Merge pull request #17322 from taosdata/FIX/TD-19239-3.0

fix: enhance subroutines to resolve WAL inconsistency after crashing or disk full
This commit is contained in:
Shengliang Guan 2022-10-17 09:06:16 +08:00 committed by GitHub
commit 4223b43a3a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
26 changed files with 727 additions and 412 deletions

View File

@ -43,6 +43,7 @@ extern "C" {
#define WAL_FILE_LEN (WAL_PATH_LEN + 32)
#define WAL_MAGIC 0xFAFBFCFDF4F3F2F1ULL
#define WAL_SCAN_BUF_SIZE (1024 * 1024 * 3)
#define WAL_RECOV_SIZE_LIMIT (100 * WAL_SCAN_BUF_SIZE)
typedef enum {
TAOS_WAL_WRITE = 1,

View File

@ -49,9 +49,15 @@ extern SDiskSpace tsTempSpace;
void osDefaultInit();
void osUpdate();
void osCleanup();
bool osLogSpaceAvailable();
bool osDataSpaceAvailable();
bool osTempSpaceAvailable();
bool osLogSpaceSufficient();
bool osDataSpaceSufficient();
bool osTempSpaceSufficient();
void osSetTimezone(const char *timezone);
void osSetSystemLocale(const char *inLocale, const char *inCharSet);
@ -59,4 +65,4 @@ void osSetSystemLocale(const char *inLocale, const char *inCharSet);
}
#endif
#endif /*_TD_OS_ENV_H_*/
#endif /*_TD_OS_ENV_H_*/

View File

@ -450,6 +450,7 @@ int32_t* taosGetErrno();
#define TSDB_CODE_WAL_INVALID_VER TAOS_DEF_ERROR_CODE(0, 0x1003)
#define TSDB_CODE_WAL_OUT_OF_MEMORY TAOS_DEF_ERROR_CODE(0, 0x1004)
#define TSDB_CODE_WAL_LOG_NOT_EXIST TAOS_DEF_ERROR_CODE(0, 0x1005)
#define TSDB_CODE_WAL_CHKSUM_MISMATCH TAOS_DEF_ERROR_CODE(0, 0x1006)
// tfs
#define TSDB_CODE_FS_INVLD_CFG TAOS_DEF_ERROR_CODE(0, 0x2201)

View File

@ -146,7 +146,7 @@ void taos_close(TAOS *taos) {
int taos_errno(TAOS_RES *res) {
if (res == NULL || TD_RES_TMQ_META(res)) {
if (terrno == TSDB_CODE_RPC_REDIRECT) terrno = TSDB_CODE_RPC_NETWORK_UNAVAIL;
if (terrno == TSDB_CODE_RPC_REDIRECT) terrno = TSDB_CODE_QRY_NOT_READY;
return terrno;
}
@ -154,13 +154,12 @@ int taos_errno(TAOS_RES *res) {
return 0;
}
return ((SRequestObj *)res)->code == TSDB_CODE_RPC_REDIRECT ? TSDB_CODE_RPC_NETWORK_UNAVAIL
: ((SRequestObj *)res)->code;
return ((SRequestObj *)res)->code == TSDB_CODE_RPC_REDIRECT ? TSDB_CODE_QRY_NOT_READY : ((SRequestObj *)res)->code;
}
const char *taos_errstr(TAOS_RES *res) {
if (res == NULL || TD_RES_TMQ_META(res)) {
if (terrno == TSDB_CODE_RPC_REDIRECT) terrno = TSDB_CODE_RPC_NETWORK_UNAVAIL;
if (terrno == TSDB_CODE_RPC_REDIRECT) terrno = TSDB_CODE_QRY_NOT_READY;
return (const char *)tstrerror(terrno);
}
@ -172,7 +171,7 @@ const char *taos_errstr(TAOS_RES *res) {
if (NULL != pRequest->msgBuf && (strlen(pRequest->msgBuf) > 0 || pRequest->code == TSDB_CODE_RPC_FQDN_ERROR)) {
return pRequest->msgBuf;
} else {
return pRequest->code == TSDB_CODE_RPC_REDIRECT ? (const char *)tstrerror(TSDB_CODE_RPC_NETWORK_UNAVAIL)
return pRequest->code == TSDB_CODE_RPC_REDIRECT ? (const char *)tstrerror(TSDB_CODE_QRY_NOT_READY)
: (const char *)tstrerror(pRequest->code);
}
}

View File

@ -51,26 +51,14 @@ static int32_t dmInitMonitor() {
static bool dmCheckDiskSpace() {
osUpdate();
if (!osDataSpaceAvailable()) {
dError("free disk size: %f GB, too little, require %f GB at least at least , quit",
(double)tsDataSpace.size.avail / 1024.0 / 1024.0 / 1024.0,
(double)tsDataSpace.reserved / 1024.0 / 1024.0 / 1024.0);
terrno = TSDB_CODE_NO_AVAIL_DISK;
return false;
if (!osDataSpaceSufficient()) {
dWarn("free data disk size: %f GB, not sufficient, expected %f GB at least", (double)tsDataSpace.size.avail / 1024.0 / 1024.0 / 1024.0, (double)tsDataSpace.reserved / 1024.0 / 1024.0 / 1024.0);
}
if (!osLogSpaceAvailable()) {
dError("free disk size: %f GB, too little, require %f GB at least at least, quit",
(double)tsLogSpace.size.avail / 1024.0 / 1024.0 / 1024.0,
(double)tsLogSpace.reserved / 1024.0 / 1024.0 / 1024.0);
terrno = TSDB_CODE_NO_AVAIL_DISK;
return false;
if (!osLogSpaceSufficient()) {
dWarn("free log disk size: %f GB, not sufficient, expected %f GB at least", (double)tsLogSpace.size.avail / 1024.0 / 1024.0 / 1024.0, (double)tsLogSpace.reserved / 1024.0 / 1024.0 / 1024.0);
}
if (!osTempSpaceAvailable()) {
dError("free disk size: %f GB, too little, require %f GB at least at least, quit",
(double)tsTempSpace.size.avail / 1024.0 / 1024.0 / 1024.0,
(double)tsTempSpace.reserved / 1024.0 / 1024.0 / 1024.0);
terrno = TSDB_CODE_NO_AVAIL_DISK;
return false;
if (!osTempSpaceSufficient()) {
dWarn("free temp disk size: %f GB, not sufficient, expected %f GB at least", (double)tsTempSpace.size.avail / 1024.0 / 1024.0 / 1024.0, (double)tsTempSpace.reserved / 1024.0 / 1024.0 / 1024.0);
}
return true;
}

View File

@ -232,7 +232,7 @@ static int32_t mndInitWal(SMnode *pMnode) {
pMnode->pWal = walOpen(path, &cfg);
if (pMnode->pWal == NULL) {
mError("failed to open wal since %s", terrstr());
mError("failed to open wal since %s. wal:%s", terrstr(), path);
return -1;
}

View File

@ -1045,7 +1045,9 @@ static int32_t tsdbEndCommit(SCommitter *pCommitter, int32_t eno) {
STsdb *pTsdb = pCommitter->pTsdb;
SMemTable *pMemTable = pTsdb->imem;
ASSERT(eno == 0);
ASSERT(eno == 0 &&
"tsdbCommit failure"
"Restart taosd");
code = tsdbFSCommit1(pTsdb, &pCommitter->fs);
TSDB_CHECK_CODE(code, lino, _exit);

View File

@ -160,6 +160,7 @@ int32_t tsdbDeleteTableData(STsdb *pTsdb, int64_t version, tb_uid_t suid, tb_uid
goto _err;
}
ASSERT(pPool != NULL);
// do delete
SDelData *pDelData = (SDelData *)vnodeBufPoolMalloc(pPool, sizeof(*pDelData));
if (pDelData == NULL) {
@ -353,6 +354,7 @@ static int32_t tsdbGetOrCreateTbData(SMemTable *pMemTable, tb_uid_t suid, tb_uid
SVBufPool *pPool = pMemTable->pTsdb->pVnode->inUse;
int8_t maxLevel = pMemTable->pTsdb->pVnode->config.tsdbCfg.slLevel;
ASSERT(pPool != NULL);
pTbData = vnodeBufPoolMalloc(pPool, sizeof(*pTbData) + SL_NODE_SIZE(maxLevel) * 2);
if (pTbData == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
@ -492,6 +494,7 @@ static int32_t tbDataDoPut(SMemTable *pMemTable, STbData *pTbData, SMemSkipListN
// node
level = tsdbMemSkipListRandLevel(&pTbData->sl);
ASSERT(pPool != NULL);
pNode = (SMemSkipListNode *)vnodeBufPoolMalloc(pPool, SL_NODE_SIZE(level) + tPutTSDBRow(NULL, pRow));
if (pNode == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;

View File

@ -112,6 +112,8 @@ void vnodeBufPoolReset(SVBufPool *pPool) {
void *vnodeBufPoolMalloc(SVBufPool *pPool, int size) {
SVBufPoolNode *pNode;
void *p = NULL;
ASSERT(pPool != NULL);
taosThreadSpinLock(&pPool->lock);
if (pPool->node.size >= pPool->ptr - pPool->node.data + size) {
// allocate from the anchor node

View File

@ -73,7 +73,7 @@ int vnodeBegin(SVnode *pVnode) {
int vnodeShouldCommit(SVnode *pVnode) {
if (pVnode->inUse) {
return pVnode->inUse->size > pVnode->inUse->node.size;
return osDataSpaceAvailable() && (pVnode->inUse->size > pVnode->inUse->node.size);
}
return false;
}
@ -89,6 +89,7 @@ int vnodeSaveInfo(const char *dir, const SVnodeInfo *pInfo) {
data = NULL;
if (vnodeEncodeInfo(pInfo, &data) < 0) {
vError("failed to encode json info.");
return -1;
}
@ -101,7 +102,7 @@ int vnodeSaveInfo(const char *dir, const SVnodeInfo *pInfo) {
}
if (taosWriteFile(pFile, data, strlen(data)) < 0) {
vError("failed to write info file:%s data:%s", fname, terrstr());
vError("failed to write info file:%s error:%s", fname, terrstr());
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
@ -233,15 +234,15 @@ int vnodeCommit(SVnode *pVnode) {
snprintf(dir, TSDB_FILENAME_LEN, "%s", pVnode->path);
}
if (vnodeSaveInfo(dir, &info) < 0) {
ASSERT(0);
vError("vgId:%d, failed to save vnode info since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
walBeginSnapshot(pVnode->pWal, pVnode->state.applied);
// preCommit
// smaSyncPreCommit(pVnode->pSma);
if (smaAsyncPreCommit(pVnode->pSma) < 0) {
ASSERT(0);
if(smaAsyncPreCommit(pVnode->pSma) < 0){
vError("vgId:%d, failed to async pre-commit sma since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
@ -250,44 +251,44 @@ int vnodeCommit(SVnode *pVnode) {
// commit each sub-system
if (metaCommit(pVnode->pMeta) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit meta since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
if (VND_IS_RSMA(pVnode)) {
if (smaAsyncCommit(pVnode->pSma) < 0) {
ASSERT(0);
vError("vgId:%d, failed to async commit sma since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
if (tsdbCommit(VND_RSMA0(pVnode)) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit tsdb rsma0 since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
if (tsdbCommit(VND_RSMA1(pVnode)) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit tsdb rsma1 since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
if (tsdbCommit(VND_RSMA2(pVnode)) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit tsdb rsma2 since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
} else {
if (tsdbCommit(pVnode->pTsdb) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit tsdb since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
}
if (tqCommit(pVnode->pTq) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit tq since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
// walCommit (TODO)
// commit info
if (vnodeCommitInfo(dir, &info) < 0) {
ASSERT(0);
vError("vgId:%d, failed to commit vnode info since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
@ -296,7 +297,7 @@ int vnodeCommit(SVnode *pVnode) {
// postCommit
// smaSyncPostCommit(pVnode->pSma);
if (smaAsyncPostCommit(pVnode->pSma) < 0) {
ASSERT(0);
vError("vgId:%d, failed to async post-commit sma since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}

View File

@ -140,7 +140,7 @@ SVnode *vnodeOpen(const char *path, STfs *pTfs, SMsgCb msgCb) {
pVnode->pWal = walOpen(tdir, &(pVnode->config.walCfg));
if (pVnode->pWal == NULL) {
vError("vgId:%d, failed to open vnode wal since %s", TD_VID(pVnode), tstrerror(terrno));
vError("vgId:%d, failed to open vnode wal since %s. wal:%s", TD_VID(pVnode), tstrerror(terrno), tdir);
goto _err;
}

View File

@ -166,6 +166,12 @@ int32_t vnodeProcessWriteMsg(SVnode *pVnode, SRpcMsg *pMsg, int64_t version, SRp
int32_t len;
int32_t ret;
if (!pVnode->inUse) {
terrno = TSDB_CODE_VND_NOT_SYNCED;
vError("vgId:%d, not ready to write since %s", TD_VID(pVnode), terrstr());
return -1;
}
vDebug("vgId:%d, start to process write request %s, index:%" PRId64, TD_VID(pVnode), TMSG_INFO(pMsg->msgType),
version);
@ -286,10 +292,16 @@ int32_t vnodeProcessWriteMsg(SVnode *pVnode, SRpcMsg *pMsg, int64_t version, SRp
_do_commit:
vInfo("vgId:%d, commit at version %" PRId64, TD_VID(pVnode), version);
// commit current change
vnodeCommit(pVnode);
if (vnodeCommit(pVnode) < 0) {
vError("vgId:%d, failed to commit vnode since %s.", TD_VID(pVnode), tstrerror(terrno));
goto _err;
}
// start a new one
vnodeBegin(pVnode);
if (vnodeBegin(pVnode) < 0) {
vError("vgId:%d, failed to begin vnode since %s.", TD_VID(pVnode), tstrerror(terrno));
goto _err;
}
}
return 0;

View File

@ -579,6 +579,7 @@ int32_t syncNodeOnAppendEntriesSnapshot2Cb(SSyncNode* ths, SyncAppendEntriesBatc
SSyncRaftEntry* pAppendEntry = (SSyncRaftEntry*)(pMsg->data + metaTableArr[i].offset);
code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry);
if (code != 0) {
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
return -1;
}
@ -710,6 +711,7 @@ int32_t syncNodeOnAppendEntriesSnapshot2Cb(SSyncNode* ths, SyncAppendEntriesBatc
SSyncRaftEntry* pAppendEntry = (SSyncRaftEntry*)(pMsg->data + metaTableArr[i].offset);
code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry);
if (code != 0) {
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
return -1;
}
@ -859,6 +861,7 @@ int32_t syncNodeOnAppendEntriesSnapshotCb(SSyncNode* ths, SyncAppendEntries* pMs
code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry);
if (code != 0) {
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
return -1;
}
@ -974,6 +977,7 @@ int32_t syncNodeOnAppendEntriesSnapshotCb(SSyncNode* ths, SyncAppendEntries* pMs
code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry);
if (code != 0) {
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
return -1;
}

View File

@ -76,7 +76,10 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) {
pEntry = (SSyncRaftEntry*)taosLRUCacheValue(pCache, h);
} else {
pEntry = pSyncNode->pLogStore->getEntry(pSyncNode->pLogStore, index);
ASSERT(pEntry != NULL);
if (pEntry == NULL) {
sError("failed to get entry since %s. index:%lld", tstrerror(terrno), index);
return;
}
}
// cannot commit, even if quorum agree. need check term!
if (pEntry->term <= pSyncNode->pRaftStore->currentTerm) {
@ -127,7 +130,9 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) {
// execute fsm
if (pSyncNode->pFsm != NULL) {
int32_t code = syncNodeCommit(pSyncNode, beginIndex, endIndex, pSyncNode->state);
ASSERT(code == 0);
if (code != 0) {
wError("failed to commit sync node since %s", tstrerror(terrno));
}
}
}
}

View File

@ -2657,7 +2657,10 @@ static int32_t syncNodeAppendNoop(SSyncNode* ths) {
if (ths->state == TAOS_SYNC_STATE_LEADER) {
int32_t code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pEntry);
ASSERT(code == 0);
if (code != 0) {
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
return -1;
}
syncNodeReplicate(ths, false);
}
@ -2733,7 +2736,7 @@ int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg, SyncI
code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pEntry);
if (code != 0) {
// del resp mgr, call FpCommitCb
ASSERT(0);
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
return -1;
}
@ -2797,8 +2800,8 @@ int32_t syncNodeOnClientRequestBatchCb(SSyncNode* ths, SyncClientRequestBatch* p
code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pEntry);
if (code != 0) {
sError("vgId:%d, failed to append log entry since %s", ths->vgId, tstrerror(terrno));
// del resp mgr, call FpCommitCb
ASSERT(0);
return -1;
}
@ -3050,7 +3053,10 @@ int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex,
pEntry = (SSyncRaftEntry*)taosLRUCacheValue(pCache, h);
} else {
code = ths->pLogStore->syncLogGetEntry(ths->pLogStore, i, &pEntry);
ASSERT(code == 0);
if (code != 0) {
sError("vgId:%d, failed to get log entry since %s. index:%lld", ths->vgId, tstrerror(terrno), i);
return -1;
}
ASSERT(pEntry != NULL);
}

View File

@ -234,8 +234,6 @@ static int32_t raftLogAppendEntry(struct SSyncLogStore* pLogStore, SSyncRaftEntr
snprintf(logBuf, sizeof(logBuf), "wal write error, index:%" PRId64 ", err:%d %X, msg:%s, syserr:%d, sysmsg:%s",
pEntry->index, err, err, errStr, sysErr, sysErrStr);
syncNodeErrorLog(pData->pSyncNode, logBuf);
ASSERT(0);
return -1;
}
pEntry->index = index;
@ -414,13 +412,11 @@ int32_t logStoreAppendEntry(SSyncLogStore* pLogStore, SSyncRaftEntry* pEntry) {
return 0;
}
SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) {
SSyncRaftEntry* logStoreGetEntryWithoutLock(SSyncLogStore* pLogStore, SyncIndex index) {
SSyncLogStoreData* pData = pLogStore->data;
SWal* pWal = pData->pWal;
if (index >= SYNC_INDEX_BEGIN && index <= logStoreLastIndex(pLogStore)) {
taosThreadMutexLock(&(pData->mutex));
// SWalReadHandle* pWalHandle = walOpenReadHandle(pWal);
SWalReader* pWalHandle = pData->pWalHandle;
ASSERT(pWalHandle != NULL);
@ -444,7 +440,8 @@ SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) {
}
} while (0);
ASSERT(0);
sError("failed to read ver since %s. index:%lld", tstrerror(terrno), index);
return NULL;
}
SSyncRaftEntry* pEntry = syncEntryBuild(pWalHandle->pHead->head.bodyLen);
@ -465,7 +462,6 @@ SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) {
terrno = saveErr;
*/
taosThreadMutexUnlock(&(pData->mutex));
return pEntry;
} else {
@ -473,6 +469,16 @@ SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) {
}
}
SSyncRaftEntry* logStoreGetEntry(SSyncLogStore* pLogStore, SyncIndex index) {
SSyncLogStoreData* pData = pLogStore->data;
SSyncRaftEntry *pEntry = NULL;
taosThreadMutexLock(&pData->mutex);
pEntry = logStoreGetEntryWithoutLock(pLogStore, index);
taosThreadMutexUnlock(&pData->mutex);
return pEntry;
}
int32_t logStoreTruncate(SSyncLogStore* pLogStore, SyncIndex fromIndex) {
SSyncLogStoreData* pData = pLogStore->data;
SWal* pWal = pData->pWal;

View File

@ -121,6 +121,7 @@ int tdbBtreeOpen(int keyLen, int valLen, SPager *pPager, char const *tbname, SPg
ret = tdbPagerWrite(pPager, pPage);
if (ret < 0) {
tdbError("failed to write page since %s", terrstr());
return -1;
}
@ -483,9 +484,8 @@ static int tdbBtreeBalanceDeeper(SBTree *pBt, SPage *pRoot, SPage **ppChild, TXN
ret = tdbPagerWrite(pPager, pChild);
if (ret < 0) {
// TODO
ASSERT(0);
return 0;
tdbError("failed to write page since %s", terrstr());
return -1;
}
// Copy the root page content to the child page
@ -556,8 +556,7 @@ static int tdbBtreeBalanceNonRoot(SBTree *pBt, SPage *pParent, int idx, TXN *pTx
ret = tdbPagerWrite(pBt->pPager, pOlds[i]);
if (ret < 0) {
// TODO
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}
}
@ -583,8 +582,7 @@ static int tdbBtreeBalanceNonRoot(SBTree *pBt, SPage *pParent, int idx, TXN *pTx
ret = tdbPagerWrite(pBt->pPager, pParent);
if (ret < 0) {
// TODO
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}
@ -719,8 +717,7 @@ static int tdbBtreeBalanceNonRoot(SBTree *pBt, SPage *pParent, int idx, TXN *pTx
ret = tdbPagerWrite(pBt->pPager, pNews[iNew]);
if (ret < 0) {
// TODO
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}
}
@ -937,7 +934,7 @@ static int tdbFetchOvflPage(SPgno *pPgno, SPage **ppOfp, TXN *pTxn, SBTree *pBt)
// mark dirty
ret = tdbPagerWrite(pBt->pPager, *ppOfp);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}
@ -1942,7 +1939,7 @@ int tdbBtcDelete(SBTC *pBtc) {
// drop the cell on the leaf
ret = tdbPagerWrite(pPager, pBtc->pPage);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}
@ -1964,7 +1961,7 @@ int tdbBtcDelete(SBTC *pBtc) {
if (idx < nCells) {
ret = tdbPagerWrite(pPager, pPage);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}
@ -2029,7 +2026,7 @@ int tdbBtcUpsert(SBTC *pBtc, const void *pKey, int kLen, const void *pData, int
// mark dirty
ret = tdbPagerWrite(pBtc->pBt->pPager, pBtc->pPage);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page since %s", terrstr());
return -1;
}

View File

@ -106,7 +106,7 @@ int32_t tdbBegin(TDB *pDb, TXN *pTxn) {
for (pPager = pDb->pgrList; pPager; pPager = pPager->pNext) {
ret = tdbPagerBegin(pPager, pTxn);
if (ret < 0) {
ASSERT(0);
tdbError("failed to begin pager since %s. dbName:%s, txnId:%d", tstrerror(terrno), pDb->dbName, pTxn->txnId);
return -1;
}
}
@ -121,7 +121,7 @@ int32_t tdbCommit(TDB *pDb, TXN *pTxn) {
for (pPager = pDb->pgrList; pPager; pPager = pPager->pNext) {
ret = tdbPagerCommit(pPager, pTxn);
if (ret < 0) {
ASSERT(0);
tdbError("failed to commit pager since %s. dbName:%s, txnId:%d", tstrerror(terrno), pDb->dbName, pTxn->txnId);
return -1;
}
}
@ -136,7 +136,7 @@ int32_t tdbAbort(TDB *pDb, TXN *pTxn) {
for (pPager = pDb->pgrList; pPager; pPager = pPager->pNext) {
ret = tdbPagerAbort(pPager, pTxn);
if (ret < 0) {
ASSERT(0);
tdbError("failed to abort pager since %s. dbName:%s, txnId:%d", tstrerror(terrno), pDb->dbName, pTxn->txnId);
return -1;
}
}

View File

@ -156,6 +156,7 @@ int tdbPagerOpenDB(SPager *pPager, SPgno *ppgno, bool toCreate, SBTree *pBt) {
ret = tdbPagerWrite(pPager, pPage);
if (ret < 0) {
tdbError("failed to write page since %s", terrstr());
return -1;
}
@ -210,7 +211,7 @@ int tdbPagerWrite(SPager *pPager, SPage *pPage) {
if (TDB_PAGE_PGNO(pPage) <= pPager->dbOrigSize) {
ret = tdbPagerWritePageToJournal(pPager, pPage);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page to journal since %s", tstrerror(terrno));
return -1;
}
}
@ -226,6 +227,8 @@ int tdbPagerBegin(SPager *pPager, TXN *pTxn) {
// Open the journal
pPager->jfd = tdbOsOpen(pPager->jFileName, TDB_O_CREAT | TDB_O_RDWR, 0755);
if (pPager->jfd < 0) {
tdbError("failed to open file due to %s. jFileName:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
@ -243,9 +246,9 @@ int tdbPagerCommit(SPager *pPager, TXN *pTxn) {
// sync the journal file
ret = tdbOsFSync(pPager->jfd);
if (ret < 0) {
// TODO
ASSERT(0);
return 0;
tdbError("failed to fsync jfd due to %s. jFileName:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
// loop to write the dirty pages to file
@ -255,7 +258,7 @@ int tdbPagerCommit(SPager *pPager, TXN *pTxn) {
pPage = (SPage *)pNode;
ret = tdbPagerWritePageToDB(pPager, pPage);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page to db since %s", tstrerror(terrno));
return -1;
}
}
@ -277,11 +280,25 @@ int tdbPagerCommit(SPager *pPager, TXN *pTxn) {
tRBTreeCreate(&pPager->rbt, pageCmpFn);
// sync the db file
tdbOsFSync(pPager->fd);
if (tdbOsFSync(pPager->fd) < 0) {
tdbError("failed to fsync fd due to %s. file:%s", strerror(errno), pPager->dbFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
// remove the journal file
tdbOsClose(pPager->jfd);
tdbOsRemove(pPager->jFileName);
if (tdbOsClose(pPager->jfd) < 0) {
tdbError("failed to close jfd due to %s. file:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
if (tdbOsRemove(pPager->jFileName) < 0 && errno != ENOENT) {
tdbError("failed to remove file due to %s. file:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
pPager->inTran = 0;
return 0;
@ -297,14 +314,14 @@ int tdbPagerAbort(SPager *pPager, TXN *pTxn) {
// 0, sync the journal file
ret = tdbOsFSync(pPager->jfd);
if (ret < 0) {
// TODO
ASSERT(0);
return 0;
tdbError("failed to fsync jfd due to %s. file:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
tdb_fd_t jfd = tdbOsOpen(pPager->jFileName, TDB_O_RDWR, 0755);
if (jfd == NULL) {
return 0;
return -1;
}
ret = tdbGetFileSize(jfd, pPager->pageSize, &journalSize);
@ -516,11 +533,15 @@ static int tdbPagerWritePageToJournal(SPager *pPager, SPage *pPage) {
ret = tdbOsWrite(pPager->jfd, &pgno, sizeof(pgno));
if (ret < 0) {
tdbError("failed to write pgno due to %s. file:%s, pgno:%u", strerror(errno), pPager->jFileName, pgno);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
ret = tdbOsWrite(pPager->jfd, pPage->pData, pPage->pageSize);
if (ret < 0) {
tdbError("failed to write page data due to %s. file:%s, pageSize:%ld", strerror(errno), pPager->jFileName, pPage->pageSize);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
@ -540,13 +561,15 @@ static int tdbPagerWritePageToDB(SPager *pPager, SPage *pPage) {
offset = (i64)pPage->pageSize * (TDB_PAGE_PGNO(pPage) - 1);
if (tdbOsLSeek(pPager->fd, offset, SEEK_SET) < 0) {
ASSERT(0);
tdbError("failed to lseek due to %s. file:%s, offset:%ld", strerror(errno), pPager->dbFileName, offset);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
ret = tdbOsWrite(pPager->fd, pPage->pData, pPage->pageSize);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write page data due to %s. file:%s, pageSize:%ld", strerror(errno), pPager->dbFileName, pPage->pageSize);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
@ -590,23 +613,38 @@ int tdbPagerRestore(SPager *pPager, SBTree *pBt) {
i64 offset = pPager->pageSize * (pgno - 1);
if (tdbOsLSeek(pPager->fd, offset, SEEK_SET) < 0) {
ASSERT(0);
tdbError("failed to lseek fd due to %s. file:%s, offset:%ld", strerror(errno), pPager->dbFileName, offset);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
ret = tdbOsWrite(pPager->fd, pageBuf, pPager->pageSize);
if (ret < 0) {
ASSERT(0);
tdbError("failed to write buf due to %s. file: %s, bufsize:%d", strerror(errno), pPager->dbFileName, pPager->pageSize);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
}
tdbOsFSync(pPager->fd);
if (tdbOsFSync(pPager->fd) < 0) {
tdbError("failed to fsync fd due to %s. dbfile:%s", strerror(errno), pPager->dbFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
tdbOsFree(pageBuf);
tdbOsClose(jfd);
tdbOsRemove(pPager->jFileName);
if (tdbOsClose(jfd) < 0) {
tdbError("failed to close jfd due to %s. jFileName:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
if (tdbOsRemove(pPager->jFileName) < 0 && errno != ENOENT) {
tdbError("failed to remove file due to %s. jFileName:%s", strerror(errno), pPager->jFileName);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
return 0;
}

View File

@ -35,113 +35,251 @@ int64_t FORCE_INLINE walGetCommittedVer(SWal* pWal) { return pWal->vers.commitVe
int64_t FORCE_INLINE walGetAppliedVer(SWal* pWal) { return pWal->vers.appliedVer; }
static FORCE_INLINE void walBuildMetaName(SWal* pWal, int metaVer, char* buf) {
sprintf(buf, "%s/meta-ver%d", pWal->path, metaVer);
static FORCE_INLINE int walBuildMetaName(SWal* pWal, int metaVer, char* buf) {
return sprintf(buf, "%s/meta-ver%d", pWal->path, metaVer);
}
static FORCE_INLINE int64_t walScanLogGetLastVer(SWal* pWal) {
int32_t sz = taosArrayGetSize(pWal->fileInfoSet);
ASSERT(sz > 0);
#if 0
for (int i = 0; i < sz; i++) {
SWalFileInfo* pFileInfo = taosArrayGet(pWal->fileInfoSet, i);
}
#endif
static FORCE_INLINE int walBuildTmpMetaName(SWal* pWal, char* buf) {
return sprintf(buf, "%s/meta-ver.tmp", pWal->path);
}
SWalFileInfo* pLastFileInfo = taosArrayGet(pWal->fileInfoSet, sz - 1);
static FORCE_INLINE int64_t walScanLogGetLastVer(SWal* pWal, int32_t fileIdx) {
int32_t sz = taosArrayGetSize(pWal->fileInfoSet);
terrno = TSDB_CODE_SUCCESS;
ASSERT(fileIdx >= 0 && fileIdx < sz);
SWalFileInfo* pFileInfo = taosArrayGet(pWal->fileInfoSet, fileIdx);
char fnameStr[WAL_FILE_LEN];
walBuildLogName(pWal, pLastFileInfo->firstVer, fnameStr);
walBuildLogName(pWal, pFileInfo->firstVer, fnameStr);
int64_t fileSize = 0;
taosStatFile(fnameStr, &fileSize, NULL);
int32_t readSize = TMIN(WAL_SCAN_BUF_SIZE, fileSize);
pLastFileInfo->fileSize = fileSize;
TdFilePtr pFile = taosOpenFile(fnameStr, TD_FILE_READ);
TdFilePtr pFile = taosOpenFile(fnameStr, TD_FILE_READ | TD_FILE_WRITE);
if (pFile == NULL) {
wError("vgId:%d, failed to open file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
// ensure size as non-negative
pFileInfo->fileSize = TMAX(0, pFileInfo->fileSize);
uint64_t magic = WAL_MAGIC;
int64_t walCkHeadSz = sizeof(SWalCkHead);
int64_t end = fileSize;
int64_t offset = 0;
int32_t capacity = 0;
int32_t readSize = 0;
char* buf = NULL;
char* found = NULL;
bool firstTrial = pFileInfo->fileSize < fileSize;
char* buf = taosMemoryMalloc(readSize + 5);
if (buf == NULL) {
taosCloseFile(&pFile);
terrno = TSDB_CODE_WAL_OUT_OF_MEMORY;
return -1;
}
int64_t offset;
offset = taosLSeekFile(pFile, -readSize, SEEK_END);
if (readSize != taosReadFile(pFile, buf, readSize)) {
taosMemoryFree(buf);
taosCloseFile(&pFile);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
char* found = NULL;
// search for the valid last WAL entry, e.g. block by block
while (1) {
offset = (firstTrial) ? pFileInfo->fileSize : TMAX(0, end - WAL_SCAN_BUF_SIZE);
ASSERT(offset <= end);
readSize = end - offset;
capacity = readSize + sizeof(magic);
int64_t limit = WAL_RECOV_SIZE_LIMIT;
if (limit < readSize) {
wError("vgId:%d, possibly corrupted WAL range exceeds size limit (i.e. %" PRId64 " bytes). offset:%" PRId64
", end:%" PRId64 ", file:%s",
pWal->cfg.vgId, limit, offset, end, fnameStr);
terrno = TSDB_CODE_WAL_SIZE_LIMIT;
goto _err;
}
void* ptr = taosMemoryRealloc(buf, capacity);
if (ptr == NULL) {
terrno = TSDB_CODE_WAL_OUT_OF_MEMORY;
goto _err;
}
buf = ptr;
int64_t ret = taosLSeekFile(pFile, offset, SEEK_SET);
if (ret < 0) {
wError("vgId:%d, failed to lseek file due to %s. offset:%" PRId64 "", pWal->cfg.vgId, strerror(errno), offset);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
if (readSize != taosReadFile(pFile, buf, readSize)) {
wError("vgId:%d, failed to read file due to %s. readSize:%" PRId64 ", file:%s", pWal->cfg.vgId, strerror(errno),
readSize, fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
char* candidate = NULL;
char* haystack = buf;
char* candidate;
while ((candidate = tmemmem(haystack, readSize - (haystack - buf), (char*)&magic, sizeof(uint64_t))) != NULL) {
// read and validate
SWalCkHead* logContent = (SWalCkHead*)candidate;
if (walValidHeadCksum(logContent) == 0 && walValidBodyCksum(logContent) == 0) {
found = candidate;
while ((candidate = tmemmem(haystack, readSize - (haystack - buf), (char*)&magic, sizeof(magic))) != NULL) {
// validate head
int64_t len = readSize - (candidate - buf);
if (len < walCkHeadSz) {
break;
}
SWalCkHead* logContent = (SWalCkHead*)candidate;
if (walValidHeadCksum(logContent) != 0) {
wError("vgId:%d, failed to validate checksum of wal entry header. offset:% %" PRId64 ", file:%s",
((char*)(logContent)-buf), fnameStr);
haystack = candidate + 1;
if (firstTrial) {
break;
} else {
continue;
}
}
// validate body
int64_t size = walCkHeadSz + logContent->head.bodyLen;
if (len < size) {
int64_t extraSize = size - len;
if (capacity < readSize + extraSize + sizeof(magic)) {
capacity += extraSize;
void* ptr = taosMemoryRealloc(buf, capacity);
if (ptr == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
goto _err;
}
buf = ptr;
}
int64_t ret = taosLSeekFile(pFile, offset + readSize, SEEK_SET);
if (ret < 0) {
wError("vgId:%d, failed to lseek file due to %s. offset:%" PRId64 "", pWal->cfg.vgId, strerror(errno),
offset);
terrno = TAOS_SYSTEM_ERROR(errno);
break;
}
if (extraSize != taosReadFile(pFile, buf + readSize, extraSize)) {
wError("vgId:%d, failed to read file due to %s. offset:%" PRId64 ", extraSize:%" PRId64 ", file:%s",
pWal->cfg.vgId, strerror(errno), offset + readSize, extraSize, fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
break;
}
}
if (walValidBodyCksum(logContent) != 0) {
terrno = TSDB_CODE_WAL_CHKSUM_MISMATCH;
wError("vgId:%d, failed to validate checksum of wal entry body. offset:% %" PRId64 ", file:%s",
((char*)(logContent)-buf), fnameStr);
haystack = candidate + 1;
if (firstTrial) {
break;
} else {
continue;
}
}
// found one
found = candidate;
haystack = candidate + 1;
}
if (found || offset == 0) break;
offset = TMIN(0, offset - readSize + sizeof(uint64_t));
int64_t offset2 = taosLSeekFile(pFile, offset, SEEK_SET);
ASSERT(offset == offset2);
if (readSize != taosReadFile(pFile, buf, readSize)) {
taosMemoryFree(buf);
taosCloseFile(&pFile);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
#if 0
if (found == buf) {
SWalCkHead* logContent = (SWalCkHead*)found;
if (walValidHeadCksum(logContent) != 0 || walValidBodyCksum(logContent) != 0) {
// file has to be deleted
taosMemoryFree(buf);
taosCloseFile(&pFile);
terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
return -1;
}
}
#endif
// go backwards, e.g. by at most one WAL scan buf size
end = offset + walCkHeadSz - 1;
firstTrial = false;
}
if (found == NULL) {
// file corrupted, no complete log
// TODO delete and search in previous files
/*ASSERT(0);*/
terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
return -1;
// determine end of last entry
SWalCkHead* lastEntry = (SWalCkHead*)found;
int64_t retVer = -1;
int64_t lastEntryBeginOffset = 0;
int64_t lastEntryEndOffset = 0;
if (lastEntry == NULL) {
terrno = TSDB_CODE_WAL_LOG_NOT_EXIST;
} else {
retVer = lastEntry->head.version;
lastEntryBeginOffset = offset + (int64_t)((char*)lastEntry - (char*)buf);
lastEntryEndOffset = lastEntryBeginOffset + sizeof(SWalCkHead) + lastEntry->head.bodyLen;
}
// truncate file
SWalCkHead* lastEntry = (SWalCkHead*)found;
int64_t retVer = lastEntry->head.version;
int64_t lastEntryBeginOffset = offset + (int64_t)((char*)found - (char*)buf);
int64_t lastEntryEndOffset = lastEntryBeginOffset + sizeof(SWalCkHead) + lastEntry->head.bodyLen;
if (lastEntryEndOffset != fileSize) {
wWarn("vgId:%d repair meta truncate file %s to %ld, orig size %ld", pWal->cfg.vgId, fnameStr, lastEntryEndOffset,
wWarn("vgId:%d, repair meta truncate file %s to %ld, orig size %ld", pWal->cfg.vgId, fnameStr, lastEntryEndOffset,
fileSize);
taosFtruncateFile(pFile, lastEntryEndOffset);
((SWalFileInfo*)taosArrayGetLast(pWal->fileInfoSet))->fileSize = lastEntryEndOffset;
pWal->totSize -= (fileSize - lastEntryEndOffset);
if (taosFtruncateFile(pFile, lastEntryEndOffset) < 0) {
wError("failed to truncate file due to %s. file:%s", strerror(errno), fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
if (taosFsyncFile(pFile) < 0) {
wError("failed to fsync file due to %s. file:%s", strerror(errno), fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
}
pFileInfo->fileSize = lastEntryEndOffset;
taosCloseFile(&pFile);
taosMemoryFree(buf);
return retVer;
_err:
taosCloseFile(&pFile);
taosMemoryFree(buf);
return -1;
}
static void walRebuildFileInfoSet(SArray* metaLogList, SArray* actualLogList) {
int metaFileNum = taosArrayGetSize(metaLogList);
int actualFileNum = taosArrayGetSize(actualLogList);
int j = 0;
// both of the lists in asc order
for (int i = 0; i < actualFileNum; i++) {
SWalFileInfo* pLogInfo = taosArrayGet(actualLogList, i);
while (j < metaFileNum) {
SWalFileInfo* pMetaInfo = taosArrayGet(metaLogList, j);
ASSERT(pMetaInfo != NULL);
if (pMetaInfo->firstVer < pLogInfo->firstVer) {
j++;
} else if (pMetaInfo->firstVer == pLogInfo->firstVer) {
(*pLogInfo) = *pMetaInfo;
j++;
break;
} else {
break;
}
}
}
taosArrayClear(metaLogList);
for (int i = 0; i < actualFileNum; i++) {
SWalFileInfo* pFileInfo = taosArrayGet(actualLogList, i);
taosArrayPush(metaLogList, pFileInfo);
}
}
void walAlignVersions(SWal* pWal) {
if (pWal->vers.firstVer > pWal->vers.snapshotVer + 1) {
wWarn("vgId:%d, firstVer:%" PRId64 " is larger than snapshotVer:%" PRId64 " + 1. align with it.", pWal->cfg.vgId,
pWal->vers.firstVer, pWal->vers.snapshotVer);
pWal->vers.firstVer = pWal->vers.snapshotVer + 1;
}
if (pWal->vers.lastVer < pWal->vers.snapshotVer) {
wWarn("vgId:%d, lastVer:%" PRId64 " is less than snapshotVer:%" PRId64 ". align with it.", pWal->cfg.vgId,
pWal->vers.lastVer, pWal->vers.snapshotVer);
pWal->vers.lastVer = pWal->vers.snapshotVer;
}
if (pWal->vers.commitVer < pWal->vers.snapshotVer) {
wWarn("vgId:%d, commitVer:%" PRId64 " is less than snapshotVer:%" PRId64 ". align with it.", pWal->cfg.vgId,
pWal->vers.commitVer, pWal->vers.snapshotVer);
pWal->vers.commitVer = pWal->vers.snapshotVer;
}
if (pWal->vers.appliedVer < pWal->vers.snapshotVer) {
wWarn("vgId:%d, appliedVer:%" PRId64 " is less than snapshotVer:%" PRId64 ". align with it.", pWal->cfg.vgId,
pWal->vers.appliedVer, pWal->vers.snapshotVer);
pWal->vers.appliedVer = pWal->vers.snapshotVer;
}
pWal->vers.commitVer = TMIN(pWal->vers.lastVer, pWal->vers.commitVer);
pWal->vers.appliedVer = TMIN(pWal->vers.commitVer, pWal->vers.appliedVer);
}
int walCheckAndRepairMeta(SWal* pWal) {
@ -150,7 +288,6 @@ int walCheckAndRepairMeta(SWal* pWal) {
const char* idxPattern = "^[0-9]+.idx$";
regex_t logRegPattern;
regex_t idxRegPattern;
bool fixed = false;
regcomp(&logRegPattern, logPattern, REG_EXTENDED);
regcomp(&idxRegPattern, idxPattern, REG_EXTENDED);
@ -184,225 +321,237 @@ int walCheckAndRepairMeta(SWal* pWal) {
taosArraySort(actualLog, compareWalFileInfo);
int metaFileNum = taosArrayGetSize(pWal->fileInfoSet);
int actualFileNum = taosArrayGetSize(actualLog);
int metaFileNum = taosArrayGetSize(pWal->fileInfoSet);
int actualFileNum = taosArrayGetSize(actualLog);
int64_t firstVerPrev = pWal->vers.firstVer;
int64_t lastVerPrev = pWal->vers.lastVer;
int64_t totSize = 0;
bool updateMeta = (metaFileNum != actualFileNum);
#if 0
for (int32_t fileNo = actualFileNum - 1; fileNo >= 0; fileNo--) {
SWalFileInfo* pFileInfo = taosArrayGet(pLogInfoArray, fileNo);
// rebuild meta of file info
walRebuildFileInfoSet(pWal->fileInfoSet, actualLog);
taosArrayDestroy(actualLog);
int32_t sz = taosArrayGetSize(pWal->fileInfoSet);
ASSERT(sz == actualFileNum);
// scan and determine the lastVer
int32_t fileIdx = sz;
while (--fileIdx >= 0) {
char fnameStr[WAL_FILE_LEN];
int64_t fileSize = 0;
SWalFileInfo* pFileInfo = taosArrayGet(pWal->fileInfoSet, fileIdx);
walBuildLogName(pWal, pFileInfo->firstVer, fnameStr);
int64_t fileSize = 0;
taosStatFile(fnameStr, &fileSize, NULL);
if (fileSize == 0) {
int32_t code = taosStatFile(fnameStr, &fileSize, NULL);
if (code < 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
wError("failed to stat file since %s. file:%s", terrstr(), fnameStr);
return -1;
}
ASSERT(pFileInfo->firstVer >= 0);
if (pFileInfo->lastVer >= pFileInfo->firstVer && fileSize == pFileInfo->fileSize) {
totSize += pFileInfo->fileSize;
continue;
}
updateMeta = true;
int64_t lastVer = walScanLogGetLastVer(pWal, fileIdx);
if (lastVer < 0) {
if (terrno != TSDB_CODE_WAL_LOG_NOT_EXIST) {
wError("failed to scan wal last ver since %s", terrstr());
return -1;
}
ASSERT(pFileInfo->fileSize == 0);
// remove the empty wal log, and its idx
taosRemoveFile(fnameStr);
walBuildIdxName(pWal, pFileInfo->firstVer, fnameStr);
taosRemoveFile(fnameStr);
taosArrayPop(pLogInfoArray);
} else {
break;
// remove its meta entry
taosArrayRemove(pWal->fileInfoSet, fileIdx);
continue;
}
// update lastVer
pFileInfo->lastVer = lastVer;
totSize += pFileInfo->fileSize;
}
actualFileNum = taosArrayGetSize(pLogInfoArray);
#endif
{
int32_t i = 0, j = 0;
while (i < actualFileNum && j < metaFileNum) {
SWalFileInfo* pActualFile = taosArrayGet(actualLog, i);
SWalFileInfo* pMetaFile = taosArrayGet(pWal->fileInfoSet, j);
if (pActualFile->firstVer < pMetaFile->firstVer) {
char fNameStr[WAL_FILE_LEN];
walBuildLogName(pWal, pActualFile->firstVer, fNameStr);
taosRemoveFile(fNameStr);
walBuildIdxName(pWal, pActualFile->firstVer, fNameStr);
taosRemoveFile(fNameStr);
i++;
} else if (pActualFile->firstVer > pMetaFile->firstVer) {
taosArrayRemove(pWal->fileInfoSet, j);
metaFileNum--;
} else {
i++;
j++;
}
}
if (i == actualFileNum && j == metaFileNum) {
if (j > 0) {
SWalFileInfo* pLastInfo = taosArrayGet(pWal->fileInfoSet, j - 1);
int64_t fsize = 0;
char fNameStr[WAL_FILE_LEN];
walBuildLogName(pWal, pLastInfo->firstVer, fNameStr);
taosStatFile(fNameStr, &fsize, NULL);
if (pLastInfo->fileSize != fsize) {
fixed = true;
pLastInfo->fileSize = fsize;
pLastInfo->lastVer = walScanLogGetLastVer(pWal);
}
}
} else {
fixed = true;
while (i < actualFileNum) {
SWalFileInfo* pActualFile = taosArrayGet(actualLog, i);
char fNameStr[WAL_FILE_LEN];
walBuildLogName(pWal, pActualFile->firstVer, fNameStr);
taosStatFile(fNameStr, &pActualFile->fileSize, NULL);
if (pActualFile->fileSize == 0) {
ASSERT(i == actualFileNum - 1);
taosRemoveFile(fNameStr);
walBuildIdxName(pWal, pActualFile->firstVer, fNameStr);
taosRemoveFile(fNameStr);
break;
}
if (i < actualFileNum - 1) {
pActualFile->lastVer = ((SWalFileInfo*)taosArrayGet(actualLog, i + 1))->firstVer - 1;
taosArrayPush(pWal->fileInfoSet, pActualFile);
i++;
} else {
pActualFile = taosArrayPush(pWal->fileInfoSet, pActualFile);
pActualFile->lastVer = walScanLogGetLastVer(pWal);
if (pActualFile->lastVer == -1) {
taosRemoveFile(fNameStr);
walBuildIdxName(pWal, pActualFile->firstVer, fNameStr);
taosRemoveFile(fNameStr);
taosArrayPop(pWal->fileInfoSet);
}
break;
}
}
}
}
#if 0
if (metaFileNum > actualFileNum) {
taosArrayPopFrontBatch(pWal->fileInfoSet, metaFileNum - actualFileNum);
} else if (metaFileNum < actualFileNum) {
for (int i = metaFileNum; i < actualFileNum; i++) {
SWalFileInfo* pFileInfo = taosArrayGet(actualLog, i);
taosArrayPush(pWal->fileInfoSet, pFileInfo);
}
}
#endif
taosArrayDestroy(actualLog);
// reset vers info and so on
actualFileNum = taosArrayGetSize(pWal->fileInfoSet);
pWal->writeCur = actualFileNum - 1;
pWal->totSize = totSize;
pWal->vers.lastVer = -1;
if (actualFileNum > 0) {
int64_t fLastVer = ((SWalFileInfo*)taosArrayGet(pWal->fileInfoSet, pWal->writeCur))->lastVer;
if (fLastVer != -1 && pWal->vers.lastVer != fLastVer) {
fixed = true;
pWal->vers.lastVer = fLastVer;
}
int64_t fFirstVer = ((SWalFileInfo*)taosArrayGet(pWal->fileInfoSet, 0))->firstVer;
if (fFirstVer != pWal->vers.firstVer) {
fixed = true;
pWal->vers.firstVer = fFirstVer;
}
pWal->vers.firstVer = ((SWalFileInfo*)taosArrayGet(pWal->fileInfoSet, 0))->firstVer;
pWal->vers.lastVer = ((SWalFileInfo*)taosArrayGetLast(pWal->fileInfoSet))->lastVer;
}
(void)walAlignVersions(pWal);
// update meta file
if (updateMeta) {
(void)walSaveMeta(pWal);
}
return 0;
}
int walReadLogHead(TdFilePtr pLogFile, int64_t offset, SWalCkHead* pCkHead) {
if (taosLSeekFile(pLogFile, offset, SEEK_SET) < 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
if (fixed) {
walSaveMeta(pWal);
if (taosReadFile(pLogFile, pCkHead, sizeof(SWalCkHead)) != sizeof(SWalCkHead)) {
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
if (walValidHeadCksum(pCkHead) != 0) {
terrno = TSDB_CODE_WAL_CHKSUM_MISMATCH;
return -1;
}
return 0;
}
int walCheckAndRepairIdx(SWal* pWal) {
int walCheckAndRepairIdxFile(SWal* pWal, int32_t fileIdx) {
int32_t sz = taosArrayGetSize(pWal->fileInfoSet);
for (int32_t i = 0; i < sz; i++) {
SWalFileInfo* pFileInfo = taosArrayGet(pWal->fileInfoSet, i);
ASSERT(fileIdx >= 0 && fileIdx < sz);
SWalFileInfo* pFileInfo = taosArrayGet(pWal->fileInfoSet, fileIdx);
char fnameStr[WAL_FILE_LEN];
walBuildIdxName(pWal, pFileInfo->firstVer, fnameStr);
char fLogNameStr[WAL_FILE_LEN];
walBuildLogName(pWal, pFileInfo->firstVer, fLogNameStr);
int64_t fileSize = 0;
char fnameStr[WAL_FILE_LEN];
walBuildIdxName(pWal, pFileInfo->firstVer, fnameStr);
int64_t fsize;
TdFilePtr pIdxFile = taosOpenFile(fnameStr, TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE);
if (pIdxFile == NULL) {
ASSERT(0);
if (taosStatFile(fnameStr, &fileSize, NULL) < 0 && errno != ENOENT) {
wError("vgId:%d, failed to stat file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
ASSERT(pFileInfo->fileSize > 0 && pFileInfo->firstVer >= 0 && pFileInfo->lastVer >= pFileInfo->firstVer);
if (fileSize == (pFileInfo->lastVer - pFileInfo->firstVer + 1) * sizeof(SWalIdxEntry)) {
return 0;
}
// start to repair
int64_t offset = fileSize - fileSize % sizeof(SWalIdxEntry);
TdFilePtr pLogFile = NULL;
TdFilePtr pIdxFile = NULL;
SWalIdxEntry idxEntry = {.ver = pFileInfo->firstVer - 1, .offset = -sizeof(SWalCkHead)};
SWalCkHead ckHead;
memset(&ckHead, 0, sizeof(ckHead));
ckHead.head.version = idxEntry.ver;
pIdxFile = taosOpenFile(fnameStr, TD_FILE_READ | TD_FILE_WRITE | TD_FILE_CREATE);
if (pIdxFile == NULL) {
wError("vgId:%d, failed to open file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
pLogFile = taosOpenFile(fLogNameStr, TD_FILE_READ);
if (pLogFile == NULL) {
terrno = TAOS_SYSTEM_ERROR(errno);
wError("vgId:%d, cannot open file %s, since %s", pWal->cfg.vgId, fLogNameStr, terrstr());
goto _err;
}
// determine the last valid entry end, i.e. offset
while ((offset -= sizeof(SWalIdxEntry)) >= 0) {
if (taosLSeekFile(pIdxFile, offset, SEEK_SET) < 0) {
wError("vgId:%d, failed to seek file due to %s. offset:" PRId64 ", file:%s", pWal->cfg.vgId, strerror(errno),
offset, fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
wError("vgId:%d, cannot open file %s, since %s", pWal->cfg.vgId, fnameStr, terrstr());
return -1;
goto _err;
}
taosFStatFile(pIdxFile, &fsize, NULL);
if (fsize == (pFileInfo->lastVer - pFileInfo->firstVer + 1) * sizeof(SWalIdxEntry)) {
taosCloseFile(&pIdxFile);
if (taosReadFile(pIdxFile, &idxEntry, sizeof(SWalIdxEntry)) != sizeof(SWalIdxEntry)) {
wError("vgId:%d, failed to read file due to %s. offset:%" PRId64 ", file:%s", pWal->cfg.vgId, strerror(errno),
offset, fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
if (idxEntry.ver > pFileInfo->lastVer) {
continue;
}
int32_t left = fsize % sizeof(SWalIdxEntry);
int64_t offset = taosLSeekFile(pIdxFile, -left, SEEK_END);
if (left != 0) {
taosFtruncateFile(pIdxFile, offset);
wWarn("vgId:%d wal truncate file %s to offset %ld since size invalid, file size %ld", pWal->cfg.vgId, fnameStr,
offset, fsize);
}
offset -= sizeof(SWalIdxEntry);
SWalIdxEntry idxEntry = {.ver = pFileInfo->firstVer};
while (1) {
if (offset < 0) {
taosLSeekFile(pIdxFile, 0, SEEK_SET);
taosWriteFile(pIdxFile, &idxEntry, sizeof(SWalIdxEntry));
break;
}
if (taosLSeekFile(pIdxFile, offset, SEEK_SET) < 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
wError("vgId:%d cannot seek offset %ld when repair idx since %s", pWal->cfg.vgId, offset, terrstr());
}
int64_t contLen = taosReadFile(pIdxFile, &idxEntry, sizeof(SWalIdxEntry));
if (contLen < 0 || contLen != sizeof(SWalIdxEntry)) {
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
if ((idxEntry.ver - pFileInfo->firstVer) * sizeof(SWalIdxEntry) != offset) {
taosFtruncateFile(pIdxFile, offset);
wWarn("vgId:%d wal truncate file %s to offset %ld since entry invalid, entry ver %ld, entry offset %ld",
pWal->cfg.vgId, fnameStr, offset, idxEntry.ver, idxEntry.offset);
offset -= sizeof(SWalIdxEntry);
} else {
break;
}
if (walReadLogHead(pLogFile, idxEntry.offset, &ckHead) < 0) {
wWarn("vgId:%d, failed to read log file since %s. file:%s, offset:%" PRId64 ", idx entry ver:%" PRId64 "",
pWal->cfg.vgId, terrstr(), fLogNameStr, idxEntry.offset, idxEntry.ver);
continue;
}
if (idxEntry.ver < pFileInfo->lastVer) {
char fLogNameStr[WAL_FILE_LEN];
walBuildLogName(pWal, pFileInfo->firstVer, fLogNameStr);
TdFilePtr pLogFile = taosOpenFile(fLogNameStr, TD_FILE_READ);
if (pLogFile == NULL) {
terrno = TAOS_SYSTEM_ERROR(errno);
wError("vgId:%d, cannot open file %s, since %s", pWal->cfg.vgId, fLogNameStr, terrstr());
return -1;
}
while (idxEntry.ver < pFileInfo->lastVer) {
if (taosLSeekFile(pLogFile, idxEntry.offset, SEEK_SET) == -1) {
terrno = TAOS_SYSTEM_ERROR(errno);
wError("vgId:%d, cannot seek file %s at %ld, since %s", pWal->cfg.vgId, fLogNameStr, idxEntry.offset,
terrstr());
return -1;
}
SWalCkHead ckHead;
taosReadFile(pLogFile, &ckHead, sizeof(SWalCkHead));
if (idxEntry.ver != ckHead.head.version) {
// todo truncate this idx also
taosCloseFile(&pLogFile);
wError("vgId:%d, invalid repair case, log seek to %ld to find ver %ld, actual ver %ld", pWal->cfg.vgId,
idxEntry.offset, idxEntry.ver, ckHead.head.version);
return -1;
}
idxEntry.ver = ckHead.head.version + 1;
idxEntry.offset = idxEntry.offset + sizeof(SWalCkHead) + ckHead.head.bodyLen;
wWarn("vgId:%d wal idx append new entry %ld %ld", pWal->cfg.vgId, idxEntry.ver, idxEntry.offset);
taosWriteFile(pIdxFile, &idxEntry, sizeof(SWalIdxEntry));
}
taosCloseFile(&pLogFile);
if (idxEntry.ver == ckHead.head.version) {
break;
}
}
offset += sizeof(SWalIdxEntry);
// ftruncate idx file
if (offset < fileSize) {
if (taosFtruncateFile(pIdxFile, offset) < 0) {
wError("vgId:%d, failed to ftruncate file due to %s. offset:%" PRId64 ", file:%s", pWal->cfg.vgId,
strerror(errno), offset, fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
}
// rebuild idx file
if (taosLSeekFile(pIdxFile, 0, SEEK_END) < 0) {
wError("vgId:%d, failed to seek file due to %s. offset:" PRId64 ", file:%s", pWal->cfg.vgId, strerror(errno),
offset, fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
while (idxEntry.ver < pFileInfo->lastVer) {
ASSERT(idxEntry.ver == ckHead.head.version);
idxEntry.ver += 1;
idxEntry.offset += sizeof(SWalCkHead) + ckHead.head.bodyLen;
if (walReadLogHead(pLogFile, idxEntry.offset, &ckHead) < 0) {
wError("vgId:%d, failed to read wal log head since %s. offset:%" PRId64 ", file:%s", pWal->cfg.vgId, terrstr(),
idxEntry.offset, fLogNameStr);
goto _err;
}
wWarn("vgId:%d wal idx append new entry %ld %ld", 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;
}
}
if (taosFsyncFile(pIdxFile) < 0) {
wError("vgId:%d, faild to fsync file since %s. file:%s", pWal->cfg.vgId, terrstr(), fnameStr);
goto _err;
}
(void)taosCloseFile(&pLogFile);
(void)taosCloseFile(&pIdxFile);
return 0;
_err:
(void)taosCloseFile(&pLogFile);
(void)taosCloseFile(&pIdxFile);
return -1;
}
int walCheckAndRepairIdx(SWal* pWal) {
int32_t sz = taosArrayGetSize(pWal->fileInfoSet);
int32_t fileIdx = sz;
while (--fileIdx >= 0) {
if (walCheckAndRepairIdxFile(pWal, fileIdx) < 0) {
wError("vgId:%d, failed to repair idx file since %s. fileIdx:%d", pWal->cfg.vgId, terrstr(), fileIdx);
return -1;
}
taosCloseFile(&pIdxFile);
}
return 0;
}
@ -495,14 +644,20 @@ int walMetaDeserialize(SWal* pWal, const char* bytes) {
ASSERT(taosArrayGetSize(pWal->fileInfoSet) == 0);
cJSON *pRoot, *pMeta, *pFiles, *pInfoJson, *pField;
pRoot = cJSON_Parse(bytes);
if (!pRoot) goto _err;
pMeta = cJSON_GetObjectItem(pRoot, "meta");
if (!pMeta) goto _err;
pField = cJSON_GetObjectItem(pMeta, "firstVer");
if (!pField) goto _err;
pWal->vers.firstVer = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pMeta, "snapshotVer");
if (!pField) goto _err;
pWal->vers.snapshotVer = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pMeta, "commitVer");
if (!pField) goto _err;
pWal->vers.commitVer = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pMeta, "lastVer");
if (!pField) goto _err;
pWal->vers.lastVer = atoll(cJSON_GetStringValue(pField));
pFiles = cJSON_GetObjectItem(pRoot, "files");
@ -512,17 +667,23 @@ int walMetaDeserialize(SWal* pWal, const char* bytes) {
taosArrayEnsureCap(pArray, sz);
SWalFileInfo* pData = pArray->pData;
for (int i = 0; i < sz; i++) {
cJSON* pInfoJson = cJSON_GetArrayItem(pFiles, i);
cJSON* pInfoJson = cJSON_GetArrayItem(pFiles, i);
if (!pInfoJson) goto _err;
SWalFileInfo* pInfo = &pData[i];
pField = cJSON_GetObjectItem(pInfoJson, "firstVer");
if (!pField) goto _err;
pInfo->firstVer = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pInfoJson, "lastVer");
if (!pField) goto _err;
pInfo->lastVer = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pInfoJson, "createTs");
if (!pField) goto _err;
pInfo->createTs = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pInfoJson, "closeTs");
if (!pField) goto _err;
pInfo->closeTs = atoll(cJSON_GetStringValue(pField));
pField = cJSON_GetObjectItem(pInfoJson, "fileSize");
if (!pField) goto _err;
pInfo->fileSize = atoll(cJSON_GetStringValue(pField));
}
taosArraySetSize(pArray, sz);
@ -530,6 +691,10 @@ int walMetaDeserialize(SWal* pWal, const char* bytes) {
pWal->writeCur = sz - 1;
cJSON_Delete(pRoot);
return 0;
_err:
cJSON_Delete(pRoot);
return -1;
}
static int walFindCurMetaVer(SWal* pWal) {
@ -565,22 +730,63 @@ static int walFindCurMetaVer(SWal* pWal) {
int walSaveMeta(SWal* pWal) {
int metaVer = walFindCurMetaVer(pWal);
char fnameStr[WAL_FILE_LEN];
walBuildMetaName(pWal, metaVer + 1, fnameStr);
TdFilePtr pMetaFile = taosOpenFile(fnameStr, TD_FILE_CREATE | TD_FILE_WRITE);
if (pMetaFile == NULL) {
char tmpFnameStr[WAL_FILE_LEN];
int n;
// fsync the idx and log file at first to ensure validity of meta
if (taosFsyncFile(pWal->pIdxFile) < 0) {
wError("vgId:%d, failed to sync idx file due to %s", pWal->cfg.vgId, strerror(errno));
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
if (taosFsyncFile(pWal->pLogFile) < 0) {
wError("vgId:%d, failed to sync log file due to %s", pWal->cfg.vgId, strerror(errno));
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
// flush to a tmpfile
n = walBuildTmpMetaName(pWal, tmpFnameStr);
ASSERT(n < sizeof(tmpFnameStr) && "Buffer overflow of file name");
TdFilePtr pMetaFile = taosOpenFile(tmpFnameStr, TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_TRUNC);
if (pMetaFile == NULL) {
wError("vgId:%d, failed to open file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), tmpFnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
char* serialized = walMetaSerialize(pWal);
int len = strlen(serialized);
if (len != taosWriteFile(pMetaFile, serialized, len)) {
// TODO:clean file
taosCloseFile(&pMetaFile);
taosRemoveFile(fnameStr);
return -1;
wError("vgId:%d, failed to write file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), tmpFnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
if (taosFsyncFile(pMetaFile) < 0) {
wError("vgId:%d, failed to sync file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), tmpFnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
if (taosCloseFile(&pMetaFile) < 0) {
wError("vgId:%d, failed to close file due to %s. file:%s", pWal->cfg.vgId, strerror(errno), tmpFnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
// rename it
n = walBuildMetaName(pWal, metaVer + 1, fnameStr);
ASSERT(n < sizeof(fnameStr) && "Buffer overflow of file name");
if (taosRenameFile(tmpFnameStr, fnameStr) < 0) {
wError("failed to rename file due to %s. dest:%s", strerror(errno), fnameStr);
terrno = TAOS_SYSTEM_ERROR(errno);
goto _err;
}
taosCloseFile(&pMetaFile);
// delete old file
if (metaVer > -1) {
walBuildMetaName(pWal, metaVer, fnameStr);
@ -588,6 +794,11 @@ int walSaveMeta(SWal* pWal) {
}
taosMemoryFree(serialized);
return 0;
_err:
taosCloseFile(&pMetaFile);
taosMemoryFree(serialized);
return -1;
}
int walLoadMeta(SWal* pWal) {
@ -629,6 +840,10 @@ int walLoadMeta(SWal* pWal) {
}
// load into fileInfoSet
int code = walMetaDeserialize(pWal, buf);
if (code < 0) {
wError("failed to deserialize wal meta. file:%s", fnameStr);
terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
}
taosCloseFile(&pFile);
taosMemoryFree(buf);
return code;

View File

@ -81,6 +81,12 @@ SWal *walOpen(const char *path, SWalCfg *pCfg) {
return NULL;
}
if (taosThreadMutexInit(&pWal->mutex, NULL) < 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
taosMemoryFree(pWal);
return NULL;
}
// set config
memcpy(&pWal->cfg, pCfg, sizeof(SWalCfg));
@ -98,15 +104,14 @@ SWal *walOpen(const char *path, SWalCfg *pCfg) {
tstrncpy(pWal->path, path, sizeof(pWal->path));
if (taosMkDir(pWal->path) != 0) {
wError("vgId:%d, path:%s, failed to create directory since %s", pWal->cfg.vgId, pWal->path, strerror(errno));
taosMemoryFree(pWal);
return NULL;
goto _err;
}
// init ref
pWal->pRefHash = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, HASH_ENTRY_LOCK);
if (pWal->pRefHash == NULL) {
taosMemoryFree(pWal);
return NULL;
wError("failed to init hash since %s", tstrerror(terrno));
goto _err;
}
// open meta
@ -117,9 +122,7 @@ SWal *walOpen(const char *path, SWalCfg *pCfg) {
pWal->fileInfoSet = taosArrayInit(8, sizeof(SWalFileInfo));
if (pWal->fileInfoSet == NULL) {
wError("vgId:%d, path:%s, failed to init taosArray %s", pWal->cfg.vgId, pWal->path, strerror(errno));
taosHashCleanup(pWal->pRefHash);
taosMemoryFree(pWal);
return NULL;
goto _err;
}
// init status
@ -131,46 +134,37 @@ SWal *walOpen(const char *path, SWalCfg *pCfg) {
pWal->writeHead.head.protoVer = WAL_PROTO_VER;
pWal->writeHead.magic = WAL_MAGIC;
if (taosThreadMutexInit(&pWal->mutex, NULL) < 0) {
taosArrayDestroy(pWal->fileInfoSet);
taosHashCleanup(pWal->pRefHash);
taosMemoryFree(pWal);
return NULL;
}
pWal->refId = taosAddRef(tsWal.refSetId, pWal);
if (pWal->refId < 0) {
taosHashCleanup(pWal->pRefHash);
taosThreadMutexDestroy(&pWal->mutex);
taosArrayDestroy(pWal->fileInfoSet);
taosMemoryFree(pWal);
return NULL;
}
walLoadMeta(pWal);
// load meta
(void)walLoadMeta(pWal);
if (walCheckAndRepairMeta(pWal) < 0) {
wError("vgId:%d cannot open wal since repair meta file failed", pWal->cfg.vgId);
taosHashCleanup(pWal->pRefHash);
taosRemoveRef(tsWal.refSetId, pWal->refId);
taosThreadMutexDestroy(&pWal->mutex);
taosArrayDestroy(pWal->fileInfoSet);
return NULL;
goto _err;
}
if (walCheckAndRepairIdx(pWal) < 0) {
wError("vgId:%d cannot open wal since repair idx file failed", pWal->cfg.vgId);
taosHashCleanup(pWal->pRefHash);
taosRemoveRef(tsWal.refSetId, pWal->refId);
taosThreadMutexDestroy(&pWal->mutex);
taosArrayDestroy(pWal->fileInfoSet);
return NULL;
goto _err;
}
// add ref
pWal->refId = taosAddRef(tsWal.refSetId, pWal);
if (pWal->refId < 0) {
wError("failed to add ref for Wal since %s", tstrerror(terrno));
goto _err;
}
wDebug("vgId:%d, wal:%p is opened, level:%d fsyncPeriod:%d", pWal->cfg.vgId, pWal, pWal->cfg.level,
pWal->cfg.fsyncPeriod);
return pWal;
_err:
taosArrayDestroy(pWal->fileInfoSet);
taosHashCleanup(pWal->pRefHash);
taosThreadMutexDestroy(&pWal->mutex);
taosMemoryFree(pWal);
pWal = NULL;
return NULL;
}
int32_t walAlter(SWal *pWal, SWalCfg *pCfg) {
@ -195,11 +189,11 @@ int32_t walAlter(SWal *pWal, SWalCfg *pCfg) {
void walClose(SWal *pWal) {
taosThreadMutexLock(&pWal->mutex);
(void)walSaveMeta(pWal);
taosCloseFile(&pWal->pLogFile);
pWal->pLogFile = NULL;
taosCloseFile(&pWal->pIdxFile);
pWal->pIdxFile = NULL;
walSaveMeta(pWal);
taosArrayDestroy(pWal->fileInfoSet);
pWal->fileInfoSet = NULL;
taosHashCleanup(pWal->pRefHash);

View File

@ -181,7 +181,11 @@ int32_t walReadSeekVerImpl(SWalReader *pReader, int64_t ver) {
SWalFileInfo tmpInfo;
tmpInfo.firstVer = ver;
SWalFileInfo *pRet = taosArraySearch(pWal->fileInfoSet, &tmpInfo, compareWalFileInfo, TD_LE);
ASSERT(pRet != NULL);
if (pRet == NULL) {
wError("failed to find WAL log file with ver:%lld", ver);
terrno = TSDB_CODE_WAL_INVALID_VER;
return -1;
}
if (pReader->curFileFirstVer != pRet->firstVer) {
// error code was set inner
if (walReadChangeFile(pReader, pRet->firstVer) < 0) {
@ -472,7 +476,8 @@ int32_t walReadVer(SWalReader *pReader, int64_t ver) {
} else {
terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
}
ASSERT(0);
wError("vgId:%d, failed to read WAL record head, index:%" PRId64 ", from log file since %s",
pReader->pWal->cfg.vgId, ver, terrstr());
taosThreadMutexUnlock(&pReader->mutex);
return -1;
}
@ -505,6 +510,8 @@ int32_t walReadVer(SWalReader *pReader, int64_t ver) {
else {
terrno = TSDB_CODE_WAL_FILE_CORRUPTED;
}
wError("vgId:%d, failed to read WAL record body, index:%" PRId64 ", from log file since %s",
pReader->pWal->cfg.vgId, ver, terrstr());
taosThreadMutexUnlock(&pReader->mutex);
return -1;
}

View File

@ -79,6 +79,11 @@ int64_t walChangeWrite(SWal* pWal, int64_t ver) {
TdFilePtr pIdxTFile, pLogTFile;
char fnameStr[WAL_FILE_LEN];
if (pWal->pLogFile != NULL) {
code = taosFsyncFile(pWal->pLogFile);
if (code != 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
code = taosCloseFile(&pWal->pLogFile);
if (code != 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
@ -86,6 +91,11 @@ int64_t walChangeWrite(SWal* pWal, int64_t ver) {
}
}
if (pWal->pIdxFile != NULL) {
code = taosFsyncFile(pWal->pIdxFile);
if (code != 0) {
terrno = TAOS_SYSTEM_ERROR(errno);
return -1;
}
code = taosCloseFile(&pWal->pIdxFile);
if (code != 0) {
terrno = TAOS_SYSTEM_ERROR(errno);

View File

@ -219,10 +219,12 @@ int32_t walRollback(SWal *pWal, int64_t ver) {
taosCloseFile(&pIdxFile);
taosCloseFile(&pLogFile);
taosFsyncFile(pWal->pLogFile);
taosFsyncFile(pWal->pIdxFile);
walSaveMeta(pWal);
code = walSaveMeta(pWal);
if (code < 0) {
wError("vgId:%d, failed to save meta since %s", pWal->cfg.vgId, terrstr());
taosThreadMutexUnlock(&pWal->mutex);
return -1;
}
// unlock
taosThreadMutexUnlock(&pWal->mutex);
@ -394,7 +396,11 @@ int32_t walRollImpl(SWal *pWal) {
pWal->lastRollSeq = walGetSeq();
walSaveMeta(pWal);
code = walSaveMeta(pWal);
if (code < 0) {
wError("vgId:%d, failed to save meta since %s", pWal->cfg.vgId, terrstr());
goto END;
}
END:
return code;
@ -550,6 +556,11 @@ int32_t walWrite(SWal *pWal, int64_t index, tmsg_t msgType, const void *body, in
void walFsync(SWal *pWal, bool forceFsync) {
if (forceFsync || (pWal->cfg.level == TAOS_WAL_FSYNC && pWal->cfg.fsyncPeriod == 0)) {
wTrace("vgId:%d, fileId:%" PRId64 ".idx, do fsync", pWal->cfg.vgId, walGetCurFileFirstVer(pWal));
if (taosFsyncFile(pWal->pIdxFile) < 0) {
wError("vgId:%d, file:%" PRId64 ".idx, fsync failed since %s", pWal->cfg.vgId, walGetCurFileFirstVer(pWal),
strerror(errno));
}
wTrace("vgId:%d, fileId:%" PRId64 ".log, do fsync", pWal->cfg.vgId, walGetCurFileFirstVer(pWal));
if (taosFsyncFile(pWal->pLogFile) < 0) {
wError("vgId:%d, file:%" PRId64 ".log, fsync failed since %s", pWal->cfg.vgId, walGetCurFileFirstVer(pWal),

View File

@ -87,11 +87,17 @@ void osUpdate() {
void osCleanup() {}
bool osLogSpaceAvailable() { return tsLogSpace.reserved <= tsLogSpace.size.avail; }
bool osLogSpaceAvailable() { return tsLogSpace.size.avail > 0; }
bool osDataSpaceAvailable() { return tsDataSpace.reserved <= tsDataSpace.size.avail; }
bool osDataSpaceAvailable() { return tsDataSpace.size.avail > 0; }
bool osTempSpaceAvailable() { return tsTempSpace.reserved <= tsTempSpace.size.avail; }
bool osTempSpaceAvailable() { return tsTempSpace.size.avail > 0; }
bool osLogSpaceSufficient() { return tsLogSpace.size.avail > tsLogSpace.reserved; }
bool osDataSpaceSufficient() { return tsDataSpace.size.avail > tsDataSpace.reserved; }
bool osTempSpaceSufficient() { return tsTempSpace.size.avail > tsTempSpace.reserved; }
void osSetTimezone(const char *timezone) { taosSetSystemTimezone(timezone, tsTimezoneStr, &tsDaylight, &tsTimezone); }

View File

@ -447,12 +447,13 @@ TAOS_DEFINE_ERROR(TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND, "TQ table schema not f
TAOS_DEFINE_ERROR(TSDB_CODE_TQ_NO_COMMITTED_OFFSET, "TQ no commited offset")
// wal
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_APP_ERROR, "Wal unexpected generic error")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_APP_ERROR, "WAL unexpected generic error")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_FILE_CORRUPTED, "WAL file is corrupted")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_SIZE_LIMIT, "WAL size exceeds limit")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_INVALID_VER, "WAL use invalid version")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_OUT_OF_MEMORY, "WAL out of memory")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_LOG_NOT_EXIST, "WAL log not exist")
TAOS_DEFINE_ERROR(TSDB_CODE_WAL_CHKSUM_MISMATCH, "WAL checksum mismatch")
// tfs
TAOS_DEFINE_ERROR(TSDB_CODE_FS_INVLD_CFG, "tfs invalid mount config")