/gtest.h>
+
+#include "mndTrans.h"
+#include "mndUser.h"
+#include "tcache.h"
+
+void reportStartup(SMgmtWrapper *pWrapper, const char *name, const char *desc) {}
+
+class MndTestTrans2 : public ::testing::Test {
+ protected:
+  static void InitLog() {
+    dDebugFlag = 143;
+    vDebugFlag = 0;
+    mDebugFlag = 207;
+    cDebugFlag = 0;
+    jniDebugFlag = 0;
+    tmrDebugFlag = 135;
+    uDebugFlag = 135;
+    rpcDebugFlag = 143;
+    qDebugFlag = 0;
+    wDebugFlag = 0;
+    sDebugFlag = 0;
+    tsdbDebugFlag = 0;
+    tsLogEmbedded = 1;
+    tsAsyncLog = 0;
+
+    const char *logpath = "/tmp/td";
+    taosRemoveDir(logpath);
+    taosMkDir(logpath);
+    tstrncpy(tsLogDir, logpath, PATH_MAX);
+    if (taosInitLog("taosdlog", 1) != 0) {
+      printf("failed to init log file\n");
+    }
+  }
+
+  static void InitMnode() {
+    static SMsgCb msgCb = {0};
+    msgCb.reportStartupFp = reportStartup;
+    msgCb.pWrapper = (SMgmtWrapper *)(&msgCb);  // hack
+    tmsgSetDefaultMsgCb(&msgCb);
+
+    SMnodeOpt opt = {0};
+    opt.deploy = 1;
+    opt.replica = 1;
+    opt.replicas[0].id = 1;
+    opt.replicas[0].port = 9040;
+    strcpy(opt.replicas[0].fqdn, "localhost");
+    opt.msgCb = msgCb;
+
+    tsTransPullupMs = 1000;
+
+    const char *mnodepath = "/tmp/mnode_test_trans";
+    taosRemoveDir(mnodepath);
+    pMnode = mndOpen(mnodepath, &opt);
+    mndStart(pMnode);
+  }
+
+  static void SetUpTestSuite() {
+    InitLog();
+    walInit();
+    InitMnode();
+  }
+
+  static void TearDownTestSuite() {
+    mndStop(pMnode);
+    mndClose(pMnode);
+    walCleanUp();
+    taosCloseLog();
+    taosStopCacheRefreshWorker();
+  }
+
+  static SMnode *pMnode;
+
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+  int32_t CreateUser(const char *acct, const char *user) {
+    SUserObj userObj = {0};
+    taosEncryptPass_c((uint8_t *)"taosdata", strlen("taosdata"), userObj.pass);
+    tstrncpy(userObj.user, user, TSDB_USER_LEN);
+    tstrncpy(userObj.acct, acct, TSDB_USER_LEN);
+    userObj.createdTime = taosGetTimestampMs();
+    userObj.updateTime = userObj.createdTime;
+    userObj.superUser = 1;
+
+    SRpcMsg  rpcMsg = {0};
+    STrans  *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_CREATE_USER, &rpcMsg);
+    SSdbRaw *pRedoRaw = mndUserActionEncode(&userObj);
+    mndTransAppendRedolog(pTrans, pRedoRaw);
+    sdbSetRawStatus(pRedoRaw, SDB_STATUS_READY);
+
+    char *param = strdup("====> test param <=====");
+    mndTransSetCb(pTrans, TEST_TRANS_START_FUNC, TEST_TRANS_STOP_FUNC, param, strlen(param) + 1);
+
+    int32_t code = mndTransPrepare(pMnode, pTrans);
+    mndTransDrop(pTrans);
+
+    return code;
+  }
+};
+
+SMnode *MndTestTrans2::pMnode;
+
+TEST_F(MndTestTrans2, 01_CbFunc) {
+  const char *acct = "root";
+  const char *acct_invalid = "root1";
+  const char *user1 = "test1";
+  const char *user2 = "test2";
+  SUserObj   *pUser1 = NULL;
+  SUserObj   *pUser2 = NULL;
+
+  ASSERT_NE(pMnode, nullptr);
+
+  // create user success
+  EXPECT_EQ(CreateUser(acct, user1), 0);
+  pUser1 = mndAcquireUser(pMnode, user1);
+  ASSERT_NE(pUser1, nullptr);
+
+  // failed to create user and rollback
+  EXPECT_EQ(CreateUser(acct_invalid, user2), 0);
+  pUser2 = mndAcquireUser(pMnode, user2);
+  ASSERT_EQ(pUser2, nullptr);
+
+  mndTransPullup(pMnode);
+}

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

@@ -31,8 +31,6 @@ extern "C" {
 #define mDebug(...) { if (mDebugFlag & DEBUG_DEBUG) { taosPrintLog("MND ", DEBUG_DEBUG, mDebugFlag, __VA_ARGS__); }}
 #define mTrace(...) { if (mDebugFlag & DEBUG_TRACE) { taosPrintLog("MND ", DEBUG_TRACE, mDebugFlag, __VA_ARGS__); }}
 
-#define SDB_MAX_SIZE (32 * 1024)
-
 typedef struct SSdbRaw {
   int8_t  type;
   int8_t  status;

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

@@ -28,12 +28,12 @@ SSdb *sdbInit(SSdbOpt *pOption) {
     return NULL;
   }
 
-  char path[PATH_MAX + 100];
-  snprintf(path, PATH_MAX + 100, "%s%sdata", pOption->path, TD_DIRSEP);
+  char path[PATH_MAX + 100] = {0};
+  snprintf(path, sizeof(path), "%s%sdata", pOption->path, TD_DIRSEP);
   pSdb->currDir = strdup(path);
-  snprintf(path, PATH_MAX + 100, "%s%ssync", pOption->path, TD_DIRSEP);
+  snprintf(path, sizeof(path), "%s%ssync", pOption->path, TD_DIRSEP);
   pSdb->syncDir = strdup(path);
-  snprintf(path, PATH_MAX + 100, "%s%stmp", pOption->path, TD_DIRSEP);
+  snprintf(path, sizeof(path), "%s%stmp", pOption->path, TD_DIRSEP);
   pSdb->tmpDir = strdup(path);
   if (pSdb->currDir == NULL || pSdb->currDir == NULL || pSdb->currDir == NULL) {
     sdbCleanup(pSdb);
@@ -50,7 +50,7 @@ SSdb *sdbInit(SSdbOpt *pOption) {
   for (ESdbType i = 0; i < SDB_MAX; ++i) {
     taosInitRWLatch(&pSdb->locks[i]);
     pSdb->maxId[i] = 0;
-    pSdb->tableVer[i] = -1;
+    pSdb->tableVer[i] = 0;
     pSdb->keyTypes[i] = SDB_KEY_INT32;
   }
 
@@ -99,7 +99,7 @@ void sdbCleanup(SSdb *pSdb) {
     taosHashClear(hash);
     taosHashCleanup(hash);
     pSdb->hashObjs[i] = NULL;
-    mDebug("sdb table:%d is cleaned up", i);
+    mDebug("sdb table:%s is cleaned up", sdbTableName(i));
   }
 
   taosMemoryFree(pSdb);

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

@@ -18,7 +18,7 @@
 #include "tchecksum.h"
 #include "wal.h"
 
-#define SDB_TABLE_SIZE 24
+#define SDB_TABLE_SIZE   24
 #define SDB_RESERVE_SIZE 512
 
 static int32_t sdbRunDeployFp(SSdb *pSdb) {
@@ -50,7 +50,7 @@ static int32_t sdbReadFileHead(SSdb *pSdb, TdFilePtr pFile) {
   }
 
   for (int32_t i = 0; i < SDB_TABLE_SIZE; ++i) {
-    int64_t maxId = -1;
+    int64_t maxId = 0;
     ret = taosReadFile(pFile, &maxId, sizeof(int64_t));
     if (ret < 0) {
       terrno = TAOS_SYSTEM_ERROR(errno);
@@ -66,7 +66,7 @@ static int32_t sdbReadFileHead(SSdb *pSdb, TdFilePtr pFile) {
   }
 
   for (int32_t i = 0; i < SDB_TABLE_SIZE; ++i) {
-    int64_t ver = -1;
+    int64_t ver = 0;
     ret = taosReadFile(pFile, &ver, sizeof(int64_t));
     if (ret < 0) {
       terrno = TAOS_SYSTEM_ERROR(errno);
@@ -102,7 +102,7 @@ static int32_t sdbWriteFileHead(SSdb *pSdb, TdFilePtr pFile) {
   }
 
   for (int32_t i = 0; i < SDB_TABLE_SIZE; ++i) {
-    int64_t maxId = -1;
+    int64_t maxId = 0;
     if (i < SDB_MAX) {
       maxId = pSdb->maxId[i];
     }
@@ -113,7 +113,7 @@ static int32_t sdbWriteFileHead(SSdb *pSdb, TdFilePtr pFile) {
   }
 
   for (int32_t i = 0; i < SDB_TABLE_SIZE; ++i) {
-    int64_t ver = -1;
+    int64_t ver = 0;
     if (i < SDB_MAX) {
       ver = pSdb->tableVer[i];
     }
@@ -165,6 +165,9 @@ int32_t sdbReadFile(SSdb *pSdb) {
     return -1;
   }
 
+  int64_t tableVer[SDB_MAX] = {0};
+  memcpy(tableVer, pSdb->tableVer, sizeof(tableVer));
+
   while (1) {
     readLen = sizeof(SSdbRaw);
     ret = taosReadFile(pFile, pRaw, readLen);
@@ -206,15 +209,16 @@ int32_t sdbReadFile(SSdb *pSdb) {
     code = sdbWriteWithoutFree(pSdb, pRaw);
     if (code != 0) {
       mError("failed to read file:%s since %s", file, terrstr());
-      goto PARSE_SDB_DATA_ERROR;
+      goto _OVER;
     }
   }
 
   code = 0;
   pSdb->lastCommitVer = pSdb->curVer;
+  memcpy(pSdb->tableVer, tableVer, sizeof(tableVer));
   mDebug("read file:%s successfully, ver:%" PRId64, file, pSdb->lastCommitVer);
 
-PARSE_SDB_DATA_ERROR:
+_OVER:
   taosCloseFile(&pFile);
   sdbFreeRaw(pRaw);
 
@@ -259,7 +263,13 @@ static int32_t sdbWriteFileImp(SSdb *pSdb) {
     SSdbRow **ppRow = taosHashIterate(hash, NULL);
     while (ppRow != NULL) {
       SSdbRow *pRow = *ppRow;
-      if (pRow == NULL || pRow->status != SDB_STATUS_READY) {
+      if (pRow == NULL) {
+        ppRow = taosHashIterate(hash, ppRow);
+        continue;
+      }
+
+      if (pRow->status != SDB_STATUS_READY && pRow->status != SDB_STATUS_DROPPING) {
+        sdbPrintOper(pSdb, pRow, "not-write");
         ppRow = taosHashIterate(hash, ppRow);
         continue;
       }

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

@@ -134,6 +134,11 @@ int32_t sdbSetRawStatus(SSdbRaw *pRaw, ESdbStatus status) {
     return -1;
   }
 
+  if (status == SDB_STATUS_INIT) {
+    terrno = TSDB_CODE_INVALID_PARA;
+    return -1;
+  }
+
   pRaw->status = status;
   return 0;
 }

source/dnode/vnode/CMakeLists.txt

@@ -8,7 +8,6 @@ target_sources(
     "src/vnd/vnodeBufPool.c"
     "src/vnd/vnodeCfg.c"
     "src/vnd/vnodeCommit.c"
-    "src/vnd/vnodeInt.c"
     "src/vnd/vnodeQuery.c"
     "src/vnd/vnodeStateMgr.c"
     "src/vnd/vnodeModule.c"
@@ -33,7 +32,6 @@ target_sources(
     "src/tsdb/tsdbMemTable.c"
     "src/tsdb/tsdbRead.c"
     "src/tsdb/tsdbReadImpl.c"
-    "src/tsdb/tsdbScan.c"
     "src/tsdb/tsdbSma.c"
     "src/tsdb/tsdbWrite.c"
 

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

@@ -43,7 +43,6 @@ int metaEncodeEntry(SCoder* pCoder, const SMetaEntry* pME);
 int metaDecodeEntry(SCoder* pCoder, SMetaEntry* pME);
 
 // metaTable ==================
-int metaDropSTable(SMeta* pMeta, int64_t verison, SVDropStbReq* pReq);
 
 // metaQuery ==================
 int metaGetTableEntryByVersion(SMetaReader* pReader, int64_t version, tb_uid_t uid);

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

@@ -40,7 +40,6 @@ static FORCE_INLINE int32_t tsdbUidStoreInit(STbUidStore **pStore) {
   return TSDB_CODE_SUCCESS;
 }
 
-
 #ifdef __cplusplus
 }
 #endif