From a6f33ba01c987682f8212ff8dfba8e6d7c176151 Mon Sep 17 00:00:00 2001 From: Minghao Li Date: Sat, 25 Jun 2022 20:31:42 +0800 Subject: [PATCH 01/10] refactor(sync): optimized one replica --- include/libs/sync/syncTools.h | 4 +- source/dnode/mnode/impl/src/mndMain.c | 10 ++--- source/dnode/vnode/src/vnd/vnodeSync.c | 60 ++++++++++++++------------ source/libs/sync/inc/syncIO.h | 2 +- source/libs/sync/inc/syncInt.h | 3 +- source/libs/sync/src/syncCommit.c | 1 + source/libs/sync/src/syncIO.c | 2 +- source/libs/sync/src/syncMain.c | 57 ++++++++++++++++++++---- 8 files changed, 92 insertions(+), 47 deletions(-) diff --git a/include/libs/sync/syncTools.h b/include/libs/sync/syncTools.h index e51b20a404..5d892352d6 100644 --- a/include/libs/sync/syncTools.h +++ b/include/libs/sync/syncTools.h @@ -524,7 +524,7 @@ void syncReconfigFinishLog2(char* s, const SyncReconfigFinish* pMsg); int32_t syncNodeOnPingCb(SSyncNode* ths, SyncPing* pMsg); int32_t syncNodeOnPingReplyCb(SSyncNode* ths, SyncPingReply* pMsg); int32_t syncNodeOnTimeoutCb(SSyncNode* ths, SyncTimeout* pMsg); -int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg); +int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg, SyncIndex* pRetIndex); int32_t syncNodeOnRequestVoteCb(SSyncNode* ths, SyncRequestVote* pMsg); int32_t syncNodeOnRequestVoteReplyCb(SSyncNode* ths, SyncRequestVoteReply* pMsg); int32_t syncNodeOnAppendEntriesCb(SSyncNode* ths, SyncAppendEntries* pMsg); @@ -541,7 +541,7 @@ int32_t syncNodeOnSnapshotRspCb(SSyncNode* ths, SyncSnapshotRsp* pMsg); // ----------------------------------------- typedef int32_t (*FpOnPingCb)(SSyncNode* ths, SyncPing* pMsg); typedef int32_t (*FpOnPingReplyCb)(SSyncNode* ths, SyncPingReply* pMsg); -typedef int32_t (*FpOnClientRequestCb)(SSyncNode* ths, SyncClientRequest* pMsg); +typedef int32_t (*FpOnClientRequestCb)(SSyncNode* ths, SyncClientRequest* pMsg, SyncIndex* pRetIndex); typedef int32_t (*FpOnRequestVoteCb)(SSyncNode* ths, SyncRequestVote* pMsg); typedef int32_t (*FpOnRequestVoteReplyCb)(SSyncNode* ths, SyncRequestVoteReply* pMsg); typedef int32_t (*FpOnAppendEntriesCb)(SSyncNode* ths, SyncAppendEntries* pMsg); diff --git a/source/dnode/mnode/impl/src/mndMain.c b/source/dnode/mnode/impl/src/mndMain.c index 000e1041d0..ead7eaaeb6 100644 --- a/source/dnode/mnode/impl/src/mndMain.c +++ b/source/dnode/mnode/impl/src/mndMain.c @@ -442,7 +442,7 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) { syncPingReplyDestroy(pSyncMsg); } else if (pMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) { SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pMsg); - code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg); + code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL); syncClientRequestDestroy(pSyncMsg); } else if (pMsg->msgType == TDMT_SYNC_REQUEST_VOTE) { SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pMsg); @@ -491,7 +491,7 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) { syncPingReplyDestroy(pSyncMsg); } else if (pMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) { SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pMsg); - code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg); + code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL); syncClientRequestDestroy(pSyncMsg); } else if (pMsg->msgType == TDMT_SYNC_REQUEST_VOTE) { SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pMsg); @@ -555,10 +555,10 @@ static int32_t mndCheckMnodeState(SRpcMsg *pMsg) { static int32_t mndCheckMsgContent(SRpcMsg *pMsg) { if (!IsReq(pMsg)) return 0; if (pMsg->contLen != 0 && pMsg->pCont != NULL) return 0; - + const STraceId *trace = &pMsg->info.traceId; mGError("msg:%p, failed to check msg, cont:%p contLen:%d, app:%p type:%s", pMsg, pMsg->pCont, pMsg->contLen, - pMsg->info.ahandle, TMSG_INFO(pMsg->msgType)); + pMsg->info.ahandle, TMSG_INFO(pMsg->msgType)); terrno = TSDB_CODE_INVALID_MSG_LEN; return -1; } @@ -723,7 +723,7 @@ int32_t mndGetMonitorInfo(SMnode *pMnode, SMonClusterInfo *pClusterInfo, SMonVgr pIter = sdbFetch(pSdb, SDB_STB, pIter, (void **)&pStb); if (pIter == NULL) break; - SMonStbDesc desc = {0}; + SMonStbDesc desc = {0}; SName name1 = {0}; tNameFromString(&name1, pStb->db, T_NAME_ACCT | T_NAME_DB | T_NAME_TABLE); diff --git a/source/dnode/vnode/src/vnd/vnodeSync.c b/source/dnode/vnode/src/vnd/vnodeSync.c index 4c56180979..ba68e7811d 100644 --- a/source/dnode/vnode/src/vnd/vnodeSync.c +++ b/source/dnode/vnode/src/vnd/vnodeSync.c @@ -137,6 +137,26 @@ void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) { vError("vgId:%d, failed to pre-process msg:%p since %s", vgId, pMsg, terrstr()); } else { code = syncPropose(pVnode->sync, pMsg, vnodeIsMsgWeak(pMsg->msgType)); + if (code == 1) { + do { + static int32_t cnt = 0; + if (cnt++ % 1000 == 1) { + vInfo("vgId:%d, msg:%p apply right now, apply index:%ld, msgtype:%s,%d", vgId, pMsg, + pMsg->info.conn.applyIndex, TMSG_INFO(pMsg->msgType), pMsg->msgType); + } + } while (0); + + SRpcMsg rsp = {.code = pMsg->code, .info = pMsg->info}; + if (vnodeProcessWriteReq(pVnode, pMsg, pMsg->info.conn.applyIndex, &rsp) < 0) { + rsp.code = terrno; + vInfo("vgId:%d, msg:%p failed to apply right now since %s", vgId, pMsg, terrstr()); + } + + if (rsp.info.handle != NULL) { + tmsgSendRsp(&rsp); + } + code = 0; + } } } @@ -260,7 +280,7 @@ int32_t vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) { SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pRpcMsg); assert(pSyncMsg != NULL); - ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg); + ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL); syncClientRequestDestroy(pSyncMsg); } else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE) { @@ -359,34 +379,18 @@ static void vnodeSyncCommitMsg(SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta c SyncIndex beginIndex = SYNC_INDEX_INVALID; char logBuf[256] = {0}; - if (pFsm->FpGetSnapshotInfo != NULL) { - (*pFsm->FpGetSnapshotInfo)(pFsm, &snapshot); - beginIndex = snapshot.lastApplyIndex; - } + snprintf(logBuf, sizeof(logBuf), + "==callback== ==CommitCb== execute, pFsm:%p, index:%ld, isWeak:%d, code:%d, state:%d %s, beginIndex :%ld\n", + pFsm, cbMeta.index, cbMeta.isWeak, cbMeta.code, cbMeta.state, syncUtilState2String(cbMeta.state), + beginIndex); + syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg); - if (cbMeta.index > beginIndex) { - snprintf( - logBuf, sizeof(logBuf), - "==callback== ==CommitCb== execute, pFsm:%p, index:%ld, isWeak:%d, code:%d, state:%d %s, beginIndex :%ld\n", - pFsm, cbMeta.index, cbMeta.isWeak, cbMeta.code, cbMeta.state, syncUtilState2String(cbMeta.state), beginIndex); - syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg); - - SRpcMsg rpcMsg = {.msgType = pMsg->msgType, .contLen = pMsg->contLen}; - rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen); - memcpy(rpcMsg.pCont, pMsg->pCont, pMsg->contLen); - syncGetAndDelRespRpc(pVnode->sync, cbMeta.seqNum, &rpcMsg.info); - rpcMsg.info.conn.applyIndex = cbMeta.index; - tmsgPutToQueue(&pVnode->msgCb, APPLY_QUEUE, &rpcMsg); - - } else { - char logBuf[256] = {0}; - snprintf(logBuf, sizeof(logBuf), - "==callback== ==CommitCb== do not execute, pFsm:%p, index:%ld, isWeak:%d, code:%d, state:%d %s, " - "beginIndex :%ld\n", - pFsm, cbMeta.index, cbMeta.isWeak, cbMeta.code, cbMeta.state, syncUtilState2String(cbMeta.state), - beginIndex); - syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg); - } + SRpcMsg rpcMsg = {.msgType = pMsg->msgType, .contLen = pMsg->contLen}; + rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen); + memcpy(rpcMsg.pCont, pMsg->pCont, pMsg->contLen); + syncGetAndDelRespRpc(pVnode->sync, cbMeta.seqNum, &rpcMsg.info); + rpcMsg.info.conn.applyIndex = cbMeta.index; + tmsgPutToQueue(&pVnode->msgCb, APPLY_QUEUE, &rpcMsg); } static void vnodeSyncPreCommitMsg(SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta cbMeta) { diff --git a/source/libs/sync/inc/syncIO.h b/source/libs/sync/inc/syncIO.h index 88d3065375..21ae40f92d 100644 --- a/source/libs/sync/inc/syncIO.h +++ b/source/libs/sync/inc/syncIO.h @@ -50,7 +50,7 @@ typedef struct SSyncIO { void *pSyncNode; int32_t (*FpOnSyncPing)(SSyncNode *pSyncNode, SyncPing *pMsg); int32_t (*FpOnSyncPingReply)(SSyncNode *pSyncNode, SyncPingReply *pMsg); - int32_t (*FpOnSyncClientRequest)(SSyncNode *pSyncNode, SyncClientRequest *pMsg); + int32_t (*FpOnSyncClientRequest)(SSyncNode *pSyncNode, SyncClientRequest *pMsg, SyncIndex* pRetIndex); int32_t (*FpOnSyncRequestVote)(SSyncNode *pSyncNode, SyncRequestVote *pMsg); int32_t (*FpOnSyncRequestVoteReply)(SSyncNode *pSyncNode, SyncRequestVoteReply *pMsg); int32_t (*FpOnSyncAppendEntries)(SSyncNode *pSyncNode, SyncAppendEntries *pMsg); diff --git a/source/libs/sync/inc/syncInt.h b/source/libs/sync/inc/syncInt.h index 1bd0671fb5..d351dc50f4 100644 --- a/source/libs/sync/inc/syncInt.h +++ b/source/libs/sync/inc/syncInt.h @@ -169,7 +169,7 @@ SSyncNode* syncNodeOpen(const SSyncInfo* pSyncInfo); void syncNodeStart(SSyncNode* pSyncNode); void syncNodeStartStandBy(SSyncNode* pSyncNode); void syncNodeClose(SSyncNode* pSyncNode); -int32_t syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak); +int32_t syncNodePropose(SSyncNode* pSyncNode, SRpcMsg* pMsg, bool isWeak); // option bool syncNodeSnapshotEnable(SSyncNode* pSyncNode); @@ -233,6 +233,7 @@ SyncIndex syncNodeGetPreIndex(SSyncNode* pSyncNode, SyncIndex index); SyncTerm syncNodeGetPreTerm(SSyncNode* pSyncNode, SyncIndex index); int32_t syncNodeGetPreIndexTerm(SSyncNode* pSyncNode, SyncIndex index, SyncIndex* pPreIndex, SyncTerm* pPreTerm); +bool syncNodeIsOptimizedOneReplica(SSyncNode* ths, SRpcMsg* pMsg); int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, uint64_t flag); int32_t syncNodeUpdateNewConfigIndex(SSyncNode* ths, SSyncCfg* pNewCfg); diff --git a/source/libs/sync/src/syncCommit.c b/source/libs/sync/src/syncCommit.c index c92edae381..982d37826a 100644 --- a/source/libs/sync/src/syncCommit.c +++ b/source/libs/sync/src/syncCommit.c @@ -102,6 +102,7 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) { } } + // maybe execute fsm if (newCommitIndex > pSyncNode->commitIndex) { SyncIndex beginIndex = pSyncNode->commitIndex + 1; SyncIndex endIndex = newCommitIndex; diff --git a/source/libs/sync/src/syncIO.c b/source/libs/sync/src/syncIO.c index 038c36c417..4243a933b0 100644 --- a/source/libs/sync/src/syncIO.c +++ b/source/libs/sync/src/syncIO.c @@ -281,7 +281,7 @@ static void *syncIOConsumerFunc(void *param) { if (io->FpOnSyncClientRequest != NULL) { SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pRpcMsg); ASSERT(pSyncMsg != NULL); - io->FpOnSyncClientRequest(io->pSyncNode, pSyncMsg); + io->FpOnSyncClientRequest(io->pSyncNode, pSyncMsg, NULL); syncClientRequestDestroy(pSyncMsg); } diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index 3100d0525c..2c00dccb60 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -50,7 +50,7 @@ static int32_t syncNodeAppendNoop(SSyncNode* ths); // process message ---- int32_t syncNodeOnPingCb(SSyncNode* ths, SyncPing* pMsg); int32_t syncNodeOnPingReplyCb(SSyncNode* ths, SyncPingReply* pMsg); -int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg); +int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg, SyncIndex* pRetIndex); // life cycle static void syncFreeNode(void* param); @@ -627,7 +627,7 @@ int32_t syncPropose(int64_t rid, SRpcMsg* pMsg, bool isWeak) { return ret; } -int32_t syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak) { +int32_t syncNodePropose(SSyncNode* pSyncNode, SRpcMsg* pMsg, bool isWeak) { int32_t ret = 0; char eventLog[128]; @@ -664,13 +664,34 @@ int32_t syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak) SRpcMsg rpcMsg; syncClientRequest2RpcMsg(pSyncMsg, &rpcMsg); - if (pSyncNode->FpEqMsg != NULL && (*pSyncNode->FpEqMsg)(pSyncNode->msgcb, &rpcMsg) == 0) { - ret = 0; + // optimized one replica + if (syncNodeIsOptimizedOneReplica(pSyncNode, pMsg)) { + SyncIndex retIndex; + int32_t code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, &retIndex); + if (code == 0) { + pMsg->info.conn.applyIndex = retIndex; + rpcFreeCont(rpcMsg.pCont); + syncRespMgrDel(pSyncNode->pSyncRespMgr, seqNum); + ret = 1; + sDebug("vgId:%d optimized index:%ld success, msgtype:%s,%d", pSyncNode->vgId, retIndex, + TMSG_INFO(pMsg->msgType), pMsg->msgType); + } else { + ret = -1; + terrno = TSDB_CODE_SYN_INTERNAL_ERROR; + sError("vgId:%d optimized index:%ld error, msgtype:%s,%d", pSyncNode->vgId, retIndex, TMSG_INFO(pMsg->msgType), + pMsg->msgType); + } + } else { - ret = -1; - terrno = TSDB_CODE_SYN_INTERNAL_ERROR; - sError("syncPropose pSyncNode->FpEqMsg is NULL"); + if (pSyncNode->FpEqMsg != NULL && (*pSyncNode->FpEqMsg)(pSyncNode->msgcb, &rpcMsg) == 0) { + ret = 0; + } else { + ret = -1; + terrno = TSDB_CODE_SYN_INTERNAL_ERROR; + sError("enqueue msg error, FpEqMsg is NULL"); + } } + syncClientRequestDestroy(pSyncMsg); goto _END; @@ -2377,7 +2398,7 @@ int32_t syncNodeOnPingReplyCb(SSyncNode* ths, SyncPingReply* pMsg) { // /\ UNCHANGED <> // -int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg) { +int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg, SyncIndex* pRetIndex) { int32_t ret = 0; syncClientRequestLog2("==syncNodeOnClientRequestCb==", pMsg); @@ -2436,6 +2457,14 @@ int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg) { rpcFreeCont(rpcMsg.pCont); } + if (pRetIndex != NULL) { + if (ret == 0 && pEntry != NULL) { + *pRetIndex = pEntry->index; + } else { + *pRetIndex = SYNC_INDEX_INVALID; + } + } + syncEntryDestory(pEntry); return ret; } @@ -2600,6 +2629,10 @@ static int32_t syncNodeProposeConfigChangeFinish(SSyncNode* ths, SyncReconfigFin return 0; } +bool syncNodeIsOptimizedOneReplica(SSyncNode* ths, SRpcMsg* pMsg) { + return (ths->replicaNum == 1 && syncUtilUserCommit(pMsg->msgType) && ths->vgId != 1); +} + int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, uint64_t flag) { int32_t code = 0; ESyncState state = flag; @@ -2621,7 +2654,13 @@ int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, syncEntry2OriginalRpc(pEntry, &rpcMsg); // user commit - if (ths->pFsm->FpCommitCb != NULL && syncUtilUserCommit(pEntry->originalRpcType)) { + bool internalExecute = (ths->pFsm->FpCommitCb != NULL) && syncUtilUserCommit(pEntry->originalRpcType); + if (ths->replicaNum == 1) { + internalExecute = syncNodeIsOptimizedOneReplica(ths, &rpcMsg) && !(ths->restoreFinish); + } + + // execute fsm in apply thread, or execute outside syncPropose + if (internalExecute) { SFsmCbMeta cbMeta = {0}; cbMeta.index = pEntry->index; cbMeta.lastConfigIndex = syncNodeGetSnapshotConfigIndex(ths, cbMeta.index); From 6eecbcd1a6500e9052cf86b43aa5f58d0cbcfe44 Mon Sep 17 00:00:00 2001 From: Minghao Li Date: Sun, 26 Jun 2022 01:16:11 +0800 Subject: [PATCH 02/10] refactor(sync): optimized one replica --- include/libs/sync/sync.h | 4 +-- source/dnode/vnode/src/vnd/vnodeSync.c | 11 +++--- source/libs/sync/inc/syncEnv.h | 12 +++---- source/libs/sync/inc/syncIO.h | 2 +- source/libs/sync/inc/syncRaftCfg.h | 10 +++--- source/libs/sync/src/syncIO.c | 6 ++-- source/libs/sync/src/syncIndexMgr.c | 2 +- source/libs/sync/src/syncMain.c | 42 +++++++++++++--------- source/libs/sync/src/syncRaftCfg.c | 8 ++--- source/libs/sync/src/syncRaftStore.c | 2 +- source/libs/sync/src/syncSnapshot.c | 10 +++--- source/libs/sync/src/syncVoteMgr.c | 4 +-- source/libs/sync/test/syncRaftLogTest2.cpp | 27 +++++++------- source/libs/sync/test/syncRaftLogTest3.cpp | 2 +- 14 files changed, 74 insertions(+), 68 deletions(-) diff --git a/include/libs/sync/sync.h b/include/libs/sync/sync.h index 52d9b6f810..5fa7eed40c 100644 --- a/include/libs/sync/sync.h +++ b/include/libs/sync/sync.h @@ -26,9 +26,9 @@ extern "C" { extern bool gRaftDetailLog; -#define SYNC_INDEX_BEGIN 0 +#define SYNC_INDEX_BEGIN 0 #define SYNC_INDEX_INVALID -1 -#define SYNC_TERM_INVALID 0xFFFFFFFFFFFFFFFF +#define SYNC_TERM_INVALID 0xFFFFFFFFFFFFFFFF typedef uint64_t SyncNodeId; typedef int32_t SyncGroupId; diff --git a/source/dnode/vnode/src/vnd/vnodeSync.c b/source/dnode/vnode/src/vnd/vnodeSync.c index ba68e7811d..b998f40195 100644 --- a/source/dnode/vnode/src/vnd/vnodeSync.c +++ b/source/dnode/vnode/src/vnd/vnodeSync.c @@ -155,7 +155,6 @@ void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) { if (rsp.info.handle != NULL) { tmsgSendRsp(&rsp); } - code = 0; } } } @@ -183,10 +182,12 @@ void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) { SRpcMsg rsp = {.code = TSDB_CODE_RPC_REDIRECT, .info = pMsg->info}; tmsgSendRedirectRsp(&rsp, &newEpSet); } else { - if (terrno != 0) code = terrno; - vError("vgId:%d, msg:%p failed to propose since %s, code:0x%x", vgId, pMsg, tstrerror(code), code); - SRpcMsg rsp = {.code = code, .info = pMsg->info}; - tmsgSendRsp(&rsp); + if (code != 1) { + if (terrno != 0) code = terrno; + vError("vgId:%d, msg:%p failed to propose since %s, code:0x%x", vgId, pMsg, tstrerror(code), code); + SRpcMsg rsp = {.code = code, .info = pMsg->info}; + tmsgSendRsp(&rsp); + } } vGTrace("vgId:%d, msg:%p is freed, code:0x%x", vgId, pMsg, code); diff --git a/source/libs/sync/inc/syncEnv.h b/source/libs/sync/inc/syncEnv.h index 2e8e0e1227..beddec64c5 100644 --- a/source/libs/sync/inc/syncEnv.h +++ b/source/libs/sync/inc/syncEnv.h @@ -28,13 +28,13 @@ extern "C" { #include "trpc.h" #include "ttimer.h" -#define TIMER_MAX_MS 0x7FFFFFFF -#define ENV_TICK_TIMER_MS 1000 -#define PING_TIMER_MS 1000 -#define ELECT_TIMER_MS_MIN 1300 -#define ELECT_TIMER_MS_MAX (ELECT_TIMER_MS_MIN * 2) +#define TIMER_MAX_MS 0x7FFFFFFF +#define ENV_TICK_TIMER_MS 1000 +#define PING_TIMER_MS 1000 +#define ELECT_TIMER_MS_MIN 1300 +#define ELECT_TIMER_MS_MAX (ELECT_TIMER_MS_MIN * 2) #define ELECT_TIMER_MS_RANGE (ELECT_TIMER_MS_MAX - ELECT_TIMER_MS_MIN) -#define HEARTBEAT_TIMER_MS 900 +#define HEARTBEAT_TIMER_MS 900 #define EMPTY_RAFT_ID ((SRaftId){.addr = 0, .vgId = 0}) diff --git a/source/libs/sync/inc/syncIO.h b/source/libs/sync/inc/syncIO.h index 21ae40f92d..f3064e668d 100644 --- a/source/libs/sync/inc/syncIO.h +++ b/source/libs/sync/inc/syncIO.h @@ -50,7 +50,7 @@ typedef struct SSyncIO { void *pSyncNode; int32_t (*FpOnSyncPing)(SSyncNode *pSyncNode, SyncPing *pMsg); int32_t (*FpOnSyncPingReply)(SSyncNode *pSyncNode, SyncPingReply *pMsg); - int32_t (*FpOnSyncClientRequest)(SSyncNode *pSyncNode, SyncClientRequest *pMsg, SyncIndex* pRetIndex); + int32_t (*FpOnSyncClientRequest)(SSyncNode *pSyncNode, SyncClientRequest *pMsg, SyncIndex *pRetIndex); int32_t (*FpOnSyncRequestVote)(SSyncNode *pSyncNode, SyncRequestVote *pMsg); int32_t (*FpOnSyncRequestVoteReply)(SSyncNode *pSyncNode, SyncRequestVoteReply *pMsg); int32_t (*FpOnSyncAppendEntries)(SSyncNode *pSyncNode, SyncAppendEntries *pMsg); diff --git a/source/libs/sync/inc/syncRaftCfg.h b/source/libs/sync/inc/syncRaftCfg.h index efe8a65b77..521ca6068d 100644 --- a/source/libs/sync/inc/syncRaftCfg.h +++ b/source/libs/sync/inc/syncRaftCfg.h @@ -49,14 +49,14 @@ int32_t raftCfgClose(SRaftCfg *pRaftCfg); int32_t raftCfgPersist(SRaftCfg *pRaftCfg); int32_t raftCfgAddConfigIndex(SRaftCfg *pRaftCfg, SyncIndex configIndex); -cJSON *syncCfg2Json(SSyncCfg *pSyncCfg); -char *syncCfg2Str(SSyncCfg *pSyncCfg); -char *syncCfg2SimpleStr(SSyncCfg *pSyncCfg); +cJSON * syncCfg2Json(SSyncCfg *pSyncCfg); +char * syncCfg2Str(SSyncCfg *pSyncCfg); +char * syncCfg2SimpleStr(SSyncCfg *pSyncCfg); int32_t syncCfgFromJson(const cJSON *pRoot, SSyncCfg *pSyncCfg); int32_t syncCfgFromStr(const char *s, SSyncCfg *pSyncCfg); -cJSON *raftCfg2Json(SRaftCfg *pRaftCfg); -char *raftCfg2Str(SRaftCfg *pRaftCfg); +cJSON * raftCfg2Json(SRaftCfg *pRaftCfg); +char * raftCfg2Str(SRaftCfg *pRaftCfg); int32_t raftCfgFromJson(const cJSON *pRoot, SRaftCfg *pRaftCfg); int32_t raftCfgFromStr(const char *s, SRaftCfg *pRaftCfg); diff --git a/source/libs/sync/src/syncIO.c b/source/libs/sync/src/syncIO.c index 4243a933b0..663745a7d7 100644 --- a/source/libs/sync/src/syncIO.c +++ b/source/libs/sync/src/syncIO.c @@ -30,7 +30,7 @@ static int32_t syncIODestroy(SSyncIO *io); static int32_t syncIOStartInternal(SSyncIO *io); static int32_t syncIOStopInternal(SSyncIO *io); -static void *syncIOConsumerFunc(void *param); +static void * syncIOConsumerFunc(void *param); static void syncIOProcessRequest(void *pParent, SRpcMsg *pMsg, SEpSet *pEpSet); static void syncIOProcessReply(void *pParent, SRpcMsg *pMsg, SEpSet *pEpSet); static int32_t syncIOAuth(void *parent, char *meterId, char *spi, char *encrypt, char *secret, char *ckey); @@ -242,9 +242,9 @@ static int32_t syncIOStopInternal(SSyncIO *io) { } static void *syncIOConsumerFunc(void *param) { - SSyncIO *io = param; + SSyncIO * io = param; STaosQall *qall; - SRpcMsg *pRpcMsg, rpcMsg; + SRpcMsg * pRpcMsg, rpcMsg; qall = taosAllocateQall(); while (1) { diff --git a/source/libs/sync/src/syncIndexMgr.c b/source/libs/sync/src/syncIndexMgr.c index 5b432aeec4..10f0e0e335 100644 --- a/source/libs/sync/src/syncIndexMgr.c +++ b/source/libs/sync/src/syncIndexMgr.c @@ -126,7 +126,7 @@ cJSON *syncIndexMgr2Json(SSyncIndexMgr *pSyncIndexMgr) { char *syncIndexMgr2Str(SSyncIndexMgr *pSyncIndexMgr) { cJSON *pJson = syncIndexMgr2Json(pSyncIndexMgr); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } diff --git a/source/libs/sync/src/syncMain.c b/source/libs/sync/src/syncMain.c index 2c00dccb60..3996193037 100644 --- a/source/libs/sync/src/syncMain.c +++ b/source/libs/sync/src/syncMain.c @@ -2654,25 +2654,33 @@ int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, syncEntry2OriginalRpc(pEntry, &rpcMsg); // user commit - bool internalExecute = (ths->pFsm->FpCommitCb != NULL) && syncUtilUserCommit(pEntry->originalRpcType); - if (ths->replicaNum == 1) { - internalExecute = syncNodeIsOptimizedOneReplica(ths, &rpcMsg) && !(ths->restoreFinish); - } + if ((ths->pFsm->FpCommitCb != NULL) && syncUtilUserCommit(pEntry->originalRpcType)) { + bool internalExecute = true; + if ((ths->replicaNum == 1) && ths->restoreFinish && (ths->vgId != 1)) { + internalExecute = false; + } - // execute fsm in apply thread, or execute outside syncPropose - if (internalExecute) { - SFsmCbMeta cbMeta = {0}; - cbMeta.index = pEntry->index; - cbMeta.lastConfigIndex = syncNodeGetSnapshotConfigIndex(ths, cbMeta.index); - cbMeta.isWeak = pEntry->isWeak; - cbMeta.code = 0; - cbMeta.state = ths->state; - cbMeta.seqNum = pEntry->seqNum; - cbMeta.term = pEntry->term; - cbMeta.currentTerm = ths->pRaftStore->currentTerm; - cbMeta.flag = flag; + do { + char logBuf[128]; + snprintf(logBuf, sizeof(logBuf), "index:%ld, internalExecute:%d", i, internalExecute); + syncNodeEventLog(ths, logBuf); + } while (0); - ths->pFsm->FpCommitCb(ths->pFsm, &rpcMsg, cbMeta); + // execute fsm in apply thread, or execute outside syncPropose + if (internalExecute) { + SFsmCbMeta cbMeta = {0}; + cbMeta.index = pEntry->index; + cbMeta.lastConfigIndex = syncNodeGetSnapshotConfigIndex(ths, cbMeta.index); + cbMeta.isWeak = pEntry->isWeak; + cbMeta.code = 0; + cbMeta.state = ths->state; + cbMeta.seqNum = pEntry->seqNum; + cbMeta.term = pEntry->term; + cbMeta.currentTerm = ths->pRaftStore->currentTerm; + cbMeta.flag = flag; + + ths->pFsm->FpCommitCb(ths->pFsm, &rpcMsg, cbMeta); + } } // config change diff --git a/source/libs/sync/src/syncRaftCfg.c b/source/libs/sync/src/syncRaftCfg.c index f40eff57d9..9d16bed6c1 100644 --- a/source/libs/sync/src/syncRaftCfg.c +++ b/source/libs/sync/src/syncRaftCfg.c @@ -101,7 +101,7 @@ cJSON *syncCfg2Json(SSyncCfg *pSyncCfg) { char *syncCfg2Str(SSyncCfg *pSyncCfg) { cJSON *pJson = syncCfg2Json(pSyncCfg); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } @@ -109,7 +109,7 @@ char *syncCfg2Str(SSyncCfg *pSyncCfg) { char *syncCfg2SimpleStr(SSyncCfg *pSyncCfg) { if (pSyncCfg != NULL) { int32_t len = 512; - char *s = taosMemoryMalloc(len); + char * s = taosMemoryMalloc(len); memset(s, 0, len); snprintf(s, len, "{replica-num:%d, my-index:%d, ", pSyncCfg->replicaNum, pSyncCfg->myIndex); @@ -205,7 +205,7 @@ cJSON *raftCfg2Json(SRaftCfg *pRaftCfg) { char *raftCfg2Str(SRaftCfg *pRaftCfg) { cJSON *pJson = raftCfg2Json(pRaftCfg); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } @@ -271,7 +271,7 @@ int32_t raftCfgFromJson(const cJSON *pRoot, SRaftCfg *pRaftCfg) { (pRaftCfg->configIndexArr)[i] = atoll(pIndex->valuestring); } - cJSON *pJsonSyncCfg = cJSON_GetObjectItem(pJson, "SSyncCfg"); + cJSON * pJsonSyncCfg = cJSON_GetObjectItem(pJson, "SSyncCfg"); int32_t code = syncCfgFromJson(pJsonSyncCfg, &(pRaftCfg->cfg)); ASSERT(code == 0); diff --git a/source/libs/sync/src/syncRaftStore.c b/source/libs/sync/src/syncRaftStore.c index a1ab95c00f..9f5cba6c66 100644 --- a/source/libs/sync/src/syncRaftStore.c +++ b/source/libs/sync/src/syncRaftStore.c @@ -216,7 +216,7 @@ cJSON *raftStore2Json(SRaftStore *pRaftStore) { char *raftStore2Str(SRaftStore *pRaftStore) { cJSON *pJson = raftStore2Json(pRaftStore); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } diff --git a/source/libs/sync/src/syncSnapshot.c b/source/libs/sync/src/syncSnapshot.c index 4fc0128400..0feaf532bb 100644 --- a/source/libs/sync/src/syncSnapshot.c +++ b/source/libs/sync/src/syncSnapshot.c @@ -314,14 +314,14 @@ cJSON *snapshotSender2Json(SSyncSnapshotSender *pSender) { char *snapshotSender2Str(SSyncSnapshotSender *pSender) { cJSON *pJson = snapshotSender2Json(pSender); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } char *snapshotSender2SimpleStr(SSyncSnapshotSender *pSender, char *event) { int32_t len = 256; - char *s = taosMemoryMalloc(len); + char * s = taosMemoryMalloc(len); SRaftId destId = pSender->pSyncNode->replicasId[pSender->replicaIndex]; char host[128]; @@ -461,7 +461,7 @@ cJSON *snapshotReceiver2Json(SSyncSnapshotReceiver *pReceiver) { cJSON_AddStringToObject(pFromId, "addr", u64buf); { uint64_t u64 = pReceiver->fromId.addr; - cJSON *pTmp = pFromId; + cJSON * pTmp = pFromId; char host[128] = {0}; uint16_t port; syncUtilU642Addr(u64, host, sizeof(host), &port); @@ -494,14 +494,14 @@ cJSON *snapshotReceiver2Json(SSyncSnapshotReceiver *pReceiver) { char *snapshotReceiver2Str(SSyncSnapshotReceiver *pReceiver) { cJSON *pJson = snapshotReceiver2Json(pReceiver); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } char *snapshotReceiver2SimpleStr(SSyncSnapshotReceiver *pReceiver, char *event) { int32_t len = 256; - char *s = taosMemoryMalloc(len); + char * s = taosMemoryMalloc(len); SRaftId fromId = pReceiver->fromId; char host[128]; diff --git a/source/libs/sync/src/syncVoteMgr.c b/source/libs/sync/src/syncVoteMgr.c index d6c2cbd34e..2c43312064 100644 --- a/source/libs/sync/src/syncVoteMgr.c +++ b/source/libs/sync/src/syncVoteMgr.c @@ -127,7 +127,7 @@ cJSON *voteGranted2Json(SVotesGranted *pVotesGranted) { char *voteGranted2Str(SVotesGranted *pVotesGranted) { cJSON *pJson = voteGranted2Json(pVotesGranted); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } @@ -256,7 +256,7 @@ cJSON *votesRespond2Json(SVotesRespond *pVotesRespond) { char *votesRespond2Str(SVotesRespond *pVotesRespond) { cJSON *pJson = votesRespond2Json(pVotesRespond); - char *serialized = cJSON_Print(pJson); + char * serialized = cJSON_Print(pJson); cJSON_Delete(pJson); return serialized; } diff --git a/source/libs/sync/test/syncRaftLogTest2.cpp b/source/libs/sync/test/syncRaftLogTest2.cpp index e3ceae0615..9e0c2ecc29 100644 --- a/source/libs/sync/test/syncRaftLogTest2.cpp +++ b/source/libs/sync/test/syncRaftLogTest2.cpp @@ -113,7 +113,7 @@ void test2() { pLogStore = logStoreCreate(pSyncNode); assert(pLogStore); pSyncNode->pLogStore = pLogStore; - //pLogStore->syncLogSetBeginIndex(pLogStore, 5); + // pLogStore->syncLogSetBeginIndex(pLogStore, 5); pLogStore->syncLogRestoreFromSnapshot(pLogStore, 4); logStoreLog2((char*)"\n\n\ntest2 ----- ", pLogStore); @@ -229,7 +229,7 @@ void test4() { assert(pLogStore); pSyncNode->pLogStore = pLogStore; logStoreLog2((char*)"\n\n\ntest4 ----- ", pLogStore); - //pLogStore->syncLogSetBeginIndex(pLogStore, 5); + // pLogStore->syncLogSetBeginIndex(pLogStore, 5); pLogStore->syncLogRestoreFromSnapshot(pLogStore, 4); for (int i = 5; i <= 9; ++i) { @@ -291,7 +291,7 @@ void test5() { assert(pLogStore); pSyncNode->pLogStore = pLogStore; logStoreLog2((char*)"\n\n\ntest5 ----- ", pLogStore); - //pLogStore->syncLogSetBeginIndex(pLogStore, 5); + // pLogStore->syncLogSetBeginIndex(pLogStore, 5); pLogStore->syncLogRestoreFromSnapshot(pLogStore, 4); for (int i = 5; i <= 9; ++i) { @@ -412,26 +412,23 @@ void test6() { do { SyncIndex firstVer = walGetFirstVer(pWal); SyncIndex lastVer = walGetLastVer(pWal); - bool isEmpty = walIsEmpty(pWal); + bool isEmpty = walIsEmpty(pWal); printf("before -------- firstVer:%ld lastVer:%ld isEmpty:%d \n", firstVer, lastVer, isEmpty); } while (0); logStoreDestory(pLogStore); cleanup(); - - // restart init(); pLogStore = logStoreCreate(pSyncNode); assert(pLogStore); pSyncNode->pLogStore = pLogStore; - do { SyncIndex firstVer = walGetFirstVer(pWal); SyncIndex lastVer = walGetLastVer(pWal); - bool isEmpty = walIsEmpty(pWal); + bool isEmpty = walIsEmpty(pWal); printf("after -------- firstVer:%ld lastVer:%ld isEmpty:%d \n", firstVer, lastVer, isEmpty); } while (0); @@ -461,13 +458,13 @@ int main(int argc, char** argv) { } sTrace("gAssert : %d", gAssert); -/* - test1(); - test2(); - test3(); - test4(); - test5(); -*/ + /* + test1(); + test2(); + test3(); + test4(); + test5(); + */ test6(); return 0; diff --git a/source/libs/sync/test/syncRaftLogTest3.cpp b/source/libs/sync/test/syncRaftLogTest3.cpp index 302e29a091..ea1788c545 100644 --- a/source/libs/sync/test/syncRaftLogTest3.cpp +++ b/source/libs/sync/test/syncRaftLogTest3.cpp @@ -312,7 +312,7 @@ void test5() { pSyncNode->pLogStore = pLogStore; logStoreLog2((char*)"\n\n\ntest5 ----- ", pLogStore); - //pSyncNode->pLogStore->syncLogSetBeginIndex(pSyncNode->pLogStore, 6); + // pSyncNode->pLogStore->syncLogSetBeginIndex(pSyncNode->pLogStore, 6); pLogStore->syncLogRestoreFromSnapshot(pSyncNode->pLogStore, 5); for (int i = 6; i <= 10; ++i) { int32_t dataLen = 10; From 393dbc5d2089abc6a8c602fe9eb8c09b153a3292 Mon Sep 17 00:00:00 2001 From: Yu Chen <74105241+yu285@users.noreply.github.com> Date: Sun, 26 Jun 2022 12:32:51 +0800 Subject: [PATCH 03/10] docs: update mnode description --- docs/en/21-tdinternal/01-arch.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/en/21-tdinternal/01-arch.md b/docs/en/21-tdinternal/01-arch.md index d7d472eb98..44651c0496 100644 --- a/docs/en/21-tdinternal/01-arch.md +++ b/docs/en/21-tdinternal/01-arch.md @@ -22,7 +22,7 @@ A complete TDengine system runs on one or more physical nodes. Logically, it inc **Virtual node (vnode)**: To better support data sharding, load balancing and prevent data from overheating or skewing, data nodes are virtualized into multiple virtual nodes (vnode, V2, V3, V4, etc. in the figure). Each vnode is a relatively independent work unit, which is the basic unit of time-series data storage and has independent running threads, memory space and persistent storage path. A vnode contains a certain number of tables (data collection points). When a new table is created, the system checks whether a new vnode needs to be created. The number of vnodes that can be created on a data node depends on the capacity of the hardware of the physical node where the data node is located. A vnode belongs to only one DB, but a DB can have multiple vnodes. In addition to the stored time-series data, a vnode also stores the schema and tag values of the included tables. A virtual node is uniquely identified in the system by the EP of the data node and the VGroup ID to which it belongs and is created and managed by the management node. -**Management node (mnode)**: A virtual logical unit responsible for monitoring and maintaining the running status of all data nodes and load balancing among nodes (M in the figure). At the same time, the management node is also responsible for the storage and management of metadata (including users, databases, tables, static tags, etc.), so it is also called Meta Node. Multiple (up to 5) mnodes can be configured in a TDengine cluster, and they are automatically constructed into a virtual management node group (M0, M1, M2 in the figure). The leader/follower mechanism is adopted for the mnode group and the data synchronization is carried out in a strongly consistent way. Any data update operation can only be executed on the leader. The creation of mnode cluster is completed automatically by the system without manual intervention. There is at most one mnode on each dnode, which is uniquely identified by the EP of the data node to which it belongs. Each dnode automatically obtains the EP of the dnode where all mnodes in the whole cluster are located, through internal messaging interaction. +**Management node (mnode)**: A virtual logical unit responsible for monitoring and maintaining the running status of all data nodes and load balancing among nodes (M in the figure). At the same time, the management node is also responsible for the storage and management of metadata (including users, databases, tables, static tags, etc.), so it is also called Meta Node. Multiple (up to 3) mnodes can be configured in a TDengine cluster, and they are automatically constructed into a virtual management node group (M0, M1, M2 in the figure). The leader/follower mechanism is adopted for the mnode group and the data synchronization is carried out in a strongly consistent way. Any data update operation can only be executed on the leader. The creation of mnode cluster is completed automatically by the system without manual intervention. There is at most one mnode on each dnode, which is uniquely identified by the EP of the data node to which it belongs. Each dnode automatically obtains the EP of the dnode where all mnodes in the whole cluster are located, through internal messaging interaction. **Virtual node group (VGroup)**: Vnodes on different data nodes can form a virtual node group to ensure the high availability of the system. The virtual node group is managed in a leader/follower mechanism. Write operations can only be performed on the leader vnode, and then replicated to follower vnodes, thus ensuring that one single replica of data is copied on multiple physical nodes. The number of virtual nodes in a vgroup equals the number of data replicas. If the number of replicas of a DB is N, the system must have at least N data nodes. The number of replicas can be specified by the parameter `“replica”` when creating a DB, and the default is 1. Using the multi-replication feature of TDengine, the same high data reliability can be achieved without the need for expensive storage devices such as disk arrays. Virtual node groups are created and managed by the management node, and the management node assigns a system unique ID, aka VGroup ID. If two virtual nodes have the same vnode group ID, it means that they belong to the same group and the data is backed up to each other. The number of virtual nodes in a virtual node group can be dynamically changed, allowing only one, that is, no data replication. VGroup ID is never changed. Even if a virtual node group is deleted, its ID will not be reused. From 56997d20575e45fc763c4a5a092b0d643219458c Mon Sep 17 00:00:00 2001 From: Yu Chen <74105241+yu285@users.noreply.github.com> Date: Sun, 26 Jun 2022 12:17:52 +0800 Subject: [PATCH 04/10] docs: update docs --- docs/zh/21-tdinternal/01-arch.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/zh/21-tdinternal/01-arch.md b/docs/zh/21-tdinternal/01-arch.md index 433cb4808b..507ccea629 100644 --- a/docs/zh/21-tdinternal/01-arch.md +++ b/docs/zh/21-tdinternal/01-arch.md @@ -23,7 +23,7 @@ TDengine 分布式架构的逻辑结构图如下： **虚拟节点（vnode）：** 为更好的支持数据分片、负载均衡，防止数据过热或倾斜，数据节点被虚拟化成多个虚拟节点（vnode，图中 V2，V3，V4 等）。每个 vnode 都是一个相对独立的工作单元，是时序数据存储的基本单元，具有独立的运行线程、内存空间与持久化存储的路径。一个 vnode 包含一定数量的表（数据采集点）。当创建一张新表时，系统会检查是否需要创建新的 vnode。一个数据节点上能创建的 vnode 的数量取决于该数据节点所在物理节点的硬件资源。一个 vnode 只属于一个 DB，但一个 DB 可以有多个 vnode。一个 vnode 除存储的时序数据外，也保存有所包含的表的 schema、标签值等。一个虚拟节点由所属的数据节点的 EP，以及所属的 VGroup ID 在系统内唯一标识，由管理节点创建并管理。 -**管理节点（mnode）：** 一个虚拟的逻辑单元，负责所有数据节点运行状态的监控和维护，以及节点之间的负载均衡（图中 M）。同时，管理节点也负责元数据（包括用户、数据库、表、静态标签等）的存储和管理，因此也称为 Meta Node。TDengine 集群中可配置多个（开源版最多不超过 3 个）mnode，它们自动构建成为一个虚拟管理节点组（图中 M0，M1，M2）。mnode 间采用 master/slave 的机制进行管理，而且采取强一致方式进行数据同步，任何数据更新操作只能在 Master 上进行。mnode 集群的创建由系统自动完成，无需人工干预。每个 dnode 上至多有一个 mnode，由所属的数据节点的 EP 来唯一标识。每个 dnode 通过内部消息交互自动获取整个集群中所有 mnode 所在的 dnode 的 EP。 +**管理节点（mnode）：** 一个虚拟的逻辑单元，负责所有数据节点运行状态的监控和维护，以及节点之间的负载均衡（图中 M）。同时，管理节点也负责元数据（包括用户、数据库、表、静态标签等）的存储和管理，因此也称为 Meta Node。TDengine 集群中可配置多个（最多不超过 3 个）mnode，它们自动构建成为一个虚拟管理节点组（图中 M0，M1，M2）。mnode 间采用 master/slave 的机制进行管理，而且采取强一致方式进行数据同步，任何数据更新操作只能在 Master 上进行。mnode 集群的创建由系统自动完成，无需人工干预。每个 dnode 上至多有一个 mnode，由所属的数据节点的 EP 来唯一标识。每个 dnode 通过内部消息交互自动获取整个集群中所有 mnode 所在的 dnode 的 EP。 **虚拟节点组（VGroup）：** 不同数据节点上的 vnode 可以组成一个虚拟节点组（vgroup）来保证系统的高可靠。虚拟节点组内采取 master/slave 的方式进行管理。写操作只能在 master vnode 上进行，系统采用异步复制的方式将数据同步到 slave vnode，这样确保了一份数据在多个物理节点上有拷贝。一个 vgroup 里虚拟节点个数就是数据的副本数。如果一个 DB 的副本数为 N，系统必须有至少 N 数据节点。副本数在创建 DB 时通过参数 replica 可以指定，缺省为 1。使用 TDengine 的多副本特性，可以不再需要昂贵的磁盘阵列等存储设备，就可以获得同样的数据高可靠性。虚拟节点组由管理节点创建、管理，并且由管理节点分配一个系统唯一的 ID，VGroup ID。如果两个虚拟节点的 VGroup ID 相同，说明他们属于同一个组，数据互为备份。虚拟节点组里虚拟节点的个数是可以动态改变的，容许只有一个，也就是没有数据复制。VGroup ID 是永远不变的，即使一个虚拟节点组被删除，它的 ID 也不会被收回重复利用。 From ff72e66dba623ace3daf0c53469e80578b406677 Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 27 Jun 2022 10:07:48 +0800 Subject: [PATCH 05/10] fix(query): support scalar function before applying interval aggregate. --- source/libs/executor/inc/executorimpl.h | 4 +-- source/libs/executor/src/executorimpl.c | 4 +-- source/libs/executor/src/scanoperator.c | 6 ++-- source/libs/executor/src/timewindowoperator.c | 31 ++++++++++++++++--- 4 files changed, 33 insertions(+), 12 deletions(-) diff --git a/source/libs/executor/inc/executorimpl.h b/source/libs/executor/inc/executorimpl.h index d8d231e952..2f613f528c 100644 --- a/source/libs/executor/inc/executorimpl.h +++ b/source/libs/executor/inc/executorimpl.h @@ -400,7 +400,7 @@ typedef struct SIntervalAggOperatorInfo { // SOptrBasicInfo should be first, SAggSupporter should be second for stream encode SOptrBasicInfo binfo; // basic info SAggSupporter aggSup; // aggregate supporter - + SExprSupp scalarSupp; // supporter for perform scalar function SGroupResInfo groupResInfo; // multiple results build supporter SInterval interval; // interval info int32_t primaryTsIndex; // primary time stamp slot id from result of downstream operator. @@ -730,7 +730,7 @@ SOperatorInfo* createSortedMergeOperatorInfo(SOperatorInfo** downstream, int32_t SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, SSDataBlock* pResBlock, SInterval* pInterval, int32_t primaryTsSlotId, - STimeWindowAggSupp *pTwAggSupp, SExecTaskInfo* pTaskInfo, bool isStream); + STimeWindowAggSupp* pTwAggSupp, SIntervalPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, bool isStream); SOperatorInfo* createMergeIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, SSDataBlock* pResBlock, SInterval* pInterval, int32_t primaryTsSlotId, diff --git a/source/libs/executor/src/executorimpl.c b/source/libs/executor/src/executorimpl.c index f352049810..6fe37feb4b 100644 --- a/source/libs/executor/src/executorimpl.c +++ b/source/libs/executor/src/executorimpl.c @@ -4216,8 +4216,8 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo int32_t tsSlotId = ((SColumnNode*)pIntervalPhyNode->window.pTspk)->slotId; bool isStream = (QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL == type); - pOptr = - createIntervalOperatorInfo(ops[0], pExprInfo, num, pResBlock, &interval, tsSlotId, &as, pTaskInfo, isStream); + pOptr = createIntervalOperatorInfo(ops[0], pExprInfo, num, pResBlock, &interval, tsSlotId, &as, pIntervalPhyNode, + pTaskInfo, isStream); } else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL == type) { SMergeAlignedIntervalPhysiNode* pIntervalPhyNode = (SMergeAlignedIntervalPhysiNode*)pPhyNode; diff --git a/source/libs/executor/src/scanoperator.c b/source/libs/executor/src/scanoperator.c index 7affac76d2..5bfb73271e 100644 --- a/source/libs/executor/src/scanoperator.c +++ b/source/libs/executor/src/scanoperator.c @@ -448,7 +448,7 @@ static SSDataBlock* doTableScanGroup(SOperatorInfo* pOperator) { qDebug("%s start to repeat ascending order scan data blocks due to query func required", GET_TASKID(pTaskInfo)); for (int32_t i = 0; i < pTableScanInfo->cond.numOfTWindows; ++i) { STimeWindow* pWin = &pTableScanInfo->cond.twindows[i]; - qDebug("%s\t qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey); + qDebug("%s qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey); } // do prepare for the next round table scan operation tsdbResetReadHandle(pTableScanInfo->dataReader, &pTableScanInfo->cond, 0); @@ -467,7 +467,7 @@ static SSDataBlock* doTableScanGroup(SOperatorInfo* pOperator) { qDebug("%s start to descending order scan data blocks due to query func required", GET_TASKID(pTaskInfo)); for (int32_t i = 0; i < pTableScanInfo->cond.numOfTWindows; ++i) { STimeWindow* pWin = &pTableScanInfo->cond.twindows[i]; - qDebug("%s\t qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey); + qDebug("%s qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey); } while (pTableScanInfo->scanTimes < total) { @@ -492,7 +492,7 @@ static SSDataBlock* doTableScanGroup(SOperatorInfo* pOperator) { GET_TASKID(pTaskInfo)); for (int32_t i = 0; i < pTableScanInfo->cond.numOfTWindows; ++i) { STimeWindow* pWin = &pTableScanInfo->cond.twindows[i]; - qDebug("%s\t qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey); + qDebug("%s qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey); } tsdbResetReadHandle(pTableScanInfo->dataReader, &pTableScanInfo->cond, 0); pTableScanInfo->curTWinIdx = 0; diff --git a/source/libs/executor/src/timewindowoperator.c b/source/libs/executor/src/timewindowoperator.c index 48b0b1c071..2f4ae6c7c7 100644 --- a/source/libs/executor/src/timewindowoperator.c +++ b/source/libs/executor/src/timewindowoperator.c @@ -956,6 +956,12 @@ static int32_t doOpenIntervalAgg(SOperatorInfo* pOperator) { getTableScanInfo(pOperator, &pInfo->order, &scanFlag); + if (pInfo->scalarSupp.pExprInfo != NULL) { + SExprSupp* pExprSup =& pInfo->scalarSupp; + projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx, + pExprSup->numOfExprs, NULL); + } + // the pDataBlock are always the same one, no need to call this again setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->order, scanFlag, true); hashIntervalAgg(pOperator, &pInfo->binfo.resultRowInfo, pBlock, scanFlag, NULL); @@ -1324,6 +1330,12 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) { break; } + if (pInfo->scalarSupp.pExprInfo != NULL) { + SExprSupp* pExprSup = &pInfo->scalarSupp; + projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx, + pExprSup->numOfExprs, NULL); + } + if (pBlock->info.type == STREAM_REPROCESS) { doClearWindows(&pInfo->aggSup, &pOperator->exprSupp, &pInfo->interval, 0, pOperator->exprSupp.numOfExprs, pBlock, NULL); @@ -1446,18 +1458,27 @@ void increaseTs(SqlFunctionCtx* pCtx) { SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols, SSDataBlock* pResBlock, SInterval* pInterval, int32_t primaryTsSlotId, - STimeWindowAggSupp* pTwAggSupp, SExecTaskInfo* pTaskInfo, bool isStream) { + STimeWindowAggSupp* pTwAggSupp, SIntervalPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, bool isStream) { SIntervalAggOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SIntervalAggOperatorInfo)); SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo)); if (pInfo == NULL || pOperator == NULL) { goto _error; } - pInfo->win = pTaskInfo->window; - pInfo->order = TSDB_ORDER_ASC; - pInfo->interval = *pInterval; + pInfo->win = pTaskInfo->window; + pInfo->order = TSDB_ORDER_ASC; + pInfo->interval = *pInterval; pInfo->execModel = pTaskInfo->execModel; - pInfo->twAggSup = *pTwAggSupp; + pInfo->twAggSup = *pTwAggSupp; + + if (pPhyNode->window.pExprs != NULL) { + int32_t numOfScalar = 0; + SExprInfo* pScalarExprInfo = createExprInfo(pPhyNode->window.pExprs, NULL, &numOfScalar); + int32_t code = initExprSupp(&pInfo->scalarSupp, pScalarExprInfo, numOfScalar); + if (code != TSDB_CODE_SUCCESS) { + goto _error; + } + } pInfo->primaryTsIndex = primaryTsSlotId; From a22814a5a7434df7f271a2a35e03afd083078997 Mon Sep 17 00:00:00 2001 From: Minghao Li Date: Mon, 27 Jun 2022 10:36:28 +0800 Subject: [PATCH 06/10] refactor(sync): optimized one replica --- source/libs/sync/inc/syncRaftLog.h | 7 +++++-- source/libs/sync/src/syncRaftLog.c | 13 +++++++++++++ 2 files changed, 18 insertions(+), 2 deletions(-) diff --git a/source/libs/sync/inc/syncRaftLog.h b/source/libs/sync/inc/syncRaftLog.h index f124867097..f3ed9e302b 100644 --- a/source/libs/sync/inc/syncRaftLog.h +++ b/source/libs/sync/inc/syncRaftLog.h @@ -29,9 +29,12 @@ extern "C" { #include "wal.h" typedef struct SSyncLogStoreData { - SSyncNode* pSyncNode; - SWal* pWal; + SSyncNode* pSyncNode; + SWal* pWal; + + TdThreadMutex mutex; SWalReadHandle* pWalHandle; + // SyncIndex beginIndex; // valid begin index, default 0, may be set beginIndex > 0 } SSyncLogStoreData; diff --git a/source/libs/sync/src/syncRaftLog.c b/source/libs/sync/src/syncRaftLog.c index cc606f4df8..d84c576714 100644 --- a/source/libs/sync/src/syncRaftLog.c +++ b/source/libs/sync/src/syncRaftLog.c @@ -257,6 +257,8 @@ static int32_t raftLogGetEntry(struct SSyncLogStore* pLogStore, SyncIndex index, return -1; } + taosThreadMutexLock(&(pData->mutex)); + code = walReadWithHandle(pWalHandle, index); if (code != 0) { int32_t err = terrno; @@ -281,6 +283,7 @@ static int32_t raftLogGetEntry(struct SSyncLogStore* pLogStore, SyncIndex index, terrno = saveErr; */ + taosThreadMutexUnlock(&(pData->mutex)); return code; } @@ -301,6 +304,7 @@ static int32_t raftLogGetEntry(struct SSyncLogStore* pLogStore, SyncIndex index, terrno = saveErr; */ + taosThreadMutexUnlock(&(pData->mutex)); return code; } @@ -364,6 +368,7 @@ SSyncLogStore* logStoreCreate(SSyncNode* pSyncNode) { pData->pWal = pSyncNode->pWal; ASSERT(pData->pWal != NULL); + taosThreadMutexInit(&(pData->mutex), NULL); pData->pWalHandle = walOpenReadHandle(pData->pWal); ASSERT(pData->pWalHandle != NULL); @@ -408,9 +413,14 @@ SSyncLogStore* logStoreCreate(SSyncNode* pSyncNode) { void logStoreDestory(SSyncLogStore* pLogStore) { if (pLogStore != NULL) { SSyncLogStoreData* pData = pLogStore->data; + + taosThreadMutexLock(&(pData->mutex)); if (pData->pWalHandle != NULL) { walCloseReadHandle(pData->pWalHandle); + pData->pWalHandle = NULL; } + taosThreadMutexUnlock(&(pData->mutex)); + taosThreadMutexDestroy(&(pData->mutex)); taosMemoryFree(pLogStore->data); taosMemoryFree(pLogStore); @@ -460,6 +470,8 @@ SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) { SWal* pWal = pData->pWal; if (index >= SYNC_INDEX_BEGIN && index <= logStoreLastIndex(pLogStore)) { + taosThreadMutexLock(&(pData->mutex)); + // SWalReadHandle* pWalHandle = walOpenReadHandle(pWal); SWalReadHandle* pWalHandle = pData->pWalHandle; ASSERT(pWalHandle != NULL); @@ -503,6 +515,7 @@ SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) { terrno = saveErr; */ + taosThreadMutexUnlock(&(pData->mutex)); return pEntry; } else { From 2d4eff202b648b5a3d35f9674ad0d7e5e442f1fa Mon Sep 17 00:00:00 2001 From: plum-lihui Date: Mon, 27 Jun 2022 10:58:38 +0800 Subject: [PATCH 07/10] test: add test case for tmq --- tests/system-test/7-tmq/tmqCheckData1.py | 180 +++++++++++++++++++++++ tests/system-test/fulltest.sh | 1 + 2 files changed, 181 insertions(+) create mode 100644 tests/system-test/7-tmq/tmqCheckData1.py diff --git a/tests/system-test/7-tmq/tmqCheckData1.py b/tests/system-test/7-tmq/tmqCheckData1.py new file mode 100644 index 0000000000..6cf849d1b9 --- /dev/null +++ b/tests/system-test/7-tmq/tmqCheckData1.py @@ -0,0 +1,180 @@ + +import taos +import sys +import time +import socket +import os +import threading + +from util.log import * +from util.sql import * +from util.cases import * +from util.dnodes import * +from util.common import * +sys.path.append("./7-tmq") +from tmqCommon import * + +class TDTestCase: + def init(self, conn, logSql): + tdLog.debug(f"start to excute {__file__}") + tdSql.init(conn.cursor()) + #tdSql.init(conn.cursor(), logSql) # output sql.txt file + + def checkFileContent(self, consumerId, queryString): + buildPath = tdCom.getBuildPath() + cfgPath = tdCom.getClientCfgPath() + dstFile = '%s/../log/dstrows_%d.txt'%(cfgPath, consumerId) + cmdStr = '%s/build/bin/taos -c %s -s "%s >> %s"'%(buildPath, cfgPath, queryString, dstFile) + tdLog.info(cmdStr) + os.system(cmdStr) + + consumeRowsFile = '%s/../log/consumerid_%d.txt'%(cfgPath, consumerId) + tdLog.info("rows file: %s, %s"%(consumeRowsFile, dstFile)) + + consumeFile = open(consumeRowsFile, mode='r') + queryFile = open(dstFile, mode='r') + + # skip first line for it is schema + queryFile.readline() + + while True: + dst = queryFile.readline() + src = consumeFile.readline() + + if dst: + if dst != src: + tdLog.exit("consumerId %d consume rows is not match the rows by direct query"%consumerId) + else: + break + return + + def tmqCase1(self): + tdLog.printNoPrefix("======== test case 1: ") + paraDict = {'dbName': 'db1', + 'dropFlag': 1, + 'event': '', + 'vgroups': 4, + 'stbName': 'stb', + 'colPrefix': 'c', + 'tagPrefix': 't', + 'colSchema': [{'type': 'INT', 'count':1}, {'type': 'binary', 'len':20, 'count':1}], + 'tagSchema': [{'type': 'INT', 'count':1}, {'type': 'binary', 'len':20, 'count':1}], + 'ctbPrefix': 'ctb', + 'ctbNum': 1, + 'rowsPerTbl': 10000, + 'batchNum': 10, + 'startTs': 1640966400000, # 2022-01-01 00:00:00.000 + 'pollDelay': 10, + 'showMsg': 1, + 'showRow': 1} + + topicNameList = ['topic1', 'topic2', 'topic3'] + expectRowsList = [] + tmqCom.initConsumerTable() + tdCom.create_database(tdSql, paraDict["dbName"],paraDict["dropFlag"], vgroups=4,replica=1) + tdLog.info("create stb") + tdCom.create_stable(tdSql, dbname=paraDict["dbName"],stbname=paraDict["stbName"], column_elm_list=paraDict['colSchema'], tag_elm_list=paraDict['tagSchema']) + tdLog.info("create ctb") + tdCom.create_ctable(tdSql, dbname=paraDict["dbName"],stbname=paraDict["stbName"],tag_elm_list=paraDict['tagSchema'],count=paraDict["ctbNum"], default_ctbname_prefix=paraDict['ctbPrefix']) + tdLog.info("insert data") + tmqCom.insert_data(tdSql,paraDict["dbName"],paraDict["ctbPrefix"],paraDict["ctbNum"],paraDict["rowsPerTbl"],paraDict["batchNum"],paraDict["startTs"]) + + tdLog.info("create topics from stb with filter") + queryString = "select ts,c1,c2 from %s.%s" %(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as stable %s.%s" %(topicNameList[0], paraDict["dbName"],paraDict["stbName"]) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + tdSql.query(queryString) + expectRowsList.append(tdSql.getRows()) + + # init consume info, and start tmq_sim, then check consume result + tdLog.info("insert consume info to consume processor") + consumerId = 0 + expectrowcnt = paraDict["rowsPerTbl"] * paraDict["ctbNum"] + topicList = topicNameList[0] + ifcheckdata = 1 + ifManualCommit = 1 + keyList = 'group.id:cgrp1, enable.auto.commit:false, auto.commit.interval.ms:6000, auto.offset.reset:earliest' + tmqCom.insertConsumerInfo(consumerId, expectrowcnt,topicList,keyList,ifcheckdata,ifManualCommit) + + tdLog.info("start consume processor") + tmqCom.startTmqSimProcess(paraDict['pollDelay'],paraDict["dbName"],paraDict['showMsg'], paraDict['showRow']) + + tdLog.info("wait the consume result") + expectRows = 1 + resultList = tmqCom.selectConsumeResult(expectRows) + + if expectRowsList[0] != resultList[0]: + tdLog.info("expect consume rows: %d, act consume rows: %d"%(expectRowsList[0], resultList[0])) + tdLog.exit("0 tmq consume rows error!") + + self.checkFileContent(consumerId, queryString) + + # reinit consume info, and start tmq_sim, then check consume result + tmqCom.initConsumerTable() + queryString = "select ts, c1, c2 from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as database %s" %(topicNameList[1], paraDict['dbName']) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + tdSql.query(queryString) + expectRowsList.append(tdSql.getRows()) + + consumerId = 1 + topicList = topicNameList[1] + tmqCom.insertConsumerInfo(consumerId, expectrowcnt,topicList,keyList,ifcheckdata,ifManualCommit) + + tdLog.info("start consume processor") + tmqCom.startTmqSimProcess(paraDict['pollDelay'],paraDict["dbName"],paraDict['showMsg'], paraDict['showRow']) + + tdLog.info("wait the consume result") + expectRows = 1 + resultList = tmqCom.selectConsumeResult(expectRows) + if expectRowsList[1] != resultList[0]: + tdLog.info("expect consume rows: %d, act consume rows: %d"%(expectRowsList[1], resultList[0])) + tdLog.exit("1 tmq consume rows error!") + + self.checkFileContent(consumerId, queryString) + + # reinit consume info, and start tmq_sim, then check consume result + tmqCom.initConsumerTable() + queryString = "select * from %s.%s"%(paraDict['dbName'], paraDict['stbName']) + sqlString = "create topic %s as %s" %(topicNameList[2], queryString) + tdLog.info("create topic sql: %s"%sqlString) + tdSql.execute(sqlString) + tdSql.query(queryString) + expectRowsList.append(tdSql.getRows()) + + consumerId = 2 + topicList = topicNameList[2] + tmqCom.insertConsumerInfo(consumerId, expectrowcnt,topicList,keyList,ifcheckdata,ifManualCommit) + + tdLog.info("start consume processor") + tmqCom.startTmqSimProcess(paraDict['pollDelay'],paraDict["dbName"],paraDict['showMsg'], paraDict['showRow']) + + tdLog.info("wait the consume result") + expectRows = 1 + resultList = tmqCom.selectConsumeResult(expectRows) + if expectRowsList[2] != resultList[0]: + tdLog.info("expect consume rows: %d, act consume rows: %d"%(expectRowsList[2], resultList[0])) + tdLog.exit("2 tmq consume rows error!") + + self.checkFileContent(consumerId, queryString) + + time.sleep(10) + for i in range(len(topicNameList)): + tdSql.query("drop topic %s"%topicNameList[i]) + + tdLog.printNoPrefix("======== test case 1 end ...... ") + + def run(self): + tdSql.prepare() + self.tmqCase1() + + def stop(self): + tdSql.close() + tdLog.success(f"{__file__} successfully executed") + +event = threading.Event() + +tdCases.addLinux(__file__, TDTestCase()) +tdCases.addWindows(__file__, TDTestCase()) diff --git a/tests/system-test/fulltest.sh b/tests/system-test/fulltest.sh index 571bb166e1..2e2e390d12 100755 --- a/tests/system-test/fulltest.sh +++ b/tests/system-test/fulltest.sh @@ -141,6 +141,7 @@ python3 ./test.py -f 7-tmq/tmqError.py python3 ./test.py -f 7-tmq/schema.py python3 ./test.py -f 7-tmq/stbFilter.py python3 ./test.py -f 7-tmq/tmqCheckData.py +python3 ./test.py -f 7-tmq/tmqCheckData1.py python3 ./test.py -f 7-tmq/tmqUdf.py #python3 ./test.py -f 7-tmq/tmq3mnodeSwitch.py -N 5 python3 ./test.py -f 7-tmq/tmqConsumerGroup.py From 02e450c5903415263e88b216c134a7259eda8d6d Mon Sep 17 00:00:00 2001 From: Haojun Liao Date: Mon, 27 Jun 2022 11:34:06 +0800 Subject: [PATCH 08/10] fix(query): fix invalid object type. --- source/libs/executor/src/timewindowoperator.c | 13 ++++++------- 1 file changed, 6 insertions(+), 7 deletions(-) diff --git a/source/libs/executor/src/timewindowoperator.c b/source/libs/executor/src/timewindowoperator.c index 5af12cafac..963e714972 100644 --- a/source/libs/executor/src/timewindowoperator.c +++ b/source/libs/executor/src/timewindowoperator.c @@ -1377,12 +1377,6 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) { break; } - if (pInfo->scalarSupp.pExprInfo != NULL) { - SExprSupp* pExprSup = &pInfo->scalarSupp; - projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx, - pExprSup->numOfExprs, NULL); - } - if (pBlock->info.type == STREAM_CLEAR) { doClearWindows(&pInfo->aggSup, &pOperator->exprSupp, &pInfo->interval, 0, pOperator->exprSupp.numOfExprs, pBlock, NULL); @@ -1393,6 +1387,11 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) { continue; } + if (pInfo->scalarSupp.pExprInfo != NULL) { + SExprSupp* pExprSup = &pInfo->scalarSupp; + projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx, pExprSup->numOfExprs, NULL); + } + // The timewindow that overlaps the timestamps of the input pBlock need to be recalculated and return to the // caller. Note that all the time window are not close till now. // the pDataBlock are always the same one, no need to call this again @@ -2494,7 +2493,7 @@ static SSDataBlock* doStreamFinalIntervalAgg(SOperatorInfo* pOperator) { if (IS_FINAL_OP(pInfo)) { int32_t childIndex = getChildIndex(pBlock); SOperatorInfo* pChildOp = taosArrayGetP(pInfo->pChildren, childIndex); - SIntervalAggOperatorInfo* pChildInfo = pChildOp->info; + SStreamFinalIntervalOperatorInfo* pChildInfo = pChildOp->info; SExprSupp* pChildSup = &pChildOp->exprSupp; doClearWindows(&pChildInfo->aggSup, pChildSup, &pChildInfo->interval, pChildInfo->primaryTsIndex, From c22cbf026f89890191f47d350f5bbbf5ebd2f4d8 Mon Sep 17 00:00:00 2001 From: tangfangzhi Date: Mon, 27 Jun 2022 11:42:25 +0800 Subject: [PATCH 09/10] enh: print host ip address in Jenkins log --- Jenkinsfile2 | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/Jenkinsfile2 b/Jenkinsfile2 index f582461fb2..f618912295 100644 --- a/Jenkinsfile2 +++ b/Jenkinsfile2 @@ -338,6 +338,14 @@ pipeline { changeRequest() } steps { + script { + def linux_node_ip = sh ( + script: 'ip addr|grep 192|grep -v virbr|awk "{print \\\$2}"|sed "s/\\/.*//"', + returnStdout: true + ).trim() + echo "${linux_node_ip}" + echo "${WKDIR}/restore.sh -p ${BRANCH_NAME} -n ${BUILD_ID} -c {container name}" + } catchError(buildResult: 'FAILURE', stageResult: 'FAILURE') { timeout(time: 120, unit: 'MINUTES'){ pre_test() From fb50977b44f4daa8f51757e3fa10661f4a3c1906 Mon Sep 17 00:00:00 2001 From: Shuduo Sang Date: Mon, 27 Jun 2022 14:07:42 +0800 Subject: [PATCH 10/10] fix: add taos_set_config() back (#14277) [TD-16828] --- include/client/taos.h | 1 + source/client/src/clientMain.c | 13 +++++++++++++ 2 files changed, 14 insertions(+) diff --git a/include/client/taos.h b/include/client/taos.h index 5e7f12de0a..d31d5c582c 100644 --- a/include/client/taos.h +++ b/include/client/taos.h @@ -128,6 +128,7 @@ typedef struct setConfRet { DLL_EXPORT void taos_cleanup(void); DLL_EXPORT int taos_options(TSDB_OPTION option, const void *arg, ...); +DLL_EXPORT setConfRet taos_set_config(const char *config); DLL_EXPORT int taos_init(void); DLL_EXPORT TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port); DLL_EXPORT TAOS *taos_connect_auth(const char *ip, const char *user, const char *auth, const char *db, uint16_t port); diff --git a/source/client/src/clientMain.c b/source/client/src/clientMain.c index bbd477fa3b..52574dcc9f 100644 --- a/source/client/src/clientMain.c +++ b/source/client/src/clientMain.c @@ -81,6 +81,19 @@ void taos_cleanup(void) { taosCloseLog(); } +static setConfRet taos_set_config_imp(const char *config){ + setConfRet ret = {SET_CONF_RET_SUCC, {0}}; + // TODO: need re-implementation + return ret; +} + +setConfRet taos_set_config(const char *config){ +// TODO pthread_mutex_lock(&setConfMutex); + setConfRet ret = taos_set_config_imp(config); +// pthread_mutex_unlock(&setConfMutex); + return ret; +} + TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port) { tscDebug("try to connect to %s:%u, user:%s db:%s", ip, port, user, db); if (user == NULL) {