Merge remote-tracking branch 'origin/3.0' into fix/mnode

cmake/cmake.define

@@ -71,8 +71,8 @@ ELSE ()
     ENDIF ()
 
     IF (${SANITIZER} MATCHES "true")
-        SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS}     -Werror -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fsanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=null -fno-sanitize=alignment -g3")
-        SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Wno-literal-suffix -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fsanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=null -fno-sanitize=alignment -g3")
+        SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS}     -Werror -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fsanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=shift-base -fno-sanitize=alignment -g3")
+        SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Werror -Wno-literal-suffix -Werror=return-type -fPIC -gdwarf-2 -fsanitize=address -fsanitize=undefined -fsanitize-recover=all -fsanitize=float-divide-by-zero -fsanitize=float-cast-overflow -fno-sanitize=shift-base -fno-sanitize=alignment -g3")
 	MESSAGE(STATUS "Will compile with Address Sanitizer!")
     ELSE ()
         SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Werror -Werror=return-type -fPIC -gdwarf-2 -g3")

include/common/tcommon.h

@@ -219,6 +219,16 @@ typedef struct {
 
 #define GET_FORWARD_DIRECTION_FACTOR(ord) (((ord) == TSDB_ORDER_ASC) ? QUERY_ASC_FORWARD_STEP : QUERY_DESC_FORWARD_STEP)
 
+#define SORT_QSORT_T 0x1
+#define SORT_SPILLED_MERGE_SORT_T 0x2
+typedef struct SSortExecInfo {
+  int32_t sortMethod;
+  int32_t sortBuffer;
+  int32_t loops;       // loop count
+  int32_t writeBytes;  // write io bytes
+  int32_t readBytes;   // read io bytes
+} SSortExecInfo;
+
 #ifdef __cplusplus
 }
 #endif

include/common/tdatablock.h

@@ -198,7 +198,7 @@ void    colDataTrim(SColumnInfoData* pColumnInfoData);
 size_t blockDataGetNumOfCols(const SSDataBlock* pBlock);
 size_t blockDataGetNumOfRows(const SSDataBlock* pBlock);
 
-int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc, SArray* pIndexMap);
+int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc);
 int32_t blockDataSplitRows(SSDataBlock* pBlock, bool hasVarCol, int32_t startIndex, int32_t* stopIndex,
                            int32_t pageSize);
 int32_t blockDataToBuf(char* buf, const SSDataBlock* pBlock);

include/common/tdataformat.h

@@ -61,9 +61,10 @@ int32_t tTSRowBuilderGetRow(STSRowBuilder *pBuilder, const STSRow2 **ppRow);
 // STag
 int32_t tTagNew(STagVal *pTagVals, int16_t nTag, STag **ppTag);
 void    tTagFree(STag *pTag);
-void    tTagGet(STag *pTag, int16_t cid, int8_t type, uint8_t **ppData, int32_t *nData);
-int32_t tEncodeTag(SEncoder *pEncoder, STag *pTag);
-int32_t tDecodeTag(SDecoder *pDecoder, const STag **ppTag);
+int32_t tTagSet(STag *pTag, SSchema *pSchema, int32_t nCols, int iCol, uint8_t *pData, uint32_t nData, STag **ppTag);
+void    tTagGet(STag *pTag, int16_t cid, int8_t type, uint8_t **ppData, uint32_t *nData);
+int32_t tEncodeTag(SEncoder *pEncoder, const STag *pTag);
+int32_t tDecodeTag(SDecoder *pDecoder, STag **ppTag);
 
 // STRUCT =================
 struct STColumn {

include/common/tmsg.h

@@ -660,8 +660,7 @@ typedef struct {
   int32_t tz;  // query client timezone
   char    intervalUnit;
   char    slidingUnit;
-  char
-      offsetUnit;  // TODO Remove it, the offset is the number of precision tickle, and it must be a immutable duration.
+  char    offsetUnit;
   int8_t  precision;
   int64_t interval;
   int64_t sliding;

include/util/taoserror.h

@@ -313,6 +313,7 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_VND_INVALID_TABLE_ACTION      TAOS_DEF_ERROR_CODE(0, 0x0519)
 #define TSDB_CODE_VND_COL_ALREADY_EXISTS        TAOS_DEF_ERROR_CODE(0, 0x051a)
 #define TSDB_CODE_VND_TABLE_COL_NOT_EXISTS      TAOS_DEF_ERROR_CODE(0, 0x051b)
+#define TSDB_CODE_VND_READ_END                  TAOS_DEF_ERROR_CODE(0, 0x051c)
 
 // tsdb
 #define TSDB_CODE_TDB_INVALID_TABLE_ID          TAOS_DEF_ERROR_CODE(0, 0x0600)

source/common/src/tdatablock.c

@@ -361,19 +361,13 @@ int32_t blockDataUpdateTsWindow(SSDataBlock* pDataBlock, int32_t tsColumnIndex)
   return 0;
 }
 
-// if pIndexMap = NULL, merger one column by on column
-int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc, SArray* pIndexMap) {
+int32_t blockDataMerge(SSDataBlock* pDest, const SSDataBlock* pSrc) {
   assert(pSrc != NULL && pDest != NULL);
   int32_t capacity = pDest->info.capacity;
 
   for (int32_t i = 0; i < pDest->info.numOfCols; ++i) {
-    int32_t mapIndex = i;
-    //    if (pIndexMap) {
-    //      mapIndex = *(int32_t*)taosArrayGet(pIndexMap, i);
-    //    }
-
     SColumnInfoData* pCol2 = taosArrayGet(pDest->pDataBlock, i);
-    SColumnInfoData* pCol1 = taosArrayGet(pSrc->pDataBlock, mapIndex);
+    SColumnInfoData* pCol1 = taosArrayGet(pSrc->pDataBlock, i);
 
     capacity = pDest->info.capacity;
     colDataMergeCol(pCol2, pDest->info.rows, &capacity, pCol1, pSrc->info.rows);

source/common/src/tdataformat.c

@@ -581,7 +581,52 @@ void tTagFree(STag *pTag) {
   if (pTag) taosMemoryFree(pTag);
 }
 
-void tTagGet(STag *pTag, int16_t cid, int8_t type, uint8_t **ppData, int32_t *nData) {
+int32_t tTagSet(STag *pTag, SSchema *pSchema, int32_t nCols, int iCol, uint8_t *pData, uint32_t nData, STag **ppTag) {
+  STagVal *pTagVals;
+  int16_t  nTags = 0;
+  SSchema *pColumn;
+  uint8_t *p;
+  uint32_t n;
+
+  pTagVals = (STagVal *)taosMemoryMalloc(sizeof(*pTagVals) * nCols);
+  if (pTagVals == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  for (int32_t i = 0; i < nCols; i++) {
+    pColumn = &pSchema[i];
+
+    if (i == iCol) {
+      p = pData;
+      n = nData;
+    } else {
+      tTagGet(pTag, pColumn->colId, pColumn->type, &p, &n);
+    }
+
+    if (p == NULL) continue;
+
+    ASSERT(IS_VAR_DATA_TYPE(pColumn->type) || n == pColumn->bytes);
+
+    pTagVals[nTags].cid = pColumn->colId;
+    pTagVals[nTags].type = pColumn->type;
+    pTagVals[nTags].nData = n;
+    pTagVals[nTags].pData = p;
+
+    nTags++;
+  }
+
+  // create new tag
+  if (tTagNew(pTagVals, nTags, ppTag) < 0) {
+    taosMemoryFree(pTagVals);
+    return -1;
+  }
+
+  taosMemoryFree(pTagVals);
+  return 0;
+}
+
+void tTagGet(STag *pTag, int16_t cid, int8_t type, uint8_t **ppData, uint32_t *nData) {
   STagIdx *pTagIdx = bsearch(&((STagIdx){.cid = cid}), pTag->idx, pTag->nTag, sizeof(STagIdx), tTagIdxCmprFn);
   if (pTagIdx == NULL) {
     *ppData = NULL;
@@ -597,18 +642,11 @@ void tTagGet(STag *pTag, int16_t cid, int8_t type, uint8_t **ppData, int32_t *nD
   }
 }
 
-int32_t tEncodeTag(SEncoder *pEncoder, STag *pTag) {
-  // return tEncodeBinary(pEncoder, (uint8_t *)pTag, pTag->len);
-  ASSERT(0);
-  return 0;
+int32_t tEncodeTag(SEncoder *pEncoder, const STag *pTag) {
+  return tEncodeBinary(pEncoder, (const uint8_t *)pTag, pTag->len);
 }
 
-int32_t tDecodeTag(SDecoder *pDecoder, const STag **ppTag) {
-  // uint32_t n;
-  // return tDecodeBinary(pDecoder, (const uint8_t **)ppTag, &n);
-  ASSERT(0);
-  return 0;
-}
+int32_t tDecodeTag(SDecoder *pDecoder, STag **ppTag) { return tDecodeBinary(pDecoder, (uint8_t **)ppTag, NULL); }
 
 #if 1  // ===================================================================================================================
 static void dataColSetNEleNull(SDataCol *pCol, int nEle);
@@ -1087,7 +1125,7 @@ SKVRow tdGetKVRowFromBuilder(SKVRowBuilder *pBuilder) {
   kvRowSetNCols(row, pBuilder->nCols);
   kvRowSetLen(row, tlen);
 
-  if(pBuilder->nCols > 0){
+  if (pBuilder->nCols > 0) {
     memcpy(kvRowColIdx(row), pBuilder->pColIdx, sizeof(SColIdx) * pBuilder->nCols);
     memcpy(kvRowValues(row), pBuilder->buf, pBuilder->size);
   }

source/dnode/vnode/CMakeLists.txt

@@ -13,6 +13,8 @@ target_sources(
     "src/vnd/vnodeModule.c"
     "src/vnd/vnodeSvr.c"
     "src/vnd/vnodeSync.c"
+    "src/vnd/vnodeSnapshot.c"
+    "src/vnd/vnodeUtil.c"
 
     # meta
     "src/meta/metaOpen.c"
@@ -22,6 +24,7 @@ target_sources(
     "src/meta/metaQuery.c"
     "src/meta/metaCommit.c"
     "src/meta/metaEntry.c"
+    "src/meta/metaSnapshot.c"
 
     # sma
     "src/sma/sma.c"
@@ -44,6 +47,7 @@ target_sources(
     "src/tsdb/tsdbReadImpl.c"
     # "src/tsdb/tsdbSma.c"
     "src/tsdb/tsdbWrite.c"
+    "src/tsdb/tsdbSnapshot.c"
 
     # tq
     "src/tq/tq.c"

source/dnode/vnode/inc/vnode.h

@@ -42,6 +42,7 @@ extern "C" {
 typedef struct SVnode           SVnode;
 typedef struct STsdbCfg         STsdbCfg;  // todo: remove
 typedef struct SVnodeCfg        SVnodeCfg;
+typedef struct SVSnapshotReader SVSnapshotReader;
 
 extern const SVnodeCfg vnodeCfgDefault;
 
@@ -59,13 +60,14 @@ int32_t vnodeProcessQueryMsg(SVnode *pVnode, SRpcMsg *pMsg);
 int32_t vnodeProcessFetchMsg(SVnode *pVnode, SRpcMsg *pMsg, SQueueInfo *pInfo);
 int32_t vnodeGetLoad(SVnode *pVnode, SVnodeLoad *pLoad);
 int32_t vnodeValidateTableHash(SVnode *pVnode, char *tableFName);
-
 int32_t vnodeStart(SVnode *pVnode);
 void    vnodeStop(SVnode *pVnode);
-
 int64_t vnodeGetSyncHandle(SVnode *pVnode);
 void    vnodeGetSnapshot(SVnode *pVnode, SSnapshot *pSnapshot);
 void    vnodeGetInfo(SVnode *pVnode, const char **dbname, int32_t *vgId);
+int32_t vnodeSnapshotReaderOpen(SVnode *pVnode, SVSnapshotReader **ppReader, int64_t sver, int64_t ever);
+int32_t vnodeSnapshotReaderClose(SVSnapshotReader *pReader);
+int32_t vnodeSnapshotRead(SVSnapshotReader *pReader, const void **ppData, uint32_t *nData);
 
 // meta
 typedef struct SMeta       SMeta;  // todo: remove

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

@@ -56,6 +56,8 @@ typedef struct SVState      SVState;
 typedef struct SVBufPool           SVBufPool;
 typedef struct SQWorker            SQHandle;
 typedef struct STsdbKeepCfg        STsdbKeepCfg;
+typedef struct SMetaSnapshotReader SMetaSnapshotReader;
+typedef struct STsdbSnapshotReader STsdbSnapshotReader;
 
 #define VNODE_META_DIR  "meta"
 #define VNODE_TSDB_DIR  "tsdb"
@@ -69,6 +71,8 @@ typedef struct STsdbKeepCfg STsdbKeepCfg;
 // vnd.h
 void*   vnodeBufPoolMalloc(SVBufPool* pPool, int size);
 void    vnodeBufPoolFree(SVBufPool* pPool, void* p);
+int32_t vnodeRealloc(void** pp, int32_t size);
+void    vnodeFree(void* p);
 
 // meta
 typedef struct SMCtbCursor SMCtbCursor;
@@ -95,6 +99,9 @@ STSma*          metaGetSmaInfoByIndex(SMeta* pMeta, int64_t indexUid);
 STSmaWrapper*   metaGetSmaInfoByTable(SMeta* pMeta, tb_uid_t uid, bool deepCopy);
 SArray*         metaGetSmaIdsByTable(SMeta* pMeta, tb_uid_t uid);
 SArray*         metaGetSmaTbUids(SMeta* pMeta);
+int32_t         metaSnapshotReaderOpen(SMeta* pMeta, SMetaSnapshotReader** ppReader, int64_t sver, int64_t ever);
+int32_t         metaSnapshotReaderClose(SMetaSnapshotReader* pReader);
+int32_t         metaSnapshotRead(SMetaSnapshotReader* pReader, void** ppData, uint32_t* nData);
 
 int32_t metaCreateTSma(SMeta* pMeta, int64_t version, SSmaCfg* pCfg);
 int32_t metaDropTSma(SMeta* pMeta, int64_t indexUid);
@@ -112,6 +119,9 @@ tsdbReaderT* tsdbQueryTables(SVnode* pVnode, SQueryTableDataCond* pCond, STableG
 tsdbReaderT  tsdbQueryCacheLastT(STsdb* tsdb, SQueryTableDataCond* pCond, STableGroupInfo* groupList, uint64_t qId,
                                  void* pMemRef);
 int32_t      tsdbGetTableGroupFromIdListT(STsdb* tsdb, SArray* pTableIdList, STableGroupInfo* pGroupInfo);
+int32_t      tsdbSnapshotReaderOpen(STsdb* pTsdb, STsdbSnapshotReader** ppReader, int64_t sver, int64_t ever);
+int32_t      tsdbSnapshotReaderClose(STsdbSnapshotReader* pReader);
+int32_t      tsdbSnapshotRead(STsdbSnapshotReader* pReader, void** ppData, uint32_t* nData);
 
 // tq
 STQ*    tqOpen(const char* path, SVnode* pVnode, SWal* pWal);

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "meta.h"
+
+struct SMetaSnapshotReader {
+  SMeta*  pMeta;
+  TBC*    pTbc;
+  int64_t sver;
+  int64_t ever;
+};
+
+int32_t metaSnapshotReaderOpen(SMeta* pMeta, SMetaSnapshotReader** ppReader, int64_t sver, int64_t ever) {
+  int32_t              code = 0;
+  int32_t              c = 0;
+  SMetaSnapshotReader* pMetaReader = NULL;
+
+  pMetaReader = (SMetaSnapshotReader*)taosMemoryCalloc(1, sizeof(*pMetaReader));
+  if (pMetaReader == NULL) {
+    code = TSDB_CODE_OUT_OF_MEMORY;
+    goto _err;
+  }
+  pMetaReader->pMeta = pMeta;
+  pMetaReader->sver = sver;
+  pMetaReader->ever = ever;
+  code = tdbTbcOpen(pMeta->pTbDb, &pMetaReader->pTbc, NULL);
+  if (code) {
+    goto _err;
+  }
+
+  code = tdbTbcMoveTo(pMetaReader->pTbc, &(STbDbKey){.version = sver, .uid = INT64_MIN}, sizeof(STbDbKey), &c);
+  if (code) {
+    goto _err;
+  }
+
+  *ppReader = pMetaReader;
+  return code;
+
+_err:
+  *ppReader = NULL;
+  return code;
+}
+
+int32_t metaSnapshotReaderClose(SMetaSnapshotReader* pReader) {
+  if (pReader) {
+    tdbTbcClose(pReader->pTbc);
+    taosMemoryFree(pReader);
+  }
+  return 0;
+}
+
+int32_t metaSnapshotRead(SMetaSnapshotReader* pReader, void** ppData, uint32_t* nDatap) {
+  const void* pKey = NULL;
+  const void* pData = NULL;
+  int32_t     nKey = 0;
+  int32_t     nData = 0;
+  int32_t     code = 0;
+
+  for (;;) {
+    code = tdbTbcGet(pReader->pTbc, &pKey, &nKey, &pData, &nData);
+    if (code || ((STbDbKey*)pData)->version > pReader->ever) {
+      return TSDB_CODE_VND_READ_END;
+    }
+
+    if (((STbDbKey*)pData)->version < pReader->sver) {
+      continue;
+    }
+
+    break;
+  }
+
+  // copy the data
+  if (vnodeRealloc(ppData, nData) < 0) {
+    code = TSDB_CODE_OUT_OF_MEMORY;
+    return code;
+  }
+
+  memcpy(*ppData, pData, nData);
+  *nDatap = nData;
+  return code;
+}
\ No newline at end of file

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

@@ -23,6 +23,7 @@ static int metaUpdateTtlIdx(SMeta *pMeta, const SMetaEntry *pME);
 static int metaSaveToSkmDb(SMeta *pMeta, const SMetaEntry *pME);
 static int metaUpdateCtbIdx(SMeta *pMeta, const SMetaEntry *pME);
 static int metaUpdateTagIdx(SMeta *pMeta, const SMetaEntry *pCtbEntry);
+static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type);
 
 int metaCreateSTable(SMeta *pMeta, int64_t version, SVCreateStbReq *pReq) {
   SMetaEntry  me = {0};
@@ -71,64 +72,71 @@ _err:
 }
 
 int metaDropSTable(SMeta *pMeta, int64_t verison, SVDropStbReq *pReq) {
-  TBC        *pNameIdxc = NULL;
-  TBC        *pUidIdxc = NULL;
-  TBC        *pCtbIdxc = NULL;
-  SCtbIdxKey *pCtbIdxKey;
-  const void *pKey = NULL;
-  int         nKey;
-  const void *pData = NULL;
-  int         nData;
-  int         c, ret;
-
-  // prepare uid idx cursor
-  tdbTbcOpen(pMeta->pUidIdx, &pUidIdxc, &pMeta->txn);
-  ret = tdbTbcMoveTo(pUidIdxc, &pReq->suid, sizeof(tb_uid_t), &c);
-  if (ret < 0 || c != 0) {
-    terrno = TSDB_CODE_VND_TB_NOT_EXIST;
-    tdbTbcClose(pUidIdxc);
-    goto _err;
-  }
+  void *pKey = NULL;
+  int   nKey = 0;
+  void *pData = NULL;
+  int   nData = 0;
+  int   c = 0;
+  int   rc = 0;
 
-  // prepare name idx cursor
-  tdbTbcOpen(pMeta->pNameIdx, &pNameIdxc, &pMeta->txn);
-  ret = tdbTbcMoveTo(pNameIdxc, pReq->name, strlen(pReq->name) + 1, &c);
-  if (ret < 0 || c != 0) {
-    ASSERT(0);
+  // check if super table exists
+  rc = tdbTbGet(pMeta->pNameIdx, pReq->name, strlen(pReq->name) + 1, &pData, &nData);
+  if (rc < 0 || *(tb_uid_t *)pData != pReq->suid) {
+    terrno = TSDB_CODE_VND_TABLE_NOT_EXIST;
+    return -1;
   }
 
-  tdbTbcDelete(pUidIdxc);
-  tdbTbcDelete(pNameIdxc);
-  tdbTbcClose(pUidIdxc);
-  tdbTbcClose(pNameIdxc);
+  // drop all child tables
+  TBC    *pCtbIdxc = NULL;
+  SArray *pArray = taosArrayInit(8, sizeof(tb_uid_t));
 
-  // loop to drop each child table
   tdbTbcOpen(pMeta->pCtbIdx, &pCtbIdxc, &pMeta->txn);
-  ret = tdbTbcMoveTo(pCtbIdxc, &(SCtbIdxKey){.suid = pReq->suid, .uid = INT64_MIN}, sizeof(SCtbIdxKey), &c);
-  if (ret < 0 || (c < 0 && tdbTbcMoveToNext(pCtbIdxc) < 0)) {
+  rc = tdbTbcMoveTo(pCtbIdxc, &(SCtbIdxKey){.suid = pReq->suid, .uid = INT64_MIN}, sizeof(SCtbIdxKey), &c);
+  if (rc < 0) {
     tdbTbcClose(pCtbIdxc);
-    goto _exit;
+    metaWLock(pMeta);
+    goto _drop_super_table;
   }
 
   for (;;) {
-    tdbTbcGet(pCtbIdxc, &pKey, &nKey, NULL, NULL);
-    pCtbIdxKey = (SCtbIdxKey *)pKey;
+    rc = tdbTbcNext(pCtbIdxc, &pKey, &nKey, NULL, NULL);
+    if (rc < 0) break;
+
+    if (((SCtbIdxKey *)pKey)->suid < pReq->suid) {
+      continue;
+    } else if (((SCtbIdxKey *)pKey)->suid > pReq->suid) {
+      break;
+    }
+
+    taosArrayPush(pArray, &(((SCtbIdxKey *)pKey)->uid));
+  }
 
-    if (pCtbIdxKey->suid > pReq->suid) break;
+  tdbTbcClose(pCtbIdxc);
 
-    // drop the child table (TODO)
+  metaWLock(pMeta);
 
-    if (tdbTbcMoveToNext(pCtbIdxc) < 0) break;
+  for (int32_t iChild = 0; iChild < taosArrayGetSize(pArray); iChild++) {
+    tb_uid_t uid = *(tb_uid_t *)taosArrayGet(pArray, iChild);
+    metaDropTableByUid(pMeta, uid, NULL);
   }
 
+  taosArrayDestroy(pArray);
+
+  // drop super table
+_drop_super_table:
+  tdbTbGet(pMeta->pUidIdx, &pReq->suid, sizeof(tb_uid_t), &pData, &nData);
+  tdbTbDelete(pMeta->pTbDb, &(STbDbKey){.version = *(int64_t *)pData, .uid = pReq->suid}, sizeof(STbDbKey),
+              &pMeta->txn);
+  tdbTbDelete(pMeta->pNameIdx, pReq->name, strlen(pReq->name) + 1, &pMeta->txn);
+  tdbTbDelete(pMeta->pUidIdx, &pReq->suid, sizeof(tb_uid_t), &pMeta->txn);
+
+  metaULock(pMeta);
+
 _exit:
+  tdbFree(pKey);
+  tdbFree(pData);
   metaDebug("vgId:%d  super table %s uid:%" PRId64 " is dropped", TD_VID(pMeta->pVnode), pReq->name, pReq->suid);
   return 0;
-
-_err:
-  metaError("vgId:%d failed to drop super table %s uid:%" PRId64 " since %s", TD_VID(pMeta->pVnode), pReq->name,
-            pReq->suid, tstrerror(terrno));
-  return -1;
 }
 
 int metaAlterSTable(SMeta *pMeta, int64_t version, SVCreateStbReq *pReq) {
@@ -256,122 +264,63 @@ _err:
 }
 
 int metaDropTable(SMeta *pMeta, int64_t version, SVDropTbReq *pReq, SArray *tbUids) {
-  TBC        *pTbDbc = NULL;
-  TBC        *pUidIdxc = NULL;
-  TBC        *pNameIdxc = NULL;
-  const void *pData;
-  int         nData;
+  void    *pData = NULL;
+  int      nData = 0;
+  int      rc = 0;
   tb_uid_t uid;
-  int64_t     tver;
-  SMetaEntry  me = {0};
-  SDecoder    coder = {0};
-  int8_t      type;
-  int64_t     ctime;
-  tb_uid_t    suid;
-  int         c = 0, ret;
-
-  // search & delete the name idx
-  tdbTbcOpen(pMeta->pNameIdx, &pNameIdxc, &pMeta->txn);
-  ret = tdbTbcMoveTo(pNameIdxc, pReq->name, strlen(pReq->name) + 1, &c);
-  if (ret < 0 || !tdbTbcIsValid(pNameIdxc) || c) {
-    tdbTbcClose(pNameIdxc);
-    terrno = TSDB_CODE_VND_TABLE_NOT_EXIST;
-    return -1;
-  }
+  int      type;
 
-  ret = tdbTbcGet(pNameIdxc, NULL, NULL, &pData, &nData);
-  if (ret < 0) {
-    ASSERT(0);
+  rc = tdbTbGet(pMeta->pNameIdx, pReq->name, strlen(pReq->name) + 1, &pData, &nData);
+  if (rc < 0) {
+    terrno = TSDB_CODE_VND_TABLE_NOT_EXIST;
     return -1;
   }
-
   uid = *(tb_uid_t *)pData;
 
-  tdbTbcDelete(pNameIdxc);
-  tdbTbcClose(pNameIdxc);
-
-  // search & delete uid idx
-  tdbTbcOpen(pMeta->pUidIdx, &pUidIdxc, &pMeta->txn);
-  ret = tdbTbcMoveTo(pUidIdxc, &uid, sizeof(uid), &c);
-  if (ret < 0 || c != 0) {
-    ASSERT(0);
-    return -1;
-  }
-
-  ret = tdbTbcGet(pUidIdxc, NULL, NULL, &pData, &nData);
-  if (ret < 0) {
-    ASSERT(0);
-    return -1;
-  }
-
-  tver = *(int64_t *)pData;
-  tdbTbcDelete(pUidIdxc);
-  tdbTbcClose(pUidIdxc);
+  metaWLock(pMeta);
+  metaDropTableByUid(pMeta, uid, &type);
+  metaULock(pMeta);
 
-  // search and get meta entry
-  tdbTbcOpen(pMeta->pTbDb, &pTbDbc, &pMeta->txn);
-  ret = tdbTbcMoveTo(pTbDbc, &(STbDbKey){.uid = uid, .version = tver}, sizeof(STbDbKey), &c);
-  if (ret < 0 || c != 0) {
-    ASSERT(0);
-    return -1;
+  if (type == TSDB_CHILD_TABLE && tbUids) {
+    taosArrayPush(tbUids, &uid);
   }
 
-  ret = tdbTbcGet(pTbDbc, NULL, NULL, &pData, &nData);
-  if (ret < 0) {
-    ASSERT(0);
-    return -1;
-  }
-
-  // decode entry
-  void *pDataCopy = taosMemoryMalloc(nData);  // remove the copy (todo)
-  memcpy(pDataCopy, pData, nData);
-  tDecoderInit(&coder, pDataCopy, nData);
-  ret = metaDecodeEntry(&coder, &me);
-  if (ret < 0) {
-    ASSERT(0);
-    return -1;
-  }
+  tdbFree(pData);
+  return 0;
+}
 
-  type = me.type;
-  if (type == TSDB_CHILD_TABLE) {
-    ctime = me.ctbEntry.ctime;
-    suid = me.ctbEntry.suid;
-    taosArrayPush(tbUids, &me.uid);
-  } else if (type == TSDB_NORMAL_TABLE) {
-    ctime = me.ntbEntry.ctime;
-    suid = 0;
-  } else {
-    ASSERT(0);
-  }
+static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type) {
+  void      *pData = NULL;
+  int        nData = 0;
+  int        rc = 0;
+  int64_t    version;
+  SMetaEntry e = {0};
+  SDecoder   dc = {0};
 
-  taosMemoryFree(pDataCopy);
-  tDecoderClear(&coder);
-  tdbTbcClose(pTbDbc);
+  rc = tdbTbGet(pMeta->pUidIdx, &uid, sizeof(uid), &pData, &nData);
+  version = *(int64_t *)pData;
 
-  if (type == TSDB_CHILD_TABLE) {
-    // remove the pCtbIdx
-    TBC *pCtbIdxc = NULL;
-    tdbTbcOpen(pMeta->pCtbIdx, &pCtbIdxc, &pMeta->txn);
+  tdbTbGet(pMeta->pTbDb, &(STbDbKey){.version = version, .uid = uid}, sizeof(STbDbKey), &pData, &nData);
 
-    ret = tdbTbcMoveTo(pCtbIdxc, &(SCtbIdxKey){.suid = suid, .uid = uid}, sizeof(SCtbIdxKey), &c);
-    if (ret < 0 || c != 0) {
-      ASSERT(0);
-      return -1;
-    }
+  tDecoderInit(&dc, pData, nData);
+  metaDecodeEntry(&dc, &e);
 
-    tdbTbcDelete(pCtbIdxc);
-    tdbTbcClose(pCtbIdxc);
+  if (type) *type = e.type;
 
-    // remove tags from pTagIdx (todo)
-  } else if (type == TSDB_NORMAL_TABLE) {
-    // remove from pSkmDb
-  } else {
-    ASSERT(0);
+  tdbTbDelete(pMeta->pTbDb, &(STbDbKey){.version = version, .uid = uid}, sizeof(STbDbKey), &pMeta->txn);
+  tdbTbDelete(pMeta->pNameIdx, e.name, strlen(e.name) + 1, &pMeta->txn);
+  tdbTbDelete(pMeta->pUidIdx, &uid, sizeof(uid), &pMeta->txn);
+  if (e.type == TSDB_CHILD_TABLE) {
+    tdbTbDelete(pMeta->pCtbIdx, &(SCtbIdxKey){.suid = e.ctbEntry.suid, .uid = uid}, sizeof(SCtbIdxKey), &pMeta->txn);
+  } else if (e.type == TSDB_NORMAL_TABLE) {
+    // drop schema.db (todo)
+    // drop ttl.idx (todo)
+  } else if (e.type == TSDB_SUPER_TABLE) {
+    // drop schema.db (todo)
   }
 
-  // remove from ttl (todo)
-  if (ctime > 0) {
-  }
+  tDecoderClear(&dc);
+  tdbFree(pData);
 
   return 0;
 }
@@ -608,14 +557,14 @@ static int metaUpdateTableTagVal(SMeta *pMeta, int64_t version, SVAlterTbReq *pA
     // TODO : need to update tag index
   }
   ctbEntry.version = version;
-  if(pTagSchema->nCols == 1 && pTagSchema->pSchema[0].type == TSDB_DATA_TYPE_JSON){
+  if (pTagSchema->nCols == 1 && pTagSchema->pSchema[0].type == TSDB_DATA_TYPE_JSON) {
     ctbEntry.ctbEntry.pTags = taosMemoryMalloc(pAlterTbReq->nTagVal);
-    if(ctbEntry.ctbEntry.pTags == NULL){
+    if (ctbEntry.ctbEntry.pTags == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       goto _err;
     }
-    memcpy((void*)ctbEntry.ctbEntry.pTags, pAlterTbReq->pTagVal, pAlterTbReq->nTagVal);
-  }else{
+    memcpy((void *)ctbEntry.ctbEntry.pTags, pAlterTbReq->pTagVal, pAlterTbReq->nTagVal);
+  } else {
     SKVRowBuilder kvrb = {0};
     const SKVRow  pOldTag = (const SKVRow)ctbEntry.ctbEntry.pTags;
     SKVRow        pNewTag = NULL;
@@ -649,7 +598,7 @@ static int metaUpdateTableTagVal(SMeta *pMeta, int64_t version, SVAlterTbReq *pA
 
   tDecoderClear(&dc1);
   tDecoderClear(&dc2);
-  if (ctbEntry.ctbEntry.pTags) taosMemoryFree((void*)ctbEntry.ctbEntry.pTags);
+  if (ctbEntry.ctbEntry.pTags) taosMemoryFree((void *)ctbEntry.ctbEntry.pTags);
   if (ctbEntry.pBuf) taosMemoryFree(ctbEntry.pBuf);
   if (stbEntry.pBuf) tdbFree(stbEntry.pBuf);
   tdbTbcClose(pTbDbc);

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

@@ -51,6 +51,47 @@ int tqExecKeyCompare(const void* pKey1, int32_t kLen1, const void* pKey2, int32_
   return strcmp(pKey1, pKey2);
 }
 
+int32_t tqStoreExec(STQ* pTq, const char* key, const STqExec* pExec) {
+  int32_t code;
+  int32_t vlen;
+  tEncodeSize(tEncodeSTqExec, pExec, vlen, code);
+  ASSERT(code == 0);
+
+  void* buf = taosMemoryCalloc(1, vlen);
+  if (buf == NULL) {
+    ASSERT(0);
+  }
+
+  SEncoder encoder;
+  tEncoderInit(&encoder, buf, vlen);
+
+  if (tEncodeSTqExec(&encoder, pExec) < 0) {
+    ASSERT(0);
+  }
+
+  TXN txn;
+
+  if (tdbTxnOpen(&txn, 0, tdbDefaultMalloc, tdbDefaultFree, NULL, TDB_TXN_WRITE | TDB_TXN_READ_UNCOMMITTED) < 0) {
+    ASSERT(0);
+  }
+
+  if (tdbBegin(pTq->pMetaStore, &txn) < 0) {
+    ASSERT(0);
+  }
+
+  if (tdbTbUpsert(pTq->pExecStore, key, (int)strlen(key), buf, vlen, &txn) < 0) {
+    ASSERT(0);
+  }
+
+  if (tdbCommit(pTq->pMetaStore, &txn) < 0) {
+    ASSERT(0);
+  }
+
+  tEncoderClear(&encoder);
+  taosMemoryFree(buf);
+  return 0;
+}
+
 STQ* tqOpen(const char* path, SVnode* pVnode, SWal* pWal) {
   STQ* pTq = taosMemoryMalloc(sizeof(STQ));
   if (pTq == NULL) {
@@ -96,8 +137,31 @@ STQ* tqOpen(const char* path, SVnode* pVnode, SWal* pWal) {
   int   vLen;
 
   tdbTbcMoveToFirst(pCur);
+  SDecoder decoder;
   while (tdbTbcNext(pCur, &pKey, &kLen, &pVal, &vLen) == 0) {
-    // create, put into execsj
+    STqExec exec;
+    tDecoderInit(&decoder, (uint8_t*)pVal, vLen);
+    tDecodeSTqExec(&decoder, &exec);
+    exec.pWalReader = walOpenReadHandle(pTq->pVnode->pWal);
+    if (exec.subType == TOPIC_SUB_TYPE__TABLE) {
+      for (int32_t i = 0; i < 5; i++) {
+        exec.pExecReader[i] = tqInitSubmitMsgScanner(pTq->pVnode->pMeta);
+
+        SReadHandle handle = {
+            .reader = exec.pExecReader[i],
+            .meta = pTq->pVnode->pMeta,
+            .pMsgCb = &pTq->pVnode->msgCb,
+        };
+        exec.task[i] = qCreateStreamExecTaskInfo(exec.qmsg, &handle);
+        ASSERT(exec.task[i]);
+      }
+    } else {
+      for (int32_t i = 0; i < 5; i++) {
+        exec.pExecReader[i] = tqInitSubmitMsgScanner(pTq->pVnode->pMeta);
+      }
+      exec.pDropTbUid = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_NO_LOCK);
+    }
+    taosHashPut(pTq->execs, pKey, kLen, &exec, sizeof(STqExec));
   }
 
   if (tdbTxnClose(&txn) < 0) {
@@ -604,7 +668,9 @@ int32_t tqProcessVgDeleteReq(STQ* pTq, char* msg, int32_t msgLen) {
     ASSERT(0);
   }
 
-  tdbTbDelete(pTq->pExecStore, pReq->subKey, (int)strlen(pReq->subKey), &txn);
+  if (tdbTbDelete(pTq->pExecStore, pReq->subKey, (int)strlen(pReq->subKey), &txn) < 0) {
+    /*ASSERT(0);*/
+  }
 
   if (tdbCommit(pTq->pMetaStore, &txn) < 0) {
     ASSERT(0);
@@ -659,60 +725,21 @@ int32_t tqProcessVgChangeReq(STQ* pTq, char* msg, int32_t msgLen) {
     }
     taosHashPut(pTq->execs, req.subKey, strlen(req.subKey), pExec, sizeof(STqExec));
 
-    int32_t code;
-    int32_t vlen;
-    tEncodeSize(tEncodeSTqExec, pExec, vlen, code);
-    ASSERT(code == 0);
-
-    void* buf = taosMemoryCalloc(1, vlen);
-    if (buf == NULL) {
-      ASSERT(0);
-    }
-
-    SEncoder encoder;
-    tEncoderInit(&encoder, buf, vlen);
-
-    if (tEncodeSTqExec(&encoder, pExec) < 0) {
-      ASSERT(0);
-    }
-
-    TXN txn;
-
-    if (tdbTxnOpen(&txn, 0, tdbDefaultMalloc, tdbDefaultFree, NULL, TDB_TXN_WRITE | TDB_TXN_READ_UNCOMMITTED) < 0) {
-      ASSERT(0);
-    }
-
-    if (tdbBegin(pTq->pMetaStore, &txn) < 0) {
-      ASSERT(0);
-    }
-
-    if (tdbTbUpsert(pTq->pExecStore, req.subKey, (int)strlen(req.subKey), buf, vlen, &txn) < 0) {
-      ASSERT(0);
-    }
-
-    if (tdbCommit(pTq->pMetaStore, &txn) < 0) {
-      ASSERT(0);
+    if (tqStoreExec(pTq, req.subKey, pExec) < 0) {
+      // TODO
     }
-
-    tEncoderClear(&encoder);
-    taosMemoryFree(buf);
-
     return 0;
   } else {
-    /*if (req.newConsumerId != -1) {*/
-    /*taosWLockLatch(&pExec->lock);*/
-    ASSERT(pExec->consumerId == req.oldConsumerId);
+    /*ASSERT(pExec->consumerId == req.oldConsumerId);*/
     // TODO handle qmsg and exec modification
     atomic_store_32(&pExec->epoch, -1);
     atomic_store_64(&pExec->consumerId, req.newConsumerId);
     atomic_add_fetch_32(&pExec->epoch, 1);
-    /*taosWUnLockLatch(&pExec->lock);*/
-    return 0;
-    /*} else {*/
+
+    if (tqStoreExec(pTq, req.subKey, pExec) < 0) {
       // TODO
-    /*taosHashRemove(pTq->tqMetaNew, req.subKey, strlen(req.subKey));*/
-    /*return 0;*/
-    /*}*/
+    }
+    return 0;
   }
 }
 

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "tsdb.h"
+
+struct STsdbSnapshotReader {
+  STsdb* pTsdb;
+  // TODO
+};
+
+int32_t tsdbSnapshotReaderOpen(STsdb* pTsdb, STsdbSnapshotReader** ppReader, int64_t sver, int64_t ever) {
+  // TODO
+  return 0;
+}
+
+int32_t tsdbSnapshotReaderClose(STsdbSnapshotReader* pReader) {
+  // TODO
+  return 0;
+}
+
+int32_t tsdbSnapshotRead(STsdbSnapshotReader* pReader, void** ppData, uint32_t* nData) {
+  // TODO
+  return 0;
+}

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "vnodeInt.h"
+
+struct SVSnapshotReader {
+  SVnode              *pVnode;
+  int64_t              sver;
+  int64_t              ever;
+  int8_t               isMetaEnd;
+  int8_t               isTsdbEnd;
+  SMetaSnapshotReader *pMetaReader;
+  STsdbSnapshotReader *pTsdbReader;
+  void                *pData;
+  int32_t              nData;
+};
+
+int32_t vnodeSnapshotReaderOpen(SVnode *pVnode, SVSnapshotReader **ppReader, int64_t sver, int64_t ever) {
+  SVSnapshotReader *pReader = NULL;
+
+  pReader = (SVSnapshotReader *)taosMemoryCalloc(1, sizeof(*pReader));
+  if (pReader == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    goto _err;
+  }
+  pReader->pVnode = pVnode;
+  pReader->sver = sver;
+  pReader->ever = ever;
+  pReader->isMetaEnd = 0;
+  pReader->isTsdbEnd = 0;
+
+  if (metaSnapshotReaderOpen(pVnode->pMeta, &pReader->pMetaReader, sver, ever) < 0) {
+    taosMemoryFree(pReader);
+    goto _err;
+  }
+
+  if (tsdbSnapshotReaderOpen(pVnode->pTsdb, &pReader->pTsdbReader, sver, ever) < 0) {
+    metaSnapshotReaderClose(pReader->pMetaReader);
+    taosMemoryFree(pReader);
+    goto _err;
+  }
+
+_exit:
+  *ppReader = pReader;
+  return 0;
+
+_err:
+  *ppReader = NULL;
+  return -1;
+}
+
+int32_t vnodeSnapshotReaderClose(SVSnapshotReader *pReader) {
+  if (pReader) {
+    vnodeFree(pReader->pData);
+    tsdbSnapshotReaderClose(pReader->pTsdbReader);
+    metaSnapshotReaderClose(pReader->pMetaReader);
+    taosMemoryFree(pReader);
+  }
+  return 0;
+}
+
+int32_t vnodeSnapshotRead(SVSnapshotReader *pReader, const void **ppData, uint32_t *nData) {
+  int32_t code = 0;
+
+  if (!pReader->isMetaEnd) {
+    code = metaSnapshotRead(pReader->pMetaReader, &pReader->pData, &pReader->nData);
+    if (code) {
+      if (code == TSDB_CODE_VND_READ_END) {
+        pReader->isMetaEnd = 1;
+      } else {
+        return code;
+      }
+    } else {
+      *ppData = pReader->pData;
+      *nData = pReader->nData;
+      return code;
+    }
+  }
+
+  if (!pReader->isTsdbEnd) {
+    code = tsdbSnapshotRead(pReader->pTsdbReader, &pReader->pData, &pReader->nData);
+    if (code) {
+      if (code == TSDB_CODE_VND_READ_END) {
+        pReader->isTsdbEnd = 1;
+      } else {
+        return code;
+      }
+    } else {
+      *ppData = pReader->pData;
+      *nData = pReader->nData;
+      return code;
+    }
+  }
+
+  code = TSDB_CODE_VND_READ_END;
+  return code;
+}
\ No newline at end of file

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

@@ -617,16 +617,18 @@ static int vnodeDebugPrintSingleSubmitMsg(SMeta *pMeta, SSubmitBlk *pBlock, SSub
   STSchema      *pSchema = NULL;
   tb_uid_t       suid = 0;
   STSRow        *row = NULL;
+  int32_t        rv = -1;
 
   tInitSubmitBlkIter(msgIter, pBlock, &blkIter);
   if (blkIter.row == NULL) return 0;
-  if (!pSchema || (suid != msgIter->suid)) {
+  if (!pSchema || (suid != msgIter->suid) || rv != TD_ROW_SVER(blkIter.row)) {
     if (pSchema) {
       taosMemoryFreeClear(pSchema);
     }
-    pSchema = metaGetTbTSchema(pMeta, msgIter->suid, 1);  // TODO: use the real schema
+    pSchema = metaGetTbTSchema(pMeta, msgIter->suid, TD_ROW_SVER(blkIter.row));  // TODO: use the real schema
     if (pSchema) {
       suid = msgIter->suid;
+      rv = TD_ROW_SVER(blkIter.row);
     }
   }
   if (!pSchema) {

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include "vnd.h"
+
+int32_t vnodeRealloc(void** pp, int32_t size) {
+  uint8_t* p = NULL;
+  int32_t  csize = 0;
+
+  if (*pp) {
+    p = (uint8_t*)(*pp) - sizeof(int32_t);
+    csize = *(int32_t*)p;
+  }
+
+  if (csize >= size) {
+    return 0;
+  }
+
+  p = (uint8_t*)taosMemoryRealloc(p, size);
+  if (p == NULL) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+  *(int32_t*)p = size;
+  *pp = p + sizeof(int32_t);
+
+  return 0;
+}
+
+void vnodeFree(void* p) {
+  if (p) {
+    taosMemoryFree(((uint8_t*)p) - sizeof(int32_t));
+  }
+}
\ No newline at end of file

source/libs/command/src/explain.c

@@ -16,6 +16,7 @@
 #include "commandInt.h"
 #include "plannodes.h"
 #include "query.h"
+#include "tcommon.h"
 
 int32_t qExplainGenerateResNode(SPhysiNode *pNode, SExplainGroup *group, SExplainResNode **pRes);
 int32_t qExplainAppendGroupResRows(void *pCtx, int32_t groupId, int32_t level);
@@ -637,13 +638,48 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
         QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
         EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       }
-      EXPLAIN_ROW_APPEND(EXPLAIN_COLUMNS_FORMAT, pSortNode->pSortKeys->length);
+
+      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, pSortNode->node.pOutputDataBlockDesc->totalRowSize);
+      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, "Sort Key: ");
+        if (pResNode->pExecInfo) {
+          for (int32_t i = 0; i < LIST_LENGTH(pSortNode->pSortKeys); ++i) {
+            SOrderByExprNode *ptn = 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, "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,
@@ -792,13 +828,8 @@ int32_t qExplainResNodeToRowsImpl(SExplainResNode *pResNode, SExplainCtx *ctx, i
         QRY_ERR_RET(qExplainBufAppendExecInfo(pResNode->pExecInfo, tbuf, &tlen));
         EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       }
-//      EXPLAIN_ROW_APPEND(EXPLAIN_FUNCTIONS_FORMAT, pPartNode->length);
-//      EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
       EXPLAIN_ROW_APPEND(EXPLAIN_WIDTH_FORMAT, pPartNode->node.pOutputDataBlockDesc->totalRowSize);
-//      if (pPartNode->pGroupKeys) {
-//        EXPLAIN_ROW_APPEND(EXPLAIN_BLANK_FORMAT);
-//        EXPLAIN_ROW_APPEND(EXPLAIN_GROUPS_FORMAT, pPartNode->pGroupKeys->length);
-//      }
+
       EXPLAIN_ROW_APPEND(EXPLAIN_RIGHT_PARENTHESIS_FORMAT);
       EXPLAIN_ROW_END();
       QRY_ERR_RET(qExplainResAppendRow(ctx, tbuf, tlen, level));

source/libs/executor/inc/executorimpl.h

@@ -628,12 +628,8 @@ typedef struct SSortOperatorInfo {
   SArray*      pColMatchInfo;  // for index map from table scan output
   int32_t      bufPageSize;
 
-  // TODO extact struct
   int64_t      startTs;       // sort start time
   uint64_t     sortElapsed;   // sort elapsed time, time to flush to disk not included.
-  uint64_t totalSize;     // total load bytes from remote
-  uint64_t totalRows;     // total number of rows
-  uint64_t totalElapsed;  // total elapsed time
 } SSortOperatorInfo;
 
 typedef struct STagFilterOperatorInfo {

source/libs/executor/inc/tsort.h

@@ -137,6 +137,14 @@ void* tsortGetValue(STupleHandle* pVHandle, int32_t colId);
  */
 SSDataBlock* tsortGetSortedDataBlock(const SSortHandle* pSortHandle);
 
+/**
+ * return the sort execution information.
+ *
+ * @param pHandle
+ * @return
+ */
+SSortExecInfo tsortGetSortExecInfo(SSortHandle* pHandle);
+
 #ifdef __cplusplus
 }
 #endif

source/libs/executor/src/executorimpl.c

@@ -1747,8 +1747,7 @@ void setFunctionResultOutput(SOptrBasicInfo* pInfo, SAggSupporter* pSup, int32_t
   SResultRow* pRow = doSetResultOutBufByKey(pSup->pResultBuf, pResultRowInfo, (char*)&tid, sizeof(tid), true, groupId,
                                             pTaskInfo, false, pSup);
 
-  ASSERT(pDataBlock->info.numOfCols == numOfExprs);
-  for (int32_t i = 0; i < pDataBlock->info.numOfCols; ++i) {
+  for (int32_t i = 0; i < numOfExprs; ++i) {
     struct SResultRowEntryInfo* pEntry = getResultCell(pRow, i, rowCellInfoOffset);
     cleanupResultRowEntry(pEntry);
 
@@ -1756,7 +1755,7 @@ void setFunctionResultOutput(SOptrBasicInfo* pInfo, SAggSupporter* pSup, int32_t
     pCtx[i].scanFlag = stage;
   }
 
-  initCtxOutputBuffer(pCtx, pDataBlock->info.numOfCols);
+  initCtxOutputBuffer(pCtx, numOfExprs);
 }
 
 void updateOutputBuf(SOptrBasicInfo* pBInfo, int32_t* bufCapacity, int32_t numOfInputRows) {

source/libs/executor/src/scanoperator.c

@@ -806,7 +806,7 @@ static SSDataBlock* getDataFromCatch(SStreamBlockScanInfo* pInfo) {
       SSDataBlock* pDB = createOneDataBlock(pInfo->pRes, false);
       blockDataFromBuf(pDB, buf);
       SSDataBlock* pSub = blockDataExtractBlock(pDB, pos->rowId, 1);
-      blockDataMerge(pInfo->pRes, pSub, NULL);
+      blockDataMerge(pInfo->pRes, pSub);
       blockDataDestroy(pDB);
       blockDataDestroy(pSub);
     }
@@ -1046,8 +1046,9 @@ static void destroySysScanOperator(void* param, int32_t numOfOutput) {
   blockDataDestroy(pInfo->pRes);
 
   const char* name = tNameGetTableName(&pInfo->name);
-  if (strncasecmp(name, TSDB_INS_TABLE_USER_TABLES, TSDB_TABLE_FNAME_LEN) == 0) {
+  if (strncasecmp(name, TSDB_INS_TABLE_USER_TABLES, TSDB_TABLE_FNAME_LEN) == 0 || pInfo->pCur != NULL) {
     metaCloseTbCursor(pInfo->pCur);
+    pInfo->pCur = NULL;
   }
 
   taosArrayDestroy(pInfo->scanCols);

source/libs/executor/src/sortoperator.c

@@ -2,6 +2,9 @@
 #include "executorimpl.h"
 
 static SSDataBlock* doSort(SOperatorInfo* pOperator);
+static int32_t doOpenSortOperator(SOperatorInfo* pOperator);
+static int32_t getExplainExecInfo(SOperatorInfo* pOptr, void** pOptrExplain, uint32_t* len);
+
 static void destroyOrderOperatorInfo(void* param, int32_t numOfOutput);
 
 SOperatorInfo* createSortOperatorInfo(SOperatorInfo* downstream, SSDataBlock* pResBlock, SArray* pSortInfo, SExprInfo* pExprInfo, int32_t numOfCols,
@@ -35,7 +38,7 @@ SOperatorInfo* createSortOperatorInfo(SOperatorInfo* downstream, SSDataBlock* pR
 
   pOperator->pTaskInfo = pTaskInfo;
   pOperator->fpSet =
-      createOperatorFpSet(operatorDummyOpenFn, doSort, NULL, NULL, destroyOrderOperatorInfo, NULL, NULL, NULL);
+      createOperatorFpSet(doOpenSortOperator, doSort, NULL, NULL, destroyOrderOperatorInfo, NULL, NULL, getExplainExecInfo);
 
   int32_t code = appendDownstream(pOperator, &downstream, 1);
   return pOperator;
@@ -121,20 +124,17 @@ void applyScalarFunction(SSDataBlock* pBlock, void* param) {
   }
 }
 
-SSDataBlock* doSort(SOperatorInfo* pOperator) {
-  if (pOperator->status == OP_EXEC_DONE) {
-    return NULL;
-  }
-
-  SExecTaskInfo*     pTaskInfo = pOperator->pTaskInfo;
+int32_t doOpenSortOperator(SOperatorInfo* pOperator) {
   SSortOperatorInfo* pInfo = pOperator->info;
+  SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
 
-  if (pOperator->status == OP_RES_TO_RETURN) {
-    return getSortedBlockData(pInfo->pSortHandle, pInfo->binfo.pRes, pOperator->resultInfo.capacity, pInfo->pColMatchInfo);
+  if (OPTR_IS_OPENED(pOperator)) {
+    return TSDB_CODE_SUCCESS;
   }
 
-//  pInfo->binfo.pRes is not equalled to the input datablock.
-//  int32_t numOfBufPage = pInfo->sortBufSize / pInfo->bufPageSize;
+  pInfo->startTs = taosGetTimestampUs();
+
+  //  pInfo->binfo.pRes is not equalled to the input datablock.
   pInfo->pSortHandle = tsortCreateSortHandle(pInfo->pSortInfo, pInfo->pColMatchInfo, SORT_SINGLESOURCE_SORT,
                                              -1, -1, NULL, pTaskInfo->id.str);
 
@@ -146,12 +146,39 @@ SSDataBlock* doSort(SOperatorInfo* pOperator) {
 
   int32_t code = tsortOpen(pInfo->pSortHandle);
   taosMemoryFreeClear(ps);
+
   if (code != TSDB_CODE_SUCCESS) {
     longjmp(pTaskInfo->env, terrno);
   }
 
+  pOperator->cost.openCost = (taosGetTimestampUs() - pInfo->startTs)/1000.0;
   pOperator->status = OP_RES_TO_RETURN;
-  return getSortedBlockData(pInfo->pSortHandle, pInfo->binfo.pRes, pOperator->resultInfo.capacity, pInfo->pColMatchInfo);
+
+  OPTR_SET_OPENED(pOperator);
+  return TSDB_CODE_SUCCESS;
+}
+
+SSDataBlock* doSort(SOperatorInfo* pOperator) {
+  if (pOperator->status == OP_EXEC_DONE) {
+    return NULL;
+  }
+
+  SExecTaskInfo*     pTaskInfo = pOperator->pTaskInfo;
+  SSortOperatorInfo* pInfo = pOperator->info;
+
+  int32_t code = pOperator->fpSet._openFn(pOperator);
+  if (code != TSDB_CODE_SUCCESS) {
+    longjmp(pTaskInfo->env, code);
+  }
+
+  SSDataBlock* pBlock = getSortedBlockData(pInfo->pSortHandle, pInfo->binfo.pRes, pOperator->resultInfo.capacity, pInfo->pColMatchInfo);
+
+  if (pBlock != NULL) {
+    pOperator->resultInfo.totalRows += pBlock->info.rows;
+  } else {
+    doSetOperatorCompleted(pOperator);
+  }
+  return pBlock;
 }
 
 void destroyOrderOperatorInfo(void* param, int32_t numOfOutput) {
@@ -161,3 +188,15 @@ void destroyOrderOperatorInfo(void* param, int32_t numOfOutput) {
   taosArrayDestroy(pInfo->pSortInfo);
   taosArrayDestroy(pInfo->pColMatchInfo);
 }
+
+int32_t getExplainExecInfo(SOperatorInfo* pOptr, void** pOptrExplain, uint32_t* len) {
+  ASSERT(pOptr != NULL);
+  SSortExecInfo* pInfo = taosMemoryCalloc(1, sizeof(SSortExecInfo));
+
+  SSortOperatorInfo *pOperatorInfo = (SSortOperatorInfo*)pOptr->info;
+
+  *pInfo = tsortGetSortExecInfo(pOperatorInfo->pSortHandle);
+  *pOptrExplain = pInfo;
+  *len = sizeof(SSortExecInfo);
+  return TSDB_CODE_SUCCESS;
+}

source/libs/executor/src/tsort.c

@@ -31,20 +31,16 @@ struct STupleHandle {
 
 struct SSortHandle {
   int32_t           type;
-
   int32_t           pageSize;
   int32_t           numOfPages;
   SDiskbasedBuf    *pBuf;
 
   SArray           *pSortInfo;
-  SArray           *pIndexMap;
   SArray           *pOrderedSource;
 
-  _sort_fetch_block_fn_t  fetchfp;
-  _sort_merge_compar_fn_t comparFn;
-  SMultiwayMergeTreeInfo *pMergeTree;
-  int64_t           startTs;
+  int32_t           loops;
   uint64_t          sortElapsed;
+  int64_t           startTs;
   uint64_t          totalElapsed;
 
   int32_t           sourceId;
@@ -53,13 +49,15 @@ struct SSortHandle {
   int32_t           numOfCompletedSources;
   bool              opened;
   const char       *idStr;
-
   bool              inMemSort;
   bool              needAdjust;
   STupleHandle      tupleHandle;
-
   void             *param;
   void (*beforeFp)(SSDataBlock* pBlock, void* param);
+
+  _sort_fetch_block_fn_t  fetchfp;
+  _sort_merge_compar_fn_t comparFn;
+  SMultiwayMergeTreeInfo *pMergeTree;
 };
 
 static int32_t msortComparFn(const void *pLeft, const void *pRight, void *param);
@@ -80,7 +78,7 @@ SSortHandle* tsortCreateSortHandle(SArray* pSortInfo, SArray* pIndexMap, int32_t
   pSortHandle->pageSize   = pageSize;
   pSortHandle->numOfPages = numOfPages;
   pSortHandle->pSortInfo  = pSortInfo;
-  pSortHandle->pIndexMap  = pIndexMap;
+  pSortHandle->loops      = 0;
 
   if (pBlock != NULL) {
     pSortHandle->pDataBlock = createOneDataBlock(pBlock, false);
@@ -415,6 +413,9 @@ static int32_t doInternalMergeSort(SSortHandle* pHandle) {
   int32_t numOfRows = blockDataGetCapacityInRow(pHandle->pDataBlock, pHandle->pageSize);
   blockDataEnsureCapacity(pHandle->pDataBlock, numOfRows);
 
+  // the initial pass + sortPass + final mergePass
+  pHandle->loops = sortPass + 2;
+
   size_t numOfSorted = taosArrayGetSize(pHandle->pOrderedSource);
   for(int32_t t = 0; t < sortPass; ++t) {
     int64_t st = taosGetTimestampUs();
@@ -502,12 +503,13 @@ static int32_t doInternalMergeSort(SSortHandle* pHandle) {
   return 0;
 }
 
-static int32_t createInitialSortedMultiSources(SSortHandle* pHandle) {
+static int32_t createInitialSources(SSortHandle* pHandle) {
   size_t sortBufSize = pHandle->numOfPages * pHandle->pageSize;
 
   if (pHandle->type == SORT_SINGLESOURCE_SORT) {
     SSortSource* source = taosArrayGetP(pHandle->pOrderedSource, 0);
     taosArrayClear(pHandle->pOrderedSource);
+    
     while (1) {
       SSDataBlock* pBlock = pHandle->fetchfp(source->param);
       if (pBlock == NULL) {
@@ -524,6 +526,7 @@ static int32_t createInitialSortedMultiSources(SSortHandle* pHandle) {
         } else {
           pHandle->pageSize = 4096;
         }
+        
         // todo!!
         pHandle->numOfPages = 1024;
         sortBufSize = pHandle->numOfPages * pHandle->pageSize;
@@ -535,7 +538,7 @@ static int32_t createInitialSortedMultiSources(SSortHandle* pHandle) {
       }
 
       // todo relocate the columns
-      int32_t code = blockDataMerge(pHandle->pDataBlock, pBlock, pHandle->pIndexMap);
+      int32_t code = blockDataMerge(pHandle->pDataBlock, pBlock);
       if (code != 0) {
         return code;
       }
@@ -569,6 +572,7 @@ static int32_t createInitialSortedMultiSources(SSortHandle* pHandle) {
         pHandle->cmpParam.numOfSources = 1;
         pHandle->inMemSort = true;
 
+        pHandle->loops = 1;
         pHandle->tupleHandle.rowIndex  = -1;
         pHandle->tupleHandle.pBlock = pHandle->pDataBlock;
         return 0;
@@ -592,7 +596,7 @@ int32_t tsortOpen(SSortHandle* pHandle) {
 
   pHandle->opened = true;
 
-  int32_t code = createInitialSortedMultiSources(pHandle);
+  int32_t code = createInitialSources(pHandle);
   if (code != TSDB_CODE_SUCCESS) {
     return code;
   }
@@ -692,3 +696,20 @@ void* tsortGetValue(STupleHandle* pVHandle, int32_t colIndex) {
   SColumnInfoData* pColInfo = TARRAY_GET_ELEM(pVHandle->pBlock->pDataBlock, colIndex);
   return colDataGetData(pColInfo, pVHandle->rowIndex);
 }
+
+SSortExecInfo tsortGetSortExecInfo(SSortHandle* pHandle) {
+  SSortExecInfo info = {0};
+
+  info.sortBuffer = pHandle->pageSize * pHandle->numOfPages;
+  info.sortMethod = pHandle->inMemSort? SORT_QSORT_T:SORT_SPILLED_MERGE_SORT_T;
+  info.loops  = pHandle->loops;
+
+  if (pHandle->pBuf != NULL) {
+    SDiskbasedBufStatis st = getDBufStatis(pHandle->pBuf);
+    info.writeBytes = st.flushBytes;
+    info.readBytes = st.loadBytes;
+  }
+
+  return info;
+}
+

source/libs/parser/src/parInsert.c

@@ -189,6 +189,7 @@ static int32_t createSName(SName* pName, SToken* pTableName, int32_t acctId, con
   const char* msg1 = "name too long";
   const char* msg2 = "invalid database name";
   const char* msg3 = "db is not specified";
+  const char* msg4 = "invalid table name";
 
   int32_t code = TSDB_CODE_SUCCESS;
   char*   p = strnchr(pTableName->z, TS_PATH_DELIMITER[0], pTableName->n, true);
@@ -207,6 +208,10 @@ static int32_t createSName(SName* pName, SToken* pTableName, int32_t acctId, con
     }
 
     int32_t tbLen = pTableName->n - dbLen - 1;
+    if (tbLen <= 0) {
+      return buildInvalidOperationMsg(pMsgBuf, msg4);
+    }
+
     char    tbname[TSDB_TABLE_FNAME_LEN] = {0};
     strncpy(tbname, p + 1, tbLen);
     /*tbLen = */ strdequote(tbname);

source/util/src/terror.c

@@ -315,6 +315,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_VND_TABLE_NOT_EXIST,          "Table does not exists
 TAOS_DEFINE_ERROR(TSDB_CODE_VND_INVALID_TABLE_ACTION,     "Invalid table action")
 TAOS_DEFINE_ERROR(TSDB_CODE_VND_COL_ALREADY_EXISTS,       "Table column already exists")
 TAOS_DEFINE_ERROR(TSDB_CODE_VND_TABLE_COL_NOT_EXISTS,     "Table column not exists")
+TAOS_DEFINE_ERROR(TSDB_CODE_VND_READ_END,                 "Read end")
 
 
 // tsdb

source/util/src/thash.c

@@ -708,7 +708,7 @@ SHashNode *doCreateHashNode(const void *key, size_t keyLen, const void *pData, s
   pNewNode->removed = 0;
   pNewNode->next = NULL;
 
-  memcpy(GET_HASH_NODE_DATA(pNewNode), pData, dsize);
+  if (pData) memcpy(GET_HASH_NODE_DATA(pNewNode), pData, dsize);
   memcpy(GET_HASH_NODE_KEY(pNewNode), key, keyLen);
 
   return pNewNode;
@@ -774,7 +774,7 @@ static void *taosHashReleaseNode(SHashObj *pHashObj, void *p, int *slot) {
         ASSERT(prevNode->next != prevNode);
       } else {
         pe->next = pOld->next;
-        SHashNode* x = pe->next;
+        SHashNode *x = pe->next;
         if (x != NULL) {
           ASSERT(x->next != x);
         }

source/util/src/tpagedbuf.c

@@ -549,12 +549,17 @@ void destroyDiskbasedBuf(SDiskbasedBuf* pBuf) {
   // print the statistics information
   {
     SDiskbasedBufStatis* ps = &pBuf->statis;
+    if (ps->loadPages == 0) {
       uDebug(
-        "Get/Release pages:%d/%d, flushToDisk:%.2f Kb (%d Pages), loadFromDisk:%.2f Kb (%d Pages), avgPageSize:%.2f "
-        "Kb\n",
+          "Get/Release pages:%d/%d, flushToDisk:%.2f Kb (%d Pages), loadFromDisk:%.2f Kb (%d Pages)",
+          ps->getPages, ps->releasePages, ps->flushBytes / 1024.0f, ps->flushPages, ps->loadBytes / 1024.0f, ps->loadPages);
+    } else {
+      uDebug(
+          "Get/Release pages:%d/%d, flushToDisk:%.2f Kb (%d Pages), loadFromDisk:%.2f Kb (%d Pages), avgPageSize:%.2f Kb",
           ps->getPages, ps->releasePages, ps->flushBytes / 1024.0f, ps->flushPages, ps->loadBytes / 1024.0f,
           ps->loadPages, ps->loadBytes / (1024.0 * ps->loadPages));
     }
+  }
 
   taosRemoveFile(pBuf->path);
   taosMemoryFreeClear(pBuf->path);

tests/script/jenkins/basic.txt

@@ -67,6 +67,7 @@
 # ---- stream
 ./test.sh -f tsim/stream/basic0.sim
 ./test.sh -f tsim/stream/basic1.sim
+./test.sh -f tsim/stream/basic2.sim
 ./test.sh -f tsim/stream/session0.sim
 ./test.sh -f tsim/stream/session1.sim