fix(sync): release SyncNode ref

Jenkinsfile2

@@ -127,6 +127,25 @@ def pre_test(){
     '''
     return 1
 }
+def pre_test_build_mac() {
+    sh '''
+        hostname
+        date
+    '''
+    sh '''
+        cd ${WK}
+        rm -rf debug
+        mkdir debug
+    '''
+    sh '''
+        cd ${WK}/debug
+        cmake ..
+        make -j8
+    '''
+    sh '''
+        date
+    '''
+}
 def pre_test_win(){
     bat '''
         hostname
@@ -334,6 +353,17 @@ pipeline {
                         }
                     }
                 }
+                stage('mac test') {
+                    agent{label " Mac_catalina "}
+                    steps {
+                        catchError(buildResult: 'FAILURE', stageResult: 'FAILURE') {
+                            timeout(time: 20, unit: 'MINUTES'){
+                                pre_test()
+                                pre_test_build_mac()
+                            }
+                        }
+                    }
+                }
                 stage('linux test') {
                     agent{label " worker03 || slave215 || slave217 || slave219 "}
                     options { skipDefaultCheckout() } 

README-CN.md

@@ -174,6 +174,8 @@ cmake .. -G "NMake Makefiles"
 nmake
 ```
 
+如果你使用的是 Visual Studio 2022 版本， 脚本 `vcvarsall.bat` 的默认安装路径是 `C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvarsall.bat`。 
+
 ### Mac OS X 系统
 
 安装 Xcode 命令行工具和 cmake. 在 Catalina 和 Big Sur 操作系统上，需要安装 XCode 11.4+ 版本。

README.md

@@ -136,7 +136,7 @@ cmake .. -DCPUTYPE=mips64 && cmake --build .
 
 ### On Windows platform
 
-If you use the Visual Studio 2013, please open a command window by executing "cmd.exe".
+If you use Visual Studio 2013, please open a command window by executing "cmd.exe".
 Please specify "amd64" for 64 bits Windows or specify "x86" is for 32 bits Windows when you execute vcvarsall.bat.
 ```cmd
 mkdir debug && cd debug
@@ -145,7 +145,7 @@ cmake .. -G "NMake Makefiles"
 nmake
 ```
 
-If you use the Visual Studio 2019 or 2017:
+If you use Visual Studio 2019 or 2017:
 
 please open a command window by executing "cmd.exe".
 Please specify "x64" for 64 bits Windows or specify "x86" is for 32 bits Windows when you execute vcvarsall.bat.
@@ -164,6 +164,8 @@ cmake .. -G "NMake Makefiles"
 nmake
 ```
 
+If you use Visual Studio 2022, the only change is the default path of `vcvarsall.bat`, which is `C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Auxiliary\Build\vcvarsall.bat`.
+
 ### On Mac OS X platform
 
 Please install XCode command line tools and cmake. Verified with XCode 11.4+ on Catalina and Big Sur.

contrib/CMakeLists.txt

@@ -118,6 +118,7 @@ execute_process(COMMAND "${CMAKE_COMMAND}" --build .
 # ================================================================================================
 # googletest
 if(${BUILD_TEST})
+    set(gtest_force_shared_crt ON CACHE BOOL "" FORCE)
     add_subdirectory(googletest EXCLUDE_FROM_ALL)
     target_include_directories(
         gtest
@@ -259,7 +260,7 @@ if(${BUILD_MSVCREGEX})
     SET_TARGET_PROPERTIES(msvcregex PROPERTIES OUTPUT_NAME msvcregex)
 endif(${BUILD_MSVCREGEX})
 
-# msvcregex
+# wcwidth
 if(${BUILD_WCWIDTH})
     add_library(wcwidth STATIC "")
     target_sources(wcwidth

include/common/tmsg.h

@@ -55,8 +55,9 @@ extern int32_t tMsgDict[];
 
 #define TMSG_SEG_CODE(TYPE) (((TYPE)&0xff00) >> 8)
 #define TMSG_SEG_SEQ(TYPE)  ((TYPE)&0xff)
-#define TMSG_INFO(TYPE)     tMsgInfo[tMsgDict[TMSG_SEG_CODE(TYPE)] + TMSG_SEG_SEQ(TYPE)]
+#define TMSG_INFO(TYPE) \
-#define TMSG_INDEX(TYPE)    (tMsgDict[TMSG_SEG_CODE(TYPE)] + TMSG_SEG_SEQ(TYPE))
+  (((TYPE) >= 0 && (TYPE) < TDMT_MAX) ? tMsgInfo[tMsgDict[TMSG_SEG_CODE(TYPE)] + TMSG_SEG_SEQ(TYPE)] : 0)
+#define TMSG_INDEX(TYPE) (tMsgDict[TMSG_SEG_CODE(TYPE)] + TMSG_SEG_SEQ(TYPE))
 
 typedef uint16_t tmsg_t;
 
@@ -708,6 +709,7 @@ typedef struct {
   int32_t buffer;        // MB
   int32_t pageSize;
   int32_t pages;
+  int32_t lastRowMem;
   int32_t daysPerFile;
   int32_t daysToKeep0;
   int32_t daysToKeep1;
@@ -736,6 +738,7 @@ typedef struct {
   int32_t buffer;
   int32_t pageSize;
   int32_t pages;
+  int32_t lastRowMem;
   int32_t daysPerFile;
   int32_t daysToKeep0;
   int32_t daysToKeep1;
@@ -1023,8 +1026,10 @@ typedef struct {
   int64_t     clusterId;
   int64_t     rebootTime;
   int64_t     updateTime;
-  int32_t     numOfCores;
+  float       numOfCores;
   int32_t     numOfSupportVnodes;
+  int64_t     memTotal;
+  int64_t     memAvail;
   char        dnodeEp[TSDB_EP_LEN];
   SMnodeLoad  mload;
   SQnodeLoad  qload;
@@ -1079,6 +1084,7 @@ typedef struct {
   int32_t  buffer;
   int32_t  pageSize;
   int32_t  pages;
+  int32_t  lastRowMem;
   int32_t  daysPerFile;
   int32_t  daysToKeep0;
   int32_t  daysToKeep1;
@@ -1131,6 +1137,7 @@ typedef struct {
   int32_t  buffer;
   int32_t  pageSize;
   int32_t  pages;
+  int32_t  lastRowMem;
   int32_t  daysPerFile;
   int32_t  daysToKeep0;
   int32_t  daysToKeep1;

include/libs/nodes/nodes.h

@@ -22,8 +22,8 @@ extern "C" {
 
 #include "tdef.h"
 
-#define nodeType(nodeptr)          (((const SNode*)(nodeptr))->type)
+#define nodeType(nodeptr)              (((const SNode*)(nodeptr))->type)
-#define setNodeType(nodeptr, type) (((SNode*)(nodeptr))->type = (type))
+#define setNodeType(nodeptr, nodetype) (((SNode*)(nodeptr))->type = (nodetype))
 
 #define LIST_LENGTH(l) (NULL != (l) ? (l)->length : 0)
 
@@ -118,6 +118,7 @@ typedef enum ENodeType {
   QUERY_NODE_DROP_TABLE_STMT,
   QUERY_NODE_DROP_SUPER_TABLE_STMT,
   QUERY_NODE_ALTER_TABLE_STMT,
+  QUERY_NODE_ALTER_SUPER_TABLE_STMT,
   QUERY_NODE_CREATE_USER_STMT,
   QUERY_NODE_ALTER_USER_STMT,
   QUERY_NODE_DROP_USER_STMT,

include/libs/nodes/plannodes.h

@@ -82,6 +82,7 @@ typedef struct SScanLogicNode {
 typedef struct SJoinLogicNode {
   SLogicNode node;
   EJoinType  joinType;
+  SNode*     pMergeCondition;
   SNode*     pOnConditions;
   bool       isSingleTableJoin;
 } SJoinLogicNode;
@@ -320,6 +321,7 @@ typedef struct SInterpFuncPhysiNode {
   SNodeList*  pFuncs;
   STimeWindow timeRange;
   int64_t     interval;
+  int8_t      intervalUnit;
   EFillMode   fillMode;
   SNode*      pFillValues;  // SNodeListNode
   SNode*      pTimeSeries;  // SColumnNode
@@ -328,6 +330,7 @@ typedef struct SInterpFuncPhysiNode {
 typedef struct SJoinPhysiNode {
   SPhysiNode node;
   EJoinType  joinType;
+  SNode*     pMergeCondition;
   SNode*     pOnConditions;
   SNodeList* pTargets;
 } SJoinPhysiNode;

include/libs/nodes/querynodes.h

@@ -376,6 +376,7 @@ void nodesRewriteSelectStmt(SSelectStmt* pSelect, ESqlClause clause, FNodeRewrit
 typedef enum ECollectColType { COLLECT_COL_TYPE_COL = 1, COLLECT_COL_TYPE_TAG, COLLECT_COL_TYPE_ALL } ECollectColType;
 int32_t nodesCollectColumns(SSelectStmt* pSelect, ESqlClause clause, const char* pTableAlias, ECollectColType type,
                             SNodeList** pCols);
+int32_t nodesCollectColumnsFromNode(SNode* node, const char* pTableAlias, ECollectColType type, SNodeList** pCols);
 
 typedef bool (*FFuncClassifier)(int32_t funcId);
 int32_t nodesCollectFuncs(SSelectStmt* pSelect, ESqlClause clause, FFuncClassifier classifier, SNodeList** pFuncs);

include/libs/scalar/scalar.h

@@ -25,6 +25,8 @@ extern "C" {
 
 typedef struct SFilterInfo SFilterInfo;
 
+int32_t scalarGetOperatorResultType(SDataType left, SDataType right, EOperatorType op, SDataType* pRes);
+
 /*
 pNode will be freed in API;
 *pRes need to freed in caller

include/libs/scheduler/scheduler.h

@@ -130,7 +130,7 @@ void schedulerStopQueryHb(void *pTrans);
  * Free the query job
  * @param pJob
  */
-void schedulerFreeJob(int64_t job, int32_t errCode);
+void schedulerFreeJob(int64_t* job, int32_t errCode);
 
 void schedulerDestroy(void);
 

include/libs/transport/trpc.h

@@ -110,12 +110,15 @@ typedef struct {
 } SRpcCtx;
 
 int32_t rpcInit();
-void    rpcCleanup();
+
-void *  rpcOpen(const SRpcInit *pRpc);
+void  rpcCleanup();
-void    rpcClose(void *);
+void *rpcOpen(const SRpcInit *pRpc);
-void *  rpcMallocCont(int32_t contLen);
+
-void    rpcFreeCont(void *pCont);
+void  rpcClose(void *);
-void *  rpcReallocCont(void *ptr, int32_t contLen);
+void  rpcCloseImpl(void *);
+void *rpcMallocCont(int32_t contLen);
+void  rpcFreeCont(void *pCont);
+void *rpcReallocCont(void *ptr, int32_t contLen);
 
 // Because taosd supports multi-process mode
 // These functions should not be used on the server side

include/util/taoserror.h

@@ -173,11 +173,12 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_DNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0341)
 #define TSDB_CODE_MND_TOO_MANY_DNODES           TAOS_DEF_ERROR_CODE(0, 0x0342)
 #define TSDB_CODE_MND_NO_ENOUGH_DNODES          TAOS_DEF_ERROR_CODE(0, 0x0343)
-#define TSDB_CODE_MND_INVALID_CLUSTER_CFG       TAOS_DEF_ERROR_CODE(0, 0x0344)
+#define TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE    TAOS_DEF_ERROR_CODE(0, 0x0344)
-#define TSDB_CODE_MND_INVALID_CLUSTER_ID        TAOS_DEF_ERROR_CODE(0, 0x0345)
+#define TSDB_CODE_MND_INVALID_CLUSTER_CFG       TAOS_DEF_ERROR_CODE(0, 0x0345)
-#define TSDB_CODE_MND_INVALID_DNODE_CFG         TAOS_DEF_ERROR_CODE(0, 0x0346)
+#define TSDB_CODE_MND_INVALID_CLUSTER_ID        TAOS_DEF_ERROR_CODE(0, 0x0346)
-#define TSDB_CODE_MND_INVALID_DNODE_EP          TAOS_DEF_ERROR_CODE(0, 0x0347)
+#define TSDB_CODE_MND_INVALID_DNODE_CFG         TAOS_DEF_ERROR_CODE(0, 0x0347)
-#define TSDB_CODE_MND_INVALID_DNODE_ID          TAOS_DEF_ERROR_CODE(0, 0x0348)
+#define TSDB_CODE_MND_INVALID_DNODE_EP          TAOS_DEF_ERROR_CODE(0, 0x0348)
+#define TSDB_CODE_MND_INVALID_DNODE_ID          TAOS_DEF_ERROR_CODE(0, 0x0349)
 
 // mnode-node
 #define TSDB_CODE_MND_MNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0350)
@@ -579,6 +580,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_INVALID_TABLE_OPTION      TAOS_DEF_ERROR_CODE(0, 0x265C)
 #define TSDB_CODE_PAR_INVALID_INTERP_CLAUSE     TAOS_DEF_ERROR_CODE(0, 0x265D)
 #define TSDB_CODE_PAR_NO_VALID_FUNC_IN_WIN      TAOS_DEF_ERROR_CODE(0, 0x265E)
+#define TSDB_CODE_PAR_ONLY_SUPPORT_SINGLE_TABLE TAOS_DEF_ERROR_CODE(0, 0x265F)
 
 //planner
 #define TSDB_CODE_PLAN_INTERNAL_ERROR           TAOS_DEF_ERROR_CODE(0, 0x2700)

include/util/tdef.h

@@ -334,6 +334,9 @@ typedef enum ELogicConditionType {
 #define TSDB_MIN_DB_CACHE_LAST_ROW      0
 #define TSDB_MAX_DB_CACHE_LAST_ROW      3
 #define TSDB_DEFAULT_CACHE_LAST_ROW     0
+#define TSDB_MIN_DB_LAST_ROW_MEM        1  // MB
+#define TSDB_MAX_DB_LAST_ROW_MEM        65536
+#define TSDB_DEFAULT_LAST_ROW_MEM       1
 #define TSDB_DB_STREAM_MODE_OFF         0
 #define TSDB_DB_STREAM_MODE_ON          1
 #define TSDB_DEFAULT_DB_STREAM_MODE     0

source/client/inc/clientInt.h

@@ -65,7 +65,7 @@ enum {
 typedef struct SAppInstInfo SAppInstInfo;
 
 typedef struct {
-  char*   key;
+  char* key;
   // statistics
   int32_t reportCnt;
   int32_t connKeyCnt;
@@ -177,14 +177,14 @@ typedef struct SReqResultInfo {
 } SReqResultInfo;
 
 typedef struct SRequestSendRecvBody {
-  tsem_t             rspSem;  // not used now
+  tsem_t            rspSem;  // not used now
-  __taos_async_fn_t  queryFp;
+  __taos_async_fn_t queryFp;
-  __taos_async_fn_t  fetchFp;
+  __taos_async_fn_t fetchFp;
-  void*              param;
+  void*             param;
-  SDataBuf           requestMsg;
+  SDataBuf          requestMsg;
-  int64_t            queryJob;  // query job, created according to sql query DAG.
+  int64_t           queryJob;  // query job, created according to sql query DAG.
-  int32_t            subplanNum;
+  int32_t           subplanNum;
-  SReqResultInfo     resInfo;
+  SReqResultInfo    resInfo;
 } SRequestSendRecvBody;
 
 typedef struct {
@@ -284,6 +284,7 @@ static FORCE_INLINE SReqResultInfo* tscGetCurResInfo(TAOS_RES* res) {
 extern SAppInfo appInfo;
 extern int32_t  clientReqRefPool;
 extern int32_t  clientConnRefPool;
+extern void*    tscQhandle;
 
 __async_send_cb_fn_t getMsgRspHandle(int32_t msgType);
 
@@ -301,7 +302,7 @@ void         destroyRequest(SRequestObj* pRequest);
 SRequestObj* acquireRequest(int64_t rid);
 int32_t      releaseRequest(int64_t rid);
 int32_t      removeRequest(int64_t rid);
-void         doDestroyRequest(void *p);
+void         doDestroyRequest(void* p);
 
 char* getDbOfConnection(STscObj* pObj);
 void  setConnectionDB(STscObj* pTscObj, const char* db);
@@ -336,8 +337,9 @@ int  hbHandleRsp(SClientHbBatchRsp* hbRsp);
 // cluster level
 SAppHbMgr* appHbMgrInit(SAppInstInfo* pAppInstInfo, char* key);
 void       appHbMgrCleanup(void);
-void       hbRemoveAppHbMrg(SAppHbMgr **pAppHbMgr);
+void       hbRemoveAppHbMrg(SAppHbMgr** pAppHbMgr);
-void       closeAllRequests(SHashObj *pRequests);
+void       destroyAllRequests(SHashObj* pRequests);
+void       stopAllRequests(SHashObj* pRequests);
 
 // conn level
 int  hbRegisterConn(SAppHbMgr* pAppHbMgr, int64_t tscRefId, int64_t clusterId, int8_t connType);
@@ -356,6 +358,9 @@ int32_t      removeMeta(STscObj* pTscObj, SArray* tbList);  // todo move to clie
 int32_t      handleAlterTbExecRes(void* res, struct SCatalog* pCatalog);  // todo move to xxx
 bool         qnodeRequired(SRequestObj* pRequest);
 
+void initTscQhandle();
+void cleanupTscQhandle();
+
 #ifdef __cplusplus
 }
 #endif

source/client/src/clientEnv.c

@@ -25,6 +25,7 @@
 #include "tmsg.h"
 #include "tref.h"
 #include "trpc.h"
+#include "tsched.h"
 #include "ttime.h"
 
 #define TSC_VAR_NOT_RELEASE 1
@@ -34,9 +35,20 @@ SAppInfo appInfo;
 int32_t  clientReqRefPool = -1;
 int32_t  clientConnRefPool = -1;
 
+void *tscQhandle = NULL;
+
 static TdThreadOnce tscinit = PTHREAD_ONCE_INIT;
 volatile int32_t    tscInitRes = 0;
 
+void initTscQhandle() {
+  // init handle
+  tscQhandle = taosInitScheduler(4096, 5, "tsc");
+}
+
+void cleanupTscQhandle() {
+  // destroy handle
+  taosCleanUpScheduler(tscQhandle);
+}
 static int32_t registerRequest(SRequestObj *pRequest) {
   STscObj *pTscObj = acquireTscObj(pRequest->pTscObj->id);
   if (NULL == pTscObj) {
@@ -121,12 +133,31 @@ void *openTransporter(const char *user, const char *auth, int32_t numOfThread) {
   return pDnodeConn;
 }
 
-void closeAllRequests(SHashObj *pRequests) {
+void destroyAllRequests(SHashObj *pRequests) {
   void *pIter = taosHashIterate(pRequests, NULL);
   while (pIter != NULL) {
     int64_t *rid = pIter;
 
-    removeRequest(*rid);
+    SRequestObj *pRequest = acquireRequest(*rid);
+    if (pRequest) {
+      destroyRequest(pRequest);
+      releaseRequest(*rid);
+    }
+
+    pIter = taosHashIterate(pRequests, pIter);
+  }
+}
+
+void stopAllRequests(SHashObj *pRequests) {
+  void *pIter = taosHashIterate(pRequests, NULL);
+  while (pIter != NULL) {
+    int64_t *rid = pIter;
+
+    SRequestObj *pRequest = acquireRequest(*rid);
+    if (pRequest) {
+      taos_stop_query(pRequest);
+      releaseRequest(*rid);
+    }
 
     pIter = taosHashIterate(pRequests, pIter);
   }
@@ -153,12 +184,18 @@ void destroyAppInst(SAppInstInfo *pAppInfo) {
 }
 
 void destroyTscObj(void *pObj) {
+  if (NULL == pObj) {
+    return;
+  }
+
   STscObj *pTscObj = pObj;
+  int64_t  tscId = pTscObj->id;
+  tscTrace("begin to destroy tscObj %" PRIx64 " p:%p", tscId, pTscObj);
 
   SClientHbKey connKey = {.tscRid = pTscObj->id, .connType = pTscObj->connType};
   hbDeregisterConn(pTscObj->pAppInfo->pAppHbMgr, connKey);
   int64_t connNum = atomic_sub_fetch_64(&pTscObj->pAppInfo->numOfConns, 1);
-  closeAllRequests(pTscObj->pRequests);
+  destroyAllRequests(pTscObj->pRequests);
   schedulerStopQueryHb(pTscObj->pAppInfo->pTransporter);
   tscDebug("connObj 0x%" PRIx64 " p:%p destroyed, remain inst totalConn:%" PRId64, pTscObj->id, pTscObj,
            pTscObj->pAppInfo->numOfConns);
@@ -167,7 +204,9 @@ void destroyTscObj(void *pObj) {
     destroyAppInst(pTscObj->pAppInfo);
   }
   taosThreadMutexDestroy(&pTscObj->mutex);
-  taosMemoryFreeClear(pTscObj);
+  taosMemoryFree(pTscObj);
+
+  tscTrace("end to destroy tscObj %" PRIx64 " p:%p", tscId, pTscObj);
 }
 
 void *createTscObj(const char *user, const char *auth, const char *db, int32_t connType, SAppInstInfo *pAppInfo) {
@@ -261,14 +300,18 @@ int32_t releaseRequest(int64_t rid) { return taosReleaseRef(clientReqRefPool, ri
 int32_t removeRequest(int64_t rid) { return taosRemoveRef(clientReqRefPool, rid); }
 
 void doDestroyRequest(void *p) {
-  assert(p != NULL);
+  if (NULL == p) {
+    return;
+  }
+
   SRequestObj *pRequest = (SRequestObj *)p;
 
+  int64_t reqId = pRequest->self;
+  tscTrace("begin to destroy request %" PRIx64 " p:%p", reqId, pRequest);
+
   taosHashRemove(pRequest->pTscObj->pRequests, &pRequest->self, sizeof(pRequest->self));
 
-  if (pRequest->body.queryJob != 0) {
+  schedulerFreeJob(&pRequest->body.queryJob, 0);
-    schedulerFreeJob(pRequest->body.queryJob, 0);
-  }
 
   taosMemoryFreeClear(pRequest->msgBuf);
   taosMemoryFreeClear(pRequest->sqlstr);
@@ -284,7 +327,9 @@ void doDestroyRequest(void *p) {
   if (pRequest->self) {
     deregisterRequest(pRequest);
   }
-  taosMemoryFreeClear(pRequest);
+  taosMemoryFree(pRequest);
+
+  tscTrace("end to destroy request %" PRIx64 " p:%p", reqId, pRequest);
 }
 
 void destroyRequest(SRequestObj *pRequest) {
@@ -292,6 +337,8 @@ void destroyRequest(SRequestObj *pRequest) {
     return;
   }
 
+  taos_stop_query(pRequest);
+
   removeRequest(pRequest->self);
 }
 
@@ -299,7 +346,7 @@ void taos_init_imp(void) {
   // In the APIs of other program language, taos_cleanup is not available yet.
   // So, to make sure taos_cleanup will be invoked to clean up the allocated resource to suppress the valgrind warning.
   atexit(taos_cleanup);
-
+  initTscQhandle();
   errno = TSDB_CODE_SUCCESS;
   taosSeedRand(taosGetTimestampSec());
 

source/client/src/clientImpl.c

@@ -25,6 +25,7 @@
 #include "tmsgtype.h"
 #include "tpagedbuf.h"
 #include "tref.h"
+#include "tsched.h"
 
 static int32_t       initEpSetFromCfg(const char* firstEp, const char* secondEp, SCorEpSet* pEpSet);
 static SMsgSendInfo* buildConnectMsg(SRequestObj* pRequest);
@@ -56,14 +57,14 @@ static char* getClusterKey(const char* user, const char* auth, const char* ip, i
 }
 
 bool chkRequestKilled(void* param) {
-  bool killed = false;
+  bool         killed = false;
   SRequestObj* pRequest = acquireRequest((int64_t)param);
   if (NULL == pRequest || pRequest->killed) {
     killed = true;
   }
 
   releaseRequest((int64_t)param);
-  
+
   return killed;
 }
 
@@ -645,9 +646,7 @@ int32_t scheduleQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNodeList
   pRequest->body.resInfo.execRes = res.res;
 
   if (code != TSDB_CODE_SUCCESS) {
-    if (pRequest->body.queryJob != 0) {
+    schedulerFreeJob(&pRequest->body.queryJob, 0);
-      schedulerFreeJob(pRequest->body.queryJob, 0);
-    }
 
     pRequest->code = code;
     terrno = code;
@@ -658,9 +657,7 @@ int32_t scheduleQuery(SRequestObj* pRequest, SQueryPlan* pDag, SArray* pNodeList
       TDMT_VND_CREATE_TABLE == pRequest->type) {
     pRequest->body.resInfo.numOfRows = res.numOfRows;
 
-    if (pRequest->body.queryJob != 0) {
+    schedulerFreeJob(&pRequest->body.queryJob, 0);
-      schedulerFreeJob(pRequest->body.queryJob, 0);
-    }
   }
 
   pRequest->code = res.code;
@@ -769,7 +766,7 @@ int32_t handleQueryExecRsp(SRequestObj* pRequest) {
       code = handleSubmitExecRes(pRequest, pRes->res, pCatalog, &epset);
       break;
     }
-    case TDMT_SCH_QUERY: 
+    case TDMT_SCH_QUERY:
     case TDMT_SCH_MERGE_QUERY: {
       code = handleQueryExecRes(pRequest, pRes->res, pCatalog, &epset);
       break;
@@ -792,10 +789,7 @@ void schedulerExecCb(SQueryResult* pResult, void* param, int32_t code) {
       TDMT_VND_CREATE_TABLE == pRequest->type) {
     pRequest->body.resInfo.numOfRows = pResult->numOfRows;
 
-    if (pRequest->body.queryJob != 0) {
+    schedulerFreeJob(&pRequest->body.queryJob, 0);
-      schedulerFreeJob(pRequest->body.queryJob, 0);
-      pRequest->body.queryJob = 0;
-    }
   }
 
   taosMemoryFree(pResult);
@@ -1239,7 +1233,16 @@ void updateTargetEpSet(SMsgSendInfo* pSendInfo, STscObj* pTscObj, SRpcMsg* pMsg,
   }
 }
 
-void processMsgFromServer(void* parent, SRpcMsg* pMsg, SEpSet* pEpSet) {
+typedef struct SchedArg {
+  SRpcMsg msg;
+  SEpSet* pEpset;
+} SchedArg;
+
+void doProcessMsgFromServer(SSchedMsg* schedMsg) {
+  SchedArg* arg = (SchedArg*)schedMsg->ahandle;
+  SRpcMsg*  pMsg = &arg->msg;
+  SEpSet*   pEpSet = arg->pEpset;
+
   SMsgSendInfo* pSendInfo = (SMsgSendInfo*)pMsg->info.ahandle;
   assert(pMsg->info.ahandle != NULL);
   STscObj* pTscObj = NULL;
@@ -1272,7 +1275,8 @@ void processMsgFromServer(void* parent, SRpcMsg* pMsg, SEpSet* pEpSet) {
 
   updateTargetEpSet(pSendInfo, pTscObj, pMsg, pEpSet);
 
-  SDataBuf buf = {.msgType = pMsg->msgType, .len = pMsg->contLen, .pData = NULL, .handle = pMsg->info.handle, .pEpSet = pEpSet};
+  SDataBuf buf = {
+      .msgType = pMsg->msgType, .len = pMsg->contLen, .pData = NULL, .handle = pMsg->info.handle, .pEpSet = pEpSet};
 
   if (pMsg->contLen > 0) {
     buf.pData = taosMemoryCalloc(1, pMsg->contLen);
@@ -1287,6 +1291,25 @@ void processMsgFromServer(void* parent, SRpcMsg* pMsg, SEpSet* pEpSet) {
   pSendInfo->fp(pSendInfo->param, &buf, pMsg->code);
   rpcFreeCont(pMsg->pCont);
   destroySendMsgInfo(pSendInfo);
+
+  taosMemoryFree(arg);
+}
+
+void processMsgFromServer(void* parent, SRpcMsg* pMsg, SEpSet* pEpSet) {
+  SSchedMsg schedMsg = {0};
+
+  SEpSet* tEpSet = pEpSet != NULL ? taosMemoryCalloc(1, sizeof(SEpSet)) : NULL;
+  if (tEpSet != NULL) {
+    *tEpSet = *pEpSet;
+  }
+
+  SchedArg* arg = taosMemoryCalloc(1, sizeof(SchedArg));
+  arg->msg = *pMsg;
+  arg->pEpset = tEpSet;
+
+  schedMsg.fp = doProcessMsgFromServer;
+  schedMsg.ahandle = arg;
+  taosScheduleTask(tscQhandle, &schedMsg);
 }
 
 TAOS* taos_connect_auth(const char* ip, const char* user, const char* auth, const char* db, uint16_t port) {
@@ -1415,7 +1438,7 @@ void* doAsyncFetchRows(SRequestObj* pRequest, bool setupOneRowPtr, bool convertU
       pParam = taosMemoryCalloc(1, sizeof(SSyncQueryParam));
       tsem_init(&pParam->sem, 0, 0);
     }
-    
+
     // convert ucs4 to native multi-bytes string
     pResultInfo->convertUcs4 = convertUcs4;
 

source/client/src/clientMain.c

@@ -47,11 +47,9 @@ int taos_options(TSDB_OPTION option, const void *arg, ...) {
   atomic_store_32(&lock, 0);
   return ret;
 }
-
 // this function may be called by user or system, or by both simultaneously.
 void taos_cleanup(void) {
-  tscInfo("start to cleanup client environment");
+  tscInfo("start  to cleanup client environment");
-
   if (atomic_val_compare_exchange_32(&sentinel, TSC_VAR_NOT_RELEASE, TSC_VAR_RELEASED) != TSC_VAR_NOT_RELEASE) {
     return;
   }
@@ -74,8 +72,8 @@ void taos_cleanup(void) {
   catalogDestroy();
   schedulerDestroy();
 
+  cleanupTscQhandle();
   rpcCleanup();
-
   tscInfo("all local resources released");
   taosCleanupCfg();
   taosCloseLog();
@@ -108,7 +106,7 @@ TAOS *taos_connect(const char *ip, const char *user, const char *pass, const cha
   if (pObj) {
     int64_t *rid = taosMemoryCalloc(1, sizeof(int64_t));
     *rid = pObj->id;
-    return (TAOS*)rid;
+    return (TAOS *)rid;
   }
 
   return NULL;
@@ -196,10 +194,10 @@ void taos_kill_query(TAOS *taos) {
   if (NULL == taos) {
     return;
   }
-  int64_t rid = *(int64_t*)taos;
+
-  
+  int64_t  rid = *(int64_t *)taos;
-  STscObj* pTscObj = acquireTscObj(rid);
+  STscObj *pTscObj = acquireTscObj(rid);
-  closeAllRequests(pTscObj->pRequests);
+  stopAllRequests(pTscObj->pRequests);
   releaseTscObj(rid);
 }
 
@@ -244,7 +242,7 @@ TAOS_ROW taos_fetch_row(TAOS_RES *res) {
 #endif
 
   } else if (TD_RES_TMQ(res)) {
-    SMqRspObj      *msg = ((SMqRspObj *)res);
+    SMqRspObj *     msg = ((SMqRspObj *)res);
     SReqResultInfo *pResultInfo;
     if (msg->resIter == -1) {
       pResultInfo = tmqGetNextResInfo(res, true);
@@ -420,7 +418,7 @@ int taos_affected_rows(TAOS_RES *res) {
     return 0;
   }
 
-  SRequestObj    *pRequest = (SRequestObj *)res;
+  SRequestObj *   pRequest = (SRequestObj *)res;
   SReqResultInfo *pResInfo = &pRequest->body.resInfo;
   return pResInfo->numOfRows;
 }
@@ -480,9 +478,7 @@ void taos_stop_query(TAOS_RES *res) {
     return;
   }
 
-  if (pRequest->body.queryJob) {
+  schedulerFreeJob(&pRequest->body.queryJob, TSDB_CODE_TSC_QUERY_KILLED);
-    schedulerFreeJob(pRequest->body.queryJob, TSDB_CODE_TSC_QUERY_KILLED);
-  }
 
   tscDebug("request %" PRIx64 " killed", pRequest->requestId);
 }
@@ -606,7 +602,7 @@ int *taos_get_column_data_offset(TAOS_RES *res, int columnIndex) {
   }
 
   SReqResultInfo *pResInfo = tscGetCurResInfo(res);
-  TAOS_FIELD     *pField = &pResInfo->userFields[columnIndex];
+  TAOS_FIELD *    pField = &pResInfo->userFields[columnIndex];
   if (!IS_VAR_DATA_TYPE(pField->type)) {
     return 0;
   }
@@ -650,8 +646,8 @@ const char *taos_get_server_info(TAOS *taos) {
 typedef struct SqlParseWrapper {
   SParseContext *pCtx;
   SCatalogReq    catalogReq;
-  SRequestObj   *pRequest;
+  SRequestObj *  pRequest;
-  SQuery        *pQuery;
+  SQuery *       pQuery;
 } SqlParseWrapper;
 
 static void destorySqlParseWrapper(SqlParseWrapper *pWrapper) {
@@ -672,8 +668,8 @@ void retrieveMetaCallback(SMetaData *pResultMeta, void *param, int32_t code) {
   tscDebug("enter meta callback, code %s", tstrerror(code));
 
   SqlParseWrapper *pWrapper = (SqlParseWrapper *)param;
-  SQuery          *pQuery = pWrapper->pQuery;
+  SQuery *         pQuery = pWrapper->pQuery;
-  SRequestObj     *pRequest = pWrapper->pRequest;
+  SRequestObj *    pRequest = pWrapper->pRequest;
 
   if (code == TSDB_CODE_SUCCESS) {
     code = qAnalyseSqlSemantic(pWrapper->pCtx, &pWrapper->catalogReq, pResultMeta, pQuery);
@@ -722,31 +718,29 @@ int32_t createParseContext(const SRequestObj *pRequest, SParseContext **pCxt) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
 
-  **pCxt = (SParseContext){
+  **pCxt = (SParseContext){.requestId = pRequest->requestId,
-      .requestId = pRequest->requestId,
+                           .requestRid = pRequest->self,
-      .requestRid = pRequest->self,
+                           .acctId = pTscObj->acctId,
-      .acctId = pTscObj->acctId,
+                           .db = pRequest->pDb,
-      .db = pRequest->pDb,
+                           .topicQuery = false,
-      .topicQuery = false,
+                           .pSql = pRequest->sqlstr,
-      .pSql = pRequest->sqlstr,
+                           .sqlLen = pRequest->sqlLen,
-      .sqlLen = pRequest->sqlLen,
+                           .pMsg = pRequest->msgBuf,
-      .pMsg = pRequest->msgBuf,
+                           .msgLen = ERROR_MSG_BUF_DEFAULT_SIZE,
-      .msgLen = ERROR_MSG_BUF_DEFAULT_SIZE,
+                           .pTransporter = pTscObj->pAppInfo->pTransporter,
-      .pTransporter = pTscObj->pAppInfo->pTransporter,
+                           .pStmtCb = NULL,
-      .pStmtCb = NULL,
+                           .pUser = pTscObj->user,
-      .pUser = pTscObj->user,
+                           .schemalessType = pTscObj->schemalessType,
-      .schemalessType = pTscObj->schemalessType,
+                           .isSuperUser = (0 == strcmp(pTscObj->user, TSDB_DEFAULT_USER)),
-      .isSuperUser = (0 == strcmp(pTscObj->user, TSDB_DEFAULT_USER)),
+                           .async = true,
-      .async = true,
+                           .svrVer = pTscObj->sVer,
-      .svrVer = pTscObj->sVer,
+                           .nodeOffline = (pTscObj->pAppInfo->onlineDnodes < pTscObj->pAppInfo->totalDnodes)};
-      .nodeOffline = (pTscObj->pAppInfo->onlineDnodes < pTscObj->pAppInfo->totalDnodes)
-  };
   return TSDB_CODE_SUCCESS;
 }
 
 void doAsyncQuery(SRequestObj *pRequest, bool updateMetaForce) {
   SParseContext *pCxt = NULL;
-  STscObj       *pTscObj = pRequest->pTscObj;
+  STscObj *      pTscObj = pRequest->pTscObj;
   int32_t        code = 0;
 
   if (pRequest->retry++ > REQUEST_TOTAL_EXEC_TIMES) {
@@ -911,10 +905,10 @@ int taos_load_table_info(TAOS *taos, const char *tableNameList) {
     return terrno;
   }
 
-  int64_t rid = *(int64_t*)taos;
+  int64_t       rid = *(int64_t *)taos;
   const int32_t MAX_TABLE_NAME_LENGTH = 12 * 1024 * 1024;  // 12MB list
   int32_t       code = 0;
-  SRequestObj  *pRequest = NULL;
+  SRequestObj * pRequest = NULL;
   SCatalogReq   catalogReq = {0};
 
   if (NULL == tableNameList) {

source/common/src/tdatablock.c

@@ -895,7 +895,7 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
         SBlockOrderInfo* pOrder = taosArrayGet(pOrderInfo, 0);
 
         int64_t p0 = taosGetTimestampUs();
-        
+
         __compar_fn_t fn = getKeyComparFunc(pColInfoData->info.type, pOrder->order);
         taosSort(pColInfoData->pData, pDataBlock->info.rows, pColInfoData->info.bytes, fn);
 
@@ -923,8 +923,9 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
     pInfo->pColData = taosArrayGet(pDataBlock->pDataBlock, pInfo->slotId);
   }
 
+  terrno = 0;
   taosqsort(index, rows, sizeof(int32_t), &helper, dataBlockCompar);
-  if(terrno) return terrno;
+  if (terrno) return terrno;
 
   int64_t p1 = taosGetTimestampUs();
 
@@ -1438,21 +1439,21 @@ static void doShiftBitmap(char* nullBitmap, size_t n, size_t total) {
   }
 }
 
-static int32_t colDataMoveVarData(SColumnInfoData* pColInfoData, size_t start, size_t end){
+static int32_t colDataMoveVarData(SColumnInfoData* pColInfoData, size_t start, size_t end) {
   int32_t dataOffset = -1;
   int32_t dataLen = 0;
   int32_t beigin = start;
-  while(beigin < end){
+  while (beigin < end) {
     int32_t offset = pColInfoData->varmeta.offset[beigin];
-    if(offset == -1) {
+    if (offset == -1) {
       beigin++;
       continue;
     }
-    if(start != 0) {
+    if (start != 0) {
       pColInfoData->varmeta.offset[beigin] = dataLen;
     }
-    char *data = pColInfoData->pData + offset;
+    char* data = pColInfoData->pData + offset;
-    if(dataOffset == -1) dataOffset = offset;   // mark the begin of data
+    if (dataOffset == -1) dataOffset = offset;  // mark the begin of data
     int32_t type = pColInfoData->info.type;
     if (type == TSDB_DATA_TYPE_JSON) {
       dataLen += getJsonValueLen(data);
@@ -1461,7 +1462,7 @@ static int32_t colDataMoveVarData(SColumnInfoData* pColInfoData, size_t start, s
     }
     beigin++;
   }
-  if(dataOffset > 0){
+  if (dataOffset > 0) {
     memmove(pColInfoData->pData, pColInfoData->pData + dataOffset, dataLen);
     memmove(pColInfoData->varmeta.offset, &pColInfoData->varmeta.offset[start], (end - start) * sizeof(int32_t));
   }

source/common/src/tmsg.c

@@ -947,8 +947,10 @@ int32_t tSerializeSStatusReq(void *buf, int32_t bufLen, SStatusReq *pReq) {
   if (tEncodeI64(&encoder, pReq->clusterId) < 0) return -1;
   if (tEncodeI64(&encoder, pReq->rebootTime) < 0) return -1;
   if (tEncodeI64(&encoder, pReq->updateTime) < 0) return -1;
-  if (tEncodeI32(&encoder, pReq->numOfCores) < 0) return -1;
+  if (tEncodeFloat(&encoder, pReq->numOfCores) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->numOfSupportVnodes) < 0) return -1;
+  if (tEncodeI64(&encoder, pReq->memTotal) < 0) return -1;
+  if (tEncodeI64(&encoder, pReq->memAvail) < 0) return -1;
   if (tEncodeCStr(&encoder, pReq->dnodeEp) < 0) return -1;
 
   // cluster cfg
@@ -1008,8 +1010,10 @@ int32_t tDeserializeSStatusReq(void *buf, int32_t bufLen, SStatusReq *pReq) {
   if (tDecodeI64(&decoder, &pReq->clusterId) < 0) return -1;
   if (tDecodeI64(&decoder, &pReq->rebootTime) < 0) return -1;
   if (tDecodeI64(&decoder, &pReq->updateTime) < 0) return -1;
-  if (tDecodeI32(&decoder, &pReq->numOfCores) < 0) return -1;
+  if (tDecodeFloat(&decoder, &pReq->numOfCores) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->numOfSupportVnodes) < 0) return -1;
+  if (tDecodeI64(&decoder, &pReq->memTotal) < 0) return -1;
+  if (tDecodeI64(&decoder, &pReq->memAvail) < 0) return -1;
   if (tDecodeCStrTo(&decoder, pReq->dnodeEp) < 0) return -1;
 
   // cluster cfg
@@ -1974,6 +1978,7 @@ int32_t tSerializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq) {
   if (tEncodeI32(&encoder, pReq->buffer) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pageSize) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pages) < 0) return -1;
+  if (tEncodeI32(&encoder, pReq->lastRowMem) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysPerFile) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep0) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep1) < 0) return -1;
@@ -2015,6 +2020,7 @@ int32_t tDeserializeSCreateDbReq(void *buf, int32_t bufLen, SCreateDbReq *pReq)
   if (tDecodeI32(&decoder, &pReq->buffer) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pageSize) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pages) < 0) return -1;
+  if (tDecodeI32(&decoder, &pReq->lastRowMem) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysPerFile) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep0) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep1) < 0) return -1;
@@ -2069,6 +2075,7 @@ int32_t tSerializeSAlterDbReq(void *buf, int32_t bufLen, SAlterDbReq *pReq) {
   if (tEncodeI32(&encoder, pReq->buffer) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pageSize) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pages) < 0) return -1;
+  if (tEncodeI32(&encoder, pReq->lastRowMem) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysPerFile) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep0) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep1) < 0) return -1;
@@ -2094,6 +2101,7 @@ int32_t tDeserializeSAlterDbReq(void *buf, int32_t bufLen, SAlterDbReq *pReq) {
   if (tDecodeI32(&decoder, &pReq->buffer) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pageSize) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pages) < 0) return -1;
+  if (tDecodeI32(&decoder, &pReq->lastRowMem) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysPerFile) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep0) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep1) < 0) return -1;
@@ -2679,10 +2687,12 @@ int32_t tDeserializeSDbCfgRsp(void *buf, int32_t bufLen, SDbCfgRsp *pRsp) {
   if (tDecodeI8(&decoder, &pRsp->strict) < 0) return -1;
   if (tDecodeI8(&decoder, &pRsp->cacheLastRow) < 0) return -1;
   if (tDecodeI32(&decoder, &pRsp->numOfRetensions) < 0) return -1;
-  pRsp->pRetensions = taosArrayInit(pRsp->numOfRetensions, sizeof(SRetention));
+  if (pRsp->numOfRetensions > 0) {
-  if (pRsp->pRetensions == NULL) {
+    pRsp->pRetensions = taosArrayInit(pRsp->numOfRetensions, sizeof(SRetention));
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    if (pRsp->pRetensions == NULL) {
-    return -1;
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      return -1;
+    }
   }
 
   for (int32_t i = 0; i < pRsp->numOfRetensions; ++i) {
@@ -3586,6 +3596,7 @@ int32_t tSerializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *pR
   if (tEncodeI32(&encoder, pReq->buffer) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pageSize) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pages) < 0) return -1;
+  if (tEncodeI32(&encoder, pReq->lastRowMem) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysPerFile) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep0) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep1) < 0) return -1;
@@ -3643,6 +3654,7 @@ int32_t tDeserializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *
   if (tDecodeI32(&decoder, &pReq->buffer) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pageSize) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pages) < 0) return -1;
+  if (tDecodeI32(&decoder, &pReq->lastRowMem) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysPerFile) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep0) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep1) < 0) return -1;
@@ -3665,7 +3677,6 @@ int32_t tDeserializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *
     SReplica *pReplica = &pReq->replicas[i];
     if (tDecodeSReplica(&decoder, pReplica) < 0) return -1;
   }
-
   if (tDecodeI32(&decoder, &pReq->numOfRetensions) < 0) return -1;
   pReq->pRetensions = taosArrayInit(pReq->numOfRetensions, sizeof(SRetention));
   if (pReq->pRetensions == NULL) {
@@ -3768,6 +3779,7 @@ int32_t tSerializeSAlterVnodeReq(void *buf, int32_t bufLen, SAlterVnodeReq *pReq
   if (tEncodeI32(&encoder, pReq->buffer) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pageSize) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->pages) < 0) return -1;
+  if (tEncodeI32(&encoder, pReq->lastRowMem) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysPerFile) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep0) < 0) return -1;
   if (tEncodeI32(&encoder, pReq->daysToKeep1) < 0) return -1;
@@ -3782,7 +3794,6 @@ int32_t tSerializeSAlterVnodeReq(void *buf, int32_t bufLen, SAlterVnodeReq *pReq
     SReplica *pReplica = &pReq->replicas[i];
     if (tEncodeSReplica(&encoder, pReplica) < 0) return -1;
   }
-
   tEndEncode(&encoder);
 
   int32_t tlen = encoder.pos;
@@ -3799,6 +3810,7 @@ int32_t tDeserializeSAlterVnodeReq(void *buf, int32_t bufLen, SAlterVnodeReq *pR
   if (tDecodeI32(&decoder, &pReq->buffer) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pageSize) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->pages) < 0) return -1;
+  if (tDecodeI32(&decoder, &pReq->lastRowMem) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysPerFile) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep0) < 0) return -1;
   if (tDecodeI32(&decoder, &pReq->daysToKeep1) < 0) return -1;

source/dnode/mgmt/mgmt_dnode/src/dmHandle.c

@@ -67,6 +67,8 @@ void dmSendStatusReq(SDnodeMgmt *pMgmt) {
   req.updateTime = pMgmt->pData->updateTime;
   req.numOfCores = tsNumOfCores;
   req.numOfSupportVnodes = tsNumOfSupportVnodes;
+  req.memTotal = tsTotalMemoryKB * 1024;
+  req.memAvail = req.memTotal - tsRpcQueueMemoryAllowed - 16 * 1024 * 1024;
   tstrncpy(req.dnodeEp, tsLocalEp, TSDB_EP_LEN);
 
   req.clusterCfg.statusInterval = tsStatusInterval;

source/dnode/mgmt/node_mgmt/src/dmProc.c

@@ -87,8 +87,8 @@ static SProcQueue *dmInitProcQueue(SProc *proc, char *ptr, int32_t size) {
 static void dmCleanupProcQueue(SProcQueue *queue) {}
 
 static inline int32_t dmPushToProcQueue(SProc *proc, SProcQueue *queue, SRpcMsg *pMsg, EProcFuncType ftype) {
-  const void   *pHead = pMsg;
+  const void *  pHead = pMsg;
-  const void   *pBody = pMsg->pCont;
+  const void *  pBody = pMsg->pCont;
   const int16_t rawHeadLen = sizeof(SRpcMsg);
   const int32_t rawBodyLen = pMsg->contLen;
   const int16_t headLen = CEIL8(rawHeadLen);
@@ -257,7 +257,7 @@ int32_t dmInitProc(struct SMgmtWrapper *pWrapper) {
   proc->wrapper = pWrapper;
   proc->name = pWrapper->name;
 
-  SShm   *shm = &proc->shm;
+  SShm *  shm = &proc->shm;
   int32_t cstart = 0;
   int32_t csize = CEIL8(shm->size / 2);
   int32_t pstart = csize;
@@ -281,13 +281,13 @@ int32_t dmInitProc(struct SMgmtWrapper *pWrapper) {
 }
 
 static void *dmConsumChildQueue(void *param) {
-  SProc        *proc = param;
+  SProc *       proc = param;
   SMgmtWrapper *pWrapper = proc->wrapper;
-  SProcQueue   *queue = proc->cqueue;
+  SProcQueue *  queue = proc->cqueue;
   int32_t       numOfMsgs = 0;
   int32_t       code = 0;
   EProcFuncType ftype = DND_FUNC_REQ;
-  SRpcMsg      *pMsg = NULL;
+  SRpcMsg *     pMsg = NULL;
 
   dDebug("node:%s, start to consume from cqueue", proc->name);
   do {
@@ -324,13 +324,13 @@ static void *dmConsumChildQueue(void *param) {
 }
 
 static void *dmConsumParentQueue(void *param) {
-  SProc        *proc = param;
+  SProc *       proc = param;
   SMgmtWrapper *pWrapper = proc->wrapper;
-  SProcQueue   *queue = proc->pqueue;
+  SProcQueue *  queue = proc->pqueue;
   int32_t       numOfMsgs = 0;
   int32_t       code = 0;
   EProcFuncType ftype = DND_FUNC_REQ;
-  SRpcMsg      *pMsg = NULL;
+  SRpcMsg *     pMsg = NULL;
 
   dDebug("node:%s, start to consume from pqueue", proc->name);
   do {
@@ -353,7 +353,7 @@ static void *dmConsumParentQueue(void *param) {
       rpcRegisterBrokenLinkArg(pMsg);
     } else if (ftype == DND_FUNC_RELEASE) {
       dmRemoveProcRpcHandle(proc, pMsg->info.handle);
-      rpcReleaseHandle(pMsg->info.handle, (int8_t)pMsg->code);
+      rpcReleaseHandle(&pMsg->info, TAOS_CONN_SERVER);
     } else {
       dError("node:%s, invalid ftype:%d from pqueue", proc->name, ftype);
       rpcFreeCont(pMsg->pCont);

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

@@ -245,7 +245,7 @@ static inline void dmReleaseHandle(SRpcHandleInfo *pHandle, int8_t type) {
     SRpcMsg msg = {.code = type, .info = *pHandle};
     dmPutToProcPQueue(&pWrapper->proc, &msg, DND_FUNC_RELEASE);
   } else {
-    rpcReleaseHandle(pHandle->handle, type);
+    rpcReleaseHandle(pHandle, type);
   }
 }
 

source/dnode/mgmt/node_util/src/dmEps.c

@@ -280,7 +280,7 @@ static void dmPrintEps(SDnodeData *pData) {
   dDebug("print dnode list, num:%d", numOfEps);
   for (int32_t i = 0; i < numOfEps; i++) {
     SDnodeEp *pEp = taosArrayGet(pData->dnodeEps, i);
-    dDebug("dnode:%d, fqdn:%s port:%u is_mnode:%d", pEp->id, pEp->ep.fqdn, pEp->ep.port, pEp->isMnode);
+    dDebug("dnode:%d, fqdn:%s port:%u isMnode:%d", pEp->id, pEp->ep.fqdn, pEp->ep.port, pEp->isMnode);
   }
 }
 

source/dnode/mgmt/test/sut/src/sut.cpp

@@ -30,12 +30,14 @@ void Testbase::InitLog(const char* path) {
   tsdbDebugFlag = 0;
   tsLogEmbedded = 1;
   tsAsyncLog = 0;
-  tsRpcQueueMemoryAllowed = 1024 * 1024 * 64;
+  
-
   taosRemoveDir(path);
   taosMkDir(path);
   tstrncpy(tsLogDir, path, PATH_MAX);
-  if (taosInitLog("taosdlog", 1) != 0) {
+
+  taosGetSystemInfo();
+  tsRpcQueueMemoryAllowed = tsTotalMemoryKB * 0.1;
+if (taosInitLog("taosdlog", 1) != 0) {
     printf("failed to init log file\n");
   }
 }

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

@@ -148,7 +148,10 @@ typedef struct {
   int32_t    accessTimes;
   int32_t    numOfVnodes;
   int32_t    numOfSupportVnodes;
-  int32_t    numOfCores;
+  float      numOfCores;
+  int64_t    memTotal;
+  int64_t    memAvail;
+  int64_t    memUsed;
   EDndReason offlineReason;
   uint16_t   port;
   char       fqdn[TSDB_FQDN_LEN];
@@ -243,6 +246,7 @@ typedef struct {
   int32_t buffer;
   int32_t pageSize;
   int32_t pages;
+  int32_t lastRowMem;
   int32_t daysPerFile;
   int32_t daysToKeep0;
   int32_t daysToKeep1;
@@ -255,8 +259,8 @@ typedef struct {
   int8_t  compression;
   int8_t  replications;
   int8_t  strict;
-  int8_t  cacheLastRow;
   int8_t  hashMethod;  // default is 1
+  int8_t  cacheLastRow;
   int32_t numOfRetensions;
   SArray* pRetensions;
   int8_t  schemaless;

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

@@ -30,6 +30,8 @@ SSdbRaw *mndVgroupActionEncode(SVgObj *pVgroup);
 SEpSet   mndGetVgroupEpset(SMnode *pMnode, const SVgObj *pVgroup);
 int32_t  mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId);
 void     mndSortVnodeGid(SVgObj *pVgroup);
+int64_t  mndGetVnodesMemory(SMnode *pMnode, int32_t dnodeId);
+int64_t  mndGetVgroupMemory(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup);
 
 SArray *mndBuildDnodesArray(SMnode *, int32_t exceptDnodeId);
 int32_t mndAllocSmaVgroup(SMnode *, SDbObj *pDb, SVgObj *pVgroup);

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

@@ -93,6 +93,7 @@ static SSdbRaw *mndDbActionEncode(SDbObj *pDb) {
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.buffer, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.pageSize, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.pages, _OVER)
+  SDB_SET_INT32(pRaw, dataPos, pDb->cfg.lastRowMem, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.daysPerFile, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.daysToKeep0, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.daysToKeep1, _OVER)
@@ -165,6 +166,7 @@ static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.buffer, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.pageSize, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.pages, _OVER)
+  SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.lastRowMem, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.daysPerFile, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.daysToKeep0, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.daysToKeep1, _OVER)
@@ -230,8 +232,9 @@ static int32_t mndDbActionUpdate(SSdb *pSdb, SDbObj *pOld, SDbObj *pNew) {
   pOld->cfgVersion = pNew->cfgVersion;
   pOld->vgVersion = pNew->vgVersion;
   pOld->cfg.buffer = pNew->cfg.buffer;
-  pOld->cfg.pages = pNew->cfg.pages;
   pOld->cfg.pageSize = pNew->cfg.pageSize;
+  pOld->cfg.pages = pNew->cfg.pages;
+  pOld->cfg.lastRowMem = pNew->cfg.lastRowMem;
   pOld->cfg.daysPerFile = pNew->cfg.daysPerFile;
   pOld->cfg.daysToKeep0 = pNew->cfg.daysToKeep0;
   pOld->cfg.daysToKeep1 = pNew->cfg.daysToKeep1;
@@ -288,6 +291,7 @@ static int32_t mndCheckDbCfg(SMnode *pMnode, SDbCfg *pCfg) {
   if (pCfg->buffer < TSDB_MIN_BUFFER_PER_VNODE || pCfg->buffer > TSDB_MAX_BUFFER_PER_VNODE) return -1;
   if (pCfg->pageSize < TSDB_MIN_PAGESIZE_PER_VNODE || pCfg->pageSize > TSDB_MAX_PAGESIZE_PER_VNODE) return -1;
   if (pCfg->pages < TSDB_MIN_PAGES_PER_VNODE || pCfg->pages > TSDB_MAX_PAGES_PER_VNODE) return -1;
+  if (pCfg->lastRowMem < TSDB_MIN_DB_LAST_ROW_MEM || pCfg->lastRowMem > TSDB_MAX_DB_LAST_ROW_MEM) return -1;
   if (pCfg->daysPerFile < TSDB_MIN_DAYS_PER_FILE || pCfg->daysPerFile > TSDB_MAX_DAYS_PER_FILE) return -1;
   if (pCfg->daysToKeep0 < TSDB_MIN_KEEP || pCfg->daysToKeep0 > TSDB_MAX_KEEP) return -1;
   if (pCfg->daysToKeep1 < TSDB_MIN_KEEP || pCfg->daysToKeep1 > TSDB_MAX_KEEP) return -1;
@@ -336,6 +340,7 @@ static void mndSetDefaultDbCfg(SDbCfg *pCfg) {
   if (pCfg->replications < 0) pCfg->replications = TSDB_DEFAULT_DB_REPLICA;
   if (pCfg->strict < 0) pCfg->strict = TSDB_DEFAULT_DB_STRICT;
   if (pCfg->cacheLastRow < 0) pCfg->cacheLastRow = TSDB_DEFAULT_CACHE_LAST_ROW;
+  if (pCfg->lastRowMem <= 0) pCfg->lastRowMem = TSDB_DEFAULT_LAST_ROW_MEM;
   if (pCfg->numOfRetensions < 0) pCfg->numOfRetensions = 0;
   if (pCfg->schemaless < 0) pCfg->schemaless = TSDB_DB_SCHEMALESS_OFF;
 }
@@ -434,6 +439,7 @@ static int32_t mndCreateDb(SMnode *pMnode, SRpcMsg *pReq, SCreateDbReq *pCreate,
       .buffer = pCreate->buffer,
       .pageSize = pCreate->pageSize,
       .pages = pCreate->pages,
+      .lastRowMem = pCreate->lastRowMem,
       .daysPerFile = pCreate->daysPerFile,
       .daysToKeep0 = pCreate->daysToKeep0,
       .daysToKeep1 = pCreate->daysToKeep1,
@@ -475,7 +481,7 @@ static int32_t mndCreateDb(SMnode *pMnode, SRpcMsg *pReq, SCreateDbReq *pCreate,
   int32_t code = -1;
   STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB, pReq);
   if (pTrans == NULL) goto _OVER;
-
+  // mndTransSetSerial(pTrans);
   mDebug("trans:%d, used to create db:%s", pTrans->id, pCreate->db);
 
   mndTransSetDbName(pTrans, dbObj.name, NULL);
@@ -622,6 +628,11 @@ static int32_t mndSetDbCfgFromAlterDbReq(SDbObj *pDb, SAlterDbReq *pAlter) {
     terrno = 0;
   }
 
+  if (pAlter->lastRowMem > 0 && pAlter->lastRowMem != pDb->cfg.lastRowMem) {
+    pDb->cfg.lastRowMem = pAlter->lastRowMem;
+    terrno = 0;
+  }
+
   if (pAlter->replications > 0 && pAlter->replications != pDb->cfg.replications) {
 #if 1
     terrno = TSDB_CODE_OPS_NOT_SUPPORT;

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

@@ -15,8 +15,8 @@
 
 #define _DEFAULT_SOURCE
 #include "mndDnode.h"
-#include "mndPrivilege.h"
 #include "mndMnode.h"
+#include "mndPrivilege.h"
 #include "mndQnode.h"
 #include "mndShow.h"
 #include "mndSnode.h"
@@ -274,15 +274,14 @@ static void mndGetDnodeData(SMnode *pMnode, SArray *pDnodeEps) {
 
     SDnodeEp dnodeEp = {0};
     dnodeEp.id = pDnode->id;
-    dnodeEp.isMnode = 0;
     dnodeEp.ep.port = pDnode->port;
     memcpy(dnodeEp.ep.fqdn, pDnode->fqdn, TSDB_FQDN_LEN);
+    sdbRelease(pSdb, pDnode);
 
+    dnodeEp.isMnode = 0;
     if (mndIsMnode(pMnode, pDnode->id)) {
       dnodeEp.isMnode = 1;
     }
-
-    sdbRelease(pSdb, pDnode);
     taosArrayPush(pDnodeEps, &dnodeEp);
   }
 }
@@ -432,7 +431,8 @@ static int32_t mndProcessStatusReq(SRpcMsg *pReq) {
     }
 
     if (!online) {
-      mInfo("dnode:%d, from offline to online", pDnode->id);
+      mInfo("dnode:%d, from offline to online, memory avail:%" PRId64 " total:%" PRId64 " cores:%.2f", pDnode->id,
+            statusReq.memAvail, statusReq.memTotal, statusReq.numOfCores);
     } else {
       mDebug("dnode:%d, send dnode epset, online:%d dnodeVer:%" PRId64 ":%" PRId64 " reboot:%d", pDnode->id, online,
              statusReq.dnodeVer, dnodeVer, reboot);
@@ -441,6 +441,8 @@ static int32_t mndProcessStatusReq(SRpcMsg *pReq) {
     pDnode->rebootTime = statusReq.rebootTime;
     pDnode->numOfCores = statusReq.numOfCores;
     pDnode->numOfSupportVnodes = statusReq.numOfSupportVnodes;
+    pDnode->memAvail = statusReq.memAvail;
+    pDnode->memTotal = statusReq.memTotal;
 
     SStatusRsp statusRsp = {0};
     statusRsp.dnodeVer = dnodeVer;
@@ -580,7 +582,7 @@ static int32_t mndProcessShowVariablesReq(SRpcMsg *pReq) {
   strcpy(info.name, "timezone");
   snprintf(info.value, TSDB_CONFIG_VALUE_LEN, "%s", tsTimezoneStr);
   taosArrayPush(rsp.variables, &info);
-  
+
   strcpy(info.name, "locale");
   snprintf(info.value, TSDB_CONFIG_VALUE_LEN, "%s", tsLocale);
   taosArrayPush(rsp.variables, &info);

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

@@ -207,6 +207,7 @@ void *mndBuildCreateVnodeReq(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVg
   createReq.buffer = pDb->cfg.buffer;
   createReq.pageSize = pDb->cfg.pageSize;
   createReq.pages = pDb->cfg.pages;
+  createReq.lastRowMem = pDb->cfg.lastRowMem;
   createReq.daysPerFile = pDb->cfg.daysPerFile;
   createReq.daysToKeep0 = pDb->cfg.daysToKeep0;
   createReq.daysToKeep1 = pDb->cfg.daysToKeep1;
@@ -274,8 +275,9 @@ void *mndBuildAlterVnodeReq(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup, int32_
   SAlterVnodeReq alterReq = {0};
   alterReq.vgVersion = pVgroup->version;
   alterReq.buffer = pDb->cfg.buffer;
-  alterReq.pages = pDb->cfg.pages;
   alterReq.pageSize = pDb->cfg.pageSize;
+  alterReq.pages = pDb->cfg.pages;
+  alterReq.lastRowMem = pDb->cfg.lastRowMem;
   alterReq.daysPerFile = pDb->cfg.daysPerFile;
   alterReq.daysToKeep0 = pDb->cfg.daysToKeep0;
   alterReq.daysToKeep1 = pDb->cfg.daysToKeep1;
@@ -392,9 +394,10 @@ static bool mndBuildDnodesArrayFp(SMnode *pMnode, void *pObj, void *p1, void *p2
   bool    online = mndIsDnodeOnline(pDnode, curMs);
   bool    isMnode = mndIsMnode(pMnode, pDnode->id);
   pDnode->numOfVnodes = mndGetVnodesNum(pMnode, pDnode->id);
+  pDnode->memUsed = mndGetVnodesMemory(pMnode, pDnode->id);
 
-  mDebug("dnode:%d, vnodes:%d support_vnodes:%d is_mnode:%d online:%d", pDnode->id, pDnode->numOfVnodes,
+  mDebug("dnode:%d, vnodes:%d supportVnodes:%d isMnode:%d online:%d memory avail:%" PRId64 " used:%" PRId64, pDnode->id,
-         pDnode->numOfSupportVnodes, isMnode, online);
+         pDnode->numOfVnodes, pDnode->numOfSupportVnodes, isMnode, online, pDnode->memAvail, pDnode->memUsed);
 
   if (isMnode) {
     pDnode->numOfVnodes++;
@@ -426,15 +429,7 @@ static int32_t mndCompareDnodeId(int32_t *dnode1Id, int32_t *dnode2Id) { return
 static int32_t mndCompareDnodeVnodes(SDnodeObj *pDnode1, SDnodeObj *pDnode2) {
   float d1Score = (float)pDnode1->numOfVnodes / pDnode1->numOfSupportVnodes;
   float d2Score = (float)pDnode2->numOfVnodes / pDnode2->numOfSupportVnodes;
-#if 0  
-  if (d1Score == d2Score) {
-    return pDnode2->id - pDnode1->id;
-  } else {
-    return d1Score >= d2Score ? 1 : 0;
-  }
-#else
   return d1Score >= d2Score ? 1 : 0;
-#endif
 }
 
 void mndSortVnodeGid(SVgObj *pVgroup) {
@@ -447,7 +442,7 @@ void mndSortVnodeGid(SVgObj *pVgroup) {
   }
 }
 
-static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {
+static int32_t mndGetAvailableDnode(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup, SArray *pArray) {
   SSdb   *pSdb = pMnode->pSdb;
   int32_t allocedVnodes = 0;
   void   *pIter = NULL;
@@ -470,6 +465,16 @@ static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pAr
       return -1;
     }
 
+    int64_t vgMem = mndGetVgroupMemory(pMnode, pDb, pVgroup);
+    if (pDnode->memAvail - vgMem - pDnode->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d, avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pDnode->id, pDnode->memAvail, pDnode->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pDnode->memUsed += vgMem;
+    }
+
     pVgid->dnodeId = pDnode->id;
     if (pVgroup->replica == 1) {
       pVgid->role = TAOS_SYNC_STATE_LEADER;
@@ -477,7 +482,8 @@ static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pAr
       pVgid->role = TAOS_SYNC_STATE_FOLLOWER;
     }
 
-    mInfo("db:%s, vgId:%d, vn:%d dnode:%d is alloced", pVgroup->dbName, pVgroup->vgId, v, pVgid->dnodeId);
+    mInfo("db:%s, vgId:%d, vn:%d is alloced, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+          pVgroup->dbName, pVgroup->vgId, v, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
     pDnode->numOfVnodes++;
   }
 
@@ -498,7 +504,7 @@ int32_t mndAllocSmaVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup) {
   pVgroup->dbUid = pDb->uid;
   pVgroup->replica = 1;
 
-  if (mndGetAvailableDnode(pMnode, pVgroup, pArray) != 0) return -1;
+  if (mndGetAvailableDnode(pMnode, pDb, pVgroup, pArray) != 0) return -1;
 
   mInfo("db:%s, sma vgId:%d is alloced", pDb->name, pVgroup->vgId);
   return 0;
@@ -546,8 +552,7 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
     pVgroup->dbUid = pDb->uid;
     pVgroup->replica = pDb->cfg.replications;
 
-    if (mndGetAvailableDnode(pMnode, pVgroup, pArray) != 0) {
+    if (mndGetAvailableDnode(pMnode, pDb, pVgroup, pArray) != 0) {
-      terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
 
@@ -728,6 +733,43 @@ int32_t mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId) {
   return numOfVnodes;
 }
 
+int64_t mndGetVgroupMemory(SMnode *pMnode, SDbObj *pDbInput, SVgObj *pVgroup) {
+  SDbObj *pDb = pDbInput;
+  if (pDbInput == NULL) {
+    pDb = mndAcquireDb(pMnode, pVgroup->dbName);
+  }
+
+  int64_t vgroupMemroy = (int64_t)pDb->cfg.buffer * 1024 * 1024 + (int64_t)pDb->cfg.pages * pDb->cfg.pageSize * 1024;
+  if (pDb->cfg.cacheLastRow > 0) {
+    vgroupMemroy += (int64_t)pDb->cfg.lastRowMem * 1024 * 1024;
+  }
+
+  if (pDbInput == NULL) {
+    mndReleaseDb(pMnode, pDb);
+  }
+  return vgroupMemroy;
+}
+
+static bool mndGetVnodeMemroyFp(SMnode *pMnode, void *pObj, void *p1, void *p2, void *p3) {
+  SVgObj  *pVgroup = pObj;
+  int32_t  dnodeId = *(int32_t *)p1;
+  int64_t *pVnodeMemory = (int64_t *)p2;
+
+  for (int32_t v = 0; v < pVgroup->replica; ++v) {
+    if (pVgroup->vnodeGid[v].dnodeId == dnodeId) {
+      *pVnodeMemory += mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    }
+  }
+
+  return true;
+}
+
+int64_t mndGetVnodesMemory(SMnode *pMnode, int32_t dnodeId) {
+  int64_t vnodeMemory = 0;
+  sdbTraverse(pMnode->pSdb, SDB_VGROUP, mndGetVnodeMemroyFp, &dnodeId, &vnodeMemory, NULL);
+  return vnodeMemory;
+}
+
 static int32_t mndRetrieveVnodes(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {
   SMnode *pMnode = pReq->info.node;
   SSdb   *pSdb = pMnode->pSdb;
@@ -807,9 +849,20 @@ int32_t mndAddVnodeToVgroup(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {
       return -1;
     }
 
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pDnode->memAvail - vgMem - pDnode->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pDnode->id, pDnode->memAvail, pDnode->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pDnode->memUsed += vgMem;
+    }
+
     pVgid->dnodeId = pDnode->id;
     pVgid->role = TAOS_SYNC_STATE_ERROR;
-    mInfo("db:%s, vgId:%d, vn:%d dnode:%d, is added", pVgroup->dbName, pVgroup->vgId, pVgroup->replica, pVgid->dnodeId);
+    mInfo("db:%s, vgId:%d, vn:%d is added, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+          pVgroup->dbName, pVgroup->vgId, pVgroup->replica, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
 
     pVgroup->replica++;
     pDnode->numOfVnodes++;
@@ -835,7 +888,10 @@ int32_t mndRemoveVnodeFromVgroup(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray
     for (int32_t vn = 0; vn < pVgroup->replica; ++vn) {
       SVnodeGid *pVgid = &pVgroup->vnodeGid[vn];
       if (pVgid->dnodeId == pDnode->id) {
-        mInfo("db:%s, vgId:%d, vn:%d dnode:%d, is removed", pVgroup->dbName, pVgroup->vgId, vn, pVgid->dnodeId);
+        int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+        pDnode->memUsed -= vgMem;
+        mInfo("db:%s, vgId:%d, vn:%d is removed, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+              pVgroup->dbName, pVgroup->vgId, vn, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
         pDnode->numOfVnodes--;
         pVgroup->replica--;
         *pDelVgid = *pVgid;
@@ -1161,6 +1217,17 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew1->memAvail - vgMem - pNew1->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew1->id, pNew1->memAvail, pNew1->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew1->memUsed += vgMem;
+    }
+
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew1->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld1->id) != 0) goto _OVER;
   }
@@ -1173,6 +1240,15 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew2->memAvail - vgMem - pNew2->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew2->id, pNew2->memAvail, pNew2->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew2->memUsed += vgMem;
+    }
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew2->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld2->id) != 0) goto _OVER;
   }
@@ -1185,6 +1261,15 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew3->memAvail - vgMem - pNew3->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew3->id, pNew3->memAvail, pNew3->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew3->memUsed += vgMem;
+    }
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew3->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld3->id) != 0) goto _OVER;
   }
@@ -1415,7 +1500,7 @@ _OVER:
   mndReleaseDb(pMnode, pDb);
 
   return code;
-#endif  
+#endif
 }
 
 int32_t mndBuildAlterVgroupAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup, SArray *pArray) {
@@ -1654,9 +1739,9 @@ static int32_t mndBalanceVgroupBetweenDnode(SMnode *pMnode, STrans *pTrans, SDno
 }
 
 static int32_t mndBalanceVgroup(SMnode *pMnode, SRpcMsg *pReq, SArray *pArray) {
-  int32_t code = -1;
+  int32_t   code = -1;
-  int32_t numOfVgroups = 0;
+  int32_t   numOfVgroups = 0;
-  STrans *pTrans = NULL;
+  STrans   *pTrans = NULL;
   SHashObj *pBalancedVgroups = NULL;
 
   pBalancedVgroups = taosHashInit(16, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false, HASH_NO_LOCK);
@@ -1721,7 +1806,7 @@ static int32_t mndProcessBalanceVgroupMsg(SRpcMsg *pReq) {
   SMnode *pMnode = pReq->info.node;
   int32_t code = -1;
   SArray *pArray = NULL;
-  void   *pIter = NULL;
+  void *pIter = NULL;
   int64_t curMs = taosGetTimestampMs();
 
   SBalanceVgroupReq req = {0};
@@ -1766,7 +1851,7 @@ _OVER:
 
   taosArrayDestroy(pArray);
   return code;
-#endif 
+#endif
 }
 
 bool mndVgroupInDb(SVgObj *pVgroup, int64_t dbUid) { return !pVgroup->isTsma && pVgroup->dbUid == dbUid; }

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

@@ -131,7 +131,7 @@ int32_t sdbSetTable(SSdb *pSdb, SSdbTable table) {
     hashType = TSDB_DATA_TYPE_BINARY;
   }
 
-  SHashObj *hash = taosHashInit(64, taosGetDefaultHashFunction(hashType), true, HASH_NO_LOCK);
+  SHashObj *hash = taosHashInit(64, taosGetDefaultHashFunction(hashType), true, HASH_ENTRY_LOCK);
   if (hash == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;

source/dnode/vnode/src/sma/smaCommit.c

@@ -83,8 +83,8 @@ int32_t smaBegin(SSma *pSma) {
 /**
  * @brief pre-commit for rollup sma.
  *  1) set trigger stat of rsma timer TASK_TRIGGER_STAT_PAUSED.
- *  2) perform persist task for qTaskInfo
+ *  2) wait all triggered fetch tasks finished
- *  3) wait all triggered fetch tasks finished
+ *  3) perform persist task for qTaskInfo
  *
  * @param pSma
  * @return int32_t
@@ -102,10 +102,7 @@ static int32_t tdProcessRSmaPreCommitImpl(SSma *pSma) {
   // step 1: set persistence task paused
   atomic_store_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_PAUSED);
 
-  // step 2: perform persist task for qTaskInfo
+  // step 2: wait all triggered fetch tasks finished
-  tdRSmaPersistExecImpl(pRSmaStat);
-
-  // step 3: wait all triggered fetch tasks finished
   int32_t nLoops = 0;
   while (1) {
     if (T_REF_VAL_GET(pStat) == 0) {
@@ -121,6 +118,9 @@ static int32_t tdProcessRSmaPreCommitImpl(SSma *pSma) {
     }
   }
 
+  // step 3: perform persist task for qTaskInfo
+  tdRSmaPersistExecImpl(pRSmaStat);
+
   smaDebug("vgId:%d, rsma pre commit succeess", SMA_VID(pSma));
 
   return TSDB_CODE_SUCCESS;

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

@@ -79,7 +79,7 @@ int vnodeEncodeConfig(const void *pObj, SJson *pJson) {
     SJson *pNodeRetentions = tjsonCreateArray();
     tjsonAddItemToObject(pJson, "retentions", pNodeRetentions);
     for (int32_t i = 0; i < nRetention; ++i) {
-      SJson      *pNodeRetention = tjsonCreateObject();
+      SJson *           pNodeRetention = tjsonCreateObject();
       const SRetention *pRetention = pCfg->tsdbCfg.retentions + i;
       tjsonAddIntegerToObject(pNodeRetention, "freq", pRetention->freq);
       tjsonAddIntegerToObject(pNodeRetention, "freqUnit", pRetention->freqUnit);
@@ -118,45 +118,45 @@ int vnodeDecodeConfig(const SJson *pJson, void *pObj) {
 
   int32_t code;
   tjsonGetNumberValue(pJson, "vgId", pCfg->vgId, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   if (tjsonGetStringValue(pJson, "dbname", pCfg->dbname) < 0) return -1;
   tjsonGetNumberValue(pJson, "dbId", pCfg->dbId, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "szPage", pCfg->szPage, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "szCache", pCfg->szCache, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "szBuf", pCfg->szBuf, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "isHeap", pCfg->isHeap, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "isWeak", pCfg->isWeak, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "isTsma", pCfg->isTsma, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "isRsma", pCfg->isRsma, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "precision", pCfg->tsdbCfg.precision, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "update", pCfg->tsdbCfg.update, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "compression", pCfg->tsdbCfg.compression, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "slLevel", pCfg->tsdbCfg.slLevel, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "daysPerFile", pCfg->tsdbCfg.days, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "minRows", pCfg->tsdbCfg.minRows, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "maxRows", pCfg->tsdbCfg.maxRows, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "keep0", pCfg->tsdbCfg.keep0, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "keep1", pCfg->tsdbCfg.keep1, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "keep2", pCfg->tsdbCfg.keep2, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
-  SJson *pNodeRetentions = tjsonGetObjectItem(pJson, "retentions");
+  SJson * pNodeRetentions = tjsonGetObjectItem(pJson, "retentions");
   int32_t nRetention = tjsonGetArraySize(pNodeRetentions);
   if (nRetention > TSDB_RETENTION_MAX) {
     nRetention = TSDB_RETENTION_MAX;
@@ -170,30 +170,30 @@ int vnodeDecodeConfig(const SJson *pJson, void *pObj) {
     tjsonGetNumberValue(pNodeRetention, "keepUnit", (pCfg->tsdbCfg.retentions)[i].keepUnit, code);
   }
   tjsonGetNumberValue(pJson, "wal.vgId", pCfg->walCfg.vgId, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "wal.fsyncPeriod", pCfg->walCfg.fsyncPeriod, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "wal.retentionPeriod", pCfg->walCfg.retentionPeriod, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "wal.rollPeriod", pCfg->walCfg.rollPeriod, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "wal.retentionSize", pCfg->walCfg.retentionSize, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "wal.segSize", pCfg->walCfg.segSize, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "wal.level", pCfg->walCfg.level, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "hashBegin", pCfg->hashBegin, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "hashEnd", pCfg->hashEnd, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "hashMethod", pCfg->hashMethod, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
 
   tjsonGetNumberValue(pJson, "syncCfg.replicaNum", pCfg->syncCfg.replicaNum, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
   tjsonGetNumberValue(pJson, "syncCfg.myIndex", pCfg->syncCfg.myIndex, code);
-  if(code < 0) return -1;
+  if (code < 0) return -1;
 
   SJson *pNodeInfoArr = tjsonGetObjectItem(pJson, "syncCfg.nodeInfo");
   int    arraySize = tjsonGetArraySize(pNodeInfoArr);

source/dnode/vnode/src/vnd/vnodeQuery.c

@@ -27,10 +27,10 @@ int vnodeGetTableMeta(SVnode *pVnode, SRpcMsg *pMsg) {
   SMetaReader    mer1 = {0};
   SMetaReader    mer2 = {0};
   char           tableFName[TSDB_TABLE_FNAME_LEN];
-  SRpcMsg        rpcMsg;
+  SRpcMsg        rpcMsg = {0};
   int32_t        code = 0;
   int32_t        rspLen = 0;
-  void          *pRsp = NULL;
+  void *         pRsp = NULL;
   SSchemaWrapper schema = {0};
   SSchemaWrapper schemaTag = {0};
 
@@ -111,6 +111,7 @@ _exit:
   rpcMsg.pCont = pRsp;
   rpcMsg.contLen = rspLen;
   rpcMsg.code = code;
+  rpcMsg.msgType = pMsg->msgType;
 
   if (code) {
     qError("get table %s meta failed cause of %s", infoReq.tbName, tstrerror(code));
@@ -130,10 +131,10 @@ int vnodeGetTableCfg(SVnode *pVnode, SRpcMsg *pMsg) {
   SMetaReader    mer1 = {0};
   SMetaReader    mer2 = {0};
   char           tableFName[TSDB_TABLE_FNAME_LEN];
-  SRpcMsg        rpcMsg;
+  SRpcMsg        rpcMsg = {0};
   int32_t        code = 0;
   int32_t        rspLen = 0;
-  void          *pRsp = NULL;
+  void *         pRsp = NULL;
   SSchemaWrapper schema = {0};
   SSchemaWrapper schemaTag = {0};
 
@@ -220,6 +221,7 @@ _exit:
   rpcMsg.pCont = pRsp;
   rpcMsg.contLen = rspLen;
   rpcMsg.code = code;
+  rpcMsg.msgType = pMsg->msgType;
 
   if (code) {
     qError("get table %s cfg failed cause of %s", cfgReq.tbName, tstrerror(code));
@@ -260,7 +262,7 @@ void vnodeGetInfo(SVnode *pVnode, const char **dbname, int32_t *vgId) {
 }
 
 // wrapper of tsdb read interface
-tsdbReaderT tsdbQueryCacheLast(SVnode *pVnode, SQueryTableDataCond *pCond, STableListInfo* tableList, uint64_t qId,
+tsdbReaderT tsdbQueryCacheLast(SVnode *pVnode, SQueryTableDataCond *pCond, STableListInfo *tableList, uint64_t qId,
                                void *pMemRef) {
 #if 0
   return tsdbQueryCacheLastT(pVnode->pTsdb, pCond, groupList, qId, pMemRef);

source/libs/command/inc/commandInt.h

@@ -29,13 +29,17 @@ extern "C" {
 #define EXPLAIN_TAG_SCAN_FORMAT "Tag Scan on %s"
 #define EXPLAIN_TBL_SCAN_FORMAT "Table Scan on %s"
 #define EXPLAIN_SYSTBL_SCAN_FORMAT "System Table Scan on %s"
+#define EXPLAIN_DISTBLK_SCAN_FORMAT "Block Dist Scan on %s"
+#define EXPLAIN_LASTROW_SCAN_FORMAT "Last Row Scan on %s"
 #define EXPLAIN_PROJECTION_FORMAT "Projection"
 #define EXPLAIN_JOIN_FORMAT "%s"
 #define EXPLAIN_AGG_FORMAT "Aggragate"
 #define EXPLAIN_INDEF_ROWS_FORMAT "Indefinite Rows Function"
 #define EXPLAIN_EXCHANGE_FORMAT "Data Exchange %d:1"
 #define EXPLAIN_SORT_FORMAT "Sort"
+#define EXPLAIN_GROUP_SORT_FORMAT "Group Sort"
 #define EXPLAIN_INTERVAL_FORMAT "Interval on Column %s"
+#define EXPLAIN_MERGE_INTERVAL_FORMAT "Merge Interval on Column %s"
 #define EXPLAIN_FILL_FORMAT "Fill"
 #define EXPLAIN_SESSION_FORMAT "Session"
 #define EXPLAIN_STATE_WINDOW_FORMAT "StateWindow on Column %s"
@@ -62,10 +66,12 @@ extern "C" {
 #define EXPLAIN_COST_FORMAT "cost=%.2f..%.2f"
 #define EXPLAIN_ROWS_FORMAT "rows=%" PRIu64
 #define EXPLAIN_COLUMNS_FORMAT "columns=%d"
+#define EXPLAIN_PSEUDO_COLUMNS_FORMAT "pseudo_columns=%d"
 #define EXPLAIN_WIDTH_FORMAT "width=%d"
 #define EXPLAIN_TABLE_SCAN_FORMAT "order=[asc|%d desc|%d]"
 #define EXPLAIN_GROUPS_FORMAT "groups=%d"
 #define EXPLAIN_WIDTH_FORMAT "width=%d"
+#define EXPLAIN_INTERVAL_VALUE_FORMAT "interval=%" PRId64 "%c"
 #define EXPLAIN_FUNCTIONS_FORMAT "functions=%d"
 #define EXPLAIN_EXECINFO_FORMAT "cost=%.3f..%.3f rows=%" PRIu64
 #define EXPLAIN_MODE_FORMAT "mode=%s"

source/libs/command/src/explain.c

@@ -199,6 +199,31 @@ int32_t qExplainGenerateResChildren(SPhysiNode *pNode, SExplainGroup *group, SNo
       pPhysiChildren = mergePhysiNode->scan.node.pChildren;
       break;
     }
+    case QUERY_NODE_PHYSICAL_PLAN_BLOCK_DIST_SCAN: {
+      SBlockDistScanPhysiNode *distPhysiNode = (SBlockDistScanPhysiNode *)pNode;
+      pPhysiChildren = distPhysiNode->node.pChildren;
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_LAST_ROW_SCAN: {
+      SLastRowScanPhysiNode *lastRowPhysiNode = (SLastRowScanPhysiNode *)pNode;
+      pPhysiChildren = lastRowPhysiNode->node.pChildren;
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT: {
+      SGroupSortPhysiNode *groupSortPhysiNode = (SGroupSortPhysiNode *)pNode;
+      pPhysiChildren = groupSortPhysiNode->node.pChildren;
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_MERGE_INTERVAL: {
+      SMergeIntervalPhysiNode *mergeIntPhysiNode = (SMergeIntervalPhysiNode *)pNode;
+      pPhysiChildren = mergeIntPhysiNode->window.node.pChildren;
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_INTERP_FUNC: {
+      SInterpFuncPhysiNode *interpPhysiNode = (SInterpFuncPhysiNode *)pNode;
+      pPhysiChildren = interpPhysiNode->node.pChildren;
+      break;
+    }
     default:
       qError("not supported physical node type %d", pNode->type);
       QRY_ERR_RET(TSDB_CODE_QRY_APP_ERROR);
@@ -378,6 +403,10 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
       }
       EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, pTagScanNode->pScanCols->length);
       EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      if (pTagScanNode->pScanPseudoCols) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pTagScanNode->pScanPseudoCols->length);
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
       EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pTagScanNode->node.pOutputDataBlockDesc->totalRowSize);
       EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
       EXPLAIN_ROW_END();
@@ -415,6 +444,10 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
 
       EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, pTblScanNode->scan.pScanCols->length);
       EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      if (pTblScanNode->scan.pScanPseudoCols) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pTblScanNode->scan.pScanPseudoCols->length);
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
       EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pTblScanNode->scan.node.pOutputDataBlockDesc->totalRowSize);
       EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       EXPLAIN_ROW_APPEND(EXPLAIN_TABLE_SCAN_FORMAT, pTblScanNode->scanSeq[0], pTblScanNode->scanSeq[1]);
@@ -516,6 +549,10 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
       }
       EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, pSTblScanNode->scan.pScanCols->length);
       EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      if (pSTblScanNode->scan.pScanPseudoCols) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pSTblScanNode->scan.pScanPseudoCols->length);
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT); 
+      }
       EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pSTblScanNode->scan.node.pOutputDataBlockDesc->totalRowSize);
       EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
@@ -1131,6 +1168,258 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
       }
       break;
     }
+    case QUERY_NODE_PHYSICAL_PLAN_BLOCK_DIST_SCAN: {
+      SBlockDistScanPhysiNode *pDistScanNode = (SBlockDistScanPhysiNode *)pNode;
+      EXPLAIN_ROW_NEW(level, EXPLAIN_DISTBLK_SCAN_FORMAT, pDistScanNode->tableName.tname);
+      EXPLAIN_ROW_APPEND(EXPLAIN_LEFT_PARENTHESIS_FORMAT);
+      if (pResNode->pExecInfo) {
+        QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+      EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, pDistScanNode->pScanCols->length);
+      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      if (pDistScanNode->pScanPseudoCols) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pDistScanNode->pScanPseudoCols->length);
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);   
+      }
+      EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pDistScanNode->node.pOutputDataBlockDesc->totalRowSize);
+      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
+      EXPLAIN_ROW_END();
+      QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+
+      if (verbose) {
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_OUTPUT_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT,
+                           nodesGetOutputNumFromSlotList(pDistScanNode->node.pOutputDataBlockDesc->pSlots));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pDistScanNode->node.pOutputDataBlockDesc->outputRowSize);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+
+        if (pDistScanNode->node.pConditions) {
+          EXPLAIN_ROW_NEW(level + 1, EXPLAIN_FILTER_FORMAT);
+          QRY_ERR_RET(nodesNodeToSQL(pDistScanNode->node.pConditions, tbuf + VARSTR_HEADER_SIZE,
+                                     TSDB_EXPLAIN_RESULT_ROW_SIZE, &tlen));
+          EXPLAIN_ROW_END();
+          QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        }
+      }
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_LAST_ROW_SCAN: {
+      SLastRowScanPhysiNode *pLastRowNode = (SLastRowScanPhysiNode *)pNode;
+      EXPLAIN_ROW_NEW(level, EXPLAIN_LASTROW_SCAN_FORMAT, pLastRowNode->tableName.tname);
+      EXPLAIN_ROW_APPEND(EXPLAIN_LEFT_PARENTHESIS_FORMAT);
+      if (pResNode->pExecInfo) {
+        QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+      EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, pLastRowNode->pScanCols->length);
+      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      if (pLastRowNode->pScanPseudoCols) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_PSEUDO_COLUMNS_FORMAT, pLastRowNode->pScanPseudoCols->length);
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+      EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pLastRowNode->node.pOutputDataBlockDesc->totalRowSize);
+      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
+      EXPLAIN_ROW_END();
+      QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+
+      if (verbose) {
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_OUTPUT_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT,
+                           nodesGetOutputNumFromSlotList(pLastRowNode->node.pOutputDataBlockDesc->pSlots));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pLastRowNode->node.pOutputDataBlockDesc->outputRowSize);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+
+        if (pLastRowNode->node.pConditions) {
+          EXPLAIN_ROW_NEW(level + 1, EXPLAIN_FILTER_FORMAT);
+          QRY_ERR_RET(nodesNodeToSQL(pLastRowNode->node.pConditions, tbuf + VARSTR_HEADER_SIZE,
+                                     TSDB_EXPLAIN_RESULT_ROW_SIZE, &tlen));
+          EXPLAIN_ROW_END();
+          QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        }
+      }
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT: {
+      SGroupSortPhysiNode *pSortNode = (SGroupSortPhysiNode *)pNode;
+      EXPLAIN_ROW_NEW(level, EXPLAIN_GROUP_SORT_FORMAT);
+      EXPLAIN_ROW_APPEND(EXPLAIN_LEFT_PARENTHESIS_FORMAT);
+      if (pResNode->pExecInfo) {
+        QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+
+      SDataBlockDescNode *pDescNode = pSortNode->node.pOutputDataBlockDesc;
+      EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, nodesGetOutputNumFromSlotList(pDescNode->pSlots));
+      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pDescNode->totalRowSize);
+      EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
+      EXPLAIN_ROW_END();
+      QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+
+      if (EXPLAIN_MODE_ANALYZE == ctx->mode) {
+        // sort key
+        EXPLAIN_ROW_NEW(level + 1, "Sort Key: ");
+        if (pResNode->pExecInfo) {
+          for (int32_t i = 0; i < LIST_LENGTH(pSortNode->pSortKeys); ++i) {
+            SOrderByExprNode *ptn = (SOrderByExprNode *)nodesListGetNode(pSortNode->pSortKeys, i);
+            EXPLAIN_ROW_APPEND("%s ", nodesGetNameFromColumnNode(ptn->pExpr));
+          }
+        }
+
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+
+        // sort method
+        EXPLAIN_ROW_NEW(level + 1, "Sort Method: ");
+
+        int32_t           nodeNum = taosArrayGetSize(pResNode->pExecInfo);
+        SExplainExecInfo *execInfo = taosArrayGet(pResNode->pExecInfo, 0);
+        SSortExecInfo    *pExecInfo = (SSortExecInfo *)execInfo->verboseInfo;
+        EXPLAIN_ROW_APPEND("%s", pExecInfo->sortMethod == SORT_QSORT_T ? "quicksort" : "merge sort");
+        if (pExecInfo->sortBuffer > 1024 * 1024) {
+          EXPLAIN_ROW_APPEND("  Buffers:%.2f Mb", pExecInfo->sortBuffer / (1024 * 1024.0));
+        } else if (pExecInfo->sortBuffer > 1024) {
+          EXPLAIN_ROW_APPEND("  Buffers:%.2f Kb", pExecInfo->sortBuffer / (1024.0));
+        } else {
+          EXPLAIN_ROW_APPEND("  Buffers:%d b", pExecInfo->sortBuffer);
+        }
+
+        EXPLAIN_ROW_APPEND("  loops:%d", pExecInfo->loops);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+      }
+
+      if (verbose) {
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_OUTPUT_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT,
+                           nodesGetOutputNumFromSlotList(pSortNode->node.pOutputDataBlockDesc->pSlots));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pSortNode->node.pOutputDataBlockDesc->outputRowSize);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+
+        if (pSortNode->node.pConditions) {
+          EXPLAIN_ROW_NEW(level + 1, EXPLAIN_FILTER_FORMAT);
+          QRY_ERR_RET(nodesNodeToSQL(pSortNode->node.pConditions, tbuf + VARSTR_HEADER_SIZE,
+                                     TSDB_EXPLAIN_RESULT_ROW_SIZE, &tlen));
+          EXPLAIN_ROW_END();
+          QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        }
+      }
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_MERGE_INTERVAL: {
+      SMergeIntervalPhysiNode *pIntNode = (SMergeIntervalPhysiNode *)pNode;
+      EXPLAIN_ROW_NEW(level, EXPLAIN_MERGE_INTERVAL_FORMAT, nodesGetNameFromColumnNode(pIntNode->window.pTspk));
+      EXPLAIN_ROW_APPEND(EXPLAIN_LEFT_PARENTHESIS_FORMAT);
+      if (pResNode->pExecInfo) {
+        QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+      EXPLAIN_ROW_APPEND(EXPLAIN_FUNCTIONS_FORMAT, pIntNode->window.pFuncs->length);
+      EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
+      EXPLAIN_ROW_END();
+      QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+
+      if (verbose) {
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_OUTPUT_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT,
+                           nodesGetOutputNumFromSlotList(pIntNode->window.node.pOutputDataBlockDesc->pSlots));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pIntNode->window.node.pOutputDataBlockDesc->outputRowSize);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        uint8_t precision = getIntervalPrecision(pIntNode);
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_TIME_WINDOWS_FORMAT,
+                        INVERAL_TIME_FROM_PRECISION_TO_UNIT(pIntNode->interval, pIntNode->intervalUnit, precision),
+                        pIntNode->intervalUnit, pIntNode->offset, getPrecisionUnit(precision),
+                        INVERAL_TIME_FROM_PRECISION_TO_UNIT(pIntNode->sliding, pIntNode->slidingUnit, precision),
+                        pIntNode->slidingUnit);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+
+        if (pIntNode->window.node.pConditions) {
+          EXPLAIN_ROW_NEW(level + 1, EXPLAIN_FILTER_FORMAT);
+          QRY_ERR_RET(nodesNodeToSQL(pIntNode->window.node.pConditions, tbuf + VARSTR_HEADER_SIZE,
+                                     TSDB_EXPLAIN_RESULT_ROW_SIZE, &tlen));
+          EXPLAIN_ROW_END();
+          QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        }
+      }
+      break;
+    }
+    case QUERY_NODE_PHYSICAL_PLAN_INTERP_FUNC: {
+      SInterpFuncPhysiNode *pInterpNode = (SInterpFuncPhysiNode *)pNode;
+      EXPLAIN_ROW_NEW(level, EXPLAIN_AGG_FORMAT);
+      EXPLAIN_ROW_APPEND(EXPLAIN_LEFT_PARENTHESIS_FORMAT);
+      if (pResNode->pExecInfo) {
+        QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+      if (pInterpNode->pFuncs) {
+        EXPLAIN_ROW_APPEND(EXPLAIN_FUNCTIONS_FORMAT, pInterpNode->pFuncs->length);
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+      }
+      
+      EXPLAIN_ROW_APPEND(EXPLAIN_MODE_FORMAT, nodesGetFillModeString(pInterpNode->fillMode));
+      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+
+      EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
+      EXPLAIN_ROW_END();
+      QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));
+
+      if (verbose) {
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_OUTPUT_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT,
+                           nodesGetOutputNumFromSlotList(pInterpNode->node.pOutputDataBlockDesc->pSlots));
+        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+        EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pInterpNode->node.pOutputDataBlockDesc->outputRowSize);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        if (pInterpNode->pFillValues) {
+          SNodeListNode *pValues = (SNodeListNode *)pInterpNode->pFillValues;
+          EXPLAIN_ROW_NEW(level + 1, EXPLAIN_FILL_VALUE_FORMAT);
+          SNode  *tNode = NULL;
+          int32_t i = 0;
+          FOREACH(tNode, pValues->pNodeList) {
+            if (i) {
+              EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
+            }
+            SValueNode *tValue = (SValueNode *)tNode;
+            char       *value = nodesGetStrValueFromNode(tValue);
+            EXPLAIN_ROW_APPEND(EXPLAIN_STRING_TYPE_FORMAT, value);
+            taosMemoryFree(value);
+            ++i;
+          }
+
+          EXPLAIN_ROW_END();
+          QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        }
+
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_INTERVAL_VALUE_FORMAT, pInterpNode->interval, pInterpNode->intervalUnit);
+        EXPLAIN_ROW_END();
+
+        EXPLAIN_ROW_NEW(level + 1, EXPLAIN_TIMERANGE_FORMAT, pInterpNode->timeRange.skey, pInterpNode->timeRange.ekey);
+        EXPLAIN_ROW_END();
+        QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+
+        if (pInterpNode->node.pConditions) {
+          EXPLAIN_ROW_NEW(level + 1, EXPLAIN_FILTER_FORMAT);
+          QRY_ERR_RET(nodesNodeToSQL(pInterpNode->node.pConditions, tbuf + VARSTR_HEADER_SIZE,
+                                     TSDB_EXPLAIN_RESULT_ROW_SIZE, &tlen));
+          EXPLAIN_ROW_END();
+          QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level + 1));
+        }
+      }
+      break;
+    }    
     default:
       qError("not supported physical node type %d", pNode->type);
       return TSDB_CODE_QRY_APP_ERROR;

source/libs/executor/inc/executorimpl.h

@@ -680,7 +680,7 @@ typedef struct SJoinOperatorInfo {
   SSDataBlock       *pRight;
   int32_t            rightPos;
   SColumnInfo        rightCol;
-  SNode             *pOnCondition;
+  SNode             *pCondAfterMerge;
 } SJoinOperatorInfo;
 
 #define OPTR_IS_OPENED(_optr)  (((_optr)->status & OP_OPENED) == OP_OPENED)

source/libs/executor/src/executorimpl.c

@@ -2843,11 +2843,18 @@ int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t* order, int32_t* scan
 
 int32_t doPrepareScan(SOperatorInfo* pOperator, uint64_t uid, int64_t ts) {
   int32_t type = pOperator->operatorType;
+
+  pOperator->status = OP_OPENED;
+
   if (type == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN) {
     SStreamBlockScanInfo* pScanInfo = pOperator->info;
     pScanInfo->blockType = STREAM_INPUT__DATA_SCAN;
 
+    pScanInfo->pSnapshotReadOp->status = OP_OPENED;
+
     STableScanInfo* pInfo = pScanInfo->pSnapshotReadOp->info;
+    ASSERT(pInfo->scanMode == TABLE_SCAN__TABLE_ORDER);
+
     if (uid == 0) {
       pInfo->noTable = 1;
       return TSDB_CODE_SUCCESS;
@@ -2861,14 +2868,6 @@ int32_t doPrepareScan(SOperatorInfo* pOperator, uint64_t uid, int64_t ts) {
     pInfo->noTable = 0;
 
     if (pInfo->lastStatus.uid != uid || pInfo->lastStatus.ts != ts) {
-      tsdbSetTableId(pInfo->dataReader, uid);
-      int64_t oldSkey = pInfo->cond.twindows[0].skey;
-      pInfo->cond.twindows[0].skey = ts + 1;
-      tsdbResetReadHandle(pInfo->dataReader, &pInfo->cond, 0);
-      pInfo->cond.twindows[0].skey = oldSkey;
-      pInfo->scanTimes = 0;
-      pInfo->curTWinIdx = 0;
-
       SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
 
       int32_t tableSz = taosArrayGetSize(pTaskInfo->tableqinfoList.pTableList);
@@ -2880,8 +2879,17 @@ int32_t doPrepareScan(SOperatorInfo* pOperator, uint64_t uid, int64_t ts) {
           pInfo->currentTable = i;
         }
       }
-      // TODO after processing drop,
+      // TODO after processing drop, found can be false
       ASSERT(found);
+
+      tsdbSetTableId(pInfo->dataReader, uid);
+      int64_t oldSkey = pInfo->cond.twindows[0].skey;
+      pInfo->cond.twindows[0].skey = ts + 1;
+      tsdbResetReadHandle(pInfo->dataReader, &pInfo->cond, 0);
+      pInfo->cond.twindows[0].skey = oldSkey;
+      pInfo->scanTimes = 0;
+      pInfo->curTWinIdx = 0;
+
       qDebug("tsdb reader offset seek to uid %ld ts %ld, table cur set to %d , all table num %d", uid, ts,
              pInfo->currentTable, tableSz);
     }

source/libs/executor/src/joinoperator.c

@@ -53,13 +53,28 @@ SOperatorInfo* createMergeJoinOperatorInfo(SOperatorInfo** pDownstream, int32_t
   pOperator->info         = pInfo;
   pOperator->pTaskInfo    = pTaskInfo;
 
-  SNode* pOnCondition = pJoinNode->pOnConditions;
+  SNode* pMergeCondition = pJoinNode->pMergeCondition;
-  if (nodeType(pOnCondition) == QUERY_NODE_OPERATOR) {
+  if (nodeType(pMergeCondition) == QUERY_NODE_OPERATOR) {
-    SOperatorNode* pNode = (SOperatorNode*)pOnCondition;
+    SOperatorNode* pNode = (SOperatorNode*)pMergeCondition;
     setJoinColumnInfo(&pInfo->leftCol, (SColumnNode*)pNode->pLeft);
     setJoinColumnInfo(&pInfo->rightCol, (SColumnNode*)pNode->pRight);
-  } else if (nodeType(pOnCondition) == QUERY_NODE_LOGIC_CONDITION) {
+  } else {
-    extractTimeCondition(pInfo, (SLogicConditionNode*)pOnCondition);
+    ASSERT(false);
+  }
+
+  if (pJoinNode->pOnConditions != NULL && pJoinNode->node.pConditions != NULL) {
+    pInfo->pCondAfterMerge = nodesMakeNode(QUERY_NODE_LOGIC_CONDITION);
+    SLogicConditionNode* pLogicCond = (SLogicConditionNode*)(pInfo->pCondAfterMerge);
+    pLogicCond->pParameterList = nodesMakeList();
+    nodesListMakeAppend(&pLogicCond->pParameterList, nodesCloneNode(pJoinNode->pOnConditions));
+    nodesListMakeAppend(&pLogicCond->pParameterList, nodesCloneNode(pJoinNode->node.pConditions));
+    pLogicCond->condType = LOGIC_COND_TYPE_AND;
+  } else if (pJoinNode->pOnConditions != NULL) {
+    pInfo->pCondAfterMerge = nodesCloneNode(pJoinNode->pOnConditions);
+  } else if (pJoinNode->node.pConditions != NULL) {
+    pInfo->pCondAfterMerge = nodesCloneNode(pJoinNode->node.pConditions);
+  } else {
+    pInfo->pCondAfterMerge = NULL;
   }
 
   pOperator->fpSet =
@@ -88,15 +103,12 @@ void setJoinColumnInfo(SColumnInfo* pColumn, const SColumnNode* pColumnNode) {
 
 void destroyMergeJoinOperator(void* param, int32_t numOfOutput) {
   SJoinOperatorInfo* pJoinOperator = (SJoinOperatorInfo*)param;
+  nodesDestroyNode(pJoinOperator->pCondAfterMerge);
 }
 
-SSDataBlock* doMergeJoin(struct SOperatorInfo* pOperator) {
+static void doMergeJoinImpl(struct SOperatorInfo* pOperator, SSDataBlock* pRes) {
   SJoinOperatorInfo* pJoinInfo = pOperator->info;
 
-  SSDataBlock* pRes = pJoinInfo->pRes;
-  blockDataCleanup(pRes);
-  blockDataEnsureCapacity(pRes, 4096);
-
   int32_t nrows = 0;
 
   while (1) {
@@ -181,7 +193,28 @@ SSDataBlock* doMergeJoin(struct SOperatorInfo* pOperator) {
       break;
     }
   }
+}
 
+SSDataBlock* doMergeJoin(struct SOperatorInfo* pOperator) {
+  SJoinOperatorInfo* pJoinInfo = pOperator->info;
+
+  SSDataBlock* pRes = pJoinInfo->pRes;
+  blockDataCleanup(pRes);
+  blockDataEnsureCapacity(pRes, 4096);
+  while (true) {
+    int32_t numOfRowsBefore = pRes->info.rows;
+    doMergeJoinImpl(pOperator, pRes);
+    int32_t numOfNewRows = pRes->info.rows - numOfRowsBefore;
+    if (numOfNewRows == 0) {
+      break;
+    }
+    if (pJoinInfo->pCondAfterMerge != NULL) {
+      doFilter(pJoinInfo->pCondAfterMerge, pRes);
+    }
+    if (pRes->info.rows >= pOperator->resultInfo.threshold) {
+      break;
+    }
+  }
   return (pRes->info.rows > 0) ? pRes : NULL;
 }
 

source/libs/function/src/builtins.c

@@ -18,6 +18,7 @@
 #include "querynodes.h"
 #include "scalar.h"
 #include "taoserror.h"
+#include "cJSON.h"
 
 static int32_t buildFuncErrMsg(char* pErrBuf, int32_t len, int32_t errCode, const char* pFormat, ...) {
   va_list vArgList;
@@ -796,6 +797,165 @@ static int32_t translateLeastSQR(SFunctionNode* pFunc, char* pErrBuf, int32_t le
   return TSDB_CODE_SUCCESS;
 }
 
+typedef enum { UNKNOWN_BIN = 0, USER_INPUT_BIN, LINEAR_BIN, LOG_BIN } EHistoBinType;
+
+static int8_t validateHistogramBinType(char* binTypeStr) {
+  int8_t binType;
+  if (strcasecmp(binTypeStr, "user_input") == 0) {
+    binType = USER_INPUT_BIN;
+  } else if (strcasecmp(binTypeStr, "linear_bin") == 0) {
+    binType = LINEAR_BIN;
+  } else if (strcasecmp(binTypeStr, "log_bin") == 0) {
+    binType = LOG_BIN;
+  } else {
+    binType = UNKNOWN_BIN;
+  }
+
+  return binType;
+}
+
+static bool validateHistogramBinDesc(char* binDescStr, int8_t binType, char* errMsg, int32_t msgLen) {
+  const char *msg1 = "HISTOGRAM function requires four parameters";
+  const char *msg3 = "HISTOGRAM function invalid format for binDesc parameter";
+  const char *msg4 = "HISTOGRAM function binDesc parameter \"count\" should be in range [1, 1000]";
+  const char *msg5 = "HISTOGRAM function bin/parameter should be in range [-DBL_MAX, DBL_MAX]";
+  const char *msg6 = "HISTOGRAM function binDesc parameter \"width\" cannot be 0";
+  const char *msg7 = "HISTOGRAM function binDesc parameter \"start\" cannot be 0 with \"log_bin\" type";
+  const char *msg8 = "HISTOGRAM function binDesc parameter \"factor\" cannot be negative or equal to 0/1";
+
+  cJSON*  binDesc = cJSON_Parse(binDescStr);
+  int32_t numOfBins;
+  double* intervals;
+  if (cJSON_IsObject(binDesc)) { /* linaer/log bins */
+    int32_t numOfParams = cJSON_GetArraySize(binDesc);
+    int32_t startIndex;
+    if (numOfParams != 4) {
+      snprintf(errMsg, msgLen, "%s", msg1);
+      return false;
+    }
+
+    cJSON* start = cJSON_GetObjectItem(binDesc, "start");
+    cJSON* factor = cJSON_GetObjectItem(binDesc, "factor");
+    cJSON* width = cJSON_GetObjectItem(binDesc, "width");
+    cJSON* count = cJSON_GetObjectItem(binDesc, "count");
+    cJSON* infinity = cJSON_GetObjectItem(binDesc, "infinity");
+
+    if (!cJSON_IsNumber(start) || !cJSON_IsNumber(count) || !cJSON_IsBool(infinity)) {
+      snprintf(errMsg, msgLen, "%s", msg3);
+      return false;
+    }
+
+    if (count->valueint <= 0 || count->valueint > 1000) {  // limit count to 1000
+      snprintf(errMsg, msgLen, "%s", msg4);
+      return false;
+    }
+
+    if (isinf(start->valuedouble) || (width != NULL && isinf(width->valuedouble)) ||
+        (factor != NULL && isinf(factor->valuedouble)) || (count != NULL && isinf(count->valuedouble))) {
+      snprintf(errMsg, msgLen, "%s", msg5);
+      return false;
+    }
+
+    int32_t counter = (int32_t)count->valueint;
+    if (infinity->valueint == false) {
+      startIndex = 0;
+      numOfBins = counter + 1;
+    } else {
+      startIndex = 1;
+      numOfBins = counter + 3;
+    }
+
+    intervals = taosMemoryCalloc(numOfBins, sizeof(double));
+    if (cJSON_IsNumber(width) && factor == NULL && binType == LINEAR_BIN) {
+      // linear bin process
+      if (width->valuedouble == 0) {
+        snprintf(errMsg, msgLen, "%s", msg6);
+        taosMemoryFree(intervals);
+        return false;
+      }
+      for (int i = 0; i < counter + 1; ++i) {
+        intervals[startIndex] = start->valuedouble + i * width->valuedouble;
+        if (isinf(intervals[startIndex])) {
+          snprintf(errMsg, msgLen, "%s", msg5);
+          taosMemoryFree(intervals);
+          return false;
+        }
+        startIndex++;
+      }
+    } else if (cJSON_IsNumber(factor) && width == NULL && binType == LOG_BIN) {
+      // log bin process
+      if (start->valuedouble == 0) {
+        snprintf(errMsg, msgLen, "%s", msg7);
+        taosMemoryFree(intervals);
+        return false;
+      }
+      if (factor->valuedouble < 0 || factor->valuedouble == 0 || factor->valuedouble == 1) {
+        snprintf(errMsg, msgLen, "%s", msg8);
+        taosMemoryFree(intervals);
+        return false;
+      }
+      for (int i = 0; i < counter + 1; ++i) {
+        intervals[startIndex] = start->valuedouble * pow(factor->valuedouble, i * 1.0);
+        if (isinf(intervals[startIndex])) {
+          snprintf(errMsg, msgLen, "%s", msg5);
+          taosMemoryFree(intervals);
+          return false;
+        }
+        startIndex++;
+      }
+    } else {
+      snprintf(errMsg, msgLen, "%s", msg3);
+      taosMemoryFree(intervals);
+      return false;
+    }
+
+    if (infinity->valueint == true) {
+      intervals[0] = -INFINITY;
+      intervals[numOfBins - 1] = INFINITY;
+      // in case of desc bin orders, -inf/inf should be swapped
+      ASSERT(numOfBins >= 4);
+      if (intervals[1] > intervals[numOfBins - 2]) {
+        TSWAP(intervals[0], intervals[numOfBins - 1]);
+      }
+    }
+  } else if (cJSON_IsArray(binDesc)) { /* user input bins */
+    if (binType != USER_INPUT_BIN) {
+      snprintf(errMsg, msgLen, "%s", msg3);
+      return false;
+    }
+    numOfBins = cJSON_GetArraySize(binDesc);
+    intervals = taosMemoryCalloc(numOfBins, sizeof(double));
+    cJSON* bin = binDesc->child;
+    if (bin == NULL) {
+      snprintf(errMsg, msgLen, "%s", msg3);
+      taosMemoryFree(intervals);
+      return false;
+    }
+    int i = 0;
+    while (bin) {
+      intervals[i] = bin->valuedouble;
+      if (!cJSON_IsNumber(bin)) {
+        snprintf(errMsg, msgLen, "%s", msg3);
+        taosMemoryFree(intervals);
+        return false;
+      }
+      if (i != 0 && intervals[i] <= intervals[i - 1]) {
+        snprintf(errMsg, msgLen, "%s", msg3);
+        taosMemoryFree(intervals);
+        return false;
+      }
+      bin = bin->next;
+      i++;
+    }
+  } else {
+    snprintf(errMsg, msgLen, "%s", msg3);
+    return false;
+  }
+
+  taosMemoryFree(intervals);
+  return true;
+}
+
 static int32_t translateHistogram(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
   int32_t numOfParams = LIST_LENGTH(pFunc->pParameterList);
   if (4 != numOfParams) {
@@ -814,6 +974,8 @@ static int32_t translateHistogram(SFunctionNode* pFunc, char* pErrBuf, int32_t l
     return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
   }
 
+  int8_t binType;
+  char*  binDesc;
   for (int32_t i = 1; i < numOfParams; ++i) {
     SNode* pParamNode = nodesListGetNode(pFunc->pParameterList, i);
     if (QUERY_NODE_VALUE != nodeType(pParamNode)) {
@@ -824,6 +986,23 @@ static int32_t translateHistogram(SFunctionNode* pFunc, char* pErrBuf, int32_t l
 
     pValue->notReserved = true;
 
+    if (i == 1) {
+      binType = validateHistogramBinType(varDataVal(pValue->datum.p));
+      if (binType == UNKNOWN_BIN) {
+        return buildFuncErrMsg(pErrBuf, len, TSDB_CODE_FUNC_FUNTION_ERROR,
+                               "HISTOGRAM function binType parameter should be "
+                               "\"user_input\", \"log_bin\" or \"linear_bin\"");
+      }
+    }
+
+    if (i == 2) {
+      char errMsg[128] = {0};
+      binDesc = varDataVal(pValue->datum.p);
+      if (!validateHistogramBinDesc(binDesc, binType, errMsg, (int32_t)sizeof(errMsg))) {
+        return buildFuncErrMsg(pErrBuf, len, TSDB_CODE_FUNC_FUNTION_ERROR, errMsg);
+      }
+    }
+
     if (i == 3 && pValue->datum.i != 1 && pValue->datum.i != 0) {
         return buildFuncErrMsg(pErrBuf, len, TSDB_CODE_FUNC_FUNTION_ERROR,
                                "HISTOGRAM function normalized parameter should be 0/1");
@@ -853,6 +1032,8 @@ static int32_t translateHistogramImpl(SFunctionNode* pFunc, char* pErrBuf, int32
       return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
     }
 
+    int8_t binType;
+    char*  binDesc;
     for (int32_t i = 1; i < numOfParams; ++i) {
       SNode* pParamNode = nodesListGetNode(pFunc->pParameterList, i);
       if (QUERY_NODE_VALUE != nodeType(pParamNode)) {
@@ -863,6 +1044,23 @@ static int32_t translateHistogramImpl(SFunctionNode* pFunc, char* pErrBuf, int32
 
       pValue->notReserved = true;
 
+      if (i == 1) {
+        binType = validateHistogramBinType(varDataVal(pValue->datum.p));
+        if (binType == UNKNOWN_BIN) {
+          return buildFuncErrMsg(pErrBuf, len, TSDB_CODE_FUNC_FUNTION_ERROR,
+                                 "HISTOGRAM function binType parameter should be "
+                                 "\"user_input\", \"log_bin\" or \"linear_bin\"");
+        }
+      }
+
+      if (i == 2) {
+        char errMsg[128] = {0};
+        binDesc = varDataVal(pValue->datum.p);
+        if (!validateHistogramBinDesc(binDesc, binType, errMsg, (int32_t)sizeof(errMsg))) {
+          return buildFuncErrMsg(pErrBuf, len, TSDB_CODE_FUNC_FUNTION_ERROR, errMsg);
+        }
+      }
+
       if (i == 3 && pValue->datum.i != 1 && pValue->datum.i != 0) {
           return buildFuncErrMsg(pErrBuf, len, TSDB_CODE_FUNC_FUNTION_ERROR,
                                  "HISTOGRAM function normalized parameter should be 0/1");
@@ -2112,7 +2310,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "histogram",
     .type = FUNCTION_TYPE_HISTOGRAM,
-    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_FORBID_FILL_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_FORBID_FILL_FUNC,
     .translateFunc = translateHistogram,
     .getEnvFunc   = getHistogramFuncEnv,
     .initFunc     = histogramFunctionSetup,

source/libs/function/src/builtinsimpl.c

@@ -719,8 +719,10 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
     ASSERT(numOfElem >= 0);
 
     pAvgRes->count += numOfElem;
-    if (IS_INTEGER_TYPE(type)) {
+    if (IS_SIGNED_NUMERIC_TYPE(type)) {
       pAvgRes->sum.isum += pAgg->sum;
+    } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
+      pAvgRes->sum.usum += pAgg->sum;
     } else if (IS_FLOAT_TYPE(type)) {
       pAvgRes->sum.dsum += GET_DOUBLE_VAL((const char*)&(pAgg->sum));
     }
@@ -784,6 +786,64 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
         break;
       }
 
+      case TSDB_DATA_TYPE_UTINYINT: {
+        uint8_t* plist = (uint8_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.usum += plist[i];
+        }
+
+        break;
+      }
+
+      case TSDB_DATA_TYPE_USMALLINT: {
+        uint16_t* plist = (uint16_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.usum += plist[i];
+        }
+        break;
+      }
+
+      case TSDB_DATA_TYPE_UINT: {
+        uint32_t* plist = (uint32_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.usum += plist[i];
+        }
+
+        break;
+      }
+
+      case TSDB_DATA_TYPE_UBIGINT: {
+        uint64_t* plist = (uint64_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.usum += plist[i];
+        }
+        break;
+      }
+
       case TSDB_DATA_TYPE_FLOAT: {
         float* plist = (float*)pCol->pData;
         for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
@@ -825,8 +885,10 @@ _avg_over:
 
 static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
   pOutput->type = pInput->type;
-  if (IS_INTEGER_TYPE(pOutput->type)) {
+  if (IS_SIGNED_NUMERIC_TYPE(pOutput->type)) {
     pOutput->sum.isum += pInput->sum.isum;
+  } else if (IS_UNSIGNED_NUMERIC_TYPE(pOutput->type)) {
+    pOutput->sum.usum += pInput->sum.usum;
   } else {
     pOutput->sum.dsum += pInput->sum.dsum;
   }
@@ -900,6 +962,22 @@ int32_t avgInvertFunction(SqlFunctionCtx* pCtx) {
       LIST_AVG_N(pAvgRes->sum.isum, int64_t);
       break;
     }
+    case TSDB_DATA_TYPE_UTINYINT: {
+      LIST_AVG_N(pAvgRes->sum.usum, uint8_t);
+      break;
+    }
+    case TSDB_DATA_TYPE_USMALLINT: {
+      LIST_AVG_N(pAvgRes->sum.usum, uint16_t);
+      break;
+    }
+    case TSDB_DATA_TYPE_UINT: {
+      LIST_AVG_N(pAvgRes->sum.usum, uint32_t);
+      break;
+    }
+    case TSDB_DATA_TYPE_UBIGINT: {
+      LIST_AVG_N(pAvgRes->sum.usum, uint64_t);
+      break;
+    }
     case TSDB_DATA_TYPE_FLOAT: {
       LIST_AVG_N(pAvgRes->sum.dsum, float);
       break;
@@ -925,8 +1003,10 @@ int32_t avgCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx) {
   SResultRowEntryInfo* pSResInfo = GET_RES_INFO(pSourceCtx);
   SAvgRes*             pSBuf = GET_ROWCELL_INTERBUF(pSResInfo);
 
-  if (IS_INTEGER_TYPE(type)) {
+  if (IS_SIGNED_NUMERIC_TYPE(type)) {
     pDBuf->sum.isum += pSBuf->sum.isum;
+  } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
+    pDBuf->sum.usum += pSBuf->sum.usum;
   } else {
     pDBuf->sum.dsum += pSBuf->sum.dsum;
   }
@@ -941,8 +1021,10 @@ int32_t avgFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
   SAvgRes* pAvgRes = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
   int32_t  type = pAvgRes->type;
 
-  if (IS_INTEGER_TYPE(type)) {
+  if (IS_SIGNED_NUMERIC_TYPE(type)) {
     pAvgRes->result = pAvgRes->sum.isum / ((double)pAvgRes->count);
+  } else if (IS_UNSIGNED_NUMERIC_TYPE(type)) {
+    pAvgRes->result = pAvgRes->sum.usum / ((double)pAvgRes->count);
   } else {
     pAvgRes->result = pAvgRes->sum.dsum / ((double)pAvgRes->count);
   }

source/libs/function/src/functionMgt.c

@@ -260,7 +260,7 @@ bool fmIsSameInOutType(int32_t funcId) {
 }
 
 static int32_t getFuncInfo(SFunctionNode* pFunc) {
-  char msg[64] = {0};
+  char msg[128] = {0};
   return fmGetFuncInfo(pFunc, msg, sizeof(msg));
 }
 

source/libs/function/src/tudf.c

@@ -1565,6 +1565,10 @@ void constructUdfService(void *argsThread) {
   //TODO return value of uv_run
   uv_run(&udfc->uvLoop, UV_RUN_DEFAULT);
   uv_loop_close(&udfc->uvLoop);
+
+  uv_walk(&udfc->uvLoop, udfUdfdCloseWalkCb, NULL);
+  uv_run(&udfc->uvLoop, UV_RUN_DEFAULT);
+  uv_loop_close(&udfc->uvLoop);
 }
 
 int32_t udfcOpen() {

source/libs/nodes/src/nodesCloneFuncs.c

@@ -368,6 +368,7 @@ static int32_t logicScanCopy(const SScanLogicNode* pSrc, SScanLogicNode* pDst) {
 static int32_t logicJoinCopy(const SJoinLogicNode* pSrc, SJoinLogicNode* pDst) {
   COPY_BASE_OBJECT_FIELD(node, logicNodeCopy);
   COPY_SCALAR_FIELD(joinType);
+  CLONE_NODE_FIELD(pMergeCondition);
   CLONE_NODE_FIELD(pOnConditions);
   COPY_SCALAR_FIELD(isSingleTableJoin);
   return TSDB_CODE_SUCCESS;

source/libs/nodes/src/nodesCodeFuncs.c

@@ -1254,6 +1254,7 @@ static int32_t jsonToLogicPlan(const SJson* pJson, void* pObj) {
 
 static const char* jkJoinLogicPlanJoinType = "JoinType";
 static const char* jkJoinLogicPlanOnConditions = "OnConditions";
+static const char* jkJoinLogicPlanMergeCondition = "MergeConditions";
 
 static int32_t logicJoinNodeToJson(const void* pObj, SJson* pJson) {
   const SJoinLogicNode* pNode = (const SJoinLogicNode*)pObj;
@@ -1262,6 +1263,9 @@ static int32_t logicJoinNodeToJson(const void* pObj, SJson* pJson) {
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddIntegerToObject(pJson, jkJoinLogicPlanJoinType, pNode->joinType);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddObject(pJson, jkJoinLogicPlanMergeCondition, nodeToJson, pNode->pMergeCondition);
+  }
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddObject(pJson, jkJoinLogicPlanOnConditions, nodeToJson, pNode->pOnConditions);
   }
@@ -1617,6 +1621,7 @@ static int32_t jsonToPhysiProjectNode(const SJson* pJson, void* pObj) {
 }
 
 static const char* jkJoinPhysiPlanJoinType = "JoinType";
+static const char* jkJoinPhysiPlanMergeCondition = "MergeCondition";
 static const char* jkJoinPhysiPlanOnConditions = "OnConditions";
 static const char* jkJoinPhysiPlanTargets = "Targets";
 
@@ -1627,6 +1632,9 @@ static int32_t physiJoinNodeToJson(const void* pObj, SJson* pJson) {
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddIntegerToObject(pJson, jkJoinPhysiPlanJoinType, pNode->joinType);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = tjsonAddObject(pJson, jkJoinPhysiPlanMergeCondition, nodeToJson, pNode->pMergeCondition);
+  }
   if (TSDB_CODE_SUCCESS == code) {
     code = tjsonAddObject(pJson, jkJoinPhysiPlanOnConditions, nodeToJson, pNode->pOnConditions);
   }
@@ -1648,6 +1656,9 @@ static int32_t jsonToPhysiJoinNode(const SJson* pJson, void* pObj) {
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeObject(pJson, jkJoinPhysiPlanOnConditions, &pNode->pOnConditions);
   }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = jsonToNodeObject(pJson, jkJoinPhysiPlanMergeCondition, &pNode->pMergeCondition);
+  }
   if (TSDB_CODE_SUCCESS == code) {
     code = jsonToNodeList(pJson, jkJoinPhysiPlanTargets, &pNode->pTargets);
   }

source/libs/nodes/src/nodesTraverseFuncs.c

@@ -470,6 +470,9 @@ static EDealRes dispatchPhysiPlan(SNode* pNode, ETraversalOrder order, FNodeWalk
     case QUERY_NODE_PHYSICAL_PLAN_MERGE_JOIN: {
       SJoinPhysiNode* pJoin = (SJoinPhysiNode*)pNode;
       res = walkPhysiNode((SPhysiNode*)pNode, order, walker, pContext);
+      if (DEAL_RES_ERROR != res && DEAL_RES_END != res) {
+        res = walkPhysiPlan(pJoin->pMergeCondition, order, walker, pContext);
+      }
       if (DEAL_RES_ERROR != res && DEAL_RES_END != res) {
         res = walkPhysiPlan(pJoin->pOnConditions, order, walker, pContext);
       }

source/libs/nodes/src/nodesUtilFuncs.c

@@ -19,8 +19,8 @@
 #include "querynodes.h"
 #include "taos.h"
 #include "taoserror.h"
-#include "thash.h"
 #include "tdatablock.h"
+#include "thash.h"
 
 static SNode* makeNode(ENodeType type, size_t size) {
   SNode* p = taosMemoryCalloc(1, size);
@@ -718,6 +718,7 @@ void nodesDestroyNode(SNode* pNode) {
     case QUERY_NODE_LOGIC_PLAN_JOIN: {
       SJoinLogicNode* pLogicNode = (SJoinLogicNode*)pNode;
       destroyLogicNode((SLogicNode*)pLogicNode);
+      nodesDestroyNode(pLogicNode->pMergeCondition);
       nodesDestroyNode(pLogicNode->pOnConditions);
       break;
     }
@@ -828,6 +829,7 @@ void nodesDestroyNode(SNode* pNode) {
     case QUERY_NODE_PHYSICAL_PLAN_MERGE_JOIN: {
       SJoinPhysiNode* pPhyNode = (SJoinPhysiNode*)pNode;
       destroyPhysiNode((SPhysiNode*)pPhyNode);
+      nodesDestroyNode(pPhyNode->pMergeCondition);
       nodesDestroyNode(pPhyNode->pOnConditions);
       nodesDestroyList(pPhyNode->pTargets);
       break;
@@ -1493,17 +1495,54 @@ int32_t nodesCollectColumns(SSelectStmt* pSelect, ESqlClause clause, const char*
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t nodesCollectColumnsFromNode(SNode* node, const char* pTableAlias, ECollectColType type, SNodeList** pCols) {
+  if (NULL == pCols) {
+    return TSDB_CODE_FAILED;
+  }
+  SCollectColumnsCxt cxt = {
+      .errCode = TSDB_CODE_SUCCESS,
+      .pTableAlias = pTableAlias,
+      .collectType = type,
+      .pCols = (NULL == *pCols ? nodesMakeList() : *pCols),
+      .pColHash = taosHashInit(128, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_NO_LOCK)};
+  if (NULL == cxt.pCols || NULL == cxt.pColHash) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+  *pCols = NULL;
+
+  nodesWalkExpr(node, collectColumns, &cxt);
+
+  taosHashCleanup(cxt.pColHash);
+  if (TSDB_CODE_SUCCESS != cxt.errCode) {
+    nodesDestroyList(cxt.pCols);
+    return cxt.errCode;
+  }
+  if (LIST_LENGTH(cxt.pCols) > 0) {
+    *pCols = cxt.pCols;
+  } else {
+    nodesDestroyList(cxt.pCols);
+  }
+
+  return TSDB_CODE_SUCCESS;
+
+}
+
 typedef struct SCollectFuncsCxt {
   int32_t         errCode;
   FFuncClassifier classifier;
   SNodeList*      pFuncs;
+  SHashObj*       pAliasName;
 } SCollectFuncsCxt;
 
 static EDealRes collectFuncs(SNode* pNode, void* pContext) {
   SCollectFuncsCxt* pCxt = (SCollectFuncsCxt*)pContext;
   if (QUERY_NODE_FUNCTION == nodeType(pNode) && pCxt->classifier(((SFunctionNode*)pNode)->funcId) &&
       !(((SExprNode*)pNode)->orderAlias)) {
-    pCxt->errCode = nodesListStrictAppend(pCxt->pFuncs, nodesCloneNode(pNode));
+    SExprNode* pExpr = (SExprNode*)pNode;
+    if (NULL == taosHashGet(pCxt->pAliasName, pExpr->aliasName, strlen(pExpr->aliasName))) {
+      pCxt->errCode = nodesListStrictAppend(pCxt->pFuncs, nodesCloneNode(pNode));
+      taosHashPut(pCxt->pAliasName, pExpr->aliasName, strlen(pExpr->aliasName), &pExpr, POINTER_BYTES);
+    }
     return (TSDB_CODE_SUCCESS == pCxt->errCode ? DEAL_RES_IGNORE_CHILD : DEAL_RES_ERROR);
   }
   return DEAL_RES_CONTINUE;
@@ -1515,23 +1554,27 @@ int32_t nodesCollectFuncs(SSelectStmt* pSelect, ESqlClause clause, FFuncClassifi
   }
 
   SCollectFuncsCxt cxt = {
-      .errCode = TSDB_CODE_SUCCESS, .classifier = classifier, .pFuncs = (NULL == *pFuncs ? nodesMakeList() : *pFuncs)};
+      .errCode = TSDB_CODE_SUCCESS,
+      .classifier = classifier,
+      .pFuncs = (NULL == *pFuncs ? nodesMakeList() : *pFuncs),
+      .pAliasName = taosHashInit(4, taosGetDefaultHashFunction(TSDB_DATA_TYPE_VARCHAR), false, false)};
   if (NULL == cxt.pFuncs) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
   *pFuncs = NULL;
   nodesWalkSelectStmt(pSelect, clause, collectFuncs, &cxt);
-  if (TSDB_CODE_SUCCESS != cxt.errCode) {
+  if (TSDB_CODE_SUCCESS == cxt.errCode) {
-    nodesDestroyList(cxt.pFuncs);
+    if (LIST_LENGTH(cxt.pFuncs) > 0) {
-    return cxt.errCode;
+      *pFuncs = cxt.pFuncs;
-  }
+    } else {
-  if (LIST_LENGTH(cxt.pFuncs) > 0) {
+      nodesDestroyList(cxt.pFuncs);
-    *pFuncs = cxt.pFuncs;
+    }
   } else {
     nodesDestroyList(cxt.pFuncs);
   }
+  taosHashCleanup(cxt.pAliasName);
 
-  return TSDB_CODE_SUCCESS;
+  return cxt.errCode;
 }
 
 typedef struct SCollectSpecialNodesCxt {

source/libs/parser/inc/parAst.h

@@ -154,6 +154,7 @@ SNode* createAlterTableDropCol(SAstCreateContext* pCxt, SNode* pRealTable, int8_
 SNode* createAlterTableRenameCol(SAstCreateContext* pCxt, SNode* pRealTable, int8_t alterType, SToken* pOldColName,
                                  SToken* pNewColName);
 SNode* createAlterTableSetTag(SAstCreateContext* pCxt, SNode* pRealTable, SToken* pTagName, SNode* pVal);
+SNode* setAlterSuperTableType(SNode* pStmt);
 SNode* createUseDatabaseStmt(SAstCreateContext* pCxt, SToken* pDbName);
 SNode* createShowStmt(SAstCreateContext* pCxt, ENodeType type);
 SNode* createShowStmtWithCond(SAstCreateContext* pCxt, ENodeType type, SNode* pDbName, SNode* pTbName,

source/libs/parser/inc/parUtil.h

@@ -53,6 +53,7 @@ typedef struct SParseMetaCache {
 } SParseMetaCache;
 
 int32_t generateSyntaxErrMsg(SMsgBuf* pBuf, int32_t errCode, ...);
+int32_t generateSyntaxErrMsgExt(SMsgBuf* pBuf, int32_t errCode, const char* pFormat, ...);
 int32_t buildInvalidOperationMsg(SMsgBuf* pMsgBuf, const char* msg);
 int32_t buildSyntaxErrMsg(SMsgBuf* pBuf, const char* additionalInfo, const char* sourceStr);
 

source/libs/parser/inc/sql.y

@@ -232,7 +232,7 @@ cmd ::= DROP TABLE multi_drop_clause(A).
 cmd ::= DROP STABLE exists_opt(A) full_table_name(B).                             { pCxt->pRootNode = createDropSuperTableStmt(pCxt, A, B); }
 
 cmd ::= ALTER TABLE alter_table_clause(A).                                        { pCxt->pRootNode = A; }
-cmd ::= ALTER STABLE alter_table_clause(A).                                       { pCxt->pRootNode = A; }
+cmd ::= ALTER STABLE alter_table_clause(A).                                       { pCxt->pRootNode = setAlterSuperTableType(A); }
 
 alter_table_clause(A) ::= full_table_name(B) alter_table_options(C).              { A = createAlterTableModifyOptions(pCxt, B, C); }
 alter_table_clause(A) ::=
@@ -259,7 +259,7 @@ multi_create_clause(A) ::= multi_create_clause(B) create_subtable_clause(C).
 
 create_subtable_clause(A) ::=
   not_exists_opt(B) full_table_name(C) USING full_table_name(D)
-  specific_tags_opt(E) TAGS NK_LP literal_list(F) NK_RP table_options(G).         { A = createCreateSubTableClause(pCxt, B, C, D, E, F, G); }
+  specific_tags_opt(E) TAGS NK_LP expression_list(F) NK_RP table_options(G).      { A = createCreateSubTableClause(pCxt, B, C, D, E, F, G); }
 
 %type multi_drop_clause                                                           { SNodeList* }
 %destructor multi_drop_clause                                                     { nodesDestroyList($$); }

source/libs/parser/src/parAstCreater.c

@@ -1127,6 +1127,11 @@ SNode* createAlterTableSetTag(SAstCreateContext* pCxt, SNode* pRealTable, SToken
   return createAlterTableStmtFinalize(pRealTable, pStmt);
 }
 
+SNode* setAlterSuperTableType(SNode* pStmt) {
+  setNodeType(pStmt, QUERY_NODE_ALTER_SUPER_TABLE_STMT);
+  return pStmt;
+}
+
 SNode* createUseDatabaseStmt(SAstCreateContext* pCxt, SToken* pDbName) {
   CHECK_PARSER_STATUS(pCxt);
   if (!checkDbName(pCxt, pDbName, false)) {

source/libs/parser/src/parAstParser.c

@@ -247,6 +247,10 @@ static int32_t collectMetaKeyFromAlterTable(SCollectMetaKeyCxt* pCxt, SAlterTabl
   return code;
 }
 
+static int32_t collectMetaKeyFromAlterStable(SCollectMetaKeyCxt* pCxt, SAlterTableStmt* pStmt) {
+  return reserveTableMetaInCache(pCxt->pParseCxt->acctId, pStmt->dbName, pStmt->tableName, pCxt->pMetaCache);
+}
+
 static int32_t collectMetaKeyFromUseDatabase(SCollectMetaKeyCxt* pCxt, SUseDatabaseStmt* pStmt) {
   return reserveDbVgVersionInCache(pCxt->pParseCxt->acctId, pStmt->dbName, pCxt->pMetaCache);
 }
@@ -483,6 +487,8 @@ static int32_t collectMetaKeyFromQuery(SCollectMetaKeyCxt* pCxt, SNode* pStmt) {
       return collectMetaKeyFromDropTable(pCxt, (SDropTableStmt*)pStmt);
     case QUERY_NODE_ALTER_TABLE_STMT:
       return collectMetaKeyFromAlterTable(pCxt, (SAlterTableStmt*)pStmt);
+    case QUERY_NODE_ALTER_SUPER_TABLE_STMT:
+      return collectMetaKeyFromAlterStable(pCxt, (SAlterTableStmt*)pStmt);
     case QUERY_NODE_USE_DATABASE_STMT:
       return collectMetaKeyFromUseDatabase(pCxt, (SUseDatabaseStmt*)pStmt);
     case QUERY_NODE_CREATE_INDEX_STMT:

source/libs/parser/src/parInsert.c

@@ -48,6 +48,12 @@
     pSql += sToken.n;                         \
   } while (TK_NK_SPACE == sToken.type)
 
+typedef struct SInsertParseBaseContext {
+  SParseContext* pComCxt;
+  char*          pSql;
+  SMsgBuf        msg;
+} SInsertParseBaseContext;
+
 typedef struct SInsertParseContext {
   SParseContext*     pComCxt;             // input
   char*              pSql;                // input
@@ -1105,6 +1111,32 @@ static int32_t storeTableMeta(SInsertParseContext* pCxt, SHashObj* pHash, SName*
   return taosHashPut(pHash, pName, len, &pBackup, POINTER_BYTES);
 }
 
+static int32_t skipParentheses(SInsertParseSyntaxCxt* pCxt) {
+  SToken  sToken;
+  int32_t expectRightParenthesis = 1;
+  while (1) {
+    NEXT_TOKEN(pCxt->pSql, sToken);
+    if (TK_NK_LP == sToken.type) {
+      ++expectRightParenthesis;
+    } else if (TK_NK_RP == sToken.type && 0 == --expectRightParenthesis) {
+      break;
+    }
+    if (0 == sToken.n) {
+      return buildSyntaxErrMsg(&pCxt->msg, ") expected", NULL);
+    }
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t skipBoundColumns(SInsertParseSyntaxCxt* pCxt) { return skipParentheses(pCxt); }
+
+static int32_t ignoreBoundColumns(SInsertParseContext* pCxt) {
+  SInsertParseSyntaxCxt cxt = {.pComCxt = pCxt->pComCxt, .pSql = pCxt->pSql, .msg = pCxt->msg, .pMetaCache = NULL};
+  int32_t               code = skipBoundColumns(&cxt);
+  pCxt->pSql = cxt.pSql;
+  return code;
+}
+
 static int32_t skipUsingClause(SInsertParseSyntaxCxt* pCxt);
 
 // pSql -> stb_name [(tag1_name, ...)] TAGS (tag1_value, ...)
@@ -1453,12 +1485,29 @@ static int32_t parseInsertBody(SInsertParseContext* pCxt) {
     tNameGetFullDbName(&name, dbFName);
     CHECK_CODE(taosHashPut(pCxt->pDbFNameHashObj, dbFName, strlen(dbFName), dbFName, sizeof(dbFName)));
 
+    bool existedUsing = false;
     // USING clause
+    if (TK_USING == sToken.type) {
+      existedUsing = true;
+      CHECK_CODE(parseUsingClause(pCxt, &name, tbFName));
+      NEXT_TOKEN(pCxt->pSql, sToken);
+      autoCreateTbl = true;
+    }
+
+    char* pBoundColsStart = NULL;
+    if (TK_NK_LP == sToken.type) {
+      // pSql -> field1_name, ...)
+      pBoundColsStart = pCxt->pSql;
+      CHECK_CODE(ignoreBoundColumns(pCxt));
+      // CHECK_CODE(parseBoundColumns(pCxt, &dataBuf->boundColumnInfo, getTableColumnSchema(pCxt->pTableMeta)));
+      NEXT_TOKEN(pCxt->pSql, sToken);
+    }
+
     if (TK_USING == sToken.type) {
       CHECK_CODE(parseUsingClause(pCxt, &name, tbFName));
       NEXT_TOKEN(pCxt->pSql, sToken);
       autoCreateTbl = true;
-    } else {
+    } else if (!existedUsing) {
       CHECK_CODE(getTableMeta(pCxt, &name, dbFName));
     }
 
@@ -1467,10 +1516,11 @@ static int32_t parseInsertBody(SInsertParseContext* pCxt) {
                                     sizeof(SSubmitBlk), getTableInfo(pCxt->pTableMeta).rowSize, pCxt->pTableMeta,
                                     &dataBuf, NULL, &pCxt->createTblReq));
 
-    if (TK_NK_LP == sToken.type) {
+    if (NULL != pBoundColsStart) {
-      // pSql -> field1_name, ...)
+      char* pCurrPos = pCxt->pSql;
+      pCxt->pSql = pBoundColsStart;
       CHECK_CODE(parseBoundColumns(pCxt, &dataBuf->boundColumnInfo, getTableColumnSchema(pCxt->pTableMeta)));
-      NEXT_TOKEN(pCxt->pSql, sToken);
+      pCxt->pSql = pCurrPos;
     }
 
     if (TK_VALUES == sToken.type) {
@@ -1610,25 +1660,6 @@ int32_t parseInsertSql(SParseContext* pContext, SQuery** pQuery, SParseMetaCache
   return code;
 }
 
-static int32_t skipParentheses(SInsertParseSyntaxCxt* pCxt) {
-  SToken  sToken;
-  int32_t expectRightParenthesis = 1;
-  while (1) {
-    NEXT_TOKEN(pCxt->pSql, sToken);
-    if (TK_NK_LP == sToken.type) {
-      ++expectRightParenthesis;
-    } else if (TK_NK_RP == sToken.type && 0 == --expectRightParenthesis) {
-      break;
-    }
-    if (0 == sToken.n) {
-      return buildSyntaxErrMsg(&pCxt->msg, ") expected", NULL);
-    }
-  }
-  return TSDB_CODE_SUCCESS;
-}
-
-static int32_t skipBoundColumns(SInsertParseSyntaxCxt* pCxt) { return skipParentheses(pCxt); }
-
 // pSql -> (field1_value, ...) [(field1_value2, ...) ...]
 static int32_t skipValuesClause(SInsertParseSyntaxCxt* pCxt) {
   int32_t numOfRows = 0;
@@ -1717,8 +1748,25 @@ static int32_t parseInsertBodySyntax(SInsertParseSyntaxCxt* pCxt) {
     SToken tbnameToken = sToken;
     NEXT_TOKEN(pCxt->pSql, sToken);
 
+    bool existedUsing = false;
     // USING clause
     if (TK_USING == sToken.type) {
+      existedUsing = true;
+      CHECK_CODE(collectAutoCreateTableMetaKey(pCxt, &tbnameToken));
+      NEXT_TOKEN(pCxt->pSql, sToken);
+      CHECK_CODE(collectTableMetaKey(pCxt, &sToken));
+      CHECK_CODE(skipUsingClause(pCxt));
+      NEXT_TOKEN(pCxt->pSql, sToken);
+    }
+
+    if (TK_NK_LP == sToken.type) {
+      // pSql -> field1_name, ...)
+      CHECK_CODE(skipBoundColumns(pCxt));
+      NEXT_TOKEN(pCxt->pSql, sToken);
+    }
+
+    if (TK_USING == sToken.type && !existedUsing) {
+      existedUsing = true;
       CHECK_CODE(collectAutoCreateTableMetaKey(pCxt, &tbnameToken));
       NEXT_TOKEN(pCxt->pSql, sToken);
       CHECK_CODE(collectTableMetaKey(pCxt, &sToken));
@@ -1728,12 +1776,6 @@ static int32_t parseInsertBodySyntax(SInsertParseSyntaxCxt* pCxt) {
       CHECK_CODE(collectTableMetaKey(pCxt, &tbnameToken));
     }
 
-    if (TK_NK_LP == sToken.type) {
-      // pSql -> field1_name, ...)
-      CHECK_CODE(skipBoundColumns(pCxt));
-      NEXT_TOKEN(pCxt->pSql, sToken);
-    }
-
     if (TK_VALUES == sToken.type) {
       // pSql -> (field1_value, ...) [(field1_value2, ...) ...]
       CHECK_CODE(skipValuesClause(pCxt));

source/libs/parser/src/parUtil.c

@@ -215,13 +215,21 @@ int32_t generateSyntaxErrMsg(SMsgBuf* pBuf, int32_t errCode, ...) {
   return errCode;
 }
 
+int32_t generateSyntaxErrMsgExt(SMsgBuf* pBuf, int32_t errCode, const char* pFormat, ...) {
+  va_list vArgList;
+  va_start(vArgList, pFormat);
+  vsnprintf(pBuf->buf, pBuf->len, pFormat, vArgList);
+  va_end(vArgList);
+  return errCode;
+}
+
 int32_t buildInvalidOperationMsg(SMsgBuf* pBuf, const char* msg) {
   strncpy(pBuf->buf, msg, pBuf->len);
   return TSDB_CODE_TSC_INVALID_OPERATION;
 }
 
 int32_t buildSyntaxErrMsg(SMsgBuf* pBuf, const char* additionalInfo, const char* sourceStr) {
-  if(pBuf == NULL) return TSDB_CODE_TSC_SQL_SYNTAX_ERROR;
+  if (pBuf == NULL) return TSDB_CODE_TSC_SQL_SYNTAX_ERROR;
   const char* msgFormat1 = "syntax error near \'%s\'";
   const char* msgFormat2 = "syntax error near \'%s\' (%s)";
   const char* msgFormat3 = "%s";

source/libs/parser/test/mockCatalog.cpp

@@ -159,7 +159,7 @@ void generatePerformanceSchema(MockCatalogService* mcs) {
  *          c4         |       column       |       DOUBLE       |    8     |
  *          c5         |       column       |       DOUBLE       |    8     |
  */
-void generateTestT1(MockCatalogService* mcs) {
+void generateTestTables(MockCatalogService* mcs) {
   ITableBuilder& builder = mcs->createTableBuilder("test", "t1", TSDB_NORMAL_TABLE, 6)
                                .setPrecision(TSDB_TIME_PRECISION_MILLI)
                                .setVgid(1)
@@ -183,23 +183,7 @@ void generateTestT1(MockCatalogService* mcs) {
  *         tag2        |        tag         |      VARCHAR       |    20    |
  *         tag3        |        tag         |     TIMESTAMP      |    8     |
  * Child Table: st1s1, st1s2
- */
+ *
-void generateTestST1(MockCatalogService* mcs) {
-  ITableBuilder& builder = mcs->createTableBuilder("test", "st1", TSDB_SUPER_TABLE, 3, 3)
-                               .setPrecision(TSDB_TIME_PRECISION_MILLI)
-                               .addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
-                               .addColumn("c1", TSDB_DATA_TYPE_INT)
-                               .addColumn("c2", TSDB_DATA_TYPE_BINARY, 20)
-                               .addTag("tag1", TSDB_DATA_TYPE_INT)
-                               .addTag("tag2", TSDB_DATA_TYPE_BINARY, 20)
-                               .addTag("tag3", TSDB_DATA_TYPE_TIMESTAMP);
-  builder.done();
-  mcs->createSubTable("test", "st1", "st1s1", 1);
-  mcs->createSubTable("test", "st1", "st1s2", 2);
-  mcs->createSubTable("test", "st1", "st1s3", 1);
-}
-
-/*
  * Super Table: st2
  *        Field        |        Type        |      DataType      |  Bytes   |
  * ==========================================================================
@@ -209,16 +193,32 @@ void generateTestST1(MockCatalogService* mcs) {
  *         jtag        |        tag         |        json        |    --    |
  * Child Table: st2s1, st2s2
  */
-void generateTestST2(MockCatalogService* mcs) {
+void generateTestStables(MockCatalogService* mcs) {
-  ITableBuilder& builder = mcs->createTableBuilder("test", "st2", TSDB_SUPER_TABLE, 3, 1)
+  {
-                               .setPrecision(TSDB_TIME_PRECISION_MILLI)
+    ITableBuilder& builder = mcs->createTableBuilder("test", "st1", TSDB_SUPER_TABLE, 3, 3)
-                               .addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
+                                 .setPrecision(TSDB_TIME_PRECISION_MILLI)
-                               .addColumn("c1", TSDB_DATA_TYPE_INT)
+                                 .addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
-                               .addColumn("c2", TSDB_DATA_TYPE_BINARY, 20)
+                                 .addColumn("c1", TSDB_DATA_TYPE_INT)
-                               .addTag("jtag", TSDB_DATA_TYPE_JSON);
+                                 .addColumn("c2", TSDB_DATA_TYPE_BINARY, 20)
-  builder.done();
+                                 .addTag("tag1", TSDB_DATA_TYPE_INT)
-  mcs->createSubTable("test", "st2", "st2s1", 1);
+                                 .addTag("tag2", TSDB_DATA_TYPE_BINARY, 20)
-  mcs->createSubTable("test", "st2", "st2s2", 2);
+                                 .addTag("tag3", TSDB_DATA_TYPE_TIMESTAMP);
+    builder.done();
+    mcs->createSubTable("test", "st1", "st1s1", 1);
+    mcs->createSubTable("test", "st1", "st1s2", 2);
+    mcs->createSubTable("test", "st1", "st1s3", 1);
+  }
+  {
+    ITableBuilder& builder = mcs->createTableBuilder("test", "st2", TSDB_SUPER_TABLE, 3, 1)
+                                 .setPrecision(TSDB_TIME_PRECISION_MILLI)
+                                 .addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
+                                 .addColumn("c1", TSDB_DATA_TYPE_INT)
+                                 .addColumn("c2", TSDB_DATA_TYPE_BINARY, 20)
+                                 .addTag("jtag", TSDB_DATA_TYPE_JSON);
+    builder.done();
+    mcs->createSubTable("test", "st2", "st2s1", 1);
+    mcs->createSubTable("test", "st2", "st2s2", 2);
+  }
 }
 
 void generateFunctions(MockCatalogService* mcs) {
@@ -233,6 +233,11 @@ void generateDnodes(MockCatalogService* mcs) {
   mcs->createDnode(3, "host3", 7030);
 }
 
+void generateDatabases(MockCatalogService* mcs) {
+  mcs->createDatabase("test");
+  mcs->createDatabase("rollup_db", true);
+}
+
 }  // namespace
 
 int32_t __catalogGetHandle(const char* clusterId, struct SCatalog** catalogHandle) { return 0; }
@@ -262,7 +267,7 @@ int32_t __catalogGetDBVgInfo(SCatalog* pCtg, SRequestConnInfo* pConn, const char
 }
 
 int32_t __catalogGetDBCfg(SCatalog* pCtg, SRequestConnInfo* pConn, const char* dbFName, SDbCfgInfo* pDbCfg) {
-  return 0;
+  return g_mockCatalogService->catalogGetDBCfg(dbFName, pDbCfg);
 }
 
 int32_t __catalogChkAuth(SCatalog* pCtg, SRequestConnInfo* pConn, const char* user, const char* dbFName, AUTH_TYPE type,
@@ -359,11 +364,11 @@ void initMetaDataEnv() {
 }
 
 void generateMetaData() {
+  generateDatabases(g_mockCatalogService.get());
   generateInformationSchema(g_mockCatalogService.get());
   generatePerformanceSchema(g_mockCatalogService.get());
-  generateTestT1(g_mockCatalogService.get());
+  generateTestTables(g_mockCatalogService.get());
-  generateTestST1(g_mockCatalogService.get());
+  generateTestStables(g_mockCatalogService.get());
-  generateTestST2(g_mockCatalogService.get());
   generateFunctions(g_mockCatalogService.get());
   generateDnodes(g_mockCatalogService.get());
   g_mockCatalogService->showTables();

source/libs/parser/test/mockCatalogService.cpp

@@ -140,6 +140,17 @@ class MockCatalogServiceImpl {
     return TSDB_CODE_SUCCESS;
   }
 
+  int32_t catalogGetDBCfg(const char* pDbFName, SDbCfgInfo* pDbCfg) const {
+    std::string                dbFName(pDbFName);
+    DbCfgCache::const_iterator it = dbCfg_.find(dbFName.substr(std::string(pDbFName).find_last_of('.') + 1));
+    if (dbCfg_.end() == it) {
+      return TSDB_CODE_FAILED;
+    }
+
+    memcpy(pDbCfg, &(it->second), sizeof(SDbCfgInfo));
+    return TSDB_CODE_SUCCESS;
+  }
+
   int32_t catalogGetUdfInfo(const std::string& funcName, SFuncInfo* pInfo) const {
     auto it = udf_.find(funcName);
     if (udf_.end() == it) {
@@ -323,12 +334,21 @@ class MockCatalogServiceImpl {
     dnode_.insert(std::make_pair(dnodeId, epSet));
   }
 
+  void createDatabase(const std::string& db, bool rollup) {
+    SDbCfgInfo cfg = {0};
+    if (rollup) {
+      cfg.pRetensions = taosArrayInit(TARRAY_MIN_SIZE, sizeof(SRetention));
+    }
+    dbCfg_.insert(std::make_pair(db, cfg));
+  }
+
  private:
   typedef std::map<std::string, std::shared_ptr<MockTableMeta>> TableMetaCache;
   typedef std::map<std::string, TableMetaCache>                 DbMetaCache;
   typedef std::map<std::string, std::shared_ptr<SFuncInfo>>     UdfMetaCache;
   typedef std::map<std::string, std::vector<STableIndexInfo>>   IndexMetaCache;
   typedef std::map<int32_t, SEpSet>                             DnodeCache;
+  typedef std::map<std::string, SDbCfgInfo>                     DbCfgCache;
 
   uint64_t getNextId() { return id_++; }
 
@@ -486,6 +506,7 @@ class MockCatalogServiceImpl {
       for (int32_t i = 0; i < ndbs; ++i) {
         SMetaRes res = {0};
         res.pRes = taosMemoryCalloc(1, sizeof(SDbCfgInfo));
+        res.code = catalogGetDBCfg((const char*)taosArrayGet(pDbCfgReq, i), (SDbCfgInfo*)res.pRes);
         taosArrayPush(*pDbCfgData, &res);
       }
     }
@@ -576,6 +597,7 @@ class MockCatalogServiceImpl {
   UdfMetaCache                  udf_;
   IndexMetaCache                index_;
   DnodeCache                    dnode_;
+  DbCfgCache                    dbCfg_;
 };
 
 MockCatalogService::MockCatalogService() : impl_(new MockCatalogServiceImpl()) {}
@@ -605,6 +627,8 @@ void MockCatalogService::createDnode(int32_t dnodeId, const std::string& host, i
   impl_->createDnode(dnodeId, host, port);
 }
 
+void MockCatalogService::createDatabase(const std::string& db, bool rollup) { impl_->createDatabase(db, rollup); }
+
 int32_t MockCatalogService::catalogGetTableMeta(const SName* pTableName, STableMeta** pTableMeta) const {
   return impl_->catalogGetTableMeta(pTableName, pTableMeta);
 }
@@ -621,6 +645,10 @@ int32_t MockCatalogService::catalogGetDBVgInfo(const char* pDbFName, SArray** pV
   return impl_->catalogGetDBVgInfo(pDbFName, pVgList);
 }
 
+int32_t MockCatalogService::catalogGetDBCfg(const char* pDbFName, SDbCfgInfo* pDbCfg) const {
+  return impl_->catalogGetDBCfg(pDbFName, pDbCfg);
+}
+
 int32_t MockCatalogService::catalogGetUdfInfo(const std::string& funcName, SFuncInfo* pInfo) const {
   return impl_->catalogGetUdfInfo(funcName, pInfo);
 }

source/libs/parser/test/mockCatalogService.h

@@ -63,11 +63,13 @@ class MockCatalogService {
   void createFunction(const std::string& func, int8_t funcType, int8_t outputType, int32_t outputLen, int32_t bufSize);
   void createSmaIndex(const SMCreateSmaReq* pReq);
   void createDnode(int32_t dnodeId, const std::string& host, int16_t port);
+  void createDatabase(const std::string& db, bool rollup = false);
 
   int32_t catalogGetTableMeta(const SName* pTableName, STableMeta** pTableMeta) const;
   int32_t catalogGetTableHashVgroup(const SName* pTableName, SVgroupInfo* vgInfo) const;
   int32_t catalogGetTableDistVgInfo(const SName* pTableName, SArray** pVgList) const;
   int32_t catalogGetDBVgInfo(const char* pDbFName, SArray** pVgList) const;
+  int32_t catalogGetDBCfg(const char* pDbFName, SDbCfgInfo* pDbCfg) const;
   int32_t catalogGetUdfInfo(const std::string& funcName, SFuncInfo* pInfo) const;
   int32_t catalogGetTableIndex(const SName* pTableName, SArray** pIndexes) const;
   int32_t catalogGetDnodeList(SArray** pDnodes) const;

source/libs/parser/test/parInitialATest.cpp

@@ -38,9 +38,9 @@ TEST_F(ParserInitialATest, alterDnode) {
 TEST_F(ParserInitialATest, alterDatabase) {
   useDb("root", "test");
 
-  run("ALTER DATABASE wxy_db CACHELAST 1 FSYNC 200 WAL 1");
+  run("ALTER DATABASE test CACHELAST 1 FSYNC 200 WAL 1");
 
-  run("ALTER DATABASE wxy_db KEEP 2400");
+  run("ALTER DATABASE test KEEP 2400");
 }
 
 TEST_F(ParserInitialATest, alterLocal) {
@@ -77,8 +77,6 @@ TEST_F(ParserInitialATest, alterLocal) {
   clearAlterLocal();
 }
 
-// todo ALTER stable
-
 /*
  * ALTER TABLE [db_name.]tb_name alter_table_clause
  *
@@ -157,7 +155,7 @@ TEST_F(ParserInitialATest, alterSTable) {
   };
 
   setCheckDdlFunc([&](const SQuery* pQuery, ParserStage stage) {
-    ASSERT_EQ(nodeType(pQuery->pRoot), QUERY_NODE_ALTER_TABLE_STMT);
+    ASSERT_EQ(nodeType(pQuery->pRoot), QUERY_NODE_ALTER_SUPER_TABLE_STMT);
     SMAlterStbReq req = {0};
     ASSERT_EQ(tDeserializeSMAlterStbReq(pQuery->pCmdMsg->pMsg, pQuery->pCmdMsg->msgLen, &req), TSDB_CODE_SUCCESS);
     ASSERT_EQ(std::string(req.name), std::string(expect.name));
@@ -181,44 +179,44 @@ TEST_F(ParserInitialATest, alterSTable) {
   });
 
   //  setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_OPTIONS, 0, nullptr, 0, 0, nullptr, nullptr, 10);
-  //  run("ALTER TABLE st1 TTL 10");
+  //  run("ALTER STABLE st1 TTL 10");
   //  clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_OPTIONS, 0, nullptr, 0, 0, nullptr, "test");
-  run("ALTER TABLE st1 COMMENT 'test'");
+  run("ALTER STABLE st1 COMMENT 'test'");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_ADD_COLUMN, 1, "cc1", TSDB_DATA_TYPE_BIGINT);
-  run("ALTER TABLE st1 ADD COLUMN cc1 BIGINT");
+  run("ALTER STABLE st1 ADD COLUMN cc1 BIGINT");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_DROP_COLUMN, 1, "c1");
-  run("ALTER TABLE st1 DROP COLUMN c1");
+  run("ALTER STABLE st1 DROP COLUMN c1");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_COLUMN_BYTES, 1, "c2", TSDB_DATA_TYPE_VARCHAR,
                      30 + VARSTR_HEADER_SIZE);
-  run("ALTER TABLE st1 MODIFY COLUMN c2 VARCHAR(30)");
+  run("ALTER STABLE st1 MODIFY COLUMN c2 VARCHAR(30)");
   clearAlterStbReq();
 
   // setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_COLUMN_NAME, 2, "c1", 0, 0, "cc1");
-  // run("ALTER TABLE st1 RENAME COLUMN c1 cc1");
+  // run("ALTER STABLE st1 RENAME COLUMN c1 cc1");
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_ADD_TAG, 1, "tag11", TSDB_DATA_TYPE_BIGINT);
-  run("ALTER TABLE st1 ADD TAG tag11 BIGINT");
+  run("ALTER STABLE st1 ADD TAG tag11 BIGINT");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_DROP_TAG, 1, "tag1");
-  run("ALTER TABLE st1 DROP TAG tag1");
+  run("ALTER STABLE st1 DROP TAG tag1");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_TAG_BYTES, 1, "tag2", TSDB_DATA_TYPE_VARCHAR,
                      30 + VARSTR_HEADER_SIZE);
-  run("ALTER TABLE st1 MODIFY TAG tag2 VARCHAR(30)");
+  run("ALTER STABLE st1 MODIFY TAG tag2 VARCHAR(30)");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_TAG_NAME, 2, "tag1", 0, 0, "tag11");
-  run("ALTER TABLE st1 RENAME TAG tag1 tag11");
+  run("ALTER STABLE st1 RENAME TAG tag1 tag11");
   clearAlterStbReq();
 
   // todo
@@ -228,11 +226,11 @@ TEST_F(ParserInitialATest, alterSTable) {
 TEST_F(ParserInitialATest, alterSTableSemanticCheck) {
   useDb("root", "test");
 
-  run("ALTER TABLE st1 RENAME COLUMN c1 cc1", TSDB_CODE_PAR_INVALID_ALTER_TABLE);
+  run("ALTER STABLE st1 RENAME COLUMN c1 cc1", TSDB_CODE_PAR_INVALID_ALTER_TABLE);
 
-  run("ALTER TABLE st1 MODIFY COLUMN c2 NCHAR(10)", TSDB_CODE_PAR_INVALID_MODIFY_COL);
+  run("ALTER STABLE st1 MODIFY COLUMN c2 NCHAR(10)", TSDB_CODE_PAR_INVALID_MODIFY_COL);
 
-  run("ALTER TABLE st1 MODIFY TAG tag2 NCHAR(10)", TSDB_CODE_PAR_INVALID_MODIFY_COL);
+  run("ALTER STABLE st1 MODIFY TAG tag2 NCHAR(10)", TSDB_CODE_PAR_INVALID_MODIFY_COL);
 }
 
 TEST_F(ParserInitialATest, alterTable) {

source/libs/parser/test/parInitialCTest.cpp

@@ -359,11 +359,11 @@ TEST_F(ParserInitialCTest, createStable) {
     memset(&expect, 0, sizeof(SMCreateStbReq));
   };
 
-  auto setCreateStbReqFunc = [&](const char* pTbname, int8_t igExists = 0, int64_t delay1 = -1, int64_t delay2 = -1,
+  auto setCreateStbReqFunc = [&](const char* pDbName, const char* pTbName, int8_t igExists = 0, int64_t delay1 = -1,
-                                 int64_t watermark1 = TSDB_DEFAULT_ROLLUP_WATERMARK,
+                                 int64_t delay2 = -1, int64_t watermark1 = TSDB_DEFAULT_ROLLUP_WATERMARK,
                                  int64_t watermark2 = TSDB_DEFAULT_ROLLUP_WATERMARK,
                                  int32_t ttl = TSDB_DEFAULT_TABLE_TTL, const char* pComment = nullptr) {
-    int32_t len = snprintf(expect.name, sizeof(expect.name), "0.test.%s", pTbname);
+    int32_t len = snprintf(expect.name, sizeof(expect.name), "0.%s.%s", pDbName, pTbName);
     expect.name[len] = '\0';
     expect.igExists = igExists;
     expect.delay1 = delay1;
@@ -454,14 +454,14 @@ TEST_F(ParserInitialCTest, createStable) {
     tFreeSMCreateStbReq(&req);
   });
 
-  setCreateStbReqFunc("t1");
+  setCreateStbReqFunc("test", "t1");
   addFieldToCreateStbReqFunc(true, "ts", TSDB_DATA_TYPE_TIMESTAMP);
   addFieldToCreateStbReqFunc(true, "c1", TSDB_DATA_TYPE_INT);
   addFieldToCreateStbReqFunc(false, "id", TSDB_DATA_TYPE_INT);
   run("CREATE STABLE t1(ts TIMESTAMP, c1 INT) TAGS(id INT)");
   clearCreateStbReq();
 
-  setCreateStbReqFunc("t1", 1, 100 * MILLISECOND_PER_SECOND, 10 * MILLISECOND_PER_MINUTE, 10,
+  setCreateStbReqFunc("rollup_db", "t1", 1, 100 * MILLISECOND_PER_SECOND, 10 * MILLISECOND_PER_MINUTE, 10,
                       1 * MILLISECOND_PER_MINUTE, 100, "test create table");
   addFieldToCreateStbReqFunc(true, "ts", TSDB_DATA_TYPE_TIMESTAMP, 0, 0);
   addFieldToCreateStbReqFunc(true, "c1", TSDB_DATA_TYPE_INT);
@@ -493,7 +493,7 @@ TEST_F(ParserInitialCTest, createStable) {
   addFieldToCreateStbReqFunc(false, "a13", TSDB_DATA_TYPE_BOOL);
   addFieldToCreateStbReqFunc(false, "a14", TSDB_DATA_TYPE_NCHAR, 30 * TSDB_NCHAR_SIZE + VARSTR_HEADER_SIZE);
   addFieldToCreateStbReqFunc(false, "a15", TSDB_DATA_TYPE_VARCHAR, 50 + VARSTR_HEADER_SIZE);
-  run("CREATE STABLE IF NOT EXISTS test.t1("
+  run("CREATE STABLE IF NOT EXISTS rollup_db.t1("
       "ts TIMESTAMP, c1 INT, c2 INT UNSIGNED, c3 BIGINT, c4 BIGINT UNSIGNED, c5 FLOAT, c6 DOUBLE, c7 BINARY(20), "
       "c8 SMALLINT, c9 SMALLINT UNSIGNED COMMENT 'test column comment', c10 TINYINT, c11 TINYINT UNSIGNED, c12 BOOL, "
       "c13 NCHAR(30), c14 VARCHAR(50)) "
@@ -507,12 +507,13 @@ TEST_F(ParserInitialCTest, createStable) {
 TEST_F(ParserInitialCTest, createStableSemanticCheck) {
   useDb("root", "test");
 
-  run("CREATE STABLE stb2 (ts TIMESTAMP, c1 INT) TAGS (tag1 INT) ROLLUP(CEIL)", TSDB_CODE_PAR_INVALID_ROLLUP_OPTION);
+  run("CREATE STABLE rollup_db.stb2 (ts TIMESTAMP, c1 INT) TAGS (tag1 INT) ROLLUP(CEIL)",
+      TSDB_CODE_PAR_INVALID_ROLLUP_OPTION);
 
-  run("CREATE STABLE stb2 (ts TIMESTAMP, c1 INT) TAGS (tag1 INT) ROLLUP(MAX) MAX_DELAY 0s WATERMARK 1m",
+  run("CREATE STABLE rollup_db.stb2 (ts TIMESTAMP, c1 INT) TAGS (tag1 INT) ROLLUP(MAX) MAX_DELAY 0s WATERMARK 1m",
       TSDB_CODE_PAR_INVALID_RANGE_OPTION);
 
-  run("CREATE STABLE stb2 (ts TIMESTAMP, c1 INT) TAGS (tag1 INT) ROLLUP(MAX) MAX_DELAY 10s WATERMARK 18m",
+  run("CREATE STABLE rollup_db.stb2 (ts TIMESTAMP, c1 INT) TAGS (tag1 INT) ROLLUP(MAX) MAX_DELAY 10s WATERMARK 18m",
       TSDB_CODE_PAR_INVALID_RANGE_OPTION);
 }
 
@@ -561,30 +562,33 @@ TEST_F(ParserInitialCTest, createStream) {
     tFreeSCMCreateStreamReq(&req);
   });
 
-  setCreateStreamReqFunc("s1", "test", "create stream s1 as select * from t1");
+  setCreateStreamReqFunc("s1", "test", "create stream s1 as select count(*) from t1 interval(10s)");
-  run("CREATE STREAM s1 AS SELECT * FROM t1");
+  run("CREATE STREAM s1 AS SELECT COUNT(*) FROM t1 INTERVAL(10S)");
   clearCreateStreamReq();
 
-  setCreateStreamReqFunc("s1", "test", "create stream if not exists s1 as select * from t1", nullptr, 1);
+  setCreateStreamReqFunc("s1", "test", "create stream if not exists s1 as select count(*) from t1 interval(10s)",
-  run("CREATE STREAM IF NOT EXISTS s1 AS SELECT * FROM t1");
+                         nullptr, 1);
+  run("CREATE STREAM IF NOT EXISTS s1 AS SELECT COUNT(*) FROM t1 INTERVAL(10S)");
   clearCreateStreamReq();
 
-  setCreateStreamReqFunc("s1", "test", "create stream s1 into st1 as select * from t1", "st1");
+  setCreateStreamReqFunc("s1", "test", "create stream s1 into st1 as select count(*) from t1 interval(10s)", "st1");
-  run("CREATE STREAM s1 INTO st1 AS SELECT * FROM t1");
+  run("CREATE STREAM s1 INTO st1 AS SELECT COUNT(*) FROM t1 INTERVAL(10S)");
   clearCreateStreamReq();
 
-  setCreateStreamReqFunc(
+  setCreateStreamReqFunc("s1", "test",
-      "s1", "test",
+                         "create stream if not exists s1 trigger max_delay 20s watermark 10s ignore expired into st1 "
-      "create stream if not exists s1 trigger max_delay 20s watermark 10s ignore expired into st1 as select * from t1",
+                         "as select count(*) from t1 interval(10s)",
-      "st1", 1, STREAM_TRIGGER_MAX_DELAY, 20 * MILLISECOND_PER_SECOND, 10 * MILLISECOND_PER_SECOND, 1);
+                         "st1", 1, STREAM_TRIGGER_MAX_DELAY, 20 * MILLISECOND_PER_SECOND, 10 * MILLISECOND_PER_SECOND,
-  run("CREATE STREAM IF NOT EXISTS s1 TRIGGER MAX_DELAY 20s WATERMARK 10s IGNORE EXPIRED INTO st1 AS SELECT * FROM t1");
+                         1);
+  run("CREATE STREAM IF NOT EXISTS s1 TRIGGER MAX_DELAY 20s WATERMARK 10s IGNORE EXPIRED INTO st1 AS SELECT COUNT(*) "
+      "FROM t1 INTERVAL(10S)");
   clearCreateStreamReq();
 }
 
 TEST_F(ParserInitialCTest, createStreamSemanticCheck) {
   useDb("root", "test");
 
-  run("CREATE STREAM s1 AS SELECT PERCENTILE(c1, 30) FROM t1", TSDB_CODE_PAR_STREAM_NOT_ALLOWED_FUNC);
+  run("CREATE STREAM s1 AS SELECT PERCENTILE(c1, 30) FROM t1 INTERVAL(10S)", TSDB_CODE_PAR_STREAM_NOT_ALLOWED_FUNC);
 }
 
 TEST_F(ParserInitialCTest, createTable) {
@@ -598,7 +602,7 @@ TEST_F(ParserInitialCTest, createTable) {
       "c13 NCHAR(30), c15 VARCHAR(50)) "
       "TTL 100 COMMENT 'test create table' SMA(c1, c2, c3)");
 
-  run("CREATE TABLE IF NOT EXISTS test.t1("
+  run("CREATE TABLE IF NOT EXISTS rollup_db.t1("
       "ts TIMESTAMP, c1 INT, c2 INT UNSIGNED, c3 BIGINT, c4 BIGINT UNSIGNED, c5 FLOAT, c6 DOUBLE, c7 BINARY(20), "
       "c8 SMALLINT, c9 SMALLINT UNSIGNED COMMENT 'test column comment', c10 TINYINT, c11 TINYINT UNSIGNED, c12 BOOL, "
       "c13 NCHAR(30), c14 VARCHAR(50)) "
@@ -617,6 +621,21 @@ TEST_F(ParserInitialCTest, createTable) {
   // run("CREATE TABLE IF NOT EXISTS t1 USING st1 TAGS(1, 'wxy', NOW + 1S)");
 }
 
+TEST_F(ParserInitialCTest, createTableSemanticCheck) {
+  useDb("root", "test");
+
+  string sql = "CREATE TABLE st1(ts TIMESTAMP, ";
+  for (int32_t i = 1; i < 4096; ++i) {
+    if (i > 1) {
+      sql.append(", ");
+    }
+    sql.append("c" + to_string(i) + " INT");
+  }
+  sql.append(") TAGS (t1 int)");
+
+  run(sql, TSDB_CODE_PAR_TOO_MANY_COLUMNS);
+}
+
 TEST_F(ParserInitialCTest, createTopic) {
   useDb("root", "test");
 

source/libs/planner/src/planOptimizer.c

@@ -480,12 +480,18 @@ static int32_t pushDownCondOptPushCondToProject(SOptimizeContext* pCxt, SProject
   return pushDownCondOptAppendCond(&pProject->node.pConditions, pCond);
 }
 
+static int32_t pushDownCondOptPushCondToJoin(SOptimizeContext* pCxt, SJoinLogicNode * pJoin, SNode** pCond) {
+  return pushDownCondOptAppendCond(&pJoin->node.pConditions, pCond);
+}
+
 static int32_t pushDownCondOptPushCondToChild(SOptimizeContext* pCxt, SLogicNode* pChild, SNode** pCond) {
   switch (nodeType(pChild)) {
     case QUERY_NODE_LOGIC_PLAN_SCAN:
       return pushDownCondOptPushCondToScan(pCxt, (SScanLogicNode*)pChild, pCond);
     case QUERY_NODE_LOGIC_PLAN_PROJECT:
       return pushDownCondOptPushCondToProject(pCxt, (SProjectLogicNode*)pChild, pCond);
+    case QUERY_NODE_LOGIC_PLAN_JOIN:
+      return pushDownCondOptPushCondToJoin(pCxt, (SJoinLogicNode*)pChild, pCond);
     default:
       break;
   }
@@ -554,13 +560,83 @@ static int32_t pushDownCondOptCheckJoinOnCond(SOptimizeContext* pCxt, SJoinLogic
   return TSDB_CODE_SUCCESS;
 }
 
+static int32_t pushDownCondOptPartJoinOnCondLogicCond(SJoinLogicNode* pJoin, SNode** ppMergeCond, SNode** ppOnCond) {
+  SLogicConditionNode* pLogicCond = (SLogicConditionNode*)(pJoin->pOnConditions);
+
+  int32_t code = TSDB_CODE_SUCCESS;
+  SNodeList* pOnConds = NULL;
+  SNode* pCond = NULL;
+  FOREACH(pCond, pLogicCond->pParameterList) {
+    if (pushDownCondOptIsPriKeyEqualCond(pJoin, pCond)) {
+      *ppMergeCond = nodesCloneNode(pCond);
+    } else {
+      code = nodesListMakeAppend(&pOnConds, nodesCloneNode(pCond));
+    }
+  }
+
+  SNode* pTempOnCond = NULL;
+  if (TSDB_CODE_SUCCESS == code) {
+    code = nodesMergeConds(&pTempOnCond, &pOnConds);
+  }
+
+  if (TSDB_CODE_SUCCESS == code && NULL != *ppMergeCond) {
+    *ppOnCond = pTempOnCond;
+    nodesDestroyNode(pJoin->pOnConditions);
+    pJoin->pOnConditions = NULL;
+    return TSDB_CODE_SUCCESS;
+  } else {
+    nodesDestroyList(pOnConds);
+    nodesDestroyNode(pTempOnCond);
+    return TSDB_CODE_PLAN_INTERNAL_ERROR;
+  }
+}
+
+static int32_t pushDownCondOptPartJoinOnCond(SJoinLogicNode* pJoin, SNode** ppMergeCond, SNode** ppOnCond) {
+  if (QUERY_NODE_LOGIC_CONDITION == nodeType(pJoin->pOnConditions) &&
+      LOGIC_COND_TYPE_AND == ((SLogicConditionNode*)(pJoin->pOnConditions))->condType) {
+    return pushDownCondOptPartJoinOnCondLogicCond(pJoin, ppMergeCond, ppOnCond);
+  }
+
+  if (pushDownCondOptIsPriKeyEqualCond(pJoin, pJoin->pOnConditions)) {
+    *ppMergeCond = nodesCloneNode(pJoin->pOnConditions);
+    *ppOnCond = NULL;
+    nodesDestroyNode(pJoin->pOnConditions);
+    pJoin->pOnConditions = NULL;
+    return TSDB_CODE_SUCCESS;
+  } else {
+    return TSDB_CODE_PLAN_INTERNAL_ERROR;
+  }
+}
+
+static int32_t pushDownCondOptJoinExtractMergeCond(SOptimizeContext* pCxt, SJoinLogicNode* pJoin) {
+  int32_t code = pushDownCondOptCheckJoinOnCond(pCxt, pJoin);
+  SNode* pJoinMergeCond = NULL;
+  SNode* pJoinOnCond = NULL;
+  if (TSDB_CODE_SUCCESS == code) {
+    code = pushDownCondOptPartJoinOnCond(pJoin, &pJoinMergeCond, &pJoinOnCond);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    pJoin->pMergeCondition = pJoinMergeCond;
+    pJoin->pOnConditions = pJoinOnCond;
+  } else {
+    nodesDestroyNode(pJoinMergeCond);
+    nodesDestroyNode(pJoinOnCond);
+  }
+  return code;
+}
+
 static int32_t pushDownCondOptDealJoin(SOptimizeContext* pCxt, SJoinLogicNode* pJoin) {
   if (OPTIMIZE_FLAG_TEST_MASK(pJoin->node.optimizedFlag, OPTIMIZE_FLAG_PUSH_DOWN_CONDE)) {
     return TSDB_CODE_SUCCESS;
   }
 
   if (NULL == pJoin->node.pConditions) {
-    return pushDownCondOptCheckJoinOnCond(pCxt, pJoin);
+    int32_t code = pushDownCondOptJoinExtractMergeCond(pCxt, pJoin);
+    if (TSDB_CODE_SUCCESS == code) {
+      OPTIMIZE_FLAG_SET_MASK(pJoin->node.optimizedFlag, OPTIMIZE_FLAG_PUSH_DOWN_CONDE);
+      pCxt->optimized = true;
+    }
+    return code;
   }
 
   SNode*  pOnCond = NULL;
@@ -579,10 +655,13 @@ static int32_t pushDownCondOptDealJoin(SOptimizeContext* pCxt, SJoinLogicNode* p
         pushDownCondOptPushCondToChild(pCxt, (SLogicNode*)nodesListGetNode(pJoin->node.pChildren, 1), &pRightChildCond);
   }
 
+  if (TSDB_CODE_SUCCESS == code) {
+    code = pushDownCondOptJoinExtractMergeCond(pCxt, pJoin);
+  }
+
   if (TSDB_CODE_SUCCESS == code) {
     OPTIMIZE_FLAG_SET_MASK(pJoin->node.optimizedFlag, OPTIMIZE_FLAG_PUSH_DOWN_CONDE);
     pCxt->optimized = true;
-    code = pushDownCondOptCheckJoinOnCond(pCxt, pJoin);
   } else {
     nodesDestroyNode(pOnCond);
     nodesDestroyNode(pLeftChildCond);
@@ -720,7 +799,8 @@ static int32_t pushDownCondOptDealAgg(SOptimizeContext* pCxt, SAggLogicNode* pAg
   // TODO: remove it after full implementation of pushing down to child
   if (1 != LIST_LENGTH(pAgg->node.pChildren) ||
       QUERY_NODE_LOGIC_PLAN_SCAN != nodeType(nodesListGetNode(pAgg->node.pChildren, 0)) &&
-      QUERY_NODE_LOGIC_PLAN_PROJECT != nodeType(nodesListGetNode(pAgg->node.pChildren, 0))) {
+      QUERY_NODE_LOGIC_PLAN_PROJECT != nodeType(nodesListGetNode(pAgg->node.pChildren, 0)) &&
+      QUERY_NODE_LOGIC_PLAN_JOIN != nodeType(nodesListGetNode(pAgg->node.pChildren, 0))) {
     return TSDB_CODE_SUCCESS;
   }
 
@@ -1251,7 +1331,7 @@ static SNode* partTagsCreateWrapperFunc(const char* pFuncName, SNode* pNode) {
   }
 
   strcpy(pFunc->functionName, pFuncName);
-  if (QUERY_NODE_COLUMN == nodeType(pNode)) {
+  if (QUERY_NODE_COLUMN == nodeType(pNode) && COLUMN_TYPE_TBNAME != ((SColumnNode*)pNode)->colType) {
     SColumnNode* pCol = (SColumnNode*)pNode;
     partTagsSetAlias(pFunc->node.aliasName, sizeof(pFunc->node.aliasName), pCol->tableAlias, pCol->colName);
   } else {
@@ -1868,6 +1948,8 @@ static EDealRes mergeProjectionsExpr(SNode** pNode, void* pContext) {
               pCxt->errCode = terrno;
               return DEAL_RES_ERROR;
             }
+            snprintf(((SExprNode*)pExpr)->aliasName, sizeof(((SExprNode*)pExpr)->aliasName), "%s",
+                     ((SExprNode*)*pNode)->aliasName);
             nodesDestroyNode(*pNode);
             *pNode = pExpr;
           }

source/libs/planner/src/planPhysiCreater.c

@@ -552,6 +552,9 @@ static int32_t createSystemTableScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan*
   if (0 == strcmp(pScanLogicNode->tableName.tname, TSDB_INS_TABLE_USER_TABLES) ||
       0 == strcmp(pScanLogicNode->tableName.tname, TSDB_INS_TABLE_USER_TABLE_DISTRIBUTED)) {
     vgroupInfoToNodeAddr(pScanLogicNode->pVgroupList->vgroups, &pSubplan->execNode);
+  } else {
+    pSubplan->execNode.nodeId = MNODE_HANDLE;
+    pSubplan->execNode.epSet = pCxt->pPlanCxt->mgmtEpSet;
   }
   SQueryNodeLoad node = {.addr = {.nodeId = MNODE_HANDLE, .epSet = pCxt->pPlanCxt->mgmtEpSet}, .load = 0};
   taosArrayPush(pCxt->pExecNodeList, &node);
@@ -609,10 +612,8 @@ static int32_t createJoinPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren
   int32_t             code = TSDB_CODE_SUCCESS;
 
   pJoin->joinType = pJoinLogicNode->joinType;
-  if (NULL != pJoinLogicNode->pOnConditions) {
+  setNodeSlotId(pCxt, pLeftDesc->dataBlockId, pRightDesc->dataBlockId, pJoinLogicNode->pMergeCondition,
-    code = setNodeSlotId(pCxt, pLeftDesc->dataBlockId, pRightDesc->dataBlockId, pJoinLogicNode->pOnConditions,
+                &pJoin->pMergeCondition);
-                         &pJoin->pOnConditions);
-  }
   if (TSDB_CODE_SUCCESS == code) {
     code = setListSlotId(pCxt, pLeftDesc->dataBlockId, pRightDesc->dataBlockId, pJoinLogicNode->node.pTargets,
                          &pJoin->pTargets);
@@ -620,6 +621,21 @@ static int32_t createJoinPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren
   if (TSDB_CODE_SUCCESS == code) {
     code = addDataBlockSlots(pCxt, pJoin->pTargets, pJoin->node.pOutputDataBlockDesc);
   }
+
+  SNodeList* condCols = nodesMakeList();
+  if (TSDB_CODE_SUCCESS == code && NULL != pJoinLogicNode->pOnConditions) {
+    code = nodesCollectColumnsFromNode(pJoinLogicNode->pOnConditions, NULL, COLLECT_COL_TYPE_ALL, &condCols);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = addDataBlockSlots(pCxt, condCols, pJoin->node.pOutputDataBlockDesc);
+    nodesDestroyList(condCols);
+  }
+
+  if (TSDB_CODE_SUCCESS == code && NULL != pJoinLogicNode->pOnConditions) {
+    code = setNodeSlotId(pCxt, ((SPhysiNode*)pJoin)->pOutputDataBlockDesc->dataBlockId, -1, pJoinLogicNode->pOnConditions,
+                         &pJoin->pOnConditions);
+  }
+
   if (TSDB_CODE_SUCCESS == code) {
     code = setConditionsSlotId(pCxt, (const SLogicNode*)pJoinLogicNode, (SPhysiNode*)pJoin);
   }

source/libs/planner/src/planSpliter.c

@@ -986,6 +986,10 @@ static bool unionIsChildSubplan(SLogicNode* pLogicNode, int32_t groupId) {
     return ((SExchangeLogicNode*)pLogicNode)->srcGroupId == groupId;
   }
 
+  if (QUERY_NODE_LOGIC_PLAN_MERGE == nodeType(pLogicNode)) {
+    return ((SMergeLogicNode*)pLogicNode)->srcGroupId == groupId;
+  }
+
   SNode* pChild;
   FOREACH(pChild, pLogicNode->pChildren) {
     bool isChild = unionIsChildSubplan((SLogicNode*)pChild, groupId);
@@ -1014,14 +1018,14 @@ static int32_t unionMountSubplan(SLogicSubplan* pParent, SNodeList* pChildren) {
   return TSDB_CODE_SUCCESS;
 }
 
-static SLogicSubplan* unionCreateSubplan(SSplitContext* pCxt, SLogicNode* pNode) {
+static SLogicSubplan* unionCreateSubplan(SSplitContext* pCxt, SLogicNode* pNode, ESubplanType subplanType) {
   SLogicSubplan* pSubplan = (SLogicSubplan*)nodesMakeNode(QUERY_NODE_LOGIC_SUBPLAN);
   if (NULL == pSubplan) {
     return NULL;
   }
   pSubplan->id.queryId = pCxt->queryId;
   pSubplan->id.groupId = pCxt->groupId;
-  pSubplan->subplanType = SUBPLAN_TYPE_SCAN;
+  pSubplan->subplanType = subplanType;
   pSubplan->pNode = pNode;
   pNode->pParent = NULL;
   return pSubplan;
@@ -1035,7 +1039,7 @@ static int32_t unionSplitSubplan(SSplitContext* pCxt, SLogicSubplan* pUnionSubpl
 
   SNode* pChild = NULL;
   FOREACH(pChild, pSplitNode->pChildren) {
-    SLogicSubplan* pNewSubplan = unionCreateSubplan(pCxt, (SLogicNode*)pChild);
+    SLogicSubplan* pNewSubplan = unionCreateSubplan(pCxt, (SLogicNode*)pChild, pUnionSubplan->subplanType);
     code = nodesListMakeStrictAppend(&pUnionSubplan->pChildren, (SNode*)pNewSubplan);
     if (TSDB_CODE_SUCCESS == code) {
       REPLACE_NODE(NULL);
@@ -1216,6 +1220,7 @@ static int32_t qnodeSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
   if (!splMatch(pCxt, pSubplan, 0, (FSplFindSplitNode)qndSplFindSplitNode, &info)) {
     return TSDB_CODE_SUCCESS;
   }
+  ((SScanLogicNode*)info.pSplitNode)->dataRequired = FUNC_DATA_REQUIRED_DATA_LOAD;
   int32_t code = splCreateExchangeNodeForSubplan(pCxt, info.pSubplan, info.pSplitNode, info.pSubplan->subplanType);
   if (TSDB_CODE_SUCCESS == code) {
     SLogicSubplan* pScanSubplan = splCreateScanSubplan(pCxt, info.pSplitNode, 0);

source/libs/planner/test/planOptimizeTest.cpp

@@ -68,6 +68,8 @@ TEST_F(PlanOptimizeTest, PartitionTags) {
   run("SELECT SUM(c1), tag1 FROM st1 GROUP BY tag1");
 
   run("SELECT SUM(c1), tag1 + 10 FROM st1 GROUP BY tag1 + 10");
+
+  run("SELECT SUM(c1), tbname FROM st1 GROUP BY tbname");
 }
 
 TEST_F(PlanOptimizeTest, eliminateProjection) {

source/libs/planner/test/planSetOpTest.cpp

@@ -97,7 +97,15 @@ TEST_F(PlanSetOpTest, unionSubquery) {
   run("SELECT * FROM (SELECT c1, c2 FROM t1 UNION SELECT c1, c2 FROM t1)");
 }
 
-TEST_F(PlanSetOpTest, bug001) {
+TEST_F(PlanSetOpTest, unionWithSubquery) {
+  useDb("root", "test");
+
+  run("SELECT c1 FROM (SELECT c1 FROM st1) UNION SELECT c2 FROM (SELECT c1 AS c2 FROM st2)");
+
+  run("SELECT c1 FROM (SELECT c1 FROM st1 ORDER BY c2) UNION SELECT c1 FROM (SELECT c1 FROM st2)");
+}
+
+TEST_F(PlanSetOpTest, unionDataTypeConversion) {
   useDb("root", "test");
 
   run("SELECT c2 FROM t1 WHERE c1 IS NOT NULL GROUP BY c2 "

source/libs/scalar/src/scalar.c

@@ -1035,3 +1035,72 @@ _return:
   sclFreeRes(ctx.pRes);
   return code;
 }
+
+int32_t scalarGetOperatorResultType(SDataType left, SDataType right, EOperatorType op, SDataType* pRes) {
+  switch (op) {
+    case OP_TYPE_ADD:
+      if (left.type == TSDB_DATA_TYPE_TIMESTAMP && right.type == TSDB_DATA_TYPE_TIMESTAMP) {
+        qError("invalid op %d, left type:%d, right type:%d", op, left.type, right.type);
+        return TSDB_CODE_TSC_INVALID_OPERATION;
+      }
+      if ((left.type == TSDB_DATA_TYPE_TIMESTAMP && (IS_INTEGER_TYPE(right.type) || right.type == TSDB_DATA_TYPE_BOOL)) ||
+          (right.type == TSDB_DATA_TYPE_TIMESTAMP && (IS_INTEGER_TYPE(left.type) || left.type == TSDB_DATA_TYPE_BOOL))) {
+        pRes->type = TSDB_DATA_TYPE_TIMESTAMP;
+        return TSDB_CODE_SUCCESS;
+      }
+      pRes->type = TSDB_DATA_TYPE_DOUBLE;
+      return TSDB_CODE_SUCCESS;
+    case OP_TYPE_SUB:
+      if ((left.type == TSDB_DATA_TYPE_TIMESTAMP && right.type == TSDB_DATA_TYPE_BIGINT) ||
+          (right.type == TSDB_DATA_TYPE_TIMESTAMP && left.type == TSDB_DATA_TYPE_BIGINT)) {
+        pRes->type = TSDB_DATA_TYPE_TIMESTAMP;
+        return TSDB_CODE_SUCCESS;
+      }
+      pRes->type = TSDB_DATA_TYPE_DOUBLE;
+      return TSDB_CODE_SUCCESS;
+    case OP_TYPE_MULTI:
+      if (left.type == TSDB_DATA_TYPE_TIMESTAMP && right.type == TSDB_DATA_TYPE_TIMESTAMP) {
+        qError("invalid op %d, left type:%d, right type:%d", op, left.type, right.type);
+        return TSDB_CODE_TSC_INVALID_OPERATION;
+      }
+    case OP_TYPE_DIV:
+      if (left.type == TSDB_DATA_TYPE_TIMESTAMP && right.type == TSDB_DATA_TYPE_TIMESTAMP) {
+        qError("invalid op %d, left type:%d, right type:%d", op, left.type, right.type);
+        return TSDB_CODE_TSC_INVALID_OPERATION;
+      }
+    case OP_TYPE_REM:
+    case OP_TYPE_MINUS:
+      pRes->type = TSDB_DATA_TYPE_DOUBLE;
+      return TSDB_CODE_SUCCESS;
+    case OP_TYPE_GREATER_THAN:
+    case OP_TYPE_GREATER_EQUAL:
+    case OP_TYPE_LOWER_THAN:
+    case OP_TYPE_LOWER_EQUAL:
+    case OP_TYPE_EQUAL:
+    case OP_TYPE_NOT_EQUAL:
+    case OP_TYPE_IN:
+    case OP_TYPE_NOT_IN:
+    case OP_TYPE_LIKE:
+    case OP_TYPE_NOT_LIKE:
+    case OP_TYPE_MATCH:
+    case OP_TYPE_NMATCH:
+    case OP_TYPE_IS_NULL:
+    case OP_TYPE_IS_NOT_NULL:
+    case OP_TYPE_IS_TRUE:
+    case OP_TYPE_JSON_CONTAINS:
+      pRes->type = TSDB_DATA_TYPE_BOOL;
+      return TSDB_CODE_SUCCESS;
+    case OP_TYPE_BIT_AND:
+    case OP_TYPE_BIT_OR:
+      pRes->type = TSDB_DATA_TYPE_BIGINT;
+      return TSDB_CODE_SUCCESS;
+    case OP_TYPE_JSON_GET_VALUE:
+      pRes->type = TSDB_DATA_TYPE_JSON;
+      return TSDB_CODE_SUCCESS;
+    default:
+      ASSERT(0);
+      return TSDB_CODE_APP_ERROR;
+  }
+}
+
+

source/libs/scalar/src/sclfunc.c

@@ -360,9 +360,6 @@ static void trtrim(char *input, char *output, int32_t type, int32_t charLen) {
 
 static int32_t doLengthFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput, _len_fn lenFn) {
   int32_t type = GET_PARAM_TYPE(pInput);
-  if (inputNum != 1 || !IS_VAR_DATA_TYPE(type)) {
-    return TSDB_CODE_FAILED;
-  }
 
   SColumnInfoData *pInputData  = pInput->columnData;
   SColumnInfoData *pOutputData = pOutput->columnData;
@@ -586,9 +583,6 @@ DONE:
 
 static int32_t doCaseConvFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput, _conv_fn convFn) {
   int32_t type = GET_PARAM_TYPE(pInput);
-  if (inputNum != 1 || !IS_VAR_DATA_TYPE(type)) {
-    return TSDB_CODE_FAILED;
-  }
 
   SColumnInfoData *pInputData  = pInput->columnData;
   SColumnInfoData *pOutputData = pOutput->columnData;
@@ -628,9 +622,6 @@ static int32_t doCaseConvFunction(SScalarParam *pInput, int32_t inputNum, SScala
 
 static int32_t doTrimFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput, _trim_fn trimFn) {
   int32_t type = GET_PARAM_TYPE(pInput);
-  if (inputNum != 1 || !IS_VAR_DATA_TYPE(type)) {
-    return TSDB_CODE_FAILED;
-  }
 
   SColumnInfoData *pInputData  = pInput->columnData;
   SColumnInfoData *pOutputData = pOutput->columnData;
@@ -664,16 +655,10 @@ static int32_t doTrimFunction(SScalarParam *pInput, int32_t inputNum, SScalarPar
 int32_t substrFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput) {
   int32_t subPos = 0;
   GET_TYPED_DATA(subPos, int32_t, GET_PARAM_TYPE(&pInput[1]), pInput[1].columnData->pData);
-  if (subPos == 0) { //subPos needs to be positive or negative values;
-    return TSDB_CODE_FAILED;
-  }
 
   int32_t subLen = INT16_MAX;
   if (inputNum == 3) {
     GET_TYPED_DATA(subLen, int32_t, GET_PARAM_TYPE(&pInput[2]), pInput[2].columnData->pData);
-    if (subLen < 0 || subLen > INT16_MAX) { //subLen cannot be negative
-      return TSDB_CODE_FAILED;
-    }
     subLen = (GET_PARAM_TYPE(pInput) == TSDB_DATA_TYPE_VARCHAR) ? subLen : subLen * TSDB_NCHAR_SIZE;
   }
 
@@ -1149,13 +1134,6 @@ int32_t toUnixtimestampFunction(SScalarParam *pInput, int32_t inputNum, SScalarP
 
 int32_t toJsonFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutput) {
   int32_t type = GET_PARAM_TYPE(pInput);
-  if (type != TSDB_DATA_TYPE_BINARY && type != TSDB_DATA_TYPE_NCHAR) {
-    return TSDB_CODE_FAILED;
-  }
-
-  if (inputNum != 1) {
-    return TSDB_CODE_FAILED;
-  }
 
   char tmp[TSDB_MAX_JSON_TAG_LEN] = {0};
   for (int32_t i = 0; i < pInput[0].numOfRows; ++i) {
@@ -1196,6 +1174,8 @@ int32_t timeTruncateFunction(SScalarParam *pInput, int32_t inputNum, SScalarPara
   int64_t factor = (timePrec == TSDB_TIME_PRECISION_MILLI) ? 1000 :
                    (timePrec == TSDB_TIME_PRECISION_MICRO ? 1000000 : 1000000000);
 
+  timeUnit = timeUnit * 1000 / factor;
+
   for (int32_t i = 0; i < pInput[0].numOfRows; ++i) {
     if (colDataIsNull_s(pInput[0].columnData, i)) {
       colDataAppendNULL(pOutput->columnData, i);
@@ -1228,7 +1208,6 @@ int32_t timeTruncateFunction(SScalarParam *pInput, int32_t inputNum, SScalarPara
     char buf[20] = {0};
     NUM_TO_STRING(TSDB_DATA_TYPE_BIGINT, &timeVal, sizeof(buf), buf);
     int32_t tsDigits = (int32_t)strlen(buf);
-    timeUnit = timeUnit * 1000 / factor;
 
     switch (timeUnit) {
       case 0: { /* 1u */
@@ -1384,6 +1363,11 @@ int32_t timeDiffFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *p
     GET_TYPED_DATA(timePrec, int64_t, GET_PARAM_TYPE(&pInput[2]), pInput[2].columnData->pData);
   }
 
+  int64_t factor = (timePrec == TSDB_TIME_PRECISION_MILLI) ? 1000 :
+                   (timePrec == TSDB_TIME_PRECISION_MICRO ? 1000000 : 1000000000);
+
+  timeUnit = timeUnit * 1000 / factor;
+
   int32_t numOfRows = 0;
   for (int32_t i = 0; i < inputNum; ++i) {
     if (pInput[i].numOfRows > numOfRows) {
@@ -1463,9 +1447,6 @@ int32_t timeDiffFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *p
         }
       }
     } else {
-      int64_t factor = (timePrec == TSDB_TIME_PRECISION_MILLI) ? 1000 :
-                       (timePrec == TSDB_TIME_PRECISION_MICRO ? 1000000 : 1000000000);
-      timeUnit = timeUnit * 1000 / factor;
       switch(timeUnit) {
         case 0: { /* 1u */
           result = result / 1000;

source/libs/scheduler/src/schJob.c

@@ -184,9 +184,7 @@ FORCE_INLINE bool schJobNeedToStop(SSchJob *pJob, int8_t *pStatus) {
   }
 
   if ((*pJob->chkKillFp)(pJob->chkKillParam)) {
-    schUpdateJobStatus(pJob, JOB_TASK_STATUS_DROPPING);
     schUpdateJobErrCode(pJob, TSDB_CODE_TSC_QUERY_KILLED);
-
     return true;
   }  
 
@@ -669,6 +667,11 @@ int32_t schSetTaskCandidateAddrs(SSchJob *pJob, SSchTask *pTask) {
     return TSDB_CODE_SUCCESS;
   }
 
+  if (SCH_IS_DATA_SRC_QRY_TASK(pTask)) {
+    SCH_TASK_ELOG("no execNode specifed for data src task, numOfEps:%d", pTask->plan->execNode.epSet.numOfEps);
+    SCH_ERR_RET(TSDB_CODE_QRY_APP_ERROR);
+  }
+
   SCH_ERR_RET(schSetAddrsFromNodeList(pJob, pTask));
 
   /*
@@ -811,14 +814,6 @@ int32_t schMoveTaskToExecList(SSchJob *pJob, SSchTask *pTask, bool *moved) {
 */
 
 int32_t schTaskCheckSetRetry(SSchJob *pJob, SSchTask *pTask, int32_t errCode, bool *needRetry) {
-  int8_t status = 0;
-
-  if (schJobNeedToStop(pJob, &status)) {
-    *needRetry = false;
-    SCH_TASK_DLOG("task no more retry cause of job status, job status:%s", jobTaskStatusStr(status));
-    return TSDB_CODE_SUCCESS;
-  }
-
   if (TSDB_CODE_SCH_TIMEOUT_ERROR == errCode) {
     pTask->maxExecTimes++;
     if (pTask->timeoutUsec < SCH_MAX_TASK_TIMEOUT_USEC) {
@@ -1277,7 +1272,7 @@ int32_t schProcessOnTaskStatusRsp(SQueryNodeEpId* pEpId, SArray* pStatusList) {
   for (int32_t i = 0; i < taskNum; ++i) {
     STaskStatus *taskStatus = taosArrayGet(pStatusList, i);
 
-    qDebug("QID:%" PRIx64 ",TID:0x%" PRIx64 ",EID:%d task status in server: %s", 
+    qDebug("QID:0x%" PRIx64 ",TID:0x%" PRIx64 ",EID:%d task status in server: %s", 
       taskStatus->queryId, taskStatus->taskId, taskStatus->execId, jobTaskStatusStr(taskStatus->status));
 
     SSchJob *pJob = schAcquireJob(taskStatus->refId);
@@ -1495,6 +1490,8 @@ void schFreeJobImpl(void *job) {
   uint64_t queryId = pJob->queryId;
   int64_t  refId = pJob->refId;
 
+  qDebug("QID:0x%" PRIx64 " begin to free sch job, refId:0x%" PRIx64 ", pointer:%p", queryId, refId, pJob);
+
   if (pJob->status == JOB_TASK_STATUS_EXECUTING) {
     schCancelJob(pJob);
   }
@@ -1535,12 +1532,12 @@ void schFreeJobImpl(void *job) {
   taosMemoryFreeClear(pJob->resData);
   taosMemoryFree(pJob);
 
-  qDebug("QID:0x%" PRIx64 " sch job freed, refId:0x%" PRIx64 ", pointer:%p", queryId, refId, pJob);
-
   int32_t jobNum = atomic_sub_fetch_32(&schMgmt.jobNum, 1);
   if (jobNum == 0) {
     schCloseJobRef();
   }
+
+  qDebug("QID:0x%" PRIx64 " sch job freed, refId:0x%" PRIx64 ", pointer:%p", queryId, refId, pJob);
 }
 
 int32_t schLaunchStaticExplainJob(SSchedulerReq *pReq, SSchJob *pJob, bool sync) {
@@ -1687,11 +1684,6 @@ _return:
 
 int32_t schDoTaskRedirect(SSchJob *pJob, SSchTask *pTask, SDataBuf* pData, int32_t rspCode) {
   int32_t code = 0;
-  int8_t  status = 0;
-  if (schJobNeedToStop(pJob, &status)) {
-    SCH_TASK_ELOG("redirect will no continue cause of job status %s", jobTaskStatusStr(status));
-    SCH_RET(atomic_load_32(&pJob->errCode));
-  }
   
   if ((pTask->execId + 1) >= pTask->maxExecTimes) {
     SCH_TASK_DLOG("task no more retry since reach max try times, execId:%d", pTask->execId);

source/libs/scheduler/src/schRemote.c

@@ -402,12 +402,16 @@ int32_t schHandleCallback(void *param, SDataBuf *pMsg, int32_t rspCode) {
     goto _return;
   }
   
-  code = schHandleResponseMsg(pJob, pTask, msgType, pMsg->pData, pMsg->len, rspCode);
+  schHandleResponseMsg(pJob, pTask, msgType, pMsg->pData, pMsg->len, rspCode);
   pMsg->pData = NULL;
 
 _return:
 
   if (pTask) {
+    if (code) {
+      schProcessOnTaskFailure(pJob, pTask, code);
+    }
+    
     SCH_UNLOCK_TASK(pTask);
   }
   

source/libs/scheduler/src/scheduler.c

@@ -225,26 +225,33 @@ void schedulerStopQueryHb(void *pTrans) {
   schCleanClusterHb(pTrans);
 }
 
-void schedulerFreeJob(int64_t job, int32_t errCode) {
+void schedulerFreeJob(int64_t* job, int32_t errCode) {
-  SSchJob *pJob = schAcquireJob(job);
+  if (0 == *job) {
+    return;
+  }
+  
+  SSchJob *pJob = schAcquireJob(*job);
   if (NULL == pJob) {
-    qError("acquire job from jobRef list failed, may be dropped, jobId:0x%" PRIx64, job);
+    qError("acquire sch job failed, may be dropped, jobId:0x%" PRIx64, *job);
+    *job = 0;
     return;
   }
 
   int32_t code = schProcessOnJobDropped(pJob, errCode);
   if (TSDB_CODE_SCH_JOB_IS_DROPPING == code) {
-    SCH_JOB_DLOG("sch job is already dropping, refId:0x%" PRIx64, job);
+    SCH_JOB_DLOG("sch job is already dropping, refId:0x%" PRIx64, *job);
+    *job = 0;
     return;
   }
 
-  SCH_JOB_DLOG("start to remove job from jobRef list, refId:0x%" PRIx64, job);
+  SCH_JOB_DLOG("start to remove job from jobRef list, refId:0x%" PRIx64, *job);
 
-  if (taosRemoveRef(schMgmt.jobRef, job)) {
+  if (taosRemoveRef(schMgmt.jobRef, *job)) {
-    SCH_JOB_ELOG("remove job from job list failed, refId:0x%" PRIx64, job);
+    SCH_JOB_ELOG("remove job from job list failed, refId:0x%" PRIx64, *job);
   }
 
-  schReleaseJob(job);
+  schReleaseJob(*job);
+  *job = 0;
 }
 
 void schedulerDestroy(void) {

source/libs/scheduler/test/schedulerTests.cpp

@@ -457,7 +457,7 @@ void schtFreeQueryJob(int32_t freeThread) {
   int64_t job = queryJobRefId;
   
   if (job && atomic_val_compare_exchange_64(&queryJobRefId, job, 0)) {
-    schedulerFreeJob(job, 0);
+    schedulerFreeJob(&job, 0);
     if (freeThread) {
       if (++freeNum % schtTestPrintNum == 0) {
         printf("FreeNum:%d\n", freeNum);
@@ -724,7 +724,7 @@ TEST(queryTest, normalCase) {
 
   schReleaseJob(job);
 
-  schedulerFreeJob(job, 0);
+  schedulerFreeJob(&job, 0);
 
   schtFreeQueryDag(&dag);
 
@@ -828,7 +828,7 @@ TEST(queryTest, readyFirstCase) {
 
   schReleaseJob(job);
 
-  schedulerFreeJob(job, 0);
+  schedulerFreeJob(&job, 0);
 
   schtFreeQueryDag(&dag);
 
@@ -940,7 +940,7 @@ TEST(queryTest, flowCtrlCase) {
 
   schReleaseJob(job);
 
-  schedulerFreeJob(job, 0);
+  schedulerFreeJob(&job, 0);
 
   schtFreeQueryDag(&dag);
 
@@ -994,7 +994,7 @@ TEST(insertTest, normalCase) {
   ASSERT_EQ(code, 0);
   ASSERT_EQ(res.numOfRows, 20);
 
-  schedulerFreeJob(insertJobRefId, 0);
+  schedulerFreeJob(&insertJobRefId, 0);
 
   schedulerDestroy();  
 }

source/libs/sync/src/syncMain.c

@@ -65,6 +65,7 @@ int32_t syncInit() {
       syncCleanUp();
       ret = -1;
     } else {
+      sDebug("sync rsetId:%" PRId64 " is open", tsNodeRefId);
       ret = syncEnvStart();
     }
   }
@@ -77,6 +78,7 @@ void syncCleanUp() {
   ASSERT(ret == 0);
 
   if (tsNodeRefId != -1) {
+    sDebug("sync rsetId:%" PRId64 " is closed", tsNodeRefId);
     taosCloseRef(tsNodeRefId);
     tsNodeRefId = -1;
   }
@@ -96,6 +98,7 @@ int64_t syncOpen(const SSyncInfo* pSyncInfo) {
     return -1;
   }
 
+  sDebug("vgId:%d, rid:%" PRId64 " is added to rsetId:%" PRId64, pSyncInfo->vgId, pSyncNode->rid, tsNodeRefId);
   return pSyncNode->rid;
 }
 
@@ -136,12 +139,14 @@ void syncStartStandBy(int64_t rid) {
 
 void syncStop(int64_t rid) {
   SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
-  if (pSyncNode != NULL) {
+  if (pSyncNode == NULL) return;
-    syncNodeClose(pSyncNode);
+
-  }
+  int32_t vgId = pSyncNode->vgId;
+  syncNodeClose(pSyncNode);
 
   taosReleaseRef(tsNodeRefId, pSyncNode->rid);
   taosRemoveRef(tsNodeRefId, rid);
+  sDebug("vgId:%d, rid:%" PRId64 " is removed from rsetId:%" PRId64, vgId, rid, tsNodeRefId);
 }
 
 int32_t syncSetStandby(int64_t rid) {

source/libs/tdb/src/db/tdbBtree.c

@@ -40,8 +40,8 @@ struct SBTree {
 #define TDB_BTREE_PAGE_IS_ROOT(PAGE)          (TDB_BTREE_PAGE_GET_FLAGS(PAGE) & TDB_BTREE_ROOT)
 #define TDB_BTREE_PAGE_IS_LEAF(PAGE)          (TDB_BTREE_PAGE_GET_FLAGS(PAGE) & TDB_BTREE_LEAF)
 #define TDB_BTREE_PAGE_IS_OVFL(PAGE)          (TDB_BTREE_PAGE_GET_FLAGS(PAGE) & TDB_BTREE_OVFL)
-#define TDB_BTREE_ASSERT_FLAG(flags)                                                 \
+#define TDB_BTREE_ASSERT_FLAG(flags)                                                     \
-  ASSERT(TDB_FLAG_IS(flags, TDB_BTREE_ROOT) || TDB_FLAG_IS(flags, TDB_BTREE_LEAF) || \
+  ASSERT(TDB_FLAG_IS(flags, TDB_BTREE_ROOT) || TDB_FLAG_IS(flags, TDB_BTREE_LEAF) ||     \
          TDB_FLAG_IS(flags, TDB_BTREE_ROOT | TDB_BTREE_LEAF) || TDB_FLAG_IS(flags, 0) || \
          TDB_FLAG_IS(flags, TDB_BTREE_OVFL))
 
@@ -58,7 +58,7 @@ typedef struct {
 
 static int tdbDefaultKeyCmprFn(const void *pKey1, int keyLen1, const void *pKey2, int keyLen2);
 static int tdbBtreeOpenImpl(SBTree *pBt);
-//static int tdbBtreeInitPage(SPage *pPage, void *arg, int init);
+// static int tdbBtreeInitPage(SPage *pPage, void *arg, int init);
 static int tdbBtreeEncodeCell(SPage *pPage, const void *pKey, int kLen, const void *pVal, int vLen, SCell *pCell,
                               int *szCell, TXN *pTxn, SBTree *pBt);
 static int tdbBtreeDecodeCell(SPage *pPage, const SCell *pCell, SCellDecoder *pDecoder, TXN *pTxn, SBTree *pBt);
@@ -321,7 +321,7 @@ static int tdbBtreeOpenImpl(SBTree *pBt) {
 
   {
     // 1. TODO: Search the main DB to check if the DB exists
-    ret = tdbPagerOpenDB(pBt->pPager, &pgno, true);
+    ret = tdbPagerOpenDB(pBt->pPager, &pgno, true, pBt);
     ASSERT(ret == 0);
   }
 
@@ -721,7 +721,8 @@ static int tdbBtreeBalanceNonRoot(SBTree *pBt, SPage *pParent, int idx, TXN *pTx
               int    szNewCell;
               SPgno  pgno;
               pgno = TDB_PAGE_PGNO(pNews[iNew]);
-              tdbBtreeEncodeCell(pParent, cd.pKey, cd.kLen, (void *)&pgno, sizeof(SPgno), pNewCell, &szNewCell, pTxn, pBt);
+              tdbBtreeEncodeCell(pParent, cd.pKey, cd.kLen, (void *)&pgno, sizeof(SPgno), pNewCell, &szNewCell, pTxn,
+                                 pBt);
               tdbPageInsertCell(pParent, sIdx++, pNewCell, szNewCell, 0);
               tdbOsFree(pNewCell);
             }
@@ -916,10 +917,10 @@ static int tdbBtreeEncodePayload(SPage *pPage, SCell *pCell, int nHeader, const
     int surplus = minLocal + (nPayload + nHeader - minLocal) % (maxLocal - sizeof(SPgno));
     int nLocal = surplus <= maxLocal ? surplus : minLocal;
 
-    //int ofpCap = tdbPageCapacity(pBt->pageSize, sizeof(SIntHdr));
+    // int ofpCap = tdbPageCapacity(pBt->pageSize, sizeof(SIntHdr));
 
     // fetch a new ofp and make it dirty
-    SPgno pgno = 0;
+    SPgno  pgno = 0;
     SPage *ofp, *nextOfp;
 
     ret = tdbFetchOvflPage(&pgno, &ofp, pTxn, pBt);
@@ -942,8 +943,8 @@ static int tdbBtreeEncodePayload(SPage *pPage, SCell *pCell, int nHeader, const
       nLeft -= kLen;
       // pack partial val to local if any space left
       if (nLocal > kLen + 4) {
-	memcpy(pCell + nHeader + kLen, pVal, nLocal - kLen - sizeof(SPgno));
+        memcpy(pCell + nHeader + kLen, pVal, nLocal - kLen - sizeof(SPgno));
-	nLeft -= nLocal - kLen - sizeof(SPgno);
+        nLeft -= nLocal - kLen - sizeof(SPgno);
       }
 
       // pack nextPgno
@@ -951,132 +952,132 @@ static int tdbBtreeEncodePayload(SPage *pPage, SCell *pCell, int nHeader, const
 
       // pack left val data to ovpages
       do {
-	lastPage = 0;
+        lastPage = 0;
-	if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
+        if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
-	  bytes = nLeft;
+          bytes = nLeft;
-	  lastPage = 1;
+          lastPage = 1;
-	} else {
+        } else {
-	  bytes = ofp->maxLocal - sizeof(SPgno);
+          bytes = ofp->maxLocal - sizeof(SPgno);
-	}
+        }
 
-	// fetch next ofp if not last page
+        // fetch next ofp if not last page
-	if (!lastPage) {
+        if (!lastPage) {
-	  // fetch a new ofp and make it dirty
+          // fetch a new ofp and make it dirty
-	  ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
+          ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
-	  if (ret < 0) {
+          if (ret < 0) {
-	    tdbFree(pBuf);
+            tdbFree(pBuf);
-	    return -1;
+            return -1;
-	  }
+          }
-	} else {
+        } else {
-	  pgno = 0;
+          pgno = 0;
-	}
+        }
 
-	memcpy(pBuf, ((SCell *)pVal) + vLen - nLeft, bytes);
+        memcpy(pBuf, ((SCell *)pVal) + vLen - nLeft, bytes);
-	memcpy(pBuf + bytes, &pgno, sizeof(pgno));
+        memcpy(pBuf + bytes, &pgno, sizeof(pgno));
 
-	ret = tdbPageInsertCell(ofp, 0, pBuf, bytes + sizeof(pgno), 0);
+        ret = tdbPageInsertCell(ofp, 0, pBuf, bytes + sizeof(pgno), 0);
-	if (ret < 0) {
+        if (ret < 0) {
-	  tdbFree(pBuf);
+          tdbFree(pBuf);
-	  return -1;
+          return -1;
-	}
+        }
 
-	ofp = nextOfp;
+        ofp = nextOfp;
-	nLeft -= bytes;
+        nLeft -= bytes;
       } while (nLeft > 0);
     } else {
       int nLeftKey = kLen;
       // pack partial key and nextPgno
       memcpy(pCell + nHeader, pKey, nLocal - 4);
       nLeft -= nLocal - 4;
-      nLeftKey -= nLocal -4;
+      nLeftKey -= nLocal - 4;
 
       memcpy(pCell + nHeader + nLocal - 4, &pgno, sizeof(pgno));
 
       int lastKeyPageSpace = 0;
       // pack left key & val to ovpages
       do {
-	// cal key to cpy
+        // cal key to cpy
-	int lastKeyPage = 0;
+        int lastKeyPage = 0;
-	if (nLeftKey <= ofp->maxLocal - sizeof(SPgno)) {
+        if (nLeftKey <= ofp->maxLocal - sizeof(SPgno)) {
-	  bytes = nLeftKey;
+          bytes = nLeftKey;
-	  lastKeyPage = 1;
+          lastKeyPage = 1;
-	  lastKeyPageSpace = ofp->maxLocal - sizeof(SPgno) - nLeftKey;
+          lastKeyPageSpace = ofp->maxLocal - sizeof(SPgno) - nLeftKey;
-	} else {
+        } else {
-	  bytes = ofp->maxLocal - sizeof(SPgno);
+          bytes = ofp->maxLocal - sizeof(SPgno);
-	}
+        }
 
-	// cpy key
+        // cpy key
-	memcpy(pBuf, ((SCell *)pKey) + kLen - nLeftKey, bytes);
+        memcpy(pBuf, ((SCell *)pKey) + kLen - nLeftKey, bytes);
 
-	if (lastKeyPage) {
+        if (lastKeyPage) {
-	  if (lastKeyPageSpace >= vLen) {
+          if (lastKeyPageSpace >= vLen) {
-	    memcpy(pBuf + kLen -nLeftKey, pVal, vLen);
+            memcpy(pBuf + kLen - nLeftKey, pVal, vLen);
 
-	    nLeft -= vLen;
+            nLeft -= vLen;
-	    pgno = 0;
+            pgno = 0;
-	  } else {
+          } else {
-	    memcpy(pBuf + kLen -nLeftKey, pVal, lastKeyPageSpace);
+            memcpy(pBuf + kLen - nLeftKey, pVal, lastKeyPageSpace);
-	    nLeft -= lastKeyPageSpace;
+            nLeft -= lastKeyPageSpace;
 
-	    // fetch next ofp, a new ofp and make it dirty
+            // fetch next ofp, a new ofp and make it dirty
-	    ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
+            ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
-	    if (ret < 0) {
+            if (ret < 0) {
-	      return -1;
+              return -1;
-	    }
+            }
-	  }
+          }
-	} else {
+        } else {
-	  // fetch next ofp, a new ofp and make it dirty
+          // fetch next ofp, a new ofp and make it dirty
-	  ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
+          ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
-	  if (ret < 0) {
+          if (ret < 0) {
-	    return -1;
+            return -1;
-	  }
+          }
-	}
+        }
 
-	memcpy(pBuf + kLen - nLeft, &pgno, sizeof(pgno));
+        memcpy(pBuf + kLen - nLeft, &pgno, sizeof(pgno));
 
-	ret = tdbPageInsertCell(ofp, 0, pBuf, bytes + sizeof(pgno), 0);
+        ret = tdbPageInsertCell(ofp, 0, pBuf, bytes + sizeof(pgno), 0);
-	if (ret < 0) {
+        if (ret < 0) {
-	  return -1;
+          return -1;
-	}
+        }
 
-	ofp = nextOfp;
+        ofp = nextOfp;
-	nLeftKey -= bytes;
+        nLeftKey -= bytes;
-	nLeft -= bytes;
+        nLeft -= bytes;
       } while (nLeftKey > 0);
 
       while (nLeft > 0) {
-	// pack left val data to ovpages
+        // pack left val data to ovpages
-	lastPage = 0;
+        lastPage = 0;
-	if (nLeft <= maxLocal - sizeof(SPgno)) {
+        if (nLeft <= maxLocal - sizeof(SPgno)) {
-	  bytes = nLeft;
+          bytes = nLeft;
-	  lastPage = 1;
+          lastPage = 1;
-	} else {
+        } else {
-	  bytes = maxLocal - sizeof(SPgno);
+          bytes = maxLocal - sizeof(SPgno);
-	}
+        }
 
-	// fetch next ofp if not last page
+        // fetch next ofp if not last page
-	if (!lastPage) {
+        if (!lastPage) {
-	  // fetch a new ofp and make it dirty
+          // fetch a new ofp and make it dirty
-	  ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
+          ret = tdbFetchOvflPage(&pgno, &nextOfp, pTxn, pBt);
-	  if (ret < 0) {
+          if (ret < 0) {
-	    tdbFree(pBuf);
+            tdbFree(pBuf);
-	    return -1;
+            return -1;
-	  }
+          }
-	} else {
+        } else {
-	  pgno = 0;
+          pgno = 0;
-	}
+        }
 
-	memcpy(pBuf, ((SCell *)pVal) + vLen - nLeft, bytes);
+        memcpy(pBuf, ((SCell *)pVal) + vLen - nLeft, bytes);
-	memcpy(pBuf + bytes, &pgno, sizeof(pgno));
+        memcpy(pBuf + bytes, &pgno, sizeof(pgno));
 
-	ret = tdbPageInsertCell(ofp, 0, pBuf, bytes + sizeof(pgno), 0);
+        ret = tdbPageInsertCell(ofp, 0, pBuf, bytes + sizeof(pgno), 0);
-	if (ret < 0) {
+        if (ret < 0) {
-	  tdbFree(pBuf);
+          tdbFree(pBuf);
-	  return -1;
+          return -1;
-	}
+        }
 
-	ofp = nextOfp;
+        ofp = nextOfp;
-	nLeft -= bytes;
+        nLeft -= bytes;
       }
     }
 
@@ -1142,7 +1143,8 @@ static int tdbBtreeEncodeCell(SPage *pPage, const void *pKey, int kLen, const vo
   return 0;
 }
 
-static int tdbBtreeDecodePayload(SPage *pPage, const SCell *pCell, int nHeader, SCellDecoder *pDecoder, TXN *pTxn, SBTree *pBt) {
+static int tdbBtreeDecodePayload(SPage *pPage, const SCell *pCell, int nHeader, SCellDecoder *pDecoder, TXN *pTxn,
+                                 SBTree *pBt) {
   int ret = 0;
   int nPayload;
   int maxLocal = pPage->maxLocal;
@@ -1171,149 +1173,149 @@ static int tdbBtreeDecodePayload(SPage *pPage, const SCell *pCell, int nHeader,
     int surplus = minLocal + (nPayload + nHeader - minLocal) % (maxLocal - sizeof(SPgno));
     int nLocal = surplus <= maxLocal ? surplus : minLocal;
 
-    int nLeft = nPayload;
+    int    nLeft = nPayload;
-    SPgno pgno = 0;
+    SPgno  pgno = 0;
     SPage *ofp;
     SCell *ofpCell;
-    int bytes;
+    int    bytes;
-    int lastPage = 0;
+    int    lastPage = 0;
 
     if (nLocal >= pDecoder->kLen + 4) {
       pDecoder->pKey = (SCell *)pCell + nHeader;
       nLeft -= kLen;
       if (nLocal > kLen + 4) {
-	// read partial val to local
+        // read partial val to local
-	pDecoder->pVal = tdbRealloc(pDecoder->pVal, vLen);
+        pDecoder->pVal = tdbRealloc(pDecoder->pVal, vLen);
-	if (pDecoder->pVal == NULL) {
+        if (pDecoder->pVal == NULL) {
-	  return -1;
+          return -1;
-	}
+        }
-	TDB_CELLDECODER_SET_FREE_VAL(pDecoder);
+        TDB_CELLDECODER_SET_FREE_VAL(pDecoder);
 
-	memcpy(pDecoder->pVal, pCell + nHeader + kLen, nLocal - kLen - sizeof(SPgno));
+        memcpy(pDecoder->pVal, pCell + nHeader + kLen, nLocal - kLen - sizeof(SPgno));
 
-	nLeft -= nLocal - kLen - sizeof(SPgno);
+        nLeft -= nLocal - kLen - sizeof(SPgno);
       }
 
       memcpy(&pgno, pCell + nHeader + nPayload - nLeft, sizeof(pgno));
 
       // unpack left val data from ovpages
-      while (pgno != 0)  {
+      while (pgno != 0) {
-	ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
+        ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
-	if (ret < 0) {
+        if (ret < 0) {
-	  return -1;
+          return -1;
-	}
+        }
 
-	ofpCell = tdbPageGetCell(ofp, 0);
+        ofpCell = tdbPageGetCell(ofp, 0);
 
-	if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
+        if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
-	  bytes = nLeft;
+          bytes = nLeft;
-	  lastPage = 1;
+          lastPage = 1;
-	} else {
+        } else {
-	  bytes = ofp->maxLocal - sizeof(SPgno);
+          bytes = ofp->maxLocal - sizeof(SPgno);
-	}
+        }
 
-	memcpy(pDecoder->pVal + vLen - nLeft, ofpCell, bytes);
+        memcpy(pDecoder->pVal + vLen - nLeft, ofpCell, bytes);
-	nLeft -= bytes;
+        nLeft -= bytes;
 
-	memcpy(&pgno, ofpCell + bytes, sizeof(pgno));
+        memcpy(&pgno, ofpCell + bytes, sizeof(pgno));
       }
     } else {
       int nLeftKey = kLen;
       // load partial key and nextPgno
       pDecoder->pKey = tdbRealloc(pDecoder->pKey, kLen);
       if (pDecoder->pKey == NULL) {
-	return -1;
+        return -1;
       }
       TDB_CELLDECODER_SET_FREE_KEY(pDecoder);
 
       memcpy(pDecoder->pKey, pCell + nHeader, nLocal - 4);
       nLeft -= nLocal - 4;
-      nLeftKey -= nLocal -4;
+      nLeftKey -= nLocal - 4;
 
       memcpy(&pgno, pCell + nHeader + nLocal - 4, sizeof(pgno));
 
       int lastKeyPageSpace = 0;
       // load left key & val to ovpages
       while (pgno != 0) {
-	ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
+        ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
-	if (ret < 0) {
+        if (ret < 0) {
-	  return -1;
+          return -1;
-	}
+        }
 
-	ofpCell = tdbPageGetCell(ofp, 0);
+        ofpCell = tdbPageGetCell(ofp, 0);
 
-	int lastKeyPage = 0;
+        int lastKeyPage = 0;
-	if (nLeftKey <= maxLocal - sizeof(SPgno)) {
+        if (nLeftKey <= maxLocal - sizeof(SPgno)) {
-	  bytes = nLeftKey;
+          bytes = nLeftKey;
-	  lastKeyPage = 1;
+          lastKeyPage = 1;
-	  lastKeyPageSpace = ofp->maxLocal - sizeof(SPgno) - nLeftKey;
+          lastKeyPageSpace = ofp->maxLocal - sizeof(SPgno) - nLeftKey;
-	} else {
+        } else {
-	  bytes = ofp->maxLocal - sizeof(SPgno);
+          bytes = ofp->maxLocal - sizeof(SPgno);
-	}
+        }
 
-	// cpy key
+        // cpy key
-	memcpy(pDecoder->pKey + kLen - nLeftKey, ofpCell, bytes);
+        memcpy(pDecoder->pKey + kLen - nLeftKey, ofpCell, bytes);
 
-	if (lastKeyPage) {
+        if (lastKeyPage) {
-	  if (lastKeyPageSpace >= vLen) {
+          if (lastKeyPageSpace >= vLen) {
-	    pDecoder->pVal = ofpCell + kLen -nLeftKey;
+            pDecoder->pVal = ofpCell + kLen - nLeftKey;
 
-	    nLeft -= vLen;
+            nLeft -= vLen;
-	    pgno = 0;
+            pgno = 0;
-	  } else {
+          } else {
-	    // read partial val to local
+            // read partial val to local
-	    pDecoder->pVal = tdbRealloc(pDecoder->pVal, vLen);
+            pDecoder->pVal = tdbRealloc(pDecoder->pVal, vLen);
-	    if (pDecoder->pVal == NULL) {
+            if (pDecoder->pVal == NULL) {
-	      return -1;
+              return -1;
-	    }
+            }
-	    TDB_CELLDECODER_SET_FREE_VAL(pDecoder);
+            TDB_CELLDECODER_SET_FREE_VAL(pDecoder);
 
-	    memcpy(pDecoder->pVal, ofpCell + kLen -nLeftKey, lastKeyPageSpace);
+            memcpy(pDecoder->pVal, ofpCell + kLen - nLeftKey, lastKeyPageSpace);
-	    nLeft -= lastKeyPageSpace;
+            nLeft -= lastKeyPageSpace;
-	  }
+          }
-	}
+        }
 
-	memcpy(&pgno, ofpCell + bytes, sizeof(pgno));
+        memcpy(&pgno, ofpCell + bytes, sizeof(pgno));
 
-	nLeftKey -= bytes;
+        nLeftKey -= bytes;
-	nLeft -= bytes;
+        nLeft -= bytes;
       }
 
       while (nLeft > 0) {
-	ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
+        ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
-	if (ret < 0) {
+        if (ret < 0) {
-	  return -1;
+          return -1;
-	}
+        }
 
-	ofpCell = tdbPageGetCell(ofp, 0);
+        ofpCell = tdbPageGetCell(ofp, 0);
 
-	// load left val data to ovpages
+        // load left val data to ovpages
-	lastPage = 0;
+        lastPage = 0;
-	if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
+        if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
-	  bytes = nLeft;
+          bytes = nLeft;
-	  lastPage = 1;
+          lastPage = 1;
-	} else {
+        } else {
-	  bytes = ofp->maxLocal - sizeof(SPgno);
+          bytes = ofp->maxLocal - sizeof(SPgno);
-	}
+        }
 
-	if (lastPage) {
+        if (lastPage) {
-	  pgno = 0;
+          pgno = 0;
-	}
+        }
 
-	if (!pDecoder->pVal) {
+        if (!pDecoder->pVal) {
-	    pDecoder->pVal = tdbRealloc(pDecoder->pVal, vLen);
+          pDecoder->pVal = tdbRealloc(pDecoder->pVal, vLen);
-	    if (pDecoder->pVal == NULL) {
+          if (pDecoder->pVal == NULL) {
-	      return -1;
+            return -1;
-	    }
+          }
-	    TDB_CELLDECODER_SET_FREE_VAL(pDecoder);
+          TDB_CELLDECODER_SET_FREE_VAL(pDecoder);
-	}
+        }
 
-	memcpy(pDecoder->pVal, ofpCell + vLen - nLeft, bytes);
+        memcpy(pDecoder->pVal, ofpCell + vLen - nLeft, bytes);
-	nLeft -= bytes;
+        nLeft -= bytes;
 
-	memcpy(&pgno, ofpCell + vLen - nLeft + bytes, sizeof(pgno));
+        memcpy(&pgno, ofpCell + vLen - nLeft + bytes, sizeof(pgno));
 
-	nLeft -= bytes;
+        nLeft -= bytes;
       }
     }
   }
@@ -1404,31 +1406,31 @@ static int tdbBtreeCellSize(const SPage *pPage, SCell *pCell, int dropOfp, TXN *
 
     // free ofp pages' cells
     if (dropOfp) {
-      int ret = 0;
+      int    ret = 0;
-      SPgno pgno = *(SPgno *) (pCell + nHeader + nLocal - sizeof(SPgno));
+      SPgno  pgno = *(SPgno *)(pCell + nHeader + nLocal - sizeof(SPgno));
-      int nLeft = nPayload - nLocal + sizeof(SPgno);
+      int    nLeft = nPayload - nLocal + sizeof(SPgno);
       SPage *ofp;
-      int bytes;
+      int    bytes;
 
       while (pgno != 0) {
-	ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
+        ret = tdbLoadOvflPage(&pgno, &ofp, pTxn, pBt);
-	if (ret < 0) {
+        if (ret < 0) {
-	  return -1;
+          return -1;
-	}
+        }
 
-	SCell *ofpCell = tdbPageGetCell(ofp, 0);
+        SCell *ofpCell = tdbPageGetCell(ofp, 0);
 
-	if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
+        if (nLeft <= ofp->maxLocal - sizeof(SPgno)) {
-	  bytes = nLeft;
+          bytes = nLeft;
-	} else {
+        } else {
-	  bytes = ofp->maxLocal - sizeof(SPgno);
+          bytes = ofp->maxLocal - sizeof(SPgno);
-	}
+        }
 
-	memcpy(&pgno, ofpCell + bytes, sizeof(pgno));
+        memcpy(&pgno, ofpCell + bytes, sizeof(pgno));
 
-	tdbPagerReturnPage(pPage->pPager, ofp, pTxn);
+        tdbPagerReturnPage(pPage->pPager, ofp, pTxn);
 
-	nLeft -= bytes;
+        nLeft -= bytes;
       }
     }
 
@@ -1932,7 +1934,7 @@ int tdbBtcUpsert(SBTC *pBtc, const void *pKey, int kLen, const void *pData, int
 
   // alloc space
   szBuf = kLen + nData + 14;
-  pBuf =  tdbRealloc(pBtc->pBt->pBuf, pBtc->pBt->pageSize > szBuf ? szBuf : pBtc->pBt->pageSize);
+  pBuf = tdbRealloc(pBtc->pBt->pBuf, pBtc->pBt->pageSize > szBuf ? szBuf : pBtc->pBt->pageSize);
   if (pBuf == NULL) {
     ASSERT(0);
     return -1;

source/libs/tdb/src/db/tdbPager.c

@@ -98,7 +98,7 @@ int tdbPagerClose(SPager *pPager) {
   return 0;
 }
 
-int tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate) {
+int tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate, SBTree *pBt) {
   SPgno  pgno;
   SPage *pPage;
   int    ret;
@@ -110,25 +110,41 @@ int tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate) {
   }
 
   {
-      // TODO: try to search the main DB to get the page number
+    // TODO: try to search the main DB to get the page number
-      // pgno = 0;
+    // pgno = 0;
   }
 
-      // if (pgno == 0 && toCreate) {
+  if (pgno == 0 && toCreate) {
-      //   ret = tdbPagerAllocPage(pPager, &pPage, &pgno);
+    // allocate a new child page
-      //   if (ret < 0) {
+    TXN txn;
-      //     return -1;
+    tdbTxnOpen(&txn, 0, tdbDefaultMalloc, tdbDefaultFree, NULL, 0);
-      //   }
 
-      //   // TODO: Need to zero the page
+    pPager->inTran = 1;
 
-      //   ret = tdbPagerWrite(pPager, pPage);
+    SBtreeInitPageArg zArg;
-      //   if (ret < 0) {
+    zArg.flags = 0x1 | 0x2;  // root leaf node;
-      //     return -1;
+    zArg.pBt = pBt;
-      //   }
+    ret = tdbPagerFetchPage(pPager, &pgno, &pPage, tdbBtreeInitPage, &zArg, &txn);
-      // }
+    if (ret < 0) {
+      return -1;
+    }
 
-      *ppgno = pgno;
+    //    ret = tdbPagerAllocPage(pPager, &pPage, &pgno);
+    // if (ret < 0) {
+    //  return -1;
+    //}
+
+    // TODO: Need to zero the page
+
+    ret = tdbPagerWrite(pPager, pPage);
+    if (ret < 0) {
+      return -1;
+    }
+
+    tdbTxnClose(&txn);
+  }
+
+  *ppgno = pgno;
   return 0;
 }
 
@@ -427,9 +443,9 @@ static int tdbPagerWritePageToDB(SPager *pPager, SPage *pPage) {
 }
 
 int tdbPagerRestore(SPager *pPager, SBTree *pBt) {
-  int ret = 0;
+  int   ret = 0;
   SPgno journalSize = 0;
-  u8 *pageBuf = NULL;
+  u8   *pageBuf = NULL;
 
   tdb_fd_t jfd = tdbOsOpen(pPager->jFileName, TDB_O_RDWR, 0755);
   if (jfd == NULL) {
@@ -454,7 +470,7 @@ int tdbPagerRestore(SPager *pPager, SBTree *pBt) {
 
   for (int pgIndex = 0; pgIndex < journalSize; ++pgIndex) {
     // read pgno & the page from journal
-    SPgno pgno;
+    SPgno  pgno;
     SPage *pPage;
 
     int ret = tdbOsRead(jfd, &pgno, sizeof(pgno));

source/libs/tdb/src/inc/tdbInt.h

@@ -128,13 +128,13 @@ typedef struct SBtInfo {
 #define TDB_CELLDECODER_FREE_VAL(pCellDecoder) ((pCellDecoder)->freeKV & TDB_CELLD_F_VAL)
 
 typedef struct {
-  int       kLen;
+  int   kLen;
-  u8       *pKey;
+  u8   *pKey;
-  int       vLen;
+  int   vLen;
-  u8       *pVal;
+  u8   *pVal;
-  SPgno     pgno;
+  SPgno pgno;
-  u8       *pBuf;
+  u8   *pBuf;
-  u8        freeKV;
+  u8    freeKV;
 } SCellDecoder;
 
 struct SBTC {
@@ -184,7 +184,7 @@ int tdbBtcUpsert(SBTC *pBtc, const void *pKey, int kLen, const void *pData, int
 
 int  tdbPagerOpen(SPCache *pCache, const char *fileName, SPager **ppPager);
 int  tdbPagerClose(SPager *pPager);
-int  tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate);
+int  tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate, SBTree *pBt);
 int  tdbPagerWrite(SPager *pPager, SPage *pPage);
 int  tdbPagerBegin(SPager *pPager, TXN *pTxn);
 int  tdbPagerCommit(SPager *pPager, TXN *pTxn);
@@ -192,7 +192,7 @@ int  tdbPagerFetchPage(SPager *pPager, SPgno *ppgno, SPage **ppPage, int (*initP
                        TXN *pTxn);
 void tdbPagerReturnPage(SPager *pPager, SPage *pPage, TXN *pTxn);
 int  tdbPagerAllocPage(SPager *pPager, SPgno *ppgno);
-int tdbPagerRestore(SPager *pPager, SBTree *pBt);
+int  tdbPagerRestore(SPager *pPager, SBTree *pBt);
 
 // tdbPCache.c ====================================
 #define TDB_PCACHE_PAGE    \
@@ -314,19 +314,18 @@ static inline int tdbTryLockPage(tdb_spinlock_t *pLock) {
 #define TDB_TRY_LOCK_PAGE(pPage)     tdbTryLockPage(&((pPage)->lock))
 
 // APIs
-#define TDB_PAGE_TOTAL_CELLS(pPage)        ((pPage)->nOverflow + (pPage)->pPageMethods->getCellNum(pPage))
+#define TDB_PAGE_TOTAL_CELLS(pPage) ((pPage)->nOverflow + (pPage)->pPageMethods->getCellNum(pPage))
-#define TDB_PAGE_USABLE_SIZE(pPage)        ((u8 *)(pPage)->pPageFtr - (pPage)->pCellIdx)
+#define TDB_PAGE_USABLE_SIZE(pPage) ((u8 *)(pPage)->pPageFtr - (pPage)->pCellIdx)
-#define TDB_PAGE_FREE_SIZE(pPage)          (*(pPage)->pPageMethods->getFreeBytes)(pPage)
+#define TDB_PAGE_FREE_SIZE(pPage)   (*(pPage)->pPageMethods->getFreeBytes)(pPage)
-#define TDB_PAGE_PGNO(pPage)               ((pPage)->pgid.pgno)
+#define TDB_PAGE_PGNO(pPage)        ((pPage)->pgid.pgno)
-#define TDB_BYTES_CELL_TAKEN(pPage, pCell) ((*(pPage)->xCellSize)(pPage, pCell, 0, NULL, NULL) + (pPage)->pPageMethods->szOffset)
+#define TDB_BYTES_CELL_TAKEN(pPage, pCell) \
-#define TDB_PAGE_OFFSET_SIZE(pPage)        ((pPage)->pPageMethods->szOffset)
+  ((*(pPage)->xCellSize)(pPage, pCell, 0, NULL, NULL) + (pPage)->pPageMethods->szOffset)
+#define TDB_PAGE_OFFSET_SIZE(pPage) ((pPage)->pPageMethods->szOffset)
 
 int  tdbPageCreate(int pageSize, SPage **ppPage, void *(*xMalloc)(void *, size_t), void *arg);
 int  tdbPageDestroy(SPage *pPage, void (*xFree)(void *arg, void *ptr), void *arg);
-void tdbPageZero(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *, int,
+void tdbPageZero(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *, int, TXN *, SBTree *pBt));
-							    TXN *, SBTree *pBt));
+void tdbPageInit(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *, int, TXN *, SBTree *pBt));
-void tdbPageInit(SPage *pPage, u8 szAmHdr, int (*xCellSize)(const SPage *, SCell *, int,
-							    TXN *, SBTree *pBt));
 int  tdbPageInsertCell(SPage *pPage, int idx, SCell *pCell, int szCell, u8 asOvfl);
 int  tdbPageDropCell(SPage *pPage, int idx, TXN *pTxn, SBTree *pBt);
 int  tdbPageUpdateCell(SPage *pPage, int idx, SCell *pCell, int szCell, TXN *pTxn, SBTree *pBt);

source/libs/tfs/src/tfs.c

@@ -291,7 +291,7 @@ int32_t tfsRmdir(STfs *pTfs, const char *rname) {
     for (int32_t id = 0; id < pTier->ndisk; id++) {
       STfsDisk *pDisk = pTier->disks[id];
       snprintf(aname, TMPNAME_LEN, "%s%s%s", pDisk->path, TD_DIRSEP, rname);
-      uInfo("====> tfs remove dir : path:%s aname:%s rname:[%s]", pDisk->path, aname, rname);
+      uInfo("tfs remove dir : path:%s aname:%s rname:[%s]", pDisk->path, aname, rname);
       taosRemoveDir(aname);
     }
   }

source/libs/transport/inc/transComm.h

@@ -253,7 +253,7 @@ int         transAsyncSend(SAsyncPool* pool, queue* mq);
   do {                                                                                                             \
     if (id > 0) {                                                                                                  \
       tTrace("handle step1");                                                                                      \
-      SExHandle* exh2 = transAcquireExHandle(id);                                                                  \
+      SExHandle* exh2 = transAcquireExHandle(transGetRefMgt(), id);                                                \
       if (exh2 == NULL || id != exh2->refId) {                                                                     \
         tTrace("handle %p except, may already freed, ignore msg, ref1: %" PRIu64 ", ref2 : %" PRIu64 "", exh1,     \
                exh2 ? exh2->refId : 0, id);                                                                        \
@@ -261,7 +261,7 @@ int         transAsyncSend(SAsyncPool* pool, queue* mq);
       }                                                                                                            \
     } else if (id == 0) {                                                                                          \
       tTrace("handle step2");                                                                                      \
-      SExHandle* exh2 = transAcquireExHandle(id);                                                                  \
+      SExHandle* exh2 = transAcquireExHandle(transGetRefMgt(), id);                                                \
       if (exh2 == NULL || id == exh2->refId) {                                                                     \
         tTrace("handle %p except, may already freed, ignore msg, ref1: %" PRIu64 ", ref2 : %" PRIu64 "", exh1, id, \
                exh2 ? exh2->refId : 0);                                                                            \
@@ -391,13 +391,16 @@ void transThreadOnce();
 void transInit();
 void transCleanup();
 
-int32_t    transOpenExHandleMgt(int size);
+int32_t transOpenRefMgt(int size, void (*func)(void*));
-void       transCloseExHandleMgt();
+void    transCloseRefMgt(int32_t refMgt);
-int64_t    transAddExHandle(void* p);
+int64_t transAddExHandle(int32_t refMgt, void* p);
-int32_t    transRemoveExHandle(int64_t refId);
+int32_t transRemoveExHandle(int32_t refMgt, int64_t refId);
-SExHandle* transAcquireExHandle(int64_t refId);
+void*   transAcquireExHandle(int32_t refMgt, int64_t refId);
-int32_t    transReleaseExHandle(int64_t refId);
+int32_t transReleaseExHandle(int32_t refMgt, int64_t refId);
-void       transDestoryExHandle(void* handle);
+void    transDestoryExHandle(void* handle);
+
+int32_t transGetRefMgt();
+int32_t transGetInstMgt();
 
 #ifdef __cplusplus
 }

source/libs/transport/inc/transportInt.h

@@ -57,7 +57,7 @@ typedef struct {
 
   void*         parent;
   void*         tcphandle;  // returned handle from TCP initialization
-  int32_t       refMgt;
+  int64_t       refId;
   TdThreadMutex mutex;
 } SRpcInfo;
 

source/libs/transport/src/trans.c

@@ -76,16 +76,23 @@ void* rpcOpen(const SRpcInit* pInit) {
   if (pInit->user) {
     memcpy(pRpc->user, pInit->user, strlen(pInit->user));
   }
-  return pRpc;
+
+  int64_t refId = transAddExHandle(transGetInstMgt(), pRpc);
+  transAcquireExHandle(transGetInstMgt(), refId);
+  pRpc->refId = refId;
+  return (void*)refId;
 }
 void rpcClose(void* arg) {
   tInfo("start to close rpc");
+  transRemoveExHandle(transGetInstMgt(), (int64_t)arg);
+  transReleaseExHandle(transGetInstMgt(), (int64_t)arg);
+  tInfo("finish to close rpc");
+  return;
+}
+void rpcCloseImpl(void* arg) {
   SRpcInfo* pRpc = (SRpcInfo*)arg;
   (*taosCloseHandle[pRpc->connType])(pRpc->tcphandle);
   taosMemoryFree(pRpc);
-  tInfo("finish to close rpc");
-
-  return;
 }
 
 void* rpcMallocCont(int32_t contLen) {
@@ -140,11 +147,10 @@ void rpcSendRecv(void* shandle, SEpSet* pEpSet, SRpcMsg* pMsg, SRpcMsg* pRsp) {
   transSendRecv(shandle, pEpSet, pMsg, pRsp);
 }
 
-void    rpcSendResponse(const SRpcMsg* pMsg) { transSendResponse(pMsg); }
+void rpcSendResponse(const SRpcMsg* pMsg) { transSendResponse(pMsg); }
 
 int32_t rpcGetConnInfo(void* thandle, SRpcConnInfo* pInfo) { return 0; }
 
-
 void rpcRefHandle(void* handle, int8_t type) {
   assert(type == TAOS_CONN_SERVER || type == TAOS_CONN_CLIENT);
   (*taosRefHandle[type])(handle);

source/libs/transport/src/transCli.c

@@ -47,6 +47,7 @@ typedef struct SCliMsg {
   queue          q;
   STransMsgType  type;
 
+  int64_t  refId;
   uint64_t st;
   int      sent;  //(0: no send, 1: alread sent)
 } SCliMsg;
@@ -262,13 +263,17 @@ static void cliReleaseUnfinishedMsg(SCliConn* conn) {
 #define REQUEST_PERSIS_HANDLE(msg)   ((msg)->info.persistHandle == 1)
 #define REQUEST_RELEASE_HANDLE(cmsg) ((cmsg)->type == Release)
 
+#define EPSET_IS_VALID(epSet)       ((epSet) != NULL && (epSet)->numOfEps != 0)
 #define EPSET_GET_SIZE(epSet)       (epSet)->numOfEps
 #define EPSET_GET_INUSE_IP(epSet)   ((epSet)->eps[(epSet)->inUse].fqdn)
 #define EPSET_GET_INUSE_PORT(epSet) ((epSet)->eps[(epSet)->inUse].port)
-#define EPSET_FORWARD_INUSE(epSet)                               \
+#define EPSET_FORWARD_INUSE(epSet)                                 \
-  do {                                                           \
+  do {                                                             \
-    (epSet)->inUse = (++((epSet)->inUse)) % ((epSet)->numOfEps); \
+    if ((epSet)->numOfEps != 0) {                                  \
+      (epSet)->inUse = (++((epSet)->inUse)) % ((epSet)->numOfEps); \
+    }                                                              \
   } while (0)
+
 #define EPSET_DEBUG_STR(epSet, tbuf)                                                                                   \
   do {                                                                                                                 \
     int len = snprintf(tbuf, sizeof(tbuf), "epset:{");                                                                 \
@@ -501,18 +506,17 @@ static SCliConn* getConnFromPool(void* pool, char* ip, uint32_t port) {
 }
 static void allocConnRef(SCliConn* conn, bool update) {
   if (update) {
-    transRemoveExHandle(conn->refId);
+    transRemoveExHandle(transGetRefMgt(), conn->refId);
     conn->refId = -1;
   }
   SExHandle* exh = taosMemoryCalloc(1, sizeof(SExHandle));
   exh->handle = conn;
   exh->pThrd = conn->hostThrd;
-  exh->refId = transAddExHandle(exh);
+  exh->refId = transAddExHandle(transGetRefMgt(), exh);
   conn->refId = exh->refId;
 }
 static void addConnToPool(void* pool, SCliConn* conn) {
   if (conn->status == ConnInPool) {
-    // assert(0);
     return;
   }
   SCliThrd* thrd = conn->hostThrd;
@@ -601,16 +605,14 @@ static void cliDestroyConn(SCliConn* conn, bool clear) {
   tTrace("%s conn %p remove from conn pool", CONN_GET_INST_LABEL(conn), conn);
   QUEUE_REMOVE(&conn->conn);
   QUEUE_INIT(&conn->conn);
-  transRemoveExHandle(conn->refId);
+  transRemoveExHandle(transGetRefMgt(), conn->refId);
   conn->refId = -1;
 
   if (clear) {
     if (!uv_is_closing((uv_handle_t*)conn->stream)) {
+      uv_read_stop(conn->stream);
       uv_close((uv_handle_t*)conn->stream, cliDestroy);
     }
-    //} else {
-    //  cliDestroy((uv_handle_t*)conn->stream);
-    //}
   }
 }
 static void cliDestroy(uv_handle_t* handle) {
@@ -619,7 +621,7 @@ static void cliDestroy(uv_handle_t* handle) {
   }
 
   SCliConn* conn = handle->data;
-  transRemoveExHandle(conn->refId);
+  transRemoveExHandle(transGetRefMgt(), conn->refId);
   taosMemoryFree(conn->ip);
   conn->stream->data = NULL;
   taosMemoryFree(conn->stream);
@@ -635,7 +637,6 @@ static bool cliHandleNoResp(SCliConn* conn) {
     SCliMsg* pMsg = transQueueGet(&conn->cliMsgs, 0);
     if (REQUEST_NO_RESP(&pMsg->msg)) {
       transQueuePop(&conn->cliMsgs);
-      // taosArrayRemove(msgs, 0);
       destroyCmsg(pMsg);
       res = true;
     }
@@ -668,7 +669,6 @@ static void cliSendCb(uv_write_t* req, int status) {
 void cliSend(SCliConn* pConn) {
   CONN_HANDLE_BROKEN(pConn);
 
-  // assert(taosArrayGetSize(pConn->cliMsgs) > 0);
   assert(!transQueueEmpty(&pConn->cliMsgs));
 
   SCliMsg* pCliMsg = NULL;
@@ -747,7 +747,7 @@ static void cliHandleQuit(SCliMsg* pMsg, SCliThrd* pThrd) {
 }
 static void cliHandleRelease(SCliMsg* pMsg, SCliThrd* pThrd) {
   int64_t    refId = (int64_t)(pMsg->msg.info.handle);
-  SExHandle* exh = transAcquireExHandle(refId);
+  SExHandle* exh = transAcquireExHandle(transGetRefMgt(), refId);
   if (exh == NULL) {
     tDebug("%" PRId64 " already release", refId);
   }
@@ -773,14 +773,14 @@ SCliConn* cliGetConn(SCliMsg* pMsg, SCliThrd* pThrd, bool* ignore) {
   SCliConn* conn = NULL;
   int64_t   refId = (int64_t)(pMsg->msg.info.handle);
   if (refId != 0) {
-    SExHandle* exh = transAcquireExHandle(refId);
+    SExHandle* exh = transAcquireExHandle(transGetRefMgt(), refId);
     if (exh == NULL) {
       *ignore = true;
       destroyCmsg(pMsg);
       return NULL;
     } else {
       conn = exh->handle;
-      transReleaseExHandle(refId);
+      transReleaseExHandle(transGetRefMgt(), refId);
     }
     return conn;
   };
@@ -810,6 +810,11 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrd* pThrd) {
   STrans*        pTransInst = pThrd->pTransInst;
 
   cliMayCvtFqdnToIp(&pCtx->epSet, &pThrd->cvtAddr);
+  if (!EPSET_IS_VALID(&pCtx->epSet)) {
+    destroyCmsg(pMsg);
+    tError("invalid epset");
+    return;
+  }
 
   bool      ignore = false;
   SCliConn* conn = cliGetConn(pMsg, pThrd, &ignore);
@@ -977,6 +982,7 @@ void cliSendQuit(SCliThrd* thrd) {
 }
 void cliWalkCb(uv_handle_t* handle, void* arg) {
   if (!uv_is_closing(handle)) {
+    uv_read_stop((uv_stream_t*)handle);
     uv_close(handle, cliDestroy);
   }
 }
@@ -1066,7 +1072,7 @@ int cliAppCb(SCliConn* pConn, STransMsg* pResp, SCliMsg* pMsg) {
   if (retry) {
     pMsg->sent = 0;
     pCtx->retryCnt += 1;
-    if (code == TSDB_CODE_RPC_NETWORK_UNAVAIL) {
+    if (code == TSDB_CODE_RPC_NETWORK_UNAVAIL || code == TSDB_CODE_RPC_BROKEN_LINK) {
       cliCompareAndSwap(&pCtx->retryLimit, TRANS_RETRY_COUNT_LIMIT, EPSET_GET_SIZE(&pCtx->epSet) * 3);
       if (pCtx->retryCnt < pCtx->retryLimit) {
         transUnrefCliHandle(pConn);
@@ -1077,12 +1083,14 @@ int cliAppCb(SCliConn* pConn, STransMsg* pResp, SCliMsg* pMsg) {
     } else {
       cliCompareAndSwap(&pCtx->retryLimit, TRANS_RETRY_COUNT_LIMIT, TRANS_RETRY_COUNT_LIMIT);
       if (pCtx->retryCnt < pCtx->retryLimit) {
-        addConnToPool(pThrd->pool, pConn);
         if (pResp->contLen == 0) {
           EPSET_FORWARD_INUSE(&pCtx->epSet);
         } else {
-          tDeserializeSEpSet(pResp->pCont, pResp->contLen, &pCtx->epSet);
+          if (tDeserializeSEpSet(pResp->pCont, pResp->contLen, &pCtx->epSet) < 0) {
+            tError("%s conn %p failed to deserialize epset", CONN_GET_INST_LABEL(pConn));
+          }
         }
+        addConnToPool(pThrd->pool, pConn);
         transFreeMsg(pResp->pCont);
         cliSchedMsgToNextNode(pMsg, pThrd);
         return -1;
@@ -1152,12 +1160,12 @@ void transUnrefCliHandle(void* handle) {
 }
 SCliThrd* transGetWorkThrdFromHandle(int64_t handle) {
   SCliThrd*  pThrd = NULL;
-  SExHandle* exh = transAcquireExHandle(handle);
+  SExHandle* exh = transAcquireExHandle(transGetRefMgt(), handle);
   if (exh == NULL) {
     return NULL;
   }
   pThrd = exh->pThrd;
-  transReleaseExHandle(handle);
+  transReleaseExHandle(transGetRefMgt(), handle);
   return pThrd;
 }
 SCliThrd* transGetWorkThrd(STrans* trans, int64_t handle) {
@@ -1184,10 +1192,13 @@ void transReleaseCliHandle(void* handle) {
 }
 
 void transSendRequest(void* shandle, const SEpSet* pEpSet, STransMsg* pReq, STransCtx* ctx) {
-  STrans*   pTransInst = (STrans*)shandle;
+  STrans* pTransInst = (STrans*)transAcquireExHandle(transGetInstMgt(), (int64_t)shandle);
+  if (pTransInst == NULL) return;
+
   SCliThrd* pThrd = transGetWorkThrd(pTransInst, (int64_t)pReq->info.handle);
   if (pThrd == NULL) {
     transFreeMsg(pReq->pCont);
+    transReleaseExHandle(transGetInstMgt(), (int64_t)shandle);
     return;
   }
 
@@ -1208,19 +1219,24 @@ void transSendRequest(void* shandle, const SEpSet* pEpSet, STransMsg* pReq, STra
   cliMsg->msg = *pReq;
   cliMsg->st = taosGetTimestampUs();
   cliMsg->type = Normal;
+  cliMsg->refId = (int64_t)shandle;
 
   STraceId* trace = &pReq->info.traceId;
   tGTrace("%s send request at thread:%08" PRId64 ", dst: %s:%d, app:%p", transLabel(pTransInst), pThrd->pid,
           EPSET_GET_INUSE_IP(&pCtx->epSet), EPSET_GET_INUSE_PORT(&pCtx->epSet), pReq->info.ahandle);
   ASSERT(transAsyncSend(pThrd->asyncPool, &(cliMsg->q)) == 0);
+  transReleaseExHandle(transGetInstMgt(), (int64_t)shandle);
   return;
 }
 
 void transSendRecv(void* shandle, const SEpSet* pEpSet, STransMsg* pReq, STransMsg* pRsp) {
-  STrans*   pTransInst = (STrans*)shandle;
+  STrans* pTransInst = (STrans*)transAcquireExHandle(transGetInstMgt(), (int64_t)shandle);
+  if (pTransInst == NULL) return;
+
   SCliThrd* pThrd = transGetWorkThrd(pTransInst, (int64_t)pReq->info.handle);
   if (pThrd == NULL) {
     transFreeMsg(pReq->pCont);
+    transReleaseExHandle(transGetInstMgt(), (int64_t)shandle);
     return;
   }
   tsem_t* sem = taosMemoryCalloc(1, sizeof(tsem_t));
@@ -1241,6 +1257,7 @@ void transSendRecv(void* shandle, const SEpSet* pEpSet, STransMsg* pReq, STransM
   cliMsg->msg = *pReq;
   cliMsg->st = taosGetTimestampUs();
   cliMsg->type = Normal;
+  cliMsg->refId = (int64_t)shandle;
 
   STraceId* trace = &pReq->info.traceId;
   tGTrace("%s send request at thread:%08" PRId64 ", dst: %s:%d, app:%p", transLabel(pTransInst), pThrd->pid,
@@ -1250,13 +1267,16 @@ void transSendRecv(void* shandle, const SEpSet* pEpSet, STransMsg* pReq, STransM
   tsem_wait(sem);
   tsem_destroy(sem);
   taosMemoryFree(sem);
+
+  transReleaseExHandle(transGetInstMgt(), (int64_t)shandle);
   return;
 }
 /*
  *
  **/
 void transSetDefaultAddr(void* shandle, const char* ip, const char* fqdn) {
-  STrans* pTransInst = shandle;
+  STrans* pTransInst = (STrans*)transAcquireExHandle(transGetInstMgt(), (int64_t)shandle);
+  if (pTransInst == NULL) return;
 
   SCvtAddr cvtAddr = {0};
   if (ip != NULL && fqdn != NULL) {
@@ -1271,11 +1291,13 @@ void transSetDefaultAddr(void* shandle, const char* ip, const char* fqdn) {
     SCliMsg* cliMsg = taosMemoryCalloc(1, sizeof(SCliMsg));
     cliMsg->ctx = pCtx;
     cliMsg->type = Update;
+    cliMsg->refId = (int64_t)shandle;
 
     SCliThrd* thrd = ((SCliObj*)pTransInst->tcphandle)->pThreadObj[i];
     tDebug("%s update epset at thread:%08" PRId64 "", pTransInst->label, thrd->pid);
 
     transAsyncSend(thrd->asyncPool, &(cliMsg->q));
   }
+  transReleaseExHandle(transGetInstMgt(), (int64_t)shandle);
 }
 #endif

source/libs/transport/src/transComm.c

@@ -19,6 +19,7 @@
 static TdThreadOnce transModuleInit = PTHREAD_ONCE_INIT;
 
 static int32_t refMgt;
+static int32_t instMgt;
 
 int transAuthenticateMsg(void* pMsg, int msgLen, void* pAuth, void* pKey) {
   T_MD5_CTX context;
@@ -479,53 +480,54 @@ bool transEpSetIsEqual(SEpSet* a, SEpSet* b) {
   }
   return true;
 }
-static int32_t transGetRefMgt() {
-  //
-  return refMgt;
-}
 
 static void transInitEnv() {
-  refMgt = transOpenExHandleMgt(50000);
+  refMgt = transOpenRefMgt(50000, transDestoryExHandle);
+  instMgt = taosOpenRef(50, rpcCloseImpl);
   uv_os_setenv("UV_TCP_SINGLE_ACCEPT", "1");
 }
 static void transDestroyEnv() {
-  // close ref
+  transCloseRefMgt(refMgt);
-  transCloseExHandleMgt();
+  transCloseRefMgt(instMgt);
 }
 
 void transInit() {
   // init env
   taosThreadOnce(&transModuleInit, transInitEnv);
 }
+
+int32_t transGetRefMgt() { return refMgt; }
+int32_t transGetInstMgt() { return instMgt; }
+
 void transCleanup() {
   // clean env
   transDestroyEnv();
 }
-int32_t transOpenExHandleMgt(int size) {
+int32_t transOpenRefMgt(int size, void (*func)(void*)) {
   // added into once later
-  return taosOpenRef(size, transDestoryExHandle);
+  return taosOpenRef(size, func);
 }
-void transCloseExHandleMgt() {
+void transCloseRefMgt(int32_t mgt) {
   // close ref
-  taosCloseRef(transGetRefMgt());
+  taosCloseRef(mgt);
 }
-int64_t transAddExHandle(void* p) {
+int64_t transAddExHandle(int32_t refMgt, void* p) {
   // acquire extern handle
-  return taosAddRef(transGetRefMgt(), p);
+  return taosAddRef(refMgt, p);
 }
-int32_t transRemoveExHandle(int64_t refId) {
+int32_t transRemoveExHandle(int32_t refMgt, int64_t refId) {
   // acquire extern handle
-  return taosRemoveRef(transGetRefMgt(), refId);
+  return taosRemoveRef(refMgt, refId);
 }
 
-SExHandle* transAcquireExHandle(int64_t refId) {
+void* transAcquireExHandle(int32_t refMgt, int64_t refId) {
   // acquire extern handle
-  return (SExHandle*)taosAcquireRef(transGetRefMgt(), refId);
+  return (void*)taosAcquireRef(refMgt, refId);
 }
 
-int32_t transReleaseExHandle(int64_t refId) {
+int32_t transReleaseExHandle(int32_t refMgt, int64_t refId) {
   // release extern handle
-  return taosReleaseRef(transGetRefMgt(), refId);
+  return taosReleaseRef(refMgt, refId);
 }
 void transDestoryExHandle(void* handle) {
   if (handle == NULL) {

source/libs/transport/src/transSvr.c

@@ -261,7 +261,7 @@ static void uvHandleReq(SSvrConn* pConn) {
   // 2. once send out data, cli conn released to conn pool immediately
   // 3. not mixed with persist
   transMsg.info.ahandle = (void*)pHead->ahandle;
-  transMsg.info.handle = (void*)transAcquireExHandle(pConn->refId);
+  transMsg.info.handle = (void*)transAcquireExHandle(transGetRefMgt(), pConn->refId);
   transMsg.info.refId = pConn->refId;
   transMsg.info.traceId = pHead->traceId;
 
@@ -279,7 +279,7 @@ static void uvHandleReq(SSvrConn* pConn) {
   pConnInfo->clientPort = ntohs(pConn->addr.sin_port);
   tstrncpy(pConnInfo->user, pConn->user, sizeof(pConnInfo->user));
 
-  transReleaseExHandle(pConn->refId);
+  transReleaseExHandle(transGetRefMgt(), pConn->refId);
 
   STrans* pTransInst = pConn->pTransInst;
   (*pTransInst->cfp)(pTransInst->parent, &transMsg, NULL);
@@ -402,8 +402,9 @@ static void uvPrepareSendData(SSvrMsg* smsg, uv_buf_t* wb) {
     } else {
       pHead->msgType = pMsg->msgType;
       // set up resp msg type
-      if (pHead->msgType == 0 && transMsgLenFromCont(pMsg->contLen) == sizeof(STransMsgHead))
+      if (pHead->msgType == 0 && transMsgLenFromCont(pMsg->contLen) == sizeof(STransMsgHead)) {
         pHead->msgType = pConn->inType + 1;
+      }
     }
   }
 
@@ -507,15 +508,15 @@ void uvWorkerAsyncCb(uv_async_t* handle) {
 
       SExHandle* exh1 = transMsg.info.handle;
       int64_t    refId = transMsg.info.refId;
-      SExHandle* exh2 = transAcquireExHandle(refId);
+      SExHandle* exh2 = transAcquireExHandle(transGetRefMgt(), refId);
       if (exh2 == NULL || exh1 != exh2) {
         tTrace("handle except msg %p, ignore it", exh1);
-        transReleaseExHandle(refId);
+        transReleaseExHandle(transGetRefMgt(), refId);
         destroySmsg(msg);
         continue;
       }
       msg->pConn = exh1->handle;
-      transReleaseExHandle(refId);
+      transReleaseExHandle(transGetRefMgt(), refId);
       (*transAsyncHandle[msg->type])(msg, pThrd);
     }
   }
@@ -757,8 +758,8 @@ static SSvrConn* createConn(void* hThrd) {
   SExHandle* exh = taosMemoryMalloc(sizeof(SExHandle));
   exh->handle = pConn;
   exh->pThrd = pThrd;
-  exh->refId = transAddExHandle(exh);
+  exh->refId = transAddExHandle(transGetRefMgt(), exh);
-  transAcquireExHandle(exh->refId);
+  transAcquireExHandle(transGetRefMgt(), exh->refId);
 
   pConn->refId = exh->refId;
   transRefSrvHandle(pConn);
@@ -789,14 +790,14 @@ static void destroyConnRegArg(SSvrConn* conn) {
   }
 }
 static int reallocConnRef(SSvrConn* conn) {
-  transReleaseExHandle(conn->refId);
+  transReleaseExHandle(transGetRefMgt(), conn->refId);
-  transRemoveExHandle(conn->refId);
+  transRemoveExHandle(transGetRefMgt(), conn->refId);
   // avoid app continue to send msg on invalid handle
   SExHandle* exh = taosMemoryMalloc(sizeof(SExHandle));
   exh->handle = conn;
   exh->pThrd = conn->hostThrd;
-  exh->refId = transAddExHandle(exh);
+  exh->refId = transAddExHandle(transGetRefMgt(), exh);
-  transAcquireExHandle(exh->refId);
+  transAcquireExHandle(transGetRefMgt(), exh->refId);
   conn->refId = exh->refId;
 
   return 0;
@@ -808,8 +809,8 @@ static void uvDestroyConn(uv_handle_t* handle) {
   }
   SWorkThrd* thrd = conn->hostThrd;
 
-  transReleaseExHandle(conn->refId);
+  transReleaseExHandle(transGetRefMgt(), conn->refId);
-  transRemoveExHandle(conn->refId);
+  transRemoveExHandle(transGetRefMgt(), conn->refId);
 
   tDebug("%s conn %p destroy", transLabel(thrd->pTransInst), conn);
   transQueueDestroy(&conn->srvMsgs);
@@ -1029,8 +1030,9 @@ void transUnrefSrvHandle(void* handle) {
 }
 
 void transReleaseSrvHandle(void* handle) {
-  SExHandle* exh = handle;
+  SRpcHandleInfo* info = handle;
-  int64_t    refId = exh->refId;
+  SExHandle*      exh = info->handle;
+  int64_t         refId = info->refId;
 
   ASYNC_CHECK_HANDLE(exh, refId);
 
@@ -1045,11 +1047,11 @@ void transReleaseSrvHandle(void* handle) {
 
   tTrace("%s conn %p start to release", transLabel(pThrd->pTransInst), exh->handle);
   transAsyncSend(pThrd->asyncPool, &m->q);
-  transReleaseExHandle(refId);
+  transReleaseExHandle(transGetRefMgt(), refId);
   return;
 _return1:
   tTrace("handle %p failed to send to release handle", exh);
-  transReleaseExHandle(refId);
+  transReleaseExHandle(transGetRefMgt(), refId);
   return;
 _return2:
   tTrace("handle %p failed to send to release handle", exh);
@@ -1074,12 +1076,12 @@ void transSendResponse(const STransMsg* msg) {
   STraceId* trace = (STraceId*)&msg->info.traceId;
   tGTrace("conn %p start to send resp (1/2)", exh->handle);
   transAsyncSend(pThrd->asyncPool, &m->q);
-  transReleaseExHandle(refId);
+  transReleaseExHandle(transGetRefMgt(), refId);
   return;
 _return1:
   tTrace("handle %p failed to send resp", exh);
   rpcFreeCont(msg->pCont);
-  transReleaseExHandle(refId);
+  transReleaseExHandle(transGetRefMgt(), refId);
   return;
 _return2:
   tTrace("handle %p failed to send resp", exh);
@@ -1103,13 +1105,13 @@ void transRegisterMsg(const STransMsg* msg) {
 
   tTrace("%s conn %p start to register brokenlink callback", transLabel(pThrd->pTransInst), exh->handle);
   transAsyncSend(pThrd->asyncPool, &m->q);
-  transReleaseExHandle(refId);
+  transReleaseExHandle(transGetRefMgt(), refId);
   return;
 
 _return1:
   tTrace("handle %p failed to register brokenlink", exh);
   rpcFreeCont(msg->pCont);
-  transReleaseExHandle(refId);
+  transReleaseExHandle(transGetRefMgt(), refId);
   return;
 _return2:
   tTrace("handle %p failed to register brokenlink", exh);

source/libs/transport/test/transUT.cpp

@@ -175,7 +175,7 @@ static void processReleaseHandleCb(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet)
   rpcMsg.code = 0;
   rpcSendResponse(&rpcMsg);
 
-  rpcReleaseHandle(pMsg->info.handle, TAOS_CONN_SERVER);
+  rpcReleaseHandle(&pMsg->info, TAOS_CONN_SERVER);
 }
 static void processRegisterFailure(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet) {
   {

source/libs/wal/src/walWrite.c

@@ -332,21 +332,25 @@ int32_t walWriteWithSyncInfo(SWal *pWal, int64_t index, tmsg_t msgType, SSyncLog
     terrno = TSDB_CODE_WAL_SIZE_LIMIT;
     return -1;
   }
+  taosThreadMutexLock(&pWal->mutex);
 
   if (index == pWal->vers.lastVer + 1) {
     if (taosArrayGetSize(pWal->fileInfoSet) == 0) {
       pWal->vers.firstVer = index;
       if (walRoll(pWal) < 0) {
+        taosThreadMutexUnlock(&pWal->mutex);
         return -1;
       }
     } else {
       int64_t passed = walGetSeq() - pWal->lastRollSeq;
       if (pWal->cfg.rollPeriod != -1 && pWal->cfg.rollPeriod != 0 && passed > pWal->cfg.rollPeriod) {
         if (walRoll(pWal) < 0) {
+          taosThreadMutexUnlock(&pWal->mutex);
           return -1;
         }
       } else if (pWal->cfg.segSize != -1 && pWal->cfg.segSize != 0 && walGetLastFileSize(pWal) > pWal->cfg.segSize) {
         if (walRoll(pWal) < 0) {
+          taosThreadMutexUnlock(&pWal->mutex);
           return -1;
         }
       }
@@ -355,6 +359,7 @@ int32_t walWriteWithSyncInfo(SWal *pWal, int64_t index, tmsg_t msgType, SSyncLog
     // reject skip log or rewrite log
     // must truncate explicitly first
     terrno = TSDB_CODE_WAL_INVALID_VER;
+    taosThreadMutexUnlock(&pWal->mutex);
     return -1;
   }
 
@@ -362,8 +367,6 @@ int32_t walWriteWithSyncInfo(SWal *pWal, int64_t index, tmsg_t msgType, SSyncLog
 
   ASSERT(pWal->writeCur >= 0);
 
-  taosThreadMutexLock(&pWal->mutex);
-
   if (pWal->pWriteIdxTFile == NULL || pWal->pWriteLogTFile == NULL) {
     walSetWrite(pWal);
     taosLSeekFile(pWal->pWriteLogTFile, 0, SEEK_END);

source/util/src/terror.c

@@ -177,6 +177,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_ALREADY_EXIST,      "Dnode already exists"
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_NOT_EXIST,          "Dnode does not exist")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_TOO_MANY_DNODES,          "Too many dnodes")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_NO_ENOUGH_DNODES,         "Out of dnodes")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE,   "No enough memory in dnode")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_CLUSTER_CFG,      "Cluster cfg inconsistent")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_CLUSTER_ID,       "Cluster id not match")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_DNODE_CFG,        "Invalid dnode cfg")

source/util/src/ttimer.c

@@ -18,6 +18,7 @@
 #include "taoserror.h"
 #include "tlog.h"
 #include "tsched.h"
+#include "tdef.h"
 
 #define tmrFatal(...)                                                     \
   {                                                                       \
@@ -110,7 +111,7 @@ typedef struct time_wheel_t {
   tmr_obj_t**     slots;
 } time_wheel_t;
 
-static int32_t tsMaxTmrCtrl = 512;
+static int32_t tsMaxTmrCtrl = TSDB_MAX_VNODES_PER_DB + 100;
 
 static TdThreadOnce  tmrModuleInit = PTHREAD_ONCE_INIT;
 static TdThreadMutex tmrCtrlMutex;

tests/script/api/stopquery.c

@@ -52,6 +52,7 @@ typedef struct {
 typedef struct SSP_CB_PARAM {
   TAOS    *taos;
   bool     fetch;
+  bool     free;
   int32_t *end;
 } SSP_CB_PARAM;
 
@@ -156,7 +157,58 @@ void sqCloseQueryCb(void *param, TAOS_RES *pRes, int code) {
   }
 }
 
-int sqSyncStopQuery(bool fetch) {
+void sqKillFetchCb(void *param, TAOS_RES *pRes, int numOfRows) {
+  SSP_CB_PARAM *qParam = (SSP_CB_PARAM *)param;
+  taos_kill_query(qParam->taos);
+
+  *qParam->end = 1;
+}
+
+void sqKillQueryCb(void *param, TAOS_RES *pRes, int code) {
+  SSP_CB_PARAM *qParam = (SSP_CB_PARAM *)param;
+  if (code == 0 && pRes) {
+    if (qParam->fetch) {
+      taos_fetch_rows_a(pRes, sqKillFetchCb, param);
+    } else {
+      taos_kill_query(qParam->taos);
+      *qParam->end = 1;
+    }
+  } else {
+    sqExit("select", taos_errstr(pRes));
+  }
+}
+
+void sqAsyncFetchCb(void *param, TAOS_RES *pRes, int numOfRows) {
+  SSP_CB_PARAM *qParam = (SSP_CB_PARAM *)param;
+  if (numOfRows > 0) {
+    taos_fetch_rows_a(pRes, sqAsyncFetchCb, param);
+  } else {
+    *qParam->end = 1;
+    if (qParam->free) {
+      taos_free_result(pRes);
+    }
+  }
+}
+
+
+void sqAsyncQueryCb(void *param, TAOS_RES *pRes, int code) {
+  SSP_CB_PARAM *qParam = (SSP_CB_PARAM *)param;
+  if (code == 0 && pRes) {
+    if (qParam->fetch) {
+      taos_fetch_rows_a(pRes, sqAsyncFetchCb, param);
+    } else {
+      if (qParam->free) {
+        taos_free_result(pRes);
+      }
+      *qParam->end = 1;
+    }
+  } else {
+    sqExit("select", taos_errstr(pRes));
+  }
+}
+
+
+int sqStopSyncQuery(bool fetch) {
   CASE_ENTER();  
   for (int32_t i = 0; i < runTimes; ++i) {
     char    sql[1024]  = {0};
@@ -189,7 +241,7 @@ int sqSyncStopQuery(bool fetch) {
   CASE_LEAVE();  
 }
 
-int sqAsyncStopQuery(bool fetch) {
+int sqStopAsyncQuery(bool fetch) {
   CASE_ENTER();  
   for (int32_t i = 0; i < runTimes; ++i) {
     char    sql[1024]  = {0};
@@ -219,7 +271,7 @@ int sqAsyncStopQuery(bool fetch) {
   CASE_LEAVE();  
 }
 
-int sqSyncFreeQuery(bool fetch) {
+int sqFreeSyncQuery(bool fetch) {
   CASE_ENTER();  
   for (int32_t i = 0; i < runTimes; ++i) {
     char    sql[1024]  = {0};
@@ -250,7 +302,7 @@ int sqSyncFreeQuery(bool fetch) {
   CASE_LEAVE();  
 }
 
-int sqAsyncFreeQuery(bool fetch) {
+int sqFreeAsyncQuery(bool fetch) {
   CASE_ENTER();  
   for (int32_t i = 0; i < runTimes; ++i) {
     char    sql[1024]  = {0};
@@ -280,7 +332,7 @@ int sqAsyncFreeQuery(bool fetch) {
   CASE_LEAVE();  
 }
 
-int sqSyncCloseQuery(bool fetch) {
+int sqCloseSyncQuery(bool fetch) {
   CASE_ENTER();  
   for (int32_t i = 0; i < runTimes; ++i) {
     char    sql[1024]  = {0};
@@ -310,7 +362,7 @@ int sqSyncCloseQuery(bool fetch) {
   CASE_LEAVE();  
 }
 
-int sqAsyncCloseQuery(bool fetch) {
+int sqCloseAsyncQuery(bool fetch) {
   CASE_ENTER();  
   for (int32_t i = 0; i < runTimes; ++i) {
     char    sql[1024]  = {0};
@@ -360,9 +412,39 @@ void *syncQueryThreadFp(void *arg) {
     taos_fetch_row(pRes);
   }
 
-  taos_free_result(pRes);
+  if (qParam->free) {
+    taos_free_result(pRes);
+  }
 }
 
+void *asyncQueryThreadFp(void *arg) {
+  SSP_CB_PARAM* qParam = (SSP_CB_PARAM*)arg;
+  char    sql[1024]  = {0};
+  int32_t code = 0;
+  TAOS   *taos = taos_connect(hostName, "root", "taosdata", NULL, 0);
+  if (taos == NULL) sqExit("taos_connect", taos_errstr(NULL));
+
+  qParam->taos = taos;
+
+  sprintf(sql, "reset query cache");
+  sqExecSQLE(taos, sql);
+  
+  sprintf(sql, "use %s", dbName);
+  sqExecSQLE(taos, sql);
+  
+  sprintf(sql, "select * from %s", tbName);
+  
+  int32_t qEnd = 0;
+  SSP_CB_PARAM param = {0};
+  param.fetch = qParam->fetch;
+  param.end = &qEnd;
+  taos_query_a(taos, sql, sqAsyncQueryCb, &param);
+  while (0 == qEnd) {
+    usleep(5000);
+  }
+}
+
+
 void *closeThreadFp(void *arg) {
   SSP_CB_PARAM* qParam = (SSP_CB_PARAM*)arg;
   while (true) {
@@ -376,7 +458,22 @@ void *closeThreadFp(void *arg) {
 }
 
 
-int sqConSyncCloseQuery(bool fetch) {
+
+void *killThreadFp(void *arg) {
+  SSP_CB_PARAM* qParam = (SSP_CB_PARAM*)arg;
+  while (true) {
+    if (qParam->taos) {
+      usleep(rand() % 10000);
+      taos_kill_query(qParam->taos);
+      break;
+    }
+    usleep(1);
+  }
+}
+
+
+
+int sqConCloseSyncQuery(bool fetch) {
   CASE_ENTER();  
   pthread_t qid, cid;
   for (int32_t i = 0; i < runTimes; ++i) {
@@ -391,26 +488,160 @@ int sqConSyncCloseQuery(bool fetch) {
   CASE_LEAVE();  
 }
 
+int sqConCloseAsyncQuery(bool fetch) {
+  CASE_ENTER();  
+  pthread_t qid, cid;
+  for (int32_t i = 0; i < runTimes; ++i) {
+    SSP_CB_PARAM param = {0};
+    param.fetch = fetch;
+    pthread_create(&qid, NULL, asyncQueryThreadFp, (void*)&param);
+    pthread_create(&cid, NULL, closeThreadFp, (void*)&param);
+    
+    pthread_join(qid, NULL);
+    pthread_join(cid, NULL);
+  }
+  CASE_LEAVE();  
+}
+
+
+int sqKillSyncQuery(bool fetch) {
+  CASE_ENTER();  
+  for (int32_t i = 0; i < runTimes; ++i) {
+    char    sql[1024]  = {0};
+    int32_t code = 0;
+    TAOS   *taos = taos_connect(hostName, "root", "taosdata", NULL, 0);
+    if (taos == NULL) sqExit("taos_connect", taos_errstr(NULL));
+
+    sprintf(sql, "reset query cache");
+    sqExecSQL(taos, sql);
+
+    sprintf(sql, "use %s", dbName);
+    sqExecSQL(taos, sql);
+
+    sprintf(sql, "select * from %s", tbName);
+    TAOS_RES* pRes = taos_query(taos, sql);
+    code = taos_errno(pRes);
+    if (code) {
+      sqExit("taos_query", taos_errstr(pRes));
+    }
+
+    if (fetch) {
+      taos_fetch_row(pRes);
+    }
+
+    taos_kill_query(taos);
+    
+    taos_close(taos);
+  }
+  CASE_LEAVE();  
+}
+
+int sqKillAsyncQuery(bool fetch) {
+  CASE_ENTER();  
+  for (int32_t i = 0; i < runTimes; ++i) {
+    char    sql[1024]  = {0};
+    int32_t code = 0;
+    TAOS   *taos = taos_connect(hostName, "root", "taosdata", NULL, 0);
+    if (taos == NULL) sqExit("taos_connect", taos_errstr(NULL));
+
+    sprintf(sql, "reset query cache");
+    sqExecSQL(taos, sql);
+
+    sprintf(sql, "use %s", dbName);
+    sqExecSQL(taos, sql);
+
+    sprintf(sql, "select * from %s", tbName);
+
+    int32_t qEnd = 0;
+    SSP_CB_PARAM param = {0};
+    param.fetch = fetch;
+    param.end = &qEnd;
+    param.taos = taos;
+    taos_query_a(taos, sql, sqKillQueryCb, &param);
+    while (0 == qEnd) {
+      usleep(5000);
+    }
+    
+    taos_close(taos);
+  }
+  CASE_LEAVE();  
+}
+
+int sqConKillSyncQuery(bool fetch) {
+  CASE_ENTER();  
+  pthread_t qid, cid;
+  for (int32_t i = 0; i < runTimes; ++i) {
+    SSP_CB_PARAM param = {0};
+    param.fetch = fetch;
+    pthread_create(&qid, NULL, syncQueryThreadFp, (void*)&param);
+    pthread_create(&cid, NULL, killThreadFp, (void*)&param);
+    
+    pthread_join(qid, NULL);
+    pthread_join(cid, NULL);
+  }
+  CASE_LEAVE();  
+}
+
+int sqConKillAsyncQuery(bool fetch) {
+  CASE_ENTER();  
+  pthread_t qid, cid;
+  for (int32_t i = 0; i < runTimes; ++i) {
+    SSP_CB_PARAM param = {0};
+    param.fetch = fetch;
+    pthread_create(&qid, NULL, asyncQueryThreadFp, (void*)&param);
+    pthread_create(&cid, NULL, killThreadFp, (void*)&param);
+    
+    pthread_join(qid, NULL);
+    pthread_join(cid, NULL);
+  }
+  CASE_LEAVE();  
+}
+
+
+
 void sqRunAllCase(void) {
 /*
-  sqSyncStopQuery(false);
+  sqStopSyncQuery(false);
-  sqSyncStopQuery(true);
+  sqStopSyncQuery(true);
-  sqAsyncStopQuery(false);
+  sqStopAsyncQuery(false);
-  sqAsyncStopQuery(true);
+  sqStopAsyncQuery(true);
 
-  sqSyncFreeQuery(false);
+  sqFreeSyncQuery(false);
-  sqSyncFreeQuery(true);
+  sqFreeSyncQuery(true);
-  sqAsyncFreeQuery(false);
+  sqFreeAsyncQuery(false);
-  sqAsyncFreeQuery(true);
+  sqFreeAsyncQuery(true);
 
-  sqSyncCloseQuery(false);
+  sqCloseSyncQuery(false);
-  sqSyncCloseQuery(true);
+  sqCloseSyncQuery(true);
-  sqAsyncCloseQuery(false);
+  sqCloseAsyncQuery(false);
-  sqAsyncCloseQuery(true);
+  sqCloseAsyncQuery(true);
-*/  
+  
-  sqConSyncCloseQuery(false);
+  sqConCloseSyncQuery(false);
-  sqConSyncCloseQuery(true);
+  sqConCloseSyncQuery(true);
+  sqConCloseAsyncQuery(false);
+  sqConCloseAsyncQuery(true);
+*/
 
+#if 0  
+
+  sqKillSyncQuery(false);
+  sqKillSyncQuery(true);
+  sqKillAsyncQuery(false);
+  sqKillAsyncQuery(true);
+#endif  
+
+  //sqConKillSyncQuery(false);
+  sqConKillSyncQuery(true);
+#if 0  
+  sqConKillAsyncQuery(false);
+  sqConKillAsyncQuery(true);
+#endif
+
+  int32_t l = 5;
+  while (l) {
+    printf("%d\n", l--);
+    sleep(1);
+  }
 }
 
 

tests/script/jenkins/basic.txt

@@ -164,6 +164,7 @@
 ./test.sh -f tsim/sma/drop_sma.sim
 ./test.sh -f tsim/sma/tsmaCreateInsertQuery.sim
 ./test.sh -f tsim/sma/rsmaCreateInsertQuery.sim
+./test.sh -f tsim/sma/rsmaPersistenceRecovery.sim
 
 # --- valgrind
 ./test.sh -f tsim/valgrind/checkError.sim -v

tests/script/sh/deploy.sh

@@ -121,7 +121,7 @@ echo "firstEp                ${HOSTNAME}:7100"   >> $TAOS_CFG
 echo "secondEp               ${HOSTNAME}:7200"   >> $TAOS_CFG
 echo "fqdn                   ${HOSTNAME}"        >> $TAOS_CFG
 echo "serverPort             ${NODE}"            >> $TAOS_CFG
-echo "supportVnodes          128"                >> $TAOS_CFG
+echo "supportVnodes          1024"                >> $TAOS_CFG
 echo "dataDir                $DATA_DIR"          >> $TAOS_CFG
 echo "logDir                 $LOG_DIR"           >> $TAOS_CFG
 echo "debugFlag              0"                  >> $TAOS_CFG
@@ -135,7 +135,7 @@ echo "jniDebugFlag           143"                >> $TAOS_CFG
 echo "qDebugFlag             143"                >> $TAOS_CFG
 echo "rpcDebugFlag           143"                >> $TAOS_CFG
 echo "tmrDebugFlag           131"                >> $TAOS_CFG
-echo "uDebugFlag             131"                >> $TAOS_CFG
+echo "uDebugFlag             143"                >> $TAOS_CFG
 echo "sDebugFlag             143"                >> $TAOS_CFG
 echo "wDebugFlag             143"                >> $TAOS_CFG
 echo "numOfLogLines          20000000"           >> $TAOS_CFG

tests/script/sh/stop_dnodes.sh

@@ -14,7 +14,7 @@ while [ -n "$PID" ]; do
   echo kill -9 $PID
   #pkill -9 taosd
   kill -9 $PID
-  echo "Killing processes locking on port 6030"
+  echo "Killing taosd processes"
   if [ "$OS_TYPE" != "Darwin" ]; then
     fuser -k -n tcp 6030
   else

tests/script/tmp/prepare.sim

@@ -7,39 +7,33 @@ system sh/deploy.sh -n dnode4 -i 4
 
 return
 
-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/deploy.sh -n dnode5 -i 5
-system sh/deploy.sh -n dnode6 -i 6
-system sh/deploy.sh -n dnode7 -i 7
-system sh/deploy.sh -n dnode8 -i 8
-system sh/deploy.sh -n dnode9 -i 9
 system sh/exec.sh -n dnode1 -s start
-system sh/exec.sh -n dnode2 -s start
+sql connect
-system sh/exec.sh -n dnode3 -s start
-system sh/exec.sh -n dnode4 -s start
-system sh/exec.sh -n dnode5 -s start
-system sh/exec.sh -n dnode6 -s start
-system sh/exec.sh -n dnode7 -s start
-system sh/exec.sh -n dnode8 -s start
-system sh/exec.sh -n dnode9 -s start
 
-sleep 2000
+$num = 200
+$i = 0
+while $i < $num 
+  $port = $i + 8000
+  $i = $i + 1
+  sql create dnode $hostname port $port
+endw 
 
-sql create dnode $hostname port 7200
+$i = 0
-sql create dnode $hostname port 7300
+while $i < $num 
-sql create dnode $hostname port 7400
+  $port = $i + 8000
-sql create dnode $hostname port 7500
+  $i = $i + 1
-sql create dnode $hostname port 7600
+  $name = dnode . $port
-sql create dnode $hostname port 7700
+  system sh/deploy.sh -n $name -i 1
-sql create dnode $hostname port 7800
+  system sh/cfg.sh -n $name -c serverPort -v $port
-sql create dnode $hostname port 7900
+  system sh/exec.sh -n $name -s start
+endw 
 
-sql show dnodes;
+return 
-print $data00 $data01
+
-print $data10 $data11
+sql create database db vgroups 1024 buffer 3;
-print $data20 $data21
+sql use db;
-print $data30 $data31
+sql create table if not exists stb (ts timestamp, c1 int, c2 float, c3 double) tags (t1 int unsigned);
-print $data40 $data41
+sql create table ct1 using stb tags(1000);
+sql create table ct2 using stb tags(1000) ;
+sql show db.tables;
+sql insert into ct1 values(now+0s, 10, 2.0, 3.0);

tests/script/tsim/sma/rsmaPersistenceRecovery.sim

@@ -0,0 +1,237 @@
+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+system sh/exec.sh -n dnode1 -s start
+sleep 50
+sql connect
+
+print =============== create database with retentions
+sql create database d0 retentions 5s:7d,5m:21d,15m:365d;
+sql use d0
+
+print =============== create super table and register rsma
+sql create table if not exists stb (ts timestamp, c1 int, c2 float) tags (city binary(20),district binary(20)) rollup(max) max_delay 5s,5s watermark 2s,3s;
+
+sql show stables
+if $rows != 1 then 
+  return -1
+endi
+
+print =============== create child table
+sql create table ct1 using stb tags("BeiJing", "ChaoYang");
+
+sql show tables
+if $rows != 1 then 
+  return -1
+endi
+
+print =============== insert data and trigger rollup
+sql insert into ct1 values(now, 10, 10.0);
+sql insert into ct1 values(now+1s, 1, 1.0);
+sql insert into ct1 values(now+2s, 100, 100.0);
+
+print =============== wait maxdelay 5+1 seconds for results
+sleep 6000
+
+print =============== select * from retention level 2 from memory
+sql select * from ct1;
+print $data00 $data01 $data02
+if $rows > 2 then
+  print retention level 2 file rows $rows > 2
+  return -1
+endi
+
+
+if $data01 != 100 then
+  if $data01 != 10 then
+    print retention level 2 file result $data01 != 100 or 10
+  return -1
+  endi
+endi
+
+print =============== select * from retention level 1 from memory
+sql select * from ct1 where ts > now-8d;
+print $data00 $data01 $data02
+if $rows > 2 then
+  print retention level 1 file rows $rows > 2
+  return -1
+endi
+
+if $data01 != 100 then
+  if $data01 != 10 then
+    print retention level 1 file result $data01 != 100 or 10
+  return -1
+  endi
+endi
+
+print =============== select * from retention level 0 from memory
+sql select * from ct1 where ts > now-3d;
+print $data00 $data01 $data02
+print $data10 $data11 $data12
+print $data20 $data21 $data22
+
+if $rows < 1 then
+  print retention level 0 file rows $rows < 1
+  return -1
+endi
+
+if $data01 != 10 then
+  print retention level 0 file result $data01 != 10
+  return -1
+endi
+
+#===================================================================
+
+
+#==================== reboot to trigger commit data to file
+system sh/exec.sh -n dnode1 -s stop -x SIGINT
+system sh/exec.sh -n dnode1 -s start
+
+print =============== select * from retention level 2 from file
+sql select * from ct1;
+print $data00 $data01 $data02
+if $rows > 2 then
+  print retention level 2 file rows $rows > 2
+  return -1
+endi
+
+if $data01 != 100 then
+  if $data01 != 10 then
+    print retention level 2 file result $data01 != 100 or 10
+  return -1
+  endi
+endi
+
+print =============== select * from retention level 1 from file
+sql select * from ct1 where ts > now-8d;
+print $data00 $data01 $data02
+if $rows > 2 then
+  print retention level 1 file rows $rows > 2
+  return -1
+endi
+
+if $data01 != 100 then
+  if $data01 != 10 then
+    print retention level 1 file result $data01 != 100 or 10
+  return -1
+  endi
+endi
+
+print =============== select * from retention level 0 from file
+sql select * from ct1 where ts > now-3d;
+print $data00 $data01 $data02
+print $data10 $data11 $data12
+print $data20 $data21 $data22
+if $rows < 1 then
+  print retention level 0 file rows $rows < 1
+  return -1
+endi
+
+if $data01 != 10 then
+  print retention level 0 file result $data01 != 10
+  return -1
+endi
+
+print =============== insert after rsma qtaskinfo recovery
+sql insert into ct1 values(now, 50, 500.0);
+sql insert into ct1 values(now+1s, 40, 40.0);
+
+print =============== wait maxdelay 5+1 seconds for results
+sleep 6000
+
+print =============== select * from retention level 2 from file and memory after rsma qtaskinfo recovery
+sql select * from ct1;
+print $data00 $data01 $data02
+if $rows > 2 then
+  print retention level 2 file/mem rows $rows > 2
+  return -1
+endi
+
+if $data01 != 100 then
+  if $data01 != 10 then
+    print retention level 2 file/mem result $data01 != 100 or 10
+  return -1
+  endi
+endi
+
+if $data02 != 500.00000 then
+  if $data02 != 100.00000 then
+    print retention level 1 file/mem result $data02 != 500.00000 or 100.00000
+  return -1
+  endi
+endi
+
+print =============== select * from retention level 1 from file and memory after rsma qtaskinfo recovery
+sql select * from ct1 where ts > now-8d;
+print $data00 $data01 $data02
+if $rows > 2 then
+  print retention level 1 file/mem rows $rows > 2
+  return -1
+endi
+
+if $data01 != 100 then
+  if $data01 != 10 then
+    print retention level 1 file/mem result $data01 != 100 or 10
+  return -1
+  endi
+endi
+
+if $data02 != 500.00000 then
+  if $data02 != 100.00000 then
+    print retention level 1 file/mem result $data02 != 500.00000 or 100.00000
+  return -1
+  endi
+endi
+
+
+print =============== select * from retention level 0 from file and memory after rsma qtaskinfo recovery
+sql select * from ct1 where ts > now-3d;
+print $data00 $data01 $data02
+print $data10 $data11 $data12
+print $data20 $data21 $data22
+print $data30 $data31 $data32
+print $data40 $data41 $data42
+
+if $rows < 1 then
+  print retention level 0 file/mem rows $rows < 1
+  return -1
+endi
+
+if $data01 != 10 then
+  print retention level 0 file/mem result $data01 != 10
+  return -1
+endi
+
+if $data11 != 1 then
+  print retention level 0 file/mem result $data11 != 1
+  return -1
+endi
+
+if $data21 != 100 then
+  print retention level 0 file/mem result $data21 != 100
+  return -1
+endi
+
+if $data31 != 50 then
+  print retention level 0 file/mem result $data31 != 50
+  return -1
+endi
+
+if $data32 != 500.00000 then
+  print retention level 0 file/mem result $data32 != 500.00000
+  return -1
+endi
+
+
+if $data41 != 40 then
+  print retention level 0 file/mem result $data41 != 40
+  return -1
+endi
+
+if $data42 != 40.00000 then
+  print retention level 0 file/mem result $data42 != 40.00000
+  return -1
+endi
+
+
+
+system sh/exec.sh -n dnode1 -s stop -x SIGINT

tests/script/tsim/valgrind/basic.sim

@@ -1,8 +1,33 @@
 system sh/stop_dnodes.sh
 system sh/deploy.sh -n dnode1 -i 1
-system sh/exec.sh -n dnode1 -s start
+system sh/exec.sh -n dnode1 -s start -v
 sql connect
 
-sql create database d0 vgroups 1;
+print =============== step1: create drop show dnodes
+$x = 0
+step1:
+	$x = $x + 1
+	sleep 1000
+	if $x == 10 then
+	  print ====> dnode not ready!
+		return -1
+	endi
+sql show dnodes
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+if $rows != 1 then
+  return -1
+endi
 
+goto _OVER
+
+print =============== step2: create alter drop show user
+sql create user u1 pass 'taosdata'
+sql show users
+sql alter user u1 sysinfo 1
+sql alter user u1 enable 1
+sql alter user u1 pass 'taosdata'
+sql drop user u1
+sql_error alter user u2 sysinfo 0
+
+_OVER:
 system sh/exec.sh -n dnode1 -s stop -x SIGINT

tests/script/tsim/valgrind/basic2.sim

@@ -0,0 +1,25 @@
+system sh/stop_dnodes.sh
+system sh/deploy.sh -n dnode1 -i 1
+system sh/exec.sh -n dnode1 -s start -v
+sql connect
+
+print =============== step1: create drop show dnodes
+$x = 0
+step1:
+	$x = $x + 1
+	sleep 1000
+	if $x == 10 then
+	  print ====> dnode not ready!
+		return -1
+	endi
+sql show dnodes
+print ===> $data00 $data01 $data02 $data03 $data04 $data05
+if $rows != 1 then
+  return -1
+endi
+
+print =============== step2: create db
+sql create database db vgroups 1
+
+_OVER:
+system sh/exec.sh -n dnode1 -s stop -x SIGINT

tests/system-test/1-insert/block_wise.py

@@ -0,0 +1,442 @@
+import datetime
+import re
+
+from dataclasses import dataclass, field
+from typing import List, Any, Tuple
+from util.log import *
+from util.sql import *
+from util.cases import *
+from util.dnodes import *
+from util.constant import *
+
+PRIMARY_COL = "ts"
+
+INT_COL = "c_int"
+BINT_COL = "c_bint"
+SINT_COL = "c_sint"
+TINT_COL = "c_tint"
+FLOAT_COL = "c_float"
+DOUBLE_COL = "c_double"
+BOOL_COL = "c_bool"
+TINT_UN_COL = "c_utint"
+SINT_UN_COL = "c_usint"
+BINT_UN_COL = "c_ubint"
+INT_UN_COL = "c_uint"
+BINARY_COL = "c_binary"
+NCHAR_COL = "c_nchar"
+TS_COL = "c_ts"
+
+NUM_COL = [INT_COL, BINT_COL, SINT_COL, TINT_COL, FLOAT_COL, DOUBLE_COL, ]
+CHAR_COL = [BINARY_COL, NCHAR_COL, ]
+BOOLEAN_COL = [BOOL_COL, ]
+TS_TYPE_COL = [TS_COL, ]
+
+INT_TAG = "t_int"
+
+ALL_COL = [PRIMARY_COL, INT_COL, BINT_COL, SINT_COL, TINT_COL, FLOAT_COL, DOUBLE_COL, BINARY_COL, NCHAR_COL, BOOL_COL, TS_COL]
+TAG_COL = [INT_TAG]
+
+# insert data args：
+TIME_STEP = 10000
+NOW = int(datetime.datetime.timestamp(datetime.datetime.now()) * 1000)
+
+# init db/table
+DBNAME  = "db"
+STBNAME = "stb1"
+CTBNAME = "ct1"
+NTBNAME = "nt1"
+
+
+@dataclass
+class DataSet:
+    ts_data     : List[int]     = field(default_factory=list)
+    int_data    : List[int]     = field(default_factory=list)
+    bint_data   : List[int]     = field(default_factory=list)
+    sint_data   : List[int]     = field(default_factory=list)
+    tint_data   : List[int]     = field(default_factory=list)
+    int_un_data : List[int]     = field(default_factory=list)
+    bint_un_data: List[int]     = field(default_factory=list)
+    sint_un_data: List[int]     = field(default_factory=list)
+    tint_un_data: List[int]     = field(default_factory=list)
+    float_data  : List[float]   = field(default_factory=list)
+    double_data : List[float]   = field(default_factory=list)
+    bool_data   : List[int]     = field(default_factory=list)
+    binary_data : List[str]     = field(default_factory=list)
+    nchar_data  : List[str]     = field(default_factory=list)
+
+
+@dataclass
+class BSMAschema:
+    creation            : str           = "CREATE"
+    tb_type             : str           = "stable"
+    tbname              : str           = STBNAME
+    cols                : Tuple[str]    = None
+    tags                : Tuple[str]    = None
+    sma_flag            : str           = "SMA"
+    sma_cols            : Tuple[str]    = None
+    create_tabel_sql    : str           = None
+    other               : Any           = None
+
+    drop                : str           = "DROP"
+    drop_flag           : str           = "INDEX"
+    querySmaOptimize    : int           = 1
+    show                : str           = "SHOW"
+    show_msg            : str           = "INDEXES"
+    show_oper           : str           = "FROM"
+    dbname              : str           = None
+    rollup_db           : bool          = False
+
+    def __post_init__(self):
+        if isinstance(self.other, dict):
+            for k,v in self.other.items():
+
+                if k.lower() == "tbname" and isinstance(v, str) and not self.tbname:
+                    self.tbname = v
+                    del self.other[k]
+
+                if k.lower() == "cols" and (isinstance(v, tuple) or isinstance(v, list)) and not self.cols:
+                    self.cols = v
+                    del self.other[k]
+
+                if k.lower() == "tags" and (isinstance(v, tuple) or isinstance(v, list)) and not self.tags:
+                    self.tags = v
+                    del self.other[k]
+
+                if k.lower() == "sma_flag" and isinstance(v, str) and not self.sma_flag:
+                    self.sma_flag = v
+                    del self.other[k]
+
+                if k.lower() == "sma_cols" and (isinstance(v, tuple) or isinstance(v, list)) and not self.sma_cols:
+                    self.sma_cols = v
+                    del self.other[k]
+
+                if k.lower() == "create_tabel_sql" and isinstance(v, str) and not self.create_tabel_sql:
+                    self.create_tabel_sql = v
+                    del self.other[k]
+
+                # bSma show and drop operator is  not completed
+                if k.lower() == "drop_flag" and isinstance(v, str) and not self.drop_flag:
+                    self.drop_flag = v
+                    del self.other[k]
+
+                if k.lower() == "show_msg" and isinstance(v, str) and not self.show_msg:
+                    self.show_msg = v
+                    del self.other[k]
+
+                if k.lower() == "dbname" and isinstance(v, str) and not self.dbname:
+                    self.dbname = v
+                    del self.other[k]
+
+                if k.lower() == "show_oper" and isinstance(v, str) and not self.show_oper:
+                    self.show_oper = v
+                    del self.other[k]
+
+                if k.lower() == "rollup_db" and isinstance(v, bool) and not self.rollup_db:
+                    self.rollup_db = v
+                    del self.other[k]
+
+
+
+# from ...pytest.util.sql import *
+# from ...pytest.util.constant import *
+
+class TDTestCase:
+
+    def init(self, conn, logSql):
+        tdLog.debug(f"start to excute {__file__}")
+        tdSql.init(conn.cursor(), False)
+        self.precision = "ms"
+        self.sma_count = 0
+        self.sma_created_index = []
+
+    def __create_sma_index(self, sma:BSMAschema):
+        if sma.create_tabel_sql:
+            sql = sma.create_tabel_sql
+        else:
+            sql = f"{sma.creation} {sma.tb_type} {sma.tbname} ({', '.join(sma.cols)}) "
+
+            if sma.tb_type == "stable" or (sma.tb_type=="table" and sma.tags):
+                sql = f"{sma.creation} {sma.tb_type} {sma.tbname} ({', '.join(sma.cols)}) tags ({', '.join(sma.tags)}) "
+
+
+        if sma.sma_flag:
+            sql += sma.sma_flag
+        if sma.sma_cols:
+            sql += f"({', '.join(sma.sma_cols)})"
+
+        if isinstance(sma.other, dict):
+            for k,v in sma.other.items():
+                if isinstance(v,tuple) or isinstance(v, list):
+                    sql += f" {k} ({' '.join(v)})"
+                else:
+                    sql += f" {k} {v}"
+        if isinstance(sma.other, tuple) or isinstance(sma.other, list):
+            sql += " ".join(sma.other)
+        if isinstance(sma.other, int) or isinstance(sma.other, float) or isinstance(sma.other, str):
+            sql += f" {sma.other}"
+
+        return sql
+
+    def __get_bsma_table_col_tag_str(self, sql:str):
+        p = re.compile(r"[(](.*)[)]", re.S)
+
+        if "tags" in (col_str := sql):
+            col_str = re.findall(p, sql.split("tags")[0])[0].split(",")
+            if (tag_str := re.findall(p, sql.split("tags")[1])[0].split(",") ):
+                col_str.extend(tag_str)
+
+        return col_str
+
+    def __get_bsma_col_tag_names(self, col_tags:list):
+        return [ col_tag.strip().split(" ")[0] for col_tag in col_tags ]
+
+    @property
+    def __get_db_tbname(self):
+        tb_list = []
+        tdSql.query("show tables")
+        for row in tdSql.queryResult:
+            tb_list.append(row[0])
+        tdSql.query("show tables")
+        for row in tdSql.queryResult:
+            tb_list.append(row[0])
+
+        return tb_list
+
+    def __bsma_create_check(self, sma:BSMAschema):
+        if not sma.creation:
+            return False
+        if not sma.create_tabel_sql and (not sma.tbname or not sma.tb_type or not sma.cols):
+            return False
+        if not sma.create_tabel_sql and (sma.tb_type == "stable" and not sma.tags):
+            return  False
+        if not sma.sma_flag or not isinstance(sma.sma_flag, str) or sma.sma_flag.upper() != "SMA":
+            return False
+        if sma.tbname in self.__get_db_tbname:
+            return False
+
+        if sma.create_tabel_sql:
+            col_tag_list = self.__get_bsma_col_tag_names(self.__get_bsma_table_col_tag_str(sma.create_tabel_sql))
+        else:
+            col_str = list(sma.cols)
+            if sma.tags:
+                col_str.extend(list(sma.tags))
+            col_tag_list = self.__get_bsma_col_tag_names(col_str)
+        if not sma.sma_cols:
+            return False
+        for col in sma.sma_cols:
+            if col not in col_tag_list:
+                return False
+
+        return True
+
+    def bsma_create_check(self, sma:BSMAschema):
+        if self.__bsma_create_check(sma):
+            tdSql.query(self.__create_sma_index(sma))
+            tdLog.info(f"current sql: {self.__create_sma_index(sma)}")
+
+        else:
+            tdSql.error(self.__create_sma_index(sma))
+
+
+    def __sma_drop_check(self, sma:BSMAschema):
+        pass
+
+    def sma_drop_check(self, sma:BSMAschema):
+        pass
+
+    def __show_sma_index(self, sma:BSMAschema):
+        pass
+
+    def __sma_show_check(self, sma:BSMAschema):
+        pass
+
+    def sma_show_check(self, sma:BSMAschema):
+        pass
+
+    @property
+    def __create_sma_sql(self):
+        err_sqls = []
+        cur_sqls = []
+        # err_set
+        ### case 1: required fields check
+        err_sqls.append( BSMAschema(creation="", tbname="stb2", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(f"{INT_TAG} int",), sma_cols=(PRIMARY_COL, INT_COL ) ) )
+        err_sqls.append( BSMAschema(tbname="", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(f"{INT_TAG} int",), sma_cols=(PRIMARY_COL, INT_COL ) ) )
+        err_sqls.append( BSMAschema(tbname="stb2", cols=(), tags=(f"{INT_TAG} int",), sma_cols=(PRIMARY_COL, INT_COL ) ) )
+        err_sqls.append( BSMAschema(tbname="stb2", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(), sma_cols=(PRIMARY_COL, INT_COL ) ) )
+        err_sqls.append( BSMAschema(tbname="stb2", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(f"{INT_TAG} int",), sma_flag="", sma_cols=(PRIMARY_COL, INT_COL ) ) )
+        err_sqls.append( BSMAschema(tbname="stb2", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(f"{INT_TAG} int",), sma_cols=() ) )
+        ### case 2:
+        err_sqls.append( BSMAschema(tbname="stb2", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(f"{INT_TAG} int",), sma_cols=({BINT_COL}) ) )
+
+        # current_set
+        cur_sqls.append( BSMAschema(tbname="stb2", cols=(f"{PRIMARY_COL} timestamp", f"{INT_COL} int"), tags=(f"{INT_TAG} int",), sma_cols=(PRIMARY_COL, INT_COL ) ) )
+
+        return err_sqls, cur_sqls
+
+    def test_create_sma(self):
+        err_sqls , cur_sqls = self.__create_sma_sql
+        for err_sql in err_sqls:
+            self.bsma_create_check(err_sql)
+        for cur_sql in cur_sqls:
+            self.bsma_create_check(cur_sql)
+
+    @property
+    def __drop_sma_sql(self):
+        err_sqls = []
+        cur_sqls = []
+        # err_set
+        ## case 1: required fields check
+        return err_sqls, cur_sqls
+
+    def test_drop_sma(self):
+        err_sqls , cur_sqls = self.__drop_sma_sql
+        for err_sql in err_sqls:
+            self.sma_drop_check(err_sql)
+        for cur_sql in cur_sqls:
+            self.sma_drop_check(cur_sql)
+
+    def all_test(self):
+        self.test_create_sma()
+
+    def __create_tb(self):
+        tdLog.printNoPrefix("==========step: create table")
+        create_stb_sql = f'''create table {STBNAME}(
+                ts timestamp, {INT_COL} int, {BINT_COL} bigint, {SINT_COL} smallint, {TINT_COL} tinyint,
+                {FLOAT_COL} float, {DOUBLE_COL} double, {BOOL_COL} bool,
+                {BINARY_COL} binary(16), {NCHAR_COL} nchar(32), {TS_COL} timestamp,
+                {TINT_UN_COL} tinyint unsigned, {SINT_UN_COL} smallint unsigned,
+                {INT_UN_COL} int unsigned, {BINT_UN_COL} bigint unsigned
+            ) tags ({INT_TAG} int)
+            '''
+        create_ntb_sql = f'''create table {NTBNAME}(
+                ts timestamp, {INT_COL} int, {BINT_COL} bigint, {SINT_COL} smallint, {TINT_COL} tinyint,
+                {FLOAT_COL} float, {DOUBLE_COL} double, {BOOL_COL} bool,
+                {BINARY_COL} binary(16), {NCHAR_COL} nchar(32), {TS_COL} timestamp,
+                {TINT_UN_COL} tinyint unsigned, {SINT_UN_COL} smallint unsigned,
+                {INT_UN_COL} int unsigned, {BINT_UN_COL} bigint unsigned
+            )
+            '''
+        tdSql.execute(create_stb_sql)
+        tdSql.execute(create_ntb_sql)
+
+        for i in range(4):
+            tdSql.execute(f'create table ct{i+1} using stb1 tags ( {i+1} )')
+
+    def __data_set(self, rows):
+        data_set = DataSet()
+
+        for i in range(rows):
+            data_set.ts_data.append(NOW + 1 * (rows - i))
+            data_set.int_data.append(rows - i)
+            data_set.bint_data.append(11111 * (rows - i))
+            data_set.sint_data.append(111 * (rows - i) % 32767)
+            data_set.tint_data.append(11 * (rows - i) % 127)
+            data_set.int_un_data.append(rows - i)
+            data_set.bint_un_data.append(11111 * (rows - i))
+            data_set.sint_un_data.append(111 * (rows - i) % 32767)
+            data_set.tint_un_data.append(11 * (rows - i) % 127)
+            data_set.float_data.append(1.11 * (rows - i))
+            data_set.double_data.append(1100.0011 * (rows - i))
+            data_set.bool_data.append((rows - i) % 2)
+            data_set.binary_data.append(f'binary{(rows - i)}')
+            data_set.nchar_data.append(f'nchar_测试_{(rows - i)}')
+
+        return data_set
+
+    def __insert_data(self):
+        tdLog.printNoPrefix("==========step: start inser data into tables now.....")
+        data = self.__data_set(rows=self.rows)
+
+        # now_time = int(datetime.datetime.timestamp(datetime.datetime.now()) * 1000)
+        null_data = '''null, null, null, null, null, null, null, null, null, null, null, null, null, null'''
+        zero_data = "0, 0, 0, 0, 0, 0, 0, 'binary_0', 'nchar_0', 0, 0, 0, 0, 0"
+
+        for i in range(self.rows):
+            row_data = f'''
+                {data.int_data[i]}, {data.bint_data[i]}, {data.sint_data[i]}, {data.tint_data[i]}, {data.float_data[i]}, {data.double_data[i]},
+                {data.bool_data[i]}, '{data.binary_data[i]}', '{data.nchar_data[i]}', {data.ts_data[i]}, {data.tint_un_data[i]},
+                {data.sint_un_data[i]}, {data.int_un_data[i]}, {data.bint_un_data[i]}
+            '''
+            neg_row_data = f'''
+                {-1 * data.int_data[i]}, {-1 * data.bint_data[i]}, {-1 * data.sint_data[i]}, {-1 * data.tint_data[i]}, {-1 * data.float_data[i]}, {-1 * data.double_data[i]},
+                {data.bool_data[i]}, '{data.binary_data[i]}', '{data.nchar_data[i]}', {data.ts_data[i]}, {1 * data.tint_un_data[i]},
+                {1 * data.sint_un_data[i]}, {1 * data.int_un_data[i]}, {1 * data.bint_un_data[i]}
+            '''
+
+            tdSql.execute(
+                f"insert into ct1 values ( {NOW - i * TIME_STEP}, {row_data} )")
+            tdSql.execute(
+                f"insert into ct2 values ( {NOW - i * int(TIME_STEP * 0.6)}, {neg_row_data} )")
+            tdSql.execute(
+                f"insert into ct4 values ( {NOW - i * int(TIME_STEP * 0.8) }, {row_data} )")
+            tdSql.execute(
+                f"insert into {NTBNAME} values ( {NOW - i * int(TIME_STEP * 1.2)}, {row_data} )")
+
+        tdSql.execute(
+            f"insert into ct2 values ( {NOW + int(TIME_STEP * 0.6)}, {null_data} )")
+        tdSql.execute(
+            f"insert into ct2 values ( {NOW - (self.rows + 1) * int(TIME_STEP * 0.6)}, {null_data} )")
+        tdSql.execute(
+            f"insert into ct2 values ( {NOW - self.rows * int(TIME_STEP * 0.29) }, {null_data} )")
+
+        tdSql.execute(
+            f"insert into ct4 values ( {NOW + int(TIME_STEP * 0.8)}, {null_data} )")
+        tdSql.execute(
+            f"insert into ct4 values ( {NOW - (self.rows + 1) * int(TIME_STEP * 0.8)}, {null_data} )")
+        tdSql.execute(
+            f"insert into ct4 values ( {NOW - self.rows * int(TIME_STEP * 0.39)}, {null_data} )")
+
+        tdSql.execute(
+            f"insert into {NTBNAME} values ( {NOW + int(TIME_STEP * 1.2)}, {null_data} )")
+        tdSql.execute(
+            f"insert into {NTBNAME} values ( {NOW - (self.rows + 1) * int(TIME_STEP * 1.2)}, {null_data} )")
+        tdSql.execute(
+            f"insert into {NTBNAME} values ( {NOW - self.rows * int(TIME_STEP * 0.59)}, {null_data} )")
+
+    def run(self):
+        self.rows = 10
+
+        tdLog.printNoPrefix("==========step0:all check")
+
+        tdLog.printNoPrefix("==========step1:create table in normal database")
+        tdSql.prepare()
+        self.__create_tb()
+        self.__insert_data()
+        self.all_test()
+
+        # drop databases, create same name db、stb and sma index
+        tdSql.prepare()
+        self.__create_tb()
+        self.__insert_data()
+        self.all_test()
+
+        tdLog.printNoPrefix("==========step2:create table in rollup database")
+        tdSql.execute("create database db3 retentions 1s:4m,2s:8m,3s:12m")
+        tdSql.execute("use db3")
+        tdSql.query(f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(first) watermark 5s max_delay 1m sma({INT_COL})")
+
+        tdSql.execute("drop database if exists db1 ")
+        tdSql.execute("drop database if exists db2 ")
+
+        tdDnodes.stop(1)
+        tdDnodes.start(1)
+
+        tdLog.printNoPrefix("==========step4:after wal, all check again ")
+        tdSql.prepare()
+        self.__create_tb()
+        self.__insert_data()
+        self.all_test()
+
+        # drop databases, create same name db、stb and sma index
+        tdSql.prepare()
+        self.__create_tb()
+        self.__insert_data()
+        self.all_test()
+
+    def stop(self):
+        tdSql.close()
+        tdLog.success(f"{__file__} successfully executed")
+
+
+tdCases.addLinux(__file__, TDTestCase())
+tdCases.addWindows(__file__, TDTestCase())

tests/system-test/1-insert/create_retentions.py

@@ -1,6 +1,6 @@
 import datetime
 
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from typing import List
 from util.log import *
 from util.sql import *
@@ -36,36 +36,20 @@ NOW = int(datetime.datetime.timestamp(datetime.datetime.now()) * 1000)
 
 @dataclass
 class DataSet:
-    ts_data         : List[int]     = None
+    ts_data     : List[int]     = field(default_factory=list)
-    int_data        : List[int]     = None
+    int_data    : List[int]     = field(default_factory=list)
-    bint_data       : List[int]     = None
+    bint_data   : List[int]     = field(default_factory=list)
-    sint_data       : List[int]     = None
+    sint_data   : List[int]     = field(default_factory=list)
-    tint_data       : List[int]     = None
+    tint_data   : List[int]     = field(default_factory=list)
-    int_un_data     : List[int]     = None
+    int_un_data : List[int]     = field(default_factory=list)
-    bint_un_data    : List[int]     = None
+    bint_un_data: List[int]     = field(default_factory=list)
-    sint_un_data    : List[int]     = None
+    sint_un_data: List[int]     = field(default_factory=list)
-    tint_un_data    : List[int]     = None
+    tint_un_data: List[int]     = field(default_factory=list)
-    float_data      : List[float]   = None
+    float_data  : List[float]   = field(default_factory=list)
-    double_data     : List[float]   = None
+    double_data : List[float]   = field(default_factory=list)
-    bool_data       : List[int]     = None
+    bool_data   : List[int]     = field(default_factory=list)
-    binary_data     : List[str]     = None
+    binary_data : List[str]     = field(default_factory=list)
-    nchar_data      : List[str]     = None
+    nchar_data  : List[str]     = field(default_factory=list)
-
-    def __post_init__(self):
-        self.ts_data        = []
-        self.int_data       = []
-        self.bint_data      = []
-        self.sint_data      = []
-        self.tint_data      = []
-        self.int_un_data    = []
-        self.bint_un_data   = []
-        self.sint_un_data   = []
-        self.tint_un_data   = []
-        self.float_data     = []
-        self.double_data    = []
-        self.bool_data      = []
-        self.binary_data    = []
-        self.nchar_data     = []
 
 
 class TDTestCase:
@@ -107,15 +91,15 @@ class TDTestCase:
             f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(count) watermark  1min",
             f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(min) max_delay -1s",
             f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(min) watermark -1m",
-            # f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) watermark 1m ",
+            f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) watermark 1m ",
-            # f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) max_delay 1m ",
+            f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) max_delay 1m ",
             f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {BINARY_COL} binary(16)) tags (tag1 int) rollup(avg) watermark 1s",
-            f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {BINARY_COL} nchar(16)) tags (tag1 int) rollup(avg) max_delay 1m",
+            f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {NCHAR_COL} nchar(16)) tags (tag1 int) rollup(avg) max_delay 1m",
-            # f"create table ntb_1 ({PRIMARY_COL} timestamp, {INT_COL} int, {BINARY_COL} nchar(16)) rollup(avg) watermark 1s max_delay 1s",
+            # f"create table ntb_1 ({PRIMARY_COL} timestamp, {INT_COL} int, {NCHAR_COL} nchar(16)) rollup(avg) watermark 1s max_delay 1s",
-            # f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {BINARY_COL} nchar(16)) tags (tag1 int) " ,
+            f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {NCHAR_COL} nchar(16)) tags (tag1 int) " ,
-            # f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) " ,
+            f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) " ,
-            # f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int) " ,
+            f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int) " ,
-            # f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {BINARY_COL} nchar(16)) " ,
+            f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int, {BINARY_COL} nchar(16)) " ,
 
             # watermark, max_delay: [0, 900000], [ms, s, m, ?]
             f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(min) max_delay 1u",
@@ -136,8 +120,9 @@ class TDTestCase:
             f"create stable stb2 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(min) watermark 5s max_delay 1m",
             f"create stable stb3 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(max) watermark 5s max_delay 1m",
             f"create stable stb4 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(sum) watermark 5s max_delay 1m",
-            # f"create stable stb5 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(last) watermark 5s max_delay 1m",
+            f"create stable stb5 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(last) watermark 5s max_delay 1m",
-            # f"create stable stb6 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(first) watermark 5s max_delay 1m",
+            f"create stable stb6 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(first) watermark 5s max_delay 1m",
+            f"create stable stb7 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(first) watermark 5s max_delay 1m sma({INT_COL})",
         ]
 
     def test_create_stb(self):
@@ -150,7 +135,7 @@ class TDTestCase:
         # assert "rollup" in tdSql.description
         tdSql.checkRows(len(self.create_stable_sql_current))
 
-        # tdSql.execute("use db")  # because db is a noraml database, not a rollup database, should not be able to create a rollup database
+        tdSql.execute("use db")  # because db is a noraml database, not a rollup database, should not be able to create a rollup stable
         # tdSql.error(f"create stable nor_db_rollup_stb ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) watermark 5s max_delay 1m")
 
 
@@ -210,20 +195,6 @@ class TDTestCase:
             data_set.binary_data.append(f'binary{(rows - i)}')
             data_set.nchar_data.append(f'nchar_测试_{(rows - i)}')
 
-            # neg_data_set.ts_data.append(-1 * i)
-            # neg_data_set.int_data.append(-i)
-            # neg_data_set.bint_data.append(-11111 * i)
-            # neg_data_set.sint_data.append(-111 * i % 32767)
-            # neg_data_set.tint_data.append(-11 * i % 127)
-            # neg_data_set.int_un_data.append(-i)
-            # neg_data_set.bint_un_data.append(-11111 * i)
-            # neg_data_set.sint_un_data.append(-111 * i % 32767)
-            # neg_data_set.tint_un_data.append(-11 * i % 127)
-            # neg_data_set.float_data.append(-1.11 * i)
-            # neg_data_set.double_data.append(-1100.0011 * i)
-            # neg_data_set.binary_data.append(f'binary{i}')
-            # neg_data_set.nchar_data.append(f'nchar_测试_{i}')
-
         return data_set
 
     def __insert_data(self):
@@ -279,9 +250,14 @@ class TDTestCase:
 
         tdLog.printNoPrefix("==========step2:create table in rollup database")
         tdSql.execute("create database db3 retentions 1s:4m,2s:8m,3s:12m")
+
+        tdSql.execute("drop database if exists db1 ")
+        tdSql.execute("drop database if exists db2 ")
+
         tdSql.execute("use db3")
-        self.__create_tb()
+        # self.__create_tb()
-        self.__insert_data()
+        # self.__insert_data()
+        self.all_test()
 
         tdSql.execute("drop database if exists db1 ")
         tdSql.execute("drop database if exists db2 ")

tests/system-test/1-insert/time_range_wise.py

@@ -325,7 +325,7 @@ class TDTestCase:
     def __sma_create_check(self, sma:SMAschema):
         if  self.updatecfgDict["querySmaOptimize"] == 0:
             return False
-        # # TODO: if database is a rollup-db, can not create sma index
+        # TODO: if database is a rollup-db, can not create sma index
         # tdSql.query("select database()")
         # if sma.rollup_db :
         #     return False
@@ -493,8 +493,8 @@ class TDTestCase:
         err_sqls , cur_sqls = self.__drop_sma_sql
         for err_sql in err_sqls:
             self.sma_drop_check(err_sql)
-        # for cur_sql in cur_sqls:
+        for cur_sql in cur_sqls:
-        #     self.sma_drop_check(cur_sql)
+            self.sma_drop_check(cur_sql)
 
     def all_test(self):
         self.test_create_sma()
@@ -605,24 +605,23 @@ class TDTestCase:
         tdLog.printNoPrefix("==========step1:create table in normal database")
         tdSql.prepare()
         self.__create_tb()
-        # self.__insert_data()
+        self.__insert_data()
         self.all_test()
 
         # drop databases, create same name db、stb and sma index
-        # tdSql.prepare()
+        tdSql.prepare()
-        # self.__create_tb()
+        self.__create_tb()
-        # self.__insert_data()
+        self.__insert_data()
-        # self.all_test()
+        self.all_test()
-
-
-
-        return
 
         tdLog.printNoPrefix("==========step2:create table in rollup database")
         tdSql.execute("create database db3 retentions 1s:4m,2s:8m,3s:12m")
         tdSql.execute("use db3")
-        self.__create_tb()
+        # self.__create_tb()
-        self.__insert_data()
+        tdSql.execute(f"create stable stb1 ({PRIMARY_COL} timestamp, {INT_COL} int) tags (tag1 int) rollup(first) watermark 5s max_delay 1m sma({INT_COL}) ")
+        self.all_test()
+
+        # self.__insert_data()
 
         tdSql.execute("drop database if exists db1 ")
         tdSql.execute("drop database if exists db2 ")

tests/system-test/2-query/abs.py

@@ -139,7 +139,7 @@ class TDTestCase:
         )
         for i in range(4):
             tdSql.execute(
-                f'create table ct{i+1} using stb1 tags ( now(), {1*i}, {11111*i}, {111*i}, {11*i}, {1.11*i}, {11.11*i}, {i%2}, "binary{i}", "nchar{i}" )')
+                f'create table ct{i+1} using stb1 tags ( now(), {1*i}, {11111*i}, {111*i}, {1*i}, {1.11*i}, {11.11*i}, {i%2}, "binary{i}", "nchar{i}" )')
 
         for i in range(9):
             tdSql.execute(

tests/system-test/2-query/avg.py

@@ -14,7 +14,7 @@ class TDTestCase:
     def init(self, conn, logSql):
         tdLog.debug(f"start to excute {__file__}")
         tdSql.init(conn.cursor(), True)
-    
+
     def prepare_datas(self):
         tdSql.execute(
             '''create table stb1
@@ -22,7 +22,7 @@ class TDTestCase:
             tags (t1 int)
             '''
         )
-        
+
         tdSql.execute(
             '''
             create table t1
@@ -64,7 +64,7 @@ class TDTestCase:
             ( '2023-02-21 01:01:01.000', NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL )
             '''
         )
-    
+
     def check_avg(self ,origin_query , check_query):
         avg_result = tdSql.getResult(origin_query)
         origin_result = tdSql.getResult(check_query)
@@ -73,13 +73,13 @@ class TDTestCase:
         for row_index , row in enumerate(avg_result):
             for col_index , elem in enumerate(row):
                 if avg_result[row_index][col_index] != origin_result[row_index][col_index]:
-                    check_status = False 
+                    check_status = False
         if not check_status:
             tdLog.notice("avg function value has not as expected , sql is \"%s\" "%origin_query )
             sys.exit(1)
         else:
             tdLog.info("avg value check pass , it work as expected ,sql is \"%s\"   "%check_query )
-        
+
     def test_errors(self):
         error_sql_lists = [
             "select avg from t1",
@@ -113,42 +113,42 @@ class TDTestCase:
         ]
         for error_sql in error_sql_lists:
             tdSql.error(error_sql)
-    
+
     def support_types(self):
         type_error_sql_lists = [
-            "select avg(ts) from t1" , 
+            "select avg(ts) from t1" ,
             "select avg(c7) from t1",
             "select avg(c8) from t1",
             "select avg(c9) from t1",
-            "select avg(ts) from ct1" , 
+            "select avg(ts) from ct1" ,
             "select avg(c7) from ct1",
             "select avg(c8) from ct1",
             "select avg(c9) from ct1",
-            "select avg(ts) from ct3" , 
+            "select avg(ts) from ct3" ,
             "select avg(c7) from ct3",
             "select avg(c8) from ct3",
             "select avg(c9) from ct3",
-            "select avg(ts) from ct4" , 
+            "select avg(ts) from ct4" ,
             "select avg(c7) from ct4",
             "select avg(c8) from ct4",
             "select avg(c9) from ct4",
-            "select avg(ts) from stb1" , 
+            "select avg(ts) from stb1" ,
             "select avg(c7) from stb1",
             "select avg(c8) from stb1",
             "select avg(c9) from stb1" ,
 
-            "select avg(ts) from stbbb1" , 
+            "select avg(ts) from stbbb1" ,
             "select avg(c7) from stbbb1",
 
             "select avg(ts) from tbname",
             "select avg(c9) from tbname"
 
         ]
-        
+
         for type_sql in type_error_sql_lists:
             tdSql.error(type_sql)
-        
+
-        
+
         type_sql_lists = [
             "select avg(c1) from t1",
             "select avg(c2) from t1",
@@ -178,16 +178,16 @@ class TDTestCase:
             "select avg(c5) from stb1",
             "select avg(c6) from stb1",
 
-            "select avg(c6) as alisb from stb1", 
+            "select avg(c6) as alisb from stb1",
-            "select avg(c6) alisb from stb1", 
+            "select avg(c6) alisb from stb1",
         ]
 
         for type_sql in type_sql_lists:
             tdSql.query(type_sql)
-    
+
     def basic_avg_function(self):
 
-        # basic query 
+        # basic query
         tdSql.query("select c1 from ct3")
         tdSql.checkRows(0)
         tdSql.query("select c1 from t1")
@@ -207,18 +207,18 @@ class TDTestCase:
         tdSql.query("select avg(c5) from ct3")
         tdSql.checkRows(0)
         tdSql.query("select avg(c6) from ct3")
-        
+
         # used for regular table
         tdSql.query("select avg(c1) from t1")
         tdSql.checkData(0, 0, 5.000000000)
-       
+
-       
+
         tdSql.query("select ts,c1, c2, c3 , c4, c5 from t1")
         tdSql.checkData(1, 5, 1.11000)
         tdSql.checkData(3, 4, 33)
         tdSql.checkData(5, 5, None)
         self.check_avg(" select avg(c1) , avg(c2) , avg(c3) from t1 " , " select sum(c1)/count(c1) , sum(c2)/count(c2) , sum(c3)/count(c3) from t1 ")
-        
+
         # used for sub table
         tdSql.query("select avg(c1) from ct1")
         tdSql.checkData(0, 0, 4.846153846)
@@ -229,8 +229,8 @@ class TDTestCase:
         self.check_avg(" select avg(abs(c1)) , avg(abs(c2)) , avg(abs(c3)) from t1 " , " select sum(abs(c1))/count(c1) , sum(abs(c2))/count(c2) , sum(abs(c3))/count(c3) from t1 ")
         self.check_avg(" select avg(abs(c1)) , avg(abs(c2)) , avg(abs(c3)) from stb1 " , " select sum(abs(c1))/count(c1) , sum(abs(c2))/count(c2) , sum(abs(c3))/count(c3) from stb1 ")
 
-        # used for stable table 
+        # used for stable table
-        
+
         tdSql.query("select avg(c1) from stb1")
         tdSql.checkRows(1)
 
@@ -241,10 +241,10 @@ class TDTestCase:
         tdSql.error("select avg(c1) from tbname")
         tdSql.error("select avg(c1) from ct5")
 
-        # mix with common col 
+        # mix with common col
         tdSql.error("select c1, avg(c1) from ct1")
         tdSql.error("select c1, avg(c1) from ct4")
-        
+
 
         # mix with common functions
         tdSql.error("select c1, avg(c1),c5, floor(c5) from ct4 ")
@@ -278,11 +278,11 @@ class TDTestCase:
         tdSql.query("select count(*) from stb1 ")
         tdSql.checkData(0,0,25)
 
-        # bug fix for compute 
+        # bug fix for compute
         tdSql.error("select c1, avg(c1) -0 ,ceil(c1)-0 from ct4 ")
         tdSql.error(" select c1, avg(c1) -0 ,avg(ceil(c1-0.1))-0.1 from ct4")
 
-        # mix with nest query 
+        # mix with nest query
         self.check_avg("select avg(col) from (select abs(c1) col from stb1)" , "select avg(abs(c1)) from stb1")
         self.check_avg("select avg(col) from (select ceil(abs(c1)) col from stb1)" , "select avg(abs(c1)) from stb1")
 
@@ -297,7 +297,7 @@ class TDTestCase:
         tdSql.query(" select avg(c1) from stb1 where c1 is null ")
         tdSql.checkRows(0)
 
-        
+
     def avg_func_filter(self):
         tdSql.execute("use db")
         tdSql.query(" select avg(c1), avg(c1) -0 ,avg(ceil(c1-0.1))-0 ,avg(floor(c1+0.1))-0.1 ,avg(ceil(log(c1,2)-0.5)) from ct4 where c1>5 ")
@@ -324,7 +324,7 @@ class TDTestCase:
 
     def avg_Arithmetic(self):
         pass
-    
+
     def check_boundary_values(self):
 
         tdSql.execute("drop database if exists bound_test")
@@ -344,11 +344,11 @@ class TDTestCase:
         tdSql.execute(
                 f"insert into sub1_bound values ( now(), 2147483646, 9223372036854775806, 32766, 126, 3.40E+38, 1.7e+308, True, 'binary_tb1', 'nchar_tb1', now() )"
             )
-        
+
         tdSql.execute(
                 f"insert into sub1_bound values ( now(), 2147483645, 9223372036854775805, 32765, 125, 3.40E+37, 1.7e+307, True, 'binary_tb1', 'nchar_tb1', now() )"
             )
-        
+
         tdSql.execute(
                 f"insert into sub1_bound values ( now(), 2147483644, 9223372036854775804, 32764, 124, 3.40E+37, 1.7e+307, True, 'binary_tb1', 'nchar_tb1', now() )"
             )
@@ -359,14 +359,14 @@ class TDTestCase:
         tdSql.execute(
                 f"insert into sub1_bound values ( now(), 2147483646, 9223372036854775806, 32766, 126, 3.40E+38, 1.7e+308, True, 'binary_tb1', 'nchar_tb1', now() )"
             )
-        
+
 
         tdSql.error(
                 f"insert into sub1_bound values ( now()+1s, 2147483648, 9223372036854775808, 32768, 128, 3.40E+38, 1.7e+308, True, 'binary_tb1', 'nchar_tb1', now() )"
             )
         self.check_avg("select avg(c1), avg(c2), avg(c3) , avg(c4), avg(c5) ,avg(c6) from sub1_bound " , " select sum(c1)/count(c1), sum(c2)/count(c2) ,sum(c3)/count(c3), sum(c4)/count(c4), sum(c5)/count(c5) ,sum(c6)/count(c6) from sub1_bound ")
 
-        
+
         # check basic elem for table per row
         tdSql.query("select avg(c1) ,avg(c2) , avg(c3) , avg(c4), avg(c5), avg(c6) from sub1_bound ")
         tdSql.checkRows(1)
@@ -376,8 +376,8 @@ class TDTestCase:
         tdSql.checkData(0,3,53.571428571)
         tdSql.checkData(0,4,5.828571332045761e+37)
         # tdSql.checkData(0,5,None)
-        
+
-       
+
         # check  + - * / in functions
         tdSql.query(" select avg(c1+1) ,avg(c2) , avg(c3*1) , avg(c4/2), avg(c5)/2, avg(c6) from sub1_bound ")
         tdSql.checkData(0,0,920350134.5714285)
@@ -386,33 +386,33 @@ class TDTestCase:
         tdSql.checkData(0,3,26.785714286)
         tdSql.checkData(0,4,2.9142856660228804e+37)
         # tdSql.checkData(0,5,None)
-        
+
-       
+
 
     def run(self):  # sourcery skip: extract-duplicate-method, remove-redundant-fstring
         tdSql.prepare()
 
         tdLog.printNoPrefix("==========step1:create table ==============")
-        
+
         self.prepare_datas()
 
-        tdLog.printNoPrefix("==========step2:test errors ==============")    
+        tdLog.printNoPrefix("==========step2:test errors ==============")
 
         self.test_errors()
-        
+
-        tdLog.printNoPrefix("==========step3:support types ============") 
+        tdLog.printNoPrefix("==========step3:support types ============")
 
         self.support_types()
 
-        tdLog.printNoPrefix("==========step4: avg basic query ============") 
+        tdLog.printNoPrefix("==========step4: avg basic query ============")
 
         self.basic_avg_function()
 
-        tdLog.printNoPrefix("==========step5: avg boundary query ============") 
+        tdLog.printNoPrefix("==========step5: avg boundary query ============")
 
         self.check_boundary_values()
 
-        tdLog.printNoPrefix("==========step6: avg filter query ============") 
+        tdLog.printNoPrefix("==========step6: avg filter query ============")
 
         self.avg_func_filter()
 

tests/system-test/2-query/distribute_agg_apercentile.py

@@ -36,7 +36,7 @@ class TDTestCase:
             '''
         )
         for i in range(20):
-            tdSql.execute(f'create table ct{i+1} using stb1 tags ( now(), {1*i}, {11111*i}, {111*i}, {11*i}, {1.11*i}, {11.11*i}, {i%2}, "binary{i}", "nchar{i}" )')
+            tdSql.execute(f'create table ct{i+1} using stb1 tags ( now(), {1*i}, {11111*i}, {111*i}, {1*i}, {1.11*i}, {11.11*i}, {i%2}, "binary{i}", "nchar{i}" )')
 
         for i in range(9):
             tdSql.execute(