Merge remote-tracking branch 'origin/develop' into feature/crash_gen

src/client/inc/tscProfile.h

@@ -26,7 +26,7 @@ void tscAddIntoSqlList(SSqlObj *pSql);
 void tscRemoveFromSqlList(SSqlObj *pSql);
 void tscAddIntoStreamList(SSqlStream *pStream);
 void tscRemoveFromStreamList(SSqlStream *pStream, SSqlObj *pSqlObj);
-char *tscBuildQueryStreamDesc(char *pMsg, STscObj *pObj);
+int  tscBuildQueryStreamDesc(void *pMsg, STscObj *pObj);
 void tscKillQuery(STscObj *pObj, uint32_t killId);
 void tscKillStream(STscObj *pObj, uint32_t killId);
 void tscKillConnection(STscObj *pObj);

src/client/inc/tscSecondaryMerge.h

@@ -63,19 +63,21 @@ typedef struct SLocalReducer {
 //  char *                 pBufForInterpo;  // intermediate buffer for interpolation
   tFilePage *            pTempBuffer;
   struct SQLFunctionCtx *pCtx;
-  int32_t                rowSize;     // size of each intermediate result.
-  int32_t                status;      // denote it is in reduce process, in reduce process, it
-  bool                   hasPrevRow;  // cannot be released
+  int32_t                rowSize;      // size of each intermediate result.
+  int32_t                finalRowSize; // final result row size
+  int32_t                status;       // denote it is in reduce process, in reduce process, it
+  bool                   hasPrevRow;   // cannot be released
   bool                   hasUnprocessedRow;
   tOrderDescriptor *     pDesc;
   SColumnModel *         resColModel;
   tExtMemBuffer **       pExtMemBuffer;      // disk-based buffer
-  SFillInfo*             pFillInfo;         // interpolation support structure
+  SFillInfo*             pFillInfo;          // interpolation support structure
   char *                 pFinalRes;          // result data after interpo
   tFilePage *            discardData;
   SResultInfo *          pResInfo;
   bool                   discard;
   int32_t                offset;             // limit offset value
+  bool                   orderPrjOnSTable;   // projection query on stable
 } SLocalReducer;
 
 typedef struct SSubqueryState {

src/client/inc/tsclient.h

@@ -285,8 +285,6 @@ typedef struct {
 typedef struct STscObj {
   void *             signature;
   void *             pTimer;
-  char               mnodeIp[TSDB_USER_LEN];
-  uint16_t           mnodePort;
   char               user[TSDB_USER_LEN];
   char               pass[TSDB_KEY_LEN];
   char               acctId[TSDB_DB_NAME_LEN];
@@ -294,6 +292,7 @@ typedef struct STscObj {
   char               sversion[TSDB_VERSION_LEN];
   char               writeAuth : 1;
   char               superAuth : 1;
+  uint32_t           connId;
   struct SSqlObj *   pHb;
   struct SSqlObj *   sqlList;
   struct SSqlStream *streamList;

src/client/src/tscFunctionImpl.c

@@ -341,16 +341,6 @@ bool stableQueryFunctChanged(int32_t funcId) {
  */
 void resetResultInfo(SResultInfo *pResInfo) { pResInfo->initialized = false; }
 
-void initResultInfo(SResultInfo *pResInfo) {
-  pResInfo->initialized = true;  // the this struct has been initialized flag
-  
-  pResInfo->complete = false;
-  pResInfo->hasResult = false;
-  pResInfo->numOfRes = 0;
-  
-  memset(pResInfo->interResultBuf, 0, (size_t)pResInfo->bufLen);
-}
-
 void setResultInfoBuf(SResultInfo *pResInfo, int32_t size, bool superTable) {
   assert(pResInfo->interResultBuf == NULL);
   
@@ -387,9 +377,7 @@ static bool function_setup(SQLFunctionCtx *pCtx) {
  */
 static void function_finalizer(SQLFunctionCtx *pCtx) {
   SResultInfo *pResInfo = GET_RES_INFO(pCtx);
-  
   if (pResInfo->hasResult != DATA_SET_FLAG) {
-    tscTrace("no result generated, result is set to NULL");
     if (pCtx->outputType == TSDB_DATA_TYPE_BINARY || pCtx->outputType == TSDB_DATA_TYPE_NCHAR) {
       setVardataNull(pCtx->aOutputBuf, pCtx->outputType);
     } else {

src/client/src/tscParseInsert.c

@@ -48,7 +48,7 @@ static int32_t tscToInteger(SSQLToken *pToken, int64_t *value, char **endPtr) {
   
   int32_t radix = 10;
   
-  int32_t radixList[3] = {16, 8, 2};
+  int32_t radixList[3] = {16, 8, 2}; // the integer number with different radix: hex, oct, bin
   if (pToken->type == TK_HEX || pToken->type == TK_OCT || pToken->type == TK_BIN) {
     radix = radixList[pToken->type - TK_HEX];
   }

src/client/src/tscPrepare.c

@@ -494,7 +494,6 @@ TAOS_STMT* taos_stmt_init(TAOS* taos) {
   tsem_init(&pSql->rspSem, 0, 0);
   pSql->signature     = pSql;
   pSql->pTscObj       = pObj;
-  //pSql->pTscObj->pSql = pSql;
   pSql->maxRetry      = TSDB_MAX_REPLICA_NUM;
 
   pStmt->pSql = pSql;
@@ -515,7 +514,7 @@ int taos_stmt_prepare(TAOS_STMT* stmt, const char* sql, unsigned long length) {
   //doAsyncQuery(pObj, pSql, waitForQueryRsp, taos, sqlstr, sqlLen);
   SSqlCmd *pCmd    = &pSql->cmd;
   SSqlRes *pRes    = &pSql->res;
-  pSql->param      = (void*)pSql;
+  pSql->param      = (void*) pSql;
   pSql->fp         = waitForQueryRsp;
   pSql->insertType = TSDB_QUERY_TYPE_STMT_INSERT;
   

src/client/src/tscProfile.c

@@ -19,6 +19,7 @@
 #include "ttime.h"
 #include "ttimer.h"
 #include "tutil.h"
+#include "taosmsg.h"
 
 void  tscSaveSlowQueryFp(void *handle, void *tmrId);
 void *tscSlowQueryConn = NULL;
@@ -96,7 +97,7 @@ void tscSaveSlowQuery(SSqlObj *pSql) {
   }
 
   tscTrace("%p query time:%" PRId64 " sql:%s", pSql, pSql->res.useconds, pSql->sqlstr);
-  int32_t sqlSize = TSDB_SHOW_SQL_LEN + size;
+  int32_t sqlSize = TSDB_SLOW_QUERY_SQL_LEN + size;
   
   char *sql = malloc(sqlSize);
   if (sql == NULL) {
@@ -106,9 +107,9 @@ void tscSaveSlowQuery(SSqlObj *pSql) {
   
   int len = snprintf(sql, size, "insert into %s.slowquery values(now, '%s', %" PRId64 ", %" PRId64 ", '", tsMonitorDbName,
           pSql->pTscObj->user, pSql->stime, pSql->res.useconds);
-  int sqlLen = snprintf(sql + len, TSDB_SHOW_SQL_LEN, "%s", pSql->sqlstr);
-  if (sqlLen > TSDB_SHOW_SQL_LEN - 1) {
-    sqlLen = len + TSDB_SHOW_SQL_LEN - 1;
+  int sqlLen = snprintf(sql + len, TSDB_SLOW_QUERY_SQL_LEN, "%s", pSql->sqlstr);
+  if (sqlLen > TSDB_SLOW_QUERY_SQL_LEN - 1) {
+    sqlLen = len + TSDB_SLOW_QUERY_SQL_LEN - 1;
   } else {
     sqlLen += len;
   }
@@ -205,28 +206,28 @@ void tscKillStream(STscObj *pObj, uint32_t killId) {
   }
 
   pthread_mutex_unlock(&pObj->mutex);
-  
+
   if (pStream) {
     tscTrace("%p stream:%p is killed, streamId:%d", pStream->pSql, pStream, killId);
+    if (pStream->callback) {
+      pStream->callback(pStream->param);
+    }
+    taos_close_stream(pStream);
+  } else {
+    tscError("failed to kill stream, streamId:%d not exist", killId);
   }
-
-  if (pStream->callback) {
-    pStream->callback(pStream->param);
-  }
-  taos_close_stream(pStream);
 }
 
-char *tscBuildQueryStreamDesc(char *pMsg, STscObj *pObj) {
-  char *  pMax = pMsg + TSDB_PAYLOAD_SIZE - 256;
+int tscBuildQueryStreamDesc(void *pMsg, STscObj *pObj) {
+  SCMHeartBeatMsg *pHeartbeat = pMsg;
+  int allocedQueriesNum = pHeartbeat->numOfQueries;
+  int allocedStreamsNum = pHeartbeat->numOfStreams;
   
-  SQqueryList *pQList = (SQqueryList *)pMsg;
-  pQList->numOfQueries = 0;
-
-  SQueryDesc *pQdesc = (SQueryDesc*)(pMsg + sizeof(SQqueryList));
+  pHeartbeat->numOfQueries = 0;
+  SQueryDesc *pQdesc = (SQueryDesc *)pHeartbeat->pData;
 
   // We extract the lock to tscBuildHeartBeatMsg function.
-  /* pthread_mutex_lock (&pObj->mutex); */
-  pMsg += sizeof(SQqueryList);
+
   SSqlObj *pSql = pObj->sqlList;
   while (pSql) {
     /*
@@ -240,47 +241,46 @@ char *tscBuildQueryStreamDesc(char *pMsg, STscObj *pObj) {
 
     strncpy(pQdesc->sql, pSql->sqlstr, TSDB_SHOW_SQL_LEN - 1);
     pQdesc->sql[TSDB_SHOW_SQL_LEN - 1] = 0;
-    pQdesc->stime = pSql->stime;
-    pQdesc->queryId = pSql->queryId;
-    pQdesc->useconds = pSql->res.useconds;
+    pQdesc->stime = htobe64(pSql->stime);
+    pQdesc->queryId = htonl(pSql->queryId);
+    pQdesc->useconds = htobe64(pSql->res.useconds);
 
-    pQList->numOfQueries++;
+    pHeartbeat->numOfQueries++;
     pQdesc++;
     pSql = pSql->next;
-    pMsg += sizeof(SQueryDesc);
-    if (pMsg > pMax) break;
+    if (pHeartbeat->numOfQueries >= allocedQueriesNum) break;
   }
 
-  SStreamList *pSList = (SStreamList *)pMsg;
-  pSList->numOfStreams = 0;
-  
-  SStreamDesc *pSdesc = (SStreamDesc*) (pMsg + sizeof(SStreamList));
+  pHeartbeat->numOfStreams = 0;
+  SStreamDesc *pSdesc = (SStreamDesc *)pQdesc;
 
-  pMsg += sizeof(SStreamList);
   SSqlStream *pStream = pObj->streamList;
   while (pStream) {
     strncpy(pSdesc->sql, pStream->pSql->sqlstr, TSDB_SHOW_SQL_LEN - 1);
     pSdesc->sql[TSDB_SHOW_SQL_LEN - 1] = 0;
-    pSdesc->streamId = pStream->streamId;
-    pSdesc->num = pStream->num;
+    pSdesc->streamId = htonl(pStream->streamId);
+    pSdesc->num = htobe64(pStream->num);
 
-    pSdesc->useconds = pStream->useconds;
-    pSdesc->stime = pStream->stime - pStream->interval;
-    pSdesc->ctime = pStream->ctime;
+    pSdesc->useconds = htobe64(pStream->useconds);
+    pSdesc->stime = htobe64(pStream->stime - pStream->interval);
+    pSdesc->ctime = htobe64(pStream->ctime);
 
-    pSdesc->slidingTime = pStream->slidingTime;
-    pSdesc->interval = pStream->interval;
+    pSdesc->slidingTime = htobe64(pStream->slidingTime);
+    pSdesc->interval = htobe64(pStream->interval);
 
-    pSList->numOfStreams++;
+    pHeartbeat->numOfStreams++;
     pSdesc++;
     pStream = pStream->next;
-    pMsg += sizeof(SStreamDesc);
-    if (pMsg > pMax) break;
+    if (pHeartbeat->numOfStreams >= allocedStreamsNum) break;
   }
 
-  /* pthread_mutex_unlock (&pObj->mutex); */
+  int32_t msgLen = pHeartbeat->numOfQueries * sizeof(SQueryDesc) + pHeartbeat->numOfStreams * sizeof(SStreamDesc) +
+                   sizeof(SCMHeartBeatMsg);
+  pHeartbeat->connId = htonl(pObj->connId);
+  pHeartbeat->numOfQueries = htonl(pHeartbeat->numOfQueries);
+  pHeartbeat->numOfStreams = htonl(pHeartbeat->numOfStreams);
 
-  return pMsg;
+  return msgLen;
 }
 
 void tscKillConnection(STscObj *pObj) {

src/client/src/tscSQLParser.c

@@ -93,7 +93,7 @@ static int32_t validateArithmeticSQLExpr(tSQLExpr* pExpr, SQueryInfo* pQueryInfo
 static int32_t validateDNodeConfig(tDCLSQL* pOptions);
 static int32_t validateLocalConfig(tDCLSQL* pOptions);
 static int32_t validateColumnName(char* name);
-static int32_t setKillInfo(SSqlObj* pSql, struct SSqlInfo* pInfo);
+static int32_t setKillInfo(SSqlObj* pSql, struct SSqlInfo* pInfo, int32_t killType);
 
 static bool validateOneTags(SSqlCmd* pCmd, TAOS_FIELD* pTagField);
 static bool hasTimestampForPointInterpQuery(SQueryInfo* pQueryInfo);
@@ -531,7 +531,7 @@ int32_t tscToSQLCmd(SSqlObj* pSql, struct SSqlInfo* pInfo) {
     case TSDB_SQL_KILL_QUERY:
     case TSDB_SQL_KILL_STREAM:
     case TSDB_SQL_KILL_CONNECTION: {
-      if ((code = setKillInfo(pSql, pInfo)) != TSDB_CODE_SUCCESS) {
+      if ((code = setKillInfo(pSql, pInfo, pInfo->type)) != TSDB_CODE_SUCCESS) {
         return code;
       }
 
@@ -1881,22 +1881,38 @@ int32_t addExprAndResultField(SQueryInfo* pQueryInfo, int32_t colIndex, tSQLExpr
   
       // functions can not be applied to normal columns
       int32_t numOfCols = tscGetNumOfColumns(pTableMetaInfo->pTableMeta);
-      if (index.columnIndex < numOfCols) {
+      if (index.columnIndex < numOfCols && index.columnIndex != TSDB_TBNAME_COLUMN_INDEX) {
         return invalidSqlErrMsg(pQueryInfo->msg, msg6);
       }
     
-      index.columnIndex -= numOfCols;
+      if (index.columnIndex > 0) {
+        index.columnIndex -= numOfCols;
+      }
       
       // 2. valid the column type
-      int16_t colType = pSchema[index.columnIndex].type;
-      if (colType == TSDB_DATA_TYPE_BOOL || colType >= TSDB_DATA_TYPE_BINARY) {
+      int16_t colType = 0;
+      if (index.columnIndex == TSDB_TBNAME_COLUMN_INDEX) {
+        colType = TSDB_DATA_TYPE_BINARY;
+      } else {
+        colType = pSchema[index.columnIndex].type;
+      }
+      
+      if (colType == TSDB_DATA_TYPE_BOOL) {
         return invalidSqlErrMsg(pQueryInfo->msg, msg1);
       }
 
       tscColumnListInsert(pTableMetaInfo->tagColList, &index);
       SSchema* pTagSchema = tscGetTableTagSchema(pTableMetaInfo->pTableMeta);
-      SSchema s = pTagSchema[index.columnIndex];
-
+      
+      SSchema s = {0};
+      if (index.columnIndex == TSDB_TBNAME_COLUMN_INDEX) {
+        s.bytes = TSDB_TABLE_NAME_LEN + VARSTR_HEADER_SIZE;
+        s.type  = TSDB_DATA_TYPE_BINARY;
+        s.colId = TSDB_TBNAME_COLUMN_INDEX;
+      } else {
+        s = pTagSchema[index.columnIndex];
+      }
+      
       int16_t bytes = 0;
       int16_t type  = 0;
       int32_t inter = 0;
@@ -2229,37 +2245,45 @@ int32_t setShowInfo(SSqlObj* pSql, struct SSqlInfo* pInfo) {
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t setKillInfo(SSqlObj* pSql, struct SSqlInfo* pInfo) {
-  const char* msg1 = "invalid ip address";
-  const char* msg2 = "invalid port";
+int32_t setKillInfo(SSqlObj* pSql, struct SSqlInfo* pInfo, int32_t killType) {
+  const char* msg1 = "invalid connection ID";
+  const char* msg2 = "invalid query ID";
+  const char* msg3 = "invalid stream ID";
 
   SSqlCmd* pCmd = &pSql->cmd;
   pCmd->command = pInfo->type;
-
-  SSQLToken* ip = &(pInfo->pDCLInfo->ip);
-  if (ip->n > TSDB_KILL_MSG_LEN) {
+  
+  SSQLToken* idStr = &(pInfo->pDCLInfo->ip);
+  if (idStr->n > TSDB_KILL_MSG_LEN) {
     return TSDB_CODE_INVALID_SQL;
   }
 
-  strncpy(pCmd->payload, ip->z, ip->n);
+  strncpy(pCmd->payload, idStr->z, idStr->n);
 
   const char delim = ':';
+  char* connIdStr = strtok(idStr->z, &delim);
+  char* queryIdStr = strtok(NULL, &delim);
 
-  char* ipStr = strtok(ip->z, &delim);
-  char* portStr = strtok(NULL, &delim);
-
-  if (!validateIpAddress(ipStr, strlen(ipStr))) {
+  int32_t connId = (int32_t)strtol(connIdStr, NULL, 10);
+  if (connId <= 0) {
     memset(pCmd->payload, 0, strlen(pCmd->payload));
-
     return invalidSqlErrMsg(tscGetErrorMsgPayload(pCmd), msg1);
   }
 
-  uint16_t port = (uint16_t)strtol(portStr, NULL, 10);
-  if (port <= 0 || port > 65535) {
-    memset(pCmd->payload, 0, strlen(pCmd->payload));
-    return invalidSqlErrMsg(tscGetErrorMsgPayload(pCmd), msg2);
+  if (killType == TSDB_SQL_KILL_CONNECTION) {
+    return TSDB_CODE_SUCCESS;
   }
 
+  int32_t queryId = (int32_t)strtol(queryIdStr, NULL, 10);
+  if (queryId <= 0) {
+    memset(pCmd->payload, 0, strlen(pCmd->payload));
+    if (killType == TSDB_SQL_KILL_QUERY) {
+      return invalidSqlErrMsg(tscGetErrorMsgPayload(pCmd), msg2);
+    } else {
+      return invalidSqlErrMsg(tscGetErrorMsgPayload(pCmd), msg3);
+    }
+  }
+  
   return TSDB_CODE_SUCCESS;
 }
 

src/client/src/tscSecondaryMerge.c

@@ -217,7 +217,7 @@ void tscCreateLocalReducer(tExtMemBuffer **pMemBuffer, int32_t numOfBuffer, tOrd
 
   pReducer->numOfBuffer = numOfFlush;
   pReducer->numOfVnode = numOfBuffer;
-
+  
   pReducer->pDesc = pDesc;
   tscTrace("%p the number of merged leaves is: %d", pSql, pReducer->numOfBuffer);
 
@@ -278,6 +278,7 @@ void tscCreateLocalReducer(tExtMemBuffer **pMemBuffer, int32_t numOfBuffer, tOrd
   SQueryInfo *pQueryInfo = tscGetQueryInfoDetail(pCmd, pCmd->clauseIndex);
 
   param->groupOrderType = pQueryInfo->groupbyExpr.orderType;
+  pReducer->orderPrjOnSTable = tscOrderedProjectionQueryOnSTable(pQueryInfo, 0);
 
   pRes->code = tLoserTreeCreate(&pReducer->pLoserTree, pReducer->numOfBuffer, param, treeComparator);
   if (pReducer->pLoserTree == NULL || pRes->code != 0) {
@@ -309,10 +310,10 @@ void tscCreateLocalReducer(tExtMemBuffer **pMemBuffer, int32_t numOfBuffer, tOrd
   pReducer->nResultBufSize = pMemBuffer[0]->pageSize * 16;
   pReducer->pResultBuf = (tFilePage *)calloc(1, pReducer->nResultBufSize + sizeof(tFilePage));
 
-  int32_t finalRowLength = tscGetResRowLength(pQueryInfo->exprList);
+  pReducer->finalRowSize = tscGetResRowLength(pQueryInfo->exprList);
   pReducer->resColModel = finalmodel;
-  pReducer->resColModel->capacity = pReducer->nResultBufSize / finalRowLength;
-  assert(finalRowLength <= pReducer->rowSize);
+  pReducer->resColModel->capacity = pReducer->nResultBufSize / pReducer->finalRowSize;
+  assert(pReducer->finalRowSize <= pReducer->rowSize);
 
   pReducer->pFinalRes = calloc(1, pReducer->rowSize * pReducer->resColModel->capacity);
 //  pReducer->pBufForInterpo = calloc(1, pReducer->nResultBufSize);
@@ -389,7 +390,7 @@ static int32_t tscFlushTmpBufferImpl(tExtMemBuffer *pMemoryBuf, tOrderDescriptor
   assert(pPage->num <= pDesc->pColumnModel->capacity);
 
   // sort before flush to disk, the data must be consecutively put on tFilePage.
-  if (pDesc->orderIdx.numOfCols > 0) {
+  if (pDesc->orderInfo.numOfCols > 0) {
     tColDataQSort(pDesc, pPage->num, 0, pPage->num - 1, pPage->data, orderType);
   }
 
@@ -590,12 +591,10 @@ static int32_t createOrderDescriptor(tOrderDescriptor **pOrderDesc, SSqlCmd *pCm
 bool isSameGroup(SSqlCmd *pCmd, SLocalReducer *pReducer, char *pPrev, tFilePage *tmpBuffer) {
   SQueryInfo *pQueryInfo = tscGetQueryInfoDetail(pCmd, pCmd->clauseIndex);
 
-  int16_t functionId = tscSqlExprGet(pQueryInfo, 0)->functionId;
-
   // disable merge procedure for column projection query
+  int16_t functionId = pReducer->pCtx[0].functionId;
   assert(functionId != TSDB_FUNC_ARITHM);
-
-  if (tscOrderedProjectionQueryOnSTable(pQueryInfo, 0)) {
+  if (pReducer->orderPrjOnSTable) {
     return true;
   }
 
@@ -604,26 +603,33 @@ bool isSameGroup(SSqlCmd *pCmd, SLocalReducer *pReducer, char *pPrev, tFilePage
   }
 
   tOrderDescriptor *pOrderDesc = pReducer->pDesc;
-  int32_t           numOfCols = pOrderDesc->orderIdx.numOfCols;
-
+  SColumnOrderInfo* orderInfo = &pOrderDesc->orderInfo;
+  
   // no group by columns, all data belongs to one group
+  int32_t numOfCols = orderInfo->numOfCols;
   if (numOfCols <= 0) {
     return true;
   }
 
-  if (pOrderDesc->orderIdx.pData[numOfCols - 1] == PRIMARYKEY_TIMESTAMP_COL_INDEX) {  //<= 0
-    // super table interval query
+  if (orderInfo->pData[numOfCols - 1] == PRIMARYKEY_TIMESTAMP_COL_INDEX) {
+    /*
+     * super table interval query
+     * if the order columns is the primary timestamp, all result data belongs to one group
+     */
     assert(pQueryInfo->intervalTime > 0);
-    pOrderDesc->orderIdx.numOfCols -= 1;
+    if (numOfCols == 1) {
+      return true;
+    }
   } else {  // simple group by query
     assert(pQueryInfo->intervalTime == 0);
   }
 
   // only one row exists
-  int32_t ret = compare_a(pOrderDesc, 1, 0, pPrev, 1, 0, tmpBuffer->data);
-  pOrderDesc->orderIdx.numOfCols = numOfCols;
-
-  return (ret == 0);
+  int32_t index = orderInfo->pData[0];
+  int32_t offset = (pOrderDesc->pColumnModel)->pFields[index].offset;
+  
+  int32_t ret = memcmp(pPrev + offset, tmpBuffer->data + offset, pOrderDesc->pColumnModel->rowSize - offset);
+  return ret == 0;
 }
 
 int32_t tscLocalReducerEnvCreate(SSqlObj *pSql, tExtMemBuffer ***pMemBuffer, tOrderDescriptor **pOrderDesc,
@@ -873,24 +879,24 @@ static void reversedCopyFromInterpolationToDstBuf(SQueryInfo *pQueryInfo, SSqlRe
  * Note: pRes->pLocalReducer may be null, due to the fact that "tscDestroyLocalReducer" is called
  * by "interuptHandler" function in shell
  */
-static void doInterpolateResult(SSqlObj *pSql, SLocalReducer *pLocalReducer, bool doneOutput) {
+static void doFillResult(SSqlObj *pSql, SLocalReducer *pLocalReducer, bool doneOutput) {
   SSqlCmd *   pCmd = &pSql->cmd;
   SSqlRes *   pRes = &pSql->res;
   
   tFilePage * pFinalDataPage = pLocalReducer->pResultBuf;
   SQueryInfo *pQueryInfo = tscGetQueryInfoDetail(pCmd, pCmd->clauseIndex);
 
-  if (pRes->pLocalReducer != pLocalReducer) {
-    /*
-     * Release the SSqlObj is called, and it is int destroying function invoked by other thread.
-     * However, the other thread will WAIT until current process fully completes.
-     * Since the flag of release struct is set by doLocalReduce function
-     */
-    assert(pRes->pLocalReducer == NULL);
-  }
+//  if (pRes->pLocalReducer != pLocalReducer) {
+//    /*
+//     * Release the SSqlObj is called, and it is int destroying function invoked by other thread.
+//     * However, the other thread will WAIT until current process fully completes.
+//     * Since the flag of release struct is set by doLocalReduce function
+//     */
+//    assert(pRes->pLocalReducer == NULL);
+//  }
 
+  // no interval query, no fill operation
   if (pQueryInfo->intervalTime == 0 || pQueryInfo->fillType == TSDB_FILL_NONE) {
-    // no interval query, no fill operation
     pRes->data = pLocalReducer->pFinalRes;
     pRes->numOfRows = pFinalDataPage->num;
     pRes->numOfClauseTotal += pRes->numOfRows;
@@ -929,9 +935,7 @@ static void doInterpolateResult(SSqlObj *pSql, SLocalReducer *pLocalReducer, boo
       savePrevRecordAndSetupInterpoInfo(pLocalReducer, pQueryInfo, pLocalReducer->pFillInfo);
     }
 
-    int32_t rowSize = tscGetResRowLength(pQueryInfo->exprList);
-    memcpy(pRes->data, pFinalDataPage->data, pRes->numOfRows * rowSize);
-
+    memcpy(pRes->data, pFinalDataPage->data, pRes->numOfRows * pLocalReducer->finalRowSize);
     pFinalDataPage->num = 0;
     return;
   }
@@ -1037,16 +1041,13 @@ static void savePreviousRow(SLocalReducer *pLocalReducer, tFilePage *tmpBuffer)
 static void doExecuteSecondaryMerge(SSqlCmd *pCmd, SLocalReducer *pLocalReducer, bool needInit) {
   // the tag columns need to be set before all functions execution
   SQueryInfo *pQueryInfo = tscGetQueryInfoDetail(pCmd, pCmd->clauseIndex);
+  
   size_t size = tscSqlExprNumOfExprs(pQueryInfo);
-
   for (int32_t j = 0; j < size; ++j) {
-    SSqlExpr *      pExpr = tscSqlExprGet(pQueryInfo, j);
     SQLFunctionCtx *pCtx = &pLocalReducer->pCtx[j];
 
-    tVariantAssign(&pCtx->param[0], &pExpr->param[0]);
-
     // tags/tags_dummy function, the tag field of SQLFunctionCtx is from the input buffer
-    int32_t functionId = pExpr->functionId;
+    int32_t functionId = pCtx->functionId;
     if (functionId == TSDB_FUNC_TAG_DUMMY || functionId == TSDB_FUNC_TAG || functionId == TSDB_FUNC_TS_DUMMY) {
       tVariantDestroy(&pCtx->tag);
       char* input = pCtx->aInputElemBuf;
@@ -1057,17 +1058,20 @@ static void doExecuteSecondaryMerge(SSqlCmd *pCmd, SLocalReducer *pLocalReducer,
       } else {
         tVariantCreateFromBinary(&pCtx->tag, input, pCtx->inputBytes, pCtx->inputType);
       }
+    } else if (functionId == TSDB_FUNC_TOP || functionId == TSDB_FUNC_BOTTOM) {
+      SSqlExpr *pExpr = tscSqlExprGet(pQueryInfo, j);
+      pCtx->param[0].i64Key = pExpr->param[0].i64Key;
     }
 
     pCtx->currentStage = SECONDARY_STAGE_MERGE;
 
     if (needInit) {
-      aAggs[pExpr->functionId].init(pCtx);
+      aAggs[pCtx->functionId].init(pCtx);
     }
   }
 
   for (int32_t j = 0; j < size; ++j) {
-    int32_t functionId = tscSqlExprGet(pQueryInfo, j)->functionId;
+    int32_t functionId = pLocalReducer->pCtx[j].functionId;
     if (functionId == TSDB_FUNC_TAG_DUMMY || functionId == TSDB_FUNC_TS_DUMMY) {
       continue;
     }
@@ -1101,8 +1105,7 @@ static int64_t getNumOfResultLocal(SQueryInfo *pQueryInfo, SQLFunctionCtx *pCtx)
      * ts, tag, tagprj function can not decide the output number of current query
      * the number of output result is decided by main output
      */
-    SSqlExpr *pExpr = tscSqlExprGet(pQueryInfo, j);
-    int32_t   functionId = pExpr->functionId;
+    int32_t functionId = pCtx[j].functionId;
     if (functionId == TSDB_FUNC_TS || functionId == TSDB_FUNC_TAG || functionId == TSDB_FUNC_TAGPRJ) {
       continue;
     }
@@ -1136,15 +1139,13 @@ static void fillMultiRowsOfTagsVal(SQueryInfo *pQueryInfo, int32_t numOfRes, SLo
 
   char *buf = malloc((size_t)maxBufSize);
   for (int32_t k = 0; k < size; ++k) {
-    SSqlExpr *pExpr = tscSqlExprGet(pQueryInfo, k);
-    if (pExpr->functionId != TSDB_FUNC_TAG) {
+    SQLFunctionCtx *pCtx = &pLocalReducer->pCtx[k];
+    if (pCtx->functionId != TSDB_FUNC_TAG) {
       continue;
     }
 
     int32_t inc = numOfRes - 1;  // tsdb_func_tag function only produce one row of result
     memset(buf, 0, (size_t)maxBufSize);
-
-    SQLFunctionCtx *pCtx = &pLocalReducer->pCtx[k];
     memcpy(buf, pCtx->aOutputBuf, (size_t)pCtx->outputBytes);
 
     for (int32_t i = 0; i < inc; ++i) {
@@ -1160,8 +1161,8 @@ int32_t finalizeRes(SQueryInfo *pQueryInfo, SLocalReducer *pLocalReducer) {
   size_t size = tscSqlExprNumOfExprs(pQueryInfo);
   
   for (int32_t k = 0; k < size; ++k) {
-    SSqlExpr *pExpr = tscSqlExprGet(pQueryInfo, k);
-    aAggs[pExpr->functionId].xFinalize(&pLocalReducer->pCtx[k]);
+    SQLFunctionCtx* pCtx = &pLocalReducer->pCtx[k];
+    aAggs[pCtx->functionId].xFinalize(pCtx);
   }
 
   pLocalReducer->hasPrevRow = false;
@@ -1182,13 +1183,13 @@ int32_t finalizeRes(SQueryInfo *pQueryInfo, SLocalReducer *pLocalReducer) {
  */
 bool needToMerge(SQueryInfo *pQueryInfo, SLocalReducer *pLocalReducer, tFilePage *tmpBuffer) {
   int32_t ret = 0;  // merge all result by default
-  int16_t functionId = tscSqlExprGet(pQueryInfo, 0)->functionId;
-
+  
+  int16_t functionId = pLocalReducer->pCtx[0].functionId;
   if (functionId == TSDB_FUNC_PRJ || functionId == TSDB_FUNC_ARITHM) {  // column projection query
     ret = 1;                                                            // disable merge procedure
   } else {
     tOrderDescriptor *pDesc = pLocalReducer->pDesc;
-    if (pDesc->orderIdx.numOfCols > 0) {
+    if (pDesc->orderInfo.numOfCols > 0) {
       if (pDesc->tsOrder == TSDB_ORDER_ASC) {  // asc
         // todo refactor comparator
         ret = compare_a(pLocalReducer->pDesc, 1, 0, pLocalReducer->prevRowOfInput, 1, 0, tmpBuffer->data);
@@ -1274,7 +1275,7 @@ bool doGenerateFinalResults(SSqlObj *pSql, SLocalReducer *pLocalReducer, bool no
     taosFillCopyInputDataFromOneFilePage(pFillInfo, pResBuf);
   }
   
-  doInterpolateResult(pSql, pLocalReducer, noMoreCurrentGroupRes);
+  doFillResult(pSql, pLocalReducer, noMoreCurrentGroupRes);
   return true;
 }
 
@@ -1341,7 +1342,7 @@ static bool doBuildFilledResultForGroup(SSqlObj *pSql) {
     // the first column must be the timestamp column
     int32_t rows = taosGetNumOfResultWithFill(pFillInfo, remain, ekey, pLocalReducer->resColModel->capacity);
     if (rows > 0) {  // do interpo
-      doInterpolateResult(pSql, pLocalReducer, false);
+      doFillResult(pSql, pLocalReducer, false);
     }
 
     return true;
@@ -1374,7 +1375,7 @@ static bool doHandleLastRemainData(SSqlObj *pSql) {
                                    pQueryInfo->slidingTimeUnit, tinfo.precision);
       int32_t rows = taosGetNumOfResultWithFill(pFillInfo, 0, etime, pLocalReducer->resColModel->capacity);
       if (rows > 0) {  // do interpo
-        doInterpolateResult(pSql, pLocalReducer, true);
+        doFillResult(pSql, pLocalReducer, true);
       }
     }
 
@@ -1408,13 +1409,11 @@ static void doProcessResultInNextWindow(SSqlObj *pSql, int32_t numOfRes) {
   size_t size = tscSqlExprNumOfExprs(pQueryInfo);
 
   for (int32_t k = 0; k < size; ++k) {
-    SSqlExpr *      pExpr = tscSqlExprGet(pQueryInfo, k);
     SQLFunctionCtx *pCtx = &pLocalReducer->pCtx[k];
-
     pCtx->aOutputBuf += pCtx->outputBytes * numOfRes;
 
     // set the correct output timestamp column position
-    if (pExpr->functionId == TSDB_FUNC_TOP || pExpr->functionId == TSDB_FUNC_BOTTOM) {
+    if (pCtx->functionId == TSDB_FUNC_TOP || pCtx->functionId == TSDB_FUNC_BOTTOM) {
       pCtx->ptsOutputBuf = ((char *)pCtx->ptsOutputBuf + TSDB_KEYSIZE * numOfRes);
     }
   }

src/client/src/tscServer.c

@@ -114,6 +114,8 @@ void tscProcessHeartBeatRsp(void *param, TAOS_RES *tres, int code) {
     if (pIpList->numOfIps > 0) 
       tscSetMgmtIpList(pIpList);
 
+    pSql->pTscObj->connId = htonl(pRsp->connId);
+
     if (pRsp->killConnection) {
       tscKillConnection(pObj);
     } else {
@@ -332,9 +334,8 @@ void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcIpSet *pIpSet) {
     rpcMsg->code = (*tscProcessMsgRsp[pCmd->command])(pSql);
   
   if (rpcMsg->code != TSDB_CODE_ACTION_IN_PROGRESS) {
-    rpcMsg->code = pRes->code ? pRes->code : pRes->numOfRows;
+    rpcMsg->code = (pRes->code == TSDB_CODE_SUCCESS) ? pRes->numOfRows: pRes->code;
     
-    tscTrace("%p SQL result:%s res:%p", pSql, tstrerror(pRes->code), pSql);
     bool shouldFree = tscShouldBeFreed(pSql);
     (*pSql->fp)(pSql->param, pSql, rpcMsg->code);
 
@@ -1129,13 +1130,6 @@ int32_t tscBuildKillMsg(SSqlObj *pSql, SSqlInfo *pInfo) {
   SSqlCmd *pCmd = &pSql->cmd;
   pCmd->payloadLen = sizeof(SCMKillQueryMsg);
 
-  if (TSDB_CODE_SUCCESS != tscAllocPayload(pCmd, pCmd->payloadLen)) {
-    tscError("%p failed to malloc for query msg", pSql);
-    return TSDB_CODE_CLI_OUT_OF_MEMORY;
-  }
-
-  SCMKillQueryMsg *pKill = (SCMKillQueryMsg*)pCmd->payload;
-  strncpy(pKill->queryId, pInfo->pDCLInfo->ip.z, pInfo->pDCLInfo->ip.n);
   switch (pCmd->command) {
     case TSDB_SQL_KILL_QUERY:
       pCmd->msgType = TSDB_MSG_TYPE_CM_KILL_QUERY;
@@ -1743,57 +1737,43 @@ int tscBuildSTableVgroupMsg(SSqlObj *pSql, SSqlInfo *pInfo) {
   return TSDB_CODE_SUCCESS;
 }
 
-int tscEstimateHeartBeatMsgLength(SSqlObj *pSql) {
-  int      size = 0;
+int tscBuildHeartBeatMsg(SSqlObj *pSql, SSqlInfo *pInfo) {
+  SSqlCmd *pCmd = &pSql->cmd;
   STscObj *pObj = pSql->pTscObj;
 
-  size += tsRpcHeadSize;
-  size += sizeof(SQqueryList);
+  pthread_mutex_lock(&pObj->mutex);
 
+  int32_t numOfQueries = 2;
   SSqlObj *tpSql = pObj->sqlList;
   while (tpSql) {
-    size += sizeof(SQueryDesc);
     tpSql = tpSql->next;
+    numOfQueries++;
   }
 
-  size += sizeof(SStreamList);
+  int32_t numOfStreams = 2;
   SSqlStream *pStream = pObj->streamList;
   while (pStream) {
-    size += sizeof(SStreamDesc);
     pStream = pStream->next;
+    numOfStreams++;
   }
 
-  return size + TSDB_EXTRA_PAYLOAD_SIZE;
-}
-
-int tscBuildHeartBeatMsg(SSqlObj *pSql, SSqlInfo *pInfo) {
-  char *pMsg, *pStart;
-  int   msgLen = 0;
-  int   size = 0;
-
-  SSqlCmd *pCmd = &pSql->cmd;
-  STscObj *pObj = pSql->pTscObj;
-
-  pthread_mutex_lock(&pObj->mutex);
-
-  size = tscEstimateHeartBeatMsgLength(pSql);
+  int size = numOfQueries * sizeof(SQueryDesc) + numOfStreams * sizeof(SStreamDesc) + sizeof(SCMHeartBeatMsg) + 100;
   if (TSDB_CODE_SUCCESS != tscAllocPayload(pCmd, size)) {
     pthread_mutex_unlock(&pObj->mutex);
     tscError("%p failed to malloc for heartbeat msg", pSql);
     return -1;
   }
 
-  pMsg = pCmd->payload;
-  pStart = pMsg;
+  SCMHeartBeatMsg *pHeartbeat = (SCMHeartBeatMsg *)pCmd->payload;
+  pHeartbeat->numOfQueries = numOfQueries;
+  pHeartbeat->numOfStreams = numOfStreams;
+  int msgLen = tscBuildQueryStreamDesc(pHeartbeat, pObj);
 
-  pMsg = tscBuildQueryStreamDesc(pMsg, pObj);
   pthread_mutex_unlock(&pObj->mutex);
 
-  msgLen = pMsg - pStart;
   pCmd->payloadLen = msgLen;
   pCmd->msgType = TSDB_MSG_TYPE_CM_HEARTBEAT;
 
-  assert(msgLen + minMsgSize() <= size);
   return TSDB_CODE_SUCCESS;
 }
 
@@ -2204,6 +2184,7 @@ int tscProcessConnectRsp(SSqlObj *pSql) {
   strcpy(pObj->sversion, pConnect->serverVersion);
   pObj->writeAuth = pConnect->writeAuth;
   pObj->superAuth = pConnect->superAuth;
+  pObj->connId = htonl(pConnect->connId);
   taosTmrReset(tscProcessActivityTimer, tsShellActivityTimer * 500, pObj, tscTmr, &pObj->pTimer);
 
   return 0;
@@ -2330,7 +2311,7 @@ int tscProcessRetrieveRspFromNode(SSqlObj *pSql) {
   }
 
   pRes->row = 0;
-  tscTrace("%p numOfRows:%d, offset:%d", pSql, pRes->numOfRows, pRes->offset);
+  tscTrace("%p numOfRows:%d, offset:%d, complete:%d", pSql, pRes->numOfRows, pRes->offset, pRes->completed);
 
   return 0;
 }

src/client/src/tscSql.c

@@ -88,7 +88,6 @@ SSqlObj *taosConnectImpl(const char *ip, const char *user, const char *pass, con
 
   strncpy(pObj->user, user, TSDB_USER_LEN);
   taosEncryptPass((uint8_t *)pass, strlen(pass), pObj->pass);
-  pObj->mnodePort = port ? port : tsDnodeShellPort;
 
   if (db) {
     int32_t len = strlen(db);

src/client/src/tscSubquery.c

@@ -1557,8 +1557,8 @@ static void tscRetrieveFromDnodeCallBack(void *param, TAOS_RES *tres, int numOfR
     assert(pRes->numOfRows == numOfRows);
     int64_t num = atomic_add_fetch_64(&pState->numOfRetrievedRows, numOfRows);
     
-//    tscTrace("%p sub:%p retrieve numOfRows:%d totalNumOfRows:%d from ip:%u,vid:%d,orderOfSub:%d", pPObj, pSql,
-//             pRes->numOfRows, pState->numOfRetrievedRows, pSvd->ip, pSvd->vnode, idx);
+    tscTrace("%p sub:%p retrieve numOfRows:%d totalNumOfRows:%d from ip:%s, orderOfSub:%d", pPObj, pSql,
+             pRes->numOfRows, pState->numOfRetrievedRows, pSql->ipList.fqdn[pSql->ipList.inUse], idx);
     
     if (num > tsMaxNumOfOrderedResults && tscIsProjectionQueryOnSTable(pQueryInfo, 0)) {
       tscError("%p sub:%p num of OrderedRes is too many, max allowed:%" PRId64 " , current:%" PRId64,
@@ -1713,6 +1713,11 @@ static void multiVnodeInsertMerge(void* param, TAOS_RES* tres, int numOfRows) {
   // increase the total inserted rows
   if (numOfRows > 0) {
     pParentObj->res.numOfRows += numOfRows;
+  } else {
+    SSqlObj* pSql = (SSqlObj*) tres;
+    assert(pSql != NULL && pSql->res.code == numOfRows);
+    
+    pParentObj->res.code = pSql->res.code;
   }
   
   taos_free_result(tres);
@@ -1947,7 +1952,8 @@ void **doSetResultRowData(SSqlObj *pSql, bool finalResult) {
   
   SQueryInfo *pQueryInfo = tscGetQueryInfoDetail(pCmd, pCmd->clauseIndex);
   
-  for (int i = 0; i < tscNumOfFields(pQueryInfo); ++i) {
+  size_t size = tscNumOfFields(pQueryInfo);
+  for (int i = 0; i < size; ++i) {
     SFieldSupInfo* pSup = tscFieldInfoGetSupp(&pQueryInfo->fieldsInfo, i);
     if (pSup->pSqlExpr != NULL) {
       tscGetResultColumnChr(pRes, &pQueryInfo->fieldsInfo, i);

src/client/src/tscUtil.c

@@ -404,8 +404,6 @@ void tscPartiallyFreeSqlObj(SSqlObj* pSql) {
   pSql->numOfSubs = 0;
   
   tscResetSqlCmdObj(pCmd);
-  
-  tscTrace("%p partially free sqlObj completed", pSql);
 }
 
 void tscFreeSqlObj(SSqlObj* pSql) {
@@ -2104,7 +2102,7 @@ void tscTryQueryNextClause(SSqlObj* pSql, void (*queryFp)()) {
 }
 
 void tscGetResultColumnChr(SSqlRes* pRes, SFieldInfo* pFieldInfo, int32_t columnIndex) {
-  SFieldSupInfo* pInfo = tscFieldInfoGetSupp(pFieldInfo, columnIndex);
+  SFieldSupInfo* pInfo = taosArrayGet(pFieldInfo->pSupportInfo, columnIndex);//tscFieldInfoGetSupp(pFieldInfo, columnIndex);
   assert(pInfo->pSqlExpr != NULL);
 
   int32_t type = pInfo->pSqlExpr->resType;

src/common/inc/tdataformat.h

@@ -251,7 +251,7 @@ void       tdFreeDataCols(SDataCols *pCols);
 void       tdAppendDataRowToDataCol(SDataRow row, STSchema *pSchema, SDataCols *pCols);
 void       tdPopDataColsPoints(SDataCols *pCols, int pointsToPop);  //!!!!
 int        tdMergeDataCols(SDataCols *target, SDataCols *src, int rowsToMerge);
-void       tdMergeTwoDataCols(SDataCols *target, SDataCols *src1, int *iter1, SDataCols *src2, int *iter2, int tRows);
+void       tdMergeTwoDataCols(SDataCols *target, SDataCols *src1, int *iter1, int limit1, SDataCols *src2, int *iter2, int limit2, int tRows);
 
 // ----------------- K-V data row structure
 /*

src/common/src/tdataformat.c

@@ -450,7 +450,8 @@ int tdMergeDataCols(SDataCols *target, SDataCols *source, int rowsToMerge) {
 
     int iter1 = 0;
     int iter2 = 0;
-    tdMergeTwoDataCols(target, pTarget, &iter1, source, &iter2, pTarget->numOfRows + rowsToMerge);
+    tdMergeTwoDataCols(target, pTarget, &iter1, pTarget->numOfRows, source, &iter2, source->numOfRows,
+                       pTarget->numOfRows + rowsToMerge);
   }
 
   tdFreeDataCols(pTarget);
@@ -461,15 +462,15 @@ _err:
   return -1;
 }
 
-void tdMergeTwoDataCols(SDataCols *target, SDataCols *src1, int *iter1, SDataCols *src2, int *iter2, int tRows) {
-  // TODO: add resolve duplicate key here
+void tdMergeTwoDataCols(SDataCols *target, SDataCols *src1, int *iter1, int limit1, SDataCols *src2, int *iter2, int limit2, int tRows) {
   tdResetDataCols(target);
+  ASSERT(limit1 <= src1->numOfRows && limit2 <= src2->numOfRows);
 
   while (target->numOfRows < tRows) {
-    if (*iter1 >= src1->numOfRows && *iter2 >= src2->numOfRows) break;
+    if (*iter1 >= limit1 && *iter2 >= limit2) break;
 
-    TSKEY key1 = (*iter1 >= src1->numOfRows) ? INT64_MAX : ((TSKEY *)(src1->cols[0].pData))[*iter1];
-    TSKEY key2 = (*iter2 >= src2->numOfRows) ? INT64_MAX : ((TSKEY *)(src2->cols[0].pData))[*iter2];
+    TSKEY key1 = (*iter1 >= limit1) ? INT64_MAX : ((TSKEY *)(src1->cols[0].pData))[*iter1];
+    TSKEY key2 = (*iter2 >= limit2) ? INT64_MAX : ((TSKEY *)(src2->cols[0].pData))[*iter2];
 
     if (key1 <= key2) {
       for (int i = 0; i < src1->numOfCols; i++) {

src/common/src/tglobal.c

@@ -112,7 +112,7 @@ int32_t tsMaxShellConns = 5000;
 char    tsDefaultDB[TSDB_DB_NAME_LEN] = {0};
 char    tsDefaultUser[64] = "root";
 char    tsDefaultPass[64] = "taosdata";
-int32_t tsMaxConnections = 50;
+int32_t tsMaxConnections = 5000;
 
 int32_t tsBalanceInterval = 300;  // seconds
 int32_t tsOfflineThreshold = 86400*100;   // seconds 10days
@@ -840,7 +840,7 @@ static void doInitGlobalConfig() {
   cfg.valType = TAOS_CFG_VTYPE_INT32;
   cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW;
   cfg.minValue = 1;
-  cfg.maxValue = 100;
+  cfg.maxValue = 100000;
   cfg.ptrLength = 0;
   cfg.unitType = TAOS_CFG_UTYPE_NONE;
   taosInitConfigOption(cfg);

src/connector/python/linux/python2/taos/cinterface.py

@@ -150,6 +150,7 @@ class CTaosInterface(object):
     libtaos.taos_fetch_lengths.restype = ctypes.c_void_p
     libtaos.taos_free_result.restype = None
     libtaos.taos_errno.restype = ctypes.c_int
+    libtaos.taos_query.restype = ctypes.POINTER(ctypes.c_void_p)
 
     def __init__(self, config=None):
         '''

src/connector/python/linux/python3/taos/cinterface.py

@@ -150,6 +150,7 @@ class CTaosInterface(object):
     libtaos.taos_fetch_lengths.restype = ctypes.c_void_p
     libtaos.taos_free_result.restype = None
     libtaos.taos_errno.restype = ctypes.c_int
+    libtaos.taos_query.restype = ctypes.POINTER(ctypes.c_void_p)
 
     def __init__(self, config=None):
         '''

src/connector/python/windows/python2/taos/cinterface.py

@@ -150,6 +150,7 @@ class CTaosInterface(object):
     libtaos.taos_fetch_lengths.restype = ctypes.c_void_p
     libtaos.taos_free_result.restype = None
     libtaos.taos_errno.restype = ctypes.c_int
+    libtaos.taos_query.restype = ctypes.POINTER(ctypes.c_void_p)
 
     def __init__(self, config=None):
         '''

src/connector/python/windows/python3/taos/cinterface.py

@@ -150,6 +150,7 @@ class CTaosInterface(object):
     libtaos.taos_fetch_lengths.restype = ctypes.c_void_p
     libtaos.taos_free_result.restype = None
     libtaos.taos_errno.restype = ctypes.c_int
+    libtaos.taos_query.restype = ctypes.POINTER(ctypes.c_void_p)
 
     def __init__(self, config=None):
         '''

src/dnode/src/dnodeMain.c

@@ -46,7 +46,7 @@ typedef struct {
   void              (*cleanup)();
 } SDnodeComponent;
 
-static const SDnodeComponent SDnodeComponents[] = {
+static const SDnodeComponent tsDnodeComponents[] = {
   {"storage", dnodeInitStorage,    dnodeCleanupStorage},
   {"vread",   dnodeInitVnodeRead,  dnodeCleanupVnodeRead},
   {"vwrite",  dnodeInitVnodeWrite, dnodeCleanupVnodeWrite},
@@ -72,14 +72,14 @@ static int dnodeCreateDir(const char *dir) {
 
 static void dnodeCleanupComponents(int32_t stepId) {
   for (int32_t i = stepId; i >= 0; i--) {
-    SDnodeComponents[i].cleanup();
+    tsDnodeComponents[i].cleanup();
   }
 }
 
 static int32_t dnodeInitComponents() {
   int32_t code = 0;
-  for (int32_t i = 0; i < sizeof(SDnodeComponents) / sizeof(SDnodeComponents[0]); i++) {
-    if (SDnodeComponents[i].init() != 0) {
+  for (int32_t i = 0; i < sizeof(tsDnodeComponents) / sizeof(tsDnodeComponents[0]); i++) {
+    if (tsDnodeComponents[i].init() != 0) {
       dnodeCleanupComponents(i);
       code = -1;
       break;
@@ -133,7 +133,7 @@ int32_t dnodeInitSystem() {
 void dnodeCleanUpSystem() {
   if (dnodeGetRunStatus() != TSDB_DNODE_RUN_STATUS_STOPPED) {
     dnodeSetRunStatus(TSDB_DNODE_RUN_STATUS_STOPPED);
-    dnodeCleanupComponents(sizeof(SDnodeComponents) / sizeof(SDnodeComponents[0]) - 1);
+    dnodeCleanupComponents(sizeof(tsDnodeComponents) / sizeof(tsDnodeComponents[0]) - 1);
     taos_cleanup();
     taosCloseLog();
   }

src/inc/taosdef.h

@@ -231,7 +231,8 @@ void tsDataSwap(void *pLeft, void *pRight, int32_t type, int32_t size);
 #define TSDB_SHELL_VNODE_BITS     24
 #define TSDB_SHELL_SID_MASK       0xFF
 #define TSDB_HTTP_TOKEN_LEN       20
-#define TSDB_SHOW_SQL_LEN         512
+#define TSDB_SHOW_SQL_LEN         64
+#define TSDB_SLOW_QUERY_SQL_LEN   512
 
 #define TSDB_METER_STATE_OFFLINE  0
 #define TSDB_METER_STATE_ONLLINE  1

src/inc/taoserror.h

@@ -150,6 +150,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_NO_DISK_PERMISSIONS,        0, 0x0405, "no disk perm
 TAOS_DEFINE_ERROR(TSDB_CODE_FILE_CORRUPTED,             0, 0x0406, "file corrupted")
 TAOS_DEFINE_ERROR(TSDB_CODE_MEMORY_CORRUPTED,           0, 0x0407, "memory corrupted")
 TAOS_DEFINE_ERROR(TSDB_CODE_NOT_SUCH_FILE_OR_DIR,       0, 0x0408, "no such file or directory")
+TAOS_DEFINE_ERROR(TSDB_CODE_TOO_MANY_SHELL_CONNS,       0, 0x0409, "too many shell conns")
 
 // client
 TAOS_DEFINE_ERROR(TSDB_CODE_INVALID_CLIENT_VERSION,     0, 0x0481, "invalid client version")

src/inc/taosmsg.h

@@ -137,6 +137,7 @@ enum _mgmt_table {
   TSDB_MGMT_TABLE_SCORES,
   TSDB_MGMT_TABLE_GRANTS,
   TSDB_MGMT_TABLE_VNODES,
+  TSDB_MGMT_TABLE_STREAMTABLES,
   TSDB_MGMT_TABLE_MAX,
 };
 
@@ -299,6 +300,9 @@ typedef struct {
   char      serverVersion[TSDB_VERSION_LEN];
   int8_t    writeAuth;
   int8_t    superAuth;
+  int8_t    reserved1;
+  int8_t    reserved2;
+  int32_t   connId;
   SRpcIpSet ipList;
 } SCMConnectRsp;
 
@@ -716,16 +720,10 @@ typedef struct {
 } SStreamDesc;
 
 typedef struct {
-  int32_t    numOfQueries;
-} SQqueryList;
-
-typedef struct {
-  int32_t     numOfStreams;
-} SStreamList;
-
-typedef struct {
-  SQqueryList qlist;
-  SStreamList slist;
+  uint32_t connId;
+  int32_t  numOfQueries;
+  int32_t  numOfStreams;
+  char     pData[];
 } SCMHeartBeatMsg;
 
 typedef struct {
@@ -733,6 +731,7 @@ typedef struct {
   uint32_t  streamId;
   uint32_t  totalDnodes;
   uint32_t  onlineDnodes;
+  uint32_t  connId;
   int8_t    killConnection;
   SRpcIpSet ipList;
 } SCMHeartBeatRsp;

src/inc/tsdb.h

@@ -188,9 +188,10 @@ typedef void *TsdbPosT;
  * @param tsdb       tsdb handle
  * @param pCond      query condition, including time window, result set order, and basic required columns for each block
  * @param groupInfo  tableId list in the form of set, seperated into different groups according to group by condition
+ * @param qinfo      query info handle from query processor
  * @return
  */
-TsdbQueryHandleT *tsdbQueryTables(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STableGroupInfo *groupInfo);
+TsdbQueryHandleT *tsdbQueryTables(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STableGroupInfo *groupInfo, void* qinfo);
 
 /**
  * Get the last row of the given query time window for all the tables in STableGroupInfo object.
@@ -202,11 +203,11 @@ TsdbQueryHandleT *tsdbQueryTables(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STable
  * @param groupInfo   tableId list.
  * @return
  */
-TsdbQueryHandleT tsdbQueryLastRow(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STableGroupInfo *groupInfo);
+TsdbQueryHandleT tsdbQueryLastRow(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STableGroupInfo *groupInfo, void* qinfo);
 
 SArray* tsdbGetQueriedTableIdList(TsdbQueryHandleT *pHandle);
 
-TsdbQueryHandleT tsdbQueryRowsInExternalWindow(TsdbRepoT *tsdb, STsdbQueryCond* pCond, STableGroupInfo *groupList);
+TsdbQueryHandleT tsdbQueryRowsInExternalWindow(TsdbRepoT *tsdb, STsdbQueryCond* pCond, STableGroupInfo *groupList, void* qinfo);
 
 /**
  * move to next block if exists

src/inc/ttokendef.h

@@ -148,8 +148,8 @@
 #define TK_SET                            130
 #define TK_KILL                           131
 #define TK_CONNECTION                     132
-#define TK_COLON                          133
-#define TK_STREAM                         134
+#define TK_STREAM                         133
+#define TK_COLON                          134
 #define TK_ABORT                          135
 #define TK_AFTER                          136
 #define TK_ATTACH                         137

src/mnode/inc/mnodeDef.h

@@ -182,8 +182,6 @@ typedef struct SUserObj {
   int8_t            updateEnd[1];
   int32_t           refCount;
   struct SAcctObj * pAcct;
-  SQqueryList *     pQList;  // query list
-  SStreamList *     pSList;  // stream list
 } SUserObj;
 
 typedef struct {

src/mnode/inc/mnodeProfile.h

@@ -21,9 +21,30 @@ extern "C" {
 #endif
 #include "mnodeDef.h"
 
+typedef struct {
+  char     user[TSDB_USER_LEN + 1];
+  int8_t   killed;
+  uint16_t port;
+  uint32_t ip;
+  uint32_t connId;
+  uint64_t stime;
+  uint64_t lastAccess;
+  uint32_t queryId;
+  uint32_t streamId;
+  int32_t  numOfQueries;
+  int32_t  numOfStreams;
+  SStreamDesc *pStreams;
+  SQueryDesc * pQueries;
+} SConnObj;
+
 int32_t mnodeInitProfile();
 void    mnodeCleanupProfile();
 
+SConnObj *mnodeCreateConn(char *user, uint32_t ip, uint16_t port);
+SConnObj *mnodeAccquireConn(uint32_t connId, char *user, uint32_t ip, uint16_t port);
+void      mnodeReleaseConn(SConnObj *pConn);
+int32_t   mnodeSaveQueryStreamList(SConnObj *pConn, SCMHeartBeatMsg *pHBMsg);
+
 #ifdef __cplusplus
 }
 #endif

src/mnode/inc/mnodeShow.h

@@ -28,6 +28,7 @@ typedef int32_t (*SShowMetaFp)(STableMetaMsg *pMeta, SShowObj *pShow, void *pCon
 typedef int32_t (*SShowRetrieveFp)(SShowObj *pShow, char *data, int32_t rows, void *pConn);
 void mnodeAddShowMetaHandle(uint8_t showType, SShowMetaFp fp);
 void mnodeAddShowRetrieveHandle(uint8_t showType, SShowRetrieveFp fp);
+void mnodeVacuumResult(char *data, int32_t numOfCols, int32_t rows, int32_t capacity, SShowObj *pShow);
 
 #ifdef __cplusplus
 }

src/mnode/src/mnodeMain.c

@@ -33,14 +33,54 @@
 #include "mnodeUser.h"
 #include "mnodeTable.h"
 #include "mnodeShow.h"
+#include "mnodeProfile.h"
+
+typedef struct {
+  const char *const name;
+  int               (*init)();
+  void              (*cleanup)();
+} SMnodeComponent;
 
 void *tsMnodeTmr;
 static bool tsMgmtIsRunning = false;
 
+static const SMnodeComponent tsMnodeComponents[] = {
+  {"profile", mnodeInitProfile, mnodeCleanupProfile},
+  {"accts",   mnodeInitAccts,   mnodeCleanupAccts},
+  {"users",   mnodeInitUsers,   mnodeCleanupUsers},
+  {"dnodes",  mnodeInitDnodes,  mnodeCleanupDnodes},
+  {"dbs",     mnodeInitDbs,     mnodeCleanupDbs},
+  {"vgroups", mnodeInitVgroups, mnodeCleanupVgroups},
+  {"tables",  mnodeInitTables,  mnodeCleanupTables},  
+  {"mnodes",  mnodeInitMnodes,  mnodeCleanupMnodes},
+  {"sdb",     sdbInit,          sdbCleanUp},
+  {"balance", balanceInit,      balanceCleanUp},
+  {"grant",   grantInit,        grantCleanUp},
+  {"show",    mnodeInitShow,    mnodeCleanUpShow}
+};
+
 static void mnodeInitTimer();
 static void mnodeCleanupTimer();
 static bool mnodeNeedStart() ;
 
+static void mnodeCleanupComponents(int32_t stepId) {
+  for (int32_t i = stepId; i >= 0; i--) {
+    tsMnodeComponents[i].cleanup();
+  }
+}
+
+static int32_t mnodeInitComponents() {
+  int32_t code = 0;
+  for (int32_t i = 0; i < sizeof(tsMnodeComponents) / sizeof(tsMnodeComponents[0]); i++) {
+    if (tsMnodeComponents[i].init() != 0) {
+      mnodeCleanupComponents(i);
+      code = -1;
+      break;
+    }
+  }
+  return code;
+}
+
 int32_t mnodeStartSystem() {
   if (tsMgmtIsRunning) {
     mPrint("mnode module already started...");
@@ -57,57 +97,7 @@ int32_t mnodeStartSystem() {
   dnodeAllocateMnodeRqueue();
   dnodeAllocateMnodePqueue();
 
-  if (mnodeInitAccts() < 0) {
-    mError("failed to init accts");
-    return -1;
-  }
-
-  if (mnodeInitUsers() < 0) {
-    mError("failed to init users");
-    return -1;
-  }
-
-  if (mnodeInitDnodes() < 0) {
-    mError("failed to init dnodes");
-    return -1;
-  }
-
-  if (mnodeInitDbs() < 0) {
-    mError("failed to init dbs");
-    return -1;
-  }
-
-  if (mnodeInitVgroups() < 0) {
-    mError("failed to init vgroups");
-    return -1;
-  }
-
-  if (mnodeInitTables() < 0) {
-    mError("failed to init tables");
-    return -1;
-  }
-
-  if (mnodeInitMnodes() < 0) {
-    mError("failed to init mnodes");
-    return -1;
-  }
-
-  if (sdbInit() < 0) {
-    mError("failed to init sdb");
-    return -1;
-  }
-
-  if (balanceInit() < 0) {
-    mError("failed to init balance")
-  }
-
-  if (grantInit() < 0) {
-    mError("failed to init grant");
-    return -1;
-  }
-
-  if (mnodeInitShow() < 0) {
-    mError("failed to init show");
+  if (mnodeInitComponents() != 0) {
     return -1;
   }
 
@@ -115,7 +105,6 @@ int32_t mnodeStartSystem() {
   tsMgmtIsRunning = true;
 
   mPrint("mnode is initialized successfully");
-
   return 0;
 }
 
@@ -133,17 +122,8 @@ void mnodeCleanupSystem() {
   dnodeFreeMnodeRqueue();
   dnodeFreeMnodePqueue();
   mnodeCleanupTimer();
-  mnodeCleanUpShow();
-  grantCleanUp();
-  balanceCleanUp();
-  sdbCleanUp();
-  mnodeCleanupMnodes();
-  mnodeCleanupTables();
-  mnodeCleanupVgroups();
-  mnodeCleanupDbs();
-  mnodeCleanupDnodes();
-  mnodeCleanupUsers();
-  mnodeCleanupAccts();
+  mnodeCleanupComponents(sizeof(tsMnodeComponents) / sizeof(tsMnodeComponents[0]) - 1);
+
   mPrint("mnode is cleaned up");
 }
 

src/mnode/src/mnodeProfile.c

@@ -18,6 +18,10 @@
 #include "taosmsg.h"
 #include "taoserror.h"
 #include "tutil.h"
+#include "ttime.h"
+#include "tcache.h"
+#include "tglobal.h"
+#include "tdataformat.h"
 #include "mnode.h"
 #include "mnodeDef.h"
 #include "mnodeInt.h"
@@ -32,153 +36,170 @@
 #include "mnodeVgroup.h"
 #include "mnodeWrite.h"
 
-int32_t mnodeSaveQueryStreamList(SCMHeartBeatMsg *pHBMsg);
-
-int32_t mnodeKillQuery(char *qidstr, void *pConn);
-int32_t mnodeKillStream(char *qidstr, void *pConn);
-int32_t mnodeKillConnection(char *qidstr, void *pConn);
-
-typedef struct {
-  char     user[TSDB_TABLE_ID_LEN + 1];
-  uint64_t stime;
-  uint32_t ip;
-  uint16_t port;
-} SConnInfo;
-
-typedef struct {
-  int       numOfConns;
-  int       index;
-  SConnInfo connInfo[];
-} SConnShow;
-
-typedef struct {
-  uint32_t ip;
-  uint16_t port;
-  char     user[TSDB_TABLE_ID_LEN+ 1];
-} SCDesc;
-
-typedef struct {
-  int32_t      index;
-  int32_t      numOfQueries;
-  SCDesc * connInfo;
-  SCDesc **cdesc;
-  SQueryDesc   qdesc[];
-} SQueryShow;
-
-typedef struct {
-  int32_t      index;
-  int32_t      numOfStreams;
-  SCDesc * connInfo;
-  SCDesc **cdesc;
-  SStreamDesc   sdesc[];
-} SStreamShow;
-
-int32_t  mgmtSaveQueryStreamList(SCMHeartBeatMsg *pHBMsg) {
-//  SAcctObj *pAcct = pConn->pAcct;
-//
-//  if (contLen <= 0 || pAcct == NULL) {
-//    return 0;
-//  }
-//
-//  pthread_mutex_lock(&pAcct->mutex);
-//
-//  if (pConn->pQList) {
-//    pAcct->acctInfo.numOfQueries -= pConn->pQList->numOfQueries;
-//    pAcct->acctInfo.numOfStreams -= pConn->pSList->numOfStreams;
-//  }
-//
-//  pConn->pQList = realloc(pConn->pQList, contLen);
-//  memcpy(pConn->pQList, cont, contLen);
-//
-//  pConn->pSList = (SStreamList *)(((char *)pConn->pQList) + pConn->pQList->numOfQueries * sizeof(SQueryDesc) + sizeof(SQqueryList));
-//
-//  pAcct->acctInfo.numOfQueries += pConn->pQList->numOfQueries;
-//  pAcct->acctInfo.numOfStreams += pConn->pSList->numOfStreams;
-//
-//  pthread_mutex_unlock(&pAcct->mutex);
+#define CONN_KEEP_TIME  (tsShellActivityTimer * 3)
+#define CONN_CHECK_TIME (tsShellActivityTimer * 2)
+#define QUERY_ID_SIZE   20
+
+extern void *tsMnodeTmr;
+static SCacheObj *tsMnodeConnCache = NULL;
+static uint32_t tsConnIndex = 0;
+
+static int32_t mnodeGetQueryMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
+static int32_t mnodeRetrieveQueries(SShowObj *pShow, char *data, int32_t rows, void *pConn);
+static int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
+static int32_t mnodeRetrieveConns(SShowObj *pShow, char *data, int32_t rows, void *pConn);
+static int32_t mnodeGetStreamMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
+static int32_t mnodeRetrieveStreams(SShowObj *pShow, char *data, int32_t rows, void *pConn);
+static void    mnodeFreeConn(void *data);
+static int32_t mnodeProcessKillQueryMsg(SMnodeMsg *pMsg);
+static int32_t mnodeProcessKillStreamMsg(SMnodeMsg *pMsg);
+static int32_t mnodeProcessKillConnectionMsg(SMnodeMsg *pMsg);
 
-  return TSDB_CODE_SUCCESS;
+int32_t mnodeInitProfile() {
+  mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_QUERIES, mnodeGetQueryMeta);
+  mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_QUERIES, mnodeRetrieveQueries);
+  mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_CONNS, mnodeGetConnsMeta);
+  mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_CONNS, mnodeRetrieveConns);
+  mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_STREAMS, mnodeGetStreamMeta);
+  mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_STREAMS, mnodeRetrieveStreams);
+
+  mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_QUERY, mnodeProcessKillQueryMsg);
+  mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_STREAM, mnodeProcessKillStreamMsg);
+  mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_CONN, mnodeProcessKillConnectionMsg);
+
+  tsMnodeConnCache = taosCacheInitWithCb(tsMnodeTmr, CONN_CHECK_TIME, mnodeFreeConn);
+  return 0;
 }
 
-int32_t mnodeGetQueries(SShowObj *pShow, void *pConn) {
-//  SAcctObj *  pAcct = pConn->pAcct;
-//  SQueryShow *pQueryShow;
-//
-//  pthread_mutex_lock(&pAcct->mutex);
-//
-//  pQueryShow = malloc(sizeof(SQueryDesc) * pAcct->acctInfo.numOfQueries + sizeof(SQueryShow));
-//  pQueryShow->numOfQueries = 0;
-//  pQueryShow->index = 0;
-//  pQueryShow->connInfo = NULL;
-//  pQueryShow->cdesc = NULL;
-//
-//  if (pAcct->acctInfo.numOfQueries > 0) {
-//    pQueryShow->connInfo = (SCDesc *)malloc(pAcct->acctInfo.numOfConns * sizeof(SCDesc));
-//    pQueryShow->cdesc = (SCDesc **)malloc(pAcct->acctInfo.numOfQueries * sizeof(SCDesc *));
-//
-//    pConn = pAcct->pConn;
-//    SQueryDesc * pQdesc = pQueryShow->qdesc;
-//    SCDesc * pCDesc = pQueryShow->connInfo;
-//    SCDesc **ppCDesc = pQueryShow->cdesc;
-//
-//    while (pConn) {
-//      if (pConn->pQList && pConn->pQList->numOfQueries > 0) {
-//        pCDesc->ip = pConn->ip;
-//        pCDesc->port = pConn->port;
-//        strcpy(pCDesc->user, pConn->pUser->user);
-//
-//        memcpy(pQdesc, pConn->pQList->qdesc, sizeof(SQueryDesc) * pConn->pQList->numOfQueries);
-//        pQdesc += pConn->pQList->numOfQueries;
-//        pQueryShow->numOfQueries += pConn->pQList->numOfQueries;
-//        for (int32_t i = 0; i < pConn->pQList->numOfQueries; ++i, ++ppCDesc) *ppCDesc = pCDesc;
-//
-//        pCDesc++;
-//      }
-//      pConn = pConn->next;
-//    }
-//  }
-//
-//  pthread_mutex_unlock(&pAcct->mutex);
-//
-//  // sorting based on useconds
-//
-//  pShow->pIter = pQueryShow;
+void mnodeCleanupProfile() {
+  if (tsMnodeConnCache != NULL) {
+    mPrint("conn cache is cleanup");
+    taosCacheCleanup(tsMnodeConnCache);
+    tsMnodeConnCache = NULL;
+  }
+}
 
-  return 0;
+SConnObj *mnodeCreateConn(char *user, uint32_t ip, uint16_t port) {
+  int32_t connSize = taosHashGetSize(tsMnodeConnCache->pHashTable);
+  if (connSize > tsMaxShellConns) {
+    mError("failed to create conn for user:%s ip:%s:%u, conns:%d larger than maxShellConns:%d, ", user, taosIpStr(ip),
+           port, connSize, tsMaxShellConns);
+    terrno = TSDB_CODE_TOO_MANY_SHELL_CONNS;
+    return NULL;
+  }
+
+  uint32_t connId = atomic_add_fetch_32(&tsConnIndex, 1);
+  if (connId == 0) atomic_add_fetch_32(&tsConnIndex, 1);
+
+  SConnObj connObj = {
+    .ip     = ip,
+    .port   = port,
+    .connId = connId,
+    .stime  = taosGetTimestampMs()
+  };
+  strcpy(connObj.user, user);
+  
+  char key[10];
+  sprintf(key, "%u", connId);  
+  SConnObj *pConn = taosCachePut(tsMnodeConnCache, key, &connObj, sizeof(connObj), CONN_KEEP_TIME);
+  
+  mTrace("connId:%d, is created, user:%s ip:%s:%u", connId, user, taosIpStr(ip), port);
+  return pConn;
 }
 
-int32_t mnodeGetQueryMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
-  int32_t cols = 0;
+void mnodeReleaseConn(SConnObj *pConn) {
+  if(pConn == NULL) return;
+  taosCacheRelease(tsMnodeConnCache, (void**)&pConn, false);
+}
+
+SConnObj *mnodeAccquireConn(uint32_t connId, char *user, uint32_t ip, uint16_t port) {
+  char key[10];
+  sprintf(key, "%u", connId);
+  uint64_t expireTime = CONN_KEEP_TIME * 1000 + (uint64_t)taosGetTimestampMs();
+
+  SConnObj *pConn = taosCacheUpdateExpireTimeByName(tsMnodeConnCache, key, expireTime);
+  if (pConn == NULL) {
+    mError("connId:%d, is already destroyed, user:%s ip:%s:%u", connId, user, taosIpStr(ip), port);
+    return NULL;
+  }
+
+  if (pConn->ip != ip || pConn->port != port /* || strcmp(pConn->user, user) != 0 */) {
+    mError("connId:%d, incoming conn user:%s ip:%s:%u, not match exist conn user:%s ip:%s:%u", connId, user,
+           taosIpStr(ip), port, pConn->user, taosIpStr(pConn->ip), pConn->port);
+    taosCacheRelease(tsMnodeConnCache, (void **)&pConn, false);
+    return NULL;
+  }
+
+  // mTrace("connId:%d, is incoming, user:%s ip:%s:%u", connId, pConn->user, taosIpStr(pConn->ip), pConn->port);
+  pConn->lastAccess = expireTime;
+  return pConn;
+}
+
+static void mnodeFreeConn(void *data) {
+  SConnObj *pConn = data;
+  tfree(pConn->pQueries);
+  tfree(pConn->pQueries);
+
+  mTrace("connId:%d, is destroyed", pConn->connId);
+}
 
+static void *mnodeGetNextConn(SHashMutableIterator *pIter, SConnObj **pConn) {
+  *pConn = NULL;
+
+  if (pIter == NULL) {
+    pIter = taosHashCreateIter(tsMnodeConnCache->pHashTable);
+  }
+
+  if (!taosHashIterNext(pIter)) {
+    taosHashDestroyIter(pIter);
+    return NULL;
+  }
+
+  SCacheDataNode **pNode = taosHashIterGet(pIter);
+  if (pNode == NULL || *pNode == NULL) {
+    taosHashDestroyIter(pIter);
+    return NULL;
+  }
+
+  *pConn = (SConnObj*)((*pNode)->data);
+  return pIter;
+}
+
+static int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
+  SUserObj *pUser = mnodeGetUserFromConn(pConn);
+  if (pUser == NULL) return 0;
+  if (strcmp(pUser->user, "root") != 0) return TSDB_CODE_NO_RIGHTS;
+  
+  int32_t cols = 0;
   SSchema *pSchema = pMeta->schema;
 
-  pShow->bytes[cols] = TSDB_USER_LEN;
+  pShow->bytes[cols] = 4;
+  pSchema[cols].type = TSDB_DATA_TYPE_INT;
+  strcpy(pSchema[cols].name, "connId");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_USER_LEN + VARSTR_HEADER_SIZE;
   pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
   strcpy(pSchema[cols].name, "user");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
-  pShow->bytes[cols] = TSDB_IPv4ADDR_LEN + 14;
+  pShow->bytes[cols] = TSDB_IPv4ADDR_LEN + 6 + VARSTR_HEADER_SIZE;
   pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
-  strcpy(pSchema[cols].name, "ip:port:id");
+  strcpy(pSchema[cols].name, "ip:port");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
   pShow->bytes[cols] = 8;
   pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
-  strcpy(pSchema[cols].name, "created_time");
+  strcpy(pSchema[cols].name, "login time");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
   pShow->bytes[cols] = 8;
-  pSchema[cols].type = TSDB_DATA_TYPE_BIGINT;
-  strcpy(pSchema[cols].name, "time(us)");
-  pSchema[cols].bytes = htons(pShow->bytes[cols]);
-  cols++;
-
-  pShow->bytes[cols] = TSDB_SHOW_SQL_LEN;
-  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
-  strcpy(pSchema[cols].name, "sql");
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "last access");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
@@ -186,314 +207,208 @@ int32_t mnodeGetQueryMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
   pShow->numOfColumns = cols;
 
   pShow->offset[0] = 0;
-  for (int32_t i = 1; i < cols; ++i) pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
 
-  pShow->numOfRows = 1000000;
-  pShow->pIter = NULL;
+  pShow->numOfRows = taosHashGetSize(tsMnodeConnCache->pHashTable);
   pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
 
-  mnodeGetQueries(pShow, pConn);
   return 0;
 }
 
-int32_t mnodeKillQuery(char *qidstr, void *pConn) {
-//  char *temp, *chr, idstr[64];
-//  strcpy(idstr, qidstr);
-//
-//  temp = idstr;
-//  chr = strchr(temp, ':');
-//  if (chr == NULL) goto _error;
-//  *chr = 0;
-//  uint32_t ip = inet_addr(temp);
-//
-//  temp = chr + 1;
-//  chr = strchr(temp, ':');
-//  if (chr == NULL) goto _error;
-//  *chr = 0;
-//  uint16_t port = htons(atoi(temp));
-//
-//  temp = chr + 1;
-//  uint32_t queryId = atoi(temp);
-//
-//  SAcctObj *pAcct = pConn->pAcct;
-//
-//  pthread_mutex_lock(&pAcct->mutex);
-//
-//  pConn = pAcct->pConn;
-//  while (pConn) {
-//    if (pConn->ip == ip && pConn->port == port && pConn->pQList) {
-//      int32_t     i;
-//      SQueryDesc *pQDesc = pConn->pQList->qdesc;
-//      for (i = 0; i < pConn->pQList->numOfQueries; ++i, ++pQDesc) {
-//        if (pQDesc->queryId == queryId) break;
-//      }
-//
-//      if (i < pConn->pQList->numOfQueries) break;
-//    }
-//
-//    pConn = pConn->next;
-//  }
-//
-//  if (pConn) pConn->queryId = queryId;
-//
-//  pthread_mutex_unlock(&pAcct->mutex);
-//
-//  if (pConn == NULL || pConn->pQList == NULL || pConn->pQList->numOfQueries == 0) goto _error;
-//
-//  mTrace("query:%s is there, kill it", qidstr);
-//  return 0;
-//
-//_error:
-//  mTrace("query:%s is not there", qidstr);
-
-  return TSDB_CODE_INVALID_QUERY_ID;
-}
-
-int32_t mnodeRetrieveQueries(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
+static int32_t mnodeRetrieveConns(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
   int32_t   numOfRows = 0;
-  char *pWrite;
+  SConnObj *pConnObj = NULL;
   int32_t   cols = 0;
-
-  SQueryShow *pQueryShow = (SQueryShow *)pShow->pIter;
-
-  if (rows > pQueryShow->numOfQueries - pQueryShow->index) rows = pQueryShow->numOfQueries - pQueryShow->index;
+  char *    pWrite;
+  char      ipStr[TSDB_IPv4ADDR_LEN + 7];
 
   while (numOfRows < rows) {
-    SQueryDesc *pNode = pQueryShow->qdesc + pQueryShow->index;
-    SCDesc *pCDesc = pQueryShow->cdesc[pQueryShow->index];
+    pShow->pIter = mnodeGetNextConn(pShow->pIter, &pConnObj);
+    if (pConnObj == NULL) break;
+    
     cols = 0;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    strcpy(pWrite, pCDesc->user);
+    *(int32_t *) pWrite = pConnObj->connId;
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    uint32_t ip = pCDesc->ip;
-    sprintf(pWrite, "%d.%d.%d.%d:%hu:%d", ip & 0xFF, (ip >> 8) & 0xFF, (ip >> 16) & 0xFF, ip >> 24, htons(pCDesc->port),
-            pNode->queryId);
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConnObj->user, TSDB_USER_LEN);
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    *(int64_t *)pWrite = pNode->stime;
+    snprintf(ipStr, TSDB_IPv4ADDR_LEN + 6, "%s:%u", taosIpStr(pConnObj->ip), pConnObj->port);
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, TSDB_IPv4ADDR_LEN + 6);
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    *(int64_t *)pWrite = pNode->useconds;
+    *(int64_t *)pWrite = pConnObj->stime;
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    strcpy(pWrite, pNode->sql);
+    *(int64_t *)pWrite = pConnObj->lastAccess;
     cols++;
 
     numOfRows++;
-    pQueryShow->index++;
-  }
-
-  if (numOfRows == 0) {
-    tfree(pQueryShow->cdesc);
-    tfree(pQueryShow->connInfo);
-    tfree(pQueryShow);
   }
 
   pShow->numOfReads += numOfRows;
+  const int32_t NUM_OF_COLUMNS = 5;
+  mnodeVacuumResult(data, NUM_OF_COLUMNS, numOfRows, rows, pShow);
+  
   return numOfRows;
 }
 
-int32_t mnodeGetStreams(SShowObj *pShow, void *pConn) {
-//  SAcctObj *   pAcct = pConn->pAcct;
-//  SStreamShow *pStreamShow;
-//
-//  pthread_mutex_lock(&pAcct->mutex);
-//
-//  pStreamShow = malloc(sizeof(SStreamDesc) * pAcct->acctInfo.numOfStreams + sizeof(SQueryShow));
-//  pStreamShow->numOfStreams = 0;
-//  pStreamShow->index = 0;
-//  pStreamShow->connInfo = NULL;
-//  pStreamShow->cdesc = NULL;
-//
-//  if (pAcct->acctInfo.numOfStreams > 0) {
-//    pStreamShow->connInfo = (SCDesc *)malloc(pAcct->acctInfo.numOfConns * sizeof(SCDesc));
-//    pStreamShow->cdesc = (SCDesc **)malloc(pAcct->acctInfo.numOfStreams * sizeof(SCDesc *));
-//
-//    pConn = pAcct->pConn;
-//    SStreamDesc * pSdesc = pStreamShow->sdesc;
-//    SCDesc * pCDesc = pStreamShow->connInfo;
-//    SCDesc **ppCDesc = pStreamShow->cdesc;
-//
-//    while (pConn) {
-//      if (pConn->pSList && pConn->pSList->numOfStreams > 0) {
-//        pCDesc->ip = pConn->ip;
-//        pCDesc->port = pConn->port;
-//        strcpy(pCDesc->user, pConn->pUser->user);
-//
-//        memcpy(pSdesc, pConn->pSList->sdesc, sizeof(SStreamDesc) * pConn->pSList->numOfStreams);
-//        pSdesc += pConn->pSList->numOfStreams;
-//        pStreamShow->numOfStreams += pConn->pSList->numOfStreams;
-//        for (int32_t i = 0; i < pConn->pSList->numOfStreams; ++i, ++ppCDesc) *ppCDesc = pCDesc;
-//
-//        pCDesc++;
-//      }
-//      pConn = pConn->next;
-//    }
-//  }
-//
-//  pthread_mutex_unlock(&pAcct->mutex);
-//
-//  // sorting based on useconds
-//
-//  pShow->pIter = pStreamShow;
+// not thread safe, need optimized
+int32_t mnodeSaveQueryStreamList(SConnObj *pConn, SCMHeartBeatMsg *pHBMsg) {
+  pConn->numOfQueries = htonl(pHBMsg->numOfQueries);
+  if (pConn->numOfQueries > 0) {
+    pConn->pQueries = calloc(sizeof(SQueryDesc), pConn->numOfQueries);
+    memcpy(pConn->pQueries, pHBMsg->pData, pConn->numOfQueries * sizeof(SQueryDesc));
+  }
 
-  return 0;
+  pConn->numOfStreams = htonl(pHBMsg->numOfStreams);
+  if (pConn->numOfStreams > 0) {
+    pConn->pStreams = calloc(sizeof(SStreamDesc), pConn->numOfStreams);
+    memcpy(pConn->pStreams, pHBMsg->pData + pConn->numOfQueries * sizeof(SQueryDesc),
+           pConn->numOfStreams * sizeof(SStreamDesc));
+  }
+
+  return TSDB_CODE_SUCCESS;
 }
 
+static int32_t mnodeGetQueryMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
+  SUserObj *pUser = mnodeGetUserFromConn(pConn);
+  if (pUser == NULL) return 0;
+  if (strcmp(pUser->user, "root") != 0) return TSDB_CODE_NO_RIGHTS;
 
-int32_t mnodeKillStream(char *qidstr, void *pConn) {
-//  char *temp, *chr, idstr[64];
-//  strcpy(idstr, qidstr);
-//
-//  temp = idstr;
-//  chr = strchr(temp, ':');
-//  if (chr == NULL) goto _error;
-//  *chr = 0;
-//  uint32_t ip = inet_addr(temp);
-//
-//  temp = chr + 1;
-//  chr = strchr(temp, ':');
-//  if (chr == NULL) goto _error;
-//  *chr = 0;
-//  uint16_t port = htons(atoi(temp));
-//
-//  temp = chr + 1;
-//  uint32_t streamId = atoi(temp);
-//
-//  SAcctObj *pAcct = pConn->pAcct;
-//
-//  pthread_mutex_lock(&pAcct->mutex);
-//
-//  pConn = pAcct->pConn;
-//  while (pConn) {
-//    if (pConn->ip == ip && pConn->port == port && pConn->pSList) {
-//      int32_t     i;
-//      SStreamDesc *pSDesc = pConn->pSList->sdesc;
-//      for (i = 0; i < pConn->pSList->numOfStreams; ++i, ++pSDesc) {
-//        if (pSDesc->streamId == streamId) break;
-//      }
-//
-//      if (i < pConn->pSList->numOfStreams) break;
-//    }
-//
-//    pConn = pConn->next;
-//  }
-//
-//  if (pConn) pConn->streamId = streamId;
-//
-//  pthread_mutex_unlock(&pAcct->mutex);
-//
-//  if (pConn == NULL || pConn->pSList == NULL || pConn->pSList->numOfStreams == 0) goto _error;
-//
-//  mTrace("stream:%s is there, kill it", qidstr);
-//  return 0;
-//
-//_error:
-//  mTrace("stream:%s is not there", qidstr);
-
-  return TSDB_CODE_INVALID_STREAM_ID;
-}
+  int32_t cols = 0;
+  SSchema *pSchema = pMeta->schema;
 
-int32_t mnodeKillConnection(char *qidstr, void *pConn) {
-//  void *pConn1 = NULL;
-//  char *    temp, *chr, idstr[64];
-//  strcpy(idstr, qidstr);
-//
-//  temp = idstr;
-//  chr = strchr(temp, ':');
-//  if (chr == NULL) goto _error;
-//  *chr = 0;
-//  uint32_t ip = inet_addr(temp);
-//
-//  temp = chr + 1;
-//  uint16_t port = htons(atoi(temp));
-//  SAcctObj *pAcct = pConn->pAcct;
-//
-//  pthread_mutex_lock(&pAcct->mutex);
-//
-//  pConn = pAcct->pConn;
-//  while (pConn) {
-//    if (pConn->ip == ip && pConn->port == port) {
-//      // there maybe two connections from a shell
-//      if (pConn1 == NULL)
-//        pConn1 = pConn;
-//      else
-//        break;
-//    }
-//
-//    pConn = pConn->next;
-//  }
-//
-//  if (pConn1) pConn1->killConnection = 1;
-//  if (pConn) pConn->killConnection = 1;
-//
-//  pthread_mutex_unlock(&pAcct->mutex);
-//
-//  if (pConn1 == NULL) goto _error;
-//
-//  mTrace("connection:%s is there, kill it", qidstr);
-//  return 0;
-//
-//_error:
-//  mTrace("connection:%s is not there", qidstr);
-
-  return TSDB_CODE_INVALID_CONNECTION;
-}
+  pShow->bytes[cols] = QUERY_ID_SIZE + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "queryId");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_USER_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "user");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_IPv4ADDR_LEN + 6 + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "ip:port");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "created time");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_BIGINT;
+  strcpy(pSchema[cols].name, "time(us)");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_SHOW_SQL_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "sql");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htons(cols);
+  pShow->numOfColumns = cols;
 
+  pShow->offset[0] = 0;
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
 
-int mnodeGetConns(SShowObj *pShow, void *pConn) {
-  //  SAcctObj * pAcct = pConn->pAcct;
-  //  SConnShow *pConnShow;
-  //
-  //  pthread_mutex_lock(&pAcct->mutex);
-  //
-  //  pConnShow = malloc(sizeof(SConnInfo) * pAcct->acctInfo.numOfConns + sizeof(SConnShow));
-  //  pConnShow->index = 0;
-  //  pConnShow->numOfConns = 0;
-  //
-  //  if (pAcct->acctInfo.numOfConns > 0) {
-  //    pConn = pAcct->pConn;
-  //    SConnInfo *pConnInfo = pConnShow->connInfo;
-  //
-  //    while (pConn && pConn->pUser) {
-  //      strcpy(pConnInfo->user, pConn->pUser->user);
-  //      pConnInfo->ip = pConn->ip;
-  //      pConnInfo->port = pConn->port;
-  //      pConnInfo->stime = pConn->stime;
-  //
-  //      pConnShow->numOfConns++;
-  //      pConnInfo++;
-  //      pConn = pConn->next;
-  //    }
-  //  }
-  //
-  //  pthread_mutex_unlock(&pAcct->mutex);
-  //
-  //  // sorting based on useconds
-  //
-  //  pShow->pIter = pConnShow;
+  pShow->numOfRows = 1000000;
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
 
   return 0;
 }
 
-int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
-  int32_t cols = 0;
+static int32_t mnodeRetrieveQueries(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
+  int32_t   numOfRows = 0;
+  SConnObj *pConnObj = NULL;
+  int32_t   cols = 0;
+  char *    pWrite;
+  char      ipStr[TSDB_IPv4ADDR_LEN + 7];
+
+  while (numOfRows < rows) {
+    pShow->pIter = mnodeGetNextConn(pShow->pIter, &pConnObj);
+    if (pConnObj == NULL) break;
+
+    for (int32_t i = 0; i < pConnObj->numOfQueries; ++i) {
+      SQueryDesc *pDesc = pConnObj->pQueries + i;
+      cols = 0;
+
+      snprintf(ipStr, QUERY_ID_SIZE + 1, "%u:%u", pConnObj->connId, htonl(pDesc->queryId));
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, QUERY_ID_SIZE);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConnObj->user, TSDB_USER_LEN);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      snprintf(ipStr, TSDB_IPv4ADDR_LEN + 6, "%s:%u", taosIpStr(pConnObj->ip), pConnObj->port);
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, TSDB_IPv4ADDR_LEN + 6);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      *(int64_t *)pWrite = htobe64(pDesc->stime);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      *(int64_t *)pWrite = htobe64(pDesc->useconds);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDesc->sql, TSDB_SHOW_SQL_LEN);
+      cols++;
 
-  pShow->bytes[cols] = TSDB_TABLE_NAME_LEN;
+      numOfRows++;
+    }
+  }
+
+  pShow->numOfReads += numOfRows;
+  const int32_t NUM_OF_COLUMNS = 6;
+  mnodeVacuumResult(data, NUM_OF_COLUMNS, numOfRows, rows, pShow);
+  return numOfRows;
+}
+
+static int32_t mnodeGetStreamMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
+  SUserObj *pUser = mnodeGetUserFromConn(pConn);
+  if (pUser == NULL) return 0;
+  if (strcmp(pUser->user, "root") != 0) return TSDB_CODE_NO_RIGHTS;
+
+  int32_t cols = 0;
   SSchema *pSchema = pMeta->schema;
 
+  pShow->bytes[cols] = QUERY_ID_SIZE + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "streamId");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_USER_LEN + VARSTR_HEADER_SIZE;
   pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
   strcpy(pSchema[cols].name, "user");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
-  pShow->bytes[cols] = TSDB_IPv4ADDR_LEN + 6;
+  pShow->bytes[cols] = TSDB_IPv4ADDR_LEN + 6 + VARSTR_HEADER_SIZE;
   pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
   strcpy(pSchema[cols].name, "ip:port");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
@@ -501,7 +416,31 @@ int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
 
   pShow->bytes[cols] = 8;
   pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
-  strcpy(pSchema[cols].name, "login time");
+  strcpy(pSchema[cols].name, "created time");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "exec time");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_BIGINT;
+  strcpy(pSchema[cols].name, "time(us)");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_SHOW_SQL_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "sql");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 4;
+  pSchema[cols].type = TSDB_DATA_TYPE_INT;
+  strcpy(pSchema[cols].name, "cycles");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
@@ -509,140 +448,148 @@ int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
   pShow->numOfColumns = cols;
 
   pShow->offset[0] = 0;
-  for (int i = 1; i < cols; ++i) pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
 
   pShow->numOfRows = 1000000;
-  pShow->pIter = NULL;
   pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
 
-  mnodeGetConns(pShow, pConn);
   return 0;
 }
 
-int32_t mnodeRetrieveConns(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
+static int32_t mnodeRetrieveStreams(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
   int32_t   numOfRows = 0;
-  char      *pWrite;
+  SConnObj *pConnObj = NULL;
   int32_t   cols = 0;
+  char *    pWrite;
+  char      ipStr[TSDB_IPv4ADDR_LEN + 7];
 
-  SConnShow *pConnShow = (SConnShow *)pShow->pIter;
+  while (numOfRows < rows) {
+    pShow->pIter = mnodeGetNextConn(pShow->pIter, &pConnObj);
+    if (pConnObj == NULL) break;
 
-  if (rows > pConnShow->numOfConns - pConnShow->index) rows = pConnShow->numOfConns - pConnShow->index;
+    for (int32_t i = 0; i < pConnObj->numOfStreams; ++i) {
+      SStreamDesc *pDesc = pConnObj->pStreams + i;
+      cols = 0;
 
-  while (numOfRows < rows) {
-    SConnInfo *pNode = pConnShow->connInfo + pConnShow->index;
-    cols = 0;
+      snprintf(ipStr, QUERY_ID_SIZE + 1, "%u:%u", pConnObj->connId, htonl(pDesc->streamId));
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, QUERY_ID_SIZE);
+      cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    strcpy(pWrite, pNode->user);
-    cols++;
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pConnObj->user, TSDB_USER_LEN);
+      cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    uint32_t ip = pNode->ip;
-    sprintf(pWrite, "%d.%d.%d.%d:%hu", ip & 0xFF, (ip >> 8) & 0xFF, (ip >> 16) & 0xFF, ip >> 24, htons(pNode->port));
-    cols++;
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      snprintf(ipStr, TSDB_IPv4ADDR_LEN + 6, "%s:%u", taosIpStr(pConnObj->ip), pConnObj->port);
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, ipStr, TSDB_IPv4ADDR_LEN + 6);
+      cols++;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    *(int64_t *)pWrite = pNode->stime;
-    cols++;
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      *(int64_t *)pWrite = htobe64(pDesc->ctime);
+      cols++;
 
-    numOfRows++;
-    pConnShow->index++;
-  }
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      *(int64_t *)pWrite = htobe64(pDesc->stime);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      *(int64_t *)pWrite = htobe64(pDesc->useconds);
+      cols++;
 
-  if (numOfRows == 0) {
-    tfree(pConnShow);
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDesc->sql, TSDB_SHOW_SQL_LEN);
+      cols++;
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      *(int32_t *)pWrite = (int32_t)htobe64(pDesc->num);
+      cols++;
+
+      numOfRows++;
+    }
   }
 
   pShow->numOfReads += numOfRows;
+  const int32_t NUM_OF_COLUMNS = 8;
+  mnodeVacuumResult(data, NUM_OF_COLUMNS, numOfRows, rows, pShow);
   return numOfRows;
 }
 
-int32_t mnodeProcessKillQueryMsg(SMnodeMsg *pMsg) {
-  // SRpcMsg rpcRsp = {.handle = pMsg->thandle, .pCont = NULL, .contLen = 0, .code = 0, .msgType = 0};
-  
-  // SUserObj *pUser = mnodeGetUserFromConn(pMsg->thandle);
-  // if (pUser == NULL) {
-  //   rpcRsp.code = TSDB_CODE_INVALID_USER;
-  //   rpcSendResponse(&rpcRsp);
-  //   return;
-  // }
-
-  // SCMKillQueryMsg *pKill = pMsg->pCont;
-  // int32_t code;
-
-  // if (!pUser->writeAuth) {
-  //   code = TSDB_CODE_NO_RIGHTS;
-  // } else {
-  //   code = mgmtKillQuery(pKill->queryId, pMsg->thandle);
-  // }
-
-  // rpcRsp.code = code;
-  // rpcSendResponse(&rpcRsp);
-  // mnodeDecUserRef(pUser);
-  return TSDB_CODE_SUCCESS;
-}
+static int32_t mnodeProcessKillQueryMsg(SMnodeMsg *pMsg) {
+  SUserObj *pUser = pMsg->pUser;
+  if (strcmp(pUser->user, "root") != 0) return TSDB_CODE_NO_RIGHTS;
 
-int32_t mnodeProcessKillStreamMsg(SMnodeMsg *pMsg) {
-  // SRpcMsg rpcRsp = {.handle = pMsg->thandle, .pCont = NULL, .contLen = 0, .code = 0, .msgType = 0};
-  
-  // SUserObj *pUser = mnodeGetUserFromConn(pMsg->thandle);
-  // if (pUser == NULL) {
-  //   rpcRsp.code = TSDB_CODE_INVALID_USER;
-  //   rpcSendResponse(&rpcRsp);
-  //   return;
-  // }
-
-  // SCMKillStreamMsg *pKill = pMsg->pCont;
-  // int32_t code;
-
-  // if (!pUser->writeAuth) {
-  //   code = TSDB_CODE_NO_RIGHTS;
-  // } else {
-  //   code = mgmtKillStream(pKill->queryId, pMsg->thandle);
-  // }
-
-  // rpcRsp.code = code;
-  // rpcSendResponse(&rpcRsp);
-  // mnodeDecUserRef(pUser);
-  return TSDB_CODE_SUCCESS;
-}
+  SCMKillQueryMsg *pKill = pMsg->rpcMsg.pCont;
+  mPrint("kill query msg is received, queryId:%s", pKill->queryId);
 
-int32_t mnodeProcessKillConnectionMsg(SMnodeMsg *pMsg) {
-  // SRpcMsg rpcRsp = {.handle = pMsg->thandle, .pCont = NULL, .contLen = 0, .code = 0, .msgType = 0};
-  
-  // SUserObj *pUser = mnodeGetUserFromConn(pMsg->thandle);
-  // if (pUser == NULL) {
-  //   rpcRsp.code = TSDB_CODE_INVALID_USER;
-  //   rpcSendResponse(&rpcRsp);
-  //   return;
-  // }
-
-  // SCMKillConnMsg *pKill = pMsg->pCont;
-  // int32_t code;
-
-  // if (!pUser->writeAuth) {
-  //   code = TSDB_CODE_NO_RIGHTS;
-  // } else {
-  //   code = mgmtKillConnection(pKill->queryId, pMsg->thandle);
-  // }
-
-  // rpcRsp.code = code;
-  // rpcSendResponse(&rpcRsp);
-  // mnodeDecUserRef(pUser);
-  return TSDB_CODE_SUCCESS;
+  const char delim = ':';
+  char* connIdStr = strtok(pKill->queryId, &delim);
+  char* queryIdStr = strtok(NULL, &delim);
+
+  if (queryIdStr == NULL || connIdStr == NULL) {
+    mPrint("failed to kill query, queryId:%s", pKill->queryId);
+   return TSDB_CODE_INVALID_QUERY_ID;
+  }
+
+  int32_t queryId = (int32_t)strtol(queryIdStr, NULL, 10);
+
+  SConnObj *pConn = taosCacheAcquireByName(tsMnodeConnCache, connIdStr);
+  if (pConn == NULL) {
+    mError("connId:%s, failed to kill queryId:%d, conn not exist", connIdStr, queryId);
+    return TSDB_CODE_INVALID_CONNECTION;
+  } else {
+    mPrint("connId:%s, queryId:%d is killed by user:%s", connIdStr, queryId, pUser->user);
+    pConn->queryId = queryId;
+    taosCacheRelease(tsMnodeConnCache, (void **)&pConn, false);
+    return TSDB_CODE_SUCCESS;
+  }
 }
 
-int32_t mnodeInitProfile() {
-  // mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_QUERIES, mnodeGetQueryMeta);
-  // mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_QUERIES, mnodeRetrieveQueries);
-  // mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_CONNS, mnodeGetConnsMeta);
-  // mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_CONNS, mnodeRetrieveConns);
+static int32_t mnodeProcessKillStreamMsg(SMnodeMsg *pMsg) {
+  SUserObj *pUser = pMsg->pUser;
+  if (strcmp(pUser->user, "root") != 0) return TSDB_CODE_NO_RIGHTS;
 
-  // mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_QUERY, mnodeProcessKillQueryMsg);
-  // mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_STREAM, mnodeProcessKillStreamMsg);
-  // mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_CONN, mnodeProcessKillConnectionMsg);
+  SCMKillQueryMsg *pKill = pMsg->rpcMsg.pCont;
+  mPrint("kill stream msg is received, streamId:%s", pKill->queryId);
 
-  return 0;
+  const char delim = ':';
+  char* connIdStr = strtok(pKill->queryId, &delim);
+  char* streamIdStr = strtok(NULL, &delim);
+
+  if (streamIdStr == NULL || connIdStr == NULL) {
+    mPrint("failed to kill stream, streamId:%s", pKill->queryId);
+   return TSDB_CODE_INVALID_STREAM_ID;
+  }
+
+  int32_t streamId = (int32_t)strtol(streamIdStr, NULL, 10);
+
+  SConnObj *pConn = taosCacheAcquireByName(tsMnodeConnCache, connIdStr);
+  if (pConn == NULL) {
+    mError("connId:%s, failed to kill streamId:%d, conn not exist", connIdStr, streamId);
+    return TSDB_CODE_INVALID_CONNECTION;
+  } else {
+    mPrint("connId:%s, streamId:%d is killed by user:%s", connIdStr, streamId, pUser->user);
+    pConn->streamId = streamId;
+    taosCacheRelease(tsMnodeConnCache, (void **)&pConn, false);
+    return TSDB_CODE_SUCCESS;
+  }
 }
 
-void mnodeCleanupProfile() {}
+static int32_t mnodeProcessKillConnectionMsg(SMnodeMsg *pMsg) {
+  SUserObj *pUser = pMsg->pUser;
+  if (strcmp(pUser->user, "root") != 0) return TSDB_CODE_NO_RIGHTS;
+
+  SCMKillConnMsg *pKill = pMsg->rpcMsg.pCont;
+  SConnObj *      pConn = taosCacheAcquireByName(tsMnodeConnCache, pKill->queryId);
+  if (pConn == NULL) {
+    mError("connId:%s, failed to kill, conn not exist", pKill->queryId);
+    return TSDB_CODE_INVALID_CONNECTION;
+  } else {
+    mPrint("connId:%s, is killed by user:%s", pKill->queryId, pUser->user);
+    pConn->killed = 1;
+    taosCacheRelease(tsMnodeConnCache, (void**)&pConn, false);
+    return TSDB_CODE_SUCCESS;
+  }
+}

src/mnode/src/mnodeShow.c

@@ -227,21 +227,47 @@ static int32_t mnodeProcessHeartBeatMsg(SMnodeMsg *pMsg) {
     return TSDB_CODE_SERV_OUT_OF_MEMORY;
   }
 
+  SCMHeartBeatMsg *pHBMsg = pMsg->rpcMsg.pCont;
+  SRpcConnInfo connInfo;
+  rpcGetConnInfo(pMsg->rpcMsg.handle, &connInfo);
+    
+  int32_t connId = htonl(pHBMsg->connId);
+  SConnObj *pConn = mnodeAccquireConn(connId, connInfo.user, connInfo.clientIp, connInfo.clientPort);
+  if (pConn == NULL) {
+    pConn = mnodeCreateConn(connInfo.user, connInfo.clientIp, connInfo.clientPort);
+  }
+
+  if (pConn == NULL) {
+    // do not close existing links, otherwise
+    // mError("failed to create connId, close connect");
+    // pHBRsp->killConnection = 1;
+  } else {
+    pHBRsp->connId = htonl(pConn->connId);
+    mnodeSaveQueryStreamList(pConn, pHBMsg);
+    
+    if (pConn->killed != 0) {
+      pHBRsp->killConnection = 1;
+    }
+
+    if (pConn->streamId != 0) {
+      pHBRsp->streamId = htonl(pConn->streamId);
+      pConn->streamId = 0;
+    }
+
+    if (pConn->queryId != 0) {
+      pHBRsp->queryId = htonl(pConn->queryId);
+      pConn->queryId = 0;
+    }
+  }
+
   pHBRsp->onlineDnodes = htonl(mnodeGetOnlinDnodesNum());
   pHBRsp->totalDnodes = htonl(mnodeGetDnodesNum());
   mnodeGetMnodeIpSetForShell(&pHBRsp->ipList);
-  
-  /*
-   * TODO
-   * Dispose kill stream or kill query message
-   */
-  pHBRsp->queryId = 0;
-  pHBRsp->streamId = 0;
-  pHBRsp->killConnection = 0;
 
   pMsg->rpcRsp.rsp = pHBRsp;
   pMsg->rpcRsp.len = sizeof(SCMHeartBeatRsp);
   
+  mnodeReleaseConn(pConn);
   return TSDB_CODE_SUCCESS;
 }
 
@@ -281,11 +307,19 @@ static int32_t mnodeProcessConnectMsg(SMnodeMsg *pMsg) {
     goto connect_over;
   }
 
+  SConnObj *pConn = mnodeCreateConn(connInfo.user, connInfo.clientIp, connInfo.clientPort);
+  if (pConn == NULL) {
+    code = terrno;
+  } else {
+    pConnectRsp->connId = htonl(pConn->connId);
+    mnodeReleaseConn(pConn);
+  }
+
   sprintf(pConnectRsp->acctId, "%x", pAcct->acctId);
   strcpy(pConnectRsp->serverVersion, version);
   pConnectRsp->writeAuth = pUser->writeAuth;
   pConnectRsp->superAuth = pUser->superAuth;
-
+  
   mnodeGetMnodeIpSetForShell(&pConnectRsp->ipList);
 
 connect_over:
@@ -358,3 +392,11 @@ static void mnodeReleaseShowObj(void *pShow, bool forceRemove) {
   mTrace("%p, show is released, force:%s", pShow, forceRemove ? "true" : "false");
   taosCacheRelease(tsMnodeShowCache, &pShow, forceRemove);
 }
+
+void mnodeVacuumResult(char *data, int32_t numOfCols, int32_t rows, int32_t capacity, SShowObj *pShow) {
+  if (rows < capacity) {
+    for (int32_t i = 0; i < numOfCols; ++i) {
+      memmove(data + pShow->offset[i] * rows, data + pShow->offset[i] * capacity, pShow->bytes[i] * rows);
+    }
+  }
+}
\ No newline at end of file

src/mnode/src/mnodeTable.c

@@ -61,8 +61,8 @@ static int32_t mnodeGetShowTableMeta(STableMetaMsg *pMeta, SShowObj *pShow, void
 static int32_t mnodeRetrieveShowTables(SShowObj *pShow, char *data, int32_t rows, void *pConn);
 static int32_t mnodeGetShowSuperTableMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
 static int32_t mnodeRetrieveShowSuperTables(SShowObj *pShow, char *data, int32_t rows, void *pConn);
-static int32_t mnodeGetStreamMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
-static int32_t mnodeRetrieveStreams(SShowObj *pShow, char *data, int32_t rows, void *pConn);
+static int32_t mnodeGetStreamTableMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
+static int32_t mnodeRetrieveStreamTables(SShowObj *pShow, char *data, int32_t rows, void *pConn);
  
 static int32_t mnodeProcessCreateTableMsg(SMnodeMsg *mnodeMsg);
 static int32_t mnodeProcessCreateSuperTableMsg(SMnodeMsg *pMsg);
@@ -568,8 +568,8 @@ int32_t mnodeInitTables() {
   mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_TABLE, mnodeRetrieveShowTables);
   mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_METRIC, mnodeGetShowSuperTableMeta);
   mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_METRIC, mnodeRetrieveShowSuperTables);
-  mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_STREAMS, mnodeGetStreamMeta);
-  mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_STREAMS, mnodeRetrieveStreams);
+  mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_STREAMTABLES, mnodeGetStreamTableMeta);
+  mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_STREAMTABLES, mnodeRetrieveStreamTables);
 
   return TSDB_CODE_SUCCESS;
 }
@@ -1284,7 +1284,7 @@ static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *pMsg) {
   char *msg = (char *)pRsp + sizeof(SCMSTableVgroupRspMsg);
 
   for (int32_t i = 0; i < numOfTable; ++i) {
-    char *stableName = (char*)pInfo + sizeof(SCMSTableVgroupMsg) + (TSDB_TABLE_ID_LEN) * i;
+    char *          stableName = (char *)pInfo + sizeof(SCMSTableVgroupMsg) + (TSDB_TABLE_ID_LEN)*i;
     SSuperTableObj *pTable = mnodeGetSuperTable(stableName);
     if (pTable == NULL) {
       mError("stable:%s, not exist while get stable vgroup info", stableName);
@@ -1294,41 +1294,48 @@ static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *pMsg) {
     if (pTable->vgHash == NULL) {
       mError("stable:%s, not vgroup exist while get stable vgroup info", stableName);
       mnodeDecTableRef(pTable);
-      continue;
-    }
 
-    SVgroupsInfo *pVgroupInfo = (SVgroupsInfo *)msg;
+      // even this super table has no corresponding table, still return
+      pRsp->numOfTables++;
 
-    SHashMutableIterator *pIter = taosHashCreateIter(pTable->vgHash);
-    int32_t vgSize = 0;
-    while (taosHashIterNext(pIter)) {
-      int32_t *pVgId = taosHashIterGet(pIter);
-      SVgObj * pVgroup = mnodeGetVgroup(*pVgId);
-      if (pVgroup == NULL) continue;
+      SVgroupsInfo *pVgroupInfo = (SVgroupsInfo *)msg;
+      pVgroupInfo->numOfVgroups = 0;
+      
+      msg += sizeof(SVgroupsInfo);
+    } else {
+      SVgroupsInfo *pVgroupInfo = (SVgroupsInfo *)msg;
 
-      pVgroupInfo->vgroups[vgSize].vgId = htonl(pVgroup->vgId);
-      for (int32_t vn = 0; vn < pVgroup->numOfVnodes; ++vn) {
-        SDnodeObj *pDnode = pVgroup->vnodeGid[vn].pDnode;
-        if (pDnode == NULL) break;
+      SHashMutableIterator *pIter = taosHashCreateIter(pTable->vgHash);
+      int32_t               vgSize = 0;
+      while (taosHashIterNext(pIter)) {
+        int32_t *pVgId = taosHashIterGet(pIter);
+        SVgObj * pVgroup = mnodeGetVgroup(*pVgId);
+        if (pVgroup == NULL) continue;
 
-        strncpy(pVgroupInfo->vgroups[vgSize].ipAddr[vn].fqdn, pDnode->dnodeFqdn, tListLen(pDnode->dnodeFqdn));
-        pVgroupInfo->vgroups[vgSize].ipAddr[vn].port = htons(pDnode->dnodePort);
+        pVgroupInfo->vgroups[vgSize].vgId = htonl(pVgroup->vgId);
+        for (int32_t vn = 0; vn < pVgroup->numOfVnodes; ++vn) {
+          SDnodeObj *pDnode = pVgroup->vnodeGid[vn].pDnode;
+          if (pDnode == NULL) break;
 
-        pVgroupInfo->vgroups[vgSize].numOfIps++;
-      }
+          strncpy(pVgroupInfo->vgroups[vgSize].ipAddr[vn].fqdn, pDnode->dnodeFqdn, tListLen(pDnode->dnodeFqdn));
+          pVgroupInfo->vgroups[vgSize].ipAddr[vn].port = htons(pDnode->dnodePort);
 
-      vgSize++;
-      mnodeDecVgroupRef(pVgroup);
-    }
+          pVgroupInfo->vgroups[vgSize].numOfIps++;
+        }
 
-    taosHashDestroyIter(pIter);
-    mnodeDecTableRef(pTable);
+        vgSize++;
+        mnodeDecVgroupRef(pVgroup);
+      }
+
+      taosHashDestroyIter(pIter);
+      mnodeDecTableRef(pTable);
 
-    pVgroupInfo->numOfVgroups = htonl(vgSize);
+      pVgroupInfo->numOfVgroups = htonl(vgSize);
 
-    // one table is done, try the next table
-    msg += sizeof(SVgroupsInfo) + vgSize * sizeof(SCMVgroupInfo);
-    pRsp->numOfTables++;
+      // one table is done, try the next table
+      msg += sizeof(SVgroupsInfo) + vgSize * sizeof(SCMVgroupInfo);
+      pRsp->numOfTables++;
+    }
   }
 
   if (pRsp->numOfTables != numOfTable) {
@@ -2111,14 +2118,6 @@ static int32_t mnodeGetShowTableMeta(STableMetaMsg *pMeta, SShowObj *pShow, void
   return 0;
 }
 
-static void mnodeVacuumResult(char *data, int32_t numOfCols, int32_t rows, int32_t capacity, SShowObj *pShow) {
-  if (rows < capacity) {
-    for (int32_t i = 0; i < numOfCols; ++i) {
-      memmove(data + pShow->offset[i] * rows, data + pShow->offset[i] * capacity, pShow->bytes[i] * rows);
-    }
-  }
-}
-
 static int32_t mnodeRetrieveShowTables(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
   SDbObj *pDb = mnodeGetDb(pShow->db);
   if (pDb == NULL) return 0;
@@ -2262,7 +2261,7 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) {
   return code;
 }
 
-static int32_t mnodeGetStreamMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
+static int32_t mnodeGetStreamTableMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
   SDbObj *pDb = mnodeGetDb(pShow->db);
   if (pDb == NULL) return TSDB_CODE_DB_NOT_SELECTED;
 
@@ -2308,7 +2307,7 @@ static int32_t mnodeGetStreamMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *p
   return 0;
 }
 
-static int32_t mnodeRetrieveStreams(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
+static int32_t mnodeRetrieveStreamTables(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
   SDbObj *pDb = mnodeGetDb(pShow->db);
   if (pDb == NULL) return 0;
 

src/plugins/monitor/src/monitorMain.c

@@ -183,7 +183,7 @@ static void dnodeBuildMonitorSql(char *sql, int32_t cmd) {
     snprintf(sql, SQL_LENGTH,
              "create table if not exists %s.slowquery(ts timestamp, username "
              "binary(%d), created_time timestamp, time bigint, sql binary(%d))",
-             tsMonitorDbName, TSDB_TABLE_ID_LEN, TSDB_SHOW_SQL_LEN);
+             tsMonitorDbName, TSDB_TABLE_ID_LEN, TSDB_SLOW_QUERY_SQL_LEN);
   } else if (cmd == MONITOR_CMD_CREATE_TB_LOG) {
     snprintf(sql, SQL_LENGTH,
              "create table if not exists %s.log(ts timestamp, level tinyint, "

src/query/inc/qExecutor.h

@@ -172,10 +172,11 @@ typedef struct SQueryRuntimeEnv {
   STSBuf*              pTSBuf;
   STSCursor            cur;
   SQueryCostInfo       summary;
-  bool                 stableQuery;  // super table query or not
+  bool                 stableQuery;      // super table query or not
   void*                pQueryHandle;
   void*                pSecQueryHandle;  // another thread for
   SDiskbasedResultBuf* pResultBuf;       // query result buffer based on blocked-wised disk file
+  bool                 topBotQuery;      // false;
 } SQueryRuntimeEnv;
 
 typedef struct SQInfo {

src/query/inc/qUtil.h

@@ -31,7 +31,8 @@ void    closeTimeWindow(SWindowResInfo* pWindowResInfo, int32_t slot);
 void    closeAllTimeWindow(SWindowResInfo* pWindowResInfo);
 void    removeRedundantWindow(SWindowResInfo *pWindowResInfo, TSKEY lastKey, int32_t order);
 SWindowResult *getWindowResult(SWindowResInfo *pWindowResInfo, int32_t slot);
-int32_t curTimeWindow(SWindowResInfo *pWindowResInfo);
+
+#define curTimeWindow(_winres)  ((_winres)->curIndex)
 bool isWindowResClosed(SWindowResInfo *pWindowResInfo, int32_t slot);
 
 void createQueryResultInfo(SQuery *pQuery, SWindowResult *pResultRow, bool isSTableQuery, SPosInfo *posInfo);

src/query/inc/qextbuffer.h

@@ -28,7 +28,7 @@ extern "C" {
 #include "tdataformat.h"
 #include "talgo.h"
 
-#define DEFAULT_PAGE_SIZE (1024L*56)  // 16k larger than the SHistoInfo
+#define DEFAULT_PAGE_SIZE (1024L*64)  // 16k larger than the SHistoInfo
 #define MAX_TMPFILE_PATH_LENGTH PATH_MAX
 #define INITIAL_ALLOCATION_BUFFER_SIZE 64
 
@@ -96,7 +96,7 @@ typedef struct SColumnOrderInfo {
 typedef struct tOrderDescriptor {
   SColumnModel *   pColumnModel;
   int32_t          tsOrder;  // timestamp order type if exists
-  SColumnOrderInfo orderIdx;
+  SColumnOrderInfo orderInfo;
 } tOrderDescriptor;
 
 typedef struct tExtMemBuffer {

src/query/inc/qresultBuf.h

@@ -85,7 +85,7 @@ SIDList getDataBufPagesIdList(SDiskbasedResultBuf* pResultBuf, int32_t groupId);
  * @param id
  * @return
  */
-tFilePage* getResultBufferPageById(SDiskbasedResultBuf* pResultBuf, int32_t id);
+#define GET_RES_BUF_PAGE_BY_ID(buf, id)  ((tFilePage*)((buf)->pBuf + DEFAULT_INTERN_BUF_PAGE_SIZE*(id)))
 
 /**
  * get the total buffer size in the format of disk file

src/query/inc/sql.y

@@ -647,9 +647,9 @@ cmd ::= ALTER TABLE ids(X) cpxName(F) SET TAG ids(Y) EQ tagitem(Z).     {
 }
 
 ////////////////////////////////////////kill statement///////////////////////////////////////
-cmd ::= KILL CONNECTION IPTOKEN(X) COLON(Z) INTEGER(Y).   {X.n += (Z.n + Y.n); setKillSQL(pInfo, TSDB_SQL_KILL_CONNECTION, &X);}
-cmd ::= KILL STREAM IPTOKEN(X) COLON(Z) INTEGER(Y) COLON(K) INTEGER(F).       {X.n += (Z.n + Y.n + K.n + F.n); setKillSQL(pInfo, TSDB_SQL_KILL_STREAM, &X);}
-cmd ::= KILL QUERY IPTOKEN(X) COLON(Z) INTEGER(Y) COLON(K) INTEGER(F).        {X.n += (Z.n + Y.n + K.n + F.n); setKillSQL(pInfo, TSDB_SQL_KILL_QUERY, &X);}
+cmd ::= KILL CONNECTION INTEGER(Y).   {setKillSQL(pInfo, TSDB_SQL_KILL_CONNECTION, &Y);}
+cmd ::= KILL STREAM INTEGER(X) COLON(Z) INTEGER(Y).       {X.n += (Z.n + Y.n); setKillSQL(pInfo, TSDB_SQL_KILL_STREAM, &X);}
+cmd ::= KILL QUERY INTEGER(X) COLON(Z) INTEGER(Y).        {X.n += (Z.n + Y.n); setKillSQL(pInfo, TSDB_SQL_KILL_QUERY, &X);}
 
 %fallback ID ABORT AFTER ASC ATTACH BEFORE BEGIN CASCADE CLUSTER CONFLICT COPY DATABASE DEFERRED
   DELIMITERS DESC DETACH EACH END EXPLAIN FAIL FOR GLOB IGNORE IMMEDIATE INITIALLY INSTEAD

src/query/inc/tsqlfunction.h

@@ -272,9 +272,18 @@ bool top_bot_datablock_filter(SQLFunctionCtx *pCtx, int32_t functionId, char *mi
 bool stableQueryFunctChanged(int32_t funcId);
 
 void resetResultInfo(SResultInfo *pResInfo);
-void initResultInfo(SResultInfo *pResInfo);
 void setResultInfoBuf(SResultInfo *pResInfo, int32_t size, bool superTable);
 
+static FORCE_INLINE void initResultInfo(SResultInfo *pResInfo) {
+  pResInfo->initialized = true;  // the this struct has been initialized flag
+  
+  pResInfo->complete = false;
+  pResInfo->hasResult = false;
+  pResInfo->numOfRes = 0;
+  
+  memset(pResInfo->interResultBuf, 0, (size_t)pResInfo->bufLen);
+}
+
 #ifdef __cplusplus
 }
 #endif

src/query/src/qExecutor.c

@@ -98,6 +98,7 @@ static bool isIntervalQuery(SQuery *pQuery) { return pQuery->intervalTime > 0; }
 static int32_t mergeIntoGroupResultImpl(SQInfo *pQInfo, SArray *group);
 
 static void setWindowResOutputBuf(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pResult);
+static void setWindowResOutputBufInitCtx(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pResult);
 static void resetMergeResultBuf(SQuery *pQuery, SQLFunctionCtx *pCtx, SResultInfo *pResultInfo);
 static bool functionNeedToExecute(SQueryRuntimeEnv *pRuntimeEnv, SQLFunctionCtx *pCtx, int32_t functionId);
 static void getNextTimeWindow(SQuery *pQuery, STimeWindow *pTimeWindow);
@@ -441,7 +442,7 @@ static int32_t addNewWindowResultBuf(SWindowResult *pWindowRes, SDiskbasedResult
     pData = getNewDataBuf(pResultBuf, sid, &pageId);
   } else {
     pageId = getLastPageId(&list);
-    pData = getResultBufferPageById(pResultBuf, pageId);
+    pData = GET_RES_BUF_PAGE_BY_ID(pResultBuf, pageId);
 
     if (pData->num >= numOfRowsPerPage) {
       pData = getNewDataBuf(pResultBuf, sid, &pageId);
@@ -484,10 +485,8 @@ static int32_t setWindowOutputBufByKey(SQueryRuntimeEnv *pRuntimeEnv, SWindowRes
 
   // set time window for current result
   pWindowRes->window = *win;
-
-  setWindowResOutputBuf(pRuntimeEnv, pWindowRes);
-  initCtxOutputBuf(pRuntimeEnv);
-
+  
+  setWindowResOutputBufInitCtx(pRuntimeEnv, pWindowRes);
   return TSDB_CODE_SUCCESS;
 }
 
@@ -651,7 +650,7 @@ static void doBlockwiseApplyFunctions(SQueryRuntimeEnv *pRuntimeEnv, SWindowStat
       pCtx[k].startOffset = (QUERY_IS_ASC_QUERY(pQuery)) ? offset : offset - (forwardStep - 1);
 
       if ((aAggs[functionId].nStatus & TSDB_FUNCSTATE_SELECTIVITY) != 0) {
-        pCtx[k].ptsList = tsBuf;
+        pCtx[k].ptsList = &tsBuf[offset];
       }
 
       // not a whole block involved in query processing, statistics data can not be used
@@ -687,47 +686,52 @@ static int32_t getNextQualifiedWindow(SQueryRuntimeEnv *pRuntimeEnv, STimeWindow
                                       SDataBlockInfo *pDataBlockInfo, TSKEY *primaryKeys,
                                       __block_search_fn_t searchFn) {
   SQuery *pQuery = pRuntimeEnv->pQuery;
+  
+  // tumbling time window query, a special case of sliding time window query
+  if (pQuery->slidingTime == pQuery->intervalTime) {
+    // todo opt
+  }
+  
+  getNextTimeWindow(pQuery, pNextWin);
+  
+  // next time window is not in current block
+  if ((pNextWin->skey > pDataBlockInfo->window.ekey && QUERY_IS_ASC_QUERY(pQuery)) ||
+      (pNextWin->ekey < pDataBlockInfo->window.skey && !QUERY_IS_ASC_QUERY(pQuery))) {
+    return -1;
+  }
 
-  while (1) {
-    if ((pNextWin->ekey > pQuery->window.ekey && QUERY_IS_ASC_QUERY(pQuery)) ||
-        (pNextWin->skey < pQuery->window.ekey && !QUERY_IS_ASC_QUERY(pQuery))) {
-      return -1;
-    }
-
-    getNextTimeWindow(pQuery, pNextWin);
-
-    // next time window is not in current block
-    if ((pNextWin->skey > pDataBlockInfo->window.ekey && QUERY_IS_ASC_QUERY(pQuery)) ||
-        (pNextWin->ekey < pDataBlockInfo->window.skey && !QUERY_IS_ASC_QUERY(pQuery))) {
-      return -1;
+  TSKEY startKey = -1;
+  if (QUERY_IS_ASC_QUERY(pQuery)) {
+    startKey = pNextWin->skey;
+    if (startKey < pQuery->window.skey) {
+      startKey = pQuery->window.skey;
     }
-
-    TSKEY startKey = -1;
-    if (QUERY_IS_ASC_QUERY(pQuery)) {
-      startKey = pNextWin->skey;
-      if (startKey < pQuery->window.skey) {
-        startKey = pQuery->window.skey;
-      }
-    } else {
-      startKey = pNextWin->ekey;
-      if (startKey > pQuery->window.skey) {
-        startKey = pQuery->window.skey;
-      }
+  } else {
+    startKey = pNextWin->ekey;
+    if (startKey > pQuery->window.skey) {
+      startKey = pQuery->window.skey;
     }
+  }
 
-    int32_t startPos = searchFn((char *)primaryKeys, pDataBlockInfo->rows, startKey, pQuery->order.order);
+  int32_t startPos = searchFn((char *)primaryKeys, pDataBlockInfo->rows, startKey, pQuery->order.order);
 
-    /*
-     * This time window does not cover any data, try next time window,
-     * this case may happen when the time window is too small
-     */
-    if ((primaryKeys[startPos] > pNextWin->ekey && QUERY_IS_ASC_QUERY(pQuery)) ||
-        (primaryKeys[startPos] < pNextWin->skey && !QUERY_IS_ASC_QUERY(pQuery))) {
-      continue;
-    }
+  /*
+   * This time window does not cover any data, try next time window,
+   * this case may happen when the time window is too small
+   */
+  if (QUERY_IS_ASC_QUERY(pQuery) && primaryKeys[startPos] > pNextWin->ekey) {
+    TSKEY next = primaryKeys[startPos];
+    
+    pNextWin->ekey += ((next - pNextWin->ekey + pQuery->slidingTime - 1)/pQuery->slidingTime) * pQuery->slidingTime;
+    pNextWin->skey = pNextWin->ekey - pQuery->intervalTime + 1;
+  } else if ((!QUERY_IS_ASC_QUERY(pQuery)) && primaryKeys[startPos] < pNextWin->skey) {
+    TSKEY next = primaryKeys[startPos];
 
-    return startPos;
+    pNextWin->skey -= ((pNextWin->skey - next + pQuery->slidingTime - 1) / pQuery->slidingTime) * pQuery->slidingTime;
+    pNextWin->ekey = pNextWin->skey + pQuery->intervalTime - 1;
   }
+  
+  return startPos;
 }
 
 static TSKEY reviseWindowEkey(SQuery *pQuery, STimeWindow *pWindow) {
@@ -1027,7 +1031,7 @@ static bool functionNeedToExecute(SQueryRuntimeEnv *pRuntimeEnv, SQLFunctionCtx
   }
   
   // in the supplementary scan, only the following functions need to be executed
-  if (IS_REVERSE_SCAN(pRuntimeEnv)) {// && (functionId == TSDB_FUNC_TAG || functionId == TSDB_FUNC_TS)) {
+  if (IS_REVERSE_SCAN(pRuntimeEnv)) {
     return false;
   }
 
@@ -1183,7 +1187,6 @@ static int32_t tableApplyFunctionsOnBlock(SQueryRuntimeEnv *pRuntimeEnv, SDataBl
   
   STableQueryInfo* pTableQInfo = pQuery->current;
   SWindowResInfo*  pWindowResInfo = &pRuntimeEnv->windowResInfo;
-  pQuery->pos = QUERY_IS_ASC_QUERY(pQuery)? 0 : pDataBlockInfo->rows - 1;
   
   if (pQuery->numOfFilterCols > 0 || pRuntimeEnv->pTSBuf != NULL || isGroupbyNormalCol(pQuery->pGroupbyExpr)) {
     rowwiseApplyFunctions(pRuntimeEnv, pStatis, pDataBlockInfo, pWindowResInfo, pDataBlock);
@@ -1598,8 +1601,7 @@ static bool onlyQueryTags(SQuery* pQuery) {
 
 /////////////////////////////////////////////////////////////////////////////////////////////
 
-void getAlignQueryTimeWindow(SQuery *pQuery, int64_t key, int64_t keyFirst, int64_t keyLast, int64_t *realSkey,
-                             int64_t *realEkey, STimeWindow *win) {
+void getAlignQueryTimeWindow(SQuery *pQuery, int64_t key, int64_t keyFirst, int64_t keyLast, STimeWindow *realWin, STimeWindow *win) {
   assert(key >= keyFirst && key <= keyLast && pQuery->slidingTime <= pQuery->intervalTime);
 
   win->skey = taosGetIntervalStartTimestamp(key, pQuery->slidingTime, pQuery->slidingTimeUnit, pQuery->precision);
@@ -1611,8 +1613,8 @@ void getAlignQueryTimeWindow(SQuery *pQuery, int64_t key, int64_t keyFirst, int6
      */
     assert(keyLast - keyFirst < pQuery->intervalTime);
 
-    *realSkey = keyFirst;
-    *realEkey = keyLast;
+    realWin->skey = keyFirst;
+    realWin->ekey = keyLast;
 
     win->ekey = INT64_MAX;
     return;
@@ -1620,17 +1622,8 @@ void getAlignQueryTimeWindow(SQuery *pQuery, int64_t key, int64_t keyFirst, int6
 
   win->ekey = win->skey + pQuery->intervalTime - 1;
 
-  if (win->skey < keyFirst) {
-    *realSkey = keyFirst;
-  } else {
-    *realSkey = win->skey;
-  }
-
-  if (win->ekey < keyLast) {
-    *realEkey = win->ekey;
-  } else {
-    *realEkey = keyLast;
-  }
+  realWin->skey = (win->skey < keyFirst)? keyFirst : win->skey;
+  realWin->ekey = (win->ekey < keyLast) ? win->ekey : keyLast;
 }
 
 static void setScanLimitationByResultBuffer(SQuery *pQuery) {
@@ -1847,31 +1840,21 @@ static int32_t getInitialPageNum(SQInfo *pQInfo) {
   return num;
 }
 
-static int32_t getRowParamForMultiRowsOutput(SQuery *pQuery, bool isSTableQuery) {
-  int32_t rowparam = 1;
+#define GET_ROW_PARAM_FOR_MULTIOUTPUT(_q, tbq, sq) (((tbq) && (!sq))? (_q)->pSelectExpr[1].base.arg->argValue.i64:1)
 
-  if (isTopBottomQuery(pQuery) && (!isSTableQuery)) {
-    rowparam = pQuery->pSelectExpr[1].base.arg->argValue.i64;
-  }
-
-  return rowparam;
-}
-
-static int32_t getNumOfRowsInResultPage(SQuery *pQuery, bool isSTableQuery) {
-  int32_t rowSize = pQuery->rowSize * getRowParamForMultiRowsOutput(pQuery, isSTableQuery);
+static FORCE_INLINE int32_t getNumOfRowsInResultPage(SQuery *pQuery, bool topBotQuery, bool isSTableQuery) {
+  int32_t rowSize = pQuery->rowSize * GET_ROW_PARAM_FOR_MULTIOUTPUT(pQuery, topBotQuery, isSTableQuery);
   return (DEFAULT_INTERN_BUF_PAGE_SIZE - sizeof(tFilePage)) / rowSize;
 }
 
 char *getPosInResultPage(SQueryRuntimeEnv *pRuntimeEnv, int32_t columnIndex, SWindowResult *pResult) {
   assert(pResult != NULL && pRuntimeEnv != NULL);
 
-  SQuery *   pQuery = pRuntimeEnv->pQuery;
-  tFilePage *page = getResultBufferPageById(pRuntimeEnv->pResultBuf, pResult->pos.pageId);
+  SQuery    *pQuery = pRuntimeEnv->pQuery;
+  tFilePage *page = GET_RES_BUF_PAGE_BY_ID(pRuntimeEnv->pResultBuf, pResult->pos.pageId);
+  int32_t realRowId = pResult->pos.rowId * GET_ROW_PARAM_FOR_MULTIOUTPUT(pQuery, pRuntimeEnv->topBotQuery, pRuntimeEnv->stableQuery);
 
-  int32_t numOfRows = getNumOfRowsInResultPage(pQuery, pRuntimeEnv->stableQuery);
-  int32_t realRowId = pResult->pos.rowId * getRowParamForMultiRowsOutput(pQuery, pRuntimeEnv->stableQuery);
-
-  return ((char *)page->data) + pRuntimeEnv->offset[columnIndex] * numOfRows +
+  return ((char *)page->data) + pRuntimeEnv->offset[columnIndex] * pRuntimeEnv->numOfRowsPerPage +
          pQuery->pSelectExpr[columnIndex].bytes * realRowId;
 }
 
@@ -2034,6 +2017,7 @@ SArray *loadDataBlockOnDemand(SQueryRuntimeEnv *pRuntimeEnv, void* pQueryHandle,
     }
   
     pRuntimeEnv->summary.totalCheckedRows += pBlockInfo->rows;
+    pRuntimeEnv->summary.loadBlocks += 1;
     pDataBlock = tsdbRetrieveDataBlock(pQueryHandle, NULL);
   }
 
@@ -2143,6 +2127,7 @@ static void ensureOutputBuffer(SQueryRuntimeEnv* pRuntimeEnv, SDataBlockInfo* pB
 static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
   SQuery *pQuery = pRuntimeEnv->pQuery;
   STableQueryInfo* pTableQueryInfo = pQuery->current;
+  SQueryCostInfo*  summary  = &pRuntimeEnv->summary;
   
   qTrace("QInfo:%p query start, qrange:%" PRId64 "-%" PRId64 ", lastkey:%" PRId64 ", order:%d",
          GET_QINFO_ADDR(pRuntimeEnv), pTableQueryInfo->win.skey, pTableQueryInfo->win.ekey, pTableQueryInfo->lastKey,
@@ -2150,7 +2135,7 @@ static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
 
   TsdbQueryHandleT pQueryHandle = IS_MASTER_SCAN(pRuntimeEnv)? pRuntimeEnv->pQueryHandle : pRuntimeEnv->pSecQueryHandle;
   while (tsdbNextDataBlock(pQueryHandle)) {
-    pRuntimeEnv->summary.totalBlocks += 1;
+    summary->totalBlocks += 1;
     if (isQueryKilled(GET_QINFO_ADDR(pRuntimeEnv))) {
       return 0;
     }
@@ -2159,19 +2144,16 @@ static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
 
     // todo extract methods
     if (isIntervalQuery(pQuery) && pRuntimeEnv->windowResInfo.prevSKey == TSKEY_INITIAL_VAL) {
-      TSKEY           skey1, ekey1;
-      STimeWindow     w = TSWINDOW_INITIALIZER;
+      STimeWindow realWin = TSWINDOW_INITIALIZER, w = TSWINDOW_INITIALIZER;
       SWindowResInfo *pWindowResInfo = &pRuntimeEnv->windowResInfo;
 
       if (QUERY_IS_ASC_QUERY(pQuery)) {
-        getAlignQueryTimeWindow(pQuery, blockInfo.window.skey, blockInfo.window.skey, pQuery->window.ekey, &skey1,
-                                &ekey1, &w);
+        getAlignQueryTimeWindow(pQuery, blockInfo.window.skey, blockInfo.window.skey, pQuery->window.ekey, &realWin, &w);
         pWindowResInfo->startTime = w.skey;
         pWindowResInfo->prevSKey = w.skey;
       } else {
         // the start position of the first time window in the endpoint that spreads beyond the queried last timestamp
-        getAlignQueryTimeWindow(pQuery, blockInfo.window.ekey, pQuery->window.ekey, blockInfo.window.ekey, &skey1,
-                                &ekey1, &w);
+        getAlignQueryTimeWindow(pQuery, blockInfo.window.ekey, pQuery->window.ekey, blockInfo.window.ekey, &realWin, &w);
 
         pWindowResInfo->startTime = pQuery->window.skey;
         pWindowResInfo->prevSKey = w.skey;
@@ -2187,9 +2169,12 @@ static int64_t doScanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv) {
 
     SDataStatis *pStatis = NULL;
     SArray *pDataBlock = loadDataBlockOnDemand(pRuntimeEnv, pQueryHandle, &blockInfo, &pStatis);
+  
+    // query start position can not move into tableApplyFunctionsOnBlock due to limit/offset condition
+    pQuery->pos = QUERY_IS_ASC_QUERY(pQuery)? 0 : blockInfo.rows - 1;
     int32_t numOfRes = tableApplyFunctionsOnBlock(pRuntimeEnv, &blockInfo, pStatis, binarySearchForKey, pDataBlock);
 
-    pRuntimeEnv->summary.totalRows += blockInfo.rows;
+    summary->totalRows += blockInfo.rows;
     qTrace("QInfo:%p check data block, brange:%" PRId64 "-%" PRId64 ", numOfRows:%d, numOfRes:%d, lastKey:%"PRId64, GET_QINFO_ADDR(pRuntimeEnv),
            blockInfo.window.skey, blockInfo.window.ekey, blockInfo.rows, numOfRes, pQuery->current->lastKey);
 
@@ -2531,7 +2516,7 @@ void copyResToQueryResultBuf(SQInfo *pQInfo, SQuery *pQuery) {
 
   int32_t total = 0;
   for (int32_t i = 0; i < list.size; ++i) {
-    tFilePage *pData = getResultBufferPageById(pResultBuf, list.pData[i]);
+    tFilePage *pData = GET_RES_BUF_PAGE_BY_ID(pResultBuf, list.pData[i]);
     total += pData->num;
   }
 
@@ -2539,7 +2524,7 @@ void copyResToQueryResultBuf(SQInfo *pQInfo, SQuery *pQuery) {
 
   int32_t offset = 0;
   for (int32_t num = 0; num < list.size; ++num) {
-    tFilePage *pData = getResultBufferPageById(pResultBuf, list.pData[num]);
+    tFilePage *pData = GET_RES_BUF_PAGE_BY_ID(pResultBuf, list.pData[num]);
 
     for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
       int32_t bytes = pRuntimeEnv->pCtx[i].outputBytes;
@@ -2558,10 +2543,8 @@ void copyResToQueryResultBuf(SQInfo *pQInfo, SQuery *pQuery) {
   pQInfo->offset += 1;
 }
 
-int64_t getNumOfResultWindowRes(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pWindowRes) {
-  SQuery *pQuery = pRuntimeEnv->pQuery;
-
-  int64_t maxOutput = 0;
+int64_t getNumOfResultWindowRes(SQuery *pQuery, SWindowResult *pWindowRes) {
+//  int64_t maxOutput = 0;
   for (int32_t j = 0; j < pQuery->numOfOutput; ++j) {
     int32_t functionId = pQuery->pSelectExpr[j].base.functionId;
 
@@ -2574,12 +2557,23 @@ int64_t getNumOfResultWindowRes(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pW
     }
 
     SResultInfo *pResultInfo = &pWindowRes->resultInfo[j];
-    if (pResultInfo != NULL && maxOutput < pResultInfo->numOfRes) {
-      maxOutput = pResultInfo->numOfRes;
+    assert(pResultInfo != NULL);
+    
+    if (pResultInfo->numOfRes > 0) {
+      return pResultInfo->numOfRes;
     }
+//    if (pResultInfo != NULL && maxOutput < pResultInfo->numOfRes) {
+//      maxOutput = pResultInfo->numOfRes;
+//
+//      if (maxOutput > 0) {
+//        break;
+//      }
+//    }
+//
+//    assert(pResultInfo != NULL);
   }
 
-  return maxOutput;
+  return 0;
 }
 
 int32_t mergeIntoGroupResultImpl(SQInfo *pQInfo, SArray *pGroup) {
@@ -2612,6 +2606,8 @@ int32_t mergeIntoGroupResultImpl(SQInfo *pQInfo, SArray *pGroup) {
 
     assert(pQInfo->numOfGroupResultPages == 0);
     return 0;
+  } else if (numOfTables == 1) { // no need to merge results since only one table in each group
+  
   }
 
   SCompSupporter cs = {pTableList, posList, pQInfo};
@@ -2636,7 +2632,7 @@ int32_t mergeIntoGroupResultImpl(SQInfo *pQInfo, SArray *pGroup) {
     TSKEY ts = GET_INT64_VAL(b);
 
     assert(ts == pWindowRes->window.skey);
-    int64_t num = getNumOfResultWindowRes(pRuntimeEnv, pWindowRes);
+    int64_t num = getNumOfResultWindowRes(pQuery, pWindowRes);
     if (num <= 0) {
       cs.position[pos] += 1;
 
@@ -2699,10 +2695,11 @@ int32_t mergeIntoGroupResultImpl(SQInfo *pQInfo, SArray *pGroup) {
   displayInterResult(pQuery->sdata, pRuntimeEnv, pQuery->sdata[0]->num);
 #endif
 
-  qTrace("QInfo:%p result merge completed, elapsed time:%" PRId64 " ms", GET_QINFO_ADDR(pQuery), endt - startt);
-  tfree(pTree);
+  qTrace("QInfo:%p result merge completed for group:%d, elapsed time:%" PRId64 " ms", pQInfo, pQInfo->groupIndex, endt - startt);
+  
   tfree(pTableList);
   tfree(posList);
+  tfree(pTree);
 
   pQInfo->offset = 0;
   for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
@@ -2928,8 +2925,13 @@ void initCtxOutputBuf(SQueryRuntimeEnv *pRuntimeEnv) {
 
   for (int32_t j = 0; j < pQuery->numOfOutput; ++j) {
     int32_t functionId = pQuery->pSelectExpr[j].base.functionId;
-
     pRuntimeEnv->pCtx[j].currentStage = 0;
+    
+    SResultInfo* pResInfo = GET_RES_INFO(&pRuntimeEnv->pCtx[j]);
+    if (pResInfo->initialized) {
+      continue;
+    }
+    
     aAggs[functionId].init(&pRuntimeEnv->pCtx[j]);
   }
 }
@@ -3078,7 +3080,7 @@ static void setEnvBeforeReverseScan(SQueryRuntimeEnv *pRuntimeEnv, SQueryStatusI
     tsdbCleanupQueryHandle(pRuntimeEnv->pSecQueryHandle);
   }
 
-  pRuntimeEnv->pSecQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &pQInfo->tableIdGroupInfo);
+  pRuntimeEnv->pSecQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &pQInfo->tableIdGroupInfo, pQInfo);
 
   setQueryStatus(pQuery, QUERY_NOT_COMPLETED);
   switchCtxOrder(pRuntimeEnv);
@@ -3150,7 +3152,7 @@ void scanAllDataBlocks(SQueryRuntimeEnv *pRuntimeEnv, TSKEY start) {
       tsdbCleanupQueryHandle(pRuntimeEnv->pSecQueryHandle);
     }
 
-    pRuntimeEnv->pSecQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &pQInfo->tableIdGroupInfo);
+    pRuntimeEnv->pSecQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &pQInfo->tableIdGroupInfo, pQInfo);
     pRuntimeEnv->windowResInfo.curIndex = qstatus.windowIndex;
 
     setQueryStatus(pQuery, QUERY_NOT_COMPLETED);
@@ -3295,13 +3297,14 @@ void setExecutionContext(SQInfo *pQInfo, STableId* pTableId, int32_t groupIndex,
   setAdditionalInfo(pQInfo, pTableId, pTableQueryInfo);
 }
 
-static void setWindowResOutputBuf(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pResult) {
+void setWindowResOutputBuf(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pResult) {
   SQuery *pQuery = pRuntimeEnv->pQuery;
 
   // Note: pResult->pos[i]->num == 0, there is only fixed number of results for each group
   for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
     SQLFunctionCtx *pCtx = &pRuntimeEnv->pCtx[i];
     pCtx->aOutputBuf = getPosInResultPage(pRuntimeEnv, i, pResult);
+    
     int32_t functionId = pQuery->pSelectExpr[i].base.functionId;
     if (functionId == TSDB_FUNC_TOP || functionId == TSDB_FUNC_BOTTOM || functionId == TSDB_FUNC_DIFF) {
       pCtx->ptsOutputBuf = pRuntimeEnv->pCtx[0].aOutputBuf;
@@ -3319,6 +3322,38 @@ static void setWindowResOutputBuf(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *
   }
 }
 
+void setWindowResOutputBufInitCtx(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pResult) {
+  SQuery *pQuery = pRuntimeEnv->pQuery;
+  
+  // Note: pResult->pos[i]->num == 0, there is only fixed number of results for each group
+  for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
+    SQLFunctionCtx *pCtx = &pRuntimeEnv->pCtx[i];
+
+    pCtx->resultInfo = &pResult->resultInfo[i];
+    if (pCtx->resultInfo->complete) {
+      continue;
+    }
+    
+    pCtx->aOutputBuf = getPosInResultPage(pRuntimeEnv, i, pResult);
+    pCtx->currentStage = 0;
+  
+    int32_t functionId = pCtx->functionId;
+    if (functionId == TSDB_FUNC_TOP || functionId == TSDB_FUNC_BOTTOM || functionId == TSDB_FUNC_DIFF) {
+      pCtx->ptsOutputBuf = pRuntimeEnv->pCtx[0].aOutputBuf;
+    }
+    
+    /*
+     * set the output buffer information and intermediate buffer
+     * not all queries require the interResultBuf, such as COUNT
+     */
+    pCtx->resultInfo->superTableQ = pRuntimeEnv->stableQuery;     // set super table query flag
+  
+    if (!pCtx->resultInfo->initialized) {
+      aAggs[functionId].init(pCtx);
+    }
+  }
+}
+
 int32_t setAdditionalInfo(SQInfo *pQInfo, STableId* pTableId, STableQueryInfo *pTableQueryInfo) {
   SQueryRuntimeEnv *pRuntimeEnv = &pQInfo->runtimeEnv;
   assert(pTableQueryInfo->lastKey >= TSKEY_INITIAL_VAL);
@@ -3374,13 +3409,12 @@ void setIntervalQueryRange(SQInfo *pQInfo, TSKEY key) {
      * In ascending query, key is the first qualified timestamp. However, in the descending order query, additional
      * operations involve.
      */
-    TSKEY           skey1, ekey1;
-    STimeWindow     w = {0};
+    STimeWindow     w = TSWINDOW_INITIALIZER, realWin = TSWINDOW_INITIALIZER;
     SWindowResInfo *pWindowResInfo = &pTableQueryInfo->windowResInfo;
 
     TSKEY sk = MIN(win.skey, win.ekey);
     TSKEY ek = MAX(win.skey, win.ekey);
-    getAlignQueryTimeWindow(pQuery, win.skey, sk, ek, &skey1, &ekey1, &w);
+    getAlignQueryTimeWindow(pQuery, win.skey, sk, ek, &realWin, &w);
     pWindowResInfo->startTime = pTableQueryInfo->win.skey;  // windowSKey may be 0 in case of 1970 timestamp
 
     if (pWindowResInfo->prevSKey == TSKEY_INITIAL_VAL) {
@@ -3738,7 +3772,7 @@ static void updateOffsetVal(SQueryRuntimeEnv *pRuntimeEnv, SDataBlockInfo *pBloc
   SColumnInfoData *pColInfoData = taosArrayGet(pDataBlock, 0);
 
   // update the pQuery->limit.offset value, and pQuery->pos value
-  TSKEY *keys = (TSKEY *)pColInfoData->pData;
+  TSKEY *keys = (TSKEY *) pColInfoData->pData;
 
   // update the offset value
   pTableQueryInfo->lastKey = keys[pQuery->pos];
@@ -3800,8 +3834,7 @@ static bool skipTimeInterval(SQueryRuntimeEnv *pRuntimeEnv, TSKEY* start) {
    */
   assert(pRuntimeEnv->windowResInfo.prevSKey == TSKEY_INITIAL_VAL);
 
-  TSKEY       skey1, ekey1;
-  STimeWindow w = TSWINDOW_INITIALIZER;
+  STimeWindow w = TSWINDOW_INITIALIZER, realWin = TSWINDOW_INITIALIZER;
   
   SWindowResInfo *pWindowResInfo = &pRuntimeEnv->windowResInfo;
   STableQueryInfo *pTableQueryInfo = pQuery->current;
@@ -3811,14 +3844,12 @@ static bool skipTimeInterval(SQueryRuntimeEnv *pRuntimeEnv, TSKEY* start) {
 
     if (QUERY_IS_ASC_QUERY(pQuery)) {
       if (pWindowResInfo->prevSKey == TSKEY_INITIAL_VAL) {
-        getAlignQueryTimeWindow(pQuery, blockInfo.window.skey, blockInfo.window.skey, pQuery->window.ekey, &skey1,
-                                &ekey1, &w);
+        getAlignQueryTimeWindow(pQuery, blockInfo.window.skey, blockInfo.window.skey, pQuery->window.ekey, &realWin, &w);
         pWindowResInfo->startTime = w.skey;
         pWindowResInfo->prevSKey = w.skey;
       }
     } else {
-      getAlignQueryTimeWindow(pQuery, blockInfo.window.ekey, pQuery->window.ekey, blockInfo.window.ekey, &skey1, &ekey1,
-                              &w);
+      getAlignQueryTimeWindow(pQuery, blockInfo.window.ekey, pQuery->window.ekey, blockInfo.window.ekey, &realWin, &w);
 
       pWindowResInfo->startTime = pQuery->window.skey;
       pWindowResInfo->prevSKey = w.skey;
@@ -3939,11 +3970,11 @@ static void setupQueryHandle(void* tsdb, SQInfo* pQInfo, bool isSTableQuery) {
   }
 
   if (isFirstLastRowQuery(pQuery)) {
-    pRuntimeEnv->pQueryHandle = tsdbQueryLastRow(tsdb, &cond, &pQInfo->tableIdGroupInfo);
+    pRuntimeEnv->pQueryHandle = tsdbQueryLastRow(tsdb, &cond, &pQInfo->tableIdGroupInfo, pQInfo);
   } else if (isPointInterpoQuery(pQuery)) {
-    pRuntimeEnv->pQueryHandle = tsdbQueryRowsInExternalWindow(tsdb, &cond, &pQInfo->tableIdGroupInfo);
+    pRuntimeEnv->pQueryHandle = tsdbQueryRowsInExternalWindow(tsdb, &cond, &pQInfo->tableIdGroupInfo, pQInfo);
   } else {
-    pRuntimeEnv->pQueryHandle = tsdbQueryTables(tsdb, &cond, &pQInfo->tableIdGroupInfo);
+    pRuntimeEnv->pQueryHandle = tsdbQueryTables(tsdb, &cond, &pQInfo->tableIdGroupInfo, pQInfo);
   }
 }
 
@@ -3999,7 +4030,7 @@ int32_t doInitQInfo(SQInfo *pQInfo, void *param, void *tsdb, int32_t vgId, bool
     return code;
   }
 
-  pRuntimeEnv->numOfRowsPerPage = getNumOfRowsInResultPage(pQuery, isSTableQuery);
+  pRuntimeEnv->numOfRowsPerPage = getNumOfRowsInResultPage(pQuery, pRuntimeEnv->topBotQuery, isSTableQuery);
 
   if (isSTableQuery) {
     int32_t rows = getInitialPageNum(pQInfo);
@@ -4057,6 +4088,7 @@ int32_t doInitQInfo(SQInfo *pQInfo, void *param, void *tsdb, int32_t vgId, bool
                                               pQuery->slidingTime, pQuery->fillType, pColInfo);
   }
   
+  pRuntimeEnv->topBotQuery = isTopBottomQuery(pQuery);
   return TSDB_CODE_SUCCESS;
 }
 
@@ -4086,10 +4118,10 @@ static void enableExecutionForNextTable(SQueryRuntimeEnv *pRuntimeEnv) {
   }
 }
 
-static int64_t queryOnDataBlocks(SQInfo *pQInfo) {
+static int64_t scanMultiTableDataBlocks(SQInfo *pQInfo) {
   SQueryRuntimeEnv *pRuntimeEnv = &pQInfo->runtimeEnv;
-  SQuery *          pQuery = pRuntimeEnv->pQuery;
-  SQueryCostInfo* summary  = &pRuntimeEnv->summary;
+  SQuery*           pQuery = pRuntimeEnv->pQuery;
+  SQueryCostInfo*   summary  = &pRuntimeEnv->summary;
   
   int64_t st = taosGetTimestampMs();
 
@@ -4144,8 +4176,9 @@ static int64_t queryOnDataBlocks(SQInfo *pQInfo) {
     summary->totalRows += blockInfo.rows;
     stableApplyFunctionsOnBlock(pRuntimeEnv, &blockInfo, pStatis, pDataBlock, binarySearchForKey);
   
-    qTrace("QInfo:%p check data block, brange:%" PRId64 "-%" PRId64 ", numOfRows:%d, lastKey:%"PRId64, GET_QINFO_ADDR(pRuntimeEnv),
-           blockInfo.window.skey, blockInfo.window.ekey, blockInfo.rows, pQuery->current->lastKey);
+    qTrace("QInfo:%p check data block, uid:%"PRId64", tid:%d, brange:%" PRId64 "-%" PRId64 ", numOfRows:%d, lastKey:%" PRId64,
+           GET_QINFO_ADDR(pRuntimeEnv), blockInfo.uid, blockInfo.tid, blockInfo.window.skey, blockInfo.window.ekey,
+           blockInfo.rows, pQuery->current->lastKey);
   }
 
   int64_t et = taosGetTimestampMs();
@@ -4186,7 +4219,7 @@ static bool multiTableMultioutputHelper(SQInfo *pQInfo, int32_t index) {
     pRuntimeEnv->pQueryHandle = NULL;
   }
 
-  pRuntimeEnv->pQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &gp);
+  pRuntimeEnv->pQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &gp, pQInfo);
   taosArrayDestroy(tx);
   taosArrayDestroy(g1);
 
@@ -4252,9 +4285,9 @@ static void sequentialTableProcess(SQInfo *pQInfo) {
       }
       
       if (isFirstLastRowQuery(pQuery)) {
-        pRuntimeEnv->pQueryHandle = tsdbQueryLastRow(pQInfo->tsdb, &cond, &gp);
+        pRuntimeEnv->pQueryHandle = tsdbQueryLastRow(pQInfo->tsdb, &cond, &gp, pQInfo);
       } else {
-        pRuntimeEnv->pQueryHandle = tsdbQueryRowsInExternalWindow(pQInfo->tsdb, &cond, &gp);
+        pRuntimeEnv->pQueryHandle = tsdbQueryRowsInExternalWindow(pQInfo->tsdb, &cond, &gp, pQInfo);
       }
       
       initCtxOutputBuf(pRuntimeEnv);
@@ -4448,7 +4481,7 @@ static void doSaveContext(SQInfo *pQInfo) {
     tsdbCleanupQueryHandle(pRuntimeEnv->pSecQueryHandle);
   }
   
-  pRuntimeEnv->pSecQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &pQInfo->tableIdGroupInfo);
+  pRuntimeEnv->pSecQueryHandle = tsdbQueryTables(pQInfo->tsdb, &cond, &pQInfo->tableIdGroupInfo, pQInfo);
   
   setQueryStatus(pQuery, QUERY_NOT_COMPLETED);
   switchCtxOrder(pRuntimeEnv);
@@ -4515,7 +4548,7 @@ static void multiTableQueryProcess(SQInfo *pQInfo) {
          pQuery->window.skey, pQuery->window.ekey, pQuery->order.order);
 
   // do check all qualified data blocks
-  int64_t el = queryOnDataBlocks(pQInfo);
+  int64_t el = scanMultiTableDataBlocks(pQInfo);
   qTrace("QInfo:%p master scan completed, elapsed time: %lldms, reverse scan start", pQInfo, el);
 
   // query error occurred or query is killed, abort current execution
@@ -4530,7 +4563,7 @@ static void multiTableQueryProcess(SQInfo *pQInfo) {
   if (needReverseScan(pQuery)) {
     doSaveContext(pQInfo);
 
-    el = queryOnDataBlocks(pQInfo);
+    el = scanMultiTableDataBlocks(pQInfo);
     qTrace("QInfo:%p reversed scan completed, elapsed time: %lldms", pQInfo, el);
 
     doRestoreContext(pQInfo);
@@ -4817,7 +4850,7 @@ static void tableQueryImpl(SQInfo *pQInfo) {
   /* check if query is killed or not */
   if (isQueryKilled(pQInfo)) {
     qTrace("QInfo:%p query is killed", pQInfo);
-  } else {// todo set the table uid and tid in log
+  } else {
     qTrace("QInfo:%p query paused, %" PRId64 " rows returned, numOfTotal:%" PRId64 " rows",
         pQInfo, pQuery->rec.rows, pQuery->rec.total + pQuery->rec.rows);
   }
@@ -4909,9 +4942,12 @@ static bool validateQuerySourceCols(SQueryTableMsg *pQueryMsg, SSqlFuncMsg** pEx
     return false;
   } else if (numOfTotal == 0) {
     for(int32_t i = 0; i < pQueryMsg->numOfOutput; ++i) {
-      if (pExprMsg[i]->functionId != TSDB_FUNC_TAGPRJ) {
-        return false;
+      if ((pExprMsg[i]->functionId == TSDB_FUNC_TAGPRJ) ||
+          (pExprMsg[i]->functionId == TSDB_FUNC_TID_TAG && pExprMsg[i]->colInfo.colId == TSDB_TBNAME_COLUMN_INDEX)) {
+        continue;
       }
+      
+      return false;
     }
   }
 
@@ -6002,53 +6038,58 @@ static void buildTagQueryResult(SQInfo* pQInfo) {
   num = taosArrayGetSize(pa);
 
   assert(num == pQInfo->groupInfo.numOfTables);
-//  int16_t type, bytes;
-
   int32_t functionId = pQuery->pSelectExpr[0].base.functionId;
   if (functionId == TSDB_FUNC_TID_TAG) { // return the tags & table Id
     assert(pQuery->numOfOutput == 1);
+    
     SExprInfo* pExprInfo = &pQuery->pSelectExpr[0];
-
     int32_t rsize = pExprInfo->bytes;
 
     for(int32_t i = 0; i < num; ++i) {
-      SGroupItem* item = taosArrayGet(pa, i);
+      SGroupItem *item = taosArrayGet(pa, i);
 
-      char* output = pQuery->sdata[0]->data + i * rsize;
+      char *output = pQuery->sdata[0]->data + i * rsize;
       varDataSetLen(output, rsize - VARSTR_HEADER_SIZE);
-      
+
       output = varDataVal(output);
-      *(int64_t*) output = item->id.uid;  // memory align problem, todo serialize
+      *(int64_t *)output = item->id.uid;  // memory align problem, todo serialize
       output += sizeof(item->id.uid);
 
-      *(int32_t*) output = item->id.tid;
+      *(int32_t *)output = item->id.tid;
       output += sizeof(item->id.tid);
 
-      *(int32_t*) output = pQInfo->vgId;
+      *(int32_t *)output = pQInfo->vgId;
       output += sizeof(pQInfo->vgId);
 
       int16_t bytes = pExprInfo->bytes;
       int16_t type = pExprInfo->type;
-      
-      char* val = tsdbGetTableTagVal(pQInfo->tsdb, &item->id, pExprInfo->base.colInfo.colId, type, bytes);
-      
-      // todo refactor
-      if (type == TSDB_DATA_TYPE_BINARY || type == TSDB_DATA_TYPE_NCHAR) {
-        if (val == NULL) {
-          setVardataNull(output, type);
-        } else {
-          memcpy(output, val, varDataTLen(val));
-        }
+
+      if (pExprInfo->base.colInfo.colId == TSDB_TBNAME_COLUMN_INDEX) {
+        char *data = tsdbGetTableName(pQInfo->tsdb, &item->id);
+        memcpy(output, data, varDataTLen(data));
       } else {
-        if (val == NULL) {
-          setNull(output, type, bytes);
+        char *val = tsdbGetTableTagVal(pQInfo->tsdb, &item->id, pExprInfo->base.colInfo.colId, type, bytes);
+
+        // todo refactor
+        if (type == TSDB_DATA_TYPE_BINARY || type == TSDB_DATA_TYPE_NCHAR) {
+          if (val == NULL) {
+            setVardataNull(output, type);
+          } else {
+            memcpy(output, val, varDataTLen(val));
+          }
         } else {
-          memcpy(output, val, bytes);
+          if (val == NULL) {
+            setNull(output, type, bytes);
+          } else {
+            memcpy(output, val, bytes);
+          }
         }
       }
     }
-
+  
+    pQInfo->tableIndex = pQInfo->groupInfo.numOfTables;
     qTrace("QInfo:%p create (tableId, tag) info completed, rows:%d", pQInfo, num);
+    
   } else {  // return only the tags|table name etc.
     for(int32_t i = 0; i < num; ++i) {
       SExprInfo* pExprInfo = pQuery->pSelectExpr;

src/query/src/qUtil.c

@@ -206,11 +206,6 @@ bool isWindowResClosed(SWindowResInfo *pWindowResInfo, int32_t slot) {
   return (getWindowResult(pWindowResInfo, slot)->status.closed == true);
 }
 
-int32_t curTimeWindow(SWindowResInfo *pWindowResInfo) {
-  assert(pWindowResInfo->curIndex >= 0 && pWindowResInfo->curIndex < pWindowResInfo->size);
-  return pWindowResInfo->curIndex;
-}
-
 void closeTimeWindow(SWindowResInfo *pWindowResInfo, int32_t slot) {
   getWindowResult(pWindowResInfo, slot)->status.closed = true;
 }

src/query/src/qextbuffer.c

@@ -356,17 +356,15 @@ static FORCE_INLINE int32_t primaryKeyComparator(int64_t f1, int64_t f2, int32_t
 static FORCE_INLINE int32_t columnValueAscendingComparator(char *f1, char *f2, int32_t type, int32_t bytes) {
   switch (type) {
     case TSDB_DATA_TYPE_INT: {
-      int32_t first = *(int32_t *)f1;
-      int32_t second = *(int32_t *)f2;
+      int32_t first  = *(int32_t *) f1;
+      int32_t second = *(int32_t *) f2;
       if (first == second) {
         return 0;
       }
       return (first < second) ? -1 : 1;
     };
     case TSDB_DATA_TYPE_DOUBLE: {
-      //double first = *(double *)f1;
-      double first = GET_DOUBLE_VAL(f1);
-      //double second = *(double *)f2;
+      double first  = GET_DOUBLE_VAL(f1);
       double second = GET_DOUBLE_VAL(f2);
       if (first == second) {
         return 0;
@@ -374,9 +372,7 @@ static FORCE_INLINE int32_t columnValueAscendingComparator(char *f1, char *f2, i
       return (first < second) ? -1 : 1;
     };
     case TSDB_DATA_TYPE_FLOAT: {
-      //float first = *(float *)f1;
-      //float second = *(float *)f2;
-      float first = GET_FLOAT_VAL(f1);
+      float first  = GET_FLOAT_VAL(f1);
       float second = GET_FLOAT_VAL(f2);
       if (first == second) {
         return 0;
@@ -439,9 +435,9 @@ int32_t compare_a(tOrderDescriptor *pDescriptor, int32_t numOfRows1, int32_t s1,
                   int32_t s2, char *data2) {
   assert(numOfRows1 == numOfRows2);
 
-  int32_t cmpCnt = pDescriptor->orderIdx.numOfCols;
+  int32_t cmpCnt = pDescriptor->orderInfo.numOfCols;
   for (int32_t i = 0; i < cmpCnt; ++i) {
-    int32_t colIdx = pDescriptor->orderIdx.pData[i];
+    int32_t colIdx = pDescriptor->orderInfo.pData[i];
 
     char *f1 = COLMODEL_GET_VAL(data1, pDescriptor->pColumnModel, numOfRows1, s1, colIdx);
     char *f2 = COLMODEL_GET_VAL(data2, pDescriptor->pColumnModel, numOfRows2, s2, colIdx);
@@ -471,9 +467,9 @@ int32_t compare_d(tOrderDescriptor *pDescriptor, int32_t numOfRows1, int32_t s1,
                   int32_t s2, char *data2) {
   assert(numOfRows1 == numOfRows2);
 
-  int32_t cmpCnt = pDescriptor->orderIdx.numOfCols;
+  int32_t cmpCnt = pDescriptor->orderInfo.numOfCols;
   for (int32_t i = 0; i < cmpCnt; ++i) {
-    int32_t colIdx = pDescriptor->orderIdx.pData[i];
+    int32_t colIdx = pDescriptor->orderInfo.pData[i];
 
     char *f1 = COLMODEL_GET_VAL(data1, pDescriptor->pColumnModel, numOfRows1, s1, colIdx);
     char *f2 = COLMODEL_GET_VAL(data2, pDescriptor->pColumnModel, numOfRows2, s2, colIdx);
@@ -563,13 +559,13 @@ static void median(tOrderDescriptor *pDescriptor, int32_t numOfRows, int32_t sta
   int32_t midIdx = ((end - start) >> 1) + start;
 
 #if defined(_DEBUG_VIEW)
-  int32_t f = pDescriptor->orderIdx.pData[0];
+  int32_t f = pDescriptor->orderInfo.pData[0];
 
   char *midx = COLMODEL_GET_VAL(data, pDescriptor->pColumnModel, numOfRows, midIdx, f);
   char *startx = COLMODEL_GET_VAL(data, pDescriptor->pColumnModel, numOfRows, start, f);
   char *endx = COLMODEL_GET_VAL(data, pDescriptor->pColumnModel, numOfRows, end, f);
 
-  int32_t colIdx = pDescriptor->orderIdx.pData[0];
+  int32_t colIdx = pDescriptor->orderInfo.pData[0];
   tSortDataPrint(pDescriptor->pColumnModel->pFields[colIdx].field.type, "before", startx, midx, endx);
 #endif
 
@@ -596,7 +592,7 @@ static void median(tOrderDescriptor *pDescriptor, int32_t numOfRows, int32_t sta
 }
 
 static UNUSED_FUNC void tRowModelDisplay(tOrderDescriptor *pDescriptor, int32_t numOfRows, char *d, int32_t len) {
-  int32_t colIdx = pDescriptor->orderIdx.pData[0];
+  int32_t colIdx = pDescriptor->orderInfo.pData[0];
 
   for (int32_t i = 0; i < len; ++i) {
     char *startx = COLMODEL_GET_VAL(d, pDescriptor->pColumnModel, numOfRows, i, colIdx);
@@ -1062,9 +1058,9 @@ tOrderDescriptor *tOrderDesCreate(const int32_t *orderColIdx, int32_t numOfOrder
   desc->pColumnModel = pModel;
   desc->tsOrder = tsOrderType;
 
-  desc->orderIdx.numOfCols = numOfOrderCols;
+  desc->orderInfo.numOfCols = numOfOrderCols;
   for (int32_t i = 0; i < numOfOrderCols; ++i) {
-    desc->orderIdx.pData[i] = orderColIdx[i];
+    desc->orderInfo.pData[i] = orderColIdx[i];
   }
 
   return desc;

src/query/src/qresultBuf.c

@@ -50,12 +50,6 @@ int32_t createDiskbasedResultBuffer(SDiskbasedResultBuf** pResultBuf, int32_t si
   return TSDB_CODE_SUCCESS;
 }
 
-tFilePage* getResultBufferPageById(SDiskbasedResultBuf* pResultBuf, int32_t id) {
-  assert(id < pResultBuf->numOfPages && id >= 0);
-
-  return (tFilePage*)(pResultBuf->pBuf + DEFAULT_INTERN_BUF_PAGE_SIZE * id);
-}
-
 int32_t getNumOfResultBufGroupId(SDiskbasedResultBuf* pResultBuf) { return taosHashGetSize(pResultBuf->idsTable); }
 
 int32_t getResBufSize(SDiskbasedResultBuf* pResultBuf) { return pResultBuf->totalBufSize; }
@@ -169,7 +163,7 @@ tFilePage* getNewDataBuf(SDiskbasedResultBuf* pResultBuf, int32_t groupId, int32
   *pageId = (pResultBuf->allocateId++);
   registerPageId(pResultBuf, groupId, *pageId);
 
-  tFilePage* page = getResultBufferPageById(pResultBuf, *pageId);
+  tFilePage* page = GET_RES_BUF_PAGE_BY_ID(pResultBuf, *pageId);
   
   // clear memory for the new page
   memset(page, 0, DEFAULT_INTERN_BUF_PAGE_SIZE);

src/query/src/sql.c

@@ -126,17 +126,17 @@ typedef union {
 #define ParseARG_FETCH SSqlInfo* pInfo = yypParser->pInfo
 #define ParseARG_STORE yypParser->pInfo = pInfo
 #define YYFALLBACK 1
-#define YYNSTATE             247
+#define YYNSTATE             241
 #define YYNRULE              220
 #define YYNTOKEN             205
-#define YY_MAX_SHIFT         246
-#define YY_MIN_SHIFTREDUCE   403
-#define YY_MAX_SHIFTREDUCE   622
-#define YY_ERROR_ACTION      623
-#define YY_ACCEPT_ACTION     624
-#define YY_NO_ACTION         625
-#define YY_MIN_REDUCE        626
-#define YY_MAX_REDUCE        845
+#define YY_MAX_SHIFT         240
+#define YY_MIN_SHIFTREDUCE   397
+#define YY_MAX_SHIFTREDUCE   616
+#define YY_ERROR_ACTION      617
+#define YY_ACCEPT_ACTION     618
+#define YY_NO_ACTION         619
+#define YY_MIN_REDUCE        620
+#define YY_MAX_REDUCE        839
 /************* End control #defines *******************************************/
 
 /* Define the yytestcase() macro to be a no-op if is not already defined
@@ -202,63 +202,63 @@ typedef union {
 **  yy_default[]       Default action for each state.
 **
 *********** Begin parsing tables **********************************************/
-#define YY_ACTTAB_COUNT (547)
+#define YY_ACTTAB_COUNT (541)
 static const YYACTIONTYPE yy_action[] = {
- /*     0 */   724,  444,  723,   11,  722,  134,  624,  246,  725,  445,
- /*    10 */   727,  726,  764,   41,   43,   21,   35,   36,  153,  244,
- /*    20 */   135,   29,  135,  444,  203,   39,   37,   40,   38,  158,
- /*    30 */   833,  445,  832,   34,   33,  139,  135,   32,   31,   30,
- /*    40 */    41,   43,  753,   35,   36,  157,  833,  166,   29,  739,
- /*    50 */   103,  203,   39,   37,   40,   38,  188,   21,  103,   99,
- /*    60 */    34,   33,  761,  155,   32,   31,   30,  404,  405,  406,
- /*    70 */   407,  408,  409,  410,  411,  412,  413,  414,  415,  245,
- /*    80 */   444,  742,   41,   43,  103,   35,   36,  103,  445,  168,
- /*    90 */    29,  738,   21,  203,   39,   37,   40,   38,   32,   31,
- /*   100 */    30,   56,   34,   33,  753,  787,   32,   31,   30,   43,
- /*   110 */   191,   35,   36,  788,  829,  198,   29,   21,  154,  203,
- /*   120 */    39,   37,   40,   38,  167,  578,  739,    8,   34,   33,
- /*   130 */    61,  113,   32,   31,   30,  665,   35,   36,  126,   59,
- /*   140 */   200,   29,   58,   17,  203,   39,   37,   40,   38,  221,
- /*   150 */    26,  739,  169,   34,   33,  220,  219,   32,   31,   30,
- /*   160 */    16,  239,  214,  238,  213,  212,  211,  237,  210,  236,
- /*   170 */   235,  209,  720,  828,  709,  710,  711,  712,  713,  714,
- /*   180 */   715,  716,  717,  718,  719,  162,  591,  234,   76,  582,
- /*   190 */   165,  585,  240,  588,  234,  162,  591,   98,  827,  582,
- /*   200 */   225,  585,   60,  588,   26,  162,  591,   12,  742,  582,
- /*   210 */   742,  585,  674,  588,   27,  126,   21,  159,  160,   34,
- /*   220 */    33,  202,  842,   32,   31,   30,  148,  159,  160,  740,
- /*   230 */   536,  539,   88,   87,  142,   18,  666,  159,  160,  126,
- /*   240 */   147,  559,  560,   39,   37,   40,   38,   50,  226,  550,
- /*   250 */   739,   34,   33,   46,  507,   32,   31,   30,  523,  531,
- /*   260 */    17,  520,  151,  521,   51,  522,  190,   26,   16,  239,
- /*   270 */   152,  238,  243,  242,   95,  237,  551,  236,  235,  177,
- /*   280 */    14,   42,  223,  222,  580,  741,  185,  187,  182,  170,
- /*   290 */   171,   42,  590,  584,  150,  587,   74,   78,   83,   86,
- /*   300 */    77,   42,  590,  161,  608,  592,   80,  589,   13,   13,
- /*   310 */   140,  583,  590,  586,  513,   47,  141,  589,   46,  798,
- /*   320 */   581,  116,  117,   68,   64,   67,  143,  589,  130,  128,
- /*   330 */    91,   90,   89,  512,   48,  207,  527,   22,  528,   22,
- /*   340 */   144,    3,   73,   72,   10,    9,  145,  525,  146,  526,
- /*   350 */    85,   84,  137,  797,  133,  138,  136,  163,  794,  524,
- /*   360 */   793,  164,  763,  733,  224,  100,  755,  780,  779,  114,
- /*   370 */    26,  115,  112,  676,  208,  131,   24,  217,  673,  218,
- /*   380 */   841,   70,  840,  838,  118,  694,   25,   93,   23,  132,
- /*   390 */   663,   79,  189,  546,  661,  192,   81,   82,  659,  658,
- /*   400 */   172,  127,  656,  196,  655,  654,  653,  652,  644,  129,
- /*   410 */   650,  648,  646,   52,  752,  767,   49,   44,  768,  781,
- /*   420 */   201,  199,  197,  195,  193,   28,  216,   75,  227,  228,
- /*   430 */   229,  230,  205,  232,  231,   53,  233,  241,  622,  149,
- /*   440 */   173,   62,   65,  174,  176,  175,  621,  178,  179,  180,
- /*   450 */   181,  657,  121,  120,  695,  125,  119,  122,  123,   92,
- /*   460 */   124,  651,    1,  106,  104,  737,   94,  105,  620,  109,
- /*   470 */   107,  108,  110,  111,    2,  184,  613,  183,  186,  190,
- /*   480 */   533,   55,  547,  156,  101,   57,  552,  194,  102,    5,
- /*   490 */     6,   63,  484,  593,    4,   19,   20,   15,  204,    7,
- /*   500 */   206,  481,  479,  478,  477,  475,  448,  215,   66,   45,
- /*   510 */    22,  509,  508,   69,  506,   54,  469,  467,  459,  465,
- /*   520 */   461,  463,  457,  455,   71,  483,  482,  480,  476,  474,
- /*   530 */    46,  446,  419,  417,  626,  625,  625,  625,  625,  625,
- /*   540 */    96,  625,  625,  625,  625,  625,   97,
+ /*     0 */   718,  438,  717,   11,  716,  134,  618,  240,  719,  439,
+ /*    10 */   721,  720,  758,   41,   43,   21,   35,   36,  153,  238,
+ /*    20 */   135,   29,  135,  438,  197,   39,   37,   40,   38,  158,
+ /*    30 */   827,  439,  826,   34,   33,  139,  135,   32,   31,   30,
+ /*    40 */    41,   43,  747,   35,   36,  157,  827,  166,   29,  733,
+ /*    50 */   103,  197,   39,   37,   40,   38,  182,   21,  103,   99,
+ /*    60 */    34,   33,  755,  155,   32,   31,   30,  398,  399,  400,
+ /*    70 */   401,  402,  403,  404,  405,  406,  407,  408,  409,  239,
+ /*    80 */   438,  736,   41,   43,  103,   35,   36,  103,  439,  168,
+ /*    90 */    29,  732,   21,  197,   39,   37,   40,   38,   32,   31,
+ /*   100 */    30,   56,   34,   33,  747,  781,   32,   31,   30,   43,
+ /*   110 */   185,   35,   36,  782,  823,  192,   29,   21,  154,  197,
+ /*   120 */    39,   37,   40,   38,  167,  572,  733,    8,   34,   33,
+ /*   130 */    61,  113,   32,   31,   30,  659,   35,   36,  126,   59,
+ /*   140 */   194,   29,   58,   17,  197,   39,   37,   40,   38,  215,
+ /*   150 */    26,  733,  169,   34,   33,  214,  213,   32,   31,   30,
+ /*   160 */    16,  233,  208,  232,  207,  206,  205,  231,  204,  230,
+ /*   170 */   229,  203,  714,  219,  703,  704,  705,  706,  707,  708,
+ /*   180 */   709,  710,  711,  712,  713,  162,  585,   50,   60,  576,
+ /*   190 */   175,  579,  165,  582,  234,  162,  585,  179,  178,  576,
+ /*   200 */    27,  579,  734,  582,   51,  162,  585,   12,   98,  576,
+ /*   210 */   736,  579,  736,  582,  228,   26,   21,  159,  160,   34,
+ /*   220 */    33,  196,  836,   32,   31,   30,  148,  159,  160,   76,
+ /*   230 */   822,  533,   88,   87,  142,  228,  668,  159,  160,  126,
+ /*   240 */   147,  553,  554,   39,   37,   40,   38,  821,  220,  544,
+ /*   250 */   733,   34,   33,   46,  501,   32,   31,   30,  517,  525,
+ /*   260 */    17,  514,  151,  515,  152,  516,  184,   26,   16,  233,
+ /*   270 */   140,  232,  237,  236,   95,  231,  660,  230,  229,  126,
+ /*   280 */   530,   42,  217,  216,  578,   18,  581,  181,  161,  170,
+ /*   290 */   171,   42,  584,  577,  150,  580,   74,   78,   83,   86,
+ /*   300 */    77,   42,  584,  574,  545,  602,   80,  583,   14,   13,
+ /*   310 */   141,  586,  584,  143,  507,   13,   47,  583,   46,   73,
+ /*   320 */    72,  116,  117,   68,   64,   67,    3,  583,  130,  128,
+ /*   330 */    91,   90,   89,  506,  201,   48,  144,   22,   22,  575,
+ /*   340 */   521,  519,  522,  520,   10,    9,   85,   84,  145,  146,
+ /*   350 */   137,  133,  138,  735,  136,  792,  791,  163,  788,  518,
+ /*   360 */   787,  164,  757,  727,  218,  100,  749,  774,  773,  114,
+ /*   370 */    26,  115,  112,  670,  202,  131,   24,  211,  667,  212,
+ /*   380 */   835,   70,  834,  832,  118,  688,   25,  183,   23,  132,
+ /*   390 */   657,   79,   93,  540,  655,  186,   81,   82,  653,  652,
+ /*   400 */   172,  190,  127,  746,  650,  649,  648,  647,  646,  638,
+ /*   410 */   129,  644,  642,   52,  640,   44,   49,  195,  761,  762,
+ /*   420 */   775,  191,  193,  189,  187,   28,  210,   75,  221,  222,
+ /*   430 */   223,  224,  225,  199,  226,  227,  235,   53,  616,  174,
+ /*   440 */   615,  149,   62,  173,   65,  176,  177,  614,  180,  651,
+ /*   450 */   607,  184,   92,  527,  645,  541,  120,  689,  121,  122,
+ /*   460 */   119,  123,  125,  124,   94,  104,    1,  731,  105,  111,
+ /*   470 */   108,  106,  107,  109,  110,    2,   55,   57,  101,  156,
+ /*   480 */   188,    5,  546,  102,   19,    6,  587,   20,    4,   15,
+ /*   490 */    63,    7,  198,  478,  200,  475,  473,  472,  471,  469,
+ /*   500 */   442,  209,   66,   45,   69,   71,   22,  503,  502,  500,
+ /*   510 */    54,  463,  461,  453,  459,  455,  457,  451,  449,  477,
+ /*   520 */   476,  474,  470,  468,   46,  440,   96,  413,  411,  620,
+ /*   530 */   619,  619,  619,  619,  619,  619,  619,  619,  619,  619,
+ /*   540 */    97,
 };
 static const YYCODETYPE yy_lookahead[] = {
  /*     0 */   225,    1,  227,  258,  229,  258,  206,  207,  233,    9,
@@ -278,44 +278,44 @@ static const YYCODETYPE yy_lookahead[] = {
  /*   140 */   262,   21,  264,   97,   24,   25,   26,   27,   28,  241,
  /*   150 */   104,  243,  126,   33,   34,  129,  130,   37,   38,   39,
  /*   160 */    85,   86,   87,   88,   89,   90,   91,   92,   93,   94,
- /*   170 */    95,   96,  225,  258,  227,  228,  229,  230,  231,  232,
- /*   180 */   233,  234,  235,  236,  237,    1,    2,   78,   72,    5,
- /*   190 */   226,    7,  226,    9,   78,    1,    2,   97,  258,    5,
- /*   200 */   209,    7,  245,    9,  104,    1,    2,   44,  244,    5,
- /*   210 */   244,    7,  213,    9,  257,  216,  209,   33,   34,   33,
- /*   220 */    34,   37,  244,   37,   38,   39,   63,   33,   34,  238,
- /*   230 */   102,   37,   69,   70,   71,  107,  213,   33,   34,  216,
- /*   240 */    77,  114,  115,   25,   26,   27,   28,  102,  241,   98,
+ /*   170 */    95,   96,  225,  209,  227,  228,  229,  230,  231,  232,
+ /*   180 */   233,  234,  235,  236,  237,    1,    2,  102,  245,    5,
+ /*   190 */   125,    7,  226,    9,  226,    1,    2,  132,  133,    5,
+ /*   200 */   257,    7,  238,    9,  119,    1,    2,   44,   97,    5,
+ /*   210 */   244,    7,  244,    9,   78,  104,  209,   33,   34,   33,
+ /*   220 */    34,   37,  244,   37,   38,   39,   63,   33,   34,   72,
+ /*   230 */   258,   37,   69,   70,   71,   78,  213,   33,   34,  216,
+ /*   240 */    77,  114,  115,   25,   26,   27,   28,  258,  241,   98,
  /*   250 */   243,   33,   34,  102,    5,   37,   38,   39,    2,   98,
- /*   260 */    97,    5,  258,    7,  119,    9,  105,  104,   85,   86,
- /*   270 */   258,   88,   60,   61,   62,   92,   98,   94,   95,  125,
- /*   280 */   102,   97,   33,   34,    1,  244,  132,  124,  134,   33,
+ /*   260 */    97,    5,  258,    7,  258,    9,  105,  104,   85,   86,
+ /*   270 */   258,   88,   60,   61,   62,   92,  213,   94,   95,  216,
+ /*   280 */   102,   97,   33,   34,    5,  107,    7,  124,   59,   33,
  /*   290 */    34,   97,  108,    5,  131,    7,   64,   65,   66,   67,
- /*   300 */    68,   97,  108,   59,   98,   98,   74,  123,  102,  102,
- /*   310 */   258,    5,  108,    7,   98,  102,  258,  123,  102,  239,
- /*   320 */    37,   64,   65,   66,   67,   68,  258,  123,   64,   65,
- /*   330 */    66,   67,   68,   98,  121,   98,    5,  102,    7,  102,
- /*   340 */   258,   97,  127,  128,  127,  128,  258,    5,  258,    7,
- /*   350 */    72,   73,  258,  239,  258,  258,  258,  239,  239,  103,
+ /*   300 */    68,   97,  108,    1,   98,   98,   74,  123,  102,  102,
+ /*   310 */   258,   98,  108,  258,   98,  102,  102,  123,  102,  127,
+ /*   320 */   128,   64,   65,   66,   67,   68,   97,  123,   64,   65,
+ /*   330 */    66,   67,   68,   98,   98,  121,  258,  102,  102,   37,
+ /*   340 */     5,    5,    7,    7,  127,  128,   72,   73,  258,  258,
+ /*   350 */   258,  258,  258,  244,  258,  239,  239,  239,  239,  103,
  /*   360 */   239,  239,  209,  240,  239,  209,  242,  265,  265,  209,
  /*   370 */   104,  209,  246,  209,  209,  209,  209,  209,  209,  209,
- /*   380 */   209,  209,  209,  209,  209,  209,  209,   59,  209,  209,
- /*   390 */   209,  209,  242,  108,  209,  261,  209,  209,  209,  209,
- /*   400 */   209,  209,  209,  261,  209,  209,  209,  209,  209,  209,
- /*   410 */   209,  209,  209,  118,  255,  210,  120,  117,  210,  210,
- /*   420 */   112,  116,  111,  110,  109,  122,   75,   84,   83,   49,
- /*   430 */    80,   82,  210,   81,   53,  210,   79,   75,    5,  210,
- /*   440 */   133,  214,  214,    5,   58,  133,    5,  133,    5,  133,
- /*   450 */    58,  210,  218,  222,  224,  217,  223,  221,  219,  211,
- /*   460 */   220,  210,  215,  252,  254,  242,  211,  253,    5,  249,
- /*   470 */   251,  250,  248,  247,  212,   58,   87,  133,  125,  105,
- /*   480 */    98,  106,   98,    1,   97,  102,   98,   97,   97,  113,
- /*   490 */   113,   72,    9,   98,   97,  102,  102,   97,   99,   97,
- /*   500 */    99,    5,    5,    5,    5,    5,   76,   15,   72,   16,
- /*   510 */   102,    5,    5,  128,   98,   97,    5,    5,    5,    5,
- /*   520 */     5,    5,    5,    5,  128,    5,    5,    5,    5,    5,
- /*   530 */   102,   76,   59,   58,    0,  269,  269,  269,  269,  269,
- /*   540 */    21,  269,  269,  269,  269,  269,   21,  269,  269,  269,
+ /*   380 */   209,  209,  209,  209,  209,  209,  209,  242,  209,  209,
+ /*   390 */   209,  209,   59,  108,  209,  261,  209,  209,  209,  209,
+ /*   400 */   209,  261,  209,  255,  209,  209,  209,  209,  209,  209,
+ /*   410 */   209,  209,  209,  118,  209,  117,  120,  112,  210,  210,
+ /*   420 */   210,  111,  116,  110,  109,  122,   75,   84,   83,   49,
+ /*   430 */    80,   82,   53,  210,   81,   79,   75,  210,    5,    5,
+ /*   440 */     5,  210,  214,  134,  214,  134,    5,    5,  125,  210,
+ /*   450 */    87,  105,  211,   98,  210,   98,  222,  224,  218,  221,
+ /*   460 */   223,  219,  217,  220,  211,  254,  215,  242,  253,  247,
+ /*   470 */   250,  252,  251,  249,  248,  212,  106,  102,   97,    1,
+ /*   480 */    97,  113,   98,   97,  102,  113,   98,  102,   97,   97,
+ /*   490 */    72,   97,   99,    9,   99,    5,    5,    5,    5,    5,
+ /*   500 */    76,   15,   72,   16,  128,  128,  102,    5,    5,   98,
+ /*   510 */    97,    5,    5,    5,    5,    5,    5,    5,    5,    5,
+ /*   520 */     5,    5,    5,    5,  102,   76,   21,   59,   58,    0,
+ /*   530 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
+ /*   540 */    21,  269,  269,  269,  269,  269,  269,  269,  269,  269,
  /*   550 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
  /*   560 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
  /*   570 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
@@ -335,84 +335,83 @@ static const YYCODETYPE yy_lookahead[] = {
  /*   710 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
  /*   720 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
  /*   730 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
- /*   740 */   269,  269,  269,  269,  269,  269,  269,  269,  269,  269,
- /*   750 */   269,  269,
+ /*   740 */   269,  269,  269,  269,  269,  269,
 };
-#define YY_SHIFT_COUNT    (246)
+#define YY_SHIFT_COUNT    (240)
 #define YY_SHIFT_MIN      (0)
-#define YY_SHIFT_MAX      (534)
+#define YY_SHIFT_MAX      (529)
 static const unsigned short int yy_shift_ofst[] = {
  /*     0 */   163,   75,  183,  184,  204,   79,   79,   79,   79,   79,
  /*    10 */    79,    0,   22,  204,  256,  256,  256,   46,   79,   79,
- /*    20 */    79,   79,   79,  116,  109,  109,  547,  194,  204,  204,
+ /*    20 */    79,   79,   79,  157,  136,  136,  541,  194,  204,  204,
  /*    30 */   204,  204,  204,  204,  204,  204,  204,  204,  204,  204,
  /*    40 */   204,  204,  204,  204,  204,  256,  256,  249,  249,  249,
- /*    50 */   249,  249,  249,   30,  249,  100,   79,   79,  127,  127,
- /*    60 */   128,   79,   79,   79,   79,   79,   79,   79,   79,   79,
+ /*    50 */   249,  249,  249,   30,  249,  111,   79,   79,  127,  127,
+ /*    60 */   178,   79,   79,   79,   79,   79,   79,   79,   79,   79,
  /*    70 */    79,   79,   79,   79,   79,   79,   79,   79,   79,   79,
  /*    80 */    79,   79,   79,   79,   79,   79,   79,   79,   79,   79,
- /*    90 */    79,   79,   79,   79,   79,   79,   79,   79,  266,  328,
- /*   100 */   328,  285,  285,  328,  295,  296,  300,  308,  305,  311,
- /*   110 */   313,  315,  303,  266,  328,  328,  351,  351,  328,  343,
- /*   120 */   345,  380,  350,  349,  381,  352,  357,  328,  362,  328,
- /*   130 */   362,  547,  547,   27,   69,   69,   69,   95,  120,  218,
+ /*    90 */    79,   79,   79,   79,   79,   79,   79,   79,  266,  333,
+ /*   100 */   333,  285,  285,  333,  295,  296,  298,  305,  306,  310,
+ /*   110 */   313,  315,  303,  266,  333,  333,  351,  351,  333,  343,
+ /*   120 */   345,  380,  350,  349,  379,  353,  356,  333,  361,  333,
+ /*   130 */   361,  541,  541,   27,   69,   69,   69,   95,  120,  218,
  /*   140 */   218,  218,  232,  186,  186,  186,  186,  257,  264,   26,
- /*   150 */   154,   61,   61,  212,  161,  151,  178,  206,  207,  288,
- /*   160 */   306,  283,  244,  213,  145,  216,  235,  237,  215,  217,
- /*   170 */   331,  342,  278,  433,  307,  438,  312,  386,  441,  314,
- /*   180 */   443,  316,  392,  463,  344,  417,  389,  353,  374,  382,
- /*   190 */   375,  383,  384,  387,  482,  390,  388,  391,  393,  376,
- /*   200 */   394,  377,  395,  397,  400,  399,  402,  401,  419,  483,
- /*   210 */   496,  497,  498,  499,  500,  430,  492,  436,  493,  385,
- /*   220 */   396,  408,  506,  507,  416,  418,  408,  511,  512,  513,
- /*   230 */   514,  515,  516,  517,  518,  520,  521,  522,  523,  524,
- /*   240 */   428,  455,  519,  525,  473,  475,  534,
+ /*   150 */    65,   61,   61,  212,  161,  151,  206,  207,  213,  279,
+ /*   160 */   288,  302,  229,  214,   85,  216,  235,  236,  192,  217,
+ /*   170 */   335,  336,  274,  433,  309,  434,  435,  311,  441,  442,
+ /*   180 */   363,  323,  346,  355,  370,  375,  357,  381,  478,  383,
+ /*   190 */   384,  386,  382,  368,  385,  372,  388,  391,  392,  393,
+ /*   200 */   394,  395,  418,  484,  490,  491,  492,  493,  494,  424,
+ /*   210 */   486,  430,  487,  376,  377,  404,  502,  503,  411,  413,
+ /*   220 */   404,  506,  507,  508,  509,  510,  511,  512,  513,  514,
+ /*   230 */   515,  516,  517,  518,  422,  449,  505,  519,  468,  470,
+ /*   240 */   529,
 };
 #define YY_REDUCE_COUNT (132)
 #define YY_REDUCE_MIN   (-255)
-#define YY_REDUCE_MAX   (262)
+#define YY_REDUCE_MAX   (263)
 static const short yy_reduce_ofst[] = {
  /*     0 */  -200,  -53, -225, -238, -222, -151, -122, -194, -117,  -92,
- /*    10 */     7, -197, -190, -236, -163,  -36,  -34, -138, -150, -159,
- /*    20 */  -125,   -9, -152,  -78,   -1,   23,  -43, -255, -253, -223,
- /*    30 */  -144,  -85,  -60,    4,   12,   52,   58,   68,   82,   88,
- /*    40 */    90,   94,   96,   97,   98,  -22,   41,   80,  114,  118,
+ /*    10 */     7, -197, -190, -236, -163,  -34,  -32, -138, -150, -159,
+ /*    20 */  -125,  -36, -152,  -78,   23,   63,  -57, -255, -253, -223,
+ /*    30 */  -144,  -28,  -11,    4,    6,   12,   52,   55,   78,   90,
+ /*    40 */    91,   92,   93,   94,   96,  -22,  109,  116,  117,  118,
  /*    50 */   119,  121,  122,  123,  125,  124,  153,  156,  102,  103,
  /*    60 */   126,  160,  162,  164,  165,  166,  167,  168,  169,  170,
  /*    70 */   171,  172,  173,  174,  175,  176,  177,  179,  180,  181,
- /*    80 */   182,  185,  187,  188,  189,  190,  191,  192,  193,  195,
- /*    90 */   196,  197,  198,  199,  200,  201,  202,  203,  150,  205,
- /*   100 */   208,  134,  142,  209,  159,  210,  214,  211,  219,  221,
- /*   110 */   220,  224,  226,  223,  222,  225,  227,  228,  229,  230,
- /*   120 */   233,  231,  234,  236,  239,  240,  238,  241,  248,  251,
- /*   130 */   255,  247,  262,
+ /*    80 */   182,  185,  187,  188,  189,  190,  191,  193,  195,  196,
+ /*    90 */   197,  198,  199,  200,  201,  202,  203,  205,  145,  208,
+ /*   100 */   209,  134,  140,  210,  148,  211,  215,  219,  221,  220,
+ /*   110 */   224,  226,  222,  225,  223,  227,  228,  230,  231,  233,
+ /*   120 */   237,  234,  240,  238,  242,  243,  245,  239,  241,  244,
+ /*   130 */   253,  251,  263,
 };
 static const YYACTIONTYPE yy_default[] = {
- /*     0 */   623,  675,  664,  835,  835,  623,  623,  623,  623,  623,
- /*    10 */   623,  765,  641,  835,  623,  623,  623,  623,  623,  623,
- /*    20 */   623,  623,  623,  677,  677,  677,  760,  623,  623,  623,
- /*    30 */   623,  623,  623,  623,  623,  623,  623,  623,  623,  623,
- /*    40 */   623,  623,  623,  623,  623,  623,  623,  623,  623,  623,
- /*    50 */   623,  623,  623,  623,  623,  623,  623,  623,  784,  784,
- /*    60 */   758,  623,  623,  623,  623,  623,  623,  623,  623,  623,
- /*    70 */   623,  623,  623,  623,  623,  623,  623,  623,  623,  662,
- /*    80 */   623,  660,  623,  623,  623,  623,  623,  623,  623,  623,
- /*    90 */   623,  623,  623,  623,  623,  649,  623,  623,  623,  643,
- /*   100 */   643,  623,  623,  643,  791,  795,  789,  777,  785,  776,
- /*   110 */   772,  771,  799,  623,  643,  643,  672,  672,  643,  693,
- /*   120 */   691,  689,  681,  687,  683,  685,  679,  643,  670,  643,
- /*   130 */   670,  708,  721,  623,  800,  834,  790,  818,  817,  830,
- /*   140 */   824,  823,  623,  822,  821,  820,  819,  623,  623,  623,
- /*   150 */   623,  826,  825,  623,  623,  623,  623,  623,  623,  623,
- /*   160 */   623,  623,  802,  796,  792,  623,  623,  623,  623,  623,
- /*   170 */   623,  623,  623,  623,  623,  623,  623,  623,  623,  623,
- /*   180 */   623,  623,  623,  623,  623,  623,  623,  623,  757,  623,
- /*   190 */   623,  766,  623,  623,  623,  623,  623,  623,  786,  623,
- /*   200 */   778,  623,  623,  623,  623,  623,  623,  734,  623,  623,
- /*   210 */   623,  623,  623,  623,  623,  623,  623,  623,  623,  623,
- /*   220 */   623,  839,  623,  623,  623,  728,  837,  623,  623,  623,
- /*   230 */   623,  623,  623,  623,  623,  623,  623,  623,  623,  623,
- /*   240 */   696,  623,  647,  645,  623,  639,  623,
+ /*     0 */   617,  669,  658,  829,  829,  617,  617,  617,  617,  617,
+ /*    10 */   617,  759,  635,  829,  617,  617,  617,  617,  617,  617,
+ /*    20 */   617,  617,  617,  671,  671,  671,  754,  617,  617,  617,
+ /*    30 */   617,  617,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*    40 */   617,  617,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*    50 */   617,  617,  617,  617,  617,  617,  617,  617,  778,  778,
+ /*    60 */   752,  617,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*    70 */   617,  617,  617,  617,  617,  617,  617,  617,  617,  656,
+ /*    80 */   617,  654,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*    90 */   617,  617,  617,  617,  617,  643,  617,  617,  617,  637,
+ /*   100 */   637,  617,  617,  637,  785,  789,  783,  771,  779,  770,
+ /*   110 */   766,  765,  793,  617,  637,  637,  666,  666,  637,  687,
+ /*   120 */   685,  683,  675,  681,  677,  679,  673,  637,  664,  637,
+ /*   130 */   664,  702,  715,  617,  794,  828,  784,  812,  811,  824,
+ /*   140 */   818,  817,  617,  816,  815,  814,  813,  617,  617,  617,
+ /*   150 */   617,  820,  819,  617,  617,  617,  617,  617,  617,  617,
+ /*   160 */   617,  617,  796,  790,  786,  617,  617,  617,  617,  617,
+ /*   170 */   617,  617,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*   180 */   617,  617,  751,  617,  617,  760,  617,  617,  617,  617,
+ /*   190 */   617,  617,  780,  617,  772,  617,  617,  617,  617,  617,
+ /*   200 */   617,  728,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*   210 */   617,  617,  617,  617,  617,  833,  617,  617,  617,  722,
+ /*   220 */   831,  617,  617,  617,  617,  617,  617,  617,  617,  617,
+ /*   230 */   617,  617,  617,  617,  690,  617,  641,  639,  617,  633,
+ /*   240 */   617,
 };
 /********** End of lemon-generated parsing tables *****************************/
 
@@ -565,8 +564,8 @@ static const YYCODETYPE yyFallback[] = {
     0,  /*        SET => nothing */
     0,  /*       KILL => nothing */
     0,  /* CONNECTION => nothing */
-    0,  /*      COLON => nothing */
     0,  /*     STREAM => nothing */
+    0,  /*      COLON => nothing */
     1,  /*      ABORT => ID */
     1,  /*      AFTER => ID */
     1,  /*     ATTACH => ID */
@@ -856,8 +855,8 @@ static const char *const yyTokenName[] = {
   /*  130 */ "SET",
   /*  131 */ "KILL",
   /*  132 */ "CONNECTION",
-  /*  133 */ "COLON",
-  /*  134 */ "STREAM",
+  /*  133 */ "STREAM",
+  /*  134 */ "COLON",
   /*  135 */ "ABORT",
   /*  136 */ "AFTER",
   /*  137 */ "ATTACH",
@@ -1216,9 +1215,9 @@ static const char *const yyRuleName[] = {
  /* 214 */ "cmd ::= ALTER TABLE ids cpxName DROP TAG ids",
  /* 215 */ "cmd ::= ALTER TABLE ids cpxName CHANGE TAG ids ids",
  /* 216 */ "cmd ::= ALTER TABLE ids cpxName SET TAG ids EQ tagitem",
- /* 217 */ "cmd ::= KILL CONNECTION IPTOKEN COLON INTEGER",
- /* 218 */ "cmd ::= KILL STREAM IPTOKEN COLON INTEGER COLON INTEGER",
- /* 219 */ "cmd ::= KILL QUERY IPTOKEN COLON INTEGER COLON INTEGER",
+ /* 217 */ "cmd ::= KILL CONNECTION INTEGER",
+ /* 218 */ "cmd ::= KILL STREAM INTEGER COLON INTEGER",
+ /* 219 */ "cmd ::= KILL QUERY INTEGER COLON INTEGER",
 };
 #endif /* NDEBUG */
 
@@ -1893,9 +1892,9 @@ static const struct {
   {  207,   -7 }, /* (214) cmd ::= ALTER TABLE ids cpxName DROP TAG ids */
   {  207,   -8 }, /* (215) cmd ::= ALTER TABLE ids cpxName CHANGE TAG ids ids */
   {  207,   -9 }, /* (216) cmd ::= ALTER TABLE ids cpxName SET TAG ids EQ tagitem */
-  {  207,   -5 }, /* (217) cmd ::= KILL CONNECTION IPTOKEN COLON INTEGER */
-  {  207,   -7 }, /* (218) cmd ::= KILL STREAM IPTOKEN COLON INTEGER COLON INTEGER */
-  {  207,   -7 }, /* (219) cmd ::= KILL QUERY IPTOKEN COLON INTEGER COLON INTEGER */
+  {  207,   -3 }, /* (217) cmd ::= KILL CONNECTION INTEGER */
+  {  207,   -5 }, /* (218) cmd ::= KILL STREAM INTEGER COLON INTEGER */
+  {  207,   -5 }, /* (219) cmd ::= KILL QUERY INTEGER COLON INTEGER */
 };
 
 static void yy_accept(yyParser*);  /* Forward Declaration */
@@ -2734,14 +2733,14 @@ static void yy_reduce(
     setSQLInfo(pInfo, pAlterTable, NULL, TSDB_SQL_ALTER_TABLE);
 }
         break;
-      case 217: /* cmd ::= KILL CONNECTION IPTOKEN COLON INTEGER */
-{yymsp[-2].minor.yy0.n += (yymsp[-1].minor.yy0.n + yymsp[0].minor.yy0.n); setKillSQL(pInfo, TSDB_SQL_KILL_CONNECTION, &yymsp[-2].minor.yy0);}
+      case 217: /* cmd ::= KILL CONNECTION INTEGER */
+{setKillSQL(pInfo, TSDB_SQL_KILL_CONNECTION, &yymsp[0].minor.yy0);}
         break;
-      case 218: /* cmd ::= KILL STREAM IPTOKEN COLON INTEGER COLON INTEGER */
-{yymsp[-4].minor.yy0.n += (yymsp[-3].minor.yy0.n + yymsp[-2].minor.yy0.n + yymsp[-1].minor.yy0.n + yymsp[0].minor.yy0.n); setKillSQL(pInfo, TSDB_SQL_KILL_STREAM, &yymsp[-4].minor.yy0);}
+      case 218: /* cmd ::= KILL STREAM INTEGER COLON INTEGER */
+{yymsp[-2].minor.yy0.n += (yymsp[-1].minor.yy0.n + yymsp[0].minor.yy0.n); setKillSQL(pInfo, TSDB_SQL_KILL_STREAM, &yymsp[-2].minor.yy0);}
         break;
-      case 219: /* cmd ::= KILL QUERY IPTOKEN COLON INTEGER COLON INTEGER */
-{yymsp[-4].minor.yy0.n += (yymsp[-3].minor.yy0.n + yymsp[-2].minor.yy0.n + yymsp[-1].minor.yy0.n + yymsp[0].minor.yy0.n); setKillSQL(pInfo, TSDB_SQL_KILL_QUERY, &yymsp[-4].minor.yy0);}
+      case 219: /* cmd ::= KILL QUERY INTEGER COLON INTEGER */
+{yymsp[-2].minor.yy0.n += (yymsp[-1].minor.yy0.n + yymsp[0].minor.yy0.n); setKillSQL(pInfo, TSDB_SQL_KILL_QUERY, &yymsp[-2].minor.yy0);}
         break;
       default:
         break;

src/rpc/src/rpcCache.c

@@ -127,7 +127,7 @@ void rpcAddConnIntoCache(void *handle, void *data, char *fqdn, uint16_t port, in
 
   hash = rpcHashConn(pCache, fqdn, port, connType);
   pNode = (SConnHash *)taosMemPoolMalloc(pCache->connHashMemPool);
-  strcpy(pNode->fqdn, fqdn);
+  tstrncpy(pNode->fqdn, fqdn, sizeof(pNode->fqdn));
   pNode->port = port;
   pNode->connType = connType;
   pNode->data = data;

src/rpc/src/rpcMain.c

@@ -60,7 +60,7 @@ typedef struct {
 
   void     *idPool;   // handle to ID pool
   void     *tmrCtrl;  // handle to timer
-  void     *hash;     // handle returned by hash utility
+  SHashObj *hash;     // handle returned by hash utility
   void     *tcphandle;// returned handle from TCP initialization
   void     *udphandle;// returned handle from UDP initialization
   void     *pCache;   // connection cache
@@ -211,7 +211,7 @@ void *rpcOpen(const SRpcInit *pInit) {
   pRpc = (SRpcInfo *)calloc(1, sizeof(SRpcInfo));
   if (pRpc == NULL) return NULL;
 
-  if(pInit->label) strcpy(pRpc->label, pInit->label);
+  if(pInit->label) tstrncpy(pRpc->label, pInit->label, sizeof(pRpc->label));
   pRpc->connType = pInit->connType;
   pRpc->idleTime = pInit->idleTime;
   pRpc->numOfThreads = pInit->numOfThreads>TSDB_MAX_RPC_THREADS ? TSDB_MAX_RPC_THREADS:pInit->numOfThreads;
@@ -228,7 +228,7 @@ void *rpcOpen(const SRpcInit *pInit) {
   size_t size = sizeof(SRpcConn) * pRpc->sessions;
   pRpc->connList = (SRpcConn *)calloc(1, size);
   if (pRpc->connList == NULL) {
-    tError("%s failed to allocate memory for taos connections, size:%d", pRpc->label, size);
+    tError("%s failed to allocate memory for taos connections, size:%ld", pRpc->label, size);
     rpcClose(pRpc);
     return NULL;
   }
@@ -459,7 +459,7 @@ int rpcGetConnInfo(void *thandle, SRpcConnInfo *pInfo) {
   pInfo->clientPort = pConn->peerPort;
   // pInfo->serverIp = pConn->destIp;
 
-  strcpy(pInfo->user, pConn->user);
+  strncpy(pInfo->user, pConn->user, sizeof(pInfo->user));
   return 0;
 }
 
@@ -503,10 +503,10 @@ static SRpcConn *rpcOpenConn(SRpcInfo *pRpc, char *peerFqdn, uint16_t peerPort,
   pConn = rpcAllocateClientConn(pRpc); 
 
   if (pConn) { 
-    strcpy(pConn->peerFqdn, peerFqdn);
+    tstrncpy(pConn->peerFqdn, peerFqdn, sizeof(pConn->peerFqdn));
     pConn->peerIp = peerIp;
     pConn->peerPort = peerPort;
-    strcpy(pConn->user, pRpc->user);
+    tstrncpy(pConn->user, pRpc->user, sizeof(pConn->user));
     pConn->connType = connType;
 
     if (taosOpenConn[connType]) {
@@ -804,7 +804,7 @@ static SRpcConn *rpcProcessMsgHead(SRpcInfo *pRpc, SRecvInfo *pRecv) {
 
   pConn = rpcGetConnObj(pRpc, sid, pRecv);
   if (pConn == NULL) {
-    tTrace("%s %p, failed to get connection obj(%s)", pRpc->label, pHead->ahandle, tstrerror(terrno)); 
+    tTrace("%s %p, failed to get connection obj(%s)", pRpc->label, (void *)pHead->ahandle, tstrerror(terrno)); 
     return NULL;
   } else {
     if (rpcIsReq(pHead->msgType)) {

src/rpc/src/rpcTcp.c

@@ -73,7 +73,7 @@ void *taosInitTcpServer(uint32_t ip, uint16_t port, char *label, int numOfThread
   pServerObj = (SServerObj *)calloc(sizeof(SServerObj), 1);
   pServerObj->ip = ip;
   pServerObj->port = port;
-  strcpy(pServerObj->label, label);
+  tstrncpy(pServerObj->label, label, sizeof(pServerObj->label));
   pServerObj->numOfThreads = numOfThreads;
 
   pServerObj->pThreadObj = (SThreadObj *)calloc(sizeof(SThreadObj), numOfThreads);
@@ -87,7 +87,7 @@ void *taosInitTcpServer(uint32_t ip, uint16_t port, char *label, int numOfThread
   pThreadObj = pServerObj->pThreadObj;
   for (int i = 0; i < numOfThreads; ++i) {
     pThreadObj->processData = fp;
-    strcpy(pThreadObj->label, label);
+    tstrncpy(pThreadObj->label, label, sizeof(pThreadObj->label));
     pThreadObj->shandle = shandle;
 
     code = pthread_mutex_init(&(pThreadObj->mutex), NULL);
@@ -247,7 +247,7 @@ void *taosInitTcpClient(uint32_t ip, uint16_t port, char *label, int num, void *
 
   pThreadObj = (SThreadObj *)malloc(sizeof(SThreadObj));
   memset(pThreadObj, 0, sizeof(SThreadObj));
-  strcpy(pThreadObj->label, label);
+  tstrncpy(pThreadObj->label, label, sizeof(pThreadObj->label));
   pThreadObj->ip = ip;
   pThreadObj->shandle = shandle;
 

src/rpc/src/rpcUdp.c

@@ -72,7 +72,7 @@ void *taosInitUdpConnection(uint32_t ip, uint16_t port, char *label, int threads
   pSet->port = port;
   pSet->shandle = shandle;
   pSet->fp = fp;
-  strcpy(pSet->label, label);
+  tstrncpy(pSet->label, label, sizeof(pSet->label));
 
   uint16_t ownPort;
   for (int i = 0; i < threads; ++i) {
@@ -99,7 +99,7 @@ void *taosInitUdpConnection(uint32_t ip, uint16_t port, char *label, int threads
       pConn->localPort = (uint16_t)ntohs(sin.sin_port);
     }
 
-    strcpy(pConn->label, label);
+    tstrncpy(pConn->label, label, sizeof(pConn->label));
     pConn->shandle = shandle;
     pConn->processData = fp;
     pConn->index = i;

src/rpc/test/rclient.c

@@ -76,6 +76,7 @@ int main(int argc, char *argv[]) {
   int      numOfReqs = 0;
   int      appThreads = 1;
   char     serverIp[40] = "127.0.0.1";
+  char     secret[TSDB_KEY_LEN] = "mypassword";
   struct   timeval systemTime;
   int64_t  startTime, endTime;
   pthread_attr_t thattr;
@@ -97,7 +98,7 @@ int main(int argc, char *argv[]) {
   rpcInit.sessions     = 100;
   rpcInit.idleTime     = tsShellActivityTimer*1000;
   rpcInit.user         = "michael";
-  rpcInit.secret       = "mypassword";
+  rpcInit.secret       = secret;
   rpcInit.ckey         = "key";
   rpcInit.spi          = 1;
   rpcInit.connType     = TAOS_CONN_CLIENT;
@@ -106,7 +107,7 @@ int main(int argc, char *argv[]) {
     if (strcmp(argv[i], "-p")==0 && i < argc-1) {
       ipSet.port[0] = atoi(argv[++i]);
     } else if (strcmp(argv[i], "-i") ==0 && i < argc-1) {
-      tstrncpy(ipSet.fqdn[0], argv[++i], sizeof(ipSet.fqdn)); 
+      tstrncpy(ipSet.fqdn[0], argv[++i], sizeof(ipSet.fqdn[0])); 
     } else if (strcmp(argv[i], "-t")==0 && i < argc-1) {
       rpcInit.numOfThreads = atoi(argv[++i]);
     } else if (strcmp(argv[i], "-m")==0 && i < argc-1) {

src/rpc/test/rsclient.c

@@ -77,6 +77,7 @@ int main(int argc, char *argv[]) {
   int      numOfReqs = 0;
   int      appThreads = 1;
   char     serverIp[40] = "127.0.0.1";
+  char     secret[TSDB_KEY_LEN] = "mypassword";
   struct   timeval systemTime;
   int64_t  startTime, endTime;
   pthread_attr_t thattr;
@@ -98,7 +99,7 @@ int main(int argc, char *argv[]) {
   rpcInit.sessions     = 100;
   rpcInit.idleTime     = tsShellActivityTimer*1000;
   rpcInit.user         = "michael";
-  rpcInit.secret       = "mypassword";
+  rpcInit.secret       = secret;
   rpcInit.ckey         = "key";
   rpcInit.spi          = 1;
   rpcInit.connType     = TAOS_CONN_CLIENT;

src/tsdb/src/tsdbRWHelper.c

@@ -977,7 +977,8 @@ static int tsdbMergeDataWithBlock(SRWHelper *pHelper, int blkIdx, SDataCols *pDa
       // tdResetDataCols(pHelper->pDataCols[1]);
       while (true) {
         if (iter1 >= pHelper->pDataCols[0]->numOfRows && iter2 >= rows3) break;
-        tdMergeTwoDataCols(pHelper->pDataCols[1], pHelper->pDataCols[0], &iter1, pDataCols, &iter2, pHelper->config.maxRowsPerFileBlock * 4 / 5);
+        tdMergeTwoDataCols(pHelper->pDataCols[1], pHelper->pDataCols[0], &iter1, pHelper->pDataCols[0]->numOfRows,
+                           pDataCols, &iter2, rowsWritten, pHelper->config.maxRowsPerFileBlock * 4 / 5);
         ASSERT(pHelper->pDataCols[1]->numOfRows > 0);
         if (tsdbWriteBlockToFile(pHelper, &(pHelper->files.dataF), pHelper->pDataCols[1],
                                  pHelper->pDataCols[1]->numOfRows, &compBlock, false, true) < 0)
@@ -989,54 +990,6 @@ static int tsdbMergeDataWithBlock(SRWHelper *pHelper, int blkIdx, SDataCols *pDa
         }
         round++;
         blkIdx++;
-        // TODO: the blkIdx here is not correct
-
-        // if (iter1 >= pHelper->pDataCols[0]->numOfRows && iter2 >= rows3) {
-        //   if (pHelper->pDataCols[1]->numOfRows > 0) {
-        //     if (tsdbWriteBlockToFile(pHelper, &pHelper->files.dataF, pHelper->pDataCols[1],
-        //                              pHelper->pDataCols[1]->numOfRows, &compBlock, false, true) < 0)
-        //       goto _err;
-        //     // TODO: the blkIdx here is not correct
-        //     tsdbAddSubBlock(pHelper, &compBlock, blkIdx, pHelper->pDataCols[1]->numOfRows);
-        //   }
-        // }
-
-        // TSKEY key1 = iter1 >= pHelper->pDataCols[0]->numOfRows
-        //                  ? INT64_MAX
-        //                  : ((int64_t *)(pHelper->pDataCols[0]->cols[0].pData))[iter1];
-        // TSKEY key2 = iter2 >= rowsWritten ? INT64_MAX : ((int64_t *)(pDataCols->cols[0].pData))[iter2];
-
-        // if (key1 < key2) {
-        //   for (int i = 0; i < pDataCols->numOfCols; i++) {
-        //     SDataCol *pDataCol = pHelper->pDataCols[1]->cols + i;
-        //     memcpy(((char *)pDataCol->pData + TYPE_BYTES[pDataCol->type] * pHelper->pDataCols[1]->numOfRows),
-        //            ((char *)pHelper->pDataCols[0]->cols[i].pData + TYPE_BYTES[pDataCol->type] * iter1),
-        //            TYPE_BYTES[pDataCol->type]);
-        //   }
-        //   pHelper->pDataCols[1]->numOfRows++;
-        //   iter1++;
-        // } else if (key1 == key2) {
-        //   // TODO: think about duplicate key cases
-        //   ASSERT(false);
-        // } else {
-        //   for (int i = 0; i < pDataCols->numOfCols; i++) {
-        //     SDataCol *pDataCol = pHelper->pDataCols[1]->cols + i;
-        //     memcpy(((char *)pDataCol->pData + TYPE_BYTES[pDataCol->type] * pHelper->pDataCols[1]->numOfRows),
-        //            ((char *)pDataCols->cols[i].pData +
-        //             TYPE_BYTES[pDataCol->type] * iter2),
-        //            TYPE_BYTES[pDataCol->type]);
-        //   }
-        //   pHelper->pDataCols[1]->numOfRows++;
-        //   iter2++;
-        // }
-
-        // if (pHelper->pDataCols[0]->numOfRows >= pHelper->config.maxRowsPerFileBlock * 4 / 5) {
-        //   if (tsdbWriteBlockToFile(pHelper, &pHelper->files.dataF, pHelper->pDataCols[1], pHelper->pDataCols[1]->numOfRows, &compBlock, false, true) < 0) goto _err;
-        //   // TODO: blkIdx here is not correct, fix it
-        //   tsdbInsertSuperBlock(pHelper, &compBlock, blkIdx);
-
-        //   tdResetDataCols(pHelper->pDataCols[1]);
-        // }
       }
     }
   }

src/tsdb/src/tsdbRead.c

@@ -17,6 +17,7 @@
 #include "tulog.h"
 #include "talgo.h"
 #include "tutil.h"
+#include "ttime.h"
 #include "tcompare.h"
 #include "exception.h"
 
@@ -71,6 +72,8 @@ typedef struct STableCheckInfo {
   int32_t       compSize;
   int32_t       numOfBlocks;    // number of qualified data blocks not the original blocks
   SDataCols*    pDataCols;
+  
+  int32_t       chosen;         // indicate which iterator should move forward
   bool          initBuf;        // whether to initialize the in-memory skip list iterator or not
   SSkipListIterator* iter;      // mem buffer skip list iterator
   SSkipListIterator* iiter;     // imem buffer skip list iterator
@@ -79,8 +82,6 @@ typedef struct STableCheckInfo {
 typedef struct STableBlockInfo {
   SCompBlock*        compBlock;
   STableCheckInfo*   pTableCheckInfo;
-//  int32_t            blockIndex;
-//  int32_t            groupIdx; /* number of group is less than the total number of tables */
 } STableBlockInfo;
 
 typedef struct SBlockOrderSupporter {
@@ -120,7 +121,7 @@ static void changeQueryHandleForInterpQuery(TsdbQueryHandleT pHandle);
 static void doMergeTwoLevelData(STsdbQueryHandle* pQueryHandle, STableCheckInfo* pCheckInfo, SCompBlock* pBlock,
                                    SArray* sa);
 static int32_t binarySearchForKey(char* pValue, int num, TSKEY key, int order);
-static int tsdbReadRowsFromCache(SSkipListIterator* pIter, STable* pTable, TSKEY maxKey, int maxRowsToRead, TSKEY* skey, TSKEY* ekey,
+static int tsdbReadRowsFromCache(STableCheckInfo* pCheckInfo, TSKEY maxKey, int maxRowsToRead, TSKEY* skey, TSKEY* ekey,
                                  STsdbQueryHandle* pQueryHandle);
 
 static void tsdbInitDataBlockLoadInfo(SDataBlockLoadInfo* pBlockLoadInfo) {
@@ -134,7 +135,7 @@ static void tsdbInitCompBlockLoadInfo(SLoadCompBlockInfo* pCompBlockLoadInfo) {
   pCompBlockLoadInfo->fileId = -1;
 }
 
-TsdbQueryHandleT* tsdbQueryTables(TsdbRepoT* tsdb, STsdbQueryCond* pCond, STableGroupInfo* groupList) {
+TsdbQueryHandleT* tsdbQueryTables(TsdbRepoT* tsdb, STsdbQueryCond* pCond, STableGroupInfo* groupList, void* qinfo) {
   // todo 1. filter not exist table
   // todo 2. add the reference count for each table that is involved in query
 
@@ -147,6 +148,7 @@ TsdbQueryHandleT* tsdbQueryTables(TsdbRepoT* tsdb, STsdbQueryCond* pCond, STable
   pQueryHandle->cur.win     = TSWINDOW_INITIALIZER;
   pQueryHandle->checkFiles  = true;//ASCENDING_TRAVERSE(pQueryHandle->order);
   pQueryHandle->activeIndex = 0;   // current active table index
+  pQueryHandle->qinfo       = qinfo;
   pQueryHandle->outputCapacity = ((STsdbRepo*)tsdb)->config.maxRowsPerFileBlock;
   
   tsdbInitReadHelper(&pQueryHandle->rhelper, (STsdbRepo*) tsdb);
@@ -201,8 +203,8 @@ TsdbQueryHandleT* tsdbQueryTables(TsdbRepoT* tsdb, STsdbQueryCond* pCond, STable
   return (TsdbQueryHandleT) pQueryHandle;
 }
 
-TsdbQueryHandleT tsdbQueryLastRow(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STableGroupInfo *groupList) {
-  STsdbQueryHandle *pQueryHandle = (STsdbQueryHandle*) tsdbQueryTables(tsdb, pCond, groupList);
+TsdbQueryHandleT tsdbQueryLastRow(TsdbRepoT *tsdb, STsdbQueryCond *pCond, STableGroupInfo *groupList, void* qinfo) {
+  STsdbQueryHandle *pQueryHandle = (STsdbQueryHandle*) tsdbQueryTables(tsdb, pCond, groupList, qinfo);
   
   pQueryHandle->type = TSDB_QUERY_TYPE_LAST;
   pQueryHandle->order = TSDB_ORDER_DESC;
@@ -227,8 +229,8 @@ SArray* tsdbGetQueriedTableIdList(TsdbQueryHandleT *pHandle) {
   return res;
 }
 
-TsdbQueryHandleT tsdbQueryRowsInExternalWindow(TsdbRepoT *tsdb, STsdbQueryCond* pCond, STableGroupInfo *groupList) {
-  STsdbQueryHandle *pQueryHandle = (STsdbQueryHandle*) tsdbQueryTables(tsdb, pCond, groupList);
+TsdbQueryHandleT tsdbQueryRowsInExternalWindow(TsdbRepoT *tsdb, STsdbQueryCond* pCond, STableGroupInfo *groupList, void* qinfo) {
+  STsdbQueryHandle *pQueryHandle = (STsdbQueryHandle*) tsdbQueryTables(tsdb, pCond, groupList, qinfo);
   
   pQueryHandle->type = TSDB_QUERY_TYPE_EXTERNAL;
 //  pQueryHandle->outputCapacity = 2;  // only allowed two rows to be loaded
@@ -303,6 +305,83 @@ static bool initTableMemIterator(STsdbQueryHandle* pHandle, STableCheckInfo* pCh
   return true;
 }
 
+SDataRow getSDataRowInTableMem(STableCheckInfo* pCheckInfo) {
+  SDataRow rmem = NULL, rimem = NULL;
+  if (pCheckInfo->iter) {
+    SSkipListNode* node = tSkipListIterGet(pCheckInfo->iter);
+    if (node != NULL) {
+      rmem = SL_GET_NODE_DATA(node);
+    }
+  }
+  
+  if (pCheckInfo->iiter) {
+    SSkipListNode* node = tSkipListIterGet(pCheckInfo->iiter);
+    if (node != NULL) {
+      rimem = SL_GET_NODE_DATA(node);
+    }
+  }
+  
+  if (rmem != NULL && rimem != NULL) {
+    if (dataRowKey(rmem) < dataRowKey(rimem)) {
+      pCheckInfo->chosen = 0;
+      return rmem;
+    } else if (dataRowKey(rmem) == dataRowKey(rimem)) {
+      // data ts are duplicated, ignore the data in mem
+      tSkipListIterNext(pCheckInfo->iter);
+      pCheckInfo->chosen = 1;
+      return rimem;
+    } else {
+      pCheckInfo->chosen = 1;
+      return rimem;
+    }
+  }
+  
+  if (rmem != NULL) {
+    pCheckInfo->chosen = 0;
+    return rmem;
+  }
+  
+  if (rimem != NULL) {
+    pCheckInfo->chosen = 1;
+    return rimem;
+  }
+  
+  return NULL;
+}
+
+bool moveToNextRow(STableCheckInfo* pCheckInfo) {
+  bool hasNext = false;
+  if (pCheckInfo->chosen == 0) {
+    if (pCheckInfo->iter != NULL) {
+      hasNext = tSkipListIterNext(pCheckInfo->iter);
+    }
+    
+    if (hasNext) {
+      return hasNext;
+    }
+    
+    if (pCheckInfo->iiter != NULL) {
+      return tSkipListIterGet(pCheckInfo->iiter) != NULL;
+    }
+  } else {
+    if (pCheckInfo->chosen == 1) {
+      if (pCheckInfo->iiter != NULL) {
+        hasNext = tSkipListIterNext(pCheckInfo->iiter);
+      }
+  
+      if (hasNext) {
+        return hasNext;
+      }
+  
+      if (pCheckInfo->iter != NULL) {
+        return tSkipListIterGet(pCheckInfo->iter) != NULL;
+      }
+    }
+  }
+  
+  return hasNext;
+}
+
 static bool hasMoreDataInCache(STsdbQueryHandle* pHandle) {
   size_t size = taosArrayGetSize(pHandle->pTableCheckInfo);
   assert(pHandle->activeIndex < size && pHandle->activeIndex >= 0 && size >= 1);
@@ -312,31 +391,13 @@ static bool hasMoreDataInCache(STsdbQueryHandle* pHandle) {
 
   STable* pTable = pCheckInfo->pTableObj;
   assert(pTable != NULL);
-
-  // no data in cache, abort
-  if (pTable->mem == NULL && pTable->imem == NULL) {
-    return false;
-  }
-  
-  if (pCheckInfo->iter == NULL && pTable->mem) {
-    pCheckInfo->iter = tSkipListCreateIterFromVal(pTable->mem->pData, (const char*) &pCheckInfo->lastKey,
-        TSDB_DATA_TYPE_TIMESTAMP, pHandle->order);
-    
-    if (pCheckInfo->iter == NULL) {
-      return false;
-    }
-  
-    if (!tSkipListIterNext(pCheckInfo->iter)) {  // buffer is empty
-      return false;
-    }
-  }
   
-  SSkipListNode* node = tSkipListIterGet(pCheckInfo->iter);
-  if (node == NULL) {
+  initTableMemIterator(pHandle, pCheckInfo);
+  SDataRow row = getSDataRowInTableMem(pCheckInfo);
+  if (row == NULL) {
     return false;
   }
-
-  SDataRow row = SL_GET_NODE_DATA(node);
+  
   pCheckInfo->lastKey = dataRowKey(row);  // first timestamp in buffer
   tsdbTrace("%p uid:%" PRId64", tid:%d check data in buffer from skey:%" PRId64 ", order:%d, %p", pHandle,
       pCheckInfo->tableId.uid, pCheckInfo->tableId.tid, pCheckInfo->lastKey, pHandle->order, pHandle->qinfo);
@@ -349,7 +410,7 @@ static bool hasMoreDataInCache(STsdbQueryHandle* pHandle) {
   
   int32_t step = ASCENDING_TRAVERSE(pHandle->order)? 1:-1;
   STimeWindow* win = &pHandle->cur.win;
-  pHandle->cur.rows = tsdbReadRowsFromCache(pCheckInfo->iter, pCheckInfo->pTableObj, pHandle->window.ekey,
+  pHandle->cur.rows = tsdbReadRowsFromCache(pCheckInfo, pHandle->window.ekey,
                                     pHandle->outputCapacity, &win->skey, &win->ekey, pHandle);  // todo refactor API
   
   // update the last key value
@@ -382,7 +443,7 @@ static int32_t getFileIdFromKey(TSKEY key, int32_t daysPerFile, int32_t precisio
   return fid;
 }
 
-static int32_t binarySearchForBlockImpl(SCompBlock* pBlock, int32_t numOfBlocks, TSKEY skey, int32_t order) {
+static int32_t binarySearchForBlock(SCompBlock* pBlock, int32_t numOfBlocks, TSKEY skey, int32_t order) {
   int32_t firstSlot = 0;
   int32_t lastSlot = numOfBlocks - 1;
   
@@ -448,7 +509,7 @@ static int32_t getFileCompInfo(STsdbQueryHandle* pQueryHandle, int32_t* numOfBlo
       TSKEY e = MAX(pCheckInfo->lastKey, pQueryHandle->window.ekey);
       
       // discard the unqualified data block based on the query time window
-      int32_t start = binarySearchForBlockImpl(pCompInfo->blocks, compIndex->numOfBlocks, s, TSDB_ORDER_ASC);
+      int32_t start = binarySearchForBlock(pCompInfo->blocks, compIndex->numOfBlocks, s, TSDB_ORDER_ASC);
       int32_t end = start;
       
       if (s > pCompInfo->blocks[start].keyLast) {
@@ -522,7 +583,9 @@ static bool doLoadFileDataBlock(STsdbQueryHandle* pQueryHandle, SCompBlock* pBlo
 
   bool    blockLoaded = false;
   SArray* sa = getDefaultLoadColumns(pQueryHandle, true);
-
+  
+  int64_t st = taosGetTimestampUs();
+  
   if (pCheckInfo->pDataCols == NULL) {
     STsdbMeta* pMeta = tsdbGetMeta(pRepo);
     pCheckInfo->pDataCols = tdNewDataCols(pMeta->maxRowBytes, pMeta->maxCols, pRepo->config.maxRowsPerFileBlock);
@@ -540,9 +603,15 @@ static bool doLoadFileDataBlock(STsdbQueryHandle* pQueryHandle, SCompBlock* pBlo
     blockLoaded = true;
   }
 
+  SDataCols* pCols = pQueryHandle->rhelper.pDataCols[0];
+  assert(pCols->numOfRows != 0);
+  
   taosArrayDestroy(sa);
   tfree(data);
-
+  
+  int64_t et = taosGetTimestampUs() - st;
+  tsdbTrace("%p load file block into buffer, elapsed time:%"PRId64 " us", pQueryHandle, et);
+  
   return blockLoaded;
 }
 
@@ -600,7 +669,7 @@ static void handleDataMergeIfNeeded(STsdbQueryHandle* pQueryHandle, SCompBlock*
       // do not load file block into buffer
       int32_t step = ASCENDING_TRAVERSE(pQueryHandle->order) ? 1 : -1;
 
-      cur->rows = tsdbReadRowsFromCache(pCheckInfo->iter, pCheckInfo->pTableObj, binfo.window.skey - step,
+      cur->rows = tsdbReadRowsFromCache(pCheckInfo, binfo.window.skey - step,
                                         pQueryHandle->outputCapacity, &cur->win.skey, &cur->win.ekey, pQueryHandle);
       pQueryHandle->realNumOfRows = cur->rows;
 
@@ -649,7 +718,7 @@ static bool loadFileDataBlock(STsdbQueryHandle* pQueryHandle, SCompBlock* pBlock
       if (!doLoadFileDataBlock(pQueryHandle, pBlock, pCheckInfo)) {
         return false;
       }
-  
+      
       SDataCols* pTSCol = pQueryHandle->rhelper.pDataCols[0];
       assert(pTSCol->cols->type == TSDB_DATA_TYPE_TIMESTAMP && pTSCol->numOfRows == pBlock->numOfRows);
       
@@ -1155,7 +1224,7 @@ static int32_t createDataBlocksInfo(STsdbQueryHandle* pQueryHandle, int32_t numO
   *numOfAllocBlocks = numOfBlocks;
 
   int32_t numOfTables = taosArrayGetSize(pQueryHandle->pTableCheckInfo);
-
+  
   SBlockOrderSupporter sup = {0};
   sup.numOfTables = numOfTables;
   sup.numOfBlocksPerTable = calloc(1, sizeof(int32_t) * numOfTables);
@@ -1192,15 +1261,26 @@ static int32_t createDataBlocksInfo(STsdbQueryHandle* pQueryHandle, int32_t numO
 
       pBlockInfo->compBlock = &pBlock[k];
       pBlockInfo->pTableCheckInfo = pTableCheck;
-      //      pBlockInfo->groupIdx = pTableCheckInfo[j]->groupIdx;     // set the group index
-      //      pBlockInfo->blockIndex = pTableCheckInfo[j]->start + k;    // set the block index in original table
       cnt++;
     }
 
     numOfQualTables++;
   }
 
-  tsdbTrace("%p create data blocks info struct completed, %d blocks in %d tables", pQueryHandle, cnt, numOfQualTables);
+  assert(numOfBlocks == cnt);
+  
+  // since there is only one table qualified, blocks are not sorted
+  if (numOfQualTables == 1) {
+    memcpy(pQueryHandle->pDataBlockInfo, sup.pDataBlockInfo[0], sizeof(STableBlockInfo) * numOfBlocks);
+    cleanBlockOrderSupporter(&sup, numOfQualTables);
+  
+    tsdbTrace("%p create data blocks info struct completed for 1 table, %d blocks not sorted %p ", pQueryHandle, cnt,
+        pQueryHandle->qinfo);
+    return TSDB_CODE_SUCCESS;
+  }
+  
+  tsdbTrace("%p create data blocks info struct completed, %d blocks in %d tables %p", pQueryHandle, cnt,
+      numOfQualTables, pQueryHandle->qinfo);
 
   assert(cnt <= numOfBlocks && numOfQualTables <= numOfTables);  // the pTableQueryInfo[j]->numOfBlocks may be 0
   sup.numOfTables = numOfQualTables;
@@ -1257,8 +1337,8 @@ static bool getDataBlocksInFilesImpl(STsdbQueryHandle* pQueryHandle) {
       break;
     }
     
-    tsdbTrace("%p %d blocks found in file for %d table(s), fid:%d", pQueryHandle, numOfBlocks,
-           numOfTables, pQueryHandle->pFileGroup->fileId);
+    tsdbTrace("%p %d blocks found in file for %d table(s), fid:%d, %p", pQueryHandle, numOfBlocks,
+           numOfTables, pQueryHandle->pFileGroup->fileId, pQueryHandle->qinfo);
     
     assert(numOfBlocks >= 0);
     if (numOfBlocks == 0) {
@@ -1565,19 +1645,22 @@ static void changeQueryHandleForInterpQuery(TsdbQueryHandleT pHandle) {
   pQueryHandle->window = (STimeWindow) {info.lastKey, TSKEY_INITIAL_VAL};
 }
 
-static int tsdbReadRowsFromCache(SSkipListIterator* pIter, STable* pTable, TSKEY maxKey, int maxRowsToRead, TSKEY* skey, TSKEY* ekey,
+static int tsdbReadRowsFromCache(STableCheckInfo* pCheckInfo, TSKEY maxKey, int maxRowsToRead, TSKEY* skey, TSKEY* ekey,
                                  STsdbQueryHandle* pQueryHandle) {
   int     numOfRows = 0;
   int32_t numOfCols = taosArrayGetSize(pQueryHandle->pColumns);
   *skey = TSKEY_INITIAL_VAL;
 
+  int64_t st = taosGetTimestampUs();
+  STSchema* pSchema = tsdbGetTableSchema(tsdbGetMeta(pQueryHandle->pTsdb), pCheckInfo->pTableObj);
+  int32_t numOfTableCols = schemaNCols(pSchema);
+  
   do {
-    SSkipListNode* node = tSkipListIterGet(pIter);
-    if (node == NULL) {
+    SDataRow row = getSDataRowInTableMem(pCheckInfo);
+    if (row == NULL) {
       break;
     }
 
-    SDataRow row = SL_GET_NODE_DATA(node);
     TSKEY key = dataRowKey(row);
     
     if ((key > maxKey && ASCENDING_TRAVERSE(pQueryHandle->order)) ||
@@ -1590,49 +1673,69 @@ static int tsdbReadRowsFromCache(SSkipListIterator* pIter, STable* pTable, TSKEY
     }
 
     if (*skey == INT64_MIN) {
-      *skey = dataRowKey(row);
+      *skey = key;
     }
 
-    *ekey = dataRowKey(row);
-
-    int32_t offset = -1;
+    *ekey = key;
     char* pData = NULL;
-  
-    STSchema* pSchema = tsdbGetTableSchema(tsdbGetMeta(pQueryHandle->pTsdb), pTable);
-    int32_t numOfTableCols = schemaNCols(pSchema);
     
-    for (int32_t i = 0; i < numOfCols; ++i) {
+    int32_t i = 0, j = 0;
+    while(i < numOfCols && j < numOfTableCols) {
       SColumnInfoData* pColInfo = taosArrayGet(pQueryHandle->pColumns, i);
-      
+      if (pSchema->columns[j].colId < pColInfo->info.colId) {
+        j++;
+        continue;
+      }
+  
       if (ASCENDING_TRAVERSE(pQueryHandle->order)) {
         pData = pColInfo->pData + numOfRows * pColInfo->info.bytes;
       } else {
         pData = pColInfo->pData + (maxRowsToRead - numOfRows - 1) * pColInfo->info.bytes;
       }
       
-      for(int32_t j = 0; j < numOfTableCols; ++j) {
-        if (pColInfo->info.colId == pSchema->columns[j].colId) {
-          offset = pSchema->columns[j].offset;
-          break;
+      if (pSchema->columns[j].colId == pColInfo->info.colId) {
+        void* value = tdGetRowDataOfCol(row, pColInfo->info.type, TD_DATA_ROW_HEAD_SIZE + pSchema->columns[j].offset);
+        if (pColInfo->info.type == TSDB_DATA_TYPE_BINARY || pColInfo->info.type == TSDB_DATA_TYPE_NCHAR) {
+          memcpy(pData, value, varDataTLen(value));
+        } else {
+          memcpy(pData, value, pColInfo->info.bytes);
         }
+        
+        j++;
+        i++;
+      } else { // pColInfo->info.colId < pSchema->columns[j].colId, it is a NULL data
+        if (pColInfo->info.type == TSDB_DATA_TYPE_BINARY || pColInfo->info.type == TSDB_DATA_TYPE_NCHAR) {
+          setVardataNull(pData, pColInfo->info.type);
+        } else {
+          setNull(pData, pColInfo->info.type, pColInfo->info.bytes);
+        }
+        i++;
       }
-      
-      assert(offset != -1); // todo handle error
-      void *value = tdGetRowDataOfCol(row, pColInfo->info.type, TD_DATA_ROW_HEAD_SIZE + offset);
-      
+    }
+    
+    while (i < numOfCols) { // the remain columns are all null data
+      SColumnInfoData* pColInfo = taosArrayGet(pQueryHandle->pColumns, i);
+      if (ASCENDING_TRAVERSE(pQueryHandle->order)) {
+        pData = pColInfo->pData + numOfRows * pColInfo->info.bytes;
+      } else {
+        pData = pColInfo->pData + (maxRowsToRead - numOfRows - 1) * pColInfo->info.bytes;
+      }
+  
       if (pColInfo->info.type == TSDB_DATA_TYPE_BINARY || pColInfo->info.type == TSDB_DATA_TYPE_NCHAR) {
-        memcpy(pData, value, varDataTLen(value));
+        setVardataNull(pData, pColInfo->info.type);
       } else {
-        memcpy(pData, value, pColInfo->info.bytes);
+        setNull(pData, pColInfo->info.type, pColInfo->info.bytes);
       }
+      
+      i++;
     }
-
+    
     if (++numOfRows >= maxRowsToRead) {
-      tSkipListIterNext(pIter);
+      moveToNextRow(pCheckInfo);
       break;
     }
     
-  } while(tSkipListIterNext(pIter));
+  } while(moveToNextRow(pCheckInfo));
 
   assert(numOfRows <= maxRowsToRead);
   
@@ -1646,6 +1749,10 @@ static int tsdbReadRowsFromCache(SSkipListIterator* pIter, STable* pTable, TSKEY
     }
   }
   
+  int64_t elapsedTime = taosGetTimestampUs() - st;
+  tsdbTrace("%p build data block from cache completed, elapsed time:%"PRId64" us, numOfRows:%d, numOfCols:%d", pQueryHandle,
+            elapsedTime, numOfRows, numOfCols);
+  
   return numOfRows;
 }
 

src/util/inc/tcache.h

@@ -111,6 +111,15 @@ void *taosCachePut(SCacheObj *pCacheObj, const char *key, const void *pData, siz
  */
 void *taosCacheAcquireByName(SCacheObj *pCacheObj, const char *key);
 
+/**
+ * update the expire time of data in cache 
+ * @param pCacheObj     cache object
+ * @param key           key
+ * @param expireTime    new expire time of data
+ * @return
+ */ 
+void* taosCacheUpdateExpireTimeByName(SCacheObj *pCacheObj, const char *key, uint64_t expireTime);
+
 /**
  * Add one reference count for the exist data, and assign this data for a new owner.
  * The new owner needs to invoke the taosCacheRelease when it does not need this data anymore.

src/util/inc/tutil.h

@@ -153,6 +153,8 @@ bool taosMbsToUcs4(char *mbs, size_t mbs_len, char *ucs4, int32_t ucs4_max_len,
 
 int tasoUcs4Compare(void* f1_ucs4, void *f2_ucs4, int bytes);
 
+void taosRandStr(char* str, int32_t size);
+
 int32_t taosUcs4ToMbs(void *ucs4, int32_t ucs4_max_len, char *mbs);
 
 bool taosValidateEncodec(const char *encodec);

src/util/src/tarray.c

@@ -92,12 +92,14 @@ void* taosArrayGet(const SArray* pArray, size_t index) {
 }
 
 void* taosArrayGetP(const SArray* pArray, size_t index) {
-  void* ret = taosArrayGet(pArray, index);
-  if (ret == NULL) {
+  assert(index < pArray->size);
+  
+  void* d = TARRAY_GET_ELEM(pArray, index);
+  if (d == NULL) {
     return NULL;
   }
   
-  return *(void**)ret;
+  return *(void**)d;
 }
 
 size_t taosArrayGetSize(const SArray* pArray) { return pArray->size; }

src/util/src/tcache.c

@@ -488,6 +488,35 @@ void *taosCacheAcquireByName(SCacheObj *pCacheObj, const char *key) {
   return (ptNode != NULL) ? (*ptNode)->data : NULL;
 }
 
+void* taosCacheUpdateExpireTimeByName(SCacheObj *pCacheObj, const char *key, uint64_t expireTime) {
+  if (pCacheObj == NULL || taosHashGetSize(pCacheObj->pHashTable) == 0) {
+    return NULL;
+  }
+  
+  uint32_t keyLen = (uint32_t)strlen(key);
+  
+  __cache_rd_lock(pCacheObj);
+  
+  SCacheDataNode **ptNode = (SCacheDataNode **)taosHashGet(pCacheObj->pHashTable, key, keyLen);
+  if (ptNode != NULL) {
+     T_REF_INC(*ptNode);
+    (*ptNode)->expiredTime = expireTime;
+  }
+  
+  __cache_unlock(pCacheObj);
+  
+  if (ptNode != NULL) {
+    atomic_add_fetch_32(&pCacheObj->statistics.hitCount, 1);
+    uTrace("key:%s expireTime is updated in cache, %p refcnt:%d", key, (*ptNode), T_REF_VAL_GET(*ptNode));
+  } else {
+    atomic_add_fetch_32(&pCacheObj->statistics.missCount, 1);
+    uTrace("key:%s not in cache, retrieved failed", key);
+  }
+  
+  atomic_add_fetch_32(&pCacheObj->statistics.totalAccess, 1);
+  return (ptNode != NULL) ? (*ptNode)->data : NULL;
+}
+
 void *taosCacheAcquireByData(SCacheObj *pCacheObj, void *data) {
   if (pCacheObj == NULL || data == NULL) return NULL;
   

src/util/src/tsocket.c

@@ -19,6 +19,7 @@
 #include "tutil.h"
 
 int taosGetFqdn(char *fqdn) {
+  int  code = 0;
   char hostname[1024];
   hostname[1023] = '\0';
   gethostname(hostname, 1023);
@@ -27,13 +28,15 @@ int taosGetFqdn(char *fqdn) {
   h = gethostbyname(hostname);
   if (h != NULL) {
     strcpy(fqdn, h->h_name);
-    return 0;
   } else {
-    uError("failed to get host name");
-    return -1;
+    uError("failed to get host name(%s)", strerror(errno));
+    code = -1;
   }
 
-  free(h);
+  // to do: free the resources
+  // free(h);
+
+  return code;
 }
 
 uint32_t taosGetIpFromFqdn(const char *fqdn) {
@@ -47,7 +50,7 @@ uint32_t ip2uint(const char *const ip_addr) {
   char ip_addr_cpy[20];
   char ip[5];
 
-  strcpy(ip_addr_cpy, ip_addr);
+  tstrncpy(ip_addr_cpy, ip_addr, sizeof(ip_addr_cpy));
 
   char *s_start, *s_end;
   s_start = ip_addr_cpy;
@@ -206,7 +209,7 @@ int taosOpenUdpSocket(uint32_t ip, uint16_t port) {
   int                reuse, nocheck;
   int                bufSize = 8192000;
 
-  uTrace("open udp socket:%s:%hu", ip, port);
+  uTrace("open udp socket:0x%x:%hu", ip, port);
 
   memset((char *)&localAddr, 0, sizeof(localAddr));
   localAddr.sin_family = AF_INET;
@@ -257,7 +260,7 @@ int taosOpenUdpSocket(uint32_t ip, uint16_t port) {
 
   /* bind socket to local address */
   if (bind(sockFd, (struct sockaddr *)&localAddr, sizeof(localAddr)) < 0) {
-    uError("failed to bind udp socket: %d (%s), %s:%hu", errno, strerror(errno), ip, port);
+    uError("failed to bind udp socket: %d (%s), 0x%x:%hu", errno, strerror(errno), ip, port);
     taosCloseSocket(sockFd);
     return -1;
   }
@@ -363,7 +366,7 @@ int taosOpenTcpServerSocket(uint32_t ip, uint16_t port) {
   int                sockFd;
   int                reuse;
 
-  uTrace("open tcp server socket:%s:%hu", ip, port);
+  uTrace("open tcp server socket:0x%x:%hu", ip, port);
 
   bzero((char *)&serverAdd, sizeof(serverAdd));
   serverAdd.sin_family = AF_INET;

src/util/src/tutil.c

@@ -27,8 +27,6 @@
 #include "tulog.h"
 #include "taoserror.h"
 
-int32_t tmpFileSerialNum = 0;
-
 int32_t strdequote(char *z) {
   if (z == NULL) {
     return 0;
@@ -433,12 +431,24 @@ void getTmpfilePath(const char *fileNamePrefix, char *dstPath) {
 #else
   char *tmpDir = "/tmp/";
 #endif
-  int64_t ts = taosGetTimestampUs();
+  
   strcpy(tmpPath, tmpDir);
   strcat(tmpPath, tdengineTmpFileNamePrefix);
   strcat(tmpPath, fileNamePrefix);
-  strcat(tmpPath, "-%d-%"PRIu64"-%u-%"PRIu64);
-  snprintf(dstPath, PATH_MAX, tmpPath, getpid(), taosGetPthreadId(), atomic_add_fetch_32(&tmpFileSerialNum, 1), ts);
+  strcat(tmpPath, "-%d-%s");
+  
+  char rand[8] = {0};
+  taosRandStr(rand, tListLen(rand) - 1);
+  snprintf(dstPath, PATH_MAX, tmpPath, getpid(), rand);
+}
+
+void taosRandStr(char* str, int32_t size) {
+  const char* set = "abcdefghijklmnopqrstuvwxyz0123456789-_.";
+  int32_t len = 39;
+  
+  for(int32_t i = 0; i < size; ++i) {
+    str[i] = set[rand()%len];
+  }
 }
 
 int tasoUcs4Compare(void* f1_ucs4, void *f2_ucs4, int bytes) {

src/wal/src/walMain.c

@@ -68,11 +68,19 @@ void *walOpen(const char *path, const SWalCfg *pCfg) {
   pWal->num = 0;
   pWal->level = pCfg->walLevel;
   pWal->keep = pCfg->keep;
-  strcpy(pWal->path, path);
+  tstrncpy(pWal->path, path, sizeof(pWal->path));
   pthread_mutex_init(&pWal->mutex, NULL);
 
-  if (access(path, F_OK) != 0) mkdir(path, 0755);
-  
+  if (access(path, F_OK) != 0) {
+    if (mkdir(path, 0755) != 0) {
+      terrno = TAOS_SYSTEM_ERROR(errno);
+      wError("wal:%s, failed to create directory(%s)", path, strerror(errno));
+      pthread_mutex_destroy(&pWal->mutex);
+      free(pWal);
+      pWal = NULL;
+    }
+  }
+     
   if (pCfg->keep == 1) return pWal;
 
   if (walHandleExistingFiles(path) == 0) 
@@ -80,7 +88,7 @@ void *walOpen(const char *path, const SWalCfg *pCfg) {
 
   if (pWal->fd <0) {
     terrno = TAOS_SYSTEM_ERROR(errno);
-    wError("wal:%s, failed to open", path);
+    wError("wal:%s, failed to open(%s)", path, strerror(errno));
     pthread_mutex_destroy(&pWal->mutex);
     free(pWal);
     pWal = NULL;
@@ -119,7 +127,8 @@ void walClose(void *handle) {
 int walRenew(void *handle) {
   if (handle == NULL) return 0;
   SWal *pWal = handle;
-  int   code = 0;
+
+  terrno = 0;
 
   pthread_mutex_lock(&pWal->mutex);
 
@@ -135,8 +144,8 @@ int walRenew(void *handle) {
   pWal->fd = open(pWal->name, O_WRONLY | O_CREAT, S_IRWXU | S_IRWXG | S_IRWXO);
 
   if (pWal->fd < 0) {
-    wError("wal:%d, failed to open(%s)", pWal->name, strerror(errno));
-    code = -1;
+    wError("wal:%s, failed to open(%s)", pWal->name, strerror(errno));
+    terrno = TAOS_SYSTEM_ERROR(errno);
   } else {
     wTrace("wal:%s, it is created", pWal->name);
 
@@ -156,14 +165,15 @@ int walRenew(void *handle) {
   
   pthread_mutex_unlock(&pWal->mutex);
 
-  return code;
+  return terrno;
 }
 
 int walWrite(void *handle, SWalHead *pHead) {
   SWal *pWal = handle;
-  int   code = 0;
   if (pWal == NULL) return -1;
 
+  terrno = 0;
+
   // no wal  
   if (pWal->level == TAOS_WAL_NOLOG) return 0;
   if (pHead->version <= pWal->version) return 0;
@@ -174,12 +184,12 @@ int walWrite(void *handle, SWalHead *pHead) {
 
   if(write(pWal->fd, pHead, contLen) != contLen) {
     wError("wal:%s, failed to write(%s)", pWal->name, strerror(errno));
-    code = -1;
+    terrno = TAOS_SYSTEM_ERROR(errno);
   } else {
     pWal->version = pHead->version;
   }
 
-  return code;
+  return terrno;
 }
 
 void walFsync(void *handle) {
@@ -196,11 +206,11 @@ void walFsync(void *handle) {
 
 int walRestore(void *handle, void *pVnode, int (*writeFp)(void *, void *, int)) {
   SWal    *pWal = handle;
-  int      code = 0;
   struct   dirent *ent;
   int      count = 0;
   uint32_t maxId = 0, minId = -1, index =0;
 
+  terrno = 0;
   int   plen = strlen(walPrefix);
   char  opath[TSDB_FILENAME_LEN+5];
    
@@ -224,30 +234,30 @@ int walRestore(void *handle, void *pVnode, int (*writeFp)(void *, void *, int))
   closedir(dir);
 
   if (count == 0) {
-    if (pWal->keep) code = walRenew(pWal);
-    return code;
+    if (pWal->keep) terrno = walRenew(pWal);
+    return terrno;
   }
 
   if ( count != (maxId-minId+1) ) {
     wError("wal:%s, messed up, count:%d max:%d min:%d", opath, count, maxId, minId);
-    code = -1;
+    terrno = TAOS_SYSTEM_ERROR(TSDB_CODE_APP_ERROR);
   } else {
     wTrace("wal:%s, %d files will be restored", opath, count);
 
     for (index = minId; index<=maxId; ++index) {
       sprintf(pWal->name, "%s/%s%d", opath, walPrefix, index);
-      code = walRestoreWalFile(pWal, pVnode, writeFp);
-      if (code < 0) break;
+      terrno = walRestoreWalFile(pWal, pVnode, writeFp);
+      if (terrno < 0) break;
     }
   }
 
-  if (code == 0) {
+  if (terrno == 0) {
     if (pWal->keep == 0) {
-      code = walRemoveWalFiles(opath);
-      if (code == 0) {
+      terrno = walRemoveWalFiles(opath);
+      if (terrno == 0) {
         if (remove(opath) < 0) {
           wError("wal:%s, failed to remove directory(%s)", opath, strerror(errno));
-          code = -1;
+          terrno = TAOS_SYSTEM_ERROR(errno);
         }
       }
     } else { 
@@ -258,12 +268,12 @@ int walRestore(void *handle, void *pVnode, int (*writeFp)(void *, void *, int))
       pWal->fd = open(pWal->name, O_WRONLY | O_CREAT | O_APPEND, S_IRWXU | S_IRWXG | S_IRWXO);
       if (pWal->fd < 0) {
         wError("wal:%s, failed to open file(%s)", pWal->name, strerror(errno));
-        code = -1;
+        terrno = TAOS_SYSTEM_ERROR(errno);
       }
     }
   }
 
-  return code;
+  return terrno;
 }
 
 int walGetWalFile(void *handle, char *name, uint32_t *index) {
@@ -292,40 +302,47 @@ int walGetWalFile(void *handle, char *name, uint32_t *index) {
 }  
 
 static int walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp) {
-  int   code = 0;
   char *name = pWal->name;
 
+  terrno = 0;
   char *buffer = malloc(1024000);  // size for one record
-  if (buffer == NULL) return -1;
+  if (buffer == NULL) {
+    terrno = TAOS_SYSTEM_ERROR(errno);   
+    return terrno;
+  }
 
   SWalHead *pHead = (SWalHead *)buffer;
 
   int fd = open(name, O_RDONLY);
   if (fd < 0) {
     wError("wal:%s, failed to open for restore(%s)", name, strerror(errno));
+    terrno = TAOS_SYSTEM_ERROR(errno);
     free(buffer);
-    return -1;
+    return terrno;
   }
 
   wTrace("wal:%s, start to restore", name);
 
   while (1) {
     int ret = read(fd, pHead, sizeof(SWalHead));
-    if ( ret == 0) { code = 0; break;}  
+    if ( ret == 0)  break;  
 
     if (ret != sizeof(SWalHead)) {
       wWarn("wal:%s, failed to read head, skip, ret:%d(%s)", name, ret, strerror(errno));
+      terrno = TAOS_SYSTEM_ERROR(errno);
       break;
     }
 
     if (!taosCheckChecksumWhole((uint8_t *)pHead, sizeof(SWalHead))) {
       wWarn("wal:%s, cksum is messed up, skip the rest of file", name);
+      terrno = TAOS_SYSTEM_ERROR(errno);
       break;
     } 
 
     ret = read(fd, pHead->cont, pHead->len);
     if ( ret != pHead->len) {
       wWarn("wal:%s, failed to read body, skip, len:%d ret:%d", name, pHead->len, ret);
+      terrno = TAOS_SYSTEM_ERROR(errno);
       break;
     }
 
@@ -336,11 +353,10 @@ static int walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp) {
   close(fd);
   free(buffer);
 
-  return code;
+  return terrno;
 }
 
 int walHandleExistingFiles(const char *path) {
-  int    code = 0;
   char   oname[TSDB_FILENAME_LEN * 3];
   char   nname[TSDB_FILENAME_LEN * 3];
   char   opath[TSDB_FILENAME_LEN];
@@ -350,6 +366,7 @@ int walHandleExistingFiles(const char *path) {
   struct dirent *ent;
   DIR   *dir = opendir(path);
   int    plen = strlen(walPrefix);
+  terrno = 0;
 
   if (access(opath, F_OK) == 0) {
     // old directory is there, it means restore process is not finished
@@ -360,13 +377,19 @@ int walHandleExistingFiles(const char *path) {
     int count = 0;
     while ((ent = readdir(dir))!= NULL) {  
       if ( strncmp(ent->d_name, walPrefix, plen) == 0) {
-        if (access(opath, F_OK) != 0) mkdir(opath, 0755);
-
         sprintf(oname, "%s/%s", path, ent->d_name);
         sprintf(nname, "%s/old/%s", path, ent->d_name);
+        if (access(opath, F_OK) != 0) {
+          if (mkdir(opath, 0755) != 0) {
+            wError("wal:%s, failed to create directory:%s(%s)", oname, opath, strerror(errno));
+            terrno = TAOS_SYSTEM_ERROR(errno);
+            break;
+          } 
+        }
+
         if (rename(oname, nname) < 0) {
           wError("wal:%s, failed to move to new:%s", oname, nname);
-          code = -1;
+          terrno = TAOS_SYSTEM_ERROR(errno);
           break;
         } 
 
@@ -378,14 +401,14 @@ int walHandleExistingFiles(const char *path) {
   }
   
   closedir(dir);
-  return code;
+  return terrno;
 }
 
 static int walRemoveWalFiles(const char *path) {
   int    plen = strlen(walPrefix);
   char   name[TSDB_FILENAME_LEN * 3];
-  int    code = 0;
-
+ 
+  terrno = 0;
   if (access(path, F_OK) != 0) return 0;
 
   struct dirent *ent;
@@ -396,13 +419,13 @@ static int walRemoveWalFiles(const char *path) {
       sprintf(name, "%s/%s", path, ent->d_name);
       if (remove(name) <0) {
         wError("wal:%s, failed to remove(%s)", name, strerror(errno));
-        code = -1; break;
+        terrno = TAOS_SYSTEM_ERROR(errno);
       }
     }
   } 
 
   closedir(dir);
 
-  return code;
+  return terrno;
 }
 

tests/pytest/query/filter.py

@@ -87,7 +87,7 @@ class TDTestCase:
 
         # <> for timestamp type
         tdSql.query("select * from db.st where ts <> '2020-05-13 10:00:00.002'")
-        #tdSql.checkRows(4)
+        # tdSql.checkRows(4)
 
         # <> for numeric type
         tdSql.query("select * from db.st where tagtype <> 2")
@@ -96,7 +96,7 @@ class TDTestCase:
         # <> for nchar type
         tdSql.query("select * from db.st where name <> 'first'")
         tdSql.checkRows(4)
-        
+
         # % for nchar type
         tdSql.query("select * from db.st where name like 'fi%'")
         tdSql.checkRows(2)

tests/pytest/query/filterCombo.py

@@ -42,14 +42,17 @@ class TDTestCase:
             ('2020-05-13 10:00:00.005', 3, 'third')""")
 
         # query with filter condition A OR condition B
-        tdSql.query("select * from db.st where ts > '2020-05-13 10:00:00.002' AND tagtype < 2")
+        tdSql.query(
+            "select * from db.st where ts > '2020-05-13 10:00:00.002' AND tagtype < 2")
         tdSql.checkRows(1)
 
         # query with filter condition A OR condition B, error expected
-        tdSql.error("select * from db.st where ts > '2020-05-13 10:00:00.002' OR tagtype < 2")
+        tdSql.error(
+            "select * from db.st where ts > '2020-05-13 10:00:00.002' OR tagtype < 2")
 
         # illegal condition
-        tdSql.error("select * from db.st where ts != '2020-05-13 10:00:00.002' OR tagtype < 2")        
+        tdSql.error(
+            "select * from db.st where ts != '2020-05-13 10:00:00.002' OR tagtype < 2")
         tdSql.error("select * from db.st where tagtype <> 1 OR tagtype < 2")
 
     def stop(self):

tests/pytest/query/queryError.py

@@ -41,18 +41,17 @@ class TDTestCase:
             ('2020-05-13 10:00:00.002', 3, 'third') dev_002 VALUES('2020-05-13 10:00:00.003', 1, 'first'), ('2020-05-13 10:00:00.004', 2, 'second'),
             ('2020-05-13 10:00:00.005', 3, 'third')""")
 
-        
         # query first .. as ..
-        tdSql.error("select first(*) as one from st")    
+        tdSql.error("select first(*) as one from st")
 
         # query last .. as ..
-        tdSql.error("select last(*) as latest from st")        
+        tdSql.error("select last(*) as latest from st")
 
         # query last row .. as ..
-        tdSql.error("select last_row as latest from st") 
+        tdSql.error("select last_row as latest from st")
 
         # query distinct on normal colnum
-        tdSql.error("select distinct tagtype from st")     
+        tdSql.error("select distinct tagtype from st")
 
         # query .. order by non-time field
         tdSql.error("select * from st order by name")

tests/pytest/query/queryNormal.py

@@ -28,17 +28,20 @@ class TDTestCase:
 
         print("==============step1")
 
-        tdSql.execute("create table stb1 (ts timestamp, c1 int, c2 float) tags(t1 int, t2 binary(10), t3 nchar(10))")
-        tdSql.execute("insert into tb1 using stb1 tags(1,'tb1', '表1') values ('2020-04-18 15:00:00.000', 1, 0.1), ('2020-04-18 15:00:01.000', 2, 0.1)")
-        tdSql.execute("insert into tb2 using stb1 tags(2,'tb2', '表2') values ('2020-04-18 15:00:02.000', 3, 2.1), ('2020-04-18 15:00:03.000', 4, 2.2)")        
-        
-        # inner join --- bug 
+        tdSql.execute(
+            "create table stb1 (ts timestamp, c1 int, c2 float) tags(t1 int, t2 binary(10), t3 nchar(10))")
+        tdSql.execute(
+            "insert into tb1 using stb1 tags(1,'tb1', '表1') values ('2020-04-18 15:00:00.000', 1, 0.1), ('2020-04-18 15:00:01.000', 2, 0.1)")
+        tdSql.execute(
+            "insert into tb2 using stb1 tags(2,'tb2', '表2') values ('2020-04-18 15:00:02.000', 3, 2.1), ('2020-04-18 15:00:03.000', 4, 2.2)")
+
+        # inner join --- bug
         tdSql.query("select * from tb1 a, tb2 b where a.ts = b.ts")
         tdSql.checkRows(1)
 
         # join 3 tables -- bug exists
         tdSql.query("select stb_t.ts, stb_t.dscrption, stb_t.temperature, stb_p.id, stb_p.dscrption, stb_p.pressure,stb_v.velocity from stb_p, stb_t, stb_v where stb_p.ts=stb_t.ts and stb_p.ts=stb_v.ts and stb_p.id = stb_t.id")
-        
+
         # query show stable
         tdSql.query("show stables")
         tdSql.checkRows(1)
@@ -51,15 +54,15 @@ class TDTestCase:
         tdSql.query("select count(*) from stb1")
         tdSql.checkData(0, 0, 4)
 
-        # query first 
+        # query first
         tdSql.query("select first(*) from stb1")
         tdSql.checkData(0, 1, 1)
-        
-        # query last 
+
+        # query last
         tdSql.query("select last(*) from stb1")
         tdSql.checkData(0, 1, 4)
 
-        # query last_row 
+        # query last_row
         tdSql.query("select last_row(*) from stb1")
         tdSql.checkData(0, 1, 4)
 
@@ -70,7 +73,7 @@ class TDTestCase:
         # query first ... as
         tdSql.query("select first(*) as begin from stb1")
         tdSql.checkData(0, 1, 1)
-        
+
         # query last ... as
         tdSql.query("select last(*) as end from stb1")
         tdSql.checkData(0, 1, 4)
@@ -93,7 +96,7 @@ class TDTestCase:
 
         # query ... alias for table ---- bug
         tdSql.query("select t.ts from tb1 t")
-        tdSql.checkRows(2)       
+        tdSql.checkRows(2)
 
         # query ... tbname
         tdSql.query("select tbname from stb1")

tests/pytest/random-test/random-test-multi-threading-3.py

@@ -111,7 +111,6 @@ class Test (threading.Thread):
                 last_tb)
             written = written + 1
 
-
     def drop_stable(self):
         tdLog.info("drop_stable")
         global last_stb
@@ -152,7 +151,6 @@ class Test (threading.Thread):
                 last_tb = ""
                 written = 0
 
-
     def query_data_from_stable(self):
         tdLog.info("query_data_from_stable")
         global last_stb
@@ -164,7 +162,6 @@ class Test (threading.Thread):
             tdLog.info("will query data from super table")
             tdSql.execute('select * from %s' % last_stb)
 
-
     def reset_query_cache(self):
         tdLog.info("reset_query_cache")
         global last_tb
@@ -232,7 +229,7 @@ class Test (threading.Thread):
                 self.threadLock.acquire()
                 tdLog.notice("first thread")
                 randDataOp = random.randint(1, 3)
-                dataOp.get(randDataOp , lambda: "ERROR")()
+                dataOp.get(randDataOp, lambda: "ERROR")()
                 self.threadLock.release()
 
         elif (self.threadId == 2):

tests/pytest/random-test/random-test-multi-threading.py

@@ -111,7 +111,6 @@ class Test (threading.Thread):
                 last_tb)
             written = written + 1
 
-
     def drop_stable(self):
         tdLog.info("drop_stable")
         global last_stb
@@ -154,7 +153,6 @@ class Test (threading.Thread):
                 last_tb = ""
                 written = 0
 
-
     def query_data_from_stable(self):
         tdLog.info("query_data_from_stable")
         global last_stb
@@ -166,7 +164,6 @@ class Test (threading.Thread):
             tdLog.info("will query data from super table")
             tdSql.execute('select * from %s' % last_stb)
 
-
     def reset_query_cache(self):
         tdLog.info("reset_query_cache")
         global last_tb
@@ -230,7 +227,7 @@ class Test (threading.Thread):
                 self.threadLock.acquire()
                 tdLog.notice("first thread")
                 randDataOp = random.randint(1, 3)
-                dataOp.get(randDataOp , lambda: "ERROR")()
+                dataOp.get(randDataOp, lambda: "ERROR")()
                 self.threadLock.release()
 
         elif (self.threadId == 2):

tests/pytest/random-test/random-test.py

@@ -112,7 +112,6 @@ class Test:
             tdSql.execute('drop table %s' % self.last_stb)
             self.last_stb = ""
 
-
     def query_data_from_stable(self):
         tdLog.info("query_data_from_stable")
         if (self.last_stb == ""):
@@ -122,20 +121,21 @@ class Test:
             tdLog.info("will query data from super table")
             tdSql.execute('select * from %s' % self.last_stb)
 
-
     def restart_database(self):
         tdLog.info("restart_databae")
         tdDnodes.stop(1)
         tdDnodes.start(1)
         tdLog.sleep(5)
 
-
     def force_restart_database(self):
         tdLog.info("force_restart_database")
         tdDnodes.forcestop(1)
         tdDnodes.start(1)
         tdLog.sleep(5)
         tdSql.prepare()
+        self.last_tb = ""
+        self.last_stb = ""
+        self.written = 0
 
     def drop_table(self):
         tdLog.info("drop_table")
@@ -159,6 +159,9 @@ class Test:
         tdDnodes.start(1)
         tdLog.sleep(5)
         tdSql.prepare()
+        self.last_tb = ""
+        self.last_stb = ""
+        self.written = 0
 
     def delete_datafiles(self):
         tdLog.info("delete_datafiles")
@@ -173,6 +176,9 @@ class Test:
         tdDnodes.start(1)
         tdLog.sleep(10)
         tdSql.prepare()
+        self.last_tb = ""
+        self.last_stb = ""
+        self.written = 0
 
 
 class TDTestCase:

tests/pytest/valgrind-test.sh

 #!/bin/bash
 # insert
-python3 ./test.py -g -f insert/basic.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/int.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/float.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/bigint.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/bool.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/double.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/smallint.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/tinyint.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/binary.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/date.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/nchar.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f insert/multi.py
-python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/basic.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/int.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/float.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/bigint.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/bool.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/double.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/smallint.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/tinyint.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/binary.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/date.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/nchar.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f insert/multi.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
 
 # table
-python3 ./test.py -g -f table/column_name.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f table/column_num.py
-python3 ./test.py -g -s && sleep 1
-python3 ./test.py -g -f table/db_table.py
-python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f table/column_name.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f table/column_num.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f table/db_table.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
 
 # import
-python3 ./test.py -g -f import_merge/importDataLastSub.py
-python3 ./test.py -g -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py -g -f import_merge/importDataLastSub.py
+PYTHONMALLOC=malloc python3 ./test.py -g -s && sleep 1
 
 #tag 
-python3 ./test.py $1 -f tag_lite/filter.py
-python3 ./test.py $1 -s && sleep 1
+PYTHONMALLOC=malloc python3 ./test.py $1 -f tag_lite/filter.py
+PYTHONMALLOC=malloc python3 ./test.py $1 -s && sleep 1

tests/script/unique/arbitrator/dn2_mn1_cache_file_sync.sim

@@ -2,11 +2,11 @@
 # step 1: start dnode1
 # step 2: start dnode2 and dnode3, and all added into cluster (Suppose dnode2 is master-vnode)
 # step 3: create db, table, insert data, and Falling disc into file (control only one file, e.g. 1841)
-# step 4: insert old data(now-15d) and new data(now+15d), control data rows  in order to save in cache, not falling disc 
+# step 4: insert old data(now-20d) and new data(now-40d), control data rows  in order to save in cache, not falling disc 
 # step 5: stop dnode2, so date rows falling disc, generate two new files 1840, 1842 in dnode2
-# step 6: insert two data rows: now-16d, now+16d
+# step 6: insert two data rows: now-21d, now-41d
 # step 7: restart dnode2, waiting sync end
-# expect: in dnode2, the files 1840 and 1842 will be removed
+# expect: in dnode2, the files 1837 and 1839 will be removed
 
 system sh/stop_dnodes.sh
 system sh/deploy.sh -n dnode1 -i 1
@@ -14,10 +14,10 @@ system sh/deploy.sh -n dnode2 -i 2
 system sh/deploy.sh -n dnode3 -i 3
 system sh/deploy.sh -n dnode4 -i 4
 
-system sh/cfg.sh -n dnode1 -c numOfMPeers -v 1
-system sh/cfg.sh -n dnode2 -c numOfMPeers -v 1
-system sh/cfg.sh -n dnode3 -c numOfMPeers -v 1
-system sh/cfg.sh -n dnode4 -c numOfMPeers -v 1
+system sh/cfg.sh -n dnode1 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode4 -c numOfMnodes -v 1
 
 system sh/cfg.sh -n dnode1 -c walLevel -v 2
 system sh/cfg.sh -n dnode2 -c walLevel -v 2
@@ -39,11 +39,6 @@ system sh/cfg.sh -n dnode2 -c alternativeRole -v 2
 system sh/cfg.sh -n dnode3 -c alternativeRole -v 2
 system sh/cfg.sh -n dnode4 -c alternativeRole -v 2
 
-system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 4
-system sh/cfg.sh -n dnode2 -c numOfTotalVnodes -v 4
-system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 4
-system sh/cfg.sh -n dnode4 -c numOfTotalVnodes -v 4
-
 system sh/cfg.sh -n dnode1 -c maxtablesPerVnode -v 4
 system sh/cfg.sh -n dnode2 -c maxtablesPerVnode -v 4
 system sh/cfg.sh -n dnode3 -c maxtablesPerVnode -v 4
@@ -64,16 +59,16 @@ sql connect
 
 print ============== step2: start dnode2/dnode3 and add into cluster , then create database with replica 2, and create table, insert data
 system sh/exec.sh -n dnode2 -s start
-#system sh/exec.sh -n dnode3 -s start
+system sh/exec.sh -n dnode3 -s start
 sql create dnode $hostname2
-#sql create dnode $hostname3
+sql create dnode $hostname3
 sleep 3000
 
 $totalTableNum = 1
 $sleepTimer = 3000
 
 $db = db
-sql create database $db replica 1 cache 1
+sql create database $db replica 2 cache 1
 sql use $db
 
 # create table , insert data
@@ -82,7 +77,7 @@ sql create table $stb (ts timestamp, c1 double) tags(t1 int)
 $rowNum = 130000
 $tblNum = $totalTableNum
 $totalRows = 0
-#$tsStart = 1420041600000
+$tsStart = 1577808000000  # 2020-01-01 00:00:00.000
 
 # insert over 2M data in order to falling disc, generate one file
 $i = 0
@@ -102,19 +97,24 @@ while $i < $tblNum
 endw 
 
 sql select count(*) from $stb
-sleep 1000
-print data00 $data00
-if $data00 != $totalRows then
-	return -1
+print rows:$rows data00:$data00
+if $rows != 1 then
+  return -1
+endi
+
+if $data00 == 0 then
+  return -1
 endi
 
+$totalRows = $data00
+
 print ============== step3: insert old data(now-15d) and new data(now+15d), control data rows  in order to save in cache, not falling disc
-sql insert into $tb values ( now - 15d , -15 )
-sql insert into $tb values ( now + 15d ,  15 )
+sql insert into $tb values ( now - 20d , -20 )
+sql insert into $tb values ( now - 40d , -40 )
 $totalRows = $totalRows + 2
 
 print ============== step4: stop dnode2, so date rows falling disc, generate two new files in dnode2
-system sh/exec.sh -n dnode2 -s stop
+system sh/exec.sh -n dnode2 -s stop -x SIGINT
 sleep $sleepTimer
 
 wait_dnode2_offline:
@@ -153,48 +153,14 @@ if $data00 != $totalRows then
 endi
 
 print ============== step5: insert two data rows: now-16d, now+16d, 
-sql insert into $tb values ( now - 16d , -16 )
-sql insert into $tb values ( now + 16d ,  16 )
+sql insert into $tb values ( now - 21d , -21 )
+sql insert into $tb values ( now - 41d , -41 )
 $totalRows = $totalRows + 2
 
-return 1
-
-
-
-print ============== step5: restart dnode2, waiting sync end
-system sh/exec.sh -n dnode2 -s start
-sleep 3000
-
-wait_dnode2_ready:
-sql show dnodes
-if $rows != 3 then
-  sleep 2000
-  goto wait_dnode2_ready
-endi
-print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
-print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
-print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
-#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
-#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
-#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
-#$dnode1Status = $data4_1  
-$dnode2Status = $data4_2
-$dnode3Status = $data4_3  
-#$dnode4Status = $data4_4 
-#$dnode5Status = $data4_5
-
-if $dnode2Status != ready then
-  sleep 2000
-  goto wait_dnode2_ready
-endi
-
-sleep $sleepTimer
-# check using select
 sql select count(*) from $stb
 print data00 $data00
 if $data00 != $totalRows then
 	return -1
 endi
 
-print ============== step6: in dnode2, the files 1840 and 1842 will be removed
-# how check in script ???
+print ============== step6: please check there should be 3 file in sim/dnode2/data/vnode/vnode2/tsdb/data/, and 1 file sim/dnode3/data/vnode/vnode2/tsdb/data/
\ No newline at end of file

tests/script/unique/arbitrator/dn2_mn1_cache_file_sync_second.sim

+# Test case describe: dnode1 is only mnode, dnode2/dnode3 are only vnode
+# step 1: start dnode1
+# step 2: start dnode2 and dnode3, and all added into cluster (Suppose dnode2 is master-vnode)
+# step 3: create db, table, insert data, and Falling disc into file (control only one file, e.g. 1841)
+# step 4: insert old data(now-20d) and new data(now-40d), control data rows  in order to save in cache, not falling disc 
+# step 5: stop dnode2, so date rows falling disc, generate two new files 1840, 1842 in dnode2
+# step 6: insert two data rows: now-21d, now-41d
+# step 7: restart dnode2, waiting sync end
+# expect: in dnode2, the files 1837 and 1839 will be removed
+
+sql connect
+sleep 3000
+
+print ============== step7: restart dnode2, waiting sync end
+system sh/exec.sh -n dnode2 -s start
+sleep 3000
+
+wait_dnode2_ready:
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode2_ready
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode2_ready
+endi
+
+sleep 1000
+# check using select
+$db = db
+$stb = stb
+sql use $db
+sql select count(*) from $stb
+print data00 $data00, should equal to dn2_mn1_cache_file_sync.sim output
+#if $data00 != $totalRows then
+#	return -1
+#endi
+
+print ============== step8: please check there should be 1 file in sim/dnode2/data/vnode/vnode2/tsdb/data/, and 1 file sim/dnode3/data/vnode/vnode2/tsdb/data/

tests/script/unique/arbitrator/dn3_mn1_replica2_wal1_AddDelDnode.sim

@@ -23,28 +23,39 @@ system sh/cfg.sh -n dnode3 -c balanceInterval -v 10
 system sh/cfg.sh -n dnode4 -c balanceInterval -v 10
 system sh/cfg.sh -n dnode5 -c balanceInterval -v 10
 
-system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 200
+system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 4
 system sh/cfg.sh -n dnode2 -c numOfTotalVnodes -v 4
 system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 4
 system sh/cfg.sh -n dnode4 -c numOfTotalVnodes -v 4
 system sh/cfg.sh -n dnode5 -c numOfTotalVnodes -v 4
 
+system sh/cfg.sh -n dnode1 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode2 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode3 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode4 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode5 -c maxtablesPerVnode -v 4
+
 system sh/cfg.sh -n dnode1 -c mnodeEqualVnodeNum -v 200
 #system sh/cfg.sh -n dnode2 -c mnodeEqualVnodeNum -v 200
 #system sh/cfg.sh -n dnode3 -c mnodeEqualVnodeNum -v 200
 #system sh/cfg.sh -n dnode4 -c mnodeEqualVnodeNum -v 200
 
+system sh/cfg.sh -n dnode1 -c alternativeRole -v 1
+system sh/cfg.sh -n dnode2 -c alternativeRole -v 2
+system sh/cfg.sh -n dnode3 -c alternativeRole -v 2
+system sh/cfg.sh -n dnode4 -c alternativeRole -v 2
+
 system sh/cfg.sh -n dnode1 -c arbitrator -v $arbitrator
 system sh/cfg.sh -n dnode2 -c arbitrator -v $arbitrator
 system sh/cfg.sh -n dnode3 -c arbitrator -v $arbitrator
 system sh/cfg.sh -n dnode4 -c arbitrator -v $arbitrator
 system sh/cfg.sh -n dnode5 -c arbitrator -v $arbitrator
 
-system sh/cfg.sh -n dnode1 -c offlineThreshold -v 20
-system sh/cfg.sh -n dnode2 -c offlineThreshold -v 20
-system sh/cfg.sh -n dnode3 -c offlineThreshold -v 20
-system sh/cfg.sh -n dnode4 -c offlineThreshold -v 20
-system sh/cfg.sh -n dnode5 -c offlineThreshold -v 20
+system sh/cfg.sh -n dnode1 -c offlineThreshold -v 10
+system sh/cfg.sh -n dnode2 -c offlineThreshold -v 10
+system sh/cfg.sh -n dnode3 -c offlineThreshold -v 10
+system sh/cfg.sh -n dnode4 -c offlineThreshold -v 10
+system sh/cfg.sh -n dnode5 -c offlineThreshold -v 10
 
 system sh/cfg.sh -n dnode1 -c enableCoreFile -v 1
 system sh/cfg.sh -n dnode2 -c enableCoreFile -v 1
@@ -63,9 +74,9 @@ sql connect
 print ============== step2: start dnode2/dnode3 and add into cluster, then create database, create table , and insert data
 system sh/exec.sh -n dnode2 -s start
 system sh/exec.sh -n dnode3 -s start
-sleep 1000
 sql create dnode $hostname2
 sql create dnode $hostname3
+sleep 3000
 
 $rowNum = 100
 $tblNum = 16
@@ -151,7 +162,7 @@ endi
 #  return -1
 #endi
 
-sleep 30000
+sleep 15000
 
 wait_drop:
 sql show dnodes
@@ -174,10 +185,69 @@ endi
 if $dnode2Status != ready then
   return -1
 endi
+if $dnode3Status != null then
+  return -1
+endi
 if $dnode4Status != ready then
   return -1
 endi
 
+print ============== step4-1: restart dnode3, adn add into cluster
+system rm -rf ../../sim/dnode3
+sleep 3000
+
+system sh/deploy.sh -n dnode3 -i 3
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode3 -c walLevel -v 1
+system sh/cfg.sh -n dnode3 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode3 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode3 -c offlineThreshold -v 10
+system sh/cfg.sh -n dnode3 -c enableCoreFile -v 1
+
+system sh/exec.sh -n dnode3 -s start
+sql create dnode $hostname3
+sleep 3000
+
+wait_dnode3_ready:
+sql show dnodes
+print rows: $rows
+if $rows != 4 then
+  sleep 3000
+  goto wait_dnode3_ready
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4  
+$dnode5Status = $data4_5
+
+if $dnode1Status != ready then
+  return -1
+endi
+if $dnode2Status != ready then
+  return -1
+endi
+if $dnode3Status != null then
+  return -1
+endi
+if $dnode4Status != ready then
+  return -1
+endi
+if $dnode5Status != ready then
+  return -1
+endi
+
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
 
 print ============== step5: start dnode5 and add into cluster , drop database
 sql drop database $db
@@ -187,7 +257,7 @@ sql create dnode $hostname5
 sleep 3000
 wait_dnode5:
 sql show dnodes
-if $rows != 4 then
+if $rows != 5 then
   sleep 3000
   goto wait_dnode5
 endi
@@ -196,10 +266,12 @@ print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
 print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
 print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
 print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
 $dnode1Status = $data4_1  
 $dnode2Status = $data4_2
 $dnode4Status = $data4_4  
-$dnode5Status = $data4_5
+$dnode5Status = $data4_5 
+$dnode6Status = $data4_6
 
 if $dnode1Status != ready then
   return -1
@@ -207,17 +279,22 @@ endi
 if $dnode2Status != ready then
   return -1
 endi
+if $dnode3Status != null then
+  return -1
+endi
 if $dnode4Status != ready then
   return -1
 endi
 if $dnode5Status != ready then
   return -1
 endi
-
+if $dnode6Status != ready then
+  return -1
+endi
 
 print ============== step6: create database and table until not free vnodes 
 $rowNum = 100
-$tblNum = 24
+$tblNum = 32
 $totalRows = 0
 $tsStart = 1420041600000
 
@@ -259,8 +336,9 @@ if $data00 != $totalRows then
 	return -1
 endi
 
-print ============== step7: drop dnode3, and system should prompt cannot drop dnodes 
-sql_error drop dnode $hostname3
-print ============== step8: add one new table, and system should prompt 'need more dnode' 
+print ============== step7: drop dnode $hostname5, system should prompt "DB error: no enough dnodes" 
+sql_error drop dnode $hostname5
+print error: $error 
+print ============== step8: create table tb_more using $stb tags( 1000 ), system should prompt 'DB error: no enough dnodes' 
 sql_error create table tb_more using $stb tags( 1000 )
-
+print error: $error 

tests/script/unique/arbitrator/dn3_mn1_replica_change.sim

@@ -89,11 +89,8 @@ if $data00 != $totalRows then
 	return -1
 endi
 
-
-
-
-print ============== step2-1: start dnode2 for falling disc, then restart dnode2, and check rows
-system sh/exec.sh -n dnode2 -s stop
+print ============== step2-1: stop dnode2 for falling disc, then restart dnode2, and check rows
+system sh/exec.sh -n dnode2 -s stop -x SIGINT
 sleep $sleepTimer
 wait_dnode2_offline_0:
 sql show dnodes
@@ -151,10 +148,6 @@ if $data00 != $totalRows then
 endi
 
 
-
-
-
-
 print ============== step3: start dnode3 and add into cluster , then alter replica from 1 to 2, and waiting sync
 system sh/exec.sh -n dnode3 -s start
 sql create dnode $hostname3

tests/script/unique/arbitrator/dn3_mn1_vnode_change.sim

+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+system sh/deploy.sh -n dnode2 -i 2
+system sh/deploy.sh -n dnode3 -i 3
+system sh/deploy.sh -n dnode4 -i 4
+
+system sh/cfg.sh -n dnode1 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 1
+
+system sh/cfg.sh -n dnode1 -c walLevel -v 2
+system sh/cfg.sh -n dnode2 -c walLevel -v 2
+system sh/cfg.sh -n dnode3 -c walLevel -v 2
+system sh/cfg.sh -n dnode4 -c walLevel -v 2
+
+system sh/cfg.sh -n dnode1 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode2 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode3 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode4 -c balanceInterval -v 10
+
+system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode2 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode4 -c numOfTotalVnodes -v 4
+
+system sh/cfg.sh -n dnode1 -c alternativeRole -v 1
+system sh/cfg.sh -n dnode2 -c alternativeRole -v 2
+system sh/cfg.sh -n dnode3 -c alternativeRole -v 2
+system sh/cfg.sh -n dnode4 -c alternativeRole -v 2
+
+system sh/cfg.sh -n dnode1 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode2 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode3 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode4 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode5 -c maxtablesPerVnode -v 4
+
+system sh/cfg.sh -n dnode1 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode2 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode3 -c arbitrator -v $arbitrator
+
+print ============== step0: start tarbitrator
+system sh/exec_tarbitrator.sh -s start
+
+print ============== step1: start dnode1, only deploy mnode
+system sh/exec.sh -n dnode1 -s start
+sleep 3000
+sql connect
+
+print ============== step2: start dnode2/dnode3/dnode4 and add into cluster , then create database with replica 3, and create table, insert data
+system sh/exec.sh -n dnode2 -s start
+system sh/exec.sh -n dnode3 -s start
+system sh/exec.sh -n dnode4 -s start
+sql create dnode $hostname2
+sql create dnode $hostname3
+sql create dnode $hostname4
+sleep 3000
+
+$totalTableNum = 100
+$sleepTimer = 3000
+
+$db = db
+sql create database $db replica 3 maxTables $totalTableNum
+sql use $db
+
+# create table , insert data
+$stb = stb
+sql create table $stb (ts timestamp, c1 int) tags(t1 int)
+$rowNum = 100
+$tblNum = $totalTableNum
+$totalRows = 0
+$tsStart = 1420041600000
+
+$i = 0
+while $i < $tblNum
+  $tb = tb . $i
+  sql create table $tb using $stb tags( $i )
+ 
+  $x = 0
+  while $x < $rowNum
+    $ts = $tsStart + $x
+    sql insert into $tb values ( $ts + 0a , $x ) ( $ts + 1a , $x ) ( $ts + 2a , $x ) ( $ts + 3a , $x ) ( $ts + 4a , $x ) ( $ts + 5a , $x ) ( $ts + 6a , $x ) ( $ts + 7a , $x ) ( $ts + 8a , $x ) ( $ts + 9a , $x ) ( $ts + 10a , $x ) ( $ts + 11a , $x ) ( $ts + 12a , $x ) ( $ts + 13a , $x ) ( $ts + 14a , $x ) ( $ts + 15a , $x ) ( $ts + 16a , $x ) ( $ts + 17a , $x ) ( $ts + 18a , $x ) ( $ts + 19a , $x ) ( $ts + 20a , $x ) ( $ts + 21a , $x ) ( $ts + 22a , $x ) ( $ts + 23a , $x ) ( $ts + 24a , $x ) ( $ts + 25a , $x ) ( $ts + 26a , $x ) ( $ts + 27a , $x ) ( $ts + 28a , $x ) ( $ts + 29a , $x ) ( $ts + 30a , $x ) ( $ts + 31a , $x ) ( $ts + 32a , $x ) ( $ts + 33a , $x ) ( $ts + 34a , $x ) ( $ts + 25a , $x ) ( $ts + 26a , $x ) ( $ts + 27a , $x ) ( $ts + 28a , $x ) ( $ts + 29a , $x ) ( $ts + 30a , $x ) ( $ts + 31a , $x ) ( $ts + 32a , $x ) ( $ts + 33a , $x ) ( $ts + 34a , $x ) ( $ts + 35a , $x ) ( $ts + 36a , $x ) ( $ts + 37a , $x ) ( $ts + 38a , $x ) ( $ts + 39a , $x ) ( $ts + 40a , $x ) ( $ts + 41a , $x ) ( $ts + 42a , $x ) ( $ts + 43a , $x ) ( $ts + 44a , $x ) ( $ts + 45a , $x ) ( $ts + 46a , $x ) ( $ts + 47a , $x ) ( $ts + 48a , $x ) ( $ts + 49a , $x ) ( $ts + 50a , $x ) ( $ts + 51a , $x ) ( $ts + 52a , $x ) ( $ts + 53a , $x ) ( $ts + 54a , $x ) ( $ts + 55a , $x ) ( $ts + 56a , $x ) ( $ts + 57a , $x ) ( $ts + 58a , $x ) ( $ts + 59a , $x ) 
+    $x = $x + 60
+  endw 
+  $totalRows = $totalRows + $x
+  print       info: inserted $x rows into $tb and totalRows: $totalRows
+  $i = $i + 1
+endw 
+
+sql select count(*) from $stb
+sleep 1000
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+
+print ============== step3: stop dnode2, and remove its vnodeX subdirector 
+system sh/exec.sh -n dnode2 -s stop -x SIGINT
+sleep $sleepTimer
+wait_dnode2_offline_0:
+sql show dnodes
+if $rows != 4 then
+  sleep 2000
+  goto wait_dnode2_offline_0
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+#$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode2Status != offline then
+  sleep 2000
+  goto wait_dnode2_offline_0
+endi
+
+system rm -rf ../../../sim/dnode2/data/vnode/*
+sleep 1000
+
+print ============== step4: restart dnode2, waiting sync end 
+system sh/exec.sh -n dnode2 -s start
+sleep $sleepTimer
+wait_dnode2_reready:
+sql show dnodes
+if $rows != 4 then
+  sleep 2000
+  goto wait_dnode2_reready
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+#$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode2_reready
+endi
+
+print ============== step5: stop dnode3/dnode4, and check rows
+system sh/exec.sh -n dnode3 -s stop
+system sh/exec.sh -n dnode4 -s stop
+sleep $sleepTimer
+
+wait_dnode34_offline:
+sql show dnodes
+if $rows != 4 then
+  sleep 2000
+  goto wait_dnode34_offline
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode34_offline
+endi
+if $dnode3Status != offline then
+  sleep 2000
+  goto wait_dnode34_offline
+endi
+if $dnode4Status != offline then
+  sleep 2000
+  goto wait_dnode34_offline
+endi
+
+# check using select
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+
+sql insert into $tb values ( now , 20000 ) ( now + 1a, 20001 ) ( now + 2a, 20002 )
+$totalRows = $totalRows + 3
+
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+

tests/script/unique/arbitrator/insert_duplicationTs.sim

+# Test case describe: dnode1 is only mnode, dnode2/dnode3 are only vnode
+# step 1: start dnode1
+# step 2: start dnode2 and dnode3, and all added into cluster (Suppose dnode2 is master-vnode)
+# step 3: create db, table, insert data, and Falling disc into file (control only one file, e.g. 1841)
+# step 4: insert old data(now-15d) and new data(now+15d), control data rows  in order to save in cache, not falling disc 
+# step 5: stop dnode2, so date rows falling disc, generate two new files 1840, 1842 in dnode2
+# step 6: insert two data rows: now-16d, now+16d
+# step 7: restart dnode2, waiting sync end
+# expect: in dnode2, the files 1840 and 1842 will be removed
+
+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+system sh/deploy.sh -n dnode2 -i 2
+system sh/deploy.sh -n dnode3 -i 3
+system sh/deploy.sh -n dnode4 -i 4
+
+system sh/cfg.sh -n dnode1 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode4 -c numOfMnodes -v 1
+
+system sh/cfg.sh -n dnode1 -c walLevel -v 2
+system sh/cfg.sh -n dnode2 -c walLevel -v 2
+system sh/cfg.sh -n dnode3 -c walLevel -v 2
+system sh/cfg.sh -n dnode4 -c walLevel -v 2
+
+system sh/cfg.sh -n dnode1 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode2 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode3 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode4 -c balanceInterval -v 10
+
+system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode2 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 4
+system sh/cfg.sh -n dnode4 -c numOfTotalVnodes -v 4
+
+system sh/cfg.sh -n dnode1 -c alternativeRole -v 1
+system sh/cfg.sh -n dnode2 -c alternativeRole -v 2
+system sh/cfg.sh -n dnode3 -c alternativeRole -v 2
+system sh/cfg.sh -n dnode4 -c alternativeRole -v 2
+
+system sh/cfg.sh -n dnode1 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode2 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode3 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode4 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode5 -c maxtablesPerVnode -v 4
+
+system sh/cfg.sh -n dnode1 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode2 -c arbitrator -v $arbitrator
+system sh/cfg.sh -n dnode3 -c arbitrator -v $arbitrator
+
+print ============== step0: start tarbitrator
+system sh/exec_tarbitrator.sh -s start
+
+print ============== step1: start dnode1, only deploy mnode
+system sh/exec.sh -n dnode1 -s start
+sleep 3000
+sql connect
+
+print ============== step2: start dnode2/dnode3 and add into cluster , then create database with replica 2, and create table, insert data
+system sh/exec.sh -n dnode2 -s start
+system sh/exec.sh -n dnode3 -s start
+sql create dnode $hostname2
+sql create dnode $hostname3
+sleep 3000
+
+$totalTableNum = 1
+$sleepTimer = 3000
+
+$db = db
+sql create database $db replica 2 cache 1
+sql use $db
+
+# create table , insert data
+$stb = stb
+sql create table $stb (ts timestamp, c1 double) tags(t1 int)
+$rowNum = 130000
+$tblNum = $totalTableNum
+$totalRows = 0
+$tsStart = 1577808000000  # 2020-01-01 00:00:00.000
+
+# insert over 2M data in order to falling disc, generate one file
+$i = 0
+while $i < $tblNum
+  $tb = tb . $i
+  sql create table $tb using $stb tags( $i )
+ 
+  $x = 0
+  while $x < $rowNum
+#    $ts = $tsStart + $x
+    sql insert into $tb values ( now + 0s , $x ) ( now + 1s , $x ) ( now + 2s , $x ) ( now + 3s , $x ) ( now + 4s , $x ) ( now + 5s , $x ) ( now + 6s , $x ) ( now + 7s , $x ) ( now + 8s , $x ) ( now + 9s , $x )
+    $x = $x + 10
+  endw 
+  $totalRows = $totalRows + $x
+  print       info: inserted $x rows into $tb and totalRows: $totalRows
+  $i = $i + 1
+endw 
+
+sql select count(*) from $stb
+print rows:$rows data00:$data00
+if $rows != 1 then
+  return -1
+endi
+
+if $data00 == 0 then
+  return -1
+endi
+
+$totalRows = $data00
+
+print ============== step3: insert old data(now-15d) and new data(now+15d), control data rows  in order to save in cache, not falling disc
+sql insert into $tb values ( now - 20d , -20 )
+sql insert into $tb values ( now - 40d , -40 )
+$totalRows = $totalRows + 2
+
+print ============== step4: stop dnode2, so date rows falling disc, generate two new files in dnode2
+system sh/exec.sh -n dnode2 -s stop -x SIGINT
+sleep $sleepTimer
+
+wait_dnode2_offline:
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode2_offline
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode2Status != offline then
+  sleep 2000
+  goto wait_dnode2_offline
+endi
+if $dnode3Status != ready then
+  sleep 2000
+  goto wait_dnode2_offline
+endi
+
+sleep $sleepTimer   # waitting for move master vnode of dnode2 to dnode3
+# check using select
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+
+print ============== step5: insert two data rows: now-16d, now+16d, 
+sql insert into $tb values ( now - 21d , -21 )
+sql insert into $tb values ( now - 41d , -41 )
+$totalRows = $totalRows + 2
+
+print ============== step5: restart dnode2, waiting sync end
+system sh/exec.sh -n dnode2 -s start
+sleep 3000
+
+wait_dnode2_ready:
+sql show dnodes
+if $rows != 3 then
+  sleep 2000
+  goto wait_dnode2_ready
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+#$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode2_ready
+endi
+
+sleep $sleepTimer
+# check using select
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+

tests/script/unique/arbitrator/replica_changeWithArbitrator.sim

@@ -2,18 +2,42 @@ system sh/stop_dnodes.sh
 system sh/deploy.sh -n dnode1 -i 1
 system sh/deploy.sh -n dnode2 -i 2
 system sh/deploy.sh -n dnode3 -i 3
+system sh/deploy.sh -n dnode4 -i 4
 
-system sh/cfg.sh -n dnode1 -c numOfMnodes -v 2
-system sh/cfg.sh -n dnode2 -c numOfMnodes -v 2
-system sh/cfg.sh -n dnode3 -c numOfMnodes -v 2
+system sh/cfg.sh -n dnode1 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 1
+system sh/cfg.sh -n dnode3 -c numOfMnodes -v 1
 
-system sh/cfg.sh -n dnode1 -c walLevel -v 1
-system sh/cfg.sh -n dnode2 -c walLevel -v 1
-system sh/cfg.sh -n dnode3 -c walLevel -v 1
+system sh/cfg.sh -n dnode1 -c walLevel -v 2
+system sh/cfg.sh -n dnode2 -c walLevel -v 2
+system sh/cfg.sh -n dnode3 -c walLevel -v 2
+system sh/cfg.sh -n dnode4 -c walLevel -v 2
 
 system sh/cfg.sh -n dnode1 -c balanceInterval -v 10
 system sh/cfg.sh -n dnode2 -c balanceInterval -v 10
 system sh/cfg.sh -n dnode3 -c balanceInterval -v 10
+system sh/cfg.sh -n dnode4 -c balanceInterval -v 10
+
+system sh/cfg.sh -n dnode1 -c numOfTotalVnodes -v 8
+system sh/cfg.sh -n dnode2 -c numOfTotalVnodes -v 8
+system sh/cfg.sh -n dnode3 -c numOfTotalVnodes -v 8
+system sh/cfg.sh -n dnode4 -c numOfTotalVnodes -v 8
+
+system sh/cfg.sh -n dnode1 -c mnodeEqualVnodeNum -v 4
+#system sh/cfg.sh -n dnode2 -c mnodeEqualVnodeNum -v 4
+#system sh/cfg.sh -n dnode3 -c mnodeEqualVnodeNum -v 4
+#system sh/cfg.sh -n dnode4 -c mnodeEqualVnodeNum -v 4
+
+system sh/cfg.sh -n dnode1 -c alternativeRole -v 0
+system sh/cfg.sh -n dnode2 -c alternativeRole -v 0
+system sh/cfg.sh -n dnode3 -c alternativeRole -v 0
+system sh/cfg.sh -n dnode4 -c alternativeRole -v 0
+
+system sh/cfg.sh -n dnode1 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode2 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode3 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode4 -c maxtablesPerVnode -v 4
+system sh/cfg.sh -n dnode5 -c maxtablesPerVnode -v 4
 
 system sh/cfg.sh -n dnode1 -c arbitrator -v $arbitrator
 system sh/cfg.sh -n dnode2 -c arbitrator -v $arbitrator
@@ -22,21 +46,23 @@ system sh/cfg.sh -n dnode3 -c arbitrator -v $arbitrator
 print ============== step0: start tarbitrator
 system sh/exec_tarbitrator.sh -s start
 
-
-print ============== step1: replica is 1, and start 1 dnode
+print ============== step1: replica is 1, and start 1 dnode, then create tables and insert data
 system sh/exec.sh -n dnode1 -s start
 sleep 3000
 sql connect
 
-$db = replica_db1
-sql create database $db replica 1 maxTables 4
+$totalTableNum = 12
+
+$db = db
+sql create database $db replica 1 maxTables  $totalTableNum
 sql use $db
 
 # create table , insert data
-$stb = repl_stb
+$stb = stb
 sql create table $stb (ts timestamp, c1 int) tags(t1 int)
 $rowNum = 10
-$tblNum = 12
+$tblNum = $totalTableNum
+$totalRows = $rowNum * $tblNum
 
 $ts0 = 1420041600000
 $ts = $ts0
@@ -55,46 +81,136 @@ while $i < $tblNum
     $x = $x + 1
   endw 
   $i = $i + 1
+  print       $tb inserted rows: $x
 endw 
 
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+
 print ============== step2: add 1 new dnode, expect balanced
 system sh/exec.sh -n dnode2 -s start
 sql create dnode $hostname2
+sleep 3000
 
 # expect after balanced, 2 vondes in dnode1, 1 vonde in dnode2
-$x = 0
-show2: 
-	$x = $x + 1
-	sleep 2000
-	if $x == 10 then
-		return -1
-	endi
-	
+wait_dnode2_ready:
 sql show dnodes
-print dnode1 openVnodes $data2_1
-print dnode2 openVnodes $data2_2
-if $data2_1 != 2 then
-	goto show2
+if $rows != 2 then
+  sleep 2000
+  goto wait_dnode2_ready
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+#$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode1Status != ready then
+  sleep 2000
+  goto wait_dnode2_ready
 endi
-if $data2_2 != 1 then
-	goto show2
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode2_ready
 endi
 
-print ============== step4: stop dnode1, and wait dnode2 master
+print ============== step3: stop dnode1/dnode2, modify cfg mpeers to 2, and restart dnode1/dnode2
 system sh/exec.sh -n dnode1 -s stop
+system sh/exec.sh -n dnode2 -s stop
+sleep 3000
 
-$x = 0
-loop_wait:
-	$x = $x + 1
-	sleep 2000
-	if $x == 10 then
-	  print ERROR: after dnode1 stop, dnode2 didn't become a master! 
-		return -1
-	endi
+system sh/cfg.sh -n dnode1 -c numOfMnodes -v 2
+system sh/cfg.sh -n dnode2 -c numOfMnodes -v 2
+
+system sh/cfg.sh -n dnode1 -c mnodeEqualVnodeNum -v 4
+system sh/cfg.sh -n dnode2 -c mnodeEqualVnodeNum -v 4
+
+system sh/cfg.sh -n dnode1 -c alternativeRole -v 1
+system sh/cfg.sh -n dnode2 -c alternativeRole -v 0
+
+system sh/exec.sh -n dnode1 -s start
+system sh/exec.sh -n dnode2 -s start
+sleep 5000
+
+
+print ============= step4: wait dnode ready
+wait_dnode_ready:
+sql show dnodes
+if $rows != 2 then
+  sleep 2000
+  goto wait_dnode_ready
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+#print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+#$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $dnode1Status != ready then
+  sleep 2000
+  goto wait_dnode_ready
+endi
+if $dnode2Status != ready then
+  sleep 2000
+  goto wait_dnode_ready
+endi
+
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
+endi
+
+print ============== step5: stop dnode1
+system sh/exec.sh -n dnode1 -s stop
+sleep 3000
+
+wait_dnode2_master:
 sql show mnodes
-$dnodeRole = $data2_1
-print dnodeRole ==> $dnodeRole
+if $rows != 2 then
+  sleep 2000
+  goto wait_dnode2_master
+endi
+print $data0_1  $data1_1  $data2_1  $data3_1  $data4_1
+print $data0_2  $data1_2  $data2_2  $data3_2  $data4_2
+#print $data0_3  $data1_3  $data2_3  $data3_3  $data4_3
+#print $data0_4  $data1_4  $data2_4  $data3_4  $data4_4
+#print $data0_5  $data1_5  $data2_5  $data3_5  $data4_5
+#print $data0_6  $data1_6  $data2_6  $data3_6  $data4_6
+$dnode1Status = $data4_1  
+$dnode2Status = $data4_2
+#$dnode3Status = $data4_3  
+#$dnode4Status = $data4_4 
+#$dnode5Status = $data4_5
+
+if $data2_1 != offline then
+  sleep 2000
+  goto wait_dnode2_master
+endi
+if $data2_2 != master then
+  sleep 2000
+  goto wait_dnode2_master
+endi
 
-if $dnodeRole != master then
-  goto loop_wait
+sql select count(*) from $stb
+print data00 $data00
+if $data00 != $totalRows then
+	return -1
 endi
+
+