From 4dca6df296a3e3882f5cbd7a1e04a867377dc900 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Mon, 9 Jan 2023 11:50:34 +0800 Subject: [PATCH 001/124] enh: optimize the error message of the situation that percentile function does not support --- source/libs/parser/src/parTranslater.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c index 05d49bb027..69a39b956f 100644 --- a/source/libs/parser/src/parTranslater.c +++ b/source/libs/parser/src/parTranslater.c @@ -1550,11 +1550,14 @@ static int32_t translateRepeatScanFunc(STranslateContext* pCxt, SFunctionNode* p // select percentile() without from clause is also valid if ((NULL != pTable && (QUERY_NODE_REAL_TABLE != nodeType(pTable) || (TSDB_CHILD_TABLE != ((SRealTableNode*)pTable)->pMeta->tableType && - TSDB_NORMAL_TABLE != ((SRealTableNode*)pTable)->pMeta->tableType))) || - NULL != pSelect->pPartitionByList) { + TSDB_NORMAL_TABLE != ((SRealTableNode*)pTable)->pMeta->tableType)))) { return generateSyntaxErrMsgExt(&pCxt->msgBuf, TSDB_CODE_PAR_ONLY_SUPPORT_SINGLE_TABLE, "%s is only supported in single table query", pFunc->functionName); } + if (NULL != pSelect->pPartitionByList) { + return generateSyntaxErrMsgExt(&pCxt->msgBuf, TSDB_CODE_PAR_NOT_ALLOWED_FUNC, + "%s function is not supported in fill query", pFunc->functionName); + } return TSDB_CODE_SUCCESS; } From 7466b5042d2249f70d4429605e4c068e81c5fac5 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Thu, 12 Jan 2023 13:50:31 +0800 Subject: [PATCH 002/124] fix: stmt memory leak --- source/libs/parser/src/parInsertStmt.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-) diff --git a/source/libs/parser/src/parInsertStmt.c b/source/libs/parser/src/parInsertStmt.c index 4ed72e6c14..f5c362edf9 100644 --- a/source/libs/parser/src/parInsertStmt.c +++ b/source/libs/parser/src/parInsertStmt.c @@ -47,7 +47,7 @@ int32_t qBindStmtTagsValue(void* pBlock, void* boundTags, int64_t suid, const ch TAOS_MULTI_BIND* bind, char* msgBuf, int32_t msgBufLen) { STableDataBlocks* pDataBlock = (STableDataBlocks*)pBlock; SMsgBuf pBuf = {.buf = msgBuf, .len = msgBufLen}; - int32_t code = TSDB_CODE_SUCCESS; + int32_t code = TSDB_CODE_SUCCESS; SParsedDataColInfo* tags = (SParsedDataColInfo*)boundTags; if (NULL == tags) { return TSDB_CODE_APP_ERROR; @@ -137,7 +137,8 @@ int32_t qBindStmtTagsValue(void* pBlock, void* boundTags, int64_t suid, const ch } SVCreateTbReq tbReq = {0}; - insBuildCreateTbReq(&tbReq, tName, pTag, suid, sTableName, tagName, pDataBlock->pTableMeta->tableInfo.numOfTags, TSDB_DEFAULT_TABLE_TTL); + insBuildCreateTbReq(&tbReq, tName, pTag, suid, sTableName, tagName, pDataBlock->pTableMeta->tableInfo.numOfTags, + TSDB_DEFAULT_TABLE_TTL); code = insBuildCreateTbMsg(pDataBlock, &tbReq); tdDestroySVCreateTbReq(&tbReq); @@ -460,9 +461,7 @@ void qFreeStmtDataBlock(void* pDataBlock) { return; } - taosMemoryFreeClear(((STableDataBlocks*)pDataBlock)->pTableMeta); - taosMemoryFreeClear(((STableDataBlocks*)pDataBlock)->pData); - taosMemoryFreeClear(pDataBlock); + insDestroyDataBlock((STableDataBlocks*)pDataBlock); } void qDestroyStmtDataBlock(void* pBlock) { From 0e061fb53cb44117f5dcb5342b45f540625c7b58 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Thu, 12 Jan 2023 13:57:47 +0800 Subject: [PATCH 003/124] fix: stmt memory leak --- source/libs/parser/src/parTranslater.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c index 69a39b956f..9991c2c6ae 100644 --- a/source/libs/parser/src/parTranslater.c +++ b/source/libs/parser/src/parTranslater.c @@ -1556,7 +1556,7 @@ static int32_t translateRepeatScanFunc(STranslateContext* pCxt, SFunctionNode* p } if (NULL != pSelect->pPartitionByList) { return generateSyntaxErrMsgExt(&pCxt->msgBuf, TSDB_CODE_PAR_NOT_ALLOWED_FUNC, - "%s function is not supported in fill query", pFunc->functionName); + "%s function is not supported in partition query", pFunc->functionName); } return TSDB_CODE_SUCCESS; } From 4359562e9cde38df77ce722abc8c16794ea29913 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Thu, 12 Jan 2023 15:01:19 +0800 Subject: [PATCH 004/124] fix: stmt memory leak --- source/client/src/clientStmt.c | 15 ++++++++------- source/libs/parser/src/parInsertStmt.c | 4 +++- 2 files changed, 11 insertions(+), 8 deletions(-) diff --git a/source/client/src/clientStmt.c b/source/client/src/clientStmt.c index 82ea9e0d8f..c040d389cd 100644 --- a/source/client/src/clientStmt.c +++ b/source/client/src/clientStmt.c @@ -152,9 +152,10 @@ int32_t stmtRestoreQueryFields(STscStmt* pStmt) { return TSDB_CODE_SUCCESS; } -int32_t stmtUpdateBindInfo(TAOS_STMT* stmt, STableMeta* pTableMeta, void* tags, SName* tbName, const char* sTableName, bool autoCreateTbl) { +int32_t stmtUpdateBindInfo(TAOS_STMT* stmt, STableMeta* pTableMeta, void* tags, SName* tbName, const char* sTableName, + bool autoCreateTbl) { STscStmt* pStmt = (STscStmt*)stmt; - char tbFName[TSDB_TABLE_FNAME_LEN]; + char tbFName[TSDB_TABLE_FNAME_LEN]; tNameExtractFullName(tbName, tbFName); memcpy(&pStmt->bInfo.sname, tbName, sizeof(*tbName)); @@ -300,7 +301,7 @@ int32_t stmtCleanExecInfo(STscStmt* pStmt, bool keepTable, bool deepClean) { continue; } - if (STMT_TYPE_MULTI_INSERT == pStmt->sql.type) { + if (pBlocks->cloned) { qFreeStmtDataBlock(pBlocks); } else { qDestroyStmtDataBlock(pBlocks); @@ -776,9 +777,9 @@ int stmtAddBatch(TAOS_STMT* stmt) { int stmtUpdateTableUid(STscStmt* pStmt, SSubmitRsp* pRsp) { tscDebug("stmt start to update tbUid, blockNum: %d", pRsp->nBlocks); - int32_t code = 0; - int32_t finalCode = 0; - size_t keyLen = 0; + int32_t code = 0; + int32_t finalCode = 0; + size_t keyLen = 0; STableDataBlocks** pIter = taosHashIterate(pStmt->exec.pBlockHash, NULL); while (pIter) { STableDataBlocks* pBlock = *pIter; @@ -848,7 +849,7 @@ int stmtUpdateTableUid(STscStmt* pStmt, SSubmitRsp* pRsp) { pMeta->uid = pTableMeta->uid; pStmt->bInfo.tbUid = pTableMeta->uid; - taosMemoryFree(pTableMeta); + taosMemoryFree(pTableMeta); } pIter = taosHashIterate(pStmt->exec.pBlockHash, pIter); diff --git a/source/libs/parser/src/parInsertStmt.c b/source/libs/parser/src/parInsertStmt.c index f5c362edf9..2466e784c1 100644 --- a/source/libs/parser/src/parInsertStmt.c +++ b/source/libs/parser/src/parInsertStmt.c @@ -461,7 +461,9 @@ void qFreeStmtDataBlock(void* pDataBlock) { return; } - insDestroyDataBlock((STableDataBlocks*)pDataBlock); + taosMemoryFreeClear(((STableDataBlocks*)pDataBlock)->pTableMeta); + taosMemoryFreeClear(((STableDataBlocks*)pDataBlock)->pData); + taosMemoryFreeClear(pDataBlock); } void qDestroyStmtDataBlock(void* pBlock) { From 4c8a2caa1238afeb57e49806efac6d82c6bf56e8 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Thu, 12 Jan 2023 17:24:04 +0800 Subject: [PATCH 005/124] fix: stmt memory leak --- include/libs/parser/parser.h | 4 +++- source/client/src/clientStmt.c | 6 +----- source/libs/parser/src/parInsertStmt.c | 13 +++++++++++++ 3 files changed, 17 insertions(+), 6 deletions(-) diff --git a/include/libs/parser/parser.h b/include/libs/parser/parser.h index 9be79a539f..e6240f64b5 100644 --- a/include/libs/parser/parser.h +++ b/include/libs/parser/parser.h @@ -90,6 +90,7 @@ int32_t qCloneStmtDataBlock(void** pDst, void* pSrc); void qFreeStmtDataBlock(void* pDataBlock); int32_t qRebuildStmtDataBlock(void** pDst, void* pSrc, uint64_t uid, int32_t vgId); void qDestroyStmtDataBlock(void* pBlock); +void qDestroyStmtDataBlockExt(void* pBlock); STableMeta* qGetTableMetaInDataBlock(void* pDataBlock); int32_t qStmtBindParams(SQuery* pQuery, TAOS_MULTI_BIND* pParams, int32_t colIdx); @@ -108,7 +109,8 @@ int32_t qCreateSName(SName* pName, const char* pTableName, int32_t acctId, char* void* smlInitHandle(SQuery* pQuery); void smlDestroyHandle(void* pHandle); int32_t smlBindData(void* handle, SArray* tags, SArray* colsSchema, SArray* cols, bool format, STableMeta* pTableMeta, - char* tableName, const char* sTableName, int32_t sTableNameLen, int32_t ttl, char* msgBuf, int16_t msgBufLen); + char* tableName, const char* sTableName, int32_t sTableNameLen, int32_t ttl, char* msgBuf, + int16_t msgBufLen); int32_t smlBuildOutput(void* handle, SHashObj* pVgHash); int32_t rewriteToVnodeModifyOpStmt(SQuery* pQuery, SArray* pBufArray); diff --git a/source/client/src/clientStmt.c b/source/client/src/clientStmt.c index c040d389cd..8a677d7a5e 100644 --- a/source/client/src/clientStmt.c +++ b/source/client/src/clientStmt.c @@ -301,11 +301,7 @@ int32_t stmtCleanExecInfo(STscStmt* pStmt, bool keepTable, bool deepClean) { continue; } - if (pBlocks->cloned) { - qFreeStmtDataBlock(pBlocks); - } else { - qDestroyStmtDataBlock(pBlocks); - } + qDestroyStmtDataBlockExt(pBlocks); taosHashRemove(pStmt->exec.pBlockHash, key, keyLen); pIter = taosHashIterate(pStmt->exec.pBlockHash, pIter); diff --git a/source/libs/parser/src/parInsertStmt.c b/source/libs/parser/src/parInsertStmt.c index 2466e784c1..8a23cd32b4 100644 --- a/source/libs/parser/src/parInsertStmt.c +++ b/source/libs/parser/src/parInsertStmt.c @@ -476,3 +476,16 @@ void qDestroyStmtDataBlock(void* pBlock) { pDataBlock->cloned = false; insDestroyDataBlock(pDataBlock); } + +void qDestroyStmtDataBlockExt(void* pBlock) { + if (pBlock == NULL) { + return; + } + + STableDataBlocks* pDataBlock = (STableDataBlocks*)pBlock; + if (pDataBlock->cloned) { + qFreeStmtDataBlock(pBlock); + } else { + qDestroyStmtDataBlock(pBlock); + } +} From 9ba8d4f935204480a562cd736649089d20f48e05 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Mon, 16 Jan 2023 09:53:43 +0800 Subject: [PATCH 006/124] fix: rollback --- include/libs/parser/parser.h | 1 - source/client/src/clientStmt.c | 7 ++++++- source/libs/parser/src/parInsertStmt.c | 13 ------------- 3 files changed, 6 insertions(+), 15 deletions(-) diff --git a/include/libs/parser/parser.h b/include/libs/parser/parser.h index e6240f64b5..52edd9708c 100644 --- a/include/libs/parser/parser.h +++ b/include/libs/parser/parser.h @@ -90,7 +90,6 @@ int32_t qCloneStmtDataBlock(void** pDst, void* pSrc); void qFreeStmtDataBlock(void* pDataBlock); int32_t qRebuildStmtDataBlock(void** pDst, void* pSrc, uint64_t uid, int32_t vgId); void qDestroyStmtDataBlock(void* pBlock); -void qDestroyStmtDataBlockExt(void* pBlock); STableMeta* qGetTableMetaInDataBlock(void* pDataBlock); int32_t qStmtBindParams(SQuery* pQuery, TAOS_MULTI_BIND* pParams, int32_t colIdx); diff --git a/source/client/src/clientStmt.c b/source/client/src/clientStmt.c index 8a677d7a5e..3be1b58bad 100644 --- a/source/client/src/clientStmt.c +++ b/source/client/src/clientStmt.c @@ -301,7 +301,12 @@ int32_t stmtCleanExecInfo(STscStmt* pStmt, bool keepTable, bool deepClean) { continue; } - qDestroyStmtDataBlockExt(pBlocks); + if (STMT_TYPE_MULTI_INSERT == pStmt->sql.type) { + qFreeStmtDataBlock(pBlocks); + } else { + qDestroyStmtDataBlock(pBlocks); + } + taosHashRemove(pStmt->exec.pBlockHash, key, keyLen); pIter = taosHashIterate(pStmt->exec.pBlockHash, pIter); diff --git a/source/libs/parser/src/parInsertStmt.c b/source/libs/parser/src/parInsertStmt.c index 8a23cd32b4..2466e784c1 100644 --- a/source/libs/parser/src/parInsertStmt.c +++ b/source/libs/parser/src/parInsertStmt.c @@ -476,16 +476,3 @@ void qDestroyStmtDataBlock(void* pBlock) { pDataBlock->cloned = false; insDestroyDataBlock(pDataBlock); } - -void qDestroyStmtDataBlockExt(void* pBlock) { - if (pBlock == NULL) { - return; - } - - STableDataBlocks* pDataBlock = (STableDataBlocks*)pBlock; - if (pDataBlock->cloned) { - qFreeStmtDataBlock(pBlock); - } else { - qDestroyStmtDataBlock(pBlock); - } -} From 95493432ed0b8980c39dfead13fe6c10c7ff3475 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Fri, 13 Jan 2023 15:59:01 +0800 Subject: [PATCH 007/124] enh: trim the last WAL idx file which is not fsynced --- source/libs/wal/src/walMeta.c | 39 ++++++++++++++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) diff --git a/source/libs/wal/src/walMeta.c b/source/libs/wal/src/walMeta.c index 44e88a4dcc..a547378967 100644 --- a/source/libs/wal/src/walMeta.c +++ b/source/libs/wal/src/walMeta.c @@ -325,6 +325,35 @@ bool walLogEntriesComplete(const SWal* pWal) { return complete; } +int walTrimIdxFile(SWal* pWal, int32_t fileIdx) { + SWalFileInfo* pFileInfo = taosArrayGet(pWal->fileInfoSet, fileIdx); + ASSERT(pFileInfo != NULL); + char fnameStr[WAL_FILE_LEN]; + walBuildIdxName(pWal, pFileInfo->firstVer, fnameStr); + + int64_t fileSize = 0; + taosStatFile(fnameStr, &fileSize, NULL); + int64_t records = TMAX(0, pFileInfo->lastVer - pFileInfo->firstVer + 1); + int64_t lastEndOffset = records * sizeof(SWalIdxEntry); + + if (fileSize <= lastEndOffset) { + return 0; + } + + TdFilePtr pFile = taosOpenFile(fnameStr, TD_FILE_READ | TD_FILE_WRITE); + if (pFile == NULL) { + terrno = TAOS_SYSTEM_ERROR(errno); + return -1; + } + + wInfo("vgId:%d, trim idx file. file: %s, size: %" PRId64 ", offset: %" PRId64, pWal->cfg.vgId, fnameStr, fileSize, + lastEndOffset); + + taosFtruncateFile(pFile, lastEndOffset); + taosCloseFile(&pFile); + return 0; +} + int walCheckAndRepairMeta(SWal* pWal) { // load log files, get first/snapshot/last version info const char* logPattern = "^[0-9]+.log$"; @@ -402,6 +431,8 @@ int walCheckAndRepairMeta(SWal* pWal) { } updateMeta = true; + (void)walTrimIdxFile(pWal, fileIdx); + int64_t lastVer = walScanLogGetLastVer(pWal, fileIdx); if (lastVer < 0) { if (terrno != TSDB_CODE_WAL_LOG_NOT_EXIST) { @@ -567,6 +598,7 @@ int walCheckAndRepairIdxFile(SWal* pWal, int32_t fileIdx) { goto _err; } + int64_t count = 0; while (idxEntry.ver < pFileInfo->lastVer) { ASSERT(idxEntry.ver == ckHead.head.version); @@ -578,11 +610,11 @@ int walCheckAndRepairIdxFile(SWal* pWal, int32_t fileIdx) { idxEntry.offset, fLogNameStr); goto _err; } - wWarn("vgId:%d, wal idx append new entry %" PRId64 " %" PRId64, pWal->cfg.vgId, idxEntry.ver, idxEntry.offset); if (taosWriteFile(pIdxFile, &idxEntry, sizeof(SWalIdxEntry)) < 0) { wError("vgId:%d, failed to append file since %s. file:%s", pWal->cfg.vgId, terrstr(), fnameStr); goto _err; } + count++; } if (taosFsyncFile(pIdxFile) < 0) { @@ -590,6 +622,11 @@ int walCheckAndRepairIdxFile(SWal* pWal, int32_t fileIdx) { goto _err; } + if (count > 0) { + wInfo("vgId:%d, rebuilt %" PRId64 " wal idx entries until lastVer: %" PRId64, pWal->cfg.vgId, count, + pFileInfo->lastVer); + } + (void)taosCloseFile(&pLogFile); (void)taosCloseFile(&pIdxFile); return 0; From d6117935371f1a0cc0d29a42ed5e772b3e65d576 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Wed, 18 Jan 2023 10:07:05 +0800 Subject: [PATCH 008/124] enh: remove unused old routines in sync --- source/libs/sync/src/syncAppendEntries.c | 292 -------------- source/libs/sync/src/syncAppendEntriesReply.c | 60 --- source/libs/sync/src/syncCommit.c | 220 +--------- source/libs/sync/src/syncElection.c | 18 +- source/libs/sync/src/syncMain.c | 379 ++---------------- source/libs/sync/src/syncPipeline.c | 5 +- source/libs/sync/src/syncReplication.c | 138 ------- 7 files changed, 47 insertions(+), 1065 deletions(-) diff --git a/source/libs/sync/src/syncAppendEntries.c b/source/libs/sync/src/syncAppendEntries.c index e77a8d4be3..835e5d248e 100644 --- a/source/libs/sync/src/syncAppendEntries.c +++ b/source/libs/sync/src/syncAppendEntries.c @@ -89,45 +89,6 @@ // /\ UNCHANGED <> // -int32_t syncNodeFollowerCommit(SSyncNode* ths, SyncIndex newCommitIndex) { - ASSERT(false && "deprecated"); - if (ths->state != TAOS_SYNC_STATE_FOLLOWER) { - sNTrace(ths, "can not do follower commit"); - return -1; - } - - // maybe update commit index, leader notice me - if (newCommitIndex > ths->commitIndex) { - // has commit entry in local - if (newCommitIndex <= ths->pLogStore->syncLogLastIndex(ths->pLogStore)) { - // advance commit index to sanpshot first - SSnapshot snapshot; - ths->pFsm->FpGetSnapshotInfo(ths->pFsm, &snapshot); - if (snapshot.lastApplyIndex >= 0 && snapshot.lastApplyIndex > ths->commitIndex) { - SyncIndex commitBegin = ths->commitIndex; - SyncIndex commitEnd = snapshot.lastApplyIndex; - ths->commitIndex = snapshot.lastApplyIndex; - sNTrace(ths, "commit by snapshot from index:%" PRId64 " to index:%" PRId64, commitBegin, commitEnd); - } - - SyncIndex beginIndex = ths->commitIndex + 1; - SyncIndex endIndex = newCommitIndex; - - // update commit index - ths->commitIndex = newCommitIndex; - - // call back Wal - int32_t code = ths->pLogStore->syncLogUpdateCommitIndex(ths->pLogStore, ths->commitIndex); - ASSERT(code == 0); - - code = syncNodeDoCommit(ths, beginIndex, endIndex, ths->state); - ASSERT(code == 0); - } - } - - return 0; -} - SSyncRaftEntry* syncBuildRaftEntryFromAppendEntries(const SyncAppendEntries* pMsg) { SSyncRaftEntry* pEntry = taosMemoryMalloc(pMsg->dataLen); if (pEntry == NULL) { @@ -232,256 +193,3 @@ _IGNORE: rpcFreeCont(rpcRsp.pCont); return 0; } - -int32_t syncNodeOnAppendEntriesOld(SSyncNode* ths, const SRpcMsg* pRpcMsg) { - SyncAppendEntries* pMsg = pRpcMsg->pCont; - SRpcMsg rpcRsp = {0}; - - // if already drop replica, do not process - if (!syncNodeInRaftGroup(ths, &(pMsg->srcId))) { - syncLogRecvAppendEntries(ths, pMsg, "not in my config"); - goto _IGNORE; - } - - // prepare response msg - int32_t code = syncBuildAppendEntriesReply(&rpcRsp, ths->vgId); - if (code != 0) { - syncLogRecvAppendEntries(ths, pMsg, "build rsp error"); - goto _IGNORE; - } - - SyncAppendEntriesReply* pReply = rpcRsp.pCont; - pReply->srcId = ths->myRaftId; - pReply->destId = pMsg->srcId; - pReply->term = ths->raftStore.currentTerm; - pReply->success = false; - // pReply->matchIndex = ths->pLogStore->syncLogLastIndex(ths->pLogStore); - pReply->matchIndex = SYNC_INDEX_INVALID; - pReply->lastSendIndex = pMsg->prevLogIndex + 1; - pReply->startTime = ths->startTime; - - if (pMsg->term < ths->raftStore.currentTerm) { - syncLogRecvAppendEntries(ths, pMsg, "reject, small term"); - goto _SEND_RESPONSE; - } - - if (pMsg->term > ths->raftStore.currentTerm) { - pReply->term = pMsg->term; - } - - syncNodeStepDown(ths, pMsg->term); - syncNodeResetElectTimer(ths); - - SyncIndex startIndex = ths->pLogStore->syncLogBeginIndex(ths->pLogStore); - SyncIndex lastIndex = ths->pLogStore->syncLogLastIndex(ths->pLogStore); - - if (pMsg->prevLogIndex > lastIndex) { - syncLogRecvAppendEntries(ths, pMsg, "reject, index not match"); - goto _SEND_RESPONSE; - } - - if (pMsg->prevLogIndex >= startIndex) { - SyncTerm myPreLogTerm = syncNodeGetPreTerm(ths, pMsg->prevLogIndex + 1); - // ASSERT(myPreLogTerm != SYNC_TERM_INVALID); - if (myPreLogTerm == SYNC_TERM_INVALID) { - syncLogRecvAppendEntries(ths, pMsg, "reject, pre-term invalid"); - goto _SEND_RESPONSE; - } - - if (myPreLogTerm != pMsg->prevLogTerm) { - syncLogRecvAppendEntries(ths, pMsg, "reject, pre-term not match"); - goto _SEND_RESPONSE; - } - } - - // accept - pReply->success = true; - bool hasAppendEntries = pMsg->dataLen > 0; - if (hasAppendEntries) { - SSyncRaftEntry* pAppendEntry = syncEntryBuildFromAppendEntries(pMsg); - ASSERT(pAppendEntry != NULL); - - SyncIndex appendIndex = pMsg->prevLogIndex + 1; - - LRUHandle* hLocal = NULL; - LRUHandle* hAppend = NULL; - - int32_t code = 0; - SSyncRaftEntry* pLocalEntry = NULL; - SLRUCache* pCache = ths->pLogStore->pCache; - hLocal = taosLRUCacheLookup(pCache, &appendIndex, sizeof(appendIndex)); - if (hLocal) { - pLocalEntry = (SSyncRaftEntry*)taosLRUCacheValue(pCache, hLocal); - code = 0; - - ths->pLogStore->cacheHit++; - sNTrace(ths, "hit cache index:%" PRId64 ", bytes:%u, %p", appendIndex, pLocalEntry->bytes, pLocalEntry); - - } else { - ths->pLogStore->cacheMiss++; - sNTrace(ths, "miss cache index:%" PRId64, appendIndex); - - code = ths->pLogStore->syncLogGetEntry(ths->pLogStore, appendIndex, &pLocalEntry); - } - - if (code == 0) { - // get local entry success - - if (pLocalEntry->term == pAppendEntry->term) { - // do nothing - sNTrace(ths, "log match, do nothing, index:%" PRId64, appendIndex); - - } else { - // truncate - code = ths->pLogStore->syncLogTruncate(ths->pLogStore, appendIndex); - if (code != 0) { - char logBuf[128]; - snprintf(logBuf, sizeof(logBuf), "ignore, truncate error, append-index:%" PRId64, appendIndex); - syncLogRecvAppendEntries(ths, pMsg, logBuf); - - if (hLocal) { - taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false); - } else { - syncEntryDestroy(pLocalEntry); - } - - if (hAppend) { - taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false); - } else { - syncEntryDestroy(pAppendEntry); - } - - goto _IGNORE; - } - - ASSERT(pAppendEntry->index == appendIndex); - - // append - code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry); - if (code != 0) { - char logBuf[128]; - snprintf(logBuf, sizeof(logBuf), "ignore, append error, append-index:%" PRId64, appendIndex); - syncLogRecvAppendEntries(ths, pMsg, logBuf); - - if (hLocal) { - taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false); - } else { - syncEntryDestroy(pLocalEntry); - } - - if (hAppend) { - taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false); - } else { - syncEntryDestroy(pAppendEntry); - } - - goto _IGNORE; - } - - syncCacheEntry(ths->pLogStore, pAppendEntry, &hAppend); - } - - } else { - if (terrno == TSDB_CODE_WAL_LOG_NOT_EXIST) { - // log not exist - - // truncate - code = ths->pLogStore->syncLogTruncate(ths->pLogStore, appendIndex); - if (code != 0) { - char logBuf[128]; - snprintf(logBuf, sizeof(logBuf), "ignore, log not exist, truncate error, append-index:%" PRId64, appendIndex); - syncLogRecvAppendEntries(ths, pMsg, logBuf); - - syncEntryDestroy(pLocalEntry); - syncEntryDestroy(pAppendEntry); - goto _IGNORE; - } - - // append - code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry); - if (code != 0) { - char logBuf[128]; - snprintf(logBuf, sizeof(logBuf), "ignore, log not exist, append error, append-index:%" PRId64, appendIndex); - syncLogRecvAppendEntries(ths, pMsg, logBuf); - - if (hLocal) { - taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false); - } else { - syncEntryDestroy(pLocalEntry); - } - - if (hAppend) { - taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false); - } else { - syncEntryDestroy(pAppendEntry); - } - - goto _IGNORE; - } - - syncCacheEntry(ths->pLogStore, pAppendEntry, &hAppend); - - } else { - // get local entry success - char logBuf[128]; - snprintf(logBuf, sizeof(logBuf), "ignore, get local entry error, append-index:%" PRId64 " err:%d", appendIndex, - terrno); - syncLogRecvAppendEntries(ths, pMsg, logBuf); - - if (hLocal) { - taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false); - } else { - syncEntryDestroy(pLocalEntry); - } - - if (hAppend) { - taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false); - } else { - syncEntryDestroy(pAppendEntry); - } - - goto _IGNORE; - } - } - - // update match index - pReply->matchIndex = pAppendEntry->index; - - if (hLocal) { - taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false); - } else { - syncEntryDestroy(pLocalEntry); - } - - if (hAppend) { - taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false); - } else { - syncEntryDestroy(pAppendEntry); - } - - } else { - // no append entries, do nothing - // maybe has extra entries, no harm - - // update match index - pReply->matchIndex = pMsg->prevLogIndex; - } - - // maybe update commit index, leader notice me - syncNodeFollowerCommit(ths, pMsg->commitIndex); - - syncLogRecvAppendEntries(ths, pMsg, "accept"); - goto _SEND_RESPONSE; - -_IGNORE: - rpcFreeCont(rpcRsp.pCont); - return 0; - -_SEND_RESPONSE: - // msg event log - syncLogSendAppendEntriesReply(ths, pReply, ""); - - // send response - syncNodeSendMsgById(&pReply->destId, ths, &rpcRsp); - return 0; -} diff --git a/source/libs/sync/src/syncAppendEntriesReply.c b/source/libs/sync/src/syncAppendEntriesReply.c index 8157a5a14f..44a29da3ea 100644 --- a/source/libs/sync/src/syncAppendEntriesReply.c +++ b/source/libs/sync/src/syncAppendEntriesReply.c @@ -89,63 +89,3 @@ int32_t syncNodeOnAppendEntriesReply(SSyncNode* ths, const SRpcMsg* pRpcMsg) { } return 0; } - -int32_t syncNodeOnAppendEntriesReplyOld(SSyncNode* ths, SyncAppendEntriesReply* pMsg) { - int32_t ret = 0; - - // if already drop replica, do not process - if (!syncNodeInRaftGroup(ths, &(pMsg->srcId))) { - syncLogRecvAppendEntriesReply(ths, pMsg, "not in my config"); - return 0; - } - - // drop stale response - if (pMsg->term < ths->raftStore.currentTerm) { - syncLogRecvAppendEntriesReply(ths, pMsg, "drop stale response"); - return 0; - } - - if (ths->state == TAOS_SYNC_STATE_LEADER) { - if (pMsg->term > ths->raftStore.currentTerm) { - syncLogRecvAppendEntriesReply(ths, pMsg, "error term"); - syncNodeStepDown(ths, pMsg->term); - return -1; - } - - ASSERT(pMsg->term == ths->raftStore.currentTerm); - - if (pMsg->success) { - SyncIndex oldMatchIndex = syncIndexMgrGetIndex(ths->pMatchIndex, &(pMsg->srcId)); - if (pMsg->matchIndex > oldMatchIndex) { - syncIndexMgrSetIndex(ths->pMatchIndex, &(pMsg->srcId), pMsg->matchIndex); - syncMaybeAdvanceCommitIndex(ths); - - // maybe update minMatchIndex - ths->minMatchIndex = syncMinMatchIndex(ths); - } - syncIndexMgrSetIndex(ths->pNextIndex, &(pMsg->srcId), pMsg->matchIndex + 1); - - } else { - SyncIndex nextIndex = syncIndexMgrGetIndex(ths->pNextIndex, &(pMsg->srcId)); - if (nextIndex > SYNC_INDEX_BEGIN) { - --nextIndex; - } - syncIndexMgrSetIndex(ths->pNextIndex, &(pMsg->srcId), nextIndex); - } - - // send next append entries - SPeerState* pState = syncNodeGetPeerState(ths, &(pMsg->srcId)); - ASSERT(pState != NULL); - - if (pMsg->lastSendIndex == pState->lastSendIndex) { - int64_t timeNow = taosGetTimestampMs(); - int64_t elapsed = timeNow - pState->lastSendTime; - sNTrace(ths, "sync-append-entries rtt elapsed:%" PRId64 ", index:%" PRId64, elapsed, pState->lastSendIndex); - - syncNodeReplicateOne(ths, &(pMsg->srcId), true); - } - } - - syncLogRecvAppendEntriesReply(ths, pMsg, "process"); - return 0; -} diff --git a/source/libs/sync/src/syncCommit.c b/source/libs/sync/src/syncCommit.c index 286cf4daf5..67ed1e0701 100644 --- a/source/libs/sync/src/syncCommit.c +++ b/source/libs/sync/src/syncCommit.c @@ -43,148 +43,6 @@ // IN commitIndex' = [commitIndex EXCEPT ![i] = newCommitIndex] // /\ UNCHANGED <> // -void syncOneReplicaAdvance(SSyncNode* pSyncNode) { - ASSERT(false && "deprecated"); - if (pSyncNode == NULL) { - sError("pSyncNode is NULL"); - return; - } - - if (pSyncNode->state != TAOS_SYNC_STATE_LEADER) { - sNError(pSyncNode, "not leader, can not advance commit index"); - return; - } - - if (pSyncNode->replicaNum != 1) { - sNError(pSyncNode, "not one replica, can not advance commit index"); - return; - } - - // advance commit index to snapshot first - SSnapshot snapshot; - pSyncNode->pFsm->FpGetSnapshotInfo(pSyncNode->pFsm, &snapshot); - if (snapshot.lastApplyIndex > 0 && snapshot.lastApplyIndex > pSyncNode->commitIndex) { - SyncIndex commitBegin = pSyncNode->commitIndex; - SyncIndex commitEnd = snapshot.lastApplyIndex; - pSyncNode->commitIndex = snapshot.lastApplyIndex; - sNTrace(pSyncNode, "commit by snapshot from index:%" PRId64 " to index:%" PRId64, commitBegin, commitEnd); - } - - // advance commit index as large as possible - SyncIndex lastIndex = syncNodeGetLastIndex(pSyncNode); - if (lastIndex > pSyncNode->commitIndex) { - sNTrace(pSyncNode, "commit by wal from index:%" PRId64 " to index:%" PRId64, pSyncNode->commitIndex + 1, lastIndex); - pSyncNode->commitIndex = lastIndex; - } - - // call back Wal - SyncIndex walCommitVer = logStoreWalCommitVer(pSyncNode->pLogStore); - if (pSyncNode->commitIndex > walCommitVer) { - pSyncNode->pLogStore->syncLogUpdateCommitIndex(pSyncNode->pLogStore, pSyncNode->commitIndex); - } -} - -void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) { - ASSERTS(false, "deprecated"); - if (pSyncNode == NULL) { - sError("pSyncNode is NULL"); - return; - } - - if (pSyncNode->state != TAOS_SYNC_STATE_LEADER) { - sNError(pSyncNode, "not leader, can not advance commit index"); - return; - } - - // advance commit index to sanpshot first - SSnapshot snapshot; - pSyncNode->pFsm->FpGetSnapshotInfo(pSyncNode->pFsm, &snapshot); - if (snapshot.lastApplyIndex > 0 && snapshot.lastApplyIndex > pSyncNode->commitIndex) { - SyncIndex commitBegin = pSyncNode->commitIndex; - SyncIndex commitEnd = snapshot.lastApplyIndex; - pSyncNode->commitIndex = snapshot.lastApplyIndex; - sNTrace(pSyncNode, "commit by snapshot from index:%" PRId64 " to index:%" PRId64, commitBegin, commitEnd); - } - - // update commit index - SyncIndex newCommitIndex = pSyncNode->commitIndex; - for (SyncIndex index = syncNodeGetLastIndex(pSyncNode); index > pSyncNode->commitIndex; --index) { - bool agree = syncAgree(pSyncNode, index); - - if (agree) { - // term - SSyncRaftEntry* pEntry = NULL; - SLRUCache* pCache = pSyncNode->pLogStore->pCache; - LRUHandle* h = taosLRUCacheLookup(pCache, &index, sizeof(index)); - if (h) { - pEntry = (SSyncRaftEntry*)taosLRUCacheValue(pCache, h); - - pSyncNode->pLogStore->cacheHit++; - sNTrace(pSyncNode, "hit cache index:%" PRId64 ", bytes:%u, %p", index, pEntry->bytes, pEntry); - - } else { - pSyncNode->pLogStore->cacheMiss++; - sNTrace(pSyncNode, "miss cache index:%" PRId64, index); - - int32_t code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, index, &pEntry); - if (code != 0) { - sNError(pSyncNode, "advance commit index error, read wal index:%" PRId64, index); - return; - } - } - // cannot commit, even if quorum agree. need check term! - if (pEntry->term <= pSyncNode->raftStore.currentTerm) { - // update commit index - newCommitIndex = index; - - if (h) { - taosLRUCacheRelease(pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - - break; - } else { - sNTrace(pSyncNode, "can not commit due to term not equal, index:%" PRId64 ", term:%" PRIu64, pEntry->index, - pEntry->term); - } - - if (h) { - taosLRUCacheRelease(pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - } - } - - // advance commit index as large as possible - SyncIndex walCommitVer = logStoreWalCommitVer(pSyncNode->pLogStore); - if (walCommitVer > newCommitIndex) { - newCommitIndex = walCommitVer; - } - - // maybe execute fsm - if (newCommitIndex > pSyncNode->commitIndex) { - SyncIndex beginIndex = pSyncNode->commitIndex + 1; - SyncIndex endIndex = newCommitIndex; - - // update commit index - pSyncNode->commitIndex = newCommitIndex; - - // call back Wal - pSyncNode->pLogStore->syncLogUpdateCommitIndex(pSyncNode->pLogStore, pSyncNode->commitIndex); - - // execute fsm - if (pSyncNode != NULL && pSyncNode->pFsm != NULL) { - int32_t code = syncNodeDoCommit(pSyncNode, beginIndex, endIndex, pSyncNode->state); - if (code != 0) { - sNError(pSyncNode, "advance commit index error, do commit begin:%" PRId64 ", end:%" PRId64, beginIndex, - endIndex); - return; - } - } - } -} bool syncAgreeIndex(SSyncNode* pSyncNode, SRaftId* pRaftId, SyncIndex index) { // I am leader, I agree @@ -210,83 +68,7 @@ static inline int64_t syncNodeAbs64(int64_t a, int64_t b) { return c; } -int32_t syncNodeDynamicQuorum(const SSyncNode* pSyncNode) { - return pSyncNode->quorum; - -#if 0 - int32_t quorum = 1; // self - - int64_t timeNow = taosGetTimestampMs(); - for (int i = 0; i < pSyncNode->peersNum; ++i) { - int64_t peerStartTime = syncIndexMgrGetStartTime(pSyncNode->pNextIndex, &(pSyncNode->peersId)[i]); - int64_t peerRecvTime = syncIndexMgrGetRecvTime(pSyncNode->pNextIndex, &(pSyncNode->peersId)[i]); - SyncIndex peerMatchIndex = syncIndexMgrGetIndex(pSyncNode->pMatchIndex, &(pSyncNode->peersId)[i]); - - int64_t recvTimeDiff = TABS(peerRecvTime - timeNow); - int64_t startTimeDiff = TABS(peerStartTime - pSyncNode->startTime); - int64_t logDiff = TABS(peerMatchIndex - syncNodeGetLastIndex(pSyncNode)); - - /* - int64_t recvTimeDiff = syncNodeAbs64(peerRecvTime, timeNow); - int64_t startTimeDiff = syncNodeAbs64(peerStartTime, pSyncNode->startTime); - int64_t logDiff = syncNodeAbs64(peerMatchIndex, syncNodeGetLastIndex(pSyncNode)); - */ - - int32_t addQuorum = 0; - - if (recvTimeDiff < SYNC_MAX_RECV_TIME_RANGE_MS) { - if (startTimeDiff < SYNC_MAX_START_TIME_RANGE_MS) { - addQuorum = 1; - } else { - if (logDiff < SYNC_ADD_QUORUM_COUNT) { - addQuorum = 1; - } else { - addQuorum = 0; - } - } - } else { - addQuorum = 0; - } - - /* - if (recvTimeDiff < SYNC_MAX_RECV_TIME_RANGE_MS) { - addQuorum = 1; - } else { - addQuorum = 0; - } - - if (startTimeDiff > SYNC_MAX_START_TIME_RANGE_MS) { - addQuorum = 0; - } - */ - - quorum += addQuorum; - } - - ASSERT(quorum <= pSyncNode->replicaNum); - - if (quorum < pSyncNode->quorum) { - quorum = pSyncNode->quorum; - } - - return quorum; -#endif -} - -/* -bool syncAgree(SSyncNode* pSyncNode, SyncIndex index) { - int agreeCount = 0; - for (int i = 0; i < pSyncNode->replicaNum; ++i) { - if (syncAgreeIndex(pSyncNode, &(pSyncNode->replicasId[i]), index)) { - ++agreeCount; - } - if (agreeCount >= syncNodeDynamicQuorum(pSyncNode)) { - return true; - } - } - return false; -} -*/ +int32_t syncNodeDynamicQuorum(const SSyncNode* pSyncNode) { return pSyncNode->quorum; } bool syncNodeAgreedUpon(SSyncNode* pNode, SyncIndex index) { int count = 0; diff --git a/source/libs/sync/src/syncElection.c b/source/libs/sync/src/syncElection.c index cd3ffc33e3..682ace83ec 100644 --- a/source/libs/sync/src/syncElection.c +++ b/source/libs/sync/src/syncElection.c @@ -43,7 +43,10 @@ static int32_t syncNodeRequestVotePeers(SSyncNode* pNode) { for (int i = 0; i < pNode->peersNum; ++i) { SRpcMsg rpcMsg = {0}; ret = syncBuildRequestVote(&rpcMsg, pNode->vgId); - ASSERT(ret == 0); + if (ret < 0) { + sError("vgId:%d, failed to build request-vote msg since %s", pNode->vgId, terrstr()); + continue; + } SyncRequestVote* pMsg = rpcMsg.pCont; pMsg->srcId = pNode->myRaftId; @@ -51,13 +54,18 @@ static int32_t syncNodeRequestVotePeers(SSyncNode* pNode) { pMsg->term = pNode->raftStore.currentTerm; ret = syncNodeGetLastIndexTerm(pNode, &pMsg->lastLogIndex, &pMsg->lastLogTerm); - ASSERT(ret == 0); + if (ret < 0) { + sError("vgId:%d, failed to get index and term of last log since %s", pNode->vgId, terrstr()); + continue; + } ret = syncNodeSendMsgById(&pNode->peersId[i], pNode, &rpcMsg); - ASSERT(ret == 0); + if (ret < 0) { + sError("vgId:%d, failed to send msg to peerId:%" PRId64, pNode->vgId, pNode->peersId[i].addr); + continue; + } } - - return ret; + return 0; } int32_t syncNodeElect(SSyncNode* pSyncNode) { diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index 02f9795cad..07e781f420 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -586,78 +586,6 @@ SSyncState syncGetState(int64_t rid) { return state; } -#if 0 -int32_t syncGetSnapshotByIndex(int64_t rid, SyncIndex index, SSnapshot* pSnapshot) { - if (index < SYNC_INDEX_BEGIN) { - return -1; - } - - SSyncNode* pSyncNode = syncNodeAcquire(rid); - if (pSyncNode == NULL) { - return -1; - } - ASSERT(rid == pSyncNode->rid); - - SSyncRaftEntry* pEntry = NULL; - int32_t code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, index, &pEntry); - if (code != 0) { - if (pEntry != NULL) { - syncEntryDestroy(pEntry); - } - syncNodeRelease(pSyncNode); - return -1; - } - ASSERT(pEntry != NULL); - - pSnapshot->data = NULL; - pSnapshot->lastApplyIndex = index; - pSnapshot->lastApplyTerm = pEntry->term; - pSnapshot->lastConfigIndex = syncNodeGetSnapshotConfigIndex(pSyncNode, index); - - syncEntryDestroy(pEntry); - syncNodeRelease(pSyncNode); - return 0; -} - -int32_t syncGetSnapshotMeta(int64_t rid, struct SSnapshotMeta* sMeta) { - SSyncNode* pSyncNode = syncNodeAcquire(rid); - if (pSyncNode == NULL) { - return -1; - } - ASSERT(rid == pSyncNode->rid); - sMeta->lastConfigIndex = pSyncNode->raftCfg.lastConfigIndex; - - sTrace("vgId:%d, get snapshot meta, lastConfigIndex:%" PRId64, pSyncNode->vgId, pSyncNode->raftCfg.lastConfigIndex); - - syncNodeRelease(pSyncNode); - return 0; -} - -int32_t syncGetSnapshotMetaByIndex(int64_t rid, SyncIndex snapshotIndex, struct SSnapshotMeta* sMeta) { - SSyncNode* pSyncNode = syncNodeAcquire(rid); - if (pSyncNode == NULL) { - return -1; - } - ASSERT(rid == pSyncNode->rid); - - ASSERT(pSyncNode->raftCfg.configIndexCount >= 1); - SyncIndex lastIndex = (pSyncNode->raftCfg.configIndexArr)[0]; - - for (int32_t i = 0; i < pSyncNode->raftCfg.configIndexCount; ++i) { - if ((pSyncNode->raftCfg.configIndexArr)[i] > lastIndex && - (pSyncNode->raftCfg.configIndexArr)[i] <= snapshotIndex) { - lastIndex = (pSyncNode->raftCfg.configIndexArr)[i]; - } - } - sMeta->lastConfigIndex = lastIndex; - sTrace("vgId:%d, get snapshot meta by index:%" PRId64 " lcindex:%" PRId64, pSyncNode->vgId, snapshotIndex, - sMeta->lastConfigIndex); - - syncNodeRelease(pSyncNode); - return 0; -} -#endif - SyncIndex syncNodeGetSnapshotConfigIndex(SSyncNode* pSyncNode, SyncIndex snapshotLastApplyIndex) { ASSERT(pSyncNode->raftCfg.configIndexCount >= 1); SyncIndex lastIndex = (pSyncNode->raftCfg.configIndexArr)[0]; @@ -1031,9 +959,12 @@ SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) { pSyncNode->commitIndex = commitIndex; sInfo("vgId:%d, sync node commitIndex initialized as %" PRId64, pSyncNode->vgId, pSyncNode->commitIndex); + // restore log store on need if (syncNodeLogStoreRestoreOnNeed(pSyncNode) < 0) { + sError("vgId:%d, failed to restore log store since %s.", pSyncNode->vgId, terrstr()); goto _error; } + // timer ms init pSyncNode->pingBaseLine = PING_TIMER_MS; pSyncNode->electBaseLine = tsElectInterval; @@ -1096,10 +1027,16 @@ SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) { pSyncNode->changing = false; // replication mgr - syncNodeLogReplMgrInit(pSyncNode); + if (syncNodeLogReplMgrInit(pSyncNode) < 0) { + sError("vgId:%d, failed to init repl mgr since %s.", pSyncNode->vgId, terrstr()); + goto _error; + } // peer state - syncNodePeerStateInit(pSyncNode); + if (syncNodePeerStateInit(pSyncNode) < 0) { + sError("vgId:%d, failed to init peer stat since %s.", pSyncNode->vgId, terrstr()); + goto _error; + } // // min match index @@ -1194,27 +1131,10 @@ int32_t syncNodeStart(SSyncNode* pSyncNode) { int32_t ret = 0; ret = syncNodeStartPingTimer(pSyncNode); - ASSERT(ret == 0); - return ret; -} - -void syncNodeStartOld(SSyncNode* pSyncNode) { - // start raft - if (pSyncNode->replicaNum == 1) { - raftStoreNextTerm(pSyncNode); - syncNodeBecomeLeader(pSyncNode, "one replica start"); - - // Raft 3.6.2 Committing entries from previous terms - syncNodeAppendNoop(pSyncNode); - syncMaybeAdvanceCommitIndex(pSyncNode); - - } else { - syncNodeBecomeFollower(pSyncNode, "first start"); + if (ret != 0) { + sError("vgId:%d, failed to start ping timer since %s", pSyncNode->vgId, terrstr()); } - - int32_t ret = 0; - ret = syncNodeStartPingTimer(pSyncNode); - ASSERT(ret == 0); + return ret; } int32_t syncNodeStartStandBy(SSyncNode* pSyncNode) { @@ -1225,11 +1145,16 @@ int32_t syncNodeStartStandBy(SSyncNode* pSyncNode) { // reset elect timer, long enough int32_t electMS = TIMER_MAX_MS; int32_t ret = syncNodeRestartElectTimer(pSyncNode, electMS); - ASSERT(ret == 0); + if (ret < 0) { + sError("vgId:%d, failed to restart elect timer since %s", pSyncNode->vgId, terrstr()); + return -1; + } - ret = 0; ret = syncNodeStartPingTimer(pSyncNode); - ASSERT(ret == 0); + if (ret < 0) { + sError("vgId:%d, failed to start ping timer since %s", pSyncNode->vgId, terrstr()); + return -1; + } return ret; } @@ -1818,12 +1743,6 @@ void syncNodeBecomeLeader(SSyncNode* pSyncNode, const char* debugStr) { pSyncNode->leaderCache = pSyncNode->myRaftId; for (int32_t i = 0; i < pSyncNode->pNextIndex->replicaNum; ++i) { - // maybe overwrite myself, no harm - // just do it! - - // pSyncNode->pNextIndex->index[i] = pSyncNode->pLogStore->getLastIndex(pSyncNode->pLogStore) + 1; - - // maybe wal is deleted SyncIndex lastIndex; SyncTerm lastTerm; int32_t code = syncNodeGetLastIndexTerm(pSyncNode, &lastIndex, &lastTerm); @@ -1885,7 +1804,11 @@ void syncNodeBecomeLeader(SSyncNode* pSyncNode, const char* debugStr) { void syncNodeCandidate2Leader(SSyncNode* pSyncNode) { ASSERT(pSyncNode->state == TAOS_SYNC_STATE_CANDIDATE); - ASSERT(voteGrantedMajority(pSyncNode->pVotesGranted)); + bool granted = voteGrantedMajority(pSyncNode->pVotesGranted); + if (!granted) { + sError("vgId:%d, not granted by majority.", pSyncNode->vgId); + return; + } syncNodeBecomeLeader(pSyncNode, "candidate to leader"); sNTrace(pSyncNode, "state change syncNodeCandidate2Leader"); @@ -1901,20 +1824,6 @@ void syncNodeCandidate2Leader(SSyncNode* pSyncNode) { pSyncNode->vgId, pSyncNode->raftStore.currentTerm, pSyncNode->commitIndex, lastIndex); } -void syncNodeCandidate2LeaderOld(SSyncNode* pSyncNode) { - ASSERT(pSyncNode->state == TAOS_SYNC_STATE_CANDIDATE); - ASSERT(voteGrantedMajority(pSyncNode->pVotesGranted)); - syncNodeBecomeLeader(pSyncNode, "candidate to leader"); - - // Raft 3.6.2 Committing entries from previous terms - syncNodeAppendNoop(pSyncNode); - syncMaybeAdvanceCommitIndex(pSyncNode); - - if (pSyncNode->replicaNum > 1) { - syncNodeReplicate(pSyncNode); - } -} - bool syncNodeIsMnode(SSyncNode* pSyncNode) { return (pSyncNode->vgId == 1); } int32_t syncNodePeerStateInit(SSyncNode* pSyncNode) { @@ -1960,7 +1869,8 @@ void syncNodeCandidate2Follower(SSyncNode* pSyncNode) { // need assert void syncNodeVoteForTerm(SSyncNode* pSyncNode, SyncTerm term, SRaftId* pRaftId) { ASSERT(term == pSyncNode->raftStore.currentTerm); - ASSERT(!raftStoreHasVoted(pSyncNode)); + bool voted = raftStoreHasVoted(pSyncNode); + ASSERT(!voted); raftStoreVote(pSyncNode, pRaftId); } @@ -2638,24 +2548,6 @@ int32_t syncNodeOnLocalCmd(SSyncNode* ths, const SRpcMsg* pRpcMsg) { return 0; } -int32_t syncNodeOnLocalCmdOld(SSyncNode* ths, const SRpcMsg* pRpcMsg) { - ASSERT(false && "deprecated"); - SyncLocalCmd* pMsg = pRpcMsg->pCont; - syncLogRecvLocalCmd(ths, pMsg, ""); - - if (pMsg->cmd == SYNC_LOCAL_CMD_STEP_DOWN) { - syncNodeStepDown(ths, pMsg->currentTerm); - - } else if (pMsg->cmd == SYNC_LOCAL_CMD_FOLLOWER_CMT) { - syncNodeFollowerCommit(ths, pMsg->commitIndex); - - } else { - sError("error local cmd"); - } - - return 0; -} - // TLA+ Spec // ClientRequest(i, v) == // /\ state[i] = Leader @@ -2700,96 +2592,6 @@ int32_t syncNodeOnClientRequest(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIn } } -int32_t syncNodeOnClientRequestOld(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIndex) { - sNTrace(ths, "on client request"); - - int32_t ret = 0; - int32_t code = 0; - - SyncIndex index = ths->pLogStore->syncLogWriteIndex(ths->pLogStore); - SyncTerm term = ths->raftStore.currentTerm; - SSyncRaftEntry* pEntry; - - if (pMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) { - pEntry = syncEntryBuildFromClientRequest(pMsg->pCont, term, index); - } else { - pEntry = syncEntryBuildFromRpcMsg(pMsg, term, index); - } - - LRUHandle* h = NULL; - - if (ths->state == TAOS_SYNC_STATE_LEADER) { - // append entry - code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pEntry); - if (code != 0) { - if (ths->replicaNum == 1) { - if (h) { - taosLRUCacheRelease(ths->pLogStore->pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - - return -1; - - } else { - // del resp mgr, call FpCommitCb - SFsmCbMeta cbMeta = { - .index = pEntry->index, - .lastConfigIndex = SYNC_INDEX_INVALID, - .isWeak = pEntry->isWeak, - .code = -1, - .state = ths->state, - .seqNum = pEntry->seqNum, - .term = pEntry->term, - .currentTerm = ths->raftStore.currentTerm, - .flag = 0, - }; - ths->pFsm->FpCommitCb(ths->pFsm, pMsg, &cbMeta); - - if (h) { - taosLRUCacheRelease(ths->pLogStore->pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - - return -1; - } - } - - syncCacheEntry(ths->pLogStore, pEntry, &h); - - // if mulit replica, start replicate right now - if (ths->replicaNum > 1) { - syncNodeReplicate(ths); - } - - // if only myself, maybe commit right now - if (ths->replicaNum == 1) { - if (syncNodeIsMnode(ths)) { - syncMaybeAdvanceCommitIndex(ths); - } else { - syncOneReplicaAdvance(ths); - } - } - } - - if (pRetIndex != NULL) { - if (ret == 0 && pEntry != NULL) { - *pRetIndex = pEntry->index; - } else { - *pRetIndex = SYNC_INDEX_INVALID; - } - } - - if (h) { - taosLRUCacheRelease(ths->pLogStore->pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - - return ret; -} - const char* syncStr(ESyncState state) { switch (state) { case TAOS_SYNC_STATE_FOLLOWER: @@ -2894,129 +2696,6 @@ bool syncNodeIsOptimizedOneReplica(SSyncNode* ths, SRpcMsg* pMsg) { return (ths->replicaNum == 1 && syncUtilUserCommit(pMsg->msgType) && ths->vgId != 1); } -int32_t syncNodeDoCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, uint64_t flag) { - ASSERT(false); - if (beginIndex > endIndex) { - return 0; - } - - if (ths == NULL) { - return -1; - } - - if (ths->pFsm != NULL && ths->pFsm->FpGetSnapshotInfo != NULL) { - // advance commit index to sanpshot first - SSnapshot snapshot = {0}; - ths->pFsm->FpGetSnapshotInfo(ths->pFsm, &snapshot); - if (snapshot.lastApplyIndex >= 0 && snapshot.lastApplyIndex >= beginIndex) { - sNTrace(ths, "commit by snapshot from index:%" PRId64 " to index:%" PRId64, beginIndex, snapshot.lastApplyIndex); - - // update begin index - beginIndex = snapshot.lastApplyIndex + 1; - } - } - - int32_t code = 0; - ESyncState state = flag; - - sNTrace(ths, "commit by wal from index:%" PRId64 " to index:%" PRId64, beginIndex, endIndex); - - // execute fsm - if (ths->pFsm != NULL) { - for (SyncIndex i = beginIndex; i <= endIndex; ++i) { - if (i != SYNC_INDEX_INVALID) { - SSyncRaftEntry* pEntry; - SLRUCache* pCache = ths->pLogStore->pCache; - LRUHandle* h = taosLRUCacheLookup(pCache, &i, sizeof(i)); - if (h) { - pEntry = (SSyncRaftEntry*)taosLRUCacheValue(pCache, h); - - ths->pLogStore->cacheHit++; - sNTrace(ths, "hit cache index:%" PRId64 ", bytes:%u, %p", i, pEntry->bytes, pEntry); - - } else { - ths->pLogStore->cacheMiss++; - sNTrace(ths, "miss cache index:%" PRId64, i); - - code = ths->pLogStore->syncLogGetEntry(ths->pLogStore, i, &pEntry); - // ASSERT(code == 0); - // ASSERT(pEntry != NULL); - if (code != 0 || pEntry == NULL) { - sNError(ths, "get log entry error"); - sFatal("vgId:%d, get log entry %" PRId64 " error when commit since %s", ths->vgId, i, terrstr()); - continue; - } - } - - SRpcMsg rpcMsg = {0}; - syncEntry2OriginalRpc(pEntry, &rpcMsg); - - sTrace("do commit index:%" PRId64 ", type:%s", i, TMSG_INFO(pEntry->msgType)); - - // user commit - if ((ths->pFsm->FpCommitCb != NULL) && syncUtilUserCommit(pEntry->originalRpcType)) { - bool internalExecute = true; - if ((ths->replicaNum == 1) && ths->restoreFinish && ths->vgId != 1) { - internalExecute = false; - } - - sNTrace(ths, "user commit index:%" PRId64 ", internal:%d, type:%s", i, internalExecute, - TMSG_INFO(pEntry->msgType)); - - // execute fsm in apply thread, or execute outside syncPropose - if (internalExecute) { - SFsmCbMeta cbMeta = { - .index = pEntry->index, - .lastConfigIndex = syncNodeGetSnapshotConfigIndex(ths, pEntry->index), - .isWeak = pEntry->isWeak, - .code = 0, - .state = ths->state, - .seqNum = pEntry->seqNum, - .term = pEntry->term, - .currentTerm = ths->raftStore.currentTerm, - .flag = flag, - }; - - syncRespMgrGetAndDel(ths->pSyncRespMgr, cbMeta.seqNum, &rpcMsg.info); - ths->pFsm->FpCommitCb(ths->pFsm, &rpcMsg, &cbMeta); - } - } - -#if 0 - // execute in pre-commit - // leader transfer - if (pEntry->originalRpcType == TDMT_SYNC_LEADER_TRANSFER) { - code = syncDoLeaderTransfer(ths, &rpcMsg, pEntry); - ASSERT(code == 0); - } -#endif - - // restore finish - // if only snapshot, a noop entry will be append, so syncLogLastIndex is always ok - if (pEntry->index == ths->pLogStore->syncLogLastIndex(ths->pLogStore)) { - if (ths->restoreFinish == false) { - if (ths->pFsm->FpRestoreFinishCb != NULL) { - ths->pFsm->FpRestoreFinishCb(ths->pFsm); - } - ths->restoreFinish = true; - - int64_t restoreDelay = taosGetTimestampMs() - ths->leaderTime; - sNTrace(ths, "restore finish, index:%" PRId64 ", elapsed:%" PRId64 " ms", pEntry->index, restoreDelay); - } - } - - rpcFreeCont(rpcMsg.pCont); - if (h) { - taosLRUCacheRelease(pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - } - } - } - return 0; -} - bool syncNodeInRaftGroup(SSyncNode* ths, SRaftId* pRaftId) { for (int32_t i = 0; i < ths->replicaNum; ++i) { if (syncUtilSameId(&((ths->replicasId)[i]), pRaftId)) { diff --git a/source/libs/sync/src/syncPipeline.c b/source/libs/sync/src/syncPipeline.c index b61fc2e90d..b95aa04e40 100644 --- a/source/libs/sync/src/syncPipeline.c +++ b/source/libs/sync/src/syncPipeline.c @@ -940,8 +940,11 @@ int32_t syncNodeLogReplMgrInit(SSyncNode* pNode) { for (int i = 0; i < TSDB_MAX_REPLICA; i++) { ASSERT(pNode->logReplMgrs[i] == NULL); pNode->logReplMgrs[i] = syncLogReplMgrCreate(); + if (pNode->logReplMgrs[i] == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + return -1; + } pNode->logReplMgrs[i]->peerId = i; - ASSERTS(pNode->logReplMgrs[i] != NULL, "Out of memory."); } return 0; } diff --git a/source/libs/sync/src/syncReplication.c b/source/libs/sync/src/syncReplication.c index 1aa476e84e..3df203221b 100644 --- a/source/libs/sync/src/syncReplication.c +++ b/source/libs/sync/src/syncReplication.c @@ -48,92 +48,6 @@ int32_t syncNodeMaybeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg); -int32_t syncNodeReplicateOne(SSyncNode* pSyncNode, SRaftId* pDestId, bool snapshot) { - ASSERT(false && "deprecated"); - // next index - SyncIndex nextIndex = syncIndexMgrGetIndex(pSyncNode->pNextIndex, pDestId); - - if (snapshot) { - // maybe start snapshot - SyncIndex logStartIndex = pSyncNode->pLogStore->syncLogBeginIndex(pSyncNode->pLogStore); - SyncIndex logEndIndex = pSyncNode->pLogStore->syncLogEndIndex(pSyncNode->pLogStore); - if (nextIndex < logStartIndex || nextIndex - 1 > logEndIndex) { - sNTrace(pSyncNode, "maybe start snapshot for next-index:%" PRId64 ", start:%" PRId64 ", end:%" PRId64, nextIndex, - logStartIndex, logEndIndex); - // start snapshot - int32_t code = syncNodeStartSnapshot(pSyncNode, pDestId); - } - } - - // pre index, pre term - SyncIndex preLogIndex = syncNodeGetPreIndex(pSyncNode, nextIndex); - SyncTerm preLogTerm = syncNodeGetPreTerm(pSyncNode, nextIndex); - - // prepare entry - SRpcMsg rpcMsg = {0}; - SyncAppendEntries* pMsg = NULL; - - SSyncRaftEntry* pEntry = NULL; - SLRUCache* pCache = pSyncNode->pLogStore->pCache; - LRUHandle* h = taosLRUCacheLookup(pCache, &nextIndex, sizeof(nextIndex)); - int32_t code = 0; - if (h) { - pEntry = (SSyncRaftEntry*)taosLRUCacheValue(pCache, h); - code = 0; - - pSyncNode->pLogStore->cacheHit++; - sNTrace(pSyncNode, "hit cache index:%" PRId64 ", bytes:%u, %p", nextIndex, pEntry->bytes, pEntry); - - } else { - pSyncNode->pLogStore->cacheMiss++; - sNTrace(pSyncNode, "miss cache index:%" PRId64, nextIndex); - - code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, nextIndex, &pEntry); - } - - if (code == 0) { - ASSERT(pEntry != NULL); - - code = syncBuildAppendEntries(&rpcMsg, (int32_t)(pEntry->bytes), pSyncNode->vgId); - ASSERT(code == 0); - - pMsg = rpcMsg.pCont; - memcpy(pMsg->data, pEntry, pEntry->bytes); - } else { - if (terrno == TSDB_CODE_WAL_LOG_NOT_EXIST) { - // no entry in log - code = syncBuildAppendEntries(&rpcMsg, 0, pSyncNode->vgId); - ASSERT(code == 0); - - pMsg = rpcMsg.pCont; - } else { - sNError(pSyncNode, "replicate to dnode:%d error, next-index:%" PRId64, DID(pDestId), nextIndex); - return -1; - } - } - - if (h) { - taosLRUCacheRelease(pCache, h, false); - } else { - syncEntryDestroy(pEntry); - } - - // prepare msg - ASSERT(pMsg != NULL); - pMsg->srcId = pSyncNode->myRaftId; - pMsg->destId = *pDestId; - pMsg->term = pSyncNode->raftStore.currentTerm; - pMsg->prevLogIndex = preLogIndex; - pMsg->prevLogTerm = preLogTerm; - pMsg->commitIndex = pSyncNode->commitIndex; - pMsg->privateTerm = 0; - // pMsg->privateTerm = syncIndexMgrGetTerm(pSyncNode->pNextIndex, pDestId); - - // send msg - syncNodeMaybeSendAppendEntries(pSyncNode, pDestId, &rpcMsg); - return 0; -} - int32_t syncNodeReplicate(SSyncNode* pNode) { SSyncLogBuffer* pBuf = pNode->pLogBuf; taosThreadMutexLock(&pBuf->mutex); @@ -156,25 +70,6 @@ int32_t syncNodeReplicateWithoutLock(SSyncNode* pNode) { return 0; } -int32_t syncNodeReplicateOld(SSyncNode* pSyncNode) { - if (pSyncNode->state != TAOS_SYNC_STATE_LEADER) { - return -1; - } - - sNTrace(pSyncNode, "do replicate"); - - int32_t ret = 0; - for (int i = 0; i < pSyncNode->peersNum; ++i) { - SRaftId* pDestId = &(pSyncNode->peersId[i]); - ret = syncNodeReplicateOne(pSyncNode, pDestId, true); - if (ret != 0) { - sError("vgId:%d, do append entries error for dnode:%d", pSyncNode->vgId, DID(pDestId)); - } - } - - return 0; -} - int32_t syncNodeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) { SyncAppendEntries* pMsg = pRpcMsg->pCont; pMsg->destId = *destRaftId; @@ -182,39 +77,6 @@ int32_t syncNodeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftI return 0; } -int32_t syncNodeSendAppendEntriesOld(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) { - int32_t ret = 0; - SyncAppendEntries* pMsg = pRpcMsg->pCont; - if (pMsg == NULL) { - sError("vgId:%d, sync-append-entries msg is NULL", pSyncNode->vgId); - return 0; - } - - SPeerState* pState = syncNodeGetPeerState(pSyncNode, destRaftId); - if (pState == NULL) { - sError("vgId:%d, replica maybe dropped", pSyncNode->vgId); - return 0; - } - - // save index, otherwise pMsg will be free by rpc - SyncIndex saveLastSendIndex = pState->lastSendIndex; - bool update = false; - if (pMsg->dataLen > 0) { - saveLastSendIndex = pMsg->prevLogIndex + 1; - update = true; - } - - syncLogSendAppendEntries(pSyncNode, pMsg, ""); - syncNodeSendMsgById(destRaftId, pSyncNode, pRpcMsg); - - if (update) { - pState->lastSendIndex = saveLastSendIndex; - pState->lastSendTime = taosGetTimestampMs(); - } - - return ret; -} - int32_t syncNodeMaybeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) { int32_t ret = 0; SyncAppendEntries* pMsg = pRpcMsg->pCont; From d6028bdde185e073b80d9ef7d9a886e8df9ec905 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Wed, 18 Jan 2023 14:59:02 +0800 Subject: [PATCH 009/124] some code --- source/dnode/vnode/src/inc/tsdb.h | 1 + source/dnode/vnode/src/inc/vnodeInt.h | 2 +- source/dnode/vnode/src/sma/smaSnapshot.c | 4 +- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 1089 +++++++++++++++----- source/dnode/vnode/src/tsdb/tsdbUtil.c | 16 +- source/dnode/vnode/src/vnd/vnodeSnapshot.c | 2 +- 6 files changed, 831 insertions(+), 283 deletions(-) diff --git a/source/dnode/vnode/src/inc/tsdb.h b/source/dnode/vnode/src/inc/tsdb.h index 5a2e462c8c..2efb00ae32 100644 --- a/source/dnode/vnode/src/inc/tsdb.h +++ b/source/dnode/vnode/src/inc/tsdb.h @@ -202,6 +202,7 @@ int32_t tsdbCmprColData(SColData *pColData, int8_t cmprAlg, SBlockCol *pBlockCol uint8_t **ppBuf); int32_t tsdbDecmprColData(uint8_t *pIn, SBlockCol *pBlockCol, int8_t cmprAlg, int32_t nVal, SColData *pColData, uint8_t **ppBuf); +int32_t tRowInfoCmprFn(const void *p1, const void *p2); // tsdbMemTable ============================================================================================== // SMemTable int32_t tsdbMemTableCreate(STsdb *pTsdb, SMemTable **ppMemTable); diff --git a/source/dnode/vnode/src/inc/vnodeInt.h b/source/dnode/vnode/src/inc/vnodeInt.h index 75367883f1..2501af7f04 100644 --- a/source/dnode/vnode/src/inc/vnodeInt.h +++ b/source/dnode/vnode/src/inc/vnodeInt.h @@ -247,7 +247,7 @@ int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader); int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData); // STsdbSnapWriter ======================================== int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWriter** ppWriter); -int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData); +int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr); int32_t tsdbSnapWriterPrepareClose(STsdbSnapWriter* pWriter); int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback); // STqSnapshotReader == diff --git a/source/dnode/vnode/src/sma/smaSnapshot.c b/source/dnode/vnode/src/sma/smaSnapshot.c index 34f884f9f9..0a6fac0fe7 100644 --- a/source/dnode/vnode/src/sma/smaSnapshot.c +++ b/source/dnode/vnode/src/sma/smaSnapshot.c @@ -423,10 +423,10 @@ int32_t rsmaSnapWrite(SRSmaSnapWriter* pWriter, uint8_t* pData, uint32_t nData) // rsma1/rsma2 if (pHdr->type == SNAP_DATA_RSMA1) { pHdr->type = SNAP_DATA_TSDB; - code = tsdbSnapWrite(pWriter->pDataWriter[0], pData, nData); + code = tsdbSnapWrite(pWriter->pDataWriter[0], pHdr); } else if (pHdr->type == SNAP_DATA_RSMA2) { pHdr->type = SNAP_DATA_TSDB; - code = tsdbSnapWrite(pWriter->pDataWriter[1], pData, nData); + code = tsdbSnapWrite(pWriter->pDataWriter[1], pHdr); } else if (pHdr->type == SNAP_DATA_QTASK) { code = rsmaSnapWriteQTaskInfo(pWriter, pData, nData); } else { diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 08d52554c6..b2574ddadd 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -15,6 +15,280 @@ #include "tsdb.h" +extern int32_t tsdbReadDataBlockEx(SDataFReader* pReader, SDataBlk* pDataBlk, SBlockData* pBlockData); + +// STsdbDataIter2 ======================================== +#define TSDB_MEM_TABLE_DATA_ITER 0 +#define TSDB_DATA_FILE_DATA_ITER 1 +#define TSDB_STT_FILE_DATA_ITER 2 + +typedef struct STsdbDataIter2 STsdbDataIter2; +struct STsdbDataIter2 { + STsdbDataIter2* next; + SRBTreeNode rbtn; + + int32_t type; + SRowInfo rowInfo; + union { + // TSDB_MEM_TABLE_DATA_ITER + struct { + SMemTable* pMemTable; + } mIter; + + // TSDB_DATA_FILE_DATA_ITER + struct { + SDataFReader* pReader; + SArray* aBlockIdx; // SArray + SMapData mDataBlk; + SBlockData bData; + int32_t iBlockIdx; + int32_t iDataBlk; + int32_t iRow; + + } dIter; + + // TSDB_STT_FILE_DATA_ITER + struct { + SDataFReader* pReader; + int32_t iStt; + SArray* aSttBlk; + SBlockData bData; + int32_t iSttBlk; + int32_t iRow; + } sIter; + }; +}; + +#define TSDB_RBTN_TO_DATA_ITER(pNode) ((STsdbDataIter2*)(((char*)pNode) - offsetof(STsdbDataIter2, rbtn))) + +/* open */ +static int32_t tsdbOpenDataFileDataIter(SDataFReader* pReader, STsdbDataIter2** ppIter) { + int32_t code = 0; + int32_t lino = 0; + + // create handle + STsdbDataIter2* pIter = (STsdbDataIter2*)taosMemoryCalloc(1, sizeof(*pIter)); + if (pIter == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + pIter->type = TSDB_DATA_FILE_DATA_ITER; + pIter->dIter.pReader = pReader; + if ((pIter->dIter.aBlockIdx = taosArrayInit(0, sizeof(SBlockIdx))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tBlockDataCreate(&pIter->dIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->dIter.iBlockIdx = -1; + pIter->dIter.iDataBlk = -1; + pIter->dIter.iRow = -1; + + // read data + code = tsdbReadBlockIdx(pReader, pIter->dIter.aBlockIdx); + TSDB_CHECK_CODE(code, lino, _exit); + + if (taosArrayGetSize(pIter->dIter.aBlockIdx) == 0) goto _clear; + +_exit: + if (code) { + if (pIter) { + _clear: + tBlockDataDestroy(&pIter->dIter.bData, 1); + taosArrayDestroy(pIter->dIter.aBlockIdx); + taosMemoryFree(pIter); + pIter = NULL; + } + } + *ppIter = pIter; + return code; +} + +static int32_t tsdbOpenSttFileDataIter(SDataFReader* pReader, int32_t iStt, STsdbDataIter2** ppIter) { + int32_t code = 0; + int32_t lino = 0; + + // create handle + STsdbDataIter2* pIter = (STsdbDataIter2*)taosMemoryCalloc(1, sizeof(*pIter)); + if (pIter == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + pIter->type = TSDB_STT_FILE_DATA_ITER; + pIter->sIter.pReader = pReader; + pIter->sIter.iStt = iStt; + pIter->sIter.aSttBlk = taosArrayInit(0, sizeof(SSttBlk)); + if (pIter->sIter.aSttBlk == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tBlockDataCreate(&pIter->sIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->sIter.iSttBlk = -1; + pIter->sIter.iRow = -1; + + // read data + code = tsdbReadSttBlk(pReader, iStt, pIter->sIter.aSttBlk); + TSDB_CHECK_CODE(code, lino, _exit); + + if (taosArrayGetSize(pIter->sIter.aSttBlk) == 0) goto _clear; + +_exit: + if (code) { + if (pIter) { + _clear: + taosArrayDestroy(pIter->sIter.aSttBlk); + tBlockDataDestroy(&pIter->sIter.bData, 1); + taosMemoryFree(pIter); + pIter = NULL; + } + } + *ppIter = pIter; + return code; +} + +/* close */ +static void tsdbCloseDataFileDataIter(STsdbDataIter2* pIter) { + tBlockDataDestroy(&pIter->dIter.bData, 1); + tMapDataClear(&pIter->dIter.mDataBlk); + taosArrayDestroy(pIter->dIter.aBlockIdx); + taosMemoryFree(pIter); +} + +static void tsdbCloseSttFileDataIter(STsdbDataIter2* pIter) { + tBlockDataDestroy(&pIter->sIter.bData, 1); + taosArrayDestroy(pIter->sIter.aSttBlk); + taosMemoryFree(pIter); +} + +static void tsdbCloseDataIter2(STsdbDataIter2* pIter) { + if (pIter->type == TSDB_MEM_TABLE_DATA_ITER) { + ASSERT(0); + } else if (pIter->type == TSDB_DATA_FILE_DATA_ITER) { + tsdbCloseDataFileDataIter(pIter); + } else if (pIter->type == TSDB_STT_FILE_DATA_ITER) { + tsdbCloseSttFileDataIter(pIter); + } else { + ASSERT(0); + } +} + +/* cmpr */ +static int32_t tsdbDataIterCmprFn(const SRBTreeNode* pNode1, const SRBTreeNode* pNode2) { + STsdbDataIter2* pIter1 = TSDB_RBTN_TO_DATA_ITER(pNode1); + STsdbDataIter2* pIter2 = TSDB_RBTN_TO_DATA_ITER(pNode2); + return tRowInfoCmprFn(&pIter1->rowInfo, &pIter2->rowInfo); +} + +/* seek */ + +/* iter next */ +static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter) { + int32_t code = 0; + int32_t lino = 0; + + for (;;) { + if (++pIter->dIter.iRow < pIter->dIter.bData.nRow) { + pIter->rowInfo.suid = pIter->dIter.bData.suid; + pIter->rowInfo.uid = pIter->dIter.bData.uid; + pIter->rowInfo.row = tsdbRowFromBlockData(&pIter->dIter.bData, pIter->dIter.iRow); + break; + } + + for (;;) { + if (++pIter->dIter.iDataBlk < pIter->dIter.mDataBlk.nItem) { + SDataBlk dataBlk; + tMapDataGetItemByIdx(&pIter->dIter.mDataBlk, pIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); + + code = tsdbReadDataBlockEx(pIter->dIter.pReader, &dataBlk, &pIter->dIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->dIter.iRow = -1; + + break; + } + + for (;;) { + if (++pIter->dIter.iBlockIdx < taosArrayGetSize(pIter->dIter.aBlockIdx)) { + SBlockIdx* pBlockIdx = taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx); + + code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->dIter.iDataBlk = -1; + + break; + } else { + pIter->rowInfo = (SRowInfo){0}; + goto _exit; + } + } + } + } + +_exit: + if (code) { + tsdbError("%s failed at line %d since %s", __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSttFileDataIterNext(STsdbDataIter2* pIter) { + int32_t code = 0; + int32_t lino = 0; + + for (;;) { + if (++pIter->sIter.iRow < pIter->sIter.bData.nRow) { + pIter->rowInfo.suid = pIter->sIter.bData.suid; + pIter->rowInfo.uid = pIter->sIter.bData.uid ? pIter->sIter.bData.uid : pIter->sIter.bData.aUid[pIter->sIter.iRow]; + pIter->rowInfo.row = tsdbRowFromBlockData(&pIter->sIter.bData, pIter->sIter.iRow); + break; + } + + if (++pIter->sIter.iSttBlk < taosArrayGetSize(pIter->sIter.aSttBlk)) { + SSttBlk* pSttBlk = taosArrayGet(pIter->sIter.aSttBlk, pIter->sIter.iSttBlk); + + code = tsdbReadSttBlockEx(pIter->sIter.pReader, pIter->sIter.iStt, pSttBlk, &pIter->sIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->sIter.iRow = -1; + } else { + pIter->rowInfo = (SRowInfo){0}; + break; + } + } + +_exit: + if (code) { + tsdbError("%s failed at line %d since %s", __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbDataIterNext2(STsdbDataIter2* pIter) { + int32_t code = 0; + + if (pIter->type == TSDB_MEM_TABLE_DATA_ITER) { + ASSERT(0); + return code; + } else if (pIter->type == TSDB_DATA_FILE_DATA_ITER) { + return tsdbDataFileDataIterNext(pIter); + } else if (pIter->type == TSDB_STT_FILE_DATA_ITER) { + return tsdbSttFileDataIterNext(pIter); + } else { + ASSERT(0); + return code; + } +} + +/* get */ + // STsdbSnapReader ======================================== typedef enum { SNAP_DATA_FILE_ITER = 0, SNAP_STT_FILE_ITER } EFIterT; typedef struct { @@ -63,8 +337,6 @@ struct STsdbSnapReader { uint8_t* aBuf[5]; }; -extern int32_t tRowInfoCmprFn(const void* p1, const void* p2); -extern int32_t tsdbReadDataBlockEx(SDataFReader* pReader, SDataBlk* pDataBlk, SBlockData* pBlockData); extern int32_t tsdbUpdateTableSchema(SMeta* pMeta, int64_t suid, int64_t uid, SSkmInfo* pSkmInfo); static int32_t tFDataIterCmprFn(const SRBTreeNode* pNode1, const SRBTreeNode* pNode2) { @@ -632,12 +904,9 @@ _exit: // STsdbSnapWriter ======================================== struct STsdbSnapWriter { - STsdb* pTsdb; - int64_t sver; - int64_t ever; - STsdbFS fs; - - // config + STsdb* pTsdb; + int64_t sver; + int64_t ever; int32_t minutes; int8_t precision; int32_t minRow; @@ -646,31 +915,31 @@ struct STsdbSnapWriter { int64_t commitID; uint8_t* aBuf[5]; - // for data file - SBlockData bData; - int32_t fid; - TABLEID id; - SSkmInfo skmTable; - struct { - SDataFReader* pReader; - SArray* aBlockIdx; - int32_t iBlockIdx; - SBlockIdx* pBlockIdx; - SMapData mDataBlk; - int32_t iDataBlk; - SBlockData bData; - int32_t iRow; - } dReader; - struct { - SDataFWriter* pWriter; - SArray* aBlockIdx; - SMapData mDataBlk; - SArray* aSttBlk; - SBlockData bData; - SBlockData sData; - } dWriter; + STsdbFS fs; - // for del file + // time-series data + SBlockData inData; + + int32_t fid; + TABLEID tbid; + SSkmInfo skmTable; + + /* reader */ + SDataFReader* pDataFReader; + STsdbDataIter2* iterList; + STsdbDataIter2* pDIter; + STsdbDataIter2* pIter; + SRBTree rbt; // SRBTree + + /* writer */ + SDataFWriter* pDataFWriter; + SArray* aBlockIdx; + SMapData mDataBlk; // SMapData + SArray* aSttBlk; // SArray + SBlockData bData; + SBlockData sData; + + // tombstone data SDelFReader* pDelFReader; SDelFWriter* pDelFWriter; int32_t iDelIdx; @@ -685,7 +954,9 @@ extern int32_t tsdbWriteSttBlock(SDataFWriter* pWriter, SBlockData* pBlockData, static int32_t tsdbSnapNextTableData(STsdbSnapWriter* pWriter) { int32_t code = 0; + int32_t lino = 0; +#if 0 ASSERT(pWriter->dReader.iRow >= pWriter->dReader.bData.nRow); if (pWriter->dReader.iBlockIdx < taosArrayGetSize(pWriter->dReader.aBlockIdx)) { @@ -702,14 +973,20 @@ static int32_t tsdbSnapNextTableData(STsdbSnapWriter* pWriter) { pWriter->dReader.iDataBlk = 0; // point to the next one tBlockDataReset(&pWriter->dReader.bData); pWriter->dReader.iRow = 0; +#endif _exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; } static int32_t tsdbSnapWriteCopyData(STsdbSnapWriter* pWriter, TABLEID* pId) { int32_t code = 0; + int32_t lino = 0; +#if 0 while (true) { if (pWriter->dReader.pBlockIdx == NULL) break; if (tTABLEIDCmprFn(pWriter->dReader.pBlockIdx, pId) >= 0) break; @@ -726,37 +1003,93 @@ static int32_t tsdbSnapWriteCopyData(STsdbSnapWriter* pWriter, TABLEID* pId) { code = tsdbSnapNextTableData(pWriter); if (code) goto _exit; } +#endif _exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; } static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pId) { int32_t code = 0; + int32_t lino = 0; + if (pWriter->pDIter) { + STsdbDataIter2* pIter = pWriter->pDIter; + for (;;) { + if (pIter->dIter.iBlockIdx + 1 >= taosArrayGetSize(pIter->dIter.aBlockIdx)) { + pWriter->pDIter = NULL; + break; + } + + SBlockIdx* pBlockIdx = (SBlockIdx*)taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx + 1); + + int32_t c = tTABLEIDCmprFn(pBlockIdx, pId); + if (c < 0) { + ++pIter->dIter.iBlockIdx; + + code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); + TSDB_CHECK_CODE(code, lino, _exit); + + SBlockIdx* pNewBlockIdx = taosArrayReserve(pWriter->aBlockIdx, 1); + if (pNewBlockIdx == NULL) { + code == TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + pNewBlockIdx->suid = pBlockIdx->suid; + pNewBlockIdx->uid = pBlockIdx->uid; + + code = tsdbWriteDataBlk(pWriter->pDataFWriter, &pIter->dIter.mDataBlk, pBlockIdx); + TSDB_CHECK_CODE(code, lino, _exit); + } else if (c == 0) { + ++pIter->dIter.iBlockIdx; + + code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->dIter.iDataBlk = -1; + + break; + } else { + break; + } + } + } + + pWriter->tbid = pId[0]; + + tMapDataReset(&pWriter->mDataBlk); + +#if 0 code = tsdbSnapWriteCopyData(pWriter, pId); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); pWriter->id.suid = pId->suid; pWriter->id.uid = pId->uid; code = tsdbUpdateTableSchema(pWriter->pTsdb->pVnode->pMeta, pId->suid, pId->uid, &pWriter->skmTable); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); tMapDataReset(&pWriter->dWriter.mDataBlk); code = tBlockDataInit(&pWriter->dWriter.bData, pId, pWriter->skmTable.pTSchema, NULL, 0); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); +#endif - return code; - -_err: - tsdbError("vgId:%d, %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; } static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { int32_t code = 0; + int32_t lino = 0; +#if 0 if (pWriter->id.suid == 0 && pWriter->id.uid == 0) return code; int32_t c = 1; @@ -807,123 +1140,174 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { pWriter->id.suid = 0; pWriter->id.uid = 0; - - return code; - -_err: - return code; -} - -static int32_t tsdbSnapWriteOpenFile(STsdbSnapWriter* pWriter, int32_t fid) { - int32_t code = 0; - STsdb* pTsdb = pWriter->pTsdb; - - ASSERT(pWriter->dWriter.pWriter == NULL); - - pWriter->fid = fid; - pWriter->id = (TABLEID){0}; - SDFileSet* pSet = taosArraySearch(pWriter->fs.aDFileSet, &(SDFileSet){.fid = fid}, tDFileSetCmprFn, TD_EQ); - - // Reader - if (pSet) { - code = tsdbDataFReaderOpen(&pWriter->dReader.pReader, pWriter->pTsdb, pSet); - if (code) goto _err; - - code = tsdbReadBlockIdx(pWriter->dReader.pReader, pWriter->dReader.aBlockIdx); - if (code) goto _err; - } else { - ASSERT(pWriter->dReader.pReader == NULL); - taosArrayClear(pWriter->dReader.aBlockIdx); - } - pWriter->dReader.iBlockIdx = 0; // point to the next one - code = tsdbSnapNextTableData(pWriter); - if (code) goto _err; - - // Writer - SHeadFile fHead = {.commitID = pWriter->commitID}; - SDataFile fData = {.commitID = pWriter->commitID}; - SSmaFile fSma = {.commitID = pWriter->commitID}; - SSttFile fStt = {.commitID = pWriter->commitID}; - SDFileSet wSet = {.fid = pWriter->fid, .pHeadF = &fHead, .pDataF = &fData, .pSmaF = &fSma}; - if (pSet) { - wSet.diskId = pSet->diskId; - fData = *pSet->pDataF; - fSma = *pSet->pSmaF; - for (int32_t iStt = 0; iStt < pSet->nSttF; iStt++) { - wSet.aSttF[iStt] = pSet->aSttF[iStt]; - } - wSet.nSttF = pSet->nSttF + 1; // TODO: fix pSet->nSttF == pTsdb->maxFile - } else { - SDiskID did = {0}; - tfsAllocDisk(pTsdb->pVnode->pTfs, 0, &did); - tfsMkdirRecurAt(pTsdb->pVnode->pTfs, pTsdb->path, did); - wSet.diskId = did; - wSet.nSttF = 1; - } - wSet.aSttF[wSet.nSttF - 1] = &fStt; - - code = tsdbDataFWriterOpen(&pWriter->dWriter.pWriter, pWriter->pTsdb, &wSet); - if (code) goto _err; - taosArrayClear(pWriter->dWriter.aBlockIdx); - tMapDataReset(&pWriter->dWriter.mDataBlk); - taosArrayClear(pWriter->dWriter.aSttBlk); - tBlockDataReset(&pWriter->dWriter.bData); - tBlockDataReset(&pWriter->dWriter.sData); - - return code; - -_err: - return code; -} - -static int32_t tsdbSnapWriteCloseFile(STsdbSnapWriter* pWriter) { - int32_t code = 0; - - ASSERT(pWriter->dWriter.pWriter); - - code = tsdbSnapWriteTableDataEnd(pWriter); - if (code) goto _err; - - // copy remain table data - TABLEID id = {.suid = INT64_MAX, .uid = INT64_MAX}; - code = tsdbSnapWriteCopyData(pWriter, &id); - if (code) goto _err; - - code = - tsdbWriteSttBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.sData, pWriter->dWriter.aSttBlk, pWriter->cmprAlg); - if (code) goto _err; - - // Indices - code = tsdbWriteBlockIdx(pWriter->dWriter.pWriter, pWriter->dWriter.aBlockIdx); - if (code) goto _err; - - code = tsdbWriteSttBlk(pWriter->dWriter.pWriter, pWriter->dWriter.aSttBlk); - if (code) goto _err; - - code = tsdbUpdateDFileSetHeader(pWriter->dWriter.pWriter); - if (code) goto _err; - - code = tsdbFSUpsertFSet(&pWriter->fs, &pWriter->dWriter.pWriter->wSet); - if (code) goto _err; - - code = tsdbDataFWriterClose(&pWriter->dWriter.pWriter, 1); - if (code) goto _err; - - if (pWriter->dReader.pReader) { - code = tsdbDataFReaderClose(&pWriter->dReader.pReader); - if (code) goto _err; - } +#endif _exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; +} -_err: +static int32_t tsdbSnapWriteOpenDataFile(STsdbSnapWriter* pWriter, int32_t fid) { + int32_t code = 0; + int32_t lino = 0; + + ASSERT(pWriter->pDataFWriter == NULL && pWriter->fid < fid); + + STsdb* pTsdb = pWriter->pTsdb; + + pWriter->fid = fid; + pWriter->tbid = (TABLEID){0}; + SDFileSet* pSet = taosArraySearch(pWriter->fs.aDFileSet, &(SDFileSet){.fid = fid}, tDFileSetCmprFn, TD_EQ); + + // open reader + pWriter->pDataFReader = NULL; + pWriter->iterList = NULL; + pWriter->pDIter = NULL; + pWriter->pIter = NULL; + tRBTreeCreate(&pWriter->rbt, tsdbDataIterCmprFn); + if (pSet) { + code = tsdbDataFReaderOpen(&pWriter->pDataFReader, pTsdb, pSet); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbOpenDataFileDataIter(pWriter->pDataFReader, &pWriter->pDIter); + TSDB_CHECK_CODE(code, lino, _exit); + if (pWriter->pDIter) { + pWriter->pDIter->next = pWriter->iterList; + pWriter->iterList = pWriter->pDIter; + } + + for (int32_t iStt = 0; iStt < pSet->nSttF; iStt++) { + code = tsdbOpenSttFileDataIter(pWriter->pDataFReader, iStt, &pWriter->pIter); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->pIter) { + code = tsdbSttFileDataIterNext(pWriter->pIter); + TSDB_CHECK_CODE(code, lino, _exit); + + // add to tree + tRBTreePut(&pWriter->rbt, &pWriter->pIter->rbtn); + + // add to list + pWriter->pIter->next = pWriter->iterList; + pWriter->iterList = pWriter->pIter; + } + } + + pWriter->pIter = NULL; + } + + // open writer + SDiskID diskId; + if (pSet) { + diskId = pSet->diskId; + } else { + tfsAllocDisk(pTsdb->pVnode->pTfs, 0 /*TODO*/, &diskId); + tfsMkdirRecurAt(pTsdb->pVnode->pTfs, pTsdb->path, diskId); + } + SDFileSet wSet = {.diskId = diskId, + .fid = fid, + .pHeadF = &(SHeadFile){.commitID = pWriter->commitID}, + .pDataF = (pSet) ? pSet->pDataF : &(SDataFile){.commitID = pWriter->commitID}, + .pSmaF = (pSet) ? pSet->pSmaF : &(SSmaFile){.commitID = pWriter->commitID}, + .nSttF = 1, + .aSttF = {&(SSttFile){.commitID = pWriter->commitID}}}; + code = tsdbDataFWriterOpen(&pWriter->pDataFWriter, pTsdb, &wSet); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->aBlockIdx) { + taosArrayClear(pWriter->aBlockIdx); + } else if ((pWriter->aBlockIdx = taosArrayInit(0, sizeof(SBlockIdx))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + tMapDataReset(&pWriter->mDataBlk); + + if (pWriter->aSttBlk) { + taosArrayClear(pWriter->aSttBlk); + } else if ((pWriter->aSttBlk = taosArrayInit(0, sizeof(SSttBlk))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + tBlockDataReset(&pWriter->bData); + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s, fid:%d", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code), + fid); + } else { + tsdbDebug("vgId:%d %s done, fid:%d", TD_VID(pTsdb->pVnode), __func__, fid); + } + return code; +} + +static int32_t tsdbSnapWriteCloseDataFile(STsdbSnapWriter* pWriter) { + int32_t code = 0; + int32_t lino = 0; + + ASSERT(pWriter->pDataFWriter); + +#if 0 + // loop write remain data + for (;;) { + SRowInfo* pRowInfo; + + code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pRowInfo == NULL) break; + + code = tsdbSnapWriteTableData(pWriter, pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbSnapWriteNextRow(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // TODO: write remain data +#endif + + // do file-level updates + code = tsdbWriteSttBlk(pWriter->pDataFWriter, pWriter->aSttBlk); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbWriteBlockIdx(pWriter->pDataFWriter, pWriter->aBlockIdx); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbUpdateDFileSetHeader(pWriter->pDataFWriter); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbFSUpsertFSet(&pWriter->fs, &pWriter->pDataFWriter->wSet); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbDataFWriterClose(&pWriter->pDataFWriter, 1); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->pDataFReader) { + code = tsdbDataFReaderClose(&pWriter->pDataFReader); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // TODO: do clear sources + {} + +_exit: + if (code) { + tsdbError("vgId:%d %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s is done", TD_VID(pWriter->pTsdb->pVnode), __func__); + } return code; } static int32_t tsdbSnapWriteToDataFile(STsdbSnapWriter* pWriter, int32_t iRow, int8_t* done) { int32_t code = 0; + int32_t lino = 0; +#if 0 SBlockData* pBData = &pWriter->bData; TABLEID id = {.suid = pBData->suid, .uid = pBData->uid ? pBData->uid : pBData->aUid[iRow]}; TSDBROW row = tsdbRowFromBlockData(pBData, iRow); @@ -942,10 +1326,10 @@ static int32_t tsdbSnapWriteToDataFile(STsdbSnapWriter* pWriter, int32_t iRow, i int32_t c = tsdbKeyCmprFn(&key, &tKey); if (c < 0) { code = tBlockDataAppendRow(&pWriter->dWriter.bData, &row, NULL, id.uid); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } else if (c > 0) { code = tBlockDataAppendRow(&pWriter->dWriter.bData, &trow, NULL, id.uid); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } else { ASSERT(0); } @@ -953,7 +1337,7 @@ static int32_t tsdbSnapWriteToDataFile(STsdbSnapWriter* pWriter, int32_t iRow, i if (pWriter->dWriter.bData.nRow >= pWriter->maxRow) { code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.bData, &pWriter->dWriter.mDataBlk, pWriter->cmprAlg); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } if (c < 0) { @@ -972,25 +1356,25 @@ static int32_t tsdbSnapWriteToDataFile(STsdbSnapWriter* pWriter, int32_t iRow, i if (c < 0) { code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.bData, &pWriter->dWriter.mDataBlk, pWriter->cmprAlg); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); code = tMapDataPutItem(&pWriter->dWriter.mDataBlk, &dataBlk, tPutDataBlk); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } else if (c > 0) { code = tBlockDataAppendRow(&pWriter->dWriter.bData, &row, NULL, id.uid); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); if (pWriter->dWriter.bData.nRow >= pWriter->maxRow) { code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.bData, &pWriter->dWriter.mDataBlk, pWriter->cmprAlg); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } *done = 1; goto _exit; } else { code = tsdbReadDataBlockEx(pWriter->dReader.pReader, &dataBlk, &pWriter->dReader.bData); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); pWriter->dReader.iRow = 0; pWriter->dReader.iDataBlk++; @@ -998,18 +1382,20 @@ static int32_t tsdbSnapWriteToDataFile(STsdbSnapWriter* pWriter, int32_t iRow, i } } } +#endif _exit: - return code; - -_err: - tsdbError("vgId:%d, %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; } static int32_t tsdbSnapWriteToSttFile(STsdbSnapWriter* pWriter, int32_t iRow) { int32_t code = 0; + int32_t lino = 0; +#if 0 TABLEID id = {.suid = pWriter->bData.suid, .uid = pWriter->bData.uid ? pWriter->bData.uid : pWriter->bData.aUid[iRow]}; TSDBROW row = tsdbRowFromBlockData(&pWriter->bData, iRow); @@ -1017,8 +1403,8 @@ static int32_t tsdbSnapWriteToSttFile(STsdbSnapWriter* pWriter, int32_t iRow) { if (pBData->suid || pBData->uid) { if (!TABLE_SAME_SCHEMA(pBData->suid, pBData->uid, id.suid, id.uid)) { - code = tsdbWriteSttBlock(pWriter->dWriter.pWriter, pBData, pWriter->dWriter.aSttBlk, pWriter->cmprAlg); - if (code) goto _err; + code = tsdbWriteSttBlock(pWriter->pDataFWriter, pBData, pWriter->aSttBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); pBData->suid = 0; pBData->uid = 0; @@ -1027,105 +1413,263 @@ static int32_t tsdbSnapWriteToSttFile(STsdbSnapWriter* pWriter, int32_t iRow) { if (pBData->suid == 0 && pBData->uid == 0) { code = tsdbUpdateTableSchema(pWriter->pTsdb->pVnode->pMeta, pWriter->id.suid, pWriter->id.uid, &pWriter->skmTable); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); TABLEID tid = {.suid = pWriter->id.suid, .uid = pWriter->id.suid ? 0 : pWriter->id.uid}; code = tBlockDataInit(pBData, &tid, pWriter->skmTable.pTSchema, NULL, 0); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } code = tBlockDataAppendRow(pBData, &row, NULL, id.uid); - if (code) goto _err; + TSDB_CHECK_CODE(code, lino, _exit); if (pBData->nRow >= pWriter->maxRow) { - code = tsdbWriteSttBlock(pWriter->dWriter.pWriter, pBData, pWriter->dWriter.aSttBlk, pWriter->cmprAlg); - if (code) goto _err; + code = tsdbWriteSttBlock(pWriter->pDataFWriter, pBData, pWriter->aSttBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); } +#endif _exit: - return code; - -_err: + if (code) { + tsdbError("vgId:%d %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); + } return code; } -static int32_t tsdbSnapWriteRowData(STsdbSnapWriter* pWriter, int32_t iRow) { +static int32_t tsdbSnapWriteNextRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowInfo) { int32_t code = 0; + int32_t lino = 0; - SBlockData* pBlockData = &pWriter->bData; - TABLEID id = {.suid = pBlockData->suid, .uid = pBlockData->uid ? pBlockData->uid : pBlockData->aUid[iRow]}; + if (pWriter->pIter) { + code = tsdbDataIterNext2(pWriter->pIter); + TSDB_CHECK_CODE(code, lino, _exit); - // End last table data write if need - if (tTABLEIDCmprFn(&pWriter->id, &id) != 0) { - code = tsdbSnapWriteTableDataEnd(pWriter); - if (code) goto _err; + if (pWriter->pIter->rowInfo.suid == 0 && pWriter->pIter->rowInfo.uid == 0) { + pWriter->pIter = NULL; + } else { + SRBTreeNode* pNode = tRBTreeMin(&pWriter->rbt); + if (pNode) { + int32_t c = tsdbDataIterCmprFn(&pWriter->pIter->rbtn, pNode); + if (c > 0) { + tRBTreePut(&pWriter->rbt, &pWriter->pIter->rbtn); + pWriter->pIter = NULL; + } else if (c == 0) { + ASSERT(0); + } + } + } } - // Start new table data write if need - if (pWriter->id.suid == 0 && pWriter->id.uid == 0) { - code = tsdbSnapWriteTableDataStart(pWriter, &id); - if (code) goto _err; + if (pWriter->pIter == NULL) { + SRBTreeNode* pNode = tRBTreeMin(&pWriter->rbt); + if (pNode) { + tRBTreeDrop(&pWriter->rbt, pNode); + pWriter->pIter = TSDB_RBTN_TO_DATA_ITER(pNode); + } } - // Merge with .data file data - int8_t done = 0; - if (pWriter->dReader.pBlockIdx && tTABLEIDCmprFn(pWriter->dReader.pBlockIdx, &id) == 0) { - code = tsdbSnapWriteToDataFile(pWriter, iRow, &done); - if (code) goto _err; - } - - // Append to the .stt data block (todo: check if need to set/reload sst block) - if (!done) { - code = tsdbSnapWriteToSttFile(pWriter, iRow); - if (code) goto _err; + if (ppRowInfo) { + if (pWriter->pIter) { + *ppRowInfo = &pWriter->pIter->rowInfo; + } else { + *ppRowInfo = NULL; + } } _exit: - return code; - -_err: - tsdbError("vgId:%d, %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; } -static int32_t tsdbSnapWriteData(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData) { - int32_t code = 0; - STsdb* pTsdb = pWriter->pTsdb; - SBlockData* pBlockData = &pWriter->bData; +static int32_t tsdbSnapWriteGetRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowInfo) { + int32_t code = 0; + int32_t lino = 0; - // Decode data - SSnapDataHdr* pHdr = (SSnapDataHdr*)pData; - code = tDecmprBlockData(pHdr->data, pHdr->size, pBlockData, pWriter->aBuf); - if (code) goto _err; + if (pWriter->pIter) { + *ppRowInfo = &pWriter->pIter->rowInfo; + goto _exit; + } - ASSERT(pBlockData->nRow > 0); + code = tsdbSnapWriteNextRow(pWriter, ppRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); - // Loop to handle each row - for (int32_t iRow = 0; iRow < pBlockData->nRow; iRow++) { - TSKEY ts = pBlockData->aTSKEY[iRow]; - int32_t fid = tsdbKeyFid(ts, pWriter->minutes, pWriter->precision); +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} - if (pWriter->dWriter.pWriter == NULL || pWriter->fid != fid) { - if (pWriter->dWriter.pWriter) { - // ASSERT(fid > pWriter->fid); +static int32_t tsdbSnapWriteRowImpl(STsdbSnapWriter* pWriter, SRowInfo* pRowInfo) { + int32_t code = 0; + int32_t lino = 0; - code = tsdbSnapWriteCloseFile(pWriter); - if (code) goto _err; - } + // TODO + ASSERT(0); - code = tsdbSnapWriteOpenFile(pWriter, fid); - if (code) goto _err; +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { + int32_t code = 0; + int32_t lino = 0; + + code = tBlockDataAppendRow(&pWriter->bData, pRow, NULL, pWriter->tbid.uid); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->bData.nRow >= pWriter->maxRow) { + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowInfo) { + int32_t code = 0; + int32_t lino = 0; + + // switch to new table if need + if (pRowInfo->uid != pWriter->tbid.uid) { + if (pRowInfo->uid) { + code = tsdbSnapWriteTableDataEnd(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); } - code = tsdbSnapWriteRowData(pWriter, iRow); - if (code) goto _err; + code = tsdbSnapWriteTableDataStart(pWriter, (TABLEID*)pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); } - return code; + // do write the row + if (pWriter->pDIter == NULL /* || false */) { + goto _write_incoming_row; + } else { + for (;;) { + while (pWriter->pDIter->dIter.iRow + 1 < pWriter->pDIter->dIter.bData.nRow) { + TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow + 1); -_err: - tsdbError("vgId:%d, vnode snapshot tsdb write data for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path, - tstrerror(code)); + int32_t c = tsdbRowCmprFn(&pRowInfo->row, &row); + if (c < 0) { + goto _write_incoming_row; + } else if (c > 0) { + ++pWriter->pDIter->dIter.iRow; + + code = tsdbSnapWriteTableRow(pWriter, &row); + TSDB_CHECK_CODE(code, lino, _exit); + } else { + ASSERT(0); + } + } + + while (pWriter->pDIter->dIter.iDataBlk < pWriter->pDIter->dIter.mDataBlk.nItem) { + SDataBlk dataBlk; + + tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk + 1, &dataBlk, + tGetDataBlk); + + int32_t c = tDataBlkCmprFn( + &dataBlk, &(SDataBlk){.minKey = TSDBROW_KEY(&pRowInfo->row), .maxKey = TSDBROW_KEY(&pRowInfo->row)}); + if (c > 0) { + goto _write_incoming_row; + } else if (c < 0) { + ++pWriter->pDIter->dIter.iDataBlk; + tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); + } else { + ++pWriter->pDIter->dIter.iDataBlk; + + code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pDIter->dIter.iRow = -1; + break; + } + } + } + } + +_write_incoming_row: + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapWriteTimeSeriesData(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { + int32_t code = 0; + int32_t lino = 0; + + code = tDecmprBlockData(pHdr->data, pHdr->size, &pWriter->inData, pWriter->aBuf); + TSDB_CHECK_CODE(code, lino, _exit); + + ASSERT(pWriter->inData.nRow > 0); + + // switch to new data file if need + int32_t fid = tsdbKeyFid(pWriter->inData.aTSKEY[0], pWriter->minutes, pWriter->precision); + if (pWriter->fid != fid) { + if (pWriter->pDataFWriter) { + code = tsdbSnapWriteCloseDataFile(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tsdbSnapWriteOpenDataFile(pWriter, fid); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // loop write each row + SRowInfo* pRowInfo; + + code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + for (int32_t iRow = 0; iRow < pWriter->bData.nRow; ++iRow) { + SRowInfo rInfo = {.suid = pWriter->inData.suid, + .uid = pWriter->inData.uid ? pWriter->inData.uid : pWriter->inData.aUid[iRow], + .row = tsdbRowFromBlockData(&pWriter->inData, iRow)}; + + for (;;) { + if (pRowInfo == NULL) { + code = tsdbSnapWriteTableData(pWriter, &rInfo); + TSDB_CHECK_CODE(code, lino, _exit); + break; + } else { + int32_t c = tRowInfoCmprFn(&rInfo, pRowInfo); + if (c < 0) { + code = tsdbSnapWriteTableData(pWriter, &rInfo); + TSDB_CHECK_CODE(code, lino, _exit); + break; + } else if (c > 0) { + code = tsdbSnapWriteTableData(pWriter, pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbSnapWriteNextRow(pWriter, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + } else { + ASSERT(0); + } + } + } + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64 " nRow:%d", TD_VID(pWriter->pTsdb->pVnode), __func__, + pWriter->inData.suid, pWriter->inData.uid, pWriter->inData.nRow); + } return code; } @@ -1159,7 +1703,7 @@ _exit: return code; } -static int32_t tsdbSnapWriteDel(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData) { +static int32_t tsdbSnapWriteDelData(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { int32_t code = 0; STsdb* pTsdb = pWriter->pTsdb; @@ -1186,10 +1730,7 @@ static int32_t tsdbSnapWriteDel(STsdbSnapWriter* pWriter, uint8_t* pData, uint32 taosArrayClear(pWriter->aDelIdxW); } - SSnapDataHdr* pHdr = (SSnapDataHdr*)pData; - TABLEID id = *(TABLEID*)pHdr->data; - - ASSERT(pHdr->size + sizeof(SSnapDataHdr) == nData); + TABLEID id = *(TABLEID*)pHdr->data; // Move write data < id code = tsdbSnapMoveWriteDelData(pWriter, &id); @@ -1208,11 +1749,11 @@ static int32_t tsdbSnapWriteDel(STsdbSnapWriter* pWriter, uint8_t* pData, uint32 taosArrayClear(pWriter->aDelData); } - int64_t n = sizeof(SSnapDataHdr) + sizeof(TABLEID); - while (n < nData) { + int64_t n = sizeof(TABLEID); + while (n < pHdr->size) { SDelData delData; - n += tGetDelData(pData + n, &delData); + n += tGetDelData(pHdr->data + n, &delData); if (taosArrayPush(pWriter->aDelData, &delData) == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; @@ -1275,12 +1816,11 @@ _err: // APIs int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWriter** ppWriter) { - int32_t code = 0; - int32_t lino = 0; - STsdbSnapWriter* pWriter = NULL; + int32_t code = 0; + int32_t lino = 0; // alloc - pWriter = (STsdbSnapWriter*)taosMemoryCalloc(1, sizeof(*pWriter)); + STsdbSnapWriter* pWriter = (STsdbSnapWriter*)taosMemoryCalloc(1, sizeof(*pWriter)); if (pWriter == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; TSDB_CHECK_CODE(code, lino, _exit); @@ -1288,11 +1828,6 @@ int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWr pWriter->pTsdb = pTsdb; pWriter->sver = sver; pWriter->ever = ever; - - code = tsdbFSCopy(pTsdb, &pWriter->fs); - TSDB_CHECK_CODE(code, lino, _exit); - - // config pWriter->minutes = pTsdb->keepCfg.days; pWriter->precision = pTsdb->keepCfg.precision; pWriter->minRow = pTsdb->pVnode->config.tsdbCfg.minRows; @@ -1300,7 +1835,19 @@ int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWr pWriter->cmprAlg = pTsdb->pVnode->config.tsdbCfg.compression; pWriter->commitID = pTsdb->pVnode->state.commitID; + code = tsdbFSCopy(pTsdb, &pWriter->fs); + TSDB_CHECK_CODE(code, lino, _exit); + // SNAP_DATA_TSDB +#if 1 + pWriter->fid = INT32_MIN; + + code = tBlockDataCreate(&pWriter->inData); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tBlockDataCreate(&pWriter->bData); + TSDB_CHECK_CODE(code, lino, _exit); +#else code = tBlockDataCreate(&pWriter->bData); TSDB_CHECK_CODE(code, lino, _exit); @@ -1330,6 +1877,7 @@ int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWr TSDB_CHECK_CODE(code, lino, _exit); code = tBlockDataCreate(&pWriter->dWriter.sData); TSDB_CHECK_CODE(code, lino, _exit); +#endif // SNAP_DATA_DEL pWriter->aDelIdxR = taosArrayInit(0, sizeof(SDelIdx)); @@ -1354,6 +1902,7 @@ _exit: *ppWriter = NULL; if (pWriter) { +#if 0 if (pWriter->aDelIdxW) taosArrayDestroy(pWriter->aDelIdxW); if (pWriter->aDelData) taosArrayDestroy(pWriter->aDelData); if (pWriter->aDelIdxR) taosArrayDestroy(pWriter->aDelIdxR); @@ -1366,6 +1915,7 @@ _exit: tBlockDataDestroy(&pWriter->bData, 1); tsdbFSDestroy(&pWriter->fs); taosMemoryFree(pWriter); +#endif } } else { tsdbInfo("vgId:%d, %s done", TD_VID(pTsdb->pVnode), __func__); @@ -1376,8 +1926,8 @@ _exit: int32_t tsdbSnapWriterPrepareClose(STsdbSnapWriter* pWriter) { int32_t code = 0; - if (pWriter->dWriter.pWriter) { - code = tsdbSnapWriteCloseFile(pWriter); + if (pWriter->pDataFWriter) { + code = tsdbSnapWriteCloseDataFile(pWriter); if (code) goto _exit; } @@ -1422,17 +1972,17 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { // SNAP_DATA_TSDB - // Writer - tBlockDataDestroy(&pWriter->dWriter.sData, 1); - tBlockDataDestroy(&pWriter->dWriter.bData, 1); - taosArrayDestroy(pWriter->dWriter.aSttBlk); - tMapDataClear(&pWriter->dWriter.mDataBlk); - taosArrayDestroy(pWriter->dWriter.aBlockIdx); + // // Writer + // tBlockDataDestroy(&pWriter->dWriter.sData, 1); + // tBlockDataDestroy(&pWriter->dWriter.bData, 1); + // taosArrayDestroy(pWriter->dWriter.aSttBlk); + // tMapDataClear(&pWriter->dWriter.mDataBlk); + // taosArrayDestroy(pWriter->dWriter.aBlockIdx); - // Reader - tBlockDataDestroy(&pWriter->dReader.bData, 1); - tMapDataClear(&pWriter->dReader.mDataBlk); - taosArrayDestroy(pWriter->dReader.aBlockIdx); + // // Reader + // tBlockDataDestroy(&pWriter->dReader.bData, 1); + // tMapDataClear(&pWriter->dReader.mDataBlk); + // taosArrayDestroy(pWriter->dReader.aBlockIdx); tBlockDataDestroy(&pWriter->bData, 1); tDestroyTSchema(pWriter->skmTable.pTSchema); @@ -1453,35 +2003,32 @@ _err: return code; } -int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, uint8_t* pData, uint32_t nData) { - int32_t code = 0; - SSnapDataHdr* pHdr = (SSnapDataHdr*)pData; +int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { + int32_t code = 0; + int32_t lino = 0; - // ts data if (pHdr->type == SNAP_DATA_TSDB) { - code = tsdbSnapWriteData(pWriter, pData, nData); - if (code) goto _err; - + code = tsdbSnapWriteTimeSeriesData(pWriter, pHdr); + TSDB_CHECK_CODE(code, lino, _exit); goto _exit; - } else { - if (pWriter->dWriter.pWriter) { - code = tsdbSnapWriteCloseFile(pWriter); - if (code) goto _err; - } + } else if (pWriter->pDataFWriter) { + code = tsdbSnapWriteCloseDataFile(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); } - // del data if (pHdr->type == SNAP_DATA_DEL) { - code = tsdbSnapWriteDel(pWriter, pData, nData); - if (code) goto _err; + code = tsdbSnapWriteDelData(pWriter, pHdr); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; } _exit: - tsdbDebug("vgId:%d, tsdb snapshot write for %s succeed", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path); - return code; - -_err: - tsdbError("vgId:%d, tsdb snapshot write for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), pWriter->pTsdb->path, - tstrerror(code)); + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s, type:%d index:%" PRId64 " size:%" PRId64, + TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code), pHdr->type, pHdr->index, pHdr->size); + } else { + tsdbDebug("vgId:%d %s done, type:%d index:%" PRId64 " size:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, + pHdr->type, pHdr->index, pHdr->size); + } return code; } diff --git a/source/dnode/vnode/src/tsdb/tsdbUtil.c b/source/dnode/vnode/src/tsdb/tsdbUtil.c index 86adc1dc80..10b3ef8562 100644 --- a/source/dnode/vnode/src/tsdb/tsdbUtil.c +++ b/source/dnode/vnode/src/tsdb/tsdbUtil.c @@ -684,7 +684,7 @@ int32_t tRowMergerInit2(SRowMerger *pMerger, STSchema *pResTSchema, TSDBROW *pRo tsdbRowGetColVal(pRow, pTSchema, jCol++, pColVal); if ((!COL_VAL_IS_NONE(pColVal)) && (!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) { uint8_t *pVal = pColVal->value.pData; - + pColVal->value.pData = NULL; code = tRealloc(&pColVal->value.pData, pColVal->value.nData); if (code) goto _exit; @@ -753,7 +753,7 @@ int32_t tRowMergerAdd(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) { code = tRealloc(&tColVal->value.pData, pColVal->value.nData); if (code) return code; - tColVal->value.nData = pColVal->value.nData; + tColVal->value.nData = pColVal->value.nData; if (pColVal->value.nData) { memcpy(tColVal->value.pData, pColVal->value.pData, pColVal->value.nData); } @@ -802,7 +802,7 @@ int32_t tRowMergerInit(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) { tsdbRowGetColVal(pRow, pTSchema, iCol, pColVal); if ((!COL_VAL_IS_NONE(pColVal)) && (!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) { uint8_t *pVal = pColVal->value.pData; - + pColVal->value.pData = NULL; code = tRealloc(&pColVal->value.pData, pColVal->value.nData); if (code) goto _exit; @@ -811,7 +811,7 @@ int32_t tRowMergerInit(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) { memcpy(pColVal->value.pData, pVal, pColVal->value.nData); } } - + if (taosArrayPush(pMerger->pArray, pColVal) == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; goto _exit; @@ -822,7 +822,7 @@ _exit: return code; } -void tRowMergerClear(SRowMerger *pMerger) { +void tRowMergerClear(SRowMerger *pMerger) { for (int32_t iCol = 1; iCol < pMerger->pTSchema->numOfCols; iCol++) { SColVal *pTColVal = taosArrayGet(pMerger->pArray, iCol); if (IS_VAR_DATA_TYPE(pTColVal->type)) { @@ -830,7 +830,7 @@ void tRowMergerClear(SRowMerger *pMerger) { } } - taosArrayDestroy(pMerger->pArray); + taosArrayDestroy(pMerger->pArray); } int32_t tRowMerge(SRowMerger *pMerger, TSDBROW *pRow) { @@ -853,7 +853,7 @@ int32_t tRowMerge(SRowMerger *pMerger, TSDBROW *pRow) { pTColVal->value.nData = pColVal->value.nData; if (pTColVal->value.nData) { - memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData); + memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData); } pTColVal->flag = 0; } else { @@ -875,7 +875,7 @@ int32_t tRowMerge(SRowMerger *pMerger, TSDBROW *pRow) { tColVal->value.nData = pColVal->value.nData; if (tColVal->value.nData) { - memcpy(tColVal->value.pData, pColVal->value.pData, tColVal->value.nData); + memcpy(tColVal->value.pData, pColVal->value.pData, tColVal->value.nData); } tColVal->flag = 0; } else { diff --git a/source/dnode/vnode/src/vnd/vnodeSnapshot.c b/source/dnode/vnode/src/vnd/vnodeSnapshot.c index e75dc24329..43f903dc48 100644 --- a/source/dnode/vnode/src/vnd/vnodeSnapshot.c +++ b/source/dnode/vnode/src/vnd/vnodeSnapshot.c @@ -455,7 +455,7 @@ int32_t vnodeSnapWrite(SVSnapWriter *pWriter, uint8_t *pData, uint32_t nData) { if (code) goto _err; } - code = tsdbSnapWrite(pWriter->pTsdbSnapWriter, pData, nData); + code = tsdbSnapWrite(pWriter->pTsdbSnapWriter, pHdr); if (code) goto _err; } break; case SNAP_DATA_TQ_HANDLE: { From d9b22b32991c072fe81e5ffd2ea4e3e8f5c3a087 Mon Sep 17 00:00:00 2001 From: slzhou Date: Wed, 18 Jan 2023 15:13:16 +0800 Subject: [PATCH 010/124] fix: add node stopped and get/accquire ctx return node stopped --- source/libs/qworker/inc/qwInt.h | 2 ++ source/libs/qworker/src/qwUtil.c | 21 +++++++++++++++++---- source/libs/qworker/src/qworker.c | 3 +++ 3 files changed, 22 insertions(+), 4 deletions(-) diff --git a/source/libs/qworker/inc/qwInt.h b/source/libs/qworker/inc/qwInt.h index aa1ce80903..bde05d4116 100644 --- a/source/libs/qworker/inc/qwInt.h +++ b/source/libs/qworker/inc/qwInt.h @@ -194,6 +194,8 @@ typedef struct SQWorker { SMsgCb msgCb; SQWStat stat; int32_t *destroyed; + + int8_t nodeStopped; } SQWorker; typedef struct SQWorkerMgmt { diff --git a/source/libs/qworker/src/qwUtil.c b/source/libs/qworker/src/qwUtil.c index fdd2775daa..7ee7c50c96 100644 --- a/source/libs/qworker/src/qwUtil.c +++ b/source/libs/qworker/src/qwUtil.c @@ -213,9 +213,15 @@ int32_t qwAcquireTaskCtx(QW_FPARAMS_DEF, SQWTaskCtx **ctx) { QW_SET_QTID(id, qId, tId, eId); *ctx = taosHashAcquire(mgmt->ctxHash, id, sizeof(id)); + int8_t nodeStopped = atomic_load_8(&mgmt->nodeStopped); if (NULL == (*ctx)) { - QW_TASK_DLOG_E("task ctx not exist, may be dropped"); - QW_ERR_RET(TSDB_CODE_QRY_TASK_CTX_NOT_EXIST); + if (!nodeStopped) { + QW_TASK_DLOG_E("task ctx not exist, may be dropped"); + QW_ERR_RET(TSDB_CODE_QRY_TASK_CTX_NOT_EXIST); + } else { + QW_TASK_DLOG_E("node stopped"); + QW_ERR_RET(TSDB_CODE_VND_STOPPED); + } } return TSDB_CODE_SUCCESS; @@ -226,9 +232,16 @@ int32_t qwGetTaskCtx(QW_FPARAMS_DEF, SQWTaskCtx **ctx) { QW_SET_QTID(id, qId, tId, eId); *ctx = taosHashGet(mgmt->ctxHash, id, sizeof(id)); + int8_t nodeStopped = atomic_load_8(&mgmt->nodeStopped); + if (NULL == (*ctx)) { - QW_TASK_DLOG_E("task ctx not exist, may be dropped"); - QW_ERR_RET(TSDB_CODE_QRY_TASK_CTX_NOT_EXIST); + if (!nodeStopped) { + QW_TASK_DLOG_E("task ctx not exist, may be dropped"); + QW_ERR_RET(TSDB_CODE_QRY_TASK_CTX_NOT_EXIST); + } else { + QW_TASK_DLOG_E("node stopped"); + QW_ERR_RET(TSDB_CODE_VND_STOPPED); + } } return TSDB_CODE_SUCCESS; diff --git a/source/libs/qworker/src/qworker.c b/source/libs/qworker/src/qworker.c index e38361d87f..fedaa96ed9 100644 --- a/source/libs/qworker/src/qworker.c +++ b/source/libs/qworker/src/qworker.c @@ -1188,6 +1188,9 @@ void qWorkerStopAllTasks(void *qWorkerMgmt) { uint64_t qId, tId, sId; int32_t eId; int64_t rId = 0; + + atomic_store_8(&mgmt->nodeStopped, 1); + void *pIter = taosHashIterate(mgmt->ctxHash, NULL); while (pIter) { SQWTaskCtx *ctx = (SQWTaskCtx *)pIter; From 1f01ba7cc0882bad4b70720d7326d52f15cefbbb Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Wed, 18 Jan 2023 16:33:52 +0800 Subject: [PATCH 011/124] fix: not keep wal log vnodeCommit'ed for single replica db --- source/libs/sync/src/syncMain.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index eeda612bcd..ebf06a5d8b 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -292,8 +292,6 @@ int32_t syncBeginSnapshot(int64_t rid, int64_t lastApplyIndex) { goto _DEL_WAL; } else { - lastApplyIndex -= SYNC_VNODE_LOG_RETENTION; - SyncIndex beginIndex = pSyncNode->pLogStore->syncLogBeginIndex(pSyncNode->pLogStore); SyncIndex endIndex = pSyncNode->pLogStore->syncLogEndIndex(pSyncNode->pLogStore); bool isEmpty = pSyncNode->pLogStore->syncLogIsEmpty(pSyncNode->pLogStore); @@ -308,6 +306,8 @@ int32_t syncBeginSnapshot(int64_t rid, int64_t lastApplyIndex) { if (pSyncNode->replicaNum > 1) { // multi replicas + lastApplyIndex -= SYNC_VNODE_LOG_RETENTION; + if (pSyncNode->state == TAOS_SYNC_STATE_LEADER) { pSyncNode->minMatchIndex = syncMinMatchIndex(pSyncNode); From 516306b598d1f963b647af8d752c3b228bd6311f Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Wed, 18 Jan 2023 17:28:18 +0800 Subject: [PATCH 012/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 332 +++++---------------- 1 file changed, 78 insertions(+), 254 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index b2574ddadd..563f488f8e 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -83,9 +83,9 @@ static int32_t tsdbOpenDataFileDataIter(SDataFReader* pReader, STsdbDataIter2** code = tBlockDataCreate(&pIter->dIter.bData); TSDB_CHECK_CODE(code, lino, _exit); - pIter->dIter.iBlockIdx = -1; - pIter->dIter.iDataBlk = -1; - pIter->dIter.iRow = -1; + pIter->dIter.iBlockIdx = 0; + pIter->dIter.iDataBlk = 0; + pIter->dIter.iRow = 0; // read data code = tsdbReadBlockIdx(pReader, pIter->dIter.aBlockIdx); @@ -130,8 +130,8 @@ static int32_t tsdbOpenSttFileDataIter(SDataFReader* pReader, int32_t iStt, STsd code = tBlockDataCreate(&pIter->sIter.bData); TSDB_CHECK_CODE(code, lino, _exit); - pIter->sIter.iSttBlk = -1; - pIter->sIter.iRow = -1; + pIter->sIter.iSttBlk = 0; + pIter->sIter.iRow = 0; // read data code = tsdbReadSttBlk(pReader, iStt, pIter->sIter.aSttBlk); @@ -194,34 +194,37 @@ static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter) { int32_t lino = 0; for (;;) { - if (++pIter->dIter.iRow < pIter->dIter.bData.nRow) { + if (pIter->dIter.iRow < pIter->dIter.bData.nRow) { pIter->rowInfo.suid = pIter->dIter.bData.suid; pIter->rowInfo.uid = pIter->dIter.bData.uid; pIter->rowInfo.row = tsdbRowFromBlockData(&pIter->dIter.bData, pIter->dIter.iRow); + pIter->dIter.iRow++; break; } for (;;) { - if (++pIter->dIter.iDataBlk < pIter->dIter.mDataBlk.nItem) { + if (pIter->dIter.iDataBlk < pIter->dIter.mDataBlk.nItem) { SDataBlk dataBlk; tMapDataGetItemByIdx(&pIter->dIter.mDataBlk, pIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); code = tsdbReadDataBlockEx(pIter->dIter.pReader, &dataBlk, &pIter->dIter.bData); TSDB_CHECK_CODE(code, lino, _exit); - pIter->dIter.iRow = -1; + pIter->dIter.iDataBlk++; + pIter->dIter.iRow = 0; break; } for (;;) { - if (++pIter->dIter.iBlockIdx < taosArrayGetSize(pIter->dIter.aBlockIdx)) { + if (pIter->dIter.iBlockIdx < taosArrayGetSize(pIter->dIter.aBlockIdx)) { SBlockIdx* pBlockIdx = taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx); code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); TSDB_CHECK_CODE(code, lino, _exit); - pIter->dIter.iDataBlk = -1; + pIter->dIter.iBlockIdx++; + pIter->dIter.iDataBlk = 0; break; } else { @@ -244,20 +247,23 @@ static int32_t tsdbSttFileDataIterNext(STsdbDataIter2* pIter) { int32_t lino = 0; for (;;) { - if (++pIter->sIter.iRow < pIter->sIter.bData.nRow) { + if (pIter->sIter.iRow < pIter->sIter.bData.nRow) { pIter->rowInfo.suid = pIter->sIter.bData.suid; pIter->rowInfo.uid = pIter->sIter.bData.uid ? pIter->sIter.bData.uid : pIter->sIter.bData.aUid[pIter->sIter.iRow]; pIter->rowInfo.row = tsdbRowFromBlockData(&pIter->sIter.bData, pIter->sIter.iRow); + pIter->sIter.iRow++; break; } - if (++pIter->sIter.iSttBlk < taosArrayGetSize(pIter->sIter.aSttBlk)) { + if (pIter->sIter.iSttBlk < taosArrayGetSize(pIter->sIter.aSttBlk)) { SSttBlk* pSttBlk = taosArrayGet(pIter->sIter.aSttBlk, pIter->sIter.iSttBlk); code = tsdbReadSttBlockEx(pIter->sIter.pReader, pIter->sIter.iStt, pSttBlk, &pIter->sIter.bData); TSDB_CHECK_CODE(code, lino, _exit); - pIter->sIter.iRow = -1; + pIter->sIter.iSttBlk++; + + pIter->sIter.iRow = 0; } else { pIter->rowInfo = (SRowInfo){0}; break; @@ -1019,16 +1025,16 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI if (pWriter->pDIter) { STsdbDataIter2* pIter = pWriter->pDIter; for (;;) { - if (pIter->dIter.iBlockIdx + 1 >= taosArrayGetSize(pIter->dIter.aBlockIdx)) { + if (pIter->dIter.iBlockIdx >= taosArrayGetSize(pIter->dIter.aBlockIdx)) { pWriter->pDIter = NULL; break; } - SBlockIdx* pBlockIdx = (SBlockIdx*)taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx + 1); + SBlockIdx* pBlockIdx = (SBlockIdx*)taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx); int32_t c = tTABLEIDCmprFn(pBlockIdx, pId); if (c < 0) { - ++pIter->dIter.iBlockIdx; + pIter->dIter.iBlockIdx++; code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); TSDB_CHECK_CODE(code, lino, _exit); @@ -1042,15 +1048,15 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI pNewBlockIdx->suid = pBlockIdx->suid; pNewBlockIdx->uid = pBlockIdx->uid; - code = tsdbWriteDataBlk(pWriter->pDataFWriter, &pIter->dIter.mDataBlk, pBlockIdx); + code = tsdbWriteDataBlk(pWriter->pDataFWriter, &pIter->dIter.mDataBlk, pNewBlockIdx); TSDB_CHECK_CODE(code, lino, _exit); } else if (c == 0) { - ++pIter->dIter.iBlockIdx; + pIter->dIter.iBlockIdx++; code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); TSDB_CHECK_CODE(code, lino, _exit); - pIter->dIter.iDataBlk = -1; + pIter->dIter.iDataBlk = 0; break; } else { @@ -1089,58 +1095,38 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { int32_t code = 0; int32_t lino = 0; -#if 0 - if (pWriter->id.suid == 0 && pWriter->id.uid == 0) return code; + // TODO: do write the row, refer line 1593 - int32_t c = 1; - if (pWriter->dReader.pBlockIdx) { - c = tTABLEIDCmprFn(pWriter->dReader.pBlockIdx, &pWriter->id); - ASSERT(c >= 0); - } + if (pWriter->bData.nRow > 0) { + if (pWriter->bData.nRow < pWriter->minRow) { + for (int32_t iRow = 0; iRow < pWriter->bData.nRow; iRow++) { + code = tBlockDataAppendRow(&pWriter->sData, &tsdbRowFromBlockData(&pWriter->bData, iRow), + pWriter->skmTable.pTSchema, pWriter->tbid.uid); + TSDB_CHECK_CODE(code, lino, _exit); - if (c == 0) { - SBlockData* pBData = &pWriter->dWriter.bData; - - for (; pWriter->dReader.iRow < pWriter->dReader.bData.nRow; pWriter->dReader.iRow++) { - TSDBROW row = tsdbRowFromBlockData(&pWriter->dReader.bData, pWriter->dReader.iRow); - - code = tBlockDataAppendRow(pBData, &row, NULL, pWriter->id.uid); - if (code) goto _err; - - if (pBData->nRow >= pWriter->maxRow) { - code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, pBData, &pWriter->dWriter.mDataBlk, pWriter->cmprAlg); - if (code) goto _err; + if (pWriter->sData.nRow >= pWriter->maxRow) { + code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } } + + tBlockDataReset(&pWriter->bData); + } else { + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); } - - code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, pBData, &pWriter->dWriter.mDataBlk, pWriter->cmprAlg); - if (code) goto _err; - - for (; pWriter->dReader.iDataBlk < pWriter->dReader.mDataBlk.nItem; pWriter->dReader.iDataBlk++) { - SDataBlk dataBlk; - tMapDataGetItemByIdx(&pWriter->dReader.mDataBlk, pWriter->dReader.iDataBlk, &dataBlk, tGetDataBlk); - - code = tMapDataPutItem(&pWriter->dWriter.mDataBlk, &dataBlk, tPutDataBlk); - if (code) goto _err; - } - - code = tsdbSnapNextTableData(pWriter); - if (code) goto _err; } - if (pWriter->dWriter.mDataBlk.nItem) { - SBlockIdx blockIdx = {.suid = pWriter->id.suid, .uid = pWriter->id.uid}; - code = tsdbWriteDataBlk(pWriter->dWriter.pWriter, &pWriter->dWriter.mDataBlk, &blockIdx); - - if (taosArrayPush(pWriter->dWriter.aBlockIdx, &blockIdx) == NULL) { + if (pWriter->mDataBlk.nItem) { + SBlockIdx* pBlockIdx = taosArrayReserve(pWriter->aBlockIdx, 1); + if (pBlockIdx == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } - } - pWriter->id.suid = 0; - pWriter->id.uid = 0; -#endif + code = tsdbWriteDataBlk(pWriter->pDataFWriter, &pWriter->mDataBlk, pBlockIdx); + TSDB_CHECK_CODE(code, lino, _exit); + } _exit: if (code) { @@ -1149,7 +1135,7 @@ _exit: return code; } -static int32_t tsdbSnapWriteOpenDataFile(STsdbSnapWriter* pWriter, int32_t fid) { +static int32_t tsdbSnapWriteFileDataStart(STsdbSnapWriter* pWriter, int32_t fid) { int32_t code = 0; int32_t lino = 0; @@ -1244,31 +1230,14 @@ _exit: return code; } -static int32_t tsdbSnapWriteCloseDataFile(STsdbSnapWriter* pWriter) { +static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { int32_t code = 0; int32_t lino = 0; ASSERT(pWriter->pDataFWriter); -#if 0 - // loop write remain data - for (;;) { - SRowInfo* pRowInfo; - - code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); - TSDB_CHECK_CODE(code, lino, _exit); - - if (pRowInfo == NULL) break; - - code = tsdbSnapWriteTableData(pWriter, pRowInfo); - TSDB_CHECK_CODE(code, lino, _exit); - - code = tsdbSnapWriteNextRow(pWriter); - TSDB_CHECK_CODE(code, lino, _exit); - } - - // TODO: write remain data -#endif + code = tsdbSnapWriteTableData(pWriter, NULL /* TODO */); + TSDB_CHECK_CODE(code, lino, _exit); // do file-level updates code = tsdbWriteSttBlk(pWriter->pDataFWriter, pWriter->aSttBlk); @@ -1303,139 +1272,6 @@ _exit: return code; } -static int32_t tsdbSnapWriteToDataFile(STsdbSnapWriter* pWriter, int32_t iRow, int8_t* done) { - int32_t code = 0; - int32_t lino = 0; - -#if 0 - SBlockData* pBData = &pWriter->bData; - TABLEID id = {.suid = pBData->suid, .uid = pBData->uid ? pBData->uid : pBData->aUid[iRow]}; - TSDBROW row = tsdbRowFromBlockData(pBData, iRow); - TSDBKEY key = TSDBROW_KEY(&row); - - *done = 0; - while (pWriter->dReader.iRow < pWriter->dReader.bData.nRow || - pWriter->dReader.iDataBlk < pWriter->dReader.mDataBlk.nItem) { - // Merge row by row - for (; pWriter->dReader.iRow < pWriter->dReader.bData.nRow; pWriter->dReader.iRow++) { - TSDBROW trow = tsdbRowFromBlockData(&pWriter->dReader.bData, pWriter->dReader.iRow); - TSDBKEY tKey = TSDBROW_KEY(&trow); - - ASSERT(pWriter->dReader.bData.suid == id.suid && pWriter->dReader.bData.uid == id.uid); - - int32_t c = tsdbKeyCmprFn(&key, &tKey); - if (c < 0) { - code = tBlockDataAppendRow(&pWriter->dWriter.bData, &row, NULL, id.uid); - TSDB_CHECK_CODE(code, lino, _exit); - } else if (c > 0) { - code = tBlockDataAppendRow(&pWriter->dWriter.bData, &trow, NULL, id.uid); - TSDB_CHECK_CODE(code, lino, _exit); - } else { - ASSERT(0); - } - - if (pWriter->dWriter.bData.nRow >= pWriter->maxRow) { - code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.bData, &pWriter->dWriter.mDataBlk, - pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } - - if (c < 0) { - *done = 1; - goto _exit; - } - } - - // Merge row by block - SDataBlk tDataBlk = {.minKey = key, .maxKey = key}; - for (; pWriter->dReader.iDataBlk < pWriter->dReader.mDataBlk.nItem; pWriter->dReader.iDataBlk++) { - SDataBlk dataBlk; - tMapDataGetItemByIdx(&pWriter->dReader.mDataBlk, pWriter->dReader.iDataBlk, &dataBlk, tGetDataBlk); - - int32_t c = tDataBlkCmprFn(&dataBlk, &tDataBlk); - if (c < 0) { - code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.bData, &pWriter->dWriter.mDataBlk, - pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - - code = tMapDataPutItem(&pWriter->dWriter.mDataBlk, &dataBlk, tPutDataBlk); - TSDB_CHECK_CODE(code, lino, _exit); - } else if (c > 0) { - code = tBlockDataAppendRow(&pWriter->dWriter.bData, &row, NULL, id.uid); - TSDB_CHECK_CODE(code, lino, _exit); - - if (pWriter->dWriter.bData.nRow >= pWriter->maxRow) { - code = tsdbWriteDataBlock(pWriter->dWriter.pWriter, &pWriter->dWriter.bData, &pWriter->dWriter.mDataBlk, - pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } - - *done = 1; - goto _exit; - } else { - code = tsdbReadDataBlockEx(pWriter->dReader.pReader, &dataBlk, &pWriter->dReader.bData); - TSDB_CHECK_CODE(code, lino, _exit); - pWriter->dReader.iRow = 0; - - pWriter->dReader.iDataBlk++; - break; - } - } - } -#endif - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - -static int32_t tsdbSnapWriteToSttFile(STsdbSnapWriter* pWriter, int32_t iRow) { - int32_t code = 0; - int32_t lino = 0; - -#if 0 - TABLEID id = {.suid = pWriter->bData.suid, - .uid = pWriter->bData.uid ? pWriter->bData.uid : pWriter->bData.aUid[iRow]}; - TSDBROW row = tsdbRowFromBlockData(&pWriter->bData, iRow); - SBlockData* pBData = &pWriter->dWriter.sData; - - if (pBData->suid || pBData->uid) { - if (!TABLE_SAME_SCHEMA(pBData->suid, pBData->uid, id.suid, id.uid)) { - code = tsdbWriteSttBlock(pWriter->pDataFWriter, pBData, pWriter->aSttBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - - pBData->suid = 0; - pBData->uid = 0; - } - } - - if (pBData->suid == 0 && pBData->uid == 0) { - code = tsdbUpdateTableSchema(pWriter->pTsdb->pVnode->pMeta, pWriter->id.suid, pWriter->id.uid, &pWriter->skmTable); - TSDB_CHECK_CODE(code, lino, _exit); - - TABLEID tid = {.suid = pWriter->id.suid, .uid = pWriter->id.suid ? 0 : pWriter->id.uid}; - code = tBlockDataInit(pBData, &tid, pWriter->skmTable.pTSchema, NULL, 0); - TSDB_CHECK_CODE(code, lino, _exit); - } - - code = tBlockDataAppendRow(pBData, &row, NULL, id.uid); - TSDB_CHECK_CODE(code, lino, _exit); - - if (pBData->nRow >= pWriter->maxRow) { - code = tsdbWriteSttBlock(pWriter->pDataFWriter, pBData, pWriter->aSttBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } -#endif - -_exit: - if (code) { - tsdbError("vgId:%d %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); - } - return code; -} - static int32_t tsdbSnapWriteNextRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowInfo) { int32_t code = 0; int32_t lino = 0; @@ -1502,20 +1338,6 @@ _exit: return code; } -static int32_t tsdbSnapWriteRowImpl(STsdbSnapWriter* pWriter, SRowInfo* pRowInfo) { - int32_t code = 0; - int32_t lino = 0; - - // TODO - ASSERT(0); - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { int32_t code = 0; int32_t lino = 0; @@ -1551,21 +1373,25 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn } // do write the row - if (pWriter->pDIter == NULL /* || false */) { - goto _write_incoming_row; + if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && + pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); } else { for (;;) { - while (pWriter->pDIter->dIter.iRow + 1 < pWriter->pDIter->dIter.bData.nRow) { - TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow + 1); + while (pWriter->pDIter->dIter.iRow < pWriter->pDIter->dIter.bData.nRow) { + TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow); int32_t c = tsdbRowCmprFn(&pRowInfo->row, &row); if (c < 0) { - goto _write_incoming_row; + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; } else if (c > 0) { - ++pWriter->pDIter->dIter.iRow; - code = tsdbSnapWriteTableRow(pWriter, &row); TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pDIter->dIter.iRow++; } else { ASSERT(0); } @@ -1573,34 +1399,32 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn while (pWriter->pDIter->dIter.iDataBlk < pWriter->pDIter->dIter.mDataBlk.nItem) { SDataBlk dataBlk; - - tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk + 1, &dataBlk, - tGetDataBlk); + tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); int32_t c = tDataBlkCmprFn( &dataBlk, &(SDataBlk){.minKey = TSDBROW_KEY(&pRowInfo->row), .maxKey = TSDBROW_KEY(&pRowInfo->row)}); if (c > 0) { - goto _write_incoming_row; + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; } else if (c < 0) { - ++pWriter->pDIter->dIter.iDataBlk; - tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); - } else { - ++pWriter->pDIter->dIter.iDataBlk; + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); + pWriter->pDIter->dIter.iDataBlk++; + } else { code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); TSDB_CHECK_CODE(code, lino, _exit); - pWriter->pDIter->dIter.iRow = -1; + pWriter->pDIter->dIter.iDataBlk++; + pWriter->pDIter->dIter.iRow = 0; break; } } } } -_write_incoming_row: - code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); - TSDB_CHECK_CODE(code, lino, _exit); - _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); @@ -1621,11 +1445,11 @@ static int32_t tsdbSnapWriteTimeSeriesData(STsdbSnapWriter* pWriter, SSnapDataHd int32_t fid = tsdbKeyFid(pWriter->inData.aTSKEY[0], pWriter->minutes, pWriter->precision); if (pWriter->fid != fid) { if (pWriter->pDataFWriter) { - code = tsdbSnapWriteCloseDataFile(pWriter); + code = tsdbSnapWriteFileDataEnd(pWriter); TSDB_CHECK_CODE(code, lino, _exit); } - code = tsdbSnapWriteOpenDataFile(pWriter, fid); + code = tsdbSnapWriteFileDataStart(pWriter, fid); TSDB_CHECK_CODE(code, lino, _exit); } @@ -1927,7 +1751,7 @@ _exit: int32_t tsdbSnapWriterPrepareClose(STsdbSnapWriter* pWriter) { int32_t code = 0; if (pWriter->pDataFWriter) { - code = tsdbSnapWriteCloseDataFile(pWriter); + code = tsdbSnapWriteFileDataEnd(pWriter); if (code) goto _exit; } @@ -2012,7 +1836,7 @@ int32_t tsdbSnapWrite(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { TSDB_CHECK_CODE(code, lino, _exit); goto _exit; } else if (pWriter->pDataFWriter) { - code = tsdbSnapWriteCloseDataFile(pWriter); + code = tsdbSnapWriteFileDataEnd(pWriter); TSDB_CHECK_CODE(code, lino, _exit); } From 720381988868bb9f5e6364ffe17d8602242b6864 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Thu, 19 Jan 2023 21:46:27 +0800 Subject: [PATCH 013/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 80 ++++++++++++++-------- 1 file changed, 51 insertions(+), 29 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 563f488f8e..5747584d26 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -44,7 +44,6 @@ struct STsdbDataIter2 { int32_t iBlockIdx; int32_t iDataBlk; int32_t iRow; - } dIter; // TSDB_STT_FILE_DATA_ITER @@ -1022,8 +1021,18 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI int32_t code = 0; int32_t lino = 0; + if (pId) { + pWriter->tbid = *pId; + } else { + pWriter->tbid = (TABLEID){INT64_MAX, INT64_MAX}; + } + if (pWriter->pDIter) { STsdbDataIter2* pIter = pWriter->pDIter; + + ASSERT(pIter->dIter.iRow >= pIter->dIter.bData.nRow); + ASSERT(pIter->dIter.iDataBlk >= pIter->dIter.mDataBlk.nItem); + for (;;) { if (pIter->dIter.iBlockIdx >= taosArrayGetSize(pIter->dIter.aBlockIdx)) { pWriter->pDIter = NULL; @@ -1032,10 +1041,8 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI SBlockIdx* pBlockIdx = (SBlockIdx*)taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx); - int32_t c = tTABLEIDCmprFn(pBlockIdx, pId); + int32_t c = tTABLEIDCmprFn(pBlockIdx, &pWriter->tbid); if (c < 0) { - pIter->dIter.iBlockIdx++; - code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); TSDB_CHECK_CODE(code, lino, _exit); @@ -1050,43 +1057,46 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI code = tsdbWriteDataBlk(pWriter->pDataFWriter, &pIter->dIter.mDataBlk, pNewBlockIdx); TSDB_CHECK_CODE(code, lino, _exit); - } else if (c == 0) { - pIter->dIter.iBlockIdx++; + pIter->dIter.iBlockIdx++; + } else if (c == 0) { code = tsdbReadDataBlk(pIter->dIter.pReader, pBlockIdx, &pIter->dIter.mDataBlk); TSDB_CHECK_CODE(code, lino, _exit); pIter->dIter.iDataBlk = 0; + pIter->dIter.iBlockIdx++; break; } else { + pIter->dIter.iDataBlk = pIter->dIter.mDataBlk.nItem; break; } } } - pWriter->tbid = pId[0]; + if (pId == NULL) { + if (pWriter->sData.nRow) { + code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + } else { + code = tsdbUpdateTableSchema(pWriter->pTsdb->pVnode->pMeta, pId->suid, pId->uid, &pWriter->skmTable); + TSDB_CHECK_CODE(code, lino, _exit); - tMapDataReset(&pWriter->mDataBlk); + tMapDataReset(&pWriter->mDataBlk); -#if 0 - code = tsdbSnapWriteCopyData(pWriter, pId); - TSDB_CHECK_CODE(code, lino, _exit); + code = tBlockDataInit(&pWriter->bData, pId, pWriter->skmTable.pTSchema, NULL, 0); + TSDB_CHECK_CODE(code, lino, _exit); - pWriter->id.suid = pId->suid; - pWriter->id.uid = pId->uid; - - code = tsdbUpdateTableSchema(pWriter->pTsdb->pVnode->pMeta, pId->suid, pId->uid, &pWriter->skmTable); - TSDB_CHECK_CODE(code, lino, _exit); - - tMapDataReset(&pWriter->dWriter.mDataBlk); - code = tBlockDataInit(&pWriter->dWriter.bData, pId, pWriter->skmTable.pTSchema, NULL, 0); - TSDB_CHECK_CODE(code, lino, _exit); -#endif + // TODO: init pWriter->sData ?? + } _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbTrace("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, pId->suid, + pId->uid); } return code; } @@ -1219,6 +1229,7 @@ static int32_t tsdbSnapWriteFileDataStart(STsdbSnapWriter* pWriter, int32_t fid) } tBlockDataReset(&pWriter->bData); + tBlockDataReset(&pWriter->sData); _exit: if (code) { @@ -1342,7 +1353,7 @@ static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { int32_t code = 0; int32_t lino = 0; - code = tBlockDataAppendRow(&pWriter->bData, pRow, NULL, pWriter->tbid.uid); + code = tBlockDataAppendRow(&pWriter->bData, pRow, pWriter->skmTable.pTSchema, pWriter->tbid.uid); TSDB_CHECK_CODE(code, lino, _exit); if (pWriter->bData.nRow >= pWriter->maxRow) { @@ -1362,8 +1373,8 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn int32_t lino = 0; // switch to new table if need - if (pRowInfo->uid != pWriter->tbid.uid) { - if (pRowInfo->uid) { + if (pRowInfo == NULL || pRowInfo->uid != pWriter->tbid.uid) { + if (pWriter->tbid.uid != 0) { code = tsdbSnapWriteTableDataEnd(pWriter); TSDB_CHECK_CODE(code, lino, _exit); } @@ -1372,6 +1383,9 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn TSDB_CHECK_CODE(code, lino, _exit); } + // end with a NULL row + if (pRowInfo == NULL) goto _exit; + // do write the row if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { @@ -1397,7 +1411,14 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn } } - while (pWriter->pDIter->dIter.iDataBlk < pWriter->pDIter->dIter.mDataBlk.nItem) { + for (;;) { + if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) { + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; + } + + // FIXME: Here can be slow, use array instead SDataBlk dataBlk; tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); @@ -1408,8 +1429,10 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn TSDB_CHECK_CODE(code, lino, _exit); goto _exit; } else if (c < 0) { - code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); + if (pWriter->bData.nRow > 0) { + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); pWriter->pDIter->dIter.iDataBlk++; @@ -1417,8 +1440,8 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); TSDB_CHECK_CODE(code, lino, _exit); - pWriter->pDIter->dIter.iDataBlk++; pWriter->pDIter->dIter.iRow = 0; + pWriter->pDIter->dIter.iDataBlk++; break; } } @@ -1455,7 +1478,6 @@ static int32_t tsdbSnapWriteTimeSeriesData(STsdbSnapWriter* pWriter, SSnapDataHd // loop write each row SRowInfo* pRowInfo; - code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); TSDB_CHECK_CODE(code, lino, _exit); for (int32_t iRow = 0; iRow < pWriter->bData.nRow; ++iRow) { From fff4a96bdf94a0f8adfafca456d9c5a71b7a7286 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Thu, 19 Jan 2023 22:18:39 +0800 Subject: [PATCH 014/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 160 ++++++++++++--------- 1 file changed, 94 insertions(+), 66 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 5747584d26..310c3de5aa 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1105,7 +1105,9 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { int32_t code = 0; int32_t lino = 0; - // TODO: do write the row, refer line 1593 + // write a NULL row to end current table data write + code = tsdbSnapWriteTableRow(pWriter, NULL); + TSDB_CHECK_CODE(code, lino, _exit); if (pWriter->bData.nRow > 0) { if (pWriter->bData.nRow < pWriter->minRow) { @@ -1120,7 +1122,7 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { } } - tBlockDataReset(&pWriter->bData); + tBlockDataClear(&pWriter->bData); } else { code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); TSDB_CHECK_CODE(code, lino, _exit); @@ -1134,6 +1136,9 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { TSDB_CHECK_CODE(code, lino, _exit); } + pBlockIdx->suid = pWriter->tbid.suid; + pBlockIdx->uid = pWriter->tbid.uid; + code = tsdbWriteDataBlk(pWriter->pDataFWriter, &pWriter->mDataBlk, pBlockIdx); TSDB_CHECK_CODE(code, lino, _exit); } @@ -1247,9 +1252,14 @@ static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { ASSERT(pWriter->pDataFWriter); - code = tsdbSnapWriteTableData(pWriter, NULL /* TODO */); + code = tsdbSnapWriteTableData(pWriter, NULL); TSDB_CHECK_CODE(code, lino, _exit); + if (pWriter->sData.nRow) { + code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + // do file-level updates code = tsdbWriteSttBlk(pWriter->pDataFWriter, pWriter->aSttBlk); TSDB_CHECK_CODE(code, lino, _exit); @@ -1271,8 +1281,12 @@ static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { TSDB_CHECK_CODE(code, lino, _exit); } - // TODO: do clear sources - {} + // clear sources + while (pWriter->iterList) { + STsdbDataIter2* pIter = pWriter->iterList; + pWriter->iterList = pIter->next; + tsdbCloseDataIter2(pIter); + } _exit: if (code) { @@ -1349,7 +1363,7 @@ _exit: return code; } -static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { +static int32_t tsdbSnapWriteTableRowImpl(STsdbSnapWriter* pWriter, TSDBROW* pRow) { int32_t code = 0; int32_t lino = 0; @@ -1368,6 +1382,79 @@ _exit: return code; } +static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { + int32_t code = 0; + int32_t lino = 0; + + // TODO: take pRow == NULL into account + + if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && + pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + } else { + for (;;) { + while (pWriter->pDIter->dIter.iRow < pWriter->pDIter->dIter.bData.nRow) { + TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow); + + int32_t c = tsdbRowCmprFn(pRow, &row); + if (c < 0) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; + } else if (c > 0) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pDIter->dIter.iRow++; + } else { + ASSERT(0); + } + } + + for (;;) { + if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; + } + + // FIXME: Here can be slow, use array instead + SDataBlk dataBlk; + tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); + + int32_t c = tDataBlkCmprFn(&dataBlk, &(SDataBlk){.minKey = TSDBROW_KEY(pRow), .maxKey = TSDBROW_KEY(pRow)}); + if (c > 0) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + goto _exit; + } else if (c < 0) { + if (pWriter->bData.nRow > 0) { + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + + tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); + pWriter->pDIter->dIter.iDataBlk++; + } else { + code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pDIter->dIter.iRow = 0; + pWriter->pDIter->dIter.iDataBlk++; + break; + } + } + } + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowInfo) { int32_t code = 0; int32_t lino = 0; @@ -1384,68 +1471,9 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn } // end with a NULL row - if (pRowInfo == NULL) goto _exit; - - // do write the row - if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && - pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { + if (pRowInfo) { code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); TSDB_CHECK_CODE(code, lino, _exit); - } else { - for (;;) { - while (pWriter->pDIter->dIter.iRow < pWriter->pDIter->dIter.bData.nRow) { - TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow); - - int32_t c = tsdbRowCmprFn(&pRowInfo->row, &row); - if (c < 0) { - code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); - TSDB_CHECK_CODE(code, lino, _exit); - goto _exit; - } else if (c > 0) { - code = tsdbSnapWriteTableRow(pWriter, &row); - TSDB_CHECK_CODE(code, lino, _exit); - - pWriter->pDIter->dIter.iRow++; - } else { - ASSERT(0); - } - } - - for (;;) { - if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) { - code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); - TSDB_CHECK_CODE(code, lino, _exit); - goto _exit; - } - - // FIXME: Here can be slow, use array instead - SDataBlk dataBlk; - tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); - - int32_t c = tDataBlkCmprFn( - &dataBlk, &(SDataBlk){.minKey = TSDBROW_KEY(&pRowInfo->row), .maxKey = TSDBROW_KEY(&pRowInfo->row)}); - if (c > 0) { - code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); - TSDB_CHECK_CODE(code, lino, _exit); - goto _exit; - } else if (c < 0) { - if (pWriter->bData.nRow > 0) { - code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } - - tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); - pWriter->pDIter->dIter.iDataBlk++; - } else { - code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); - TSDB_CHECK_CODE(code, lino, _exit); - - pWriter->pDIter->dIter.iRow = 0; - pWriter->pDIter->dIter.iDataBlk++; - break; - } - } - } } _exit: From fae4f2c4ed96427cfefa25c6609455ed804d1e46 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Fri, 20 Jan 2023 18:08:34 +0800 Subject: [PATCH 015/124] refactor: do some internal refactor. --- source/libs/executor/inc/executorimpl.h | 3 ++- source/libs/executor/src/exchangeoperator.c | 30 ++++++--------------- source/libs/executor/src/executil.c | 4 +++ source/libs/executor/src/projectoperator.c | 29 +++----------------- source/libs/executor/src/scanoperator.c | 16 +++++------ source/libs/executor/src/sortoperator.c | 28 +++++-------------- 6 files changed, 32 insertions(+), 78 deletions(-) diff --git a/source/libs/executor/inc/executorimpl.h b/source/libs/executor/inc/executorimpl.h index c68f7c4697..4ae178d508 100644 --- a/source/libs/executor/inc/executorimpl.h +++ b/source/libs/executor/inc/executorimpl.h @@ -704,9 +704,10 @@ void doBuildResultDatablock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SG SDiskbasedBuf* pBuf); bool hasLimitOffsetInfo(SLimitInfo* pLimitInfo); +bool hasSlimitOffsetInfo(SLimitInfo* pLimitInfo); void initLimitInfo(const SNode* pLimit, const SNode* pSLimit, SLimitInfo* pLimitInfo); void resetLimitInfoForNextGroup(SLimitInfo* pLimitInfo); -bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo, SOperatorInfo* pOperator); +bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo); void applyAggFunctionOnPartialTuples(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, SColumnInfoData* pTimeWindowData, int32_t offset, int32_t forwardStep, int32_t numOfTotal, int32_t numOfOutput); diff --git a/source/libs/executor/src/exchangeoperator.c b/source/libs/executor/src/exchangeoperator.c index 037b33dc9f..08b7d371e2 100644 --- a/source/libs/executor/src/exchangeoperator.c +++ b/source/libs/executor/src/exchangeoperator.c @@ -707,6 +707,8 @@ int32_t prepareLoadRemoteData(SOperatorInfo* pOperator) { } int32_t handleLimitOffset(SOperatorInfo* pOperator, SLimitInfo* pLimitInfo, SSDataBlock* pBlock, bool holdDataInBuf) { + SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo; + if (pLimitInfo->remainGroupOffset > 0) { if (pLimitInfo->currentGroupId == 0) { // it is the first group pLimitInfo->currentGroupId = pBlock->info.id.groupId; @@ -750,36 +752,20 @@ int32_t handleLimitOffset(SOperatorInfo* pOperator, SLimitInfo* pLimitInfo, SSDa // set current group id pLimitInfo->currentGroupId = pBlock->info.id.groupId; - if (pLimitInfo->remainOffset >= pBlock->info.rows) { - pLimitInfo->remainOffset -= pBlock->info.rows; - blockDataCleanup(pBlock); + bool limitReached = applyLimitOffset(pLimitInfo, pBlock, pTaskInfo); + if (pBlock->info.rows == 0) { return PROJECT_RETRIEVE_CONTINUE; - } else if (pLimitInfo->remainOffset < pBlock->info.rows && pLimitInfo->remainOffset > 0) { - blockDataTrimFirstNRows(pBlock, pLimitInfo->remainOffset); - pLimitInfo->remainOffset = 0; - } - - // check for the limitation in each group - if (pLimitInfo->limit.limit >= 0 && pLimitInfo->numOfOutputRows + pBlock->info.rows >= pLimitInfo->limit.limit) { - int32_t keepRows = (int32_t)(pLimitInfo->limit.limit - pLimitInfo->numOfOutputRows); - blockDataKeepFirstNRows(pBlock, keepRows); - if (pLimitInfo->slimit.limit > 0 && pLimitInfo->slimit.limit <= pLimitInfo->numOfOutputGroups) { + } else { + if (limitReached && (pLimitInfo->slimit.limit > 0 && pLimitInfo->slimit.limit <= pLimitInfo->numOfOutputGroups)) { setOperatorCompleted(pOperator); - } else { - // current group limitation is reached, and future blocks of this group need to be discarded. - if (pBlock->info.rows == 0) { - return PROJECT_RETRIEVE_CONTINUE; - } + return PROJECT_RETRIEVE_DONE; } - - return PROJECT_RETRIEVE_DONE; } // todo optimize performance // If there are slimit/soffset value exists, multi-round result can not be packed into one group, since the // they may not belong to the same group the limit/offset value is not valid in this case. - if ((!holdDataInBuf) || (pBlock->info.rows >= pOperator->resultInfo.threshold) || pLimitInfo->slimit.offset != -1 || - pLimitInfo->slimit.limit != -1) { + if ((!holdDataInBuf) || (pBlock->info.rows >= pOperator->resultInfo.threshold) || hasSlimitOffsetInfo(pLimitInfo)) { return PROJECT_RETRIEVE_DONE; } else { // not full enough, continue to accumulate the output data in the buffer. return PROJECT_RETRIEVE_CONTINUE; diff --git a/source/libs/executor/src/executil.c b/source/libs/executor/src/executil.c index 757324a773..92d52fbb0a 100644 --- a/source/libs/executor/src/executil.c +++ b/source/libs/executor/src/executil.c @@ -1749,6 +1749,10 @@ bool hasLimitOffsetInfo(SLimitInfo* pLimitInfo) { pLimitInfo->slimit.offset != -1); } +bool hasSlimitOffsetInfo(SLimitInfo* pLimitInfo) { + return (pLimitInfo->slimit.limit != -1 || pLimitInfo->slimit.offset != -1); +} + void initLimitInfo(const SNode* pLimit, const SNode* pSLimit, SLimitInfo* pLimitInfo) { SLimit limit = {.limit = getLimit(pLimit), .offset = getOffset(pLimit)}; SLimit slimit = {.limit = getLimit(pSLimit), .offset = getOffset(pSLimit)}; diff --git a/source/libs/executor/src/projectoperator.c b/source/libs/executor/src/projectoperator.c index 3e3610827b..b1dc217bf5 100644 --- a/source/libs/executor/src/projectoperator.c +++ b/source/libs/executor/src/projectoperator.c @@ -185,36 +185,15 @@ static int32_t doIngroupLimitOffset(SLimitInfo* pLimitInfo, uint64_t groupId, SS SOperatorInfo* pOperator) { // set current group id pLimitInfo->currentGroupId = groupId; - - if (pLimitInfo->remainOffset >= pBlock->info.rows) { - pLimitInfo->remainOffset -= pBlock->info.rows; - blockDataCleanup(pBlock); + bool limitReached = applyLimitOffset(pLimitInfo, pBlock, pOperator->pTaskInfo); + if (pBlock->info.rows == 0) { return PROJECT_RETRIEVE_CONTINUE; - } else if (pLimitInfo->remainOffset < pBlock->info.rows && pLimitInfo->remainOffset > 0) { - blockDataTrimFirstNRows(pBlock, pLimitInfo->remainOffset); - pLimitInfo->remainOffset = 0; - } - - // check for the limitation in each group - if (pLimitInfo->limit.limit >= 0 && pLimitInfo->numOfOutputRows + pBlock->info.rows >= pLimitInfo->limit.limit) { - int32_t keepRows = (int32_t)(pLimitInfo->limit.limit - pLimitInfo->numOfOutputRows); - blockDataKeepFirstNRows(pBlock, keepRows); - - // TODO: optimize it later when partition by + limit - // all retrieved requirement has been fulfilled, let's finish this - if ((pLimitInfo->slimit.limit == -1 && pLimitInfo->currentGroupId == 0) || - (pLimitInfo->slimit.limit > 0 && pLimitInfo->slimit.limit <= pLimitInfo->numOfOutputGroups)) { + } else { + if (limitReached && (pLimitInfo->slimit.limit > 0 && pLimitInfo->slimit.limit <= pLimitInfo->numOfOutputGroups)) { setOperatorCompleted(pOperator); - } else { - // Even current group is done, there may be many vgroups remain existed, and we need to continue to retrieve data - // from next group. So let's continue this retrieve process - if (keepRows == 0) { - return PROJECT_RETRIEVE_CONTINUE; - } } } - pLimitInfo->numOfOutputRows += pBlock->info.rows; return PROJECT_RETRIEVE_DONE; } diff --git a/source/libs/executor/src/scanoperator.c b/source/libs/executor/src/scanoperator.c index 813763fffa..2813ef3505 100644 --- a/source/libs/executor/src/scanoperator.c +++ b/source/libs/executor/src/scanoperator.c @@ -256,12 +256,11 @@ static void doSetTagColumnData(STableScanBase* pTableScanInfo, SSDataBlock* pBlo } } -// todo handle the slimit info -bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo, SOperatorInfo* pOperator) { +bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo) { SLimit* pLimit = &pLimitInfo->limit; const char* id = GET_TASKID(pTaskInfo); - if (pLimit->offset > 0 && pLimitInfo->remainOffset > 0) { + if (pLimitInfo->remainOffset > 0) { if (pLimitInfo->remainOffset >= pBlock->info.rows) { pLimitInfo->remainOffset -= pBlock->info.rows; blockDataEmpty(pBlock); @@ -276,12 +275,14 @@ bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo if (pLimit->limit != -1 && pLimit->limit <= (pLimitInfo->numOfOutputRows + pBlock->info.rows)) { // limit the output rows int32_t keep = (int32_t)(pLimit->limit - pLimitInfo->numOfOutputRows); - blockDataKeepFirstNRows(pBlock, keep); + + pLimitInfo->numOfOutputRows += pBlock->info.rows; qDebug("output limit %" PRId64 " has reached, %s", pLimit->limit, id); return true; } + pLimitInfo->numOfOutputRows += pBlock->info.rows; return false; } @@ -393,13 +394,12 @@ static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanBase* pTableSca } } - bool limitReached = applyLimitOffset(&pTableScanInfo->limitInfo, pBlock, pTaskInfo, pOperator); + bool limitReached = applyLimitOffset(&pTableScanInfo->limitInfo, pBlock, pTaskInfo); if (limitReached) { // set operator flag is done setOperatorCompleted(pOperator); } pCost->totalRows += pBlock->info.rows; - pTableScanInfo->limitInfo.numOfOutputRows = pCost->totalRows; return TSDB_CODE_SUCCESS; } @@ -2714,9 +2714,7 @@ SSDataBlock* getSortedTableMergeScanBlockData(SSortHandle* pHandle, SSDataBlock* } } - applyLimitOffset(&pInfo->limitInfo, pResBlock, pTaskInfo, pOperator); - pInfo->limitInfo.numOfOutputRows += pResBlock->info.rows; - + applyLimitOffset(&pInfo->limitInfo, pResBlock, pTaskInfo); qDebug("%s get sorted row block, rows:%d, limit:%"PRId64, GET_TASKID(pTaskInfo), pResBlock->info.rows, pInfo->limitInfo.numOfOutputRows); diff --git a/source/libs/executor/src/sortoperator.c b/source/libs/executor/src/sortoperator.c index 97b4fd9dc4..e91d41897d 100644 --- a/source/libs/executor/src/sortoperator.c +++ b/source/libs/executor/src/sortoperator.c @@ -222,6 +222,7 @@ SSDataBlock* doSort(SOperatorInfo* pOperator) { T_LONG_JMP(pTaskInfo->env, code); } + // multi-group case not handle here SSDataBlock* pBlock = NULL; while (1) { pBlock = getSortedBlockData(pInfo->pSortHandle, pInfo->binfo.pRes, pOperator->resultInfo.capacity, @@ -236,28 +237,13 @@ SSDataBlock* doSort(SOperatorInfo* pOperator) { continue; } - // todo add the limit/offset info - if (pInfo->limitInfo.remainOffset > 0) { - if (pInfo->limitInfo.remainOffset >= blockDataGetNumOfRows(pBlock)) { - pInfo->limitInfo.remainOffset -= pBlock->info.rows; - continue; - } - - blockDataTrimFirstNRows(pBlock, pInfo->limitInfo.remainOffset); - pInfo->limitInfo.remainOffset = 0; + bool limitReached = applyLimitOffset(&pInfo->limitInfo, pBlock, pTaskInfo); + if (limitReached) { + resetLimitInfoForNextGroup(&pInfo->limitInfo); } - if (pInfo->limitInfo.limit.limit > 0 && - pInfo->limitInfo.limit.limit <= pInfo->limitInfo.numOfOutputRows + blockDataGetNumOfRows(pBlock)) { - int32_t remain = pInfo->limitInfo.limit.limit - pInfo->limitInfo.numOfOutputRows; - blockDataKeepFirstNRows(pBlock, remain); - } - - size_t numOfRows = blockDataGetNumOfRows(pBlock); - pInfo->limitInfo.numOfOutputRows += numOfRows; - pOperator->resultInfo.totalRows += numOfRows; - - if (numOfRows > 0) { + pOperator->resultInfo.totalRows += pBlock->info.rows; + if (pBlock->info.rows > 0) { break; } } @@ -680,7 +666,7 @@ SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pData break; } - bool limitReached = applyLimitOffset(&pInfo->limitInfo, p, pTaskInfo, pOperator); + bool limitReached = applyLimitOffset(&pInfo->limitInfo, p, pTaskInfo); if (limitReached) { resetLimitInfoForNextGroup(&pInfo->limitInfo); } From 7803104b7e31ab9283b29aa33ec391804b721123 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Fri, 20 Jan 2023 22:50:35 +0800 Subject: [PATCH 016/124] fix(query): do some internal refactor, and identify a bug. --- source/common/src/tdatablock.c | 1 + source/libs/executor/src/projectoperator.c | 11 ++++++++++- source/libs/executor/src/sortoperator.c | 2 ++ 3 files changed, 13 insertions(+), 1 deletion(-) diff --git a/source/common/src/tdatablock.c b/source/common/src/tdatablock.c index 43f272d599..f41eb1adaf 100644 --- a/source/common/src/tdatablock.c +++ b/source/common/src/tdatablock.c @@ -1431,6 +1431,7 @@ SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData) { pBlock->info.rows = 0; pBlock->info.capacity = 0; pBlock->info.rowSize = 0; + pBlock->info.id = pDataBlock->info.id; size_t numOfCols = taosArrayGetSize(pDataBlock->pDataBlock); for (int32_t i = 0; i < numOfCols; ++i) { diff --git a/source/libs/executor/src/projectoperator.c b/source/libs/executor/src/projectoperator.c index b1dc217bf5..d641810cee 100644 --- a/source/libs/executor/src/projectoperator.c +++ b/source/libs/executor/src/projectoperator.c @@ -90,7 +90,16 @@ SOperatorInfo* createProjectOperatorInfo(SOperatorInfo* downstream, SProjectPhys pInfo->binfo.pRes = pResBlock; pInfo->pFinalRes = createOneDataBlock(pResBlock, false); - pInfo->mergeDataBlocks = (pTaskInfo->execModel == OPTR_EXEC_MODEL_STREAM) ? false : pProjPhyNode->mergeDataBlock; + + if (pTaskInfo->execModel == OPTR_EXEC_MODEL_STREAM) { + pInfo->mergeDataBlocks = false; + } else { + if (!pProjPhyNode->ignoreGroupId) { + pInfo->mergeDataBlocks = false; + } else { + pInfo->mergeDataBlocks = pProjPhyNode->mergeDataBlock; + } + } int32_t numOfRows = 4096; size_t keyBufSize = sizeof(int64_t) + sizeof(int64_t) + POINTER_BYTES; diff --git a/source/libs/executor/src/sortoperator.c b/source/libs/executor/src/sortoperator.c index e91d41897d..6d3da3e111 100644 --- a/source/libs/executor/src/sortoperator.c +++ b/source/libs/executor/src/sortoperator.c @@ -237,6 +237,7 @@ SSDataBlock* doSort(SOperatorInfo* pOperator) { continue; } + // there are bugs? bool limitReached = applyLimitOffset(&pInfo->limitInfo, pBlock, pTaskInfo); if (limitReached) { resetLimitInfoForNextGroup(&pInfo->limitInfo); @@ -666,6 +667,7 @@ SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pData break; } + // todo fix it: we need to decide whether this block is belonged to previous group or not . bool limitReached = applyLimitOffset(&pInfo->limitInfo, p, pTaskInfo); if (limitReached) { resetLimitInfoForNextGroup(&pInfo->limitInfo); From a898be4f7d1bc5fb9c528353d4c2ddf039c86b1e Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Fri, 20 Jan 2023 23:38:31 +0800 Subject: [PATCH 017/124] fix(query): set correct total rsp rows. --- source/libs/executor/src/exchangeoperator.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/source/libs/executor/src/exchangeoperator.c b/source/libs/executor/src/exchangeoperator.c index 08b7d371e2..e5089ab4a9 100644 --- a/source/libs/executor/src/exchangeoperator.c +++ b/source/libs/executor/src/exchangeoperator.c @@ -218,10 +218,7 @@ static SSDataBlock* loadRemoteData(SOperatorInfo* pOperator) { if (status == PROJECT_RETRIEVE_CONTINUE) { continue; } else if (status == PROJECT_RETRIEVE_DONE) { - size_t rows = pBlock->info.rows; - pExchangeInfo->limitInfo.numOfOutputRows += rows; - - if (rows == 0) { + if (pBlock->info.rows == 0) { setOperatorCompleted(pOperator); return NULL; } else { From 4f0a605cc08ab3d50597e8b98b40a4a64f3dbe81 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sun, 22 Jan 2023 17:20:26 +0800 Subject: [PATCH 018/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 310c3de5aa..caf3463fdc 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1255,6 +1255,8 @@ static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { code = tsdbSnapWriteTableData(pWriter, NULL); TSDB_CHECK_CODE(code, lino, _exit); + // TODO: ?? + if (pWriter->sData.nRow) { code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); TSDB_CHECK_CODE(code, lino, _exit); From a902d412eb002b43002a12f5d45bf4414a0f58c3 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Thu, 26 Jan 2023 20:20:39 +0800 Subject: [PATCH 019/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 38 ++++++++++------------ source/dnode/vnode/src/tsdb/tsdbUtil.c | 2 +- 2 files changed, 19 insertions(+), 21 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index caf3463fdc..387f5e1763 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1030,6 +1030,7 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI if (pWriter->pDIter) { STsdbDataIter2* pIter = pWriter->pDIter; + // assert last table data end ASSERT(pIter->dIter.iRow >= pIter->dIter.bData.nRow); ASSERT(pIter->dIter.iDataBlk >= pIter->dIter.mDataBlk.nItem); @@ -1095,8 +1096,8 @@ _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); } else { - tsdbTrace("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, pId->suid, - pId->uid); + tsdbTrace("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, + pWriter->tbid.suid, pWriter->tbid.uid); } return code; } @@ -1388,24 +1389,21 @@ static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { int32_t code = 0; int32_t lino = 0; - // TODO: take pRow == NULL into account + TSDBKEY inKey = pRow ? TSDBROW_KEY(pRow) : TSDBKEY_MAX; if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { - code = tsdbSnapWriteTableRowImpl(pWriter, pRow); - TSDB_CHECK_CODE(code, lino, _exit); + goto _write_row; } else { for (;;) { while (pWriter->pDIter->dIter.iRow < pWriter->pDIter->dIter.bData.nRow) { TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow); - int32_t c = tsdbRowCmprFn(pRow, &row); + int32_t c = tsdbKeyCmprFn(&inKey, &TSDBROW_KEY(&row)); if (c < 0) { - code = tsdbSnapWriteTableRowImpl(pWriter, pRow); - TSDB_CHECK_CODE(code, lino, _exit); - goto _exit; + goto _write_row; } else if (c > 0) { - code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + code = tsdbSnapWriteTableRowImpl(pWriter, &row); TSDB_CHECK_CODE(code, lino, _exit); pWriter->pDIter->dIter.iRow++; @@ -1415,21 +1413,15 @@ static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { } for (;;) { - if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) { - code = tsdbSnapWriteTableRowImpl(pWriter, pRow); - TSDB_CHECK_CODE(code, lino, _exit); - goto _exit; - } + if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) goto _write_row; // FIXME: Here can be slow, use array instead SDataBlk dataBlk; tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); - int32_t c = tDataBlkCmprFn(&dataBlk, &(SDataBlk){.minKey = TSDBROW_KEY(pRow), .maxKey = TSDBROW_KEY(pRow)}); + int32_t c = tDataBlkCmprFn(&dataBlk, &(SDataBlk){.minKey = inKey, .maxKey = inKey}); if (c > 0) { - code = tsdbSnapWriteTableRowImpl(pWriter, pRow); - TSDB_CHECK_CODE(code, lino, _exit); - goto _exit; + goto _write_row; } else if (c < 0) { if (pWriter->bData.nRow > 0) { code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); @@ -1450,6 +1442,12 @@ static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { } } +_write_row: + if (pRow) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + } + _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); @@ -1510,7 +1508,7 @@ static int32_t tsdbSnapWriteTimeSeriesData(STsdbSnapWriter* pWriter, SSnapDataHd SRowInfo* pRowInfo; code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); TSDB_CHECK_CODE(code, lino, _exit); - for (int32_t iRow = 0; iRow < pWriter->bData.nRow; ++iRow) { + for (int32_t iRow = 0; iRow < pWriter->inData.nRow; ++iRow) { SRowInfo rInfo = {.suid = pWriter->inData.suid, .uid = pWriter->inData.uid ? pWriter->inData.uid : pWriter->inData.aUid[iRow], .row = tsdbRowFromBlockData(&pWriter->inData, iRow)}; diff --git a/source/dnode/vnode/src/tsdb/tsdbUtil.c b/source/dnode/vnode/src/tsdb/tsdbUtil.c index 4e6adb2dbf..e2d4b92836 100644 --- a/source/dnode/vnode/src/tsdb/tsdbUtil.c +++ b/source/dnode/vnode/src/tsdb/tsdbUtil.c @@ -757,7 +757,7 @@ int32_t tRowMergerAdd(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) { pTColVal->value.nData = pColVal->value.nData; if (pTColVal->value.nData) { - memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData); + memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData); } pTColVal->flag = 0; } else { From ec886154a4d4f564fbfa34ecc8fa2643bcf35c24 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Thu, 26 Jan 2023 21:48:54 +0800 Subject: [PATCH 020/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 387f5e1763..ebb4986e23 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1075,12 +1075,7 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI } } - if (pId == NULL) { - if (pWriter->sData.nRow) { - code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } - } else { + if (pId) { code = tsdbUpdateTableSchema(pWriter->pTsdb->pVnode->pMeta, pId->suid, pId->uid, &pWriter->skmTable); TSDB_CHECK_CODE(code, lino, _exit); @@ -1088,8 +1083,17 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI code = tBlockDataInit(&pWriter->bData, pId, pWriter->skmTable.pTSchema, NULL, 0); TSDB_CHECK_CODE(code, lino, _exit); + } - // TODO: init pWriter->sData ?? + if (!TABLE_SAME_SCHEMA(pWriter->tbid.suid, pWriter->tbid.uid, pWriter->sData.suid, pWriter->sData.uid)) { + if ((pWriter->sData.nRow > 0)) { + code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + + TABLEID id = {.suid = pWriter->tbid.suid, .uid = pWriter->tbid.suid ? 0 : pWriter->tbid.uid}; + code = tBlockDataInit(&pWriter->sData, &id, pWriter->skmTable.pTSchema, NULL, 0); + TSDB_CHECK_CODE(code, lino, _exit); } _exit: From 2f5eeb8000f4778d07537ac8b1e9837207692ade Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Thu, 26 Jan 2023 21:57:22 +0800 Subject: [PATCH 021/124] make it compile --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 252 ++++++++++----------- 1 file changed, 122 insertions(+), 130 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index ebb4986e23..46ffcefe08 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1106,6 +1106,95 @@ _exit: return code; } +static int32_t tsdbSnapWriteTableRowImpl(STsdbSnapWriter* pWriter, TSDBROW* pRow) { + int32_t code = 0; + int32_t lino = 0; + + code = tBlockDataAppendRow(&pWriter->bData, pRow, pWriter->skmTable.pTSchema, pWriter->tbid.uid); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->bData.nRow >= pWriter->maxRow) { + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { + int32_t code = 0; + int32_t lino = 0; + + TSDBKEY inKey = pRow ? TSDBROW_KEY(pRow) : TSDBKEY_MAX; + + if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && + pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { + goto _write_row; + } else { + for (;;) { + while (pWriter->pDIter->dIter.iRow < pWriter->pDIter->dIter.bData.nRow) { + TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow); + + int32_t c = tsdbKeyCmprFn(&inKey, &TSDBROW_KEY(&row)); + if (c < 0) { + goto _write_row; + } else if (c > 0) { + code = tsdbSnapWriteTableRowImpl(pWriter, &row); + TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pDIter->dIter.iRow++; + } else { + ASSERT(0); + } + } + + for (;;) { + if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) goto _write_row; + + // FIXME: Here can be slow, use array instead + SDataBlk dataBlk; + tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); + + int32_t c = tDataBlkCmprFn(&dataBlk, &(SDataBlk){.minKey = inKey, .maxKey = inKey}); + if (c > 0) { + goto _write_row; + } else if (c < 0) { + if (pWriter->bData.nRow > 0) { + code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); + TSDB_CHECK_CODE(code, lino, _exit); + } + + tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); + pWriter->pDIter->dIter.iDataBlk++; + } else { + code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pDIter->dIter.iRow = 0; + pWriter->pDIter->dIter.iDataBlk++; + break; + } + } + } + } + +_write_row: + if (pRow) { + code = tsdbSnapWriteTableRowImpl(pWriter, pRow); + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { int32_t code = 0; int32_t lino = 0; @@ -1251,6 +1340,34 @@ _exit: return code; } +static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowInfo) { + int32_t code = 0; + int32_t lino = 0; + + // switch to new table if need + if (pRowInfo == NULL || pRowInfo->uid != pWriter->tbid.uid) { + if (pWriter->tbid.uid != 0) { + code = tsdbSnapWriteTableDataEnd(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tsdbSnapWriteTableDataStart(pWriter, (TABLEID*)pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // end with a NULL row + if (pRowInfo) { + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { int32_t code = 0; int32_t lino = 0; @@ -1370,123 +1487,6 @@ _exit: return code; } -static int32_t tsdbSnapWriteTableRowImpl(STsdbSnapWriter* pWriter, TSDBROW* pRow) { - int32_t code = 0; - int32_t lino = 0; - - code = tBlockDataAppendRow(&pWriter->bData, pRow, pWriter->skmTable.pTSchema, pWriter->tbid.uid); - TSDB_CHECK_CODE(code, lino, _exit); - - if (pWriter->bData.nRow >= pWriter->maxRow) { - code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - -static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { - int32_t code = 0; - int32_t lino = 0; - - TSDBKEY inKey = pRow ? TSDBROW_KEY(pRow) : TSDBKEY_MAX; - - if (pWriter->pDIter == NULL || (pWriter->pDIter->dIter.iRow >= pWriter->pDIter->dIter.bData.nRow && - pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem)) { - goto _write_row; - } else { - for (;;) { - while (pWriter->pDIter->dIter.iRow < pWriter->pDIter->dIter.bData.nRow) { - TSDBROW row = tsdbRowFromBlockData(&pWriter->pDIter->dIter.bData, pWriter->pDIter->dIter.iRow); - - int32_t c = tsdbKeyCmprFn(&inKey, &TSDBROW_KEY(&row)); - if (c < 0) { - goto _write_row; - } else if (c > 0) { - code = tsdbSnapWriteTableRowImpl(pWriter, &row); - TSDB_CHECK_CODE(code, lino, _exit); - - pWriter->pDIter->dIter.iRow++; - } else { - ASSERT(0); - } - } - - for (;;) { - if (pWriter->pDIter->dIter.iDataBlk >= pWriter->pDIter->dIter.mDataBlk.nItem) goto _write_row; - - // FIXME: Here can be slow, use array instead - SDataBlk dataBlk; - tMapDataGetItemByIdx(&pWriter->pDIter->dIter.mDataBlk, pWriter->pDIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); - - int32_t c = tDataBlkCmprFn(&dataBlk, &(SDataBlk){.minKey = inKey, .maxKey = inKey}); - if (c > 0) { - goto _write_row; - } else if (c < 0) { - if (pWriter->bData.nRow > 0) { - code = tsdbWriteDataBlock(pWriter->pDataFWriter, &pWriter->bData, &pWriter->mDataBlk, pWriter->cmprAlg); - TSDB_CHECK_CODE(code, lino, _exit); - } - - tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); - pWriter->pDIter->dIter.iDataBlk++; - } else { - code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); - TSDB_CHECK_CODE(code, lino, _exit); - - pWriter->pDIter->dIter.iRow = 0; - pWriter->pDIter->dIter.iDataBlk++; - break; - } - } - } - } - -_write_row: - if (pRow) { - code = tsdbSnapWriteTableRowImpl(pWriter, pRow); - TSDB_CHECK_CODE(code, lino, _exit); - } - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - -static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowInfo) { - int32_t code = 0; - int32_t lino = 0; - - // switch to new table if need - if (pRowInfo == NULL || pRowInfo->uid != pWriter->tbid.uid) { - if (pWriter->tbid.uid != 0) { - code = tsdbSnapWriteTableDataEnd(pWriter); - TSDB_CHECK_CODE(code, lino, _exit); - } - - code = tsdbSnapWriteTableDataStart(pWriter, (TABLEID*)pRowInfo); - TSDB_CHECK_CODE(code, lino, _exit); - } - - // end with a NULL row - if (pRowInfo) { - code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); - TSDB_CHECK_CODE(code, lino, _exit); - } - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - static int32_t tsdbSnapWriteTimeSeriesData(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { int32_t code = 0; int32_t lino = 0; @@ -1849,21 +1849,13 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { taosArrayDestroy(pWriter->aDelIdxR); // SNAP_DATA_TSDB - - // // Writer - // tBlockDataDestroy(&pWriter->dWriter.sData, 1); - // tBlockDataDestroy(&pWriter->dWriter.bData, 1); - // taosArrayDestroy(pWriter->dWriter.aSttBlk); - // tMapDataClear(&pWriter->dWriter.mDataBlk); - // taosArrayDestroy(pWriter->dWriter.aBlockIdx); - - // // Reader - // tBlockDataDestroy(&pWriter->dReader.bData, 1); - // tMapDataClear(&pWriter->dReader.mDataBlk); - // taosArrayDestroy(pWriter->dReader.aBlockIdx); - + tBlockDataDestroy(&pWriter->sData, 1); tBlockDataDestroy(&pWriter->bData, 1); + taosArrayDestroy(pWriter->aSttBlk); + tMapDataClear(&pWriter->mDataBlk); + taosArrayDestroy(pWriter->aBlockIdx); tDestroyTSchema(pWriter->skmTable.pTSchema); + tBlockDataDestroy(&pWriter->inData, 1); for (int32_t iBuf = 0; iBuf < sizeof(pWriter->aBuf) / sizeof(uint8_t*); iBuf++) { tFree(pWriter->aBuf[iBuf]); From ec1592390af6c346c0d56e750f12fcdc5cd6f247 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Sat, 28 Jan 2023 10:35:21 +0800 Subject: [PATCH 022/124] fix: limit push down error --- source/libs/planner/src/planSpliter.c | 32 +++++++++++++++++---------- 1 file changed, 20 insertions(+), 12 deletions(-) diff --git a/source/libs/planner/src/planSpliter.c b/source/libs/planner/src/planSpliter.c index a7eac2c853..d85e4ca10d 100644 --- a/source/libs/planner/src/planSpliter.c +++ b/source/libs/planner/src/planSpliter.c @@ -348,7 +348,8 @@ static bool stbSplIsPartTableAgg(SAggLogicNode* pAgg) { return false; } if (NULL != pAgg->pGroupKeys) { - return stbSplHasPartTbname(pAgg->pGroupKeys) && stbSplNotSystemScan((SLogicNode*)nodesListGetNode(pAgg->node.pChildren, 0)); + return stbSplHasPartTbname(pAgg->pGroupKeys) && + stbSplNotSystemScan((SLogicNode*)nodesListGetNode(pAgg->node.pChildren, 0)); } return stbSplHasPartTbname(stbSplGetPartKeys((SLogicNode*)nodesListGetNode(pAgg->node.pChildren, 0))); } @@ -1025,21 +1026,29 @@ static int32_t stbSplSplitSortNode(SSplitContext* pCxt, SStableSplitInfo* pInfo) return code; } -static int32_t stbSplSplitScanNodeWithoutPartTags(SSplitContext* pCxt, SStableSplitInfo* pInfo) { - SLogicNode* pSplitNode = pInfo->pSplitNode; +static int32_t stbSplGetSplitNodeForScan(SStableSplitInfo* pInfo, SLogicNode** pSplitNode) { + *pSplitNode = pInfo->pSplitNode; if (NULL != pInfo->pSplitNode->pParent && QUERY_NODE_LOGIC_PLAN_PROJECT == nodeType(pInfo->pSplitNode->pParent) && NULL == pInfo->pSplitNode->pParent->pLimit && NULL == pInfo->pSplitNode->pParent->pSlimit) { - pSplitNode = pInfo->pSplitNode->pParent; + *pSplitNode = pInfo->pSplitNode->pParent; if (NULL != pInfo->pSplitNode->pLimit) { - pSplitNode->pLimit = nodesCloneNode(pInfo->pSplitNode->pLimit); - if (NULL == pSplitNode->pLimit) { + (*pSplitNode)->pLimit = nodesCloneNode(pInfo->pSplitNode->pLimit); + if (NULL == (*pSplitNode)->pLimit) { return TSDB_CODE_OUT_OF_MEMORY; } ((SLimitNode*)pInfo->pSplitNode->pLimit)->limit += ((SLimitNode*)pInfo->pSplitNode->pLimit)->offset; ((SLimitNode*)pInfo->pSplitNode->pLimit)->offset = 0; } } - int32_t code = splCreateExchangeNodeForSubplan(pCxt, pInfo->pSubplan, pSplitNode, SUBPLAN_TYPE_MERGE); + return TSDB_CODE_SUCCESS; +} + +static int32_t stbSplSplitScanNodeWithoutPartTags(SSplitContext* pCxt, SStableSplitInfo* pInfo) { + SLogicNode* pSplitNode = NULL; + int32_t code = stbSplGetSplitNodeForScan(pInfo, &pSplitNode); + if (TSDB_CODE_SUCCESS == code) { + code = splCreateExchangeNodeForSubplan(pCxt, pInfo->pSubplan, pSplitNode, SUBPLAN_TYPE_MERGE); + } if (TSDB_CODE_SUCCESS == code) { code = nodesListMakeStrictAppend(&pInfo->pSubplan->pChildren, (SNode*)splCreateScanSubplan(pCxt, pSplitNode, SPLIT_FLAG_STABLE_SPLIT)); @@ -1049,12 +1058,11 @@ static int32_t stbSplSplitScanNodeWithoutPartTags(SSplitContext* pCxt, SStableSp } static int32_t stbSplSplitScanNodeWithPartTags(SSplitContext* pCxt, SStableSplitInfo* pInfo) { - SLogicNode* pSplitNode = pInfo->pSplitNode; - if (NULL != pInfo->pSplitNode->pParent && QUERY_NODE_LOGIC_PLAN_PROJECT == nodeType(pInfo->pSplitNode->pParent) && - NULL == pInfo->pSplitNode->pParent->pLimit && NULL == pInfo->pSplitNode->pParent->pSlimit) { - pSplitNode = pInfo->pSplitNode->pParent; + SLogicNode* pSplitNode = NULL; + int32_t code = stbSplGetSplitNodeForScan(pInfo, &pSplitNode); + if (TSDB_CODE_SUCCESS == code) { + code = stbSplCreateMergeNode(pCxt, pInfo->pSubplan, pSplitNode, NULL, pSplitNode, true); } - int32_t code = stbSplCreateMergeNode(pCxt, pInfo->pSubplan, pSplitNode, NULL, pSplitNode, true); if (TSDB_CODE_SUCCESS == code) { code = nodesListMakeStrictAppend(&pInfo->pSubplan->pChildren, (SNode*)splCreateScanSubplan(pCxt, pSplitNode, SPLIT_FLAG_STABLE_SPLIT)); From beb3de8530ddd38e29efdc094badaf46158e360b Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Sat, 28 Jan 2023 10:53:43 +0800 Subject: [PATCH 023/124] fix: limit push down error --- source/libs/planner/src/planSpliter.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/source/libs/planner/src/planSpliter.c b/source/libs/planner/src/planSpliter.c index d85e4ca10d..4c8b996a75 100644 --- a/source/libs/planner/src/planSpliter.c +++ b/source/libs/planner/src/planSpliter.c @@ -560,6 +560,8 @@ static int32_t stbSplCreateMergeNode(SSplitContext* pCxt, SLogicSubplan* pSubpla if (NULL == pMerge->node.pLimit) { code = TSDB_CODE_OUT_OF_MEMORY; } + ((SLimitNode*)pSplitNode->pLimit)->limit += ((SLimitNode*)pSplitNode->pLimit)->offset; + ((SLimitNode*)pSplitNode->pLimit)->offset = 0; } if (TSDB_CODE_SUCCESS == code) { if (NULL == pSubplan) { From 8258c68b6dd76e7c14b50ceb3f4fde4183ecb79f Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Sat, 28 Jan 2023 11:11:37 +0800 Subject: [PATCH 024/124] fix: subquery output ignores group id --- source/libs/planner/src/planLogicCreater.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/planner/src/planLogicCreater.c b/source/libs/planner/src/planLogicCreater.c index 084d99cae5..bd1823a770 100644 --- a/source/libs/planner/src/planLogicCreater.c +++ b/source/libs/planner/src/planLogicCreater.c @@ -1016,7 +1016,7 @@ static int32_t createProjectLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSel TSWAP(pProject->node.pLimit, pSelect->pLimit); TSWAP(pProject->node.pSlimit, pSelect->pSlimit); - pProject->ignoreGroupId = (NULL == pSelect->pPartitionByList); + pProject->ignoreGroupId = pSelect->isSubquery ? true : (NULL == pSelect->pPartitionByList); pProject->node.groupAction = (!pSelect->isSubquery && pCxt->pPlanCxt->streamQuery) ? GROUP_ACTION_KEEP : GROUP_ACTION_CLEAR; pProject->node.requireDataOrder = DATA_ORDER_LEVEL_NONE; From 8d5630651780891ef19b6be9762c1f2e8036bc86 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sat, 28 Jan 2023 11:31:32 +0800 Subject: [PATCH 025/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 97 ++++++---------------- 1 file changed, 24 insertions(+), 73 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 46ffcefe08..0a04d09a2c 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1049,7 +1049,7 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI SBlockIdx* pNewBlockIdx = taosArrayReserve(pWriter->aBlockIdx, 1); if (pNewBlockIdx == NULL) { - code == TSDB_CODE_OUT_OF_MEMORY; + code = TSDB_CODE_OUT_OF_MEMORY; TSDB_CHECK_CODE(code, lino, _exit); } @@ -1091,9 +1091,11 @@ static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pI TSDB_CHECK_CODE(code, lino, _exit); } - TABLEID id = {.suid = pWriter->tbid.suid, .uid = pWriter->tbid.suid ? 0 : pWriter->tbid.uid}; - code = tBlockDataInit(&pWriter->sData, &id, pWriter->skmTable.pTSchema, NULL, 0); - TSDB_CHECK_CODE(code, lino, _exit); + if (pId) { + TABLEID id = {.suid = pWriter->tbid.suid, .uid = pWriter->tbid.suid ? 0 : pWriter->tbid.uid}; + code = tBlockDataInit(&pWriter->sData, &id, pWriter->skmTable.pTSchema, NULL, 0); + TSDB_CHECK_CODE(code, lino, _exit); + } } _exit: @@ -1206,8 +1208,8 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { if (pWriter->bData.nRow > 0) { if (pWriter->bData.nRow < pWriter->minRow) { for (int32_t iRow = 0; iRow < pWriter->bData.nRow; iRow++) { - code = tBlockDataAppendRow(&pWriter->sData, &tsdbRowFromBlockData(&pWriter->bData, iRow), - pWriter->skmTable.pTSchema, pWriter->tbid.uid); + code = + tBlockDataAppendRow(&pWriter->sData, &tsdbRowFromBlockData(&pWriter->bData, iRow), NULL, pWriter->tbid.uid); TSDB_CHECK_CODE(code, lino, _exit); if (pWriter->sData.nRow >= pWriter->maxRow) { @@ -1374,13 +1376,17 @@ static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { ASSERT(pWriter->pDataFWriter); - code = tsdbSnapWriteTableData(pWriter, NULL); + // consume remain data and end with a NULL table row + SRowInfo* pRowInfo; + code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); TSDB_CHECK_CODE(code, lino, _exit); + for (;;) { + code = tsdbSnapWriteTableData(pWriter, pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); - // TODO: ?? + if (pRowInfo == NULL) break; - if (pWriter->sData.nRow) { - code = tsdbWriteSttBlock(pWriter->pDataFWriter, &pWriter->sData, pWriter->aSttBlk, pWriter->cmprAlg); + code = tsdbSnapWriteNextRow(pWriter, &pRowInfo); TSDB_CHECK_CODE(code, lino, _exit); } @@ -1717,88 +1723,33 @@ int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWr TSDB_CHECK_CODE(code, lino, _exit); // SNAP_DATA_TSDB -#if 1 - pWriter->fid = INT32_MIN; - code = tBlockDataCreate(&pWriter->inData); TSDB_CHECK_CODE(code, lino, _exit); - code = tBlockDataCreate(&pWriter->bData); - TSDB_CHECK_CODE(code, lino, _exit); -#else - code = tBlockDataCreate(&pWriter->bData); - TSDB_CHECK_CODE(code, lino, _exit); - pWriter->fid = INT32_MIN; - pWriter->id = (TABLEID){0}; - // Reader - pWriter->dReader.aBlockIdx = taosArrayInit(0, sizeof(SBlockIdx)); - if (pWriter->dReader.aBlockIdx == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - code = tBlockDataCreate(&pWriter->dReader.bData); + + code = tBlockDataCreate(&pWriter->bData); TSDB_CHECK_CODE(code, lino, _exit); - // Writer - pWriter->dWriter.aBlockIdx = taosArrayInit(0, sizeof(SBlockIdx)); - if (pWriter->dWriter.aBlockIdx == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - pWriter->dWriter.aSttBlk = taosArrayInit(0, sizeof(SSttBlk)); - if (pWriter->dWriter.aSttBlk == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - code = tBlockDataCreate(&pWriter->dWriter.bData); + code = tBlockDataCreate(&pWriter->sData); TSDB_CHECK_CODE(code, lino, _exit); - code = tBlockDataCreate(&pWriter->dWriter.sData); - TSDB_CHECK_CODE(code, lino, _exit); -#endif // SNAP_DATA_DEL - pWriter->aDelIdxR = taosArrayInit(0, sizeof(SDelIdx)); - if (pWriter->aDelIdxR == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - pWriter->aDelData = taosArrayInit(0, sizeof(SDelData)); - if (pWriter->aDelData == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - pWriter->aDelIdxW = taosArrayInit(0, sizeof(SDelIdx)); - if (pWriter->aDelIdxW == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } _exit: if (code) { tsdbError("vgId:%d, %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); - *ppWriter = NULL; - if (pWriter) { -#if 0 - if (pWriter->aDelIdxW) taosArrayDestroy(pWriter->aDelIdxW); - if (pWriter->aDelData) taosArrayDestroy(pWriter->aDelData); - if (pWriter->aDelIdxR) taosArrayDestroy(pWriter->aDelIdxR); - tBlockDataDestroy(&pWriter->dWriter.sData, 1); - tBlockDataDestroy(&pWriter->dWriter.bData, 1); - if (pWriter->dWriter.aSttBlk) taosArrayDestroy(pWriter->dWriter.aSttBlk); - if (pWriter->dWriter.aBlockIdx) taosArrayDestroy(pWriter->dWriter.aBlockIdx); - tBlockDataDestroy(&pWriter->dReader.bData, 1); - if (pWriter->dReader.aBlockIdx) taosArrayDestroy(pWriter->dReader.aBlockIdx); + tBlockDataDestroy(&pWriter->sData, 1); tBlockDataDestroy(&pWriter->bData, 1); + tBlockDataDestroy(&pWriter->inData, 1); tsdbFSDestroy(&pWriter->fs); - taosMemoryFree(pWriter); -#endif + pWriter = NULL; } } else { - tsdbInfo("vgId:%d, %s done", TD_VID(pTsdb->pVnode), __func__); - *ppWriter = pWriter; + tsdbInfo("vgId:%d %s done, sver:%" PRId64 " ever:%" PRId64, TD_VID(pTsdb->pVnode), __func__, sver, ever); } + *ppWriter = pWriter; return code; } From 545e4d7f02669d1db271db3771dab369df51f52c Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sat, 28 Jan 2023 13:30:15 +0800 Subject: [PATCH 026/124] make it compile --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 114 ++++++++++----------- 1 file changed, 57 insertions(+), 57 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 0a04d09a2c..e602226457 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1370,63 +1370,6 @@ _exit: return code; } -static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { - int32_t code = 0; - int32_t lino = 0; - - ASSERT(pWriter->pDataFWriter); - - // consume remain data and end with a NULL table row - SRowInfo* pRowInfo; - code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); - TSDB_CHECK_CODE(code, lino, _exit); - for (;;) { - code = tsdbSnapWriteTableData(pWriter, pRowInfo); - TSDB_CHECK_CODE(code, lino, _exit); - - if (pRowInfo == NULL) break; - - code = tsdbSnapWriteNextRow(pWriter, &pRowInfo); - TSDB_CHECK_CODE(code, lino, _exit); - } - - // do file-level updates - code = tsdbWriteSttBlk(pWriter->pDataFWriter, pWriter->aSttBlk); - TSDB_CHECK_CODE(code, lino, _exit); - - code = tsdbWriteBlockIdx(pWriter->pDataFWriter, pWriter->aBlockIdx); - TSDB_CHECK_CODE(code, lino, _exit); - - code = tsdbUpdateDFileSetHeader(pWriter->pDataFWriter); - TSDB_CHECK_CODE(code, lino, _exit); - - code = tsdbFSUpsertFSet(&pWriter->fs, &pWriter->pDataFWriter->wSet); - TSDB_CHECK_CODE(code, lino, _exit); - - code = tsdbDataFWriterClose(&pWriter->pDataFWriter, 1); - TSDB_CHECK_CODE(code, lino, _exit); - - if (pWriter->pDataFReader) { - code = tsdbDataFReaderClose(&pWriter->pDataFReader); - TSDB_CHECK_CODE(code, lino, _exit); - } - - // clear sources - while (pWriter->iterList) { - STsdbDataIter2* pIter = pWriter->iterList; - pWriter->iterList = pIter->next; - tsdbCloseDataIter2(pIter); - } - -_exit: - if (code) { - tsdbError("vgId:%d %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); - } else { - tsdbDebug("vgId:%d %s is done", TD_VID(pWriter->pTsdb->pVnode), __func__); - } - return code; -} - static int32_t tsdbSnapWriteNextRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowInfo) { int32_t code = 0; int32_t lino = 0; @@ -1493,6 +1436,63 @@ _exit: return code; } +static int32_t tsdbSnapWriteFileDataEnd(STsdbSnapWriter* pWriter) { + int32_t code = 0; + int32_t lino = 0; + + ASSERT(pWriter->pDataFWriter); + + // consume remain data and end with a NULL table row + SRowInfo* pRowInfo; + code = tsdbSnapWriteGetRow(pWriter, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + for (;;) { + code = tsdbSnapWriteTableData(pWriter, pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pRowInfo == NULL) break; + + code = tsdbSnapWriteNextRow(pWriter, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // do file-level updates + code = tsdbWriteSttBlk(pWriter->pDataFWriter, pWriter->aSttBlk); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbWriteBlockIdx(pWriter->pDataFWriter, pWriter->aBlockIdx); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbUpdateDFileSetHeader(pWriter->pDataFWriter); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbFSUpsertFSet(&pWriter->fs, &pWriter->pDataFWriter->wSet); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbDataFWriterClose(&pWriter->pDataFWriter, 1); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->pDataFReader) { + code = tsdbDataFReaderClose(&pWriter->pDataFReader); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // clear sources + while (pWriter->iterList) { + STsdbDataIter2* pIter = pWriter->iterList; + pWriter->iterList = pIter->next; + tsdbCloseDataIter2(pIter); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s is done", TD_VID(pWriter->pTsdb->pVnode), __func__); + } + return code; +} + static int32_t tsdbSnapWriteTimeSeriesData(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { int32_t code = 0; int32_t lino = 0; From 30a6488f0d757f312f9299eaece032823a512cd0 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Sat, 28 Jan 2023 13:38:39 +0800 Subject: [PATCH 027/124] fix: create topic before inserting data for two testcases of 7-tmq/tmqUpdate --- tests/system-test/7-tmq/tmqUpdate-1ctb.py | 31 ++++++++++------- .../7-tmq/tmqUpdate-multiCtb-snapshot0.py | 34 +++++++++++-------- 2 files changed, 37 insertions(+), 28 deletions(-) diff --git a/tests/system-test/7-tmq/tmqUpdate-1ctb.py b/tests/system-test/7-tmq/tmqUpdate-1ctb.py index b974e4a41a..f03fa84335 100644 --- a/tests/system-test/7-tmq/tmqUpdate-1ctb.py +++ b/tests/system-test/7-tmq/tmqUpdate-1ctb.py @@ -100,6 +100,14 @@ class TDTestCase: # update to half tables paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) + + tdLog.info("create topics from stb1") + topicFromStb1 = 'topic_stb1' + queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as %s" %(topicFromStb1, queryString) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + # tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix="ctbx", # ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], # startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) @@ -107,12 +115,6 @@ class TDTestCase: ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) - tdLog.info("create topics from stb1") - topicFromStb1 = 'topic_stb1' - queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) - sqlString = "create topic %s as %s" %(topicFromStb1, queryString) - tdLog.info("create topic sql: %s"%sqlString) - tdSql.execute(sqlString) # paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl @@ -156,6 +158,7 @@ class TDTestCase: def tmqCase2(self): tdLog.printNoPrefix("======== test case 2: ") + paraDict = {'dbName': 'dbt', 'dropFlag': 1, 'event': '', @@ -175,7 +178,7 @@ class TDTestCase: 'showMsg': 1, 'showRow': 1, 'snapshot': 0} - + self.snapshot = 1 paraDict['snapshot'] = self.snapshot paraDict['vgroups'] = self.vgroups paraDict['ctbNum'] = self.ctbNum @@ -187,12 +190,6 @@ class TDTestCase: # update to half tables paraDict['startTs'] = paraDict['startTs'] + int(self.rowsPerTbl / 2) paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) - tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict["ctbPrefix"], - ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], - startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) - # tmqCom.insert_data_interlaceByMultiTbl(tsql=tdSql,dbName=paraDict["dbName"],ctbPrefix=paraDict["ctbPrefix"], - # ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], - # startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) tmqCom.initConsumerTable() tdLog.info("create topics from stb1") @@ -202,6 +199,14 @@ class TDTestCase: tdLog.info("create topic sql: %s"%sqlString) tdSql.execute(sqlString) + tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict["ctbPrefix"], + ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], + startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) + # tmqCom.insert_data_interlaceByMultiTbl(tsql=tdSql,dbName=paraDict["dbName"],ctbPrefix=paraDict["ctbPrefix"], + # ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], + # startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) + + # paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl consumerId = 1 diff --git a/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py b/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py index d5df88cf43..7310466abe 100644 --- a/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py +++ b/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py @@ -101,19 +101,20 @@ class TDTestCase: # update to half tables paraDict['ctbNum'] = int(self.ctbNum/2) paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) - tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=self.autoCtbPrefix, - ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], - startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) - tmqCom.insert_data_interlaceByMultiTbl(tsql=tdSql,dbName=paraDict["dbName"],ctbPrefix=paraDict["ctbPrefix"], - ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], - startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) - + tdLog.info("create topics from stb1") topicFromStb1 = 'topic_stb1' queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) sqlString = "create topic %s as %s" %(topicFromStb1, queryString) tdLog.info("create topic sql: %s"%sqlString) tdSql.execute(sqlString) + + tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=self.autoCtbPrefix, + ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], + startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) + tmqCom.insert_data_interlaceByMultiTbl(tsql=tdSql,dbName=paraDict["dbName"],ctbPrefix=paraDict["ctbPrefix"], + ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], + startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl @@ -176,6 +177,7 @@ class TDTestCase: 'showRow': 1, 'snapshot': 0} + self.snapshot = 1 paraDict['snapshot'] = self.snapshot paraDict['vgroups'] = self.vgroups paraDict['ctbNum'] = self.ctbNum @@ -188,6 +190,16 @@ class TDTestCase: paraDict['ctbNum'] = int(self.ctbNum/2) paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) paraDict['startTs'] = paraDict['startTs'] + int(self.rowsPerTbl / 2) + + tmqCom.initConsumerTable() + tdLog.info("create topics from stb1") + topicFromStb1 = 'topic_stb1' + queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as %s" %(topicFromStb1, queryString) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + + tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=self.autoCtbPrefix, ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']+int(self.ctbNum/2)) @@ -200,14 +212,6 @@ class TDTestCase: ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']+int(self.ctbNum/2)) - tmqCom.initConsumerTable() - tdLog.info("create topics from stb1") - topicFromStb1 = 'topic_stb1' - queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) - sqlString = "create topic %s as %s" %(topicFromStb1, queryString) - tdLog.info("create topic sql: %s"%sqlString) - tdSql.execute(sqlString) - # paraDict['ctbNum'] = self.ctbNum paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl From e8dfa66750ca5c5d7f1d4193d0a389c14ea483cf Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Sat, 28 Jan 2023 14:24:04 +0800 Subject: [PATCH 028/124] fix: taosbenchmark ignore drop db failure for cloud (#19671) * fix: taosbenchmark ignore drop db failure for cloud * fix: update taos-tools 0cd564a for main --- cmake/taostools_CMakeLists.txt.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/taostools_CMakeLists.txt.in b/cmake/taostools_CMakeLists.txt.in index a69e1578c0..13a81f88ea 100644 --- a/cmake/taostools_CMakeLists.txt.in +++ b/cmake/taostools_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taos-tools ExternalProject_Add(taos-tools GIT_REPOSITORY https://github.com/taosdata/taos-tools.git - GIT_TAG 7d24ed5 + GIT_TAG 0cd564a SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools" BINARY_DIR "" #BUILD_IN_SOURCE TRUE From 2a71d47668208ec814269fad359bf95a3c0d41a9 Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Sat, 28 Jan 2023 16:11:17 +0800 Subject: [PATCH 029/124] fix: wal ref --- include/libs/wal/wal.h | 1 + source/dnode/vnode/src/tq/tq.c | 4 +++- source/libs/wal/src/walRef.c | 29 +++++++++++++++++++++++++++-- 3 files changed, 31 insertions(+), 3 deletions(-) diff --git a/include/libs/wal/wal.h b/include/libs/wal/wal.h index a1ae1e429d..bef7301a07 100644 --- a/include/libs/wal/wal.h +++ b/include/libs/wal/wal.h @@ -201,6 +201,7 @@ int32_t walFetchHead(SWalReader *pRead, int64_t ver, SWalCkHead *pHead); int32_t walFetchBody(SWalReader *pRead, SWalCkHead **ppHead); int32_t walSkipFetchBody(SWalReader *pRead, const SWalCkHead *pHead); +SWalRef *walRefFirstVer(SWal *); SWalRef *walRefCommittedVer(SWal *); SWalRef *walOpenRef(SWal *); diff --git a/source/dnode/vnode/src/tq/tq.c b/source/dnode/vnode/src/tq/tq.c index 1d5fae33eb..7649e8a006 100644 --- a/source/dnode/vnode/src/tq/tq.c +++ b/source/dnode/vnode/src/tq/tq.c @@ -521,7 +521,9 @@ int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg) { tqOffsetResetToData(&fetchOffsetNew, 0, 0); } } else { - tqOffsetResetToLog(&fetchOffsetNew, walGetFirstVer(pTq->pVnode->pWal)); + int64_t firstVer = walGetFirstVer(pTq->pVnode->pWal); + walRefVer(pHandle->pRef, firstVer); + tqOffsetResetToLog(&fetchOffsetNew, firstVer - 1); } } else if (reqOffset.type == TMQ_OFFSET__RESET_LATEST) { if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__COLUMN) { diff --git a/source/libs/wal/src/walRef.c b/source/libs/wal/src/walRef.c index e86111109c..f5cfe9abae 100644 --- a/source/libs/wal/src/walRef.c +++ b/source/libs/wal/src/walRef.c @@ -77,14 +77,39 @@ void walUnrefVer(SWalRef *pRef) { } #endif -SWalRef *walRefCommittedVer(SWal *pWal) { +SWalRef *walRefFirstVer(SWal *pWal) { SWalRef *pRef = walOpenRef(pWal); if (pRef == NULL) { return NULL; } taosThreadMutexLock(&pWal->mutex); - int64_t ver = walGetCommittedVer(pWal); + int64_t ver = walGetFirstVer(pWal); + + wDebug("vgId:%d, wal ref version %" PRId64 " for first", pWal->cfg.vgId, ver); + + pRef->refVer = ver; + // bsearch in fileSet + SWalFileInfo tmpInfo; + tmpInfo.firstVer = ver; + SWalFileInfo *pRet = taosArraySearch(pWal->fileInfoSet, &tmpInfo, compareWalFileInfo, TD_LE); + ASSERT(pRet != NULL); + pRef->refFile = pRet->firstVer; + + taosThreadMutexUnlock(&pWal->mutex); + return pRef; +} + +SWalRef *walRefCommittedVer(SWal *pWal) { + SWalRef *pRef = walOpenRef(pWal); + if (pRef == NULL) { + return NULL; + } + taosThreadMutexLock(&pWal->mutex); + + int64_t ver = walGetCommittedVer(pWal); + + wDebug("vgId:%d, wal ref version %" PRId64 " for committed", pWal->cfg.vgId, ver); pRef->refVer = ver; // bsearch in fileSet From fb2b42f968de39118eabe85f753e7d3f9d1547e0 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sat, 28 Jan 2023 16:51:22 +0800 Subject: [PATCH 030/124] refact more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 460 ++++++++++++++------- 1 file changed, 307 insertions(+), 153 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index e602226457..6096936c29 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -21,6 +21,7 @@ extern int32_t tsdbReadDataBlockEx(SDataFReader* pReader, SDataBlk* pDataBlk, SB #define TSDB_MEM_TABLE_DATA_ITER 0 #define TSDB_DATA_FILE_DATA_ITER 1 #define TSDB_STT_FILE_DATA_ITER 2 +#define TSDB_TOMB_FILE_DATA_ITER 3 typedef struct STsdbDataIter2 STsdbDataIter2; struct STsdbDataIter2 { @@ -55,6 +56,14 @@ struct STsdbDataIter2 { int32_t iSttBlk; int32_t iRow; } sIter; + // TSDB_TOMB_FILE_DATA_ITER + struct { + SDelFReader* pReader; + SArray* aDelIdx; + SArray* aDelData; + int32_t iDelIdx; + int32_t iDelData; + } tIter; }; }; @@ -152,6 +161,49 @@ _exit: return code; } +static int32_t tsdbOpenTombFileDataIter(SDelFReader* pReader, STsdbDataIter2** ppIter) { + int32_t code = 0; + int32_t lino = 0; + + STsdbDataIter2* pIter = (STsdbDataIter2*)taosMemoryCalloc(1, sizeof(*pIter)); + if (pIter == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + pIter->type = TSDB_TOMB_FILE_DATA_ITER; + + pIter->tIter.pReader = pReader; + if ((pIter->tIter.aDelIdx = taosArrayInit(0, sizeof(SDelIdx))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + if ((pIter->tIter.aDelData = taosArrayInit(0, sizeof(SDelData))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tsdbReadDelIdx(pReader, pIter->tIter.aDelIdx); + TSDB_CHECK_CODE(code, lino, _exit); + + if (taosArrayGetSize(pIter->tIter.aDelIdx) == 0) goto _clear; + + pIter->tIter.iDelIdx = 0; + pIter->tIter.iDelData = 0; + +_exit: + if (code) { + if (pIter) { + _clear: + taosArrayDestroy(pIter->tIter.aDelIdx); + taosArrayDestroy(pIter->tIter.aDelData); + taosMemoryFree(pIter); + pIter = NULL; + } + } + *ppIter = pIter; + return code; +} + /* close */ static void tsdbCloseDataFileDataIter(STsdbDataIter2* pIter) { tBlockDataDestroy(&pIter->dIter.bData, 1); @@ -921,19 +973,19 @@ struct STsdbSnapWriter { uint8_t* aBuf[5]; STsdbFS fs; + TABLEID tbid; // time-series data SBlockData inData; int32_t fid; - TABLEID tbid; SSkmInfo skmTable; /* reader */ SDataFReader* pDataFReader; STsdbDataIter2* iterList; STsdbDataIter2* pDIter; - STsdbDataIter2* pIter; + STsdbDataIter2* pSIter; SRBTree rbt; // SRBTree /* writer */ @@ -945,12 +997,14 @@ struct STsdbSnapWriter { SBlockData sData; // tombstone data - SDelFReader* pDelFReader; + /* reader */ + SDelFReader* pDelFReader; + STsdbDataIter2* pTIter; + + /* writer */ SDelFWriter* pDelFWriter; - int32_t iDelIdx; - SArray* aDelIdxR; + SArray* aDelIdx; SArray* aDelData; - SArray* aDelIdxW; }; // SNAP_DATA_TSDB @@ -1207,6 +1261,7 @@ static int32_t tsdbSnapWriteTableDataEnd(STsdbSnapWriter* pWriter) { if (pWriter->bData.nRow > 0) { if (pWriter->bData.nRow < pWriter->minRow) { + ASSERT(TABLE_SAME_SCHEMA(pWriter->sData.suid, pWriter->sData.uid, pWriter->tbid.suid, pWriter->tbid.uid)); for (int32_t iRow = 0; iRow < pWriter->bData.nRow; iRow++) { code = tBlockDataAppendRow(&pWriter->sData, &tsdbRowFromBlockData(&pWriter->bData, iRow), NULL, pWriter->tbid.uid); @@ -1262,7 +1317,7 @@ static int32_t tsdbSnapWriteFileDataStart(STsdbSnapWriter* pWriter, int32_t fid) pWriter->pDataFReader = NULL; pWriter->iterList = NULL; pWriter->pDIter = NULL; - pWriter->pIter = NULL; + pWriter->pSIter = NULL; tRBTreeCreate(&pWriter->rbt, tsdbDataIterCmprFn); if (pSet) { code = tsdbDataFReaderOpen(&pWriter->pDataFReader, pTsdb, pSet); @@ -1276,23 +1331,23 @@ static int32_t tsdbSnapWriteFileDataStart(STsdbSnapWriter* pWriter, int32_t fid) } for (int32_t iStt = 0; iStt < pSet->nSttF; iStt++) { - code = tsdbOpenSttFileDataIter(pWriter->pDataFReader, iStt, &pWriter->pIter); + code = tsdbOpenSttFileDataIter(pWriter->pDataFReader, iStt, &pWriter->pSIter); TSDB_CHECK_CODE(code, lino, _exit); - if (pWriter->pIter) { - code = tsdbSttFileDataIterNext(pWriter->pIter); + if (pWriter->pSIter) { + code = tsdbSttFileDataIterNext(pWriter->pSIter); TSDB_CHECK_CODE(code, lino, _exit); // add to tree - tRBTreePut(&pWriter->rbt, &pWriter->pIter->rbtn); + tRBTreePut(&pWriter->rbt, &pWriter->pSIter->rbtn); // add to list - pWriter->pIter->next = pWriter->iterList; - pWriter->iterList = pWriter->pIter; + pWriter->pSIter->next = pWriter->iterList; + pWriter->iterList = pWriter->pSIter; } } - pWriter->pIter = NULL; + pWriter->pSIter = NULL; } // open writer @@ -1348,7 +1403,7 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn // switch to new table if need if (pRowInfo == NULL || pRowInfo->uid != pWriter->tbid.uid) { - if (pWriter->tbid.uid != 0) { + if (pWriter->tbid.uid) { code = tsdbSnapWriteTableDataEnd(pWriter); TSDB_CHECK_CODE(code, lino, _exit); } @@ -1357,11 +1412,10 @@ static int32_t tsdbSnapWriteTableData(STsdbSnapWriter* pWriter, SRowInfo* pRowIn TSDB_CHECK_CODE(code, lino, _exit); } - // end with a NULL row - if (pRowInfo) { - code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); - TSDB_CHECK_CODE(code, lino, _exit); - } + if (pRowInfo == NULL) goto _exit; + + code = tsdbSnapWriteTableRow(pWriter, &pRowInfo->row); + TSDB_CHECK_CODE(code, lino, _exit); _exit: if (code) { @@ -1374,19 +1428,19 @@ static int32_t tsdbSnapWriteNextRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowIn int32_t code = 0; int32_t lino = 0; - if (pWriter->pIter) { - code = tsdbDataIterNext2(pWriter->pIter); + if (pWriter->pSIter) { + code = tsdbDataIterNext2(pWriter->pSIter); TSDB_CHECK_CODE(code, lino, _exit); - if (pWriter->pIter->rowInfo.suid == 0 && pWriter->pIter->rowInfo.uid == 0) { - pWriter->pIter = NULL; + if (pWriter->pSIter->rowInfo.suid == 0 && pWriter->pSIter->rowInfo.uid == 0) { + pWriter->pSIter = NULL; } else { SRBTreeNode* pNode = tRBTreeMin(&pWriter->rbt); if (pNode) { - int32_t c = tsdbDataIterCmprFn(&pWriter->pIter->rbtn, pNode); + int32_t c = tsdbDataIterCmprFn(&pWriter->pSIter->rbtn, pNode); if (c > 0) { - tRBTreePut(&pWriter->rbt, &pWriter->pIter->rbtn); - pWriter->pIter = NULL; + tRBTreePut(&pWriter->rbt, &pWriter->pSIter->rbtn); + pWriter->pSIter = NULL; } else if (c == 0) { ASSERT(0); } @@ -1394,17 +1448,17 @@ static int32_t tsdbSnapWriteNextRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowIn } } - if (pWriter->pIter == NULL) { + if (pWriter->pSIter == NULL) { SRBTreeNode* pNode = tRBTreeMin(&pWriter->rbt); if (pNode) { tRBTreeDrop(&pWriter->rbt, pNode); - pWriter->pIter = TSDB_RBTN_TO_DATA_ITER(pNode); + pWriter->pSIter = TSDB_RBTN_TO_DATA_ITER(pNode); } } if (ppRowInfo) { - if (pWriter->pIter) { - *ppRowInfo = &pWriter->pIter->rowInfo; + if (pWriter->pSIter) { + *ppRowInfo = &pWriter->pSIter->rowInfo; } else { *ppRowInfo = NULL; } @@ -1421,8 +1475,8 @@ static int32_t tsdbSnapWriteGetRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowInf int32_t code = 0; int32_t lino = 0; - if (pWriter->pIter) { - *ppRowInfo = &pWriter->pIter->rowInfo; + if (pWriter->pSIter) { + *ppRowInfo = &pWriter->pSIter->rowInfo; goto _exit; } @@ -1558,143 +1612,238 @@ _exit: } // SNAP_DATA_DEL -static int32_t tsdbSnapMoveWriteDelData(STsdbSnapWriter* pWriter, TABLEID* pId) { +static int32_t tsdbSnapWriteDelTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pId) { int32_t code = 0; + int32_t lino = 0; - while (true) { - if (pWriter->iDelIdx >= taosArrayGetSize(pWriter->aDelIdxR)) break; + if (pId) { + pWriter->tbid = *pId; + } else { + pWriter->tbid = (TABLEID){.suid = INT64_MAX, .uid = INT64_MAX}; + } - SDelIdx* pDelIdx = (SDelIdx*)taosArrayGet(pWriter->aDelIdxR, pWriter->iDelIdx); + if (pWriter->pTIter) { + ASSERT(pWriter->pTIter->tIter.iDelData >= taosArrayGetSize(pWriter->pTIter->tIter.aDelData)); - if (tTABLEIDCmprFn(pDelIdx, pId) >= 0) break; + for (;;) { + if (pWriter->pTIter->tIter.iDelIdx >= taosArrayGetSize(pWriter->pTIter->tIter.aDelIdx)) { + break; + } - code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->aDelData); - if (code) goto _exit; + SDelIdx* pDelIdx = taosArrayGet(pWriter->pTIter->tIter.aDelIdx, pWriter->pTIter->tIter.iDelIdx); - SDelIdx delIdx = *pDelIdx; - code = tsdbWriteDelData(pWriter->pDelFWriter, pWriter->aDelData, &delIdx); - if (code) goto _exit; + int32_t c = tTABLEIDCmprFn(pDelIdx, &pWriter->tbid); + if (c < 0) { + code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->pTIter->tIter.aDelData); + TSDB_CHECK_CODE(code, lino, _exit); - if (taosArrayPush(pWriter->aDelIdxW, &delIdx) == NULL) { + SDelIdx* pDelIdxNew = taosArrayReserve(pWriter->pTIter->tIter.aDelIdx, 1); + if (pDelIdxNew == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + pDelIdxNew->suid = pDelIdx->suid; + pDelIdxNew->uid = pDelIdx->uid; + + code = tsdbWriteDelData(pWriter->pDelFWriter, pWriter->pTIter->tIter.aDelData, pDelIdxNew); + TSDB_CHECK_CODE(code, lino, _exit); + + pWriter->pTIter->tIter.iDelIdx++; + } else if (c == 0) { + code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->pTIter->tIter.aDelData); + TSDB_CHECK_CODE(code, lino, _exit); + + if (taosArrayAddBatch(pWriter->aDelData, pWriter->pTIter->tIter.aDelData->pData, + taosArrayGetSize(pWriter->pTIter->tIter.aDelData)) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + pWriter->pTIter->tIter.iDelData = taosArrayGetSize(pWriter->pTIter->tIter.aDelData); + pWriter->pTIter->tIter.iDelIdx++; + break; + } else { + pWriter->pTIter->tIter.iDelData = taosArrayGetSize(pWriter->pTIter->tIter.aDelData); + break; + } + } + } + + taosArrayClear(pWriter->aDelData); + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbTrace("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, pId->suid, + pId->uid); + } + return code; +} + +static int32_t tsdbSnapWriteDelTableDataEnd(STsdbSnapWriter* pWriter) { + int32_t code = 0; + int32_t lino = 0; + + if (taosArrayGetSize(pWriter->aDelData) > 0) { + SDelIdx* pDelIdx = taosArrayReserve(pWriter->aDelIdx, 1); + if (pDelIdx == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; - goto _exit; + TSDB_CHECK_CODE(code, lino, _exit); } - pWriter->iDelIdx++; + pDelIdx->suid = pWriter->tbid.suid; + pDelIdx->uid = pWriter->tbid.uid; + + code = tsdbWriteDelData(pWriter->pDelFWriter, pWriter->aDelData, pDelIdx); + TSDB_CHECK_CODE(code, lino, _exit); } _exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbTrace("vgId:%d %s done", TD_VID(pWriter->pTsdb->pVnode), __func__); + } + return code; +} + +static int32_t tsdbSnapWriteDelTableData(STsdbSnapWriter* pWriter, TABLEID* pId, uint8_t* pData, int64_t size) { + int32_t code = 0; + int32_t lino = 0; + + if (pId == NULL || pId->uid != pWriter->tbid.uid) { + if (pWriter->tbid.uid) { + code = tsdbSnapWriteDelTableDataEnd(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tsdbSnapWriteDelTableDataStart(pWriter, pId); + TSDB_CHECK_CODE(code, lino, _exit); + } + + if (pId == NULL) goto _exit; + + int64_t n = 0; + while (n < size) { + SDelData delData; + n += tGetDelData(pData + n, &delData); + + if (taosArrayPush(pWriter->aDelData, &delData) < 0) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + } + + ASSERT(n == size); + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapWriteDelDataStart(STsdbSnapWriter* pWriter) { + int32_t code = 0; + int32_t lino = 0; + + STsdb* pTsdb = pWriter->pTsdb; + SDelFile* pDelFile = pWriter->fs.pDelFile; + + pWriter->tbid = (TABLEID){0}; + + // reader + if (pDelFile) { + code = tsdbDelFReaderOpen(&pWriter->pDelFReader, pDelFile, pTsdb); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbOpenTombFileDataIter(pWriter->pDelFReader, &pWriter->pTIter); + TSDB_CHECK_CODE(code, lino, _exit); + } + + // writer + code = tsdbDelFWriterOpen(&pWriter->pDelFWriter, &(SDelFile){.commitID = pWriter->commitID}, pTsdb); + TSDB_CHECK_CODE(code, lino, _exit); + + if ((pWriter->aDelIdx = taosArrayInit(0, sizeof(SDelIdx))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + if ((pWriter->aDelData = taosArrayInit(0, sizeof(SDelData))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s done", TD_VID(pTsdb->pVnode), __func__); + } + return code; +} + +static int32_t tsdbSnapWriteDelDataEnd(STsdbSnapWriter* pWriter) { + int32_t code = 0; + int32_t lino = 0; + + STsdb* pTsdb = pWriter->pTsdb; + + // end remaining table (TODO) + code = tsdbSnapWriteDelTableData(pWriter, NULL, NULL, 0); + TSDB_CHECK_CODE(code, lino, _exit); + + // update file-level info + code = tsdbWriteDelIdx(pWriter->pDelFWriter, pWriter->aDelIdx); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbUpdateDelFileHdr(pWriter->pDelFWriter); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbFSUpsertDelFile(&pWriter->fs, &pWriter->pDelFWriter->fDel); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbDelFWriterClose(&pWriter->pDelFWriter, 1); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pWriter->pDelFReader) { + code = tsdbDelFReaderClose(&pWriter->pDelFReader); + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbInfo("vgId:%d %s done", TD_VID(pTsdb->pVnode), __func__); + } return code; } static int32_t tsdbSnapWriteDelData(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr) { int32_t code = 0; - STsdb* pTsdb = pWriter->pTsdb; + int32_t lino = 0; - // Open del file if not opened yet + STsdb* pTsdb = pWriter->pTsdb; + + // start to write del data if need if (pWriter->pDelFWriter == NULL) { - SDelFile* pDelFile = pWriter->fs.pDelFile; - - // reader - if (pDelFile) { - code = tsdbDelFReaderOpen(&pWriter->pDelFReader, pDelFile, pTsdb); - if (code) goto _err; - - code = tsdbReadDelIdx(pWriter->pDelFReader, pWriter->aDelIdxR); - if (code) goto _err; - } else { - taosArrayClear(pWriter->aDelIdxR); - } - pWriter->iDelIdx = 0; - - // writer - SDelFile delFile = {.commitID = pWriter->commitID}; - code = tsdbDelFWriterOpen(&pWriter->pDelFWriter, &delFile, pTsdb); - if (code) goto _err; - taosArrayClear(pWriter->aDelIdxW); + code = tsdbSnapWriteDelDataStart(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); } - TABLEID id = *(TABLEID*)pHdr->data; + code = tsdbSnapWriteDelTableData(pWriter, (TABLEID*)pHdr->data, pHdr->data + sizeof(TABLEID), + pHdr->size - sizeof(TABLEID)); + TSDB_CHECK_CODE(code, lino, _exit); - // Move write data < id - code = tsdbSnapMoveWriteDelData(pWriter, &id); - if (code) goto _err; - - // Merge incoming data with current - if (pWriter->iDelIdx < taosArrayGetSize(pWriter->aDelIdxR) && - tTABLEIDCmprFn(taosArrayGet(pWriter->aDelIdxR, pWriter->iDelIdx), &id) == 0) { - SDelIdx* pDelIdx = (SDelIdx*)taosArrayGet(pWriter->aDelIdxR, pWriter->iDelIdx); - - code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->aDelData); - if (code) goto _err; - - pWriter->iDelIdx++; +_exit: + if (code) { + tsdbError("vgId:%d %s failed since %s", TD_VID(pTsdb->pVnode), __func__, tstrerror(code)); } else { - taosArrayClear(pWriter->aDelData); + tsdbTrace("vgId:%d %s done", TD_VID(pTsdb->pVnode), __func__); } - - int64_t n = sizeof(TABLEID); - while (n < pHdr->size) { - SDelData delData; - - n += tGetDelData(pHdr->data + n, &delData); - - if (taosArrayPush(pWriter->aDelData, &delData) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - } - - SDelIdx delIdx = {.suid = id.suid, .uid = id.uid}; - code = tsdbWriteDelData(pWriter->pDelFWriter, pWriter->aDelData, &delIdx); - if (code) goto _err; - - if (taosArrayPush(pWriter->aDelIdxW, &delIdx) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _err; - } - - return code; - -_err: - tsdbError("vgId:%d, vnode snapshot tsdb write del for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path, - tstrerror(code)); - return code; -} - -static int32_t tsdbSnapWriteDelEnd(STsdbSnapWriter* pWriter) { - int32_t code = 0; - STsdb* pTsdb = pWriter->pTsdb; - - if (pWriter->pDelFWriter == NULL) return code; - - TABLEID id = {.suid = INT64_MAX, .uid = INT64_MAX}; - code = tsdbSnapMoveWriteDelData(pWriter, &id); - if (code) goto _err; - - code = tsdbWriteDelIdx(pWriter->pDelFWriter, pWriter->aDelIdxW); - if (code) goto _err; - - code = tsdbUpdateDelFileHdr(pWriter->pDelFWriter); - if (code) goto _err; - - code = tsdbFSUpsertDelFile(&pWriter->fs, &pWriter->pDelFWriter->fDel); - if (code) goto _err; - - code = tsdbDelFWriterClose(&pWriter->pDelFWriter, 1); - if (code) goto _err; - - if (pWriter->pDelFReader) { - code = tsdbDelFReaderClose(&pWriter->pDelFReader); - if (code) goto _err; - } - - tsdbInfo("vgId:%d, vnode snapshot tsdb write del for %s end", TD_VID(pTsdb->pVnode), pTsdb->path); - return code; - -_err: - tsdbError("vgId:%d, vnode snapshot tsdb write del end for %s failed since %s", TD_VID(pTsdb->pVnode), pTsdb->path, - tstrerror(code)); return code; } @@ -1738,7 +1887,7 @@ int32_t tsdbSnapWriterOpen(STsdb* pTsdb, int64_t sver, int64_t ever, STsdbSnapWr _exit: if (code) { - tsdbError("vgId:%d, %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); if (pWriter) { tBlockDataDestroy(&pWriter->sData, 1); tBlockDataDestroy(&pWriter->bData, 1); @@ -1755,20 +1904,26 @@ _exit: int32_t tsdbSnapWriterPrepareClose(STsdbSnapWriter* pWriter) { int32_t code = 0; + int32_t lino = 0; + if (pWriter->pDataFWriter) { code = tsdbSnapWriteFileDataEnd(pWriter); - if (code) goto _exit; + TSDB_CHECK_CODE(code, lino, _exit); } - code = tsdbSnapWriteDelEnd(pWriter); - if (code) goto _exit; + if (pWriter->pDelFWriter) { + code = tsdbSnapWriteDelDataEnd(pWriter); + TSDB_CHECK_CODE(code, lino, _exit); + } code = tsdbFSPrepareCommit(pWriter->pTsdb, &pWriter->fs); - if (code) goto _exit; + TSDB_CHECK_CODE(code, lino, _exit); _exit: if (code) { - tsdbError("vgId:%d, %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, tstrerror(code)); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s done", TD_VID(pWriter->pTsdb->pVnode), __func__); } return code; } @@ -1795,9 +1950,8 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { } // SNAP_DATA_DEL - taosArrayDestroy(pWriter->aDelIdxW); taosArrayDestroy(pWriter->aDelData); - taosArrayDestroy(pWriter->aDelIdxR); + taosArrayDestroy(pWriter->aDelIdx); // SNAP_DATA_TSDB tBlockDataDestroy(&pWriter->sData, 1); From f5c4ca3380ba7361b8b2a89910896078303c6f8a Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Sat, 28 Jan 2023 17:17:10 +0800 Subject: [PATCH 031/124] fix: wal ref --- include/libs/wal/wal.h | 2 +- source/dnode/vnode/src/tq/tq.c | 11 ++++++++--- source/libs/wal/src/walRef.c | 8 +++++--- 3 files changed, 14 insertions(+), 7 deletions(-) diff --git a/include/libs/wal/wal.h b/include/libs/wal/wal.h index bef7301a07..a0f421212a 100644 --- a/include/libs/wal/wal.h +++ b/include/libs/wal/wal.h @@ -201,7 +201,7 @@ int32_t walFetchHead(SWalReader *pRead, int64_t ver, SWalCkHead *pHead); int32_t walFetchBody(SWalReader *pRead, SWalCkHead **ppHead); int32_t walSkipFetchBody(SWalReader *pRead, const SWalCkHead *pHead); -SWalRef *walRefFirstVer(SWal *); +SWalRef *walRefFirstVer(SWal *, SWalRef *); SWalRef *walRefCommittedVer(SWal *); SWalRef *walOpenRef(SWal *); diff --git a/source/dnode/vnode/src/tq/tq.c b/source/dnode/vnode/src/tq/tq.c index 7649e8a006..b195cfafb0 100644 --- a/source/dnode/vnode/src/tq/tq.c +++ b/source/dnode/vnode/src/tq/tq.c @@ -521,9 +521,12 @@ int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg) { tqOffsetResetToData(&fetchOffsetNew, 0, 0); } } else { - int64_t firstVer = walGetFirstVer(pTq->pVnode->pWal); - walRefVer(pHandle->pRef, firstVer); - tqOffsetResetToLog(&fetchOffsetNew, firstVer - 1); + pHandle->pRef = walRefFirstVer(pTq->pVnode->pWal, pHandle->pRef); + if (pHandle->pRef == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + return -1; + } + tqOffsetResetToLog(&fetchOffsetNew, pHandle->pRef->refVer - 1); } } else if (reqOffset.type == TMQ_OFFSET__RESET_LATEST) { if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__COLUMN) { @@ -721,6 +724,8 @@ int32_t tqProcessPollReq(STQ* pTq, SRpcMsg* pMsg) { int32_t tqProcessDeleteSubReq(STQ* pTq, int64_t version, char* msg, int32_t msgLen) { SMqVDeleteReq* pReq = (SMqVDeleteReq*)msg; + tqDebug("vgId:%d, delete sub: %s", pTq->pVnode->config.vgId, pReq->subKey); + taosWLockLatch(&pTq->pushLock); int32_t code = taosHashRemove(pTq->pPushMgr, pReq->subKey, strlen(pReq->subKey)); if (code != 0) { diff --git a/source/libs/wal/src/walRef.c b/source/libs/wal/src/walRef.c index f5cfe9abae..43470f4c82 100644 --- a/source/libs/wal/src/walRef.c +++ b/source/libs/wal/src/walRef.c @@ -77,10 +77,12 @@ void walUnrefVer(SWalRef *pRef) { } #endif -SWalRef *walRefFirstVer(SWal *pWal) { - SWalRef *pRef = walOpenRef(pWal); +SWalRef *walRefFirstVer(SWal *pWal, SWalRef *pRef) { if (pRef == NULL) { - return NULL; + pRef = walOpenRef(pWal); + if (pRef == NULL) { + return NULL; + } } taosThreadMutexLock(&pWal->mutex); From 1a5adfc547c3fb25d4a6339b8047483bcc76c587 Mon Sep 17 00:00:00 2001 From: xinsheng Ren <285808407@qq.com> Date: Sat, 28 Jan 2023 17:29:38 +0800 Subject: [PATCH 032/124] fix:getTimeOfDay on windows (#19680) --- source/os/src/osTime.c | 47 ++++++++++++++---------------------------- 1 file changed, 15 insertions(+), 32 deletions(-) diff --git a/source/os/src/osTime.c b/source/os/src/osTime.c index cd4324a592..685693a709 100644 --- a/source/os/src/osTime.c +++ b/source/os/src/osTime.c @@ -33,6 +33,11 @@ #include //#define TM_YEAR_BASE 1970 //origin #define TM_YEAR_BASE 1900 // slguan + +// This magic number is the number of 100 nanosecond intervals since January 1, 1601 (UTC) +// until 00:00:00 January 1, 1970 +static const uint64_t TIMEEPOCH = ((uint64_t)116444736000000000ULL); + /* * We do not implement alternate representations. However, we always * check whether a given modifier is allowed for a certain conversion. @@ -341,15 +346,17 @@ char *taosStrpTime(const char *buf, const char *fmt, struct tm *tm) { int32_t taosGetTimeOfDay(struct timeval *tv) { #ifdef WINDOWS - time_t t; - t = taosGetTimestampSec(); - SYSTEMTIME st; - GetLocalTime(&st); + LARGE_INTEGER t; + FILETIME f; - tv->tv_sec = (long)t; - tv->tv_usec = st.wMilliseconds * 1000; + GetSystemTimeAsFileTime(&f); + t.QuadPart = f.dwHighDateTime; + t.QuadPart <<= 32; + t.QuadPart |= f.dwLowDateTime; - return 0; + t.QuadPart -= TIMEEPOCH; + tv->tv_sec = t.QuadPart / 10000000; + tv->tv_usec = (t.QuadPart % 10000000) / 10; #else return gettimeofday(tv, NULL); #endif @@ -550,37 +557,13 @@ int32_t taosClockGetTime(int clock_id, struct timespec *pTS) { #ifdef WINDOWS LARGE_INTEGER t; FILETIME f; - static FILETIME ff; - static SYSTEMTIME ss; - static LARGE_INTEGER offset; - - static int8_t offsetInit = 0; - static volatile bool offsetInitFinished = false; - int8_t old = atomic_val_compare_exchange_8(&offsetInit, 0, 1); - if (0 == old) { - ss.wYear = 1970; - ss.wMonth = 1; - ss.wDay = 1; - ss.wHour = 0; - ss.wMinute = 0; - ss.wSecond = 0; - ss.wMilliseconds = 0; - SystemTimeToFileTime(&ss, &ff); - offset.QuadPart = ff.dwHighDateTime; - offset.QuadPart <<= 32; - offset.QuadPart |= ff.dwLowDateTime; - offsetInitFinished = true; - } else { - while (!offsetInitFinished) - ; // Ensure initialization is completed. - } GetSystemTimeAsFileTime(&f); t.QuadPart = f.dwHighDateTime; t.QuadPart <<= 32; t.QuadPart |= f.dwLowDateTime; - t.QuadPart -= offset.QuadPart; + t.QuadPart -= TIMEEPOCH; pTS->tv_sec = t.QuadPart / 10000000; pTS->tv_nsec = (t.QuadPart % 10000000) * 100; return (0); From b996618e65abd0ca3aa881a92df2ae99c3b30961 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sat, 28 Jan 2023 18:03:13 +0800 Subject: [PATCH 033/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 312 ++++++--------------- 1 file changed, 93 insertions(+), 219 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 6096936c29..c519e92727 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -16,6 +16,9 @@ #include "tsdb.h" extern int32_t tsdbReadDataBlockEx(SDataFReader* pReader, SDataBlk* pDataBlk, SBlockData* pBlockData); +extern int32_t tsdbUpdateTableSchema(SMeta* pMeta, int64_t suid, int64_t uid, SSkmInfo* pSkmInfo); +extern int32_t tsdbWriteDataBlock(SDataFWriter* pWriter, SBlockData* pBlockData, SMapData* mDataBlk, int8_t cmprAlg); +extern int32_t tsdbWriteSttBlock(SDataFWriter* pWriter, SBlockData* pBlockData, SArray* aSttBlk, int8_t cmprAlg); // STsdbDataIter2 ======================================== #define TSDB_MEM_TABLE_DATA_ITER 0 @@ -218,6 +221,12 @@ static void tsdbCloseSttFileDataIter(STsdbDataIter2* pIter) { taosMemoryFree(pIter); } +static void tsdbCloseTombFileDataIter(STsdbDataIter2* pIter) { + taosArrayDestroy(pIter->tIter.aDelData); + taosArrayDestroy(pIter->tIter.aDelIdx); + taosMemoryFree(pIter); +} + static void tsdbCloseDataIter2(STsdbDataIter2* pIter) { if (pIter->type == TSDB_MEM_TABLE_DATA_ITER) { ASSERT(0); @@ -225,6 +234,8 @@ static void tsdbCloseDataIter2(STsdbDataIter2* pIter) { tsdbCloseDataFileDataIter(pIter); } else if (pIter->type == TSDB_STT_FILE_DATA_ITER) { tsdbCloseSttFileDataIter(pIter); + } else if (pIter->type == TSDB_TOMB_FILE_DATA_ITER) { + tsdbCloseTombFileDataIter(pIter); } else { ASSERT(0); } @@ -347,54 +358,31 @@ static int32_t tsdbDataIterNext2(STsdbDataIter2* pIter) { /* get */ // STsdbSnapReader ======================================== -typedef enum { SNAP_DATA_FILE_ITER = 0, SNAP_STT_FILE_ITER } EFIterT; -typedef struct { - SRBTreeNode n; - SRowInfo rInfo; - EFIterT type; - union { - struct { - SArray* aBlockIdx; - int32_t iBlockIdx; - SBlockIdx* pBlockIdx; - SMapData mBlock; - int32_t iBlock; - }; // .data file - struct { - int32_t iStt; - SArray* aSttBlk; - int32_t iSttBlk; - }; // .stt file - }; - SBlockData bData; - int32_t iRow; -} SFDataIter; - struct STsdbSnapReader { - STsdb* pTsdb; - int64_t sver; - int64_t ever; - STsdbFS fs; - int8_t type; - // for data file - int8_t dataDone; - int32_t fid; - SDataFReader* pDataFReader; - SFDataIter* pIter; - SRBTree rbt; - SFDataIter aFDataIter[TSDB_MAX_STT_TRIGGER + 1]; - SBlockData bData; - SSkmInfo skmTable; - // for del file - int8_t delDone; - SDelFReader* pDelFReader; - SArray* aDelIdx; // SArray - int32_t iDelIdx; - SArray* aDelData; // SArray - uint8_t* aBuf[5]; -}; + STsdb* pTsdb; + int64_t sver; + int64_t ever; + int8_t type; + uint8_t* aBuf[5]; -extern int32_t tsdbUpdateTableSchema(SMeta* pMeta, int64_t suid, int64_t uid, SSkmInfo* pSkmInfo); + STsdbFS fs; + SSkmInfo skmTable; + + // timeseries data + int8_t dataDone; + int32_t fid; + + SDataFReader* pDataFReader; + STsdbDataIter2* iterList; + STsdbDataIter2* pIter; + SRBTree rbt; + SBlockData bData; + + // tombstone data + int8_t delDone; + SDelFReader* pDelFReader; + STsdbDataIter2* pTIter; +}; static int32_t tFDataIterCmprFn(const SRBTreeNode* pNode1, const SRBTreeNode* pNode2) { SFDataIter* pIter1 = (SFDataIter*)(((uint8_t*)pNode1) - offsetof(SFDataIter, n)); @@ -647,7 +635,7 @@ _exit: return code; } -static int32_t tsdbSnapReadData(STsdbSnapReader* pReader, uint8_t** ppData) { +static int32_t tsdbSnapReadTimeSeriesData(STsdbSnapReader* pReader, uint8_t** ppData) { int32_t code = 0; int32_t lino = 0; @@ -705,7 +693,7 @@ _exit: return code; } -static int32_t tsdbSnapReadDel(STsdbSnapReader* pReader, uint8_t** ppData) { +static int32_t tsdbSnapReadTombstoneData(STsdbSnapReader* pReader, uint8_t** ppData) { int32_t code = 0; int32_t lino = 0; @@ -790,12 +778,11 @@ _exit: } int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, int8_t type, STsdbSnapReader** ppReader) { - int32_t code = 0; - int32_t lino = 0; - STsdbSnapReader* pReader = NULL; + int32_t code = 0; + int32_t lino = 0; // alloc - pReader = (STsdbSnapReader*)taosMemoryCalloc(1, sizeof(*pReader)); + STsdbSnapReader* pReader = (STsdbSnapReader*)taosMemoryCalloc(1, sizeof(*pReader)); if (pReader == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; TSDB_CHECK_CODE(code, lino, _exit); @@ -805,118 +792,78 @@ int32_t tsdbSnapReaderOpen(STsdb* pTsdb, int64_t sver, int64_t ever, int8_t type pReader->ever = ever; pReader->type = type; - code = taosThreadRwlockRdlock(&pTsdb->rwLock); - if (code) { - code = TAOS_SYSTEM_ERROR(code); - TSDB_CHECK_CODE(code, lino, _exit); - } - + taosThreadRwlockRdlock(&pTsdb->rwLock); code = tsdbFSRef(pTsdb, &pReader->fs); if (code) { taosThreadRwlockUnlock(&pTsdb->rwLock); TSDB_CHECK_CODE(code, lino, _exit); } + taosThreadRwlockUnlock(&pTsdb->rwLock); - code = taosThreadRwlockUnlock(&pTsdb->rwLock); - if (code) { - code = TAOS_SYSTEM_ERROR(code); - TSDB_CHECK_CODE(code, lino, _exit); - } - - // data + // init pReader->fid = INT32_MIN; - for (int32_t iIter = 0; iIter < sizeof(pReader->aFDataIter) / sizeof(pReader->aFDataIter[0]); iIter++) { - SFDataIter* pIter = &pReader->aFDataIter[iIter]; - - if (iIter == 0) { - pIter->aBlockIdx = taosArrayInit(0, sizeof(SBlockIdx)); - if (pIter->aBlockIdx == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - } else { - pIter->aSttBlk = taosArrayInit(0, sizeof(SSttBlk)); - if (pIter->aSttBlk == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - } - - code = tBlockDataCreate(&pIter->bData); - TSDB_CHECK_CODE(code, lino, _exit); - } code = tBlockDataCreate(&pReader->bData); TSDB_CHECK_CODE(code, lino, _exit); - // del - pReader->aDelIdx = taosArrayInit(0, sizeof(SDelIdx)); - if (pReader->aDelIdx == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - pReader->aDelData = taosArrayInit(0, sizeof(SDelData)); - if (pReader->aDelData == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - _exit: if (code) { - tsdbError("vgId:%d, %s failed at line %d since %s, TSDB path: %s", TD_VID(pTsdb->pVnode), __func__, lino, - tstrerror(code), pTsdb->path); - *ppReader = NULL; - + tsdbError("vgId:%d %s failed at line %d since %s, sver:%" PRId64 " ever:%" PRId64 " type:%d", TD_VID(pTsdb->pVnode), + __func__, lino, tstrerror(code), sver, ever, type); if (pReader) { - taosArrayDestroy(pReader->aDelData); - taosArrayDestroy(pReader->aDelIdx); tBlockDataDestroy(&pReader->bData, 1); - tsdbFSDestroy(&pReader->fs); + tsdbFSUnref(pTsdb, &pReader->fs); taosMemoryFree(pReader); + pReader = NULL; } } else { - *ppReader = pReader; - tsdbInfo("vgId:%d, vnode snapshot tsdb reader opened for %s", TD_VID(pTsdb->pVnode), pTsdb->path); + tsdbInfo("vgId:%d %s done, sver:%" PRId64 " ever:%" PRId64 " type:%d", TD_VID(pTsdb->pVnode), __func__, sver, ever, + type); } + *ppReader = pReader; return code; } int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader) { - int32_t code = 0; + int32_t code = 0; + int32_t lino = 0; + STsdbSnapReader* pReader = *ppReader; - // data - if (pReader->pDataFReader) tsdbDataFReaderClose(&pReader->pDataFReader); - for (int32_t iIter = 0; iIter < sizeof(pReader->aFDataIter) / sizeof(pReader->aFDataIter[0]); iIter++) { - SFDataIter* pIter = &pReader->aFDataIter[iIter]; - - if (iIter == 0) { - taosArrayDestroy(pIter->aBlockIdx); - tMapDataClear(&pIter->mBlock); - } else { - taosArrayDestroy(pIter->aSttBlk); - } - - tBlockDataDestroy(&pIter->bData, 1); + // tombstone + if (pReader->pTIter) { + tsdbCloseDataIter2(pReader->pTIter); + pReader->pTIter = NULL; + } + if (pReader->pDelFReader) { + tsdbDelFReaderClose(&pReader->pDelFReader); } + // timeseries tBlockDataDestroy(&pReader->bData, 1); + while (pReader->iterList) { + STsdbDataIter2* pIter = pReader->iterList; + pReader->iterList = pIter->next; + tsdbCloseDataIter2(pIter); + } + if (pReader->pDataFReader) { + tsdbDataFReaderClose(&pReader->pDataFReader); + } + + // other tDestroyTSchema(pReader->skmTable.pTSchema); - - // del - if (pReader->pDelFReader) tsdbDelFReaderClose(&pReader->pDelFReader); - taosArrayDestroy(pReader->aDelIdx); - taosArrayDestroy(pReader->aDelData); - tsdbFSUnref(pReader->pTsdb, &pReader->fs); - - tsdbInfo("vgId:%d, vnode snapshot tsdb reader closed for %s", TD_VID(pReader->pTsdb->pVnode), pReader->pTsdb->path); - for (int32_t iBuf = 0; iBuf < sizeof(pReader->aBuf) / sizeof(pReader->aBuf[0]); iBuf++) { tFree(pReader->aBuf[iBuf]); } - taosMemoryFree(pReader); + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s done", TD_VID(pReader->pTsdb->pVnode), __func__); + } *ppReader = NULL; return code; } @@ -929,7 +876,7 @@ int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData) { // read data file if (!pReader->dataDone) { - code = tsdbSnapReadData(pReader, ppData); + code = tsdbSnapReadTimeSeriesData(pReader, ppData); TSDB_CHECK_CODE(code, lino, _exit); if (*ppData) { goto _exit; @@ -940,7 +887,7 @@ int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData) { // read del file if (!pReader->delDone) { - code = tsdbSnapReadDel(pReader, ppData); + code = tsdbSnapReadTombstoneData(pReader, ppData); TSDB_CHECK_CODE(code, lino, _exit); if (*ppData) { goto _exit; @@ -951,10 +898,9 @@ int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData) { _exit: if (code) { - tsdbError("vgId:%d, %s failed since %s, path:%s", TD_VID(pReader->pTsdb->pVnode), __func__, tstrerror(code), - pReader->pTsdb->path); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); } else { - tsdbDebug("vgId:%d, %s done, path:%s", TD_VID(pReader->pTsdb->pVnode), __func__, pReader->pTsdb->path); + tsdbDebug("vgId:%d %s done", TD_VID(pReader->pTsdb->pVnode), __func__); } return code; } @@ -1008,69 +954,6 @@ struct STsdbSnapWriter { }; // SNAP_DATA_TSDB -extern int32_t tsdbWriteDataBlock(SDataFWriter* pWriter, SBlockData* pBlockData, SMapData* mDataBlk, int8_t cmprAlg); -extern int32_t tsdbWriteSttBlock(SDataFWriter* pWriter, SBlockData* pBlockData, SArray* aSttBlk, int8_t cmprAlg); - -static int32_t tsdbSnapNextTableData(STsdbSnapWriter* pWriter) { - int32_t code = 0; - int32_t lino = 0; - -#if 0 - ASSERT(pWriter->dReader.iRow >= pWriter->dReader.bData.nRow); - - if (pWriter->dReader.iBlockIdx < taosArrayGetSize(pWriter->dReader.aBlockIdx)) { - pWriter->dReader.pBlockIdx = (SBlockIdx*)taosArrayGet(pWriter->dReader.aBlockIdx, pWriter->dReader.iBlockIdx); - - code = tsdbReadDataBlk(pWriter->dReader.pReader, pWriter->dReader.pBlockIdx, &pWriter->dReader.mDataBlk); - if (code) goto _exit; - - pWriter->dReader.iBlockIdx++; - } else { - pWriter->dReader.pBlockIdx = NULL; - tMapDataReset(&pWriter->dReader.mDataBlk); - } - pWriter->dReader.iDataBlk = 0; // point to the next one - tBlockDataReset(&pWriter->dReader.bData); - pWriter->dReader.iRow = 0; -#endif - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - -static int32_t tsdbSnapWriteCopyData(STsdbSnapWriter* pWriter, TABLEID* pId) { - int32_t code = 0; - int32_t lino = 0; - -#if 0 - while (true) { - if (pWriter->dReader.pBlockIdx == NULL) break; - if (tTABLEIDCmprFn(pWriter->dReader.pBlockIdx, pId) >= 0) break; - - SBlockIdx blkIdx = *pWriter->dReader.pBlockIdx; - code = tsdbWriteDataBlk(pWriter->dWriter.pWriter, &pWriter->dReader.mDataBlk, &blkIdx); - if (code) goto _exit; - - if (taosArrayPush(pWriter->dWriter.aBlockIdx, &blkIdx) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - goto _exit; - } - - code = tsdbSnapNextTableData(pWriter); - if (code) goto _exit; - } -#endif - -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); - } - return code; -} - static int32_t tsdbSnapWriteTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pId) { int32_t code = 0; int32_t lino = 0; @@ -1622,14 +1505,10 @@ static int32_t tsdbSnapWriteDelTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pWriter->tbid = (TABLEID){.suid = INT64_MAX, .uid = INT64_MAX}; } + taosArrayClear(pWriter->aDelData); + if (pWriter->pTIter) { - ASSERT(pWriter->pTIter->tIter.iDelData >= taosArrayGetSize(pWriter->pTIter->tIter.aDelData)); - - for (;;) { - if (pWriter->pTIter->tIter.iDelIdx >= taosArrayGetSize(pWriter->pTIter->tIter.aDelIdx)) { - break; - } - + while (pWriter->pTIter->tIter.iDelIdx < taosArrayGetSize(pWriter->pTIter->tIter.aDelIdx)) { SDelIdx* pDelIdx = taosArrayGet(pWriter->pTIter->tIter.aDelIdx, pWriter->pTIter->tIter.iDelIdx); int32_t c = tTABLEIDCmprFn(pDelIdx, &pWriter->tbid); @@ -1637,7 +1516,7 @@ static int32_t tsdbSnapWriteDelTableDataStart(STsdbSnapWriter* pWriter, TABLEID* code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->pTIter->tIter.aDelData); TSDB_CHECK_CODE(code, lino, _exit); - SDelIdx* pDelIdxNew = taosArrayReserve(pWriter->pTIter->tIter.aDelIdx, 1); + SDelIdx* pDelIdxNew = taosArrayReserve(pWriter->aDelIdx, 1); if (pDelIdxNew == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; TSDB_CHECK_CODE(code, lino, _exit); @@ -1651,27 +1530,17 @@ static int32_t tsdbSnapWriteDelTableDataStart(STsdbSnapWriter* pWriter, TABLEID* pWriter->pTIter->tIter.iDelIdx++; } else if (c == 0) { - code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->pTIter->tIter.aDelData); + code = tsdbReadDelData(pWriter->pDelFReader, pDelIdx, pWriter->aDelData); TSDB_CHECK_CODE(code, lino, _exit); - if (taosArrayAddBatch(pWriter->aDelData, pWriter->pTIter->tIter.aDelData->pData, - taosArrayGetSize(pWriter->pTIter->tIter.aDelData)) == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - - pWriter->pTIter->tIter.iDelData = taosArrayGetSize(pWriter->pTIter->tIter.aDelData); pWriter->pTIter->tIter.iDelIdx++; break; } else { - pWriter->pTIter->tIter.iDelData = taosArrayGetSize(pWriter->pTIter->tIter.aDelData); break; } } } - taosArrayClear(pWriter->aDelData); - _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); @@ -1735,7 +1604,6 @@ static int32_t tsdbSnapWriteDelTableData(STsdbSnapWriter* pWriter, TABLEID* pId, TSDB_CHECK_CODE(code, lino, _exit); } } - ASSERT(n == size); _exit: @@ -1791,7 +1659,7 @@ static int32_t tsdbSnapWriteDelDataEnd(STsdbSnapWriter* pWriter) { STsdb* pTsdb = pWriter->pTsdb; - // end remaining table (TODO) + // end remaining table with NULL data code = tsdbSnapWriteDelTableData(pWriter, NULL, NULL, 0); TSDB_CHECK_CODE(code, lino, _exit); @@ -1813,6 +1681,11 @@ static int32_t tsdbSnapWriteDelDataEnd(STsdbSnapWriter* pWriter) { TSDB_CHECK_CODE(code, lino, _exit); } + if (pWriter->pTIter) { + tsdbCloseDataIter2(pWriter->pTIter); + pWriter->pTIter = NULL; + } + _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); @@ -1834,6 +1707,7 @@ static int32_t tsdbSnapWriteDelData(STsdbSnapWriter* pWriter, SSnapDataHdr* pHdr TSDB_CHECK_CODE(code, lino, _exit); } + // do write del data code = tsdbSnapWriteDelTableData(pWriter, (TABLEID*)pHdr->data, pHdr->data + sizeof(TABLEID), pHdr->size - sizeof(TABLEID)); TSDB_CHECK_CODE(code, lino, _exit); From 9ec79d7ef84a8cb8a91db661e475e1a1828f3d6f Mon Sep 17 00:00:00 2001 From: plum-lihui Date: Sat, 28 Jan 2023 18:15:11 +0800 Subject: [PATCH 034/124] test:modify expect consume rows --- tests/system-test/7-tmq/tmqUpdate-1ctb.py | 33 +++++++--------- .../7-tmq/tmqUpdate-multiCtb-snapshot0.py | 38 +++++++++---------- 2 files changed, 31 insertions(+), 40 deletions(-) diff --git a/tests/system-test/7-tmq/tmqUpdate-1ctb.py b/tests/system-test/7-tmq/tmqUpdate-1ctb.py index f03fa84335..db2ec3285d 100644 --- a/tests/system-test/7-tmq/tmqUpdate-1ctb.py +++ b/tests/system-test/7-tmq/tmqUpdate-1ctb.py @@ -100,14 +100,6 @@ class TDTestCase: # update to half tables paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) - - tdLog.info("create topics from stb1") - topicFromStb1 = 'topic_stb1' - queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) - sqlString = "create topic %s as %s" %(topicFromStb1, queryString) - tdLog.info("create topic sql: %s"%sqlString) - tdSql.execute(sqlString) - # tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix="ctbx", # ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], # startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) @@ -115,6 +107,12 @@ class TDTestCase: ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) + tdLog.info("create topics from stb1") + topicFromStb1 = 'topic_stb1' + queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as %s" %(topicFromStb1, queryString) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) # paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl @@ -158,7 +156,6 @@ class TDTestCase: def tmqCase2(self): tdLog.printNoPrefix("======== test case 2: ") - paraDict = {'dbName': 'dbt', 'dropFlag': 1, 'event': '', @@ -178,7 +175,7 @@ class TDTestCase: 'showMsg': 1, 'showRow': 1, 'snapshot': 0} - self.snapshot = 1 + paraDict['snapshot'] = self.snapshot paraDict['vgroups'] = self.vgroups paraDict['ctbNum'] = self.ctbNum @@ -190,6 +187,12 @@ class TDTestCase: # update to half tables paraDict['startTs'] = paraDict['startTs'] + int(self.rowsPerTbl / 2) paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) + tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict["ctbPrefix"], + ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], + startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) + # tmqCom.insert_data_interlaceByMultiTbl(tsql=tdSql,dbName=paraDict["dbName"],ctbPrefix=paraDict["ctbPrefix"], + # ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], + # startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) tmqCom.initConsumerTable() tdLog.info("create topics from stb1") @@ -199,19 +202,11 @@ class TDTestCase: tdLog.info("create topic sql: %s"%sqlString) tdSql.execute(sqlString) - tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict["ctbPrefix"], - ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], - startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) - # tmqCom.insert_data_interlaceByMultiTbl(tsql=tdSql,dbName=paraDict["dbName"],ctbPrefix=paraDict["ctbPrefix"], - # ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], - # startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) - - # paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl consumerId = 1 if self.snapshot == 0: - expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (2)) + expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (1/2)) elif self.snapshot == 1: expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (1)) diff --git a/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py b/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py index 7310466abe..daffff44c1 100644 --- a/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py +++ b/tests/system-test/7-tmq/tmqUpdate-multiCtb-snapshot0.py @@ -101,14 +101,6 @@ class TDTestCase: # update to half tables paraDict['ctbNum'] = int(self.ctbNum/2) paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) - - tdLog.info("create topics from stb1") - topicFromStb1 = 'topic_stb1' - queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) - sqlString = "create topic %s as %s" %(topicFromStb1, queryString) - tdLog.info("create topic sql: %s"%sqlString) - tdSql.execute(sqlString) - tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=self.autoCtbPrefix, ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) @@ -116,6 +108,13 @@ class TDTestCase: ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']) + tdLog.info("create topics from stb1") + topicFromStb1 = 'topic_stb1' + queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as %s" %(topicFromStb1, queryString) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl consumerId = 0 @@ -177,7 +176,6 @@ class TDTestCase: 'showRow': 1, 'snapshot': 0} - self.snapshot = 1 paraDict['snapshot'] = self.snapshot paraDict['vgroups'] = self.vgroups paraDict['ctbNum'] = self.ctbNum @@ -190,16 +188,6 @@ class TDTestCase: paraDict['ctbNum'] = int(self.ctbNum/2) paraDict['rowsPerTbl'] = int(self.rowsPerTbl / 2) paraDict['startTs'] = paraDict['startTs'] + int(self.rowsPerTbl / 2) - - tmqCom.initConsumerTable() - tdLog.info("create topics from stb1") - topicFromStb1 = 'topic_stb1' - queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) - sqlString = "create topic %s as %s" %(topicFromStb1, queryString) - tdLog.info("create topic sql: %s"%sqlString) - tdSql.execute(sqlString) - - tmqCom.insert_data_with_autoCreateTbl(tsql=tdSql,dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=self.autoCtbPrefix, ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']+int(self.ctbNum/2)) @@ -212,14 +200,22 @@ class TDTestCase: ctbNum=paraDict["ctbNum"],rowsPerTbl=paraDict["rowsPerTbl"],batchNum=paraDict["batchNum"], startTs=paraDict["startTs"],ctbStartIdx=paraDict['ctbStartIdx']+int(self.ctbNum/2)) + tmqCom.initConsumerTable() + tdLog.info("create topics from stb1") + topicFromStb1 = 'topic_stb1' + queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as %s" %(topicFromStb1, queryString) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + # paraDict['ctbNum'] = self.ctbNum paraDict['ctbNum'] = self.ctbNum paraDict['rowsPerTbl'] = self.rowsPerTbl consumerId = 1 if self.snapshot == 0: - expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (2 + 1/2*1/2*2 + 1/2*1/2)) + expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (1/2) * (1/2*3)) elif self.snapshot == 1: - expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (2 + 1/2*1/2)) + expectrowcnt = int(paraDict["rowsPerTbl"] * paraDict["ctbNum"] * (1 + 1/2)) topicList = topicFromStb1 ifcheckdata = 1 From ded3d6e7c9e2bd80ac64520b6724e9446b3cbb3b Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sun, 29 Jan 2023 09:08:02 +0800 Subject: [PATCH 035/124] more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 588 +++++++++++---------- 1 file changed, 310 insertions(+), 278 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index c519e92727..b14fbdf189 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -26,7 +26,9 @@ extern int32_t tsdbWriteSttBlock(SDataFWriter* pWriter, SBlockData* pBlockData, #define TSDB_STT_FILE_DATA_ITER 2 #define TSDB_TOMB_FILE_DATA_ITER 3 -typedef struct STsdbDataIter2 STsdbDataIter2; +typedef struct STsdbDataIter2 STsdbDataIter2; +typedef struct STsdbFilterInfo STsdbFilterInfo; + struct STsdbDataIter2 { STsdbDataIter2* next; SRBTreeNode rbtn; @@ -70,6 +72,13 @@ struct STsdbDataIter2 { }; }; +#define TSDB_FILTER_FLAG_BY_VERSION 0x1 +struct STsdbFilterInfo { + int32_t flag; + int64_t sver; + int64_t ever; +}; + #define TSDB_RBTN_TO_DATA_ITER(pNode) ((STsdbDataIter2*)(((char*)pNode) - offsetof(STsdbDataIter2, rbtn))) /* open */ @@ -251,24 +260,44 @@ static int32_t tsdbDataIterCmprFn(const SRBTreeNode* pNode1, const SRBTreeNode* /* seek */ /* iter next */ -static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter) { +static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo* pFilterInfo) { int32_t code = 0; int32_t lino = 0; for (;;) { - if (pIter->dIter.iRow < pIter->dIter.bData.nRow) { + while (pIter->dIter.iRow < pIter->dIter.bData.nRow) { + if (pFilterInfo) { + if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) { + if (pIter->dIter.bData.aVersion[pIter->dIter.iRow] < pFilterInfo->sver || + pIter->dIter.bData.aVersion[pIter->dIter.iRow] > pFilterInfo->ever) { + pIter->dIter.iRow++; + continue; + } + } + } + pIter->rowInfo.suid = pIter->dIter.bData.suid; pIter->rowInfo.uid = pIter->dIter.bData.uid; pIter->rowInfo.row = tsdbRowFromBlockData(&pIter->dIter.bData, pIter->dIter.iRow); pIter->dIter.iRow++; - break; + goto _exit; } for (;;) { - if (pIter->dIter.iDataBlk < pIter->dIter.mDataBlk.nItem) { + while (pIter->dIter.iDataBlk < pIter->dIter.mDataBlk.nItem) { SDataBlk dataBlk; tMapDataGetItemByIdx(&pIter->dIter.mDataBlk, pIter->dIter.iDataBlk, &dataBlk, tGetDataBlk); + // filter + if (pFilterInfo) { + if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) { + if (pFilterInfo->sver > dataBlk.maxVer || pFilterInfo->ever < dataBlk.minVer) { + pIter->dIter.iDataBlk++; + continue; + } + } + } + code = tsdbReadDataBlockEx(pIter->dIter.pReader, &dataBlk, &pIter->dIter.bData); TSDB_CHECK_CODE(code, lino, _exit); @@ -278,6 +307,8 @@ static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter) { break; } + if (pIter->dIter.iRow < pIter->dIter.bData.nRow) break; + for (;;) { if (pIter->dIter.iBlockIdx < taosArrayGetSize(pIter->dIter.aBlockIdx)) { SBlockIdx* pBlockIdx = taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx); @@ -304,31 +335,52 @@ _exit: return code; } -static int32_t tsdbSttFileDataIterNext(STsdbDataIter2* pIter) { +static int32_t tsdbSttFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo* pFilterInfo) { int32_t code = 0; int32_t lino = 0; for (;;) { - if (pIter->sIter.iRow < pIter->sIter.bData.nRow) { + while (pIter->sIter.iRow < pIter->sIter.bData.nRow) { + if (pFilterInfo) { + if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) { + if (pFilterInfo->sver > pIter->sIter.bData.aVersion[pIter->sIter.iRow] || + pFilterInfo->ever < pIter->sIter.bData.aVersion[pIter->sIter.iRow]) { + pIter->sIter.iRow++; + continue; + } + } + } + pIter->rowInfo.suid = pIter->sIter.bData.suid; pIter->rowInfo.uid = pIter->sIter.bData.uid ? pIter->sIter.bData.uid : pIter->sIter.bData.aUid[pIter->sIter.iRow]; pIter->rowInfo.row = tsdbRowFromBlockData(&pIter->sIter.bData, pIter->sIter.iRow); pIter->sIter.iRow++; - break; + goto _exit; } - if (pIter->sIter.iSttBlk < taosArrayGetSize(pIter->sIter.aSttBlk)) { - SSttBlk* pSttBlk = taosArrayGet(pIter->sIter.aSttBlk, pIter->sIter.iSttBlk); + for (;;) { + if (pIter->sIter.iSttBlk < taosArrayGetSize(pIter->sIter.aSttBlk)) { + SSttBlk* pSttBlk = taosArrayGet(pIter->sIter.aSttBlk, pIter->sIter.iSttBlk); - code = tsdbReadSttBlockEx(pIter->sIter.pReader, pIter->sIter.iStt, pSttBlk, &pIter->sIter.bData); - TSDB_CHECK_CODE(code, lino, _exit); + if (pFilterInfo) { + if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) { + if (pFilterInfo->sver > pSttBlk->maxVer || pFilterInfo->ever < pSttBlk->minVer) { + pIter->sIter.iSttBlk++; + continue; + } + } + } - pIter->sIter.iSttBlk++; + code = tsdbReadSttBlockEx(pIter->sIter.pReader, pIter->sIter.iStt, pSttBlk, &pIter->sIter.bData); + TSDB_CHECK_CODE(code, lino, _exit); - pIter->sIter.iRow = 0; - } else { - pIter->rowInfo = (SRowInfo){0}; - break; + pIter->sIter.iRow = 0; + pIter->sIter.iSttBlk++; + break; + } else { + pIter->rowInfo = (SRowInfo){0}; + goto _exit; + } } } @@ -339,16 +391,16 @@ _exit: return code; } -static int32_t tsdbDataIterNext2(STsdbDataIter2* pIter) { +static int32_t tsdbDataIterNext2(STsdbDataIter2* pIter, STsdbFilterInfo* pFilterInfo) { int32_t code = 0; if (pIter->type == TSDB_MEM_TABLE_DATA_ITER) { ASSERT(0); return code; } else if (pIter->type == TSDB_DATA_FILE_DATA_ITER) { - return tsdbDataFileDataIterNext(pIter); + return tsdbDataFileDataIterNext(pIter, pFilterInfo); } else if (pIter->type == TSDB_STT_FILE_DATA_ITER) { - return tsdbSttFileDataIterNext(pIter); + return tsdbSttFileDataIterNext(pIter, pFilterInfo); } else { ASSERT(0); return code; @@ -366,6 +418,7 @@ struct STsdbSnapReader { uint8_t* aBuf[5]; STsdbFS fs; + TABLEID tbid; SSkmInfo skmTable; // timeseries data @@ -382,224 +435,151 @@ struct STsdbSnapReader { int8_t delDone; SDelFReader* pDelFReader; STsdbDataIter2* pTIter; + SArray* aDelData; }; -static int32_t tFDataIterCmprFn(const SRBTreeNode* pNode1, const SRBTreeNode* pNode2) { - SFDataIter* pIter1 = (SFDataIter*)(((uint8_t*)pNode1) - offsetof(SFDataIter, n)); - SFDataIter* pIter2 = (SFDataIter*)(((uint8_t*)pNode2) - offsetof(SFDataIter, n)); - - return tRowInfoCmprFn(&pIter1->rInfo, &pIter2->rInfo); -} - -static int32_t tsdbSnapReadOpenFile(STsdbSnapReader* pReader) { +static int32_t tsdbSnapReadFileDataStart(STsdbSnapReader* pReader) { int32_t code = 0; int32_t lino = 0; - SDFileSet dFileSet = {.fid = pReader->fid}; - SDFileSet* pSet = taosArraySearch(pReader->fs.aDFileSet, &dFileSet, tDFileSetCmprFn, TD_GT); - if (pSet == NULL) return code; + SDFileSet* pSet = taosArraySearch(pReader->fs.aDFileSet, &(SDFileSet){.fid = pReader->fid}, tDFileSetCmprFn, TD_GT); + if (pSet == NULL) { + pReader->fid = INT32_MAX; + goto _exit; + } pReader->fid = pSet->fid; + + tRBTreeCreate(&pReader->rbt, tsdbDataIterCmprFn); + code = tsdbDataFReaderOpen(&pReader->pDataFReader, pReader->pTsdb, pSet); TSDB_CHECK_CODE(code, lino, _exit); - pReader->pIter = NULL; - tRBTreeCreate(&pReader->rbt, tFDataIterCmprFn); - - // .data file - SFDataIter* pIter = &pReader->aFDataIter[0]; - pIter->type = SNAP_DATA_FILE_ITER; - - code = tsdbReadBlockIdx(pReader->pDataFReader, pIter->aBlockIdx); + code = tsdbOpenDataFileDataIter(pReader->pDataFReader, &pReader->pIter); TSDB_CHECK_CODE(code, lino, _exit); - for (pIter->iBlockIdx = 0; pIter->iBlockIdx < taosArrayGetSize(pIter->aBlockIdx); pIter->iBlockIdx++) { - pIter->pBlockIdx = (SBlockIdx*)taosArrayGet(pIter->aBlockIdx, pIter->iBlockIdx); - - code = tsdbReadDataBlk(pReader->pDataFReader, pIter->pBlockIdx, &pIter->mBlock); + if (pReader->pIter) { + // iter to next with filter info (sver, ever) + code = tsdbDataIterNext2(pReader->pIter, + &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, // flag + .sver = pReader->sver, + .ever = pReader->ever}); TSDB_CHECK_CODE(code, lino, _exit); - for (pIter->iBlock = 0; pIter->iBlock < pIter->mBlock.nItem; pIter->iBlock++) { - SDataBlk dataBlk; - tMapDataGetItemByIdx(&pIter->mBlock, pIter->iBlock, &dataBlk, tGetDataBlk); + if (pReader->pIter->rowInfo.suid || pReader->pIter->rowInfo.uid) { + // add to rbtree + tRBTreePut(&pReader->rbt, &pReader->pIter->rbtn); - if (dataBlk.minVer > pReader->ever || dataBlk.maxVer < pReader->sver) continue; - - code = tsdbReadDataBlockEx(pReader->pDataFReader, &dataBlk, &pIter->bData); - TSDB_CHECK_CODE(code, lino, _exit); - - ASSERT(pIter->pBlockIdx->suid == pIter->bData.suid); - ASSERT(pIter->pBlockIdx->uid == pIter->bData.uid); - - for (pIter->iRow = 0; pIter->iRow < pIter->bData.nRow; pIter->iRow++) { - int64_t rowVer = pIter->bData.aVersion[pIter->iRow]; - - if (rowVer >= pReader->sver && rowVer <= pReader->ever) { - pIter->rInfo.suid = pIter->pBlockIdx->suid; - pIter->rInfo.uid = pIter->pBlockIdx->uid; - pIter->rInfo.row = tsdbRowFromBlockData(&pIter->bData, pIter->iRow); - goto _add_iter_and_break; - } - } + // add to iterList + pReader->pIter->next = pReader->iterList; + pReader->iterList = pReader->pIter; + } else { + tsdbCloseDataIter2(pReader->pIter); } - - continue; - - _add_iter_and_break: - tRBTreePut(&pReader->rbt, (SRBTreeNode*)pIter); - break; } - // .stt file - pIter = &pReader->aFDataIter[1]; - for (int32_t iStt = 0; iStt < pSet->nSttF; iStt++) { - pIter->type = SNAP_STT_FILE_ITER; - pIter->iStt = iStt; - - code = tsdbReadSttBlk(pReader->pDataFReader, iStt, pIter->aSttBlk); + for (int32_t iStt = 0; iStt < pSet->nSttF; ++iStt) { + code = tsdbOpenSttFileDataIter(pReader->pDataFReader, iStt, &pReader->pIter); TSDB_CHECK_CODE(code, lino, _exit); - for (pIter->iSttBlk = 0; pIter->iSttBlk < taosArrayGetSize(pIter->aSttBlk); pIter->iSttBlk++) { - SSttBlk* pSttBlk = (SSttBlk*)taosArrayGet(pIter->aSttBlk, pIter->iSttBlk); - - if (pSttBlk->minVer > pReader->ever) continue; - if (pSttBlk->maxVer < pReader->sver) continue; - - code = tsdbReadSttBlockEx(pReader->pDataFReader, iStt, pSttBlk, &pIter->bData); + if (pReader->pIter) { + // iter to valid row + code = tsdbDataIterNext2(pReader->pIter, + &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, // flag + .sver = pReader->sver, + .ever = pReader->ever}); TSDB_CHECK_CODE(code, lino, _exit); - for (pIter->iRow = 0; pIter->iRow < pIter->bData.nRow; pIter->iRow++) { - int64_t rowVer = pIter->bData.aVersion[pIter->iRow]; + if (pReader->pIter->rowInfo.suid || pReader->pIter->rowInfo.uid) { + // add to rbtree + tRBTreePut(&pReader->rbt, &pReader->pIter->rbtn); - if (rowVer >= pReader->sver && rowVer <= pReader->ever) { - pIter->rInfo.suid = pIter->bData.suid; - pIter->rInfo.uid = pIter->bData.uid ? pIter->bData.uid : pIter->bData.aUid[pIter->iRow]; - pIter->rInfo.row = tsdbRowFromBlockData(&pIter->bData, pIter->iRow); - goto _add_iter; - } + // add to iterList + pReader->pIter->next = pReader->iterList; + pReader->iterList = pReader->pIter; + } else { + tsdbCloseDataIter2(pReader->pIter); } } - - continue; - - _add_iter: - tRBTreePut(&pReader->rbt, (SRBTreeNode*)pIter); - pIter++; } + pReader->pIter = NULL; + _exit: if (code) { - tsdbError("vgId:%d, %s failed since %s", TD_VID(pReader->pTsdb->pVnode), __func__, tstrerror(code)); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); } else { - tsdbInfo("vgId:%d, %s done, path:%s, fid:%d", TD_VID(pReader->pTsdb->pVnode), __func__, pReader->pTsdb->path, - pReader->fid); + tsdbInfo("vgId:%d %s done, fid:%d", TD_VID(pReader->pTsdb->pVnode), __func__, pReader->fid); } return code; } -static int32_t tsdbSnapNextRow(STsdbSnapReader* pReader) { +static void tsdbSnapReadFileDataEnd(STsdbSnapReader* pReader) { + while (pReader->iterList) { + STsdbDataIter2* pIter = pReader->iterList; + pReader->iterList = pIter->next; + tsdbCloseDataIter2(pIter); + } + + tsdbDataFReaderClose(&pReader->pDataFReader); +} + +static int32_t tsdbSnapReadNextRow(STsdbSnapReader* pReader, SRowInfo** ppRowInfo) { int32_t code = 0; + int32_t lino = 0; if (pReader->pIter) { - SFDataIter* pIter = NULL; - while (true) { - _find_row: - pIter = pReader->pIter; - for (pIter->iRow++; pIter->iRow < pIter->bData.nRow; pIter->iRow++) { - int64_t rowVer = pIter->bData.aVersion[pIter->iRow]; + code = tsdbDataIterNext2(pReader->pIter, &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, // flag + .sver = pReader->sver, + .ever = pReader->ever}); + TSDB_CHECK_CODE(code, lino, _exit); - if (rowVer >= pReader->sver && rowVer <= pReader->ever) { - pIter->rInfo.suid = pIter->bData.suid; - pIter->rInfo.uid = pIter->bData.uid ? pIter->bData.uid : pIter->bData.aUid[pIter->iRow]; - pIter->rInfo.row = tsdbRowFromBlockData(&pIter->bData, pIter->iRow); - goto _out; + if (pReader->pIter->rowInfo.suid == 0 && pReader->pIter->rowInfo.uid == 0) { + pReader->pIter = NULL; + } else { + SRBTreeNode* pNode = tRBTreeMin(&pReader->rbt); + if (pNode) { + int32_t c = tsdbDataIterCmprFn(&pReader->pIter->rbtn, pNode); + if (c > 0) { + tRBTreePut(&pReader->rbt, &pReader->pIter->rbtn); + pReader->pIter = NULL; + } else if (c == 0) { + ASSERT(0); } } - - if (pIter->type == SNAP_DATA_FILE_ITER) { - while (true) { - for (pIter->iBlock++; pIter->iBlock < pIter->mBlock.nItem; pIter->iBlock++) { - SDataBlk dataBlk; - tMapDataGetItemByIdx(&pIter->mBlock, pIter->iBlock, &dataBlk, tGetDataBlk); - - if (dataBlk.minVer > pReader->ever || dataBlk.maxVer < pReader->sver) continue; - - code = tsdbReadDataBlockEx(pReader->pDataFReader, &dataBlk, &pIter->bData); - if (code) goto _err; - - pIter->iRow = -1; - goto _find_row; - } - - pIter->iBlockIdx++; - if (pIter->iBlockIdx >= taosArrayGetSize(pIter->aBlockIdx)) break; - - pIter->pBlockIdx = (SBlockIdx*)taosArrayGet(pIter->aBlockIdx, pIter->iBlockIdx); - code = tsdbReadDataBlk(pReader->pDataFReader, pIter->pBlockIdx, &pIter->mBlock); - if (code) goto _err; - pIter->iBlock = -1; - } - - pReader->pIter = NULL; - break; - } else if (pIter->type == SNAP_STT_FILE_ITER) { - for (pIter->iSttBlk++; pIter->iSttBlk < taosArrayGetSize(pIter->aSttBlk); pIter->iSttBlk++) { - SSttBlk* pSttBlk = (SSttBlk*)taosArrayGet(pIter->aSttBlk, pIter->iSttBlk); - - if (pSttBlk->minVer > pReader->ever || pSttBlk->maxVer < pReader->sver) continue; - - code = tsdbReadSttBlockEx(pReader->pDataFReader, pIter->iStt, pSttBlk, &pIter->bData); - if (code) goto _err; - - pIter->iRow = -1; - goto _find_row; - } - - pReader->pIter = NULL; - break; - } else { - ASSERT(0); - } - } - - _out: - pIter = (SFDataIter*)tRBTreeMin(&pReader->rbt); - if (pReader->pIter && pIter) { - int32_t c = tRowInfoCmprFn(&pReader->pIter->rInfo, &pIter->rInfo); - if (c > 0) { - tRBTreePut(&pReader->rbt, (SRBTreeNode*)pReader->pIter); - pReader->pIter = NULL; - } else { - ASSERT(c); - } } } if (pReader->pIter == NULL) { - pReader->pIter = (SFDataIter*)tRBTreeMin(&pReader->rbt); - if (pReader->pIter) { - tRBTreeDrop(&pReader->rbt, (SRBTreeNode*)pReader->pIter); + SRBTreeNode* pNode = tRBTreeMin(&pReader->rbt); + if (pNode) { + tRBTreeDrop(&pReader->rbt, pNode); + pReader->pIter = TSDB_RBTN_TO_DATA_ITER(pNode); } } - return code; + if (ppRowInfo) { + if (pReader->pIter) { + *ppRowInfo = &pReader->pIter->rowInfo; + } else { + *ppRowInfo = NULL; + } + } -_err: +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } return code; } -static SRowInfo* tsdbSnapGetRow(STsdbSnapReader* pReader) { +static int32_t tsdbSnapReadGetRow(STsdbSnapReader* pReader, SRowInfo** ppRowInfo) { if (pReader->pIter) { - return &pReader->pIter->rInfo; - } else { - tsdbSnapNextRow(pReader); - - if (pReader->pIter) { - return &pReader->pIter->rInfo; - } else { - return NULL; - } + *ppRowInfo = &pReader->pIter->rowInfo; + return 0; } + + return tsdbSnapReadNextRow(pReader, ppRowInfo); } static int32_t tsdbSnapCmprData(STsdbSnapReader* pReader, uint8_t** ppData) { @@ -641,138 +621,190 @@ static int32_t tsdbSnapReadTimeSeriesData(STsdbSnapReader* pReader, uint8_t** pp STsdb* pTsdb = pReader->pTsdb; - while (true) { + tBlockDataClear(&pReader->bData); + + for (;;) { + // start a new file read if need if (pReader->pDataFReader == NULL) { - code = tsdbSnapReadOpenFile(pReader); + code = tsdbSnapReadFileDataStart(pReader); TSDB_CHECK_CODE(code, lino, _exit); } if (pReader->pDataFReader == NULL) break; - SRowInfo* pRowInfo = tsdbSnapGetRow(pReader); + SRowInfo* pRowInfo; + code = tsdbSnapReadGetRow(pReader, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + if (pRowInfo == NULL) { - tsdbDataFReaderClose(&pReader->pDataFReader); + tsdbSnapReadFileDataEnd(pReader); continue; } - TABLEID id = {.suid = pRowInfo->suid, .uid = pRowInfo->uid}; - SBlockData* pBlockData = &pReader->bData; - - code = tsdbUpdateTableSchema(pTsdb->pVnode->pMeta, id.suid, id.uid, &pReader->skmTable); + code = tsdbUpdateTableSchema(pTsdb->pVnode->pMeta, pRowInfo->suid, pRowInfo->uid, &pReader->skmTable); TSDB_CHECK_CODE(code, lino, _exit); - code = tBlockDataInit(pBlockData, &id, pReader->skmTable.pTSchema, NULL, 0); + code = tBlockDataInit(&pReader->bData, (TABLEID*)pRowInfo, pReader->skmTable.pTSchema, NULL, 0); TSDB_CHECK_CODE(code, lino, _exit); - while (pRowInfo->suid == id.suid && pRowInfo->uid == id.uid) { - code = tBlockDataAppendRow(pBlockData, &pRowInfo->row, NULL, pRowInfo->uid); - TSDB_CHECK_CODE(code, lino, _exit); + do { + if (!TABLE_SAME_SCHEMA(pReader->bData.suid, pReader->bData.uid, pRowInfo->suid, pRowInfo->uid)) break; - code = tsdbSnapNextRow(pReader); - TSDB_CHECK_CODE(code, lino, _exit); + if (pReader->bData.uid && pReader->bData.uid != pRowInfo->uid) { + code = tRealloc((uint8_t**)&pReader->bData.aUid, sizeof(int64_t) * (pReader->bData.nRow + 1)); + TSDB_CHECK_CODE(code, lino, _exit); - pRowInfo = tsdbSnapGetRow(pReader); - if (pRowInfo == NULL) { - tsdbDataFReaderClose(&pReader->pDataFReader); - break; + for (int32_t iRow = 0; iRow < pReader->bData.nRow; ++iRow) { + pReader->bData.aUid[iRow] = pReader->bData.uid; + } + pReader->bData.uid = 0; } - if (pBlockData->nRow >= 4096) break; - } + code = tBlockDataAppendRow(&pReader->bData, &pRowInfo->row, NULL, pRowInfo->uid); + TSDB_CHECK_CODE(code, lino, _exit); - code = tsdbSnapCmprData(pReader, ppData); - TSDB_CHECK_CODE(code, lino, _exit); + code = tsdbSnapReadNextRow(pReader, &pRowInfo); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pReader->bData.nRow >= 4096) break; + } while (pRowInfo); + + ASSERT(pReader->bData.nRow > 0); break; } + if (pReader->bData.nRow > 0) { + code = tsdbSnapCmprData(pReader, ppData); + TSDB_CHECK_CODE(code, lino, _exit); + } + _exit: if (code) { - tsdbError("vgId:%d, %s failed since %s, path:%s", TD_VID(pTsdb->pVnode), __func__, tstrerror(code), pTsdb->path); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); } return code; } -static int32_t tsdbSnapReadTombstoneData(STsdbSnapReader* pReader, uint8_t** ppData) { +static int32_t tsdbSnapCmprTombData(STsdbSnapReader* pReader, uint8_t** ppData) { int32_t code = 0; int32_t lino = 0; - STsdb* pTsdb = pReader->pTsdb; - SDelFile* pDelFile = pReader->fs.pDelFile; - - if (pReader->pDelFReader == NULL) { - if (pDelFile == NULL) { - goto _exit; - } - - // open - code = tsdbDelFReaderOpen(&pReader->pDelFReader, pDelFile, pTsdb); - TSDB_CHECK_CODE(code, lino, _exit); - - // read index - code = tsdbReadDelIdx(pReader->pDelFReader, pReader->aDelIdx); - TSDB_CHECK_CODE(code, lino, _exit); - - pReader->iDelIdx = 0; + int64_t size = sizeof(TABLEID); + for (int32_t iDelData = 0; iDelData < taosArrayGetSize(pReader->aDelData); ++iDelData) { + size += tPutDelData(NULL, taosArrayGet(pReader->aDelData, iDelData)); } - while (true) { - if (pReader->iDelIdx >= taosArrayGetSize(pReader->aDelIdx)) { - tsdbDelFReaderClose(&pReader->pDelFReader); - break; - } - - SDelIdx* pDelIdx = (SDelIdx*)taosArrayGet(pReader->aDelIdx, pReader->iDelIdx); - - pReader->iDelIdx++; - - code = tsdbReadDelData(pReader->pDelFReader, pDelIdx, pReader->aDelData); + uint8_t* pData = (uint8_t*)taosMemoryMalloc(sizeof(SSnapDataHdr) + size); + if (pData == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; TSDB_CHECK_CODE(code, lino, _exit); + } - int32_t size = 0; - for (int32_t iDelData = 0; iDelData < taosArrayGetSize(pReader->aDelData); iDelData++) { - SDelData* pDelData = (SDelData*)taosArrayGet(pReader->aDelData, iDelData); + SSnapDataHdr* pHdr = (SSnapDataHdr*)pData; + pHdr->type = SNAP_DATA_DEL; + pHdr->size = size; - if (pDelData->version >= pReader->sver && pDelData->version <= pReader->ever) { - size += tPutDelData(NULL, pDelData); - } - } - if (size == 0) continue; + TABLEID* pId = (TABLEID*)(pData + sizeof(SSnapDataHdr)); + *pId = pReader->tbid; - // org data - size = sizeof(TABLEID) + size; - *ppData = taosMemoryMalloc(sizeof(SSnapDataHdr) + size); - if (*ppData == NULL) { - code = TSDB_CODE_OUT_OF_MEMORY; - TSDB_CHECK_CODE(code, lino, _exit); - } - - SSnapDataHdr* pHdr = (SSnapDataHdr*)(*ppData); - pHdr->type = SNAP_DATA_DEL; - pHdr->size = size; - - TABLEID* pId = (TABLEID*)(&pHdr[1]); - pId->suid = pDelIdx->suid; - pId->uid = pDelIdx->uid; - int32_t n = sizeof(SSnapDataHdr) + sizeof(TABLEID); - for (int32_t iDelData = 0; iDelData < taosArrayGetSize(pReader->aDelData); iDelData++) { - SDelData* pDelData = (SDelData*)taosArrayGet(pReader->aDelData, iDelData); - - if (pDelData->version < pReader->sver) continue; - if (pDelData->version > pReader->ever) continue; - - n += tPutDelData((*ppData) + n, pDelData); - } - - tsdbInfo("vgId:%d, vnode snapshot tsdb read del data for %s, suid:%" PRId64 " uid:%" PRId64 " size:%d", - TD_VID(pTsdb->pVnode), pTsdb->path, pDelIdx->suid, pDelIdx->uid, size); - - break; + size = sizeof(SSnapDataHdr) + sizeof(TABLEID); + for (int32_t iDelData = 0; iDelData < taosArrayGetSize(pReader->aDelData); ++iDelData) { + size += tPutDelData(pData + size, taosArrayGet(pReader->aDelData, iDelData)); } _exit: if (code) { - tsdbError("vgId:%d, %s failed since %s, path:%s", TD_VID(pTsdb->pVnode), __func__, tstrerror(code), pTsdb->path); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); + if (pData) { + taosMemoryFree(pData); + pData = NULL; + } + } + *ppData = pData; + return code; +} + +static int32_t tsdbSnapReadGetTombData(STsdbSnapReader* pReader, void* pDelInfo) { + int32_t code = 0; + int32_t lino = 0; + + ASSERT(0); + // TODO +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapReadNextTombData(STsdbSnapReader* pReader, void* pDelInfo) { + int32_t code = 0; + int32_t lino = 0; + // TODO +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); + } + return code; +} + +static int32_t tsdbSnapReadTombData(STsdbSnapReader* pReader, uint8_t** ppData) { + int32_t code = 0; + int32_t lino = 0; + + STsdb* pTsdb = pReader->pTsdb; + + if (pReader->pDelFReader == NULL) { + if (pReader->fs.pDelFile == NULL) goto _exit; + + // open + code = tsdbDelFReaderOpen(&pReader->pDelFReader, pReader->fs.pDelFile, pTsdb); + TSDB_CHECK_CODE(code, lino, _exit); + + code = tsdbOpenTombFileDataIter(pReader->pDelFReader, &pReader->pTIter); + TSDB_CHECK_CODE(code, lino, _exit); + } + + struct { + int64_t suid; + int64_t uid; + SDelData dData; + }* pDelInfo; + code = tsdbSnapReadGetTombData(pReader, &pDelInfo); + TSDB_CHECK_CODE(code, lino, _exit); + + if (pDelInfo == NULL) goto _exit; + + pReader->tbid = *(TABLEID*)pDelInfo; + + if (pReader->aDelData) { + taosArrayClear(pReader->aDelData); + } else if ((pReader->aDelData = taosArrayInit(16, sizeof(SDelData))) == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + while (pDelInfo && pDelInfo->suid == pReader->tbid.suid && pDelInfo->uid == pReader->tbid.uid) { + if (taosArrayPush(pReader->aDelData, &pDelInfo->dData) < 0) { + code = TSDB_CODE_OUT_OF_MEMORY; + TSDB_CHECK_CODE(code, lino, _exit); + } + + code = tsdbSnapReadNextTombData(pReader, &pDelInfo); + TSDB_CHECK_CODE(code, lino, _exit); + } + + if (taosArrayGetSize(pReader->aDelData) > 0) { + code = tsdbSnapCmprTombData(pReader, ppData); + TSDB_CHECK_CODE(code, lino, _exit); + } + +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbDebug("vgId:%d %s done", TD_VID(pTsdb->pVnode), __func__); } return code; } @@ -887,7 +919,7 @@ int32_t tsdbSnapRead(STsdbSnapReader* pReader, uint8_t** ppData) { // read del file if (!pReader->delDone) { - code = tsdbSnapReadTombstoneData(pReader, ppData); + code = tsdbSnapReadTombData(pReader, ppData); TSDB_CHECK_CODE(code, lino, _exit); if (*ppData) { goto _exit; @@ -1218,7 +1250,7 @@ static int32_t tsdbSnapWriteFileDataStart(STsdbSnapWriter* pWriter, int32_t fid) TSDB_CHECK_CODE(code, lino, _exit); if (pWriter->pSIter) { - code = tsdbSttFileDataIterNext(pWriter->pSIter); + code = tsdbSttFileDataIterNext(pWriter->pSIter, NULL); TSDB_CHECK_CODE(code, lino, _exit); // add to tree @@ -1312,7 +1344,7 @@ static int32_t tsdbSnapWriteNextRow(STsdbSnapWriter* pWriter, SRowInfo** ppRowIn int32_t lino = 0; if (pWriter->pSIter) { - code = tsdbDataIterNext2(pWriter->pSIter); + code = tsdbDataIterNext2(pWriter->pSIter, NULL); TSDB_CHECK_CODE(code, lino, _exit); if (pWriter->pSIter->rowInfo.suid == 0 && pWriter->pSIter->rowInfo.uid == 0) { From 555ee2b556b582c5609b9f56db96437b240f66a1 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Sun, 29 Jan 2023 09:43:03 +0800 Subject: [PATCH 036/124] fix(query): add qsort for alpine. --- source/os/src/osMath.c | 18 ++++++++++++++---- source/util/src/talgo.c | 16 ---------------- 2 files changed, 14 insertions(+), 20 deletions(-) diff --git a/source/os/src/osMath.c b/source/os/src/osMath.c index dddadd5ff6..5ac0fce006 100644 --- a/source/os/src/osMath.c +++ b/source/os/src/osMath.c @@ -15,7 +15,6 @@ #define ALLOW_FORBID_FUNC #define _DEFAULT_SOURCE -#include "os.h" #include #ifdef WINDOWS @@ -32,7 +31,18 @@ void swapStr(char* j, char* J, int width) { } #endif -// todo refactor: 1) move away; 2) use merge sort instead; 3) qsort is not a stable sort actually. -void taosSort(void* arr, int64_t sz, int64_t width, __compar_fn_t compar) { - qsort(arr, sz, width, compar); +int32_t qsortHelper(const void* p1, const void* p2, const void* param) { + __compar_fn_t comparFn = param; + return comparFn(p1, p2); } + +// todo refactor: 1) move away; 2) use merge sort instead; 3) qsort is not a stable sort actually. +void taosSort(void* base, int64_t sz, int64_t width, __compar_fn_t compar) { +#ifdef _ALPINE + void* param = compar; + taosqsort(base, width, sz, param, qsortHelper); +#else + qsort(base, sz, width, compar); +#endif +} + diff --git a/source/util/src/talgo.c b/source/util/src/talgo.c index d9319485b7..a06aac6afe 100644 --- a/source/util/src/talgo.c +++ b/source/util/src/talgo.c @@ -41,12 +41,6 @@ static void median(void *src, int64_t size, int64_t s, int64_t e, const void *pa ASSERT(comparFn(elePtrAt(src, size, mid), elePtrAt(src, size, s), param) <= 0 && comparFn(elePtrAt(src, size, s), elePtrAt(src, size, e), param) <= 0); - -#ifdef _DEBUG_VIEW -// tTagsPrints(src[s], pOrderDesc->pColumnModel, &pOrderDesc->orderIdx); -// tTagsPrints(src[mid], pOrderDesc->pColumnModel, &pOrderDesc->orderIdx); -// tTagsPrints(src[e], pOrderDesc->pColumnModel, &pOrderDesc->orderIdx); -#endif } static void tInsertSort(void *src, int64_t size, int32_t s, int32_t e, const void *param, __ext_compar_fn_t comparFn, @@ -278,14 +272,4 @@ void taosheapsort(void *base, int32_t size, int32_t len, const void *parcompar, } taosMemoryFree(buf); - /* - char *buf = taosMemoryCalloc(1, size); - - for (i = len - 1; i > 0; i--) { - doswap(elePtrAt(base, size, 0), elePtrAt(base, size, i)); - taosheapadjust(base, size, 0, i - 1, parcompar, compar, parswap, swap, maxroot); - } - - taosMemoryFreeClear(buf); - */ } From e0b0ce23951d1aadf94ba409424ecb3aaece4bd3 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sun, 29 Jan 2023 10:21:39 +0800 Subject: [PATCH 037/124] refact more code --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 107 +++++++++++++++++---- 1 file changed, 86 insertions(+), 21 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index b14fbdf189..befaf4a3e0 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -29,12 +29,19 @@ extern int32_t tsdbWriteSttBlock(SDataFWriter* pWriter, SBlockData* pBlockData, typedef struct STsdbDataIter2 STsdbDataIter2; typedef struct STsdbFilterInfo STsdbFilterInfo; +typedef struct { + int64_t suid; + int64_t uid; + SDelData delData; +} SDelInfo; + struct STsdbDataIter2 { STsdbDataIter2* next; SRBTreeNode rbtn; int32_t type; SRowInfo rowInfo; + SDelInfo delInfo; union { // TSDB_MEM_TABLE_DATA_ITER struct { @@ -391,6 +398,54 @@ _exit: return code; } +static int32_t tsdbTombFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo* pFilterInfo) { + int32_t code = 0; + int32_t lino = 0; + + for (;;) { + while (pIter->tIter.iDelData < taosArrayGetSize(pIter->tIter.aDelData)) { + SDelData* pDelData = taosArrayGet(pIter->tIter.aDelData, pIter->tIter.iDelData); + + if (pFilterInfo) { + if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) { + if (pFilterInfo->sver > pDelData->version || pFilterInfo->ever < pDelData->version) { + pIter->tIter.iDelData++; + continue; + } + } + } + + pIter->delInfo.delData = *pDelData; + pIter->tIter.iDelData++; + goto _exit; + } + + for (;;) { + if (pIter->tIter.iDelIdx < taosArrayGetSize(pIter->tIter.aDelIdx)) { + SDelIdx* pDelIdx = taosArrayGet(pIter->tIter.aDelIdx, pIter->tIter.iDelIdx); + + code = tsdbReadDelData(pIter->tIter.pReader, pDelIdx, pIter->tIter.aDelData); + TSDB_CHECK_CODE(code, lino, _exit); + + pIter->delInfo.suid = pDelIdx->suid; + pIter->delInfo.uid = pDelIdx->uid; + pIter->tIter.iDelData = 0; + pIter->tIter.iDelIdx++; + break; + } else { + pIter->delInfo = (SDelInfo){0}; + goto _exit; + } + } + } + +_exit: + if (code) { + tsdbError("%s failed at line %d since %s", __func__, lino, tstrerror(code)); + } + return code; +} + static int32_t tsdbDataIterNext2(STsdbDataIter2* pIter, STsdbFilterInfo* pFilterInfo) { int32_t code = 0; @@ -401,6 +456,8 @@ static int32_t tsdbDataIterNext2(STsdbDataIter2* pIter, STsdbFilterInfo* pFilter return tsdbDataFileDataIterNext(pIter, pFilterInfo); } else if (pIter->type == TSDB_STT_FILE_DATA_ITER) { return tsdbSttFileDataIterNext(pIter, pFilterInfo); + } else if (pIter->type == TSDB_TOMB_FILE_DATA_ITER) { + return tsdbTombFileDataIterNext(pIter, pFilterInfo); } else { ASSERT(0); return code; @@ -725,23 +782,27 @@ _exit: return code; } -static int32_t tsdbSnapReadGetTombData(STsdbSnapReader* pReader, void* pDelInfo) { - int32_t code = 0; - int32_t lino = 0; - - ASSERT(0); - // TODO -_exit: - if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); +static void tsdbSnapReadGetTombData(STsdbSnapReader* pReader, SDelInfo** ppDelInfo) { + if (pReader->pTIter == NULL || (pReader->pTIter->delInfo.suid == 0 && pReader->pTIter->delInfo.uid == 0)) { + *ppDelInfo = NULL; + } else { + *ppDelInfo = &pReader->pTIter->delInfo; } - return code; } -static int32_t tsdbSnapReadNextTombData(STsdbSnapReader* pReader, void* pDelInfo) { +static int32_t tsdbSnapReadNextTombData(STsdbSnapReader* pReader, SDelInfo** ppDelInfo) { int32_t code = 0; int32_t lino = 0; - // TODO + + code = tsdbDataIterNext2( + pReader->pTIter, + &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, .sver = pReader->sver, .ever = pReader->ever}); + TSDB_CHECK_CODE(code, lino, _exit); + + if (ppDelInfo) { + tsdbSnapReadGetTombData(pReader, ppDelInfo); + } + _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); @@ -755,6 +816,7 @@ static int32_t tsdbSnapReadTombData(STsdbSnapReader* pReader, uint8_t** ppData) STsdb* pTsdb = pReader->pTsdb; + // open tombstone data iter if need if (pReader->pDelFReader == NULL) { if (pReader->fs.pDelFile == NULL) goto _exit; @@ -764,15 +826,16 @@ static int32_t tsdbSnapReadTombData(STsdbSnapReader* pReader, uint8_t** ppData) code = tsdbOpenTombFileDataIter(pReader->pDelFReader, &pReader->pTIter); TSDB_CHECK_CODE(code, lino, _exit); + + if (pReader->pTIter) { + code = tsdbSnapReadNextTombData(pReader, NULL); + TSDB_CHECK_CODE(code, lino, _exit); + } } - struct { - int64_t suid; - int64_t uid; - SDelData dData; - }* pDelInfo; - code = tsdbSnapReadGetTombData(pReader, &pDelInfo); - TSDB_CHECK_CODE(code, lino, _exit); + // loop to get tombstone data + SDelInfo* pDelInfo; + tsdbSnapReadGetTombData(pReader, &pDelInfo); if (pDelInfo == NULL) goto _exit; @@ -786,7 +849,7 @@ static int32_t tsdbSnapReadTombData(STsdbSnapReader* pReader, uint8_t** ppData) } while (pDelInfo && pDelInfo->suid == pReader->tbid.suid && pDelInfo->uid == pReader->tbid.uid) { - if (taosArrayPush(pReader->aDelData, &pDelInfo->dData) < 0) { + if (taosArrayPush(pReader->aDelData, &pDelInfo->delData) < 0) { code = TSDB_CODE_OUT_OF_MEMORY; TSDB_CHECK_CODE(code, lino, _exit); } @@ -795,6 +858,7 @@ static int32_t tsdbSnapReadTombData(STsdbSnapReader* pReader, uint8_t** ppData) TSDB_CHECK_CODE(code, lino, _exit); } + // encode tombstone data if (taosArrayGetSize(pReader->aDelData) > 0) { code = tsdbSnapCmprTombData(pReader, ppData); TSDB_CHECK_CODE(code, lino, _exit); @@ -870,9 +934,9 @@ int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader) { if (pReader->pDelFReader) { tsdbDelFReaderClose(&pReader->pDelFReader); } + taosArrayDestroy(pReader->aDelData); // timeseries - tBlockDataDestroy(&pReader->bData, 1); while (pReader->iterList) { STsdbDataIter2* pIter = pReader->iterList; pReader->iterList = pIter->next; @@ -881,6 +945,7 @@ int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader) { if (pReader->pDataFReader) { tsdbDataFReaderClose(&pReader->pDataFReader); } + tBlockDataDestroy(&pReader->bData, 1); // other tDestroyTSchema(pReader->skmTable.pTSchema); From ef8b68bc80ca0af3091f51ab4d883b8d1390d787 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Sun, 29 Jan 2023 13:31:43 +0800 Subject: [PATCH 038/124] fix(query): fix syntax error on windows. --- source/os/src/osMath.c | 1 + 1 file changed, 1 insertion(+) diff --git a/source/os/src/osMath.c b/source/os/src/osMath.c index 5ac0fce006..067b130b57 100644 --- a/source/os/src/osMath.c +++ b/source/os/src/osMath.c @@ -16,6 +16,7 @@ #define ALLOW_FORBID_FUNC #define _DEFAULT_SOURCE #include +#include "osDef.h" #ifdef WINDOWS void swapStr(char* j, char* J, int width) { From 995b75f77f9ab598cce093904e6c3a2b69d32a13 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Sun, 29 Jan 2023 13:44:30 +0800 Subject: [PATCH 039/124] fix(query): fix syntax error on windows. --- source/os/src/osMath.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/os/src/osMath.c b/source/os/src/osMath.c index 067b130b57..9b24cce8c3 100644 --- a/source/os/src/osMath.c +++ b/source/os/src/osMath.c @@ -32,7 +32,7 @@ void swapStr(char* j, char* J, int width) { } #endif -int32_t qsortHelper(const void* p1, const void* p2, const void* param) { +int qsortHelper(const void* p1, const void* p2, const void* param) { __compar_fn_t comparFn = param; return comparFn(p1, p2); } From 4892740cb10033a8862554780e340c710bcc6b62 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Sun, 29 Jan 2023 14:00:00 +0800 Subject: [PATCH 040/124] fix(query): fix error on windows. --- source/os/src/osMath.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/os/src/osMath.c b/source/os/src/osMath.c index 9b24cce8c3..41c8c9257a 100644 --- a/source/os/src/osMath.c +++ b/source/os/src/osMath.c @@ -15,8 +15,8 @@ #define ALLOW_FORBID_FUNC #define _DEFAULT_SOURCE +#include "os.h" #include -#include "osDef.h" #ifdef WINDOWS void swapStr(char* j, char* J, int width) { From f6335a795ba73832aa9b3224978fa29c60ffad2a Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sun, 29 Jan 2023 14:47:45 +0800 Subject: [PATCH 041/124] fix: assert coredump --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index befaf4a3e0..ded4a20860 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -678,7 +678,7 @@ static int32_t tsdbSnapReadTimeSeriesData(STsdbSnapReader* pReader, uint8_t** pp STsdb* pTsdb = pReader->pTsdb; - tBlockDataClear(&pReader->bData); + tBlockDataReset(&pReader->bData); for (;;) { // start a new file read if need From 27f3b0219628e771945f67d6333713cfa8df18ea Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Sun, 29 Jan 2023 15:09:13 +0800 Subject: [PATCH 042/124] more code fix --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 24 ++++++++++++---------- 1 file changed, 13 insertions(+), 11 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index ded4a20860..3ef0cdc56f 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -925,6 +925,7 @@ int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader) { int32_t lino = 0; STsdbSnapReader* pReader = *ppReader; + STsdb* pTsdb = pReader->pTsdb; // tombstone if (pReader->pTIter) { @@ -957,9 +958,9 @@ int32_t tsdbSnapReaderClose(STsdbSnapReader** ppReader) { _exit: if (code) { - tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pReader->pTsdb->pVnode), __func__, lino, tstrerror(code)); + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); } else { - tsdbDebug("vgId:%d %s done", TD_VID(pReader->pTsdb->pVnode), __func__); + tsdbDebug("vgId:%d %s done", TD_VID(pTsdb->pVnode), __func__); } *ppReader = NULL; return code; @@ -1900,7 +1901,9 @@ _exit: } int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { - int32_t code = 0; + int32_t code = 0; + int32_t lino = 0; + STsdbSnapWriter* pWriter = *ppWriter; STsdb* pTsdb = pWriter->pTsdb; @@ -1913,7 +1916,7 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { code = tsdbFSCommit(pWriter->pTsdb); if (code) { taosThreadRwlockUnlock(&pTsdb->rwLock); - goto _err; + TSDB_CHECK_CODE(code, lino, _exit); } // unlock @@ -1936,16 +1939,15 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { for (int32_t iBuf = 0; iBuf < sizeof(pWriter->aBuf) / sizeof(uint8_t*); iBuf++) { tFree(pWriter->aBuf[iBuf]); } - tsdbInfo("vgId:%d, %s done", TD_VID(pWriter->pTsdb->pVnode), __func__); taosMemoryFree(pWriter); *ppWriter = NULL; - return code; -_err: - tsdbError("vgId:%d, vnode snapshot tsdb writer close for %s failed since %s", TD_VID(pWriter->pTsdb->pVnode), - pWriter->pTsdb->path, tstrerror(code)); - taosMemoryFree(pWriter); - *ppWriter = NULL; +_exit: + if (code) { + tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pTsdb->pVnode), __func__, lino, tstrerror(code)); + } else { + tsdbInfo("vgId:%d %s done", TD_VID(pTsdb->pVnode), __func__); + } return code; } From 96f5249782bcbab49c7828bce822e08c3472de17 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Sun, 29 Jan 2023 15:54:25 +0800 Subject: [PATCH 043/124] enh: flush database in testcase alter_replica_13.sim to trigger snapshot repl --- tests/script/tsim/db/alter_replica_13.sim | 1 + 1 file changed, 1 insertion(+) diff --git a/tests/script/tsim/db/alter_replica_13.sim b/tests/script/tsim/db/alter_replica_13.sim index d75acb50ad..a9dc1741a1 100644 --- a/tests/script/tsim/db/alter_replica_13.sim +++ b/tests/script/tsim/db/alter_replica_13.sim @@ -79,6 +79,7 @@ sql insert into db.ctb6 values(now, 6, "6") sql insert into db.ctb7 values(now, 7, "7") sql insert into db.ctb8 values(now, 8, "8") sql insert into db.ctb9 values(now, 9, "9") +sql flush database db; print =============== step3: create dnodes sql create dnode $hostname port 7300 From 82baa875c623d8bc32fe3bec853310803ae29b52 Mon Sep 17 00:00:00 2001 From: chenhaoran Date: Sun, 29 Jan 2023 17:05:13 +0800 Subject: [PATCH 044/124] ci:add taosx compilation in ci --- tests/parallel_test/container_build.sh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tests/parallel_test/container_build.sh b/tests/parallel_test/container_build.sh index 221e549056..ff854449bb 100755 --- a/tests/parallel_test/container_build.sh +++ b/tests/parallel_test/container_build.sh @@ -55,7 +55,7 @@ fi date docker run \ -v $REP_MOUNT_PARAM \ - --rm --ulimit core=-1 taos_test:v1.0 sh -c "cd $REP_DIR;rm -rf debug;mkdir -p debug;cd debug;cmake .. -DBUILD_HTTP=false -DBUILD_TOOLS=true -DBUILD_TEST=true -DWEBSOCKET=true;make -j || exit 1" + --rm --ulimit core=-1 taos_test:v1.0 sh -c "cd $REP_DIR;rm -rf debug;mkdir -p debug;cd debug;cmake .. -DBUILD_HTTP=false -DBUILD_TOOLS=true -DBUILD_TEST=true -DWEBSOCKET=true -DBUILD_TAOSX=true;make -j || exit 1" if [[ -d ${WORKDIR}/debugNoSan ]] ;then echo "delete ${WORKDIR}/debugNoSan" @@ -70,7 +70,7 @@ mv ${REP_REAL_PATH}/debug ${WORKDIR}/debugNoSan date docker run \ -v $REP_MOUNT_PARAM \ - --rm --ulimit core=-1 taos_test:v1.0 sh -c "cd $REP_DIR;rm -rf debug;mkdir -p debug;cd debug;cmake .. -DBUILD_HTTP=false -DBUILD_TOOLS=true -DBUILD_TEST=true -DWEBSOCKET=true -DBUILD_SANITIZER=1 -DTOOLS_SANITIZE=true -DTOOLS_BUILD_TYPE=Debug;make -j || exit 1 " + --rm --ulimit core=-1 taos_test:v1.0 sh -c "cd $REP_DIR;rm -rf debug;mkdir -p debug;cd debug;cmake .. -DBUILD_HTTP=false -DBUILD_TOOLS=true -DBUILD_TEST=true -DWEBSOCKET=true -DBUILD_SANITIZER=1 -DTOOLS_SANITIZE=true -DTOOLS_BUILD_TYPE=Debug -DBUILD_TAOSX=true;make -j || exit 1 " mv ${REP_REAL_PATH}/debug ${WORKDIR}/debugSan From bd23a36f86b0392037bcbd4207ab8501e40a6d38 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Sun, 29 Jan 2023 18:03:09 +0800 Subject: [PATCH 045/124] fix: round lastApplyIndex up to beginIndex -1 in syncBeginSnapshot --- source/libs/sync/src/syncMain.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index b9ca6af912..ea22ac7bb5 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -306,7 +306,7 @@ int32_t syncBeginSnapshot(int64_t rid, int64_t lastApplyIndex) { if (pSyncNode->replicaNum > 1) { // multi replicas - lastApplyIndex -= SYNC_VNODE_LOG_RETENTION; + lastApplyIndex = TMAX(lastApplyIndex - SYNC_VNODE_LOG_RETENTION, beginIndex - 1); if (pSyncNode->state == TAOS_SYNC_STATE_LEADER) { pSyncNode->minMatchIndex = syncMinMatchIndex(pSyncNode); From 1741e98a4e8fd5353aa541ca9a06def8546f5349 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 30 Jan 2023 10:13:54 +0800 Subject: [PATCH 046/124] fix(query): reset the del file index when beginning last file check. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 44 ++++++++++++++++---------- 1 file changed, 28 insertions(+), 16 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 85282a2340..a77eb8298e 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -215,6 +215,7 @@ static int32_t doBuildDataBlock(STsdbReader* pReader); static TSDBKEY getCurrentKeyInBuf(STableBlockScanInfo* pScanInfo, STsdbReader* pReader); static bool hasDataInFileBlock(const SBlockData* pBlockData, const SFileBlockDumpInfo* pDumpInfo); static void initBlockDumpInfo(STsdbReader* pReader, SDataBlockIter* pBlockIter); +static int32_t getInitialDelIndex(const SArray* pDelSkyline, int32_t order); static bool outOfTimeWindow(int64_t ts, STimeWindow* pWindow) { return (ts > pWindow->ekey) || (ts < pWindow->skey); } @@ -2526,6 +2527,14 @@ _end: void setComposedBlockFlag(STsdbReader* pReader, bool composed) { pReader->status.composedDataBlock = composed; } +int32_t getInitialDelIndex(const SArray* pDelSkyline, int32_t order) { + if (pDelSkyline == NULL) { + return 0; + } + + return ASCENDING_TRAVERSE(order) ? 0 : taosArrayGetSize(pDelSkyline) - 1; +} + int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, STsdbReader* pReader, STbData* pMemTbData, STbData* piMemTbData) { if (pBlockScanInfo->delSkyline != NULL) { @@ -2543,7 +2552,6 @@ int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, STsdbReader* if (pIdx != NULL) { code = tsdbReadDelData(pReader->pDelFReader, pIdx, pDelData); } - if (code != TSDB_CODE_SUCCESS) { goto _err; } @@ -2572,11 +2580,13 @@ int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, STsdbReader* } taosArrayDestroy(pDelData); - pBlockScanInfo->iter.index = - ASCENDING_TRAVERSE(pReader->order) ? 0 : taosArrayGetSize(pBlockScanInfo->delSkyline) - 1; - pBlockScanInfo->iiter.index = pBlockScanInfo->iter.index; - pBlockScanInfo->fileDelIndex = pBlockScanInfo->iter.index; - pBlockScanInfo->lastBlockDelIndex = pBlockScanInfo->iter.index; + int32_t index = getInitialDelIndex(pBlockScanInfo->delSkyline, pReader->order); + + pBlockScanInfo->iter.index = index; + pBlockScanInfo->iiter.index = index; + pBlockScanInfo->fileDelIndex = index; + pBlockScanInfo->lastBlockDelIndex = index; + return code; _err: @@ -2676,13 +2686,17 @@ static int32_t uidComparFunc(const void* p1, const void* p2) { } } -static void extractOrderedTableUidList(SUidOrderCheckInfo* pOrderCheckInfo, SReaderStatus* pStatus) { +static void extractOrderedTableUidList(SUidOrderCheckInfo* pOrderCheckInfo, SReaderStatus* pStatus, int32_t order) { int32_t index = 0; int32_t total = taosHashGetSize(pStatus->pTableMap); void* p = taosHashIterate(pStatus->pTableMap, NULL); while (p != NULL) { STableBlockScanInfo* pScanInfo = *(STableBlockScanInfo**)p; + + // reset the last del file index + pScanInfo->lastBlockDelIndex = getInitialDelIndex(pScanInfo->delSkyline, order); + pOrderCheckInfo->tableUidList[index++] = pScanInfo->uid; p = taosHashIterate(pStatus->pTableMap, p); } @@ -2690,7 +2704,9 @@ static void extractOrderedTableUidList(SUidOrderCheckInfo* pOrderCheckInfo, SRea taosSort(pOrderCheckInfo->tableUidList, total, sizeof(uint64_t), uidComparFunc); } -static int32_t initOrderCheckInfo(SUidOrderCheckInfo* pOrderCheckInfo, SReaderStatus* pStatus) { +static int32_t initOrderCheckInfo(SUidOrderCheckInfo* pOrderCheckInfo, STsdbReader* pReader) { + SReaderStatus* pStatus = &pReader->status; + int32_t total = taosHashGetSize(pStatus->pTableMap); if (total == 0) { return TSDB_CODE_SUCCESS; @@ -2703,7 +2719,7 @@ static int32_t initOrderCheckInfo(SUidOrderCheckInfo* pOrderCheckInfo, SReaderSt return TSDB_CODE_OUT_OF_MEMORY; } - extractOrderedTableUidList(pOrderCheckInfo, pStatus); + extractOrderedTableUidList(pOrderCheckInfo, pStatus, pReader->order); uint64_t uid = pOrderCheckInfo->tableUidList[0]; pStatus->pTableIter = taosHashGet(pStatus->pTableMap, &uid, sizeof(uid)); } else { @@ -2720,7 +2736,7 @@ static int32_t initOrderCheckInfo(SUidOrderCheckInfo* pOrderCheckInfo, SReaderSt } pOrderCheckInfo->tableUidList = p; - extractOrderedTableUidList(pOrderCheckInfo, pStatus); + extractOrderedTableUidList(pOrderCheckInfo, pStatus, pReader->order); uid = pOrderCheckInfo->tableUidList[0]; pStatus->pTableIter = taosHashGet(pStatus->pTableMap, &uid, sizeof(uid)); @@ -2740,11 +2756,7 @@ static bool moveToNextTable(SUidOrderCheckInfo* pOrderedCheckInfo, SReaderStatus uint64_t uid = pOrderedCheckInfo->tableUidList[pOrderedCheckInfo->currentIndex]; pStatus->pTableIter = taosHashGet(pStatus->pTableMap, &uid, sizeof(uid)); - if (pStatus->pTableIter == NULL) { - return false; - } - - return true; + return (pStatus->pTableIter != NULL); } static int32_t doLoadLastBlockSequentially(STsdbReader* pReader) { @@ -2752,7 +2764,7 @@ static int32_t doLoadLastBlockSequentially(STsdbReader* pReader) { SLastBlockReader* pLastBlockReader = pStatus->fileIter.pLastBlockReader; SUidOrderCheckInfo* pOrderedCheckInfo = &pStatus->uidCheckInfo; - int32_t code = initOrderCheckInfo(pOrderedCheckInfo, pStatus); + int32_t code = initOrderCheckInfo(pOrderedCheckInfo, pReader); if (code != TSDB_CODE_SUCCESS || (taosHashGetSize(pStatus->pTableMap) == 0)) { return code; } From 9502fb3633c8d79bb92c27339692391de09f2494 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Mon, 30 Jan 2023 11:40:41 +0800 Subject: [PATCH 047/124] fix memory leak --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 1 + 1 file changed, 1 insertion(+) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 3ef0cdc56f..1bfd630331 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1939,6 +1939,7 @@ int32_t tsdbSnapWriterClose(STsdbSnapWriter** ppWriter, int8_t rollback) { for (int32_t iBuf = 0; iBuf < sizeof(pWriter->aBuf) / sizeof(uint8_t*); iBuf++) { tFree(pWriter->aBuf[iBuf]); } + tsdbFSDestroy(&pWriter->fs); taosMemoryFree(pWriter); *ppWriter = NULL; From 1ce1f8143cd638ec9cab541aba160c2e2ddfaa73 Mon Sep 17 00:00:00 2001 From: 54liuyao <54liuyao@163.com> Date: Mon, 30 Jan 2023 10:45:12 +0800 Subject: [PATCH 048/124] fix:process data with incorrect timestamp --- source/libs/executor/src/timewindowoperator.c | 14 ++- tests/script/tsim/stream/basic1.sim | 53 +++++++++++ tests/script/tsim/stream/triggerInterval0.sim | 88 +++++++++++++++---- 3 files changed, 135 insertions(+), 20 deletions(-) diff --git a/source/libs/executor/src/timewindowoperator.c b/source/libs/executor/src/timewindowoperator.c index d78e9c4edf..d320ef6e9e 100644 --- a/source/libs/executor/src/timewindowoperator.c +++ b/source/libs/executor/src/timewindowoperator.c @@ -2477,7 +2477,19 @@ static void doStreamIntervalAggImpl(SOperatorInfo* pOperatorInfo, SSDataBlock* p pInfo->delKey = key; } int32_t prevEndPos = (forwardRows - 1) * step + startPos; - ASSERT(pSDataBlock->info.window.skey > 0 && pSDataBlock->info.window.ekey > 0); + if (pSDataBlock->info.window.skey <= 0 || pSDataBlock->info.window.ekey <= 0) { + qError("table uid %" PRIu64 " data block timestamp range may not be calculated! minKey %" PRId64 + ",maxKey %" PRId64, + pSDataBlock->info.id.uid, pSDataBlock->info.window.skey, pSDataBlock->info.window.ekey); + blockDataUpdateTsWindow(pSDataBlock, 0); + + // timestamp of the data is incorrect + if (pSDataBlock->info.window.skey <= 0 || pSDataBlock->info.window.ekey <= 0) { + qError("table uid %" PRIu64 " data block timestamp is out of range! minKey %" PRId64 ",maxKey %" PRId64, + pSDataBlock->info.id.uid, pSDataBlock->info.window.skey, pSDataBlock->info.window.ekey); + } + } + if (IS_FINAL_OP(pInfo)) { startPos = getNextQualifiedFinalWindow(&pInfo->interval, &nextWin, &pSDataBlock->info, tsCols, prevEndPos); } else { diff --git a/tests/script/tsim/stream/basic1.sim b/tests/script/tsim/stream/basic1.sim index 7bf10df637..c61c7667f8 100644 --- a/tests/script/tsim/stream/basic1.sim +++ b/tests/script/tsim/stream/basic1.sim @@ -834,4 +834,57 @@ endi print ====== test _wstart end +print insert into ts1 values(-1648791211000,1,2,3) + +sql create database test7 vgroups 1; +sql use test7; +sql create stable st(ts timestamp, a int, b int , c int) tags(ta int,tb int,tc int); +sql create table ts1 using st tags(1,1,1); +sql create stream streams7 trigger at_once into streamt7 as select _wstart, count(*) from ts1 interval(10s) ; + +sql insert into ts1 values(1648791211000,1,2,3); +sql_error insert into ts1 values(-1648791211000,1,2,3); + +loop18: + +sleep 200 +sql select * from streamt7; + +$loop_count = $loop_count + 1 +if $loop_count == 10 then + return -1 +endi + +if $rows != 1 then + print =====rows=$rows + goto loop18 +endi + +if $data01 != 1 then + print =====data01=$data01 + goto loop18 +endi + +sql_error insert into ts1 values(-1648791211001,1,2,3) (1648791211001,1,2,3); + +sql select _wstart, count(*) from ts1 interval(10s) ; + +print $data00 $data01 +print $data10 $data11 + +loop19: + +sleep 200 +sql select * from streamt7; + +$loop_count = $loop_count + 1 +if $loop_count == 10 then + return -1 +endi + +if $rows != 1 then + print =====rows=$rows + goto loop19 +endi + system sh/exec.sh -n dnode1 -s stop -x SIGINT diff --git a/tests/script/tsim/stream/triggerInterval0.sim b/tests/script/tsim/stream/triggerInterval0.sim index 7353f026bb..b522dcf035 100644 --- a/tests/script/tsim/stream/triggerInterval0.sim +++ b/tests/script/tsim/stream/triggerInterval0.sim @@ -29,69 +29,119 @@ sql insert into t1 values(1648791223001,2,2,3,1.1); sql insert into t1 values(1648791223002,2,2,3,1.1); sql insert into t1 values(1648791223003,2,2,3,1.1); sql insert into t1 values(1648791223001,2,2,3,1.1); + +print step 0 + +$loop_count = 0 + +loop0: sleep 300 + +$loop_count = $loop_count + 1 +if $loop_count == 10 then + return -1 +endi + sql select * from streamt; + if $rows != 1 then print ======$rows - return -1 + goto loop0 endi if $data01 != 1 then print ======$data01 - return -1 + goto loop0 endi sql insert into t1 values(1648791233001,2,2,3,1.1); + +print step 1 + +$loop_count = 0 + +loop1: sleep 300 + +$loop_count = $loop_count + 1 +if $loop_count == 10 then + return -1 +endi + sql select * from streamt; if $rows != 2 then print ======$rows - return -1 + goto loop1 endi if $data01 != 1 then print ======$data01 - return -1 + goto loop1 endi if $data11 != 3 then print ======$data11 - return -1 + goto loop1 endi sql insert into t1 values(1648791223004,2,2,3,1.1); sql insert into t1 values(1648791223004,2,2,3,1.1); sql insert into t1 values(1648791223005,2,2,3,1.1); + +print step 2 + +$loop_count = 0 + +loop2: sleep 300 + +$loop_count = $loop_count + 1 +if $loop_count == 10 then + return -1 +endi + sql select * from streamt; if $rows != 2 then print ======$rows - return -1 + goto loop2 endi + if $data01 != 1 then print ======$data01 - return -1 + goto loop2 endi if $data11 != 5 then print ======$data11 - return -1 + goto loop2 endi sql insert into t1 values(1648791233002,3,2,3,2.1); sql insert into t1 values(1648791213002,4,2,3,3.1) sql insert into t1 values(1648791213002,4,2,3,4.1); + +print step 3 + +$loop_count = 0 + +loop3: sleep 300 -sql select * from streamt; -if $rows != 2 then - print ======$rows - return -1 -endi -if $data01 != 2 then - print ======$data01 - return -1 -endi -if $data11 != 5 then - print ======$data11 + +$loop_count = $loop_count + 1 +if $loop_count == 10 then return -1 endi +sql select * from streamt; +if $rows != 2 then + print ======$rows + goto loop3 +endi +if $data01 != 2 then + print ======$data01 + goto loop3 +endi +if $data11 != 5 then + print ======$data11 + goto loop3 +endi + system sh/exec.sh -n dnode1 -s stop -x SIGINT \ No newline at end of file From ac9cdf6c58eaf5f4e4303a52fe855485a7f321eb Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Tue, 31 Jan 2023 00:46:58 +0800 Subject: [PATCH 049/124] fix: install script refine sentence for main (#19703) * fix: packaging/tools/install.sh for main * fix: refine sentence * fix: dpkg grep ii to detech deb package --- packaging/tools/install.sh | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/packaging/tools/install.sh b/packaging/tools/install.sh index 4be179b04d..dfdbaa6fdd 100755 --- a/packaging/tools/install.sh +++ b/packaging/tools/install.sh @@ -210,8 +210,8 @@ function install_bin() { [ -x ${install_main_dir}/bin/${serverName} ] && ${csudo}ln -s ${install_main_dir}/bin/${serverName} ${bin_link_dir}/${serverName} || : [ -x ${install_main_dir}/bin/${udfdName} ] && ${csudo}ln -s ${install_main_dir}/bin/${udfdName} ${bin_link_dir}/${udfdName} || : [ -x ${install_main_dir}/bin/${adapterName} ] && ${csudo}ln -s ${install_main_dir}/bin/${adapterName} ${bin_link_dir}/${adapterName} || : - [ -x ${install_main_dir}/bin/${benchmarkName} ] && ${csudo}ln -s ${install_main_dir}/bin/${benchmarkName} ${bin_link_dir}/${demoName} || : - [ -x ${install_main_dir}/bin/${benchmarkName} ] && ${csudo}ln -s ${install_main_dir}/bin/${benchmarkName} ${bin_link_dir}/${benchmarkName} || : + [ -x ${install_main_dir}/bin/${benchmarkName} ] && ${csudo}ln -sf ${install_main_dir}/bin/${benchmarkName} ${bin_link_dir}/${demoName} || : + [ -x ${install_main_dir}/bin/${benchmarkName} ] && ${csudo}ln -sf ${install_main_dir}/bin/${benchmarkName} ${bin_link_dir}/${benchmarkName} || : [ -x ${install_main_dir}/bin/${dumpName} ] && ${csudo}ln -s ${install_main_dir}/bin/${dumpName} ${bin_link_dir}/${dumpName} || : [ -x ${install_main_dir}/bin/${xname} ] && ${csudo}ln -s ${install_main_dir}/bin/${xname} ${bin_link_dir}/${xname} || : [ -x ${install_main_dir}/bin/TDinsight.sh ] && ${csudo}ln -s ${install_main_dir}/bin/TDinsight.sh ${bin_link_dir}/TDinsight.sh || : @@ -746,7 +746,7 @@ function is_version_compatible() { deb_erase() { confirm="" while [ "" == "${confirm}" ]; do - echo -e -n "${RED}Exist tdengine deb detected, do you want to remove it? [yes|no] ${NC}:" + echo -e -n "${RED}Existing TDengine deb is detected, do you want to remove it? [yes|no] ${NC}:" read confirm if [ "yes" == "$confirm" ]; then ${csudo}dpkg --remove tdengine ||: @@ -760,7 +760,7 @@ deb_erase() { rpm_erase() { confirm="" while [ "" == "${confirm}" ]; do - echo -e -n "${RED}Exist tdengine rpm detected, do you want to remove it? [yes|no] ${NC}:" + echo -e -n "${RED}Existing TDengine rpm is detected, do you want to remove it? [yes|no] ${NC}:" read confirm if [ "yes" == "$confirm" ]; then ${csudo}rpm -e tdengine ||: @@ -787,7 +787,7 @@ function updateProduct() { if echo $osinfo | grep -qwi "centos"; then rpm -q tdengine 2>&1 > /dev/null && rpm_erase tdengine ||: elif echo $osinfo | grep -qwi "ubuntu"; then - dpkg -l tdengine 2>&1 > /dev/null && deb_erase tdengine ||: + dpkg -l tdengine 2>&1 | grep ii > /dev/null && deb_erase tdengine ||: fi tar -zxf ${tarName} From 889019b253c7236e9d922689330e3ff04761e36f Mon Sep 17 00:00:00 2001 From: sunpeng Date: Tue, 31 Jan 2023 09:37:50 +0800 Subject: [PATCH 050/124] build: add python demo to ci (#19699) * build: add python demo to ci * build: fix python demo --- docs/examples/go/go.mod | 6 - docs/examples/python/conn_native_pandas.py | 7 +- docs/examples/python/conn_rest_pandas.py | 6 +- docs/examples/python/connect_rest_examples.py | 23 ++-- .../connection_usage_native_reference.py | 8 +- docs/examples/python/fast_write_example.py | 119 ++++++++++++------ docs/examples/python/kafka_example.py | 80 +++++++++--- docs/examples/python/mockdatasource.py | 19 ++- docs/examples/python/sql_writer.py | 24 +++- docs/examples/python/tmq_example.py | 95 +++++++------- tests/docs-examples-test/python.sh | 43 ++++++- 11 files changed, 298 insertions(+), 132 deletions(-) delete mode 100644 docs/examples/go/go.mod diff --git a/docs/examples/go/go.mod b/docs/examples/go/go.mod deleted file mode 100644 index 2bc1a74cb6..0000000000 --- a/docs/examples/go/go.mod +++ /dev/null @@ -1,6 +0,0 @@ -module goexample - -go 1.17 - -require github.com/taosdata/driver-go/v3 3.0 - diff --git a/docs/examples/python/conn_native_pandas.py b/docs/examples/python/conn_native_pandas.py index 56942ef570..f3bab15efb 100644 --- a/docs/examples/python/conn_native_pandas.py +++ b/docs/examples/python/conn_native_pandas.py @@ -1,8 +1,11 @@ import pandas -from sqlalchemy import create_engine +from sqlalchemy import create_engine, text engine = create_engine("taos://root:taosdata@localhost:6030/power") -df = pandas.read_sql("SELECT * FROM meters", engine) +conn = engine.connect() +df = pandas.read_sql(text("SELECT * FROM power.meters"), conn) +conn.close() + # print index print(df.index) diff --git a/docs/examples/python/conn_rest_pandas.py b/docs/examples/python/conn_rest_pandas.py index 0164080cd5..1b207d6ff1 100644 --- a/docs/examples/python/conn_rest_pandas.py +++ b/docs/examples/python/conn_rest_pandas.py @@ -1,8 +1,10 @@ import pandas -from sqlalchemy import create_engine +from sqlalchemy import create_engine, text engine = create_engine("taosrest://root:taosdata@localhost:6041") -df: pandas.DataFrame = pandas.read_sql("SELECT * FROM power.meters", engine) +conn = engine.connect() +df: pandas.DataFrame = pandas.read_sql(text("SELECT * FROM power.meters"), conn) +conn.close() # print index print(df.index) diff --git a/docs/examples/python/connect_rest_examples.py b/docs/examples/python/connect_rest_examples.py index 900ec1022e..0f8625ae53 100644 --- a/docs/examples/python/connect_rest_examples.py +++ b/docs/examples/python/connect_rest_examples.py @@ -1,24 +1,25 @@ # ANCHOR: connect from taosrest import connect, TaosRestConnection, TaosRestCursor -conn: TaosRestConnection = connect(url="http://localhost:6041", - user="root", - password="taosdata", - timeout=30) +conn = connect(url="http://localhost:6041", + user="root", + password="taosdata", + timeout=30) # ANCHOR_END: connect # ANCHOR: basic # create STable -cursor: TaosRestCursor = conn.cursor() +cursor = conn.cursor() cursor.execute("DROP DATABASE IF EXISTS power") cursor.execute("CREATE DATABASE power") -cursor.execute("CREATE STABLE power.meters (ts TIMESTAMP, current FLOAT, voltage INT, phase FLOAT) TAGS (location BINARY(64), groupId INT)") +cursor.execute( + "CREATE STABLE power.meters (ts TIMESTAMP, current FLOAT, voltage INT, phase FLOAT) TAGS (location BINARY(64), groupId INT)") # insert data -cursor.execute("""INSERT INTO power.d1001 USING power.meters TAGS(California.SanFrancisco, 2) VALUES ('2018-10-03 14:38:05.000', 10.30000, 219, 0.31000) ('2018-10-03 14:38:15.000', 12.60000, 218, 0.33000) ('2018-10-03 14:38:16.800', 12.30000, 221, 0.31000) - power.d1002 USING power.meters TAGS(California.SanFrancisco, 3) VALUES ('2018-10-03 14:38:16.650', 10.30000, 218, 0.25000) - power.d1003 USING power.meters TAGS(California.LosAngeles, 2) VALUES ('2018-10-03 14:38:05.500', 11.80000, 221, 0.28000) ('2018-10-03 14:38:16.600', 13.40000, 223, 0.29000) - power.d1004 USING power.meters TAGS(California.LosAngeles, 3) VALUES ('2018-10-03 14:38:05.000', 10.80000, 223, 0.29000) ('2018-10-03 14:38:06.500', 11.50000, 221, 0.35000)""") +cursor.execute("""INSERT INTO power.d1001 USING power.meters TAGS('California.SanFrancisco', 2) VALUES ('2018-10-03 14:38:05.000', 10.30000, 219, 0.31000) ('2018-10-03 14:38:15.000', 12.60000, 218, 0.33000) ('2018-10-03 14:38:16.800', 12.30000, 221, 0.31000) + power.d1002 USING power.meters TAGS('California.SanFrancisco', 3) VALUES ('2018-10-03 14:38:16.650', 10.30000, 218, 0.25000) + power.d1003 USING power.meters TAGS('California.LosAngeles', 2) VALUES ('2018-10-03 14:38:05.500', 11.80000, 221, 0.28000) ('2018-10-03 14:38:16.600', 13.40000, 223, 0.29000) + power.d1004 USING power.meters TAGS('California.LosAngeles', 3) VALUES ('2018-10-03 14:38:05.000', 10.80000, 223, 0.29000) ('2018-10-03 14:38:06.500', 11.50000, 221, 0.35000)""") print("inserted row count:", cursor.rowcount) # query data @@ -28,7 +29,7 @@ print("queried row count:", cursor.rowcount) # get column names from cursor column_names = [meta[0] for meta in cursor.description] # get rows -data: list[tuple] = cursor.fetchall() +data = cursor.fetchall() print(column_names) for row in data: print(row) diff --git a/docs/examples/python/connection_usage_native_reference.py b/docs/examples/python/connection_usage_native_reference.py index 4803511e42..0a23c5f95b 100644 --- a/docs/examples/python/connection_usage_native_reference.py +++ b/docs/examples/python/connection_usage_native_reference.py @@ -8,7 +8,7 @@ conn.execute("CREATE DATABASE test") # change database. same as execute "USE db" conn.select_db("test") conn.execute("CREATE STABLE weather(ts TIMESTAMP, temperature FLOAT) TAGS (location INT)") -affected_row: int = conn.execute("INSERT INTO t1 USING weather TAGS(1) VALUES (now, 23.5) (now+1m, 23.5) (now+2m 24.4)") +affected_row = conn.execute("INSERT INTO t1 USING weather TAGS(1) VALUES (now, 23.5) (now+1m, 23.5) (now+2m, 24.4)") print("affected_row", affected_row) # output: # affected_row 3 @@ -16,10 +16,10 @@ print("affected_row", affected_row) # ANCHOR: query # Execute a sql and get its result set. It's useful for SELECT statement -result: taos.TaosResult = conn.query("SELECT * from weather") +result = conn.query("SELECT * from weather") # Get fields from result -fields: taos.field.TaosFields = result.fields +fields = result.fields for field in fields: print(field) # {name: ts, type: 9, bytes: 8} @@ -42,4 +42,4 @@ print(data) # ANCHOR_END: query -conn.close() +conn.close() \ No newline at end of file diff --git a/docs/examples/python/fast_write_example.py b/docs/examples/python/fast_write_example.py index c9d606388f..626e3310b1 100644 --- a/docs/examples/python/fast_write_example.py +++ b/docs/examples/python/fast_write_example.py @@ -1,15 +1,14 @@ # install dependencies: # recommend python >= 3.8 -# pip3 install faster-fifo # import logging import math +import multiprocessing import sys import time import os -from multiprocessing import Process -from faster_fifo import Queue +from multiprocessing import Process, Queue from mockdatasource import MockDataSource from queue import Empty from typing import List @@ -22,8 +21,7 @@ TABLE_COUNT = 1000 QUEUE_SIZE = 1000000 MAX_BATCH_SIZE = 3000 -read_processes = [] -write_processes = [] +_DONE_MESSAGE = '__DONE__' def get_connection(): @@ -44,41 +42,64 @@ def get_connection(): # ANCHOR: read -def run_read_task(task_id: int, task_queues: List[Queue]): +def run_read_task(task_id: int, task_queues: List[Queue], infinity): table_count_per_task = TABLE_COUNT // READ_TASK_COUNT - data_source = MockDataSource(f"tb{task_id}", table_count_per_task) + data_source = MockDataSource(f"tb{task_id}", table_count_per_task, infinity) try: for batch in data_source: + if isinstance(batch, tuple): + batch = [batch] for table_id, rows in batch: # hash data to different queue i = table_id % len(task_queues) # block putting forever when the queue is full - task_queues[i].put_many(rows, block=True, timeout=-1) + for row in rows: + task_queues[i].put(row) + if not infinity: + for queue in task_queues: + queue.put(_DONE_MESSAGE) except KeyboardInterrupt: pass + finally: + logging.info('read task over') # ANCHOR_END: read + # ANCHOR: write -def run_write_task(task_id: int, queue: Queue): +def run_write_task(task_id: int, queue: Queue, done_queue: Queue): from sql_writer import SQLWriter log = logging.getLogger(f"WriteTask-{task_id}") writer = SQLWriter(get_connection) lines = None try: while True: - try: - # get as many as possible - lines = queue.get_many(block=False, max_messages_to_get=MAX_BATCH_SIZE) + over = False + lines = [] + for _ in range(MAX_BATCH_SIZE): + try: + line = queue.get_nowait() + if line == _DONE_MESSAGE: + over = True + break + if line: + lines.append(line) + except Empty: + time.sleep(0.1) + if len(lines) > 0: writer.process_lines(lines) - except Empty: - time.sleep(0.01) + if over: + done_queue.put(_DONE_MESSAGE) + break except KeyboardInterrupt: pass except BaseException as e: log.debug(f"lines={lines}") raise e + finally: + writer.close() + log.debug('write task over') # ANCHOR_END: write @@ -103,47 +124,64 @@ def set_global_config(): # ANCHOR: monitor -def run_monitor_process(): +def run_monitor_process(done_queue: Queue): log = logging.getLogger("DataBaseMonitor") - conn = get_connection() - conn.execute("DROP DATABASE IF EXISTS test") - conn.execute("CREATE DATABASE test") - conn.execute("CREATE STABLE test.meters (ts TIMESTAMP, current FLOAT, voltage INT, phase FLOAT) " - "TAGS (location BINARY(64), groupId INT)") + conn = None + try: + conn = get_connection() - def get_count(): - res = conn.query("SELECT count(*) FROM test.meters") - rows = res.fetch_all() - return rows[0][0] if rows else 0 + def get_count(): + res = conn.query("SELECT count(*) FROM test.meters") + rows = res.fetch_all() + return rows[0][0] if rows else 0 - last_count = 0 - while True: - time.sleep(10) - count = get_count() - log.info(f"count={count} speed={(count - last_count) / 10}") - last_count = count + last_count = 0 + while True: + try: + done = done_queue.get_nowait() + if done == _DONE_MESSAGE: + break + except Empty: + pass + time.sleep(10) + count = get_count() + log.info(f"count={count} speed={(count - last_count) / 10}") + last_count = count + finally: + conn.close() # ANCHOR_END: monitor # ANCHOR: main -def main(): +def main(infinity): set_global_config() logging.info(f"READ_TASK_COUNT={READ_TASK_COUNT}, WRITE_TASK_COUNT={WRITE_TASK_COUNT}, " f"TABLE_COUNT={TABLE_COUNT}, QUEUE_SIZE={QUEUE_SIZE}, MAX_BATCH_SIZE={MAX_BATCH_SIZE}") - monitor_process = Process(target=run_monitor_process) + conn = get_connection() + conn.execute("DROP DATABASE IF EXISTS test") + conn.execute("CREATE DATABASE IF NOT EXISTS test") + conn.execute("CREATE STABLE IF NOT EXISTS test.meters (ts TIMESTAMP, current FLOAT, voltage INT, phase FLOAT) " + "TAGS (location BINARY(64), groupId INT)") + conn.close() + + done_queue = Queue() + monitor_process = Process(target=run_monitor_process, args=(done_queue,)) monitor_process.start() - time.sleep(3) # waiting for database ready. + logging.debug(f"monitor task started with pid {monitor_process.pid}") task_queues: List[Queue] = [] + write_processes = [] + read_processes = [] + # create task queues for i in range(WRITE_TASK_COUNT): - queue = Queue(max_size_bytes=QUEUE_SIZE) + queue = Queue() task_queues.append(queue) # create write processes for i in range(WRITE_TASK_COUNT): - p = Process(target=run_write_task, args=(i, task_queues[i])) + p = Process(target=run_write_task, args=(i, task_queues[i], done_queue)) p.start() logging.debug(f"WriteTask-{i} started with pid {p.pid}") write_processes.append(p) @@ -151,13 +189,19 @@ def main(): # create read processes for i in range(READ_TASK_COUNT): queues = assign_queues(i, task_queues) - p = Process(target=run_read_task, args=(i, queues)) + p = Process(target=run_read_task, args=(i, queues, infinity)) p.start() logging.debug(f"ReadTask-{i} started with pid {p.pid}") read_processes.append(p) try: monitor_process.join() + for p in read_processes: + p.join() + for p in write_processes: + p.join() + time.sleep(1) + return except KeyboardInterrupt: monitor_process.terminate() [p.terminate() for p in read_processes] @@ -176,5 +220,6 @@ def assign_queues(read_task_id, task_queues): if __name__ == '__main__': - main() + multiprocessing.set_start_method('spawn') + main(False) # ANCHOR_END: main diff --git a/docs/examples/python/kafka_example.py b/docs/examples/python/kafka_example.py index 735059eec0..a89287d372 100644 --- a/docs/examples/python/kafka_example.py +++ b/docs/examples/python/kafka_example.py @@ -26,7 +26,8 @@ class Consumer(object): 'bath_consume': True, 'batch_size': 1000, 'async_model': True, - 'workers': 10 + 'workers': 10, + 'testing': False } LOCATIONS = ['California.SanFrancisco', 'California.LosAngles', 'California.SanDiego', 'California.SanJose', @@ -46,11 +47,12 @@ class Consumer(object): def __init__(self, **configs): self.config: dict = self.DEFAULT_CONFIGS self.config.update(configs) - self.consumer = KafkaConsumer( - self.config.get('kafka_topic'), # topic - bootstrap_servers=self.config.get('kafka_brokers'), - group_id=self.config.get('kafka_group_id'), - ) + if not self.config.get('testing'): + self.consumer = KafkaConsumer( + self.config.get('kafka_topic'), # topic + bootstrap_servers=self.config.get('kafka_brokers'), + group_id=self.config.get('kafka_group_id'), + ) self.taos = taos.connect( host=self.config.get('taos_host'), user=self.config.get('taos_user'), @@ -60,7 +62,7 @@ class Consumer(object): ) if self.config.get('async_model'): self.pool = ThreadPoolExecutor(max_workers=self.config.get('workers')) - self.tasks: list[Future] = [] + self.tasks = [] # tags and table mapping # key: {location}_{groupId} value: self.tag_table_mapping = {} i = 0 @@ -104,8 +106,8 @@ class Consumer(object): for task in self.tasks: while not task.done(): pass - if self.pool is not None: - self.pool.shutdown() + if self.pool is not None: + self.pool.shutdown() # clean data if self.config.get('clean_after_testing'): @@ -115,14 +117,14 @@ class Consumer(object): if self.taos is not None: self.taos.close() - def _run(self, f: Callable[[ConsumerRecord], bool]): + def _run(self, f): for message in self.consumer: if self.config.get('async_model'): self.pool.submit(f(message)) else: f(message) - def _run_batch(self, f: Callable[[list[list[ConsumerRecord]]], None]): + def _run_batch(self, f): while True: messages = self.consumer.poll(timeout_ms=500, max_records=self.config.get('batch_size')) if messages: @@ -140,7 +142,7 @@ class Consumer(object): logging.info('## insert sql %s', sql) return self.taos.execute(sql=sql) == 1 - def _to_taos_batch(self, messages: list[list[ConsumerRecord]]): + def _to_taos_batch(self, messages): sql = self._build_sql_batch(messages=messages) if len(sql) == 0: # decode error, skip return @@ -162,7 +164,7 @@ class Consumer(object): table_name = self._get_table_name(location=location, group_id=group_id) return self.INSERT_PART_SQL.format(table_name, ts, current, voltage, phase) - def _build_sql_batch(self, messages: list[list[ConsumerRecord]]) -> str: + def _build_sql_batch(self, messages) -> str: sql_list = [] for partition_messages in messages: for message in partition_messages: @@ -186,7 +188,55 @@ def _get_location_and_group(key: str) -> (str, int): return fields[0], fields[1] +def test_to_taos(consumer: Consumer): + msg = { + 'location': 'California.SanFrancisco', + 'groupId': 1, + 'ts': '2022-12-06 15:13:38.643', + 'current': 3.41, + 'voltage': 105, + 'phase': 0.02027, + } + record = ConsumerRecord(checksum=None, headers=None, offset=1, key=None, value=json.dumps(msg), partition=1, + topic='test', serialized_key_size=None, serialized_header_size=None, + serialized_value_size=None, timestamp=time.time(), timestamp_type=None) + assert consumer._to_taos(message=record) + + +def test_to_taos_batch(consumer: Consumer): + records = [ + [ + ConsumerRecord(checksum=None, headers=None, offset=1, key=None, + value=json.dumps({'location': 'California.SanFrancisco', + 'groupId': 1, + 'ts': '2022-12-06 15:13:38.643', + 'current': 3.41, + 'voltage': 105, + 'phase': 0.02027, }), + partition=1, topic='test', serialized_key_size=None, serialized_header_size=None, + serialized_value_size=None, timestamp=time.time(), timestamp_type=None), + ConsumerRecord(checksum=None, headers=None, offset=1, key=None, + value=json.dumps({'location': 'California.LosAngles', + 'groupId': 2, + 'ts': '2022-12-06 15:13:39.643', + 'current': 3.41, + 'voltage': 102, + 'phase': 0.02027, }), + partition=1, topic='test', serialized_key_size=None, serialized_header_size=None, + serialized_value_size=None, timestamp=time.time(), timestamp_type=None), + ] + ] + + consumer._to_taos_batch(messages=records) + + if __name__ == '__main__': - consumer = Consumer(async_model=True) + consumer = Consumer(async_model=True, testing=True) + # init env consumer.init_env() - consumer.consume() \ No newline at end of file + # consumer.consume() + # test build sql + # test build sql batch + test_to_taos(consumer) + test_to_taos_batch(consumer) + \ No newline at end of file diff --git a/docs/examples/python/mockdatasource.py b/docs/examples/python/mockdatasource.py index 1c516a800e..15a7d2ff8c 100644 --- a/docs/examples/python/mockdatasource.py +++ b/docs/examples/python/mockdatasource.py @@ -10,13 +10,14 @@ class MockDataSource: "9.4,118,0.141,California.SanFrancisco,4" ] - def __init__(self, tb_name_prefix, table_count): + def __init__(self, tb_name_prefix, table_count, infinity=True): self.table_name_prefix = tb_name_prefix + "_" self.table_count = table_count self.max_rows = 10000000 self.current_ts = round(time.time() * 1000) - self.max_rows * 100 # [(tableId, tableName, values),] self.data = self._init_data() + self.infinity = infinity def _init_data(self): lines = self.samples * (self.table_count // 5 + 1) @@ -28,14 +29,19 @@ class MockDataSource: def __iter__(self): self.row = 0 - return self + if not self.infinity: + return iter(self._iter_data()) + else: + return self def __next__(self): """ next 1000 rows for each table. return: {tableId:[row,...]} """ - # generate 1000 timestamps + return self._iter_data() + + def _iter_data(self): ts = [] for _ in range(1000): self.current_ts += 100 @@ -47,3 +53,10 @@ class MockDataSource: rows = [table_name + ',' + t + ',' + values for t in ts] result.append((table_id, rows)) return result + + +if __name__ == '__main__': + datasource = MockDataSource('t', 10, False) + for data in datasource: + print(data) + \ No newline at end of file diff --git a/docs/examples/python/sql_writer.py b/docs/examples/python/sql_writer.py index 758167376b..db51bb7174 100644 --- a/docs/examples/python/sql_writer.py +++ b/docs/examples/python/sql_writer.py @@ -10,6 +10,7 @@ class SQLWriter: self._tb_tags = {} self._conn = get_connection_func() self._max_sql_length = self.get_max_sql_length() + self._conn.execute("create database if not exists test") self._conn.execute("USE test") def get_max_sql_length(self): @@ -20,7 +21,7 @@ class SQLWriter: return int(r[1]) return 1024 * 1024 - def process_lines(self, lines: str): + def process_lines(self, lines: [str]): """ :param lines: [[tbName,ts,current,voltage,phase,location,groupId]] """ @@ -60,6 +61,7 @@ class SQLWriter: buf.append(q) sql_len += len(q) sql += " ".join(buf) + self.create_tables() self.execute_sql(sql) self._tb_values.clear() @@ -88,3 +90,23 @@ class SQLWriter: except BaseException as e: self.log.error("Execute SQL: %s", sql) raise e + + def close(self): + if self._conn: + self._conn.close() + + +if __name__ == '__main__': + def get_connection_func(): + conn = taos.connect() + return conn + + + writer = SQLWriter(get_connection_func=get_connection_func) + writer.execute_sql( + "create stable if not exists meters (ts timestamp, current float, voltage int, phase float) " + "tags (location binary(64), groupId int)") + writer.execute_sql( + "INSERT INTO d21001 USING meters TAGS ('California.SanFrancisco', 2) " + "VALUES ('2021-07-13 14:06:32.272', 10.2, 219, 0.32)") + \ No newline at end of file diff --git a/docs/examples/python/tmq_example.py b/docs/examples/python/tmq_example.py index a4625ca11a..32778e9f25 100644 --- a/docs/examples/python/tmq_example.py +++ b/docs/examples/python/tmq_example.py @@ -1,58 +1,55 @@ +from taos.tmq import Consumer import taos -from taos.tmq import * - -conn = taos.connect() - -print("init") -conn.execute("drop topic if exists topic_ctb_column") -conn.execute("drop database if exists py_tmq") -conn.execute("create database if not exists py_tmq vgroups 2") -conn.select_db("py_tmq") -conn.execute( - "create stable if not exists stb1 (ts timestamp, c1 int, c2 float, c3 binary(10)) tags(t1 int)" -) -conn.execute("create table if not exists tb1 using stb1 tags(1)") -conn.execute("create table if not exists tb2 using stb1 tags(2)") -conn.execute("create table if not exists tb3 using stb1 tags(3)") - -print("create topic") -conn.execute( - "create topic if not exists topic_ctb_column as select ts, c1, c2, c3 from stb1" -) - -print("build consumer") -conf = TaosTmqConf() -conf.set("group.id", "tg2") -conf.set("td.connect.user", "root") -conf.set("td.connect.pass", "taosdata") -conf.set("enable.auto.commit", "true") -def tmq_commit_cb_print(tmq, resp, offset, param=None): - print(f"commit: {resp}, tmq: {tmq}, offset: {offset}, param: {param}") +def init_tmq_env(db, topic): + conn = taos.connect() + conn.execute("drop topic if exists {}".format(topic)) + conn.execute("drop database if exists {}".format(db)) + conn.execute("create database if not exists {}".format(db)) + conn.select_db(db) + conn.execute( + "create stable if not exists stb1 (ts timestamp, c1 int, c2 float, c3 varchar(16)) tags(t1 int, t3 varchar(16))") + conn.execute("create table if not exists tb1 using stb1 tags(1, 't1')") + conn.execute("create table if not exists tb2 using stb1 tags(2, 't2')") + conn.execute("create table if not exists tb3 using stb1 tags(3, 't3')") + conn.execute("create topic if not exists {} as select ts, c1, c2, c3 from stb1".format(topic)) + conn.execute("insert into tb1 values (now, 1, 1.0, 'tmq test')") + conn.execute("insert into tb2 values (now, 2, 2.0, 'tmq test')") + conn.execute("insert into tb3 values (now, 3, 3.0, 'tmq test')") -conf.set_auto_commit_cb(tmq_commit_cb_print, None) -tmq = conf.new_consumer() +def cleanup(db, topic): + conn = taos.connect() + conn.execute("drop topic if exists {}".format(topic)) + conn.execute("drop database if exists {}".format(db)) -print("build topic list") -topic_list = TaosTmqList() -topic_list.append("topic_ctb_column") +if __name__ == '__main__': + init_tmq_env("tmq_test", "tmq_test_topic") # init env + consumer = Consumer( + { + "group.id": "tg2", + "td.connect.user": "root", + "td.connect.pass": "taosdata", + "enable.auto.commit": "true", + } + ) + consumer.subscribe(["tmq_test_topic"]) -print("basic consume loop") -tmq.subscribe(topic_list) + try: + while True: + res = consumer.poll(1) + if not res: + break + err = res.error() + if err is not None: + raise err + val = res.value() -sub_list = tmq.subscription() - -print("subscribed topics: ", sub_list) - -while 1: - res = tmq.poll(1000) - if res: - topic = res.get_topic_name() - vg = res.get_vgroup_id() - db = res.get_db_name() - print(f"topic: {topic}\nvgroup id: {vg}\ndb: {db}") - for row in res: - print(row) + for block in val: + print(block.fetchall()) + finally: + consumer.unsubscribe() + consumer.close() + cleanup("tmq_test", "tmq_test_topic") \ No newline at end of file diff --git a/tests/docs-examples-test/python.sh b/tests/docs-examples-test/python.sh index 140d05395b..ccb391b752 100644 --- a/tests/docs-examples-test/python.sh +++ b/tests/docs-examples-test/python.sh @@ -23,7 +23,7 @@ python3 bind_param_example.py # 4 taos -s "drop database power" -python3 multi_bind_example.py +python3 multi_bind_example.py # 5 python3 query_example.py @@ -44,4 +44,43 @@ taos -s "drop database test" python3 json_protocol_example.py # 10 -# python3 subscribe_demo.py +pip install SQLAlchemy +pip install pandas +taosBenchmark -y -d power -t 10 -n 10 +python3 conn_native_pandas.py +python3 conn_rest_pandas.py +taos -s "drop database if exists power" + +# 11 +taos -s "create database if not exists test" +python3 connect_native_reference.py + +# 12 +python3 connect_rest_examples.py + +# 13 +python3 handle_exception.py + +# 14 +taosBenchmark -y -d power -t 2 -n 10 +python3 rest_client_example.py +taos -s "drop database if exists power" + +# 15 +python3 result_set_examples.py + +# 16 +python3 tmq_example.py + +# 17 +python3 sql_writer.py + +# 18 +python3 mockdatasource.py + +# 19 +python3 fast_write_example.py + +# 20 +pip3 install kafka-python +python3 kafka_example.py From cabf3e86839de09bbfa45a623319b7d43f57d0e2 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Tue, 31 Jan 2023 11:22:06 +0800 Subject: [PATCH 051/124] fix: invalid read --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 1bfd630331..98c9c0fdda 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1643,8 +1643,8 @@ _exit: if (code) { tsdbError("vgId:%d %s failed at line %d since %s", TD_VID(pWriter->pTsdb->pVnode), __func__, lino, tstrerror(code)); } else { - tsdbTrace("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, pId->suid, - pId->uid); + tsdbTrace("vgId:%d %s done, suid:%" PRId64 " uid:%" PRId64, TD_VID(pWriter->pTsdb->pVnode), __func__, + pWriter->tbid.suid, pWriter->tbid.uid); } return code; } From a84e990c4e1f99cd2f68c7913c2c56197cba2f7f Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Tue, 31 Jan 2023 11:29:46 +0800 Subject: [PATCH 052/124] fix: not try to propose commit on vmCloseVnode --- source/dnode/mgmt/mgmt_vnode/src/vmInt.c | 2 -- 1 file changed, 2 deletions(-) diff --git a/source/dnode/mgmt/mgmt_vnode/src/vmInt.c b/source/dnode/mgmt/mgmt_vnode/src/vmInt.c index 99ba9b9b3b..951544c766 100644 --- a/source/dnode/mgmt/mgmt_vnode/src/vmInt.c +++ b/source/dnode/mgmt/mgmt_vnode/src/vmInt.c @@ -79,8 +79,6 @@ int32_t vmOpenVnode(SVnodeMgmt *pMgmt, SWrapperCfg *pCfg, SVnode *pImpl) { void vmCloseVnode(SVnodeMgmt *pMgmt, SVnodeObj *pVnode) { char path[TSDB_FILENAME_LEN] = {0}; - vnodeProposeCommitOnNeed(pVnode->pImpl); - taosThreadRwlockWrlock(&pMgmt->lock); taosHashRemove(pMgmt->hash, &pVnode->vgId, sizeof(int32_t)); taosThreadRwlockUnlock(&pMgmt->lock); From b56db2026b7c7a30cc94196edf5b1b887ae52996 Mon Sep 17 00:00:00 2001 From: Xuefeng Tan <1172915550@qq.com> Date: Tue, 31 Jan 2023 15:14:48 +0800 Subject: [PATCH 053/124] fix(taosAdapter): set option `TSDB_OPTION_USE_ADAPTER = true` (#19711) --- cmake/taosadapter_CMakeLists.txt.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/taosadapter_CMakeLists.txt.in b/cmake/taosadapter_CMakeLists.txt.in index 13b247770e..d156057459 100644 --- a/cmake/taosadapter_CMakeLists.txt.in +++ b/cmake/taosadapter_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taosadapter ExternalProject_Add(taosadapter GIT_REPOSITORY https://github.com/taosdata/taosadapter.git - GIT_TAG 213f8b3 + GIT_TAG 3e08996 SOURCE_DIR "${TD_SOURCE_DIR}/tools/taosadapter" BINARY_DIR "" #BUILD_IN_SOURCE TRUE From c8394cbf6e6c67f4679d7949792dfe20d0e3866b Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Tue, 31 Jan 2023 16:38:29 +0800 Subject: [PATCH 054/124] fix: add get tb from mem time info --- source/dnode/vnode/src/tsdb/tsdbRead.c | 15 ++++++++++++--- source/libs/executor/src/tfill.c | 26 -------------------------- 2 files changed, 12 insertions(+), 29 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 85282a2340..e3240ec7d7 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -79,6 +79,8 @@ typedef struct SIOCostSummary { int64_t composedBlocks; double buildComposedBlockTime; double createScanInfoList; + double getTbFromMemTime; + double getTbFromIMemTime; } SIOCostSummary; typedef struct SBlockLoadSuppInfo { @@ -2177,10 +2179,13 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea } int32_t backward = (!ASCENDING_TRAVERSE(pReader->order)); - + int64_t st = 0; + STbData* d = NULL; if (pReader->pReadSnap->pMem != NULL) { + st = taosGetTimestampUs(); d = tsdbGetTbDataFromMemTable(pReader->pReadSnap->pMem, pReader->suid, pBlockScanInfo->uid); + pReader->cost.getTbFromMemTime += (taosGetTimestampUs() - st) / 1000.0; if (d != NULL) { code = tsdbTbDataIterCreate(d, &startKey, backward, &pBlockScanInfo->iter.iter); if (code == TSDB_CODE_SUCCESS) { @@ -2201,7 +2206,9 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea STbData* di = NULL; if (pReader->pReadSnap->pIMem != NULL) { + st = taosGetTimestampUs(); di = tsdbGetTbDataFromMemTable(pReader->pReadSnap->pIMem, pReader->suid, pBlockScanInfo->uid); + pReader->cost.getTbFromIMemTime += (taosGetTimestampUs() - st) / 1000.0; if (di != NULL) { code = tsdbTbDataIterCreate(di, &startKey, backward, &pBlockScanInfo->iiter.iter); if (code == TSDB_CODE_SUCCESS) { @@ -4005,11 +4012,13 @@ void tsdbReaderClose(STsdbReader* pReader) { ", fileBlocks-load-time:%.2f ms, " "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 - ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms, %s", + ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms," + ", getTbFromMem-time:%.2f ms, getTbFromIMem-time:%.2f ms, %s", pReader, pCost->headFileLoad, pCost->headFileLoadTime, pCost->smaDataLoad, pCost->smaLoadTime, pCost->numOfBlocks, pCost->blockLoadTime, pCost->buildmemBlock, pCost->lastBlockLoad, pCost->lastBlockLoadTime, pCost->composedBlocks, pCost->buildComposedBlockTime, - numOfTables * sizeof(STableBlockScanInfo) / 1000.0, pCost->createScanInfoList, pReader->idStr); + numOfTables * sizeof(STableBlockScanInfo) / 1000.0, pCost->createScanInfoList, + pCost->getTbFromMemTime, pCost->getTbFromIMemTime, pReader->idStr); taosMemoryFree(pReader->idStr); taosMemoryFree(pReader->pSchema); diff --git a/source/libs/executor/src/tfill.c b/source/libs/executor/src/tfill.c index 2d921d43d3..826af20224 100644 --- a/source/libs/executor/src/tfill.c +++ b/source/libs/executor/src/tfill.c @@ -447,32 +447,6 @@ struct SFillInfo* taosCreateFillInfo(TSKEY skey, int32_t numOfFillCols, int32_t taosResetFillInfo(pFillInfo, skey); - switch (fillType) { - case FILL_MODE_NONE: - pFillInfo->type = TSDB_FILL_NONE; - break; - case FILL_MODE_PREV: - pFillInfo->type = TSDB_FILL_PREV; - break; - case FILL_MODE_NULL: - pFillInfo->type = TSDB_FILL_NULL; - break; - case FILL_MODE_LINEAR: - pFillInfo->type = TSDB_FILL_LINEAR; - break; - case FILL_MODE_NEXT: - pFillInfo->type = TSDB_FILL_NEXT; - break; - case FILL_MODE_VALUE: - pFillInfo->type = TSDB_FILL_SET_VALUE; - break; - default: { - taosMemoryFree(pFillInfo); - terrno = TSDB_CODE_INVALID_PARA; - return NULL; - } - } - pFillInfo->type = fillType; pFillInfo->pFillCol = pCol; pFillInfo->numOfCols = numOfFillCols + numOfNotFillCols; From f6df973d4d2e3ae42da47fc82f9ead04b353fff2 Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Wed, 1 Feb 2023 10:15:42 +0800 Subject: [PATCH 055/124] fix: no cache auto create table error --- source/libs/parser/src/parInsertSql.c | 3 +++ 1 file changed, 3 insertions(+) diff --git a/source/libs/parser/src/parInsertSql.c b/source/libs/parser/src/parInsertSql.c index 98c6aed829..1e1821842f 100644 --- a/source/libs/parser/src/parInsertSql.c +++ b/source/libs/parser/src/parInsertSql.c @@ -1760,6 +1760,9 @@ static int32_t getTableSchemaFromMetaData(SInsertParseContext* pCxt, const SMeta if (TSDB_CODE_SUCCESS == code && !isStb && TSDB_SUPER_TABLE == pStmt->pTableMeta->tableType) { code = buildInvalidOperationMsg(&pCxt->msg, "insert data into super table is not supported"); } + if (TSDB_CODE_SUCCESS == code && isStb) { + code = storeTableMeta(pCxt, pStmt); + } if (TSDB_CODE_SUCCESS == code) { code = getTableVgroupFromMetaData(pMetaData->pTableHash, pStmt, isStb); } From c4531512303fb4129b22cde6552bfe193cce80c5 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Wed, 1 Feb 2023 15:56:58 +0800 Subject: [PATCH 056/124] test(query): add regress test case. --- tests/script/tsim/parser/regressiontest.sim | 34 +++++++++++++++++++++ 1 file changed, 34 insertions(+) diff --git a/tests/script/tsim/parser/regressiontest.sim b/tests/script/tsim/parser/regressiontest.sim index 1b127155cb..3ce2b47b44 100644 --- a/tests/script/tsim/parser/regressiontest.sim +++ b/tests/script/tsim/parser/regressiontest.sim @@ -63,4 +63,38 @@ if $rows != 8198 then return -1 endi +print ===========================> TD-22077 && TD-21877 +sql drop database if exists $db -x step1 +sql create database $db vgroups 1; + +sql use $db +sql create stable st1 (ts timestamp, c int) tags(a int); +sql create table t1 using st1 tags(1); +sql create table t2 using st1 tags(2); + +$i = 0 +$ts = 1674977959000 +$rowNum = 200 + +$x = 0 +while $x < $rowNum +$xs = $x * $delta +$ts = $ts0 + $xs +sql insert into t1 values ( $ts , $x ) +sql insert into t2 values ( $ts + 1000a, $x ) +$x = $x + 1 +$ts = $ts + 1000 +endw + +sql flush database $db + +sql insert into t1 values('2018-09-17 09:00:26', 26); +sql insert into t2 values('2018-09-17 09:00:25', 25); + +sql insert into t2 values('2018-09-17 09:00:30', 30); +sql flush database reg_db0; + +sql delete from st1 where ts<='2018-9-17 09:00:26'; +sql select * from st1; + system sh/exec.sh -n dnode1 -s stop -x SIGINT From a3a2af4b3fb35f9c62ad43a86b388a43814bbcf0 Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Wed, 1 Feb 2023 15:59:29 +0800 Subject: [PATCH 057/124] fix: taosbenchmark schemaless refine for main (#19714) * fix: taos-tools 143b9e4 for taosbenchmark schemaless refine for main * fix: update taos-tools 723f696 * test: fix ../tests/develop-test/5-taos-tools/taosbenchmark/sml_json_alltypes.py * test: check nchar temporarily as diff behavior between main and 3.0 branch * fix: update taos-tools 181bcac --- cmake/taostools_CMakeLists.txt.in | 2 +- .../taosbenchmark/sml_json_alltypes.py | 62 ++++++++++++------- 2 files changed, 42 insertions(+), 22 deletions(-) diff --git a/cmake/taostools_CMakeLists.txt.in b/cmake/taostools_CMakeLists.txt.in index 13a81f88ea..926d0c63e7 100644 --- a/cmake/taostools_CMakeLists.txt.in +++ b/cmake/taostools_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taos-tools ExternalProject_Add(taos-tools GIT_REPOSITORY https://github.com/taosdata/taos-tools.git - GIT_TAG 0cd564a + GIT_TAG 181bcac SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools" BINARY_DIR "" #BUILD_IN_SOURCE TRUE diff --git a/tests/develop-test/5-taos-tools/taosbenchmark/sml_json_alltypes.py b/tests/develop-test/5-taos-tools/taosbenchmark/sml_json_alltypes.py index 2c6d09b0f5..1b65e38d72 100644 --- a/tests/develop-test/5-taos-tools/taosbenchmark/sml_json_alltypes.py +++ b/tests/develop-test/5-taos-tools/taosbenchmark/sml_json_alltypes.py @@ -19,32 +19,38 @@ from util.dnodes import * class TDTestCase: def caseDescription(self): - ''' + """ [TD-11510] taosBenchmark test cases - ''' - return + """ def init(self, conn, logSql, replicaVar=1): - self.replicaVar = int(replicaVar) tdLog.debug("start to execute %s" % __file__) + self.replicaVar = int(replicaVar) tdSql.init(conn.cursor(), logSql) def getPath(self, tool="taosBenchmark"): selfPath = os.path.dirname(os.path.realpath(__file__)) - if ("community" in selfPath): - projPath = selfPath[:selfPath.find("community")] + if "community" in selfPath: + projPath = selfPath[: selfPath.find("community")] + elif "src" in selfPath: + projPath = selfPath[: selfPath.find("src")] + elif "/tools/" in selfPath: + projPath = selfPath[: selfPath.find("/tools/")] + elif "/tests/" in selfPath: + projPath = selfPath[: selfPath.find("/tests/")] else: - projPath = selfPath[:selfPath.find("tests")] + tdLog.info("cannot found %s in path: %s, use system's" % (tool, selfPath)) + projPath = "/usr/local/taos/bin/" paths = [] - for root, dirs, files in os.walk(projPath): - if ((tool) in files): + for root, dummy, files in os.walk(projPath): + if (tool) in files: rootRealPath = os.path.dirname(os.path.realpath(root)) - if ("packaging" not in rootRealPath): + if "packaging" not in rootRealPath: paths.append(os.path.join(root, tool)) break - if (len(paths) == 0): + if len(paths) == 0: tdLog.exit("taosBenchmark not found!") return else: @@ -52,31 +58,45 @@ class TDTestCase: return paths[0] def run(self): + tdSql.query("select client_version()") + client_ver = "".join(tdSql.queryResult[0]) + major_ver = client_ver.split(".")[0] + binPath = self.getPath() - cmd = "%s -f ./5-taos-tools/taosbenchmark/json/sml_json_alltypes.json" %binPath + cmd = "%s -f ./5-taos-tools/taosbenchmark/json/sml_json_alltypes.json" % binPath tdLog.info("%s" % cmd) os.system("%s" % cmd) tdSql.execute("reset query cache") tdSql.query("describe db.stb1") tdSql.checkData(1, 1, "BOOL") tdSql.query("describe db.stb2") - tdSql.checkData(1, 1, "TINYINT") + tdSql.checkData(1, 1, "DOUBLE") tdSql.query("describe db.stb3") - tdSql.checkData(1, 1, "SMALLINT") + tdSql.checkData(1, 1, "DOUBLE") tdSql.query("describe db.stb4") - tdSql.checkData(1, 1, "INT") + tdSql.checkData(1, 1, "DOUBLE") tdSql.query("describe db.stb5") - tdSql.checkData(1, 1, "BIGINT") + tdSql.checkData(1, 1, "DOUBLE") tdSql.query("describe db.stb6") - tdSql.checkData(1, 1, "FLOAT") + tdSql.checkData(1, 1, "DOUBLE") tdSql.query("describe db.stb7") tdSql.checkData(1, 1, "DOUBLE") tdSql.query("describe db.stb8") - tdSql.checkData(1, 1, "VARCHAR") - tdSql.checkData(1, 2, 16) + if major_ver == "3": + tdSql.checkData(1, 1, "NCHAR") + tdSql.checkData(1, 2, 16) + else: + tdSql.checkData(1, 1, "NCHAR") + tdSql.checkData(1, 2, 8) + tdSql.query("describe db.stb9") - tdSql.checkData(1, 1, "NCHAR") - tdSql.checkData(1, 2, 16) + if major_ver == "3": + tdSql.checkData(1, 1, "NCHAR") + tdSql.checkData(1, 2, 16) + else: + tdSql.checkData(1, 1, "NCHAR") + tdSql.checkData(1, 2, 8) + tdSql.query("select count(*) from db.stb1") tdSql.checkData(0, 0, 160) tdSql.query("select count(*) from db.stb2") From 869ab0d394e08e19839a22300bbd55e059166462 Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Wed, 1 Feb 2023 17:27:56 +0800 Subject: [PATCH 058/124] fix: snode timer --- source/dnode/snode/src/snode.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/source/dnode/snode/src/snode.c b/source/dnode/snode/src/snode.c index 860db20fa8..1d2f4da26b 100644 --- a/source/dnode/snode/src/snode.c +++ b/source/dnode/snode/src/snode.c @@ -93,6 +93,8 @@ int32_t sndExpandTask(SSnode *pSnode, SStreamTask *pTask, int64_t ver) { pTask->exec.executor = qCreateStreamExecTaskInfo(pTask->exec.qmsg, &mgHandle); ASSERT(pTask->exec.executor); + streamSetupTrigger(pTask); + return 0; } From df30b51a7517d63d89c9730d8e52aed920c6de49 Mon Sep 17 00:00:00 2001 From: Minglei Jin Date: Wed, 1 Feb 2023 17:59:39 +0800 Subject: [PATCH 059/124] fix(stream): use tdb page replacement with stream state --- source/libs/stream/src/streamState.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/source/libs/stream/src/streamState.c b/source/libs/stream/src/streamState.c index 6670bf463e..2460da25f4 100644 --- a/source/libs/stream/src/streamState.c +++ b/source/libs/stream/src/streamState.c @@ -192,7 +192,7 @@ void streamStateClose(SStreamState* pState) { } int32_t streamStateBegin(SStreamState* pState) { - if (tdbBegin(pState->pTdbState->db, &pState->pTdbState->txn, tdbDefaultMalloc, tdbDefaultFree, NULL, + if (tdbBegin(pState->pTdbState->db, &pState->pTdbState->txn, NULL, NULL, NULL, TDB_TXN_WRITE | TDB_TXN_READ_UNCOMMITTED) < 0) { tdbAbort(pState->pTdbState->db, pState->pTdbState->txn); return -1; @@ -208,7 +208,7 @@ int32_t streamStateCommit(SStreamState* pState) { return -1; } - if (tdbBegin(pState->pTdbState->db, &pState->pTdbState->txn, tdbDefaultMalloc, tdbDefaultFree, NULL, + if (tdbBegin(pState->pTdbState->db, &pState->pTdbState->txn, NULL, NULL, NULL, TDB_TXN_WRITE | TDB_TXN_READ_UNCOMMITTED) < 0) { return -1; } @@ -220,7 +220,7 @@ int32_t streamStateAbort(SStreamState* pState) { return -1; } - if (tdbBegin(pState->pTdbState->db, &pState->pTdbState->txn, tdbDefaultMalloc, tdbDefaultFree, NULL, + if (tdbBegin(pState->pTdbState->db, &pState->pTdbState->txn, NULL, NULL, NULL, TDB_TXN_WRITE | TDB_TXN_READ_UNCOMMITTED) < 0) { return -1; } From 8e4c6ccbb3c49a92296dc6d424d26af947c467a7 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Wed, 1 Feb 2023 18:15:35 +0800 Subject: [PATCH 060/124] fix(query): move the reset to other place. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 19 +++++++++++++++---- source/util/src/talgo.c | 8 ++++---- 2 files changed, 19 insertions(+), 8 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index a77eb8298e..0bb9ebc40e 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -2693,10 +2693,6 @@ static void extractOrderedTableUidList(SUidOrderCheckInfo* pOrderCheckInfo, SRea void* p = taosHashIterate(pStatus->pTableMap, NULL); while (p != NULL) { STableBlockScanInfo* pScanInfo = *(STableBlockScanInfo**)p; - - // reset the last del file index - pScanInfo->lastBlockDelIndex = getInitialDelIndex(pScanInfo->delSkyline, order); - pOrderCheckInfo->tableUidList[index++] = pScanInfo->uid; p = taosHashIterate(pStatus->pTableMap, p); } @@ -2704,6 +2700,18 @@ static void extractOrderedTableUidList(SUidOrderCheckInfo* pOrderCheckInfo, SRea taosSort(pOrderCheckInfo->tableUidList, total, sizeof(uint64_t), uidComparFunc); } +// reset the last del file index +static void resetScanBlockLastBlockDelIndex(SReaderStatus* pStatus, int32_t order) { + void* p = taosHashIterate(pStatus->pTableMap, NULL); + while (p != NULL) { + STableBlockScanInfo* pScanInfo = *(STableBlockScanInfo**)p; + + // reset the last del file index + pScanInfo->lastBlockDelIndex = getInitialDelIndex(pScanInfo->delSkyline, order); + p = taosHashIterate(pStatus->pTableMap, p); + } +} + static int32_t initOrderCheckInfo(SUidOrderCheckInfo* pOrderCheckInfo, STsdbReader* pReader) { SReaderStatus* pStatus = &pReader->status; @@ -3052,6 +3060,7 @@ static int32_t buildBlockFromFiles(STsdbReader* pReader) { // this file does not have data files, let's start check the last block file if exists if (pBlockIter->numOfBlocks == 0) { + resetScanBlockLastBlockDelIndex(&pReader->status, pReader->order); goto _begin; } } @@ -3083,6 +3092,7 @@ static int32_t buildBlockFromFiles(STsdbReader* pReader) { // data blocks in current file are exhausted, let's try the next file now tBlockDataReset(&pReader->status.fileBlockData); resetDataBlockIterator(pBlockIter, pReader->order); + resetScanBlockLastBlockDelIndex(&pReader->status, pReader->order); goto _begin; } else { code = initForFirstBlockInFile(pReader, pBlockIter); @@ -3094,6 +3104,7 @@ static int32_t buildBlockFromFiles(STsdbReader* pReader) { // this file does not have blocks, let's start check the last block file if (pBlockIter->numOfBlocks == 0) { + resetScanBlockLastBlockDelIndex(&pReader->status, pReader->order); goto _begin; } } diff --git a/source/util/src/talgo.c b/source/util/src/talgo.c index a06aac6afe..e373850b3c 100644 --- a/source/util/src/talgo.c +++ b/source/util/src/talgo.c @@ -28,14 +28,14 @@ static void median(void *src, int64_t size, int64_t s, int64_t e, const void *pa void *buf) { int32_t mid = ((int32_t)(e - s) >> 1u) + (int32_t)s; - if (comparFn(elePtrAt(src, size, mid), elePtrAt(src, size, s), param) == 1) { + if (comparFn(elePtrAt(src, size, mid), elePtrAt(src, size, s), param) > 0) { doswap(elePtrAt(src, size, mid), elePtrAt(src, size, s), size, buf); } - if (comparFn(elePtrAt(src, size, mid), elePtrAt(src, size, e), param) == 1) { + if (comparFn(elePtrAt(src, size, mid), elePtrAt(src, size, e), param) > 0) { doswap(elePtrAt(src, size, mid), elePtrAt(src, size, s), size, buf); doswap(elePtrAt(src, size, mid), elePtrAt(src, size, e), size, buf); - } else if (comparFn(elePtrAt(src, size, s), elePtrAt(src, size, e), param) == 1) { + } else if (comparFn(elePtrAt(src, size, s), elePtrAt(src, size, e), param) > 0) { doswap(elePtrAt(src, size, s), elePtrAt(src, size, e), size, buf); } @@ -47,7 +47,7 @@ static void tInsertSort(void *src, int64_t size, int32_t s, int32_t e, const voi void *buf) { for (int32_t i = s + 1; i <= e; ++i) { for (int32_t j = i; j > s; --j) { - if (comparFn(elePtrAt(src, size, j), elePtrAt(src, size, j - 1), param) == -1) { + if (comparFn(elePtrAt(src, size, j), elePtrAt(src, size, j - 1), param) < 0) { doswap(elePtrAt(src, size, j), elePtrAt(src, size, j - 1), size, buf); } else { break; From 043fc8d98046afbfea2ae8cb84899391f88b83de Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Wed, 1 Feb 2023 19:11:51 +0800 Subject: [PATCH 061/124] enh: improve logging msgs for sync snapshot repl --- source/libs/sync/src/syncUtil.c | 133 +++++++++++++++----------------- 1 file changed, 63 insertions(+), 70 deletions(-) diff --git a/source/libs/sync/src/syncUtil.c b/source/libs/sync/src/syncUtil.c index b246d9a79d..6a50572cba 100644 --- a/source/libs/sync/src/syncUtil.c +++ b/source/libs/sync/src/syncUtil.c @@ -141,20 +141,15 @@ static void syncLogReplMgrStates2Str(SSyncNode* pSyncNode, char* buf, int32_t bu } static void syncPeerState2Str(SSyncNode* pSyncNode, char* buf, int32_t bufLen) { - int32_t len = 1; - + int32_t len = 0; + len += snprintf(buf + len, bufLen - len, "%s", "{"); for (int32_t i = 0; i < pSyncNode->replicaNum; ++i) { SPeerState* pState = syncNodeGetPeerState(pSyncNode, &(pSyncNode->replicasId[i])); if (pState == NULL) break; - - if (i < pSyncNode->replicaNum - 1) { - len += snprintf(buf + len, bufLen - len, "%d:%" PRId64 " %" PRId64 ", ", i, pState->lastSendIndex, - pState->lastSendTime); - } else { - len += snprintf(buf + len, bufLen - len, "%d:%" PRId64 " %" PRId64 "}", i, pState->lastSendIndex, - pState->lastSendTime); - } + len += snprintf(buf + len, bufLen - len, "%d:%" PRId64 " %" PRId64 "%s", i, pState->lastSendIndex, + pState->lastSendTime, (i < pSyncNode->replicaNum - 1) ? ", " : ""); } + len += snprintf(buf + len, bufLen - len, "%s", "}"); } void syncPrintNodeLog(const char* flags, ELogLevel level, int32_t dflag, SSyncNode* pNode, const char* format, ...) { @@ -245,7 +240,7 @@ void syncPrintSnapshotSenderLog(const char* flags, ELogLevel level, int32_t dfla char cfgStr[1024] = ""; syncCfg2SimpleStr(&pNode->raftCfg.cfg, cfgStr, sizeof(cfgStr)); - char peerStr[1024] = "{"; + char peerStr[1024] = ""; syncPeerState2Str(pNode, peerStr, sizeof(peerStr)); char eventLog[512]; // {0}; @@ -255,20 +250,21 @@ void syncPrintSnapshotSenderLog(const char* flags, ELogLevel level, int32_t dfla va_end(argpointer); taosPrintLog(flags, level, dflag, - "vgId:%d, %s, sync:%s, {%p s-param:%" PRId64 " e-param:%" PRId64 " laindex:%" PRId64 " laterm:%" PRIu64 - " lcindex:%" PRId64 - " seq:%d ack:%d finish:%d replica-index:%d dnode:%d}" - ", tm:%" PRIu64 ", cmt:%" PRId64 ", fst:%" PRId64 ", lst:%" PRId64 ", min:%" PRId64 ", snap:%" PRId64 - ", snap-tm:%" PRIu64 ", sby:%d, stgy:%d, bch:%d, r-num:%d, lcfg:%" PRId64 - ", chging:%d, rsto:%d, dquorum:%d, elt:%" PRId64 ", hb:%" PRId64 ", %s, %s", + "vgId:%d, %s, sync:%s, snap-sender:{%p start:%" PRId64 " end:%" PRId64 " last-index:%" PRId64 + " last-term:%" PRIu64 " last-cfg:%" PRId64 + ", seq:%d ack:%d finish:%d, as:%d dnode:%d}" + ", term:%" PRIu64 ", commit-index:%" PRId64 ", firstver:%" PRId64 ", lastver:%" PRId64 + ", min-match:%" PRId64 ", snap:{last-index:%" PRId64 ", term:%" PRIu64 + "}, standby:%d, batch-sz:%d, replicas:%d, last-cfg:%" PRId64 + ", chging:%d, restore:%d, quorum:%d, lc-timer:{elect:%" PRId64 ", hb:%" PRId64 "}, peer:%s, cfg:%s", pNode->vgId, eventLog, syncStr(pNode->state), pSender, pSender->snapshotParam.start, pSender->snapshotParam.end, pSender->snapshot.lastApplyIndex, pSender->snapshot.lastApplyTerm, pSender->snapshot.lastConfigIndex, pSender->seq, pSender->ack, pSender->finish, pSender->replicaIndex, DID(&pNode->replicasId[pSender->replicaIndex]), pNode->raftStore.currentTerm, pNode->commitIndex, logBeginIndex, logLastIndex, pNode->minMatchIndex, snapshot.lastApplyIndex, snapshot.lastApplyTerm, - pNode->raftCfg.isStandBy, pNode->raftCfg.snapshotStrategy, pNode->raftCfg.batchSize, pNode->replicaNum, - pNode->raftCfg.lastConfigIndex, pNode->changing, pNode->restoreFinish, syncNodeDynamicQuorum(pNode), - pNode->electTimerLogicClock, pNode->heartbeatTimerLogicClockUser, peerStr, cfgStr); + pNode->raftCfg.isStandBy, pNode->raftCfg.batchSize, pNode->replicaNum, pNode->raftCfg.lastConfigIndex, + pNode->changing, pNode->restoreFinish, syncNodeDynamicQuorum(pNode), pNode->electTimerLogicClock, + pNode->heartbeatTimerLogicClockUser, peerStr, cfgStr); } void syncPrintSnapshotReceiverLog(const char* flags, ELogLevel level, int32_t dflag, SSyncSnapshotReceiver* pReceiver, @@ -291,7 +287,7 @@ void syncPrintSnapshotReceiverLog(const char* flags, ELogLevel level, int32_t df char cfgStr[1024] = ""; syncCfg2SimpleStr(&pNode->raftCfg.cfg, cfgStr, sizeof(cfgStr)); - char peerStr[1024] = "{"; + char peerStr[1024] = ""; syncPeerState2Str(pNode, peerStr, sizeof(peerStr)); char eventLog[512]; // {0}; @@ -300,22 +296,22 @@ void syncPrintSnapshotReceiverLog(const char* flags, ELogLevel level, int32_t df int32_t writeLen = vsnprintf(eventLog, sizeof(eventLog), format, argpointer); va_end(argpointer); - taosPrintLog(flags, level, dflag, - "vgId:%d, %s, sync:%s," - " {%p start:%d ack:%d term:%" PRIu64 " start-time:%" PRId64 " from dnode:%d s-param:%" PRId64 - " e-param:%" PRId64 " laindex:%" PRId64 " laterm:%" PRIu64 " lcindex:%" PRId64 - "}" - ", tm:%" PRIu64 ", cmt:%" PRId64 ", fst:%" PRId64 ", lst:%" PRId64 ", min:%" PRId64 ", snap:%" PRId64 - ", snap-tm:%" PRIu64 ", sby:%d, stgy:%d, bch:%d, r-num:%d, lcfg:%" PRId64 - ", chging:%d, rsto:%d, dquorum:%d, elt:%" PRId64 ", hb:%" PRId64 ", %s, %s", - pNode->vgId, eventLog, syncStr(pNode->state), pReceiver, pReceiver->start, pReceiver->ack, - pReceiver->term, pReceiver->startTime, DID(&pReceiver->fromId), pReceiver->snapshotParam.start, - pReceiver->snapshotParam.end, pReceiver->snapshot.lastApplyIndex, pReceiver->snapshot.lastApplyTerm, - pReceiver->snapshot.lastConfigIndex, pNode->raftStore.currentTerm, pNode->commitIndex, logBeginIndex, - logLastIndex, pNode->minMatchIndex, snapshot.lastApplyIndex, snapshot.lastApplyTerm, - pNode->raftCfg.isStandBy, pNode->raftCfg.snapshotStrategy, pNode->raftCfg.batchSize, pNode->replicaNum, - pNode->raftCfg.lastConfigIndex, pNode->changing, pNode->restoreFinish, syncNodeDynamicQuorum(pNode), - pNode->electTimerLogicClock, pNode->heartbeatTimerLogicClockUser, peerStr, cfgStr); + taosPrintLog( + flags, level, dflag, + "vgId:%d, %s, sync:%s," + " snap-receiver:{%p started:%d acked:%d term:%" PRIu64 " start-time:%" PRId64 " from-dnode:%d, start:%" PRId64 + " end:%" PRId64 " last-index:%" PRId64 " last-term:%" PRIu64 " last-cfg:%" PRId64 + "}" + ", term:%" PRIu64 ", commit-index:%" PRId64 ", firstver:%" PRId64 ", lastver:%" PRId64 ", min-match:%" PRId64 + ", snap:{last-index:%" PRId64 ", last-term:%" PRIu64 "}, standby:%d, batch-sz:%d, replicas:%d, last-cfg:%" PRId64 + ", chging:%d, restore:%d, quorum:%d, lc-timers:{elect:%" PRId64 ", hb:%" PRId64 "}, peer:%s, cfg:%s", + pNode->vgId, eventLog, syncStr(pNode->state), pReceiver, pReceiver->start, pReceiver->ack, pReceiver->term, + pReceiver->startTime, DID(&pReceiver->fromId), pReceiver->snapshotParam.start, pReceiver->snapshotParam.end, + pReceiver->snapshot.lastApplyIndex, pReceiver->snapshot.lastApplyTerm, pReceiver->snapshot.lastConfigIndex, + pNode->raftStore.currentTerm, pNode->commitIndex, logBeginIndex, logLastIndex, pNode->minMatchIndex, + snapshot.lastApplyIndex, snapshot.lastApplyTerm, pNode->raftCfg.isStandBy, pNode->raftCfg.batchSize, + pNode->replicaNum, pNode->raftCfg.lastConfigIndex, pNode->changing, pNode->restoreFinish, + syncNodeDynamicQuorum(pNode), pNode->electTimerLogicClock, pNode->heartbeatTimerLogicClockUser, peerStr, cfgStr); } void syncLogRecvTimer(SSyncNode* pSyncNode, const SyncTimeout* pMsg, const char* s) { @@ -351,13 +347,13 @@ void syncLogSendHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, bool int64_t execTime) { if (printX) { sNTrace(pSyncNode, - "send sync-heartbeat to dnode:%d {term:%" PRId64 ", cmt:%" PRId64 ", min-match:%" PRId64 ", ts:%" PRId64 - "}, x", + "send sync-heartbeat to dnode:%d {term:%" PRId64 ", commit-index:%" PRId64 ", min-match:%" PRId64 + ", ts:%" PRId64 "}, x", DID(&pMsg->destId), pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp); } else { sNTrace(pSyncNode, - "send sync-heartbeat to dnode:%d {term:%" PRId64 ", cmt:%" PRId64 ", min-match:%" PRId64 ", ts:%" PRId64 - "}, timer-elapsed:%" PRId64 ", next-exec:%" PRId64, + "send sync-heartbeat to dnode:%d {term:%" PRId64 ", commit-index:%" PRId64 ", min-match:%" PRId64 + ", ts:%" PRId64 "}, timer-elapsed:%" PRId64 ", next-exec:%" PRId64, DID(&pMsg->destId), pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, timerElapsed, execTime); } @@ -368,14 +364,14 @@ void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64 pSyncNode->hbSlowNum++; sNInfo(pSyncNode, - "recv sync-heartbeat from dnode:%d slow {term:%" PRId64 ", cmt:%" PRId64 ", min-match:%" PRId64 + "recv sync-heartbeat from dnode:%d slow {term:%" PRId64 ", commit-index:%" PRId64 ", min-match:%" PRId64 ", ts:%" PRId64 "}, %s, net elapsed:%" PRId64, DID(&pMsg->srcId), pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, s, timeDiff); } sNTrace(pSyncNode, - "recv sync-heartbeat from dnode:%d {term:%" PRId64 ", cmt:%" PRId64 ", min-match:%" PRId64 ", ts:%" PRId64 - "}, %s, net elapsed:%" PRId64, + "recv sync-heartbeat from dnode:%d {term:%" PRId64 ", commit-index:%" PRId64 ", min-match:%" PRId64 + ", ts:%" PRId64 "}, %s, net elapsed:%" PRId64, DID(&pMsg->srcId), pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, s, timeDiff); } @@ -400,67 +396,64 @@ void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* p void syncLogSendSyncSnapshotSend(SSyncNode* pSyncNode, const SyncSnapshotSend* pMsg, const char* s) { sNDebug(pSyncNode, - "send sync-snapshot-send to dnode:%d, %s, seq:%d, term:%" PRId64 ", begin:%" PRId64 ", end:%" PRId64 - ", lterm:%" PRId64 ", stime:%" PRId64, + "send sync-snapshot-send to dnode:%d, %s, seq:%d, term:%" PRId64 ", begin-index:%" PRId64 + ", last-index:%" PRId64 ", last-term:%" PRId64 ", start-time:%" PRId64, DID(&pMsg->destId), s, pMsg->seq, pMsg->term, pMsg->beginIndex, pMsg->lastIndex, pMsg->lastTerm, pMsg->startTime); } void syncLogRecvSyncSnapshotSend(SSyncNode* pSyncNode, const SyncSnapshotSend* pMsg, const char* s) { sNDebug(pSyncNode, - "recv sync-snapshot-send from dnode:%d, %s, seq:%d, term:%" PRId64 ", begin:%" PRId64 ", lst:%" PRId64 - ", lterm:%" PRId64 ", stime:%" PRId64 ", len:%u", + "recv sync-snapshot-send from dnode:%d, %s, seq:%d, term:%" PRId64 ", begin-index:%" PRId64 + ", last-index:%" PRId64 ", last-term:%" PRId64 ", start-time:%" PRId64 ", data-len:%u", DID(&pMsg->srcId), s, pMsg->seq, pMsg->term, pMsg->beginIndex, pMsg->lastIndex, pMsg->lastTerm, pMsg->startTime, pMsg->dataLen); } void syncLogSendSyncSnapshotRsp(SSyncNode* pSyncNode, const SyncSnapshotRsp* pMsg, const char* s) { sNDebug(pSyncNode, - "send sync-snapshot-rsp to dnode:%d, %s, ack:%d, term:%" PRId64 ", begin:%" PRId64 ", lst:%" PRId64 - ", lterm:%" PRId64 ", stime:%" PRId64, + "send sync-snapshot-rsp to dnode:%d, %s, acked:%d, term:%" PRId64 ", begin-index:%" PRId64 + ", last-index:%" PRId64 ", last-term:%" PRId64 ", start-time:%" PRId64, DID(&pMsg->destId), s, pMsg->ack, pMsg->term, pMsg->snapBeginIndex, pMsg->lastIndex, pMsg->lastTerm, pMsg->startTime); } void syncLogRecvSyncSnapshotRsp(SSyncNode* pSyncNode, const SyncSnapshotRsp* pMsg, const char* s) { sNDebug(pSyncNode, - "recv sync-snapshot-rsp from dnode:%d, %s, ack:%d, term:%" PRId64 ", begin:%" PRId64 ", lst:%" PRId64 - ", lterm:%" PRId64 ", stime:%" PRId64, + "recv sync-snapshot-rsp from dnode:%d, %s, ack:%d, term:%" PRId64 ", begin-index:%" PRId64 + ", last-index:%" PRId64 ", last-term:%" PRId64 ", start-time:%" PRId64, DID(&pMsg->srcId), s, pMsg->ack, pMsg->term, pMsg->snapBeginIndex, pMsg->lastIndex, pMsg->lastTerm, pMsg->startTime); } void syncLogRecvAppendEntries(SSyncNode* pSyncNode, const SyncAppendEntries* pMsg, const char* s) { sNTrace(pSyncNode, - "recv sync-append-entries from dnode:%d {term:%" PRId64 ", pre-index:%" PRId64 ", pre-term:%" PRId64 - ", cmt:%" PRId64 ", pterm:%" PRId64 ", datalen:%d}, %s", - DID(&pMsg->srcId), pMsg->term, pMsg->prevLogIndex, pMsg->prevLogTerm, pMsg->commitIndex, pMsg->privateTerm, - pMsg->dataLen, s); + "recv sync-append-entries from dnode:%d {term:%" PRId64 ", prev-log:{index:%" PRId64 ", term:%" PRId64 + "}, commit-index:%" PRId64 ", datalen:%d}, %s", + DID(&pMsg->srcId), pMsg->term, pMsg->prevLogIndex, pMsg->prevLogTerm, pMsg->commitIndex, pMsg->dataLen, s); } void syncLogSendAppendEntries(SSyncNode* pSyncNode, const SyncAppendEntries* pMsg, const char* s) { sNTrace(pSyncNode, - "send sync-append-entries to dnode:%d, {term:%" PRId64 ", pre-index:%" PRId64 ", pre-term:%" PRId64 - ", lsend-index:%" PRId64 ", cmt:%" PRId64 ", datalen:%d}, %s", + "send sync-append-entries to dnode:%d, {term:%" PRId64 ", prev-log:{index:%" PRId64 ", term:%" PRId64 + "}, index:%" PRId64 ", commit-index:%" PRId64 ", datalen:%d}, %s", DID(&pMsg->destId), pMsg->term, pMsg->prevLogIndex, pMsg->prevLogTerm, (pMsg->prevLogIndex + 1), pMsg->commitIndex, pMsg->dataLen, s); } -void syncLogRecvRequestVote(SSyncNode* pSyncNode, const SyncRequestVote* pMsg, int32_t voteGranted, const char* s) { - if (voteGranted == -1) { - sNInfo(pSyncNode, - "recv sync-request-vote from dnode:%d, {term:%" PRId64 ", lindex:%" PRId64 ", lterm:%" PRId64 "}, %s", - DID(&pMsg->srcId), pMsg->term, pMsg->lastLogIndex, pMsg->lastLogTerm, s); - } else { - sNInfo(pSyncNode, - "recv sync-request-vote from dnode:%d, {term:%" PRId64 ", lindex:%" PRId64 ", lterm:%" PRId64 - "}, granted:%d", - DID(&pMsg->srcId), pMsg->term, pMsg->lastLogIndex, pMsg->lastLogTerm, voteGranted); - } +void syncLogRecvRequestVote(SSyncNode* pSyncNode, const SyncRequestVote* pMsg, int32_t voteGranted, + const char* errmsg) { + char statusMsg[64]; + snprintf(statusMsg, sizeof(statusMsg), "granted:%d", voteGranted); + sNInfo(pSyncNode, + "recv sync-request-vote from dnode:%d, {term:%" PRId64 ", last-index:%" PRId64 ", last-term:%" PRId64 "}, %s", + DID(&pMsg->srcId), pMsg->term, pMsg->lastLogIndex, pMsg->lastLogTerm, + (voteGranted != -1) ? statusMsg : errmsg); } void syncLogSendRequestVote(SSyncNode* pNode, const SyncRequestVote* pMsg, const char* s) { - sNInfo(pNode, "send sync-request-vote to dnode:%d {term:%" PRId64 ", lindex:%" PRId64 ", lterm:%" PRId64 "}, %s", + sNInfo(pNode, + "send sync-request-vote to dnode:%d {term:%" PRId64 ", last-index:%" PRId64 ", last-term:%" PRId64 "}, %s", DID(&pMsg->destId), pMsg->term, pMsg->lastLogIndex, pMsg->lastLogTerm, s); } From 3665ce26e7ca2ab3f6c38747f3ef0c407418ec43 Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Wed, 1 Feb 2023 21:32:34 +0800 Subject: [PATCH 062/124] fix: update taos-tools a0234fe for main (#19737) --- cmake/taostools_CMakeLists.txt.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/taostools_CMakeLists.txt.in b/cmake/taostools_CMakeLists.txt.in index 926d0c63e7..1053caf4ef 100644 --- a/cmake/taostools_CMakeLists.txt.in +++ b/cmake/taostools_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taos-tools ExternalProject_Add(taos-tools GIT_REPOSITORY https://github.com/taosdata/taos-tools.git - GIT_TAG 181bcac + GIT_TAG a0234fe SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools" BINARY_DIR "" #BUILD_IN_SOURCE TRUE From 9fe1512c553e0c9f33d858bc85ae26fa48a614e2 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Wed, 1 Feb 2023 22:41:39 +0800 Subject: [PATCH 063/124] refactor: add some logs. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 0bb9ebc40e..db52d7604f 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -1113,9 +1113,9 @@ static int32_t copyBlockDataToSDataBlock(STsdbReader* pReader, STableBlockScanIn int32_t unDumpedRows = asc ? pBlock->nRow - pDumpInfo->rowIndex : pDumpInfo->rowIndex + 1; tsdbDebug("%p copy file block to sdatablock, global index:%d, table index:%d, brange:%" PRId64 "-%" PRId64 - ", rows:%d, remain:%d, minVer:%" PRId64 ", maxVer:%" PRId64 ", elapsed time:%.2f ms, %s", + ", rows:%d, remain:%d, minVer:%" PRId64 ", maxVer:%" PRId64 ", uid:%"PRIu64" elapsed time:%.2f ms, %s", pReader, pBlockIter->index, pBlockInfo->tbBlockIdx, pBlock->minKey.ts, pBlock->maxKey.ts, dumpedRows, - unDumpedRows, pBlock->minVer, pBlock->maxVer, elapsedTime, pReader->idStr); + unDumpedRows, pBlock->minVer, pBlock->maxVer, pBlockInfo->uid, elapsedTime, pReader->idStr); return TSDB_CODE_SUCCESS; } @@ -2187,17 +2187,17 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea if (code == TSDB_CODE_SUCCESS) { pBlockScanInfo->iter.hasVal = (tsdbTbDataIterGet(pBlockScanInfo->iter.iter) != NULL); - tsdbDebug("%p uid:%" PRId64 ", check data in mem from skey:%" PRId64 ", order:%d, ts range in buf:%" PRId64 + tsdbDebug("%p uid:%" PRIu64 ", check data in mem from skey:%" PRId64 ", order:%d, ts range in buf:%" PRId64 "-%" PRId64 " %s", pReader, pBlockScanInfo->uid, startKey.ts, pReader->order, d->minKey, d->maxKey, pReader->idStr); } else { - tsdbError("%p uid:%" PRId64 ", failed to create iterator for imem, code:%s, %s", pReader, pBlockScanInfo->uid, + tsdbError("%p uid:%" PRIu64 ", failed to create iterator for imem, code:%s, %s", pReader, pBlockScanInfo->uid, tstrerror(code), pReader->idStr); return code; } } } else { - tsdbDebug("%p uid:%" PRId64 ", no data in mem, %s", pReader, pBlockScanInfo->uid, pReader->idStr); + tsdbDebug("%p uid:%" PRIu64 ", no data in mem, %s", pReader, pBlockScanInfo->uid, pReader->idStr); } STbData* di = NULL; @@ -2208,17 +2208,17 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea if (code == TSDB_CODE_SUCCESS) { pBlockScanInfo->iiter.hasVal = (tsdbTbDataIterGet(pBlockScanInfo->iiter.iter) != NULL); - tsdbDebug("%p uid:%" PRId64 ", check data in imem from skey:%" PRId64 ", order:%d, ts range in buf:%" PRId64 + tsdbDebug("%p uid:%" PRIu64 ", check data in imem from skey:%" PRId64 ", order:%d, ts range in buf:%" PRId64 "-%" PRId64 " %s", pReader, pBlockScanInfo->uid, startKey.ts, pReader->order, di->minKey, di->maxKey, pReader->idStr); } else { - tsdbError("%p uid:%" PRId64 ", failed to create iterator for mem, code:%s, %s", pReader, pBlockScanInfo->uid, + tsdbError("%p uid:%" PRIu64 ", failed to create iterator for mem, code:%s, %s", pReader, pBlockScanInfo->uid, tstrerror(code), pReader->idStr); return code; } } } else { - tsdbDebug("%p uid:%" PRId64 ", no data in imem, %s", pReader, pBlockScanInfo->uid, pReader->idStr); + tsdbDebug("%p uid:%" PRIu64 ", no data in imem, %s", pReader, pBlockScanInfo->uid, pReader->idStr); } initDelSkylineIterator(pBlockScanInfo, pReader, d, di); @@ -2837,6 +2837,8 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { SFileDataBlockInfo* pBlockInfo = getCurrentBlockInfo(pBlockIter); SLastBlockReader* pLastBlockReader = pReader->status.fileIter.pLastBlockReader; + ASSERT(pBlockInfo != NULL); + if (pBlockInfo != NULL) { pScanInfo = *(STableBlockScanInfo**)taosHashGet(pReader->status.pTableMap, &pBlockInfo->uid, sizeof(pBlockInfo->uid)); @@ -2857,7 +2859,7 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { initLastBlockReader(pLastBlockReader, pScanInfo, pReader); TSDBKEY keyInBuf = getCurrentKeyInBuf(pScanInfo, pReader); - if (pBlockInfo == NULL) { // build data block from last data file + /*if (pBlockInfo == NULL) { // build data block from last data file SBlockData* pBData = &pReader->status.fileBlockData; tBlockDataReset(pBData); @@ -2889,7 +2891,7 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { pReader, pResBlock->info.id.uid, pResBlock->info.window.skey, pResBlock->info.window.ekey, pResBlock->info.rows, el, pReader->idStr); } - } else if (fileBlockShouldLoad(pReader, pBlockInfo, pBlock, pScanInfo, keyInBuf, pLastBlockReader)) { + } else*/ if (fileBlockShouldLoad(pReader, pBlockInfo, pBlock, pScanInfo, keyInBuf, pLastBlockReader)) { code = doLoadFileBlockData(pReader, pBlockIter, &pStatus->fileBlockData, pScanInfo->uid); if (code != TSDB_CODE_SUCCESS) { return code; From 9a261ab2d74c9ef4231b73e939707d758250b497 Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Thu, 2 Feb 2023 08:45:38 +0800 Subject: [PATCH 064/124] improvement: optimize build table delete skyline performance --- source/dnode/vnode/src/tsdb/tsdbRead.c | 7 +- source/dnode/vnode/src/tsdb/tsdbUtil.c | 138 ++++++++++++++++++++++++- 2 files changed, 142 insertions(+), 3 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index e3240ec7d7..564ae46a26 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -81,6 +81,7 @@ typedef struct SIOCostSummary { double createScanInfoList; double getTbFromMemTime; double getTbFromIMemTime; + double initDelSkylineIterTime; } SIOCostSummary; typedef struct SBlockLoadSuppInfo { @@ -2227,7 +2228,9 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea tsdbDebug("%p uid:%" PRId64 ", no data in imem, %s", pReader, pBlockScanInfo->uid, pReader->idStr); } + st = taosGetTimestampUs(); initDelSkylineIterator(pBlockScanInfo, pReader, d, di); + pReader->cost.initDelSkylineIterTime += (taosGetTimestampUs() - st) / 1000.0; pBlockScanInfo->iterInit = true; return TSDB_CODE_SUCCESS; @@ -4013,12 +4016,12 @@ void tsdbReaderClose(STsdbReader* pReader) { "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms," - ", getTbFromMem-time:%.2f ms, getTbFromIMem-time:%.2f ms, %s", + ", getTbFromMem-time:%.2f ms, getTbFromIMem-time:%.2f ms, initDelSkylineIterTime:%.2f ms, %s", pReader, pCost->headFileLoad, pCost->headFileLoadTime, pCost->smaDataLoad, pCost->smaLoadTime, pCost->numOfBlocks, pCost->blockLoadTime, pCost->buildmemBlock, pCost->lastBlockLoad, pCost->lastBlockLoadTime, pCost->composedBlocks, pCost->buildComposedBlockTime, numOfTables * sizeof(STableBlockScanInfo) / 1000.0, pCost->createScanInfoList, - pCost->getTbFromMemTime, pCost->getTbFromIMemTime, pReader->idStr); + pCost->getTbFromMemTime, pCost->getTbFromIMemTime, pCost->initDelSkylineIterTime, pReader->idStr); taosMemoryFree(pReader->idStr); taosMemoryFree(pReader->pSchema); diff --git a/source/dnode/vnode/src/tsdb/tsdbUtil.c b/source/dnode/vnode/src/tsdb/tsdbUtil.c index e2d4b92836..a252c6deb6 100644 --- a/source/dnode/vnode/src/tsdb/tsdbUtil.c +++ b/source/dnode/vnode/src/tsdb/tsdbUtil.c @@ -929,8 +929,9 @@ int32_t tRowMergerGetRow(SRowMerger *pMerger, STSRow **ppRow) { return code; } +/* // delete skyline ====================================================== -static int32_t tsdbMergeSkyline(SArray *aSkyline1, SArray *aSkyline2, SArray *aSkyline) { +static int32_t tsdbMergeSkyline2(SArray *aSkyline1, SArray *aSkyline2, SArray *aSkyline) { int32_t code = 0; int32_t i1 = 0; int32_t n1 = taosArrayGetSize(aSkyline1); @@ -996,7 +997,141 @@ static int32_t tsdbMergeSkyline(SArray *aSkyline1, SArray *aSkyline2, SArray *aS _exit: return code; } +*/ + +// delete skyline ====================================================== +static int32_t tsdbMergeSkyline(SArray *pSkyline1, SArray *pSkyline2, SArray *pSkyline) { + int32_t code = 0; + int32_t i1 = 0; + int32_t n1 = taosArrayGetSize(pSkyline1); + int32_t i2 = 0; + int32_t n2 = taosArrayGetSize(pSkyline2); + TSDBKEY *pKey1; + TSDBKEY *pKey2; + int64_t version1 = 0; + int64_t version2 = 0; + + ASSERT(n1 > 0 && n2 > 0); + + taosArrayClear(pSkyline); + TSDBKEY **pItem = TARRAY_GET_ELEM(pSkyline, 0); + + while (i1 < n1 && i2 < n2) { + pKey1 = (TSDBKEY *)taosArrayGetP(pSkyline1, i1); + pKey2 = (TSDBKEY *)taosArrayGetP(pSkyline2, i2); + + if (pKey1->ts < pKey2->ts) { + version1 = pKey1->version; + *pItem = pKey1; + i1++; + } else if (pKey1->ts > pKey2->ts) { + version2 = pKey2->version; + *pItem = pKey2; + i2++; + } else { + version1 = pKey1->version; + version2 = pKey2->version; + *pItem = pKey1; + i1++; + i2++; + } + + (*pItem)->version = TMAX(version1, version2); + pItem++; + } + + while (i1 < n1) { + pKey1 = (TSDBKEY *)taosArrayGetP(pSkyline1, i1); + *pItem = pKey1; + pItem++; + i1++; + } + + while (i2 < n2) { + pKey2 = (TSDBKEY *)taosArrayGetP(pSkyline2, i2); + *pItem = pKey2; + pItem++; + i2++; + } + + taosArraySetSize(pSkyline, TARRAY_ELEM_IDX(pSkyline, pItem)); + +_exit: + return code; +} + + +int32_t tsdbBuildDeleteSkylineImpl(SArray *aSkyline, int32_t sidx, int32_t eidx, SArray *pSkyline) { + int32_t code = 0; + SDelData *pDelData; + int32_t midx; + + taosArrayClear(pSkyline); + if (sidx == eidx) { + TSDBKEY *pItem1 = taosArrayGet(aSkyline, sidx * 2); + TSDBKEY *pItem2 = taosArrayGet(aSkyline, sidx * 2 + 1); + taosArrayPush(pSkyline, &pItem1); + taosArrayPush(pSkyline, &pItem2); + } else { + SArray *pSkyline1 = NULL; + SArray *pSkyline2 = NULL; + midx = (sidx + eidx) / 2; + + pSkyline1 = taosArrayInit((midx - sidx + 1) * 2, POINTER_BYTES); + pSkyline2 = taosArrayInit((eidx - midx) * 2, POINTER_BYTES); + if (pSkyline1 == NULL || pSkyline1 == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _clear; + } + + code = tsdbBuildDeleteSkylineImpl(aSkyline, sidx, midx, pSkyline1); + if (code) goto _clear; + + code = tsdbBuildDeleteSkylineImpl(aSkyline, midx + 1, eidx, pSkyline2); + if (code) goto _clear; + + code = tsdbMergeSkyline(pSkyline1, pSkyline2, pSkyline); + + _clear: + taosArrayDestroy(pSkyline1); + taosArrayDestroy(pSkyline2); + } + + return code; +} + + int32_t tsdbBuildDeleteSkyline(SArray *aDelData, int32_t sidx, int32_t eidx, SArray *aSkyline) { + SDelData *pDelData; + int32_t code = 0; + int32_t dataNum = eidx - sidx + 1; + SArray *aTmpSkyline = taosArrayInit(dataNum * 2, sizeof(TSDBKEY)); + SArray *pSkyline = taosArrayInit(dataNum * 2, POINTER_BYTES); + + for (int32_t i = sidx; i <= eidx; ++i) { + pDelData = (SDelData *)taosArrayGet(aDelData, i); + taosArrayPush(aTmpSkyline, &(TSDBKEY){.ts = pDelData->sKey, .version = pDelData->version}); + taosArrayPush(aTmpSkyline, &(TSDBKEY){.ts = pDelData->eKey, .version = 0}); + } + + code = tsdbBuildDeleteSkylineImpl(aTmpSkyline, sidx, eidx, pSkyline); + if (code) goto _clear; + + int32_t skylineNum = taosArrayGetSize(pSkyline); + for (int32_t i = 0; i < skylineNum; ++i) { + TSDBKEY *p = taosArrayGetP(pSkyline, i); + taosArrayPush(aSkyline, p); + } + +_clear: + taosArrayDestroy(aTmpSkyline); + taosArrayDestroy(pSkyline); + + return code; +} + +/* +int32_t tsdbBuildDeleteSkyline2(SArray *aDelData, int32_t sidx, int32_t eidx, SArray *aSkyline) { int32_t code = 0; SDelData *pDelData; int32_t midx; @@ -1033,6 +1168,7 @@ int32_t tsdbBuildDeleteSkyline(SArray *aDelData, int32_t sidx, int32_t eidx, SAr return code; } +*/ // SBlockData ====================================================== int32_t tBlockDataCreate(SBlockData *pBlockData) { From 041d0d817a63b514f22637f4478d5eace116f15c Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Thu, 2 Feb 2023 09:11:17 +0800 Subject: [PATCH 065/124] fix: tq open --- source/dnode/vnode/src/tq/tq.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/source/dnode/vnode/src/tq/tq.c b/source/dnode/vnode/src/tq/tq.c index b195cfafb0..cca241a1cf 100644 --- a/source/dnode/vnode/src/tq/tq.c +++ b/source/dnode/vnode/src/tq/tq.c @@ -93,21 +93,21 @@ STQ* tqOpen(const char* path, SVnode* pVnode) { taosHashSetFreeFp(pTq->pCheckInfo, (FDelete)tDeleteSTqCheckInfo); if (tqMetaOpen(pTq) < 0) { - ASSERT(0); + return NULL; } pTq->pOffsetStore = tqOffsetOpen(pTq); if (pTq->pOffsetStore == NULL) { - ASSERT(0); + return NULL; } pTq->pStreamMeta = streamMetaOpen(path, pTq, (FTaskExpand*)tqExpandTask, pTq->pVnode->config.vgId); if (pTq->pStreamMeta == NULL) { - ASSERT(0); + return NULL; } if (streamLoadTasks(pTq->pStreamMeta) < 0) { - ASSERT(0); + return NULL; } return pTq; From 1e17f1d482d65daba29c3f5d017d604383578913 Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Thu, 2 Feb 2023 09:42:06 +0800 Subject: [PATCH 066/124] fix: mem leak --- source/client/src/clientTmq.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/source/client/src/clientTmq.c b/source/client/src/clientTmq.c index 90c405c204..22739108e2 100644 --- a/source/client/src/clientTmq.c +++ b/source/client/src/clientTmq.c @@ -1891,9 +1891,6 @@ int32_t tmq_consumer_close(tmq_t* tmq) { } tmq_list_destroy(lst); - - /*return rsp;*/ - return 0; } taosRemoveRef(tmqMgmt.rsetId, tmq->refId); return 0; From 69d6fb5bfe504f2a661530762395ec2f22f45a9b Mon Sep 17 00:00:00 2001 From: Minglei Jin Date: Thu, 2 Feb 2023 10:20:34 +0800 Subject: [PATCH 067/124] fix(meta): use metaAbort instead of tdbAbort with metaClose --- source/dnode/vnode/src/meta/metaCommit.c | 5 ++++- source/dnode/vnode/src/meta/metaOpen.c | 2 +- 2 files changed, 5 insertions(+), 2 deletions(-) diff --git a/source/dnode/vnode/src/meta/metaCommit.c b/source/dnode/vnode/src/meta/metaCommit.c index f61930b84c..f597c100d0 100644 --- a/source/dnode/vnode/src/meta/metaCommit.c +++ b/source/dnode/vnode/src/meta/metaCommit.c @@ -56,4 +56,7 @@ int metaPrepareAsyncCommit(SMeta *pMeta) { } // abort the meta txn -int metaAbort(SMeta *pMeta) { return tdbAbort(pMeta->pEnv, pMeta->txn); } +int metaAbort(SMeta *pMeta) { + if (!pMeta->txn) return 0; + return tdbAbort(pMeta->pEnv, pMeta->txn); +} diff --git a/source/dnode/vnode/src/meta/metaOpen.c b/source/dnode/vnode/src/meta/metaOpen.c index 867b481bcc..35677d6f07 100644 --- a/source/dnode/vnode/src/meta/metaOpen.c +++ b/source/dnode/vnode/src/meta/metaOpen.c @@ -203,7 +203,7 @@ _err: int metaClose(SMeta *pMeta) { if (pMeta) { - if (pMeta->pEnv) tdbAbort(pMeta->pEnv, pMeta->txn); + if (pMeta->pEnv) metaAbort(pMeta); if (pMeta->pCache) metaCacheClose(pMeta); if (pMeta->pIdx) metaCloseIdx(pMeta); if (pMeta->pStreamDb) tdbTbClose(pMeta->pStreamDb); From 7f571b1ebdb122d8ce9b0defc29e79076cf47c7f Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Thu, 2 Feb 2023 10:59:47 +0800 Subject: [PATCH 068/124] fix: a time point can be fill --- source/libs/parser/src/parTranslater.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c index 9991c2c6ae..183d1b0c6e 100644 --- a/source/libs/parser/src/parTranslater.c +++ b/source/libs/parser/src/parTranslater.c @@ -2978,7 +2978,7 @@ static int32_t checkFill(STranslateContext* pCxt, SFillNode* pFill, SValueNode* intervalRange = pInterval->datum.i; } - if ((timeRange == 0) || (timeRange / intervalRange) >= MAX_INTERVAL_TIME_WINDOW) { + if ((timeRange / intervalRange) >= MAX_INTERVAL_TIME_WINDOW) { return generateSyntaxErrMsg(&pCxt->msgBuf, TSDB_CODE_PAR_INVALID_FILL_TIME_RANGE); } From e234dda2de599b0f39ab9f272e46131e0727cb61 Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Thu, 2 Feb 2023 11:44:08 +0800 Subject: [PATCH 069/124] fix: interp support filter and scalar calc --- source/libs/executor/src/timesliceoperator.c | 12 ++++++++ tests/script/tsim/parser/interp.sim | 29 ++++++++++++++++++++ 2 files changed, 41 insertions(+) diff --git a/source/libs/executor/src/timesliceoperator.c b/source/libs/executor/src/timesliceoperator.c index 94c98f41c9..6561e810bb 100644 --- a/source/libs/executor/src/timesliceoperator.c +++ b/source/libs/executor/src/timesliceoperator.c @@ -433,6 +433,11 @@ static SSDataBlock* doTimeslice(SOperatorInfo* pOperator) { break; } + if (pSliceInfo->scalarSup.pExprInfo != NULL) { + SExprSupp* pExprSup = &pSliceInfo->scalarSup; + projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx, pExprSup->numOfExprs, NULL); + } + int32_t code = initKeeperInfo(pSliceInfo, pBlock); if (code != TSDB_CODE_SUCCESS) { T_LONG_JMP(pTaskInfo->env, code); @@ -531,6 +536,8 @@ static SSDataBlock* doTimeslice(SOperatorInfo* pOperator) { taosTimeAdd(pSliceInfo->current, pInterval->interval, pInterval->intervalUnit, pInterval->precision); } + doFilter(pResBlock, pOperator->exprSupp.pFilterInfo, NULL); + // restore the value setTaskStatus(pOperator->pTaskInfo, TASK_COMPLETED); if (pResBlock->info.rows == 0) { @@ -566,6 +573,11 @@ SOperatorInfo* createTimeSliceOperatorInfo(SOperatorInfo* downstream, SPhysiNode } } + code = filterInitFromNode((SNode*)pInterpPhyNode->node.pConditions, &pOperator->exprSupp.pFilterInfo, 0); + if (code != TSDB_CODE_SUCCESS) { + goto _error; + } + pInfo->tsCol = extractColumnFromColumnNode((SColumnNode*)pInterpPhyNode->pTimeSeries); pInfo->fillType = convertFillType(pInterpPhyNode->fillMode); initResultSizeInfo(&pOperator->resultInfo, 4096); diff --git a/tests/script/tsim/parser/interp.sim b/tests/script/tsim/parser/interp.sim index 1b7878178c..e6512a22d7 100644 --- a/tests/script/tsim/parser/interp.sim +++ b/tests/script/tsim/parser/interp.sim @@ -72,4 +72,33 @@ sql_error select interp(*) from nt5931 where ts=now sql_error select interp(*) from st5931 where ts=now sql_error select interp(*) from ct5931 where ts=now +sql create stable sta (ts timestamp, f1 double, f2 binary(200)) tags(t1 int); +sql create table tba1 using sta tags(1); +sql insert into tba1 values ('2022-04-26 15:15:01', -3.0, "a"); +sql insert into tba1 values ('2022-04-26 15:15:05', 3.0, "b"); +sql select a from (select interp(f1) as a from tba1 where ts >= '2022-04-26 15:15:01' and ts <= '2022-04-26 15:15:05' range('2022-04-26 15:15:01','2022-04-26 15:15:05') every(1s) fill(linear)) where a > 0; +if $rows != 2 then + return -1 +endi +if $data00 != 1.500000000 then + return -1 +endi +if $data10 != 3.000000000 then + return -1 +endi + +sql select a from (select interp(f1+1) as a from tba1 where ts >= '2022-04-26 15:15:01' and ts <= '2022-04-26 15:15:05' range('2022-04-26 15:15:01','2022-04-26 15:15:05') every(1s) fill(linear)) where a > 0; +if $rows != 3 then + return -1 +endi +if $data00 != 1.000000000 then + return -1 +endi +if $data10 != 2.500000000 then + return -1 +endi +if $data20 != 4.000000000 then + return -1 +endi + system sh/exec.sh -n dnode1 -s stop -x SIGINT From 3f577c5019a8ea8ccd1c74f2dc95edc9863f7a6b Mon Sep 17 00:00:00 2001 From: shenglian zhou Date: Thu, 2 Feb 2023 11:57:22 +0800 Subject: [PATCH 070/124] fix: show table distributed only works on super table, child table, and normal table --- include/libs/function/functionMgt.h | 1 + source/libs/function/src/functionMgt.c | 7 ++++++ source/libs/parser/src/parTranslater.c | 28 ++++++++++++++++++++++-- tests/script/tsim/compute/block_dist.sim | 4 ++++ 4 files changed, 38 insertions(+), 2 deletions(-) diff --git a/include/libs/function/functionMgt.h b/include/libs/function/functionMgt.h index 9ca6a7a9fa..411c7d5e8b 100644 --- a/include/libs/function/functionMgt.h +++ b/include/libs/function/functionMgt.h @@ -218,6 +218,7 @@ bool fmIsKeepOrderFunc(int32_t funcId); bool fmIsCumulativeFunc(int32_t funcId); bool fmIsInterpPseudoColumnFunc(int32_t funcId); bool fmIsGroupKeyFunc(int32_t funcId); +bool fmIsBlockDistFunc(int32_t funcId); void getLastCacheDataType(SDataType* pType); diff --git a/source/libs/function/src/functionMgt.c b/source/libs/function/src/functionMgt.c index 6ab91a4483..e3127fcd7b 100644 --- a/source/libs/function/src/functionMgt.c +++ b/source/libs/function/src/functionMgt.c @@ -262,6 +262,13 @@ bool fmIsGroupKeyFunc(int32_t funcId) { return FUNCTION_TYPE_GROUP_KEY == funcMgtBuiltins[funcId].type; } +bool fmIsBlockDistFunc(int32_t funcId) { + if (funcId < 0 || funcId >= funcMgtBuiltinsNum) { + return false; + } + return FUNCTION_TYPE_BLOCK_DIST == funcMgtBuiltins[funcId].type; +} + void fmFuncMgtDestroy() { void* m = gFunMgtService.pFuncNameHashTable; if (m != NULL && atomic_val_compare_exchange_ptr((void**)&gFunMgtService.pFuncNameHashTable, m, 0) == m) { diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c index 05d49bb027..b65d7e98d5 100644 --- a/source/libs/parser/src/parTranslater.c +++ b/source/libs/parser/src/parTranslater.c @@ -1558,6 +1558,26 @@ static int32_t translateRepeatScanFunc(STranslateContext* pCxt, SFunctionNode* p return TSDB_CODE_SUCCESS; } +static int32_t translateBlockDistFunc(STranslateContext* pCtx, SFunctionNode* pFunc) { + if (!fmIsBlockDistFunc(pFunc->funcId)) { + return TSDB_CODE_SUCCESS; + } + if (!isSelectStmt(pCtx->pCurrStmt)) { + return generateSyntaxErrMsgExt(&pCtx->msgBuf, TSDB_CODE_PAR_ONLY_SUPPORT_SINGLE_TABLE, + "%s is only supported in single table query", pFunc->functionName); + } + SSelectStmt* pSelect = (SSelectStmt*)pCtx->pCurrStmt; + SNode* pTable = pSelect->pFromTable; + if (NULL != pTable && (QUERY_NODE_REAL_TABLE != nodeType(pTable) || + (TSDB_SUPER_TABLE != ((SRealTableNode*)pTable)->pMeta->tableType && + TSDB_CHILD_TABLE != ((SRealTableNode*)pTable)->pMeta->tableType && + TSDB_NORMAL_TABLE != ((SRealTableNode*)pTable)->pMeta->tableType))) { + return generateSyntaxErrMsgExt(&pCtx->msgBuf, TSDB_CODE_PAR_NOT_ALLOWED_FUNC, + "%s is only supported on super table, child table or normal table", pFunc->functionName); + } + return TSDB_CODE_SUCCESS; +} + static bool isStar(SNode* pNode) { return (QUERY_NODE_COLUMN == nodeType(pNode)) && ('\0' == ((SColumnNode*)pNode)->tableAlias[0]) && (0 == strcmp(((SColumnNode*)pNode)->colName, "*")); @@ -1717,7 +1737,7 @@ static int32_t rewriteSystemInfoFunc(STranslateContext* pCxt, SNode** pNode) { return TSDB_CODE_PAR_INTERNAL_ERROR; } -static int32_t translateNoramlFunction(STranslateContext* pCxt, SFunctionNode* pFunc) { +static int32_t translateNormalFunction(STranslateContext* pCxt, SFunctionNode* pFunc) { int32_t code = translateAggFunc(pCxt, pFunc); if (TSDB_CODE_SUCCESS == code) { code = translateScanPseudoColumnFunc(pCxt, pFunc); @@ -1749,6 +1769,9 @@ static int32_t translateNoramlFunction(STranslateContext* pCxt, SFunctionNode* p if (TSDB_CODE_SUCCESS == code) { code = translateTimelineFunc(pCxt, pFunc); } + if (TSDB_CODE_SUCCESS == code) { + code = translateBlockDistFunc(pCxt, pFunc); + } if (TSDB_CODE_SUCCESS == code) { setFuncClassification(pCxt->pCurrStmt, pFunc); } @@ -1809,7 +1832,7 @@ static int32_t translateFunctionImpl(STranslateContext* pCxt, SFunctionNode** pF if (fmIsClientPseudoColumnFunc((*pFunc)->funcId)) { return rewriteClientPseudoColumnFunc(pCxt, (SNode**)pFunc); } - return translateNoramlFunction(pCxt, *pFunc); + return translateNormalFunction(pCxt, *pFunc); } static EDealRes translateFunction(STranslateContext* pCxt, SFunctionNode** pFunc) { @@ -6307,6 +6330,7 @@ static int32_t createSelectStmtForShow(ENodeType showType, SSelectStmt** pStmt) } static int32_t createSelectStmtForShowTableDist(SShowTableDistributedStmt* pStmt, SSelectStmt** pOutput) { + return createSimpleSelectStmt(pStmt->dbName, pStmt->tableName, 0, NULL, pOutput); } diff --git a/tests/script/tsim/compute/block_dist.sim b/tests/script/tsim/compute/block_dist.sim index 2d0a4e8902..4fdcf63e34 100644 --- a/tests/script/tsim/compute/block_dist.sim +++ b/tests/script/tsim/compute/block_dist.sim @@ -91,6 +91,10 @@ print ============== TD-5998 sql_error select _block_dist() from (select * from $nt) sql_error select _block_dist() from (select * from $mt) +print ============== TD-22140 & TD-22165 +sql_error show table distributed information_schema.ins_databases +sql_error show table distributed performance_schema.perf_apps + print =============== clear sql drop database $db sql select * from information_schema.ins_databases From b9ee0f983f6f7b849d14bc8db3d023ae77fe343d Mon Sep 17 00:00:00 2001 From: chenhaoran Date: Thu, 2 Feb 2023 15:32:02 +0800 Subject: [PATCH 071/124] ci:add ci docker file to TD-rep --- tests/ci/Dockerfile | 48 ++++++++++++++++++++++++++++++ tests/ci/build_image.sh | 4 +++ tests/ci/daily_build_image.sh | 56 +++++++++++++++++++++++++++++++++++ 3 files changed, 108 insertions(+) create mode 100644 tests/ci/Dockerfile create mode 100755 tests/ci/build_image.sh create mode 100755 tests/ci/daily_build_image.sh diff --git a/tests/ci/Dockerfile b/tests/ci/Dockerfile new file mode 100644 index 0000000000..594bcc902d --- /dev/null +++ b/tests/ci/Dockerfile @@ -0,0 +1,48 @@ +FROM python:3.8 +RUN pip3 config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple +RUN pip3 install pandas psutil fabric2 requests faker simplejson toml pexpect tzlocal distro +RUN apt-get update +RUN apt-get install -y psmisc sudo tree libjansson-dev libsnappy-dev liblzma-dev libz-dev zlib1g pkg-config build-essential valgrind \ + vim libjemalloc-dev openssh-server screen sshpass net-tools dirmngr gnupg apt-transport-https ca-certificates software-properties-common r-base iputils-ping +RUN apt-key adv --keyserver keyserver.ubuntu.com --recv-keys E298A3A825C0D65DFD57CBB651716619E084DAB9 +RUN add-apt-repository 'deb https://cloud.r-project.org/bin/linux/ubuntu focal-cran40/' +RUN apt install -y r-base +ADD go1.17.6.linux-amd64.tar.gz /usr/local/ +ADD jdk-8u144-linux-x64.tar.gz /usr/local/ +ADD apache-maven-3.8.4-bin.tar.gz /usr/local/ +RUN apt-get install wget -y \ + && wget https://packages.microsoft.com/config/ubuntu/18.04/packages-microsoft-prod.deb -O packages-microsoft-prod.deb \ + && dpkg -i packages-microsoft-prod.deb \ + && rm packages-microsoft-prod.deb \ + && apt-get update && apt-get install -y dotnet-sdk-5.0 && apt-get install -y dotnet-sdk-6.0 +ADD node-v12.20.0-linux-x64.tar.gz /usr/local/ +RUN sh -c "rm -f /etc/localtime;ln -s /usr/share/zoneinfo/Asia/Shanghai /etc/localtime;echo \"Asia/Shanghai\" >/etc/timezone" +COPY id_rsa /root/.ssh/id_rsa +COPY .m2 /root/.m2 +COPY .nuget /root/.nuget +COPY .dotnet /root/.dotnet +COPY .cargo /root/.cargo +COPY go /root/go +ADD cmake-3.21.5-linux-x86_64.tar.gz /usr/local/ +RUN echo " export RUSTUP_DIST_SERVER=\"https://rsproxy.cn\" " >> /root/.bashrc +RUN echo " export RUSTUP_UPDATE_ROOT=\"https://rsproxy.cn/rustup\" " >> /root/.bashrc +RUN curl https://sh.rustup.rs -o /tmp/rustup-init.sh +RUN sh /tmp/rustup-init.sh -y +ENV PATH /usr/local/go/bin:/usr/local/node-v12.20.0-linux-x64/bin:/usr/local/apache-maven-3.8.4/bin:/usr/local/jdk1.8.0_144/bin:/usr/local/cmake-3.21.5-linux-x86_64/bin:/root/.cargo/bin:$PATH +ENV JAVA_HOME /usr/local/jdk1.8.0_144 +RUN go env -w GOPROXY=https://goproxy.cn +RUN echo "StrictHostKeyChecking no" >>/etc/ssh/ssh_config +RUN npm config -g set unsafe-perm +RUN npm config -g set registry https://registry.npm.taobao.org +COPY .npm /root/.npm +RUN R CMD javareconf JAVA_HOME=${JAVA_HOME} JAVA=${JAVA_HOME}/bin/java JAVAC=${JAVA_HOME}/bin/javac JAVAH=${JAVA_HOME}/bin/javah JAR=${JAVA_HOME}/bin/jar +RUN echo "install.packages(\"RJDBC\", repos=\"http://cran.us.r-project.org\")"|R --no-save +COPY .gitconfig /root/.gitconfig +RUN mkdir -p /run/sshd +COPY id_rsa.pub /root/.ssh/id_rsa.pub +COPY id_rsa.pub /root/.ssh/authorized_keys +RUN pip3 uninstall -y taostest +COPY repository/TDinternal /home/TDinternal +COPY repository/taos-connector-python /home/taos-connector-python +RUN sh -c "cd /home/taos-connector-python; pip3 install ." +COPY setup.sh /home/setup.sh \ No newline at end of file diff --git a/tests/ci/build_image.sh b/tests/ci/build_image.sh new file mode 100755 index 0000000000..1864df35db --- /dev/null +++ b/tests/ci/build_image.sh @@ -0,0 +1,4 @@ +#!/bin/bash + +docker build --no-cache -t taos_test:v1.0 . + diff --git a/tests/ci/daily_build_image.sh b/tests/ci/daily_build_image.sh new file mode 100755 index 0000000000..47f0c971d9 --- /dev/null +++ b/tests/ci/daily_build_image.sh @@ -0,0 +1,56 @@ +#!/bin/bash + +set -x + +script_dir=`dirname $0` +cd $script_dir +script_dir=`pwd` +cd $script_dir/repository/taos-connector-python +git pull + +cd $script_dir/repository/TDinternal +git clean -fxd +git pull + +cd $script_dir/repository/TDinternal/community +git clean -fxd +git pull +git submodule update --init --recursive + +cd $script_dir +./build_image.sh || exit 1 +docker image prune -f +ips="\ +192.168.1.47 \ +192.168.1.48 \ +192.168.1.49 \ +192.168.1.52 \ +192.168.0.215 \ +192.168.0.217 \ +192.168.0.219 \ +" + +image=taos_image.tar + +docker save taos_test:v1.0 -o $image + +for ip in $ips; do + echo "scp $image root@$ip:/home/ &" + scp $image root@$ip:/home/ & +done +wait + +for ip in $ips; do + echo "ssh root@$ip docker load -i /home/$image &" + ssh root@$ip docker load -i /home/$image & +done +wait + +for ip in $ips; do + echo "ssh root@$ip rm -f /home/$image &" + ssh root@$ip rm -f /home/$image & +done +wait + +rm -rf taos_image.tar + From 57a0c8abd20ef2291388221c5d6876dd2469e5a1 Mon Sep 17 00:00:00 2001 From: chenhaoran Date: Thu, 2 Feb 2023 15:52:12 +0800 Subject: [PATCH 072/124] ci:add ci docker file to TD-rep --- tests/ci/daily_build_image.sh | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/tests/ci/daily_build_image.sh b/tests/ci/daily_build_image.sh index 47f0c971d9..01148a3aae 100755 --- a/tests/ci/daily_build_image.sh +++ b/tests/ci/daily_build_image.sh @@ -14,10 +14,14 @@ git pull cd $script_dir/repository/TDinternal/community git clean -fxd -git pull +git checkout main +git pull origin main git submodule update --init --recursive cd $script_dir +cp $script_dir/repository/TDinternal/community/tests/ci/build_image.sh . +cp $script_dir/repository/TDinternal/community/tests/ci/daily_build_image.sh . + ./build_image.sh || exit 1 docker image prune -f ips="\ From a32c8520399e1b5f8694f61d03e77b5e0562d298 Mon Sep 17 00:00:00 2001 From: wangmm0220 Date: Thu, 2 Feb 2023 16:13:17 +0800 Subject: [PATCH 073/124] fix:add ttl drop response for mnode --- source/dnode/mgmt/mgmt_mnode/src/mmHandle.c | 1 + source/dnode/mnode/impl/src/mndStb.c | 6 ++++++ 2 files changed, 7 insertions(+) diff --git a/source/dnode/mgmt/mgmt_mnode/src/mmHandle.c b/source/dnode/mgmt/mgmt_mnode/src/mmHandle.c index 16fe6c1b91..cf4eaaf7d1 100644 --- a/source/dnode/mgmt/mgmt_mnode/src/mmHandle.c +++ b/source/dnode/mgmt/mgmt_mnode/src/mmHandle.c @@ -167,6 +167,7 @@ SArray *mmGetMsgHandles() { if (dmSetMgmtHandle(pArray, TDMT_VND_CREATE_STB_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; if (dmSetMgmtHandle(pArray, TDMT_VND_ALTER_STB_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; if (dmSetMgmtHandle(pArray, TDMT_VND_DROP_STB_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; + if (dmSetMgmtHandle(pArray, TDMT_VND_DROP_TTL_TABLE_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; if (dmSetMgmtHandle(pArray, TDMT_VND_CREATE_SMA_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; if (dmSetMgmtHandle(pArray, TDMT_VND_DROP_SMA_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; if (dmSetMgmtHandle(pArray, TDMT_VND_TMQ_SUBSCRIBE_RSP, mmPutMsgToWriteQueue, 0) == NULL) goto _OVER; diff --git a/source/dnode/mnode/impl/src/mndStb.c b/source/dnode/mnode/impl/src/mndStb.c index d504a94700..471c53b2f5 100644 --- a/source/dnode/mnode/impl/src/mndStb.c +++ b/source/dnode/mnode/impl/src/mndStb.c @@ -41,6 +41,7 @@ static int32_t mndProcessTtlTimer(SRpcMsg *pReq); static int32_t mndProcessCreateStbReq(SRpcMsg *pReq); static int32_t mndProcessAlterStbReq(SRpcMsg *pReq); static int32_t mndProcessDropStbReq(SRpcMsg *pReq); +static int32_t mndProcessDropTtltbReq(SRpcMsg *pReq); static int32_t mndProcessTableMetaReq(SRpcMsg *pReq); static int32_t mndRetrieveStb(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows); static void mndCancelGetNextStb(SMnode *pMnode, void *pIter); @@ -64,6 +65,7 @@ int32_t mndInitStb(SMnode *pMnode) { mndSetMsgHandle(pMnode, TDMT_MND_ALTER_STB, mndProcessAlterStbReq); mndSetMsgHandle(pMnode, TDMT_MND_DROP_STB, mndProcessDropStbReq); mndSetMsgHandle(pMnode, TDMT_VND_CREATE_STB_RSP, mndTransProcessRsp); + mndSetMsgHandle(pMnode, TDMT_VND_DROP_TTL_TABLE_RSP, mndProcessDropTtltbReq); mndSetMsgHandle(pMnode, TDMT_VND_ALTER_STB_RSP, mndTransProcessRsp); mndSetMsgHandle(pMnode, TDMT_VND_DROP_STB_RSP, mndTransProcessRsp); mndSetMsgHandle(pMnode, TDMT_MND_TABLE_META, mndProcessTableMetaReq); @@ -2176,6 +2178,10 @@ static int32_t mndCheckDropStbForStream(SMnode *pMnode, const char *stbFullName, return 0; } +static int32_t mndProcessDropTtltbReq(SRpcMsg *pRsp) { + return 0; +} + static int32_t mndProcessDropStbReq(SRpcMsg *pReq) { SMnode *pMnode = pReq->info.node; int32_t code = -1; From c4366e4dbcc28d802969f7b8e44740f3c84c1b3a Mon Sep 17 00:00:00 2001 From: shenglian zhou Date: Thu, 2 Feb 2023 16:34:20 +0800 Subject: [PATCH 074/124] fix: minior, remove blank line --- source/libs/parser/src/parTranslater.c | 1 - 1 file changed, 1 deletion(-) diff --git a/source/libs/parser/src/parTranslater.c b/source/libs/parser/src/parTranslater.c index b65d7e98d5..172b2b44ab 100644 --- a/source/libs/parser/src/parTranslater.c +++ b/source/libs/parser/src/parTranslater.c @@ -6330,7 +6330,6 @@ static int32_t createSelectStmtForShow(ENodeType showType, SSelectStmt** pStmt) } static int32_t createSelectStmtForShowTableDist(SShowTableDistributedStmt* pStmt, SSelectStmt** pOutput) { - return createSimpleSelectStmt(pStmt->dbName, pStmt->tableName, 0, NULL, pOutput); } From aab31f655c14a06df56be761a397fb51b566b240 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Thu, 2 Feb 2023 16:57:05 +0800 Subject: [PATCH 075/124] fix(query): fix bug in multi-group limit/offset of the merge sort . --- source/libs/executor/src/sortoperator.c | 32 +++++++++++++------------ 1 file changed, 17 insertions(+), 15 deletions(-) diff --git a/source/libs/executor/src/sortoperator.c b/source/libs/executor/src/sortoperator.c index 6d3da3e111..98ef6b8a36 100644 --- a/source/libs/executor/src/sortoperator.c +++ b/source/libs/executor/src/sortoperator.c @@ -544,7 +544,6 @@ typedef struct SMultiwayMergeOperatorInfo { SSDataBlock* pIntermediateBlock; // to hold the intermediate result int64_t startTs; // sort start time bool groupSort; - bool hasGroupId; uint64_t groupId; STupleHandle* prefetchedTuple; } SMultiwayMergeOperatorInfo; @@ -591,7 +590,9 @@ int32_t openMultiwayMergeOperator(SOperatorInfo* pOperator) { return TSDB_CODE_SUCCESS; } -static void doGetSortedBlockData(SMultiwayMergeOperatorInfo* pInfo, SSortHandle* pHandle, int32_t capacity, SSDataBlock* p) { +static void doGetSortedBlockData(SMultiwayMergeOperatorInfo* pInfo, SSortHandle* pHandle, int32_t capacity, + SSDataBlock* p, bool* newgroup) { + *newgroup = false; while (1) { STupleHandle* pTupleHandle = NULL; @@ -600,8 +601,12 @@ static void doGetSortedBlockData(SMultiwayMergeOperatorInfo* pInfo, SSortHandle* pTupleHandle = tsortNextTuple(pHandle); } else { pTupleHandle = pInfo->prefetchedTuple; - pInfo->groupId = tsortGetGroupId(pTupleHandle); pInfo->prefetchedTuple = NULL; + uint64_t gid = tsortGetGroupId(pTupleHandle); + if (gid != pInfo->groupId) { + *newgroup = true; + pInfo->groupId = gid; + } } } else { pTupleHandle = tsortNextTuple(pHandle); @@ -614,12 +619,10 @@ static void doGetSortedBlockData(SMultiwayMergeOperatorInfo* pInfo, SSortHandle* if (pInfo->groupSort) { uint64_t tupleGroupId = tsortGetGroupId(pTupleHandle); - if (!pInfo->hasGroupId) { + if (pInfo->groupId == 0 || pInfo->groupId == tupleGroupId) { + appendOneRowToDataBlock(p, pTupleHandle); + p->info.id.groupId = tupleGroupId; pInfo->groupId = tupleGroupId; - pInfo->hasGroupId = true; - appendOneRowToDataBlock(p, pTupleHandle); - } else if (pInfo->groupId == tupleGroupId) { - appendOneRowToDataBlock(p, pTupleHandle); } else { pInfo->prefetchedTuple = pTupleHandle; break; @@ -632,11 +635,6 @@ static void doGetSortedBlockData(SMultiwayMergeOperatorInfo* pInfo, SSortHandle* break; } } - - if (pInfo->groupSort) { - pInfo->hasGroupId = false; - } - } SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pDataBlock, SArray* pColMatchInfo, @@ -660,14 +658,18 @@ SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pData } SSDataBlock* p = pInfo->pIntermediateBlock; + bool newgroup = false; while (1) { - doGetSortedBlockData(pInfo, pHandle, capacity, p); + doGetSortedBlockData(pInfo, pHandle, capacity, p, &newgroup); if (p->info.rows == 0) { break; } - // todo fix it: we need to decide whether this block is belonged to previous group or not . + if (newgroup) { + resetLimitInfoForNextGroup(&pInfo->limitInfo); + } + bool limitReached = applyLimitOffset(&pInfo->limitInfo, p, pTaskInfo); if (limitReached) { resetLimitInfoForNextGroup(&pInfo->limitInfo); From 615fe4869e1ccac80a58dff18660ee752c1bc69f Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Thu, 2 Feb 2023 17:46:58 +0800 Subject: [PATCH 076/124] fix: fix not msg to send --- source/libs/transport/src/transCli.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index 1a99db5f99..be687b6193 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -800,9 +800,12 @@ static void cliSendCb(uv_write_t* req, int status) { } void cliSend(SCliConn* pConn) { - bool empty = transQueueEmpty(&pConn->cliMsgs); - ASSERTS(empty == false, "trans-cli get invalid msg"); - if (empty == true) { + SCliThrd* pThrd = pConn->hostThrd; + STrans* pTransInst = pThrd->pTransInst; + + if (transQueueEmpty(&pConn->cliMsgs)) { + tError("%s conn %p not msg to send", pTransInst->label, pConn); + cliHandleExecept(pConn); return; } @@ -812,9 +815,6 @@ void cliSend(SCliConn* pConn) { STransConnCtx* pCtx = pCliMsg->ctx; - SCliThrd* pThrd = pConn->hostThrd; - STrans* pTransInst = pThrd->pTransInst; - STransMsg* pMsg = (STransMsg*)(&pCliMsg->msg); if (pMsg->pCont == 0) { pMsg->pCont = (void*)rpcMallocCont(0); From 6cdfa6d60f3fd46989ff084b2793d229dd0254ec Mon Sep 17 00:00:00 2001 From: Minglei Jin Date: Thu, 2 Feb 2023 17:51:04 +0800 Subject: [PATCH 077/124] fix(tdb): return success if txn is commited --- source/libs/tdb/src/db/tdbPager.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/source/libs/tdb/src/db/tdbPager.c b/source/libs/tdb/src/db/tdbPager.c index 62d82edeb1..63e88b0a12 100644 --- a/source/libs/tdb/src/db/tdbPager.c +++ b/source/libs/tdb/src/db/tdbPager.c @@ -457,6 +457,11 @@ int tdbPagerAbort(SPager *pPager, TXN *pTxn) { SPgno journalSize = 0; int ret; + if (pTxn->jfd == 0) { + // txn is commited + return 0; + } + // sync the journal file ret = tdbOsFSync(pTxn->jfd); if (ret < 0) { From 746a6bc243fd110ddd81e1c108244d224e33740b Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Thu, 2 Feb 2023 18:24:25 +0800 Subject: [PATCH 078/124] fix: taosbenchmark handle mem better patch2 for main (#19743) * fix: update taos-tools a0234fe for main * fix: update taos-tools c4a567b --- cmake/taostools_CMakeLists.txt.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/taostools_CMakeLists.txt.in b/cmake/taostools_CMakeLists.txt.in index 1053caf4ef..4da2d80fbb 100644 --- a/cmake/taostools_CMakeLists.txt.in +++ b/cmake/taostools_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taos-tools ExternalProject_Add(taos-tools GIT_REPOSITORY https://github.com/taosdata/taos-tools.git - GIT_TAG a0234fe + GIT_TAG c4a567b SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools" BINARY_DIR "" #BUILD_IN_SOURCE TRUE From b6b0b3439e8c519a2210842fdc52089bb5f135be Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Thu, 2 Feb 2023 19:34:18 +0800 Subject: [PATCH 079/124] enh: rename syncNodeOnSnapshotPre to syncNodeOnSnapshotPrep --- source/libs/sync/inc/syncSnapshot.h | 2 +- source/libs/sync/src/syncSnapshot.c | 26 +++++++++++++------------- 2 files changed, 14 insertions(+), 14 deletions(-) diff --git a/source/libs/sync/inc/syncSnapshot.h b/source/libs/sync/inc/syncSnapshot.h index 5277e7818f..063b4f51f5 100644 --- a/source/libs/sync/inc/syncSnapshot.h +++ b/source/libs/sync/inc/syncSnapshot.h @@ -24,7 +24,7 @@ extern "C" { #define SYNC_SNAPSHOT_SEQ_INVALID -2 #define SYNC_SNAPSHOT_SEQ_FORCE_CLOSE -3 -#define SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT -1 +#define SYNC_SNAPSHOT_SEQ_PREP_SNAPSHOT -1 #define SYNC_SNAPSHOT_SEQ_BEGIN 0 #define SYNC_SNAPSHOT_SEQ_END 0x7FFFFFFF diff --git a/source/libs/sync/src/syncSnapshot.c b/source/libs/sync/src/syncSnapshot.c index 880c76e4dd..e61bcc9ffc 100644 --- a/source/libs/sync/src/syncSnapshot.c +++ b/source/libs/sync/src/syncSnapshot.c @@ -112,7 +112,7 @@ int32_t snapshotSenderStart(SSyncSnapshotSender *pSender) { pMsg->lastConfigIndex = pSender->snapshot.lastConfigIndex; pMsg->lastConfig = pSender->lastConfig; pMsg->startTime = pSender->startTime; - pMsg->seq = SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT; + pMsg->seq = SYNC_SNAPSHOT_SEQ_PREP_SNAPSHOT; // event log syncLogSendSyncSnapshotSend(pSender->pSyncNode, pMsg, "snapshot sender start"); @@ -379,7 +379,7 @@ void snapshotReceiverStart(SSyncSnapshotReceiver *pReceiver, SyncSnapshotSend *p } pReceiver->start = true; - pReceiver->ack = SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT; + pReceiver->ack = SYNC_SNAPSHOT_SEQ_PREP_SNAPSHOT; pReceiver->term = pReceiver->pSyncNode->raftStore.currentTerm; pReceiver->fromId = pPreMsg->srcId; pReceiver->startTime = pPreMsg->startTime; @@ -527,7 +527,7 @@ SyncIndex syncNodeGetSnapBeginIndex(SSyncNode *ths) { return snapStart; } -static int32_t syncNodeOnSnapshotPre(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { +static int32_t syncNodeOnSnapshotPrep(SSyncNode *pSyncNode, SyncSnapshotSend *pMsg) { SSyncSnapshotReceiver *pReceiver = pSyncNode->pNewNodeReceiver; int64_t timeNow = taosGetTimestampMs(); int32_t code = 0; @@ -565,7 +565,7 @@ _START_RECEIVER: } else { // waiting for clock match while (timeNow < pMsg->startTime) { - sRInfo(pReceiver, "snapshot receiver pre waitting for true time, now:%" PRId64 ", stime:%" PRId64, timeNow, + sRInfo(pReceiver, "snapshot receiver pre waitting for true time, now:%" PRId64 ", startTime:%" PRId64, timeNow, pMsg->startTime); taosMsleep(10); timeNow = taosGetTimestampMs(); @@ -765,7 +765,7 @@ static int32_t syncNodeOnSnapshotEnd(SSyncNode *pSyncNode, SyncSnapshotSend *pMs // receiver on message // -// condition 1, recv SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT +// condition 1, recv SYNC_SNAPSHOT_SEQ_PREP_SNAPSHOT // if receiver already start // if sender.start-time > receiver.start-time, restart receiver(reply snapshot start) // if sender.start-time = receiver.start-time, maybe duplicate msg @@ -809,9 +809,9 @@ int32_t syncNodeOnSnapshot(SSyncNode *pSyncNode, const SRpcMsg *pRpcMsg) { int32_t code = 0; if (pSyncNode->state == TAOS_SYNC_STATE_FOLLOWER) { if (pMsg->term == pSyncNode->raftStore.currentTerm) { - if (pMsg->seq == SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT) { + if (pMsg->seq == SYNC_SNAPSHOT_SEQ_PREP_SNAPSHOT) { syncLogRecvSyncSnapshotSend(pSyncNode, pMsg, "process seq pre-snapshot"); - code = syncNodeOnSnapshotPre(pSyncNode, pMsg); + code = syncNodeOnSnapshotPrep(pSyncNode, pMsg); } else if (pMsg->seq == SYNC_SNAPSHOT_SEQ_BEGIN) { syncLogRecvSyncSnapshotSend(pSyncNode, pMsg, "process seq begin"); code = syncNodeOnSnapshotBegin(pSyncNode, pMsg); @@ -848,7 +848,7 @@ int32_t syncNodeOnSnapshot(SSyncNode *pSyncNode, const SRpcMsg *pRpcMsg) { return code; } -static int32_t syncNodeOnSnapshotPreRsp(SSyncNode *pSyncNode, SSyncSnapshotSender *pSender, SyncSnapshotRsp *pMsg) { +static int32_t syncNodeOnSnapshotPrepRsp(SSyncNode *pSyncNode, SSyncSnapshotSender *pSender, SyncSnapshotRsp *pMsg) { SSnapshot snapshot = {0}; pSyncNode->pFsm->FpGetSnapshotInfo(pSyncNode->pFsm, &snapshot); @@ -945,8 +945,8 @@ int32_t syncNodeOnSnapshotRsp(SSyncNode *pSyncNode, const SRpcMsg *pRpcMsg) { if (pMsg->startTime != pSender->startTime) { syncLogRecvSyncSnapshotRsp(pSyncNode, pMsg, "snapshot sender and receiver time not match"); - sSError(pSender, "sender:%" PRId64 " receiver:%" PRId64 " time not match, code:0x%x", pMsg->startTime, - pSender->startTime, pMsg->code); + sSError(pSender, "sender:%" PRId64 " receiver:%" PRId64 " time not match, error:%s 0x%x", pMsg->startTime, + pSender->startTime, tstrerror(pMsg->code), pMsg->code); terrno = TSDB_CODE_SYN_INTERNAL_ERROR; goto _ERROR; } @@ -961,15 +961,15 @@ int32_t syncNodeOnSnapshotRsp(SSyncNode *pSyncNode, const SRpcMsg *pRpcMsg) { if (pMsg->code != 0) { syncLogRecvSyncSnapshotRsp(pSyncNode, pMsg, "receive error code"); - sSError(pSender, "snapshot sender receive error code:0x%x and stop sender", pMsg->code); + sSError(pSender, "snapshot sender receive error:%s 0x%x and stop sender", tstrerror(pMsg->code), pMsg->code); terrno = pMsg->code; goto _ERROR; } // prepare , send begin msg - if (pMsg->ack == SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT) { + if (pMsg->ack == SYNC_SNAPSHOT_SEQ_PREP_SNAPSHOT) { syncLogRecvSyncSnapshotRsp(pSyncNode, pMsg, "process seq pre-snapshot"); - return syncNodeOnSnapshotPreRsp(pSyncNode, pSender, pMsg); + return syncNodeOnSnapshotPrepRsp(pSyncNode, pSender, pMsg); } if (pSender->pReader == NULL || pSender->finish) { From 40b745932ce8b17f1381ae12231c5391b16b1b78 Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Thu, 2 Feb 2023 19:39:08 +0800 Subject: [PATCH 080/124] fix: scalar expr memory leak issue --- source/libs/executor/src/timesliceoperator.c | 1 + 1 file changed, 1 insertion(+) diff --git a/source/libs/executor/src/timesliceoperator.c b/source/libs/executor/src/timesliceoperator.c index 6561e810bb..bca87079ec 100644 --- a/source/libs/executor/src/timesliceoperator.c +++ b/source/libs/executor/src/timesliceoperator.c @@ -634,6 +634,7 @@ void destroyTimeSliceOperatorInfo(void* param) { taosMemoryFree(pKey->end.val); } taosArrayDestroy(pInfo->pLinearInfo); + cleanupExprSupp(&pInfo->scalarSup); taosMemoryFree(pInfo->pFillColInfo); taosMemoryFreeClear(param); From 1766e46fca0333ceded8506f4d45c146ea7729ea Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Thu, 2 Feb 2023 21:46:48 +0800 Subject: [PATCH 081/124] fix compile error --- source/libs/transport/src/transCli.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index be687b6193..401f9c5120 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -805,7 +805,7 @@ void cliSend(SCliConn* pConn) { if (transQueueEmpty(&pConn->cliMsgs)) { tError("%s conn %p not msg to send", pTransInst->label, pConn); - cliHandleExecept(pConn); + cliHandleExcept(pConn); return; } From 561ee87e9dd9a5ef2f1a6e9e3edba00b624b01d7 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Fri, 3 Feb 2023 09:22:00 +0800 Subject: [PATCH 082/124] enh: change the level of sync probe logging msg to trace --- source/libs/sync/src/syncPipeline.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/sync/src/syncPipeline.c b/source/libs/sync/src/syncPipeline.c index 6cc517fda0..bb3bb0d6a4 100644 --- a/source/libs/sync/src/syncPipeline.c +++ b/source/libs/sync/src/syncPipeline.c @@ -830,7 +830,7 @@ int32_t syncLogReplMgrReplicateProbe(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sy pMgr->endIndex = index + 1; SSyncLogBuffer* pBuf = pNode->pLogBuf; - sInfo("vgId:%d, probe peer:%" PRIx64 " with msg of index:%" PRId64 " term: %" PRId64 ". mgr (rs:%d): [%" PRId64 + sTrace("vgId:%d, probe peer:%" PRIx64 " with msg of index:%" PRId64 " term: %" PRId64 ". mgr (rs:%d): [%" PRId64 " %" PRId64 ", %" PRId64 "), buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pDestId->addr, index, term, pMgr->restored, pMgr->startIndex, pMgr->matchIndex, pMgr->endIndex, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); From d83a94bb195bcbd6e0c68ebcbb65cb511b7672f6 Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Fri, 3 Feb 2023 09:48:50 +0800 Subject: [PATCH 083/124] fix: check dup rebalance --- source/dnode/mnode/impl/src/mndConsumer.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/source/dnode/mnode/impl/src/mndConsumer.c b/source/dnode/mnode/impl/src/mndConsumer.c index 3bbf4a4279..1aa2fa997b 100644 --- a/source/dnode/mnode/impl/src/mndConsumer.c +++ b/source/dnode/mnode/impl/src/mndConsumer.c @@ -836,10 +836,13 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer, char *addedTopic = strdup(taosArrayGetP(pNewConsumer->rebNewTopics, 0)); // not exist in current topic + bool existing = false; #if 1 for (int32_t i = 0; i < taosArrayGetSize(pOldConsumer->currentTopics); i++) { char *topic = taosArrayGetP(pOldConsumer->currentTopics, i); - ASSERT(strcmp(topic, addedTopic) != 0); + if (strcmp(topic, addedTopic) == 0) { + existing = true; + } } #endif @@ -854,8 +857,10 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer, } // add to current topic - taosArrayPush(pOldConsumer->currentTopics, &addedTopic); - taosArraySort(pOldConsumer->currentTopics, taosArrayCompareString); + if (!existing) { + taosArrayPush(pOldConsumer->currentTopics, &addedTopic); + taosArraySort(pOldConsumer->currentTopics, taosArrayCompareString); + } // set status if (taosArrayGetSize(pOldConsumer->rebNewTopics) == 0 && taosArrayGetSize(pOldConsumer->rebRemovedTopics) == 0) { From ca16ebc6e77c0c5a8cd49cf3d935418c29a23a58 Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Fri, 3 Feb 2023 10:36:47 +0800 Subject: [PATCH 084/124] fix: add get ip addr result validation --- source/libs/transport/src/transCli.c | 19 ++++++++++++++++++- source/os/src/osSocket.c | 2 +- 2 files changed, 19 insertions(+), 2 deletions(-) diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index 1a99db5f99..ef64e8e040 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -1045,6 +1045,12 @@ static FORCE_INLINE uint32_t cliGetIpFromFqdnCache(SHashObj* cache, char* fqdn) uint32_t* v = taosHashGet(cache, fqdn, strlen(fqdn)); if (v == NULL) { addr = taosGetIpv4FromFqdn(fqdn); + if (addr == 0xffffffff) { + terrno = TAOS_SYSTEM_ERROR(errno); + tError("failed to get ip from fqdn:%s since %s", fqdn, terrstr()); + return addr; + } + taosHashPut(cache, fqdn, strlen(fqdn), &addr, sizeof(addr)); } else { addr = *v; @@ -1116,9 +1122,20 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrd* pThrd) { conn->ip = strdup(EPSET_GET_INUSE_IP(&pCtx->epSet)); conn->port = EPSET_GET_INUSE_PORT(&pCtx->epSet); + uint32_t ipaddr = cliGetIpFromFqdnCache(pThrd->fqdn2ipCache, conn->ip); + if (ipaddr == 0xffffffff) { + uv_timer_stop(conn->timer); + conn->timer->data = NULL; + taosArrayPush(pThrd->timerList, &conn->timer); + conn->timer = NULL; + + cliHandleExcept(conn); + return; + } + struct sockaddr_in addr; addr.sin_family = AF_INET; - addr.sin_addr.s_addr = cliGetIpFromFqdnCache(pThrd->fqdn2ipCache, conn->ip); + addr.sin_addr.s_addr = ipaddr; addr.sin_port = (uint16_t)htons((uint16_t)conn->port); STraceId* trace = &(pMsg->msg.info.traceId); diff --git a/source/os/src/osSocket.c b/source/os/src/osSocket.c index 6611a937f2..de628c7596 100644 --- a/source/os/src/osSocket.c +++ b/source/os/src/osSocket.c @@ -944,7 +944,7 @@ uint32_t taosGetIpv4FromFqdn(const char *fqdn) { iResult = WSAStartup(MAKEWORD(2, 2), &wsaData); if (iResult != 0) { // printf("WSAStartup failed: %d\n", iResult); - return 1; + return 0xFFFFFFFF; } #endif struct addrinfo hints = {0}; From cdf362b58861a6c36c3c14c8bdacb36add09986c Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Fri, 3 Feb 2023 12:08:16 +0800 Subject: [PATCH 085/124] fix: fix fd limit crash --- source/client/src/clientImpl.c | 1 + source/libs/transport/src/transCli.c | 23 ++++++++++++++++++++--- source/libs/transport/src/transComm.c | 21 +++++++++++++++++---- 3 files changed, 38 insertions(+), 7 deletions(-) diff --git a/source/client/src/clientImpl.c b/source/client/src/clientImpl.c index b5b99e92b0..09f614ddb2 100644 --- a/source/client/src/clientImpl.c +++ b/source/client/src/clientImpl.c @@ -1460,6 +1460,7 @@ void processMsgFromServer(void* parent, SRpcMsg* pMsg, SEpSet* pEpSet) { tscError("failed to sched msg to tsc, tsc ready to quit"); rpcFreeCont(pMsg->pCont); taosMemoryFree(arg->pEpset); + destroySendMsgInfo(pMsg->info.ahandle); taosMemoryFree(arg); } } diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index 1a99db5f99..954303e3c1 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -1275,7 +1275,11 @@ void* transInitClient(uint32_t ip, uint32_t port, char* label, int numOfThreads, for (int i = 0; i < cli->numOfThreads; i++) { SCliThrd* pThrd = createThrdObj(shandle); - int err = taosThreadCreate(&pThrd->thread, NULL, cliWorkThread, (void*)(pThrd)); + if (pThrd == NULL) { + return NULL; + } + + int err = taosThreadCreate(&pThrd->thread, NULL, cliWorkThread, (void*)(pThrd)); if (err == 0) { tDebug("success to create tranport-cli thread:%d", i); } @@ -1332,9 +1336,22 @@ static SCliThrd* createThrdObj(void* trans) { taosThreadMutexInit(&pThrd->msgMtx, NULL); pThrd->loop = (uv_loop_t*)taosMemoryMalloc(sizeof(uv_loop_t)); - uv_loop_init(pThrd->loop); - + int err = uv_loop_init(pThrd->loop); + if (err != 0) { + tError("failed to init uv_loop, reason:%s", uv_err_name(err)); + taosMemoryFree(pThrd->loop); + taosThreadMutexDestroy(&pThrd->msgMtx); + taosMemoryFree(pThrd); + return NULL; + } pThrd->asyncPool = transAsyncPoolCreate(pThrd->loop, 8, pThrd, cliAsyncCb); + if (pThrd->asyncPool == NULL) { + uv_loop_close(pThrd->loop); + taosMemoryFree(pThrd->loop); + taosThreadMutexDestroy(&pThrd->msgMtx); + taosMemoryFree(pThrd); + return NULL; + } pThrd->prepare = taosMemoryCalloc(1, sizeof(uv_prepare_t)); uv_prepare_init(pThrd->loop, pThrd->prepare); diff --git a/source/libs/transport/src/transComm.c b/source/libs/transport/src/transComm.c index 1161ed7c00..8c9e8f5a60 100644 --- a/source/libs/transport/src/transComm.c +++ b/source/libs/transport/src/transComm.c @@ -214,24 +214,37 @@ SAsyncPool* transAsyncPoolCreate(uv_loop_t* loop, int sz, void* arg, AsyncCB cb) pool->nAsync = sz; pool->asyncs = taosMemoryCalloc(1, sizeof(uv_async_t) * pool->nAsync); - for (int i = 0; i < pool->nAsync; i++) { + int i = 0, err = 0; + for (i = 0; i < pool->nAsync; i++) { + uv_async_t* async = &(pool->asyncs[i]); + SAsyncItem* item = taosMemoryCalloc(1, sizeof(SAsyncItem)); item->pThrd = arg; QUEUE_INIT(&item->qmsg); taosThreadMutexInit(&item->mtx, NULL); - uv_async_t* async = &(pool->asyncs[i]); - uv_async_init(loop, async, cb); async->data = item; + err = uv_async_init(loop, async, cb); + if (err != 0) { + tError("failed to init async, reason:%s", uv_err_name(err)); + break; + } } + + if (i != pool->nAsync) { + transAsyncPoolDestroy(pool); + pool = NULL; + } + return pool; } void transAsyncPoolDestroy(SAsyncPool* pool) { for (int i = 0; i < pool->nAsync; i++) { uv_async_t* async = &(pool->asyncs[i]); - SAsyncItem* item = async->data; + if (item == NULL) continue; + taosThreadMutexDestroy(&item->mtx); taosMemoryFree(item); } From 9f62cfd6f330599b6424c82e495557b195712395 Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Fri, 3 Feb 2023 12:12:58 +0800 Subject: [PATCH 086/124] ref log --- source/libs/transport/src/transCli.c | 1 + 1 file changed, 1 insertion(+) diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index 954303e3c1..9626c92216 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -1346,6 +1346,7 @@ static SCliThrd* createThrdObj(void* trans) { } pThrd->asyncPool = transAsyncPoolCreate(pThrd->loop, 8, pThrd, cliAsyncCb); if (pThrd->asyncPool == NULL) { + tError("failed to init async pool"); uv_loop_close(pThrd->loop); taosMemoryFree(pThrd->loop); taosThreadMutexDestroy(&pThrd->msgMtx); From e14994c23f9db8fcf0498ded13244d283492d71d Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Fri, 3 Feb 2023 12:45:20 +0800 Subject: [PATCH 087/124] fix: fail fast --- source/client/src/clientImpl.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/source/client/src/clientImpl.c b/source/client/src/clientImpl.c index 09f614ddb2..3c26e674b4 100644 --- a/source/client/src/clientImpl.c +++ b/source/client/src/clientImpl.c @@ -138,6 +138,12 @@ STscObj* taos_connect_internal(const char* ip, const char* user, const char* pas p->mgmtEp = epSet; taosThreadMutexInit(&p->qnodeMutex, NULL); p->pTransporter = openTransporter(user, secretEncrypt, tsNumOfCores / 2); + if (p->pTransporter == NULL) { + taosThreadMutexUnlock(&appInfo.mutex); + taosMemoryFreeClear(key); + taosMemoryFree(p); + return NULL; + } p->pAppHbMgr = appHbMgrInit(p, key); if (NULL == p->pAppHbMgr) { destroyAppInst(p); From e7a72d5a284b6a204bcb3165e250a321917fbf8b Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Fri, 3 Feb 2023 13:57:26 +0800 Subject: [PATCH 088/124] fix: no resp when host machine shutdown --- source/libs/transport/src/transCli.c | 29 ++++++++++++++++++++++----- source/libs/transport/src/transComm.c | 4 ++++ source/os/src/osSocket.c | 8 ++++---- 3 files changed, 32 insertions(+), 9 deletions(-) diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index 1a99db5f99..01929b237f 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -671,7 +671,7 @@ static SCliConn* cliCreateConn(SCliThrd* pThrd) { conn->stream = (uv_stream_t*)taosMemoryMalloc(sizeof(uv_tcp_t)); uv_tcp_init(pThrd->loop, (uv_tcp_t*)(conn->stream)); conn->stream->data = conn; - transSetConnOption((uv_tcp_t*)conn->stream); + // transSetConnOption((uv_tcp_t*)conn->stream); uv_timer_t* timer = taosArrayGetSize(pThrd->timerList) > 0 ? *(uv_timer_t**)taosArrayPop(pThrd->timerList) : NULL; if (timer == NULL) { @@ -1061,9 +1061,10 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrd* pThrd) { STransConnCtx* pCtx = pMsg->ctx; cliMayCvtFqdnToIp(&pCtx->epSet, &pThrd->cvtAddr); + STraceId* trace = &pMsg->msg.info.traceId; if (!EPSET_IS_VALID(&pCtx->epSet)) { - tError("invalid epset"); + tGError("%s, msg %s sent with invalid epset", pTransInst->label, TMSG_INFO(pMsg->msg.msgType)); destroyCmsg(pMsg); return; } @@ -1121,10 +1122,29 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrd* pThrd) { addr.sin_addr.s_addr = cliGetIpFromFqdnCache(pThrd->fqdn2ipCache, conn->ip); addr.sin_port = (uint16_t)htons((uint16_t)conn->port); - STraceId* trace = &(pMsg->msg.info.traceId); tGTrace("%s conn %p try to connect to %s:%d", pTransInst->label, conn, conn->ip, conn->port); + int32_t fd = taosCreateSocketWithTimeout(TRANS_CONN_TIMEOUT * 4); + if (fd == -1) { + tGError("%s conn %p failed to create socket, reason:%s", transLabel(pTransInst), conn, + tstrerror(TAOS_SYSTEM_ERROR(errno))); + cliHandleExcept(conn); + errno = 0; + return; + } + int ret = uv_tcp_open((uv_tcp_t*)conn->stream, fd); + if (ret != 0) { + tGError("%s conn %p failed to set stream, reason:%s", transLabel(pTransInst), conn, uv_err_name(ret)); + cliHandleExcept(conn); + return; + } + ret = transSetConnOption((uv_tcp_t*)conn->stream); + if (ret != 0) { + tGError("%s conn %p failed to set socket opt, reason:%s", transLabel(pTransInst), conn, uv_err_name(ret)); + cliHandleExcept(conn); + return; + } - int ret = uv_tcp_connect(&conn->connReq, (uv_tcp_t*)(conn->stream), (const struct sockaddr*)&addr, cliConnCb); + ret = uv_tcp_connect(&conn->connReq, (uv_tcp_t*)(conn->stream), (const struct sockaddr*)&addr, cliConnCb); if (ret != 0) { tGError("%s conn %p failed to connect to %s:%d, reason:%s", pTransInst->label, conn, conn->ip, conn->port, uv_err_name(ret)); @@ -1139,7 +1159,6 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrd* pThrd) { } uv_timer_start(conn->timer, cliConnTimeout, TRANS_CONN_TIMEOUT, 0); } - STraceId* trace = &pMsg->msg.info.traceId; tGTrace("%s conn %p ready", pTransInst->label, conn); } static void cliAsyncCb(uv_async_t* handle) { diff --git a/source/libs/transport/src/transComm.c b/source/libs/transport/src/transComm.c index 1161ed7c00..41233a6526 100644 --- a/source/libs/transport/src/transComm.c +++ b/source/libs/transport/src/transComm.c @@ -205,6 +205,10 @@ bool transReadComplete(SConnBuffer* connBuf) { } int transSetConnOption(uv_tcp_t* stream) { +#if defined(WINDOWS) || defined(DARWIN) +#else + uv_tcp_keepalive(stream, 1, 20); +#endif return uv_tcp_nodelay(stream, 1); // int ret = uv_tcp_keepalive(stream, 5, 60); } diff --git a/source/os/src/osSocket.c b/source/os/src/osSocket.c index 6611a937f2..3c7a1e64ee 100644 --- a/source/os/src/osSocket.c +++ b/source/os/src/osSocket.c @@ -55,7 +55,7 @@ typedef struct TdSocket { #endif int refId; SocketFd fd; -} * TdSocketPtr, TdSocket; +} *TdSocketPtr, TdSocket; typedef struct TdSocketServer { #if SOCKET_WITH_LOCK @@ -63,7 +63,7 @@ typedef struct TdSocketServer { #endif int refId; SocketFd fd; -} * TdSocketServerPtr, TdSocketServer; +} *TdSocketServerPtr, TdSocketServer; typedef struct TdEpoll { #if SOCKET_WITH_LOCK @@ -71,7 +71,7 @@ typedef struct TdEpoll { #endif int refId; EpollFd fd; -} * TdEpollPtr, TdEpoll; +} *TdEpollPtr, TdEpoll; #if 0 int32_t taosSendto(TdSocketPtr pSocket, void *buf, int len, unsigned int flags, const struct sockaddr *dest_addr, @@ -1005,7 +1005,7 @@ int32_t taosGetFqdn(char *fqdn) { // immediately // hints.ai_family = AF_INET; strcpy(fqdn, hostname); - strcpy(fqdn+strlen(hostname), ".local"); + strcpy(fqdn + strlen(hostname), ".local"); #else // __APPLE__ struct addrinfo hints = {0}; struct addrinfo *result = NULL; From 383bcea0cbf51e1270f8b26b063539e030e5089a Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Fri, 3 Feb 2023 14:17:36 +0800 Subject: [PATCH 089/124] fix err log --- include/common/tmsg.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/include/common/tmsg.h b/include/common/tmsg.h index 0cc9fb8619..c4a3cad28a 100644 --- a/include/common/tmsg.h +++ b/include/common/tmsg.h @@ -58,7 +58,7 @@ extern int32_t tMsgDict[]; #define TMSG_INFO(TYPE) \ ((TYPE) < TDMT_DND_MAX_MSG || (TYPE) < TDMT_MND_MAX_MSG || (TYPE) < TDMT_VND_MAX_MSG || (TYPE) < TDMT_SCH_MAX_MSG || \ (TYPE) < TDMT_STREAM_MAX_MSG || (TYPE) < TDMT_MON_MAX_MSG || (TYPE) < TDMT_SYNC_MAX_MSG) || \ - (TYPE) < TDMT_VND_STREAM_MSG || (TYPE) < TDMT_VND_TMQ_MSG \ + (TYPE) < TDMT_VND_STREAM_MSG || (TYPE) < TDMT_VND_TMQ_MSG || (TYPE) < TDMT_VND_TMQ_MAX_MSG \ ? tMsgInfo[tMsgDict[TMSG_SEG_CODE(TYPE)] + TMSG_SEG_SEQ(TYPE)] \ : 0 From 49631f6e9647617ef13c230d7e3f31d9b7c27a18 Mon Sep 17 00:00:00 2001 From: Hongze Cheng Date: Fri, 3 Feb 2023 14:34:55 +0800 Subject: [PATCH 090/124] fix: snapshot coredump caused by type --- source/dnode/vnode/src/tsdb/tsdbSnapshot.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c index 98c9c0fdda..e8181f922f 100644 --- a/source/dnode/vnode/src/tsdb/tsdbSnapshot.c +++ b/source/dnode/vnode/src/tsdb/tsdbSnapshot.c @@ -1205,7 +1205,7 @@ static int32_t tsdbSnapWriteTableRow(STsdbSnapWriter* pWriter, TSDBROW* pRow) { TSDB_CHECK_CODE(code, lino, _exit); } - tMapDataPutItem(&pWriter->pDIter->dIter.mDataBlk, &dataBlk, tPutDataBlk); + tMapDataPutItem(&pWriter->mDataBlk, &dataBlk, tPutDataBlk); pWriter->pDIter->dIter.iDataBlk++; } else { code = tsdbReadDataBlockEx(pWriter->pDataFReader, &dataBlk, &pWriter->pDIter->dIter.bData); From b21ed4b617423cd37e09af65979257a2bd719aaf Mon Sep 17 00:00:00 2001 From: Shengliang Guan Date: Fri, 3 Feb 2023 15:49:51 +0800 Subject: [PATCH 091/124] fix: not apply to mnode if trans createTime not matched --- source/dnode/mnode/impl/src/mndSync.c | 6 +++++- source/dnode/mnode/impl/src/mndTrans.c | 16 ++++++++++++++-- 2 files changed, 19 insertions(+), 3 deletions(-) diff --git a/source/dnode/mnode/impl/src/mndSync.c b/source/dnode/mnode/impl/src/mndSync.c index 6b675586e4..d458ffaed4 100644 --- a/source/dnode/mnode/impl/src/mndSync.c +++ b/source/dnode/mnode/impl/src/mndSync.c @@ -85,7 +85,11 @@ int32_t mndProcessWriteMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta pRaw, pMgmt->transSec, pMgmt->transSeq); if (pMeta->code == 0) { - sdbWriteWithoutFree(pMnode->pSdb, pRaw); + int32_t code = sdbWriteWithoutFree(pMnode->pSdb, pRaw); + if (code != 0) { + mError("trans:%d, failed to write to sdb since %s", transId, terrstr()); + return 0; + } sdbSetApplyInfo(pMnode->pSdb, pMeta->index, pMeta->term, pMeta->lastConfigIndex); } diff --git a/source/dnode/mnode/impl/src/mndTrans.c b/source/dnode/mnode/impl/src/mndTrans.c index dfcd55bcba..0b28a6eb43 100644 --- a/source/dnode/mnode/impl/src/mndTrans.c +++ b/source/dnode/mnode/impl/src/mndTrans.c @@ -572,8 +572,20 @@ static void mndTransUpdateActions(SArray *pOldArray, SArray *pNewArray) { } static int32_t mndTransActionUpdate(SSdb *pSdb, STrans *pOld, STrans *pNew) { - mTrace("trans:%d, perform update action, old row:%p stage:%s, new row:%p stage:%s", pOld->id, pOld, - mndTransStr(pOld->stage), pNew, mndTransStr(pNew->stage)); + mTrace("trans:%d, perform update action, old row:%p stage:%s create:%" PRId64 ", new row:%p stage:%s create:%" PRId64, + pOld->id, pOld, mndTransStr(pOld->stage), pOld->createdTime, pNew, mndTransStr(pNew->stage), + pNew->createdTime); + + if (pOld->createdTime != pNew->createdTime) { + mError("trans:%d, failed to perform update action since createTime not match, old row:%p stage:%s create:%" PRId64 + ", new row:%p stage:%s create:%" PRId64, + pOld->id, pOld, mndTransStr(pOld->stage), pOld->createdTime, pNew, mndTransStr(pNew->stage), + pNew->createdTime); + // only occured while sync timeout + terrno = TSDB_CODE_MND_TRNAS_SYNC_TIMEOUT; + return -1; + } + mndTransUpdateActions(pOld->redoActions, pNew->redoActions); mndTransUpdateActions(pOld->undoActions, pNew->undoActions); mndTransUpdateActions(pOld->commitActions, pNew->commitActions); From f803fa44b578211558a170eb61807e652a9b73fb Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Fri, 3 Feb 2023 16:50:13 +0800 Subject: [PATCH 092/124] fix(query): set the correct number of rows when calculating using sma for min/max function. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 51 ++++----------------- source/libs/function/src/detail/tminmax.c | 13 ++---- tests/script/tsim/parser/regressiontest.sim | 39 +++++++++++++--- 3 files changed, 44 insertions(+), 59 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index db52d7604f..4e78dbe392 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -2839,12 +2839,11 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { ASSERT(pBlockInfo != NULL); - if (pBlockInfo != NULL) { - pScanInfo = - *(STableBlockScanInfo**)taosHashGet(pReader->status.pTableMap, &pBlockInfo->uid, sizeof(pBlockInfo->uid)); - } else { - pScanInfo = *pReader->status.pTableIter; - } +// if (pBlockInfo != NULL) { + pScanInfo = *(STableBlockScanInfo**)taosHashGet(pReader->status.pTableMap, &pBlockInfo->uid, sizeof(pBlockInfo->uid)); + // } else { + // pScanInfo = *pReader->status.pTableIter; + // } if (pScanInfo == NULL) { tsdbError("failed to get table scan-info, %s", pReader->idStr); @@ -2852,46 +2851,14 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { return code; } - if (pBlockInfo != NULL) { - pBlock = getCurrentBlock(pBlockIter); - } +// if (pBlockInfo != NULL) { + pBlock = getCurrentBlock(pBlockIter); +// } initLastBlockReader(pLastBlockReader, pScanInfo, pReader); TSDBKEY keyInBuf = getCurrentKeyInBuf(pScanInfo, pReader); - /*if (pBlockInfo == NULL) { // build data block from last data file - SBlockData* pBData = &pReader->status.fileBlockData; - tBlockDataReset(pBData); - - SSDataBlock* pResBlock = pReader->pResBlock; - tsdbDebug("load data in last block firstly, due to desc scan data, %s", pReader->idStr); - - int64_t st = taosGetTimestampUs(); - - while (1) { - bool hasBlockLData = hasDataInLastBlock(pLastBlockReader); - - // no data in last block and block, no need to proceed. - if (hasBlockLData == false) { - break; - } - - buildComposedDataBlockImpl(pReader, pScanInfo, &pReader->status.fileBlockData, pLastBlockReader); - if (pResBlock->info.rows >= pReader->capacity) { - break; - } - } - - double el = (taosGetTimestampUs() - st) / 1000.0; - updateComposedBlockInfo(pReader, el, pScanInfo); - - if (pResBlock->info.rows > 0) { - tsdbDebug("%p uid:%" PRIu64 ", composed data block created, brange:%" PRIu64 "-%" PRIu64 - " rows:%d, elapsed time:%.2f ms %s", - pReader, pResBlock->info.id.uid, pResBlock->info.window.skey, pResBlock->info.window.ekey, - pResBlock->info.rows, el, pReader->idStr); - } - } else*/ if (fileBlockShouldLoad(pReader, pBlockInfo, pBlock, pScanInfo, keyInBuf, pLastBlockReader)) { + if (fileBlockShouldLoad(pReader, pBlockInfo, pBlock, pScanInfo, keyInBuf, pLastBlockReader)) { code = doLoadFileBlockData(pReader, pBlockIter, &pStatus->fileBlockData, pScanInfo->uid); if (code != TSDB_CODE_SUCCESS) { return code; diff --git a/source/libs/function/src/detail/tminmax.c b/source/libs/function/src/detail/tminmax.c index 847c738655..a511ca97f1 100644 --- a/source/libs/function/src/detail/tminmax.c +++ b/source/libs/function/src/detail/tminmax.c @@ -714,26 +714,18 @@ int32_t doMinMaxHelper(SqlFunctionCtx* pCtx, int32_t isMinFunc, int32_t* nElems) pBuf->type = type; if (IS_NULL_TYPE(type)) { - numOfElems = 0; goto _over; } // data in current data block are qualified to the query if (pInput->colDataSMAIsSet) { numOfElems = pInput->numOfRows - pAgg->numOfNull; - if (numOfElems == 0) { goto _over; } - void* tval = NULL; int16_t index = 0; - - if (isMinFunc) { - tval = &pInput->pColumnDataAgg[0]->min; - } else { - tval = &pInput->pColumnDataAgg[0]->max; - } + void* tval = (isMinFunc) ? &pInput->pColumnDataAgg[0]->min : &pInput->pColumnDataAgg[0]->max; if (!pBuf->assign) { if (type == TSDB_DATA_TYPE_FLOAT) { @@ -824,8 +816,9 @@ int32_t doMinMaxHelper(SqlFunctionCtx* pCtx, int32_t isMinFunc, int32_t* nElems) } } + numOfElems = 1; pBuf->assign = true; - return TSDB_CODE_SUCCESS; + goto _over; } int32_t start = pInput->startRowIndex; diff --git a/tests/script/tsim/parser/regressiontest.sim b/tests/script/tsim/parser/regressiontest.sim index 3ce2b47b44..1d84ae88cb 100644 --- a/tests/script/tsim/parser/regressiontest.sim +++ b/tests/script/tsim/parser/regressiontest.sim @@ -31,13 +31,8 @@ sql insert into $tb values ( $ts , $x ) $x = $x + 1 endw -print ================== restart server to commit data into disk -system sh/exec.sh -n dnode1 -s stop -x SIGINT -system sh/exec.sh -n dnode1 -s start -print ================== server restart completed -sql connect +sql flush database $db -sql use $db sql delete from $tb where ts=1537146000000 sql delete from $tb where ts=1537146409500 @@ -63,7 +58,6 @@ if $rows != 8198 then return -1 endi -print ===========================> TD-22077 && TD-21877 sql drop database if exists $db -x step1 sql create database $db vgroups 1; @@ -88,6 +82,8 @@ endw sql flush database $db +print ===========================> TD-22077 && TD-21877 + sql insert into t1 values('2018-09-17 09:00:26', 26); sql insert into t2 values('2018-09-17 09:00:25', 25); @@ -97,4 +93,33 @@ sql flush database reg_db0; sql delete from st1 where ts<='2018-9-17 09:00:26'; sql select * from st1; +sql drop table t1 +sql drop table t2 + +print =========================================>TD-22196 +sql create table t1 using st1 tags(1); + +$i = 0 +$ts = 1674977959000 +$rowNum = 200 + +$x = 0 +while $x < $rowNum +$xs = $x * $delta +$ts = $ts0 + $xs +sql insert into t1 values ( $ts , $x ) +$x = $x + 1 +$ts = $ts + 1000 +endw + +sql flush database $db +sql select min(c),max(c) from t1 +if $data00 != 0 then + return -1 +endi + +if $data01 != 199 then + return -1 +endi + system sh/exec.sh -n dnode1 -s stop -x SIGINT From 845530e1251febfd96370d801d11e84c8675f470 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Fri, 3 Feb 2023 17:33:35 +0800 Subject: [PATCH 093/124] other: add some logs. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 6 +++++- source/libs/executor/src/scanoperator.c | 4 ++-- 2 files changed, 7 insertions(+), 3 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 4e78dbe392..1fd934066a 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -2904,7 +2904,7 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { if (pResBlock->info.rows > 0) { tsdbDebug("%p uid:%" PRIu64 ", composed data block created, brange:%" PRIu64 "-%" PRIu64 - " rows:%d, elapsed time:%.2f ms %s", + " rows:%d, elapsed time:%.2f ms %s", pReader, pResBlock->info.id.uid, pResBlock->info.window.skey, pResBlock->info.window.ekey, pResBlock->info.rows, el, pReader->idStr); } @@ -2918,6 +2918,10 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { // update the last key for the corresponding table pScanInfo->lastKey = ASCENDING_TRAVERSE(pReader->order) ? pInfo->window.ekey : pInfo->window.skey; + tsdbDebug("%p uid:%" PRIu64 " clean file block retrieved from file, global index:%d, " + "table index:%d, rows:%d, brange:%" PRId64 "-%" PRId64 ", %s", + pReader, pScanInfo->uid, pBlockIter->index, pBlockInfo->tbBlockIdx, pBlock->nRow, pBlock->minKey.ts, + pBlock->maxKey.ts, pReader->idStr); } } diff --git a/source/libs/executor/src/scanoperator.c b/source/libs/executor/src/scanoperator.c index d54f5bec5a..169693b8c4 100644 --- a/source/libs/executor/src/scanoperator.c +++ b/source/libs/executor/src/scanoperator.c @@ -311,8 +311,8 @@ static int32_t loadDataBlock(SOperatorInfo* pOperator, STableScanBase* pTableSca pCost->totalRows += pBlock->info.rows; return TSDB_CODE_SUCCESS; } else if (*status == FUNC_DATA_REQUIRED_NOT_LOAD) { - qDebug("%s data block skipped, brange:%" PRId64 "-%" PRId64 ", rows:%d", GET_TASKID(pTaskInfo), - pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows); + qDebug("%s data block skipped, brange:%" PRId64 "-%" PRId64 ", rows:%d, uid:%"PRIu64, GET_TASKID(pTaskInfo), + pBlockInfo->window.skey, pBlockInfo->window.ekey, pBlockInfo->rows, pBlockInfo->id.uid); doSetTagColumnData(pTableScanInfo, pBlock, pTaskInfo, 1); pCost->skipBlocks += 1; return TSDB_CODE_SUCCESS; From 5a5dca31133bcdcbb607df5bb20b9cb77e66ea95 Mon Sep 17 00:00:00 2001 From: sunpeng Date: Fri, 3 Feb 2023 17:40:53 +0800 Subject: [PATCH 094/124] docs: revert go mod file (#19791) * docs: revert go mod file * delete indirect --- docs/examples/go/go.mod | 6 ++++++ 1 file changed, 6 insertions(+) create mode 100644 docs/examples/go/go.mod diff --git a/docs/examples/go/go.mod b/docs/examples/go/go.mod new file mode 100644 index 0000000000..716a0ef5dc --- /dev/null +++ b/docs/examples/go/go.mod @@ -0,0 +1,6 @@ +module goexample + +go 1.17 + +require github.com/taosdata/driver-go/v3 v3.1.0 + From e0412d30576261889835b76cb96d087f3ace5841 Mon Sep 17 00:00:00 2001 From: Benguang Zhao Date: Fri, 3 Feb 2023 18:49:55 +0800 Subject: [PATCH 095/124] fix: request snapshot from maximum of commit vers of tsdb and wal --- source/libs/sync/src/syncSnapshot.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) diff --git a/source/libs/sync/src/syncSnapshot.c b/source/libs/sync/src/syncSnapshot.c index e61bcc9ffc..18f263cc95 100644 --- a/source/libs/sync/src/syncSnapshot.c +++ b/source/libs/sync/src/syncSnapshot.c @@ -510,16 +510,8 @@ SyncIndex syncNodeGetSnapBeginIndex(SSyncNode *ths) { SSyncLogStoreData *pData = ths->pLogStore->data; SWal *pWal = pData->pWal; - bool isEmpty = ths->pLogStore->syncLogIsEmpty(ths->pLogStore); int64_t walCommitVer = walGetCommittedVer(pWal); - - if (!isEmpty && ths->commitIndex != walCommitVer) { - sNError(ths, "commit not same, wal-commit:%" PRId64 ", commit:%" PRId64 ", ignore", walCommitVer, - ths->commitIndex); - snapStart = walCommitVer + 1; - } else { - snapStart = ths->commitIndex + 1; - } + snapStart = TMAX(ths->commitIndex, walCommitVer) + 1; sNInfo(ths, "snapshot begin index is %" PRId64, snapStart); } From b36b34c1bb856451e82647d2fb19be42c81ea3b3 Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Sat, 4 Feb 2023 12:56:22 +0800 Subject: [PATCH 096/124] change transport param --- source/libs/transport/src/transCli.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/source/libs/transport/src/transCli.c b/source/libs/transport/src/transCli.c index ef64e8e040..11f3f070c0 100644 --- a/source/libs/transport/src/transCli.c +++ b/source/libs/transport/src/transCli.c @@ -778,7 +778,7 @@ static void cliSendCb(uv_write_t* req, int status) { SCliMsg* pMsg = !transQueueEmpty(&pConn->cliMsgs) ? transQueueGet(&pConn->cliMsgs, 0) : NULL; if (pMsg != NULL) { int64_t cost = taosGetTimestampUs() - pMsg->st; - if (cost > 1000) { + if (cost > 1000 * 20) { tWarn("%s conn %p send cost:%dus, send exception", CONN_GET_INST_LABEL(pConn), pConn, (int)cost); } } @@ -1049,8 +1049,8 @@ static FORCE_INLINE uint32_t cliGetIpFromFqdnCache(SHashObj* cache, char* fqdn) terrno = TAOS_SYSTEM_ERROR(errno); tError("failed to get ip from fqdn:%s since %s", fqdn, terrstr()); return addr; - } - + } + taosHashPut(cache, fqdn, strlen(fqdn), &addr, sizeof(addr)); } else { addr = *v; @@ -1132,7 +1132,7 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrd* pThrd) { cliHandleExcept(conn); return; } - + struct sockaddr_in addr; addr.sin_family = AF_INET; addr.sin_addr.s_addr = ipaddr; From 2a7732b1594db8d522e548551bfcc2fcd112e73b Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Sat, 4 Feb 2023 22:35:41 +0800 Subject: [PATCH 097/124] fix: update taos-tools 6a2d9fc for main (#19802) --- cmake/taostools_CMakeLists.txt.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/taostools_CMakeLists.txt.in b/cmake/taostools_CMakeLists.txt.in index 4da2d80fbb..37dd448b87 100644 --- a/cmake/taostools_CMakeLists.txt.in +++ b/cmake/taostools_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taos-tools ExternalProject_Add(taos-tools GIT_REPOSITORY https://github.com/taosdata/taos-tools.git - GIT_TAG c4a567b + GIT_TAG 6a2d9fc SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools" BINARY_DIR "" #BUILD_IN_SOURCE TRUE From 115160bbd0de92288007602b25d15a3bf680c55a Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Sun, 5 Feb 2023 10:11:26 +0800 Subject: [PATCH 098/124] fix: insert syntax error --- source/libs/parser/inc/parInsertUtil.h | 10 +++--- source/libs/parser/inc/parToken.h | 2 +- source/libs/parser/src/parInsertSql.c | 43 ++++++++++++++++---------- source/libs/parser/src/parTokenizer.c | 6 +++- source/libs/parser/src/parser.c | 4 +-- 5 files changed, 40 insertions(+), 25 deletions(-) diff --git a/source/libs/parser/inc/parInsertUtil.h b/source/libs/parser/inc/parInsertUtil.h index 5cc72f8692..9dd4daec3b 100644 --- a/source/libs/parser/inc/parInsertUtil.h +++ b/source/libs/parser/inc/parInsertUtil.h @@ -22,11 +22,11 @@ struct SToken; #define IS_DATA_COL_ORDERED(spd) ((spd->orderStatus) == (int8_t)ORDER_STATUS_ORDERED) -#define NEXT_TOKEN(pSql, sToken) \ - do { \ - int32_t index = 0; \ - sToken = tStrGetToken(pSql, &index, false); \ - pSql += index; \ +#define NEXT_TOKEN(pSql, sToken) \ + do { \ + int32_t index = 0; \ + sToken = tStrGetToken(pSql, &index, false, NULL); \ + pSql += index; \ } while (0) #define CHECK_CODE(expr) \ diff --git a/source/libs/parser/inc/parToken.h b/source/libs/parser/inc/parToken.h index fb4b46aa35..d539e8b37b 100644 --- a/source/libs/parser/inc/parToken.h +++ b/source/libs/parser/inc/parToken.h @@ -55,7 +55,7 @@ uint32_t tGetToken(const char *z, uint32_t *tokenType); * @param isPrevOptr * @return */ -SToken tStrGetToken(const char *str, int32_t *i, bool isPrevOptr); +SToken tStrGetToken(const char *str, int32_t *i, bool isPrevOptr, bool *pIgnoreComma); /** * check if it is a keyword or not diff --git a/source/libs/parser/src/parInsertSql.c b/source/libs/parser/src/parInsertSql.c index 1e1821842f..2e6c3a5fb1 100644 --- a/source/libs/parser/src/parInsertSql.c +++ b/source/libs/parser/src/parInsertSql.c @@ -18,16 +18,23 @@ #include "tglobal.h" #include "ttime.h" -#define NEXT_TOKEN_WITH_PREV(pSql, token) \ - do { \ - int32_t index = 0; \ - token = tStrGetToken(pSql, &index, true); \ - pSql += index; \ +#define NEXT_TOKEN_WITH_PREV(pSql, token) \ + do { \ + int32_t index = 0; \ + token = tStrGetToken(pSql, &index, true, NULL); \ + pSql += index; \ } while (0) -#define NEXT_TOKEN_KEEP_SQL(pSql, token, index) \ - do { \ - token = tStrGetToken(pSql, &index, false); \ +#define NEXT_TOKEN_WITH_PREV_EXT(pSql, token, pIgnoreComma) \ + do { \ + int32_t index = 0; \ + token = tStrGetToken(pSql, &index, true, pIgnoreComma); \ + pSql += index; \ + } while (0) + +#define NEXT_TOKEN_KEEP_SQL(pSql, token, index) \ + do { \ + token = tStrGetToken(pSql, &index, false, NULL); \ } while (0) #define NEXT_VALID_TOKEN(pSql, token) \ @@ -302,12 +309,12 @@ static int parseTime(const char** end, SToken* pToken, int16_t timePrec, int64_t * e.g., now+12a, now-5h */ index = 0; - SToken token = tStrGetToken(pTokenEnd, &index, false); + SToken token = tStrGetToken(pTokenEnd, &index, false, NULL); pTokenEnd += index; if (token.type == TK_NK_MINUS || token.type == TK_NK_PLUS) { index = 0; - SToken valueToken = tStrGetToken(pTokenEnd, &index, false); + SToken valueToken = tStrGetToken(pTokenEnd, &index, false, NULL); pTokenEnd += index; if (valueToken.n < 2) { @@ -1240,10 +1247,14 @@ static int parseOneRow(SInsertParseContext* pCxt, const char** pSql, STableDataB int32_t code = tdSRowResetBuf(pBuilder, row); // 1. set the parsed value from sql string for (int i = 0; i < pCols->numOfBound && TSDB_CODE_SUCCESS == code; ++i) { - NEXT_TOKEN_WITH_PREV(*pSql, *pToken); - SSchema* pSchema = &pSchemas[pCols->boundColumns[i]]; + const char* pOrigSql = *pSql; + bool ignoreComma = false; + NEXT_TOKEN_WITH_PREV_EXT(*pSql, *pToken, &ignoreComma); + if (ignoreComma) { + code = buildSyntaxErrMsg(&pCxt->msg, "invalid data or symbol", pOrigSql); + } - if (pToken->type == TK_NK_QUESTION) { + if (TSDB_CODE_SUCCESS == code && pToken->type == TK_NK_QUESTION) { isParseBindParam = true; if (NULL == pCxt->pComCxt->pStmtCb) { code = buildSyntaxErrMsg(&pCxt->msg, "? only used in stmt", pToken->z); @@ -1260,10 +1271,10 @@ static int parseOneRow(SInsertParseContext* pCxt, const char** pSql, STableDataB } if (TSDB_CODE_SUCCESS == code) { - param.schema = pSchema; + param.schema = &pSchemas[pCols->boundColumns[i]]; insGetSTSRowAppendInfo(pBuilder->rowType, pCols, i, ¶m.toffset, ¶m.colIdx); - code = parseValueToken(pCxt, pSql, pToken, pSchema, getTableInfo(pDataBuf->pTableMeta).precision, insMemRowAppend, - ¶m); + code = parseValueToken(pCxt, pSql, pToken, param.schema, getTableInfo(pDataBuf->pTableMeta).precision, + insMemRowAppend, ¶m); } if (TSDB_CODE_SUCCESS == code && i < pCols->numOfBound - 1) { diff --git a/source/libs/parser/src/parTokenizer.c b/source/libs/parser/src/parTokenizer.c index e62b2f0f5a..6a96ff9c50 100644 --- a/source/libs/parser/src/parTokenizer.c +++ b/source/libs/parser/src/parTokenizer.c @@ -620,7 +620,7 @@ uint32_t tGetToken(const char* z, uint32_t* tokenId) { return 0; } -SToken tStrGetToken(const char* str, int32_t* i, bool isPrevOptr) { +SToken tStrGetToken(const char* str, int32_t* i, bool isPrevOptr, bool* pIgnoreComma) { SToken t0 = {0}; // here we reach the end of sql string, null-terminated string @@ -641,6 +641,10 @@ SToken tStrGetToken(const char* str, int32_t* i, bool isPrevOptr) { return t0; } + if (NULL != pIgnoreComma && t == ',') { + *pIgnoreComma = true; + } + t = str[++(*i)]; } diff --git a/source/libs/parser/src/parser.c b/source/libs/parser/src/parser.c index cf338d63ff..47482db740 100644 --- a/source/libs/parser/src/parser.c +++ b/source/libs/parser/src/parser.c @@ -27,7 +27,7 @@ bool qIsInsertValuesSql(const char* pStr, size_t length) { const char* pSql = pStr; int32_t index = 0; - SToken t = tStrGetToken((char*)pStr, &index, false); + SToken t = tStrGetToken((char*)pStr, &index, false, NULL); if (TK_INSERT != t.type && TK_IMPORT != t.type) { return false; } @@ -35,7 +35,7 @@ bool qIsInsertValuesSql(const char* pStr, size_t length) { do { pStr += index; index = 0; - t = tStrGetToken((char*)pStr, &index, false); + t = tStrGetToken((char*)pStr, &index, false, NULL); if (TK_USING == t.type || TK_VALUES == t.type || TK_FILE == t.type) { return true; } else if (TK_SELECT == t.type) { From 121afc60470fc3005e9ad4a1da6c36418731e6e5 Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Mon, 6 Feb 2023 10:23:25 +0800 Subject: [PATCH 099/124] rm invalid param on macos --- source/os/src/osSocket.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) diff --git a/source/os/src/osSocket.c b/source/os/src/osSocket.c index 3c7a1e64ee..66cddd4dbe 100644 --- a/source/os/src/osSocket.c +++ b/source/os/src/osSocket.c @@ -1060,7 +1060,7 @@ int32_t taosCreateSocketWithTimeout(uint32_t timeout) { #if defined(WINDOWS) SOCKET fd; #else - int fd; + int fd; #endif if ((fd = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP)) == INVALID_SOCKET) { return -1; @@ -1071,11 +1071,12 @@ int32_t taosCreateSocketWithTimeout(uint32_t timeout) { return -1; } #elif defined(_TD_DARWIN_64) - uint32_t conn_timeout_ms = timeout * 1000; - if (0 != setsockopt(fd, IPPROTO_TCP, TCP_CONNECTIONTIMEOUT, (char *)&conn_timeout_ms, sizeof(conn_timeout_ms))) { - taosCloseSocketNoCheck1(fd); - return -1; - } + // invalid config + // uint32_t conn_timeout_ms = timeout * 1000; + // if (0 != setsockopt(fd, IPPROTO_TCP, TCP_CONNECTIONTIMEOUT, (char *)&conn_timeout_ms, sizeof(conn_timeout_ms))) { + // taosCloseSocketNoCheck1(fd); + // return -1; + //} #else // Linux like systems uint32_t conn_timeout_ms = timeout * 1000; if (0 != setsockopt(fd, IPPROTO_TCP, TCP_USER_TIMEOUT, (char *)&conn_timeout_ms, sizeof(conn_timeout_ms))) { From 89b1843382f1c9b37224b6e73efa1809d0ee9a7b Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 6 Feb 2023 10:59:45 +0800 Subject: [PATCH 100/124] fix(query): add null ptr check when retrieving the block info. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 1fd934066a..15432b01c8 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -4398,7 +4398,7 @@ int64_t tsdbGetNumOfRowsInMemTable(STsdbReader* pReader) { STableBlockScanInfo* pBlockScanInfo = *(STableBlockScanInfo**)pStatus->pTableIter; STbData* d = NULL; - if (pReader->pTsdb->mem != NULL) { + if (pReader->pReadSnap->pMem != NULL) { d = tsdbGetTbDataFromMemTable(pReader->pReadSnap->pMem, pReader->suid, pBlockScanInfo->uid); if (d != NULL) { rows += tsdbGetNRowsInTbData(d); @@ -4406,7 +4406,7 @@ int64_t tsdbGetNumOfRowsInMemTable(STsdbReader* pReader) { } STbData* di = NULL; - if (pReader->pTsdb->imem != NULL) { + if (pReader->pReadSnap->pIMem != NULL) { di = tsdbGetTbDataFromMemTable(pReader->pReadSnap->pIMem, pReader->suid, pBlockScanInfo->uid); if (di != NULL) { rows += tsdbGetNRowsInTbData(di); From 2b4adceb930c6939bbad7569f30e02d6d892579a Mon Sep 17 00:00:00 2001 From: Xiaoyu Wang Date: Mon, 6 Feb 2023 11:07:51 +0800 Subject: [PATCH 101/124] fix: insert syntax error --- source/libs/parser/src/parInsertSql.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) diff --git a/source/libs/parser/src/parInsertSql.c b/source/libs/parser/src/parInsertSql.c index 2e6c3a5fb1..abee9416fa 100644 --- a/source/libs/parser/src/parInsertSql.c +++ b/source/libs/parser/src/parInsertSql.c @@ -1252,25 +1252,26 @@ static int parseOneRow(SInsertParseContext* pCxt, const char** pSql, STableDataB NEXT_TOKEN_WITH_PREV_EXT(*pSql, *pToken, &ignoreComma); if (ignoreComma) { code = buildSyntaxErrMsg(&pCxt->msg, "invalid data or symbol", pOrigSql); + break; } - if (TSDB_CODE_SUCCESS == code && pToken->type == TK_NK_QUESTION) { + if (pToken->type == TK_NK_QUESTION) { isParseBindParam = true; if (NULL == pCxt->pComCxt->pStmtCb) { code = buildSyntaxErrMsg(&pCxt->msg, "? only used in stmt", pToken->z); + break; + } + } else { + if (TK_NK_RP == pToken->type) { + code = generateSyntaxErrMsg(&pCxt->msg, TSDB_CODE_PAR_INVALID_COLUMNS_NUM); + break; } - continue; - } - if (TSDB_CODE_SUCCESS == code && TK_NK_RP == pToken->type) { - code = generateSyntaxErrMsg(&pCxt->msg, TSDB_CODE_PAR_INVALID_COLUMNS_NUM); - } + if (isParseBindParam) { + code = buildInvalidOperationMsg(&pCxt->msg, "no mix usage for ? and values"); + break; + } - if (TSDB_CODE_SUCCESS == code && isParseBindParam) { - code = buildInvalidOperationMsg(&pCxt->msg, "no mix usage for ? and values"); - } - - if (TSDB_CODE_SUCCESS == code) { param.schema = &pSchemas[pCols->boundColumns[i]]; insGetSTSRowAppendInfo(pBuilder->rowType, pCols, i, ¶m.toffset, ¶m.colIdx); code = parseValueToken(pCxt, pSql, pToken, param.schema, getTableInfo(pDataBuf->pTableMeta).precision, From e615053f4a90e7aa8fe569dfe2ab7823ece30858 Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Mon, 6 Feb 2023 10:40:33 +0800 Subject: [PATCH 102/124] enh: stream number limit --- include/util/taoserror.h | 1 + source/dnode/mnode/impl/src/mndStream.c | 30 ++++ source/util/src/terror.c | 1 + tests/script/tsim/stream/drop_stream.sim | 220 +++++++++++------------ 4 files changed, 142 insertions(+), 110 deletions(-) diff --git a/include/util/taoserror.h b/include/util/taoserror.h index 51f336c3aa..21d299f28d 100644 --- a/include/util/taoserror.h +++ b/include/util/taoserror.h @@ -355,6 +355,7 @@ int32_t* taosGetErrno(); #define TSDB_CODE_MND_STREAM_MUST_BE_DELETED TAOS_DEF_ERROR_CODE(0, 0x03F3) #define TSDB_CODE_MND_STREAM_TASK_DROPPED TAOS_DEF_ERROR_CODE(0, 0x03F4) #define TSDB_CODE_MND_MULTI_REPLICA_SOURCE_DB TAOS_DEF_ERROR_CODE(0, 0x03F5) +#define TSDB_CODE_MND_TOO_MANY_STREAMS TAOS_DEF_ERROR_CODE(0, 0x03F6) // mnode-sma #define TSDB_CODE_MND_SMA_ALREADY_EXIST TAOS_DEF_ERROR_CODE(0, 0x0480) diff --git a/source/dnode/mnode/impl/src/mndStream.c b/source/dnode/mnode/impl/src/mndStream.c index 83ed6eea78..cb6b54ceed 100644 --- a/source/dnode/mnode/impl/src/mndStream.c +++ b/source/dnode/mnode/impl/src/mndStream.c @@ -31,6 +31,8 @@ #define MND_STREAM_VER_NUMBER 2 #define MND_STREAM_RESERVE_SIZE 64 +#define MND_STREAM_MAX_NUM 10 + static int32_t mndStreamActionInsert(SSdb *pSdb, SStreamObj *pStream); static int32_t mndStreamActionDelete(SSdb *pSdb, SStreamObj *pStream); static int32_t mndStreamActionUpdate(SSdb *pSdb, SStreamObj *pStream, SStreamObj *pNewStream); @@ -625,6 +627,34 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) { goto _OVER; } + { + int32_t numOfStream = 0; + + SStreamObj *pStream = NULL; + void *pIter = NULL; + + while (1) { + pIter = sdbFetch(pMnode->pSdb, SDB_STREAM, pIter, (void **)&pStream); + if (pIter == NULL) { + if (numOfStream > MND_STREAM_MAX_NUM) { + mError("too many streams, no more than 10 for each database"); + terrno = TSDB_CODE_MND_TOO_MANY_STREAMS; + goto _OVER; + } + break; + } + + if (pStream->sourceDbUid == streamObj.sourceDbUid) { + ++numOfStream; + } + if (numOfStream > MND_STREAM_MAX_NUM) { + mError("too many streams, no more than 10 for each database"); + terrno = TSDB_CODE_MND_TOO_MANY_STREAMS; + goto _OVER; + } + } + } + pDb = mndAcquireDb(pMnode, streamObj.sourceDb); if (pDb->cfg.replications != 1) { mError("stream source db must have only 1 replica, but %s has %d", pDb->name, pDb->cfg.replications); diff --git a/source/util/src/terror.c b/source/util/src/terror.c index 14c3df38a7..d610d26f97 100644 --- a/source/util/src/terror.c +++ b/source/util/src/terror.c @@ -287,6 +287,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_MND_STREAM_NOT_EXIST, "Stream not exist") TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_STREAM_OPTION, "Invalid stream option") TAOS_DEFINE_ERROR(TSDB_CODE_MND_STREAM_MUST_BE_DELETED, "Stream must be dropped first") TAOS_DEFINE_ERROR(TSDB_CODE_MND_MULTI_REPLICA_SOURCE_DB, "Stream temporarily does not support source db having replica > 1") +TAOS_DEFINE_ERROR(TSDB_CODE_MND_TOO_MANY_STREAMS, "Too many streams") // mnode-sma TAOS_DEFINE_ERROR(TSDB_CODE_MND_SMA_ALREADY_EXIST, "SMA already exists") diff --git a/tests/script/tsim/stream/drop_stream.sim b/tests/script/tsim/stream/drop_stream.sim index 1a474bd9ae..8d193bc0a0 100644 --- a/tests/script/tsim/stream/drop_stream.sim +++ b/tests/script/tsim/stream/drop_stream.sim @@ -58,59 +58,59 @@ sql create stream stb_atan_stream trigger at_once into output_atan_stb as select sql create stream ctb_atan_stream trigger at_once into output_atan_ctb as select ts, atan(c1), atan(c2), c3 from scalar_ct1; sql create stream tb_atan_stream trigger at_once into output_atan_tb as select ts, atan(c1), atan(c2), c3 from scalar_tb; sql create stream stb_ceil_stream trigger at_once into output_ceil_stb as select ts, ceil(c1), ceil(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_ceil_stream trigger at_once into output_ceil_ctb as select ts, ceil(c1), ceil(c2), c3 from scalar_ct1; -sql create stream tb_ceil_stream trigger at_once into output_ceil_tb as select ts, ceil(c1), ceil(c2), c3 from scalar_tb; -sql create stream stb_cos_stream trigger at_once into output_cos_stb as select ts, cos(c1), cos(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_cos_stream trigger at_once into output_cos_ctb as select ts, cos(c1), cos(c2), c3 from scalar_ct1; -sql create stream tb_cos_stream trigger at_once into output_cos_tb as select ts, cos(c1), cos(c2), c3 from scalar_tb; -sql create stream stb_floor_stream trigger at_once into output_floor_stb as select ts, floor(c1), floor(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_floor_stream trigger at_once into output_floor_ctb as select ts, floor(c1), floor(c2), c3 from scalar_ct1; -sql create stream tb_floor_stream trigger at_once into output_floor_tb as select ts, floor(c1), floor(c2), c3 from scalar_tb; -sql create stream stb_log_stream trigger at_once into output_log_stb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_stb partition by tbname; -sql create stream ctb_log_stream trigger at_once into output_log_ctb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_ct1; -sql create stream tb_log_stream trigger at_once into output_log_tb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_tb; -sql create stream stb_pow_stream trigger at_once into output_pow_stb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_stb partition by tbname; -sql create stream ctb_pow_stream trigger at_once into output_pow_ctb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_ct1; -sql create stream tb_pow_stream trigger at_once into output_pow_tb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_tb; -sql create stream stb_round_stream trigger at_once into output_round_stb as select ts, round(c1), round(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_round_stream trigger at_once into output_round_ctb as select ts, round(c1), round(c2), c3 from scalar_ct1; -sql create stream tb_round_stream trigger at_once into output_round_tb as select ts, round(c1), round(c2), c3 from scalar_tb; -sql create stream stb_sin_stream trigger at_once into output_sin_stb as select ts, sin(c1), sin(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_sin_stream trigger at_once into output_sin_ctb as select ts, sin(c1), sin(c2), c3 from scalar_ct1; -sql create stream tb_sin_stream trigger at_once into output_sin_tb as select ts, sin(c1), sin(c2), c3 from scalar_tb; -sql create stream stb_sqrt_stream trigger at_once into output_sqrt_stb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_sqrt_stream trigger at_once into output_sqrt_ctb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_ct1; -sql create stream tb_sqrt_stream trigger at_once into output_sqrt_tb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_tb; -sql create stream stb_tan_stream trigger at_once into output_tan_stb as select ts, tan(c1), tan(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_tan_stream trigger at_once into output_tan_ctb as select ts, tan(c1), tan(c2), c3 from scalar_ct1; -sql create stream tb_tan_stream trigger at_once into output_tan_tb as select ts, tan(c1), tan(c2), c3 from scalar_tb; -sql create stream stb_char_length_stream into output_char_length_stb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_stb partition by tbname; -sql create stream ctb_char_length_stream into output_char_length_ctb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_ct1; -sql create stream tb_char_length_stream into output_char_length_tb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_tb; -sql create stream stb_concat_stream into output_concat_stb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_stb partition by tbname; -sql create stream ctb_concat_stream into output_concat_ctb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_ct1; -sql create stream tb_concat_stream into output_concat_tb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_tb; -sql create stream stb_concat_ws_stream into output_concat_ws_stb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_stb partition by tbname; -sql create stream ctb_concat_ws_stream into output_concat_ws_ctb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_ct1; -sql create stream tb_concat_ws_stream into output_concat_ws_tb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_tb; -sql create stream stb_length_stream into output_length_stb as select ts, length(c3), length(c4), length(c5) from scalar_stb partition by tbname; -sql create stream ctb_length_stream into output_length_ctb as select ts, length(c3), length(c4), length(c5) from scalar_ct1; -sql create stream tb_length_stream into output_length_tb as select ts, length(c3), length(c4), length(c5) from scalar_tb; -sql create stream stb_lower_stream into output_lower_stb as select ts, lower(c3), lower(c4), lower(c5) from scalar_stb partition by tbname; -sql create stream ctb_lower_stream into output_lower_ctb as select ts, lower(c3), lower(c4), lower(c5) from scalar_ct1; -sql create stream tb_lower_stream into output_lower_tb as select ts, lower(c3), lower(c4), lower(c5) from scalar_tb; -sql create stream stb_ltrim_stream into output_ltrim_stb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_stb partition by tbname; -sql create stream ctb_ltrim_stream into output_ltrim_ctb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_ct1; -sql create stream tb_ltrim_stream into output_ltrim_tb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_tb; -sql create stream stb_rtrim_stream into output_rtrim_stb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_stb partition by tbname; -sql create stream ctb_rtrim_stream into output_rtrim_ctb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_ct1; -sql create stream tb_rtrim_stream into output_rtrim_tb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_tb; -sql create stream stb_substr_stream into output_substr_stb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_stb partition by tbname; -sql create stream ctb_substr_stream into output_substr_ctb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_ct1; -sql create stream tb_substr_stream into output_substr_tb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_tb; -sql create stream stb_upper_stream into output_upper_stb as select ts, upper(c3), upper(c4), upper(c5) from scalar_stb partition by tbname; -sql create stream ctb_upper_stream into output_upper_ctb as select ts, upper(c3), upper(c4), upper(c5) from scalar_ct1; -sql create stream tb_upper_stream into output_upper_tb as select ts, upper(c3), upper(c4), upper(c5) from scalar_tb; +# sql create stream ctb_ceil_stream trigger at_once into output_ceil_ctb as select ts, ceil(c1), ceil(c2), c3 from scalar_ct1; +# sql create stream tb_ceil_stream trigger at_once into output_ceil_tb as select ts, ceil(c1), ceil(c2), c3 from scalar_tb; +# sql create stream stb_cos_stream trigger at_once into output_cos_stb as select ts, cos(c1), cos(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_cos_stream trigger at_once into output_cos_ctb as select ts, cos(c1), cos(c2), c3 from scalar_ct1; +# sql create stream tb_cos_stream trigger at_once into output_cos_tb as select ts, cos(c1), cos(c2), c3 from scalar_tb; +# sql create stream stb_floor_stream trigger at_once into output_floor_stb as select ts, floor(c1), floor(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_floor_stream trigger at_once into output_floor_ctb as select ts, floor(c1), floor(c2), c3 from scalar_ct1; +# sql create stream tb_floor_stream trigger at_once into output_floor_tb as select ts, floor(c1), floor(c2), c3 from scalar_tb; +# sql create stream stb_log_stream trigger at_once into output_log_stb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_log_stream trigger at_once into output_log_ctb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_ct1; +# sql create stream tb_log_stream trigger at_once into output_log_tb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_tb; +# sql create stream stb_pow_stream trigger at_once into output_pow_stb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_pow_stream trigger at_once into output_pow_ctb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_ct1; +# sql create stream tb_pow_stream trigger at_once into output_pow_tb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_tb; +# sql create stream stb_round_stream trigger at_once into output_round_stb as select ts, round(c1), round(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_round_stream trigger at_once into output_round_ctb as select ts, round(c1), round(c2), c3 from scalar_ct1; +# sql create stream tb_round_stream trigger at_once into output_round_tb as select ts, round(c1), round(c2), c3 from scalar_tb; +# sql create stream stb_sin_stream trigger at_once into output_sin_stb as select ts, sin(c1), sin(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_sin_stream trigger at_once into output_sin_ctb as select ts, sin(c1), sin(c2), c3 from scalar_ct1; +# sql create stream tb_sin_stream trigger at_once into output_sin_tb as select ts, sin(c1), sin(c2), c3 from scalar_tb; +# sql create stream stb_sqrt_stream trigger at_once into output_sqrt_stb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_sqrt_stream trigger at_once into output_sqrt_ctb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_ct1; +# sql create stream tb_sqrt_stream trigger at_once into output_sqrt_tb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_tb; +# sql create stream stb_tan_stream trigger at_once into output_tan_stb as select ts, tan(c1), tan(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_tan_stream trigger at_once into output_tan_ctb as select ts, tan(c1), tan(c2), c3 from scalar_ct1; +# sql create stream tb_tan_stream trigger at_once into output_tan_tb as select ts, tan(c1), tan(c2), c3 from scalar_tb; +# sql create stream stb_char_length_stream into output_char_length_stb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_stb partition by tbname; +# sql create stream ctb_char_length_stream into output_char_length_ctb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_ct1; +# sql create stream tb_char_length_stream into output_char_length_tb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_tb; +# sql create stream stb_concat_stream into output_concat_stb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_stb partition by tbname; +# sql create stream ctb_concat_stream into output_concat_ctb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_ct1; +# sql create stream tb_concat_stream into output_concat_tb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_tb; +# sql create stream stb_concat_ws_stream into output_concat_ws_stb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_stb partition by tbname; +# sql create stream ctb_concat_ws_stream into output_concat_ws_ctb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_ct1; +# sql create stream tb_concat_ws_stream into output_concat_ws_tb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_tb; +# sql create stream stb_length_stream into output_length_stb as select ts, length(c3), length(c4), length(c5) from scalar_stb partition by tbname; +# sql create stream ctb_length_stream into output_length_ctb as select ts, length(c3), length(c4), length(c5) from scalar_ct1; +# sql create stream tb_length_stream into output_length_tb as select ts, length(c3), length(c4), length(c5) from scalar_tb; +# sql create stream stb_lower_stream into output_lower_stb as select ts, lower(c3), lower(c4), lower(c5) from scalar_stb partition by tbname; +# sql create stream ctb_lower_stream into output_lower_ctb as select ts, lower(c3), lower(c4), lower(c5) from scalar_ct1; +# sql create stream tb_lower_stream into output_lower_tb as select ts, lower(c3), lower(c4), lower(c5) from scalar_tb; +# sql create stream stb_ltrim_stream into output_ltrim_stb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_stb partition by tbname; +# sql create stream ctb_ltrim_stream into output_ltrim_ctb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_ct1; +# sql create stream tb_ltrim_stream into output_ltrim_tb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_tb; +# sql create stream stb_rtrim_stream into output_rtrim_stb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_stb partition by tbname; +# sql create stream ctb_rtrim_stream into output_rtrim_ctb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_ct1; +# sql create stream tb_rtrim_stream into output_rtrim_tb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_tb; +# sql create stream stb_substr_stream into output_substr_stb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_stb partition by tbname; +# sql create stream ctb_substr_stream into output_substr_ctb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_ct1; +# sql create stream tb_substr_stream into output_substr_tb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_tb; +# sql create stream stb_upper_stream into output_upper_stb as select ts, upper(c3), upper(c4), upper(c5) from scalar_stb partition by tbname; +# sql create stream ctb_upper_stream into output_upper_ctb as select ts, upper(c3), upper(c4), upper(c5) from scalar_ct1; +# sql create stream tb_upper_stream into output_upper_tb as select ts, upper(c3), upper(c4), upper(c5) from scalar_tb; sql insert into scalar_ct1 values (1656668180503, 100, 100.1, "beijing", "taos", "Taos"); sql insert into scalar_ct1 values (1656668180503+1s, -50, -50.1, "tianjin", "taosdata", "Taosdata"); sql insert into scalar_ct1 values (1656668180503+2s, 0, Null, "hebei", "TDengine", Null); @@ -146,62 +146,62 @@ sql create stream stb_asin_stream trigger at_once into output_asin_stb as select sql create stream ctb_asin_stream trigger at_once into output_asin_ctb as select ts, asin(c1), asin(c2), c3 from scalar_ct1; sql create stream tb_asin_stream trigger at_once into output_asin_tb as select ts, asin(c1), asin(c2), c3 from scalar_tb; sql create stream stb_atan_stream trigger at_once into output_atan_stb as select ts, atan(c1), atan(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_atan_stream trigger at_once into output_atan_ctb as select ts, atan(c1), atan(c2), c3 from scalar_ct1; -sql create stream tb_atan_stream trigger at_once into output_atan_tb as select ts, atan(c1), atan(c2), c3 from scalar_tb; -sql create stream stb_ceil_stream trigger at_once into output_ceil_stb as select ts, ceil(c1), ceil(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_ceil_stream trigger at_once into output_ceil_ctb as select ts, ceil(c1), ceil(c2), c3 from scalar_ct1; -sql create stream tb_ceil_stream trigger at_once into output_ceil_tb as select ts, ceil(c1), ceil(c2), c3 from scalar_tb; -sql create stream stb_cos_stream trigger at_once into output_cos_stb as select ts, cos(c1), cos(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_cos_stream trigger at_once into output_cos_ctb as select ts, cos(c1), cos(c2), c3 from scalar_ct1; -sql create stream tb_cos_stream trigger at_once into output_cos_tb as select ts, cos(c1), cos(c2), c3 from scalar_tb; -sql create stream stb_floor_stream trigger at_once into output_floor_stb as select ts, floor(c1), floor(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_floor_stream trigger at_once into output_floor_ctb as select ts, floor(c1), floor(c2), c3 from scalar_ct1; -sql create stream tb_floor_stream trigger at_once into output_floor_tb as select ts, floor(c1), floor(c2), c3 from scalar_tb; -sql create stream stb_log_stream trigger at_once into output_log_stb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_stb partition by tbname; -sql create stream ctb_log_stream trigger at_once into output_log_ctb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_ct1; -sql create stream tb_log_stream trigger at_once into output_log_tb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_tb; -sql create stream stb_pow_stream trigger at_once into output_pow_stb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_stb partition by tbname; -sql create stream ctb_pow_stream trigger at_once into output_pow_ctb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_ct1; -sql create stream tb_pow_stream trigger at_once into output_pow_tb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_tb; -sql create stream stb_round_stream trigger at_once into output_round_stb as select ts, round(c1), round(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_round_stream trigger at_once into output_round_ctb as select ts, round(c1), round(c2), c3 from scalar_ct1; -sql create stream tb_round_stream trigger at_once into output_round_tb as select ts, round(c1), round(c2), c3 from scalar_tb; -sql create stream stb_sin_stream trigger at_once into output_sin_stb as select ts, sin(c1), sin(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_sin_stream trigger at_once into output_sin_ctb as select ts, sin(c1), sin(c2), c3 from scalar_ct1; -sql create stream tb_sin_stream trigger at_once into output_sin_tb as select ts, sin(c1), sin(c2), c3 from scalar_tb; -sql create stream stb_sqrt_stream trigger at_once into output_sqrt_stb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_sqrt_stream trigger at_once into output_sqrt_ctb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_ct1; -sql create stream tb_sqrt_stream trigger at_once into output_sqrt_tb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_tb; -sql create stream stb_tan_stream trigger at_once into output_tan_stb as select ts, tan(c1), tan(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_tan_stream trigger at_once into output_tan_ctb as select ts, tan(c1), tan(c2), c3 from scalar_ct1; -sql create stream tb_tan_stream trigger at_once into output_tan_tb as select ts, tan(c1), tan(c2), c3 from scalar_tb; -sql create stream stb_char_length_stream into output_char_length_stb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_stb partition by tbname; -sql create stream ctb_char_length_stream into output_char_length_ctb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_ct1; -sql create stream tb_char_length_stream into output_char_length_tb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_tb; -sql create stream stb_concat_stream into output_concat_stb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_stb partition by tbname; -sql create stream ctb_concat_stream into output_concat_ctb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_ct1; -sql create stream tb_concat_stream into output_concat_tb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_tb; -sql create stream stb_concat_ws_stream into output_concat_ws_stb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_stb partition by tbname; -sql create stream ctb_concat_ws_stream into output_concat_ws_ctb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_ct1; -sql create stream tb_concat_ws_stream into output_concat_ws_tb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_tb; -sql create stream stb_length_stream into output_length_stb as select ts, length(c3), length(c4), length(c5) from scalar_stb partition by tbname; -sql create stream ctb_length_stream into output_length_ctb as select ts, length(c3), length(c4), length(c5) from scalar_ct1; -sql create stream tb_length_stream into output_length_tb as select ts, length(c3), length(c4), length(c5) from scalar_tb; -sql create stream stb_lower_stream into output_lower_stb as select ts, lower(c3), lower(c4), lower(c5) from scalar_stb partition by tbname; -sql create stream ctb_lower_stream into output_lower_ctb as select ts, lower(c3), lower(c4), lower(c5) from scalar_ct1; -sql create stream tb_lower_stream into output_lower_tb as select ts, lower(c3), lower(c4), lower(c5) from scalar_tb; -sql create stream stb_ltrim_stream into output_ltrim_stb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_stb partition by tbname; -sql create stream ctb_ltrim_stream into output_ltrim_ctb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_ct1; -sql create stream tb_ltrim_stream into output_ltrim_tb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_tb; -sql create stream stb_rtrim_stream into output_rtrim_stb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_stb partition by tbname; -sql create stream ctb_rtrim_stream into output_rtrim_ctb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_ct1; -sql create stream tb_rtrim_stream into output_rtrim_tb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_tb; -sql create stream stb_substr_stream into output_substr_stb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_stb partition by tbname; -sql create stream ctb_substr_stream into output_substr_ctb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_ct1; -sql create stream tb_substr_stream into output_substr_tb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_tb; -sql create stream stb_upper_stream into output_upper_stb as select ts, upper(c3), upper(c4), upper(c5) from scalar_stb partition by tbname; -sql create stream ctb_upper_stream into output_upper_ctb as select ts, upper(c3), upper(c4), upper(c5) from scalar_ct1; -sql create stream tb_upper_stream into output_upper_tb as select ts, upper(c3), upper(c4), upper(c5) from scalar_tb; +# sql create stream ctb_atan_stream trigger at_once into output_atan_ctb as select ts, atan(c1), atan(c2), c3 from scalar_ct1; +# sql create stream tb_atan_stream trigger at_once into output_atan_tb as select ts, atan(c1), atan(c2), c3 from scalar_tb; +# sql create stream stb_ceil_stream trigger at_once into output_ceil_stb as select ts, ceil(c1), ceil(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_ceil_stream trigger at_once into output_ceil_ctb as select ts, ceil(c1), ceil(c2), c3 from scalar_ct1; +# sql create stream tb_ceil_stream trigger at_once into output_ceil_tb as select ts, ceil(c1), ceil(c2), c3 from scalar_tb; +# sql create stream stb_cos_stream trigger at_once into output_cos_stb as select ts, cos(c1), cos(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_cos_stream trigger at_once into output_cos_ctb as select ts, cos(c1), cos(c2), c3 from scalar_ct1; +# sql create stream tb_cos_stream trigger at_once into output_cos_tb as select ts, cos(c1), cos(c2), c3 from scalar_tb; +# sql create stream stb_floor_stream trigger at_once into output_floor_stb as select ts, floor(c1), floor(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_floor_stream trigger at_once into output_floor_ctb as select ts, floor(c1), floor(c2), c3 from scalar_ct1; +# sql create stream tb_floor_stream trigger at_once into output_floor_tb as select ts, floor(c1), floor(c2), c3 from scalar_tb; +# sql create stream stb_log_stream trigger at_once into output_log_stb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_log_stream trigger at_once into output_log_ctb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_ct1; +# sql create stream tb_log_stream trigger at_once into output_log_tb as select ts, log(c1, 2), log(c2, 2), c3 from scalar_tb; +# sql create stream stb_pow_stream trigger at_once into output_pow_stb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_pow_stream trigger at_once into output_pow_ctb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_ct1; +# sql create stream tb_pow_stream trigger at_once into output_pow_tb as select ts, pow(c1, 2), pow(c2, 2), c3 from scalar_tb; +# sql create stream stb_round_stream trigger at_once into output_round_stb as select ts, round(c1), round(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_round_stream trigger at_once into output_round_ctb as select ts, round(c1), round(c2), c3 from scalar_ct1; +# sql create stream tb_round_stream trigger at_once into output_round_tb as select ts, round(c1), round(c2), c3 from scalar_tb; +# sql create stream stb_sin_stream trigger at_once into output_sin_stb as select ts, sin(c1), sin(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_sin_stream trigger at_once into output_sin_ctb as select ts, sin(c1), sin(c2), c3 from scalar_ct1; +# sql create stream tb_sin_stream trigger at_once into output_sin_tb as select ts, sin(c1), sin(c2), c3 from scalar_tb; +# sql create stream stb_sqrt_stream trigger at_once into output_sqrt_stb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_sqrt_stream trigger at_once into output_sqrt_ctb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_ct1; +# sql create stream tb_sqrt_stream trigger at_once into output_sqrt_tb as select ts, sqrt(c1), sqrt(c2), c3 from scalar_tb; +# sql create stream stb_tan_stream trigger at_once into output_tan_stb as select ts, tan(c1), tan(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_tan_stream trigger at_once into output_tan_ctb as select ts, tan(c1), tan(c2), c3 from scalar_ct1; +# sql create stream tb_tan_stream trigger at_once into output_tan_tb as select ts, tan(c1), tan(c2), c3 from scalar_tb; +# sql create stream stb_char_length_stream into output_char_length_stb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_stb partition by tbname; +# sql create stream ctb_char_length_stream into output_char_length_ctb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_ct1; +# sql create stream tb_char_length_stream into output_char_length_tb as select ts, char_length(c3), char_length(c4), char_length(c5) from scalar_tb; +# sql create stream stb_concat_stream into output_concat_stb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_stb partition by tbname; +# sql create stream ctb_concat_stream into output_concat_ctb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_ct1; +# sql create stream tb_concat_stream into output_concat_tb as select ts, concat(c3, c4), concat(c3, c5), concat(c4, c5), concat(c3, c4, c5) from scalar_tb; +# sql create stream stb_concat_ws_stream into output_concat_ws_stb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_stb partition by tbname; +# sql create stream ctb_concat_ws_stream into output_concat_ws_ctb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_ct1; +# sql create stream tb_concat_ws_stream into output_concat_ws_tb as select ts, concat_ws("aND", c3, c4), concat_ws("and", c3, c5), concat_ws("And", c4, c5), concat_ws("AND", c3, c4, c5) from scalar_tb; +# sql create stream stb_length_stream into output_length_stb as select ts, length(c3), length(c4), length(c5) from scalar_stb partition by tbname; +# sql create stream ctb_length_stream into output_length_ctb as select ts, length(c3), length(c4), length(c5) from scalar_ct1; +# sql create stream tb_length_stream into output_length_tb as select ts, length(c3), length(c4), length(c5) from scalar_tb; +# sql create stream stb_lower_stream into output_lower_stb as select ts, lower(c3), lower(c4), lower(c5) from scalar_stb partition by tbname; +# sql create stream ctb_lower_stream into output_lower_ctb as select ts, lower(c3), lower(c4), lower(c5) from scalar_ct1; +# sql create stream tb_lower_stream into output_lower_tb as select ts, lower(c3), lower(c4), lower(c5) from scalar_tb; +# sql create stream stb_ltrim_stream into output_ltrim_stb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_stb partition by tbname; +# sql create stream ctb_ltrim_stream into output_ltrim_ctb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_ct1; +# sql create stream tb_ltrim_stream into output_ltrim_tb as select ts, ltrim(c3), ltrim(c4), ltrim(c5) from scalar_tb; +# sql create stream stb_rtrim_stream into output_rtrim_stb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_stb partition by tbname; +# sql create stream ctb_rtrim_stream into output_rtrim_ctb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_ct1; +# sql create stream tb_rtrim_stream into output_rtrim_tb as select ts, rtrim(c3), rtrim(c4), rtrim(c5) from scalar_tb; +# sql create stream stb_substr_stream into output_substr_stb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_stb partition by tbname; +# sql create stream ctb_substr_stream into output_substr_ctb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_ct1; +# sql create stream tb_substr_stream into output_substr_tb as select ts, substr(c3, 2), substr(c3, 2, 2), substr(c4, 5, 1), substr(c5, 3, 4) from scalar_tb; +# sql create stream stb_upper_stream into output_upper_stb as select ts, upper(c3), upper(c4), upper(c5) from scalar_stb partition by tbname; +# sql create stream ctb_upper_stream into output_upper_ctb as select ts, upper(c3), upper(c4), upper(c5) from scalar_ct1; +# sql create stream tb_upper_stream into output_upper_tb as select ts, upper(c3), upper(c4), upper(c5) from scalar_tb; sql insert into scalar_ct1 values (1656668180503, 100, 100.1, "beijing", "taos", "Taos"); sql insert into scalar_ct1 values (1656668180503+1s, -50, -50.1, "tianjin", "taosdata", "Taosdata"); sql insert into scalar_ct1 values (1656668180503+2s, 0, Null, "hebei", "TDengine", Null); @@ -273,4 +273,4 @@ print ========== step7 system sh/exec.sh -n dnode1 -s stop -x SIGINT system sh/exec.sh -n dnode2 -s stop -x SIGINT system sh/exec.sh -n dnode3 -s stop -x SIGINT -system sh/exec.sh -n dnode4 -s stop -x SIGINT \ No newline at end of file +system sh/exec.sh -n dnode4 -s stop -x SIGINT From 4939f5470723cc354b20a283ed155c07440c8bf5 Mon Sep 17 00:00:00 2001 From: yihaoDeng Date: Mon, 6 Feb 2023 12:15:45 +0800 Subject: [PATCH 103/124] fix invalid code --- source/libs/executor/src/groupoperator.c | 27 ++++++++++++++++++------ 1 file changed, 21 insertions(+), 6 deletions(-) diff --git a/source/libs/executor/src/groupoperator.c b/source/libs/executor/src/groupoperator.c index bf4b9a2599..5febfbb2f3 100644 --- a/source/libs/executor/src/groupoperator.c +++ b/source/libs/executor/src/groupoperator.c @@ -492,8 +492,8 @@ _error: static void doHashPartition(SOperatorInfo* pOperator, SSDataBlock* pBlock) { SPartitionOperatorInfo* pInfo = pOperator->info; - SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo; - + SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo; + for (int32_t j = 0; j < pBlock->info.rows; ++j) { recordNewGroupKeys(pInfo->pGroupCols, pInfo->pGroupColVals, pBlock, j); int32_t len = buildGroupKeys(pInfo->keyBuf, pInfo->pGroupColVals); @@ -690,8 +690,8 @@ static int compareDataGroupInfo(const void* group1, const void* group2) { static SSDataBlock* buildPartitionResult(SOperatorInfo* pOperator) { SPartitionOperatorInfo* pInfo = pOperator->info; - SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo; - + SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo; + SDataGroupInfo* pGroupInfo = (pInfo->groupIndex != -1) ? taosArrayGet(pInfo->sortedGroupArray, pInfo->groupIndex) : NULL; if (pInfo->groupIndex == -1 || pInfo->pageIndex >= taosArrayGetSize(pGroupInfo->pPageList)) { @@ -713,7 +713,7 @@ static SSDataBlock* buildPartitionResult(SOperatorInfo* pOperator) { qError("failed to get buffer, code:%s, %s", tstrerror(terrno), GET_TASKID(pTaskInfo)); T_LONG_JMP(pTaskInfo->env, terrno); } - + blockDataEnsureCapacity(pInfo->binfo.pRes, pInfo->rowCapacity); blockDataFromBuf1(pInfo->binfo.pRes, page, pInfo->rowCapacity); @@ -829,6 +829,8 @@ SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartition SPartitionOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SPartitionOperatorInfo)); SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo)); if (pInfo == NULL || pOperator == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + pTaskInfo->code = terrno; goto _error; } @@ -841,6 +843,8 @@ SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartition SExprInfo* pExprInfo1 = createExprInfo(pPartNode->pExprs, NULL, &num); int32_t code = initExprSupp(&pInfo->scalarSup, pExprInfo1, num); if (code != TSDB_CODE_SUCCESS) { + terrno = code; + pTaskInfo->code = terrno; goto _error; } } @@ -848,6 +852,8 @@ SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartition _hash_fn_t hashFn = taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY); pInfo->pGroupSet = taosHashInit(100, hashFn, false, HASH_NO_LOCK); if (pInfo->pGroupSet == NULL) { + terrno = TSDB_CODE_OUT_OF_MEMORY; + pTaskInfo->code = terrno; goto _error; } @@ -866,6 +872,8 @@ SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartition int32_t code = createDiskbasedBuf(&pInfo->pBuf, defaultPgsz, defaultBufsz, pTaskInfo->id.str, tsTempDir); if (code != TSDB_CODE_SUCCESS) { + terrno = code; + pTaskInfo->code = code; goto _error; } @@ -873,6 +881,8 @@ SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartition pInfo->columnOffset = setupColumnOffset(pInfo->binfo.pRes, pInfo->rowCapacity); code = initGroupOptrInfo(&pInfo->pGroupColVals, &pInfo->groupKeyLen, &pInfo->keyBuf, pInfo->pGroupCols); if (code != TSDB_CODE_SUCCESS) { + terrno = code; + pTaskInfo->code = code; goto _error; } @@ -885,10 +895,15 @@ SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartition createOperatorFpSet(optrDummyOpenFn, hashPartition, NULL, destroyPartitionOperatorInfo, optrDefaultBufFn, NULL); code = appendDownstream(pOperator, &downstream, 1); + if (code != TSDB_CODE_SUCCESS) { + terrno = code; + pTaskInfo->code = code; + goto _error; + } + return pOperator; _error: - pTaskInfo->code = TSDB_CODE_OUT_OF_MEMORY; if (pInfo != NULL) { destroyPartitionOperatorInfo(pInfo); } From 24cdf0e7585ab94cf30ea37cf831d15de1f37874 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 6 Feb 2023 13:40:51 +0800 Subject: [PATCH 104/124] enh(query): add check for limit reach status. --- source/libs/executor/src/projectoperator.c | 1 + source/libs/executor/src/scanoperator.c | 16 +++++++++++----- 2 files changed, 12 insertions(+), 5 deletions(-) diff --git a/source/libs/executor/src/projectoperator.c b/source/libs/executor/src/projectoperator.c index d641810cee..4a3a3f4de4 100644 --- a/source/libs/executor/src/projectoperator.c +++ b/source/libs/executor/src/projectoperator.c @@ -190,6 +190,7 @@ static int32_t setInfoForNewGroup(SSDataBlock* pBlock, SLimitInfo* pLimitInfo, S return PROJECT_RETRIEVE_DONE; } +// todo refactor static int32_t doIngroupLimitOffset(SLimitInfo* pLimitInfo, uint64_t groupId, SSDataBlock* pBlock, SOperatorInfo* pOperator) { // set current group id diff --git a/source/libs/executor/src/scanoperator.c b/source/libs/executor/src/scanoperator.c index 169693b8c4..b939ccdcab 100644 --- a/source/libs/executor/src/scanoperator.c +++ b/source/libs/executor/src/scanoperator.c @@ -617,6 +617,10 @@ static SSDataBlock* doTableScanImpl(SOperatorInfo* pOperator) { T_LONG_JMP(pTaskInfo->env, pTaskInfo->code); } + if (pOperator->status == OP_EXEC_DONE) { + break; + } + // process this data block based on the probabilities bool processThisBlock = processBlockWithProbability(&pTableScanInfo->sample); if (!processThisBlock) { @@ -628,9 +632,8 @@ static SSDataBlock* doTableScanImpl(SOperatorInfo* pOperator) { uint32_t status = 0; int32_t code = loadDataBlock(pOperator, &pTableScanInfo->base, pBlock, &status); - // int32_t code = loadDataBlockOnDemand(pOperator->pRuntimeEnv, pTableScanInfo, pBlock, &status); if (code != TSDB_CODE_SUCCESS) { - T_LONG_JMP(pOperator->pTaskInfo->env, code); + T_LONG_JMP(pTaskInfo->env, code); } // current block is filter out according to filter condition, continue load the next block @@ -2540,7 +2543,7 @@ static SSDataBlock* getTableDataBlockImpl(void* param) { } uint32_t status = 0; - loadDataBlock(pOperator, &pInfo->base, pBlock, &status); + code = loadDataBlock(pOperator, &pInfo->base, pBlock, &status); // code = loadDataBlockFromOneTable(pOperator, pTableScanInfo, pBlock, &status); if (code != TSDB_CODE_SUCCESS) { T_LONG_JMP(pTaskInfo->env, code); @@ -2714,10 +2717,13 @@ SSDataBlock* getSortedTableMergeScanBlockData(SSortHandle* pHandle, SSDataBlock* } } - applyLimitOffset(&pInfo->limitInfo, pResBlock, pTaskInfo); + bool limitReached = applyLimitOffset(&pInfo->limitInfo, pResBlock, pTaskInfo); qDebug("%s get sorted row block, rows:%d, limit:%"PRId64, GET_TASKID(pTaskInfo), pResBlock->info.rows, - pInfo->limitInfo.numOfOutputRows); + pInfo->limitInfo.numOfOutputRows); + if (limitReached) { + resetLimitInfoForNextGroup(&pInfo->limitInfo); + } return (pResBlock->info.rows > 0) ? pResBlock : NULL; } From 18e3450c110e717edbfa71c384b8f5baff179f45 Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Mon, 6 Feb 2023 14:01:43 +0800 Subject: [PATCH 105/124] fix: client update epset crash issue --- source/client/src/clientImpl.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/source/client/src/clientImpl.c b/source/client/src/clientImpl.c index b5b99e92b0..07046eb041 100644 --- a/source/client/src/clientImpl.c +++ b/source/client/src/clientImpl.c @@ -1402,8 +1402,6 @@ int32_t doProcessMsgFromServer(void* param) { tscError("0x%" PRIx64 " rsp msg:%s, code:%s rspLen:%d, elapsed time:%d ms, reqId:0x%" PRIx64, pRequest->self, TMSG_INFO(pMsg->msgType), tstrerror(pMsg->code), pMsg->contLen, elapsed / 1000, pRequest->requestId); } - - taosReleaseRef(clientReqRefPool, pSendInfo->requestObjRefId); } } @@ -1423,6 +1421,11 @@ int32_t doProcessMsgFromServer(void* param) { } pSendInfo->fp(pSendInfo->param, &buf, pMsg->code); + + if (pTscObj) { + taosReleaseRef(clientReqRefPool, pSendInfo->requestObjRefId); + } + rpcFreeCont(pMsg->pCont); destroySendMsgInfo(pSendInfo); From 023b8377fb7da1be37c7264e0c1ceda63bcb8a23 Mon Sep 17 00:00:00 2001 From: Minglei Jin Date: Mon, 6 Feb 2023 14:12:57 +0800 Subject: [PATCH 106/124] enh(tsdb/cache): new block index cache for querying --- source/dnode/vnode/src/inc/tsdb.h | 5 + source/dnode/vnode/src/tsdb/tsdbCache.c | 117 +++++++++++++++++++++ source/dnode/vnode/src/tsdb/tsdbRead.c | 130 ++++++++++++------------ 3 files changed, 187 insertions(+), 65 deletions(-) diff --git a/source/dnode/vnode/src/inc/tsdb.h b/source/dnode/vnode/src/inc/tsdb.h index 2efb00ae32..ec66e581ef 100644 --- a/source/dnode/vnode/src/inc/tsdb.h +++ b/source/dnode/vnode/src/inc/tsdb.h @@ -321,6 +321,8 @@ struct STsdb { STsdbFS fs; SLRUCache *lruCache; TdThreadMutex lruMutex; + SLRUCache *biCache; + TdThreadMutex biMutex; }; struct TSDBKEY { @@ -746,6 +748,9 @@ int32_t tsdbCacheGetLastH(SLRUCache *pCache, tb_uid_t uid, SCacheRowsReader *pr, int32_t tsdbCacheGetLastrowH(SLRUCache *pCache, tb_uid_t uid, SCacheRowsReader *pr, LRUHandle **h); int32_t tsdbCacheRelease(SLRUCache *pCache, LRUHandle *h); +int32_t tsdbCacheGetBlockIdx(SLRUCache *pCache, SDataFReader *pFileReader, LRUHandle **handle); +int32_t tsdbBICacheRelease(SLRUCache *pCache, LRUHandle *h); + int32_t tsdbCacheDeleteLastrow(SLRUCache *pCache, tb_uid_t uid, TSKEY eKey); int32_t tsdbCacheDeleteLast(SLRUCache *pCache, tb_uid_t uid, TSKEY eKey); int32_t tsdbCacheDelete(SLRUCache *pCache, tb_uid_t uid, TSKEY eKey); diff --git a/source/dnode/vnode/src/tsdb/tsdbCache.c b/source/dnode/vnode/src/tsdb/tsdbCache.c index 0fc5b617bb..13ef8b79a5 100644 --- a/source/dnode/vnode/src/tsdb/tsdbCache.c +++ b/source/dnode/vnode/src/tsdb/tsdbCache.c @@ -15,6 +15,34 @@ #include "tsdb.h" +static int32_t tsdbOpenBICache(STsdb *pTsdb) { + int32_t code = 0; + SLRUCache *pCache = taosLRUCacheInit(1 * 1024 * 1024, -1, .5); + if (pCache == NULL) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + + taosLRUCacheSetStrictCapacity(pCache, false); + + taosThreadMutexInit(&pTsdb->biMutex, NULL); + +_err: + pTsdb->biCache = pCache; + return code; +} + +static void tsdbCloseBICache(STsdb *pTsdb) { + SLRUCache *pCache = pTsdb->biCache; + if (pCache) { + taosLRUCacheEraseUnrefEntries(pCache); + + taosLRUCacheCleanup(pCache); + + taosThreadMutexDestroy(&pTsdb->biMutex); + } +} + int32_t tsdbOpenCache(STsdb *pTsdb) { int32_t code = 0; SLRUCache *pCache = NULL; @@ -26,6 +54,12 @@ int32_t tsdbOpenCache(STsdb *pTsdb) { goto _err; } + code = tsdbOpenBICache(pTsdb); + if (code != TSDB_CODE_SUCCESS) { + code = TSDB_CODE_OUT_OF_MEMORY; + goto _err; + } + taosLRUCacheSetStrictCapacity(pCache, false); taosThreadMutexInit(&pTsdb->lruMutex, NULL); @@ -44,6 +78,8 @@ void tsdbCloseCache(STsdb *pTsdb) { taosThreadMutexDestroy(&pTsdb->lruMutex); } + + tsdbCloseBICache(pTsdb); } static void getTableCacheKey(tb_uid_t uid, int cacheType, char *key, int *len) { @@ -1475,3 +1511,84 @@ size_t tsdbCacheGetUsage(SVnode *pVnode) { return usage; } + +static void getBICacheKey(int32_t fid, int64_t commitID, char *key, int *len) { + struct { + int32_t fid; + int64_t commitID; + } biKey = {0}; + + biKey.fid = fid; + biKey.commitID = commitID; + + *len = sizeof(biKey); + memcpy(key, &biKey, *len); +} + +static int32_t tsdbCacheLoadBlockIdx(SDataFReader *pFileReader, SArray **aBlockIdx) { + SArray *pArray = taosArrayInit(8, sizeof(SBlockIdx)); + int32_t code = tsdbReadBlockIdx(pFileReader, pArray); + + if (code != TSDB_CODE_SUCCESS) { + taosArrayDestroy(pArray); + code = TSDB_CODE_OUT_OF_MEMORY; + return code; + } + + *aBlockIdx = pArray; + + return code; +} + +static void deleteBICache(const void *key, size_t keyLen, void *value) { + SArray *pArray = (SArray *)value; + + taosArrayDestroy(pArray); +} + +int32_t tsdbCacheGetBlockIdx(SLRUCache *pCache, SDataFReader *pFileReader, LRUHandle **handle) { + int32_t code = 0; + char key[128] = {0}; + int keyLen = 0; + + getBICacheKey(pFileReader->pSet->fid, pFileReader->pSet->pHeadF->commitID, key, &keyLen); + LRUHandle *h = taosLRUCacheLookup(pCache, key, keyLen); + if (!h) { + STsdb *pTsdb = pFileReader->pTsdb; + taosThreadMutexLock(&pTsdb->biMutex); + + h = taosLRUCacheLookup(pCache, key, keyLen); + if (!h) { + SArray *pArray = NULL; + code = tsdbCacheLoadBlockIdx(pFileReader, &pArray); + // if table's empty or error, return code of -1 + if (code != TSDB_CODE_SUCCESS || pArray == NULL) { + taosThreadMutexUnlock(&pTsdb->biMutex); + + *handle = NULL; + return 0; + } + + size_t charge = pArray->capacity * pArray->elemSize + sizeof(*pArray); + _taos_lru_deleter_t deleter = deleteBICache; + LRUStatus status = taosLRUCacheInsert(pCache, key, keyLen, pArray, charge, deleter, &h, TAOS_LRU_PRIORITY_LOW); + if (status != TAOS_LRU_STATUS_OK) { + code = -1; + } + } + + taosThreadMutexUnlock(&pTsdb->biMutex); + } + + *handle = h; + + return code; +} + +int32_t tsdbBICacheRelease(SLRUCache *pCache, LRUHandle *h) { + int32_t code = 0; + + taosLRUCacheRelease(pCache, h, false); + + return code; +} diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 52beab9748..b67da5c0f7 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -85,13 +85,13 @@ typedef struct SIOCostSummary { } SIOCostSummary; typedef struct SBlockLoadSuppInfo { - SArray* pColAgg; - SColumnDataAgg tsColAgg; - int16_t* colId; - int16_t* slotId; - int32_t numOfCols; - char** buildBuf; // build string tmp buffer, todo remove it later after all string format being updated. - bool smaValid; // the sma on all queried columns are activated + SArray* pColAgg; + SColumnDataAgg tsColAgg; + int16_t* colId; + int16_t* slotId; + int32_t numOfCols; + char** buildBuf; // build string tmp buffer, todo remove it later after all string format being updated. + bool smaValid; // the sma on all queried columns are activated } SBlockLoadSuppInfo; typedef struct SLastBlockReader { @@ -171,11 +171,11 @@ struct STsdbReader { SBlockLoadSuppInfo suppInfo; STsdbReadSnap* pReadSnap; SIOCostSummary cost; - STSchema* pSchema; // the newest version schema - STSchema* pMemSchema; // the previous schema for in-memory data, to avoid load schema too many times - SDataFReader* pFileReader; // the file reader - SDelFReader* pDelFReader; // the del file reader - SArray* pDelIdx; // del file block index; + STSchema* pSchema; // the newest version schema + STSchema* pMemSchema; // the previous schema for in-memory data, to avoid load schema too many times + SDataFReader* pFileReader; // the file reader + SDelFReader* pDelFReader; // the del file reader + SArray* pDelIdx; // del file block index; SVersionRange verRange; SBlockInfoBuf blockInfoBuf; int32_t step; @@ -222,17 +222,18 @@ static int32_t getInitialDelIndex(const SArray* pDelSkyline, int32_t order static bool outOfTimeWindow(int64_t ts, STimeWindow* pWindow) { return (ts > pWindow->ekey) || (ts < pWindow->skey); } -static int32_t setColumnIdSlotList(SBlockLoadSuppInfo* pSupInfo, SColumnInfo* pCols, const int32_t* pSlotIdList, int32_t numOfCols) { +static int32_t setColumnIdSlotList(SBlockLoadSuppInfo* pSupInfo, SColumnInfo* pCols, const int32_t* pSlotIdList, + int32_t numOfCols) { pSupInfo->smaValid = true; pSupInfo->numOfCols = numOfCols; - pSupInfo->colId = taosMemoryMalloc(numOfCols * (sizeof(int16_t)*2 + POINTER_BYTES)); + pSupInfo->colId = taosMemoryMalloc(numOfCols * (sizeof(int16_t) * 2 + POINTER_BYTES)); if (pSupInfo->colId == NULL) { taosMemoryFree(pSupInfo->colId); return TSDB_CODE_OUT_OF_MEMORY; } pSupInfo->slotId = (int16_t*)((char*)pSupInfo->colId + (sizeof(int16_t) * numOfCols)); - pSupInfo->buildBuf = (char**) ((char*)pSupInfo->slotId + (sizeof(int16_t) * numOfCols)); + pSupInfo->buildBuf = (char**)((char*)pSupInfo->slotId + (sizeof(int16_t) * numOfCols)); for (int32_t i = 0; i < numOfCols; ++i) { pSupInfo->colId[i] = pCols[i].colId; pSupInfo->slotId[i] = pSlotIdList[i]; @@ -250,7 +251,7 @@ static int32_t setColumnIdSlotList(SBlockLoadSuppInfo* pSupInfo, SColumnInfo* pC static int32_t updateBlockSMAInfo(STSchema* pSchema, SBlockLoadSuppInfo* pSupInfo) { int32_t i = 0, j = 0; - while(i < pSchema->numOfCols && j < pSupInfo->numOfCols) { + while (i < pSchema->numOfCols && j < pSupInfo->numOfCols) { STColumn* pTCol = &pSchema->columns[i]; if (pTCol->colId == pSupInfo->colId[j]) { if (!IS_BSMA_ON(pTCol)) { @@ -315,7 +316,8 @@ static void* getPosInBlockInfoBuf(SBlockInfoBuf* pBuf, int32_t index) { } // NOTE: speedup the whole processing by preparing the buffer for STableBlockScanInfo in batch model -static SHashObj* createDataBlockScanInfo(STsdbReader* pTsdbReader, SBlockInfoBuf* pBuf, const STableKeyInfo* idList, int32_t numOfTables) { +static SHashObj* createDataBlockScanInfo(STsdbReader* pTsdbReader, SBlockInfoBuf* pBuf, const STableKeyInfo* idList, + int32_t numOfTables) { // allocate buffer in order to load data blocks from file // todo use simple hash instead, optimize the memory consumption SHashObj* pTableMap = @@ -401,9 +403,7 @@ static void destroyAllBlockScanInfo(SHashObj* pTableMap) { taosHashCleanup(pTableMap); } -static bool isEmptyQueryTimeWindow(STimeWindow* pWindow) { - return pWindow->skey > pWindow->ekey; -} +static bool isEmptyQueryTimeWindow(STimeWindow* pWindow) { return pWindow->skey > pWindow->ekey; } // Update the query time window according to the data time to live(TTL) information, in order to avoid to return // the expired data to client, even it is queried already. @@ -647,15 +647,18 @@ _end: } static int32_t doLoadBlockIndex(STsdbReader* pReader, SDataFReader* pFileReader, SArray* pIndexList) { - SArray* aBlockIdx = taosArrayInit(8, sizeof(SBlockIdx)); + // SArray* aBlockIdx = taosArrayInit(8, sizeof(SBlockIdx)); int64_t st = taosGetTimestampUs(); - int32_t code = tsdbReadBlockIdx(pFileReader, aBlockIdx); - if (code != TSDB_CODE_SUCCESS) { + // int32_t code = tsdbReadBlockIdx(pFileReader, aBlockIdx); + LRUHandle* handle = NULL; + int32_t code = tsdbCacheGetBlockIdx(pFileReader->pTsdb->biCache, pFileReader, &handle); + if (code != TSDB_CODE_SUCCESS || handle == NULL) { goto _end; } - size_t num = taosArrayGetSize(aBlockIdx); + SArray* aBlockIdx = (SArray*)taosLRUCacheValue(pFileReader->pTsdb->biCache, handle); + size_t num = taosArrayGetSize(aBlockIdx); if (num == 0) { taosArrayDestroy(aBlockIdx); return TSDB_CODE_SUCCESS; @@ -693,7 +696,8 @@ static int32_t doLoadBlockIndex(STsdbReader* pReader, SDataFReader* pFileReader, pReader->cost.headFileLoadTime += (et1 - st) / 1000.0; _end: - taosArrayDestroy(aBlockIdx); + // taosArrayDestroy(aBlockIdx); + tsdbBICacheRelease(pFileReader->pTsdb->biCache, handle); return code; } @@ -772,7 +776,6 @@ static int32_t doLoadFileBlock(STsdbReader* pReader, SArray* pIndexList, SBlockN numOfTables, pBlockNum->numOfBlocks, numOfQTable, pBlockNum->numOfLastFiles, sizeInDisk / 1000.0, el, pReader->idStr); - pReader->cost.numOfBlocks += total; pReader->cost.headFileLoadTime += el; @@ -906,7 +909,7 @@ static void copyPrimaryTsCol(const SBlockData* pBlockData, SFileBlockDumpInfo* p // a faster version of copy procedure. static void copyNumericCols(const SColData* pData, SFileBlockDumpInfo* pDumpInfo, SColumnInfoData* pColData, - int32_t dumpedRows, bool asc) { + int32_t dumpedRows, bool asc) { uint8_t* p = NULL; if (asc) { p = pData->pData + tDataTypes[pData->type].bytes * pDumpInfo->rowIndex; @@ -915,22 +918,21 @@ static void copyNumericCols(const SColData* pData, SFileBlockDumpInfo* pDumpInfo p = pData->pData + tDataTypes[pData->type].bytes * startIndex; } - int32_t step = asc? 1:-1; + int32_t step = asc ? 1 : -1; // make sure it is aligned to 8bit, the allocated memory address is aligned to 256bit -// ASSERT((((uint64_t)pColData->pData) & (0x8 - 1)) == 0); + // ASSERT((((uint64_t)pColData->pData) & (0x8 - 1)) == 0); // 1. copy data in a batch model memcpy(pColData->pData, p, dumpedRows * tDataTypes[pData->type].bytes); // 2. reverse the array list in case of descending order scan data block if (!asc) { - switch(pColData->info.type) { + switch (pColData->info.type) { case TSDB_DATA_TYPE_TIMESTAMP: case TSDB_DATA_TYPE_DOUBLE: case TSDB_DATA_TYPE_BIGINT: - case TSDB_DATA_TYPE_UBIGINT: - { + case TSDB_DATA_TYPE_UBIGINT: { int32_t mid = dumpedRows >> 1u; int64_t* pts = (int64_t*)pColData->pData; for (int32_t j = 0; j < mid; ++j) { @@ -944,7 +946,7 @@ static void copyNumericCols(const SColData* pData, SFileBlockDumpInfo* pDumpInfo case TSDB_DATA_TYPE_BOOL: case TSDB_DATA_TYPE_TINYINT: case TSDB_DATA_TYPE_UTINYINT: { - int32_t mid = dumpedRows >> 1u; + int32_t mid = dumpedRows >> 1u; int8_t* pts = (int8_t*)pColData->pData; for (int32_t j = 0; j < mid; ++j) { int8_t t = pts[j]; @@ -1116,7 +1118,7 @@ static int32_t copyBlockDataToSDataBlock(STsdbReader* pReader, STableBlockScanIn int32_t unDumpedRows = asc ? pBlock->nRow - pDumpInfo->rowIndex : pDumpInfo->rowIndex + 1; tsdbDebug("%p copy file block to sdatablock, global index:%d, table index:%d, brange:%" PRId64 "-%" PRId64 - ", rows:%d, remain:%d, minVer:%" PRId64 ", maxVer:%" PRId64 ", uid:%"PRIu64" elapsed time:%.2f ms, %s", + ", rows:%d, remain:%d, minVer:%" PRId64 ", maxVer:%" PRId64 ", uid:%" PRIu64 " elapsed time:%.2f ms, %s", pReader, pBlockIter->index, pBlockInfo->tbBlockIdx, pBlock->minKey.ts, pBlock->maxKey.ts, dumpedRows, unDumpedRows, pBlock->minVer, pBlock->maxVer, pBlockInfo->uid, elapsedTime, pReader->idStr); @@ -1770,7 +1772,7 @@ static int32_t doMergeBufAndFileRows(STsdbReader* pReader, STableBlockScanInfo* tRowMergerAdd(&merge, pRow, pSchema); } else { init = true; - int32_t code = tRowMergerInit(&merge, pRow, pSchema); + int32_t code = tRowMergerInit(&merge, pRow, pSchema); if (code != TSDB_CODE_SUCCESS) { return code; } @@ -2182,7 +2184,7 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea int32_t backward = (!ASCENDING_TRAVERSE(pReader->order)); int64_t st = 0; - + STbData* d = NULL; if (pReader->pReadSnap->pMem != NULL) { st = taosGetTimestampUs(); @@ -2307,7 +2309,7 @@ static bool hasDataInLastBlock(SLastBlockReader* pLastBlockReader) { return pLas bool hasDataInFileBlock(const SBlockData* pBlockData, const SFileBlockDumpInfo* pDumpInfo) { if ((pBlockData->nRow > 0) && (pBlockData->nRow != pDumpInfo->totalRows)) { - return false; // this is an invalid result. + return false; // this is an invalid result. } return pBlockData->nRow > 0 && (!pDumpInfo->allDumped); } @@ -2458,7 +2460,7 @@ static int32_t buildComposedDataBlock(STsdbReader* pReader) { TSDBKEY keyInBuf = getCurrentKeyInBuf(pBlockScanInfo, pReader); // it is a clean block, load it directly - if (isCleanFileDataBlock(pReader, pBlockInfo, pBlock, pBlockScanInfo, keyInBuf, pLastBlockReader) && + if (isCleanFileDataBlock(pReader, pBlockInfo, pBlock, pBlockScanInfo, keyInBuf, pLastBlockReader) && pBlock->nRow <= pReader->capacity) { if (asc || ((!asc) && (!hasDataInLastBlock(pLastBlockReader)))) { copyBlockDataToSDataBlock(pReader, pBlockScanInfo); @@ -2661,7 +2663,6 @@ static int32_t moveToNextFile(STsdbReader* pReader, SBlockNumber* pBlockNum) { taosArrayDestroy(pIndexList); if (pReader->pReadSnap != NULL) { - SDelFile* pDelFile = pReader->pReadSnap->fs.pDelFile; if (pReader->pDelFReader == NULL && pDelFile != NULL) { int32_t code = tsdbDelFReaderOpen(&pReader->pDelFReader, pDelFile, pReader->pTsdb); @@ -2849,7 +2850,7 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { ASSERT(pBlockInfo != NULL); -// if (pBlockInfo != NULL) { + // if (pBlockInfo != NULL) { pScanInfo = *(STableBlockScanInfo**)taosHashGet(pReader->status.pTableMap, &pBlockInfo->uid, sizeof(pBlockInfo->uid)); // } else { // pScanInfo = *pReader->status.pTableIter; @@ -2861,9 +2862,9 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { return code; } -// if (pBlockInfo != NULL) { + // if (pBlockInfo != NULL) { pBlock = getCurrentBlock(pBlockIter); -// } + // } initLastBlockReader(pLastBlockReader, pScanInfo, pReader); TSDBKEY keyInBuf = getCurrentKeyInBuf(pScanInfo, pReader); @@ -2928,7 +2929,8 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { // update the last key for the corresponding table pScanInfo->lastKey = ASCENDING_TRAVERSE(pReader->order) ? pInfo->window.ekey : pInfo->window.skey; - tsdbDebug("%p uid:%" PRIu64 " clean file block retrieved from file, global index:%d, " + tsdbDebug("%p uid:%" PRIu64 + " clean file block retrieved from file, global index:%d, " "table index:%d, rows:%d, brange:%" PRId64 "-%" PRId64 ", %s", pReader, pScanInfo->uid, pBlockIter->index, pBlockInfo->tbBlockIdx, pBlock->nRow, pBlock->minKey.ts, pBlock->maxKey.ts, pReader->idStr); @@ -3183,7 +3185,8 @@ bool hasBeenDropped(const SArray* pDelList, int32_t* index, TSDBKEY* pKey, int32 return false; } else if (pKey->ts == last->ts) { TSDBKEY* prev = taosArrayGet(pDelList, num - 2); - return (prev->version >= pKey->version && prev->version <= pVerRange->maxVer && prev->version >= pVerRange->minVer); + return (prev->version >= pKey->version && prev->version <= pVerRange->maxVer && + prev->version >= pVerRange->minVer); } } else { TSDBKEY* pCurrent = taosArrayGet(pDelList, *index); @@ -3367,7 +3370,7 @@ static int32_t checkForNeighborFileBlock(STsdbReader* pReader, STableBlockScanIn *state = CHECK_FILEBLOCK_QUIT; int32_t step = ASCENDING_TRAVERSE(pReader->order) ? 1 : -1; - bool loadNeighbor = true; + bool loadNeighbor = true; int32_t code = loadNeighborIfOverlap(pFBlock, pScanInfo, pReader, &loadNeighbor); if (loadNeighbor && (code == TSDB_CODE_SUCCESS)) { @@ -3619,7 +3622,7 @@ int32_t doAppendRowFromTSRow(SSDataBlock* pBlock, STsdbReader* pReader, STSRow* SColVal colVal = {0}; int32_t i = 0, j = 0; - if (pSupInfo->colId[i]== PRIMARYKEY_TIMESTAMP_COL_ID) { + if (pSupInfo->colId[i] == PRIMARYKEY_TIMESTAMP_COL_ID) { SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, pSupInfo->slotId[i]); ((int64_t*)pColData->pData)[outputRowIndex] = pTSRow->ts; i += 1; @@ -3664,7 +3667,7 @@ int32_t doAppendRowFromFileBlock(SSDataBlock* pResBlock, STsdbReader* pReader, S int32_t outputRowIndex = pResBlock->info.rows; SBlockLoadSuppInfo* pSupInfo = &pReader->suppInfo; - if (pReader->suppInfo.colId[i]== PRIMARYKEY_TIMESTAMP_COL_ID) { + if (pReader->suppInfo.colId[i] == PRIMARYKEY_TIMESTAMP_COL_ID) { SColumnInfoData* pColData = taosArrayGet(pResBlock->pDataBlock, pSupInfo->slotId[i]); ((int64_t*)pColData->pData)[outputRowIndex] = pBlockData->aTSKEY[rowIndex]; i += 1; @@ -4006,18 +4009,17 @@ void tsdbReaderClose(STsdbReader* pReader) { taosMemoryFree(pLReader); } - tsdbDebug("%p :io-cost summary: head-file:%" PRIu64 ", head-file time:%.2f ms, SMA:%" PRId64 - " SMA-time:%.2f ms, fileBlocks:%" PRId64 - ", fileBlocks-load-time:%.2f ms, " - "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 - ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 - ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms," - ", getTbFromMem-time:%.2f ms, getTbFromIMem-time:%.2f ms, initDelSkylineIterTime:%.2f ms, %s", - pReader, pCost->headFileLoad, pCost->headFileLoadTime, pCost->smaDataLoad, pCost->smaLoadTime, - pCost->numOfBlocks, pCost->blockLoadTime, pCost->buildmemBlock, pCost->lastBlockLoad, - pCost->lastBlockLoadTime, pCost->composedBlocks, pCost->buildComposedBlockTime, - numOfTables * sizeof(STableBlockScanInfo) / 1000.0, pCost->createScanInfoList, - pCost->getTbFromMemTime, pCost->getTbFromIMemTime, pCost->initDelSkylineIterTime, pReader->idStr); + tsdbDebug( + "%p :io-cost summary: head-file:%" PRIu64 ", head-file time:%.2f ms, SMA:%" PRId64 + " SMA-time:%.2f ms, fileBlocks:%" PRId64 + ", fileBlocks-load-time:%.2f ms, " + "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 + ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms," + ", getTbFromMem-time:%.2f ms, getTbFromIMem-time:%.2f ms, initDelSkylineIterTime:%.2f ms, %s", + pReader, pCost->headFileLoad, pCost->headFileLoadTime, pCost->smaDataLoad, pCost->smaLoadTime, pCost->numOfBlocks, + pCost->blockLoadTime, pCost->buildmemBlock, pCost->lastBlockLoad, pCost->lastBlockLoadTime, pCost->composedBlocks, + pCost->buildComposedBlockTime, numOfTables * sizeof(STableBlockScanInfo) / 1000.0, pCost->createScanInfoList, + pCost->getTbFromMemTime, pCost->getTbFromIMemTime, pCost->initDelSkylineIterTime, pReader->idStr); taosMemoryFree(pReader->idStr); taosMemoryFree(pReader->pSchema); @@ -4033,7 +4035,7 @@ static bool doTsdbNextDataBlock(STsdbReader* pReader) { blockDataCleanup(pBlock); SReaderStatus* pStatus = &pReader->status; - if (taosHashGetSize(pStatus->pTableMap) == 0){ + if (taosHashGetSize(pStatus->pTableMap) == 0) { return false; } @@ -4123,12 +4125,10 @@ void tsdbRetrieveDataBlockInfo(const STsdbReader* pReader, int32_t* rows, uint64 } } - -static void doFillNullColSMA(SBlockLoadSuppInfo* pSup, int32_t numOfRows, int32_t numOfCols, - SColumnDataAgg* pTsAgg) { +static void doFillNullColSMA(SBlockLoadSuppInfo* pSup, int32_t numOfRows, int32_t numOfCols, SColumnDataAgg* pTsAgg) { // do fill all null column value SMA info int32_t i = 0, j = 0; - int32_t size = (int32_t) taosArrayGetSize(pSup->pColAgg); + int32_t size = (int32_t)taosArrayGetSize(pSup->pColAgg); taosArrayInsert(pSup->pColAgg, 0, pTsAgg); while (j < numOfCols && i < size) { @@ -4141,7 +4141,7 @@ static void doFillNullColSMA(SBlockLoadSuppInfo* pSup, int32_t numOfRows, int32_ } else if (pSup->colId[j] < pAgg->colId) { if (pSup->colId[j] != PRIMARYKEY_TIMESTAMP_COL_ID) { SColumnDataAgg nullColAgg = {.colId = pSup->colId[j], .numOfNull = numOfRows}; - taosArrayInsert(pSup->pColAgg, i ,&nullColAgg); + taosArrayInsert(pSup->pColAgg, i, &nullColAgg); } j += 1; } From a6d52aed4f2b93080ec669064a0966f0f59b560a Mon Sep 17 00:00:00 2001 From: Minglei Jin Date: Mon, 6 Feb 2023 14:18:49 +0800 Subject: [PATCH 107/124] fix(tsdb/cache): use 5M cache for block index instead of 1M --- source/dnode/vnode/src/tsdb/tsdbCache.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbCache.c b/source/dnode/vnode/src/tsdb/tsdbCache.c index 13ef8b79a5..fb2efda8e4 100644 --- a/source/dnode/vnode/src/tsdb/tsdbCache.c +++ b/source/dnode/vnode/src/tsdb/tsdbCache.c @@ -17,7 +17,7 @@ static int32_t tsdbOpenBICache(STsdb *pTsdb) { int32_t code = 0; - SLRUCache *pCache = taosLRUCacheInit(1 * 1024 * 1024, -1, .5); + SLRUCache *pCache = taosLRUCacheInit(5 * 1024 * 1024, -1, .5); if (pCache == NULL) { code = TSDB_CODE_OUT_OF_MEMORY; goto _err; From b83f93d564f1bd8ac4f7a3b06cd37acc11cd2449 Mon Sep 17 00:00:00 2001 From: Minglei Jin Date: Mon, 6 Feb 2023 14:41:17 +0800 Subject: [PATCH 108/124] fix(query): release lru handle before returning from block idx loading --- source/dnode/vnode/src/tsdb/tsdbRead.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index b67da5c0f7..e543565010 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -660,7 +660,8 @@ static int32_t doLoadBlockIndex(STsdbReader* pReader, SDataFReader* pFileReader, SArray* aBlockIdx = (SArray*)taosLRUCacheValue(pFileReader->pTsdb->biCache, handle); size_t num = taosArrayGetSize(aBlockIdx); if (num == 0) { - taosArrayDestroy(aBlockIdx); + tsdbBICacheRelease(pFileReader->pTsdb->biCache, handle); + // taosArrayDestroy(aBlockIdx); return TSDB_CODE_SUCCESS; } From ab328a7c1e1eaf52fdb51475014be073cabee376 Mon Sep 17 00:00:00 2001 From: dapan1121 Date: Mon, 6 Feb 2023 15:26:07 +0800 Subject: [PATCH 109/124] fix: query res freed when fetch handling drop event --- source/libs/qworker/src/qworker.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/source/libs/qworker/src/qworker.c b/source/libs/qworker/src/qworker.c index fedaa96ed9..6486b4b684 100644 --- a/source/libs/qworker/src/qworker.c +++ b/source/libs/qworker/src/qworker.c @@ -517,7 +517,7 @@ int32_t qwHandlePostPhaseEvents(QW_FPARAMS_DEF, int8_t phase, SQWPhaseInput *inp } if (QW_EVENT_RECEIVED(ctx, QW_EVENT_DROP)) { - if (QW_PHASE_POST_FETCH != phase || qwTaskNotInExec(ctx)) { + if (QW_PHASE_POST_FETCH != phase || ((!QW_QUERY_RUNNING(ctx)) && qwTaskNotInExec(ctx))) { QW_ERR_JRET(qwDropTask(QW_FPARAMS())); QW_ERR_JRET(ctx->rspCode); } From e7b97cc08654da001039ded31ca639c9dd6d1476 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 6 Feb 2023 15:59:47 +0800 Subject: [PATCH 110/124] fix(query): set the correct timestamp value in case that only imem has data. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 45 +++++++++++++++++--------- 1 file changed, 29 insertions(+), 16 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index 15432b01c8..6532a99746 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -1552,8 +1552,7 @@ static bool fileBlockShouldLoad(STsdbReader* pReader, SFileDataBlockInfo* pBlock // log the reason why load the datablock for profile if (loadDataBlock) { - tsdbDebug("%p uid:%" PRIu64 - " need to load the datablock, overlapwithneighborblock:%d, hasDup:%d, partiallyRequired:%d, " + tsdbDebug("%p uid:%" PRIu64 " need to load the datablock, overlapneighbor:%d, hasDup:%d, partiallyRequired:%d, " "overlapWithKey:%d, greaterThanBuf:%d, overlapWithDel:%d, overlapWithlastBlock:%d, %s", pReader, pBlockInfo->uid, info.overlapWithNeighborBlock, info.hasDupTs, info.partiallyRequired, info.overlapWithKeyInBuf, info.moreThanCapcity, info.overlapWithDelInfo, info.overlapWithLastBlock, @@ -2595,21 +2594,37 @@ _err: } TSDBKEY getCurrentKeyInBuf(STableBlockScanInfo* pScanInfo, STsdbReader* pReader) { - TSDBKEY key = {.ts = TSKEY_INITIAL_VAL}; + bool asc = ASCENDING_TRAVERSE(pReader->order); + TSKEY initialVal = asc? TSKEY_MIN:TSKEY_MAX; + + TSDBKEY key = {.ts = initialVal}, ikey = {.ts = initialVal}; + + bool hasKey = false, hasIKey = false; TSDBROW* pRow = getValidMemRow(&pScanInfo->iter, pScanInfo->delSkyline, pReader); if (pRow != NULL) { + hasKey = true; key = TSDBROW_KEY(pRow); } - pRow = getValidMemRow(&pScanInfo->iiter, pScanInfo->delSkyline, pReader); - if (pRow != NULL) { - TSDBKEY k = TSDBROW_KEY(pRow); - if (key.ts > k.ts) { - key = k; - } + TSDBROW* pIRow = getValidMemRow(&pScanInfo->iiter, pScanInfo->delSkyline, pReader); + if (pIRow != NULL) { + hasIKey = true; + ikey = TSDBROW_KEY(pIRow); + } + + if (hasKey) { + if (hasIKey) { // has data in mem & imem + if (asc) { + return key.ts <= ikey.ts ? key : ikey; + } else { + return key.ts <= ikey.ts ? ikey: key; + } + } else { // no data in imem + return key; + } + } else { // no data in mem & imem, return the initial value + return hasIKey? ikey:key; } - - return key; } static int32_t moveToNextFile(STsdbReader* pReader, SBlockNumber* pBlockNum) { @@ -3997,11 +4012,9 @@ void tsdbReaderClose(STsdbReader* pReader) { } tsdbDebug("%p :io-cost summary: head-file:%" PRIu64 ", head-file time:%.2f ms, SMA:%" PRId64 - " SMA-time:%.2f ms, fileBlocks:%" PRId64 - ", fileBlocks-load-time:%.2f ms, " - "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 - ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 - ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms, %s", + " SMA-time:%.2f ms, fileBlocks:%" PRId64 ", fileBlocks-load-time:%.2f ms, " + "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 + ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, createTime:%.2f ms, %s", pReader, pCost->headFileLoad, pCost->headFileLoadTime, pCost->smaDataLoad, pCost->smaLoadTime, pCost->numOfBlocks, pCost->blockLoadTime, pCost->buildmemBlock, pCost->lastBlockLoad, pCost->lastBlockLoadTime, pCost->composedBlocks, pCost->buildComposedBlockTime, From 08b3ce7cc06def2e45d26e6ed3e680661a69ac90 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 6 Feb 2023 16:08:41 +0800 Subject: [PATCH 111/124] fix(query): remove timing code. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 13 ++++--------- 1 file changed, 4 insertions(+), 9 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index aa5d8fa3a4..c3f5e0f3c5 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -79,8 +79,8 @@ typedef struct SIOCostSummary { int64_t composedBlocks; double buildComposedBlockTime; double createScanInfoList; - double getTbFromMemTime; - double getTbFromIMemTime; +// double getTbFromMemTime; +// double getTbFromIMemTime; double initDelSkylineIterTime; } SIOCostSummary; @@ -2187,9 +2187,7 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea STbData* d = NULL; if (pReader->pReadSnap->pMem != NULL) { - st = taosGetTimestampUs(); d = tsdbGetTbDataFromMemTable(pReader->pReadSnap->pMem, pReader->suid, pBlockScanInfo->uid); - pReader->cost.getTbFromMemTime += (taosGetTimestampUs() - st) / 1000.0; if (d != NULL) { code = tsdbTbDataIterCreate(d, &startKey, backward, &pBlockScanInfo->iter.iter); if (code == TSDB_CODE_SUCCESS) { @@ -2210,9 +2208,7 @@ static int32_t initMemDataIterator(STableBlockScanInfo* pBlockScanInfo, STsdbRea STbData* di = NULL; if (pReader->pReadSnap->pIMem != NULL) { - st = taosGetTimestampUs(); di = tsdbGetTbDataFromMemTable(pReader->pReadSnap->pIMem, pReader->suid, pBlockScanInfo->uid); - pReader->cost.getTbFromIMemTime += (taosGetTimestampUs() - st) / 1000.0; if (di != NULL) { code = tsdbTbDataIterCreate(di, &startKey, backward, &pBlockScanInfo->iiter.iter); if (code == TSDB_CODE_SUCCESS) { @@ -4030,12 +4026,11 @@ void tsdbReaderClose(STsdbReader* pReader) { " SMA-time:%.2f ms, fileBlocks:%" PRId64 ", fileBlocks-load-time:%.2f ms, " "build in-memory-block-time:%.2f ms, lastBlocks:%" PRId64 ", lastBlocks-time:%.2f ms, composed-blocks:%" PRId64 - ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, creatTime:%.2f ms," - "getTbFromMem-time:%.2f ms, getTbFromIMem-time:%.2f ms, initDelSkylineIterTime:%.2f ms, %s", + ", composed-blocks-time:%.2fms, STableBlockScanInfo size:%.2f Kb, createTime:%.2f ms,initDelSkylineIterTime:%.2f ms, %s", pReader, pCost->headFileLoad, pCost->headFileLoadTime, pCost->smaDataLoad, pCost->smaLoadTime, pCost->numOfBlocks, pCost->blockLoadTime, pCost->buildmemBlock, pCost->lastBlockLoad, pCost->lastBlockLoadTime, pCost->composedBlocks, pCost->buildComposedBlockTime, numOfTables * sizeof(STableBlockScanInfo) / 1000.0, pCost->createScanInfoList, - pCost->getTbFromMemTime, pCost->getTbFromIMemTime, pCost->initDelSkylineIterTime, pReader->idStr); + pCost->initDelSkylineIterTime, pReader->idStr); taosMemoryFree(pReader->idStr); taosMemoryFree(pReader->pSchema); From 55f78b4109d8d551273729e63e2decdfd3aac972 Mon Sep 17 00:00:00 2001 From: Liu Jicong Date: Mon, 6 Feb 2023 14:55:38 +0800 Subject: [PATCH 112/124] fix case --- source/dnode/mnode/impl/src/mndStream.c | 1 + tests/script/tsim/stream/drop_stream.sim | 6 +++--- 2 files changed, 4 insertions(+), 3 deletions(-) diff --git a/source/dnode/mnode/impl/src/mndStream.c b/source/dnode/mnode/impl/src/mndStream.c index cb6b54ceed..8f6c7e19ed 100644 --- a/source/dnode/mnode/impl/src/mndStream.c +++ b/source/dnode/mnode/impl/src/mndStream.c @@ -647,6 +647,7 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) { if (pStream->sourceDbUid == streamObj.sourceDbUid) { ++numOfStream; } + sdbRelease(pMnode->pSdb, pStream); if (numOfStream > MND_STREAM_MAX_NUM) { mError("too many streams, no more than 10 for each database"); terrno = TSDB_CODE_MND_TOO_MANY_STREAMS; diff --git a/tests/script/tsim/stream/drop_stream.sim b/tests/script/tsim/stream/drop_stream.sim index 8d193bc0a0..7ff9632a3e 100644 --- a/tests/script/tsim/stream/drop_stream.sim +++ b/tests/script/tsim/stream/drop_stream.sim @@ -55,9 +55,9 @@ sql create stream stb_asin_stream trigger at_once into output_asin_stb as select sql create stream ctb_asin_stream trigger at_once into output_asin_ctb as select ts, asin(c1), asin(c2), c3 from scalar_ct1; sql create stream tb_asin_stream trigger at_once into output_asin_tb as select ts, asin(c1), asin(c2), c3 from scalar_tb; sql create stream stb_atan_stream trigger at_once into output_atan_stb as select ts, atan(c1), atan(c2), c3 from scalar_stb partition by tbname; -sql create stream ctb_atan_stream trigger at_once into output_atan_ctb as select ts, atan(c1), atan(c2), c3 from scalar_ct1; -sql create stream tb_atan_stream trigger at_once into output_atan_tb as select ts, atan(c1), atan(c2), c3 from scalar_tb; -sql create stream stb_ceil_stream trigger at_once into output_ceil_stb as select ts, ceil(c1), ceil(c2), c3 from scalar_stb partition by tbname; +# sql create stream ctb_atan_stream trigger at_once into output_atan_ctb as select ts, atan(c1), atan(c2), c3 from scalar_ct1; +# sql create stream tb_atan_stream trigger at_once into output_atan_tb as select ts, atan(c1), atan(c2), c3 from scalar_tb; +# sql create stream stb_ceil_stream trigger at_once into output_ceil_stb as select ts, ceil(c1), ceil(c2), c3 from scalar_stb partition by tbname; # sql create stream ctb_ceil_stream trigger at_once into output_ceil_ctb as select ts, ceil(c1), ceil(c2), c3 from scalar_ct1; # sql create stream tb_ceil_stream trigger at_once into output_ceil_tb as select ts, ceil(c1), ceil(c2), c3 from scalar_tb; # sql create stream stb_cos_stream trigger at_once into output_cos_stb as select ts, cos(c1), cos(c2), c3 from scalar_stb partition by tbname; From 5e2bb7527508f4f9d9ea7bef1a8e85176f599b55 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 6 Feb 2023 17:56:45 +0800 Subject: [PATCH 113/124] fix(query):set correct initial key value for mem/imem data. --- source/dnode/vnode/src/tsdb/tsdbRead.c | 17 ++++++----------- 1 file changed, 6 insertions(+), 11 deletions(-) diff --git a/source/dnode/vnode/src/tsdb/tsdbRead.c b/source/dnode/vnode/src/tsdb/tsdbRead.c index c3f5e0f3c5..717b76046a 100644 --- a/source/dnode/vnode/src/tsdb/tsdbRead.c +++ b/source/dnode/vnode/src/tsdb/tsdbRead.c @@ -2604,9 +2604,9 @@ _err: TSDBKEY getCurrentKeyInBuf(STableBlockScanInfo* pScanInfo, STsdbReader* pReader) { bool asc = ASCENDING_TRAVERSE(pReader->order); - TSKEY initialVal = asc? TSKEY_MIN:TSKEY_MAX; +// TSKEY initialVal = asc? TSKEY_MIN:TSKEY_MAX; - TSDBKEY key = {.ts = initialVal}, ikey = {.ts = initialVal}; + TSDBKEY key = {.ts = TSKEY_INITIAL_VAL}, ikey = {.ts = TSKEY_INITIAL_VAL}; bool hasKey = false, hasIKey = false; TSDBROW* pRow = getValidMemRow(&pScanInfo->iter, pScanInfo->delSkyline, pReader); @@ -2631,8 +2631,10 @@ TSDBKEY getCurrentKeyInBuf(STableBlockScanInfo* pScanInfo, STsdbReader* pReader) } else { // no data in imem return key; } - } else { // no data in mem & imem, return the initial value - return hasIKey? ikey:key; + } else { + // no data in mem & imem, return the initial value + // only imem has data, return ikey + return ikey; } } @@ -2862,21 +2864,14 @@ static int32_t doBuildDataBlock(STsdbReader* pReader) { ASSERT(pBlockInfo != NULL); - // if (pBlockInfo != NULL) { pScanInfo = *(STableBlockScanInfo**)taosHashGet(pReader->status.pTableMap, &pBlockInfo->uid, sizeof(pBlockInfo->uid)); - // } else { - // pScanInfo = *pReader->status.pTableIter; - // } - if (pScanInfo == NULL) { tsdbError("failed to get table scan-info, %s", pReader->idStr); code = TSDB_CODE_INVALID_PARA; return code; } - // if (pBlockInfo != NULL) { pBlock = getCurrentBlock(pBlockIter); - // } initLastBlockReader(pLastBlockReader, pScanInfo, pReader); TSDBKEY keyInBuf = getCurrentKeyInBuf(pScanInfo, pReader); From 2d951c1690b01c569f10c2a506e8111ca97014b9 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 6 Feb 2023 22:34:27 +0800 Subject: [PATCH 114/124] fix(query):set null for if the queried table is dropped when retrieving the corresponding tags. --- source/libs/executor/src/scanoperator.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) diff --git a/source/libs/executor/src/scanoperator.c b/source/libs/executor/src/scanoperator.c index b939ccdcab..6d3eb67f16 100644 --- a/source/libs/executor/src/scanoperator.c +++ b/source/libs/executor/src/scanoperator.c @@ -446,6 +446,16 @@ static STableCachedVal* createTableCacheVal(const SMetaReader* pMetaReader) { // const void *key, size_t keyLen, void *value static void freeCachedMetaItem(const void* key, size_t keyLen, void* value) { freeTableCachedVal(value); } + +static void doSetNullValue(SSDataBlock* pBlock, const SExprInfo* pExpr, int32_t numOfExpr) { + for (int32_t j = 0; j < numOfExpr; ++j) { + int32_t dstSlotId = pExpr[j].base.resSchema.slotId; + + SColumnInfoData* pColInfoData = taosArrayGet(pBlock->pDataBlock, dstSlotId); + colDataAppendNNULL(pColInfoData, 0, pBlock->info.rows); + } +} + int32_t addTagPseudoColumnData(SReadHandle* pHandle, const SExprInfo* pExpr, int32_t numOfExpr, SSDataBlock* pBlock, int32_t rows, const char* idStr, STableMetaCacheInfo* pCache) { // currently only the tbname pseudo column @@ -465,14 +475,22 @@ int32_t addTagPseudoColumnData(SReadHandle* pHandle, const SExprInfo* pExpr, int SMetaReader mr = {0}; LRUHandle* h = NULL; + // todo refactor: extract method + // the handling of the null data should be packed in the extracted method + // 1. check if it is existed in meta cache if (pCache == NULL) { metaReaderInit(&mr, pHandle->meta, 0); code = metaGetTableEntryByUidCache(&mr, pBlock->info.id.uid); if (code != TSDB_CODE_SUCCESS) { + + // when encounter the TSDB_CODE_PAR_TABLE_NOT_EXIST error, we proceed. if (terrno == TSDB_CODE_PAR_TABLE_NOT_EXIST) { qWarn("failed to get table meta, table may have been dropped, uid:0x%" PRIx64 ", code:%s, %s", pBlock->info.id.uid, tstrerror(terrno), idStr); + + // append null value before return to caller, since the caller will ignore this error code and proceed + doSetNullValue(pBlock, pExpr, numOfExpr); } else { qError("failed to get table meta, uid:0x%" PRIx64 ", code:%s, %s", pBlock->info.id.uid, tstrerror(terrno), idStr); @@ -498,6 +516,8 @@ int32_t addTagPseudoColumnData(SReadHandle* pHandle, const SExprInfo* pExpr, int if (terrno == TSDB_CODE_PAR_TABLE_NOT_EXIST) { qWarn("failed to get table meta, table may have been dropped, uid:0x%" PRIx64 ", code:%s, %s", pBlock->info.id.uid, tstrerror(terrno), idStr); + // append null value before return to caller, since the caller will ignore this error code and proceed + doSetNullValue(pBlock, pExpr, numOfExpr); } else { qError("failed to get table meta, uid:0x%" PRIx64 ", code:%s, %s", pBlock->info.id.uid, tstrerror(terrno), idStr); From 917d1165a779bc5adc545dba91b9386242a9dd37 Mon Sep 17 00:00:00 2001 From: Shengliang Guan Date: Mon, 6 Feb 2023 22:49:13 +0800 Subject: [PATCH 115/124] fix: set transId to 0 while timeout or in follower state --- source/dnode/mgmt/mgmt_mnode/src/mmWorker.c | 2 +- source/dnode/mnode/impl/src/mndSync.c | 8 ++++++-- source/dnode/mnode/impl/src/mndTrans.c | 12 +++++++++--- 3 files changed, 16 insertions(+), 6 deletions(-) diff --git a/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c b/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c index 095857825d..b0810d528f 100644 --- a/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c +++ b/source/dnode/mgmt/mgmt_mnode/src/mmWorker.c @@ -49,7 +49,7 @@ static void mmProcessRpcMsg(SQueueInfo *pInfo, SRpcMsg *pMsg) { pMsg->info.node = pMgmt->pMnode; const STraceId *trace = &pMsg->info.traceId; - dGTrace("msg:%p, get from mnode queue", pMsg); + dGTrace("msg:%p, get from mnode queue, type:%s", pMsg, TMSG_INFO(pMsg->msgType)); int32_t code = mndProcessRpcMsg(pMsg); diff --git a/source/dnode/mnode/impl/src/mndSync.c b/source/dnode/mnode/impl/src/mndSync.c index d458ffaed4..387a141c49 100644 --- a/source/dnode/mnode/impl/src/mndSync.c +++ b/source/dnode/mnode/impl/src/mndSync.c @@ -114,7 +114,11 @@ int32_t mndProcessWriteMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta taosThreadMutexUnlock(&pMgmt->lock); STrans *pTrans = mndAcquireTrans(pMnode, transId); if (pTrans != NULL) { - mInfo("trans:%d, execute in mnode which not leader or sync timeout", transId); + mInfo("trans:%d, execute in mnode which not leader or sync timeout, createTime:%" PRId64 " saved trans:%d", + transId, pTrans->createdTime, pMgmt->transId); + pMgmt->transId = 0; + pMgmt->transSec = 0; + pMgmt->transSeq = 0; mndTransExecute(pMnode, pTrans); mndReleaseTrans(pMnode, pTrans); // sdbWriteFile(pMnode->pSdb, SDB_WRITE_DELTA); @@ -372,7 +376,7 @@ int32_t mndSyncPropose(SMnode *pMnode, SSdbRaw *pRaw, int32_t transId) { taosThreadMutexLock(&pMgmt->lock); pMgmt->errCode = 0; - if (pMgmt->transId != 0) { + if (pMgmt->transId != 0 && pMgmt->transId != transId) { mError("trans:%d, can't be proposed since trans:%d already waiting for confirm", transId, pMgmt->transId); taosThreadMutexUnlock(&pMgmt->lock); rpcFreeCont(req.pCont); diff --git a/source/dnode/mnode/impl/src/mndTrans.c b/source/dnode/mnode/impl/src/mndTrans.c index 0b28a6eb43..15c28f8e8f 100644 --- a/source/dnode/mnode/impl/src/mndTrans.c +++ b/source/dnode/mnode/impl/src/mndTrans.c @@ -794,7 +794,8 @@ static int32_t mndTransSync(SMnode *pMnode, STrans *pTrans) { mInfo("trans:%d, sync to other mnodes, stage:%s", pTrans->id, mndTransStr(pTrans->stage)); int32_t code = mndSyncPropose(pMnode, pRaw, pTrans->id); if (code != 0) { - mError("trans:%d, failed to sync, errno:%s code:%s", pTrans->id, terrstr(), tstrerror(code)); + mError("trans:%d, failed to sync, errno:%s code:%s createTime:%" PRId64 " saved trans:%d", pTrans->id, terrstr(), + tstrerror(code), pTrans->createdTime, pMnode->syncMgmt.transId); sdbFreeRaw(pRaw); return -1; } @@ -1495,7 +1496,11 @@ static bool mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans) { mError("trans:%d, failed to write sdb since %s", pTrans->id, terrstr()); } - mInfo("trans:%d, execute finished, code:0x%x, failedTimes:%d", pTrans->id, pTrans->code, pTrans->failedTimes); + mInfo("trans:%d, execute finished, code:0x%x, failedTimes:%d createTime:%" PRId64, pTrans->id, pTrans->code, + pTrans->failedTimes, pTrans->createdTime); + pMnode->syncMgmt.transId = 0; + pMnode->syncMgmt.transSec = 0; + pMnode->syncMgmt.transSeq = 0; return continueExec; } @@ -1503,7 +1508,8 @@ void mndTransExecute(SMnode *pMnode, STrans *pTrans) { bool continueExec = true; while (continueExec) { - mInfo("trans:%d, continue to execute, stage:%s", pTrans->id, mndTransStr(pTrans->stage)); + mInfo("trans:%d, continue to execute, stage:%s createTime:%" PRId64, pTrans->id, mndTransStr(pTrans->stage), + pTrans->createdTime); pTrans->lastExecTime = taosGetTimestampMs(); switch (pTrans->stage) { case TRN_STAGE_PREPARE: From 151dfea99c265409e9cef64f459826ebb67522ca Mon Sep 17 00:00:00 2001 From: Shengliang Guan Date: Tue, 7 Feb 2023 00:07:05 +0800 Subject: [PATCH 116/124] fix: restore some changes --- source/dnode/mnode/impl/src/mndSync.c | 5 +---- source/dnode/mnode/impl/src/mndTrans.c | 8 +++----- 2 files changed, 4 insertions(+), 9 deletions(-) diff --git a/source/dnode/mnode/impl/src/mndSync.c b/source/dnode/mnode/impl/src/mndSync.c index 387a141c49..97ad541c1f 100644 --- a/source/dnode/mnode/impl/src/mndSync.c +++ b/source/dnode/mnode/impl/src/mndSync.c @@ -116,9 +116,6 @@ int32_t mndProcessWriteMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta if (pTrans != NULL) { mInfo("trans:%d, execute in mnode which not leader or sync timeout, createTime:%" PRId64 " saved trans:%d", transId, pTrans->createdTime, pMgmt->transId); - pMgmt->transId = 0; - pMgmt->transSec = 0; - pMgmt->transSeq = 0; mndTransExecute(pMnode, pTrans); mndReleaseTrans(pMnode, pTrans); // sdbWriteFile(pMnode->pSdb, SDB_WRITE_DELTA); @@ -376,7 +373,7 @@ int32_t mndSyncPropose(SMnode *pMnode, SSdbRaw *pRaw, int32_t transId) { taosThreadMutexLock(&pMgmt->lock); pMgmt->errCode = 0; - if (pMgmt->transId != 0 && pMgmt->transId != transId) { + if (pMgmt->transId != 0 /* && pMgmt->transId != transId*/) { mError("trans:%d, can't be proposed since trans:%d already waiting for confirm", transId, pMgmt->transId); taosThreadMutexUnlock(&pMgmt->lock); rpcFreeCont(req.pCont); diff --git a/source/dnode/mnode/impl/src/mndTrans.c b/source/dnode/mnode/impl/src/mndTrans.c index 15c28f8e8f..f931ca121a 100644 --- a/source/dnode/mnode/impl/src/mndTrans.c +++ b/source/dnode/mnode/impl/src/mndTrans.c @@ -791,7 +791,8 @@ static int32_t mndTransSync(SMnode *pMnode, STrans *pTrans) { } (void)sdbSetRawStatus(pRaw, SDB_STATUS_READY); - mInfo("trans:%d, sync to other mnodes, stage:%s", pTrans->id, mndTransStr(pTrans->stage)); + mInfo("trans:%d, sync to other mnodes, stage:%s createTime:%" PRId64, pTrans->id, mndTransStr(pTrans->stage), + pTrans->createdTime); int32_t code = mndSyncPropose(pMnode, pRaw, pTrans->id); if (code != 0) { mError("trans:%d, failed to sync, errno:%s code:%s createTime:%" PRId64 " saved trans:%d", pTrans->id, terrstr(), @@ -801,7 +802,7 @@ static int32_t mndTransSync(SMnode *pMnode, STrans *pTrans) { } sdbFreeRaw(pRaw); - mInfo("trans:%d, sync finished", pTrans->id); + mInfo("trans:%d, sync finished, createTime:%" PRId64, pTrans->id, pTrans->createdTime); return 0; } @@ -1498,9 +1499,6 @@ static bool mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans) { mInfo("trans:%d, execute finished, code:0x%x, failedTimes:%d createTime:%" PRId64, pTrans->id, pTrans->code, pTrans->failedTimes, pTrans->createdTime); - pMnode->syncMgmt.transId = 0; - pMnode->syncMgmt.transSec = 0; - pMnode->syncMgmt.transSeq = 0; return continueExec; } From 4d20469e2b3cfe92cf7e37e3e227e5277d102503 Mon Sep 17 00:00:00 2001 From: Shengliang Guan Date: Tue, 7 Feb 2023 00:12:05 +0800 Subject: [PATCH 117/124] fix: adjust log formats --- source/libs/sync/src/syncCommit.c | 2 +- source/libs/sync/src/syncMain.c | 20 ++++----- source/libs/sync/src/syncPipeline.c | 66 ++++++++++++++-------------- source/libs/sync/src/syncRaftEntry.c | 2 +- source/libs/sync/src/syncUtil.c | 2 +- 5 files changed, 46 insertions(+), 46 deletions(-) diff --git a/source/libs/sync/src/syncCommit.c b/source/libs/sync/src/syncCommit.c index 67ed1e0701..6d256a735d 100644 --- a/source/libs/sync/src/syncCommit.c +++ b/source/libs/sync/src/syncCommit.c @@ -110,7 +110,7 @@ int64_t syncNodeCheckCommitIndex(SSyncNode* ths, SyncIndex indexLikely) { if (indexLikely > ths->commitIndex && syncNodeAgreedUpon(ths, indexLikely)) { SyncIndex commitIndex = indexLikely; syncNodeUpdateCommitIndex(ths, commitIndex); - sTrace("vgId:%d, agreed upon. role:%d, term:%" PRId64 ", index: %" PRId64 "", ths->vgId, ths->state, + sTrace("vgId:%d, agreed upon. role:%d, term:%" PRId64 ", index:%" PRId64 "", ths->vgId, ths->state, ths->raftStore.currentTerm, commitIndex); } return ths->commitIndex; diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index ea22ac7bb5..803df6b662 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -85,7 +85,7 @@ int64_t syncOpen(SSyncInfo* pSyncInfo) { int32_t syncStart(int64_t rid) { SSyncNode* pSyncNode = syncNodeAcquire(rid); if (pSyncNode == NULL) { - sError("failed to acquire rid: %" PRId64 " of tsNodeReftId for pSyncNode", rid); + sError("failed to acquire rid:%" PRId64 " of tsNodeReftId for pSyncNode", rid); return -1; } @@ -756,7 +756,7 @@ int32_t syncNodeLogStoreRestoreOnNeed(SSyncNode* pNode) { SyncIndex lastVer = pNode->pLogStore->syncLogLastIndex(pNode->pLogStore); if (lastVer < commitIndex || firstVer > commitIndex + 1) { if (pNode->pLogStore->syncLogRestoreFromSnapshot(pNode->pLogStore, commitIndex)) { - sError("vgId:%d, failed to restore log store from snapshot since %s. lastVer: %" PRId64 ", snapshotVer: %" PRId64, + sError("vgId:%d, failed to restore log store from snapshot since %s. lastVer:%" PRId64 ", snapshotVer:%" PRId64, pNode->vgId, terrstr(), lastVer, commitIndex); return -1; } @@ -1101,7 +1101,7 @@ int32_t syncNodeRestore(SSyncNode* pSyncNode) { SyncIndex endIndex = pSyncNode->pLogBuf->endIndex; if (lastVer != -1 && endIndex != lastVer + 1) { terrno = TSDB_CODE_WAL_LOG_INCOMPLETE; - sError("vgId:%d, failed to restore sync node since %s. expected lastLogIndex: %" PRId64 ", lastVer: %" PRId64 "", + sError("vgId:%d, failed to restore sync node since %s. expected lastLogIndex:%" PRId64 ", lastVer:%" PRId64 "", pSyncNode->vgId, terrstr(), endIndex - 1, lastVer); return -1; } @@ -1820,7 +1820,7 @@ void syncNodeCandidate2Leader(SSyncNode* pSyncNode) { SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore); ASSERT(lastIndex >= 0); - sInfo("vgId:%d, become leader. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64 "", + sInfo("vgId:%d, become leader. term:%" PRId64 ", commit index:%" PRId64 ", last index:%" PRId64 "", pSyncNode->vgId, pSyncNode->raftStore.currentTerm, pSyncNode->commitIndex, lastIndex); } @@ -1839,7 +1839,7 @@ void syncNodeFollower2Candidate(SSyncNode* pSyncNode) { ASSERT(pSyncNode->state == TAOS_SYNC_STATE_FOLLOWER); pSyncNode->state = TAOS_SYNC_STATE_CANDIDATE; SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore); - sInfo("vgId:%d, become candidate from follower. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64, + sInfo("vgId:%d, become candidate from follower. term:%" PRId64 ", commit index:%" PRId64 ", last index:%" PRId64, pSyncNode->vgId, pSyncNode->raftStore.currentTerm, pSyncNode->commitIndex, lastIndex); sNTrace(pSyncNode, "follower to candidate"); @@ -1849,7 +1849,7 @@ void syncNodeLeader2Follower(SSyncNode* pSyncNode) { ASSERT(pSyncNode->state == TAOS_SYNC_STATE_LEADER); syncNodeBecomeFollower(pSyncNode, "leader to follower"); SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore); - sInfo("vgId:%d, become follower from leader. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64, + sInfo("vgId:%d, become follower from leader. term:%" PRId64 ", commit index:%" PRId64 ", last index:%" PRId64, pSyncNode->vgId, pSyncNode->raftStore.currentTerm, pSyncNode->commitIndex, lastIndex); sNTrace(pSyncNode, "leader to follower"); @@ -1859,7 +1859,7 @@ void syncNodeCandidate2Follower(SSyncNode* pSyncNode) { ASSERT(pSyncNode->state == TAOS_SYNC_STATE_CANDIDATE); syncNodeBecomeFollower(pSyncNode, "candidate to follower"); SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore); - sInfo("vgId:%d, become follower from candidate. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64, + sInfo("vgId:%d, become follower from candidate. term:%" PRId64 ", commit index:%" PRId64 ", last index:%" PRId64, pSyncNode->vgId, pSyncNode->raftStore.currentTerm, pSyncNode->commitIndex, lastIndex); sNTrace(pSyncNode, "candidate to follower"); @@ -2299,7 +2299,7 @@ int32_t syncNodeAppend(SSyncNode* ths, SSyncRaftEntry* pEntry) { // proceed match index, with replicating on needed SyncIndex matchIndex = syncLogBufferProceed(ths->pLogBuf, ths, NULL); - sTrace("vgId:%d, append raft entry. index: %" PRId64 ", term: %" PRId64 " pBuf: [%" PRId64 " %" PRId64 " %" PRId64 + sTrace("vgId:%d, append raft entry. index:%" PRId64 ", term:%" PRId64 " pBuf: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", ths->vgId, pEntry->index, pEntry->term, ths->pLogBuf->startIndex, ths->pLogBuf->commitIndex, ths->pLogBuf->matchIndex, ths->pLogBuf->endIndex); @@ -2472,7 +2472,7 @@ int32_t syncNodeOnHeartbeat(SSyncNode* ths, const SRpcMsg* pRpcMsg) { sError("vgId:%d, sync enqueue step-down msg error, code:%d", ths->vgId, code); rpcFreeCont(rpcMsgLocalCmd.pCont); } else { - sTrace("vgId:%d, sync enqueue step-down msg, new-term: %" PRId64, ths->vgId, pSyncMsg->currentTerm); + sTrace("vgId:%d, sync enqueue step-down msg, new-term:%" PRId64, ths->vgId, pSyncMsg->currentTerm); } } } @@ -2538,7 +2538,7 @@ int32_t syncNodeOnLocalCmd(SSyncNode* ths, const SRpcMsg* pRpcMsg) { (void)syncNodeUpdateCommitIndex(ths, pMsg->commitIndex); } if (syncLogBufferCommit(ths->pLogBuf, ths, ths->commitIndex) < 0) { - sError("vgId:%d, failed to commit raft log since %s. commit index: %" PRId64 "", ths->vgId, terrstr(), + sError("vgId:%d, failed to commit raft log since %s. commit index:%" PRId64 "", ths->vgId, terrstr(), ths->commitIndex); } } else { diff --git a/source/libs/sync/src/syncPipeline.c b/source/libs/sync/src/syncPipeline.c index bb3bb0d6a4..b3eb5684cf 100644 --- a/source/libs/sync/src/syncPipeline.c +++ b/source/libs/sync/src/syncPipeline.c @@ -132,16 +132,16 @@ SSyncRaftEntry* syncEntryBuildDummy(SyncTerm term, SyncIndex index, int32_t vgId int32_t syncLogValidateAlignmentOfCommit(SSyncNode* pNode, SyncIndex commitIndex) { SyncIndex firstVer = pNode->pLogStore->syncLogBeginIndex(pNode->pLogStore); if (firstVer > commitIndex + 1) { - sError("vgId:%d, firstVer of WAL log greater than tsdb commit version + 1. firstVer: %" PRId64 - ", tsdb commit version: %" PRId64 "", + sError("vgId:%d, firstVer of WAL log greater than tsdb commit version + 1. firstVer:%" PRId64 + ", tsdb commit version:%" PRId64 "", pNode->vgId, firstVer, commitIndex); return -1; } SyncIndex lastVer = pNode->pLogStore->syncLogLastIndex(pNode->pLogStore); if (lastVer < commitIndex) { - sError("vgId:%d, lastVer of WAL log less than tsdb commit version. lastVer: %" PRId64 - ", tsdb commit version: %" PRId64 "", + sError("vgId:%d, lastVer of WAL log less than tsdb commit version. lastVer:%" PRId64 + ", tsdb commit version:%" PRId64 "", pNode->vgId, lastVer, commitIndex); return -1; } @@ -293,7 +293,7 @@ int32_t syncLogBufferAccept(SSyncLogBuffer* pBuf, SSyncNode* pNode, SSyncRaftEnt bool inBuf = true; if (index <= pBuf->commitIndex) { - sTrace("vgId:%d, already committed. index: %" PRId64 ", term: %" PRId64 ". log buffer: [%" PRId64 " %" PRId64 + sTrace("vgId:%d, already committed. index:%" PRId64 ", term:%" PRId64 ". log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pEntry->index, pEntry->term, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); @@ -306,7 +306,7 @@ int32_t syncLogBufferAccept(SSyncLogBuffer* pBuf, SSyncNode* pNode, SSyncRaftEnt } if (index - pBuf->startIndex >= pBuf->size) { - sWarn("vgId:%d, out of buffer range. index: %" PRId64 ", term: %" PRId64 ". log buffer: [%" PRId64 " %" PRId64 + sWarn("vgId:%d, out of buffer range. index:%" PRId64 ", term:%" PRId64 ". log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pEntry->index, pEntry->term, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); @@ -314,8 +314,8 @@ int32_t syncLogBufferAccept(SSyncLogBuffer* pBuf, SSyncNode* pNode, SSyncRaftEnt } if (index > pBuf->matchIndex && lastMatchTerm != prevTerm) { - sWarn("vgId:%d, not ready to accept. index: %" PRId64 ", term: %" PRId64 ": prevterm: %" PRId64 - " != lastmatch: %" PRId64 ". log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", + sWarn("vgId:%d, not ready to accept. index:%" PRId64 ", term:%" PRId64 ": prevterm:%" PRId64 + " != lastmatch:%" PRId64 ". log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pEntry->index, pEntry->term, prevTerm, lastMatchTerm, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); goto _out; @@ -328,7 +328,7 @@ int32_t syncLogBufferAccept(SSyncLogBuffer* pBuf, SSyncNode* pNode, SSyncRaftEnt if (pEntry->term != pExist->term) { (void)syncLogBufferRollback(pBuf, pNode, index); } else { - sTrace("vgId:%d, duplicate log entry received. index: %" PRId64 ", term: %" PRId64 ". log buffer: [%" PRId64 + sTrace("vgId:%d, duplicate log entry received. index:%" PRId64 ", term:%" PRId64 ". log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pEntry->index, pEntry->term, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); @@ -434,7 +434,7 @@ int64_t syncLogBufferProceed(SSyncLogBuffer* pBuf, SSyncNode* pNode, SyncTerm* p // increase match index pBuf->matchIndex = index; - sTrace("vgId:%d, log buffer proceed. start index: %" PRId64 ", match index: %" PRId64 ", end index: %" PRId64, + sTrace("vgId:%d, log buffer proceed. start index:%" PRId64 ", match index:%" PRId64 ", end index:%" PRId64, pNode->vgId, pBuf->startIndex, pBuf->matchIndex, pBuf->endIndex); // replicate on demand @@ -475,7 +475,7 @@ int32_t syncLogFsmExecute(SSyncNode* pNode, SSyncFSM* pFsm, ESyncState role, Syn } if (pEntry->originalRpcType == TDMT_VND_COMMIT) { - sInfo("vgId:%d, fsm execute vnode commit. index: %" PRId64 ", term: %" PRId64 "", pNode->vgId, pEntry->index, + sInfo("vgId:%d, fsm execute vnode commit. index:%" PRId64 ", term:%" PRId64 "", pNode->vgId, pEntry->index, pEntry->term); } @@ -528,7 +528,7 @@ int32_t syncLogBufferCommit(SSyncLogBuffer* pBuf, SSyncNode* pNode, int64_t comm goto _out; } - sTrace("vgId:%d, commit. log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 "), role: %d, term: %" PRId64, + sTrace("vgId:%d, commit. log buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 "), role:%d, term:%" PRId64, pNode->vgId, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex, role, term); // execute in fsm @@ -541,19 +541,19 @@ int32_t syncLogBufferCommit(SSyncLogBuffer* pBuf, SSyncNode* pNode, int64_t comm // execute it if (!syncUtilUserCommit(pEntry->originalRpcType)) { - sInfo("vgId:%d, commit sync barrier. index: %" PRId64 ", term:%" PRId64 ", type: %s", vgId, pEntry->index, + sInfo("vgId:%d, commit sync barrier. index:%" PRId64 ", term:%" PRId64 ", type:%s", vgId, pEntry->index, pEntry->term, TMSG_INFO(pEntry->originalRpcType)); } if (syncLogFsmExecute(pNode, pFsm, role, term, pEntry, 0) != 0) { sError("vgId:%d, failed to execute sync log entry. index:%" PRId64 ", term:%" PRId64 - ", role: %d, current term: %" PRId64, + ", role:%d, current term:%" PRId64, vgId, pEntry->index, pEntry->term, role, term); goto _out; } pBuf->commitIndex = index; - sTrace("vgId:%d, committed index: %" PRId64 ", term: %" PRId64 ", role: %d, current term: %" PRId64 "", pNode->vgId, + sTrace("vgId:%d, committed index:%" PRId64 ", term:%" PRId64 ", role:%d, current term:%" PRId64 "", pNode->vgId, pEntry->index, pEntry->term, role, term); if (!inBuf) { @@ -614,7 +614,7 @@ int32_t syncLogReplMgrRetryOnNeed(SSyncLogReplMgr* pMgr, SSyncNode* pNode) { SRaftId* pDestId = &pNode->replicasId[pMgr->peerId]; if (pMgr->retryBackoff == SYNC_MAX_RETRY_BACKOFF) { syncLogReplMgrReset(pMgr); - sWarn("vgId:%d, reset sync log repl mgr since retry backoff exceeding limit. peer: %" PRIx64, pNode->vgId, + sWarn("vgId:%d, reset sync log repl mgr since retry backoff exceeding limit. peer:%" PRIx64, pNode->vgId, pDestId->addr); return -1; } @@ -639,7 +639,7 @@ int32_t syncLogReplMgrRetryOnNeed(SSyncLogReplMgr* pMgr, SSyncNode* pNode) { if (pMgr->states[pos].acked) { if (pMgr->matchIndex < index && pMgr->states[pos].timeMs + (syncGetRetryMaxWaitMs() << 3) < nowMs) { syncLogReplMgrReset(pMgr); - sWarn("vgId:%d, reset sync log repl mgr since stagnation. index: %" PRId64 ", peer: %" PRIx64, pNode->vgId, + sWarn("vgId:%d, reset sync log repl mgr since stagnation. index:%" PRId64 ", peer:%" PRIx64, pNode->vgId, index, pDestId->addr); goto _out; } @@ -648,7 +648,7 @@ int32_t syncLogReplMgrRetryOnNeed(SSyncLogReplMgr* pMgr, SSyncNode* pNode) { bool barrier = false; if (syncLogReplMgrReplicateOneTo(pMgr, pNode, index, &term, pDestId, &barrier) < 0) { - sError("vgId:%d, failed to replicate sync log entry since %s. index: %" PRId64 ", dest: %" PRIx64 "", pNode->vgId, + sError("vgId:%d, failed to replicate sync log entry since %s. index:%" PRId64 ", dest:%" PRIx64 "", pNode->vgId, terrstr(), index, pDestId->addr); goto _out; } @@ -670,8 +670,8 @@ _out: if (retried) { pMgr->retryBackoff = syncLogGetNextRetryBackoff(pMgr); SSyncLogBuffer* pBuf = pNode->pLogBuf; - sInfo("vgId:%d, resend %d sync log entries. dest: %" PRIx64 ", indexes: %" PRId64 " ..., terms: ... %" PRId64 - ", retryWaitMs: %" PRId64 ", mgr: [%" PRId64 " %" PRId64 ", %" PRId64 "), buffer: [%" PRId64 " %" PRId64 + sInfo("vgId:%d, resend %d sync log entries. dest:%" PRIx64 ", indexes:%" PRId64 " ..., terms: ... %" PRId64 + ", retryWaitMs:%" PRId64 ", mgr: [%" PRId64 " %" PRId64 ", %" PRId64 "), buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, count, pDestId->addr, firstIndex, term, retryWaitMs, pMgr->startIndex, pMgr->matchIndex, pMgr->endIndex, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); @@ -714,7 +714,7 @@ int32_t syncLogReplMgrProcessReplyAsRecovery(SSyncLogReplMgr* pMgr, SSyncNode* p } if (pMsg->success == false && pMsg->matchIndex >= pMsg->lastSendIndex) { - sWarn("vgId:%d, failed to rollback match index. peer: dnode:%d, match index: %" PRId64 ", last sent: %" PRId64, + sWarn("vgId:%d, failed to rollback match index. peer: dnode:%d, match index:%" PRId64 ", last sent:%" PRId64, pNode->vgId, DID(&destId), pMsg->matchIndex, pMsg->lastSendIndex); if (syncNodeStartSnapshot(pNode, &destId) < 0) { sError("vgId:%d, failed to start snapshot for peer dnode:%d", pNode->vgId, DID(&destId)); @@ -761,7 +761,7 @@ int32_t syncLogReplMgrProcessHeartbeatReply(SSyncLogReplMgr* pMgr, SSyncNode* pN SSyncLogBuffer* pBuf = pNode->pLogBuf; taosThreadMutexLock(&pBuf->mutex); if (pMsg->startTime != 0 && pMsg->startTime != pMgr->peerStartTime) { - sInfo("vgId:%d, reset sync log repl mgr in heartbeat. peer: %" PRIx64 ", start time:%" PRId64 ", old:%" PRId64 "", + sInfo("vgId:%d, reset sync log repl mgr in heartbeat. peer:%" PRIx64 ", start time:%" PRId64 ", old:%" PRId64 "", pNode->vgId, pMsg->srcId.addr, pMsg->startTime, pMgr->peerStartTime); syncLogReplMgrReset(pMgr); pMgr->peerStartTime = pMsg->startTime; @@ -774,7 +774,7 @@ int32_t syncLogReplMgrProcessReply(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sync SSyncLogBuffer* pBuf = pNode->pLogBuf; taosThreadMutexLock(&pBuf->mutex); if (pMsg->startTime != pMgr->peerStartTime) { - sInfo("vgId:%d, reset sync log repl mgr in appendlog reply. peer: %" PRIx64 ", start time:%" PRId64 + sInfo("vgId:%d, reset sync log repl mgr in appendlog reply. peer:%" PRIx64 ", start time:%" PRId64 ", old:%" PRId64, pNode->vgId, pMsg->srcId.addr, pMsg->startTime, pMgr->peerStartTime); syncLogReplMgrReset(pMgr); @@ -815,7 +815,7 @@ int32_t syncLogReplMgrReplicateProbe(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sy bool barrier = false; SyncTerm term = -1; if (syncLogReplMgrReplicateOneTo(pMgr, pNode, index, &term, pDestId, &barrier) < 0) { - sError("vgId:%d, failed to replicate log entry since %s. index: %" PRId64 ", dest: 0x%016" PRIx64 "", pNode->vgId, + sError("vgId:%d, failed to replicate log entry since %s. index:%" PRId64 ", dest: 0x%016" PRIx64 "", pNode->vgId, terrstr(), index, pDestId->addr); return -1; } @@ -830,7 +830,7 @@ int32_t syncLogReplMgrReplicateProbe(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sy pMgr->endIndex = index + 1; SSyncLogBuffer* pBuf = pNode->pLogBuf; - sTrace("vgId:%d, probe peer:%" PRIx64 " with msg of index:%" PRId64 " term: %" PRId64 ". mgr (rs:%d): [%" PRId64 + sTrace("vgId:%d, probe peer:%" PRIx64 " with msg of index:%" PRId64 " term:%" PRId64 ". mgr (rs:%d): [%" PRId64 " %" PRId64 ", %" PRId64 "), buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pDestId->addr, index, term, pMgr->restored, pMgr->startIndex, pMgr->matchIndex, pMgr->endIndex, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); @@ -860,7 +860,7 @@ int32_t syncLogReplMgrReplicateAttempt(SSyncLogReplMgr* pMgr, SSyncNode* pNode) bool barrier = false; SyncTerm term = -1; if (syncLogReplMgrReplicateOneTo(pMgr, pNode, index, &term, pDestId, &barrier) < 0) { - sError("vgId:%d, failed to replicate log entry since %s. index: %" PRId64 ", dest: 0x%016" PRIx64 "", pNode->vgId, + sError("vgId:%d, failed to replicate log entry since %s. index:%" PRId64 ", dest: 0x%016" PRIx64 "", pNode->vgId, terrstr(), index, pDestId->addr); return -1; } @@ -874,7 +874,7 @@ int32_t syncLogReplMgrReplicateAttempt(SSyncLogReplMgr* pMgr, SSyncNode* pNode) pMgr->endIndex = index + 1; if (barrier) { - sInfo("vgId:%d, replicated sync barrier to dest: %" PRIx64 ". index: %" PRId64 ", term: %" PRId64 + sInfo("vgId:%d, replicated sync barrier to dest:%" PRIx64 ". index:%" PRId64 ", term:%" PRId64 ", repl mgr: rs(%d) [%" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, pDestId->addr, index, term, pMgr->restored, pMgr->startIndex, pMgr->matchIndex, pMgr->endIndex); @@ -885,7 +885,7 @@ int32_t syncLogReplMgrReplicateAttempt(SSyncLogReplMgr* pMgr, SSyncNode* pNode) syncLogReplMgrRetryOnNeed(pMgr, pNode); SSyncLogBuffer* pBuf = pNode->pLogBuf; - sTrace("vgId:%d, replicated %d msgs to peer: %" PRIx64 ". indexes: %" PRId64 "..., terms: ...%" PRId64 + sTrace("vgId:%d, replicated %d msgs to peer:%" PRIx64 ". indexes:%" PRId64 "..., terms: ...%" PRId64 ", mgr: (rs:%d) [%" PRId64 " %" PRId64 ", %" PRId64 "), buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, count, pDestId->addr, firstIndex, term, pMgr->restored, pMgr->startIndex, pMgr->matchIndex, @@ -1028,7 +1028,7 @@ int32_t syncLogBufferRollback(SSyncLogBuffer* pBuf, SSyncNode* pNode, SyncIndex return 0; } - sInfo("vgId:%d, rollback sync log buffer. toindex: %" PRId64 ", buffer: [%" PRId64 " %" PRId64 " %" PRId64 + sInfo("vgId:%d, rollback sync log buffer. toindex:%" PRId64 ", buffer: [%" PRId64 " %" PRId64 " %" PRId64 ", %" PRId64 ")", pNode->vgId, toIndex, pBuf->startIndex, pBuf->commitIndex, pBuf->matchIndex, pBuf->endIndex); @@ -1119,11 +1119,11 @@ int32_t syncLogReplMgrReplicateOneTo(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sy pEntry = syncLogBufferGetOneEntry(pBuf, pNode, index, &inBuf); if (pEntry == NULL) { - sError("vgId:%d, failed to get raft entry for index: %" PRId64 "", pNode->vgId, index); + sError("vgId:%d, failed to get raft entry for index:%" PRId64 "", pNode->vgId, index); if (terrno == TSDB_CODE_WAL_LOG_NOT_EXIST) { SSyncLogReplMgr* pMgr = syncNodeGetLogReplMgr(pNode, pDestId); if (pMgr) { - sInfo("vgId:%d, reset sync log repl mgr of peer: %" PRIx64 " since %s. index: %" PRId64, pNode->vgId, + sInfo("vgId:%d, reset sync log repl mgr of peer:%" PRIx64 " since %s. index:%" PRId64, pNode->vgId, pDestId->addr, terrstr(), index); (void)syncLogReplMgrReset(pMgr); } @@ -1134,7 +1134,7 @@ int32_t syncLogReplMgrReplicateOneTo(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sy prevLogTerm = syncLogReplMgrGetPrevLogTerm(pMgr, pNode, index); if (prevLogTerm < 0) { - sError("vgId:%d, failed to get prev log term since %s. index: %" PRId64 "", pNode->vgId, terrstr(), index); + sError("vgId:%d, failed to get prev log term since %s. index:%" PRId64 "", pNode->vgId, terrstr(), index); goto _err; } if (pTerm) *pTerm = pEntry->term; @@ -1147,7 +1147,7 @@ int32_t syncLogReplMgrReplicateOneTo(SSyncLogReplMgr* pMgr, SSyncNode* pNode, Sy (void)syncNodeSendAppendEntries(pNode, pDestId, &msgOut); - sTrace("vgId:%d, replicate one msg index: %" PRId64 " term: %" PRId64 " prevterm: %" PRId64 " to dest: 0x%016" PRIx64, + sTrace("vgId:%d, replicate one msg index:%" PRId64 " term:%" PRId64 " prevterm:%" PRId64 " to dest: 0x%016" PRIx64, pNode->vgId, pEntry->index, pEntry->term, prevLogTerm, pDestId->addr); if (!inBuf) { diff --git a/source/libs/sync/src/syncRaftEntry.c b/source/libs/sync/src/syncRaftEntry.c index 988a86cc67..623f1b77a4 100644 --- a/source/libs/sync/src/syncRaftEntry.c +++ b/source/libs/sync/src/syncRaftEntry.c @@ -91,7 +91,7 @@ SSyncRaftEntry* syncEntryBuildNoop(SyncTerm term, SyncIndex index, int32_t vgId) void syncEntryDestroy(SSyncRaftEntry* pEntry) { if (pEntry != NULL) { - sTrace("free entry: %p", pEntry); + sTrace("free entry:%p", pEntry); taosMemoryFree(pEntry); } } diff --git a/source/libs/sync/src/syncUtil.c b/source/libs/sync/src/syncUtil.c index 6a50572cba..97641b8f41 100644 --- a/source/libs/sync/src/syncUtil.c +++ b/source/libs/sync/src/syncUtil.c @@ -43,7 +43,7 @@ void syncUtilNodeInfo2EpSet(const SNodeInfo* pInfo, SEpSet* pEpSet) { bool syncUtilNodeInfo2RaftId(const SNodeInfo* pInfo, SyncGroupId vgId, SRaftId* raftId) { uint32_t ipv4 = taosGetIpv4FromFqdn(pInfo->nodeFqdn); if (ipv4 == 0xFFFFFFFF || ipv4 == 1) { - sError("failed to resolve ipv4 addr, fqdn: %s", pInfo->nodeFqdn); + sError("failed to resolve ipv4 addr, fqdn:%s", pInfo->nodeFqdn); terrno = TSDB_CODE_TSC_INVALID_FQDN; return false; } From 2dcb2811086418430363f2f56b6bfb79e0445e63 Mon Sep 17 00:00:00 2001 From: Shengliang Guan Date: Tue, 7 Feb 2023 02:08:31 +0800 Subject: [PATCH 118/124] fix: should not execute trans if not leader --- source/dnode/mnode/impl/inc/mndTrans.h | 2 +- source/dnode/mnode/impl/src/mndSync.c | 2 +- source/dnode/mnode/impl/src/mndTrans.c | 28 +++++++++++++++++--------- 3 files changed, 21 insertions(+), 11 deletions(-) diff --git a/source/dnode/mnode/impl/inc/mndTrans.h b/source/dnode/mnode/impl/inc/mndTrans.h index 07066d2251..d6e5d322ba 100644 --- a/source/dnode/mnode/impl/inc/mndTrans.h +++ b/source/dnode/mnode/impl/inc/mndTrans.h @@ -81,7 +81,7 @@ int32_t mndTransPrepare(SMnode *pMnode, STrans *pTrans); int32_t mndTransProcessRsp(SRpcMsg *pRsp); void mndTransPullup(SMnode *pMnode); int32_t mndKillTrans(SMnode *pMnode, STrans *pTrans); -void mndTransExecute(SMnode *pMnode, STrans *pTrans); +void mndTransExecute(SMnode *pMnode, STrans *pTrans, bool isLeader); int32_t mndSetRpcInfoForDbTrans(SMnode *pMnode, SRpcMsg *pMsg, EOperType oper, const char *dbname); #ifdef __cplusplus diff --git a/source/dnode/mnode/impl/src/mndSync.c b/source/dnode/mnode/impl/src/mndSync.c index 97ad541c1f..1dc9ae9c4b 100644 --- a/source/dnode/mnode/impl/src/mndSync.c +++ b/source/dnode/mnode/impl/src/mndSync.c @@ -116,7 +116,7 @@ int32_t mndProcessWriteMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta if (pTrans != NULL) { mInfo("trans:%d, execute in mnode which not leader or sync timeout, createTime:%" PRId64 " saved trans:%d", transId, pTrans->createdTime, pMgmt->transId); - mndTransExecute(pMnode, pTrans); + mndTransExecute(pMnode, pTrans, false); mndReleaseTrans(pMnode, pTrans); // sdbWriteFile(pMnode->pSdb, SDB_WRITE_DELTA); } else { diff --git a/source/dnode/mnode/impl/src/mndTrans.c b/source/dnode/mnode/impl/src/mndTrans.c index f931ca121a..6bb286cd6f 100644 --- a/source/dnode/mnode/impl/src/mndTrans.c +++ b/source/dnode/mnode/impl/src/mndTrans.c @@ -905,7 +905,7 @@ int32_t mndTransPrepare(SMnode *pMnode, STrans *pTrans) { pTrans->rpcRsp = NULL; pTrans->rpcRspLen = 0; - mndTransExecute(pMnode, pNew); + mndTransExecute(pMnode, pNew, true); mndReleaseTrans(pMnode, pNew); return 0; } @@ -1068,7 +1068,7 @@ int32_t mndTransProcessRsp(SRpcMsg *pRsp) { mInfo("trans:%d, %s:%d response is received, code:0x%x, accept:0x%x retry:0x%x", transId, mndTransStr(pAction->stage), action, pRsp->code, pAction->acceptableCode, pAction->retryCode); - mndTransExecute(pMnode, pTrans); + mndTransExecute(pMnode, pTrans, true); _OVER: mndReleaseTrans(pMnode, pTrans); @@ -1502,12 +1502,12 @@ static bool mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans) { return continueExec; } -void mndTransExecute(SMnode *pMnode, STrans *pTrans) { +void mndTransExecute(SMnode *pMnode, STrans *pTrans, bool isLeader) { bool continueExec = true; while (continueExec) { - mInfo("trans:%d, continue to execute, stage:%s createTime:%" PRId64, pTrans->id, mndTransStr(pTrans->stage), - pTrans->createdTime); + mInfo("trans:%d, continue to execute, stage:%s createTime:%" PRId64 " leader:%d", pTrans->id, + mndTransStr(pTrans->stage), pTrans->createdTime, isLeader); pTrans->lastExecTime = taosGetTimestampMs(); switch (pTrans->stage) { case TRN_STAGE_PREPARE: @@ -1517,13 +1517,23 @@ void mndTransExecute(SMnode *pMnode, STrans *pTrans) { continueExec = mndTransPerformRedoActionStage(pMnode, pTrans); break; case TRN_STAGE_COMMIT: - continueExec = mndTransPerformCommitStage(pMnode, pTrans); + if (isLeader) { + continueExec = mndTransPerformCommitStage(pMnode, pTrans); + } else { + mInfo("trans:%d, can not commit since not leader", pTrans->id); + continueExec = false; + } break; case TRN_STAGE_COMMIT_ACTION: continueExec = mndTransPerformCommitActionStage(pMnode, pTrans); break; case TRN_STAGE_ROLLBACK: - continueExec = mndTransPerformRollbackStage(pMnode, pTrans); + if (isLeader) { + continueExec = mndTransPerformRollbackStage(pMnode, pTrans); + } else { + mInfo("trans:%d, can not rollback since not leader", pTrans->id); + continueExec = false; + } break; case TRN_STAGE_UNDO_ACTION: continueExec = mndTransPerformUndoActionStage(pMnode, pTrans); @@ -1566,7 +1576,7 @@ int32_t mndKillTrans(SMnode *pMnode, STrans *pTrans) { pAction->errCode = 0; } - mndTransExecute(pMnode, pTrans); + mndTransExecute(pMnode, pTrans, true); return 0; } @@ -1624,7 +1634,7 @@ void mndTransPullup(SMnode *pMnode) { int32_t *pTransId = taosArrayGet(pArray, i); STrans *pTrans = mndAcquireTrans(pMnode, *pTransId); if (pTrans != NULL) { - mndTransExecute(pMnode, pTrans); + mndTransExecute(pMnode, pTrans, true); } mndReleaseTrans(pMnode, pTrans); } From 96cd93db629e84a1b1322485de65db8adaa27f0c Mon Sep 17 00:00:00 2001 From: Xuefeng Tan <1172915550@qq.com> Date: Tue, 7 Feb 2023 10:26:31 +0800 Subject: [PATCH 119/124] enh(taosAdapter): add whether to disable the collection clientIP configuration option (#19819) --- cmake/taosadapter_CMakeLists.txt.in | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/cmake/taosadapter_CMakeLists.txt.in b/cmake/taosadapter_CMakeLists.txt.in index d156057459..8df50af03c 100644 --- a/cmake/taosadapter_CMakeLists.txt.in +++ b/cmake/taosadapter_CMakeLists.txt.in @@ -2,7 +2,7 @@ # taosadapter ExternalProject_Add(taosadapter GIT_REPOSITORY https://github.com/taosdata/taosadapter.git - GIT_TAG 3e08996 + GIT_TAG db6c843 SOURCE_DIR "${TD_SOURCE_DIR}/tools/taosadapter" BINARY_DIR "" #BUILD_IN_SOURCE TRUE From 45d20e3e865baba12090f835c62df7282753ddcd Mon Sep 17 00:00:00 2001 From: Ganlin Zhao <36554565+glzhao89@users.noreply.github.com> Date: Tue, 7 Feb 2023 10:32:02 +0800 Subject: [PATCH 120/124] docs: add more details for interp sub clause zh --- docs/zh/12-taos-sql/10-function.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/zh/12-taos-sql/10-function.md b/docs/zh/12-taos-sql/10-function.md index afe90b8a93..d6ca91ce11 100644 --- a/docs/zh/12-taos-sql/10-function.md +++ b/docs/zh/12-taos-sql/10-function.md @@ -875,9 +875,9 @@ INTERP(expr) - INTERP 用于在指定时间断面获取指定列的记录值，如果该时间断面不存在符合条件的行数据，那么会根据 FILL 参数的设定进行插值。 - INTERP 的输入数据为指定列的数据，可以通过条件语句（where 子句）来对原始列数据进行过滤，如果没有指定过滤条件则输入为全部数据。 - INTERP 需要同时与 RANGE，EVERY 和 FILL 关键字一起使用。 -- INTERP 的输出时间范围根据 RANGE(timestamp1,timestamp2)字段来指定，需满足 timestamp1<=timestamp2。其中 timestamp1（必选值）为输出时间范围的起始值，即如果 timestamp1 时刻符合插值条件则 timestamp1 为输出的第一条记录，timestamp2（必选值）为输出时间范围的结束值，即输出的最后一条记录的 timestamp 不能大于 timestamp2。 -- INTERP 根据 EVERY 字段来确定输出时间范围内的结果条数，即从 timestamp1 开始每隔固定长度的时间（EVERY 值）进行插值。 -- INTERP 根据 FILL 字段来决定在每个符合输出条件的时刻如何进行插值。 +- INTERP 的输出时间范围根据 RANGE(timestamp1,timestamp2)字段来指定，需满足 timestamp1 < timestamp2。其中 timestamp1（必选值）为输出时间范围的起始值，即如果 timestamp1 时刻符合插值条件则 timestamp1 为输出的第一条记录，timestamp2（必选值）为输出时间范围的结束值，即输出的最后一条记录的 timestamp 不能大于 timestamp2。 +- INTERP 根据 EVERY(time_unit) 字段来确定输出时间范围内的结果条数，即从 timestamp1 开始每隔固定长度的时间（time_unit 值）进行插值，time_unit 可取值时间单位：1a(毫秒)，1s(秒)，1m(分)，1h(小时)，1d(天), 1w(周)。 +- INTERP 根据 FILL 字段来决定在每个符合输出条件的时刻如何进行插值。关于 FILL 子句如何使用请参考 [FILL 子句](./distinguished/#fill-子句) - INTERP 只能在一个时间序列内进行插值，因此当作用于超级表时必须跟 partition by tbname 一起使用。 - INTERP 可以与伪列 _irowts 一起使用，返回插值点所对应的时间戳(3.0.1.4版本以后支持)。 From 90ac7efc4529ab16e82be1098dc17152f02aa955 Mon Sep 17 00:00:00 2001 From: Ganlin Zhao <36554565+glzhao89@users.noreply.github.com> Date: Tue, 7 Feb 2023 10:46:43 +0800 Subject: [PATCH 121/124] docs: add more details about interp subclause en --- docs/en/12-taos-sql/10-function.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/docs/en/12-taos-sql/10-function.md b/docs/en/12-taos-sql/10-function.md index f0daf4b82a..d79b73e917 100644 --- a/docs/en/12-taos-sql/10-function.md +++ b/docs/en/12-taos-sql/10-function.md @@ -872,9 +872,9 @@ INTERP(expr) - `INTERP` is used to get the value that matches the specified time slice from a column. If no such value exists an interpolation value will be returned based on `FILL` parameter. - The input data of `INTERP` is the value of the specified column and a `where` clause can be used to filter the original data. If no `where` condition is specified then all original data is the input. - `INTERP` must be used along with `RANGE`, `EVERY`, `FILL` keywords. -- The output time range of `INTERP` is specified by `RANGE(timestamp1,timestamp2)` parameter, with timestamp1<=timestamp2. timestamp1 is the starting point of the output time range and must be specified. timestamp2 is the ending point of the output time range and must be specified. -- The number of rows in the result set of `INTERP` is determined by the parameter `EVERY`. Starting from timestamp1, one interpolation is performed for every time interval specified `EVERY` parameter. The parameter `EVERY` must be an integer, with no quotes, with a time unit of: b(nanosecond), u(microsecond), a(millisecond)), s(second), m(minute), h(hour), d(day), or w(week). For example, `EVERY(500a)` will interpolate every 500 milliseconds. -- Interpolation is performed based on `FILL` parameter. +- The output time range of `INTERP` is specified by `RANGE(timestamp1,timestamp2)` parameter, with timestamp1 < timestamp2. timestamp1 is the starting point of the output time range and must be specified. timestamp2 is the ending point of the output time range and must be specified. +- The number of rows in the result set of `INTERP` is determined by the parameter `EVERY(time_unit)`. Starting from timestamp1, one interpolation is performed for every time interval specified `EVERY` parameter. The parameter `EVERY` must be an integer, with no quotes, with a time unit of: a(millisecond)), s(second), m(minute), h(hour), d(day), or w(week). For example, `EVERY(500a)` will interpolate every 500 milliseconds. +- Interpolation is performed based on `FILL` parameter. For more information about FILL clause, see [FILL Clause](./distinguished/#fill-clause). - `INTERP` can only be used to interpolate in single timeline. So it must be used with `partition by tbname` when it's used on a STable. - Pseudo column `_irowts` can be used along with `INTERP` to return the timestamps associated with interpolation points(support after version 3.0.1.4). From 35675aeb2a3d1ed30784b69e2320337ff6e3129c Mon Sep 17 00:00:00 2001 From: Ganlin Zhao <36554565+glzhao89@users.noreply.github.com> Date: Tue, 7 Feb 2023 10:49:42 +0800 Subject: [PATCH 122/124] Update 10-function.md --- docs/en/12-taos-sql/10-function.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/12-taos-sql/10-function.md b/docs/en/12-taos-sql/10-function.md index d79b73e917..1d1a4560f1 100644 --- a/docs/en/12-taos-sql/10-function.md +++ b/docs/en/12-taos-sql/10-function.md @@ -873,7 +873,7 @@ INTERP(expr) - The input data of `INTERP` is the value of the specified column and a `where` clause can be used to filter the original data. If no `where` condition is specified then all original data is the input. - `INTERP` must be used along with `RANGE`, `EVERY`, `FILL` keywords. - The output time range of `INTERP` is specified by `RANGE(timestamp1,timestamp2)` parameter, with timestamp1 < timestamp2. timestamp1 is the starting point of the output time range and must be specified. timestamp2 is the ending point of the output time range and must be specified. -- The number of rows in the result set of `INTERP` is determined by the parameter `EVERY(time_unit)`. Starting from timestamp1, one interpolation is performed for every time interval specified `EVERY` parameter. The parameter `EVERY` must be an integer, with no quotes, with a time unit of: a(millisecond)), s(second), m(minute), h(hour), d(day), or w(week). For example, `EVERY(500a)` will interpolate every 500 milliseconds. +- The number of rows in the result set of `INTERP` is determined by the parameter `EVERY(time_unit)`. Starting from timestamp1, one interpolation is performed for every time interval specified `time_unit` parameter. The parameter `time_unit` must be an integer, with no quotes, with a time unit of: a(millisecond)), s(second), m(minute), h(hour), d(day), or w(week). For example, `EVERY(500a)` will interpolate every 500 milliseconds. - Interpolation is performed based on `FILL` parameter. For more information about FILL clause, see [FILL Clause](./distinguished/#fill-clause). - `INTERP` can only be used to interpolate in single timeline. So it must be used with `partition by tbname` when it's used on a STable. - Pseudo column `_irowts` can be used along with `INTERP` to return the timestamps associated with interpolation points(support after version 3.0.1.4). From 60c0f0175a15d974cbf221d069f750088c8a3fc7 Mon Sep 17 00:00:00 2001 From: Ganlin Zhao <36554565+glzhao89@users.noreply.github.com> Date: Tue, 7 Feb 2023 10:54:19 +0800 Subject: [PATCH 123/124] Update 10-function.md --- docs/zh/12-taos-sql/10-function.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/zh/12-taos-sql/10-function.md b/docs/zh/12-taos-sql/10-function.md index d6ca91ce11..c2fe0f0cb6 100644 --- a/docs/zh/12-taos-sql/10-function.md +++ b/docs/zh/12-taos-sql/10-function.md @@ -876,7 +876,7 @@ INTERP(expr) - INTERP 的输入数据为指定列的数据，可以通过条件语句（where 子句）来对原始列数据进行过滤，如果没有指定过滤条件则输入为全部数据。 - INTERP 需要同时与 RANGE，EVERY 和 FILL 关键字一起使用。 - INTERP 的输出时间范围根据 RANGE(timestamp1,timestamp2)字段来指定，需满足 timestamp1 < timestamp2。其中 timestamp1（必选值）为输出时间范围的起始值，即如果 timestamp1 时刻符合插值条件则 timestamp1 为输出的第一条记录，timestamp2（必选值）为输出时间范围的结束值，即输出的最后一条记录的 timestamp 不能大于 timestamp2。 -- INTERP 根据 EVERY(time_unit) 字段来确定输出时间范围内的结果条数，即从 timestamp1 开始每隔固定长度的时间（time_unit 值）进行插值，time_unit 可取值时间单位：1a(毫秒)，1s(秒)，1m(分)，1h(小时)，1d(天), 1w(周)。 +- INTERP 根据 EVERY(time_unit) 字段来确定输出时间范围内的结果条数，即从 timestamp1 开始每隔固定长度的时间（time_unit 值）进行插值，time_unit 可取值时间单位：1a(毫秒)，1s(秒)，1m(分)，1h(小时)，1d(天)，1w(周)。例如 EVERY(500a) 将对于指定数据每500毫秒间隔进行一次插值. - INTERP 根据 FILL 字段来决定在每个符合输出条件的时刻如何进行插值。关于 FILL 子句如何使用请参考 [FILL 子句](./distinguished/#fill-子句) - INTERP 只能在一个时间序列内进行插值，因此当作用于超级表时必须跟 partition by tbname 一起使用。 - INTERP 可以与伪列 _irowts 一起使用，返回插值点所对应的时间戳(3.0.1.4版本以后支持)。 From 5f89fe7652a7c5b5bf0bec1415629d266fe17c8b Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Tue, 7 Feb 2023 11:01:52 +0800 Subject: [PATCH 124/124] fix(query):add check for no disk error. --- source/libs/executor/src/tsort.c | 21 ++++++--- source/util/src/tpagedbuf.c | 75 +++++++++++++++----------------- 2 files changed, 50 insertions(+), 46 deletions(-) diff --git a/source/libs/executor/src/tsort.c b/source/libs/executor/src/tsort.c index 661e9f97b7..90f20f40b8 100644 --- a/source/libs/executor/src/tsort.c +++ b/source/libs/executor/src/tsort.c @@ -190,8 +190,9 @@ static int32_t doAddToBuf(SSDataBlock* pDataBlock, SSortHandle* pHandle) { qError("Add to buf failed since %s", terrstr(terrno)); return terrno; } + int32_t code = createDiskbasedBuf(&pHandle->pBuf, pHandle->pageSize, pHandle->numOfPages * pHandle->pageSize, - "doAddToBuf", tsTempDir); + "sortExternalBuf", tsTempDir); dBufSetPrintInfo(pHandle->pBuf); if (code != TSDB_CODE_SUCCESS) { return code; @@ -635,6 +636,7 @@ int32_t getProperSortPageSize(size_t rowSize, uint32_t numOfCols) { static int32_t createInitialSources(SSortHandle* pHandle) { size_t sortBufSize = pHandle->numOfPages * pHandle->pageSize; + int32_t code = 0; if (pHandle->type == SORT_SINGLESOURCE_SORT) { SSortSource** pSource = taosArrayGet(pHandle->pOrderedSource, 0); @@ -663,8 +665,8 @@ static int32_t createInitialSources(SSortHandle* pHandle) { pHandle->beforeFp(pBlock, pHandle->param); } - int32_t code = blockDataMerge(pHandle->pDataBlock, pBlock); - if (code != 0) { + code = blockDataMerge(pHandle->pDataBlock, pBlock); + if (code != TSDB_CODE_SUCCESS) { if (source->param && !source->onlyRef) { taosMemoryFree(source->param); } @@ -689,6 +691,7 @@ static int32_t createInitialSources(SSortHandle* pHandle) { blockDataDestroy(source->src.pBlock); source->src.pBlock = NULL; } + taosMemoryFree(source); return code; } @@ -696,13 +699,17 @@ static int32_t createInitialSources(SSortHandle* pHandle) { int64_t el = taosGetTimestampUs() - p; pHandle->sortElapsed += el; - doAddToBuf(pHandle->pDataBlock, pHandle); + code = doAddToBuf(pHandle->pDataBlock, pHandle); + if (code != TSDB_CODE_SUCCESS) { + return code; + } } } if (source->param && !source->onlyRef) { taosMemoryFree(source->param); } + taosMemoryFree(source); if (pHandle->pDataBlock != NULL && pHandle->pDataBlock->info.rows > 0) { @@ -711,7 +718,7 @@ static int32_t createInitialSources(SSortHandle* pHandle) { // Perform the in-memory sort and then flush data in the buffer into disk. int64_t p = taosGetTimestampUs(); - int32_t code = blockDataSort(pHandle->pDataBlock, pHandle->pSortInfo); + code = blockDataSort(pHandle->pDataBlock, pHandle->pSortInfo); if (code != 0) { return code; } @@ -729,12 +736,12 @@ static int32_t createInitialSources(SSortHandle* pHandle) { pHandle->tupleHandle.pBlock = pHandle->pDataBlock; return 0; } else { - doAddToBuf(pHandle->pDataBlock, pHandle); + code = doAddToBuf(pHandle->pDataBlock, pHandle); } } } - return TSDB_CODE_SUCCESS; + return code; } int32_t tsortOpen(SSortHandle* pHandle) { diff --git a/source/util/src/tpagedbuf.c b/source/util/src/tpagedbuf.c index 7c60862c56..7ed7b2fbfe 100644 --- a/source/util/src/tpagedbuf.c +++ b/source/util/src/tpagedbuf.c @@ -126,6 +126,30 @@ static uint64_t allocateNewPositionInFile(SDiskbasedBuf* pBuf, size_t size) { static FORCE_INLINE size_t getAllocPageSize(int32_t pageSize) { return pageSize + POINTER_BYTES + sizeof(SFilePage); } +static int32_t doFlushBufPageImpl(SDiskbasedBuf* pBuf, int64_t offset, const char* pData, int32_t size) { + int32_t ret = taosLSeekFile(pBuf->pFile, offset, SEEK_SET); + if (ret == -1) { + terrno = TAOS_SYSTEM_ERROR(errno); + return terrno; + } + + ret = (int32_t)taosWriteFile(pBuf->pFile, pData, size); + if (ret != size) { + terrno = TAOS_SYSTEM_ERROR(errno); + return terrno; + } + + // extend the file + if (pBuf->fileSize < offset + size) { + pBuf->fileSize = offset + size; + } + + pBuf->statis.flushBytes += size; + pBuf->statis.flushPages += 1; + + return TSDB_CODE_SUCCESS; +} + static char* doFlushBufPage(SDiskbasedBuf* pBuf, SPageInfo* pg) { if (pg->pData == NULL || pg->used) { uError("invalid params in paged buffer process when flushing buf to disk, %s", pBuf->id); @@ -134,7 +158,9 @@ static char* doFlushBufPage(SDiskbasedBuf* pBuf, SPageInfo* pg) { } int32_t size = pBuf->pageSize; - char* t = NULL; + int64_t offset = pg->offset; + + char* t = NULL; if ((!HAS_DATA_IN_DISK(pg)) || pg->dirty) { void* payload = GET_PAYLOAD_DATA(pg); t = doCompressData(payload, pBuf->pageSize, &size, pBuf); @@ -147,59 +173,29 @@ static char* doFlushBufPage(SDiskbasedBuf* pBuf, SPageInfo* pg) { // this page is flushed to disk for the first time if (pg->dirty) { if (!HAS_DATA_IN_DISK(pg)) { - pg->offset = allocateNewPositionInFile(pBuf, size); + offset = allocateNewPositionInFile(pBuf, size); pBuf->nextPos += size; - int32_t ret = taosLSeekFile(pBuf->pFile, pg->offset, SEEK_SET); - if (ret == -1) { - terrno = TAOS_SYSTEM_ERROR(errno); + int32_t code = doFlushBufPageImpl(pBuf, offset, t, size); + if (code != TSDB_CODE_SUCCESS) { return NULL; } - - ret = (int32_t)taosWriteFile(pBuf->pFile, t, size); - if (ret != size) { - terrno = TAOS_SYSTEM_ERROR(errno); - return NULL; - } - - // extend the file size - if (pBuf->fileSize < pg->offset + size) { - pBuf->fileSize = pg->offset + size; - } - - pBuf->statis.flushBytes += size; - pBuf->statis.flushPages += 1; } else { // length becomes greater, current space is not enough, allocate new place, otherwise, do nothing if (pg->length < size) { // 1. add current space to free list - SPageDiskInfo dinfo = {.length = pg->length, .offset = pg->offset}; + SPageDiskInfo dinfo = {.length = pg->length, .offset = offset}; taosArrayPush(pBuf->pFree, &dinfo); // 2. allocate new position, and update the info - pg->offset = allocateNewPositionInFile(pBuf, size); + offset = allocateNewPositionInFile(pBuf, size); pBuf->nextPos += size; } - // 3. write to disk. - int32_t ret = taosLSeekFile(pBuf->pFile, pg->offset, SEEK_SET); - if (ret == -1) { - terrno = TAOS_SYSTEM_ERROR(errno); + int32_t code = doFlushBufPageImpl(pBuf, offset, t, size); + if (code != TSDB_CODE_SUCCESS) { return NULL; } - - ret = (int32_t)taosWriteFile(pBuf->pFile, t, size); - if (ret != size) { - terrno = TAOS_SYSTEM_ERROR(errno); - return NULL; - } - - if (pBuf->fileSize < pg->offset + size) { - pBuf->fileSize = pg->offset + size; - } - - pBuf->statis.flushBytes += size; - pBuf->statis.flushPages += 1; } } else { // NOTE: the size may be -1, the this recycle page has not been flushed to disk yet. size = pg->length; @@ -209,9 +205,10 @@ static char* doFlushBufPage(SDiskbasedBuf* pBuf, SPageInfo* pg) { memset(pDataBuf, 0, getAllocPageSize(pBuf->pageSize)); #ifdef BUF_PAGE_DEBUG - uDebug("page_flush %p, pageId:%d, offset:%d", pDataBuf, pg->pageId, pg->offset); + uDebug("page_flush %p, pageId:%d, offset:%d", pDataBuf, pg->pageId, offset); #endif + pg->offset = offset; pg->length = size; // on disk size return pDataBuf; }