Merge branch '3.0' into feature/TD-11274-3.0

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 setNodeType(nodeptr, type) (((SNode*)(nodeptr))->type = (type))
+#define nodeType(nodeptr)              (((const SNode*)(nodeptr))->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

@@ -320,6 +320,7 @@ typedef struct SInterpFuncPhysiNode {
   SNodeList*  pFuncs;
   STimeWindow timeRange;
   int64_t     interval;
+  int8_t      intervalUnit;
   EFillMode   fillMode;
   SNode*      pFillValues;  // SNodeListNode
   SNode*      pTimeSeries;  // SColumnNode

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        snapshotEnable;
-  SyncGroupId vgId;
-  int32_t     batchSize;
-  SSyncCfg    syncCfg;
-  char        path[TSDB_FILENAME_LEN];
-  SWal*       pWal;
-  SSyncFSM*   pFsm;
-  SMsgCb*     msgcb;
+  bool          isStandBy;
+  ESyncStrategy snapshotStrategy;
+  SyncGroupId   vgId;
+  int32_t       batchSize;
+  SSyncCfg      syncCfg;
+  char          path[TSDB_FILENAME_LEN];
+  SWal*         pWal;
+  SSyncFSM*     pFsm;
+  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 originalRpcType;  // user RpcMsg msgType
+  uint32_t msgType;          // TDMT_SYNC_CLIENT_REQUEST
+  uint32_t originalRpcType;  // origin RpcMsg msgType
   uint64_t seqNum;
   bool     isWeak;
-  uint32_t dataLen;  // user RpcMsg.contLen
-  char     data[];   // user RpcMsg.pCont
+  uint32_t dataLen;  // origin RpcMsg.contLen
+  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
-  char     data[];   // user RpcMsg.pCont
+  uint32_t dataLen;
+  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
-// block2: SOffsetAndContLen Array
-// block3: SRpcMsg Array
-// block4: SRpcMsg pCont Array
+// data:
+// block1: SOffsetAndContLen Array
+// block2: entry 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
-  SyncTerm  lastTerm;   // lastTerm of snapshot
-  SyncIndex lastConfigIndex;
+  SyncIndex beginIndex;       // snapshot.beginIndex
+  SyncIndex lastIndex;        // snapshot.lastIndex
+  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    rpcClose(void *);
-void *  rpcMallocCont(int32_t contLen);
-void    rpcFreeCont(void *pCont);
-void *  rpcReallocCont(void *ptr, int32_t contLen);
+
+void  rpcCleanup();
+void *rpcOpen(const SRpcInit *pRpc);
+
+void  rpcClose(void *);
+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_INVALID_CLUSTER_ID        TAOS_DEF_ERROR_CODE(0, 0x0345)
-#define TSDB_CODE_MND_INVALID_DNODE_CFG         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_ID          TAOS_DEF_ERROR_CODE(0, 0x0348)
+#define TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE    TAOS_DEF_ERROR_CODE(0, 0x0344)
+#define TSDB_CODE_MND_INVALID_CLUSTER_CFG       TAOS_DEF_ERROR_CODE(0, 0x0345)
+#define TSDB_CODE_MND_INVALID_CLUSTER_ID        TAOS_DEF_ERROR_CODE(0, 0x0346)
+#define TSDB_CODE_MND_INVALID_DNODE_CFG         TAOS_DEF_ERROR_CODE(0, 0x0347)
+#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)

include/util/tdef.h

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

source/client/inc/clientInt.h

@@ -65,7 +65,7 @@ enum {
 typedef struct SAppInstInfo SAppInstInfo;
 
 typedef struct {
-  char*   key;
+  char* key;
   // statistics
   int32_t reportCnt;
   int32_t connKeyCnt;
@@ -177,14 +177,14 @@ typedef struct SReqResultInfo {
 } SReqResultInfo;
 
 typedef struct SRequestSendRecvBody {
-  tsem_t             rspSem;  // not used now
-  __taos_async_fn_t  queryFp;
-  __taos_async_fn_t  fetchFp;
-  void*              param;
-  SDataBuf           requestMsg;
-  int64_t            queryJob;  // query job, created according to sql query DAG.
-  int32_t            subplanNum;
-  SReqResultInfo     resInfo;
+  tsem_t            rspSem;  // not used now
+  __taos_async_fn_t queryFp;
+  __taos_async_fn_t fetchFp;
+  void*             param;
+  SDataBuf          requestMsg;
+  int64_t           queryJob;  // query job, created according to sql query DAG.
+  int32_t           subplanNum;
+  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       closeAllRequests(SHashObj *pRequests);
+void       hbRemoveAppHbMrg(SAppHbMgr** pAppHbMgr);
+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;
+
+    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;
 
-    removeRequest(*rid);
+    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);
-      pRequest->body.queryJob = 0;
-    }
+    schedulerFreeJob(&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;
-  
-  STscObj* pTscObj = acquireTscObj(rid);
-  closeAllRequests(pTscObj->pRequests);
+
+  int64_t  rid = *(int64_t *)taos;
+  STscObj *pTscObj = acquireTscObj(rid);
+  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;
-  SQuery        *pQuery;
+  SRequestObj *  pRequest;
+  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;
-  SRequestObj     *pRequest = pWrapper->pRequest;
+  SQuery *         pQuery = pWrapper->pQuery;
+  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){
-      .requestId = pRequest->requestId,
-      .requestRid = pRequest->self,
-      .acctId = pTscObj->acctId,
-      .db = pRequest->pDb,
-      .topicQuery = false,
-      .pSql = pRequest->sqlstr,
-      .sqlLen = pRequest->sqlLen,
-      .pMsg = pRequest->msgBuf,
-      .msgLen = ERROR_MSG_BUF_DEFAULT_SIZE,
-      .pTransporter = pTscObj->pAppInfo->pTransporter,
-      .pStmtCb = NULL,
-      .pUser = pTscObj->user,
-      .schemalessType = pTscObj->schemalessType,
-      .isSuperUser = (0 == strcmp(pTscObj->user, TSDB_DEFAULT_USER)),
-      .async = true,
-      .svrVer = pTscObj->sVer,
-      .nodeOffline = (pTscObj->pAppInfo->onlineDnodes < pTscObj->pAppInfo->totalDnodes)
-  };
+  **pCxt = (SParseContext){.requestId = pRequest->requestId,
+                           .requestRid = pRequest->self,
+                           .acctId = pTscObj->acctId,
+                           .db = pRequest->pDb,
+                           .topicQuery = false,
+                           .pSql = pRequest->sqlstr,
+                           .sqlLen = pRequest->sqlLen,
+                           .pMsg = pRequest->msgBuf,
+                           .msgLen = ERROR_MSG_BUF_DEFAULT_SIZE,
+                           .pTransporter = pTscObj->pAppInfo->pTransporter,
+                           .pStmtCb = NULL,
+                           .pUser = pTscObj->user,
+                           .schemalessType = pTscObj->schemalessType,
+                           .isSuperUser = (0 == strcmp(pTscObj->user, TSDB_DEFAULT_USER)),
+                           .async = true,
+                           .svrVer = pTscObj->sVer,
+                           .nodeOffline = (pTscObj->pAppInfo->onlineDnodes < pTscObj->pAppInfo->totalDnodes)};
   return TSDB_CODE_SUCCESS;
 }
 
 void doAsyncQuery(SRequestObj *pRequest, bool updateMetaForce) {
   SParseContext *pCxt = NULL;
-  STscObj       *pTscObj = pRequest->pTscObj;
+  STscObj *      pTscObj = pRequest->pTscObj;
   int32_t        code = 0;
 
   if (pRequest->retry++ > REQUEST_TOTAL_EXEC_TIMES) {
@@ -911,10 +905,10 @@ int taos_load_table_info(TAOS *taos, const char *tableNameList) {
     return terrno;
   }
 
-  int64_t rid = *(int64_t*)taos;
+  int64_t       rid = *(int64_t *)taos;
   const int32_t MAX_TABLE_NAME_LENGTH = 12 * 1024 * 1024;  // 12MB list
   int32_t       code = 0;
-  SRequestObj  *pRequest = NULL;
+  SRequestObj * pRequest = NULL;
   SCatalogReq   catalogReq = {0};
 
   if (NULL == tableNameList) {

source/common/src/tdatablock.c

@@ -895,7 +895,7 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
         SBlockOrderInfo* pOrder = taosArrayGet(pOrderInfo, 0);
 
         int64_t p0 = taosGetTimestampUs();
-        
+
         __compar_fn_t fn = getKeyComparFunc(pColInfoData->info.type, pOrder->order);
         taosSort(pColInfoData->pData, pDataBlock->info.rows, pColInfoData->info.bytes, fn);
 
@@ -923,8 +923,9 @@ int32_t blockDataSort(SSDataBlock* pDataBlock, SArray* pOrderInfo) {
     pInfo->pColData = taosArrayGet(pDataBlock->pDataBlock, pInfo->slotId);
   }
 
+  terrno = 0;
   taosqsort(index, rows, sizeof(int32_t), &helper, dataBlockCompar);
-  if(terrno) return terrno;
+  if (terrno) return terrno;
 
   int64_t p1 = taosGetTimestampUs();
 
@@ -1438,21 +1439,21 @@ static void doShiftBitmap(char* nullBitmap, size_t n, size_t total) {
   }
 }
 
-static int32_t colDataMoveVarData(SColumnInfoData* pColInfoData, size_t start, size_t end){
+static int32_t colDataMoveVarData(SColumnInfoData* pColInfoData, size_t start, size_t end) {
   int32_t dataOffset = -1;
   int32_t dataLen = 0;
   int32_t beigin = start;
-  while(beigin < end){
+  while (beigin < end) {
     int32_t offset = pColInfoData->varmeta.offset[beigin];
-    if(offset == -1) {
+    if (offset == -1) {
       beigin++;
       continue;
     }
-    if(start != 0) {
+    if (start != 0) {
       pColInfoData->varmeta.offset[beigin] = dataLen;
     }
-    char *data = pColInfoData->pData + offset;
-    if(dataOffset == -1) dataOffset = offset;   // mark the begin of data
+    char* data = pColInfoData->pData + offset;
+    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;
@@ -3586,6 +3594,7 @@ 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 +3652,7 @@ 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 +3675,6 @@ 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 +3777,7 @@ 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 +3792,6 @@ 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 +3808,7 @@ 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_util/src/dmEps.c

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

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

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

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

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

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

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

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

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

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

@@ -15,8 +15,8 @@
 
 #define _DEFAULT_SOURCE
 #include "mndDnode.h"
-#include "mndPrivilege.h"
 #include "mndMnode.h"
+#include "mndPrivilege.h"
 #include "mndQnode.h"
 #include "mndShow.h"
 #include "mndSnode.h"
@@ -432,7 +432,8 @@ 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 +442,8 @@ 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 +583,7 @@ static int32_t mndProcessShowVariablesReq(SRpcMsg *pReq) {
   strcpy(info.name, "timezone");
   snprintf(info.value, TSDB_CONFIG_VALUE_LEN, "%s", tsTimezoneStr);
   taosArrayPush(rsp.variables, &info);
-  
+
   strcpy(info.name, "locale");
   snprintf(info.value, TSDB_CONFIG_VALUE_LEN, "%s", tsLocale);
   taosArrayPush(rsp.variables, &info);

source/dnode/mnode/impl/src/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/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,
-         pDnode->numOfSupportVnodes, isMnode, online);
+  mDebug("dnode:%d, vnodes:%d supportVnodes:%d isMnode:%d online:%d memory avail:%" PRId64 " used:%" PRId64, pDnode->id,
+         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) {
-      terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
+    if (mndGetAvailableDnode(pMnode, pDb, pVgroup, pArray) != 0) {
       goto _OVER;
     }
 
@@ -728,6 +733,43 @@ int32_t mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId) {
   return numOfVnodes;
 }
 
+int64_t mndGetVgroupMemory(SMnode *pMnode, SDbObj *pDbInput, SVgObj *pVgroup) {
+  SDbObj *pDb = pDbInput;
+  if (pDbInput == NULL) {
+    pDb = mndAcquireDb(pMnode, pVgroup->dbName);
+  }
+
+  int64_t vgroupMemroy = (int64_t)pDb->cfg.buffer * 1024 * 1024 + (int64_t)pDb->cfg.pages * pDb->cfg.pageSize * 1024;
+  if (pDb->cfg.cacheLastRow > 0) {
+    vgroupMemroy += (int64_t)pDb->cfg.lastRowMem * 1024 * 1024;
+  }
+
+  if (pDbInput == NULL) {
+    mndReleaseDb(pMnode, pDb);
+  }
+  return vgroupMemroy;
+}
+
+static bool mndGetVnodeMemroyFp(SMnode *pMnode, void *pObj, void *p1, void *p2, void *p3) {
+  SVgObj  *pVgroup = pObj;
+  int32_t  dnodeId = *(int32_t *)p1;
+  int64_t *pVnodeMemory = (int64_t *)p2;
+
+  for (int32_t v = 0; v < pVgroup->replica; ++v) {
+    if (pVgroup->vnodeGid[v].dnodeId == dnodeId) {
+      *pVnodeMemory += mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    }
+  }
+
+  return true;
+}
+
+int64_t mndGetVnodesMemory(SMnode *pMnode, int32_t dnodeId) {
+  int64_t vnodeMemory = 0;
+  sdbTraverse(pMnode->pSdb, SDB_VGROUP, mndGetVnodeMemroyFp, &dnodeId, &vnodeMemory, NULL);
+  return vnodeMemory;
+}
+
 static int32_t mndRetrieveVnodes(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {
   SMnode *pMnode = pReq->info.node;
   SSdb   *pSdb = pMnode->pSdb;
@@ -807,9 +849,20 @@ int32_t mndAddVnodeToVgroup(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {
       return -1;
     }
 
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pDnode->memAvail - vgMem - pDnode->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pDnode->id, pDnode->memAvail, pDnode->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pDnode->memUsed += vgMem;
+    }
+
     pVgid->dnodeId = pDnode->id;
     pVgid->role = TAOS_SYNC_STATE_ERROR;
-    mInfo("db:%s, vgId:%d, vn:%d dnode:%d, is added", pVgroup->dbName, pVgroup->vgId, pVgroup->replica, pVgid->dnodeId);
+    mInfo("db:%s, vgId:%d, vn:%d is added, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+          pVgroup->dbName, pVgroup->vgId, pVgroup->replica, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
 
     pVgroup->replica++;
     pDnode->numOfVnodes++;
@@ -835,7 +888,10 @@ int32_t mndRemoveVnodeFromVgroup(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray
     for (int32_t vn = 0; vn < pVgroup->replica; ++vn) {
       SVnodeGid *pVgid = &pVgroup->vnodeGid[vn];
       if (pVgid->dnodeId == pDnode->id) {
-        mInfo("db:%s, vgId:%d, vn:%d dnode:%d, is removed", pVgroup->dbName, pVgroup->vgId, vn, pVgid->dnodeId);
+        int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+        pDnode->memUsed -= vgMem;
+        mInfo("db:%s, vgId:%d, vn:%d is removed, memory:%" PRId64 ", dnode:%d avail:%" PRId64 " used:%" PRId64,
+              pVgroup->dbName, pVgroup->vgId, vn, vgMem, pVgid->dnodeId, pDnode->memAvail, pDnode->memUsed);
         pDnode->numOfVnodes--;
         pVgroup->replica--;
         *pDelVgid = *pVgid;
@@ -1161,6 +1217,17 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew1->memAvail - vgMem - pNew1->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew1->id, pNew1->memAvail, pNew1->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew1->memUsed += vgMem;
+    }
+
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew1->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld1->id) != 0) goto _OVER;
   }
@@ -1173,6 +1240,15 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew2->memAvail - vgMem - pNew2->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew2->id, pNew2->memAvail, pNew2->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew2->memUsed += vgMem;
+    }
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew2->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld2->id) != 0) goto _OVER;
   }
@@ -1185,6 +1261,15 @@ static int32_t mndRedistributeVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb,
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
       goto _OVER;
     }
+    int64_t vgMem = mndGetVgroupMemory(pMnode, NULL, pVgroup);
+    if (pNew3->memAvail - vgMem - pNew3->memUsed <= 0) {
+      mError("db:%s, vgId:%d, no enough memory:%" PRId64 " in dnode:%d avail:%" PRId64 " used:%" PRId64,
+             pVgroup->dbName, pVgroup->vgId, vgMem, pNew3->id, pNew3->memAvail, pNew3->memUsed);
+      terrno = TSDB_CODE_MND_NO_ENOUGH_MEM_IN_DNODE;
+      return -1;
+    } else {
+      pNew3->memUsed += vgMem;
+    }
     if (mndAddIncVgroupReplicaToTrans(pMnode, pTrans, pDb, &newVg, pNew3->id) != 0) goto _OVER;
     if (mndAddDecVgroupReplicaFromTrans(pMnode, pTrans, pDb, &newVg, pOld3->id) != 0) goto _OVER;
   }
@@ -1415,7 +1500,7 @@ _OVER:
   mndReleaseDb(pMnode, pDb);
 
   return code;
-#endif  
+#endif
 }
 
 int32_t mndBuildAlterVgroupAction(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SVgObj *pVgroup, SArray *pArray) {
@@ -1654,9 +1739,9 @@ static int32_t mndBalanceVgroupBetweenDnode(SMnode *pMnode, STrans *pTrans, SDno
 }
 
 static int32_t mndBalanceVgroup(SMnode *pMnode, SRpcMsg *pReq, SArray *pArray) {
-  int32_t code = -1;
-  int32_t numOfVgroups = 0;
-  STrans *pTrans = NULL;
+  int32_t   code = -1;
+  int32_t   numOfVgroups = 0;
+  STrans   *pTrans = NULL;
   SHashObj *pBalancedVgroups = NULL;
 
   pBalancedVgroups = taosHashInit(16, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false, HASH_NO_LOCK);
@@ -1721,7 +1806,7 @@ static int32_t mndProcessBalanceVgroupMsg(SRpcMsg *pReq) {
   SMnode *pMnode = pReq->info.node;
   int32_t code = -1;
   SArray *pArray = NULL;
-  void   *pIter = NULL;
+  void *pIter = NULL;
   int64_t curMs = taosGetTimestampMs();
 
   SBalanceVgroupReq req = {0};
@@ -1766,7 +1851,7 @@ _OVER:
 
   taosArrayDestroy(pArray);
   return code;
-#endif 
+#endif
 }
 
 bool mndVgroupInDb(SVgObj *pVgroup, int64_t dbUid) { return !pVgroup->isTsma && pVgroup->dbUid == dbUid; }
\ No newline at end of file

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) {
-      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_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) {
-      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);
+    // ToDo: ugly! use function pointer
+    // use different strategy
+    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);
+
+      } 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_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) {
+        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);
+        ret = -1;
+      }
 
-    } 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;
+      // use wal first strategy
+
+      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/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/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/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);
@@ -1497,13 +1497,18 @@ 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 +1520,27 @@ 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) {
-    nodesDestroyList(cxt.pFuncs);
-    return cxt.errCode;
-  }
-  if (LIST_LENGTH(cxt.pFuncs) > 0) {
-    *pFuncs = cxt.pFuncs;
+  if (TSDB_CODE_SUCCESS == cxt.errCode) {
+    if (LIST_LENGTH(cxt.pFuncs) > 0) {
+      *pFuncs = cxt.pFuncs;
+    } else {
+      nodesDestroyList(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;
-    } else {
+    }
+
+    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 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) {
-      // pSql -> field1_name, ...)
+    if (NULL != pBoundColsStart) {
+      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;
@@ -1717,15 +1748,15 @@ 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);
-    } else {
-      CHECK_CODE(collectTableMetaKey(pCxt, &tbnameToken));
     }
 
     if (TK_NK_LP == sToken.type) {
@@ -1734,6 +1765,17 @@ static int32_t parseInsertBodySyntax(SInsertParseSyntaxCxt* 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));
+      CHECK_CODE(skipUsingClause(pCxt));
+      NEXT_TOKEN(pCxt->pSql, sToken);
+    } else {
+      CHECK_CODE(collectTableMetaKey(pCxt, &tbnameToken));
+    }
+
     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/parInitialATest.cpp

@@ -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/planner/src/planOptimizer.c

@@ -720,7 +720,7 @@ 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))) {
     return TSDB_CODE_SUCCESS;
   }
 
@@ -1251,7 +1251,7 @@ 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 +1868,8 @@ 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);

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);

source/libs/planner/test/planOptimizeTest.cpp

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

source/libs/planner/test/planSetOpTest.cpp

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

source/libs/scalar/src/scalar.c

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

source/libs/scheduler/src/schRemote.c

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

source/libs/scheduler/src/scheduler.c

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

source/libs/sync/inc/syncAppendEntriesReply.h

@@ -44,11 +44,6 @@ int32_t syncNodeOnAppendEntriesReplyCb(SSyncNode* ths, SyncAppendEntriesReply* p
 int32_t syncNodeOnAppendEntriesReplySnapshotCb(SSyncNode* ths, SyncAppendEntriesReply* pMsg);
 int32_t syncNodeOnAppendEntriesReplySnapshot2Cb(SSyncNode* ths, SyncAppendEntriesReply* pMsg);
 
-typedef struct SReaderParam {
-  SyncIndex start;
-  SyncIndex end;
-} SReaderParam;
-
 #ifdef __cplusplus
 }
 #endif

taos-tools @ 1163c0f6

-Subproject commit 7105027650b51e701cfa1dac11b8fb42d447dd01
+Subproject commit 1163c0f60aa65d6cc58283247c8bf8c56ba43b92