diff --git a/cmake/install.inc b/cmake/install.inc index 746e493a17..4b2d4828f8 100755 --- a/cmake/install.inc +++ b/cmake/install.inc @@ -32,7 +32,7 @@ ELSEIF (TD_WINDOWS) #INSTALL(TARGETS taos RUNTIME DESTINATION driver) #INSTALL(TARGETS shell RUNTIME DESTINATION .) IF (TD_MVN_INSTALLED) - INSTALL(FILES ${LIBRARY_OUTPUT_PATH}/taos-jdbcdriver-2.0.8-dist.jar DESTINATION connector/jdbc) + INSTALL(FILES ${LIBRARY_OUTPUT_PATH}/taos-jdbcdriver-2.0.13-dist.jar DESTINATION connector/jdbc) ENDIF () ELSEIF (TD_DARWIN) SET(TD_MAKE_INSTALL_SH "${TD_COMMUNITY_DIR}/packaging/tools/make_install.sh") diff --git a/cmake/version.inc b/cmake/version.inc index 741f76da43..2f0ec81aea 100644 --- a/cmake/version.inc +++ b/cmake/version.inc @@ -4,7 +4,7 @@ PROJECT(TDengine) IF (DEFINED VERNUMBER) SET(TD_VER_NUMBER ${VERNUMBER}) ELSE () - SET(TD_VER_NUMBER "2.0.7.0") + SET(TD_VER_NUMBER "2.0.8.0") ENDIF () IF (DEFINED VERCOMPATIBLE) diff --git a/snap/snapcraft.yaml b/snap/snapcraft.yaml index 1738ff7ec8..059c0650c2 100644 --- a/snap/snapcraft.yaml +++ b/snap/snapcraft.yaml @@ -1,6 +1,6 @@ name: tdengine base: core18 -version: '2.0.7.0' +version: '2.0.8.0' icon: snap/gui/t-dengine.svg summary: an open-source big data platform designed and optimized for IoT. description: | @@ -72,7 +72,7 @@ parts: - usr/bin/taosd - usr/bin/taos - usr/bin/taosdemo - - usr/lib/libtaos.so.2.0.6.0 + - usr/lib/libtaos.so.2.0.8.0 - usr/lib/libtaos.so.1 - usr/lib/libtaos.so diff --git a/src/balance/src/balance.c b/src/balance/src/balance.c index df78f4fe27..b172867929 100644 --- a/src/balance/src/balance.c +++ b/src/balance/src/balance.c @@ -162,7 +162,6 @@ int32_t balanceAllocVnodes(SVgObj *pVgroup) { pDnode->openVnodes, pDnode->diskAvailable, pDnode->alternativeRole); mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); if (mnodeGetOnlineDnodesNum() == 0) { return TSDB_CODE_MND_NOT_READY; @@ -377,15 +376,13 @@ static bool balanceMonitorBalance() { srcScore, pDestDnode->score, destScore); balanceAddVnode(pVgroup, pSrcDnode, pDestDnode); mnodeDecVgroupRef(pVgroup); - sdbFreeIter(pIter); + mnodeCancelGetNextVgroup(pIter); return true; } } mnodeDecVgroupRef(pVgroup); } - - sdbFreeIter(pIter); } return false; @@ -413,8 +410,6 @@ void balanceReset() { mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); - tsAccessSquence = 0; } @@ -443,12 +438,11 @@ static int32_t balanceMonitorVgroups() { mnodeDecVgroupRef(pVgroup); if (code == TSDB_CODE_SUCCESS) { + mnodeCancelGetNextVgroup(pIter); break; } } - sdbFreeIter(pIter); - return hasUpdatingVgroup; } @@ -465,11 +459,12 @@ static bool balanceMonitorDnodeDropping(SDnodeObj *pDnode) { hasThisDnode = balanceCheckDnodeInVgroup(pDnode, pVgroup); mnodeDecVgroupRef(pVgroup); - if (hasThisDnode) break; + if (hasThisDnode) { + mnodeCancelGetNextVgroup(pIter); + break; + } } - sdbFreeIter(pIter); - if (!hasThisDnode) { mInfo("dnode:%d, dropped for all vnodes are moving to other dnodes", pDnode->dnodeId); mnodeDropDnode(pDnode, NULL); @@ -499,20 +494,18 @@ static bool balanceMontiorDropping() { pDnode->status = TAOS_DN_STATUS_DROPPING; mnodeUpdateDnode(pDnode); mnodeDecDnodeRef(pDnode); - sdbFreeIter(pIter); + mnodeCancelGetNextDnode(pIter); return true; } if (pDnode->status == TAOS_DN_STATUS_DROPPING) { bool ret = balanceMonitorDnodeDropping(pDnode); mnodeDecDnodeRef(pDnode); - sdbFreeIter(pIter); + mnodeCancelGetNextDnode(pIter); return ret; } } - sdbFreeIter(pIter); - return false; } @@ -556,8 +549,6 @@ static void balanceSetVgroupOffline(SDnodeObj* pDnode) { } mnodeDecVgroupRef(pVgroup); } - - sdbFreeIter(pIter); } static void balanceCheckDnodeAccess() { @@ -578,8 +569,6 @@ static void balanceCheckDnodeAccess() { } mnodeDecDnodeRef(pDnode); } - - sdbFreeIter(pIter); } static void balanceProcessBalanceTimer(void *handle, void *tmrId) { @@ -630,6 +619,7 @@ void balanceAsyncNotify() { int32_t balanceInit() { mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_SCORES, balanceGetScoresMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_SCORES, balanceRetrieveScores); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_SCORES, mnodeCancelGetNextDnode); pthread_mutex_init(&tsBalanceMutex, NULL); balanceInitDnodeList(); @@ -667,8 +657,6 @@ int32_t balanceDropDnode(SDnodeObj *pDnode) { mnodeDecDnodeRef(pTempDnode); } - sdbFreeIter(pIter); - if (pDnode->openVnodes > totalFreeVnodes) { mError("dnode:%d, openVnodes:%d totalFreeVnodes:%d no enough dnodes", pDnode->dnodeId, pDnode->openVnodes, totalFreeVnodes); return TSDB_CODE_MND_NO_ENOUGH_DNODES; @@ -780,8 +768,12 @@ void balanceAccquireDnodeList() { SDnodeObj *pDnode = NULL; int32_t dnodeIndex = 0; - while (1) { - if (dnodeIndex >= dnodesNum) break; + while (1) { + if (dnodeIndex >= dnodesNum) { + mnodeCancelGetNextDnode(pIter); + break; + } + pIter = mnodeGetNextDnode(pIter, &pDnode); if (pDnode == NULL) break; if (pDnode->status == TAOS_DN_STATUS_OFFLINE) { @@ -802,8 +794,6 @@ void balanceAccquireDnodeList() { dnodeIndex++; } - sdbFreeIter(pIter); - tsBalanceDnodeListSize = dnodeIndex; } diff --git a/src/client/inc/tsclient.h b/src/client/inc/tsclient.h index ff36cf0f5a..c482375bff 100644 --- a/src/client/inc/tsclient.h +++ b/src/client/inc/tsclient.h @@ -282,7 +282,7 @@ typedef struct { int8_t dataSourceType; // load data from file or not int8_t submitSchema; // submit block is built with table schema - STagData tagData; + STagData *pTagData; // NOTE: pTagData->data is used as a variant length array SHashObj *pTableList; // referred table involved in sql SArray *pDataBlocks; // SArray submit data blocks after parsing sql } SSqlCmd; diff --git a/src/client/src/tscParseInsert.c b/src/client/src/tscParseInsert.c index a44a158f93..6de8195d73 100644 --- a/src/client/src/tscParseInsert.c +++ b/src/client/src/tscParseInsert.c @@ -790,9 +790,6 @@ static int32_t tscCheckIfCreateTable(char **sqlstr, SSqlObj *pSql) { sql += index; tscAllocPayload(pCmd, sizeof(STagData)); - STagData *pTag = &pCmd->tagData; - - memset(pTag, 0, sizeof(STagData)); //the source super table is moved to the secondary position of the pTableMetaInfo list if (pQueryInfo->numOfTables < 2) { @@ -805,7 +802,14 @@ static int32_t tscCheckIfCreateTable(char **sqlstr, SSqlObj *pSql) { return code; } + STagData *pTag = realloc(pCmd->pTagData, offsetof(STagData, data)); + if (pTag == NULL) { + return TSDB_CODE_TSC_OUT_OF_MEMORY; + } + memset(pTag, 0, offsetof(STagData, data)); tstrncpy(pTag->name, pSTableMeterMetaInfo->name, sizeof(pTag->name)); + pCmd->pTagData = pTag; + code = tscGetTableMeta(pSql, pSTableMeterMetaInfo); if (code != TSDB_CODE_SUCCESS) { return code; @@ -934,7 +938,13 @@ static int32_t tscCheckIfCreateTable(char **sqlstr, SSqlObj *pSql) { return TSDB_CODE_TSC_OUT_OF_MEMORY; } tdSortKVRowByColIdx(row); - pTag->dataLen = kvRowLen(row); + + pTag = (STagData*)realloc(pCmd->pTagData, offsetof(STagData, data) + kvRowLen(row)); + if (pTag == NULL) { + return TSDB_CODE_TSC_OUT_OF_MEMORY; + } + pCmd->pTagData = pTag; + pTag->dataLen = htonl(kvRowLen(row)); kvRowCpy(pTag->data, row); free(row); @@ -945,8 +955,6 @@ static int32_t tscCheckIfCreateTable(char **sqlstr, SSqlObj *pSql) { return tscSQLSyntaxErrMsg(pCmd->payload, ") expected", sToken.z); } - pTag->dataLen = htonl(pTag->dataLen); - if (tscValidateName(&tableToken) != TSDB_CODE_SUCCESS) { return tscInvalidSQLErrMsg(pCmd->payload, "invalid table name", *sqlstr); } diff --git a/src/client/src/tscServer.c b/src/client/src/tscServer.c index 15859674ab..9735e26ece 100644 --- a/src/client/src/tscServer.c +++ b/src/client/src/tscServer.c @@ -1565,11 +1565,11 @@ int tscBuildTableMetaMsg(SSqlObj *pSql, SSqlInfo *pInfo) { char *pMsg = (char *)pInfoMsg + sizeof(STableInfoMsg); - size_t len = htonl(pCmd->tagData.dataLen); - if (pSql->cmd.autoCreated) { + if (pCmd->autoCreated && pCmd->pTagData != NULL) { + int len = htonl(pCmd->pTagData->dataLen); if (len > 0) { - len += sizeof(pCmd->tagData.name) + sizeof(pCmd->tagData.dataLen); - memcpy(pInfoMsg->tags, &pCmd->tagData, len); + len += sizeof(pCmd->pTagData->name) + sizeof(pCmd->pTagData->dataLen); + memcpy(pInfoMsg->tags, pCmd->pTagData, len); pMsg += len; } } @@ -2239,8 +2239,6 @@ static int32_t getTableMetaFromMgmt(SSqlObj *pSql, STableMetaInfo *pTableMetaInf pNew->signature = pNew; pNew->cmd.command = TSDB_SQL_META; - registerSqlObj(pNew); - tscAddSubqueryInfo(&pNew->cmd); SQueryInfo *pNewQueryInfo = tscGetQueryInfoDetailSafely(&pNew->cmd, 0); @@ -2248,8 +2246,7 @@ static int32_t getTableMetaFromMgmt(SSqlObj *pSql, STableMetaInfo *pTableMetaInf pNew->cmd.autoCreated = pSql->cmd.autoCreated; // create table if not exists if (TSDB_CODE_SUCCESS != tscAllocPayload(&pNew->cmd, TSDB_DEFAULT_PAYLOAD_SIZE + pSql->cmd.payloadLen)) { tscError("%p malloc failed for payload to get table meta", pSql); - free(pNew); - + tscFreeSqlObj(pNew); return TSDB_CODE_TSC_OUT_OF_MEMORY; } @@ -2257,12 +2254,25 @@ static int32_t getTableMetaFromMgmt(SSqlObj *pSql, STableMetaInfo *pTableMetaInf assert(pNew->cmd.numOfClause == 1 && pNewQueryInfo->numOfTables == 1); tstrncpy(pNewMeterMetaInfo->name, pTableMetaInfo->name, sizeof(pNewMeterMetaInfo->name)); - memcpy(&pNew->cmd.tagData, &pSql->cmd.tagData, sizeof(pSql->cmd.tagData)); + + if (pSql->cmd.pTagData != NULL) { + int size = offsetof(STagData, data) + htonl(pSql->cmd.pTagData->dataLen); + pNew->cmd.pTagData = calloc(1, size); + if (pNew->cmd.pTagData == NULL) { + tscError("%p malloc failed for new tag data to get table meta", pSql); + tscFreeSqlObj(pNew); + return TSDB_CODE_TSC_OUT_OF_MEMORY; + } + memcpy(pNew->cmd.pTagData, pSql->cmd.pTagData, size); + } + tscDebug("%p new pSqlObj:%p to get tableMeta, auto create:%d", pSql, pNew, pNew->cmd.autoCreated); pNew->fp = tscTableMetaCallBack; pNew->param = pSql; + registerSqlObj(pNew); + int32_t code = tscProcessSql(pNew); if (code == TSDB_CODE_SUCCESS) { code = TSDB_CODE_TSC_ACTION_IN_PROGRESS; // notify upper application that current process need to be terminated diff --git a/src/client/src/tscSub.c b/src/client/src/tscSub.c index 2c81bd7c7c..a782b53e75 100644 --- a/src/client/src/tscSub.c +++ b/src/client/src/tscSub.c @@ -442,7 +442,12 @@ TAOS_RES *taos_consume(TAOS_SUB *tsub) { size_t size = taosArrayGetSize(pSub->progress) * sizeof(STableIdInfo); size += sizeof(SQueryTableMsg) + 4096; - tscAllocPayload(&pSql->cmd, (int)size); + int code = tscAllocPayload(&pSql->cmd, (int)size); + if (code != TSDB_CODE_SUCCESS) { + tscError("failed to alloc payload"); + return NULL; + } + for (int retry = 0; retry < 3; retry++) { tscRemoveFromSqlList(pSql); diff --git a/src/client/src/tscUtil.c b/src/client/src/tscUtil.c index a75f6d8296..d22c2588cc 100644 --- a/src/client/src/tscUtil.c +++ b/src/client/src/tscUtil.c @@ -408,7 +408,7 @@ void tscResetSqlCmdObj(SSqlCmd* pCmd, bool removeFromCache) { pCmd->pTableList = NULL; pCmd->pDataBlocks = tscDestroyBlockArrayList(pCmd->pDataBlocks); - + tscFreeQueryInfo(pCmd, removeFromCache); } @@ -512,6 +512,8 @@ void tscFreeSqlObj(SSqlObj* pSql) { tscFreeSqlResult(pSql); tscResetSqlCmdObj(pCmd, false); + tfree(pCmd->pTagData); + memset(pCmd->payload, 0, (size_t)pCmd->allocSize); tfree(pCmd->payload); pCmd->allocSize = 0; @@ -1909,7 +1911,17 @@ SSqlObj* createSimpleSubObj(SSqlObj* pSql, void (*fp)(), void* param, int32_t cm pCmd->command = cmd; pCmd->parseFinished = 1; pCmd->autoCreated = pSql->cmd.autoCreated; - memcpy(&pCmd->tagData, &pSql->cmd.tagData, sizeof(pCmd->tagData)); + + if (pSql->cmd.pTagData != NULL) { + int size = offsetof(STagData, data) + htonl(pSql->cmd.pTagData->dataLen); + pNew->cmd.pTagData = calloc(1, size); + if (pNew->cmd.pTagData == NULL) { + tscError("%p new subquery failed, unable to malloc tag data, tableIndex:%d", pSql, 0); + free(pNew); + return NULL; + } + memcpy(pNew->cmd.pTagData, pSql->cmd.pTagData, size); + } if (tscAddSubqueryInfo(pCmd) != TSDB_CODE_SUCCESS) { tscFreeSqlObj(pNew); diff --git a/src/common/inc/tglobal.h b/src/common/inc/tglobal.h index 4087f638a9..efe3d7678a 100644 --- a/src/common/inc/tglobal.h +++ b/src/common/inc/tglobal.h @@ -97,6 +97,7 @@ extern int32_t tsAlternativeRole; extern int32_t tsBalanceInterval; extern int32_t tsOfflineThreshold; extern int32_t tsMnodeEqualVnodeNum; +extern int32_t tsFlowCtrl; // restful extern int32_t tsEnableHttpModule; diff --git a/src/common/src/tglobal.c b/src/common/src/tglobal.c index 4495c3d928..d26746a5d1 100644 --- a/src/common/src/tglobal.c +++ b/src/common/src/tglobal.c @@ -133,6 +133,7 @@ int32_t tsAlternativeRole = 0; int32_t tsBalanceInterval = 300; // seconds int32_t tsOfflineThreshold = 86400*100; // seconds 10days int32_t tsMnodeEqualVnodeNum = 4; +int32_t tsFlowCtrl = 1; // restful int32_t tsEnableHttpModule = 1; @@ -199,7 +200,7 @@ int32_t tsNumOfLogLines = 10000000; int32_t mDebugFlag = 135; int32_t sdbDebugFlag = 135; int32_t dDebugFlag = 135; -int32_t vDebugFlag = 131; +int32_t vDebugFlag = 135; int32_t cDebugFlag = 131; int32_t jniDebugFlag = 131; int32_t odbcDebugFlag = 131; @@ -971,6 +972,17 @@ static void doInitGlobalConfig(void) { cfg.maxValue = 1000; cfg.ptrLength = 0; cfg.unitType = TAOS_CFG_UTYPE_NONE; + taosInitConfigOption(cfg); + + // module configs + cfg.option = "flowctrl"; + cfg.ptr = &tsFlowCtrl; + cfg.valType = TAOS_CFG_VTYPE_INT32; + cfg.cfgType = TSDB_CFG_CTYPE_B_CONFIG | TSDB_CFG_CTYPE_B_SHOW; + cfg.minValue = 0; + cfg.maxValue = 1; + cfg.ptrLength = 0; + cfg.unitType = TAOS_CFG_UTYPE_NONE; taosInitConfigOption(cfg); cfg.option = "http"; diff --git a/src/connector/go b/src/connector/go index 050667e5b4..8c58c512b6 160000 --- a/src/connector/go +++ b/src/connector/go @@ -1 +1 @@ -Subproject commit 050667e5b4d0eafa5387e4283e713559b421203f +Subproject commit 8c58c512b6acda8bcdfa48fdc7140227b5221766 diff --git a/src/connector/grafanaplugin b/src/connector/grafanaplugin index ec77d9049a..d598db167e 160000 --- a/src/connector/grafanaplugin +++ b/src/connector/grafanaplugin @@ -1 +1 @@ -Subproject commit ec77d9049a719dabfd1a7c1122a209e201861944 +Subproject commit d598db167eb256fe67409b7bb3d0eb7fffc3ff8c diff --git a/src/connector/jdbc/CMakeLists.txt b/src/connector/jdbc/CMakeLists.txt index c565853ab0..701a39b209 100644 --- a/src/connector/jdbc/CMakeLists.txt +++ b/src/connector/jdbc/CMakeLists.txt @@ -8,7 +8,7 @@ IF (TD_MVN_INSTALLED) ADD_CUSTOM_COMMAND(OUTPUT ${JDBC_CMD_NAME} POST_BUILD COMMAND mvn -Dmaven.test.skip=true install -f ${CMAKE_CURRENT_SOURCE_DIR}/pom.xml - COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/target/taos-jdbcdriver-2.0.8-dist.jar ${LIBRARY_OUTPUT_PATH} + COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/target/taos-jdbcdriver-2.0.13-dist.jar ${LIBRARY_OUTPUT_PATH} COMMAND mvn -Dmaven.test.skip=true clean -f ${CMAKE_CURRENT_SOURCE_DIR}/pom.xml COMMENT "build jdbc driver") ADD_CUSTOM_TARGET(${JDBC_TARGET_NAME} ALL WORKING_DIRECTORY ${EXECUTABLE_OUTPUT_PATH} DEPENDS ${JDBC_CMD_NAME}) diff --git a/src/connector/jdbc/deploy-pom.xml b/src/connector/jdbc/deploy-pom.xml index 51db837c7b..1dc2625e62 100755 --- a/src/connector/jdbc/deploy-pom.xml +++ b/src/connector/jdbc/deploy-pom.xml @@ -5,7 +5,7 @@ com.taosdata.jdbc taos-jdbcdriver - 2.0.10 + 2.0.13 jar JDBCDriver @@ -49,17 +49,29 @@ - - org.apache.commons - commons-lang3 - 3.5 - junit junit 4.13 test + + + + org.apache.httpcomponents + httpclient + 4.5.8 + + + org.apache.commons + commons-lang3 + 3.9 + + + com.alibaba + fastjson + 1.2.58 + diff --git a/src/connector/jdbc/pom.xml b/src/connector/jdbc/pom.xml index e7124a0599..3d1f402435 100755 --- a/src/connector/jdbc/pom.xml +++ b/src/connector/jdbc/pom.xml @@ -3,7 +3,7 @@ 4.0.0 com.taosdata.jdbc taos-jdbcdriver - 2.0.8 + 2.0.13 jar JDBCDriver https://github.com/taosdata/TDengine/tree/master/src/connector/jdbc @@ -112,6 +112,13 @@ maven-surefire-plugin 2.12.4 + + **/*Test.java + + + **/BatchInsertTest.java + **/FailOverTest.java + true diff --git a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulConnection.java b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulConnection.java index b82efca3ef..6b0937a9b7 100644 --- a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulConnection.java +++ b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulConnection.java @@ -15,7 +15,6 @@ public class RestfulConnection implements Connection { private final String database; private final String url; - public RestfulConnection(String host, String port, Properties props, String database, String url) { this.host = host; this.port = Integer.parseInt(port); @@ -28,7 +27,7 @@ public class RestfulConnection implements Connection { public Statement createStatement() throws SQLException { if (isClosed()) throw new SQLException(TSDBConstants.WrapErrMsg("restful TDengine connection is closed.")); - return new RestfulStatement(this, this.database); + return new RestfulStatement(this, database); } @Override @@ -104,22 +103,28 @@ public class RestfulConnection implements Connection { @Override public void setTransactionIsolation(int level) throws SQLException { - + //transaction is not supported + throw new SQLFeatureNotSupportedException("transactions are not supported"); } + /** + * + */ @Override public int getTransactionIsolation() throws SQLException { - return 0; + //Connection.TRANSACTION_NONE specifies that transactions are not supported. + return Connection.TRANSACTION_NONE; } @Override public SQLWarning getWarnings() throws SQLException { + //TODO: getWarnings not implemented return null; } @Override public void clearWarnings() throws SQLException { - + throw new SQLFeatureNotSupportedException("clearWarnings not supported."); } @Override @@ -209,22 +214,26 @@ public class RestfulConnection implements Connection { @Override public Clob createClob() throws SQLException { - return null; + //TODO: not supported + throw new SQLFeatureNotSupportedException(); } @Override public Blob createBlob() throws SQLException { - return null; + //TODO: not supported + throw new SQLFeatureNotSupportedException(); } @Override public NClob createNClob() throws SQLException { - return null; + //TODO: not supported + throw new SQLFeatureNotSupportedException(); } @Override public SQLXML createSQLXML() throws SQLException { - return null; + //TODO: not supported + throw new SQLFeatureNotSupportedException(); } @Override @@ -254,12 +263,14 @@ public class RestfulConnection implements Connection { @Override public Array createArrayOf(String typeName, Object[] elements) throws SQLException { - return null; + //TODO: not supported + throw new SQLFeatureNotSupportedException(); } @Override public Struct createStruct(String typeName, Object[] attributes) throws SQLException { - return null; + //TODO: not supported + throw new SQLFeatureNotSupportedException(); } @Override @@ -289,12 +300,16 @@ public class RestfulConnection implements Connection { @Override public T unwrap(Class iface) throws SQLException { - return null; + try { + return iface.cast(this); + } catch (ClassCastException cce) { + throw new SQLException("Unable to unwrap to " + iface.toString()); + } } @Override public boolean isWrapperFor(Class iface) throws SQLException { - return false; + return iface.isInstance(this); } public String getHost() { diff --git a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulDriver.java b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulDriver.java index c267f660de..9e87cfa680 100644 --- a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulDriver.java +++ b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulDriver.java @@ -35,7 +35,7 @@ public class RestfulDriver extends AbstractTaosDriver { Properties props = parseURL(url, info); String host = props.getProperty(TSDBDriver.PROPERTY_KEY_HOST, "localhost"); String port = props.getProperty(TSDBDriver.PROPERTY_KEY_PORT, "6041"); - String database = props.getProperty(TSDBDriver.PROPERTY_KEY_DBNAME); + String database = props.containsKey(TSDBDriver.PROPERTY_KEY_DBNAME) ? props.getProperty(TSDBDriver.PROPERTY_KEY_DBNAME) : null; String loginUrl = "http://" + props.getProperty(TSDBDriver.PROPERTY_KEY_HOST) + ":" + props.getProperty(TSDBDriver.PROPERTY_KEY_PORT) + "/rest/login/" @@ -86,6 +86,7 @@ public class RestfulDriver extends AbstractTaosDriver { @Override public Logger getParentLogger() throws SQLFeatureNotSupportedException { - return null; + //TODO SQLFeatureNotSupportedException + throw new SQLFeatureNotSupportedException(); } } diff --git a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulStatement.java b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulStatement.java index 20510f0135..30b56638d8 100644 --- a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulStatement.java +++ b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/rs/RestfulStatement.java @@ -4,6 +4,7 @@ import com.alibaba.fastjson.JSON; import com.alibaba.fastjson.JSONObject; import com.taosdata.jdbc.TSDBConstants; import com.taosdata.jdbc.rs.util.HttpClientPoolUtil; +import com.taosdata.jdbc.utils.SqlSyntaxValidator; import java.sql.*; import java.util.Arrays; @@ -11,19 +12,23 @@ import java.util.List; public class RestfulStatement implements Statement { - private final String catalog; + private boolean closed; + private String database; private final RestfulConnection conn; - public RestfulStatement(RestfulConnection c, String catalog) { + public RestfulStatement(RestfulConnection c, String database) { this.conn = c; - this.catalog = catalog; + this.database = database; } @Override public ResultSet executeQuery(String sql) throws SQLException { + if (isClosed()) + throw new SQLException("statement already closed"); + if (!SqlSyntaxValidator.isSelectSql(sql)) + throw new SQLException("not a select sql for executeQuery: " + sql); - final String url = "http://" + conn.getHost() + ":"+conn.getPort()+"/rest/sql"; - + final String url = "http://" + conn.getHost() + ":" + conn.getPort() + "/rest/sql"; String result = HttpClientPoolUtil.execute(url, sql); String fields = ""; List words = Arrays.asList(sql.split(" ")); @@ -65,12 +70,29 @@ public class RestfulStatement implements Statement { @Override public int executeUpdate(String sql) throws SQLException { - return 0; + if (isClosed()) + throw new SQLException("statement already closed"); + if (!SqlSyntaxValidator.isValidForExecuteUpdate(sql)) + throw new SQLException("not a valid sql for executeUpdate: " + sql); + + if (this.database == null) + throw new SQLException("Database not specified or available"); + + final String url = "http://" + conn.getHost() + ":" + conn.getPort() + "/rest/sql"; + HttpClientPoolUtil.execute(url, "use " + conn.getDatabase()); + String result = HttpClientPoolUtil.execute(url, sql); + JSONObject jsonObject = JSON.parseObject(result); + if (jsonObject.getString("status").equals("error")) { + throw new SQLException(TSDBConstants.WrapErrMsg("SQL execution error: " + + jsonObject.getString("desc") + "\n" + + "error code: " + jsonObject.getString("code"))); + } + return Integer.parseInt(jsonObject.getString("rows")); } @Override public void close() throws SQLException { - + this.closed = true; } @Override @@ -115,6 +137,7 @@ public class RestfulStatement implements Statement { @Override public SQLWarning getWarnings() throws SQLException { + //TODO: getWarnings not Implemented return null; } @@ -130,7 +153,29 @@ public class RestfulStatement implements Statement { @Override public boolean execute(String sql) throws SQLException { - return false; + if (isClosed()) { + throw new SQLException("Invalid method call on a closed statement."); + } + //如果执行了use操作应该将当前Statement的catalog设置为新的database + if (SqlSyntaxValidator.isUseSql(sql)) { + this.database = sql.trim().replace("use", "").trim(); + } + if (this.database == null) + throw new SQLException("Database not specified or available"); + + final String url = "http://" + conn.getHost() + ":" + conn.getPort() + "/rest/sql"; + // use database + HttpClientPoolUtil.execute(url, "use " + conn.getDatabase()); + // execute sql + String result = HttpClientPoolUtil.execute(url, sql); + // parse result + JSONObject jsonObject = JSON.parseObject(result); + if (jsonObject.getString("status").equals("error")) { + throw new SQLException(TSDBConstants.WrapErrMsg("SQL execution error: " + + jsonObject.getString("desc") + "\n" + + "error code: " + jsonObject.getString("code"))); + } + return true; } @Override @@ -245,7 +290,7 @@ public class RestfulStatement implements Statement { @Override public boolean isClosed() throws SQLException { - return false; + return closed; } @Override @@ -270,11 +315,15 @@ public class RestfulStatement implements Statement { @Override public T unwrap(Class iface) throws SQLException { - return null; + try { + return iface.cast(this); + } catch (ClassCastException cce) { + throw new SQLException("Unable to unwrap to " + iface.toString()); + } } @Override public boolean isWrapperFor(Class iface) throws SQLException { - return false; + return iface.isInstance(this); } } diff --git a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/utils/SqlSyntaxValidator.java b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/utils/SqlSyntaxValidator.java index 066dfad5d5..388c3978be 100644 --- a/src/connector/jdbc/src/main/java/com/taosdata/jdbc/utils/SqlSyntaxValidator.java +++ b/src/connector/jdbc/src/main/java/com/taosdata/jdbc/utils/SqlSyntaxValidator.java @@ -22,6 +22,9 @@ import java.sql.SQLException; public class SqlSyntaxValidator { + private static final String[] updateSQL = {"insert", "update", "delete", "create", "alter", "drop", "show", "describe", "use"}; + private static final String[] querySQL = {"select"}; + private TSDBConnection tsdbConnection; public SqlSyntaxValidator(Connection connection) { @@ -34,7 +37,7 @@ public class SqlSyntaxValidator { if (tsdbConnection == null || tsdbConnection.isClosed()) { throw new SQLException("invalid connection"); } else { - TSDBJNIConnector jniConnector = tsdbConnection.getConnection(); + TSDBJNIConnector jniConnector = tsdbConnection.getConnection(); if (jniConnector == null) { throw new SQLException("jniConnector is null"); } else { @@ -43,4 +46,28 @@ public class SqlSyntaxValidator { } return res; } + + public static boolean isValidForExecuteUpdate(String sql) { + for (String prefix : updateSQL) { + if (sql.trim().toLowerCase().startsWith(prefix)) + return true; + } + return false; + } + + public static boolean isUseSql(String sql) { + return sql.trim().toLowerCase().startsWith(updateSQL[8]) || sql.trim().toLowerCase().matches("create\\s*database.*") || sql.toLowerCase().toLowerCase().matches("drop\\s*database.*"); + } + + public static boolean isUpdateSql(String sql) { + return sql.trim().toLowerCase().startsWith(updateSQL[1]); + } + + public static boolean isInsertSql(String sql) { + return sql.trim().toLowerCase().startsWith(updateSQL[0]); + } + + public static boolean isSelectSql(String sql) { + return sql.trim().toLowerCase().startsWith(querySQL[0]); + } } diff --git a/src/connector/jdbc/src/test/java/com/taosdata/jdbc/cases/AppMemoryLeakTest.java b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/cases/AppMemoryLeakTest.java new file mode 100644 index 0000000000..8de2e3b442 --- /dev/null +++ b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/cases/AppMemoryLeakTest.java @@ -0,0 +1,44 @@ +package com.taosdata.jdbc.cases; + +import org.junit.Test; + +import java.sql.Connection; +import java.sql.DriverManager; +import java.sql.SQLException; +import java.sql.Statement; + +public class AppMemoryLeakTest { + + @Test(expected = SQLException.class) + public void testCreateTooManyConnection() throws ClassNotFoundException, SQLException { + Class.forName("com.taosdata.jdbc.TSDBDriver"); + int conCnt = 0; + while (true) { + Connection conn = DriverManager.getConnection("jdbc:TAOS://localhost:6030/?user=root&password=taosdata"); + System.out.println(conCnt++ + " : " + conn); + } + } + + @Test + public void testCreateTooManyStatement() throws ClassNotFoundException, SQLException { + Class.forName("com.taosdata.jdbc.TSDBDriver"); + int stmtCnt = 0; + Connection conn = DriverManager.getConnection("jdbc:TAOS://localhost:6030/?user=root&password=taosdata"); + while (true) { + Statement stmt = conn.createStatement(); + System.out.println(++stmtCnt + " : " + stmt); + } + } + + public static void main(String[] args) throws ClassNotFoundException, SQLException { + Class.forName("com.taosdata.jdbc.TSDBDriver"); + int stmtCnt = 0; + Connection conn = DriverManager.getConnection("jdbc:TAOS://localhost:6030/?user=root&password=taosdata"); + while (true) { + Statement stmt = conn.createStatement(); + System.out.println(++stmtCnt + " : " + stmt); + } + } + + +} diff --git a/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulDriverTest.java b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulDriverTest.java index a91d1c2d6b..d07a6a2179 100644 --- a/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulDriverTest.java +++ b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulDriverTest.java @@ -8,33 +8,43 @@ import java.sql.*; public class RestfulDriverTest { @Test - public void testCase001() { - try { - Class.forName("com.taosdata.jdbc.rs.RestfulDriver"); - Connection connection = DriverManager.getConnection("jdbc:TAOS-RS://master:6041/?user=root&password=taosdata"); - Statement statement = connection.createStatement(); - ResultSet resultSet = statement.executeQuery("select * from log.log"); - ResultSetMetaData metaData = resultSet.getMetaData(); - while (resultSet.next()) { - for (int i = 1; i <= metaData.getColumnCount(); i++) { - String column = metaData.getColumnLabel(i); - String value = resultSet.getString(i); - System.out.print(column + ":" + value + "\t"); - } - System.out.println(); - } - statement.close(); - connection.close(); - } catch (SQLException | ClassNotFoundException e) { - e.printStackTrace(); - } + public void connect() { + } @Test - public void testAcceptUrl() throws SQLException { + public void acceptsURL() throws SQLException { Driver driver = new RestfulDriver(); boolean isAccept = driver.acceptsURL("jdbc:TAOS-RS://master:6041"); Assert.assertTrue(isAccept); + isAccept = driver.acceptsURL("jdbc:TAOS://master:6041"); + Assert.assertFalse(isAccept); } + @Test + public void getPropertyInfo() throws SQLException { + Driver driver = new RestfulDriver(); + final String url = ""; + DriverPropertyInfo[] propertyInfo = driver.getPropertyInfo(url, null); + } + + @Test + public void getMajorVersion() { + Assert.assertEquals(2, new RestfulDriver().getMajorVersion()); + } + + @Test + public void getMinorVersion() { + Assert.assertEquals(0, new RestfulDriver().getMinorVersion()); + } + + @Test + public void jdbcCompliant() { + Assert.assertFalse(new RestfulDriver().jdbcCompliant()); + } + + @Test(expected = SQLFeatureNotSupportedException.class) + public void getParentLogger() throws SQLFeatureNotSupportedException { + new RestfulDriver().getParentLogger(); + } } diff --git a/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulJDBCTest.java b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulJDBCTest.java new file mode 100644 index 0000000000..d13475b96d --- /dev/null +++ b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/rs/RestfulJDBCTest.java @@ -0,0 +1,108 @@ +package com.taosdata.jdbc.rs; + +import org.junit.*; +import org.junit.runners.MethodSorters; + +import java.sql.*; +import java.util.Random; + +@FixMethodOrder(MethodSorters.NAME_ASCENDING) +public class RestfulJDBCTest { + + private Connection connection; + + @Before + public void before() throws ClassNotFoundException, SQLException { + Class.forName("com.taosdata.jdbc.rs.RestfulDriver"); + connection = DriverManager.getConnection("jdbc:TAOS-RS://master:6041/restful_test?user=root&password=taosdata"); + } + + @After + public void after() throws SQLException { + if (connection != null) + connection.close(); + } + + + /** + * 查询所有log.log + **/ + @Test + public void testCase001() { + try { + Statement statement = connection.createStatement(); + ResultSet resultSet = statement.executeQuery("select * from log.log"); + ResultSetMetaData metaData = resultSet.getMetaData(); + while (resultSet.next()) { + for (int i = 1; i <= metaData.getColumnCount(); i++) { + String column = metaData.getColumnLabel(i); + String value = resultSet.getString(i); + System.out.print(column + ":" + value + "\t"); + } + System.out.println(); + } + statement.close(); + } catch (SQLException e) { + e.printStackTrace(); + } + } + + /** + * create database + */ + @Test + public void testCase002() { + try (Statement stmt = connection.createStatement()) { + stmt.execute("drop database if exists restful_test"); + stmt.execute("create database if not exists restful_test"); + stmt.execute("use restful_test"); + } catch (SQLException e) { + e.printStackTrace(); + } + } + + /** + * create super table + ***/ + @Test + public void testCase003() { + try (Statement stmt = connection.createStatement()) { + stmt.execute("create table weather(ts timestamp, temperature float, humidity int) tags(location nchar(64), groupId int)"); + } catch (SQLException e) { + e.printStackTrace(); + } + } + + @Test + public void testCase004() { + try (Statement stmt = connection.createStatement()) { + for (int i = 1; i <= 100; i++) { + stmt.execute("create table t" + i + " using weather tags('beijing', '" + i + "')"); + } + } catch (SQLException e) { + e.printStackTrace(); + } + } + + private Random random = new Random(System.currentTimeMillis()); + + @Test + public void testCase005() { + try (Statement stmt = connection.createStatement()) { + int rows = 0; + for (int i = 0; i < 10; i++) { + for (int j = 1; j <= 100; j++) { + long currentTimeMillis = System.currentTimeMillis(); + int affectRows = stmt.executeUpdate("insert into t" + j + " values(" + currentTimeMillis + "," + (random.nextFloat() * 50) + "," + random.nextInt(100) + ")"); + Assert.assertEquals(1, affectRows); + rows += affectRows; + } + } + Assert.assertEquals(1000, rows); + } catch (SQLException e) { + e.printStackTrace(); + } + } + + +} diff --git a/src/connector/jdbc/src/test/java/com/taosdata/jdbc/utils/SqlSyntaxValidatorTest.java b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/utils/SqlSyntaxValidatorTest.java new file mode 100644 index 0000000000..fb570e16c4 --- /dev/null +++ b/src/connector/jdbc/src/test/java/com/taosdata/jdbc/utils/SqlSyntaxValidatorTest.java @@ -0,0 +1,28 @@ +package com.taosdata.jdbc.utils; + +import org.junit.Assert; +import org.junit.Test; + +public class SqlSyntaxValidatorTest { + + @Test + public void validateSqlSyntax() { + } + + @Test + public void isSelectSQL() { + Assert.assertTrue(SqlSyntaxValidator.isSelectSql("select * from test.weather")); + Assert.assertTrue(SqlSyntaxValidator.isSelectSql(" select * from test.weather")); + Assert.assertTrue(SqlSyntaxValidator.isSelectSql(" select * from test.weather ")); + Assert.assertFalse(SqlSyntaxValidator.isSelectSql("insert into test.weather values(now, 1.1, 2)")); + } + + @Test + public void isUseSQL() { + Assert.assertTrue(SqlSyntaxValidator.isUseSql("use database test")); + Assert.assertTrue(SqlSyntaxValidator.isUseSql("create database test")); + Assert.assertTrue(SqlSyntaxValidator.isUseSql("create database if not exist test")); + Assert.assertTrue(SqlSyntaxValidator.isUseSql("drop database test")); + Assert.assertTrue(SqlSyntaxValidator.isUseSql("drop database if exist test")); + } +} \ No newline at end of file diff --git a/src/connector/python/linux/python2/setup.py b/src/connector/python/linux/python2/setup.py index 2e4f80b8f0..b3daa98bdc 100644 --- a/src/connector/python/linux/python2/setup.py +++ b/src/connector/python/linux/python2/setup.py @@ -5,7 +5,7 @@ with open("README.md", "r") as fh: setuptools.setup( name="taos", - version="2.0.0", + version="2.0.2", author="Taosdata Inc.", author_email="support@taosdata.com", description="TDengine python client package", diff --git a/src/connector/python/linux/python2/taos/cinterface.py b/src/connector/python/linux/python2/taos/cinterface.py index 269326535c..6f0435722f 100644 --- a/src/connector/python/linux/python2/taos/cinterface.py +++ b/src/connector/python/linux/python2/taos/cinterface.py @@ -18,7 +18,7 @@ def _crow_timestamp_to_python(data, num_of_rows, nbytes=None, micro=False): _timestamp_converter = _convert_microsecond_to_datetime if num_of_rows > 0: - return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)][::1])) + return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)])) else: return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)])) @@ -26,7 +26,7 @@ def _crow_bool_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bool row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_bool))[:abs(num_of_rows)] ] @@ -34,7 +34,7 @@ def _crow_tinyint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C tinyint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] @@ -42,7 +42,7 @@ def _crow_smallint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C smallint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)][::1]] + return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)]] else: return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)] ] @@ -50,7 +50,7 @@ def _crow_int_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C int row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] @@ -58,7 +58,7 @@ def _crow_bigint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bigint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)] ] @@ -66,7 +66,7 @@ def _crow_float_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C float row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] @@ -74,7 +74,7 @@ def _crow_double_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C double row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] @@ -82,7 +82,7 @@ def _crow_binary_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C binary row to python row """ if num_of_rows > 0: - return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)][::1]] + return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] else: return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] @@ -90,9 +90,7 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C nchar row to python row """ assert(nbytes is not None) - - res = [] - + res=[] for i in range(abs(num_of_rows)): try: if num_of_rows >= 0: @@ -103,17 +101,49 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): except ValueError: res.append(None) + return res + +def _crow_binary_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C binary row to python row + """ + res=[] + if num_of_rows > 0: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + return res + +def _crow_nchar_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C nchar row to python row + """ + assert(nbytes is not None) + res=[] + if num_of_rows >= 0: + for i in range(abs(num_of_rows)): + try: + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode() ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + res.append( (ctypes.cast(data+nbytes*i+2, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) + except ValueError: + res.append(None) return res - # if num_of_rows > 0: - # for i in range(abs(num_of_rows)): - # try: - # res.append( (ctypes.cast(data+nbytes*i, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) - # except ValueError: - # res.append(None) - # return res - # # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)][::-1]] - # else: - # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)]] _CONVERT_FUNC = { FieldType.C_BOOL: _crow_bool_to_python, @@ -128,6 +158,19 @@ _CONVERT_FUNC = { FieldType.C_NCHAR : _crow_nchar_to_python } +_CONVERT_FUNC_BLOCK = { + FieldType.C_BOOL: _crow_bool_to_python, + FieldType.C_TINYINT : _crow_tinyint_to_python, + FieldType.C_SMALLINT : _crow_smallint_to_python, + FieldType.C_INT : _crow_int_to_python, + FieldType.C_BIGINT : _crow_bigint_to_python, + FieldType.C_FLOAT : _crow_float_to_python, + FieldType.C_DOUBLE : _crow_double_to_python, + FieldType.C_BINARY: _crow_binary_to_python_block, + FieldType.C_TIMESTAMP : _crow_timestamp_to_python, + FieldType.C_NCHAR : _crow_nchar_to_python_block +} + # Corresponding TAOS_FIELD structure in C class TaosField(ctypes.Structure): _fields_ = [('name', ctypes.c_char * 65), @@ -227,8 +270,8 @@ class CTaosInterface(object): print('connect to TDengine failed') raise ConnectionError("connect to TDengine failed") # sys.exit(1) - else: - print('connect to TDengine success') + #else: + # print('connect to TDengine success') return connection @@ -237,7 +280,7 @@ class CTaosInterface(object): '''Close the TDengine handle ''' CTaosInterface.libtaos.taos_close(connection) - print('connection is closed') + #print('connection is closed') @staticmethod def query(connection, sql): @@ -310,6 +353,24 @@ class CTaosInterface(object): @staticmethod def fetchBlock(result, fields): + pblock = ctypes.c_void_p(0) + num_of_rows = CTaosInterface.libtaos.taos_fetch_block( + result, ctypes.byref(pblock)) + if num_of_rows == 0: + return None, 0 + isMicro = (CTaosInterface.libtaos.taos_result_precision(result) == FieldType.C_TIMESTAMP_MICRO) + blocks = [None] * len(fields) + fieldL = CTaosInterface.libtaos.taos_fetch_lengths(result) + fieldLen = [ele for ele in ctypes.cast(fieldL, ctypes.POINTER(ctypes.c_int))[:len(fields)]] + for i in range(len(fields)): + data = ctypes.cast(pblock, ctypes.POINTER(ctypes.c_void_p))[i] + if fields[i]['type'] not in _CONVERT_FUNC_BLOCK: + raise DatabaseError("Invalid data type returned from database") + blocks[i] = _CONVERT_FUNC_BLOCK[fields[i]['type']](data, num_of_rows, fieldLen[i], isMicro) + + return blocks, abs(num_of_rows) + @staticmethod + def fetchRow(result, fields): pblock = ctypes.c_void_p(0) pblock = CTaosInterface.libtaos.taos_fetch_row(result) if pblock : @@ -393,7 +454,7 @@ class CTaosInterface(object): def errStr(result): """Return the error styring """ - return CTaosInterface.libtaos.taos_errstr(result) + return CTaosInterface.libtaos.taos_errstr(result).decode('utf-8') if __name__ == '__main__': @@ -410,4 +471,4 @@ if __name__ == '__main__': print(data) cinter.freeResult(result) - cinter.close(conn) \ No newline at end of file + cinter.close(conn) diff --git a/src/connector/python/linux/python2/taos/cursor.py b/src/connector/python/linux/python2/taos/cursor.py index bc6670ca77..eee5e39488 100644 --- a/src/connector/python/linux/python2/taos/cursor.py +++ b/src/connector/python/linux/python2/taos/cursor.py @@ -49,7 +49,7 @@ class TDengineCursor(object): raise OperationalError("Invalid use of fetch iterator") if self._block_rows <= self._block_iter: - block, self._block_rows = CTaosInterface.fetchBlock( + block, self._block_rows = CTaosInterface.fetchRow( self._result, self._fields) if self._block_rows == 0: raise StopIteration @@ -190,6 +190,23 @@ class TDengineCursor(object): if self._result is None or self._fields is None: raise OperationalError("Invalid use of fetchall") + buffer = [[] for i in range(len(self._fields))] + self._rowcount = 0 + while True: + block, num_of_fields = CTaosInterface.fetchRow(self._result, self._fields) + errno = CTaosInterface.libtaos.taos_errno(self._result) + if errno != 0: + raise ProgrammingError(CTaosInterface.errStr(self._result), errno) + if num_of_fields == 0: + break + self._rowcount += num_of_fields + for i in range(len(self._fields)): + buffer[i].extend(block[i]) + return list(map(tuple, zip(*buffer))) + def fetchall_block(self): + if self._result is None or self._fields is None: + raise OperationalError("Invalid use of fetchall") + buffer = [[] for i in range(len(self._fields))] self._rowcount = 0 while True: @@ -203,7 +220,6 @@ class TDengineCursor(object): for i in range(len(self._fields)): buffer[i].extend(block[i]) return list(map(tuple, zip(*buffer))) - def nextset(self): """ """ diff --git a/src/connector/python/linux/python3/setup.py b/src/connector/python/linux/python3/setup.py index 03a49fc1c5..f49ebe2b6d 100644 --- a/src/connector/python/linux/python3/setup.py +++ b/src/connector/python/linux/python3/setup.py @@ -5,7 +5,7 @@ with open("README.md", "r") as fh: setuptools.setup( name="taos", - version="2.0.0", + version="2.0.2", author="Taosdata Inc.", author_email="support@taosdata.com", description="TDengine python client package", diff --git a/src/connector/python/linux/python3/taos/cinterface.py b/src/connector/python/linux/python3/taos/cinterface.py index be5b99d8c1..bbf5a0c714 100644 --- a/src/connector/python/linux/python3/taos/cinterface.py +++ b/src/connector/python/linux/python3/taos/cinterface.py @@ -18,7 +18,7 @@ def _crow_timestamp_to_python(data, num_of_rows, nbytes=None, micro=False): _timestamp_converter = _convert_microsecond_to_datetime if num_of_rows > 0: - return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)][::1])) + return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)])) else: return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)])) @@ -26,7 +26,7 @@ def _crow_bool_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bool row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_bool))[:abs(num_of_rows)] ] @@ -34,7 +34,7 @@ def _crow_tinyint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C tinyint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] @@ -42,7 +42,7 @@ def _crow_smallint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C smallint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)][::1]] + return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)]] else: return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)] ] @@ -50,7 +50,7 @@ def _crow_int_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C int row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] @@ -58,7 +58,7 @@ def _crow_bigint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bigint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_long))[:abs(num_of_rows)] ] @@ -66,7 +66,7 @@ def _crow_float_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C float row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] @@ -74,7 +74,7 @@ def _crow_double_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C double row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] @@ -82,7 +82,7 @@ def _crow_binary_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C binary row to python row """ if num_of_rows > 0: - return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)][::1]] + return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] else: return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] @@ -90,9 +90,7 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C nchar row to python row """ assert(nbytes is not None) - - res = [] - + res=[] for i in range(abs(num_of_rows)): try: if num_of_rows >= 0: @@ -103,17 +101,49 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): except ValueError: res.append(None) + return res + +def _crow_binary_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C binary row to python row + """ + res=[] + if num_of_rows > 0: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + return res + +def _crow_nchar_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C nchar row to python row + """ + assert(nbytes is not None) + res=[] + if num_of_rows >= 0: + for i in range(abs(num_of_rows)): + try: + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode() ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + res.append( (ctypes.cast(data+nbytes*i+2, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) + except ValueError: + res.append(None) return res - # if num_of_rows > 0: - # for i in range(abs(num_of_rows)): - # try: - # res.append( (ctypes.cast(data+nbytes*i, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) - # except ValueError: - # res.append(None) - # return res - # # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)][::1]] - # else: - # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)]] _CONVERT_FUNC = { FieldType.C_BOOL: _crow_bool_to_python, @@ -128,6 +158,19 @@ _CONVERT_FUNC = { FieldType.C_NCHAR : _crow_nchar_to_python } +_CONVERT_FUNC_BLOCK = { + FieldType.C_BOOL: _crow_bool_to_python, + FieldType.C_TINYINT : _crow_tinyint_to_python, + FieldType.C_SMALLINT : _crow_smallint_to_python, + FieldType.C_INT : _crow_int_to_python, + FieldType.C_BIGINT : _crow_bigint_to_python, + FieldType.C_FLOAT : _crow_float_to_python, + FieldType.C_DOUBLE : _crow_double_to_python, + FieldType.C_BINARY: _crow_binary_to_python_block, + FieldType.C_TIMESTAMP : _crow_timestamp_to_python, + FieldType.C_NCHAR : _crow_nchar_to_python_block +} + # Corresponding TAOS_FIELD structure in C class TaosField(ctypes.Structure): _fields_ = [('name', ctypes.c_char * 65), @@ -253,7 +296,7 @@ class CTaosInterface(object): raise AttributeError("sql is expected as a string") # finally: # CTaosInterface.libtaos.close(connection) - + @staticmethod def affectedRows(result): """The affected rows after runing query @@ -308,29 +351,26 @@ class CTaosInterface(object): return fields - # @staticmethod - # def fetchBlock(result, fields): - # pblock = ctypes.c_void_p(0) - # num_of_rows = CTaosInterface.libtaos.taos_fetch_block( - # result, ctypes.byref(pblock)) - # if num_of_rows == 0: - # return None, 0 - - # isMicro = (CTaosInterface.libtaos.taos_result_precision(result) == FieldType.C_TIMESTAMP_MICRO) - # blocks = [None] * len(fields) - # fieldL = CTaosInterface.libtaos.taos_fetch_lengths(result) - # fieldLen = [ele for ele in ctypes.cast(fieldL, ctypes.POINTER(ctypes.c_int))[:len(fields)]] - # for i in range(len(fields)): - # data = ctypes.cast(pblock, ctypes.POINTER(ctypes.c_void_p))[i] - - # if fields[i]['type'] not in _CONVERT_FUNC: - # raise DatabaseError("Invalid data type returned from database") - # print('====================',fieldLen[i]) - # blocks[i] = _CONVERT_FUNC[fields[i]['type']](data, num_of_rows, fieldLen[i], isMicro) - - # return blocks, abs(num_of_rows) @staticmethod def fetchBlock(result, fields): + pblock = ctypes.c_void_p(0) + num_of_rows = CTaosInterface.libtaos.taos_fetch_block( + result, ctypes.byref(pblock)) + if num_of_rows == 0: + return None, 0 + isMicro = (CTaosInterface.libtaos.taos_result_precision(result) == FieldType.C_TIMESTAMP_MICRO) + blocks = [None] * len(fields) + fieldL = CTaosInterface.libtaos.taos_fetch_lengths(result) + fieldLen = [ele for ele in ctypes.cast(fieldL, ctypes.POINTER(ctypes.c_int))[:len(fields)]] + for i in range(len(fields)): + data = ctypes.cast(pblock, ctypes.POINTER(ctypes.c_void_p))[i] + if fields[i]['type'] not in _CONVERT_FUNC_BLOCK: + raise DatabaseError("Invalid data type returned from database") + blocks[i] = _CONVERT_FUNC_BLOCK[fields[i]['type']](data, num_of_rows, fieldLen[i], isMicro) + + return blocks, abs(num_of_rows) + @staticmethod + def fetchRow(result, fields): pblock = ctypes.c_void_p(0) pblock = CTaosInterface.libtaos.taos_fetch_row(result) if pblock : @@ -350,6 +390,7 @@ class CTaosInterface(object): else: return None, 0 return blocks, abs(num_of_rows) + @staticmethod def freeResult(result): CTaosInterface.libtaos.taos_free_result(result) diff --git a/src/connector/python/linux/python3/taos/cursor.py b/src/connector/python/linux/python3/taos/cursor.py index eb10bed485..d8184668c8 100644 --- a/src/connector/python/linux/python3/taos/cursor.py +++ b/src/connector/python/linux/python3/taos/cursor.py @@ -5,7 +5,6 @@ import threading # querySeqNum = 0 - class TDengineCursor(object): """Database cursor which is used to manage the context of a fetch operation. @@ -51,7 +50,7 @@ class TDengineCursor(object): raise OperationalError("Invalid use of fetch iterator") if self._block_rows <= self._block_iter: - block, self._block_rows = CTaosInterface.fetchBlock( + block, self._block_rows = CTaosInterface.fetchRow( self._result, self._fields) if self._block_rows == 0: raise StopIteration @@ -196,18 +195,13 @@ class TDengineCursor(object): def fetchall(self): """Fetch all (remaining) rows of a query result, returning them as a sequence of sequences (e.g. a list of tuples). Note that the cursor's arraysize attribute can affect the performance of this operation. """ - # if threading.get_ident() != self._threadId: - # info ="[WARNING] Cursor fetchall:Thread ID not match,creater:"+str(self._threadId)+" caller:"+str(threading.get_ident()) - # raise OperationalError(info) - # print(info) - # return None if self._result is None or self._fields is None: raise OperationalError("Invalid use of fetchall") buffer = [[] for i in range(len(self._fields))] self._rowcount = 0 while True: - block, num_of_fields = CTaosInterface.fetchBlock(self._result, self._fields) + block, num_of_fields = CTaosInterface.fetchRow(self._result, self._fields) errno = CTaosInterface.libtaos.taos_errno(self._result) if errno != 0: raise ProgrammingError(CTaosInterface.errStr(self._result), errno) @@ -218,6 +212,22 @@ class TDengineCursor(object): buffer[i].extend(block[i]) return list(map(tuple, zip(*buffer))) + def fetchall_block(self): + if self._result is None or self._fields is None: + raise OperationalError("Invalid use of fetchall") + + buffer = [[] for i in range(len(self._fields))] + self._rowcount = 0 + while True: + block, num_of_fields = CTaosInterface.fetchBlock(self._result, self._fields) + errno = CTaosInterface.libtaos.taos_errno(self._result) + if errno != 0: + raise ProgrammingError(CTaosInterface.errStr(self._result), errno) + if num_of_fields == 0: break + self._rowcount += num_of_fields + for i in range(len(self._fields)): + buffer[i].extend(block[i]) + return list(map(tuple, zip(*buffer))) def nextset(self): """ """ diff --git a/src/connector/python/windows/python2/setup.py b/src/connector/python/windows/python2/setup.py index fd82a55650..34cb7a04d7 100644 --- a/src/connector/python/windows/python2/setup.py +++ b/src/connector/python/windows/python2/setup.py @@ -5,7 +5,7 @@ with open("README.md", "r") as fh: setuptools.setup( name="taos", - version="2.0.0", + version="2.0.2", author="Taosdata Inc.", author_email="support@taosdata.com", description="TDengine python client package", diff --git a/src/connector/python/windows/python2/taos/cinterface.py b/src/connector/python/windows/python2/taos/cinterface.py index 084d38e41c..d4bf1a0350 100644 --- a/src/connector/python/windows/python2/taos/cinterface.py +++ b/src/connector/python/windows/python2/taos/cinterface.py @@ -18,7 +18,7 @@ def _crow_timestamp_to_python(data, num_of_rows, nbytes=None, micro=False): _timestamp_converter = _convert_microsecond_to_datetime if num_of_rows > 0: - return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)][::1])) + return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)])) else: return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)])) @@ -26,7 +26,7 @@ def _crow_bool_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bool row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_bool))[:abs(num_of_rows)] ] @@ -34,7 +34,7 @@ def _crow_tinyint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C tinyint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] @@ -42,7 +42,7 @@ def _crow_smallint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C smallint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)][::1]] + return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)]] else: return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)] ] @@ -50,7 +50,7 @@ def _crow_int_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C int row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] @@ -58,7 +58,7 @@ def _crow_bigint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bigint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)] ] @@ -66,7 +66,7 @@ def _crow_float_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C float row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] @@ -74,7 +74,7 @@ def _crow_double_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C double row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] @@ -82,7 +82,7 @@ def _crow_binary_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C binary row to python row """ if num_of_rows > 0: - return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)][::1]] + return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] else: return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] @@ -90,9 +90,7 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C nchar row to python row """ assert(nbytes is not None) - - res = [] - + res=[] for i in range(abs(num_of_rows)): try: if num_of_rows >= 0: @@ -103,17 +101,49 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): except ValueError: res.append(None) + return res + +def _crow_binary_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C binary row to python row + """ + res=[] + if num_of_rows > 0: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + return res + +def _crow_nchar_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C nchar row to python row + """ + assert(nbytes is not None) + res=[] + if num_of_rows >= 0: + for i in range(abs(num_of_rows)): + try: + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode() ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + res.append( (ctypes.cast(data+nbytes*i+2, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) + except ValueError: + res.append(None) return res - # if num_of_rows > 0: - # for i in range(abs(num_of_rows)): - # try: - # res.append( (ctypes.cast(data+nbytes*i, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) - # except ValueError: - # res.append(None) - # return res - # # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)][::-1]] - # else: - # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)]] _CONVERT_FUNC = { FieldType.C_BOOL: _crow_bool_to_python, @@ -128,6 +158,19 @@ _CONVERT_FUNC = { FieldType.C_NCHAR : _crow_nchar_to_python } +_CONVERT_FUNC_BLOCK = { + FieldType.C_BOOL: _crow_bool_to_python, + FieldType.C_TINYINT : _crow_tinyint_to_python, + FieldType.C_SMALLINT : _crow_smallint_to_python, + FieldType.C_INT : _crow_int_to_python, + FieldType.C_BIGINT : _crow_bigint_to_python, + FieldType.C_FLOAT : _crow_float_to_python, + FieldType.C_DOUBLE : _crow_double_to_python, + FieldType.C_BINARY: _crow_binary_to_python_block, + FieldType.C_TIMESTAMP : _crow_timestamp_to_python, + FieldType.C_NCHAR : _crow_nchar_to_python_block +} + # Corresponding TAOS_FIELD structure in C class TaosField(ctypes.Structure): _fields_ = [('name', ctypes.c_char * 65), @@ -225,9 +268,10 @@ class CTaosInterface(object): if connection.value == None: print('connect to TDengine failed') + raise ConnectionError("connect to TDengine failed") # sys.exit(1) - else: - print('connect to TDengine success') + #else: + # print('connect to TDengine success') return connection @@ -236,7 +280,7 @@ class CTaosInterface(object): '''Close the TDengine handle ''' CTaosInterface.libtaos.taos_close(connection) - print('connection is closed') + #print('connection is closed') @staticmethod def query(connection, sql): @@ -252,7 +296,7 @@ class CTaosInterface(object): raise AttributeError("sql is expected as a string") # finally: # CTaosInterface.libtaos.close(connection) - + @staticmethod def affectedRows(result): """The affected rows after runing query @@ -309,6 +353,24 @@ class CTaosInterface(object): @staticmethod def fetchBlock(result, fields): + pblock = ctypes.c_void_p(0) + num_of_rows = CTaosInterface.libtaos.taos_fetch_block( + result, ctypes.byref(pblock)) + if num_of_rows == 0: + return None, 0 + isMicro = (CTaosInterface.libtaos.taos_result_precision(result) == FieldType.C_TIMESTAMP_MICRO) + blocks = [None] * len(fields) + fieldL = CTaosInterface.libtaos.taos_fetch_lengths(result) + fieldLen = [ele for ele in ctypes.cast(fieldL, ctypes.POINTER(ctypes.c_int))[:len(fields)]] + for i in range(len(fields)): + data = ctypes.cast(pblock, ctypes.POINTER(ctypes.c_void_p))[i] + if fields[i]['type'] not in _CONVERT_FUNC_BLOCK: + raise DatabaseError("Invalid data type returned from database") + blocks[i] = _CONVERT_FUNC_BLOCK[fields[i]['type']](data, num_of_rows, fieldLen[i], isMicro) + + return blocks, abs(num_of_rows) + @staticmethod + def fetchRow(result, fields): pblock = ctypes.c_void_p(0) pblock = CTaosInterface.libtaos.taos_fetch_row(result) if pblock : @@ -393,7 +455,7 @@ class CTaosInterface(object): def errStr(result): """Return the error styring """ - return CTaosInterface.libtaos.taos_errstr(result) + return CTaosInterface.libtaos.taos_errstr(result).decode('utf-8') if __name__ == '__main__': @@ -410,4 +472,4 @@ if __name__ == '__main__': print(data) cinter.freeResult(result) - cinter.close(conn) \ No newline at end of file + cinter.close(conn) diff --git a/src/connector/python/windows/python2/taos/cursor.py b/src/connector/python/windows/python2/taos/cursor.py index 35846cbe11..71651afee1 100644 --- a/src/connector/python/windows/python2/taos/cursor.py +++ b/src/connector/python/windows/python2/taos/cursor.py @@ -50,7 +50,7 @@ class TDengineCursor(object): raise OperationalError("Invalid use of fetch iterator") if self._block_rows <= self._block_iter: - block, self._block_rows = CTaosInterface.fetchBlock(self._result, self._fields) + block, self._block_rows = CTaosInterface.fetchRow(self._result, self._fields) if self._block_rows == 0: raise StopIteration self._block = list(map(tuple, zip(*block))) @@ -143,7 +143,25 @@ class TDengineCursor(object): """ if self._result is None or self._fields is None: raise OperationalError("Invalid use of fetchall") + + buffer = [[] for i in range(len(self._fields))] + self._rowcount = 0 + while True: + block, num_of_fields = CTaosInterface.fetchRow(self._result, self._fields) + errno = CTaosInterface.libtaos.taos_errno(self._result) + if errno != 0: + raise ProgrammingError(CTaosInterface.errStr(self._result), errno) + if num_of_fields == 0: + break + self._rowcount += num_of_fields + for i in range(len(self._fields)): + buffer[i].extend(block[i]) + return list(map(tuple, zip(*buffer))) + def fetchall_block(self): + if self._result is None or self._fields is None: + raise OperationalError("Invalid use of fetchall") + buffer = [[] for i in range(len(self._fields))] self._rowcount = 0 while True: @@ -178,7 +196,7 @@ class TDengineCursor(object): self._description = None self._rowcount = -1 if self._result is not None: - CTaosInterface.freeResult(self._result) + CTaosInterface.freeResult(self._result) self._result = None self._fields = None self._block = None diff --git a/src/connector/python/windows/python3/setup.py b/src/connector/python/windows/python3/setup.py index 9abdace5a9..c09644d330 100644 --- a/src/connector/python/windows/python3/setup.py +++ b/src/connector/python/windows/python3/setup.py @@ -5,7 +5,7 @@ with open("README.md", "r") as fh: setuptools.setup( name="taos", - version="2.0.0", + version="2.0.2", author="Taosdata Inc.", author_email="support@taosdata.com", description="TDengine python client package", diff --git a/src/connector/python/windows/python3/taos/cinterface.py b/src/connector/python/windows/python3/taos/cinterface.py index 68adf191c9..fb1b65fc15 100644 --- a/src/connector/python/windows/python3/taos/cinterface.py +++ b/src/connector/python/windows/python3/taos/cinterface.py @@ -18,7 +18,7 @@ def _crow_timestamp_to_python(data, num_of_rows, nbytes=None, micro=False): _timestamp_converter = _convert_microsecond_to_datetime if num_of_rows > 0: - return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)][::1])) + return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)])) else: return list(map(_timestamp_converter, ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)])) @@ -26,7 +26,7 @@ def _crow_bool_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bool row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BOOL_NULL else bool(ele) for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_bool))[:abs(num_of_rows)] ] @@ -34,7 +34,7 @@ def _crow_tinyint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C tinyint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_TINYINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_byte))[:abs(num_of_rows)] ] @@ -42,7 +42,7 @@ def _crow_smallint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C smallint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)][::1]] + return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)]] else: return [ None if ele == FieldType.C_SMALLINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_short))[:abs(num_of_rows)] ] @@ -50,7 +50,7 @@ def _crow_int_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C int row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_INT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_int))[:abs(num_of_rows)] ] @@ -58,7 +58,7 @@ def _crow_bigint_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C bigint row to python row """ if num_of_rows > 0: - return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)][::1] ] + return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)] ] else: return [ None if ele == FieldType.C_BIGINT_NULL else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_longlong))[:abs(num_of_rows)] ] @@ -66,7 +66,7 @@ def _crow_float_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C float row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_float))[:abs(num_of_rows)] ] @@ -74,7 +74,7 @@ def _crow_double_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C double row to python row """ if num_of_rows > 0: - return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)][::1] ] + return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] else: return [ None if math.isnan(ele) else ele for ele in ctypes.cast(data, ctypes.POINTER(ctypes.c_double))[:abs(num_of_rows)] ] @@ -82,7 +82,7 @@ def _crow_binary_to_python(data, num_of_rows, nbytes=None, micro=False): """Function to convert C binary row to python row """ if num_of_rows > 0: - return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)][::1]] + return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] else: return [ None if ele.value[0:1] == FieldType.C_BINARY_NULL else ele.value.decode('utf-8') for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_char * nbytes)))[:abs(num_of_rows)]] @@ -104,16 +104,48 @@ def _crow_nchar_to_python(data, num_of_rows, nbytes=None, micro=False): res.append(None) return res - # if num_of_rows > 0: - # for i in range(abs(num_of_rows)): - # try: - # res.append( (ctypes.cast(data+nbytes*i, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) - # except ValueError: - # res.append(None) - # return res - # # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)][::-1]] - # else: - # return [ele.value for ele in (ctypes.cast(data, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[:abs(num_of_rows)]] + +def _crow_binary_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C binary row to python row + """ + res=[] + if num_of_rows > 0: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + rbyte=ctypes.cast(data+nbytes*i,ctypes.POINTER(ctypes.c_short))[:1].pop() + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode()[0:rbyte] ) + except ValueError: + res.append(None) + return res + +def _crow_nchar_to_python_block(data, num_of_rows, nbytes=None, micro=False): + """Function to convert C nchar row to python row + """ + assert(nbytes is not None) + res=[] + if num_of_rows >= 0: + for i in range(abs(num_of_rows)): + try: + tmpstr = ctypes.c_char_p(data+nbytes*i+2) + res.append( tmpstr.value.decode() ) + except ValueError: + res.append(None) + else: + for i in range(abs(num_of_rows)): + try: + res.append( (ctypes.cast(data+nbytes*i+2, ctypes.POINTER(ctypes.c_wchar * (nbytes//4))))[0].value ) + except ValueError: + res.append(None) + return res _CONVERT_FUNC = { FieldType.C_BOOL: _crow_bool_to_python, @@ -128,6 +160,19 @@ _CONVERT_FUNC = { FieldType.C_NCHAR : _crow_nchar_to_python } +_CONVERT_FUNC_BLOCK = { + FieldType.C_BOOL: _crow_bool_to_python, + FieldType.C_TINYINT : _crow_tinyint_to_python, + FieldType.C_SMALLINT : _crow_smallint_to_python, + FieldType.C_INT : _crow_int_to_python, + FieldType.C_BIGINT : _crow_bigint_to_python, + FieldType.C_FLOAT : _crow_float_to_python, + FieldType.C_DOUBLE : _crow_double_to_python, + FieldType.C_BINARY: _crow_binary_to_python_block, + FieldType.C_TIMESTAMP : _crow_timestamp_to_python, + FieldType.C_NCHAR : _crow_nchar_to_python_block +} + # Corresponding TAOS_FIELD structure in C class TaosField(ctypes.Structure): _fields_ = [('name', ctypes.c_char * 65), @@ -227,8 +272,8 @@ class CTaosInterface(object): print('connect to TDengine failed') raise ConnectionError("connect to TDengine failed") # sys.exit(1) - else: - print('connect to TDengine success') + #else: + # print('connect to TDengine success') return connection @@ -237,7 +282,7 @@ class CTaosInterface(object): '''Close the TDengine handle ''' CTaosInterface.libtaos.taos_close(connection) - print('connection is closed') + #print('connection is closed') @staticmethod def query(connection, sql): @@ -310,6 +355,24 @@ class CTaosInterface(object): @staticmethod def fetchBlock(result, fields): + pblock = ctypes.c_void_p(0) + num_of_rows = CTaosInterface.libtaos.taos_fetch_block( + result, ctypes.byref(pblock)) + if num_of_rows == 0: + return None, 0 + isMicro = (CTaosInterface.libtaos.taos_result_precision(result) == FieldType.C_TIMESTAMP_MICRO) + blocks = [None] * len(fields) + fieldL = CTaosInterface.libtaos.taos_fetch_lengths(result) + fieldLen = [ele for ele in ctypes.cast(fieldL, ctypes.POINTER(ctypes.c_int))[:len(fields)]] + for i in range(len(fields)): + data = ctypes.cast(pblock, ctypes.POINTER(ctypes.c_void_p))[i] + if fields[i]['type'] not in _CONVERT_FUNC_BLOCK: + raise DatabaseError("Invalid data type returned from database") + blocks[i] = _CONVERT_FUNC_BLOCK[fields[i]['type']](data, num_of_rows, fieldLen[i], isMicro) + + return blocks, abs(num_of_rows) + @staticmethod + def fetchRow(result, fields): pblock = ctypes.c_void_p(0) pblock = CTaosInterface.libtaos.taos_fetch_row(result) if pblock : diff --git a/src/connector/python/windows/python3/taos/cursor.py b/src/connector/python/windows/python3/taos/cursor.py index b58b494cad..b813bba357 100644 --- a/src/connector/python/windows/python3/taos/cursor.py +++ b/src/connector/python/windows/python3/taos/cursor.py @@ -51,7 +51,7 @@ class TDengineCursor(object): raise OperationalError("Invalid use of fetch iterator") if self._block_rows <= self._block_iter: - block, self._block_rows = CTaosInterface.fetchBlock(self._result, self._fields) + block, self._block_rows = CTaosInterface.fetchRow(self._result, self._fields) if self._block_rows == 0: raise StopIteration self._block = list(map(tuple, zip(*block))) @@ -144,7 +144,25 @@ class TDengineCursor(object): """ if self._result is None or self._fields is None: raise OperationalError("Invalid use of fetchall") - + + buffer = [[] for i in range(len(self._fields))] + self._rowcount = 0 + while True: + block, num_of_fields = CTaosInterface.fetchRow(self._result, self._fields) + errno = CTaosInterface.libtaos.taos_errno(self._result) + if errno != 0: + raise ProgrammingError(CTaosInterface.errStr(self._result), errno) + if num_of_fields == 0: + break + self._rowcount += num_of_fields + for i in range(len(self._fields)): + buffer[i].extend(block[i]) + return list(map(tuple, zip(*buffer))) + + def fetchall_block(self): + if self._result is None or self._fields is None: + raise OperationalError("Invalid use of fetchall") + buffer = [[] for i in range(len(self._fields))] self._rowcount = 0 while True: diff --git a/src/dnode/src/dnodeEps.c b/src/dnode/src/dnodeEps.c index 9c90c39181..83f294e05e 100644 --- a/src/dnode/src/dnodeEps.c +++ b/src/dnode/src/dnodeEps.c @@ -33,7 +33,7 @@ static void dnodePrintEps(SDnodeEps *eps); int32_t dnodeInitEps() { pthread_mutex_init(&tsEpsMutex, NULL); - tsEpsHash = taosHashInit(4, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, true); + tsEpsHash = taosHashInit(4, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_ENTRY_LOCK); dnodeResetEps(NULL); int32_t ret = dnodeReadEps(); if (ret == 0) { diff --git a/src/dnode/src/dnodeMPeer.c b/src/dnode/src/dnodeMPeer.c index 05b37bd338..ee6dc5212e 100644 --- a/src/dnode/src/dnodeMPeer.c +++ b/src/dnode/src/dnodeMPeer.c @@ -162,7 +162,7 @@ static void *dnodeProcessMPeerQueue(void *param) { break; } - dDebug("msg:%s will be processed in mpeer queue", taosMsg[pPeerMsg->rpcMsg.msgType]); + dTrace("msg:%s will be processed in mpeer queue", taosMsg[pPeerMsg->rpcMsg.msgType]); int32_t code = mnodeProcessPeerReq(pPeerMsg); dnodeSendRpcMPeerRsp(pPeerMsg, code); } diff --git a/src/dnode/src/dnodeMRead.c b/src/dnode/src/dnodeMRead.c index c14c7a8158..65f3af7b3b 100644 --- a/src/dnode/src/dnodeMRead.c +++ b/src/dnode/src/dnodeMRead.c @@ -168,7 +168,8 @@ static void *dnodeProcessMReadQueue(void *param) { break; } - dDebug("%p, msg:%s will be processed in mread queue", pRead->rpcMsg.ahandle, taosMsg[pRead->rpcMsg.msgType]); + dTrace("msg:%p, app:%p type:%s will be processed in mread queue", pRead->rpcMsg.ahandle, pRead, + taosMsg[pRead->rpcMsg.msgType]); int32_t code = mnodeProcessRead(pRead); dnodeSendRpcMReadRsp(pRead, code); } diff --git a/src/dnode/src/dnodeMWrite.c b/src/dnode/src/dnodeMWrite.c index bde4b95bc6..ef2d49ef42 100644 --- a/src/dnode/src/dnodeMWrite.c +++ b/src/dnode/src/dnodeMWrite.c @@ -127,7 +127,7 @@ void dnodeDispatchToMWriteQueue(SRpcMsg *pMsg) { dnodeSendRedirectMsg(pMsg, true); } else { SMnodeMsg *pWrite = mnodeCreateMsg(pMsg); - dDebug("app:%p:%p, msg:%s is put into mwrite queue:%p", pWrite->rpcMsg.ahandle, pWrite, + dDebug("msg:%p, app:%p type:%s is put into mwrite queue:%p", pWrite, pWrite->rpcMsg.ahandle, taosMsg[pWrite->rpcMsg.msgType], tsMWriteQueue); taosWriteQitem(tsMWriteQueue, TAOS_QTYPE_RPC, pWrite); } @@ -136,7 +136,7 @@ void dnodeDispatchToMWriteQueue(SRpcMsg *pMsg) { } static void dnodeFreeMWriteMsg(SMnodeMsg *pWrite) { - dDebug("app:%p:%p, msg:%s is freed from mwrite queue:%p", pWrite->rpcMsg.ahandle, pWrite, + dDebug("msg:%p, app:%p type:%s is freed from mwrite queue:%p", pWrite, pWrite->rpcMsg.ahandle, taosMsg[pWrite->rpcMsg.msgType], tsMWriteQueue); mnodeCleanupMsg(pWrite); @@ -174,7 +174,7 @@ static void *dnodeProcessMWriteQueue(void *param) { break; } - dDebug("app:%p:%p, msg:%s will be processed in mwrite queue", pWrite->rpcMsg.ahandle, pWrite, + dTrace("msg:%p, app:%p type:%s will be processed in mwrite queue", pWrite, pWrite->rpcMsg.ahandle, taosMsg[pWrite->rpcMsg.msgType]); int32_t code = mnodeProcessWrite(pWrite); @@ -188,13 +188,13 @@ void dnodeReprocessMWriteMsg(void *pMsg) { SMnodeMsg *pWrite = pMsg; if (!mnodeIsRunning() || tsMWriteQueue == NULL) { - dDebug("app:%p:%p, msg:%s is redirected for mnode not running, retry times:%d", pWrite->rpcMsg.ahandle, pWrite, + dDebug("msg:%p, app:%p type:%s is redirected for mnode not running, retry times:%d", pWrite, pWrite->rpcMsg.ahandle, taosMsg[pWrite->rpcMsg.msgType], pWrite->retry); dnodeSendRedirectMsg(pMsg, true); dnodeFreeMWriteMsg(pWrite); } else { - dDebug("app:%p:%p, msg:%s is reput into mwrite queue:%p, retry times:%d", pWrite->rpcMsg.ahandle, pWrite, + dDebug("msg:%p, app:%p type:%s is reput into mwrite queue:%p, retry times:%d", pWrite, pWrite->rpcMsg.ahandle, taosMsg[pWrite->rpcMsg.msgType], tsMWriteQueue, pWrite->retry); taosWriteQitem(tsMWriteQueue, TAOS_QTYPE_RPC, pWrite); diff --git a/src/dnode/src/dnodeMain.c b/src/dnode/src/dnodeMain.c index 130be0af20..9f52dbd331 100644 --- a/src/dnode/src/dnodeMain.c +++ b/src/dnode/src/dnodeMain.c @@ -19,6 +19,7 @@ #include "tutil.h" #include "tconfig.h" #include "tglobal.h" +#include "tfile.h" #include "twal.h" #include "trpc.h" #include "dnode.h" @@ -55,6 +56,7 @@ typedef struct { } SDnodeComponent; static const SDnodeComponent tsDnodeComponents[] = { + {"tfile", tfInit, tfCleanup}, {"rpc", rpcInit, rpcCleanup}, {"storage", dnodeInitStorage, dnodeCleanupStorage}, {"dnodecfg", dnodeInitCfg, dnodeCleanupCfg}, diff --git a/src/dnode/src/dnodeMgmt.c b/src/dnode/src/dnodeMgmt.c index dcb48f7833..da1852e05e 100644 --- a/src/dnode/src/dnodeMgmt.c +++ b/src/dnode/src/dnodeMgmt.c @@ -216,7 +216,7 @@ static void *dnodeProcessMgmtQueue(void *param) { } pMsg = &pMgmt->rpcMsg; - dDebug("%p, msg:%p:%s will be processed", pMsg->ahandle, pMgmt, taosMsg[pMsg->msgType]); + dDebug("msg:%p, ahandle:%p type:%s will be processed", pMgmt, pMsg->ahandle, taosMsg[pMsg->msgType]); if (dnodeProcessMgmtMsgFp[pMsg->msgType]) { rsp.code = (*dnodeProcessMgmtMsgFp[pMsg->msgType])(pMsg); } else { diff --git a/src/dnode/src/dnodePeer.c b/src/dnode/src/dnodePeer.c index 4c44924cd0..6bf22cee4e 100644 --- a/src/dnode/src/dnodePeer.c +++ b/src/dnode/src/dnodePeer.c @@ -151,9 +151,9 @@ void dnodeCleanupClient() { } static void dnodeProcessRspFromDnode(SRpcMsg *pMsg, SRpcEpSet *pEpSet) { - if (dnodeGetRunStatus() != TSDB_RUN_STATUS_RUNING) { + if (dnodeGetRunStatus() == TSDB_RUN_STATUS_STOPPED) { if (pMsg == NULL || pMsg->pCont == NULL) return; - dDebug("msg:%p is ignored since dnode not running", pMsg); + dDebug("msg:%p is ignored since dnode is stopping", pMsg); rpcFreeCont(pMsg->pCont); return; } diff --git a/src/dnode/src/dnodeVRead.c b/src/dnode/src/dnodeVRead.c index e1c1b1585d..b42a627a3a 100644 --- a/src/dnode/src/dnodeVRead.c +++ b/src/dnode/src/dnodeVRead.c @@ -175,7 +175,7 @@ static void *dnodeProcessReadQueue(void *pWorker) { break; } - dDebug("%p, msg:%p:%s will be processed in vread queue, qtype:%d", pRead->rpcAhandle, pRead, + dTrace("msg:%p, app:%p type:%s will be processed in vread queue, qtype:%d", pRead, pRead->rpcAhandle, taosMsg[pRead->msgType], qtype); int32_t code = vnodeProcessRead(pVnode, pRead); diff --git a/src/dnode/src/dnodeVWrite.c b/src/dnode/src/dnodeVWrite.c index da1a902fb3..6d4b50ee54 100644 --- a/src/dnode/src/dnodeVWrite.c +++ b/src/dnode/src/dnodeVWrite.c @@ -176,7 +176,7 @@ void dnodeSendRpcVWriteRsp(void *pVnode, void *wparam, int32_t code) { if (count <= 1) return; SRpcMsg rpcRsp = { - .handle = pWrite->rpcHandle, + .handle = pWrite->rpcMsg.handle, .pCont = pWrite->rspRet.rsp, .contLen = pWrite->rspRet.len, .code = pWrite->code, @@ -205,8 +205,8 @@ static void *dnodeProcessVWriteQueue(void *wparam) { bool forceFsync = false; for (int32_t i = 0; i < numOfMsgs; ++i) { taosGetQitem(pWorker->qall, &qtype, (void **)&pWrite); - dTrace("%p, msg:%p:%s will be processed in vwrite queue, qtype:%s hver:%" PRIu64, pWrite->rpcAhandle, pWrite, - taosMsg[pWrite->pHead->msgType], qtypeStr[qtype], pWrite->pHead->version); + dTrace("msg:%p, app:%p type:%s will be processed in vwrite queue, qtype:%s hver:%" PRIu64, pWrite, + pWrite->rpcMsg.ahandle, taosMsg[pWrite->pHead->msgType], qtypeStr[qtype], pWrite->pHead->version); pWrite->code = vnodeProcessWrite(pVnode, pWrite->pHead, qtype, &pWrite->rspRet); if (pWrite->code <= 0) pWrite->processedCount = 1; diff --git a/src/inc/taoserror.h b/src/inc/taoserror.h index ff91989e5f..5fde6d40d5 100644 --- a/src/inc/taoserror.h +++ b/src/inc/taoserror.h @@ -205,6 +205,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_VND_INVALID_VRESION_FILE, 0, 0x050A, "Invalid ve TAOS_DEFINE_ERROR(TSDB_CODE_VND_IS_FULL, 0, 0x050B, "Vnode memory is full because commit failed") TAOS_DEFINE_ERROR(TSDB_CODE_VND_NOT_SYNCED, 0, 0x0511, "Database suspended") TAOS_DEFINE_ERROR(TSDB_CODE_VND_NO_WRITE_AUTH, 0, 0x0512, "Write operation denied") +TAOS_DEFINE_ERROR(TSDB_CODE_VND_SYNCING, 0, 0x0513, "Database is syncing") // tsdb TAOS_DEFINE_ERROR(TSDB_CODE_TDB_INVALID_TABLE_ID, 0, 0x0600, "Invalid table ID") @@ -259,6 +260,10 @@ TAOS_DEFINE_ERROR(TSDB_CODE_GRANT_CPU_LIMITED, 0, 0x080B, "CPU cores TAOS_DEFINE_ERROR(TSDB_CODE_SYN_INVALID_CONFIG, 0, 0x0900, "Invalid Sync Configuration") TAOS_DEFINE_ERROR(TSDB_CODE_SYN_NOT_ENABLED, 0, 0x0901, "Sync module not enabled") TAOS_DEFINE_ERROR(TSDB_CODE_SYN_INVALID_VERSION, 0, 0x0902, "Invalid Sync version") +TAOS_DEFINE_ERROR(TSDB_CODE_SYN_CONFIRM_EXPIRED, 0, 0x0903, "Sync confirm expired") +TAOS_DEFINE_ERROR(TSDB_CODE_SYN_VND_COMMITING, 0, 0x0904, "Vnode is commiting") +TAOS_DEFINE_ERROR(TSDB_CODE_SYN_FILE_CHNAGED, 0, 0x0905, "Vnode file is changed") +TAOS_DEFINE_ERROR(TSDB_CODE_SYN_APP_ERROR, 0, 0x1000, "Unexpected generic error in sync") // wal TAOS_DEFINE_ERROR(TSDB_CODE_WAL_APP_ERROR, 0, 0x1000, "Unexpected generic error in wal") @@ -365,7 +370,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_HTTP_OP_TAG_VALUE_TOO_LONG, 0, 0x11A4, "tag value TAOS_DEFINE_ERROR(TSDB_CODE_HTTP_OP_VALUE_NULL, 0, 0x11A5, "value not find") TAOS_DEFINE_ERROR(TSDB_CODE_HTTP_OP_VALUE_TYPE, 0, 0x11A6, "value type should be boolean, number or string") - +// odbc TAOS_DEFINE_ERROR(TSDB_CODE_ODBC_OOM, 0, 0x2100, "out of memory") TAOS_DEFINE_ERROR(TSDB_CODE_ODBC_CONV_CHAR_NOT_NUM, 0, 0x2101, "convertion not a valid literal input") TAOS_DEFINE_ERROR(TSDB_CODE_ODBC_CONV_UNDEF, 0, 0x2102, "convertion undefined") @@ -389,7 +394,6 @@ TAOS_DEFINE_ERROR(TSDB_CODE_ODBC_CONV_SRC_BAD_SEQ, 0, 0x2113, "src bad se TAOS_DEFINE_ERROR(TSDB_CODE_ODBC_CONV_SRC_INCOMPLETE, 0, 0x2114, "src incomplete") TAOS_DEFINE_ERROR(TSDB_CODE_ODBC_CONV_SRC_GENERAL, 0, 0x2115, "src general") - #ifdef TAOS_ERROR_C }; #endif diff --git a/src/inc/tsync.h b/src/inc/tsync.h index d57433eba9..293e382519 100644 --- a/src/inc/tsync.h +++ b/src/inc/tsync.h @@ -23,19 +23,19 @@ extern "C" { #define TAOS_SYNC_MAX_REPLICA 5 #define TAOS_SYNC_MAX_INDEX 0x7FFFFFFF -typedef enum _TAOS_SYNC_ROLE { - TAOS_SYNC_ROLE_OFFLINE, - TAOS_SYNC_ROLE_UNSYNCED, - TAOS_SYNC_ROLE_SYNCING, - TAOS_SYNC_ROLE_SLAVE, - TAOS_SYNC_ROLE_MASTER, +typedef enum { + TAOS_SYNC_ROLE_OFFLINE = 0, + TAOS_SYNC_ROLE_UNSYNCED = 1, + TAOS_SYNC_ROLE_SYNCING = 2, + TAOS_SYNC_ROLE_SLAVE = 3, + TAOS_SYNC_ROLE_MASTER = 4 } ESyncRole; -typedef enum _TAOS_SYNC_STATUS { - TAOS_SYNC_STATUS_INIT, - TAOS_SYNC_STATUS_START, - TAOS_SYNC_STATUS_FILE, - TAOS_SYNC_STATUS_CACHE, +typedef enum { + TAOS_SYNC_STATUS_INIT = 0, + TAOS_SYNC_STATUS_START = 1, + TAOS_SYNC_STATUS_FILE = 2, + TAOS_SYNC_STATUS_CACHE = 3 } ESyncStatus; typedef struct { @@ -64,33 +64,35 @@ typedef struct { if name is provided(name[0] is not zero), get the named file at the specified index. If not there, return zero. If it is there, set the size to file size, and return file magic number. Index shall not be updated. */ -typedef uint32_t (*FGetFileInfo)(void *ahandle, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion); +typedef uint32_t (*FGetFileInfo)(int32_t vgId, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion); // get the wal file from index or after // return value, -1: error, 1:more wal files, 0:last WAL. if name[0]==0, no WAL file -typedef int32_t (*FGetWalInfo)(void *ahandle, char *fileName, int64_t *fileId); +typedef int32_t (*FGetWalInfo)(int32_t vgId, char *fileName, int64_t *fileId); // when a forward pkt is received, call this to handle data -typedef int32_t (*FWriteToCache)(void *ahandle, void *pHead, int32_t qtype, void *pMsg); +typedef int32_t (*FWriteToCache)(int32_t vgId, void *pHead, int32_t qtype, void *pMsg); // when forward is confirmed by peer, master call this API to notify app -typedef void (*FConfirmForward)(void *ahandle, void *mhandle, int32_t code); +typedef void (*FConfirmForward)(int32_t vgId, void *mhandle, int32_t code); // when role is changed, call this to notify app -typedef void (*FNotifyRole)(void *ahandle, int8_t role); +typedef void (*FNotifyRole)(int32_t vgId, int8_t role); // if a number of retrieving data failed, call this to start flow control -typedef void (*FNotifyFlowCtrl)(void *ahandle, int32_t mseconds); +typedef void (*FNotifyFlowCtrl)(int32_t vgId, int32_t level); // when data file is synced successfully, notity app -typedef int32_t (*FNotifyFileSynced)(void *ahandle, uint64_t fversion); +typedef int32_t (*FNotifyFileSynced)(int32_t vgId, uint64_t fversion); + +// get file version +typedef int32_t (*FGetFileVersion)(int32_t vgId, uint64_t *fver); typedef struct { int32_t vgId; // vgroup ID uint64_t version; // initial version SSyncCfg syncCfg; // configuration from mgmt - char path[128]; // path to the file - void * ahandle; // handle provided by APP + char path[TSDB_FILENAME_LEN]; // path to the file FGetFileInfo getFileInfo; FGetWalInfo getWalInfo; FWriteToCache writeToCache; @@ -98,6 +100,7 @@ typedef struct { FNotifyRole notifyRole; FNotifyFlowCtrl notifyFlowCtrl; FNotifyFileSynced notifyFileSynced; + FGetFileVersion getFileVersion; } SSyncInfo; typedef void *tsync_h; diff --git a/src/inc/twal.h b/src/inc/twal.h index 8dd3a8a912..1645de77aa 100644 --- a/src/inc/twal.h +++ b/src/inc/twal.h @@ -51,9 +51,8 @@ typedef struct { typedef void * twalh; // WAL HANDLE typedef int32_t FWalWrite(void *ahandle, void *pHead, int32_t qtype, void *pMsg); -int32_t walInit(); -void walCleanUp(); - +int32_t walInit(); +void walCleanUp(); twalh walOpen(char *path, SWalCfg *pCfg); int32_t walAlter(twalh pWal, SWalCfg *pCfg); void walStop(twalh); diff --git a/src/inc/vnode.h b/src/inc/vnode.h index 4e8389498b..5f643295d6 100644 --- a/src/inc/vnode.h +++ b/src/inc/vnode.h @@ -20,6 +20,7 @@ extern "C" { #endif +#include "trpc.h" #include "twal.h" typedef enum _VN_STATUS { @@ -51,8 +52,9 @@ typedef struct { typedef struct { int32_t code; int32_t processedCount; - void * rpcHandle; - void * rpcAhandle; + int32_t qtype; + void * pVnode; + SRpcMsg rpcMsg; SRspRet rspRet; char reserveForSync[16]; SWalHead pHead[]; diff --git a/src/kit/taosdump/taosdump.c b/src/kit/taosdump/taosdump.c index 88f07ee602..a7258c9724 100644 --- a/src/kit/taosdump/taosdump.c +++ b/src/kit/taosdump/taosdump.c @@ -2146,7 +2146,7 @@ int taosDumpInOneFile(TAOS * taos, FILE* fp, char* fcharset, char* encode, c char * line = NULL; size_t line_len = 0; - cmd = (char *)malloc(COMMAND_SIZE); + cmd = (char *)malloc(TSDB_MAX_ALLOWED_SQL_LEN); if (cmd == NULL) { fprintf(stderr, "failed to allocate memory\n"); return -1; @@ -2155,7 +2155,7 @@ int taosDumpInOneFile(TAOS * taos, FILE* fp, char* fcharset, char* encode, c int lineNo = 0; while ((read_len = getline(&line, &line_len, fp)) != -1) { ++lineNo; - if (read_len >= COMMAND_SIZE) continue; + if (read_len >= TSDB_MAX_ALLOWED_SQL_LEN) continue; line[--read_len] = '\0'; //if (read_len == 0 || isCommentLine(line)) { // line starts with # @@ -2176,7 +2176,7 @@ int taosDumpInOneFile(TAOS * taos, FILE* fp, char* fcharset, char* encode, c fprintf(stderr, "error sql: linenu:%d, file:%s\n", lineNo, fileName); } - memset(cmd, 0, COMMAND_SIZE); + memset(cmd, 0, TSDB_MAX_ALLOWED_SQL_LEN); cmd_len = 0; } diff --git a/src/mnode/inc/mnodeAcct.h b/src/mnode/inc/mnodeAcct.h index 744a62f948..595dcca413 100644 --- a/src/mnode/inc/mnodeAcct.h +++ b/src/mnode/inc/mnodeAcct.h @@ -27,6 +27,7 @@ void mnodeCleanupAccts(); void mnodeGetStatOfAllAcct(SAcctInfo* pAcctInfo); void * mnodeGetAcct(char *acctName); void * mnodeGetNextAcct(void *pIter, SAcctObj **pAcct); +void mnodeCancelGetNextAcct(void *pIter); void mnodeIncAcctRef(SAcctObj *pAcct); void mnodeDecAcctRef(SAcctObj *pAcct); void mnodeAddDbToAcct(SAcctObj *pAcct, SDbObj *pDb); diff --git a/src/mnode/inc/mnodeDb.h b/src/mnode/inc/mnodeDb.h index 7cbd08ed92..9354b923d7 100644 --- a/src/mnode/inc/mnodeDb.h +++ b/src/mnode/inc/mnodeDb.h @@ -34,6 +34,7 @@ int64_t mnodeGetDbNum(); SDbObj *mnodeGetDb(char *db); SDbObj *mnodeGetDbByTableId(char *db); void * mnodeGetNextDb(void *pIter, SDbObj **pDb); +void mnodeCancelGetNextDb(void *pIter); void mnodeIncDbRef(SDbObj *pDb); void mnodeDecDbRef(SDbObj *pDb); bool mnodeCheckIsMonitorDB(char *db, char *monitordb); diff --git a/src/mnode/inc/mnodeDnode.h b/src/mnode/inc/mnodeDnode.h index b6ddb7a9bf..8bc29ef9ef 100644 --- a/src/mnode/inc/mnodeDnode.h +++ b/src/mnode/inc/mnodeDnode.h @@ -65,6 +65,7 @@ int32_t mnodeGetDnodesNum(); int32_t mnodeGetOnlinDnodesCpuCoreNum(); int32_t mnodeGetOnlineDnodesNum(); void * mnodeGetNextDnode(void *pIter, SDnodeObj **pDnode); +void mnodeCancelGetNextDnode(void *pIter); void mnodeIncDnodeRef(SDnodeObj *pDnode); void mnodeDecDnodeRef(SDnodeObj *pDnode); void * mnodeGetDnode(int32_t dnodeId); diff --git a/src/mnode/inc/mnodeMnode.h b/src/mnode/inc/mnodeMnode.h index a28a03ea40..10cbcebe22 100644 --- a/src/mnode/inc/mnodeMnode.h +++ b/src/mnode/inc/mnodeMnode.h @@ -38,6 +38,7 @@ void mnodeDropMnodeLocal(int32_t dnodeId); void * mnodeGetMnode(int32_t mnodeId); int32_t mnodeGetMnodesNum(); void * mnodeGetNextMnode(void *pIter, struct SMnodeObj **pMnode); +void mnodeCancelGetNextMnode(void *pIter); void mnodeIncMnodeRef(struct SMnodeObj *pMnode); void mnodeDecMnodeRef(struct SMnodeObj *pMnode); diff --git a/src/mnode/inc/mnodeSdb.h b/src/mnode/inc/mnodeSdb.h index f4854f69a0..81629ede78 100644 --- a/src/mnode/inc/mnodeSdb.h +++ b/src/mnode/inc/mnodeSdb.h @@ -20,7 +20,8 @@ extern "C" { #endif -struct SMnodeMsg; +#include "mnode.h" +#include "twal.h" typedef enum { SDB_TABLE_CLUSTER = 0, @@ -36,44 +37,46 @@ typedef enum { } ESdbTable; typedef enum { - SDB_KEY_STRING, - SDB_KEY_INT, - SDB_KEY_AUTO, - SDB_KEY_VAR_STRING, + SDB_KEY_STRING = 0, + SDB_KEY_INT = 1, + SDB_KEY_AUTO = 2, + SDB_KEY_VAR_STRING = 3, } ESdbKey; typedef enum { - SDB_OPER_GLOBAL, - SDB_OPER_LOCAL + SDB_OPER_GLOBAL = 0, + SDB_OPER_LOCAL = 1 } ESdbOper; -typedef struct SSdbOper { - ESdbOper type; - int32_t rowSize; - int32_t retCode; // for callback in sdb queue - int32_t processedCount; // for sync fwd callback - int32_t (*reqFp)(struct SMnodeMsg *pMsg); - int32_t (*writeCb)(struct SMnodeMsg *pMsg, int32_t code); - void * table; - void * pObj; - void * rowData; - struct SMnodeMsg *pMsg; -} SSdbOper; +typedef struct SSdbRow { + ESdbOper type; + int32_t processedCount; // for sync fwd callback + int32_t code; // for callback in sdb queue + int32_t rowSize; + void * rowData; + void * pObj; + void * pTable; + SMnodeMsg *pMsg; + int32_t (*fpReq)(SMnodeMsg *pMsg); + int32_t (*fpRsp)(SMnodeMsg *pMsg, int32_t code); + char reserveForSync[16]; + SWalHead pHead[]; +} SSdbRow; typedef struct { - char *tableName; - int32_t hashSessions; - int32_t maxRowSize; - int32_t refCountPos; - ESdbTable tableId; + char * name; + int32_t hashSessions; + int32_t maxRowSize; + int32_t refCountPos; + ESdbTable id; ESdbKey keyType; - int32_t (*insertFp)(SSdbOper *pOper); - int32_t (*deleteFp)(SSdbOper *pOper); - int32_t (*updateFp)(SSdbOper *pOper); - int32_t (*encodeFp)(SSdbOper *pOper); - int32_t (*decodeFp)(SSdbOper *pDesc); - int32_t (*destroyFp)(SSdbOper *pDesc); - int32_t (*restoredFp)(); + int32_t (*fpInsert)(SSdbRow *pRow); + int32_t (*fpDelete)(SSdbRow *pRow); + int32_t (*fpUpdate)(SSdbRow *pRow); + int32_t (*fpEncode)(SSdbRow *pRow); + int32_t (*fpDecode)(SSdbRow *pRow); + int32_t (*fpDestroy)(SSdbRow *pRow); + int32_t (*fpRestored)(); } SSdbTableDesc; int32_t sdbInit(); @@ -84,20 +87,20 @@ bool sdbIsMaster(); bool sdbIsServing(); void sdbUpdateMnodeRoles(); -int32_t sdbInsertRow(SSdbOper *pOper); -int32_t sdbDeleteRow(SSdbOper *pOper); -int32_t sdbUpdateRow(SSdbOper *pOper); -int32_t sdbInsertRowImp(SSdbOper *pOper); +int32_t sdbInsertRow(SSdbRow *pRow); +int32_t sdbDeleteRow(SSdbRow *pRow); +int32_t sdbUpdateRow(SSdbRow *pRow); +int32_t sdbInsertRowToQueue(SSdbRow *pRow); -void *sdbGetRow(void *handle, void *key); -void *sdbFetchRow(void *handle, void *pIter, void **ppRow); -void sdbFreeIter(void *pIter); -void sdbIncRef(void *thandle, void *pRow); -void sdbDecRef(void *thandle, void *pRow); -int64_t sdbGetNumOfRows(void *handle); -int32_t sdbGetId(void *handle); +void * sdbGetRow(void *pTable, void *key); +void * sdbFetchRow(void *pTable, void *pIter, void **ppRow); +void sdbFreeIter(void *pTable, void *pIter); +void sdbIncRef(void *pTable, void *pRow); +void sdbDecRef(void *pTable, void *pRow); +int64_t sdbGetNumOfRows(void *pTable); +int32_t sdbGetId(void *pTable); uint64_t sdbGetVersion(); -bool sdbCheckRowDeleted(void *thandle, void *pRow); +bool sdbCheckRowDeleted(void *pTable, void *pRow); #ifdef __cplusplus } diff --git a/src/mnode/inc/mnodeShow.h b/src/mnode/inc/mnodeShow.h index da66e71678..f985fe792d 100644 --- a/src/mnode/inc/mnodeShow.h +++ b/src/mnode/inc/mnodeShow.h @@ -26,8 +26,10 @@ void mnodeCleanUpShow(); typedef int32_t (*SShowMetaFp)(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn); typedef int32_t (*SShowRetrieveFp)(SShowObj *pShow, char *data, int32_t rows, void *pConn); +typedef void (*SShowFreeIterFp)(void *pIter); void mnodeAddShowMetaHandle(uint8_t showType, SShowMetaFp fp); void mnodeAddShowRetrieveHandle(uint8_t showType, SShowRetrieveFp fp); +void mnodeAddShowFreeIterHandle(uint8_t msgType, SShowFreeIterFp fp); void mnodeVacuumResult(char *data, int32_t numOfCols, int32_t rows, int32_t capacity, SShowObj *pShow); #ifdef __cplusplus diff --git a/src/mnode/inc/mnodeTable.h b/src/mnode/inc/mnodeTable.h index 7c0077aa60..bf04f26a90 100644 --- a/src/mnode/inc/mnodeTable.h +++ b/src/mnode/inc/mnodeTable.h @@ -31,6 +31,8 @@ void mnodeIncTableRef(void *pTable); void mnodeDecTableRef(void *pTable); void * mnodeGetNextChildTable(void *pIter, SCTableObj **pTable); void * mnodeGetNextSuperTable(void *pIter, SSTableObj **pTable); +void mnodeCancelGetNextChildTable(void *pIter); +void mnodeCancelGetNextSuperTable(void *pIter); void mnodeDropAllChildTables(SDbObj *pDropDb); void mnodeDropAllSuperTables(SDbObj *pDropDb); void mnodeDropAllChildTablesInVgroups(SVgObj *pVgroup); diff --git a/src/mnode/inc/mnodeUser.h b/src/mnode/inc/mnodeUser.h index 073460f9d3..156bc7aaeb 100644 --- a/src/mnode/inc/mnodeUser.h +++ b/src/mnode/inc/mnodeUser.h @@ -25,6 +25,7 @@ int32_t mnodeInitUsers(); void mnodeCleanupUsers(); SUserObj *mnodeGetUser(char *name); void * mnodeGetNextUser(void *pIter, SUserObj **pUser); +void mnodeCancelGetNextUser(void *pIter); void mnodeIncUserRef(SUserObj *pUser); void mnodeDecUserRef(SUserObj *pUser); SUserObj *mnodeGetUserFromConn(void *pConn); diff --git a/src/mnode/inc/mnodeVgroup.h b/src/mnode/inc/mnodeVgroup.h index 0e6d9dfde4..ee9ec7ae93 100644 --- a/src/mnode/inc/mnodeVgroup.h +++ b/src/mnode/inc/mnodeVgroup.h @@ -34,6 +34,7 @@ void mnodeDropAllDnodeVgroups(SDnodeObj *pDropDnode); //void mnodeUpdateAllDbVgroups(SDbObj *pAlterDb); void * mnodeGetNextVgroup(void *pIter, SVgObj **pVgroup); +void mnodeCancelGetNextVgroup(void *pIter); void mnodeUpdateVgroup(SVgObj *pVgroup); void mnodeUpdateVgroupStatus(SVgObj *pVgroup, SDnodeObj *pDnode, SVnodeLoad *pVload); void mnodeCheckUnCreatedVgroup(SDnodeObj *pDnode, SVnodeLoad *pVloads, int32_t openVnodes); diff --git a/src/mnode/src/mnodeAcct.c b/src/mnode/src/mnodeAcct.c index 365cf656de..3da889b284 100644 --- a/src/mnode/src/mnodeAcct.c +++ b/src/mnode/src/mnodeAcct.c @@ -16,6 +16,7 @@ #define _DEFAULT_SOURCE #include "os.h" #include "taoserror.h" +#include "tglobal.h" #include "dnode.h" #include "mnodeDef.h" #include "mnodeInt.h" @@ -25,36 +26,34 @@ #include "mnodeUser.h" #include "mnodeVgroup.h" -#include "tglobal.h" - void * tsAcctSdb = NULL; static int32_t tsAcctUpdateSize; static int32_t mnodeCreateRootAcct(); -static int32_t mnodeAcctActionDestroy(SSdbOper *pOper) { - SAcctObj *pAcct = pOper->pObj; +static int32_t mnodeAcctActionDestroy(SSdbRow *pRow) { + SAcctObj *pAcct = pRow->pObj; pthread_mutex_destroy(&pAcct->mutex); - tfree(pOper->pObj); + tfree(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeAcctActionInsert(SSdbOper *pOper) { - SAcctObj *pAcct = pOper->pObj; +static int32_t mnodeAcctActionInsert(SSdbRow *pRow) { + SAcctObj *pAcct = pRow->pObj; memset(&pAcct->acctInfo, 0, sizeof(SAcctInfo)); pAcct->acctInfo.accessState = TSDB_VN_ALL_ACCCESS; pthread_mutex_init(&pAcct->mutex, NULL); return TSDB_CODE_SUCCESS; } -static int32_t mnodeAcctActionDelete(SSdbOper *pOper) { - SAcctObj *pAcct = pOper->pObj; +static int32_t mnodeAcctActionDelete(SSdbRow *pRow) { + SAcctObj *pAcct = pRow->pObj; mnodeDropAllUsers(pAcct); mnodeDropAllDbs(pAcct); return TSDB_CODE_SUCCESS; } -static int32_t mnodeAcctActionUpdate(SSdbOper *pOper) { - SAcctObj *pAcct = pOper->pObj; +static int32_t mnodeAcctActionUpdate(SSdbRow *pRow) { + SAcctObj *pAcct = pRow->pObj; SAcctObj *pSaved = mnodeGetAcct(pAcct->user); if (pAcct != pSaved) { memcpy(pSaved, pAcct, tsAcctUpdateSize); @@ -64,19 +63,19 @@ static int32_t mnodeAcctActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeAcctActionEncode(SSdbOper *pOper) { - SAcctObj *pAcct = pOper->pObj; - memcpy(pOper->rowData, pAcct, tsAcctUpdateSize); - pOper->rowSize = tsAcctUpdateSize; +static int32_t mnodeAcctActionEncode(SSdbRow *pRow) { + SAcctObj *pAcct = pRow->pObj; + memcpy(pRow->rowData, pAcct, tsAcctUpdateSize); + pRow->rowSize = tsAcctUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeAcctActionDecode(SSdbOper *pOper) { +static int32_t mnodeAcctActionDecode(SSdbRow *pRow) { SAcctObj *pAcct = (SAcctObj *) calloc(1, sizeof(SAcctObj)); if (pAcct == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pAcct, pOper->rowData, tsAcctUpdateSize); - pOper->pObj = pAcct; + memcpy(pAcct, pRow->rowData, tsAcctUpdateSize); + pRow->pObj = pAcct; return TSDB_CODE_SUCCESS; } @@ -99,29 +98,29 @@ int32_t mnodeInitAccts() { SAcctObj tObj; tsAcctUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_ACCOUNT, - .tableName = "accounts", + SSdbTableDesc desc = { + .id = SDB_TABLE_ACCOUNT, + .name = "accounts", .hashSessions = TSDB_DEFAULT_ACCOUNTS_HASH_SIZE, .maxRowSize = tsAcctUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_STRING, - .insertFp = mnodeAcctActionInsert, - .deleteFp = mnodeAcctActionDelete, - .updateFp = mnodeAcctActionUpdate, - .encodeFp = mnodeAcctActionEncode, - .decodeFp = mnodeAcctActionDecode, - .destroyFp = mnodeAcctActionDestroy, - .restoredFp = mnodeAcctActionRestored + .fpInsert = mnodeAcctActionInsert, + .fpDelete = mnodeAcctActionDelete, + .fpUpdate = mnodeAcctActionUpdate, + .fpEncode = mnodeAcctActionEncode, + .fpDecode = mnodeAcctActionDecode, + .fpDestroy = mnodeAcctActionDestroy, + .fpRestored = mnodeAcctActionRestored }; - tsAcctSdb = sdbOpenTable(&tableDesc); + tsAcctSdb = sdbOpenTable(&desc); if (tsAcctSdb == NULL) { - mError("table:%s, failed to create hash", tableDesc.tableName); + mError("table:%s, failed to create hash", desc.name); return -1; } - mDebug("table:%s, hash is created", tableDesc.tableName); + mDebug("table:%s, hash is created", desc.name); return TSDB_CODE_SUCCESS; } @@ -145,7 +144,6 @@ void mnodeGetStatOfAllAcct(SAcctInfo* pAcctInfo) { pAcctInfo->numOfTimeSeries += pAcct->acctInfo.numOfTimeSeries; mnodeDecAcctRef(pAcct); } - sdbFreeIter(pIter); SVgObj *pVgroup = NULL; pIter = NULL; @@ -159,7 +157,6 @@ void mnodeGetStatOfAllAcct(SAcctInfo* pAcctInfo) { pAcctInfo->totalPoints += pVgroup->pointsWritten; mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); } void *mnodeGetAcct(char *name) { @@ -170,6 +167,10 @@ void *mnodeGetNextAcct(void *pIter, SAcctObj **pAcct) { return sdbFetchRow(tsAcctSdb, pIter, (void **)pAcct); } +void mnodeCancelGetNextAcct(void *pIter) { + sdbFreeIter(tsAcctSdb, pIter); +} + void mnodeIncAcctRef(SAcctObj *pAcct) { sdbIncRef(tsAcctSdb, pAcct); } @@ -226,13 +227,13 @@ static int32_t mnodeCreateRootAcct() { pAcct->acctId = sdbGetId(tsAcctSdb); pAcct->createdTime = taosGetTimestampMs(); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsAcctSdb, - .pObj = pAcct, + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsAcctSdb, + .pObj = pAcct, }; - return sdbInsertRow(&oper); + return sdbInsertRow(&row); } #ifndef _ACCT diff --git a/src/mnode/src/mnodeCluster.c b/src/mnode/src/mnodeCluster.c index 98587cf53d..839407c5b0 100644 --- a/src/mnode/src/mnodeCluster.c +++ b/src/mnode/src/mnodeCluster.c @@ -31,37 +31,38 @@ static int32_t mnodeCreateCluster(); static int32_t mnodeGetClusterMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn); static int32_t mnodeRetrieveClusters(SShowObj *pShow, char *data, int32_t rows, void *pConn); +static void mnodeCancelGetNextCluster(void *pIter); -static int32_t mnodeClusterActionDestroy(SSdbOper *pOper) { - tfree(pOper->pObj); +static int32_t mnodeClusterActionDestroy(SSdbRow *pRow) { + tfree(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeClusterActionInsert(SSdbOper *pOper) { +static int32_t mnodeClusterActionInsert(SSdbRow *pRow) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeClusterActionDelete(SSdbOper *pOper) { +static int32_t mnodeClusterActionDelete(SSdbRow *pRow) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeClusterActionUpdate(SSdbOper *pOper) { +static int32_t mnodeClusterActionUpdate(SSdbRow *pRow) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeClusterActionEncode(SSdbOper *pOper) { - SClusterObj *pCluster = pOper->pObj; - memcpy(pOper->rowData, pCluster, tsClusterUpdateSize); - pOper->rowSize = tsClusterUpdateSize; +static int32_t mnodeClusterActionEncode(SSdbRow *pRow) { + SClusterObj *pCluster = pRow->pObj; + memcpy(pRow->rowData, pCluster, tsClusterUpdateSize); + pRow->rowSize = tsClusterUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeClusterActionDecode(SSdbOper *pOper) { +static int32_t mnodeClusterActionDecode(SSdbRow *pRow) { SClusterObj *pCluster = (SClusterObj *) calloc(1, sizeof(SClusterObj)); if (pCluster == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pCluster, pOper->rowData, tsClusterUpdateSize); - pOper->pObj = pCluster; + memcpy(pCluster, pRow->rowData, tsClusterUpdateSize); + pRow->pObj = pCluster; return TSDB_CODE_SUCCESS; } @@ -84,32 +85,33 @@ int32_t mnodeInitCluster() { SClusterObj tObj; tsClusterUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_CLUSTER, - .tableName = "cluster", + SSdbTableDesc desc = { + .id = SDB_TABLE_CLUSTER, + .name = "cluster", .hashSessions = TSDB_DEFAULT_CLUSTER_HASH_SIZE, .maxRowSize = tsClusterUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_STRING, - .insertFp = mnodeClusterActionInsert, - .deleteFp = mnodeClusterActionDelete, - .updateFp = mnodeClusterActionUpdate, - .encodeFp = mnodeClusterActionEncode, - .decodeFp = mnodeClusterActionDecode, - .destroyFp = mnodeClusterActionDestroy, - .restoredFp = mnodeClusterActionRestored + .fpInsert = mnodeClusterActionInsert, + .fpDelete = mnodeClusterActionDelete, + .fpUpdate = mnodeClusterActionUpdate, + .fpEncode = mnodeClusterActionEncode, + .fpDecode = mnodeClusterActionDecode, + .fpDestroy = mnodeClusterActionDestroy, + .fpRestored = mnodeClusterActionRestored }; - tsClusterSdb = sdbOpenTable(&tableDesc); + tsClusterSdb = sdbOpenTable(&desc); if (tsClusterSdb == NULL) { - mError("table:%s, failed to create hash", tableDesc.tableName); + mError("table:%s, failed to create hash", desc.name); return -1; } mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_CLUSTER, mnodeGetClusterMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_CLUSTER, mnodeRetrieveClusters); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_CLUSTER, mnodeCancelGetNextCluster); - mDebug("table:%s, hash is created", tableDesc.tableName); + mDebug("table:%s, hash is created", desc.name); return TSDB_CODE_SUCCESS; } @@ -122,6 +124,10 @@ void *mnodeGetNextCluster(void *pIter, SClusterObj **pCluster) { return sdbFetchRow(tsClusterSdb, pIter, (void **)pCluster); } +void mnodeCancelGetNextCluster(void *pIter) { + sdbFreeIter(tsClusterSdb, pIter); +} + void mnodeIncClusterRef(SClusterObj *pCluster) { sdbIncRef(tsClusterSdb, pCluster); } @@ -145,13 +151,13 @@ static int32_t mnodeCreateCluster() { mDebug("uid is %s", pCluster->uid); } - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsClusterSdb, - .pObj = pCluster, + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsClusterSdb, + .pObj = pCluster, }; - return sdbInsertRow(&oper); + return sdbInsertRow(&row); } const char* mnodeGetClusterId() { @@ -167,7 +173,7 @@ void mnodeUpdateClusterId() { } mnodeDecClusterRef(pCluster); - sdbFreeIter(pIter); + mnodeCancelGetNextCluster(pIter); } static int32_t mnodeGetClusterMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) { diff --git a/src/mnode/src/mnodeDb.c b/src/mnode/src/mnodeDb.c index 3b3a21aeb1..77f3b93eb1 100644 --- a/src/mnode/src/mnodeDb.c +++ b/src/mnode/src/mnodeDb.c @@ -56,8 +56,8 @@ static void mnodeDestroyDb(SDbObj *pDb) { tfree(pDb); } -static int32_t mnodeDbActionDestroy(SSdbOper *pOper) { - mnodeDestroyDb(pOper->pObj); +static int32_t mnodeDbActionDestroy(SSdbRow *pRow) { + mnodeDestroyDb(pRow->pObj); return TSDB_CODE_SUCCESS; } @@ -65,8 +65,8 @@ int64_t mnodeGetDbNum() { return sdbGetNumOfRows(tsDbSdb); } -static int32_t mnodeDbActionInsert(SSdbOper *pOper) { - SDbObj *pDb = pOper->pObj; +static int32_t mnodeDbActionInsert(SSdbRow *pRow) { + SDbObj *pDb = pRow->pObj; SAcctObj *pAcct = mnodeGetAcct(pDb->acct); pthread_mutex_init(&pDb->mutex, NULL); @@ -91,8 +91,8 @@ static int32_t mnodeDbActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeDbActionDelete(SSdbOper *pOper) { - SDbObj *pDb = pOper->pObj; +static int32_t mnodeDbActionDelete(SSdbRow *pRow) { + SDbObj *pDb = pRow->pObj; SAcctObj *pAcct = mnodeGetAcct(pDb->acct); mnodeDropAllChildTables(pDb); @@ -107,11 +107,11 @@ static int32_t mnodeDbActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeDbActionUpdate(SSdbOper *pOper) { - SDbObj *pNew = pOper->pObj; +static int32_t mnodeDbActionUpdate(SSdbRow *pRow) { + SDbObj *pNew = pRow->pObj; SDbObj *pDb = mnodeGetDb(pNew->name); if (pDb != NULL && pNew != pDb) { - memcpy(pDb, pNew, pOper->rowSize); + memcpy(pDb, pNew, pRow->rowSize); free(pNew->vgList); free(pNew); } @@ -120,19 +120,19 @@ static int32_t mnodeDbActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeDbActionEncode(SSdbOper *pOper) { - SDbObj *pDb = pOper->pObj; - memcpy(pOper->rowData, pDb, tsDbUpdateSize); - pOper->rowSize = tsDbUpdateSize; +static int32_t mnodeDbActionEncode(SSdbRow *pRow) { + SDbObj *pDb = pRow->pObj; + memcpy(pRow->rowData, pDb, tsDbUpdateSize); + pRow->rowSize = tsDbUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeDbActionDecode(SSdbOper *pOper) { +static int32_t mnodeDbActionDecode(SSdbRow *pRow) { SDbObj *pDb = (SDbObj *) calloc(1, sizeof(SDbObj)); if (pDb == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pDb, pOper->rowData, tsDbUpdateSize); - pOper->pObj = pDb; + memcpy(pDb, pRow->rowData, tsDbUpdateSize); + pRow->pObj = pDb; return TSDB_CODE_SUCCESS; } @@ -144,23 +144,23 @@ int32_t mnodeInitDbs() { SDbObj tObj; tsDbUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_DB, - .tableName = "dbs", + SSdbTableDesc desc = { + .id = SDB_TABLE_DB, + .name = "dbs", .hashSessions = TSDB_DEFAULT_DBS_HASH_SIZE, .maxRowSize = tsDbUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_STRING, - .insertFp = mnodeDbActionInsert, - .deleteFp = mnodeDbActionDelete, - .updateFp = mnodeDbActionUpdate, - .encodeFp = mnodeDbActionEncode, - .decodeFp = mnodeDbActionDecode, - .destroyFp = mnodeDbActionDestroy, - .restoredFp = mnodeDbActionRestored + .fpInsert = mnodeDbActionInsert, + .fpDelete = mnodeDbActionDelete, + .fpUpdate = mnodeDbActionUpdate, + .fpEncode = mnodeDbActionEncode, + .fpDecode = mnodeDbActionDecode, + .fpDestroy = mnodeDbActionDestroy, + .fpRestored = mnodeDbActionRestored }; - tsDbSdb = sdbOpenTable(&tableDesc); + tsDbSdb = sdbOpenTable(&desc); if (tsDbSdb == NULL) { mError("failed to init db data"); return -1; @@ -171,6 +171,7 @@ int32_t mnodeInitDbs() { mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_DROP_DB, mnodeProcessDropDbMsg); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_DB, mnodeGetDbMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_DB, mnodeRetrieveDbs); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_DB, mnodeCancelGetNextDb); mDebug("table:dbs table is created"); return 0; @@ -180,6 +181,10 @@ void *mnodeGetNextDb(void *pIter, SDbObj **pDb) { return sdbFetchRow(tsDbSdb, pIter, (void **)pDb); } +void mnodeCancelGetNextDb(void *pIter) { + sdbFreeIter(tsDbSdb, pIter); +} + SDbObj *mnodeGetDb(char *db) { return (SDbObj *)sdbGetRow(tsDbSdb, db); } @@ -412,16 +417,16 @@ static int32_t mnodeCreateDb(SAcctObj *pAcct, SCreateDbMsg *pCreate, SMnodeMsg * pMsg->pDb = pDb; mnodeIncDbRef(pDb); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDbSdb, - .pObj = pDb, - .rowSize = sizeof(SDbObj), - .pMsg = pMsg, - .writeCb = mnodeCreateDbCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDbSdb, + .pObj = pDb, + .rowSize = sizeof(SDbObj), + .pMsg = pMsg, + .fpRsp = mnodeCreateDbCb }; - code = sdbInsertRow(&oper); + code = sdbInsertRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("db:%s, failed to create, reason:%s", pDb->name, tstrerror(code)); pMsg->pDb = NULL; @@ -440,8 +445,8 @@ bool mnodeCheckIsMonitorDB(char *db, char *monitordb) { } #if 0 -void mnodePrintVgroups(SDbObj *pDb, char *oper) { - mInfo("db:%s, vgroup link from head, oper:%s", pDb->name, oper); +void mnodePrintVgroups(SDbObj *pDb, char *row) { + mInfo("db:%s, vgroup link from head, row:%s", pDb->name, row); SVgObj *pVgroup = pDb->pHead; while (pVgroup != NULL) { mInfo("vgId:%d", pVgroup->vgId); @@ -807,13 +812,13 @@ static int32_t mnodeSetDbDropping(SDbObj *pDb) { if (pDb->status) return TSDB_CODE_SUCCESS; pDb->status = true; - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDbSdb, - .pObj = pDb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDbSdb, + .pObj = pDb }; - int32_t code = sdbUpdateRow(&oper); + int32_t code = sdbUpdateRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("db:%s, failed to set dropping state, reason:%s", pDb->name, tstrerror(code)); } @@ -986,8 +991,8 @@ static int32_t mnodeAlterDbCb(SMnodeMsg *pMsg, int32_t code) { SDbObj *pDb = pMsg->pDb; void *pIter = NULL; - while (1) { - SVgObj *pVgroup = NULL; + SVgObj *pVgroup = NULL; + while (1) { pIter = mnodeGetNextVgroup(pIter, &pVgroup); if (pVgroup == NULL) break; if (pVgroup->pDb == pDb) { @@ -995,7 +1000,6 @@ static int32_t mnodeAlterDbCb(SMnodeMsg *pMsg, int32_t code) { } mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); mDebug("db:%s, all vgroups is altered", pDb->name); mLInfo("db:%s, is alterd by %s", pDb->name, mnodeGetUserFromMsg(pMsg)); @@ -1019,15 +1023,15 @@ static int32_t mnodeAlterDb(SDbObj *pDb, SAlterDbMsg *pAlter, void *pMsg) { if (memcmp(&newCfg, &pDb->cfg, sizeof(SDbCfg)) != 0) { pDb->cfg = newCfg; pDb->cfgVersion++; - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDbSdb, - .pObj = pDb, - .pMsg = pMsg, - .writeCb = mnodeAlterDbCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDbSdb, + .pObj = pDb, + .pMsg = pMsg, + .fpRsp = mnodeAlterDbCb }; - code = sdbUpdateRow(&oper); + code = sdbUpdateRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("db:%s, failed to alter, reason:%s", pDb->name, tstrerror(code)); } @@ -1071,15 +1075,15 @@ static int32_t mnodeDropDb(SMnodeMsg *pMsg) { SDbObj *pDb = pMsg->pDb; mInfo("db:%s, drop db from sdb", pDb->name); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDbSdb, - .pObj = pDb, - .pMsg = pMsg, - .writeCb = mnodeDropDbCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDbSdb, + .pObj = pDb, + .pMsg = pMsg, + .fpRsp = mnodeDropDbCb }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("db:%s, failed to drop, reason:%s", pDb->name, tstrerror(code)); } @@ -1134,19 +1138,17 @@ void mnodeDropAllDbs(SAcctObj *pAcct) { if (pDb->pAcct == pAcct) { mInfo("db:%s, drop db from sdb for acct:%s is dropped", pDb->name, pAcct->user); - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsDbSdb, - .pObj = pDb + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsDbSdb, + .pObj = pDb }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfDbs++; } mnodeDecDbRef(pDb); } - sdbFreeIter(pIter); - mInfo("acct:%s, all dbs:%d is dropped from sdb", pAcct->user, numOfDbs); } diff --git a/src/mnode/src/mnodeDnode.c b/src/mnode/src/mnodeDnode.c index 7e34e08373..0e9b56e459 100644 --- a/src/mnode/src/mnodeDnode.c +++ b/src/mnode/src/mnodeDnode.c @@ -87,13 +87,13 @@ static char* offlineReason[] = { "unknown", }; -static int32_t mnodeDnodeActionDestroy(SSdbOper *pOper) { - tfree(pOper->pObj); +static int32_t mnodeDnodeActionDestroy(SSdbRow *pRow) { + tfree(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeDnodeActionInsert(SSdbOper *pOper) { - SDnodeObj *pDnode = pOper->pObj; +static int32_t mnodeDnodeActionInsert(SSdbRow *pRow) { + SDnodeObj *pDnode = pRow->pObj; if (pDnode->status != TAOS_DN_STATUS_DROPPING) { pDnode->status = TAOS_DN_STATUS_OFFLINE; pDnode->lastAccess = tsAccessSquence; @@ -107,8 +107,8 @@ static int32_t mnodeDnodeActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeDnodeActionDelete(SSdbOper *pOper) { - SDnodeObj *pDnode = pOper->pObj; +static int32_t mnodeDnodeActionDelete(SSdbRow *pRow) { + SDnodeObj *pDnode = pRow->pObj; #ifndef _SYNC mnodeDropAllDnodeVgroups(pDnode); @@ -121,11 +121,11 @@ static int32_t mnodeDnodeActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeDnodeActionUpdate(SSdbOper *pOper) { - SDnodeObj *pNew = pOper->pObj; +static int32_t mnodeDnodeActionUpdate(SSdbRow *pRow) { + SDnodeObj *pNew = pRow->pObj; SDnodeObj *pDnode = mnodeGetDnode(pNew->dnodeId); if (pDnode != NULL && pNew != pDnode) { - memcpy(pDnode, pNew, pOper->rowSize); + memcpy(pDnode, pNew, pRow->rowSize); free(pNew); } mnodeDecDnodeRef(pDnode); @@ -134,19 +134,19 @@ static int32_t mnodeDnodeActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeDnodeActionEncode(SSdbOper *pOper) { - SDnodeObj *pDnode = pOper->pObj; - memcpy(pOper->rowData, pDnode, tsDnodeUpdateSize); - pOper->rowSize = tsDnodeUpdateSize; +static int32_t mnodeDnodeActionEncode(SSdbRow *pRow) { + SDnodeObj *pDnode = pRow->pObj; + memcpy(pRow->rowData, pDnode, tsDnodeUpdateSize); + pRow->rowSize = tsDnodeUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeDnodeActionDecode(SSdbOper *pOper) { +static int32_t mnodeDnodeActionDecode(SSdbRow *pRow) { SDnodeObj *pDnode = (SDnodeObj *) calloc(1, sizeof(SDnodeObj)); if (pDnode == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pDnode, pOper->rowData, tsDnodeUpdateSize); - pOper->pObj = pDnode; + memcpy(pDnode, pRow->rowData, tsDnodeUpdateSize); + pRow->pObj = pDnode; return TSDB_CODE_SUCCESS; } @@ -171,23 +171,23 @@ int32_t mnodeInitDnodes() { tsDnodeUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; pthread_mutex_init(&tsDnodeEpsMutex, NULL); - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_DNODE, - .tableName = "dnodes", + SSdbTableDesc desc = { + .id = SDB_TABLE_DNODE, + .name = "dnodes", .hashSessions = TSDB_DEFAULT_DNODES_HASH_SIZE, .maxRowSize = tsDnodeUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_AUTO, - .insertFp = mnodeDnodeActionInsert, - .deleteFp = mnodeDnodeActionDelete, - .updateFp = mnodeDnodeActionUpdate, - .encodeFp = mnodeDnodeActionEncode, - .decodeFp = mnodeDnodeActionDecode, - .destroyFp = mnodeDnodeActionDestroy, - .restoredFp = mnodeDnodeActionRestored + .fpInsert = mnodeDnodeActionInsert, + .fpDelete = mnodeDnodeActionDelete, + .fpUpdate = mnodeDnodeActionUpdate, + .fpEncode = mnodeDnodeActionEncode, + .fpDecode = mnodeDnodeActionDecode, + .fpDestroy = mnodeDnodeActionDestroy, + .fpRestored = mnodeDnodeActionRestored }; - tsDnodeSdb = sdbOpenTable(&tableDesc); + tsDnodeSdb = sdbOpenTable(&desc); if (tsDnodeSdb == NULL) { mError("failed to init dnodes data"); return -1; @@ -206,6 +206,7 @@ int32_t mnodeInitDnodes() { mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_VNODES, mnodeRetrieveVnodes); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_DNODE, mnodeGetDnodeMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_DNODE, mnodeRetrieveDnodes); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_DNODE, mnodeCancelGetNextDnode); mDebug("table:dnodes table is created"); return 0; @@ -223,6 +224,10 @@ void *mnodeGetNextDnode(void *pIter, SDnodeObj **pDnode) { return sdbFetchRow(tsDnodeSdb, pIter, (void **)pDnode); } +void mnodeCancelGetNextDnode(void *pIter) { + sdbFreeIter(tsDnodeSdb, pIter); +} + int32_t mnodeGetDnodesNum() { return sdbGetNumOfRows(tsDnodeSdb); } @@ -241,8 +246,6 @@ int32_t mnodeGetOnlinDnodesCpuCoreNum() { mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); - if (cpuCores < 2) cpuCores = 2; return cpuCores; } @@ -259,8 +262,6 @@ int32_t mnodeGetOnlineDnodesNum() { mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); - return onlineDnodes; } @@ -276,13 +277,12 @@ void *mnodeGetDnodeByEp(char *ep) { pIter = mnodeGetNextDnode(pIter, &pDnode); if (pDnode == NULL) break; if (strcmp(ep, pDnode->dnodeEp) == 0) { - sdbFreeIter(pIter); + mnodeCancelGetNextDnode(pIter); return pDnode; } mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); return NULL; } @@ -296,13 +296,13 @@ void mnodeDecDnodeRef(SDnodeObj *pDnode) { } void mnodeUpdateDnode(SDnodeObj *pDnode) { - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDnodeSdb, - .pObj = pDnode + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDnodeSdb, + .pObj = pDnode }; - int32_t code = sdbUpdateRow(&oper); + int32_t code = sdbUpdateRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("dnodeId:%d, failed update", pDnode->dnodeId); } @@ -464,7 +464,10 @@ static void mnodeUpdateDnodeEps() { while (1) { pIter = mnodeGetNextDnode(pIter, &pDnode); if (pDnode == NULL) break; - if (dnodesNum >= totalDnodes) break; + if (dnodesNum >= totalDnodes) { + mnodeCancelGetNextDnode(pIter); + break; + } SDnodeEp *pEp = &tsDnodeEps->dnodeEps[dnodesNum]; dnodesNum++; @@ -474,7 +477,6 @@ static void mnodeUpdateDnodeEps() { mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); pthread_mutex_unlock(&tsDnodeEpsMutex); } @@ -644,15 +646,15 @@ static int32_t mnodeCreateDnode(char *ep, SMnodeMsg *pMsg) { tstrncpy(pDnode->dnodeEp, ep, TSDB_EP_LEN); taosGetFqdnPortFromEp(ep, pDnode->dnodeFqdn, &pDnode->dnodePort); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDnodeSdb, - .pObj = pDnode, + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDnodeSdb, + .pObj = pDnode, .rowSize = sizeof(SDnodeObj), - .pMsg = pMsg + .pMsg = pMsg }; - int32_t code = sdbInsertRow(&oper); + int32_t code = sdbInsertRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { int dnodeId = pDnode->dnodeId; tfree(pDnode); @@ -665,14 +667,14 @@ static int32_t mnodeCreateDnode(char *ep, SMnodeMsg *pMsg) { } int32_t mnodeDropDnode(SDnodeObj *pDnode, void *pMsg) { - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsDnodeSdb, - .pObj = pDnode, - .pMsg = pMsg + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsDnodeSdb, + .pObj = pDnode, + .pMsg = pMsg }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("dnode:%d, failed to drop from cluster, result:%s", pDnode->dnodeId, tstrerror(code)); } else { @@ -1100,7 +1102,7 @@ static int32_t mnodeGetVnodeMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pC pDnode = mnodeGetDnodeByEp(pShow->payload); } else { void *pIter = mnodeGetNextDnode(NULL, (SDnodeObj **)&pDnode); - sdbFreeIter(pIter); + mnodeCancelGetNextDnode(pIter); } if (pDnode != NULL) { @@ -1141,14 +1143,13 @@ static int32_t mnodeRetrieveVnodes(SShowObj *pShow, char *data, int32_t rows, vo cols++; pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows; - strcpy(pWrite, mnodeGetMnodeRoleStr(pVgid->role)); + strcpy(pWrite, syncRole[pVgid->role]); cols++; } } mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); } else { numOfRows = 0; } @@ -1217,8 +1218,6 @@ int32_t balanceAllocVnodes(SVgObj *pVgroup) { mnodeDecDnodeRef(pDnode); } - sdbFreeIter(pIter); - if (pSelDnode == NULL) { mError("failed to alloc vnode to vgroup"); return TSDB_CODE_MND_NO_ENOUGH_DNODES; diff --git a/src/mnode/src/mnodeInt.c b/src/mnode/src/mnodeInt.c index 8431baf4f4..534deec09a 100644 --- a/src/mnode/src/mnodeInt.c +++ b/src/mnode/src/mnodeInt.c @@ -48,7 +48,7 @@ void *mnodeCreateMsg(SRpcMsg *pRpcMsg) { int32_t mnodeInitMsg(SMnodeMsg *pMsg) { if (pMsg->pUser != NULL) { - mDebug("app:%p:%p, user info already inited", pMsg->rpcMsg.ahandle, pMsg); + mTrace("msg:%p, app:%p user info already inited", pMsg, pMsg->rpcMsg.ahandle); return TSDB_CODE_SUCCESS; } diff --git a/src/mnode/src/mnodeMnode.c b/src/mnode/src/mnodeMnode.c index 092f246c13..8e3ad0a248 100644 --- a/src/mnode/src/mnodeMnode.c +++ b/src/mnode/src/mnodeMnode.c @@ -58,13 +58,13 @@ static int32_t mnodeRetrieveMnodes(SShowObj *pShow, char *data, int32_t rows, vo #define mnodeMnodeDestroyLock() pthread_mutex_destroy(&tsMnodeLock) #endif -static int32_t mnodeMnodeActionDestroy(SSdbOper *pOper) { - tfree(pOper->pObj); +static int32_t mnodeMnodeActionDestroy(SSdbRow *pRow) { + tfree(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeMnodeActionInsert(SSdbOper *pOper) { - SMnodeObj *pMnode = pOper->pObj; +static int32_t mnodeMnodeActionInsert(SSdbRow *pRow) { + SMnodeObj *pMnode = pRow->pObj; SDnodeObj *pDnode = mnodeGetDnode(pMnode->mnodeId); if (pDnode == NULL) return TSDB_CODE_MND_DNODE_NOT_EXIST; @@ -76,8 +76,8 @@ static int32_t mnodeMnodeActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeMnodeActionDelete(SSdbOper *pOper) { - SMnodeObj *pMnode = pOper->pObj; +static int32_t mnodeMnodeActionDelete(SSdbRow *pRow) { + SMnodeObj *pMnode = pRow->pObj; SDnodeObj *pDnode = mnodeGetDnode(pMnode->mnodeId); if (pDnode == NULL) return TSDB_CODE_MND_DNODE_NOT_EXIST; @@ -88,30 +88,30 @@ static int32_t mnodeMnodeActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeMnodeActionUpdate(SSdbOper *pOper) { - SMnodeObj *pMnode = pOper->pObj; +static int32_t mnodeMnodeActionUpdate(SSdbRow *pRow) { + SMnodeObj *pMnode = pRow->pObj; SMnodeObj *pSaved = mnodeGetMnode(pMnode->mnodeId); if (pMnode != pSaved) { - memcpy(pSaved, pMnode, pOper->rowSize); + memcpy(pSaved, pMnode, pRow->rowSize); free(pMnode); } mnodeDecMnodeRef(pSaved); return TSDB_CODE_SUCCESS; } -static int32_t mnodeMnodeActionEncode(SSdbOper *pOper) { - SMnodeObj *pMnode = pOper->pObj; - memcpy(pOper->rowData, pMnode, tsMnodeUpdateSize); - pOper->rowSize = tsMnodeUpdateSize; +static int32_t mnodeMnodeActionEncode(SSdbRow *pRow) { + SMnodeObj *pMnode = pRow->pObj; + memcpy(pRow->rowData, pMnode, tsMnodeUpdateSize); + pRow->rowSize = tsMnodeUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeMnodeActionDecode(SSdbOper *pOper) { +static int32_t mnodeMnodeActionDecode(SSdbRow *pRow) { SMnodeObj *pMnode = calloc(1, sizeof(SMnodeObj)); if (pMnode == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pMnode, pOper->rowData, tsMnodeUpdateSize); - pOper->pObj = pMnode; + memcpy(pMnode, pRow->rowData, tsMnodeUpdateSize); + pRow->pObj = pMnode; return TSDB_CODE_SUCCESS; } @@ -123,7 +123,7 @@ static int32_t mnodeMnodeActionRestored() { pMnode->role = TAOS_SYNC_ROLE_MASTER; mnodeDecMnodeRef(pMnode); } - sdbFreeIter(pIter); + mnodeCancelGetNextMnode(pIter); } mnodeUpdateMnodeEpSet(); @@ -137,23 +137,23 @@ int32_t mnodeInitMnodes() { SMnodeObj tObj; tsMnodeUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_MNODE, - .tableName = "mnodes", + SSdbTableDesc desc = { + .id = SDB_TABLE_MNODE, + .name = "mnodes", .hashSessions = TSDB_DEFAULT_MNODES_HASH_SIZE, .maxRowSize = tsMnodeUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_INT, - .insertFp = mnodeMnodeActionInsert, - .deleteFp = mnodeMnodeActionDelete, - .updateFp = mnodeMnodeActionUpdate, - .encodeFp = mnodeMnodeActionEncode, - .decodeFp = mnodeMnodeActionDecode, - .destroyFp = mnodeMnodeActionDestroy, - .restoredFp = mnodeMnodeActionRestored + .fpInsert = mnodeMnodeActionInsert, + .fpDelete = mnodeMnodeActionDelete, + .fpUpdate = mnodeMnodeActionUpdate, + .fpEncode = mnodeMnodeActionEncode, + .fpDecode = mnodeMnodeActionDecode, + .fpDestroy = mnodeMnodeActionDestroy, + .fpRestored = mnodeMnodeActionRestored }; - tsMnodeSdb = sdbOpenTable(&tableDesc); + tsMnodeSdb = sdbOpenTable(&desc); if (tsMnodeSdb == NULL) { mError("failed to init mnodes data"); return -1; @@ -161,6 +161,7 @@ int32_t mnodeInitMnodes() { mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_MNODE, mnodeGetMnodeMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_MNODE, mnodeRetrieveMnodes); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_MNODE, mnodeCancelGetNextMnode); mDebug("table:mnodes table is created"); return TSDB_CODE_SUCCESS; @@ -192,8 +193,8 @@ void *mnodeGetNextMnode(void *pIter, SMnodeObj **pMnode) { return sdbFetchRow(tsMnodeSdb, pIter, (void **)pMnode); } -char *mnodeGetMnodeRoleStr(int32_t role) { - return syncRole[role]; +void mnodeCancelGetNextMnode(void *pIter) { + sdbFreeIter(tsMnodeSdb, pIter); } void mnodeUpdateMnodeEpSet() { @@ -243,8 +244,6 @@ void mnodeUpdateMnodeEpSet() { tsMnodeEpSetForShell.numOfEps = index; tsMnodeEpSetForPeer.numOfEps = index; - sdbFreeIter(pIter); - mnodeMnodeUnLock(); } @@ -329,11 +328,11 @@ void mnodeCreateMnode(int32_t dnodeId, char *dnodeEp, bool needConfirm) { pMnode->mnodeId = dnodeId; pMnode->createdTime = taosGetTimestampMs(); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsMnodeSdb, - .pObj = pMnode, - .writeCb = mnodeCreateMnodeCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsMnodeSdb, + .pObj = pMnode, + .fpRsp = mnodeCreateMnodeCb }; int32_t code = TSDB_CODE_SUCCESS; @@ -346,7 +345,7 @@ void mnodeCreateMnode(int32_t dnodeId, char *dnodeEp, bool needConfirm) { return; } - code = sdbInsertRow(&oper); + code = sdbInsertRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("dnode:%d, failed to create mnode, ep:%s reason:%s", dnodeId, dnodeEp, tstrerror(code)); tfree(pMnode); @@ -356,8 +355,8 @@ void mnodeCreateMnode(int32_t dnodeId, char *dnodeEp, bool needConfirm) { void mnodeDropMnodeLocal(int32_t dnodeId) { SMnodeObj *pMnode = mnodeGetMnode(dnodeId); if (pMnode != NULL) { - SSdbOper oper = {.type = SDB_OPER_LOCAL, .table = tsMnodeSdb, .pObj = pMnode}; - sdbDeleteRow(&oper); + SSdbRow row = {.type = SDB_OPER_LOCAL, .pTable = tsMnodeSdb, .pObj = pMnode}; + sdbDeleteRow(&row); mnodeDecMnodeRef(pMnode); } @@ -371,13 +370,13 @@ int32_t mnodeDropMnode(int32_t dnodeId) { return TSDB_CODE_MND_DNODE_NOT_EXIST; } - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsMnodeSdb, - .pObj = pMnode + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsMnodeSdb, + .pObj = pMnode }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); sdbDecRef(tsMnodeSdb, pMnode); @@ -469,7 +468,7 @@ static int32_t mnodeRetrieveMnodes(SShowObj *pShow, char *data, int32_t rows, vo cols++; pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows; - char* roles = mnodeGetMnodeRoleStr(pMnode->role); + char* roles = syncRole[pMnode->role]; STR_WITH_MAXSIZE_TO_VARSTR(pWrite, roles, pShow->bytes[cols]); cols++; diff --git a/src/mnode/src/mnodePeer.c b/src/mnode/src/mnodePeer.c index e43f8c1b78..f13ef75398 100644 --- a/src/mnode/src/mnodePeer.c +++ b/src/mnode/src/mnodePeer.c @@ -47,7 +47,7 @@ void mnodeAddPeerRspHandle(uint8_t msgType, void (*fp)(SRpcMsg *rpcMsg)) { int32_t mnodeProcessPeerReq(SMnodeMsg *pMsg) { if (pMsg->rpcMsg.pCont == NULL) { - mError("%p, msg:%s in mpeer queue, content is null", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); + mError("msg:%p, ahandle:%p type:%s in mpeer queue, content is null", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_INVALID_MSG_LEN; } @@ -58,8 +58,8 @@ int32_t mnodeProcessPeerReq(SMnodeMsg *pMsg) { rpcRsp->rsp = epSet; rpcRsp->len = sizeof(SRpcEpSet); - mDebug("%p, msg:%s in mpeer queue will be redirected, numOfEps:%d inUse:%d", pMsg->rpcMsg.ahandle, - taosMsg[pMsg->rpcMsg.msgType], epSet->numOfEps, epSet->inUse); + mDebug("msg:%p, ahandle:%p type:%s in mpeer queue will be redirected, numOfEps:%d inUse:%d", pMsg, + pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], epSet->numOfEps, epSet->inUse); for (int32_t i = 0; i < epSet->numOfEps; ++i) { if (strcmp(epSet->fqdn[i], tsLocalFqdn) == 0 && htons(epSet->port[i]) == tsServerPort + TSDB_PORT_DNODEDNODE) { epSet->inUse = (i + 1) % epSet->numOfEps; @@ -73,7 +73,8 @@ int32_t mnodeProcessPeerReq(SMnodeMsg *pMsg) { } if (tsMnodeProcessPeerMsgFp[pMsg->rpcMsg.msgType] == NULL) { - mError("%p, msg:%s in mpeer queue, not processed", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); + mError("msg:%p, ahandle:%p type:%s in mpeer queue, not processed", pMsg, pMsg->rpcMsg.ahandle, + taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_MSG_NOT_PROCESSED; } @@ -82,13 +83,14 @@ int32_t mnodeProcessPeerReq(SMnodeMsg *pMsg) { void mnodeProcessPeerRsp(SRpcMsg *pMsg) { if (!sdbIsMaster()) { - mError("%p, msg:%s is not processed for it is not master", pMsg->ahandle, taosMsg[pMsg->msgType]); + mError("msg:%p, ahandle:%p type:%s is not processed for it is not master", pMsg, pMsg->ahandle, + taosMsg[pMsg->msgType]); return; } if (tsMnodeProcessPeerRspFp[pMsg->msgType]) { (*tsMnodeProcessPeerRspFp[pMsg->msgType])(pMsg); } else { - mError("%p, msg:%s is not processed", pMsg->ahandle, taosMsg[pMsg->msgType]); + mError("msg:%p, ahandle:%p type:%s is not processed", pMsg, pMsg->ahandle, taosMsg[pMsg->msgType]); } } diff --git a/src/mnode/src/mnodeProfile.c b/src/mnode/src/mnodeProfile.c index 12ac64854c..8467adfc17 100644 --- a/src/mnode/src/mnodeProfile.c +++ b/src/mnode/src/mnodeProfile.c @@ -42,6 +42,7 @@ static int32_t mnodeGetQueryMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pC static int32_t mnodeRetrieveQueries(SShowObj *pShow, char *data, int32_t rows, void *pConn); static int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn); static int32_t mnodeRetrieveConns(SShowObj *pShow, char *data, int32_t rows, void *pConn); +static void mnodeCancelGetNextConn(void *pIter); static int32_t mnodeGetStreamMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn); static int32_t mnodeRetrieveStreams(SShowObj *pShow, char *data, int32_t rows, void *pConn); static void mnodeFreeConn(void *data); @@ -52,10 +53,13 @@ static int32_t mnodeProcessKillConnectionMsg(SMnodeMsg *pMsg); int32_t mnodeInitProfile() { mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_QUERIES, mnodeGetQueryMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_QUERIES, mnodeRetrieveQueries); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_QUERIES, mnodeCancelGetNextConn); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_CONNS, mnodeGetConnsMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_CONNS, mnodeRetrieveConns); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_CONNS, mnodeCancelGetNextConn); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_STREAMS, mnodeGetStreamMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_STREAMS, mnodeRetrieveStreams); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_STREAMS, mnodeCancelGetNextConn); mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_QUERY, mnodeProcessKillQueryMsg); mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_KILL_STREAM, mnodeProcessKillStreamMsg); @@ -137,21 +141,15 @@ static void mnodeFreeConn(void *data) { mDebug("connId:%d, is destroyed", pConn->connId); } -static void *mnodeGetNextConn(SHashMutableIterator *pIter, SConnObj **pConn) { +static void *mnodeGetNextConn(void *pIter, SConnObj **pConn) { *pConn = NULL; - if (pIter == NULL) { - pIter = taosHashCreateIter(tsMnodeConnCache->pHashTable); - } + pIter = taosHashIterate(tsMnodeConnCache->pHashTable, pIter); + if (pIter == NULL) return NULL; - if (!taosHashIterNext(pIter)) { - taosHashDestroyIter(pIter); - return NULL; - } - - SCacheDataNode **pNode = taosHashIterGet(pIter); + SCacheDataNode **pNode = pIter; if (pNode == NULL || *pNode == NULL) { - taosHashDestroyIter(pIter); + taosHashCancelIterate(tsMnodeConnCache->pHashTable, pIter); return NULL; } @@ -159,6 +157,10 @@ static void *mnodeGetNextConn(SHashMutableIterator *pIter, SConnObj **pConn) { return pIter; } +static void mnodeCancelGetNextConn(void *pIter) { + taosHashCancelIterate(tsMnodeConnCache->pHashTable, pIter); +} + static int32_t mnodeGetConnsMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) { SUserObj *pUser = mnodeGetUserFromConn(pConn); if (pUser == NULL) return 0; diff --git a/src/mnode/src/mnodeRead.c b/src/mnode/src/mnodeRead.c index 93b944febb..ea7ce783e8 100644 --- a/src/mnode/src/mnodeRead.c +++ b/src/mnode/src/mnodeRead.c @@ -43,7 +43,7 @@ void mnodeAddReadMsgHandle(uint8_t msgType, int32_t (*fp)(SMnodeMsg *pMsg)) { int32_t mnodeProcessRead(SMnodeMsg *pMsg) { if (pMsg->rpcMsg.pCont == NULL) { - mError("%p, msg:%s in mread queue, content is null", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); + mError("msg:%p, app:%p type:%s in mread queue, content is null", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_INVALID_MSG_LEN; } @@ -52,7 +52,7 @@ int32_t mnodeProcessRead(SMnodeMsg *pMsg) { SRpcEpSet *epSet = rpcMallocCont(sizeof(SRpcEpSet)); mnodeGetMnodeEpSetForShell(epSet); - mDebug("%p, msg:%s in mread queue will be redirected, numOfEps:%d inUse:%d", pMsg->rpcMsg.ahandle, + mDebug("msg:%p, app:%p type:%s in mread queue will be redirected, numOfEps:%d inUse:%d", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], epSet->numOfEps, epSet->inUse); for (int32_t i = 0; i < epSet->numOfEps; ++i) { if (strcmp(epSet->fqdn[i], tsLocalFqdn) == 0 && htons(epSet->port[i]) == tsServerPort) { @@ -70,13 +70,15 @@ int32_t mnodeProcessRead(SMnodeMsg *pMsg) { } if (tsMnodeProcessReadMsgFp[pMsg->rpcMsg.msgType] == NULL) { - mError("%p, msg:%s in mread queue, not processed", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); + mError("msg:%p, app:%p type:%s in mread queue, not processed", pMsg, pMsg->rpcMsg.ahandle, + taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_MSG_NOT_PROCESSED; } int32_t code = mnodeInitMsg(pMsg); if (code != TSDB_CODE_SUCCESS) { - mError("%p, msg:%s in mread queue, not processed reason:%s", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], tstrerror(code)); + mError("msg:%p, app:%p type:%s in mread queue, not processed reason:%s", pMsg, pMsg->rpcMsg.ahandle, + taosMsg[pMsg->rpcMsg.msgType], tstrerror(code)); return code; } diff --git a/src/mnode/src/mnodeSdb.c b/src/mnode/src/mnodeSdb.c index 003ecd0d24..44e367244c 100644 --- a/src/mnode/src/mnodeSdb.c +++ b/src/mnode/src/mnodeSdb.c @@ -34,23 +34,30 @@ #include "mnodeSdb.h" #define SDB_TABLE_LEN 12 -#define SDB_SYNC_HACK 16 +#define MAX_QUEUED_MSG_NUM 10000 typedef enum { - SDB_ACTION_INSERT, - SDB_ACTION_DELETE, - SDB_ACTION_UPDATE + SDB_ACTION_INSERT = 0, + SDB_ACTION_DELETE = 1, + SDB_ACTION_UPDATE = 2 } ESdbAction; typedef enum { - SDB_STATUS_OFFLINE, - SDB_STATUS_SERVING, - SDB_STATUS_CLOSING + SDB_STATUS_OFFLINE = 0, + SDB_STATUS_SERVING = 1, + SDB_STATUS_CLOSING = 2 } ESdbStatus; -typedef struct _SSdbTable { - char tableName[SDB_TABLE_LEN]; - ESdbTable tableId; +char *actStr[] = { + "insert", + "delete", + "update", + "invalid" +}; + +typedef struct SSdbTable { + char name[SDB_TABLE_LEN]; + ESdbTable id; ESdbKey keyType; int32_t hashSessions; int32_t maxRowSize; @@ -58,13 +65,13 @@ typedef struct _SSdbTable { int32_t autoIndex; int64_t numOfRows; void * iHandle; - int32_t (*insertFp)(SSdbOper *pDesc); - int32_t (*deleteFp)(SSdbOper *pOper); - int32_t (*updateFp)(SSdbOper *pOper); - int32_t (*decodeFp)(SSdbOper *pOper); - int32_t (*encodeFp)(SSdbOper *pOper); - int32_t (*destroyFp)(SSdbOper *pOper); - int32_t (*restoredFp)(); + int32_t (*fpInsert)(SSdbRow *pRow); + int32_t (*fpDelete)(SSdbRow *pRow); + int32_t (*fpUpdate)(SSdbRow *pRow); + int32_t (*fpDecode)(SSdbRow *pRow); + int32_t (*fpEncode)(SSdbRow *pRow); + int32_t (*fpDestroy)(SSdbRow *pRow); + int32_t (*fpRestored)(); pthread_mutex_t mutex; } SSdbTable; @@ -75,60 +82,61 @@ typedef struct { int64_t sync; void * wal; SSyncCfg cfg; + int32_t queuedMsg; int32_t numOfTables; SSdbTable *tableList[SDB_TABLE_MAX]; pthread_mutex_t mutex; -} SSdbObject; +} SSdbMgmt; typedef struct { pthread_t thread; int32_t workerId; -} SSdbWriteWorker; +} SSdbWorker; typedef struct { int32_t num; - SSdbWriteWorker *writeWorker; -} SSdbWriteWorkerPool; + SSdbWorker *worker; +} SSdbWorkerPool; extern void * tsMnodeTmr; -static void * tsUpdateSyncTmr; -static SSdbObject tsSdbObj = {0}; -static taos_qset tsSdbWriteQset; -static taos_qall tsSdbWriteQall; -static taos_queue tsSdbWriteQueue; -static SSdbWriteWorkerPool tsSdbPool; +static void * tsSdbTmr; +static SSdbMgmt tsSdbMgmt = {0}; +static taos_qset tsSdbWQset; +static taos_qall tsSdbWQall; +static taos_queue tsSdbWQueue; +static SSdbWorkerPool tsSdbPool; -static int32_t sdbWrite(void *param, void *data, int32_t type, void *pMsg); -static int32_t sdbWriteToQueue(void *param, void *data, int32_t type, void *pMsg); -static void * sdbWorkerFp(void *param); -static int32_t sdbInitWriteWorker(); -static void sdbCleanupWriteWorker(); -static int32_t sdbAllocWriteQueue(); -static void sdbFreeWritequeue(); -static int32_t sdbUpdateRowImp(SSdbOper *pOper); -static int32_t sdbDeleteRowImp(SSdbOper *pOper); -static int32_t sdbInsertHash(SSdbTable *pTable, SSdbOper *pOper); -static int32_t sdbUpdateHash(SSdbTable *pTable, SSdbOper *pOper); -static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbOper *pOper); +static int32_t sdbProcessWrite(void *pRow, void *pHead, int32_t qtype, void *unused); +static int32_t sdbWriteFwdToQueue(int32_t vgId, void *pHead, int32_t qtype, void *rparam); +static int32_t sdbWriteRowToQueue(SSdbRow *pRow, int32_t action); +static void sdbFreeFromQueue(SSdbRow *pRow); +static void * sdbWorkerFp(void *pWorker); +static int32_t sdbInitWorker(); +static void sdbCleanupWorker(); +static int32_t sdbAllocQueue(); +static void sdbFreeQueue(); +static int32_t sdbInsertHash(SSdbTable *pTable, SSdbRow *pRow); +static int32_t sdbUpdateHash(SSdbTable *pTable, SSdbRow *pRow); +static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbRow *pRow); -int32_t sdbGetId(void *handle) { - return ((SSdbTable *)handle)->autoIndex; +int32_t sdbGetId(void *pTable) { + return ((SSdbTable *)pTable)->autoIndex; } -int64_t sdbGetNumOfRows(void *handle) { - return ((SSdbTable *)handle)->numOfRows; +int64_t sdbGetNumOfRows(void *pTable) { + return ((SSdbTable *)pTable)->numOfRows; } uint64_t sdbGetVersion() { - return tsSdbObj.version; + return tsSdbMgmt.version; } bool sdbIsMaster() { - return tsSdbObj.role == TAOS_SYNC_ROLE_MASTER; + return tsSdbMgmt.role == TAOS_SYNC_ROLE_MASTER; } bool sdbIsServing() { - return tsSdbObj.status == SDB_STATUS_SERVING; + return tsSdbMgmt.status == SDB_STATUS_SERVING; } static void *sdbGetObjKey(SSdbTable *pTable, void *key) { @@ -139,18 +147,6 @@ static void *sdbGetObjKey(SSdbTable *pTable, void *key) { return key; } -static char *sdbGetActionStr(int32_t action) { - switch (action) { - case SDB_ACTION_INSERT: - return "insert"; - case SDB_ACTION_DELETE: - return "delete"; - case SDB_ACTION_UPDATE: - return "update"; - } - return "invalid"; -} - static char *sdbGetKeyStr(SSdbTable *pTable, void *key) { static char str[16]; switch (pTable->keyType) { @@ -166,28 +162,28 @@ static char *sdbGetKeyStr(SSdbTable *pTable, void *key) { } } -static char *sdbGetKeyStrFromObj(SSdbTable *pTable, void *key) { +static char *sdbGetRowStr(SSdbTable *pTable, void *key) { return sdbGetKeyStr(pTable, sdbGetObjKey(pTable, key)); } static void *sdbGetTableFromId(int32_t tableId) { - return tsSdbObj.tableList[tableId]; + return tsSdbMgmt.tableList[tableId]; } static int32_t sdbInitWal() { SWalCfg walCfg = {.vgId = 1, .walLevel = TAOS_WAL_FSYNC, .keep = TAOS_WAL_KEEP, .fsyncPeriod = 0}; - char temp[TSDB_FILENAME_LEN]; + char temp[TSDB_FILENAME_LEN] = {0}; sprintf(temp, "%s/wal", tsMnodeDir); - tsSdbObj.wal = walOpen(temp, &walCfg); - if (tsSdbObj.wal == NULL) { - sdbError("failed to open sdb wal in %s", tsMnodeDir); + tsSdbMgmt.wal = walOpen(temp, &walCfg); + if (tsSdbMgmt.wal == NULL) { + sdbError("vgId:1, failed to open wal in %s", tsMnodeDir); return -1; } - sdbInfo("open sdb wal for restore"); - int code = walRestore(tsSdbObj.wal, NULL, sdbWrite); + sdbInfo("vgId:1, open wal for restore"); + int32_t code = walRestore(tsSdbMgmt.wal, NULL, sdbProcessWrite); if (code != TSDB_CODE_SUCCESS) { - sdbError("failed to open wal for restore, reason:%s", tstrerror(code)); + sdbError("vgId:1, failed to open wal for restore since %s", tstrerror(code)); return -1; } return 0; @@ -199,31 +195,31 @@ static void sdbRestoreTables() { for (int32_t tableId = 0; tableId < SDB_TABLE_MAX; ++tableId) { SSdbTable *pTable = sdbGetTableFromId(tableId); if (pTable == NULL) continue; - if (pTable->restoredFp) { - (*pTable->restoredFp)(); + if (pTable->fpRestored) { + (*pTable->fpRestored)(); } totalRows += pTable->numOfRows; numOfTables++; - sdbDebug("table:%s, is restored, numOfRows:%" PRId64, pTable->tableName, pTable->numOfRows); + sdbDebug("vgId:1, sdb:%s is restored, rows:%" PRId64, pTable->name, pTable->numOfRows); } - sdbInfo("sdb is restored, ver:%" PRId64 " totalRows:%d numOfTables:%d", tsSdbObj.version, totalRows, numOfTables); + sdbInfo("vgId:1, sdb is restored, mver:%" PRIu64 " rows:%d tables:%d", tsSdbMgmt.version, totalRows, numOfTables); } void sdbUpdateMnodeRoles() { - if (tsSdbObj.sync <= 0) return; + if (tsSdbMgmt.sync <= 0) return; SNodesRole roles = {0}; - syncGetNodesRole(tsSdbObj.sync, &roles); + if (syncGetNodesRole(tsSdbMgmt.sync, &roles) != 0) return; - sdbInfo("update mnodes sync roles, total:%d", tsSdbObj.cfg.replica); - for (int32_t i = 0; i < tsSdbObj.cfg.replica; ++i) { + sdbInfo("vgId:1, update mnodes role, replica:%d", tsSdbMgmt.cfg.replica); + for (int32_t i = 0; i < tsSdbMgmt.cfg.replica; ++i) { SMnodeObj *pMnode = mnodeGetMnode(roles.nodeId[i]); if (pMnode != NULL) { pMnode->role = roles.role[i]; - sdbInfo("mnode:%d, role:%s", pMnode->mnodeId, mnodeGetMnodeRoleStr(pMnode->role)); - if (pMnode->mnodeId == dnodeGetDnodeId()) tsSdbObj.role = pMnode->role; + sdbInfo("vgId:1, mnode:%d, role:%s", pMnode->mnodeId, syncRole[pMnode->role]); + if (pMnode->mnodeId == dnodeGetDnodeId()) tsSdbMgmt.role = pMnode->role; mnodeDecMnodeRef(pMnode); } } @@ -232,93 +228,80 @@ void sdbUpdateMnodeRoles() { mnodeUpdateMnodeEpSet(); } -static uint32_t sdbGetFileInfo(void *ahandle, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion) { +static uint32_t sdbGetFileInfo(int32_t vgId, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion) { sdbUpdateMnodeRoles(); return 0; } -static int32_t sdbGetWalInfo(void *ahandle, char *fileName, int64_t *fileId) { - return walGetWalFile(tsSdbObj.wal, fileName, fileId); +static int32_t sdbGetWalInfo(int32_t vgId, char *fileName, int64_t *fileId) { + return walGetWalFile(tsSdbMgmt.wal, fileName, fileId); } -static void sdbNotifyRole(void *ahandle, int8_t role) { - sdbInfo("mnode role changed from %s to %s", mnodeGetMnodeRoleStr(tsSdbObj.role), mnodeGetMnodeRoleStr(role)); +static void sdbNotifyRole(int32_t vgId, int8_t role) { + sdbInfo("vgId:1, mnode role changed from %s to %s", syncRole[tsSdbMgmt.role], syncRole[role]); - if (role == TAOS_SYNC_ROLE_MASTER && tsSdbObj.role != TAOS_SYNC_ROLE_MASTER) { + if (role == TAOS_SYNC_ROLE_MASTER && tsSdbMgmt.role != TAOS_SYNC_ROLE_MASTER) { balanceReset(); } - tsSdbObj.role = role; + tsSdbMgmt.role = role; sdbUpdateMnodeRoles(); } +// failed to forward, need revert insert +static void sdbHandleFailedConfirm(SSdbRow *pRow) { + SWalHead *pHead = pRow->pHead; + int32_t action = pHead->msgType % 10; + + sdbError("vgId:1, row:%p:%s hver:%" PRIu64 " action:%s, failed to foward since %s", pRow->pObj, + sdbGetKeyStr(pRow->pTable, pHead->cont), pHead->version, actStr[action], tstrerror(pRow->code)); + + // It's better to create a table in two stages, create it first and then set it success + if (action == SDB_ACTION_INSERT) { + SSdbRow row = {.type = SDB_OPER_GLOBAL, .pTable = pRow->pTable, .pObj = pRow->pObj}; + sdbDeleteRow(&row); + } +} + FORCE_INLINE -static void sdbConfirmForward(void *ahandle, void *param, int32_t code) { - assert(param); - SSdbOper * pOper = param; - SMnodeMsg *pMsg = pOper->pMsg; - if (code <= 0) pOper->retCode = code; +static void sdbConfirmForward(int32_t vgId, void *wparam, int32_t code) { + if (wparam == NULL) return; + SSdbRow *pRow = wparam; + SMnodeMsg * pMsg = pRow->pMsg; - int32_t processedCount = atomic_add_fetch_32(&pOper->processedCount, 1); - if (processedCount <= 1) { - if (pMsg != NULL) { - sdbDebug("app:%p:%p, waiting for confirm this operation, count:%d result:%s", pMsg->rpcMsg.ahandle, pMsg, - processedCount, tstrerror(code)); - } + if (code <= 0) pRow->code = code; + int32_t count = atomic_add_fetch_32(&pRow->processedCount, 1); + if (count <= 1) { + if (pMsg != NULL) sdbTrace("vgId:1, msg:%p waiting for confirm, count:%d code:%x", pMsg, count, code); return; + } else { + if (pMsg != NULL) sdbTrace("vgId:1, msg:%p is confirmed, code:%x", pMsg, code); } - if (pMsg != NULL) { - sdbDebug("app:%p:%p, is confirmed and will do callback func, result:%s", pMsg->rpcMsg.ahandle, pMsg, - tstrerror(code)); + if (pRow->code != TSDB_CODE_SUCCESS) sdbHandleFailedConfirm(pRow); + + if (pRow->fpRsp != NULL) { + pRow->code = (*pRow->fpRsp)(pMsg, pRow->code); } - // failed to forward, need revert insert - if (pOper->retCode != TSDB_CODE_SUCCESS) { - SWalHead *pHead = (void *)pOper + sizeof(SSdbOper) + SDB_SYNC_HACK; - int32_t action = pHead->msgType % 10; - sdbError("table:%s record:%p:%s ver:%" PRIu64 ", action:%d failed to foward reason:%s", - ((SSdbTable *)pOper->table)->tableName, pOper->pObj, sdbGetKeyStr(pOper->table, pHead->cont), - pHead->version, action, tstrerror(pOper->retCode)); - if (action == SDB_ACTION_INSERT) { - // It's better to create a table in two stages, create it first and then set it success - //sdbDeleteHash(pOper->table, pOper); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = pOper->table, - .pObj = pOper->pObj - }; - sdbDeleteRow(&oper); - } - } - - if (pOper->writeCb != NULL) { - pOper->retCode = (*pOper->writeCb)(pMsg, pOper->retCode); - } - dnodeSendRpcMWriteRsp(pMsg, pOper->retCode); - - // if ahandle, means this func is called by sdb write - if (ahandle == NULL) { - sdbDecRef(pOper->table, pOper->pObj); - } - - taosFreeQitem(pOper); + dnodeSendRpcMWriteRsp(pMsg, pRow->code); + sdbFreeFromQueue(pRow); } static void sdbUpdateSyncTmrFp(void *param, void *tmrId) { sdbUpdateSync(NULL); } void sdbUpdateAsync() { - taosTmrReset(sdbUpdateSyncTmrFp, 200, NULL, tsMnodeTmr, &tsUpdateSyncTmr); + taosTmrReset(sdbUpdateSyncTmrFp, 200, NULL, tsMnodeTmr, &tsSdbTmr); } void sdbUpdateSync(void *pMnodes) { SMnodeInfos *mnodes = pMnodes; if (!mnodeIsRunning()) { - mDebug("mnode not start yet, update sync config later"); + mDebug("vgId:1, mnode not start yet, update sync config later"); return; } - mDebug("update sync config in sync module, mnodes:%p", pMnodes); + mDebug("vgId:1, update sync config in sync module, mnodes:%p", pMnodes); SSyncCfg syncCfg = {0}; int32_t index = 0; @@ -342,9 +325,8 @@ void sdbUpdateSync(void *pMnodes) { mnodeDecDnodeRef(pDnode); mnodeDecMnodeRef(pMnode); } - sdbFreeIter(pIter); syncCfg.replica = index; - mDebug("mnodes info not input, use infos in sdb, numOfMnodes:%d", syncCfg.replica); + mDebug("vgId:1, mnodes info not input, use infos in sdb, numOfMnodes:%d", syncCfg.replica); } else { for (index = 0; index < mnodes->mnodeNum; ++index) { SMnodeInfo *node = &mnodes->mnodeInfos[index]; @@ -353,7 +335,7 @@ void sdbUpdateSync(void *pMnodes) { syncCfg.nodeInfo[index].nodePort += TSDB_PORT_SYNC; } syncCfg.replica = index; - mDebug("mnodes info input, numOfMnodes:%d", syncCfg.replica); + mDebug("vgId:1, mnodes info input, numOfMnodes:%d", syncCfg.replica); } syncCfg.quorum = (syncCfg.replica == 1) ? 1 : 2; @@ -367,18 +349,19 @@ void sdbUpdateSync(void *pMnodes) { } if (!hasThisDnode) { - sdbDebug("update sync config, this dnode not exist"); + sdbDebug("vgId:1, update sync config, this dnode not exist"); return; } - if (memcmp(&syncCfg, &tsSdbObj.cfg, sizeof(SSyncCfg)) == 0) { - sdbDebug("update sync config, info not changed"); + if (memcmp(&syncCfg, &tsSdbMgmt.cfg, sizeof(SSyncCfg)) == 0) { + sdbDebug("vgId:1, update sync config, info not changed"); return; } - sdbInfo("work as mnode, replica:%d", syncCfg.replica); + sdbInfo("vgId:1, work as mnode, replica:%d", syncCfg.replica); for (int32_t i = 0; i < syncCfg.replica; ++i) { - sdbInfo("mnode:%d, %s:%d", syncCfg.nodeInfo[i].nodeId, syncCfg.nodeInfo[i].nodeFqdn, syncCfg.nodeInfo[i].nodePort); + sdbInfo("vgId:1, mnode:%d, %s:%d", syncCfg.nodeInfo[i].nodeId, syncCfg.nodeInfo[i].nodeFqdn, + syncCfg.nodeInfo[i].nodePort); } SSyncInfo syncInfo = {0}; @@ -386,26 +369,26 @@ void sdbUpdateSync(void *pMnodes) { syncInfo.version = sdbGetVersion(); syncInfo.syncCfg = syncCfg; sprintf(syncInfo.path, "%s", tsMnodeDir); - syncInfo.ahandle = NULL; syncInfo.getWalInfo = sdbGetWalInfo; syncInfo.getFileInfo = sdbGetFileInfo; - syncInfo.writeToCache = sdbWriteToQueue; + syncInfo.writeToCache = sdbWriteFwdToQueue; syncInfo.confirmForward = sdbConfirmForward; syncInfo.notifyRole = sdbNotifyRole; - tsSdbObj.cfg = syncCfg; + tsSdbMgmt.cfg = syncCfg; - if (tsSdbObj.sync) { - syncReconfig(tsSdbObj.sync, &syncCfg); + if (tsSdbMgmt.sync) { + syncReconfig(tsSdbMgmt.sync, &syncCfg); } else { - tsSdbObj.sync = syncStart(&syncInfo); + tsSdbMgmt.sync = syncStart(&syncInfo); } + sdbUpdateMnodeRoles(); } int32_t sdbInit() { - pthread_mutex_init(&tsSdbObj.mutex, NULL); + pthread_mutex_init(&tsSdbMgmt.mutex, NULL); - if (sdbInitWriteWorker() != 0) { + if (sdbInitWorker() != 0) { return -1; } @@ -416,56 +399,56 @@ int32_t sdbInit() { sdbRestoreTables(); if (mnodeGetMnodesNum() == 1) { - tsSdbObj.role = TAOS_SYNC_ROLE_MASTER; + tsSdbMgmt.role = TAOS_SYNC_ROLE_MASTER; } - tsSdbObj.status = SDB_STATUS_SERVING; + tsSdbMgmt.status = SDB_STATUS_SERVING; return TSDB_CODE_SUCCESS; } void sdbCleanUp() { - if (tsSdbObj.status != SDB_STATUS_SERVING) return; + if (tsSdbMgmt.status != SDB_STATUS_SERVING) return; - tsSdbObj.status = SDB_STATUS_CLOSING; - - sdbCleanupWriteWorker(); - sdbDebug("sdb will be closed, ver:%" PRId64, tsSdbObj.version); + tsSdbMgmt.status = SDB_STATUS_CLOSING; - if (tsSdbObj.sync) { - syncStop(tsSdbObj.sync); - tsSdbObj.sync = -1; + sdbCleanupWorker(); + sdbDebug("vgId:1, sdb will be closed, mver:%" PRIu64, tsSdbMgmt.version); + + if (tsSdbMgmt.sync) { + syncStop(tsSdbMgmt.sync); + tsSdbMgmt.sync = -1; } - if (tsSdbObj.wal) { - walClose(tsSdbObj.wal); - tsSdbObj.wal = NULL; + if (tsSdbMgmt.wal) { + walClose(tsSdbMgmt.wal); + tsSdbMgmt.wal = NULL; } - pthread_mutex_destroy(&tsSdbObj.mutex); + pthread_mutex_destroy(&tsSdbMgmt.mutex); } -void sdbIncRef(void *handle, void *pObj) { - if (pObj == NULL || handle == NULL) return; +void sdbIncRef(void *tparam, void *pRow) { + if (pRow == NULL || tparam == NULL) return; - SSdbTable *pTable = handle; - int32_t * pRefCount = (int32_t *)(pObj + pTable->refCountPos); + SSdbTable *pTable = tparam; + int32_t * pRefCount = (int32_t *)(pRow + pTable->refCountPos); int32_t refCount = atomic_add_fetch_32(pRefCount, 1); - sdbTrace("add ref to table:%s record:%p:%s:%d", pTable->tableName, pObj, sdbGetKeyStrFromObj(pTable, pObj), refCount); + sdbTrace("vgId:1, sdb:%s, inc ref to row:%p:%s:%d", pTable->name, pRow, sdbGetRowStr(pTable, pRow), refCount); } -void sdbDecRef(void *handle, void *pObj) { - if (pObj == NULL || handle == NULL) return; +void sdbDecRef(void *tparam, void *pRow) { + if (pRow == NULL || tparam == NULL) return; - SSdbTable *pTable = handle; - int32_t * pRefCount = (int32_t *)(pObj + pTable->refCountPos); + SSdbTable *pTable = tparam; + int32_t * pRefCount = (int32_t *)(pRow + pTable->refCountPos); int32_t refCount = atomic_sub_fetch_32(pRefCount, 1); - sdbTrace("def ref of table:%s record:%p:%s:%d", pTable->tableName, pObj, sdbGetKeyStrFromObj(pTable, pObj), refCount); + sdbTrace("vgId:1, sdb:%s, dec ref to row:%p:%s:%d", pTable->name, pRow, sdbGetRowStr(pTable, pRow), refCount); - int32_t *updateEnd = pObj + pTable->refCountPos - 4; + int32_t *updateEnd = pRow + pTable->refCountPos - 4; if (refCount <= 0 && *updateEnd) { - sdbTrace("table:%s, record:%p:%s:%d is destroyed", pTable->tableName, pObj, sdbGetKeyStrFromObj(pTable, pObj), refCount); - SSdbOper oper = {.pObj = pObj}; - (*pTable->destroyFp)(&oper); + sdbTrace("vgId:1, sdb:%s, row:%p:%s:%d destroyed", pTable->name, pRow, sdbGetRowStr(pTable, pRow), refCount); + SSdbRow row = {.pObj = pRow}; + (*pTable->fpDestroy)(&row); } } @@ -487,12 +470,12 @@ static void *sdbGetRowMetaFromObj(SSdbTable *pTable, void *key) { return sdbGetRowMeta(pTable, sdbGetObjKey(pTable, key)); } -void *sdbGetRow(void *handle, void *key) { - SSdbTable *pTable = handle; +void *sdbGetRow(void *tparam, void *key) { + SSdbTable *pTable = tparam; pthread_mutex_lock(&pTable->mutex); - void *pRow = sdbGetRowMeta(handle, key); - if (pRow) sdbIncRef(handle, pRow); + void *pRow = sdbGetRowMeta(pTable, key); + if (pRow) sdbIncRef(pTable, pRow); pthread_mutex_unlock(&pTable->mutex); return pRow; @@ -502,8 +485,8 @@ static void *sdbGetRowFromObj(SSdbTable *pTable, void *key) { return sdbGetRow(pTable, sdbGetObjKey(pTable, key)); } -static int32_t sdbInsertHash(SSdbTable *pTable, SSdbOper *pOper) { - void * key = sdbGetObjKey(pTable, pOper->pObj); +static int32_t sdbInsertHash(SSdbTable *pTable, SSdbRow *pRow) { + void * key = sdbGetObjKey(pTable, pRow->pObj); int32_t keySize = sizeof(int32_t); if (pTable->keyType == SDB_KEY_STRING || pTable->keyType == SDB_KEY_VAR_STRING) { @@ -511,43 +494,43 @@ static int32_t sdbInsertHash(SSdbTable *pTable, SSdbOper *pOper) { } pthread_mutex_lock(&pTable->mutex); - taosHashPut(pTable->iHandle, key, keySize, &pOper->pObj, sizeof(int64_t)); + taosHashPut(pTable->iHandle, key, keySize, &pRow->pObj, sizeof(int64_t)); pthread_mutex_unlock(&pTable->mutex); - sdbIncRef(pTable, pOper->pObj); + sdbIncRef(pTable, pRow->pObj); atomic_add_fetch_32(&pTable->numOfRows, 1); if (pTable->keyType == SDB_KEY_AUTO) { - pTable->autoIndex = MAX(pTable->autoIndex, *((uint32_t *)pOper->pObj)); + pTable->autoIndex = MAX(pTable->autoIndex, *((uint32_t *)pRow->pObj)); } else { atomic_add_fetch_32(&pTable->autoIndex, 1); } - sdbDebug("table:%s, insert record:%s to hash, rowSize:%d numOfRows:%" PRId64 ", msg:%p", pTable->tableName, - sdbGetKeyStrFromObj(pTable, pOper->pObj), pOper->rowSize, pTable->numOfRows, pOper->pMsg); + sdbDebug("vgId:1, sdb:%s, insert key:%s to hash, rowSize:%d rows:%" PRId64 ", msg:%p", pTable->name, + sdbGetRowStr(pTable, pRow->pObj), pRow->rowSize, pTable->numOfRows, pRow->pMsg); - int32_t code = (*pTable->insertFp)(pOper); + int32_t code = (*pTable->fpInsert)(pRow); if (code != TSDB_CODE_SUCCESS) { - sdbError("table:%s, failed to insert record:%s to hash, remove it", pTable->tableName, - sdbGetKeyStrFromObj(pTable, pOper->pObj)); - sdbDeleteHash(pTable, pOper); + sdbError("vgId:1, sdb:%s, failed to insert key:%s to hash, remove it", pTable->name, + sdbGetRowStr(pTable, pRow->pObj)); + sdbDeleteHash(pTable, pRow); } return TSDB_CODE_SUCCESS; } -static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbOper *pOper) { - int32_t *updateEnd = pOper->pObj + pTable->refCountPos - 4; +static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbRow *pRow) { + int32_t *updateEnd = pRow->pObj + pTable->refCountPos - 4; bool set = atomic_val_compare_exchange_32(updateEnd, 0, 1) == 0; if (!set) { - sdbError("table:%s, failed to delete record:%s from hash, for it already removed", pTable->tableName, - sdbGetKeyStrFromObj(pTable, pOper->pObj)); + sdbError("vgId:1, sdb:%s, failed to delete key:%s from hash, for it already removed", pTable->name, + sdbGetRowStr(pTable, pRow->pObj)); return TSDB_CODE_MND_SDB_OBJ_NOT_THERE; } - (*pTable->deleteFp)(pOper); + (*pTable->fpDelete)(pRow); - void * key = sdbGetObjKey(pTable, pOper->pObj); + void * key = sdbGetObjKey(pTable, pRow->pObj); int32_t keySize = sizeof(int32_t); if (pTable->keyType == SDB_KEY_STRING || pTable->keyType == SDB_KEY_VAR_STRING) { keySize = strlen((char *)key); @@ -558,339 +541,264 @@ static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbOper *pOper) { pthread_mutex_unlock(&pTable->mutex); atomic_sub_fetch_32(&pTable->numOfRows, 1); - - sdbDebug("table:%s, delete record:%s from hash, numOfRows:%" PRId64 ", msg:%p", pTable->tableName, - sdbGetKeyStrFromObj(pTable, pOper->pObj), pTable->numOfRows, pOper->pMsg); - sdbDecRef(pTable, pOper->pObj); + sdbDebug("vgId:1, sdb:%s, delete key:%s from hash, numOfRows:%" PRId64 ", msg:%p", pTable->name, + sdbGetRowStr(pTable, pRow->pObj), pTable->numOfRows, pRow->pMsg); + + sdbDecRef(pTable, pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t sdbUpdateHash(SSdbTable *pTable, SSdbOper *pOper) { - sdbDebug("table:%s, update record:%s in hash, numOfRows:%" PRId64 ", msg:%p", pTable->tableName, - sdbGetKeyStrFromObj(pTable, pOper->pObj), pTable->numOfRows, pOper->pMsg); +static int32_t sdbUpdateHash(SSdbTable *pTable, SSdbRow *pRow) { + sdbDebug("vgId:1, sdb:%s, update key:%s in hash, numOfRows:%" PRId64 ", msg:%p", pTable->name, + sdbGetRowStr(pTable, pRow->pObj), pTable->numOfRows, pRow->pMsg); - (*pTable->updateFp)(pOper); + (*pTable->fpUpdate)(pRow); return TSDB_CODE_SUCCESS; } -static int sdbWrite(void *param, void *data, int32_t type, void *pMsg) { - SSdbOper *pOper = param; - SWalHead *pHead = data; +static int32_t sdbPerformInsertAction(SWalHead *pHead, SSdbTable *pTable) { + SSdbRow row = {.rowSize = pHead->len, .rowData = pHead->cont, .pTable = pTable}; + (*pTable->fpDecode)(&row); + return sdbInsertHash(pTable, &row); +} + +static int32_t sdbPerformDeleteAction(SWalHead *pHead, SSdbTable *pTable) { + void *pObj = sdbGetRowMeta(pTable, pHead->cont); + if (pObj == NULL) { + sdbDebug("vgId:1, sdb:%s, object:%s not exist in hash, ignore delete action", pTable->name, + sdbGetKeyStr(pTable, pHead->cont)); + return TSDB_CODE_SUCCESS; + } + SSdbRow row = {.pTable = pTable, .pObj = pObj}; + return sdbDeleteHash(pTable, &row); +} + +static int32_t sdbPerformUpdateAction(SWalHead *pHead, SSdbTable *pTable) { + void *pObj = sdbGetRowMeta(pTable, pHead->cont); + if (pObj == NULL) { + sdbDebug("vgId:1, sdb:%s, object:%s not exist in hash, ignore update action", pTable->name, + sdbGetKeyStr(pTable, pHead->cont)); + return TSDB_CODE_SUCCESS; + } + SSdbRow row = {.rowSize = pHead->len, .rowData = pHead->cont, .pTable = pTable}; + (*pTable->fpDecode)(&row); + return sdbUpdateHash(pTable, &row); +} + +static int32_t sdbProcessWrite(void *wparam, void *hparam, int32_t qtype, void *unused) { + SSdbRow *pRow = wparam; + SWalHead *pHead = hparam; int32_t tableId = pHead->msgType / 10; int32_t action = pHead->msgType % 10; SSdbTable *pTable = sdbGetTableFromId(tableId); assert(pTable != NULL); - pthread_mutex_lock(&tsSdbObj.mutex); + if (qtype == TAOS_QTYPE_QUERY) return sdbPerformDeleteAction(pHead, pTable); + + pthread_mutex_lock(&tsSdbMgmt.mutex); if (pHead->version == 0) { // assign version - tsSdbObj.version++; - pHead->version = tsSdbObj.version; + tsSdbMgmt.version++; + pHead->version = tsSdbMgmt.version; } else { // for data from WAL or forward, version may be smaller - if (pHead->version <= tsSdbObj.version) { - pthread_mutex_unlock(&tsSdbObj.mutex); - sdbDebug("table:%s, failed to restore %s record:%s from source(%d), ver:%" PRId64 " too large, sdb ver:%" PRId64, - pTable->tableName, sdbGetActionStr(action), sdbGetKeyStr(pTable, pHead->cont), type, pHead->version, tsSdbObj.version); + if (pHead->version <= tsSdbMgmt.version) { + pthread_mutex_unlock(&tsSdbMgmt.mutex); + sdbDebug("vgId:1, sdb:%s, failed to restore %s key:%s from source(%d), hver:%" PRIu64 " too large, mver:%" PRIu64, + pTable->name, actStr[action], sdbGetKeyStr(pTable, pHead->cont), qtype, pHead->version, tsSdbMgmt.version); return TSDB_CODE_SUCCESS; - } else if (pHead->version != tsSdbObj.version + 1) { - pthread_mutex_unlock(&tsSdbObj.mutex); - sdbError("table:%s, failed to restore %s record:%s from source(%d), ver:%" PRId64 " too large, sdb ver:%" PRId64, - pTable->tableName, sdbGetActionStr(action), sdbGetKeyStr(pTable, pHead->cont), type, pHead->version, tsSdbObj.version); + } else if (pHead->version != tsSdbMgmt.version + 1) { + pthread_mutex_unlock(&tsSdbMgmt.mutex); + sdbError("vgId:1, sdb:%s, failed to restore %s key:%s from source(%d), hver:%" PRIu64 " too large, mver:%" PRIu64, + pTable->name, actStr[action], sdbGetKeyStr(pTable, pHead->cont), qtype, pHead->version, tsSdbMgmt.version); return TSDB_CODE_SYN_INVALID_VERSION; } else { - tsSdbObj.version = pHead->version; + tsSdbMgmt.version = pHead->version; } } - int32_t code = walWrite(tsSdbObj.wal, pHead); + int32_t code = walWrite(tsSdbMgmt.wal, pHead); if (code < 0) { - pthread_mutex_unlock(&tsSdbObj.mutex); + pthread_mutex_unlock(&tsSdbMgmt.mutex); return code; } - pthread_mutex_unlock(&tsSdbObj.mutex); + pthread_mutex_unlock(&tsSdbMgmt.mutex); - // from app, oper is created - if (pOper != NULL) { + // from app, row is created + if (pRow != NULL) { // forward to peers - pOper->processedCount = 0; - int32_t syncCode = syncForwardToPeer(tsSdbObj.sync, pHead, pOper, TAOS_QTYPE_RPC); - if (syncCode <= 0) pOper->processedCount = 1; + pRow->processedCount = 0; + int32_t syncCode = syncForwardToPeer(tsSdbMgmt.sync, pHead, pRow, TAOS_QTYPE_RPC); + if (syncCode <= 0) pRow->processedCount = 1; if (syncCode < 0) { - sdbError("table:%s, failed to forward request, result:%s action:%s record:%s ver:%" PRId64 ", msg:%p", pTable->tableName, - tstrerror(syncCode), sdbGetActionStr(action), sdbGetKeyStr(pTable, pHead->cont), pHead->version, pOper->pMsg); + sdbError("vgId:1, sdb:%s, failed to forward req since %s action:%s key:%s hver:%" PRIu64 ", msg:%p", pTable->name, + tstrerror(syncCode), actStr[action], sdbGetKeyStr(pTable, pHead->cont), pHead->version, pRow->pMsg); } else if (syncCode > 0) { - sdbDebug("table:%s, forward request is sent, action:%s record:%s ver:%" PRId64 ", msg:%p", pTable->tableName, - sdbGetActionStr(action), sdbGetKeyStr(pTable, pHead->cont), pHead->version, pOper->pMsg); + sdbDebug("vgId:1, sdb:%s, forward req is sent, action:%s key:%s hver:%" PRIu64 ", msg:%p", pTable->name, + actStr[action], sdbGetKeyStr(pTable, pHead->cont), pHead->version, pRow->pMsg); } else { - sdbTrace("table:%s, no need to send fwd request, action:%s record:%s ver:%" PRId64 ", msg:%p", pTable->tableName, - sdbGetActionStr(action), sdbGetKeyStr(pTable, pHead->cont), pHead->version, pOper->pMsg); + sdbTrace("vgId:1, sdb:%s, no need to send fwd req, action:%s key:%s hver:%" PRIu64 ", msg:%p", pTable->name, + actStr[action], sdbGetKeyStr(pTable, pHead->cont), pHead->version, pRow->pMsg); } return syncCode; } - sdbDebug("table:%s, record from wal/fwd is disposed, action:%s record:%s ver:%" PRId64, pTable->tableName, - sdbGetActionStr(action), sdbGetKeyStr(pTable, pHead->cont), pHead->version); + sdbDebug("vgId:1, sdb:%s, record from wal/fwd is disposed, action:%s key:%s hver:%" PRIu64, pTable->name, + actStr[action], sdbGetKeyStr(pTable, pHead->cont), pHead->version); // even it is WAL/FWD, it shall be called to update version in sync - syncForwardToPeer(tsSdbObj.sync, pHead, pOper, TAOS_QTYPE_RPC); + syncForwardToPeer(tsSdbMgmt.sync, pHead, pRow, TAOS_QTYPE_RPC); - // from wal or forward msg, oper not created, should add into hash + // from wal or forward msg, row not created, should add into hash if (action == SDB_ACTION_INSERT) { - SSdbOper oper = {.rowSize = pHead->len, .rowData = pHead->cont, .table = pTable}; - code = (*pTable->decodeFp)(&oper); - return sdbInsertHash(pTable, &oper); + return sdbPerformInsertAction(pHead, pTable); } else if (action == SDB_ACTION_DELETE) { - void *pRow = sdbGetRowMeta(pTable, pHead->cont); - if (pRow == NULL) { - sdbDebug("table:%s, object:%s not exist in hash, ignore delete action", pTable->tableName, - sdbGetKeyStr(pTable, pHead->cont)); + if (qtype == TAOS_QTYPE_FWD) { + // Drop database/stable may take a long time and cause a timeout, so we confirm first then reput it into queue + sdbWriteFwdToQueue(1, hparam, TAOS_QTYPE_QUERY, unused); return TSDB_CODE_SUCCESS; + } else { + return sdbPerformDeleteAction(pHead, pTable); } - SSdbOper oper = {.table = pTable, .pObj = pRow}; - return sdbDeleteHash(pTable, &oper); } else if (action == SDB_ACTION_UPDATE) { - void *pRow = sdbGetRowMeta(pTable, pHead->cont); - if (pRow == NULL) { - sdbDebug("table:%s, object:%s not exist in hash, ignore update action", pTable->tableName, - sdbGetKeyStr(pTable, pHead->cont)); - return TSDB_CODE_SUCCESS; - } - SSdbOper oper = {.rowSize = pHead->len, .rowData = pHead->cont, .table = pTable}; - code = (*pTable->decodeFp)(&oper); - return sdbUpdateHash(pTable, &oper); + return sdbPerformUpdateAction(pHead, pTable); } else { return TSDB_CODE_MND_INVALID_MSG_TYPE; } } -int32_t sdbInsertRow(SSdbOper *pOper) { - SSdbTable *pTable = (SSdbTable *)pOper->table; +int32_t sdbInsertRow(SSdbRow *pRow) { + SSdbTable *pTable = pRow->pTable; if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; - if (sdbGetRowFromObj(pTable, pOper->pObj)) { - sdbError("table:%s, failed to insert record:%s, already exist", pTable->tableName, - sdbGetKeyStrFromObj(pTable, pOper->pObj)); - sdbDecRef(pTable, pOper->pObj); + if (sdbGetRowFromObj(pTable, pRow->pObj)) { + sdbError("vgId:1, sdb:%s, failed to insert:%s since it exist", pTable->name, sdbGetRowStr(pTable, pRow->pObj)); + sdbDecRef(pTable, pRow->pObj); return TSDB_CODE_MND_SDB_OBJ_ALREADY_THERE; } if (pTable->keyType == SDB_KEY_AUTO) { - *((uint32_t *)pOper->pObj) = atomic_add_fetch_32(&pTable->autoIndex, 1); + *((uint32_t *)pRow->pObj) = atomic_add_fetch_32(&pTable->autoIndex, 1); // let vgId increase from 2 - if (pTable->autoIndex == 1 && strcmp(pTable->tableName, "vgroups") == 0) { - *((uint32_t *)pOper->pObj) = atomic_add_fetch_32(&pTable->autoIndex, 1); + if (pTable->autoIndex == 1 && pTable->id == SDB_TABLE_VGROUP) { + *((uint32_t *)pRow->pObj) = atomic_add_fetch_32(&pTable->autoIndex, 1); } } - int32_t code = sdbInsertHash(pTable, pOper); + int32_t code = sdbInsertHash(pTable, pRow); if (code != TSDB_CODE_SUCCESS) { - sdbError("table:%s, failed to insert into hash", pTable->tableName); + sdbError("vgId:1, sdb:%s, failed to insert:%s into hash", pTable->name, sdbGetRowStr(pTable, pRow->pObj)); return code; } // just insert data into memory - if (pOper->type != SDB_OPER_GLOBAL) { + if (pRow->type != SDB_OPER_GLOBAL) { return TSDB_CODE_SUCCESS; } - if (pOper->reqFp) { - return (*pOper->reqFp)(pOper->pMsg); + if (pRow->fpReq) { + return (*pRow->fpReq)(pRow->pMsg); } else { - return sdbInsertRowImp(pOper); + return sdbWriteRowToQueue(pRow, SDB_ACTION_INSERT); } } -int32_t sdbInsertRowImp(SSdbOper *pOper) { - SSdbTable *pTable = (SSdbTable *)pOper->table; - if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; - - int32_t size = sizeof(SSdbOper) + sizeof(SWalHead) + pTable->maxRowSize + SDB_SYNC_HACK; - SSdbOper *pNewOper = taosAllocateQitem(size); - - SWalHead *pHead = (void *)pNewOper + sizeof(SSdbOper) + SDB_SYNC_HACK; - pHead->version = 0; - pHead->len = pOper->rowSize; - pHead->msgType = pTable->tableId * 10 + SDB_ACTION_INSERT; - - pOper->rowData = pHead->cont; - (*pTable->encodeFp)(pOper); - pHead->len = pOper->rowSize; - - memcpy(pNewOper, pOper, sizeof(SSdbOper)); - - if (pNewOper->pMsg != NULL) { - sdbDebug("app:%p:%p, table:%s record:%p:%s, insert action is add to sdb queue", pNewOper->pMsg->rpcMsg.ahandle, - pNewOper->pMsg, pTable->tableName, pOper->pObj, sdbGetKeyStrFromObj(pTable, pOper->pObj)); - } - - sdbIncRef(pNewOper->table, pNewOper->pObj); - taosWriteQitem(tsSdbWriteQueue, TAOS_QTYPE_RPC, pNewOper); - - return TSDB_CODE_MND_ACTION_IN_PROGRESS; -} - -bool sdbCheckRowDeleted(void *pTableInput, void *pRow) { - SSdbTable *pTable = pTableInput; +bool sdbCheckRowDeleted(void *tparam, void *pRow) { + SSdbTable *pTable = tparam; if (pTable == NULL) return false; int32_t *updateEnd = pRow + pTable->refCountPos - 4; return atomic_val_compare_exchange_32(updateEnd, 1, 1) == 1; } -int32_t sdbDeleteRow(SSdbOper *pOper) { - SSdbTable *pTable = (SSdbTable *)pOper->table; +int32_t sdbDeleteRow(SSdbRow *pRow) { + SSdbTable *pTable = pRow->pTable; if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; - void *pRow = sdbGetRowMetaFromObj(pTable, pOper->pObj); - if (pRow == NULL) { - sdbDebug("table:%s, record is not there, delete failed", pTable->tableName); + void *pObj = sdbGetRowMetaFromObj(pTable, pRow->pObj); + if (pObj == NULL) { + sdbDebug("vgId:1, sdb:%s, record is not there, delete failed", pTable->name); return TSDB_CODE_MND_SDB_OBJ_NOT_THERE; } - sdbIncRef(pTable, pOper->pObj); - - int32_t code = sdbDeleteHash(pTable, pOper); + int32_t code = sdbDeleteHash(pTable, pRow); if (code != TSDB_CODE_SUCCESS) { - sdbError("table:%s, failed to delete from hash", pTable->tableName); - sdbDecRef(pTable, pOper->pObj); + sdbError("vgId:1, sdb:%s, failed to delete from hash", pTable->name); return code; } // just delete data from memory - if (pOper->type != SDB_OPER_GLOBAL) { - sdbDecRef(pTable, pOper->pObj); + if (pRow->type != SDB_OPER_GLOBAL) { return TSDB_CODE_SUCCESS; } - if (pOper->reqFp) { - return (*pOper->reqFp)(pOper->pMsg); + if (pRow->fpReq) { + return (*pRow->fpReq)(pRow->pMsg); } else { - return sdbDeleteRowImp(pOper); + return sdbWriteRowToQueue(pRow, SDB_ACTION_DELETE); } } -int32_t sdbDeleteRowImp(SSdbOper *pOper) { - SSdbTable *pTable = (SSdbTable *)pOper->table; +int32_t sdbUpdateRow(SSdbRow *pRow) { + SSdbTable *pTable = pRow->pTable; if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; - int32_t size = sizeof(SSdbOper) + sizeof(SWalHead) + pTable->maxRowSize + SDB_SYNC_HACK; - SSdbOper *pNewOper = taosAllocateQitem(size); - - SWalHead *pHead = (void *)pNewOper + sizeof(SSdbOper) + SDB_SYNC_HACK; - pHead->version = 0; - pHead->msgType = pTable->tableId * 10 + SDB_ACTION_DELETE; - - pOper->rowData = pHead->cont; - (*pTable->encodeFp)(pOper); - pHead->len = pOper->rowSize; - - memcpy(pNewOper, pOper, sizeof(SSdbOper)); - - if (pNewOper->pMsg != NULL) { - sdbDebug("app:%p:%p, table:%s record:%p:%s, delete action is add to sdb queue", pNewOper->pMsg->rpcMsg.ahandle, - pNewOper->pMsg, pTable->tableName, pOper->pObj, sdbGetKeyStrFromObj(pTable, pOper->pObj)); - } - - taosWriteQitem(tsSdbWriteQueue, TAOS_QTYPE_RPC, pNewOper); - - return TSDB_CODE_MND_ACTION_IN_PROGRESS; -} - -int32_t sdbUpdateRow(SSdbOper *pOper) { - SSdbTable *pTable = (SSdbTable *)pOper->table; - if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; - - void *pRow = sdbGetRowMetaFromObj(pTable, pOper->pObj); - if (pRow == NULL) { - sdbDebug("table:%s, record is not there, update failed", pTable->tableName); + void *pObj = sdbGetRowMetaFromObj(pTable, pRow->pObj); + if (pObj == NULL) { + sdbDebug("vgId:1, sdb:%s, record is not there, update failed", pTable->name); return TSDB_CODE_MND_SDB_OBJ_NOT_THERE; } - int32_t code = sdbUpdateHash(pTable, pOper); + int32_t code = sdbUpdateHash(pTable, pRow); if (code != TSDB_CODE_SUCCESS) { - sdbError("table:%s, failed to update hash", pTable->tableName); + sdbError("vgId:1, sdb:%s, failed to update hash", pTable->name); return code; } // just update data in memory - if (pOper->type != SDB_OPER_GLOBAL) { + if (pRow->type != SDB_OPER_GLOBAL) { return TSDB_CODE_SUCCESS; } - if (pOper->reqFp) { - return (*pOper->reqFp)(pOper->pMsg); + if (pRow->fpReq) { + return (*pRow->fpReq)(pRow->pMsg); } else { - return sdbUpdateRowImp(pOper); + return sdbWriteRowToQueue(pRow, SDB_ACTION_UPDATE); } } -int32_t sdbUpdateRowImp(SSdbOper *pOper) { - SSdbTable *pTable = (SSdbTable *)pOper->table; - if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; - - int32_t size = sizeof(SSdbOper) + sizeof(SWalHead) + pTable->maxRowSize + SDB_SYNC_HACK; - SSdbOper *pNewOper = taosAllocateQitem(size); - - SWalHead *pHead = (void *)pNewOper + sizeof(SSdbOper) + SDB_SYNC_HACK; - pHead->version = 0; - pHead->msgType = pTable->tableId * 10 + SDB_ACTION_UPDATE; - - pOper->rowData = pHead->cont; - (*pTable->encodeFp)(pOper); - pHead->len = pOper->rowSize; - - memcpy(pNewOper, pOper, sizeof(SSdbOper)); - - if (pNewOper->pMsg != NULL) { - sdbDebug("app:%p:%p, table:%s record:%p:%s, update action is add to sdb queue", pNewOper->pMsg->rpcMsg.ahandle, - pNewOper->pMsg, pTable->tableName, pOper->pObj, sdbGetKeyStrFromObj(pTable, pOper->pObj)); - } - - sdbIncRef(pNewOper->table, pNewOper->pObj); - taosWriteQitem(tsSdbWriteQueue, TAOS_QTYPE_RPC, pNewOper); - - return TSDB_CODE_MND_ACTION_IN_PROGRESS; -} - -void *sdbFetchRow(void *handle, void *pNode, void **ppRow) { - SSdbTable *pTable = (SSdbTable *)handle; +void *sdbFetchRow(void *tparam, void *pIter, void **ppRow) { + SSdbTable *pTable = tparam; *ppRow = NULL; if (pTable == NULL) return NULL; - SHashMutableIterator *pIter = pNode; - if (pIter == NULL) { - pIter = taosHashCreateIter(pTable->iHandle); - } + pIter = taosHashIterate(pTable->iHandle, pIter); + if (pIter == NULL) return NULL; - if (!taosHashIterNext(pIter)) { - taosHashDestroyIter(pIter); - return NULL; - } - - void **ppMetaRow = taosHashIterGet(pIter); + void **ppMetaRow = pIter; if (ppMetaRow == NULL) { - taosHashDestroyIter(pIter); + taosHashCancelIterate(pTable->iHandle, pIter); return NULL; } *ppRow = *ppMetaRow; - sdbIncRef(handle, *ppMetaRow); + sdbIncRef(pTable, *ppMetaRow); return pIter; } -void sdbFreeIter(void *pIter) { - if (pIter != NULL) { - taosHashDestroyIter(pIter); - } +void sdbFreeIter(void *tparam, void *pIter) { + SSdbTable *pTable = tparam; + if (pTable == NULL || pIter == NULL) return; + + taosHashCancelIterate(pTable->iHandle, pIter); } void *sdbOpenTable(SSdbTableDesc *pDesc) { @@ -899,28 +807,28 @@ void *sdbOpenTable(SSdbTableDesc *pDesc) { if (pTable == NULL) return NULL; pthread_mutex_init(&pTable->mutex, NULL); - tstrncpy(pTable->tableName, pDesc->tableName, SDB_TABLE_LEN); + tstrncpy(pTable->name, pDesc->name, SDB_TABLE_LEN); pTable->keyType = pDesc->keyType; - pTable->tableId = pDesc->tableId; + pTable->id = pDesc->id; pTable->hashSessions = pDesc->hashSessions; pTable->maxRowSize = pDesc->maxRowSize; pTable->refCountPos = pDesc->refCountPos; - pTable->insertFp = pDesc->insertFp; - pTable->deleteFp = pDesc->deleteFp; - pTable->updateFp = pDesc->updateFp; - pTable->encodeFp = pDesc->encodeFp; - pTable->decodeFp = pDesc->decodeFp; - pTable->destroyFp = pDesc->destroyFp; - pTable->restoredFp = pDesc->restoredFp; + pTable->fpInsert = pDesc->fpInsert; + pTable->fpDelete = pDesc->fpDelete; + pTable->fpUpdate = pDesc->fpUpdate; + pTable->fpEncode = pDesc->fpEncode; + pTable->fpDecode = pDesc->fpDecode; + pTable->fpDestroy = pDesc->fpDestroy; + pTable->fpRestored = pDesc->fpRestored; _hash_fn_t hashFp = taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT); if (pTable->keyType == SDB_KEY_STRING || pTable->keyType == SDB_KEY_VAR_STRING) { hashFp = taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY); } - pTable->iHandle = taosHashInit(pTable->hashSessions, hashFp, true, true); + pTable->iHandle = taosHashInit(pTable->hashSessions, hashFp, true, HASH_ENTRY_LOCK); - tsSdbObj.numOfTables++; - tsSdbObj.tableList[pTable->tableId] = pTable; + tsSdbMgmt.numOfTables++; + tsSdbMgmt.tableList[pTable->id] = pTable; return pTable; } @@ -928,87 +836,88 @@ void sdbCloseTable(void *handle) { SSdbTable *pTable = (SSdbTable *)handle; if (pTable == NULL) return; - tsSdbObj.numOfTables--; - tsSdbObj.tableList[pTable->tableId] = NULL; + tsSdbMgmt.numOfTables--; + tsSdbMgmt.tableList[pTable->id] = NULL; - SHashMutableIterator *pIter = taosHashCreateIter(pTable->iHandle); - while (taosHashIterNext(pIter)) { - void **ppRow = taosHashIterGet(pIter); + void *pIter = taosHashIterate(pTable->iHandle, NULL); + while (pIter) { + void **ppRow = pIter; + pIter = taosHashIterate(pTable->iHandle, pIter); if (ppRow == NULL) continue; - SSdbOper oper = { + SSdbRow row = { .pObj = *ppRow, - .table = pTable, + .pTable = pTable, }; - (*pTable->destroyFp)(&oper); + (*pTable->fpDestroy)(&row); } - taosHashDestroyIter(pIter); + taosHashCancelIterate(pTable->iHandle, pIter); taosHashCleanup(pTable->iHandle); pthread_mutex_destroy(&pTable->mutex); - sdbDebug("table:%s, is closed, numOfTables:%d", pTable->tableName, tsSdbObj.numOfTables); + sdbDebug("vgId:1, sdb:%s, is closed, numOfTables:%d", pTable->name, tsSdbMgmt.numOfTables); free(pTable); } -int32_t sdbInitWriteWorker() { +static int32_t sdbInitWorker() { tsSdbPool.num = 1; - tsSdbPool.writeWorker = (SSdbWriteWorker *)calloc(sizeof(SSdbWriteWorker), tsSdbPool.num); + tsSdbPool.worker = calloc(sizeof(SSdbWorker), tsSdbPool.num); - if (tsSdbPool.writeWorker == NULL) return -1; + if (tsSdbPool.worker == NULL) return -1; for (int32_t i = 0; i < tsSdbPool.num; ++i) { - SSdbWriteWorker *pWorker = tsSdbPool.writeWorker + i; + SSdbWorker *pWorker = tsSdbPool.worker + i; pWorker->workerId = i; } - sdbAllocWriteQueue(); + sdbAllocQueue(); - mInfo("sdb write is opened"); + mInfo("vgId:1, sdb write is opened"); return 0; } -void sdbCleanupWriteWorker() { +static void sdbCleanupWorker() { for (int32_t i = 0; i < tsSdbPool.num; ++i) { - SSdbWriteWorker *pWorker = tsSdbPool.writeWorker + i; + SSdbWorker *pWorker = tsSdbPool.worker + i; if (pWorker->thread) { - taosQsetThreadResume(tsSdbWriteQset); + taosQsetThreadResume(tsSdbWQset); } } for (int32_t i = 0; i < tsSdbPool.num; ++i) { - SSdbWriteWorker *pWorker = tsSdbPool.writeWorker + i; + SSdbWorker *pWorker = tsSdbPool.worker + i; if (pWorker->thread) { pthread_join(pWorker->thread, NULL); } } - sdbFreeWritequeue(); - tfree(tsSdbPool.writeWorker); + sdbFreeQueue(); + tfree(tsSdbPool.worker); - mInfo("sdb write is closed"); + mInfo("vgId:1, sdb write is closed"); } -int32_t sdbAllocWriteQueue() { - tsSdbWriteQueue = taosOpenQueue(); - if (tsSdbWriteQueue == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; +static int32_t sdbAllocQueue() { + tsSdbWQueue = taosOpenQueue(); + if (tsSdbWQueue == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - tsSdbWriteQset = taosOpenQset(); - if (tsSdbWriteQset == NULL) { - taosCloseQueue(tsSdbWriteQueue); + tsSdbWQset = taosOpenQset(); + if (tsSdbWQset == NULL) { + taosCloseQueue(tsSdbWQueue); return TSDB_CODE_MND_OUT_OF_MEMORY; } - taosAddIntoQset(tsSdbWriteQset, tsSdbWriteQueue, NULL); + taosAddIntoQset(tsSdbWQset, tsSdbWQueue, NULL); - tsSdbWriteQall = taosAllocateQall(); - if (tsSdbWriteQall == NULL) { - taosCloseQset(tsSdbWriteQset); - taosCloseQueue(tsSdbWriteQueue); + tsSdbWQall = taosAllocateQall(); + if (tsSdbWQall == NULL) { + taosCloseQset(tsSdbWQset); + taosCloseQueue(tsSdbWQueue); return TSDB_CODE_MND_OUT_OF_MEMORY; } for (int32_t i = 0; i < tsSdbPool.num; ++i) { - SSdbWriteWorker *pWorker = tsSdbPool.writeWorker + i; + SSdbWorker *pWorker = tsSdbPool.worker + i; pWorker->workerId = i; pthread_attr_t thAttr; @@ -1017,9 +926,9 @@ int32_t sdbAllocWriteQueue() { if (pthread_create(&pWorker->thread, &thAttr, sdbWorkerFp, pWorker) != 0) { mError("failed to create thread to process sdb write queue, reason:%s", strerror(errno)); - taosFreeQall(tsSdbWriteQall); - taosCloseQset(tsSdbWriteQset); - taosCloseQueue(tsSdbWriteQueue); + taosFreeQall(tsSdbWQall); + taosCloseQset(tsSdbWQset); + taosCloseQueue(tsSdbWQueue); return TSDB_CODE_MND_OUT_OF_MEMORY; } @@ -1027,85 +936,128 @@ int32_t sdbAllocWriteQueue() { mDebug("sdb write worker:%d is launched, total:%d", pWorker->workerId, tsSdbPool.num); } - mDebug("sdb write queue:%p is allocated", tsSdbWriteQueue); + mDebug("sdb write queue:%p is allocated", tsSdbWQueue); return TSDB_CODE_SUCCESS; } -void sdbFreeWritequeue() { - taosCloseQueue(tsSdbWriteQueue); - taosFreeQall(tsSdbWriteQall); - taosCloseQset(tsSdbWriteQset); - tsSdbWriteQall = NULL; - tsSdbWriteQset = NULL; - tsSdbWriteQueue = NULL; +static void sdbFreeQueue() { + taosCloseQueue(tsSdbWQueue); + taosFreeQall(tsSdbWQall); + taosCloseQset(tsSdbWQset); + tsSdbWQall = NULL; + tsSdbWQset = NULL; + tsSdbWQueue = NULL; } -int32_t sdbWriteToQueue(void *param, void *data, int32_t qtype, void *pMsg) { - SWalHead *pHead = data; - int32_t size = sizeof(SWalHead) + pHead->len; - SWalHead *pWal = taosAllocateQitem(size); - memcpy(pWal, pHead, size); +static int32_t sdbWriteToQueue(SSdbRow *pRow, int32_t qtype) { + SWalHead *pHead = pRow->pHead; - taosWriteQitem(tsSdbWriteQueue, qtype, pWal); - return 0; + if (pHead->len > TSDB_MAX_WAL_SIZE) { + sdbError("vgId:1, wal len:%d exceeds limit, hver:%" PRIu64, pHead->len, pHead->version); + taosFreeQitem(pRow); + return TSDB_CODE_WAL_SIZE_LIMIT; + } + + int32_t queued = atomic_add_fetch_32(&tsSdbMgmt.queuedMsg, 1); + if (queued > MAX_QUEUED_MSG_NUM) { + sdbDebug("vgId:1, too many msg:%d in sdb queue, flow control", queued); + taosMsleep(1); + } + + sdbIncRef(pRow->pTable, pRow->pObj); + + sdbTrace("vgId:1, msg:%p qtype:%s write into to sdb queue, queued:%d", pRow->pMsg, qtypeStr[qtype], queued); + taosWriteQitem(tsSdbWQueue, qtype, pRow); + + return TSDB_CODE_MND_ACTION_IN_PROGRESS; } -static void *sdbWorkerFp(void *param) { - SWalHead *pHead; - SSdbOper *pOper; - int32_t type; - int32_t numOfMsgs; - void * item; - void * unUsed; +static void sdbFreeFromQueue(SSdbRow *pRow) { + int32_t queued = atomic_sub_fetch_32(&tsSdbMgmt.queuedMsg, 1); + sdbTrace("vgId:1, msg:%p free from sdb queue, queued:%d", pRow->pMsg, queued); + + sdbDecRef(pRow->pTable, pRow->pObj); + taosFreeQitem(pRow); +} + +static int32_t sdbWriteFwdToQueue(int32_t vgId, void *wparam, int32_t qtype, void *rparam) { + SWalHead *pHead = wparam; + + int32_t size = sizeof(SSdbRow) + sizeof(SWalHead) + pHead->len; + SSdbRow *pRow = taosAllocateQitem(size); + if (pRow == NULL) { + return TSDB_CODE_VND_OUT_OF_MEMORY; + } + + memcpy(pRow->pHead, pHead, sizeof(SWalHead) + pHead->len); + pRow->rowData = pRow->pHead->cont; + + return sdbWriteToQueue(pRow, qtype); +} + +static int32_t sdbWriteRowToQueue(SSdbRow *pInputRow, int32_t action) { + SSdbTable *pTable = pInputRow->pTable; + if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE; + + int32_t size = sizeof(SSdbRow) + sizeof(SWalHead) + pTable->maxRowSize; + SSdbRow *pRow = taosAllocateQitem(size); + if (pRow == NULL) { + return TSDB_CODE_VND_OUT_OF_MEMORY; + } + + memcpy(pRow, pInputRow, sizeof(SSdbRow)); + pRow->processedCount = 1; + + SWalHead *pHead = pRow->pHead; + pRow->rowData = pHead->cont; + (*pTable->fpEncode)(pRow); + + pHead->len = pRow->rowSize; + pHead->version = 0; + pHead->msgType = pTable->id * 10 + action; + + return sdbWriteToQueue(pRow, TAOS_QTYPE_RPC); +} + +int32_t sdbInsertRowToQueue(SSdbRow *pRow) { return sdbWriteRowToQueue(pRow, SDB_ACTION_INSERT); } + +static void *sdbWorkerFp(void *pWorker) { + SSdbRow *pRow; + int32_t qtype; + void * unUsed; while (1) { - numOfMsgs = taosReadAllQitemsFromQset(tsSdbWriteQset, tsSdbWriteQall, &unUsed); + int32_t numOfMsgs = taosReadAllQitemsFromQset(tsSdbWQset, tsSdbWQall, &unUsed); if (numOfMsgs == 0) { - sdbDebug("qset:%p, sdb got no message from qset, exiting", tsSdbWriteQset); + sdbDebug("qset:%p, sdb got no message from qset, exiting", tsSdbWQset); break; } for (int32_t i = 0; i < numOfMsgs; ++i) { - taosGetQitem(tsSdbWriteQall, &type, &item); - if (type == TAOS_QTYPE_RPC) { - pOper = (SSdbOper *)item; - pOper->processedCount = 1; - pHead = (void *)pOper + sizeof(SSdbOper) + SDB_SYNC_HACK; - if (pOper->pMsg != NULL) { - sdbDebug("app:%p:%p, table:%s record:%p:%s ver:%" PRIu64 ", will be processed in sdb queue", - pOper->pMsg->rpcMsg.ahandle, pOper->pMsg, ((SSdbTable *)pOper->table)->tableName, pOper->pObj, - sdbGetKeyStr(pOper->table, pHead->cont), pHead->version); - } - } else { - pHead = (SWalHead *)item; - pOper = NULL; - } + taosGetQitem(tsSdbWQall, &qtype, (void **)&pRow); + sdbTrace("vgId:1, msg:%p, row:%p hver:%" PRIu64 ", will be processed in sdb queue", pRow->pMsg, pRow->pObj, + pRow->pHead->version); - int32_t code = sdbWrite(pOper, pHead, type, NULL); - if (code > 0) code = 0; - if (pOper) { - pOper->retCode = code; - } else { - pHead->len = code; // hackway - } + pRow->code = sdbProcessWrite((qtype == TAOS_QTYPE_RPC) ? pRow : NULL, pRow->pHead, qtype, NULL); + if (pRow->code > 0) pRow->code = 0; + + sdbTrace("vgId:1, msg:%p is processed in sdb queue, code:%x", pRow->pMsg, pRow->code); } - walFsync(tsSdbObj.wal, true); + walFsync(tsSdbMgmt.wal, true); // browse all items, and process them one by one - taosResetQitems(tsSdbWriteQall); + taosResetQitems(tsSdbWQall); for (int32_t i = 0; i < numOfMsgs; ++i) { - taosGetQitem(tsSdbWriteQall, &type, &item); + taosGetQitem(tsSdbWQall, &qtype, (void **)&pRow); - if (type == TAOS_QTYPE_RPC) { - pOper = (SSdbOper *)item; - sdbConfirmForward(NULL, pOper, pOper->retCode); - } else if (type == TAOS_QTYPE_FWD) { - pHead = (SWalHead *)item; - syncConfirmForward(tsSdbObj.sync, pHead->version, pHead->len); - taosFreeQitem(item); + if (qtype == TAOS_QTYPE_RPC) { + sdbConfirmForward(1, pRow, pRow->code); } else { - taosFreeQitem(item); + if (qtype == TAOS_QTYPE_FWD) { + syncConfirmForward(tsSdbMgmt.sync, pRow->pHead->version, pRow->code); + } + sdbFreeFromQueue(pRow); } } } diff --git a/src/mnode/src/mnodeShow.c b/src/mnode/src/mnodeShow.c index 0d53fa9617..2621441cb5 100644 --- a/src/mnode/src/mnodeShow.c +++ b/src/mnode/src/mnodeShow.c @@ -57,6 +57,7 @@ static void *tsMnodeShowCache = NULL; static int32_t tsShowObjIndex = 0; static SShowMetaFp tsMnodeShowMetaFp[TSDB_MGMT_TABLE_MAX] = {0}; static SShowRetrieveFp tsMnodeShowRetrieveFp[TSDB_MGMT_TABLE_MAX] = {0}; +static SShowFreeIterFp tsMnodeShowFreeIterFp[TSDB_MGMT_TABLE_MAX] = {0}; int32_t mnodeInitShow() { mnodeAddReadMsgHandle(TSDB_MSG_TYPE_CM_SHOW, mnodeProcessShowMsg); @@ -85,6 +86,10 @@ void mnodeAddShowRetrieveHandle(uint8_t msgType, SShowRetrieveFp fp) { tsMnodeShowRetrieveFp[msgType] = fp; } +void mnodeAddShowFreeIterHandle(uint8_t msgType, SShowFreeIterFp fp) { + tsMnodeShowFreeIterFp[msgType] = fp; +} + static char *mnodeGetShowType(int32_t showType) { switch (showType) { case TSDB_MGMT_TABLE_ACCT: return "show accounts"; @@ -412,7 +417,9 @@ static void* mnodePutShowObj(SShowObj *pShow) { static void mnodeFreeShowObj(void *data) { SShowObj *pShow = *(SShowObj **)data; - sdbFreeIter(pShow->pIter); + if (tsMnodeShowFreeIterFp[pShow->type] != NULL && pShow->pIter != NULL) { + (*tsMnodeShowFreeIterFp[pShow->type])(pShow->pIter); + } mDebug("%p, show is destroyed, data:%p index:%d", pShow, data, pShow->index); tfree(pShow); diff --git a/src/mnode/src/mnodeTable.c b/src/mnode/src/mnodeTable.c index 2b5e6455c0..cc7ffb81ad 100644 --- a/src/mnode/src/mnodeTable.c +++ b/src/mnode/src/mnodeTable.c @@ -99,13 +99,13 @@ static void mnodeDestroyChildTable(SCTableObj *pTable) { tfree(pTable); } -static int32_t mnodeChildTableActionDestroy(SSdbOper *pOper) { - mnodeDestroyChildTable(pOper->pObj); +static int32_t mnodeChildTableActionDestroy(SSdbRow *pRow) { + mnodeDestroyChildTable(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeChildTableActionInsert(SSdbOper *pOper) { - SCTableObj *pTable = pOper->pObj; +static int32_t mnodeChildTableActionInsert(SSdbRow *pRow) { + SCTableObj *pTable = pRow->pObj; SVgObj *pVgroup = mnodeGetVgroup(pTable->vgId); if (pVgroup == NULL) { @@ -153,8 +153,8 @@ static int32_t mnodeChildTableActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeChildTableActionDelete(SSdbOper *pOper) { - SCTableObj *pTable = pOper->pObj; +static int32_t mnodeChildTableActionDelete(SSdbRow *pRow) { + SCTableObj *pTable = pRow->pObj; if (pTable->vgId == 0) { return TSDB_CODE_MND_VGROUP_NOT_EXIST; } @@ -189,8 +189,8 @@ static int32_t mnodeChildTableActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeChildTableActionUpdate(SSdbOper *pOper) { - SCTableObj *pNew = pOper->pObj; +static int32_t mnodeChildTableActionUpdate(SSdbRow *pRow) { + SCTableObj *pNew = pRow->pObj; SCTableObj *pTable = mnodeGetChildTable(pNew->info.tableId); if (pTable != pNew) { void *oldTableId = pTable->info.tableId; @@ -216,50 +216,50 @@ static int32_t mnodeChildTableActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeChildTableActionEncode(SSdbOper *pOper) { - SCTableObj *pTable = pOper->pObj; - assert(pTable != NULL && pOper->rowData != NULL); +static int32_t mnodeChildTableActionEncode(SSdbRow *pRow) { + SCTableObj *pTable = pRow->pObj; + assert(pTable != NULL && pRow->rowData != NULL); int32_t len = strlen(pTable->info.tableId); if (len >= TSDB_TABLE_FNAME_LEN) return TSDB_CODE_MND_INVALID_TABLE_ID; - memcpy(pOper->rowData, pTable->info.tableId, len); - memset(pOper->rowData + len, 0, 1); + memcpy(pRow->rowData, pTable->info.tableId, len); + memset(pRow->rowData + len, 0, 1); len++; - memcpy(pOper->rowData + len, (char*)pTable + sizeof(char *), tsChildTableUpdateSize); + memcpy(pRow->rowData + len, (char*)pTable + sizeof(char *), tsChildTableUpdateSize); len += tsChildTableUpdateSize; if (pTable->info.type != TSDB_CHILD_TABLE) { int32_t schemaSize = pTable->numOfColumns * sizeof(SSchema); - memcpy(pOper->rowData + len, pTable->schema, schemaSize); + memcpy(pRow->rowData + len, pTable->schema, schemaSize); len += schemaSize; if (pTable->sqlLen != 0) { - memcpy(pOper->rowData + len, pTable->sql, pTable->sqlLen); + memcpy(pRow->rowData + len, pTable->sql, pTable->sqlLen); len += pTable->sqlLen; } } - pOper->rowSize = len; + pRow->rowSize = len; return TSDB_CODE_SUCCESS; } -static int32_t mnodeChildTableActionDecode(SSdbOper *pOper) { - assert(pOper->rowData != NULL); +static int32_t mnodeChildTableActionDecode(SSdbRow *pRow) { + assert(pRow->rowData != NULL); SCTableObj *pTable = calloc(1, sizeof(SCTableObj)); if (pTable == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - int32_t len = strlen(pOper->rowData); + int32_t len = strlen(pRow->rowData); if (len >= TSDB_TABLE_FNAME_LEN) { free(pTable); return TSDB_CODE_MND_INVALID_TABLE_ID; } - pTable->info.tableId = strdup(pOper->rowData); + pTable->info.tableId = strdup(pRow->rowData); len++; - memcpy((char*)pTable + sizeof(char *), pOper->rowData + len, tsChildTableUpdateSize); + memcpy((char*)pTable + sizeof(char *), pRow->rowData + len, tsChildTableUpdateSize); len += tsChildTableUpdateSize; if (pTable->info.type != TSDB_CHILD_TABLE) { @@ -269,7 +269,7 @@ static int32_t mnodeChildTableActionDecode(SSdbOper *pOper) { mnodeDestroyChildTable(pTable); return TSDB_CODE_MND_INVALID_TABLE_TYPE; } - memcpy(pTable->schema, pOper->rowData + len, schemaSize); + memcpy(pTable->schema, pRow->rowData + len, schemaSize); len += schemaSize; if (pTable->sqlLen != 0) { @@ -278,11 +278,11 @@ static int32_t mnodeChildTableActionDecode(SSdbOper *pOper) { mnodeDestroyChildTable(pTable); return TSDB_CODE_MND_OUT_OF_MEMORY; } - memcpy(pTable->sql, pOper->rowData + len, pTable->sqlLen); + memcpy(pTable->sql, pRow->rowData + len, pTable->sqlLen); } } - pOper->pObj = pTable; + pRow->pObj = pTable; return TSDB_CODE_SUCCESS; } @@ -297,7 +297,7 @@ static int32_t mnodeChildTableActionRestored() { SDbObj *pDb = mnodeGetDbByTableId(pTable->info.tableId); if (pDb == NULL || pDb->status != TSDB_DB_STATUS_READY) { mError("ctable:%s, failed to get db or db in dropping, discard it", pTable->info.tableId); - SSdbOper desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .table = tsChildTableSdb}; + SSdbRow desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .pTable = tsChildTableSdb}; sdbDeleteRow(&desc); mnodeDecTableRef(pTable); mnodeDecDbRef(pDb); @@ -309,7 +309,7 @@ static int32_t mnodeChildTableActionRestored() { if (pVgroup == NULL) { mError("ctable:%s, failed to get vgId:%d tid:%d, discard it", pTable->info.tableId, pTable->vgId, pTable->tid); pTable->vgId = 0; - SSdbOper desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .table = tsChildTableSdb}; + SSdbRow desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .pTable = tsChildTableSdb}; sdbDeleteRow(&desc); mnodeDecTableRef(pTable); continue; @@ -320,7 +320,7 @@ static int32_t mnodeChildTableActionRestored() { mError("ctable:%s, db:%s not match with vgId:%d db:%s sid:%d, discard it", pTable->info.tableId, pDb->name, pTable->vgId, pVgroup->dbName, pTable->tid); pTable->vgId = 0; - SSdbOper desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .table = tsChildTableSdb}; + SSdbRow desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .pTable = tsChildTableSdb}; sdbDeleteRow(&desc); mnodeDecTableRef(pTable); continue; @@ -331,7 +331,7 @@ static int32_t mnodeChildTableActionRestored() { if (pSuperTable == NULL) { mError("ctable:%s, stable:%" PRIu64 " not exist", pTable->info.tableId, pTable->suid); pTable->vgId = 0; - SSdbOper desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .table = tsChildTableSdb}; + SSdbRow desc = {.type = SDB_OPER_LOCAL, .pObj = pTable, .pTable = tsChildTableSdb}; sdbDeleteRow(&desc); mnodeDecTableRef(pTable); continue; @@ -342,8 +342,7 @@ static int32_t mnodeChildTableActionRestored() { mnodeDecTableRef(pTable); } - sdbFreeIter(pIter); - + mnodeCancelGetNextChildTable(pIter); return 0; } @@ -352,19 +351,19 @@ static int32_t mnodeInitChildTables() { tsChildTableUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj.info.type; SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_CTABLE, - .tableName = "ctables", + .id = SDB_TABLE_CTABLE, + .name = "ctables", .hashSessions = TSDB_DEFAULT_CTABLES_HASH_SIZE, .maxRowSize = sizeof(SCTableObj) + sizeof(SSchema) * (TSDB_MAX_TAGS + TSDB_MAX_COLUMNS + 16) + TSDB_TABLE_FNAME_LEN + TSDB_CQ_SQL_SIZE, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_VAR_STRING, - .insertFp = mnodeChildTableActionInsert, - .deleteFp = mnodeChildTableActionDelete, - .updateFp = mnodeChildTableActionUpdate, - .encodeFp = mnodeChildTableActionEncode, - .decodeFp = mnodeChildTableActionDecode, - .destroyFp = mnodeChildTableActionDestroy, - .restoredFp = mnodeChildTableActionRestored + .fpInsert = mnodeChildTableActionInsert, + .fpDelete = mnodeChildTableActionDelete, + .fpUpdate = mnodeChildTableActionUpdate, + .fpEncode = mnodeChildTableActionEncode, + .fpDecode = mnodeChildTableActionDecode, + .fpDestroy = mnodeChildTableActionDestroy, + .fpRestored = mnodeChildTableActionRestored }; tsChildTableSdb = sdbOpenTable(&tableDesc); @@ -394,7 +393,7 @@ static void mnodeAddTableIntoStable(SSTableObj *pStable, SCTableObj *pCtable) { atomic_add_fetch_32(&pStable->numOfTables, 1); if (pStable->vgHash == NULL) { - pStable->vgHash = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, false); + pStable->vgHash = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_NO_LOCK); } if (pStable->vgHash != NULL) { @@ -413,7 +412,7 @@ static void mnodeRemoveTableFromStable(SSTableObj *pStable, SCTableObj *pCtable) SVgObj *pVgroup = mnodeGetVgroup(pCtable->vgId); if (pVgroup == NULL) { - taosHashRemove(pStable->vgHash, (char *)&pCtable->vgId, sizeof(pCtable->vgId)); + taosHashRemove(pStable->vgHash, &pCtable->vgId, sizeof(pCtable->vgId)); mDebug("table:%s, vgId:%d is remove from stable vgList, sizeOfVgList:%d", pStable->info.tableId, pCtable->vgId, (int32_t)taosHashGetSize(pStable->vgHash)); } @@ -430,13 +429,13 @@ static void mnodeDestroySuperTable(SSTableObj *pStable) { tfree(pStable); } -static int32_t mnodeSuperTableActionDestroy(SSdbOper *pOper) { - mnodeDestroySuperTable(pOper->pObj); +static int32_t mnodeSuperTableActionDestroy(SSdbRow *pRow) { + mnodeDestroySuperTable(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeSuperTableActionInsert(SSdbOper *pOper) { - SSTableObj *pStable = pOper->pObj; +static int32_t mnodeSuperTableActionInsert(SSdbRow *pRow) { + SSTableObj *pStable = pRow->pObj; SDbObj *pDb = mnodeGetDbByTableId(pStable->info.tableId); if (pDb != NULL && pDb->status == TSDB_DB_STATUS_READY) { mnodeAddSuperTableIntoDb(pDb); @@ -446,8 +445,8 @@ static int32_t mnodeSuperTableActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeSuperTableActionDelete(SSdbOper *pOper) { - SSTableObj *pStable = pOper->pObj; +static int32_t mnodeSuperTableActionDelete(SSdbRow *pRow) { + SSTableObj *pStable = pRow->pObj; SDbObj *pDb = mnodeGetDbByTableId(pStable->info.tableId); if (pDb != NULL) { mnodeRemoveSuperTableFromDb(pDb); @@ -458,8 +457,8 @@ static int32_t mnodeSuperTableActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeSuperTableActionUpdate(SSdbOper *pOper) { - SSTableObj *pNew = pOper->pObj; +static int32_t mnodeSuperTableActionUpdate(SSdbRow *pRow) { + SSTableObj *pNew = pRow->pObj; SSTableObj *pTable = mnodeGetSuperTable(pNew->info.tableId); if (pTable != NULL && pTable != pNew) { void *oldTableId = pTable->info.tableId; @@ -483,43 +482,43 @@ static int32_t mnodeSuperTableActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeSuperTableActionEncode(SSdbOper *pOper) { - SSTableObj *pStable = pOper->pObj; - assert(pOper->pObj != NULL && pOper->rowData != NULL); +static int32_t mnodeSuperTableActionEncode(SSdbRow *pRow) { + SSTableObj *pStable = pRow->pObj; + assert(pRow->pObj != NULL && pRow->rowData != NULL); int32_t len = strlen(pStable->info.tableId); if (len >= TSDB_TABLE_FNAME_LEN) len = TSDB_CODE_MND_INVALID_TABLE_ID; - memcpy(pOper->rowData, pStable->info.tableId, len); - memset(pOper->rowData + len, 0, 1); + memcpy(pRow->rowData, pStable->info.tableId, len); + memset(pRow->rowData + len, 0, 1); len++; - memcpy(pOper->rowData + len, (char*)pStable + sizeof(char *), tsSuperTableUpdateSize); + memcpy(pRow->rowData + len, (char*)pStable + sizeof(char *), tsSuperTableUpdateSize); len += tsSuperTableUpdateSize; int32_t schemaSize = sizeof(SSchema) * (pStable->numOfColumns + pStable->numOfTags); - memcpy(pOper->rowData + len, pStable->schema, schemaSize); + memcpy(pRow->rowData + len, pStable->schema, schemaSize); len += schemaSize; - pOper->rowSize = len; + pRow->rowSize = len; return TSDB_CODE_SUCCESS; } -static int32_t mnodeSuperTableActionDecode(SSdbOper *pOper) { - assert(pOper->rowData != NULL); +static int32_t mnodeSuperTableActionDecode(SSdbRow *pRow) { + assert(pRow->rowData != NULL); SSTableObj *pStable = (SSTableObj *) calloc(1, sizeof(SSTableObj)); if (pStable == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - int32_t len = strlen(pOper->rowData); + int32_t len = strlen(pRow->rowData); if (len >= TSDB_TABLE_FNAME_LEN){ free(pStable); return TSDB_CODE_MND_INVALID_TABLE_ID; } - pStable->info.tableId = strdup(pOper->rowData); + pStable->info.tableId = strdup(pRow->rowData); len++; - memcpy((char*)pStable + sizeof(char *), pOper->rowData + len, tsSuperTableUpdateSize); + memcpy((char*)pStable + sizeof(char *), pRow->rowData + len, tsSuperTableUpdateSize); len += tsSuperTableUpdateSize; int32_t schemaSize = sizeof(SSchema) * (pStable->numOfColumns + pStable->numOfTags); @@ -529,9 +528,9 @@ static int32_t mnodeSuperTableActionDecode(SSdbOper *pOper) { return TSDB_CODE_MND_NOT_SUPER_TABLE; } - memcpy(pStable->schema, pOper->rowData + len, schemaSize); + memcpy(pStable->schema, pRow->rowData + len, schemaSize); - pOper->pObj = pStable; + pRow->pObj = pStable; return TSDB_CODE_SUCCESS; } @@ -545,19 +544,19 @@ static int32_t mnodeInitSuperTables() { tsSuperTableUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj.info.type; SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_STABLE, - .tableName = "stables", + .id = SDB_TABLE_STABLE, + .name = "stables", .hashSessions = TSDB_DEFAULT_STABLES_HASH_SIZE, .maxRowSize = sizeof(SSTableObj) + sizeof(SSchema) * (TSDB_MAX_TAGS + TSDB_MAX_COLUMNS + 16) + TSDB_TABLE_FNAME_LEN, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_VAR_STRING, - .insertFp = mnodeSuperTableActionInsert, - .deleteFp = mnodeSuperTableActionDelete, - .updateFp = mnodeSuperTableActionUpdate, - .encodeFp = mnodeSuperTableActionEncode, - .decodeFp = mnodeSuperTableActionDecode, - .destroyFp = mnodeSuperTableActionDestroy, - .restoredFp = mnodeSuperTableActionRestored + .fpInsert = mnodeSuperTableActionInsert, + .fpDelete = mnodeSuperTableActionDelete, + .fpUpdate = mnodeSuperTableActionUpdate, + .fpEncode = mnodeSuperTableActionEncode, + .fpDecode = mnodeSuperTableActionDecode, + .fpDestroy = mnodeSuperTableActionDestroy, + .fpRestored = mnodeSuperTableActionRestored }; tsSuperTableSdb = sdbOpenTable(&tableDesc); @@ -602,10 +601,13 @@ int32_t mnodeInitTables() { mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_TABLE, mnodeGetShowTableMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_TABLE, mnodeRetrieveShowTables); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_TABLE, mnodeCancelGetNextChildTable); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_METRIC, mnodeGetShowSuperTableMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_METRIC, mnodeRetrieveShowSuperTables); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_METRIC, mnodeCancelGetNextSuperTable); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_STREAMTABLES, mnodeGetStreamTableMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_STREAMTABLES, mnodeRetrieveStreamTables); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_STREAMTABLES, mnodeCancelGetNextChildTable); return TSDB_CODE_SUCCESS; } @@ -626,14 +628,12 @@ static void *mnodeGetSuperTableByUid(uint64_t uid) { pIter = mnodeGetNextSuperTable(pIter, &pStable); if (pStable == NULL) break; if (pStable->uid == uid) { - sdbFreeIter(pIter); + mnodeCancelGetNextSuperTable(pIter); return pStable; } mnodeDecTableRef(pStable); } - sdbFreeIter(pIter); - return NULL; } @@ -655,10 +655,18 @@ void *mnodeGetNextChildTable(void *pIter, SCTableObj **pTable) { return sdbFetchRow(tsChildTableSdb, pIter, (void **)pTable); } +void mnodeCancelGetNextChildTable(void *pIter) { + sdbFreeIter(tsChildTableSdb, pIter); +} + void *mnodeGetNextSuperTable(void *pIter, SSTableObj **pTable) { return sdbFetchRow(tsSuperTableSdb, pIter, (void **)pTable); } +void mnodeCancelGetNextSuperTable(void *pIter) { + sdbFreeIter(tsSuperTableSdb, pIter); +} + void mnodeIncTableRef(void *p1) { STableObj *pTable = (STableObj *)p1; if (pTable->type == TSDB_SUPER_TABLE) { @@ -703,7 +711,7 @@ static int32_t mnodeProcessCreateTableMsg(SMnodeMsg *pMsg) { if (pMsg->pDb == NULL) pMsg->pDb = mnodeGetDb(pCreate->db); if (pMsg->pDb == NULL) { - mError("app:%p:%p, table:%s, failed to create, db not selected", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mError("msg:%p, app:%p table:%s, failed to create, db not selected", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return TSDB_CODE_MND_DB_NOT_SELECTED; } @@ -715,24 +723,24 @@ static int32_t mnodeProcessCreateTableMsg(SMnodeMsg *pMsg) { if (pMsg->pTable == NULL) pMsg->pTable = mnodeGetTable(pCreate->tableId); if (pMsg->pTable != NULL && pMsg->retry == 0) { if (pCreate->getMeta) { - mDebug("app:%p:%p, table:%s, continue to get meta", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mDebug("msg:%p, app:%p table:%s, continue to get meta", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return mnodeGetChildTableMeta(pMsg); } else if (pCreate->igExists) { - mDebug("app:%p:%p, table:%s, is already exist", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mDebug("msg:%p, app:%p table:%s, is already exist", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return TSDB_CODE_SUCCESS; } else { - mError("app:%p:%p, table:%s, failed to create, table already exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to create, table already exist", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return TSDB_CODE_MND_TABLE_ALREADY_EXIST; } } if (pCreate->numOfTags != 0) { - mDebug("app:%p:%p, table:%s, create stable msg is received from thandle:%p", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, create stable msg is received from thandle:%p", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId, pMsg->rpcMsg.handle); return mnodeProcessCreateSuperTableMsg(pMsg); } else { - mDebug("app:%p:%p, table:%s, create ctable msg is received from thandle:%p", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, create ctable msg is received from thandle:%p", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId, pMsg->rpcMsg.handle); return mnodeProcessCreateChildTableMsg(pMsg); } @@ -742,17 +750,18 @@ static int32_t mnodeProcessDropTableMsg(SMnodeMsg *pMsg) { SCMDropTableMsg *pDrop = pMsg->rpcMsg.pCont; if (pMsg->pDb == NULL) pMsg->pDb = mnodeGetDbByTableId(pDrop->tableId); if (pMsg->pDb == NULL) { - mError("app:%p:%p, table:%s, failed to drop table, db not selected or db in dropping", pMsg->rpcMsg.ahandle, pMsg, pDrop->tableId); + mError("msg:%p, app:%p table:%s, failed to drop table, db not selected or db in dropping", pMsg, + pMsg->rpcMsg.ahandle, pDrop->tableId); return TSDB_CODE_MND_DB_NOT_SELECTED; } - + if (pMsg->pDb->status != TSDB_DB_STATUS_READY) { mError("db:%s, status:%d, in dropping", pMsg->pDb->name, pMsg->pDb->status); return TSDB_CODE_MND_DB_IN_DROPPING; } if (mnodeCheckIsMonitorDB(pMsg->pDb->name, tsMonitorDbName)) { - mError("app:%p:%p, table:%s, failed to drop table, in monitor database", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to drop table, in monitor database", pMsg, pMsg->rpcMsg.ahandle, pDrop->tableId); return TSDB_CODE_MND_MONITOR_DB_FORBIDDEN; } @@ -760,22 +769,22 @@ static int32_t mnodeProcessDropTableMsg(SMnodeMsg *pMsg) { if (pMsg->pTable == NULL) pMsg->pTable = mnodeGetTable(pDrop->tableId); if (pMsg->pTable == NULL) { if (pDrop->igNotExists) { - mDebug("app:%p:%p, table:%s, table is not exist, treat as success", pMsg->rpcMsg.ahandle, pMsg, pDrop->tableId); + mDebug("msg:%p, app:%p table:%s is not exist, treat as success", pMsg, pMsg->rpcMsg.ahandle, pDrop->tableId); return TSDB_CODE_SUCCESS; } else { - mError("app:%p:%p, table:%s, failed to drop table, table not exist", pMsg->rpcMsg.ahandle, pMsg, pDrop->tableId); + mError("msg:%p, app:%p table:%s, failed to drop, table not exist", pMsg, pMsg->rpcMsg.ahandle, pDrop->tableId); return TSDB_CODE_MND_INVALID_TABLE_NAME; } } if (pMsg->pTable->type == TSDB_SUPER_TABLE) { SSTableObj *pSTable = (SSTableObj *)pMsg->pTable; - mInfo("app:%p:%p, table:%s, start to drop stable, uid:%" PRIu64 ", numOfChildTables:%d, sizeOfVgList:%d", - pMsg->rpcMsg.ahandle, pMsg, pDrop->tableId, pSTable->uid, pSTable->numOfTables, (int32_t)taosHashGetSize(pSTable->vgHash)); + mInfo("msg:%p, app:%p table:%s, start to drop stable, uid:%" PRIu64 ", numOfChildTables:%d, sizeOfVgList:%d", + pMsg, pMsg->rpcMsg.ahandle, pDrop->tableId, pSTable->uid, pSTable->numOfTables, (int32_t)taosHashGetSize(pSTable->vgHash)); return mnodeProcessDropSuperTableMsg(pMsg); } else { SCTableObj *pCTable = (SCTableObj *)pMsg->pTable; - mInfo("app:%p:%p, table:%s, start to drop ctable, vgId:%d tid:%d uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg, + mInfo("msg:%p, app:%p table:%s, start to drop ctable, vgId:%d tid:%d uid:%" PRIu64, pMsg, pMsg->rpcMsg.ahandle, pDrop->tableId, pCTable->vgId, pCTable->tid, pCTable->uid); return mnodeProcessDropChildTableMsg(pMsg); } @@ -784,12 +793,12 @@ static int32_t mnodeProcessDropTableMsg(SMnodeMsg *pMsg) { static int32_t mnodeProcessTableMetaMsg(SMnodeMsg *pMsg) { STableInfoMsg *pInfo = pMsg->rpcMsg.pCont; pInfo->createFlag = htons(pInfo->createFlag); - mDebug("app:%p:%p, table:%s, table meta msg is received from thandle:%p, createFlag:%d", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, table meta msg is received from thandle:%p, createFlag:%d", pMsg, pMsg->rpcMsg.ahandle, pInfo->tableId, pMsg->rpcMsg.handle, pInfo->createFlag); if (pMsg->pDb == NULL) pMsg->pDb = mnodeGetDbByTableId(pInfo->tableId); if (pMsg->pDb == NULL) { - mError("app:%p:%p, table:%s, failed to get table meta, db not selected", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to get table meta, db not selected", pMsg, pMsg->rpcMsg.ahandle, pInfo->tableId); return TSDB_CODE_MND_DB_NOT_SELECTED; } @@ -802,11 +811,11 @@ static int32_t mnodeProcessTableMetaMsg(SMnodeMsg *pMsg) { if (pMsg->pTable == NULL) pMsg->pTable = mnodeGetTable(pInfo->tableId); if (pMsg->pTable == NULL) { if (!pInfo->createFlag) { - mError("app:%p:%p, table:%s, failed to get table meta, table not exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to get table meta, table not exist", pMsg, pMsg->rpcMsg.ahandle, pInfo->tableId); return TSDB_CODE_MND_INVALID_TABLE_NAME; } else { - mDebug("app:%p:%p, table:%s, failed to get table meta, start auto create table ", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, failed to get table meta, start auto create table ", pMsg, pMsg->rpcMsg.ahandle, pInfo->tableId); return mnodeAutoCreateChildTable(pMsg); } @@ -826,9 +835,9 @@ static int32_t mnodeCreateSuperTableCb(SMnodeMsg *pMsg, int32_t code) { if (code == TSDB_CODE_SUCCESS) { mLInfo("stable:%s, is created in sdb, uid:%" PRIu64, pTable->info.tableId, pTable->uid); } else { - mError("app:%p:%p, stable:%s, failed to create in sdb, reason:%s", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mError("msg:%p, app:%p stable:%s, failed to create in sdb, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, tstrerror(code)); - SSdbOper desc = {.type = SDB_OPER_GLOBAL, .pObj = pTable, .table = tsSuperTableSdb}; + SSdbRow desc = {.type = SDB_OPER_GLOBAL, .pObj = pTable, .pTable = tsSuperTableSdb}; sdbDeleteRow(&desc); } @@ -841,7 +850,7 @@ static int32_t mnodeProcessCreateSuperTableMsg(SMnodeMsg *pMsg) { SCMCreateTableMsg *pCreate = pMsg->rpcMsg.pCont; SSTableObj * pStable = calloc(1, sizeof(SSTableObj)); if (pStable == NULL) { - mError("app:%p:%p, table:%s, failed to create, no enough memory", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mError("msg:%p, app:%p table:%s, failed to create, no enough memory", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return TSDB_CODE_MND_OUT_OF_MEMORY; } @@ -860,7 +869,7 @@ static int32_t mnodeProcessCreateSuperTableMsg(SMnodeMsg *pMsg) { pStable->schema = (SSchema *)calloc(1, schemaSize); if (pStable->schema == NULL) { free(pStable); - mError("app:%p:%p, table:%s, failed to create, no schema input", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mError("msg:%p, app:%p table:%s, failed to create, no schema input", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return TSDB_CODE_MND_INVALID_TABLE_NAME; } memcpy(pStable->schema, pCreate->schema, numOfCols * sizeof(SSchema)); @@ -878,20 +887,20 @@ static int32_t mnodeProcessCreateSuperTableMsg(SMnodeMsg *pMsg) { pMsg->pTable = (STableObj *)pStable; mnodeIncTableRef(pMsg->pTable); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, .rowSize = sizeof(SSTableObj) + schemaSize, - .pMsg = pMsg, - .writeCb = mnodeCreateSuperTableCb + .pMsg = pMsg, + .fpRsp = mnodeCreateSuperTableCb }; - int32_t code = sdbInsertRow(&oper); + int32_t code = sdbInsertRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mnodeDestroySuperTable(pStable); pMsg->pTable = NULL; - mError("app:%p:%p, table:%s, failed to create, sdb error", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mError("msg:%p, app:%p table:%s, failed to create, sdb error", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); } return code; @@ -900,9 +909,9 @@ static int32_t mnodeProcessCreateSuperTableMsg(SMnodeMsg *pMsg) { static int32_t mnodeDropSuperTableCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pTable = (SSTableObj *)pMsg->pTable; if (code != TSDB_CODE_SUCCESS) { - mError("app:%p:%p, stable:%s, failed to drop, sdb error", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId); + mError("msg:%p, app:%p stable:%s, failed to drop, sdb error", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); } else { - mLInfo("app:%p:%p, stable:%s, is dropped from sdb", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId); + mLInfo("msg:%p, app:%p stable:%s, is dropped from sdb", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); } return code; @@ -913,10 +922,10 @@ static int32_t mnodeProcessDropSuperTableMsg(SMnodeMsg *pMsg) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; if (pStable->vgHash != NULL /*pStable->numOfTables != 0*/) { - SHashMutableIterator *pIter = taosHashCreateIter(pStable->vgHash); - while (taosHashIterNext(pIter)) { - int32_t *pVgId = taosHashIterGet(pIter); + int32_t *pVgId = taosHashIterate(pStable->vgHash, NULL); + while (pVgId) { SVgObj *pVgroup = mnodeGetVgroup(*pVgId); + pVgId = taosHashIterate(pStable->vgHash, pVgId); if (pVgroup == NULL) break; SDropSTableMsg *pDrop = rpcMallocCont(sizeof(SDropSTableMsg)); @@ -925,29 +934,30 @@ static int32_t mnodeProcessDropSuperTableMsg(SMnodeMsg *pMsg) { pDrop->uid = htobe64(pStable->uid); mnodeExtractTableName(pStable->info.tableId, pDrop->tableId); - mInfo("app:%p:%p, stable:%s, send drop stable msg to vgId:%d", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, - pVgroup->vgId); + mInfo("msg:%p, app:%p stable:%s, send drop stable msg to vgId:%d", pMsg, pMsg->rpcMsg.ahandle, + pStable->info.tableId, pVgroup->vgId); SRpcEpSet epSet = mnodeGetEpSetFromVgroup(pVgroup); SRpcMsg rpcMsg = {.pCont = pDrop, .contLen = sizeof(SDropSTableMsg), .msgType = TSDB_MSG_TYPE_MD_DROP_STABLE}; dnodeSendMsgToDnode(&epSet, &rpcMsg); mnodeDecVgroupRef(pVgroup); } - taosHashDestroyIter(pIter); + + taosHashCancelIterate(pStable->vgHash, pVgId); mnodeDropAllChildTablesInStable(pStable); } - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeDropSuperTableCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeDropSuperTableCb }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { - mError("app:%p:%p, table:%s, failed to drop, reason:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mError("msg:%p, app:%p table:%s, failed to drop, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); } @@ -967,7 +977,7 @@ static int32_t mnodeFindSuperTableTagIndex(SSTableObj *pStable, const char *tagN static int32_t mnodeAddSuperTableTagCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, stable %s, add tag result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mLInfo("msg:%p, app:%p stable %s, add tag result:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); return code; @@ -976,20 +986,20 @@ static int32_t mnodeAddSuperTableTagCb(SMnodeMsg *pMsg, int32_t code) { static int32_t mnodeAddSuperTableTag(SMnodeMsg *pMsg, SSchema schema[], int32_t ntags) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; if (pStable->numOfTags + ntags > TSDB_MAX_TAGS) { - mError("app:%p:%p, stable:%s, add tag, too many tags", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId); + mError("msg:%p, app:%p stable:%s, add tag, too many tags", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId); return TSDB_CODE_MND_TOO_MANY_TAGS; } for (int32_t i = 0; i < ntags; i++) { if (mnodeFindSuperTableColumnIndex(pStable, schema[i].name) > 0) { - mError("app:%p:%p, stable:%s, add tag, column:%s already exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p stable:%s, add tag, column:%s already exist", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, schema[i].name); return TSDB_CODE_MND_TAG_ALREAY_EXIST; } if (mnodeFindSuperTableTagIndex(pStable, schema[i].name) > 0) { - mError("app:%p:%p, stable:%s, add tag, tag:%s already exist", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, - schema[i].name); + mError("msg:%p, app:%p stable:%s, add tag, tag:%s already exist", pMsg, pMsg->rpcMsg.ahandle, + pStable->info.tableId, schema[i].name); return TSDB_CODE_MND_FIELD_ALREAY_EXIST; } } @@ -1007,23 +1017,23 @@ static int32_t mnodeAddSuperTableTag(SMnodeMsg *pMsg, SSchema schema[], int32_t pStable->numOfTags += ntags; pStable->tversion++; - mInfo("app:%p:%p, stable %s, start to add tag %s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mInfo("msg:%p, app:%p stable %s, start to add tag %s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, schema[0].name); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeAddSuperTableTagCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeAddSuperTableTagCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeDropSuperTableTagCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, stable %s, drop tag result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mLInfo("msg:%p, app:%p stable %s, drop tag result:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); return code; } @@ -1032,7 +1042,7 @@ static int32_t mnodeDropSuperTableTag(SMnodeMsg *pMsg, char *tagName) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; int32_t col = mnodeFindSuperTableTagIndex(pStable, tagName); if (col < 0) { - mError("app:%p:%p, stable:%s, drop tag, tag:%s not exist", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mError("msg:%p, app:%p stable:%s, drop tag, tag:%s not exist", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tagName); return TSDB_CODE_MND_TAG_NOT_EXIST; } @@ -1042,22 +1052,22 @@ static int32_t mnodeDropSuperTableTag(SMnodeMsg *pMsg, char *tagName) { pStable->numOfTags--; pStable->tversion++; - mInfo("app:%p:%p, stable %s, start to drop tag %s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, tagName); + mInfo("msg:%p, app:%p stable %s, start to drop tag %s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tagName); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeDropSuperTableTagCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeDropSuperTableTagCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeModifySuperTableTagNameCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, stable %s, modify tag result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mLInfo("msg:%p, app:%p stable %s, modify tag result:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); return code; } @@ -1066,7 +1076,7 @@ static int32_t mnodeModifySuperTableTagName(SMnodeMsg *pMsg, char *oldTagName, c SSTableObj *pStable = (SSTableObj *)pMsg->pTable; int32_t col = mnodeFindSuperTableTagIndex(pStable, oldTagName); if (col < 0) { - mError("app:%p:%p, stable:%s, failed to modify table tag, oldName: %s, newName: %s", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p stable:%s, failed to modify table tag, oldName: %s, newName: %s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, oldTagName, newTagName); return TSDB_CODE_MND_TAG_NOT_EXIST; } @@ -1085,18 +1095,18 @@ static int32_t mnodeModifySuperTableTagName(SMnodeMsg *pMsg, char *oldTagName, c SSchema *schema = (SSchema *) (pStable->schema + pStable->numOfColumns + col); tstrncpy(schema->name, newTagName, sizeof(schema->name)); - mInfo("app:%p:%p, stable %s, start to modify tag %s to %s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mInfo("msg:%p, app:%p stable %s, start to modify tag %s to %s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, oldTagName, newTagName); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeModifySuperTableTagNameCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeModifySuperTableTagNameCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeFindSuperTableColumnIndex(SSTableObj *pStable, char *colName) { @@ -1112,7 +1122,7 @@ static int32_t mnodeFindSuperTableColumnIndex(SSTableObj *pStable, char *colName static int32_t mnodeAddSuperTableColumnCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, stable %s, add column result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mLInfo("msg:%p, app:%p stable %s, add column result:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); return code; } @@ -1121,19 +1131,19 @@ static int32_t mnodeAddSuperTableColumn(SMnodeMsg *pMsg, SSchema schema[], int32 SDbObj *pDb = pMsg->pDb; SSTableObj *pStable = (SSTableObj *)pMsg->pTable; if (ncols <= 0) { - mError("app:%p:%p, stable:%s, add column, ncols:%d <= 0", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, ncols); + mError("msg:%p, app:%p stable:%s, add column, ncols:%d <= 0", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, ncols); return TSDB_CODE_MND_APP_ERROR; } for (int32_t i = 0; i < ncols; i++) { if (mnodeFindSuperTableColumnIndex(pStable, schema[i].name) > 0) { - mError("app:%p:%p, stable:%s, add column, column:%s already exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p stable:%s, add column, column:%s already exist", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, schema[i].name); return TSDB_CODE_MND_FIELD_ALREAY_EXIST; } if (mnodeFindSuperTableTagIndex(pStable, schema[i].name) > 0) { - mError("app:%p:%p, stable:%s, add column, tag:%s already exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p stable:%s, add column, tag:%s already exist", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, schema[i].name); return TSDB_CODE_MND_TAG_ALREAY_EXIST; } @@ -1160,22 +1170,22 @@ static int32_t mnodeAddSuperTableColumn(SMnodeMsg *pMsg, SSchema schema[], int32 mnodeDecAcctRef(pAcct); } - mInfo("app:%p:%p, stable %s, start to add column", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId); + mInfo("msg:%p, app:%p stable %s, start to add column", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeAddSuperTableColumnCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeAddSuperTableColumnCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeDropSuperTableColumnCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, stable %s, delete column result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mLInfo("msg:%p, app:%p stable %s, delete column result:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); return code; } @@ -1185,7 +1195,7 @@ static int32_t mnodeDropSuperTableColumn(SMnodeMsg *pMsg, char *colName) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; int32_t col = mnodeFindSuperTableColumnIndex(pStable, colName); if (col <= 0) { - mError("app:%p:%p, stable:%s, drop column, column:%s not exist", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mError("msg:%p, app:%p stable:%s, drop column, column:%s not exist", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, colName); return TSDB_CODE_MND_FIELD_NOT_EXIST; } @@ -1205,22 +1215,22 @@ static int32_t mnodeDropSuperTableColumn(SMnodeMsg *pMsg, char *colName) { mnodeDecAcctRef(pAcct); } - mInfo("app:%p:%p, stable %s, start to delete column", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId); + mInfo("msg:%p, app:%p stable %s, start to delete column", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeDropSuperTableColumnCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeDropSuperTableColumnCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeChangeSuperTableColumnCb(SMnodeMsg *pMsg, int32_t code) { SSTableObj *pStable = (SSTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, stable %s, change column result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mLInfo("msg:%p, app:%p stable %s, change column result:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, tstrerror(code)); return code; } @@ -1229,7 +1239,7 @@ static int32_t mnodeChangeSuperTableColumn(SMnodeMsg *pMsg, char *oldName, char SSTableObj *pStable = (SSTableObj *)pMsg->pTable; int32_t col = mnodeFindSuperTableColumnIndex(pStable, oldName); if (col < 0) { - mError("app:%p:%p, stable:%s, change column, oldName: %s, newName: %s", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p stable:%s, change column, oldName:%s, newName:%s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, oldName, newName); return TSDB_CODE_MND_FIELD_NOT_EXIST; } @@ -1248,18 +1258,18 @@ static int32_t mnodeChangeSuperTableColumn(SMnodeMsg *pMsg, char *oldName, char SSchema *schema = (SSchema *) (pStable->schema + col); tstrncpy(schema->name, newName, sizeof(schema->name)); - mInfo("app:%p:%p, stable %s, start to modify column %s to %s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId, + mInfo("msg:%p, app:%p stable %s, start to modify column %s to %s", pMsg, pMsg->rpcMsg.ahandle, pStable->info.tableId, oldName, newName); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsSuperTableSdb, - .pObj = pStable, - .pMsg = pMsg, - .writeCb = mnodeChangeSuperTableColumnCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsSuperTableSdb, + .pObj = pStable, + .pMsg = pMsg, + .fpRsp = mnodeChangeSuperTableColumnCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } // show super tables @@ -1417,20 +1427,18 @@ void mnodeDropAllSuperTables(SDbObj *pDropDb) { if (pTable == NULL) break; if (strncmp(prefix, pTable->info.tableId, prefixLen) == 0) { - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsSuperTableSdb, - .pObj = pTable, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsSuperTableSdb, + .pObj = pTable, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfTables ++; } mnodeDecTableRef(pTable); } - sdbFreeIter(pIter); - mInfo("db:%s, all super tables:%d is dropped from sdb", pDropDb->name, numOfTables); } @@ -1470,7 +1478,7 @@ static int32_t mnodeGetSuperTableMeta(SMnodeMsg *pMsg) { pMsg->rpcRsp.rsp = pMeta; - mDebug("app:%p:%p, stable:%s, uid:%" PRIu64 " table meta is retrieved", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p stable:%s, uid:%" PRIu64 " table meta is retrieved", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->uid); return TSDB_CODE_SUCCESS; } @@ -1503,12 +1511,12 @@ static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *pMsg) { char * stableName = (char *)pInfo + sizeof(SSTableVgroupMsg) + (TSDB_TABLE_FNAME_LEN)*i; SSTableObj *pTable = mnodeGetSuperTable(stableName); if (pTable == NULL) { - mError("app:%p:%p, stable:%s, not exist while get stable vgroup info", pMsg->rpcMsg.ahandle, pMsg, stableName); + mError("msg:%p, app:%p stable:%s, not exist while get stable vgroup info", pMsg, pMsg->rpcMsg.ahandle, stableName); mnodeDecTableRef(pTable); continue; } if (pTable->vgHash == NULL) { - mDebug("app:%p:%p, stable:%s, no vgroup exist while get stable vgroup info", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p stable:%s, no vgroup exist while get stable vgroup info", pMsg, pMsg->rpcMsg.ahandle, stableName); mnodeDecTableRef(pTable); @@ -1522,11 +1530,11 @@ static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *pMsg) { } else { SVgroupsMsg *pVgroupMsg = (SVgroupsMsg *)msg; - SHashMutableIterator *pIter = taosHashCreateIter(pTable->vgHash); - int32_t vgSize = 0; - while (taosHashIterNext(pIter)) { - int32_t *pVgId = taosHashIterGet(pIter); - SVgObj * pVgroup = mnodeGetVgroup(*pVgId); + int32_t *pVgId = taosHashIterate(pTable->vgHash, NULL); + int32_t vgSize = 0; + while (pVgId) { + SVgObj *pVgroup = mnodeGetVgroup(*pVgId); + pVgId = taosHashIterate(pTable->vgHash, pVgId); if (pVgroup == NULL) continue; pVgroupMsg->vgroups[vgSize].vgId = htonl(pVgroup->vgId); @@ -1546,7 +1554,7 @@ static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *pMsg) { mnodeDecVgroupRef(pVgroup); } - taosHashDestroyIter(pIter); + taosHashCancelIterate(pTable->vgHash, pVgId); mnodeDecTableRef(pTable); pVgroupMsg->numOfVgroups = htonl(vgSize); @@ -1650,7 +1658,7 @@ static int32_t mnodeDoCreateChildTableFp(SMnodeMsg *pMsg) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; assert(pTable); - mDebug("app:%p:%p, table:%s, created in mnode, vgId:%d sid:%d, uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, created in mnode, vgId:%d sid:%d, uid:%" PRIu64, pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid); SCMCreateTableMsg *pCreate = pMsg->rpcMsg.pCont; @@ -1679,22 +1687,22 @@ static int32_t mnodeDoCreateChildTableCb(SMnodeMsg *pMsg, int32_t code) { if (code == TSDB_CODE_SUCCESS) { if (pCreate->getMeta) { - mDebug("app:%p:%p, table:%s, created in dnode and continue to get meta, thandle:%p", pMsg->rpcMsg.ahandle, pMsg, - pTable->info.tableId, pMsg->rpcMsg.handle); + mDebug("msg:%p, app:%p table:%s, created in dnode and continue to get meta, thandle:%p", pMsg, + pMsg->rpcMsg.ahandle, pTable->info.tableId, pMsg->rpcMsg.handle); pMsg->retry = 0; dnodeReprocessMWriteMsg(pMsg); } else { - mDebug("app:%p:%p, table:%s, created in dnode, thandle:%p", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mDebug("msg:%p, app:%p table:%s, created in dnode, thandle:%p", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pMsg->rpcMsg.handle); dnodeSendRpcMWriteRsp(pMsg, TSDB_CODE_SUCCESS); } return TSDB_CODE_MND_ACTION_IN_PROGRESS; } else { - mError("app:%p:%p, table:%s, failed to create table sid:%d, uid:%" PRIu64 ", reason:%s", pMsg->rpcMsg.ahandle, pMsg, - pTable->info.tableId, pTable->tid, pTable->uid, tstrerror(code)); - SSdbOper desc = {.type = SDB_OPER_GLOBAL, .pObj = pTable, .table = tsChildTableSdb}; + mError("msg:%p, app:%p table:%s, failed to create table sid:%d, uid:%" PRIu64 ", reason:%s", pMsg, + pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->tid, pTable->uid, tstrerror(code)); + SSdbRow desc = {.type = SDB_OPER_GLOBAL, .pObj = pTable, .pTable = tsChildTableSdb}; sdbDeleteRow(&desc); return code; } @@ -1705,7 +1713,7 @@ static int32_t mnodeDoCreateChildTable(SMnodeMsg *pMsg, int32_t tid) { SCMCreateTableMsg *pCreate = pMsg->rpcMsg.pCont; SCTableObj *pTable = calloc(1, sizeof(SCTableObj)); if (pTable == NULL) { - mError("app:%p:%p, table:%s, failed to alloc memory", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mError("msg:%p, app:%p table:%s, failed to alloc memory", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return TSDB_CODE_MND_OUT_OF_MEMORY; } @@ -1724,7 +1732,7 @@ static int32_t mnodeDoCreateChildTable(SMnodeMsg *pMsg, int32_t tid) { STagData *pTagData = (STagData *)pCreate->schema; // it is a tag key if (pMsg->pSTable == NULL) pMsg->pSTable = mnodeGetSuperTable(pTagData->name); if (pMsg->pSTable == NULL) { - mError("app:%p:%p, table:%s, corresponding super table:%s does not exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, corresponding super table:%s does not exist", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId, pTagData->name); mnodeDestroyChildTable(pTable); return TSDB_CODE_MND_INVALID_TABLE_NAME; @@ -1772,7 +1780,7 @@ static int32_t mnodeDoCreateChildTable(SMnodeMsg *pMsg, int32_t tid) { } memcpy(pTable->sql, (char *) (pCreate->schema) + numOfCols * sizeof(SSchema), pTable->sqlLen); pTable->sql[pTable->sqlLen - 1] = 0; - mDebug("app:%p:%p, table:%s, stream sql len:%d sql:%s", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mDebug("msg:%p, app:%p table:%s, stream sql len:%d sql:%s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->sqlLen, pTable->sql); } } @@ -1780,22 +1788,22 @@ static int32_t mnodeDoCreateChildTable(SMnodeMsg *pMsg, int32_t tid) { pMsg->pTable = (STableObj *)pTable; mnodeIncTableRef(pMsg->pTable); - SSdbOper desc = { - .type = SDB_OPER_GLOBAL, - .pObj = pTable, - .table = tsChildTableSdb, - .pMsg = pMsg, - .reqFp = mnodeDoCreateChildTableFp + SSdbRow desc = { + .type = SDB_OPER_GLOBAL, + .pObj = pTable, + .pTable = tsChildTableSdb, + .pMsg = pMsg, + .fpReq = mnodeDoCreateChildTableFp }; int32_t code = sdbInsertRow(&desc); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mnodeDestroyChildTable(pTable); pMsg->pTable = NULL; - mError("app:%p:%p, table:%s, failed to create, reason:%s", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId, + mError("msg:%p, app:%p table:%s, failed to create, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId, tstrerror(code)); } else { - mDebug("app:%p:%p, table:%s, allocated in vgroup, vgId:%d sid:%d uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, allocated in vgroup, vgId:%d sid:%d uid:%" PRIu64, pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pVgroup->vgId, pTable->tid, pTable->uid); } @@ -1806,7 +1814,7 @@ static int32_t mnodeProcessCreateChildTableMsg(SMnodeMsg *pMsg) { SCMCreateTableMsg *pCreate = pMsg->rpcMsg.pCont; int32_t code = grantCheck(TSDB_GRANT_TIMESERIES); if (code != TSDB_CODE_SUCCESS) { - mError("app:%p:%p, table:%s, failed to create, grant timeseries failed", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to create, grant timeseries failed", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return code; } @@ -1817,7 +1825,7 @@ static int32_t mnodeProcessCreateChildTableMsg(SMnodeMsg *pMsg) { int32_t tid = 0; code = mnodeGetAvailableVgroup(pMsg, &pVgroup, &tid); if (code != TSDB_CODE_SUCCESS) { - mDebug("app:%p:%p, table:%s, failed to get available vgroup, reason:%s", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, failed to get available vgroup, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId, tstrerror(code)); return code; } @@ -1836,22 +1844,22 @@ static int32_t mnodeProcessCreateChildTableMsg(SMnodeMsg *pMsg) { } if (pMsg->pTable == NULL) { - mError("app:%p:%p, table:%s, object not found, retry:%d reason:%s", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId, pMsg->retry, + mError("msg:%p, app:%p table:%s, object not found, retry:%d reason:%s", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId, pMsg->retry, tstrerror(terrno)); return terrno; } else { - mDebug("app:%p:%p, table:%s, send create msg to vnode again", pMsg->rpcMsg.ahandle, pMsg, pCreate->tableId); + mDebug("msg:%p, app:%p table:%s, send create msg to vnode again", pMsg, pMsg->rpcMsg.ahandle, pCreate->tableId); return mnodeDoCreateChildTableFp(pMsg); } } static int32_t mnodeSendDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; - mLInfo("app:%p:%p, ctable:%s, is dropped from sdb", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId); + mLInfo("msg:%p, app:%p ctable:%s, is dropped from sdb", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); SMDDropTableMsg *pDrop = rpcMallocCont(sizeof(SMDDropTableMsg)); if (pDrop == NULL) { - mError("app:%p:%p, ctable:%s, failed to drop ctable, no enough memory", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p ctable:%s, failed to drop ctable, no enough memory", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); return TSDB_CODE_MND_OUT_OF_MEMORY; } @@ -1864,7 +1872,7 @@ static int32_t mnodeSendDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) { SRpcEpSet epSet = mnodeGetEpSetFromVgroup(pMsg->pVgroup); - mInfo("app:%p:%p, ctable:%s, send drop ctable msg, vgId:%d sid:%d uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg, + mInfo("msg:%p, app:%p ctable:%s, send drop ctable msg, vgId:%d sid:%d uid:%" PRIu64, pMsg, pMsg->rpcMsg.ahandle, pDrop->tableId, pTable->vgId, pTable->tid, pTable->uid); SRpcMsg rpcMsg = { @@ -1885,7 +1893,7 @@ static int32_t mnodeSendDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) { static int32_t mnodeDropChildTableCb(SMnodeMsg *pMsg, int32_t code) { if (code != TSDB_CODE_SUCCESS) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; - mError("app:%p:%p, ctable:%s, failed to drop, sdb error", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId); + mError("msg:%p, app:%p ctable:%s, failed to drop, sdb error", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); return code; } @@ -1896,22 +1904,22 @@ static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; if (pMsg->pVgroup == NULL) pMsg->pVgroup = mnodeGetVgroup(pTable->vgId); if (pMsg->pVgroup == NULL) { - mError("app:%p:%p, table:%s, failed to drop ctable, vgroup not exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to drop ctable, vgroup not exist", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); return TSDB_CODE_MND_APP_ERROR; } - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsChildTableSdb, - .pObj = pTable, - .pMsg = pMsg, - .writeCb = mnodeDropChildTableCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsChildTableSdb, + .pObj = pTable, + .pMsg = pMsg, + .fpRsp = mnodeDropChildTableCb }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { - mError("app:%p:%p, ctable:%s, failed to drop, reason:%s", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mError("msg:%p, app:%p ctable:%s, failed to drop, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, tstrerror(code)); } @@ -1932,7 +1940,7 @@ static int32_t mnodeFindNormalTableColumnIndex(SCTableObj *pTable, char *colName static int32_t mnodeAlterNormalTableColumnCb(SMnodeMsg *pMsg, int32_t code) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; if (code != TSDB_CODE_SUCCESS) { - mError("app:%p:%p, ctable %s, failed to alter column, reason:%s", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mError("msg:%p, app:%p ctable %s, failed to alter column, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, tstrerror(code)); return code; } @@ -1946,7 +1954,7 @@ static int32_t mnodeAlterNormalTableColumnCb(SMnodeMsg *pMsg, int32_t code) { pMsg->pVgroup = mnodeGetVgroup(pTable->vgId); if (pMsg->pVgroup == NULL) { rpcFreeCont(pMDCreate); - mError("app:%p:%p, ctable %s, vgId:%d not exist in mnode", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mError("msg:%p, app:%p ctable %s, vgId:%d not exist in mnode", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->vgId); return TSDB_CODE_MND_VGROUP_NOT_EXIST; } @@ -1961,7 +1969,7 @@ static int32_t mnodeAlterNormalTableColumnCb(SMnodeMsg *pMsg, int32_t code) { .msgType = TSDB_MSG_TYPE_MD_ALTER_TABLE }; - mDebug("app:%p:%p, ctable %s, send alter column msg to vgId:%d", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mDebug("msg:%p, app:%p ctable %s, send alter column msg to vgId:%d", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pMsg->pVgroup->vgId); dnodeSendMsgToDnode(&epSet, &rpcMsg); @@ -1972,13 +1980,13 @@ static int32_t mnodeAddNormalTableColumn(SMnodeMsg *pMsg, SSchema schema[], int3 SCTableObj *pTable = (SCTableObj *)pMsg->pTable; SDbObj *pDb = pMsg->pDb; if (ncols <= 0) { - mError("app:%p:%p, ctable:%s, add column, ncols:%d <= 0", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, ncols); + mError("msg:%p, app:%p ctable:%s, add column, ncols:%d <= 0", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, ncols); return TSDB_CODE_MND_APP_ERROR; } for (int32_t i = 0; i < ncols; i++) { if (mnodeFindNormalTableColumnIndex(pTable, schema[i].name) > 0) { - mError("app:%p:%p, ctable:%s, add column, column:%s already exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p ctable:%s, add column, column:%s already exist", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, schema[i].name); return TSDB_CODE_MND_FIELD_ALREAY_EXIST; } @@ -2003,17 +2011,17 @@ static int32_t mnodeAddNormalTableColumn(SMnodeMsg *pMsg, SSchema schema[], int3 mnodeDecAcctRef(pAcct); } - mInfo("app:%p:%p, ctable %s, start to add column", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId); + mInfo("msg:%p, app:%p ctable %s, start to add column", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsChildTableSdb, - .pObj = pTable, - .pMsg = pMsg, - .writeCb = mnodeAlterNormalTableColumnCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsChildTableSdb, + .pObj = pTable, + .pMsg = pMsg, + .fpRsp = mnodeAlterNormalTableColumnCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeDropNormalTableColumn(SMnodeMsg *pMsg, char *colName) { @@ -2021,8 +2029,8 @@ static int32_t mnodeDropNormalTableColumn(SMnodeMsg *pMsg, char *colName) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; int32_t col = mnodeFindNormalTableColumnIndex(pTable, colName); if (col <= 0) { - mError("app:%p:%p, ctable:%s, drop column, column:%s not exist", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, - colName); + mError("msg:%p, app:%p ctable:%s, drop column, column:%s not exist", pMsg, pMsg->rpcMsg.ahandle, + pTable->info.tableId, colName); return TSDB_CODE_MND_FIELD_NOT_EXIST; } @@ -2036,24 +2044,24 @@ static int32_t mnodeDropNormalTableColumn(SMnodeMsg *pMsg, char *colName) { mnodeDecAcctRef(pAcct); } - mInfo("app:%p:%p, ctable %s, start to drop column %s", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, colName); + mInfo("msg:%p, app:%p ctable %s, start to drop column %s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, colName); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsChildTableSdb, - .pObj = pTable, - .pMsg = pMsg, - .writeCb = mnodeAlterNormalTableColumnCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsChildTableSdb, + .pObj = pTable, + .pMsg = pMsg, + .fpRsp = mnodeAlterNormalTableColumnCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeChangeNormalTableColumn(SMnodeMsg *pMsg, char *oldName, char *newName) { SCTableObj *pTable = (SCTableObj *)pMsg->pTable; int32_t col = mnodeFindNormalTableColumnIndex(pTable, oldName); if (col < 0) { - mError("app:%p:%p, ctable:%s, change column, oldName: %s, newName: %s", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p ctable:%s, change column, oldName: %s, newName: %s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, oldName, newName); return TSDB_CODE_MND_FIELD_NOT_EXIST; } @@ -2072,18 +2080,18 @@ static int32_t mnodeChangeNormalTableColumn(SMnodeMsg *pMsg, char *oldName, char SSchema *schema = (SSchema *) (pTable->schema + col); tstrncpy(schema->name, newName, sizeof(schema->name)); - mInfo("app:%p:%p, ctable %s, start to modify column %s to %s", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, + mInfo("msg:%p, app:%p ctable %s, start to modify column %s to %s", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, oldName, newName); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsChildTableSdb, - .pObj = pTable, - .pMsg = pMsg, - .writeCb = mnodeAlterNormalTableColumnCb + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsChildTableSdb, + .pObj = pTable, + .pMsg = pMsg, + .fpRsp = mnodeAlterNormalTableColumnCb }; - return sdbUpdateRow(&oper); + return sdbUpdateRow(&row); } static int32_t mnodeSetSchemaFromNormalTable(SSchema *pSchema, SCTableObj *pTable) { @@ -2128,7 +2136,7 @@ static int32_t mnodeDoGetChildTableMeta(SMnodeMsg *pMsg, STableMetaMsg *pMeta) { if (pMsg->pVgroup == NULL) pMsg->pVgroup = mnodeGetVgroup(pTable->vgId); if (pMsg->pVgroup == NULL) { - mError("app:%p:%p, table:%s, failed to get table meta, vgroup not exist", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to get table meta, vgroup not exist", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); return TSDB_CODE_MND_VGROUP_NOT_EXIST; } @@ -2143,7 +2151,7 @@ static int32_t mnodeDoGetChildTableMeta(SMnodeMsg *pMsg, STableMetaMsg *pMeta) { } pMeta->vgroup.vgId = htonl(pMsg->pVgroup->vgId); - mDebug("app:%p:%p, table:%s, uid:%" PRIu64 " table meta is retrieved, vgId:%d sid:%d", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, uid:%" PRIu64 " table meta is retrieved, vgId:%d sid:%d", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->uid, pTable->vgId, pTable->tid); return TSDB_CODE_SUCCESS; @@ -2154,16 +2162,16 @@ static int32_t mnodeAutoCreateChildTable(SMnodeMsg *pMsg) { STagData *pTags = (STagData *)pInfo->tags; int32_t tagLen = htonl(pTags->dataLen); if (pTags->name[0] == 0) { - mError("app:%p:%p, table:%s, failed to create table on demand for stable is empty, tagLen:%d", pMsg->rpcMsg.ahandle, - pMsg, pInfo->tableId, tagLen); - return TSDB_CODE_MND_INVALID_STABLE_NAME; + mError("msg:%p, app:%p table:%s, failed to create table on demand for stable is empty, tagLen:%d", pMsg, + pMsg->rpcMsg.ahandle, pInfo->tableId, tagLen); + return TSDB_CODE_MND_INVALID_STABLE_NAME; } int32_t contLen = sizeof(SCMCreateTableMsg) + offsetof(STagData, data) + tagLen; SCMCreateTableMsg *pCreateMsg = calloc(1, contLen); if (pCreateMsg == NULL) { - mError("app:%p:%p, table:%s, failed to create table while get meta info, no enough memory", pMsg->rpcMsg.ahandle, - pMsg, pInfo->tableId); + mError("msg:%p, app:%p table:%s, failed to create table while get meta info, no enough memory", pMsg, + pMsg->rpcMsg.ahandle, pInfo->tableId); return TSDB_CODE_MND_OUT_OF_MEMORY; } @@ -2175,8 +2183,8 @@ static int32_t mnodeAutoCreateChildTable(SMnodeMsg *pMsg) { pCreateMsg->contLen = htonl(contLen); memcpy(pCreateMsg->schema, pTags, contLen - sizeof(SCMCreateTableMsg)); - mDebug("app:%p:%p, table:%s, start to create on demand, tagLen:%d stable:%s", - pMsg->rpcMsg.ahandle, pMsg, pInfo->tableId, tagLen, pTags->name); + mDebug("msg:%p, app:%p table:%s, start to create on demand, tagLen:%d stable:%s", pMsg, pMsg->rpcMsg.ahandle, + pInfo->tableId, tagLen, pTags->name); if (pMsg->rpcMsg.pCont != pMsg->pCont) { tfree(pMsg->rpcMsg.pCont); @@ -2192,7 +2200,7 @@ static int32_t mnodeGetChildTableMeta(SMnodeMsg *pMsg) { STableMetaMsg *pMeta = rpcMallocCont(sizeof(STableMetaMsg) + sizeof(SSchema) * (TSDB_MAX_TAGS + TSDB_MAX_COLUMNS + 16)); if (pMeta == NULL) { - mError("app:%p:%p, table:%s, failed to get table meta, no enough memory", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p table:%s, failed to get table meta, no enough memory", pMsg, pMsg->rpcMsg.ahandle, pMsg->pTable->tableId); return TSDB_CODE_MND_OUT_OF_MEMORY; } @@ -2218,19 +2226,17 @@ void mnodeDropAllChildTablesInVgroups(SVgObj *pVgroup) { if (pTable == NULL) break; if (pTable->vgId == pVgroup->vgId) { - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsChildTableSdb, - .pObj = pTable, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsChildTableSdb, + .pObj = pTable, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfTables++; } mnodeDecTableRef(pTable); } - sdbFreeIter(pIter); - mInfo("vgId:%d, all child tables is dropped from sdb", pVgroup->vgId); } @@ -2251,19 +2257,17 @@ void mnodeDropAllChildTables(SDbObj *pDropDb) { if (pTable == NULL) break; if (strncmp(prefix, pTable->info.tableId, prefixLen) == 0) { - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsChildTableSdb, - .pObj = pTable, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsChildTableSdb, + .pObj = pTable, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfTables++; } mnodeDecTableRef(pTable); } - sdbFreeIter(pIter); - mInfo("db:%s, all child tables:%d is dropped from sdb", pDropDb->name, numOfTables); } @@ -2280,20 +2284,18 @@ static void mnodeDropAllChildTablesInStable(SSTableObj *pStable) { if (pTable == NULL) break; if (pTable->superTable == pStable) { - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsChildTableSdb, - .pObj = pTable, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsChildTableSdb, + .pObj = pTable, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfTables++; } mnodeDecTableRef(pTable); } - sdbFreeIter(pIter); - mInfo("stable:%s, all child tables:%d is dropped from sdb", pStable->info.tableId, numOfTables); } @@ -2317,12 +2319,12 @@ static int32_t mnodeProcessTableCfgMsg(SMnodeMsg *pMsg) { pCfg->dnodeId = htonl(pCfg->dnodeId); pCfg->vgId = htonl(pCfg->vgId); pCfg->sid = htonl(pCfg->sid); - mDebug("app:%p:%p, dnode:%d, vgId:%d sid:%d, receive table config msg", pMsg->rpcMsg.ahandle, pMsg, pCfg->dnodeId, + mDebug("msg:%p, app:%p dnode:%d, vgId:%d sid:%d, receive table config msg", pMsg, pMsg->rpcMsg.ahandle, pCfg->dnodeId, pCfg->vgId, pCfg->sid); SCTableObj *pTable = mnodeGetTableByPos(pCfg->vgId, pCfg->sid); if (pTable == NULL) { - mError("app:%p:%p, dnode:%d, vgId:%d sid:%d, table not found", pMsg->rpcMsg.ahandle, pMsg, pCfg->dnodeId, + mError("msg:%p, app:%p dnode:%d, vgId:%d sid:%d, table not found", pMsg, pMsg->rpcMsg.ahandle, pCfg->dnodeId, pCfg->vgId, pCfg->sid); return TSDB_CODE_MND_INVALID_TABLE_ID; } @@ -2343,38 +2345,37 @@ static int32_t mnodeProcessTableCfgMsg(SMnodeMsg *pMsg) { static void mnodeProcessDropChildTableRsp(SRpcMsg *rpcMsg) { if (rpcMsg->ahandle == NULL) return; - SMnodeMsg *mnodeMsg = rpcMsg->ahandle; - mnodeMsg->received++; + SMnodeMsg *pMsg = rpcMsg->ahandle; + pMsg->received++; - SCTableObj *pTable = (SCTableObj *)mnodeMsg->pTable; + SCTableObj *pTable = (SCTableObj *)pMsg->pTable; assert(pTable); - mInfo("app:%p:%p, table:%s, drop table rsp received, vgId:%d sid:%d uid:%" PRIu64 ", thandle:%p result:%s", - mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid, - mnodeMsg->rpcMsg.handle, tstrerror(rpcMsg->code)); + mInfo("msg:%p, app:%p table:%s, drop table rsp received, vgId:%d sid:%d uid:%" PRIu64 ", thandle:%p result:%s", pMsg, + pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid, pMsg->rpcMsg.handle, + tstrerror(rpcMsg->code)); if (rpcMsg->code != TSDB_CODE_SUCCESS) { - mError("app:%p:%p, table:%s, failed to drop in dnode, vgId:%d sid:%d uid:%" PRIu64 ", reason:%s", - mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid, - tstrerror(rpcMsg->code)); - dnodeSendRpcMWriteRsp(mnodeMsg, rpcMsg->code); + mError("msg:%p, app:%p table:%s, failed to drop in dnode, vgId:%d sid:%d uid:%" PRIu64 ", reason:%s", pMsg, + pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid, tstrerror(rpcMsg->code)); + dnodeSendRpcMWriteRsp(pMsg, rpcMsg->code); return; } - if (mnodeMsg->pVgroup == NULL) mnodeMsg->pVgroup = mnodeGetVgroup(pTable->vgId); - if (mnodeMsg->pVgroup == NULL) { - mError("app:%p:%p, table:%s, failed to get vgroup", mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId); - dnodeSendRpcMWriteRsp(mnodeMsg, TSDB_CODE_MND_VGROUP_NOT_EXIST); + if (pMsg->pVgroup == NULL) pMsg->pVgroup = mnodeGetVgroup(pTable->vgId); + if (pMsg->pVgroup == NULL) { + mError("msg:%p, app:%p table:%s, failed to get vgroup", pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId); + dnodeSendRpcMWriteRsp(pMsg, TSDB_CODE_MND_VGROUP_NOT_EXIST); return; } - if (mnodeMsg->pVgroup->numOfTables <= 0) { - mInfo("app:%p:%p, vgId:%d, all tables is dropped, drop vgroup", mnodeMsg->rpcMsg.ahandle, mnodeMsg, - mnodeMsg->pVgroup->vgId); - mnodeDropVgroup(mnodeMsg->pVgroup, NULL); + if (pMsg->pVgroup->numOfTables <= 0) { + mInfo("msg:%p, app:%p vgId:%d, all tables is dropped, drop vgroup", pMsg, pMsg->rpcMsg.ahandle, + pMsg->pVgroup->vgId); + mnodeDropVgroup(pMsg->pVgroup, NULL); } - dnodeSendRpcMWriteRsp(mnodeMsg, TSDB_CODE_SUCCESS); + dnodeSendRpcMWriteRsp(pMsg, TSDB_CODE_SUCCESS); } /* @@ -2384,70 +2385,70 @@ static void mnodeProcessDropChildTableRsp(SRpcMsg *rpcMsg) { static void mnodeProcessCreateChildTableRsp(SRpcMsg *rpcMsg) { if (rpcMsg->ahandle == NULL) return; - SMnodeMsg *mnodeMsg = rpcMsg->ahandle; - mnodeMsg->received++; + SMnodeMsg *pMsg = rpcMsg->ahandle; + pMsg->received++; - SCTableObj *pTable = (SCTableObj *)mnodeMsg->pTable; + SCTableObj *pTable = (SCTableObj *)pMsg->pTable; assert(pTable); // If the table is deleted by another thread during creation, stop creating and send drop msg to vnode if (sdbCheckRowDeleted(tsChildTableSdb, pTable)) { - mDebug("app:%p:%p, table:%s, create table rsp received, but a deleting opertion incoming, vgId:%d sid:%d uid:%" PRIu64, - mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid); + mDebug("msg:%p, app:%p table:%s, create table rsp received, but a deleting opertion incoming, vgId:%d sid:%d uid:%" PRIu64, + pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid); // if the vgroup is already dropped from hash, it can't be accquired by pTable->vgId // so the refCount of vgroup can not be decreased // SVgObj *pVgroup = mnodeGetVgroup(pTable->vgId); // if (pVgroup == NULL) { - // mnodeRemoveTableFromVgroup(mnodeMsg->pVgroup, pTable); + // mnodeRemoveTableFromVgroup(pMsg->pVgroup, pTable); // } // mnodeDecVgroupRef(pVgroup); - mnodeSendDropChildTableMsg(mnodeMsg, false); + mnodeSendDropChildTableMsg(pMsg, false); rpcMsg->code = TSDB_CODE_SUCCESS; - dnodeSendRpcMWriteRsp(mnodeMsg, rpcMsg->code); + dnodeSendRpcMWriteRsp(pMsg, rpcMsg->code); return; } if (rpcMsg->code == TSDB_CODE_SUCCESS || rpcMsg->code == TSDB_CODE_TDB_TABLE_ALREADY_EXIST) { - SSdbOper desc = { - .type = SDB_OPER_GLOBAL, - .pObj = pTable, - .table = tsChildTableSdb, - .pMsg = mnodeMsg, - .writeCb = mnodeDoCreateChildTableCb + SSdbRow desc = { + .type = SDB_OPER_GLOBAL, + .pObj = pTable, + .pTable = tsChildTableSdb, + .pMsg = pMsg, + .fpRsp = mnodeDoCreateChildTableCb }; - int32_t code = sdbInsertRowImp(&desc); + int32_t code = sdbInsertRowToQueue(&desc); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { - mnodeMsg->pTable = NULL; + pMsg->pTable = NULL; mnodeDestroyChildTable(pTable); - dnodeSendRpcMWriteRsp(mnodeMsg, code); + dnodeSendRpcMWriteRsp(pMsg, code); } } else { - mnodeMsg->retry++; + pMsg->retry++; int32_t sec = taosGetTimestampSec(); - if (mnodeMsg->retry < CREATE_CTABLE_RETRY_TIMES && ABS(sec - mnodeMsg->incomingTs) < CREATE_CTABLE_RETRY_SEC) { - mDebug("app:%p:%p, table:%s, create table rsp received, need retry, times:%d vgId:%d sid:%d uid:%" PRIu64 + if (pMsg->retry < CREATE_CTABLE_RETRY_TIMES && ABS(sec - pMsg->incomingTs) < CREATE_CTABLE_RETRY_SEC) { + mDebug("msg:%p, app:%p table:%s, create table rsp received, need retry, times:%d vgId:%d sid:%d uid:%" PRIu64 " result:%s thandle:%p", - mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId, mnodeMsg->retry, pTable->vgId, pTable->tid, - pTable->uid, tstrerror(rpcMsg->code), mnodeMsg->rpcMsg.handle); + pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pMsg->retry, pTable->vgId, pTable->tid, pTable->uid, + tstrerror(rpcMsg->code), pMsg->rpcMsg.handle); - dnodeDelayReprocessMWriteMsg(mnodeMsg); + dnodeDelayReprocessMWriteMsg(pMsg); } else { - mError("app:%p:%p, table:%s, failed to create in dnode, vgId:%d sid:%d uid:%" PRIu64 + mError("msg:%p, app:%p table:%s, failed to create in dnode, vgId:%d sid:%d uid:%" PRIu64 ", result:%s thandle:%p incomingTs:%d curTs:%d retryTimes:%d", - mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid, - tstrerror(rpcMsg->code), mnodeMsg->rpcMsg.handle, mnodeMsg->incomingTs, sec, mnodeMsg->retry); + pMsg, pMsg->rpcMsg.ahandle, pTable->info.tableId, pTable->vgId, pTable->tid, pTable->uid, + tstrerror(rpcMsg->code), pMsg->rpcMsg.handle, pMsg->incomingTs, sec, pMsg->retry); - SSdbOper oper = {.type = SDB_OPER_GLOBAL, .table = tsChildTableSdb, .pObj = pTable}; - sdbDeleteRow(&oper); + SSdbRow row = {.type = SDB_OPER_GLOBAL, .pTable = tsChildTableSdb, .pObj = pTable}; + sdbDeleteRow(&row); if (rpcMsg->code == TSDB_CODE_APP_NOT_READY) { //Avoid retry again in client rpcMsg->code = TSDB_CODE_MND_VGROUP_NOT_READY; } - dnodeSendRpcMWriteRsp(mnodeMsg, rpcMsg->code); + dnodeSendRpcMWriteRsp(pMsg, rpcMsg->code); } } } @@ -2455,28 +2456,28 @@ static void mnodeProcessCreateChildTableRsp(SRpcMsg *rpcMsg) { static void mnodeProcessAlterTableRsp(SRpcMsg *rpcMsg) { if (rpcMsg->ahandle == NULL) return; - SMnodeMsg *mnodeMsg = rpcMsg->ahandle; - mnodeMsg->received++; + SMnodeMsg *pMsg = rpcMsg->ahandle; + pMsg->received++; - SCTableObj *pTable = (SCTableObj *)mnodeMsg->pTable; + SCTableObj *pTable = (SCTableObj *)pMsg->pTable; assert(pTable); if (rpcMsg->code == TSDB_CODE_SUCCESS || rpcMsg->code == TSDB_CODE_TDB_TABLE_ALREADY_EXIST) { - mDebug("app:%p:%p, ctable:%s, altered in dnode, thandle:%p result:%s", mnodeMsg->rpcMsg.ahandle, mnodeMsg, - pTable->info.tableId, mnodeMsg->rpcMsg.handle, tstrerror(rpcMsg->code)); + mDebug("msg:%p, app:%p ctable:%s, altered in dnode, thandle:%p result:%s", pMsg, pMsg->rpcMsg.ahandle, + pTable->info.tableId, pMsg->rpcMsg.handle, tstrerror(rpcMsg->code)); - dnodeSendRpcMWriteRsp(mnodeMsg, TSDB_CODE_SUCCESS); + dnodeSendRpcMWriteRsp(pMsg, TSDB_CODE_SUCCESS); } else { - if (mnodeMsg->retry++ < ALTER_CTABLE_RETRY_TIMES) { - mDebug("app:%p:%p, table:%s, alter table rsp received, need retry, times:%d result:%s thandle:%p", - mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId, mnodeMsg->retry, tstrerror(rpcMsg->code), - mnodeMsg->rpcMsg.handle); + if (pMsg->retry++ < ALTER_CTABLE_RETRY_TIMES) { + mDebug("msg:%p, app:%p table:%s, alter table rsp received, need retry, times:%d result:%s thandle:%p", + pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId, pMsg->retry, tstrerror(rpcMsg->code), + pMsg->rpcMsg.handle); - dnodeDelayReprocessMWriteMsg(mnodeMsg); + dnodeDelayReprocessMWriteMsg(pMsg); } else { - mError("app:%p:%p, table:%s, failed to alter in dnode, result:%s thandle:%p", mnodeMsg->rpcMsg.ahandle, mnodeMsg, - pTable->info.tableId, tstrerror(rpcMsg->code), mnodeMsg->rpcMsg.handle); - dnodeSendRpcMWriteRsp(mnodeMsg, rpcMsg->code); + mError("msg:%p, app:%p table:%s, failed to alter in dnode, result:%s thandle:%p", pMsg, pMsg->rpcMsg.ahandle, + pTable->info.tableId, tstrerror(rpcMsg->code), pMsg->rpcMsg.handle); + dnodeSendRpcMWriteRsp(pMsg, rpcMsg->code); } } } @@ -2719,12 +2720,12 @@ static int32_t mnodeRetrieveShowTables(SShowObj *pShow, char *data, int32_t rows static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) { SAlterTableMsg *pAlter = pMsg->rpcMsg.pCont; - mDebug("app:%p:%p, table:%s, alter table msg is received from thandle:%p", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p table:%s, alter table msg is received from thandle:%p", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId, pMsg->rpcMsg.handle); if (pMsg->pDb == NULL) pMsg->pDb = mnodeGetDbByTableId(pAlter->tableId); if (pMsg->pDb == NULL) { - mError("app:%p:%p, table:%s, failed to alter table, db not selected", pMsg->rpcMsg.ahandle, pMsg, pAlter->tableId); + mError("msg:%p, app:%p table:%s, failed to alter table, db not selected", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId); return TSDB_CODE_MND_DB_NOT_SELECTED; } @@ -2734,13 +2735,13 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) { } if (mnodeCheckIsMonitorDB(pMsg->pDb->name, tsMonitorDbName)) { - mError("app:%p:%p, table:%s, failed to alter table, its log db", pMsg->rpcMsg.ahandle, pMsg, pAlter->tableId); + mError("msg:%p, app:%p table:%s, failed to alter table, its log db", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId); return TSDB_CODE_MND_MONITOR_DB_FORBIDDEN; } if (pMsg->pTable == NULL) pMsg->pTable = mnodeGetTable(pAlter->tableId); if (pMsg->pTable == NULL) { - mError("app:%p:%p, table:%s, failed to alter table, table not exist", pMsg->rpcMsg.ahandle, pMsg, pAlter->tableId); + mError("msg:%p, app:%p table:%s, failed to alter table, table not exist", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId); return TSDB_CODE_MND_INVALID_TABLE_NAME; } @@ -2749,7 +2750,7 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) { pAlter->tagValLen = htonl(pAlter->tagValLen); if (pAlter->numOfCols > 2) { - mError("app:%p:%p, table:%s, error numOfCols:%d in alter table", pMsg->rpcMsg.ahandle, pMsg, pAlter->tableId, + mError("msg:%p, app:%p table:%s, error numOfCols:%d in alter table", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId, pAlter->numOfCols); return TSDB_CODE_MND_APP_ERROR; } @@ -2760,7 +2761,7 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) { int32_t code = TSDB_CODE_COM_OPS_NOT_SUPPORT; if (pMsg->pTable->type == TSDB_SUPER_TABLE) { - mDebug("app:%p:%p, table:%s, start to alter stable", pMsg->rpcMsg.ahandle, pMsg, pAlter->tableId); + mDebug("msg:%p, app:%p table:%s, start to alter stable", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId); if (pAlter->type == TSDB_ALTER_TABLE_ADD_TAG_COLUMN) { code = mnodeAddSuperTableTag(pMsg, pAlter->schema, 1); } else if (pAlter->type == TSDB_ALTER_TABLE_DROP_TAG_COLUMN) { @@ -2776,7 +2777,7 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) { } else { } } else { - mDebug("app:%p:%p, table:%s, start to alter ctable", pMsg->rpcMsg.ahandle, pMsg, pAlter->tableId); + mDebug("msg:%p, app:%p table:%s, start to alter ctable", pMsg, pMsg->rpcMsg.ahandle, pAlter->tableId); if (pAlter->type == TSDB_ALTER_TABLE_UPDATE_TAG_VAL) { return TSDB_CODE_COM_OPS_NOT_SUPPORT; } else if (pAlter->type == TSDB_ALTER_TABLE_ADD_COLUMN) { diff --git a/src/mnode/src/mnodeUser.c b/src/mnode/src/mnodeUser.c index cd02d5a935..aee167631f 100644 --- a/src/mnode/src/mnodeUser.c +++ b/src/mnode/src/mnodeUser.c @@ -42,13 +42,13 @@ static int32_t mnodeProcessAlterUserMsg(SMnodeMsg *pMsg); static int32_t mnodeProcessDropUserMsg(SMnodeMsg *pMsg); static int32_t mnodeProcessAuthMsg(SMnodeMsg *pMsg); -static int32_t mnodeUserActionDestroy(SSdbOper *pOper) { - tfree(pOper->pObj); +static int32_t mnodeUserActionDestroy(SSdbRow *pRow) { + tfree(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeUserActionInsert(SSdbOper *pOper) { - SUserObj *pUser = pOper->pObj; +static int32_t mnodeUserActionInsert(SSdbRow *pRow) { + SUserObj *pUser = pRow->pObj; SAcctObj *pAcct = mnodeGetAcct(pUser->acct); if (pAcct != NULL) { @@ -62,8 +62,8 @@ static int32_t mnodeUserActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeUserActionDelete(SSdbOper *pOper) { - SUserObj *pUser = pOper->pObj; +static int32_t mnodeUserActionDelete(SSdbRow *pRow) { + SUserObj *pUser = pRow->pObj; SAcctObj *pAcct = mnodeGetAcct(pUser->acct); if (pAcct != NULL) { @@ -74,8 +74,8 @@ static int32_t mnodeUserActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeUserActionUpdate(SSdbOper *pOper) { - SUserObj *pUser = pOper->pObj; +static int32_t mnodeUserActionUpdate(SSdbRow *pRow) { + SUserObj *pUser = pRow->pObj; SUserObj *pSaved = mnodeGetUser(pUser->user); if (pUser != pSaved) { memcpy(pSaved, pUser, tsUserUpdateSize); @@ -85,19 +85,19 @@ static int32_t mnodeUserActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeUserActionEncode(SSdbOper *pOper) { - SUserObj *pUser = pOper->pObj; - memcpy(pOper->rowData, pUser, tsUserUpdateSize); - pOper->rowSize = tsUserUpdateSize; +static int32_t mnodeUserActionEncode(SSdbRow *pRow) { + SUserObj *pUser = pRow->pObj; + memcpy(pRow->rowData, pUser, tsUserUpdateSize); + pRow->rowSize = tsUserUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeUserActionDecode(SSdbOper *pOper) { +static int32_t mnodeUserActionDecode(SSdbRow *pRow) { SUserObj *pUser = (SUserObj *)calloc(1, sizeof(SUserObj)); if (pUser == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pUser, pOper->rowData, tsUserUpdateSize); - pOper->pObj = pUser; + memcpy(pUser, pRow->rowData, tsUserUpdateSize); + pRow->pObj = pUser; return TSDB_CODE_SUCCESS; } @@ -123,7 +123,6 @@ static void mnodePrintUserAuth() { } fflush(fp); - sdbFreeIter(pIter); fclose(fp); } @@ -150,25 +149,25 @@ int32_t mnodeInitUsers() { SUserObj tObj; tsUserUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_USER, - .tableName = "users", + SSdbTableDesc desc = { + .id = SDB_TABLE_USER, + .name = "users", .hashSessions = TSDB_DEFAULT_USERS_HASH_SIZE, .maxRowSize = tsUserUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_STRING, - .insertFp = mnodeUserActionInsert, - .deleteFp = mnodeUserActionDelete, - .updateFp = mnodeUserActionUpdate, - .encodeFp = mnodeUserActionEncode, - .decodeFp = mnodeUserActionDecode, - .destroyFp = mnodeUserActionDestroy, - .restoredFp = mnodeUserActionRestored + .fpInsert = mnodeUserActionInsert, + .fpDelete = mnodeUserActionDelete, + .fpUpdate = mnodeUserActionUpdate, + .fpEncode = mnodeUserActionEncode, + .fpDecode = mnodeUserActionDecode, + .fpDestroy = mnodeUserActionDestroy, + .fpRestored = mnodeUserActionRestored }; - tsUserSdb = sdbOpenTable(&tableDesc); + tsUserSdb = sdbOpenTable(&desc); if (tsUserSdb == NULL) { - mError("table:%s, failed to create hash", tableDesc.tableName); + mError("table:%s, failed to create hash", desc.name); return -1; } @@ -177,9 +176,11 @@ int32_t mnodeInitUsers() { mnodeAddWriteMsgHandle(TSDB_MSG_TYPE_CM_DROP_USER, mnodeProcessDropUserMsg); mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_USER, mnodeGetUserMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_USER, mnodeRetrieveUsers); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_USER, mnodeCancelGetNextUser); + mnodeAddPeerMsgHandle(TSDB_MSG_TYPE_DM_AUTH, mnodeProcessAuthMsg); - mDebug("table:%s, hash is created", tableDesc.tableName); + mDebug("table:%s, hash is created", desc.name); return 0; } @@ -196,6 +197,10 @@ void *mnodeGetNextUser(void *pIter, SUserObj **pUser) { return sdbFetchRow(tsUserSdb, pIter, (void **)pUser); } +void mnodeCancelGetNextUser(void *pIter) { + sdbFreeIter(tsUserSdb, pIter); +} + void mnodeIncUserRef(SUserObj *pUser) { return sdbIncRef(tsUserSdb, pUser); } @@ -205,14 +210,14 @@ void mnodeDecUserRef(SUserObj *pUser) { } static int32_t mnodeUpdateUser(SUserObj *pUser, void *pMsg) { - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsUserSdb, - .pObj = pUser, - .pMsg = pMsg + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsUserSdb, + .pObj = pUser, + .pMsg = pMsg }; - int32_t code = sdbUpdateRow(&oper); + int32_t code = sdbUpdateRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("user:%s, failed to alter by %s, reason:%s", pUser->user, mnodeGetUserFromMsg(pMsg), tstrerror(code)); } else { @@ -259,15 +264,15 @@ int32_t mnodeCreateUser(SAcctObj *pAcct, char *name, char *pass, void *pMsg) { pUser->superAuth = 1; } - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsUserSdb, - .pObj = pUser, - .rowSize = sizeof(SUserObj), - .pMsg = pMsg + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsUserSdb, + .pObj = pUser, + .rowSize = sizeof(SUserObj), + .pMsg = pMsg }; - code = sdbInsertRow(&oper); + code = sdbInsertRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("user:%s, failed to create by %s, reason:%s", pUser->user, mnodeGetUserFromMsg(pMsg), tstrerror(code)); tfree(pUser); @@ -279,14 +284,14 @@ int32_t mnodeCreateUser(SAcctObj *pAcct, char *name, char *pass, void *pMsg) { } static int32_t mnodeDropUser(SUserObj *pUser, void *pMsg) { - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsUserSdb, - .pObj = pUser, - .pMsg = pMsg + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsUserSdb, + .pObj = pUser, + .pMsg = pMsg }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("user:%s, failed to drop by %s, reason:%s", pUser->user, mnodeGetUserFromMsg(pMsg), tstrerror(code)); } else { @@ -562,20 +567,18 @@ void mnodeDropAllUsers(SAcctObj *pAcct) { if (pUser == NULL) break; if (strncmp(pUser->acct, pAcct->user, acctNameLen) == 0) { - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsUserSdb, - .pObj = pUser, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsUserSdb, + .pObj = pUser, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfUsers++; } mnodeDecUserRef(pUser); } - sdbFreeIter(pIter); - mDebug("acct:%s, all users:%d is dropped from sdb", pAcct->user, numOfUsers); } diff --git a/src/mnode/src/mnodeVgroup.c b/src/mnode/src/mnodeVgroup.c index 9e398e94f1..a8343083aa 100644 --- a/src/mnode/src/mnodeVgroup.c +++ b/src/mnode/src/mnodeVgroup.c @@ -72,13 +72,13 @@ static void mnodeDestroyVgroup(SVgObj *pVgroup) { tfree(pVgroup); } -static int32_t mnodeVgroupActionDestroy(SSdbOper *pOper) { - mnodeDestroyVgroup(pOper->pObj); +static int32_t mnodeVgroupActionDestroy(SSdbRow *pRow) { + mnodeDestroyVgroup(pRow->pObj); return TSDB_CODE_SUCCESS; } -static int32_t mnodeVgroupActionInsert(SSdbOper *pOper) { - SVgObj *pVgroup = pOper->pObj; +static int32_t mnodeVgroupActionInsert(SSdbRow *pRow) { + SVgObj *pVgroup = pRow->pObj; // refer to db SDbObj *pDb = mnodeGetDb(pVgroup->dbName); @@ -115,8 +115,8 @@ static int32_t mnodeVgroupActionInsert(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeVgroupActionDelete(SSdbOper *pOper) { - SVgObj *pVgroup = pOper->pObj; +static int32_t mnodeVgroupActionDelete(SSdbRow *pRow) { + SVgObj *pVgroup = pRow->pObj; if (pVgroup->pDb == NULL) { mError("vgId:%d, db:%s is not exist while insert into hash", pVgroup->vgId, pVgroup->dbName); @@ -137,8 +137,8 @@ static int32_t mnodeVgroupActionDelete(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeVgroupActionUpdate(SSdbOper *pOper) { - SVgObj *pNew = pOper->pObj; +static int32_t mnodeVgroupActionUpdate(SSdbRow *pRow) { + SVgObj *pNew = pRow->pObj; SVgObj *pVgroup = mnodeGetVgroup(pNew->vgId); if (pVgroup != pNew) { @@ -176,25 +176,25 @@ static int32_t mnodeVgroupActionUpdate(SSdbOper *pOper) { return TSDB_CODE_SUCCESS; } -static int32_t mnodeVgroupActionEncode(SSdbOper *pOper) { - SVgObj *pVgroup = pOper->pObj; - memcpy(pOper->rowData, pVgroup, tsVgUpdateSize); - SVgObj *pTmpVgroup = pOper->rowData; +static int32_t mnodeVgroupActionEncode(SSdbRow *pRow) { + SVgObj *pVgroup = pRow->pObj; + memcpy(pRow->rowData, pVgroup, tsVgUpdateSize); + SVgObj *pTmpVgroup = pRow->rowData; for (int32_t i = 0; i < TSDB_MAX_REPLICA; ++i) { pTmpVgroup->vnodeGid[i].pDnode = NULL; pTmpVgroup->vnodeGid[i].role = 0; } - pOper->rowSize = tsVgUpdateSize; + pRow->rowSize = tsVgUpdateSize; return TSDB_CODE_SUCCESS; } -static int32_t mnodeVgroupActionDecode(SSdbOper *pOper) { +static int32_t mnodeVgroupActionDecode(SSdbRow *pRow) { SVgObj *pVgroup = (SVgObj *) calloc(1, sizeof(SVgObj)); if (pVgroup == NULL) return TSDB_CODE_MND_OUT_OF_MEMORY; - memcpy(pVgroup, pOper->rowData, tsVgUpdateSize); - pOper->pObj = pVgroup; + memcpy(pVgroup, pRow->rowData, tsVgUpdateSize); + pRow->pObj = pVgroup; return TSDB_CODE_SUCCESS; } @@ -206,23 +206,23 @@ int32_t mnodeInitVgroups() { SVgObj tObj; tsVgUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj; - SSdbTableDesc tableDesc = { - .tableId = SDB_TABLE_VGROUP, - .tableName = "vgroups", + SSdbTableDesc desc = { + .id = SDB_TABLE_VGROUP, + .name = "vgroups", .hashSessions = TSDB_DEFAULT_VGROUPS_HASH_SIZE, .maxRowSize = tsVgUpdateSize, .refCountPos = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj, .keyType = SDB_KEY_AUTO, - .insertFp = mnodeVgroupActionInsert, - .deleteFp = mnodeVgroupActionDelete, - .updateFp = mnodeVgroupActionUpdate, - .encodeFp = mnodeVgroupActionEncode, - .decodeFp = mnodeVgroupActionDecode, - .destroyFp = mnodeVgroupActionDestroy, - .restoredFp = mnodeVgroupActionRestored, + .fpInsert = mnodeVgroupActionInsert, + .fpDelete = mnodeVgroupActionDelete, + .fpUpdate = mnodeVgroupActionUpdate, + .fpEncode = mnodeVgroupActionEncode, + .fpDecode = mnodeVgroupActionDecode, + .fpDestroy = mnodeVgroupActionDestroy, + .fpRestored = mnodeVgroupActionRestored, }; - tsVgroupSdb = sdbOpenTable(&tableDesc); + tsVgroupSdb = sdbOpenTable(&desc); if (tsVgroupSdb == NULL) { mError("failed to init vgroups data"); return -1; @@ -230,6 +230,7 @@ int32_t mnodeInitVgroups() { mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_VGROUP, mnodeGetVgroupMeta); mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_VGROUP, mnodeRetrieveVgroups); + mnodeAddShowFreeIterHandle(TSDB_MGMT_TABLE_VGROUP, mnodeCancelGetNextVgroup); mnodeAddPeerRspHandle(TSDB_MSG_TYPE_MD_CREATE_VNODE_RSP, mnodeProcessCreateVnodeRsp); mnodeAddPeerRspHandle(TSDB_MSG_TYPE_MD_ALTER_VNODE_RSP, mnodeProcessAlterVnodeRsp); mnodeAddPeerRspHandle(TSDB_MSG_TYPE_MD_DROP_VNODE_RSP, mnodeProcessDropVnodeRsp); @@ -253,13 +254,13 @@ SVgObj *mnodeGetVgroup(int32_t vgId) { } void mnodeUpdateVgroup(SVgObj *pVgroup) { - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsVgroupSdb, - .pObj = pVgroup + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup }; - int32_t code = sdbUpdateRow(&oper); + int32_t code = sdbUpdateRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mError("vgId:%d, failed to update vgroup", pVgroup->vgId); } @@ -304,7 +305,7 @@ void mnodeCheckUnCreatedVgroup(SDnodeObj *pDnode, SVnodeLoad *pVloads, int32_t o mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); + mnodeCancelGetNextVgroup(pIter); } void mnodeUpdateVgroupStatus(SVgObj *pVgroup, SDnodeObj *pDnode, SVnodeLoad *pVload) { @@ -421,7 +422,7 @@ int32_t mnodeGetAvailableVgroup(SMnodeMsg *pMsg, SVgObj **ppVgroup, int32_t *pSi int32_t sid = taosAllocateId(pVgroup->idPool); if (sid <= 0) { - mDebug("app:%p:%p, db:%s, no enough sid in vgId:%d", pMsg->rpcMsg.ahandle, pMsg, pDb->name, pVgroup->vgId); + mDebug("msg:%p, app:%p db:%s, no enough sid in vgId:%d", pMsg, pMsg->rpcMsg.ahandle, pDb->name, pVgroup->vgId); continue; } @@ -442,8 +443,8 @@ int32_t mnodeGetAvailableVgroup(SMnodeMsg *pMsg, SVgObj **ppVgroup, int32_t *pSi int32_t code = TSDB_CODE_MND_NO_ENOUGH_DNODES; if (pDb->numOfVgroups < maxVgroupsPerDb) { - mDebug("app:%p:%p, db:%s, try to create a new vgroup, numOfVgroups:%d maxVgroupsPerDb:%d", pMsg->rpcMsg.ahandle, - pMsg, pDb->name, pDb->numOfVgroups, maxVgroupsPerDb); + mDebug("msg:%p, app:%p db:%s, try to create a new vgroup, numOfVgroups:%d maxVgroupsPerDb:%d", pMsg, + pMsg->rpcMsg.ahandle, pDb->name, pDb->numOfVgroups, maxVgroupsPerDb); pthread_mutex_unlock(&pDb->mutex); code = mnodeCreateVgroup(pMsg); if (code == TSDB_CODE_MND_ACTION_IN_PROGRESS) { @@ -455,8 +456,8 @@ int32_t mnodeGetAvailableVgroup(SMnodeMsg *pMsg, SVgObj **ppVgroup, int32_t *pSi if (pDb->numOfVgroups < 1) { pthread_mutex_unlock(&pDb->mutex); - mDebug("app:%p:%p, db:%s, failed create new vgroup since:%s, numOfVgroups:%d maxVgroupsPerDb:%d ", - pMsg->rpcMsg.ahandle, pMsg, pDb->name, tstrerror(code), pDb->numOfVgroups, maxVgroupsPerDb); + mDebug("msg:%p, app:%p db:%s, failed create new vgroup since:%s, numOfVgroups:%d maxVgroupsPerDb:%d ", pMsg, + pMsg->rpcMsg.ahandle, pDb->name, tstrerror(code), pDb->numOfVgroups, maxVgroupsPerDb); return code; } @@ -474,7 +475,7 @@ int32_t mnodeGetAvailableVgroup(SMnodeMsg *pMsg, SVgObj **ppVgroup, int32_t *pSi int32_t sid = taosAllocateId(pVgroup->idPool); if (sid <= 0) { - mError("app:%p:%p, db:%s, no enough sid in vgId:%d", pMsg->rpcMsg.ahandle, pMsg, pDb->name, pVgroup->vgId); + mError("msg:%p, app:%p db:%s, no enough sid in vgId:%d", pMsg, pMsg->rpcMsg.ahandle, pDb->name, pVgroup->vgId); pthread_mutex_unlock(&pDb->mutex); return TSDB_CODE_MND_NO_ENOUGH_DNODES; } @@ -491,15 +492,19 @@ void *mnodeGetNextVgroup(void *pIter, SVgObj **pVgroup) { return sdbFetchRow(tsVgroupSdb, pIter, (void **)pVgroup); } +void mnodeCancelGetNextVgroup(void *pIter) { + sdbFreeIter(tsVgroupSdb, pIter); +} + static int32_t mnodeCreateVgroupFp(SMnodeMsg *pMsg) { SVgObj *pVgroup = pMsg->pVgroup; SDbObj *pDb = pMsg->pDb; assert(pVgroup); - mInfo("app:%p:%p, vgId:%d, is created in mnode, db:%s replica:%d", pMsg->rpcMsg.ahandle, pMsg, pVgroup->vgId, + mInfo("msg:%p, app:%p vgId:%d, is created in mnode, db:%s replica:%d", pMsg, pMsg->rpcMsg.ahandle, pVgroup->vgId, pDb->name, pVgroup->numOfVnodes); for (int32_t i = 0; i < pVgroup->numOfVnodes; ++i) { - mInfo("app:%p:%p, vgId:%d, index:%d, dnode:%d", pMsg->rpcMsg.ahandle, pMsg, pVgroup->vgId, i, + mInfo("msg:%p, app:%p vgId:%d, index:%d, dnode:%d", pMsg, pMsg->rpcMsg.ahandle, pVgroup->vgId, i, pVgroup->vnodeGid[i].dnodeId); } @@ -517,30 +522,30 @@ static int32_t mnodeCreateVgroupCb(SMnodeMsg *pMsg, int32_t code) { assert(pVgroup); if (code != TSDB_CODE_SUCCESS) { - mError("app:%p:%p, vgId:%d, failed to create in sdb, reason:%s", pMsg->rpcMsg.ahandle, pMsg, pVgroup->vgId, + mError("msg:%p, app:%p vgId:%d, failed to create in sdb, reason:%s", pMsg, pMsg->rpcMsg.ahandle, pVgroup->vgId, tstrerror(code)); - SSdbOper desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .table = tsVgroupSdb}; + SSdbRow desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .pTable = tsVgroupSdb}; sdbDeleteRow(&desc); return code; } else { - mInfo("app:%p:%p, vgId:%d, is created in sdb, db:%s replica:%d", pMsg->rpcMsg.ahandle, pMsg, pVgroup->vgId, - pDb->name, pVgroup->numOfVnodes); + mInfo("msg:%p, app:%p vgId:%d, is created in sdb, db:%s replica:%d", pMsg, pMsg->rpcMsg.ahandle, pVgroup->vgId, + pDb->name, pVgroup->numOfVnodes); pVgroup->status = TAOS_VG_STATUS_READY; - SSdbOper desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .table = tsVgroupSdb}; + SSdbRow desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .pTable = tsVgroupSdb}; (void)sdbUpdateRow(&desc); dnodeReprocessMWriteMsg(pMsg); return TSDB_CODE_MND_ACTION_IN_PROGRESS; // if (pVgroup->status == TAOS_VG_STATUS_CREATING || pVgroup->status == TAOS_VG_STATUS_READY) { - // mInfo("app:%p:%p, vgId:%d, is created in sdb, db:%s replica:%d", pMsg->rpcMsg.ahandle, pMsg, pVgroup->vgId, + // mInfo("msg:%p, app:%p vgId:%d, is created in sdb, db:%s replica:%d", pMsg, pMsg->rpcMsg.ahandle, pVgroup->vgId, // pDb->name, pVgroup->numOfVnodes); // pVgroup->status = TAOS_VG_STATUS_READY; - // SSdbOper desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .table = tsVgroupSdb}; + // SSdbRow desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .pTable = tsVgroupSdb}; // (void)sdbUpdateRow(&desc); // dnodeReprocessMWriteMsg(pMsg); // return TSDB_CODE_MND_ACTION_IN_PROGRESS; // } else { - // mError("app:%p:%p, vgId:%d, is created in sdb, db:%s replica:%d, but vgroup is dropping", pMsg->rpcMsg.ahandle, + // mError("msg:%p, app:%p vgId:%d, is created in sdb, db:%s replica:%d, but vgroup is dropping", pMsg->rpcMsg.ahandle, // pMsg, pVgroup->vgId, pDb->name, pVgroup->numOfVnodes); // return TSDB_CODE_MND_VGROUP_NOT_EXIST; // } @@ -571,16 +576,16 @@ int32_t mnodeCreateVgroup(SMnodeMsg *pMsg) { pMsg->pVgroup = pVgroup; mnodeIncVgroupRef(pVgroup); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsVgroupSdb, - .pObj = pVgroup, - .rowSize = sizeof(SVgObj), - .pMsg = pMsg, - .reqFp = mnodeCreateVgroupFp + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup, + .rowSize = sizeof(SVgObj), + .pMsg = pMsg, + .fpReq = mnodeCreateVgroupFp }; - code = sdbInsertRow(&oper); + code = sdbInsertRow(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { pMsg->pVgroup = NULL; mnodeDestroyVgroup(pVgroup); @@ -595,12 +600,12 @@ void mnodeDropVgroup(SVgObj *pVgroup, void *ahandle) { } else { mDebug("vgId:%d, replica:%d is deleting from sdb", pVgroup->vgId, pVgroup->numOfVnodes); mnodeSendDropVgroupMsg(pVgroup, NULL); - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsVgroupSdb, - .pObj = pVgroup + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); } } @@ -770,7 +775,7 @@ static int32_t mnodeRetrieveVgroups(SShowObj *pShow, char *data, int32_t rows, v SDnodeObj * pDnode = pVgroup->vnodeGid[i].pDnode; const char *role = "NULL"; if (pDnode != NULL) { - role = mnodeGetMnodeRoleStr(pVgroup->vnodeGid[i].role); + role = syncRole[pVgroup->vnodeGid[i].role]; } pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows; @@ -957,28 +962,28 @@ static void mnodeProcessCreateVnodeRsp(SRpcMsg *rpcMsg) { if (mnodeMsg->received != mnodeMsg->expected) return; if (mnodeMsg->received == mnodeMsg->successed) { - SSdbOper oper = { + SSdbRow row = { .type = SDB_OPER_GLOBAL, - .table = tsVgroupSdb, + .pTable = tsVgroupSdb, .pObj = pVgroup, .rowSize = sizeof(SVgObj), .pMsg = mnodeMsg, - .writeCb = mnodeCreateVgroupCb + .fpRsp = mnodeCreateVgroupCb }; - int32_t code = sdbInsertRowImp(&oper); + int32_t code = sdbInsertRowToQueue(&row); if (code != TSDB_CODE_SUCCESS && code != TSDB_CODE_MND_ACTION_IN_PROGRESS) { mnodeMsg->pVgroup = NULL; mnodeDestroyVgroup(pVgroup); dnodeSendRpcMWriteRsp(mnodeMsg, code); } } else { - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsVgroupSdb, - .pObj = pVgroup + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); dnodeSendRpcMWriteRsp(mnodeMsg, mnodeMsg->code); } } @@ -1031,12 +1036,12 @@ static void mnodeProcessDropVnodeRsp(SRpcMsg *rpcMsg) { if (mnodeMsg->received != mnodeMsg->expected) return; - SSdbOper oper = { - .type = SDB_OPER_GLOBAL, - .table = tsVgroupSdb, - .pObj = pVgroup + SSdbRow row = { + .type = SDB_OPER_GLOBAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup }; - int32_t code = sdbDeleteRow(&oper); + int32_t code = sdbDeleteRow(&row); if (code != 0) { code = TSDB_CODE_MND_SDB_ERROR; } @@ -1084,19 +1089,17 @@ void mnodeDropAllDnodeVgroups(SDnodeObj *pDropDnode) { if (pVgroup->vnodeGid[0].dnodeId == pDropDnode->dnodeId) { mnodeDropAllChildTablesInVgroups(pVgroup); - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsVgroupSdb, - .pObj = pVgroup, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfVgroups++; } mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); - mInfo("dnode:%d, all vgroups:%d is dropped from sdb", pDropDnode->dnodeId, numOfVgroups); } @@ -1118,8 +1121,6 @@ void mnodeUpdateAllDbVgroups(SDbObj *pAlterDb) { mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); - mInfo("db:%s, all vgroups is updated in sdb", pAlterDb->name); } #endif @@ -1135,20 +1136,18 @@ void mnodeDropAllDbVgroups(SDbObj *pDropDb) { if (pVgroup == NULL) break; if (pVgroup->pDb == pDropDb) { - SSdbOper oper = { - .type = SDB_OPER_LOCAL, - .table = tsVgroupSdb, - .pObj = pVgroup, + SSdbRow row = { + .type = SDB_OPER_LOCAL, + .pTable = tsVgroupSdb, + .pObj = pVgroup, }; - sdbDeleteRow(&oper); + sdbDeleteRow(&row); numOfVgroups++; } mnodeDecVgroupRef(pVgroup); } - sdbFreeIter(pIter); - mInfo("db:%s, all vgroups:%d is dropped from sdb", pDropDb->name, numOfVgroups); } @@ -1170,7 +1169,5 @@ void mnodeSendDropAllDbVgroupsMsg(SDbObj *pDropDb) { numOfVgroups++; } - sdbFreeIter(pIter); - mInfo("db:%s, all vgroups:%d drop msg is sent to dnode", pDropDb->name, numOfVgroups); } diff --git a/src/mnode/src/mnodeWrite.c b/src/mnode/src/mnodeWrite.c index d021745d2b..8893316ffc 100644 --- a/src/mnode/src/mnodeWrite.c +++ b/src/mnode/src/mnodeWrite.c @@ -43,7 +43,7 @@ void mnodeAddWriteMsgHandle(uint8_t msgType, int32_t (*fp)(SMnodeMsg *mnodeMsg)) int32_t mnodeProcessWrite(SMnodeMsg *pMsg) { if (pMsg->rpcMsg.pCont == NULL) { - mError("app:%p:%p, msg:%s content is null", pMsg->rpcMsg.ahandle, pMsg, taosMsg[pMsg->rpcMsg.msgType]); + mError("msg:%p, app:%p type:%s content is null", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_INVALID_MSG_LEN; } @@ -54,15 +54,15 @@ int32_t mnodeProcessWrite(SMnodeMsg *pMsg) { rpcRsp->rsp = epSet; rpcRsp->len = sizeof(SRpcEpSet); - mDebug("app:%p:%p, msg:%s in write queue, will be redirected, numOfEps:%d inUse:%d", pMsg->rpcMsg.ahandle, pMsg, + mDebug("msg:%p, app:%p type:%s in write queue, will be redirected, numOfEps:%d inUse:%d", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], epSet->numOfEps, epSet->inUse); for (int32_t i = 0; i < epSet->numOfEps; ++i) { if (strcmp(epSet->fqdn[i], tsLocalFqdn) == 0 && htons(epSet->port[i]) == tsServerPort) { epSet->inUse = (i + 1) % epSet->numOfEps; - mDebug("app:%p:%p, mnode index:%d ep:%s:%d, set inUse to %d", pMsg->rpcMsg.ahandle, pMsg, i, epSet->fqdn[i], + mDebug("msg:%p, app:%p mnode index:%d ep:%s:%d, set inUse to %d", pMsg, pMsg->rpcMsg.ahandle, i, epSet->fqdn[i], htons(epSet->port[i]), epSet->inUse); } else { - mDebug("app:%p:%p, mnode index:%d ep:%s:%d", pMsg->rpcMsg.ahandle, pMsg, i, epSet->fqdn[i], + mDebug("msg:%p, app:%p mnode index:%d ep:%s:%d", pMsg, pMsg->rpcMsg.ahandle, i, epSet->fqdn[i], htons(epSet->port[i])); } } @@ -71,19 +71,19 @@ int32_t mnodeProcessWrite(SMnodeMsg *pMsg) { } if (tsMnodeProcessWriteMsgFp[pMsg->rpcMsg.msgType] == NULL) { - mError("app:%p:%p, msg:%s not processed", pMsg->rpcMsg.ahandle, pMsg, taosMsg[pMsg->rpcMsg.msgType]); + mError("msg:%p, app:%p type:%s not processed", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_MSG_NOT_PROCESSED; } int32_t code = mnodeInitMsg(pMsg); if (code != TSDB_CODE_SUCCESS) { - mError("app:%p:%p, msg:%s not processed, reason:%s", pMsg->rpcMsg.ahandle, pMsg, taosMsg[pMsg->rpcMsg.msgType], + mError("msg:%p, app:%p type:%s not processed, reason:%s", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], tstrerror(code)); return code; } if (!pMsg->pUser->writeAuth) { - mError("app:%p:%p, msg:%s not processed, no write auth", pMsg->rpcMsg.ahandle, pMsg, + mError("msg:%p, app:%p type:%s not processed, no write auth", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]); return TSDB_CODE_MND_NO_RIGHTS; } diff --git a/src/os/inc/osAlpine.h b/src/os/inc/osAlpine.h index d939adfb6d..eba9459395 100644 --- a/src/os/inc/osAlpine.h +++ b/src/os/inc/osAlpine.h @@ -77,6 +77,7 @@ extern "C" { #include #include #include +#include typedef int(*__compar_fn_t)(const void *, const void *); void error (int, int, const char *); diff --git a/src/os/inc/osArm32.h b/src/os/inc/osArm32.h index 17b4d2dbd5..24ff95522e 100644 --- a/src/os/inc/osArm32.h +++ b/src/os/inc/osArm32.h @@ -76,6 +76,7 @@ extern "C" { #include #include #include +#include #define TAOS_OS_FUNC_LZ4 #define BUILDIN_CLZL(val) __builtin_clzll(val) diff --git a/src/os/inc/osArm64.h b/src/os/inc/osArm64.h index 3ae08b45f4..22f0000e96 100644 --- a/src/os/inc/osArm64.h +++ b/src/os/inc/osArm64.h @@ -77,6 +77,7 @@ extern "C" { #include #include #include +#include #ifdef __cplusplus } diff --git a/src/os/inc/osDarwin.h b/src/os/inc/osDarwin.h index 7bb844831e..1461ec6d3b 100644 --- a/src/os/inc/osDarwin.h +++ b/src/os/inc/osDarwin.h @@ -70,6 +70,7 @@ extern "C" { #include #include #include +#include #define TAOS_OS_FUNC_FILE_SENDIFLE diff --git a/src/os/inc/osFile.h b/src/os/inc/osFile.h index 62e44d8eb0..c9b3b9cd76 100644 --- a/src/os/inc/osFile.h +++ b/src/os/inc/osFile.h @@ -20,17 +20,6 @@ extern "C" { #endif -#define tread(fd, buf, count) read(fd, buf, count) -#define twrite(fd, buf, count) write(fd, buf, count) -#define tlseek(fd, offset, whence) lseek(fd, offset, whence) -#define tclose(fd) \ - { \ - if (FD_VALID(fd)) { \ - close(fd); \ - fd = FD_INITIALIZER; \ - } \ - } - int64_t taosReadImp(int32_t fd, void *buf, int64_t count); int64_t taosWriteImp(int32_t fd, void *buf, int64_t count); int64_t taosLSeekImp(int32_t fd, int64_t offset, int32_t whence); @@ -39,7 +28,13 @@ int32_t taosRenameFile(char *fullPath, char *suffix, char delimiter, char **dstP #define taosRead(fd, buf, count) taosReadImp(fd, buf, count) #define taosWrite(fd, buf, count) taosWriteImp(fd, buf, count) #define taosLSeek(fd, offset, whence) taosLSeekImp(fd, offset, whence) -#define taosClose(x) tclose(x) +#define taosClose(fd) \ + { \ + if (FD_VALID(fd)) { \ + close(fd); \ + fd = FD_INITIALIZER; \ + } \ + } // TAOS_OS_FUNC_FILE_SENDIFLE int64_t taosSendFile(int32_t dfd, int32_t sfd, int64_t *offset, int64_t size); diff --git a/src/os/inc/osLinux32.h b/src/os/inc/osLinux32.h index 93e917e797..cfef05368f 100644 --- a/src/os/inc/osLinux32.h +++ b/src/os/inc/osLinux32.h @@ -76,6 +76,7 @@ extern "C" { #include #include #include +#include #define TAOS_OS_FUNC_LZ4 #define BUILDIN_CLZL(val) __builtin_clzll(val) diff --git a/src/os/inc/osLinux64.h b/src/os/inc/osLinux64.h index c0841c41bd..a2febd51b7 100644 --- a/src/os/inc/osLinux64.h +++ b/src/os/inc/osLinux64.h @@ -79,6 +79,7 @@ extern "C" { #include #endif #include +#include #ifdef __cplusplus } diff --git a/src/os/inc/osWindows.h b/src/os/inc/osWindows.h index 5003e48c44..5a1e642572 100644 --- a/src/os/inc/osWindows.h +++ b/src/os/inc/osWindows.h @@ -40,6 +40,7 @@ #include #include #include +#include #include "msvcProcess.h" #include "msvcDirect.h" #include "msvcFcntl.h" diff --git a/src/os/src/detail/osFile.c b/src/os/src/detail/osFile.c index 23fc88b8e1..2e6886aa21 100644 --- a/src/os/src/detail/osFile.c +++ b/src/os/src/detail/osFile.c @@ -116,7 +116,7 @@ int64_t taosWriteImp(int32_t fd, void *buf, int64_t n) { } int64_t taosLSeekImp(int32_t fd, int64_t offset, int32_t whence) { - return (int64_t)tlseek(fd, (long)offset, whence); + return (int64_t)lseek(fd, (long)offset, whence); } #ifndef TAOS_OS_FUNC_FILE_SENDIFLE diff --git a/src/plugins/http/inc/httpInt.h b/src/plugins/http/inc/httpInt.h index affc0e838e..ebdfabf310 100644 --- a/src/plugins/http/inc/httpInt.h +++ b/src/plugins/http/inc/httpInt.h @@ -118,7 +118,7 @@ typedef struct { typedef struct { char *module; - bool (*decodeFp)(struct HttpContext *pContext); + bool (*fpDecode)(struct HttpContext *pContext); } HttpDecodeMethod; typedef struct { diff --git a/src/plugins/http/src/httpHandle.c b/src/plugins/http/src/httpHandle.c index b50217cfc4..7c56507514 100644 --- a/src/plugins/http/src/httpHandle.c +++ b/src/plugins/http/src/httpHandle.c @@ -21,11 +21,11 @@ #include "httpHandle.h" bool httpDecodeRequest(HttpContext* pContext) { - if (pContext->decodeMethod->decodeFp == NULL) { + if (pContext->decodeMethod->fpDecode == NULL) { return false; } - return (*pContext->decodeMethod->decodeFp)(pContext); + return (*pContext->decodeMethod->fpDecode)(pContext); } /** diff --git a/src/query/inc/qUtil.h b/src/query/inc/qUtil.h index bc23ffadd4..8b84ac0182 100644 --- a/src/query/inc/qUtil.h +++ b/src/query/inc/qUtil.h @@ -27,7 +27,7 @@ int32_t getOutputInterResultBufSize(SQuery* pQuery); void clearResultRow(SQueryRuntimeEnv* pRuntimeEnv, SResultRow* pRow, int16_t type); -void copyResultRow(SQueryRuntimeEnv* pRuntimeEnv, SResultRow* dst, const SResultRow* src); +void copyResultRow(SQueryRuntimeEnv* pRuntimeEnv, SResultRow* dst, const SResultRow* src, int16_t type); SResultRowCellInfo* getResultCell(SQueryRuntimeEnv* pRuntimeEnv, const SResultRow* pRow, int32_t index); int32_t initWindowResInfo(SWindowResInfo* pWindowResInfo, int32_t size, int32_t threshold, int16_t type); diff --git a/src/query/src/qExecutor.c b/src/query/src/qExecutor.c index 7e63edc83c..712b6fa051 100644 --- a/src/query/src/qExecutor.c +++ b/src/query/src/qExecutor.c @@ -6076,9 +6076,9 @@ static int32_t convertQueryMsg(SQueryTableMsg *pQueryMsg, SArray **pTableIdList, pQueryMsg->interval.interval = htobe64(pQueryMsg->interval.interval); pQueryMsg->interval.sliding = htobe64(pQueryMsg->interval.sliding); pQueryMsg->interval.offset = htobe64(pQueryMsg->interval.offset); - pQueryMsg->interval.intervalUnit = pQueryMsg->interval.intervalUnit; - pQueryMsg->interval.slidingUnit = pQueryMsg->interval.slidingUnit; - pQueryMsg->interval.offsetUnit = pQueryMsg->interval.offsetUnit; + // pQueryMsg->interval.intervalUnit = pQueryMsg->interval.intervalUnit; + // pQueryMsg->interval.slidingUnit = pQueryMsg->interval.slidingUnit; + // pQueryMsg->interval.offsetUnit = pQueryMsg->interval.offsetUnit; pQueryMsg->limit = htobe64(pQueryMsg->limit); pQueryMsg->offset = htobe64(pQueryMsg->offset); diff --git a/src/query/src/qResultbuf.c b/src/query/src/qResultbuf.c index 55a7aea53a..bc7243830d 100644 --- a/src/query/src/qResultbuf.c +++ b/src/query/src/qResultbuf.c @@ -423,9 +423,8 @@ void destroyResultBuf(SDiskbasedResultBuf* pResultBuf) { unlink(pResultBuf->path); tfree(pResultBuf->path); - SHashMutableIterator* iter = taosHashCreateIter(pResultBuf->groupSet); - while(taosHashIterNext(iter)) { - SArray** p = (SArray**) taosHashIterGet(iter); + SArray** p = taosHashIterate(pResultBuf->groupSet, NULL); + while(p) { size_t n = taosArrayGetSize(*p); for(int32_t i = 0; i < n; ++i) { SPageInfo* pi = taosArrayGetP(*p, i); @@ -434,10 +433,9 @@ void destroyResultBuf(SDiskbasedResultBuf* pResultBuf) { } taosArrayDestroy(*p); + p = taosHashIterate(pResultBuf->groupSet, p); } - taosHashDestroyIter(iter); - tdListFree(pResultBuf->lruList); taosArrayDestroy(pResultBuf->emptyDummyIdList); taosHashCleanup(pResultBuf->groupSet); diff --git a/src/query/src/qUtil.c b/src/query/src/qUtil.c index d93d02cc0f..2c41d9bfc5 100644 --- a/src/query/src/qUtil.c +++ b/src/query/src/qUtil.c @@ -133,7 +133,7 @@ void clearFirstNWindowRes(SQueryRuntimeEnv *pRuntimeEnv, int32_t num) { // clear all the closed windows from the window list for (int32_t k = 0; k < remain; ++k) { - copyResultRow(pRuntimeEnv, pWindowResInfo->pResult[k], pWindowResInfo->pResult[num + k]); + copyResultRow(pRuntimeEnv, pWindowResInfo->pResult[k], pWindowResInfo->pResult[num + k], type); } // move the unclosed window in the front of the window list @@ -272,9 +272,15 @@ void clearResultRow(SQueryRuntimeEnv *pRuntimeEnv, SResultRow *pWindowRes, int16 * since the attribute of "Pos" is bound to each window result when the window result is created in the * disk-based result buffer. */ -void copyResultRow(SQueryRuntimeEnv *pRuntimeEnv, SResultRow *dst, const SResultRow *src) { +void copyResultRow(SQueryRuntimeEnv *pRuntimeEnv, SResultRow *dst, const SResultRow *src, int16_t type) { dst->numOfRows = src->numOfRows; - dst->win = src->win; + + if (type == TSDB_DATA_TYPE_BINARY || type == TSDB_DATA_TYPE_NCHAR) { + dst->key = realloc(dst->key, varDataTLen(src->key)); + varDataCopy(dst->key, src->key); + } else { + dst->win = src->win; + } dst->closed = src->closed; int32_t nOutputCols = pRuntimeEnv->pQuery->numOfOutput; diff --git a/src/rpc/src/rpcTcp.c b/src/rpc/src/rpcTcp.c index 7b8cf3cda2..2850046d05 100644 --- a/src/rpc/src/rpcTcp.c +++ b/src/rpc/src/rpcTcp.c @@ -371,10 +371,13 @@ void taosCloseTcpConnection(void *chandle) { int taosSendTcpData(uint32_t ip, uint16_t port, void *data, int len, void *chandle) { SFdObj *pFdObj = chandle; - if (pFdObj == NULL || pFdObj->signature != pFdObj) return -1; + SThreadObj *pThreadObj = pFdObj->pThreadObj; - return taosWriteMsg(pFdObj->fd, data, len); + int ret = taosWriteMsg(pFdObj->fd, data, len); + tTrace("%s %p TCP data is sent, FD:%p fd:%d bytes:%d", pThreadObj->label, pFdObj->thandle, pFdObj, pFdObj->fd, ret); + + return ret; } static void taosReportBrokenLink(SFdObj *pFdObj) { @@ -409,7 +412,7 @@ static int taosReadTcpData(SFdObj *pFdObj, SRecvInfo *pInfo) { headLen = taosReadMsg(pFdObj->fd, &rpcHead, sizeof(SRpcHead)); if (headLen != sizeof(SRpcHead)) { - tDebug("%s %p read error, headLen:%d", pThreadObj->label, pFdObj->thandle, headLen); + tDebug("%s %p read error, FD:%p headLen:%d", pThreadObj->label, pFdObj->thandle, pFdObj, headLen); return -1; } @@ -420,7 +423,7 @@ static int taosReadTcpData(SFdObj *pFdObj, SRecvInfo *pInfo) { tError("%s %p TCP malloc(size:%d) fail", pThreadObj->label, pFdObj->thandle, msgLen); return -1; } else { - tTrace("TCP malloc mem:%p size:%d", buffer, size); + tTrace("%s %p read data, FD:%p fd:%d TCP malloc mem:%p", pThreadObj->label, pFdObj->thandle, pFdObj, pFdObj->fd, buffer); } msg = buffer + tsRpcOverhead; @@ -583,8 +586,8 @@ static void taosFreeFdObj(SFdObj *pFdObj) { pthread_mutex_unlock(&pThreadObj->mutex); - tDebug("%s %p TCP connection is closed, FD:%p numOfFds:%d", - pThreadObj->label, pFdObj->thandle, pFdObj, pThreadObj->numOfFds); + tDebug("%s %p TCP connection is closed, FD:%p fd:%d numOfFds:%d", + pThreadObj->label, pFdObj->thandle, pFdObj, pFdObj->fd, pThreadObj->numOfFds); tfree(pFdObj); } diff --git a/src/sync/inc/syncInt.h b/src/sync/inc/syncInt.h index 309d5b1a75..7dc04d70d0 100644 --- a/src/sync/inc/syncInt.h +++ b/src/sync/inc/syncInt.h @@ -27,16 +27,22 @@ extern "C" { #define sDebug(...) { if (sDebugFlag & DEBUG_DEBUG) { taosPrintLog("SYN ", sDebugFlag, __VA_ARGS__); }} #define sTrace(...) { if (sDebugFlag & DEBUG_TRACE) { taosPrintLog("SYN ", sDebugFlag, __VA_ARGS__); }} -#define TAOS_SMSG_SYNC_DATA 1 -#define TAOS_SMSG_FORWARD 2 -#define TAOS_SMSG_FORWARD_RSP 3 -#define TAOS_SMSG_SYNC_REQ 4 -#define TAOS_SMSG_SYNC_RSP 5 -#define TAOS_SMSG_SYNC_MUST 6 -#define TAOS_SMSG_STATUS 7 +typedef enum { + TAOS_SMSG_SYNC_DATA = 1, + TAOS_SMSG_FORWARD = 2, + TAOS_SMSG_FORWARD_RSP = 3, + TAOS_SMSG_SYNC_REQ = 4, + TAOS_SMSG_SYNC_RSP = 5, + TAOS_SMSG_SYNC_MUST = 6, + TAOS_SMSG_STATUS = 7, + TAOS_SMSG_SYNC_DATA_RSP = 8, +} ESyncMsgType; #define SYNC_MAX_SIZE (TSDB_MAX_WAL_SIZE + sizeof(SWalHead) + sizeof(SSyncHead) + 16) #define SYNC_RECV_BUFFER_SIZE (5*1024*1024) +#define SYNC_FWD_TIMER 300 +#define SYNC_ROLE_TIMER 10000 +#define SYNC_WAIT_AFTER_CHOOSE_MASTER 3 #define nodeRole pNode->peerInfo[pNode->selfIndex]->role #define nodeVersion pNode->peerInfo[pNode->selfIndex]->version @@ -60,6 +66,10 @@ typedef struct { int32_t sourceId; // only for arbitrator } SFirstPkt; +typedef struct { + int8_t sync; +} SFirstPktRsp; + typedef struct { int8_t role; uint64_t version; @@ -68,6 +78,9 @@ typedef struct { typedef struct { int8_t role; int8_t ack; + int8_t type; + int8_t reserved[3]; + uint16_t tranId; uint64_t version; SPeerStatus peersStatus[]; } SPeersStatus; @@ -120,12 +133,13 @@ typedef struct SsyncPeer { int32_t nodeId; uint32_t ip; uint16_t port; + int8_t role; + int8_t sstatus; // sync status char fqdn[TSDB_FQDN_LEN]; // peer ip string char id[TSDB_EP_LEN + 32]; // peer vgId + end point - int8_t role; - int8_t sstatus; // sync status uint64_t version; - uint64_t sversion; // track the peer version in retrieve process + uint64_t sversion; // track the peer version in retrieve process + uint64_t lastVer; // track the file version while retrieve int32_t syncFd; int32_t peerFd; // forward FD int32_t numOfRetrieves; // number of retrieves tried @@ -135,7 +149,7 @@ typedef struct SsyncPeer { int32_t notifyFd; int32_t watchNum; int32_t *watchFd; - int8_t refCount; // reference count + int32_t refCount; // reference count struct SSyncNode *pSyncNode; } SSyncPeer; @@ -143,29 +157,30 @@ typedef struct SSyncNode { char path[TSDB_FILENAME_LEN]; int8_t replica; int8_t quorum; + int8_t selfIndex; uint32_t vgId; int64_t rid; - void *ahandle; - int8_t selfIndex; - SSyncPeer *peerInfo[TAOS_SYNC_MAX_REPLICA+1]; // extra one for arbitrator - SSyncPeer *pMaster; - int8_t refCount; + SSyncPeer * peerInfo[TAOS_SYNC_MAX_REPLICA + 1]; // extra one for arbitrator + SSyncPeer * pMaster; SRecvBuffer *pRecv; - SSyncFwds *pSyncFwds; // saved forward info if quorum >1 - void *pFwdTimer; - FGetFileInfo getFileInfo; - FGetWalInfo getWalInfo; - FWriteToCache writeToCache; - FConfirmForward confirmForward; - FNotifyRole notifyRole; - FNotifyFlowCtrl notifyFlowCtrl; + SSyncFwds * pSyncFwds; // saved forward info if quorum >1 + void * pFwdTimer; + void * pRoleTimer; + FGetFileInfo getFileInfo; + FGetWalInfo getWalInfo; + FWriteToCache writeToCache; + FConfirmForward confirmForward; + FNotifyRole notifyRole; + FNotifyFlowCtrl notifyFlowCtrl; FNotifyFileSynced notifyFileSynced; - pthread_mutex_t mutex; + FGetFileVersion getFileVersion; + pthread_mutex_t mutex; } SSyncNode; // sync module global extern int32_t tsSyncNum; extern char tsNodeFqdn[TSDB_FQDN_LEN]; +extern char * syncStatus[]; void *syncRetrieveData(void *param); void *syncRestoreData(void *param); diff --git a/src/sync/src/syncMain.c b/src/sync/src/syncMain.c index 7df7fd37e0..ce85c940ff 100644 --- a/src/sync/src/syncMain.c +++ b/src/sync/src/syncMain.c @@ -13,8 +13,7 @@ * along with this program. If not, see . */ -//#include -//#include +#define _DEFAULT_SOURCE #include "os.h" #include "hash.h" #include "tlog.h" @@ -50,7 +49,7 @@ static int32_t tsSyncRefId = -1; static void syncProcessSyncRequest(char *pMsg, SSyncPeer *pPeer); static void syncRecoverFromMaster(SSyncPeer *pPeer); static void syncCheckPeerConnection(void *param, void *tmrId); -static void syncSendPeersStatusMsgToPeer(SSyncPeer *pPeer, char ack); +static void syncSendPeersStatusMsgToPeer(SSyncPeer *pPeer, char ack, int8_t type, uint16_t tranId); static void syncProcessBrokenLink(void *param); static int32_t syncProcessPeerMsg(void *param, void *buffer); static void syncProcessIncommingConnection(int32_t connFd, uint32_t sourceIp); @@ -59,6 +58,7 @@ static void syncAddArbitrator(SSyncNode *pNode); static void syncFreeNode(void *); static void syncRemoveConfirmedFwdInfo(SSyncNode *pNode); static void syncMonitorFwdInfos(void *param, void *tmrId); +static void syncMonitorNodeRole(void *param, void *tmrId); static void syncProcessFwdAck(SSyncNode *pNode, SFwdInfo *pFwdInfo, int32_t code); static void syncSaveFwdInfo(SSyncNode *pNode, uint64_t version, void *mhandle); static void syncRestartPeer(SSyncPeer *pPeer); @@ -73,8 +73,42 @@ char* syncRole[] = { "master" }; +char *syncStatus[] = { + "init", + "start", + "file", + "cache", + "invalid" +}; + +typedef enum { + SYNC_STATUS_BROADCAST, + SYNC_STATUS_BROADCAST_RSP, + SYNC_STATUS_SETUP_CONN, + SYNC_STATUS_SETUP_CONN_RSP, + SYNC_STATUS_EXCHANGE_DATA, + SYNC_STATUS_EXCHANGE_DATA_RSP, + SYNC_STATUS_CHECK_ROLE, + SYNC_STATUS_CHECK_ROLE_RSP +} ESyncStatusType; + +char *statusType[] = { + "broadcast", + "broadcast-rsp", + "setup-conn", + "setup-conn-rsp", + "exchange-data", + "exchange-data-rsp", + "check-role", + "check-role-rsp" +}; + +uint16_t syncGenTranId() { + return taosRand() & 0XFFFF; +} + int32_t syncInit() { - SPoolInfo info; + SPoolInfo info = {0}; info.numOfThreads = tsSyncTcpThreads; info.serverIp = 0; @@ -98,7 +132,7 @@ int32_t syncInit() { return -1; } - tsVgIdHash = taosHashInit(TSDB_MIN_VNODES, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, true); + tsVgIdHash = taosHashInit(TSDB_MIN_VNODES, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_ENTRY_LOCK); if (tsVgIdHash == NULL) { sError("failed to init tsVgIdHash"); taosTmrCleanUp(tsSyncTmrCtrl); @@ -145,7 +179,7 @@ void syncCleanUp() { int64_t syncStart(const SSyncInfo *pInfo) { const SSyncCfg *pCfg = &pInfo->syncCfg; - SSyncNode *pNode = (SSyncNode *)calloc(sizeof(SSyncNode), 1); + SSyncNode *pNode = calloc(sizeof(SSyncNode), 1); if (pNode == NULL) { sError("no memory to allocate syncNode"); terrno = TAOS_SYSTEM_ERROR(errno); @@ -155,7 +189,6 @@ int64_t syncStart(const SSyncInfo *pInfo) { tstrncpy(pNode->path, pInfo->path, sizeof(pNode->path)); pthread_mutex_init(&pNode->mutex, NULL); - pNode->ahandle = pInfo->ahandle; pNode->getFileInfo = pInfo->getFileInfo; pNode->getWalInfo = pInfo->getWalInfo; pNode->writeToCache = pInfo->writeToCache; @@ -163,6 +196,7 @@ int64_t syncStart(const SSyncInfo *pInfo) { pNode->confirmForward = pInfo->confirmForward; pNode->notifyFlowCtrl = pInfo->notifyFlowCtrl; pNode->notifyFileSynced = pInfo->notifyFileSynced; + pNode->getFileVersion = pInfo->getFileVersion; pNode->selfIndex = -1; pNode->vgId = pInfo->vgId; @@ -176,18 +210,18 @@ int64_t syncStart(const SSyncInfo *pInfo) { return -1; } - for (int32_t i = 0; i < pCfg->replica; ++i) { - const SNodeInfo *pNodeInfo = pCfg->nodeInfo + i; - pNode->peerInfo[i] = syncAddPeer(pNode, pNodeInfo); - if (pNode->peerInfo[i] == NULL) { - sError("vgId:%d, node:%d fqdn:%s port:%u is not configured, stop taosd", pNode->vgId, pNodeInfo->nodeId, pNodeInfo->nodeFqdn, - pNodeInfo->nodePort); + for (int32_t index = 0; index < pCfg->replica; ++index) { + const SNodeInfo *pNodeInfo = pCfg->nodeInfo + index; + pNode->peerInfo[index] = syncAddPeer(pNode, pNodeInfo); + if (pNode->peerInfo[index] == NULL) { + sError("vgId:%d, node:%d fqdn:%s port:%u is not configured, stop taosd", pNode->vgId, pNodeInfo->nodeId, + pNodeInfo->nodeFqdn, pNodeInfo->nodePort); syncStop(pNode->rid); exit(1); } if ((strcmp(pNodeInfo->nodeFqdn, tsNodeFqdn) == 0) && (pNodeInfo->nodePort == tsSyncPort)) { - pNode->selfIndex = i; + pNode->selfIndex = index; } } @@ -211,18 +245,25 @@ int64_t syncStart(const SSyncInfo *pInfo) { return -1; } - pNode->pFwdTimer = taosTmrStart(syncMonitorFwdInfos, 300, (void *)pNode->rid, tsSyncTmrCtrl); + pNode->pFwdTimer = taosTmrStart(syncMonitorFwdInfos, SYNC_FWD_TIMER, (void *)pNode->rid, tsSyncTmrCtrl); if (pNode->pFwdTimer == NULL) { - sError("vgId:%d, failed to allocate timer", pNode->vgId); + sError("vgId:%d, failed to allocate fwd timer", pNode->vgId); + syncStop(pNode->rid); + return -1; + } + + pNode->pRoleTimer = taosTmrStart(syncMonitorNodeRole, SYNC_ROLE_TIMER, (void *)pNode->rid, tsSyncTmrCtrl); + if (pNode->pRoleTimer == NULL) { + sError("vgId:%d, failed to allocate role timer", pNode->vgId); syncStop(pNode->rid); return -1; } syncAddArbitrator(pNode); - taosHashPut(tsVgIdHash, (const char *)&pNode->vgId, sizeof(int32_t), (char *)(&pNode), sizeof(SSyncNode *)); + taosHashPut(tsVgIdHash, &pNode->vgId, sizeof(int32_t), &pNode, sizeof(SSyncNode *)); if (pNode->notifyRole) { - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); } return pNode->rid; @@ -236,20 +277,21 @@ void syncStop(int64_t rid) { sInfo("vgId:%d, cleanup sync", pNode->vgId); - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); - if (tsVgIdHash) taosHashRemove(tsVgIdHash, (const char *)&pNode->vgId, sizeof(int32_t)); + if (tsVgIdHash) taosHashRemove(tsVgIdHash, &pNode->vgId, sizeof(int32_t)); if (pNode->pFwdTimer) taosTmrStop(pNode->pFwdTimer); + if (pNode->pRoleTimer) taosTmrStop(pNode->pRoleTimer); - for (int32_t i = 0; i < pNode->replica; ++i) { - pPeer = pNode->peerInfo[i]; + for (int32_t index = 0; index < pNode->replica; ++index) { + pPeer = pNode->peerInfo[index]; if (pPeer) syncRemovePeer(pPeer); } pPeer = pNode->peerInfo[TAOS_SYNC_MAX_REPLICA]; if (pPeer) syncRemovePeer(pPeer); - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); taosReleaseRef(tsSyncRefId, rid); taosRemoveRef(tsSyncRefId, rid); @@ -264,7 +306,7 @@ int32_t syncReconfig(int64_t rid, const SSyncCfg *pNewCfg) { sInfo("vgId:%d, reconfig, role:%s replica:%d old:%d", pNode->vgId, syncRole[nodeRole], pNewCfg->replica, pNode->replica); - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); for (i = 0; i < pNode->replica; ++i) { for (j = 0; j < pNewCfg->replica; ++j) { @@ -314,17 +356,15 @@ int32_t syncReconfig(int64_t rid, const SSyncCfg *pNewCfg) { if (pNewCfg->replica <= 1) { sInfo("vgId:%d, no peers are configured, work as master!", pNode->vgId); nodeRole = TAOS_SYNC_ROLE_MASTER; - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); } - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); - sInfo("vgId:%d, %d replicas are configured, quorum:%d role:%s", pNode->vgId, pNode->replica, pNode->quorum, - syncRole[nodeRole]); + sInfo("vgId:%d, %d replicas are configured, quorum:%d", pNode->vgId, pNode->replica, pNode->quorum); syncBroadcastStatus(pNode); taosReleaseRef(tsSyncRefId, rid); - return 0; } @@ -356,10 +396,10 @@ void syncConfirmForward(int64_t rid, uint64_t version, int32_t code) { pFwdRsp->code = code; int32_t msgLen = sizeof(SSyncHead) + sizeof(SFwdRsp); - int32_t retLen = write(pPeer->peerFd, msg, msgLen); + int32_t retLen = taosWriteMsg(pPeer->peerFd, msg, msgLen); if (retLen == msgLen) { - sDebug("%s, forward-rsp is sent, code:%x hver:%" PRIu64, pPeer->id, code, version); + sTrace("%s, forward-rsp is sent, code:%x hver:%" PRIu64, pPeer->id, code, version); } else { sDebug("%s, failed to send forward ack, restart", pPeer->id); syncRestartConnection(pPeer); @@ -369,6 +409,7 @@ void syncConfirmForward(int64_t rid, uint64_t version, int32_t code) { taosReleaseRef(tsSyncRefId, rid); } +#if 0 void syncRecover(int64_t rid) { SSyncPeer *pPeer; @@ -379,22 +420,23 @@ void syncRecover(int64_t rid) { // if take this node to unsync state, the whole system may not work nodeRole = TAOS_SYNC_ROLE_UNSYNCED; - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); nodeVersion = 0; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); for (int32_t i = 0; i < pNode->replica; ++i) { - pPeer = (SSyncPeer *)pNode->peerInfo[i]; + pPeer = pNode->peerInfo[i]; if (pPeer->peerFd >= 0) { syncRestartConnection(pPeer); } } - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); taosReleaseRef(tsSyncRefId, rid); } +#endif int32_t syncGetNodesRole(int64_t rid, SNodesRole *pNodesRole) { SSyncNode *pNode = taosAcquireRef(tsSyncRefId, rid); @@ -449,10 +491,10 @@ static void syncFreeNode(void *param) { tfree(pNode); } -void syncAddPeerRef(SSyncPeer *pPeer) { atomic_add_fetch_8(&pPeer->refCount, 1); } +void syncAddPeerRef(SSyncPeer *pPeer) { atomic_add_fetch_32(&pPeer->refCount, 1); } int32_t syncDecPeerRef(SSyncPeer *pPeer) { - if (atomic_sub_fetch_8(&pPeer->refCount, 1) == 0) { + if (atomic_sub_fetch_32(&pPeer->refCount, 1) == 0) { taosReleaseRef(tsSyncRefId, pPeer->pSyncNode->rid); sDebug("%s, resource is freed", pPeer->id); @@ -465,6 +507,8 @@ int32_t syncDecPeerRef(SSyncPeer *pPeer) { } static void syncClosePeerConn(SSyncPeer *pPeer) { + sDebug("%s, pfd:%d sfd:%d will be closed", pPeer->id, pPeer->peerFd, pPeer->syncFd); + taosTmrStopA(&pPeer->timer); taosClose(pPeer->syncFd); if (pPeer->peerFd >= 0) { @@ -497,7 +541,7 @@ static SSyncPeer *syncAddPeer(SSyncNode *pNode, const SNodeInfo *pInfo) { pPeer->ip = ip; pPeer->port = pInfo->nodePort; pPeer->fqdn[sizeof(pPeer->fqdn) - 1] = 0; - snprintf(pPeer->id, sizeof(pPeer->id), "vgId:%d, peer:%s:%u", pNode->vgId, pPeer->fqdn, pPeer->port); + snprintf(pPeer->id, sizeof(pPeer->id), "vgId:%d, nodeId:%d", pNode->vgId, pPeer->nodeId); pPeer->peerFd = -1; pPeer->syncFd = -1; @@ -510,7 +554,7 @@ static SSyncPeer *syncAddPeer(SSyncNode *pNode, const SNodeInfo *pInfo) { if (pPeer->nodeId == 0 || (ret > 0) || (ret == 0 && pPeer->port > tsSyncPort)) { int32_t checkMs = 100 + (pNode->vgId * 10) % 100; if (pNode->vgId > 1) checkMs = tsStatusInterval * 1000 + checkMs; - sDebug("%s, start to check peer connection after %d ms", pPeer->id, checkMs); + sDebug("%s, check peer connection after %d ms", pPeer->id, checkMs); taosTmrReset(syncCheckPeerConnection, checkMs, pPeer, tsSyncTmrCtrl, &pPeer->timer); } @@ -519,22 +563,20 @@ static SSyncPeer *syncAddPeer(SSyncNode *pNode, const SNodeInfo *pInfo) { } void syncBroadcastStatus(SSyncNode *pNode) { - SSyncPeer *pPeer; - - for (int32_t i = 0; i < pNode->replica; ++i) { - if (i == pNode->selfIndex) continue; - pPeer = pNode->peerInfo[i]; - syncSendPeersStatusMsgToPeer(pPeer, 1); + for (int32_t index = 0; index < pNode->replica; ++index) { + if (index == pNode->selfIndex) continue; + SSyncPeer *pPeer = pNode->peerInfo[index]; + syncSendPeersStatusMsgToPeer(pPeer, 1, SYNC_STATUS_BROADCAST, syncGenTranId()); } } static void syncResetFlowCtrl(SSyncNode *pNode) { - for (int32_t i = 0; i < pNode->replica; ++i) { - pNode->peerInfo[i]->numOfRetrieves = 0; + for (int32_t index = 0; index < pNode->replica; ++index) { + pNode->peerInfo[index]->numOfRetrieves = 0; } if (pNode->notifyFlowCtrl) { - (*pNode->notifyFlowCtrl)(pNode->ahandle, 0); + (*pNode->notifyFlowCtrl)(pNode->vgId, 0); } } @@ -544,8 +586,6 @@ static void syncChooseMaster(SSyncNode *pNode) { int32_t index = -1; int32_t replica = pNode->replica; - sDebug("vgId:%d, choose master", pNode->vgId); - for (int32_t i = 0; i < pNode->replica; ++i) { if (pNode->peerInfo[i]->role != TAOS_SYNC_ROLE_OFFLINE) { onlineNum++; @@ -587,19 +627,11 @@ static void syncChooseMaster(SSyncNode *pNode) { sInfo("vgId:%d, start to work as master", pNode->vgId); nodeRole = TAOS_SYNC_ROLE_MASTER; -#if 0 - for (int32_t i = 0; i < pNode->replica; ++i) { - if (i == index) continue; - pPeer = pNode->peerInfo[i]; - if (pPeer->version == nodeVersion) { - pPeer->role = TAOS_SYNC_ROLE_SLAVE; - pPeer->sstatus = TAOS_SYNC_STATUS_CACHE; - sInfo("%s, it shall work as slave", pPeer->id); - } - } -#endif + // Wait for other nodes to receive status to avoid version inconsistency + taosMsleep(SYNC_WAIT_AFTER_CHOOSE_MASTER); + syncResetFlowCtrl(pNode); - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); } else { pPeer = pNode->peerInfo[index]; sInfo("%s, it shall work as master", pPeer->id); @@ -611,11 +643,11 @@ static void syncChooseMaster(SSyncNode *pNode) { static SSyncPeer *syncCheckMaster(SSyncNode *pNode) { int32_t onlineNum = 0; - int32_t index = -1; + int32_t masterIndex = -1; int32_t replica = pNode->replica; - for (int32_t i = 0; i < pNode->replica; ++i) { - if (pNode->peerInfo[i]->role != TAOS_SYNC_ROLE_OFFLINE) { + for (int32_t index = 0; index < pNode->replica; ++index) { + if (pNode->peerInfo[index]->role != TAOS_SYNC_ROLE_OFFLINE) { onlineNum++; } } @@ -630,97 +662,99 @@ static SSyncPeer *syncCheckMaster(SSyncNode *pNode) { if (onlineNum <= replica * 0.5) { if (nodeRole != TAOS_SYNC_ROLE_UNSYNCED) { nodeRole = TAOS_SYNC_ROLE_UNSYNCED; - // pNode->peerInfo[pNode->selfIndex]->role = nodeRole; - (*pNode->notifyRole)(pNode->ahandle, nodeRole); - sInfo("vgId:%d, change to unsynced state, online:%d replica:%d", pNode->vgId, onlineNum, replica); + (*pNode->notifyRole)(pNode->vgId, nodeRole); + sInfo("vgId:%d, self change to unsynced state, online:%d replica:%d", pNode->vgId, onlineNum, replica); } } else { - for (int32_t i = 0; i < pNode->replica; ++i) { - SSyncPeer *pTemp = pNode->peerInfo[i]; + for (int32_t index = 0; index < pNode->replica; ++index) { + SSyncPeer *pTemp = pNode->peerInfo[index]; if (pTemp->role != TAOS_SYNC_ROLE_MASTER) continue; - if (index < 0) { - index = i; + if (masterIndex < 0) { + masterIndex = index; } else { // multiple masters, it shall not happen - if (i == pNode->selfIndex) { + if (masterIndex == pNode->selfIndex) { sError("%s, peer is master, work as slave instead", pTemp->id); nodeRole = TAOS_SYNC_ROLE_SLAVE; - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); } } } } - SSyncPeer *pMaster = (index >= 0) ? pNode->peerInfo[index] : NULL; + SSyncPeer *pMaster = (masterIndex >= 0) ? pNode->peerInfo[masterIndex] : NULL; return pMaster; } static int32_t syncValidateMaster(SSyncPeer *pPeer) { SSyncNode *pNode = pPeer->pSyncNode; - int32_t code = 0; + int32_t code = 0; if (nodeRole == TAOS_SYNC_ROLE_MASTER && nodeVersion < pPeer->version) { - sDebug("%s, slave has higher version, restart all connections!!!", pPeer->id); + sDebug("%s, peer has higher sver:%" PRIu64 ", restart all peer connections", pPeer->id, pPeer->version); nodeRole = TAOS_SYNC_ROLE_UNSYNCED; - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); code = -1; - for (int32_t i = 0; i < pNode->replica; ++i) { - if (i == pNode->selfIndex) continue; - syncRestartPeer(pNode->peerInfo[i]); + for (int32_t index = 0; index < pNode->replica; ++index) { + if (index == pNode->selfIndex) continue; + syncRestartPeer(pNode->peerInfo[index]); } } return code; } -static void syncCheckRole(SSyncPeer *pPeer, SPeerStatus peersStatus[], int8_t newRole) { +static void syncCheckRole(SSyncPeer *pPeer, SPeerStatus* peersStatus, int8_t newPeerRole) { SSyncNode *pNode = pPeer->pSyncNode; - int8_t peerOldRole = pPeer->role; - int8_t selfOldRole = nodeRole; - int8_t i, syncRequired = 0; + int8_t oldPeerRole = pPeer->role; + int8_t oldSelfRole = nodeRole; + int8_t syncRequired = 0; - // pNode->peerInfo[pNode->selfIndex]->version = nodeVersion; - pPeer->role = newRole; - - sDebug("%s, own role:%s, new peer role:%s", pPeer->id, syncRole[nodeRole], syncRole[pPeer->role]); + pPeer->role = newPeerRole; + sDebug("%s, peer role:%s change to %s", pPeer->id, syncRole[oldPeerRole], syncRole[newPeerRole]); SSyncPeer *pMaster = syncCheckMaster(pNode); if (pMaster) { // master is there pNode->pMaster = pMaster; - sDebug("%s, it is the master, ver:%" PRIu64, pMaster->id, pMaster->version); + sDebug("%s, it is the master, sver:%" PRIu64, pMaster->id, pMaster->version); if (syncValidateMaster(pPeer) < 0) return; if (nodeRole == TAOS_SYNC_ROLE_UNSYNCED) { if (nodeVersion < pMaster->version) { + sDebug("%s, is master, sync required, self sver:%" PRIu64, pMaster->id, nodeVersion); syncRequired = 1; } else { - sInfo("%s is master, work as slave, ver:%" PRIu64, pMaster->id, pMaster->version); + sInfo("%s, is master, work as slave, self sver:%" PRIu64, pMaster->id, nodeVersion); nodeRole = TAOS_SYNC_ROLE_SLAVE; - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); } } else if (nodeRole == TAOS_SYNC_ROLE_SLAVE && pMaster == pPeer) { - // nodeVersion = pMaster->version; + sDebug("%s, is master, continue work as slave, self sver:%" PRIu64, pMaster->id, nodeVersion); } } else { // master not there, if all peer's state and version are consistent, choose the master int32_t consistent = 0; - if (peersStatus) { - for (i = 0; i < pNode->replica; ++i) { - SSyncPeer *pTemp = pNode->peerInfo[i]; - if (pTemp->role != peersStatus[i].role) break; - if ((pTemp->role != TAOS_SYNC_ROLE_OFFLINE) && (pTemp->version != peersStatus[i].version)) break; + int32_t index = 0; + if (peersStatus != NULL) { + for (index = 0; index < pNode->replica; ++index) { + SSyncPeer *pTemp = pNode->peerInfo[index]; + if (pTemp->role != peersStatus[index].role) break; + if ((pTemp->role != TAOS_SYNC_ROLE_OFFLINE) && (pTemp->version != peersStatus[index].version)) break; } - if (i >= pNode->replica) consistent = 1; + if (index >= pNode->replica) consistent = 1; } else { if (pNode->replica == 2) consistent = 1; } if (consistent) { + sDebug("vgId:%d, choose master", pNode->vgId); syncChooseMaster(pNode); + } else { + sDebug("vgId:%d, cannot choose master since roles inequality", pNode->vgId); } } @@ -728,7 +762,8 @@ static void syncCheckRole(SSyncPeer *pPeer, SPeerStatus peersStatus[], int8_t ne syncRecoverFromMaster(pMaster); } - if (peerOldRole != newRole || nodeRole != selfOldRole) { + if (oldPeerRole != newPeerRole || nodeRole != oldSelfRole) { + sDebug("vgId:%d, roles changed, broadcast status", pNode->vgId); syncBroadcastStatus(pNode); } @@ -738,14 +773,16 @@ static void syncCheckRole(SSyncPeer *pPeer, SPeerStatus peersStatus[], int8_t ne } static void syncRestartPeer(SSyncPeer *pPeer) { - sDebug("%s, restart connection", pPeer->id); + sDebug("%s, restart peer connection, last sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); syncClosePeerConn(pPeer); pPeer->sstatus = TAOS_SYNC_STATUS_INIT; + sDebug("%s, peer conn is restart and set sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); int32_t ret = strcmp(pPeer->fqdn, tsNodeFqdn); if (ret > 0 || (ret == 0 && pPeer->port > tsSyncPort)) { + sDebug("%s, check peer connection in 1000 ms", pPeer->id); taosTmrReset(syncCheckPeerConnection, tsSyncTimer * 1000, pPeer, tsSyncTmrCtrl, &pPeer->timer); } } @@ -770,7 +807,7 @@ static void syncProcessSyncRequest(char *msg, SSyncPeer *pPeer) { } if (pPeer->sstatus != TAOS_SYNC_STATUS_INIT) { - sDebug("%s, sync is already started", pPeer->id); + sDebug("%s, sync is already started for sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); return; // already started } @@ -784,11 +821,11 @@ static void syncProcessSyncRequest(char *msg, SSyncPeer *pPeer) { pthread_attr_destroy(&thattr); if (ret != 0) { - sError("%s, failed to create sync thread(%s)", pPeer->id, strerror(errno)); + sError("%s, failed to create sync thread since %s", pPeer->id, strerror(errno)); syncDecPeerRef(pPeer); } else { pPeer->sstatus = TAOS_SYNC_STATUS_START; - sDebug("%s, thread is created to retrieve data", pPeer->id); + sDebug("%s, thread is created to retrieve data, set sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); } } @@ -796,27 +833,28 @@ static void syncNotStarted(void *param, void *tmrId) { SSyncPeer *pPeer = param; SSyncNode *pNode = pPeer->pSyncNode; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); pPeer->timer = NULL; - sInfo("%s, sync connection is still not up, restart", pPeer->id); + pPeer->sstatus = TAOS_SYNC_STATUS_INIT; + sInfo("%s, sync conn is still not up, restart and set sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); syncRestartConnection(pPeer); - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); } static void syncTryRecoverFromMaster(void *param, void *tmrId) { SSyncPeer *pPeer = param; SSyncNode *pNode = pPeer->pSyncNode; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); syncRecoverFromMaster(pPeer); - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); } static void syncRecoverFromMaster(SSyncPeer *pPeer) { SSyncNode *pNode = pPeer->pSyncNode; if (nodeSStatus != TAOS_SYNC_STATUS_INIT) { - sDebug("%s, sync is already started, status:%d", pPeer->id, nodeSStatus); + sDebug("%s, sync is already started since sstatus:%s", pPeer->id, syncStatus[nodeSStatus]); return; } @@ -840,11 +878,11 @@ static void syncRecoverFromMaster(SSyncPeer *pPeer) { firstPkt.port = tsSyncPort; taosTmrReset(syncNotStarted, tsSyncTimer * 1000, pPeer, tsSyncTmrCtrl, &pPeer->timer); - if (write(pPeer->peerFd, &firstPkt, sizeof(firstPkt)) != sizeof(firstPkt)) { + if (taosWriteMsg(pPeer->peerFd, &firstPkt, sizeof(firstPkt)) != sizeof(firstPkt)) { sError("%s, failed to send sync-req to peer", pPeer->id); } else { nodeSStatus = TAOS_SYNC_STATUS_START; - sInfo("%s, sync-req is sent", pPeer->id); + sInfo("%s, sync-req is sent to peer, set sstatus:%s", pPeer->id, syncStatus[nodeSStatus]); } } @@ -854,7 +892,7 @@ static void syncProcessFwdResponse(char *cont, SSyncPeer *pPeer) { SSyncFwds *pSyncFwds = pNode->pSyncFwds; SFwdInfo * pFwdInfo; - sDebug("%s, forward-rsp is received, code:%x ver:%" PRIu64, pPeer->id, pFwdRsp->code, pFwdRsp->version); + sTrace("%s, forward-rsp is received, code:%x hver:%" PRIu64, pPeer->id, pFwdRsp->code, pFwdRsp->version); SFwdInfo *pFirst = pSyncFwds->fwdInfo + pSyncFwds->first; if (pFirst->version <= pFwdRsp->version && pSyncFwds->fwds > 0) { @@ -873,16 +911,17 @@ static void syncProcessForwardFromPeer(char *cont, SSyncPeer *pPeer) { SSyncNode *pNode = pPeer->pSyncNode; SWalHead * pHead = (SWalHead *)cont; - sDebug("%s, forward is received, hver:%" PRIu64 ", len:%d", pPeer->id, pHead->version, pHead->len); + sTrace("%s, forward is received, hver:%" PRIu64 ", len:%d", pPeer->id, pHead->version, pHead->len); if (nodeRole == TAOS_SYNC_ROLE_SLAVE) { // nodeVersion = pHead->version; - (*pNode->writeToCache)(pNode->ahandle, pHead, TAOS_QTYPE_FWD, NULL); + (*pNode->writeToCache)(pNode->vgId, pHead, TAOS_QTYPE_FWD, NULL); } else { if (nodeSStatus != TAOS_SYNC_STATUS_INIT) { syncSaveIntoBuffer(pPeer, pHead); } else { - sError("%s, forward discarded, hver:%" PRIu64, pPeer->id, pHead->version); + sError("%s, forward discarded since sstatus:%s, hver:%" PRIu64, pPeer->id, syncStatus[nodeSStatus], + pHead->version); } } } @@ -891,14 +930,15 @@ static void syncProcessPeersStatusMsg(char *cont, SSyncPeer *pPeer) { SSyncNode * pNode = pPeer->pSyncNode; SPeersStatus *pPeersStatus = (SPeersStatus *)cont; - sDebug("%s, status msg is received, self:%s ver:%" PRIu64 " peer:%s ver:%" PRIu64 ", ack:%d", pPeer->id, - syncRole[nodeRole], nodeVersion, syncRole[pPeersStatus->role], pPeersStatus->version, pPeersStatus->ack); + sDebug("%s, status is received, self:%s:%s:%" PRIu64 ", peer:%s:%" PRIu64 ", ack:%d tranId:%u type:%s pfd:%d", + pPeer->id, syncRole[nodeRole], syncStatus[nodeSStatus], nodeVersion, syncRole[pPeersStatus->role], + pPeersStatus->version, pPeersStatus->ack, pPeersStatus->tranId, statusType[pPeersStatus->type], pPeer->peerFd); pPeer->version = pPeersStatus->version; syncCheckRole(pPeer, pPeersStatus->peersStatus, pPeersStatus->role); if (pPeersStatus->ack) { - syncSendPeersStatusMsgToPeer(pPeer, 0); + syncSendPeersStatusMsgToPeer(pPeer, 0, pPeersStatus->type + 1, pPeersStatus->tranId); } } @@ -933,7 +973,7 @@ static int32_t syncProcessPeerMsg(void *param, void *buffer) { char * cont = buffer; SSyncNode *pNode = pPeer->pSyncNode; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); int32_t code = syncReadPeerMsg(pPeer, &head, cont); @@ -949,14 +989,14 @@ static int32_t syncProcessPeerMsg(void *param, void *buffer) { } } - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); return code; } #define statusMsgLen sizeof(SSyncHead) + sizeof(SPeersStatus) + sizeof(SPeerStatus) * TAOS_SYNC_MAX_REPLICA -static void syncSendPeersStatusMsgToPeer(SSyncPeer *pPeer, char ack) { +static void syncSendPeersStatusMsgToPeer(SSyncPeer *pPeer, char ack, int8_t type, uint16_t tranId) { SSyncNode *pNode = pPeer->pSyncNode; char msg[statusMsgLen] = {0}; @@ -971,22 +1011,24 @@ static void syncSendPeersStatusMsgToPeer(SSyncPeer *pPeer, char ack) { pPeersStatus->version = nodeVersion; pPeersStatus->role = nodeRole; pPeersStatus->ack = ack; + pPeersStatus->type = type; + pPeersStatus->tranId = tranId; for (int32_t i = 0; i < pNode->replica; ++i) { pPeersStatus->peersStatus[i].role = pNode->peerInfo[i]->role; pPeersStatus->peersStatus[i].version = pNode->peerInfo[i]->version; } - int32_t retLen = write(pPeer->peerFd, msg, statusMsgLen); + int32_t retLen = taosWriteMsg(pPeer->peerFd, msg, statusMsgLen); if (retLen == statusMsgLen) { - sDebug("%s, status msg is sent, self:%s ver:%" PRIu64 ", ack:%d", pPeer->id, syncRole[pPeersStatus->role], - pPeersStatus->version, pPeersStatus->ack); + sDebug("%s, status is sent, self:%s:%s:%" PRIu64 ", peer:%s:%s:%" PRIu64 ", ack:%d tranId:%u type:%s pfd:%d", + pPeer->id, syncRole[nodeRole], syncStatus[nodeSStatus], nodeVersion, syncRole[pPeer->role], + syncStatus[pPeer->sstatus], pPeer->version, pPeersStatus->ack, pPeersStatus->tranId, + statusType[pPeersStatus->type], pPeer->peerFd); } else { sDebug("%s, failed to send status msg, restart", pPeer->id); syncRestartConnection(pPeer); } - - return; } static void syncSetupPeerConnection(SSyncPeer *pPeer) { @@ -995,13 +1037,13 @@ static void syncSetupPeerConnection(SSyncPeer *pPeer) { taosTmrStopA(&pPeer->timer); if (pPeer->peerFd >= 0) { sDebug("%s, send role version to peer", pPeer->id); - syncSendPeersStatusMsgToPeer(pPeer, 1); + syncSendPeersStatusMsgToPeer(pPeer, 1, SYNC_STATUS_SETUP_CONN, syncGenTranId()); return; } int32_t connFd = taosOpenTcpClientSocket(pPeer->ip, pPeer->port, 0); if (connFd < 0) { - sDebug("%s, failed to open tcp socket(%s)", pPeer->id, strerror(errno)); + sDebug("%s, failed to open tcp socket since %s", pPeer->id, strerror(errno)); taosTmrReset(syncCheckPeerConnection, tsSyncTimer * 1000, pPeer, tsSyncTmrCtrl, &pPeer->timer); return; } @@ -1014,15 +1056,15 @@ static void syncSetupPeerConnection(SSyncPeer *pPeer) { firstPkt.port = tsSyncPort; firstPkt.sourceId = pNode->vgId; // tell arbitrator its vgId - if (write(connFd, &firstPkt, sizeof(firstPkt)) == sizeof(firstPkt)) { - sDebug("%s, connection to peer server is setup", pPeer->id); + if (taosWriteMsg(connFd, &firstPkt, sizeof(firstPkt)) == sizeof(firstPkt)) { + sDebug("%s, connection to peer server is setup, pfd:%d sfd:%d", pPeer->id, connFd, pPeer->syncFd); pPeer->peerFd = connFd; pPeer->role = TAOS_SYNC_ROLE_UNSYNCED; pPeer->pConn = taosAllocateTcpConn(tsTcpPool, pPeer, connFd); syncAddPeerRef(pPeer); } else { - sDebug("try later"); - close(connFd); + sDebug("%s, failed to setup peer connection to server since %s, try later", pPeer->id, strerror(errno)); + taosClose(connFd); taosTmrReset(syncCheckPeerConnection, tsSyncTimer * 1000, pPeer, tsSyncTmrCtrl, &pPeer->timer); } } @@ -1031,12 +1073,12 @@ static void syncCheckPeerConnection(void *param, void *tmrId) { SSyncPeer *pPeer = param; SSyncNode *pNode = pPeer->pSyncNode; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); sDebug("%s, check peer connection", pPeer->id); syncSetupPeerConnection(pPeer); - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); } static void syncCreateRestoreDataThread(SSyncPeer *pPeer) { @@ -1069,13 +1111,13 @@ static void syncProcessIncommingConnection(int32_t connFd, uint32_t sourceIp) { SFirstPkt firstPkt; if (taosReadMsg(connFd, &firstPkt, sizeof(firstPkt)) != sizeof(firstPkt)) { - sError("failed to read peer first pkt from ip:%s(%s)", ipstr, strerror(errno)); + sError("failed to read peer first pkt from ip:%s since %s", ipstr, strerror(errno)); taosCloseSocket(connFd); return; } int32_t vgId = firstPkt.syncHead.vgId; - SSyncNode **ppNode = (SSyncNode **)taosHashGet(tsVgIdHash, (const char *)&vgId, sizeof(int32_t)); + SSyncNode **ppNode = taosHashGet(tsVgIdHash, &vgId, sizeof(int32_t)); if (ppNode == NULL || *ppNode == NULL) { sError("vgId:%d, vgId could not be found", vgId); taosCloseSocket(connFd); @@ -1083,7 +1125,7 @@ static void syncProcessIncommingConnection(int32_t connFd, uint32_t sourceIp) { } SSyncNode *pNode = *ppNode; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); SSyncPeer *pPeer; for (i = 0; i < pNode->replica; ++i) { @@ -1102,17 +1144,17 @@ static void syncProcessIncommingConnection(int32_t connFd, uint32_t sourceIp) { pPeer->syncFd = connFd; syncCreateRestoreDataThread(pPeer); } else { - sDebug("%s, TCP connection is already up, close one", pPeer->id); + sDebug("%s, TCP connection is up, pfd:%d sfd:%d, old pfd:%d", pPeer->id, connFd, pPeer->syncFd, pPeer->peerFd); syncClosePeerConn(pPeer); pPeer->peerFd = connFd; pPeer->pConn = taosAllocateTcpConn(tsTcpPool, pPeer, connFd); syncAddPeerRef(pPeer); sDebug("%s, ready to exchange data", pPeer->id); - syncSendPeersStatusMsgToPeer(pPeer, 1); + syncSendPeersStatusMsgToPeer(pPeer, 1, SYNC_STATUS_EXCHANGE_DATA, syncGenTranId()); } } - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); } static void syncProcessBrokenLink(void *param) { @@ -1121,22 +1163,22 @@ static void syncProcessBrokenLink(void *param) { SSyncNode *pNode = pPeer->pSyncNode; if (taosAcquireRef(tsSyncRefId, pNode->rid) == NULL) return; - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); - sDebug("%s, TCP link is broken(%s)", pPeer->id, strerror(errno)); + sDebug("%s, TCP link is broken since %s, pfd:%d sfd:%d", pPeer->id, strerror(errno), pPeer->peerFd, pPeer->syncFd); pPeer->peerFd = -1; if (syncDecPeerRef(pPeer) != 0) { syncRestartConnection(pPeer); } - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); taosReleaseRef(tsSyncRefId, pNode->rid); } static void syncSaveFwdInfo(SSyncNode *pNode, uint64_t version, void *mhandle) { SSyncFwds *pSyncFwds = pNode->pSyncFwds; - uint64_t time = taosGetTimestampMs(); + int64_t time = taosGetTimestampMs(); if (pSyncFwds->fwds >= tsMaxFwdInfo) { pSyncFwds->first = (pSyncFwds->first + 1) % tsMaxFwdInfo; @@ -1154,7 +1196,7 @@ static void syncSaveFwdInfo(SSyncNode *pNode, uint64_t version, void *mhandle) { pFwdInfo->time = time; pSyncFwds->fwds++; - sDebug("vgId:%d, fwd info is saved, ver:%" PRIu64 " fwds:%d ", pNode->vgId, version, pSyncFwds->fwds); + sTrace("vgId:%d, fwd info is saved, hver:%" PRIu64 " fwds:%d ", pNode->vgId, version, pSyncFwds->fwds); } static void syncRemoveConfirmedFwdInfo(SSyncNode *pNode) { @@ -1168,7 +1210,7 @@ static void syncRemoveConfirmedFwdInfo(SSyncNode *pNode) { pSyncFwds->first = (pSyncFwds->first + 1) % tsMaxFwdInfo; pSyncFwds->fwds--; if (pSyncFwds->fwds == 0) pSyncFwds->first = pSyncFwds->last; - // sDebug("vgId:%d, fwd info is removed, ver:%d, fwds:%d", + // sDebug("vgId:%d, fwd info is removed, hver:%d, fwds:%d", // pNode->vgId, pFwdInfo->version, pSyncFwds->fwds); memset(pFwdInfo, 0, sizeof(SFwdInfo)); } @@ -1191,14 +1233,36 @@ static void syncProcessFwdAck(SSyncNode *pNode, SFwdInfo *pFwdInfo, int32_t code } if (confirm && pFwdInfo->confirmed == 0) { - sDebug("vgId:%d, forward is confirmed, ver:%" PRIu64 " code:%x", pNode->vgId, pFwdInfo->version, pFwdInfo->code); - (*pNode->confirmForward)(pNode->ahandle, pFwdInfo->mhandle, pFwdInfo->code); + sTrace("vgId:%d, forward is confirmed, hver:%" PRIu64 " code:%x", pNode->vgId, pFwdInfo->version, pFwdInfo->code); + (*pNode->confirmForward)(pNode->vgId, pFwdInfo->mhandle, pFwdInfo->code); pFwdInfo->confirmed = 1; } } +static void syncMonitorNodeRole(void *param, void *tmrId) { + int64_t rid = (int64_t)param; + SSyncNode *pNode = taosAcquireRef(tsSyncRefId, rid); + if (pNode == NULL) return; + + for (int32_t index = 0; index < pNode->replica; index++) { + if (index == pNode->selfIndex) continue; + + SSyncPeer *pPeer = pNode->peerInfo[index]; + if (/*pPeer->role > TAOS_SYNC_ROLE_UNSYNCED && */ nodeRole > TAOS_SYNC_ROLE_UNSYNCED) continue; + if (/*pPeer->sstatus > TAOS_SYNC_STATUS_INIT || */ nodeSStatus > TAOS_SYNC_STATUS_INIT) continue; + + sDebug("%s, check roles since self:%s sstatus:%s, peer:%s sstatus:%s", pPeer->id, syncRole[pPeer->role], + syncStatus[pPeer->sstatus], syncRole[nodeRole], syncStatus[nodeSStatus]); + syncSendPeersStatusMsgToPeer(pPeer, 1, SYNC_STATUS_CHECK_ROLE, syncGenTranId()); + break; + } + + pNode->pRoleTimer = taosTmrStart(syncMonitorNodeRole, SYNC_ROLE_TIMER, (void *)pNode->rid, tsSyncTmrCtrl); + taosReleaseRef(tsSyncRefId, rid); +} + static void syncMonitorFwdInfos(void *param, void *tmrId) { - int64_t rid = (int64_t) param; + int64_t rid = (int64_t)param; SSyncNode *pNode = taosAcquireRef(tsSyncRefId, rid); if (pNode == NULL) return; @@ -1208,21 +1272,21 @@ static void syncMonitorFwdInfos(void *param, void *tmrId) { int64_t time = taosGetTimestampMs(); if (pSyncFwds->fwds > 0) { - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); for (int32_t i = 0; i < pSyncFwds->fwds; ++i) { SFwdInfo *pFwdInfo = pSyncFwds->fwdInfo + (pSyncFwds->first + i) % tsMaxFwdInfo; if (ABS(time - pFwdInfo->time) < 2000) break; sDebug("vgId:%d, forward info expired, hver:%" PRIu64 " curtime:%" PRIu64 " savetime:%" PRIu64, pNode->vgId, pFwdInfo->version, time, pFwdInfo->time); - syncProcessFwdAck(pNode, pFwdInfo, TSDB_CODE_RPC_NETWORK_UNAVAIL); + syncProcessFwdAck(pNode, pFwdInfo, TSDB_CODE_SYN_CONFIRM_EXPIRED); } syncRemoveConfirmedFwdInfo(pNode); - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); } - pNode->pFwdTimer = taosTmrStart(syncMonitorFwdInfos, 300, (void *)pNode->rid, tsSyncTmrCtrl); + pNode->pFwdTimer = taosTmrStart(syncMonitorFwdInfos, SYNC_FWD_TIMER, (void *)pNode->rid, tsSyncTmrCtrl); } taosReleaseRef(tsSyncRefId, rid); @@ -1235,9 +1299,8 @@ static int32_t syncForwardToPeerImpl(SSyncNode *pNode, void *data, void *mhandle int32_t fwdLen; int32_t code = 0; - if (pWalHead->version > nodeVersion + 1) { - sError("vgId:%d, hver:%" PRIu64 ", inconsistent with ver:%" PRIu64, pNode->vgId, pWalHead->version, nodeVersion); + sError("vgId:%d, hver:%" PRIu64 ", inconsistent with sver:%" PRIu64, pNode->vgId, pWalHead->version, nodeVersion); if (nodeRole == TAOS_SYNC_ROLE_SLAVE) { sInfo("vgId:%d, restart connection", pNode->vgId); for (int32_t i = 0; i < pNode->replica; ++i) { @@ -1250,23 +1313,23 @@ static int32_t syncForwardToPeerImpl(SSyncNode *pNode, void *data, void *mhandle } // always update version - nodeVersion = pWalHead->version; sTrace("vgId:%d, forward to peer, replica:%d role:%s qtype:%s hver:%" PRIu64, pNode->vgId, pNode->replica, syncRole[nodeRole], qtypeStr[qtype], pWalHead->version); + nodeVersion = pWalHead->version; if (pNode->replica == 1 || nodeRole != TAOS_SYNC_ROLE_MASTER) return 0; // only pkt from RPC or CQ can be forwarded if (qtype != TAOS_QTYPE_RPC && qtype != TAOS_QTYPE_CQ) return 0; - // a hacker way to improve the performance + // a hacker way to improve the performance pSyncHead = (SSyncHead *)(((char *)pWalHead) - sizeof(SSyncHead)); pSyncHead->type = TAOS_SMSG_FORWARD; pSyncHead->pversion = 0; pSyncHead->len = sizeof(SWalHead) + pWalHead->len; fwdLen = pSyncHead->len + sizeof(SSyncHead); // include the WAL and SYNC head - pthread_mutex_lock(&(pNode->mutex)); + pthread_mutex_lock(&pNode->mutex); for (int32_t i = 0; i < pNode->replica; ++i) { pPeer = pNode->peerInfo[i]; @@ -1278,16 +1341,18 @@ static int32_t syncForwardToPeerImpl(SSyncNode *pNode, void *data, void *mhandle code = 1; } - int32_t retLen = write(pPeer->peerFd, pSyncHead, fwdLen); + int32_t retLen = taosWriteMsg(pPeer->peerFd, pSyncHead, fwdLen); if (retLen == fwdLen) { - sDebug("%s, forward is sent, ver:%" PRIu64 " contLen:%d", pPeer->id, pWalHead->version, pWalHead->len); + sTrace("%s, forward is sent, role:%s sstatus:%s hver:%" PRIu64 " contLen:%d", pPeer->id, syncRole[pPeer->role], + syncStatus[pPeer->sstatus], pWalHead->version, pWalHead->len); } else { - sError("%s, failed to forward, ver:%" PRIu64 " retLen:%d", pPeer->id, pWalHead->version, retLen); + sError("%s, failed to forward, role:%s sstatus:%s hver:%" PRIu64 " retLen:%d", pPeer->id, syncRole[pPeer->role], + syncStatus[pPeer->sstatus], pWalHead->version, retLen); syncRestartConnection(pPeer); } } - pthread_mutex_unlock(&(pNode->mutex)); + pthread_mutex_unlock(&pNode->mutex); return code; } diff --git a/src/sync/src/syncRestore.c b/src/sync/src/syncRestore.c index 5d7b9eac9b..d156c93865 100644 --- a/src/sync/src/syncRestore.c +++ b/src/sync/src/syncRestore.c @@ -13,7 +13,9 @@ * along with this program. If not, see . */ +#define _DEFAULT_SOURCE #include "os.h" +#include "taoserror.h" #include "tlog.h" #include "tutil.h" #include "ttimer.h" @@ -36,7 +38,7 @@ static void syncRemoveExtraFile(SSyncPeer *pPeer, int32_t sindex, int32_t eindex while (1) { name[0] = 0; - magic = (*pNode->getFileInfo)(pNode->ahandle, name, &index, eindex, &size, &fversion); + magic = (*pNode->getFileInfo)(pNode->vgId, name, &index, eindex, &size, &fversion); if (magic == 0) break; snprintf(fname, sizeof(fname), "%s/%s", pNode->path, name); @@ -50,12 +52,12 @@ static void syncRemoveExtraFile(SSyncPeer *pPeer, int32_t sindex, int32_t eindex static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { SSyncNode *pNode = pPeer->pSyncNode; - SFileInfo minfo; memset(&minfo, 0, sizeof(minfo)); /* = {0}; */ // master file info - SFileInfo sinfo; memset(&sinfo, 0, sizeof(sinfo)); /* = {0}; */ // slave file info - SFileAck fileAck; + SFileInfo minfo; memset(&minfo, 0, sizeof(SFileInfo)); /* = {0}; */ + SFileInfo sinfo; memset(&sinfo, 0, sizeof(SFileInfo)); /* = {0}; */ + SFileAck fileAck = {0}; int32_t code = -1; char name[TSDB_FILENAME_LEN * 2] = {0}; - uint32_t pindex = 0; // index in last restore + uint32_t pindex = 0; // index in last restore bool fileChanged = false; *fversion = 0; @@ -63,7 +65,10 @@ static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { while (1) { // read file info int32_t ret = taosReadMsg(pPeer->syncFd, &(minfo), sizeof(minfo)); - if (ret < 0) break; + if (ret < 0) { + sError("%s, failed to read file info while restore file since %s", pPeer->id, strerror(errno)); + break; + } // if no more file from master, break; if (minfo.name[0] == 0 || minfo.magic == 0) { @@ -81,8 +86,8 @@ static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { // check the file info sinfo = minfo; - sDebug("%s, get file info:%s", pPeer->id, minfo.name); - sinfo.magic = (*pNode->getFileInfo)(pNode->ahandle, sinfo.name, &sinfo.index, TAOS_SYNC_MAX_INDEX, &sinfo.size, + sDebug("%s, get file:%s info size:%" PRId64, pPeer->id, minfo.name, minfo.size); + sinfo.magic = (*pNode->getFileInfo)(pNode->vgId, sinfo.name, &sinfo.index, TAOS_SYNC_MAX_INDEX, &sinfo.size, &sinfo.fversion); // if file not there or magic is not the same, file shall be synced @@ -90,8 +95,11 @@ static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { fileAck.sync = (sinfo.magic != minfo.magic || sinfo.name[0] == 0) ? 1 : 0; // send file ack - ret = taosWriteMsg(pPeer->syncFd, &(fileAck), sizeof(fileAck)); - if (ret < 0) break; + ret = taosWriteMsg(pPeer->syncFd, &fileAck, sizeof(fileAck)); + if (ret < 0) { + sError("%s, failed to write file:%s ack while restore file since %s", pPeer->id, minfo.name, strerror(errno)); + break; + } // if sync is not required, continue if (fileAck.sync == 0) { @@ -106,14 +114,17 @@ static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { int32_t dfd = open(name, O_WRONLY | O_CREAT | O_TRUNC, S_IRWXU | S_IRWXG | S_IRWXO); if (dfd < 0) { - sError("%s, failed to open file:%s", pPeer->id, name); + sError("%s, failed to open file:%s while restore file since %s", pPeer->id, minfo.name, strerror(errno)); break; } ret = taosCopyFds(pPeer->syncFd, dfd, minfo.size); fsync(dfd); close(dfd); - if (ret < 0) break; + if (ret < 0) { + sError("%s, failed to copy file:%s while restore file since %s", pPeer->id, minfo.name, strerror(errno)); + break; + } fileChanged = true; sDebug("%s, %s is received, size:%" PRId64, pPeer->id, minfo.name, minfo.size); @@ -123,10 +134,11 @@ static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { // data file is changed, code shall be set to 1 *fversion = minfo.fversion; code = 1; + sDebug("%s, file changed after restore file, fver:%" PRIu64, pPeer->id, *fversion); } if (code < 0) { - sError("%s, failed to restore %s(%s)", pPeer->id, name, strerror(errno)); + sError("%s, failed to restore %s since %s", pPeer->id, name, strerror(errno)); } return code; @@ -135,33 +147,46 @@ static int32_t syncRestoreFile(SSyncPeer *pPeer, uint64_t *fversion) { static int32_t syncRestoreWal(SSyncPeer *pPeer) { SSyncNode *pNode = pPeer->pSyncNode; int32_t ret, code = -1; + uint64_t lastVer = 0; - void *buffer = calloc(SYNC_MAX_SIZE, 1); // size for one record - if (buffer == NULL) return -1; - - SWalHead *pHead = (SWalHead *)buffer; + SWalHead *pHead = calloc(SYNC_MAX_SIZE, 1); // size for one record + if (pHead == NULL) return -1; while (1) { ret = taosReadMsg(pPeer->syncFd, pHead, sizeof(SWalHead)); - if (ret < 0) break; + if (ret < 0) { + sError("%s, failed to read walhead while restore wal since %s", pPeer->id, strerror(errno)); + break; + } if (pHead->len == 0) { + sDebug("%s, wal is synced over, last wver:%" PRIu64, pPeer->id, lastVer); code = 0; break; } // wal sync over ret = taosReadMsg(pPeer->syncFd, pHead->cont, pHead->len); - if (ret < 0) break; + if (ret < 0) { + sError("%s, failed to read walcont, len:%d while restore wal since %s", pPeer->id, pHead->len, strerror(errno)); + break; + } - sDebug("%s, restore a record, qtype:wal hver:%" PRIu64, pPeer->id, pHead->version); - (*pNode->writeToCache)(pNode->ahandle, pHead, TAOS_QTYPE_WAL, NULL); + sTrace("%s, restore a record, qtype:wal len:%d hver:%" PRIu64, pPeer->id, pHead->len, pHead->version); + + if (lastVer == pHead->version) { + sError("%s, failed to restore record, same hver:%" PRIu64 ", wal sync failed" PRIu64, pPeer->id, lastVer); + break; + } + lastVer = pHead->version; + + (*pNode->writeToCache)(pNode->vgId, pHead, TAOS_QTYPE_WAL, NULL); } if (code < 0) { - sError("%s, failed to restore wal(%s)", pPeer->id, strerror(errno)); + sError("%s, failed to restore wal from syncFd:%d since %s", pPeer->id, pPeer->syncFd, strerror(errno)); } - free(buffer); + free(pHead); return code; } @@ -169,7 +194,7 @@ static char *syncProcessOneBufferedFwd(SSyncPeer *pPeer, char *offset) { SSyncNode *pNode = pPeer->pSyncNode; SWalHead * pHead = (SWalHead *)offset; - (*pNode->writeToCache)(pNode->ahandle, pHead, TAOS_QTYPE_FWD, NULL); + (*pNode->writeToCache)(pNode->vgId, pHead, TAOS_QTYPE_FWD, NULL); offset += pHead->len + sizeof(SWalHead); return offset; @@ -206,15 +231,18 @@ static int32_t syncProcessBufferedFwd(SSyncPeer *pPeer) { int32_t syncSaveIntoBuffer(SSyncPeer *pPeer, SWalHead *pHead) { SSyncNode * pNode = pPeer->pSyncNode; SRecvBuffer *pRecv = pNode->pRecv; - - if (pRecv == NULL) return -1; int32_t len = pHead->len + sizeof(SWalHead); + if (pRecv == NULL) { + sError("%s, recv buffer is not create yet", pPeer->id); + return -1; + } + if (pRecv->bufferSize - (pRecv->offset - pRecv->buffer) >= len) { memcpy(pRecv->offset, pHead, len); pRecv->offset += len; pRecv->forwards++; - sDebug("%s, fwd is saved into queue, ver:%" PRIu64 " fwds:%d", pPeer->id, pHead->version, pRecv->forwards); + sTrace("%s, fwd is saved into queue, hver:%" PRIu64 " fwds:%d", pPeer->id, pHead->version, pRecv->forwards); } else { sError("%s, buffer size:%d is too small", pPeer->id, pRecv->bufferSize); pRecv->code = -1; // set error code @@ -257,7 +285,14 @@ static int32_t syncRestoreDataStepByStep(SSyncPeer *pPeer) { nodeSStatus = TAOS_SYNC_STATUS_FILE; uint64_t fversion = 0; - sDebug("%s, start to restore file", pPeer->id); + sInfo("%s, start to restore, sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); + SFirstPktRsp firstPktRsp = {.sync = 1}; + if (taosWriteMsg(pPeer->syncFd, &firstPktRsp, sizeof(SFirstPktRsp)) < 0) { + sError("%s, failed to send sync firstPkt rsp since %s", pPeer->id, strerror(errno)); + return -1; + } + + sInfo("%s, start to restore file, set sstatus:%s", pPeer->id, syncStatus[nodeSStatus]); int32_t code = syncRestoreFile(pPeer, &fversion); if (code < 0) { sError("%s, failed to restore file", pPeer->id); @@ -266,7 +301,7 @@ static int32_t syncRestoreDataStepByStep(SSyncPeer *pPeer) { // if code > 0, data file is changed, notify app, and pass the version if (code > 0 && pNode->notifyFileSynced) { - if ((*pNode->notifyFileSynced)(pNode->ahandle, fversion) < 0) { + if ((*pNode->notifyFileSynced)(pNode->vgId, fversion) < 0) { sError("%s, app not in ready state", pPeer->id); return -1; } @@ -274,14 +309,14 @@ static int32_t syncRestoreDataStepByStep(SSyncPeer *pPeer) { nodeVersion = fversion; - sDebug("%s, start to restore wal", pPeer->id); + sInfo("%s, start to restore wal", pPeer->id); if (syncRestoreWal(pPeer) < 0) { sError("%s, failed to restore wal", pPeer->id); return -1; } nodeSStatus = TAOS_SYNC_STATUS_CACHE; - sDebug("%s, start to insert buffered points", pPeer->id); + sInfo("%s, start to insert buffered points, set sstatus:%s", pPeer->id, syncStatus[nodeSStatus]); if (syncProcessBufferedFwd(pPeer) < 0) { sError("%s, failed to insert buffered points", pPeer->id); return -1; @@ -291,16 +326,17 @@ static int32_t syncRestoreDataStepByStep(SSyncPeer *pPeer) { } void *syncRestoreData(void *param) { - SSyncPeer *pPeer = (SSyncPeer *)param; + SSyncPeer *pPeer = param; SSyncNode *pNode = pPeer->pSyncNode; taosBlockSIGPIPE(); __sync_fetch_and_add(&tsSyncNum, 1); - (*pNode->notifyRole)(pNode->ahandle, TAOS_SYNC_ROLE_SYNCING); + (*pNode->notifyRole)(pNode->vgId, TAOS_SYNC_ROLE_SYNCING); if (syncOpenRecvBuffer(pNode) < 0) { - sError("%s, failed to allocate recv buffer", pPeer->id); + sError("%s, failed to allocate recv buffer, restart connection", pPeer->id); + syncRestartConnection(pPeer); } else { if (syncRestoreDataStepByStep(pPeer) == 0) { sInfo("%s, it is synced successfully", pPeer->id); @@ -313,9 +349,11 @@ void *syncRestoreData(void *param) { } } - (*pNode->notifyRole)(pNode->ahandle, nodeRole); + (*pNode->notifyRole)(pNode->vgId, nodeRole); nodeSStatus = TAOS_SYNC_STATUS_INIT; + sInfo("%s, sync over, set sstatus:%s", pPeer->id, syncStatus[nodeSStatus]); + taosClose(pPeer->syncFd); syncCloseRecvBuffer(pNode); __sync_fetch_and_sub(&tsSyncNum, 1); diff --git a/src/sync/src/syncRetrieve.c b/src/sync/src/syncRetrieve.c index 21151f1199..305ec492b9 100644 --- a/src/sync/src/syncRetrieve.c +++ b/src/sync/src/syncRetrieve.c @@ -13,11 +13,10 @@ * along with this program. If not, see . */ -#include -#include +#define _DEFAULT_SOURCE #include -#include #include "os.h" +#include "taoserror.h" #include "tlog.h" #include "tutil.h" #include "tglobal.h" @@ -27,117 +26,70 @@ #include "tsync.h" #include "syncInt.h" -static int32_t syncAddIntoWatchList(SSyncPeer *pPeer, char *name) { - sDebug("%s, start to monitor:%s", pPeer->id, name); +static int32_t syncAreFilesModified(SSyncNode *pNode, SSyncPeer *pPeer) { + if (pNode->getFileVersion == NULL) return TSDB_CODE_SUCCESS; - if (pPeer->notifyFd <= 0) { - pPeer->watchNum = 0; - pPeer->notifyFd = inotify_init1(IN_NONBLOCK); - if (pPeer->notifyFd < 0) { - sError("%s, failed to init inotify(%s)", pPeer->id, strerror(errno)); - return -1; - } - - if (pPeer->watchFd == NULL) pPeer->watchFd = malloc(sizeof(int32_t) * tsMaxWatchFiles); - if (pPeer->watchFd == NULL) { - sError("%s, failed to allocate watchFd", pPeer->id); - return -1; - } - - memset(pPeer->watchFd, -1, sizeof(int32_t) * tsMaxWatchFiles); + uint64_t fver = 0; + int32_t code = (*pNode->getFileVersion)(pNode->vgId, &fver); + if (code != 0) { + sInfo("%s, vnode is commiting while retrieve, last fver:%" PRIu64, pPeer->id, pPeer->lastVer); + pPeer->fileChanged = 1; + return TSDB_CODE_SYN_VND_COMMITING; } - int32_t *wd = pPeer->watchFd + pPeer->watchNum; - - if (*wd >= 0) { - if (inotify_rm_watch(pPeer->notifyFd, *wd) < 0) { - sError("%s, failed to remove wd:%d(%s)", pPeer->id, *wd, strerror(errno)); - return -1; - } + if (fver != pPeer->lastVer) { + sInfo("%s, files are modified while retrieve, fver:%" PRIu64 ", last fver:%" PRIu64, pPeer->id, fver, pPeer->lastVer); + pPeer->fileChanged = 1; + return TSDB_CODE_SYN_FILE_CHNAGED; } - *wd = inotify_add_watch(pPeer->notifyFd, name, IN_MODIFY | IN_DELETE); - if (*wd == -1) { - sError("%s, failed to add %s(%s)", pPeer->id, name, strerror(errno)); - return -1; - } else { - sDebug("%s, monitor %s, wd:%d watchNum:%d", pPeer->id, name, *wd, pPeer->watchNum); - } - - pPeer->watchNum = (pPeer->watchNum + 1) % tsMaxWatchFiles; - - return 0; -} - -static int32_t syncAreFilesModified(SSyncPeer *pPeer) { - if (pPeer->notifyFd <= 0) return 0; - - char buf[2048]; - int32_t len = read(pPeer->notifyFd, buf, sizeof(buf)); - if (len < 0 && errno != EAGAIN) { - sError("%s, failed to read notify FD(%s)", pPeer->id, strerror(errno)); - return -1; - } - - int32_t code = 0; - if (len > 0) { - const struct inotify_event *event; - char *ptr; - for (ptr = buf; ptr < buf + len; ptr += sizeof(struct inotify_event) + event->len) { - event = (const struct inotify_event *)ptr; - if ((event->mask & IN_MODIFY) || (event->mask & IN_DELETE)) { - sDebug("%s, processed file is changed", pPeer->id); - pPeer->fileChanged = 1; - code = 1; - break; - } - } - } - - return code; + pPeer->fileChanged = 0; + return TSDB_CODE_SUCCESS; } static int32_t syncRetrieveFile(SSyncPeer *pPeer) { SSyncNode *pNode = pPeer->pSyncNode; - SFileInfo fileInfo; - SFileAck fileAck; - int32_t code = -1; + SFileInfo fileInfo; memset(&fileInfo, 0, sizeof(SFileInfo)); + SFileAck fileAck = {0}; + int32_t code = TSDB_CODE_SYN_APP_ERROR; char name[TSDB_FILENAME_LEN * 2] = {0}; - memset(&fileInfo, 0, sizeof(fileInfo)); - memset(&fileAck, 0, sizeof(fileAck)); + if (pNode->getFileVersion) (*pNode->getFileVersion)(pNode->vgId, &pPeer->lastVer); while (1) { // retrieve file info fileInfo.name[0] = 0; - fileInfo.magic = (*pNode->getFileInfo)(pNode->ahandle, fileInfo.name, &fileInfo.index, TAOS_SYNC_MAX_INDEX, + fileInfo.magic = (*pNode->getFileInfo)(pNode->vgId, fileInfo.name, &fileInfo.index, TAOS_SYNC_MAX_INDEX, &fileInfo.size, &fileInfo.fversion); // fileInfo.size = htonl(size); + sDebug("%s, file:%s info is sent, size:%" PRId64, pPeer->id, fileInfo.name, fileInfo.size); // send the file info int32_t ret = taosWriteMsg(pPeer->syncFd, &(fileInfo), sizeof(fileInfo)); - if (ret < 0) break; + if (ret < 0) { + code = TAOS_SYSTEM_ERROR(errno); + sError("%s, failed to write file:%s info while retrieve file since %s", pPeer->id, fileInfo.name, strerror(errno)); + break; + } // if no file anymore, break if (fileInfo.magic == 0 || fileInfo.name[0] == 0) { + code = TSDB_CODE_SUCCESS; sDebug("%s, no more files to sync", pPeer->id); - code = 0; break; } // wait for the ack from peer - ret = taosReadMsg(pPeer->syncFd, &(fileAck), sizeof(fileAck)); - if (ret < 0) break; + ret = taosReadMsg(pPeer->syncFd, &fileAck, sizeof(fileAck)); + if (ret < 0) { + code = TAOS_SYSTEM_ERROR(errno); + sError("%s, failed to read file:%s ack while retrieve file since %s", pPeer->id, fileInfo.name, strerror(errno)); + break; + } // set the peer sync version pPeer->sversion = fileInfo.fversion; - // get the full path to file - snprintf(name, sizeof(name), "%s/%s", pNode->path, fileInfo.name); - - // add the file into watch list - if (syncAddIntoWatchList(pPeer, name) < 0) break; - // if sync is not required, continue if (fileAck.sync == 0) { fileInfo.index++; @@ -145,23 +97,35 @@ static int32_t syncRetrieveFile(SSyncPeer *pPeer) { continue; } + // get the full path to file + snprintf(name, sizeof(name), "%s/%s", pNode->path, fileInfo.name); + // send the file to peer int32_t sfd = open(name, O_RDONLY); - if (sfd < 0) break; + if (sfd < 0) { + code = TAOS_SYSTEM_ERROR(errno); + sError("%s, failed to open file:%s while retrieve file since %s", pPeer->id, fileInfo.name, strerror(errno)); + break; + } ret = taosSendFile(pPeer->syncFd, sfd, NULL, fileInfo.size); close(sfd); - if (ret < 0) break; + if (ret < 0) { + code = TAOS_SYSTEM_ERROR(errno); + sError("%s, failed to send file:%s while retrieve file since %s", pPeer->id, fileInfo.name, strerror(errno)); + break; + } - sDebug("%s, %s is sent, size:%" PRId64, pPeer->id, name, fileInfo.size); + sDebug("%s, file:%s is sent, size:%" PRId64, pPeer->id, fileInfo.name, fileInfo.size); fileInfo.index++; // check if processed files are modified - if (syncAreFilesModified(pPeer) != 0) break; + code = syncAreFilesModified(pNode, pPeer); + if (code != TSDB_CODE_SUCCESS) break; } - if (code < 0) { - sError("%s, failed to retrieve file(%s)", pPeer->id, strerror(errno)); + if (code != TSDB_CODE_SUCCESS) { + sError("%s, failed to retrieve file since %s", pPeer->id, tstrerror(code)); } return code; @@ -201,7 +165,7 @@ static int32_t syncMonitorLastWal(SSyncPeer *pPeer, char *name) { taosClose(pPeer->notifyFd); pPeer->notifyFd = inotify_init1(IN_NONBLOCK); if (pPeer->notifyFd < 0) { - sError("%s, failed to init inotify(%s)", pPeer->id, strerror(errno)); + sError("%s, failed to init inotify since %s", pPeer->id, strerror(errno)); return -1; } @@ -216,7 +180,7 @@ static int32_t syncMonitorLastWal(SSyncPeer *pPeer, char *name) { *wd = inotify_add_watch(pPeer->notifyFd, name, IN_MODIFY | IN_CLOSE_WRITE); if (*wd == -1) { - sError("%s, failed to watch last wal(%s)", pPeer->id, strerror(errno)); + sError("%s, failed to watch last wal since %s", pPeer->id, strerror(errno)); return -1; } @@ -227,7 +191,7 @@ static int32_t syncCheckLastWalChanges(SSyncPeer *pPeer, uint32_t *pEvent) { char buf[2048]; int32_t len = read(pPeer->notifyFd, buf, sizeof(buf)); if (len < 0 && errno != EAGAIN) { - sError("%s, failed to read notify FD(%s)", pPeer->id, strerror(errno)); + sError("%s, failed to read notify FD since %s", pPeer->id, strerror(errno)); return -1; } @@ -268,7 +232,7 @@ static int32_t syncRetrieveLastWal(SSyncPeer *pPeer, char *name, uint64_t fversi break; } - sDebug("%s, last wal is forwarded, ver:%" PRIu64, pPeer->id, pHead->version); + sTrace("%s, last wal is forwarded, hver:%" PRIu64, pPeer->id, pHead->version); int32_t ret = taosWriteMsg(pPeer->syncFd, pHead, wsize); if (ret != wsize) break; pPeer->sversion = pHead->version; @@ -292,9 +256,9 @@ static int32_t syncRetrieveLastWal(SSyncPeer *pPeer, char *name, uint64_t fversi static int32_t syncProcessLastWal(SSyncPeer *pPeer, char *wname, int64_t index) { SSyncNode *pNode = pPeer->pSyncNode; int32_t code = -1; - char fname[TSDB_FILENAME_LEN * 2]; // full path to wal file + char fname[TSDB_FILENAME_LEN * 2] = {0}; // full path to wal file - if (syncAreFilesModified(pPeer) != 0) return -1; + if (syncAreFilesModified(pNode, pPeer) != 0) return -1; while (1) { int32_t once = 0; // last WAL has once ever been processed @@ -320,6 +284,7 @@ static int32_t syncProcessLastWal(SSyncPeer *pPeer, char *wname, int64_t index) if (((event & IN_MODIFY) == 0) || once) { if (fversion == 0) { pPeer->sstatus = TAOS_SYNC_STATUS_CACHE; // start to forward pkt + sDebug("%s, fversion is 0 then set sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); fversion = nodeVersion; // must read data to fversion } } @@ -356,7 +321,7 @@ static int32_t syncProcessLastWal(SSyncPeer *pPeer, char *wname, int64_t index) index++; wname[0] = 0; - code = (*pNode->getWalInfo)(pNode->ahandle, wname, &index); + code = (*pNode->getWalInfo)(pNode->vgId, wname, &index); if (code < 0) break; if (wname[0] == 0) { code = 0; @@ -384,7 +349,7 @@ static int32_t syncRetrieveWal(SSyncPeer *pPeer) { while (1) { // retrieve wal info wname[0] = 0; - code = (*pNode->getWalInfo)(pNode->ahandle, wname, &index); + code = (*pNode->getWalInfo)(pNode->vgId, wname, &index); if (code < 0) break; // error if (wname[0] == 0) { // no wal file sDebug("%s, no wal file", pPeer->id); @@ -412,25 +377,24 @@ static int32_t syncRetrieveWal(SSyncPeer *pPeer) { close(sfd); if (code < 0) break; - index++; - - if (syncAreFilesModified(pPeer) != 0) break; + if (syncAreFilesModified(pNode, pPeer) != 0) break; } if (code == 0) { - sDebug("%s, wal retrieve is finished", pPeer->id); pPeer->sstatus = TAOS_SYNC_STATUS_CACHE; + sInfo("%s, wal retrieve is finished, set sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); + SWalHead walHead; memset(&walHead, 0, sizeof(walHead)); code = taosWriteMsg(pPeer->syncFd, &walHead, sizeof(walHead)); } else { - sError("%s, failed to send wal(%s)", pPeer->id, strerror(errno)); + sError("%s, failed to send wal since %s", pPeer->id, strerror(errno)); } return code; } -static int32_t syncRetrieveDataStepByStep(SSyncPeer *pPeer) { +static int32_t syncRetrieveFirstPkt(SSyncPeer *pPeer) { SSyncNode *pNode = pPeer->pSyncNode; SFirstPkt firstPkt; @@ -440,25 +404,41 @@ static int32_t syncRetrieveDataStepByStep(SSyncPeer *pPeer) { tstrncpy(firstPkt.fqdn, tsNodeFqdn, sizeof(firstPkt.fqdn)); firstPkt.port = tsSyncPort; - if (write(pPeer->syncFd, (char *)&firstPkt, sizeof(firstPkt)) < 0) { - sError("%s, failed to send syncCmd", pPeer->id); + if (taosWriteMsg(pPeer->syncFd, &firstPkt, sizeof(firstPkt)) < 0) { + sError("%s, failed to send sync firstPkt since %s", pPeer->id, strerror(errno)); + return -1; + } + + SFirstPktRsp firstPktRsp; + if (taosReadMsg(pPeer->syncFd, &firstPktRsp, sizeof(SFirstPktRsp)) < 0) { + sError("%s, failed to read sync firstPkt rsp since %s", pPeer->id, strerror(errno)); + return -1; + } + + return 0; +} + +static int32_t syncRetrieveDataStepByStep(SSyncPeer *pPeer) { + sInfo("%s, start to retrieve, sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); + if (syncRetrieveFirstPkt(pPeer) < 0) { + sError("%s, failed to start retrieve", pPeer->id); return -1; } pPeer->sversion = 0; pPeer->sstatus = TAOS_SYNC_STATUS_FILE; - sDebug("%s, start to retrieve file", pPeer->id); + sInfo("%s, start to retrieve files, set sstatus:%s", pPeer->id, syncStatus[pPeer->sstatus]); if (syncRetrieveFile(pPeer) < 0) { - sError("%s, failed to retrieve file", pPeer->id); + sError("%s, failed to retrieve files", pPeer->id); return -1; } // if no files are synced, there must be wal to sync, sversion must be larger than one if (pPeer->sversion == 0) pPeer->sversion = 1; - sDebug("%s, start to retrieve wal", pPeer->id); + sInfo("%s, start to retrieve wals", pPeer->id); if (syncRetrieveWal(pPeer) < 0) { - sError("%s, failed to retrieve wal", pPeer->id); + sError("%s, failed to retrieve wals", pPeer->id); return -1; } @@ -470,7 +450,8 @@ void *syncRetrieveData(void *param) { SSyncNode *pNode = pPeer->pSyncNode; taosBlockSIGPIPE(); - pPeer->fileChanged = 0; + if (pNode->notifyFlowCtrl) (*pNode->notifyFlowCtrl)(pNode->vgId, pPeer->numOfRetrieves); + pPeer->syncFd = taosOpenTcpClientSocket(pPeer->ip, pPeer->port, 0); if (pPeer->syncFd < 0) { sError("%s, failed to open socket to sync", pPeer->id); @@ -478,7 +459,7 @@ void *syncRetrieveData(void *param) { sInfo("%s, sync tcp is setup", pPeer->id); if (syncRetrieveDataStepByStep(pPeer) == 0) { - sDebug("%s, sync retrieve process is successful", pPeer->id); + sInfo("%s, sync retrieve process is successful", pPeer->id); } else { sError("%s, failed to retrieve data, restart connection", pPeer->id); syncRestartConnection(pPeer); @@ -486,13 +467,10 @@ void *syncRetrieveData(void *param) { } if (pPeer->fileChanged) { - // if file is changed 3 times continuously, start flow control pPeer->numOfRetrieves++; - if (pPeer->numOfRetrieves >= 2 && pNode->notifyFlowCtrl) - (*pNode->notifyFlowCtrl)(pNode->ahandle, 4 << (pPeer->numOfRetrieves - 2)); } else { pPeer->numOfRetrieves = 0; - if (pNode->notifyFlowCtrl) (*pNode->notifyFlowCtrl)(pNode->ahandle, 0); + if (pNode->notifyFlowCtrl) (*pNode->notifyFlowCtrl)(pNode->vgId, 0); } pPeer->fileChanged = 0; diff --git a/src/sync/src/taosTcpPool.c b/src/sync/src/taosTcpPool.c index 875528e66b..eb05cf7c6f 100644 --- a/src/sync/src/taosTcpPool.c +++ b/src/sync/src/taosTcpPool.c @@ -150,7 +150,7 @@ void *taosAllocateTcpConn(void *param, void *pPeer, int32_t connFd) { } void taosFreeTcpConn(void *param) { - SConnObj * pConn = (SConnObj *)param; + SConnObj * pConn = param; SThreadObj *pThread = pConn->pThread; sDebug("%p TCP connection will be closed, fd:%d", pThread, pConn->fd); @@ -203,16 +203,19 @@ static void *taosProcessTcpData(void *param) { assert(pConn); if (events[i].events & EPOLLERR) { + sDebug("conn is broken since EPOLLERR"); taosProcessBrokenLink(pConn); continue; } if (events[i].events & EPOLLHUP) { + sDebug("conn is broken since EPOLLHUP"); taosProcessBrokenLink(pConn); continue; } if (events[i].events & EPOLLRDHUP) { + sDebug("conn is broken since EPOLLRDHUP"); taosProcessBrokenLink(pConn); continue; } diff --git a/src/sync/src/tarbitrator.c b/src/sync/src/tarbitrator.c index b7f819a3cd..4016042de2 100644 --- a/src/sync/src/tarbitrator.c +++ b/src/sync/src/tarbitrator.c @@ -115,14 +115,14 @@ static void arbProcessIncommingConnection(int32_t connFd, uint32_t sourceIp) { SFirstPkt firstPkt; if (taosReadMsg(connFd, &firstPkt, sizeof(firstPkt)) != sizeof(firstPkt)) { - sError("failed to read peer first pkt from ip:%s(%s)", ipstr, strerror(errno)); + sError("failed to read peer first pkt from ip:%s since %s", ipstr, strerror(errno)); taosCloseSocket(connFd); return; } - SNodeConn *pNode = (SNodeConn *)calloc(sizeof(SNodeConn), 1); + SNodeConn *pNode = calloc(sizeof(SNodeConn), 1); if (pNode == NULL) { - sError("failed to allocate memory(%s)", strerror(errno)); + sError("failed to allocate memory since %s", strerror(errno)); taosCloseSocket(connFd); return; } @@ -146,7 +146,7 @@ static void arbProcessIncommingConnection(int32_t connFd, uint32_t sourceIp) { static void arbProcessBrokenLink(void *param) { SNodeConn *pNode = param; - sDebug("%s, TCP link is broken(%s), close connection", pNode->id, strerror(errno)); + sDebug("%s, TCP link is broken since %s, close connection", pNode->id, strerror(errno)); tfree(pNode); } diff --git a/src/sync/test/syncServer.c b/src/sync/test/syncServer.c index 9dd3feb461..5a64a0a36a 100644 --- a/src/sync/test/syncServer.c +++ b/src/sync/test/syncServer.c @@ -70,7 +70,7 @@ int writeIntoWal(SWalHead *pHead) { return 0; } -void confirmForward(void *ahandle, void *mhandle, int32_t code) { +void confirmForward(int32_t vgId, void *mhandle, int32_t code) { SRpcMsg * pMsg = (SRpcMsg *)mhandle; SWalHead *pHead = (SWalHead *)(((char *)pMsg->pCont) - sizeof(SWalHead)); @@ -227,7 +227,7 @@ void processRequestMsg(SRpcMsg *pMsg, SRpcEpSet *pEpSet) { taosWriteQitem(qhandle, TAOS_QTYPE_RPC, pTemp); } -uint32_t getFileInfo(void *ahandle, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion) { +uint32_t getFileInfo(int32_t vgId, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion) { uint32_t magic; struct stat fstat; char aname[280]; @@ -254,7 +254,7 @@ uint32_t getFileInfo(void *ahandle, char *name, uint32_t *index, uint32_t eindex return magic; } -int getWalInfo(void *ahandle, char *name, int64_t *index) { +int getWalInfo(int32_t vgId, char *name, int64_t *index) { struct stat fstat; char aname[280]; @@ -272,7 +272,7 @@ int getWalInfo(void *ahandle, char *name, int64_t *index) { return 1; } -int writeToCache(void *ahandle, void *data, int type) { +int writeToCache(int32_t vgId, void *data, int type) { SWalHead *pHead = data; uDebug("pkt from peer is received, ver:%" PRIu64 " len:%d type:%d", pHead->version, pHead->len, type); @@ -285,9 +285,9 @@ int writeToCache(void *ahandle, void *data, int type) { return 0; } -void confirmFwd(void *ahandle, int64_t version) { return; } +void confirmFwd(int32_t vgId, int64_t version) { return; } -void notifyRole(void *ahandle, int8_t r) { +void notifyRole(int32_t vgId, int8_t r) { role = r; printf("current role:%s\n", syncRole[role]); } @@ -296,7 +296,6 @@ void initSync() { pCfg->replica = 1; pCfg->quorum = 1; syncInfo.vgId = 1; - syncInfo.ahandle = &syncInfo; syncInfo.getFileInfo = getFileInfo; syncInfo.getWalInfo = getWalInfo; syncInfo.writeToCache = writeToCache; diff --git a/src/tsdb/src/tsdbMemTable.c b/src/tsdb/src/tsdbMemTable.c index 71944c87c6..4c51608b5d 100644 --- a/src/tsdb/src/tsdbMemTable.c +++ b/src/tsdb/src/tsdbMemTable.c @@ -871,10 +871,10 @@ static void tsdbFreeRows(STsdbRepo *pRepo, void **rows, int rowCounter) { listNEles(pRepo->mem->bufBlockList), pBufBlock->offset, pBufBlock->remain); if (pBufBlock->offset == 0) { // return the block to buffer pool - tsdbLockRepo(pRepo); + if (tsdbLockRepo(pRepo) < 0) return; SListNode *pNode = tdListPopTail(pRepo->mem->bufBlockList); tdListPrependNode(pBufPool->bufBlockList, pNode); - tsdbUnlockRepo(pRepo); + if (tsdbUnlockRepo(pRepo) < 0) return; } } else { ASSERT(listNEles(pRepo->mem->extraBuffList) > 0); diff --git a/src/util/inc/hash.h b/src/util/inc/hash.h index 42bc136584..b0319d3e13 100644 --- a/src/util/inc/hash.h +++ b/src/util/inc/hash.h @@ -31,16 +31,18 @@ extern "C" { typedef void (*_hash_free_fn_t)(void *param); typedef struct SHashNode { -// char *key; struct SHashNode *next; uint32_t hashVal; // the hash value of key uint32_t keyLen; // length of the key -// char *data; + size_t dataLen; // length of data + int8_t count; // reference count + int8_t removed; // flag to indicate removed + char data[]; } SHashNode; -#define GET_HASH_NODE_KEY(_n) ((char*)(_n) + sizeof(SHashNode)) -#define GET_HASH_NODE_DATA(_n) ((char*)(_n) + sizeof(SHashNode) + (_n)->keyLen) - +#define GET_HASH_NODE_KEY(_n) ((char*)(_n) + sizeof(SHashNode) + (_n)->dataLen) +#define GET_HASH_NODE_DATA(_n) ((char*)(_n) + sizeof(SHashNode)) +#define GET_HASH_PNODE(_n) ((char*)(_n) - sizeof(SHashNode)); typedef enum SHashLockTypeE { HASH_NO_LOCK = 0, HASH_ENTRY_LOCK = 1, @@ -65,15 +67,6 @@ typedef struct SHashObj { SArray *pMemBlock; // memory block allocated for SHashEntry } SHashObj; -typedef struct SHashMutableIterator { - SHashObj *pHashObj; - int32_t entryIndex; - SHashNode *pCur; - SHashNode *pNext; // current node can be deleted for mutable iterator, so keep the next one before return current - size_t numOfChecked; // already check number of elements in hash table - size_t numOfEntries; // number of entries while the iterator is created -} SHashMutableIterator; - /** * init the hash table * @@ -142,33 +135,9 @@ int32_t taosHashCondTraverse(SHashObj *pHashObj, bool (*fp)(void *, void *), voi */ void taosHashCleanup(SHashObj *pHashObj); -/** - * - * @param pHashObj - * @return - */ -SHashMutableIterator* taosHashCreateIter(SHashObj *pHashObj); - -/** - * - * @param iter - * @return - */ -bool taosHashIterNext(SHashMutableIterator *iter); - -/** - * - * @param iter - * @return - */ -void *taosHashIterGet(SHashMutableIterator *iter); - -/** - * - * @param iter - * @return - */ -void* taosHashDestroyIter(SHashMutableIterator* iter); +/* +void *SHashMutableIterator* taosHashCreateIter(SHashObj *pHashObj, void *); +*/ /** * @@ -179,6 +148,9 @@ int32_t taosHashGetMaxOverflowLinkLength(const SHashObj *pHashObj); size_t taosHashGetMemSize(const SHashObj *pHashObj); +void *taosHashIterate(SHashObj *pHashObj, void *p); +void taosHashCancelIterate(SHashObj *pHashObj, void *p); + #ifdef __cplusplus } #endif diff --git a/src/util/inc/tfile.h b/src/util/inc/tfile.h index 10b7c1df35..066040170e 100644 --- a/src/util/inc/tfile.h +++ b/src/util/inc/tfile.h @@ -20,23 +20,26 @@ extern "C" { #endif -#include - // init taos file module -int32_t tfinit(); +int32_t tfInit(); // clean up taos file module -void tfcleanup(); +void tfCleanup(); // the same syntax as UNIX standard open/close/read/write // but FD is int64_t and will never be reused -int64_t tfopen(const char *pathname, int32_t flags); -int64_t tfclose(int64_t tfd); -int64_t tfwrite(int64_t tfd, void *buf, int64_t count); -int64_t tfread(int64_t tfd, void *buf, int64_t count); +int64_t tfOpen(const char *pathname, int32_t flags); +int64_t tfOpenM(const char *pathname, int32_t flags, mode_t mode); +int64_t tfClose(int64_t tfd); +int64_t tfWrite(int64_t tfd, void *buf, int64_t count); +int64_t tfRead(int64_t tfd, void *buf, int64_t count); +int32_t tfFsync(int64_t tfd); +bool tfValid(int64_t tfd); +int64_t tfLseek(int64_t tfd, int64_t offset, int32_t whence); +int32_t tfFtruncate(int64_t tfd, int64_t length); #ifdef __cplusplus } #endif -#endif // TDENGINE_TREF_H +#endif // TDENGINE_TFILE_H diff --git a/src/util/src/hash.c b/src/util/src/hash.c index 03a7342497..3ac54eedd8 100644 --- a/src/util/src/hash.c +++ b/src/util/src/hash.c @@ -76,7 +76,7 @@ static FORCE_INLINE int32_t taosHashCapacity(int32_t length) { static FORCE_INLINE SHashNode *doSearchInEntryList(SHashEntry *pe, const void *key, size_t keyLen, uint32_t hashVal) { SHashNode *pNode = pe->next; while (pNode) { - if ((pNode->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY(pNode), key, keyLen) == 0)) { + if ((pNode->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY(pNode), key, keyLen) == 0) && pNode->removed == 0) { assert(pNode->hashVal == hashVal); break; } @@ -114,15 +114,25 @@ static SHashNode *doCreateHashNode(const void *key, size_t keyLen, const void *p * @param dsize size of actual data * @return hash node */ -static FORCE_INLINE SHashNode *doUpdateHashNode(SHashEntry* pe, SHashNode* prev, SHashNode *pNode, SHashNode *pNewNode) { +static FORCE_INLINE SHashNode *doUpdateHashNode(SHashObj *pHashObj, SHashEntry* pe, SHashNode* prev, SHashNode *pNode, SHashNode *pNewNode) { assert(pNode->keyLen == pNewNode->keyLen); + + pNode->count--; if (prev != NULL) { prev->next = pNewNode; } else { pe->next = pNewNode; } - pNewNode->next = pNode->next; + if (pNode->count <= 0) { + pNewNode->next = pNode->next; + DO_FREE_HASH_NODE(pNode); + } else { + pNewNode->next = pNode; + pe->num++; + atomic_add_fetch_64(&pHashObj->size, 1); + } + return pNewNode; } @@ -139,7 +149,6 @@ static void pushfrontNodeInEntryList(SHashEntry *pEntry, SHashNode *pNode); * @param pIter * @return */ -static SHashNode *getNextHashNode(SHashMutableIterator *pIter); SHashObj *taosHashInit(size_t capacity, _hash_fn_t fn, bool update, SHashLockTypeE type) { if (capacity == 0 || fn == NULL) { @@ -213,7 +222,7 @@ int32_t taosHashPut(SHashObj *pHashObj, const void *key, size_t keyLen, void *da SHashNode* prev = NULL; while (pNode) { - if ((pNode->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY(pNode), key, keyLen) == 0)) { + if ((pNode->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY(pNode), key, keyLen) == 0) && pNode->removed == 0) { assert(pNode->hashVal == hashVal); break; } @@ -244,8 +253,7 @@ int32_t taosHashPut(SHashObj *pHashObj, const void *key, size_t keyLen, void *da } else { // not support the update operation, return error if (pHashObj->enableUpdate) { - doUpdateHashNode(pe, prev, pNode, pNewNode); - DO_FREE_HASH_NODE(pNode); + doUpdateHashNode(pHashObj, pe, prev, pNode, pNewNode); } else { DO_FREE_HASH_NODE(pNewNode); } @@ -335,22 +343,10 @@ int32_t taosHashRemoveWithData(SHashObj *pHashObj, const void *key, size_t keyLe int32_t slot = HASH_INDEX(hashVal, pHashObj->capacity); SHashEntry *pe = pHashObj->hashList[slot]; - // no data, return directly - if (pe->num == 0) { - __rd_unlock(&pHashObj->lock, pHashObj->type); - return -1; - } - if (pHashObj->type == HASH_ENTRY_LOCK) { taosWLockLatch(&pe->latch); } - if (pe->num == 0) { - assert(pe->next == NULL); - } else { - assert(pe->next != NULL); - } - // double check after locked if (pe->num == 0) { assert(pe->next == NULL); @@ -360,37 +356,37 @@ int32_t taosHashRemoveWithData(SHashObj *pHashObj, const void *key, size_t keyLe return -1; } + int code = -1; SHashNode *pNode = pe->next; - SHashNode *pRes = NULL; + SHashNode *prevNode = NULL; - // remove it - if ((pNode->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY(pNode), key, keyLen) == 0)) { - pe->num -= 1; - pRes = pNode; - pe->next = pNode->next; - } else { - while (pNode->next != NULL) { - if (((pNode->next)->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY((pNode->next)), key, keyLen) == 0)) { - assert((pNode->next)->hashVal == hashVal); - break; + while (pNode) { + if ((pNode->keyLen == keyLen) && (memcmp(GET_HASH_NODE_KEY(pNode), key, keyLen) == 0) && pNode->removed == 0) + break; + + prevNode = pNode; + pNode = pNode->next; + } + + if (pNode) { + code = 0; // it is found + + pNode->count--; + pNode->removed = 1; + if (pNode->count <= 0) { + if (prevNode) { + prevNode->next = pNode->next; + } else { + pe->next = pNode->next; } + + if (data) memcpy(data, GET_HASH_NODE_DATA(pNode), dsize); - pNode = pNode->next; + pe->num--; + atomic_sub_fetch_64(&pHashObj->size, 1); + FREE_HASH_NODE(pHashObj, pNode); } - - - if (pNode->next != NULL) { - pe->num -= 1; - pRes = pNode->next; - pNode->next = pNode->next->next; - } - } - - if (pe->num == 0) { - assert(pe->next == NULL); - } else { - assert(pe->next != NULL); - } + } if (pHashObj->type == HASH_ENTRY_LOCK) { taosWUnLockLatch(&pe->latch); @@ -398,17 +394,7 @@ int32_t taosHashRemoveWithData(SHashObj *pHashObj, const void *key, size_t keyLe __rd_unlock(&pHashObj->lock, pHashObj->type); - if (data != NULL && pRes != NULL) { - memcpy(data, GET_HASH_NODE_DATA(pRes), dsize); - } - - if (pRes != NULL) { - atomic_sub_fetch_64(&pHashObj->size, 1); - FREE_HASH_NODE(pHashObj, pRes); - return 0; - } else { - return -1; - } + return code; } int32_t taosHashCondTraverse(SHashObj *pHashObj, bool (*fp)(void *, void *), void *param) { @@ -531,98 +517,6 @@ void taosHashCleanup(SHashObj *pHashObj) { free(pHashObj); } -SHashMutableIterator *taosHashCreateIter(SHashObj *pHashObj) { - SHashMutableIterator *pIter = calloc(1, sizeof(SHashMutableIterator)); - if (pIter == NULL) { - return NULL; - } - - pIter->pHashObj = pHashObj; - - // keep it in local variable, in case the resize operation expand the size - pIter->numOfEntries = pHashObj->capacity; - return pIter; -} - -bool taosHashIterNext(SHashMutableIterator *pIter) { - if (pIter == NULL) { - return false; - } - - size_t size = taosHashGetSize(pIter->pHashObj); - if (size == 0) { - return false; - } - - // check the first one - if (pIter->numOfChecked == 0) { - assert(pIter->pCur == NULL && pIter->pNext == NULL); - - while (1) { - SHashEntry *pEntry = pIter->pHashObj->hashList[pIter->entryIndex]; - if (pEntry->num == 0) { - assert(pEntry->next == NULL); - - pIter->entryIndex++; - continue; - } - - if (pIter->pHashObj->type == HASH_ENTRY_LOCK) { - taosRLockLatch(&pEntry->latch); - } - - pIter->pCur = pEntry->next; - - if (pIter->pCur->next) { - pIter->pNext = pIter->pCur->next; - - if (pIter->pHashObj->type == HASH_ENTRY_LOCK) { - taosRUnLockLatch(&pEntry->latch); - } - } else { - if (pIter->pHashObj->type == HASH_ENTRY_LOCK) { - taosRUnLockLatch(&pEntry->latch); - } - - pIter->pNext = getNextHashNode(pIter); - } - - break; - } - - pIter->numOfChecked++; - return true; - } else { - assert(pIter->pCur != NULL); - if (pIter->pNext) { - pIter->pCur = pIter->pNext; - } else { // no more data in the hash list - return false; - } - - pIter->numOfChecked++; - - if (pIter->pCur->next) { - pIter->pNext = pIter->pCur->next; - } else { - pIter->pNext = getNextHashNode(pIter); - } - - return true; - } -} - -void *taosHashIterGet(SHashMutableIterator *iter) { return (iter == NULL) ? NULL : GET_HASH_NODE_DATA(iter->pCur); } - -void *taosHashDestroyIter(SHashMutableIterator *iter) { - if (iter == NULL) { - return NULL; - } - - free(iter); - return NULL; -} - // for profile only int32_t taosHashGetMaxOverflowLinkLength(const SHashObj *pHashObj) { if (pHashObj == NULL || pHashObj->size == 0) { @@ -759,6 +653,8 @@ SHashNode *doCreateHashNode(const void *key, size_t keyLen, const void *pData, s pNewNode->keyLen = (uint32_t)keyLen; pNewNode->hashVal = hashVal; + pNewNode->dataLen = dsize; + pNewNode->count = 1; memcpy(GET_HASH_NODE_DATA(pNewNode), pData, dsize); memcpy(GET_HASH_NODE_KEY(pNewNode), key, keyLen); @@ -775,35 +671,6 @@ void pushfrontNodeInEntryList(SHashEntry *pEntry, SHashNode *pNode) { pEntry->num += 1; } -SHashNode *getNextHashNode(SHashMutableIterator *pIter) { - assert(pIter != NULL); - - pIter->entryIndex++; - SHashNode *p = NULL; - - while (pIter->entryIndex < pIter->numOfEntries) { - SHashEntry *pEntry = pIter->pHashObj->hashList[pIter->entryIndex]; - if (pEntry->num == 0) { - pIter->entryIndex++; - continue; - } - - if (pIter->pHashObj->type == HASH_ENTRY_LOCK) { - taosRLockLatch(&pEntry->latch); - } - - p = pEntry->next; - - if (pIter->pHashObj->type == HASH_ENTRY_LOCK) { - taosRUnLockLatch(&pEntry->latch); - } - - return p; - } - - return NULL; -} - size_t taosHashGetMemSize(const SHashObj *pHashObj) { if (pHashObj == NULL) { return 0; @@ -811,3 +678,129 @@ size_t taosHashGetMemSize(const SHashObj *pHashObj) { return (pHashObj->capacity * (sizeof(SHashEntry) + POINTER_BYTES)) + sizeof(SHashNode) * taosHashGetSize(pHashObj) + sizeof(SHashObj); } + +// release the pNode, return next pNode, and lock the current entry +static void *taosHashReleaseNode(SHashObj *pHashObj, void *p, int *slot) { + + SHashNode *pOld = (SHashNode *)GET_HASH_PNODE(p); + SHashNode *prevNode = NULL; + + *slot = HASH_INDEX(pOld->hashVal, pHashObj->capacity); + SHashEntry *pe = pHashObj->hashList[*slot]; + + // lock entry + if (pHashObj->type == HASH_ENTRY_LOCK) { + taosWLockLatch(&pe->latch); + } + + SHashNode *pNode = pe->next; + + while (pNode) { + if (pNode == pOld) + break; + + prevNode = pNode; + pNode = pNode->next; + } + + if (pNode) { + pNode = pNode->next; + while (pNode) { + if (pNode->removed == 0) break; + pNode = pNode->next; + } + + pOld->count--; + if (pOld->count <=0) { + if (prevNode) { + prevNode->next = pOld->next; + } else { + pe->next = pOld->next; + } + + pe->num--; + atomic_sub_fetch_64(&pHashObj->size, 1); + FREE_HASH_NODE(pHashObj, pOld); + } + } else { + uError("pNode:%p data:%p is not there!!!", pNode, p); + } + + return pNode; +} + +void *taosHashIterate(SHashObj *pHashObj, void *p) { + if (pHashObj == NULL) return NULL; + + int slot = 0; + char *data = NULL; + + // only add the read lock to disable the resize process + __rd_lock(&pHashObj->lock, pHashObj->type); + + SHashNode *pNode = NULL; + if (p) { + pNode = taosHashReleaseNode(pHashObj, p, &slot); + if (pNode == NULL) { + SHashEntry *pe = pHashObj->hashList[slot]; + if (pHashObj->type == HASH_ENTRY_LOCK) { + taosWUnLockLatch(&pe->latch); + } + + slot = slot + 1; + } + } + + if (pNode == NULL) { + for (; slot < pHashObj->capacity; ++slot) { + SHashEntry *pe = pHashObj->hashList[slot]; + + // lock entry + if (pHashObj->type == HASH_ENTRY_LOCK) { + taosWLockLatch(&pe->latch); + } + + pNode = pe->next; + while (pNode) { + if (pNode->removed == 0) break; + pNode = pNode->next; + } + + if (pNode) break; + + if (pHashObj->type == HASH_ENTRY_LOCK) { + taosWUnLockLatch(&pe->latch); + } + } + } + + if (pNode) { + SHashEntry *pe = pHashObj->hashList[slot]; + pNode->count++; + data = GET_HASH_NODE_DATA(pNode); + if (pHashObj->type == HASH_ENTRY_LOCK) { + taosWUnLockLatch(&pe->latch); + } + } + + __rd_unlock(&pHashObj->lock, pHashObj->type); + return data; + +} + +void taosHashCancelIterate(SHashObj *pHashObj, void *p) { + if (pHashObj == NULL || p == NULL) return; + + // only add the read lock to disable the resize process + __rd_lock(&pHashObj->lock, pHashObj->type); + + int slot; + taosHashReleaseNode(pHashObj, p, &slot); + + SHashEntry *pe = pHashObj->hashList[slot]; + if (pHashObj->type == HASH_ENTRY_LOCK) { + taosWUnLockLatch(&pe->latch); + } + + __rd_unlock(&pHashObj->lock, pHashObj->type); +} diff --git a/src/util/src/tfile.c b/src/util/src/tfile.c index 27ba30fe81..1fa6a3d096 100644 --- a/src/util/src/tfile.c +++ b/src/util/src/tfile.c @@ -13,6 +13,7 @@ * along with this program. If not, see . */ +#define _DEFAULT_SOURCE #include "os.h" #include "taoserror.h" #include "tulog.h" @@ -21,40 +22,52 @@ static int32_t tsFileRsetId = -1; -static void taosCloseFile(void *p) { +static void tfCloseFile(void *p) { close((int32_t)(uintptr_t)p); } -int32_t tfinit() { - tsFileRsetId = taosOpenRef(2000, taosCloseFile); - return tsFileRsetId; +int32_t tfInit() { + tsFileRsetId = taosOpenRef(2000, tfCloseFile); + if (tsFileRsetId > 0) { + return 0; + } else { + return -1; + } } -void tfcleanup() { +void tfCleanup() { if (tsFileRsetId >= 0) taosCloseRef(tsFileRsetId); tsFileRsetId = -1; } -int64_t tfopen(const char *pathname, int32_t flags) { - int32_t fd = open(pathname, flags); - +static int64_t tfOpenImp(int32_t fd) { if (fd < 0) { terrno = TAOS_SYSTEM_ERROR(errno); return -1; - } + } - void *p = (void *)(int64_t)fd; + void * p = (void *)(int64_t)fd; int64_t rid = taosAddRef(tsFileRsetId, p); if (rid < 0) close(fd); return rid; } -int64_t tfclose(int64_t tfd) { +int64_t tfOpen(const char *pathname, int32_t flags) { + int32_t fd = open(pathname, flags); + return tfOpenImp(fd); +} + +int64_t tfOpenM(const char *pathname, int32_t flags, mode_t mode) { + int32_t fd = open(pathname, flags, mode); + return tfOpenImp(fd); +} + +int64_t tfClose(int64_t tfd) { return taosRemoveRef(tsFileRsetId, tfd); } -int64_t tfwrite(int64_t tfd, void *buf, int64_t count) { +int64_t tfWrite(int64_t tfd, void *buf, int64_t count) { void *p = taosAcquireRef(tsFileRsetId, tfd); if (p == NULL) return -1; @@ -67,7 +80,7 @@ int64_t tfwrite(int64_t tfd, void *buf, int64_t count) { return ret; } -int64_t tfread(int64_t tfd, void *buf, int64_t count) { +int64_t tfRead(int64_t tfd, void *buf, int64_t count) { void *p = taosAcquireRef(tsFileRsetId, tfd); if (p == NULL) return -1; @@ -79,3 +92,32 @@ int64_t tfread(int64_t tfd, void *buf, int64_t count) { taosReleaseRef(tsFileRsetId, tfd); return ret; } + +int64_t tfFsync(int64_t tfd) { + void *p = taosAcquireRef(tsFileRsetId, tfd); + if (p == NULL) return -1; + + int32_t fd = (int32_t)(uintptr_t)p; + return fsync(fd); +} + +bool tfValid(int64_t tfd) { + void *p = taosAcquireRef(tsFileRsetId, tfd); + return p != NULL; +} + +int64_t tfLseek(int64_t tfd, int64_t offset, int32_t whence) { + void *p = taosAcquireRef(tsFileRsetId, tfd); + if (p == NULL) return -1; + + int32_t fd = (int32_t)(uintptr_t)p; + return taosLSeek(fd, offset, whence); +} + +int32_t tfFtruncate(int64_t tfd, int64_t length) { + void *p = taosAcquireRef(tsFileRsetId, tfd); + if (p == NULL) return -1; + + int32_t fd = (int32_t)(uintptr_t)p; + return taosFtruncate(fd, length); +} diff --git a/src/util/src/tkvstore.c b/src/util/src/tkvstore.c index 31641ac9a7..2b1d13c78b 100644 --- a/src/util/src/tkvstore.c +++ b/src/util/src/tkvstore.c @@ -529,7 +529,7 @@ static int tdRestoreKVStore(SKVStore *pStore) { void * buf = NULL; int64_t maxBufSize = 0; SKVRecord rInfo = {0}; - SHashMutableIterator *pIter = NULL; + SKVRecord *pRecord = NULL; ASSERT(TD_KVSTORE_HEADER_SIZE == lseek(pStore->fd, 0, SEEK_CUR)); ASSERT(pStore->info.size == TD_KVSTORE_HEADER_SIZE); @@ -582,16 +582,8 @@ static int tdRestoreKVStore(SKVStore *pStore) { goto _err; } - pIter = taosHashCreateIter(pStore->map); - if (pIter == NULL) { - uError("failed to create hash iter while opening KV store %s", pStore->fname); - terrno = TSDB_CODE_COM_OUT_OF_MEMORY; - goto _err; - } - - while (taosHashIterNext(pIter)) { - SKVRecord *pRecord = taosHashIterGet(pIter); - + pRecord = taosHashIterate(pStore->map, NULL); + while (pRecord) { if (lseek(pStore->fd, (off_t)(pRecord->offset + sizeof(SKVRecord)), SEEK_SET) < 0) { uError("failed to lseek file %s since %s, offset %" PRId64, pStore->fname, strerror(errno), pRecord->offset); terrno = TAOS_SYSTEM_ERROR(errno); @@ -613,16 +605,17 @@ static int tdRestoreKVStore(SKVStore *pStore) { goto _err; } } + + pRecord = taosHashIterate(pStore->map, pRecord); } if (pStore->aFunc) (*pStore->aFunc)(pStore->appH); - taosHashDestroyIter(pIter); tfree(buf); return 0; _err: - taosHashDestroyIter(pIter); + taosHashCancelIterate(pStore->map, pRecord); tfree(buf); return -1; } diff --git a/src/util/src/tqueue.c b/src/util/src/tqueue.c index d3c17e7626..143ce7c474 100644 --- a/src/util/src/tqueue.c +++ b/src/util/src/tqueue.c @@ -79,15 +79,12 @@ void taosCloseQueue(taos_queue param) { if (queue->qset) taosRemoveFromQset(qset, queue); - pthread_mutex_lock(&queue->mutex); - while (pNode) { pTemp = pNode; pNode = pNode->next; free (pTemp); } - pthread_mutex_unlock(&queue->mutex); pthread_mutex_destroy(&queue->mutex); free(queue); diff --git a/src/util/src/tskiplist.c b/src/util/src/tskiplist.c index a36f7f0261..20bec16954 100644 --- a/src/util/src/tskiplist.c +++ b/src/util/src/tskiplist.c @@ -291,7 +291,10 @@ bool tSkipListIterNext(SSkipListIterator *iter) { iter->next = SL_NODE_GET_FORWARD_POINTER(iter->cur, 0); iter->step++; } else { - if (iter->cur == pSkipList->pHead) return false; + if (iter->cur == pSkipList->pHead) { + tSkipListUnlock(pSkipList); + return false; + } iter->cur = SL_NODE_GET_BACKWARD_POINTER(iter->cur, 0); // a new node is inserted into between iter->cur and iter->next, ignore it diff --git a/src/util/src/tsocket.c b/src/util/src/tsocket.c index 1be79b7bbd..2543c81708 100644 --- a/src/util/src/tsocket.c +++ b/src/util/src/tsocket.c @@ -107,7 +107,7 @@ int32_t taosWriteMsg(SOCKET fd, void *buf, int32_t nbytes) { while (nleft > 0) { nwritten = (int32_t)taosWriteSocket(fd, (char *)ptr, (size_t)nleft); if (nwritten <= 0) { - if (errno == EINTR || errno == EAGAIN || errno == EWOULDBLOCK) + if (errno == EINTR /* || errno == EAGAIN || errno == EWOULDBLOCK */) continue; else return -1; @@ -133,7 +133,7 @@ int32_t taosReadMsg(SOCKET fd, void *buf, int32_t nbytes) { if (nread == 0) { break; } else if (nread < 0) { - if (errno == EINTR || errno == EAGAIN || errno == EWOULDBLOCK) { + if (errno == EINTR/* || errno == EAGAIN || errno == EWOULDBLOCK*/) { continue; } else { return -1; diff --git a/src/vnode/inc/vnodeInt.h b/src/vnode/inc/vnodeInt.h index 021831a644..e468c2e83e 100644 --- a/src/vnode/inc/vnodeInt.h +++ b/src/vnode/inc/vnodeInt.h @@ -39,11 +39,12 @@ typedef struct { int32_t refCount; // reference count int32_t queuedWMsg; int32_t queuedRMsg; - int32_t delayMs; + int32_t flowctrlLevel; int8_t status; int8_t role; int8_t accessState; int8_t isFull; + int8_t isCommiting; uint64_t version; // current version uint64_t fversion; // version on saved data file void *wqueue; diff --git a/src/vnode/src/vnodeMain.c b/src/vnode/src/vnodeMain.c index 3f20efb150..3d3abde3c7 100644 --- a/src/vnode/src/vnodeMain.c +++ b/src/vnode/src/vnodeMain.c @@ -30,19 +30,22 @@ static SHashObj*tsVnodesHash; static void vnodeCleanUp(SVnodeObj *pVnode); -static int vnodeProcessTsdbStatus(void *arg, int status, int eno); -static uint32_t vnodeGetFileInfo(void *ahandle, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion); -static int vnodeGetWalInfo(void *ahandle, char *fileName, int64_t *fileId); -static void vnodeNotifyRole(void *ahandle, int8_t role); -static void vnodeCtrlFlow(void *handle, int32_t mseconds); -static int vnodeNotifyFileSynced(void *ahandle, uint64_t fversion); +static int32_t vnodeProcessTsdbStatus(void *arg, int32_t status, int32_t eno); +static uint32_t vnodeGetFileInfo(int32_t vgId, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion); +static int32_t vnodeGetWalInfo(int32_t vgId, char *fileName, int64_t *fileId); +static void vnodeNotifyRole(int32_t vgId, int8_t role); +static void vnodeCtrlFlow(int32_t vgId, int32_t level); +static int32_t vnodeNotifyFileSynced(int32_t vgId, uint64_t fversion); +static void vnodeConfirmForard(int32_t vgId, void *wparam, int32_t code); +static int32_t vnodeWriteToCache(int32_t vgId, void *wparam, int32_t qtype, void *rparam); +static int32_t vnodeGetFileVersion(int32_t vgId, uint64_t *fver); #ifndef _SYNC int64_t syncStart(const SSyncInfo *info) { return NULL; } -int32_t syncForwardToPeer(int64_t rid, void *pHead, void *mhandle, int qtype) { return 0; } +int32_t syncForwardToPeer(int64_t rid, void *pHead, void *mhandle, int32_t qtype) { return 0; } void syncStop(int64_t rid) {} -int32_t syncReconfig(int64_t rid, const SSyncCfg * cfg) { return 0; } -int syncGetNodesRole(int64_t rid, SNodesRole * cfg) { return 0; } +int32_t syncReconfig(int64_t rid, const SSyncCfg *cfg) { return 0; } +int32_t syncGetNodesRole(int64_t rid, SNodesRole *cfg) { return 0; } void syncConfirmForward(int64_t rid, uint64_t version, int32_t code) {} #endif @@ -55,13 +58,13 @@ char* vnodeStatus[] = { }; int32_t vnodeInitResources() { - int code = syncInit(); + int32_t code = syncInit(); if (code != 0) return code; vnodeInitWriteFp(); vnodeInitReadFp(); - tsVnodesHash = taosHashInit(TSDB_MIN_VNODES, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, true); + tsVnodesHash = taosHashInit(TSDB_MIN_VNODES, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_ENTRY_LOCK); if (tsVnodesHash == NULL) { vError("failed to init vnode list"); return TSDB_CODE_VND_OUT_OF_MEMORY; @@ -173,8 +176,8 @@ int32_t vnodeDrop(int32_t vgId) { return TSDB_CODE_SUCCESS; } -int32_t vnodeAlter(void *param, SCreateVnodeMsg *pVnodeCfg) { - SVnodeObj *pVnode = param; +int32_t vnodeAlter(void *vparam, SCreateVnodeMsg *pVnodeCfg) { + SVnodeObj *pVnode = vparam; // vnode in non-ready state and still needs to return success instead of TSDB_CODE_VND_INVALID_STATUS // cfgVersion can be corrected by status msg @@ -325,19 +328,32 @@ int32_t vnodeOpen(int32_t vnode, char *rootDir) { walRemoveAllOldFiles(pVnode->wal); walRenew(pVnode->wal); + pVnode->qMgmt = qOpenQueryMgmt(pVnode->vgId); + if (pVnode->qMgmt == NULL) { + vnodeCleanUp(pVnode); + return terrno; + } + + pVnode->events = NULL; + + vDebug("vgId:%d, vnode is opened in %s, pVnode:%p", pVnode->vgId, rootDir, pVnode); + tsdbIncCommitRef(pVnode->vgId); + + taosHashPut(tsVnodesHash, &pVnode->vgId, sizeof(int32_t), &pVnode, sizeof(SVnodeObj *)); + SSyncInfo syncInfo; syncInfo.vgId = pVnode->vgId; syncInfo.version = pVnode->version; syncInfo.syncCfg = pVnode->syncCfg; sprintf(syncInfo.path, "%s", rootDir); - syncInfo.ahandle = pVnode; syncInfo.getWalInfo = vnodeGetWalInfo; syncInfo.getFileInfo = vnodeGetFileInfo; - syncInfo.writeToCache = vnodeWriteToWQueue; - syncInfo.confirmForward = dnodeSendRpcVWriteRsp; + syncInfo.writeToCache = vnodeWriteToCache; + syncInfo.confirmForward = vnodeConfirmForard; syncInfo.notifyRole = vnodeNotifyRole; syncInfo.notifyFlowCtrl = vnodeCtrlFlow; syncInfo.notifyFileSynced = vnodeNotifyFileSynced; + syncInfo.getFileVersion = vnodeGetFileVersion; pVnode->sync = syncStart(&syncInfo); #ifndef _SYNC @@ -346,24 +362,13 @@ int32_t vnodeOpen(int32_t vnode, char *rootDir) { if (pVnode->sync <= 0) { vError("vgId:%d, failed to open sync module, replica:%d reason:%s", pVnode->vgId, pVnode->syncCfg.replica, tstrerror(terrno)); + vnodeRelease(pVnode); vnodeCleanUp(pVnode); return terrno; } -#endif +#endif - pVnode->qMgmt = qOpenQueryMgmt(pVnode->vgId); - if (pVnode->qMgmt == NULL) { - vnodeCleanUp(pVnode); - return terrno; - } - - pVnode->events = NULL; pVnode->status = TAOS_VN_STATUS_READY; - vDebug("vgId:%d, vnode is opened in %s, pVnode:%p", pVnode->vgId, rootDir, pVnode); - - tsdbIncCommitRef(pVnode->vgId); - taosHashPut(tsVnodesHash, (const char *)&pVnode->vgId, sizeof(int32_t), (char *)(&pVnode), sizeof(SVnodeObj *)); - return TSDB_CODE_SUCCESS; } @@ -389,7 +394,7 @@ void vnodeRelease(void *vparam) { assert(refCount >= 0); if (refCount > 0) { - if (pVnode->status == TAOS_VN_STATUS_RESET && refCount == 2) { + if (pVnode->status == TAOS_VN_STATUS_RESET && refCount <= 3) { tsem_post(&pVnode->sem); } return; @@ -517,11 +522,10 @@ static void vnodeBuildVloadMsg(SVnodeObj *pVnode, SStatusMsg *pStatus) { } int32_t vnodeGetVnodeList(int32_t vnodeList[], int32_t *numOfVnodes) { - SHashMutableIterator *pIter = taosHashCreateIter(tsVnodesHash); - while (taosHashIterNext(pIter)) { - SVnodeObj **pVnode = taosHashIterGet(pIter); - if (pVnode == NULL) continue; - if (*pVnode == NULL) continue; + void *pIter = taosHashIterate(tsVnodesHash, NULL); + while (pIter) { + SVnodeObj **pVnode = pIter; + if (*pVnode) { (*numOfVnodes)++; if (*numOfVnodes >= TSDB_MAX_VNODES) { @@ -530,25 +534,25 @@ int32_t vnodeGetVnodeList(int32_t vnodeList[], int32_t *numOfVnodes) { } else { vnodeList[*numOfVnodes - 1] = (*pVnode)->vgId; } - } - taosHashDestroyIter(pIter); + } + + pIter = taosHashIterate(tsVnodesHash, pIter); + } return TSDB_CODE_SUCCESS; } void vnodeBuildStatusMsg(void *param) { SStatusMsg *pStatus = param; - SHashMutableIterator *pIter = taosHashCreateIter(tsVnodesHash); - while (taosHashIterNext(pIter)) { - SVnodeObj **pVnode = taosHashIterGet(pIter); - if (pVnode == NULL) continue; - if (*pVnode == NULL) continue; - - vnodeBuildVloadMsg(*pVnode, pStatus); + void *pIter = taosHashIterate(tsVnodesHash, NULL); + while (pIter) { + SVnodeObj **pVnode = pIter; + if (*pVnode) { + vnodeBuildVloadMsg(*pVnode, pStatus); + } + pIter = taosHashIterate(tsVnodesHash, pIter); } - - taosHashDestroyIter(pIter); } void vnodeSetAccess(SVgroupAccess *pAccess, int32_t numOfVnodes) { @@ -567,11 +571,11 @@ void vnodeSetAccess(SVgroupAccess *pAccess, int32_t numOfVnodes) { static void vnodeCleanUp(SVnodeObj *pVnode) { // remove from hash, so new messages wont be consumed - taosHashRemove(tsVnodesHash, (const char *)&pVnode->vgId, sizeof(int32_t)); + taosHashRemove(tsVnodesHash, &pVnode->vgId, sizeof(int32_t)); if (pVnode->status != TAOS_VN_STATUS_INIT) { // it may be in updateing or reset state, then it shall wait - int i = 0; + int32_t i = 0; while (atomic_val_compare_exchange_8(&pVnode->status, TAOS_VN_STATUS_READY, TAOS_VN_STATUS_CLOSING) != TAOS_VN_STATUS_READY) { if (++i % 1000 == 0) { @@ -594,51 +598,75 @@ static void vnodeCleanUp(SVnodeObj *pVnode) { vnodeRelease(pVnode); } -// TODO: this is a simple implement -static int vnodeProcessTsdbStatus(void *arg, int status, int eno) { +static int32_t vnodeProcessTsdbStatus(void *arg, int32_t status, int32_t eno) { SVnodeObj *pVnode = arg; if (eno != TSDB_CODE_SUCCESS) { vError("vgId:%d, failed to commit since %s, fver:%" PRIu64 " vver:%" PRIu64, pVnode->vgId, tstrerror(eno), pVnode->fversion, pVnode->version); + pVnode->isCommiting = 0; pVnode->isFull = 1; return 0; } if (status == TSDB_STATUS_COMMIT_START) { + pVnode->isCommiting = 1; pVnode->fversion = pVnode->version; vDebug("vgId:%d, start commit, fver:%" PRIu64 " vver:%" PRIu64, pVnode->vgId, pVnode->fversion, pVnode->version); - if (pVnode->status == TAOS_VN_STATUS_INIT) { - return 0; - } else { + if (pVnode->status != TAOS_VN_STATUS_INIT) { return walRenew(pVnode->wal); } + return 0; } if (status == TSDB_STATUS_COMMIT_OVER) { vDebug("vgId:%d, commit over, fver:%" PRIu64 " vver:%" PRIu64, pVnode->vgId, pVnode->fversion, pVnode->version); + pVnode->isCommiting = 0; pVnode->isFull = 0; - walRemoveOneOldFile(pVnode->wal); + if (pVnode->status != TAOS_VN_STATUS_INIT) { + walRemoveOneOldFile(pVnode->wal); + } return vnodeSaveVersion(pVnode); } return 0; } -static uint32_t vnodeGetFileInfo(void *ahandle, char *name, uint32_t *index, uint32_t eindex, int64_t *size, +static uint32_t vnodeGetFileInfo(int32_t vgId, char *name, uint32_t *index, uint32_t eindex, int64_t *size, uint64_t *fversion) { - SVnodeObj *pVnode = ahandle; + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vError("vgId:%d, vnode not found while get file info", vgId); + return 0; + } + *fversion = pVnode->fversion; - return tsdbGetFileInfo(pVnode->tsdb, name, index, eindex, size); + uint32_t ret = tsdbGetFileInfo(pVnode->tsdb, name, index, eindex, size); + + vnodeRelease(pVnode); + return ret; } -static int vnodeGetWalInfo(void *ahandle, char *fileName, int64_t *fileId) { - SVnodeObj *pVnode = ahandle; - return walGetWalFile(pVnode->wal, fileName, fileId); +static int32_t vnodeGetWalInfo(int32_t vgId, char *fileName, int64_t *fileId) { + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vError("vgId:%d, vnode not found while get wal info", vgId); + return -1; + } + + int32_t code = walGetWalFile(pVnode->wal, fileName, fileId); + + vnodeRelease(pVnode); + return code; } -static void vnodeNotifyRole(void *ahandle, int8_t role) { - SVnodeObj *pVnode = ahandle; +static void vnodeNotifyRole(int32_t vgId, int8_t role) { + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vTrace("vgId:%d, vnode not found while notify role", vgId); + return; + } + vInfo("vgId:%d, sync role changed from %s to %s", pVnode->vgId, syncRole[pVnode->role], syncRole[role]); pVnode->role = role; dnodeSendStatusMsgToMnode(); @@ -648,17 +676,26 @@ static void vnodeNotifyRole(void *ahandle, int8_t role) { } else { cqStop(pVnode->cq); } + + vnodeRelease(pVnode); } -static void vnodeCtrlFlow(void *ahandle, int32_t mseconds) { - SVnodeObj *pVnode = ahandle; - if (pVnode->delayMs != mseconds) { - pVnode->delayMs = mseconds; - vDebug("vgId:%d, sync flow control, mseconds:%d", pVnode->vgId, mseconds); +static void vnodeCtrlFlow(int32_t vgId, int32_t level) { + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vTrace("vgId:%d, vnode not found while flow ctrl", vgId); + return; } + + if (pVnode->flowctrlLevel != level) { + vDebug("vgId:%d, set flowctrl level from %d to %d", pVnode->vgId, pVnode->flowctrlLevel, level); + pVnode->flowctrlLevel = level; + } + + vnodeRelease(pVnode); } -static int vnodeResetTsdb(SVnodeObj *pVnode) { +static int32_t vnodeResetTsdb(SVnodeObj *pVnode) { char rootDir[128] = "\0"; sprintf(rootDir, "%s/tsdb", pVnode->rootDir); @@ -668,11 +705,11 @@ static int vnodeResetTsdb(SVnodeObj *pVnode) { void *tsdb = pVnode->tsdb; pVnode->tsdb = NULL; - + // acquire vnode int32_t refCount = atomic_add_fetch_32(&pVnode->refCount, 1); - if (refCount > 2) { + if (refCount > 3) { tsem_wait(&pVnode->sem); } @@ -692,14 +729,63 @@ static int vnodeResetTsdb(SVnodeObj *pVnode) { return 0; } -static int vnodeNotifyFileSynced(void *ahandle, uint64_t fversion) { - SVnodeObj *pVnode = ahandle; +static int32_t vnodeNotifyFileSynced(int32_t vgId, uint64_t fversion) { + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vError("vgId:%d, vnode not found while notify file synced", vgId); + return 0; + } pVnode->fversion = fversion; pVnode->version = fversion; vnodeSaveVersion(pVnode); - vDebug("vgId:%d, data file is synced, fver:%" PRIu64 " vver:%" PRIu64, pVnode->vgId, pVnode->fversion, - pVnode->version); - return vnodeResetTsdb(pVnode); + vDebug("vgId:%d, data file is synced, fver:%" PRIu64 " vver:%" PRIu64, vgId, fversion, fversion); + int32_t code = vnodeResetTsdb(pVnode); + + vnodeRelease(pVnode); + return code; +} + +void vnodeConfirmForard(int32_t vgId, void *wparam, int32_t code) { + void *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vError("vgId:%d, vnode not found while confirm forward", vgId); + return; + } + + dnodeSendRpcVWriteRsp(pVnode, wparam, code); + vnodeRelease(pVnode); +} + +static int32_t vnodeWriteToCache(int32_t vgId, void *wparam, int32_t qtype, void *rparam) { + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vError("vgId:%d, vnode not found while write to cache", vgId); + return TSDB_CODE_VND_INVALID_VGROUP_ID; + } + + int32_t code = vnodeWriteToWQueue(pVnode, wparam, qtype, rparam); + + vnodeRelease(pVnode); + return code; +} + +static int32_t vnodeGetFileVersion(int32_t vgId, uint64_t *fver) { + SVnodeObj *pVnode = vnodeAcquire(vgId); + if (pVnode == NULL) { + vError("vgId:%d, vnode not found while write to cache", vgId); + return -1; + } + + int32_t code = 0; + if (pVnode->isCommiting) { + vDebug("vgId:%d, vnode is commiting while get file version", vgId); + code = -1; + } else { + *fver = pVnode->fversion; + } + + vnodeRelease(pVnode); + return code; } diff --git a/src/vnode/src/vnodeRead.c b/src/vnode/src/vnodeRead.c index e10d62a0c9..5ef79cfbf0 100644 --- a/src/vnode/src/vnodeRead.c +++ b/src/vnode/src/vnodeRead.c @@ -53,8 +53,7 @@ int32_t vnodeProcessRead(void *vparam, SVReadMsg *pRead) { return (*vnodeProcessReadMsgFp[msgType])(pVnode, pRead); } -static int32_t vnodeCheckRead(void *vparam) { - SVnodeObj *pVnode = vparam; +static int32_t vnodeCheckRead(SVnodeObj *pVnode) { if (pVnode->status != TAOS_VN_STATUS_READY) { vDebug("vgId:%d, vnode status is %s, refCount:%d pVnode:%p", pVnode->vgId, vnodeStatus[pVnode->status], pVnode->refCount, pVnode); diff --git a/src/vnode/src/vnodeWrite.c b/src/vnode/src/vnodeWrite.c index 2d2be602ed..268d1fb53b 100644 --- a/src/vnode/src/vnodeWrite.c +++ b/src/vnode/src/vnodeWrite.c @@ -17,19 +17,23 @@ #include "os.h" #include "taosmsg.h" #include "taoserror.h" +#include "tglobal.h" #include "tqueue.h" #include "trpc.h" #include "tsdb.h" #include "twal.h" #include "tsync.h" +#include "ttimer.h" #include "tdataformat.h" #include "vnode.h" #include "vnodeInt.h" #include "syncInt.h" #include "tcq.h" +#include "dnode.h" #define MAX_QUEUED_MSG_NUM 10000 +extern void * tsDnodeTmr; static int32_t (*vnodeProcessWriteMsgFp[TSDB_MSG_TYPE_MAX])(SVnodeObj *, void *pCont, SRspRet *); static int32_t vnodeProcessSubmitMsg(SVnodeObj *pVnode, void *pCont, SRspRet *); static int32_t vnodeProcessCreateTableMsg(SVnodeObj *pVnode, void *pCont, SRspRet *); @@ -37,6 +41,7 @@ static int32_t vnodeProcessDropTableMsg(SVnodeObj *pVnode, void *pCont, SRspRet static int32_t vnodeProcessAlterTableMsg(SVnodeObj *pVnode, void *pCont, SRspRet *); static int32_t vnodeProcessDropStableMsg(SVnodeObj *pVnode, void *pCont, SRspRet *); static int32_t vnodeProcessUpdateTagValMsg(SVnodeObj *pVnode, void *pCont, SRspRet *); +static int32_t vnodePerformFlowCtrl(SVWriteMsg *pWrite); void vnodeInitWriteFp(void) { vnodeProcessWriteMsgFp[TSDB_MSG_TYPE_SUBMIT] = vnodeProcessSubmitMsg; @@ -48,10 +53,10 @@ void vnodeInitWriteFp(void) { } int32_t vnodeProcessWrite(void *vparam, void *wparam, int32_t qtype, void *rparam) { - int32_t code = 0; - SVnodeObj * pVnode = vparam; - SWalHead * pHead = wparam; - SRspRet * pRspRet = rparam; + int32_t code = 0; + SVnodeObj *pVnode = vparam; + SWalHead * pHead = wparam; + SRspRet * pRspRet = rparam; if (vnodeProcessWriteMsgFp[pHead->msgType] == NULL) { vError("vgId:%d, msg:%s not processed since no handle, qtype:%s hver:%" PRIu64, pVnode->vgId, @@ -77,8 +82,6 @@ int32_t vnodeProcessWrite(void *vparam, void *wparam, int32_t qtype, void *rpara // assign version pHead->version = pVnode->version + 1; - if (pVnode->delayMs) taosMsleep(pVnode->delayMs); - } else { // from wal or forward // for data from WAL or forward, version may be smaller if (pHead->version <= pVnode->version) return 0; @@ -218,9 +221,10 @@ static int32_t vnodeProcessUpdateTagValMsg(SVnodeObj *pVnode, void *pCont, SRspR int32_t vnodeWriteToWQueue(void *vparam, void *wparam, int32_t qtype, void *rparam) { SVnodeObj *pVnode = vparam; SWalHead * pHead = wparam; + int32_t code = 0; if (qtype == TAOS_QTYPE_RPC) { - int32_t code = vnodeCheckWrite(pVnode); + code = vnodeCheckWrite(pVnode); if (code != TSDB_CODE_SUCCESS) return code; } @@ -237,11 +241,12 @@ int32_t vnodeWriteToWQueue(void *vparam, void *wparam, int32_t qtype, void *rpar if (rparam != NULL) { SRpcMsg *pRpcMsg = rparam; - pWrite->rpcHandle = pRpcMsg->handle; - pWrite->rpcAhandle = pRpcMsg->ahandle; + pWrite->rpcMsg = *pRpcMsg; } memcpy(pWrite->pHead, pHead, sizeof(SWalHead) + pHead->len); + pWrite->pVnode = pVnode; + pWrite->qtype = qtype; atomic_add_fetch_32(&pVnode->refCount, 1); @@ -251,6 +256,9 @@ int32_t vnodeWriteToWQueue(void *vparam, void *wparam, int32_t qtype, void *rpar taosMsleep(1); } + code = vnodePerformFlowCtrl(pWrite); + if (code != 0) return 0; + vTrace("vgId:%d, write into vwqueue, refCount:%d queued:%d", pVnode->vgId, pVnode->refCount, pVnode->queuedWMsg); taosWriteQitem(pVnode->wqueue, qtype, pWrite); @@ -260,9 +268,52 @@ int32_t vnodeWriteToWQueue(void *vparam, void *wparam, int32_t qtype, void *rpar void vnodeFreeFromWQueue(void *vparam, SVWriteMsg *pWrite) { SVnodeObj *pVnode = vparam; - atomic_sub_fetch_32(&pVnode->queuedWMsg, 1); - vTrace("vgId:%d, free from vwqueue, refCount:%d queued:%d", pVnode->vgId, pVnode->refCount, pVnode->queuedWMsg); + int32_t queued = atomic_sub_fetch_32(&pVnode->queuedWMsg, 1); + vTrace("vgId:%d, msg:%p, app:%p, free from vwqueue, queued:%d", pVnode->vgId, pWrite, pWrite->rpcMsg.ahandle, queued); taosFreeQitem(pWrite); vnodeRelease(pVnode); } + +static void vnodeFlowCtrlMsgToWQueue(void *param, void *tmrId) { + SVWriteMsg *pWrite = param; + SVnodeObj * pVnode = pWrite->pVnode; + int32_t code = TSDB_CODE_VND_SYNCING; + + pWrite->processedCount++; + if (pWrite->processedCount > 100) { + vError("vgId:%d, msg:%p, failed to process since %s, retry:%d", pVnode->vgId, pWrite, tstrerror(code), + pWrite->processedCount); + pWrite->processedCount = 1; + dnodeSendRpcVWriteRsp(pWrite->pVnode, pWrite, code); + } else { + code = vnodePerformFlowCtrl(pWrite); + if (code == 0) { + vDebug("vgId:%d, msg:%p, write into vwqueue after flowctrl, retry:%d", pVnode->vgId, pWrite, + pWrite->processedCount); + pWrite->processedCount = 0; + taosWriteQitem(pVnode->wqueue, pWrite->qtype, pWrite); + } + } +} + +static int32_t vnodePerformFlowCtrl(SVWriteMsg *pWrite) { + SVnodeObj *pVnode = pWrite->pVnode; + if (pVnode->flowctrlLevel <= 0) return 0; + if (pWrite->qtype != TAOS_QTYPE_RPC) return 0; + + if (tsFlowCtrl == 0) { + int32_t ms = pow(2, pVnode->flowctrlLevel + 2); + if (ms > 100) ms = 100; + vTrace("vgId:%d, msg:%p, app:%p, perform flowctrl for %d ms", pVnode->vgId, pWrite, pWrite->rpcMsg.ahandle, ms); + taosMsleep(ms); + return 0; + } else { + void *unUsed = NULL; + taosTmrReset(vnodeFlowCtrlMsgToWQueue, 100, pWrite, tsDnodeTmr, &unUsed); + + vTrace("vgId:%d, msg:%p, app:%p, perform flowctrl, retry:%d", pVnode->vgId, pWrite, pWrite->rpcMsg.ahandle, + pWrite->processedCount); + return TSDB_CODE_VND_ACTION_IN_PROGRESS; + } +} diff --git a/src/wal/inc/walInt.h b/src/wal/inc/walInt.h index b0edabfbd8..06748d885f 100644 --- a/src/wal/inc/walInt.h +++ b/src/wal/inc/walInt.h @@ -44,8 +44,8 @@ typedef struct { uint64_t version; int64_t fileId; int64_t rid; + int64_t tfd; int32_t vgId; - int32_t fd; int32_t keep; int32_t level; int32_t fsyncPeriod; diff --git a/src/wal/src/walMgmt.c b/src/wal/src/walMgmt.c index 36c190be3e..72ea239817 100644 --- a/src/wal/src/walMgmt.c +++ b/src/wal/src/walMgmt.c @@ -17,6 +17,7 @@ #include "os.h" #include "taoserror.h" #include "tref.h" +#include "tfile.h" #include "twal.h" #include "walInt.h" @@ -61,7 +62,7 @@ void *walOpen(char *path, SWalCfg *pCfg) { } pWal->vgId = pCfg->vgId; - pWal->fd = -1; + pWal->tfd = -1; pWal->fileId = -1; pWal->level = pCfg->walLevel; pWal->keep = pCfg->keep; @@ -124,7 +125,7 @@ void walClose(void *handle) { SWal *pWal = handle; pthread_mutex_lock(&pWal->mutex); - taosClose(pWal->fd); + tfClose(pWal->tfd); pthread_mutex_unlock(&pWal->mutex); taosRemoveRef(tsWal.refId, pWal->rid); } @@ -143,7 +144,7 @@ static void walFreeObj(void *wal) { SWal *pWal = wal; wDebug("vgId:%d, wal:%p is freed", pWal->vgId, pWal); - taosClose(pWal->fd); + tfClose(pWal->tfd); pthread_mutex_destroy(&pWal->mutex); tfree(pWal); } @@ -172,7 +173,7 @@ static void walFsyncAll() { while (pWal) { if (walNeedFsync(pWal)) { wTrace("vgId:%d, do fsync, level:%d seq:%d rseq:%d", pWal->vgId, pWal->level, pWal->fsyncSeq, tsWal.seq); - int32_t code = fsync(pWal->fd); + int32_t code = tfFsync(pWal->tfd); if (code != 0) { wError("vgId:%d, file:%s, failed to fsync since %s", pWal->vgId, pWal->name, strerror(code)); } diff --git a/src/wal/src/walWrite.c b/src/wal/src/walWrite.c index 72464d4309..10e1b4dd61 100644 --- a/src/wal/src/walWrite.c +++ b/src/wal/src/walWrite.c @@ -18,6 +18,7 @@ #include "os.h" #include "taoserror.h" #include "tchecksum.h" +#include "tfile.h" #include "twal.h" #include "walInt.h" @@ -36,8 +37,8 @@ int32_t walRenew(void *handle) { pthread_mutex_lock(&pWal->mutex); - if (pWal->fd >= 0) { - tclose(pWal->fd); + if (tfValid(pWal->tfd)) { + tfClose(pWal->tfd); wDebug("vgId:%d, file:%s, it is closed", pWal->vgId, pWal->name); } @@ -49,9 +50,9 @@ int32_t walRenew(void *handle) { } snprintf(pWal->name, sizeof(pWal->name), "%s/%s%" PRId64, pWal->path, WAL_PREFIX, pWal->fileId); - pWal->fd = open(pWal->name, O_WRONLY | O_CREAT, S_IRWXU | S_IRWXG | S_IRWXO); + pWal->tfd = tfOpenM(pWal->name, O_WRONLY | O_CREAT, S_IRWXU | S_IRWXG | S_IRWXO); - if (pWal->fd < 0) { + if (!tfValid(pWal->tfd)) { code = TAOS_SYSTEM_ERROR(errno); wError("vgId:%d, file:%s, failed to open since %s", pWal->vgId, pWal->name, strerror(errno)); } else { @@ -67,7 +68,7 @@ void walRemoveOneOldFile(void *handle) { SWal *pWal = handle; if (pWal == NULL) return; if (pWal->keep == TAOS_WAL_KEEP) return; - if (pWal->fd <= 0) return; + if (!tfValid(pWal->tfd)) return; pthread_mutex_lock(&pWal->mutex); @@ -113,7 +114,7 @@ int32_t walWrite(void *handle, SWalHead *pHead) { int32_t code = 0; // no wal - if (pWal->fd <= 0) return 0; + if (!tfValid(pWal->tfd)) return 0; if (pWal->level == TAOS_WAL_NOLOG) return 0; if (pHead->version <= pWal->version) return 0; @@ -123,12 +124,12 @@ int32_t walWrite(void *handle, SWalHead *pHead) { pthread_mutex_lock(&pWal->mutex); - if (taosWrite(pWal->fd, pHead, contLen) != contLen) { + if (tfWrite(pWal->tfd, pHead, contLen) != contLen) { code = TAOS_SYSTEM_ERROR(errno); wError("vgId:%d, file:%s, failed to write since %s", pWal->vgId, pWal->name, strerror(errno)); } else { - wTrace("vgId:%d, write wal, fileId:%" PRId64 " fd:%d hver:%" PRId64 " wver:%" PRIu64 " len:%d", pWal->vgId, - pWal->fileId, pWal->fd, pHead->version, pWal->version, pHead->len); + wTrace("vgId:%d, write wal, fileId:%" PRId64 " tfd:%" PRId64 " hver:%" PRId64 " wver:%" PRIu64 " len:%d", pWal->vgId, + pWal->fileId, pWal->tfd, pHead->version, pWal->version, pHead->len); pWal->version = pHead->version; } @@ -141,12 +142,12 @@ int32_t walWrite(void *handle, SWalHead *pHead) { void walFsync(void *handle, bool forceFsync) { SWal *pWal = handle; - if (pWal == NULL || pWal->fd < 0) return; + if (pWal == NULL || !tfValid(pWal->tfd)) return; if (forceFsync || (pWal->level == TAOS_WAL_FSYNC && pWal->fsyncPeriod == 0)) { - wTrace("vgId:%d, file:%s, do fsync", pWal->vgId, pWal->name); - if (fsync(pWal->fd) < 0) { - wError("vgId:%d, file:%s, fsync failed since %s", pWal->vgId, pWal->name, strerror(errno)); + wTrace("vgId:%d, fileId:%" PRId64 ", do fsync", pWal->vgId, pWal->fileId); + if (tfFsync(pWal->tfd) < 0) { + wError("vgId:%d, fileId:%" PRId64 ", fsync failed since %s", pWal->vgId, pWal->fileId, strerror(errno)); } } } @@ -186,8 +187,8 @@ int32_t walRestore(void *handle, void *pVnode, FWalWrite writeFp) { // open the existing WAL file in append mode pWal->fileId = 0; snprintf(pWal->name, sizeof(pWal->name), "%s/%s%" PRId64, pWal->path, WAL_PREFIX, pWal->fileId); - pWal->fd = open(pWal->name, O_WRONLY | O_CREAT | O_APPEND, S_IRWXU | S_IRWXG | S_IRWXO); - if (pWal->fd < 0) { + pWal->tfd = tfOpenM(pWal->name, O_WRONLY | O_CREAT | O_APPEND, S_IRWXU | S_IRWXG | S_IRWXO); + if (!tfValid(pWal->tfd)) { wError("vgId:%d, file:%s, failed to open since %s", pWal->vgId, pWal->name, strerror(errno)); return TAOS_SYSTEM_ERROR(errno); } @@ -211,28 +212,28 @@ int32_t walGetWalFile(void *handle, char *fileName, int64_t *fileId) { code = (*fileId == pWal->fileId) ? 0 : 1; } - wTrace("vgId:%d, get wal file, code:%d curId:%" PRId64 " outId:%" PRId64, pWal->vgId, code, pWal->fileId, *fileId); + wDebug("vgId:%d, get wal file, code:%d curId:%" PRId64 " outId:%" PRId64, pWal->vgId, code, pWal->fileId, *fileId); pthread_mutex_unlock(&(pWal->mutex)); return code; } -static void walFtruncate(SWal *pWal, int32_t fd, int64_t offset) { - taosFtruncate(fd, offset); - fsync(fd); +static void walFtruncate(SWal *pWal, int64_t tfd, int64_t offset) { + tfFtruncate(tfd, offset); + tfFsync(tfd); } -static int32_t walSkipCorruptedRecord(SWal *pWal, SWalHead *pHead, int32_t fd, int64_t *offset) { +static int32_t walSkipCorruptedRecord(SWal *pWal, SWalHead *pHead, int64_t tfd, int64_t *offset) { int64_t pos = *offset; while (1) { pos++; - if (lseek(fd, pos, SEEK_SET) < 0) { + if (tfLseek(tfd, pos, SEEK_SET) < 0) { wError("vgId:%d, failed to seek from corrupted wal file since %s", pWal->vgId, strerror(errno)); return TSDB_CODE_WAL_FILE_CORRUPTED; } - if (taosRead(fd, pHead, sizeof(SWalHead)) <= 0) { + if (tfRead(tfd, pHead, sizeof(SWalHead)) <= 0) { wError("vgId:%d, read to end of corrupted wal file, offset:%" PRId64, pWal->vgId, pos); return TSDB_CODE_WAL_FILE_CORRUPTED; } @@ -259,8 +260,8 @@ static int32_t walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp, ch return TAOS_SYSTEM_ERROR(errno); } - int32_t fd = open(name, O_RDWR); - if (fd < 0) { + int64_t tfd = tfOpen(name, O_RDWR); + if (!tfValid(tfd)) { wError("vgId:%d, file:%s, failed to open for restore since %s", pWal->vgId, name, strerror(errno)); tfree(buffer); return TAOS_SYSTEM_ERROR(errno); @@ -273,7 +274,7 @@ static int32_t walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp, ch SWalHead *pHead = buffer; while (1) { - int32_t ret = taosRead(fd, pHead, sizeof(SWalHead)); + int32_t ret = tfRead(tfd, pHead, sizeof(SWalHead)); if (ret == 0) break; if (ret < 0) { @@ -284,16 +285,16 @@ static int32_t walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp, ch if (ret < sizeof(SWalHead)) { wError("vgId:%d, file:%s, failed to read wal head, ret is %d", pWal->vgId, name, ret); - walFtruncate(pWal, fd, offset); + walFtruncate(pWal, tfd, offset); break; } if (!taosCheckChecksumWhole((uint8_t *)pHead, sizeof(SWalHead))) { wError("vgId:%d, file:%s, wal head cksum is messed up, hver:%" PRIu64 " len:%d offset:%" PRId64, pWal->vgId, name, pHead->version, pHead->len, offset); - code = walSkipCorruptedRecord(pWal, pHead, fd, &offset); + code = walSkipCorruptedRecord(pWal, pHead, tfd, &offset); if (code != TSDB_CODE_SUCCESS) { - walFtruncate(pWal, fd, offset); + walFtruncate(pWal, tfd, offset); break; } } @@ -310,7 +311,7 @@ static int32_t walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp, ch pHead = buffer; } - ret = taosRead(fd, pHead->cont, pHead->len); + ret = tfRead(tfd, pHead->cont, pHead->len); if (ret < 0) { wError("vgId:%d, file:%s, failed to read wal body since %s", pWal->vgId, name, strerror(errno)); code = TAOS_SYSTEM_ERROR(errno); @@ -325,14 +326,14 @@ static int32_t walRestoreWalFile(SWal *pWal, void *pVnode, FWalWrite writeFp, ch offset = offset + sizeof(SWalHead) + pHead->len; - wTrace("vgId:%d, restore wal, fileId:%" PRId64 " hver:%" PRIu64 " wver:%" PRIu64 " len:%d", pWal->vgId, + wDebug("vgId:%d, restore wal, fileId:%" PRId64 " hver:%" PRIu64 " wver:%" PRIu64 " len:%d", pWal->vgId, fileId, pHead->version, pWal->version, pHead->len); pWal->version = pHead->version; (*writeFp)(pVnode, pHead, TAOS_QTYPE_WAL, NULL); } - tclose(fd); + tfClose(tfd); tfree(buffer); return code; diff --git a/tests/examples/c/apitest.c b/tests/examples/c/apitest.c index be60a88ad7..e08d667d6b 100644 --- a/tests/examples/c/apitest.c +++ b/tests/examples/c/apitest.c @@ -79,11 +79,11 @@ static int print_result(TAOS_RES* res, int blockFetch) { if (blockFetch) { int rows = 0; while ((rows = taos_fetch_block(res, &row))) { - for (int i = 0; i < rows; i++) { - char temp[256]; - taos_print_row(temp, row + i, fields, num_fields); - puts(temp); - } + //for (int i = 0; i < rows; i++) { + // char temp[256]; + // taos_print_row(temp, row + i, fields, num_fields); + // puts(temp); + //} nRows += rows; } } else { @@ -498,4 +498,4 @@ int main(int argc, char *argv[]) { taos_close(taos); taos_cleanup(); -} \ No newline at end of file +} diff --git a/tests/examples/c/subscribe.c b/tests/examples/c/subscribe.c index 8368af18f7..cdd8ddaf7f 100644 --- a/tests/examples/c/subscribe.c +++ b/tests/examples/c/subscribe.c @@ -19,10 +19,10 @@ void print_result(TAOS_RES* res, int blockFetch) { if (blockFetch) { nRows = taos_fetch_block(res, &row); - for (int i = 0; i < nRows; i++) { - taos_print_row(buf, row + i, fields, num_fields); - puts(buf); - } + //for (int i = 0; i < nRows; i++) { + // taos_print_row(buf, row + i, fields, num_fields); + // puts(buf); + //} } else { while ((row = taos_fetch_row(res))) { taos_print_row(buf, row, fields, num_fields); diff --git a/tests/pytest/cluster/clusterSetup.py b/tests/pytest/cluster/clusterSetup.py index 36af8ac42e..dbda5657b6 100644 --- a/tests/pytest/cluster/clusterSetup.py +++ b/tests/pytest/cluster/clusterSetup.py @@ -31,6 +31,23 @@ class Node: self.homeDir = homeDir self.conn = Connection("{}@{}".format(username, hostName), connect_kwargs={"password": "{}".format(password)}) + def buildTaosd(self): + try: + self.conn.cd("/root/TDinternal/community") + self.conn.run("git checkout develop") + self.conn.run("git pull") + self.conn.cd("/root/TDinternal") + self.conn.run("git checkout develop") + self.conn.run("git pull") + self.conn.cd("/root/TDinternal/debug") + self.conn.run("cmake ..") + self.conn.run("make") + self.conn.run("make install") + except Exception as e: + print("Build Taosd error for node %d " % self.index) + logging.exception(e) + pass + def startTaosd(self): try: self.conn.run("sudo systemctl start taosd") @@ -50,7 +67,7 @@ class Node: self.conn.run("sudo systemctl restart taosd") except Exception as e: print("Stop Taosd error for node %d " % self.index) - logging.exception(e) + logging.exception(e) def removeTaosd(self): try: @@ -105,9 +122,11 @@ class Node: class Nodes: def __init__(self): - self.node1 = Node(1, 'ubuntu', '192.168.1.52', 'node1', 'tbase125!', '/home/ubuntu') - self.node2 = Node(2, 'ubuntu', '192.168.1.53', 'node2', 'tbase125!', '/home/ubuntu') - self.node3 = Node(3, 'ubuntu', '192.168.1.54', 'node3', 'tbase125!', '/home/ubuntu') + self.node1 = Node(1, 'root', '52.151.60.239', 'node1', 'r', '/root/') + self.node2 = Node(2, 'root', '52.183.32.246', 'node1', 'r', '/root/') + self.node3 = Node(3, 'root', '51.143.46.79', 'node1', 'r', '/root/') + self.node4 = Node(4, 'root', '52.183.2.76', 'node1', 'r', '/root/') + self.node5 = Node(5, 'root', '13.66.225.87', 'node1', 'r', '/root/') def stopAllTaosd(self): self.node1.stopTaosd() diff --git a/tests/pytest/concurrent_inquiry.py b/tests/pytest/concurrent_inquiry.py index 5d1e9a7537..6f4258312d 100644 --- a/tests/pytest/concurrent_inquiry.py +++ b/tests/pytest/concurrent_inquiry.py @@ -146,7 +146,7 @@ class ConcurrentInquiry: col_list=self.stb_stru_list[tbi-1] tag_list=self.stb_tag_list[tbi-1] is_stb=1 - tlist=col_list+tag_list + tlist=col_list+tag_list+['abc'] #增加不存在的域'abc'，是否会引起新bug con_rand=random.randint(0,len(condition_list)) func_rand=random.randint(0,len(func_list)) col_rand=random.randint(0,len(col_list)) diff --git a/tests/pytest/fulltest.sh b/tests/pytest/fulltest.sh index 525fbad6c1..0e3b482e3d 100755 --- a/tests/pytest/fulltest.sh +++ b/tests/pytest/fulltest.sh @@ -158,6 +158,7 @@ python3 ./test.py -f query/bug1471.py python3 ./test.py -f query/bug1874.py python3 ./test.py -f query/bug1875.py python3 ./test.py -f query/bug1876.py +python3 ./test.py -f query/bug2218.py #stream python3 ./test.py -f stream/metric_1.py @@ -205,7 +206,9 @@ python3 test.py -f query/queryInterval.py python3 test.py -f query/queryFillTest.py # tools -python3 test.py -f tools/taosdemo.py +python3 test.py -f tools/taosdemoTest.py +python3 test.py -f tools/taosdumpTest.py +python3 test.py -f tools/lowaTest.py # subscribe python3 test.py -f subscribe/singlemeter.py diff --git a/tests/pytest/insert/restfulInsert.py b/tests/pytest/insert/restfulInsert.py index 9fa1f33a24..e3a963f1d4 100644 --- a/tests/pytest/insert/restfulInsert.py +++ b/tests/pytest/insert/restfulInsert.py @@ -18,10 +18,10 @@ import time import argparse class RestfulInsert: - def __init__(self, host, dbname, threads, tables, records, batchSize, tbNamePerfix, outOfOrder): + def __init__(self, host, startTimestamp, dbname, threads, tables, records, batchSize, tbNamePerfix, outOfOrder): self.header = {'Authorization': 'Basic cm9vdDp0YW9zZGF0YQ=='} self.url = "http://%s:6041/rest/sql" % host - self.ts = 1500000000000 + self.ts = startTimestamp self.dbname = dbname self.numOfThreads = threads self.numOfTables = tables @@ -36,8 +36,10 @@ class RestfulInsert: for i in range(tablesPerThread): tableID = threadID * tablesPerThread name = 'beijing' if tableID % 2 == 0 else 'shanghai' - data = "create table %s.%s%d using %s.meters tags(%d, '%s')" % (self.dbname, self.tableNamePerfix, tableID + i, self.dbname, tableID + i, name) - requests.post(self.url, data, headers = self.header) + data = "create table if not exists %s.%s%d using %s.meters tags(%d, '%s')" % (self.dbname, self.tableNamePerfix, tableID + i, self.dbname, tableID + i, name) + response = requests.post(self.url, data, headers = self.header) + if response.status_code != 200: + print(response.content) def insertData(self, threadID): print("thread %d started" % threadID) @@ -50,14 +52,19 @@ class RestfulInsert: values = [] for k in range(self.batchSize): data += "(%d, %d, %d, %d)" % (start + j * self.batchSize + k, random.randint(1, 100), random.randint(1, 100), random.randint(1, 100)) - requests.post(self.url, data, headers = self.header) + response = requests.post(self.url, data, headers = self.header) + if response.status_code != 200: + print(response.content) def insertUnlimitedData(self, threadID): print("thread %d started" % threadID) tablesPerThread = int (self.numOfTables / self.numOfThreads) + + count = 0 while True: i = 0 - start = self.ts + start = self.ts + count * self.batchSize + count = count + 1 for i in range(tablesPerThread): tableID = i + threadID * tablesPerThread @@ -65,7 +72,7 @@ class RestfulInsert: data = "insert into %s.%s%d values" % (self.dbname, self.tableNamePerfix, tableID) values = [] for k in range(self.batchSize): - values.append("(%d, %d, %d, %d)" % (start + j * self.batchSize + k, random.randint(1, 100), random.randint(1, 100), random.randint(1, 100))) + values.append("(%d, %d, %d, %d)" % (start + k, random.randint(1, 100), random.randint(1, 100), random.randint(1, 100))) if(self.outOfOrder == False): for k in range(len(values)): @@ -73,15 +80,15 @@ class RestfulInsert: else: random.shuffle(values) for k in range(len(values)): - data += values[k] - requests.post(self.url, data, headers = self.header) + data += values[k] + response = requests.post(self.url, data, headers = self.header) + if response.status_code != 200: + print(response.content) - def run(self): - data = "drop database if exists %s" % self.dbname + def run(self): + data = "create database if not exists %s" % self.dbname requests.post(self.url, data, headers = self.header) - data = "create database %s" % self.dbname - requests.post(self.url, data, headers = self.header) - data = "create table %s.meters(ts timestamp, f1 int, f2 int, f3 int) tags(id int, loc nchar(20))" % self.dbname + data = "create table if not exists %s.meters(ts timestamp, f1 int, f2 int, f3 int) tags(id int, loc nchar(20))" % self.dbname requests.post(self.url, data, headers = self.header) threads = [] @@ -117,6 +124,13 @@ parser.add_argument( default='127.0.0.1', type=str, help='host name to be connected (default: 127.0.0.1)') +parser.add_argument( + '-S', + '--start-timestamp', + action='store', + default=1500000000000, + type=int, + help='insert data from timestamp (default: 1500000000000)') parser.add_argument( '-d', '--db-name', @@ -166,5 +180,5 @@ parser.add_argument( help='The order of test data (default: False)') args = parser.parse_args() -ri = RestfulInsert(args.host_name, args.db_name, args.number_of_threads, args.number_of_tables, args.number_of_records, args.batch_size, args.table_name_prefix, args.out_of_order) +ri = RestfulInsert(args.host_name, args.start_timestamp, args.db_name, args.number_of_threads, args.number_of_tables, args.number_of_records, args.batch_size, args.table_name_prefix, args.out_of_order) ri.run() \ No newline at end of file diff --git a/tests/pytest/query/bug2218.py b/tests/pytest/query/bug2218.py new file mode 100644 index 0000000000..bb92e5d9ce --- /dev/null +++ b/tests/pytest/query/bug2218.py @@ -0,0 +1,54 @@ +################################################################### +# Copyright (c) 2016 by TAOS Technologies, Inc. +# All rights reserved. +# +# This file is proprietary and confidential to TAOS Technologies. +# No part of this file may be reproduced, stored, transmitted, +# disclosed or used in any form or by any means other than as +# expressly provided by the written permission from Jianhui Tao +# +################################################################### + +# -*- coding: utf-8 -*- + +import sys +from util.log import * +from util.cases import * +from util.sql import * +from util.dnodes import * +class TDTestCase: + def init(self, conn, logSql): + tdLog.debug("start to execute %s" % __file__) + tdSql.init(conn.cursor(), logSql) + + def run(self): + tdSql.prepare() + print("==========step1") + print("create table && insert data") + + tdSql.execute("create table mt0 (ts timestamp, c1 int, c2 float, c3 bigint, c4 smallint, c5 tinyint, c6 double, c7 bool)") + insertRows = 1000 + t0 = 1604298064000 + tdLog.info("insert %d rows" % (insertRows)) + for i in range(insertRows): + ret = tdSql.execute( + "insert into mt0 values (%d , %d,%d,%d,%d,%d,%d,%d)" % + (t0+i,i%100,i/2,i%100,i%100,i%100,i*1.0,i%2)) + print("==========step2") + print("test col*1*1 desc ") + tdSql.query('select c1,c1*1*1,c2*1*1,c3*1*1,c4*1*1,c5*1*1,c6*1*1 from mt0 order by ts desc limit 2') + tdSql.checkData(0,0,99) + tdSql.checkData(0,1,0.0) + tdSql.checkData(0,2,0.0) + tdSql.checkData(0,3,0.0) + tdSql.checkData(0,4,0.0) + tdSql.checkData(0,5,0.0) + tdSql.checkData(0,6,0.0) + + + def stop(self): + tdSql.close() + tdLog.success("%s successfully executed" % __file__) + +tdCases.addWindows(__file__, TDTestCase()) +tdCases.addLinux(__file__, TDTestCase()) \ No newline at end of file diff --git a/tests/pytest/query/filterOtherTypes.py b/tests/pytest/query/filterOtherTypes.py index 85d5a67bef..f80552138d 100644 --- a/tests/pytest/query/filterOtherTypes.py +++ b/tests/pytest/query/filterOtherTypes.py @@ -376,11 +376,9 @@ class TDTestCase: tdSql.execute("insert into t1 values(1538548685000, 1) (1538548685001, 2) (1538548685002, 3)") tdSql.execute("insert into t2 values(1538548685000, 4) (1538548685001, 5) (1538548685002, 6)") - tdSql.query("select * from t1 where tag1 like '%g'") - tdSql.checkRows(3) + tdSql.error("select * from t1 where tag1 like '%g'") - tdSql.query("select * from t2 where tag1 like '%g'") - tdSql.checkRows(3) + tdSql.error("select * from t2 where tag1 like '%g'") tdSql.query("select * from meters where tag1 like '%g'") tdSql.checkRows(6) @@ -396,20 +394,16 @@ class TDTestCase: tdSql.execute("insert into t5 values(1538548685000, 1) (1538548685001, 2) (1538548685002, 3)") tdSql.execute("insert into t6 values(1538548685000, 1) (1538548685001, 2) (1538548685002, 3)") - tdSql.query("select * from t3 where tag1 like '%京'") - tdSql.checkRows(3) + tdSql.error("select * from t3 where tag1 like '%京'") - tdSql.query("select * from t4 where tag1 like '%京'") - tdSql.checkRows(3) + tdSql.error("select * from t4 where tag1 like '%京'") tdSql.query("select * from meters1 where tag1 like '%京'") tdSql.checkRows(6) - tdSql.query("select * from t5 where tag1 like '%g'") - tdSql.checkRows(3) + tdSql.error("select * from t5 where tag1 like '%g'") - tdSql.query("select * from t6 where tag1 like '%g'") - tdSql.checkRows(3) + tdSql.error("select * from t6 where tag1 like '%g'") tdSql.query("select * from meters1 where tag1 like '%g'") tdSql.checkRows(6) diff --git a/tests/pytest/query/queryInterval.py b/tests/pytest/query/queryInterval.py index ec3255ba7f..98b8c9fbef 100644 --- a/tests/pytest/query/queryInterval.py +++ b/tests/pytest/query/queryInterval.py @@ -32,30 +32,44 @@ class TDTestCase: tdSql.execute("insert into t0 using st tags('beijing') values(%d, 220) (%d, 221) (%d, 225) (%d, 228) (%d, 222)" % (self.ts, self.ts + 1000000000, self.ts + 2000000000, self.ts + 3000000000, self.ts + 6000000000)) tdSql.execute("insert into t1 using st tags('shanghai') values(%d, 220) (%d, 221) (%d, 225) (%d, 228) (%d, 222)" - % (self.ts, self.ts + 2000000000, self.ts + 4000000000, self.ts + 5000000000, self.ts + 7000000000)) + % (self.ts, self.ts + 2000000000, self.ts + 4000000000, self.ts + 5000000000, self.ts + 7000000000)) + tdSql.query("select avg(voltage) from st interval(1n)") - tdSql.checkRows(4) - tdSql.checkData(0, 1, 220.0) - tdSql.checkData(1, 1, 222.33333333333334) - tdSql.checkData(2, 1, 227.0) - tdSql.checkData(3, 1, 222.0) + tdSql.checkRows(3) + tdSql.checkData(0, 0, "2020-07-01 00:00:00") + tdSql.checkData(0, 1, 221.4) + tdSql.checkData(1, 0, "2020-08-01 00:00:00") + tdSql.checkData(1, 1, 227.0) + tdSql.checkData(2, 0, "2020-09-01 00:00:00") + tdSql.checkData(2, 1, 222.0) tdSql.query("select avg(voltage) from st interval(1n, 15d)") tdSql.checkRows(4) + tdSql.checkData(0, 0, "2020-06-16 00:00:00") tdSql.checkData(0, 1, 220.333333) + tdSql.checkData(1, 0, "2020-07-16 00:00:00") tdSql.checkData(1, 1, 224.666666) + tdSql.checkData(2, 0, "2020-08-16 00:00:00") tdSql.checkData(2, 1, 225.0) + tdSql.checkData(3, 0, "2020-09-16 00:00:00") tdSql.checkData(3, 1, 222.0) tdSql.query("select avg(voltage) from st interval(1n, 15d) group by loc") tdSql.checkRows(7) + tdSql.checkData(0, 0, "2020-06-16 00:00:00") tdSql.checkData(0, 1, 220.5) + tdSql.checkData(1, 0, "2020-07-16 00:00:00") tdSql.checkData(1, 1, 226.5) + tdSql.checkData(2, 0, "2020-08-16 00:00:00") tdSql.checkData(2, 1, 222.0) + tdSql.checkData(3, 0, "2020-06-16 00:00:00") tdSql.checkData(3, 1, 220.0) + tdSql.checkData(4, 0, "2020-07-16 00:00:00") tdSql.checkData(4, 1, 221.0) + tdSql.checkData(5, 0, "2020-08-16 00:00:00") tdSql.checkData(5, 1, 226.5) + tdSql.checkData(6, 0, "2020-09-16 00:00:00") tdSql.checkData(6, 1, 222.0) def stop(self): diff --git a/tests/pytest/query/queryJoin.py b/tests/pytest/query/queryJoin.py index 5ad49a265e..59e01615b4 100644 --- a/tests/pytest/query/queryJoin.py +++ b/tests/pytest/query/queryJoin.py @@ -141,6 +141,42 @@ class TDTestCase: tdSql.query("select * from meters1, meters3 where meters1.ts = meters3.ts and meters1.tag1 = meters3.tag1") tdSql.checkRows(0) + tdSql.execute("create table join_mt0(ts timestamp, c1 int, c2 float, c3 bigint, c4 smallint, c5 tinyint, c6 double, c7 bool, c8 binary(10), c9 nchar(9)) tags(t1 int, t2 binary(12))") + tdSql.execute("create table join_mt1(ts timestamp, c1 int, c2 float, c3 bigint, c4 smallint, c5 tinyint, c6 double, c7 bool, c8 binary(10), c9 nchar(9)) tags(t1 int, t2 binary(12), t3 int)") + + ts = 1538548685000 + for i in range(3): + tdSql.execute("create table join_tb%d using join_mt0 tags(%d, 'abc')" % (i, i)) + sql = "insert into join_tb%d values" % i + for j in range(500): + val = j % 100 + sql += "(%d, %d, %f, %d, %d, %d, %f, %d, 'binary%d', 'nchar%d')" % (ts + j, val, val * 1.0, val, val, val, val * 1.0, val % 2, val, val) + tdSql.execute(sql) + sql = "insert into join_tb%d values" % i + for j in range(500, 1000): + val = j % 100 + sql += "(%d, %d, %f, %d, %d, %d, %f, %d, 'binary%d', 'nchar%d')" % (ts + 500 + j, val, val * 1.0, val, val, val, val * 1.0, val % 2, val, val) + tdSql.execute(sql) + + for i in range(3): + tdSql.execute("create table join_1_tb%d using join_mt1 tags(%d, 'abc%d', %d)" % (i, i, i, i)) + sql = "insert into join_1_tb%d values" % i + for j in range(500): + val = j % 100 + sql += "(%d, %d, %f, %d, %d, %d, %f, %d, 'binary%d', 'nchar%d')" % (ts + j, val, val * 1.0, val, val, val, val * 1.0, val % 2, val, val) + tdSql.execute(sql) + sql = "insert into join_1_tb%d values" % i + for j in range(500, 1000): + val = j % 100 + sql += "(%d, %d, %f, %d, %d, %d, %f, %d, 'binary%d', 'nchar%d')" % (ts + 500 + j, val, val * 1.0, val, val, val, val * 1.0, val % 2, val, val) + tdSql.execute(sql) + + tdSql.error("select count(join_mt0.c1), sum(join_mt1.c2), first(join_mt0.c5), last(join_mt1.c7) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts interval(10a) group by join_mt0.t1 order by join_mt0.ts desc") + tdSql.error("select count(join_mt0.c1), first(join_mt0.c1)-first(join_mt1.c1), first(join_mt1.c9) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts") + tdSql.error("select count(join_mt0.c1), first(join_mt0.c1), first(join_mt1.c9) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts interval(10a) group by join_mt0.t1, join_mt0.t2 order by join_mt0.t1 desc slimit 3") + tdSql.error("select count(join_mt0.c1), first(join_mt0.c1) from join_mt0, join_mt1 where join_mt0.t1=join_mt1.t1 and join_mt0.ts=join_mt1.ts interval(10a) group by join_mt0.t1, join_mt0.t2, join_mt1.t1 order by join_mt0.ts desc, join_mt1.ts asc limit 10;") + tdSql.error("select join_mt1.c1,join_mt0.c1 from join_mt1,join_mt0 where join_mt1.ts = join_mt0.ts and join_mt1.t1 = join_mt0.t1 order by t") + def stop(self): tdSql.close() diff --git a/tests/pytest/tools/lowa.py b/tests/pytest/tools/lowaTest.py similarity index 100% rename from tests/pytest/tools/lowa.py rename to tests/pytest/tools/lowaTest.py diff --git a/tests/pytest/tools/taosdemo.py b/tests/pytest/tools/taosdemoTest.py similarity index 90% rename from tests/pytest/tools/taosdemo.py rename to tests/pytest/tools/taosdemoTest.py index 5bf8ebaf03..1a976aef59 100644 --- a/tests/pytest/tools/taosdemo.py +++ b/tests/pytest/tools/taosdemoTest.py @@ -55,6 +55,12 @@ class TDTestCase: tdSql.query("select count(*) from meters") tdSql.checkData(0, 0, self.numberOfTables * self.numberOfRecords) + tdSql.query("select sum(f1) from test.meters interval(1h) sliding(30m)") + tdSql.checkRows(2) + + tdSql.query("select apercentile(f1, 1) from test.meters interval(10s)") + tdSql.checkRows(11) + def stop(self): tdSql.close() tdLog.success("%s successfully executed" % __file__) diff --git a/tests/pytest/tools/taosdumpTest.py b/tests/pytest/tools/taosdumpTest.py new file mode 100644 index 0000000000..534a477b34 --- /dev/null +++ b/tests/pytest/tools/taosdumpTest.py @@ -0,0 +1,89 @@ +################################################################### +# Copyright (c) 2016 by TAOS Technologies, Inc. +# All rights reserved. +# +# This file is proprietary and confidential to TAOS Technologies. +# No part of this file may be reproduced, stored, transmitted, +# disclosed or used in any form or by any means other than as +# expressly provided by the written permission from Jianhui Tao +# +################################################################### + +# -*- coding: utf-8 -*- + +import sys +import os +from util.log import * +from util.cases import * +from util.sql import * +from util.dnodes import * + + +class TDTestCase: + def init(self, conn, logSql): + tdLog.debug("start to execute %s" % __file__) + tdSql.init(conn.cursor(), logSql) + + self.ts = 1538548685000 + self.numberOfTables = 10000 + self.numberOfRecords = 100 + + def run(self): + tdSql.prepare() + + tdSql.execute("create table st(ts timestamp, c1 int, c2 nchar(10)) tags(t1 int, t2 binary(10))") + tdSql.execute("create table t1 using st tags(1, 'beijing')") + sql = "insert into t1 values" + currts = self.ts + for i in range(100): + sql += "(%d, %d, 'nchar%d')" % (currts + i, i % 100, i % 100) + tdSql.execute(sql) + + tdSql.execute("create table t2 using st tags(2, 'shanghai')") + sql = "insert into t2 values" + currts = self.ts + for i in range(100): + sql += "(%d, %d, 'nchar%d')" % (currts + i, i % 100, i % 100) + tdSql.execute(sql) + + os.system("taosdump --databases db -o /tmp") + + tdSql.execute("drop database db") + tdSql.query("show databases") + tdSql.checkRows(0) + + os.system("taosdump -i /tmp") + + tdSql.query("show databases") + tdSql.checkRows(1) + tdSql.checkData(0, 0, 'db') + + tdSql.execute("use db") + tdSql.query("show stables") + tdSql.checkRows(1) + tdSql.checkData(0, 0, 'st') + + tdSql.query("show tables") + tdSql.checkRows(2) + tdSql.checkData(0, 0, 't2') + tdSql.checkData(1, 0, 't1') + + tdSql.query("select * from t1") + tdSql.checkRows(100) + for i in range(100): + tdSql.checkData(i, 1, i) + tdSql.checkData(i, 2, "nchar%d" % i) + + tdSql.query("select * from t2") + tdSql.checkRows(100) + for i in range(100): + tdSql.checkData(i, 1, i) + tdSql.checkData(i, 2, "nchar%d" % i) + + def stop(self): + tdSql.close() + tdLog.success("%s successfully executed" % __file__) + + +tdCases.addWindows(__file__, TDTestCase()) +tdCases.addLinux(__file__, TDTestCase()) \ No newline at end of file diff --git a/tests/script/basicSuite.sim b/tests/script/basicSuite.sim index 61a9d68d26..5e22e02297 100644 --- a/tests/script/basicSuite.sim +++ b/tests/script/basicSuite.sim @@ -14,7 +14,6 @@ run general/table/vgroup.sim run general/user/authority.sim run general/vector/metrics_mix.sim run general/vector/table_field.sim -run general/user/authority.sim run general/tag/set.sim run general/table/delete_writing.sim run general/stable/disk.sim diff --git a/tests/script/general/wal/sync.sim b/tests/script/general/wal/sync.sim index abaf22f919..c6f7402b87 100644 --- a/tests/script/general/wal/sync.sim +++ b/tests/script/general/wal/sync.sim @@ -82,6 +82,7 @@ restful d1 table_rest 1591772800 30000 restful d1 table_rest 1591872800 30000 restful d1 table_rest 1591972800 30000 +sleep 1000 sql select * from table_rest; print rows: $rows if $rows != 300000 then diff --git a/tests/script/tmp/mnodes.sim b/tests/script/tmp/mnodes.sim index 48dbc19cb2..e11140028d 100644 --- a/tests/script/tmp/mnodes.sim +++ b/tests/script/tmp/mnodes.sim @@ -20,6 +20,14 @@ system sh/cfg.sh -n dnode1 -c maxTablesPerVnode -v 20000 system sh/cfg.sh -n dnode2 -c maxTablesPerVnode -v 20000 system sh/cfg.sh -n dnode3 -c maxTablesPerVnode -v 20000 +system sh/cfg.sh -n dnode1 -c minTablesPerVnode -v 1000 +system sh/cfg.sh -n dnode2 -c minTablesPerVnode -v 1000 +system sh/cfg.sh -n dnode3 -c minTablesPerVnode -v 1000 + +system sh/cfg.sh -n dnode1 -c maxVgroupsPerDb -v 20 +system sh/cfg.sh -n dnode2 -c maxVgroupsPerDb -v 20 +system sh/cfg.sh -n dnode3 -c maxVgroupsPerDb -v 20 + system sh/cfg.sh -n dnode1 -c replica -v 3 system sh/cfg.sh -n dnode2 -c replica -v 3 system sh/cfg.sh -n dnode3 -c replica -v 3 diff --git a/tests/script/unique/cluster/flowctrl.sim b/tests/script/unique/cluster/flowctrl.sim new file mode 100644 index 0000000000..6dc60d9fba --- /dev/null +++ b/tests/script/unique/cluster/flowctrl.sim @@ -0,0 +1,131 @@ +system sh/stop_dnodes.sh + +system sh/deploy.sh -n dnode1 -i 1 +system sh/deploy.sh -n dnode2 -i 2 +system sh/deploy.sh -n dnode3 -i 3 + +system sh/cfg.sh -n dnode1 -c numOfMnodes -v 3 +system sh/cfg.sh -n dnode2 -c numOfMnodes -v 3 +system sh/cfg.sh -n dnode3 -c numOfMnodes -v 3 + +system sh/cfg.sh -n dnode1 -c mnodeEqualVnodeNum -v 4 +system sh/cfg.sh -n dnode2 -c mnodeEqualVnodeNum -v 4 +system sh/cfg.sh -n dnode3 -c mnodeEqualVnodeNum -v 4 + +system sh/cfg.sh -n dnode1 -c http -v 0 +system sh/cfg.sh -n dnode2 -c http -v 0 +system sh/cfg.sh -n dnode3 -c http -v 0 + +system sh/cfg.sh -n dnode1 -c maxTablesPerVnode -v 20000 +system sh/cfg.sh -n dnode2 -c maxTablesPerVnode -v 20000 +system sh/cfg.sh -n dnode3 -c maxTablesPerVnode -v 20000 + +system sh/cfg.sh -n dnode1 -c maxVgroupsPerDb -v 20 +system sh/cfg.sh -n dnode2 -c maxVgroupsPerDb -v 20 +system sh/cfg.sh -n dnode3 -c maxVgroupsPerDb -v 20 + +system sh/cfg.sh -n dnode1 -c replica -v 3 +system sh/cfg.sh -n dnode2 -c replica -v 3 +system sh/cfg.sh -n dnode3 -c replica -v 3 + +print ============== deploy + +system sh/exec.sh -n dnode1 -s start +sleep 5001 +sql connect + +sql create dnode $hostname2 +sql create dnode $hostname3 +system sh/exec.sh -n dnode2 -s start +system sh/exec.sh -n dnode3 -s start + +print =============== step1 +$x = 0 +show1: + $x = $x + 1 + sleep 2000 + if $x == 5 then + return -1 + endi +sql show mnodes -x show1 +$mnode1Role = $data2_1 +print mnode1Role $mnode1Role +$mnode2Role = $data2_2 +print mnode2Role $mnode2Role +$mnode3Role = $data2_3 +print mnode3Role $mnode3Role + +if $mnode1Role != master then + goto show1 +endi +if $mnode2Role != slave then + goto show1 +endi +if $mnode3Role != slave then + goto show1 +endi + +print =============== step2 + +sql create database db replica 3 +sql use db +sql create table tb (ts timestamp, test int) + +$x = 0 +while $x < 100 + $ms = $x . s + sql insert into tb values (now + $ms , $x ) + $x = $x + 1 +endw + +print =============== step3 +sleep 3000 + +system sh/exec.sh -n dnode1 -s stop -x SIGINT +system sh/exec.sh -n dnode2 -s stop -x SIGINT +system sh/exec.sh -n dnode3 -s stop -x SIGINT + +print =============== step4 +sleep 5000 +system sh/exec.sh -n dnode1 -s start +system sh/exec.sh -n dnode2 -s start +system sh/exec.sh -n dnode3 -s start + +print =============== step5 +sleep 8000 +while $x < 200 + $ms = $x . s + sql insert into tb values (now + $ms , $x ) + $x = $x + 1 +endw + +print =============== step6 +system sh/exec.sh -n dnode2 -s stop -x SIGINT +sleep 3000 +while $x < 300 + $ms = $x . s + sql insert into tb values (now + $ms , $x ) + $x = $x + 1 +endw + +system sh/exec.sh -n dnode2 -s start + +sleep 6000 +print =============== step7 +while $x < 400 + $ms = $x . s + sql insert into tb values (now + $ms , $x ) + $x = $x + 1 + sleep 1 +endw + +print =============== step8 +sql select * from tb +print rows $rows +if $rows != 400 then + return -1 +endi + +system sh/exec.sh -n dnode1 -s stop -x SIGINT +system sh/exec.sh -n dnode2 -s stop -x SIGINT +system sh/exec.sh -n dnode3 -s stop -x SIGINT \ No newline at end of file diff --git a/tests/test-all.sh b/tests/test-all.sh index ff47cbfd71..5897978bce 100755 --- a/tests/test-all.sh +++ b/tests/test-all.sh @@ -17,9 +17,9 @@ function runSimCaseOneByOne { echo -e "${GREEN}$case success${NC}" | tee -a out.log || \ echo -e "${RED}$case failed${NC}" | tee -a out.log out_log=`tail -1 out.log ` - if [[ $out_log =~ 'failed' ]];then - exit 8 - fi + # if [[ $out_log =~ 'failed' ]];then + # exit 8 + # fi end_time=`date +%s` echo execution time of $case was `expr $end_time - $start_time`s. | tee -a out.log fi @@ -42,9 +42,9 @@ function runPyCaseOneByOne { echo -e "${RED}$case failed${NC}" | tee -a pytest-out.log end_time=`date +%s` out_log=`tail -1 pytest-out.log ` - if [[ $out_log =~ 'failed' ]];then - exit 8 - fi + # if [[ $out_log =~ 'failed' ]];then + # exit 8 + # fi echo execution time of $case was `expr $end_time - $start_time`s. | tee -a pytest-out.log else $line > /dev/null 2>&1 diff --git a/tests/test/c/CMakeLists.txt b/tests/test/c/CMakeLists.txt index 26aa20e647..11480a8ba2 100644 --- a/tests/test/c/CMakeLists.txt +++ b/tests/test/c/CMakeLists.txt @@ -31,8 +31,8 @@ IF (TD_LINUX) #add_executable(createTablePerformance createTablePerformance.c) #target_link_libraries(createTablePerformance taos_static tutil common pthread) - #add_executable(createNormalTable createNormalTable.c) - #target_link_libraries(createNormalTable taos_static tutil common pthread) + add_executable(createNormalTable createNormalTable.c) + target_link_libraries(createNormalTable taos_static tutil common pthread) #add_executable(queryPerformance queryPerformance.c) #target_link_libraries(queryPerformance taos_static tutil common pthread) @@ -45,5 +45,8 @@ IF (TD_LINUX) #add_executable(invalidTableId invalidTableId.c) #target_link_libraries(invalidTableId taos_static tutil common pthread) + + add_executable(hashIterator hashIterator.c) + target_link_libraries(hashIterator taos_static tutil common pthread) ENDIF() diff --git a/tests/test/c/hashIterator.c b/tests/test/c/hashIterator.c new file mode 100644 index 0000000000..cbd8a0895e --- /dev/null +++ b/tests/test/c/hashIterator.c @@ -0,0 +1,72 @@ +/* + * Copyright (c) 2019 TAOS Data, Inc. + * + * 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 . + */ + +#define _DEFAULT_SOURCE +#include "os.h" +#include "taos.h" +#include "tulog.h" +#include "tutil.h" +#include "hash.h" + +typedef struct HashTestRow { + int32_t keySize; + char key[100]; +} HashTestRow; + +int main(int argc, char *argv[]) { + _hash_fn_t hashFp = taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY); + void * hashHandle = taosHashInit(100, hashFp, true, HASH_ENTRY_LOCK); + + pPrint("insert 3 rows to hash"); + for (int32_t t = 0; t < 3; ++t) { + HashTestRow row = {0}; + row.keySize = sprintf(row.key, "0.db.st%d", t); + + taosHashPut(hashHandle, row.key, row.keySize, &row, sizeof(HashTestRow)); + } + + pPrint("start iterator"); + HashTestRow *row = taosHashIterate(hashHandle, NULL); + while (row) { + pPrint("drop key:%s", row->key); + taosHashRemove(hashHandle, row->key, row->keySize); + + pPrint("get rows from hash"); + for (int32_t t = 0; t < 3; ++t) { + HashTestRow r = {0}; + r.keySize = sprintf(r.key, "0.db.st%d", t); + + void *result = taosHashGet(hashHandle, r.key, r.keySize); + pPrint("get key:%s result:%p", r.key, result); + } + + //Before getting the next iterator, the object just deleted can be obtained + row = taosHashIterate(hashHandle, row); + } + + pPrint("stop iterator"); + taosHashCancelIterate(hashHandle, row); + + pPrint("get rows from hash"); + for (int32_t t = 0; t < 3; ++t) { + HashTestRow r = {0}; + r.keySize = sprintf(r.key, "0.db.st%d", t); + + void *result = taosHashGet(hashHandle, r.key, r.keySize); + pPrint("get key:%s result:%p", r.key, result); + } + + return 0; +} \ No newline at end of file