#3548

Committed 04 Dec 2024 01:03PM UTC coverage: 59.846% (-0.8%) from 60.691%

Build # #3548

Build Type

push

travis-ci

Committed by

web-flow

Commit Message

Merge pull request #29033 from taosdata/fix/calculate-vnode-memory-used

fix/calculate-vnode-memory-used

Run Details

118484 of 254183 branches covered (46.61%)

Branch coverage included in aggregate %.

199691 of 277471 relevant lines covered (71.97%)

18794141.86 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

75.21

/source/libs/sync/src/syncReplication.c

/*
 * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
 *
 * This program is free software: you can use, redistribute, and/or modify
 * it under the terms of the GNU Affero General Public License, version 3
 * or later ("AGPL"), as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.
 *
 * You should have received a copy of the GNU Affero General Public License
 * along with this program. If not, see <http://www.gnu.org/licenses/>.
 */

#define _DEFAULT_SOURCE
#include "syncReplication.h"
#include "syncIndexMgr.h"
#include "syncPipeline.h"
#include "syncRaftEntry.h"
#include "syncRaftStore.h"
#include "syncUtil.h"

// TLA+ Spec
// AppendEntries(i, j) ==
//    /\ i /= j
//    /\ state[i] = Leader
//    /\ LET prevLogIndex == nextIndex[i][j] - 1
//           prevLogTerm == IF prevLogIndex > 0 THEN
//                              log[i][prevLogIndex].term
//                          ELSE
//                              0
//           \* Send up to 1 entry, constrained by the end of the log.
//           lastEntry == Min({Len(log[i]), nextIndex[i][j]})
//           entries == SubSeq(log[i], nextIndex[i][j], lastEntry)
//       IN Send([mtype          |-> AppendEntriesRequest,
//                mterm          |-> currentTerm[i],
//                mprevLogIndex  |-> prevLogIndex,
//                mprevLogTerm   |-> prevLogTerm,
//                mentries       |-> entries,
//                \* mlog is used as a history variable for the proof.
//                \* It would not exist in a real implementation.
//                mlog           |-> log[i],
//                mcommitIndex   |-> Min({commitIndex[i], lastEntry}),
//                msource        |-> i,
//                mdest          |-> j])
//    /\ UNCHANGED <<serverVars, candidateVars, leaderVars, logVars>>

int32_t syncNodeReplicateReset(SSyncNode* pNode, SRaftId* pDestId) {
  SSyncLogBuffer* pBuf = pNode->pLogBuf;
  (void)taosThreadMutexLock(&pBuf->mutex);
  SSyncLogReplMgr* pMgr = syncNodeGetLogReplMgr(pNode, pDestId);
  syncLogReplReset(pMgr);
  (void)taosThreadMutexUnlock(&pBuf->mutex);

  TAOS_RETURN(TSDB_CODE_SUCCESS);
}

int32_t syncNodeReplicate(SSyncNode* pNode) {
  SSyncLogBuffer* pBuf = pNode->pLogBuf;
  (void)taosThreadMutexLock(&pBuf->mutex);
  int32_t ret = syncNodeReplicateWithoutLock(pNode);
  (void)taosThreadMutexUnlock(&pBuf->mutex);

  TAOS_RETURN(ret);
}

int32_t syncNodeReplicateWithoutLock(SSyncNode* pNode) {
  if ((pNode->state != TAOS_SYNC_STATE_LEADER && pNode->state != TAOS_SYNC_STATE_ASSIGNED_LEADER) ||
      pNode->raftCfg.cfg.totalReplicaNum == 1) {
    TAOS_RETURN(TSDB_CODE_SUCCESS);
  }
  for (int32_t i = 0; i < pNode->totalReplicaNum; i++) {
    if (syncUtilSameId(&pNode->replicasId[i], &pNode->myRaftId)) {
      continue;
    }
    SSyncLogReplMgr* pMgr = pNode->logReplMgrs[i];
    int32_t          ret = 0;
    if ((ret = syncLogReplStart(pMgr, pNode)) != 0) {
      sWarn("vgId:%d, failed to start log replication to dnode:%d since %s", pNode->vgId, DID(&(pNode->replicasId[i])),
            tstrerror(ret));
    }
  }

  TAOS_RETURN(TSDB_CODE_SUCCESS);
}

int32_t syncNodeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) {
  SyncAppendEntries* pMsg = pRpcMsg->pCont;
  pMsg->destId = *destRaftId;
  TAOS_CHECK_RETURN(syncNodeSendMsgById(destRaftId, pSyncNode, pRpcMsg));

  int32_t nRef = 0;
  if (pSyncNode != NULL) {
    nRef = atomic_add_fetch_32(&pSyncNode->sendCount, 1);
    if (nRef <= 0) {
      sError("vgId:%d, send count is %d", pSyncNode->vgId, nRef);
    }
  }

  SSyncLogReplMgr* mgr = syncNodeGetLogReplMgr(pSyncNode, (SRaftId*)destRaftId);
  if (mgr != NULL) {
    nRef = atomic_add_fetch_32(&mgr->sendCount, 1);
    if (nRef <= 0) {
      sError("vgId:%d, send count is %d", pSyncNode->vgId, nRef);
    }
  }

  TAOS_RETURN(TSDB_CODE_SUCCESS);
}

int32_t syncNodeSendHeartbeat(SSyncNode* pSyncNode, const SRaftId* destId, SRpcMsg* pMsg) {
  SRaftId destIdTmp = *destId;
  TAOS_CHECK_RETURN(syncNodeSendMsgById(destId, pSyncNode, pMsg));

  int64_t tsMs = taosGetTimestampMs();
  syncIndexMgrSetSentTime(pSyncNode->pMatchIndex, &destIdTmp, tsMs);

  return TSDB_CODE_SUCCESS;
}

int32_t syncNodeHeartbeatPeers(SSyncNode* pSyncNode) {
  int64_t ts = taosGetTimestampMs();
  for (int32_t i = 0; i < pSyncNode->peersNum; ++i) {
    SRpcMsg rpcMsg = {0};
    if (syncBuildHeartbeat(&rpcMsg, pSyncNode->vgId) != 0) {
      sError("vgId:%d, build sync-heartbeat error", pSyncNode->vgId);
      continue;
    }

    SyncHeartbeat* pSyncMsg = rpcMsg.pCont;
    pSyncMsg->srcId = pSyncNode->myRaftId;
    pSyncMsg->destId = pSyncNode->peersId[i];
    pSyncMsg->term = raftStoreGetTerm(pSyncNode);
    pSyncMsg->commitIndex = pSyncNode->commitIndex;
    pSyncMsg->minMatchIndex = syncMinMatchIndex(pSyncNode);
    pSyncMsg->privateTerm = 0;
    pSyncMsg->timeStamp = ts;

    // send msg
    TRACE_SET_MSGID(&(rpcMsg.info.traceId), tGenIdPI64());
    STraceId* trace = &(rpcMsg.info.traceId);
    sGTrace("vgId:%d, send sync-heartbeat to dnode:%d", pSyncNode->vgId, DID(&(pSyncMsg->destId)));
    syncLogSendHeartbeat(pSyncNode, pSyncMsg, true, 0, 0);
    int32_t ret = syncNodeSendHeartbeat(pSyncNode, &pSyncMsg->destId, &rpcMsg);
    if (ret != 0) {
      sError("vgId:%d, failed to send sync-heartbeat since %s", pSyncNode->vgId, tstrerror(ret));
    }
  }

  TAOS_RETURN(TSDB_CODE_SUCCESS);
}

1	/*
2	* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
3	*
4	* This program is free software: you can use, redistribute, and/or modify
5	* it under the terms of the GNU Affero General Public License, version 3
6	* or later ("AGPL"), as published by the Free Software Foundation.
7	*
8	* This program is distributed in the hope that it will be useful, but WITHOUT
9	* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10	* FITNESS FOR A PARTICULAR PURPOSE.
11	*
12	* You should have received a copy of the GNU Affero General Public License
13	* along with this program. If not, see <http://www.gnu.org/licenses/>.
14	*/
15
16	#define _DEFAULT_SOURCE
17	#include "syncReplication.h"
18	#include "syncIndexMgr.h"
19	#include "syncPipeline.h"
20	#include "syncRaftEntry.h"
21	#include "syncRaftStore.h"
22	#include "syncUtil.h"
23
24	// TLA+ Spec
25	// AppendEntries(i, j) ==
26	// /\ i /= j
27	// /\ state[i] = Leader
28	// /\ LET prevLogIndex == nextIndex[i][j] - 1
29	// prevLogTerm == IF prevLogIndex > 0 THEN
30	// log[i][prevLogIndex].term
31	// ELSE
32	// 0
33	// \* Send up to 1 entry, constrained by the end of the log.
34	// lastEntry == Min({Len(log[i]), nextIndex[i][j]})
35	// entries == SubSeq(log[i], nextIndex[i][j], lastEntry)
36	// IN Send([mtype \|-> AppendEntriesRequest,
37	// mterm \|-> currentTerm[i],
38	// mprevLogIndex \|-> prevLogIndex,
39	// mprevLogTerm \|-> prevLogTerm,
40	// mentries \|-> entries,
41	// \* mlog is used as a history variable for the proof.
42	// \* It would not exist in a real implementation.
43	// mlog \|-> log[i],
44	// mcommitIndex \|-> Min({commitIndex[i], lastEntry}),
45	// msource \|-> i,
46	// mdest \|-> j])
47	// /\ UNCHANGED <<serverVars, candidateVars, leaderVars, logVars>>
48
49	int32_t syncNodeReplicateReset(SSyncNode* pNode, SRaftId* pDestId) {	69✔
50	SSyncLogBuffer* pBuf = pNode->pLogBuf;	69✔
51	(void)taosThreadMutexLock(&pBuf->mutex);	69✔
52	SSyncLogReplMgr* pMgr = syncNodeGetLogReplMgr(pNode, pDestId);	69✔
53	syncLogReplReset(pMgr);	69✔
54	(void)taosThreadMutexUnlock(&pBuf->mutex);	69✔
55
56	TAOS_RETURN(TSDB_CODE_SUCCESS);	69✔
57	}
58
59	int32_t syncNodeReplicate(SSyncNode* pNode) {	208,910✔
60	SSyncLogBuffer* pBuf = pNode->pLogBuf;	208,910✔
61	(void)taosThreadMutexLock(&pBuf->mutex);	208,910✔
62	int32_t ret = syncNodeReplicateWithoutLock(pNode);	209,298✔
63	(void)taosThreadMutexUnlock(&pBuf->mutex);	209,028✔
64
65	TAOS_RETURN(ret);	209,337✔
66	}
67
68	int32_t syncNodeReplicateWithoutLock(SSyncNode* pNode) {	13,368,690✔
69	if ((pNode->state != TAOS_SYNC_STATE_LEADER && pNode->state != TAOS_SYNC_STATE_ASSIGNED_LEADER) \|\|	13,368,690!
70	pNode->raftCfg.cfg.totalReplicaNum == 1) {	10,538,024✔
71	TAOS_RETURN(TSDB_CODE_SUCCESS);	13,111,172✔
72	}
73	for (int32_t i = 0; i < pNode->totalReplicaNum; i++) {	896,568✔
74	if (syncUtilSameId(&pNode->replicasId[i], &pNode->myRaftId)) {	639,052✔
75	continue;	257,517✔
76	}
77	SSyncLogReplMgr* pMgr = pNode->logReplMgrs[i];	381,532✔
78	int32_t ret = 0;	381,532✔
79	if ((ret = syncLogReplStart(pMgr, pNode)) != 0) {	381,532✔
80	sWarn("vgId:%d, failed to start log replication to dnode:%d since %s", pNode->vgId, DID(&(pNode->replicasId[i])),	46!
81	tstrerror(ret));
82	}
83	}
84
85	TAOS_RETURN(TSDB_CODE_SUCCESS);	257,516✔
86	}
87
88	int32_t syncNodeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) {	2,851,418✔
89	SyncAppendEntries* pMsg = pRpcMsg->pCont;	2,851,418✔
90	pMsg->destId = *destRaftId;	2,851,418✔
91	TAOS_CHECK_RETURN(syncNodeSendMsgById(destRaftId, pSyncNode, pRpcMsg));	2,851,418!
92
93	int32_t nRef = 0;	2,851,420✔
94	if (pSyncNode != NULL) {	2,851,420!
95	nRef = atomic_add_fetch_32(&pSyncNode->sendCount, 1);	2,851,420✔
96	if (nRef <= 0) {	2,851,421!
97	sError("vgId:%d, send count is %d", pSyncNode->vgId, nRef);	×
98	}
99	}
100
101	SSyncLogReplMgr* mgr = syncNodeGetLogReplMgr(pSyncNode, (SRaftId*)destRaftId);	2,851,421✔
102	if (mgr != NULL) {	2,851,420!
103	nRef = atomic_add_fetch_32(&mgr->sendCount, 1);	2,851,420✔
104	if (nRef <= 0) {	2,851,421!
105	sError("vgId:%d, send count is %d", pSyncNode->vgId, nRef);	×
106	}
107	}
108
109	TAOS_RETURN(TSDB_CODE_SUCCESS);	2,851,421✔
110	}
111
112	int32_t syncNodeSendHeartbeat(SSyncNode* pSyncNode, const SRaftId* destId, SRpcMsg* pMsg) {	36,960✔
113	SRaftId destIdTmp = *destId;	36,960✔
114	TAOS_CHECK_RETURN(syncNodeSendMsgById(destId, pSyncNode, pMsg));	36,960✔
115
116	int64_t tsMs = taosGetTimestampMs();	36,905✔
117	syncIndexMgrSetSentTime(pSyncNode->pMatchIndex, &destIdTmp, tsMs);	36,905✔
118
119	return TSDB_CODE_SUCCESS;	36,905✔
120	}
121
122	int32_t syncNodeHeartbeatPeers(SSyncNode* pSyncNode) {	11,202✔
123	int64_t ts = taosGetTimestampMs();	11,202✔
124	for (int32_t i = 0; i < pSyncNode->peersNum; ++i) {	13,230✔
125	SRpcMsg rpcMsg = {0};	2,028✔
126	if (syncBuildHeartbeat(&rpcMsg, pSyncNode->vgId) != 0) {	2,028!
127	sError("vgId:%d, build sync-heartbeat error", pSyncNode->vgId);	×
128	continue;	×
129	}
130
131	SyncHeartbeat* pSyncMsg = rpcMsg.pCont;	2,028✔
132	pSyncMsg->srcId = pSyncNode->myRaftId;	2,028✔
133	pSyncMsg->destId = pSyncNode->peersId[i];	2,028✔
134	pSyncMsg->term = raftStoreGetTerm(pSyncNode);	2,028✔
135	pSyncMsg->commitIndex = pSyncNode->commitIndex;	2,028✔
136	pSyncMsg->minMatchIndex = syncMinMatchIndex(pSyncNode);	2,028✔
137	pSyncMsg->privateTerm = 0;	2,028✔
138	pSyncMsg->timeStamp = ts;	2,028✔
139
140	// send msg
141	TRACE_SET_MSGID(&(rpcMsg.info.traceId), tGenIdPI64());	2,028✔
142	STraceId* trace = &(rpcMsg.info.traceId);	2,028✔
143	sGTrace("vgId:%d, send sync-heartbeat to dnode:%d", pSyncNode->vgId, DID(&(pSyncMsg->destId)));	2,028!
144	syncLogSendHeartbeat(pSyncNode, pSyncMsg, true, 0, 0);	2,028✔
145	int32_t ret = syncNodeSendHeartbeat(pSyncNode, &pSyncMsg->destId, &rpcMsg);	2,028✔
146	if (ret != 0) {	2,028!
147	sError("vgId:%d, failed to send sync-heartbeat since %s", pSyncNode->vgId, tstrerror(ret));	×
148	}
149	}
150
151	TAOS_RETURN(TSDB_CODE_SUCCESS);	11,202✔
152	}

taosdata / TDengine / #3548

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous