Merge branch '3.0' into feature/TD-11463-3.0

.clang-format

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

2.0/src/query/inc/qExecutor.h

@@ -574,11 +574,11 @@ typedef struct SMultiwayMergeInfo {
 } SMultiwayMergeInfo;
 
 // todo support the disk-based sort
-typedef struct SOrderOperatorInfo {
+typedef struct SSortOperatorInfo {
   int32_t      colIndex;
   int32_t      order;
   SSDataBlock *pDataBlock;
-} SOrderOperatorInfo;
+} SSortOperatorInfo;
 
 void appendUpstream(SOperatorInfo* p, SOperatorInfo* pUpstream);
 
@@ -609,7 +609,7 @@ SOperatorInfo* createFilterOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperator
                                         int32_t numOfOutput, SColumnInfo* pCols, int32_t numOfFilter);
 
 SOperatorInfo* createJoinOperatorInfo(SOperatorInfo** pUpstream, int32_t numOfUpstream, SSchema* pSchema, int32_t numOfOutput);
-SOperatorInfo* createOrderOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperatorInfo* upstream, SExprInfo* pExpr, int32_t numOfOutput, SOrderVal* pOrderVal);
+SOperatorInfo* createSortOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperatorInfo* upstream, SExprInfo* pExpr, int32_t numOfOutput, SOrderVal* pOrderVal);
 
 SSDataBlock* doGlobalAggregate(void* param, bool* newgroup);
 SSDataBlock* doMultiwayMergeSort(void* param, bool* newgroup);

2.0/src/query/src/qExecutor.c

@@ -2301,7 +2301,7 @@ static int32_t setupQueryRuntimeEnv(SQueryRuntimeEnv *pRuntimeEnv, int32_t numOf
       }
 
       case OP_Order: {
-        pRuntimeEnv->proot = createOrderOperatorInfo(pRuntimeEnv, pRuntimeEnv->proot, pQueryAttr->pExpr1, pQueryAttr->numOfOutput, &pQueryAttr->order);
+        pRuntimeEnv->proot = createSortOperatorInfo(pRuntimeEnv, pRuntimeEnv->proot, pQueryAttr->pExpr1, pQueryAttr->numOfOutput, &pQueryAttr->order);
         break;
       }
 
@@ -5516,7 +5516,7 @@ static SSDataBlock* doSort(void* param, bool* newgroup) {
     return NULL;
   }
 
-  SOrderOperatorInfo* pInfo = pOperator->info;
+  SSortOperatorInfo* pInfo = pOperator->info;
 
   SSDataBlock* pBlock = NULL;
   while(1) {
@@ -5556,8 +5556,8 @@ static SSDataBlock* doSort(void* param, bool* newgroup) {
   return (pInfo->pDataBlock->info.rows > 0)? pInfo->pDataBlock:NULL;
 }
 
-SOperatorInfo *createOrderOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperatorInfo* upstream, SExprInfo* pExpr, int32_t numOfOutput, SOrderVal* pOrderVal) {
+SOperatorInfo *createSortOperatorInfo(SQueryRuntimeEnv* pRuntimeEnv, SOperatorInfo* upstream, SExprInfo* pExpr, int32_t numOfOutput, SOrderVal* pOrderVal) {
-  SOrderOperatorInfo* pInfo = calloc(1, sizeof(SOrderOperatorInfo));
+  SSortOperatorInfo* pInfo = calloc(1, sizeof(SSortOperatorInfo));
 
   {
       SSDataBlock* pDataBlock = calloc(1, sizeof(SSDataBlock));
@@ -6611,7 +6611,7 @@ static void destroyTagScanOperatorInfo(void* param, int32_t numOfOutput) {
 }
 
 static void destroyOrderOperatorInfo(void* param, int32_t numOfOutput) {
-  SOrderOperatorInfo* pInfo = (SOrderOperatorInfo*) param;
+  SSortOperatorInfo* pInfo = (SSortOperatorInfo*) param;
   pInfo->pDataBlock = blockDataDestroy(pInfo->pDataBlock);
 }
 

example/src/tmq.c

@@ -80,8 +80,9 @@ int32_t create_topic() {
   }
   taos_free_result(pRes);
 
-  const char* sql = "select * from tu1";
+  /*const char* sql = "select * from tu1";*/
-  pRes = tmq_create_topic(pConn, "test_stb_topic_1", sql, strlen(sql));
+  /*pRes = tmq_create_topic(pConn, "test_stb_topic_1", sql, strlen(sql));*/
+  pRes = taos_query(pConn, "create topic test_stb_topic_1 as select * from tu1");
   if (taos_errno(pRes) != 0) {
     printf("failed to create topic test_stb_topic_1, reason:%s\n", taos_errstr(pRes));
     return -1;

example/src/tstream.c

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

include/client/consumer/consumer.h

@@ -1,78 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef _TD_CONSUMER_H_
-#define _TD_CONSUMER_H_
-
-#include "tlist.h"
-#include "tarray.h"
-#include "hash.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-  //consumer handle
-  struct tmq_consumer_t;
-  typedef struct tmq_consumer_t tmq_consumer_t;
-
-  //consumer config
-  struct tmq_consumer_config_t;
-  typedef struct tmq_consumer_config_t tmq_consumer_config_t;
-
-  //response err
-  struct tmq_resp_err_t;
-  typedef struct tmq_resp_err_t tmq_resp_err_t;
-
-  struct tmq_message_t;
-  typedef struct tmq_message_t tmq_message_t;
-
-  struct tmq_col_batch_t;
-  typedef struct tmq_col_batch_t tmq_col_batch_t;
-
-  //get content of message
-  tmq_col_batch_t* tmq_get_msg_col_by_idx(tmq_message_t*, int32_t col_id);
-  tmq_col_batch_t* tmq_get_msg_col_by_name(tmq_message_t*, const char*);
-
-  //consumer config
-  int32_t tmq_conf_set(tmq_consumer_config_t* , const char* config_key, const char* config_value, char* errstr, int32_t errstr_cap);
-
-  //consumer initialization
-  //resouces are supposed to be free by users by calling tmq_consumer_destroy
-  tmq_consumer_t* tmq_consumer_new(tmq_consumer_config_t* , char* errstr, int32_t errstr_cap);
-
-  //subscribe
-  tmq_resp_err_t tmq_subscribe(tmq_consumer_t*, const SList*);
-  tmq_resp_err_t tmq_unsubscribe(tmq_consumer_t*);
-
-  //consume
-  //resouces are supposed to be free by users by calling tmq_message_destroy
-  tmq_message_t* tmq_consume_poll(tmq_consumer_t*, int64_t blocking_time);
-
-  //destroy message and free memory
-  void tmq_message_destroy(tmq_message_t*);
-
-  //close consumer
-  int32_t tmq_consumer_close(tmq_consumer_t*);
-
-  //destroy consumer
-  void tmq_consumer_destroy(tmq_message_t*);
-
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_CONSUMER_H_*/

include/client/taos.h

@@ -26,9 +26,11 @@ extern "C" {
 typedef void   TAOS;
 typedef void   TAOS_STMT;
 typedef void   TAOS_RES;
-typedef void   TAOS_STREAM;
-typedef void   TAOS_SUB;
 typedef void **TAOS_ROW;
+#if 0
+typedef void   TAOS_STREAM;
+#endif
+typedef void   TAOS_SUB;
 
 // Data type definition
 #define TSDB_DATA_TYPE_NULL       0   // 1 bytes
@@ -122,8 +124,25 @@ typedef struct TAOS_MULTI_BIND {
   int       num;
 } TAOS_MULTI_BIND;
 
+typedef enum {
+  SET_CONF_RET_SUCC = 0,
+  SET_CONF_RET_ERR_PART = -1,
+  SET_CONF_RET_ERR_INNER = -2,
+  SET_CONF_RET_ERR_JSON_INVALID = -3,
+  SET_CONF_RET_ERR_JSON_PARSE = -4,
+  SET_CONF_RET_ERR_ONLY_ONCE = -5,
+  SET_CONF_RET_ERR_TOO_LONG = -6
+} SET_CONF_RET_CODE;
+
+#define RET_MSG_LENGTH 1024
+typedef struct setConfRet {
+  SET_CONF_RET_CODE retCode;
+  char   retMsg[RET_MSG_LENGTH];
+} setConfRet;
+
 DLL_EXPORT void  taos_cleanup(void);
 DLL_EXPORT int   taos_options(TSDB_OPTION option, const void *arg, ...);
+DLL_EXPORT setConfRet   taos_set_config(const char *config);
 DLL_EXPORT TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port);
 DLL_EXPORT TAOS *taos_connect_l(const char *ip, int ipLen, const char *user, int userLen, const char *pass, int passLen,
                                 const char *db, int dbLen, uint16_t port);
@@ -149,6 +168,7 @@ DLL_EXPORT int       taos_stmt_execute(TAOS_STMT *stmt);
 DLL_EXPORT TAOS_RES *taos_stmt_use_result(TAOS_STMT *stmt);
 DLL_EXPORT int       taos_stmt_close(TAOS_STMT *stmt);
 DLL_EXPORT char     *taos_stmt_errstr(TAOS_STMT *stmt);
+DLL_EXPORT int       taos_stmt_affected_rows(TAOS_STMT *stmt);
 
 DLL_EXPORT TAOS_RES *taos_query(TAOS *taos, const char *sql);
 DLL_EXPORT TAOS_RES *taos_query_l(TAOS *taos, const char *sql, int sqlLen);
@@ -165,10 +185,13 @@ DLL_EXPORT int         taos_select_db(TAOS *taos, const char *db);
 DLL_EXPORT int         taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields);
 DLL_EXPORT void        taos_stop_query(TAOS_RES *res);
 DLL_EXPORT bool        taos_is_null(TAOS_RES *res, int32_t row, int32_t col);
+DLL_EXPORT bool        taos_is_update_query(TAOS_RES *res);
 DLL_EXPORT int         taos_fetch_block(TAOS_RES *res, TAOS_ROW *rows);
 DLL_EXPORT int         taos_validate_sql(TAOS *taos, const char *sql);
+DLL_EXPORT void taos_reset_current_db(TAOS *taos);
 
 DLL_EXPORT int *taos_fetch_lengths(TAOS_RES *res);
+DLL_EXPORT TAOS_ROW *taos_result_block(TAOS_RES *res);
 
 DLL_EXPORT const char *taos_get_server_info(TAOS *taos);
 DLL_EXPORT const char *taos_get_client_info();
@@ -179,15 +202,20 @@ DLL_EXPORT int         taos_errno(TAOS_RES *tres);
 DLL_EXPORT void taos_query_a(TAOS *taos, const char *sql, __taos_async_fn_t fp, void *param);
 DLL_EXPORT void taos_fetch_rows_a(TAOS_RES *res, __taos_async_fn_t fp, void *param);
 
+// Shuduo: temporary enable for app build
+#if 1
 typedef void (*__taos_sub_fn_t)(TAOS_SUB *tsub, TAOS_RES *res, void *param, int code);
 DLL_EXPORT TAOS_SUB *taos_subscribe(TAOS *taos, int restart, const char *topic, const char *sql, __taos_sub_fn_t fp,
                                     void *param, int interval);
 DLL_EXPORT TAOS_RES *taos_consume(TAOS_SUB *tsub);
 DLL_EXPORT void      taos_unsubscribe(TAOS_SUB *tsub, int keepProgress);
+#endif
 
+#if 0
 DLL_EXPORT TAOS_STREAM *taos_open_stream(TAOS *taos, const char *sql, void (*fp)(void *param, TAOS_RES *, TAOS_ROW row),
                                          int64_t stime, void *param, void (*callback)(void *));
 DLL_EXPORT void         taos_close_stream(TAOS_STREAM *tstr);
+#endif
 
 DLL_EXPORT int       taos_load_table_info(TAOS *taos, const char *tableNameList);
 DLL_EXPORT TAOS_RES *taos_schemaless_insert(TAOS *taos, char *lines[], int numLines, int protocol, int precision);
@@ -215,8 +243,12 @@ DLL_EXPORT tmq_list_t *tmq_list_new();
 DLL_EXPORT int32_t     tmq_list_append(tmq_list_t *, const char *);
 DLL_EXPORT void        tmq_list_destroy(tmq_list_t *);
 
-DLL_EXPORT tmq_t      *tmq_consumer_new(void *conn, tmq_conf_t *conf, char *errstr, int32_t errstrLen);
+// will be removed in 3.0
-DLL_EXPORT tmq_t      *tmq_consumer_new1(tmq_conf_t *conf, char *errstr, int32_t errstrLen);
+DLL_EXPORT tmq_t *tmq_consumer_new(void *conn, tmq_conf_t *conf, char *errstr, int32_t errstrLen);
+
+// will replace last one
+DLL_EXPORT tmq_t *tmq_consumer_new1(tmq_conf_t *conf, char *errstr, int32_t errstrLen);
+
 DLL_EXPORT const char *tmq_err2str(tmq_resp_err_t);
 
 /* ------------------------TMQ CONSUMER INTERFACE------------------------ */
@@ -255,20 +287,26 @@ int32_t tmqGetSkipLogNum(tmq_message_t *tmq_message);
 
 /* -------------------------TMQ MSG HANDLE INTERFACE---------------------- */
 
-DLL_EXPORT TAOS_ROW tmq_get_row(tmq_message_t *message);
+DLL_EXPORT TAOS_ROW    tmq_get_row(tmq_message_t *message);
-DLL_EXPORT char    *tmq_get_topic_name(tmq_message_t *message);
+DLL_EXPORT char       *tmq_get_topic_name(tmq_message_t *message);
-DLL_EXPORT void    *tmq_get_topic_schema(tmq_t *tmq, const char *topic);
+DLL_EXPORT int32_t     tmq_get_vgroup_id(tmq_message_t *message);
-DLL_EXPORT void     tmq_message_destroy(tmq_message_t *tmq_message);
+DLL_EXPORT int64_t     tmq_get_request_offset(tmq_message_t *message);
+DLL_EXPORT int64_t     tmq_get_response_offset(tmq_message_t *message);
+DLL_EXPORT TAOS_FIELD *tmq_get_fields(tmq_t *tmq, const char *topic);
+DLL_EXPORT int32_t     tmq_field_count(tmq_t *tmq, const char *topic);
+DLL_EXPORT void        tmq_message_destroy(tmq_message_t *tmq_message);
 
 /* --------------------TMPORARY INTERFACE FOR TESTING--------------------- */
+#if 0
 DLL_EXPORT TAOS_RES *tmq_create_topic(TAOS *taos, const char *name, const char *sql, int sqlLen);
+#endif
 
 DLL_EXPORT TAOS_RES *tmq_create_stream(TAOS *taos, const char *streamName, const char *tbName, const char *sql);
 
-/* -------------------------------- OTHER -------------------------------- */
+/* ------------------------------ TMQ END -------------------------------- */
+#if 1 // Shuduo: temporary enable for app build
 typedef void (*TAOS_SUBSCRIBE_CALLBACK)(TAOS_SUB *tsub, TAOS_RES *res, void *param, int code);
-
+#endif
-DLL_EXPORT int taos_stmt_affected_rows(TAOS_STMT *stmt);
 
 #ifdef __cplusplus
 }

include/common/tcommon.h

@@ -197,6 +197,11 @@ typedef struct SGroupbyExpr {
   bool    groupbyTag;  // group by tag or column
 } SGroupbyExpr;
 
+enum {
+  FUNC_PARAM_TYPE_VALUE = 0,
+  FUNC_PARAM_TYPE_COLUMN,
+};
+
 typedef struct SFunctParam {
   int32_t  type;
   SColumn* pCol;

include/common/tdatablock.h

@@ -29,8 +29,9 @@ typedef struct SCorEpSet {
 } SCorEpSet;
 
 typedef struct SBlockOrderInfo {
+  bool             nullFirst;
   int32_t          order;
-  int32_t          colIndex;
+  int32_t          slotId;
   SColumnInfoData* pColData;
 } SBlockOrderInfo;
 
@@ -132,7 +133,8 @@ static FORCE_INLINE int32_t colDataAppendInt32(SColumnInfoData* pColumnInfoData,
 }
 
 static FORCE_INLINE int32_t colDataAppendInt64(SColumnInfoData* pColumnInfoData, uint32_t currentRow, int64_t* v) {
-  ASSERT(pColumnInfoData->info.type == TSDB_DATA_TYPE_BIGINT || pColumnInfoData->info.type == TSDB_DATA_TYPE_UBIGINT);
+  int32_t type = pColumnInfoData->info.type;
+  ASSERT(type == TSDB_DATA_TYPE_BIGINT || type == TSDB_DATA_TYPE_UBIGINT || type == TSDB_DATA_TYPE_TIMESTAMP);
   char* p = pColumnInfoData->pData + pColumnInfoData->info.bytes * currentRow;
   *(int64_t*)p = *(int64_t*)v;
 }
@@ -174,17 +176,16 @@ size_t blockDataGetRowSize(SSDataBlock* pBlock);
 double blockDataGetSerialRowSize(const SSDataBlock* pBlock);
 size_t blockDataGetSerialMetaSize(const SSDataBlock* pBlock);
 
-SSchema* blockDataExtractSchema(const SSDataBlock* pBlock, int32_t* numOfCols);
+int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo);
-
-int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullFirst);
 int32_t blockDataSort_rv(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullFirst);
 
-int32_t      blockDataEnsureColumnCapacity(SColumnInfoData* pColumn, uint32_t numOfRows);
+int32_t blockDataEnsureColumnCapacity(SColumnInfoData* pColumn, uint32_t numOfRows);
-int32_t      blockDataEnsureCapacity(SSDataBlock* pDataBlock, uint32_t numOfRows);
+int32_t blockDataEnsureCapacity(SSDataBlock* pDataBlock, uint32_t numOfRows);
-void         blockDataCleanup(SSDataBlock* pDataBlock);
+void    blockDataCleanup(SSDataBlock* pDataBlock);
+size_t  blockDataGetCapacityInRow(const SSDataBlock* pBlock, size_t pageSize);
+void*   blockDataDestroy(SSDataBlock* pBlock);
+
 SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock);
-size_t       blockDataGetCapacityInRow(const SSDataBlock* pBlock, size_t pageSize);
-void*        blockDataDestroy(SSDataBlock* pBlock);
 
 void blockDebugShowData(const SArray* dataBlocks);
 

include/common/tglobal.h

@@ -51,7 +51,7 @@ extern int32_t tsCompatibleModel;
 extern bool    tsEnableSlaveQuery;
 extern bool    tsPrintAuth;
 extern int64_t tsTickPerDay[3];
-extern int32_t tsMultiProcess;
+extern bool    tsMultiProcess;
 
 // monitor
 extern bool     tsEnableMonitor;

include/common/tmsg.h

@@ -469,8 +469,7 @@ typedef struct {
   int32_t tz;  // query client timezone
   char    intervalUnit;
   char    slidingUnit;
-  char
+  char    offsetUnit;  // TODO Remove it, the offset is the number of precision tickle, and it must be a immutable duration.
-      offsetUnit;  // TODO Remove it, the offset is the number of precision tickle, and it must be a immutable duration.
   int8_t  precision;
   int64_t interval;
   int64_t sliding;
@@ -801,7 +800,10 @@ typedef struct SVgroupInfo {
   uint32_t hashBegin;
   uint32_t hashEnd;
   SEpSet   epSet;
-  int32_t  numOfTable;  // unit is TSDB_TABLE_NUM_UNIT
+  union {
+    int32_t numOfTable;  // unit is TSDB_TABLE_NUM_UNIT
+    int32_t taskId;      // used in stream
+  };
 } SVgroupInfo;
 
 typedef struct {
@@ -2185,23 +2187,6 @@ static FORCE_INLINE void* tDecodeTSmaWrapper(void* buf, STSmaWrapper* pSW) {
   return buf;
 }
 
-typedef struct {
-  int64_t uid;
-  int32_t numOfRows;
-  char*   colData;
-} SMqTbData;
-
-typedef struct {
-  char       topicName[TSDB_TOPIC_FNAME_LEN];
-  int64_t    committedOffset;
-  int64_t    reqOffset;
-  int64_t    rspOffset;
-  int32_t    skipLogNum;
-  int32_t    bodyLen;
-  int32_t    numOfTb;
-  SMqTbData* tbData;
-} SMqTopicData;
-
 typedef struct {
   int8_t  mqMsgType;
   int32_t code;
@@ -2229,8 +2214,11 @@ typedef struct {
 } SMqSubVgEp;
 
 typedef struct {
-  char    topic[TSDB_TOPIC_FNAME_LEN];
+  char        topic[TSDB_TOPIC_FNAME_LEN];
-  SArray* vgs;  // SArray<SMqSubVgEp>
+  int8_t      isSchemaAdaptive;
+  SArray*     vgs;  // SArray<SMqSubVgEp>
+  int32_t     numOfFields;
+  TAOS_FIELD* fields;
 } SMqSubTopicEp;
 
 typedef struct {
@@ -2251,32 +2239,6 @@ typedef struct {
   SArray*    topics;  // SArray<SMqSubTopicEp>
 } SMqCMGetSubEpRsp;
 
-typedef struct {
-  int32_t curBlock;
-  int32_t curRow;
-  void**  uData;
-} SMqRowIter;
-
-struct tmq_message_t {
-  SMqPollRsp msg;
-  void*      vg;
-  SMqRowIter iter;
-};
-
-#if 0
-struct tmq_message_t {
-  SMqRspHead head;
-  union {
-    SMqPollRsp       consumeRsp;
-    SMqCMGetSubEpRsp getEpRsp;
-  };
-  void*   extra;
-  int32_t curBlock;
-  int32_t curRow;
-  void**  uData;
-};
-#endif
-
 static FORCE_INLINE void tDeleteSMqSubTopicEp(SMqSubTopicEp* pSubTopicEp) { taosArrayDestroy(pSubTopicEp->vgs); }
 
 static FORCE_INLINE int32_t tEncodeSMqSubVgEp(void** buf, const SMqSubVgEp* pVgEp) {
@@ -2301,17 +2263,21 @@ static FORCE_INLINE void tDeleteSMqCMGetSubEpRsp(SMqCMGetSubEpRsp* pRsp) {
 static FORCE_INLINE int32_t tEncodeSMqSubTopicEp(void** buf, const SMqSubTopicEp* pTopicEp) {
   int32_t tlen = 0;
   tlen += taosEncodeString(buf, pTopicEp->topic);
+  tlen += taosEncodeFixedI8(buf, pTopicEp->isSchemaAdaptive);
   int32_t sz = taosArrayGetSize(pTopicEp->vgs);
   tlen += taosEncodeFixedI32(buf, sz);
   for (int32_t i = 0; i < sz; i++) {
     SMqSubVgEp* pVgEp = (SMqSubVgEp*)taosArrayGet(pTopicEp->vgs, i);
     tlen += tEncodeSMqSubVgEp(buf, pVgEp);
   }
+  tlen += taosEncodeFixedI32(buf, pTopicEp->numOfFields);
+  // tlen += taosEncodeBinary(buf, pTopicEp->fields, pTopicEp->numOfFields * sizeof(TAOS_FIELD));
   return tlen;
 }
 
 static FORCE_INLINE void* tDecodeSMqSubTopicEp(void* buf, SMqSubTopicEp* pTopicEp) {
   buf = taosDecodeStringTo(buf, pTopicEp->topic);
+  buf = taosDecodeFixedI8(buf, &pTopicEp->isSchemaAdaptive);
   int32_t sz;
   buf = taosDecodeFixedI32(buf, &sz);
   pTopicEp->vgs = taosArrayInit(sz, sizeof(SMqSubVgEp));
@@ -2323,6 +2289,8 @@ static FORCE_INLINE void* tDecodeSMqSubTopicEp(void* buf, SMqSubTopicEp* pTopicE
     buf = tDecodeSMqSubVgEp(buf, &vgEp);
     taosArrayPush(pTopicEp->vgs, &vgEp);
   }
+  buf = taosDecodeFixedI32(buf, &pTopicEp->numOfFields);
+  // buf = taosDecodeBinary(buf, (void**)&pTopicEp->fields, pTopicEp->numOfFields * sizeof(TAOS_FIELD));
   return buf;
 }
 

include/common/tmsgcb.h

@@ -25,6 +25,7 @@ extern "C" {
 typedef struct SRpcMsg      SRpcMsg;
 typedef struct SEpSet       SEpSet;
 typedef struct SMgmtWrapper SMgmtWrapper;
+
 typedef enum {
   QUERY_QUEUE,
   FETCH_QUEUE,
@@ -38,24 +39,31 @@ typedef enum {
 
 typedef int32_t (*PutToQueueFp)(SMgmtWrapper* pWrapper, SRpcMsg* pReq);
 typedef int32_t (*GetQueueSizeFp)(SMgmtWrapper* pWrapper, int32_t vgId, EQueueType qtype);
-typedef int32_t (*SendReqFp)(SMgmtWrapper* pWrapper, SEpSet* epSet, SRpcMsg* pReq);
+typedef int32_t (*SendReqFp)(SMgmtWrapper* pWrapper, const SEpSet* epSet, SRpcMsg* pReq);
 typedef int32_t (*SendMnodeReqFp)(SMgmtWrapper* pWrapper, SRpcMsg* pReq);
-typedef void (*SendRspFp)(SMgmtWrapper* pWrapper, SRpcMsg* pRsp);
+typedef void (*SendRspFp)(SMgmtWrapper* pWrapper, const SRpcMsg* pRsp);
+typedef void (*RegisterBrokenLinkArgFp)(SMgmtWrapper* pWrapper, SRpcMsg* pMsg);
+typedef void (*ReleaseHandleFp)(SMgmtWrapper* pWrapper, void* handle, int8_t type);
 
 typedef struct {
-  SMgmtWrapper*  pWrapper;
+  SMgmtWrapper*           pWrapper;
-  PutToQueueFp   queueFps[QUEUE_MAX];
+  PutToQueueFp            queueFps[QUEUE_MAX];
-  GetQueueSizeFp qsizeFp;
+  GetQueueSizeFp          qsizeFp;
-  SendReqFp      sendReqFp;
+  SendReqFp               sendReqFp;
-  SendMnodeReqFp sendMnodeReqFp;
+  SendMnodeReqFp          sendMnodeReqFp;
-  SendRspFp      sendRspFp;
+  SendRspFp               sendRspFp;
+  RegisterBrokenLinkArgFp registerBrokenLinkArgFp;
+  ReleaseHandleFp         releaseHandleFp;
 } SMsgCb;
 
+void    tmsgSetDefaultMsgCb(const SMsgCb* pMsgCb);
 int32_t tmsgPutToQueue(const SMsgCb* pMsgCb, EQueueType qtype, SRpcMsg* pReq);
 int32_t tmsgGetQueueSize(const SMsgCb* pMsgCb, int32_t vgId, EQueueType qtype);
-int32_t tmsgSendReq(const SMsgCb* pMsgCb, SEpSet* epSet, SRpcMsg* pReq);
+int32_t tmsgSendReq(const SMsgCb* pMsgCb, const SEpSet* epSet, SRpcMsg* pReq);
 int32_t tmsgSendMnodeReq(const SMsgCb* pMsgCb, SRpcMsg* pReq);
-void    tmsgSendRsp(const SMsgCb* pMsgCb, SRpcMsg* pRsp);
+void    tmsgSendRsp(const SRpcMsg* pRsp);
+void    tmsgRegisterBrokenLinkArg(const SMsgCb* pMsgCb, SRpcMsg* pMsg);
+void    tmsgReleaseHandle(void* handle, int8_t type);
 
 #ifdef __cplusplus
 }

include/common/ttokendef.h

@@ -138,47 +138,59 @@
 #define TK_INTERVAL                       120
 #define TK_TOPIC                          121
 #define TK_AS                             122
-#define TK_NK_BOOL                        123
+#define TK_EXPLAIN                        123
-#define TK_NK_VARIABLE                    124
+#define TK_ANALYZE                        124
-#define TK_BETWEEN                        125
+#define TK_VERBOSE                        125
-#define TK_IS                             126
+#define TK_NK_BOOL                        126
-#define TK_NULL                           127
+#define TK_RATIO                          127
-#define TK_NK_LT                          128
+#define TK_NULL                           128
-#define TK_NK_GT                          129
+#define TK_NK_VARIABLE                    129
-#define TK_NK_LE                          130
+#define TK_NK_UNDERLINE                   130
-#define TK_NK_GE                          131
+#define TK_ROWTS                          131
-#define TK_NK_NE                          132
+#define TK_TBNAME                         132
-#define TK_MATCH                          133
+#define TK_QSTARTTS                       133
-#define TK_NMATCH                         134
+#define TK_QENDTS                         134
-#define TK_IN                             135
+#define TK_WSTARTTS                       135
-#define TK_JOIN                           136
+#define TK_WENDTS                         136
-#define TK_INNER                          137
+#define TK_WDURATION                      137
-#define TK_SELECT                         138
+#define TK_BETWEEN                        138
-#define TK_DISTINCT                       139
+#define TK_IS                             139
-#define TK_WHERE                          140
+#define TK_NK_LT                          140
-#define TK_PARTITION                      141
+#define TK_NK_GT                          141
-#define TK_BY                             142
+#define TK_NK_LE                          142
-#define TK_SESSION                        143
+#define TK_NK_GE                          143
-#define TK_STATE_WINDOW                   144
+#define TK_NK_NE                          144
-#define TK_SLIDING                        145
+#define TK_MATCH                          145
-#define TK_FILL                           146
+#define TK_NMATCH                         146
-#define TK_VALUE                          147
+#define TK_IN                             147
-#define TK_NONE                           148
+#define TK_JOIN                           148
-#define TK_PREV                           149
+#define TK_INNER                          149
-#define TK_LINEAR                         150
+#define TK_SELECT                         150
-#define TK_NEXT                           151
+#define TK_DISTINCT                       151
-#define TK_GROUP                          152
+#define TK_WHERE                          152
-#define TK_HAVING                         153
+#define TK_PARTITION                      153
-#define TK_ORDER                          154
+#define TK_BY                             154
-#define TK_SLIMIT                         155
+#define TK_SESSION                        155
-#define TK_SOFFSET                        156
+#define TK_STATE_WINDOW                   156
-#define TK_LIMIT                          157
+#define TK_SLIDING                        157
-#define TK_OFFSET                         158
+#define TK_FILL                           158
-#define TK_ASC                            159
+#define TK_VALUE                          159
-#define TK_DESC                           160
+#define TK_NONE                           160
-#define TK_NULLS                          161
+#define TK_PREV                           161
-#define TK_FIRST                          162
+#define TK_LINEAR                         162
-#define TK_LAST                           163
+#define TK_NEXT                           163
+#define TK_GROUP                          164
+#define TK_HAVING                         165
+#define TK_ORDER                          166
+#define TK_SLIMIT                         167
+#define TK_SOFFSET                        168
+#define TK_LIMIT                          169
+#define TK_OFFSET                         170
+#define TK_ASC                            171
+#define TK_DESC                           172
+#define TK_NULLS                          173
+#define TK_FIRST                          174
+#define TK_LAST                           175
 
 #define TK_NK_SPACE                       300
 #define TK_NK_COMMENT                     301

include/dnode/mgmt/dnode.h

@@ -46,11 +46,12 @@ typedef struct {
   char      localFqdn[TSDB_FQDN_LEN];
   char      firstEp[TSDB_EP_LEN];
   char      secondEp[TSDB_EP_LEN];
-  SDiskCfg *pDisks;
+  SDiskCfg *disks;
   int32_t   numOfDisks;
+  int8_t    ntype;
 } SDnodeOpt;
 
-typedef enum { DND_EVENT_START, DND_EVENT_STOP = 1, DND_EVENT_RELOAD } EDndEvent;
+typedef enum { DND_EVENT_START, DND_EVENT_STOP = 1, DND_EVENT_CHILD } EDndEvent;
 
 /**
  * @brief Initialize and start the dnode.

include/libs/executor/executor.h

@@ -20,8 +20,8 @@
 extern "C" {
 #endif
 
-#include "tcommon.h"
 #include "query.h"
+#include "tcommon.h"
 
 typedef void* qTaskInfo_t;
 typedef void* DataSinkHandle;
@@ -36,7 +36,12 @@ typedef struct SReadHandle {
 
 #define STREAM_DATA_TYPE_SUBMIT_BLOCK 0x1
 #define STREAM_DATA_TYPE_SSDATA_BLOCK 0x2
-  
+
+typedef enum {
+  OPTR_EXEC_MODEL_BATCH  = 0x1,
+  OPTR_EXEC_MODEL_STREAM = 0x2,
+} EOPTR_EXEC_MODEL;
+
 /**
  * Create the exec task for streaming mode
  * @param pMsg
@@ -54,6 +59,16 @@ qTaskInfo_t qCreateStreamExecTaskInfo(void* msg, void* streamReadHandle);
  */
 int32_t qSetStreamInput(qTaskInfo_t tinfo, const void* input, int32_t type);
 
+/**
+ * Set multiple input data blocks for the stream scan.
+ * @param tinfo
+ * @param pBlocks
+ * @param numOfInputBlock
+ * @param type
+ * @return
+ */
+int32_t qSetMultiStreamInput(qTaskInfo_t tinfo, const void* pBlocks, size_t numOfBlocks, int32_t type);
+
 /**
  * Update the table id list, add or remove.
  *
@@ -74,7 +89,7 @@ int32_t qUpdateQualifiedTableId(qTaskInfo_t tinfo, SArray* tableIdList, bool isA
  * @return
  */
 int32_t qCreateExecTask(SReadHandle* readHandle, int32_t vgId, uint64_t taskId, struct SSubplan* pPlan,
-                        qTaskInfo_t* pTaskInfo, DataSinkHandle* handle);
+                        qTaskInfo_t* pTaskInfo, DataSinkHandle* handle, EOPTR_EXEC_MODEL model);
 
 /**
  * The main task execution function, including query on both table and multiple tables,
@@ -86,16 +101,6 @@ int32_t qCreateExecTask(SReadHandle* readHandle, int32_t vgId, uint64_t taskId,
  */
 int32_t qExecTask(qTaskInfo_t tinfo, SSDataBlock** pRes, uint64_t* useconds);
 
-/**
- * Retrieve the produced results information, if current query is not paused or completed,
- * this function will be blocked to wait for the query execution completed or paused,
- * in which case enough results have been produced already.
- *
- * @param tinfo
- * @return
- */
-int32_t qRetrieveQueryResultInfo(qTaskInfo_t tinfo, bool* buildRes, void* pRspContext);
-
 /**
  * kill the ongoing query and free the query handle and corresponding resources automatically
  * @param tinfo  qhandle
@@ -158,50 +163,6 @@ int32_t qGetQualifiedTableIdList(void* pTableList, const char* tagCond, int32_t
  */
 int32_t qUpdateQueriedTableIdList(qTaskInfo_t tinfo, int64_t uid, int32_t type);
 
-//================================================================================================
-// query handle management
-/**
- * Query handle mgmt object
- * @param vgId
- * @return
- */
-void* qOpenTaskMgmt(int32_t vgId);
-
-/**
- * broadcast the close information and wait for all query stop.
- * @param pExecutor
- */
-void qTaskMgmtNotifyClosing(void* pExecutor);
-
-/**
- * Re-open the query handle management module when opening the vnode again.
- * @param pExecutor
- */
-void qQueryMgmtReOpen(void* pExecutor);
-
-/**
- * Close query mgmt and clean up resources.
- * @param pExecutor
- */
-void qCleanupTaskMgmt(void* pExecutor);
-
-/**
- * Add the query into the query mgmt object
- * @param pMgmt
- * @param qId
- * @param qInfo
- * @return
- */
-void** qRegisterTask(void* pMgmt, uint64_t qId, void* qInfo);
-
-/**
- * acquire the query handle according to the key from query mgmt object.
- * @param pMgmt
- * @param key
- * @return
- */
-void** qAcquireTask(void* pMgmt, uint64_t key);
-
 /**
  * release the query handle and decrease the reference count in cache
  * @param pMgmt
@@ -211,14 +172,6 @@ void** qAcquireTask(void* pMgmt, uint64_t key);
  */
 void** qReleaseTask(void* pMgmt, void* pQInfo, bool freeHandle);
 
-/**
- * De-register the query handle from the management module and free it immediately.
- * @param pMgmt
- * @param pQInfo
- * @return
- */
-void** qDeregisterQInfo(void* pMgmt, void* pQInfo);
-
 void qProcessFetchRsp(void* parent, struct SRpcMsg* pMsg, struct SEpSet* pEpSet);
 
 #ifdef __cplusplus

include/libs/function/function.h

@@ -41,6 +41,7 @@ typedef void (*FExecFinalize)(struct SqlFunctionCtx *pCtx);
 typedef int32_t (*FScalarExecProcess)(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 
 typedef struct SScalarFuncExecFuncs {
+  FExecGetEnv getEnv;
   FScalarExecProcess process;
 } SScalarFuncExecFuncs;
 
@@ -241,7 +242,6 @@ typedef struct tExprNode {
   };
 } tExprNode;
 
-void exprTreeToBinary(SBufferWriter* bw, tExprNode* pExprTree);
 void tExprTreeDestroy(tExprNode *pNode, void (*fp)(void *));
 
 typedef struct SAggFunctionInfo {
@@ -267,28 +267,6 @@ struct SScalarParam {
   int32_t            numOfRows;
 };
 
-typedef struct SMultiFunctionsDesc {
-  bool stableQuery;
-  bool groupbyColumn;
-  bool agg;
-  bool arithmeticOnAgg;
-  bool projectionQuery;
-  bool hasFilter;
-  bool onlyTagQuery;
-  bool orderProjectQuery;
-  bool globalMerge;
-  bool multigroupResult;
-  bool blockDistribution;
-  bool stateWindow;
-  bool timewindow;
-  bool sessionWindow;
-  bool topbotQuery;
-  bool interpQuery;
-  bool distinct;
-  bool join;
-  bool continueQuery;
-} SMultiFunctionsDesc;
-
 int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionId, int32_t param, SResultDataInfo* pInfo, int16_t extLength,
                           bool isSuperTable);
 
@@ -296,8 +274,6 @@ bool qIsValidUdf(SArray* pUdfInfo, const char* name, int32_t len, int32_t* funct
 
 tExprNode* exprTreeFromBinary(const void* data, size_t size);
 
-void extractFunctionDesc(SArray* pFunctionIdList, SMultiFunctionsDesc* pDesc);
-
 tExprNode* exprdup(tExprNode* pTree);
 
 void resetResultRowEntryResult(SqlFunctionCtx* pCtx, int32_t num);

include/libs/function/functionMgt.h

@@ -96,7 +96,16 @@ typedef enum EFunctionType {
   FUNCTION_TYPE_SERVER_SERSION,
   FUNCTION_TYPE_SERVER_STATUS,
   FUNCTION_TYPE_CURRENT_USER,
-  FUNCTION_TYPE_USER
+  FUNCTION_TYPE_USER,
+
+  // pseudo column function
+  FUNCTION_TYPE_ROWTS = 3500,
+  FUNCTION_TYPE_TBNAME,
+  FUNCTION_TYPE_QSTARTTS,
+  FUNCTION_TYPE_QENDTS,
+  FUNCTION_TYPE_WSTARTTS,
+  FUNCTION_TYPE_WENDTS,
+  FUNCTION_TYPE_WDURATION
 } EFunctionType;
 
 struct SqlFunctionCtx;
@@ -118,6 +127,8 @@ bool fmIsStringFunc(int32_t funcId);
 bool fmIsDatetimeFunc(int32_t funcId);
 bool fmIsTimelineFunc(int32_t funcId);
 bool fmIsTimeorderFunc(int32_t funcId);
+bool fmIsWindowPseudoColumnFunc(int32_t funcId);
+bool fmIsWindowClauseFunc(int32_t funcId);
 
 int32_t fmFuncScanType(int32_t funcId);
 

include/libs/nodes/nodes.h

@@ -70,6 +70,7 @@ typedef enum ENodeType {
   QUERY_NODE_DATABASE_OPTIONS,
   QUERY_NODE_TABLE_OPTIONS,
   QUERY_NODE_INDEX_OPTIONS,
+  QUERY_NODE_EXPLAIN_OPTIONS,
 
   // Statement nodes are used in parser and planner module.
   QUERY_NODE_SET_OPERATOR,
@@ -99,6 +100,7 @@ typedef enum ENodeType {
   QUERY_NODE_CREATE_TOPIC_STMT,
   QUERY_NODE_DROP_TOPIC_STMT,
   QUERY_NODE_ALTER_LOCAL_STMT,
+  QUERY_NODE_EXPLAIN_STMT,
   QUERY_NODE_SHOW_DATABASES_STMT,
   QUERY_NODE_SHOW_TABLES_STMT,
   QUERY_NODE_SHOW_STABLES_STMT,
@@ -121,6 +123,7 @@ typedef enum ENodeType {
   QUERY_NODE_LOGIC_PLAN_EXCHANGE,
   QUERY_NODE_LOGIC_PLAN_WINDOW,
   QUERY_NODE_LOGIC_PLAN_SORT,
+  QUERY_NODE_LOGIC_PLAN_PARTITION,
   QUERY_NODE_LOGIC_SUBPLAN,
   QUERY_NODE_LOGIC_PLAN,
 

include/libs/nodes/plannodes.h

@@ -49,6 +49,7 @@ typedef struct SScanLogicNode {
   STimeWindow scanRange;
   SName tableName;
   bool showRewrite;
+  double ratio;
 } SScanLogicNode;
 
 typedef struct SJoinLogicNode {
@@ -66,7 +67,11 @@ typedef struct SAggLogicNode {
 typedef struct SProjectLogicNode {
   SLogicNode node;
   SNodeList* pProjections;
-  char stmtName[TSDB_TABLE_NAME_LEN]; 
+  char stmtName[TSDB_TABLE_NAME_LEN];
+  int64_t limit;
+  int64_t offset;
+  int64_t slimit;
+  int64_t soffset;
 } SProjectLogicNode;
 
 typedef struct SVnodeModifLogicNode {
@@ -106,6 +111,11 @@ typedef struct SSortLogicNode {
   SNodeList* pSortKeys;
 } SSortLogicNode;
 
+typedef struct SPartitionLogicNode {
+  SLogicNode node;
+  SNodeList* pPartitionKeys;
+} SPartitionLogicNode;
+
 typedef enum ESubplanType {
   SUBPLAN_TYPE_MERGE = 1,
   SUBPLAN_TYPE_PARTIAL,
@@ -150,7 +160,8 @@ typedef struct SDataBlockDescNode {
   ENodeType type;
   int16_t dataBlockId;
   SNodeList* pSlots;
-  int32_t resultRowSize;
+  int32_t totalRowSize;
+  int32_t outputRowSize;
   int16_t precision;
 } SDataBlockDescNode;
 
@@ -187,7 +198,7 @@ typedef struct STableScanPhysiNode {
   SScanPhysiNode scan;
   uint8_t scanFlag;         // denotes reversed scan of data or not
   STimeWindow scanRange;
-  SNode* pScanConditions;
+  double ratio;
 } STableScanPhysiNode;
 
 typedef STableScanPhysiNode STableSeqScanPhysiNode;
@@ -195,6 +206,10 @@ typedef STableScanPhysiNode STableSeqScanPhysiNode;
 typedef struct SProjectPhysiNode {
   SPhysiNode node;
   SNodeList* pProjections;
+  int64_t limit;
+  int64_t offset;
+  int64_t slimit;
+  int64_t soffset;
 } SProjectPhysiNode;
 
 typedef struct SJoinPhysiNode {
@@ -238,6 +253,7 @@ typedef struct SIntervalPhysiNode {
   int64_t    sliding;
   int8_t     intervalUnit;
   int8_t     slidingUnit;
+  uint8_t precision;
   SFillNode* pFill;
 } SIntervalPhysiNode;
 
@@ -283,11 +299,23 @@ typedef struct SSubplan {
   SDataSinkNode* pDataSink;    // data of the subplan flow into the datasink
 } SSubplan;
 
+typedef enum EExplainMode {
+  EXPLAIN_MODE_DISABLE = 1,
+  EXPLAIN_MODE_STATIC,
+  EXPLAIN_MODE_ANALYZE
+} EExplainMode;
+
+typedef struct SExplainInfo {
+  EExplainMode mode;
+  bool verbose;
+} SExplainInfo;
+
 typedef struct SQueryPlan {
   ENodeType type;
   uint64_t queryId;
   int32_t numOfSubplans;
   SNodeList* pSubplans; // Element is SNodeListNode. The execution level of subplan, starting from 0.
+  SExplainInfo explainInfo;
 } SQueryPlan;
 
 #ifdef __cplusplus

include/libs/nodes/querynodes.h

@@ -131,6 +131,7 @@ typedef struct SRealTableNode {
   struct STableMeta* pMeta;
   SVgroupsInfo* pVgroupList;
   char useDbName[TSDB_DB_NAME_LEN];
+  double ratio;
 } SRealTableNode;
 
 typedef struct STempTableNode {
@@ -180,8 +181,8 @@ typedef struct SOrderByExprNode {
 
 typedef struct SLimitNode {
   ENodeType type; // QUERY_NODE_LIMIT
-  uint64_t limit;
+  int64_t limit;
-  uint64_t offset;
+  int64_t offset;
 } SLimitNode;
 
 typedef struct SStateWindowNode {
@@ -282,6 +283,19 @@ typedef struct SVnodeModifOpStmt {
   const char* sql;                 // current sql statement position
 } SVnodeModifOpStmt;
 
+typedef struct SExplainOptions {
+  ENodeType type;
+  bool verbose;
+  double ratio;
+} SExplainOptions;
+
+typedef struct SExplainStmt {
+  ENodeType type;
+  bool analyze;
+  SExplainOptions* pOptions;
+  SNode* pQuery;
+} SExplainStmt;
+
 void nodesWalkSelectStmt(SSelectStmt* pSelect, ESqlClause clause, FNodeWalker walker, void* pContext);
 void nodesRewriteSelectStmt(SSelectStmt* pSelect, ESqlClause clause, FNodeRewriter rewriter, void* pContext);
 

include/libs/scalar/scalar.h

@@ -58,6 +58,14 @@ int32_t ceilFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
 int32_t floorFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 int32_t roundFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
 
+bool getTimePseudoFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
+
+int32_t winStartTsFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t winEndTsFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t winDurFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t qStartTsFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+int32_t qEndTsFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput);
+
 #ifdef __cplusplus
 }
 #endif

include/libs/stream/tstream.h

@@ -30,6 +30,11 @@ enum {
   STREAM_TASK_STATUS__STOP,
 };
 
+enum {
+  STREAM_CREATED_BY__USER = 1,
+  STREAM_CREATED_BY__SMA,
+};
+
 #if 0
 // pipe  -> fetch/pipe queue
 // merge -> merge      queue

include/libs/transport/trpc.h

@@ -49,11 +49,13 @@ typedef struct SRpcMsg {
 } SRpcMsg;
 
 typedef struct {
-  char    user[TSDB_USER_LEN];
+  char     user[TSDB_USER_LEN];
-  SRpcMsg rpcMsg;
+  uint32_t clientIp;
-  int32_t rspLen;
+  uint16_t clientPort;
-  void *  pRsp;
+  SRpcMsg  rpcMsg;
-  void *  pNode;
+  int32_t  rspLen;
+  void    *pRsp;
+  void    *pNode;
 } SNodeMsg;
 
 typedef struct SRpcInit {
@@ -105,20 +107,20 @@ void    rpcClose(void *);
 void *  rpcMallocCont(int contLen);
 void    rpcFreeCont(void *pCont);
 void *  rpcReallocCont(void *ptr, int contLen);
-void    rpcSendRequest(void *thandle, const SEpSet *pEpSet, SRpcMsg *pMsg, int64_t *rid);
-void    rpcSendRequestWithCtx(void *thandle, const SEpSet *pEpSet, SRpcMsg *pMsg, int64_t *rid, SRpcCtx *ctx);
 
+// Because taosd supports multi-process mode
+// These functions should not be used on the server side
+// Please use tmsg<xx> functions, which are defined in tmsgcb.h
+void rpcSendRequest(void *thandle, const SEpSet *pEpSet, SRpcMsg *pMsg, int64_t *rid);
 void rpcSendResponse(const SRpcMsg *pMsg);
+void rpcRegisterBrokenLinkArg(SRpcMsg *msg);
+void rpcReleaseHandle(void *handle, int8_t type);  // just release client conn to rpc instance, no close sock
+
+// These functions will not be called in the child process
 void rpcSendRedirectRsp(void *pConn, const SEpSet *pEpSet);
+void rpcSendRequestWithCtx(void *thandle, const SEpSet *pEpSet, SRpcMsg *pMsg, int64_t *rid, SRpcCtx *ctx);
 int  rpcGetConnInfo(void *thandle, SRpcConnInfo *pInfo);
 void rpcSendRecv(void *shandle, SEpSet *pEpSet, SRpcMsg *pReq, SRpcMsg *pRsp);
-int  rpcReportProgress(void *pConn, char *pCont, int contLen);
-void rpcCancelRequest(int64_t rid);
-void rpcRegisterBrokenLinkArg(SRpcMsg *msg);
-// just release client conn to rpc instance, no close sock
-void rpcReleaseHandle(void *handle, int8_t type);  //
-void rpcRefHandle(void *handle, int8_t type);
-void rpcUnrefHandle(void *handle, int8_t type);
 
 #ifdef __cplusplus
 }

include/os/os.h

@@ -42,6 +42,9 @@ extern "C" {
 #include <sys/types.h>
 #include <termios.h>
 #include <sys/statvfs.h>
+#include <sys/prctl.h>
+#include <sys/shm.h>
+#include <sys/wait.h>
 
 #if defined(DARWIN)
 #else
@@ -70,7 +73,6 @@ extern "C" {
 #include <wchar.h>
 #include <wctype.h>
 
-
 #include "osAtomic.h"
 #include "osDef.h"
 #include "osDir.h"
@@ -84,6 +86,7 @@ extern "C" {
 #include "osThread.h"
 #include "osSemaphore.h"
 #include "osSignal.h"
+#include "osShm.h"
 #include "osSleep.h"
 #include "osSocket.h"
 #include "osString.h"

include/os/osShm.h

@@ -12,21 +12,27 @@
  * You should have received a copy of the GNU Affero General Public License
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
-#ifndef TDENGINE_TSCALARFUNCTION_H
+
-#define TDENGINE_TSCALARFUNCTION_H
+#ifndef _TD_OS_SHM_H_
+#define _TD_OS_SHM_H_
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-#include "function.h"
+typedef struct {
-#include "scalar.h"
+  int32_t id;
-
+  int32_t size;
-
+  void*   ptr;
+} SShm;
 
+int32_t taosCreateShm(SShm *pShm, int32_t shmsize) ;
+void    taosDropShm(SShm *pShm);
+int32_t taosAttachShm(SShm *pShm);
+void    taosDetachShm(SShm *pShm);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif  // TDENGINE_TSCALARFUNCTION_H
+#endif /*_TD_OS_SHM_H_*/

include/util/taoserror.h

@@ -16,7 +16,7 @@
 #ifndef _TD_UTIL_TAOS_ERROR_H_
 #define _TD_UTIL_TAOS_ERROR_H_
 
-#include "os.h"
+#include <stdint.h>
 
 #ifdef __cplusplus
 extern "C" {

include/util/tdef.h

@@ -381,6 +381,14 @@ typedef enum ELogicConditionType {
 #define TSDB_MAX_DB_DELAY     10
 #define TSDB_DEFAULT_DB_DELAY 2
 
+#define TSDB_DEFAULT_EXPLAIN_VERBOSE false
+
+#define TSDB_MIN_EXPLAIN_RATIO       0
+#define TSDB_MAX_EXPLAIN_RATIO       1
+#define TSDB_DEFAULT_EXPLAIN_RATIO   0.001
+
+#define TSDB_EXPLAIN_RESULT_ROW_SIZE 1024
+
 #define TSDB_MAX_JOIN_TABLE_NUM 10
 #define TSDB_MAX_UNION_CLAUSE   5
 

include/util/tprocess.h

@@ -22,11 +22,14 @@
 extern "C" {
 #endif
 
+typedef enum { PROC_REQ, PROC_RSP, PROC_REG, PROC_RELEASE } ProcFuncType;
+
 typedef struct SProcQueue SProcQueue;
 typedef struct SProcObj   SProcObj;
 typedef void *(*ProcMallocFp)(int32_t contLen);
 typedef void *(*ProcFreeFp)(void *pCont);
-typedef void *(*ProcConsumeFp)(void *pParent, void *pHead, int32_t headLen, void *pBody, int32_t bodyLen);
+typedef void *(*ProcConsumeFp)(void *pParent, void *pHead, int16_t headLen, void *pBody, int32_t bodyLen,
+                               ProcFuncType ftype);
 
 typedef struct {
   int32_t       childQueueSize;
@@ -52,9 +55,10 @@ int32_t   taosProcRun(SProcObj *pProc);
 void      taosProcStop(SProcObj *pProc);
 bool      taosProcIsChild(SProcObj *pProc);
 int32_t   taosProcChildId(SProcObj *pProc);
-
+int32_t   taosProcPutToChildQ(SProcObj *pProc, const void *pHead, int16_t headLen, const void *pBody, int32_t bodyLen,
-int32_t taosProcPutToChildQueue(SProcObj *pProc, void *pHead, int32_t headLen, void *pBody, int32_t bodyLen);
+                              ProcFuncType ftype);
-int32_t taosProcPutToParentQueue(SProcObj *pProc, void *pHead, int32_t headLen, void *pBody, int32_t bodyLen);
+int32_t   taosProcPutToParentQ(SProcObj *pProc, const void *pHead, int16_t headLen, const void *pBody, int32_t bodyLen,
+                               ProcFuncType ftype);
 
 #ifdef __cplusplus
 }

source/client/consumer/consumer.c

@@ -1,16 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#include "consumer.h"

source/client/src/clientEnv.c

@@ -200,6 +200,9 @@ static void doDestroyRequest(void *p) {
     taosArrayDestroy(pRequest->body.showInfo.pArray);
   }
 
+  taosArrayDestroy(pRequest->tableList);
+  taosArrayDestroy(pRequest->dbList);
+
   deregisterRequest(pRequest);
   taosMemoryFreeClear(pRequest);
 }

source/client/src/clientMain.c

@@ -1,12 +1,27 @@
-#include "os.h"
+/*
-#include "tref.h"
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
-#include "trpc.h"
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "catalog.h"
 #include "clientInt.h"
 #include "clientLog.h"
+#include "os.h"
 #include "query.h"
-#include "tmsg.h"
 #include "tglobal.h"
-#include "catalog.h"
+#include "tmsg.h"
+#include "tref.h"
+#include "trpc.h"
 #include "version.h"
 
 #define TSC_VAR_NOT_RELEASE 1
@@ -24,7 +39,7 @@ int taos_options(TSDB_OPTION option, const void *arg, ...) {
     }
   }
 
-  int ret = taos_options_imp(option, (const char*)arg);
+  int ret = taos_options_imp(option, (const char *)arg);
   atomic_store_32(&lock, 0);
   return ret;
 }
@@ -56,6 +71,12 @@ void taos_cleanup(void) {
   tscInfo("all local resources released");
 }
 
+setConfRet   taos_set_config(const char *config) {
+  // TODO
+  setConfRet ret = {SET_CONF_RET_SUCC, {0}};
+  return ret;
+}
+
 TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port) {
   tscDebug("try to connect to %s:%u, user:%s db:%s", ip, port, user, db);
   if (user == NULL) {
@@ -69,13 +90,13 @@ TAOS *taos_connect(const char *ip, const char *user, const char *pass, const cha
   return taos_connect_internal(ip, user, pass, NULL, db, port);
 }
 
-void taos_close(TAOS* taos) {
+void taos_close(TAOS *taos) {
   if (taos == NULL) {
     return;
   }
 
   STscObj *pTscObj = (STscObj *)taos;
-  tscDebug("0x%"PRIx64" try to close connection, numOfReq:%d", pTscObj->id, pTscObj->numOfReqs);
+  tscDebug("0x%" PRIx64 " try to close connection, numOfReq:%d", pTscObj->id, pTscObj->numOfReqs);
 
   /*taosRemoveRef(clientConnRefPool, pTscObj->id);*/
 }
@@ -85,48 +106,46 @@ int taos_errno(TAOS_RES *tres) {
     return terrno;
   }
 
-  return ((SRequestObj*) tres)->code;
+  return ((SRequestObj *)tres)->code;
 }
 
 const char *taos_errstr(TAOS_RES *res) {
-  SRequestObj *pRequest = (SRequestObj *) res;
+  SRequestObj *pRequest = (SRequestObj *)res;
 
   if (pRequest == NULL) {
-    return (const char*) tstrerror(terrno);
+    return (const char *)tstrerror(terrno);
   }
 
   if (strlen(pRequest->msgBuf) > 0 || pRequest->code == TSDB_CODE_RPC_FQDN_ERROR) {
     return pRequest->msgBuf;
   } else {
-    return (const char*)tstrerror(pRequest->code);
+    return (const char *)tstrerror(pRequest->code);
   }
 }
 
 void taos_free_result(TAOS_RES *res) {
-  SRequestObj* pRequest = (SRequestObj*) res;
+  SRequestObj *pRequest = (SRequestObj *)res;
   destroyRequest(pRequest);
 }
 
-int  taos_field_count(TAOS_RES *res) {
+int taos_field_count(TAOS_RES *res) {
   if (res == NULL) {
     return 0;
   }
 
-  SRequestObj* pRequest = (SRequestObj*) res;
+  SRequestObj    *pRequest = (SRequestObj *)res;
-  SReqResultInfo* pResInfo = &pRequest->body.resInfo;
+  SReqResultInfo *pResInfo = &pRequest->body.resInfo;
   return pResInfo->numOfCols;
 }
 
-int  taos_num_fields(TAOS_RES *res) {
+int taos_num_fields(TAOS_RES *res) { return taos_field_count(res); }
-  return taos_field_count(res);
-}
 
 TAOS_FIELD *taos_fetch_fields(TAOS_RES *res) {
   if (taos_num_fields(res) == 0) {
     return NULL;
   }
 
-  SReqResultInfo* pResInfo = &(((SRequestObj*) res)->body.resInfo);
+  SReqResultInfo *pResInfo = &(((SRequestObj *)res)->body.resInfo);
   return pResInfo->fields;
 }
 
@@ -135,7 +154,7 @@ TAOS_RES *taos_query(TAOS *taos, const char *sql) {
     return NULL;
   }
 
-  return taos_query_l(taos, sql, (int32_t) strlen(sql));
+  return taos_query_l(taos, sql, (int32_t)strlen(sql));
 }
 
 TAOS_ROW taos_fetch_row(TAOS_RES *res) {
@@ -143,18 +162,16 @@ TAOS_ROW taos_fetch_row(TAOS_RES *res) {
     return NULL;
   }
 
-  SRequestObj *pRequest = (SRequestObj *) res;
+  SRequestObj *pRequest = (SRequestObj *)res;
-  if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT ||
+  if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT || pRequest->type == TSDB_SQL_INSERT ||
-      pRequest->type == TSDB_SQL_INSERT ||
+      pRequest->code != TSDB_CODE_SUCCESS || taos_num_fields(res) == 0) {
-      pRequest->code != TSDB_CODE_SUCCESS ||
-      taos_num_fields(res) == 0) {
     return NULL;
   }
 
   return doFetchRow(pRequest, true);
 }
 
-int  taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields) {
+int taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields) {
   int32_t len = 0;
   for (int i = 0; i < num_fields; ++i) {
     if (i > 0) {
@@ -213,7 +230,7 @@ int  taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields)
 
       case TSDB_DATA_TYPE_BINARY:
       case TSDB_DATA_TYPE_NCHAR: {
-        int32_t charLen = varDataLen((char*)row[i] - VARSTR_HEADER_SIZE);
+        int32_t charLen = varDataLen((char *)row[i] - VARSTR_HEADER_SIZE);
         if (fields[i].type == TSDB_DATA_TYPE_BINARY) {
           assert(charLen <= fields[i].bytes && charLen >= 0);
         } else {
@@ -238,31 +255,49 @@ int  taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields)
   return len;
 }
 
-int* taos_fetch_lengths(TAOS_RES *res) {
+int *taos_fetch_lengths(TAOS_RES *res) {
   if (res == NULL) {
     return NULL;
   }
 
-  return ((SRequestObj*) res)->body.resInfo.length;
+  return ((SRequestObj *)res)->body.resInfo.length;
+}
+
+TAOS_ROW *taos_result_block(TAOS_RES *res) {
+  // TODO
+  return NULL;
 }
 
 // todo intergrate with tDataTypes
 const char *taos_data_type(int type) {
   switch (type) {
-    case TSDB_DATA_TYPE_NULL:            return "TSDB_DATA_TYPE_NULL";
+    case TSDB_DATA_TYPE_NULL:
-    case TSDB_DATA_TYPE_BOOL:            return "TSDB_DATA_TYPE_BOOL";
+      return "TSDB_DATA_TYPE_NULL";
-    case TSDB_DATA_TYPE_TINYINT:         return "TSDB_DATA_TYPE_TINYINT";
+    case TSDB_DATA_TYPE_BOOL:
-    case TSDB_DATA_TYPE_SMALLINT:        return "TSDB_DATA_TYPE_SMALLINT";
+      return "TSDB_DATA_TYPE_BOOL";
-    case TSDB_DATA_TYPE_INT:             return "TSDB_DATA_TYPE_INT";
+    case TSDB_DATA_TYPE_TINYINT:
-    case TSDB_DATA_TYPE_BIGINT:          return "TSDB_DATA_TYPE_BIGINT";
+      return "TSDB_DATA_TYPE_TINYINT";
-    case TSDB_DATA_TYPE_FLOAT:           return "TSDB_DATA_TYPE_FLOAT";
+    case TSDB_DATA_TYPE_SMALLINT:
-    case TSDB_DATA_TYPE_DOUBLE:          return "TSDB_DATA_TYPE_DOUBLE";
+      return "TSDB_DATA_TYPE_SMALLINT";
-    case TSDB_DATA_TYPE_VARCHAR:         return "TSDB_DATA_TYPE_VARCHAR";
+    case TSDB_DATA_TYPE_INT:
-//    case TSDB_DATA_TYPE_BINARY:          return "TSDB_DATA_TYPE_VARCHAR";
+      return "TSDB_DATA_TYPE_INT";
-    case TSDB_DATA_TYPE_TIMESTAMP:       return "TSDB_DATA_TYPE_TIMESTAMP";
+    case TSDB_DATA_TYPE_BIGINT:
-    case TSDB_DATA_TYPE_NCHAR:           return "TSDB_DATA_TYPE_NCHAR";
+      return "TSDB_DATA_TYPE_BIGINT";
-    case TSDB_DATA_TYPE_JSON:            return "TSDB_DATA_TYPE_JSON";
+    case TSDB_DATA_TYPE_FLOAT:
-    default: return "UNKNOWN";
+      return "TSDB_DATA_TYPE_FLOAT";
+    case TSDB_DATA_TYPE_DOUBLE:
+      return "TSDB_DATA_TYPE_DOUBLE";
+    case TSDB_DATA_TYPE_VARCHAR:
+      return "TSDB_DATA_TYPE_VARCHAR";
+      //    case TSDB_DATA_TYPE_BINARY:          return "TSDB_DATA_TYPE_VARCHAR";
+    case TSDB_DATA_TYPE_TIMESTAMP:
+      return "TSDB_DATA_TYPE_TIMESTAMP";
+    case TSDB_DATA_TYPE_NCHAR:
+      return "TSDB_DATA_TYPE_NCHAR";
+    case TSDB_DATA_TYPE_JSON:
+      return "TSDB_DATA_TYPE_JSON";
+    default:
+      return "UNKNOWN";
   }
 }
 
@@ -273,8 +308,8 @@ int taos_affected_rows(TAOS_RES *res) {
     return 0;
   }
 
-  SRequestObj* pRequest = (SRequestObj*) res;
+  SRequestObj    *pRequest = (SRequestObj *)res;
-  SReqResultInfo* pResInfo = &pRequest->body.resInfo;
+  SReqResultInfo *pResInfo = &pRequest->body.resInfo;
   return pResInfo->numOfRows;
 }
 
@@ -295,8 +330,8 @@ int taos_select_db(TAOS *taos, const char *db) {
   char sql[256] = {0};
   snprintf(sql, tListLen(sql), "use %s", db);
 
-  TAOS_RES* pRequest = taos_query(taos, sql);
+  TAOS_RES *pRequest = taos_query(taos, sql);
-  int32_t code = taos_errno(pRequest);
+  int32_t   code = taos_errno(pRequest);
 
   taos_free_result(pRequest);
   return code;
@@ -307,53 +342,59 @@ void taos_stop_query(TAOS_RES *res) {
     return;
   }
 
-  SRequestObj* pRequest = (SRequestObj*) res;
+  SRequestObj *pRequest = (SRequestObj *)res;
-  int32_t numOfFields = taos_num_fields(pRequest);
+  int32_t      numOfFields = taos_num_fields(pRequest);
 
   // It is not a query, no need to stop.
   if (numOfFields == 0) {
     return;
   }
 
-//  scheduleCancelJob(pRequest->body.pQueryJob);
+  //  scheduleCancelJob(pRequest->body.pQueryJob);
 }
 
 bool taos_is_null(TAOS_RES *res, int32_t row, int32_t col) {
-  SRequestObj* pRequestObj = res;
+  SRequestObj    *pRequestObj = res;
-  SReqResultInfo* pResultInfo = &pRequestObj->body.resInfo;
+  SReqResultInfo *pResultInfo = &pRequestObj->body.resInfo;
   if (col >= pResultInfo->numOfCols || col < 0 || row >= pResultInfo->numOfRows || row < 0) {
     return true;
   }
 
-  SResultColumn* pCol = &pRequestObj->body.resInfo.pCol[col];
+  SResultColumn *pCol = &pRequestObj->body.resInfo.pCol[col];
   return colDataIsNull_f(pCol->nullbitmap, row);
 }
 
-int  taos_fetch_block(TAOS_RES *res, TAOS_ROW *rows) {
+bool taos_is_update_query(TAOS_RES *res) {
+  // TODO
+  return true;
+}
+
+int taos_fetch_block(TAOS_RES *res, TAOS_ROW *rows) {
   if (res == NULL) {
     return 0;
   }
 
-  SRequestObj *pRequest = (SRequestObj *) res;
+  SRequestObj *pRequest = (SRequestObj *)res;
-  if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT ||
+  if (pRequest->type == TSDB_SQL_RETRIEVE_EMPTY_RESULT || pRequest->type == TSDB_SQL_INSERT ||
-      pRequest->type == TSDB_SQL_INSERT ||
+      pRequest->code != TSDB_CODE_SUCCESS || taos_num_fields(res) == 0) {
-      pRequest->code != TSDB_CODE_SUCCESS ||
-      taos_num_fields(res) == 0) {
     return 0;
   }
 
   doFetchRow(pRequest, false);
 
   // TODO refactor
-  SReqResultInfo* pResultInfo = &pRequest->body.resInfo;
+  SReqResultInfo *pResultInfo = &pRequest->body.resInfo;
   pResultInfo->current = pResultInfo->numOfRows;
   *rows = pResultInfo->row;
 
   return pResultInfo->numOfRows;
 }
 
-int  taos_validate_sql(TAOS *taos, const char *sql) {
+int taos_validate_sql(TAOS *taos, const char *sql) { return true; }
-  return true;
+
+void taos_reset_current_db(TAOS *taos) {
+  // TODO
+  return;
 }
 
 const char *taos_get_server_info(TAOS *taos) {
@@ -361,7 +402,7 @@ const char *taos_get_server_info(TAOS *taos) {
     return NULL;
   }
 
-  STscObj* pTscObj = (STscObj*) taos;
+  STscObj *pTscObj = (STscObj *)taos;
   return pTscObj->ver;
 }
 
@@ -387,52 +428,67 @@ void taos_unsubscribe(TAOS_SUB *tsub, int keepProgress) {
     // TODO
 }
 
-TAOS_STMT* taos_stmt_init(TAOS* taos) {
+int taos_load_table_info(TAOS *taos, const char *tableNameList) {
-    // TODO
+  // TODO
-    return NULL;
+  return -1;
 }
 
-int taos_stmt_close(TAOS_STMT* stmt) {
+TAOS_STMT *taos_stmt_init(TAOS *taos) {
-    // TODO
+  // TODO
-    return -1;
+  return NULL;
 }
 
-int taos_stmt_execute(TAOS_STMT* stmt) {
+int taos_stmt_close(TAOS_STMT *stmt) {
-    // TODO
+  // TODO
-    return -1;
+  return -1;
+}
+
+int taos_stmt_execute(TAOS_STMT *stmt) {
+  // TODO
+  return -1;
 }
 
 char *taos_stmt_errstr(TAOS_STMT *stmt) {
-    // TODO
+  // TODO
-    return NULL;
+  return NULL;
 }
 
-int taos_stmt_affected_rows(TAOS_STMT* stmt) {
+int taos_stmt_affected_rows(TAOS_STMT *stmt) {
+  // TODO
+  return -1;
+}
+
+TAOS_RES *taos_schemaless_insert(TAOS *taos, char *lines[], int numLines, int protocol, int precision) {
+  // TODO
+  return NULL;
+}
+
+int taos_stmt_bind_param(TAOS_STMT *stmt, TAOS_BIND *bind) {
+  // TODO
+  return -1;
+}
+
+int taos_stmt_prepare(TAOS_STMT *stmt, const char *sql, unsigned long length) {
+  // TODO
+  return -1;
+}
+
+int taos_stmt_set_tbname_tags(TAOS_STMT *stmt, const char *name, TAOS_BIND *tags) {
+  // TODO
+  return -1;
+}
+
+int taos_stmt_set_tbname(TAOS_STMT *stmt, const char *name) {
+  // TODO
+  return -1;
+}
+
+int taos_stmt_is_insert(TAOS_STMT *stmt, int *insert) {
     // TODO
     return -1;
 }
 
-TAOS_RES* taos_schemaless_insert(TAOS* taos, char* lines[], int numLines, int protocol, int precision) {
+int taos_stmt_num_params(TAOS_STMT *stmt, int *nums) {
-    // TODO
-    return NULL;
-}
-
-int taos_stmt_bind_param(TAOS_STMT* stmt, TAOS_BIND* bind) {
-    // TODO
-    return -1;
-}
-
-int taos_stmt_prepare(TAOS_STMT* stmt, const char* sql, unsigned long length) {
-    // TODO
-    return -1;
-}
-
-int taos_stmt_set_tbname_tags(TAOS_STMT* stmt, const char* name, TAOS_BIND* tags) {
-    // TODO
-    return -1;
-}
-
-int taos_stmt_set_tbname(TAOS_STMT* stmt, const char* name) {
     // TODO
     return -1;
 }
@@ -442,6 +498,11 @@ int taos_stmt_add_batch(TAOS_STMT* stmt) {
     return -1;
 }
 
+TAOS_RES *taos_stmt_use_result(TAOS_STMT *stmt) {
+    // TODO
+    return NULL;
+}
+
 int taos_stmt_bind_param_batch(TAOS_STMT* stmt, TAOS_MULTI_BIND* bind) {
     // TODO
     return -1;

source/client/src/tmq.c

@@ -26,6 +26,18 @@
 #include "tqueue.h"
 #include "tref.h"
 
+typedef struct {
+  int32_t curBlock;
+  int32_t curRow;
+  void**  uData;
+} SMqRowIter;
+
+struct tmq_message_t {
+  SMqPollRsp msg;
+  void*      vg;
+  SMqRowIter iter;
+};
+
 struct tmq_list_t {
   SArray container;
 };
@@ -99,13 +111,14 @@ typedef struct {
 
 typedef struct {
   // subscribe info
-  int32_t        sqlLen;
+  int32_t     sqlLen;
-  char*          sql;
+  char*       sql;
-  char*          topicName;
+  char*       topicName;
-  int64_t        topicId;
+  int64_t     topicId;
-  int32_t        nextVgIdx;
+  SArray*     vgs;  // SArray<SMqClientVg>
-  SArray*        vgs;  // SArray<SMqClientVg>
+  int8_t      isSchemaAdaptive;
-  SSchemaWrapper schema;
+  int32_t     numOfFields;
+  TAOS_FIELD* fields;
 } SMqClientTopic;
 
 typedef struct {
@@ -130,11 +143,11 @@ typedef struct {
 } SMqPollCbParam;
 
 typedef struct {
-  tmq_t* tmq;
+  tmq_t*         tmq;
-  /*SMqClientVg* pVg;*/
   int32_t        async;
   tsem_t         rspSem;
   tmq_resp_err_t rspErr;
+  /*SMqClientVg* pVg;*/
 } SMqCommitCbParam;
 
 tmq_conf_t* tmq_conf_new() {
@@ -228,6 +241,10 @@ void tmq_list_destroy(tmq_list_t* list) {
   taosArrayDestroyEx(container, (void (*)(void*))taosMemoryFree);
 }
 
+static int32_t tmqMakeTopicVgKey(char* dst, const char* topicName, int32_t vg) {
+  return sprintf(dst, "%s:%d", topicName, vg);
+}
+
 void tmqClearUnhandleMsg(tmq_t* tmq) {
   tmq_message_t* msg = NULL;
   while (1) {
@@ -471,7 +488,12 @@ tmq_resp_err_t tmq_subscribe(tmq_t* tmq, tmq_list_t* topic_list) {
     tNameExtractFullName(&name, topicFname);
     tscDebug("subscribe topic: %s", topicFname);
     SMqClientTopic topic = {
-        .nextVgIdx = 0, .sql = NULL, .sqlLen = 0, .topicId = 0, .topicName = topicFname, .vgs = NULL};
+        .sql = NULL,
+        .sqlLen = 0,
+        .topicId = 0,
+        .topicName = topicFname,
+        .vgs = NULL,
+    };
     topic.vgs = taosArrayInit(0, sizeof(SMqClientVg));
     taosArrayPush(tmq->clientTopics, &topic);
     taosArrayPush(req.topicNames, &topicFname);
@@ -615,6 +637,7 @@ _return:
   return pRequest;
 }
 
+#if 0
 TAOS_RES* tmq_create_topic(TAOS* taos, const char* topicName, const char* sql, int sqlLen) {
   STscObj*     pTscObj = (STscObj*)taos;
   SRequestObj* pRequest = NULL;
@@ -700,6 +723,7 @@ _return:
 
   return pRequest;
 }
+#endif
 
 static char* formatTimestamp(char* buf, int64_t val, int precision) {
   time_t  tt;
@@ -807,7 +831,7 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) {
   SMqClientVg*    pVg = pParam->pVg;
   tmq_t*          tmq = pParam->tmq;
   if (code != 0) {
-    printf("msg discard, code:%x\n", code);
+    tscWarn("msg discard, code:%x", code);
     goto WRITE_QUEUE_FAIL;
   }
 
@@ -815,12 +839,12 @@ int32_t tmqPollCb(void* param, const SDataBuf* pMsg, int32_t code) {
   int32_t tmqEpoch = atomic_load_32(&tmq->epoch);
   if (msgEpoch < tmqEpoch) {
     tsem_post(&tmq->rspSem);
-    printf("discard rsp epoch %d, current epoch %d\n", msgEpoch, tmqEpoch);
+    tscWarn("discard rsp epoch %d, current epoch %d", msgEpoch, tmqEpoch);
     return 0;
   }
 
   if (msgEpoch != tmqEpoch) {
-    printf("mismatch rsp epoch %d, current epoch %d\n", msgEpoch, tmqEpoch);
+    tscWarn("mismatch rsp epoch %d, current epoch %d", msgEpoch, tmqEpoch);
   } else {
     atomic_sub_fetch_32(&tmq->waitingRequest, 1);
   }
@@ -879,19 +903,54 @@ WRITE_QUEUE_FAIL:
 bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
   /*printf("call update ep %d\n", epoch);*/
   bool    set = false;
-  int32_t sz = taosArrayGetSize(pRsp->topics);
+  int32_t topicNumGet = taosArrayGetSize(pRsp->topics);
-  SArray* newTopics = taosArrayInit(sz, sizeof(SMqClientTopic));
+  char    vgKey[TSDB_TOPIC_FNAME_LEN + 22];
-  for (int32_t i = 0; i < sz; i++) {
+  SArray* newTopics = taosArrayInit(topicNumGet, sizeof(SMqClientTopic));
+  if (newTopics == NULL) {
+    return false;
+  }
+  SHashObj* pHash = taosHashInit(64, MurmurHash3_32, false, HASH_NO_LOCK);
+  if (pHash == NULL) {
+    taosArrayDestroy(newTopics);
+    return false;
+  }
+
+  // find topic, build hash
+  for (int32_t i = 0; i < topicNumGet; i++) {
     SMqClientTopic topic = {0};
     SMqSubTopicEp* pTopicEp = taosArrayGet(pRsp->topics, i);
+    taosHashClear(pHash);
     topic.topicName = strdup(pTopicEp->topic);
-    int32_t vgSz = taosArrayGetSize(pTopicEp->vgs);
+
-    topic.vgs = taosArrayInit(vgSz, sizeof(SMqClientVg));
+    int32_t topicNumCur = taosArrayGetSize(tmq->clientTopics);
-    for (int32_t j = 0; j < vgSz; j++) {
+    for (int32_t j = 0; j < topicNumCur; j++) {
+      // find old topic
+      SMqClientTopic* pTopicCur = taosArrayGet(tmq->clientTopics, j);
+      if (pTopicCur->vgs && strcmp(pTopicCur->topicName, pTopicEp->topic) == 0) {
+        int32_t vgNumCur = taosArrayGetSize(pTopicCur->vgs);
+        if (vgNumCur == 0) break;
+        for (int32_t k = 0; k < vgNumCur; k++) {
+          SMqClientVg* pVgCur = taosArrayGet(pTopicCur->vgs, k);
+          sprintf(vgKey, "%s:%d", topic.topicName, pVgCur->vgId);
+          taosHashPut(pHash, vgKey, strlen(vgKey), &pVgCur->currentOffset, sizeof(int64_t));
+        }
+        break;
+      }
+    }
+
+    int32_t vgNumGet = taosArrayGetSize(pTopicEp->vgs);
+    topic.vgs = taosArrayInit(vgNumGet, sizeof(SMqClientVg));
+    for (int32_t j = 0; j < vgNumGet; j++) {
       SMqSubVgEp* pVgEp = taosArrayGet(pTopicEp->vgs, j);
+      sprintf(vgKey, "%s:%d", topic.topicName, pVgEp->vgId);
+      int64_t* pOffset = taosHashGet(pHash, vgKey, strlen(vgKey));
+      int64_t  offset = pVgEp->offset;
+      if (pOffset != NULL) {
+        offset = *pOffset;
+      }
       SMqClientVg clientVg = {
           .pollCnt = 0,
-          .currentOffset = pVgEp->offset,
+          .currentOffset = offset,
           .vgId = pVgEp->vgId,
           .epSet = pVgEp->epSet,
           .vgStatus = TMQ_VG_STATUS__IDLE,
@@ -902,6 +961,7 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqCMGetSubEpRsp* pRsp) {
     taosArrayPush(newTopics, &topic);
   }
   if (tmq->clientTopics) taosArrayDestroy(tmq->clientTopics);
+  taosHashCleanup(pHash);
   tmq->clientTopics = newTopics;
   atomic_store_32(&tmq->epoch, epoch);
   return set;
@@ -911,7 +971,7 @@ int32_t tmqAskEpCb(void* param, const SDataBuf* pMsg, int32_t code) {
   SMqAskEpCbParam* pParam = (SMqAskEpCbParam*)param;
   tmq_t*           tmq = pParam->tmq;
   if (code != 0) {
-    printf("get topic endpoint error, not ready, wait:%d\n", pParam->sync);
+    tscError("get topic endpoint error, not ready, wait:%d\n", pParam->sync);
     goto END;
   }
 
@@ -1282,6 +1342,7 @@ tmq_message_t* tmq_consumer_poll(tmq_t* tmq, int64_t blocking_time) {
 
   while (1) {
     /*printf("cycle\n");*/
+    tmqAskEp(tmq, false);
     tmqPollImpl(tmq, blocking_time);
 
     tsem_wait(&tmq->rspSem);

source/client/test/tmqTest.cpp

@@ -35,7 +35,7 @@ int main(int argc, char** argv) {
 
 TEST(testCase, driverInit_Test) {
   // taosInitGlobalCfg();
-//  taos_init();
+  //  taos_init();
 }
 
 TEST(testCase, create_topic_ctb_Test) {
@@ -55,8 +55,9 @@ TEST(testCase, create_topic_ctb_Test) {
 
   taos_free_result(pRes);
 
-  char* sql = "select * from tu";
+  // char* sql = "select * from tu";
-  pRes = tmq_create_topic(pConn, "test_ctb_topic_1", sql, strlen(sql));
+  // pRes = tmq_create_topic(pConn, "test_ctb_topic_1", sql, strlen(sql));
+  pRes = taos_query(pConn, "create test_ctb_topic_1 as select * from tu");
   taos_free_result(pRes);
   taos_close(pConn);
 }
@@ -69,7 +70,7 @@ TEST(testCase, create_topic_stb_Test) {
   if (taos_errno(pRes) != 0) {
     printf("error in use db, reason:%s\n", taos_errstr(pRes));
   }
-  //taos_free_result(pRes);
+  // taos_free_result(pRes);
 
   TAOS_FIELD* pFields = taos_fetch_fields(pRes);
   ASSERT_TRUE(pFields == nullptr);
@@ -79,8 +80,9 @@ TEST(testCase, create_topic_stb_Test) {
 
   taos_free_result(pRes);
 
-  char* sql = "select * from st1";
+  // char* sql = "select * from st1";
-  pRes = tmq_create_topic(pConn, "test_stb_topic_1", sql, strlen(sql));
+  // pRes = tmq_create_topic(pConn, "test_stb_topic_1", sql, strlen(sql));
+  pRes = taos_query(pConn, "create test_ctb_topic_1 as select * from st1");
   taos_free_result(pRes);
   taos_close(pConn);
 }

source/common/src/tdatablock.c

@@ -331,7 +331,6 @@ int32_t blockDataUpdateTsWindow(SSDataBlock* pDataBlock) {
     return 0;
   }
 
-  ASSERT(pColInfoData->nullbitmap == NULL);
   pDataBlock->info.window.skey = *(TSKEY*)colDataGetData(pColInfoData, 0);
   pDataBlock->info.window.ekey = *(TSKEY*)colDataGetData(pColInfoData, (pDataBlock->info.rows - 1));
   return 0;
@@ -609,22 +608,6 @@ size_t blockDataGetSerialMetaSize(const SSDataBlock* pBlock) {
   return sizeof(int32_t) + pBlock->info.numOfCols * sizeof(int32_t);
 }
 
-SSchema* blockDataExtractSchema(const SSDataBlock* pBlock, int32_t* numOfCols) {
-  SSchema* pSchema = taosMemoryCalloc(pBlock->info.numOfCols, sizeof(SSchema));
-  for (int32_t i = 0; i < pBlock->info.numOfCols; ++i) {
-    SColumnInfoData* pColInfoData = taosArrayGet(pBlock->pDataBlock, i);
-    pSchema[i].bytes = pColInfoData->info.bytes;
-    pSchema[i].type = pColInfoData->info.type;
-    pSchema[i].colId = pColInfoData->info.colId;
-  }
-
-  if (numOfCols != NULL) {
-    *numOfCols = pBlock->info.numOfCols;
-  }
-
-  return pSchema;
-}
-
 double blockDataGetSerialRowSize(const SSDataBlock* pBlock) {
   ASSERT(pBlock != NULL);
   double rowSize = 0;
@@ -647,7 +630,6 @@ double blockDataGetSerialRowSize(const SSDataBlock* pBlock) {
 typedef struct SSDataBlockSortHelper {
   SArray*      orderInfo;  // SArray<SBlockOrderInfo>
   SSDataBlock* pDataBlock;
-  bool         nullFirst;
 } SSDataBlockSortHelper;
 
 int32_t dataBlockCompar(const void* p1, const void* p2, const void* param) {
@@ -672,11 +654,11 @@ int32_t dataBlockCompar(const void* p1, const void* p2, const void* param) {
       }
 
       if (rightNull) {
-        return pHelper->nullFirst ? 1 : -1;
+        return pOrder->nullFirst ? 1 : -1;
       }
 
       if (leftNull) {
-        return pHelper->nullFirst ? -1 : 1;
+        return pOrder->nullFirst ? -1 : 1;
       }
     }
 
@@ -907,7 +889,7 @@ static __compar_fn_t getComparFn(int32_t type, int32_t order) {
   }
 }
 
-int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullFirst) {
+int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
   ASSERT(pDataBlock != NULL && pOrderInfo != NULL);
   if (pDataBlock->info.rows <= 1) {
     return TSDB_CODE_SUCCESS;
@@ -922,7 +904,7 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullFirs
   for (int32_t i = 0; i < taosArrayGetSize(pOrderInfo); ++i) {
     SBlockOrderInfo* pInfo = taosArrayGet(pOrderInfo, i);
 
-    SColumnInfoData* pColInfoData = taosArrayGet(pDataBlock->pDataBlock, pInfo->colIndex);
+    SColumnInfoData* pColInfoData = taosArrayGet(pDataBlock->pDataBlock, pInfo->slotId);
     if (pColInfoData->hasNull) {
       sortColumnHasNull = true;
     }
@@ -961,10 +943,10 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullFirs
 
   int64_t p0 = taosGetTimestampUs();
 
-  SSDataBlockSortHelper helper = {.nullFirst = nullFirst, .pDataBlock = pDataBlock, .orderInfo = pOrderInfo};
+  SSDataBlockSortHelper helper = {.pDataBlock = pDataBlock, .orderInfo = pOrderInfo};
   for (int32_t i = 0; i < taosArrayGetSize(helper.orderInfo); ++i) {
     struct SBlockOrderInfo* pInfo = taosArrayGet(helper.orderInfo, i);
-    pInfo->pColData = taosArrayGet(pDataBlock->pDataBlock, pInfo->colIndex);
+    pInfo->pColData = taosArrayGet(pDataBlock->pDataBlock, pInfo->slotId);
   }
 
   taosqsort(index, rows, sizeof(int32_t), &helper, dataBlockCompar);
@@ -1012,7 +994,7 @@ SHelper* createTupleIndex_rv(int32_t numOfRows, SArray* pOrderInfo, SSDataBlock*
 
   for (int32_t i = 0; i < numOfCols; ++i) {
     SBlockOrderInfo* pInfo = taosArrayGet(pOrderInfo, i);
-    SColumnInfoData* pColInfo = taosArrayGet(pBlock->pDataBlock, pInfo->colIndex);
+    SColumnInfoData* pColInfo = taosArrayGet(pBlock->pDataBlock, pInfo->slotId);
     pInfo->pColData = pColInfo;
     sortValLengthPerRow += pColInfo->info.bytes;
   }
@@ -1106,7 +1088,7 @@ int32_t blockDataSort_rv(SSDataBlock* pDataBlock, SArray* pOrderInfo, bool nullF
   // Allocate the additional buffer.
   int64_t p0 = taosGetTimestampUs();
 
-  SSDataBlockSortHelper helper = {.nullFirst = nullFirst, .pDataBlock = pDataBlock, .orderInfo = pOrderInfo};
+  SSDataBlockSortHelper helper = {.pDataBlock = pDataBlock, .orderInfo = pOrderInfo};
 
   uint32_t rows = pDataBlock->info.rows;
   SHelper* index = createTupleIndex_rv(rows, helper.orderInfo, pDataBlock);

source/common/src/tglobal.c

@@ -45,7 +45,7 @@ float   tsRatioOfQueryCores = 1.0f;
 int32_t tsMaxBinaryDisplayWidth = 30;
 bool    tsEnableSlaveQuery = 1;
 bool    tsPrintAuth = 0;
-int32_t tsMultiProcess = 0;
+bool    tsMultiProcess = 0;
 
 // monitor
 bool     tsEnableMonitor = 1;
@@ -347,7 +347,7 @@ static int32_t taosAddServerCfg(SConfig *pCfg) {
   if (cfgAddBool(pCfg, "printAuth", tsPrintAuth, 0) != 0) return -1;
   if (cfgAddBool(pCfg, "slaveQuery", tsEnableSlaveQuery, 0) != 0) return -1;
   if (cfgAddBool(pCfg, "deadLockKillQuery", tsDeadLockKillQuery, 0) != 0) return -1;
-  if (cfgAddInt32(pCfg, "multiProcess", tsMultiProcess, 0, 2, 0) != 0) return -1;
+  if (cfgAddBool(pCfg, "multiProcess", tsMultiProcess, 0) != 0) return -1;
 
   if (cfgAddBool(pCfg, "monitor", tsEnableMonitor, 0) != 0) return -1;
   if (cfgAddInt32(pCfg, "monitorInterval", tsMonitorInterval, 1, 360000, 0) != 0) return -1;
@@ -466,7 +466,7 @@ static int32_t taosSetServerCfg(SConfig *pCfg) {
   tsPrintAuth = cfgGetItem(pCfg, "printAuth")->bval;
   tsEnableSlaveQuery = cfgGetItem(pCfg, "slaveQuery")->bval;
   tsDeadLockKillQuery = cfgGetItem(pCfg, "deadLockKillQuery")->bval;
-  tsMultiProcess = cfgGetItem(pCfg, "multiProcess")->i32;
+  tsMultiProcess = cfgGetItem(pCfg, "multiProcess")->bval;
 
   tsEnableMonitor = cfgGetItem(pCfg, "monitor")->bval;
   tsMonitorInterval = cfgGetItem(pCfg, "monitorInterval")->i32;

source/common/src/tmsgcb.c

@@ -16,6 +16,10 @@
 #define _DEFAULT_SOURCE
 #include "tmsgcb.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);
 }
@@ -24,7 +28,7 @@ int32_t tmsgGetQueueSize(const SMsgCb* pMsgCb, int32_t vgId, EQueueType qtype) {
   return (*pMsgCb->qsizeFp)(pMsgCb->pWrapper, vgId, qtype);
 }
 
-int32_t tmsgSendReq(const SMsgCb* pMsgCb, SEpSet* epSet, SRpcMsg* pReq) {
+int32_t tmsgSendReq(const SMsgCb* pMsgCb, const SEpSet* epSet, SRpcMsg* pReq) {
   return (*pMsgCb->sendReqFp)(pMsgCb->pWrapper, epSet, pReq);
 }
 
@@ -32,4 +36,12 @@ int32_t tmsgSendMnodeReq(const SMsgCb* pMsgCb, SRpcMsg* pReq) {
   return (*pMsgCb->sendMnodeReqFp)(pMsgCb->pWrapper, pReq);
 }
 
-void tmsgSendRsp(const SMsgCb* pMsgCb, SRpcMsg* pRsp) { return (*pMsgCb->sendRspFp)(pMsgCb->pWrapper, pRsp); }
+void tmsgSendRsp(const SRpcMsg* pRsp) { return (*tsDefaultMsgCb.sendRspFp)(tsDefaultMsgCb.pWrapper, pRsp); }
+
+void tmsgRegisterBrokenLinkArg(const SMsgCb* pMsgCb, SRpcMsg* pMsg) {
+  (*pMsgCb->registerBrokenLinkArgFp)(pMsgCb->pWrapper, pMsg);
+}
+
+void tmsgReleaseHandle(void* handle, int8_t type) {
+  (*tsDefaultMsgCb.releaseHandleFp)(tsDefaultMsgCb.pWrapper, handle, type);
+}

source/common/test/commonTests.cpp

@@ -167,10 +167,10 @@ TEST(testCase, Datablock_test) {
   printf("the second row of binary:%s, length:%d\n", (char*)varDataVal(pData), varDataLen(pData));
 
   SArray* pOrderInfo = taosArrayInit(3, sizeof(SBlockOrderInfo));
-  SBlockOrderInfo order = {.order = TSDB_ORDER_ASC, .colIndex = 0};
+  SBlockOrderInfo order = {.nullFirst = true, .order = TSDB_ORDER_ASC, .slotId = 0};
   taosArrayPush(pOrderInfo, &order);
 
-  blockDataSort(b, pOrderInfo, true);
+  blockDataSort(b, pOrderInfo);
   blockDataDestroy(b);
 
   taosArrayDestroy(pOrderInfo);

source/dnode/mgmt/CMakeLists.txt

@@ -1,12 +1,10 @@
-aux_source_directory(src DNODE_SRC)
+aux_source_directory(dm/src DNODE_SRC)
-aux_source_directory(dnode/src DNODE_SRC)
+aux_source_directory(qm/src DNODE_SRC)
-aux_source_directory(qnode/src DNODE_SRC)
+aux_source_directory(bm/src DNODE_SRC)
-aux_source_directory(bnode/src DNODE_SRC)
+aux_source_directory(sm/src DNODE_SRC)
-aux_source_directory(snode/src DNODE_SRC)
+aux_source_directory(vm/src DNODE_SRC)
-aux_source_directory(vnode/src DNODE_SRC)
+aux_source_directory(mm/src DNODE_SRC)
-aux_source_directory(mnode/src DNODE_SRC)
+aux_source_directory(main/src DNODE_SRC)
-aux_source_directory(container/src DNODE_SRC)
-
 add_library(dnode STATIC ${DNODE_SRC})
 target_link_libraries(
     dnode cjson mnode vnode qnode snode bnode wal sync taos tfs monitor
@@ -14,16 +12,22 @@ target_link_libraries(
 target_include_directories(
     dnode
     PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/mgmt"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/dnode/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/dm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/qnode/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/qm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/bnode/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/bm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/snode/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/sm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/vnode/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/vm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/mnode/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/mm/inc"
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/container/inc"
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/main/inc"
 )
 
-add_subdirectory(main)
+aux_source_directory(main/exe EXEC_SRC)
+add_executable(taosd ${EXEC_SRC})
+target_include_directories(
+    taosd
+    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/main/inc"
+)
+target_link_libraries(taosd dnode)
 
 if(${BUILD_TEST})
     add_subdirectory(test)

source/dnode/mgmt/bm/src/bmInt.c

@@ -24,6 +24,7 @@ static void bmInitOption(SBnodeMgmt *pMgmt, SBnodeOpt *pOption) {
   msgCb.sendReqFp = dndSendReqToDnode;
   msgCb.sendMnodeReqFp = dndSendReqToMnode;
   msgCb.sendRspFp = dndSendRsp;
+  msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
   pOption->msgCb = msgCb;
 }
 

source/dnode/mgmt/container/src/dndFile.c

@@ -1,103 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http:www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "dndInt.h"
-
-#define MAXLEN 1024
-
-int32_t dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed) {
-  int32_t       code = TSDB_CODE_NODE_PARSE_FILE_ERROR;
-  int32_t       len = 0;
-  const int32_t maxLen = MAXLEN;
-  char          content[MAXLEN + 1] = {0};
-  cJSON        *root = NULL;
-  char          file[PATH_MAX];
-  TdFilePtr     pFile = NULL;
-
-  snprintf(file, sizeof(file), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
-  pFile = taosOpenFile(file, TD_FILE_READ);
-  if (pFile == NULL) {
-    dDebug("file %s not exist", file);
-    code = 0;
-    goto _OVER;
-  }
-
-  len = (int32_t)taosReadFile(pFile, content, maxLen);
-  if (len <= 0) {
-    dError("failed to read %s since content is null", file);
-    goto _OVER;
-  }
-
-  content[len] = 0;
-  root = cJSON_Parse(content);
-  if (root == NULL) {
-    dError("failed to read %s since invalid json format", file);
-    goto _OVER;
-  }
-
-  cJSON *deployed = cJSON_GetObjectItem(root, "deployed");
-  if (!deployed || deployed->type != cJSON_Number) {
-    dError("failed to read %s since deployed not found", file);
-    goto _OVER;
-  }
-  *pDeployed = deployed->valueint != 0;
-
-  code = 0;
-  dDebug("succcessed to read file %s, deployed:%d", file, *pDeployed);
-
-_OVER:
-  if (root != NULL) cJSON_Delete(root);
-  if (pFile != NULL) taosCloseFile(&pFile);
-
-  terrno = code;
-  return code;
-}
-
-int32_t dndWriteFile(SMgmtWrapper *pWrapper, bool deployed) {
-  char file[PATH_MAX] = {0};
-  snprintf(file, sizeof(file), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
-
-  TdFilePtr pFile = taosOpenFile(file, TD_FILE_CTEATE | TD_FILE_WRITE | TD_FILE_TRUNC);
-  if (pFile == NULL) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    dError("failed to write %s since %s", file, terrstr());
-    return -1;
-  }
-
-  int32_t       len = 0;
-  const int32_t maxLen = MAXLEN;
-  char          content[MAXLEN + 1] = {0};
-
-  len += snprintf(content + len, maxLen - len, "{\n");
-  len += snprintf(content + len, maxLen - len, "  \"deployed\": %d\n", deployed);
-  len += snprintf(content + len, maxLen - len, "}\n");
-
-  taosWriteFile(pFile, content, len);
-  taosFsyncFile(pFile);
-  taosCloseFile(&pFile);
-
-  char realfile[PATH_MAX] = {0};
-  snprintf(realfile, sizeof(realfile), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
-
-  if (taosRenameFile(file, realfile) != 0) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    dError("failed to rename %s since %s", file, terrstr());
-    return -1;
-  }
-
-  dInfo("successed to write %s, deployed:%d", realfile, deployed);
-  return 0;
-}

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

@@ -29,7 +29,7 @@ void dmInitMsgHandles(SMgmtWrapper *pWrapper);
 
 void    dmGetMnodeEpSet(SDnodeMgmt *pMgmt, SEpSet *pEpSet);
 void    dmUpdateMnodeEpSet(SDnodeMgmt *pMgmt, SEpSet *pEpSet);
-void    dmSendRedirectRsp(SDnodeMgmt *pMgmt, SRpcMsg *pMsg);
+void    dmSendRedirectRsp(SDnodeMgmt *pMgmt, const SRpcMsg *pMsg);
 
 #ifdef __cplusplus
 }

source/dnode/mgmt/dm/src/dmInt.c

@@ -54,7 +54,7 @@ void dmGetDnodeEp(SMgmtWrapper *pWrapper, int32_t dnodeId, char *pEp, char *pFqd
   taosRUnLockLatch(&pMgmt->latch);
 }
 
-void dmSendRedirectRsp(SDnodeMgmt *pMgmt, SRpcMsg *pReq) {
+void dmSendRedirectRsp(SDnodeMgmt *pMgmt, const SRpcMsg *pReq) {
   SDnode *pDnode = pMgmt->pDnode;
 
   SEpSet epSet = {0};

source/dnode/mgmt/main/CMakeLists.txt

@@ -1,9 +0,0 @@
-aux_source_directory(src EXEC_SRC)
-add_executable(taosd ${EXEC_SRC})
-
-target_include_directories(
-    taosd
-    PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/inc"
-)
-
-target_link_libraries(taosd dnode)

source/dnode/mgmt/main/exe/dndMain.c

@@ -0,0 +1,217 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "dnd.h"
+#include "tconfig.h"
+
+static struct {
+  bool      dumpConfig;
+  bool      generateGrant;
+  bool      printAuth;
+  bool      printVersion;
+  char      envFile[PATH_MAX];
+  char      apolloUrl[PATH_MAX];
+  SArray   *pArgs;  // SConfigPair
+  SDnode   *pDnode;
+  ENodeType ntype;
+} global = {0};
+
+static void dndStopDnode(int signum, void *info, void *ctx) {
+  dInfo("signal:%d is received", signum);
+  SDnode *pDnode = atomic_val_compare_exchange_ptr(&global.pDnode, 0, global.pDnode);
+  if (pDnode != NULL) {
+    dndHandleEvent(pDnode, DND_EVENT_STOP);
+  }
+}
+
+static void dndHandleChild(int signum, void *info, void *ctx) {
+  dInfo("signal:%d is received", signum);
+  dndHandleEvent(global.pDnode, DND_EVENT_CHILD);
+}
+
+static void dndSetSignalHandle() {
+  taosSetSignal(SIGTERM, dndStopDnode);
+  taosSetSignal(SIGHUP, dndStopDnode);
+  taosSetSignal(SIGINT, dndStopDnode);
+  taosSetSignal(SIGABRT, dndStopDnode);
+  taosSetSignal(SIGBREAK, dndStopDnode);
+
+  if (!tsMultiProcess) {
+    // Set the single process signal
+  } else if (global.ntype == DNODE) {
+    // Set the parent process signal
+    // When the child process exits, the parent process receives a signal
+    taosSetSignal(SIGCHLD, dndHandleChild);
+  } else {
+    // Set child process signal
+    // When the parent process exits, the child process will receive the SIGKILL signal
+    prctl(PR_SET_PDEATHSIG, SIGKILL);
+  }
+}
+
+static int32_t dndParseArgs(int32_t argc, char const *argv[]) {
+  for (int32_t i = 1; i < argc; ++i) {
+    if (strcmp(argv[i], "-c") == 0) {
+      if (i < argc - 1) {
+        if (strlen(argv[++i]) >= PATH_MAX) {
+          printf("config file path overflow");
+          return -1;
+        }
+        tstrncpy(configDir, argv[i], PATH_MAX);
+      } else {
+        printf("'-c' requires a parameter, default is %s\n", configDir);
+        return -1;
+      }
+    } else if (strcmp(argv[i], "-a") == 0) {
+      tstrncpy(global.apolloUrl, argv[++i], PATH_MAX);
+    } else if (strcmp(argv[i], "-e") == 0) {
+      tstrncpy(global.envFile, argv[++i], PATH_MAX);
+    } else if (strcmp(argv[i], "-k") == 0) {
+      global.generateGrant = true;
+    } else if (strcmp(argv[i], "-n") == 0) {
+      global.ntype = atoi(argv[++i]);
+    } else if (strcmp(argv[i], "-C") == 0) {
+      global.dumpConfig = true;
+    } else if (strcmp(argv[i], "-V") == 0) {
+      global.printVersion = true;
+    } else {
+    }
+  }
+
+  return 0;
+}
+
+static void dndGenerateGrant() {
+  // grantParseParameter();
+  printf("this feature is not implemented yet\n");
+}
+
+static void dndPrintVersion() {
+#ifdef TD_ENTERPRISE
+  char *releaseName = "enterprise";
+#else
+  char *releaseName = "community";
+#endif
+  printf("%s version: %s compatible_version: %s\n", releaseName, version, compatible_version);
+  printf("gitinfo: %s\n", gitinfo);
+  printf("buildInfo: %s\n", buildinfo);
+}
+
+static void dndDumpCfg() {
+  SConfig *pCfg = taosGetCfg();
+  cfgDumpCfg(pCfg, 0, 1);
+}
+
+static SDnodeOpt dndGetOpt() {
+  SConfig  *pCfg = taosGetCfg();
+  SDnodeOpt option = {0};
+
+  option.numOfSupportVnodes = cfgGetItem(pCfg, "supportVnodes")->i32;
+  tstrncpy(option.dataDir, tsDataDir, sizeof(option.dataDir));
+  tstrncpy(option.firstEp, tsFirst, sizeof(option.firstEp));
+  tstrncpy(option.secondEp, tsSecond, sizeof(option.firstEp));
+  option.serverPort = tsServerPort;
+  tstrncpy(option.localFqdn, tsLocalFqdn, sizeof(option.localFqdn));
+  snprintf(option.localEp, sizeof(option.localEp), "%s:%u", option.localFqdn, option.serverPort);
+  option.disks = tsDiskCfg;
+  option.numOfDisks = tsDiskCfgNum;
+  option.ntype = global.ntype;
+  return option;
+}
+
+static int32_t dndInitLog() {
+  char logName[12] = {0};
+  snprintf(logName, sizeof(logName), "%slog", dndNodeLogStr(global.ntype));
+  return taosCreateLog(logName, 1, configDir, global.envFile, global.apolloUrl, global.pArgs, 0);
+}
+
+static void dndSetProcName(char **argv) {
+  if (global.ntype != DNODE) {
+    const char *name = dndNodeProcStr(global.ntype);
+    prctl(PR_SET_NAME, name);
+    strcpy(argv[0], name);
+  }
+}
+
+static int32_t dndRunDnode() {
+  if (dndInit() != 0) {
+    dError("failed to initialize environment since %s", terrstr());
+    return -1;
+  }
+
+  SDnodeOpt option = dndGetOpt();
+  SDnode   *pDnode = dndCreate(&option);
+  if (pDnode == NULL) {
+    dError("failed to to create dnode object since %s", terrstr());
+    return -1;
+  } else {
+    global.pDnode = pDnode;
+    dndSetSignalHandle();
+  }
+
+  dInfo("start the service");
+  int32_t code = dndRun(pDnode);
+  dInfo("start shutting down the service");
+
+  global.pDnode = NULL;
+  dndClose(pDnode);
+  dndCleanup();
+  taosCloseLog();
+  taosCleanupCfg();
+  return code;
+}
+
+int main(int argc, char const *argv[]) {
+  if (!taosCheckSystemIsSmallEnd()) {
+    printf("failed to start since on non-small-end machines\n");
+    return -1;
+  }
+
+  if (dndParseArgs(argc, argv) != 0) {
+    printf("failed to start since parse args error\n");
+    return -1;
+  }
+
+  if (global.generateGrant) {
+    dndGenerateGrant();
+    return 0;
+  }
+
+  if (global.printVersion) {
+    dndPrintVersion();
+    return 0;
+  }
+
+  if (dndInitLog() != 0) {
+    printf("failed to start since init log error\n");
+    return -1;
+  }
+
+  if (taosInitCfg(configDir, global.envFile, global.apolloUrl, global.pArgs, 0) != 0) {
+    dError("failed to start since read config error");
+    return -1;
+  }
+
+  if (global.dumpConfig) {
+    dndDumpCfg();
+    taosCleanupCfg();
+    taosCloseLog();
+    return 0;
+  }
+
+  dndSetProcName((char **)argv);
+  return dndRunDnode();
+}

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

@@ -33,6 +33,7 @@
 #include "tthread.h"
 #include "ttime.h"
 #include "tworker.h"
+#include "tmsgcb.h"
 
 #include "dnode.h"
 #include "monitor.h"
@@ -95,10 +96,11 @@ typedef struct SMgmtWrapper {
   bool        required;
   EProcType   procType;
   SProcObj   *pProc;
+  SShm        shm;
   void       *pMgmt;
   SDnode     *pDnode;
   NodeMsgFp   msgFps[TDMT_MAX];
-  int32_t     msgVgIds[TDMT_MAX];  // Handle the case where the same message type is distributed to qnode or vnode
+  int8_t      msgVgIds[TDMT_MAX];  // Handle the case where the same message type is distributed to qnode or vnode
   SMgmtFp     fp;
 } SMgmtWrapper;
 
@@ -118,27 +120,32 @@ typedef struct SDnode {
   char        *firstEp;
   char        *secondEp;
   char        *dataDir;
-  SDiskCfg    *pDisks;
+  SDiskCfg    *disks;
   int32_t      numOfDisks;
   uint16_t     serverPort;
   bool         dropped;
+  ENodeType    ntype;
   EDndStatus   status;
   EDndEvent    event;
   SStartupReq  startup;
-  TdFilePtr    pLockFile;
+  TdFilePtr    runtimeFile;
   STransMgmt   trans;
   SMgmtWrapper wrappers[NODE_MAX];
 } SDnode;
 
-EDndStatus dndGetStatus(SDnode *pDnode);
+const char *dndNodeLogStr(ENodeType ntype);
-void       dndSetStatus(SDnode *pDnode, EDndStatus stat);
+const char *dndNodeProcStr(ENodeType ntype);
-void       dndSetMsgHandle(SMgmtWrapper *pWrapper, int32_t msgType, NodeMsgFp nodeMsgFp, int32_t vgId);
+EDndStatus  dndGetStatus(SDnode *pDnode);
-void       dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc);
+void        dndSetStatus(SDnode *pDnode, EDndStatus stat);
-void       dndSendMonitorReport(SDnode *pDnode);
+void        dndSetMsgHandle(SMgmtWrapper *pWrapper, tmsg_t msgType, NodeMsgFp nodeMsgFp, int8_t vgId);
+void        dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc);
+void        dndSendMonitorReport(SDnode *pDnode);
 
 int32_t dndSendReqToMnode(SMgmtWrapper *pWrapper, SRpcMsg *pMsg);
-int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, SEpSet *pEpSet, SRpcMsg *pMsg);
+int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, const SEpSet *pEpSet, SRpcMsg *pMsg);
-void    dndSendRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp);
+void    dndSendRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp);
+void    dndRegisterBrokenLinkArg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg);
+SMsgCb  dndCreateMsgcb(SMgmtWrapper *pWrapper);
 
 int32_t dndProcessNodeMsg(SDnode *pDnode, SNodeMsg *pMsg);
 int32_t dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed);

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

@@ -34,7 +34,6 @@ int32_t     dndInit();
 void        dndCleanup();
 const char *dndStatStr(EDndStatus stat);
 void        dndGetStartup(SDnode *pDnode, SStartupReq *pStartup);
-TdFilePtr   dndCheckRunning(const char *dataDir);
 void        dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pMsg);
 
 // dndMsg.c
@@ -56,6 +55,12 @@ void    dndCleanupServer(SDnode *pDnode);
 int32_t dndInitClient(SDnode *pDnode);
 void    dndCleanupClient(SDnode *pDnode);
 int32_t dndInitMsgHandle(SDnode *pDnode);
+void    dndSendRpcRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp);
+
+// dndFile.c
+int32_t dndOpenRuntimeFile(SDnode *pDnode);
+int32_t dndWriteRuntimeFile(SDnode *pDnode);
+void    dndCloseRuntimeFile(SDnode *pDnode);
 
 #ifdef __cplusplus
 }

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

@@ -1,48 +0,0 @@
-
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef _TD_DND_MAIN_H_
-#define _TD_DND_MAIN_H_
-
-#include "dnode.h"
-
-#include "taoserror.h"
-#include "tconfig.h"
-#include "tglobal.h"
-#include "tlog.h"
-#include "version.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#define dFatal(...) { if (dDebugFlag & DEBUG_FATAL) { taosPrintLog("DND FATAL ", DEBUG_FATAL, 255, __VA_ARGS__); }}
-#define dError(...) { if (dDebugFlag & DEBUG_ERROR) { taosPrintLog("DND ERROR ", DEBUG_ERROR, 255, __VA_ARGS__); }}
-#define dWarn(...)  { if (dDebugFlag & DEBUG_WARN)  { taosPrintLog("DND WARN ", DEBUG_WARN, 255, __VA_ARGS__); }}
-#define dInfo(...)  { if (dDebugFlag & DEBUG_INFO)  { taosPrintLog("DND ", DEBUG_INFO, 255, __VA_ARGS__); }}
-#define dDebug(...) { if (dDebugFlag & DEBUG_DEBUG) { taosPrintLog("DND ", DEBUG_DEBUG, dDebugFlag, __VA_ARGS__); }}
-#define dTrace(...) { if (dDebugFlag & DEBUG_TRACE) { taosPrintLog("DND ", DEBUG_TRACE, dDebugFlag, __VA_ARGS__); }}
-
-void      dndDumpCfg();
-void      dndPrintVersion();
-void      dndGenerateGrant();
-SDnodeOpt dndGetOpt();
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_DND_MAIN_H_*/

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

@@ -77,6 +77,8 @@ static int32_t dndRunInSingleProcess(SDnode *pDnode) {
     SMgmtWrapper *pWrapper = &pDnode->wrappers[n];
     pWrapper->required = dndRequireNode(pWrapper);
     if (!pWrapper->required) continue;
+    SMsgCb msgCb = dndCreateMsgcb(pWrapper);
+    tmsgSetDefaultMsgCb(&msgCb);
 
     if (taosMkDir(pWrapper->path) != 0) {
       terrno = TAOS_SYSTEM_ERROR(errno);
@@ -116,11 +118,11 @@ static void dndClearNodesExecpt(SDnode *pDnode, ENodeType except) {
   }
 }
 
-static void dndConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int32_t msgLen, void *pCont, int32_t contLen) {
+static void dndConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
+                                 ProcFuncType ftype) {
   SRpcMsg *pRpc = &pMsg->rpcMsg;
   pRpc->pCont = pCont;
-  dTrace("msg:%p, get from child process queue, type:%s handle:%p app:%p", pMsg, TMSG_INFO(pRpc->msgType),
+  dTrace("msg:%p, get from child queue, handle:%p app:%p", pMsg, pRpc->handle, pRpc->ahandle);
-         pRpc->handle, pRpc->ahandle);
 
   NodeMsgFp msgFp = pWrapper->msgFps[TMSG_INDEX(pRpc->msgType)];
   int32_t   code = (*msgFp)(pWrapper, pMsg);
@@ -138,13 +140,27 @@ static void dndConsumeChildQueue(SMgmtWrapper *pWrapper, SNodeMsg *pMsg, int32_t
   }
 }
 
-static void dndConsumeParentQueue(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, int32_t msgLen, void *pCont, int32_t contLen) {
+static void dndConsumeParentQueue(SMgmtWrapper *pWrapper, SRpcMsg *pMsg, int16_t msgLen, void *pCont, int32_t contLen,
-  pRpc->pCont = pCont;
+                                  ProcFuncType ftype) {
-  dTrace("msg:%p, get from parent process queue, type:%s handle:%p app:%p", pRpc, TMSG_INFO(pRpc->msgType),
+  pMsg->pCont = pCont;
-         pRpc->handle, pRpc->ahandle);
+  dTrace("msg:%p, get from parent queue, handle:%p app:%p", pMsg, pMsg->handle, pMsg->ahandle);
 
-  dndSendRsp(pWrapper, pRpc);
+  switch (ftype) {
-  taosMemoryFree(pRpc);
+    case PROC_REG:
+      rpcRegisterBrokenLinkArg(pMsg);
+      break;
+    case PROC_RELEASE:
+      rpcReleaseHandle(pMsg->handle, (int8_t)pMsg->code);
+      rpcFreeCont(pCont);
+      break;
+    case PROC_REQ:
+      // todo send to dnode
+      dndSendReqToMnode(pWrapper, pMsg);
+    default:
+      dndSendRpcRsp(pWrapper, pMsg);
+      break;
+  }
+  taosMemoryFree(pMsg);
 }
 
 static int32_t dndRunInMultiProcess(SDnode *pDnode) {
@@ -155,6 +171,9 @@ static int32_t dndRunInMultiProcess(SDnode *pDnode) {
     pWrapper->required = dndRequireNode(pWrapper);
     if (!pWrapper->required) continue;
 
+    SMsgCb msgCb = dndCreateMsgcb(pWrapper);
+    tmsgSetDefaultMsgCb(&msgCb);
+
     if (taosMkDir(pWrapper->path) != 0) {
       terrno = TAOS_SYSTEM_ERROR(errno);
       dError("failed to create dir:%s since %s", pWrapper->path, terrstr());
@@ -202,15 +221,23 @@ static int32_t dndRunInMultiProcess(SDnode *pDnode) {
       dndClearNodesExecpt(pDnode, n);
 
       dInfo("node:%s, will be initialized in child process", pWrapper->name);
-      dndOpenNode(pWrapper);
+      if (dndOpenNode(pWrapper) != 0) {
+        dInfo("node:%s, failed to init in child process since %s", pWrapper->name, terrstr());
+        return -1;
+      }
+
+      if (taosProcRun(pProc) != 0) {
+        dError("node:%s, failed to run proc since %s", pWrapper->name, terrstr());
+        return -1;
+      }
+      break;
     } else {
       dInfo("node:%s, will not start in parent process, child pid:%d", pWrapper->name, taosProcChildId(pProc));
       pWrapper->procType = PROC_PARENT;
-    }
+      if (taosProcRun(pProc) != 0) {
-
+        dError("node:%s, failed to run proc since %s", pWrapper->name, terrstr());
-    if (taosProcRun(pProc) != 0) {
+        return -1;
-      dError("node:%s, failed to run proc since %s", pWrapper->name, terrstr());
+      }
-      return -1;
     }
   }
 
@@ -232,7 +259,7 @@ static int32_t dndRunInMultiProcess(SDnode *pDnode) {
 }
 
 int32_t dndRun(SDnode *pDnode) {
-  if (tsMultiProcess == 0) {
+  if (!tsMultiProcess) {
     if (dndRunInSingleProcess(pDnode) != 0) {
       dError("failed to run dnode in single process mode since %s", terrstr());
       return -1;

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

@@ -0,0 +1,264 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http:www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "dndInt.h"
+
+#define MAXLEN 1024
+
+int32_t dndReadFile(SMgmtWrapper *pWrapper, bool *pDeployed) {
+  int32_t   code = TSDB_CODE_NODE_PARSE_FILE_ERROR;
+  int64_t   len = 0;
+  char      content[MAXLEN + 1] = {0};
+  cJSON    *root = NULL;
+  char      file[PATH_MAX];
+  TdFilePtr pFile = NULL;
+
+  snprintf(file, sizeof(file), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
+  pFile = taosOpenFile(file, TD_FILE_READ);
+  if (pFile == NULL) {
+    dDebug("file %s not exist", file);
+    code = 0;
+    goto _OVER;
+  }
+
+  len = taosReadFile(pFile, content, MAXLEN);
+  if (len <= 0) {
+    dError("failed to read %s since content is null", file);
+    goto _OVER;
+  }
+
+  root = cJSON_Parse(content);
+  if (root == NULL) {
+    dError("failed to read %s since invalid json format", file);
+    goto _OVER;
+  }
+
+  cJSON *deployed = cJSON_GetObjectItem(root, "deployed");
+  if (!deployed || deployed->type != cJSON_Number) {
+    dError("failed to read %s since deployed not found", file);
+    goto _OVER;
+  }
+  *pDeployed = deployed->valueint != 0;
+
+  dDebug("succcessed to read file %s, deployed:%d", file, *pDeployed);
+  code = 0;
+
+_OVER:
+  if (root != NULL) cJSON_Delete(root);
+  if (pFile != NULL) taosCloseFile(&pFile);
+
+  terrno = code;
+  return code;
+}
+
+int32_t dndWriteFile(SMgmtWrapper *pWrapper, bool deployed) {
+  int32_t   code = -1;
+  int32_t   len = 0;
+  char      content[MAXLEN + 1] = {0};
+  char      file[PATH_MAX] = {0};
+  char      realfile[PATH_MAX] = {0};
+  TdFilePtr pFile = NULL;
+
+  snprintf(file, sizeof(file), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
+  snprintf(realfile, sizeof(realfile), "%s%s%s.json", pWrapper->path, TD_DIRSEP, pWrapper->name);
+
+  pFile = taosOpenFile(file, TD_FILE_CTEATE | TD_FILE_WRITE | TD_FILE_TRUNC);
+  if (pFile == NULL) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to write %s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  len += snprintf(content + len, MAXLEN - len, "{\n");
+  len += snprintf(content + len, MAXLEN - len, "  \"deployed\": %d\n", deployed);
+  len += snprintf(content + len, MAXLEN - len, "}\n");
+
+  if (taosWriteFile(pFile, content, len) != len) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to write file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  if (taosFsyncFile(pFile) != 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to fsync file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  taosCloseFile(&pFile);
+
+  if (taosRenameFile(file, realfile) != 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to rename %s since %s", file, terrstr());
+    return -1;
+  }
+
+  dInfo("successed to write %s, deployed:%d", realfile, deployed);
+  code = 0;
+
+_OVER:
+  if (pFile != NULL) {
+    taosCloseFile(&pFile);
+  }
+
+  return code;
+}
+
+int32_t dndOpenRuntimeFile(SDnode *pDnode) {
+  int32_t   code = -1;
+  char      itemName[24] = {0};
+  char      content[MAXLEN + 1] = {0};
+  char      file[PATH_MAX] = {0};
+  cJSON    *root = NULL;
+  TdFilePtr pFile = NULL;
+
+  snprintf(file, sizeof(file), "%s%s.running", pDnode->dataDir, TD_DIRSEP);
+  pFile = taosOpenFile(file, TD_FILE_CTEATE | TD_FILE_WRITE | TD_FILE_TRUNC);
+  if (pFile == NULL) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to open file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  if (taosLockFile(pFile) != 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to lock file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  if (taosReadFile(pFile, content, MAXLEN) > 0) {
+    root = cJSON_Parse(content);
+    if (root == NULL) {
+      terrno = TSDB_CODE_NODE_PARSE_FILE_ERROR;
+      dError("failed to read %s since invalid json format", file);
+      goto _OVER;
+    }
+
+    for (ENodeType ntype = DNODE + 1; ntype < NODE_MAX; ++ntype) {
+      snprintf(itemName, sizeof(itemName), "%s_shmid", dndNodeProcStr(ntype));
+      cJSON *shmid = cJSON_GetObjectItem(root, itemName);
+      if (shmid && shmid->type == cJSON_Number) {
+        pDnode->wrappers[ntype].shm.id = shmid->valueint;
+      }
+
+      snprintf(itemName, sizeof(itemName), "%s_shmsize", dndNodeProcStr(ntype));
+      cJSON *shmsize = cJSON_GetObjectItem(root, itemName);
+      if (shmsize && shmsize->type == cJSON_Number) {
+        pDnode->wrappers[ntype].shm.size = shmsize->valueint;
+      }
+    }
+  }
+
+  if (tsMultiProcess || pDnode->ntype == DNODE) {
+    for (ENodeType ntype = DNODE; ntype < NODE_MAX; ++ntype) {
+      SMgmtWrapper *pWrapper = &pDnode->wrappers[pDnode->ntype];
+      if (pWrapper->shm.id > 0) {
+        dDebug("shmid:%d, is closed, size:%d", pWrapper->shm.id, pWrapper->shm.size);
+        taosDropShm(&pWrapper->shm);
+      }
+    }
+  } else {
+    SMgmtWrapper *pWrapper = &pDnode->wrappers[pDnode->ntype];
+    if (taosAttachShm(&pWrapper->shm) != 0) {
+      terrno = TAOS_SYSTEM_ERROR(errno);
+      dError("shmid:%d, failed to attach since %s", pWrapper->shm.id, terrstr());
+      goto _OVER;
+    }
+    dDebug("shmid:%d, is attached, size:%d", pWrapper->shm.id, pWrapper->shm.size);
+  }
+
+  dDebug("successed to open %s", file);
+  code = 0;
+
+_OVER:
+  if (root != NULL) cJSON_Delete(root);
+  if (code != 0) {
+    if (pFile != NULL) taosCloseFile(&pFile);
+  } else {
+    pDnode->runtimeFile = pFile;
+  }
+
+  return code;
+}
+
+int32_t dndWriteRuntimeFile(SDnode *pDnode) {
+  int32_t   code = -1;
+  int32_t   len = 0;
+  char      content[MAXLEN + 1] = {0};
+  char      file[PATH_MAX] = {0};
+  char      realfile[PATH_MAX] = {0};
+  TdFilePtr pFile = NULL;
+
+  snprintf(file, sizeof(file), "%s%s.running.bak", pDnode->dataDir, TD_DIRSEP);
+  snprintf(realfile, sizeof(realfile), "%s%s.running", pDnode->dataDir, TD_DIRSEP);
+
+  pFile = taosOpenFile(file, TD_FILE_CTEATE | TD_FILE_WRITE | TD_FILE_TRUNC);
+  if (pFile == NULL) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to open file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  len += snprintf(content + len, MAXLEN - len, "{\n");
+  for (ENodeType ntype = DNODE + 1; ntype < NODE_MAX; ++ntype) {
+    SMgmtWrapper *pWrapper = &pDnode->wrappers[pDnode->ntype];
+    len += snprintf(content + len, MAXLEN - len, "  \"%s_shmid\": %d,\n", dndNodeProcStr(ntype), pWrapper->shm.id);
+    if (ntype == NODE_MAX - 1) {
+      len += snprintf(content + len, MAXLEN - len, "  \"%s_shmsize\": %d\n", dndNodeProcStr(ntype), pWrapper->shm.size);
+    } else {
+      len += snprintf(content + len, MAXLEN - len, "  \"%s_shmsize\": %d,\n", dndNodeProcStr(ntype), pWrapper->shm.size);
+    }
+  }
+  len += snprintf(content + len, MAXLEN - len, "}\n");
+
+  if (taosWriteFile(pFile, content, len) != len) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to write file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  if (taosFsyncFile(pFile) != 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to fsync file:%s since %s", file, terrstr());
+    goto _OVER;
+  }
+
+  taosCloseFile(&pFile);
+
+  if (taosRenameFile(file, realfile) != 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    dError("failed to rename %s to %s since %s", file, realfile, terrstr());
+    return -1;
+  }
+
+  dDebug("successed to write %s", realfile);
+  code = 0;
+
+_OVER:
+  if (pFile != NULL) {
+    taosCloseFile(&pFile);
+  }
+
+  return code;
+}
+
+void dndCloseRuntimeFile(SDnode *pDnode) {
+  if (pDnode->runtimeFile) {
+    taosUnLockFile(pDnode->runtimeFile);
+    taosCloseFile(&pDnode->runtimeFile);
+    pDnode->runtimeFile = NULL;
+  }
+}

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

@@ -20,7 +20,7 @@
 static int8_t once = DND_ENV_INIT;
 
 int32_t dndInit() {
-  dInfo("start to init dnode env");
+  dDebug("start to init dnode env");
   if (atomic_val_compare_exchange_8(&once, DND_ENV_INIT, DND_ENV_READY) != DND_ENV_INIT) {
     terrno = TSDB_CODE_REPEAT_INIT;
     dError("failed to init dnode env since %s", terrstr());
@@ -31,12 +31,6 @@ int32_t dndInit() {
   taosBlockSIGPIPE();
   taosResolveCRC();
 
-  if (rpcInit() != 0) {
-    dError("failed to init rpc since %s", terrstr());
-    dndCleanup();
-    return -1;
-  }
-
   SMonCfg monCfg = {0};
   monCfg.maxLogs = tsMonitorMaxLogs;
   monCfg.port = tsMonitorPort;
@@ -44,29 +38,27 @@ int32_t dndInit() {
   monCfg.comp = tsMonitorComp;
   if (monInit(&monCfg) != 0) {
     dError("failed to init monitor since %s", terrstr());
-    dndCleanup();
     return -1;
   }
 
-  dInfo("dnode env is initialized");
+  dDebug("dnode env is initialized");
   return 0;
 }
 
 void dndCleanup() {
-  dInfo("start to cleanup dnode env");
+  dDebug("start to cleanup dnode env");
   if (atomic_val_compare_exchange_8(&once, DND_ENV_READY, DND_ENV_CLEANUP) != DND_ENV_READY) {
     dError("dnode env is already cleaned up");
     return;
   }
 
   monCleanup();
-  rpcCleanup();
   walCleanUp();
   taosStopCacheRefreshWorker();
-  dInfo("dnode env is cleaned up");
+  dDebug("dnode env is cleaned up");
 }
 
-void dndSetMsgHandle(SMgmtWrapper *pWrapper, int32_t msgType, NodeMsgFp nodeMsgFp, int32_t vgId) {
+void dndSetMsgHandle(SMgmtWrapper *pWrapper, tmsg_t msgType, NodeMsgFp nodeMsgFp, int8_t vgId) {
   pWrapper->msgFps[TMSG_INDEX(msgType)] = nodeMsgFp;
   pWrapper->msgVgIds[TMSG_INDEX(msgType)] = vgId;
 }
@@ -80,19 +72,6 @@ void dndSetStatus(SDnode *pDnode, EDndStatus status) {
   }
 }
 
-const char *dndStatStr(EDndStatus status) {
-  switch (status) {
-    case DND_STAT_INIT:
-      return "init";
-    case DND_STAT_RUNNING:
-      return "running";
-    case DND_STAT_STOPPED:
-      return "stopped";
-    default:
-      return "unknown";
-  }
-}
-
 void dndReportStartup(SDnode *pDnode, const char *pName, const char *pDesc) {
   SStartupReq *pStartup = &pDnode->startup;
   tstrncpy(pStartup->name, pName, TSDB_STEP_NAME_LEN);
@@ -105,29 +84,6 @@ void dndGetStartup(SDnode *pDnode, SStartupReq *pStartup) {
   pStartup->finished = (dndGetStatus(pDnode) == DND_STAT_RUNNING);
 }
 
-TdFilePtr dndCheckRunning(const char *dataDir) {
-  char filepath[PATH_MAX] = {0};
-  snprintf(filepath, sizeof(filepath), "%s/.running", dataDir);
-
-  TdFilePtr pFile = taosOpenFile(filepath, TD_FILE_CTEATE | TD_FILE_WRITE | TD_FILE_TRUNC);
-  if (pFile == NULL) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    dError("failed to lock file:%s since %s", filepath, terrstr());
-    return NULL;
-  }
-
-  int32_t ret = taosLockFile(pFile);
-  if (ret != 0) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    dError("failed to lock file:%s since %s", filepath, terrstr());
-    taosCloseFile(&pFile);
-    return NULL;
-  }
-
-  dDebug("file:%s is locked", filepath);
-  return pFile;
-}
-
 void dndProcessStartupReq(SDnode *pDnode, SRpcMsg *pReq) {
   dDebug("startup req is received");
   SStartupReq *pStartup = rpcMallocCont(sizeof(SStartupReq));

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

@@ -1,138 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "dndMain.h"
-
-static struct {
-  bool    dumpConfig;
-  bool    generateGrant;
-  bool    printAuth;
-  bool    printVersion;
-  char    envFile[PATH_MAX];
-  char    apolloUrl[PATH_MAX];
-  SDnode *pDnode;
-} global = {0};
-
-static void dndSigintHandle(int signum, void *info, void *ctx) {
-  dInfo("signal:%d is received", signum);
-  SDnode *pDnode = atomic_val_compare_exchange_ptr(&global.pDnode, 0, global.pDnode);
-  if (pDnode != NULL) {
-    dndHandleEvent(pDnode, DND_EVENT_STOP);
-  }
-}
-
-static void dndSetSignalHandle() {
-  taosSetSignal(SIGTERM, dndSigintHandle);
-  taosSetSignal(SIGHUP, dndSigintHandle);
-  taosSetSignal(SIGINT, dndSigintHandle);
-  taosSetSignal(SIGABRT, dndSigintHandle);
-  taosSetSignal(SIGBREAK, dndSigintHandle);
-}
-
-static int32_t dndParseOption(int32_t argc, char const *argv[]) {
-  for (int32_t i = 1; i < argc; ++i) {
-    if (strcmp(argv[i], "-c") == 0) {
-      if (i < argc - 1) {
-        if (strlen(argv[++i]) >= PATH_MAX) {
-          printf("config file path overflow");
-          return -1;
-        }
-        tstrncpy(configDir, argv[i], PATH_MAX);
-      } else {
-        printf("'-c' requires a parameter, default is %s\n", configDir);
-        return -1;
-      }
-    } else if (strcmp(argv[i], "-C") == 0) {
-      global.dumpConfig = true;
-    } else if (strcmp(argv[i], "-k") == 0) {
-      global.generateGrant = true;
-    } else if (strcmp(argv[i], "-V") == 0) {
-      global.printVersion = true;
-    } else {
-    }
-  }
-
-  return 0;
-}
-
-static int32_t dndRunDnode() {
-  if (dndInit() != 0) {
-    dInfo("failed to initialize dnode environment since %s", terrstr());
-    return -1;
-  }
-
-  SDnodeOpt option = dndGetOpt();
-
-  SDnode *pDnode = dndCreate(&option);
-  if (pDnode == NULL) {
-    dError("failed to to create dnode object since %s", terrstr());
-    return -1;
-  } else {
-    global.pDnode = pDnode;
-    dndSetSignalHandle();
-  }
-
-  dInfo("start the TDengine service");
-  int32_t code = dndRun(pDnode);
-  dInfo("start shutting down the TDengine service");
-
-  global.pDnode = NULL;
-  dndClose(pDnode);
-  dndCleanup();
-  taosCloseLog();
-  taosCleanupCfg();
-  return code;
-}
-
-int main(int argc, char const *argv[]) {
-  if (!taosCheckSystemIsSmallEnd()) {
-    dError("failed to start TDengine since on non-small-end machines");
-    return -1;
-  }
-
-  if (dndParseOption(argc, argv) != 0) {
-    return -1;
-  }
-
-  if (global.generateGrant) {
-    dndGenerateGrant();
-    return 0;
-  }
-
-  if (global.printVersion) {
-    dndPrintVersion();
-    return 0;
-  }
-
-  if (taosCreateLog("taosdlog", 1, configDir, global.envFile, global.apolloUrl, NULL, 0) != 0) {
-    dError("failed to start TDengine since read log config error");
-    return -1;
-  }
-
-  if (taosInitCfg(configDir, global.envFile, global.apolloUrl, NULL, 0) != 0) {
-    dError("failed to start TDengine since read config error");
-    return -1;
-  }
-
-  if (global.dumpConfig) {
-    dndDumpCfg();
-    taosCleanupCfg();
-    taosCloseLog();
-    return 0;
-  }
-
-  return dndRunDnode();
-}

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

@@ -42,6 +42,8 @@ static inline int32_t dndBuildMsg(SNodeMsg *pMsg, SRpcMsg *pRpc) {
   }
 
   memcpy(pMsg->user, connInfo.user, TSDB_USER_LEN);
+  pMsg->clientIp = connInfo.clientIp;
+  pMsg->clientPort = connInfo.clientPort;
   memcpy(&pMsg->rpcMsg, pRpc, sizeof(SRpcMsg));
   return 0;
 }
@@ -66,7 +68,7 @@ void dndProcessRpcMsg(SMgmtWrapper *pWrapper, SRpcMsg *pRpc, SEpSet *pEpSet) {
   } else if (pWrapper->procType == PROC_PARENT) {
     dTrace("msg:%p, is created and will put into child queue, handle:%p app:%p user:%s", pMsg, pRpc->handle,
            pRpc->ahandle, pMsg->user);
-    code = taosProcPutToChildQueue(pWrapper->pProc, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen);
+    code = taosProcPutToChildQ(pWrapper->pProc, pMsg, sizeof(SNodeMsg), pRpc->pCont, pRpc->contLen, PROC_REQ);
   } else {
     dTrace("msg:%p, should not processed in child process, handle:%p app:%p user:%s", pMsg, pRpc->handle, pRpc->ahandle,
            pMsg->user);

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

@@ -16,7 +16,7 @@
 #define _DEFAULT_SOURCE
 #include "dndInt.h"
 
-static int32_t dndInitMemory(SDnode *pDnode, const SDnodeOpt *pOption) {
+static int32_t dndInitVars(SDnode *pDnode, const SDnodeOpt *pOption) {
   pDnode->numOfSupportVnodes = pOption->numOfSupportVnodes;
   pDnode->serverPort = pOption->serverPort;
   pDnode->dataDir = strdup(pOption->dataDir);
@@ -24,8 +24,9 @@ static int32_t dndInitMemory(SDnode *pDnode, const SDnodeOpt *pOption) {
   pDnode->localFqdn = strdup(pOption->localFqdn);
   pDnode->firstEp = strdup(pOption->firstEp);
   pDnode->secondEp = strdup(pOption->secondEp);
-  pDnode->pDisks = pOption->pDisks;
+  pDnode->disks = pOption->disks;
   pDnode->numOfDisks = pOption->numOfDisks;
+  pDnode->ntype = pOption->ntype;
   pDnode->rebootTime = taosGetTimestampMs();
 
   if (pDnode->dataDir == NULL || pDnode->localEp == NULL || pDnode->localFqdn == NULL || pDnode->firstEp == NULL ||
@@ -36,16 +37,12 @@ static int32_t dndInitMemory(SDnode *pDnode, const SDnodeOpt *pOption) {
   return 0;
 }
 
-static void dndClearMemory(SDnode *pDnode) {
+static void dndClearVars(SDnode *pDnode) {
   for (ENodeType n = 0; n < NODE_MAX; ++n) {
     SMgmtWrapper *pMgmt = &pDnode->wrappers[n];
     taosMemoryFreeClear(pMgmt->path);
   }
-  if (pDnode->pLockFile != NULL) {
+  dndCloseRuntimeFile(pDnode);
-    taosUnLockFile(pDnode->pLockFile);
-    taosCloseFile(&pDnode->pLockFile);
-    pDnode->pLockFile = NULL;
-  }
   taosMemoryFreeClear(pDnode->localEp);
   taosMemoryFreeClear(pDnode->localFqdn);
   taosMemoryFreeClear(pDnode->firstEp);
@@ -67,13 +64,21 @@ SDnode *dndCreate(const SDnodeOpt *pOption) {
     goto _OVER;
   }
 
-  if (dndInitMemory(pDnode, pOption) != 0) {
+  if (dndInitVars(pDnode, pOption) != 0) {
+    dError("failed to init variables since %s", terrstr());
     goto _OVER;
   }
 
   dndSetStatus(pDnode, DND_STAT_INIT);
-  pDnode->pLockFile = dndCheckRunning(pDnode->dataDir);
+  dmGetMgmtFp(&pDnode->wrappers[DNODE]);
-  if (pDnode->pLockFile == NULL) {
+  mmGetMgmtFp(&pDnode->wrappers[MNODE]);
+  vmGetMgmtFp(&pDnode->wrappers[VNODES]);
+  qmGetMgmtFp(&pDnode->wrappers[QNODE]);
+  smGetMgmtFp(&pDnode->wrappers[SNODE]);
+  bmGetMgmtFp(&pDnode->wrappers[BNODE]);
+
+  if (dndOpenRuntimeFile(pDnode) != 0) {
+    dError("failed to open runtime file since %s", terrstr());
     goto _OVER;
   }
 
@@ -87,13 +92,6 @@ SDnode *dndCreate(const SDnodeOpt *pOption) {
     goto _OVER;
   }
 
-  dmGetMgmtFp(&pDnode->wrappers[DNODE]);
-  mmGetMgmtFp(&pDnode->wrappers[MNODE]);
-  vmGetMgmtFp(&pDnode->wrappers[VNODES]);
-  qmGetMgmtFp(&pDnode->wrappers[QNODE]);
-  smGetMgmtFp(&pDnode->wrappers[SNODE]);
-  bmGetMgmtFp(&pDnode->wrappers[BNODE]);
-
   if (dndInitMsgHandle(pDnode) != 0) {
     goto _OVER;
   }
@@ -116,7 +114,7 @@ SDnode *dndCreate(const SDnodeOpt *pOption) {
 
 _OVER:
   if (code != 0 && pDnode) {
-    dndClearMemory(pDnode);
+    dndClearVars(pDnode);
     pDnode = NULL;
     dError("failed to create dnode object since %s", terrstr());
   } else {
@@ -145,7 +143,7 @@ void dndClose(SDnode *pDnode) {
     dndCloseNode(pWrapper);
   }
 
-  dndClearMemory(pDnode);
+  dndClearVars(pDnode);
   dInfo("dnode object is closed, data:%p", pDnode);
 }
 

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

@@ -0,0 +1,64 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "dndInt.h"
+
+const char *dndStatStr(EDndStatus status) {
+  switch (status) {
+    case DND_STAT_INIT:
+      return "init";
+    case DND_STAT_RUNNING:
+      return "running";
+    case DND_STAT_STOPPED:
+      return "stopped";
+    default:
+      return "unknown";
+  }
+}
+
+const char *dndNodeLogStr(ENodeType ntype) {
+  switch (ntype) {
+    case VNODES:
+      return "vnode";
+    case QNODE:
+      return "qnode";
+    case SNODE:
+      return "snode";
+    case MNODE:
+      return "mnode";
+    case BNODE:
+      return "bnode";
+    default:
+      return "taosd";
+  }
+}
+
+const char *dndNodeProcStr(ENodeType ntype) {
+  switch (ntype) {
+    case VNODES:
+      return "taosv";
+    case QNODE:
+      return "taosq";
+    case SNODE:
+      return "taoss";
+    case MNODE:
+      return "taosm";
+    case BNODE:
+      return "taosb";
+    default:
+      return "taosd";
+  }
+}

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

@@ -309,7 +309,7 @@ int32_t dndInitMsgHandle(SDnode *pDnode) {
   return 0;
 }
 
-static int32_t dndSendRpcReq(STransMgmt *pMgmt, SEpSet *pEpSet, SRpcMsg *pReq) {
+static int32_t dndSendRpcReq(STransMgmt *pMgmt, const SEpSet *pEpSet, SRpcMsg *pReq) {
   if (pMgmt->clientRpc == NULL) {
     terrno = TSDB_CODE_DND_OFFLINE;
     return -1;
@@ -319,9 +319,8 @@ static int32_t dndSendRpcReq(STransMgmt *pMgmt, SEpSet *pEpSet, SRpcMsg *pReq) {
   return 0;
 }
 
-int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, SEpSet *pEpSet, SRpcMsg *pReq) {
+int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, const SEpSet *pEpSet, SRpcMsg *pReq) {
-  if (pWrapper->procType == PROC_CHILD) {
+  if (pWrapper->procType != PROC_CHILD) {
-  } else {
     SDnode *pDnode = pWrapper->pDnode;
     if (dndGetStatus(pDnode) != DND_STAT_RUNNING) {
       terrno = TSDB_CODE_DND_OFFLINE;
@@ -329,26 +328,27 @@ int32_t dndSendReqToDnode(SMgmtWrapper *pWrapper, SEpSet *pEpSet, SRpcMsg *pReq)
       return -1;
     }
     return dndSendRpcReq(&pDnode->trans, pEpSet, pReq);
+  } else {
+    while (taosProcPutToParentQ(pWrapper->pProc, pReq, sizeof(SRpcMsg), pReq->pCont, pReq->contLen, PROC_REQ) != 0) {
+      taosMsleep(1);
+    }
   }
 }
 
 int32_t dndSendReqToMnode(SMgmtWrapper *pWrapper, SRpcMsg *pReq) {
-  if (pWrapper->procType == PROC_CHILD) {
+  SDnode     *pDnode = pWrapper->pDnode;
-  } else {
+  STransMgmt *pTrans = &pDnode->trans;
-    SDnode     *pDnode = pWrapper->pDnode;
+  SEpSet      epSet = {0};
-    STransMgmt *pTrans = &pDnode->trans;
-    SEpSet      epSet = {0};
 
-    SMgmtWrapper *pWrapper = dndAcquireWrapper(pDnode, DNODE);
+  SMgmtWrapper *pWrapper2 = dndAcquireWrapper(pDnode, DNODE);
-    if (pWrapper != NULL) {
+  if (pWrapper2 != NULL) {
-      dmGetMnodeEpSet(pWrapper->pMgmt, &epSet);
+    dmGetMnodeEpSet(pWrapper2->pMgmt, &epSet);
-      dndReleaseWrapper(pWrapper);
+    dndReleaseWrapper(pWrapper2);
-    }
-    return dndSendRpcReq(pTrans, &epSet, pReq);
   }
+  return dndSendRpcReq(pTrans, &epSet, pReq);
 }
 
-static void dndSendRpcRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp) {
+void dndSendRpcRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
   if (pRsp->code == TSDB_CODE_APP_NOT_READY) {
     SMgmtWrapper *pDnodeWrapper = dndAcquireWrapper(pWrapper->pDnode, DNODE);
     if (pDnodeWrapper != NULL) {
@@ -362,16 +362,45 @@ static void dndSendRpcRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp) {
   }
 }
 
-void dndSendRsp(SMgmtWrapper *pWrapper, SRpcMsg *pRsp) {
+void dndSendRsp(SMgmtWrapper *pWrapper, const SRpcMsg *pRsp) {
-  if (pWrapper->procType == PROC_CHILD) {
+  if (pWrapper->procType != PROC_CHILD) {
-    int32_t code = -1;
-    do {
-      code = taosProcPutToParentQueue(pWrapper->pProc, pRsp, sizeof(SRpcMsg), pRsp->pCont, pRsp->contLen);
-      if (code != 0) {
-        taosMsleep(10);
-      }
-    } while (code != 0);
-  } else {
     dndSendRpcRsp(pWrapper, pRsp);
+  } else {
+    while (taosProcPutToParentQ(pWrapper->pProc, pRsp, sizeof(SRpcMsg), pRsp->pCont, pRsp->contLen, PROC_RSP) != 0) {
+      taosMsleep(1);
+    }
   }
 }
+
+void dndRegisterBrokenLinkArg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg) {
+  if (pWrapper->procType != PROC_CHILD) {
+    rpcRegisterBrokenLinkArg(pMsg);
+  } else {
+    while (taosProcPutToParentQ(pWrapper->pProc, pMsg, sizeof(SRpcMsg), pMsg->pCont, pMsg->contLen, PROC_REG) != 0) {
+      taosMsleep(1);
+    }
+  }
+}
+
+static void dndReleaseHandle(SMgmtWrapper *pWrapper, void *handle, int8_t type) {
+  if (pWrapper->procType != PROC_CHILD) {
+    rpcReleaseHandle(handle, type);
+  } else {
+    SRpcMsg msg = {.handle = handle, .code = type};
+    while (taosProcPutToParentQ(pWrapper->pProc, &msg, sizeof(SRpcMsg), NULL, 0, PROC_RELEASE) != 0) {
+      taosMsleep(1);
+    }
+  }
+}
+
+SMsgCb dndCreateMsgcb(SMgmtWrapper *pWrapper) {
+  SMsgCb msgCb = {
+      .pWrapper = pWrapper,
+      .registerBrokenLinkArgFp = dndRegisterBrokenLinkArg,
+      .releaseHandleFp = dndReleaseHandle,
+      .sendMnodeReqFp = dndSendReqToMnode,
+      .sendReqFp = dndSendReqToDnode,
+      .sendRspFp = dndSendRsp,
+  };
+  return msgCb;
+}

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

@@ -1,55 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "dndMain.h"
-
-void dndGenerateGrant() {
-#if 0
-  grantParseParameter();
-#endif
-}
-
-void dndPrintVersion() {
-#ifdef TD_ENTERPRISE
-  char *releaseName = "enterprise";
-#else
-  char *releaseName = "community";
-#endif
-  printf("%s version: %s compatible_version: %s\n", releaseName, version, compatible_version);
-  printf("gitinfo: %s\n", gitinfo);
-  printf("builuInfo: %s\n", buildinfo);
-}
-
-void dndDumpCfg() {
-  SConfig *pCfg = taosGetCfg();
-  cfgDumpCfg(pCfg, 0, 1);
-}
-
-SDnodeOpt dndGetOpt() {
-  SConfig  *pCfg = taosGetCfg();
-  SDnodeOpt option = {0};
-
-  option.numOfSupportVnodes = cfgGetItem(pCfg, "supportVnodes")->i32;
-  tstrncpy(option.dataDir, tsDataDir, sizeof(option.dataDir));
-  tstrncpy(option.firstEp, tsFirst, sizeof(option.firstEp));
-  tstrncpy(option.secondEp, tsSecond, sizeof(option.firstEp));
-  option.serverPort = tsServerPort;
-  tstrncpy(option.localFqdn, tsLocalFqdn, sizeof(option.localFqdn));
-  snprintf(option.localEp, sizeof(option.localEp), "%s:%u", option.localFqdn, option.serverPort);
-  option.pDisks = tsDiskCfg;
-  option.numOfDisks = tsDiskCfgNum;
-  return option;
-}

source/dnode/mgmt/mm/src/mmInt.c

@@ -52,6 +52,7 @@ static void mmInitOption(SMnodeMgmt *pMgmt, SMnodeOpt *pOption) {
   msgCb.sendReqFp = dndSendReqToDnode;
   msgCb.sendMnodeReqFp = dndSendReqToMnode;
   msgCb.sendRspFp = dndSendRsp;
+  msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
   pOption->msgCb = msgCb;
 }
 

source/dnode/mgmt/qm/src/qmInt.c

@@ -27,6 +27,7 @@ static void qmInitOption(SQnodeMgmt *pMgmt, SQnodeOpt *pOption) {
   msgCb.sendReqFp = dndSendReqToDnode;
   msgCb.sendMnodeReqFp = dndSendReqToMnode;
   msgCb.sendRspFp = dndSendRsp;
+  msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
   pOption->msgCb = msgCb;
 }
 

source/dnode/mgmt/sm/src/smInt.c

@@ -24,6 +24,7 @@ static void smInitOption(SSnodeMgmt *pMgmt, SSnodeOpt *pOption) {
   msgCb.sendReqFp = dndSendReqToDnode;
   msgCb.sendMnodeReqFp = dndSendReqToMnode;
   msgCb.sendRspFp = dndSendRsp;
+  msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
   pOption->msgCb = msgCb;
 }
 

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

@@ -137,6 +137,7 @@ static void *vmOpenVnodeFunc(void *param) {
     msgCb.sendReqFp = dndSendReqToDnode;
     msgCb.sendMnodeReqFp = dndSendReqToMnode;
     msgCb.sendRspFp = dndSendRsp;
+    msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
     SVnodeCfg cfg = {.msgCb = msgCb, .pTfs = pMgmt->pTfs, .vgId = pCfg->vgId, .dbId = pCfg->dbUid};
     SVnode   *pImpl = vnodeOpen(pCfg->path, &cfg);
     if (pImpl == NULL) {
@@ -284,7 +285,7 @@ static int32_t vmInit(SMgmtWrapper *pWrapper) {
   tstrncpy(dCfg.dir, pDnode->dataDir, TSDB_FILENAME_LEN);
   dCfg.level = 0;
   dCfg.primary = 1;
-  SDiskCfg *pDisks = pDnode->pDisks;
+  SDiskCfg *pDisks = pDnode->disks;
   int32_t   numOfDisks = pDnode->numOfDisks;
   if (numOfDisks <= 0 || pDisks == NULL) {
     pDisks = &dCfg;

source/dnode/mgmt/vm/src/vmMsg.c

@@ -91,6 +91,7 @@ int32_t vmProcessCreateVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pMsg) {
   msgCb.sendReqFp = dndSendReqToDnode;
   msgCb.sendMnodeReqFp = dndSendReqToMnode;
   msgCb.sendRspFp = dndSendRsp;
+  msgCb.registerBrokenLinkArgFp = dndRegisterBrokenLinkArg;
 
   vnodeCfg.msgCb = msgCb;
   vnodeCfg.pTfs = pMgmt->pTfs;
@@ -244,44 +245,45 @@ int32_t vmProcessCompactVnodeReq(SVnodesMgmt *pMgmt, SNodeMsg *pMsg) {
 
 void vmInitMsgHandles(SMgmtWrapper *pWrapper) {
   // Requests handled by VNODE
-  dndSetMsgHandle(pWrapper, TDMT_VND_SUBMIT, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_SUBMIT, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_QUERY, vmProcessQueryMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_QUERY, (NodeMsgFp)vmProcessQueryMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_QUERY_CONTINUE, vmProcessQueryMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_QUERY_CONTINUE, (NodeMsgFp)vmProcessQueryMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_FETCH, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_FETCH, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_FETCH_RSP, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_FETCH_RSP, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_ALTER_TABLE, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_ALTER_TABLE, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_UPDATE_TAG_VAL, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_UPDATE_TAG_VAL, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_TABLE_META, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TABLE_META, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_TABLES_META, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TABLES_META, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_CONSUME, vmProcessQueryMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_CONSUME, (NodeMsgFp)vmProcessQueryMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_QUERY, vmProcessQueryMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_QUERY, (NodeMsgFp)vmProcessQueryMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_CONNECT, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_CONNECT, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_DISCONNECT, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_DISCONNECT, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_SET_CUR, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_SET_CUR, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_RES_READY, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_RES_READY, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_TASKS_STATUS, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TASKS_STATUS, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_CANCEL_TASK, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_CANCEL_TASK, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_TASK, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_TASK, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_CREATE_STB, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_CREATE_STB, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_ALTER_STB, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_ALTER_STB, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_STB, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_STB, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_CREATE_TABLE, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_CREATE_TABLE, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_ALTER_TABLE, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_ALTER_TABLE, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_TABLE, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_TABLE, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_CREATE_SMA, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_CREATE_SMA, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_CANCEL_SMA, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_CANCEL_SMA, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_SMA, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_DROP_SMA, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_SHOW_TABLES, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_SHOW_TABLES, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_SHOW_TABLES_FETCH, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_SHOW_TABLES_FETCH, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_SET_CONN, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_SET_CONN, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_REB, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_REB, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_SET_CUR, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_MQ_SET_CUR, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_CONSUME, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_CONSUME, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_DEPLOY, vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_DEPLOY, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_QUERY_HEARTBEAT, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_QUERY_HEARTBEAT, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_PIPE_EXEC, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_PIPE_EXEC, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_MERGE_EXEC, vmProcessMergeMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_MERGE_EXEC, (NodeMsgFp)vmProcessMergeMsg, VND_VGID);
-  dndSetMsgHandle(pWrapper, TDMT_VND_STREAM_TRIGGER, vmProcessFetchMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_TASK_WRITE_EXEC, (NodeMsgFp)vmProcessWriteMsg, VND_VGID);
+  dndSetMsgHandle(pWrapper, TDMT_VND_STREAM_TRIGGER, (NodeMsgFp)vmProcessFetchMsg, VND_VGID);
 
   dndSetMsgHandle(pWrapper, TDMT_DND_CREATE_VNODE, vmProcessMgmtMsg, VND_VGID);
   dndSetMsgHandle(pWrapper, TDMT_DND_ALTER_VNODE, vmProcessMgmtMsg, VND_VGID);

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

@@ -76,7 +76,7 @@ static void vmProcessFetchQueue(SQueueInfo *pInfo, SNodeMsg *pMsg) {
   SVnodeObj *pVnode = pInfo->ahandle;
 
   dTrace("msg:%p, will be processed in vnode-fetch queue", pMsg);
-  int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, &pMsg->rpcMsg);
+  int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, &pMsg->rpcMsg, pInfo);
   if (code != 0) {
     vmSendRsp(pVnode->pWrapper, pMsg, code);
     dTrace("msg:%p, is freed, result:0x%04x:%s", pMsg, code & 0XFFFF, tstrerror(code));
@@ -168,7 +168,7 @@ static void vmProcessMergeQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
     taosGetQitem(qall, (void **)&pMsg);
 
     dTrace("msg:%p, will be processed in vnode-merge queue", pMsg);
-    int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, &pMsg->rpcMsg);
+    int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, &pMsg->rpcMsg, pInfo);
     if (code != 0) {
       vmSendRsp(pVnode->pWrapper, pMsg, code);
       dTrace("msg:%p, is freed, result:0x%04x:%s", pMsg, code & 0XFFFF, tstrerror(code));
@@ -414,8 +414,7 @@ int32_t vmStartWorker(SVnodesMgmt *pMgmt) {
   pWPool->max = maxMergeThreads;
   if (tWWorkerInit(pWPool) != 0) return -1;
 
-  SSingleWorkerCfg cfg = {
+  SSingleWorkerCfg cfg = {.min = 1, .max = 1, .name = "vnode-mgmt", .fp = (FItem)vmProcessMgmtQueue, .param = pMgmt};
-      .min = 1, .max = 1, .name = "vnode-mgmt", .fp = (FItem)vmProcessMgmtQueue, .param = pMgmt};
   if (tSingleWorkerInit(&pMgmt->mgmtWorker, &cfg) != 0) {
     dError("failed to start vnode-mgmt worker since %s", terrstr());
     return -1;

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

@@ -731,10 +731,10 @@ typedef struct {
   int32_t  vgNum;
   SRWLatch lock;
   int8_t   status;
-  int8_t   sourceType;
-  int8_t   sinkType;
   // int32_t  sqlLen;
-  int32_t        sinkVgId;  // 0 for automatic
+  int8_t         createdBy;      // STREAM_CREATED_BY__USER or SMA
+  int32_t        fixedSinkVgId;  // 0 for shuffle
+  int64_t        smaId;          // 0 for unused
   char*          sql;
   char*          logicalPlan;
   char*          physicalPlan;

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

@@ -27,7 +27,7 @@ void    mndCleanupScheduler(SMnode* pMnode);
 
 int32_t mndSchedInitSubEp(SMnode* pMnode, const SMqTopicObj* pTopic, SMqSubscribeObj* pSub);
 
-int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream, int64_t smaId);
+int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream);
 
 #ifdef __cplusplus
 }

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

@@ -31,7 +31,7 @@ void        mndReleaseStream(SMnode *pMnode, SStreamObj *pStream);
 SSdbRaw *mndStreamActionEncode(SStreamObj *pStream);
 SSdbRow *mndStreamActionDecode(SSdbRaw *pRaw);
 
-int32_t mndAddStreamToTrans(SMnode *pMnode, SStreamObj *pStream, const char *ast, STrans *pTrans, int64_t smaId);
+int32_t mndAddStreamToTrans(SMnode *pMnode, SStreamObj *pStream, const char *ast, STrans *pTrans);
 
 #ifdef __cplusplus
 }

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

@@ -26,8 +26,11 @@ int32_t tEncodeSStreamObj(SCoder *pEncoder, const SStreamObj *pObj) {
   if (tEncodeI64(pEncoder, pObj->dbUid) < 0) return -1;
   if (tEncodeI32(pEncoder, pObj->version) < 0) return -1;
   if (tEncodeI8(pEncoder, pObj->status) < 0) return -1;
+  if (tEncodeI8(pEncoder, pObj->createdBy) < 0) return -1;
+  if (tEncodeI32(pEncoder, pObj->fixedSinkVgId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pObj->smaId) < 0) return -1;
   if (tEncodeCStr(pEncoder, pObj->sql) < 0) return -1;
-  if (tEncodeCStr(pEncoder, pObj->logicalPlan) < 0) return -1;
+  /*if (tEncodeCStr(pEncoder, pObj->logicalPlan) < 0) return -1;*/
   if (tEncodeCStr(pEncoder, pObj->physicalPlan) < 0) return -1;
   // TODO encode tasks
   if (pObj->tasks) {
@@ -69,8 +72,11 @@ int32_t tDecodeSStreamObj(SCoder *pDecoder, SStreamObj *pObj) {
   if (tDecodeI64(pDecoder, &pObj->dbUid) < 0) return -1;
   if (tDecodeI32(pDecoder, &pObj->version) < 0) return -1;
   if (tDecodeI8(pDecoder, &pObj->status) < 0) return -1;
+  if (tDecodeI8(pDecoder, &pObj->createdBy) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pObj->fixedSinkVgId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pObj->smaId) < 0) return -1;
   if (tDecodeCStrAlloc(pDecoder, &pObj->sql) < 0) return -1;
-  if (tDecodeCStrAlloc(pDecoder, &pObj->logicalPlan) < 0) return -1;
+  /*if (tDecodeCStrAlloc(pDecoder, &pObj->logicalPlan) < 0) return -1;*/
   if (tDecodeCStrAlloc(pDecoder, &pObj->physicalPlan) < 0) return -1;
   pObj->tasks = NULL;
   int32_t sz;

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

@@ -20,138 +20,178 @@
 #define SYSTABLE_SCH_DB_NAME_LEN    ((TSDB_DB_NAME_LEN - 1) + VARSTR_HEADER_SIZE)
 
 //!!!! Note: only APPEND columns in below tables, NO insert !!!!
-static const SInfosTableSchema dnodesSchema[] = {{.name = "id",             .bytes = 2,   .type = TSDB_DATA_TYPE_SMALLINT},
+static const SInfosTableSchema dnodesSchema[] = {
-                                                 {.name = "endpoint",       .bytes = TSDB_EP_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "id", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "vnodes",         .bytes = 2,   .type = TSDB_DATA_TYPE_SMALLINT},
+    {.name = "endpoint", .bytes = TSDB_EP_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "max_vnodes",     .bytes = 2,   .type = TSDB_DATA_TYPE_SMALLINT},
+    {.name = "vnodes", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "status",         .bytes = 10 + VARSTR_HEADER_SIZE,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "max_vnodes", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "create_time",    .bytes = 8,   .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "status", .bytes = 10 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "note",           .bytes = 256 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                };
+    {.name = "note", .bytes = 256 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-static const SInfosTableSchema mnodesSchema[] = {{.name = "id",             .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+};
-                                                 {.name = "endpoint",       .bytes = TSDB_EP_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
+static const SInfosTableSchema mnodesSchema[] = {
-                                                 {.name = "role",           .bytes = 12 + VARSTR_HEADER_SIZE,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                 {.name = "role_time",      .bytes = 8,   .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "endpoint", .bytes = TSDB_EP_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "create_time",    .bytes = 8,   .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "role", .bytes = 12 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                };
+    {.name = "role_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-static const SInfosTableSchema modulesSchema[] = {{.name = "id",             .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                 {.name = "endpoint",        .bytes = 134, .type = TSDB_DATA_TYPE_BINARY},
+};
-                                                 {.name = "module",          .bytes = 10,  .type = TSDB_DATA_TYPE_BINARY},
+static const SInfosTableSchema modulesSchema[] = {
-                                                };
+    {.name = "id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-static const SInfosTableSchema qnodesSchema[] = {{.name = "id",             .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "endpoint", .bytes = 134, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "endpoint",       .bytes = 134, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "module", .bytes = 10, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "create_time",    .bytes = 8,   .type = TSDB_DATA_TYPE_TIMESTAMP},
+};
-                                                };
+static const SInfosTableSchema qnodesSchema[] = {
-static const SInfosTableSchema userDBSchema[] = {{.name = "name",             .bytes = SYSTABLE_SCH_DB_NAME_LEN,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                 {.name = "create_time",      .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "endpoint", .bytes = 134, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "vgroups",          .bytes = 2,    .type = TSDB_DATA_TYPE_SMALLINT},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                 {.name = "ntables",          .bytes = 8,    .type = TSDB_DATA_TYPE_BIGINT},
+};
-                                                 {.name = "replica",          .bytes = 2,    .type = TSDB_DATA_TYPE_SMALLINT},
+static const SInfosTableSchema userDBSchema[] = {
-                                                 {.name = "quorum",           .bytes = 2,    .type = TSDB_DATA_TYPE_SMALLINT},
+    {.name = "name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "days",             .bytes = 2,    .type = TSDB_DATA_TYPE_SMALLINT},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                 {.name = "keep",             .bytes = 24 + VARSTR_HEADER_SIZE,    .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "vgroups", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "cache",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "ntables", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
-                                                 {.name = "blocks",           .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "replica", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "minrows",          .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "quorum", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "maxrows",          .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "days", .bytes = 2, .type = TSDB_DATA_TYPE_SMALLINT},
-                                                 {.name = "wallevel",         .bytes = 1,    .type = TSDB_DATA_TYPE_TINYINT},
+    {.name = "keep", .bytes = 24 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                 {.name = "fsync",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "cache", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                 {.name = "comp",             .bytes = 1,    .type = TSDB_DATA_TYPE_TINYINT},
+    {.name = "blocks", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                 {.name = "cachelast",        .bytes = 1,    .type = TSDB_DATA_TYPE_TINYINT},
+    {.name = "minrows", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                 {.name = "precision",        .bytes = 3 + VARSTR_HEADER_SIZE,    .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "maxrows", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-//                                                 {.name = "update",           .bytes = 1,   .type = TSDB_DATA_TYPE_TINYINT},  // disable update
+    {.name = "wallevel", .bytes = 1, .type = TSDB_DATA_TYPE_TINYINT},
-                                                };
+    {.name = "fsync", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-static const SInfosTableSchema userFuncSchema[] = {{.name = "name",           .bytes = 32,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "comp", .bytes = 1, .type = TSDB_DATA_TYPE_TINYINT},
-                                                   {.name = "create_time",    .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "cachelast", .bytes = 1, .type = TSDB_DATA_TYPE_TINYINT},
-                                                   {.name = "ntables",        .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "precision", .bytes = 3 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "precision",      .bytes = 2,    .type = TSDB_DATA_TYPE_BINARY},
+    //                                                 {.name = "update",           .bytes = 1,   .type =
-                                                   {.name = "status",         .bytes = 10,   .type = TSDB_DATA_TYPE_BINARY},
+    //                                                 TSDB_DATA_TYPE_TINYINT},  // disable update
-                                                  };
+};
-static const SInfosTableSchema userIdxSchema[] = {{.name = "db_name",          .bytes = 32,   .type = TSDB_DATA_TYPE_BINARY},
+static const SInfosTableSchema userFuncSchema[] = {
-                                                  {.name = "table_name",       .bytes = SYSTABLE_SCH_DB_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "name", .bytes = 32, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "index_database",   .bytes = 32,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                  {.name = "index_name",       .bytes = SYSTABLE_SCH_DB_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "ntables", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "column_name",      .bytes = 64,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "precision", .bytes = 2, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "index_type",       .bytes = 10,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "status", .bytes = 10, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "index_extensions", .bytes = 256,  .type = TSDB_DATA_TYPE_BINARY},
+};
-                                                 };
+static const SInfosTableSchema userIdxSchema[] = {
-static const SInfosTableSchema userStbsSchema[] = {{.name = "stable_name",     .bytes = SYSTABLE_SCH_TABLE_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "db_name", .bytes = 32, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "db_name",         .bytes = SYSTABLE_SCH_DB_NAME_LEN,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "table_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "create_time",     .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "index_database", .bytes = 32, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "columns",         .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "index_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "tags",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "column_name", .bytes = 64, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "tables",          .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "index_type", .bytes = 10, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "last_update",     .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "index_extensions", .bytes = 256, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "table_comment",   .bytes = 1024 + VARSTR_HEADER_SIZE,    .type = TSDB_DATA_TYPE_INT},
+};
-                                                  };
+static const SInfosTableSchema userStbsSchema[] = {
-static const SInfosTableSchema userStreamsSchema[] = {{.name = "stream_name",  .bytes = SYSTABLE_SCH_DB_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "stable_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                      {.name = "user_name",    .bytes = 23,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "db_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                      {.name = "dest_table",   .bytes = SYSTABLE_SCH_DB_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                      {.name = "create_time",  .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "columns", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                      {.name = "sql",          .bytes = 1024, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "tags", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                     };
+    {.name = "tables", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "last_update", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "table_comment", .bytes = 1024 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_INT},
+};
+static const SInfosTableSchema userStreamsSchema[] = {
+    {.name = "stream_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "user_name", .bytes = 23, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "dest_table", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "sql", .bytes = 1024, .type = TSDB_DATA_TYPE_BINARY},
+};
 static const SInfosTableSchema userTblsSchema[] = {
-                                                   {.name = "table_name",      .bytes = SYSTABLE_SCH_TABLE_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "table_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "db_name",         .bytes = SYSTABLE_SCH_DB_NAME_LEN,   .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "db_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "create_time",     .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                                   {.name = "columns",         .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "columns", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                   {.name = "stable_name",     .bytes = SYSTABLE_SCH_TABLE_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "stable_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                   {.name = "uid",             .bytes = 8,    .type = TSDB_DATA_TYPE_BIGINT},
+    {.name = "uid", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
-                                                   {.name = "vgroup_id",       .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                   {.name = "ttl",             .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "ttl", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                   {.name = "table_comment",    .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "table_comment", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  };
+};
-static const SInfosTableSchema userTblDistSchema[] = {{.name = "db_name",                .bytes = 32,   .type = TSDB_DATA_TYPE_BINARY},
+static const SInfosTableSchema userTblDistSchema[] = {
-                                                      {.name = "table_name",             .bytes = SYSTABLE_SCH_DB_NAME_LEN,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "db_name", .bytes = 32, .type = TSDB_DATA_TYPE_BINARY},
-                                                      {.name = "distributed_histogram",  .bytes = 500,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "table_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                      {.name = "min_of_rows",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "distributed_histogram", .bytes = 500, .type = TSDB_DATA_TYPE_BINARY},
-                                                      {.name = "max_of_rows",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "min_of_rows", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                      {.name = "avg_of_rows",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "max_of_rows", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                      {.name = "stddev_of_rows",         .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "avg_of_rows", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                      {.name = "rows",                   .bytes = 8,    .type = TSDB_DATA_TYPE_BIGINT},
+    {.name = "stddev_of_rows", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                      {.name = "blocks",                 .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "rows", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
-                                                      {.name = "storage_size",           .bytes = 8,    .type = TSDB_DATA_TYPE_BIGINT},
+    {.name = "blocks", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                      {.name = "compression_ratio",      .bytes = 8,    .type = TSDB_DATA_TYPE_DOUBLE},
+    {.name = "storage_size", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
-                                                      {.name = "rows_in_mem",            .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "compression_ratio", .bytes = 8, .type = TSDB_DATA_TYPE_DOUBLE},
-                                                      {.name = "seek_header_time",       .bytes = 4,    .type = TSDB_DATA_TYPE_INT},
+    {.name = "rows_in_mem", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                     };
+    {.name = "seek_header_time", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-static const SInfosTableSchema userUsersSchema[] = {{.name = "name",           .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE,   .type = TSDB_DATA_TYPE_BINARY},
+};
-                                                    {.name = "privilege",      .bytes = 10 + VARSTR_HEADER_SIZE,  .type = TSDB_DATA_TYPE_BINARY},
+static const SInfosTableSchema userUsersSchema[] = {
-                                                    {.name = "create_time",    .bytes = 8,    .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "name", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                    {.name = "account",        .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE,    .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "privilege", .bytes = 10 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                   };
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-static const SInfosTableSchema vgroupsSchema[] = {{.name = "vgroup_id",        .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "account", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "db_name",          .bytes = SYSTABLE_SCH_DB_NAME_LEN,   .type = TSDB_DATA_TYPE_BINARY},
+};
-                                                  {.name = "tables",           .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+static const SInfosTableSchema vgroupsSchema[] = {
-                                                  {.name = "status",           .bytes = 12 + VARSTR_HEADER_SIZE,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "onlines",          .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "db_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "v1_dnode",         .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "tables", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "v1_status",        .bytes = 10,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "status", .bytes = 12 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "v2_dnode",         .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "onlines", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "v2_status",        .bytes = 10,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "v1_dnode", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "v3_dnode",         .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "v1_status", .bytes = 10, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "v3_status",        .bytes = 10,  .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "v2_dnode", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "compacting",       .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "v2_status", .bytes = 10, .type = TSDB_DATA_TYPE_BINARY},
-                                                  {.name = "nfiles",           .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "v3_dnode", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                                  {.name = "file_size",        .bytes = 4,   .type = TSDB_DATA_TYPE_INT},
+    {.name = "v3_status", .bytes = 10, .type = TSDB_DATA_TYPE_BINARY},
-                                                 };
+    {.name = "compacting", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "nfiles", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "file_size", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+};
 
-static const SInfosTableMeta infosMeta[] = {{TSDB_INS_TABLE_DNODES, dnodesSchema, tListLen(dnodesSchema)},
+// TODO put into perf schema
-                                            {TSDB_INS_TABLE_MNODES, mnodesSchema, tListLen(mnodesSchema)},
+static const SInfosTableSchema topicSchema[] = {
-                                            {TSDB_INS_TABLE_MODULES, modulesSchema, tListLen(modulesSchema)},
+    {.name = "topic_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                            {TSDB_INS_TABLE_QNODES, qnodesSchema, tListLen(qnodesSchema)},
+    {.name = "db_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-                                            {TSDB_INS_TABLE_USER_DATABASES, userDBSchema, tListLen(userDBSchema)},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-                                            {TSDB_INS_TABLE_USER_FUNCTIONS, userFuncSchema, tListLen(userFuncSchema)},
+    {.name = "sql", .bytes = 1024, .type = TSDB_DATA_TYPE_BINARY},
-                                            {TSDB_INS_TABLE_USER_INDEXES, userIdxSchema, tListLen(userIdxSchema)},
+    {.name = "row_len", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-                                            {TSDB_INS_TABLE_USER_STABLES, userStbsSchema, tListLen(userStbsSchema)},
+};
-                                            {TSDB_INS_TABLE_USER_STREAMS, userStreamsSchema, tListLen(userStreamsSchema)},
-                                            {TSDB_INS_TABLE_USER_TABLES, userTblsSchema, tListLen(userTblsSchema)},
-                                            {TSDB_INS_TABLE_USER_TABLE_DISTRIBUTED, userTblDistSchema, tListLen(userTblDistSchema)},
-                                            {TSDB_INS_TABLE_USER_USERS, userUsersSchema, tListLen(userUsersSchema)},
-                                            {TSDB_INS_TABLE_VGROUPS, vgroupsSchema, tListLen(vgroupsSchema)},
-                                 };
 
-//connection/application/
+static const SInfosTableSchema consumerSchema[] = {
+    {.name = "client_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "pid", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "status", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    // ep
+    // up time
+    // topics
+};
+
+static const SInfosTableSchema subscribeSchema[] = {
+    {.name = "topic_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "client_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+};
+
+static const SInfosTableMeta infosMeta[] = {
+    {TSDB_INS_TABLE_DNODES, dnodesSchema, tListLen(dnodesSchema)},
+    {TSDB_INS_TABLE_MNODES, mnodesSchema, tListLen(mnodesSchema)},
+    {TSDB_INS_TABLE_MODULES, modulesSchema, tListLen(modulesSchema)},
+    {TSDB_INS_TABLE_QNODES, qnodesSchema, tListLen(qnodesSchema)},
+    {TSDB_INS_TABLE_USER_DATABASES, userDBSchema, tListLen(userDBSchema)},
+    {TSDB_INS_TABLE_USER_FUNCTIONS, userFuncSchema, tListLen(userFuncSchema)},
+    {TSDB_INS_TABLE_USER_INDEXES, userIdxSchema, tListLen(userIdxSchema)},
+    {TSDB_INS_TABLE_USER_STABLES, userStbsSchema, tListLen(userStbsSchema)},
+    {TSDB_INS_TABLE_USER_STREAMS, userStreamsSchema, tListLen(userStreamsSchema)},
+    {TSDB_INS_TABLE_USER_TABLES, userTblsSchema, tListLen(userTblsSchema)},
+    {TSDB_INS_TABLE_USER_TABLE_DISTRIBUTED, userTblDistSchema, tListLen(userTblDistSchema)},
+    {TSDB_INS_TABLE_USER_USERS, userUsersSchema, tListLen(userUsersSchema)},
+    {TSDB_INS_TABLE_VGROUPS, vgroupsSchema, tListLen(vgroupsSchema)},
+};
+
+// connection/application/
 int32_t mndInitInfosTableSchema(const SInfosTableSchema *pSrc, int32_t colNum, SSchema **pDst) {
   SSchema *schema = taosMemoryCalloc(colNum, sizeof(SSchema));
   if (NULL == schema) {
@@ -161,7 +201,7 @@ int32_t mndInitInfosTableSchema(const SInfosTableSchema *pSrc, int32_t colNum, S
 
   for (int32_t i = 0; i < colNum; ++i) {
     strcpy(schema[i].name, pSrc[i].name);
-    
+
     schema[i].type = pSrc[i].type;
     schema[i].colId = i + 1;
     schema[i].bytes = pSrc[i].bytes;
@@ -182,11 +222,11 @@ int32_t mndInsInitMeta(SHashObj *hash) {
   for (int32_t i = 0; i < tListLen(infosMeta); ++i) {
     strcpy(meta.tbName, infosMeta[i].name);
     meta.numOfColumns = infosMeta[i].colNum;
-    
+
     if (mndInitInfosTableSchema(infosMeta[i].schema, infosMeta[i].colNum, &meta.pSchemas)) {
       return -1;
     }
-    
+
     if (taosHashPut(hash, meta.tbName, strlen(meta.tbName), &meta, sizeof(meta))) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
@@ -210,7 +250,7 @@ int32_t mndBuildInsTableSchema(SMnode *pMnode, const char *dbFName, const char *
   }
 
   *pRsp = *meta;
-  
+
   pRsp->pSchemas = taosMemoryCalloc(meta->numOfColumns, sizeof(SSchema));
   if (pRsp->pSchemas == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
@@ -237,13 +277,13 @@ void mndCleanupInfos(SMnode *pMnode) {
   if (NULL == pMnode->infosMeta) {
     return;
   }
-  
+
   void *pIter = taosHashIterate(pMnode->infosMeta, NULL);
   while (pIter) {
     STableMetaRsp *meta = (STableMetaRsp *)pIter;
 
     taosMemoryFreeClear(meta->pSchemas);
-    
+
     pIter = taosHashIterate(pMnode->infosMeta, pIter);
   }
 
@@ -251,5 +291,3 @@ void mndCleanupInfos(SMnode *pMnode) {
   pMnode->infosMeta = NULL;
 }
 
-
-

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

@@ -44,7 +44,8 @@ typedef struct {
   SQueryDesc *pQueries;
 } SConnObj;
 
-static SConnObj *mndCreateConn(SMnode *pMnode, SRpcConnInfo *pInfo, int32_t pid, const char *app, int64_t startTime);
+static SConnObj *mndCreateConn(SMnode *pMnode, const char *user, uint32_t ip, uint16_t port, int32_t pid,
+                               const char *app, int64_t startTime);
 static void      mndFreeConn(SConnObj *pConn);
 static SConnObj *mndAcquireConn(SMnode *pMnode, int32_t connId);
 static void      mndReleaseConn(SMnode *pMnode, SConnObj *pConn);
@@ -94,7 +95,8 @@ void mndCleanupProfile(SMnode *pMnode) {
   }
 }
 
-static SConnObj *mndCreateConn(SMnode *pMnode, SRpcConnInfo *pInfo, int32_t pid, const char *app, int64_t startTime) {
+static SConnObj *mndCreateConn(SMnode *pMnode, const char *user, uint32_t ip, uint16_t port, int32_t pid,
+                               const char *app, int64_t startTime) {
   SProfileMgmt *pMgmt = &pMnode->profileMgmt;
 
   int32_t connId = atomic_add_fetch_32(&pMgmt->connId, 1);
@@ -104,8 +106,8 @@ static SConnObj *mndCreateConn(SMnode *pMnode, SRpcConnInfo *pInfo, int32_t pid,
   SConnObj connObj = {.id = connId,
                       .appStartTimeMs = startTime,
                       .pid = pid,
-                      .ip = pInfo->clientIp,
+                      .ip = ip,
-                      .port = pInfo->clientPort,
+                      .port = port,
                       .killed = 0,
                       .loginTimeMs = taosGetTimestampMs(),
                       .lastAccessTimeMs = 0,
@@ -114,17 +116,17 @@ static SConnObj *mndCreateConn(SMnode *pMnode, SRpcConnInfo *pInfo, int32_t pid,
                       .pQueries = NULL};
 
   connObj.lastAccessTimeMs = connObj.loginTimeMs;
-  tstrncpy(connObj.user, pInfo->user, TSDB_USER_LEN);
+  tstrncpy(connObj.user, user, TSDB_USER_LEN);
   tstrncpy(connObj.app, app, TSDB_APP_NAME_LEN);
 
   int32_t   keepTime = tsShellActivityTimer * 3;
   SConnObj *pConn = taosCachePut(pMgmt->cache, &connId, sizeof(int32_t), &connObj, sizeof(connObj), keepTime * 1000);
   if (pConn == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
-    mError("conn:%d, failed to put into cache since %s, user:%s", connId, pInfo->user, terrstr());
+    mError("conn:%d, failed to put into cache since %s, user:%s", connId, user, terrstr());
     return NULL;
   } else {
-    mTrace("conn:%d, is created, data:%p user:%s", pConn->id, pConn, pInfo->user);
+    mTrace("conn:%d, is created, data:%p user:%s", pConn->id, pConn, user);
     return pConn;
   }
 }
@@ -184,20 +186,14 @@ static int32_t mndProcessConnectReq(SNodeMsg *pReq) {
   SConnObj   *pConn = NULL;
   int32_t     code = -1;
   SConnectReq connReq = {0};
+  char        ip[30] = {0};
 
   if (tDeserializeSConnectReq(pReq->rpcMsg.pCont, pReq->rpcMsg.contLen, &connReq) != 0) {
     terrno = TSDB_CODE_INVALID_MSG;
     goto CONN_OVER;
   }
 
-  SRpcConnInfo info = {0};
+  taosIp2String(pReq->clientIp, ip);
-  if (rpcGetConnInfo(pReq->rpcMsg.handle, &info) != 0) {
-    mError("user:%s, failed to login while get connection info since %s", pReq->user, terrstr());
-    goto CONN_OVER;
-  }
-
-  char ip[30];
-  taosIp2String(info.clientIp, ip);
 
   pUser = mndAcquireUser(pMnode, pReq->user);
   if (pUser == NULL) {
@@ -216,7 +212,8 @@ static int32_t mndProcessConnectReq(SNodeMsg *pReq) {
     }
   }
 
-  pConn = mndCreateConn(pMnode, &info, connReq.pid, connReq.app, connReq.startTime);
+  pConn =
+      mndCreateConn(pMnode, pReq->user, pReq->clientIp, pReq->clientPort, connReq.pid, connReq.app, connReq.startTime);
   if (pConn == NULL) {
     mError("user:%s, failed to login from %s while create connection since %s", pReq->user, ip, terrstr());
     goto CONN_OVER;
@@ -241,7 +238,7 @@ static int32_t mndProcessConnectReq(SNodeMsg *pReq) {
   pReq->rspLen = contLen;
   pReq->pRsp = pRsp;
 
-  mDebug("user:%s, login from %s, conn:%d, app:%s", info.user, ip, pConn->id, connReq.app);
+  mDebug("user:%s, login from %s, conn:%d, app:%s", pReq->user, ip, pConn->id, connReq.app);
 
   code = 0;
 

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

@@ -14,15 +14,15 @@
  */
 
 #include "mndQuery.h"
-#include "mndMnode.h"
 #include "executor.h"
+#include "mndMnode.h"
 #include "qworker.h"
 
 int32_t mndProcessQueryMsg(SNodeMsg *pReq) {
-  mTrace("message in query queue is processing");
+  SMnode     *pMnode = pReq->pNode;
-  SMnode *pMnode = pReq->pNode;
   SReadHandle handle = {0};
 
+  mTrace("msg:%p, in query queue is processing", pReq);
   switch (pReq->rpcMsg.msgType) {
     case TDMT_VND_QUERY:
       return qWorkerProcessQueryMsg(&handle, pMnode->pQuery, &pReq->rpcMsg);
@@ -35,9 +35,9 @@ int32_t mndProcessQueryMsg(SNodeMsg *pReq) {
 }
 
 int32_t mndProcessFetchMsg(SNodeMsg *pReq) {
-  mTrace("message in fetch queue is processing");
   SMnode *pMnode = pReq->pNode;
-  
+  mTrace("msg:%p, in fetch queue is processing", pReq);
+
   switch (pReq->rpcMsg.msgType) {
     case TDMT_VND_FETCH:
       return qWorkerProcessFetchMsg(pMnode, pMnode->pQuery, &pReq->rpcMsg);
@@ -52,9 +52,9 @@ int32_t mndProcessFetchMsg(SNodeMsg *pReq) {
 }
 
 int32_t mndInitQuery(SMnode *pMnode) {
-  int32_t code = qWorkerInit(NODE_TYPE_MNODE, MND_VGID, NULL, (void **)&pMnode->pQuery, &pMnode->msgCb);
+  if (qWorkerInit(NODE_TYPE_MNODE, MND_VGID, NULL, (void **)&pMnode->pQuery, &pMnode->msgCb) != 0) {
-  if (code) {
+    mError("failed to init qworker in mnode since %s", terrstr());
-    return code;
+    return -1;
   }
 
   mndSetMsgHandle(pMnode, TDMT_VND_QUERY, mndProcessQueryMsg);
@@ -67,4 +67,3 @@ int32_t mndInitQuery(SMnode *pMnode) {
 }
 
 void mndCleanupQuery(SMnode *pMnode) { qWorkerDestroy((void **)&pMnode->pQuery); }
-

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

@@ -119,7 +119,92 @@ SVgObj* mndSchedFetchOneVg(SMnode* pMnode, int64_t dbUid) {
   return pVgroup;
 }
 
-int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream, int64_t smaId) {
+int32_t mndAddShuffledSinkToStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream) {
+  SSdb*   pSdb = pMnode->pSdb;
+  void*   pIter = NULL;
+  SArray* tasks = taosArrayGetP(pStream->tasks, 0);
+
+  ASSERT(taosArrayGetSize(pStream->tasks) == 1);
+
+  while (1) {
+    SVgObj* pVgroup;
+    pIter = sdbFetch(pSdb, SDB_VGROUP, pIter, (void**)&pVgroup);
+    if (pIter == NULL) break;
+    if (pVgroup->dbUid != pStream->dbUid) {
+      sdbRelease(pSdb, pVgroup);
+      continue;
+    }
+    SStreamTask* pTask = tNewSStreamTask(pStream->uid);
+    if (pTask == NULL) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      return -1;
+    }
+    taosArrayPush(tasks, &pTask);
+
+    pTask->nodeId = pVgroup->vgId;
+    pTask->epSet = mndGetVgroupEpset(pMnode, pVgroup);
+
+    // source
+    pTask->sourceType = TASK_SOURCE__MERGE;
+
+    // exec
+    pTask->execType = TASK_EXEC__NONE;
+
+    // sink
+    if (pStream->createdBy == STREAM_CREATED_BY__SMA) {
+      pTask->sinkType = TASK_SINK__SMA;
+      pTask->smaSink.smaId = pStream->smaId;
+    } else {
+      pTask->sinkType = TASK_SINK__TABLE;
+    }
+
+    // dispatch
+    pTask->dispatchType = TASK_DISPATCH__NONE;
+
+    mndPersistTaskDeployReq(pTrans, pTask, &pTask->epSet, TDMT_VND_TASK_DEPLOY, pVgroup->vgId);
+  }
+  return 0;
+}
+
+int32_t mndAddFixedSinkToStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream) {
+  ASSERT(pStream->fixedSinkVgId != 0);
+  SArray*      tasks = taosArrayGetP(pStream->tasks, 0);
+  SStreamTask* pTask = tNewSStreamTask(pStream->uid);
+  if (pTask == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  taosArrayPush(tasks, &pTask);
+
+  pTask->nodeId = pStream->fixedSinkVgId;
+  SVgObj* pVgroup = mndAcquireVgroup(pMnode, pStream->fixedSinkVgId);
+  if (pVgroup == NULL) {
+    return -1;
+  }
+  pTask->epSet = mndGetVgroupEpset(pMnode, pVgroup);
+  // source
+  pTask->sourceType = TASK_SOURCE__MERGE;
+
+  // exec
+  pTask->execType = TASK_EXEC__NONE;
+
+  // sink
+  if (pStream->createdBy == STREAM_CREATED_BY__SMA) {
+    pTask->sinkType = TASK_SINK__SMA;
+    pTask->smaSink.smaId = pStream->smaId;
+  } else {
+    pTask->sinkType = TASK_SINK__TABLE;
+  }
+  //
+  // dispatch
+  pTask->dispatchType = TASK_DISPATCH__NONE;
+
+  mndPersistTaskDeployReq(pTrans, pTask, &pTask->epSet, TDMT_VND_TASK_DEPLOY, pVgroup->vgId);
+
+  return 0;
+}
+
+int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream) {
   SSdb*       pSdb = pMnode->pSdb;
   SQueryPlan* pPlan = qStringToQueryPlan(pStream->physicalPlan);
   if (pPlan == NULL) {
@@ -132,6 +217,19 @@ int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream, i
   ASSERT(totLevel <= 2);
   pStream->tasks = taosArrayInit(totLevel, sizeof(void*));
 
+  bool hasExtraSink = false;
+  if (totLevel == 2) {
+    SArray* taskOneLevel = taosArrayInit(0, sizeof(void*));
+    taosArrayPush(pStream->tasks, &taskOneLevel);
+    // add extra sink
+    hasExtraSink = true;
+    if (pStream->fixedSinkVgId == 0) {
+      mndAddShuffledSinkToStream(pMnode, pTrans, pStream);
+    } else {
+      mndAddFixedSinkToStream(pMnode, pTrans, pStream);
+    }
+  }
+
   for (int32_t level = 0; level < totLevel; level++) {
     SArray*        taskOneLevel = taosArrayInit(0, sizeof(void*));
     SNodeListNode* inner = nodesListGetNode(pPlan->pSubplans, level);
@@ -164,9 +262,15 @@ int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream, i
           // only for inplace
           pTask->sinkType = TASK_SINK__SHOW;
           pTask->showSink.reserved = 0;
-          if (smaId != -1) {
+          if (!hasExtraSink) {
-            pTask->sinkType = TASK_SINK__SMA;
+#if 1
-            pTask->smaSink.smaId = smaId;
+            if (pStream->createdBy == STREAM_CREATED_BY__SMA) {
+              pTask->sinkType = TASK_SINK__SMA;
+              pTask->smaSink.smaId = pStream->smaId;
+            } else {
+              pTask->sinkType = TASK_SINK__TABLE;
+            }
+#endif
           }
         } else {
           pTask->sinkType = TASK_SINK__NONE;
@@ -175,17 +279,15 @@ int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream, i
         // dispatch part
         if (level == 0) {
           pTask->dispatchType = TASK_DISPATCH__NONE;
-          // if inplace sink, no dispatcher
-          // if fixed ep, add fixed ep dispatcher
-          // if shuffle, add shuffle dispatcher
         } else {
           // add fixed ep dispatcher
           int32_t lastLevel = level - 1;
           ASSERT(lastLevel == 0);
+          if (hasExtraSink) lastLevel++;
           SArray* pArray = taosArrayGetP(pStream->tasks, lastLevel);
           // one merge only
           ASSERT(taosArrayGetSize(pArray) == 1);
-          SStreamTask* lastLevelTask = taosArrayGetP(pArray, lastLevel);
+          SStreamTask* lastLevelTask = taosArrayGetP(pArray, 0);
           pTask->dispatchMsgType = TDMT_VND_TASK_MERGE_EXEC;
           pTask->dispatchType = TASK_DISPATCH__FIXED;
 
@@ -222,18 +324,55 @@ int32_t mndScheduleStream(SMnode* pMnode, STrans* pTrans, SStreamObj* pStream, i
       /*pTask->sinkType = TASK_SINK__NONE;*/
 
       // dispatch part
-      pTask->dispatchType = TASK_DISPATCH__NONE;
+      ASSERT(hasExtraSink);
-#if 0
+      /*pTask->dispatchType = TASK_DISPATCH__NONE;*/
-      pTask->dispatchType = TASK_DISPATCH__SHUFFLE;
+#if 1
-      pTask->dispatchMsgType = TDMT_VND_TASK_WRITE_EXEC;
+
-      SDbObj* pDb = mndAcquireDb(pMnode, pStream->db);
+      if (hasExtraSink) {
-      ASSERT(pDb);
+        // add dispatcher
-      if (mndExtractDbInfo(pMnode, pDb, &pTask->shuffleDispatcher.dbInfo, NULL) < 0) {
+        if (pStream->fixedSinkVgId == 0) {
-        sdbRelease(pSdb, pDb);
+          pTask->dispatchType = TASK_DISPATCH__SHUFFLE;
-        qDestroyQueryPlan(pPlan);
+
-        return -1;
+          pTask->dispatchMsgType = TDMT_VND_TASK_WRITE_EXEC;
+          SDbObj* pDb = mndAcquireDb(pMnode, pStream->db);
+          ASSERT(pDb);
+          if (mndExtractDbInfo(pMnode, pDb, &pTask->shuffleDispatcher.dbInfo, NULL) < 0) {
+            sdbRelease(pSdb, pDb);
+            qDestroyQueryPlan(pPlan);
+            return -1;
+          }
+          sdbRelease(pSdb, pDb);
+
+          // put taskId to useDbRsp
+          // TODO: optimize
+          SArray* pVgs = pTask->shuffleDispatcher.dbInfo.pVgroupInfos;
+          int32_t sz = taosArrayGetSize(pVgs);
+          SArray* sinkLv = taosArrayGetP(pStream->tasks, 0);
+          int32_t sinkLvSize = taosArrayGetSize(sinkLv);
+          for (int32_t i = 0; i < sz; i++) {
+            SVgroupInfo* pVgInfo = taosArrayGet(pVgs, i);
+            for (int32_t j = 0; j < sinkLvSize; j++) {
+              SStreamTask* pLastLevelTask = taosArrayGetP(sinkLv, j);
+              /*printf("vgid %d node id %d\n", pVgInfo->vgId, pTask->nodeId);*/
+              if (pLastLevelTask->nodeId == pVgInfo->vgId) {
+                pVgInfo->taskId = pLastLevelTask->taskId;
+                /*printf("taskid %d set to %d\n", pVgInfo->taskId, pTask->taskId);*/
+                break;
+              }
+            }
+          }
+        } else {
+          pTask->dispatchType = TASK_DISPATCH__FIXED;
+          pTask->dispatchMsgType = TDMT_VND_TASK_WRITE_EXEC;
+          SArray* pArray = taosArrayGetP(pStream->tasks, 0);
+          // one sink only
+          ASSERT(taosArrayGetSize(pArray) == 1);
+          SStreamTask* lastLevelTask = taosArrayGetP(pArray, 0);
+          pTask->fixedEpDispatcher.taskId = lastLevelTask->taskId;
+          pTask->fixedEpDispatcher.nodeId = lastLevelTask->nodeId;
+          pTask->fixedEpDispatcher.epSet = lastLevelTask->epSet;
+        }
       }
-      sdbRelease(pSdb, pDb);
 #endif
 
       // exec part

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

@@ -415,6 +415,10 @@ static int32_t mndCreateSma(SMnode *pMnode, SNodeMsg *pReq, SMCreateSmaReq *pCre
   streamObj.dbUid = pDb->uid;
   streamObj.version = 1;
   streamObj.sql = pCreate->sql;
+  streamObj.createdBy = STREAM_CREATED_BY__SMA;
+  // TODO
+  streamObj.fixedSinkVgId = 0;
+  streamObj.smaId = smaObj.uid;
   /*streamObj.physicalPlan = "";*/
   streamObj.logicalPlan = "not implemented";
 
@@ -428,7 +432,7 @@ static int32_t mndCreateSma(SMnode *pMnode, SNodeMsg *pReq, SMCreateSmaReq *pCre
   if (mndSetCreateSmaRedoLogs(pMnode, pTrans, &smaObj) != 0) goto _OVER;
   if (mndSetCreateSmaCommitLogs(pMnode, pTrans, &smaObj) != 0) goto _OVER;
   if (mndSetCreateSmaRedoActions(pMnode, pTrans, pDb, &smaObj) != 0) goto _OVER;
-  if (mndAddStreamToTrans(pMnode, &streamObj, pCreate->ast, pTrans, smaObj.uid) != 0) goto _OVER;
+  if (mndAddStreamToTrans(pMnode, &streamObj, pCreate->ast, pTrans) != 0) goto _OVER;
   if (mndTransPrepare(pMnode, pTrans) != 0) goto _OVER;
 
   code = 0;

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

@@ -246,7 +246,7 @@ static int32_t mndStreamGetPlanString(const char *ast, char **pStr) {
   return code;
 }
 
-int32_t mndAddStreamToTrans(SMnode *pMnode, SStreamObj *pStream, const char *ast, STrans *pTrans, int64_t smaId) {
+int32_t mndAddStreamToTrans(SMnode *pMnode, SStreamObj *pStream, const char *ast, STrans *pTrans) {
   SNode *pAst = NULL;
 
   if (nodesStringToNode(ast, &pAst) < 0) {
@@ -271,7 +271,7 @@ int32_t mndAddStreamToTrans(SMnode *pMnode, SStreamObj *pStream, const char *ast
     return -1;
   }
 
-  if (mndScheduleStream(pMnode, pTrans, pStream, smaId) < 0) {
+  if (mndScheduleStream(pMnode, pTrans, pStream) < 0) {
     mError("stream:%ld, schedule stream since %s", pStream->uid, terrstr());
     return -1;
   }
@@ -300,6 +300,10 @@ static int32_t mndCreateStream(SMnode *pMnode, SNodeMsg *pReq, SCMCreateStreamRe
   streamObj.dbUid = pDb->uid;
   streamObj.version = 1;
   streamObj.sql = pCreate->sql;
+  streamObj.createdBy = STREAM_CREATED_BY__USER;
+  // TODO
+  streamObj.fixedSinkVgId = 0;
+  streamObj.smaId = 0;
   /*streamObj.physicalPlan = "";*/
   streamObj.logicalPlan = "not implemented";
 
@@ -310,7 +314,7 @@ static int32_t mndCreateStream(SMnode *pMnode, SNodeMsg *pReq, SCMCreateStreamRe
   }
   mDebug("trans:%d, used to create stream:%s", pTrans->id, pCreate->name);
 
-  if (mndAddStreamToTrans(pMnode, &streamObj, pCreate->ast, pTrans, -1) != 0) {
+  if (mndAddStreamToTrans(pMnode, &streamObj, pCreate->ast, pTrans) != 0) {
     mError("trans:%d, failed to add stream since %s", pTrans->id, terrstr());
     mndTransDrop(pTrans);
     return -1;

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

@@ -62,8 +62,6 @@ static int32_t mndPersistMqSetConnReq(SMnode *pMnode, STrans *pTrans, const SMqT
 
 static int32_t mndPersistRebalanceMsg(SMnode *pMnode, STrans *pTrans, const SMqConsumerEp *pConsumerEp);
 
-static int32_t mndInitUnassignedVg(SMnode *pMnode, const SMqTopicObj *pTopic, SMqSubscribeObj *pSub);
-
 int32_t mndInitSubscribe(SMnode *pMnode) {
   SSdbTable table = {.sdbType = SDB_SUBSCRIBE,
                      .keyType = SDB_KEY_BINARY,
@@ -98,14 +96,6 @@ static SMqSubscribeObj *mndCreateSubscription(SMnode *pMnode, const SMqTopicObj
     return NULL;
   }
 
-#if 0
-  if (mndInitUnassignedVg(pMnode, pTopic, pSub) < 0) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    tDeleteSMqSubscribeObj(pSub);
-    taosMemoryFree(pSub);
-    return NULL;
-  }
-#endif
   // TODO: disable alter subscribed table
   return pSub;
 }
@@ -210,45 +200,6 @@ static int32_t mndPersistCancelConnReq(SMnode *pMnode, STrans *pTrans, const SMq
   return 0;
 }
 
-#if 0
-static int32_t mndProcessResetOffsetReq(SNodeMsg *pMsg) {
-  SMnode             *pMnode = pMsg->pNode;
-  uint8_t            *str = pMsg->rpcMsg.pCont;
-  SMqCMResetOffsetReq req;
-
-  SCoder decoder;
-  tCoderInit(&decoder, TD_LITTLE_ENDIAN, str, pMsg->rpcMsg.contLen, TD_DECODER);
-  tDecodeSMqCMResetOffsetReq(&decoder, &req);
-
-  SHashObj *pHash = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false, HASH_NO_LOCK);
-  if (pHash == NULL) {
-    return -1;
-  }
-
-  for (int32_t i = 0; i < req.num; i++) {
-    SMqOffset    *pOffset = &req.offsets[i];
-    SMqVgOffsets *pVgOffset = taosHashGet(pHash, &pOffset->vgId, sizeof(int32_t));
-    if (pVgOffset == NULL) {
-      pVgOffset = taosMemoryMalloc(sizeof(SMqVgOffsets));
-      if (pVgOffset == NULL) {
-        return -1;
-      }
-      pVgOffset->offsets = taosArrayInit(0, sizeof(void *));
-      taosArrayPush(pVgOffset->offsets, &pOffset);
-    }
-    taosHashPut(pHash, &pOffset->vgId, sizeof(int32_t), &pVgOffset, sizeof(void *));
-  }
-
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, &pMsg->rpcMsg);
-  if (pTrans == NULL) {
-    mError("mq-reset-offset: failed since %s", terrstr());
-    return -1;
-  }
-
-  return 0;
-}
-#endif
-
 static int32_t mndProcessGetSubEpReq(SNodeMsg *pMsg) {
   SMnode           *pMnode = pMsg->pNode;
   SMqCMGetSubEpReq *pReq = (SMqCMGetSubEpReq *)pMsg->rpcMsg.pCont;
@@ -574,251 +525,6 @@ static int32_t mndProcessDoRebalanceMsg(SNodeMsg *pMsg) {
   return 0;
 }
 
-#if 0
-      for (int32_t j = 0; j < consumerNum; j++) {
-        bool            changed = false;
-        bool            unfished = false;
-
-        bool            canUseLeft = imbalanceSolved < imbalanceVg;
-        bool            mustUseLeft = canUseLeft && (imbalanceVg - imbalanceSolved >= consumerNum - j);
-        ASSERT(imbalanceVg - imbalanceSolved <= consumerNum - j);
-
-        int32_t maxVg = vgEachConsumer + canUseLeft;
-        int32_t minVg = vgEachConsumer + mustUseLeft;
-
-        SMqSubConsumer *pSubConsumer = taosArrayGet(pSub->consumers, j);
-        int32_t         vgThisConsumerBeforeRb = taosArrayGetSize(pSubConsumer->vgInfo);
-        int32_t         vgThisConsumerAfterRb;
-        if (vgThisConsumerBeforeRb > maxVg) {
-          vgThisConsumerAfterRb = maxVg; 
-          imbalanceSolved++;
-          changed = true;
-        } else if (vgThisConsumerBeforeRb < minVg) {
-          vgThisConsumerAfterRb = minVg;
-          if (mustUseLeft) imbalanceSolved++;
-          changed = true;
-        } else {
-          vgThisConsumerAfterRb = vgThisConsumerBeforeRb;
-        }
-
-        if (vgThisConsumerBeforeRb > vgThisConsumerAfterRb) {
-          while (taosArrayGetSize(pSubConsumer->vgInfo) > vgThisConsumerAfterRb) {
-            // put into unassigned
-            SMqConsumerEp *pConsumerEp = taosArrayPop(pSubConsumer->vgInfo);
-            ASSERT(pConsumerEp != NULL);
-            ASSERT(pConsumerEp->consumerId == pSubConsumer->consumerId);
-            taosArrayPush(unassignedVgStash, pConsumerEp);
-          }
-
-        } else if (vgThisConsumerBeforeRb < vgThisConsumerAfterRb) {
-          // assign from unassigned
-          while (taosArrayGetSize(pSubConsumer->vgInfo) < vgThisConsumerAfterRb) {
-            // if no unassgined, save j
-            if (taosArrayGetSize(unassignedVgStash) == 0) {
-              taosArrayPush(unassignedConsumerIdx, &j);
-              unfished = true;
-              break;
-            }
-            // assign vg to consumer
-            SMqConsumerEp *pConsumerEp = taosArrayPop(unassignedVgStash);
-            ASSERT(pConsumerEp != NULL);
-            pConsumerEp->oldConsumerId = pConsumerEp->consumerId;
-            pConsumerEp->consumerId = pSubConsumer->consumerId;
-            taosArrayPush(pSubConsumer->vgInfo, pConsumerEp);
-            // build msg and persist into trans
-          }
-        }
-
-        if (changed && !unfished) {
-          SMqConsumerObj *pRebConsumer = mndAcquireConsumer(pMnode, pSubConsumer->consumerId);
-          pRebConsumer->epoch++;
-          if (vgThisConsumerAfterRb != 0) {
-            atomic_store_32(&pRebConsumer->status, MQ_CONSUMER_STATUS__ACTIVE);
-          } else {
-            atomic_store_32(&pRebConsumer->status, MQ_CONSUMER_STATUS__IDLE);
-          }
-          SSdbRaw *pConsumerRaw = mndConsumerActionEncode(pRebConsumer);
-          sdbSetRawStatus(pConsumerRaw, SDB_STATUS_READY);
-          mndTransAppendRedolog(pTrans, pConsumerRaw);
-          mndReleaseConsumer(pMnode, pRebConsumer);
-          // TODO: save history
-        }
-      }
-
-      for (int32_t j = 0; j < taosArrayGetSize(unassignedConsumerIdx); j++) {
-        bool            canUseLeft = imbalanceSolved < imbalanceVg;
-        int32_t         consumerIdx = *(int32_t *)taosArrayGet(unassignedConsumerIdx, j);
-        SMqSubConsumer *pSubConsumer = taosArrayGet(pSub->consumers, consumerIdx);
-        if (canUseLeft) imbalanceSolved++;
-        // must use
-        int32_t vgThisConsumerAfterRb = taosArrayGetSize(pSubConsumer->vgInfo) + canUseLeft;
-        while (taosArrayGetSize(pSubConsumer->vgInfo) < vgEachConsumer + canUseLeft) {
-          // assign vg to consumer
-          SMqConsumerEp *pConsumerEp = taosArrayPop(unassignedVgStash);
-          ASSERT(pConsumerEp != NULL);
-          pConsumerEp->oldConsumerId = pConsumerEp->consumerId;
-          pConsumerEp->consumerId = pSubConsumer->consumerId;
-          taosArrayPush(pSubConsumer->vgInfo, pConsumerEp);
-          // build msg and persist into trans
-        }
-        SMqConsumerObj *pRebConsumer = mndAcquireConsumer(pMnode, pSubConsumer->consumerId);
-        pRebConsumer->epoch++;
-        atomic_store_32(&pRebConsumer->status, MQ_CONSUMER_STATUS__ACTIVE);
-        SSdbRaw *pConsumerRaw = mndConsumerActionEncode(pRebConsumer);
-        sdbSetRawStatus(pConsumerRaw, SDB_STATUS_READY);
-        mndTransAppendRedolog(pTrans, pConsumerRaw);
-        mndReleaseConsumer(pMnode, pRebConsumer);
-        // TODO: save history
-      }
-#endif
-
-#if 0
-    //update consumer status for the subscribption
-    for (int32_t i = 0; i < taosArrayGetSize(pSub->assigned); i++) {
-      SMqConsumerEp *pCEp = taosArrayGet(pSub->assigned, i);
-      int64_t        consumerId = pCEp->consumerId;
-      if (pCEp->status != -1) {
-        int32_t consumerHbStatus = atomic_fetch_add_32(&pCEp->consumerHbStatus, 1);
-        if (consumerHbStatus < MND_SUBSCRIBE_REBALANCE_CNT) {
-          continue;
-        }
-        // put consumer into lostConsumer
-        SMqConsumerEp* lostConsumer = taosArrayPush(pSub->lostConsumer, pCEp);
-        lostConsumer->qmsg = NULL;
-        // put vg into unassigned
-        taosArrayPush(pSub->unassignedVg, pCEp);
-        // remove from assigned
-        // TODO: swap with last one, reduce size and reset i
-        taosArrayRemove(pSub->assigned, i);
-        // remove from available consumer
-        for (int32_t j = 0; j < taosArrayGetSize(pSub->availConsumer); j++) {
-          if (*(int64_t *)taosArrayGet(pSub->availConsumer, i) == pCEp->consumerId) {
-            taosArrayRemove(pSub->availConsumer, j);
-            break;
-          }
-          // TODO: acquire consumer, set status to unavail
-        }
-#if 0
-        SMqConsumerObj* pConsumer = mndAcquireConsumer(pMnode, consumerId);
-        pConsumer->epoch++;
-        printf("current epoch %ld size %ld", pConsumer->epoch, pConsumer->topics->size);
-        SSdbRaw* pRaw = mndConsumerActionEncode(pConsumer);
-        sdbSetRawStatus(pRaw, SDB_STATUS_READY);
-        sdbWriteNotFree(pMnode->pSdb, pRaw);
-        mndReleaseConsumer(pMnode, pConsumer);
-#endif
-      }
-    }
-    // no available consumer, skip rebalance
-    if (taosArrayGetSize(pSub->availConsumer) == 0) {
-      continue;
-    }
-    taosArrayGet(pSub->availConsumer, 0);
-    // rebalance condition1 : have unassigned vg
-    // assign vg to a consumer, trying to find the least assigned one
-    if ((sz = taosArrayGetSize(pSub->unassignedVg)) > 0) {
-      char *topic = NULL;
-      char *cgroup = NULL;
-      mndSplitSubscribeKey(pSub->key, &topic, &cgroup);
-
-      SMqTopicObj *pTopic = mndAcquireTopic(pMnode, topic);
-      STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
-      for (int32_t i = 0; i < sz; i++) {
-        int64_t        consumerId = *(int64_t *)taosArrayGet(pSub->availConsumer, pSub->nextConsumerIdx);
-        pSub->nextConsumerIdx = (pSub->nextConsumerIdx + 1) % taosArrayGetSize(pSub->availConsumer);
-
-        SMqConsumerEp *pCEp = taosArrayPop(pSub->unassignedVg);
-        pCEp->oldConsumerId = pCEp->consumerId;
-        pCEp->consumerId = consumerId;
-        taosArrayPush(pSub->assigned, pCEp);
-
-        SMqConsumerObj *pConsumer = mndAcquireConsumer(pMnode, consumerId);
-        pConsumer->epoch++;
-        SSdbRaw* pConsumerRaw = mndConsumerActionEncode(pConsumer);
-        sdbSetRawStatus(pConsumerRaw, SDB_STATUS_READY);
-        sdbWrite(pMnode->pSdb, pConsumerRaw);
-        mndReleaseConsumer(pMnode, pConsumer);
-
-        void* msg;
-        int32_t msgLen;
-        mndBuildRebalanceMsg(&msg, &msgLen, pTopic, pCEp, cgroup, topic);
-
-        // persist msg
-        STransAction action = {0};
-        action.epSet = pCEp->epSet;
-        action.pCont = msg;
-        action.contLen = sizeof(SMsgHead) + msgLen;
-        action.msgType = TDMT_VND_MQ_SET_CONN;
-        mndTransAppendRedoAction(pTrans, &action);
-
-        // persist data
-        SSdbRaw *pRaw = mndSubActionEncode(pSub);
-        sdbSetRawStatus(pRaw, SDB_STATUS_READY);
-        mndTransAppendRedolog(pTrans, pRaw);
-      }
-
-      if (mndTransPrepare(pMnode, pTrans) != 0) {
-        mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
-      }
-      mndReleaseTopic(pMnode, pTopic);
-      mndTransDrop(pTrans);
-      taosMemoryFreeClear(topic);
-      taosMemoryFreeClear(cgroup);
-    }
-    // rebalance condition2 : imbalance assignment
-  }
-  return 0;
-}
-#endif
-
-#if 0
-static int32_t mndInitUnassignedVg(SMnode *pMnode, const SMqTopicObj *pTopic, SMqSubscribeObj *pSub) {
-  SSdb      *pSdb = pMnode->pSdb;
-  SVgObj    *pVgroup = NULL;
-  SQueryPlan *pPlan = qStringToQueryPlan(pTopic->physicalPlan);
-  SArray    *pArray = NULL;
-  SNodeListNode    *inner = (SNodeListNode*)nodesListGetNode(pPlan->pSubplans, 0);
-  SSubplan  *plan = (SSubplan*)nodesListGetNode(inner->pNodeList, 0);
-  SArray    *unassignedVg = pSub->unassignedVg;
-
-  void *pIter = NULL;
-  while (1) {
-    pIter = sdbFetch(pSdb, SDB_VGROUP, pIter, (void **)&pVgroup);
-    if (pIter == NULL) break;
-    if (pVgroup->dbUid != pTopic->dbUid) {
-      sdbRelease(pSdb, pVgroup);
-      continue;
-    }
-
-    pSub->vgNum++;
-    plan->execNode.nodeId = pVgroup->vgId;
-    plan->execNode.epset = mndGetVgroupEpset(pMnode, pVgroup);
-
-    if (schedulerConvertDagToTaskList(pPlan, &pArray) < 0) {
-      terrno = TSDB_CODE_MND_UNSUPPORTED_TOPIC;
-      mError("unsupport topic: %s, sql: %s", pTopic->name, pTopic->sql);
-      return -1;
-    }
-
-    SMqConsumerEp consumerEp = {0};
-    consumerEp.status = 0;
-    consumerEp.consumerId = -1;
-    STaskInfo *pTaskInfo = taosArrayGet(pArray, 0);
-    consumerEp.epSet = pTaskInfo->addr.epset;
-    consumerEp.vgId = pTaskInfo->addr.nodeId;
-
-    ASSERT(consumerEp.vgId == pVgroup->vgId);
-    consumerEp.qmsg = strdup(pTaskInfo->msg->msg);
-    taosArrayPush(unassignedVg, &consumerEp);
-    // TODO: free taskInfo
-    taosArrayDestroy(pArray);
-  }
-
-  /*qDestroyQueryDag(pDag);*/
-  return 0;
-}
-#endif
-
 static int32_t mndPersistMqSetConnReq(SMnode *pMnode, STrans *pTrans, const SMqTopicObj *pTopic, const char *cgroup,
                                       const SMqConsumerEp *pConsumerEp) {
   ASSERT(pConsumerEp->oldConsumerId == -1);
@@ -1140,7 +846,7 @@ static int32_t mndProcessSubscribeReq(SNodeMsg *pMsg) {
     }
   }
 
-  if (oldSub) taosArrayDestroyEx(oldSub, (void (*)(void*))taosMemoryFree);
+  if (oldSub) taosArrayDestroyEx(oldSub, (void (*)(void *))taosMemoryFree);
 
   // persist consumerObj
   SSdbRaw *pConsumerRaw = mndConsumerActionEncode(pConsumer);

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

@@ -26,7 +26,7 @@
 #include "parser.h"
 #include "tname.h"
 
-#define MND_TOPIC_VER_NUMBER   1
+#define MND_TOPIC_VER_NUMBER 1
 #define MND_TOPIC_RESERVE_SIZE 64
 
 static int32_t mndTopicActionInsert(SSdb *pSdb, SMqTopicObj *pTopic);
@@ -65,9 +65,11 @@ void mndCleanupTopic(SMnode *pMnode) {}
 SSdbRaw *mndTopicActionEncode(SMqTopicObj *pTopic) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
 
-  int32_t  logicalPlanLen = strlen(pTopic->logicalPlan) + 1;
+  int32_t logicalPlanLen = strlen(pTopic->logicalPlan) + 1;
-  int32_t  physicalPlanLen = strlen(pTopic->physicalPlan) + 1;
+  int32_t physicalPlanLen = strlen(pTopic->physicalPlan) + 1;
-  int32_t  size = sizeof(SMqTopicObj) + logicalPlanLen + physicalPlanLen + pTopic->sqlLen + MND_TOPIC_RESERVE_SIZE;
+  int32_t swLen = taosEncodeSSchemaWrapper(NULL, &pTopic->schema);
+  int32_t size =
+      sizeof(SMqTopicObj) + logicalPlanLen + physicalPlanLen + pTopic->sqlLen + swLen + MND_TOPIC_RESERVE_SIZE;
   SSdbRaw *pRaw = sdbAllocRaw(SDB_TOPIC, MND_TOPIC_VER_NUMBER, size);
   if (pRaw == NULL) goto TOPIC_ENCODE_OVER;
 
@@ -86,8 +88,7 @@ SSdbRaw *mndTopicActionEncode(SMqTopicObj *pTopic) {
   SDB_SET_INT32(pRaw, dataPos, physicalPlanLen, TOPIC_ENCODE_OVER);
   SDB_SET_BINARY(pRaw, dataPos, pTopic->physicalPlan, physicalPlanLen, TOPIC_ENCODE_OVER);
 
-  int32_t swLen = taosEncodeSSchemaWrapper(NULL, &pTopic->schema);
+  void *swBuf = taosMemoryMalloc(swLen);
-  void   *swBuf = taosMemoryMalloc(swLen);
   if (swBuf == NULL) {
     goto TOPIC_ENCODE_OVER;
   }

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

@@ -54,7 +54,7 @@ static bool    mndTransPerformRollbackStage(SMnode *pMnode, STrans *pTrans);
 static bool    mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans);
 
 static void    mndTransExecute(SMnode *pMnode, STrans *pTrans);
-static void    mndTransSendRpcRsp(STrans *pTrans);
+static void    mndTransSendRpcRsp(SMnode *pMnode, STrans *pTrans);
 static int32_t mndProcessTransReq(SNodeMsg *pReq);
 static int32_t mndProcessKillTransReq(SNodeMsg *pReq);
 
@@ -737,7 +737,7 @@ static int32_t mndTransRollback(SMnode *pMnode, STrans *pTrans) {
   return 0;
 }
 
-static void mndTransSendRpcRsp(STrans *pTrans) {
+static void mndTransSendRpcRsp(SMnode *pMnode, STrans *pTrans) {
   bool sendRsp = false;
 
   if (pTrans->stage == TRN_STAGE_FINISHED) {
@@ -771,7 +771,7 @@ static void mndTransSendRpcRsp(STrans *pTrans) {
                       .ahandle = pTrans->rpcAHandle,
                       .pCont = rpcCont,
                       .contLen = pTrans->rpcRspLen};
-    rpcSendResponse(&rspMsg);
+    tmsgSendRsp(&rspMsg);
     pTrans->rpcHandle = NULL;
     pTrans->rpcRsp = NULL;
     pTrans->rpcRspLen = 0;
@@ -898,7 +898,7 @@ static int32_t mndTransSendActionMsg(SMnode *pMnode, STrans *pTrans, SArray *pAr
       pAction->msgReceived = 0;
       pAction->errCode = 0;
     } else {
-      mDebug("trans:%d, action:%d not send since %s", pTrans->id, action, terrstr());
+      mError("trans:%d, action:%d not send since %s", pTrans->id, action, terrstr());
       return -1;
     }
   }
@@ -938,7 +938,7 @@ static int32_t mndTransExecuteActions(SMnode *pMnode, STrans *pTrans, SArray *pA
       return errCode;
     }
   } else {
-    mDebug("trans:%d, %d of %d actions executed, code:0x%04x", pTrans->id, numOfReceived, numOfActions, errCode & 0XFFFF);
+    mDebug("trans:%d, %d of %d actions executing", pTrans->id, numOfReceived, numOfActions);
     return TSDB_CODE_MND_ACTION_IN_PROGRESS;
   }
 }
@@ -1158,7 +1158,7 @@ static void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
     }
   }
 
-  mndTransSendRpcRsp(pTrans);
+  mndTransSendRpcRsp(pMnode, pTrans);
 }
 
 static int32_t mndProcessTransReq(SNodeMsg *pReq) {