Merge branch '3.0' into feat/TD-22746

cmake/cmake.define

@@ -119,6 +119,9 @@ ELSE ()
         SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS}     -Werror -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fsanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=shift-base -fno-sanitize=alignment -g3 -Wformat=0")
         SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-literal-suffix -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fsanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=shift-base -fno-sanitize=alignment -g3 -Wformat=0")
 	      MESSAGE(STATUS "Compile with Address Sanitizer!")
+    ELSEIF (${BUILD_RELEASE})
+        SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror -Werror=return-type -fPIC -O3 -Wformat=2 -Wno-format-nonliteral -Wno-format-truncation -Wno-format-y2k")
+        SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Wno-reserved-user-defined-literal -Wno-literal-suffix -Werror=return-type -fPIC -O3 -Wformat=2 -Wno-format-nonliteral -Wno-format-truncation -Wno-format-y2k")
     ELSE ()
         SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror -Werror=return-type -fPIC -gdwarf-2 -g3 -Wformat=2 -Wno-format-nonliteral -Wno-format-truncation -Wno-format-y2k")
         SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-reserved-user-defined-literal -Wno-literal-suffix -Werror=return-type -fPIC -gdwarf-2 -g3 -Wformat=2 -Wno-format-nonliteral -Wno-format-truncation -Wno-format-y2k")

cmake/cmake.options

@@ -171,3 +171,8 @@ option(
    ON 
 )
 
+option(
+   BUILD_RELEASE
+   "If build release version"
+   OFF
+)

cmake/rocksdb_CMakeLists.txt.in

@@ -1,8 +1,8 @@
 
 # rocksdb
 ExternalProject_Add(rocksdb
-        GIT_REPOSITORY https://github.com/taosdata-contrib/rocksdb.git
+        GIT_REPOSITORY https://github.com/facebook/rocksdb.git
-        GIT_TAG v6.23.3
+        GIT_TAG v8.1.1
         SOURCE_DIR "${TD_CONTRIB_DIR}/rocksdb"
         CONFIGURE_COMMAND ""
         BUILD_COMMAND ""

include/client/taos.h

@@ -310,6 +310,7 @@ DLL_EXPORT void           tmq_conf_set_auto_commit_cb(tmq_conf_t *conf, tmq_comm
 DLL_EXPORT const char *tmq_get_topic_name(TAOS_RES *res);
 DLL_EXPORT const char *tmq_get_db_name(TAOS_RES *res);
 DLL_EXPORT int32_t     tmq_get_vgroup_id(TAOS_RES *res);
+DLL_EXPORT int64_t     tmq_get_vgroup_offset(TAOS_RES* res);
 
 /* ------------------------------ TAOSX -----------------------------------*/
 // note: following apis are unstable

include/common/tmsg.h

@@ -952,6 +952,9 @@ int32_t tSerializeSVDropTtlTableReq(void* buf, int32_t bufLen, SVDropTtlTableReq
 int32_t tDeserializeSVDropTtlTableReq(void* buf, int32_t bufLen, SVDropTtlTableReq* pReq);
 
 typedef struct {
+  char    db[TSDB_DB_FNAME_LEN];
+  int64_t dbId;
+  int32_t cfgVersion;
   int32_t numOfVgroups;
   int32_t numOfStables;
   int32_t buffer;
@@ -984,8 +987,13 @@ typedef struct {
   int16_t sstTrigger;
 } SDbCfgRsp;
 
+typedef SDbCfgRsp     SDbCfgInfo;
+
+int32_t tSerializeSDbCfgRspImpl(SEncoder *encoder, const SDbCfgRsp *pRsp);
 int32_t tSerializeSDbCfgRsp(void* buf, int32_t bufLen, const SDbCfgRsp* pRsp);
 int32_t tDeserializeSDbCfgRsp(void* buf, int32_t bufLen, SDbCfgRsp* pRsp);
+int32_t tDeserializeSDbCfgRspImpl(SDecoder* decoder, SDbCfgRsp *pRsp);
+void tFreeSDbCfgRsp(SDbCfgRsp *pRsp);
 
 typedef struct {
   int32_t rowNum;
@@ -1042,12 +1050,17 @@ int32_t tDeserializeSDnodeListRsp(void* buf, int32_t bufLen, SDnodeListRsp* pRsp
 void    tFreeSDnodeListRsp(SDnodeListRsp* pRsp);
 
 typedef struct {
-  SArray* pArray;  // Array of SUseDbRsp
+  SUseDbRsp *useDbRsp;
-} SUseDbBatchRsp;
+  SDbCfgRsp *cfgRsp;
+} SDbHbRsp;
 
-int32_t tSerializeSUseDbBatchRsp(void* buf, int32_t bufLen, SUseDbBatchRsp* pRsp);
+typedef struct {
-int32_t tDeserializeSUseDbBatchRsp(void* buf, int32_t bufLen, SUseDbBatchRsp* pRsp);
+  SArray* pArray;  // Array of SDbHbRsp
-void    tFreeSUseDbBatchRsp(SUseDbBatchRsp* pRsp);
+} SDbHbBatchRsp;
+
+int32_t tSerializeSDbHbBatchRsp(void* buf, int32_t bufLen, SDbHbBatchRsp* pRsp);
+int32_t tDeserializeSDbHbBatchRsp(void* buf, int32_t bufLen, SDbHbBatchRsp* pRsp);
+void    tFreeSDbHbBatchRsp(SDbHbBatchRsp* pRsp);
 
 typedef struct {
   SArray* pArray;  // Array of SGetUserAuthRsp

include/libs/catalog/catalog.h

@@ -129,13 +129,14 @@ typedef struct SSTableVersion {
   int32_t  smaVer;
 } SSTableVersion;
 
-typedef struct SDbVgVersion {
+typedef struct SDbCacheInfo {
   char    dbFName[TSDB_DB_FNAME_LEN];
   int64_t dbId;
   int32_t vgVersion;
+  int32_t cfgVersion;
   int32_t numOfTable;  // unit is TSDB_TABLE_NUM_UNIT
   int64_t stateTs;
-} SDbVgVersion;
+} SDbCacheInfo;
 
 typedef struct STbSVersion {
   char*   tbFName;
@@ -148,7 +149,6 @@ typedef struct SUserAuthVersion {
   int32_t version;
 } SUserAuthVersion;
 
-typedef SDbCfgRsp     SDbCfgInfo;
 typedef SUserIndexRsp SIndexInfo;
 
 typedef void (*catalogCallback)(SMetaData* pResult, void* param, int32_t code);
@@ -180,6 +180,8 @@ int32_t catalogGetDBVgInfo(SCatalog* pCtg, SRequestConnInfo* pConn, const char*
 
 int32_t catalogUpdateDBVgInfo(SCatalog* pCatalog, const char* dbName, uint64_t dbId, SDBVgInfo* dbInfo);
 
+int32_t catalogUpdateDbCfg(SCatalog* pCtg, const char* dbFName, uint64_t dbId, SDbCfgInfo* cfgInfo);
+
 int32_t catalogRemoveDB(SCatalog* pCatalog, const char* dbName, uint64_t dbId);
 
 int32_t catalogRemoveTableMeta(SCatalog* pCtg, SName* pTableName);
@@ -304,7 +306,7 @@ int32_t catalogGetDnodeList(SCatalog* pCatalog, SRequestConnInfo* pConn, SArray*
 
 int32_t catalogGetExpiredSTables(SCatalog* pCatalog, SSTableVersion** stables, uint32_t* num);
 
-int32_t catalogGetExpiredDBs(SCatalog* pCatalog, SDbVgVersion** dbs, uint32_t* num);
+int32_t catalogGetExpiredDBs(SCatalog* pCatalog, SDbCacheInfo** dbs, uint32_t* num);
 
 int32_t catalogGetExpiredUsers(SCatalog* pCtg, SUserAuthVersion** users, uint32_t* num);
 

include/libs/function/functionMgt.h

@@ -230,6 +230,7 @@ typedef enum EFuncDataRequired {
   FUNC_DATA_REQUIRED_SMA_LOAD,
   FUNC_DATA_REQUIRED_NOT_LOAD,
   FUNC_DATA_REQUIRED_FILTEROUT,
+  FUNC_DATA_REQUIRED_ALL_FILTEROUT,
 } EFuncDataRequired;
 
 EFuncDataRequired fmFuncDataRequired(SFunctionNode* pFunc, STimeWindow* pTimeWindow);

include/libs/nodes/nodes.h

@@ -313,6 +313,7 @@ void    nodesDestroyAllocator(int64_t allocatorId);
 
 SNode* nodesMakeNode(ENodeType type);
 void   nodesDestroyNode(SNode* pNode);
+void   nodesFree(void* p);
 
 SNodeList* nodesMakeList();
 int32_t    nodesListAppend(SNodeList* pList, SNode* pNode);

include/libs/qcom/query.h

@@ -256,6 +256,7 @@ int32_t cloneTableMeta(STableMeta* pSrc, STableMeta** pDst);
 int32_t cloneDbVgInfo(SDBVgInfo* pSrc, SDBVgInfo** pDst);
 int32_t cloneSVreateTbReq(SVCreateTbReq* pSrc, SVCreateTbReq** pDst);
 void    freeVgInfo(SDBVgInfo* vgInfo);
+void    freeDbCfgInfo(SDbCfgInfo *pInfo);
 
 extern int32_t (*queryBuildMsg[TDMT_MAX])(void* input, char** msg, int32_t msgSize, int32_t* msgLen,
                                           void* (*mallocFp)(int64_t));

include/libs/stream/tstream.h

@@ -134,7 +134,6 @@ typedef struct {
 typedef struct {
   int8_t       type;
   int64_t      ver;
-  int32_t*     dataRef;
   SSDataBlock* pBlock;
 } SStreamRefDataBlock;
 

include/libs/wal/wal.h

@@ -135,6 +135,7 @@ typedef struct {
   int8_t scanUncommited;
   int8_t scanNotApplied;
   int8_t scanMeta;
+  int8_t deleteMsg;
   int8_t enableRef;
 } SWalFilterCond;
 
@@ -212,8 +213,7 @@ SWalRef *walRefCommittedVer(SWal *);
 
 SWalRef *walOpenRef(SWal *);
 void     walCloseRef(SWal *pWal, int64_t refId);
-int32_t  walRefVer(SWalRef *, int64_t ver);
+int32_t  walSetRefVer(SWalRef *, int64_t ver);
-void     walUnrefVer(SWalRef *);
 
 // helper function for raft
 bool walLogExist(SWal *, int64_t ver);

source/client/inc/clientLog.h

@@ -25,6 +25,7 @@ extern "C" {
 // clang-format off
 #define tscFatal(...)  do { if (cDebugFlag & DEBUG_FATAL) { taosPrintLog("TSC FATAL ", DEBUG_FATAL, cDebugFlag, __VA_ARGS__); }} while(0)
 #define tscError(...)  do { if (cDebugFlag & DEBUG_ERROR) { taosPrintLog("TSC ERROR ", DEBUG_ERROR, cDebugFlag, __VA_ARGS__); }} while(0)
+#define tscErrorL(...) do { if (cDebugFlag & DEBUG_ERROR) { taosPrintLongString("TSC ERROR ", DEBUG_ERROR, cDebugFlag, __VA_ARGS__); }} while(0)
 #define tscWarn(...)   do { if (cDebugFlag & DEBUG_WARN)  { taosPrintLog("TSC WARN ", DEBUG_WARN, cDebugFlag, __VA_ARGS__); }}  while(0)
 #define tscWarnL(...)  do { if (cDebugFlag & DEBUG_WARN)  { taosPrintLongString("TSC WARN ", DEBUG_WARN, cDebugFlag, __VA_ARGS__); }}  while(0)
 #define tscInfo(...)   do { if (cDebugFlag & DEBUG_INFO)  { taosPrintLog("TSC ", DEBUG_INFO, cDebugFlag, __VA_ARGS__); }} while(0)
@@ -32,6 +33,8 @@ extern "C" {
 #define tscTrace(...)  do { if (cDebugFlag & DEBUG_TRACE) { taosPrintLog("TSC ", DEBUG_TRACE, cDebugFlag, __VA_ARGS__); }} while(0)
 #define tscDebugL(...) do { if (cDebugFlag & DEBUG_DEBUG) { taosPrintLongString("TSC ", DEBUG_DEBUG, cDebugFlag, __VA_ARGS__); }} while(0)
 #define tscPerf(...)   do { if (cDebugFlag & DEBUG_INFO)  { taosPrintLog("TSC ", 0, cDebugFlag, __VA_ARGS__); }} while(0)
+#define tscLog(...)    do { taosPrintLog("TSC ", 0, DEBUG_FILE, __VA_ARGS__); } while(0)
+#define tscLogL(...)   do { taosPrintLongString("TSC ", 0, DEBUG_FILE, __VA_ARGS__); } while(0)
 // clang-format on
 
 #ifdef __cplusplus

source/client/src/clientEnv.c

@@ -110,7 +110,7 @@ static void deregisterRequest(SRequestObj *pRequest) {
   if (duration >= (tsSlowLogThreshold * 1000000UL)) {
     atomic_add_fetch_64((int64_t *)&pActivity->numOfSlowQueries, 1);
     if (tsSlowLogScope & reqType) {
-      taosPrintSlowLog("PID:%d, Conn:%u, QID:0x%" PRIx64 ", Start:%" PRId64 ", Duration:%" PRId64 "us, SQL:%s", 
+      taosPrintSlowLog("PID:%d, Conn:%u, QID:0x%" PRIx64 ", Start:%" PRId64 ", Duration:%" PRId64 "us, SQL:%s",
         taosGetPId(), pTscObj->connId, pRequest->requestId, pRequest->metric.start, duration, pRequest->sqlstr);
     }
   }
@@ -449,6 +449,7 @@ static void *tscCrashReportThreadFp(void *param) {
         tscError("failed to send crash report");
         if (pFile) {
           taosReleaseCrashLogFile(pFile, false);
+          pFile = NULL;
           continue;
         }
       } else {
@@ -468,6 +469,7 @@ static void *tscCrashReportThreadFp(void *param) {
 
     if (pFile) {
       taosReleaseCrashLogFile(pFile, truncateFile);
+      pFile = NULL;
       truncateFile = false;
     }
 

source/client/src/clientHb.c

@@ -24,6 +24,8 @@ typedef struct {
     struct {
       int64_t clusterId;
       int32_t passKeyCnt;
+      int32_t passVer;
+      int32_t reqCnt;
     };
   };
 } SHbParam;
@@ -149,48 +151,52 @@ _return:
 static int32_t hbProcessDBInfoRsp(void *value, int32_t valueLen, struct SCatalog *pCatalog) {
   int32_t code = 0;
 
-  SUseDbBatchRsp batchUseRsp = {0};
+  SDbHbBatchRsp batchRsp = {0};
-  if (tDeserializeSUseDbBatchRsp(value, valueLen, &batchUseRsp) != 0) {
+  if (tDeserializeSDbHbBatchRsp(value, valueLen, &batchRsp) != 0) {
     terrno = TSDB_CODE_INVALID_MSG;
-    return -1;
+    code = terrno;
+    goto _return;
   }
 
-  int32_t numOfBatchs = taosArrayGetSize(batchUseRsp.pArray);
+  int32_t numOfBatchs = taosArrayGetSize(batchRsp.pArray);
   for (int32_t i = 0; i < numOfBatchs; ++i) {
-    SUseDbRsp *rsp = taosArrayGet(batchUseRsp.pArray, i);
+    SDbHbRsp *rsp = taosArrayGet(batchRsp.pArray, i);
-    tscDebug("hb db rsp, db:%s, vgVersion:%d, stateTs:%" PRId64 ", uid:%" PRIx64, rsp->db, rsp->vgVersion, rsp->stateTs,
+    if (rsp->useDbRsp) {
-             rsp->uid);
+      tscDebug("hb use db rsp, db:%s, vgVersion:%d, stateTs:%" PRId64 ", uid:%" PRIx64,
+        rsp->useDbRsp->db, rsp->useDbRsp->vgVersion, rsp->useDbRsp->stateTs, rsp->useDbRsp->uid);
 
-    if (rsp->vgVersion < 0) {
+      if (rsp->useDbRsp->vgVersion < 0) {
-      code = catalogRemoveDB(pCatalog, rsp->db, rsp->uid);
+        code = catalogRemoveDB(pCatalog, rsp->useDbRsp->db, rsp->useDbRsp->uid);
-    } else {
+      } else {
-      SDBVgInfo *vgInfo = NULL;
+        SDBVgInfo *vgInfo = NULL;
-      code = hbGenerateVgInfoFromRsp(&vgInfo, rsp);
+        code = hbGenerateVgInfoFromRsp(&vgInfo, rsp->useDbRsp);
-      if (TSDB_CODE_SUCCESS != code) {
-        goto _return;
-      }
-
-      catalogUpdateDBVgInfo(pCatalog, rsp->db, rsp->uid, vgInfo);
-
-      if (IS_SYS_DBNAME(rsp->db)) {
-        code = hbGenerateVgInfoFromRsp(&vgInfo, rsp);
         if (TSDB_CODE_SUCCESS != code) {
           goto _return;
         }
 
-        catalogUpdateDBVgInfo(pCatalog, (rsp->db[0] == 'i') ? TSDB_PERFORMANCE_SCHEMA_DB : TSDB_INFORMATION_SCHEMA_DB,
+        catalogUpdateDBVgInfo(pCatalog, rsp->useDbRsp->db, rsp->useDbRsp->uid, vgInfo);
-                              rsp->uid, vgInfo);
+
+        if (IS_SYS_DBNAME(rsp->useDbRsp->db)) {
+          code = hbGenerateVgInfoFromRsp(&vgInfo, rsp->useDbRsp);
+          if (TSDB_CODE_SUCCESS != code) {
+            goto _return;
+          }
+
+          catalogUpdateDBVgInfo(pCatalog, (rsp->useDbRsp->db[0] == 'i') ? TSDB_PERFORMANCE_SCHEMA_DB : TSDB_INFORMATION_SCHEMA_DB, rsp->useDbRsp->uid, vgInfo);
+        }
       }
     }
 
-    if (code) {
+    if (rsp->cfgRsp) {
-      goto _return;
+      tscDebug("hb db cfg rsp, db:%s, cfgVersion:%d", rsp->cfgRsp->db, rsp->cfgRsp->cfgVersion);
+      catalogUpdateDbCfg(pCatalog, rsp->cfgRsp->db, rsp->cfgRsp->dbId, rsp->cfgRsp);
+      rsp->cfgRsp = NULL;
     }
   }
 
 _return:
 
-  tFreeSUseDbBatchRsp(&batchUseRsp);
+  tFreeSDbHbBatchRsp(&batchRsp);
   return code;
 }
 
@@ -536,14 +542,20 @@ int32_t hbGetQueryBasicInfo(SClientHbKey *connKey, SClientHbReq *req) {
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t hbGetUserBasicInfo(SClientHbKey *connKey, SClientHbReq *req) {
+static int32_t hbGetUserBasicInfo(SClientHbKey *connKey, SHbParam *param, SClientHbReq *req) {
   STscObj *pTscObj = (STscObj *)acquireTscObj(connKey->tscRid);
   if (!pTscObj) {
     tscWarn("tscObj rid %" PRIx64 " not exist", connKey->tscRid);
     return TSDB_CODE_APP_ERROR;
   }
 
-  int32_t           code = 0;
+  int32_t code = 0;
+
+  if (param && (param->passVer != INT32_MIN) && (param->passVer <= pTscObj->passInfo.ver)) {
+    tscDebug("hb got user basic info, no need since passVer %d <= %d", param->passVer, pTscObj->passInfo.ver);
+    goto _return;
+  }
+
   SUserPassVersion *user = taosMemoryMalloc(sizeof(SUserPassVersion));
   if (!user) {
     code = TSDB_CODE_OUT_OF_MEMORY;
@@ -570,6 +582,11 @@ static int32_t hbGetUserBasicInfo(SClientHbKey *connKey, SClientHbReq *req) {
     goto _return;
   }
 
+  // assign the passVer
+  if (param) {
+    param->passVer = pTscObj->passInfo.ver;
+  }
+
 _return:
   releaseTscObj(connKey->tscRid);
   if (code) {
@@ -617,7 +634,7 @@ int32_t hbGetExpiredUserInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, S
 }
 
 int32_t hbGetExpiredDBInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, SClientHbReq *req) {
-  SDbVgVersion *dbs = NULL;
+  SDbCacheInfo *dbs = NULL;
   uint32_t      dbNum = 0;
   int32_t       code = 0;
 
@@ -632,19 +649,20 @@ int32_t hbGetExpiredDBInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, SCl
   }
 
   for (int32_t i = 0; i < dbNum; ++i) {
-    SDbVgVersion *db = &dbs[i];
+    SDbCacheInfo *db = &dbs[i];
-    tscDebug("the %dth expired dbFName:%s, dbId:%" PRId64 ", vgVersion:%d, numOfTable:%d, startTs:%" PRId64, i,
+    tscDebug("the %dth expired dbFName:%s, dbId:%" PRId64 ", vgVersion:%d, cfgVersion:%d, numOfTable:%d, startTs:%" PRId64,
-             db->dbFName, db->dbId, db->vgVersion, db->numOfTable, db->stateTs);
+      i, db->dbFName, db->dbId, db->vgVersion, db->cfgVersion, db->numOfTable, db->stateTs);
 
     db->dbId = htobe64(db->dbId);
     db->vgVersion = htonl(db->vgVersion);
+    db->cfgVersion = htonl(db->cfgVersion);
     db->numOfTable = htonl(db->numOfTable);
     db->stateTs = htobe64(db->stateTs);
   }
 
   SKv kv = {
       .key = HEARTBEAT_KEY_DBINFO,
-      .valueLen = sizeof(SDbVgVersion) * dbNum,
+      .valueLen = sizeof(SDbCacheInfo) * dbNum,
       .value = dbs,
   };
 
@@ -714,13 +732,16 @@ int32_t hbGetAppInfo(int64_t clusterId, SClientHbReq *req) {
 }
 
 int32_t hbQueryHbReqHandle(SClientHbKey *connKey, void *param, SClientHbReq *req) {
-  SHbParam        *hbParam = (SHbParam *)param;
+  int32_t   code = 0;
-  struct SCatalog *pCatalog = NULL;
+  SHbParam *hbParam = (SHbParam *)param;
+  SCatalog *pCatalog = NULL;
 
-  int32_t code = catalogGetHandle(hbParam->clusterId, &pCatalog);
+  if (hbParam->reqCnt == 0) {
-  if (code != TSDB_CODE_SUCCESS) {
+    code = catalogGetHandle(hbParam->clusterId, &pCatalog);
-    tscWarn("catalogGetHandle failed, clusterId:%" PRIx64 ", error:%s", hbParam->clusterId, tstrerror(code));
+    if (code != TSDB_CODE_SUCCESS) {
-    return code;
+      tscWarn("catalogGetHandle failed, clusterId:%" PRIx64 ", error:%s", hbParam->clusterId, tstrerror(code));
+      return code;
+    }
   }
 
   hbGetAppInfo(hbParam->clusterId, req);
@@ -728,23 +749,27 @@ int32_t hbQueryHbReqHandle(SClientHbKey *connKey, void *param, SClientHbReq *req
   hbGetQueryBasicInfo(connKey, req);
 
   if (hbParam->passKeyCnt > 0) {
-    hbGetUserBasicInfo(connKey, req);
+    hbGetUserBasicInfo(connKey, hbParam, req);
   }
 
-  code = hbGetExpiredUserInfo(connKey, pCatalog, req);
+  if (hbParam->reqCnt == 0) {
-  if (TSDB_CODE_SUCCESS != code) {
+    code = hbGetExpiredUserInfo(connKey, pCatalog, req);
-    return code;
+    if (TSDB_CODE_SUCCESS != code) {
+      return code;
+    }
+
+    code = hbGetExpiredDBInfo(connKey, pCatalog, req);
+    if (TSDB_CODE_SUCCESS != code) {
+      return code;
+    }
+
+    code = hbGetExpiredStbInfo(connKey, pCatalog, req);
+    if (TSDB_CODE_SUCCESS != code) {
+      return code;
+    }
   }
 
-  code = hbGetExpiredDBInfo(connKey, pCatalog, req);
+  ++hbParam->reqCnt; // success to get catalog info
-  if (TSDB_CODE_SUCCESS != code) {
-    return code;
-  }
-
-  code = hbGetExpiredStbInfo(connKey, pCatalog, req);
-  if (TSDB_CODE_SUCCESS != code) {
-    return code;
-  }
 
   return TSDB_CODE_SUCCESS;
 }
@@ -766,55 +791,47 @@ SClientHbBatchReq *hbGatherAllInfo(SAppHbMgr *pAppHbMgr) {
   }
   int32_t connKeyCnt = atomic_load_32(&pAppHbMgr->connKeyCnt);
   pBatchReq->reqs = taosArrayInit(connKeyCnt, sizeof(SClientHbReq));
-
+  if (!pBatchReq->reqs) {
-  int64_t rid = -1;
-  int32_t code = 0;
-
-  void *pIter = taosHashIterate(pAppHbMgr->activeInfo, NULL);
-
-  SClientHbReq *pOneReq = pIter;
-  SClientHbKey *connKey = pOneReq ? &pOneReq->connKey : NULL;
-  if (connKey != NULL) rid = connKey->tscRid;
-
-  STscObj *pTscObj = (STscObj *)acquireTscObj(rid);
-  if (pTscObj == NULL) {
     tFreeClientHbBatchReq(pBatchReq);
     return NULL;
   }
 
-  while (pIter != NULL) {
+  void    *pIter = NULL;
+  SHbParam param = {0};
+  while ((pIter = taosHashIterate(pAppHbMgr->activeInfo, pIter))) {
+    SClientHbReq *pOneReq = pIter;
+    SClientHbKey *connKey = &pOneReq->connKey;
+    STscObj      *pTscObj = (STscObj *)acquireTscObj(connKey->tscRid);
+
+    if (!pTscObj) {
+      continue;
+    }
+
     pOneReq = taosArrayPush(pBatchReq->reqs, pOneReq);
-    SHbParam param;
+
-    switch (pOneReq->connKey.connType) {
+    switch (connKey->connType) {
       case CONN_TYPE__QUERY: {
-        param.clusterId = pOneReq->clusterId;
+        if (param.clusterId == 0) {
+          // init
+          param.clusterId = pOneReq->clusterId;
+          param.passVer = INT32_MIN;
+        }
         param.passKeyCnt = atomic_load_32(&pAppHbMgr->passKeyCnt);
         break;
       }
       default:
         break;
     }
-    if (clientHbMgr.reqHandle[pOneReq->connKey.connType]) {
+    if (clientHbMgr.reqHandle[connKey->connType]) {
-      code = (*clientHbMgr.reqHandle[pOneReq->connKey.connType])(&pOneReq->connKey, &param, pOneReq);
+      int32_t code = (*clientHbMgr.reqHandle[connKey->connType])(connKey, &param, pOneReq);
       if (code) {
         tscWarn("hbGatherAllInfo failed since %s, tscRid:%" PRIi64 ", connType:%" PRIi8, tstrerror(code),
-                pOneReq->connKey.tscRid, pOneReq->connKey.connType);
+                connKey->tscRid, connKey->connType);
       }
     }
-    break;
 
-#if 0
+    releaseTscObj(connKey->tscRid);
-    if (code) {
-      pIter = taosHashIterate(pAppHbMgr->activeInfo, pIter);
-      pOneReq = pIter;
-      continue;
-    }
-
-    pIter = taosHashIterate(pAppHbMgr->activeInfo, pIter);
-    pOneReq = pIter;
-#endif
   }
-  releaseTscObj(rid);
 
   return pBatchReq;
 }
@@ -885,7 +902,6 @@ static void *hbThreadFunc(void *param) {
       hbGatherAppInfo();
     }
 
-    SArray *mgr = taosArrayInit(sz, sizeof(void *));
     for (int i = 0; i < sz; i++) {
       SAppHbMgr *pAppHbMgr = taosArrayGetP(clientHbMgr.appHbMgrs, i);
       if (pAppHbMgr == NULL) {
@@ -894,7 +910,6 @@ static void *hbThreadFunc(void *param) {
 
       int32_t connCnt = atomic_load_32(&pAppHbMgr->connKeyCnt);
       if (connCnt == 0) {
-        taosArrayPush(mgr, &pAppHbMgr);
         continue;
       }
       SClientHbBatchReq *pReq = hbGatherAllInfo(pAppHbMgr);
@@ -908,7 +923,6 @@ static void *hbThreadFunc(void *param) {
         terrno = TSDB_CODE_OUT_OF_MEMORY;
         tFreeClientHbBatchReq(pReq);
         // hbClearReqInfo(pAppHbMgr);
-        taosArrayPush(mgr, &pAppHbMgr);
         break;
       }
 
@@ -920,7 +934,6 @@ static void *hbThreadFunc(void *param) {
         tFreeClientHbBatchReq(pReq);
         // hbClearReqInfo(pAppHbMgr);
         taosMemoryFree(buf);
-        taosArrayPush(mgr, &pAppHbMgr);
         break;
       }
       pInfo->fp = hbAsyncCallBack;
@@ -941,12 +954,8 @@ static void *hbThreadFunc(void *param) {
       // hbClearReqInfo(pAppHbMgr);
 
       atomic_add_fetch_32(&pAppHbMgr->reportCnt, 1);
-      taosArrayPush(mgr, &pAppHbMgr);
     }
 
-    taosArrayDestroy(clientHbMgr.appHbMgrs);
-    clientHbMgr.appHbMgrs = mgr;
-
     taosThreadMutexUnlock(&clientHbMgr.lock);
 
     taosMsleep(HEARTBEAT_INTERVAL);

source/client/src/clientTmq.c

@@ -147,8 +147,9 @@ typedef struct {
   int32_t       vgId;
   int32_t       vgStatus;
   int32_t       vgSkipCnt;            // here used to mark the slow vgroups
-  bool          receiveInfo;
+  bool          receivedInfoFromVnode;// has already received info from vnode
   int64_t       emptyBlockReceiveTs;  // once empty block is received, idle for ignoreCnt then start to poll data
+  bool          seekUpdated;          // offset is updated by seek operator, therefore, not update by vnode rsp.
   SEpSet        epSet;
 } SMqClientVg;
 
@@ -1429,6 +1430,9 @@ static void initClientTopicFromRsp(SMqClientTopic* pTopic, SMqSubTopicEp* pTopic
     clientVg.offsetInfo.committedOffset = offsetNew;
     clientVg.offsetInfo.walVerBegin = -1;
     clientVg.offsetInfo.walVerEnd = -1;
+    clientVg.seekUpdated = false;
+    clientVg.receivedInfoFromVnode = false;
+
     taosArrayPush(pTopic->vgs, &clientVg);
   }
 }
@@ -1704,6 +1708,7 @@ static int32_t doTmqPollImpl(tmq_t* pTmq, SMqClientTopic* pTopic, SMqClientVg* p
   asyncSendMsgToServer(pTmq->pTscObj->pAppInfo->pTransporter, &pVg->epSet, &transporterId, sendInfo);
 
   pVg->pollCnt++;
+  pVg->seekUpdated = false;   // reset this flag.
   pTmq->pollCnt++;
 
   return TSDB_CODE_SUCCESS;
@@ -1806,8 +1811,11 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
           pVg->epSet = *pollRspWrapper->pEpset;
         }
 
-        // update the local offset value only for the returned values.
+        // update the local offset value only for the returned values, only when the local offset is NOT updated
-        pVg->offsetInfo.currentOffset = pDataRsp->rspOffset;
+        // by tmq_offset_seek function
+        if (!pVg->seekUpdated) {
+          pVg->offsetInfo.currentOffset = pDataRsp->rspOffset;
+        }
 
         // update the status
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
@@ -1815,7 +1823,7 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
         // update the valid wal version range
         pVg->offsetInfo.walVerBegin = pDataRsp->head.walsver;
         pVg->offsetInfo.walVerEnd = pDataRsp->head.walever;
-        pVg->receiveInfo = true;
+        pVg->receivedInfoFromVnode = true;
 
         char buf[80];
         tFormatOffset(buf, 80, &pDataRsp->rspOffset);
@@ -1853,9 +1861,10 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
 
       if (pollRspWrapper->metaRsp.head.epoch == consumerEpoch) {
         SMqClientVg* pVg = pollRspWrapper->vgHandle;
-        if(pollRspWrapper->metaRsp.rspOffset.type != 0){    // if offset is validate
+        if (!pVg->seekUpdated) {
           pVg->offsetInfo.currentOffset = pollRspWrapper->metaRsp.rspOffset;
         }
+
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
         // build rsp
         SMqMetaRspObj* pRsp = tmqBuildMetaRspFromWrapper(pollRspWrapper);
@@ -1873,9 +1882,10 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
 
       if (pollRspWrapper->taosxRsp.head.epoch == consumerEpoch) {
         SMqClientVg* pVg = pollRspWrapper->vgHandle;
-        if(pollRspWrapper->taosxRsp.rspOffset.type != 0){    // if offset is validate
+        if (!pVg->seekUpdated) {  // if offset is validate
           pVg->offsetInfo.currentOffset = pollRspWrapper->taosxRsp.rspOffset;
         }
+
         atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
 
         if (pollRspWrapper->taosxRsp.blockNum == 0) {
@@ -2109,6 +2119,29 @@ int32_t tmq_get_vgroup_id(TAOS_RES* res) {
   }
 }
 
+int64_t tmq_get_vgroup_offset(TAOS_RES* res) {
+  if (TD_RES_TMQ(res)) {
+    SMqRspObj* pRspObj = (SMqRspObj*) res;
+    STqOffsetVal* pOffset = &pRspObj->rsp.rspOffset;
+    if (pOffset->type == TMQ_OFFSET__LOG) {
+      return pRspObj->rsp.rspOffset.version;
+    }
+  } else if (TD_RES_TMQ_META(res)) {
+    SMqMetaRspObj* pRspObj = (SMqMetaRspObj*)res;
+    if (pRspObj->metaRsp.rspOffset.type == TMQ_OFFSET__LOG) {
+      return pRspObj->metaRsp.rspOffset.version;
+    }
+  } else if (TD_RES_TMQ_METADATA(res)) {
+    SMqTaosxRspObj* pRspObj = (SMqTaosxRspObj*) res;
+    if (pRspObj->rsp.rspOffset.type == TMQ_OFFSET__LOG) {
+      return pRspObj->rsp.rspOffset.version;
+    }
+  }
+
+  // data from tsdb, no valid offset info
+  return -1;
+}
+
 const char* tmq_get_table_name(TAOS_RES* res) {
   if (TD_RES_TMQ(res)) {
     SMqRspObj* pRspObj = (SMqRspObj*)res;
@@ -2410,7 +2443,7 @@ int32_t tmq_get_topic_assignment(tmq_t* tmq, const char* pTopicName, tmq_topic_a
 
   for (int32_t j = 0; j < (*numOfAssignment); ++j) {
     SMqClientVg* pClientVg = taosArrayGet(pTopic->vgs, j);
-    if (!pClientVg->receiveInfo) {
+    if (!pClientVg->receivedInfoFromVnode) {
       needFetch = true;
       break;
     }
@@ -2600,6 +2633,7 @@ int32_t tmq_offset_seek(tmq_t* tmq, const char* pTopicName, int32_t vgId, int64_
   if (pOffsetInfo->currentOffset.type == TMQ_OFFSET__LOG) {
     pOffsetInfo->currentOffset.version = offset;
     pOffsetInfo->committedOffset.version = INT64_MIN;
+    pVg->seekUpdated = true;
   }
 
   SMqRspObj rspObj = {.resType = RES_TYPE__TMQ, .vgId = pVg->vgId};

source/common/src/tmsg.c

@@ -2805,7 +2805,26 @@ int32_t tSerializeSUseDbRsp(void *buf, int32_t bufLen, const SUseDbRsp *pRsp) {
   return tlen;
 }
 
-int32_t tSerializeSUseDbBatchRsp(void *buf, int32_t bufLen, SUseDbBatchRsp *pRsp) {
+int32_t tSerializeSDbHbRspImp(SEncoder *pEncoder, const SDbHbRsp *pRsp) {
+  if (pRsp->useDbRsp) {
+    if (tEncodeI8(pEncoder, 1) < 0) return -1;
+    if (tSerializeSUseDbRspImp(pEncoder, pRsp->useDbRsp) < 0) return -1;
+  } else {
+    if (tEncodeI8(pEncoder, 0) < 0) return -1;
+  }
+
+  if (pRsp->cfgRsp) {
+    if (tEncodeI8(pEncoder, 1) < 0) return -1;
+    if (tSerializeSDbCfgRspImpl(pEncoder, pRsp->cfgRsp) < 0) return -1;
+  } else {
+    if (tEncodeI8(pEncoder, 0) < 0) return -1;
+  }
+
+  return 0;
+}
+
+
+int32_t tSerializeSDbHbBatchRsp(void *buf, int32_t bufLen, SDbHbBatchRsp *pRsp) {
   SEncoder encoder = {0};
   tEncoderInit(&encoder, buf, bufLen);
 
@@ -2814,8 +2833,8 @@ int32_t tSerializeSUseDbBatchRsp(void *buf, int32_t bufLen, SUseDbBatchRsp *pRsp
   int32_t numOfBatch = taosArrayGetSize(pRsp->pArray);
   if (tEncodeI32(&encoder, numOfBatch) < 0) return -1;
   for (int32_t i = 0; i < numOfBatch; ++i) {
-    SUseDbRsp *pUsedbRsp = taosArrayGet(pRsp->pArray, i);
+    SDbHbRsp *pDbRsp = taosArrayGet(pRsp->pArray, i);
-    if (tSerializeSUseDbRspImp(&encoder, pUsedbRsp) < 0) return -1;
+    if (tSerializeSDbHbRspImp(&encoder, pDbRsp) < 0) return -1;
   }
   tEndEncode(&encoder);
 
@@ -2868,7 +2887,25 @@ int32_t tDeserializeSUseDbRsp(void *buf, int32_t bufLen, SUseDbRsp *pRsp) {
   return 0;
 }
 
-int32_t tDeserializeSUseDbBatchRsp(void *buf, int32_t bufLen, SUseDbBatchRsp *pRsp) {
+int32_t tDeserializeSDbHbRspImp(SDecoder* decoder, SDbHbRsp* pRsp) {
+  int8_t flag = 0;
+  if (tDecodeI8(decoder, &flag) < 0) return -1;
+  if (flag) {
+    pRsp->useDbRsp = taosMemoryCalloc(1, sizeof(SUseDbRsp));
+    if (NULL == pRsp->useDbRsp) return -1;
+    if (tDeserializeSUseDbRspImp(decoder, pRsp->useDbRsp) < 0) return -1;
+  }
+  if (tDecodeI8(decoder, &flag) < 0) return -1;
+  if (flag) {
+    pRsp->cfgRsp = taosMemoryCalloc(1, sizeof(SDbCfgRsp));
+    if (NULL == pRsp->cfgRsp) return -1;
+    if (tDeserializeSDbCfgRspImpl(decoder, pRsp->cfgRsp) < 0) return -1;
+  }
+
+  return 0;
+}
+
+int32_t tDeserializeSDbHbBatchRsp(void *buf, int32_t bufLen, SDbHbBatchRsp *pRsp) {
   SDecoder decoder = {0};
   tDecoderInit(&decoder, buf, bufLen);
 
@@ -2877,19 +2914,19 @@ int32_t tDeserializeSUseDbBatchRsp(void *buf, int32_t bufLen, SUseDbBatchRsp *pR
   int32_t numOfBatch = taosArrayGetSize(pRsp->pArray);
   if (tDecodeI32(&decoder, &numOfBatch) < 0) return -1;
 
-  pRsp->pArray = taosArrayInit(numOfBatch, sizeof(SUseDbRsp));
+  pRsp->pArray = taosArrayInit(numOfBatch, sizeof(SDbHbRsp));
   if (pRsp->pArray == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
   for (int32_t i = 0; i < numOfBatch; ++i) {
-    SUseDbRsp usedbRsp = {0};
+    SDbHbRsp rsp = {0};
-    if (tDeserializeSUseDbRspImp(&decoder, &usedbRsp) < 0) {
+    if (tDeserializeSDbHbRspImp(&decoder, &rsp) < 0) {
       tDecoderClear(&decoder);
       return -1;
     }
-    taosArrayPush(pRsp->pArray, &usedbRsp);
+    taosArrayPush(pRsp->pArray, &rsp);
   }
   tEndDecode(&decoder);
 
@@ -2899,11 +2936,27 @@ int32_t tDeserializeSUseDbBatchRsp(void *buf, int32_t bufLen, SUseDbBatchRsp *pR
 
 void tFreeSUsedbRsp(SUseDbRsp *pRsp) { taosArrayDestroy(pRsp->pVgroupInfos); }
 
-void tFreeSUseDbBatchRsp(SUseDbBatchRsp *pRsp) {
+void tFreeSDbHbRsp(SDbHbRsp *pDbRsp) {
+  if (NULL == pDbRsp) {
+    return;
+  }
+
+  if (pDbRsp->useDbRsp) {
+    tFreeSUsedbRsp(pDbRsp->useDbRsp);
+    taosMemoryFree(pDbRsp->useDbRsp);
+  }
+
+  if (pDbRsp->cfgRsp) {
+    tFreeSDbCfgRsp(pDbRsp->cfgRsp);
+    taosMemoryFree(pDbRsp->cfgRsp);
+  }
+}
+
+void tFreeSDbHbBatchRsp(SDbHbBatchRsp *pRsp) {
   int32_t numOfBatch = taosArrayGetSize(pRsp->pArray);
   for (int32_t i = 0; i < numOfBatch; ++i) {
-    SUseDbRsp *pUsedbRsp = taosArrayGet(pRsp->pArray, i);
+    SDbHbRsp *pDbRsp = taosArrayGet(pRsp->pArray, i);
-    tFreeSUsedbRsp(pUsedbRsp);
+    tFreeSDbHbRsp(pDbRsp);
   }
 
   taosArrayDestroy(pRsp->pArray);
@@ -3119,85 +3172,93 @@ int32_t tDeserializeSVDropTtlTableReq(void *buf, int32_t bufLen, SVDropTtlTableR
   return 0;
 }
 
+int32_t tSerializeSDbCfgRspImpl(SEncoder *encoder, const SDbCfgRsp *pRsp) {
+  if (tEncodeCStr(encoder, pRsp->db) < 0) return -1;
+  if (tEncodeI64(encoder, pRsp->dbId) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->cfgVersion) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->numOfVgroups) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->numOfStables) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->buffer) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->cacheSize) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->pageSize) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->pages) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->daysPerFile) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->daysToKeep0) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->daysToKeep1) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->daysToKeep2) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->minRows) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->maxRows) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->walFsyncPeriod) < 0) return -1;
+  if (tEncodeI16(encoder, pRsp->hashPrefix) < 0) return -1;
+  if (tEncodeI16(encoder, pRsp->hashSuffix) < 0) return -1;
+  if (tEncodeI8(encoder, pRsp->walLevel) < 0) return -1;
+  if (tEncodeI8(encoder, pRsp->precision) < 0) return -1;
+  if (tEncodeI8(encoder, pRsp->compression) < 0) return -1;
+  if (tEncodeI8(encoder, pRsp->replications) < 0) return -1;
+  if (tEncodeI8(encoder, pRsp->strict) < 0) return -1;
+  if (tEncodeI8(encoder, pRsp->cacheLast) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->tsdbPageSize) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->walRetentionPeriod) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->walRollPeriod) < 0) return -1;
+  if (tEncodeI64(encoder, pRsp->walRetentionSize) < 0) return -1;
+  if (tEncodeI64(encoder, pRsp->walSegmentSize) < 0) return -1;
+  if (tEncodeI32(encoder, pRsp->numOfRetensions) < 0) return -1;
+  for (int32_t i = 0; i < pRsp->numOfRetensions; ++i) {
+    SRetention *pRetension = taosArrayGet(pRsp->pRetensions, i);
+    if (tEncodeI64(encoder, pRetension->freq) < 0) return -1;
+    if (tEncodeI64(encoder, pRetension->keep) < 0) return -1;
+    if (tEncodeI8(encoder, pRetension->freqUnit) < 0) return -1;
+    if (tEncodeI8(encoder, pRetension->keepUnit) < 0) return -1;
+  }
+  if (tEncodeI8(encoder, pRsp->schemaless) < 0) return -1;
+  if (tEncodeI16(encoder, pRsp->sstTrigger) < 0) return -1;
+
+  return 0;
+}
+
 int32_t tSerializeSDbCfgRsp(void *buf, int32_t bufLen, const SDbCfgRsp *pRsp) {
   SEncoder encoder = {0};
   tEncoderInit(&encoder, buf, bufLen);
 
   if (tStartEncode(&encoder) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->numOfVgroups) < 0) return -1;
+  tSerializeSDbCfgRspImpl(&encoder, pRsp);
-  if (tEncodeI32(&encoder, pRsp->numOfStables) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->buffer) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->cacheSize) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->pageSize) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->pages) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->daysPerFile) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->daysToKeep0) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->daysToKeep1) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->daysToKeep2) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->minRows) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->maxRows) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->walFsyncPeriod) < 0) return -1;
-  if (tEncodeI16(&encoder, pRsp->hashPrefix) < 0) return -1;
-  if (tEncodeI16(&encoder, pRsp->hashSuffix) < 0) return -1;
-  if (tEncodeI8(&encoder, pRsp->walLevel) < 0) return -1;
-  if (tEncodeI8(&encoder, pRsp->precision) < 0) return -1;
-  if (tEncodeI8(&encoder, pRsp->compression) < 0) return -1;
-  if (tEncodeI8(&encoder, pRsp->replications) < 0) return -1;
-  if (tEncodeI8(&encoder, pRsp->strict) < 0) return -1;
-  if (tEncodeI8(&encoder, pRsp->cacheLast) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->tsdbPageSize) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->walRetentionPeriod) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->walRollPeriod) < 0) return -1;
-  if (tEncodeI64(&encoder, pRsp->walRetentionSize) < 0) return -1;
-  if (tEncodeI64(&encoder, pRsp->walSegmentSize) < 0) return -1;
-  if (tEncodeI32(&encoder, pRsp->numOfRetensions) < 0) return -1;
-  for (int32_t i = 0; i < pRsp->numOfRetensions; ++i) {
-    SRetention *pRetension = taosArrayGet(pRsp->pRetensions, i);
-    if (tEncodeI64(&encoder, pRetension->freq) < 0) return -1;
-    if (tEncodeI64(&encoder, pRetension->keep) < 0) return -1;
-    if (tEncodeI8(&encoder, pRetension->freqUnit) < 0) return -1;
-    if (tEncodeI8(&encoder, pRetension->keepUnit) < 0) return -1;
-  }
-  if (tEncodeI8(&encoder, pRsp->schemaless) < 0) return -1;
-  if (tEncodeI16(&encoder, pRsp->sstTrigger) < 0) return -1;
   tEndEncode(&encoder);
   int32_t tlen = encoder.pos;
   tEncoderClear(&encoder);
   return tlen;
 }
 
-int32_t tDeserializeSDbCfgRsp(void *buf, int32_t bufLen, SDbCfgRsp *pRsp) {
+int32_t tDeserializeSDbCfgRspImpl(SDecoder* decoder, SDbCfgRsp *pRsp) {
-  SDecoder decoder = {0};
+  if (tDecodeCStrTo(decoder, pRsp->db) < 0) return -1;
-  tDecoderInit(&decoder, buf, bufLen);
+  if (tDecodeI64(decoder, &pRsp->dbId) < 0) return -1;
-
+  if (tDecodeI32(decoder, &pRsp->cfgVersion) < 0) return -1;
-  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->numOfVgroups) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->numOfVgroups) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->numOfStables) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->numOfStables) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->buffer) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->buffer) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->cacheSize) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->cacheSize) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->pageSize) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->pageSize) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->pages) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->pages) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->daysPerFile) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->daysPerFile) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->daysToKeep0) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->daysToKeep0) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->daysToKeep1) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->daysToKeep1) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->daysToKeep2) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->daysToKeep2) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->minRows) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->minRows) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->maxRows) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->maxRows) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->walFsyncPeriod) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->walFsyncPeriod) < 0) return -1;
+  if (tDecodeI16(decoder, &pRsp->hashPrefix) < 0) return -1;
-  if (tDecodeI16(&decoder, &pRsp->hashPrefix) < 0) return -1;
+  if (tDecodeI16(decoder, &pRsp->hashSuffix) < 0) return -1;
-  if (tDecodeI16(&decoder, &pRsp->hashSuffix) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->walLevel) < 0) return -1;
-  if (tDecodeI8(&decoder, &pRsp->walLevel) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->precision) < 0) return -1;
-  if (tDecodeI8(&decoder, &pRsp->precision) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->compression) < 0) return -1;
-  if (tDecodeI8(&decoder, &pRsp->compression) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->replications) < 0) return -1;
-  if (tDecodeI8(&decoder, &pRsp->replications) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->strict) < 0) return -1;
-  if (tDecodeI8(&decoder, &pRsp->strict) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->cacheLast) < 0) return -1;
-  if (tDecodeI8(&decoder, &pRsp->cacheLast) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->tsdbPageSize) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->tsdbPageSize) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->walRetentionPeriod) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->walRetentionPeriod) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->walRollPeriod) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->walRollPeriod) < 0) return -1;
+  if (tDecodeI64(decoder, &pRsp->walRetentionSize) < 0) return -1;
-  if (tDecodeI64(&decoder, &pRsp->walRetentionSize) < 0) return -1;
+  if (tDecodeI64(decoder, &pRsp->walSegmentSize) < 0) return -1;
-  if (tDecodeI64(&decoder, &pRsp->walSegmentSize) < 0) return -1;
+  if (tDecodeI32(decoder, &pRsp->numOfRetensions) < 0) return -1;
-  if (tDecodeI32(&decoder, &pRsp->numOfRetensions) < 0) return -1;
   if (pRsp->numOfRetensions > 0) {
     pRsp->pRetensions = taosArrayInit(pRsp->numOfRetensions, sizeof(SRetention));
     if (pRsp->pRetensions == NULL) {
@@ -3208,23 +3269,41 @@ int32_t tDeserializeSDbCfgRsp(void *buf, int32_t bufLen, SDbCfgRsp *pRsp) {
 
   for (int32_t i = 0; i < pRsp->numOfRetensions; ++i) {
     SRetention rentension = {0};
-    if (tDecodeI64(&decoder, &rentension.freq) < 0) return -1;
+    if (tDecodeI64(decoder, &rentension.freq) < 0) return -1;
-    if (tDecodeI64(&decoder, &rentension.keep) < 0) return -1;
+    if (tDecodeI64(decoder, &rentension.keep) < 0) return -1;
-    if (tDecodeI8(&decoder, &rentension.freqUnit) < 0) return -1;
+    if (tDecodeI8(decoder, &rentension.freqUnit) < 0) return -1;
-    if (tDecodeI8(&decoder, &rentension.keepUnit) < 0) return -1;
+    if (tDecodeI8(decoder, &rentension.keepUnit) < 0) return -1;
     if (taosArrayPush(pRsp->pRetensions, &rentension) == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
     }
   }
-  if (tDecodeI8(&decoder, &pRsp->schemaless) < 0) return -1;
+  if (tDecodeI8(decoder, &pRsp->schemaless) < 0) return -1;
-  if (tDecodeI16(&decoder, &pRsp->sstTrigger) < 0) return -1;
+  if (tDecodeI16(decoder, &pRsp->sstTrigger) < 0) return -1;
+
+  return 0;
+}
+
+int32_t tDeserializeSDbCfgRsp(void *buf, int32_t bufLen, SDbCfgRsp *pRsp) {
+  SDecoder decoder = {0};
+  tDecoderInit(&decoder, buf, bufLen);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDeserializeSDbCfgRspImpl(&decoder, pRsp) < 0) return -1;
   tEndDecode(&decoder);
 
   tDecoderClear(&decoder);
   return 0;
 }
 
+void tFreeSDbCfgRsp(SDbCfgRsp *pRsp) {
+  if (NULL == pRsp) {
+    return;
+  }
+
+  taosArrayDestroy(pRsp->pRetensions);
+}
+
 int32_t tSerializeSUserIndexReq(void *buf, int32_t bufLen, SUserIndexReq *pReq) {
   SEncoder encoder = {0};
   tEncoderInit(&encoder, buf, bufLen);

source/dnode/mgmt/exe/dmMain.c

@@ -87,6 +87,18 @@ static void dmStopDnode(int signum, void *sigInfo, void *context) {
 }
 
 void dmLogCrash(int signum, void *sigInfo, void *context) {
+  // taosIgnSignal(SIGTERM);
+  // taosIgnSignal(SIGHUP);
+  // taosIgnSignal(SIGINT);
+  // taosIgnSignal(SIGBREAK);
+
+#ifndef WINDOWS
+  taosIgnSignal(SIGBUS);
+#endif
+  taosIgnSignal(SIGABRT);
+  taosIgnSignal(SIGFPE);
+  taosIgnSignal(SIGSEGV);
+
   char       *pMsg = NULL;
   const char *flags = "UTL FATAL ";
   ELogLevel   level = DEBUG_FATAL;

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

@@ -91,6 +91,7 @@ static void *dmCrashReportThreadFp(void *param) {
         dError("failed to send crash report");
         if (pFile) {
           taosReleaseCrashLogFile(pFile, false);
+          pFile = NULL;
           continue;
         }
       } else {
@@ -110,6 +111,7 @@ static void *dmCrashReportThreadFp(void *param) {
     
     if (pFile) {
       taosReleaseCrashLogFile(pFile, truncateFile);
+      pFile = NULL;
       truncateFile = false;
     }
     

source/dnode/mgmt/mgmt_mnode/src/mmWorker.c

@@ -120,6 +120,11 @@ int32_t mmPutMsgToReadQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) {
 }
 
 int32_t mmPutMsgToQueryQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) {
+  if (NULL == pMgmt->pMnode) {
+    const STraceId *trace = &pMsg->info.traceId;
+    dGError("msg:%p, stop to pre-process in mnode since mnode is NULL, type:%s", pMsg, TMSG_INFO(pMsg->msgType));
+    return -1;
+  }
   pMsg->info.node = pMgmt->pMnode;
   if (mndPreProcessQueryMsg(pMsg) != 0) {
     const STraceId *trace = &pMsg->info.traceId;

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

@@ -26,7 +26,7 @@ int32_t mndInitDb(SMnode *pMnode);
 void    mndCleanupDb(SMnode *pMnode);
 SDbObj *mndAcquireDb(SMnode *pMnode, const char *db);
 void    mndReleaseDb(SMnode *pMnode, SDbObj *pDb);
-int32_t mndValidateDbInfo(SMnode *pMnode, SDbVgVersion *pDbs, int32_t numOfDbs, void **ppRsp, int32_t *pRspLen);
+int32_t mndValidateDbInfo(SMnode *pMnode, SDbCacheInfo *pDbs, int32_t numOfDbs, void **ppRsp, int32_t *pRspLen);
 int32_t mndExtractDbInfo(SMnode *pMnode, SDbObj *pDb, SUseDbRsp *pRsp, const SUseDbReq *pReq);
 bool    mndIsDbReady(SMnode *pMnode, SDbObj *pDb);
 

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

@@ -898,6 +898,42 @@ _OVER:
   return code;
 }
 
+static void mndDumpDbCfgInfo(SDbCfgRsp *cfgRsp, SDbObj *pDb) {
+  strcpy(cfgRsp->db, pDb->name);
+  cfgRsp->dbId = pDb->uid;
+  cfgRsp->cfgVersion = pDb->cfgVersion;
+  cfgRsp->numOfVgroups = pDb->cfg.numOfVgroups;
+  cfgRsp->numOfStables = pDb->cfg.numOfStables;
+  cfgRsp->buffer = pDb->cfg.buffer;
+  cfgRsp->cacheSize = pDb->cfg.cacheLastSize;
+  cfgRsp->pageSize = pDb->cfg.pageSize;
+  cfgRsp->pages = pDb->cfg.pages;
+  cfgRsp->daysPerFile = pDb->cfg.daysPerFile;
+  cfgRsp->daysToKeep0 = pDb->cfg.daysToKeep0;
+  cfgRsp->daysToKeep1 = pDb->cfg.daysToKeep1;
+  cfgRsp->daysToKeep2 = pDb->cfg.daysToKeep2;
+  cfgRsp->minRows = pDb->cfg.minRows;
+  cfgRsp->maxRows = pDb->cfg.maxRows;
+  cfgRsp->walFsyncPeriod = pDb->cfg.walFsyncPeriod;
+  cfgRsp->hashPrefix = pDb->cfg.hashPrefix;
+  cfgRsp->hashSuffix = pDb->cfg.hashSuffix;
+  cfgRsp->walLevel = pDb->cfg.walLevel;
+  cfgRsp->precision = pDb->cfg.precision;
+  cfgRsp->compression = pDb->cfg.compression;
+  cfgRsp->replications = pDb->cfg.replications;
+  cfgRsp->strict = pDb->cfg.strict;
+  cfgRsp->cacheLast = pDb->cfg.cacheLast;
+  cfgRsp->tsdbPageSize = pDb->cfg.tsdbPageSize;
+  cfgRsp->walRetentionPeriod = pDb->cfg.walRetentionPeriod;
+  cfgRsp->walRollPeriod = pDb->cfg.walRollPeriod;
+  cfgRsp->walRetentionSize = pDb->cfg.walRetentionSize;
+  cfgRsp->walSegmentSize = pDb->cfg.walSegmentSize;
+  cfgRsp->numOfRetensions = pDb->cfg.numOfRetensions;
+  cfgRsp->pRetensions = pDb->cfg.pRetensions;
+  cfgRsp->schemaless = pDb->cfg.schemaless;
+  cfgRsp->sstTrigger = pDb->cfg.sstTrigger;
+}
+
 static int32_t mndProcessGetDbCfgReq(SRpcMsg *pReq) {
   SMnode   *pMnode = pReq->info.node;
   int32_t   code = -1;
@@ -910,41 +946,15 @@ static int32_t mndProcessGetDbCfgReq(SRpcMsg *pReq) {
     goto _OVER;
   }
 
-  pDb = mndAcquireDb(pMnode, cfgReq.db);
+  if (strcasecmp(cfgReq.db, TSDB_INFORMATION_SCHEMA_DB) && strcasecmp(cfgReq.db, TSDB_PERFORMANCE_SCHEMA_DB)) {
-  if (pDb == NULL) {
+    pDb = mndAcquireDb(pMnode, cfgReq.db);
-    goto _OVER;
+    if (pDb == NULL) {
-  }
+      goto _OVER;
+    }
 
-  cfgRsp.numOfVgroups = pDb->cfg.numOfVgroups;
+    mndDumpDbCfgInfo(&cfgRsp, pDb);
-  cfgRsp.numOfStables = pDb->cfg.numOfStables;
+  }
-  cfgRsp.buffer = pDb->cfg.buffer;
+  
-  cfgRsp.cacheSize = pDb->cfg.cacheLastSize;
-  cfgRsp.pageSize = pDb->cfg.pageSize;
-  cfgRsp.pages = pDb->cfg.pages;
-  cfgRsp.daysPerFile = pDb->cfg.daysPerFile;
-  cfgRsp.daysToKeep0 = pDb->cfg.daysToKeep0;
-  cfgRsp.daysToKeep1 = pDb->cfg.daysToKeep1;
-  cfgRsp.daysToKeep2 = pDb->cfg.daysToKeep2;
-  cfgRsp.minRows = pDb->cfg.minRows;
-  cfgRsp.maxRows = pDb->cfg.maxRows;
-  cfgRsp.walFsyncPeriod = pDb->cfg.walFsyncPeriod;
-  cfgRsp.hashPrefix = pDb->cfg.hashPrefix;
-  cfgRsp.hashSuffix = pDb->cfg.hashSuffix;
-  cfgRsp.walLevel = pDb->cfg.walLevel;
-  cfgRsp.precision = pDb->cfg.precision;
-  cfgRsp.compression = pDb->cfg.compression;
-  cfgRsp.replications = pDb->cfg.replications;
-  cfgRsp.strict = pDb->cfg.strict;
-  cfgRsp.cacheLast = pDb->cfg.cacheLast;
-  cfgRsp.tsdbPageSize = pDb->cfg.tsdbPageSize;
-  cfgRsp.walRetentionPeriod = pDb->cfg.walRetentionPeriod;
-  cfgRsp.walRollPeriod = pDb->cfg.walRollPeriod;
-  cfgRsp.walRetentionSize = pDb->cfg.walRetentionSize;
-  cfgRsp.walSegmentSize = pDb->cfg.walSegmentSize;
-  cfgRsp.numOfRetensions = pDb->cfg.numOfRetensions;
-  cfgRsp.pRetensions = pDb->cfg.pRetensions;
-  cfgRsp.schemaless = pDb->cfg.schemaless;
-  cfgRsp.sstTrigger = pDb->cfg.sstTrigger;
   int32_t contLen = tSerializeSDbCfgRsp(NULL, 0, &cfgRsp);
   void   *pRsp = rpcMallocCont(contLen);
   if (pRsp == NULL) {
@@ -1341,103 +1351,119 @@ _OVER:
   return code;
 }
 
-int32_t mndValidateDbInfo(SMnode *pMnode, SDbVgVersion *pDbs, int32_t numOfDbs, void **ppRsp, int32_t *pRspLen) {
+int32_t mndValidateDbInfo(SMnode *pMnode, SDbCacheInfo *pDbs, int32_t numOfDbs, void **ppRsp, int32_t *pRspLen) {
-  SUseDbBatchRsp batchUseRsp = {0};
+  SDbHbBatchRsp batchRsp = {0};
-  batchUseRsp.pArray = taosArrayInit(numOfDbs, sizeof(SUseDbRsp));
+  batchRsp.pArray = taosArrayInit(numOfDbs, sizeof(SDbHbRsp));
-  if (batchUseRsp.pArray == NULL) {
+  if (batchRsp.pArray == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
   for (int32_t i = 0; i < numOfDbs; ++i) {
-    SDbVgVersion *pDbVgVersion = &pDbs[i];
+    SDbCacheInfo *pDbCacheInfo = &pDbs[i];
-    pDbVgVersion->dbId = be64toh(pDbVgVersion->dbId);
+    pDbCacheInfo->dbId = be64toh(pDbCacheInfo->dbId);
-    pDbVgVersion->vgVersion = htonl(pDbVgVersion->vgVersion);
+    pDbCacheInfo->vgVersion = htonl(pDbCacheInfo->vgVersion);
-    pDbVgVersion->numOfTable = htonl(pDbVgVersion->numOfTable);
+    pDbCacheInfo->cfgVersion = htonl(pDbCacheInfo->cfgVersion);
-    pDbVgVersion->stateTs = be64toh(pDbVgVersion->stateTs);
+    pDbCacheInfo->numOfTable = htonl(pDbCacheInfo->numOfTable);
+    pDbCacheInfo->stateTs = be64toh(pDbCacheInfo->stateTs);
 
-    SUseDbRsp usedbRsp = {0};
+    SDbHbRsp rsp = {0};
 
-    if ((0 == strcasecmp(pDbVgVersion->dbFName, TSDB_INFORMATION_SCHEMA_DB) ||
+    if ((0 == strcasecmp(pDbCacheInfo->dbFName, TSDB_INFORMATION_SCHEMA_DB) ||
-         (0 == strcasecmp(pDbVgVersion->dbFName, TSDB_PERFORMANCE_SCHEMA_DB)))) {
+         (0 == strcasecmp(pDbCacheInfo->dbFName, TSDB_PERFORMANCE_SCHEMA_DB)))) {
-      memcpy(usedbRsp.db, pDbVgVersion->dbFName, TSDB_DB_FNAME_LEN);
       int32_t vgVersion = mndGetGlobalVgroupVersion(pMnode);
-      if (pDbVgVersion->vgVersion < vgVersion) {
+      if (pDbCacheInfo->vgVersion >= vgVersion) {
-        usedbRsp.pVgroupInfos = taosArrayInit(10, sizeof(SVgroupInfo));
+        continue;
-
-        mndBuildDBVgroupInfo(NULL, pMnode, usedbRsp.pVgroupInfos);
-        usedbRsp.vgVersion = vgVersion++;
-      } else {
-        usedbRsp.vgVersion = pDbVgVersion->vgVersion;
       }
-      usedbRsp.vgNum = taosArrayGetSize(usedbRsp.pVgroupInfos);
+      
+      rsp.useDbRsp = taosMemoryCalloc(1, sizeof(SUseDbRsp));
+      memcpy(rsp.useDbRsp->db, pDbCacheInfo->dbFName, TSDB_DB_FNAME_LEN);
+      rsp.useDbRsp->pVgroupInfos = taosArrayInit(10, sizeof(SVgroupInfo));
 
-      taosArrayPush(batchUseRsp.pArray, &usedbRsp);
+      mndBuildDBVgroupInfo(NULL, pMnode, rsp.useDbRsp->pVgroupInfos);
+      rsp.useDbRsp->vgVersion = vgVersion++;
+      
+      rsp.useDbRsp->vgNum = taosArrayGetSize(rsp.useDbRsp->pVgroupInfos);
+
+      taosArrayPush(batchRsp.pArray, &rsp);
 
       continue;
     }
 
-    SDbObj *pDb = mndAcquireDb(pMnode, pDbVgVersion->dbFName);
+    SDbObj *pDb = mndAcquireDb(pMnode, pDbCacheInfo->dbFName);
     if (pDb == NULL) {
-      mTrace("db:%s, no exist", pDbVgVersion->dbFName);
+      mTrace("db:%s, no exist", pDbCacheInfo->dbFName);
-      memcpy(usedbRsp.db, pDbVgVersion->dbFName, TSDB_DB_FNAME_LEN);
+      rsp.useDbRsp = taosMemoryCalloc(1, sizeof(SUseDbRsp));
-      usedbRsp.uid = pDbVgVersion->dbId;
+      memcpy(rsp.useDbRsp->db, pDbCacheInfo->dbFName, TSDB_DB_FNAME_LEN);
-      usedbRsp.vgVersion = -1;
+      rsp.useDbRsp->uid = pDbCacheInfo->dbId;
-      taosArrayPush(batchUseRsp.pArray, &usedbRsp);
+      rsp.useDbRsp->vgVersion = -1;
+      taosArrayPush(batchRsp.pArray, &rsp);
       continue;
     }
 
     int32_t numOfTable = mndGetDBTableNum(pDb, pMnode);
 
-    if (pDbVgVersion->vgVersion >= pDb->vgVersion && numOfTable == pDbVgVersion->numOfTable &&
+    if (pDbCacheInfo->vgVersion >= pDb->vgVersion && 
-        pDbVgVersion->stateTs == pDb->stateTs) {
+        pDbCacheInfo->cfgVersion >= pDb->cfgVersion &&
-      mTrace("db:%s, valid dbinfo, vgVersion:%d stateTs:%" PRId64
+        numOfTable == pDbCacheInfo->numOfTable &&
-             " numOfTables:%d, not changed vgVersion:%d stateTs:%" PRId64 " numOfTables:%d",
+        pDbCacheInfo->stateTs == pDb->stateTs) {
-             pDbVgVersion->dbFName, pDbVgVersion->vgVersion, pDbVgVersion->stateTs, pDbVgVersion->numOfTable,
+      mTrace("db:%s, valid dbinfo, vgVersion:%d cfgVersion:%d stateTs:%" PRId64
-             pDb->vgVersion, pDb->stateTs, numOfTable);
+             " numOfTables:%d, not changed vgVersion:%d cfgVersion:%d stateTs:%" PRId64 " numOfTables:%d",
+             pDbCacheInfo->dbFName, pDbCacheInfo->vgVersion, pDbCacheInfo->cfgVersion, pDbCacheInfo->stateTs, pDbCacheInfo->numOfTable,
+             pDb->vgVersion, pDb->cfgVersion, pDb->stateTs, numOfTable);
       mndReleaseDb(pMnode, pDb);
       continue;
     } else {
-      mInfo("db:%s, valid dbinfo, vgVersion:%d stateTs:%" PRId64
+      mInfo("db:%s, valid dbinfo, vgVersion:%d cfgVersion:%d stateTs:%" PRId64
-            " numOfTables:%d, changed to vgVersion:%d stateTs:%" PRId64 " numOfTables:%d",
+            " numOfTables:%d, changed to vgVersion:%d cfgVersion:%d stateTs:%" PRId64 " numOfTables:%d",
-            pDbVgVersion->dbFName, pDbVgVersion->vgVersion, pDbVgVersion->stateTs, pDbVgVersion->numOfTable,
+            pDbCacheInfo->dbFName, pDbCacheInfo->vgVersion, pDbCacheInfo->cfgVersion, pDbCacheInfo->stateTs, pDbCacheInfo->numOfTable,
-            pDb->vgVersion, pDb->stateTs, numOfTable);
+            pDb->vgVersion, pDb->cfgVersion, pDb->stateTs, numOfTable);
     }
 
-    usedbRsp.pVgroupInfos = taosArrayInit(pDb->cfg.numOfVgroups, sizeof(SVgroupInfo));
+    if (pDbCacheInfo->cfgVersion < pDb->cfgVersion) {
-    if (usedbRsp.pVgroupInfos == NULL) {
+      rsp.cfgRsp = taosMemoryCalloc(1, sizeof(SDbCfgRsp));
-      mndReleaseDb(pMnode, pDb);
+      mndDumpDbCfgInfo(rsp.cfgRsp, pDb);
-      mError("db:%s, failed to malloc usedb response", pDb->name);
-      continue;
     }
 
-    mndBuildDBVgroupInfo(pDb, pMnode, usedbRsp.pVgroupInfos);
+    if (pDbCacheInfo->vgVersion < pDb->vgVersion || 
-    memcpy(usedbRsp.db, pDb->name, TSDB_DB_FNAME_LEN);
+        numOfTable != pDbCacheInfo->numOfTable ||
-    usedbRsp.uid = pDb->uid;
+        pDbCacheInfo->stateTs != pDb->stateTs) {
-    usedbRsp.vgVersion = pDb->vgVersion;
+      rsp.useDbRsp = taosMemoryCalloc(1, sizeof(SUseDbRsp));
-    usedbRsp.stateTs = pDb->stateTs;
+      rsp.useDbRsp->pVgroupInfos = taosArrayInit(pDb->cfg.numOfVgroups, sizeof(SVgroupInfo));
-    usedbRsp.vgNum = (int32_t)taosArrayGetSize(usedbRsp.pVgroupInfos);
+      if (rsp.useDbRsp->pVgroupInfos == NULL) {
-    usedbRsp.hashMethod = pDb->cfg.hashMethod;
+        mndReleaseDb(pMnode, pDb);
-    usedbRsp.hashPrefix = pDb->cfg.hashPrefix;
+        mError("db:%s, failed to malloc usedb response", pDb->name);
-    usedbRsp.hashSuffix = pDb->cfg.hashSuffix;
+        continue;
+      }
 
-    taosArrayPush(batchUseRsp.pArray, &usedbRsp);
+      mndBuildDBVgroupInfo(pDb, pMnode, rsp.useDbRsp->pVgroupInfos);
+      memcpy(rsp.useDbRsp->db, pDb->name, TSDB_DB_FNAME_LEN);
+      rsp.useDbRsp->uid = pDb->uid;
+      rsp.useDbRsp->vgVersion = pDb->vgVersion;
+      rsp.useDbRsp->stateTs = pDb->stateTs;
+      rsp.useDbRsp->vgNum = (int32_t)taosArrayGetSize(rsp.useDbRsp->pVgroupInfos);
+      rsp.useDbRsp->hashMethod = pDb->cfg.hashMethod;
+      rsp.useDbRsp->hashPrefix = pDb->cfg.hashPrefix;
+      rsp.useDbRsp->hashSuffix = pDb->cfg.hashSuffix;
+    }
+
+    taosArrayPush(batchRsp.pArray, &rsp);
     mndReleaseDb(pMnode, pDb);
   }
 
-  int32_t rspLen = tSerializeSUseDbBatchRsp(NULL, 0, &batchUseRsp);
+  int32_t rspLen = tSerializeSDbHbBatchRsp(NULL, 0, &batchRsp);
   void   *pRsp = taosMemoryMalloc(rspLen);
   if (pRsp == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
-    tFreeSUseDbBatchRsp(&batchUseRsp);
+    tFreeSDbHbBatchRsp(&batchRsp);
     return -1;
   }
-  tSerializeSUseDbBatchRsp(pRsp, rspLen, &batchUseRsp);
+  tSerializeSDbHbBatchRsp(pRsp, rspLen, &batchRsp);
 
   *ppRsp = pRsp;
   *pRspLen = rspLen;
 
-  tFreeSUseDbBatchRsp(&batchUseRsp);
+  tFreeSDbHbBatchRsp(&batchRsp);
   return 0;
 }
 

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

@@ -534,7 +534,7 @@ static int32_t mndProcessQueryHeartBeat(SMnode *pMnode, SRpcMsg *pMsg, SClientHb
       case HEARTBEAT_KEY_DBINFO: {
         void   *rspMsg = NULL;
         int32_t rspLen = 0;
-        mndValidateDbInfo(pMnode, kv->value, kv->valueLen / sizeof(SDbVgVersion), &rspMsg, &rspLen);
+        mndValidateDbInfo(pMnode, kv->value, kv->valueLen / sizeof(SDbCacheInfo), &rspMsg, &rspLen);
         if (rspMsg && rspLen > 0) {
           SKv kv1 = {.key = HEARTBEAT_KEY_DBINFO, .valueLen = rspLen, .value = rspMsg};
           taosArrayPush(hbRsp.info, &kv1);

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

@@ -2742,7 +2742,7 @@ void mndExtractTbNameFromStbFullName(const char *stbFullName, char *dst, int32_t
 //      varDataSetLen(stbName, strlen(&stbName[VARSTR_HEADER_SIZE]));
 //
 //      SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)stbName, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)stbName, false);
 //
 //      char db[TSDB_DB_NAME_LEN + VARSTR_HEADER_SIZE] = {0};
 //      tNameFromString(&name, pStb->db, T_NAME_ACCT | T_NAME_DB);
@@ -2750,29 +2750,29 @@ void mndExtractTbNameFromStbFullName(const char *stbFullName, char *dst, int32_t
 //      varDataSetLen(db, strlen(varDataVal(db)));
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)db, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)db, false);
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)&pStb->createdTime, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)&pStb->createdTime, false);
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)&pStb->numOfColumns, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)&pStb->numOfColumns, false);
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)&pStb->numOfTags, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)&pStb->numOfTags, false);
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)&pStb->updateTime, false);  // number of tables
+//      colDataSetVal(pColInfo, numOfRows, (const char *)&pStb->updateTime, false);  // number of tables
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
 //      if (pStb->commentLen > 0) {
 //        char comment[TSDB_TB_COMMENT_LEN + VARSTR_HEADER_SIZE] = {0};
 //        STR_TO_VARSTR(comment, pStb->comment);
-//        colDataAppend(pColInfo, numOfRows, comment, false);
+//        colDataSetVal(pColInfo, numOfRows, comment, false);
 //      } else if (pStb->commentLen == 0) {
 //        char comment[VARSTR_HEADER_SIZE + VARSTR_HEADER_SIZE] = {0};
 //        STR_TO_VARSTR(comment, "");
-//        colDataAppend(pColInfo, numOfRows, comment, false);
+//        colDataSetVal(pColInfo, numOfRows, comment, false);
 //      } else {
 //        colDataSetNULL(pColInfo, numOfRows);
 //      }
@@ -2782,14 +2782,14 @@ void mndExtractTbNameFromStbFullName(const char *stbFullName, char *dst, int32_t
 //      varDataSetLen(watermark, strlen(varDataVal(watermark)));
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)watermark, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)watermark, false);
 //
 //      char maxDelay[64 + VARSTR_HEADER_SIZE] = {0};
 //      sprintf(varDataVal(maxDelay), "%" PRId64 "a,%" PRId64 "a", pStb->maxdelay[0], pStb->maxdelay[1]);
 //      varDataSetLen(maxDelay, strlen(varDataVal(maxDelay)));
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)maxDelay, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)maxDelay, false);
 //
 //      char    rollup[160 + VARSTR_HEADER_SIZE] = {0};
 //      int32_t rollupNum = (int32_t)taosArrayGetSize(pStb->pFuncs);
@@ -2808,7 +2808,7 @@ void mndExtractTbNameFromStbFullName(const char *stbFullName, char *dst, int32_t
 //      varDataSetLen(rollup, strlen(varDataVal(rollup)));
 //
 //      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//      colDataAppend(pColInfo, numOfRows, (const char *)rollup, false);
+//      colDataSetVal(pColInfo, numOfRows, (const char *)rollup, false);
 //
 //      numOfRows++;
 //      sdbRelease(pSdb, pStb);
@@ -3067,20 +3067,20 @@ static int32_t buildDbColsInfoBlock(const SSDataBlock *p, const SSysTableMeta *p
     for (int32_t j = 0; j < pm->colNum; j++) {
       // table name
       SColumnInfoData *pColInfoData = taosArrayGet(p->pDataBlock, 0);
-      colDataAppend(pColInfoData, numOfRows, tName, false);
+      colDataSetVal(pColInfoData, numOfRows, tName, false);
 
       // database name
       pColInfoData = taosArrayGet(p->pDataBlock, 1);
-      colDataAppend(pColInfoData, numOfRows, dName, false);
+      colDataSetVal(pColInfoData, numOfRows, dName, false);
 
       pColInfoData = taosArrayGet(p->pDataBlock, 2);
-      colDataAppend(pColInfoData, numOfRows, typeName, false);
+      colDataSetVal(pColInfoData, numOfRows, typeName, false);
 
       // col name
       char colName[TSDB_COL_NAME_LEN + VARSTR_HEADER_SIZE] = {0};
       STR_TO_VARSTR(colName, pm->schema[j].name);
       pColInfoData = taosArrayGet(p->pDataBlock, 3);
-      colDataAppend(pColInfoData, numOfRows, colName, false);
+      colDataSetVal(pColInfoData, numOfRows, colName, false);
 
       // col type
       int8_t colType = pm->schema[j].type;
@@ -3095,10 +3095,10 @@ static int32_t buildDbColsInfoBlock(const SSDataBlock *p, const SSysTableMeta *p
                               (int32_t)((pm->schema[j].bytes - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE));
       }
       varDataSetLen(colTypeStr, colTypeLen);
-      colDataAppend(pColInfoData, numOfRows, (char *)colTypeStr, false);
+      colDataSetVal(pColInfoData, numOfRows, (char *)colTypeStr, false);
 
       pColInfoData = taosArrayGet(p->pDataBlock, 5);
-      colDataAppend(pColInfoData, numOfRows, (const char *)&pm->schema[j].bytes, false);
+      colDataSetVal(pColInfoData, numOfRows, (const char *)&pm->schema[j].bytes, false);
       for (int32_t k = 6; k <= 8; ++k) {
         pColInfoData = taosArrayGet(p->pDataBlock, k);
         colDataSetNULL(pColInfoData, numOfRows);
@@ -3192,19 +3192,19 @@ static int32_t mndRetrieveStbCol(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pB
     for (int i = 0; i < pStb->numOfColumns; i++) {
       int32_t          cols = 0;
       SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-      colDataAppend(pColInfo, numOfRows, (const char *)stbName, false);
+      colDataSetVal(pColInfo, numOfRows, (const char *)stbName, false);
 
       pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-      colDataAppend(pColInfo, numOfRows, (const char *)db, false);
+      colDataSetVal(pColInfo, numOfRows, (const char *)db, false);
 
       pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-      colDataAppend(pColInfo, numOfRows, typeName, false);
+      colDataSetVal(pColInfo, numOfRows, typeName, false);
 
       // col name
       char colName[TSDB_COL_NAME_LEN + VARSTR_HEADER_SIZE] = {0};
       STR_TO_VARSTR(colName, pStb->pColumns[i].name);
       pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-      colDataAppend(pColInfo, numOfRows, colName, false);
+      colDataSetVal(pColInfo, numOfRows, colName, false);
 
       // col type
       int8_t colType = pStb->pColumns[i].type;
@@ -3219,10 +3219,10 @@ static int32_t mndRetrieveStbCol(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pB
                               (int32_t)((pStb->pColumns[i].bytes - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE));
       }
       varDataSetLen(colTypeStr, colTypeLen);
-      colDataAppend(pColInfo, numOfRows, (char *)colTypeStr, false);
+      colDataSetVal(pColInfo, numOfRows, (char *)colTypeStr, false);
 
       pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-      colDataAppend(pColInfo, numOfRows, (const char *)&pStb->pColumns[i].bytes, false);
+      colDataSetVal(pColInfo, numOfRows, (const char *)&pStb->pColumns[i].bytes, false);
       while (cols < pShow->numOfColumns) {
         pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
         colDataSetNULL(pColInfo, numOfRows);

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

@@ -1306,7 +1306,7 @@ int32_t mndPauseAllStreamTasks(STrans *pTrans, SStreamObj *pStream) {
     int32_t sz = taosArrayGetSize(pTasks);
     for (int32_t j = 0; j < sz; j++) {
       SStreamTask *pTask = taosArrayGetP(pTasks, j);
-      if (mndPauseStreamTask(pTrans, pTask) < 0) {
+      if (pTask->taskLevel == TASK_LEVEL__SOURCE && mndPauseStreamTask(pTrans, pTask) < 0) {
         return -1;
       }
     }
@@ -1430,7 +1430,7 @@ int32_t mndResumeAllStreamTasks(STrans *pTrans, SStreamObj *pStream, int8_t igUn
     int32_t sz = taosArrayGetSize(pTasks);
     for (int32_t j = 0; j < sz; j++) {
       SStreamTask *pTask = taosArrayGetP(pTasks, j);
-      if (mndResumeStreamTask(pTrans, pTask, igUntreated) < 0) {
+      if (pTask->taskLevel == TASK_LEVEL__SOURCE && mndResumeStreamTask(pTrans, pTask, igUntreated) < 0) {
         return -1;
       }
     }

source/dnode/vnode/inc/vnode.h

@@ -125,6 +125,11 @@ int32_t  metaUidFilterCachePut(SMeta *pMeta, uint64_t suid, const void *pKey, in
                                int32_t payloadLen, double selectivityRatio);
 int32_t  metaUidCacheClear(SMeta *pMeta, uint64_t suid);
 tb_uid_t metaGetTableEntryUidByName(SMeta *pMeta, const char *name);
+int32_t  metaTbGroupCacheClear(SMeta* pMeta, uint64_t suid);
+int32_t  metaGetCachedTbGroup(SMeta* pMeta, tb_uid_t suid, const uint8_t* pKey, int32_t keyLen, SArray** pList);
+int32_t  metaPutTbGroupToCache(SMeta* pMeta, uint64_t suid, const void* pKey, int32_t keyLen, void* pPayload,
+                              int32_t payloadLen);
+
 int64_t  metaGetTbNum(SMeta *pMeta);
 int64_t  metaGetNtbNum(SMeta *pMeta);
 typedef struct {
@@ -177,9 +182,9 @@ typedef struct STsdbReader STsdbReader;
 #define CACHESCAN_RETRIEVE_LAST_ROW    0x4
 #define CACHESCAN_RETRIEVE_LAST        0x8
 
-int32_t tsdbSetTableList(STsdbReader *pReader, const void *pTableList, int32_t num);
+int32_t      tsdbReaderOpen(SVnode *pVnode, SQueryTableDataCond *pCond, void *pTableList, int32_t numOfTables,
-int32_t tsdbReaderOpen(SVnode *pVnode, SQueryTableDataCond *pCond, void *pTableList, int32_t numOfTables,
+                            SSDataBlock *pResBlock, STsdbReader **ppReader, const char *idstr, bool countOnly, SHashObj** pIgnoreTables);
-                       SSDataBlock *pResBlock, STsdbReader **ppReader, const char *idstr, bool countOnly);
+int32_t      tsdbSetTableList(STsdbReader *pReader, const void *pTableList, int32_t num);
 
 void         tsdbReaderSetId(STsdbReader *pReader, const char *idstr);
 void         tsdbReaderClose(STsdbReader *pReader);
@@ -197,6 +202,7 @@ int32_t      tsdbSetTableList(STsdbReader *pReader, const void *pTableList, int3
 void         tsdbReaderSetId(STsdbReader *pReader, const char *idstr);
 void         tsdbReaderSetCloseFlag(STsdbReader *pReader);
 
+int32_t tsdbReuseCacherowsReader(void* pReader, void* pTableIdList, int32_t numOfTables);
 int32_t tsdbCacherowsReaderOpen(void *pVnode, int32_t type, void *pTableIdList, int32_t numOfTables, int32_t numOfCols,
                                 SArray *pCidList, int32_t *pSlotIds, uint64_t suid, void **pReader, const char *idstr);
 int32_t tsdbRetrieveCacheRows(void *pReader, SSDataBlock *pResBlock, const int32_t *slotIds, const int32_t *dstSlotIds,
@@ -264,7 +270,7 @@ int32_t tqSeekVer(STqReader *pReader, int64_t ver, const char *id);
 int32_t tqNextBlockInWal(STqReader* pReader);
 bool    tqNextBlockImpl(STqReader *pReader, const char* idstr);
 
-int32_t extractSubmitMsgFromWal(SWalReader *pReader, SPackedData *pPackedData);
+int32_t extractMsgFromWal(SWalReader* pReader, void** pItem, const char* id);
 int32_t tqReaderSetSubmitMsg(STqReader *pReader, void *msgStr, int32_t msgLen, int64_t ver);
 bool    tqNextDataBlockFilterOut(STqReader *pReader, SHashObj *filterOutUids);
 int32_t tqRetrieveDataBlock(STqReader *pReader, const char* idstr);

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

@@ -46,23 +46,23 @@ typedef struct STqOffsetStore STqOffsetStore;
 
 // tqPush
 
-typedef struct {
+//typedef struct {
-  // msg info
+//  // msg info
-  int64_t consumerId;
+//  int64_t consumerId;
-  int64_t reqOffset;
+//  int64_t reqOffset;
-  int64_t processedVer;
+//  int64_t processedVer;
-  int32_t epoch;
+//  int32_t epoch;
-  // rpc info
+//  // rpc info
-  int64_t        reqId;
+//  int64_t        reqId;
-  SRpcHandleInfo rpcInfo;
+//  SRpcHandleInfo rpcInfo;
-  tmr_h          timerId;
+//  tmr_h          timerId;
-  int8_t         tmrStopped;
+//  int8_t         tmrStopped;
-  // exec
+//  // exec
-  int8_t       inputStatus;
+//  int8_t       inputStatus;
-  int8_t       execStatus;
+//  int8_t       execStatus;
-  SStreamQueue inputQ;
+//  SStreamQueue inputQ;
-  SRWLatch     lock;
+//  SRWLatch     lock;
-} STqPushHandle;
+//} STqPushHandle;
 
 // tqExec
 
@@ -90,6 +90,11 @@ typedef struct {
   int32_t      numOfCols;  // number of out pout column, temporarily used
 } STqExecHandle;
 
+typedef enum tq_handle_status{
+  TMQ_HANDLE_STATUS_IDLE = 0,
+  TMQ_HANDLE_STATUS_EXEC = 1,
+}tq_handle_status;
+
 typedef struct {
   char          subKey[TSDB_SUBSCRIBE_KEY_LEN];
   int64_t       consumerId;
@@ -98,11 +103,11 @@ typedef struct {
   int64_t       snapshotVer;
   SWalReader*   pWalReader;
   SWalRef*      pRef;
-  STqPushHandle pushHandle;    // push
+//  STqPushHandle pushHandle;    // push
   STqExecHandle execHandle;    // exec
   SRpcMsg*      msg;
   int32_t       noDataPollCnt;
-  int8_t        exec;
+  tq_handle_status        status;
 } STqHandle;
 
 struct STQ {
@@ -177,13 +182,13 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t ver);
 int32_t tqStreamTasksScanWal(STQ* pTq);
 
 // tq util
+int32_t extractDelDataBlock(const void* pData, int32_t len, int64_t ver, SStreamRefDataBlock** pRefBlock);
 char*   createStreamTaskIdStr(int64_t streamId, int32_t taskId);
-int32_t tqAddInputBlockNLaunchTask(SStreamTask* pTask, SStreamQueueItem* pQueueItem, int64_t ver);
+int32_t tqAddInputBlockNLaunchTask(SStreamTask* pTask, SStreamQueueItem* pQueueItem);
 int32_t tqExtractDataForMq(STQ* pTq, STqHandle* pHandle, const SMqPollReq* pRequest, SRpcMsg* pMsg);
 int32_t tqDoSendDataRsp(const SRpcHandleInfo* pRpcHandleInfo, const SMqDataRsp* pRsp, int32_t epoch, int64_t consumerId,
                         int32_t type, int64_t sver, int64_t ever);
 int32_t tqInitDataRsp(SMqDataRsp* pRsp, const SMqPollReq* pReq);
-bool    tqIsHandleExecuting(STqHandle* pHandle);
 #ifdef __cplusplus
 }
 #endif

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

@@ -220,7 +220,7 @@ int32_t tqProcessTaskResumeReq(STQ* pTq, int64_t version, char* msg, int32_t msg
 int32_t tqProcessStreamTaskCheckReq(STQ* pTq, SRpcMsg* pMsg);
 int32_t tqProcessStreamTaskCheckRsp(STQ* pTq, int64_t version, char* msg, int32_t msgLen);
 int32_t tqProcessSubmitReqForSubscribe(STQ* pTq);
-int32_t tqProcessDelReq(STQ* pTq, void* pReq, int32_t len, int64_t ver);
+int32_t tqProcessDeleteDataReq(STQ* pTq, void* pReq, int32_t len, int64_t ver);
 int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg);
 int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg, bool exec);
 int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg);

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

@@ -59,6 +59,13 @@ struct SMetaCache {
     SHashObj*     pTableEntry;
     SLRUCache*    pUidResCache;
   } sTagFilterResCache;
+
+  struct STbGroupResCache {
+    TdThreadMutex lock;
+    uint32_t      accTimes;
+    SHashObj*     pTableEntry;
+    SLRUCache*    pResCache;
+  } STbGroupResCache;
 };
 
 static void entryCacheClose(SMeta* pMeta) {
@@ -144,6 +151,25 @@ int32_t metaCacheOpen(SMeta* pMeta) {
   taosHashSetFreeFp(pCache->sTagFilterResCache.pTableEntry, freeCacheEntryFp);
   taosThreadMutexInit(&pCache->sTagFilterResCache.lock, NULL);
 
+
+  pCache->STbGroupResCache.pResCache = taosLRUCacheInit(5 * 1024 * 1024, -1, 0.5);
+  if (pCache->STbGroupResCache.pResCache == NULL) {
+    code = TSDB_CODE_OUT_OF_MEMORY;
+    goto _err2;
+  }
+
+  pCache->STbGroupResCache.accTimes = 0;
+  pCache->STbGroupResCache.pTableEntry =
+      taosHashInit(1024, taosGetDefaultHashFunction(TSDB_DATA_TYPE_VARCHAR), false, HASH_NO_LOCK);
+  if (pCache->STbGroupResCache.pTableEntry == NULL) {
+    code = TSDB_CODE_OUT_OF_MEMORY;
+    goto _err2;
+  }
+
+  taosHashSetFreeFp(pCache->STbGroupResCache.pTableEntry, freeCacheEntryFp);
+  taosThreadMutexInit(&pCache->STbGroupResCache.lock, NULL);
+
+
   pMeta->pCache = pCache;
   return code;
 
@@ -165,6 +191,10 @@ void metaCacheClose(SMeta* pMeta) {
     taosThreadMutexDestroy(&pMeta->pCache->sTagFilterResCache.lock);
     taosHashCleanup(pMeta->pCache->sTagFilterResCache.pTableEntry);
 
+    taosLRUCacheCleanup(pMeta->pCache->STbGroupResCache.pResCache);
+    taosThreadMutexDestroy(&pMeta->pCache->STbGroupResCache.lock);
+    taosHashCleanup(pMeta->pCache->STbGroupResCache.pTableEntry);
+
     taosMemoryFree(pMeta->pCache);
     pMeta->pCache = NULL;
   }
@@ -520,7 +550,7 @@ int32_t metaGetCachedTableUidList(SMeta* pMeta, tb_uid_t suid, const uint8_t* pK
   return TSDB_CODE_SUCCESS;
 }
 
-static void freePayload(const void* key, size_t keyLen, void* value) {
+static void freeUidCachePayload(const void* key, size_t keyLen, void* value) {
   if (value == NULL) {
     return;
   }
@@ -626,7 +656,7 @@ int32_t metaUidFilterCachePut(SMeta* pMeta, uint64_t suid, const void* pKey, int
   }
 
   // add to cache.
-  taosLRUCacheInsert(pCache, key, TAG_FILTER_RES_KEY_LEN, pPayload, payloadLen, freePayload, NULL,
+  taosLRUCacheInsert(pCache, key, TAG_FILTER_RES_KEY_LEN, pPayload, payloadLen, freeUidCachePayload, NULL,
                      TAOS_LRU_PRIORITY_LOW);
 _end:
   taosThreadMutexUnlock(pLock);
@@ -671,3 +701,180 @@ int32_t metaUidCacheClear(SMeta* pMeta, uint64_t suid) {
   metaDebug("vgId:%d suid:%"PRId64" cached related tag filter uid list cleared", vgId, suid);
   return TSDB_CODE_SUCCESS;
 }
+
+int32_t metaGetCachedTbGroup(SMeta* pMeta, tb_uid_t suid, const uint8_t* pKey, int32_t keyLen, SArray** pList) {
+  int32_t vgId = TD_VID(pMeta->pVnode);
+
+  // generate the composed key for LRU cache
+  SLRUCache*     pCache = pMeta->pCache->STbGroupResCache.pResCache;
+  SHashObj*      pTableMap = pMeta->pCache->STbGroupResCache.pTableEntry;
+  TdThreadMutex* pLock = &pMeta->pCache->STbGroupResCache.lock;
+
+  *pList = NULL;
+  uint64_t key[4];
+  initCacheKey(key, pTableMap, suid, (const char*)pKey, keyLen);
+
+  taosThreadMutexLock(pLock);
+  pMeta->pCache->STbGroupResCache.accTimes += 1;
+
+  LRUHandle* pHandle = taosLRUCacheLookup(pCache, key, TAG_FILTER_RES_KEY_LEN);
+  if (pHandle == NULL) {
+    taosThreadMutexUnlock(pLock);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  STagFilterResEntry** pEntry = taosHashGet(pTableMap, &suid, sizeof(uint64_t));
+  if (NULL == pEntry) {
+    metaDebug("suid %" PRIu64 " not in tb group cache", suid);
+    return TSDB_CODE_FAILED;
+  }
+
+  *pList = taosArrayDup(taosLRUCacheValue(pCache, pHandle), NULL);
+
+  (*pEntry)->hitTimes += 1;
+
+  uint32_t acc = pMeta->pCache->STbGroupResCache.accTimes;
+  if ((*pEntry)->hitTimes % 5000 == 0 && (*pEntry)->hitTimes > 0) {
+    metaInfo("vgId:%d tb group cache hit:%d, total acc:%d, rate:%.2f", vgId, (*pEntry)->hitTimes, acc, ((double)(*pEntry)->hitTimes) / acc);
+  }
+
+  taosLRUCacheRelease(pCache, pHandle, false);
+
+  // unlock meta
+  taosThreadMutexUnlock(pLock);
+  return TSDB_CODE_SUCCESS;
+}
+
+
+static void freeTbGroupCachePayload(const void* key, size_t keyLen, void* value) {
+  if (value == NULL) {
+    return;
+  }
+
+  const uint64_t* p = key;
+  if (keyLen != sizeof(int64_t) * 4) {
+    metaError("tb group key length is invalid, length:%d, expect:%d", (int32_t)keyLen, (int32_t)sizeof(uint64_t) * 2);
+    return;
+  }
+
+  SHashObj* pHashObj = (SHashObj*)p[0];
+
+  STagFilterResEntry** pEntry = taosHashGet(pHashObj, &p[1], sizeof(uint64_t));
+
+  if (pEntry != NULL && (*pEntry) != NULL) {
+    int64_t st = taosGetTimestampUs();
+
+    SListIter iter = {0};
+    tdListInitIter((SList*)&((*pEntry)->list), &iter, TD_LIST_FORWARD);
+
+    SListNode* pNode = NULL;
+    while ((pNode = tdListNext(&iter)) != NULL) {
+      uint64_t* digest = (uint64_t*)pNode->data;
+      if (digest[0] == p[2] && digest[1] == p[3]) {
+        void* tmp = tdListPopNode(&((*pEntry)->list), pNode);
+        taosMemoryFree(tmp);
+
+        double el = (taosGetTimestampUs() - st) / 1000.0;
+        metaDebug("clear one item in tb group cache, remain cached item:%d, elapsed time:%.2fms", listNEles(&((*pEntry)->list)),
+                 el);
+        break;
+      }
+    }
+  }
+
+  taosArrayDestroy((SArray*)value);
+}
+
+
+int32_t metaPutTbGroupToCache(SMeta* pMeta, uint64_t suid, const void* pKey, int32_t keyLen, void* pPayload,
+                              int32_t payloadLen) {
+  int32_t code = 0;
+  int32_t vgId = TD_VID(pMeta->pVnode);
+
+  if (payloadLen > tsTagFilterResCacheSize) {
+    metaDebug("vgId:%d, suid:%" PRIu64
+              " ignore to add to tb group cache, due to payload length %d greater than threshold %d",
+              vgId, suid, payloadLen, tsTagFilterResCacheSize);
+    taosArrayDestroy((SArray*)pPayload);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SLRUCache*     pCache = pMeta->pCache->STbGroupResCache.pResCache;
+  SHashObj*      pTableEntry = pMeta->pCache->STbGroupResCache.pTableEntry;
+  TdThreadMutex* pLock = &pMeta->pCache->STbGroupResCache.lock;
+
+  uint64_t key[4] = {0};
+  initCacheKey(key, pTableEntry, suid, pKey, keyLen);
+
+  taosThreadMutexLock(pLock);
+  STagFilterResEntry** pEntry = taosHashGet(pTableEntry, &suid, sizeof(uint64_t));
+  if (pEntry == NULL) {
+    code = addNewEntry(pTableEntry, pKey, keyLen, suid);
+    if (code != TSDB_CODE_SUCCESS) {
+      goto _end;
+    }
+  } else { // check if it exists or not
+    size_t size = listNEles(&(*pEntry)->list);
+    if (size == 0) {
+      tdListAppend(&(*pEntry)->list, pKey);
+    } else {
+      SListNode* pNode = listHead(&(*pEntry)->list);
+      uint64_t*  p = (uint64_t*)pNode->data;
+      if (p[1] == ((uint64_t*)pKey)[1] && p[0] == ((uint64_t*)pKey)[0]) {
+        // we have already found the existed items, no need to added to cache anymore.
+        taosThreadMutexUnlock(pLock);
+        return TSDB_CODE_SUCCESS;
+      } else {  // not equal, append it
+        tdListAppend(&(*pEntry)->list, pKey);
+      }
+    }
+  }
+
+  // add to cache.
+  taosLRUCacheInsert(pCache, key, TAG_FILTER_RES_KEY_LEN, pPayload, payloadLen, freeTbGroupCachePayload, NULL,
+                     TAOS_LRU_PRIORITY_LOW);
+_end:
+  taosThreadMutexUnlock(pLock);
+  metaDebug("vgId:%d, suid:%" PRIu64 " tb group added into cache, total:%d, tables:%d", vgId, suid,
+            (int32_t)taosLRUCacheGetUsage(pCache), taosHashGetSize(pTableEntry));
+
+  return code;
+}
+
+// remove the lru cache that are expired due to the tags value update, or creating, or dropping, of child tables
+int32_t metaTbGroupCacheClear(SMeta* pMeta, uint64_t suid) {
+  uint64_t  p[4] = {0};
+  int32_t   vgId = TD_VID(pMeta->pVnode);
+  SHashObj* pEntryHashMap = pMeta->pCache->STbGroupResCache.pTableEntry;
+
+  uint64_t dummy[2] = {0};
+  initCacheKey(p, pEntryHashMap, suid, (char*) &dummy[0], 16);
+
+  TdThreadMutex* pLock = &pMeta->pCache->STbGroupResCache.lock;
+  taosThreadMutexLock(pLock);
+
+  STagFilterResEntry** pEntry = taosHashGet(pEntryHashMap, &suid, sizeof(uint64_t));
+  if (pEntry == NULL || listNEles(&(*pEntry)->list) == 0) {
+    taosThreadMutexUnlock(pLock);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  (*pEntry)->hitTimes = 0;
+
+  SListIter iter = {0};
+  tdListInitIter(&(*pEntry)->list, &iter, TD_LIST_FORWARD);
+
+  SListNode* pNode = NULL;
+  while ((pNode = tdListNext(&iter)) != NULL) {
+    setMD5DigestInKey(p, pNode->data, 2 * sizeof(uint64_t));
+    taosLRUCacheErase(pMeta->pCache->STbGroupResCache.pResCache, p, TAG_FILTER_RES_KEY_LEN);
+  }
+
+  tdListEmpty(&(*pEntry)->list);
+  taosThreadMutexUnlock(pLock);
+
+  metaDebug("vgId:%d suid:%"PRId64" cached related tb group cleared", vgId, suid);
+  return TSDB_CODE_SUCCESS;
+}
+
+

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

@@ -767,6 +767,7 @@ int metaCreateTable(SMeta *pMeta, int64_t version, SVCreateTbReq *pReq, STableMe
     metaWLock(pMeta);
     metaUpdateStbStats(pMeta, me.ctbEntry.suid, 1);
     metaUidCacheClear(pMeta, me.ctbEntry.suid);
+    metaTbGroupCacheClear(pMeta, me.ctbEntry.suid);
     metaULock(pMeta);
   } else {
     me.ntbEntry.ctime = pReq->ctime;
@@ -998,6 +999,7 @@ static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type) {
 
     metaUpdateStbStats(pMeta, e.ctbEntry.suid, -1);
     metaUidCacheClear(pMeta, e.ctbEntry.suid);
+    metaTbGroupCacheClear(pMeta, e.ctbEntry.suid);    
   } else if (e.type == TSDB_NORMAL_TABLE) {
     // drop schema.db (todo)
 
@@ -1009,6 +1011,7 @@ static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type) {
 
     metaStatsCacheDrop(pMeta, uid);
     metaUidCacheClear(pMeta, uid);
+    metaTbGroupCacheClear(pMeta, uid);        
     --pMeta->pVnode->config.vndStats.numOfSTables;
   }
 
@@ -1429,6 +1432,7 @@ static int metaUpdateTableTagVal(SMeta *pMeta, int64_t version, SVAlterTbReq *pA
               ((STag *)(ctbEntry.ctbEntry.pTags))->len, pMeta->txn);
 
   metaUidCacheClear(pMeta, ctbEntry.ctbEntry.suid);
+  metaTbGroupCacheClear(pMeta, ctbEntry.ctbEntry.suid);        
 
   metaULock(pMeta);
 

source/dnode/vnode/src/tq/tq.c

@@ -22,6 +22,10 @@
 
 static int32_t tqInitialize(STQ* pTq);
 
+static FORCE_INLINE bool tqIsHandleExec(STqHandle* pHandle) { return TMQ_HANDLE_STATUS_EXEC == pHandle->status; }
+static FORCE_INLINE void tqSetHandleExec(STqHandle* pHandle) {pHandle->status = TMQ_HANDLE_STATUS_EXEC;}
+static FORCE_INLINE void tqSetHandleIdle(STqHandle* pHandle) {pHandle->status = TMQ_HANDLE_STATUS_IDLE;}
+
 int32_t tqInit() {
   int8_t old;
   while (1) {
@@ -61,6 +65,7 @@ void tqCleanUp() {
 static void destroyTqHandle(void* data) {
   STqHandle* pData = (STqHandle*)data;
   qDestroyTask(pData->execHandle.task);
+
   if (pData->execHandle.subType == TOPIC_SUB_TYPE__COLUMN) {
     taosMemoryFreeClear(pData->execHandle.execCol.qmsg);
   } else if (pData->execHandle.subType == TOPIC_SUB_TYPE__DB) {
@@ -302,13 +307,6 @@ int32_t tqProcessOffsetCommitReq(STQ* pTq, int64_t sversion, char* msg, int32_t
     return -1;
   }
 
-  if (pOffset->val.type == TMQ_OFFSET__LOG) {
-    STqHandle* pHandle = taosHashGet(pTq->pHandle, pOffset->subKey, strlen(pOffset->subKey));
-    if (pHandle && (walRefVer(pHandle->pRef, pOffset->val.version) < 0)) {
-      return -1;
-    }
-  }
-
   return 0;
 }
 
@@ -319,11 +317,15 @@ int32_t tqProcessSeekReq(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen)
   SDecoder decoder;
   tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
   if (tDecodeMqVgOffset(&decoder, &vgOffset) < 0) {
+    tqError("vgId:%d failed to decode seek msg", vgId);
     return -1;
   }
 
   tDecoderClear(&decoder);
 
+  tqDebug("topic:%s, vgId:%d process offset seek by consumer:0x%" PRIx64 ", req offset:%" PRId64,
+          vgOffset.offset.subKey, vgId, vgOffset.consumerId, vgOffset.offset.val.version);
+
   STqOffset* pOffset = &vgOffset.offset;
   if (pOffset->val.type != TMQ_OFFSET__LOG) {
     tqError("vgId:%d, subKey:%s invalid seek offset type:%d", vgId, pOffset->subKey, pOffset->val.type);
@@ -385,6 +387,9 @@ int32_t tqProcessSeekReq(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen)
     return -1;
   }
 
+  tqDebug("topic:%s, vgId:%d consumer:0x%" PRIx64 " offset is update to:%" PRId64, vgOffset.offset.subKey, vgId,
+          vgOffset.consumerId, vgOffset.offset.val.version);
+
   return 0;
 }
 
@@ -427,41 +432,48 @@ int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg) {
   STqOffsetVal reqOffset = req.reqOffset;
   int32_t      vgId = TD_VID(pTq->pVnode);
 
+  taosWLockLatch(&pTq->lock);
   // 1. find handle
   STqHandle* pHandle = taosHashGet(pTq->pHandle, req.subKey, strlen(req.subKey));
   if (pHandle == NULL) {
     tqError("tmq poll: consumer:0x%" PRIx64 " vgId:%d subkey %s not found", consumerId, vgId, req.subKey);
     terrno = TSDB_CODE_INVALID_MSG;
+    taosWUnLockLatch(&pTq->lock);
     return -1;
   }
 
+  while (tqIsHandleExec(pHandle)) {
+    tqDebug("tmq poll: consumer:0x%" PRIx64 "vgId:%d, topic:%s, subscription is executing, wait for 5ms and retry", consumerId, vgId, req.subKey);
+    taosMsleep(5);
+  }
+
   // 2. check re-balance status
-  taosRLockLatch(&pTq->lock);
   if (pHandle->consumerId != consumerId) {
     tqDebug("ERROR tmq poll: consumer:0x%" PRIx64 " vgId:%d, subkey %s, mismatch for saved handle consumer:0x%" PRIx64,
             consumerId, vgId, req.subKey, pHandle->consumerId);
     terrno = TSDB_CODE_TMQ_CONSUMER_MISMATCH;
-    taosRUnLockLatch(&pTq->lock);
+    taosWUnLockLatch(&pTq->lock);
     return -1;
   }
-  taosRUnLockLatch(&pTq->lock);
+  tqSetHandleExec(pHandle);
+  taosWUnLockLatch(&pTq->lock);
 
   // 3. update the epoch value
-  taosWLockLatch(&pTq->lock);
   int32_t savedEpoch = pHandle->epoch;
   if (savedEpoch < reqEpoch) {
     tqDebug("tmq poll: consumer:0x%" PRIx64 " epoch update from %d to %d by poll req", consumerId, savedEpoch,
             reqEpoch);
     pHandle->epoch = reqEpoch;
   }
-  taosWUnLockLatch(&pTq->lock);
 
   char buf[80];
   tFormatOffset(buf, 80, &reqOffset);
   tqDebug("tmq poll: consumer:0x%" PRIx64 " (epoch %d), subkey %s, recv poll req vgId:%d, req:%s, reqId:0x%" PRIx64,
           consumerId, req.epoch, pHandle->subKey, vgId, buf, req.reqId);
 
-  return tqExtractDataForMq(pTq, pHandle, &req, pMsg);
+  int code = tqExtractDataForMq(pTq, pHandle, &req, pMsg);
+  tqSetHandleIdle(pHandle);
+  return code;
 }
 
 int32_t tqProcessVgWalInfoReq(STQ* pTq, SRpcMsg* pMsg) {
@@ -549,25 +561,19 @@ int32_t tqProcessDeleteSubReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
 
   tqDebug("vgId:%d, tq process delete sub req %s", vgId, pReq->subKey);
   int32_t code = 0;
-//  taosWLockLatch(&pTq->lock);
-//  int32_t code = taosHashRemove(pTq->pPushMgr, pReq->subKey, strlen(pReq->subKey));
-//  if (code != 0) {
-//    tqDebug("vgId:%d, tq remove push handle %s", pTq->pVnode->config.vgId, pReq->subKey);
-//  }
-//  taosWUnLockLatch(&pTq->lock);
 
+  taosWLockLatch(&pTq->lock);
   STqHandle* pHandle = taosHashGet(pTq->pHandle, pReq->subKey, strlen(pReq->subKey));
   if (pHandle) {
-    // walCloseRef(pHandle->pWalReader->pWal, pHandle->pRef->refId);
+    while (tqIsHandleExec(pHandle)) {
-    if (pHandle->pRef) {
-      walCloseRef(pTq->pVnode->pWal, pHandle->pRef->refId);
-    }
-
-    while (tqIsHandleExecuting(pHandle)) {
       tqDebug("vgId:%d, topic:%s, subscription is executing, wait for 5ms and retry", vgId, pHandle->subKey);
       taosMsleep(5);
     }
 
+    if (pHandle->pRef) {
+      walCloseRef(pTq->pVnode->pWal, pHandle->pRef->refId);
+    }
+
     code = taosHashRemove(pTq->pHandle, pReq->subKey, strlen(pReq->subKey));
     if (code != 0) {
       tqError("cannot process tq delete req %s, since no such handle", pReq->subKey);
@@ -582,6 +588,8 @@ int32_t tqProcessDeleteSubReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
   if (tqMetaDeleteHandle(pTq, pReq->subKey) < 0) {
     tqError("cannot process tq delete req %s, since no such offset in tdb", pReq->subKey);
   }
+  taosWUnLockLatch(&pTq->lock);
+
   return 0;
 }
 
@@ -628,6 +636,7 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
   tqDebug("vgId:%d, tq process sub req:%s, Id:0x%" PRIx64 " -> Id:0x%" PRIx64, pVnode->config.vgId, req.subKey,
           req.oldConsumerId, req.newConsumerId);
 
+  taosWLockLatch(&pTq->lock);
   STqHandle* pHandle = taosHashGet(pTq->pHandle, req.subKey, strlen(req.subKey));
   if (pHandle == NULL) {
     if (req.oldConsumerId != -1) {
@@ -679,7 +688,7 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
       pHandle->execHandle.pTqReader = tqReaderOpen(pVnode);
 
       pHandle->execHandle.execDb.pFilterOutTbUid =
-          taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_NO_LOCK);
+          taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_ENTRY_LOCK);
       buildSnapContext(handle.meta, handle.version, 0, pHandle->execHandle.subType, pHandle->fetchMeta,
                        (SSnapContext**)(&handle.sContext));
 
@@ -710,6 +719,11 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
     ret = tqMetaSaveHandle(pTq, req.subKey, pHandle);
     goto end;
   } else {
+    while (tqIsHandleExec(pHandle)) {
+      tqDebug("sub req vgId:%d, topic:%s, subscription is executing, wait for 5ms and retry", vgId, pHandle->subKey);
+      taosMsleep(5);
+    }
+
     if (pHandle->consumerId == req.newConsumerId) {  // do nothing
       tqInfo("vgId:%d consumer:0x%" PRIx64 " remains, no switch occurs", req.vgId, req.newConsumerId);
       atomic_add_fetch_32(&pHandle->epoch, 1);
@@ -725,22 +739,17 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
     if (pTaskInfo != NULL) {
       qKillTask(pTaskInfo, TSDB_CODE_SUCCESS);
     }
-
-    taosWLockLatch(&pTq->lock);
-    // remove if it has been register in the push manager, and return one empty block to consumer
-    tqUnregisterPushHandle(pTq, pHandle);
-
-
     if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__COLUMN) {
       qStreamCloseTsdbReader(pTaskInfo);
     }
-
+    // remove if it has been register in the push manager, and return one empty block to consumer
-    taosWUnLockLatch(&pTq->lock);
+    tqUnregisterPushHandle(pTq, pHandle);
     ret = tqMetaSaveHandle(pTq, req.subKey, pHandle);
     goto end;
   }
 
 end:
+  taosWUnLockLatch(&pTq->lock);
   taosMemoryFree(req.qmsg);
   return ret;
 }
@@ -828,7 +837,8 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t ver) {
   }
 
   if (pTask->taskLevel == TASK_LEVEL__SOURCE) {
-    pTask->exec.pWalReader = walOpenReader(pTq->pVnode->pWal, NULL);
+    SWalFilterCond cond = {.deleteMsg = 1};
+    pTask->exec.pWalReader = walOpenReader(pTq->pVnode->pWal, &cond);
   }
 
   streamSetupTrigger(pTask);
@@ -1032,7 +1042,7 @@ int32_t tqProcessTaskRecover1Req(STQ* pTq, SRpcMsg* pMsg) {
   memcpy(serializedReq, &req, len);
 
   // dispatch msg
-  tqDebug("s-task:%s step 1 finished, send msg to start blocking recover stage(step 2)", pTask->id.idStr);
+  tqDebug("s-task:%s start recover block stage", pTask->id.idStr);
 
   SRpcMsg rpcMsg = {
       .code = 0, .contLen = len, .msgType = TDMT_VND_STREAM_RECOVER_BLOCKING_STAGE, .pCont = serializedReq};
@@ -1060,6 +1070,9 @@ int32_t tqProcessTaskRecover2Req(STQ* pTq, int64_t sversion, char* msg, int32_t
     return -1;
   }
 
+  qDebug("s-task:%s set the start wal offset to be:%"PRId64, pTask->id.idStr, sversion);
+  walReaderSeekVer(pTask->exec.pWalReader, sversion);
+
   if (atomic_load_8(&pTask->status.taskStatus) == TASK_STATUS__DROPPING) {
     streamMetaReleaseTask(pTq->pStreamMeta, pTask);
     return 0;
@@ -1129,7 +1142,60 @@ int32_t tqProcessTaskRecoverFinishRsp(STQ* pTq, SRpcMsg* pMsg) {
   return 0;
 }
 
-int32_t tqProcessDelReq(STQ* pTq, void* pReq, int32_t len, int64_t ver) {
+int32_t extractDelDataBlock(const void* pData, int32_t len, int64_t ver, SStreamRefDataBlock** pRefBlock) {
+  SDecoder*   pCoder = &(SDecoder){0};
+  SDeleteRes* pRes = &(SDeleteRes){0};
+
+  *pRefBlock = NULL;
+
+  pRes->uidList = taosArrayInit(0, sizeof(tb_uid_t));
+  if (pRes->uidList == NULL) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  tDecoderInit(pCoder, (uint8_t*)pData, len);
+  tDecodeDeleteRes(pCoder, pRes);
+  tDecoderClear(pCoder);
+
+  int32_t numOfTables = taosArrayGetSize(pRes->uidList);
+  if (numOfTables == 0 || pRes->affectedRows == 0) {
+    taosArrayDestroy(pRes->uidList);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SSDataBlock* pDelBlock = createSpecialDataBlock(STREAM_DELETE_DATA);
+  blockDataEnsureCapacity(pDelBlock, numOfTables);
+  pDelBlock->info.rows = numOfTables;
+  pDelBlock->info.version = ver;
+
+  for (int32_t i = 0; i < numOfTables; i++) {
+    // start key column
+    SColumnInfoData* pStartCol = taosArrayGet(pDelBlock->pDataBlock, START_TS_COLUMN_INDEX);
+    colDataSetVal(pStartCol, i, (const char*)&pRes->skey, false);  // end key column
+    SColumnInfoData* pEndCol = taosArrayGet(pDelBlock->pDataBlock, END_TS_COLUMN_INDEX);
+    colDataSetVal(pEndCol, i, (const char*)&pRes->ekey, false);
+    // uid column
+    SColumnInfoData* pUidCol = taosArrayGet(pDelBlock->pDataBlock, UID_COLUMN_INDEX);
+    int64_t*         pUid = taosArrayGet(pRes->uidList, i);
+    colDataSetVal(pUidCol, i, (const char*)pUid, false);
+
+    colDataSetNULL(taosArrayGet(pDelBlock->pDataBlock, GROUPID_COLUMN_INDEX), i);
+    colDataSetNULL(taosArrayGet(pDelBlock->pDataBlock, CALCULATE_START_TS_COLUMN_INDEX), i);
+    colDataSetNULL(taosArrayGet(pDelBlock->pDataBlock, CALCULATE_END_TS_COLUMN_INDEX), i);
+  }
+
+  taosArrayDestroy(pRes->uidList);
+  *pRefBlock = taosAllocateQitem(sizeof(SStreamRefDataBlock), DEF_QITEM, 0);
+  if (pRefBlock == NULL) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  (*pRefBlock)->type = STREAM_INPUT__REF_DATA_BLOCK;
+  (*pRefBlock)->pBlock = pDelBlock;
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t tqProcessDeleteDataReq(STQ* pTq, void* pReq, int32_t len, int64_t ver) {
   bool        failed = false;
   SDecoder*   pCoder = &(SDecoder){0};
   SDeleteRes* pRes = &(SDeleteRes){0};
@@ -1149,6 +1215,7 @@ int32_t tqProcessDelReq(STQ* pTq, void* pReq, int32_t len, int64_t ver) {
     taosArrayDestroy(pRes->uidList);
     return 0;
   }
+
   SSDataBlock* pDelBlock = createSpecialDataBlock(STREAM_DELETE_DATA);
   blockDataEnsureCapacity(pDelBlock, sz);
   pDelBlock->info.rows = sz;
@@ -1195,8 +1262,6 @@ int32_t tqProcessDelReq(STQ* pTq, void* pReq, int32_t len, int64_t ver) {
       SStreamRefDataBlock* pRefBlock = taosAllocateQitem(sizeof(SStreamRefDataBlock), DEF_QITEM, 0);
       pRefBlock->type = STREAM_INPUT__REF_DATA_BLOCK;
       pRefBlock->pBlock = pDelBlock;
-      pRefBlock->dataRef = pRef;
-      atomic_add_fetch_32(pRefBlock->dataRef, 1);
 
       if (tAppendDataToInputQueue(pTask, (SStreamQueueItem*)pRefBlock) < 0) {
         atomic_sub_fetch_32(pRef, 1);
@@ -1375,6 +1440,7 @@ int32_t tqProcessTaskResumeReq(STQ* pTq, int64_t sversion, char* msg, int32_t ms
     // no lock needs to secure the access of the version
     if (pReq->igUntreated) {  // discard all the data  when the stream task is suspended.
       pTask->chkInfo.currentVer = sversion;
+      walReaderSeekVer(pTask->exec.pWalReader, sversion);
       tqDebug("vgId:%d s-task:%s resume to normal from the latest version:%" PRId64 ", vnode ver:%" PRId64, pTq->pStreamMeta->vgId,
               pTask->id.idStr, pTask->chkInfo.currentVer, sversion);
     } else {  // from the previous paused version and go on

source/dnode/vnode/src/tq/tqMeta.c

@@ -54,7 +54,7 @@ int32_t tDecodeSTqHandle(SDecoder* pDecoder, STqHandle* pHandle) {
     if (tDecodeCStrAlloc(pDecoder, &pHandle->execHandle.execCol.qmsg) < 0) return -1;
   } else if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__DB) {
     pHandle->execHandle.execDb.pFilterOutTbUid =
-        taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_NO_LOCK);
+        taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_ENTRY_LOCK);
     int32_t size = 0;
     if (tDecodeI32(pDecoder, &size) < 0) return -1;
     for (int32_t i = 0; i < size; i++) {
@@ -295,7 +295,7 @@ int32_t tqMetaRestoreHandle(STQ* pTq) {
       code = -1;
       goto end;
     }
-    walRefVer(handle.pRef, handle.snapshotVer);
+    walSetRefVer(handle.pRef, handle.snapshotVer);
 
     SReadHandle reader = {
         .meta = pTq->pVnode->pMeta,
@@ -352,7 +352,9 @@ int32_t tqMetaRestoreHandle(STQ* pTq) {
       handle.execHandle.task = qCreateQueueExecTaskInfo(NULL, &reader, vgId, NULL, 0);
     }
     tqDebug("tq restore %s consumer %" PRId64 " vgId:%d", handle.subKey, handle.consumerId, vgId);
+    taosWLockLatch(&pTq->lock);
     taosHashPut(pTq->pHandle, pKey, kLen, &handle, sizeof(STqHandle));
+    taosWUnLockLatch(&pTq->lock);
   }
 
 end:

source/dnode/vnode/src/tq/tqOffset.c

@@ -78,13 +78,15 @@ int32_t tqOffsetRestoreFromFile(STqOffsetStore* pStore, const char* fname) {
 
     // todo remove this
     if (offset.val.type == TMQ_OFFSET__LOG) {
+      taosWLockLatch(&pStore->pTq->lock);
       STqHandle* pHandle = taosHashGet(pStore->pTq->pHandle, offset.subKey, strlen(offset.subKey));
       if (pHandle) {
-        if (walRefVer(pHandle->pRef, offset.val.version) < 0) {
+        if (walSetRefVer(pHandle->pRef, offset.val.version) < 0) {
 //          tqError("vgId: %d, tq handle %s ref ver %" PRId64 "error", pStore->pTq->pVnode->config.vgId, pHandle->subKey,
 //                  offset.val.version);
         }
       }
+      taosWUnLockLatch(&pStore->pTq->lock);
     }
 
     taosMemoryFree(pMemBuf);

source/dnode/vnode/src/tq/tqPush.c

@@ -34,12 +34,12 @@ int32_t tqPushMsg(STQ* pTq, void* msg, int32_t msgLen, tmsg_t msgType, int64_t v
       return 0;
     }
 
-    if (msgType == TDMT_VND_SUBMIT) {
+    if (msgType == TDMT_VND_SUBMIT || msgType == TDMT_VND_DELETE) {
       tqStartStreamTasks(pTq);
     }
 
     if (msgType == TDMT_VND_DELETE) {
-      tqProcessDelReq(pTq, POINTER_SHIFT(msg, sizeof(SMsgHead)), msgLen - sizeof(SMsgHead), ver);
+//      tqProcessDeleteDataReq(pTq, POINTER_SHIFT(msg, sizeof(SMsgHead)), msgLen - sizeof(SMsgHead), ver);
     }
   }
 

source/dnode/vnode/src/tq/tqRead.c

@@ -302,25 +302,44 @@ int32_t tqSeekVer(STqReader* pReader, int64_t ver, const char* id) {
   return 0;
 }
 
-int32_t extractSubmitMsgFromWal(SWalReader* pReader, SPackedData* pPackedData) {
+int32_t extractMsgFromWal(SWalReader* pReader, void** pItem, const char* id) {
-  if (walNextValidMsg(pReader) < 0) {
+  int32_t code = walNextValidMsg(pReader);
-    return -1;
+  if (code != TSDB_CODE_SUCCESS) {
+    return code;
   }
 
-  void*   pBody = POINTER_SHIFT(pReader->pHead->head.body, sizeof(SSubmitReq2Msg));
-  int32_t len = pReader->pHead->head.bodyLen - sizeof(SSubmitReq2Msg);
   int64_t ver = pReader->pHead->head.version;
 
-  void* data = taosMemoryMalloc(len);
+  if (pReader->pHead->head.msgType == TDMT_VND_SUBMIT) {
-  if (data == NULL) {
+    void*   pBody = POINTER_SHIFT(pReader->pHead->head.body, sizeof(SSubmitReq2Msg));
-    // todo: for all stream in this vnode, keep this offset in the offset files, and wait for a moment, and then retry
+    int32_t len = pReader->pHead->head.bodyLen - sizeof(SSubmitReq2Msg);
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
+
-    tqError("vgId:%d, failed to copy submit data for stream processing, since out of memory", 0);
+    void* data = taosMemoryMalloc(len);
-    return -1;
+    if (data == NULL) {
+      // todo: for all stream in this vnode, keep this offset in the offset files, and wait for a moment, and then retry
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      tqError("vgId:%d, failed to copy submit data for stream processing, since out of memory", 0);
+      return -1;
+    }
+
+    memcpy(data, pBody, len);
+    SPackedData data1 = (SPackedData){.ver = ver, .msgLen = len, .msgStr = data};
+
+    *pItem = (SStreamQueueItem*)streamDataSubmitNew(data1, STREAM_INPUT__DATA_SUBMIT);
+    if (*pItem == NULL) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      tqError("%s failed to create data submit for stream since out of memory", id);
+      return terrno;
+    }
+  } else if (pReader->pHead->head.msgType == TDMT_VND_DELETE) {
+    void*   pBody = POINTER_SHIFT(pReader->pHead->head.body, sizeof(SMsgHead));
+    int32_t len = pReader->pHead->head.bodyLen - sizeof(SMsgHead);
+
+    extractDelDataBlock(pBody, len, ver, (SStreamRefDataBlock**)pItem);
+  } else {
+    ASSERT(0);
   }
 
-  memcpy(data, pBody, len);
-  *pPackedData = (SPackedData){.ver = ver, .msgLen = len, .msgStr = data};
   return 0;
 }
 
@@ -503,9 +522,76 @@ int32_t tqMaskBlock(SSchemaWrapper* pDst, SSDataBlock* pBlock, const SSchemaWrap
   return 0;
 }
 
-int32_t tqRetrieveDataBlock(STqReader* pReader, const char* idstr) {
+static int32_t buildResSDataBlock(SSDataBlock* pBlock, SSchemaWrapper* pSchema, const SArray* pColIdList) {
-  tqDebug("tq reader retrieve data block %p, index:%d/%d, %s", pReader->msg.msgStr, pReader->nextBlk,
+  if (blockDataGetNumOfCols(pBlock) > 0) {
-          (int32_t)taosArrayGetSize(pReader->submit.aSubmitTbData), idstr);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  int32_t numOfCols = taosArrayGetSize(pColIdList);
+
+  if (numOfCols == 0) {  // all columns are required
+    for (int32_t i = 0; i < pSchema->nCols; ++i) {
+      SSchema*        pColSchema = &pSchema->pSchema[i];
+      SColumnInfoData colInfo = createColumnInfoData(pColSchema->type, pColSchema->bytes, pColSchema->colId);
+
+      int32_t code = blockDataAppendColInfo(pBlock, &colInfo);
+      if (code != TSDB_CODE_SUCCESS) {
+        blockDataFreeRes(pBlock);
+        return TSDB_CODE_OUT_OF_MEMORY;
+      }
+    }
+  } else {
+    if (numOfCols > pSchema->nCols) {
+      numOfCols = pSchema->nCols;
+    }
+
+    int32_t i = 0;
+    int32_t j = 0;
+    while (i < pSchema->nCols && j < numOfCols) {
+      SSchema* pColSchema = &pSchema->pSchema[i];
+      col_id_t colIdSchema = pColSchema->colId;
+
+      col_id_t colIdNeed = *(col_id_t*)taosArrayGet(pColIdList, j);
+      if (colIdSchema < colIdNeed) {
+        i++;
+      } else if (colIdSchema > colIdNeed) {
+        j++;
+      } else {
+        SColumnInfoData colInfo = createColumnInfoData(pColSchema->type, pColSchema->bytes, pColSchema->colId);
+        int32_t         code = blockDataAppendColInfo(pBlock, &colInfo);
+        if (code != TSDB_CODE_SUCCESS) {
+          return -1;
+        }
+        i++;
+        j++;
+      }
+    }
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t doSetVal(SColumnInfoData* pColumnInfoData, int32_t rowIndex, SColVal* pColVal) {
+  int32_t code = TSDB_CODE_SUCCESS;
+
+  if (IS_STR_DATA_TYPE(pColVal->type)) {
+    char val[65535 + 2] = {0};
+    if (pColVal->value.pData != NULL) {
+      memcpy(varDataVal(val), pColVal->value.pData, pColVal->value.nData);
+      varDataSetLen(val, pColVal->value.nData);
+      code = colDataSetVal(pColumnInfoData, rowIndex, val, !COL_VAL_IS_VALUE(pColVal));
+    } else {
+      colDataSetNULL(pColumnInfoData, rowIndex);
+    }
+  } else {
+    code = colDataSetVal(pColumnInfoData, rowIndex, (void*)&pColVal->value.val, !COL_VAL_IS_VALUE(pColVal));
+  }
+
+  return code;
+}
+
+int32_t tqRetrieveDataBlock(STqReader* pReader, const char* id) {
+  tqDebug("tq reader retrieve data block %p, index:%d", pReader->msg.msgStr, pReader->nextBlk);
 
   SSubmitTbData* pSubmitTbData = taosArrayGet(pReader->submit.aSubmitTbData, pReader->nextBlk++);
 
@@ -538,53 +624,11 @@ int32_t tqRetrieveDataBlock(STqReader* pReader, const char* idstr) {
     pReader->cachedSchemaSuid = suid;
     pReader->cachedSchemaVer = sversion;
 
-    SSchemaWrapper* pSchemaWrapper = pReader->pSchemaWrapper;
+    ASSERT(pReader->cachedSchemaVer == pReader->pSchemaWrapper->version);
-    if (blockDataGetNumOfCols(pBlock) > 0) {
+    if (blockDataGetNumOfCols(pBlock) == 0) {
-      blockDataDestroy(pReader->pResBlock);
+      int32_t code = buildResSDataBlock(pReader->pResBlock, pReader->pSchemaWrapper, pReader->pColIdList);
-      pReader->pResBlock = createDataBlock();
+      if (code != TSDB_CODE_SUCCESS) {
-      pBlock = pReader->pResBlock;
+        return code;
-
-      pBlock->info.id.uid = uid;
-      pBlock->info.version = pReader->msg.ver;
-    }
-
-    int32_t numOfCols = taosArrayGetSize(pReader->pColIdList);
-    if (numOfCols == 0) {  // all columns are required
-      for (int32_t i = 0; i < pSchemaWrapper->nCols; ++i) {
-        SSchema*        pColSchema = &pSchemaWrapper->pSchema[i];
-        SColumnInfoData colInfo = createColumnInfoData(pColSchema->type, pColSchema->bytes, pColSchema->colId);
-
-        int32_t code = blockDataAppendColInfo(pBlock, &colInfo);
-        if (code != TSDB_CODE_SUCCESS) {
-          blockDataFreeRes(pBlock);
-          return -1;
-        }
-      }
-    } else {
-      if (numOfCols > pSchemaWrapper->nCols) {
-        numOfCols = pSchemaWrapper->nCols;
-      }
-
-      int32_t i = 0;
-      int32_t j = 0;
-      while (i < pSchemaWrapper->nCols && j < numOfCols) {
-        SSchema* pColSchema = &pSchemaWrapper->pSchema[i];
-        col_id_t colIdSchema = pColSchema->colId;
-
-        col_id_t colIdNeed = *(col_id_t*)taosArrayGet(pReader->pColIdList, j);
-        if (colIdSchema < colIdNeed) {
-          i++;
-        } else if (colIdSchema > colIdNeed) {
-          j++;
-        } else {
-          SColumnInfoData colInfo = createColumnInfoData(pColSchema->type, pColSchema->bytes, pColSchema->colId);
-          int32_t         code = blockDataAppendColInfo(pBlock, &colInfo);
-          if (code != TSDB_CODE_SUCCESS) {
-            return -1;
-          }
-          i++;
-          j++;
-        }
       }
     }
   }
@@ -632,30 +676,15 @@ int32_t tqRetrieveDataBlock(STqReader* pReader, const char* idstr) {
       } else if (pCol->cid == pColData->info.colId) {
         for (int32_t i = 0; i < pCol->nVal; i++) {
           tColDataGetValue(pCol, i, &colVal);
-          if (IS_STR_DATA_TYPE(colVal.type)) {
+          int32_t code = doSetVal(pColData, i, &colVal);
-            if (colVal.value.pData != NULL) {
+          if (code != TSDB_CODE_SUCCESS) {
-              char val[65535 + 2] = {0};
+            return code;
-              memcpy(varDataVal(val), colVal.value.pData, colVal.value.nData);
-              varDataSetLen(val, colVal.value.nData);
-              if (colDataAppend(pColData, i, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
-                return -1;
-              }
-            } else {
-              colDataSetNULL(pColData, i);
-            }
-          } else {
-            if (colDataAppend(pColData, i, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
-              return -1;
-            }
           }
         }
         sourceIdx++;
         targetIdx++;
       } else {
-        for (int32_t i = 0; i < pCol->nVal; i++) {
+        colDataSetNNULL(pColData, 0, pCol->nVal);
-          colDataSetNULL(pColData, i);
-        }
-
         targetIdx++;
       }
     }
@@ -679,21 +708,9 @@ int32_t tqRetrieveDataBlock(STqReader* pReader, const char* idstr) {
             sourceIdx++;
             continue;
           } else if (colVal.cid == pColData->info.colId) {
-            if (IS_STR_DATA_TYPE(colVal.type)) {
+            int32_t code = doSetVal(pColData, i, &colVal);
-              if (colVal.value.pData != NULL) {
+            if (code != TSDB_CODE_SUCCESS) {
-                char val[65535 + 2] = {0};
+              return code;
-                memcpy(varDataVal(val), colVal.value.pData, colVal.value.nData);
-                varDataSetLen(val, colVal.value.nData);
-                if (colDataAppend(pColData, i, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
-                  return -1;
-                }
-              } else {
-                colDataSetNULL(pColData, i);
-              }
-            } else {
-              if (colDataAppend(pColData, i, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
-                return -1;
-              }
             }
 
             sourceIdx++;
@@ -831,14 +848,14 @@ int32_t tqRetrieveTaosxBlock(STqReader* pReader, SArray* blocks, SArray* schemas
               char val[65535 + 2];
               memcpy(varDataVal(val), colVal.value.pData, colVal.value.nData);
               varDataSetLen(val, colVal.value.nData);
-              if (colDataAppend(pColData, curRow - lastRow, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
+              if (colDataSetVal(pColData, curRow - lastRow, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
                 goto FAIL;
               }
             } else {
               colDataSetNULL(pColData, curRow - lastRow);
             }
           } else {
-            if (colDataAppend(pColData, curRow - lastRow, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
+            if (colDataSetVal(pColData, curRow - lastRow, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
               goto FAIL;
             }
           }
@@ -928,14 +945,14 @@ int32_t tqRetrieveTaosxBlock(STqReader* pReader, SArray* blocks, SArray* schemas
               char val[65535 + 2];
               memcpy(varDataVal(val), colVal.value.pData, colVal.value.nData);
               varDataSetLen(val, colVal.value.nData);
-              if (colDataAppend(pColData, curRow - lastRow, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
+              if (colDataSetVal(pColData, curRow - lastRow, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
                 goto FAIL;
               }
             } else {
               colDataSetNULL(pColData, curRow - lastRow);
             }
           } else {
-            if (colDataAppend(pColData, curRow - lastRow, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
+            if (colDataSetVal(pColData, curRow - lastRow, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
               goto FAIL;
             }
           }
@@ -1015,6 +1032,7 @@ int32_t tqUpdateTbUidList(STQ* pTq, const SArray* tbUidList, bool isAdd) {
   int32_t vgId = TD_VID(pTq->pVnode);
 
   // update the table list for each consumer handle
+  taosWLockLatch(&pTq->lock);
   while (1) {
     pIter = taosHashIterate(pTq->pHandle, pIter);
     if (pIter == NULL) {
@@ -1071,6 +1089,7 @@ int32_t tqUpdateTbUidList(STQ* pTq, const SArray* tbUidList, bool isAdd) {
       }
     }
   }
+  taosWUnLockLatch(&pTq->lock);
 
   // update the table list handle for each stream scanner/wal reader
   taosWLockLatch(&pTq->pStreamMeta->lock);

source/dnode/vnode/src/tq/tqRestore.c

@@ -120,13 +120,11 @@ int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle) {
         continue;
       }
 
-
-
       // append the data for the stream
       tqDebug("vgId:%d s-task:%s wal reader seek to ver:%" PRId64, vgId, pTask->id.idStr, pTask->chkInfo.currentVer);
     } else {
       int64_t currentVer = walReaderGetCurrentVer(pTask->exec.pWalReader);
-      if (currentVer != pTask->chkInfo.currentVer) {
+      if (currentVer == -1) {
         int32_t code = walReaderSeekVer(pTask->exec.pWalReader, pTask->chkInfo.currentVer);
         if (code != TSDB_CODE_SUCCESS) {  // no data in wal, quit
           streamMetaReleaseTask(pStreamMeta, pTask);
@@ -134,28 +132,27 @@ int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle) {
         }
 
         // append the data for the stream
-        tqDebug("vgId:%d s-task:%s wal reader seek to ver:%" PRId64, vgId, pTask->id.idStr, pTask->chkInfo.currentVer);
+        tqDebug("vgId:%d s-task:%s wal reader initial seek to ver:%" PRId64, vgId, pTask->id.idStr, pTask->chkInfo.currentVer);
       }
     }
 
-    SPackedData packData = {0};
+    // append the data for the stream
-    int32_t     code = extractSubmitMsgFromWal(pTask->exec.pWalReader, &packData);
+    SStreamQueueItem* pItem = NULL;
+    int32_t code = extractMsgFromWal(pTask->exec.pWalReader, (void**) &pItem, pTask->id.idStr);
     if (code != TSDB_CODE_SUCCESS) {  // failed, continue
       streamMetaReleaseTask(pStreamMeta, pTask);
       continue;
     }
 
-    SStreamDataSubmit2* p = streamDataSubmitNew(packData, STREAM_INPUT__DATA_SUBMIT);
+    // delete ignore
-    if (p == NULL) {
+    if (pItem == NULL) {
-      terrno = TSDB_CODE_OUT_OF_MEMORY;
-      tqError("%s failed to create data submit for stream since out of memory", pTask->id.idStr);
       streamMetaReleaseTask(pStreamMeta, pTask);
       continue;
     }
 
     noNewDataInWal = false;
 
-    code = tqAddInputBlockNLaunchTask(pTask, (SStreamQueueItem*)p, packData.ver);
+    code = tqAddInputBlockNLaunchTask(pTask, pItem);
     if (code == TSDB_CODE_SUCCESS) {
       pTask->chkInfo.currentVer = walReaderGetCurrentVer(pTask->exec.pWalReader);
       tqDebug("s-task:%s set the ver:%" PRId64 " from WALReader after extract block from WAL", pTask->id.idStr,
@@ -164,8 +161,7 @@ int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle) {
       tqError("s-task:%s append input queue failed, ver:%" PRId64, pTask->id.idStr, pTask->chkInfo.currentVer);
     }
 
-    streamDataSubmitDestroy(p);
+
-    taosFreeQitem(p);
     streamMetaReleaseTask(pStreamMeta, pTask);
   }
 

source/dnode/vnode/src/tq/tqScan.c

@@ -82,6 +82,7 @@ int32_t tqScanData(STQ* pTq, const STqHandle* pHandle, SMqDataRsp* pRsp, STqOffs
     SSDataBlock* pDataBlock = NULL;
     uint64_t     ts = 0;
     qStreamSetOpen(task);
+
     tqDebug("consumer:0x%" PRIx64 " vgId:%d, tmq one task start execute", pHandle->consumerId, vgId);
     if (qExecTask(task, &pDataBlock, &ts) != TSDB_CODE_SUCCESS) {
       tqError("consumer:0x%" PRIx64 " vgId:%d, task exec error since %s", pHandle->consumerId, vgId, terrstr());

source/dnode/vnode/src/tq/tqUtil.c

@@ -26,10 +26,10 @@ char* createStreamTaskIdStr(int64_t streamId, int32_t taskId) {
   return taosStrdup(buf);
 }
 
-int32_t tqAddInputBlockNLaunchTask(SStreamTask* pTask, SStreamQueueItem* pQueueItem, int64_t ver) {
+int32_t tqAddInputBlockNLaunchTask(SStreamTask* pTask, SStreamQueueItem* pQueueItem) {
   int32_t code = tAppendDataToInputQueue(pTask, pQueueItem);
   if (code < 0) {
-    tqError("s-task:%s failed to put into queue, too many, next start ver:%" PRId64, pTask->id.idStr, ver);
+    tqError("s-task:%s failed to put into queue, too many", pTask->id.idStr);
     return -1;
   }
 
@@ -162,8 +162,6 @@ static int32_t extractResetOffsetVal(STqOffsetVal* pOffsetVal, STQ* pTq, STqHand
   return 0;
 }
 
-bool tqIsHandleExecuting(STqHandle* pHandle) { return 1 == atomic_load_8(&pHandle->exec); }
-
 static int32_t extractDataAndRspForNormalSubscribe(STQ* pTq, STqHandle* pHandle, const SMqPollReq* pRequest,
                                                    SRpcMsg* pMsg, STqOffsetVal* pOffset) {
   uint64_t consumerId = pRequest->consumerId;
@@ -172,20 +170,6 @@ static int32_t extractDataAndRspForNormalSubscribe(STQ* pTq, STqHandle* pHandle,
 
   SMqDataRsp dataRsp = {0};
   tqInitDataRsp(&dataRsp, pRequest);
-//  qTaskInfo_t task = pHandle->execHandle.task;
-//  if (qTaskIsExecuting(task)) {
-//    code = tqSendDataRsp(pHandle, pMsg, pRequest, &dataRsp, TMQ_MSG_TYPE__POLL_RSP, vgId);
-//    tDeleteMqDataRsp(&dataRsp);
-//    return code;
-//  }
-
-  // todo add more status check to avoid race condition
-  while (tqIsHandleExecuting(pHandle)) {
-    tqDebug("vgId:%d, topic:%s, subscription is executing, wait for 5ms and retry", vgId, pHandle->subKey);
-    taosMsleep(5);
-  }
-
-  atomic_store_8(&pHandle->exec, 1);
 
   qSetTaskId(pHandle->execHandle.task, consumerId, pRequest->reqId);
   code = tqScanData(pTq, pHandle, &dataRsp, pOffset);
@@ -203,7 +187,6 @@ static int32_t extractDataAndRspForNormalSubscribe(STQ* pTq, STqHandle* pHandle,
       code = tqRegisterPushHandle(pTq, pHandle, pMsg);
       taosWUnLockLatch(&pTq->lock);
       tDeleteMqDataRsp(&dataRsp);
-      atomic_store_8(&pHandle->exec, 0);
       return code;
     } else {
       pHandle->noDataPollCnt++;
@@ -222,7 +205,6 @@ static int32_t extractDataAndRspForNormalSubscribe(STQ* pTq, STqHandle* pHandle,
     tDeleteMqDataRsp(&dataRsp);
   }
 
-  atomic_store_8(&pHandle->exec, 0);
   return code;
 }
 
@@ -233,19 +215,6 @@ static int32_t extractDataAndRspForDbStbSubscribe(STQ* pTq, STqHandle* pHandle,
   SMqMetaRsp metaRsp = {0};
   STaosxRsp taosxRsp = {0};
   tqInitTaosxRsp(&taosxRsp, pRequest);
-//  qTaskInfo_t task = pHandle->execHandle.task;
-//  if(qTaskIsExecuting(task)){
-//    code = tqSendDataRsp(pHandle, pMsg, pRequest, (SMqDataRsp*)&taosxRsp, TMQ_MSG_TYPE__TAOSX_RSP, vgId);
-//    tDeleteSTaosxRsp(&taosxRsp);
-//    return code;
-//  }
-
-  while (tqIsHandleExecuting(pHandle)) {
-    tqDebug("vgId:%d, topic:%s, subscription is executing, wait for 5ms and retry", vgId, pHandle->subKey);
-    taosMsleep(5);
-  }
-
-  atomic_store_8(&pHandle->exec, 1);
 
   if (offset->type != TMQ_OFFSET__LOG) {
     if (tqScanTaosx(pTq, pHandle, &taosxRsp, &metaRsp, offset) < 0) {
@@ -255,18 +224,22 @@ static int32_t extractDataAndRspForDbStbSubscribe(STQ* pTq, STqHandle* pHandle,
 
     if (metaRsp.metaRspLen > 0) {
       code = tqSendMetaPollRsp(pHandle, pMsg, pRequest, &metaRsp, vgId);
-      tqDebug("tmq poll: consumer:0x%" PRIx64 " subkey:%s vgId:%d, send meta offset type:%d,uid:%" PRId64 ",ts:%" PRId64,
+      tqDebug("tmq poll: consumer:0x%" PRIx64 " subkey:%s vgId:%d, send meta offset type:%d,uid:%" PRId64
-              pRequest->consumerId, pHandle->subKey, vgId, metaRsp.rspOffset.type, metaRsp.rspOffset.uid, metaRsp.rspOffset.ts);
+              ",ts:%" PRId64,
+              pRequest->consumerId, pHandle->subKey, vgId, metaRsp.rspOffset.type, metaRsp.rspOffset.uid,
+              metaRsp.rspOffset.ts);
       taosMemoryFree(metaRsp.metaRsp);
       goto end;
     }
 
     tqDebug("taosx poll: consumer:0x%" PRIx64 " subkey:%s vgId:%d, send data blockNum:%d, offset type:%d,uid:%" PRId64
-                ",ts:%" PRId64,pRequest->consumerId, pHandle->subKey, vgId, taosxRsp.blockNum, taosxRsp.rspOffset.type, taosxRsp.rspOffset.uid,taosxRsp.rspOffset.ts);
+            ",ts:%" PRId64,
+            pRequest->consumerId, pHandle->subKey, vgId, taosxRsp.blockNum, taosxRsp.rspOffset.type,
+            taosxRsp.rspOffset.uid, taosxRsp.rspOffset.ts);
     if (taosxRsp.blockNum > 0) {
       code = tqSendDataRsp(pHandle, pMsg, pRequest, (SMqDataRsp*)&taosxRsp, TMQ_MSG_TYPE__TAOSX_RSP, vgId);
       goto end;
-    }else {
+    } else {
       *offset = taosxRsp.rspOffset;
     }
   }
@@ -343,7 +316,6 @@ static int32_t extractDataAndRspForDbStbSubscribe(STQ* pTq, STqHandle* pHandle,
   }
 
 end:
-  atomic_store_8(&pHandle->exec, 0);
 
   tDeleteSTaosxRsp(&taosxRsp);
   taosMemoryFreeClear(pCkHead);

source/dnode/vnode/src/tsdb/tsdbCacheRead.c

@@ -120,6 +120,18 @@ static int32_t setTableSchema(SCacheRowsReader* p, uint64_t suid, const char* id
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t tsdbReuseCacherowsReader(void* reader, void* pTableIdList, int32_t numOfTables) {
+  SCacheRowsReader* pReader = (SCacheRowsReader*)reader;
+
+  pReader->pTableList = pTableIdList;
+  pReader->numOfTables = numOfTables;
+  pReader->lastTs = INT64_MIN;
+
+  resetLastBlockLoadInfo(pReader->pLoadInfo);
+
+  return TSDB_CODE_SUCCESS;
+}
+
 int32_t tsdbCacherowsReaderOpen(void* pVnode, int32_t type, void* pTableIdList, int32_t numOfTables, int32_t numOfCols,
                                 SArray* pCidList, int32_t* pSlotIds, uint64_t suid, void** pReader, const char* idstr) {
   *pReader = NULL;

source/dnode/vnode/src/tsdb/tsdbRead.c

@@ -210,6 +210,7 @@ struct STsdbReader {
   SBlockLoadSuppInfo suppInfo;
   STsdbReadSnap*     pReadSnap;
   SIOCostSummary     cost;
+  SHashObj**         pIgnoreTables;
   STSchema*          pSchema;      // the newest version schema
   SSHashObj*         pSchemaMap;   // keep the retrieved schema info, to avoid the overhead by repeatly load schema
   SDataFReader*      pFileReader;  // the file reader
@@ -2785,15 +2786,21 @@ static int32_t buildComposedDataBlock(STsdbReader* pReader) {
   int64_t st = taosGetTimestampUs();
   int32_t step = asc ? 1 : -1;
   double  el = 0;
+  SDataBlk* pBlock = getCurrentBlock(&pReader->status.blockIter);
+  SFileBlockDumpInfo* pDumpInfo = &pReader->status.fBlockDumpInfo;
 
   STableBlockScanInfo* pBlockScanInfo = NULL;
   if (pBlockInfo != NULL) {
+    if (pReader->pIgnoreTables && taosHashGet(*pReader->pIgnoreTables, &pBlockInfo->uid, sizeof(pBlockInfo->uid))) {
+      setBlockAllDumped(pDumpInfo, pBlock->maxKey.ts, pReader->order);
+      return code;
+    }
+    
     pBlockScanInfo = getTableBlockScanInfo(pReader->status.pTableMap, pBlockInfo->uid, pReader->idStr);
     if (pBlockScanInfo == NULL) {
       goto _end;
     }
 
-    SDataBlk* pBlock = getCurrentBlock(&pReader->status.blockIter);
     TSDBKEY   keyInBuf = getCurrentKeyInBuf(pBlockScanInfo, pReader);
 
     // it is a clean block, load it directly
@@ -2812,9 +2819,12 @@ static int32_t buildComposedDataBlock(STsdbReader* pReader) {
     }
   } else {  // file blocks not exist
     pBlockScanInfo = *pReader->status.pTableIter;
+    if (pReader->pIgnoreTables && taosHashGet(*pReader->pIgnoreTables, &pBlockScanInfo->uid, sizeof(pBlockScanInfo->uid))) {
+      setBlockAllDumped(pDumpInfo, pBlock->maxKey.ts, pReader->order);
+      return code;
+    }
   }
 
-  SFileBlockDumpInfo* pDumpInfo = &pReader->status.fBlockDumpInfo;
   SBlockData*         pBlockData = &pReader->status.fileBlockData;
 
   while (1) {
@@ -3101,6 +3111,14 @@ static int32_t doLoadLastBlockSequentially(STsdbReader* pReader) {
 
     // load the last data block of current table
     STableBlockScanInfo* pScanInfo = *(STableBlockScanInfo**)pStatus->pTableIter;
+    if (pReader->pIgnoreTables && taosHashGet(*pReader->pIgnoreTables, &pScanInfo->uid, sizeof(pScanInfo->uid))) {
+      bool hasNexTable = moveToNextTable(pUidList, pStatus);
+      if (!hasNexTable) {
+        return TSDB_CODE_SUCCESS;
+      }
+
+      continue;
+    }
 
     // reset the index in last block when handing a new file
     doCleanupTableScanInfo(pScanInfo);
@@ -3156,13 +3174,18 @@ static int32_t doLoadLastBlockSequentially(STsdbReader* pReader) {
 
 static int32_t doBuildDataBlock(STsdbReader* pReader) {
   int32_t   code = TSDB_CODE_SUCCESS;
-  SDataBlk* pBlock = NULL;
 
   SReaderStatus*       pStatus = &pReader->status;
   SDataBlockIter*      pBlockIter = &pStatus->blockIter;
   STableBlockScanInfo* pScanInfo = NULL;
   SFileDataBlockInfo*  pBlockInfo = getCurrentBlockInfo(pBlockIter);
   SLastBlockReader*    pLastBlockReader = pReader->status.fileIter.pLastBlockReader;
+  SDataBlk*            pBlock = getCurrentBlock(pBlockIter);
+
+  if (pReader->pIgnoreTables && taosHashGet(*pReader->pIgnoreTables, &pBlockInfo->uid, sizeof(pBlockInfo->uid))) {
+    setBlockAllDumped(&pStatus->fBlockDumpInfo, pBlock->maxKey.ts, pReader->order);
+    return code;
+  }
 
   if (pReader->code != TSDB_CODE_SUCCESS) {
     return pReader->code;
@@ -3173,7 +3196,6 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) {
     return terrno;
   }
 
-  pBlock = getCurrentBlock(pBlockIter);
 
   initLastBlockReader(pLastBlockReader, pScanInfo, pReader);
   TSDBKEY keyInBuf = getCurrentKeyInBuf(pScanInfo, pReader);
@@ -3406,6 +3428,13 @@ static int32_t buildBlockFromBufferSequentially(STsdbReader* pReader) {
     }
 
     STableBlockScanInfo** pBlockScanInfo = pStatus->pTableIter;
+    if (pReader->pIgnoreTables && taosHashGet(*pReader->pIgnoreTables, &(*pBlockScanInfo)->uid, sizeof((*pBlockScanInfo)->uid))) {
+      bool hasNexTable = moveToNextTable(pUidList, pStatus);
+      if (!hasNexTable) {
+        return TSDB_CODE_SUCCESS;
+      }
+    }
+    
     initMemDataIterator(*pBlockScanInfo, pReader);
 
     int64_t endKey = (ASCENDING_TRAVERSE(pReader->order)) ? INT64_MAX : INT64_MIN;
@@ -4360,7 +4389,7 @@ static void freeSchemaFunc(void* param) {
 
 // ====================================== EXPOSED APIs ======================================
 int32_t tsdbReaderOpen(SVnode* pVnode, SQueryTableDataCond* pCond, void* pTableList, int32_t numOfTables,
-                       SSDataBlock* pResBlock, STsdbReader** ppReader, const char* idstr, bool countOnly) {
+                       SSDataBlock* pResBlock, STsdbReader** ppReader, const char* idstr, bool countOnly, SHashObj** pIgnoreTables) {
   STimeWindow window = pCond->twindows;
 
   int32_t capacity = pVnode->config.tsdbCfg.maxRows;
@@ -4470,6 +4499,8 @@ int32_t tsdbReaderOpen(SVnode* pVnode, SQueryTableDataCond* pCond, void* pTableL
     pReader->readMode = READ_MODE_COUNT_ONLY;
   }
 
+  pReader->pIgnoreTables = pIgnoreTables;
+
   tsdbDebug("%p total numOfTable:%d in this query %s", pReader, numOfTables, pReader->idStr);
   return code;
 

source/libs/catalog/inc/catalogInt.h

@@ -86,6 +86,7 @@ enum {
 
 enum {
   CTG_OP_UPDATE_VGROUP = 0,
+  CTG_OP_UPDATE_DB_CFG,
   CTG_OP_UPDATE_TB_META,
   CTG_OP_DROP_DB_CACHE,
   CTG_OP_DROP_DB_VGROUP,
@@ -253,21 +254,27 @@ typedef struct SCtgVgCache {
   SDBVgInfo* vgInfo;
 } SCtgVgCache;
 
+typedef struct SCtgCfgCache {
+  SRWLatch    cfgLock;
+  SDbCfgInfo* cfgInfo;
+} SCtgCfgCache;
+
 typedef struct SCtgDBCache {
-  SRWLatch    dbLock;  // RC between destroy tbCache/stbCache and all reads
+  SRWLatch     dbLock;  // RC between destroy tbCache/stbCache and all reads
-  uint64_t    dbId;
+  uint64_t     dbId;
-  int8_t      deleted;
+  int8_t       deleted;
-  SCtgVgCache vgCache;
+  SCtgVgCache  vgCache;
-  SHashObj*   tbCache;   // key:tbname, value:SCtgTbCache
+  SCtgCfgCache cfgCache;
-  SHashObj*   stbCache;  // key:suid, value:char*
+  SHashObj*    tbCache;   // key:tbname, value:SCtgTbCache
-  uint64_t    dbCacheNum[CTG_CI_MAX_VALUE];
+  SHashObj*    stbCache;  // key:suid, value:char*
-  uint64_t    dbCacheSize;
+  uint64_t     dbCacheNum[CTG_CI_MAX_VALUE];
+  uint64_t     dbCacheSize;
 } SCtgDBCache;
 
 typedef struct SCtgRentSlot {
   SRWLatch lock;
   bool     needSort;
-  SArray*  meta;  // element is SDbVgVersion or SSTableVersion
+  SArray*  meta;  // element is SDbCacheInfo or SSTableVersion
 } SCtgRentSlot;
 
 typedef struct SCtgRentMgmt {
@@ -435,6 +442,13 @@ typedef struct SCtgUpdateVgMsg {
   SDBVgInfo* dbInfo;
 } SCtgUpdateVgMsg;
 
+typedef struct SCtgUpdateDbCfgMsg {
+  SCatalog*   pCtg;
+  char        dbFName[TSDB_DB_FNAME_LEN];
+  uint64_t    dbId;
+  SDbCfgInfo* cfgInfo;
+} SCtgUpdateDbCfgMsg;
+
 typedef struct SCtgUpdateTbMetaMsg {
   SCatalog*         pCtg;
   STableMetaOutput* pMeta;
@@ -847,8 +861,10 @@ int32_t ctgRemoveTbMetaFromCache(SCatalog* pCtg, SName* pTableName, bool syncReq
 int32_t ctgGetTbMetaFromCache(SCatalog* pCtg, SCtgTbMetaCtx* ctx, STableMeta** pTableMeta);
 int32_t ctgGetTbMetasFromCache(SCatalog* pCtg, SRequestConnInfo* pConn, SCtgTbMetasCtx* ctx, int32_t dbIdx,
                                int32_t* fetchIdx, int32_t baseResIdx, SArray* pList);
+void*   ctgCloneDbCfgInfo(void* pSrc);
 
 int32_t ctgOpUpdateVgroup(SCtgCacheOperation* action);
+int32_t ctgOpUpdateDbCfg(SCtgCacheOperation *operation);
 int32_t ctgOpUpdateTbMeta(SCtgCacheOperation* action);
 int32_t ctgOpDropDbCache(SCtgCacheOperation* action);
 int32_t ctgOpDropDbVgroup(SCtgCacheOperation* action);
@@ -870,6 +886,7 @@ int32_t ctgDropStbMetaEnqueue(SCatalog* pCtg, const char* dbFName, int64_t dbId,
                               bool syncReq);
 int32_t ctgDropTbMetaEnqueue(SCatalog* pCtg, const char* dbFName, int64_t dbId, const char* tbName, bool syncReq);
 int32_t ctgUpdateVgroupEnqueue(SCatalog* pCtg, const char* dbFName, int64_t dbId, SDBVgInfo* dbInfo, bool syncReq);
+int32_t ctgUpdateDbCfgEnqueue(SCatalog *pCtg, const char *dbFName, int64_t dbId, SDbCfgInfo *cfgInfo, bool syncOp);
 int32_t ctgUpdateTbMetaEnqueue(SCatalog* pCtg, STableMetaOutput* output, bool syncReq);
 int32_t ctgUpdateUserEnqueue(SCatalog* pCtg, SGetUserAuthRsp* pAuth, bool syncReq);
 int32_t ctgUpdateVgEpsetEnqueue(SCatalog* pCtg, char* dbFName, int32_t vgId, SEpSet* pEpSet);
@@ -941,9 +958,9 @@ int32_t ctgGetVgIdsFromHashValue(SCatalog* pCtg, SDBVgInfo* dbInfo, char* dbFNam
 void    ctgResetTbMetaTask(SCtgTask* pTask);
 void    ctgFreeDbCache(SCtgDBCache* dbCache);
 int32_t ctgStbVersionSortCompare(const void* key1, const void* key2);
-int32_t ctgDbVgVersionSortCompare(const void* key1, const void* key2);
+int32_t ctgDbCacheInfoSortCompare(const void* key1, const void* key2);
 int32_t ctgStbVersionSearchCompare(const void* key1, const void* key2);
-int32_t ctgDbVgVersionSearchCompare(const void* key1, const void* key2);
+int32_t ctgDbCacheInfoSearchCompare(const void* key1, const void* key2);
 void    ctgFreeSTableMetaOutput(STableMetaOutput* pOutput);
 int32_t ctgUpdateMsgCtx(SCtgMsgCtx* pCtx, int32_t reqType, void* out, char* target);
 int32_t ctgAddMsgCtx(SArray* pCtxs, int32_t reqType, void* out, char* target);
@@ -964,6 +981,9 @@ SName*  ctgGetFetchName(SArray* pNames, SCtgFetch* pFetch);
 int32_t ctgdGetOneHandle(SCatalog** pHandle);
 int     ctgVgInfoComp(const void* lp, const void* rp);
 int32_t ctgMakeVgArray(SDBVgInfo* dbInfo);
+int32_t ctgChkSetAuthRes(SCatalog *pCtg, SCtgAuthReq *req, SCtgAuthRsp* res);
+int32_t ctgReadDBCfgFromCache(SCatalog *pCtg, const char* dbFName, SDbCfgInfo* pDbCfg);
+
 int32_t ctgAcquireVgMetaFromCache(SCatalog* pCtg, const char* dbFName, const char* tbName, SCtgDBCache** pDb,
                                   SCtgTbCache** pTb);
 int32_t ctgCopyTbMeta(SCatalog* pCtg, SCtgTbMetaCtx* ctx, SCtgDBCache** pDb, SCtgTbCache** pTb, STableMeta** pTableMeta,

source/libs/catalog/src/catalog.c

@@ -698,6 +698,22 @@ _return:
   CTG_API_NLEAVE();
 }
 
+int32_t ctgGetDBCfg(SCatalog* pCtg, SRequestConnInfo* pConn, const char* dbFName, SDbCfgInfo* pDbCfg) {
+  CTG_ERR_RET(ctgReadDBCfgFromCache(pCtg, dbFName, pDbCfg));
+
+  if (pDbCfg->cfgVersion < 0) {
+    CTG_ERR_RET(ctgGetDBCfgFromMnode(pCtg, pConn, dbFName, pDbCfg, NULL));
+    SDbCfgInfo *pCfg = ctgCloneDbCfgInfo(pDbCfg);    
+    if (NULL == pCfg) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
+    
+    CTG_ERR_RET(ctgUpdateDbCfgEnqueue(pCtg, dbFName, pDbCfg->dbId, pCfg, false));
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
 
 int32_t catalogInit(SCatalogCfg* cfg) {
   if (gCtgMgmt.pCluster) {
@@ -1001,6 +1017,23 @@ _return:
   CTG_API_LEAVE(code);
 }
 
+int32_t catalogUpdateDbCfg(SCatalog* pCtg, const char* dbFName, uint64_t dbId, SDbCfgInfo* cfgInfo) {
+  CTG_API_ENTER();
+
+  int32_t code = 0;
+
+  if (NULL == pCtg || NULL == dbFName || NULL == cfgInfo) {
+    freeDbCfgInfo(cfgInfo);
+    CTG_ERR_JRET(TSDB_CODE_CTG_INVALID_INPUT);
+  }
+
+  code = ctgUpdateDbCfgEnqueue(pCtg, dbFName, dbId, cfgInfo, false);
+
+_return:
+
+  CTG_API_LEAVE(code);
+}
+
 int32_t catalogRemoveDB(SCatalog* pCtg, const char* dbFName, uint64_t dbId) {
   CTG_API_ENTER();
 
@@ -1430,14 +1463,14 @@ int32_t catalogGetExpiredSTables(SCatalog* pCtg, SSTableVersion** stables, uint3
   CTG_API_LEAVE(ctgMetaRentGet(&pCtg->stbRent, (void**)stables, num, sizeof(SSTableVersion)));
 }
 
-int32_t catalogGetExpiredDBs(SCatalog* pCtg, SDbVgVersion** dbs, uint32_t* num) {
+int32_t catalogGetExpiredDBs(SCatalog* pCtg, SDbCacheInfo** dbs, uint32_t* num) {
   CTG_API_ENTER();
 
   if (NULL == pCtg || NULL == dbs || NULL == num) {
     CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
   }
 
-  CTG_API_LEAVE(ctgMetaRentGet(&pCtg->dbRent, (void**)dbs, num, sizeof(SDbVgVersion)));
+  CTG_API_LEAVE(ctgMetaRentGet(&pCtg->dbRent, (void**)dbs, num, sizeof(SDbCacheInfo)));
 }
 
 int32_t catalogGetExpiredUsers(SCatalog* pCtg, SUserAuthVersion** users, uint32_t* num) {
@@ -1485,9 +1518,7 @@ int32_t catalogGetDBCfg(SCatalog* pCtg, SRequestConnInfo* pConn, const char* dbF
     CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
   }
 
-  CTG_CACHE_NHIT_INC(CTG_CI_DB_CFG, 1);
+  CTG_API_LEAVE(ctgGetDBCfg(pCtg, pConn, dbFName, pDbCfg));
-
-  CTG_API_LEAVE(ctgGetDBCfgFromMnode(pCtg, pConn, dbFName, pDbCfg, NULL));
 }
 
 int32_t catalogGetIndexMeta(SCatalog* pCtg, SRequestConnInfo* pConn, const char* indexName, SIndexInfo* pInfo) {

source/libs/catalog/src/ctgAsync.c

@@ -1587,7 +1587,6 @@ int32_t ctgHandleGetTbIndexRsp(SCtgTaskReq* tReq, int32_t reqType, const SDataBu
   CTG_ERR_JRET(ctgCloneTableIndex(pOut->pIndex, &pInfo));
   pTask->res = pInfo;
 
-  SCtgTbIndexCtx* ctx = pTask->taskCtx;
   CTG_ERR_JRET(ctgUpdateTbIndexEnqueue(pTask->pJob->pCtg, (STableIndex**)&pTask->msgCtx.out, false));
 
 _return:
@@ -1660,8 +1659,14 @@ _return:
 int32_t ctgHandleGetDbCfgRsp(SCtgTaskReq* tReq, int32_t reqType, const SDataBuf* pMsg, int32_t rspCode) {
   int32_t   code = 0;
   SCtgTask* pTask = tReq->pTask;
+  SCtgDbCfgCtx* ctx = pTask->taskCtx;
+  
   CTG_ERR_JRET(ctgProcessRspMsg(pTask->msgCtx.out, reqType, pMsg->pData, pMsg->len, rspCode, pTask->msgCtx.target));
 
+  SDbCfgInfo* pCfg = ctgCloneDbCfgInfo(pTask->msgCtx.out);
+
+  CTG_ERR_RET(ctgUpdateDbCfgEnqueue(pTask->pJob->pCtg, ctx->dbFName, pCfg->dbId, pCfg, false));
+
   TSWAP(pTask->res, pTask->msgCtx.out);
 
 _return:
@@ -2195,13 +2200,24 @@ int32_t ctgLaunchGetDbCfgTask(SCtgTask* pTask) {
   SCtgDbCfgCtx*     pCtx = (SCtgDbCfgCtx*)pTask->taskCtx;
   SCtgJob*          pJob = pTask->pJob;
   SCtgMsgCtx*       pMsgCtx = CTG_GET_TASK_MSGCTX(pTask, -1);
+  SDbCfgInfo        cfgInfo;
   if (NULL == pMsgCtx->pBatchs) {
     pMsgCtx->pBatchs = pJob->pBatchs;
   }
 
-  CTG_CACHE_NHIT_INC(CTG_CI_DB_CFG, 1);
+  CTG_ERR_RET(ctgReadDBCfgFromCache(pCtg, pCtx->dbFName, &cfgInfo));
 
-  CTG_ERR_RET(ctgGetDBCfgFromMnode(pCtg, pConn, pCtx->dbFName, NULL, pTask));
+  if (cfgInfo.cfgVersion < 0) {
+    CTG_ERR_RET(ctgGetDBCfgFromMnode(pCtg, pConn, pCtx->dbFName, NULL, pTask));
+  } else {
+    pTask->res = taosMemoryCalloc(1, sizeof(SDbCfgInfo));
+    if (NULL == pTask->res) {
+      CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
+    }
+
+    memcpy(pTask->res, &cfgInfo, sizeof(cfgInfo));
+    CTG_ERR_RET(ctgHandleTaskEnd(pTask, 0));
+  }
 
   return TSDB_CODE_SUCCESS;
 }

source/libs/catalog/src/ctgCache.c

@@ -20,6 +20,7 @@
 #include "trpc.h"
 
 SCtgOperation gCtgCacheOperation[CTG_OP_MAX] = {{CTG_OP_UPDATE_VGROUP, "update vgInfo", ctgOpUpdateVgroup},
+                                                {CTG_OP_UPDATE_DB_CFG, "update dbCfg", ctgOpUpdateDbCfg},
                                                 {CTG_OP_UPDATE_TB_META, "update tbMeta", ctgOpUpdateTbMeta},
                                                 {CTG_OP_DROP_DB_CACHE, "drop DB", ctgOpDropDbCache},
                                                 {CTG_OP_DROP_DB_VGROUP, "drop DBVgroup", ctgOpDropDbVgroup},
@@ -90,10 +91,15 @@ int32_t ctgWLockVgInfo(SCatalog *pCtg, SCtgDBCache *dbCache) {
   return TSDB_CODE_SUCCESS;
 }
 
-void ctgRUnlockVgInfo(SCtgDBCache *dbCache) { CTG_UNLOCK(CTG_READ, &dbCache->vgCache.vgLock); }
+void ctgRLockDbCfgInfo(SCtgDBCache *dbCache) {  CTG_LOCK(CTG_READ, &dbCache->cfgCache.cfgLock); }
+void ctgWLockDbCfgInfo(SCtgDBCache *dbCache) {  CTG_LOCK(CTG_WRITE, &dbCache->cfgCache.cfgLock); }
 
+void ctgRUnlockVgInfo(SCtgDBCache *dbCache) { CTG_UNLOCK(CTG_READ, &dbCache->vgCache.vgLock); }
 void ctgWUnlockVgInfo(SCtgDBCache *dbCache) { CTG_UNLOCK(CTG_WRITE, &dbCache->vgCache.vgLock); }
 
+void ctgRUnlockDbCfgInfo(SCtgDBCache *dbCache) { CTG_UNLOCK(CTG_READ, &dbCache->cfgCache.cfgLock); }
+void ctgWUnlockDbCfgInfo(SCtgDBCache *dbCache) { CTG_UNLOCK(CTG_WRITE, &dbCache->cfgCache.cfgLock); }
+
 void ctgReleaseDBCache(SCatalog *pCtg, SCtgDBCache *dbCache) {
   CTG_UNLOCK(CTG_READ, &dbCache->dbLock);
   taosHashRelease(pCtg->dbCache, dbCache);
@@ -704,6 +710,43 @@ _return:
   CTG_RET(code);
 }
 
+int32_t ctgReadDBCfgFromCache(SCatalog *pCtg, const char* dbFName, SDbCfgInfo* pDbCfg) {
+  int32_t code = 0;
+  SCtgDBCache *dbCache = NULL;
+  ctgAcquireDBCache(pCtg, dbFName, &dbCache);
+  if (NULL == dbCache) {
+    ctgDebug("db %s not in cache", dbFName);
+    pDbCfg->cfgVersion = -1;
+    CTG_CACHE_NHIT_INC(CTG_CI_DB_CFG, 1);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  CTG_LOCK(CTG_READ, &dbCache->cfgCache.cfgLock);
+
+  if (dbCache->cfgCache.cfgInfo) {
+    SDbCfgInfo *pInfo = ctgCloneDbCfgInfo(dbCache->cfgCache.cfgInfo);
+    if (NULL == pInfo) {
+      CTG_ERR_JRET(TSDB_CODE_OUT_OF_MEMORY);
+    }
+    
+    memcpy(pDbCfg, pInfo, sizeof(*pInfo));
+    taosMemoryFree(pInfo);
+    CTG_CACHE_HIT_INC(CTG_CI_DB_CFG, 1);
+  } else {
+    pDbCfg->cfgVersion = -1;
+    CTG_CACHE_NHIT_INC(CTG_CI_DB_CFG, 1);
+  }
+
+_return:
+
+  if (dbCache) {
+    CTG_UNLOCK(CTG_READ, &dbCache->cfgCache.cfgLock);
+    ctgReleaseDBCache(pCtg, dbCache);
+  }
+  
+  return code;
+}
+
 int32_t ctgGetCachedStbNameFromSuid(SCatalog* pCtg, char* dbFName, uint64_t suid, char **stbName) {
   *stbName = NULL;
 
@@ -727,6 +770,7 @@ int32_t ctgGetCachedStbNameFromSuid(SCatalog* pCtg, char* dbFName, uint64_t suid
   return TSDB_CODE_SUCCESS;
 }
 
+
 int32_t ctgChkAuthFromCache(SCatalog *pCtg, SUserAuthInfo *pReq, bool *inCache, SCtgAuthRsp *pRes) {
   int32_t code = 0;
   if (IS_SYS_DBNAME(pReq->tbName.dbname)) {
@@ -1007,6 +1051,44 @@ _return:
   CTG_RET(code);
 }
 
+
+int32_t ctgUpdateDbCfgEnqueue(SCatalog *pCtg, const char *dbFName, int64_t dbId, SDbCfgInfo *cfgInfo, bool syncOp) {
+  int32_t             code = 0;
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_UPDATE_DB_CFG;
+  op->syncOp = syncOp;
+
+  SCtgUpdateDbCfgMsg *msg = taosMemoryMalloc(sizeof(SCtgUpdateDbCfgMsg));
+  if (NULL == msg) {
+    ctgError("malloc %d failed", (int32_t)sizeof(SCtgUpdateDbCfgMsg));
+    taosMemoryFree(op);
+    freeDbCfgInfo(cfgInfo);
+    CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
+  }
+
+  char *p = strchr(dbFName, '.');
+  if (p && IS_SYS_DBNAME(p + 1)) {
+    dbFName = p + 1;
+  }
+
+  tstrncpy(msg->dbFName, dbFName, sizeof(msg->dbFName));
+  msg->pCtg = pCtg;
+  msg->dbId = dbId;
+  msg->cfgInfo = cfgInfo;
+
+  op->data = msg;
+
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
+
+  return TSDB_CODE_SUCCESS;
+
+_return:
+
+  freeDbCfgInfo(cfgInfo);
+  CTG_RET(code);
+}
+
+
 int32_t ctgUpdateTbMetaEnqueue(SCatalog *pCtg, STableMetaOutput *output, bool syncOp) {
   int32_t             code = 0;
   SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
@@ -1426,15 +1508,15 @@ int32_t ctgAddNewDBCache(SCatalog *pCtg, const char *dbFName, uint64_t dbId) {
 
   CTG_CACHE_NUM_INC(CTG_CI_DB, 1);
 
-  SDbVgVersion vgVersion = {.dbId = newDBCache.dbId, .vgVersion = -1, .stateTs = 0};
+  SDbCacheInfo dbCacheInfo = {.dbId = newDBCache.dbId, .vgVersion = -1, .stateTs = 0, .cfgVersion = -1};
-  tstrncpy(vgVersion.dbFName, dbFName, sizeof(vgVersion.dbFName));
+  tstrncpy(dbCacheInfo.dbFName, dbFName, sizeof(dbCacheInfo.dbFName));
 
   ctgDebug("db added to cache, dbFName:%s, dbId:0x%" PRIx64, dbFName, dbId);
 
   if (!IS_SYS_DBNAME(dbFName)) {
-    CTG_ERR_RET(ctgMetaRentAdd(&pCtg->dbRent, &vgVersion, dbId, sizeof(SDbVgVersion)));
+    CTG_ERR_RET(ctgMetaRentAdd(&pCtg->dbRent, &dbCacheInfo, dbId, sizeof(SDbCacheInfo)));
 
-    ctgDebug("db added to rent, dbFName:%s, vgVersion:%d, dbId:0x%" PRIx64, dbFName, vgVersion.vgVersion, dbId);
+    ctgDebug("db added to rent, dbFName:%s, vgVersion:%d, dbId:0x%" PRIx64, dbFName, dbCacheInfo.vgVersion, dbId);
   }
 
   return TSDB_CODE_SUCCESS;
@@ -1478,7 +1560,7 @@ int32_t ctgRemoveDBFromCache(SCatalog *pCtg, SCtgDBCache *dbCache, const char *d
 
   CTG_UNLOCK(CTG_WRITE, &dbCache->dbLock);
 
-  CTG_ERR_RET(ctgMetaRentRemove(&pCtg->dbRent, dbId, ctgDbVgVersionSortCompare, ctgDbVgVersionSearchCompare));
+  CTG_ERR_RET(ctgMetaRentRemove(&pCtg->dbRent, dbId, ctgDbCacheInfoSortCompare, ctgDbCacheInfoSearchCompare));
   ctgDebug("db removed from rent, dbFName:%s, dbId:0x%" PRIx64, dbFName, dbId);
 
   if (taosHashRemove(pCtg->dbCache, dbFName, strlen(dbFName))) {
@@ -1791,8 +1873,8 @@ int32_t ctgOpUpdateVgroup(SCtgCacheOperation *operation) {
   }
 
   bool         newAdded = false;
-  SDbVgVersion vgVersion = {
+  SDbCacheInfo dbCacheInfo = {
-      .dbId = msg->dbId, .vgVersion = dbInfo->vgVersion, .numOfTable = dbInfo->numOfTable, .stateTs = dbInfo->stateTs};
+      .dbId = msg->dbId, .vgVersion = dbInfo->vgVersion, .cfgVersion = -1, .numOfTable = dbInfo->numOfTable, .stateTs = dbInfo->stateTs};
 
   SCtgDBCache *dbCache = NULL;
   CTG_ERR_JRET(ctgGetAddDBCache(msg->pCtg, dbFName, msg->dbId, &dbCache));
@@ -1834,12 +1916,16 @@ int32_t ctgOpUpdateVgroup(SCtgCacheOperation *operation) {
     CTG_DB_NUM_RESET(CTG_CI_DB_VGROUP);
   }
 
+  if (dbCache->cfgCache.cfgInfo) {
+    dbCacheInfo.cfgVersion = dbCache->cfgCache.cfgInfo->cfgVersion;
+  }
+
   vgCache->vgInfo = dbInfo;
   msg->dbInfo = NULL;
   CTG_DB_NUM_SET(CTG_CI_DB_VGROUP);
 
   ctgDebug("db vgInfo updated, dbFName:%s, vgVer:%d, stateTs:%" PRId64 ", dbId:0x%" PRIx64, dbFName,
-           vgVersion.vgVersion, vgVersion.stateTs, vgVersion.dbId);
+           dbCacheInfo.vgVersion, dbCacheInfo.stateTs, dbCacheInfo.dbId);
 
   ctgWUnlockVgInfo(dbCache);
 
@@ -1850,9 +1936,9 @@ int32_t ctgOpUpdateVgroup(SCtgCacheOperation *operation) {
   dbCache = NULL;
 
   // if (!IS_SYS_DBNAME(dbFName)) {
-  tstrncpy(vgVersion.dbFName, dbFName, sizeof(vgVersion.dbFName));
+  tstrncpy(dbCacheInfo.dbFName, dbFName, sizeof(dbCacheInfo.dbFName));
-  CTG_ERR_JRET(ctgMetaRentUpdate(&msg->pCtg->dbRent, &vgVersion, vgVersion.dbId, sizeof(SDbVgVersion),
+  CTG_ERR_JRET(ctgMetaRentUpdate(&msg->pCtg->dbRent, &dbCacheInfo, dbCacheInfo.dbId, sizeof(SDbCacheInfo),
-                                 ctgDbVgVersionSortCompare, ctgDbVgVersionSearchCompare));
+                                 ctgDbCacheInfoSortCompare, ctgDbCacheInfoSearchCompare));
   //}
 
 _return:
@@ -1863,6 +1949,67 @@ _return:
   CTG_RET(code);
 }
 
+int32_t ctgOpUpdateDbCfg(SCtgCacheOperation *operation) {
+  int32_t          code = 0;
+  SCtgUpdateDbCfgMsg *msg = operation->data;
+  SDbCfgInfo       *cfgInfo = msg->cfgInfo;
+  char            *dbFName = msg->dbFName;
+  SCatalog        *pCtg = msg->pCtg;
+
+  if (pCtg->stopUpdate || NULL == cfgInfo) {
+    goto _return;
+  }
+
+  if (cfgInfo->cfgVersion < 0) {
+    ctgDebug("invalid db cfgInfo, dbFName:%s, cfgVersion:%d", dbFName, cfgInfo->cfgVersion);
+    CTG_ERR_JRET(TSDB_CODE_APP_ERROR);
+  }
+
+  SCtgDBCache *dbCache = NULL;
+  CTG_ERR_JRET(ctgGetAddDBCache(msg->pCtg, dbFName, msg->dbId, &dbCache));
+  if (NULL == dbCache) {
+    ctgInfo("conflict db update, ignore this update, dbFName:%s, dbId:0x%" PRIx64, dbFName, msg->dbId);
+    CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);
+  }
+
+  SDbCacheInfo cacheInfo = {0};
+  cacheInfo.dbId = dbCache->dbId;
+  tstrncpy(cacheInfo.dbFName, dbFName, sizeof(cacheInfo.dbFName));
+  cacheInfo.cfgVersion = cfgInfo->cfgVersion;
+
+  SCtgVgCache *vgCache = &dbCache->vgCache;
+  if (vgCache->vgInfo) {
+    cacheInfo.vgVersion = vgCache->vgInfo->vgVersion;
+    cacheInfo.numOfTable = vgCache->vgInfo->numOfTable;
+    cacheInfo.stateTs = vgCache->vgInfo->stateTs;
+  } else {
+    cacheInfo.vgVersion = -1;
+  }
+
+  ctgWLockDbCfgInfo(dbCache);
+
+  freeDbCfgInfo(dbCache->cfgCache.cfgInfo);
+  dbCache->cfgCache.cfgInfo = cfgInfo;
+  cfgInfo = NULL;
+  
+  ctgWUnlockDbCfgInfo(dbCache);
+
+  ctgDebug("db cfgInfo updated, dbFName:%s, cfgVer:%d", dbFName, dbCache->cfgCache.cfgInfo->cfgVersion);
+
+  // if (!IS_SYS_DBNAME(dbFName)) {
+  CTG_ERR_JRET(ctgMetaRentUpdate(&msg->pCtg->dbRent, &cacheInfo, cacheInfo.dbId, sizeof(SDbCacheInfo),
+                                 ctgDbCacheInfoSortCompare, ctgDbCacheInfoSearchCompare));
+  //}
+
+_return:
+
+  freeDbCfgInfo(cfgInfo);
+  taosMemoryFreeClear(msg);
+
+  CTG_RET(code);
+}
+
+
 int32_t ctgOpDropDbCache(SCtgCacheOperation *operation) {
   int32_t        code = 0;
   SCtgDropDBMsg *msg = operation->data;
@@ -2124,27 +2271,19 @@ int32_t ctgOpUpdateUser(SCtgCacheOperation *operation) {
   CTG_LOCK(CTG_WRITE, &pUser->lock);
 
   taosHashCleanup(pUser->userAuth.createdDbs);
-  pUser->userAuth.createdDbs = msg->userAuth.createdDbs;
-  msg->userAuth.createdDbs = NULL;
-
   taosHashCleanup(pUser->userAuth.readDbs);
-  pUser->userAuth.readDbs = msg->userAuth.readDbs;
-  msg->userAuth.readDbs = NULL;
-
   taosHashCleanup(pUser->userAuth.writeDbs);
-  pUser->userAuth.writeDbs = msg->userAuth.writeDbs;
-  msg->userAuth.writeDbs = NULL;
-
   taosHashCleanup(pUser->userAuth.readTbs);
-  pUser->userAuth.readTbs = msg->userAuth.readTbs;
-  msg->userAuth.readTbs = NULL;
-
   taosHashCleanup(pUser->userAuth.writeTbs);
-  pUser->userAuth.writeTbs = msg->userAuth.writeTbs;
-  msg->userAuth.writeTbs = NULL;
-
   taosHashCleanup(pUser->userAuth.useDbs);
-  pUser->userAuth.useDbs = msg->userAuth.useDbs;
+
+  memcpy(&pUser->userAuth, &msg->userAuth, sizeof(msg->userAuth));
+
+  msg->userAuth.createdDbs = NULL;
+  msg->userAuth.readDbs = NULL;
+  msg->userAuth.writeDbs = NULL;
+  msg->userAuth.readTbs = NULL;
+  msg->userAuth.writeTbs = NULL;
   msg->userAuth.useDbs = NULL;
 
   CTG_UNLOCK(CTG_WRITE, &pUser->lock);

source/libs/catalog/src/ctgDbg.c

@@ -501,6 +501,25 @@ void ctgdShowDBCache(SCatalog *pCtg, SHashObj *dbHash) {
       }
     }
 
+    if (dbCache->cfgCache.cfgInfo) {
+      SDbCfgInfo *pCfg = dbCache->cfgCache.cfgInfo;
+      ctgDebug("[%d] db [%.*s][0x%" PRIx64
+               "] %s: cfgVersion:%d, numOfVgroups:%d, numOfStables:%d, buffer:%d, cacheSize:%d, pageSize:%d, pages:%d"
+               ", daysPerFile:%d, daysToKeep0:%d, daysToKeep1:%d, daysToKeep2:%d, minRows:%d, maxRows:%d, walFsyncPeriod:%d"
+               ", hashPrefix:%d, hashSuffix:%d, walLevel:%d, precision:%d, compression:%d, replications:%d, strict:%d"
+               ", cacheLast:%d, tsdbPageSize:%d, walRetentionPeriod:%d, walRollPeriod:%d, walRetentionSize:%" PRId64 ""
+               ", walSegmentSize:%" PRId64 ", numOfRetensions:%d, schemaless:%d, sstTrigger:%d",
+               i, (int32_t)len, dbFName, dbCache->dbId, dbCache->deleted ? "deleted" : "", 
+               pCfg->cfgVersion, pCfg->numOfVgroups, pCfg->numOfStables, pCfg->buffer,
+               pCfg->cacheSize, pCfg->pageSize, pCfg->pages, pCfg->daysPerFile, pCfg->daysToKeep0,
+               pCfg->daysToKeep1, pCfg->daysToKeep2, pCfg->minRows, pCfg->maxRows, pCfg->walFsyncPeriod,
+               pCfg->hashPrefix, pCfg->hashSuffix, pCfg->walLevel, pCfg->precision, pCfg->compression,
+               pCfg->replications, pCfg->strict, pCfg->cacheLast, pCfg->tsdbPageSize, pCfg->walRetentionPeriod,
+               pCfg->walRollPeriod, pCfg->walRetentionSize, pCfg->walSegmentSize, pCfg->numOfRetensions,
+               pCfg->schemaless, pCfg->sstTrigger);
+    }
+
+    ++i;
     pIter = taosHashIterate(dbHash, pIter);
   }
 }

source/libs/catalog/src/ctgUtil.c

@@ -251,6 +251,7 @@ void ctgFreeTbCache(SCtgDBCache* dbCache) {
 }
 
 void ctgFreeVgInfoCache(SCtgDBCache* dbCache) { freeVgInfo(dbCache->vgCache.vgInfo); }
+void ctgFreeCfgInfoCache(SCtgDBCache* dbCache) { freeDbCfgInfo(dbCache->cfgCache.cfgInfo); }
 
 void ctgFreeDbCache(SCtgDBCache* dbCache) {
   if (NULL == dbCache) {
@@ -258,6 +259,7 @@ void ctgFreeDbCache(SCtgDBCache* dbCache) {
   }
 
   ctgFreeVgInfoCache(dbCache);
+  ctgFreeCfgInfoCache(dbCache);
   ctgFreeStbMetaCache(dbCache);
   ctgFreeTbCache(dbCache);
 }
@@ -1153,10 +1155,10 @@ int32_t ctgStbVersionSearchCompare(const void* key1, const void* key2) {
   }
 }
 
-int32_t ctgDbVgVersionSearchCompare(const void* key1, const void* key2) {
+int32_t ctgDbCacheInfoSearchCompare(const void* key1, const void* key2) {
-  if (*(int64_t*)key1 < ((SDbVgVersion*)key2)->dbId) {
+  if (*(int64_t*)key1 < ((SDbCacheInfo*)key2)->dbId) {
     return -1;
-  } else if (*(int64_t*)key1 > ((SDbVgVersion*)key2)->dbId) {
+  } else if (*(int64_t*)key1 > ((SDbCacheInfo*)key2)->dbId) {
     return 1;
   } else {
     return 0;
@@ -1173,10 +1175,10 @@ int32_t ctgStbVersionSortCompare(const void* key1, const void* key2) {
   }
 }
 
-int32_t ctgDbVgVersionSortCompare(const void* key1, const void* key2) {
+int32_t ctgDbCacheInfoSortCompare(const void* key1, const void* key2) {
-  if (((SDbVgVersion*)key1)->dbId < ((SDbVgVersion*)key2)->dbId) {
+  if (((SDbCacheInfo*)key1)->dbId < ((SDbCacheInfo*)key2)->dbId) {
     return -1;
-  } else if (((SDbVgVersion*)key1)->dbId > ((SDbVgVersion*)key2)->dbId) {
+  } else if (((SDbCacheInfo*)key1)->dbId > ((SDbCacheInfo*)key2)->dbId) {
     return 1;
   } else {
     return 0;
@@ -1347,16 +1349,20 @@ static void* ctgCloneDbVgroup(void* pSrc) { return taosArrayDup((const SArray*)p
 
 static void ctgFreeDbVgroup(void* p) { taosArrayDestroy((SArray*)((SMetaRes*)p)->pRes); }
 
-static void* ctgCloneDbCfgInfo(void* pSrc) {
+void* ctgCloneDbCfgInfo(void* pSrc) {
   SDbCfgInfo* pDst = taosMemoryMalloc(sizeof(SDbCfgInfo));
   if (NULL == pDst) {
     return NULL;
   }
   memcpy(pDst, pSrc, sizeof(SDbCfgInfo));
+  pDst->pRetensions = taosArrayDup(((SDbCfgInfo *)pSrc)->pRetensions, NULL);
   return pDst;
 }
 
-static void ctgFreeDbCfgInfo(void* p) { taosMemoryFree(((SMetaRes*)p)->pRes); }
+static void ctgFreeDbCfgInfo(void* p) { 
+  SDbCfgInfo* pDst = (SDbCfgInfo *)((SMetaRes*)p)->pRes;
+  freeDbCfgInfo(pDst);
+}
 
 static void* ctgCloneDbInfo(void* pSrc) {
   SDbInfo* pDst = taosMemoryMalloc(sizeof(SDbInfo));

source/libs/catalog/test/catalogTests.cpp

@@ -1331,7 +1331,7 @@ TEST(tableMeta, normalTable) {
   ASSERT_EQ(tableMeta->tableInfo.precision, 1);
   ASSERT_EQ(tableMeta->tableInfo.rowSize, 12);
 
-  SDbVgVersion   *dbs = NULL;
+  SDbCacheInfo   *dbs = NULL;
   SSTableVersion *stb = NULL;
   uint32_t        dbNum = 0, stbNum = 0, allDbNum = 0, allStbNum = 0;
   int32_t         i = 0;
@@ -1443,7 +1443,7 @@ TEST(tableMeta, childTableCase) {
 
   taosMemoryFree(tableMeta);
 
-  SDbVgVersion   *dbs = NULL;
+  SDbCacheInfo   *dbs = NULL;
   SSTableVersion *stb = NULL;
   uint32_t        dbNum = 0, stbNum = 0, allDbNum = 0, allStbNum = 0;
   int32_t         i = 0;
@@ -1584,7 +1584,7 @@ TEST(tableMeta, superTableCase) {
 
   taosMemoryFree(tableMeta);
 
-  SDbVgVersion   *dbs = NULL;
+  SDbCacheInfo   *dbs = NULL;
   SSTableVersion *stb = NULL;
   uint32_t        dbNum = 0, stbNum = 0, allDbNum = 0, allStbNum = 0;
   int32_t         i = 0;
@@ -2682,7 +2682,7 @@ TEST(rentTest, allRent) {
   SDBVgInfo         dbVgroup = {0};
   SArray           *vgList = NULL;
   ctgTestStop = false;
-  SDbVgVersion   *dbs = NULL;
+  SDbCacheInfo   *dbs = NULL;
   SSTableVersion *stable = NULL;
   uint32_t        num = 0;
 

source/libs/executor/inc/executil.h

@@ -86,8 +86,25 @@ typedef struct SColMatchInfo {
   int32_t matchType;  // determinate the source according to col id or slot id
 } SColMatchInfo;
 
-typedef struct SExecTaskInfo  SExecTaskInfo;
+typedef struct SExecTaskInfo SExecTaskInfo;
-typedef struct STableListInfo STableListInfo;
+
+typedef struct STableListIdInfo {
+  uint64_t suid;
+  uint64_t uid;
+  int32_t  tableType;
+} STableListIdInfo;
+
+// If the numOfOutputGroups is 1, the data blocks that belongs to different groups will be provided randomly
+// The numOfOutputGroups is specified by physical plan. and will not be affect by numOfGroups
+typedef struct STableListInfo {
+  bool             oneTableForEachGroup;
+  int32_t          numOfOuputGroups;  // the data block will be generated one by one
+  int32_t*         groupOffset;       // keep the offset value for each group in the tableList
+  SArray*          pTableList;
+  SHashObj*        map;     // speedup acquire the tableQueryInfo by table uid
+  STableListIdInfo idInfo;  // this maybe the super table or ordinary table
+} STableListInfo;
+
 struct SqlFunctionCtx;
 
 int32_t createScanTableListInfo(SScanPhysiNode* pScanNode, SNodeList* pGroupTags, bool groupSort, SReadHandle* pHandle,

source/libs/executor/inc/executorInt.h

@@ -213,6 +213,7 @@ typedef struct STableScanInfo {
   SScanInfo       scanInfo;
   int32_t         scanTimes;
   SSDataBlock*    pResBlock;
+  SHashObj*       pIgnoreTables;
   SSampleExecInfo sample;  // sample execution info
   int32_t         currentGroupId;
   int32_t         currentTable;

source/libs/executor/src/cachescanoperator.c

@@ -238,13 +238,17 @@ SSDataBlock* doScanCache(SOperatorInfo* pOperator) {
         T_LONG_JMP(pTaskInfo->env, code);
       }
 
-      code = tsdbCacherowsReaderOpen(pInfo->readHandle.vnode, pInfo->retrieveType, pList, num,
+      if (NULL == pInfo->pLastrowReader) {
-                                     taosArrayGetSize(pInfo->matchInfo.pList), pInfo->pCidList, pInfo->pSlotIds, suid,
+        code = tsdbCacherowsReaderOpen(pInfo->readHandle.vnode, pInfo->retrieveType, pList, num,
-                                     &pInfo->pLastrowReader, pTaskInfo->id.str);
+                                       taosArrayGetSize(pInfo->matchInfo.pList), pInfo->pCidList, pInfo->pSlotIds, suid, &pInfo->pLastrowReader,
-      if (code != TSDB_CODE_SUCCESS) {
+                                       pTaskInfo->id.str);
-        pInfo->currentGroupIndex += 1;
+        if (code != TSDB_CODE_SUCCESS) {
-        taosArrayClear(pInfo->pUidList);
+          pInfo->currentGroupIndex += 1;
-        continue;
+          taosArrayClear(pInfo->pUidList);
+          continue;
+        }
+      } else {
+        tsdbReuseCacherowsReader(pInfo->pLastrowReader, pList, num);
       }
 
       taosArrayClear(pInfo->pUidList);
@@ -276,13 +280,14 @@ SSDataBlock* doScanCache(SOperatorInfo* pOperator) {
           }
         }
 
-        pInfo->pLastrowReader = tsdbCacherowsReaderClose(pInfo->pLastrowReader);
+        //pInfo->pLastrowReader = tsdbCacherowsReaderClose(pInfo->pLastrowReader);
         return pInfo->pRes;
       } else {
-        pInfo->pLastrowReader = tsdbCacherowsReaderClose(pInfo->pLastrowReader);
+        //pInfo->pLastrowReader = tsdbCacherowsReaderClose(pInfo->pLastrowReader);
       }
     }
 
+    pInfo->pLastrowReader = tsdbCacherowsReaderClose(pInfo->pLastrowReader);
     setOperatorCompleted(pOperator);
     return NULL;
   }

source/libs/executor/src/dataDispatcher.c

@@ -198,9 +198,6 @@ static int32_t getDataBlock(SDataSinkHandle* pHandle, SOutputData* pOutput) {
   pOutput->numOfCols = pEntry->numOfCols;
   pOutput->compressed = pEntry->compressed;
 
-  //  ASSERT(pEntry->numOfRows == *(int32_t*)(pEntry->data + 8));
-  //  ASSERT(pEntry->numOfCols == *(int32_t*)(pEntry->data + 8 + 4));
-
   atomic_sub_fetch_64(&pDispatcher->cachedSize, pEntry->dataLen);
   atomic_sub_fetch_64(&gDataSinkStat.cachedSize, pEntry->dataLen);
 

source/libs/executor/src/executil.c

@@ -28,23 +28,6 @@
 #include "querytask.h"
 #include "tcompression.h"
 
-typedef struct STableListIdInfo {
-  uint64_t suid;
-  uint64_t uid;
-  int32_t  tableType;
-} STableListIdInfo;
-
-// If the numOfOutputGroups is 1, the data blocks that belongs to different groups will be provided randomly
-// The numOfOutputGroups is specified by physical plan. and will not be affect by numOfGroups
-struct STableListInfo {
-  bool             oneTableForEachGroup;
-  int32_t          numOfOuputGroups;  // the data block will be generated one by one
-  int32_t*         groupOffset;       // keep the offset value for each group in the tableList
-  SArray*          pTableList;
-  SHashObj*        map;     // speedup acquire the tableQueryInfo by table uid
-  STableListIdInfo idInfo;  // this maybe the super table or ordinary table
-};
-
 typedef struct tagFilterAssist {
   SHashObj* colHash;
   int32_t   index;
@@ -62,7 +45,7 @@ static int32_t        optimizeTbnameInCond(void* metaHandle, int64_t suid, SArra
 static int32_t        optimizeTbnameInCondImpl(void* metaHandle, SArray* list, SNode* pTagCond);
 
 static int32_t      getTableList(void* metaHandle, void* pVnode, SScanPhysiNode* pScanNode, SNode* pTagCond,
-                                 SNode* pTagIndexCond, STableListInfo* pListInfo, const char* idstr);
+                                 SNode* pTagIndexCond, STableListInfo* pListInfo, uint8_t* digest, const char* idstr);
 static SSDataBlock* createTagValBlockForFilter(SArray* pColList, int32_t numOfTables, SArray* pUidTagList,
                                                void* metaHandle);
 
@@ -436,12 +419,49 @@ void freeItem(void* p) {
   }
 }
 
-int32_t getColInfoResultForGroupby(void* metaHandle, SNodeList* group, STableListInfo* pTableListInfo) {
+static void genTagFilterDigest(const SNode* pTagCond, T_MD5_CTX* pContext) {
+  if (pTagCond == NULL) {
+    return;
+  }
+
+  char*   payload = NULL;
+  int32_t len = 0;
+  nodesNodeToMsg(pTagCond, &payload, &len);
+
+  tMD5Init(pContext);
+  tMD5Update(pContext, (uint8_t*)payload, (uint32_t)len);
+  tMD5Final(pContext);
+
+  taosMemoryFree(payload);
+}
+
+
+static void genTbGroupDigest(const SNode* pGroup, uint8_t* filterDigest, T_MD5_CTX* pContext) {
+  char*   payload = NULL;
+  int32_t len = 0;
+  nodesNodeToMsg(pGroup, &payload, &len);
+  if (filterDigest[0]) {
+    payload = taosMemoryRealloc(payload, len + tListLen(pContext->digest));
+    memcpy(payload + len, filterDigest + 1, tListLen(pContext->digest));
+    len += tListLen(pContext->digest);
+  }
+
+  tMD5Init(pContext);
+  tMD5Update(pContext, (uint8_t*)payload, (uint32_t)len);
+  tMD5Final(pContext);
+
+  taosMemoryFree(payload);
+}
+
+
+int32_t getColInfoResultForGroupby(void* metaHandle, SNodeList* group, STableListInfo* pTableListInfo, uint8_t *digest) {
   int32_t      code = TSDB_CODE_SUCCESS;
   SArray*      pBlockList = NULL;
   SSDataBlock* pResBlock = NULL;
   void*        keyBuf = NULL;
   SArray*      groupData = NULL;
+  SArray*      pUidTagList = NULL;
+  SArray*      tableList = NULL;
 
   int32_t rows = taosArrayGetSize(pTableListInfo->pTableList);
   if (rows == 0) {
@@ -468,7 +488,23 @@ int32_t getColInfoResultForGroupby(void* metaHandle, SNodeList* group, STableLis
     REPLACE_NODE(pNode);
   }
 
-  SArray* pUidTagList = taosArrayInit(8, sizeof(STUidTagInfo));
+  T_MD5_CTX context = {0};
+  if (tsTagFilterCache) {
+    SNodeListNode* listNode = (SNodeListNode*)nodesMakeNode(QUERY_NODE_NODE_LIST);
+    listNode->pNodeList = group;
+    genTbGroupDigest((SNode *)listNode, digest, &context);
+    nodesFree(listNode);
+    
+    metaGetCachedTbGroup(metaHandle, pTableListInfo->idInfo.suid, context.digest, tListLen(context.digest), &tableList);
+    if (tableList) {
+      taosArrayDestroy(pTableListInfo->pTableList);
+      pTableListInfo->pTableList = tableList;
+      qDebug("retrieve tb group list from cache, numOfTables:%d", (int32_t)taosArrayGetSize(pTableListInfo->pTableList));
+      goto end;
+    }
+  }
+  
+  pUidTagList = taosArrayInit(8, sizeof(STUidTagInfo));
   for (int32_t i = 0; i < rows; ++i) {
     STableKeyInfo* pkeyInfo = taosArrayGet(pTableListInfo->pTableList, i);
     STUidTagInfo   info = {.uid = pkeyInfo->uid};
@@ -594,6 +630,11 @@ int32_t getColInfoResultForGroupby(void* metaHandle, SNodeList* group, STableLis
     info->groupId = calcGroupId(keyBuf, len);
   }
 
+  if (tsTagFilterCache) {
+    tableList = taosArrayDup(pTableListInfo->pTableList, NULL);
+    metaPutTbGroupToCache(metaHandle, pTableListInfo->idInfo.suid, context.digest, tListLen(context.digest), tableList, taosArrayGetSize(tableList) * sizeof(STableKeyInfo));
+  }
+  
   //  int64_t st2 = taosGetTimestampUs();
   //  qDebug("calculate tag block rows:%d, cost:%ld us", rows, st2-st1);
 
@@ -798,21 +839,6 @@ static int32_t optimizeTbnameInCondImpl(void* metaHandle, SArray* pExistedUidLis
   return -1;
 }
 
-static void genTagFilterDigest(const SNode* pTagCond, T_MD5_CTX* pContext) {
-  if (pTagCond == NULL) {
-    return;
-  }
-
-  char*   payload = NULL;
-  int32_t len = 0;
-  nodesNodeToMsg(pTagCond, &payload, &len);
-
-  tMD5Init(pContext);
-  tMD5Update(pContext, (uint8_t*)payload, (uint32_t)len);
-  tMD5Final(pContext);
-
-  taosMemoryFree(payload);
-}
 
 static SSDataBlock* createTagValBlockForFilter(SArray* pColList, int32_t numOfTables, SArray* pUidTagList,
                                                void* metaHandle) {
@@ -1027,7 +1053,7 @@ end:
 }
 
 int32_t getTableList(void* metaHandle, void* pVnode, SScanPhysiNode* pScanNode, SNode* pTagCond, SNode* pTagIndexCond,
-                     STableListInfo* pListInfo, const char* idstr) {
+                     STableListInfo* pListInfo, uint8_t* digest, const char* idstr) {
   int32_t code = TSDB_CODE_SUCCESS;
   size_t  numOfTables = 0;
 
@@ -1057,6 +1083,8 @@ int32_t getTableList(void* metaHandle, void* pVnode, SScanPhysiNode* pScanNode,
       metaGetCachedTableUidList(metaHandle, pScanNode->suid, context.digest, tListLen(context.digest), pUidList,
                                 &acquired);
       if (acquired) {
+        digest[0] = 1;
+        memcpy(digest + 1, context.digest, tListLen(context.digest));
         qDebug("retrieve table uid list from cache, numOfTables:%d", (int32_t)taosArrayGetSize(pUidList));
         goto _end;
       }
@@ -1100,6 +1128,8 @@ int32_t getTableList(void* metaHandle, void* pVnode, SScanPhysiNode* pScanNode,
       }
 
       metaUidFilterCachePut(metaHandle, pScanNode->suid, context.digest, tListLen(context.digest), pPayload, size, 1);
+      digest[0] = 1;
+      memcpy(digest + 1, context.digest, tListLen(context.digest));
     }
   }
 
@@ -2003,8 +2033,8 @@ static int32_t sortTableGroup(STableListInfo* pTableListInfo) {
   return TDB_CODE_SUCCESS;
 }
 
-int32_t buildGroupIdMapForAllTables(STableListInfo* pTableListInfo, SReadHandle* pHandle, SNodeList* group,
+int32_t buildGroupIdMapForAllTables(STableListInfo* pTableListInfo, SReadHandle* pHandle, SScanPhysiNode* pScanNode, SNodeList* group,
-                                    bool groupSort) {
+                                    bool groupSort, uint8_t *digest) {
   int32_t code = TSDB_CODE_SUCCESS;
 
   bool   groupByTbname = groupbyTbname(group);
@@ -2024,7 +2054,7 @@ int32_t buildGroupIdMapForAllTables(STableListInfo* pTableListInfo, SReadHandle*
       pTableListInfo->numOfOuputGroups = 1;
     }
   } else {
-    code = getColInfoResultForGroupby(pHandle->meta, group, pTableListInfo);
+    code = getColInfoResultForGroupby(pHandle->meta, group, pTableListInfo, digest);
     if (code != TSDB_CODE_SUCCESS) {
       return code;
     }
@@ -2055,7 +2085,8 @@ int32_t createScanTableListInfo(SScanPhysiNode* pScanNode, SNodeList* pGroupTags
     return TSDB_CODE_INVALID_PARA;
   }
 
-  int32_t code = getTableList(pHandle->meta, pHandle->vnode, pScanNode, pTagCond, pTagIndexCond, pTableListInfo, idStr);
+  uint8_t digest[17] = {0};
+  int32_t code = getTableList(pHandle->meta, pHandle->vnode, pScanNode, pTagCond, pTagIndexCond, pTableListInfo, digest, idStr);
   if (code != TSDB_CODE_SUCCESS) {
     qError("failed to getTableList, code: %s", tstrerror(code));
     return code;
@@ -2073,7 +2104,7 @@ int32_t createScanTableListInfo(SScanPhysiNode* pScanNode, SNodeList* pGroupTags
     return TSDB_CODE_SUCCESS;
   }
 
-  code = buildGroupIdMapForAllTables(pTableListInfo, pHandle, pGroupTags, groupSort);
+  code = buildGroupIdMapForAllTables(pTableListInfo, pHandle, pScanNode, pGroupTags, groupSort, digest);
   if (code != TSDB_CODE_SUCCESS) {
     return code;
   }

source/libs/executor/src/executor.c

@@ -1060,7 +1060,10 @@ void qStreamSetOpen(qTaskInfo_t tinfo) {
 
 void verifyOffset(void *pWalReader, STqOffsetVal* pOffset){
   // if offset version is small than first version , let's seek to first version
+  taosThreadMutexLock(&((SWalReader*)pWalReader)->pWal->mutex);
   int64_t firstVer = walGetFirstVer(((SWalReader*)pWalReader)->pWal);
+  taosThreadMutexUnlock(&((SWalReader*)pWalReader)->pWal->mutex);
+
   if (pOffset->version + 1 < firstVer){
     pOffset->version = firstVer - 1;
   }
@@ -1150,7 +1153,7 @@ int32_t qStreamPrepareScan(qTaskInfo_t tinfo, STqOffsetVal* pOffset, int8_t subT
 
       if (pScanBaseInfo->dataReader == NULL) {
         int32_t code = tsdbReaderOpen(pScanBaseInfo->readHandle.vnode, &pScanBaseInfo->cond, &keyInfo, 1,
-                                      pScanInfo->pResBlock, &pScanBaseInfo->dataReader, id, false);
+                                      pScanInfo->pResBlock, &pScanBaseInfo->dataReader, id, false, NULL);
         if (code != TSDB_CODE_SUCCESS) {
           qError("prepare read tsdb snapshot failed, uid:%" PRId64 ", code:%s %s", pOffset->uid, tstrerror(code), id);
           terrno = code;
@@ -1209,7 +1212,7 @@ int32_t qStreamPrepareScan(qTaskInfo_t tinfo, STqOffsetVal* pOffset, int8_t subT
       int32_t        size = tableListGetSize(pTableListInfo);
 
       tsdbReaderOpen(pInfo->vnode, &pTaskInfo->streamInfo.tableCond, pList, size, NULL, &pInfo->dataReader, NULL,
-                     false);
+                     false, NULL);
 
       cleanupQueryTableDataCond(&pTaskInfo->streamInfo.tableCond);
       strcpy(pTaskInfo->streamInfo.tbName, mtInfo.tbName);

source/libs/executor/src/executorInt.c

@@ -82,7 +82,7 @@ static void    extractQualifiedTupleByFilterResult(SSDataBlock* pBlock, const SC
 static int32_t doSetInputDataBlock(SExprSupp* pExprSup, SSDataBlock* pBlock, int32_t order, int32_t scanFlag,
                                    bool createDummyCol);
 static int32_t doCopyToSDataBlock(SExecTaskInfo* pTaskInfo, SSDataBlock* pBlock, SExprSupp* pSup, SDiskbasedBuf* pBuf,
-                                  SGroupResInfo* pGroupResInfo, int32_t threshold);
+                                  SGroupResInfo* pGroupResInfo, int32_t threshold, bool ignoreGroup);
 
 SResultRow* getNewResultRow(SDiskbasedBuf* pResultBuf, int32_t* currentPageId, int32_t interBufSize) {
   SFilePage* pData = NULL;
@@ -777,7 +777,7 @@ int32_t finalizeResultRows(SDiskbasedBuf* pBuf, SResultRowPosition* resultRowPos
 }
 
 int32_t doCopyToSDataBlock(SExecTaskInfo* pTaskInfo, SSDataBlock* pBlock, SExprSupp* pSup, SDiskbasedBuf* pBuf,
-                           SGroupResInfo* pGroupResInfo, int32_t threshold) {
+                           SGroupResInfo* pGroupResInfo, int32_t threshold, bool ignoreGroup) {
   SExprInfo*      pExprInfo = pSup->pExprInfo;
   int32_t         numOfExprs = pSup->numOfExprs;
   int32_t*        rowEntryOffset = pSup->rowEntryInfoOffset;
@@ -804,20 +804,23 @@ int32_t doCopyToSDataBlock(SExecTaskInfo* pTaskInfo, SSDataBlock* pBlock, SExprS
       continue;
     }
 
-    if (pBlock->info.id.groupId == 0) {
+    if (!ignoreGroup) {
-      pBlock->info.id.groupId = pPos->groupId;
+      if (pBlock->info.id.groupId == 0) {
-    } else {
+        pBlock->info.id.groupId = pPos->groupId;
-      // current value belongs to different group, it can't be packed into one datablock
+      } else {
-      if (pBlock->info.id.groupId != pPos->groupId) {
+        // current value belongs to different group, it can't be packed into one datablock
-        releaseBufPage(pBuf, page);
+        if (pBlock->info.id.groupId != pPos->groupId) {
-        break;
+          releaseBufPage(pBuf, page);
+          break;
+        }
       }
     }
 
     if (pBlock->info.rows + pRow->numOfRows > pBlock->info.capacity) {
-      blockDataEnsureCapacity(pBlock, pBlock->info.rows + pRow->numOfRows);
+      uint32_t newSize = pBlock->info.rows + pRow->numOfRows + (numOfRows - i) > 1 ? 1 : 0;
-      qDebug("datablock capacity not sufficient, expand to required:%" PRId64 ", current capacity:%d, %s",
+      blockDataEnsureCapacity(pBlock, newSize);
-             (pRow->numOfRows + pBlock->info.rows), pBlock->info.capacity, GET_TASKID(pTaskInfo));
+      qDebug("datablock capacity not sufficient, expand to required:%d, current capacity:%d, %s",
+             newSize, pBlock->info.capacity, GET_TASKID(pTaskInfo));
       // todo set the pOperator->resultInfo size
     }
 
@@ -854,7 +857,7 @@ void doBuildStreamResBlock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SGr
   // clear the existed group id
   pBlock->info.id.groupId = 0;
   ASSERT(!pbInfo->mergeResultBlock);
-  doCopyToSDataBlock(pTaskInfo, pBlock, &pOperator->exprSupp, pBuf, pGroupResInfo, pOperator->resultInfo.threshold);
+  doCopyToSDataBlock(pTaskInfo, pBlock, &pOperator->exprSupp, pBuf, pGroupResInfo, pOperator->resultInfo.threshold, false);
 
   void* tbname = NULL;
   if (streamStateGetParName(pTaskInfo->streamInfo.pState, pBlock->info.id.groupId, &tbname) < 0) {
@@ -881,10 +884,10 @@ void doBuildResultDatablock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SG
   // clear the existed group id
   pBlock->info.id.groupId = 0;
   if (!pbInfo->mergeResultBlock) {
-    doCopyToSDataBlock(pTaskInfo, pBlock, &pOperator->exprSupp, pBuf, pGroupResInfo, pOperator->resultInfo.threshold);
+    doCopyToSDataBlock(pTaskInfo, pBlock, &pOperator->exprSupp, pBuf, pGroupResInfo, pOperator->resultInfo.threshold, false);
   } else {
     while (hasRemainResults(pGroupResInfo)) {
-      doCopyToSDataBlock(pTaskInfo, pBlock, &pOperator->exprSupp, pBuf, pGroupResInfo, pOperator->resultInfo.threshold);
+      doCopyToSDataBlock(pTaskInfo, pBlock, &pOperator->exprSupp, pBuf, pGroupResInfo, pOperator->resultInfo.threshold, true);
       if (pBlock->info.rows >= pOperator->resultInfo.threshold) {
         break;
       }

source/libs/executor/src/scanoperator.c

@@ -195,9 +195,24 @@ static SResultRow* getTableGroupOutputBuf(SOperatorInfo* pOperator, uint64_t gro
   return (SResultRow*)((char*)(*pPage) + p1->offset);
 }
 
+static int32_t insertTableToScanIgnoreList(STableScanInfo* pTableScanInfo, uint64_t uid) {
+  if (NULL == pTableScanInfo->pIgnoreTables) {
+    int32_t tableNum = taosArrayGetSize(pTableScanInfo->base.pTableListInfo->pTableList);
+    pTableScanInfo->pIgnoreTables = taosHashInit(tableNum,  taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, HASH_NO_LOCK);
+    if (NULL == pTableScanInfo->pIgnoreTables) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
+  }
+  
+  taosHashPut(pTableScanInfo->pIgnoreTables, &uid, sizeof(uid), &pTableScanInfo->scanTimes, sizeof(pTableScanInfo->scanTimes));
+
+  return TSDB_CODE_SUCCESS;
+}
+
 static int32_t doDynamicPruneDataBlock(SOperatorInfo* pOperator, SDataBlockInfo* pBlockInfo, uint32_t* status) {
   STableScanInfo* pTableScanInfo = pOperator->info;
-
+  int32_t code = TSDB_CODE_SUCCESS;
+  
   if (pTableScanInfo->base.pdInfo.pExprSup == NULL) {
     return TSDB_CODE_SUCCESS;
   }
@@ -229,9 +244,10 @@ static int32_t doDynamicPruneDataBlock(SOperatorInfo* pOperator, SDataBlockInfo*
 
   if (notLoadBlock) {
     *status = FUNC_DATA_REQUIRED_NOT_LOAD;
+    code = insertTableToScanIgnoreList(pTableScanInfo, pBlockInfo->id.uid);
   }
 
-  return TSDB_CODE_SUCCESS;
+  return code;
 }
 
 static bool doFilterByBlockSMA(SFilterInfo* pFilterInfo, SColumnDataAgg** pColsAgg, int32_t numOfCols,
@@ -383,7 +399,13 @@ static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanBase* pTableSca
            GET_TASKID(pTaskInfo), pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows);
     pCost->skipBlocks += 1;
     tsdbReleaseDataBlock(pTableScanInfo->dataReader);
-    *status = FUNC_DATA_REQUIRED_FILTEROUT;
+
+    STableScanInfo* pTableScanInfo = pOperator->info;
+    if (taosHashGetSize(pTableScanInfo->pIgnoreTables) == taosArrayGetSize(pTableScanInfo->base.pTableListInfo->pTableList)) {
+      *status = FUNC_DATA_REQUIRED_ALL_FILTEROUT;
+    } else {
+      *status = FUNC_DATA_REQUIRED_FILTEROUT;
+    }
     return TSDB_CODE_SUCCESS;
   }
 
@@ -698,6 +720,10 @@ static SSDataBlock* doTableScanImpl(SOperatorInfo* pOperator) {
       T_LONG_JMP(pTaskInfo->env, code);
     }
 
+    if (status == FUNC_DATA_REQUIRED_ALL_FILTEROUT) {
+      break;
+    }
+
     // current block is filter out according to filter condition, continue load the next block
     if (status == FUNC_DATA_REQUIRED_FILTEROUT || pBlock->info.rows == 0) {
       continue;
@@ -737,6 +763,7 @@ static SSDataBlock* doGroupedTableScan(SOperatorInfo* pOperator) {
     }
 
     pTableScanInfo->scanTimes += 1;
+    taosHashClear(pTableScanInfo->pIgnoreTables);
 
     if (pTableScanInfo->scanTimes < pTableScanInfo->scanInfo.numOfAsc) {
       setTaskStatus(pTaskInfo, TASK_NOT_COMPLETED);
@@ -764,6 +791,7 @@ static SSDataBlock* doGroupedTableScan(SOperatorInfo* pOperator) {
       }
 
       pTableScanInfo->scanTimes += 1;
+      taosHashClear(pTableScanInfo->pIgnoreTables);
 
       if (pTableScanInfo->scanTimes < total) {
         setTaskStatus(pTaskInfo, TASK_NOT_COMPLETED);
@@ -828,7 +856,7 @@ static SSDataBlock* doTableScan(SOperatorInfo* pOperator) {
       ASSERT(pInfo->base.dataReader == NULL);
 
       int32_t code = tsdbReaderOpen(pInfo->base.readHandle.vnode, &pInfo->base.cond, pList, num, pInfo->pResBlock,
-                                    (STsdbReader**)&pInfo->base.dataReader, GET_TASKID(pTaskInfo), pInfo->countOnly);
+                                    (STsdbReader**)&pInfo->base.dataReader, GET_TASKID(pTaskInfo), pInfo->countOnly, &pInfo->pIgnoreTables);
       if (code != TSDB_CODE_SUCCESS) {
         T_LONG_JMP(pTaskInfo->env, code);
       }
@@ -897,6 +925,7 @@ static void destroyTableScanBase(STableScanBase* pBase) {
 static void destroyTableScanOperatorInfo(void* param) {
   STableScanInfo* pTableScanInfo = (STableScanInfo*)param;
   blockDataDestroy(pTableScanInfo->pResBlock);
+  taosHashCleanup(pTableScanInfo->pIgnoreTables);
   destroyTableScanBase(&pTableScanInfo->base);
   taosMemoryFreeClear(param);
 }
@@ -1063,7 +1092,7 @@ static SSDataBlock* readPreVersionData(SOperatorInfo* pTableScanOp, uint64_t tbU
   SSDataBlock* pBlock = pTableScanInfo->pResBlock;
   STsdbReader* pReader = NULL;
   int32_t      code = tsdbReaderOpen(pTableScanInfo->base.readHandle.vnode, &cond, &tblInfo, 1, pBlock,
-                                     (STsdbReader**)&pReader, GET_TASKID(pTaskInfo), false);
+                                     (STsdbReader**)&pReader, GET_TASKID(pTaskInfo), false, NULL);
   if (code != TSDB_CODE_SUCCESS) {
     terrno = code;
     T_LONG_JMP(pTaskInfo->env, code);
@@ -1790,7 +1819,7 @@ static SSDataBlock* doStreamScan(SOperatorInfo* pOperator) {
   SExecTaskInfo*   pTaskInfo = pOperator->pTaskInfo;
   SStreamScanInfo* pInfo = pOperator->info;
 
-  qDebug("stream scan called");
+  qDebug("stream scan started, %s", GET_TASKID(pTaskInfo));
 
   if (pTaskInfo->streamInfo.recoverStep == STREAM_RECOVER_STEP__PREPARE1 ||
       pTaskInfo->streamInfo.recoverStep == STREAM_RECOVER_STEP__PREPARE2) {
@@ -1799,13 +1828,13 @@ static SSDataBlock* doStreamScan(SOperatorInfo* pOperator) {
     if (pTaskInfo->streamInfo.recoverStep == STREAM_RECOVER_STEP__PREPARE1) {
       pTSInfo->base.cond.startVersion = 0;
       pTSInfo->base.cond.endVersion = pTaskInfo->streamInfo.fillHistoryVer1;
-      qDebug("stream recover step 1, from %" PRId64 " to %" PRId64, pTSInfo->base.cond.startVersion,
+      qDebug("stream recover step1, verRange:%" PRId64 " - %" PRId64, pTSInfo->base.cond.startVersion,
              pTSInfo->base.cond.endVersion);
       pTaskInfo->streamInfo.recoverStep = STREAM_RECOVER_STEP__SCAN1;
     } else {
       pTSInfo->base.cond.startVersion = pTaskInfo->streamInfo.fillHistoryVer1 + 1;
       pTSInfo->base.cond.endVersion = pTaskInfo->streamInfo.fillHistoryVer2;
-      qDebug("stream recover step 2, from %" PRId64 " to %" PRId64, pTSInfo->base.cond.startVersion,
+      qDebug("stream recover step2, verRange:%" PRId64 " - %" PRId64, pTSInfo->base.cond.startVersion,
              pTSInfo->base.cond.endVersion);
       pTaskInfo->streamInfo.recoverStep = STREAM_RECOVER_STEP__SCAN2;
     }
@@ -1907,7 +1936,6 @@ FETCH_NEXT_BLOCK:
   if (pInfo->blockType == STREAM_INPUT__DATA_BLOCK) {
     if (pInfo->validBlockIndex >= total) {
       doClearBufferedBlocks(pInfo);
-      /*pOperator->status = OP_EXEC_DONE;*/
       return NULL;
     }
 
@@ -2688,7 +2716,7 @@ static SSDataBlock* getTableDataBlockImpl(void* param) {
   SReadHandle* pHandle = &pInfo->base.readHandle;
 
   if (NULL == source->dataReader || !source->multiReader) {
-    code = tsdbReaderOpen(pHandle->vnode, pQueryCond, p, 1, pBlock, &source->dataReader, GET_TASKID(pTaskInfo), false);
+    code = tsdbReaderOpen(pHandle->vnode, pQueryCond, p, 1, pBlock, &source->dataReader, GET_TASKID(pTaskInfo), false, NULL);
     if (code != 0) {
       T_LONG_JMP(pTaskInfo->env, code);
     }
@@ -2736,6 +2764,10 @@ static SSDataBlock* getTableDataBlockImpl(void* param) {
       T_LONG_JMP(pTaskInfo->env, code);
     }
 
+    if (status == FUNC_DATA_REQUIRED_ALL_FILTEROUT) {
+      break;
+    }
+
     // current block is filter out according to filter condition, continue load the next block
     if (status == FUNC_DATA_REQUIRED_FILTEROUT || pBlock->info.rows == 0) {
       continue;

source/libs/executor/src/sysscanoperator.c

@@ -2268,7 +2268,7 @@ SOperatorInfo* createDataBlockInfoScanOperator(SReadHandle* readHandle, SBlockDi
     size_t num = tableListGetSize(pTableListInfo);
     void*  pList = tableListGetInfo(pTableListInfo, 0);
 
-    code = tsdbReaderOpen(readHandle->vnode, &cond, pList, num, pInfo->pResBlock, &pInfo->pHandle, pTaskInfo->id.str, false);
+    code = tsdbReaderOpen(readHandle->vnode, &cond, pList, num, pInfo->pResBlock, &pInfo->pHandle, pTaskInfo->id.str, false, NULL);
     cleanupQueryTableDataCond(&cond);
     if (code != 0) {
       goto _error;

source/libs/executor/src/timesliceoperator.c

@@ -310,6 +310,11 @@ static bool genInterpolationResult(STimeSliceOperatorInfo* pSliceInfo, SExprSupp
           break;
         }
 
+        if (end.key != INT64_MIN && end.key < pSliceInfo->current) {
+          hasInterp = false;
+          break;
+        }
+
         if (start.key == INT64_MIN || end.key == INT64_MIN) {
           colDataSetNULL(pDst, rows);
           break;
@@ -894,8 +899,10 @@ void destroyTimeSliceOperatorInfo(void* param) {
   }
   taosArrayDestroy(pInfo->pLinearInfo);
 
-  taosMemoryFree(pInfo->pPrevGroupKey->pData);
+  if (pInfo->pPrevGroupKey) {
-  taosMemoryFree(pInfo->pPrevGroupKey);
+    taosMemoryFree(pInfo->pPrevGroupKey->pData);
+    taosMemoryFree(pInfo->pPrevGroupKey);
+  }
 
   cleanupExprSupp(&pInfo->scalarSup);
 

source/libs/function/src/udfd.c

@@ -54,16 +54,15 @@ int32_t udfdCPluginOpen(SScriptUdfEnvItem *items, int numItems) { return 0; }
 int32_t udfdCPluginClose() { return 0; }
 
 const char *udfdCPluginUdfInitLoadInitDestoryFuncs(SUdfCPluginCtx *udfCtx, const char *udfName) {
-  char  initFuncName[TSDB_FUNC_NAME_LEN + 5] = {0};
+  char  initFuncName[TSDB_FUNC_NAME_LEN + 6] = {0};
   char *initSuffix = "_init";
-  strcpy(initFuncName, udfName);
+  snprintf(initFuncName, sizeof(initFuncName), "%s%s", udfName, initSuffix);
-  strncat(initFuncName, initSuffix, strlen(initSuffix));
   uv_dlsym(&udfCtx->lib, initFuncName, (void **)(&udfCtx->initFunc));
 
-  char  destroyFuncName[TSDB_FUNC_NAME_LEN + 5] = {0};
+  char  destroyFuncName[TSDB_FUNC_NAME_LEN + 9] = {0};
   char *destroySuffix = "_destroy";
   strcpy(destroyFuncName, udfName);
-  strncat(destroyFuncName, destroySuffix, strlen(destroySuffix));
+  snprintf(destroyFuncName, sizeof(destroyFuncName), "%s%s", udfName, destroySuffix);
   uv_dlsym(&udfCtx->lib, destroyFuncName, (void **)(&udfCtx->destroyFunc));
   return udfName;
 }
@@ -73,22 +72,19 @@ void udfdCPluginUdfInitLoadAggFuncs(SUdfCPluginCtx *udfCtx, const char *udfName)
   strcpy(processFuncName, udfName);
   uv_dlsym(&udfCtx->lib, processFuncName, (void **)(&udfCtx->aggProcFunc));
 
-  char  startFuncName[TSDB_FUNC_NAME_LEN + 6] = {0};
+  char  startFuncName[TSDB_FUNC_NAME_LEN + 7] = {0};
   char *startSuffix = "_start";
-  strncpy(startFuncName, processFuncName, sizeof(startFuncName));
+  snprintf(startFuncName, sizeof(startFuncName), "%s%s", processFuncName, startSuffix);
-  strncat(startFuncName, startSuffix, strlen(startSuffix));
   uv_dlsym(&udfCtx->lib, startFuncName, (void **)(&udfCtx->aggStartFunc));
 
-  char  finishFuncName[TSDB_FUNC_NAME_LEN + 7] = {0};
+  char  finishFuncName[TSDB_FUNC_NAME_LEN + 8] = {0};
   char *finishSuffix = "_finish";
-  strncpy(finishFuncName, processFuncName, sizeof(finishFuncName));
+  snprintf(finishFuncName, sizeof(finishFuncName), "%s%s", processFuncName, finishSuffix);
-  strncat(finishFuncName, finishSuffix, strlen(finishSuffix));
   uv_dlsym(&udfCtx->lib, finishFuncName, (void **)(&udfCtx->aggFinishFunc));
 
-  char  mergeFuncName[TSDB_FUNC_NAME_LEN + 6] = {0};
+  char  mergeFuncName[TSDB_FUNC_NAME_LEN + 7] = {0};
   char *mergeSuffix = "_merge";
-  strncpy(mergeFuncName, processFuncName, sizeof(mergeFuncName));
+  snprintf(mergeFuncName, sizeof(mergeFuncName), "%s%s", processFuncName, mergeSuffix);
-  strncat(mergeFuncName, mergeSuffix, strlen(mergeSuffix));
   uv_dlsym(&udfCtx->lib, mergeFuncName, (void **)(&udfCtx->aggMergeFunc));
 }
 

source/libs/nodes/src/nodesUtilFuncs.c

@@ -87,7 +87,7 @@ static void* nodesCalloc(int32_t num, int32_t size) {
   return (char*)p + 1;
 }
 
-static void nodesFree(void* p) {
+void nodesFree(void* p) {
   char* ptr = (char*)p - 1;
   if (0 == *ptr) {
     taosMemoryFree(ptr);

source/libs/planner/src/planOptimizer.c

@@ -2628,7 +2628,9 @@ static bool tbCntScanOptIsEligibleConds(STbCntScanOptInfo* pInfo, SNode* pCondit
   if (NULL == pConditions) {
     return true;
   }
-
+  if (LIST_LENGTH(pInfo->pAgg->pGroupKeys) != 0) {
+    return false;
+  }
   if (QUERY_NODE_LOGIC_CONDITION == nodeType(pConditions)) {
     return tbCntScanOptIsEligibleLogicCond(pInfo, (SLogicConditionNode*)pConditions);
   }

source/libs/qcom/src/queryUtil.c

@@ -548,3 +548,11 @@ int32_t cloneSVreateTbReq(SVCreateTbReq* pSrc, SVCreateTbReq** pDst) {
 
   return TSDB_CODE_SUCCESS;
 }
+
+void freeDbCfgInfo(SDbCfgInfo *pInfo) {
+  if (pInfo) {
+    taosArrayDestroy(pInfo->pRetensions);
+  }
+  taosMemoryFree(pInfo);
+}
+

source/libs/qworker/inc/qwInt.h

@@ -120,7 +120,8 @@ typedef struct SQWTaskCtx {
   int8_t   explain;
   int8_t   needFetch;
   int8_t   localExec;
-  int32_t  msgType;
+  int32_t  queryMsgType;
+  int32_t  fetchMsgType;
   int32_t  level;
   uint64_t sId;
 

source/libs/qworker/src/qwDbg.c

@@ -126,10 +126,10 @@ void qwDbgDumpTasksInfo(SQWorker *mgmt) {
     void       *key = taosHashGetKey(pIter, NULL);
     QW_GET_QTID(key, qId, tId, eId);
     
-    QW_TASK_DLOG("%p lock:%x, phase:%d, type:%d, explain:%d, needFetch:%d, localExec:%d, msgType:%d, "
+    QW_TASK_DLOG("%p lock:%x, phase:%d, type:%d, explain:%d, needFetch:%d, localExec:%d, queryMsgType:%d, "
       "sId:%" PRId64 ", level:%d, queryGotData:%d, queryRsped:%d, queryEnd:%d, queryContinue:%d, queryInQueue:%d, "
       "rspCode:%x, affectedRows:%" PRId64 ", taskHandle:%p, sinkHandle:%p, tbFName:%s, sver:%d, tver:%d, events:%d,%d,%d,%d,%d",
-      ctx, ctx->lock, ctx->phase, ctx->taskType, ctx->explain, ctx->needFetch, ctx->localExec, ctx->msgType,
+      ctx, ctx->lock, ctx->phase, ctx->taskType, ctx->explain, ctx->needFetch, ctx->localExec, ctx->queryMsgType,
       ctx->sId, ctx->level, ctx->queryGotData, ctx->queryRsped, ctx->queryEnd, ctx->queryContinue, 
       ctx->queryInQueue, ctx->rspCode, ctx->affectedRows, ctx->taskHandle, ctx->sinkHandle, ctx->tbInfo.tbFName,
       ctx->tbInfo.sversion, ctx->tbInfo.tversion, ctx->events[QW_EVENT_CANCEL], ctx->events[QW_EVENT_READY], 
@@ -259,9 +259,9 @@ void qwDbgSimulateDead(QW_FPARAMS_DEF, SQWTaskCtx *ctx, bool *rsped) {
   static int32_t ignoreTime = 0;
 
   if (++ignoreTime > 10 && 0 == taosRand() % 9) {
-    if (ctx->msgType == TDMT_SCH_FETCH) {
+    if (ctx->fetchMsgType == TDMT_SCH_FETCH) {
       qwBuildAndSendErrorRsp(TDMT_SCH_LINK_BROKEN, &ctx->ctrlConnInfo, TSDB_CODE_RPC_BROKEN_LINK);
-      qwBuildAndSendErrorRsp(ctx->msgType + 1, &ctx->dataConnInfo, TSDB_CODE_QRY_TASK_CTX_NOT_EXIST);
+      qwBuildAndSendErrorRsp(ctx->fetchMsgType + 1, &ctx->dataConnInfo, TSDB_CODE_QRY_TASK_CTX_NOT_EXIST);
       *rsped = true;
       
       taosSsleep(3);

source/libs/qworker/src/qworker.c

@@ -254,6 +254,7 @@ bool qwTaskNotInExec(SQWTaskCtx *ctx) {
   return false;
 }
 
+
 int32_t qwGenerateSchHbRsp(SQWorker *mgmt, SQWSchStatus *sch, SQWHbInfo *hbInfo) {
   int32_t taskNum = 0;
 
@@ -436,6 +437,40 @@ int32_t qwGetDeleteResFromSink(QW_FPARAMS_DEF, SQWTaskCtx *ctx, SDeleteRes *pRes
   return TSDB_CODE_SUCCESS;
 }
 
+
+int32_t qwQuickRspFetchReq(QW_FPARAMS_DEF, SQWTaskCtx    * ctx, SQWMsg *qwMsg, int32_t code) {
+  if (QUERY_RSP_POLICY_QUICK == tsQueryRspPolicy && ctx != NULL) {
+    if (QW_EVENT_RECEIVED(ctx, QW_EVENT_FETCH)) {
+      void       *rsp = NULL;
+      int32_t     dataLen = 0;
+      SOutputData sOutput = {0};
+      if (qwGetQueryResFromSink(QW_FPARAMS(), ctx, &dataLen, &rsp, &sOutput)) {
+        return TSDB_CODE_SUCCESS;
+      }
+
+      if (rsp) {
+        bool qComplete = (DS_BUF_EMPTY == sOutput.bufStatus && sOutput.queryEnd);
+
+        qwBuildFetchRsp(rsp, &sOutput, dataLen, qComplete);
+        if (qComplete) {
+          atomic_store_8((int8_t *)&ctx->queryEnd, true);
+        }
+
+        qwMsg->connInfo = ctx->dataConnInfo;
+        QW_SET_EVENT_PROCESSED(ctx, QW_EVENT_FETCH);
+
+        qwBuildAndSendFetchRsp(ctx->fetchMsgType + 1, &qwMsg->connInfo, rsp, dataLen, code);
+        rsp = NULL;
+
+        QW_TASK_DLOG("fetch rsp send, handle:%p, code:%x - %s, dataLen:%d", qwMsg->connInfo.handle, code, tstrerror(code),
+                     dataLen);
+      }
+    } 
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
 int32_t qwHandlePrePhaseEvents(QW_FPARAMS_DEF, int8_t phase, SQWPhaseInput *input, SQWPhaseOutput *output) {
   int32_t     code = 0;
   SQWTaskCtx *ctx = NULL;
@@ -584,7 +619,7 @@ _return:
 
   if (QW_PHASE_POST_QUERY == phase && ctx && !ctx->queryRsped) {
     bool   rsped = false;
-    SQWMsg qwMsg = {.msgType = ctx->msgType, .connInfo = ctx->ctrlConnInfo};
+    SQWMsg qwMsg = {.msgType = ctx->queryMsgType, .connInfo = ctx->ctrlConnInfo};
     qwDbgSimulateRedirect(&qwMsg, ctx, &rsped);
     qwDbgSimulateDead(QW_FPARAMS(), ctx, &rsped);
     if (!rsped) {
@@ -634,6 +669,8 @@ int32_t qwPreprocessQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg) {
 
   QW_ERR_JRET(qwAddTaskStatus(QW_FPARAMS(), JOB_TASK_STATUS_INIT));
 
+  qwSendQueryRsp(QW_FPARAMS(), qwMsg->msgType + 1, ctx, code, true);
+
 _return:
 
   if (ctx) {
@@ -660,7 +697,7 @@ int32_t qwProcessQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg, char *sql) {
   ctx->taskType = qwMsg->msgInfo.taskType;
   ctx->explain = qwMsg->msgInfo.explain;
   ctx->needFetch = qwMsg->msgInfo.needFetch;
-  ctx->msgType = qwMsg->msgType;
+  ctx->queryMsgType = qwMsg->msgType;
   ctx->localExec = false;
 
   // QW_TASK_DLOGL("subplan json string, len:%d, %s", qwMsg->msgLen, qwMsg->msg);
@@ -684,7 +721,7 @@ int32_t qwProcessQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg, char *sql) {
     QW_ERR_JRET(TSDB_CODE_APP_ERROR);
   }
 
-  qwSendQueryRsp(QW_FPARAMS(), qwMsg->msgType + 1, ctx, code, true);
+  //qwSendQueryRsp(QW_FPARAMS(), qwMsg->msgType + 1, ctx, code, true);
 
   ctx->level = plan->level;
   atomic_store_ptr(&ctx->taskHandle, pTaskInfo);
@@ -701,32 +738,7 @@ _return:
   input.msgType = qwMsg->msgType;
   code = qwHandlePostPhaseEvents(QW_FPARAMS(), QW_PHASE_POST_QUERY, &input, NULL);
 
-  if (QUERY_RSP_POLICY_QUICK == tsQueryRspPolicy && ctx != NULL && QW_EVENT_RECEIVED(ctx, QW_EVENT_FETCH)) {
+  qwQuickRspFetchReq(QW_FPARAMS(), ctx, qwMsg, code);
-    void       *rsp = NULL;
-    int32_t     dataLen = 0;
-    SOutputData sOutput = {0};
-    if (qwGetQueryResFromSink(QW_FPARAMS(), ctx, &dataLen, &rsp, &sOutput)) {
-      return TSDB_CODE_SUCCESS;
-    }
-
-    if (rsp) {
-      bool qComplete = (DS_BUF_EMPTY == sOutput.bufStatus && sOutput.queryEnd);
-
-      qwBuildFetchRsp(rsp, &sOutput, dataLen, qComplete);
-      if (qComplete) {
-        atomic_store_8((int8_t *)&ctx->queryEnd, true);
-      }
-
-      qwMsg->connInfo = ctx->dataConnInfo;
-      QW_SET_EVENT_PROCESSED(ctx, QW_EVENT_FETCH);
-
-      qwBuildAndSendFetchRsp(ctx->msgType + 1, &qwMsg->connInfo, rsp, dataLen, code);
-      rsp = NULL;
-
-      QW_TASK_DLOG("fetch rsp send, handle:%p, code:%x - %s, dataLen:%d", qwMsg->connInfo.handle, code, tstrerror(code),
-                   dataLen);
-    }
-  }
 
   QW_RET(TSDB_CODE_SUCCESS);
 }
@@ -750,8 +762,10 @@ int32_t qwProcessCQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg) {
     atomic_store_8((int8_t *)&ctx->queryInQueue, 0);
     atomic_store_8((int8_t *)&ctx->queryContinue, 0);
 
-    QW_ERR_JRET(qwExecTask(QW_FPARAMS(), ctx, &queryStop));
+    if (!queryStop) {
-
+      QW_ERR_JRET(qwExecTask(QW_FPARAMS(), ctx, &queryStop));
+    }
+    
     if (QW_EVENT_RECEIVED(ctx, QW_EVENT_FETCH)) {
       SOutputData sOutput = {0};
       QW_ERR_JRET(qwGetQueryResFromSink(QW_FPARAMS(), ctx, &dataLen, &rsp, &sOutput));
@@ -774,7 +788,7 @@ int32_t qwProcessCQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg) {
         qwMsg->connInfo = ctx->dataConnInfo;
         QW_SET_EVENT_PROCESSED(ctx, QW_EVENT_FETCH);
 
-        qwBuildAndSendFetchRsp(ctx->msgType + 1, &qwMsg->connInfo, rsp, dataLen, code);
+        qwBuildAndSendFetchRsp(ctx->fetchMsgType + 1, &qwMsg->connInfo, rsp, dataLen, code);
         rsp = NULL;
 
         QW_TASK_DLOG("fetch rsp send, handle:%p, code:%x - %s, dataLen:%d", qwMsg->connInfo.handle, code,
@@ -796,7 +810,7 @@ int32_t qwProcessCQuery(QW_FPARAMS_DEF, SQWMsg *qwMsg) {
       rsp = NULL;
 
       qwMsg->connInfo = ctx->dataConnInfo;
-      qwBuildAndSendFetchRsp(ctx->msgType + 1, &qwMsg->connInfo, NULL, 0, code);
+      qwBuildAndSendFetchRsp(ctx->fetchMsgType + 1, &qwMsg->connInfo, NULL, 0, code);
       QW_TASK_DLOG("fetch rsp send, handle:%p, code:%x - %s, dataLen:%d", qwMsg->connInfo.handle, code, tstrerror(code),
                    0);
     }
@@ -830,7 +844,7 @@ int32_t qwProcessFetch(QW_FPARAMS_DEF, SQWMsg *qwMsg) {
 
   QW_ERR_JRET(qwGetTaskCtx(QW_FPARAMS(), &ctx));
 
-  ctx->msgType = qwMsg->msgType;
+  ctx->fetchMsgType = qwMsg->msgType;
   ctx->dataConnInfo = qwMsg->connInfo;
 
   SOutputData sOutput = {0};
@@ -896,6 +910,8 @@ _return:
       qwFreeFetchRsp(rsp);
       rsp = NULL;
     }
+  } else {    
+    //qwQuickRspFetchReq(QW_FPARAMS(), ctx, qwMsg, code);
   }
 
   QW_RET(TSDB_CODE_SUCCESS);
@@ -1307,7 +1323,7 @@ int32_t qWorkerProcessLocalQuery(void *pMgmt, uint64_t sId, uint64_t qId, uint64
   ctx->taskType = qwMsg->msgInfo.taskType;
   ctx->explain = qwMsg->msgInfo.explain;
   ctx->needFetch = qwMsg->msgInfo.needFetch;
-  ctx->msgType = qwMsg->msgType;
+  ctx->queryMsgType = qwMsg->msgType;
   ctx->localExec = true;
   ctx->explainRes = explainRes;
 
@@ -1362,7 +1378,7 @@ int32_t qWorkerProcessLocalFetch(void *pMgmt, uint64_t sId, uint64_t qId, uint64
 
   QW_ERR_JRET(qwGetTaskCtx(QW_FPARAMS(), &ctx));
 
-  ctx->msgType = TDMT_SCH_MERGE_FETCH;
+  ctx->fetchMsgType = TDMT_SCH_MERGE_FETCH;
   ctx->explainRes = explainRes;
 
   SOutputData sOutput = {0};

source/libs/stream/src/stream.c

@@ -16,8 +16,10 @@
 #include "streamInc.h"
 #include "ttimer.h"
 
-#define STREAM_TASK_INPUT_QUEUEU_CAPACITY 20480
+#define STREAM_TASK_INPUT_QUEUEU_CAPACITY          20480
-#define STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE (100)
+#define STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE (50)
+#define ONE_MB_F                                  (1048576.0)
+#define QUEUE_MEM_SIZE_IN_MB(_q)                  (taosQueueMemorySize(_q)/ONE_MB_F)
 
 int32_t streamInit() {
   int8_t old;
@@ -281,33 +283,27 @@ int32_t streamProcessRetrieveReq(SStreamTask* pTask, SStreamRetrieveReq* pReq, S
 }
 
 bool tInputQueueIsFull(const SStreamTask* pTask) {
-  return taosQueueItemSize((pTask->inputQueue->queue)) >= STREAM_TASK_INPUT_QUEUEU_CAPACITY;
+  bool isFull = taosQueueItemSize((pTask->inputQueue->queue)) >= STREAM_TASK_INPUT_QUEUEU_CAPACITY;
+  double size = QUEUE_MEM_SIZE_IN_MB(pTask->inputQueue->queue);
+  return (isFull || size >= STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE);
 }
 
 int32_t tAppendDataToInputQueue(SStreamTask* pTask, SStreamQueueItem* pItem) {
-  int8_t type = pItem->type;
+  int8_t  type = pItem->type;
+  int32_t total = taosQueueItemSize(pTask->inputQueue->queue) + 1;
+  double  size = QUEUE_MEM_SIZE_IN_MB(pTask->inputQueue->queue);
 
   if (type == STREAM_INPUT__DATA_SUBMIT) {
-    SStreamDataSubmit2* pSubmitBlock = streamSubmitBlockClone((SStreamDataSubmit2*)pItem);
+    SStreamDataSubmit2* pSubmitBlock = (SStreamDataSubmit2*)pItem;
-    if (pSubmitBlock == NULL) {
-      qDebug("task %d %p submit enqueue failed since out of memory", pTask->id.taskId, pTask);
-      terrno = TSDB_CODE_OUT_OF_MEMORY;
-      atomic_store_8(&pTask->inputStatus, TASK_INPUT_STATUS__FAILED);
-      return -1;
-    }
-
-    int32_t numOfBlocks = taosQueueItemSize(pTask->inputQueue->queue) + 1;
-    double size = taosQueueMemorySize(pTask->inputQueue->queue) / 1048576.0;
-
     qDebug("s-task:%s submit enqueue msgLen:%d ver:%" PRId64 ", total in queue:%d, size:%.2fMiB", pTask->id.idStr,
-           pSubmitBlock->submit.msgLen, pSubmitBlock->submit.ver, numOfBlocks, size);
+           pSubmitBlock->submit.msgLen, pSubmitBlock->submit.ver, total, size);
 
-    if ((pTask->taskLevel == TASK_LEVEL__SOURCE) &&
+    if ((pTask->taskLevel == TASK_LEVEL__SOURCE) && (tInputQueueIsFull(pTask))) {
-        (numOfBlocks > STREAM_TASK_INPUT_QUEUEU_CAPACITY || (size >= STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE))) {
       qError("s-task:%s input queue is full, capacity(size:%d num:%dMiB), current(blocks:%d, size:%.2fMiB) abort", pTask->id.idStr,
              STREAM_TASK_INPUT_QUEUEU_CAPACITY, STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE,
-             numOfBlocks, size);
+             total, size);
       streamDataSubmitDestroy(pSubmitBlock);
+      taosFreeQitem(pSubmitBlock);
       return -1;
     }
 
@@ -315,10 +311,8 @@ int32_t tAppendDataToInputQueue(SStreamTask* pTask, SStreamQueueItem* pItem) {
   } else if (type == STREAM_INPUT__DATA_BLOCK || type == STREAM_INPUT__DATA_RETRIEVE ||
              type == STREAM_INPUT__REF_DATA_BLOCK) {
     int32_t numOfBlocks = taosQueueItemSize(pTask->inputQueue->queue) + 1;
-    double size = taosQueueMemorySize(pTask->inputQueue->queue) / 1048576.0;
 
-    if ((pTask->taskLevel == TASK_LEVEL__SOURCE) &&
+    if ((pTask->taskLevel == TASK_LEVEL__SOURCE) && (tInputQueueIsFull(pTask))) {
-        (numOfBlocks > STREAM_TASK_INPUT_QUEUEU_CAPACITY || (size >= STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE))) {
       qError("s-task:%s input queue is full, capacity:%d size:%d MiB, current(blocks:%d, size:%.2fMiB) abort",
              pTask->id.idStr, STREAM_TASK_INPUT_QUEUEU_CAPACITY, STREAM_TASK_INPUT_QUEUEU_CAPACITY_IN_SIZE, numOfBlocks,
              size);

source/libs/stream/src/streamData.c

@@ -200,13 +200,7 @@ void streamFreeQitem(SStreamQueueItem* data) {
     taosFreeQitem(pMerge);
   } else if (type == STREAM_INPUT__REF_DATA_BLOCK) {
     SStreamRefDataBlock* pRefBlock = (SStreamRefDataBlock*)data;
-
+    blockDataDestroy(pRefBlock->pBlock);
-    int32_t ref = atomic_sub_fetch_32(pRefBlock->dataRef, 1);
-    ASSERT(ref >= 0);
-    if (ref == 0) {
-      blockDataDestroy(pRefBlock->pBlock);
-      taosMemoryFree(pRefBlock->dataRef);
-    }
     taosFreeQitem(pRefBlock);
   }
 }

source/libs/stream/src/streamExec.c

@@ -35,7 +35,7 @@ static int32_t streamTaskExecImpl(SStreamTask* pTask, const void* data, SArray*
 
   while (pTask->taskLevel == TASK_LEVEL__SOURCE) {
     int8_t status = atomic_load_8(&pTask->status.taskStatus);
-    if (status != TASK_STATUS__NORMAL) {
+    if (status != TASK_STATUS__NORMAL && status != TASK_STATUS__PAUSE) {
       qError("stream task wait for the end of fill history, s-task:%s, status:%d", pTask->id.idStr,
              atomic_load_8(&pTask->status.taskStatus));
       taosMsleep(2);
@@ -67,7 +67,7 @@ static int32_t streamTaskExecImpl(SStreamTask* pTask, const void* data, SArray*
 
     SArray* pBlockList = pMerged->submits;
     int32_t numOfBlocks = taosArrayGetSize(pBlockList);
-    qDebug("s-task:%s %p set submit input (merged), numOfblocks:%d", pTask->id.idStr, pTask, numOfBlocks);
+    qDebug("s-task:%s %p set submit input (merged), batch num:%d", pTask->id.idStr, pTask, numOfBlocks);
     qSetMultiStreamInput(pExecutor, pBlockList->pData, numOfBlocks, STREAM_INPUT__MERGED_SUBMIT);
   } else if (pItem->type == STREAM_INPUT__REF_DATA_BLOCK) {
     const SStreamRefDataBlock* pRefBlock = (const SStreamRefDataBlock*)data;
@@ -408,7 +408,7 @@ int32_t streamTryExec(SStreamTask* pTask) {
     atomic_store_8(&pTask->status.schedStatus, TASK_SCHED_STATUS__INACTIVE);
     qDebug("s-task:%s exec completed", pTask->id.idStr);
 
-    if (!taosQueueEmpty(pTask->inputQueue->queue) && (!streamTaskShouldStop(&pTask->status))) {
+    if (!taosQueueEmpty(pTask->inputQueue->queue) && (!streamTaskShouldStop(&pTask->status)) && (!streamTaskShouldPause(&pTask->status))) {
       streamSchedExec(pTask);
     }
   }

source/libs/stream/src/streamRecover.c

@@ -224,12 +224,19 @@ int32_t streamBuildSourceRecover2Req(SStreamTask* pTask, SStreamRecoverStep2Req*
 
 int32_t streamSourceRecoverScanStep2(SStreamTask* pTask, int64_t ver) {
   void* exec = pTask->exec.pExecutor;
+  const char* id = pTask->id.idStr;
 
-  qDebug("s-task:%s recover step2 (blocking stage) started", pTask->id.idStr);
+  int64_t st = taosGetTimestampMs();
+  qDebug("s-task:%s recover step2(blocking stage) started", id);
   if (qStreamSourceRecoverStep2(exec, ver) < 0) {
   }
 
-  return streamScanExec(pTask, 100);
+  int32_t code = streamScanExec(pTask, 100);
+
+  double el = (taosGetTimestampMs() - st) / 1000.0;
+  qDebug("s-task:%s recover step2(blocking stage) ended, elapsed time:%.2fs", id,  el);
+
+  return code;
 }
 
 int32_t streamDispatchRecoverFinishReq(SStreamTask* pTask) {

source/libs/transport/src/transCli.c

@@ -966,7 +966,7 @@ static void cliSendCb(uv_write_t* req, int status) {
   }
 
   if (status == 0) {
-    tTrace("%s conn %p data already was written out", CONN_GET_INST_LABEL(pConn), pConn);
+    tDebug("%s conn %p data already was written out", CONN_GET_INST_LABEL(pConn), pConn);
   } else {
     if (!uv_is_closing((uv_handle_t*)&pConn->stream)) {
       tError("%s conn %p failed to write:%s", CONN_GET_INST_LABEL(pConn), pConn, uv_err_name(status));

source/libs/transport/src/transSvr.c

@@ -236,7 +236,7 @@ static bool uvHandleReq(SSvrConn* pConn) {
   if (pConn->status == ConnNormal && pHead->noResp == 0) {
     transRefSrvHandle(pConn);
     if (cost >= EXCEPTION_LIMIT_US) {
-      tGWarn("%s conn %p %s received from %s, local info:%s, len:%d, cost:%dus, recv exception", transLabel(pTransInst),
+      tGDebug("%s conn %p %s received from %s, local info:%s, len:%d, cost:%dus, recv exception", transLabel(pTransInst),
              pConn, TMSG_INFO(transMsg.msgType), pConn->dst, pConn->src, msgLen, (int)cost);
     } else {
       tGDebug("%s conn %p %s received from %s, local info:%s, len:%d, cost:%dus", transLabel(pTransInst), pConn,
@@ -244,7 +244,7 @@ static bool uvHandleReq(SSvrConn* pConn) {
     }
   } else {
     if (cost >= EXCEPTION_LIMIT_US) {
-      tGWarn("%s conn %p %s received from %s, local info:%s, len:%d, noResp:%d, code:%d, cost:%dus, recv exception",
+      tGDebug("%s conn %p %s received from %s, local info:%s, len:%d, noResp:%d, code:%d, cost:%dus, recv exception",
              transLabel(pTransInst), pConn, TMSG_INFO(transMsg.msgType), pConn->dst, pConn->src, msgLen, pHead->noResp,
              transMsg.code, (int)(cost));
     } else {

source/libs/wal/src/walRead.c

@@ -62,9 +62,6 @@ SWalReader *walOpenReader(SWal *pWal, SWalFilterCond *cond) {
 void walCloseReader(SWalReader *pReader) {
   taosCloseFile(&pReader->pIdxFile);
   taosCloseFile(&pReader->pLogFile);
-  /*if (pReader->cond.enableRef) {*/
-  /*taosHashRemove(pReader->pWal->pRefHash, &pReader->readerId, sizeof(int64_t));*/
-  /*}*/
   taosMemoryFreeClear(pReader->pHead);
   taosMemoryFree(pReader);
 }
@@ -74,22 +71,25 @@ int32_t walNextValidMsg(SWalReader *pReader) {
   int64_t lastVer = walGetLastVer(pReader->pWal);
   int64_t committedVer = walGetCommittedVer(pReader->pWal);
   int64_t appliedVer = walGetAppliedVer(pReader->pWal);
+
   if(appliedVer < committedVer){   // wait apply ver equal to commit ver, otherwise may lost data when consume data [TD-24010]
     wDebug("vgId:%d, wal apply ver:%"PRId64" smaller than commit ver:%"PRId64, pReader->pWal->cfg.vgId, appliedVer, committedVer);
-//    taosMsleep(10);
   }
-//  int64_t endVer = pReader->cond.scanUncommited ? lastVer : committedVer;
+
   int64_t endVer = TMIN(appliedVer, committedVer);
 
   wDebug("vgId:%d, wal start to fetch, index:%" PRId64 ", last index:%" PRId64 " commit index:%" PRId64
          ", applied index:%" PRId64", end index:%" PRId64,
          pReader->pWal->cfg.vgId, fetchVer, lastVer, committedVer, appliedVer, endVer);
+
   while (fetchVer <= endVer) {
     if (walFetchHeadNew(pReader, fetchVer) < 0) {
       return -1;
     }
-    if (pReader->pHead->head.msgType == TDMT_VND_SUBMIT ||
+
-        (IS_META_MSG(pReader->pHead->head.msgType) && pReader->cond.scanMeta)) {
+    int32_t type = pReader->pHead->head.msgType;
+    if (type == TDMT_VND_SUBMIT || ((type == TDMT_VND_DELETE) && (pReader->cond.deleteMsg == 1)) ||
+        (IS_META_MSG(type) && pReader->cond.scanMeta)) {
       if (walFetchBodyNew(pReader) < 0) {
         return -1;
       }
@@ -98,9 +98,11 @@ int32_t walNextValidMsg(SWalReader *pReader) {
       if (walSkipFetchBodyNew(pReader) < 0) {
         return -1;
       }
+
       fetchVer = pReader->curVersion;
     }
   }
+
   return -1;
 }
 
@@ -256,7 +258,9 @@ static int32_t walFetchHeadNew(SWalReader *pRead, int64_t fetchVer) {
     if (contLen == sizeof(SWalCkHead)) {
       break;
     } else if (contLen == 0 && !seeked) {
-      walReadSeekVerImpl(pRead, fetchVer);
+      if(walReadSeekVerImpl(pRead, fetchVer) < 0){
+        return -1;
+      }
       seeked = true;
       continue;
     } else {
@@ -285,6 +289,7 @@ static int32_t walFetchBodyNew(SWalReader *pReader) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
     }
+
     pReader->pHead = ptr;
     pReadHead = &pReader->pHead->head;
     pReader->capacity = pReadHead->bodyLen;
@@ -300,14 +305,11 @@ static int32_t walFetchBodyNew(SWalReader *pReader) {
              pReader->pWal->cfg.vgId, pReader->pHead->head.version, ver);
       terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
     }
-//    pRead->curInvalid = 1;
     return -1;
   }
 
   if (walValidBodyCksum(pReader->pHead) != 0) {
     wError("vgId:%d, wal fetch body error:%" PRId64 ", since body checksum not passed", pReader->pWal->cfg.vgId, ver);
-//    pRead->curInvalid = 1;
-
     terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
     return -1;
   }
@@ -363,7 +365,9 @@ int32_t walFetchHead(SWalReader *pRead, int64_t ver, SWalCkHead *pHead) {
     if (contLen == sizeof(SWalCkHead)) {
       break;
     } else if (contLen == 0 && !seeked) {
-      walReadSeekVerImpl(pRead, ver);
+      if(walReadSeekVerImpl(pRead, ver) < 0){
+        return -1;
+      }
       seeked = true;
       continue;
     } else {
@@ -497,7 +501,10 @@ int32_t walReadVer(SWalReader *pReader, int64_t ver) {
     if (contLen == sizeof(SWalCkHead)) {
       break;
     } else if (contLen == 0 && !seeked) {
-      walReadSeekVerImpl(pReader, ver);
+      if(walReadSeekVerImpl(pReader, ver) < 0){
+        taosThreadMutexUnlock(&pReader->mutex);
+        return -1;
+      }
       seeked = true;
       continue;
     } else {

source/libs/wal/src/walRef.c

@@ -45,7 +45,7 @@ void walCloseRef(SWal *pWal, int64_t refId) {
   taosMemoryFree(pRef);
 }
 
-int32_t walRefVer(SWalRef *pRef, int64_t ver) {
+int32_t walSetRefVer(SWalRef *pRef, int64_t ver) {
   SWal *pWal = pRef->pWal;
   wDebug("vgId:%d, wal ref version %" PRId64 ", refId %" PRId64, pWal->cfg.vgId, ver, pRef->refId);
   if (pRef->refVer != ver) {
@@ -57,26 +57,12 @@ int32_t walRefVer(SWalRef *pRef, int64_t ver) {
     }
 
     pRef->refVer = ver;
-    // bsearch in fileSet
-    //    SWalFileInfo tmpInfo;
-    //    tmpInfo.firstVer = ver;
-    //    SWalFileInfo *pRet = taosArraySearch(pWal->fileInfoSet, &tmpInfo, compareWalFileInfo, TD_LE);
-    //    ASSERT(pRet != NULL);
-    //    pRef->refFile = pRet->firstVer;
-
     taosThreadMutexUnlock(&pWal->mutex);
   }
 
   return 0;
 }
 
-#if 1
-void walUnrefVer(SWalRef *pRef) {
-  pRef->refId = -1;
-  //  pRef->refFile = -1;
-}
-#endif
-
 SWalRef *walRefFirstVer(SWal *pWal, SWalRef *pRef) {
   if (pRef == NULL) {
     pRef = walOpenRef(pWal);
@@ -87,12 +73,6 @@ SWalRef *walRefFirstVer(SWal *pWal, SWalRef *pRef) {
   taosThreadMutexLock(&pWal->mutex);
   int64_t ver = walGetFirstVer(pWal);
   pRef->refVer = ver;
-  // bsearch in fileSet
-  //  SWalFileInfo tmpInfo;
-  //  tmpInfo.firstVer = ver;
-  //  SWalFileInfo *pRet = taosArraySearch(pWal->fileInfoSet, &tmpInfo, compareWalFileInfo, TD_LE);
-  //  ASSERT(pRet != NULL);
-  //  pRef->refFile = pRet->firstVer;
 
   taosThreadMutexUnlock(&pWal->mutex);
   wDebug("vgId:%d, wal ref version %" PRId64 " for first", pWal->cfg.vgId, ver);

source/util/src/tarray.c

@@ -316,6 +316,10 @@ SArray* taosArrayFromList(const void* src, size_t size, size_t elemSize) {
 }
 
 SArray* taosArrayDup(const SArray* pSrc, __array_item_dup_fn_t fn) {
+  if (NULL == pSrc) {
+    return NULL;
+  }
+  
   if (pSrc->size == 0) {  // empty array list
     return taosArrayInit(8, pSrc->elemSize);
   }

source/util/src/tcompression.c

@@ -912,11 +912,11 @@ int32_t tsCompressDoubleImp(const char *const input, const int32_t nelements, ch
   return opos;
 }
 
-uint64_t decodeDoubleValue(const char *const input, int32_t *const ipos, uint8_t flag) {
+FORCE_INLINE uint64_t decodeDoubleValue(const char *const input, int32_t *const ipos, uint8_t flag) {
   uint64_t diff = 0ul;
-  int32_t  nbytes = (flag & INT8MASK(3)) + 1;
+  int32_t  nbytes = (flag & 0x7) + 1;
   for (int32_t i = 0; i < nbytes; i++) {
-    diff = diff | ((INT64MASK(8) & input[(*ipos)++]) << BITS_PER_BYTE * i);
+    diff |= (((uint64_t)0xff & input[(*ipos)++]) << BITS_PER_BYTE * i);
   }
   int32_t shift_width = (LONG_BYTES * BITS_PER_BYTE - nbytes * BITS_PER_BYTE) * (flag >> 3);
   diff <<= shift_width;
@@ -936,25 +936,22 @@ int32_t tsDecompressDoubleImp(const char *const input, const int32_t nelements,
   uint8_t  flags = 0;
   int32_t  ipos = 1;
   int32_t  opos = 0;
-  uint64_t prev_value = 0;
+  uint64_t diff = 0;
+  union {
+    uint64_t bits;
+    double   real;
+  } curr;
+
+  curr.bits = 0;
 
   for (int32_t i = 0; i < nelements; i++) {
     if ((i & 0x01) == 0) {
       flags = input[ipos++];
     }
 
-    uint8_t flag = flags & INT8MASK(4);
+    diff = decodeDoubleValue(input, &ipos, flags & 0x0f);
     flags >>= 4;
-
+    curr.bits ^= diff;
-    uint64_t diff = decodeDoubleValue(input, &ipos, flag);
-    union {
-      uint64_t bits;
-      double   real;
-    } curr;
-
-    uint64_t predicted = prev_value;
-    curr.bits = predicted ^ diff;
-    prev_value = curr.bits;
 
     ostream[opos++] = curr.real;
   }

source/util/src/tlog.c

@@ -664,7 +664,7 @@ static int32_t taosPushLogBuffer(SLogBuff *pLogBuf, const char *msg, int32_t msg
   int32_t        end = 0;
   int32_t        remainSize = 0;
   static int64_t lostLine = 0;
-  char           tmpBuf[128] = {0};
+  char           tmpBuf[128];
   int32_t        tmpBufLen = 0;
 
   if (pLogBuf == NULL || pLogBuf->stop) return -1;

tests/parallel_test/cases.task

@@ -495,7 +495,8 @@
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/stbFilter.py
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqCheckData.py
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqCheckData1.py
-,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqConsumerGroup.py
+#,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqConsumerGroup.py
+,,n,system-test,python3 ./test.py -f 7-tmq/tmqConsumerGroup.py
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqAlterSchema.py
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqConsFromTsdb.py -N 3 -n 3
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmqConsFromTsdb1.py -N 3 -n 3
@@ -535,6 +536,7 @@
 ,,y,system-test,./pytest.sh python3 ./test.py -f 7-tmq/tmq3mnodeSwitch.py -N 6 -M 3 -n 3 
 ,,y,system-test,./pytest.sh python3 ./test.py -f 99-TDcase/TD-19201.py
 ,,y,system-test,./pytest.sh python3 ./test.py -f 99-TDcase/TD-21561.py
+,,y,system-test,./pytest.sh python3 ./test.py -f 99-TDcase/TS-3404.py
 
 ,,y,system-test,./pytest.sh python3 ./test.py -f 0-others/balance_vgroups_r1.py -N 6
 ,,y,system-test,./pytest.sh python3 ./test.py -f 0-others/taosShell.py

tests/script/tsim/alter/table.sim

@@ -657,36 +657,33 @@ if $data20 != null then
   return -1
 endi
 
-#print =============== error for normal table
+print =============== error for normal table
-#sql create table tb2023(ts timestamp, f int);
+sql create table tb2023(ts timestamp, f int);
-#sql_error alter table tb2023 add column v varchar(65535);
+sql_error alter table tb2023 add column v varchar(65518);
-#sql_error alter table tb2023 add column v varchar(65535);
+sql_error alter table tb2023 add column v varchar(65531);
-#sql_error alter table tb2023 add column v varchar(65530);
+sql_error alter table tb2023 add column v varchar(65535);
-#sql alter table tb2023 add column v varchar(16374);
+sql alter table tb2023 add column v varchar(65517);
-#sql_error alter table tb2023 modify column v varchar(65536);
+sql_error alter table tb2023 modify column v varchar(65518);
-#sql desc tb2023
+sql desc tb2023
-#sql alter table tb2023 drop column v
+sql alter table tb2023 drop column v
-#sql_error alter table tb2023 add column v nchar(16384);
+sql_error alter table tb2023 add column v nchar(16380);
-#sql alter table tb2023 add column v nchar(4093);
+sql alter table tb2023 add column v nchar(16379);
-#sql_error alter table tb2023 modify column v nchar(16384);
+sql_error alter table tb2023 modify column v nchar(16380);
-#sql_error alter table tb2023 add column v nchar(16384);
+sql desc tb2023
-#sql alter table tb2023 drop column v
+
-#sql alter table tb2023 add column v nchar(16374);
+print =============== error for super table
-#sql desc tb2023
+sql create table stb2023(ts timestamp, f int) tags(t1 int);
-#
+sql_error alter table stb2023 add column v varchar(65518);
-#print =============== error for super table
+sql_error alter table stb2023 add column v varchar(65531);
-#sql create table stb2023(ts timestamp, f int) tags(t1 int);
+sql_error alter table stb2023 add column v varchar(65535);
-#sql_error alter table stb2023 add column v varchar(65535);
+sql alter table stb2023 add column v varchar(65517);
-#sql_error alter table stb2023 add column v varchar(65536);
+sql_error alter table stb2023 modify column v varchar(65518);
-#sql_error alter table stb2023 add column v varchar(33100);
+sql desc stb2023
-#sql alter table stb2023 add column v varchar(16374);
+sql alter table stb2023 drop column v
-#sql_error alter table stb2023 modify column v varchar(16375);
+sql_error alter table stb2023 add column v nchar(16380);
-#sql desc stb2023
+sql alter table stb2023 add column v nchar(16379);
-#sql alter table stb2023 drop column v
+sql_error alter table stb2023 modify column v nchar(16380);
-#sql_error alter table stb2023 add column v nchar(4094);
+sql desc stb2023
-#sql alter table stb2023 add column v nchar(4093);
-#sql_error alter table stb2023 modify column v nchar(4094);
-#sql desc stb2023
 
 print ======= over
 sql drop database d1

tests/script/tsim/parser/alter_column.sim

@@ -48,7 +48,7 @@ sql_error alter table tb modify column c2 binary(10);
 sql_error alter table tb modify column c2 binary(9);
 sql_error alter table tb modify column c2 binary(-9);
 sql_error alter table tb modify column c2 binary(0);
-sql_error alter table tb modify column c2 binary(65600);
+sql_error alter table tb modify column c2 binary(65436);
 sql_error alter table tb modify column c2 nchar(30);
 sql_error alter table tb modify column c3 double;
 sql_error alter table tb modify column c3 nchar(10);

tests/script/tsim/query/tableCount.sim

@@ -103,7 +103,14 @@ endi
 if $data62 != 5 then
   return -1
 endi
-
+sql select count(table_name) from information_schema.ins_tables where db_name='db1' and stable_name='sta' group by stable_name
+print $rows , $data00
+if $rows != 1 then
+  return -1
+endi
+if $data00 != 8 then
+  return -1
+endi
 sql select distinct db_name from information_schema.ins_tables;
 print $rows
 if $rows != 4 then

tests/system-test/7-tmq/tmqCommon.py

@@ -145,29 +145,25 @@ class TMQCom:
             processID = subprocess.check_output(psCmd, shell=True).decode("utf-8")
         tdLog.debug("%s is stopped by kill -INT" % (processorName))
 
-    def getStartConsumeNotifyFromTmqsim(self,cdbName='cdb'):
+    def getStartConsumeNotifyFromTmqsim(self,cdbName='cdb',rows=1):
         loopFlag = 1
         while loopFlag:
-            tdSql.query("select * from %s.notifyinfo"%cdbName)
+            tdSql.query("select * from %s.notifyinfo where cmdid = 0"%cdbName)
-            #tdLog.info("row: %d, %l64d, %l64d"%(tdSql.getData(0, 1),tdSql.getData(0, 2),tdSql.getData(0, 3))
             actRows = tdSql.getRows()
-            for i in range(actRows):
+            tdLog.info("row: %d"%(actRows))
-                if tdSql.getData(i, 1) == 0:
+            if (actRows >= rows):
                     loopFlag = 0
-                    break
             time.sleep(0.02)
         return
 
-    def getStartCommitNotifyFromTmqsim(self,cdbName='cdb'):
+    def getStartCommitNotifyFromTmqsim(self,cdbName='cdb',rows=1):
         loopFlag = 1
         while loopFlag:
-            tdSql.query("select * from %s.notifyinfo"%cdbName)
+            tdSql.query("select * from %s.notifyinfo where cmdid = 1"%cdbName)
-            #tdLog.info("row: %d, %l64d, %l64d"%(tdSql.getData(0, 1),tdSql.getData(0, 2),tdSql.getData(0, 3))
             actRows = tdSql.getRows()
-            for i in range(actRows):
+            tdLog.info("row: %d"%(actRows))
-                if tdSql.getData(i, 1) == 1:
+            if (actRows >= rows):
-                    loopFlag = 0
+                loopFlag = 0
-                    break
             time.sleep(0.02)
         return
 

tests/system-test/7-tmq/tmqConsumerGroup.py

@@ -100,7 +100,7 @@ class TDTestCase:
 
         tdLog.info("wait consumer commit notify")
         # tmqCom.getStartCommitNotifyFromTmqsim(rows=4)
-        tmqCom.getStartConsumeNotifyFromTmqsim()
+        tmqCom.getStartConsumeNotifyFromTmqsim(rows=2)
 
         tdLog.info("pkill one consume processor")
         tmqCom.stopTmqSimProcess('tmq_sim_new')

tests/system-test/99-TDcase/TS-3404.py

@@ -0,0 +1,101 @@
+import taos
+import sys
+import time
+import socket
+import os
+import threading
+
+from util.log import *
+from util.sql import *
+from util.cases import *
+from util.dnodes import *
+
+class TDTestCase:
+    hostname = socket.gethostname()
+
+    def init(self, conn, logSql, replicaVar=1):
+        self.replicaVar = int(replicaVar)
+        tdLog.debug(f"start to excute {__file__}")
+        #tdSql.init(conn.cursor())
+        tdSql.init(conn.cursor(), logSql)  # output sql.txt file
+
+    def getBuildPath(self):
+        selfPath = os.path.dirname(os.path.realpath(__file__))
+
+        if ("community" in selfPath):
+            projPath = selfPath[:selfPath.find("community")]
+        else:
+            projPath = selfPath[:selfPath.find("tests")]
+
+        for root, dirs, files in os.walk(projPath):
+            if ("taosd" in files or "taosd.exe" in files):
+                rootRealPath = os.path.dirname(os.path.realpath(root))
+                if ("packaging" not in rootRealPath):
+                    buildPath = root[:len(root) - len("/build/bin")]
+                    break
+        return buildPath
+
+    def create_tables(self):
+        tdSql.execute(f"CREATE STABLE `stb5` (`ts` TIMESTAMP, `ip_value` FLOAT, `ip_quality` INT) TAGS (`t1` INT)")
+        tdSql.execute(f"CREATE TABLE `t_11` USING `stb5` (`t1`) TAGS (1)")
+
+    def insert_data(self):
+        tdLog.debug("start to insert data ............")
+
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-10 09:30:47.722', 10.30000, 100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-10 09:30:56.383', 12.30000, 100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-10 09:48:55.778', 13.30000, 100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-10 09:51:50.821', 9.30000,  100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-10 09:58:07.162', 9.30000,  100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-10 13:41:16.075', 9.30000,  100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-13 14:12:58.318', 21.00000, 100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-13 14:13:21.328', 1.10000,  100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-13 14:35:24.258', 1.30000,  100)")
+        tdSql.execute(f"INSERT INTO `t_11` VALUES ('2023-05-13 16:56:49.033', 1.80000,  100)")
+
+        tdLog.debug("insert data ............ [OK]")
+
+    def run(self):
+        tdSql.prepare()
+        self.create_tables()
+        self.insert_data()
+        tdLog.printNoPrefix("======== test TS-3404")
+
+        tdSql.query(f"select _irowts, interp(ip_value) from t_11 range('2023-05-13 14:00:00', '2023-05-13 15:00:00') every(300s) fill(linear);")
+        tdSql.checkRows(13)
+
+        tdSql.checkData(0,  0, '2023-05-13 14:00:00.000')
+        tdSql.checkData(1,  0, '2023-05-13 14:05:00.000')
+        tdSql.checkData(2,  0, '2023-05-13 14:10:00.000')
+        tdSql.checkData(3,  0, '2023-05-13 14:15:00.000')
+        tdSql.checkData(4,  0, '2023-05-13 14:20:00.000')
+        tdSql.checkData(5,  0, '2023-05-13 14:25:00.000')
+        tdSql.checkData(6,  0, '2023-05-13 14:30:00.000')
+        tdSql.checkData(7,  0, '2023-05-13 14:35:00.000')
+        tdSql.checkData(8,  0, '2023-05-13 14:40:00.000')
+        tdSql.checkData(9,  0, '2023-05-13 14:45:00.000')
+        tdSql.checkData(10, 0, '2023-05-13 14:50:00.000')
+        tdSql.checkData(11, 0, '2023-05-13 14:55:00.000')
+        tdSql.checkData(12, 0, '2023-05-13 15:00:00.000')
+
+        tdSql.checkData(0,  1, 20.96512)
+        tdSql.checkData(1,  1, 20.97857)
+        tdSql.checkData(2,  1, 20.99201)
+        tdSql.checkData(3,  1, 1.114917)
+        tdSql.checkData(4,  1, 1.160271)
+        tdSql.checkData(5,  1, 1.205625)
+        tdSql.checkData(6,  1, 1.250978)
+        tdSql.checkData(7,  1, 1.296333)
+        tdSql.checkData(8,  1, 1.316249)
+        tdSql.checkData(9,  1, 1.333927)
+        tdSql.checkData(10, 1, 1.351607)
+        tdSql.checkData(11, 1, 1.369285)
+        tdSql.checkData(12, 1, 1.386964)
+
+
+    def stop(self):
+        tdSql.close()
+        tdLog.success(f"{__file__} successfully executed")
+
+tdCases.addLinux(__file__, TDTestCase())
+tdCases.addWindows(__file__, TDTestCase())