Merge branch '3.0' into test/chr/TD-14699

Jenkinsfile2

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

README-CN.md

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

README.md

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

contrib/CMakeLists.txt

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

include/common/tmsg.h

@@ -708,6 +708,7 @@ typedef struct {
   int32_t buffer;        // MB
   int32_t pageSize;
   int32_t pages;
+  int32_t lastRowMem;
   int32_t daysPerFile;
   int32_t daysToKeep0;
   int32_t daysToKeep1;
@@ -736,6 +737,7 @@ typedef struct {
   int32_t buffer;
   int32_t pageSize;
   int32_t pages;
+  int32_t lastRowMem;
   int32_t daysPerFile;
   int32_t daysToKeep0;
   int32_t daysToKeep1;
@@ -1023,8 +1025,10 @@ typedef struct {
   int64_t     clusterId;
   int64_t     rebootTime;
   int64_t     updateTime;
-  int32_t     numOfCores;
+  float       numOfCores;
   int32_t     numOfSupportVnodes;
+  int64_t     memTotal;
+  int64_t     memAvail;
   char        dnodeEp[TSDB_EP_LEN];
   SMnodeLoad  mload;
   SQnodeLoad  qload;
@@ -1079,6 +1083,7 @@ typedef struct {
   int32_t  buffer;
   int32_t  pageSize;
   int32_t  pages;
+  int32_t  lastRowMem;
   int32_t  daysPerFile;
   int32_t  daysToKeep0;
   int32_t  daysToKeep1;
@@ -1131,6 +1136,7 @@ typedef struct {
   int32_t  buffer;
   int32_t  pageSize;
   int32_t  pages;
+  int32_t  lastRowMem;
   int32_t  daysPerFile;
   int32_t  daysToKeep0;
   int32_t  daysToKeep1;

include/libs/nodes/nodes.h

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

include/libs/nodes/plannodes.h

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

include/libs/nodes/querynodes.h

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

include/libs/scalar/scalar.h

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

include/libs/scheduler/scheduler.h

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

include/libs/sync/sync.h

@@ -31,6 +31,12 @@ extern bool gRaftDetailLog;
 #define SYNC_INDEX_INVALID -1
 #define SYNC_TERM_INVALID 0xFFFFFFFFFFFFFFFF
 
+typedef enum {
+  SYNC_STRATEGY_NO_SNAPSHOT = 0,
+  SYNC_STRATEGY_STANDARD_SNAPSHOT = 1,
+  SYNC_STRATEGY_WAL_FIRST = 2,
+} ESyncStrategy;
+
 typedef uint64_t SyncNodeId;
 typedef int32_t  SyncGroupId;
 typedef int64_t  SyncIndex;
@@ -48,11 +54,6 @@ typedef enum {
   TAOS_SYNC_STATE_ERROR = 103,
 } ESyncState;
 
-typedef enum {
-  TAOS_SYNC_FSM_CB_SUCCESS = 0,
-  TAOS_SYNC_FSM_CB_OTHER_ERROR = 1,
-} ESyncFsmCbCode;
-
 typedef struct SNodeInfo {
   uint16_t nodePort;
   char     nodeFqdn[TSDB_FQDN_LEN];
@@ -96,6 +97,11 @@ typedef struct SReConfigCbMeta {
 
 } SReConfigCbMeta;
 
+typedef struct SSnapshotParam {
+  SyncIndex start;
+  SyncIndex end;
+} SSnapshotParam;
+
 typedef struct SSnapshot {
   void*     data;
   SyncIndex lastApplyIndex;
@@ -125,7 +131,7 @@ typedef struct SSyncFSM {
   int32_t (*FpSnapshotStopRead)(struct SSyncFSM* pFsm, void* pReader);
   int32_t (*FpSnapshotDoRead)(struct SSyncFSM* pFsm, void* pReader, void** ppBuf, int32_t* len);
 
-  int32_t (*FpSnapshotStartWrite)(struct SSyncFSM* pFsm, void** ppWriter);
+  int32_t (*FpSnapshotStartWrite)(struct SSyncFSM* pFsm, void* pWriterParam, void** ppWriter);
   int32_t (*FpSnapshotStopWrite)(struct SSyncFSM* pFsm, void* pWriter, bool isApply);
   int32_t (*FpSnapshotDoWrite)(struct SSyncFSM* pFsm, void* pWriter, void* pBuf, int32_t len);
 
@@ -178,15 +184,15 @@ typedef struct SSyncLogStore {
 } SSyncLogStore;
 
 typedef struct SSyncInfo {
-  bool        isStandBy;
+  bool          isStandBy;
-  bool        snapshotEnable;
+  ESyncStrategy snapshotStrategy;
-  SyncGroupId vgId;
+  SyncGroupId   vgId;
-  int32_t     batchSize;
+  int32_t       batchSize;
-  SSyncCfg    syncCfg;
+  SSyncCfg      syncCfg;
-  char        path[TSDB_FILENAME_LEN];
+  char          path[TSDB_FILENAME_LEN];
-  SWal*       pWal;
+  SWal*         pWal;
-  SSyncFSM*   pFsm;
+  SSyncFSM*     pFsm;
-  SMsgCb*     msgcb;
+  SMsgCb*       msgcb;
   int32_t (*FpSendMsg)(const SEpSet* pEpSet, SRpcMsg* pMsg);
   int32_t (*FpEqMsg)(const SMsgCb* msgcb, SRpcMsg* pMsg);
 } SSyncInfo;
@@ -205,7 +211,7 @@ SyncGroupId syncGetVgId(int64_t rid);
 void        syncGetEpSet(int64_t rid, SEpSet* pEpSet);
 void        syncGetRetryEpSet(int64_t rid, SEpSet* pEpSet);
 int32_t     syncPropose(int64_t rid, SRpcMsg* pMsg, bool isWeak);
-// int32_t     syncProposeBatch(int64_t rid, SRpcMsg* pMsgArr, bool* pIsWeakArr, int32_t arrSize);
+int32_t     syncProposeBatch(int64_t rid, SRpcMsg* pMsgArr, bool* pIsWeakArr, int32_t arrSize);
 bool        syncEnvIsStart();
 const char* syncStr(ESyncState state);
 bool        syncIsRestoreFinish(int64_t rid);

include/libs/sync/syncTools.h

@@ -191,12 +191,12 @@ void syncTimeoutLog2(char* s, const SyncTimeout* pMsg);
 typedef struct SyncClientRequest {
   uint32_t bytes;
   int32_t  vgId;
-  uint32_t msgType;          // SyncClientRequest msgType
+  uint32_t msgType;          // TDMT_SYNC_CLIENT_REQUEST
-  uint32_t originalRpcType;  // user RpcMsg msgType
+  uint32_t originalRpcType;  // origin RpcMsg msgType
   uint64_t seqNum;
   bool     isWeak;
-  uint32_t dataLen;  // user RpcMsg.contLen
+  uint32_t dataLen;  // origin RpcMsg.contLen
-  char     data[];   // user RpcMsg.pCont
+  char     data[];   // origin RpcMsg.pCont
 } SyncClientRequest;
 
 SyncClientRequest* syncClientRequestBuild(uint32_t dataLen);
@@ -220,11 +220,6 @@ void syncClientRequestLog(const SyncClientRequest* pMsg);
 void syncClientRequestLog2(char* s, const SyncClientRequest* pMsg);
 
 // ---------------------------------------------
-typedef struct SOffsetAndContLen {
-  int32_t offset;
-  int32_t contLen;
-} SOffsetAndContLen;
-
 typedef struct SRaftMeta {
   uint64_t seqNum;
   bool     isWeak;
@@ -232,20 +227,33 @@ typedef struct SRaftMeta {
 
 // block1:
 // block2: SRaftMeta array
-// block3: rpc msg array (with pCont)
+// block3: rpc msg array (with pCont pointer)
 
 typedef struct SyncClientRequestBatch {
   uint32_t bytes;
   int32_t  vgId;
-  uint32_t msgType;  // SyncClientRequestBatch msgType
+  uint32_t msgType;  // TDMT_SYNC_CLIENT_REQUEST_BATCH
   uint32_t dataCount;
-  uint32_t dataLen;  // user RpcMsg.contLen
+  uint32_t dataLen;
-  char     data[];   // user RpcMsg.pCont
+  char     data[];  // block2, block3
 } SyncClientRequestBatch;
 
 SyncClientRequestBatch* syncClientRequestBatchBuild(SRpcMsg* rpcMsgArr, SRaftMeta* raftArr, int32_t arrSize,
                                                     int32_t vgId);
 void                    syncClientRequestBatch2RpcMsg(const SyncClientRequestBatch* pSyncMsg, SRpcMsg* pRpcMsg);
+void                    syncClientRequestBatchDestroy(SyncClientRequestBatch* pMsg);
+void                    syncClientRequestBatchDestroyDeep(SyncClientRequestBatch* pMsg);
+SRaftMeta*              syncClientRequestBatchMetaArr(const SyncClientRequestBatch* pSyncMsg);
+SRpcMsg*                syncClientRequestBatchRpcMsgArr(const SyncClientRequestBatch* pSyncMsg);
+SyncClientRequestBatch* syncClientRequestBatchFromRpcMsg(const SRpcMsg* pRpcMsg);
+cJSON*                  syncClientRequestBatch2Json(const SyncClientRequestBatch* pMsg);
+char*                   syncClientRequestBatch2Str(const SyncClientRequestBatch* pMsg);
+
+// for debug ----------------------
+void syncClientRequestBatchPrint(const SyncClientRequestBatch* pMsg);
+void syncClientRequestBatchPrint2(char* s, const SyncClientRequestBatch* pMsg);
+void syncClientRequestBatchLog(const SyncClientRequestBatch* pMsg);
+void syncClientRequestBatchLog2(char* s, const SyncClientRequestBatch* pMsg);
 
 // ---------------------------------------------
 typedef struct SyncClientRequestReply {
@@ -318,12 +326,15 @@ void syncRequestVoteReplyLog(const SyncRequestVoteReply* pMsg);
 void syncRequestVoteReplyLog2(char* s, const SyncRequestVoteReply* pMsg);
 
 // ---------------------------------------------
+// data: entry
+
 typedef struct SyncAppendEntries {
   uint32_t bytes;
   int32_t  vgId;
   uint32_t msgType;
   SRaftId  srcId;
   SRaftId  destId;
+
   // private data
   SyncTerm  term;
   SyncIndex prevLogIndex;
@@ -354,18 +365,14 @@ void syncAppendEntriesLog2(char* s, const SyncAppendEntries* pMsg);
 
 // ---------------------------------------------
 
-// define ahead
-/*
 typedef struct SOffsetAndContLen {
   int32_t offset;
   int32_t contLen;
 } SOffsetAndContLen;
-*/
 
-// block1: SOffsetAndContLen
+// data:
-// block2: SOffsetAndContLen Array
+// block1: SOffsetAndContLen Array
-// block3: SRpcMsg Array
+// block2: entry Array
-// block4: SRpcMsg pCont Array
 
 typedef struct SyncAppendEntriesBatch {
   uint32_t bytes;
@@ -382,10 +389,11 @@ typedef struct SyncAppendEntriesBatch {
   SyncTerm  privateTerm;
   int32_t   dataCount;
   uint32_t  dataLen;
-  char      data[];
+  char      data[];  // block1, block2
 } SyncAppendEntriesBatch;
 
-SyncAppendEntriesBatch* syncAppendEntriesBatchBuild(SRpcMsg* rpcMsgArr, int32_t arrSize, int32_t vgId);
+SyncAppendEntriesBatch* syncAppendEntriesBatchBuild(SSyncRaftEntry** entryPArr, int32_t arrSize, int32_t vgId);
+SOffsetAndContLen*      syncAppendEntriesBatchMetaTableArray(SyncAppendEntriesBatch* pMsg);
 void                    syncAppendEntriesBatchDestroy(SyncAppendEntriesBatch* pMsg);
 void                    syncAppendEntriesBatchSerialize(const SyncAppendEntriesBatch* pMsg, char* buf, uint32_t bufLen);
 void                    syncAppendEntriesBatchDeserialize(const char* buf, uint32_t len, SyncAppendEntriesBatch* pMsg);
@@ -396,8 +404,6 @@ void                    syncAppendEntriesBatchFromRpcMsg(const SRpcMsg* pRpcMsg,
 SyncAppendEntriesBatch* syncAppendEntriesBatchFromRpcMsg2(const SRpcMsg* pRpcMsg);
 cJSON*                  syncAppendEntriesBatch2Json(const SyncAppendEntriesBatch* pMsg);
 char*                   syncAppendEntriesBatch2Str(const SyncAppendEntriesBatch* pMsg);
-void syncAppendEntriesBatch2RpcMsgArray(SyncAppendEntriesBatch* pSyncMsg, SRpcMsg* rpcMsgArr, int32_t maxArrSize,
-                                        int32_t* pRetArrSize);
 
 // for debug ----------------------
 void syncAppendEntriesBatchPrint(const SyncAppendEntriesBatch* pMsg);
@@ -477,9 +483,10 @@ typedef struct SyncSnapshotSend {
   SRaftId  destId;
 
   SyncTerm  term;
-  SyncIndex lastIndex;  // lastIndex of snapshot
+  SyncIndex beginIndex;       // snapshot.beginIndex
-  SyncTerm  lastTerm;   // lastTerm of snapshot
+  SyncIndex lastIndex;        // snapshot.lastIndex
-  SyncIndex lastConfigIndex;
+  SyncTerm  lastTerm;         // snapshot.lastTerm
+  SyncIndex lastConfigIndex;  // snapshot.lastConfigIndex
   SSyncCfg  lastConfig;
   SyncTerm  privateTerm;
   int32_t   seq;
@@ -617,6 +624,9 @@ int32_t syncNodeOnRequestVoteReplySnapshotCb(SSyncNode* ths, SyncRequestVoteRepl
 int32_t syncNodeOnAppendEntriesSnapshotCb(SSyncNode* ths, SyncAppendEntries* pMsg);
 int32_t syncNodeOnAppendEntriesReplySnapshotCb(SSyncNode* ths, SyncAppendEntriesReply* pMsg);
 
+int32_t syncNodeOnAppendEntriesSnapshot2Cb(SSyncNode* ths, SyncAppendEntriesBatch* pMsg);
+int32_t syncNodeOnAppendEntriesReplySnapshot2Cb(SSyncNode* ths, SyncAppendEntriesReply* pMsg);
+
 int32_t syncNodeOnSnapshotSendCb(SSyncNode* ths, SyncSnapshotSend* pMsg);
 int32_t syncNodeOnSnapshotRspCb(SSyncNode* ths, SyncSnapshotRsp* pMsg);
 
@@ -633,7 +643,8 @@ typedef int32_t (*FpOnSnapshotSendCb)(SSyncNode* ths, SyncSnapshotSend* pMsg);
 typedef int32_t (*FpOnSnapshotRspCb)(SSyncNode* ths, SyncSnapshotRsp* pMsg);
 
 // option ----------------------------------
-bool syncNodeSnapshotEnable(SSyncNode* pSyncNode);
+bool          syncNodeSnapshotEnable(SSyncNode* pSyncNode);
+ESyncStrategy syncNodeStrategy(SSyncNode* pSyncNode);
 
 // ---------------------------------------------
 

include/libs/transport/trpc.h

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

include/util/taoserror.h

@@ -173,11 +173,12 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_DNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0341)
 #define TSDB_CODE_MND_TOO_MANY_DNODES           TAOS_DEF_ERROR_CODE(0, 0x0342)
 #define TSDB_CODE_MND_NO_ENOUGH_DNODES          TAOS_DEF_ERROR_CODE(0, 0x0343)
-#define TSDB_CODE_MND_INVALID_CLUSTER_CFG       TAOS_DEF_ERROR_CODE(0, 0x0344)
+#define TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE    TAOS_DEF_ERROR_CODE(0, 0x0344)
-#define TSDB_CODE_MND_INVALID_CLUSTER_ID        TAOS_DEF_ERROR_CODE(0, 0x0345)
+#define TSDB_CODE_MND_INVALID_CLUSTER_CFG       TAOS_DEF_ERROR_CODE(0, 0x0345)
-#define TSDB_CODE_MND_INVALID_DNODE_CFG         TAOS_DEF_ERROR_CODE(0, 0x0346)
+#define TSDB_CODE_MND_INVALID_CLUSTER_ID        TAOS_DEF_ERROR_CODE(0, 0x0346)
-#define TSDB_CODE_MND_INVALID_DNODE_EP          TAOS_DEF_ERROR_CODE(0, 0x0347)
+#define TSDB_CODE_MND_INVALID_DNODE_CFG         TAOS_DEF_ERROR_CODE(0, 0x0347)
-#define TSDB_CODE_MND_INVALID_DNODE_ID          TAOS_DEF_ERROR_CODE(0, 0x0348)
+#define TSDB_CODE_MND_INVALID_DNODE_EP          TAOS_DEF_ERROR_CODE(0, 0x0348)
+#define TSDB_CODE_MND_INVALID_DNODE_ID          TAOS_DEF_ERROR_CODE(0, 0x0349)
 
 // mnode-node
 #define TSDB_CODE_MND_MNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0350)
@@ -428,6 +429,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_SYN_PROPOSE_NOT_READY         TAOS_DEF_ERROR_CODE(0, 0x0911)
 #define TSDB_CODE_SYN_STANDBY_NOT_READY         TAOS_DEF_ERROR_CODE(0, 0x0912)
 #define TSDB_CODE_SYN_BATCH_ERROR               TAOS_DEF_ERROR_CODE(0, 0x0913)
+#define TSDB_CODE_SYN_TIMEOUT                   TAOS_DEF_ERROR_CODE(0, 0x0914)
 #define TSDB_CODE_SYN_INTERNAL_ERROR            TAOS_DEF_ERROR_CODE(0, 0x09FF)
 
 // tq
@@ -578,6 +580,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_PAR_INVALID_TABLE_OPTION      TAOS_DEF_ERROR_CODE(0, 0x265C)
 #define TSDB_CODE_PAR_INVALID_INTERP_CLAUSE     TAOS_DEF_ERROR_CODE(0, 0x265D)
 #define TSDB_CODE_PAR_NO_VALID_FUNC_IN_WIN      TAOS_DEF_ERROR_CODE(0, 0x265E)
+#define TSDB_CODE_PAR_ONLY_SUPPORT_SINGLE_TABLE TAOS_DEF_ERROR_CODE(0, 0x265F)
 
 //planner
 #define TSDB_CODE_PLAN_INTERNAL_ERROR           TAOS_DEF_ERROR_CODE(0, 0x2700)
@@ -627,6 +630,8 @@ int32_t* taosGetErrno();
 
 //index
 #define TSDB_CODE_INDEX_REBUILDING               TAOS_DEF_ERROR_CODE(0, 0x3200)
+#define TSDB_CODE_INDEX_INVALID_FILE             TAOS_DEF_ERROR_CODE(0, 0x3201)
+
 
 //tmq
 #define TSDB_CODE_TMQ_INVALID_MSG                TAOS_DEF_ERROR_CODE(0, 0x4000)

include/util/tdef.h

@@ -287,7 +287,7 @@ typedef enum ELogicConditionType {
 #define TSDB_MULTI_TABLEMETA_MAX_NUM 100000  // maximum batch size allowed to load table meta
 
 #define TSDB_MIN_VNODES_PER_DB          1
-#define TSDB_MAX_VNODES_PER_DB          4096
+#define TSDB_MAX_VNODES_PER_DB          1024
 #define TSDB_DEFAULT_VN_PER_DB          2
 #define TSDB_MIN_BUFFER_PER_VNODE       3      // unit MB
 #define TSDB_MAX_BUFFER_PER_VNODE       16384  // unit MB
@@ -334,6 +334,9 @@ typedef enum ELogicConditionType {
 #define TSDB_MIN_DB_CACHE_LAST_ROW      0
 #define TSDB_MAX_DB_CACHE_LAST_ROW      3
 #define TSDB_DEFAULT_CACHE_LAST_ROW     0
+#define TSDB_MIN_DB_LAST_ROW_MEM        1  // MB
+#define TSDB_MAX_DB_LAST_ROW_MEM        65536
+#define TSDB_DEFAULT_LAST_ROW_MEM       1
 #define TSDB_DB_STREAM_MODE_OFF         0
 #define TSDB_DB_STREAM_MODE_ON          1
 #define TSDB_DEFAULT_DB_STREAM_MODE     0

source/client/inc/clientInt.h

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

source/client/src/clientEnv.c

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

source/client/src/clientImpl.c

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

source/client/src/clientMain.c

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

source/common/src/systable.c

@@ -108,13 +108,11 @@ static const SSysDbTableSchema userFuncSchema[] = {
 };
 
 static const SSysDbTableSchema userIdxSchema[] = {
+    {.name = "index_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "db_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "table_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "index_database", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
+    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-    {.name = "index_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
+    {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-    {.name = "column_name", .bytes = SYSTABLE_SCH_COL_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "index_type", .bytes = 10, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "index_extensions", .bytes = 256, .type = TSDB_DATA_TYPE_VARCHAR},
 };
 
 static const SSysDbTableSchema userStbsSchema[] = {
@@ -313,7 +311,7 @@ static const SSysDbTableSchema querySchema[] = {
     {.name = "query_id", .bytes = TSDB_QUERY_ID_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "req_id", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
     {.name = "connId", .bytes = 4, .type = TSDB_DATA_TYPE_UINT},
-    {.name = "app", .bytes = TSDB_APP_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},    
+    {.name = "app", .bytes = TSDB_APP_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "pid", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "user", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "end_point", .bytes = TSDB_IPv4ADDR_LEN + 6 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
@@ -329,17 +327,17 @@ static const SSysDbTableSchema appSchema[] = {
     {.name = "app_id", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
     {.name = "ip", .bytes = TSDB_IPv4ADDR_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "pid", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-    {.name = "name", .bytes = TSDB_APP_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},    
+    {.name = "name", .bytes = TSDB_APP_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "start_time", .bytes = 8 , .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "start_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-    {.name = "insert_req", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "insert_req", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "insert_row", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "insert_row", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "insert_time", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "insert_time", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "insert_bytes", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "insert_bytes", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "fetch_bytes", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "fetch_bytes", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "query_time", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "query_time", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "show_query", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "show_query", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "total_req", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "total_req", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
-    {.name = "current_req", .bytes = 8 , .type = TSDB_DATA_TYPE_UBIGINT},
+    {.name = "current_req", .bytes = 8, .type = TSDB_DATA_TYPE_UBIGINT},
     {.name = "last_access", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
 };
 
@@ -353,8 +351,7 @@ static const SSysTableMeta perfsMeta[] = {
     {TSDB_PERFS_TABLE_TRANS, transSchema, tListLen(transSchema)},
     {TSDB_PERFS_TABLE_SMAS, smaSchema, tListLen(smaSchema)},
     {TSDB_PERFS_TABLE_STREAMS, streamSchema, tListLen(streamSchema)},
-    {TSDB_PERFS_TABLE_APPS, appSchema, tListLen(appSchema)}
+    {TSDB_PERFS_TABLE_APPS, appSchema, tListLen(appSchema)}};
-};
 
 void getInfosDbMeta(const SSysTableMeta** pInfosTableMeta, size_t* size) {
   *pInfosTableMeta = infosMeta;

source/common/src/tdatablock.c

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

source/common/src/tmsg.c

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

@@ -93,6 +93,7 @@ static SSdbRaw *mndDbActionEncode(SDbObj *pDb) {
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.buffer, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.pageSize, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.pages, _OVER)
+  SDB_SET_INT32(pRaw, dataPos, pDb->cfg.lastRowMem, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.daysPerFile, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.daysToKeep0, _OVER)
   SDB_SET_INT32(pRaw, dataPos, pDb->cfg.daysToKeep1, _OVER)
@@ -165,6 +166,7 @@ static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.buffer, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.pageSize, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.pages, _OVER)
+  SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.lastRowMem, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.daysPerFile, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.daysToKeep0, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &pDb->cfg.daysToKeep1, _OVER)
@@ -230,8 +232,9 @@ static int32_t mndDbActionUpdate(SSdb *pSdb, SDbObj *pOld, SDbObj *pNew) {
   pOld->cfgVersion = pNew->cfgVersion;
   pOld->vgVersion = pNew->vgVersion;
   pOld->cfg.buffer = pNew->cfg.buffer;
-  pOld->cfg.pages = pNew->cfg.pages;
   pOld->cfg.pageSize = pNew->cfg.pageSize;
+  pOld->cfg.pages = pNew->cfg.pages;
+  pOld->cfg.lastRowMem = pNew->cfg.lastRowMem;
   pOld->cfg.daysPerFile = pNew->cfg.daysPerFile;
   pOld->cfg.daysToKeep0 = pNew->cfg.daysToKeep0;
   pOld->cfg.daysToKeep1 = pNew->cfg.daysToKeep1;
@@ -288,6 +291,7 @@ static int32_t mndCheckDbCfg(SMnode *pMnode, SDbCfg *pCfg) {
   if (pCfg->buffer < TSDB_MIN_BUFFER_PER_VNODE || pCfg->buffer > TSDB_MAX_BUFFER_PER_VNODE) return -1;
   if (pCfg->pageSize < TSDB_MIN_PAGESIZE_PER_VNODE || pCfg->pageSize > TSDB_MAX_PAGESIZE_PER_VNODE) return -1;
   if (pCfg->pages < TSDB_MIN_PAGES_PER_VNODE || pCfg->pages > TSDB_MAX_PAGES_PER_VNODE) return -1;
+  if (pCfg->lastRowMem < TSDB_MIN_DB_LAST_ROW_MEM || pCfg->lastRowMem > TSDB_MAX_DB_LAST_ROW_MEM) return -1;
   if (pCfg->daysPerFile < TSDB_MIN_DAYS_PER_FILE || pCfg->daysPerFile > TSDB_MAX_DAYS_PER_FILE) return -1;
   if (pCfg->daysToKeep0 < TSDB_MIN_KEEP || pCfg->daysToKeep0 > TSDB_MAX_KEEP) return -1;
   if (pCfg->daysToKeep1 < TSDB_MIN_KEEP || pCfg->daysToKeep1 > TSDB_MAX_KEEP) return -1;
@@ -336,6 +340,7 @@ static void mndSetDefaultDbCfg(SDbCfg *pCfg) {
   if (pCfg->replications < 0) pCfg->replications = TSDB_DEFAULT_DB_REPLICA;
   if (pCfg->strict < 0) pCfg->strict = TSDB_DEFAULT_DB_STRICT;
   if (pCfg->cacheLastRow < 0) pCfg->cacheLastRow = TSDB_DEFAULT_CACHE_LAST_ROW;
+  if (pCfg->lastRowMem <= 0) pCfg->lastRowMem = TSDB_DEFAULT_LAST_ROW_MEM;
   if (pCfg->numOfRetensions < 0) pCfg->numOfRetensions = 0;
   if (pCfg->schemaless < 0) pCfg->schemaless = TSDB_DB_SCHEMALESS_OFF;
 }
@@ -434,6 +439,7 @@ static int32_t mndCreateDb(SMnode *pMnode, SRpcMsg *pReq, SCreateDbReq *pCreate,
       .buffer = pCreate->buffer,
       .pageSize = pCreate->pageSize,
       .pages = pCreate->pages,
+      .lastRowMem = pCreate->lastRowMem,
       .daysPerFile = pCreate->daysPerFile,
       .daysToKeep0 = pCreate->daysToKeep0,
       .daysToKeep1 = pCreate->daysToKeep1,
@@ -475,7 +481,7 @@ static int32_t mndCreateDb(SMnode *pMnode, SRpcMsg *pReq, SCreateDbReq *pCreate,
   int32_t code = -1;
   STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB, pReq);
   if (pTrans == NULL) goto _OVER;
-
+  // mndTransSetSerial(pTrans);
   mDebug("trans:%d, used to create db:%s", pTrans->id, pCreate->db);
 
   mndTransSetDbName(pTrans, dbObj.name, NULL);
@@ -549,18 +555,33 @@ static int32_t mndSetDbCfgFromAlterDbReq(SDbObj *pDb, SAlterDbReq *pAlter) {
   terrno = TSDB_CODE_MND_DB_OPTION_UNCHANGED;
 
   if (pAlter->buffer > 0 && pAlter->buffer != pDb->cfg.buffer) {
+#if 1
+    terrno = TSDB_CODE_OPS_NOT_SUPPORT;
+    return terrno;
+#else
     pDb->cfg.buffer = pAlter->buffer;
     terrno = 0;
+#endif
   }
 
   if (pAlter->pages > 0 && pAlter->pages != pDb->cfg.pages) {
+#if 1
+    terrno = TSDB_CODE_OPS_NOT_SUPPORT;
+    return terrno;
+#else
     pDb->cfg.pages = pAlter->pages;
     terrno = 0;
+#endif
   }
 
   if (pAlter->pageSize > 0 && pAlter->pageSize != pDb->cfg.pageSize) {
+#if 1
+    terrno = TSDB_CODE_OPS_NOT_SUPPORT;
+    return terrno;
+#else
     pDb->cfg.pageSize = pAlter->pageSize;
     terrno = 0;
+#endif
   }
 
   if (pAlter->daysPerFile > 0 && pAlter->daysPerFile != pDb->cfg.daysPerFile) {
@@ -594,8 +615,12 @@ static int32_t mndSetDbCfgFromAlterDbReq(SDbObj *pDb, SAlterDbReq *pAlter) {
   }
 
   if (pAlter->strict >= 0 && pAlter->strict != pDb->cfg.strict) {
+#if 1
+    terrno = TSDB_CODE_OPS_NOT_SUPPORT;
+#else
     pDb->cfg.strict = pAlter->strict;
     terrno = 0;
+#endif
   }
 
   if (pAlter->cacheLastRow >= 0 && pAlter->cacheLastRow != pDb->cfg.cacheLastRow) {
@@ -603,10 +628,19 @@ static int32_t mndSetDbCfgFromAlterDbReq(SDbObj *pDb, SAlterDbReq *pAlter) {
     terrno = 0;
   }
 
+  if (pAlter->lastRowMem > 0 && pAlter->lastRowMem != pDb->cfg.lastRowMem) {
+    pDb->cfg.lastRowMem = pAlter->lastRowMem;
+    terrno = 0;
+  }
+
   if (pAlter->replications > 0 && pAlter->replications != pDb->cfg.replications) {
+#if 1
+    terrno = TSDB_CODE_OPS_NOT_SUPPORT;
+#else
     pDb->cfg.replications = pAlter->replications;
     pDb->vgVersion++;
     terrno = 0;
+#endif
   }
 
   return terrno;

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

@@ -15,8 +15,8 @@
 
 #define _DEFAULT_SOURCE
 #include "mndDnode.h"
-#include "mndPrivilege.h"
 #include "mndMnode.h"
+#include "mndPrivilege.h"
 #include "mndQnode.h"
 #include "mndShow.h"
 #include "mndSnode.h"
@@ -274,15 +274,14 @@ static void mndGetDnodeData(SMnode *pMnode, SArray *pDnodeEps) {
 
     SDnodeEp dnodeEp = {0};
     dnodeEp.id = pDnode->id;
-    dnodeEp.isMnode = 0;
     dnodeEp.ep.port = pDnode->port;
     memcpy(dnodeEp.ep.fqdn, pDnode->fqdn, TSDB_FQDN_LEN);
+    sdbRelease(pSdb, pDnode);
 
+    dnodeEp.isMnode = 0;
     if (mndIsMnode(pMnode, pDnode->id)) {
       dnodeEp.isMnode = 1;
     }
-
-    sdbRelease(pSdb, pDnode);
     taosArrayPush(pDnodeEps, &dnodeEp);
   }
 }
@@ -432,7 +431,8 @@ static int32_t mndProcessStatusReq(SRpcMsg *pReq) {
     }
 
     if (!online) {
-      mInfo("dnode:%d, from offline to online", pDnode->id);
+      mInfo("dnode:%d, from offline to online, memory avail:%" PRId64 " total:%" PRId64 " cores:%.2f", pDnode->id,
+            statusReq.memAvail, statusReq.memTotal, statusReq.numOfCores);
     } else {
       mDebug("dnode:%d, send dnode epset, online:%d dnodeVer:%" PRId64 ":%" PRId64 " reboot:%d", pDnode->id, online,
              statusReq.dnodeVer, dnodeVer, reboot);
@@ -441,6 +441,8 @@ static int32_t mndProcessStatusReq(SRpcMsg *pReq) {
     pDnode->rebootTime = statusReq.rebootTime;
     pDnode->numOfCores = statusReq.numOfCores;
     pDnode->numOfSupportVnodes = statusReq.numOfSupportVnodes;
+    pDnode->memAvail = statusReq.memAvail;
+    pDnode->memTotal = statusReq.memTotal;
 
     SStatusRsp statusRsp = {0};
     statusRsp.dnodeVer = dnodeVer;
@@ -580,7 +582,7 @@ static int32_t mndProcessShowVariablesReq(SRpcMsg *pReq) {
   strcpy(info.name, "timezone");
   snprintf(info.value, TSDB_CONFIG_VALUE_LEN, "%s", tsTimezoneStr);
   taosArrayPush(rsp.variables, &info);
-  
+
   strcpy(info.name, "locale");
   snprintf(info.value, TSDB_CONFIG_VALUE_LEN, "%s", tsLocale);
   taosArrayPush(rsp.variables, &info);
@@ -758,6 +760,11 @@ static int32_t mndProcessDropDnodeReq(SRpcMsg *pReq) {
     }
   }
 
+  if (numOfVnodes > 0) {
+    terrno = TSDB_CODE_OPS_NOT_SUPPORT;
+    goto _OVER;
+  }
+
   code = mndDropDnode(pMnode, pReq, pDnode, pMObj, pQObj, pSObj, numOfVnodes);
   if (code == 0) code = TSDB_CODE_ACTION_IN_PROGRESS;
 

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

@@ -58,7 +58,7 @@ static void *mndBuildTimerMsg(int32_t *pContLen) {
 
 static void mndPullupTrans(SMnode *pMnode) {
   int32_t contLen = 0;
-  void *  pReq = mndBuildTimerMsg(&contLen);
+  void   *pReq = mndBuildTimerMsg(&contLen);
   if (pReq != NULL) {
     SRpcMsg rpcMsg = {.msgType = TDMT_MND_TRANS_TIMER, .pCont = pReq, .contLen = contLen};
     tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
@@ -67,14 +67,14 @@ static void mndPullupTrans(SMnode *pMnode) {
 
 static void mndTtlTimer(SMnode *pMnode) {
   int32_t contLen = 0;
-  void *  pReq = mndBuildTimerMsg(&contLen);
+  void   *pReq = mndBuildTimerMsg(&contLen);
   SRpcMsg rpcMsg = {.msgType = TDMT_MND_TTL_TIMER, .pCont = pReq, .contLen = contLen};
   tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
 }
 
 static void mndCalMqRebalance(SMnode *pMnode) {
   int32_t contLen = 0;
-  void *  pReq = mndBuildTimerMsg(&contLen);
+  void   *pReq = mndBuildTimerMsg(&contLen);
   if (pReq != NULL) {
     SRpcMsg rpcMsg = {.msgType = TDMT_MND_MQ_TIMER, .pCont = pReq, .contLen = contLen};
     tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
@@ -83,7 +83,7 @@ static void mndCalMqRebalance(SMnode *pMnode) {
 
 static void mndPullupTelem(SMnode *pMnode) {
   int32_t contLen = 0;
-  void *  pReq = mndBuildTimerMsg(&contLen);
+  void   *pReq = mndBuildTimerMsg(&contLen);
   if (pReq != NULL) {
     SRpcMsg rpcMsg = {.msgType = TDMT_MND_TELEM_TIMER, .pCont = pReq, .contLen = contLen};
     tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
@@ -395,7 +395,7 @@ void mndStop(SMnode *pMnode) {
 }
 
 int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
-  SMnode *   pMnode = pMsg->info.node;
+  SMnode    *pMnode = pMsg->info.node;
   SSyncMgmt *pMgmt = &pMnode->syncMgmt;
   int32_t    code = 0;
 
@@ -413,7 +413,7 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
   }
 
   do {
-    char *         syncNodeStr = sync2SimpleStr(pMgmt->sync);
+    char          *syncNodeStr = sync2SimpleStr(pMgmt->sync);
     static int64_t mndTick = 0;
     if (++mndTick % 10 == 1) {
       mTrace("vgId:%d, sync trace msg:%s, %s", syncGetVgId(pMgmt->sync), TMSG_INFO(pMsg->msgType), syncNodeStr);
@@ -427,7 +427,7 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
   } while (0);
 
   // ToDo: ugly! use function pointer
-  if (syncNodeSnapshotEnable(pSyncNode)) {
+  if (syncNodeStrategy(pSyncNode) == SYNC_STRATEGY_STANDARD_SNAPSHOT) {
     if (pMsg->msgType == TDMT_SYNC_TIMEOUT) {
       SyncTimeout *pSyncMsg = syncTimeoutFromRpcMsg2(pMsg);
       code = syncNodeOnTimeoutCb(pSyncNode, pSyncMsg);
@@ -579,7 +579,7 @@ static int32_t mndCheckMsgContent(SRpcMsg *pMsg) {
 }
 
 int32_t mndProcessRpcMsg(SRpcMsg *pMsg) {
-  SMnode *        pMnode = pMsg->info.node;
+  SMnode         *pMnode = pMsg->info.node;
   const STraceId *trace = &pMsg->info.traceId;
 
   MndMsgFp fp = pMnode->msgFp[TMSG_INDEX(pMsg->msgType)];
@@ -632,7 +632,7 @@ int32_t mndGetMonitorInfo(SMnode *pMnode, SMonClusterInfo *pClusterInfo, SMonVgr
                           SMonStbInfo *pStbInfo, SMonGrantInfo *pGrantInfo) {
   if (mndAcquireRpcRef(pMnode) != 0) return -1;
 
-  SSdb *  pSdb = pMnode->pSdb;
+  SSdb   *pSdb = pMnode->pSdb;
   int64_t ms = taosGetTimestampMs();
 
   pClusterInfo->dnodes = taosArrayInit(sdbGetSize(pSdb, SDB_DNODE), sizeof(SMonDnodeDesc));
@@ -713,7 +713,7 @@ int32_t mndGetMonitorInfo(SMnode *pMnode, SMonClusterInfo *pClusterInfo, SMonVgr
     pGrantInfo->timeseries_used += pVgroup->numOfTimeSeries;
     tstrncpy(desc.status, "unsynced", sizeof(desc.status));
     for (int32_t i = 0; i < pVgroup->replica; ++i) {
-      SVnodeGid *    pVgid = &pVgroup->vnodeGid[i];
+      SVnodeGid     *pVgid = &pVgroup->vnodeGid[i];
       SMonVnodeDesc *pVnDesc = &desc.vnodes[i];
       pVnDesc->dnode_id = pVgid->dnodeId;
       tstrncpy(pVnDesc->vnode_role, syncStr(pVgid->role), sizeof(pVnDesc->vnode_role));

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

@@ -71,7 +71,7 @@ static int32_t convertToRetrieveType(char *name, int32_t len) {
   } else if (strncasecmp(name, TSDB_INS_TABLE_USER_FUNCTIONS, len) == 0) {
     type = TSDB_MGMT_TABLE_FUNC;
   } else if (strncasecmp(name, TSDB_INS_TABLE_USER_INDEXES, len) == 0) {
-    //    type = TSDB_MGMT_TABLE_INDEX;
+    type = TSDB_MGMT_TABLE_INDEX;
   } else if (strncasecmp(name, TSDB_INS_TABLE_USER_STABLES, len) == 0) {
     type = TSDB_MGMT_TABLE_STB;
   } else if (strncasecmp(name, TSDB_INS_TABLE_USER_TABLES, len) == 0) {

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

@@ -1147,29 +1147,32 @@ static int32_t mndRetrieveSma(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBloc
 
     SName smaName = {0};
     tNameFromString(&smaName, pSma->name, T_NAME_ACCT | T_NAME_DB | T_NAME_TABLE);
+    char n1[TSDB_TABLE_FNAME_LEN + VARSTR_HEADER_SIZE] = {0};
+    STR_TO_VARSTR(n1, (char *)tNameGetTableName(&smaName));
 
-    char n[TSDB_TABLE_FNAME_LEN + VARSTR_HEADER_SIZE] = {0};
+    char n2[TSDB_DB_FNAME_LEN + VARSTR_HEADER_SIZE] = {0};
-    STR_TO_VARSTR(n, (char *)tNameGetTableName(&smaName));
+    STR_TO_VARSTR(n2, (char *)mndGetDbStr(pDb->name));
-    cols++;
 
     SName stbName = {0};
     tNameFromString(&stbName, pSma->stb, T_NAME_ACCT | T_NAME_DB | T_NAME_TABLE);
-
+    char n3[TSDB_TABLE_FNAME_LEN + VARSTR_HEADER_SIZE] = {0};
-    char n1[TSDB_TABLE_FNAME_LEN + VARSTR_HEADER_SIZE] = {0};
+    STR_TO_VARSTR(n3, (char *)tNameGetTableName(&stbName));
-    STR_TO_VARSTR(n1, (char *)tNameGetTableName(&stbName));
 
     SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-    colDataAppend(pColInfo, numOfRows, (const char *)n, false);
-
-    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
-    colDataAppend(pColInfo, numOfRows, (const char *)&pSma->createdTime, false);
-
-    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)n1, false);
 
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)n2, false);
+
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)n3, false);
+
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
     colDataAppend(pColInfo, numOfRows, (const char *)&pSma->dstVgId, false);
 
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pSma->createdTime, false);
+
     numOfRows++;
     sdbRelease(pSdb, pSma);
   }

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

@@ -273,6 +273,9 @@ _OVER:
 }
 
 static int32_t mndProcessCreateSnodeReq(SRpcMsg *pReq) {
+#if 1
+  return TSDB_CODE_OPS_NOT_SUPPORT;
+#else
   SMnode          *pMnode = pReq->info.node;
   int32_t          code = -1;
   SSnodeObj       *pObj = NULL;
@@ -315,6 +318,7 @@ _OVER:
   mndReleaseSnode(pMnode, pObj);
   mndReleaseDnode(pMnode, pDnode);
   return code;
+#endif
 }
 
 static int32_t mndSetDropSnodeRedoLogs(STrans *pTrans, SSnodeObj *pObj) {
@@ -386,9 +390,12 @@ _OVER:
 }
 
 static int32_t mndProcessDropSnodeReq(SRpcMsg *pReq) {
-  SMnode        *pMnode = pReq->info.node;
+#if 1
-  int32_t        code = -1;
+  return TSDB_CODE_OPS_NOT_SUPPORT;
-  SSnodeObj     *pObj = NULL;
+#else
+  SMnode *pMnode = pReq->info.node;
+  int32_t code = -1;
+  SSnodeObj *pObj = NULL;
   SMDropSnodeReq dropReq = {0};
 
   if (tDeserializeSCreateDropMQSBNodeReq(pReq->pCont, pReq->contLen, &dropReq) != 0) {
@@ -422,6 +429,7 @@ _OVER:
 
   mndReleaseSnode(pMnode, pObj);
   return code;
+#endif
 }
 
 static int32_t mndRetrieveSnodes(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {

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

@@ -247,7 +247,6 @@ static int32_t mndBuildStreamObjFromCreateReq(SMnode *pMnode, SStreamObj *pObj,
   pObj->uid = mndGenerateUid(pObj->name, strlen(pObj->name));
   pObj->status = 0;
 
-  // TODO
   pObj->igExpired = pCreate->igExpired;
   pObj->trigger = pCreate->triggerType;
   pObj->triggerParam = pCreate->maxDelay;

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

@@ -134,7 +134,7 @@ int32_t mndSnapshotDoRead(struct SSyncFSM *pFsm, void *pReader, void **ppBuf, in
   return sdbDoRead(pMnode->pSdb, pReader, ppBuf, len);
 }
 
-int32_t mndSnapshotStartWrite(struct SSyncFSM *pFsm, void **ppWriter) {
+int32_t mndSnapshotStartWrite(struct SSyncFSM *pFsm, void *pParam, void **ppWriter) {
   mInfo("start to apply snapshot to sdb");
   SMnode *pMnode = pFsm->data;
   return sdbStartWrite(pMnode->pSdb, (SSdbIter **)ppWriter);
@@ -178,7 +178,7 @@ int32_t mndInitSync(SMnode *pMnode) {
   syncInfo.pWal = pMnode->pWal;
   syncInfo.pFsm = mndSyncMakeFsm(pMnode);
   syncInfo.isStandBy = pMgmt->standby;
-  syncInfo.snapshotEnable = true;
+  syncInfo.snapshotStrategy = SYNC_STRATEGY_STANDARD_SNAPSHOT;
 
   mInfo("start to open mnode sync, standby:%d", pMgmt->standby);
   if (pMgmt->standby || pMgmt->replica.id > 0) {

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

@@ -207,6 +207,7 @@ void *mndBuildCreateVnodeReq(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVg
   createReq.buffer = pDb->cfg.buffer;
   createReq.pageSize = pDb->cfg.pageSize;
   createReq.pages = pDb->cfg.pages;
+  createReq.lastRowMem = pDb->cfg.lastRowMem;
   createReq.daysPerFile = pDb->cfg.daysPerFile;
   createReq.daysToKeep0 = pDb->cfg.daysToKeep0;
   createReq.daysToKeep1 = pDb->cfg.daysToKeep1;
@@ -274,8 +275,9 @@ void *mndBuildAlterVnodeReq(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup, int32_
   SAlterVnodeReq alterReq = {0};
   alterReq.vgVersion = pVgroup->version;
   alterReq.buffer = pDb->cfg.buffer;
-  alterReq.pages = pDb->cfg.pages;
   alterReq.pageSize = pDb->cfg.pageSize;
+  alterReq.pages = pDb->cfg.pages;
+  alterReq.lastRowMem = pDb->cfg.lastRowMem;
   alterReq.daysPerFile = pDb->cfg.daysPerFile;
   alterReq.daysToKeep0 = pDb->cfg.daysToKeep0;
   alterReq.daysToKeep1 = pDb->cfg.daysToKeep1;
@@ -392,9 +394,10 @@ static bool mndBuildDnodesArrayFp(SMnode *pMnode, void *pObj, void *p1, void *p2
   bool    online = mndIsDnodeOnline(pDnode, curMs);
   bool    isMnode = mndIsMnode(pMnode, pDnode->id);
   pDnode->numOfVnodes = mndGetVnodesNum(pMnode, pDnode->id);
+  pDnode->memUsed = mndGetVnodesMemory(pMnode, pDnode->id);
 
-  mDebug("dnode:%d, vnodes:%d support_vnodes:%d is_mnode:%d online:%d", pDnode->id, pDnode->numOfVnodes,
+  mDebug("dnode:%d, vnodes:%d supportVnodes:%d isMnode:%d online:%d memory avail:%" PRId64 " used:%" PRId64, pDnode->id,
-         pDnode->numOfSupportVnodes, isMnode, online);
+         pDnode->numOfVnodes, pDnode->numOfSupportVnodes, isMnode, online, pDnode->memAvail, pDnode->memUsed);
 
   if (isMnode) {
     pDnode->numOfVnodes++;
@@ -426,15 +429,7 @@ static int32_t mndCompareDnodeId(int32_t *dnode1Id, int32_t *dnode2Id) { return
 static int32_t mndCompareDnodeVnodes(SDnodeObj *pDnode1, SDnodeObj *pDnode2) {
   float d1Score = (float)pDnode1->numOfVnodes / pDnode1->numOfSupportVnodes;
   float d2Score = (float)pDnode2->numOfVnodes / pDnode2->numOfSupportVnodes;
-#if 0  
-  if (d1Score == d2Score) {
-    return pDnode2->id - pDnode1->id;
-  } else {
-    return d1Score >= d2Score ? 1 : 0;
-  }
-#else
   return d1Score >= d2Score ? 1 : 0;
-#endif
 }
 
 void mndSortVnodeGid(SVgObj *pVgroup) {
@@ -447,7 +442,7 @@ void mndSortVnodeGid(SVgObj *pVgroup) {
   }
 }
 
-static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {
+static int32_t mndGetAvailableDnode(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup, SArray *pArray) {
   SSdb   *pSdb = pMnode->pSdb;
   int32_t allocedVnodes = 0;
   void   *pIter = NULL;
@@ -470,6 +465,16 @@ static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pAr
       return -1;
     }
 
+    int64_t vgMem = mndGetVgroupMemory(pMnode, pDb, pVgroup);
+    if (pDnode->memAvail - vgMem - pDnode->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d, avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pDnode->id, pDnode->memAvail, pDnode->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pDnode->memUsed += vgMem;
+    }
+
     pVgid->dnodeId = pDnode->id;
     if (pVgroup->replica == 1) {
       pVgid->role = TAOS_SYNC_STATE_LEADER;
@@ -477,7 +482,8 @@ static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pAr
       pVgid->role = TAOS_SYNC_STATE_FOLLOWER;
     }
 
-    mInfo("db:%s, vgId:%d, vn:%d dnode:%d is alloced", pVgroup->dbName, pVgroup->vgId, v, pVgid->dnodeId);
+    mInfo("db:%s, vgId:%d, vn:%d is alloced, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+          pVgroup->dbName, pVgroup->vgId, v, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
     pDnode->numOfVnodes++;
   }
 
@@ -498,7 +504,7 @@ int32_t mndAllocSmaVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj *pVgroup) {
   pVgroup->dbUid = pDb->uid;
   pVgroup->replica = 1;
 
-  if (mndGetAvailableDnode(pMnode, pVgroup, pArray) != 0) return -1;
+  if (mndGetAvailableDnode(pMnode, pDb, pVgroup, pArray) != 0) return -1;
 
   mInfo("db:%s, sma vgId:%d is alloced", pDb->name, pVgroup->vgId);
   return 0;
@@ -546,8 +552,7 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
     pVgroup->dbUid = pDb->uid;
     pVgroup->replica = pDb->cfg.replications;
 
-    if (mndGetAvailableDnode(pMnode, pVgroup, pArray) != 0) {
+    if (mndGetAvailableDnode(pMnode, pDb, pVgroup, pArray) != 0) {
-      terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
 
@@ -728,6 +733,43 @@ int32_t mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId) {
   return numOfVnodes;
 }
 
+int64_t mndGetVgroupMemory(SMnode *pMnode, SDbObj *pDbInput, SVgObj *pVgroup) {
+  SDbObj *pDb = pDbInput;
+  if (pDbInput == NULL) {
+    pDb = mndAcquireDb(pMnode, pVgroup->dbName);
+  }
+
+  int64_t vgroupMemroy = (int64_t)pDb->cfg.buffer * 1024 * 1024 + (int64_t)pDb->cfg.pages * pDb->cfg.pageSize * 1024;
+  if (pDb->cfg.cacheLastRow > 0) {
+    vgroupMemroy += (int64_t)pDb->cfg.lastRowMem * 1024 * 1024;
+  }
+
+  if (pDbInput == NULL) {
+    mndReleaseDb(pMnode, pDb);
+  }
+  return vgroupMemroy;
+}
+
+static bool mndGetVnodeMemroyFp(SMnode *pMnode, void *pObj, void *p1, void *p2, void *p3) {
+  SVgObj  *pVgroup = pObj;
+  int32_t  dnodeId = *(int32_t *)p1;
+  int64_t *pVnodeMemory = (int64_t *)p2;
+
+  for (int32_t v = 0; v < pVgroup->replica; ++v) {
+    if (pVgroup->vnodeGid[v].dnodeId == dnodeId) {
+      *pVnodeMemory += mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    }
+  }
+
+  return true;
+}
+
+int64_t mndGetVnodesMemory(SMnode *pMnode, int32_t dnodeId) {
+  int64_t vnodeMemory = 0;
+  sdbTraverse(pMnode->pSdb, SDB_VGROUP, mndGetVnodeMemroyFp, &dnodeId, &vnodeMemory, NULL);
+  return vnodeMemory;
+}
+
 static int32_t mndRetrieveVnodes(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {
   SMnode *pMnode = pReq->info.node;
   SSdb   *pSdb = pMnode->pSdb;
@@ -807,9 +849,20 @@ int32_t mndAddVnodeToVgroup(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {
       return -1;
     }
 
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pDnode->memAvail - vgMem - pDnode->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pDnode->id, pDnode->memAvail, pDnode->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pDnode->memUsed += vgMem;
+    }
+
     pVgid->dnodeId = pDnode->id;
     pVgid->role = TAOS_SYNC_STATE_ERROR;
-    mInfo("db:%s, vgId:%d, vn:%d dnode:%d, is added", pVgroup->dbName, pVgroup->vgId, pVgroup->replica, pVgid->dnodeId);
+    mInfo("db:%s, vgId:%d, vn:%d is added, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+          pVgroup->dbName, pVgroup->vgId, pVgroup->replica, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
 
     pVgroup->replica++;
     pDnode->numOfVnodes++;
@@ -835,7 +888,10 @@ int32_t mndRemoveVnodeFromVgroup(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray
     for (int32_t vn = 0; vn < pVgroup->replica; ++vn) {
       SVnodeGid *pVgid = &pVgroup->vnodeGid[vn];
       if (pVgid->dnodeId == pDnode->id) {
-        mInfo("db:%s, vgId:%d, vn:%d dnode:%d, is removed", pVgroup->dbName, pVgroup->vgId, vn, pVgid->dnodeId);
+        int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+        pDnode->memUsed -= vgMem;
+        mInfo("db:%s, vgId:%d, vn:%d is removed, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+              pVgroup->dbName, pVgroup->vgId, vn, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
         pDnode->numOfVnodes--;
         pVgroup->replica--;
         *pDelVgid = *pVgid;
@@ -1161,6 +1217,17 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew1->memAvail - vgMem - pNew1->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew1->id, pNew1->memAvail, pNew1->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew1->memUsed += vgMem;
+    }
+
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew1->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld1->id) != 0) goto _OVER;
   }
@@ -1173,6 +1240,15 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew2->memAvail - vgMem - pNew2->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew2->id, pNew2->memAvail, pNew2->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew2->memUsed += vgMem;
+    }
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew2->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld2->id) != 0) goto _OVER;
   }
@@ -1185,6 +1261,15 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew3->memAvail - vgMem - pNew3->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew3->id, pNew3->memAvail, pNew3->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew3->memUsed += vgMem;
+    }
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew3->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld3->id) != 0) goto _OVER;
   }
@@ -1219,6 +1304,9 @@ _OVER:
 }
 
 static int32_t mndProcessRedistributeVgroupMsg(SRpcMsg *pReq) {
+#if 1
+  return TSDB_CODE_OPS_NOT_SUPPORT;
+#else
   SMnode    *pMnode = pReq->info.node;
   SDnodeObj *pNew1 = NULL;
   SDnodeObj *pNew2 = NULL;
@@ -1412,6 +1500,7 @@ _OVER:
   mndReleaseDb(pMnode, pDb);
 
   return code;
+#endif
 }
 
 int32_t mndBuildAlterVgroupAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup, SArray *pArray) {
@@ -1650,9 +1739,9 @@ static int32_t mndBalanceVgroupBetweenDnode(SMnode *pMnode, STrans *pTrans, SDno
 }
 
 static int32_t mndBalanceVgroup(SMnode *pMnode, SRpcMsg *pReq, SArray *pArray) {
-  int32_t code = -1;
+  int32_t   code = -1;
-  int32_t numOfVgroups = 0;
+  int32_t   numOfVgroups = 0;
-  STrans *pTrans = NULL;
+  STrans   *pTrans = NULL;
   SHashObj *pBalancedVgroups = NULL;
 
   pBalancedVgroups = taosHashInit(16, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false, HASH_NO_LOCK);
@@ -1711,10 +1800,13 @@ _OVER:
 }
 
 static int32_t mndProcessBalanceVgroupMsg(SRpcMsg *pReq) {
+#if 1
+  return TSDB_CODE_OPS_NOT_SUPPORT;
+#else
   SMnode *pMnode = pReq->info.node;
   int32_t code = -1;
   SArray *pArray = NULL;
-  void   *pIter = NULL;
+  void *pIter = NULL;
   int64_t curMs = taosGetTimestampMs();
 
   SBalanceVgroupReq req = {0};
@@ -1759,6 +1851,7 @@ _OVER:
 
   taosArrayDestroy(pArray);
   return code;
+#endif
 }
 
 bool mndVgroupInDb(SVgObj *pVgroup, int64_t dbUid) { return !pVgroup->isTsma && pVgroup->dbUid == dbUid; }

source/dnode/mnode/impl/test/db/db.cpp

@@ -93,7 +93,7 @@ TEST_F(MndTestDb, 02_Create_Alter_Drop_Db) {
 
     SRpcMsg* pRsp = test.SendReq(TDMT_MND_ALTER_DB, pReq, contLen);
     ASSERT_NE(pRsp, nullptr);
-    ASSERT_EQ(pRsp->code, 0);
+    ASSERT_EQ(pRsp->code, TSDB_CODE_OPS_NOT_SUPPORT);
   }
 
   test.SendShowReq(TSDB_MGMT_TABLE_DB, "user_databases", "");

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

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

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

@@ -62,12 +62,10 @@ struct STSmaStat {
 
 struct SRSmaStat {
   SSma     *pSma;
-  int64_t   refId;         // shared by persistence/fetch tasks
+  int64_t   refId;         // shared by fetch tasks
-  void     *tmrHandle;     // for persistence task
+  void     *tmrHandle;     // shared by fetch tasks
-  tmr_h     tmrId;         // for persistence task
+  int8_t    triggerStat;   // shared by fetch tasks
-  int32_t   tmrSeconds;    // for persistence task
+  int8_t    runningStat;   // for persistence task 
-  int8_t    triggerStat;   // for persistence task
-  int8_t    runningStat;   // for persistence task
   SHashObj *rsmaInfoHash;  // key: stbUid, value: SRSmaInfo;
 };
 
@@ -82,7 +80,6 @@ struct SSmaStat {
 #define SMA_TSMA_STAT(s)     (&(s)->tsmaStat)
 #define SMA_RSMA_STAT(s)     (&(s)->rsmaStat)
 #define RSMA_INFO_HASH(r)    ((r)->rsmaInfoHash)
-#define RSMA_TMR_ID(r)       ((r)->tmrId)
 #define RSMA_TMR_HANDLE(r)   ((r)->tmrHandle)
 #define RSMA_TRIGGER_STAT(r) (&(r)->triggerStat)
 #define RSMA_RUNNING_STAT(r) (&(r)->runningStat)
@@ -185,9 +182,11 @@ static FORCE_INLINE void tdSmaStatSetDropped(STSmaStat *pTStat) {
 static int32_t tdDestroySmaState(SSmaStat *pSmaStat, int8_t smaType);
 void          *tdFreeSmaState(SSmaStat *pSmaStat, int8_t smaType);
 void          *tdFreeRSmaInfo(SRSmaInfo *pInfo);
+int32_t        tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat);
 
 int32_t tdProcessRSmaCreateImpl(SSma *pSma, SRSmaParam *param, int64_t suid, const char *tbName);
 int32_t tdProcessRSmaRestoreImpl(SSma *pSma);
+
 int32_t tdProcessTSmaCreateImpl(SSma *pSma, int64_t version, const char *pMsg);
 int32_t tdProcessTSmaInsertImpl(SSma *pSma, int64_t indexUid, const char *msg);
 int32_t tdProcessTSmaGetDaysImpl(SVnodeCfg *pCfg, void *pCont, uint32_t contLen, int32_t *days);
@@ -244,8 +243,8 @@ void    tdUpdateTFileMagic(STFile *pTFile, void *pCksm);
 void    tdCloseTFile(STFile *pTFile);
 void    tdDestroyTFile(STFile *pTFile);
 
-void tdGetVndFileName(int32_t vgId, const char *dname, const char *fname, int64_t version, char *outputName);
+void tdGetVndFileName(int32_t vgId, const char *pdname, const char *dname, const char *fname, int64_t version, char *outputName);
-void tdGetVndDirName(int32_t vgId, const char *dname, char *outputName);
+void tdGetVndDirName(int32_t vgId,const char *pdname,  const char *dname, bool endWithSep, char *outputName);
 
 #ifdef __cplusplus
 }

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

@@ -163,8 +163,8 @@ SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pSchema, bool
 
 // sma
 int32_t smaOpen(SVnode* pVnode);
-int32_t smaCloseEnv(SSma* pSma);
+int32_t smaClose(SSma* pSma);
-int32_t smaCloseEx(SSma* pSma);
+int32_t smaBegin(SSma* pSma);
 int32_t smaPreCommit(SSma* pSma);
 int32_t smaCommit(SSma* pSma);
 int32_t smaPostCommit(SSma* pSma);

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

@@ -43,13 +43,48 @@ int32_t smaCommit(SSma *pSma) { return tdProcessRSmaCommitImpl(pSma); }
  */
 int32_t smaPostCommit(SSma *pSma) { return tdProcessRSmaPostCommitImpl(pSma); }
 
+/**
+ * @brief set rsma trigger stat active
+ *
+ * @param pSma
+ * @return int32_t
+ */
+int32_t smaBegin(SSma *pSma) {
+  SSmaEnv *pSmaEnv = SMA_RSMA_ENV(pSma);
+  if (!pSmaEnv) {
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SSmaStat  *pStat = SMA_ENV_STAT(pSmaEnv);
+  SRSmaStat *pRSmaStat = SMA_RSMA_STAT(pStat);
+
+  int8_t rsmaTriggerStat =
+      atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_PAUSED, TASK_TRIGGER_STAT_ACTIVE);
+  switch (rsmaTriggerStat) {
+    case TASK_TRIGGER_STAT_PAUSED: {
+      smaDebug("vgId:%d rsma trigger stat from paused to active", SMA_VID(pSma));
+      break;
+    }
+    case TASK_TRIGGER_STAT_INIT: {
+      atomic_store_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_ACTIVE);
+      smaDebug("vgId:%d rsma trigger stat from init to active", SMA_VID(pSma));
+      break;
+    }
+    default: {
+      atomic_store_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_ACTIVE);
+      smaWarn("vgId:%d rsma trigger stat %" PRIi8 " is unexpected", SMA_VID(pSma), rsmaTriggerStat);
+      ASSERT(0);
+      break;
+    }
+  }
+  return TSDB_CODE_SUCCESS;
+}
+
 /**
  * @brief pre-commit for rollup sma.
  *  1) set trigger stat of rsma timer TASK_TRIGGER_STAT_PAUSED.
- *  2) perform persist task for qTaskInfo
+ *  2) wait all triggered fetch tasks finished
- *  3) wait all triggered fetch tasks finished
+ *  3) perform persist task for qTaskInfo
- *  4) set trigger stat of rsma timer TASK_TRIGGER_STAT_ACTIVE.
- *  5) finish
  *
  * @param pSma
  * @return int32_t
@@ -63,10 +98,30 @@ static int32_t tdProcessRSmaPreCommitImpl(SSma *pSma) {
   SSmaStat  *pStat = SMA_ENV_STAT(pSmaEnv);
   SRSmaStat *pRSmaStat = SMA_RSMA_STAT(pStat);
 
-  // step 1
+
+  // step 1: set persistence task paused
   atomic_store_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_PAUSED);
 
-  // step 2
+  // step 2: wait all triggered fetch tasks finished
+  int32_t nLoops = 0;
+  while (1) {
+    if (T_REF_VAL_GET(pStat) == 0) {
+      smaDebug("vgId:%d, rsma fetch tasks all finished", SMA_VID(pSma));
+      break;
+    } else {
+      smaDebug("vgId:%d, rsma fetch tasks not all finished yet", SMA_VID(pSma));
+    }
+    ++nLoops;
+    if (nLoops > 1000) {
+      sched_yield();
+      nLoops = 0;
+    }
+  }
+
+  // step 3: perform persist task for qTaskInfo
+  tdRSmaPersistExecImpl(pRSmaStat);
+
+  smaDebug("vgId:%d, rsma pre commit succeess", SMA_VID(pSma));
 
   return TSDB_CODE_SUCCESS;
 }
@@ -103,48 +158,68 @@ static int32_t tdProcessRSmaPostCommitImpl(SSma *pSma) {
   TdDirPtr      pDir = NULL;
   TdDirEntryPtr pDirEntry = NULL;
   char          dir[TSDB_FILENAME_LEN];
-  char          bname[TSDB_FILENAME_LEN];
+  const char   *pattern = "v[0-9]+qtaskinfo\\.ver([0-9]+)?$";
-  const char   *pattern = "^v[0-9]+qtaskinfo\\.ver([0-9]+)?$";
   regex_t       regex;
+  int           code = 0;
 
-  tdGetVndDirName(TD_VID(pVnode), VNODE_RSMA_DIR, dir);
+  tdGetVndDirName(TD_VID(pVnode), tfsGetPrimaryPath(pVnode->pTfs), VNODE_RSMA_DIR, true, dir);
 
   // Resource allocation and init
-  regcomp(&regex, pattern, REG_EXTENDED);
+  if ((code = regcomp(&regex, pattern, REG_EXTENDED)) != 0) {
-
+    char errbuf[128];
-  if ((pDir = taosOpenDir(dir)) == NULL) {
+    regerror(code, &regex, errbuf, sizeof(errbuf));
-    terrno = TAOS_SYSTEM_ERROR(errno);
+    smaWarn("vgId:%d, rsma post commit, regcomp for %s failed since %s", TD_VID(pVnode), dir, errbuf);
-    smaWarn("rsma post-commit open dir %s failed since %s", dir, terrstr());
     return TSDB_CODE_FAILED;
   }
 
+  if ((pDir = taosOpenDir(dir)) == NULL) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    smaWarn("vgId:%d, rsma post commit, open dir %s failed since %s", TD_VID(pVnode), dir, terrstr());
+    return TSDB_CODE_FAILED;
+  }
+
+  int32_t    dirLen = strlen(dir);
+  char      *dirEnd = POINTER_SHIFT(dir, dirLen);
   regmatch_t regMatch[2];
   while ((pDirEntry = taosReadDir(pDir)) != NULL) {
     char *entryName = taosGetDirEntryName(pDirEntry);
     if (!entryName) {
       continue;
     }
-    char *fileName = taosDirEntryBaseName(entryName);
+
-    int   code = regexec(&regex, bname, 2, regMatch, 0);
+    code = regexec(&regex, entryName, 2, regMatch, 0);
 
     if (code == 0) {
       // match
-      printf("match 0 = %s\n", (char *)POINTER_SHIFT(fileName, regMatch[0].rm_so));
+      int64_t version = -1;
-      printf("match 1 = %s\n", (char *)POINTER_SHIFT(fileName, regMatch[1].rm_so));
+      sscanf((const char *)POINTER_SHIFT(entryName, regMatch[1].rm_so), "%" PRIi64, &version);
+      if ((version < committed) && (version > -1)) {
+        strncpy(dirEnd, entryName, TSDB_FILENAME_LEN - dirLen);
+        if (taosRemoveFile(dir) != 0) {
+          terrno = TAOS_SYSTEM_ERROR(errno);
+          smaWarn("vgId:%d, committed version:%" PRIi64 ", failed to remove %s since %s", TD_VID(pVnode), committed,
+                  dir, terrstr());
+        } else {
+          smaDebug("vgId:%d, committed version:%" PRIi64 ", success to remove %s", TD_VID(pVnode), committed, dir);
+        }
+      }
     } else if (code == REG_NOMATCH) {
       // not match
-      smaInfo("rsma post-commit not match %s", fileName);
+      smaTrace("vgId:%d, rsma post commit, not match %s", TD_VID(pVnode), entryName);
       continue;
     } else {
       // has other error
-      terrno = TAOS_SYSTEM_ERROR(code);
+      char errbuf[128];
-      smaWarn("rsma post-commit regexec failed since %s", terrstr());
+      regerror(code, &regex, errbuf, sizeof(errbuf));
+      smaWarn("vgId:%d, rsma post commit, regexec failed since %s", TD_VID(pVnode), errbuf);
 
       taosCloseDir(&pDir);
       regfree(&regex);
       return TSDB_CODE_FAILED;
     }
   }
+
   taosCloseDir(&pDir);
+  regfree(&regex);
   return TSDB_CODE_SUCCESS;
 }

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

@@ -132,6 +132,7 @@ static int32_t tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType, const SSma *pS
     if (smaType == TSDB_SMA_TYPE_ROLLUP) {
       SRSmaStat *pRSmaStat = (SRSmaStat *)(*pSmaStat);
       pRSmaStat->pSma = (SSma *)pSma;
+      atomic_store_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_INIT);
 
       // init smaMgmt
       smaMgmt.smaRef = taosOpenRef(SMA_MGMT_REF_NUM, tdDestroyRSmaStat);
@@ -192,22 +193,20 @@ static void *tdFreeTSmaStat(STSmaStat *pStat) {
 static void tdDestroyRSmaStat(void *pRSmaStat) {
   if (pRSmaStat) {
     SRSmaStat *pStat = (SRSmaStat *)pRSmaStat;
-    smaDebug("vgId:%d %s:%d destroy rsma stat %p", SMA_VID(pStat->pSma), __func__, __LINE__, pRSmaStat);
+    SSma      *pSma = pStat->pSma;
-    // step 1: set persistence task cancelled
+    smaDebug("vgId:%d, destroy rsma stat %p", SMA_VID(pSma), pRSmaStat);
+    // step 1: set rsma trigger stat cancelled
     atomic_store_8(RSMA_TRIGGER_STAT(pStat), TASK_TRIGGER_STAT_CANCELLED);
 
-    // step 2: stop the persistence timer
+    // step 2: wait the persistence thread to finish
-    taosTmrStopA(&RSMA_TMR_ID(pStat));
-
-    // step 3: wait the persistence thread to finish
     int32_t nLoops = 0;
     if (atomic_load_8(RSMA_RUNNING_STAT(pStat)) == 1) {
       while (1) {
         if (atomic_load_8(RSMA_TRIGGER_STAT(pStat)) == TASK_TRIGGER_STAT_FINISHED) {
-          smaDebug("rsma, persist task finished already");
+          smaDebug("vgId:%d, rsma persist task finished already", SMA_VID(pSma));
           break;
         } else {
-          smaDebug("rsma, persist task not finished yet since rsma stat in %" PRIi8,
+          smaDebug("vgId:%d, rsma persist task not finished yet since rsma stat in %" PRIi8, SMA_VID(pSma),
                    atomic_load_8(RSMA_TRIGGER_STAT(pStat)));
         }
         ++nLoops;
@@ -218,13 +217,15 @@ static void tdDestroyRSmaStat(void *pRSmaStat) {
       }
     }
 
-    // step 4: destroy the rsma info and associated fetch tasks
+    // step 3: destroy the rsma info and associated fetch tasks
     // TODO: use taosHashSetFreeFp when taosHashSetFreeFp is ready.
-    void *infoHash = taosHashIterate(RSMA_INFO_HASH(pStat), NULL);
+    if (taosHashGetSize(RSMA_INFO_HASH(pStat)) > 0) {
-    while (infoHash) {
+      void *infoHash = taosHashIterate(RSMA_INFO_HASH(pStat), NULL);
-      SRSmaInfo *pSmaInfo = *(SRSmaInfo **)infoHash;
+      while (infoHash) {
-      tdFreeRSmaInfo(pSmaInfo);
+        SRSmaInfo *pSmaInfo = *(SRSmaInfo **)infoHash;
-      infoHash = taosHashIterate(RSMA_INFO_HASH(pStat), infoHash);
+        tdFreeRSmaInfo(pSmaInfo);
+        infoHash = taosHashIterate(RSMA_INFO_HASH(pStat), infoHash);
+      }
     }
     taosHashCleanup(RSMA_INFO_HASH(pStat));
 
@@ -232,10 +233,10 @@ static void tdDestroyRSmaStat(void *pRSmaStat) {
     nLoops = 0;
     while (1) {
       if (T_REF_VAL_GET((SSmaStat *)pStat) == 0) {
-        smaDebug("rsma, all fetch task finished already");
+        smaDebug("vgId:%d, rsma fetch tasks all finished", SMA_VID(pSma));
         break;
       } else {
-        smaDebug("rsma, fetch tasks not all finished yet");
+        smaDebug("vgId:%d, rsma fetch tasks not all finished yet", SMA_VID(pSma));
       }
       ++nLoops;
       if (nLoops > 1000) {
@@ -275,7 +276,7 @@ int32_t tdDestroySmaState(SSmaStat *pSmaStat, int8_t smaType) {
     } else if (smaType == TSDB_SMA_TYPE_ROLLUP) {
       SRSmaStat *pRSmaStat = SMA_RSMA_STAT(pSmaStat);
       if (taosRemoveRef(smaMgmt.smaRef, RSMA_REF_ID(pRSmaStat)) < 0) {
-        smaError("remove refId from smaRef failed, refId:0x%" PRIx64, RSMA_REF_ID(pRSmaStat));
+        smaError("remove refId from rsmaRef:0x%" PRIx64 " failed since %s", RSMA_REF_ID(pRSmaStat), terrstr());
       }
     } else {
       ASSERT(0);

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

@@ -135,17 +135,11 @@ _err:
   return -1;
 }
 
-int32_t smaCloseEnv(SSma *pSma) {
+int32_t smaClose(SSma *pSma) {
-  if (pSma) {
-    SMA_TSMA_ENV(pSma) = tdFreeSmaEnv(SMA_TSMA_ENV(pSma));
-    SMA_RSMA_ENV(pSma) = tdFreeSmaEnv(SMA_RSMA_ENV(pSma));
-  }
-  return 0;
-}
-
-int32_t smaCloseEx(SSma *pSma) {
   if (pSma) {
     taosThreadMutexDestroy(&pSma->mutex);
+    SMA_TSMA_ENV(pSma) = tdFreeSmaEnv(SMA_TSMA_ENV(pSma));
+    SMA_RSMA_ENV(pSma) = tdFreeSmaEnv(SMA_RSMA_ENV(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));

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

@@ -15,9 +15,8 @@
 
 #include "sma.h"
 
-#define RSMA_QTASKINFO_PERSIST_MS 7200000
+#define RSMA_QTASKINFO_BUFSIZE  32768
-#define RSMA_QTASKINFO_BUFSIZE    32768
+#define RSMA_QTASKINFO_HEAD_LEN (sizeof(int32_t) + sizeof(int8_t) + sizeof(int64_t))  // len + type + suid
-#define RSMA_QTASKINFO_HEAD_LEN   (sizeof(int32_t) + sizeof(int8_t) + sizeof(int64_t))  // len + type + suid
 
 SSmaMgmt smaMgmt = {
     .smaRef = -1,
@@ -43,9 +42,9 @@ static int32_t tdRSmaQTaskInfoIterNextBlock(SRSmaQTaskInfoIter *pIter, bool *isF
 static int32_t tdRSmaQTaskInfoRestore(SSma *pSma, SRSmaQTaskInfoIter *pIter);
 static int32_t tdRSmaQTaskInfoItemRestore(SSma *pSma, const SRSmaQTaskInfoItem *infoItem);
 
-static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma);
+static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma, int64_t *nTables);
-static int32_t tdRSmaRestoreQTaskInfoReload(SSma *pSma);
+static int32_t tdRSmaRestoreQTaskInfoReload(SSma *pSma, int64_t *committed);
-static int32_t tdRSmaRestoreTSDataReload(SSma *pSma);
+static int32_t tdRSmaRestoreTSDataReload(SSma *pSma, int64_t committed);
 
 struct SRSmaInfoItem {
   SRSmaInfo *pRsmaInfo;
@@ -88,7 +87,7 @@ struct SRSmaQTaskInfoIter {
 };
 
 static void tdRSmaQTaskInfoGetFName(int32_t vgId, int64_t version, char *outputName) {
-  tdGetVndFileName(vgId, VNODE_RSMA_DIR, TD_QTASKINFO_FNAME_PREFIX, version, outputName);
+  tdGetVndFileName(vgId, NULL, VNODE_RSMA_DIR, TD_QTASKINFO_FNAME_PREFIX, version, outputName);
 }
 
 static FORCE_INLINE int32_t tdRSmaQTaskInfoContLen(int32_t lenWithHead) {
@@ -114,12 +113,14 @@ void *tdFreeRSmaInfo(SRSmaInfo *pInfo) {
     for (int32_t i = 0; i < TSDB_RETENTION_L2; ++i) {
       SRSmaInfoItem *pItem = &pInfo->items[i];
       if (pItem->taskInfo) {
-        smaDebug("vgId:%d, stb %" PRIi64 " stop fetch-timer %p level %d", SMA_VID(pSma), pInfo->suid, pItem->tmrId,
+        if (pItem->tmrId) {
-                 i + 1);
+          smaDebug("vgId:%d, table %" PRIi64 " stop fetch timer %p level %d", SMA_VID(pSma), pInfo->suid, pItem->tmrId,
-        taosTmrStopA(&pItem->tmrId);
+                   i + 1);
+          taosTmrStopA(&pItem->tmrId);
+        }
         tdFreeTaskHandle(&pItem->taskInfo, SMA_VID(pSma), i + 1);
       } else {
-        smaDebug("vgId:%d, stb %" PRIi64 " no need to destroy rsma info level %d since empty taskInfo", SMA_VID(pSma),
+        smaDebug("vgId:%d, table %" PRIi64 " no need to destroy rsma info level %d since empty taskInfo", SMA_VID(pSma),
                  pInfo->suid, i + 1);
       }
     }
@@ -358,13 +359,7 @@ int32_t tdProcessRSmaCreateImpl(SSma *pSma, SRSmaParam *param, int64_t suid, con
     goto _err;
   }
 
-  smaDebug("vgId:%d, register rsma info succeed for suid:%" PRIi64, SMA_VID(pSma), suid);
+  smaDebug("vgId:%d, register rsma info succeed for table %" PRIi64, SMA_VID(pSma), suid);
-
-  // start the persist timer
-  if (TASK_TRIGGER_STAT_INIT ==
-      atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pStat), TASK_TRIGGER_STAT_INIT, TASK_TRIGGER_STAT_ACTIVE)) {
-    taosTmrStart(tdRSmaPersistTrigger, RSMA_QTASKINFO_PERSIST_MS, pStat, RSMA_TMR_HANDLE(pStat));
-  }
 
   return TSDB_CODE_SUCCESS;
 _err:
@@ -748,7 +743,7 @@ int32_t tdProcessRSmaSubmit(SSma *pSma, void *pMsg, int32_t inputType) {
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma) {
+static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma, int64_t *nTables) {
   SVnode *pVnode = pSma->pVnode;
 
   SArray *suidList = taosArrayInit(1, sizeof(tb_uid_t));
@@ -758,7 +753,12 @@ static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma) {
     return TSDB_CODE_FAILED;
   }
 
-  int32_t arrSize = taosArrayGetSize(suidList);
+  int64_t arrSize = taosArrayGetSize(suidList);
+
+  if (nTables) {
+    *nTables = arrSize;
+  }
+
   if (arrSize == 0) {
     taosArrayDestroy(suidList);
     smaDebug("vgId:%d, no need to restore rsma env since empty stb id list", TD_VID(pVnode));
@@ -767,9 +767,9 @@ static int32_t tdRSmaRestoreQTaskInfoInit(SSma *pSma) {
 
   SMetaReader mr = {0};
   metaReaderInit(&mr, SMA_META(pSma), 0);
-  for (int32_t i = 0; i < arrSize; ++i) {
+  for (int64_t i = 0; i < arrSize; ++i) {
     tb_uid_t suid = *(tb_uid_t *)taosArrayGet(suidList, i);
-    smaDebug("vgId:%d, rsma restore, suid[%d] is %" PRIi64, TD_VID(pVnode), i, suid);
+    smaDebug("vgId:%d, rsma restore, suid is %" PRIi64, TD_VID(pVnode), suid);
     if (metaGetTableEntryByUid(&mr, suid) < 0) {
       smaError("vgId:%d, rsma restore, failed to get table meta for %" PRIi64 " since %s", TD_VID(pVnode), suid,
                terrstr());
@@ -803,7 +803,7 @@ _err:
   return TSDB_CODE_FAILED;
 }
 
-static int32_t tdRSmaRestoreQTaskInfoReload(SSma *pSma) {
+static int32_t tdRSmaRestoreQTaskInfoReload(SSma *pSma, int64_t *committed) {
   SVnode *pVnode = pSma->pVnode;
   STFile  tFile = {0};
   char    qTaskInfoFName[TSDB_FILENAME_LEN] = {0};
@@ -814,6 +814,13 @@ static int32_t tdRSmaRestoreQTaskInfoReload(SSma *pSma) {
   }
 
   if (!taosCheckExistFile(TD_TFILE_FULL_NAME(&tFile))) {
+    if (pVnode->state.committed > 0) {
+      smaWarn("vgId:%d, rsma restore for version %" PRIi64 ", not start as %s not exist", TD_VID(pVnode),
+              pVnode->state.committed, TD_TFILE_FULL_NAME(&tFile));
+    } else {
+      smaDebug("vgId:%d, rsma restore for version %" PRIi64 ", no need as %s not exist", TD_VID(pVnode),
+               pVnode->state.committed, TD_TFILE_FULL_NAME(&tFile));
+    }
     return TSDB_CODE_SUCCESS;
   }
 
@@ -839,9 +846,14 @@ static int32_t tdRSmaRestoreQTaskInfoReload(SSma *pSma) {
   tdRSmaQTaskInfoIterDestroy(&fIter);
   tdCloseTFile(&tFile);
   tdDestroyTFile(&tFile);
+
+  // restored successfully from committed
+  *committed = pVnode->state.committed;
+
   return TSDB_CODE_SUCCESS;
 _err:
-  smaError("rsma restore, qtaskinfo reload failed since %s", terrstr());
+  smaError("vgId:%d, rsma restore for version %" PRIi64 ", qtaskinfo reload failed since %s", TD_VID(pVnode),
+           pVnode->state.committed, terrstr());
   return TSDB_CODE_FAILED;
 }
 
@@ -849,35 +861,45 @@ _err:
  * @brief reload ts data from checkpoint
  *
  * @param pSma
+ * @param committed restore from committed version
  * @return int32_t
  */
-static int32_t tdRSmaRestoreTSDataReload(SSma *pSma) {
+static int32_t tdRSmaRestoreTSDataReload(SSma *pSma, int64_t committed) {
   // TODO
+  smaDebug("vgId:%d, rsma restore from %" PRIi64 ", ts data reload success", SMA_VID(pSma), committed);
   return TSDB_CODE_SUCCESS;
 _err:
-  smaError("rsma restore, ts data reload failed since %s", terrstr());
+  smaError("vgId:%d, rsma restore from %" PRIi64 ", ts data reload failed since %s", SMA_VID(pSma), committed,
+           terrstr());
   return TSDB_CODE_FAILED;
 }
 
 int32_t tdProcessRSmaRestoreImpl(SSma *pSma) {
   // step 1: iterate all stables to restore the rsma env
-  if (tdRSmaRestoreQTaskInfoInit(pSma) < 0) {
+  int64_t nTables = 0;
+  if (tdRSmaRestoreQTaskInfoInit(pSma, &nTables) < 0) {
     goto _err;
   }
 
+  if (nTables <= 0) {
+    smaDebug("vgId:%d, no need to restore rsma task since no tables", SMA_VID(pSma));
+    return TSDB_CODE_SUCCESS;
+  }
+
   // step 2: retrieve qtaskinfo items from the persistence file(rsma/qtaskinfo) and restore
-  if (tdRSmaRestoreQTaskInfoReload(pSma) < 0) {
+  int64_t committed = -1;
+  if (tdRSmaRestoreQTaskInfoReload(pSma, &committed) < 0) {
     goto _err;
   }
 
   // step 3: reload ts data from checkpoint
-  if (tdRSmaRestoreTSDataReload(pSma) < 0) {
+  if (tdRSmaRestoreTSDataReload(pSma, committed) < 0) {
     goto _err;
   }
 
   return TSDB_CODE_SUCCESS;
 _err:
-  smaError("failed to restore rsma task since %s", terrstr());
+  smaError("vgId:%d failed to restore rsma task since %s", SMA_VID(pSma), terrstr());
   return TSDB_CODE_FAILED;
 }
 
@@ -893,9 +915,9 @@ static int32_t tdRSmaQTaskInfoItemRestore(SSma *pSma, const SRSmaQTaskInfoItem *
     return TSDB_CODE_SUCCESS;
   }
 
-  if (pItem->type == 1) {
+  if (pItem->type == TSDB_RETENTION_L1) {
     qTaskInfo = pRSmaInfo->items[0].taskInfo;
-  } else if (pItem->type == 2) {
+  } else if (pItem->type == TSDB_RETENTION_L2) {
     qTaskInfo = pRSmaInfo->items[1].taskInfo;
   } else {
     ASSERT(0);
@@ -1017,7 +1039,8 @@ static int32_t tdRSmaQTaskInfoRestore(SSma *pSma, SRSmaQTaskInfoIter *pIter) {
     pIter->qBuf = taosDecodeFixedI32(pIter->qBuf, &qTaskInfoLenWithHead);
     if (qTaskInfoLenWithHead < RSMA_QTASKINFO_HEAD_LEN) {
       terrno = TSDB_CODE_TDB_FILE_CORRUPTED;
-      smaError("restore rsma qtaskinfo file %s failed since %s", TD_TFILE_FULL_NAME(pIter->pTFile), terrstr());
+      smaError("vgId:%d, restore rsma qtaskinfo file %s failed since %s", SMA_VID(pSma),
+               TD_TFILE_FULL_NAME(pIter->pTFile), terrstr());
       return TSDB_CODE_FAILED;
     }
 
@@ -1054,13 +1077,17 @@ static int32_t tdRSmaQTaskInfoRestore(SSma *pSma, SRSmaQTaskInfoIter *pIter) {
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat) {
+int32_t tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat) {
   SSma   *pSma = pRSmaStat->pSma;
   SVnode *pVnode = pSma->pVnode;
   int32_t vid = SMA_VID(pSma);
   int64_t toffset = 0;
   bool    isFileCreated = false;
 
+  if (taosHashGetSize(RSMA_INFO_HASH(pRSmaStat)) <= 0) {
+    return TSDB_CODE_SUCCESS;
+  }
+
   void *infoHash = taosHashIterate(RSMA_INFO_HASH(pRSmaStat), NULL);
   if (!infoHash) {
     return TSDB_CODE_SUCCESS;
@@ -1099,11 +1126,15 @@ static int32_t tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat) {
         char qTaskInfoFName[TSDB_FILENAME_LEN];
         tdRSmaQTaskInfoGetFName(vid, pSma->pVnode->state.applied, qTaskInfoFName);
         if (tdInitTFile(&tFile, tfsGetPrimaryPath(pVnode->pTfs), qTaskInfoFName) < 0) {
+          smaError("vgId:%d, rsma persit, init %s failed since %s", vid, qTaskInfoFName, terrstr());
           goto _err;
         }
         if (tdCreateTFile(&tFile, true, -1) < 0) {
+          smaError("vgId:%d, rsma persit, create %s failed since %s", vid, TD_TFILE_FULL_NAME(&tFile), terrstr());
           goto _err;
         }
+        smaDebug("vgId:%d, rsma, table %" PRIi64 " level %d serialize qTaskInfo, file %s created", vid, pRSmaInfo->suid,
+                 i + 1, TD_TFILE_FULL_NAME(&tFile));
 
         isFileCreated = true;
       }
@@ -1143,6 +1174,7 @@ static int32_t tdRSmaPersistExecImpl(SRSmaStat *pRSmaStat) {
   }
   return TSDB_CODE_SUCCESS;
 _err:
+  smaError("vgId:%d, rsma persit failed since %s", vid, terrstr());
   if (isFileCreated) {
     tdRemoveTFile(&tFile);
     tdDestroyTFile(&tFile);
@@ -1201,7 +1233,6 @@ static void tdRSmaPersistTask(SRSmaStat *pRSmaStat) {
     } else {
       smaWarn("vgId:%d, persist task in abnormal stat %" PRIi8, SMA_VID(pRSmaStat->pSma),
               atomic_load_8(RSMA_TRIGGER_STAT(pRSmaStat)));
-      ASSERT(0);
     }
     atomic_store_8(RSMA_RUNNING_STAT(pRSmaStat), 0);
     taosReleaseRef(smaMgmt.smaRef, pRSmaStat->refId);
@@ -1238,8 +1269,8 @@ static void tdRSmaPersistTrigger(void *param, void *tmrId) {
         // start persist task
         tdRSmaPersistTask(pRSmaStat);
 
-        taosTmrReset(tdRSmaPersistTrigger, RSMA_QTASKINFO_PERSIST_MS, pRSmaStat, pRSmaStat->tmrHandle,
+        // taosTmrReset(tdRSmaPersistTrigger, 5000, pRSmaStat, pRSmaStat->tmrHandle,
-                     &pRSmaStat->tmrId);
+        //              RSMA_TMR_ID(pRSmaStat));
       } else {
         atomic_store_8(RSMA_RUNNING_STAT(pRSmaStat), 0);
       }

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

@@ -140,7 +140,7 @@ int64_t tdAppendTFile(STFile *pTFile, void *buf, int64_t nbyte, int64_t *offset)
     return -1;
   }
 
-#if 1
+#if 0
   smaDebug("append to file %s, offset:%" PRIi64 " nbyte:%" PRIi64 " fsize:%" PRIi64, TD_TFILE_FULL_NAME(pTFile),
            toffset, nbyte, toffset + nbyte);
 #endif
@@ -181,16 +181,43 @@ void tdCloseTFile(STFile *pTFile) {
 
 void tdDestroyTFile(STFile *pTFile) { taosMemoryFreeClear(TD_TFILE_FULL_NAME(pTFile)); }
 
-void tdGetVndFileName(int32_t vgId, const char *dname, const char *fname, int64_t version, char *outputName) {
+void tdGetVndFileName(int32_t vgId, const char *pdname, const char *dname, const char *fname, int64_t version,
+                      char *outputName) {
   if (version < 0) {
-    snprintf(outputName, TSDB_FILENAME_LEN, "vnode/vnode%d/%s/v%d%s", vgId, dname, vgId, fname);
+    if (pdname) {
+      snprintf(outputName, TSDB_FILENAME_LEN, "%s%svnode%svnode%d%s%s%sv%d%s", pdname, TD_DIRSEP, TD_DIRSEP, vgId,
+               TD_DIRSEP, dname, TD_DIRSEP, vgId, fname);
+    } else {
+      snprintf(outputName, TSDB_FILENAME_LEN, "vnode%svnode%d%s%s%sv%d%s", TD_DIRSEP, vgId, TD_DIRSEP, dname, TD_DIRSEP,
+               vgId, fname);
+    }
   } else {
-    snprintf(outputName, TSDB_FILENAME_LEN, "vnode/vnode%d/%s/v%d%s%" PRIi64, vgId, dname, vgId, fname, version);
+    if (pdname) {
+      snprintf(outputName, TSDB_FILENAME_LEN, "%s%svnode%svnode%d%s%s%sv%d%s%" PRIi64, pdname, TD_DIRSEP, TD_DIRSEP,
+               vgId, TD_DIRSEP, dname, TD_DIRSEP, vgId, fname, version);
+    } else {
+      snprintf(outputName, TSDB_FILENAME_LEN, "vnode%svnode%d%s%s%sv%d%s%" PRIi64, TD_DIRSEP, vgId, TD_DIRSEP, dname,
+               TD_DIRSEP, vgId, fname, version);
+    }
   }
 }
 
-void tdGetVndDirName(int32_t vgId, const char *dname, char *outputName) {
+void tdGetVndDirName(int32_t vgId, const char *pdname, const char *dname, bool endWithSep, char *outputName) {
-  snprintf(outputName, TSDB_FILENAME_LEN, "vnode/vnode%d/%s", vgId, dname);
+  if (pdname) {
+    if (endWithSep) {
+      snprintf(outputName, TSDB_FILENAME_LEN, "%s%svnode%svnode%d%s%s%s", pdname, TD_DIRSEP, TD_DIRSEP, vgId, TD_DIRSEP,
+               dname, TD_DIRSEP);
+    } else {
+      snprintf(outputName, TSDB_FILENAME_LEN, "%s%svnode%svnode%d%s%s", pdname, TD_DIRSEP, TD_DIRSEP, vgId, TD_DIRSEP,
+               dname);
+    }
+  } else {
+    if (endWithSep) {
+      snprintf(outputName, TSDB_FILENAME_LEN, "vnode%svnode%d%s%s%s", TD_DIRSEP, vgId, TD_DIRSEP, dname, TD_DIRSEP);
+    } else {
+      snprintf(outputName, TSDB_FILENAME_LEN, "vnode%svnode%d%s%s", TD_DIRSEP, vgId, TD_DIRSEP, dname);
+    }
+  }
 }
 
 int32_t tdInitTFile(STFile *pTFile, const char *dname, const char *fname) {
@@ -215,35 +242,36 @@ int32_t tdInitTFile(STFile *pTFile, const char *dname, const char *fname) {
 
 int32_t tdCreateTFile(STFile *pTFile, bool updateHeader, int8_t fType) {
   ASSERT(pTFile->info.fsize == 0 && pTFile->info.magic == TD_FILE_INIT_MAGIC);
-
   pTFile->pFile = taosOpenFile(TD_TFILE_FULL_NAME(pTFile), TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_TRUNC);
   if (pTFile->pFile == NULL) {
     if (errno == ENOENT) {
       // Try to create directory recursively
-      if (taosMulMkDir(taosDirName(TD_TFILE_FULL_NAME(pTFile))) != 0) {
+      char *s = strdup(TD_TFILE_FULL_NAME(pTFile));
+      if (taosMulMkDir(taosDirName(s)) != 0) {
+        terrno = TAOS_SYSTEM_ERROR(errno);
+        taosMemoryFree(s);
+        return -1;
+      }
+      taosMemoryFree(s);
+      pTFile->pFile = taosOpenFile(TD_TFILE_FULL_NAME(pTFile), TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_TRUNC);
+      if (pTFile->pFile == NULL) {
         terrno = TAOS_SYSTEM_ERROR(errno);
         return -1;
-      } else {
-        pTFile->pFile = taosOpenFile(TD_TFILE_FULL_NAME(pTFile), TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_TRUNC);
-        if (pTFile->pFile == NULL) {
-          terrno = TAOS_SYSTEM_ERROR(errno);
-          return -1;
-        }
       }
     }
+  }
 
-    if (!updateHeader) {
+  if (!updateHeader) {
-      return 0;
+    return 0;
-    }
+  }
 
-    pTFile->info.fsize += TD_FILE_HEAD_SIZE;
+  pTFile->info.fsize += TD_FILE_HEAD_SIZE;
-    pTFile->info.fver = 0;
+  pTFile->info.fver = 0;
 
-    if (tdUpdateTFileHeader(pTFile) < 0) {
+  if (tdUpdateTFileHeader(pTFile) < 0) {
-      tdCloseTFile(pTFile);
+    tdCloseTFile(pTFile);
-      tdRemoveTFile(pTFile);
+    tdRemoveTFile(pTFile);
-      return -1;
+    return -1;
-    }
   }
 
   return 0;

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

@@ -406,193 +406,6 @@ OVER:
   return code;
 }
 
-#if 0
-int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg, int32_t workerId) {
-  SMqPollReq* pReq = pMsg->pCont;
-  int64_t     consumerId = pReq->consumerId;
-  int64_t     timeout = pReq->timeout;
-  int32_t     reqEpoch = pReq->epoch;
-  int64_t     fetchOffset;
-  int32_t     code = 0;
-
-  // get offset to fetch message
-  if (pReq->currentOffset >= 0) {
-    fetchOffset = pReq->currentOffset + 1;
-  } else {
-    STqOffset* pOffset = tqOffsetRead(pTq->pOffsetStore, pReq->subKey);
-    if (pOffset != NULL) {
-      ASSERT(pOffset->val.type == TMQ_OFFSET__LOG);
-      tqDebug("consumer %ld, restore offset of %s on vg %d, offset(type:log) version: %ld", consumerId, pReq->subKey,
-              TD_VID(pTq->pVnode), pOffset->val.version);
-      fetchOffset = pOffset->val.version + 1;
-    } else {
-      if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__EARLIEAST) {
-        fetchOffset = walGetFirstVer(pTq->pWal);
-      } else if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__LATEST) {
-        fetchOffset = walGetCommittedVer(pTq->pWal);
-      } else if (pReq->currentOffset == TMQ_CONF__RESET_OFFSET__NONE) {
-        tqError("tmq poll: no offset committed for consumer %ld in vg %d, subkey %s", consumerId, TD_VID(pTq->pVnode),
-                pReq->subKey);
-        terrno = TSDB_CODE_TQ_NO_COMMITTED_OFFSET;
-        return -1;
-      }
-      tqDebug("consumer %ld, restore offset of %s on vg %d failed, config is %ld, set to %ld", consumerId, pReq->subKey,
-              TD_VID(pTq->pVnode), pReq->currentOffset, fetchOffset);
-    }
-  }
-
-  tqDebug("tmq poll: consumer %ld (epoch %d) recv poll req in vg %d, req offset %ld fetch offset %ld", consumerId,
-          pReq->epoch, TD_VID(pTq->pVnode), pReq->currentOffset, fetchOffset);
-
-  STqHandle* pHandle = taosHashGet(pTq->handles, pReq->subKey, strlen(pReq->subKey));
-  /*ASSERT(pHandle);*/
-  if (pHandle == NULL) {
-    tqError("tmq poll: no consumer handle for consumer %ld in vg %d, subkey %s", consumerId, TD_VID(pTq->pVnode),
-            pReq->subKey);
-    return -1;
-  }
-
-  if (pHandle->consumerId != consumerId) {
-    tqError("tmq poll: consumer handle mismatch for consumer %ld in vg %d, subkey %s, handle consumer id %ld",
-            consumerId, TD_VID(pTq->pVnode), pReq->subKey, pHandle->consumerId);
-    return -1;
-  }
-
-  int32_t consumerEpoch = atomic_load_32(&pHandle->epoch);
-  while (consumerEpoch < reqEpoch) {
-    consumerEpoch = atomic_val_compare_exchange_32(&pHandle->epoch, consumerEpoch, reqEpoch);
-  }
-
-  SMqDataBlkRsp rsp = {0};
-  rsp.reqOffset = pReq->currentOffset;
-
-  rsp.blockData = taosArrayInit(0, sizeof(void*));
-  rsp.blockDataLen = taosArrayInit(0, sizeof(int32_t));
-
-  if (rsp.blockData == NULL || rsp.blockDataLen == NULL) {
-    return -1;
-  }
-
-  rsp.withTbName = pReq->withTbName;
-  if (rsp.withTbName) {
-    rsp.blockTbName = taosArrayInit(0, sizeof(void*));
-  }
-
-  if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__COLUMN) {
-    rsp.withSchema = false;
-  } else {
-    rsp.withSchema = true;
-    rsp.blockSchema = taosArrayInit(0, sizeof(void*));
-  }
-
-#if 1
-  if (pReq->useSnapshot) {
-    // TODO set ver into snapshot
-    int64_t lastVer = walGetCommittedVer(pTq->pWal);
-    if (rsp.reqOffset < lastVer) {
-      tqInfo("retrieve using snapshot req offset %ld last ver %ld", rsp.reqOffset, lastVer);
-      tqScanSnapshot(pTq, &pHandle->execHandle, &rsp, workerId);
-
-      if (rsp.blockNum != 0) {
-        rsp.withTbName = false;
-        rsp.rspOffset = lastVer;
-        tqInfo("direct send by snapshot req offset %ld rsp offset %ld", rsp.reqOffset, rsp.rspOffset);
-        fetchOffset = lastVer;
-        goto SEND_RSP;
-      }
-    }
-  }
-#endif
-
-  SWalHead* pHeadWithCkSum = taosMemoryMalloc(sizeof(SWalHead) + 2048);
-  if (pHeadWithCkSum == NULL) {
-    return -1;
-  }
-
-  walSetReaderCapacity(pHandle->pWalReader, 2048);
-
-  while (1) {
-    consumerEpoch = atomic_load_32(&pHandle->epoch);
-    if (consumerEpoch > reqEpoch) {
-      tqWarn("tmq poll: consumer %ld (epoch %d) vg %d offset %ld, found new consumer epoch %d, discard req epoch %d",
-             consumerId, pReq->epoch, TD_VID(pTq->pVnode), fetchOffset, consumerEpoch, reqEpoch);
-      break;
-    }
-
-    if (tqFetchLog(pTq, pHandle, &fetchOffset, &pHeadWithCkSum) < 0) {
-      // TODO add push mgr
-      break;
-    }
-
-    SWalCont* pHead = &pHeadWithCkSum->head;
-
-    tqDebug("tmq poll: consumer %ld (epoch %d) iter log, vg %d offset %ld msgType %d", consumerId, pReq->epoch,
-            TD_VID(pTq->pVnode), fetchOffset, pHead->msgType);
-
-    if (pHead->msgType == TDMT_VND_SUBMIT) {
-      SSubmitReq* pCont = (SSubmitReq*)&pHead->body;
-
-      if (tqDataExec(pTq, &pHandle->execHandle, pCont, &rsp, workerId) < 0) {
-        /*ASSERT(0);*/
-      }
-    } else {
-      ASSERT(pHandle->fetchMeta);
-      ASSERT(IS_META_MSG(pHead->msgType));
-      tqInfo("fetch meta msg, ver: %ld, type: %d", pHead->version, pHead->msgType);
-      SMqMetaRsp metaRsp = {0};
-      metaRsp.reqOffset = pReq->currentOffset;
-      metaRsp.rspOffset = fetchOffset;
-      metaRsp.resMsgType = pHead->msgType;
-      metaRsp.metaRspLen = pHead->bodyLen;
-      metaRsp.metaRsp = pHead->body;
-      if (tqSendMetaPollRsp(pTq, pMsg, pReq, &metaRsp) < 0) {
-        code = -1;
-        goto OVER;
-      }
-      code = 0;
-      goto OVER;
-    }
-
-    // TODO batch optimization:
-    // TODO continue scan until meeting batch requirement
-    if (rsp.blockNum > 0 /* threshold */) {
-      break;
-    } else {
-      fetchOffset++;
-    }
-  }
-
-  taosMemoryFree(pHeadWithCkSum);
-
-SEND_RSP:
-  ASSERT(taosArrayGetSize(rsp.blockData) == rsp.blockNum);
-  ASSERT(taosArrayGetSize(rsp.blockDataLen) == rsp.blockNum);
-  if (rsp.withSchema) {
-    ASSERT(taosArrayGetSize(rsp.blockSchema) == rsp.blockNum);
-  }
-
-  rsp.rspOffset = fetchOffset;
-
-  if (tqSendDataRsp(pTq, pMsg, pReq, &rsp) < 0) {
-    code = -1;
-  }
-OVER:
-  // TODO wrap in destroy func
-  taosArrayDestroy(rsp.blockDataLen);
-  taosArrayDestroyP(rsp.blockData, (FDelete)taosMemoryFree);
-
-  if (rsp.withSchema) {
-    taosArrayDestroyP(rsp.blockSchema, (FDelete)tDeleteSSchemaWrapper);
-  }
-
-  if (rsp.withTbName) {
-    taosArrayDestroyP(rsp.blockTbName, (FDelete)taosMemoryFree);
-  }
-
-  return code;
-}
-#endif
-
 int32_t tqProcessVgDeleteReq(STQ* pTq, char* msg, int32_t msgLen) {
   SMqVDeleteReq* pReq = (SMqVDeleteReq*)msg;
 

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

@@ -69,6 +69,9 @@ int vnodeBegin(SVnode *pVnode) {
     }
   }
 
+  // begin sma
+  smaBegin(pVnode->pSma);  // TODO: refactor to include the rsma1/rsma2 tsdbBegin() after tsdb_refact branch merged
+
   return 0;
 }
 
@@ -230,8 +233,8 @@ int vnodeCommit(SVnode *pVnode) {
   }
 
   // preCommit
-  // TODO
+  smaPreCommit(pVnode->pSma);
-  
+
   // commit each sub-system
   if (metaCommit(pVnode->pMeta) < 0) {
     ASSERT(0);

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

@@ -152,12 +152,11 @@ SVnode *vnodeOpen(const char *path, STfs *pTfs, SMsgCb msgCb) {
   return pVnode;
 
 _err:
-  if (pVnode->pSma) smaCloseEnv(pVnode->pSma);
   if (pVnode->pQuery) vnodeQueryClose(pVnode);
   if (pVnode->pTq) tqClose(pVnode->pTq);
   if (pVnode->pWal) walClose(pVnode->pWal);
   if (pVnode->pTsdb) tsdbClose(&pVnode->pTsdb);
-  if (pVnode->pSma) smaCloseEx(pVnode->pSma);
+  if (pVnode->pSma) smaClose(pVnode->pSma);
   if (pVnode->pMeta) metaClose(pVnode->pMeta);
 
   tsem_destroy(&(pVnode->canCommit));
@@ -167,14 +166,13 @@ _err:
 
 void vnodeClose(SVnode *pVnode) {
   if (pVnode) {
-    smaCloseEnv(pVnode->pSma);
     vnodeCommit(pVnode);
     vnodeSyncClose(pVnode);
     vnodeQueryClose(pVnode);
     walClose(pVnode->pWal);
     tqClose(pVnode->pTq);
     if (pVnode->pTsdb) tsdbClose(&pVnode->pTsdb);
-    smaCloseEx(pVnode->pSma);
+    smaClose(pVnode->pSma);
     metaClose(pVnode->pMeta);
     vnodeCloseBufPool(pVnode);
     // destroy handle

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

@@ -256,70 +256,133 @@ int32_t vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
 
     SRpcMsg *pRpcMsg = pMsg;
 
-    if (pRpcMsg->msgType == TDMT_SYNC_TIMEOUT) {
+    // ToDo: ugly! use function pointer
-      SyncTimeout *pSyncMsg = syncTimeoutFromRpcMsg2(pRpcMsg);
+    // use different strategy
-      assert(pSyncMsg != NULL);
+    if (syncNodeStrategy(pSyncNode) == SYNC_STRATEGY_NO_SNAPSHOT) {
+      if (pRpcMsg->msgType == TDMT_SYNC_TIMEOUT) {
+        SyncTimeout *pSyncMsg = syncTimeoutFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnTimeoutCb(pSyncNode, pSyncMsg);
+        syncTimeoutDestroy(pSyncMsg);
 
-      ret = syncNodeOnTimeoutCb(pSyncNode, pSyncMsg);
+      } else if (pRpcMsg->msgType == TDMT_SYNC_PING) {
-      syncTimeoutDestroy(pSyncMsg);
+        SyncPing *pSyncMsg = syncPingFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnPingCb(pSyncNode, pSyncMsg);
+        syncPingDestroy(pSyncMsg);
 
-    } else if (pRpcMsg->msgType == TDMT_SYNC_PING) {
+      } else if (pRpcMsg->msgType == TDMT_SYNC_PING_REPLY) {
-      SyncPing *pSyncMsg = syncPingFromRpcMsg2(pRpcMsg);
+        SyncPingReply *pSyncMsg = syncPingReplyFromRpcMsg2(pRpcMsg);
-      assert(pSyncMsg != NULL);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnPingReplyCb(pSyncNode, pSyncMsg);
+        syncPingReplyDestroy(pSyncMsg);
 
-      ret = syncNodeOnPingCb(pSyncNode, pSyncMsg);
+      } else if (pRpcMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) {
-      syncPingDestroy(pSyncMsg);
+        SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL);
+        syncClientRequestDestroy(pSyncMsg);
 
-    } else if (pRpcMsg->msgType == TDMT_SYNC_PING_REPLY) {
+      } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE) {
-      SyncPingReply *pSyncMsg = syncPingReplyFromRpcMsg2(pRpcMsg);
+        SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pRpcMsg);
-      assert(pSyncMsg != NULL);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnRequestVoteCb(pSyncNode, pSyncMsg);
+        syncRequestVoteDestroy(pSyncMsg);
 
-      ret = syncNodeOnPingReplyCb(pSyncNode, pSyncMsg);
+      } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE_REPLY) {
-      syncPingReplyDestroy(pSyncMsg);
+        SyncRequestVoteReply *pSyncMsg = syncRequestVoteReplyFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnRequestVoteReplyCb(pSyncNode, pSyncMsg);
+        syncRequestVoteReplyDestroy(pSyncMsg);
 
-    } else if (pRpcMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) {
+      } else if (pRpcMsg->msgType == TDMT_SYNC_APPEND_ENTRIES) {
-      SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pRpcMsg);
+        SyncAppendEntries *pSyncMsg = syncAppendEntriesFromRpcMsg2(pRpcMsg);
-      assert(pSyncMsg != NULL);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnAppendEntriesCb(pSyncNode, pSyncMsg);
+        syncAppendEntriesDestroy(pSyncMsg);
 
-      ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL);
+      } else if (pRpcMsg->msgType == TDMT_SYNC_APPEND_ENTRIES_REPLY) {
-      syncClientRequestDestroy(pSyncMsg);
+        SyncAppendEntriesReply *pSyncMsg = syncAppendEntriesReplyFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnAppendEntriesReplyCb(pSyncNode, pSyncMsg);
+        syncAppendEntriesReplyDestroy(pSyncMsg);
 
-    } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE) {
+      } else if (pRpcMsg->msgType == TDMT_SYNC_SET_VNODE_STANDBY) {
-      SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pRpcMsg);
+        ret = vnodeSetStandBy(pVnode);
-      assert(pSyncMsg != NULL);
+        if (ret != 0 && terrno != 0) ret = terrno;
+        SRpcMsg rsp = {.code = ret, .info = pMsg->info};
+        tmsgSendRsp(&rsp);
+      } else {
+        vError("==vnodeProcessSyncReq== error msg type:%d", pRpcMsg->msgType);
+        ret = -1;
+      }
 
-      ret = syncNodeOnRequestVoteCb(pSyncNode, pSyncMsg);
-      syncRequestVoteDestroy(pSyncMsg);
-
-    } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE_REPLY) {
-      SyncRequestVoteReply *pSyncMsg = syncRequestVoteReplyFromRpcMsg2(pRpcMsg);
-      assert(pSyncMsg != NULL);
-
-      ret = syncNodeOnRequestVoteReplyCb(pSyncNode, pSyncMsg);
-      syncRequestVoteReplyDestroy(pSyncMsg);
-
-    } else if (pRpcMsg->msgType == TDMT_SYNC_APPEND_ENTRIES) {
-      SyncAppendEntries *pSyncMsg = syncAppendEntriesFromRpcMsg2(pRpcMsg);
-      assert(pSyncMsg != NULL);
-
-      ret = syncNodeOnAppendEntriesCb(pSyncNode, pSyncMsg);
-      syncAppendEntriesDestroy(pSyncMsg);
-
-    } else if (pRpcMsg->msgType == TDMT_SYNC_APPEND_ENTRIES_REPLY) {
-      SyncAppendEntriesReply *pSyncMsg = syncAppendEntriesReplyFromRpcMsg2(pRpcMsg);
-      assert(pSyncMsg != NULL);
-
-      ret = syncNodeOnAppendEntriesReplyCb(pSyncNode, pSyncMsg);
-      syncAppendEntriesReplyDestroy(pSyncMsg);
-
-    } else if (pRpcMsg->msgType == TDMT_SYNC_SET_VNODE_STANDBY) {
-      ret = vnodeSetStandBy(pVnode);
-      if (ret != 0 && terrno != 0) ret = terrno;
-      SRpcMsg rsp = {.code = ret, .info = pMsg->info};
-      tmsgSendRsp(&rsp);
     } else {
-      vError("==vnodeProcessSyncReq== error msg type:%d", pRpcMsg->msgType);
+      // use wal first strategy
-      ret = -1;
+
+      if (pRpcMsg->msgType == TDMT_SYNC_TIMEOUT) {
+        SyncTimeout *pSyncMsg = syncTimeoutFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnTimeoutCb(pSyncNode, pSyncMsg);
+        syncTimeoutDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_PING) {
+        SyncPing *pSyncMsg = syncPingFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnPingCb(pSyncNode, pSyncMsg);
+        syncPingDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_PING_REPLY) {
+        SyncPingReply *pSyncMsg = syncPingReplyFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnPingReplyCb(pSyncNode, pSyncMsg);
+        syncPingReplyDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) {
+        SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL);
+        syncClientRequestDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_CLIENT_REQUEST_BATCH) {
+        SyncClientRequestBatch *pSyncMsg = syncClientRequestBatchFromRpcMsg(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnClientRequestBatchCb(pSyncNode, pSyncMsg);
+        syncClientRequestBatchDestroyDeep(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE) {
+        SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnRequestVoteCb(pSyncNode, pSyncMsg);
+        syncRequestVoteDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE_REPLY) {
+        SyncRequestVoteReply *pSyncMsg = syncRequestVoteReplyFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnRequestVoteReplyCb(pSyncNode, pSyncMsg);
+        syncRequestVoteReplyDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_APPEND_ENTRIES_BATCH) {
+        SyncAppendEntriesBatch *pSyncMsg = syncAppendEntriesBatchFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnAppendEntriesSnapshot2Cb(pSyncNode, pSyncMsg);
+        syncAppendEntriesBatchDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_APPEND_ENTRIES_REPLY) {
+        SyncAppendEntriesReply *pSyncMsg = syncAppendEntriesReplyFromRpcMsg2(pRpcMsg);
+        ASSERT(pSyncMsg != NULL);
+        ret = syncNodeOnAppendEntriesReplySnapshot2Cb(pSyncNode, pSyncMsg);
+        syncAppendEntriesReplyDestroy(pSyncMsg);
+
+      } else if (pRpcMsg->msgType == TDMT_SYNC_SET_VNODE_STANDBY) {
+        ret = vnodeSetStandBy(pVnode);
+        if (ret != 0 && terrno != 0) ret = terrno;
+        SRpcMsg rsp = {.code = ret, .info = pMsg->info};
+        tmsgSendRsp(&rsp);
+      } else {
+        vError("==vnodeProcessSyncReq== error msg type:%d", pRpcMsg->msgType);
+        ret = -1;
+      }
     }
 
     syncNodeRelease(pSyncNode);
@@ -415,7 +478,7 @@ static int32_t vnodeSnapshotStopRead(struct SSyncFSM *pFsm, void *pReader) { ret
 
 static int32_t vnodeSnapshotDoRead(struct SSyncFSM *pFsm, void *pReader, void **ppBuf, int32_t *len) { return 0; }
 
-static int32_t vnodeSnapshotStartWrite(struct SSyncFSM *pFsm, void **ppWriter) { return 0; }
+static int32_t vnodeSnapshotStartWrite(struct SSyncFSM *pFsm, void *pParam, void **ppWriter) { return 0; }
 
 static int32_t vnodeSnapshotStopWrite(struct SSyncFSM *pFsm, void *pWriter, bool isApply) { return 0; }
 
@@ -442,7 +505,8 @@ static SSyncFSM *vnodeSyncMakeFsm(SVnode *pVnode) {
 
 int32_t vnodeSyncOpen(SVnode *pVnode, char *path) {
   SSyncInfo syncInfo = {
-      .snapshotEnable = false,
+      .snapshotStrategy = SYNC_STRATEGY_NO_SNAPSHOT,
+      .batchSize = 10,
       .vgId = pVnode->config.vgId,
       .isStandBy = pVnode->config.standby,
       .syncCfg = pVnode->config.syncCfg,

source/libs/command/inc/commandInt.h

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

source/libs/command/src/explain.c

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

source/libs/executor/inc/executorimpl.h

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

source/libs/executor/src/executorimpl.c

@@ -1422,7 +1422,7 @@ void setExecutionContext(SOperatorInfo* pOperator, int32_t numOfOutput, uint64_t
     return;
   }
 #ifdef BUF_PAGE_DEBUG
-  qDebug("page_setbuf, groupId:%"PRIu64, groupId);
+  qDebug("page_setbuf, groupId:%" PRIu64, groupId);
 #endif
   doSetTableGroupOutputBuf(pOperator, pAggInfo, numOfOutput, groupId);
 
@@ -1570,9 +1570,9 @@ int32_t doCopyToSDataBlock(SExecTaskInfo* pTaskInfo, SSDataBlock* pBlock, SExprI
 
     releaseBufPage(pBuf, page);
     pBlock->info.rows += pRow->numOfRows;
-//    if (pBlock->info.rows >= pBlock->info.capacity) {  // output buffer is full
+    //    if (pBlock->info.rows >= pBlock->info.capacity) {  // output buffer is full
-//      break;
+    //      break;
-//    }
+    //    }
   }
 
   qDebug("%s result generated, rows:%d, groupId:%" PRIu64, GET_TASKID(pTaskInfo), pBlock->info.rows,
@@ -2843,11 +2843,18 @@ int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t* order, int32_t* scan
 
 int32_t doPrepareScan(SOperatorInfo* pOperator, uint64_t uid, int64_t ts) {
   int32_t type = pOperator->operatorType;
+
+  pOperator->status = OP_OPENED;
+
   if (type == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN) {
     SStreamBlockScanInfo* pScanInfo = pOperator->info;
     pScanInfo->blockType = STREAM_INPUT__DATA_SCAN;
 
+    pScanInfo->pSnapshotReadOp->status = OP_OPENED;
+
     STableScanInfo* pInfo = pScanInfo->pSnapshotReadOp->info;
+    ASSERT(pInfo->scanMode == TABLE_SCAN__TABLE_ORDER);
+
     if (uid == 0) {
       pInfo->noTable = 1;
       return TSDB_CODE_SUCCESS;
@@ -2861,6 +2868,20 @@ int32_t doPrepareScan(SOperatorInfo* pOperator, uint64_t uid, int64_t ts) {
     pInfo->noTable = 0;
 
     if (pInfo->lastStatus.uid != uid || pInfo->lastStatus.ts != ts) {
+      SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
+
+      int32_t tableSz = taosArrayGetSize(pTaskInfo->tableqinfoList.pTableList);
+      bool    found = false;
+      for (int32_t i = 0; i < tableSz; i++) {
+        STableKeyInfo* pTableInfo = taosArrayGet(pTaskInfo->tableqinfoList.pTableList, i);
+        if (pTableInfo->uid == uid) {
+          found = true;
+          pInfo->currentTable = i;
+        }
+      }
+      // TODO after processing drop, found can be false
+      ASSERT(found);
+
       tsdbSetTableId(pInfo->dataReader, uid);
       int64_t oldSkey = pInfo->cond.twindows[0].skey;
       pInfo->cond.twindows[0].skey = ts + 1;
@@ -2868,7 +2889,11 @@ int32_t doPrepareScan(SOperatorInfo* pOperator, uint64_t uid, int64_t ts) {
       pInfo->cond.twindows[0].skey = oldSkey;
       pInfo->scanTimes = 0;
       pInfo->curTWinIdx = 0;
+
+      qDebug("tsdb reader offset seek to uid %ld ts %ld, table cur set to %d , all table num %d", uid, ts,
+             pInfo->currentTable, tableSz);
     }
+
     return TSDB_CODE_SUCCESS;
 
   } else {
@@ -4107,8 +4132,8 @@ int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle,
       } else {
         isNull[index++] = 0;
         char* data = nodesGetValueFromNode(pValue);
-        if (pValue->node.resType.type == TSDB_DATA_TYPE_JSON){
+        if (pValue->node.resType.type == TSDB_DATA_TYPE_JSON) {
-          if(tTagIsJson(data)){
+          if (tTagIsJson(data)) {
             terrno = TSDB_CODE_QRY_JSON_IN_GROUP_ERROR;
             taosMemoryFree(keyBuf);
             nodesClearList(groupNew);
@@ -4173,7 +4198,7 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
     } else if (QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN == type) {
       STableMergeScanPhysiNode* pTableScanNode = (STableMergeScanPhysiNode*)pPhyNode;
       int32_t code = createScanTableListInfo(pTableScanNode, pHandle, pTableListInfo, queryId, taskId);
-      if(code){
+      if (code) {
         pTaskInfo->code = code;
         return NULL;
       }
@@ -4202,7 +4227,7 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
       };
       if (pHandle) {
         int32_t code = createScanTableListInfo(pTableScanNode, pHandle, pTableListInfo, queryId, taskId);
-        if(code){
+        if (code) {
           pTaskInfo->code = code;
           return NULL;
         }

source/libs/executor/src/joinoperator.c

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

source/libs/executor/src/scanoperator.c

@@ -1329,6 +1329,13 @@ static void destroySysScanOperator(void* param, int32_t numOfOutput) {
   taosArrayDestroy(pInfo->scanCols);
 }
 
+static int32_t getSysTableDbNameColId(const char* pTable) {
+  // if (0 == strcmp(TSDB_INS_TABLE_USER_INDEXES, pTable)) {
+  //   return 1;
+  // }
+  return TSDB_INS_USER_STABLES_DBNAME_COLID;
+}
+
 EDealRes getDBNameFromConditionWalker(SNode* pNode, void* pContext) {
   int32_t   code = TSDB_CODE_SUCCESS;
   ENodeType nType = nodeType(pNode);
@@ -1350,7 +1357,7 @@ EDealRes getDBNameFromConditionWalker(SNode* pNode, void* pContext) {
       }
 
       SColumnNode* node = (SColumnNode*)pNode;
-      if (TSDB_INS_USER_STABLES_DBNAME_COLID == node->colId) {
+      if (getSysTableDbNameColId(node->tableName) == node->colId) {
         *(int32_t*)pContext = 2;
         return DEAL_RES_CONTINUE;
       }

source/libs/function/src/builtins.c

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

source/libs/function/src/builtinsimpl.c

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

source/libs/function/src/functionMgt.c

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

source/libs/function/src/tudf.c

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

source/libs/index/inc/indexFst.h

@@ -21,7 +21,7 @@ extern "C" {
 #endif
 
 #include "indexFstAutomation.h"
-#include "indexFstCountingWriter.h"
+#include "indexFstFile.h"
 #include "indexFstNode.h"
 #include "indexFstRegistry.h"
 #include "indexFstUtil.h"
@@ -90,8 +90,8 @@ FstBuilderNode*     fstUnFinishedNodesPopEmpty(FstUnFinishedNodes* nodes);
 uint64_t fstUnFinishedNodesFindCommPrefixAndSetOutput(FstUnFinishedNodes* node, FstSlice bs, Output in, Output* out);
 
 typedef struct FstBuilder {
-  FstCountingWriter*  wrt;         // The FST raw data is written directly to `wtr`.
+  IdxFstFile*         wrt;         // The FST raw data is written directly to `wtr`.
-  FstUnFinishedNodes* unfinished;  // The stack of unfinished nodes
+  FstUnFinishedNodes* unfinished;  // The stack of unfinished  nodes
   FstRegistry*        registry;    // A map of finished nodes.
   FstSlice            last;        // The last word added
   CompiledAddr        lastAddr;    // The address of the last compiled node
@@ -125,9 +125,9 @@ FstState fstStateCreateFrom(FstSlice* data, CompiledAddr addr);
 FstState fstStateCreate(State state);
 
 // compile
-void fstStateCompileForOneTransNext(FstCountingWriter* w, CompiledAddr addr, uint8_t inp);
+void fstStateCompileForOneTransNext(IdxFstFile* w, CompiledAddr addr, uint8_t inp);
-void fstStateCompileForOneTrans(FstCountingWriter* w, CompiledAddr addr, FstTransition* trn);
+void fstStateCompileForOneTrans(IdxFstFile* w, CompiledAddr addr, FstTransition* trn);
-void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuilderNode* node);
+void fstStateCompileForAnyTrans(IdxFstFile* w, CompiledAddr addr, FstBuilderNode* node);
 
 // set_comm_input
 void fstStateSetCommInput(FstState* state, uint8_t inp);
@@ -282,7 +282,7 @@ FStmSt* stmBuilderIntoStm(FStmBuilder* sb);
 bool fstVerify(Fst* fst);
 
 // refactor this function
-bool fstBuilderNodeCompileTo(FstBuilderNode* b, FstCountingWriter* wrt, CompiledAddr lastAddr, CompiledAddr startAddr);
+bool fstBuilderNodeCompileTo(FstBuilderNode* b, IdxFstFile* wrt, CompiledAddr lastAddr, CompiledAddr startAddr);
 
 typedef struct StreamState {
   FstNode*  node;

source/libs/index/inc/indexFstCountingWriter.h

@@ -1,96 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef __INDEX_FST_COUNTING_WRITER_H__
-#define __INDEX_FST_COUNTING_WRITER_H__
-
-#include "indexInt.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-//#define USE_MMAP 1
-
-#define DefaultMem 1024 * 1024
-
-static char tmpFile[] = "./index";
-typedef enum WriterType { TMemory, TFile } WriterType;
-
-typedef struct WriterCtx {
-  int (*write)(struct WriterCtx* ctx, uint8_t* buf, int len);
-  int (*read)(struct WriterCtx* ctx, uint8_t* buf, int len);
-  int (*flush)(struct WriterCtx* ctx);
-  int (*readFrom)(struct WriterCtx* ctx, uint8_t* buf, int len, int32_t offset);
-  int (*size)(struct WriterCtx* ctx);
-  WriterType type;
-  union {
-    struct {
-      TdFilePtr pFile;
-      bool readOnly;
-      char buf[256];
-      int  size;
-#ifdef USE_MMAP
-      char* ptr;
-#endif
-    } file;
-    struct {
-      int32_t capa;
-      char*   buf;
-    } mem;
-  };
-  int32_t offset;
-  int32_t limit;
-} WriterCtx;
-
-static int writeCtxDoWrite(WriterCtx* ctx, uint8_t* buf, int len);
-static int writeCtxDoRead(WriterCtx* ctx, uint8_t* buf, int len);
-static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t offset);
-static int writeCtxDoFlush(WriterCtx* ctx);
-
-WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity);
-void       writerCtxDestroy(WriterCtx* w, bool remove);
-
-typedef uint32_t CheckSummer;
-
-typedef struct FstCountingWriter {
-  void*       wrt;  // wrap any writer that counts and checksum bytes written
-  uint64_t    count;
-  CheckSummer summer;
-} FstCountingWriter;
-
-int fstCountingWriterWrite(FstCountingWriter* write, uint8_t* buf, uint32_t len);
-
-int fstCountingWriterRead(FstCountingWriter* write, uint8_t* buf, uint32_t len);
-
-int fstCountingWriterFlush(FstCountingWriter* write);
-
-uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write);
-
-FstCountingWriter* fstCountingWriterCreate(void* wtr);
-void               fstCountingWriterDestroy(FstCountingWriter* w);
-
-void    fstCountingWriterPackUintIn(FstCountingWriter* writer, uint64_t n, uint8_t nBytes);
-uint8_t fstCountingWriterPackUint(FstCountingWriter* writer, uint64_t n);
-
-#define FST_WRITER_COUNT(writer) (writer->count)
-#define FST_WRITER_INTER_WRITER(writer) (writer->wtr)
-#define FST_WRITE_CHECK_SUMMER(writer) (writer->summer)
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif

source/libs/index/inc/indexFstFile.h

@@ -0,0 +1,96 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __INDEX_FST_FILE_H__
+#define __INDEX_FST_FILE_H__
+
+#include "indexInt.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+//#define USE_MMAP 1
+
+#define DefaultMem 1024 * 1024
+
+static char tmpFile[] = "./index";
+typedef enum WriterType { TMemory, TFile } WriterType;
+
+typedef struct IFileCtx {
+  int (*write)(struct IFileCtx* ctx, uint8_t* buf, int len);
+  int (*read)(struct IFileCtx* ctx, uint8_t* buf, int len);
+  int (*flush)(struct IFileCtx* ctx);
+  int (*readFrom)(struct IFileCtx* ctx, uint8_t* buf, int len, int32_t offset);
+  int (*size)(struct IFileCtx* ctx);
+  WriterType type;
+  union {
+    struct {
+      TdFilePtr pFile;
+      bool      readOnly;
+      char      buf[256];
+      int64_t   size;
+#ifdef USE_MMAP
+      char* ptr;
+#endif
+    } file;
+    struct {
+      int32_t cap;
+      char*   buf;
+    } mem;
+  };
+  int32_t offset;
+  int32_t limit;
+} IFileCtx;
+
+static int idxFileCtxDoWrite(IFileCtx* ctx, uint8_t* buf, int len);
+static int idxFileCtxDoRead(IFileCtx* ctx, uint8_t* buf, int len);
+static int idxFileCtxDoReadFrom(IFileCtx* ctx, uint8_t* buf, int len, int32_t offset);
+static int idxFileCtxDoFlush(IFileCtx* ctx);
+
+IFileCtx* idxFileCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity);
+void      idxFileCtxDestroy(IFileCtx* w, bool remove);
+
+typedef uint32_t CheckSummer;
+
+typedef struct IdxFstFile {
+  void*       wrt;  // wrap any writer that counts and checksum bytes written
+  uint64_t    count;
+  CheckSummer summer;
+} IdxFstFile;
+
+int idxFileWrite(IdxFstFile* write, uint8_t* buf, uint32_t len);
+
+int idxFileRead(IdxFstFile* write, uint8_t* buf, uint32_t len);
+
+int idxFileFlush(IdxFstFile* write);
+
+uint32_t idxFileMaskedCheckSum(IdxFstFile* write);
+
+IdxFstFile* idxFileCreate(void* wtr);
+void        idxFileDestroy(IdxFstFile* w);
+
+void    idxFilePackUintIn(IdxFstFile* writer, uint64_t n, uint8_t nBytes);
+uint8_t idxFilePackUint(IdxFstFile* writer, uint64_t n);
+
+#define FST_WRITER_COUNT(writer)        (writer->count)
+#define FST_WRITER_INTER_WRITER(writer) (writer->wtr)
+#define FST_WRITE_CHECK_SUMMER(writer)  (writer->summer)
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

source/libs/index/inc/indexFstNode.h

@@ -20,12 +20,12 @@
 extern "C" {
 #endif
 
-#include "indexFstCountingWriter.h"
+#include "indexFstFile.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 
-#define FST_BUILDER_NODE_IS_FINAL(bn) (bn->isFinal)
+#define FST_BUILDER_NODE_IS_FINAL(bn)           (bn->isFinal)
-#define FST_BUILDER_NODE_TRANS_ISEMPTY(bn) (taosArrayGetSize(bn->trans) == 0)
+#define FST_BUILDER_NODE_TRANS_ISEMPTY(bn)      (taosArrayGetSize(bn->trans) == 0)
 #define FST_BUILDER_NODE_FINALOUTPUT_ISZERO(bn) (bn->finalOutput == 0)
 
 typedef struct FstTransition {
@@ -46,7 +46,7 @@ FstBuilderNode* fstBuilderNodeClone(FstBuilderNode* src);
 
 void fstBuilderNodeCloneFrom(FstBuilderNode* dst, FstBuilderNode* src);
 
-// bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt,
+// bool fstBuilderNodeCompileTo(FstBuilderNode *b, IdxFile' *wrt,
 // CompiledAddr lastAddr, CompiledAddr startAddr);
 bool fstBuilderNodeEqual(FstBuilderNode* n1, FstBuilderNode* n2);
 

source/libs/index/inc/indexTfile.h

@@ -16,7 +16,7 @@
 #define __INDEX_TFILE_H__
 
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
+#include "indexFstFile.h"
 #include "indexInt.h"
 #include "indexTfile.h"
 #include "indexUtil.h"
@@ -59,7 +59,7 @@ typedef struct TFileCache {
 
 typedef struct TFileWriter {
   FstBuilder* fb;
-  WriterCtx*  ctx;
+  IFileCtx*   ctx;
   TFileHeader header;
   uint32_t    offset;
 } TFileWriter;
@@ -68,7 +68,7 @@ typedef struct TFileWriter {
 typedef struct TFileReader {
   T_REF_DECLARE()
   Fst*        fst;
-  WriterCtx*  ctx;
+  IFileCtx*   ctx;
   TFileHeader header;
   bool        remove;
 } TFileReader;
@@ -103,7 +103,7 @@ void         tfileCachePut(TFileCache* tcache, ICacheKey* key, TFileReader* read
 TFileReader* tfileGetReaderByCol(IndexTFile* tf, uint64_t suid, char* colName);
 
 TFileReader* tfileReaderOpen(char* path, uint64_t suid, int64_t version, const char* colName);
-TFileReader* tfileReaderCreate(WriterCtx* ctx);
+TFileReader* tfileReaderCreate(IFileCtx* ctx);
 void         tfileReaderDestroy(TFileReader* reader);
 int          tfileReaderSearch(TFileReader* reader, SIndexTermQuery* query, SIdxTRslt* tr);
 void         tfileReaderRef(TFileReader* reader);
@@ -111,7 +111,7 @@ void         tfileReaderUnRef(TFileReader* reader);
 
 TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int64_t version, const char* colName, uint8_t type);
 void         tfileWriterClose(TFileWriter* tw);
-TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header);
+TFileWriter* tfileWriterCreate(IFileCtx* ctx, TFileHeader* header);
 void         tfileWriterDestroy(TFileWriter* tw);
 int          tfileWriterPut(TFileWriter* tw, void* data, bool order);
 int          tfileWriterFinish(TFileWriter* tw);

source/libs/index/src/index.c

@@ -39,7 +39,7 @@
 #define INDEX_DATA_BIGINT_NULL    0x8000000000000000LL
 #define INDEX_DATA_TIMESTAMP_NULL TSDB_DATA_BIGINT_NULL
 
-#define INDEX_DATA_FLOAT_NULL    0x7FF00000           // it is an NAN
+#define INDEX_DATA_FLOAT_NULL    0x7FF00000            // it is an NAN
 #define INDEX_DATA_DOUBLE_NULL   0x7FFFFF0000000000LL  // an NAN
 #define INDEX_DATA_NCHAR_NULL    0xFFFFFFFF
 #define INDEX_DATA_BINARY_NULL   0xFF
@@ -614,7 +614,7 @@ static int idxGenTFile(SIndex* sIdx, IndexCache* cache, SArray* batch) {
   return ret;
 END:
   if (tw != NULL) {
-    writerCtxDestroy(tw->ctx, true);
+    idxFileCtxDestroy(tw->ctx, true);
     taosMemoryFree(tw);
   }
   return -1;

source/libs/index/src/indexFst.c

@@ -19,11 +19,11 @@
 #include "tchecksum.h"
 #include "tcoding.h"
 
-static void fstPackDeltaIn(FstCountingWriter* wrt, CompiledAddr nodeAddr, CompiledAddr transAddr, uint8_t nBytes) {
+static void fstPackDeltaIn(IdxFstFile* wrt, CompiledAddr nodeAddr, CompiledAddr transAddr, uint8_t nBytes) {
   CompiledAddr deltaAddr = (transAddr == EMPTY_ADDRESS) ? EMPTY_ADDRESS : nodeAddr - transAddr;
-  fstCountingWriterPackUintIn(wrt, deltaAddr, nBytes);
+  idxFilePackUintIn(wrt, deltaAddr, nBytes);
 }
-static uint8_t fstPackDetla(FstCountingWriter* wrt, CompiledAddr nodeAddr, CompiledAddr transAddr) {
+static uint8_t fstPackDetla(IdxFstFile* wrt, CompiledAddr nodeAddr, CompiledAddr transAddr) {
   uint8_t nBytes = packDeltaSize(nodeAddr, transAddr);
   fstPackDeltaIn(wrt, nodeAddr, transAddr, nBytes);
   return nBytes;
@@ -208,7 +208,7 @@ FstState fstStateCreate(State state) {
   return fstStateDict[idx];
 }
 // compile
-void fstStateCompileForOneTransNext(FstCountingWriter* w, CompiledAddr addr, uint8_t inp) {
+void fstStateCompileForOneTransNext(IdxFstFile* w, CompiledAddr addr, uint8_t inp) {
   FstState s = fstStateCreate(OneTransNext);
   fstStateSetCommInput(&s, inp);
 
@@ -216,21 +216,21 @@ void fstStateCompileForOneTransNext(FstCountingWriter* w, CompiledAddr addr, uin
   uint8_t v = fstStateCommInput(&s, &null);
   if (null) {
     // w->write_all(&[inp])
-    fstCountingWriterWrite(w, &inp, 1);
+    idxFileWrite(w, &inp, 1);
   }
-  fstCountingWriterWrite(w, &(s.val), 1);
+  idxFileWrite(w, &(s.val), 1);
   // w->write_all(&[s.val])
   return;
 }
-void fstStateCompileForOneTrans(FstCountingWriter* w, CompiledAddr addr, FstTransition* trn) {
+void fstStateCompileForOneTrans(IdxFstFile* w, CompiledAddr addr, FstTransition* trn) {
   Output    out = trn->out;
-  uint8_t   outPackSize = (out == 0 ? 0 : fstCountingWriterPackUint(w, out));
+  uint8_t   outPackSize = (out == 0 ? 0 : idxFilePackUint(w, out));
   uint8_t   transPackSize = fstPackDetla(w, addr, trn->addr);
   PackSizes packSizes = 0;
 
   FST_SET_OUTPUT_PACK_SIZE(packSizes, outPackSize);
   FST_SET_TRANSITION_PACK_SIZE(packSizes, transPackSize);
-  fstCountingWriterWrite(w, (char*)&packSizes, sizeof(packSizes));
+  idxFileWrite(w, (char*)&packSizes, sizeof(packSizes));
 
   FstState st = fstStateCreate(OneTrans);
 
@@ -239,12 +239,12 @@ void fstStateCompileForOneTrans(FstCountingWriter* w, CompiledAddr addr, FstTran
   bool    null = false;
   uint8_t inp = fstStateCommInput(&st, &null);
   if (null == true) {
-    fstCountingWriterWrite(w, (char*)&trn->inp, sizeof(trn->inp));
+    idxFileWrite(w, (char*)&trn->inp, sizeof(trn->inp));
   }
-  fstCountingWriterWrite(w, (char*)(&(st.val)), sizeof(st.val));
+  idxFileWrite(w, (char*)(&(st.val)), sizeof(st.val));
   return;
 }
-void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuilderNode* node) {
+void fstStateCompileForAnyTrans(IdxFstFile* w, CompiledAddr addr, FstBuilderNode* node) {
   int32_t sz = taosArrayGetSize(node->trans);
   assert(sz <= 256);
 
@@ -275,11 +275,11 @@ void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuil
 
   if (anyOuts) {
     if (FST_BUILDER_NODE_IS_FINAL(node)) {
-      fstCountingWriterPackUintIn(w, node->finalOutput, oSize);
+      idxFilePackUintIn(w, node->finalOutput, oSize);
     }
     for (int32_t i = sz - 1; i >= 0; i--) {
       FstTransition* t = taosArrayGet(node->trans, i);
-      fstCountingWriterPackUintIn(w, t->out, oSize);
+      idxFilePackUintIn(w, t->out, oSize);
     }
   }
   for (int32_t i = sz - 1; i >= 0; i--) {
@@ -288,7 +288,7 @@ void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuil
   }
   for (int32_t i = sz - 1; i >= 0; i--) {
     FstTransition* t = taosArrayGet(node->trans, i);
-    fstCountingWriterWrite(w, (char*)&t->inp, 1);
+    idxFileWrite(w, (char*)&t->inp, 1);
     // fstPackDeltaIn(w, addr, t->addr, tSize);
   }
   if (sz > TRANS_INDEX_THRESHOLD) {
@@ -306,10 +306,10 @@ void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuil
       index[t->inp] = i;
       // fstPackDeltaIn(w, addr, t->addr, tSize);
     }
-    fstCountingWriterWrite(w, (char*)index, 256);
+    idxFileWrite(w, (char*)index, 256);
     taosMemoryFree(index);
   }
-  fstCountingWriterWrite(w, (char*)&packSizes, 1);
+  idxFileWrite(w, (char*)&packSizes, 1);
   bool null = false;
   fstStateStateNtrans(&st, &null);
   if (null == true) {
@@ -318,12 +318,12 @@ void fstStateCompileForAnyTrans(FstCountingWriter* w, CompiledAddr addr, FstBuil
     // encoded in the state byte.
     uint8_t v = 1;
     if (sz == 256) {
-      fstCountingWriterWrite(w, (char*)&v, 1);
+      idxFileWrite(w, (char*)&v, 1);
     } else {
-      fstCountingWriterWrite(w, (char*)&sz, 1);
+      idxFileWrite(w, (char*)&sz, 1);
     }
   }
-  fstCountingWriterWrite(w, (char*)(&(st.val)), 1);
+  idxFileWrite(w, (char*)(&(st.val)), 1);
   return;
 }
 
@@ -753,7 +753,7 @@ bool fstNodeCompile(FstNode* node, void* w, CompiledAddr lastAddr, CompiledAddr
   return true;
 }
 
-bool fstBuilderNodeCompileTo(FstBuilderNode* b, FstCountingWriter* wrt, CompiledAddr lastAddr, CompiledAddr startAddr) {
+bool fstBuilderNodeCompileTo(FstBuilderNode* b, IdxFstFile* wrt, CompiledAddr lastAddr, CompiledAddr startAddr) {
   return fstNodeCompile(NULL, wrt, lastAddr, startAddr, b);
 }
 
@@ -763,7 +763,7 @@ FstBuilder* fstBuilderCreate(void* w, FstType ty) {
     return b;
   }
 
-  b->wrt = fstCountingWriterCreate(w);
+  b->wrt = idxFileCreate(w);
   b->unfinished = fstUnFinishedNodesCreate();
   b->registry = fstRegistryCreate(10000, 2);
   b->last = fstSliceCreate(NULL, 0);
@@ -773,12 +773,12 @@ FstBuilder* fstBuilderCreate(void* w, FstType ty) {
   char  buf64[8] = {0};
   void* pBuf64 = buf64;
   taosEncodeFixedU64(&pBuf64, VERSION);
-  fstCountingWriterWrite(b->wrt, buf64, sizeof(buf64));
+  idxFileWrite(b->wrt, buf64, sizeof(buf64));
 
   pBuf64 = buf64;
   memset(buf64, 0, sizeof(buf64));
   taosEncodeFixedU64(&pBuf64, ty);
-  fstCountingWriterWrite(b->wrt, buf64, sizeof(buf64));
+  idxFileWrite(b->wrt, buf64, sizeof(buf64));
 
   return b;
 }
@@ -787,7 +787,7 @@ void fstBuilderDestroy(FstBuilder* b) {
     return;
   }
 
-  fstCountingWriterDestroy(b->wrt);
+  idxFileDestroy(b->wrt);
   fstUnFinishedNodesDestroy(b->unfinished);
   fstRegistryDestroy(b->registry);
   fstSliceDestroy(&b->last);
@@ -905,21 +905,19 @@ void* fstBuilderInsertInner(FstBuilder* b) {
 
   void* pBuf64 = buf64;
   taosEncodeFixedU64(&pBuf64, b->len);
-  fstCountingWriterWrite(b->wrt, buf64, sizeof(buf64));
+  idxFileWrite(b->wrt, buf64, sizeof(buf64));
 
   pBuf64 = buf64;
   taosEncodeFixedU64(&pBuf64, rootAddr);
-  fstCountingWriterWrite(b->wrt, buf64, sizeof(buf64));
+  idxFileWrite(b->wrt, buf64, sizeof(buf64));
 
   char     buf32[4] = {0};
   void*    pBuf32 = buf32;
-  uint32_t sum = fstCountingWriterMaskedCheckSum(b->wrt);
+  uint32_t sum = idxFileMaskedCheckSum(b->wrt);
   taosEncodeFixedU32(&pBuf32, sum);
-  fstCountingWriterWrite(b->wrt, buf32, sizeof(buf32));
+  idxFileWrite(b->wrt, buf32, sizeof(buf32));
 
-  fstCountingWriterFlush(b->wrt);
+  idxFileFlush(b->wrt);
-  // fstCountingWriterDestroy(b->wrt);
-  // b->wrt = NULL;
   return b->wrt;
 }
 void fstBuilderFinish(FstBuilder* b) { fstBuilderInsertInner(b); }

source/libs/index/src/indexFstDfa.c

@@ -61,9 +61,10 @@ void dfaBuilderDestroy(FstDfaBuilder *builder) {
     pIter = taosHashIterate(builder->cache, pIter);
   }
   taosHashCleanup(builder->cache);
+  taosMemoryFree(builder);
 }
 
-FstDfa *dfaBuilderBuild(FstDfaBuilder *builder) {
+FstDfa *dfaBuilder(FstDfaBuilder *builder) {
   uint32_t      sz = taosArrayGetSize(builder->dfa->insts);
   FstSparseSet *cur = sparSetCreate(sz);
   FstSparseSet *nxt = sparSetCreate(sz);

source/libs/index/src/indexFstFile.c

@@ -13,13 +13,13 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#include "indexFstCountingWriter.h"
+#include "indexFstFile.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 #include "os.h"
 #include "tutil.h"
 
-static int writeCtxDoWrite(WriterCtx* ctx, uint8_t* buf, int len) {
+static int idxFileCtxDoWrite(IFileCtx* ctx, uint8_t* buf, int len) {
   if (ctx->type == TFile) {
     assert(len == taosWriteFile(ctx->file.pFile, buf, len));
   } else {
@@ -28,7 +28,7 @@ static int writeCtxDoWrite(WriterCtx* ctx, uint8_t* buf, int len) {
   ctx->offset += len;
   return len;
 }
-static int writeCtxDoRead(WriterCtx* ctx, uint8_t* buf, int len) {
+static int idxFileCtxDoRead(IFileCtx* ctx, uint8_t* buf, int len) {
   int nRead = 0;
   if (ctx->type == TFile) {
 #ifdef USE_MMAP
@@ -44,7 +44,7 @@ static int writeCtxDoRead(WriterCtx* ctx, uint8_t* buf, int len) {
 
   return nRead;
 }
-static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t offset) {
+static int idxFileCtxDoReadFrom(IFileCtx* ctx, uint8_t* buf, int len, int32_t offset) {
   int nRead = 0;
   if (ctx->type == TFile) {
     // tfLseek(ctx->file.pFile, offset, 0);
@@ -61,7 +61,7 @@ static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t off
   }
   return nRead;
 }
-static int writeCtxGetSize(WriterCtx* ctx) {
+static int idxFileCtxGetSize(IFileCtx* ctx) {
   if (ctx->type == TFile) {
     int64_t file_size = 0;
     taosStatFile(ctx->file.buf, &file_size, NULL);
@@ -69,7 +69,7 @@ static int writeCtxGetSize(WriterCtx* ctx) {
   }
   return 0;
 }
-static int writeCtxDoFlush(WriterCtx* ctx) {
+static int idxFileCtxDoFlush(IFileCtx* ctx) {
   if (ctx->type == TFile) {
     // taosFsyncFile(ctx->file.pFile);
     taosFsyncFile(ctx->file.pFile);
@@ -80,8 +80,8 @@ static int writeCtxDoFlush(WriterCtx* ctx) {
   return 1;
 }
 
-WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity) {
+IFileCtx* idxFileCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity) {
-  WriterCtx* ctx = taosMemoryCalloc(1, sizeof(WriterCtx));
+  IFileCtx* ctx = taosMemoryCalloc(1, sizeof(IFileCtx));
   if (ctx == NULL) {
     return NULL;
   }
@@ -90,39 +90,36 @@ WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int
   if (ctx->type == TFile) {
     // ugly code, refactor later
     ctx->file.readOnly = readOnly;
+    memcpy(ctx->file.buf, path, strlen(path));
     if (readOnly == false) {
-      // ctx->file.pFile = open(path, O_WRONLY | O_CREAT | O_APPEND, S_IRWXU | S_IRWXG | S_IRWXO);
       ctx->file.pFile = taosOpenFile(path, TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_APPEND);
       taosFtruncateFile(ctx->file.pFile, 0);
-      int64_t file_size;
+      taosStatFile(path, &ctx->file.size, NULL);
-      taosStatFile(path, &file_size, NULL);
+      // ctx->file.size = (int)size;
-      ctx->file.size = (int)file_size;
 
     } else {
-      // ctx->file.pFile = open(path, O_RDONLY, S_IRWXU | S_IRWXG | S_IRWXO);
       ctx->file.pFile = taosOpenFile(path, TD_FILE_READ);
 
-      int64_t file_size = 0;
+      int64_t size = 0;
-      taosFStatFile(ctx->file.pFile, &file_size, NULL);
+      taosFStatFile(ctx->file.pFile, &ctx->file.size, NULL);
-      ctx->file.size = (int)file_size;
+      ctx->file.size = (int)size;
 #ifdef USE_MMAP
       ctx->file.ptr = (char*)tfMmapReadOnly(ctx->file.pFile, ctx->file.size);
 #endif
     }
-    memcpy(ctx->file.buf, path, strlen(path));
     if (ctx->file.pFile == NULL) {
       indexError("failed to open file, error %d", errno);
       goto END;
     }
   } else if (ctx->type == TMemory) {
     ctx->mem.buf = taosMemoryCalloc(1, sizeof(char) * capacity);
-    ctx->mem.capa = capacity;
+    ctx->mem.cap = capacity;
   }
-  ctx->write = writeCtxDoWrite;
+  ctx->write = idxFileCtxDoWrite;
-  ctx->read = writeCtxDoRead;
+  ctx->read = idxFileCtxDoRead;
-  ctx->flush = writeCtxDoFlush;
+  ctx->flush = idxFileCtxDoFlush;
-  ctx->readFrom = writeCtxDoReadFrom;
+  ctx->readFrom = idxFileCtxDoReadFrom;
-  ctx->size = writeCtxGetSize;
+  ctx->size = idxFileCtxGetSize;
 
   ctx->offset = 0;
   ctx->limit = capacity;
@@ -135,7 +132,7 @@ END:
   taosMemoryFree(ctx);
   return NULL;
 }
-void writerCtxDestroy(WriterCtx* ctx, bool remove) {
+void idxFileCtxDestroy(IFileCtx* ctx, bool remove) {
   if (ctx->type == TMemory) {
     taosMemoryFree(ctx->mem.buf);
   } else {
@@ -149,9 +146,6 @@ void writerCtxDestroy(WriterCtx* ctx, bool remove) {
     if (ctx->file.readOnly == false) {
       int64_t file_size = 0;
       taosStatFile(ctx->file.buf, &file_size, NULL);
-      // struct stat fstat;
-      // stat(ctx->file.buf, &fstat);
-      // indexError("write file size: %d", (int)(fstat.st_size));
     }
     if (remove) {
       unlink(ctx->file.buf);
@@ -160,30 +154,29 @@ void writerCtxDestroy(WriterCtx* ctx, bool remove) {
   taosMemoryFree(ctx);
 }
 
-FstCountingWriter* fstCountingWriterCreate(void* wrt) {
+IdxFstFile* idxFileCreate(void* wrt) {
-  FstCountingWriter* cw = taosMemoryCalloc(1, sizeof(FstCountingWriter));
+  IdxFstFile* cw = taosMemoryCalloc(1, sizeof(IdxFstFile));
   if (cw == NULL) {
     return NULL;
   }
 
   cw->wrt = wrt;
-  //(void *)(writerCtxCreate(TFile, readOnly));
   return cw;
 }
-void fstCountingWriterDestroy(FstCountingWriter* cw) {
+void idxFileDestroy(IdxFstFile* cw) {
   // free wrt object: close fd or free mem
-  fstCountingWriterFlush(cw);
+  idxFileFlush(cw);
-  // writerCtxDestroy((WriterCtx *)(cw->wrt));
+  // idxFileCtxDestroy((IFileCtx *)(cw->wrt));
   taosMemoryFree(cw);
 }
 
-int fstCountingWriterWrite(FstCountingWriter* write, uint8_t* buf, uint32_t len) {
+int idxFileWrite(IdxFstFile* write, uint8_t* buf, uint32_t len) {
   if (write == NULL) {
     return 0;
   }
   // update checksum
   // write data to file/socket or mem
-  WriterCtx* ctx = write->wrt;
+  IFileCtx* ctx = write->wrt;
 
   int nWrite = ctx->write(ctx, buf, len);
   assert(nWrite == len);
@@ -192,42 +185,41 @@ int fstCountingWriterWrite(FstCountingWriter* write, uint8_t* buf, uint32_t len)
   write->summer = taosCalcChecksum(write->summer, buf, len);
   return len;
 }
-int fstCountingWriterRead(FstCountingWriter* write, uint8_t* buf, uint32_t len) {
+int idxFileRead(IdxFstFile* write, uint8_t* buf, uint32_t len) {
   if (write == NULL) {
     return 0;
   }
-  WriterCtx* ctx = write->wrt;
+  IFileCtx* ctx = write->wrt;
-  int        nRead = ctx->read(ctx, buf, len);
+  int       nRead = ctx->read(ctx, buf, len);
   // assert(nRead == len);
   return nRead;
 }
 
-uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write) {
+uint32_t idxFileMaskedCheckSum(IdxFstFile* write) {
   // opt
   return write->summer;
 }
 
-int fstCountingWriterFlush(FstCountingWriter* write) {
+int idxFileFlush(IdxFstFile* write) {
-  WriterCtx* ctx = write->wrt;
+  IFileCtx* ctx = write->wrt;
   ctx->flush(ctx);
-  // write->wtr->flush
   return 1;
 }
 
-void fstCountingWriterPackUintIn(FstCountingWriter* writer, uint64_t n, uint8_t nBytes) {
+void idxFilePackUintIn(IdxFstFile* writer, uint64_t n, uint8_t nBytes) {
   assert(1 <= nBytes && nBytes <= 8);
   uint8_t* buf = taosMemoryCalloc(8, sizeof(uint8_t));
   for (uint8_t i = 0; i < nBytes; i++) {
     buf[i] = (uint8_t)n;
     n = n >> 8;
   }
-  fstCountingWriterWrite(writer, buf, nBytes);
+  idxFileWrite(writer, buf, nBytes);
   taosMemoryFree(buf);
   return;
 }
 
-uint8_t fstCountingWriterPackUint(FstCountingWriter* writer, uint64_t n) {
+uint8_t idxFilePackUint(IdxFstFile* writer, uint64_t n) {
   uint8_t nBytes = packSize(n);
-  fstCountingWriterPackUintIn(writer, n, nBytes);
+  idxFilePackUintIn(writer, n, nBytes);
   return nBytes;
 }

source/libs/index/src/indexFstNode.c

@@ -95,7 +95,7 @@ void fstBuilderNodeCloneFrom(FstBuilderNode* dst, FstBuilderNode* src) {
   }
 }
 
-// bool fstBuilderNodeCompileTo(FstBuilderNode *b, FstCountingWriter *wrt, CompiledAddr lastAddr, CompiledAddr
+// bool fstBuilderNodeCompileTo(FstBuilderNode *b, IdxFile *wrt, CompiledAddr lastAddr, CompiledAddr
 // startAddr) {
 
 // size_t sz = taosArrayGetSize(b->trans);

source/libs/index/src/indexFstUtil.c

@@ -75,7 +75,6 @@ CompiledAddr unpackDelta(char* data, uint64_t len, uint64_t nodeAddr) {
 }
 
 // fst slice func
-//
 
 FstSlice fstSliceCreate(uint8_t* data, uint64_t len) {
   FstString* str = (FstString*)taosMemoryMalloc(sizeof(FstString));
@@ -164,16 +163,3 @@ int fstSliceCompare(FstSlice* a, FstSlice* b) {
     return 0;
   }
 }
-
-// FstStack* fstStackCreate(size_t elemSize, StackFreeElem freeFn) {
-//  FstStack *s = taosMemoryCalloc(1, sizeof(FstStack));
-//  if (s == NULL) { return NULL; }
-//  s->
-//  s->freeFn
-//
-//}
-// void  *fstStackPush(FstStack *s, void *elem);
-// void  *fstStackTop(FstStack *s);
-// size_t fstStackLen(FstStack *s);
-// void  *fstStackGetAt(FstStack *s, size_t i);
-// void   fstStackDestory(FstStack *);

source/libs/index/src/indexTfile.c

@@ -16,7 +16,7 @@
 #include "index.h"
 #include "indexComm.h"
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
+#include "indexFstFile.h"
 #include "indexUtil.h"
 #include "taosdef.h"
 #include "taoserror.h"
@@ -103,7 +103,7 @@ TFileCache* tfileCacheCreate(const char* path) {
   for (size_t i = 0; i < taosArrayGetSize(files); i++) {
     char* file = taosArrayGetP(files, i);
 
-    WriterCtx* wc = writerCtxCreate(TFile, file, true, 1024 * 1024 * 64);
+    IFileCtx* wc = idxFileCtxCreate(TFile, file, true, 1024 * 1024 * 64);
     if (wc == NULL) {
       indexError("failed to open index:%s", file);
       goto End;
@@ -175,7 +175,7 @@ void tfileCachePut(TFileCache* tcache, ICacheKey* key, TFileReader* reader) {
   tfileReaderRef(reader);
   return;
 }
-TFileReader* tfileReaderCreate(WriterCtx* ctx) {
+TFileReader* tfileReaderCreate(IFileCtx* ctx) {
   TFileReader* reader = taosMemoryCalloc(1, sizeof(TFileReader));
   if (reader == NULL) {
     return NULL;
@@ -216,7 +216,7 @@ void tfileReaderDestroy(TFileReader* reader) {
   } else {
     indexInfo("%s is not removed", reader->ctx->file.buf);
   }
-  writerCtxDestroy(reader->ctx, reader->remove);
+  idxFileCtxDestroy(reader->ctx, reader->remove);
 
   taosMemoryFree(reader);
 }
@@ -490,7 +490,7 @@ TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int64_t version, const c
   char fullname[256] = {0};
   tfileGenFileFullName(fullname, path, suid, colName, version);
   // indexInfo("open write file name %s", fullname);
-  WriterCtx* wcx = writerCtxCreate(TFile, fullname, false, 1024 * 1024 * 64);
+  IFileCtx* wcx = idxFileCtxCreate(TFile, fullname, false, 1024 * 1024 * 64);
   if (wcx == NULL) {
     return NULL;
   }
@@ -507,18 +507,18 @@ TFileReader* tfileReaderOpen(char* path, uint64_t suid, int64_t version, const c
   char fullname[256] = {0};
   tfileGenFileFullName(fullname, path, suid, colName, version);
 
-  WriterCtx* wc = writerCtxCreate(TFile, fullname, true, 1024 * 1024 * 1024);
+  IFileCtx* wc = idxFileCtxCreate(TFile, fullname, true, 1024 * 1024 * 1024);
   if (wc == NULL) {
     terrno = TAOS_SYSTEM_ERROR(errno);
     indexError("failed to open readonly file: %s, reason: %s", fullname, terrstr());
     return NULL;
   }
-  indexTrace("open read file name:%s, file size: %d", wc->file.buf, wc->file.size);
+  indexTrace("open read file name:%s, file size: %" PRId64 "", wc->file.buf, wc->file.size);
 
   TFileReader* reader = tfileReaderCreate(wc);
   return reader;
 }
-TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header) {
+TFileWriter* tfileWriterCreate(IFileCtx* ctx, TFileHeader* header) {
   TFileWriter* tw = taosMemoryCalloc(1, sizeof(TFileWriter));
   if (tw == NULL) {
     indexError("index: %" PRIu64 " failed to alloc TFilerWriter", header->suid);
@@ -609,14 +609,14 @@ void tfileWriterClose(TFileWriter* tw) {
   if (tw == NULL) {
     return;
   }
-  writerCtxDestroy(tw->ctx, false);
+  idxFileCtxDestroy(tw->ctx, false);
   taosMemoryFree(tw);
 }
 void tfileWriterDestroy(TFileWriter* tw) {
   if (tw == NULL) {
     return;
   }
-  writerCtxDestroy(tw->ctx, false);
+  idxFileCtxDestroy(tw->ctx, false);
   taosMemoryFree(tw);
 }
 
@@ -892,8 +892,8 @@ static int tfileReaderLoadHeader(TFileReader* reader) {
   return 0;
 }
 static int tfileReaderLoadFst(TFileReader* reader) {
-  WriterCtx* ctx = reader->ctx;
+  IFileCtx* ctx = reader->ctx;
-  int        size = ctx->size(ctx);
+  int       size = ctx->size(ctx);
 
   // current load fst into memory, refactor it later
   int   fstSize = size - reader->header.fstOffset - sizeof(tfileMagicNumber);
@@ -905,8 +905,9 @@ static int tfileReaderLoadFst(TFileReader* reader) {
   int64_t ts = taosGetTimestampUs();
   int32_t nread = ctx->readFrom(ctx, buf, fstSize, reader->header.fstOffset);
   int64_t cost = taosGetTimestampUs() - ts;
-  indexInfo("nread = %d, and fst offset=%d, fst size: %d, filename: %s, file size: %d, time cost: %" PRId64 "us", nread,
+  indexInfo("nread = %d, and fst offset=%d, fst size: %d, filename: %s, file size: %" PRId64 ", time cost: %" PRId64
-            reader->header.fstOffset, fstSize, ctx->file.buf, ctx->file.size, cost);
+            "us",
+            nread, reader->header.fstOffset, fstSize, ctx->file.buf, ctx->file.size, cost);
   // we assuse fst size less than FST_MAX_SIZE
   assert(nread > 0 && nread <= fstSize);
 
@@ -919,7 +920,7 @@ static int tfileReaderLoadFst(TFileReader* reader) {
 }
 static int tfileReaderLoadTableIds(TFileReader* reader, int32_t offset, SArray* result) {
   // TODO(yihao): opt later
-  WriterCtx* ctx = reader->ctx;
+  IFileCtx* ctx = reader->ctx;
   // add block cache
   char    block[4096] = {0};
   int32_t nread = ctx->readFrom(ctx, block, sizeof(block), offset);
@@ -952,7 +953,7 @@ static int tfileReaderLoadTableIds(TFileReader* reader, int32_t offset, SArray*
 }
 static int tfileReaderVerify(TFileReader* reader) {
   // just validate header and Footer, file corrupted also shuild be verified later
-  WriterCtx* ctx = reader->ctx;
+  IFileCtx* ctx = reader->ctx;
 
   uint64_t tMagicNumber = 0;
 

source/libs/index/test/fstTest.cc

@@ -7,7 +7,6 @@
 #include "index.h"
 #include "indexCache.h"
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 #include "indexTfile.h"
@@ -20,7 +19,7 @@ class FstWriter {
  public:
   FstWriter() {
     taosRemoveFile(fileName.c_str());
-    _wc = writerCtxCreate(TFile, fileName.c_str(), false, 64 * 1024 * 1024);
+    _wc = idxFileCtxCreate(TFile, fileName.c_str(), false, 64 * 1024 * 1024);
     _b = fstBuilderCreate(_wc, 0);
   }
   bool Put(const std::string& key, uint64_t val) {
@@ -38,25 +37,25 @@ class FstWriter {
     fstBuilderFinish(_b);
     fstBuilderDestroy(_b);
 
-    writerCtxDestroy(_wc, false);
+    idxFileCtxDestroy(_wc, false);
   }
 
  private:
   FstBuilder* _b;
-  WriterCtx*  _wc;
+  IFileCtx*   _wc;
 };
 
 class FstReadMemory {
  public:
   FstReadMemory(int32_t size, const std::string& fileName = TD_TMP_DIR_PATH "tindex.tindex") {
-    _wc = writerCtxCreate(TFile, fileName.c_str(), true, 64 * 1024);
+    _wc = idxFileCtxCreate(TFile, fileName.c_str(), true, 64 * 1024);
-    _w = fstCountingWriterCreate(_wc);
+    _w = idxFileCreate(_wc);
     _size = size;
     memset((void*)&_s, 0, sizeof(_s));
   }
   bool init() {
     char* buf = (char*)taosMemoryCalloc(1, sizeof(char) * _size);
-    int   nRead = fstCountingWriterRead(_w, (uint8_t*)buf, _size);
+    int   nRead = idxFileRead(_w, (uint8_t*)buf, _size);
     if (nRead <= 0) {
       return false;
     }
@@ -141,18 +140,18 @@ class FstReadMemory {
   }
 
   ~FstReadMemory() {
-    fstCountingWriterDestroy(_w);
+    idxFileDestroy(_w);
     fstDestroy(_fst);
     fstSliceDestroy(&_s);
-    writerCtxDestroy(_wc, false);
+    idxFileCtxDestroy(_wc, false);
   }
 
  private:
-  FstCountingWriter* _w;
+  IdxFstFile* _w;
-  Fst*               _fst;
+  Fst*        _fst;
-  FstSlice           _s;
+  FstSlice    _s;
-  WriterCtx*         _wc;
+  IFileCtx*   _wc;
-  int32_t            _size;
+  int32_t     _size;
 };
 
 #define L 100

source/libs/index/test/fstUT.cc

@@ -8,7 +8,6 @@
 #include "index.h"
 #include "indexCache.h"
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 #include "indexTfile.h"
@@ -40,7 +39,7 @@ static void EnvCleanup() {}
 class FstWriter {
  public:
   FstWriter() {
-    _wc = writerCtxCreate(TFile, tindex, false, 64 * 1024 * 1024);
+    _wc = idxFileCtxCreate(TFile, tindex, false, 64 * 1024 * 1024);
     _b = fstBuilderCreate(_wc, 0);
   }
   bool Put(const std::string& key, uint64_t val) {
@@ -58,25 +57,25 @@ class FstWriter {
     fstBuilderFinish(_b);
     fstBuilderDestroy(_b);
 
-    writerCtxDestroy(_wc, false);
+    idxFileCtxDestroy(_wc, false);
   }
 
  private:
   FstBuilder* _b;
-  WriterCtx*  _wc;
+  IFileCtx*   _wc;
 };
 
 class FstReadMemory {
  public:
   FstReadMemory(size_t size) {
-    _wc = writerCtxCreate(TFile, tindex, true, 64 * 1024);
+    _wc = idxFileCtxCreate(TFile, tindex, true, 64 * 1024);
-    _w = fstCountingWriterCreate(_wc);
+    _w = idxFileCreate(_wc);
     _size = size;
     memset((void*)&_s, 0, sizeof(_s));
   }
   bool init() {
     char* buf = (char*)taosMemoryCalloc(1, sizeof(char) * _size);
-    int   nRead = fstCountingWriterRead(_w, (uint8_t*)buf, _size);
+    int   nRead = idxFileRead(_w, (uint8_t*)buf, _size);
     if (nRead <= 0) {
       return false;
     }
@@ -130,18 +129,18 @@ class FstReadMemory {
   }
 
   ~FstReadMemory() {
-    fstCountingWriterDestroy(_w);
+    idxFileDestroy(_w);
     fstDestroy(_fst);
     fstSliceDestroy(&_s);
-    writerCtxDestroy(_wc, false);
+    idxFileCtxDestroy(_wc, false);
   }
 
  private:
-  FstCountingWriter* _w;
+  IdxFstFile* _w;
-  Fst*               _fst;
+  Fst*        _fst;
-  FstSlice           _s;
+  FstSlice    _s;
-  WriterCtx*         _wc;
+  IFileCtx*   _wc;
-  size_t             _size;
+  size_t      _size;
 };
 
 class FstWriterEnv : public ::testing::Test {

source/libs/index/test/indexTests.cc

@@ -20,7 +20,6 @@
 #include "index.h"
 #include "indexCache.h"
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 #include "indexTfile.h"
@@ -51,7 +50,7 @@ class DebugInfo {
 class FstWriter {
  public:
   FstWriter() {
-    _wc = writerCtxCreate(TFile, TD_TMP_DIR_PATH "tindex", false, 64 * 1024 * 1024);
+    _wc = idxFileCtxCreate(TFile, TD_TMP_DIR_PATH "tindex", false, 64 * 1024 * 1024);
     _b = fstBuilderCreate(NULL, 0);
   }
   bool Put(const std::string& key, uint64_t val) {
@@ -64,25 +63,25 @@ class FstWriter {
     fstBuilderFinish(_b);
     fstBuilderDestroy(_b);
 
-    writerCtxDestroy(_wc, false);
+    idxFileCtxDestroy(_wc, false);
   }
 
  private:
   FstBuilder* _b;
-  WriterCtx*  _wc;
+  IFileCtx*   _wc;
 };
 
 class FstReadMemory {
  public:
   FstReadMemory(size_t size) {
-    _wc = writerCtxCreate(TFile, TD_TMP_DIR_PATH "tindex", true, 64 * 1024);
+    _wc = idxFileCtxCreate(TFile, TD_TMP_DIR_PATH "tindex", true, 64 * 1024);
-    _w = fstCountingWriterCreate(_wc);
+    _w = idxFileCreate(_wc);
     _size = size;
     memset((void*)&_s, 0, sizeof(_s));
   }
   bool init() {
     char* buf = (char*)taosMemoryCalloc(1, sizeof(char) * _size);
-    int   nRead = fstCountingWriterRead(_w, (uint8_t*)buf, _size);
+    int   nRead = idxFileRead(_w, (uint8_t*)buf, _size);
     if (nRead <= 0) {
       return false;
     }
@@ -124,18 +123,18 @@ class FstReadMemory {
   }
 
   ~FstReadMemory() {
-    fstCountingWriterDestroy(_w);
+    idxFileDestroy(_w);
     fstDestroy(_fst);
     fstSliceDestroy(&_s);
-    writerCtxDestroy(_wc, true);
+    idxFileCtxDestroy(_wc, true);
   }
 
  private:
-  FstCountingWriter* _w;
+  IdxFstFile* _w;
-  Fst*               _fst;
+  Fst*        _fst;
-  FstSlice           _s;
+  FstSlice    _s;
-  WriterCtx*         _wc;
+  IFileCtx*   _wc;
-  size_t             _size;
+  size_t      _size;
 };
 
 #define L 100
@@ -392,13 +391,13 @@ class TFileObj {
 
     fileName_ = path;
 
-    WriterCtx* ctx = writerCtxCreate(TFile, path.c_str(), false, 64 * 1024 * 1024);
+    IFileCtx* ctx = idxFileCtxCreate(TFile, path.c_str(), false, 64 * 1024 * 1024);
 
     writer_ = tfileWriterCreate(ctx, &header);
     return writer_ != NULL ? true : false;
   }
   bool InitReader() {
-    WriterCtx* ctx = writerCtxCreate(TFile, fileName_.c_str(), true, 64 * 1024 * 1024);
+    IFileCtx* ctx = idxFileCtxCreate(TFile, fileName_.c_str(), true, 64 * 1024 * 1024);
     reader_ = tfileReaderCreate(ctx);
     return reader_ != NULL ? true : false;
   }

source/libs/index/test/jsonUT.cc

@@ -7,7 +7,6 @@
 #include "index.h"
 #include "indexCache.h"
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 #include "indexTfile.h"

source/libs/index/test/utilUT.cc

@@ -8,7 +8,6 @@
 #include "indexCache.h"
 #include "indexComm.h"
 #include "indexFst.h"
-#include "indexFstCountingWriter.h"
 #include "indexFstUtil.h"
 #include "indexInt.h"
 #include "indexTfile.h"

source/libs/nodes/src/nodesCloneFuncs.c

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

source/libs/nodes/src/nodesCodeFuncs.c

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

source/libs/nodes/src/nodesTraverseFuncs.c

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

source/libs/nodes/src/nodesUtilFuncs.c

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

source/libs/parser/inc/parAst.h

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

source/libs/parser/inc/parUtil.h

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

source/libs/parser/inc/sql.y

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

source/libs/parser/src/parAstCreater.c

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

source/libs/parser/src/parAstParser.c

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

source/libs/parser/src/parInsert.c

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

source/libs/parser/src/parUtil.c

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

source/libs/parser/test/mockCatalog.cpp

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

source/libs/parser/test/mockCatalogService.cpp

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

source/libs/parser/test/mockCatalogService.h

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

source/libs/parser/test/parInitialATest.cpp

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

source/libs/parser/test/parInitialCTest.cpp

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

source/libs/planner/src/planOptimizer.c

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

source/libs/planner/src/planPhysiCreater.c

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

source/libs/planner/src/planSpliter.c

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

source/libs/planner/test/planOptimizeTest.cpp

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