homework-jianmu/source/libs/sync/src/syncReplication.c

530 lines
16 KiB
C

/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "syncReplication.h"
#include "syncIndexMgr.h"
#include "syncMessage.h"
#include "syncRaftCfg.h"
#include "syncRaftEntry.h"
#include "syncRaftLog.h"
#include "syncRaftStore.h"
#include "syncSnapshot.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 syncNodeAppendEntriesPeers(SSyncNode* pSyncNode) {
ASSERT(pSyncNode->state == TAOS_SYNC_STATE_LEADER);
syncIndexMgrLog2("==syncNodeAppendEntriesPeers== pNextIndex", pSyncNode->pNextIndex);
syncIndexMgrLog2("==syncNodeAppendEntriesPeers== pMatchIndex", pSyncNode->pMatchIndex);
logStoreSimpleLog2("==syncNodeAppendEntriesPeers==", pSyncNode->pLogStore);
int32_t ret = 0;
for (int i = 0; i < pSyncNode->peersNum; ++i) {
SRaftId* pDestId = &(pSyncNode->peersId[i]);
// set prevLogIndex
SyncIndex nextIndex = syncIndexMgrGetIndex(pSyncNode->pNextIndex, pDestId);
SyncIndex preLogIndex = nextIndex - 1;
// set preLogTerm
SyncTerm preLogTerm = 0;
if (preLogIndex >= SYNC_INDEX_BEGIN) {
SSyncRaftEntry* pPreEntry = pSyncNode->pLogStore->getEntry(pSyncNode->pLogStore, preLogIndex);
ASSERT(pPreEntry != NULL);
preLogTerm = pPreEntry->term;
syncEntryDestory(pPreEntry);
}
// batch optimized
// SyncIndex lastIndex = syncUtilMinIndex(pSyncNode->pLogStore->getLastIndex(pSyncNode->pLogStore), nextIndex);
SyncAppendEntries* pMsg = NULL;
SSyncRaftEntry* pEntry = pSyncNode->pLogStore->getEntry(pSyncNode->pLogStore, nextIndex);
if (pEntry != NULL) {
pMsg = syncAppendEntriesBuild(pEntry->bytes, pSyncNode->vgId);
ASSERT(pMsg != NULL);
// add pEntry into msg
uint32_t len;
char* serialized = syncEntrySerialize(pEntry, &len);
ASSERT(len == pEntry->bytes);
memcpy(pMsg->data, serialized, len);
taosMemoryFree(serialized);
syncEntryDestory(pEntry);
} else {
// maybe overflow, send empty record
pMsg = syncAppendEntriesBuild(0, pSyncNode->vgId);
ASSERT(pMsg != NULL);
}
ASSERT(pMsg != NULL);
pMsg->srcId = pSyncNode->myRaftId;
pMsg->destId = *pDestId;
pMsg->term = pSyncNode->pRaftStore->currentTerm;
pMsg->prevLogIndex = preLogIndex;
pMsg->prevLogTerm = preLogTerm;
pMsg->commitIndex = pSyncNode->commitIndex;
syncAppendEntriesLog2("==syncNodeAppendEntriesPeers==", pMsg);
// send AppendEntries
syncNodeAppendEntries(pSyncNode, pDestId, pMsg);
syncAppendEntriesDestroy(pMsg);
}
return ret;
}
int32_t syncNodeAppendEntriesOnePeer(SSyncNode* pSyncNode, SRaftId* pDestId, SyncIndex nextIndex) {
int32_t ret = 0;
// pre index, pre term
SyncIndex preLogIndex = syncNodeGetPreIndex(pSyncNode, nextIndex);
SyncTerm preLogTerm = syncNodeGetPreTerm(pSyncNode, nextIndex);
if (preLogTerm == SYNC_TERM_INVALID) {
SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
// SyncIndex newNextIndex = nextIndex + 1;
syncIndexMgrSetIndex(pSyncNode->pNextIndex, pDestId, newNextIndex);
syncIndexMgrSetIndex(pSyncNode->pMatchIndex, pDestId, SYNC_INDEX_INVALID);
sError("vgId:%d, sync get pre term error, nextIndex:%" PRId64 ", update next-index:%" PRId64
", match-index:%d, raftid:%" PRId64,
pSyncNode->vgId, nextIndex, newNextIndex, SYNC_INDEX_INVALID, pDestId->addr);
return -1;
}
// entry pointer array
SSyncRaftEntry* entryPArr[SYNC_MAX_BATCH_SIZE];
memset(entryPArr, 0, sizeof(entryPArr));
// get entry batch
int32_t getCount = 0;
SyncIndex getEntryIndex = nextIndex;
for (int32_t i = 0; i < pSyncNode->pRaftCfg->batchSize; ++i) {
SSyncRaftEntry* pEntry = NULL;
int32_t code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, getEntryIndex, &pEntry);
if (code == 0) {
ASSERT(pEntry != NULL);
entryPArr[i] = pEntry;
getCount++;
getEntryIndex++;
} else {
break;
}
}
// event log
do {
char logBuf[128];
char host[64];
uint16_t port;
syncUtilU642Addr(pDestId->addr, host, sizeof(host), &port);
snprintf(logBuf, sizeof(logBuf), "build batch:%d for %s:%d", getCount, host, port);
syncNodeEventLog(pSyncNode, logBuf);
} while (0);
// build msg
SyncAppendEntriesBatch* pMsg = syncAppendEntriesBatchBuild(entryPArr, getCount, pSyncNode->vgId);
ASSERT(pMsg != NULL);
// free entries
for (int32_t i = 0; i < pSyncNode->pRaftCfg->batchSize; ++i) {
SSyncRaftEntry* pEntry = entryPArr[i];
if (pEntry != NULL) {
syncEntryDestory(pEntry);
entryPArr[i] = NULL;
}
}
// prepare msg
pMsg->srcId = pSyncNode->myRaftId;
pMsg->destId = *pDestId;
pMsg->term = pSyncNode->pRaftStore->currentTerm;
pMsg->prevLogIndex = preLogIndex;
pMsg->prevLogTerm = preLogTerm;
pMsg->commitIndex = pSyncNode->commitIndex;
pMsg->privateTerm = 0;
pMsg->dataCount = getCount;
// send msg
syncNodeAppendEntriesBatch(pSyncNode, pDestId, pMsg);
// speed up
if (pMsg->dataCount > 0 && pSyncNode->commitIndex - pMsg->prevLogIndex > SYNC_SLOW_DOWN_RANGE) {
ret = 1;
#if 0
do {
char logBuf[128];
char host[64];
uint16_t port;
syncUtilU642Addr(pDestId->addr, host, sizeof(host), &port);
snprintf(logBuf, sizeof(logBuf), "maybe speed up for %s:%d, pre-index:%ld", host, port, pMsg->prevLogIndex);
syncNodeEventLog(pSyncNode, logBuf);
} while (0);
#endif
}
syncAppendEntriesBatchDestroy(pMsg);
return ret;
}
int32_t syncNodeAppendEntriesPeersSnapshot2(SSyncNode* pSyncNode) {
if (pSyncNode->state != TAOS_SYNC_STATE_LEADER) {
return -1;
}
int32_t ret = 0;
for (int i = 0; i < pSyncNode->peersNum; ++i) {
SRaftId* pDestId = &(pSyncNode->peersId[i]);
// next index
SyncIndex nextIndex = syncIndexMgrGetIndex(pSyncNode->pNextIndex, pDestId);
ret = syncNodeAppendEntriesOnePeer(pSyncNode, pDestId, nextIndex);
}
return ret;
}
#if 0
int32_t syncNodeAppendEntriesPeersSnapshot2(SSyncNode* pSyncNode) {
if (pSyncNode->state != TAOS_SYNC_STATE_LEADER) {
return -1;
}
int32_t ret = 0;
for (int i = 0; i < pSyncNode->peersNum; ++i) {
SRaftId* pDestId = &(pSyncNode->peersId[i]);
// next index
SyncIndex nextIndex = syncIndexMgrGetIndex(pSyncNode->pNextIndex, pDestId);
// pre index, pre term
SyncIndex preLogIndex = syncNodeGetPreIndex(pSyncNode, nextIndex);
SyncTerm preLogTerm = syncNodeGetPreTerm(pSyncNode, nextIndex);
if (preLogTerm == SYNC_TERM_INVALID) {
SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
// SyncIndex newNextIndex = nextIndex + 1;
syncIndexMgrSetIndex(pSyncNode->pNextIndex, pDestId, newNextIndex);
syncIndexMgrSetIndex(pSyncNode->pMatchIndex, pDestId, SYNC_INDEX_INVALID);
sError("vgId:%d, sync get pre term error, nextIndex:%" PRId64 ", update next-index:%" PRId64
", match-index:%d, raftid:%" PRId64,
pSyncNode->vgId, nextIndex, newNextIndex, SYNC_INDEX_INVALID, pDestId->addr);
return -1;
}
// entry pointer array
SSyncRaftEntry* entryPArr[SYNC_MAX_BATCH_SIZE];
memset(entryPArr, 0, sizeof(entryPArr));
// get entry batch
int32_t getCount = 0;
SyncIndex getEntryIndex = nextIndex;
for (int32_t i = 0; i < pSyncNode->pRaftCfg->batchSize; ++i) {
SSyncRaftEntry* pEntry = NULL;
int32_t code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, getEntryIndex, &pEntry);
if (code == 0) {
ASSERT(pEntry != NULL);
entryPArr[i] = pEntry;
getCount++;
getEntryIndex++;
} else {
break;
}
}
// event log
do {
char logBuf[128];
char host[64];
uint16_t port;
syncUtilU642Addr(pDestId->addr, host, sizeof(host), &port);
snprintf(logBuf, sizeof(logBuf), "build batch:%d for %s:%d", getCount, host, port);
syncNodeEventLog(pSyncNode, logBuf);
} while (0);
// build msg
SyncAppendEntriesBatch* pMsg = syncAppendEntriesBatchBuild(entryPArr, getCount, pSyncNode->vgId);
ASSERT(pMsg != NULL);
// free entries
for (int32_t i = 0; i < pSyncNode->pRaftCfg->batchSize; ++i) {
SSyncRaftEntry* pEntry = entryPArr[i];
if (pEntry != NULL) {
syncEntryDestory(pEntry);
entryPArr[i] = NULL;
}
}
// prepare msg
pMsg->srcId = pSyncNode->myRaftId;
pMsg->destId = *pDestId;
pMsg->term = pSyncNode->pRaftStore->currentTerm;
pMsg->prevLogIndex = preLogIndex;
pMsg->prevLogTerm = preLogTerm;
pMsg->commitIndex = pSyncNode->commitIndex;
pMsg->privateTerm = 0;
pMsg->dataCount = getCount;
// send msg
syncNodeAppendEntriesBatch(pSyncNode, pDestId, pMsg);
// speed up
if (pMsg->dataCount > 0 && pSyncNode->commitIndex - pMsg->prevLogIndex > SYNC_SLOW_DOWN_RANGE) {
ret = 1;
#if 0
do {
char logBuf[128];
char host[64];
uint16_t port;
syncUtilU642Addr(pDestId->addr, host, sizeof(host), &port);
snprintf(logBuf, sizeof(logBuf), "maybe speed up for %s:%d, pre-index:%ld", host, port, pMsg->prevLogIndex);
syncNodeEventLog(pSyncNode, logBuf);
} while (0);
#endif
}
syncAppendEntriesBatchDestroy(pMsg);
}
return ret;
}
#endif
int32_t syncNodeAppendEntriesPeersSnapshot(SSyncNode* pSyncNode) {
ASSERT(pSyncNode->state == TAOS_SYNC_STATE_LEADER);
syncIndexMgrLog2("begin append entries peers pNextIndex:", pSyncNode->pNextIndex);
syncIndexMgrLog2("begin append entries peers pMatchIndex:", pSyncNode->pMatchIndex);
logStoreSimpleLog2("begin append entries peers LogStore:", pSyncNode->pLogStore);
int32_t ret = 0;
for (int i = 0; i < pSyncNode->peersNum; ++i) {
SRaftId* pDestId = &(pSyncNode->peersId[i]);
// next index
SyncIndex nextIndex = syncIndexMgrGetIndex(pSyncNode->pNextIndex, pDestId);
// pre index, pre term
SyncIndex preLogIndex = syncNodeGetPreIndex(pSyncNode, nextIndex);
SyncTerm preLogTerm = syncNodeGetPreTerm(pSyncNode, nextIndex);
if (preLogTerm == SYNC_TERM_INVALID) {
SyncIndex newNextIndex = syncNodeGetLastIndex(pSyncNode) + 1;
// SyncIndex newNextIndex = nextIndex + 1;
syncIndexMgrSetIndex(pSyncNode->pNextIndex, pDestId, newNextIndex);
syncIndexMgrSetIndex(pSyncNode->pMatchIndex, pDestId, SYNC_INDEX_INVALID);
sError("vgId:%d, sync get pre term error, nextIndex:%" PRId64 ", update next-index:%" PRId64
", match-index:%d, raftid:%" PRId64,
pSyncNode->vgId, nextIndex, newNextIndex, SYNC_INDEX_INVALID, pDestId->addr);
return -1;
}
// prepare entry
SyncAppendEntries* pMsg = NULL;
SSyncRaftEntry* pEntry;
int32_t code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, nextIndex, &pEntry);
if (code == 0) {
ASSERT(pEntry != NULL);
pMsg = syncAppendEntriesBuild(pEntry->bytes, pSyncNode->vgId);
ASSERT(pMsg != NULL);
// add pEntry into msg
uint32_t len;
char* serialized = syncEntrySerialize(pEntry, &len);
ASSERT(len == pEntry->bytes);
memcpy(pMsg->data, serialized, len);
taosMemoryFree(serialized);
syncEntryDestory(pEntry);
} else {
if (terrno == TSDB_CODE_WAL_LOG_NOT_EXIST) {
// no entry in log
pMsg = syncAppendEntriesBuild(0, pSyncNode->vgId);
ASSERT(pMsg != NULL);
} else {
syncNodeLog3("", pSyncNode);
ASSERT(0);
}
}
// prepare msg
ASSERT(pMsg != NULL);
pMsg->srcId = pSyncNode->myRaftId;
pMsg->destId = *pDestId;
pMsg->term = pSyncNode->pRaftStore->currentTerm;
pMsg->prevLogIndex = preLogIndex;
pMsg->prevLogTerm = preLogTerm;
pMsg->commitIndex = pSyncNode->commitIndex;
pMsg->privateTerm = 0;
// pMsg->privateTerm = syncIndexMgrGetTerm(pSyncNode->pNextIndex, pDestId);
// send msg
syncNodeAppendEntries(pSyncNode, pDestId, pMsg);
syncAppendEntriesDestroy(pMsg);
}
return ret;
}
int32_t syncNodeReplicate(SSyncNode* pSyncNode, bool isTimer) {
// start replicate
int32_t ret = 0;
switch (pSyncNode->pRaftCfg->snapshotStrategy) {
case SYNC_STRATEGY_NO_SNAPSHOT:
ret = syncNodeAppendEntriesPeers(pSyncNode);
break;
case SYNC_STRATEGY_STANDARD_SNAPSHOT:
ret = syncNodeAppendEntriesPeersSnapshot(pSyncNode);
break;
case SYNC_STRATEGY_WAL_FIRST:
ret = syncNodeAppendEntriesPeersSnapshot2(pSyncNode);
break;
default:
ret = syncNodeAppendEntriesPeers(pSyncNode);
break;
}
// start delay
int64_t timeNow = taosGetTimestampMs();
int64_t startDelay = timeNow - pSyncNode->startTime;
// replicate delay
int64_t replicateDelay = timeNow - pSyncNode->lastReplicateTime;
pSyncNode->lastReplicateTime = timeNow;
if (ret > 0 && isTimer && startDelay > SYNC_SPEED_UP_AFTER_MS) {
// speed up replicate
int32_t ms =
pSyncNode->heartbeatTimerMS < SYNC_SPEED_UP_HB_TIMER ? pSyncNode->heartbeatTimerMS : SYNC_SPEED_UP_HB_TIMER;
syncNodeRestartNowHeartbeatTimerMS(pSyncNode, ms);
#if 0
do {
char logBuf[128];
snprintf(logBuf, sizeof(logBuf), "replicate speed up");
syncNodeEventLog(pSyncNode, logBuf);
} while (0);
#endif
} else {
syncNodeRestartHeartbeatTimer(pSyncNode);
#if 0
do {
char logBuf[128];
snprintf(logBuf, sizeof(logBuf), "replicate slow down");
syncNodeEventLog(pSyncNode, logBuf);
} while (0);
#endif
}
return ret;
}
int32_t syncNodeAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, const SyncAppendEntries* pMsg) {
int32_t ret = 0;
syncLogSendAppendEntries(pSyncNode, pMsg, "");
SRpcMsg rpcMsg;
syncAppendEntries2RpcMsg(pMsg, &rpcMsg);
syncNodeSendMsgById(destRaftId, pSyncNode, &rpcMsg);
return ret;
}
int32_t syncNodeAppendEntriesBatch(SSyncNode* pSyncNode, const SRaftId* destRaftId,
const SyncAppendEntriesBatch* pMsg) {
syncLogSendAppendEntriesBatch(pSyncNode, pMsg, "");
SRpcMsg rpcMsg;
syncAppendEntriesBatch2RpcMsg(pMsg, &rpcMsg);
syncNodeSendMsgById(destRaftId, pSyncNode, &rpcMsg);
return 0;
}
int32_t syncNodeHeartbeat(SSyncNode* pSyncNode, const SRaftId* destRaftId, const SyncHeartbeat* pMsg) {
int32_t ret = 0;
syncLogSendHeartbeat(pSyncNode, pMsg, "");
SRpcMsg rpcMsg;
syncHeartbeat2RpcMsg(pMsg, &rpcMsg);
syncNodeSendMsgById(&(pMsg->destId), pSyncNode, &rpcMsg);
return ret;
}
int32_t syncNodeHeartbeatPeers(SSyncNode* pSyncNode) {
for (int32_t i = 0; i < pSyncNode->peersNum; ++i) {
SyncHeartbeat* pSyncMsg = syncHeartbeatBuild(pSyncNode->vgId);
pSyncMsg->srcId = pSyncNode->myRaftId;
pSyncMsg->destId = pSyncNode->peersId[i];
pSyncMsg->term = pSyncNode->pRaftStore->currentTerm;
pSyncMsg->commitIndex = pSyncNode->commitIndex;
pSyncMsg->privateTerm = 0;
SRpcMsg rpcMsg;
syncHeartbeat2RpcMsg(pSyncMsg, &rpcMsg);
// send msg
syncNodeHeartbeat(pSyncNode, &(pSyncMsg->destId), pSyncMsg);
syncHeartbeatDestroy(pSyncMsg);
}
return 0;
}