Merge remote-tracking branch 'origin/3.0' into test/distributed_aggregate

.gitmodules

@@ -16,3 +16,6 @@
 [submodule "tools/taos-tools"]
 	path = tools/taos-tools
 	url = https://github.com/taosdata/taos-tools
+[submodule "tools/taosadapter"]
+	path = tools/taosadapter
+	url = https://github.com/taosdata/taosadapter.git

cmake/cmake.define

@@ -18,6 +18,33 @@ if (NOT DEFINED TD_GRANT)
   SET(TD_GRANT FALSE)
 endif()
 
+IF ("${BUILD_HTTP}" STREQUAL "")
+  IF (TD_LINUX)
+    IF (TD_ARM_32)
+      SET(TD_BUILD_HTTP TRUE)
+    ELSE ()
+      SET(TD_BUILD_HTTP TRUE)
+    ENDIF ()
+  ELSEIF (TD_DARWIN)
+    SET(TD_BUILD_HTTP TRUE)
+  ELSE ()
+    SET(TD_BUILD_HTTP TRUE)
+  ENDIF ()
+ELSEIF (${BUILD_HTTP} MATCHES "false")
+  SET(TD_BUILD_HTTP FALSE)
+ELSEIF (${BUILD_HTTP} MATCHES "true")
+  SET(TD_BUILD_HTTP TRUE)
+ELSEIF (${BUILD_HTTP} MATCHES "internal")
+  SET(TD_BUILD_HTTP FALSE)
+  SET(TD_BUILD_TAOSA_INTERNAL TRUE)
+ELSE ()
+  SET(TD_BUILD_HTTP TRUE)
+ENDIF ()
+
+IF (TD_BUILD_HTTP)
+  ADD_DEFINITIONS(-DHTTP_EMBEDDED)
+ENDIF ()
+
 IF ("${BUILD_TOOLS}" STREQUAL "")
   IF (TD_LINUX)
     IF (TD_ARM_32)

include/common/tdatablock.h

@@ -234,9 +234,6 @@ int32_t buildSubmitReqFromDataBlock(SSubmitReq** pReq, const SArray* pDataBlocks
 
 char* buildCtbNameByGroupId(const char* stbName, uint64_t groupId);
 
-SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pSchema, bool createTb, int64_t suid,
-                            const char* stbFullName, int32_t vgId);
-
 static FORCE_INLINE int32_t blockGetEncodeSize(const SSDataBlock* pBlock) {
   return blockDataGetSerialMetaSize(pBlock->info.numOfCols) + blockDataGetSize(pBlock);
 }

include/common/tmsg.h

@@ -1134,14 +1134,16 @@ void    tFreeSMAlterStbRsp(SMAlterStbRsp* pRsp);
 int32_t tSerializeSTableMetaRsp(void* buf, int32_t bufLen, STableMetaRsp* pRsp);
 int32_t tDeserializeSTableMetaRsp(void* buf, int32_t bufLen, STableMetaRsp* pRsp);
 void    tFreeSTableMetaRsp(STableMetaRsp* pRsp);
+void tFreeSTableIndexRsp(void *info);
 
 typedef struct {
-  SArray* pArray;  // Array of STableMetaRsp
+  SArray*         pMetaRsp;  // Array of STableMetaRsp
-} STableMetaBatchRsp;
+  SArray*         pIndexRsp;  // Array of STableIndexRsp;
+} SSTbHbRsp;
 
-int32_t tSerializeSTableMetaBatchRsp(void* buf, int32_t bufLen, STableMetaBatchRsp* pRsp);
+int32_t tSerializeSSTbHbRsp(void* buf, int32_t bufLen, SSTbHbRsp* pRsp);
-int32_t tDeserializeSTableMetaBatchRsp(void* buf, int32_t bufLen, STableMetaBatchRsp* pRsp);
+int32_t tDeserializeSSTbHbRsp(void* buf, int32_t bufLen, SSTbHbRsp* pRsp);
-void    tFreeSTableMetaBatchRsp(STableMetaBatchRsp* pRsp);
+void    tFreeSSTbHbRsp(SSTbHbRsp* pRsp);
 
 typedef struct {
   int32_t numOfTables;
@@ -2306,6 +2308,29 @@ int32_t tDecodeSMqOffset(SDecoder* decoder, SMqOffset* pOffset);
 int32_t tEncodeSMqCMCommitOffsetReq(SEncoder* encoder, const SMqCMCommitOffsetReq* pReq);
 int32_t tDecodeSMqCMCommitOffsetReq(SDecoder* decoder, SMqCMCommitOffsetReq* pReq);
 
+// tqOffset
+enum {
+  TMQ_OFFSET__SNAPSHOT = 1,
+  TMQ_OFFSET__LOG,
+};
+
+typedef struct {
+  int8_t type;
+  union {
+    struct {
+      int64_t uid;
+      int64_t ts;
+    };
+    struct {
+      int64_t version;
+    };
+  };
+  char subKey[TSDB_SUBSCRIBE_KEY_LEN];
+} STqOffset;
+
+int32_t tEncodeSTqOffset(SEncoder* pEncoder, const STqOffset* pOffset);
+int32_t tDecodeSTqOffset(SDecoder* pDecoder, STqOffset* pOffset);
+
 typedef struct {
   char    name[TSDB_TABLE_FNAME_LEN];
   char    stb[TSDB_TABLE_FNAME_LEN];
@@ -2402,6 +2427,7 @@ typedef struct {
 
 static FORCE_INLINE void tDestroyTSma(STSma* pSma) {
   if (pSma) {
+    taosMemoryFreeClear(pSma->dstTbName);
     taosMemoryFreeClear(pSma->expr);
     taosMemoryFreeClear(pSma->tagsFilter);
   }
@@ -2430,7 +2456,7 @@ int32_t tEncodeSVCreateTSmaReq(SEncoder* pCoder, const SVCreateTSmaReq* pReq);
 int32_t tDecodeSVCreateTSmaReq(SDecoder* pCoder, SVCreateTSmaReq* pReq);
 
 int32_t tEncodeTSma(SEncoder* pCoder, const STSma* pSma);
-int32_t tDecodeTSma(SDecoder* pCoder, STSma* pSma);
+int32_t tDecodeTSma(SDecoder* pCoder, STSma* pSma, bool deepCopy);
 
 static int32_t tEncodeTSmaWrapper(SEncoder* pEncoder, const STSmaWrapper* pReq) {
   if (tEncodeI32(pEncoder, pReq->number) < 0) return -1;
@@ -2440,10 +2466,10 @@ static int32_t tEncodeTSmaWrapper(SEncoder* pEncoder, const STSmaWrapper* pReq)
   return 0;
 }
 
-static int32_t tDecodeTSmaWrapper(SDecoder* pDecoder, STSmaWrapper* pReq) {
+static int32_t tDecodeTSmaWrapper(SDecoder* pDecoder, STSmaWrapper* pReq, bool deepCopy) {
   if (tDecodeI32(pDecoder, &pReq->number) < 0) return -1;
   for (int32_t i = 0; i < pReq->number; ++i) {
-    tDecodeTSma(pDecoder, pReq->tSma + i);
+    tDecodeTSma(pDecoder, pReq->tSma + i, deepCopy);
   }
   return 0;
 }
@@ -2502,7 +2528,11 @@ typedef struct {
 } STableIndexInfo;
 
 typedef struct {
-  SArray* pIndex;
+  char     tbName[TSDB_TABLE_NAME_LEN];
+  char     dbFName[TSDB_DB_FNAME_LEN];
+  uint64_t suid;
+  int32_t  version;
+  SArray*  pIndex;
 } STableIndexRsp;
 
 int32_t tSerializeSTableIndexRsp(void* buf, int32_t bufLen, const STableIndexRsp* pRsp);

include/common/tmsgdef.h

@@ -97,7 +97,6 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_MNODE, "create-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_MNODE, "alter-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_MNODE, "drop-mnode", NULL, NULL)
-  TD_DEF_MSG_TYPE(TDMT_MND_SET_STANDBY, "set-mnode-standby", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_QNODE, "create-qnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_QNODE, "alter-qnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_QNODE, "drop-qnode", NULL, NULL)
@@ -141,7 +140,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_MQ_CONSUMER_RECOVER, "consumer-recover", SMqConsumerRecoverMsg, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_MQ_DO_REBALANCE, "do-rebalance", SMqDoRebalanceMsg, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_MQ_DROP_CGROUP, "drop-cgroup", SMqDropCGroupReq, SMqDropCGroupRsp)
-  TD_DEF_MSG_TYPE(TDMT_MND_MQ_COMMIT_OFFSET, "commit-offset", SMqCMCommitOffsetReq, SMqCMCommitOffsetRsp)
+  TD_DEF_MSG_TYPE(TDMT_MND_MQ_COMMIT_OFFSET, "mnode-commit-offset", SMqCMCommitOffsetReq, SMqCMCommitOffsetRsp)
   TD_DEF_MSG_TYPE(TDMT_MND_MQ_TIMER, "mq-tmr", SMTimerReq, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_TELEM_TIMER, "telem-tmr", SMTimerReq, SMTimerReq)
   TD_DEF_MSG_TYPE(TDMT_MND_TRANS_TIMER, "trans-tmr", NULL, NULL)
@@ -177,6 +176,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_DROP_STB, "vnode-drop-stb", SVDropStbReq, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_VG_CHANGE, "vnode-mq-vg-change", SMqRebVgReq, SMqRebVgRsp)
   TD_DEF_MSG_TYPE(TDMT_VND_MQ_VG_DELETE, "vnode-mq-vg-delete", SMqVDeleteReq, SMqVDeleteRsp)
+  TD_DEF_MSG_TYPE(TDMT_VND_MQ_COMMIT_OFFSET, "vnode-commit-offset", STqOffset, STqOffset)
   TD_DEF_MSG_TYPE(TDMT_VND_CANCEL_TASK, "vnode-cancel-task", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_DROP_TASK, "vnode-drop-task", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_CREATE_TOPIC, "vnode-create-topic", NULL, NULL)
@@ -238,6 +238,8 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_SYNC_SNAPSHOT_SEND, "sync-snapshot-send", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_SYNC_SNAPSHOT_RSP, "sync-snapshot-rsp", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_SYNC_LEADER_TRANSFER, "sync-leader-transfer", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_SYNC_SET_MNODE_STANDBY, "set-mnode-standby", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_SYNC_SET_VNODE_STANDBY, "set-vnode-standby", NULL, NULL)
   
 #if defined(TD_MSG_NUMBER_)
   TDMT_MAX

include/dnode/mnode/mnode.h

@@ -59,7 +59,13 @@ void mndClose(SMnode *pMnode);
  * @param pMnode The mnode object.
  */
 int32_t mndStart(SMnode *pMnode);
-void    mndStop(SMnode *pMnode);
+
+/**
+ * @brief Stop mnode
+ *
+ * @param pMnode The mnode object.
+ */
+void mndStop(SMnode *pMnode);
 
 /**
  * @brief Get mnode monitor info.
@@ -71,17 +77,25 @@ void    mndStop(SMnode *pMnode);
  * @return int32_t 0 for success, -1 for failure.
  */
 int32_t mndGetMonitorInfo(SMnode *pMnode, SMonClusterInfo *pCluster, SMonVgroupInfo *pVgroup, SMonGrantInfo *pGrant);
+
+/**
+ * @brief Get mnode loads for status msg.
+ *
+ * @param pMnode The mnode object.
+ * @param pLoad
+ * @return int32_t 0 for success, -1 for failure.
+ */
 int32_t mndGetLoad(SMnode *pMnode, SMnodeLoad *pLoad);
 
 /**
- * @brief Process the read, write, sync request.
+ * @brief Process the rpc, sync request.
  *
  * @param pMsg The request msg.
  * @return int32_t 0 for success, -1 for failure.
  */
 int32_t mndProcessRpcMsg(SRpcMsg *pMsg);
 int32_t mndProcessSyncMsg(SRpcMsg *pMsg);
-int32_t mndPreprocessQueryMsg(SMnode * pMnode, SRpcMsg * pMsg);
+int32_t mndPreProcessMsg(SRpcMsg *pMsg);
 
 /**
  * @brief Generate machine code

include/libs/catalog/catalog.h

@@ -98,14 +98,15 @@ typedef struct SCatalogCfg {
   uint32_t stbRentSec;
 } SCatalogCfg;
 
-typedef struct SSTableMetaVersion {
+typedef struct SSTableVersion {
   char     dbFName[TSDB_DB_FNAME_LEN];
   char     stbName[TSDB_TABLE_NAME_LEN];
   uint64_t dbId;
   uint64_t suid;
   int16_t  sversion;
-  int16_t  tversion;
+  int16_t  tversion;  
-} SSTableMetaVersion;
+  int32_t  smaVer;
+} SSTableVersion;
 
 typedef struct SDbVgVersion {
   char    dbFName[TSDB_DB_FNAME_LEN];
@@ -267,7 +268,7 @@ int32_t catalogAsyncGetAllMeta(SCatalog* pCtg, SRequestConnInfo* pConn, uint64_t
 
 int32_t catalogGetQnodeList(SCatalog* pCatalog, SRequestConnInfo* pConn, SArray* pQnodeList);
 
-int32_t catalogGetExpiredSTables(SCatalog* pCatalog, SSTableMetaVersion** stables, uint32_t* num);
+int32_t catalogGetExpiredSTables(SCatalog* pCatalog, SSTableVersion **stables, uint32_t *num);
 
 int32_t catalogGetExpiredDBs(SCatalog* pCatalog, SDbVgVersion** dbs, uint32_t* num);
 
@@ -279,6 +280,8 @@ int32_t catalogGetIndexMeta(SCatalog* pCtg, SRequestConnInfo* pConn, const char*
 
 int32_t catalogGetTableIndex(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName, SArray** pRes);
 
+int32_t catalogUpdateTableIndex(SCatalog* pCtg, STableIndexRsp *pRsp);
+
 int32_t catalogGetUdfInfo(SCatalog* pCtg, SRequestConnInfo* pConn, const char* funcName, SFuncInfo* pInfo);
 
 int32_t catalogChkAuth(SCatalog* pCtg, SRequestConnInfo* pConn, const char* user, const char* dbFName, AUTH_TYPE type, bool *pass);

include/libs/function/functionMgt.h

@@ -142,6 +142,10 @@ typedef enum EFunctionType {
   FUNCTION_TYPE_FIRST_MERGE,
   FUNCTION_TYPE_LAST_PARTIAL,
   FUNCTION_TYPE_LAST_MERGE,
+  FUNCTION_TYPE_AVG_PARTIAL,
+  FUNCTION_TYPE_AVG_MERGE,
+  FUNCTION_TYPE_STDDEV_PARTIAL,
+  FUNCTION_TYPE_STDDEV_MERGE,
 
   // user defined funcion
   FUNCTION_TYPE_UDF = 10000

include/libs/parser/parser.h

@@ -65,6 +65,7 @@ void qDestroyQuery(SQuery* pQueryNode);
 
 int32_t qExtractResultSchema(const SNode* pRoot, int32_t* numOfCols, SSchema** pSchema);
 int32_t qSetSTableIdForRsma(SNode* pStmt, int64_t uid);
+void    qCleanupKeywordsTable();
 
 int32_t     qBuildStmtOutput(SQuery* pQuery, SHashObj* pVgHash, SHashObj* pBlockHash);
 int32_t     qResetStmtDataBlock(void* block, bool keepBuf);

include/libs/sync/sync.h

@@ -47,8 +47,9 @@ typedef enum {
 typedef enum {
   TAOS_SYNC_PROPOSE_SUCCESS = 0,
   TAOS_SYNC_PROPOSE_NOT_LEADER = 1,
-  TAOS_SYNC_PROPOSE_OTHER_ERROR = 2,
+  TAOS_SYNC_ONLY_ONE_REPLICA = 2,
-  TAOS_SYNC_ONLY_ONE_REPLICA = 3,
+  TAOS_SYNC_NOT_IN_NEW_CONFIG = 3,
+  TAOS_SYNC_OTHER_ERROR = 100,
 } ESyncProposeCode;
 
 typedef enum {
@@ -110,6 +111,7 @@ typedef struct SSyncFSM {
 
   void (*FpRestoreFinishCb)(struct SSyncFSM* pFsm);
   void (*FpReConfigCb)(struct SSyncFSM* pFsm, const SRpcMsg* pMsg, SReConfigCbMeta cbMeta);
+  void (*FpLeaderTransferCb)(struct SSyncFSM* pFsm, const SRpcMsg* pMsg, SFsmCbMeta cbMeta);
 
   int32_t (*FpGetSnapshot)(struct SSyncFSM* pFsm, SSnapshot* pSnapshot);
 
@@ -199,15 +201,13 @@ bool        syncIsRestoreFinish(int64_t rid);
 int32_t     syncGetSnapshotMeta(int64_t rid, struct SSnapshotMeta* sMeta);
 
 int32_t syncReconfig(int64_t rid, const SSyncCfg* pNewCfg);
-int32_t syncReconfigRaw(int64_t rid, const SSyncCfg* pNewCfg, SRpcMsg* pRpcMsg);
+
+// build SRpcMsg, need to call syncPropose with SRpcMsg
+int32_t syncReconfigBuild(int64_t rid, const SSyncCfg* pNewCfg, SRpcMsg* pRpcMsg);
 
 int32_t syncLeaderTransfer(int64_t rid);
 int32_t syncLeaderTransferTo(int64_t rid, SNodeInfo newLeader);
 
-// to be moved to static
-void syncStartNormal(int64_t rid);
-void syncStartStandBy(int64_t rid);
-
 #ifdef __cplusplus
 }
 #endif

include/libs/sync/syncTools.h

@@ -467,6 +467,7 @@ typedef struct SyncLeaderTransfer {
    SRaftId  srcId;
    SRaftId  destId;
    */
+  SNodeInfo newNodeInfo;
   SRaftId newLeaderId;
 } SyncLeaderTransfer;
 

include/util/taoserror.h

@@ -653,6 +653,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_INVALID_DELETE_WHERE      TAOS_DEF_ERROR_CODE(0, 0x2655)
 #define TSDB_CODE_PAR_INVALID_REDISTRIBUTE_VG   TAOS_DEF_ERROR_CODE(0, 0x2656)
 #define TSDB_CODE_PAR_FILL_NOT_ALLOWED_FUNC     TAOS_DEF_ERROR_CODE(0, 0x2657)
+#define TSDB_CODE_PAR_INVALID_WINDOW_PC         TAOS_DEF_ERROR_CODE(0, 0x2658)
 
 //planner
 #define TSDB_CODE_PLAN_INTERNAL_ERROR           TAOS_DEF_ERROR_CODE(0, 0x2700)
@@ -684,17 +685,23 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_SML_INVALID_DB_CONF           TAOS_DEF_ERROR_CODE(0, 0x3003)
 
 //tsma
-#define TSDB_CODE_TSMA_ALREADY_EXIST             TAOS_DEF_ERROR_CODE(0, 0x3100)
+#define TSDB_CODE_TSMA_INIT_FAILED               TAOS_DEF_ERROR_CODE(0, 0x3100)
-#define TSDB_CODE_TSMA_NO_INDEX_IN_META          TAOS_DEF_ERROR_CODE(0, 0x3101)
+#define TSDB_CODE_TSMA_ALREADY_EXIST             TAOS_DEF_ERROR_CODE(0, 0x3101)
-#define TSDB_CODE_TSMA_INVALID_ENV               TAOS_DEF_ERROR_CODE(0, 0x3102)
+#define TSDB_CODE_TSMA_NO_INDEX_IN_META          TAOS_DEF_ERROR_CODE(0, 0x3102)
-#define TSDB_CODE_TSMA_INVALID_STAT              TAOS_DEF_ERROR_CODE(0, 0x3103)
+#define TSDB_CODE_TSMA_INVALID_ENV               TAOS_DEF_ERROR_CODE(0, 0x3103)
-#define TSDB_CODE_TSMA_NO_INDEX_IN_CACHE         TAOS_DEF_ERROR_CODE(0, 0x3104)
+#define TSDB_CODE_TSMA_INVALID_STAT              TAOS_DEF_ERROR_CODE(0, 0x3104)
-#define TSDB_CODE_TSMA_RM_SKEY_IN_HASH           TAOS_DEF_ERROR_CODE(0, 0x3105)
+#define TSDB_CODE_TSMA_INVALID_PTR               TAOS_DEF_ERROR_CODE(0, 0x3105)
+#define TSDB_CODE_TSMA_INVALID_PARA              TAOS_DEF_ERROR_CODE(0, 0x3106)
+#define TSDB_CODE_TSMA_NO_INDEX_IN_CACHE         TAOS_DEF_ERROR_CODE(0, 0x3107)
+
 
 //rsma
 #define TSDB_CODE_RSMA_INVALID_ENV               TAOS_DEF_ERROR_CODE(0, 0x3150)
 #define TSDB_CODE_RSMA_INVALID_STAT              TAOS_DEF_ERROR_CODE(0, 0x3151)
 
+//index
+#define TSDB_CODE_INDEX_REBUILDING               TAOS_DEF_ERROR_CODE(0, 0x3200)
+
 
 #ifdef __cplusplus
 }

source/client/src/clientHb.c

@@ -99,15 +99,15 @@ static int32_t hbProcessDBInfoRsp(void *value, int32_t valueLen, struct SCatalog
 static int32_t hbProcessStbInfoRsp(void *value, int32_t valueLen, struct SCatalog *pCatalog) {
   int32_t code = 0;
 
-  STableMetaBatchRsp batchMetaRsp = {0};
+  SSTbHbRsp hbRsp = {0};
-  if (tDeserializeSTableMetaBatchRsp(value, valueLen, &batchMetaRsp) != 0) {
+  if (tDeserializeSSTbHbRsp(value, valueLen, &hbRsp) != 0) {
     terrno = TSDB_CODE_INVALID_MSG;
     return -1;
   }
 
-  int32_t numOfBatchs = taosArrayGetSize(batchMetaRsp.pArray);
+  int32_t numOfMeta = taosArrayGetSize(hbRsp.pMetaRsp);
-  for (int32_t i = 0; i < numOfBatchs; ++i) {
+  for (int32_t i = 0; i < numOfMeta; ++i) {
-    STableMetaRsp *rsp = taosArrayGet(batchMetaRsp.pArray, i);
+    STableMetaRsp *rsp = taosArrayGet(hbRsp.pMetaRsp, i);
 
     if (rsp->numOfColumns < 0) {
       tscDebug("hb remove stb, db:%s, stb:%s", rsp->dbFName, rsp->stbName);
@@ -116,7 +116,7 @@ static int32_t hbProcessStbInfoRsp(void *value, int32_t valueLen, struct SCatalo
       tscDebug("hb update stb, db:%s, stb:%s", rsp->dbFName, rsp->stbName);
       if (rsp->pSchemas[0].colId != PRIMARYKEY_TIMESTAMP_COL_ID) {
         tscError("invalid colId[%" PRIi16 "] for the first column in table meta rsp msg", rsp->pSchemas[0].colId);
-        tFreeSTableMetaBatchRsp(&batchMetaRsp);
+        tFreeSSTbHbRsp(&hbRsp);
         return TSDB_CODE_TSC_INVALID_VALUE;
       }
 
@@ -124,7 +124,17 @@ static int32_t hbProcessStbInfoRsp(void *value, int32_t valueLen, struct SCatalo
     }
   }
 
-  tFreeSTableMetaBatchRsp(&batchMetaRsp);
+  int32_t numOfIndex = taosArrayGetSize(hbRsp.pIndexRsp);
+  for (int32_t i = 0; i < numOfIndex; ++i) {
+    STableIndexRsp *rsp = taosArrayGet(hbRsp.pIndexRsp, i);
+
+    catalogUpdateTableIndex(pCatalog, rsp);
+  }
+
+  taosArrayDestroy(hbRsp.pIndexRsp);
+  hbRsp.pIndexRsp = NULL;
+
+  tFreeSSTbHbRsp(&hbRsp);
   return TSDB_CODE_SUCCESS;
 }
 
@@ -455,7 +465,7 @@ int32_t hbGetExpiredDBInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, SCl
 }
 
 int32_t hbGetExpiredStbInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, SClientHbReq *req) {
-  SSTableMetaVersion *stbs = NULL;
+  SSTableVersion *stbs = NULL;
   uint32_t            stbNum = 0;
   int32_t             code = 0;
 
@@ -469,15 +479,16 @@ int32_t hbGetExpiredStbInfo(SClientHbKey *connKey, struct SCatalog *pCatalog, SC
   }
 
   for (int32_t i = 0; i < stbNum; ++i) {
-    SSTableMetaVersion *stb = &stbs[i];
+    SSTableVersion *stb = &stbs[i];
     stb->suid = htobe64(stb->suid);
     stb->sversion = htons(stb->sversion);
     stb->tversion = htons(stb->tversion);
+    stb->smaVer = htonl(stb->smaVer);
   }
 
   SKv kv = {
       .key = HEARTBEAT_KEY_STBINFO,
-      .valueLen = sizeof(SSTableMetaVersion) * stbNum,
+      .valueLen = sizeof(SSTableVersion) * stbNum,
       .value = stbs,
   };
 

source/client/src/clientImpl.c

@@ -206,7 +206,7 @@ int32_t execLocalCmd(SRequestObj* pRequest, SQuery* pQuery) {
   SRetrieveTableRsp* pRsp = NULL;
   int32_t            code = qExecCommand(pQuery->pRoot, &pRsp);
   if (TSDB_CODE_SUCCESS == code && NULL != pRsp) {
-    code = setQueryResultFromRsp(&pRequest->body.resInfo, pRsp, false, false);
+    code = setQueryResultFromRsp(&pRequest->body.resInfo, pRsp, false, true);
   }
 
   return code;
@@ -1415,7 +1415,7 @@ int32_t setResultDataPtr(SReqResultInfo* pResultInfo, TAOS_FIELD* pFields, int32
     int32_t bytes = *(int32_t*)p;
     p += sizeof(int32_t);
 
-    //    ASSERT(type == pFields[i].type && bytes == pFields[i].bytes);
+    /*ASSERT(type == pFields[i].type && bytes == pFields[i].bytes);*/
   }
 
   int32_t* colLength = (int32_t*)p;

source/client/src/clientMain.c

@@ -25,6 +25,7 @@
 #include "tref.h"
 #include "trpc.h"
 #include "version.h"
+#include "functionMgt.h"
 
 #define TSC_VAR_NOT_RELEASE 1
 #define TSC_VAR_RELEASED    0
@@ -61,6 +62,9 @@ void taos_cleanup(void) {
 
   cleanupTaskQueue();
 
+  fmFuncMgtDestroy();
+  qCleanupKeywordsTable();
+
   id = clientConnRefPool;
   clientConnRefPool = -1;
   taosCloseRef(id);

source/client/src/clientSml.c

@@ -2355,34 +2355,34 @@ static int smlProcess(SSmlHandle *info, char *lines[], int numLines) {
 }
 
 static int32_t isSchemalessDb(STscObj *taos, SRequestObj *request) {
-  SCatalog *catalog = NULL;
+//  SCatalog *catalog = NULL;
-  int32_t   code = catalogGetHandle(((STscObj *)taos)->pAppInfo->clusterId, &catalog);
+//  int32_t   code = catalogGetHandle(((STscObj *)taos)->pAppInfo->clusterId, &catalog);
-  if (code != TSDB_CODE_SUCCESS) {
+//  if (code != TSDB_CODE_SUCCESS) {
-    uError("SML get catalog error %d", code);
+//    uError("SML get catalog error %d", code);
-    return code;
+//    return code;
-  }
+//  }
-
+//
-  SName name;
+//  SName name;
-  tNameSetDbName(&name, taos->acctId, taos->db, strlen(taos->db));
+//  tNameSetDbName(&name, taos->acctId, taos->db, strlen(taos->db));
-  char dbFname[TSDB_DB_FNAME_LEN] = {0};
+//  char dbFname[TSDB_DB_FNAME_LEN] = {0};
-  tNameGetFullDbName(&name, dbFname);
+//  tNameGetFullDbName(&name, dbFname);
-  SDbCfgInfo pInfo = {0};
+//  SDbCfgInfo pInfo = {0};
-
+//
-  SRequestConnInfo conn = {0};
+//  SRequestConnInfo conn = {0};
-  conn.pTrans = taos->pAppInfo->pTransporter;
+//  conn.pTrans = taos->pAppInfo->pTransporter;
-  conn.requestId = request->requestId;
+//  conn.requestId = request->requestId;
-  conn.requestObjRefId = request->self;
+//  conn.requestObjRefId = request->self;
-  conn.mgmtEps = getEpSet_s(&taos->pAppInfo->mgmtEp);
+//  conn.mgmtEps = getEpSet_s(&taos->pAppInfo->mgmtEp);
-
+//
-  code = catalogGetDBCfg(catalog, &conn, dbFname, &pInfo);
+//  code = catalogGetDBCfg(catalog, &conn, dbFname, &pInfo);
-  if (code != TSDB_CODE_SUCCESS) {
+//  if (code != TSDB_CODE_SUCCESS) {
-    return code;
+//    return code;
-  }
+//  }
-  taosArrayDestroy(pInfo.pRetensions);
+//  taosArrayDestroy(pInfo.pRetensions);
-
+//
-  if (!pInfo.schemaless) {
+//  if (!pInfo.schemaless) {
-    return TSDB_CODE_SML_INVALID_DB_CONF;
+//    return TSDB_CODE_SML_INVALID_DB_CONF;
-  }
+//  }
   return TSDB_CODE_SUCCESS;
 }
 

source/client/src/tmq.c

@@ -132,6 +132,7 @@ typedef struct {
   // statistics
   int64_t pollCnt;
   // offset
+  int64_t committedOffset;
   int64_t currentOffset;
   // connection info
   int32_t vgId;
@@ -193,6 +194,26 @@ typedef struct {
   void*          userParam;
 } SMqCommitCbParam;
 
+typedef struct {
+  tmq_t*         tmq;
+  int8_t         automatic;
+  int8_t         async;
+  int8_t         freeOffsets;
+  int32_t        waitingRspNum;
+  int32_t        totalRspNum;
+  tmq_resp_err_t rspErr;
+  tmq_commit_cb* userCb;
+  SArray*        successfulOffsets;
+  SArray*        failedOffsets;
+  void*          userParam;
+  tsem_t         rspSem;
+} SMqCommitCbParamSet;
+
+typedef struct {
+  SMqCommitCbParamSet* params;
+  STqOffset*           pOffset;
+} SMqCommitCbParam2;
+
 tmq_conf_t* tmq_conf_new() {
   tmq_conf_t* conf = taosMemoryCalloc(1, sizeof(tmq_conf_t));
   conf->withTbName = false;
@@ -343,6 +364,139 @@ int32_t tmqCommitCb(void* param, const SDataBuf* pMsg, int32_t code) {
   return 0;
 }
 
+int32_t tmqCommitCb2(void* param, const SDataBuf* pBuf, int32_t code) {
+  SMqCommitCbParam2*   pParam = (SMqCommitCbParam2*)param;
+  SMqCommitCbParamSet* pParamSet = (SMqCommitCbParamSet*)pParam->params;
+  // push into array
+  if (code == 0) {
+    taosArrayPush(pParamSet->failedOffsets, &pParam->pOffset);
+  } else {
+    taosArrayPush(pParamSet->successfulOffsets, &pParam->pOffset);
+  }
+
+  // count down waiting rsp
+  int32_t waitingRspNum = atomic_sub_fetch_32(&pParamSet->waitingRspNum, 1);
+  ASSERT(waitingRspNum >= 0);
+
+  if (waitingRspNum == 0) {
+    // if no more waiting rsp
+    if (pParamSet->async) {
+      // call async cb func
+      if (pParamSet->automatic && pParamSet->tmq->commitCb) {
+        pParamSet->tmq->commitCb(pParamSet->tmq, pParamSet->rspErr, NULL, pParamSet->tmq->commitCbUserParam);
+      } else if (!pParamSet->automatic && pParamSet->userCb) {
+        // sem post
+        pParamSet->userCb(pParamSet->tmq, pParamSet->rspErr, NULL, pParamSet->userParam);
+      }
+    }
+
+    taosArrayDestroyP(pParamSet->successfulOffsets, taosMemoryFree);
+    taosArrayDestroyP(pParamSet->failedOffsets, taosMemoryFree);
+  }
+  return 0;
+}
+
+int32_t tmqCommitInner2(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8_t automatic, int8_t async,
+                        tmq_commit_cb* userCb, void* userParam) {
+  int32_t code = -1;
+
+  SMqCommitCbParamSet* pParamSet = taosMemoryCalloc(1, sizeof(SMqCommitCbParamSet));
+  if (pParamSet == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pParamSet->tmq = tmq;
+  pParamSet->automatic = automatic;
+  pParamSet->async = async;
+  pParamSet->freeOffsets = 1;
+  pParamSet->userCb = userCb;
+  pParamSet->userParam = userParam;
+  tsem_init(&pParamSet->rspSem, 0, 0);
+
+  for (int32_t i = 0; i < taosArrayGetSize(tmq->clientTopics); i++) {
+    SMqClientTopic* pTopic = taosArrayGet(tmq->clientTopics, i);
+    for (int32_t j = 0; j < taosArrayGetSize(pTopic->vgs); j++) {
+      SMqClientVg* pVg = taosArrayGet(pTopic->vgs, i);
+      STqOffset*   pOffset = taosMemoryCalloc(1, sizeof(STqOffset));
+      if (pOffset == NULL) {
+        terrno = TSDB_CODE_OUT_OF_MEMORY;
+        return -1;
+      }
+      int32_t tlen = strlen(tmq->groupId);
+      memcpy(pOffset->subKey, tmq->groupId, tlen);
+      pOffset->subKey[tlen] = TMQ_SEPARATOR;
+      strcpy(pOffset->subKey + tlen + 1, pTopic->topicName);
+      int32_t len;
+      int32_t code;
+      tEncodeSize(tEncodeSTqOffset, pOffset, len, code);
+      if (code < 0) {
+        ASSERT(0);
+      }
+      void* buf = taosMemoryCalloc(1, sizeof(SMsgHead) + len);
+      ((SMsgHead*)buf)->vgId = htonl(pVg->vgId);
+
+      void* abuf = POINTER_SHIFT(buf, sizeof(SMsgHead));
+
+      SEncoder encoder;
+      tEncoderInit(&encoder, abuf, len);
+      tEncodeSTqOffset(&encoder, pOffset);
+
+      // build param
+      SMqCommitCbParam2* pParam = taosMemoryCalloc(1, sizeof(SMqCommitCbParam2));
+      pParam->params = pParamSet;
+      pParam->pOffset = pOffset;
+
+      // build send info
+      SMsgSendInfo* pMsgSendInfo = taosMemoryCalloc(1, sizeof(SMsgSendInfo));
+      if (pMsgSendInfo == NULL) {
+        // TODO
+        continue;
+      }
+      pMsgSendInfo->msgInfo = (SDataBuf){
+          .pData = buf,
+          .len = len,
+          .handle = NULL,
+      };
+
+      pMsgSendInfo->requestId = generateRequestId();
+      pMsgSendInfo->requestObjRefId = 0;
+      pMsgSendInfo->param = pParam;
+      pMsgSendInfo->fp = tmqCommitCb2;
+      pMsgSendInfo->msgType = TDMT_MND_MQ_COMMIT_OFFSET;
+      // send msg
+
+      SEpSet  epSet = getEpSet_s(&tmq->pTscObj->pAppInfo->mgmtEp);
+      int64_t transporterId = 0;
+      asyncSendMsgToServer(tmq->pTscObj->pAppInfo->pTransporter, &epSet, &transporterId, pMsgSendInfo);
+      pParamSet->waitingRspNum++;
+      pParamSet->totalRspNum++;
+    }
+  }
+
+  if (!async) {
+    tsem_wait(&pParamSet->rspSem);
+    code = pParamSet->rspErr;
+    tsem_destroy(&pParamSet->rspSem);
+  } else {
+    code = 0;
+  }
+
+  if (code != 0 && async) {
+    if (automatic) {
+      tmq->commitCb(tmq, code, NULL, tmq->commitCbUserParam);
+    } else {
+      userCb(tmq, code, NULL, userParam);
+    }
+  }
+
+  if (!async) {
+    taosArrayDestroyP(pParamSet->successfulOffsets, taosMemoryFree);
+    taosArrayDestroyP(pParamSet->failedOffsets, taosMemoryFree);
+  }
+
+  return 0;
+}
+
 int32_t tmqCommitInner(tmq_t* tmq, const tmq_topic_vgroup_list_t* offsets, int8_t automatic, int8_t async,
                        tmq_commit_cb* userCb, void* userParam) {
   SMqCMCommitOffsetReq req;
@@ -890,12 +1044,13 @@ bool tmqUpdateEp(tmq_t* tmq, int32_t epoch, SMqAskEpRsp* pRsp) {
       sprintf(vgKey, "%s:%d", topic.topicName, pVgEp->vgId);
       int64_t* pOffset = taosHashGet(pHash, vgKey, strlen(vgKey));
       int64_t  offset = pVgEp->offset;
-      tscDebug("consumer %ld epoch %d vg %d offset og to %ld", tmq->consumerId, epoch, pVgEp->vgId, offset);
+      tscDebug("consumer %ld(epoch %d) original offset of vg %d is %ld", tmq->consumerId, epoch, pVgEp->vgId, offset);
       if (pOffset != NULL) {
         offset = *pOffset;
-        tscDebug("consumer %ld epoch %d vg %d found %s", tmq->consumerId, epoch, pVgEp->vgId, vgKey);
+        tscDebug("consumer %ld(epoch %d) receive offset of vg %d, full key is %s", tmq->consumerId, epoch, pVgEp->vgId,
+                 vgKey);
       }
-      tscDebug("consumer %ld epoch %d vg %d offset set to %ld", tmq->consumerId, epoch, pVgEp->vgId, offset);
+      tscDebug("consumer %ld(epoch %d) offset of vg %d updated to %ld", tmq->consumerId, epoch, pVgEp->vgId, offset);
       SMqClientVg clientVg = {
           .pollCnt = 0,
           .currentOffset = offset,
@@ -1226,9 +1381,8 @@ SMqRspObj* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
     if (rspWrapper->tmqRspType == TMQ_MSG_TYPE__POLL_RSP) {
       SMqPollRspWrapper* pollRspWrapper = (SMqPollRspWrapper*)rspWrapper;
       /*atomic_sub_fetch_32(&tmq->readyRequest, 1);*/
-      /*printf("handle poll rsp %d\n", rspMsg->head.mqMsgType);*/
+      int32_t consumerEpoch = atomic_load_32(&tmq->epoch);
-      if (pollRspWrapper->msg.head.epoch == atomic_load_32(&tmq->epoch)) {
+      if (pollRspWrapper->msg.head.epoch == consumerEpoch) {
-        /*printf("epoch match\n");*/
         SMqClientVg* pVg = pollRspWrapper->vgHandle;
         /*printf("vg %d offset %ld up to %ld\n", pVg->vgId, pVg->currentOffset, rspMsg->msg.rspOffset);*/
         pVg->currentOffset = pollRspWrapper->msg.rspOffset;
@@ -1243,7 +1397,8 @@ SMqRspObj* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
         taosFreeQitem(pollRspWrapper);
         return pRsp;
       } else {
-        /*printf("epoch mismatch\n");*/
+        tscDebug("msg discard since epoch mismatch: msg epoch %d, consumer epoch %d\n", pollRspWrapper->msg.head.epoch,
+                 consumerEpoch);
         taosFreeQitem(pollRspWrapper);
       }
     } else {
@@ -1263,10 +1418,14 @@ TAOS_RES* tmq_consumer_poll(tmq_t* tmq, int64_t timeout) {
   SMqRspObj* rspObj;
   int64_t    startTime = taosGetTimestampMs();
 
+#if 0
+  tmqHandleAllDelayedTask(tmq);
+  tmqPollImpl(tmq, timeout);
   rspObj = tmqHandleAllRsp(tmq, timeout, false);
   if (rspObj) {
     return (TAOS_RES*)rspObj;
   }
+#endif
 
   // in no topic status also need process delayed task
   if (atomic_load_8(&tmq->status) == TMQ_CONSUMER_STATUS__INIT) {
@@ -1359,8 +1518,7 @@ const char* tmq_get_table_name(TAOS_RES* res) {
         pRspObj->resIter >= pRspObj->rsp.blockNum) {
       return NULL;
     }
-    const char* name = taosArrayGetP(pRspObj->rsp.blockTbName, pRspObj->resIter);
+    return (const char*)taosArrayGetP(pRspObj->rsp.blockTbName, pRspObj->resIter);
-    return name;
   }
   return NULL;
 }

source/common/src/systable.c

@@ -91,8 +91,8 @@ static const SSysDbTableSchema userDBSchema[] = {
     {.name = "precision", .bytes = 2 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "single_stable_model", .bytes = 1, .type = TSDB_DATA_TYPE_BOOL},
     {.name = "status", .bytes = 10 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
-//    {.name = "schemaless", .bytes = 1, .type = TSDB_DATA_TYPE_BOOL},
+    //    {.name = "schemaless", .bytes = 1, .type = TSDB_DATA_TYPE_BOOL},
-    {.name = "retension", .bytes = 60 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
+    {.name = "retention", .bytes = 60 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
 
     // {.name = "update", .bytes = 1, .type = TSDB_DATA_TYPE_TINYINT},  // disable update
 };
@@ -137,7 +137,7 @@ static const SSysDbTableSchema streamSchema[] = {
     {.name = "target_table", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "watermark", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
     {.name = "trigger", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-    };
+};
 
 static const SSysDbTableSchema userTblsSchema[] = {
     {.name = "table_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
@@ -221,7 +221,9 @@ static const SSysDbTableSchema transSchema[] = {
     {.name = "db", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "failed_times", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "last_exec_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-    {.name = "last_action_info", .bytes = (TSDB_TRANS_ERROR_LEN - 1) + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
+    {.name = "last_action_info",
+     .bytes = (TSDB_TRANS_ERROR_LEN - 1) + VARSTR_HEADER_SIZE,
+     .type = TSDB_DATA_TYPE_VARCHAR},
 };
 
 static const SSysDbTableSchema configSchema[] = {
@@ -314,8 +316,6 @@ static const SSysDbTableSchema querySchema[] = {
     {.name = "sql", .bytes = TSDB_SHOW_SQL_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
 };
 
-
-
 static const SSysTableMeta perfsMeta[] = {
     {TSDB_PERFS_TABLE_CONNECTIONS, connectionsSchema, tListLen(connectionsSchema)},
     {TSDB_PERFS_TABLE_QUERIES, querySchema, tListLen(querySchema)},

source/common/src/tdatablock.c

@@ -1708,6 +1708,7 @@ char* buildCtbNameByGroupId(const char* stbName, uint64_t groupId) {
   pTag->keyLen = strlen(pTag->key);
   pTag->type = TSDB_DATA_TYPE_UBIGINT;
   pTag->u = groupId;
+  pTag->length = sizeof(uint64_t);
   taosArrayPush(tags, &pTag);
 
   void* cname = taosMemoryCalloc(1, TSDB_TABLE_NAME_LEN + 1);
@@ -1728,173 +1729,6 @@ char* buildCtbNameByGroupId(const char* stbName, uint64_t groupId) {
   return rname.childTableName;
 }
 
-SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pTSchema, bool createTb, int64_t suid,
-                            const char* stbFullName, int32_t vgId) {
-  SSubmitReq* ret = NULL;
-  SArray*     tagArray = taosArrayInit(1, sizeof(STagVal));
-  if (!tagArray) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  }
-
-  // cal size
-  int32_t cap = sizeof(SSubmitReq);
-  int32_t sz = taosArrayGetSize(pBlocks);
-  for (int32_t i = 0; i < sz; i++) {
-    SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
-    int32_t      rows = pDataBlock->info.rows;
-    // TODO min
-    int32_t rowSize = pDataBlock->info.rowSize;
-    int32_t maxLen = TD_ROW_MAX_BYTES_FROM_SCHEMA(pTSchema);
-    int32_t schemaLen = 0;
-
-    if (createTb) {
-      SVCreateTbReq createTbReq = {0};
-      char*         cname = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
-      createTbReq.name = cname;
-      createTbReq.flags = 0;
-      createTbReq.type = TSDB_CHILD_TABLE;
-      createTbReq.ctb.suid = suid;
-
-      STagVal tagVal = {.cid = 1,
-                        .type = TSDB_DATA_TYPE_UBIGINT,
-                        .pData = (uint8_t*)&pDataBlock->info.groupId,
-                        .nData = sizeof(uint64_t)};
-      STag*   pTag = NULL;
-      taosArrayClear(tagArray);
-      taosArrayPush(tagArray, &tagVal);
-      tTagNew(tagArray, 1, false, &pTag);
-      if (pTag == NULL) {
-        tdDestroySVCreateTbReq(&createTbReq);
-        taosArrayDestroy(tagArray);
-        return NULL;
-      }
-      createTbReq.ctb.pTag = (uint8_t*)pTag;
-
-      int32_t code;
-      tEncodeSize(tEncodeSVCreateTbReq, &createTbReq, schemaLen, code);
-
-      tdDestroySVCreateTbReq(&createTbReq);
-      if (code < 0) {
-        taosArrayDestroy(tagArray);
-        return NULL;
-      }
-    }
-
-    cap += sizeof(SSubmitBlk) + schemaLen + rows * maxLen;
-  }
-
-  // assign data
-  // TODO
-  ret = taosMemoryCalloc(1, cap + 46);
-  ret = POINTER_SHIFT(ret, 46);
-  ret->header.vgId = vgId;
-  ret->version = htonl(1);
-  ret->length = sizeof(SSubmitReq);
-  ret->numOfBlocks = htonl(sz);
-
-  void* submitBlk = POINTER_SHIFT(ret, sizeof(SSubmitReq));
-  for (int32_t i = 0; i < sz; i++) {
-    SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
-
-    SSubmitBlk* blkHead = submitBlk;
-    blkHead->numOfRows = htons(pDataBlock->info.rows);
-    blkHead->sversion = htonl(pTSchema->version);
-    // TODO
-    blkHead->suid = htobe64(suid);
-    // uid is assigned by vnode
-    blkHead->uid = 0;
-
-    int32_t rows = pDataBlock->info.rows;
-    /*int32_t maxLen = TD_ROW_MAX_BYTES_FROM_SCHEMA(pTSchema);*/
-    /*blkHead->dataLen = htonl(rows * maxLen);*/
-    blkHead->dataLen = 0;
-
-    void* blockData = POINTER_SHIFT(submitBlk, sizeof(SSubmitBlk));
-
-    int32_t schemaLen = 0;
-    if (createTb) {
-      SVCreateTbReq createTbReq = {0};
-      char*         cname = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
-      createTbReq.name = cname;
-      createTbReq.flags = 0;
-      createTbReq.type = TSDB_CHILD_TABLE;
-      createTbReq.ctb.suid = suid;
-
-      STagVal tagVal = {.cid = 1,
-                        .type = TSDB_DATA_TYPE_UBIGINT,
-                        .pData = (uint8_t*)&pDataBlock->info.groupId,
-                        .nData = sizeof(uint64_t)};
-      taosArrayClear(tagArray);
-      taosArrayPush(tagArray, &tagVal);
-      STag* pTag = NULL;
-      tTagNew(tagArray, 1, false, &pTag);
-      if (pTag == NULL) {
-        tdDestroySVCreateTbReq(&createTbReq);
-        taosArrayDestroy(tagArray);
-        taosMemoryFreeClear(ret);
-        return NULL;
-      }
-      createTbReq.ctb.pTag = (uint8_t*)pTag;
-
-      int32_t code;
-      tEncodeSize(tEncodeSVCreateTbReq, &createTbReq, schemaLen, code);
-      if (code < 0) {
-        tdDestroySVCreateTbReq(&createTbReq);
-        taosArrayDestroy(tagArray);
-        taosMemoryFreeClear(ret);
-        return NULL;
-      }
-
-      SEncoder encoder = {0};
-      tEncoderInit(&encoder, blockData, schemaLen);
-      code = tEncodeSVCreateTbReq(&encoder, &createTbReq);
-      tEncoderClear(&encoder);
-      tdDestroySVCreateTbReq(&createTbReq);
-
-      if (code < 0) {
-        taosArrayDestroy(tagArray);
-        taosMemoryFreeClear(ret);
-        return NULL;
-      }
-    }
-    blkHead->schemaLen = htonl(schemaLen);
-
-    STSRow* rowData = POINTER_SHIFT(blockData, schemaLen);
-
-    for (int32_t j = 0; j < rows; j++) {
-      SRowBuilder rb = {0};
-      tdSRowInit(&rb, pTSchema->version);
-      tdSRowSetTpInfo(&rb, pTSchema->numOfCols, pTSchema->flen);
-      tdSRowResetBuf(&rb, rowData);
-
-      for (int32_t k = 0; k < pTSchema->numOfCols; k++) {
-        const STColumn*  pColumn = &pTSchema->columns[k];
-        SColumnInfoData* pColData = taosArrayGet(pDataBlock->pDataBlock, k);
-        if (colDataIsNull_s(pColData, j)) {
-          tdAppendColValToRow(&rb, pColumn->colId, pColumn->type, TD_VTYPE_NONE, NULL, false, pColumn->offset, k);
-        } else {
-          void* data = colDataGetData(pColData, j);
-          tdAppendColValToRow(&rb, pColumn->colId, pColumn->type, TD_VTYPE_NORM, data, true, pColumn->offset, k);
-        }
-      }
-      int32_t rowLen = TD_ROW_LEN(rowData);
-      rowData = POINTER_SHIFT(rowData, rowLen);
-      blkHead->dataLen += rowLen;
-    }
-    int32_t dataLen = blkHead->dataLen;
-    blkHead->dataLen = htonl(dataLen);
-
-    ret->length += sizeof(SSubmitBlk) + schemaLen + dataLen;
-    blkHead = POINTER_SHIFT(blkHead, schemaLen + dataLen);
-    /*submitBlk = blkHead;*/
-  }
-
-  ret->length = htonl(ret->length);
-  taosArrayDestroy(tagArray);
-  return ret;
-}
-
 void blockCompressEncode(const SSDataBlock* pBlock, char* data, int32_t* dataLen, int32_t numOfCols,
                          int8_t needCompress) {
   // todo extract method

source/common/src/tmsg.c

@@ -2438,6 +2438,10 @@ int32_t tSerializeSTableIndexRsp(void *buf, int32_t bufLen, const STableIndexRsp
   tEncoderInit(&encoder, buf, bufLen);
 
   if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->tbName) < 0) return -1;
+  if (tEncodeCStr(&encoder, pRsp->dbFName) < 0) return -1;
+  if (tEncodeU64(&encoder, pRsp->suid) < 0) return -1;
+  if (tEncodeI32(&encoder, pRsp->version) < 0) return -1;
   int32_t num = taosArrayGetSize(pRsp->pIndex);
   if (tEncodeI32(&encoder, num) < 0) return -1;
   if (num > 0) {
@@ -2472,6 +2476,10 @@ int32_t tDeserializeSTableIndexRsp(void *buf, int32_t bufLen, STableIndexRsp *pR
   tDecoderInit(&decoder, buf, bufLen);
 
   if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->tbName) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pRsp->dbFName) < 0) return -1;
+  if (tDecodeU64(&decoder, &pRsp->suid) < 0) return -1;
+  if (tDecodeI32(&decoder, &pRsp->version) < 0) return -1;
   int32_t num = 0;
   if (tDecodeI32(&decoder, &num) < 0) return -1;
   if (num > 0) {
@@ -2631,18 +2639,35 @@ int32_t tSerializeSTableMetaRsp(void *buf, int32_t bufLen, STableMetaRsp *pRsp)
   return tlen;
 }
 
-int32_t tSerializeSTableMetaBatchRsp(void *buf, int32_t bufLen, STableMetaBatchRsp *pRsp) {
+int32_t tSerializeSSTbHbRsp(void *buf, int32_t bufLen, SSTbHbRsp *pRsp) {
   SEncoder encoder = {0};
   tEncoderInit(&encoder, buf, bufLen);
 
   if (tStartEncode(&encoder) < 0) return -1;
 
-  int32_t numOfBatch = taosArrayGetSize(pRsp->pArray);
+  int32_t numOfMeta = taosArrayGetSize(pRsp->pMetaRsp);
-  if (tEncodeI32(&encoder, numOfBatch) < 0) return -1;
+  if (tEncodeI32(&encoder, numOfMeta) < 0) return -1;
-  for (int32_t i = 0; i < numOfBatch; ++i) {
+  for (int32_t i = 0; i < numOfMeta; ++i) {
-    STableMetaRsp *pMetaRsp = taosArrayGet(pRsp->pArray, i);
+    STableMetaRsp *pMetaRsp = taosArrayGet(pRsp->pMetaRsp, i);
     if (tEncodeSTableMetaRsp(&encoder, pMetaRsp) < 0) return -1;
   }
+
+  int32_t numOfIndex = taosArrayGetSize(pRsp->pIndexRsp);
+  if (tEncodeI32(&encoder, numOfIndex) < 0) return -1;
+  for (int32_t i = 0; i < numOfIndex; ++i) {
+    STableIndexRsp *pIndexRsp = taosArrayGet(pRsp->pIndexRsp, i);
+    if (tEncodeCStr(&encoder, pIndexRsp->tbName) < 0) return -1;
+    if (tEncodeCStr(&encoder, pIndexRsp->dbFName) < 0) return -1;
+    if (tEncodeU64(&encoder, pIndexRsp->suid) < 0) return -1;
+    if (tEncodeI32(&encoder, pIndexRsp->version) < 0) return -1;
+    int32_t num = taosArrayGetSize(pIndexRsp->pIndex);
+    if (tEncodeI32(&encoder, num) < 0) return -1;
+    for (int32_t i = 0; i < num; ++i) {
+      STableIndexInfo *pInfo = (STableIndexInfo *)taosArrayGet(pIndexRsp->pIndex, i);
+      if (tSerializeSTableIndexInfo(&encoder, pInfo) < 0) return -1;
+    }
+  }
+
   tEndEncode(&encoder);
 
   int32_t tlen = encoder.pos;
@@ -2662,26 +2687,58 @@ int32_t tDeserializeSTableMetaRsp(void *buf, int32_t bufLen, STableMetaRsp *pRsp
   return 0;
 }
 
-int32_t tDeserializeSTableMetaBatchRsp(void *buf, int32_t bufLen, STableMetaBatchRsp *pRsp) {
+int32_t tDeserializeSSTbHbRsp(void *buf, int32_t bufLen, SSTbHbRsp *pRsp) {
   SDecoder decoder = {0};
   tDecoderInit(&decoder, buf, bufLen);
 
   if (tStartDecode(&decoder) < 0) return -1;
 
-  int32_t numOfBatch = taosArrayGetSize(pRsp->pArray);
+  int32_t numOfMeta = 0;
-  if (tDecodeI32(&decoder, &numOfBatch) < 0) return -1;
+  if (tDecodeI32(&decoder, &numOfMeta) < 0) return -1;
-
+  pRsp->pMetaRsp = taosArrayInit(numOfMeta, sizeof(STableMetaRsp));
-  pRsp->pArray = taosArrayInit(numOfBatch, sizeof(STableMetaRsp));
+  if (pRsp->pMetaRsp == NULL) {
-  if (pRsp->pArray == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
-  for (int32_t i = 0; i < numOfBatch; ++i) {
+  for (int32_t i = 0; i < numOfMeta; ++i) {
     STableMetaRsp tableMetaRsp = {0};
     if (tDecodeSTableMetaRsp(&decoder, &tableMetaRsp) < 0) return -1;
-    taosArrayPush(pRsp->pArray, &tableMetaRsp);
+    taosArrayPush(pRsp->pMetaRsp, &tableMetaRsp);
   }
+
+  int32_t numOfIndex = 0;
+  if (tDecodeI32(&decoder, &numOfIndex) < 0) return -1;
+
+  pRsp->pIndexRsp = taosArrayInit(numOfIndex, sizeof(STableIndexRsp));
+  if (pRsp->pIndexRsp == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  for (int32_t i = 0; i < numOfIndex; ++i) {
+    STableIndexRsp tableIndexRsp = {0};
+    if (tDecodeCStrTo(&decoder, tableIndexRsp.tbName) < 0) return -1;
+    if (tDecodeCStrTo(&decoder, tableIndexRsp.dbFName) < 0) return -1;
+    if (tDecodeU64(&decoder, &tableIndexRsp.suid) < 0) return -1;
+    if (tDecodeI32(&decoder, &tableIndexRsp.version) < 0) return -1;
+    int32_t num = 0;
+    if (tDecodeI32(&decoder, &num) < 0) return -1;
+    if (num > 0) {
+      tableIndexRsp.pIndex = taosArrayInit(num, sizeof(STableIndexInfo));
+      if (NULL == tableIndexRsp.pIndex) return -1;
+      STableIndexInfo info;
+      for (int32_t i = 0; i < num; ++i) {
+        if (tDeserializeSTableIndexInfo(&decoder, &info) < 0) return -1;
+        if (NULL == taosArrayPush(tableIndexRsp.pIndex, &info)) {
+          taosMemoryFree(info.expr);
+          return -1;
+        }
+      }
+    }
+    taosArrayPush(pRsp->pIndexRsp, &tableIndexRsp);
+  }
+
   tEndDecode(&decoder);
 
   tDecoderClear(&decoder);
@@ -2690,14 +2747,32 @@ int32_t tDeserializeSTableMetaBatchRsp(void *buf, int32_t bufLen, STableMetaBatc
 
 void tFreeSTableMetaRsp(STableMetaRsp *pRsp) { taosMemoryFreeClear(pRsp->pSchemas); }
 
-void tFreeSTableMetaBatchRsp(STableMetaBatchRsp *pRsp) {
+void tFreeSTableIndexRsp(void *info) {
-  int32_t numOfBatch = taosArrayGetSize(pRsp->pArray);
+  if (NULL == info) {
-  for (int32_t i = 0; i < numOfBatch; ++i) {
+    return;
-    STableMetaRsp *pMetaRsp = taosArrayGet(pRsp->pArray, i);
+  }
+
+  STableIndexRsp *pInfo = (STableIndexRsp *)info;
+
+  taosArrayDestroyEx(pInfo->pIndex, tFreeSTableIndexInfo);
+}
+
+void tFreeSSTbHbRsp(SSTbHbRsp *pRsp) {
+  int32_t numOfMeta = taosArrayGetSize(pRsp->pMetaRsp);
+  for (int32_t i = 0; i < numOfMeta; ++i) {
+    STableMetaRsp *pMetaRsp = taosArrayGet(pRsp->pMetaRsp, i);
     tFreeSTableMetaRsp(pMetaRsp);
   }
 
-  taosArrayDestroy(pRsp->pArray);
+  taosArrayDestroy(pRsp->pMetaRsp);
+
+  int32_t numOfIndex = taosArrayGetSize(pRsp->pIndexRsp);
+  for (int32_t i = 0; i < numOfIndex; ++i) {
+    STableIndexRsp *pIndexRsp = taosArrayGet(pRsp->pIndexRsp, i);
+    tFreeSTableIndexRsp(pIndexRsp);
+  }
+
+  taosArrayDestroy(pRsp->pIndexRsp);
 }
 
 int32_t tSerializeSShowRsp(void *buf, int32_t bufLen, SShowRsp *pRsp) {
@@ -3925,7 +4000,7 @@ int32_t tEncodeTSma(SEncoder *pCoder, const STSma *pSma) {
   return 0;
 }
 
-int32_t tDecodeTSma(SDecoder *pCoder, STSma *pSma) {
+int32_t tDecodeTSma(SDecoder *pCoder, STSma *pSma, bool deepCopy) {
   if (tDecodeI8(pCoder, &pSma->version) < 0) return -1;
   if (tDecodeI8(pCoder, &pSma->intervalUnit) < 0) return -1;
   if (tDecodeI8(pCoder, &pSma->slidingUnit) < 0) return -1;
@@ -3937,17 +4012,30 @@ int32_t tDecodeTSma(SDecoder *pCoder, STSma *pSma) {
   if (tDecodeI64(pCoder, &pSma->indexUid) < 0) return -1;
   if (tDecodeI64(pCoder, &pSma->tableUid) < 0) return -1;
   if (tDecodeI64(pCoder, &pSma->dstTbUid) < 0) return -1;
-  if (tDecodeCStr(pCoder, &pSma->dstTbName) < 0) return -1;
+  if (deepCopy) {
+    if (tDecodeCStrAlloc(pCoder, &pSma->dstTbName) < 0) return -1;
+  } else {
+    if (tDecodeCStr(pCoder, &pSma->dstTbName) < 0) return -1;
+  }
+
   if (tDecodeI64(pCoder, &pSma->interval) < 0) return -1;
   if (tDecodeI64(pCoder, &pSma->offset) < 0) return -1;
   if (tDecodeI64(pCoder, &pSma->sliding) < 0) return -1;
   if (pSma->exprLen > 0) {
-    if (tDecodeCStr(pCoder, &pSma->expr) < 0) return -1;
+    if (deepCopy) {
+      if (tDecodeCStrAlloc(pCoder, &pSma->expr) < 0) return -1;
+    } else {
+      if (tDecodeCStr(pCoder, &pSma->expr) < 0) return -1;
+    }
   } else {
     pSma->expr = NULL;
   }
   if (pSma->tagsFilterLen > 0) {
-    if (tDecodeCStr(pCoder, &pSma->tagsFilter) < 0) return -1;
+    if (deepCopy) {
+      if (tDecodeCStrAlloc(pCoder, &pSma->tagsFilter) < 0) return -1;
+    } else {
+      if (tDecodeCStr(pCoder, &pSma->tagsFilter) < 0) return -1;
+    }
   } else {
     pSma->tagsFilter = NULL;
   }
@@ -3970,7 +4058,7 @@ int32_t tEncodeSVCreateTSmaReq(SEncoder *pCoder, const SVCreateTSmaReq *pReq) {
 int32_t tDecodeSVCreateTSmaReq(SDecoder *pCoder, SVCreateTSmaReq *pReq) {
   if (tStartDecode(pCoder) < 0) return -1;
 
-  tDecodeTSma(pCoder, pReq);
+  tDecodeTSma(pCoder, pReq, false);
 
   tEndDecode(pCoder);
   return 0;
@@ -4776,3 +4864,31 @@ void tFreeSMAlterStbRsp(SMAlterStbRsp *pRsp) {
     taosMemoryFree(pRsp->pMeta);
   }
 }
+
+int32_t tEncodeSTqOffset(SEncoder *pEncoder, const STqOffset *pOffset) {
+  if (tEncodeI8(pEncoder, pOffset->type) < 0) return -1;
+  if (pOffset->type == TMQ_OFFSET__SNAPSHOT) {
+    if (tEncodeI64(pEncoder, pOffset->uid) < 0) return -1;
+    if (tEncodeI64(pEncoder, pOffset->ts) < 0) return -1;
+  } else if (pOffset->type == TMQ_OFFSET__LOG) {
+    if (tEncodeI64(pEncoder, pOffset->version) < 0) return -1;
+  } else {
+    ASSERT(0);
+  }
+  if (tEncodeCStr(pEncoder, pOffset->subKey) < 0) return -1;
+  return 0;
+}
+
+int32_t tDecodeSTqOffset(SDecoder *pDecoder, STqOffset *pOffset) {
+  if (tDecodeI8(pDecoder, &pOffset->type) < 0) return -1;
+  if (pOffset->type == TMQ_OFFSET__SNAPSHOT) {
+    if (tDecodeI64(pDecoder, &pOffset->uid) < 0) return -1;
+    if (tDecodeI64(pDecoder, &pOffset->ts) < 0) return -1;
+  } else if (pOffset->type == TMQ_OFFSET__LOG) {
+    if (tDecodeI64(pDecoder, &pOffset->version) < 0) return -1;
+  } else {
+    ASSERT(0);
+  }
+  if (tDecodeCStrTo(pDecoder, pOffset->subKey) < 0) return -1;
+  return 0;
+}

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

@@ -155,7 +155,6 @@ SArray *mmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_MND_CREATE_MNODE, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_ALTER_MNODE, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_ALTER_MNODE_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
-  if (dmSetMgmtHandle(pArray, TDMT_MND_SET_STANDBY_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_DROP_MNODE, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_CREATE_QNODE, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_DROP_QNODE, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
@@ -236,7 +235,10 @@ SArray *mmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_APPEND_ENTRIES_REPLY, mmPutMsgToSyncQueue, 1) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_SNAPSHOT_SEND, mmPutMsgToSyncQueue, 1) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_SNAPSHOT_RSP, mmPutMsgToSyncQueue, 1) == NULL) goto _OVER;
-  if (dmSetMgmtHandle(pArray, TDMT_MND_SET_STANDBY, mmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_SYNC_SET_MNODE_STANDBY, mmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
+
+  if (dmSetMgmtHandle(pArray, TDMT_SYNC_SET_MNODE_STANDBY_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_SYNC_SET_VNODE_STANDBY_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
 
   code = 0;
 

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

@@ -114,7 +114,8 @@ int32_t mmPutMsgToReadQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) {
 }
 
 int32_t mmPutMsgToQueryQueue(SMnodeMgmt *pMgmt, SRpcMsg *pMsg) {
-  if (mndPreprocessQueryMsg(pMgmt->pMnode, pMsg) != 0) {
+  pMsg->info.node = pMgmt->pMnode;
+  if (mndPreProcessMsg(pMsg) != 0) {
     dError("msg:%p, failed to pre-process in mnode since %s, type:%s", pMsg, terrstr(), TMSG_INFO(pMsg->msgType));
     return -1;
   }

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

@@ -344,6 +344,7 @@ SArray *vmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_VND_SUBMIT_RSMA, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_VND_MQ_VG_CHANGE, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_VND_MQ_VG_DELETE, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_VND_MQ_COMMIT_OFFSET, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_VND_CONSUME, vmPutMsgToFetchQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_VND_DELETE, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_VND_QUERY_HEARTBEAT, vmPutMsgToFetchQueue, 0) == NULL) goto _OVER;
@@ -373,6 +374,7 @@ SArray *vmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_REQUEST_VOTE_REPLY, vmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_APPEND_ENTRIES, vmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_APPEND_ENTRIES_REPLY, vmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_SYNC_SET_VNODE_STANDBY, vmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
 
   code = 0;
 

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

@@ -124,7 +124,7 @@ static int32_t dmProcessCreateNodeReq(EDndNodeType ntype, SRpcMsg *pMsg) {
   } else {
     dInfo("node:%s, has been created", pWrapper->name);
     code = dmOpenNode(pWrapper);
-    if (code != 0) {
+    if (code == 0) {
       code = dmStartNode(pWrapper);
     }
     pWrapper->deployed = true;

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

@@ -124,8 +124,6 @@ void    mndReleaseRpcRef(SMnode *pMnode);
 void    mndSetRestore(SMnode *pMnode, bool restored);
 void    mndSetStop(SMnode *pMnode);
 bool    mndGetStop(SMnode *pMnode);
-int32_t mndAcquireSyncRef(SMnode *pMnode);
-void    mndReleaseSyncRef(SMnode *pMnode);
 
 #ifdef __cplusplus
 }

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

@@ -26,6 +26,9 @@ int32_t  mndInitSma(SMnode *pMnode);
 void     mndCleanupSma(SMnode *pMnode);
 SSmaObj *mndAcquireSma(SMnode *pMnode, char *smaName);
 void     mndReleaseSma(SMnode *pMnode, SSmaObj *pSma);
+int32_t  mndDropSmasByStb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb);
+int32_t  mndDropSmasByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb);
+int32_t  mndGetTableSma(SMnode *pMnode, char *tbFName, STableIndexRsp *rsp, bool *exist);
 
 #ifdef __cplusplus
 }

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

@@ -27,7 +27,7 @@ void     mndCleanupStb(SMnode *pMnode);
 SStbObj *mndAcquireStb(SMnode *pMnode, char *stbName);
 void     mndReleaseStb(SMnode *pMnode, SStbObj *pStb);
 SSdbRaw *mndStbActionEncode(SStbObj *pStb);
-int32_t  mndValidateStbInfo(SMnode *pMnode, SSTableMetaVersion *pStbs, int32_t numOfStbs, void **ppRsp,
+int32_t  mndValidateStbInfo(SMnode *pMnode, SSTableVersion *pStbs, int32_t numOfStbs, void **ppRsp,
                             int32_t *pRspLen);
 int32_t  mndGetNumOfStbs(SMnode *pMnode, char *dbName, int32_t *pNumOfStbs);
 

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

@@ -183,12 +183,12 @@ static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
     pDb->cfg.pRetensions = taosArrayInit(pDb->cfg.numOfRetensions, sizeof(SRetention));
     if (pDb->cfg.pRetensions == NULL) goto _OVER;
     for (int32_t i = 0; i < pDb->cfg.numOfRetensions; ++i) {
-      SRetention retension = {0};
+      SRetention retention = {0};
-      SDB_GET_INT64(pRaw, dataPos, &retension.freq, _OVER)
+      SDB_GET_INT64(pRaw, dataPos, &retention.freq, _OVER)
-      SDB_GET_INT64(pRaw, dataPos, &retension.keep, _OVER)
+      SDB_GET_INT64(pRaw, dataPos, &retention.keep, _OVER)
-      SDB_GET_INT8(pRaw, dataPos, &retension.freqUnit, _OVER)
+      SDB_GET_INT8(pRaw, dataPos, &retention.freqUnit, _OVER)
-      SDB_GET_INT8(pRaw, dataPos, &retension.keepUnit, _OVER)
+      SDB_GET_INT8(pRaw, dataPos, &retention.keepUnit, _OVER)
-      if (taosArrayPush(pDb->cfg.pRetensions, &retension) == NULL) {
+      if (taosArrayPush(pDb->cfg.pRetensions, &retention) == NULL) {
         goto _OVER;
       }
     }
@@ -472,7 +472,7 @@ static int32_t mndCreateDb(SMnode *pMnode, SRpcMsg *pReq, SCreateDbReq *pCreate,
   }
 
   int32_t code = -1;
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_CONFLICT_DB, pReq);
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB, pReq);
   if (pTrans == NULL) goto _OVER;
 
   mDebug("trans:%d, used to create db:%s", pTrans->id, pCreate->db);
@@ -935,6 +935,7 @@ static int32_t mndDropDb(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb) {
   if (mndDropOffsetByDB(pMnode, pTrans, pDb) != 0) goto _OVER;
   if (mndDropSubByDB(pMnode, pTrans, pDb) != 0) goto _OVER;
   if (mndDropTopicByDB(pMnode, pTrans, pDb) != 0) goto _OVER;
+  if (mndDropSmasByDb(pMnode, pTrans, pDb) != 0) goto _OVER;
   if (mndSetDropDbRedoActions(pMnode, pTrans, pDb) != 0) goto _OVER;
 
   SUserObj *pUser = mndAcquireUser(pMnode, pDb->createUser);
@@ -1366,7 +1367,7 @@ char *buildRetension(SArray *pRetension) {
 }
 
 static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, int32_t rows, int64_t numOfTables,
-                           bool sysDb) {
+                           bool sysDb, ESdbStatus objStatus) {
   int32_t cols = 0;
 
   int32_t     bytes = pShow->pMeta->pSchemas[cols].bytes;
@@ -1379,7 +1380,9 @@ static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, in
   }
 
   char *status = "ready";
-  char  statusB[24] = {0};
+  if (objStatus == SDB_STATUS_CREATING) status = "creating";
+  if (objStatus == SDB_STATUS_DROPPING) status = "dropping";
+  char statusB[24] = {0};
   STR_WITH_SIZE_TO_VARSTR(statusB, status, strlen(status));
 
   if (sysDb) {
@@ -1503,8 +1506,8 @@ static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, in
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, rows, (const char *)statusB, false);
 
-//    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    //    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-//    colDataAppend(pColInfo, rows, (const char *)&pDb->cfg.schemaless, false);
+    //    colDataAppend(pColInfo, rows, (const char *)&pDb->cfg.schemaless, false);
 
     char *p = buildRetension(pDb->cfg.pRetensions);
 
@@ -1548,29 +1551,30 @@ static bool mndGetTablesOfDbFp(SMnode *pMnode, void *pObj, void *p1, void *p2, v
 }
 
 static int32_t mndRetrieveDbs(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rowsCapacity) {
-  SMnode *pMnode = pReq->info.node;
+  SMnode    *pMnode = pReq->info.node;
-  SSdb   *pSdb = pMnode->pSdb;
+  SSdb      *pSdb = pMnode->pSdb;
-  int32_t numOfRows = 0;
+  int32_t    numOfRows = 0;
-  SDbObj *pDb = NULL;
+  SDbObj    *pDb = NULL;
+  ESdbStatus objStatus = 0;
 
   // Append the information_schema database into the result.
   if (!pShow->sysDbRsp) {
     SDbObj infoschemaDb = {0};
     setInformationSchemaDbCfg(&infoschemaDb);
-    dumpDbInfoData(pBlock, &infoschemaDb, pShow, numOfRows, 14, true);
+    dumpDbInfoData(pBlock, &infoschemaDb, pShow, numOfRows, 14, true, 0);
 
     numOfRows += 1;
 
     SDbObj perfschemaDb = {0};
     setPerfSchemaDbCfg(&perfschemaDb);
-    dumpDbInfoData(pBlock, &perfschemaDb, pShow, numOfRows, 3, true);
+    dumpDbInfoData(pBlock, &perfschemaDb, pShow, numOfRows, 3, true, 0);
 
     numOfRows += 1;
     pShow->sysDbRsp = true;
   }
 
   while (numOfRows < rowsCapacity) {
-    pShow->pIter = sdbFetch(pSdb, SDB_DB, pShow->pIter, (void **)&pDb);
+    pShow->pIter = sdbFetchAll(pSdb, SDB_DB, pShow->pIter, (void **)&pDb, &objStatus);
     if (pShow->pIter == NULL) {
       break;
     }
@@ -1578,7 +1582,7 @@ static int32_t mndRetrieveDbs(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBloc
     int32_t numOfTables = 0;
     sdbTraverse(pSdb, SDB_VGROUP, mndGetTablesOfDbFp, &numOfTables, NULL, NULL);
 
-    dumpDbInfoData(pBlock, pDb, pShow, numOfRows, numOfTables, false);
+    dumpDbInfoData(pBlock, pDb, pShow, numOfRows, numOfTables, false, objStatus);
     numOfRows++;
     sdbRelease(pSdb, pDb);
   }

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

@@ -380,22 +380,17 @@ void mndStop(SMnode *pMnode) {
 int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
   SMnode    *pMnode = pMsg->info.node;
   SSyncMgmt *pMgmt = &pMnode->syncMgmt;
-  int32_t    code = TAOS_SYNC_PROPOSE_OTHER_ERROR;
+  int32_t    code = TAOS_SYNC_OTHER_ERROR;
 
   if (!syncEnvIsStart()) {
     mError("failed to process sync msg:%p type:%s since syncEnv stop", pMsg, TMSG_INFO(pMsg->msgType));
-    return TAOS_SYNC_PROPOSE_OTHER_ERROR;
+    return TAOS_SYNC_OTHER_ERROR;
   }
 
   SSyncNode *pSyncNode = syncNodeAcquire(pMgmt->sync);
   if (pSyncNode == NULL) {
     mError("failed to process sync msg:%p type:%s since syncNode is null", pMsg, TMSG_INFO(pMsg->msgType));
-    return TAOS_SYNC_PROPOSE_OTHER_ERROR;
+    return TAOS_SYNC_OTHER_ERROR;
-  }
-
-  if (mndAcquireSyncRef(pMnode) != 0) {
-    mError("failed to process sync msg:%p type:%s since %s", pMsg, TMSG_INFO(pMsg->msgType), terrstr());
-    return TAOS_SYNC_PROPOSE_OTHER_ERROR;
   }
 
   char  logBuf[512] = {0};
@@ -450,13 +445,13 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
       SyncSnapshotRsp *pSyncMsg = syncSnapshotRspFromRpcMsg2(pMsg);
       code = syncNodeOnSnapshotRspCb(pSyncNode, pSyncMsg);
       syncSnapshotRspDestroy(pSyncMsg);
-    } else if (pMsg->msgType == TDMT_MND_SET_STANDBY) {
+    } else if (pMsg->msgType == TDMT_SYNC_SET_MNODE_STANDBY) {
       code = syncSetStandby(pMgmt->sync);
       SRpcMsg rsp = {.code = code, .info = pMsg->info};
       tmsgSendRsp(&rsp);
     } else {
       mError("failed to process msg:%p since invalid type:%s", pMsg, TMSG_INFO(pMsg->msgType));
-      code = TAOS_SYNC_PROPOSE_OTHER_ERROR;
+      code = TAOS_SYNC_OTHER_ERROR;
     }
   } else {
     if (pMsg->msgType == TDMT_SYNC_TIMEOUT) {
@@ -491,17 +486,16 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
       SyncAppendEntriesReply *pSyncMsg = syncAppendEntriesReplyFromRpcMsg2(pMsg);
       code = syncNodeOnAppendEntriesReplyCb(pSyncNode, pSyncMsg);
       syncAppendEntriesReplyDestroy(pSyncMsg);
-    } else if (pMsg->msgType == TDMT_MND_SET_STANDBY) {
+    } else if (pMsg->msgType == TDMT_SYNC_SET_MNODE_STANDBY) {
       code = syncSetStandby(pMgmt->sync);
       SRpcMsg rsp = {.code = code, .info = pMsg->info};
       tmsgSendRsp(&rsp);
     } else {
       mError("failed to process msg:%p since invalid type:%s", pMsg, TMSG_INFO(pMsg->msgType));
-      code = TAOS_SYNC_PROPOSE_OTHER_ERROR;
+      code = TAOS_SYNC_OTHER_ERROR;
     }
   }
 
-  mndReleaseSyncRef(pMnode);
   return code;
 }
 
@@ -754,24 +748,3 @@ void mndSetStop(SMnode *pMnode) {
 }
 
 bool mndGetStop(SMnode *pMnode) { return pMnode->stopped; }
-
-int32_t mndAcquireSyncRef(SMnode *pMnode) {
-  int32_t code = 0;
-  taosThreadRwlockRdlock(&pMnode->lock);
-  if (pMnode->stopped) {
-    terrno = TSDB_CODE_APP_NOT_READY;
-    code = -1;
-  } else {
-    int32_t ref = atomic_add_fetch_32(&pMnode->syncRef, 1);
-    // mTrace("mnode sync is acquired, ref:%d", ref);
-  }
-  taosThreadRwlockUnlock(&pMnode->lock);
-  return code;
-}
-
-void mndReleaseSyncRef(SMnode *pMnode) {
-  taosThreadRwlockRdlock(&pMnode->lock);
-  int32_t ref = atomic_sub_fetch_32(&pMnode->syncRef, 1);
-  // mTrace("mnode sync is released, ref:%d", ref);
-  taosThreadRwlockUnlock(&pMnode->lock);
-}

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

@@ -55,7 +55,8 @@ int32_t mndInitMnode(SMnode *pMnode) {
   mndSetMsgHandle(pMnode, TDMT_MND_ALTER_MNODE_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_MND_DROP_MNODE, mndProcessDropMnodeReq);
   mndSetMsgHandle(pMnode, TDMT_DND_DROP_MNODE_RSP, mndTransProcessRsp);
-  mndSetMsgHandle(pMnode, TDMT_MND_SET_STANDBY_RSP, mndTransProcessRsp);
+  mndSetMsgHandle(pMnode, TDMT_SYNC_SET_MNODE_STANDBY_RSP, mndTransProcessRsp);
+  mndSetMsgHandle(pMnode, TDMT_SYNC_SET_VNODE_STANDBY_RSP, mndTransProcessRsp);
 
   mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_MNODE, mndRetrieveMnodes);
   mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_MNODE, mndCancelGetNextMnode);
@@ -511,7 +512,7 @@ static int32_t mndSetDropMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDnode
         .epSet = dropEpSet,
         .pCont = pReq,
         .contLen = contLen,
-        .msgType = TDMT_MND_SET_STANDBY,
+        .msgType = TDMT_SYNC_SET_MNODE_STANDBY,
         .acceptableCode = TSDB_CODE_NODE_NOT_DEPLOYED,
     };
 
@@ -659,7 +660,7 @@ static int32_t mndRetrieveMnodes(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pB
   int32_t    numOfRows = 0;
   int32_t    cols = 0;
   SMnodeObj *pObj = NULL;
-  ESdbStatus objStatus;
+  ESdbStatus objStatus = 0;
   char      *pWrite;
   int64_t    curMs = taosGetTimestampMs();
 

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

@@ -433,7 +433,7 @@ static int32_t mndProcessQueryHeartBeat(SMnode *pMnode, SRpcMsg *pMsg, SClientHb
       case HEARTBEAT_KEY_STBINFO: {
         void   *rspMsg = NULL;
         int32_t rspLen = 0;
-        mndValidateStbInfo(pMnode, kv->value, kv->valueLen / sizeof(SSTableMetaVersion), &rspMsg, &rspLen);
+        mndValidateStbInfo(pMnode, kv->value, kv->valueLen / sizeof(SSTableVersion), &rspMsg, &rspLen);
         if (rspMsg && rspLen > 0) {
           SKv kv1 = {.key = HEARTBEAT_KEY_STBINFO, .valueLen = rspLen, .value = rspMsg};
           taosArrayPush(hbRsp.info, &kv1);

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

@@ -18,11 +18,10 @@
 #include "mndMnode.h"
 #include "qworker.h"
 
-int32_t mndPreprocessQueryMsg(SMnode * pMnode, SRpcMsg * pMsg) {
+int32_t mndPreProcessMsg(SRpcMsg *pMsg) {
-  if (TDMT_VND_QUERY != pMsg->msgType) {
+  if (TDMT_VND_QUERY != pMsg->msgType) return 0;
-    return 0;
-  }
 
+  SMnode *pMnode = pMsg->info.node;
   return qWorkerPreprocessQueryMsg(pMnode->pQuery, pMsg);
 }
 

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

@@ -38,12 +38,13 @@ static int32_t  mndSmaActionInsert(SSdb *pSdb, SSmaObj *pSma);
 static int32_t  mndSmaActionDelete(SSdb *pSdb, SSmaObj *pSpSmatb);
 static int32_t  mndSmaActionUpdate(SSdb *pSdb, SSmaObj *pOld, SSmaObj *pNew);
 static int32_t  mndSmaGetVgEpSet(SMnode *pMnode, SDbObj *pDb, SVgEpSet **ppVgEpSet, int32_t *numOfVgroups);
-static int32_t  mndProcessMCreateSmaReq(SRpcMsg *pReq);
+static int32_t  mndProcessCreateSmaReq(SRpcMsg *pReq);
-static int32_t  mndProcessMDropSmaReq(SRpcMsg *pReq);
+static int32_t  mndProcessDropSmaReq(SRpcMsg *pReq);
 static int32_t  mndProcessGetSmaReq(SRpcMsg *pReq);
-static int32_t mndProcessGetTbSmaReq(SRpcMsg *pReq);
+static int32_t  mndProcessGetTbSmaReq(SRpcMsg *pReq);
 static int32_t  mndRetrieveSma(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
 static void     mndCancelGetNextSma(SMnode *pMnode, void *pIter);
+static void     mndDestroySmaObj(SSmaObj *pSmaObj);
 
 int32_t mndInitSma(SMnode *pMnode) {
   SSdbTable table = {
@@ -56,8 +57,8 @@ int32_t mndInitSma(SMnode *pMnode) {
       .deleteFp = (SdbDeleteFp)mndSmaActionDelete,
   };
 
-  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_SMA, mndProcessMCreateSmaReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_SMA, mndProcessCreateSmaReq);
-  mndSetMsgHandle(pMnode, TDMT_MND_DROP_SMA, mndProcessMDropSmaReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_SMA, mndProcessDropSmaReq);
   mndSetMsgHandle(pMnode, TDMT_VND_CREATE_SMA_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_VND_DROP_SMA_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_MND_GET_INDEX, mndProcessGetSmaReq);
@@ -79,7 +80,6 @@ static SSdbRaw *mndSmaActionEncode(SSmaObj *pSma) {
   if (pRaw == NULL) goto _OVER;
 
   int32_t dataPos = 0;
-
   SDB_SET_BINARY(pRaw, dataPos, pSma->name, TSDB_TABLE_FNAME_LEN, _OVER)
   SDB_SET_BINARY(pRaw, dataPos, pSma->stb, TSDB_TABLE_FNAME_LEN, _OVER)
   SDB_SET_BINARY(pRaw, dataPos, pSma->db, TSDB_DB_FNAME_LEN, _OVER)
@@ -100,6 +100,7 @@ static SSdbRaw *mndSmaActionEncode(SSmaObj *pSma) {
   SDB_SET_INT32(pRaw, dataPos, pSma->tagsFilterLen, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pSma->sqlLen, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pSma->astLen, _OVER)
+
   if (pSma->exprLen > 0) {
     SDB_SET_BINARY(pRaw, dataPos, pSma->expr, pSma->exprLen, _OVER)
   }
@@ -115,6 +116,7 @@ static SSdbRaw *mndSmaActionEncode(SSmaObj *pSma) {
 
   SDB_SET_RESERVE(pRaw, dataPos, TSDB_SMA_RESERVE_SIZE, _OVER)
   SDB_SET_DATALEN(pRaw, dataPos, _OVER)
+
   terrno = 0;
 
 _OVER:
@@ -193,6 +195,7 @@ static SSdbRow *mndSmaActionDecode(SSdbRaw *pRaw) {
   }
 
   SDB_GET_RESERVE(pRaw, dataPos, TSDB_SMA_RESERVE_SIZE, _OVER)
+
   terrno = 0;
 
 _OVER:
@@ -383,6 +386,27 @@ static int32_t mndSetCreateSmaVgroupCommitLogs(SMnode *pMnode, STrans *pTrans, S
   return 0;
 }
 
+static int32_t mndSetUpdateSmaStbCommitLogs(SMnode *pMnode, STrans *pTrans, SStbObj *pStb) {
+  SStbObj stbObj = {0};
+  taosRLockLatch(&pStb->lock);
+  memcpy(&stbObj, pStb, sizeof(SStbObj));
+  taosRUnLockLatch(&pStb->lock);
+  stbObj.numOfColumns = 0;
+  stbObj.pColumns = NULL;
+  stbObj.numOfTags = 0;
+  stbObj.pTags = NULL;
+  stbObj.updateTime = taosGetTimestampMs();
+  stbObj.lock = 0;
+  stbObj.smaVer++;
+
+  SSdbRaw *pCommitRaw = mndStbActionEncode(&stbObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+
+  return 0;
+}
+
 static int32_t mndSetCreateSmaRedoActions(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SSmaObj *pSma) {
   SSdb   *pSdb = pMnode->pSdb;
   SVgObj *pVgroup = NULL;
@@ -457,7 +481,6 @@ static int32_t mndSetCreateSmaVgroupRedoActions(SMnode *pMnode, STrans *pTrans,
   pSma->schemaTag.pSchema[0].flags = 0;
   snprintf(pSma->schemaTag.pSchema[0].name, TSDB_COL_NAME_LEN, "groupId");
 
-
   int32_t smaContLen = 0;
   void   *pSmaReq = mndBuildVCreateSmaReq(pMnode, pVgroup, pSma, &smaContLen);
   if (pSmaReq == NULL) return -1;
@@ -481,6 +504,13 @@ static int32_t mndSetCreateSmaVgroupRedoActions(SMnode *pMnode, STrans *pTrans,
   return 0;
 }
 
+static void mndDestroySmaObj(SSmaObj *pSmaObj) {
+  if (pSmaObj) {
+    taosMemoryFreeClear(pSmaObj->schemaRow.pSchema);
+    taosMemoryFreeClear(pSmaObj->schemaTag.pSchema);
+  }
+}
+
 static int32_t mndCreateSma(SMnode *pMnode, SRpcMsg *pReq, SMCreateSmaReq *pCreate, SDbObj *pDb, SStbObj *pStb) {
   SSmaObj smaObj = {0};
   memcpy(smaObj.name, pCreate->name, TSDB_TABLE_FNAME_LEN);
@@ -504,34 +534,23 @@ static int32_t mndCreateSma(SMnode *pMnode, SRpcMsg *pReq, SMCreateSmaReq *pCrea
   smaObj.tagsFilterLen = pCreate->tagsFilterLen;
   smaObj.sqlLen = pCreate->sqlLen;
   smaObj.astLen = pCreate->astLen;
-
   if (smaObj.exprLen > 0) {
-    smaObj.expr = taosMemoryMalloc(smaObj.exprLen);
+    smaObj.expr = pCreate->expr;
-    if (smaObj.expr == NULL) goto _OVER;
-    memcpy(smaObj.expr, pCreate->expr, smaObj.exprLen);
   }
-
   if (smaObj.tagsFilterLen > 0) {
-    smaObj.tagsFilter = taosMemoryMalloc(smaObj.tagsFilterLen);
+    smaObj.tagsFilter = pCreate->tagsFilter;
-    if (smaObj.tagsFilter == NULL) goto _OVER;
-    memcpy(smaObj.tagsFilter, pCreate->tagsFilter, smaObj.tagsFilterLen);
   }
-
   if (smaObj.sqlLen > 0) {
-    smaObj.sql = taosMemoryMalloc(smaObj.sqlLen);
+    smaObj.sql = pCreate->sql;
-    if (smaObj.sql == NULL) goto _OVER;
-    memcpy(smaObj.sql, pCreate->sql, smaObj.sqlLen);
   }
-
   if (smaObj.astLen > 0) {
-    smaObj.ast = taosMemoryMalloc(smaObj.astLen);
+    smaObj.ast = pCreate->ast;
-    if (smaObj.ast == NULL) goto _OVER;
-    memcpy(smaObj.ast, pCreate->ast, smaObj.astLen);
   }
 
   SStreamObj streamObj = {0};
   tstrncpy(streamObj.name, pCreate->name, TSDB_STREAM_FNAME_LEN);
   tstrncpy(streamObj.sourceDb, pDb->name, TSDB_DB_FNAME_LEN);
+  tstrncpy(streamObj.targetDb, streamObj.sourceDb, TSDB_DB_FNAME_LEN);
   streamObj.createTime = taosGetTimestampMs();
   streamObj.updateTime = streamObj.createTime;
   streamObj.uid = mndGenerateUid(pCreate->name, strlen(pCreate->name));
@@ -559,6 +578,7 @@ static int32_t mndCreateSma(SMnode *pMnode, SRpcMsg *pReq, SMCreateSmaReq *pCrea
   if (mndSetCreateSmaVgroupRedoLogs(pMnode, pTrans, &streamObj.fixedSinkVg) != 0) goto _OVER;
   if (mndSetCreateSmaCommitLogs(pMnode, pTrans, &smaObj) != 0) goto _OVER;
   if (mndSetCreateSmaVgroupCommitLogs(pMnode, pTrans, &streamObj.fixedSinkVg) != 0) goto _OVER;
+  if (mndSetUpdateSmaStbCommitLogs(pMnode, pTrans, pStb) != 0) goto _OVER;
   if (mndSetCreateSmaRedoActions(pMnode, pTrans, pDb, &smaObj) != 0) goto _OVER;
   if (mndSetCreateSmaVgroupRedoActions(pMnode, pTrans, pDb, &streamObj.fixedSinkVg, &smaObj) != 0) goto _OVER;
   if (mndAddStreamToTrans(pMnode, &streamObj, pCreate->ast, STREAM_TRIGGER_AT_ONCE, 0, pTrans) != 0) goto _OVER;
@@ -567,6 +587,7 @@ static int32_t mndCreateSma(SMnode *pMnode, SRpcMsg *pReq, SMCreateSmaReq *pCrea
   code = 0;
 
 _OVER:
+  mndDestroySmaObj(&smaObj);
   mndTransDrop(pTrans);
   return code;
 }
@@ -599,7 +620,7 @@ static int32_t mndCheckCreateSmaReq(SMCreateSmaReq *pCreate) {
   return 0;
 }
 
-static int32_t mndProcessMCreateSmaReq(SRpcMsg *pReq) {
+static int32_t mndProcessCreateSmaReq(SRpcMsg *pReq) {
   SMnode        *pMnode = pReq->info.node;
   int32_t        code = -1;
   SStbObj       *pStb = NULL;
@@ -781,13 +802,17 @@ static int32_t mndSetDropSmaVgroupRedoActions(SMnode *pMnode, STrans *pTrans, SD
 }
 
 static int32_t mndDropSma(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SSmaObj *pSma) {
-  int32_t code = -1;
+  int32_t  code = -1;
-  SVgObj *pVgroup = NULL;
+  SVgObj  *pVgroup = NULL;
-  STrans *pTrans = NULL;
+  SStbObj *pStb = NULL;
+  STrans  *pTrans = NULL;
 
   pVgroup = mndAcquireVgroup(pMnode, pSma->dstVgId);
   if (pVgroup == NULL) goto _OVER;
 
+  pStb = mndAcquireStb(pMnode, pSma->stb);
+  if (pStb == NULL) goto _OVER;
+
   pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_CONFLICT_DB, pReq);
   if (pTrans == NULL) goto _OVER;
 
@@ -798,6 +823,7 @@ static int32_t mndDropSma(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SSmaObj *p
   if (mndSetDropSmaVgroupRedoLogs(pMnode, pTrans, pVgroup) != 0) goto _OVER;
   if (mndSetDropSmaCommitLogs(pMnode, pTrans, pSma) != 0) goto _OVER;
   if (mndSetDropSmaVgroupCommitLogs(pMnode, pTrans, pVgroup) != 0) goto _OVER;
+  if (mndSetUpdateSmaStbCommitLogs(pMnode, pTrans, pStb) != 0) goto _OVER;
   if (mndSetDropSmaRedoActions(pMnode, pTrans, pDb, pSma) != 0) goto _OVER;
   if (mndSetDropSmaVgroupRedoActions(pMnode, pTrans, pDb, pVgroup) != 0) goto _OVER;
   if (mndTransPrepare(pMnode, pTrans) != 0) goto _OVER;
@@ -807,10 +833,78 @@ static int32_t mndDropSma(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SSmaObj *p
 _OVER:
   mndTransDrop(pTrans);
   mndReleaseVgroup(pMnode, pVgroup);
+  mndReleaseStb(pMnode, pStb);
   return code;
 }
 
-static int32_t mndProcessMDropSmaReq(SRpcMsg *pReq) {
+int32_t mndDropSmasByStb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb) {
+  SSdb    *pSdb = pMnode->pSdb;
+  SSmaObj *pSma = NULL;
+  void    *pIter = NULL;
+  SVgObj  *pVgroup = NULL;
+  int32_t  code = -1;
+
+  while (1) {
+    pIter = sdbFetch(pSdb, SDB_SMA, pIter, (void **)&pSma);
+    if (pIter == NULL) break;
+
+    if (pSma->stbUid == pStb->uid) {
+      pVgroup = mndAcquireVgroup(pMnode, pSma->dstVgId);
+      if (pVgroup == NULL) goto _OVER;
+      if (mndSetDropSmaVgroupCommitLogs(pMnode, pTrans, pVgroup) != 0) goto _OVER;
+      if (mndSetDropSmaVgroupRedoActions(pMnode, pTrans, pDb, pVgroup) != 0) goto _OVER;
+      if (mndSetDropSmaCommitLogs(pMnode, pTrans, pSma) != 0) goto _OVER;
+      if (mndSetDropSmaRedoActions(pMnode, pTrans, pDb, pSma) != 0) goto _OVER;
+      mndReleaseVgroup(pMnode, pVgroup);
+      pVgroup = NULL;
+    }
+
+    sdbRelease(pSdb, pSma);
+  }
+
+  code = 0;
+
+_OVER:
+  sdbCancelFetch(pSdb, pIter);
+  sdbRelease(pSdb, pSma);
+  mndReleaseVgroup(pMnode, pVgroup);
+  return code;
+}
+
+int32_t mndDropSmasByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb) {
+  SSdb    *pSdb = pMnode->pSdb;
+  SSmaObj *pSma = NULL;
+  void    *pIter = NULL;
+  SVgObj  *pVgroup = NULL;
+  int32_t  code = -1;
+
+  while (1) {
+    pIter = sdbFetch(pSdb, SDB_SMA, pIter, (void **)&pSma);
+    if (pIter == NULL) break;
+
+    if (pSma->dbUid == pDb->uid) {
+      pVgroup = mndAcquireVgroup(pMnode, pSma->dstVgId);
+      if (pVgroup == NULL) goto _OVER;
+      if (mndSetDropSmaVgroupCommitLogs(pMnode, pTrans, pVgroup) != 0) goto _OVER;
+      if (mndSetDropSmaVgroupRedoActions(pMnode, pTrans, pDb, pVgroup) != 0) goto _OVER;
+      if (mndSetDropSmaCommitLogs(pMnode, pTrans, pSma) != 0) goto _OVER;
+      mndReleaseVgroup(pMnode, pVgroup);
+      pVgroup = NULL;
+    }
+
+    sdbRelease(pSdb, pSma);
+  }
+
+  code = 0;
+
+_OVER:
+  sdbCancelFetch(pSdb, pIter);
+  sdbRelease(pSdb, pSma);
+  mndReleaseVgroup(pMnode, pVgroup);
+  return code;
+}
+
+static int32_t mndProcessDropSmaReq(SRpcMsg *pReq) {
   SMnode      *pMnode = pReq->info.node;
   int32_t      code = -1;
   SUserObj    *pUser = NULL;
@@ -899,18 +993,30 @@ static int32_t mndGetSma(SMnode *pMnode, SUserIndexReq *indexReq, SUserIndexRsp
   return code;
 }
 
-static int32_t mndGetTableSma(SMnode *pMnode, STableIndexReq *indexReq, STableIndexRsp *rsp, bool *exist) {
+int32_t mndGetTableSma(SMnode *pMnode, char *tbFName, STableIndexRsp *rsp, bool *exist) {
-  int32_t  code = 0;
+  int32_t         code = 0;
-  SSmaObj *pSma = NULL;
+  SSmaObj        *pSma = NULL;
-  SSdb   *pSdb = pMnode->pSdb;
+  SSdb           *pSdb = pMnode->pSdb;
-  void   *pIter = NULL;
+  void           *pIter = NULL;
   STableIndexInfo info;
 
+  SStbObj *pStb = mndAcquireStb(pMnode, tbFName);
+  if (NULL == pStb) {
+    *exist = false;
+    return TSDB_CODE_SUCCESS;
+  }
+
+  strcpy(rsp->dbFName, pStb->db);
+  strcpy(rsp->tbName, pStb->name + strlen(pStb->db) + 1);
+  rsp->suid = pStb->uid;
+  rsp->version = pStb->smaVer;
+  mndReleaseStb(pMnode, pStb);
+
   while (1) {
     pIter = sdbFetch(pSdb, SDB_SMA, pIter, (void **)&pSma);
     if (pIter == NULL) break;
 
-    if (pSma->stb[0] != indexReq->tbFName[0] || strcmp(pSma->stb, indexReq->tbFName)) {
+    if (pSma->stb[0] != tbFName[0] || strcmp(pSma->stb, tbFName)) {
       continue;
     }
 
@@ -922,14 +1028,14 @@ static int32_t mndGetTableSma(SMnode *pMnode, STableIndexReq *indexReq, STableIn
     info.dstTbUid = pSma->dstTbUid;
     info.dstVgId = pSma->dstVgId;
 
-    SVgObj* pVg = mndAcquireVgroup(pMnode, pSma->dstVgId);
+    SVgObj *pVg = mndAcquireVgroup(pMnode, pSma->dstVgId);
     if (pVg == NULL) {
       code = -1;
       sdbRelease(pSdb, pSma);
       return code;
     }
     info.epSet = mndGetVgroupEpset(pMnode, pVg);
-    
+
     info.expr = taosMemoryMalloc(pSma->exprLen + 1);
     if (info.expr == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
@@ -953,7 +1059,7 @@ static int32_t mndGetTableSma(SMnode *pMnode, STableIndexReq *indexReq, STableIn
 
     sdbRelease(pSdb, pSma);
   }
-  
+
   return code;
 }
 
@@ -1005,10 +1111,10 @@ _OVER:
 
 static int32_t mndProcessGetTbSmaReq(SRpcMsg *pReq) {
   STableIndexReq indexReq = {0};
-  SMnode       *pMnode = pReq->info.node;
+  SMnode        *pMnode = pReq->info.node;
-  int32_t       code = -1;
+  int32_t        code = -1;
   STableIndexRsp rsp = {0};
-  bool          exist = false;
+  bool           exist = false;
 
   if (tDeserializeSTableIndexReq(pReq->pCont, pReq->contLen, &indexReq) != 0) {
     terrno = TSDB_CODE_INVALID_MSG;
@@ -1022,7 +1128,7 @@ static int32_t mndProcessGetTbSmaReq(SRpcMsg *pReq) {
     goto _OVER;
   }
 
-  code = mndGetTableSma(pMnode, &indexReq, &rsp, &exist);
+  code = mndGetTableSma(pMnode, indexReq.tbFName, &rsp, &exist);
   if (code) {
     goto _OVER;
   }
@@ -1055,7 +1161,6 @@ _OVER:
   return code;
 }
 
-
 static int32_t mndRetrieveSma(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {
   SMnode  *pMnode = pReq->info.node;
   SSdb    *pSdb = pMnode->pSdb;
@@ -1114,4 +1219,4 @@ static int32_t mndRetrieveSma(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBloc
 static void mndCancelGetNextSma(SMnode *pMnode, void *pIter) {
   SSdb *pSdb = pMnode->pSdb;
   sdbCancelFetch(pSdb, pIter);
-}
+}

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

@@ -23,10 +23,12 @@
 #include "mndPerfSchema.h"
 #include "mndScheduler.h"
 #include "mndShow.h"
+#include "mndSma.h"
 #include "mndTopic.h"
 #include "mndTrans.h"
 #include "mndUser.h"
 #include "mndVgroup.h"
+#include "mndSma.h"
 #include "tname.h"
 
 #define STB_VER_NUMBER   1
@@ -36,9 +38,9 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw);
 static int32_t  mndStbActionInsert(SSdb *pSdb, SStbObj *pStb);
 static int32_t  mndStbActionDelete(SSdb *pSdb, SStbObj *pStb);
 static int32_t  mndStbActionUpdate(SSdb *pSdb, SStbObj *pOld, SStbObj *pNew);
-static int32_t  mndProcessMCreateStbReq(SRpcMsg *pReq);
+static int32_t  mndProcessCreateStbReq(SRpcMsg *pReq);
-static int32_t  mndProcessMAlterStbReq(SRpcMsg *pReq);
+static int32_t  mndProcessAlterStbReq(SRpcMsg *pReq);
-static int32_t  mndProcessMDropStbReq(SRpcMsg *pReq);
+static int32_t  mndProcessDropStbReq(SRpcMsg *pReq);
 static int32_t  mndProcessTableMetaReq(SRpcMsg *pReq);
 static int32_t  mndRetrieveStb(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
 static void     mndCancelGetNextStb(SMnode *pMnode, void *pIter);
@@ -54,9 +56,9 @@ int32_t mndInitStb(SMnode *pMnode) {
       .deleteFp = (SdbDeleteFp)mndStbActionDelete,
   };
 
-  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_STB, mndProcessMCreateStbReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_STB, mndProcessCreateStbReq);
-  mndSetMsgHandle(pMnode, TDMT_MND_ALTER_STB, mndProcessMAlterStbReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_ALTER_STB, mndProcessAlterStbReq);
-  mndSetMsgHandle(pMnode, TDMT_MND_DROP_STB, mndProcessMDropStbReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_STB, mndProcessDropStbReq);
   mndSetMsgHandle(pMnode, TDMT_VND_CREATE_STB_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_VND_ALTER_STB_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_VND_DROP_STB_RSP, mndTransProcessRsp);
@@ -318,6 +320,7 @@ static int32_t mndStbActionUpdate(SSdb *pSdb, SStbObj *pOld, SStbObj *pNew) {
   pOld->updateTime = pNew->updateTime;
   pOld->tagVer = pNew->tagVer;
   pOld->colVer = pNew->colVer;
+  pOld->smaVer = pNew->smaVer;
   pOld->nextColId = pNew->nextColId;
   pOld->ttl = pNew->ttl;
   pOld->numOfColumns = pNew->numOfColumns;
@@ -361,7 +364,7 @@ SDbObj *mndAcquireDbByStb(SMnode *pMnode, const char *stbName) {
   return mndAcquireDb(pMnode, db);
 }
 
-static FORCE_INLINE int schemaExColIdCompare(const void *colId, const void *pSchema) {
+static FORCE_INLINE int32_t schemaExColIdCompare(const void *colId, const void *pSchema) {
   if (*(col_id_t *)colId < ((SSchema *)pSchema)->colId) {
     return -1;
   } else if (*(col_id_t *)colId > ((SSchema *)pSchema)->colId) {
@@ -395,14 +398,14 @@ static void *mndBuildVCreateStbReq(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pSt
     req.pRSmaParam.xFilesFactor = pStb->xFilesFactor;
     req.pRSmaParam.delay = pStb->delay;
     if (pStb->ast1Len > 0) {
-      if (mndConvertRsmaTask(&req.pRSmaParam.qmsg1, &req.pRSmaParam.qmsg1Len, pStb->pAst1, pStb->uid, STREAM_TRIGGER_AT_ONCE, 0,
+      if (mndConvertRsmaTask(&req.pRSmaParam.qmsg1, &req.pRSmaParam.qmsg1Len, pStb->pAst1, pStb->uid,
-                             req.pRSmaParam.xFilesFactor) != TSDB_CODE_SUCCESS) {
+                             STREAM_TRIGGER_AT_ONCE, 0, req.pRSmaParam.xFilesFactor) != TSDB_CODE_SUCCESS) {
         return NULL;
       }
     }
     if (pStb->ast2Len > 0) {
-      if (mndConvertRsmaTask(&req.pRSmaParam.qmsg2, &req.pRSmaParam.qmsg2Len, pStb->pAst2, pStb->uid, STREAM_TRIGGER_AT_ONCE, 0,
+      if (mndConvertRsmaTask(&req.pRSmaParam.qmsg2, &req.pRSmaParam.qmsg2Len, pStb->pAst2, pStb->uid,
-                             req.pRSmaParam.xFilesFactor) != TSDB_CODE_SUCCESS) {
+                             STREAM_TRIGGER_AT_ONCE, 0, req.pRSmaParam.xFilesFactor) != TSDB_CODE_SUCCESS) {
         return NULL;
       }
     }
@@ -761,7 +764,7 @@ int32_t mndAddStbToTrans(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *p
   return 0;
 }
 
-static int32_t mndProcessMCreateStbReq(SRpcMsg *pReq) {
+static int32_t mndProcessCreateStbReq(SRpcMsg *pReq) {
   SMnode        *pMnode = pReq->info.node;
   int32_t        code = -1;
   SStbObj       *pStb = NULL;
@@ -1271,7 +1274,7 @@ static int32_t mndBuildStbSchemaImp(SDbObj *pDb, SStbObj *pStb, const char *tbNa
   return 0;
 }
 
-static int32_t mndBuildStbSchema(SMnode *pMnode, const char *dbFName, const char *tbName, STableMetaRsp *pRsp) {
+static int32_t mndBuildStbSchema(SMnode *pMnode, const char *dbFName, const char *tbName, STableMetaRsp *pRsp, int32_t *smaVer) {
   char tbFName[TSDB_TABLE_FNAME_LEN] = {0};
   snprintf(tbFName, sizeof(tbFName), "%s.%s", dbFName, tbName);
 
@@ -1288,6 +1291,10 @@ static int32_t mndBuildStbSchema(SMnode *pMnode, const char *dbFName, const char
     return -1;
   }
 
+  if (smaVer) {
+    *smaVer = pStb->smaVer;
+  }
+
   int32_t code = mndBuildStbSchemaImp(pDb, pStb, tbName, pRsp);
   mndReleaseDb(pMnode, pDb);
   mndReleaseStb(pMnode, pStb);
@@ -1296,7 +1303,7 @@ static int32_t mndBuildStbSchema(SMnode *pMnode, const char *dbFName, const char
 
 static int32_t mndBuildSMAlterStbRsp(SDbObj *pDb, const SMAlterStbReq *pAlter, SStbObj *pObj, void **pCont,
                                      int32_t *pLen) {
-  int           ret;
+  int32_t       ret;
   SEncoder      ec = {0};
   uint32_t      contLen = 0;
   SMAlterStbRsp alterRsp = {0};
@@ -1415,7 +1422,7 @@ _OVER:
   return code;
 }
 
-static int32_t mndProcessMAlterStbReq(SRpcMsg *pReq) {
+static int32_t mndProcessAlterStbReq(SRpcMsg *pReq) {
   SMnode       *pMnode = pReq->info.node;
   int32_t       code = -1;
   SDbObj       *pDb = NULL;
@@ -1545,6 +1552,7 @@ static int32_t mndDropStb(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SStbObj *p
   if (mndSetDropStbRedoLogs(pMnode, pTrans, pStb) != 0) goto _OVER;
   if (mndSetDropStbCommitLogs(pMnode, pTrans, pStb) != 0) goto _OVER;
   if (mndSetDropStbRedoActions(pMnode, pTrans, pDb, pStb) != 0) goto _OVER;
+  if (mndDropSmasByStb(pMnode, pTrans, pDb, pStb) != 0) goto _OVER;
   if (mndTransPrepare(pMnode, pTrans) != 0) goto _OVER;
 
   code = 0;
@@ -1554,7 +1562,7 @@ _OVER:
   return code;
 }
 
-static int32_t mndProcessMDropStbReq(SRpcMsg *pReq) {
+static int32_t mndProcessDropStbReq(SRpcMsg *pReq) {
   SMnode      *pMnode = pReq->info.node;
   int32_t      code = -1;
   SUserObj    *pUser = NULL;
@@ -1634,7 +1642,7 @@ static int32_t mndProcessTableMetaReq(SRpcMsg *pReq) {
     }
   } else {
     mDebug("stb:%s.%s, start to retrieve meta", infoReq.dbFName, infoReq.tbName);
-    if (mndBuildStbSchema(pMnode, infoReq.dbFName, infoReq.tbName, &metaRsp) != 0) {
+    if (mndBuildStbSchema(pMnode, infoReq.dbFName, infoReq.tbName, &metaRsp, NULL) != 0) {
       goto _OVER;
     }
   }
@@ -1667,51 +1675,86 @@ _OVER:
   return code;
 }
 
-int32_t mndValidateStbInfo(SMnode *pMnode, SSTableMetaVersion *pStbVersions, int32_t numOfStbs, void **ppRsp,
+int32_t mndValidateStbInfo(SMnode *pMnode, SSTableVersion *pStbVersions, int32_t numOfStbs, void **ppRsp,
                            int32_t *pRspLen) {
-  STableMetaBatchRsp batchMetaRsp = {0};
+  SSTbHbRsp hbRsp = {0};
-  batchMetaRsp.pArray = taosArrayInit(numOfStbs, sizeof(STableMetaRsp));
+  hbRsp.pMetaRsp = taosArrayInit(numOfStbs, sizeof(STableMetaRsp));
-  if (batchMetaRsp.pArray == NULL) {
+  if (hbRsp.pMetaRsp == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
+  hbRsp.pIndexRsp = taosArrayInit(numOfStbs, sizeof(STableIndexRsp));
+  if (NULL == hbRsp.pIndexRsp) {
+    taosArrayDestroy(hbRsp.pMetaRsp);
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  
   for (int32_t i = 0; i < numOfStbs; ++i) {
-    SSTableMetaVersion *pStbVersion = &pStbVersions[i];
+    SSTableVersion *pStbVersion = &pStbVersions[i];
     pStbVersion->suid = be64toh(pStbVersion->suid);
     pStbVersion->sversion = ntohs(pStbVersion->sversion);
     pStbVersion->tversion = ntohs(pStbVersion->tversion);
+    pStbVersion->smaVer = ntohl(pStbVersion->smaVer);
 
     STableMetaRsp metaRsp = {0};
+    int32_t smaVer = 0;
     mDebug("stb:%s.%s, start to retrieve meta", pStbVersion->dbFName, pStbVersion->stbName);
-    if (mndBuildStbSchema(pMnode, pStbVersion->dbFName, pStbVersion->stbName, &metaRsp) != 0) {
+    if (mndBuildStbSchema(pMnode, pStbVersion->dbFName, pStbVersion->stbName, &metaRsp, &smaVer) != 0) {
       metaRsp.numOfColumns = -1;
       metaRsp.suid = pStbVersion->suid;
+      taosArrayPush(hbRsp.pMetaRsp, &metaRsp);
+      continue;
     }
 
     if (pStbVersion->sversion != metaRsp.sversion || pStbVersion->tversion != metaRsp.tversion) {
-      taosArrayPush(batchMetaRsp.pArray, &metaRsp);
+      taosArrayPush(hbRsp.pMetaRsp, &metaRsp);
     } else {
       tFreeSTableMetaRsp(&metaRsp);
     }
+
+    if (pStbVersion->smaVer && pStbVersion->smaVer != smaVer) {
+      bool exist = false;
+      char tbFName[TSDB_TABLE_FNAME_LEN];
+      STableIndexRsp indexRsp = {0};
+      indexRsp.pIndex = taosArrayInit(10, sizeof(STableIndexInfo));
+      if (NULL == indexRsp.pIndex) {
+        terrno = TSDB_CODE_OUT_OF_MEMORY;
+        return -1;
+      }
+      
+      sprintf(tbFName, "%s.%s", pStbVersion->dbFName, pStbVersion->stbName);
+      int32_t code = mndGetTableSma(pMnode, tbFName, &indexRsp, &exist);
+      if (code || !exist) {
+        indexRsp.suid = pStbVersion->suid;
+        indexRsp.version = -1;
+        indexRsp.pIndex = NULL;
+      }
+
+      strcpy(indexRsp.dbFName, pStbVersion->dbFName);
+      strcpy(indexRsp.tbName, pStbVersion->stbName);
+
+      taosArrayPush(hbRsp.pIndexRsp, &indexRsp);
+    }
   }
 
-  int32_t rspLen = tSerializeSTableMetaBatchRsp(NULL, 0, &batchMetaRsp);
+  int32_t rspLen = tSerializeSSTbHbRsp(NULL, 0, &hbRsp);
   if (rspLen < 0) {
-    tFreeSTableMetaBatchRsp(&batchMetaRsp);
+    tFreeSSTbHbRsp(&hbRsp);
     terrno = TSDB_CODE_INVALID_MSG;
     return -1;
   }
 
   void *pRsp = taosMemoryMalloc(rspLen);
   if (pRsp == NULL) {
-    tFreeSTableMetaBatchRsp(&batchMetaRsp);
+    tFreeSSTbHbRsp(&hbRsp);
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
   }
 
-  tSerializeSTableMetaBatchRsp(pRsp, rspLen, &batchMetaRsp);
+  tSerializeSSTbHbRsp(pRsp, rspLen, &hbRsp);
-  tFreeSTableMetaBatchRsp(&batchMetaRsp);
+  tFreeSSTbHbRsp(&hbRsp);
   *ppRsp = pRsp;
   *pRspLen = rspLen;
   return 0;

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

@@ -236,7 +236,7 @@ int32_t mndSyncPropose(SMnode *pMnode, SSdbRaw *pRaw, int32_t transId) {
     tsem_wait(&pMgmt->syncSem);
   } else if (code == TAOS_SYNC_PROPOSE_NOT_LEADER) {
     terrno = TSDB_CODE_APP_NOT_READY;
-  } else if (code == TAOS_SYNC_PROPOSE_OTHER_ERROR) {
+  } else if (code == TAOS_SYNC_OTHER_ERROR) {
     terrno = TSDB_CODE_SYN_INTERNAL_ERROR;
   } else {
     terrno = TSDB_CODE_APP_ERROR;
@@ -254,13 +254,16 @@ int32_t mndSyncPropose(SMnode *pMnode, SSdbRaw *pRaw, int32_t transId) {
 void mndSyncStart(SMnode *pMnode) {
   SSyncMgmt *pMgmt = &pMnode->syncMgmt;
   syncSetMsgCb(pMgmt->sync, &pMnode->msgCb);
+  syncStart(pMgmt->sync);
+  mDebug("mnode sync started, id:%" PRId64 " standby:%d", pMgmt->sync, pMgmt->standby);
 
+/*
   if (pMgmt->standby) {
     syncStartStandBy(pMgmt->sync);
   } else {
     syncStart(pMgmt->sync);
   }
-  mDebug("mnode sync started, id:%" PRId64 " standby:%d", pMgmt->sync, pMgmt->standby);
+*/  
 }
 
 void mndSyncStop(SMnode *pMnode) {}

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

@@ -1347,13 +1347,11 @@ int32_t mndKillTrans(SMnode *pMnode, STrans *pTrans) {
 
   for (int32_t i = 0; i < taosArrayGetSize(pArray); ++i) {
     STransAction *pAction = taosArrayGet(pArray, i);
-    if (pAction->errCode != 0) {
+    mInfo("trans:%d, %s:%d set processed for kill msg received, errCode from %s to success", pTrans->id,
-      mInfo("trans:%d, %s:%d set processed for kill msg received, errCode from %s to success", pTrans->id,
+          mndTransStr(pAction->stage), i, tstrerror(pAction->errCode));
-            mndTransStr(pAction->stage), i, tstrerror(pAction->errCode));
+    pAction->msgSent = 1;
-      pAction->msgSent = 1;
+    pAction->msgReceived = 1;
-      pAction->msgReceived = 1;
+    pAction->errCode = 0;
-      pAction->errCode = 0;
-    }
   }
 
   mndTransExecute(pMnode, pTrans);

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

@@ -322,6 +322,33 @@ void *mndBuildAlterVnodeReq(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup, int32_
   return pReq;
 }
 
+void *mndBuildSetVnodeStandbyReq(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup, int32_t *pContLen) {
+  SSetStandbyReq standbyReq = {0};
+  standbyReq.dnodeId = pDnode->id;
+  standbyReq.standby = 1;
+
+  int32_t contLen = tSerializeSSetStandbyReq(NULL, 0, &standbyReq);
+  if (contLen < 0) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  contLen += sizeof(SMsgHead);
+  void *pReq = taosMemoryMalloc(contLen);
+  if (pReq == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  tSerializeSSetStandbyReq((char *)pReq + sizeof(SMsgHead), contLen, &standbyReq);
+  SMsgHead *pHead = pReq;
+  pHead->contLen = htonl(contLen);
+  pHead->vgId = htonl(pVgroup->vgId);
+
+  *pContLen = contLen;
+  return pReq;
+}
+
 void *mndBuildDropVnodeReq(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup, int32_t *pContLen) {
   SDropVnodeReq dropReq = {0};
   dropReq.dnodeId = pDnode->id;
@@ -898,6 +925,39 @@ int32_t mndAddAlterVnodeAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgO
   return 0;
 }
 
+static int32_t mndAddSetVnodeStandByAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup,
+                                           SVnodeGid *pVgid, bool isRedo) {
+  STransAction action = {0};
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pVgid->dnodeId);
+  if (pDnode == NULL) return -1;
+  action.epSet = mndGetDnodeEpset(pDnode);
+  mndReleaseDnode(pMnode, pDnode);
+
+  int32_t contLen = 0;
+  void   *pReq = mndBuildSetVnodeStandbyReq(pMnode, pDnode, pDb, pVgroup, &contLen);
+  if (pReq == NULL) return -1;
+
+  action.pCont = pReq;
+  action.contLen = contLen;
+  action.msgType = TDMT_SYNC_SET_VNODE_STANDBY;
+  action.acceptableCode = TSDB_CODE_NODE_NOT_DEPLOYED;
+
+  if (isRedo) {
+    if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+      taosMemoryFree(pReq);
+      return -1;
+    }
+  } else {
+    if (mndTransAppendUndoAction(pTrans, &action) != 0) {
+      taosMemoryFree(pReq);
+      return -1;
+    }
+  }
+
+  return 0;
+}
+
 int32_t mndAddDropVnodeAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup, SVnodeGid *pVgid,
                               bool isRedo) {
   STransAction action = {0};
@@ -952,6 +1012,7 @@ int32_t mndSetMoveVgroupInfoToTrans(SMnode *pMnode, STrans *pTrans, SDbObj *pDb,
   SVnodeGid del = newVg.vnodeGid[vnIndex];
   newVg.vnodeGid[vnIndex] = newVg.vnodeGid[newVg.replica];
   memset(&newVg.vnodeGid[newVg.replica], 0, sizeof(SVnodeGid));
+  if (mndAddSetVnodeStandByAction(pMnode, pTrans, pDb, pVgroup, &del, true) != 0) return -1;
   if (mndAddAlterVnodeAction(pMnode, pTrans, pDb, &newVg, TDMT_VND_ALTER_REPLICA) != 0) return -1;
   if (mndAddDropVnodeAction(pMnode, pTrans, pDb, &newVg, &del, true) != 0) return -1;
   if (mndAddAlterVnodeConfirmAction(pMnode, pTrans, pDb, &newVg) != 0) return -1;
@@ -1031,6 +1092,7 @@ static int32_t mndAddDecVgroupReplicaFromTrans(SMnode *pMnode, STrans *pTrans, S
   memcpy(pGid, &pVgroup->vnodeGid[pVgroup->replica], sizeof(SVnodeGid));
   memset(&pVgroup->vnodeGid[pVgroup->replica], 0, sizeof(SVnodeGid));
 
+  if (mndAddSetVnodeStandByAction(pMnode, pTrans, pDb, pVgroup, &delGid, true) != 0) return -1;
   if (mndAddAlterVnodeAction(pMnode, pTrans, pDb, pVgroup, TDMT_VND_ALTER_REPLICA) != 0) return -1;
   if (mndAddDropVnodeAction(pMnode, pTrans, pDb, pVgroup, &delGid, true) != 0) return -1;
   if (mndAddAlterVnodeConfirmAction(pMnode, pTrans, pDb, pVgroup) != 0) return -1;
@@ -1341,12 +1403,14 @@ int32_t mndBuildAlterVgroupAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, S
 
       SVnodeGid del1 = {0};
       if (mndRemoveVnodeFromVgroup(pMnode, &newVgroup, pArray, &del1) != 0) return -1;
+      if (mndAddSetVnodeStandByAction(pMnode, pTrans, pDb, pVgroup, &del1, true) != 0) return -1;
       if (mndAddAlterVnodeAction(pMnode, pTrans, pDb, &newVgroup, TDMT_VND_ALTER_REPLICA) != 0) return -1;
       if (mndAddDropVnodeAction(pMnode, pTrans, pDb, &newVgroup, &del1, true) != 0) return -1;
       if (mndAddAlterVnodeConfirmAction(pMnode, pTrans, pDb, &newVgroup) != 0) return -1;
 
       SVnodeGid del2 = {0};
       if (mndRemoveVnodeFromVgroup(pMnode, &newVgroup, pArray, &del2) != 0) return -1;
+      if (mndAddSetVnodeStandByAction(pMnode, pTrans, pDb, pVgroup, &del2, true) != 0) return -1;
       if (mndAddAlterVnodeAction(pMnode, pTrans, pDb, &newVgroup, TDMT_VND_ALTER_REPLICA) != 0) return -1;
       if (mndAddDropVnodeAction(pMnode, pTrans, pDb, &newVgroup, &del2, true) != 0) return -1;
       if (mndAddAlterVnodeConfirmAction(pMnode, pTrans, pDb, &newVgroup) != 0) return -1;
@@ -1396,6 +1460,7 @@ static int32_t mndSplitVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SVgObj
   } else if (newVg1.replica == 3) {
     SVnodeGid del1 = {0};
     if (mndRemoveVnodeFromVgroup(pMnode, &newVg1, pArray, &del1) != 0) goto _OVER;
+    if (mndAddSetVnodeStandByAction(pMnode, pTrans, pDb, pVgroup, &del1, true) != 0) return -1;
     if (mndAddAlterVnodeAction(pMnode, pTrans, pDb, &newVg1, TDMT_VND_ALTER_REPLICA) != 0) goto _OVER;
     if (mndAddDropVnodeAction(pMnode, pTrans, pDb, &newVg1, &del1, true) != 0) goto _OVER;
     if (mndAddAlterVnodeConfirmAction(pMnode, pTrans, pDb, &newVg1) != 0) goto _OVER;

source/dnode/vnode/CMakeLists.txt

@@ -31,7 +31,7 @@ target_sources(
     "src/sma/smaEnv.c"
     "src/sma/smaOpen.c"
     "src/sma/smaRollup.c"
-    "src/sma/smaTimeRange2.c"
+    "src/sma/smaTimeRange.c"
 
     # tsdb
     "src/tsdb/tsdbCommit.c"

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

@@ -38,8 +38,6 @@ typedef struct SSmaStatItem SSmaStatItem;
 typedef struct SSmaKey      SSmaKey;
 typedef struct SRSmaInfo    SRSmaInfo;
 
-#define SMA_IVLD_FID INT_MIN
-
 struct SSmaEnv {
   TdThreadRwlock lock;
   int8_t         type;
@@ -49,45 +47,38 @@ struct SSmaEnv {
 #define SMA_ENV_LOCK(env)       ((env)->lock)
 #define SMA_ENV_TYPE(env)       ((env)->type)
 #define SMA_ENV_STAT(env)       ((env)->pStat)
-#define SMA_ENV_STAT_ITEMS(env) ((env)->pStat->smaStatItems)
+#define SMA_ENV_STAT_ITEM(env) ((env)->pStat->tsmaStatItem)
 
 struct SSmaStatItem {
-  int8_t state;  // ETsdbSmaStat
+  int8_t    state;  // ETsdbSmaStat
-  STSma *pTSma;  // cache schema
+  STSma    *pTSma;  // cache schema
+  STSchema *pTSchema;
 };
 
 struct SSmaStat {
   union {
-    SHashObj *smaStatItems;  // key: indexUid, value: SSmaStatItem for tsma
+    SSmaStatItem tsmaStatItem;
-    SHashObj *rsmaInfoHash;  // key: stbUid, value: SRSmaInfo;
+    SHashObj    *rsmaInfoHash;  // key: stbUid, value: SRSmaInfo;
   };
   T_REF_DECLARE()
 };
-#define SMA_STAT_ITEMS(s)     ((s)->smaStatItems)
+#define SMA_STAT_ITEM(s)      ((s)->tsmaStatItem)
 #define SMA_STAT_INFO_HASH(s) ((s)->rsmaInfoHash)
 
 void  tdDestroySmaEnv(SSmaEnv *pSmaEnv);
 void *tdFreeSmaEnv(SSmaEnv *pSmaEnv);
-#if 0
-int32_t tbGetTSmaStatus(SSma *pSma, STSma *param, void *result);
-int32_t tbRemoveTSmaData(SSma *pSma, STSma *param, STimeWindow *pWin);
-#endif
 
-int32_t tdInitSma(SSma *pSma);
 int32_t tdDropTSma(SSma *pSma, char *pMsg);
 int32_t tdDropTSmaData(SSma *pSma, int64_t indexUid);
 int32_t tdInsertRSmaData(SSma *pSma, char *msg);
 
 int32_t tdRefSmaStat(SSma *pSma, SSmaStat *pStat);
 int32_t tdUnRefSmaStat(SSma *pSma, SSmaStat *pStat);
-int32_t tdCheckAndInitSmaEnv(SSma *pSma, int8_t smaType, bool onlyCheck);
+int32_t tdCheckAndInitSmaEnv(SSma *pSma, int8_t smaType);
 
 int32_t tdLockSma(SSma *pSma);
 int32_t tdUnLockSma(SSma *pSma);
 
-static FORCE_INLINE int16_t tdTSmaAdd(SSma *pSma, int16_t n) { return atomic_add_fetch_16(&SMA_TSMA_NUM(pSma), n); }
-static FORCE_INLINE int16_t tdTSmaSub(SSma *pSma, int16_t n) { return atomic_sub_fetch_16(&SMA_TSMA_NUM(pSma), n); }
-
 static FORCE_INLINE int32_t tdRLockSmaEnv(SSmaEnv *pEnv) {
   int code = taosThreadRwlockRdlock(&(pEnv->lock));
   if (code != 0) {
@@ -160,11 +151,10 @@ static FORCE_INLINE void tdSmaStatSetDropped(SSmaStatItem *pStatItem) {
   }
 }
 
-static int32_t  tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType);
+static int32_t tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType);
-void           *tdFreeSmaStatItem(SSmaStatItem *pSmaStatItem);
+void          *tdFreeSmaStatItem(SSmaStatItem *pSmaStatItem);
-static int32_t  tdDestroySmaState(SSmaStat *pSmaStat, int8_t smaType);
+static int32_t tdDestroySmaState(SSmaStat *pSmaStat, int8_t smaType);
-static SSmaEnv *tdNewSmaEnv(const SSma *pSma, int8_t smaType, const char *path, SDiskID did);
+void          *tdFreeSmaState(SSmaStat *pSmaStat, int8_t smaType);
-static int32_t  tdInitSmaEnv(SSma *pSma, int8_t smaType, const char *path, SDiskID did, SSmaEnv **pEnv);
 
 void *tdFreeRSmaInfo(SRSmaInfo *pInfo);
 

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

@@ -41,7 +41,6 @@ extern "C" {
 #define tqTrace(...) do { if (tqDebugFlag & DEBUG_TRACE) { taosPrintLog("TQ  ", DEBUG_TRACE, tqDebugFlag, __VA_ARGS__); }} while(0)
 // clang-format on
 
-typedef struct STqOffsetCfg   STqOffsetCfg;
 typedef struct STqOffsetStore STqOffsetStore;
 
 // tqRead
@@ -127,14 +126,15 @@ typedef struct {
 } STqHandle;
 
 struct STQ {
-  char*     path;
+  char*           path;
-  SHashObj* pushMgr;       // consumerId -> STqHandle*
+  SHashObj*       pushMgr;       // consumerId -> STqHandle*
-  SHashObj* handles;       // subKey -> STqHandle
+  SHashObj*       handles;       // subKey -> STqHandle
-  SHashObj* pStreamTasks;  // taksId -> SStreamTask
+  SHashObj*       pStreamTasks;  // taksId -> SStreamTask
-  SVnode*   pVnode;
+  STqOffsetStore* pOffsetStore;
-  SWal*     pWal;
+  SVnode*         pVnode;
-  TDB*      pMetaStore;
+  SWal*           pWal;
-  TTB*      pExecStore;
+  TDB*            pMetaStore;
+  TTB*            pExecStore;
 };
 
 typedef struct {
@@ -157,17 +157,19 @@ int32_t tqMetaClose(STQ* pTq);
 int32_t tqMetaSaveHandle(STQ* pTq, const char* key, const STqHandle* pHandle);
 int32_t tqMetaDeleteHandle(STQ* pTq, const char* key);
 
+typedef struct {
+  int32_t size;
+} STqOffsetHead;
+
+STqOffsetStore* tqOffsetOpen();
+void            tqOffsetClose(STqOffsetStore*);
+STqOffset*      tqOffsetRead(STqOffsetStore* pStore, const char* subscribeKey);
+int32_t         tqOffsetWrite(STqOffsetStore* pStore, const STqOffset* pOffset);
+int32_t         tqOffsetSnapshot(STqOffsetStore* pStore);
+
 // tqSink
 void tqTableSink(SStreamTask* pTask, void* vnode, int64_t ver, void* data);
 
-// tqOffset
-STqOffsetStore* tqOffsetOpen(STqOffsetCfg*);
-void            tqOffsetClose(STqOffsetStore*);
-int64_t         tqOffsetFetch(STqOffsetStore* pStore, const char* subscribeKey);
-int32_t         tqOffsetCommit(STqOffsetStore* pStore, const char* subscribeKey, int64_t offset);
-int32_t         tqOffsetPersist(STqOffsetStore* pStore, const char* subscribeKey);
-int32_t         tqOffsetPersistAll(STqOffsetStore* pStore);
-
 #ifdef __cplusplus
 }
 #endif

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

@@ -137,6 +137,7 @@ int     tqCommit(STQ*);
 int32_t tqUpdateTbUidList(STQ* pTq, const SArray* tbUidList, bool isAdd);
 int32_t tqProcessVgChangeReq(STQ* pTq, char* msg, int32_t msgLen);
 int32_t tqProcessVgDeleteReq(STQ* pTq, char* msg, int32_t msgLen);
+int32_t tqProcessOffsetCommitReq(STQ* pTq, char* msg, int32_t msgLen);
 int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg, int32_t workerId);
 int32_t tqProcessTaskDeploy(STQ* pTq, char* msg, int32_t msgLen);
 int32_t tqProcessStreamTrigger(STQ* pTq, SSubmitReq* data);
@@ -146,6 +147,9 @@ int32_t tqProcessTaskRecoverReq(STQ* pTq, SRpcMsg* pMsg);
 int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg);
 int32_t tqProcessTaskRecoverRsp(STQ* pTq, SRpcMsg* pMsg);
 
+SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pSchema, bool createTb, int64_t suid,
+                            const char* stbFullName, int32_t vgId);
+
 // sma
 int32_t smaOpen(SVnode* pVnode);
 int32_t smaClose(SSma* pSma);
@@ -244,7 +248,6 @@ struct STbUidStore {
 };
 
 struct SSma {
-  int16_t       nTSma;
   bool          locked;
   TdThreadMutex mutex;
   SVnode*       pVnode;
@@ -260,7 +263,6 @@ struct SSma {
 #define SMA_META(s)       ((s)->pVnode->pMeta)
 #define SMA_VID(s)        TD_VID((s)->pVnode)
 #define SMA_TFS(s)        ((s)->pVnode->pTfs)
-#define SMA_TSMA_NUM(s)   ((s)->nTSma)
 #define SMA_TSMA_ENV(s)   ((s)->pTSmaEnv)
 #define SMA_RSMA_ENV(s)   ((s)->pRSmaEnv)
 #define SMA_RSMA_TSDB0(s) ((s)->pVnode->pTsdb)

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

@@ -75,7 +75,7 @@ int metaDecodeEntry(SDecoder *pCoder, SMetaEntry *pME) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
     }
-    if (tDecodeTSma(pCoder, pME->smaEntry.tsma) < 0) return -1;
+    if (tDecodeTSma(pCoder, pME->smaEntry.tsma, true) < 0) return -1;
   } else {
     ASSERT(0);
   }

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

@@ -663,12 +663,23 @@ int32_t metaFilteTableIds(SMeta *pMeta, SMetaFltParam *param, SArray *pUids) {
 
   void *  entryKey = NULL, *entryVal = NULL;
   int32_t nEntryKey, nEntryVal;
+  bool    first = true;
   while (1) {
     valid = tdbTbcGet(pCursor->pCur, (const void **)&entryKey, &nEntryKey, (const void **)&entryVal, &nEntryVal);
     if (valid < 0) {
       break;
     }
     STagIdxKey *p = entryKey;
+    if (p->type != pCursor->type) {
+      if (first) {
+        valid = param->reverse ? tdbTbcMoveToPrev(pCursor->pCur) : tdbTbcMoveToNext(pCursor->pCur);
+        if (valid < 0) break;
+        continue;
+      } else {
+        break;
+      }
+    }
+    first = false;
     if (p != NULL) {
       int32_t cmp = (*param->filterFunc)(p->data, pKey->data, pKey->type);
       if (cmp == 0) {

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

@@ -44,3 +44,209 @@ int32_t smaGetTSmaDays(SVnodeCfg* pCfg, void* pCont, uint32_t contLen, int32_t*
   smaDebug("vgId:%d, get tsma days %d", pCfg->vgId, *days);
   return code;
 }
+
+#if 0
+
+/**
+ * @brief TODO: Assume that the final generated result it less than 3M
+ *
+ * @param pReq
+ * @param pDataBlocks
+ * @param vgId
+ * @param suid  // TODO: check with Liao whether suid response is reasonable
+ *
+ * TODO: colId should be set
+ */
+int32_t buildSubmitReqFromDataBlock(SSubmitReq** pReq, const SArray* pDataBlocks, STSchema* pTSchema, int32_t vgId,
+                                    tb_uid_t suid, const char* stbName, bool isCreateCtb) {
+  int32_t sz = taosArrayGetSize(pDataBlocks);
+  int32_t bufSize = sizeof(SSubmitReq);
+  for (int32_t i = 0; i < sz; ++i) {
+    SDataBlockInfo* pBlkInfo = &((SSDataBlock*)taosArrayGet(pDataBlocks, i))->info;
+    bufSize += pBlkInfo->rows * (TD_ROW_HEAD_LEN + pBlkInfo->rowSize + BitmapLen(pBlkInfo->numOfCols));
+    bufSize += sizeof(SSubmitBlk);
+  }
+
+  *pReq = taosMemoryCalloc(1, bufSize);
+  if (!(*pReq)) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return TSDB_CODE_FAILED;
+  }
+  void* pDataBuf = *pReq;
+
+  SArray*     pTagArray = NULL;
+  int32_t     msgLen = sizeof(SSubmitReq);
+  int32_t     numOfBlks = 0;
+  int32_t     schemaLen = 0;
+  SRowBuilder rb = {0};
+  tdSRowInit(&rb, pTSchema->version);
+
+  for (int32_t i = 0; i < sz; ++i) {
+    SSDataBlock*    pDataBlock = taosArrayGet(pDataBlocks, i);
+    SDataBlockInfo* pDataBlkInfo = &pDataBlock->info;
+    int32_t         colNum = pDataBlkInfo->numOfCols;
+    int32_t         rows = pDataBlkInfo->rows;
+    int32_t         rowSize = pDataBlkInfo->rowSize;
+    int64_t         groupId = pDataBlkInfo->groupId;
+
+    if (rb.nCols != colNum) {
+      tdSRowSetTpInfo(&rb, colNum, pTSchema->flen);
+    }
+
+   if(isCreateCtb) {
+     SMetaReader  mr = {0};
+     const char*  ctbName = buildCtbNameByGroupId(stbName, pDataBlock->info.groupId);
+     if (metaGetTableEntryByName(&mr, ctbName) != 0) {
+       smaDebug("vgId:%d, no tsma ctb %s exists", vgId, ctbName);
+     }
+     SVCreateTbReq ctbReq = {0};
+     ctbReq.name = ctbName;
+     ctbReq.type = TSDB_CHILD_TABLE;
+     ctbReq.ctb.suid = suid;
+
+     STagVal tagVal = {.cid = colNum + PRIMARYKEY_TIMESTAMP_COL_ID,
+                       .type = TSDB_DATA_TYPE_BIGINT,
+                       .i64 = groupId};
+     STag*   pTag = NULL;
+     if(!pTagArray) {
+       pTagArray = taosArrayInit(1, sizeof(STagVal));
+       if (!pTagArray) goto _err;
+     }
+     taosArrayClear(pTagArray);
+     taosArrayPush(pTagArray, &tagVal);
+     tTagNew(pTagArray, 1, false, &pTag);
+     if (pTag == NULL) {
+       tdDestroySVCreateTbReq(&ctbReq);
+       goto _err;
+      }
+      ctbReq.ctb.pTag = (uint8_t*)pTag;
+
+      int32_t code;
+      tEncodeSize(tEncodeSVCreateTbReq, &ctbReq, schemaLen, code);
+
+      tdDestroySVCreateTbReq(&ctbReq);
+      if (code < 0) {
+        goto _err;
+      }
+   }
+
+
+
+    SSubmitBlk* pSubmitBlk = POINTER_SHIFT(pDataBuf, msgLen);
+    pSubmitBlk->suid = suid;
+    pSubmitBlk->uid = groupId;
+    pSubmitBlk->numOfRows = rows;
+
+    msgLen += sizeof(SSubmitBlk);
+    int32_t dataLen = 0;
+    for (int32_t j = 0; j < rows; ++j) {                     // iterate by row
+      tdSRowResetBuf(&rb, POINTER_SHIFT(pDataBuf, msgLen));  // set row buf
+      bool    isStartKey = false;
+      int32_t offset = 0;
+      for (int32_t k = 0; k < colNum; ++k) {  // iterate by column
+        SColumnInfoData* pColInfoData = taosArrayGet(pDataBlock->pDataBlock, k);
+        STColumn*        pCol = &pTSchema->columns[k];
+        void*            var = POINTER_SHIFT(pColInfoData->pData, j * pColInfoData->info.bytes);
+        switch (pColInfoData->info.type) {
+          case TSDB_DATA_TYPE_TIMESTAMP:
+            if (!isStartKey) {
+              isStartKey = true;
+              tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID, TSDB_DATA_TYPE_TIMESTAMP, TD_VTYPE_NORM, var, true,
+                                  offset, k);
+
+            } else {
+              tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID + k, TSDB_DATA_TYPE_TIMESTAMP, TD_VTYPE_NORM, var,
+                                  true, offset, k);
+            }
+            break;
+          case TSDB_DATA_TYPE_NCHAR: {
+            tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID + k, TSDB_DATA_TYPE_NCHAR, TD_VTYPE_NORM, var, true,
+                                offset, k);
+            break;
+          }
+          case TSDB_DATA_TYPE_VARCHAR: {  // TSDB_DATA_TYPE_BINARY
+            tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID + k, TSDB_DATA_TYPE_VARCHAR, TD_VTYPE_NORM, var, true,
+                                offset, k);
+            break;
+          }
+          case TSDB_DATA_TYPE_VARBINARY:
+          case TSDB_DATA_TYPE_DECIMAL:
+          case TSDB_DATA_TYPE_BLOB:
+          case TSDB_DATA_TYPE_JSON:
+          case TSDB_DATA_TYPE_MEDIUMBLOB:
+            uError("the column type %" PRIi16 " is defined but not implemented yet", pColInfoData->info.type);
+            TASSERT(0);
+            break;
+          default:
+            if (pColInfoData->info.type < TSDB_DATA_TYPE_MAX && pColInfoData->info.type > TSDB_DATA_TYPE_NULL) {
+              if (pCol->type == pColInfoData->info.type) {
+                tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID + k, pCol->type, TD_VTYPE_NORM, var, true, offset,
+                                    k);
+              } else {
+                char tv[8] = {0};
+                if (pColInfoData->info.type == TSDB_DATA_TYPE_FLOAT) {
+                  float v = 0;
+                  GET_TYPED_DATA(v, float, pColInfoData->info.type, var);
+                  SET_TYPED_DATA(&tv, pCol->type, v);
+                } else if (pColInfoData->info.type == TSDB_DATA_TYPE_DOUBLE) {
+                  double v = 0;
+                  GET_TYPED_DATA(v, double, pColInfoData->info.type, var);
+                  SET_TYPED_DATA(&tv, pCol->type, v);
+                } else if (IS_SIGNED_NUMERIC_TYPE(pColInfoData->info.type)) {
+                  int64_t v = 0;
+                  GET_TYPED_DATA(v, int64_t, pColInfoData->info.type, var);
+                  SET_TYPED_DATA(&tv, pCol->type, v);
+                } else {
+                  uint64_t v = 0;
+                  GET_TYPED_DATA(v, uint64_t, pColInfoData->info.type, var);
+                  SET_TYPED_DATA(&tv, pCol->type, v);
+                }
+                tdAppendColValToRow(&rb, PRIMARYKEY_TIMESTAMP_COL_ID + k, pCol->type, TD_VTYPE_NORM, tv, true, offset,
+                                    k);
+              }
+            } else {
+              uError("the column type %" PRIi16 " is undefined\n", pColInfoData->info.type);
+              TASSERT(0);
+            }
+            break;
+        }
+        offset += TYPE_BYTES[pCol->type];  // sum/avg would convert to int64_t/uint64_t/double during aggregation
+      }
+      dataLen += TD_ROW_LEN(rb.pBuf);
+#ifdef TD_DEBUG_PRINT_ROW
+      tdSRowPrint(rb.pBuf, pTSchema, __func__);
+#endif
+    }
+
+    ++numOfBlks;
+
+    pSubmitBlk->dataLen = dataLen;
+    msgLen += pSubmitBlk->dataLen;
+  }
+
+  (*pReq)->length = msgLen;
+
+  (*pReq)->header.vgId = htonl(vgId);
+  (*pReq)->header.contLen = htonl(msgLen);
+  (*pReq)->length = (*pReq)->header.contLen;
+  (*pReq)->numOfBlocks = htonl(numOfBlks);
+  SSubmitBlk* blk = (SSubmitBlk*)((*pReq) + 1);
+  while (numOfBlks--) {
+    int32_t dataLen = blk->dataLen;
+    blk->uid = htobe64(blk->uid);
+    blk->suid = htobe64(blk->suid);
+    blk->padding = htonl(blk->padding);
+    blk->sversion = htonl(blk->sversion);
+    blk->dataLen = htonl(blk->dataLen);
+    blk->schemaLen = htonl(blk->schemaLen);
+    blk->numOfRows = htons(blk->numOfRows);
+    blk = (SSubmitBlk*)(blk->data + dataLen);
+  }
+  return TSDB_CODE_SUCCESS;
+_err:
+  taosMemoryFreeClear(*pReq);
+  taosArrayDestroy(pTagArray);
+
+  return TSDB_CODE_FAILED;
+}
+#endif

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

@@ -17,123 +17,17 @@
 
 typedef struct SSmaStat SSmaStat;
 
-static const char *TSDB_SMA_DNAME[] = {
-    "",      // TSDB_SMA_TYPE_BLOCK
-    "tsma",  // TSDB_SMA_TYPE_TIME_RANGE
-    "rsma",  // TSDB_SMA_TYPE_ROLLUP
-};
-
-#define SMA_TEST_INDEX_NAME "smaTestIndexName"  // TODO: just for test
-#define SMA_TEST_INDEX_UID  2000000001          // TODO: just for test
-#define SMA_STATE_HASH_SLOT 4
-
 #define RSMA_TASK_INFO_HASH_SLOT 8
 
-typedef struct SPoolMem {
-  int64_t          size;
-  struct SPoolMem *prev;
-  struct SPoolMem *next;
-} SPoolMem;
-
 // declaration of static functions
 
-// insert data
+static int32_t  tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType);
-
+static SSmaEnv *tdNewSmaEnv(const SSma *pSma, int8_t smaType, const char *path);
-static void tdGetSmaDir(int32_t vgId, ETsdbSmaType smaType, char dirName[]);
+static int32_t  tdInitSmaEnv(SSma *pSma, int8_t smaType, const char *path, SSmaEnv **pEnv);
-
-// Pool Memory
-static SPoolMem *openPool();
-static void      clearPool(SPoolMem *pPool);
-static void      closePool(SPoolMem *pPool);
-static void     *poolMalloc(void *arg, size_t size);
-static void      poolFree(void *arg, void *ptr);
 
 // implementation
 
-static SPoolMem *openPool() {
+static SSmaEnv *tdNewSmaEnv(const SSma *pSma, int8_t smaType, const char *path) {
-  SPoolMem *pPool = (SPoolMem *)taosMemoryMalloc(sizeof(*pPool));
-
-  pPool->prev = pPool->next = pPool;
-  pPool->size = 0;
-
-  return pPool;
-}
-
-static void clearPool(SPoolMem *pPool) {
-  if (!pPool) return;
-
-  SPoolMem *pMem;
-
-  do {
-    pMem = pPool->next;
-
-    if (pMem == pPool) break;
-
-    pMem->next->prev = pMem->prev;
-    pMem->prev->next = pMem->next;
-    pPool->size -= pMem->size;
-
-    taosMemoryFree(pMem);
-  } while (1);
-
-  assert(pPool->size == 0);
-}
-
-static void closePool(SPoolMem *pPool) {
-  if (pPool) {
-    clearPool(pPool);
-    taosMemoryFree(pPool);
-  }
-}
-
-static void *poolMalloc(void *arg, size_t size) {
-  void     *ptr = NULL;
-  SPoolMem *pPool = (SPoolMem *)arg;
-  SPoolMem *pMem;
-
-  pMem = (SPoolMem *)taosMemoryMalloc(sizeof(*pMem) + size);
-  if (!pMem) {
-    assert(0);
-  }
-
-  pMem->size = sizeof(*pMem) + size;
-  pMem->next = pPool->next;
-  pMem->prev = pPool;
-
-  pPool->next->prev = pMem;
-  pPool->next = pMem;
-  pPool->size += pMem->size;
-
-  ptr = (void *)(&pMem[1]);
-  return ptr;
-}
-
-static void poolFree(void *arg, void *ptr) {
-  SPoolMem *pPool = (SPoolMem *)arg;
-  SPoolMem *pMem;
-
-  pMem = &(((SPoolMem *)ptr)[-1]);
-
-  pMem->next->prev = pMem->prev;
-  pMem->prev->next = pMem->next;
-  pPool->size -= pMem->size;
-
-  taosMemoryFree(pMem);
-}
-
-int32_t tdInitSma(SSma *pSma) {
-  int32_t numOfTSma = taosArrayGetSize(metaGetSmaTbUids(SMA_META(pSma)));
-  if (numOfTSma > 0) {
-    atomic_store_16(&SMA_TSMA_NUM(pSma), (int16_t)numOfTSma);
-  }
-  return TSDB_CODE_SUCCESS;
-}
-
-static void tdGetSmaDir(int32_t vgId, ETsdbSmaType smaType, char dirName[]) {
-  snprintf(dirName, TSDB_FILENAME_LEN, "vnode%svnode%d%s%s", TD_DIRSEP, vgId, TD_DIRSEP, TSDB_SMA_DNAME[smaType]);
-}
-
-static SSmaEnv *tdNewSmaEnv(const SSma *pSma, int8_t smaType, const char *path, SDiskID did) {
   SSmaEnv *pEnv = NULL;
 
   pEnv = (SSmaEnv *)taosMemoryCalloc(1, sizeof(SSmaEnv));
@@ -156,18 +50,17 @@ static SSmaEnv *tdNewSmaEnv(const SSma *pSma, int8_t smaType, const char *path,
     return NULL;
   }
 
-
   return pEnv;
 }
 
-static int32_t tdInitSmaEnv(SSma *pSma, int8_t smaType, const char *path, SDiskID did, SSmaEnv **pEnv) {
+static int32_t tdInitSmaEnv(SSma *pSma, int8_t smaType, const char *path, SSmaEnv **pEnv) {
   if (!pEnv) {
     terrno = TSDB_CODE_INVALID_PTR;
     return TSDB_CODE_FAILED;
   }
 
   if (!(*pEnv)) {
-    if (!(*pEnv = tdNewSmaEnv(pSma, smaType, path, did))) {
+    if (!(*pEnv = tdNewSmaEnv(pSma, smaType, path))) {
       return TSDB_CODE_FAILED;
     }
   }
@@ -183,15 +76,16 @@ static int32_t tdInitSmaEnv(SSma *pSma, int8_t smaType, const char *path, SDiskI
  */
 void tdDestroySmaEnv(SSmaEnv *pSmaEnv) {
   if (pSmaEnv) {
-    tdDestroySmaState(pSmaEnv->pStat, SMA_ENV_TYPE(pSmaEnv));
+    pSmaEnv->pStat = tdFreeSmaState(pSmaEnv->pStat, SMA_ENV_TYPE(pSmaEnv));
-    taosMemoryFreeClear(pSmaEnv->pStat);
     taosThreadRwlockDestroy(&(pSmaEnv->lock));
   }
 }
 
 void *tdFreeSmaEnv(SSmaEnv *pSmaEnv) {
-  tdDestroySmaEnv(pSmaEnv);
+  if (pSmaEnv) {
-  taosMemoryFreeClear(pSmaEnv);
+    tdDestroySmaEnv(pSmaEnv);
+    taosMemoryFreeClear(pSmaEnv);
+  }
   return NULL;
 }
 
@@ -239,13 +133,7 @@ static int32_t tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType) {
         return TSDB_CODE_FAILED;
       }
     } else if (smaType == TSDB_SMA_TYPE_TIME_RANGE) {
-      SMA_STAT_ITEMS(*pSmaStat) =
+      // TODO
-          taosHashInit(SMA_STATE_HASH_SLOT, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
-
-      if (!SMA_STAT_ITEMS(*pSmaStat)) {
-        taosMemoryFreeClear(*pSmaStat);
-        return TSDB_CODE_FAILED;
-      }
     } else {
       ASSERT(0);
     }
@@ -262,6 +150,12 @@ void *tdFreeSmaStatItem(SSmaStatItem *pSmaStatItem) {
   return NULL;
 }
 
+void* tdFreeSmaState(SSmaStat *pSmaStat, int8_t smaType) {
+  tdDestroySmaState(pSmaStat, smaType);
+  taosMemoryFreeClear(pSmaStat);
+  return NULL;
+}
+
 /**
  * @brief Release resources allocated for its member fields, not including itself.
  *
@@ -270,16 +164,10 @@ void *tdFreeSmaStatItem(SSmaStatItem *pSmaStatItem) {
  */
 int32_t tdDestroySmaState(SSmaStat *pSmaStat, int8_t smaType) {
   if (pSmaStat) {
-    // TODO: use taosHashSetFreeFp when taosHashSetFreeFp is ready.
     if (smaType == TSDB_SMA_TYPE_TIME_RANGE) {
-      void *item = taosHashIterate(SMA_STAT_ITEMS(pSmaStat), NULL);
+      tdFreeSmaStatItem(&pSmaStat->tsmaStatItem);
-      while (item) {
-        SSmaStatItem *pItem = *(SSmaStatItem **)item;
-        tdFreeSmaStatItem(pItem);
-        item = taosHashIterate(SMA_STAT_ITEMS(pSmaStat), item);
-      }
-      taosHashCleanup(SMA_STAT_ITEMS(pSmaStat));
     } else if (smaType == TSDB_SMA_TYPE_ROLLUP) {
+      // TODO: use taosHashSetFreeFp when taosHashSetFreeFp is ready.
       void *infoHash = taosHashIterate(SMA_STAT_INFO_HASH(pSmaStat), NULL);
       while (infoHash) {
         SRSmaInfo *pInfoHash = *(SRSmaInfo **)infoHash;
@@ -317,7 +205,7 @@ int32_t tdUnLockSma(SSma *pSma) {
   return 0;
 }
 
-int32_t tdCheckAndInitSmaEnv(SSma *pSma, int8_t smaType, bool onlyCheck) {
+int32_t tdCheckAndInitSmaEnv(SSma *pSma, int8_t smaType) {
   SSmaEnv *pEnv = NULL;
 
   // return if already init
@@ -344,26 +232,7 @@ int32_t tdCheckAndInitSmaEnv(SSma *pSma, int8_t smaType, bool onlyCheck) {
   if (!pEnv) {
     char rname[TSDB_FILENAME_LEN] = {0};
 
-    SDiskID did = {0};
+    if (tdInitSmaEnv(pSma, smaType, rname, &pEnv) < 0) {
-    if (tfsAllocDisk(SMA_TFS(pSma), TFS_PRIMARY_LEVEL, &did) < 0) {
-      tdUnLockSma(pSma);
-      return TSDB_CODE_FAILED;
-    }
-
-    if (did.level < 0 || did.id < 0) {
-      tdUnLockSma(pSma);
-      smaError("vgId:%d, init sma env failed since invalid did(%d,%d)", SMA_VID(pSma), did.level, did.id);
-      return TSDB_CODE_FAILED;
-    }
-
-    tdGetSmaDir(SMA_VID(pSma), smaType, rname);
-
-    if (tfsMkdirRecurAt(SMA_TFS(pSma), rname, did) < 0) {
-      tdUnLockSma(pSma);
-      return TSDB_CODE_FAILED;
-    }
-
-    if (tdInitSmaEnv(pSma, smaType, rname, did, &pEnv) < 0) {
       tdUnLockSma(pSma);
       return TSDB_CODE_FAILED;
     }

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

@@ -132,7 +132,9 @@ int32_t smaClose(SSma *pSma) {
     if SMA_RSMA_TSDB0 (pSma) tsdbClose(&SMA_RSMA_TSDB0(pSma));
     if SMA_RSMA_TSDB1 (pSma) tsdbClose(&SMA_RSMA_TSDB1(pSma));
     if SMA_RSMA_TSDB2 (pSma) tsdbClose(&SMA_RSMA_TSDB2(pSma));
-    taosMemoryFree(pSma);
+    // SMA_TSMA_ENV(pSma) = tdFreeSmaEnv(SMA_TSMA_ENV(pSma));
+    // SMA_RSMA_ENV(pSma) = tdFreeSmaEnv(SMA_RSMA_ENV(pSma));
+    taosMemoryFreeClear(pSma);
   }
   return 0;
 }

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

@@ -181,7 +181,7 @@ int32_t tdProcessRSmaCreate(SVnode *pVnode, SVCreateStbReq *pReq) {
     return TSDB_CODE_SUCCESS;
   }
 
-  if (tdCheckAndInitSmaEnv(pSma, TSDB_SMA_TYPE_ROLLUP, false) != TSDB_CODE_SUCCESS) {
+  if (tdCheckAndInitSmaEnv(pSma, TSDB_SMA_TYPE_ROLLUP) != TSDB_CODE_SUCCESS) {
     terrno = TSDB_CODE_TDB_INIT_FAILED;
     return TSDB_CODE_FAILED;
   }

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

@@ -142,7 +142,6 @@ int32_t tdProcessTSmaInsertImpl(SSma *pSma, int64_t indexUid, const char *msg) {
   ASSERT(pItem);
 
   if (!pItem->pTSma) {
-    // cache smaMeta
     STSma *pTSma = metaGetSmaInfoByIndex(SMA_META(pSma), indexUid);
     if (!pTSma) {
       terrno = TSDB_CODE_TSMA_NO_INDEX_IN_META;
@@ -150,27 +149,28 @@ int32_t tdProcessTSmaInsertImpl(SSma *pSma, int64_t indexUid, const char *msg) {
       return TSDB_CODE_FAILED;
     }
     pItem->pTSma = pTSma;
+    pItem->pTSchema = metaGetTbTSchema(SMA_META(pSma), pTSma->dstTbUid, -1);
+    ASSERT(pItem->pTSchema);  // TODO
   }
 
-  STSma *pTSma = pItem->pTSma;
+  ASSERT(pItem->pTSma->indexUid == indexUid);
-
-  ASSERT(pTSma->indexUid == indexUid);
-
-  SMetaReader mr = {0};
-
-  const char *dbName = "testDb";
-  if (metaGetTableEntryByName(&mr, dbName) != 0) {
-    smaDebug("vgId:%d, tsma no table testTb exists for smaIndex %" PRIi64 " since %s", SMA_VID(pSma), indexUid, tstrerror(terrno));
-    SVCreateStbReq pReq = {0};
-    pReq.name = dbName;
-    pReq.suid = pTSma->dstTbUid;
-    pReq.schemaRow = pCfg->schemaRow;
-    pReq.schemaTag = pCfg->schemaTag;
-  }
 
   SSubmitReq *pSubmitReq = NULL;
-  buildSubmitReqFromDataBlock(&pSubmitReq, (const SArray *)msg, NULL, pItem->pTSma->dstVgId,
+
-                              pItem->pTSma->dstTbUid);
+  pSubmitReq = tdBlockToSubmit((const SArray *)msg, pItem->pTSchema, true, pItem->pTSma->dstTbUid,
+                               pItem->pTSma->dstTbName, pItem->pTSma->dstVgId);
+
+  ASSERT(pSubmitReq);  // TODO
+
+  ASSERT(!strncasecmp("td.tsma.rst.tb", pItem->pTSma->dstTbName, 14));
+
+  SRpcMsg submitReqMsg = {
+      .msgType = TDMT_VND_SUBMIT,
+      .pCont = pSubmitReq,
+      .contLen = ntohl(pSubmitReq->length),
+  };
+
+  ASSERT(tmsgPutToQueue(&pSma->pVnode->msgCb, WRITE_QUEUE, &submitReqMsg) == 0);
 
   tdUnRefSmaStat(pSma, pStat);
 

source/dnode/vnode/src/sma/smaTimeRange2.c

@@ -1,170 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#include "sma.h"
-#include "tsdb.h"
-
-typedef STsdbCfg STSmaKeepCfg;
-
-#undef _TEST_SMA_PRINT_DEBUG_LOG_
-#define SMA_STORAGE_MINUTES_MAX  86400
-#define SMA_STORAGE_MINUTES_DAY  1440
-#define SMA_STORAGE_MINUTES_MIN  1440
-#define SMA_STORAGE_TSDB_MINUTES 86400
-#define SMA_STORAGE_TSDB_TIMES   10
-#define SMA_STORAGE_SPLIT_FACTOR 14400  // least records in tsma file TODO: the feasible value?
-#define SMA_KEY_LEN              16     // TSKEY+groupId 8+8
-#define SMA_DROP_EXPIRED_TIME    10     // default is 10 seconds
-
-#define SMA_STATE_ITEM_HASH_SLOT 32
-
-// static func
-
-/**
- * @brief Judge the tsma file split days
- *
- * @param pCfg
- * @param pCont
- * @param contLen
- * @param days unit is minute
- * @return int32_t
- */
-int32_t tdProcessTSmaGetDaysImpl(SVnodeCfg *pCfg, void *pCont, uint32_t contLen, int32_t *days) {
-  SDecoder coder = {0};
-  tDecoderInit(&coder, pCont, contLen);
-
-  STSma tsma = {0};
-  if (tDecodeSVCreateTSmaReq(&coder, &tsma) < 0) {
-    terrno = TSDB_CODE_MSG_DECODE_ERROR;
-    goto _err;
-  }
-  STsdbCfg *pTsdbCfg = &pCfg->tsdbCfg;
-  int64_t   sInterval = convertTimeFromPrecisionToUnit(tsma.interval, pTsdbCfg->precision, TIME_UNIT_SECOND);
-  if (sInterval <= 0) {
-    *days = pTsdbCfg->days;
-    return 0;
-  }
-  int64_t records = pTsdbCfg->days * 60 / sInterval;
-  if (records >= SMA_STORAGE_SPLIT_FACTOR) {
-    *days = pTsdbCfg->days;
-  } else {
-    int64_t mInterval = convertTimeFromPrecisionToUnit(tsma.interval, pTsdbCfg->precision, TIME_UNIT_MINUTE);
-    int64_t daysPerFile = mInterval * SMA_STORAGE_MINUTES_DAY * 2;
-
-    if (daysPerFile > SMA_STORAGE_MINUTES_MAX) {
-      *days = SMA_STORAGE_MINUTES_MAX;
-    } else {
-      *days = (int32_t)daysPerFile;
-    }
-
-    if (*days < pTsdbCfg->days) {
-      *days = pTsdbCfg->days;
-    }
-  }
-  tDecoderClear(&coder);
-  return 0;
-_err:
-  tDecoderClear(&coder);
-  return -1;
-}
-
-// read data
-
-// implementation
-
-/**
- * @brief Insert/Update Time-range-wise SMA data.
- *  - If interval < SMA_STORAGE_SPLIT_HOURS(e.g. 24), save the SMA data as a part of DFileSet to e.g.
- * v3f1900.tsma.${sma_index_name}. The days is the same with that for TS data files.
- *  - If interval >= SMA_STORAGE_SPLIT_HOURS, save the SMA data to e.g. vnode3/tsma/v3f632.tsma.${sma_index_name}. The
- * days is 30 times of the interval, and the minimum days is SMA_STORAGE_TSDB_DAYS(30d).
- *  - The destination file of one data block for some interval is determined by its start TS key.
- *
- * @param pSma
- * @param msg
- * @return int32_t
- */
-int32_t tdProcessTSmaInsertImpl(SSma *pSma, int64_t indexUid, const char *msg) {
-  STsdbCfg *pCfg = SMA_TSDB_CFG(pSma);
-
-  const SArray *pDataBlocks = (const SArray *)msg;
-
-  // TODO: destroy SSDataBlocks(msg)
-
-  // For super table aggregation, the sma data is stored in vgroup calculated from the hash value of stable name. Thus
-  // the sma data would arrive ahead of the update-expired-window msg.
-  if (tdCheckAndInitSmaEnv(pSma, TSDB_SMA_TYPE_TIME_RANGE, false) != TSDB_CODE_SUCCESS) {
-    terrno = TSDB_CODE_TDB_INIT_FAILED;
-    return TSDB_CODE_FAILED;
-  }
-
-  if (!pDataBlocks) {
-    terrno = TSDB_CODE_INVALID_PTR;
-    smaWarn("vgId:%d, insert tsma data failed since pDataBlocks is NULL", SMA_VID(pSma));
-    return terrno;
-  }
-
-  if (taosArrayGetSize(pDataBlocks) <= 0) {
-    terrno = TSDB_CODE_INVALID_PARA;
-    smaWarn("vgId:%d, insert tsma data failed since pDataBlocks is empty", SMA_VID(pSma));
-    return TSDB_CODE_FAILED;
-  }
-
-  SSmaEnv      *pEnv = SMA_TSMA_ENV(pSma);
-  SSmaStat     *pStat = SMA_ENV_STAT(pEnv);
-  SSmaStatItem *pItem = NULL;
-
-  tdRefSmaStat(pSma, pStat);
-
-  if (pStat && SMA_STAT_ITEMS(pStat)) {
-    pItem = taosHashGet(SMA_STAT_ITEMS(pStat), &indexUid, sizeof(indexUid));
-  }
-
-  if (!pItem || !(pItem = *(SSmaStatItem **)pItem) || tdSmaStatIsDropped(pItem)) {
-    terrno = TSDB_CODE_TSMA_INVALID_STAT;
-    tdUnRefSmaStat(pSma, pStat);
-    return TSDB_CODE_FAILED;
-  }
-
-  STSma *pTSma = pItem->pTSma;
-
-  tdUnRefSmaStat(pSma, pStat);
-
-  return TSDB_CODE_SUCCESS;
-}
-
-int32_t tdProcessTSmaCreateImpl(SSma *pSma, int64_t version, const char *pMsg) {
-  SSmaCfg *pCfg = (SSmaCfg *)pMsg;
-
-  if (metaCreateTSma(SMA_META(pSma), version, pCfg) < 0) {
-    return -1;
-  }
-
-  if (TD_VID(pSma->pVnode) == pCfg->dstVgId) {
-    // create stable to save tsma result in dstVgId
-    SVCreateStbReq pReq = {0};
-    pReq.name = pCfg->dstTbName;
-    pReq.suid = pCfg->dstTbUid;
-    pReq.schemaRow = pCfg->schemaRow;
-    pReq.schemaTag = pCfg->schemaTag;
-
-    if (metaCreateSTable(SMA_META(pSma), version, &pReq) < 0) {
-      return -1;
-    }
-  }
-
-  tdTSmaAdd(pSma, 1);
-  return 0;
-}

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

@@ -47,7 +47,7 @@ void tqCleanUp() {
 }
 
 STQ* tqOpen(const char* path, SVnode* pVnode, SWal* pWal) {
-  STQ* pTq = taosMemoryMalloc(sizeof(STQ));
+  STQ* pTq = taosMemoryCalloc(1, sizeof(STQ));
   if (pTq == NULL) {
     terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return NULL;
@@ -66,19 +66,23 @@ STQ* tqOpen(const char* path, SVnode* pVnode, SWal* pWal) {
     ASSERT(0);
   }
 
+  if (tqOffsetOpen(pTq) < 0) {
+    ASSERT(0);
+  }
+
   return pTq;
 }
 
 void tqClose(STQ* pTq) {
   if (pTq) {
-    taosMemoryFreeClear(pTq->path);
+    tqOffsetClose(pTq->pOffsetStore);
     taosHashCleanup(pTq->handles);
     taosHashCleanup(pTq->pStreamTasks);
     taosHashCleanup(pTq->pushMgr);
+    taosMemoryFree(pTq->path);
     tqMetaClose(pTq);
     taosMemoryFree(pTq);
   }
-  // TODO
 }
 
 int32_t tqSendPollRsp(STQ* pTq, const SRpcMsg* pMsg, const SMqPollReq* pReq, const SMqDataBlkRsp* pRsp) {
@@ -109,6 +113,33 @@ int32_t tqSendPollRsp(STQ* pTq, const SRpcMsg* pMsg, const SMqPollReq* pReq, con
   return 0;
 }
 
+int32_t tqProcessOffsetCommitReq(STQ* pTq, char* msg, int32_t msgLen) {
+  STqOffset offset = {0};
+  SDecoder  decoder;
+  tDecoderInit(&decoder, msg, msgLen);
+  if (tDecodeSTqOffset(&decoder, &offset) < 0) {
+    ASSERT(0);
+    return -1;
+  }
+  tDecoderClear(&decoder);
+
+  if (offset.type == TMQ_OFFSET__SNAPSHOT) {
+    tqDebug("receive offset commit msg to %s, offset(type:snapshot) uid: %ld, ts: %ld", offset.subKey, offset.uid,
+            offset.ts);
+  } else if (offset.type == TMQ_OFFSET__LOG) {
+    tqDebug("receive offset commit msg to %s, offset(type:log) version: %ld", offset.subKey, offset.version);
+  } else {
+    ASSERT(0);
+  }
+
+  if (tqOffsetWrite(pTq->pOffsetStore, &offset) < 0) {
+    ASSERT(0);
+    return -1;
+  }
+
+  return 0;
+}
+
 int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg, int32_t workerId) {
   SMqPollReq* pReq = pMsg->pCont;
   int64_t     consumerId = pReq->consumerId;

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

@@ -16,26 +16,128 @@
 
 #include "tq.h"
 
-enum ETqOffsetPersist {
-  TQ_OFFSET_PERSIST__LAZY = 1,
-  TQ_OFFSET_PERSIST__EAGER,
-};
-
-struct STqOffsetCfg {
-  int8_t persistPolicy;
-};
-
 struct STqOffsetStore {
-  STqOffsetCfg cfg;
+  char*     fname;
-  SHashObj*    pHash;  // SHashObj<subscribeKey, offset>
+  STQ*      pTq;
+  SHashObj* pHash;  // SHashObj<subscribeKey, offset>
 };
 
-STqOffsetStore* tqOffsetOpen(STqOffsetCfg* pCfg) {
+static char* buildFileName(const char* path) {
-  STqOffsetStore* pStore = taosMemoryMalloc(sizeof(STqOffsetStore));
+  int32_t len = strlen(path);
+  char*   fname = taosMemoryCalloc(1, len + 20);
+  snprintf(fname, len + 20, "%s/offset", path);
+  return fname;
+}
+
+STqOffsetStore* tqOffsetOpen(STQ* pTq) {
+  STqOffsetStore* pStore = taosMemoryCalloc(1, sizeof(STqOffsetStore));
   if (pStore == NULL) {
     return NULL;
   }
-  memcpy(&pStore->cfg, pCfg, sizeof(STqOffsetCfg));
+  pStore->pTq = pTq;
+  pTq->pOffsetStore = pStore;
+
   pStore->pHash = taosHashInit(64, MurmurHash3_32, true, HASH_NO_LOCK);
+  if (pStore->pHash == NULL) {
+    if (pStore->pHash) taosHashCleanup(pStore->pHash);
+    return NULL;
+  }
+  char*     fname = buildFileName(pStore->pTq->path);
+  TdFilePtr pFile = taosOpenFile(fname, TD_FILE_READ);
+  if (pFile != NULL) {
+    STqOffsetHead head = {0};
+    int64_t       code;
+
+    while (1) {
+      if ((code = taosReadFile(pFile, &head, sizeof(STqOffsetHead))) != sizeof(STqOffsetHead)) {
+        if (code == 0) {
+          break;
+        } else {
+          ASSERT(0);
+          // TODO handle error
+        }
+      }
+      int32_t size = htonl(head.size);
+      void*   memBuf = taosMemoryCalloc(1, size);
+      if ((code = taosReadFile(pFile, memBuf, size)) != size) {
+        ASSERT(0);
+        // TODO handle error
+      }
+      STqOffset offset;
+      SDecoder  decoder;
+      tDecoderInit(&decoder, memBuf, size);
+      if (tDecodeSTqOffset(&decoder, &offset) < 0) {
+        ASSERT(0);
+      }
+      tDecoderClear(&decoder);
+      if (taosHashPut(pStore->pHash, offset.subKey, strlen(offset.subKey), &offset, sizeof(STqOffset)) < 0) {
+        ASSERT(0);
+        // TODO
+      }
+    }
+
+    taosCloseFile(&pFile);
+    taosMemoryFree(fname);
+  }
   return pStore;
 }
+
+void tqOffsetClose(STqOffsetStore* pStore) {
+  tqOffsetSnapshot(pStore);
+  taosHashCleanup(pStore->pHash);
+}
+
+STqOffset* tqOffsetRead(STqOffsetStore* pStore, const char* subscribeKey) {
+  return (STqOffset*)taosHashGet(pStore->pHash, subscribeKey, strlen(subscribeKey));
+}
+
+int32_t tqOffsetWrite(STqOffsetStore* pStore, const STqOffset* pOffset) {
+  return taosHashPut(pStore->pHash, pOffset->subKey, strlen(pOffset->subKey), pOffset, sizeof(STqOffset));
+}
+
+int32_t tqOffsetSnapshot(STqOffsetStore* pStore) {
+  // open file
+  // TODO file name should be with a version
+  char*     fname = buildFileName(pStore->pTq->path);
+  TdFilePtr pFile = taosOpenFile(fname, TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_APPEND);
+  if (pFile == NULL) {
+    ASSERT(0);
+    return -1;
+  }
+  void* pIter = NULL;
+  while (1) {
+    pIter = taosHashIterate(pStore->pHash, pIter);
+    if (pIter == NULL) break;
+    STqOffset* pOffset = (STqOffset*)pIter;
+    int32_t    bodyLen;
+    int32_t    code;
+    tEncodeSize(tEncodeSTqOffset, pOffset, bodyLen, code);
+    ASSERT(code == 0);
+    if (code < 0) {
+      ASSERT(0);
+      taosHashCancelIterate(pStore->pHash, pIter);
+      return -1;
+    }
+
+    int32_t totLen = sizeof(STqOffsetHead) + bodyLen;
+    void*   buf = taosMemoryCalloc(1, totLen);
+    void*   abuf = POINTER_SHIFT(buf, sizeof(STqOffsetHead));
+
+    ((STqOffsetHead*)buf)->size = htonl(bodyLen);
+    SEncoder encoder;
+    tEncoderInit(&encoder, abuf, bodyLen);
+    tEncodeSTqOffset(&encoder, pOffset);
+    // write file
+    int64_t writeLen;
+    if ((writeLen = taosWriteFile(pFile, buf, totLen)) != bodyLen) {
+      ASSERT(0);
+      tqError("write offset incomplete, len %d, write len %ld", bodyLen, writeLen);
+      taosHashCancelIterate(pStore->pHash, pIter);
+      return -1;
+    }
+  }
+  // close and rename file
+  taosCloseFile(&pFile);
+  taosMemoryFree(fname);
+  return 0;
+}

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

@@ -15,6 +15,174 @@
 
 #include "tq.h"
 
+SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pTSchema, bool createTb, int64_t suid,
+                                   const char* stbFullName, int32_t vgId) {
+  SSubmitReq* ret = NULL;
+  SArray*     tagArray = taosArrayInit(1, sizeof(STagVal));
+  if (!tagArray) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  // cal size
+  int32_t cap = sizeof(SSubmitReq);
+  int32_t sz = taosArrayGetSize(pBlocks);
+  for (int32_t i = 0; i < sz; i++) {
+    SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
+    int32_t      rows = pDataBlock->info.rows;
+    // TODO min
+    int32_t rowSize = pDataBlock->info.rowSize;
+    int32_t maxLen = TD_ROW_MAX_BYTES_FROM_SCHEMA(pTSchema);
+    int32_t schemaLen = 0;
+
+    if (createTb) {
+      SVCreateTbReq createTbReq = {0};
+      char*         cname = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
+      createTbReq.name = cname;
+      createTbReq.flags = 0;
+      createTbReq.type = TSDB_CHILD_TABLE;
+      createTbReq.ctb.suid = suid;
+
+      STagVal tagVal = {
+          .cid = pDataBlock->info.numOfCols + 1,
+          .type = TSDB_DATA_TYPE_UBIGINT,
+          .i64 = (int64_t)pDataBlock->info.groupId,
+      };
+      STag* pTag = NULL;
+      taosArrayClear(tagArray);
+      taosArrayPush(tagArray, &tagVal);
+      tTagNew(tagArray, 1, false, &pTag);
+      if (pTag == NULL) {
+        tdDestroySVCreateTbReq(&createTbReq);
+        taosArrayDestroy(tagArray);
+        return NULL;
+      }
+      createTbReq.ctb.pTag = (uint8_t*)pTag;
+
+      int32_t code;
+      tEncodeSize(tEncodeSVCreateTbReq, &createTbReq, schemaLen, code);
+
+      tdDestroySVCreateTbReq(&createTbReq);
+      if (code < 0) {
+        taosArrayDestroy(tagArray);
+        return NULL;
+      }
+    }
+
+    cap += sizeof(SSubmitBlk) + schemaLen + rows * maxLen;
+  }
+
+  // assign data
+  // TODO
+  ret = rpcMallocCont(cap);
+  ret->header.vgId = vgId;
+  ret->version = htonl(1);
+  ret->length = sizeof(SSubmitReq);
+  ret->numOfBlocks = htonl(sz);
+
+  void* submitBlk = POINTER_SHIFT(ret, sizeof(SSubmitReq));
+  for (int32_t i = 0; i < sz; i++) {
+    SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
+
+    SSubmitBlk* blkHead = submitBlk;
+    blkHead->numOfRows = htons(pDataBlock->info.rows);
+    blkHead->sversion = htonl(pTSchema->version);
+    // TODO
+    blkHead->suid = htobe64(suid);
+    // uid is assigned by vnode
+    blkHead->uid = 0;
+
+    int32_t rows = pDataBlock->info.rows;
+    /*int32_t maxLen = TD_ROW_MAX_BYTES_FROM_SCHEMA(pTSchema);*/
+    /*blkHead->dataLen = htonl(rows * maxLen);*/
+    blkHead->dataLen = 0;
+
+    void* blockData = POINTER_SHIFT(submitBlk, sizeof(SSubmitBlk));
+
+    int32_t schemaLen = 0;
+    if (createTb) {
+      SVCreateTbReq createTbReq = {0};
+      char*         cname = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
+      createTbReq.name = cname;
+      createTbReq.flags = 0;
+      createTbReq.type = TSDB_CHILD_TABLE;
+      createTbReq.ctb.suid = suid;
+
+      STagVal tagVal = {
+          .cid = pDataBlock->info.numOfCols + 1,
+          .type = TSDB_DATA_TYPE_UBIGINT,
+          .i64 = (int64_t)pDataBlock->info.groupId,
+      };
+      taosArrayClear(tagArray);
+      taosArrayPush(tagArray, &tagVal);
+      STag* pTag = NULL;
+      tTagNew(tagArray, 1, false, &pTag);
+      if (pTag == NULL) {
+        tdDestroySVCreateTbReq(&createTbReq);
+        taosArrayDestroy(tagArray);
+        taosMemoryFreeClear(ret);
+        return NULL;
+      }
+      createTbReq.ctb.pTag = (uint8_t*)pTag;
+
+      int32_t code;
+      tEncodeSize(tEncodeSVCreateTbReq, &createTbReq, schemaLen, code);
+      if (code < 0) {
+        tdDestroySVCreateTbReq(&createTbReq);
+        taosArrayDestroy(tagArray);
+        taosMemoryFreeClear(ret);
+        return NULL;
+      }
+
+      SEncoder encoder = {0};
+      tEncoderInit(&encoder, blockData, schemaLen);
+      code = tEncodeSVCreateTbReq(&encoder, &createTbReq);
+      tEncoderClear(&encoder);
+      tdDestroySVCreateTbReq(&createTbReq);
+
+      if (code < 0) {
+        taosArrayDestroy(tagArray);
+        taosMemoryFreeClear(ret);
+        return NULL;
+      }
+    }
+    blkHead->schemaLen = htonl(schemaLen);
+
+    STSRow* rowData = POINTER_SHIFT(blockData, schemaLen);
+
+    for (int32_t j = 0; j < rows; j++) {
+      SRowBuilder rb = {0};
+      tdSRowInit(&rb, pTSchema->version);
+      tdSRowSetTpInfo(&rb, pTSchema->numOfCols, pTSchema->flen);
+      tdSRowResetBuf(&rb, rowData);
+
+      for (int32_t k = 0; k < pTSchema->numOfCols; k++) {
+        const STColumn*  pColumn = &pTSchema->columns[k];
+        SColumnInfoData* pColData = taosArrayGet(pDataBlock->pDataBlock, k);
+        if (colDataIsNull_s(pColData, j)) {
+          tdAppendColValToRow(&rb, pColumn->colId, pColumn->type, TD_VTYPE_NONE, NULL, false, pColumn->offset, k);
+        } else {
+          void* data = colDataGetData(pColData, j);
+          tdAppendColValToRow(&rb, pColumn->colId, pColumn->type, TD_VTYPE_NORM, data, true, pColumn->offset, k);
+        }
+      }
+      int32_t rowLen = TD_ROW_LEN(rowData);
+      rowData = POINTER_SHIFT(rowData, rowLen);
+      blkHead->dataLen += rowLen;
+    }
+    int32_t dataLen = blkHead->dataLen;
+    blkHead->dataLen = htonl(dataLen);
+
+    ret->length += sizeof(SSubmitBlk) + schemaLen + dataLen;
+    blkHead = POINTER_SHIFT(blkHead, schemaLen + dataLen);
+    /*submitBlk = blkHead;*/
+  }
+
+  ret->length = htonl(ret->length);
+  taosArrayDestroy(tagArray);
+  return ret;
+}
+
 void tqTableSink(SStreamTask* pTask, void* vnode, int64_t ver, void* data) {
   const SArray* pRes = (const SArray*)data;
   SVnode*       pVnode = (SVnode*)vnode;

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

@@ -148,17 +148,24 @@ int32_t vnodeProcessWriteReq(SVnode *pVnode, SRpcMsg *pMsg, int64_t version, SRp
     case TDMT_VND_MQ_VG_CHANGE:
       if (tqProcessVgChangeReq(pVnode->pTq, POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead)),
                                pMsg->contLen - sizeof(SMsgHead)) < 0) {
-        // TODO: handle error
+        goto _err;
       }
       break;
     case TDMT_VND_MQ_VG_DELETE:
       if (tqProcessVgDeleteReq(pVnode->pTq, pMsg->pCont, pMsg->contLen) < 0) {
-        // TODO: handle error
+        goto _err;
+      }
+      break;
+    case TDMT_VND_MQ_COMMIT_OFFSET:
+      if (tqProcessOffsetCommitReq(pVnode->pTq, POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead)),
+                                   pMsg->contLen - sizeof(SMsgHead)) < 0) {
+        goto _err;
       }
       break;
     case TDMT_STREAM_TASK_DEPLOY: {
       if (tqProcessTaskDeploy(pVnode->pTq, POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead)),
                               pMsg->contLen - sizeof(SMsgHead)) < 0) {
+        goto _err;
       }
     } break;
     case TDMT_VND_ALTER_CONFIRM:
@@ -277,7 +284,7 @@ int32_t vnodeProcessWriteMsg(SVnode *pVnode, int64_t version, SRpcMsg *pMsg, SRp
 void smaHandleRes(void *pVnode, int64_t smaId, const SArray *data) {
   // TODO
 
-  // blockDebugShowData(data, __func__);
+  blockDebugShowData(data, __func__);
   tdProcessTSmaInsert(((SVnode *)pVnode)->pSma, smaId, (const char *)data);
 }
 
@@ -289,7 +296,7 @@ void vnodeUpdateMetaRsp(SVnode *pVnode, STableMetaRsp *pMetaRsp) {
 }
 
 int32_t vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
-  int32_t ret = TAOS_SYNC_PROPOSE_OTHER_ERROR;
+  int32_t ret = TAOS_SYNC_OTHER_ERROR;
 
   if (syncEnvIsStart()) {
     SSyncNode *pSyncNode = syncNodeAcquire(pVnode->sync);
@@ -368,15 +375,19 @@ int32_t vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
       ret = syncNodeOnAppendEntriesReplyCb(pSyncNode, pSyncMsg);
       syncAppendEntriesReplyDestroy(pSyncMsg);
 
+    } else if (pRpcMsg->msgType == TDMT_SYNC_SET_MNODE_STANDBY) {
+      ret = syncSetStandby(pVnode->sync);
+      SRpcMsg rsp = {.code = ret, .info = pMsg->info};
+      tmsgSendRsp(&rsp);
     } else {
       vError("==vnodeProcessSyncReq== error msg type:%d", pRpcMsg->msgType);
-      ret = TAOS_SYNC_PROPOSE_OTHER_ERROR;
+      ret = TAOS_SYNC_OTHER_ERROR;
     }
 
     syncNodeRelease(pSyncNode);
   } else {
     vError("==vnodeProcessSyncReq== error syncEnv stop");
-    ret = TAOS_SYNC_PROPOSE_OTHER_ERROR;
+    ret = TAOS_SYNC_OTHER_ERROR;
   }
 
   return ret;
@@ -901,8 +912,8 @@ static int32_t vnodeProcessAlterConfirmReq(SVnode *pVnode, int64_t version, void
 static int32_t vnodeProcessAlterHasnRangeReq(SVnode *pVnode, int64_t version, void *pReq, int32_t len, SRpcMsg *pRsp) {
   vInfo("vgId:%d, alter hashrange msg will be processed", TD_VID(pVnode));
 
-  // todo 
+  // todo
-  // 1. stop work 
+  // 1. stop work
   // 2. adjust hash range / compact / remove wals / rename vgroups
   // 3. reload sync
   return 0;

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

@@ -66,7 +66,13 @@ static int32_t vnodeProcessAlterReplicaReq(SVnode *pVnode, SRpcMsg *pMsg) {
     vInfo("vgId:%d, replica:%d %s:%u", TD_VID(pVnode), r, pNode->nodeFqdn, pNode->nodePort);
   }
 
-  return syncReconfig(pVnode->sync, &cfg);
+  SRpcMsg rpcMsg = {.info = pMsg->info};
+  if (syncReconfigBuild(pVnode->sync, &cfg, &rpcMsg) != 0) {
+    vError("vgId:%d, failed to build reconfig msg since %s", TD_VID(pVnode), terrstr());
+    return -1;
+  }
+
+  return syncPropose(pVnode->sync, &rpcMsg, false);
 }
 
 void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
@@ -241,6 +247,30 @@ static void vnodeSyncRollBackMsg(SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta
   syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg);
 }
 
+int32_t vnodeSnapshotStartRead(struct SSyncFSM *pFsm, void **ppReader) {
+  return 0;
+}
+
+int32_t vnodeSnapshotStopRead(struct SSyncFSM *pFsm, void *pReader) {
+  return 0;
+}
+
+int32_t vnodeSnapshotDoRead(struct SSyncFSM *pFsm, void *pReader, void **ppBuf, int32_t *len) {
+  return 0;
+}
+
+int32_t vnodeSnapshotStartWrite(struct SSyncFSM *pFsm, void **ppWriter) {
+  return 0;
+}
+
+int32_t vnodeSnapshotStopWrite(struct SSyncFSM *pFsm, void *pWriter, bool isApply) {
+  return 0;
+}
+
+int32_t vnodeSnapshotDoWrite(struct SSyncFSM *pFsm, void *pWriter, void *pBuf, int32_t len) {
+  return 0;
+}
+
 static SSyncFSM *vnodeSyncMakeFsm(SVnode *pVnode) {
   SSyncFSM *pFsm = taosMemoryCalloc(1, sizeof(SSyncFSM));
   pFsm->data = pVnode;
@@ -250,6 +280,14 @@ static SSyncFSM *vnodeSyncMakeFsm(SVnode *pVnode) {
   pFsm->FpGetSnapshot = vnodeSyncGetSnapshot;
   pFsm->FpRestoreFinishCb = NULL;
   pFsm->FpReConfigCb = vnodeSyncReconfig;
+
+  pFsm->FpSnapshotStartRead = vnodeSnapshotStartRead;
+  pFsm->FpSnapshotStopRead = vnodeSnapshotStopRead;
+  pFsm->FpSnapshotDoRead = vnodeSnapshotDoRead;
+  pFsm->FpSnapshotStartWrite = vnodeSnapshotStartWrite;
+  pFsm->FpSnapshotStopWrite = vnodeSnapshotStopWrite;
+  pFsm->FpSnapshotDoWrite = vnodeSnapshotDoWrite;
+
   return pFsm;
 }
 
@@ -283,11 +321,14 @@ int32_t vnodeSyncOpen(SVnode *pVnode, char *path) {
 
 void vnodeSyncStart(SVnode *pVnode) {
   syncSetMsgCb(pVnode->sync, &pVnode->msgCb);
+  syncStart(pVnode->sync);
+  /*
   if (pVnode->config.standby) {
     syncStartStandBy(pVnode->sync);
   } else {
     syncStart(pVnode->sync);
   }
+  */
 }
 
 void vnodeSyncClose(SVnode *pVnode) { syncStop(pVnode->sync); }

source/dnode/vnode/test/tsdbSmaTest.cpp

@@ -121,7 +121,7 @@ TEST(testCase, tSma_Meta_Encode_Decode_Test) {
 
   // decode
   STSmaWrapper dstTSmaWrapper = {0};
-  void        *result = tDecodeTSmaWrapper(pSW, &dstTSmaWrapper);
+  void        *result = tDecodeTSmaWrapper(pSW, &dstTSmaWrapper, false);
   EXPECT_NE(result, nullptr);
 
   EXPECT_EQ(tSmaWrapper.number, dstTSmaWrapper.number);

source/libs/catalog/inc/catalogInt.h

@@ -57,6 +57,8 @@ enum {
   CTG_OP_DROP_TB_META,
   CTG_OP_UPDATE_USER,
   CTG_OP_UPDATE_VG_EPSET,
+  CTG_OP_UPDATE_TB_INDEX,
+  CTG_OP_DROP_TB_INDEX,
   CTG_OP_MAX
 };
 
@@ -128,25 +130,33 @@ typedef struct SCtgUserCtx {
   SUserAuthInfo user;
 } SCtgUserCtx;
 
-typedef struct SCtgTbMetaCache {
+typedef STableIndexRsp STableIndex;
-  SRWLatch  stbLock;
+
-  SRWLatch  metaLock;        // RC between cache destroy and all other operations
+typedef struct SCtgTbCache {
-  SHashObj *metaCache;       //key:tbname, value:STableMeta
+  SRWLatch     metaLock;
-  SHashObj *stbCache;        //key:suid, value:STableMeta*
+  STableMeta  *pMeta;
-} SCtgTbMetaCache;
+  SRWLatch     indexLock;
+  STableIndex *pIndex;
+} SCtgTbCache;
+
+typedef struct SCtgVgCache {
+  SRWLatch         vgLock;
+  SDBVgInfo       *vgInfo;  
+} SCtgVgCache;
 
 typedef struct SCtgDBCache {
-  SRWLatch         vgLock;
+  SRWLatch         dbLock;          // RC between destroy tbCache/stbCache and all reads
   uint64_t         dbId;
   int8_t           deleted;
-  SDBVgInfo       *vgInfo;  
+  SCtgVgCache      vgCache;
-  SCtgTbMetaCache  tbCache;
+  SHashObj        *tbCache;         // key:tbname, value:SCtgTbCache
+  SHashObj        *stbCache;        // key:suid, value:STableMeta*
 } SCtgDBCache;
 
 typedef struct SCtgRentSlot {
   SRWLatch lock;
   bool     needSort;
-  SArray  *meta;  // element is SDbVgVersion or SSTableMetaVersion
+  SArray  *meta;  // element is SDbVgVersion or SSTableVersion
 } SCtgRentSlot;
 
 typedef struct SCtgRentMgmt {
@@ -245,8 +255,10 @@ typedef struct SCtgCacheStat {
   uint64_t userNum;
   uint64_t vgHitNum;
   uint64_t vgMissNum;
-  uint64_t tblHitNum;
+  uint64_t tbMetaHitNum;
-  uint64_t tblMissNum;
+  uint64_t tbMetaMissNum;
+  uint64_t tbIndexHitNum;
+  uint64_t tbIndexMissNum;  
   uint64_t userHitNum;
   uint64_t userMissNum;
 } SCtgCacheStat;
@@ -268,10 +280,10 @@ typedef struct SCtgUpdateVgMsg {
   SDBVgInfo* dbInfo;
 } SCtgUpdateVgMsg;
 
-typedef struct SCtgUpdateTblMsg {
+typedef struct SCtgUpdateTbMetaMsg {
-  SCatalog* pCtg;
+  SCatalog*         pCtg;
-  STableMetaOutput* output;
+  STableMetaOutput* pMeta;
-} SCtgUpdateTblMsg;
+} SCtgUpdateTbMetaMsg;
 
 typedef struct SCtgDropDBMsg {
   SCatalog* pCtg;
@@ -305,6 +317,17 @@ typedef struct SCtgUpdateUserMsg {
   SGetUserAuthRsp userAuth;
 } SCtgUpdateUserMsg;
 
+typedef struct SCtgUpdateTbIndexMsg {
+  SCatalog*    pCtg;
+  STableIndex* pIndex;
+} SCtgUpdateTbIndexMsg;
+
+typedef struct SCtgDropTbIndexMsg {
+  SCatalog*    pCtg;
+  char         dbFName[TSDB_DB_FNAME_LEN];
+  char         tbName[TSDB_TABLE_NAME_LEN];
+} SCtgDropTbIndexMsg;
+
 typedef struct SCtgUpdateEpsetMsg {
   SCatalog* pCtg;
   char  dbFName[TSDB_DB_FNAME_LEN];
@@ -465,12 +488,11 @@ int32_t ctgOpUpdateUser(SCtgCacheOperation *action);
 int32_t ctgOpUpdateEpset(SCtgCacheOperation *operation);
 int32_t ctgAcquireVgInfoFromCache(SCatalog* pCtg, const char *dbFName, SCtgDBCache **pCache);
 void ctgReleaseDBCache(SCatalog *pCtg, SCtgDBCache *dbCache);
-void ctgReleaseVgInfo(SCtgDBCache *dbCache);
+void ctgRUnlockVgInfo(SCtgDBCache *dbCache);
-int32_t ctgAcquireVgInfoFromCache(SCatalog* pCtg, const char *dbFName, SCtgDBCache **pCache);
 int32_t ctgTbMetaExistInCache(SCatalog* pCtg, char *dbFName, char* tbName, int32_t *exist);
 int32_t ctgReadTbMetaFromCache(SCatalog* pCtg, SCtgTbMetaCtx* ctx, STableMeta** pTableMeta);
-int32_t ctgReadTbVerFromCache(SCatalog *pCtg, const SName *pTableName, int32_t *sver, int32_t *tver, int32_t *tbType, uint64_t *suid, char *stbName);
+int32_t ctgReadTbVerFromCache(SCatalog *pCtg, SName *pTableName, int32_t *sver, int32_t *tver, int32_t *tbType, uint64_t *suid, char *stbName);
-int32_t ctgChkAuthFromCache(SCatalog* pCtg, const char* user, const char* dbFName, AUTH_TYPE type, bool *inCache, bool *pass);
+int32_t ctgChkAuthFromCache(SCatalog* pCtg, char* user, char* dbFName, AUTH_TYPE type, bool *inCache, bool *pass);
 int32_t ctgDropDbCacheEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId);
 int32_t ctgDropDbVgroupEnqueue(SCatalog* pCtg, const char *dbFName, bool syncReq);
 int32_t ctgDropStbMetaEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId, const char *stbName, uint64_t suid, bool syncReq);
@@ -479,12 +501,18 @@ int32_t ctgUpdateVgroupEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId
 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);
+int32_t ctgUpdateTbIndexEnqueue(SCatalog* pCtg, STableIndex **pIndex, bool syncOp);
 int32_t ctgMetaRentInit(SCtgRentMgmt *mgmt, uint32_t rentSec, int8_t type);
 int32_t ctgMetaRentAdd(SCtgRentMgmt *mgmt, void *meta, int64_t id, int32_t size);
 int32_t ctgMetaRentGet(SCtgRentMgmt *mgmt, void **res, uint32_t *num, int32_t size);
 int32_t ctgUpdateTbMetaToCache(SCatalog* pCtg, STableMetaOutput* pOut, bool syncReq);
 int32_t ctgStartUpdateThread();
 int32_t ctgRelaunchGetTbMetaTask(SCtgTask *pTask);
+void ctgReleaseVgInfoToCache(SCatalog* pCtg, SCtgDBCache *dbCache);
+int32_t ctgReadTbIndexFromCache(SCatalog* pCtg, SName* pTableName, SArray** pRes);
+int32_t ctgDropTbIndexEnqueue(SCatalog* pCtg, SName* pName, bool syncOp);
+int32_t ctgOpDropTbIndex(SCtgCacheOperation *operation);
+int32_t ctgOpUpdateTbIndex(SCtgCacheOperation *operation);
 
 
 
@@ -493,7 +521,7 @@ int32_t ctgGetDBVgInfoFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SBuildU
 int32_t ctgGetQnodeListFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SArray *out, SCtgTask* pTask);
 int32_t ctgGetDBCfgFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, const char *dbFName, SDbCfgInfo *out, SCtgTask* pTask);
 int32_t ctgGetIndexInfoFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, const char *indexName, SIndexInfo *out, SCtgTask* pTask);
-int32_t ctgGetTbIndexFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SName* name, SArray** out, SCtgTask* pTask);
+int32_t ctgGetTbIndexFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SName *name, STableIndex* out, SCtgTask* pTask);
 int32_t ctgGetUdfInfoFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, const char *funcName, SFuncInfo *out, SCtgTask* pTask);
 int32_t ctgGetUserDbAuthFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, const char *user, SGetUserAuthRsp *out, SCtgTask* pTask);
 int32_t ctgGetTbMetaFromMnodeImpl(SCatalog* pCtg, SRequestConnInfo *pConn, char *dbFName, char* tbName, STableMetaOutput* out, SCtgTask* pTask);
@@ -521,6 +549,8 @@ void ctgFreeSTableMetaOutput(STableMetaOutput* pOutput);
 int32_t ctgUpdateMsgCtx(SCtgMsgCtx* pCtx, int32_t reqType, void* out, char* target);
 char *ctgTaskTypeStr(CTG_TASK_TYPE type);
 int32_t ctgUpdateSendTargetInfo(SMsgSendInfo *pMsgSendInfo, int32_t msgType, SCtgTask* pTask);
+int32_t ctgCloneTableIndex(SArray* pIndex, SArray** pRes);
+void ctgFreeSTableIndex(void *info);
 
 
 extern SCatalogMgmt gCtgMgmt;

source/libs/catalog/src/catalog.c

@@ -96,8 +96,7 @@ int32_t ctgRefreshDBVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char*
   if (NULL != dbCache) {
     input.dbId = dbCache->dbId;
 
-    ctgReleaseVgInfo(dbCache);
+    ctgReleaseVgInfoToCache(pCtg, dbCache);
-    ctgReleaseDBCache(pCtg, dbCache);
   }
 
   input.vgVersion = CTG_DEFAULT_INVALID_VERSION;
@@ -349,8 +348,8 @@ int32_t ctgChkAuth(SCatalog* pCtg, SRequestConnInfo *pConn, const char* user, co
   int32_t code = 0;
 
   *pass = false;
-
+  
-  CTG_ERR_RET(ctgChkAuthFromCache(pCtg, user, dbFName, type, &inCache, pass));
+  CTG_ERR_RET(ctgChkAuthFromCache(pCtg, (char*)user, (char*)dbFName, type, &inCache, pass));
 
   if (inCache) {
     return TSDB_CODE_SUCCESS;
@@ -382,6 +381,45 @@ _return:
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t ctgGetTbIndex(SCatalog* pCtg, SRequestConnInfo *pConn, SName* pTableName, SArray** pRes) {
+  CTG_ERR_RET(ctgReadTbIndexFromCache(pCtg, pTableName, pRes));
+  if (*pRes) {
+    return TSDB_CODE_SUCCESS;
+  }
+
+  STableIndex *pIndex = taosMemoryCalloc(1, sizeof(STableIndex));
+  if (NULL == pIndex) {
+    CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
+  }
+  
+  int32_t code = ctgGetTbIndexFromMnode(pCtg, pConn, (SName*)pTableName, pIndex, NULL);
+  if (TSDB_CODE_MND_DB_INDEX_NOT_EXIST == code) {
+    code = 0;
+    goto _return;
+  }
+  CTG_ERR_JRET(code);
+  
+  SArray* pInfo = NULL;
+  CTG_ERR_JRET(ctgCloneTableIndex(pIndex->pIndex, &pInfo));
+
+  *pRes = pInfo;
+
+  CTG_ERR_JRET(ctgUpdateTbIndexEnqueue(pCtg, &pIndex, false));
+
+  return TSDB_CODE_SUCCESS;
+
+_return:
+
+  tFreeSTableIndexRsp(pIndex);
+  taosMemoryFree(pIndex);
+
+  taosArrayDestroyEx(*pRes, tFreeSTableIndexInfo);
+  *pRes = NULL;
+
+  CTG_RET(code);
+}
+
+
 int32_t ctgGetTbDistVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, SName* pTableName, SArray** pVgList) {
   STableMeta *tbMeta = NULL;
   int32_t code = 0;
@@ -404,7 +442,7 @@ int32_t ctgGetTbDistVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, SName* pTabl
   CTG_ERR_JRET(ctgGetDBVgInfo(pCtg, pConn, db, &dbCache, &vgInfo));
 
   if (dbCache) {
-    vgHash = dbCache->vgInfo->vgHash;
+    vgHash = dbCache->vgCache.vgInfo->vgHash;
   } else {
     vgHash = vgInfo->vgHash;
   }
@@ -442,7 +480,7 @@ int32_t ctgGetTbDistVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, SName* pTabl
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgRUnlockVgInfo(dbCache);
     ctgReleaseDBCache(pCtg, dbCache);
   }
 
@@ -631,12 +669,11 @@ int32_t catalogGetDBVgVersion(SCatalog* pCtg, const char* dbFName, int32_t* vers
     CTG_API_LEAVE(TSDB_CODE_SUCCESS);
   }
 
-  *version = dbCache->vgInfo->vgVersion;
+  *version = dbCache->vgCache.vgInfo->vgVersion;
   *dbId = dbCache->dbId;
-  *tableNum = dbCache->vgInfo->numOfTable;
+  *tableNum = dbCache->vgCache.vgInfo->numOfTable;
 
-  ctgReleaseVgInfo(dbCache);
+  ctgReleaseVgInfoToCache(pCtg, dbCache);
-  ctgReleaseDBCache(pCtg, dbCache);
 
   ctgDebug("Got db vgVersion from cache, dbFName:%s, vgVersion:%d", dbFName, *version);
 
@@ -661,7 +698,7 @@ int32_t catalogGetDBVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char*
   SDBVgInfo *vgInfo = NULL;
   CTG_ERR_JRET(ctgGetDBVgInfo(pCtg, pConn, dbFName, &dbCache, &vgInfo));
   if (dbCache) {
-    vgHash = dbCache->vgInfo->vgHash;
+    vgHash = dbCache->vgCache.vgInfo->vgHash;
   } else {
     vgHash = vgInfo->vgHash;
   }
@@ -674,7 +711,7 @@ int32_t catalogGetDBVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char*
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgRUnlockVgInfo(dbCache);
     ctgReleaseDBCache(pCtg, dbCache);
   }
 
@@ -741,6 +778,30 @@ _return:
   CTG_API_LEAVE(code);
 }
 
+int32_t catalogUpdateTableIndex(SCatalog* pCtg, STableIndexRsp *pRsp) {
+  CTG_API_ENTER();
+
+  int32_t code = 0;
+  
+  if (NULL == pCtg || NULL == pRsp) {
+    CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
+  }
+
+  STableIndex* pIndex = taosMemoryCalloc(1, sizeof(STableIndex));
+  if (NULL == pIndex) {
+    CTG_API_LEAVE(TSDB_CODE_OUT_OF_MEMORY);
+  }
+
+  memcpy(pIndex, pRsp, sizeof(STableIndex));
+
+  CTG_ERR_JRET(ctgUpdateTbIndexEnqueue(pCtg, &pIndex, false));
+
+_return:
+  
+  CTG_API_LEAVE(code);
+}
+
+
 int32_t catalogRemoveTableMeta(SCatalog* pCtg, SName* pTableName) {
   CTG_API_ENTER();
 
@@ -932,12 +993,12 @@ int32_t catalogGetTableHashVgroup(SCatalog *pCtg, SRequestConnInfo *pConn, const
   SDBVgInfo *vgInfo = NULL;
   CTG_ERR_JRET(ctgGetDBVgInfo(pCtg, pConn, db, &dbCache, &vgInfo));
 
-  CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, vgInfo ? vgInfo : dbCache->vgInfo, pTableName, pVgroup));
+  CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, vgInfo ? vgInfo : dbCache->vgCache.vgInfo, pTableName, pVgroup));
 
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgRUnlockVgInfo(dbCache);
     ctgReleaseDBCache(pCtg, dbCache);
   }
 
@@ -1060,14 +1121,14 @@ _return:
   CTG_API_LEAVE(TSDB_CODE_SUCCESS);
 }
 
-int32_t catalogGetExpiredSTables(SCatalog* pCtg, SSTableMetaVersion** stables, uint32_t* num) {
+int32_t catalogGetExpiredSTables(SCatalog* pCtg, SSTableVersion **stables, uint32_t *num) {
   CTG_API_ENTER();
 
   if (NULL == pCtg || NULL == stables || NULL == num) {
     CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
   }
 
-  CTG_API_LEAVE(ctgMetaRentGet(&pCtg->stbRent, (void**)stables, num, sizeof(SSTableMetaVersion)));
+  CTG_API_LEAVE(ctgMetaRentGet(&pCtg->stbRent, (void **)stables, num, sizeof(SSTableVersion)));
 }
 
 int32_t catalogGetExpiredDBs(SCatalog* pCtg, SDbVgVersion** dbs, uint32_t* num) {
@@ -1138,7 +1199,12 @@ int32_t catalogGetTableIndex(SCatalog* pCtg, SRequestConnInfo *pConn, const SNam
     CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
   }
 
-  CTG_API_LEAVE(ctgGetTbIndexFromMnode(pCtg, pConn, (SName*)pTableName, pRes, NULL));
+  int32_t code = 0;
+  CTG_ERR_JRET(ctgGetTbIndex(pCtg, pConn, (SName*)pTableName, pRes));
+  
+_return:
+
+  CTG_API_LEAVE(code);
 }
 
 int32_t catalogGetUdfInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char* funcName, SFuncInfo* pInfo) {

source/libs/catalog/src/ctgAsync.c

@@ -344,6 +344,11 @@ int32_t ctgHandleForceUpdate(SCatalog* pCtg, SCtgJob *pJob, const SCatalogReq* p
     }
   }
 
+  for (int32_t i = 0; i < pJob->tbIndexNum; ++i) {
+    SName* name = taosArrayGet(pReq->pTableIndex, i);
+    ctgDropTbIndexEnqueue(pCtg, name, true);
+  }
+
   return TSDB_CODE_SUCCESS;
 }
 
@@ -680,15 +685,14 @@ int32_t ctgHandleGetTbMetaRsp(SCtgTask* pTask, int32_t reqType, const SDataBuf *
           CTG_ERR_RET(ctgAcquireVgInfoFromCache(pCtg, dbFName, &dbCache));
           if (NULL != dbCache) {
             SVgroupInfo vgInfo = {0};
-            CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, dbCache->vgInfo, ctx->pName, &vgInfo));
+            CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, dbCache->vgCache.vgInfo, ctx->pName, &vgInfo));
           
             ctgDebug("will refresh tbmeta, not supposed to be stb, tbName:%s, flag:%d", tNameGetTableName(ctx->pName), ctx->flag);
 
             ctx->vgId = vgInfo.vgId;          
             CTG_ERR_JRET(ctgGetTbMetaFromVnode(pCtg, pConn, ctx->pName, &vgInfo, NULL, pTask));
 
-            ctgReleaseVgInfo(dbCache);
+            ctgReleaseVgInfoToCache(pCtg, dbCache);
-            ctgReleaseDBCache(pCtg, dbCache);
           } else {
             SBuildUseDBInput input = {0};
           
@@ -786,7 +790,7 @@ int32_t ctgHandleGetTbMetaRsp(SCtgTask* pTask, int32_t reqType, const SDataBuf *
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgRUnlockVgInfo(dbCache);
     ctgReleaseDBCache(pCtg, dbCache);
   }
 
@@ -866,9 +870,15 @@ _return:
 
 int32_t ctgHandleGetTbIndexRsp(SCtgTask* pTask, int32_t reqType, const SDataBuf *pMsg, int32_t rspCode) {
   int32_t code = 0;
-  CTG_ERR_JRET(ctgProcessRspMsg(pTask->msgCtx.out, reqType, pMsg->pData, pMsg->len, rspCode, pTask->msgCtx.target));
+  CTG_ERR_JRET(ctgProcessRspMsg(&pTask->msgCtx.out, reqType, pMsg->pData, pMsg->len, rspCode, pTask->msgCtx.target));
 
-  TSWAP(pTask->res, pTask->msgCtx.out);
+  STableIndex* pOut = (STableIndex*)pTask->msgCtx.out;
+  SArray* pInfo = NULL;
+  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:
   if (TSDB_CODE_MND_DB_INDEX_NOT_EXIST == code) {
@@ -1008,7 +1018,7 @@ int32_t ctgAsyncRefreshTbMeta(SCtgTask *pTask) {
   CTG_ERR_RET(ctgAcquireVgInfoFromCache(pCtg, dbFName, &dbCache));
   if (dbCache) {
     SVgroupInfo vgInfo = {0};
-    CTG_ERR_RET(ctgGetVgInfoFromHashValue(pCtg, dbCache->vgInfo, ctx->pName, &vgInfo));
+    CTG_ERR_RET(ctgGetVgInfoFromHashValue(pCtg, dbCache->vgCache.vgInfo, ctx->pName, &vgInfo));
 
     ctgDebug("will refresh tbmeta, not supposed to be stb, tbName:%s, flag:%d", tNameGetTableName(ctx->pName), ctx->flag);
 
@@ -1026,8 +1036,7 @@ int32_t ctgAsyncRefreshTbMeta(SCtgTask *pTask) {
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgReleaseVgInfoToCache(pCtg, dbCache);
-    ctgReleaseDBCache(pCtg, dbCache);
   }
 
   CTG_RET(code);
@@ -1057,7 +1066,7 @@ int32_t ctgLaunchGetDbVgTask(SCtgTask *pTask) {
   
   CTG_ERR_RET(ctgAcquireVgInfoFromCache(pCtg, pCtx->dbFName, &dbCache));
   if (NULL != dbCache) {
-    CTG_ERR_JRET(ctgGenerateVgList(pCtg, dbCache->vgInfo->vgHash, (SArray**)&pTask->res));
+    CTG_ERR_JRET(ctgGenerateVgList(pCtg, dbCache->vgCache.vgInfo->vgHash, (SArray**)&pTask->res));
     
     CTG_ERR_JRET(ctgHandleTaskEnd(pTask, 0));
   } else {
@@ -1072,8 +1081,7 @@ int32_t ctgLaunchGetDbVgTask(SCtgTask *pTask) {
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgReleaseVgInfoToCache(pCtg, dbCache);
-    ctgReleaseDBCache(pCtg, dbCache);
   }
 
   CTG_RET(code);
@@ -1092,7 +1100,7 @@ int32_t ctgLaunchGetTbHashTask(SCtgTask *pTask) {
     if (NULL == pTask->res) {
       CTG_ERR_JRET(TSDB_CODE_OUT_OF_MEMORY);
     }
-    CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, dbCache->vgInfo, pCtx->pName, (SVgroupInfo*)pTask->res));
+    CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pCtg, dbCache->vgCache.vgInfo, pCtx->pName, (SVgroupInfo*)pTask->res));
     
     CTG_ERR_JRET(ctgHandleTaskEnd(pTask, 0));
   } else {
@@ -1107,8 +1115,7 @@ int32_t ctgLaunchGetTbHashTask(SCtgTask *pTask) {
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgReleaseVgInfoToCache(pCtg, dbCache);
-    ctgReleaseDBCache(pCtg, dbCache);
   }
 
   CTG_RET(code);
@@ -1119,6 +1126,15 @@ int32_t ctgLaunchGetTbIndexTask(SCtgTask *pTask) {
   SCatalog* pCtg = pTask->pJob->pCtg; 
   SRequestConnInfo* pConn = &pTask->pJob->conn;
   SCtgTbIndexCtx* pCtx = (SCtgTbIndexCtx*)pTask->taskCtx;
+  SArray* pRes = NULL;
+
+  CTG_ERR_RET(ctgReadTbIndexFromCache(pCtg, pCtx->pName, &pRes));
+  if (pRes) {
+    pTask->res = pRes;
+    
+    CTG_ERR_RET(ctgHandleTaskEnd(pTask, 0));
+    return TSDB_CODE_SUCCESS;
+  }
   
   CTG_ERR_RET(ctgGetTbIndexFromMnode(pCtg, pConn, pCtx->pName, NULL, pTask));
   return TSDB_CODE_SUCCESS;
@@ -1157,9 +1173,9 @@ int32_t ctgLaunchGetDbInfoTask(SCtgTask *pTask) {
   SDbInfo* pInfo = (SDbInfo*)pTask->res;
   CTG_ERR_RET(ctgAcquireVgInfoFromCache(pCtg, pCtx->dbFName, &dbCache));
   if (NULL != dbCache) {
-    pInfo->vgVer = dbCache->vgInfo->vgVersion;
+    pInfo->vgVer = dbCache->vgCache.vgInfo->vgVersion;
     pInfo->dbId = dbCache->dbId;
-    pInfo->tbNum = dbCache->vgInfo->numOfTable;
+    pInfo->tbNum = dbCache->vgCache.vgInfo->numOfTable;
   } else {
     pInfo->vgVer = CTG_DEFAULT_INVALID_VERSION;
   }
@@ -1169,8 +1185,7 @@ int32_t ctgLaunchGetDbInfoTask(SCtgTask *pTask) {
 _return:
 
   if (dbCache) {
-    ctgReleaseVgInfo(dbCache);
+    ctgReleaseVgInfoToCache(pCtg, dbCache);
-    ctgReleaseDBCache(pCtg, dbCache);
   }
 
   CTG_RET(code);

source/libs/catalog/src/ctgDbg.c

@@ -19,7 +19,7 @@
 #include "catalogInt.h"
 
 extern SCatalogMgmt gCtgMgmt;
-SCtgDebug gCTGDebug = {.apiEnable = true};
+SCtgDebug gCTGDebug = {0};
 
 void ctgdUserCallback(SMetaData* pResult, void* param, int32_t code) {
   ASSERT(*(int32_t*)param == 1);
@@ -266,11 +266,11 @@ int32_t ctgdGetStatNum(char *option, void *res) {
 }
 
 int32_t ctgdGetTbMetaNum(SCtgDBCache *dbCache) {
-  return dbCache->tbCache.metaCache ? (int32_t)taosHashGetSize(dbCache->tbCache.metaCache) : 0;
+  return dbCache->tbCache ? (int32_t)taosHashGetSize(dbCache->tbCache) : 0;
 }
 
 int32_t ctgdGetStbNum(SCtgDBCache *dbCache) {
-  return dbCache->tbCache.stbCache ? (int32_t)taosHashGetSize(dbCache->tbCache.stbCache) : 0;
+  return dbCache->stbCache ? (int32_t)taosHashGetSize(dbCache->stbCache) : 0;
 }
 
 int32_t ctgdGetRentNum(SCtgRentMgmt *rent) {
@@ -363,17 +363,17 @@ void ctgdShowDBCache(SCatalog* pCtg, SHashObj *dbHash) {
 
     dbFName = taosHashGetKey(pIter, &len);
 
-    int32_t metaNum = dbCache->tbCache.metaCache ? taosHashGetSize(dbCache->tbCache.metaCache) : 0;
+    int32_t metaNum = dbCache->tbCache ? taosHashGetSize(dbCache->tbCache) : 0;
-    int32_t stbNum = dbCache->tbCache.stbCache ? taosHashGetSize(dbCache->tbCache.stbCache) : 0;
+    int32_t stbNum = dbCache->stbCache ? taosHashGetSize(dbCache->stbCache) : 0;
     int32_t vgVersion = CTG_DEFAULT_INVALID_VERSION;
     int32_t hashMethod = -1;
     int32_t vgNum = 0;
 
-    if (dbCache->vgInfo) {
+    if (dbCache->vgCache.vgInfo) {
-      vgVersion = dbCache->vgInfo->vgVersion;
+      vgVersion = dbCache->vgCache.vgInfo->vgVersion;
-      hashMethod = dbCache->vgInfo->hashMethod;
+      hashMethod = dbCache->vgCache.vgInfo->hashMethod;
-      if (dbCache->vgInfo->vgHash) {
+      if (dbCache->vgCache.vgInfo->vgHash) {
-        vgNum = taosHashGetSize(dbCache->vgInfo->vgHash);
+        vgNum = taosHashGetSize(dbCache->vgCache.vgInfo->vgHash);
       }
     }
     

source/libs/catalog/src/ctgRemote.c

@@ -431,7 +431,7 @@ int32_t ctgGetIndexInfoFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, const
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t ctgGetTbIndexFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SName *name, SArray** out, SCtgTask* pTask) {
+int32_t ctgGetTbIndexFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SName *name, STableIndex* out, SCtgTask* pTask) {
   char *msg = NULL;
   int32_t msgLen = 0;
   int32_t reqType = TDMT_MND_GET_TABLE_INDEX;
@@ -448,10 +448,11 @@ int32_t ctgGetTbIndexFromMnode(SCatalog* pCtg, SRequestConnInfo *pConn, SName *n
   }
 
   if (pTask) {
-    void* pOut = taosMemoryCalloc(1, POINTER_BYTES);
+    void* pOut = taosMemoryCalloc(1, sizeof(STableIndex));
     if (NULL == pOut) {
       CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
     }
+    
     CTG_ERR_RET(ctgUpdateMsgCtx(&pTask->msgCtx, reqType, pOut, (char*)tbFName));
     
     CTG_RET(ctgAsyncSendMsg(pCtg, pConn, pTask, reqType, msg, msgLen));

source/libs/catalog/src/ctgUtil.c

@@ -44,6 +44,16 @@ char *ctgTaskTypeStr(CTG_TASK_TYPE type) {
   }
 }
 
+void ctgFreeSTableIndex(void *info) {
+  if (NULL == info) {
+    return;
+  }
+
+  STableIndex *pInfo = (STableIndex *)info;
+
+  taosArrayDestroyEx(pInfo->pIndex, tFreeSTableIndexInfo);
+}
+
 void ctgFreeSMetaData(SMetaData* pData) {
   taosArrayDestroy(pData->pTableMeta);
   pData->pTableMeta = NULL;
@@ -110,25 +120,39 @@ void ctgFreeMetaRent(SCtgRentMgmt *mgmt) {
   taosMemoryFreeClear(mgmt->slots);
 }
 
-
+void ctgFreeStbMetaCache(SCtgDBCache *dbCache) {
-void ctgFreeTbMetaCache(SCtgTbMetaCache *cache) {
+  if (NULL == dbCache->stbCache) {
-  CTG_LOCK(CTG_WRITE, &cache->stbLock);
+    return;
-  if (cache->stbCache) {
-    int32_t stblNum = taosHashGetSize(cache->stbCache);  
-    taosHashCleanup(cache->stbCache);
-    cache->stbCache = NULL;
-    CTG_CACHE_STAT_DEC(stblNum, stblNum);
   }
-  CTG_UNLOCK(CTG_WRITE, &cache->stbLock);
 
-  CTG_LOCK(CTG_WRITE, &cache->metaLock);
+  int32_t stblNum = taosHashGetSize(dbCache->stbCache);  
-  if (cache->metaCache) {
+  taosHashCleanup(dbCache->stbCache);
-    int32_t tblNum = taosHashGetSize(cache->metaCache);
+  dbCache->stbCache = NULL;
-    taosHashCleanup(cache->metaCache);
+  CTG_CACHE_STAT_DEC(stblNum, stblNum);
-    cache->metaCache = NULL;
+}
-    CTG_CACHE_STAT_DEC(tblNum, tblNum);
+
+void ctgFreeTbCacheImpl(SCtgTbCache *pCache) {
+  taosMemoryFreeClear(pCache->pMeta);
+  if (pCache->pIndex) {
+    taosArrayDestroyEx(pCache->pIndex->pIndex, tFreeSTableIndexInfo);
+    taosMemoryFreeClear(pCache->pIndex);
   }
-  CTG_UNLOCK(CTG_WRITE, &cache->metaLock);
+}
+
+void ctgFreeTbCache(SCtgDBCache *dbCache) {
+  if (NULL == dbCache->tbCache) {
+    return;
+  }
+
+  int32_t tblNum = taosHashGetSize(dbCache->tbCache);
+  SCtgTbCache *pCache = taosHashIterate(dbCache->tbCache, NULL);
+  while (NULL != pCache) {
+    ctgFreeTbCacheImpl(pCache);
+    pCache = taosHashIterate(dbCache->tbCache, pCache);
+  }
+  taosHashCleanup(dbCache->tbCache);
+  dbCache->tbCache = NULL;
+  CTG_CACHE_STAT_DEC(tblNum, tblNum);
 }
 
 void ctgFreeVgInfo(SDBVgInfo *vgInfo) {
@@ -144,16 +168,18 @@ void ctgFreeVgInfo(SDBVgInfo *vgInfo) {
   taosMemoryFreeClear(vgInfo);
 }
 
+void ctgFreeVgInfoCache(SCtgDBCache *dbCache) {
+  ctgFreeVgInfo(dbCache->vgCache.vgInfo);
+}
+
 void ctgFreeDbCache(SCtgDBCache *dbCache) {
   if (NULL == dbCache) {
     return;
   }
 
-  CTG_LOCK(CTG_WRITE, &dbCache->vgLock);
+  ctgFreeVgInfoCache(dbCache);
-  ctgFreeVgInfo (dbCache->vgInfo);
+  ctgFreeStbMetaCache(dbCache);
-  CTG_UNLOCK(CTG_WRITE, &dbCache->vgLock);
+  ctgFreeTbCache(dbCache);
-
-  ctgFreeTbMetaCache(&dbCache->tbCache);
 }
 
 
@@ -167,16 +193,13 @@ void ctgFreeHandle(SCatalog* pCtg) {
     void *pIter = taosHashIterate(pCtg->dbCache, NULL);
     while (pIter) {
       SCtgDBCache *dbCache = pIter;
-
       atomic_store_8(&dbCache->deleted, 1);
-
       ctgFreeDbCache(dbCache);
             
       pIter = taosHashIterate(pCtg->dbCache, pIter);
     }  
 
     taosHashCleanup(pCtg->dbCache);
-    
     CTG_CACHE_STAT_DEC(dbNum, dbNum);
   }
 
@@ -186,14 +209,12 @@ void ctgFreeHandle(SCatalog* pCtg) {
     void *pIter = taosHashIterate(pCtg->userCache, NULL);
     while (pIter) {
       SCtgUserAuth *userCache = pIter;
-
       ctgFreeSCtgUserAuth(userCache);
 
       pIter = taosHashIterate(pCtg->userCache, pIter);
     }  
 
     taosHashCleanup(pCtg->userCache);
-
     CTG_CACHE_STAT_DEC(userNum, userNum);
   }
 
@@ -252,9 +273,9 @@ void ctgFreeMsgCtx(SCtgMsgCtx* pCtx) {
       break;
     }
     case TDMT_MND_GET_TABLE_INDEX: {
-      SArray** pOut = (SArray**)pCtx->out;
+      STableIndex* pOut = (STableIndex*)pCtx->out;
       if (pOut) {
-        taosArrayDestroyEx(*pOut, tFreeSTableIndexInfo);
+        taosArrayDestroyEx(pOut->pIndex, tFreeSTableIndexInfo);
         taosMemoryFreeClear(pCtx->out);
       }
       break;
@@ -535,9 +556,9 @@ int32_t ctgGetVgInfoFromHashValue(SCatalog *pCtg, SDBVgInfo *dbInfo, const SName
 }
 
 int32_t ctgStbVersionSearchCompare(const void* key1, const void* key2) {
-  if (*(uint64_t *)key1 < ((SSTableMetaVersion*)key2)->suid) {
+  if (*(uint64_t *)key1 < ((SSTableVersion*)key2)->suid) {
     return -1;
-  } else if (*(uint64_t *)key1 > ((SSTableMetaVersion*)key2)->suid) {
+  } else if (*(uint64_t *)key1 > ((SSTableVersion*)key2)->suid) {
     return 1;
   } else {
     return 0;
@@ -555,9 +576,9 @@ int32_t ctgDbVgVersionSearchCompare(const void* key1, const void* key2) {
 }
 
 int32_t ctgStbVersionSortCompare(const void* key1, const void* key2) {
-  if (((SSTableMetaVersion*)key1)->suid < ((SSTableMetaVersion*)key2)->suid) {
+  if (((SSTableVersion*)key1)->suid < ((SSTableVersion*)key2)->suid) {
     return -1;
-  } else if (((SSTableMetaVersion*)key1)->suid > ((SSTableMetaVersion*)key2)->suid) {
+  } else if (((SSTableVersion*)key1)->suid > ((SSTableVersion*)key2)->suid) {
     return 1;
   } else {
     return 0;
@@ -640,6 +661,27 @@ int32_t ctgCloneMetaOutput(STableMetaOutput *output, STableMetaOutput **pOutput)
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t ctgCloneTableIndex(SArray* pIndex, SArray** pRes) {
+  if (NULL == pIndex) {
+    *pRes = NULL;
+    return TSDB_CODE_SUCCESS;
+  }
+
+  int32_t num = taosArrayGetSize(pIndex);
+  *pRes = taosArrayInit(num, sizeof(STableIndexInfo));
+  if (NULL == *pRes) {
+    CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
+  }
+
+  for (int32_t i = 0; i < num; ++i) {
+    STableIndexInfo *pInfo = taosArrayGet(pIndex, i);
+    taosArrayPush(*pRes, pInfo);
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+
 int32_t ctgUpdateSendTargetInfo(SMsgSendInfo *pMsgSendInfo, int32_t msgType, SCtgTask* pTask) {
   if (msgType == TDMT_VND_TABLE_META) {
     SCtgTbMetaCtx* ctx = (SCtgTbMetaCtx*)pTask->taskCtx;

source/libs/catalog/test/catalogTests.cpp

@@ -895,9 +895,9 @@ void *ctgTestSetCtableMetaThread(void *param) {
     output = (STableMetaOutput *)taosMemoryMalloc(sizeof(STableMetaOutput));
     ctgTestBuildCTableMetaOutput(output);
 
-    SCtgUpdateTblMsg *msg = (SCtgUpdateTblMsg *)taosMemoryMalloc(sizeof(SCtgUpdateTblMsg));
+    SCtgUpdateTbMetaMsg *msg = (SCtgUpdateTbMetaMsg *)taosMemoryMalloc(sizeof(SCtgUpdateTbMetaMsg));
     msg->pCtg = pCtg;
-    msg->output = output;
+    msg->pMeta = output;
     operation.data = msg;
 
     code = ctgOpUpdateTbMeta(&operation);
@@ -989,7 +989,7 @@ TEST(tableMeta, normalTable) {
   ASSERT_EQ(tableMeta->tableInfo.rowSize, 12);
 
   SDbVgVersion       *dbs = NULL;
-  SSTableMetaVersion *stb = NULL;
+  SSTableVersion *stb = NULL;
   uint32_t            dbNum = 0, stbNum = 0, allDbNum = 0, allStbNum = 0;
   int32_t             i = 0;
   while (i < 5) {
@@ -1098,7 +1098,7 @@ TEST(tableMeta, childTableCase) {
   ASSERT_EQ(tableMeta->tableInfo.rowSize, 12);
 
   SDbVgVersion       *dbs = NULL;
-  SSTableMetaVersion *stb = NULL;
+  SSTableVersion *stb = NULL;
   uint32_t            dbNum = 0, stbNum = 0, allDbNum = 0, allStbNum = 0;
   int32_t             i = 0;
   while (i < 5) {
@@ -1220,7 +1220,7 @@ TEST(tableMeta, superTableCase) {
   ASSERT_EQ(tableMeta->tableInfo.rowSize, 12);
 
   SDbVgVersion       *dbs = NULL;
-  SSTableMetaVersion *stb = NULL;
+  SSTableVersion *stb = NULL;
   uint32_t            dbNum = 0, stbNum = 0, allDbNum = 0, allStbNum = 0;
   int32_t             i = 0;
   while (i < 5) {
@@ -2299,7 +2299,7 @@ TEST(rentTest, allRent) {
   SArray          *vgList = NULL;
   ctgTestStop = false;
   SDbVgVersion       *dbs = NULL;
-  SSTableMetaVersion *stable = NULL;
+  SSTableVersion *stable = NULL;
   uint32_t            num = 0;
 
   ctgTestInitLogFile();

source/libs/executor/inc/executorInt.h

@@ -27,10 +27,7 @@ typedef struct {
   int32_t         bytes;
 } SGroupKeys, SStateKeys;
 
-int32_t initGroupOptrInfo(SArray** pGroupColVals, int32_t* keyLen, char** keyBuf, const SArray* pGroupColList);
 uint64_t calcGroupId(char* pData, int32_t len);
-void recordNewGroupKeys(SArray* pGroupCols, SArray* pGroupColVals, SSDataBlock* pBlock, int32_t rowIndex);
-int32_t buildGroupKeys(void* pKey, const SArray* pGroupColVals);
 #ifdef __cplusplus
 }
 #endif

source/libs/executor/inc/executorimpl.h

@@ -336,12 +336,6 @@ typedef struct STableScanInfo {
   int32_t         dataBlockLoadFlag;
   SInterval       interval;     // if the upstream is an interval operator, the interval info is also kept here to get the time window to check if current data block needs to be loaded.
 
-  SArray*        pGroupCols;
-  SArray*        pGroupColVals;  // current group column values, SArray<SGroupKeys>
-  char*          keyBuf;         // group by keys for hash
-  int32_t        groupKeyLen;    // total group by column width
-  SHashObj*      pGroupSet;      // quick locate the window object for each result
-
   SSampleExecInfo sample;       // sample execution info
   int32_t         curTWinIdx;
 } STableScanInfo;
@@ -789,7 +783,7 @@ SResultRow* doSetResultOutBufByKey(SDiskbasedBuf* pResultBuf, SResultRowInfo* pR
 
 SOperatorInfo* createExchangeOperatorInfo(void* pTransporter, SExchangePhysiNode* pExNode, SExecTaskInfo* pTaskInfo);
 
-SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode, tsdbReaderT pDataReader, SReadHandle* pHandle, SArray* groupKyes, SExecTaskInfo* pTaskInfo);
+SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode, tsdbReaderT pDataReader, SReadHandle* pHandle, SExecTaskInfo* pTaskInfo);
 SOperatorInfo* createTagScanOperatorInfo(SReadHandle* pReadHandle, STagScanPhysiNode* pPhyNode,
                                          STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
 SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSystemTableScanPhysiNode *pScanPhyNode, SExecTaskInfo* pTaskInfo);
@@ -827,7 +821,7 @@ SOperatorInfo* createGroupOperatorInfo(SOperatorInfo* downstream, SExprInfo* pEx
                                        SExprInfo* pScalarExprInfo, int32_t numOfScalarExpr, SExecTaskInfo* pTaskInfo);
 SOperatorInfo* createDataBlockInfoScanOperator(void* dataReader, SExecTaskInfo* pTaskInfo);
 
-SOperatorInfo* createStreamScanOperatorInfo(void* pDataReader, SReadHandle* pHandle, SArray* pTableIdList,
+SOperatorInfo* createStreamScanOperatorInfo(void* pDataReader, SReadHandle* pHandle,
     STableScanPhysiNode* pTableScanNode, SExecTaskInfo* pTaskInfo, STimeWindowAggSupp* pTwSup);
 
 

source/libs/executor/src/executorimpl.c

@@ -4558,7 +4558,6 @@ static tsdbReaderT doCreateDataReader(STableScanPhysiNode* pTableScanNode, SRead
 
 static int32_t getTableList(void* metaHandle, int32_t tableType, uint64_t tableUid, STableListInfo* pListInfo,
                             SNode* pTagCond);
-static SArray* extractTableIdList(const STableListInfo* pTableGroupInfo);
 static SArray* extractColumnInfo(SNodeList* pNodeList);
 
 static SArray* createSortInfo(SNodeList* pNodeList);
@@ -4592,6 +4591,85 @@ int32_t extractTableSchemaVersion(SReadHandle* pHandle, uint64_t uid, SExecTaskI
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle, SArray* groupKey){
+  if(groupKey == NULL) {
+    return TDB_CODE_SUCCESS;
+  }
+
+  pTableListInfo->map = taosHashInit(32, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
+  if (pTableListInfo->map == NULL) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+  int32_t keyLen = 0;
+  void *keyBuf = NULL;
+  int32_t numOfGroupCols = taosArrayGetSize(groupKey);
+  for (int32_t j = 0; j < numOfGroupCols; ++j) {
+    SColumn* pCol = taosArrayGet(groupKey, j);
+    keyLen += pCol->bytes; // actual data + null_flag
+  }
+
+  int32_t nullFlagSize = sizeof(int8_t) * numOfGroupCols;
+  keyLen += nullFlagSize;
+
+  keyBuf = taosMemoryCalloc(1, keyLen);
+  if (keyBuf == NULL) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  for(int32_t i = 0; i < taosArrayGetSize(pTableListInfo->pTableList); i++){
+    STableKeyInfo *info = taosArrayGet(pTableListInfo->pTableList, i);
+    SMetaReader mr = {0};
+    metaReaderInit(&mr, pHandle->meta, 0);
+    metaGetTableEntryByUid(&mr, info->uid);
+
+    char* isNull = (char*)keyBuf;
+    char* pStart = (char*)keyBuf + sizeof(int8_t) * numOfGroupCols;
+    for (int32_t j = 0; j < numOfGroupCols; ++j) {
+      SColumn* pCol = taosArrayGet(groupKey, j);
+
+      if(strcmp(pCol->name, "tbname") == 0){
+        isNull[i] = 0;
+        memcpy(pStart, mr.me.name, strlen(mr.me.name));
+        pStart += strlen(mr.me.name);
+      }else{
+        STagVal tagVal = {0};
+        tagVal.cid = pCol->colId;
+        const char* p = metaGetTableTagVal(&mr.me, pCol->type, &tagVal);
+        if(p == NULL){
+          isNull[j] = 1;
+          continue;
+        }
+        isNull[i] = 0;
+        if (pCol->type == TSDB_DATA_TYPE_JSON) {
+//          int32_t dataLen = getJsonValueLen(pkey->pData);
+//          memcpy(pStart, (pkey->pData), dataLen);
+//          pStart += dataLen;
+        } else if (IS_VAR_DATA_TYPE(pCol->type)) {
+          memcpy(pStart, tagVal.pData, tagVal.nData);
+          pStart += tagVal.nData;
+          ASSERT(tagVal.nData <= pCol->bytes);
+        } else {
+          memcpy(pStart, &(tagVal.i64), pCol->bytes);
+          pStart += pCol->bytes;
+        }
+      }
+    }
+
+    int32_t len = (int32_t) (pStart - (char*)keyBuf);
+    uint64_t* groupId = taosHashGet(pTableListInfo->map, keyBuf, len);
+    if (groupId) {
+      taosHashPut(pTableListInfo->map, &(info->uid), sizeof(uint64_t), groupId, sizeof(uint64_t));
+    } else {
+      uint64_t tmpId = calcGroupId(keyBuf, len);
+      taosHashPut(pTableListInfo->map, &(info->uid), sizeof(uint64_t), &tmpId, sizeof(uint64_t));
+    }
+
+    metaReaderClear(&mr);
+  }
+  taosMemoryFree(keyBuf);
+  return TDB_CODE_SUCCESS;
+}
+
 SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, SReadHandle* pHandle,
                                   uint64_t queryId, uint64_t taskId, STableListInfo* pTableListInfo, SNode* pTagCond) {
   int32_t type = nodeType(pPhyNode);
@@ -4605,15 +4683,23 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
       if (pDataReader == NULL && terrno != 0) {
         return NULL;
       }
-      SArray* groupKyes = extractPartitionColInfo(pTableScanNode->pPartitionKeys);
+
       int32_t code = extractTableSchemaVersion(pHandle, pTableScanNode->scan.uid, pTaskInfo);
       if (code) {
         tsdbCleanupReadHandle(pDataReader);
         return NULL;
       }
-      
+
+      SArray* groupKeys = extractPartitionColInfo(pTableScanNode->pPartitionKeys);
+      code = generateGroupIdMap(pTableListInfo, pHandle, groupKeys); //todo for json
+      taosArrayDestroy(groupKeys);
+      if (code){
+        tsdbCleanupReadHandle(pDataReader);
+        return NULL;
+      }
+
       SOperatorInfo* pOperator =
-          createTableScanOperatorInfo(pTableScanNode, pDataReader, pHandle, groupKyes, pTaskInfo);
+          createTableScanOperatorInfo(pTableScanNode, pDataReader, pHandle, pTaskInfo);
 
       STableScanInfo* pScanInfo = pOperator->info;
       pTaskInfo->cost.pRecoder = &pScanInfo->readRecorder;
@@ -4639,12 +4725,18 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
       } else {
         qDebug("%s pDataReader is not NULL", GET_TASKID(pTaskInfo));
       }
-      SArray* tableIdList = extractTableIdList(pTableListInfo);
+
+      SArray* groupKeys = extractPartitionColInfo(pTableScanNode->pPartitionKeys);
+      int32_t code = generateGroupIdMap(pTableListInfo, pHandle, groupKeys); //todo for json
+      taosArrayDestroy(groupKeys);
+      if (code){
+        tsdbCleanupReadHandle(pDataReader);
+        return NULL;
+      }
 
       SOperatorInfo* pOperator =
-          createStreamScanOperatorInfo(pDataReader, pHandle, tableIdList, pTableScanNode, pTaskInfo, &twSup);
+          createStreamScanOperatorInfo(pDataReader, pHandle, pTableScanNode, pTaskInfo, &twSup);
 
-      taosArrayDestroy(tableIdList);
       return pOperator;
     } else if (QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN == type) {
       SSystemTableScanPhysiNode* pSysScanPhyNode = (SSystemTableScanPhysiNode*)pPhyNode;
@@ -4969,6 +5061,7 @@ SArray* extractColumnInfo(SNodeList* pNodeList) {
 }
 
 SArray* extractPartitionColInfo(SNodeList* pNodeList) {
+  if(!pNodeList) return NULL;
   size_t  numOfCols = LIST_LENGTH(pNodeList);
   SArray* pList = taosArrayInit(numOfCols, sizeof(SColumn));
   if (pList == NULL) {
@@ -5073,7 +5166,9 @@ int32_t getTableList(void* metaHandle, int32_t tableType, uint64_t tableUid, STa
 
       SArray* res = taosArrayInit(8, sizeof(uint64_t));
       code = doFilterTag(pTagCond, &metaArg, res);
-      if (code != TSDB_CODE_SUCCESS) {
+      if (code == TSDB_CODE_INDEX_REBUILDING){    // todo
+        // doFilter();
+      } else if (code != TSDB_CODE_SUCCESS) {
         qError("failed  to  get tableIds, reason: %s, suid: %" PRIu64 "", tstrerror(code), tableUid);
         taosArrayDestroy(res);
         terrno = code;
@@ -5081,6 +5176,7 @@ int32_t getTableList(void* metaHandle, int32_t tableType, uint64_t tableUid, STa
       } else {
         qDebug("sucess to  get tableIds, size: %d, suid: %" PRIu64 "", (int)taosArrayGetSize(res), tableUid);
       }
+
       for (int i = 0; i < taosArrayGetSize(res); i++) {
         STableKeyInfo info = {.lastKey = TSKEY_INITIAL_VAL, .uid = *(uint64_t*)taosArrayGet(res, i)};
         taosArrayPush(pListInfo->pTableList, &info);
@@ -5097,18 +5193,6 @@ int32_t getTableList(void* metaHandle, int32_t tableType, uint64_t tableUid, STa
   return code;
 }
 
-SArray* extractTableIdList(const STableListInfo* pTableGroupInfo) {
-  SArray* tableIdList = taosArrayInit(4, sizeof(uint64_t));
-
-  // Transfer the Array of STableKeyInfo into uid list.
-  for (int32_t i = 0; i < taosArrayGetSize(pTableGroupInfo->pTableList); ++i) {
-    STableKeyInfo* pkeyInfo = taosArrayGet(pTableGroupInfo->pTableList, i);
-    taosArrayPush(tableIdList, &pkeyInfo->uid);
-  }
-
-  return tableIdList;
-}
-
 tsdbReaderT doCreateDataReader(STableScanPhysiNode* pTableScanNode, SReadHandle* pHandle,
                                STableListInfo* pTableListInfo, uint64_t queryId, uint64_t taskId, SNode* pTagCond) {
   int32_t code =

source/libs/executor/src/groupoperator.c

@@ -37,7 +37,7 @@ static void destroyGroupOperatorInfo(void* param, int32_t numOfOutput) {
   taosArrayDestroy(pInfo->pGroupColVals);
 }
 
-int32_t initGroupOptrInfo(SArray** pGroupColVals, int32_t* keyLen, char** keyBuf, const SArray* pGroupColList) {
+static int32_t initGroupOptrInfo(SArray** pGroupColVals, int32_t* keyLen, char** keyBuf, const SArray* pGroupColList) {
   *pGroupColVals = taosArrayInit(4, sizeof(SGroupKeys));
   if ((*pGroupColVals) == NULL) {
     return TSDB_CODE_OUT_OF_MEMORY;
@@ -118,7 +118,7 @@ static bool groupKeyCompare(SArray* pGroupCols, SArray* pGroupColVals, SSDataBlo
   return true;
 }
 
-void recordNewGroupKeys(SArray* pGroupCols, SArray* pGroupColVals, SSDataBlock* pBlock, int32_t rowIndex) {
+static void recordNewGroupKeys(SArray* pGroupCols, SArray* pGroupColVals, SSDataBlock* pBlock, int32_t rowIndex) {
   SColumnDataAgg* pColAgg = NULL;
 
   size_t numOfGroupCols = taosArrayGetSize(pGroupCols);
@@ -150,7 +150,7 @@ void recordNewGroupKeys(SArray* pGroupCols, SArray* pGroupColVals, SSDataBlock*
   }
 }
 
-int32_t buildGroupKeys(void* pKey, const SArray* pGroupColVals) {
+static int32_t buildGroupKeys(void* pKey, const SArray* pGroupColVals) {
   ASSERT(pKey != NULL);
   size_t numOfGroupCols = taosArrayGetSize(pGroupColVals);
 

source/libs/executor/src/scanoperator.c

@@ -391,22 +391,16 @@ static SSDataBlock* doTableScanImpl(SOperatorInfo* pOperator) {
       longjmp(pOperator->pTaskInfo->env, code);
     }
 
-    recordNewGroupKeys(pTableScanInfo->pGroupCols, pTableScanInfo->pGroupColVals, pBlock, 0);
-    int32_t len = buildGroupKeys(pTableScanInfo->keyBuf, pTableScanInfo->pGroupColVals);
-
-    uint64_t* groupId = taosHashGet(pTableScanInfo->pGroupSet, pTableScanInfo->keyBuf, len);
-    if (groupId) {
-      pBlock->info.groupId = *groupId;
-    } else if (len != 0) {
-      pBlock->info.groupId = calcGroupId(pTableScanInfo->keyBuf, len);
-      taosHashPut(pTableScanInfo->pGroupSet, pTableScanInfo->keyBuf, len, &pBlock->info.groupId, sizeof(uint64_t));
-    }
-
     // 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;
     }
 
+    uint64_t* groupId = taosHashGet(pOperator->pTaskInfo->tableqinfoList.map, &pBlock->info.uid, sizeof(int64_t));
+    if (groupId) {
+      pBlock->info.groupId = *groupId;
+    }
+
     pOperator->resultInfo.totalRows = pTableScanInfo->readRecorder.totalRows;
     pTableScanInfo->readRecorder.elapsedTime += (taosGetTimestampUs() - st) / 1000.0;
 
@@ -530,21 +524,13 @@ static void destroyTableScanOperatorInfo(void* param, int32_t numOfOutput) {
 
   tsdbCleanupReadHandle(pTableScanInfo->dataReader);
 
-  taosArrayDestroy(pTableScanInfo->pGroupCols);
-  for (int i = 0; i < taosArrayGetSize(pTableScanInfo->pGroupColVals); i++) {
-    SGroupKeys key = *(SGroupKeys*)taosArrayGet(pTableScanInfo->pGroupColVals, i);
-    taosMemoryFree(key.pData);
-  }
-  taosArrayDestroy(pTableScanInfo->pGroupColVals);
-  taosMemoryFree(pTableScanInfo->keyBuf);
-  taosHashCleanup(pTableScanInfo->pGroupSet);
   if (pTableScanInfo->pColMatchInfo != NULL) {
     taosArrayDestroy(pTableScanInfo->pColMatchInfo);
   }
 }
 
 SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode, tsdbReaderT pDataReader,
-                                           SReadHandle* readHandle, SArray* groupKyes, SExecTaskInfo* pTaskInfo) {
+                                           SReadHandle* readHandle, SExecTaskInfo* pTaskInfo) {
   STableScanInfo* pInfo = taosMemoryCalloc(1, sizeof(STableScanInfo));
   SOperatorInfo*  pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
   if (pInfo == NULL || pOperator == NULL) {
@@ -591,18 +577,6 @@ SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode,
   pOperator->numOfExprs = numOfCols;
   pOperator->pTaskInfo = pTaskInfo;
 
-  // for table group
-  pInfo->pGroupCols = groupKyes;
-  _hash_fn_t hashFn = taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY);
-  pInfo->pGroupSet = taosHashInit(100, hashFn, false, HASH_NO_LOCK);
-  if (pInfo->pGroupSet == NULL) {
-    goto _error;
-  }
-  code = initGroupOptrInfo(&pInfo->pGroupColVals, &pInfo->groupKeyLen, &pInfo->keyBuf, groupKyes);
-  if (code != TSDB_CODE_SUCCESS) {
-    goto _error;
-  }
-
   pOperator->fpSet = createOperatorFpSet(operatorDummyOpenFn, doTableScan, NULL, NULL, destroyTableScanOperatorInfo,
                                          NULL, NULL, getTableScannerExecInfo);
 
@@ -912,6 +886,11 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
         pInfo->pRes->info.groupId = groupId;
       }
 
+      uint64_t* groupIdPre = taosHashGet(pOperator->pTaskInfo->tableqinfoList.map, &uid, sizeof(int64_t));
+      if (groupIdPre) {
+        pInfo->pRes->info.groupId = *groupIdPre;
+      }
+
       for (int32_t i = 0; i < taosArrayGetSize(pInfo->pColMatchInfo); ++i) {
         SColMatchInfo* pColMatchInfo = taosArrayGet(pInfo->pColMatchInfo, i);
         if (!pColMatchInfo->output) {
@@ -979,11 +958,24 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
   }
 }
 
-SOperatorInfo* createStreamScanOperatorInfo(void* pDataReader, SReadHandle* pHandle, SArray* pTableIdList,
+static SArray* extractTableIdList(const STableListInfo* pTableGroupInfo) {
+  SArray* tableIdList = taosArrayInit(4, sizeof(uint64_t));
+
+  // Transfer the Array of STableKeyInfo into uid list.
+  for (int32_t i = 0; i < taosArrayGetSize(pTableGroupInfo->pTableList); ++i) {
+    STableKeyInfo* pkeyInfo = taosArrayGet(pTableGroupInfo->pTableList, i);
+    taosArrayPush(tableIdList, &pkeyInfo->uid);
+  }
+
+  return tableIdList;
+}
+
+SOperatorInfo* createStreamScanOperatorInfo(void* pDataReader, SReadHandle* pHandle,
                                             STableScanPhysiNode* pTableScanNode, SExecTaskInfo* pTaskInfo,
                                             STimeWindowAggSupp* pTwSup) {
   SStreamBlockScanInfo* pInfo = taosMemoryCalloc(1, sizeof(SStreamBlockScanInfo));
   SOperatorInfo*        pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
+
   if (pInfo == NULL || pOperator == NULL) {
     terrno = TSDB_CODE_QRY_OUT_OF_MEMORY;
     goto _error;
@@ -992,7 +984,7 @@ SOperatorInfo* createStreamScanOperatorInfo(void* pDataReader, SReadHandle* pHan
   SScanPhysiNode* pScanPhyNode = &pTableScanNode->scan;
 
   SDataBlockDescNode* pDescNode = pScanPhyNode->node.pOutputDataBlockDesc;
-  SOperatorInfo* pTableScanDummy = createTableScanOperatorInfo(pTableScanNode, pDataReader, pHandle, NULL, pTaskInfo);
+  SOperatorInfo* pTableScanDummy = createTableScanOperatorInfo(pTableScanNode, pDataReader, pHandle, pTaskInfo);
 
   STableScanInfo* pSTInfo = (STableScanInfo*)pTableScanDummy->info;
 
@@ -1014,10 +1006,13 @@ SOperatorInfo* createStreamScanOperatorInfo(void* pDataReader, SReadHandle* pHan
 
   // set the extract column id to streamHandle
   tqReadHandleSetColIdList((STqReadHandle*)pHandle->reader, pColIds);
-  int32_t code = tqReadHandleSetTbUidList(pHandle->reader, pTableIdList);
+  SArray* tableIdList = extractTableIdList(&pTaskInfo->tableqinfoList);
+  int32_t code = tqReadHandleSetTbUidList(pHandle->reader, tableIdList);
   if (code != 0) {
+    taosArrayDestroy(tableIdList);
     goto _error;
   }
+  taosArrayDestroy(tableIdList);
 
   pInfo->pBlockLists = taosArrayInit(4, POINTER_BYTES);
   if (pInfo->pBlockLists == NULL) {

source/libs/function/inc/builtinsimpl.h

@@ -55,16 +55,22 @@ int32_t maxCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx);
 bool getAvgFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 bool avgFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
 int32_t avgFunction(SqlFunctionCtx* pCtx);
+int32_t avgFunctionMerge(SqlFunctionCtx* pCtx);
 int32_t avgFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
+int32_t avgPartialFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
 int32_t avgInvertFunction(SqlFunctionCtx* pCtx);
 int32_t avgCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx);
+int32_t getAvgInfoSize();
 
 bool getStddevFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 bool stddevFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);
 int32_t stddevFunction(SqlFunctionCtx* pCtx);
+int32_t stddevFunctionMerge(SqlFunctionCtx* pCtx);
 int32_t stddevFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
+int32_t stddevPartialFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock);
 int32_t stddevInvertFunction(SqlFunctionCtx* pCtx);
 int32_t stddevCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx);
+int32_t getStddevInfoSize();
 
 bool getLeastSQRFuncEnv(struct SFunctionNode* pFunc, SFuncExecEnv* pEnv);
 bool leastSQRFunctionSetup(SqlFunctionCtx *pCtx, SResultRowEntryInfo* pResultInfo);

source/libs/function/src/builtins.c

@@ -155,6 +155,64 @@ static int32_t translateSum(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
   return TSDB_CODE_SUCCESS;
 }
 
+static int32_t translateAvgPartial(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
+  if (1 != LIST_LENGTH(pFunc->pParameterList)) {
+    return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  uint8_t paraType = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
+  if (!IS_NUMERIC_TYPE(paraType) && !IS_NULL_TYPE(paraType)) {
+    return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  pFunc->node.resType = (SDataType){.bytes = getAvgInfoSize() + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY};
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t translateAvgMerge(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
+  if (1 != LIST_LENGTH(pFunc->pParameterList)) {
+    return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  uint8_t paraType = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
+  if (TSDB_DATA_TYPE_BINARY != paraType) {
+    return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  pFunc->node.resType = (SDataType){.bytes = tDataTypes[TSDB_DATA_TYPE_DOUBLE].bytes, .type = TSDB_DATA_TYPE_DOUBLE};
+
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t translateStddevPartial(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
+  if (1 != LIST_LENGTH(pFunc->pParameterList)) {
+    return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  uint8_t paraType = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
+  if (!IS_NUMERIC_TYPE(paraType) && !IS_NULL_TYPE(paraType)) {
+    return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  pFunc->node.resType = (SDataType){.bytes = getStddevInfoSize() + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY};
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t translateStddevMerge(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
+  if (1 != LIST_LENGTH(pFunc->pParameterList)) {
+    return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  uint8_t paraType = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
+  if (TSDB_DATA_TYPE_BINARY != paraType) {
+    return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
+  }
+
+  pFunc->node.resType = (SDataType){.bytes = tDataTypes[TSDB_DATA_TYPE_DOUBLE].bytes, .type = TSDB_DATA_TYPE_DOUBLE};
+
+  return TSDB_CODE_SUCCESS;
+}
+
 static int32_t translateWduration(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
   // pseudo column do not need to check parameters
   pFunc->node.resType = (SDataType){.bytes = sizeof(int64_t), .type = TSDB_DATA_TYPE_BIGINT};
@@ -393,7 +451,8 @@ static int32_t translateTopBotImpl(SFunctionNode* pFunc, char* pErrBuf, int32_t
     pValue->notReserved = true;
 
     // set result type
-    pFunc->node.resType = (SDataType){.bytes = getTopBotInfoSize(pValue->datum.i) + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY};
+    pFunc->node.resType =
+        (SDataType){.bytes = getTopBotInfoSize(pValue->datum.i) + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY};
   } else {
     if (1 != numOfParams) {
       return invaildFuncParaNumErrMsg(pErrBuf, len, pFunc->functionName);
@@ -956,11 +1015,11 @@ static int32_t translateFirstLast(SFunctionNode* pFunc, char* pErrBuf, int32_t l
 
 static int32_t translateFirstLastImpl(SFunctionNode* pFunc, char* pErrBuf, int32_t len, bool isPartial) {
   // first(col_list) will be rewritten as first(col)
-  if (2 != LIST_LENGTH(pFunc->pParameterList)) { //input has two params c0,ts, is this a bug?
+  if (2 != LIST_LENGTH(pFunc->pParameterList)) {  // input has two params c0,ts, is this a bug?
     return TSDB_CODE_SUCCESS;
   }
 
-  SNode* pPara = nodesListGetNode(pFunc->pParameterList, 0);
+  SNode*  pPara = nodesListGetNode(pFunc->pParameterList, 0);
   uint8_t paraType = ((SExprNode*)pPara)->resType.type;
   int32_t paraBytes = ((SExprNode*)pPara)->resType.bytes;
   if (isPartial) {
@@ -969,8 +1028,8 @@ static int32_t translateFirstLastImpl(SFunctionNode* pFunc, char* pErrBuf, int32
                              "The parameters of first/last can only be columns");
     }
 
-    pFunc->node.resType = (SDataType){.bytes = getFirstLastInfoSize(paraBytes) + VARSTR_HEADER_SIZE,
+    pFunc->node.resType =
-                                      .type = TSDB_DATA_TYPE_BINARY};
+        (SDataType){.bytes = getFirstLastInfoSize(paraBytes) + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY};
   } else {
     if (TSDB_DATA_TYPE_BINARY != paraType) {
       return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
@@ -1493,6 +1552,32 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .finalizeFunc = stddevFinalize,
     .invertFunc   = stddevInvertFunction,
     .combineFunc  = stddevCombine,
+    .pPartialFunc = "_stddev_partial",
+    .pMergeFunc   = "_stddev_merge"
+  },
+  {
+    .name = "_stddev_partial",
+    .type = FUNCTION_TYPE_STDDEV_PARTIAL,
+    .classification = FUNC_MGT_AGG_FUNC,
+    .translateFunc = translateStddevPartial,
+    .getEnvFunc   = getStddevFuncEnv,
+    .initFunc     = stddevFunctionSetup,
+    .processFunc  = stddevFunction,
+    .finalizeFunc = stddevPartialFinalize,
+    .invertFunc   = stddevInvertFunction,
+    .combineFunc  = stddevCombine,
+  },
+  {
+    .name = "_stddev_merge",
+    .type = FUNCTION_TYPE_STDDEV_MERGE,
+    .classification = FUNC_MGT_AGG_FUNC,
+    .translateFunc = translateStddevMerge,
+    .getEnvFunc   = getStddevFuncEnv,
+    .initFunc     = stddevFunctionSetup,
+    .processFunc  = stddevFunctionMerge,
+    .finalizeFunc = stddevFinalize,
+    .invertFunc   = stddevInvertFunction,
+    .combineFunc  = stddevCombine,
   },
   {
     .name = "leastsquares",
@@ -1517,11 +1602,37 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
     .finalizeFunc = avgFinalize,
     .invertFunc   = avgInvertFunction,
     .combineFunc  = avgCombine,
+    .pPartialFunc = "_avg_partial",
+    .pMergeFunc   = "_avg_merge"
+  },
+  {
+    .name = "_avg_partial",
+    .type = FUNCTION_TYPE_AVG_PARTIAL,
+    .classification = FUNC_MGT_AGG_FUNC,
+    .translateFunc = translateAvgPartial,
+    .getEnvFunc   = getAvgFuncEnv,
+    .initFunc     = avgFunctionSetup,
+    .processFunc  = avgFunction,
+    .finalizeFunc = avgPartialFinalize,
+    .invertFunc   = avgInvertFunction,
+    .combineFunc  = avgCombine,
+  },
+  {
+    .name = "_avg_merge",
+    .type = FUNCTION_TYPE_AVG_MERGE,
+    .classification = FUNC_MGT_AGG_FUNC,
+    .translateFunc = translateAvgMerge,
+    .getEnvFunc   = getAvgFuncEnv,
+    .initFunc     = avgFunctionSetup,
+    .processFunc  = avgFunctionMerge,
+    .finalizeFunc = avgFinalize,
+    .invertFunc   = avgInvertFunction,
+    .combineFunc  = avgCombine,
   },
   {
     .name = "percentile",
     .type = FUNCTION_TYPE_PERCENTILE,
-    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_REPEAT_SCAN_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_REPEAT_SCAN_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translatePercentile,
     .getEnvFunc   = getPercentileFuncEnv,
     .initFunc     = percentileFunctionSetup,
@@ -1682,7 +1793,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "elapsed",
     .type = FUNCTION_TYPE_ELAPSED,
-    .classification = FUNC_MGT_AGG_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .dataRequiredFunc = statisDataRequired,
     .translateFunc = translateElapsed,
     .getEnvFunc   = getElapsedFuncEnv,
@@ -1803,7 +1914,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "twa",
     .type = FUNCTION_TYPE_TWA,
-    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateInNumOutDou,
     .getEnvFunc    = getTwaFuncEnv,
     .initFunc      = twaFunctionSetup,
@@ -1889,7 +2000,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "diff",
     .type = FUNCTION_TYPE_DIFF,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateDiff,
     .getEnvFunc   = getDiffFuncEnv,
     .initFunc     = diffFunctionSetup,
@@ -1899,7 +2010,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "statecount",
     .type = FUNCTION_TYPE_STATE_COUNT,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateStateCount,
     .getEnvFunc   = getStateFuncEnv,
     .initFunc     = functionSetup,
@@ -1909,7 +2020,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "stateduration",
     .type = FUNCTION_TYPE_STATE_DURATION,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateStateDuration,
     .getEnvFunc   = getStateFuncEnv,
     .initFunc     = functionSetup,
@@ -1919,7 +2030,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "csum",
     .type = FUNCTION_TYPE_CSUM,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateCsum,
     .getEnvFunc   = getCsumFuncEnv,
     .initFunc     = functionSetup,
@@ -1929,7 +2040,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "mavg",
     .type = FUNCTION_TYPE_MAVG,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateMavg,
     .getEnvFunc   = getMavgFuncEnv,
     .initFunc     = mavgFunctionSetup,
@@ -1939,7 +2050,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "sample",
     .type = FUNCTION_TYPE_SAMPLE,
-    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_INDEFINITE_ROWS_FUNC,
+    .classification = FUNC_MGT_AGG_FUNC | FUNC_MGT_SELECT_FUNC | FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateSample,
     .getEnvFunc   = getSampleFuncEnv,
     .initFunc     = sampleFunctionSetup,
@@ -1949,7 +2060,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "tail",
     .type = FUNCTION_TYPE_TAIL,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateTail,
     .getEnvFunc   = getTailFuncEnv,
     .initFunc     = tailFunctionSetup,
@@ -1959,7 +2070,7 @@ const SBuiltinFuncDefinition funcMgtBuiltins[] = {
   {
     .name = "unique",
     .type = FUNCTION_TYPE_UNIQUE,
-    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC,
+    .classification = FUNC_MGT_INDEFINITE_ROWS_FUNC | FUNC_MGT_TIMELINE_FUNC | FUNC_MGT_FORBID_STREAM_FUNC,
     .translateFunc = translateUnique,
     .getEnvFunc   = getUniqueFuncEnv,
     .initFunc     = uniqueFunctionSetup,

source/libs/function/src/builtinsimpl.c

@@ -51,6 +51,7 @@ typedef struct SAvgRes {
   double  result;
   SSumRes sum;
   int64_t count;
+  int16_t type;  // store the original input type, used in merge function
 } SAvgRes;
 
 typedef struct STuplePos {
@@ -88,6 +89,7 @@ typedef struct SStddevRes {
     double  dsum;
     int64_t isum;
   };
+  int16_t type;
 } SStddevRes;
 
 typedef struct SLeastSQRInfo {
@@ -624,6 +626,8 @@ bool getSumFuncEnv(SFunctionNode* UNUSED_PARAM(pFunc), SFuncExecEnv* pEnv) {
   return true;
 }
 
+int32_t getAvgInfoSize() { return (int32_t)sizeof(SAvgRes); }
+
 bool getAvgFuncEnv(SFunctionNode* UNUSED_PARAM(pFunc), SFuncExecEnv* pEnv) {
   pEnv->calcMemSize = sizeof(SAvgRes);
   return true;
@@ -642,11 +646,12 @@ bool avgFunctionSetup(SqlFunctionCtx* pCtx, SResultRowEntryInfo* pResultInfo) {
 int32_t avgFunction(SqlFunctionCtx* pCtx) {
   int32_t numOfElem = 0;
 
-  // Only the pre-computing information loaded and actual data does not loaded
   SInputColumnInfoData* pInput = &pCtx->input;
+  SColumnDataAgg*       pAgg = pInput->pColumnDataAgg[0];
   int32_t               type = pInput->pData[0]->info.type;
 
   SAvgRes* pAvgRes = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+  pAvgRes->type = type;
 
   // computing based on the true data block
   SColumnInfoData* pCol = pInput->pData[0];
@@ -660,95 +665,107 @@ int32_t avgFunction(SqlFunctionCtx* pCtx) {
     goto _avg_over;
   }
 
-  switch (type) {
+  if (pInput->colDataAggIsSet) {
-    case TSDB_DATA_TYPE_TINYINT: {
+    numOfElem = numOfRows - pAgg->numOfNull;
-      int8_t* plist = (int8_t*)pCol->pData;
+    ASSERT(numOfElem >= 0);
-      for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+
-        if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+    pAvgRes->count += numOfElem;
-          continue;
+    if (IS_INTEGER_TYPE(type)) {
+      pAvgRes->sum.isum += pAgg->sum;
+    } else if (IS_FLOAT_TYPE(type)) {
+      pAvgRes->sum.dsum += GET_DOUBLE_VAL((const char*)&(pAgg->sum));
+    }
+  } else {  // computing based on the true data block
+    switch (type) {
+      case TSDB_DATA_TYPE_TINYINT: {
+        int8_t* plist = (int8_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.isum += plist[i];
         }
 
-        numOfElem += 1;
+        break;
-        pAvgRes->count += 1;
-        pAvgRes->sum.isum += plist[i];
       }
 
-      break;
+      case TSDB_DATA_TYPE_SMALLINT: {
-    }
+        int16_t* plist = (int16_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
 
-    case TSDB_DATA_TYPE_SMALLINT: {
+          numOfElem += 1;
-      int16_t* plist = (int16_t*)pCol->pData;
+          pAvgRes->count += 1;
-      for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          pAvgRes->sum.isum += plist[i];
-        if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+        }
-          continue;
+        break;
+      }
+
+      case TSDB_DATA_TYPE_INT: {
+        int32_t* plist = (int32_t*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.isum += plist[i];
         }
 
-        numOfElem += 1;
+        break;
-        pAvgRes->count += 1;
-        pAvgRes->sum.isum += plist[i];
       }
-      break;
-    }
 
-    case TSDB_DATA_TYPE_INT: {
+      case TSDB_DATA_TYPE_BIGINT: {
-      int32_t* plist = (int32_t*)pCol->pData;
+        int64_t* plist = (int64_t*)pCol->pData;
-      for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
-        if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
-          continue;
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.isum += plist[i];
         }
-
+        break;
-        numOfElem += 1;
-        pAvgRes->count += 1;
-        pAvgRes->sum.isum += plist[i];
       }
 
-      break;
+      case TSDB_DATA_TYPE_FLOAT: {
-    }
+        float* plist = (float*)pCol->pData;
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+            continue;
+          }
 
-    case TSDB_DATA_TYPE_BIGINT: {
+          numOfElem += 1;
-      int64_t* plist = (int64_t*)pCol->pData;
+          pAvgRes->count += 1;
-      for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+          pAvgRes->sum.dsum += plist[i];
-        if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
-          continue;
         }
-
+        break;
-        numOfElem += 1;
-        pAvgRes->count += 1;
-        pAvgRes->sum.isum += plist[i];
       }
-      break;
-    }
 
-    case TSDB_DATA_TYPE_FLOAT: {
+      case TSDB_DATA_TYPE_DOUBLE: {
-      float* plist = (float*)pCol->pData;
+        double* plist = (double*)pCol->pData;
-      for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
+        for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
-        if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
+          if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
-          continue;
+            continue;
+          }
+
+          numOfElem += 1;
+          pAvgRes->count += 1;
+          pAvgRes->sum.dsum += plist[i];
         }
-
+        break;
-        numOfElem += 1;
-        pAvgRes->count += 1;
-        pAvgRes->sum.dsum += plist[i];
       }
-      break;
+
+      default:
+        break;
     }
-
-    case TSDB_DATA_TYPE_DOUBLE: {
-      double* plist = (double*)pCol->pData;
-      for (int32_t i = start; i < numOfRows + pInput->startRowIndex; ++i) {
-        if (pCol->hasNull && colDataIsNull_f(pCol->nullbitmap, i)) {
-          continue;
-        }
-
-        numOfElem += 1;
-        pAvgRes->count += 1;
-        pAvgRes->sum.dsum += plist[i];
-      }
-      break;
-    }
-
-    default:
-      break;
   }
 
 _avg_over:
@@ -757,6 +774,37 @@ _avg_over:
   return TSDB_CODE_SUCCESS;
 }
 
+static void avgTransferInfo(SAvgRes* pInput, SAvgRes* pOutput) {
+  pOutput->type = pInput->type;
+  if (IS_INTEGER_TYPE(pOutput->type)) {
+    pOutput->sum.isum += pInput->sum.isum;
+  } else {
+    pOutput->sum.dsum += pInput->sum.dsum;
+  }
+
+  pOutput->count += pInput->count;
+
+  return;
+}
+
+int32_t avgFunctionMerge(SqlFunctionCtx* pCtx) {
+  SInputColumnInfoData* pInput = &pCtx->input;
+  SColumnInfoData*      pCol = pInput->pData[0];
+  ASSERT(pCol->info.type == TSDB_DATA_TYPE_BINARY);
+
+  SAvgRes* pInfo = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+
+  int32_t start = pInput->startRowIndex;
+  char*   data = colDataGetData(pCol, start);
+  SAvgRes* pInputInfo = (SAvgRes*)varDataVal(data);
+
+  avgTransferInfo(pInputInfo, pInfo);
+
+  SET_VAL(GET_RES_INFO(pCtx), 1, 1);
+
+  return TSDB_CODE_SUCCESS;
+}
+
 #define LIST_AVG_N(sumT, T)                                               \
   do {                                                                    \
     T* plist = (T*)pCol->pData;                                           \
@@ -841,8 +889,8 @@ int32_t avgCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx) {
 int32_t avgFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
   SInputColumnInfoData* pInput = &pCtx->input;
 
-  int32_t  type = pInput->pData[0]->info.type;
   SAvgRes* pAvgRes = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+  int32_t  type = pAvgRes->type;
 
   if (IS_INTEGER_TYPE(type)) {
     pAvgRes->result = pAvgRes->sum.isum / ((double)pAvgRes->count);
@@ -858,6 +906,24 @@ int32_t avgFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
   return functionFinalize(pCtx, pBlock);
 }
 
+int32_t avgPartialFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
+  SResultRowEntryInfo* pResInfo = GET_RES_INFO(pCtx);
+  SAvgRes* pInfo = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+  int32_t  resultBytes = getAvgInfoSize();
+  char*    res = taosMemoryCalloc(resultBytes + VARSTR_HEADER_SIZE, sizeof(char));
+
+  memcpy(varDataVal(res), pInfo, resultBytes);
+  varDataSetLen(res, resultBytes);
+
+  int32_t          slotId = pCtx->pExpr->base.resSchema.slotId;
+  SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, slotId);
+
+  colDataAppend(pCol, pBlock->info.rows, res, false);
+
+  taosMemoryFree(res);
+  return pResInfo->numOfRes;
+}
+
 EFuncDataRequired statisDataRequired(SFunctionNode* pFunc, STimeWindow* pTimeWindow) {
   return FUNC_DATA_REQUIRED_STATIS_LOAD;
 }
@@ -1423,6 +1489,8 @@ int32_t maxCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx) {
   return minMaxCombine(pDestCtx, pSourceCtx, 0);
 }
 
+int32_t getStddevInfoSize() { return (int32_t)sizeof(SStddevRes); }
+
 bool getStddevFuncEnv(SFunctionNode* pFunc, SFuncExecEnv* pEnv) {
   pEnv->calcMemSize = sizeof(SStddevRes);
   return true;
@@ -1446,6 +1514,7 @@ int32_t stddevFunction(SqlFunctionCtx* pCtx) {
   int32_t               type = pInput->pData[0]->info.type;
 
   SStddevRes* pStddevRes = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+  pStddevRes->type = type;
 
   // computing based on the true data block
   SColumnInfoData* pCol = pInput->pData[0];
@@ -1562,6 +1631,39 @@ _stddev_over:
   return TSDB_CODE_SUCCESS;
 }
 
+static void stddevTransferInfo(SStddevRes* pInput, SStddevRes* pOutput) {
+  pOutput->type = pInput->type;
+  if (IS_INTEGER_TYPE(pOutput->type)) {
+    pOutput->quadraticISum += pInput->quadraticISum;
+    pOutput->isum += pInput->isum;
+  } else {
+    pOutput->quadraticDSum += pInput->quadraticDSum;
+    pOutput->dsum += pInput->dsum;
+  }
+
+  pOutput->count += pInput->count;
+
+  return;
+}
+
+int32_t stddevFunctionMerge(SqlFunctionCtx* pCtx) {
+  SInputColumnInfoData* pInput = &pCtx->input;
+  SColumnInfoData*      pCol = pInput->pData[0];
+  ASSERT(pCol->info.type == TSDB_DATA_TYPE_BINARY);
+
+  SStddevRes* pInfo = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+
+  int32_t start = pInput->startRowIndex;
+  char*   data = colDataGetData(pCol, start);
+  SStddevRes* pInputInfo = (SStddevRes*)varDataVal(data);
+
+  stddevTransferInfo(pInputInfo, pInfo);
+
+  SET_VAL(GET_RES_INFO(pCtx), 1, 1);
+
+  return TSDB_CODE_SUCCESS;
+}
+
 #define LIST_STDDEV_SUB_N(sumT, T)                                 \
   do {                                                             \
     T* plist = (T*)pCol->pData;                                    \
@@ -1627,9 +1729,10 @@ int32_t stddevInvertFunction(SqlFunctionCtx* pCtx) {
 
 int32_t stddevFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
   SInputColumnInfoData* pInput = &pCtx->input;
-  int32_t               type = pInput->pData[0]->info.type;
   SStddevRes*           pStddevRes = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+  int32_t               type = pStddevRes->type;
   double                avg;
+
   if (IS_INTEGER_TYPE(type)) {
     avg = pStddevRes->isum / ((double)pStddevRes->count);
     pStddevRes->result = sqrt(pStddevRes->quadraticISum / ((double)pStddevRes->count) - avg * avg);
@@ -1641,6 +1744,24 @@ int32_t stddevFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
   return functionFinalize(pCtx, pBlock);
 }
 
+int32_t stddevPartialFinalize(SqlFunctionCtx* pCtx, SSDataBlock* pBlock) {
+  SResultRowEntryInfo* pResInfo = GET_RES_INFO(pCtx);
+  SStddevRes* pInfo = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
+  int32_t  resultBytes = getStddevInfoSize();
+  char*    res = taosMemoryCalloc(resultBytes + VARSTR_HEADER_SIZE, sizeof(char));
+
+  memcpy(varDataVal(res), pInfo, resultBytes);
+  varDataSetLen(res, resultBytes);
+
+  int32_t          slotId = pCtx->pExpr->base.resSchema.slotId;
+  SColumnInfoData* pCol = taosArrayGet(pBlock->pDataBlock, slotId);
+
+  colDataAppend(pCol, pBlock->info.rows, res, false);
+
+  taosMemoryFree(res);
+  return pResInfo->numOfRes;
+}
+
 int32_t stddevCombine(SqlFunctionCtx* pDestCtx, SqlFunctionCtx* pSourceCtx) {
   SResultRowEntryInfo* pDResInfo = GET_RES_INFO(pDestCtx);
   SStddevRes*          pDBuf = GET_ROWCELL_INTERBUF(pDResInfo);
@@ -3226,11 +3347,10 @@ int32_t spreadFunctionMerge(SqlFunctionCtx* pCtx) {
   ASSERT(pCol->info.type == TSDB_DATA_TYPE_BINARY);
 
   SSpreadInfo* pInfo = GET_ROWCELL_INTERBUF(GET_RES_INFO(pCtx));
-  SSpreadInfo* pInputInfo;
 
   int32_t start = pInput->startRowIndex;
   char*   data = colDataGetData(pCol, start);
-  pInputInfo = (SSpreadInfo*)varDataVal(data);
+  SSpreadInfo* pInputInfo = (SSpreadInfo*)varDataVal(data);
 
   spreadTransferInfo(pInputInfo, pInfo);
 

source/libs/index/src/indexFilter.c

@@ -97,7 +97,14 @@ static int32_t sifGetOperParamNum(EOperatorType ty) {
   }
   return 2;
 }
-static int32_t sifValidateColumn(SColumnNode *cn) {
+static int32_t sifValidOp(EOperatorType ty) {
+  if ((ty >= OP_TYPE_ADD && ty <= OP_TYPE_BIT_OR) || (ty == OP_TYPE_IN || ty == OP_TYPE_NOT_IN) ||
+      (ty == OP_TYPE_LIKE || ty == OP_TYPE_NOT_LIKE || ty == OP_TYPE_MATCH || ty == OP_TYPE_NMATCH)) {
+    return -1;
+  }
+  return 0;
+}
+static int32_t sifValidColumn(SColumnNode *cn) {
   // add more check
   if (cn == NULL) {
     return TSDB_CODE_QRY_INVALID_INPUT;
@@ -176,6 +183,7 @@ static int32_t sifInitJsonParam(SNode *node, SIFParam *param, SIFCtx *ctx) {
   memcpy(param->colName, r->literal, strlen(r->literal));
   // sprintf(param->colName, "%s_%s", l->colName, r->literal);
   param->colValType = r->typeData;
+  param->status = SFLT_COARSE_INDEX;
   return 0;
   // memcpy(param->colName, l->colName, sizeof(l->colName));
 }
@@ -197,7 +205,7 @@ static int32_t sifInitParam(SNode *node, SIFParam *param, SIFCtx *ctx) {
     case QUERY_NODE_COLUMN: {
       SColumnNode *cn = (SColumnNode *)node;
       /*only support tag column*/
-      SIF_ERR_RET(sifValidateColumn(cn));
+      SIF_ERR_RET(sifValidColumn(cn));
 
       param->colId = cn->colId;
       param->colValType = cn->node.resType.type;
@@ -247,11 +255,13 @@ static int32_t sifInitOperParams(SIFParam **params, SOperatorNode *node, SIFCtx
     return code;
   }
   SIFParam *paramList = taosMemoryCalloc(nParam, sizeof(SIFParam));
+
   if (NULL == paramList) {
     SIF_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
   }
 
-  if (nodeType(node->pLeft) == QUERY_NODE_OPERATOR) {
+  if (nodeType(node->pLeft) == QUERY_NODE_OPERATOR &&
+      (((SOperatorNode *)(node->pLeft))->opType == OP_TYPE_JSON_GET_VALUE)) {
     SNode *interNode = (node->pLeft);
     SIF_ERR_JRET(sifInitJsonParam(interNode, &paramList[0], ctx));
     if (nParam > 1) {
@@ -505,6 +515,11 @@ static int32_t sifGetOperFn(int32_t funcId, sif_func_t *func, SIdxFltStatus *sta
 
 static int32_t sifExecOper(SOperatorNode *node, SIFCtx *ctx, SIFParam *output) {
   int32_t code = 0;
+  if (sifValidOp(node->opType) < 0) {
+    output->status = SFLT_NOT_INDEX;
+    return code;
+  }
+
   int32_t nParam = sifGetOperParamNum(node->opType);
   if (nParam <= 1) {
     output->status = SFLT_NOT_INDEX;
@@ -513,9 +528,15 @@ static int32_t sifExecOper(SOperatorNode *node, SIFCtx *ctx, SIFParam *output) {
   if (node->opType == OP_TYPE_JSON_GET_VALUE) {
     return code;
   }
-  SIFParam *params = NULL;
 
+  SIFParam *params = NULL;
   SIF_ERR_RET(sifInitOperParams(&params, node, ctx));
+
+  if (params[0].status == SFLT_NOT_INDEX || (nParam > 1 && params[1].status == SFLT_NOT_INDEX)) {
+    output->status = SFLT_NOT_INDEX;
+    return code;
+  }
+
   // ugly code, refactor later
   output->arg = ctx->arg;
   sif_func_t operFn = sifNullFunc;

source/libs/parser/src/parAstCreater.c

@@ -230,9 +230,13 @@ SNode* releaseRawExprNode(SAstCreateContext* pCxt, SNode* pNode) {
   SRawExprNode* pRawExpr = (SRawExprNode*)pNode;
   SNode*        pExpr = pRawExpr->pNode;
   if (nodesIsExprNode(pExpr)) {
-    int32_t len = TMIN(sizeof(((SExprNode*)pExpr)->aliasName) - 1, pRawExpr->n);
+    if (QUERY_NODE_COLUMN == nodeType(pExpr)) {
-    strncpy(((SExprNode*)pExpr)->aliasName, pRawExpr->p, len);
+      strcpy(((SExprNode*)pExpr)->aliasName, ((SColumnNode*)pExpr)->colName);
-    ((SExprNode*)pExpr)->aliasName[len] = '\0';
+    } else {
+      int32_t len = TMIN(sizeof(((SExprNode*)pExpr)->aliasName) - 1, pRawExpr->n);
+      strncpy(((SExprNode*)pExpr)->aliasName, pRawExpr->p, len);
+      ((SExprNode*)pExpr)->aliasName[len] = '\0';
+    }
   }
   taosMemoryFreeClear(pNode);
   return pExpr;
@@ -800,10 +804,10 @@ SNode* setDatabaseOption(SAstCreateContext* pCxt, SNode* pOptions, EDatabaseOpti
     case DB_OPTION_RETENTIONS:
       ((SDatabaseOptions*)pOptions)->pRetentions = pVal;
       break;
-    case DB_OPTION_SCHEMALESS:
+//    case DB_OPTION_SCHEMALESS:
 //      ((SDatabaseOptions*)pOptions)->schemaless = taosStr2Int8(((SToken*)pVal)->z, NULL, 10);
-      ((SDatabaseOptions*)pOptions)->schemaless = 1;
+//      ((SDatabaseOptions*)pOptions)->schemaless = 0;
-      break;
+//      break;
     default:
       break;
   }

source/libs/parser/src/parTranslater.c

@@ -537,13 +537,15 @@ static int32_t findAndSetColumn(STranslateContext* pCxt, SColumnNode** pColRef,
     SNode*     pNode;
     FOREACH(pNode, pProjectList) {
       SExprNode* pExpr = (SExprNode*)pNode;
-      if (0 == strcmp(pCol->colName, pExpr->aliasName) ||
+      if (0 == strcmp(pCol->colName, pExpr->aliasName)) {
-          (isPrimaryKey((STempTableNode*)pTable, pNode) && isInternalPrimaryKey(pCol))) {
         if (*pFound) {
           return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_AMBIGUOUS_COLUMN, pCol->colName);
         }
         setColumnInfoByExpr(pTable, pExpr, pColRef);
         *pFound = true;
+      } else if (isPrimaryKey((STempTableNode*)pTable, pNode) && isInternalPrimaryKey(pCol)) {
+        setColumnInfoByExpr(pTable, pExpr, pColRef);
+        *pFound = true;
       }
     }
   }
@@ -662,9 +664,24 @@ static int32_t parseTimeFromValueNode(STranslateContext* pCxt, SValueNode* pVal)
     }
     char* pEnd = NULL;
     pVal->datum.i = taosStr2Int64(pVal->literal, &pEnd, 10);
-    return (NULL != pEnd && '\0' == *pEnd) ? TSDB_CODE_SUCCESS : TSDB_CODE_FAILED;
+    return (NULL != pEnd && '\0' == *pEnd) ? TSDB_CODE_SUCCESS : TSDB_CODE_PAR_WRONG_VALUE_TYPE;
   } else {
-    return TSDB_CODE_FAILED;
+    return TSDB_CODE_PAR_WRONG_VALUE_TYPE;
+  }
+}
+
+static int32_t parseBoolFromValueNode(STranslateContext* pCxt, SValueNode* pVal) {
+  if (IS_VAR_DATA_TYPE(pVal->node.resType.type) || TSDB_DATA_TYPE_BOOL == pVal->node.resType.type) {
+    pVal->datum.b = (0 == strcasecmp(pVal->literal, "true"));
+    return TSDB_CODE_SUCCESS;
+  } else if (IS_INTEGER_TYPE(pVal->node.resType.type)) {
+    pVal->datum.b = (0 != taosStr2Int64(pVal->literal, NULL, 10));
+    return TSDB_CODE_SUCCESS;
+  } else if (IS_FLOAT_TYPE(pVal->node.resType.type)) {
+    pVal->datum.b = (0 != taosStr2Double(pVal->literal, NULL));
+    return TSDB_CODE_SUCCESS;
+  } else {
+    return TSDB_CODE_PAR_WRONG_VALUE_TYPE;
   }
 }
 
@@ -685,7 +702,9 @@ static EDealRes translateValueImpl(STranslateContext* pCxt, SValueNode* pVal, SD
       case TSDB_DATA_TYPE_NULL:
         break;
       case TSDB_DATA_TYPE_BOOL:
-        pVal->datum.b = (0 == strcasecmp(pVal->literal, "true"));
+        if (TSDB_CODE_SUCCESS != parseBoolFromValueNode(pCxt, pVal)) {
+          return generateDealNodeErrMsg(pCxt, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pVal->literal);
+        }
         *(bool*)&pVal->typeData = pVal->datum.b;
         break;
       case TSDB_DATA_TYPE_TINYINT: {
@@ -1068,6 +1087,16 @@ static int32_t translateForbidFillFunc(STranslateContext* pCxt, SFunctionNode* p
   return TSDB_CODE_SUCCESS;
 }
 
+static int32_t translateWindowPseudoColumnFunc(STranslateContext* pCxt, SFunctionNode* pFunc) {
+  if (!fmIsWindowPseudoColumnFunc(pFunc->funcId)) {
+    return TSDB_CODE_SUCCESS;
+  }
+  if (NULL == pCxt->pCurrSelectStmt->pWindow) {
+    return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_WINDOW_PC);
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
 static void setFuncClassification(SSelectStmt* pSelect, SFunctionNode* pFunc) {
   if (NULL != pSelect) {
     pSelect->hasAggFuncs = pSelect->hasAggFuncs ? true : fmIsAggFunc(pFunc->funcId);
@@ -1097,6 +1126,9 @@ static EDealRes translateFunction(STranslateContext* pCxt, SFunctionNode* pFunc)
   if (TSDB_CODE_SUCCESS == pCxt->errCode) {
     pCxt->errCode = translateForbidFillFunc(pCxt, pFunc);
   }
+  if (TSDB_CODE_SUCCESS == pCxt->errCode) {
+    pCxt->errCode = translateWindowPseudoColumnFunc(pCxt, pFunc);
+  }
   if (TSDB_CODE_SUCCESS == pCxt->errCode) {
     setFuncClassification(pCxt->pCurrSelectStmt, pFunc);
   }
@@ -3106,7 +3138,11 @@ static int32_t translateDropSuperTable(STranslateContext* pCxt, SDropSuperTableS
                                    pStmt->ignoreNotExists);
 }
 
-static int32_t setAlterTableField(SAlterTableStmt* pStmt, SMAlterStbReq* pAlterReq) {
+static int32_t buildAlterSuperTableReq(STranslateContext* pCxt, SAlterTableStmt* pStmt, SMAlterStbReq* pAlterReq) {
+  SName tableName;
+  tNameExtractFullName(toName(pCxt->pParseCxt->acctId, pStmt->dbName, pStmt->tableName, &tableName), pAlterReq->name);
+  pAlterReq->alterType = pStmt->alterType;
+
   if (TSDB_ALTER_TABLE_UPDATE_OPTIONS == pStmt->alterType) {
     pAlterReq->ttl = pStmt->pOptions->ttl;
     if ('\0' != pStmt->pOptions->comment[0]) {
@@ -3154,15 +3190,45 @@ static int32_t setAlterTableField(SAlterTableStmt* pStmt, SMAlterStbReq* pAlterR
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t translateAlterTable(STranslateContext* pCxt, SAlterTableStmt* pStmt) {
+static SSchema* getColSchema(STableMeta* pTableMeta, const char* pTagName) {
-  SMAlterStbReq alterReq = {0};
+  int32_t numOfFields = getNumOfTags(pTableMeta) + getNumOfColumns(pTableMeta);
-  SName         tableName;
+  for (int32_t i = 0; i < numOfFields; ++i) {
-  tNameExtractFullName(toName(pCxt->pParseCxt->acctId, pStmt->dbName, pStmt->tableName, &tableName), alterReq.name);
+    SSchema* pTagSchema = pTableMeta->schema + i;
-  alterReq.alterType = pStmt->alterType;
+    if (0 == strcmp(pTagName, pTagSchema->name)) {
+      return pTagSchema;
+    }
+  }
+  return NULL;
+}
+
+static int32_t checkAlterSuperTable(STranslateContext* pCxt, SAlterTableStmt* pStmt) {
   if (TSDB_ALTER_TABLE_UPDATE_TAG_VAL == pStmt->alterType || TSDB_ALTER_TABLE_UPDATE_COLUMN_NAME == pStmt->alterType) {
     return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_ALTER_TABLE);
   }
-  int32_t code = setAlterTableField(pStmt, &alterReq);
+  if (TSDB_ALTER_TABLE_UPDATE_COLUMN_BYTES == pStmt->alterType ||
+      TSDB_ALTER_TABLE_UPDATE_TAG_BYTES == pStmt->alterType) {
+    STableMeta* pTableMeta = NULL;
+    int32_t     code = getTableMeta(pCxt, pStmt->dbName, pStmt->tableName, &pTableMeta);
+    if (TSDB_CODE_SUCCESS == code) {
+      SSchema* pSchema = getColSchema(pTableMeta, pStmt->colName);
+      if (NULL == pSchema) {
+        code = generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_COLUMN, pStmt->colName);
+      } else if (!IS_VAR_DATA_TYPE(pSchema->type) || pSchema->type != pStmt->dataType.type ||
+                 pSchema->bytes >= calcTypeBytes(pStmt->dataType)) {
+        code = generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_MODIFY_COL);
+      }
+    }
+    return code;
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t translateAlterSuperTable(STranslateContext* pCxt, SAlterTableStmt* pStmt) {
+  SMAlterStbReq alterReq = {0};
+  int32_t       code = checkAlterSuperTable(pCxt, pStmt);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = buildAlterSuperTableReq(pCxt, pStmt, &alterReq);
+  }
   if (TSDB_CODE_SUCCESS == code) {
     code = buildCmdMsg(pCxt, TDMT_MND_ALTER_STB, (FSerializeFunc)tSerializeSMAlterStbReq, &alterReq);
   }
@@ -3835,7 +3901,7 @@ static int32_t translateQuery(STranslateContext* pCxt, SNode* pNode) {
       code = translateDropSuperTable(pCxt, (SDropSuperTableStmt*)pNode);
       break;
     case QUERY_NODE_ALTER_TABLE_STMT:
-      code = translateAlterTable(pCxt, (SAlterTableStmt*)pNode);
+      code = translateAlterSuperTable(pCxt, (SAlterTableStmt*)pNode);
       break;
     case QUERY_NODE_CREATE_USER_STMT:
       code = translateCreateUser(pCxt, (SCreateUserStmt*)pNode);
@@ -4445,10 +4511,35 @@ static int32_t translateTagVal(STranslateContext* pCxt, uint8_t precision, SSche
                 ? pCxt->errCode
                 : TSDB_CODE_SUCCESS);
   } else {
+    // return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_WRONG_VALUE_TYPE, ((SExprNode*)pNode)->aliasName);
     return TSDB_CODE_FAILED;
   }
 }
 
+static int32_t buildJsonTagVal(STranslateContext* pCxt, SSchema* pTagSchema, SValueNode* pVal, SArray* pTagArray,
+                               STag** ppTag) {
+  if (pVal->literal && strlen(pVal->literal) > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE) {
+    return buildSyntaxErrMsg(&pCxt->msgBuf, "json string too long than 4095", pVal->literal);
+  }
+
+  return parseJsontoTagData(pVal->literal, pTagArray, ppTag, &pCxt->msgBuf);
+}
+
+static int32_t buildNormalTagVal(STranslateContext* pCxt, SSchema* pTagSchema, SValueNode* pVal, SArray* pTagArray) {
+  if (pVal->node.resType.type != TSDB_DATA_TYPE_NULL) {
+    void*   nodeVal = nodesGetValueFromNode(pVal);
+    STagVal val = {.cid = pTagSchema->colId, .type = pTagSchema->type};
+    if (IS_VAR_DATA_TYPE(pTagSchema->type)) {
+      val.pData = varDataVal(nodeVal);
+      val.nData = varDataLen(nodeVal);
+    } else {
+      memcpy(&val.i64, nodeVal, pTagSchema->bytes);
+    }
+    taosArrayPush(pTagArray, &val);
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
 static int32_t buildKVRowForBindTags(STranslateContext* pCxt, SCreateSubTableClause* pStmt, STableMeta* pSuperTableMeta,
                                      STag** ppTag) {
   int32_t numOfTags = getNumOfTags(pSuperTableMeta);
@@ -4458,11 +4549,10 @@ static int32_t buildKVRowForBindTags(STranslateContext* pCxt, SCreateSubTableCla
   }
 
   SArray* pTagArray = taosArrayInit(LIST_LENGTH(pStmt->pValsOfTags), sizeof(STagVal));
-  if (!pTagArray) {
+  if (NULL == pTagArray) {
-    return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_TSC_OUT_OF_MEMORY);
+    return TSDB_CODE_OUT_OF_MEMORY;
   }
   int32_t  code = TSDB_CODE_SUCCESS;
-  int16_t  nTags = 0, nBufPos = 0;
   SSchema* pTagSchema = getTableTagSchema(pSuperTableMeta);
   SNode *  pTag = NULL, *pNode = NULL;
   bool     isJson = false;
@@ -4490,27 +4580,12 @@ static int32_t buildKVRowForBindTags(STranslateContext* pCxt, SCreateSubTableCla
     } else {
       REPLACE_LIST2_NODE(pVal);
     }
-    if (pTagSchema->type == TSDB_DATA_TYPE_JSON) {
-      if (pVal->literal && strlen(pVal->literal) > (TSDB_MAX_JSON_TAG_LEN - VARSTR_HEADER_SIZE) / TSDB_NCHAR_SIZE) {
-        code = buildSyntaxErrMsg(&pCxt->msgBuf, "json string too long than 4095", pVal->literal);
-        goto end;
-      }
 
+    if (pSchema->type == TSDB_DATA_TYPE_JSON) {
       isJson = true;
-      code = parseJsontoTagData(pVal->literal, pTagArray, ppTag, &pCxt->msgBuf);
+      code = buildJsonTagVal(pCxt, pSchema, pVal, pTagArray, ppTag);
-      if (code != TSDB_CODE_SUCCESS) {
-        goto end;
-      }
     } else if (pVal->node.resType.type != TSDB_DATA_TYPE_NULL) {
-      void*   nodeVal = nodesGetValueFromNode(pVal);
+      code = buildNormalTagVal(pCxt, pSchema, pVal, pTagArray);
-      STagVal val = {.cid = pTagSchema->colId, .type = pTagSchema->type};
-      if (IS_VAR_DATA_TYPE(pTagSchema->type)) {
-        val.pData = varDataVal(nodeVal);
-        val.nData = varDataLen(nodeVal);
-      } else {
-        memcpy(&val.i64, nodeVal, pTagSchema->bytes);
-      }
-      taosArrayPush(pTagArray, &val);
     }
   }
 
@@ -4825,17 +4900,6 @@ static int32_t rewriteDropTable(STranslateContext* pCxt, SQuery* pQuery) {
   return rewriteToVnodeModifyOpStmt(pQuery, pBufArray);
 }
 
-static SSchema* getColSchema(STableMeta* pTableMeta, const char* pTagName) {
-  int32_t numOfFields = getNumOfTags(pTableMeta) + getNumOfColumns(pTableMeta);
-  for (int32_t i = 0; i < numOfFields; ++i) {
-    SSchema* pTagSchema = pTableMeta->schema + i;
-    if (0 == strcmp(pTagName, pTagSchema->name)) {
-      return pTagSchema;
-    }
-  }
-  return NULL;
-}
-
 static int32_t buildUpdateTagValReq(STranslateContext* pCxt, SAlterTableStmt* pStmt, STableMeta* pTableMeta,
                                     SVAlterTbReq* pReq) {
   SSchema* pSchema = getColSchema(pTableMeta, pStmt->colName);
@@ -4853,6 +4917,10 @@ static int32_t buildUpdateTagValReq(STranslateContext* pCxt, SAlterTableStmt* pS
     return pCxt->errCode;
   }
 
+  if (IS_VAR_DATA_TYPE(pSchema->type) && strlen(pStmt->pVal->literal) > pSchema->bytes) {
+    return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_WRONG_VALUE_TYPE, pStmt->pVal->literal);
+  }
+
   pReq->isNull = (TSDB_DATA_TYPE_NULL == pStmt->pVal->node.resType.type);
   if (pStmt->pVal->node.resType.type == TSDB_DATA_TYPE_JSON) {
     if (pStmt->pVal->literal &&

source/libs/parser/src/parUtil.c

@@ -186,6 +186,8 @@ static char* getSyntaxErrFormat(int32_t errCode) {
       return "The REDISTRIBUTE VGROUP statement only support 1 to 3 dnodes";
     case TSDB_CODE_PAR_FILL_NOT_ALLOWED_FUNC:
       return "%s function not allowed in fill query";
+    case TSDB_CODE_PAR_INVALID_WINDOW_PC:
+      return "_WSTARTTS, _WENDTS and _WDURATION can only be used in window queries";
     case TSDB_CODE_OUT_OF_MEMORY:
       return "Out of memory";
     default:

source/libs/parser/src/parser.c

@@ -207,6 +207,8 @@ int32_t qSetSTableIdForRsma(SNode* pStmt, int64_t uid) {
   return TSDB_CODE_FAILED;
 }
 
+void qCleanupKeywordsTable() { taosCleanupKeywordsTable(); }
+
 int32_t qStmtBindParams(SQuery* pQuery, TAOS_MULTI_BIND* pParams, int32_t colIdx) {
   int32_t code = TSDB_CODE_SUCCESS;
 

source/libs/parser/test/mockCatalog.cpp

@@ -141,16 +141,18 @@ void generateTestT1(MockCatalogService* mcs) {
  *          c2         |       column       |      VARCHAR       |    20    |
  *         tag1        |        tag         |        INT         |    4     |
  *         tag2        |        tag         |      VARCHAR       |    20    |
+ *         tag3        |        tag         |     TIMESTAMP      |    8     |
  * Child Table: st1s1, st1s2
  */
 void generateTestST1(MockCatalogService* mcs) {
-  ITableBuilder& builder = mcs->createTableBuilder("test", "st1", TSDB_SUPER_TABLE, 3, 2)
+  ITableBuilder& builder = mcs->createTableBuilder("test", "st1", TSDB_SUPER_TABLE, 3, 3)
                                .setPrecision(TSDB_TIME_PRECISION_MILLI)
                                .addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
                                .addColumn("c1", TSDB_DATA_TYPE_INT)
                                .addColumn("c2", TSDB_DATA_TYPE_BINARY, 20)
                                .addTag("tag1", TSDB_DATA_TYPE_INT)
-                               .addTag("tag2", TSDB_DATA_TYPE_BINARY, 20);
+                               .addTag("tag2", TSDB_DATA_TYPE_BINARY, 20)
+                               .addTag("tag3", TSDB_DATA_TYPE_TIMESTAMP);
   builder.done();
   mcs->createSubTable("test", "st1", "st1s1", 1);
   mcs->createSubTable("test", "st1", "st1s2", 2);
@@ -189,17 +191,17 @@ void generateFunctions(MockCatalogService* mcs) {
 
 int32_t __catalogGetHandle(const char* clusterId, struct SCatalog** catalogHandle) { return 0; }
 
-int32_t __catalogGetTableMeta(struct SCatalog* pCatalog, SRequestConnInfo *pConn, const SName* pTableName,
+int32_t __catalogGetTableMeta(struct SCatalog* pCatalog, SRequestConnInfo* pConn, const SName* pTableName,
                               STableMeta** pTableMeta) {
   return g_mockCatalogService->catalogGetTableMeta(pTableName, pTableMeta);
 }
 
-int32_t __catalogGetTableHashVgroup(struct SCatalog* pCatalog, SRequestConnInfo *pConn,
+int32_t __catalogGetTableHashVgroup(struct SCatalog* pCatalog, SRequestConnInfo* pConn, const SName* pTableName,
-                                    const SName* pTableName, SVgroupInfo* vgInfo) {
+                                    SVgroupInfo* vgInfo) {
   return g_mockCatalogService->catalogGetTableHashVgroup(pTableName, vgInfo);
 }
 
-int32_t __catalogGetTableDistVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const SName* pTableName,
+int32_t __catalogGetTableDistVgInfo(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName,
                                     SArray** pVgList) {
   return g_mockCatalogService->catalogGetTableDistVgInfo(pTableName, pVgList);
 }
@@ -209,28 +211,26 @@ int32_t __catalogGetDBVgVersion(SCatalog* pCtg, const char* dbFName, int32_t* ve
   return 0;
 }
 
-int32_t __catalogGetDBVgInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char* dbFName,
+int32_t __catalogGetDBVgInfo(SCatalog* pCtg, SRequestConnInfo* pConn, const char* dbFName, SArray** pVgList) {
-                             SArray** pVgList) {
   return g_mockCatalogService->catalogGetDBVgInfo(dbFName, pVgList);
 }
 
-int32_t __catalogGetDBCfg(SCatalog* pCtg, SRequestConnInfo *pConn, const char* dbFName, SDbCfgInfo* pDbCfg) {
+int32_t __catalogGetDBCfg(SCatalog* pCtg, SRequestConnInfo* pConn, const char* dbFName, SDbCfgInfo* pDbCfg) {
   return 0;
 }
 
-int32_t __catalogChkAuth(SCatalog* pCtg, SRequestConnInfo *pConn, const char* user, const char* dbFName,
+int32_t __catalogChkAuth(SCatalog* pCtg, SRequestConnInfo* pConn, const char* user, const char* dbFName, AUTH_TYPE type,
-                         AUTH_TYPE type, bool* pass) {
+                         bool* pass) {
   *pass = true;
   return 0;
 }
 
-int32_t __catalogGetUdfInfo(SCatalog* pCtg, SRequestConnInfo *pConn, const char* funcName,
+int32_t __catalogGetUdfInfo(SCatalog* pCtg, SRequestConnInfo* pConn, const char* funcName, SFuncInfo* pInfo) {
-                            SFuncInfo* pInfo) {
   return g_mockCatalogService->catalogGetUdfInfo(funcName, pInfo);
 }
 
-int32_t __catalogRefreshGetTableMeta(SCatalog* pCatalog, SRequestConnInfo *pConn,
+int32_t __catalogRefreshGetTableMeta(SCatalog* pCatalog, SRequestConnInfo* pConn, const SName* pTableName,
-                                     const SName* pTableName, STableMeta** pTableMeta, int32_t isSTable) {
+                                     STableMeta** pTableMeta, int32_t isSTable) {
   return g_mockCatalogService->catalogGetTableMeta(pTableName, pTableMeta);
 }
 

source/libs/parser/test/parInitialATest.cpp

@@ -163,9 +163,9 @@ TEST_F(ParserInitialATest, alterSTable) {
   run("ALTER TABLE st1 DROP COLUMN c1");
   clearAlterStbReq();
 
-  setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_COLUMN_BYTES, 1, "c1", TSDB_DATA_TYPE_VARCHAR,
+  setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_COLUMN_BYTES, 1, "c2", TSDB_DATA_TYPE_VARCHAR,
-                     20 + VARSTR_HEADER_SIZE);
+                     30 + VARSTR_HEADER_SIZE);
-  run("ALTER TABLE st1 MODIFY COLUMN c1 VARCHAR(20)");
+  run("ALTER TABLE st1 MODIFY COLUMN c2 VARCHAR(30)");
   clearAlterStbReq();
 
   // setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_COLUMN_NAME, 2, "c1", 0, 0, "cc1");
@@ -179,9 +179,9 @@ TEST_F(ParserInitialATest, alterSTable) {
   run("ALTER TABLE st1 DROP TAG tag1");
   clearAlterStbReq();
 
-  setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_TAG_BYTES, 1, "tag1", TSDB_DATA_TYPE_VARCHAR,
+  setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_TAG_BYTES, 1, "tag2", TSDB_DATA_TYPE_VARCHAR,
-                     20 + VARSTR_HEADER_SIZE);
+                     30 + VARSTR_HEADER_SIZE);
-  run("ALTER TABLE st1 MODIFY TAG tag1 VARCHAR(20)");
+  run("ALTER TABLE st1 MODIFY TAG tag2 VARCHAR(30)");
   clearAlterStbReq();
 
   setAlterStbReqFunc("st1", TSDB_ALTER_TABLE_UPDATE_TAG_NAME, 2, "tag1", 0, 0, "tag11");
@@ -196,6 +196,10 @@ TEST_F(ParserInitialATest, alterSTableSemanticCheck) {
   useDb("root", "test");
 
   run("ALTER TABLE st1 RENAME COLUMN c1 cc1", TSDB_CODE_PAR_INVALID_ALTER_TABLE);
+
+  run("ALTER TABLE st1 MODIFY COLUMN c2 NCHAR(10)", TSDB_CODE_PAR_INVALID_MODIFY_COL);
+
+  run("ALTER TABLE st1 MODIFY TAG tag2 NCHAR(10)", TSDB_CODE_PAR_INVALID_MODIFY_COL);
 }
 
 TEST_F(ParserInitialATest, alterTable) {
@@ -336,6 +340,8 @@ TEST_F(ParserInitialATest, alterTableSemanticCheck) {
   useDb("root", "test");
 
   run("ALTER TABLE st1s1 RENAME COLUMN c1 cc1", TSDB_CODE_PAR_INVALID_ALTER_TABLE);
+
+  run("ALTER TABLE st1s1 SET TAG tag2 =  '123456789012345678901'", TSDB_CODE_PAR_WRONG_VALUE_TYPE);
 }
 
 TEST_F(ParserInitialATest, alterUser) {

source/libs/parser/test/parInitialCTest.cpp

@@ -155,7 +155,7 @@ TEST_F(ParserInitialCTest, createDatabase) {
     ASSERT_EQ(req.replications, expect.replications);
     ASSERT_EQ(req.strict, expect.strict);
     ASSERT_EQ(req.cacheLastRow, expect.cacheLastRow);
-    ASSERT_EQ(req.schemaless, expect.schemaless);
+    //ASSERT_EQ(req.schemaless, expect.schemaless);
     ASSERT_EQ(req.ignoreExist, expect.ignoreExist);
     ASSERT_EQ(req.numOfRetensions, expect.numOfRetensions);
     if (expect.numOfRetensions > 0) {
@@ -548,12 +548,14 @@ TEST_F(ParserInitialCTest, createTable) {
       "a14 NCHAR(30), a15 VARCHAR(50)) "
       "TTL 100 COMMENT 'test create table' SMA(c1, c2, c3) ROLLUP (MIN) FILE_FACTOR 0.1");
 
-  run("CREATE TABLE IF NOT EXISTS t1 USING st1 TAGS(1, 'wxy')");
+  run("CREATE TABLE IF NOT EXISTS t1 USING st1 TAGS(1, 'wxy', NOW)");
 
   run("CREATE TABLE "
       "IF NOT EXISTS test.t1 USING test.st1 (tag1, tag2) TAGS(1, 'abc') "
       "IF NOT EXISTS test.t2 USING test.st1 (tag1, tag2) TAGS(2, 'abc') "
       "IF NOT EXISTS test.t3 USING test.st1 (tag1, tag2) TAGS(3, 'abc') ");
+
+  // run("CREATE TABLE IF NOT EXISTS t1 USING st1 TAGS(1, 'wxy', NOW + 1S)");
 }
 
 TEST_F(ParserInitialCTest, createTopic) {

source/libs/parser/test/parInsertTest.cpp

@@ -245,8 +245,8 @@ TEST_F(InsertTest, autoCreateTableTest) {
   setDatabase("root", "test");
 
   bind(
-      "insert into st1s1 using st1 tags(1, 'wxy') values (now, 1, \"beijing\")(now+1s, 2, \"shanghai\")(now+2s, 3, "
+      "insert into st1s1 using st1 tags(1, 'wxy', now) "
-      "\"guangzhou\")");
+      "values (now, 1, \"beijing\")(now+1s, 2, \"shanghai\")(now+2s, 3, \"guangzhou\")");
   ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
   dumpReslut();
   checkReslut(1, 3);
@@ -257,8 +257,8 @@ TEST_F(InsertTest, autoCreateTableTest) {
   ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
 
   bind(
-      "insert into st1s1 using st1 tags(1, 'wxy') values (now, 1, \"beijing\")(now+1s, 2, \"shanghai\")(now+2s, 3, "
+      "insert into st1s1 using st1 tags(1, 'wxy', now) "
-      "\"guangzhou\")");
+      "values (now, 1, \"beijing\")(now+1s, 2, \"shanghai\")(now+2s, 3, \"guangzhou\")");
   ASSERT_EQ(runAsync(), TSDB_CODE_SUCCESS);
 
   bind(

source/libs/parser/test/parSelectTest.cpp

@@ -219,6 +219,7 @@ TEST_F(ParserSelectTest, intervalSemanticCheck) {
   run("SELECT HISTOGRAM(c1, 'log_bin', '{\"start\": -33,\"factor\": 55,\"count\": 5,\"infinity\": false}', 1) FROM t1 "
       "WHERE ts > TIMESTAMP '2022-04-01 00:00:00' and ts < TIMESTAMP '2022-04-30 23:59:59' INTERVAL(10s) FILL(NULL)",
       TSDB_CODE_PAR_FILL_NOT_ALLOWED_FUNC);
+  run("SELECT _WSTARTTS, _WENDTS, _WDURATION, sum(c1) FROM t1", TSDB_CODE_PAR_INVALID_WINDOW_PC);
 }
 
 TEST_F(ParserSelectTest, subquery) {
@@ -231,13 +232,22 @@ TEST_F(ParserSelectTest, subquery) {
   run("SELECT SUM(a) FROM (SELECT MAX(c1) a, ts FROM st1s1 PARTITION BY TBNAME INTERVAL(1m)) INTERVAL(1n)");
 
   run("SELECT SUM(a) FROM (SELECT MAX(c1) a, _wstartts FROM st1s1 PARTITION BY TBNAME INTERVAL(1m)) INTERVAL(1n)");
+
+  run("SELECT _C0 FROM (SELECT _ROWTS, ts FROM st1s1)");
+
+  run("SELECT ts FROM (SELECT t1.ts FROM st1s1 t1)");
 }
 
 TEST_F(ParserSelectTest, subquerySemanticCheck) {
   useDb("root", "test");
 
-  run("SELECT SUM(a) FROM (SELECT MAX(c1) a FROM st1s1 INTERVAL(1m)) INTERVAL(1n)", TSDB_CODE_PAR_NOT_ALLOWED_WIN_QUERY,
+  run("SELECT SUM(a) FROM (SELECT MAX(c1) a FROM st1s1 INTERVAL(1m)) INTERVAL(1n)",
-      PARSER_STAGE_TRANSLATE);
+      TSDB_CODE_PAR_NOT_ALLOWED_WIN_QUERY);
+
+  run("SELECT ts FROM (SELECT t1.ts AS ts, t2.ts FROM st1s1 t1, st1s2 t2 WHERE t1.ts = t2.ts)",
+      TSDB_CODE_PAR_AMBIGUOUS_COLUMN);
+
+  run("SELECT ts FROM (SELECT ts AS c1 FROM st1s1 t1)", TSDB_CODE_PAR_INVALID_COLUMN);
 }
 
 TEST_F(ParserSelectTest, semanticCheck) {

source/libs/parser/test/parTestUtil.cpp

@@ -62,7 +62,7 @@ int32_t getLogLevel() { return g_logLevel; }
 
 class ParserTestBaseImpl {
  public:
-  ParserTestBaseImpl(ParserTestBase* pBase) : pBase_(pBase) {}
+  ParserTestBaseImpl(ParserTestBase* pBase) : pBase_(pBase), sqlNo_(0) {}
 
   void login(const std::string& user) { caseEnv_.user_ = user; }
 
@@ -73,6 +73,7 @@ class ParserTestBaseImpl {
   }
 
   void run(const string& sql, int32_t expect, ParserStage checkStage) {
+    ++sqlNo_;
     if (caseEnv_.nsql_ > 0) {
       --(caseEnv_.nsql_);
       return;
@@ -174,7 +175,7 @@ class ParserTestBaseImpl {
   }
 
   void dump() {
-    cout << "==========================================sql : [" << stmtEnv_.sql_ << "]" << endl;
+    cout << "========================================== " << sqlNo_ << " sql : [" << stmtEnv_.sql_ << "]" << endl;
     if (!res_.parsedAst_.empty()) {
       cout << "raw syntax tree : " << endl;
       cout << res_.parsedAst_ << endl;
@@ -425,6 +426,7 @@ class ParserTestBaseImpl {
   stmtEnv         stmtEnv_;
   stmtRes         res_;
   ParserTestBase* pBase_;
+  int32_t         sqlNo_;
 };
 
 ParserTestBase::ParserTestBase() : impl_(new ParserTestBaseImpl(this)) {}

source/libs/planner/src/planSpliter.c

@@ -15,6 +15,7 @@
 
 #include "functionMgt.h"
 #include "planInt.h"
+#include "tglobal.h"
 
 #define SPLIT_FLAG_MASK(n) (1 << n)
 
@@ -37,7 +38,8 @@ typedef struct SSplitRule {
   FSplit splitFunc;
 } SSplitRule;
 
-typedef bool (*FSplFindSplitNode)(SSplitContext* pCxt, SLogicSubplan* pSubplan, void* pInfo);
+// typedef bool (*FSplFindSplitNode)(SSplitContext* pCxt, SLogicSubplan* pSubplan, void* pInfo);
+typedef bool (*FSplFindSplitNode)(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode, void* pInfo);
 
 static void splSetSubplanVgroups(SLogicSubplan* pSubplan, SLogicNode* pNode) {
   if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pNode)) {
@@ -95,9 +97,23 @@ static int32_t splCreateExchangeNodeForSubplan(SSplitContext* pCxt, SLogicSubpla
   return code;
 }
 
+static bool splMatchByNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode, FSplFindSplitNode func,
+                           void* pInfo) {
+  if (func(pCxt, pSubplan, pNode, pInfo)) {
+    return true;
+  }
+  SNode* pChild;
+  FOREACH(pChild, pNode->pChildren) {
+    if (splMatchByNode(pCxt, pSubplan, (SLogicNode*)pChild, func, pInfo)) {
+      return true;
+    }
+  }
+  return NULL;
+}
+
 static bool splMatch(SSplitContext* pCxt, SLogicSubplan* pSubplan, int32_t flag, FSplFindSplitNode func, void* pInfo) {
   if (!SPLIT_FLAG_TEST_MASK(pSubplan->splitFlag, flag)) {
-    if (func(pCxt, pSubplan, pInfo)) {
+    if (splMatchByNode(pCxt, pSubplan, pSubplan->pNode, func, pInfo)) {
       return true;
     }
   }
@@ -110,6 +126,11 @@ static bool splMatch(SSplitContext* pCxt, SLogicSubplan* pSubplan, int32_t flag,
   return false;
 }
 
+static void splSetParent(SLogicNode* pNode) {
+  SNode* pChild = NULL;
+  FOREACH(pChild, pNode->pChildren) { ((SLogicNode*)pChild)->pParent = pNode; }
+}
+
 typedef struct SStableSplitInfo {
   SLogicNode*    pSplitNode;
   SLogicSubplan* pSubplan;
@@ -136,11 +157,21 @@ static bool stbSplHasMultiTbScan(bool streamQuery, SLogicNode* pNode) {
     return false;
   }
   SNode* pChild = nodesListGetNode(pNode->pChildren, 0);
+  if (QUERY_NODE_LOGIC_PLAN_PARTITION == nodeType(pChild)) {
+    if (1 != LIST_LENGTH(((SLogicNode*)pChild)->pChildren)) {
+      return false;
+    }
+    pChild = nodesListGetNode(((SLogicNode*)pChild)->pChildren, 0);
+  }
   return (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pChild) && stbSplIsMultiTbScan(streamQuery, (SScanLogicNode*)pChild));
 }
 
 static bool stbSplNeedSplit(bool streamQuery, SLogicNode* pNode) {
   switch (nodeType(pNode)) {
+    case QUERY_NODE_LOGIC_PLAN_SCAN:
+      return stbSplIsMultiTbScan(streamQuery, (SScanLogicNode*)pNode);
+    // case QUERY_NODE_LOGIC_PLAN_JOIN:
+    //   return !(((SJoinLogicNode*)pNode)->isSingleTableJoin);
     case QUERY_NODE_LOGIC_PLAN_AGG:
       return !stbSplHasGatherExecFunc(((SAggLogicNode*)pNode)->pAggFuncs) && stbSplHasMultiTbScan(streamQuery, pNode);
     case QUERY_NODE_LOGIC_PLAN_WINDOW: {
@@ -152,35 +183,20 @@ static bool stbSplNeedSplit(bool streamQuery, SLogicNode* pNode) {
     }
     case QUERY_NODE_LOGIC_PLAN_SORT:
       return stbSplHasMultiTbScan(streamQuery, pNode);
-    case QUERY_NODE_LOGIC_PLAN_SCAN:
-      return stbSplIsMultiTbScan(streamQuery, (SScanLogicNode*)pNode);
     default:
       break;
   }
   return false;
 }
 
-static SLogicNode* stbSplMatchByNode(bool streamQuery, SLogicNode* pNode) {
+static bool stbSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode,
-  if (stbSplNeedSplit(streamQuery, pNode)) {
+                                SStableSplitInfo* pInfo) {
-    return pNode;
+  if (stbSplNeedSplit(pCxt->pPlanCxt->streamQuery, pNode)) {
-  }
+    pInfo->pSplitNode = pNode;
-  SNode* pChild;
-  FOREACH(pChild, pNode->pChildren) {
-    SLogicNode* pSplitNode = stbSplMatchByNode(streamQuery, (SLogicNode*)pChild);
-    if (NULL != pSplitNode) {
-      return pSplitNode;
-    }
-  }
-  return NULL;
-}
-
-static bool stbSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SStableSplitInfo* pInfo) {
-  SLogicNode* pSplitNode = stbSplMatchByNode(pCxt->pPlanCxt->streamQuery, pSubplan->pNode);
-  if (NULL != pSplitNode) {
-    pInfo->pSplitNode = pSplitNode;
     pInfo->pSubplan = pSubplan;
+    return true;
   }
-  return NULL != pSplitNode;
+  return false;
 }
 
 static int32_t stbSplRewriteFuns(const SNodeList* pFuncs, SNodeList** pPartialFuncs, SNodeList** pMergeFuncs) {
@@ -258,6 +274,7 @@ static int32_t stbSplCreatePartWindowNode(SWindowLogicNode* pMergeWindow, SLogic
   if (TSDB_CODE_SUCCESS == code) {
     pMergeWindow->node.pTargets = pTargets;
     pPartWin->node.pChildren = pChildren;
+    splSetParent((SLogicNode*)pPartWin);
     code = stbSplRewriteFuns(pFunc, &pPartWin->pFuncs, &pMergeWindow->pFuncs);
   }
   int32_t index = 0;
@@ -285,13 +302,24 @@ static int32_t stbSplCreatePartWindowNode(SWindowLogicNode* pMergeWindow, SLogic
   return code;
 }
 
+static int32_t stbSplGetNumOfVgroups(SLogicNode* pNode) {
+  if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pNode)) {
+    return ((SScanLogicNode*)pNode)->pVgroupList->numOfVgroups;
+  } else {
+    if (1 == LIST_LENGTH(pNode->pChildren)) {
+      return stbSplGetNumOfVgroups((SLogicNode*)nodesListGetNode(pNode->pChildren, 0));
+    }
+  }
+  return 0;
+}
+
 static int32_t stbSplCreateMergeNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pSplitNode,
                                      SNodeList* pMergeKeys, SLogicNode* pPartChild) {
   SMergeLogicNode* pMerge = (SMergeLogicNode*)nodesMakeNode(QUERY_NODE_LOGIC_PLAN_MERGE);
   if (NULL == pMerge) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
-  pMerge->numOfChannels = ((SScanLogicNode*)nodesListGetNode(pPartChild->pChildren, 0))->pVgroupList->numOfVgroups;
+  pMerge->numOfChannels = stbSplGetNumOfVgroups(pPartChild);
   pMerge->srcGroupId = pCxt->groupId;
   pMerge->node.precision = pPartChild->precision;
   pMerge->pMergeKeys = pMergeKeys;
@@ -329,12 +357,12 @@ static int32_t stbSplCreateExchangeNode(SSplitContext* pCxt, SLogicNode* pParent
   return code;
 }
 
-static int32_t stbSplCreateMergeKeysForInterval(SNode* pWStartTs, SNodeList** pMergeKeys) {
+static int32_t stbSplCreateMergeKeysByPrimaryKey(SNode* pPrimaryKey, SNodeList** pMergeKeys) {
   SOrderByExprNode* pMergeKey = (SOrderByExprNode*)nodesMakeNode(QUERY_NODE_ORDER_BY_EXPR);
   if (NULL == pMergeKey) {
     return TSDB_CODE_OUT_OF_MEMORY;
   }
-  pMergeKey->pExpr = nodesCloneNode(pWStartTs);
+  pMergeKey->pExpr = nodesCloneNode(pPrimaryKey);
   if (NULL == pMergeKey->pExpr) {
     nodesDestroyNode((SNode*)pMergeKey);
     return TSDB_CODE_OUT_OF_MEMORY;
@@ -351,7 +379,7 @@ static int32_t stbSplSplitIntervalForBatch(SSplitContext* pCxt, SStableSplitInfo
     ((SWindowLogicNode*)pPartWindow)->intervalAlgo = INTERVAL_ALGO_HASH;
     ((SWindowLogicNode*)pInfo->pSplitNode)->intervalAlgo = INTERVAL_ALGO_MERGE;
     SNodeList* pMergeKeys = NULL;
-    code = stbSplCreateMergeKeysForInterval(((SWindowLogicNode*)pInfo->pSplitNode)->pTspk, &pMergeKeys);
+    code = stbSplCreateMergeKeysByPrimaryKey(((SWindowLogicNode*)pInfo->pSplitNode)->pTspk, &pMergeKeys);
     if (TSDB_CODE_SUCCESS == code) {
       code = stbSplCreateMergeNode(pCxt, NULL, pInfo->pSplitNode, pMergeKeys, pPartWindow);
     }
@@ -416,6 +444,8 @@ static int32_t stbSplCreatePartAggNode(SAggLogicNode* pMergeAgg, SLogicNode** pO
   pMergeAgg->node.pTargets = NULL;
   SNodeList* pChildren = pMergeAgg->node.pChildren;
   pMergeAgg->node.pChildren = NULL;
+  SNode* pConditions = pMergeAgg->node.pConditions;
+  pMergeAgg->node.pConditions = NULL;
 
   int32_t        code = TSDB_CODE_SUCCESS;
   SAggLogicNode* pPartAgg = (SAggLogicNode*)nodesCloneNode((SNode*)pMergeAgg);
@@ -434,8 +464,10 @@ static int32_t stbSplCreatePartAggNode(SAggLogicNode* pMergeAgg, SLogicNode** pO
     }
   }
   if (TSDB_CODE_SUCCESS == code) {
+    pMergeAgg->node.pConditions = pConditions;
     pMergeAgg->node.pTargets = pTargets;
     pPartAgg->node.pChildren = pChildren;
+    splSetParent((SLogicNode*)pPartAgg);
 
     code = stbSplRewriteFuns(pFunc, &pPartAgg->pAggFuncs, &pMergeAgg->pAggFuncs);
   }
@@ -550,6 +582,7 @@ static int32_t stbSplCreatePartSortNode(SSortLogicNode* pSort, SLogicNode** pOut
   SNodeList* pMergeKeys = NULL;
   if (TSDB_CODE_SUCCESS == code) {
     pPartSort->node.pChildren = pChildren;
+    splSetParent((SLogicNode*)pPartSort);
     pPartSort->pSortKeys = pSortKeys;
     code = stbSplCreateMergeKeys(pPartSort->pSortKeys, pPartSort->node.pTargets, &pMergeKeys);
   }
@@ -589,6 +622,56 @@ static int32_t stbSplSplitScanNode(SSplitContext* pCxt, SStableSplitInfo* pInfo)
   return code;
 }
 
+static SNode* stbSplFindPrimaryKeyFromScan(SScanLogicNode* pScan) {
+  SNode* pCol = NULL;
+  FOREACH(pCol, pScan->pScanCols) {
+    if (PRIMARYKEY_TIMESTAMP_COL_ID == ((SColumnNode*)pCol)->colId) {
+      return pCol;
+    }
+  }
+  return NULL;
+}
+
+static int32_t stbSplSplitScanNodeForJoin(SSplitContext* pCxt, SLogicSubplan* pSubplan, SScanLogicNode* pScan) {
+  SNodeList* pMergeKeys = NULL;
+  int32_t    code = stbSplCreateMergeKeysByPrimaryKey(stbSplFindPrimaryKeyFromScan(pScan), &pMergeKeys);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = stbSplCreateMergeNode(pCxt, pSubplan, (SLogicNode*)pScan, pMergeKeys, (SLogicNode*)pScan);
+  }
+  if (TSDB_CODE_SUCCESS == code) {
+    code = nodesListMakeStrictAppend(&pSubplan->pChildren,
+                                     (SNode*)splCreateScanSubplan(pCxt, (SLogicNode*)pScan, SPLIT_FLAG_STABLE_SPLIT));
+  }
+  return code;
+}
+
+static int32_t stbSplSplitJoinNodeImpl(SSplitContext* pCxt, SLogicSubplan* pSubplan, SJoinLogicNode* pJoin) {
+  int32_t code = TSDB_CODE_SUCCESS;
+  SNode*  pChild = NULL;
+  FOREACH(pChild, pJoin->node.pChildren) {
+    if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pChild)) {
+      code = stbSplSplitScanNodeForJoin(pCxt, pSubplan, (SScanLogicNode*)pChild);
+    } else if (QUERY_NODE_LOGIC_PLAN_JOIN == nodeType(pChild)) {
+      code = stbSplSplitJoinNodeImpl(pCxt, pSubplan, (SJoinLogicNode*)pChild);
+    } else {
+      code = TSDB_CODE_PLAN_INTERNAL_ERROR;
+    }
+    if (TSDB_CODE_SUCCESS != code) {
+      break;
+    }
+  }
+  return code;
+}
+
+static int32_t stbSplSplitJoinNode(SSplitContext* pCxt, SStableSplitInfo* pInfo) {
+  int32_t code = stbSplSplitJoinNodeImpl(pCxt, pInfo->pSubplan, (SJoinLogicNode*)pInfo->pSplitNode);
+  if (TSDB_CODE_SUCCESS == code) {
+    pInfo->pSubplan->subplanType = SUBPLAN_TYPE_MERGE;
+    SPLIT_FLAG_SET_MASK(pInfo->pSubplan->splitFlag, SPLIT_FLAG_STABLE_SPLIT);
+  }
+  return code;
+}
+
 static int32_t stableSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
   if (pCxt->pPlanCxt->rSmaQuery) {
     return TSDB_CODE_SUCCESS;
@@ -601,6 +684,12 @@ static int32_t stableSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
 
   int32_t code = TSDB_CODE_SUCCESS;
   switch (nodeType(info.pSplitNode)) {
+    case QUERY_NODE_LOGIC_PLAN_SCAN:
+      code = stbSplSplitScanNode(pCxt, &info);
+      break;
+    case QUERY_NODE_LOGIC_PLAN_JOIN:
+      code = stbSplSplitJoinNode(pCxt, &info);
+      break;
     case QUERY_NODE_LOGIC_PLAN_AGG:
       code = stbSplSplitAggNode(pCxt, &info);
       break;
@@ -610,9 +699,6 @@ static int32_t stableSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
     case QUERY_NODE_LOGIC_PLAN_SORT:
       code = stbSplSplitSortNode(pCxt, &info);
       break;
-    case QUERY_NODE_LOGIC_PLAN_SCAN:
-      code = stbSplSplitScanNode(pCxt, &info);
-      break;
     default:
       break;
   }
@@ -628,7 +714,12 @@ typedef struct SSigTbJoinSplitInfo {
   SLogicSubplan*  pSubplan;
 } SSigTbJoinSplitInfo;
 
-static bool sigTbJoinSplNeedSplit(SJoinLogicNode* pJoin) {
+static bool sigTbJoinSplNeedSplit(SLogicNode* pNode) {
+  if (QUERY_NODE_LOGIC_PLAN_JOIN != nodeType(pNode)) {
+    return false;
+  }
+
+  SJoinLogicNode* pJoin = (SJoinLogicNode*)pNode;
   if (!pJoin->isSingleTableJoin) {
     return false;
   }
@@ -636,28 +727,15 @@ static bool sigTbJoinSplNeedSplit(SJoinLogicNode* pJoin) {
          QUERY_NODE_LOGIC_PLAN_EXCHANGE != nodeType(nodesListGetNode(pJoin->node.pChildren, 1));
 }
 
-static SJoinLogicNode* sigTbJoinSplMatchByNode(SLogicNode* pNode) {
+static bool sigTbJoinSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode,
-  if (QUERY_NODE_LOGIC_PLAN_JOIN == nodeType(pNode) && sigTbJoinSplNeedSplit((SJoinLogicNode*)pNode)) {
+                                      SSigTbJoinSplitInfo* pInfo) {
-    return (SJoinLogicNode*)pNode;
+  if (sigTbJoinSplNeedSplit(pNode)) {
-  }
+    pInfo->pJoin = (SJoinLogicNode*)pNode;
-  SNode* pChild;
+    pInfo->pSplitNode = (SLogicNode*)nodesListGetNode(pNode->pChildren, 1);
-  FOREACH(pChild, pNode->pChildren) {
-    SJoinLogicNode* pSplitNode = sigTbJoinSplMatchByNode((SLogicNode*)pChild);
-    if (NULL != pSplitNode) {
-      return pSplitNode;
-    }
-  }
-  return NULL;
-}
-
-static bool sigTbJoinSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SSigTbJoinSplitInfo* pInfo) {
-  SJoinLogicNode* pJoin = sigTbJoinSplMatchByNode(pSubplan->pNode);
-  if (NULL != pJoin) {
-    pInfo->pJoin = pJoin;
-    pInfo->pSplitNode = (SLogicNode*)nodesListGetNode(pJoin->node.pChildren, 1);
     pInfo->pSubplan = pSubplan;
+    return true;
   }
-  return NULL != pJoin;
+  return false;
 }
 
 static int32_t singleTableJoinSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
@@ -750,27 +828,14 @@ typedef struct SUnionAllSplitInfo {
   SLogicSubplan*     pSubplan;
 } SUnionAllSplitInfo;
 
-static SLogicNode* unAllSplMatchByNode(SLogicNode* pNode) {
+static bool unAllSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode,
+                                  SUnionAllSplitInfo* pInfo) {
   if (QUERY_NODE_LOGIC_PLAN_PROJECT == nodeType(pNode) && LIST_LENGTH(pNode->pChildren) > 1) {
-    return pNode;
+    pInfo->pProject = (SProjectLogicNode*)pNode;
-  }
-  SNode* pChild;
-  FOREACH(pChild, pNode->pChildren) {
-    SLogicNode* pSplitNode = unAllSplMatchByNode((SLogicNode*)pChild);
-    if (NULL != pSplitNode) {
-      return pSplitNode;
-    }
-  }
-  return NULL;
-}
-
-static bool unAllSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SUnionAllSplitInfo* pInfo) {
-  SLogicNode* pSplitNode = unAllSplMatchByNode(pSubplan->pNode);
-  if (NULL != pSplitNode) {
-    pInfo->pProject = (SProjectLogicNode*)pSplitNode;
     pInfo->pSubplan = pSubplan;
+    return true;
   }
-  return NULL != pSplitNode;
+  return false;
 }
 
 static int32_t unAllSplCreateExchangeNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SProjectLogicNode* pProject) {
@@ -825,20 +890,6 @@ typedef struct SUnionDistinctSplitInfo {
   SLogicSubplan* pSubplan;
 } SUnionDistinctSplitInfo;
 
-static SLogicNode* unDistSplMatchByNode(SLogicNode* pNode) {
-  if (QUERY_NODE_LOGIC_PLAN_AGG == nodeType(pNode) && LIST_LENGTH(pNode->pChildren) > 1) {
-    return pNode;
-  }
-  SNode* pChild;
-  FOREACH(pChild, pNode->pChildren) {
-    SLogicNode* pSplitNode = unDistSplMatchByNode((SLogicNode*)pChild);
-    if (NULL != pSplitNode) {
-      return pSplitNode;
-    }
-  }
-  return NULL;
-}
-
 static int32_t unDistSplCreateExchangeNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SAggLogicNode* pAgg) {
   SExchangeLogicNode* pExchange = (SExchangeLogicNode*)nodesMakeNode(QUERY_NODE_LOGIC_PLAN_EXCHANGE);
   if (NULL == pExchange) {
@@ -856,13 +907,14 @@ static int32_t unDistSplCreateExchangeNode(SSplitContext* pCxt, SLogicSubplan* p
   return nodesListMakeAppend(&pAgg->node.pChildren, (SNode*)pExchange);
 }
 
-static bool unDistSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SUnionDistinctSplitInfo* pInfo) {
+static bool unDistSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode,
-  SLogicNode* pSplitNode = unDistSplMatchByNode(pSubplan->pNode);
+                                   SUnionDistinctSplitInfo* pInfo) {
-  if (NULL != pSplitNode) {
+  if (QUERY_NODE_LOGIC_PLAN_AGG == nodeType(pNode) && LIST_LENGTH(pNode->pChildren) > 1) {
-    pInfo->pAgg = (SAggLogicNode*)pSplitNode;
+    pInfo->pAgg = (SAggLogicNode*)pNode;
     pInfo->pSubplan = pSubplan;
+    return true;
   }
-  return NULL != pSplitNode;
+  return false;
 }
 
 static int32_t unionDistinctSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
@@ -885,27 +937,14 @@ typedef struct SSmaIndexSplitInfo {
   SLogicSubplan*   pSubplan;
 } SSmaIndexSplitInfo;
 
-static SLogicNode* smaIdxSplMatchByNode(SLogicNode* pNode) {
+static bool smaIdxSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode,
+                                   SSmaIndexSplitInfo* pInfo) {
   if (QUERY_NODE_LOGIC_PLAN_MERGE == nodeType(pNode) && LIST_LENGTH(pNode->pChildren) > 1) {
-    return pNode;
+    pInfo->pMerge = (SMergeLogicNode*)pNode;
-  }
-  SNode* pChild;
-  FOREACH(pChild, pNode->pChildren) {
-    SLogicNode* pSplitNode = smaIdxSplMatchByNode((SLogicNode*)pChild);
-    if (NULL != pSplitNode) {
-      return pSplitNode;
-    }
-  }
-  return NULL;
-}
-
-static bool smaIdxSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SSmaIndexSplitInfo* pInfo) {
-  SLogicNode* pSplitNode = smaIdxSplMatchByNode(pSubplan->pNode);
-  if (NULL != pSplitNode) {
-    pInfo->pMerge = (SMergeLogicNode*)pSplitNode;
     pInfo->pSubplan = pSubplan;
+    return true;
   }
-  return NULL != pSplitNode;
+  return false;
 }
 
 static int32_t smaIndexSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
@@ -923,13 +962,47 @@ static int32_t smaIndexSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
   return code;
 }
 
+typedef struct SQnodeSplitInfo {
+  SLogicNode*    pSplitNode;
+  SLogicSubplan* pSubplan;
+} SQnodeSplitInfo;
+
+static bool qndSplFindSplitNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pNode,
+                                SQnodeSplitInfo* pInfo) {
+  if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pNode) && NULL != pNode->pParent) {
+    pInfo->pSplitNode = pNode;
+    pInfo->pSubplan = pSubplan;
+    return true;
+  }
+  return false;
+}
+
+static int32_t qnodeSplit(SSplitContext* pCxt, SLogicSubplan* pSubplan) {
+  if (QUERY_POLICY_QNODE != tsQueryPolicy) {
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SQnodeSplitInfo info = {0};
+  if (!splMatch(pCxt, pSubplan, 0, (FSplFindSplitNode)qndSplFindSplitNode, &info)) {
+    return TSDB_CODE_SUCCESS;
+  }
+  int32_t code = splCreateExchangeNodeForSubplan(pCxt, info.pSubplan, info.pSplitNode, info.pSubplan->subplanType);
+  if (TSDB_CODE_SUCCESS == code) {
+    code = nodesListMakeStrictAppend(&info.pSubplan->pChildren, (SNode*)splCreateScanSubplan(pCxt, info.pSplitNode, 0));
+  }
+  ++(pCxt->groupId);
+  pCxt->split = true;
+  return code;
+}
+
 // clang-format off
 static const SSplitRule splitRuleSet[] = {
   {.pName = "SuperTableSplit",      .splitFunc = stableSplit},
   {.pName = "SingleTableJoinSplit", .splitFunc = singleTableJoinSplit},
   {.pName = "UnionAllSplit",        .splitFunc = unionAllSplit},
   {.pName = "UnionDistinctSplit",   .splitFunc = unionDistinctSplit},
-  {.pName = "SmaIndexSplit",        .splitFunc = smaIndexSplit}
+  {.pName = "SmaIndexSplit",        .splitFunc = smaIndexSplit},
+  {.pName = "QnodeSplit",           .splitFunc = qnodeSplit}
 };
 // clang-format on
 

source/libs/planner/test/planGroupByTest.cpp

@@ -82,4 +82,8 @@ TEST_F(PlanGroupByTest, stable) {
   run("SELECT COUNT(*) FROM st1");
 
   run("SELECT COUNT(*) FROM st1 GROUP BY c1");
+
+  run("SELECT COUNT(*) FROM st1 PARTITION BY c2 GROUP BY c1");
+
+  run("SELECT SUM(c1) FROM st1 GROUP BY c2 HAVING SUM(c1) IS NOT NULL");
 }

source/libs/planner/test/planIntervalTest.cpp

@@ -60,4 +60,6 @@ TEST_F(PlanIntervalTest, stable) {
   run("SELECT COUNT(*) FROM st1 INTERVAL(10s)");
 
   run("SELECT _WSTARTTS, COUNT(*) FROM st1 INTERVAL(10s)");
+
+  run("SELECT _WSTARTTS, COUNT(*) FROM st1 PARTITION BY TBNAME INTERVAL(10s)");
 }

source/libs/planner/test/planJoinTest.cpp

@@ -50,3 +50,9 @@ TEST_F(PlanJoinTest, multiJoin) {
 
   run("SELECT t1.c1, t2.c1 FROM st1s1 t1 JOIN st1s2 t2 ON t1.ts = t2.ts JOIN st1s3 t3 ON t1.ts = t3.ts");
 }
+
+TEST_F(PlanJoinTest, stable) {
+  useDb("root", "test");
+
+  run("SELECT t1.c1, t2.c1 FROM st1 t1 JOIN st2 t2 ON t1.ts = t2.ts ");
+}

source/libs/planner/test/planOrderByTest.cpp

@@ -49,4 +49,6 @@ TEST_F(PlanOrderByTest, stable) {
 
   // ORDER BY key is not in the projection list
   run("SELECT c2 FROM st1 ORDER BY c1");
+
+  run("SELECT c2 FROM st1 PARTITION BY c2 ORDER BY c1");
 }

source/libs/planner/test/planOtherTest.cpp

@@ -83,3 +83,10 @@ TEST_F(PlanOtherTest, delete) {
 
   run("DELETE FROM st1 WHERE ts > now - 2d and ts < now - 1d AND tag1 = 10");
 }
+
+TEST_F(PlanOtherTest, queryPolicy) {
+  useDb("root", "test");
+
+  tsQueryPolicy = QUERY_POLICY_QNODE;
+  run("SELECT COUNT(*) FROM st1");
+}

source/libs/planner/test/planTestUtil.h

@@ -18,6 +18,10 @@
 
 #include <gtest/gtest.h>
 
+#define ALLOW_FORBID_FUNC
+
+#include "planInt.h"
+
 class PlannerTestBaseImpl;
 struct TAOS_MULTI_BIND;
 

source/libs/qcom/src/querymsg.c

@@ -484,13 +484,11 @@ int32_t queryProcessGetTbIndexRsp(void *output, char *msg, int32_t msgSize) {
     return TSDB_CODE_TSC_INVALID_INPUT;
   }
 
-  STableIndexRsp out = {0};
+  STableIndexRsp *out = (STableIndexRsp*)output;
-  if (tDeserializeSTableIndexRsp(msg, msgSize, &out) != 0) {
+  if (tDeserializeSTableIndexRsp(msg, msgSize, out) != 0) {
     qError("tDeserializeSTableIndexRsp failed, msgSize:%d", msgSize);
     return TSDB_CODE_INVALID_MSG;
   }
-
-  *(void **)output = out.pIndex;
   
   return TSDB_CODE_SUCCESS;
 }

source/libs/qworker/src/qworker.c

@@ -837,7 +837,6 @@ void qwProcessHbTimerEvent(void *param, void *tmrId) {
   SQWorker *mgmt = qwAcquire(refId);
   if (NULL == mgmt) {
     QW_DLOG("qwAcquire %" PRIx64 "failed", refId);
-    taosMemoryFree(param);
     return;
   }
 

source/libs/scalar/src/filter.c

@@ -1476,6 +1476,11 @@ void filterDumpInfoToString(SFilterInfo *info, const char *msg, int32_t options)
       for (uint32_t i = 0; i < info->fields[FLD_TYPE_VALUE].num; ++i) {
         SFilterField *field = &info->fields[FLD_TYPE_VALUE].fields[i];
         if (field->desc) {
+          if (QUERY_NODE_VALUE != nodeType(field->desc)) {
+            qDebug("VAL%d => [type:not value node][val:NIL]", i); //TODO
+            continue;
+          }
+
           SValueNode *var = (SValueNode *)field->desc;
           SDataType *dType = &var->node.resType;
           if (dType->type == TSDB_DATA_TYPE_VALUE_ARRAY) {

source/libs/sync/inc/syncInt.h

@@ -168,6 +168,7 @@ SSyncNode* syncNodeOpen(const SSyncInfo* pSyncInfo);
 void       syncNodeStart(SSyncNode* pSyncNode);
 void       syncNodeStartStandBy(SSyncNode* pSyncNode);
 void       syncNodeClose(SSyncNode* pSyncNode);
+int32_t    syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak);
 
 // option
 bool syncNodeSnapshotEnable(SSyncNode* pSyncNode);
@@ -232,6 +233,9 @@ int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex,
 bool                 syncNodeInRaftGroup(SSyncNode* ths, SRaftId* pRaftId);
 SSyncSnapshotSender* syncNodeGetSnapshotSender(SSyncNode* ths, SRaftId* pDestId);
 
+void syncStartNormal(int64_t rid);
+void syncStartStandBy(int64_t rid);
+
 // for debug --------------
 void syncNodePrint(SSyncNode* pObj);
 void syncNodePrint2(char* s, SSyncNode* pObj);

source/libs/sync/src/syncAppendEntries.c

@@ -995,7 +995,7 @@ int32_t syncNodeOnAppendEntriesSnapshotCb(SSyncNode* ths, SyncAppendEntries* pMs
             ths->commitIndex = snapshot.lastApplyIndex;
 
             sDebug("vgId:%d sync event commit by snapshot from index:%ld to index:%ld, %s", ths->vgId, commitBegin,
-                  commitEnd, syncUtilState2String(ths->state));
+                   commitEnd, syncUtilState2String(ths->state));
           }
 
           SyncIndex beginIndex = ths->commitIndex + 1;

source/libs/sync/src/syncCommit.c

@@ -57,7 +57,7 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) {
     pSyncNode->commitIndex = snapshot.lastApplyIndex;
 
     sDebug("vgId:%d sync event commit by snapshot from index:%ld to index:%ld, %s", pSyncNode->vgId,
-          pSyncNode->commitIndex, snapshot.lastApplyIndex, syncUtilState2String(pSyncNode->state));
+           pSyncNode->commitIndex, snapshot.lastApplyIndex, syncUtilState2String(pSyncNode->state));
   }
 
   // update commit index

source/libs/sync/src/syncMain.c

@@ -149,12 +149,12 @@ void syncStop(int64_t rid) {
 int32_t syncSetStandby(int64_t rid) {
   SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
   if (pSyncNode == NULL) {
-    return -1;
+    return TAOS_SYNC_OTHER_ERROR;
   }
 
   if (pSyncNode->state == TAOS_SYNC_STATE_LEADER) {
     taosReleaseRef(tsNodeRefId, pSyncNode->rid);
-    return -1;
+    return TAOS_SYNC_OTHER_ERROR;
   }
 
   // state change
@@ -174,14 +174,88 @@ int32_t syncSetStandby(int64_t rid) {
   return 0;
 }
 
-int32_t syncReconfig(int64_t rid, const SSyncCfg* pSyncCfg) {
+int32_t syncReconfigBuild(int64_t rid, const SSyncCfg* pNewCfg, SRpcMsg* pRpcMsg) {
-  int32_t ret = 0;
+  SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
-  char*   newconfig = syncCfg2Str((SSyncCfg*)pSyncCfg);
+  if (pSyncNode == NULL) {
+    return TAOS_SYNC_OTHER_ERROR;
+  }
+  ASSERT(rid == pSyncNode->rid);
 
+  int32_t ret = 0;
+  bool    IamInNew = false;
+  for (int i = 0; i < pNewCfg->replicaNum; ++i) {
+    if (strcmp((pNewCfg->nodeInfo)[i].nodeFqdn, pSyncNode->myNodeInfo.nodeFqdn) == 0 &&
+        (pNewCfg->nodeInfo)[i].nodePort == pSyncNode->myNodeInfo.nodePort) {
+      IamInNew = true;
+    }
+
+    /*
+        SRaftId newId;
+        newId.addr = syncUtilAddr2U64((pNewCfg->nodeInfo)[i].nodeFqdn, (pNewCfg->nodeInfo)[i].nodePort);
+        newId.vgId = pSyncNode->vgId;
+        if (syncUtilSameId(&(pSyncNode->myRaftId), &newId)) {
+          IamInNew = true;
+        }
+    */
+  }
+
+  if (!IamInNew) {
+    taosReleaseRef(tsNodeRefId, pSyncNode->rid);
+    return TAOS_SYNC_NOT_IN_NEW_CONFIG;
+  }
+
+  char* newconfig = syncCfg2Str((SSyncCfg*)pNewCfg);
+  pRpcMsg->msgType = TDMT_SYNC_CONFIG_CHANGE;
+  pRpcMsg->info.noResp = 1;
+  pRpcMsg->contLen = strlen(newconfig) + 1;
+  pRpcMsg->pCont = rpcMallocCont(pRpcMsg->contLen);
+  snprintf(pRpcMsg->pCont, pRpcMsg->contLen, "%s", newconfig);
+  taosMemoryFree(newconfig);
+
+  taosReleaseRef(tsNodeRefId, pSyncNode->rid);
+  return ret;
+}
+
+int32_t syncReconfig(int64_t rid, const SSyncCfg* pNewCfg) {
+  SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
+  if (pSyncNode == NULL) {
+    return TAOS_SYNC_OTHER_ERROR;
+  }
+  ASSERT(rid == pSyncNode->rid);
+
+  bool IamInNew = false;
+  for (int i = 0; i < pNewCfg->replicaNum; ++i) {
+    if (strcmp((pNewCfg->nodeInfo)[i].nodeFqdn, pSyncNode->myNodeInfo.nodeFqdn) == 0 &&
+        (pNewCfg->nodeInfo)[i].nodePort == pSyncNode->myNodeInfo.nodePort) {
+      IamInNew = true;
+    }
+
+    /*
+        // some problem in inet_addr
+
+        SRaftId newId = EMPTY_RAFT_ID;
+        newId.addr = syncUtilAddr2U64((pNewCfg->nodeInfo)[i].nodeFqdn, (pNewCfg->nodeInfo)[i].nodePort);
+        newId.vgId = pSyncNode->vgId;
+
+        if (syncUtilSameId(&(pSyncNode->myRaftId), &newId)) {
+          IamInNew = true;
+        }
+      */
+  }
+
+  if (!IamInNew) {
+    sError("sync reconfig error, not in new config");
+    taosReleaseRef(tsNodeRefId, pSyncNode->rid);
+    return TAOS_SYNC_NOT_IN_NEW_CONFIG;
+  }
+
+  char* newconfig = syncCfg2Str((SSyncCfg*)pNewCfg);
   if (gRaftDetailLog) {
     sInfo("==syncReconfig== newconfig:%s", newconfig);
   }
 
+  int32_t ret = 0;
+
   SRpcMsg rpcMsg = {0};
   rpcMsg.msgType = TDMT_SYNC_CONFIG_CHANGE;
   rpcMsg.info.noResp = 1;
@@ -189,58 +263,59 @@ int32_t syncReconfig(int64_t rid, const SSyncCfg* pSyncCfg) {
   rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen);
   snprintf(rpcMsg.pCont, rpcMsg.contLen, "%s", newconfig);
   taosMemoryFree(newconfig);
-  ret = syncPropose(rid, &rpcMsg, false);
+  ret = syncNodePropose(pSyncNode, &rpcMsg, false);
+
+  taosReleaseRef(tsNodeRefId, pSyncNode->rid);
   return ret;
 }
 
 int32_t syncLeaderTransfer(int64_t rid) {
-  int32_t ret = 0;
+  SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
+  if (pSyncNode == NULL) {
+    return TAOS_SYNC_OTHER_ERROR;
+  }
+  ASSERT(rid == pSyncNode->rid);
 
+  if (pSyncNode->peersNum == 0) {
+    taosReleaseRef(tsNodeRefId, pSyncNode->rid);
+    return TAOS_SYNC_OTHER_ERROR;
+  }
+
+  SNodeInfo newLeader = (pSyncNode->peersNodeInfo)[0];
+  taosReleaseRef(tsNodeRefId, pSyncNode->rid);
+
+  int32_t ret = syncLeaderTransferTo(rid, newLeader);
   return ret;
 }
 
 int32_t syncLeaderTransferTo(int64_t rid, SNodeInfo newLeader) {
   SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
   if (pSyncNode == NULL) {
-    return false;
+    return TAOS_SYNC_OTHER_ERROR;
   }
-  assert(rid == pSyncNode->rid);
+  ASSERT(rid == pSyncNode->rid);
   int32_t ret = 0;
 
   if (pSyncNode->replicaNum == 1) {
-    taosReleaseRef(tsNodeRefId, pSyncNode->rid);
     sError("only one replica, cannot drop leader");
+    taosReleaseRef(tsNodeRefId, pSyncNode->rid);
     return TAOS_SYNC_ONLY_ONE_REPLICA;
   }
 
   SyncLeaderTransfer* pMsg = syncLeaderTransferBuild(pSyncNode->vgId);
   pMsg->newLeaderId.addr = syncUtilAddr2U64(newLeader.nodeFqdn, newLeader.nodePort);
   pMsg->newLeaderId.vgId = pSyncNode->vgId;
+  pMsg->newNodeInfo = newLeader;
   ASSERT(pMsg != NULL);
   SRpcMsg rpcMsg = {0};
   syncLeaderTransfer2RpcMsg(pMsg, &rpcMsg);
   syncLeaderTransferDestroy(pMsg);
 
-  ret = syncPropose(rid, &rpcMsg, false);
+  ret = syncNodePropose(pSyncNode, &rpcMsg, false);
-
   taosReleaseRef(tsNodeRefId, pSyncNode->rid);
   return ret;
 }
 
-int32_t syncReconfigRaw(int64_t rid, const SSyncCfg* pNewCfg, SRpcMsg* pRpcMsg) {
-  int32_t ret = 0;
-  char*   newconfig = syncCfg2Str((SSyncCfg*)pNewCfg);
-
-  pRpcMsg->msgType = TDMT_SYNC_CONFIG_CHANGE;
-  pRpcMsg->info.noResp = 1;
-  pRpcMsg->contLen = strlen(newconfig) + 1;
-  pRpcMsg->pCont = rpcMallocCont(pRpcMsg->contLen);
-  snprintf(pRpcMsg->pCont, pRpcMsg->contLen, "%s", newconfig);
-  taosMemoryFree(newconfig);
-
-  return ret;
-}
-
 bool syncCanLeaderTransfer(int64_t rid) {
   SSyncNode* pSyncNode = (SSyncNode*)taosAcquireRef(tsNodeRefId, rid);
   if (pSyncNode == NULL) {
@@ -273,8 +348,6 @@ bool syncCanLeaderTransfer(int64_t rid) {
   return matchOK;
 }
 
-int32_t syncGiveUpLeader(int64_t rid) { return 0; }
-
 int32_t syncForwardToPeer(int64_t rid, const SRpcMsg* pMsg, bool isWeak) {
   int32_t ret = syncPropose(rid, pMsg, isWeak);
   return ret;
@@ -469,17 +542,25 @@ int32_t syncPropose(int64_t rid, const SRpcMsg* pMsg, bool isWeak) {
 
   SSyncNode* pSyncNode = taosAcquireRef(tsNodeRefId, rid);
   if (pSyncNode == NULL) {
-    return TAOS_SYNC_PROPOSE_OTHER_ERROR;
+    return TAOS_SYNC_OTHER_ERROR;
   }
   assert(rid == pSyncNode->rid);
+  sDebug("vgId:%d sync event propose msgType:%s", pSyncNode->vgId, TMSG_INFO(pMsg->msgType));
+  ret = syncNodePropose(pSyncNode, pMsg, isWeak);
+
+  taosReleaseRef(tsNodeRefId, pSyncNode->rid);
+  return ret;
+}
+
+int32_t syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak) {
+  int32_t ret = TAOS_SYNC_PROPOSE_SUCCESS;
+  sDebug("vgId:%d sync event propose msgType:%s", pSyncNode->vgId, TMSG_INFO(pMsg->msgType));
 
   if (pSyncNode->state == TAOS_SYNC_STATE_LEADER) {
     SRespStub stub;
     stub.createTime = taosGetTimestampMs();
     stub.rpcMsg = *pMsg;
     uint64_t seqNum = syncRespMgrAdd(pSyncNode->pSyncRespMgr, &stub);
-    sDebug("vgId:%d, sync event propose, type:%s seq:%" PRIu64 " handle:%p", pSyncNode->vgId, TMSG_INFO(pMsg->msgType),
-           seqNum, pMsg->info.handle);
 
     SyncClientRequest* pSyncMsg = syncClientRequestBuild2(pMsg, seqNum, isWeak, pSyncNode->vgId);
     SRpcMsg            rpcMsg;
@@ -488,16 +569,14 @@ int32_t syncPropose(int64_t rid, const SRpcMsg* pMsg, bool isWeak) {
     if (pSyncNode->FpEqMsg != NULL && (*pSyncNode->FpEqMsg)(pSyncNode->msgcb, &rpcMsg) == 0) {
       ret = TAOS_SYNC_PROPOSE_SUCCESS;
     } else {
-      sTrace("syncPropose pSyncNode->FpEqMsg is NULL");
+      sError("syncPropose pSyncNode->FpEqMsg is NULL");
     }
     syncClientRequestDestroy(pSyncMsg);
   } else {
-    sDebug("vgId:%d, failed to propose since not leader, type:%s handle:%p %s", pSyncNode->vgId,
+    sError("syncPropose not leader, %s", syncUtilState2String(pSyncNode->state));
-           TMSG_INFO(pMsg->msgType), pMsg->info.handle, syncUtilState2String(pSyncNode->state));
     ret = TAOS_SYNC_PROPOSE_NOT_LEADER;
   }
 
-  taosReleaseRef(tsNodeRefId, pSyncNode->rid);
   return ret;
 }
 
@@ -690,7 +769,7 @@ SSyncNode* syncNodeOpen(const SSyncInfo* pOldSyncInfo) {
   }
 
   // tools
-  pSyncNode->pSyncRespMgr = syncRespMgrCreate(NULL, 0);
+  pSyncNode->pSyncRespMgr = syncRespMgrCreate(pSyncNode, 0);
   assert(pSyncNode->pSyncRespMgr != NULL);
 
   // restore state
@@ -1162,8 +1241,14 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* newConfig, SyncIndex l
   int32_t oldReplicaNum = pSyncNode->replicaNum;
   SRaftId oldReplicasId[TSDB_MAX_REPLICA];
   memcpy(oldReplicasId, pSyncNode->replicasId, sizeof(oldReplicasId));
-  SSyncSnapshotSender*   oldSenders[TSDB_MAX_REPLICA];
+  SSyncSnapshotSender* oldSenders[TSDB_MAX_REPLICA];
-  memcpy(oldSenders, pSyncNode->senders, sizeof(oldSenders));
+  for (int i = 0; i < TSDB_MAX_REPLICA; ++i) {
+    oldSenders[i] = (pSyncNode->senders)[i];
+    sDebug("vgId:%d sync event save senders %d, %p", pSyncNode->vgId, i, oldSenders[i]);
+    if (gRaftDetailLog) {
+      ;
+    }
+  }
 
   // init internal
   pSyncNode->myNodeInfo = pSyncNode->pRaftCfg->cfg.nodeInfo[pSyncNode->pRaftCfg->cfg.myIndex];
@@ -1195,24 +1280,51 @@ void syncNodeUpdateConfig(SSyncNode* pSyncNode, SSyncCfg* newConfig, SyncIndex l
 
   pSyncNode->quorum = syncUtilQuorum(pSyncNode->pRaftCfg->cfg.replicaNum);
 
-  // reset snapshot senders, memory leak
+  // reset snapshot senders
+
+  // clear new
   for (int i = 0; i < TSDB_MAX_REPLICA; ++i) {
     (pSyncNode->senders)[i] = NULL;
   }
+
+  // reset new
   for (int i = 0; i < pSyncNode->replicaNum; ++i) {
+    // reset sender
+    bool reset = false;
     for (int j = 0; j < TSDB_MAX_REPLICA; ++j) {
       if (syncUtilSameId(&(pSyncNode->replicasId)[i], &oldReplicasId[j])) {
-        char host[128];
+        char     host[128];
         uint16_t port;
         syncUtilU642Addr((pSyncNode->replicasId)[i].addr, host, sizeof(host), &port);
-        sDebug("vgId:%d sync event reset sender for %lu, %s:%d", pSyncNode->vgId, (pSyncNode->replicasId)[i].addr, host, port);
+        sDebug("vgId:%d sync event reset sender for %lu, newIndex:%d, %s:%d, %p", pSyncNode->vgId,
+               (pSyncNode->replicasId)[i].addr, i, host, port, oldSenders[j]);
         (pSyncNode->senders)[i] = oldSenders[j];
+        oldSenders[j] = NULL;
+        reset = true;
+
+        // reset replicaIndex
+        int32_t oldreplicaIndex = (pSyncNode->senders)[i]->replicaIndex;
+        (pSyncNode->senders)[i]->replicaIndex = i;
+        sDebug("vgId:%d sync event udpate replicaIndex from %d to %d, %s:%d, %p, reset:%d", pSyncNode->vgId,
+               oldreplicaIndex, i, host, port, (pSyncNode->senders)[i], reset);
       }
     }
   }
+
+  // create new
   for (int i = 0; i < TSDB_MAX_REPLICA; ++i) {
     if ((pSyncNode->senders)[i] == NULL) {
       (pSyncNode->senders)[i] = snapshotSenderCreate(pSyncNode, i);
+      sDebug("vgId:%d sync event create new sender %p replicaIndex:%d", pSyncNode->vgId, (pSyncNode->senders)[i], i);
+    }
+  }
+
+  // free old
+  for (int i = 0; i < TSDB_MAX_REPLICA; ++i) {
+    if (oldSenders[i] != NULL) {
+      snapshotSenderDestroy(oldSenders[i]);
+      sDebug("vgId:%d sync event delete old sender %p replicaIndex:%d", pSyncNode->vgId, oldSenders[i], i);
+      oldSenders[i] = NULL;
     }
   }
 
@@ -1272,8 +1384,8 @@ void syncNodeUpdateTerm(SSyncNode* pSyncNode, SyncTerm term) {
 }
 
 void syncNodeBecomeFollower(SSyncNode* pSyncNode, const char* debugStr) {
-  sDebug("vgId:%d sync event become follower, isStandBy:%d, %s", pSyncNode->vgId, pSyncNode->pRaftCfg->isStandBy,
+  sDebug("vgId:%d sync event become follower, isStandBy:%d, replicaNum:%d, %s", pSyncNode->vgId,
-         debugStr);
+         pSyncNode->pRaftCfg->isStandBy, pSyncNode->replicaNum, debugStr);
 
   // maybe clear leader cache
   if (pSyncNode->state == TAOS_SYNC_STATE_LEADER) {
@@ -1307,8 +1419,8 @@ void syncNodeBecomeFollower(SSyncNode* pSyncNode, const char* debugStr) {
 //     /\ UNCHANGED <<messages, currentTerm, votedFor, candidateVars, logVars>>
 //
 void syncNodeBecomeLeader(SSyncNode* pSyncNode, const char* debugStr) {
-  sDebug("vgId:%d sync event become leader, isStandBy:%d, %s", pSyncNode->vgId, pSyncNode->pRaftCfg->isStandBy,
+  sDebug("vgId:%d sync event become leader, isStandBy:%d, replicaNum:%d %s", pSyncNode->vgId,
-         debugStr);
+         pSyncNode->pRaftCfg->isStandBy, pSyncNode->replicaNum, debugStr);
 
   // state change
   pSyncNode->state = TAOS_SYNC_STATE_LEADER;
@@ -1857,10 +1969,52 @@ const char* syncStr(ESyncState state) {
 }
 
 static int32_t syncDoLeaderTransfer(SSyncNode* ths, SRpcMsg* pRpcMsg, SSyncRaftEntry* pEntry) {
-  SyncLeaderTransfer* pSyncLeaderTransfer;
+  SyncLeaderTransfer* pSyncLeaderTransfer = syncLeaderTransferFromRpcMsg2(pRpcMsg);
-  if (syncUtilSameId(&(pSyncLeaderTransfer->newLeaderId), &(ths->myRaftId))) {
+
+  /*
+    char     host[128];
+    uint16_t port;
+    syncUtilU642Addr(pSyncLeaderTransfer->newLeaderId.addr, host, sizeof(host), &port);
+    sDebug("vgId:%d sync event, maybe leader transfer to %s:%d %lu", ths->vgId, host, port,
+           pSyncLeaderTransfer->newLeaderId.addr);
+  */
+
+  sDebug("vgId:%d sync event, begin leader transfer", ths->vgId);
+
+  if (strcmp(pSyncLeaderTransfer->newNodeInfo.nodeFqdn, ths->myNodeInfo.nodeFqdn) == 0 &&
+      pSyncLeaderTransfer->newNodeInfo.nodePort == ths->myNodeInfo.nodePort) {
+    sDebug("vgId:%d sync event, maybe leader transfer to %s:%d %lu", ths->vgId,
+           pSyncLeaderTransfer->newNodeInfo.nodeFqdn, pSyncLeaderTransfer->newNodeInfo.nodePort,
+           pSyncLeaderTransfer->newLeaderId.addr);
+
+    // reset elect timer now!
+    int32_t electMS = 1;
+    int32_t ret = syncNodeRestartElectTimer(ths, electMS);
+    ASSERT(ret == 0);
   }
 
+  /*
+    if (syncUtilSameId(&(pSyncLeaderTransfer->newLeaderId), &(ths->myRaftId))) {
+      // reset elect timer now!
+      int32_t electMS = 1;
+      int32_t ret = syncNodeRestartElectTimer(ths, electMS);
+      ASSERT(ret == 0);
+    }
+  */
+  if (ths->pFsm->FpLeaderTransferCb != NULL) {
+    SFsmCbMeta cbMeta;
+    cbMeta.code = 0;
+    cbMeta.currentTerm = ths->pRaftStore->currentTerm;
+    cbMeta.flag = 0;
+    cbMeta.index = pEntry->index;
+    cbMeta.isWeak = pEntry->isWeak;
+    cbMeta.seqNum = pEntry->seqNum;
+    cbMeta.state = ths->state;
+    cbMeta.term = pEntry->term;
+    ths->pFsm->FpLeaderTransferCb(ths->pFsm, pRpcMsg, cbMeta);
+  }
+
+  syncLeaderTransferDestroy(pSyncLeaderTransfer);
   return 0;
 }
 
@@ -1884,26 +2038,31 @@ static int32_t syncNodeConfigChange(SSyncNode* ths, SRpcMsg* pRpcMsg, SSyncRaftE
 
   bool isDrop;
 
-  //if (IamInNew || (!IamInNew && ths->state != TAOS_SYNC_STATE_LEADER)) {
+  // if (IamInNew || (!IamInNew && ths->state != TAOS_SYNC_STATE_LEADER)) {
   if (IamInNew) {
     syncNodeUpdateConfig(ths, &newSyncCfg, pEntry->index, &isDrop);
 
     // change isStandBy to normal
     if (!isDrop) {
+      char tmpbuf[128];
+      snprintf(tmpbuf, sizeof(tmpbuf), "config change from %d to %d", oldSyncCfg.replicaNum, newSyncCfg.replicaNum);
       if (ths->state == TAOS_SYNC_STATE_LEADER) {
-        syncNodeBecomeLeader(ths, "config change");
+        syncNodeBecomeLeader(ths, tmpbuf);
       } else {
-        syncNodeBecomeFollower(ths, "config change");
+        syncNodeBecomeFollower(ths, tmpbuf);
       }
     }
   } else {
-    syncNodeBecomeFollower(ths, "config change2");
+    char tmpbuf[128];
+    snprintf(tmpbuf, sizeof(tmpbuf), "config change2 from %d to %d", oldSyncCfg.replicaNum, newSyncCfg.replicaNum);
+    syncNodeBecomeFollower(ths, tmpbuf);
   }
-  
+
   if (gRaftDetailLog) {
     char* sOld = syncCfg2Str(&oldSyncCfg);
     char* sNew = syncCfg2Str(&newSyncCfg);
-    sInfo("==config change== 0x11 old:%s new:%s isDrop:%d index:%ld \n", sOld, sNew, isDrop, pEntry->index);
+    sInfo("==config change== 0x11 old:%s new:%s isDrop:%d index:%ld IamInNew:%d \n", sOld, sNew, isDrop, pEntry->index,
+          IamInNew);
     taosMemoryFree(sOld);
     taosMemoryFree(sNew);
   }
@@ -1965,7 +2124,7 @@ int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex,
           ASSERT(code == 0);
         }
 
-        // config change
+        // leader transfer
         if (pEntry->originalRpcType == TDMT_SYNC_LEADER_TRANSFER) {
           code = syncDoLeaderTransfer(ths, &rpcMsg, pEntry);
           ASSERT(code == 0);

source/libs/sync/src/syncRespMgr.c

@@ -44,6 +44,10 @@ int64_t syncRespMgrAdd(SSyncRespMgr *pObj, SRespStub *pStub) {
   uint64_t keyCode = ++(pObj->seqNum);
   taosHashPut(pObj->pRespHash, &keyCode, sizeof(keyCode), pStub, sizeof(SRespStub));
 
+  SSyncNode *pSyncNode = pObj->data;
+  sDebug("vgId:%d sync event resp mgr add, type:%s seq:%lu handle:%p", pSyncNode->vgId,
+         TMSG_INFO(pStub->rpcMsg.msgType), keyCode, pStub->rpcMsg.info.handle);
+
   taosThreadMutexUnlock(&(pObj->mutex));
   return keyCode;
 }
@@ -63,6 +67,11 @@ int32_t syncRespMgrGet(SSyncRespMgr *pObj, uint64_t index, SRespStub *pStub) {
   void *pTmp = taosHashGet(pObj->pRespHash, &index, sizeof(index));
   if (pTmp != NULL) {
     memcpy(pStub, pTmp, sizeof(SRespStub));
+
+    SSyncNode *pSyncNode = pObj->data;
+    sDebug("vgId:%d sync event resp mgr get, type:%s seq:%lu handle:%p", pSyncNode->vgId,
+           TMSG_INFO(pStub->rpcMsg.msgType), index, pStub->rpcMsg.info.handle);
+
     taosThreadMutexUnlock(&(pObj->mutex));
     return 1;  // get one object
   }
@@ -76,6 +85,11 @@ int32_t syncRespMgrGetAndDel(SSyncRespMgr *pObj, uint64_t index, SRespStub *pStu
   void *pTmp = taosHashGet(pObj->pRespHash, &index, sizeof(index));
   if (pTmp != NULL) {
     memcpy(pStub, pTmp, sizeof(SRespStub));
+
+    SSyncNode *pSyncNode = pObj->data;
+    sDebug("vgId:%d sync event resp mgr get and del, type:%s seq:%lu handle:%p", pSyncNode->vgId,
+           TMSG_INFO(pStub->rpcMsg.msgType), index, pStub->rpcMsg.info.handle);
+
     taosHashRemove(pObj->pRespHash, &index, sizeof(index));
     taosThreadMutexUnlock(&(pObj->mutex));
     return 1;  // get one object

source/libs/sync/src/syncSnapshot.c

@@ -393,7 +393,7 @@ cJSON *snapshotSender2Json(SSyncSnapshotSender *pSender) {
 
 char *snapshotSender2Str(SSyncSnapshotSender *pSender) {
   cJSON *pJson = snapshotSender2Json(pSender);
-  char * serialized = cJSON_Print(pJson);
+  char  *serialized = cJSON_Print(pJson);
   cJSON_Delete(pJson);
   return serialized;
 }
@@ -514,7 +514,7 @@ cJSON *snapshotReceiver2Json(SSyncSnapshotReceiver *pReceiver) {
     cJSON_AddStringToObject(pFromId, "addr", u64buf);
     {
       uint64_t u64 = pReceiver->fromId.addr;
-      cJSON *  pTmp = pFromId;
+      cJSON   *pTmp = pFromId;
       char     host[128] = {0};
       uint16_t port;
       syncUtilU642Addr(u64, host, sizeof(host), &port);
@@ -538,7 +538,7 @@ cJSON *snapshotReceiver2Json(SSyncSnapshotReceiver *pReceiver) {
 
 char *snapshotReceiver2Str(SSyncSnapshotReceiver *pReceiver) {
   cJSON *pJson = snapshotReceiver2Json(pReceiver);
-  char * serialized = cJSON_Print(pJson);
+  char  *serialized = cJSON_Print(pJson);
   cJSON_Delete(pJson);
   return serialized;
 }
@@ -588,6 +588,8 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) {
 
         // maybe update lastconfig
         if (pMsg->lastConfigIndex >= SYNC_INDEX_BEGIN) {
+          int32_t oldReplicaNum = pSyncNode->replicaNum;
+
           // update new config myIndex
           bool     IamInNew = false;
           SSyncCfg newSyncCfg = pMsg->lastConfig;
@@ -614,10 +616,12 @@ int32_t syncNodeOnSnapshotSendCb(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) {
 
           // change isStandBy to normal
           if (!isDrop) {
+            char tmpbuf[128];
+            snprintf(tmpbuf, sizeof(tmpbuf), "config change3 from %d to %d", oldReplicaNum, newSyncCfg.replicaNum);
             if (pSyncNode->state == TAOS_SYNC_STATE_LEADER) {
-              syncNodeBecomeLeader(pSyncNode, "config change");
+              syncNodeBecomeLeader(pSyncNode, tmpbuf);
             } else {
-              syncNodeBecomeFollower(pSyncNode, "config change");
+              syncNodeBecomeFollower(pSyncNode, tmpbuf);
             }
           }
         }

source/libs/sync/src/syncUtil.c

@@ -14,6 +14,8 @@
  */
 
 #include "syncUtil.h"
+#include <stdio.h>
+
 #include "syncEnv.h"
 
 void addEpIntoEpSet(SEpSet* pEpSet, const char* fqdn, uint16_t port);
@@ -21,8 +23,31 @@ void addEpIntoEpSet(SEpSet* pEpSet, const char* fqdn, uint16_t port);
 // ---- encode / decode
 uint64_t syncUtilAddr2U64(const char* host, uint16_t port) {
   uint64_t u64;
+
+  uint32_t hostU32 = taosGetIpv4FromFqdn(host);
+  if (hostU32 == (uint32_t)-1) {
+    sError("Get IP address error");
+    return -1;
+  }
+
+  /*
   uint32_t hostU32 = (uint32_t)taosInetAddr(host);
-  // assert(hostU32 != (uint32_t)-1);
+  if (hostU32 == (uint32_t)-1) {
+    struct hostent* hostEnt = gethostbyname(host);
+    if (hostEnt == NULL) {
+      sError("Get IP address error");
+      return -1;
+    }
+
+    const char* newHost = taosInetNtoa(*(struct in_addr*)(hostEnt->h_addr_list[0]));
+    hostU32 = (uint32_t)taosInetAddr(newHost);
+    if (hostU32 == (uint32_t)-1) {
+      sError("change %s to id, error", newHost);
+    }
+    // ASSERT(hostU32 != (uint32_t)-1);
+  }
+  */
+
   u64 = (((uint64_t)hostU32) << 32) | (((uint32_t)port) << 16);
   return u64;
 }

source/libs/sync/test/CMakeLists.txt

@@ -1,4 +1,5 @@
 add_executable(syncTest "")
+add_executable(syncRaftIdCheck "")
 add_executable(syncEnvTest "")
 add_executable(syncPingTimerTest "")
 add_executable(syncIOTickQTest "")
@@ -54,6 +55,10 @@ target_sources(syncTest
     PRIVATE
     "syncTest.cpp"
 )
+target_sources(syncRaftIdCheck
+    PRIVATE
+    "syncRaftIdCheck.cpp"
+)
 target_sources(syncEnvTest
     PRIVATE
     "syncEnvTest.cpp"
@@ -257,6 +262,11 @@ target_include_directories(syncTest
     "${TD_SOURCE_DIR}/include/libs/sync"
     "${CMAKE_CURRENT_SOURCE_DIR}/../inc"
 )
+target_include_directories(syncRaftIdCheck
+    PUBLIC
+    "${TD_SOURCE_DIR}/include/libs/sync"
+    "${CMAKE_CURRENT_SOURCE_DIR}/../inc"
+)
 target_include_directories(syncEnvTest
     PUBLIC
     "${TD_SOURCE_DIR}/include/libs/sync"
@@ -508,6 +518,10 @@ target_link_libraries(syncTest
     sync
     gtest_main
 )
+target_link_libraries(syncRaftIdCheck
+    sync
+    gtest_main
+)
 target_link_libraries(syncEnvTest
     sync
     gtest_main