Merge pull request #13495 from taosdata/feature/qnode

enh: remove offline scheduler info

include/common/tmsg.h

@@ -2485,6 +2485,32 @@ typedef struct {
 int32_t tSerializeSUserIndexRsp(void* buf, int32_t bufLen, const SUserIndexRsp* pRsp);
 int32_t tDeserializeSUserIndexRsp(void* buf, int32_t bufLen, SUserIndexRsp* pRsp);
 
+typedef struct {
+  char tbFName[TSDB_TABLE_FNAME_LEN];
+} STableIndexReq;
+
+int32_t tSerializeSTableIndexReq(void* buf, int32_t bufLen, STableIndexReq* pReq);
+int32_t tDeserializeSTableIndexReq(void* buf, int32_t bufLen, STableIndexReq* pReq);
+
+typedef struct {
+  int8_t    intervalUnit;
+  int8_t    slidingUnit;
+  int64_t   interval;
+  int64_t   offset;
+  int64_t   sliding;
+  int64_t   dstTbUid;  
+  int32_t   dstVgId;  // for stream  
+  char*     expr;
+} STableIndexInfo;
+
+typedef struct {
+  SArray* pIndex;
+} STableIndexRsp;
+
+int32_t tSerializeSTableIndexRsp(void* buf, int32_t bufLen, const STableIndexRsp* pRsp);
+int32_t tDeserializeSTableIndexRsp(void* buf, int32_t bufLen, STableIndexRsp* pRsp);
+
+
 typedef struct {
   int8_t  mqMsgType;
   int32_t code;

include/common/tmsgdef.h

@@ -130,6 +130,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_INDEX, "create-index", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_INDEX, "drop-index", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_GET_INDEX, "get-index", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_GET_TABLE_INDEX, "get-table-index", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_TOPIC, "create-topic", SMCreateTopicReq, SMCreateTopicRsp)
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_TOPIC, "alter-topic", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_TOPIC, "drop-topic", NULL, NULL)

include/libs/catalog/catalog.h

@@ -272,6 +272,8 @@ int32_t catalogGetDBCfg(SCatalog* pCtg, void *pRpc, const SEpSet* pMgmtEps, cons
 
 int32_t catalogGetIndexMeta(SCatalog* pCtg, void *pRpc, const SEpSet* pMgmtEps, const char* indexName, SIndexInfo* pInfo);
 
+int32_t catalogGetTableIndex(SCatalog* pCtg, void *pTrans, const SEpSet* pMgmtEps, const char* tbFName, SArray** pRes);
+
 int32_t catalogGetUdfInfo(SCatalog* pCtg, void *pRpc, const SEpSet* pMgmtEps, const char* funcName, SFuncInfo* pInfo);
 
 int32_t catalogChkAuth(SCatalog* pCtg, void *pRpc, const SEpSet* pMgmtEps, const char* user, const char* dbFName, AUTH_TYPE type, bool *pass);
@@ -280,7 +282,6 @@ int32_t catalogUpdateUserAuthInfo(SCatalog* pCtg, SGetUserAuthRsp* pAuth);
 
 int32_t catalogUpdateVgEpSet(SCatalog* pCtg, const char* dbFName, int32_t vgId, SEpSet *epSet);
 
-
 int32_t ctgdLaunchAsyncCall(SCatalog* pCtg, void *pTrans, const SEpSet* pMgmtEps, uint64_t reqId, bool forceUpdate);
 
 

source/common/src/tmsg.c

@@ -2394,6 +2394,102 @@ int32_t tDeserializeSUserIndexRsp(void *buf, int32_t bufLen, SUserIndexRsp *pRsp
   return 0;
 }
 
+int32_t tSerializeSTableIndexReq(void *buf, int32_t bufLen, STableIndexReq *pReq) {
+  SEncoder encoder = {0};
+  tEncoderInit(&encoder, buf, bufLen);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeCStr(&encoder, pReq->tbFName) < 0) return -1;
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tEncoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSTableIndexReq(void *buf, int32_t bufLen, STableIndexReq *pReq) {
+  SDecoder decoder = {0};
+  tDecoderInit(&decoder, buf, bufLen);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pReq->tbFName) < 0) return -1;
+  tEndDecode(&decoder);
+
+  tDecoderClear(&decoder);
+  return 0;
+}
+
+int32_t tSerializeSTableIndexInfo(SEncoder *pEncoder, STableIndexInfo* pInfo) {
+  if (tEncodeI8(pEncoder, pInfo->intervalUnit) < 0) return -1;
+  if (tEncodeI8(pEncoder, pInfo->slidingUnit) < 0) return -1;
+  if (tEncodeI64(pEncoder, pInfo->interval) < 0) return -1;
+  if (tEncodeI64(pEncoder, pInfo->offset) < 0) return -1;
+  if (tEncodeI64(pEncoder, pInfo->sliding) < 0) return -1;
+  if (tEncodeI64(pEncoder, pInfo->dstTbUid) < 0) return -1;
+  if (tEncodeI32(pEncoder, pInfo->dstVgId) < 0) return -1;
+  if (tEncodeCStr(pEncoder, pInfo->expr) < 0) return -1;
+  return 0;
+}
+
+int32_t tSerializeSTableIndexRsp(void *buf, int32_t bufLen, const STableIndexRsp *pRsp) {
+  SEncoder encoder = {0};
+  tEncoderInit(&encoder, buf, bufLen);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  int32_t num = taosArrayGetSize(pRsp->pIndex);
+  if (tEncodeI32(&encoder, num) < 0) return -1;
+  if (num > 0) {
+    for (int32_t i = 0; i < num; ++i) {
+      STableIndexInfo* pInfo = (STableIndexInfo*)taosArrayGet(pRsp->pIndex, i);
+      if (tSerializeSTableIndexInfo(&encoder, pInfo) < 0) return -1;
+    }
+  }
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tEncoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSTableIndexInfo(SDecoder *pDecoder, STableIndexInfo *pInfo) {
+  if (tDecodeI8(pDecoder, &pInfo->intervalUnit) < 0) return -1;
+  if (tDecodeI8(pDecoder, &pInfo->slidingUnit) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pInfo->interval) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pInfo->offset) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pInfo->sliding) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pInfo->dstTbUid) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pInfo->dstVgId) < 0) return -1;
+  if (tDecodeCStrAlloc(pDecoder, &pInfo->expr) < 0) return -1;
+
+  return 0;
+}
+
+int32_t tDeserializeSTableIndexRsp(void *buf, int32_t bufLen, STableIndexRsp *pRsp) {
+  SDecoder decoder = {0};
+  tDecoderInit(&decoder, buf, bufLen);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  int32_t num = 0;
+  if (tDecodeI32(&decoder, &num) < 0) return -1;
+  if (num > 0) {
+    pRsp->pIndex = taosArrayInit(num, sizeof(STableIndexInfo));
+    if (NULL == pRsp->pIndex) return -1;
+    STableIndexInfo info;
+    for (int32_t i = 0; i < num; ++i) {
+      if (tDeserializeSTableIndexInfo(&decoder, &info) < 0) return -1;
+      if (NULL == taosArrayPush(pRsp->pIndex, &info)) {
+        taosMemoryFree(info.expr);
+        return -1;
+      }
+    }
+  }
+  tEndDecode(&decoder);
+
+  tDecoderClear(&decoder);
+  return 0;
+}
+
+
 int32_t tSerializeSShowReq(void *buf, int32_t bufLen, SShowReq *pReq) {
   SEncoder encoder = {0};
   tEncoderInit(&encoder, buf, bufLen);

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

@@ -181,6 +181,7 @@ SArray *mmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_MND_DROP_SMA, mmPutNodeMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_CREATE_STREAM, mmPutNodeMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_GET_INDEX, mmPutNodeMsgToReadQueue, 0) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_MND_GET_TABLE_INDEX, mmPutNodeMsgToReadQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_CREATE_TOPIC, mmPutNodeMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_ALTER_TOPIC, mmPutNodeMsgToWriteQueue, 0) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_MND_DROP_TOPIC, mmPutNodeMsgToWriteQueue, 0) == NULL) goto _OVER;

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

@@ -309,6 +309,7 @@ typedef struct {
   int8_t    slidingUnit;
   int8_t    timezone;
   int32_t   dstVgId;  // for stream
+  int64_t   dstTbUid;
   int64_t   interval;
   int64_t   offset;
   int64_t   sliding;

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

@@ -40,6 +40,7 @@ static int32_t  mndSmaGetVgEpSet(SMnode *pMnode, SDbObj *pDb, SVgEpSet **ppVgEpS
 static int32_t  mndProcessMCreateSmaReq(SRpcMsg *pReq);
 static int32_t  mndProcessMDropSmaReq(SRpcMsg *pReq);
 static int32_t  mndProcessGetSmaReq(SRpcMsg *pReq);
+static int32_t mndProcessGetTbSmaReq(SRpcMsg *pReq);
 static int32_t  mndRetrieveSma(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
 static void     mndCancelGetNextSma(SMnode *pMnode, void *pIter);
 
@@ -59,6 +60,7 @@ int32_t mndInitSma(SMnode *pMnode) {
   mndSetMsgHandle(pMnode, TDMT_VND_CREATE_SMA_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_VND_DROP_SMA_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_MND_GET_INDEX, mndProcessGetSmaReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_GET_TABLE_INDEX, mndProcessGetTbSmaReq);
 
   mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_INDEX, mndRetrieveSma);
   mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_INDEX, mndCancelGetNextSma);
@@ -870,6 +872,55 @@ static int32_t mndGetSma(SMnode *pMnode, SUserIndexReq *indexReq, SUserIndexRsp
   return code;
 }
 
+static int32_t mndGetTableSma(SMnode *pMnode, STableIndexReq *indexReq, STableIndexRsp *rsp, bool *exist) {
+  int32_t  code = 0;
+  SSmaObj *pSma = NULL;
+  SSdb   *pSdb = pMnode->pSdb;
+  void   *pIter = NULL;
+  STableIndexInfo info;
+
+  while (1) {
+    pIter = sdbFetch(pSdb, SDB_SMA, pIter, (void **)&pSma);
+    if (pIter == NULL) break;
+
+    if (pSma->stb[0] != indexReq->tbFName[0] || strcmp(pSma->stb, indexReq->tbFName)) {
+      continue;
+    }
+
+    info.intervalUnit = pSma->intervalUnit;
+    info.slidingUnit = pSma->slidingUnit;
+    info.interval = pSma->interval;
+    info.offset = pSma->offset;
+    info.sliding = pSma->sliding;
+    info.dstTbUid = pSma->dstTbUid;
+    info.dstVgId = pSma->dstVgId;
+    info.expr = taosMemoryMalloc(pSma->exprLen + 1);
+    if (info.expr == NULL) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      code = -1;
+      sdbRelease(pSdb, pSma);
+      return code;
+    }
+
+    memcpy(info.expr, pSma->expr, pSma->exprLen);
+    info.expr[pSma->exprLen] = 0;
+
+    if (NULL == taosArrayPush(rsp->pIndex, &info)) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      code = -1;
+      taosMemoryFree(info.expr);
+      sdbRelease(pSdb, pSma);
+      return code;
+    }
+
+    *exist = true;
+
+    sdbRelease(pSdb, pSma);
+  }
+  
+  return code;
+}
+
 static int32_t mndProcessGetSmaReq(SRpcMsg *pReq) {
   SUserIndexReq indexReq = {0};
   SMnode       *pMnode = pReq->info.node;
@@ -916,6 +967,59 @@ _OVER:
   return code;
 }
 
+static int32_t mndProcessGetTbSmaReq(SRpcMsg *pReq) {
+  STableIndexReq indexReq = {0};
+  SMnode       *pMnode = pReq->info.node;
+  int32_t       code = -1;
+  STableIndexRsp rsp = {0};
+  bool          exist = false;
+
+  if (tDeserializeSTableIndexReq(pReq->pCont, pReq->contLen, &indexReq) != 0) {
+    terrno = TSDB_CODE_INVALID_MSG;
+    goto _OVER;
+  }
+
+  rsp.pIndex = taosArrayInit(10, sizeof(STableIndexInfo));
+  if (NULL == rsp.pIndex) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    code = -1;
+    goto _OVER;
+  }
+
+  code = mndGetTableSma(pMnode, &indexReq, &rsp, &exist);
+  if (code) {
+    goto _OVER;
+  }
+
+  if (!exist) {
+    code = -1;
+    terrno = TSDB_CODE_MND_DB_INDEX_NOT_EXIST;
+  } else {
+    int32_t contLen = tSerializeSTableIndexRsp(NULL, 0, &rsp);
+    void   *pRsp = rpcMallocCont(contLen);
+    if (pRsp == NULL) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      code = -1;
+      goto _OVER;
+    }
+
+    tSerializeSTableIndexRsp(pRsp, contLen, &rsp);
+
+    pReq->info.rsp = pRsp;
+    pReq->info.rspLen = contLen;
+
+    code = 0;
+  }
+
+_OVER:
+  if (code != 0) {
+    mError("failed to get table index %s since %s", indexReq.tbFName, terrstr());
+  }
+
+  return code;
+}
+
+
 static int32_t mndRetrieveSma(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {
   SMnode  *pMnode = pReq->info.node;
   SSdb    *pSdb = pMnode->pSdb;

source/libs/catalog/inc/catalogInt.h

@@ -310,22 +310,19 @@ typedef struct SCtgCacheOperation {
   int32_t  opId;
   void    *data;
   bool     syncOp;
-  uint64_t seqId;
+  tsem_t   rspSem;  
 } SCtgCacheOperation;
 
 typedef struct SCtgQNode {
-  SCtgCacheOperation     op;
+  SCtgCacheOperation    *op;
   struct SCtgQNode      *next;
 } SCtgQNode;
 
 typedef struct SCtgQueue {
   SRWLatch              qlock;
-  uint64_t              seqId;
-  uint64_t              seqDone;
   SCtgQNode            *head;
   SCtgQNode            *tail;
   tsem_t                reqSem;  
-  tsem_t                rspSem;  
   uint64_t              qRemainNum;
 } SCtgQueue;
 
@@ -493,6 +490,7 @@ int32_t ctgGetDBVgInfoFromMnode(CTG_PARAMS, SBuildUseDBInput *input, SUseDbOutpu
 int32_t ctgGetQnodeListFromMnode(CTG_PARAMS, SArray *out, SCtgTask* pTask);
 int32_t ctgGetDBCfgFromMnode(CTG_PARAMS, const char *dbFName, SDbCfgInfo *out, SCtgTask* pTask);
 int32_t ctgGetIndexInfoFromMnode(CTG_PARAMS, const char *indexName, SIndexInfo *out, SCtgTask* pTask);
+int32_t ctgGetTbIndexFromMnode(CTG_PARAMS, const char *tbFName, SArray** out, SCtgTask* pTask);
 int32_t ctgGetUdfInfoFromMnode(CTG_PARAMS, const char *funcName, SFuncInfo *out, SCtgTask* pTask);
 int32_t ctgGetUserDbAuthFromMnode(CTG_PARAMS, const char *user, SGetUserAuthRsp *out, SCtgTask* pTask);
 int32_t ctgGetTbMetaFromMnodeImpl(CTG_PARAMS, char *dbFName, char* tbName, STableMetaOutput* out, SCtgTask* pTask);

source/libs/catalog/src/catalog.c

@@ -506,11 +506,6 @@ int32_t catalogInit(SCatalogCfg *cfg) {
     CTG_ERR_RET(TSDB_CODE_CTG_SYS_ERROR);
   }
   
-  if (tsem_init(&gCtgMgmt.queue.rspSem, 0, 0)) {
-    qError("tsem_init failed, error:%s", tstrerror(TAOS_SYSTEM_ERROR(errno)));
-    CTG_ERR_RET(TSDB_CODE_CTG_SYS_ERROR);
-  }
-
   gCtgMgmt.queue.head = taosMemoryCalloc(1, sizeof(SCtgQNode));
   if (NULL == gCtgMgmt.queue.head) {
     qError("calloc %d failed", (int32_t)sizeof(SCtgQNode));
@@ -1141,6 +1136,17 @@ int32_t catalogGetIndexMeta(SCatalog* pCtg, void *pTrans, const SEpSet* pMgmtEps
   CTG_API_LEAVE(ctgGetIndexInfoFromMnode(CTG_PARAMS_LIST(), indexName, pInfo, NULL));
 }
 
+int32_t catalogGetTableIndex(SCatalog* pCtg, void *pTrans, const SEpSet* pMgmtEps, const char* tbFName, SArray** pRes) {
+  CTG_API_ENTER();
+  
+  if (NULL == pCtg || NULL == pTrans || NULL == pMgmtEps || NULL == tbFName || NULL == pRes) {
+    CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
+  }
+
+  CTG_API_LEAVE(ctgGetTbIndexFromMnode(CTG_PARAMS_LIST(), tbFName, pRes, NULL));
+}
+
+
 int32_t catalogGetUdfInfo(SCatalog* pCtg, void *pTrans, const SEpSet* pMgmtEps, const char* funcName, SFuncInfo* pInfo) {
   CTG_API_ENTER();
   
@@ -1194,10 +1200,6 @@ void catalogDestroy(void) {
   if (tsem_post(&gCtgMgmt.queue.reqSem)) {
     qError("tsem_post failed, error:%s", tstrerror(TAOS_SYSTEM_ERROR(errno)));
   }
-  
-  if (tsem_post(&gCtgMgmt.queue.rspSem)) {
-    qError("tsem_post failed, error:%s", tstrerror(TAOS_SYSTEM_ERROR(errno)));
-  }
 
   while (CTG_IS_LOCKED(&gCtgMgmt.lock)) {
     taosUsleep(1);

source/libs/catalog/src/ctgCache.c

@@ -501,25 +501,6 @@ _return:
   return TSDB_CODE_SUCCESS;
 }
 
-
-void ctgWaitOpDone(SCtgCacheOperation *action) {
-  while (true) {
-    tsem_wait(&gCtgMgmt.queue.rspSem);
-    
-    if (atomic_load_8((int8_t*)&gCtgMgmt.exit)) {
-      tsem_post(&gCtgMgmt.queue.rspSem);
-      break;
-    }
-
-    if (gCtgMgmt.queue.seqDone >= action->seqId) {
-      break;
-    }
-
-    tsem_post(&gCtgMgmt.queue.rspSem);
-    sched_yield();
-  }
-}
-
 void ctgDequeue(SCtgCacheOperation **op) {
   SCtgQNode *orig = gCtgMgmt.queue.head;
   
@@ -530,7 +511,7 @@ void ctgDequeue(SCtgCacheOperation **op) {
   
   taosMemoryFreeClear(orig);
 
-  *op = &node->op;
+  *op = node->op;
 }
 
 
@@ -541,9 +522,11 @@ int32_t ctgEnqueue(SCatalog* pCtg, SCtgCacheOperation *operation) {
     CTG_RET(TSDB_CODE_CTG_MEM_ERROR);
   }
 
-  operation->seqId = atomic_add_fetch_64(&gCtgMgmt.queue.seqId, 1);
+  if (operation->syncOp) {
+    tsem_init(&operation->rspSem, 0, 0);
+  }
   
-  node->op = *operation;
+  node->op = operation;
 
   CTG_LOCK(CTG_WRITE, &gCtgMgmt.queue.qlock);
   gCtgMgmt.queue.tail->next = node;
@@ -558,7 +541,8 @@ int32_t ctgEnqueue(SCatalog* pCtg, SCtgCacheOperation *operation) {
   ctgDebug("action [%s] added into queue", gCtgCacheOperation[operation->opId].name);
 
   if (operation->syncOp) {
-    ctgWaitOpDone(operation);
+    tsem_wait(&operation->rspSem);
+    taosMemoryFree(operation);
   }
 
   return TSDB_CODE_SUCCESS;
@@ -567,7 +551,9 @@ int32_t ctgEnqueue(SCatalog* pCtg, SCtgCacheOperation *operation) {
 
 int32_t ctgDropDbCacheEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_DROP_DB_CACHE};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_DROP_DB_CACHE;
+  
   SCtgDropDBMsg *msg = taosMemoryMalloc(sizeof(SCtgDropDBMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgDropDBMsg));
@@ -583,21 +569,24 @@ int32_t ctgDropDbCacheEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId)
   strncpy(msg->dbFName, dbFName, sizeof(msg->dbFName));
   msg->dbId = dbId;
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
 
 _return:
 
-  taosMemoryFreeClear(action.data);
+  taosMemoryFreeClear(op->data);
   CTG_RET(code);
 }
 
 int32_t ctgDropDbVgroupEnqueue(SCatalog* pCtg, const char *dbFName, bool syncOp) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_DROP_DB_VGROUP, .syncOp = syncOp};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_DROP_DB_VGROUP;
+  op->syncOp = syncOp;
+  
   SCtgDropDbVgroupMsg *msg = taosMemoryMalloc(sizeof(SCtgDropDbVgroupMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgDropDbVgroupMsg));
@@ -612,15 +601,15 @@ int32_t ctgDropDbVgroupEnqueue(SCatalog* pCtg, const char *dbFName, bool syncOp)
   msg->pCtg = pCtg;
   strncpy(msg->dbFName, dbFName, sizeof(msg->dbFName));
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
 
 _return:
 
-  taosMemoryFreeClear(action.data);
+  taosMemoryFreeClear(op->data);
   CTG_RET(code);
 }
 
@@ -628,7 +617,10 @@ _return:
 
 int32_t ctgDropStbMetaEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId, const char *stbName, uint64_t suid, bool syncOp) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_DROP_STB_META, .syncOp = syncOp};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_DROP_STB_META;
+  op->syncOp = syncOp;
+  
   SCtgDropStbMetaMsg *msg = taosMemoryMalloc(sizeof(SCtgDropStbMetaMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgDropStbMetaMsg));
@@ -641,15 +633,15 @@ int32_t ctgDropStbMetaEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId,
   msg->dbId = dbId;
   msg->suid = suid;
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
 
 _return:
 
-  taosMemoryFreeClear(action.data);
+  taosMemoryFreeClear(op->data);
   CTG_RET(code);
 }
 
@@ -657,7 +649,10 @@ _return:
 
 int32_t ctgDropTbMetaEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId, const char *tbName, bool syncOp) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_DROP_TB_META, .syncOp = syncOp};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_DROP_TB_META;
+  op->syncOp = syncOp;
+  
   SCtgDropTblMetaMsg *msg = taosMemoryMalloc(sizeof(SCtgDropTblMetaMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgDropTblMetaMsg));
@@ -669,21 +664,24 @@ int32_t ctgDropTbMetaEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId,
   strncpy(msg->tbName, tbName, sizeof(msg->tbName));
   msg->dbId = dbId;
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
 
 _return:
 
-  taosMemoryFreeClear(action.data);
+  taosMemoryFreeClear(op->data);
   CTG_RET(code);
 }
 
 int32_t ctgUpdateVgroupEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId, SDBVgInfo* dbInfo, bool syncOp) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_UPDATE_VGROUP, .syncOp = syncOp};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_UPDATE_VGROUP;
+  op->syncOp = syncOp;
+  
   SCtgUpdateVgMsg *msg = taosMemoryMalloc(sizeof(SCtgUpdateVgMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgUpdateVgMsg));
@@ -701,22 +699,25 @@ int32_t ctgUpdateVgroupEnqueue(SCatalog* pCtg, const char *dbFName, int64_t dbId
   msg->dbId = dbId;
   msg->dbInfo = dbInfo;
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
 
 _return:
 
   ctgFreeVgInfo(dbInfo);
-  taosMemoryFreeClear(action.data);
+  taosMemoryFreeClear(op->data);
   CTG_RET(code);
 }
 
 int32_t ctgUpdateTbMetaEnqueue(SCatalog* pCtg, STableMetaOutput *output, bool syncOp) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_UPDATE_TB_META, .syncOp = syncOp};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_UPDATE_TB_META;
+  op->syncOp = syncOp;
+  
   SCtgUpdateTblMsg *msg = taosMemoryMalloc(sizeof(SCtgUpdateTblMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgUpdateTblMsg));
@@ -731,9 +732,9 @@ int32_t ctgUpdateTbMetaEnqueue(SCatalog* pCtg, STableMetaOutput *output, bool sy
   msg->pCtg = pCtg;
   msg->output = output;
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
   
@@ -746,7 +747,9 @@ _return:
 
 int32_t ctgUpdateVgEpsetEnqueue(SCatalog* pCtg, char *dbFName, int32_t vgId, SEpSet* pEpSet) {
   int32_t code = 0;
-  SCtgCacheOperation operation= {.opId = CTG_OP_UPDATE_VG_EPSET};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_UPDATE_VG_EPSET;
+  
   SCtgUpdateEpsetMsg *msg = taosMemoryMalloc(sizeof(SCtgUpdateEpsetMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgUpdateEpsetMsg));
@@ -758,9 +761,9 @@ int32_t ctgUpdateVgEpsetEnqueue(SCatalog* pCtg, char *dbFName, int32_t vgId, SEp
   msg->vgId = vgId;
   msg->epSet = *pEpSet;
 
-  operation.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &operation));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
 
   return TSDB_CODE_SUCCESS;
   
@@ -775,7 +778,10 @@ _return:
 
 int32_t ctgUpdateUserEnqueue(SCatalog* pCtg, SGetUserAuthRsp *pAuth, bool syncOp) {
   int32_t code = 0;
-  SCtgCacheOperation action= {.opId = CTG_OP_UPDATE_USER, .syncOp = syncOp};
+  SCtgCacheOperation *op = taosMemoryCalloc(1, sizeof(SCtgCacheOperation));
+  op->opId = CTG_OP_UPDATE_USER;
+  op->syncOp = syncOp;
+  
   SCtgUpdateUserMsg *msg = taosMemoryMalloc(sizeof(SCtgUpdateUserMsg));
   if (NULL == msg) {
     ctgError("malloc %d failed", (int32_t)sizeof(SCtgUpdateUserMsg));
@@ -785,9 +791,9 @@ int32_t ctgUpdateUserEnqueue(SCatalog* pCtg, SGetUserAuthRsp *pAuth, bool syncOp
   msg->pCtg = pCtg;
   msg->userAuth = *pAuth;
 
-  action.data = msg;
+  op->data = msg;
 
-  CTG_ERR_JRET(ctgEnqueue(pCtg, &action));
+  CTG_ERR_JRET(ctgEnqueue(pCtg, op));
   
   return TSDB_CODE_SUCCESS;
   
@@ -1607,6 +1613,39 @@ void ctgUpdateThreadUnexpectedStopped(void) {
   if (CTG_IS_LOCKED(&gCtgMgmt.lock) > 0) CTG_UNLOCK(CTG_READ, &gCtgMgmt.lock);
 }
 
+void ctgCleanupCacheQueue(void) {
+  SCtgQNode *node = NULL;
+  SCtgQNode *nodeNext = NULL;
+
+  while (true) {
+    node = gCtgMgmt.queue.head->next;
+    while (node) {
+      if (node->op) {
+        taosMemoryFree(node->op->data);
+        if (node->op->syncOp) {
+          tsem_post(&node->op->rspSem);
+        } else {
+          taosMemoryFree(node->op);
+        }
+      }
+
+      nodeNext = node->next;
+      taosMemoryFree(node);
+      
+      node = nodeNext;
+    }
+
+    if (CTG_IS_LOCKED(&gCtgMgmt.lock)) {
+      taosUsleep(1);
+    } else {
+      break;
+    }
+  }
+
+  taosMemoryFreeClear(gCtgMgmt.queue.head);
+  gCtgMgmt.queue.tail = NULL;
+}
+
 void* ctgUpdateThreadFunc(void* param) {
   setThreadName("catalog");
 #ifdef WINDOWS
@@ -1622,7 +1661,8 @@ void* ctgUpdateThreadFunc(void* param) {
     }
     
     if (atomic_load_8((int8_t*)&gCtgMgmt.exit)) {
-      tsem_post(&gCtgMgmt.queue.rspSem);
+      CTG_UNLOCK(CTG_READ, &gCtgMgmt.lock);
+      ctgCleanupCacheQueue();
       break;
     }
 
@@ -1634,10 +1674,8 @@ void* ctgUpdateThreadFunc(void* param) {
     
     (*gCtgCacheOperation[operation->opId].func)(operation);
 
-    gCtgMgmt.queue.seqDone = operation->seqId;
-
     if (operation->syncOp) {
-      tsem_post(&gCtgMgmt.queue.rspSem);
+      tsem_post(&operation->rspSem);
     }
 
     CTG_RT_STAT_INC(qDoneNum, 1); 

source/libs/catalog/src/ctgRemote.c

@@ -85,6 +85,21 @@ int32_t ctgProcessRspMsg(void* out, int32_t reqType, char* msg, int32_t msgSize,
       qDebug("Got index from mnode, indexName:%s", target);
       break;
     }
+    case TDMT_MND_GET_TABLE_INDEX: {
+      if (TSDB_CODE_SUCCESS != rspCode) {
+        qError("error rsp for get table index, error:%s, tbFName:%s", tstrerror(rspCode), target);
+        CTG_ERR_RET(rspCode);
+      }
+      
+      code = queryProcessMsgRsp[TMSG_INDEX(reqType)](out, msg, msgSize);
+      if (code) {
+        qError("Process get table index rsp failed, error:%s, tbFName:%s", tstrerror(code), target);
+        CTG_ERR_RET(code);
+      }
+      
+      qDebug("Got table index from mnode, tbFName:%s", target);
+      break;
+    }
     case TDMT_MND_RETRIEVE_FUNC: {
       if (TSDB_CODE_SUCCESS != rspCode) {
         qError("error rsp for get udf, error:%s, funcName:%s", tstrerror(rspCode), target);
@@ -412,6 +427,44 @@ int32_t ctgGetIndexInfoFromMnode(CTG_PARAMS, const char *indexName, SIndexInfo *
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t ctgGetTbIndexFromMnode(CTG_PARAMS, const char *tbFName, SArray** out, SCtgTask* pTask) {
+  char *msg = NULL;
+  int32_t msgLen = 0;
+  int32_t reqType = TDMT_MND_GET_TABLE_INDEX;
+  void*(*mallocFp)(int32_t) = pTask ? taosMemoryMalloc : rpcMallocCont;
+
+  ctgDebug("try to get tb index from mnode, tbFName:%s", tbFName);
+
+  int32_t code = queryBuildMsg[TMSG_INDEX(reqType)]((void *)tbFName, &msg, 0, &msgLen, mallocFp);
+  if (code) {
+    ctgError("Build get index msg failed, code:%s, tbFName:%s", tstrerror(code), tbFName);
+    CTG_ERR_RET(code);
+  }
+
+  if (pTask) {
+    void* pOut = taosMemoryCalloc(1, POINTER_BYTES);
+    if (NULL == pOut) {
+      CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
+    }
+    CTG_ERR_RET(ctgUpdateMsgCtx(&pTask->msgCtx, reqType, pOut, (char*)tbFName));
+    
+    CTG_RET(ctgAsyncSendMsg(CTG_PARAMS_LIST(), pTask, reqType, msg, msgLen));
+  }
+  
+  SRpcMsg rpcMsg = {
+      .msgType = reqType,
+      .pCont   = msg,
+      .contLen = msgLen,
+  };
+
+  SRpcMsg rpcRsp = {0};
+  rpcSendRecv(pTrans, (SEpSet*)pMgmtEps, &rpcMsg, &rpcRsp);
+
+  CTG_ERR_RET(ctgProcessRspMsg(out, reqType, rpcRsp.pCont, rpcRsp.contLen, rpcRsp.code, (char*)tbFName));
+  
+  return TSDB_CODE_SUCCESS;
+}
+
 int32_t ctgGetUdfInfoFromMnode(CTG_PARAMS, const char *funcName, SFuncInfo *out, SCtgTask* pTask) {
   char *msg = NULL;
   int32_t msgLen = 0;

source/libs/qcom/src/querymsg.c

@@ -203,6 +203,24 @@ int32_t queryBuildGetUserAuthMsg(void *input, char **msg, int32_t msgSize, int32
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t queryBuildGetTbIndexMsg(void *input, char **msg, int32_t msgSize, int32_t *msgLen, void*(*mallcFp)(int32_t)) {
+  if (NULL == msg || NULL == msgLen) {
+    return TSDB_CODE_TSC_INVALID_INPUT;
+  }
+
+  STableIndexReq indexReq = {0};
+  strcpy(indexReq.tbFName, input);
+
+  int32_t bufLen = tSerializeSTableIndexReq(NULL, 0, &indexReq);
+  void   *pBuf = (*mallcFp)(bufLen);
+  tSerializeSTableIndexReq(pBuf, bufLen, &indexReq);
+
+  *msg = pBuf;
+  *msgLen = bufLen;
+
+  return TSDB_CODE_SUCCESS;
+}
+
 
 int32_t queryProcessUseDBRsp(void *output, char *msg, int32_t msgSize) {
   SUseDbOutput *pOut = output;
@@ -459,26 +477,43 @@ int32_t queryProcessGetUserAuthRsp(void *output, char *msg, int32_t msgSize) {
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t queryProcessGetTbIndexRsp(void *output, char *msg, int32_t msgSize) {
+  if (NULL == output || NULL == msg || msgSize <= 0) {
+    return TSDB_CODE_TSC_INVALID_INPUT;
+  }
 
-void initQueryModuleMsgHandle() {
-  queryBuildMsg[TMSG_INDEX(TDMT_VND_TABLE_META)] = queryBuildTableMetaReqMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_TABLE_META)] = queryBuildTableMetaReqMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_USE_DB)]     = queryBuildUseDbMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_QNODE_LIST)] = queryBuildQnodeListMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_DB_CFG)] = queryBuildGetDBCfgMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_INDEX)]  = queryBuildGetIndexMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_RETRIEVE_FUNC)]  = queryBuildRetrieveFuncMsg;
-  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_USER_AUTH)]  = queryBuildGetUserAuthMsg;
+  STableIndexRsp out = {0};
+  if (tDeserializeSTableIndexRsp(msg, msgSize, &out) != 0) {
+    qError("tDeserializeSTableIndexRsp failed, msgSize:%d", msgSize);
+    return TSDB_CODE_INVALID_MSG;
+  }
+
+  *(void **)output = out.pIndex;
   
+  return TSDB_CODE_SUCCESS;
+}
+
 
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_VND_TABLE_META)] = queryProcessTableMetaRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_TABLE_META)] = queryProcessTableMetaRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_USE_DB)]     = queryProcessUseDBRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_QNODE_LIST)] = queryProcessQnodeListRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_DB_CFG)] = queryProcessGetDbCfgRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_INDEX)]  = queryProcessGetIndexRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_RETRIEVE_FUNC)]  = queryProcessRetrieveFuncRsp;
-  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_USER_AUTH)]  = queryProcessGetUserAuthRsp;
+void initQueryModuleMsgHandle() {
+  queryBuildMsg[TMSG_INDEX(TDMT_VND_TABLE_META)]       = queryBuildTableMetaReqMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_TABLE_META)]       = queryBuildTableMetaReqMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_USE_DB)]           = queryBuildUseDbMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_QNODE_LIST)]       = queryBuildQnodeListMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_DB_CFG)]       = queryBuildGetDBCfgMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_INDEX)]        = queryBuildGetIndexMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_RETRIEVE_FUNC)]    = queryBuildRetrieveFuncMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_USER_AUTH)]    = queryBuildGetUserAuthMsg;
+  queryBuildMsg[TMSG_INDEX(TDMT_MND_GET_TABLE_INDEX)]  = queryBuildGetTbIndexMsg;
+
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_VND_TABLE_META)]      = queryProcessTableMetaRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_TABLE_META)]      = queryProcessTableMetaRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_USE_DB)]          = queryProcessUseDBRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_QNODE_LIST)]      = queryProcessQnodeListRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_DB_CFG)]      = queryProcessGetDbCfgRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_INDEX)]       = queryProcessGetIndexRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_RETRIEVE_FUNC)]   = queryProcessRetrieveFuncRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_USER_AUTH)]   = queryProcessGetUserAuthRsp;
+  queryProcessMsgRsp[TMSG_INDEX(TDMT_MND_GET_TABLE_INDEX)] = queryProcessGetTbIndexRsp;
 }
 
 #pragma GCC diagnostic pop

source/libs/qworker/inc/qwInt.h

@@ -356,7 +356,7 @@ int32_t qwOpenRef(void);
 void qwSetHbParam(int64_t refId, SQWHbParam **pParam);
 int32_t qwUpdateTimeInQueue(SQWorker *mgmt, int64_t ts, EQueueType type);
 int64_t qwGetTimeInQueue(SQWorker *mgmt, EQueueType type);
-void qwClearExpiredSch(SArray* pExpiredSch);
+void qwClearExpiredSch(SQWorker *mgmt, SArray* pExpiredSch);
 int32_t qwAcquireScheduler(SQWorker *mgmt, uint64_t sId, int32_t rwType, SQWSchStatus **sch);
 void qwFreeTaskCtx(QW_FPARAMS_DEF, SQWTaskCtx *ctx);
 

source/libs/qworker/src/qwUtil.c

@@ -539,8 +539,23 @@ int64_t qwGetTimeInQueue(SQWorker *mgmt, EQueueType type) {
 }
 
 
-void qwClearExpiredSch(SArray* pExpiredSch) {
+void qwClearExpiredSch(SQWorker *mgmt, SArray* pExpiredSch) {
+  int32_t num = taosArrayGetSize(pExpiredSch);
+  for (int32_t i = 0; i < num; ++i) {
+    uint64_t *sId = taosArrayGet(pExpiredSch, i);
+    SQWSchStatus *pSch = NULL;
+    if (qwAcquireScheduler(mgmt, *sId, QW_WRITE, &pSch)) {
+      continue;
+    }
+
+    if (taosHashGetSize(pSch->tasksHash) <= 0) {
+      qwDestroySchStatus(pSch);
+      taosHashRemove(mgmt->schHash, sId, sizeof(*sId));
+      qError("sch %" PRIx64 "destroyed", *sId);
+    }
 
+    qwReleaseScheduler(QW_WRITE, mgmt);
+  }
 }
 
 

source/libs/qworker/src/qworker.c

@@ -790,9 +790,10 @@ int32_t qwProcessHb(SQWorker *mgmt, SQWMsg *qwMsg, SSchedulerHbReq *req) {
   }
 
   QW_ERR_JRET(qwAcquireAddScheduler(mgmt, req->sId, QW_READ, &sch));
-
   QW_ERR_JRET(qwRegisterHbBrokenLinkArg(mgmt, req->sId, &qwMsg->connInfo));
 
+  sch->hbBrokenTs = 0;
+
   QW_LOCK(QW_WRITE, &sch->hbConnLock);
 
   if (sch->hbConnInfo.handle) {
@@ -912,7 +913,7 @@ _return:
   }
 
   if (taosArrayGetSize(pExpiredSch) > 0) {
-    qwClearExpiredSch(pExpiredSch);
+    qwClearExpiredSch(mgmt, pExpiredSch);
   }
 
   taosMemoryFreeClear(rspList);

source/libs/scheduler/inc/schedulerInt.h

@@ -207,6 +207,7 @@ typedef struct SSchJob {
   SArray          *dataSrcTasks; // SArray<SQueryTask*>
   int32_t          levelIdx;
   SEpSet           dataSrcEps;
+  SHashObj        *taskList;
   SHashObj        *execTasks; // executing tasks, key:taskid, value:SQueryTask*
   SHashObj        *succTasks; // succeed tasks, key:taskid, value:SQueryTask*
   SHashObj        *failTasks; // failed tasks, key:taskid, value:SQueryTask*

source/libs/scheduler/src/schJob.c

@@ -63,6 +63,13 @@ int32_t schInitJob(SSchJob **pSchJob, SQueryPlan *pDag, void *pTrans, SArray *pN
   if (pNodeList != NULL) {
     pJob->nodeList = taosArrayDup(pNodeList);
   }
+  
+  pJob->taskList =
+      taosHashInit(pDag->numOfSubplans, taosGetDefaultHashFunction(TSDB_DATA_TYPE_UBIGINT), false, HASH_ENTRY_LOCK);
+  if (NULL == pJob->taskList) {
+    SCH_JOB_ELOG("taosHashInit %d taskList failed", pDag->numOfSubplans);
+    SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+  }
 
   SCH_ERR_JRET(schValidateAndBuildJob(pDag, pJob));
 
@@ -486,23 +493,26 @@ int32_t schValidateAndBuildJob(SQueryPlan *pDag, SSchJob *pJob) {
       SCH_SET_JOB_TYPE(pJob, plan->subplanType);
 
       SSchTask  task = {0};
-      SSchTask *pTask = &task;
-
       SCH_ERR_JRET(schInitTask(pJob, &task, plan, pLevel));
 
-      void *p = taosArrayPush(pLevel->subTasks, &task);
-      if (NULL == p) {
+      SSchTask *pTask = taosArrayPush(pLevel->subTasks, &task);
+      if (NULL == pTask) {
         SCH_TASK_ELOG("taosArrayPush task to level failed, level:%d, taskIdx:%d", pLevel->level, n);
         SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
       }
 
-      SCH_ERR_JRET(schRecordQueryDataSrc(pJob, p));
+      SCH_ERR_JRET(schRecordQueryDataSrc(pJob, pTask));
 
-      if (0 != taosHashPut(planToTask, &plan, POINTER_BYTES, &p, POINTER_BYTES)) {
+      if (0 != taosHashPut(planToTask, &plan, POINTER_BYTES, &pTask, POINTER_BYTES)) {
         SCH_TASK_ELOG("taosHashPut to planToTaks failed, taskIdx:%d", n);
         SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
       }
 
+      if (0 != taosHashPut(pJob->taskList, &pTask->taskId, sizeof(pTask->taskId), &pTask, POINTER_BYTES)) {
+        SCH_TASK_ELOG("taosHashPut to taskList failed, taskIdx:%d", n);
+        SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      }
+
       ++pJob->taskNum;
     }
 
@@ -1276,14 +1286,10 @@ int32_t schGetTaskFromList(SHashObj *pTaskList, uint64_t taskId, SSchTask **pTas
 }
 
 int32_t schGetTaskInJob(SSchJob *pJob, uint64_t taskId, SSchTask **pTask) {
-  schGetTaskFromList(pJob->execTasks, taskId, pTask);
+  schGetTaskFromList(pJob->taskList, taskId, pTask);
   if (NULL == *pTask) {
-    schGetTaskFromList(pJob->succTasks, taskId, pTask);
-
-    if (NULL == *pTask) {
-      SCH_JOB_ELOG("task not found in execList & succList, taskId:%" PRIx64, taskId);
-      SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
-    }
+    SCH_JOB_ELOG("task not found in job task list, taskId:%" PRIx64, taskId);
+    SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
   }
 
   return TSDB_CODE_SUCCESS;
@@ -1445,7 +1451,8 @@ void schFreeJobImpl(void *job) {
   taosHashCleanup(pJob->execTasks);
   taosHashCleanup(pJob->failTasks);
   taosHashCleanup(pJob->succTasks);
-
+  taosHashCleanup(pJob->taskList);
+  
   taosArrayDestroy(pJob->levels);
   taosArrayDestroy(pJob->nodeList);
   taosArrayDestroy(pJob->dataSrcTasks);