Merge branch 'develop' into feature/sangshuduo/perftest-script-improve

.travis.yml

@@ -63,7 +63,7 @@ matrix:
           pkill -TERM -x taosd
           fuser -k -n tcp 6030
           sleep 1
-          ./crash_gen.sh -a -p -t 4 -s 25|| travis_terminate $?
+          ./crash_gen.sh -a -p -t 4 -s 2000|| travis_terminate $?
           sleep 1
 
           cd ${TRAVIS_BUILD_DIR}/tests/pytest

packaging/cfg/taos.cfg

@@ -15,28 +15,15 @@
 # arbitrator            arbitrator_hostname:6030     
 
 # the full-qualified domain name (FQDN) of dnode 
-# fqdn                  hostname:6030
+# fqdn                  hostname
 
-# port for MNode connect to Client, default udp[6030-6055] tcp[6030]
+# port for MNode connect to Client, default udp/tcp [6030-6040]
 # serverPort            6030
 
-# http service port, default tcp[6020]
+# http service port, default tcp [6020]
 # httpPort              6020
 
-# set socket type ("udp" and "tcp")
-# the server and client should have the same socket type. Otherwise, connect will fail
-# sockettype            udp
-
-# for the cluster version, data file's directory is configured this way 
-# option                mount_path           tier_level
-# dataDir               /mnt/disk1/taos      0
-# dataDir               /mnt/disk2/taos      0
-# dataDir               /mnt/disk3/taos      0
-# dataDir               /mnt/disk4/taos      0
-# dataDir               /mnt/disk5/taos      0
-# dataDir               /mnt/disk6/taos      1
-# dataDir               /mnt/disk7/taos      1
-# for the stand-alone version, data file's directory is configured this way 
+# data file's directory
 # dataDir               /var/lib/taos
 
 # log file's directory
@@ -46,12 +33,12 @@
 # numOfMnodes           3
 
 # optional roles for dnode. 0 - any, 1 - mnode, 2 - dnode
-# alternativeRole       0
+# role                  0
 
 # number of threads per CPU core
 # numOfThreadsPerCore   1.0
 
-# number of vgroups per db
+# max number of vgroups per db
 # maxVgroupsPerDb       0
 
 # max number of tables per vnode
@@ -60,9 +47,6 @@
 # the ratio of threads responsible for querying in the total thread
 # ratioOfQueryThreads   0.5
 
-# interval of check load balance when the management node is in normal operation
-# balanceInterval       300
-
 # interval of DNode report status to MNode, unit is Second, for cluster version only 
 # statusInterval        1
 
@@ -82,13 +66,13 @@
 # tableMetaKeepTimer    7200
 
 # Minimum sliding window time
-# minSlidingTime         10
+# minSlidingTime        10
 
 # Time window minimum
-# minIntervalTime        10
+# minIntervalTime       10
 
 # max length of an SQL
-# maxSQLLength          65380
+# maxSQLLength          65480
 
 # Support the maximum number of records allowed for super table time sorting
 # maxNumOfOrderedRes    100000
@@ -153,11 +137,8 @@
 # Stop writing data when the disk size of the log folder is less than this value
 # minimalDataDirGB      0.1
 
-# mnode take into account while balance, for cluster version only 
-# mnodeEqualVnodeNum    4
-
 # number of seconds allowed for a dnode to be offline, for cluster version only 
-# offlineThreshold      864000
+# offlineThreshold      8640000
 
 # start http service
 # http                  1
@@ -165,7 +146,7 @@
 # start system monitor module
 # monitor               1
 
-# start http service
+# start muqq service
 # mqtt                  0
 
 # mqtt uri  
@@ -201,19 +182,22 @@
 # 131: output warning and error，135: output info, warning and error to log.
 # 199: output debug, info, warning and error to both screen and file
 
-# debug flag for basic utils
-# debugFlag             131
+# debug flag for all log type, take effect when non-zero value
+# debugFlag             0
 
 # debug flag for meta management messages
 # mDebugFlag            135
 
 # debug flag for dnode messages
-# dDebugFlag            131
+# dDebugFlag            135
 
-# debug flag for TDengine SDB
+# debug flag for sync module
 # sDebugFlag            135
 
-# debug flag for TDengine SDB
+# debug flag for WAL
+# wDebugFlag            135
+
+# debug flag for SDB
 # sdbDebugFlag          135
 
 # debug flag for RPC 
@@ -246,6 +230,9 @@
 # debug flag for query
 # qDebugflag            131
 
+# debug flag for vnode
+# vDebugflag            131
+
 # debug flag for http server
 # tsdbDebugFlag         131
 

src/client/inc/tscLocalMerge.h

@@ -20,8 +20,8 @@
 extern "C" {
 #endif
 
-#include "qextbuffer.h"
-#include "qfill.h"
+#include "qExtbuffer.h"
+#include "qFill.h"
 #include "taosmsg.h"
 #include "tlosertree.h"
 #include "tsclient.h"

src/client/inc/tscUtil.h

@@ -23,11 +23,11 @@ extern "C" {
 /*
  * @date   2018/09/30
  */
-#include "os.h"
-#include "tbuffer.h"
 #include "exception.h"
-#include "qextbuffer.h"
+#include "os.h"
+#include "qExtbuffer.h"
 #include "taosdef.h"
+#include "tbuffer.h"
 #include "tscLocalMerge.h"
 #include "tsclient.h"
 
@@ -268,7 +268,7 @@ bool hasMoreClauseToTry(SSqlObj* pSql);
 void tscTryQueryNextVnode(SSqlObj *pSql, __async_cb_func_t fp);
 void tscAsyncQuerySingleRowForNextVnode(void *param, TAOS_RES *tres, int numOfRows);
 void tscTryQueryNextClause(SSqlObj* pSql, __async_cb_func_t fp);
-int  tscSetMgmtIpListFromCfg(const char *first, const char *second);
+int  tscSetMgmtEpSetFromCfg(const char *first, const char *second);
 
 void* malloc_throw(size_t size);
 void* calloc_throw(size_t nmemb, size_t size);

src/client/inc/tsclient.h

@@ -31,8 +31,8 @@ extern "C" {
 #include "tutil.h"
 
 #include "qExecutor.h"
+#include "qTsbuf.h"
 #include "qsqlparser.h"
-#include "qtsbuf.h"
 #include "tcmdtype.h"
 
 // forward declaration
@@ -52,12 +52,20 @@ typedef struct STableComInfo {
   int32_t rowSize;
 } STableComInfo;
 
+typedef struct SCMCorVgroupInfo {
+  int32_t version;
+  int8_t inUse;
+  int8_t  numOfEps;
+  SEpAddr epAddr[TSDB_MAX_REPLICA];
+} SCMCorVgroupInfo;
+
 typedef struct STableMeta {
   STableComInfo tableInfo;
   uint8_t       tableType;
   int16_t       sversion;
   int16_t       tversion;
-  SCMVgroupInfo vgroupInfo;
+  SCMVgroupInfo  vgroupInfo;
+  SCMCorVgroupInfo  corVgroupInfo;
   int32_t       sid;       // the index of one table in a virtual node
   uint64_t      uid;       // unique id of a table
   SSchema       schema[];  // if the table is TSDB_CHILD_TABLE, schema is acquired by super table meta info
@@ -306,7 +314,7 @@ typedef struct SSqlObj {
   char *           sqlstr;
   char             retry;
   char             maxRetry;
-  SRpcIpSet        ipList;
+  SRpcEpSet        epSet;
   char             listed;
   tsem_t           rspSem;
   SSqlCmd          cmd;
@@ -350,7 +358,7 @@ void    tscInitMsgsFp();
 
 int tsParseSql(SSqlObj *pSql, bool initial);
 
-void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcIpSet *pIpSet);
+void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcEpSet *pEpSet);
 int  tscProcessSql(SSqlObj *pSql);
 
 int  tscRenewTableMeta(SSqlObj *pSql, char *tableId);
@@ -456,7 +464,8 @@ extern void *    tscQhandle;
 extern int       tscKeepConn[];
 extern int       tsInsertHeadSize;
 extern int       tscNumOfThreads;
-extern SRpcIpSet tscMgmtIpSet;
+  
+extern SRpcCorEpSet tscMgmtEpSet;
 
 extern int (*tscBuildMsg[TSDB_SQL_MAX])(SSqlObj *pSql, SSqlInfo *pInfo);
 

src/client/src/tscFunctionImpl.c

@@ -14,15 +14,15 @@
  */
 
 #include "os.h"
-#include "qextbuffer.h"
-#include "qfill.h"
-#include "qhistogram.h"
-#include "qpercentile.h"
-#include "qsyntaxtreefunction.h"
-#include "qtsbuf.h"
+#include "qAst.h"
+#include "qExtbuffer.h"
+#include "qFill.h"
+#include "qHistogram.h"
+#include "qPercentile.h"
+#include "qSyntaxtreefunction.h"
+#include "qTsbuf.h"
 #include "taosdef.h"
 #include "taosmsg.h"
-#include "qast.h"
 #include "tscLog.h"
 #include "tscSubquery.h"
 #include "tscompression.h"
@@ -74,7 +74,7 @@ for (int32_t i = 0; i < (ctx)->tagInfo.numOfTagCols; ++i) {                  \
 void noop1(SQLFunctionCtx *UNUSED_PARAM(pCtx)) {}
 void noop2(SQLFunctionCtx *UNUSED_PARAM(pCtx), int32_t UNUSED_PARAM(index)) {}
 
-void doFinalizer(SQLFunctionCtx *pCtx) { resetResultInfo(GET_RES_INFO(pCtx)); }
+void doFinalizer(SQLFunctionCtx *pCtx) { RESET_RESULT_INFO(GET_RES_INFO(pCtx)); }
 
 typedef struct tValuePair {
   tVariant v;
@@ -330,12 +330,6 @@ int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionI
   return TSDB_CODE_SUCCESS;
 }
 
-/**
- * the numOfRes should be kept, since it may be used later
- * and allow the ResultInfo to be re initialized
- */
-void resetResultInfo(SResultInfo *pResInfo) { pResInfo->initialized = false; }
-
 void setResultInfoBuf(SResultInfo *pResInfo, int32_t size, bool superTable, char* buf) {
   assert(pResInfo->interResultBuf == NULL);
   

src/client/src/tscLocal.c

@@ -16,14 +16,14 @@
 #include "os.h"
 #include "taosmsg.h"
 
-#include "tcache.h"
-#include "tscUtil.h"
-#include "tsclient.h"
+#include "qExtbuffer.h"
 #include "taosdef.h"
+#include "tcache.h"
+#include "tname.h"
 #include "tscLog.h"
-#include "qextbuffer.h"
+#include "tscUtil.h"
 #include "tschemautil.h"
-#include "tname.h"
+#include "tsclient.h"
 
 static void tscSetLocalQueryResult(SSqlObj *pSql, const char *val, const char *columnName, int16_t type, size_t valueLength);
 

src/client/src/tscSQLParser.c

@@ -18,9 +18,9 @@
 #define _DEFAULT_SOURCE
 
 #include "os.h"
+#include "qAst.h"
 #include "taos.h"
 #include "taosmsg.h"
-#include "qast.h"
 #include "tcompare.h"
 #include "tname.h"
 #include "tscLog.h"
@@ -358,7 +358,7 @@ int32_t tscToSQLCmd(SSqlObj* pSql, struct SSqlInfo* pInfo) {
     }
 
     case TSDB_SQL_CFG_DNODE: {
-      const char* msg2 = "invalid configure options or values";
+      const char* msg2 = "invalid configure options or values, such as resetlog / debugFlag 135 / balance 'vnode:2-dnode:2' / monitor 1 ";
       const char* msg3 = "invalid dnode ep";
 
       /* validate the ip address */
@@ -2355,9 +2355,9 @@ bool validateIpAddress(const char* ip, size_t size) {
 
   strncpy(tmp, ip, size);
 
-  in_addr_t ipAddr = inet_addr(tmp);
+  in_addr_t epAddr = inet_addr(tmp);
 
-  return ipAddr != INADDR_NONE;
+  return epAddr != INADDR_NONE;
 }
 
 int32_t tscTansformSQLFuncForSTableQuery(SQueryInfo* pQueryInfo) {
@@ -4674,26 +4674,42 @@ int32_t validateDNodeConfig(tDCLSQL* pOptions) {
     return TSDB_CODE_TSC_INVALID_SQL;
   }
 
-  const int DNODE_DYNAMIC_CFG_OPTIONS_SIZE = 19;
-  const SDNodeDynConfOption DNODE_DYNAMIC_CFG_OPTIONS[] = {
-      {"resetLog", 8},       {"resetQueryCache", 15},  {"debugFlag", 9},     {"mDebugFlag", 10},
-      {"dDebugFlag", 10},    {"sdbDebugFlag", 12},     {"vDebugFlag", 10},   {"cDebugFlag", 10},
-      {"httpDebugFlag", 13}, {"monitorDebugFlag", 16}, {"rpcDebugFlag", 12}, {"uDebugFlag", 10},
-      {"tmrDebugFlag", 12},  {"qDebugflag", 10},       {"sDebugflag", 10},   {"tsdbDebugFlag", 13},
-      {"mqttDebugFlag", 13}, {"wDebugFlag", 10},       {"monitor", 7}};
+  const int tokenLogEnd = 2;
+  const int tokenBalance = 2;
+  const int tokenMonitor = 3;
+  const int tokenDebugFlag = 4;
+  const int tokenDebugFlagEnd = 20;
+  const SDNodeDynConfOption cfgOptions[] = {
+      {"resetLog", 8},    {"resetQueryCache", 15},  {"balance", 7},     {"monitor", 7},
+      {"debugFlag", 9},   {"monitorDebugFlag", 16}, {"vDebugFlag", 10}, {"mDebugFlag", 10},
+      {"cDebugFlag", 10}, {"httpDebugFlag", 13},    {"qDebugflag", 10}, {"sdbDebugFlag", 12},
+      {"uDebugFlag", 10}, {"tsdbDebugFlag", 13},    {"sDebugflag", 10}, {"rpcDebugFlag", 12},
+      {"dDebugFlag", 10}, {"mqttDebugFlag", 13},    {"wDebugFlag", 10}, {"tmrDebugFlag", 12},
+  };
 
   SSQLToken* pOptionToken = &pOptions->a[1];
 
   if (pOptions->nTokens == 2) {
     // reset log and reset query cache does not need value
-    for (int32_t i = 0; i < 2; ++i) {
-      const SDNodeDynConfOption* pOption = &DNODE_DYNAMIC_CFG_OPTIONS[i];
+    for (int32_t i = 0; i < tokenLogEnd; ++i) {
+      const SDNodeDynConfOption* pOption = &cfgOptions[i];
       if ((strncasecmp(pOption->name, pOptionToken->z, pOptionToken->n) == 0) && (pOption->len == pOptionToken->n)) {
         return TSDB_CODE_SUCCESS;
       }
     }
-  } else if ((strncasecmp(DNODE_DYNAMIC_CFG_OPTIONS[DNODE_DYNAMIC_CFG_OPTIONS_SIZE - 1].name, pOptionToken->z, pOptionToken->n) == 0) &&
-             (DNODE_DYNAMIC_CFG_OPTIONS[DNODE_DYNAMIC_CFG_OPTIONS_SIZE - 1].len == pOptionToken->n)) {
+  } else if ((strncasecmp(cfgOptions[tokenBalance].name, pOptionToken->z, pOptionToken->n) == 0) &&
+             (cfgOptions[tokenBalance].len == pOptionToken->n)) {
+    SSQLToken* pValToken = &pOptions->a[2];
+    int32_t vnodeId = 0;
+    int32_t dnodeId = 0;
+    strdequote(pValToken->z);
+    bool parseOk = taosCheckBalanceCfgOptions(pValToken->z, &vnodeId, &dnodeId);
+    if (!parseOk) {
+      return TSDB_CODE_TSC_INVALID_SQL;  // options value is invalid
+    }
+    return TSDB_CODE_SUCCESS;
+  } else if ((strncasecmp(cfgOptions[tokenMonitor].name, pOptionToken->z, pOptionToken->n) == 0) &&
+             (cfgOptions[tokenMonitor].len == pOptionToken->n)) {
     SSQLToken* pValToken = &pOptions->a[2];
     int32_t    val = strtol(pValToken->z, NULL, 10);
     if (val != 0 && val != 1) {
@@ -4709,8 +4725,8 @@ int32_t validateDNodeConfig(tDCLSQL* pOptions) {
       return TSDB_CODE_TSC_INVALID_SQL;
     }
 
-    for (int32_t i = 2; i < DNODE_DYNAMIC_CFG_OPTIONS_SIZE - 1; ++i) {
-      const SDNodeDynConfOption* pOption = &DNODE_DYNAMIC_CFG_OPTIONS[i];
+    for (int32_t i = tokenDebugFlag; i < tokenDebugFlagEnd; ++i) {
+      const SDNodeDynConfOption* pOption = &cfgOptions[i];
 
       if ((strncasecmp(pOption->name, pOptionToken->z, pOptionToken->n) == 0) && (pOption->len == pOptionToken->n)) {
         /* options is valid */

src/client/src/tscSchemaUtil.c

@@ -140,7 +140,15 @@ struct SSchema tscGetTbnameColumnSchema() {
   strcpy(s.name, TSQL_TBNAME_L);
   return s;
 }
-
+static void tscInitCorVgroupInfo(SCMCorVgroupInfo *corVgroupInfo, SCMVgroupInfo *vgroupInfo) {
+  corVgroupInfo->version = 0;
+  corVgroupInfo->inUse = 0;
+  corVgroupInfo->numOfEps = vgroupInfo->numOfEps;
+  for (int32_t i = 0; i < corVgroupInfo->numOfEps; i++) {
+    strncpy(corVgroupInfo->epAddr[i].fqdn, vgroupInfo->epAddr[i].fqdn, TSDB_FQDN_LEN);
+    corVgroupInfo->epAddr[i].port = vgroupInfo->epAddr[i].port;
+  }
+}
 STableMeta* tscCreateTableMetaFromMsg(STableMetaMsg* pTableMetaMsg, size_t* size) {
   assert(pTableMetaMsg != NULL);
   
@@ -157,6 +165,9 @@ STableMeta* tscCreateTableMetaFromMsg(STableMetaMsg* pTableMetaMsg, size_t* size
   pTableMeta->sid = pTableMetaMsg->sid;
   pTableMeta->uid = pTableMetaMsg->uid;
   pTableMeta->vgroupInfo = pTableMetaMsg->vgroup;
+
+  tscInitCorVgroupInfo(&pTableMeta->corVgroupInfo, &pTableMeta->vgroupInfo);
+
   pTableMeta->sversion = pTableMetaMsg->sversion;
   pTableMeta->tversion = pTableMetaMsg->tversion;
   

src/client/src/tscServer.c

@@ -26,11 +26,12 @@
 #include "ttime.h"
 #include "ttimer.h"
 #include "tutil.h"
+#include "tlockfree.h"
 
 #define TSC_MGMT_VNODE 999
 
-SRpcIpSet  tscMgmtIpSet;
-SRpcIpSet  tscDnodeIpSet;
+SRpcCorEpSet  tscMgmtEpSet;
+SRpcEpSet  tscDnodeEpSet;
 
 int (*tscBuildMsg[TSDB_SQL_MAX])(SSqlObj *pSql, SSqlInfo *pInfo) = {0};
 
@@ -44,47 +45,89 @@ void tscSaveSubscriptionProgress(void* sub);
 
 static int32_t minMsgSize() { return tsRpcHeadSize + 100; }
 
-static void tscSetDnodeIpList(SSqlObj* pSql, SCMVgroupInfo* pVgroupInfo) {
-  SRpcIpSet* pIpList = &pSql->ipList;
-  pIpList->inUse    = 0;
+static void tscSetDnodeEpSet(SSqlObj* pSql, SCMVgroupInfo* pVgroupInfo) {
+  SRpcEpSet* pEpSet = &pSql->epSet;
+  pEpSet->inUse    = 0;
   if (pVgroupInfo == NULL) {
-    pIpList->numOfIps = 0;
+    pEpSet->numOfEps = 0;
     return;
   }
-  
-  pIpList->numOfIps = pVgroupInfo->numOfIps;
-  for(int32_t i = 0; i < pVgroupInfo->numOfIps; ++i) {
-    strcpy(pIpList->fqdn[i], pVgroupInfo->ipAddr[i].fqdn);
-    pIpList->port[i] = pVgroupInfo->ipAddr[i].port;
-  }
-}
 
-void tscPrintMgmtIp() {
-  if (tscMgmtIpSet.numOfIps <= 0) {
-    tscError("invalid mnode IP list:%d", tscMgmtIpSet.numOfIps);
+  pEpSet->numOfEps = pVgroupInfo->numOfEps;
+  for(int32_t i = 0; i < pVgroupInfo->numOfEps; ++i) {
+    strcpy(pEpSet->fqdn[i], pVgroupInfo->epAddr[i].fqdn);
+    pEpSet->port[i] = pVgroupInfo->epAddr[i].port;
+  }
+}
+static void tscDumpMgmtEpSet(SRpcEpSet *epSet) {
+  taosCorBeginRead(&tscMgmtEpSet.version);
+  *epSet = tscMgmtEpSet.epSet;
+  taosCorEndRead(&tscMgmtEpSet.version);
+}  
+static void tscEpSetHtons(SRpcEpSet *s) {
+   for (int32_t i = 0; i < s->numOfEps; i++) {
+      s->port[i] = htons(s->port[i]);    
+   }
+} 
+bool tscEpSetIsEqual(SRpcEpSet *s1, SRpcEpSet *s2) {
+   if (s1->numOfEps != s2->numOfEps || s1->inUse != s2->inUse) {
+     return false;
+   } 
+   for (int32_t i = 0; i < s1->numOfEps; i++) {
+     if (s1->port[i] != s2->port[i] 
+        || strncmp(s1->fqdn[i], s2->fqdn[i], TSDB_FQDN_LEN) != 0)
+        return false;
+   }
+   return true;
+}
+void tscUpdateMgmtEpSet(SRpcEpSet *pEpSet) {
+  // no need to update if equal
+  taosCorBeginWrite(&tscMgmtEpSet.version);
+  tscMgmtEpSet.epSet = *pEpSet;
+  taosCorEndWrite(&tscMgmtEpSet.version);
+}
+static void tscDumpEpSetFromVgroupInfo(SCMCorVgroupInfo *pVgroupInfo, SRpcEpSet *pEpSet) {
+  if (pVgroupInfo == NULL) { return;}
+  taosCorBeginRead(&pVgroupInfo->version);
+  int8_t inUse = pVgroupInfo->inUse;
+  pEpSet->inUse = (inUse >= 0 && inUse < TSDB_MAX_REPLICA) ? inUse: 0; 
+  pEpSet->numOfEps = pVgroupInfo->numOfEps;  
+  for (int32_t i = 0; i < pVgroupInfo->numOfEps; ++i) {
+    strncpy(pEpSet->fqdn[i], pVgroupInfo->epAddr[i].fqdn, TSDB_FQDN_LEN);
+    pEpSet->port[i] = pVgroupInfo->epAddr[i].port;
+  }
+  taosCorEndRead(&pVgroupInfo->version);
+}
+
+static void tscUpdateVgroupInfo(SSqlObj *pObj, SRpcEpSet *pEpSet) {
+  SSqlCmd *pCmd = &pObj->cmd;
+  STableMetaInfo *pTableMetaInfo = tscGetTableMetaInfoFromCmd(pCmd, pCmd->clauseIndex, 0);
+  if (pTableMetaInfo == NULL || pTableMetaInfo->pTableMeta == NULL) { return;}
+  SCMCorVgroupInfo *pVgroupInfo = &pTableMetaInfo->pTableMeta->corVgroupInfo;
+
+  taosCorBeginWrite(&pVgroupInfo->version);
+  tscDebug("before: Endpoint in use: %d", pVgroupInfo->inUse);
+  pVgroupInfo->inUse = pEpSet->inUse;
+  pVgroupInfo->numOfEps = pEpSet->numOfEps;
+  for (int32_t i = 0; i < pVgroupInfo->numOfEps; i++) {
+    strncpy(pVgroupInfo->epAddr[i].fqdn, pEpSet->fqdn[i], TSDB_FQDN_LEN);
+    pVgroupInfo->epAddr[i].port = pEpSet->port[i];
+  }
+  tscDebug("after: EndPoint in use: %d", pVgroupInfo->inUse);
+  taosCorEndWrite(&pVgroupInfo->version);
+}
+void tscPrintMgmtEp() {
+  SRpcEpSet dump;
+  tscDumpMgmtEpSet(&dump);
+  if (dump.numOfEps <= 0) {
+    tscError("invalid mnode EP list:%d", dump.numOfEps);
   } else {
-    for (int i = 0; i < tscMgmtIpSet.numOfIps; ++i) {
-      tscDebug("mnode index:%d %s:%d", i, tscMgmtIpSet.fqdn[i], tscMgmtIpSet.port[i]);
+    for (int i = 0; i < dump.numOfEps; ++i) {
+      tscDebug("mnode index:%d %s:%d", i, dump.fqdn[i], dump.port[i]);
     }
   }
 }
 
-void tscSetMgmtIpList(SRpcIpSet *pIpList) {
-  tscMgmtIpSet.numOfIps = pIpList->numOfIps;
-  tscMgmtIpSet.inUse = pIpList->inUse;
-  for (int32_t i = 0; i < tscMgmtIpSet.numOfIps; ++i) {
-    tscMgmtIpSet.port[i] = htons(pIpList->port[i]);
-  }
-}
-
-void tscUpdateIpSet(void *ahandle, SRpcIpSet *pIpSet) {
-  tscMgmtIpSet = *pIpSet;
-  tscDebug("mnode IP list is changed for ufp is called, numOfIps:%d inUse:%d", tscMgmtIpSet.numOfIps, tscMgmtIpSet.inUse);
-  for (int32_t i = 0; i < tscMgmtIpSet.numOfIps; ++i) {
-    tscDebug("index:%d fqdn:%s port:%d", i, tscMgmtIpSet.fqdn[i], tscMgmtIpSet.port[i]);
-  }
-}
-
 /*
  * For each management node, try twice at least in case of poor network situation.
  * If the client start to connect to a non-management node from the client, and the first retry may fail due to
@@ -95,7 +138,9 @@ void tscUpdateIpSet(void *ahandle, SRpcIpSet *pIpSet) {
 UNUSED_FUNC
 static int32_t tscGetMgmtConnMaxRetryTimes() {
   int32_t factor = 2;
-  return tscMgmtIpSet.numOfIps * factor;
+  SRpcEpSet dump;
+  tscDumpMgmtEpSet(&dump);
+  return dump.numOfEps * factor;
 }
 
 void tscProcessHeartBeatRsp(void *param, TAOS_RES *tres, int code) {
@@ -111,9 +156,11 @@ void tscProcessHeartBeatRsp(void *param, TAOS_RES *tres, int code) {
 
   if (code == 0) {
     SCMHeartBeatRsp *pRsp = (SCMHeartBeatRsp *)pRes->pRsp;
-    SRpcIpSet *      pIpList = &pRsp->ipList;
-    if (pIpList->numOfIps > 0) 
-      tscSetMgmtIpList(pIpList);
+    SRpcEpSet *      epSet = &pRsp->epSet;
+    if (epSet->numOfEps > 0) {
+      tscEpSetHtons(epSet);
+      tscUpdateMgmtEpSet(epSet);
+    } 
 
     pSql->pTscObj->connId = htonl(pRsp->connId);
 
@@ -185,7 +232,7 @@ int tscSendMsgToServer(SSqlObj *pSql) {
 
   // set the mgmt ip list
   if (pSql->cmd.command >= TSDB_SQL_MGMT) {
-    pSql->ipList = tscMgmtIpSet;
+    tscDumpMgmtEpSet(&pSql->epSet);
   }
 
   memcpy(pMsg, pSql->cmd.payload, pSql->cmd.payloadLen);
@@ -203,11 +250,11 @@ int tscSendMsgToServer(SSqlObj *pSql) {
   // Otherwise, the pSql object may have been released already during the response function, which is
   // processMsgFromServer function. In the meanwhile, the assignment of the rpc context to sql object will absolutely
   // cause crash.
-  rpcSendRequest(pObj->pDnodeConn, &pSql->ipList, &rpcMsg);
+  rpcSendRequest(pObj->pDnodeConn, &pSql->epSet, &rpcMsg);
   return TSDB_CODE_SUCCESS;
 }
 
-void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcIpSet *pIpSet) {
+void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcEpSet *pEpSet) {
   SSqlObj *pSql = (SSqlObj *)rpcMsg->ahandle;
   if (pSql == NULL || pSql->signature != pSql) {
     tscError("%p sql is already released", pSql);
@@ -236,58 +283,41 @@ void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcIpSet *pIpSet) {
     return;
   }
 
-  if (pCmd->command < TSDB_SQL_MGMT) {
-    if (pIpSet) pSql->ipList = *pIpSet;
-  } else {
-    if (pIpSet) tscMgmtIpSet = *pIpSet;
+  if (pEpSet) { 
+    if (!tscEpSetIsEqual(&pSql->epSet, pEpSet)) {
+      if(pCmd->command < TSDB_SQL_MGMT)  { 
+        tscUpdateVgroupInfo(pSql, pEpSet); 
+      } else {
+        tscUpdateMgmtEpSet(pEpSet);
+    }
+    }
   }
 
-  if (rpcMsg->pCont == NULL) {
-    rpcMsg->code = TSDB_CODE_RPC_NETWORK_UNAVAIL;
-  } else {
-    STableMetaInfo *pTableMetaInfo = tscGetTableMetaInfoFromCmd(pCmd, pCmd->clauseIndex, 0);
-    // if (rpcMsg->code != TSDB_CODE_RPC_NETWORK_UNAVAIL) {
-    //   if (pCmd->command == TSDB_SQL_CONNECT) {
-    //     rpcMsg->code = TSDB_CODE_RPC_NETWORK_UNAVAIL;
-    //     rpcFreeCont(rpcMsg->pCont);
-    //     return;
-    //   }
-
-    //   if (pCmd->command == TSDB_SQL_HB) {
-    //     rpcMsg->code = TSDB_CODE_RPC_NOT_READY;
-    //     rpcFreeCont(rpcMsg->pCont);
-    //     return;
-    //   }
-
-    //   if (pCmd->command == TSDB_SQL_META || pCmd->command == TSDB_SQL_DESCRIBE_TABLE ||
-    //       pCmd->command == TSDB_SQL_STABLEVGROUP || pCmd->command == TSDB_SQL_SHOW ||
-    //       pCmd->command == TSDB_SQL_RETRIEVE) {
-    //     // get table meta/vgroup query will not retry, do nothing
-    //   }
-    // }
-
-    if ((pCmd->command == TSDB_SQL_SELECT || pCmd->command == TSDB_SQL_FETCH || pCmd->command == TSDB_SQL_INSERT ||
-         pCmd->command == TSDB_SQL_UPDATE_TAGS_VAL) &&
-        (rpcMsg->code == TSDB_CODE_TDB_INVALID_TABLE_ID || rpcMsg->code == TSDB_CODE_VND_INVALID_VGROUP_ID ||
-         rpcMsg->code == TSDB_CODE_RPC_NETWORK_UNAVAIL || rpcMsg->code == TSDB_CODE_TDB_TABLE_RECONFIGURE)) {
-      tscWarn("%p it shall renew table meta, code:%s, retry:%d", pSql, tstrerror(rpcMsg->code), ++pSql->retry);
-      // set the flag to denote that sql string needs to be re-parsed and build submit block with table schema
-      if (rpcMsg->code == TSDB_CODE_TDB_TABLE_RECONFIGURE) {
-        pSql->cmd.submitSchema = 1;
-      }
+  STableMetaInfo *pTableMetaInfo = tscGetTableMetaInfoFromCmd(pCmd, pCmd->clauseIndex, 0);
 
-      pSql->res.code = rpcMsg->code;  // keep the previous error code
-      if (pSql->retry > pSql->maxRetry) {
-        tscError("%p max retry %d reached, give up", pSql, pSql->maxRetry);
-      } else {
-        rpcMsg->code = tscRenewTableMeta(pSql, pTableMetaInfo->name);
-
-        // if there is an error occurring, proceed to the following error handling procedure.
-        // todo add test cases
-        if (rpcMsg->code == TSDB_CODE_TSC_ACTION_IN_PROGRESS) {
-          rpcFreeCont(rpcMsg->pCont);
-          return;
-        }
+  int32_t cmd = pCmd->command;
+  if ((cmd == TSDB_SQL_SELECT || cmd == TSDB_SQL_FETCH || cmd == TSDB_SQL_INSERT || cmd == TSDB_SQL_UPDATE_TAGS_VAL) &&
+      (rpcMsg->code == TSDB_CODE_TDB_INVALID_TABLE_ID ||
+       rpcMsg->code == TSDB_CODE_VND_INVALID_VGROUP_ID ||
+       rpcMsg->code == TSDB_CODE_RPC_NETWORK_UNAVAIL ||
+       rpcMsg->code == TSDB_CODE_TDB_TABLE_RECONFIGURE)) {
+    tscWarn("%p it shall renew table meta, code:%s, retry:%d", pSql, tstrerror(rpcMsg->code), ++pSql->retry);
+
+    // set the flag to denote that sql string needs to be re-parsed and build submit block with table schema
+    if (rpcMsg->code == TSDB_CODE_TDB_TABLE_RECONFIGURE) {
+      pSql->cmd.submitSchema = 1;
+    }
+
+    pSql->res.code = rpcMsg->code;  // keep the previous error code
+    if (pSql->retry > pSql->maxRetry) {
+      tscError("%p max retry %d reached, give up", pSql, pSql->maxRetry);
+    } else {
+      rpcMsg->code = tscRenewTableMeta(pSql, pTableMetaInfo->name);
+
+      // if there is an error occurring, proceed to the following error handling procedure.
+      if (rpcMsg->code == TSDB_CODE_TSC_ACTION_IN_PROGRESS) {
+        rpcFreeCont(rpcMsg->pCont);
+        return;
       }
     }
   }
@@ -421,7 +451,8 @@ int tscProcessSql(SSqlObj *pSql) {
       return pSql->res.code;
     }
   } else if (pCmd->command < TSDB_SQL_LOCAL) {
-    pSql->ipList = tscMgmtIpSet;
+
+    //pSql->epSet = tscMgmtEpSet;
   } else {  // local handler
     return (*tscProcessMsgRsp[pCmd->command])(pSql);
   }
@@ -525,10 +556,10 @@ int tscBuildSubmitMsg(SSqlObj *pSql, SSqlInfo *pInfo) {
 
   // pSql->cmd.payloadLen is set during copying data into payload
   pSql->cmd.msgType = TSDB_MSG_TYPE_SUBMIT;
-  tscSetDnodeIpList(pSql, &pTableMeta->vgroupInfo);
-  
-  tscDebug("%p build submit msg, vgId:%d numOfTables:%d numberOfIP:%d", pSql, vgId, pSql->cmd.numOfTablesInSubmit,
-      pSql->ipList.numOfIps);
+  tscDumpEpSetFromVgroupInfo(&pTableMeta->corVgroupInfo, &pSql->epSet);
+
+  tscDebug("%p build submit msg, vgId:%d numOfTables:%d numberOfEP:%d", pSql, vgId, pSql->cmd.numOfTablesInSubmit,
+      pSql->epSet.numOfEps);
   return TSDB_CODE_SUCCESS;
 }
 
@@ -567,11 +598,11 @@ static char *doSerializeTableInfo(SQueryTableMsg* pQueryMsg, SSqlObj *pSql, char
     } else {
       pVgroupInfo = &pTableMeta->vgroupInfo;
     }
+    tscSetDnodeEpSet(pSql, pVgroupInfo);
 
-    tscSetDnodeIpList(pSql, pVgroupInfo);
     if (pVgroupInfo != NULL) {
       pQueryMsg->head.vgId = htonl(pVgroupInfo->vgId);
-    }
+    } 
 
     STableIdInfo *pTableIdInfo = (STableIdInfo *)pMsg;
     pTableIdInfo->tid = htonl(pTableMeta->sid);
@@ -580,7 +611,7 @@ static char *doSerializeTableInfo(SQueryTableMsg* pQueryMsg, SSqlObj *pSql, char
 
     pQueryMsg->numOfTables = htonl(1);  // set the number of tables
     pMsg += sizeof(STableIdInfo);
-  } else { // it is a subquery of the super table query, this IP info is acquired from vgroupInfo
+  } else { // it is a subquery of the super table query, this EP info is acquired from vgroupInfo
     int32_t index = pTableMetaInfo->vgroupIndex;
     int32_t numOfVgroups = taosArrayGetSize(pTableMetaInfo->pVgroupTables);
     assert(index >= 0 && index < numOfVgroups);
@@ -588,9 +619,9 @@ static char *doSerializeTableInfo(SQueryTableMsg* pQueryMsg, SSqlObj *pSql, char
     tscDebug("%p query on stable, vgIndex:%d, numOfVgroups:%d", pSql, index, numOfVgroups);
 
     SVgroupTableInfo* pTableIdList = taosArrayGet(pTableMetaInfo->pVgroupTables, index);
-    
-    // set the vgroup info
-    tscSetDnodeIpList(pSql, &pTableIdList->vgInfo);
+
+    // set the vgroup info 
+    tscSetDnodeEpSet(pSql, &pTableIdList->vgInfo);
     pQueryMsg->head.vgId = htonl(pTableIdList->vgInfo.vgId);
     
     int32_t numOfTables = taosArrayGetSize(pTableIdList->itemList);
@@ -1136,11 +1167,11 @@ int32_t tscBuildShowMsg(SSqlObj *pSql, SSqlInfo *pInfo) {
       pShowMsg->payloadLen = htons(pPattern->n);
     }
   } else {
-    SSQLToken *pIpAddr = &pShowInfo->prefix;
-    assert(pIpAddr->n > 0 && pIpAddr->type > 0);
+    SSQLToken *pEpAddr = &pShowInfo->prefix;
+    assert(pEpAddr->n > 0 && pEpAddr->type > 0);
 
-    strncpy(pShowMsg->payload, pIpAddr->z, pIpAddr->n);
-    pShowMsg->payloadLen = htons(pIpAddr->n);
+    strncpy(pShowMsg->payload, pEpAddr->z, pEpAddr->n);
+    pShowMsg->payloadLen = htons(pEpAddr->n);
   }
 
   pCmd->payloadLen = sizeof(SCMShowMsg) + pShowMsg->payloadLen;
@@ -1323,7 +1354,7 @@ int tscBuildUpdateTagMsg(SSqlObj* pSql, SSqlInfo *pInfo) {
   SQueryInfo *    pQueryInfo = tscGetQueryInfoDetail(pCmd, 0);
   STableMetaInfo *pTableMetaInfo = tscGetMetaInfo(pQueryInfo, 0);
 
-  tscSetDnodeIpList(pSql, &pTableMetaInfo->pTableMeta->vgroupInfo);
+  tscDumpEpSetFromVgroupInfo(&pTableMetaInfo->pTableMeta->corVgroupInfo, &pSql->epSet);
 
   return TSDB_CODE_SUCCESS;
 }
@@ -1658,8 +1689,8 @@ int tscProcessTableMetaRsp(SSqlObj *pSql) {
   pMetaMsg->contLen = htons(pMetaMsg->contLen);
   pMetaMsg->numOfColumns = htons(pMetaMsg->numOfColumns);
 
-  if (pMetaMsg->sid < 0 || pMetaMsg->vgroup.numOfIps < 0) {
-    tscError("invalid meter vgId:%d, sid%d", pMetaMsg->vgroup.numOfIps, pMetaMsg->sid);
+  if (pMetaMsg->sid < 0 || pMetaMsg->vgroup.numOfEps < 0) {
+    tscError("invalid meter vgId:%d, sid%d", pMetaMsg->vgroup.numOfEps, pMetaMsg->sid);
     return TSDB_CODE_TSC_INVALID_VALUE;
   }
 
@@ -1673,8 +1704,8 @@ int tscProcessTableMetaRsp(SSqlObj *pSql) {
     return TSDB_CODE_TSC_INVALID_VALUE;
   }
 
-  for (int i = 0; i < pMetaMsg->vgroup.numOfIps; ++i) {
-    pMetaMsg->vgroup.ipAddr[i].port = htons(pMetaMsg->vgroup.ipAddr[i].port);
+  for (int i = 0; i < pMetaMsg->vgroup.numOfEps; ++i) {
+    pMetaMsg->vgroup.epAddr[i].port = htons(pMetaMsg->vgroup.epAddr[i].port);
   }
 
   SSchema* pSchema = pMetaMsg->schema;
@@ -1847,13 +1878,14 @@ int tscProcessSTableVgroupRsp(SSqlObj *pSql) {
 
     memcpy(pInfo->vgroupList, pVgroupInfo, size);
     for (int32_t j = 0; j < pInfo->vgroupList->numOfVgroups; ++j) {
+      //just init, no need to lock
       SCMVgroupInfo *pVgroups = &pInfo->vgroupList->vgroups[j];
-
       pVgroups->vgId = htonl(pVgroups->vgId);
-      assert(pVgroups->numOfIps >= 1);
+      assert(pVgroups->numOfEps >= 1);
+
+      for (int32_t k = 0; k < pVgroups->numOfEps; ++k) {
+        pVgroups->epAddr[k].port = htons(pVgroups->epAddr[k].port);
 
-      for (int32_t k = 0; k < pVgroups->numOfIps; ++k) {
-        pVgroups->ipAddr[k].port = htons(pVgroups->ipAddr[k].port);
       }
 
       pMsg += size;
@@ -1946,8 +1978,10 @@ int tscProcessConnectRsp(SSqlObj *pSql) {
   assert(len <= sizeof(pObj->db));
   tstrncpy(pObj->db, temp, sizeof(pObj->db));
   
-  if (pConnect->ipList.numOfIps > 0) 
-    tscSetMgmtIpList(&pConnect->ipList);
+  if (pConnect->epSet.numOfEps > 0) {
+    tscEpSetHtons(&pConnect->epSet);
+    tscUpdateMgmtEpSet(&pConnect->epSet);
+  } 
 
   strcpy(pObj->sversion, pConnect->serverVersion);
   pObj->writeAuth = pConnect->writeAuth;

src/client/src/tscSql.c

@@ -15,7 +15,7 @@
 
 #include "hash.h"
 #include "os.h"
-#include "qast.h"
+#include "qAst.h"
 #include "tcache.h"
 #include "tnote.h"
 #include "trpc.h"
@@ -62,8 +62,8 @@ SSqlObj *taosConnectImpl(const char *ip, const char *user, const char *pass, con
   }
 
   if (ip) {
-    if (tscSetMgmtIpListFromCfg(ip, NULL) < 0) return NULL;
-    if (port) tscMgmtIpSet.port[0] = port;
+    if (tscSetMgmtEpSetFromCfg(ip, NULL) < 0) return NULL;
+    if (port) tscMgmtEpSet.epSet.port[0] = port;
   } 
  
   void *pDnodeConn = NULL;
@@ -724,6 +724,13 @@ int taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields)
   return len;
 }
 
+static void asyncCallback(void *param, TAOS_RES *tres, int code) {
+  assert(param != NULL);
+  SSqlObj *pSql = ((SSqlObj *)param);
+  pSql->res.code = code;
+  sem_post(&pSql->rspSem);
+}
+
 int taos_validate_sql(TAOS *taos, const char *sql) {
   STscObj *pObj = (STscObj *)taos;
   if (pObj == NULL || pObj->signature != pObj) {
@@ -732,7 +739,8 @@ int taos_validate_sql(TAOS *taos, const char *sql) {
   }
 
   SSqlObj* pSql = calloc(1, sizeof(SSqlObj));
-  
+  pSql->pTscObj = taos;
+  pSql->signature = pSql;
   SSqlRes *pRes = &pSql->res;
   SSqlCmd *pCmd = &pSql->cmd;
   
@@ -766,10 +774,17 @@ int taos_validate_sql(TAOS *taos, const char *sql) {
     pCmd->pTableList = NULL;
   }
 
-  pRes->code = (uint8_t)tsParseSql(pSql, false);
-  int code = pRes->code;
-
-  tscDebug("%p Valid SQL result:%d, %s pObj:%p", pSql, pRes->code, taos_errstr(taos), pObj);
+  pSql->fp = asyncCallback;
+  pSql->fetchFp = asyncCallback;
+  pSql->param = pSql;
+  int code = tsParseSql(pSql, true);
+  if (code == TSDB_CODE_TSC_ACTION_IN_PROGRESS) {
+    sem_wait(&pSql->rspSem);
+    code = pSql->res.code;
+  }
+  if (code != TSDB_CODE_SUCCESS) {
+    tscDebug("%p Valid SQL result:%d, %s pObj:%p", pSql, code, taos_errstr(taos), pObj);
+  }
   taos_free_result(pSql);
 
   return code;
@@ -865,6 +880,8 @@ int taos_load_table_info(TAOS *taos, const char *tableNameList) {
   }
 
   SSqlObj* pSql = calloc(1, sizeof(SSqlObj));
+  pSql->pTscObj = taos;
+  pSql->signature = pSql;
   SSqlRes *pRes = &pSql->res;
 
   pRes->numOfTotal = 0;  // the number of getting table meta from server

src/client/src/tscStream.c

@@ -122,7 +122,7 @@ static void tscProcessStreamTimer(void *handle, void *tmrId) {
       pQueryInfo->window.ekey = pStream->etime;
     }
   } else {
-    pQueryInfo->window.skey = pStream->stime - pStream->interval;
+    pQueryInfo->window.skey = pStream->stime;
     int64_t etime = taosGetTimestamp(pStream->precision);
     // delay to wait all data in last time window
     if (pStream->precision == TSDB_TIME_PRECISION_MICRO) {
@@ -232,6 +232,9 @@ static void tscProcessStreamRetrieveResult(void *param, TAOS_RES *res, int numOf
       (*pStream->fp)(pStream->param, res, row);
     }
 
+    if (!pStream->isProject) {
+      pStream->stime += pStream->slidingTime;
+    }
     // actually only one row is returned. this following is not necessary
     taos_fetch_rows_a(res, tscProcessStreamRetrieveResult, pStream);
   } else {  // numOfRows == 0, all data has been retrieved
@@ -432,6 +435,7 @@ static int64_t tscGetStreamStartTimestamp(SSqlObj *pSql, SSqlStream *pStream, in
   } else {             // timewindow based aggregation stream
     if (stime == 0) {  // no data in meter till now
       stime = ((int64_t)taosGetTimestamp(pStream->precision) / pStream->interval) * pStream->interval;
+      stime -= pStream->interval;
       tscWarn("%p stream:%p, last timestamp:0, reset to:%" PRId64, pSql, pStream, stime);
     } else {
       int64_t newStime = (stime / pStream->interval) * pStream->interval;

src/client/src/tscSub.c

@@ -34,6 +34,7 @@ typedef struct SSubscriptionProgress {
 typedef struct SSub {
   void *                  signature;
   char                    topic[32];
+  sem_t                   sem;
   int64_t                 lastSyncTime;
   int64_t                 lastConsumeTime;
   TAOS *                  taos;
@@ -83,84 +84,108 @@ void tscUpdateSubscriptionProgress(void* sub, int64_t uid, TSKEY ts) {
 
 static void asyncCallback(void *param, TAOS_RES *tres, int code) {
   assert(param != NULL);
-  SSqlObj *pSql = ((SSqlObj *)param);
-  
-  pSql->res.code = code;
-  sem_post(&pSql->rspSem);
+  SSub *pSub = ((SSub *)param);
+  pSub->pSql->res.code = code;
+  sem_post(&pSub->sem);
 }
 
 
 static SSub* tscCreateSubscription(STscObj* pObj, const char* topic, const char* sql) {
-  SSub* pSub = NULL;
-
-  TRY( 8 ) {
-    SSqlObj* pSql = calloc_throw(1, sizeof(SSqlObj));
-    CLEANUP_PUSH_FREE(true, pSql);
-    SSqlCmd *pCmd = &pSql->cmd;
-    SSqlRes *pRes = &pSql->res;
-
-    if (tsem_init(&pSql->rspSem, 0, 0) == -1) {
-      THROW(TAOS_SYSTEM_ERROR(errno));
-    }
-    CLEANUP_PUSH_INT_PTR(true, tsem_destroy, &pSql->rspSem);
+  int code = TSDB_CODE_SUCCESS, line = __LINE__;
+  SSqlObj* pSql = NULL;
 
-    pSql->signature = pSql;
-    pSql->param = pSql;
-    pSql->pTscObj = pObj;
-    pSql->maxRetry = TSDB_MAX_REPLICA;
-    pSql->fp = asyncCallback;
-
-    int code = tscAllocPayload(pCmd, TSDB_DEFAULT_PAYLOAD_SIZE);
-    if (code != TSDB_CODE_SUCCESS) {
-      THROW(code);
-    }
-    CLEANUP_PUSH_FREE(true, pCmd->payload);
+  SSub* pSub = calloc(1, sizeof(SSub));
+  if (pSub == NULL) {
+    line = __LINE__;
+    code = TSDB_CODE_TSC_OUT_OF_MEMORY;
+    goto fail;
+  }
+  pSub->signature = pSub;
+  if (tsem_init(&pSub->sem, 0, 0) == -1) {
+    line = __LINE__;
+    code = TAOS_SYSTEM_ERROR(errno);
+    goto fail;
+  }
+  tstrncpy(pSub->topic, topic, sizeof(pSub->topic));
+  pSub->progress = taosArrayInit(32, sizeof(SSubscriptionProgress));
+  if (pSub->progress == NULL) {
+    line = __LINE__;
+    code = TSDB_CODE_TSC_OUT_OF_MEMORY;
+    goto fail;
+  }
 
-    pRes->qhandle = 0;
-    pRes->numOfRows = 1;
+  pSql = calloc(1, sizeof(SSqlObj));
+  if (pSql == NULL) {
+    line = __LINE__;
+    code = TSDB_CODE_TSC_OUT_OF_MEMORY;
+    goto fail;
+  }
+  pSql->signature = pSql;
+  pSql->pTscObj = pObj;
+  pSql->pSubscription = pSub;
+  pSub->pSql = pSql;
 
-    pSql->sqlstr = strdup_throw(sql);
-    CLEANUP_PUSH_FREE(true, pSql->sqlstr);
-    strtolower(pSql->sqlstr, pSql->sqlstr);
+  SSqlCmd* pCmd = &pSql->cmd;
+  SSqlRes* pRes = &pSql->res;
+  if (tsem_init(&pSql->rspSem, 0, 0) == -1) {
+    line = __LINE__;
+    code = TAOS_SYSTEM_ERROR(errno);
+    goto fail;
+  }
 
-    code = tsParseSql(pSql, false);
-    if (code == TSDB_CODE_TSC_ACTION_IN_PROGRESS) {
-      // wait for the callback function to post the semaphore
-      sem_wait(&pSql->rspSem);
-      code = pSql->res.code;
-    }
-    if (code != TSDB_CODE_SUCCESS) {
-      tscError("failed to parse sql statement: %s, error: %s", pSub->topic, tstrerror(code));
-      THROW( code );
-    }
+  pSql->param = pSub;
+  pSql->maxRetry = TSDB_MAX_REPLICA;
+  pSql->fp = asyncCallback;
+  pSql->fetchFp = asyncCallback;
+  pSql->sqlstr = strdup(sql);
+  if (pSql->sqlstr == NULL) {
+    line = __LINE__;
+    code = TSDB_CODE_TSC_OUT_OF_MEMORY;
+    goto fail;
+  }
+  strtolower(pSql->sqlstr, pSql->sqlstr);
+  pRes->qhandle = 0;
+  pRes->numOfRows = 1;
+
+  code = tscAllocPayload(pCmd, TSDB_DEFAULT_PAYLOAD_SIZE);
+  if (code != TSDB_CODE_SUCCESS) {
+    line = __LINE__;
+    goto fail;
+  }
 
-    if (pSql->cmd.command != TSDB_SQL_SELECT) {
-      tscError("only 'select' statement is allowed in subscription: %s", pSub->topic);
-      THROW( -1 );  // TODO
-    }
+  code = tsParseSql(pSql, false);
+  if (code == TSDB_CODE_TSC_ACTION_IN_PROGRESS) {
+    sem_wait(&pSub->sem);
+    code = pSql->res.code;
+  }
+  if (code != TSDB_CODE_SUCCESS) {
+    line = __LINE__;
+    goto fail;
+  }
 
-    pSub = calloc_throw(1, sizeof(SSub));
-    CLEANUP_PUSH_FREE(true, pSub);
-    pSql->pSubscription = pSub;
-    pSub->pSql = pSql;
-    pSub->signature = pSub;
-    strncpy(pSub->topic, topic, sizeof(pSub->topic));
-    pSub->topic[sizeof(pSub->topic) - 1] = 0;
-    pSub->progress = taosArrayInit(32, sizeof(SSubscriptionProgress));
-    if (pSub->progress == NULL) {
-      THROW(TSDB_CODE_TSC_OUT_OF_MEMORY);
-    }
+  if (pSql->cmd.command != TSDB_SQL_SELECT) {
+    line = __LINE__;
+    code = TSDB_CODE_TSC_INVALID_SQL;
+    goto fail;
+  }
 
-    CLEANUP_EXECUTE();
+  return pSub;
 
-  } CATCH( code ) {
-    tscError("failed to create subscription object: %s", tstrerror(code));
-    CLEANUP_EXECUTE();
+fail:
+  tscError("tscCreateSubscription failed at line %d, reason: %s", line, tstrerror(code));
+  if (pSql != NULL) {
+    tscFreeSqlObj(pSql);
+    pSql = NULL;
+  }
+  if (pSub != NULL) {
+    taosArrayDestroy(pSub->progress);
+    tsem_destroy(&pSub->sem);
+    free(pSub);
     pSub = NULL;
+  }
 
-  } END_TRY
-
-  return pSub;
+  terrno = code;
+  return NULL;
 }
 
 
@@ -231,11 +256,12 @@ static int tscUpdateSubscription(STscObj* pObj, SSub* pSub) {
   }
   size_t numOfTables = taosArrayGetSize(tables);
 
+  SQueryInfo* pQueryInfo = tscGetQueryInfoDetail(pCmd, 0);
   SArray* progress = taosArrayInit(numOfTables, sizeof(SSubscriptionProgress));
   for( size_t i = 0; i < numOfTables; i++ ) {
     STidTags* tt = taosArrayGet( tables, i );
     SSubscriptionProgress p = { .uid = tt->uid };
-    p.key = tscGetSubscriptionProgress(pSub, tt->uid, INT64_MIN);
+    p.key = tscGetSubscriptionProgress(pSub, tt->uid, pQueryInfo->window.skey);
     taosArrayPush(progress, &p);
   }
   taosArraySort(progress, tscCompareSubscriptionProgress);
@@ -405,9 +431,10 @@ TAOS_RES *taos_consume(TAOS_SUB *tsub) {
     tscGetTableMetaInfoFromCmd(&pSql->cmd, 0, 0)->vgroupIndex = 0;
 
     pSql->fp = asyncCallback;
-    pSql->param = pSql;
+    pSql->fetchFp = asyncCallback;
+    pSql->param = pSub;
     tscDoQuery(pSql);
-    sem_wait(&pSql->rspSem);
+    sem_wait(&pSub->sem);
 
     if (pRes->code != TSDB_CODE_SUCCESS) {
       continue;
@@ -437,7 +464,9 @@ void taos_unsubscribe(TAOS_SUB *tsub, int keepProgress) {
   }
 
   if (keepProgress) {
-    tscSaveSubscriptionProgress(pSub);
+    if (pSub->progress != NULL) {
+      tscSaveSubscriptionProgress(pSub);
+    }
   } else {
     char path[256];
     sprintf(path, "%s/subscribe/%s", tsDataDir, pSub->topic);
@@ -448,6 +477,7 @@ void taos_unsubscribe(TAOS_SUB *tsub, int keepProgress) {
 
   tscFreeSqlObj(pSub->pSql);
   taosArrayDestroy(pSub->progress);
+  tsem_destroy(&pSub->sem);
   memset(pSub, 0, sizeof(*pSub));
   free(pSub);
 }

src/client/src/tscSubquery.c

@@ -14,8 +14,8 @@
  */
 #include "os.h"
 
-#include "qtsbuf.h"
-#include "qast.h"
+#include "qAst.h"
+#include "qTsbuf.h"
 #include "tcompare.h"
 #include "tscLog.h"
 #include "tscSubquery.h"
@@ -458,7 +458,7 @@ void tscBuildVgroupTableInfo(SSqlObj* pSql, STableMetaInfo* pTableMetaInfo, SArr
           break;
         }
       }
-      assert(info.vgInfo.numOfIps != 0);
+      assert(info.vgInfo.numOfEps != 0);
 
       vgTables = taosArrayInit(4, sizeof(STableIdInfo));
       info.itemList = vgTables;
@@ -1600,8 +1600,8 @@ static void tscAllDataRetrievedFromDnode(SRetrieveSupport *trsupport, SSqlObj* p
   
   // data in from current vnode is stored in cache and disk
   uint32_t numOfRowsFromSubquery = trsupport->pExtMemBuffer[idx]->numOfTotalElems + trsupport->localBuffer->num;
-    tscDebug("%p sub:%p all data retrieved from ip:%s, vgId:%d, numOfRows:%d, orderOfSub:%d", pParentSql, pSql,
-        pTableMetaInfo->vgroupList->vgroups[0].ipAddr[0].fqdn, pTableMetaInfo->vgroupList->vgroups[0].vgId,
+    tscDebug("%p sub:%p all data retrieved from ep:%s, vgId:%d, numOfRows:%d, orderOfSub:%d", pParentSql, pSql,
+        pTableMetaInfo->vgroupList->vgroups[0].epAddr[0].fqdn, pTableMetaInfo->vgroupList->vgroups[0].vgId,
         numOfRowsFromSubquery, idx);
   
   tColModelCompact(pDesc->pColumnModel, trsupport->localBuffer, pDesc->pColumnModel->capacity);
@@ -1719,8 +1719,8 @@ static void tscRetrieveFromDnodeCallBack(void *param, TAOS_RES *tres, int numOfR
     assert(pRes->numOfRows == numOfRows);
     int64_t num = atomic_add_fetch_64(&pState->numOfRetrievedRows, numOfRows);
     
-    tscDebug("%p sub:%p retrieve numOfRows:%" PRId64 " totalNumOfRows:%" PRIu64 " from ip:%s, orderOfSub:%d", pParentSql, pSql,
-             pRes->numOfRows, pState->numOfRetrievedRows, pSql->ipList.fqdn[pSql->ipList.inUse], idx);
+    tscDebug("%p sub:%p retrieve numOfRows:%" PRId64 " totalNumOfRows:%" PRIu64 " from ep:%s, orderOfSub:%d", pParentSql, pSql,
+             pRes->numOfRows, pState->numOfRetrievedRows, pSql->epSet.fqdn[pSql->epSet.inUse], idx);
 
     if (num > tsMaxNumOfOrderedResults && tscIsProjectionQueryOnSTable(pQueryInfo, 0)) {
       tscError("%p sub:%p num of OrderedRes is too many, max allowed:%" PRId32 " , current:%" PRId64,
@@ -1828,8 +1828,8 @@ void tscRetrieveDataRes(void *param, TAOS_RES *tres, int code) {
     return;
   }
 
-  tscTrace("%p sub:%p query complete, ip:%s, vgId:%d, orderOfSub:%d, retrieve data", trsupport->pParentSql, pSql,
-             pVgroup->ipAddr[0].fqdn, pVgroup->vgId, trsupport->subqueryIndex);
+  tscTrace("%p sub:%p query complete, ep:%s, vgId:%d, orderOfSub:%d, retrieve data", trsupport->pParentSql, pSql,
+             pVgroup->epAddr[0].fqdn, pVgroup->vgId, trsupport->subqueryIndex);
 
   if (pSql->res.qhandle == 0) { // qhandle is NULL, code is TSDB_CODE_SUCCESS means no results generated from this vnode
     tscRetrieveFromDnodeCallBack(param, pSql, 0);

src/client/src/tscSystem.c

@@ -41,7 +41,7 @@ int tscNumOfThreads;
 
 static pthread_once_t tscinit = PTHREAD_ONCE_INIT;
 void taosInitNote(int numOfNoteLines, int maxNotes, char* lable);
-void tscUpdateIpSet(void *ahandle, SRpcIpSet *pIpSet);
+//void tscUpdateEpSet(void *ahandle, SRpcEpSet *pEpSet);
 
 void tscCheckDiskUsage(void *UNUSED_PARAM(para), void* UNUSED_PARAM(param)) {
   taosGetDisk();
@@ -116,8 +116,8 @@ void taos_init_imp() {
     taosInitNote(tsNumOfLogLines / 10, 1, (char*)"tsc_note");
   }
 
-  if (tscSetMgmtIpListFromCfg(tsFirst, tsSecond) < 0) {
-    tscError("failed to init mnode IP list");
+  if (tscSetMgmtEpSetFromCfg(tsFirst, tsSecond) < 0) {
+    tscError("failed to init mnode EP list");
     return;
   } 
 

src/client/src/tscUtil.c

@@ -13,11 +13,11 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#include "os.h"
-#include "hash.h"
 #include "tscUtil.h"
+#include "hash.h"
+#include "os.h"
+#include "qAst.h"
 #include "taosmsg.h"
-#include "qast.h"
 #include "tcache.h"
 #include "tkey.h"
 #include "tmd5.h"
@@ -2145,17 +2145,20 @@ char* strdup_throw(const char* str) {
   return p;
 }
 
-int tscSetMgmtIpListFromCfg(const char *first, const char *second) {
-  tscMgmtIpSet.numOfIps = 0;
-  tscMgmtIpSet.inUse = 0;
+int tscSetMgmtEpSetFromCfg(const char *first, const char *second) {
+  // init mgmt ip set 
+  tscMgmtEpSet.version = 0;
+  SRpcEpSet *mgmtEpSet = &(tscMgmtEpSet.epSet);
+  mgmtEpSet->numOfEps = 0;
+  mgmtEpSet->inUse = 0;
 
   if (first && first[0] != 0) {
     if (strlen(first) >= TSDB_EP_LEN) {
       terrno = TSDB_CODE_TSC_INVALID_FQDN;
       return -1;
     }
-    taosGetFqdnPortFromEp(first, tscMgmtIpSet.fqdn[tscMgmtIpSet.numOfIps], &tscMgmtIpSet.port[tscMgmtIpSet.numOfIps]);
-    tscMgmtIpSet.numOfIps++;
+    taosGetFqdnPortFromEp(first, mgmtEpSet->fqdn[mgmtEpSet->numOfEps], &(mgmtEpSet->port[mgmtEpSet->numOfEps]));
+    mgmtEpSet->numOfEps++;
   }
 
   if (second && second[0] != 0) {
@@ -2163,11 +2166,11 @@ int tscSetMgmtIpListFromCfg(const char *first, const char *second) {
       terrno = TSDB_CODE_TSC_INVALID_FQDN;
       return -1;
     }
-    taosGetFqdnPortFromEp(second, tscMgmtIpSet.fqdn[tscMgmtIpSet.numOfIps], &tscMgmtIpSet.port[tscMgmtIpSet.numOfIps]);
-    tscMgmtIpSet.numOfIps++;
+    taosGetFqdnPortFromEp(second, mgmtEpSet->fqdn[mgmtEpSet->numOfEps], &(mgmtEpSet->port[mgmtEpSet->numOfEps]));
+    mgmtEpSet->numOfEps++;
   }
 
-  if ( tscMgmtIpSet.numOfIps == 0) {
+  if (mgmtEpSet->numOfEps == 0) {
     terrno = TSDB_CODE_TSC_INVALID_FQDN;
     return -1;
   }

src/common/inc/tglobal.h

@@ -174,6 +174,7 @@ bool taosCheckGlobalCfg();
 void taosSetAllDebugFlag();
 bool taosCfgDynamicOptions(char *msg);
 int  taosGetFqdnPortFromEp(const char *ep, char *fqdn, uint16_t *port);
+bool taosCheckBalanceCfgOptions(const char *option, int32_t *vnodeId, int32_t *dnodeId);
  
 #ifdef __cplusplus
 }

src/common/src/tdataformat.c

@@ -318,7 +318,7 @@ SDataCols *tdNewDataCols(int maxRowSize, int maxCols, int maxRows) {
   pCols->maxPoints = maxRows;
   pCols->bufSize = maxRowSize * maxRows;
 
-  pCols->buf = calloc(1, pCols->bufSize);
+  pCols->buf = malloc(pCols->bufSize);
   if (pCols->buf == NULL) {
     free(pCols);
     return NULL;

src/common/src/tglobal.c

@@ -198,6 +198,7 @@ int32_t tsdbDebugFlag = 131;
 
 int32_t (*monitorStartSystemFp)() = NULL;
 void (*monitorStopSystemFp)() = NULL;
+void (*monitorExecuteSQLFp)(char *sql) = NULL;
 
 static pthread_once_t tsInitGlobalCfgOnce = PTHREAD_ONCE_INIT;
 
@@ -252,11 +253,15 @@ bool taosCfgDynamicOptions(char *msg) {
         if (monitorStartSystemFp) {
           (*monitorStartSystemFp)();
           uInfo("monitor is enabled");
+        } else {
+          uError("monitor can't be updated, for monitor not initialized");
         }
       } else {
         if (monitorStopSystemFp) {
           (*monitorStopSystemFp)();
           uInfo("monitor is disabled");
+        } else {
+          uError("monitor can't be updated, for monitor not initialized");
         }
       }
       return true;
@@ -276,7 +281,12 @@ bool taosCfgDynamicOptions(char *msg) {
   }
 
   if (strncasecmp(option, "resetQueryCache", 15) == 0) {
-    uError("reset query cache can't be executed, for monitor not initialized");
+    if (monitorExecuteSQLFp) {
+      (*monitorExecuteSQLFp)("resetQueryCache");
+      uInfo("resetquerycache is executed");
+    } else {
+      uError("resetquerycache can't be executed, for monitor not started");
+    }
   }
 
   return false;
@@ -1300,3 +1310,32 @@ int taosGetFqdnPortFromEp(const char *ep, char *fqdn, uint16_t *port) {
 
   return 0; 
 }
+
+/*
+ * alter dnode 1 balance "vnode:1-dnode:2"
+ */
+
+bool taosCheckBalanceCfgOptions(const char *option, int32_t *vnodeId, int32_t *dnodeId) {
+  int len = strlen(option);
+  if (strncasecmp(option, "vnode:", 6) != 0) {
+    return false;
+  }
+
+  int pos = 0;
+  for (; pos < len; ++pos) {
+    if (option[pos] == '-') break;
+  }
+
+  if (++pos >= len) return false;
+  if (strncasecmp(option + pos, "dnode:", 6) != 0) {
+    return false;
+  }
+
+  *vnodeId = strtol(option + 6, NULL, 10);
+  *dnodeId = strtol(option + pos + 6, NULL, 10);
+  if (*vnodeId <= 1 || *dnodeId <= 0) {
+    return false;
+  }
+
+  return true;
+}
\ No newline at end of file

src/connector/grafana/tdengine/package-lock.json

 {
   "name": "TDengine",
-  "version": "1.0.0",
+  "version": "2.0.0",
   "lockfileVersion": 1,
   "requires": true,
   "dependencies": {
@@ -2302,9 +2302,9 @@
       }
     },
     "lodash": {
-      "version": "4.17.13",
-      "resolved": "https://registry.npmjs.org/lodash/-/lodash-4.17.13.tgz",
-      "integrity": "sha512-vm3/XWXfWtRua0FkUyEHBZy8kCPjErNBT9fJx8Zvs+U6zjqPbTUOpkaoum3O5uiA8sm+yNMHXfYkTUHFoMxFNA=="
+      "version": "4.17.19",
+      "resolved": "https://registry.npmjs.org/lodash/-/lodash-4.17.19.tgz",
+      "integrity": "sha512-JNvd8XER9GQX0v2qJgsaN/mzFCNA5BRe/j8JN9d+tWyGLSodKQHKFicdwNYzWwI3wjRnaKPsGj1XkBjx/F96DQ=="
     },
     "log-symbols": {
       "version": "2.2.0",

src/connector/grafana/tdengine/package.json

@@ -38,7 +38,7 @@
     "q": "^1.5.0"
   },
   "dependencies": {
-    "lodash": "^4.17.13",
+    "lodash": "^4.17.19",
     "yarn": "^1.22.0"
   },
   "homepage": "https://github.com/taosdata/TDengine/tree/develop/src/connector/grafana/tdengine"

src/connector/grafana/tdengine/yarn.lock

@@ -1888,14 +1888,10 @@ locate-path@^3.0.0:
     p-locate "^3.0.0"
     path-exists "^3.0.0"
 
-lodash@^4.17.10, lodash@^4.17.13, lodash@^4.17.4, lodash@^4.2.0, lodash@~4.17.10, lodash@~4.17.5:
-  version "4.17.13"
-  resolved "https://registry.yarnpkg.com/lodash/-/lodash-4.17.13.tgz#0bdc3a6adc873d2f4e0c4bac285df91b64fc7b93"
-
-lodash@^4.17.15:
-  version "4.17.15"
-  resolved "https://registry.yarnpkg.com/lodash/-/lodash-4.17.15.tgz#b447f6670a0455bbfeedd11392eff330ea097548"
-  integrity sha512-8xOcRHvCjnocdS5cpwXQXVzmmh5e5+saE2QGoeQmbKmRS6J3VQppPOIt0MnmE+4xlZoumy0GPG0D0MVIQbNA1A==
+lodash@^4.17.10, lodash@^4.17.15, lodash@^4.17.19, lodash@^4.17.4, lodash@^4.2.0, lodash@~4.17.10, lodash@~4.17.5:
+  version "4.17.19"
+  resolved "https://registry.yarnpkg.com/lodash/-/lodash-4.17.19.tgz#e48ddedbe30b3321783c5b4301fbd353bc1e4a4b"
+  integrity sha512-JNvd8XER9GQX0v2qJgsaN/mzFCNA5BRe/j8JN9d+tWyGLSodKQHKFicdwNYzWwI3wjRnaKPsGj1XkBjx/F96DQ==
 
 log-symbols@2.2.0:
   version "2.2.0"

src/connector/jdbc/src/main/java/com/taosdata/jdbc/utils/TDNode.java

 package com.taosdata.jdbc.utils;
 
+import java.io.BufferedReader;
 import java.io.File;
+import java.io.InputStreamReader;
 import java.util.*;
 import java.util.concurrent.TimeUnit;
 
@@ -31,6 +33,10 @@ public class TDNode {
         this.testCluster = testCluster;
     }
 
+    public void setRunning(int running) {
+        this.running = running;
+    }
+
     public void searchTaosd(File dir, ArrayList<String> taosdPath) {        
         File[] fileList = dir.listFiles();
         
@@ -102,15 +108,46 @@ public class TDNode {
         this.running = 1;        
     }
 
-    public void stop() {
-        String toBeKilled = "taosd";
+    public Integer getTaosdPid() {
+        String cmd = "ps -ef|grep -w taosd| grep -v grep | awk '{print $2}'"; 
+        String[] cmds = {"sh", "-c", cmd};
+        try {
+            Process process = Runtime.getRuntime().exec(cmds);
+            BufferedReader reader = new BufferedReader(new InputStreamReader(process.getInputStream()));
+            String line = null;
+            Integer res = null;
+            while((line = reader.readLine()) != null) {
+                if(!line.isEmpty()) {
+                    res = Integer.valueOf(line);
+                    break;
+                }
+            }
+            
+            return res;
+        } catch (Exception e) {
+            e.printStackTrace();
+        } 
+        return null;
+    }
+
+    public void stop() {        
 
         if (this.running != 0) {
-            String killCmd = "pkill -kill -x " + toBeKilled;
-            String[] killCmds = {"sh", "-c", killCmd};
-            try {                                
-                Runtime.getRuntime().exec(killCmds).waitFor();
+            Integer pid = null;                        
+            while((pid = getTaosdPid()) != null) {
+                
+                String killCmd = "kill -term " + pid;                                      
+                String[] killCmds = {"sh", "-c", killCmd};
+                try {                                
+                    Runtime.getRuntime().exec(killCmds).waitFor();                    
+
+                    TimeUnit.SECONDS.sleep(2);
+                } catch (Exception e) {
+                    e.printStackTrace();
+                }                
+            }
 
+            try {
                 for(int port = 6030; port < 6041; port ++) {
                     String fuserCmd = "fuser -k -n tcp " + port;
                     Runtime.getRuntime().exec(fuserCmd).waitFor();
@@ -120,7 +157,7 @@ public class TDNode {
             }
 
             this.running = 0;
-            System.out.println("dnode:" + this.index + " is stopped by pkill");
+            System.out.println("dnode:" + this.index + " is stopped by kill -term");
         }
     }
 

src/connector/jdbc/src/main/java/com/taosdata/jdbc/utils/TDNodes.java

@@ -14,33 +14,6 @@ public class TDNodes {
         }               
     }
 
-    public void setPath(String path) {
-        try {
-            String killCmd = "pkill -kill -x taosd";
-            String[] killCmds = {"sh", "-c", killCmd};
-            Runtime.getRuntime().exec(killCmds).waitFor();        
-
-            String binPath = System.getProperty("user.dir");
-            binPath += "/../../../debug";
-            System.out.println("binPath: " + binPath);
-        
-            File file = new File(path);
-            binPath = file.getCanonicalPath();
-            System.out.println("binPath real path: " + binPath);
-                
-            if(path.isEmpty()){
-                file = new File(path + "/../../");                
-                path = file.getCanonicalPath();
-            }            
-
-            for(int i = 0; i < tdNodes.size(); i++) {
-                tdNodes.get(i).setPath(path);
-            }
-        } catch (Exception e) {
-            e.printStackTrace();
-        }
-    }
-    
     public void setTestCluster(boolean testCluster) {
         this.testCluster = testCluster;
     }
@@ -70,6 +43,11 @@ public class TDNodes {
         check(index);
         tdNodes.get(index - 1).setCfgConfig(option, value);
     }
+
+    public TDNode getTDNode(int index) {
+        check(index);
+        return tdNodes.get(index - 1);
+    }
     
     public void start(int index) {
         check(index);

src/connector/jdbc/src/test/java/com/taosdata/jdbc/BaseTest.java

 package com.taosdata.jdbc;
 
-import java.io.File;
 import com.taosdata.jdbc.utils.TDNodes;
 
 import org.junit.AfterClass;
@@ -8,31 +7,30 @@ import org.junit.BeforeClass;
 
 public class BaseTest {
 
-    private static boolean testCluster = false;
-    private static String deployPath = System.getProperty("user.dir");         
-    private static TDNodes tdNodes = new TDNodes();    
+    private static boolean testCluster = false;           
+    private static TDNodes nodes = new TDNodes();    
     
     
     @BeforeClass
     public static void setupEnv() {
-        try{
-            File file = new File(deployPath + "/../../../");
-            String rootPath = file.getCanonicalPath();
-                       
-            tdNodes.setPath(rootPath);
-            tdNodes.setTestCluster(testCluster);            
+        try{                         
+            if (nodes.getTDNode(1).getTaosdPid() != null) {
+                System.out.println("Kill taosd before running JDBC test");
+                nodes.getTDNode(1).setRunning(1);
+                nodes.stop(1);
+            }
 
-            tdNodes.deploy(1);
-            tdNodes.start(1);  
+            nodes.setTestCluster(testCluster);  
+            nodes.deploy(1);
+            nodes.start(1);  
 
         } catch (Exception e) {
-            e.printStackTrace();
-            System.out.println("Base Test Exception");
+            e.printStackTrace();            
         }
     }
 
     @AfterClass
     public static void cleanUpEnv() {
-        tdNodes.stop(1);
+        nodes.stop(1);
     }
 }
\ No newline at end of file

src/cq/src/cqMain.c

@@ -103,9 +103,6 @@ void cqClose(void *handle) {
   SCqContext *pContext = handle;
   if (handle == NULL) return;
 
-  taosTmrCleanUp(pContext->tmrCtrl);
-  pContext->tmrCtrl = NULL;
-
   // stop all CQs
   cqStop(pContext);
 
@@ -125,6 +122,9 @@ void cqClose(void *handle) {
 
   pthread_mutex_destroy(&pContext->mutex);
 
+  taosTmrCleanUp(pContext->tmrCtrl);
+  pContext->tmrCtrl = NULL;
+
   cTrace("vgId:%d, CQ is closed", pContext->vgId);
   free(pContext);
 }

src/dnode/inc/dnodeMgmt.h

@@ -35,8 +35,8 @@ void*   dnodeGetVnodeTsdb(void *pVnode);
 void    dnodeReleaseVnode(void *pVnode);
 
 void    dnodeSendRedirectMsg(SRpcMsg *rpcMsg, bool forShell);
-void    dnodeGetMnodeIpSetForPeer(void *ipSet);
-void    dnodeGetMnodeIpSetForShell(void *ipSet);
+void    dnodeGetMnodeEpSetForPeer(void *epSet);
+void    dnodeGetMnodeEpSetForShell(void *epSet);
 
 #ifdef __cplusplus
 }

src/dnode/src/dnodeMgmt.c

@@ -52,7 +52,7 @@ void *          tsDnodeTmr = NULL;
 static void *   tsStatusTimer = NULL;
 static uint32_t tsRebootTime;
 
-static SRpcIpSet     tsDMnodeIpSet = {0};
+static SRpcEpSet     tsDMnodeEpSet = {0};
 static SDMMnodeInfos tsDMnodeInfos = {0};
 static SDMDnodeCfg   tsDnodeCfg = {0};
 static taos_qset     tsMgmtQset = NULL;
@@ -90,21 +90,21 @@ int32_t dnodeInitMgmt() {
   tsRebootTime = taosGetTimestampSec();
 
   if (!dnodeReadMnodeInfos()) {
-    memset(&tsDMnodeIpSet, 0, sizeof(SRpcIpSet));
+    memset(&tsDMnodeEpSet, 0, sizeof(SRpcEpSet));
     memset(&tsDMnodeInfos, 0, sizeof(SDMMnodeInfos));
 
-    tsDMnodeIpSet.numOfIps = 1;
-    taosGetFqdnPortFromEp(tsFirst, tsDMnodeIpSet.fqdn[0], &tsDMnodeIpSet.port[0]);
+    tsDMnodeEpSet.numOfEps = 1;
+    taosGetFqdnPortFromEp(tsFirst, tsDMnodeEpSet.fqdn[0], &tsDMnodeEpSet.port[0]);
     
     if (strcmp(tsSecond, tsFirst) != 0) {
-      tsDMnodeIpSet.numOfIps = 2;
-      taosGetFqdnPortFromEp(tsSecond, tsDMnodeIpSet.fqdn[1], &tsDMnodeIpSet.port[1]);
+      tsDMnodeEpSet.numOfEps = 2;
+      taosGetFqdnPortFromEp(tsSecond, tsDMnodeEpSet.fqdn[1], &tsDMnodeEpSet.port[1]);
     }
   } else {
-    tsDMnodeIpSet.inUse = tsDMnodeInfos.inUse;
-    tsDMnodeIpSet.numOfIps = tsDMnodeInfos.nodeNum;
+    tsDMnodeEpSet.inUse = tsDMnodeInfos.inUse;
+    tsDMnodeEpSet.numOfEps = tsDMnodeInfos.nodeNum;
     for (int32_t i = 0; i < tsDMnodeInfos.nodeNum; i++) {
-      taosGetFqdnPortFromEp(tsDMnodeInfos.nodeInfos[i].nodeEp, tsDMnodeIpSet.fqdn[i], &tsDMnodeIpSet.port[i]);
+      taosGetFqdnPortFromEp(tsDMnodeInfos.nodeInfos[i].nodeEp, tsDMnodeEpSet.fqdn[i], &tsDMnodeEpSet.port[i]);
     }
   }
 
@@ -450,27 +450,27 @@ static int32_t dnodeProcessConfigDnodeMsg(SRpcMsg *pMsg) {
   return taosCfgDynamicOptions(pCfg->config);
 }
 
-void dnodeUpdateMnodeIpSetForPeer(SRpcIpSet *pIpSet) {
-  dInfo("mnode IP list for is changed, numOfIps:%d inUse:%d", pIpSet->numOfIps, pIpSet->inUse);
-  for (int i = 0; i < pIpSet->numOfIps; ++i) {
-    pIpSet->port[i] -= TSDB_PORT_DNODEDNODE;
-    dInfo("mnode index:%d %s:%u", i, pIpSet->fqdn[i], pIpSet->port[i])
+void dnodeUpdateMnodeEpSetForPeer(SRpcEpSet *pEpSet) {
+  dInfo("mnode EP list for is changed, numOfEps:%d inUse:%d", pEpSet->numOfEps, pEpSet->inUse);
+  for (int i = 0; i < pEpSet->numOfEps; ++i) {
+    pEpSet->port[i] -= TSDB_PORT_DNODEDNODE;
+    dInfo("mnode index:%d %s:%u", i, pEpSet->fqdn[i], pEpSet->port[i])
   }
 
-  tsDMnodeIpSet = *pIpSet;
+  tsDMnodeEpSet = *pEpSet;
 }
 
-void dnodeGetMnodeIpSetForPeer(void *ipSetRaw) {
-  SRpcIpSet *ipSet = ipSetRaw;
-  *ipSet = tsDMnodeIpSet;
+void dnodeGetMnodeEpSetForPeer(void *epSetRaw) {
+  SRpcEpSet *epSet = epSetRaw;
+  *epSet = tsDMnodeEpSet;
 
-  for (int i=0; i<ipSet->numOfIps; ++i) 
-    ipSet->port[i] += TSDB_PORT_DNODEDNODE;
+  for (int i=0; i<epSet->numOfEps; ++i) 
+    epSet->port[i] += TSDB_PORT_DNODEDNODE;
 }
 
-void dnodeGetMnodeIpSetForShell(void *ipSetRaw) {
-  SRpcIpSet *ipSet = ipSetRaw;
-  *ipSet = tsDMnodeIpSet;
+void dnodeGetMnodeEpSetForShell(void *epSetRaw) {
+  SRpcEpSet *epSet = epSetRaw;
+  *epSet = tsDMnodeEpSet;
 }
 
 static void dnodeProcessStatusRsp(SRpcMsg *pMsg) {
@@ -492,6 +492,7 @@ static void dnodeProcessStatusRsp(SRpcMsg *pMsg) {
   pCfg->numOfVnodes  = htonl(pCfg->numOfVnodes);
   pCfg->moduleStatus = htonl(pCfg->moduleStatus);
   pCfg->dnodeId      = htonl(pCfg->dnodeId);
+  pCfg->clusterId    = htonl(pCfg->clusterId);
 
   for (int32_t i = 0; i < pMnodes->nodeNum; ++i) {
     SDMMnodeInfo *pMnodeInfo = &pMnodes->nodeInfos[i];
@@ -536,10 +537,10 @@ static void dnodeUpdateMnodeInfos(SDMMnodeInfos *pMnodes) {
     dInfo("mnode index:%d, %s", tsDMnodeInfos.nodeInfos[i].nodeId, tsDMnodeInfos.nodeInfos[i].nodeEp);
   }
 
-  tsDMnodeIpSet.inUse = tsDMnodeInfos.inUse;
-  tsDMnodeIpSet.numOfIps = tsDMnodeInfos.nodeNum;
+  tsDMnodeEpSet.inUse = tsDMnodeInfos.inUse;
+  tsDMnodeEpSet.numOfEps = tsDMnodeInfos.nodeNum;
   for (int32_t i = 0; i < tsDMnodeInfos.nodeNum; i++) {
-    taosGetFqdnPortFromEp(tsDMnodeInfos.nodeInfos[i].nodeEp, tsDMnodeIpSet.fqdn[i], &tsDMnodeIpSet.port[i]);
+    taosGetFqdnPortFromEp(tsDMnodeInfos.nodeInfos[i].nodeEp, tsDMnodeEpSet.fqdn[i], &tsDMnodeEpSet.port[i]);
   }
 
   dnodeSaveMnodeInfos();
@@ -549,10 +550,10 @@ static void dnodeUpdateMnodeInfos(SDMMnodeInfos *pMnodes) {
 static bool dnodeReadMnodeInfos() {
   char ipFile[TSDB_FILENAME_LEN*2] = {0};
   
-  sprintf(ipFile, "%s/mnodeIpList.json", tsDnodeDir);
+  sprintf(ipFile, "%s/mnodeEpSet.json", tsDnodeDir);
   FILE *fp = fopen(ipFile, "r");
   if (!fp) {
-    dDebug("failed to read mnodeIpList.json, file not exist");
+    dDebug("failed to read mnodeEpSet.json, file not exist");
     return false;
   }
 
@@ -563,40 +564,40 @@ static bool dnodeReadMnodeInfos() {
   if (len <= 0) {
     free(content);
     fclose(fp);
-    dError("failed to read mnodeIpList.json, content is null");
+    dError("failed to read mnodeEpSet.json, content is null");
     return false;
   }
 
   content[len] = 0;
   cJSON* root = cJSON_Parse(content);
   if (root == NULL) {
-    dError("failed to read mnodeIpList.json, invalid json format");
+    dError("failed to read mnodeEpSet.json, invalid json format");
     goto PARSE_OVER;
   }
 
   cJSON* inUse = cJSON_GetObjectItem(root, "inUse");
   if (!inUse || inUse->type != cJSON_Number) {
-    dError("failed to read mnodeIpList.json, inUse not found");
+    dError("failed to read mnodeEpSet.json, inUse not found");
     goto PARSE_OVER;
   }
   tsDMnodeInfos.inUse = inUse->valueint;
 
   cJSON* nodeNum = cJSON_GetObjectItem(root, "nodeNum");
   if (!nodeNum || nodeNum->type != cJSON_Number) {
-    dError("failed to read mnodeIpList.json, nodeNum not found");
+    dError("failed to read mnodeEpSet.json, nodeNum not found");
     goto PARSE_OVER;
   }
   tsDMnodeInfos.nodeNum = nodeNum->valueint;
 
   cJSON* nodeInfos = cJSON_GetObjectItem(root, "nodeInfos");
   if (!nodeInfos || nodeInfos->type != cJSON_Array) {
-    dError("failed to read mnodeIpList.json, nodeInfos not found");
+    dError("failed to read mnodeEpSet.json, nodeInfos not found");
     goto PARSE_OVER;
   }
 
   int size = cJSON_GetArraySize(nodeInfos);
   if (size != tsDMnodeInfos.nodeNum) {
-    dError("failed to read mnodeIpList.json, nodeInfos size not matched");
+    dError("failed to read mnodeEpSet.json, nodeInfos size not matched");
     goto PARSE_OVER;
   }
 
@@ -606,14 +607,14 @@ static bool dnodeReadMnodeInfos() {
 
     cJSON *nodeId = cJSON_GetObjectItem(nodeInfo, "nodeId");
     if (!nodeId || nodeId->type != cJSON_Number) {
-      dError("failed to read mnodeIpList.json, nodeId not found");
+      dError("failed to read mnodeEpSet.json, nodeId not found");
       goto PARSE_OVER;
     }
     tsDMnodeInfos.nodeInfos[i].nodeId = nodeId->valueint;
 
     cJSON *nodeEp = cJSON_GetObjectItem(nodeInfo, "nodeEp");
     if (!nodeEp || nodeEp->type != cJSON_String || nodeEp->valuestring == NULL) {
-      dError("failed to read mnodeIpList.json, nodeName not found");
+      dError("failed to read mnodeEpSet.json, nodeName not found");
       goto PARSE_OVER;
     }
     strncpy(tsDMnodeInfos.nodeInfos[i].nodeEp, nodeEp->valuestring, TSDB_EP_LEN);
@@ -621,7 +622,7 @@ static bool dnodeReadMnodeInfos() {
 
   ret = true;
 
-  dInfo("read mnode iplist successed, numOfIps:%d inUse:%d", tsDMnodeInfos.nodeNum, tsDMnodeInfos.inUse);
+  dInfo("read mnode epSet successed, numOfEps:%d inUse:%d", tsDMnodeInfos.nodeNum, tsDMnodeInfos.inUse);
   for (int32_t i = 0; i < tsDMnodeInfos.nodeNum; i++) {
     dInfo("mnode:%d, %s", tsDMnodeInfos.nodeInfos[i].nodeId, tsDMnodeInfos.nodeInfos[i].nodeEp);
   }
@@ -635,7 +636,7 @@ PARSE_OVER:
 
 static void dnodeSaveMnodeInfos() {
   char ipFile[TSDB_FILENAME_LEN] = {0};
-  sprintf(ipFile, "%s/mnodeIpList.json", tsDnodeDir);
+  sprintf(ipFile, "%s/mnodeEpSet.json", tsDnodeDir);
   FILE *fp = fopen(ipFile, "w");
   if (!fp) return;
 
@@ -663,11 +664,11 @@ static void dnodeSaveMnodeInfos() {
   fclose(fp);
   free(content);
   
-  dInfo("save mnode iplist successed");
+  dInfo("save mnode epSet successed");
 }
 
 char *dnodeGetMnodeMasterEp() {
-  return tsDMnodeInfos.nodeInfos[tsDMnodeIpSet.inUse].nodeEp;
+  return tsDMnodeInfos.nodeInfos[tsDMnodeEpSet.inUse].nodeEp;
 }
 
 void* dnodeGetMnodeInfos() {
@@ -697,6 +698,7 @@ static void dnodeSendStatusMsg(void *handle, void *tmrId) {
   //strcpy(pStatus->dnodeName, tsDnodeName);
   pStatus->version          = htonl(tsVersion);
   pStatus->dnodeId          = htonl(tsDnodeCfg.dnodeId);
+  pStatus->clusterId        = htonl(tsDnodeCfg.clusterId);
   strcpy(pStatus->dnodeEp, tsLocalEp);
   pStatus->lastReboot       = htonl(tsRebootTime);
   pStatus->numOfCores       = htons((uint16_t) tsNumOfCores);
@@ -726,9 +728,9 @@ static void dnodeSendStatusMsg(void *handle, void *tmrId) {
     .msgType = TSDB_MSG_TYPE_DM_STATUS
   };
 
-  SRpcIpSet ipSet;
-  dnodeGetMnodeIpSetForPeer(&ipSet);
-  dnodeSendMsgToDnode(&ipSet, &rpcMsg);
+  SRpcEpSet epSet;
+  dnodeGetMnodeEpSetForPeer(&epSet);
+  dnodeSendMsgToDnode(&epSet, &rpcMsg);
 }
 
 static bool dnodeReadDnodeCfg() {
@@ -767,6 +769,13 @@ static bool dnodeReadDnodeCfg() {
   }
   tsDnodeCfg.dnodeId = dnodeId->valueint;
 
+  cJSON* clusterId = cJSON_GetObjectItem(root, "clusterId");
+  if (!clusterId || clusterId->type != cJSON_Number) {
+    dError("failed to read dnodeCfg.json, clusterId not found");
+    goto PARSE_CFG_OVER;
+  }
+  tsDnodeCfg.clusterId = clusterId->valueint;
+
   ret = true;
 
   dInfo("read numOfVnodes successed, dnodeId:%d", tsDnodeCfg.dnodeId);
@@ -790,7 +799,8 @@ static void dnodeSaveDnodeCfg() {
   char *  content = calloc(1, maxLen + 1);
 
   len += snprintf(content + len, maxLen - len, "{\n");
-  len += snprintf(content + len, maxLen - len, "  \"dnodeId\": %d\n", tsDnodeCfg.dnodeId);
+  len += snprintf(content + len, maxLen - len, "  \"dnodeId\": %d,\n", tsDnodeCfg.dnodeId);
+  len += snprintf(content + len, maxLen - len, "  \"clusterId\": %d\n", tsDnodeCfg.clusterId);
   len += snprintf(content + len, maxLen - len, "}\n"); 
 
   fwrite(content, 1, len, fp);
@@ -803,8 +813,9 @@ static void dnodeSaveDnodeCfg() {
 
 void dnodeUpdateDnodeCfg(SDMDnodeCfg *pCfg) {
   if (tsDnodeCfg.dnodeId == 0) {
-    dInfo("dnodeId is set to %d", pCfg->dnodeId);  
+    dInfo("dnodeId is set to %d, clusterId is set to %d", pCfg->dnodeId, pCfg->clusterId);  
     tsDnodeCfg.dnodeId = pCfg->dnodeId;
+    tsDnodeCfg.clusterId = pCfg->clusterId;
     dnodeSaveDnodeCfg();
   }
 }
@@ -817,20 +828,20 @@ void dnodeSendRedirectMsg(SRpcMsg *rpcMsg, bool forShell) {
   SRpcConnInfo connInfo = {0};
   rpcGetConnInfo(rpcMsg->handle, &connInfo);
 
-  SRpcIpSet ipSet = {0};
+  SRpcEpSet epSet = {0};
   if (forShell) {
-    dnodeGetMnodeIpSetForShell(&ipSet);
+    dnodeGetMnodeEpSetForShell(&epSet);
   } else {
-    dnodeGetMnodeIpSetForPeer(&ipSet);
+    dnodeGetMnodeEpSetForPeer(&epSet);
   }
   
-  dDebug("msg:%s will be redirected, dnodeIp:%s user:%s, numOfIps:%d inUse:%d", taosMsg[rpcMsg->msgType],
-         taosIpStr(connInfo.clientIp), connInfo.user, ipSet.numOfIps, ipSet.inUse);
+  dDebug("msg:%s will be redirected, dnodeIp:%s user:%s, numOfEps:%d inUse:%d", taosMsg[rpcMsg->msgType],
+         taosIpStr(connInfo.clientIp), connInfo.user, epSet.numOfEps, epSet.inUse);
 
-  for (int i = 0; i < ipSet.numOfIps; ++i) {
-    dDebug("mnode index:%d %s:%d", i, ipSet.fqdn[i], ipSet.port[i]);
-    ipSet.port[i] = htons(ipSet.port[i]);
+  for (int i = 0; i < epSet.numOfEps; ++i) {
+    dDebug("mnode index:%d %s:%d", i, epSet.fqdn[i], epSet.port[i]);
+    epSet.port[i] = htons(epSet.port[i]);
   }
 
-  rpcSendRedirectRsp(rpcMsg->handle, &ipSet);
+  rpcSendRedirectRsp(rpcMsg->handle, &epSet);
 }

src/dnode/src/dnodePeer.c

@@ -29,11 +29,11 @@
 #include "dnodeVWrite.h"
 #include "dnodeMPeer.h"
 
-extern void dnodeUpdateMnodeIpSetForPeer(SRpcIpSet *pIpSet);
+extern void dnodeUpdateMnodeEpSetForPeer(SRpcEpSet *pEpSet);
 static void (*dnodeProcessReqMsgFp[TSDB_MSG_TYPE_MAX])(SRpcMsg *);
-static void dnodeProcessReqMsgFromDnode(SRpcMsg *pMsg, SRpcIpSet *);
+static void dnodeProcessReqMsgFromDnode(SRpcMsg *pMsg, SRpcEpSet *);
 static void (*dnodeProcessRspMsgFp[TSDB_MSG_TYPE_MAX])(SRpcMsg *rpcMsg);
-static void dnodeProcessRspFromDnode(SRpcMsg *pMsg, SRpcIpSet *pIpSet);
+static void dnodeProcessRspFromDnode(SRpcMsg *pMsg, SRpcEpSet *pEpSet);
 static void *tsDnodeServerRpc = NULL;
 static void *tsDnodeClientRpc = NULL;
 
@@ -83,7 +83,7 @@ void dnodeCleanupServer() {
   }
 }
 
-static void dnodeProcessReqMsgFromDnode(SRpcMsg *pMsg, SRpcIpSet *pIpSet) {
+static void dnodeProcessReqMsgFromDnode(SRpcMsg *pMsg, SRpcEpSet *pEpSet) {
   SRpcMsg rspMsg = {
     .handle  = pMsg->handle,
     .pCont   = NULL,
@@ -148,9 +148,9 @@ void dnodeCleanupClient() {
   }
 }
 
-static void dnodeProcessRspFromDnode(SRpcMsg *pMsg, SRpcIpSet *pIpSet) {
-  if (pMsg->msgType == TSDB_MSG_TYPE_DM_STATUS_RSP && pIpSet) {
-    dnodeUpdateMnodeIpSetForPeer(pIpSet);
+static void dnodeProcessRspFromDnode(SRpcMsg *pMsg, SRpcEpSet *pEpSet) {
+  if (pMsg->msgType == TSDB_MSG_TYPE_DM_STATUS_RSP && pEpSet) {
+    dnodeUpdateMnodeEpSetForPeer(pEpSet);
   }
 
   if (dnodeProcessRspMsgFp[pMsg->msgType]) {    
@@ -166,12 +166,12 @@ void dnodeAddClientRspHandle(uint8_t msgType, void (*fp)(SRpcMsg *rpcMsg)) {
   dnodeProcessRspMsgFp[msgType] = fp;
 }
 
-void dnodeSendMsgToDnode(SRpcIpSet *ipSet, SRpcMsg *rpcMsg) {
-  rpcSendRequest(tsDnodeClientRpc, ipSet, rpcMsg);
+void dnodeSendMsgToDnode(SRpcEpSet *epSet, SRpcMsg *rpcMsg) {
+  rpcSendRequest(tsDnodeClientRpc, epSet, rpcMsg);
 }
 
 void dnodeSendMsgToDnodeRecv(SRpcMsg *rpcMsg, SRpcMsg *rpcRsp) {
-  SRpcIpSet ipSet = {0};
-  dnodeGetMnodeIpSetForPeer(&ipSet);
-  rpcSendRecv(tsDnodeClientRpc, &ipSet, rpcMsg, rpcRsp);
+  SRpcEpSet epSet = {0};
+  dnodeGetMnodeEpSetForPeer(&epSet);
+  rpcSendRecv(tsDnodeClientRpc, &epSet, rpcMsg, rpcRsp);
 }

src/dnode/src/dnodeShell.c

@@ -31,7 +31,7 @@
 #include "dnodeShell.h"
 
 static void  (*dnodeProcessShellMsgFp[TSDB_MSG_TYPE_MAX])(SRpcMsg *);
-static void    dnodeProcessMsgFromShell(SRpcMsg *pMsg, SRpcIpSet *);
+static void    dnodeProcessMsgFromShell(SRpcMsg *pMsg, SRpcEpSet *);
 static int     dnodeRetrieveUserAuthInfo(char *user, char *spi, char *encrypt, char *secret, char *ckey);
 static void  * tsDnodeShellRpc = NULL;
 static int32_t tsDnodeQueryReqNum  = 0;
@@ -108,7 +108,7 @@ void dnodeCleanupShell() {
   }
 }
 
-void dnodeProcessMsgFromShell(SRpcMsg *pMsg, SRpcIpSet *pIpSet) {
+void dnodeProcessMsgFromShell(SRpcMsg *pMsg, SRpcEpSet *pEpSet) {
   SRpcMsg rpcMsg = {
     .handle  = pMsg->handle,
     .pCont   = NULL,

src/inc/dnode.h

@@ -39,13 +39,13 @@ SDnodeStatisInfo dnodeGetStatisInfo();
 
 bool    dnodeIsFirstDeploy();
 char *  dnodeGetMnodeMasterEp();
-void    dnodeGetMnodeIpSetForPeer(void *ipSet);
-void    dnodeGetMnodeIpSetForShell(void *ipSet);
+void    dnodeGetMnodeEpSetForPeer(void *epSet);
+void    dnodeGetMnodeEpSetForShell(void *epSet);
 void *  dnodeGetMnodeInfos();
 int32_t dnodeGetDnodeId();
 
 void  dnodeAddClientRspHandle(uint8_t msgType, void (*fp)(SRpcMsg *rpcMsg));
-void  dnodeSendMsgToDnode(SRpcIpSet *ipSet, SRpcMsg *rpcMsg);
+void  dnodeSendMsgToDnode(SRpcEpSet *epSet, SRpcMsg *rpcMsg);
 void  dnodeSendMsgToDnodeRecv(SRpcMsg *rpcMsg, SRpcMsg *rpcRsp);
 void *dnodeSendCfgTableToRecv(int32_t vgId, int32_t sid);
 

src/inc/taos.h

@@ -67,8 +67,6 @@ DLL_EXPORT void  taos_init();
 DLL_EXPORT void  taos_cleanup();
 DLL_EXPORT int   taos_options(TSDB_OPTION option, const void *arg, ...);
 DLL_EXPORT TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port);
-DLL_EXPORT TAOS *taos_connect_c(const char *ip, uint8_t ipLen, const char *user, uint8_t userLen,
-                 const char *pass, uint8_t passLen, const char *db, uint8_t dbLen, uint16_t port);
 DLL_EXPORT void  taos_close(TAOS *taos);
 
 typedef struct TAOS_BIND {
@@ -90,7 +88,6 @@ TAOS_RES * taos_stmt_use_result(TAOS_STMT *stmt);
 int        taos_stmt_close(TAOS_STMT *stmt);
 
 DLL_EXPORT TAOS_RES *taos_query(TAOS *taos, const char *sql);
-DLL_EXPORT TAOS_RES *taos_query_c(TAOS *taos, const char *sql, uint32_t sqlLen);
 DLL_EXPORT TAOS_ROW taos_fetch_row(TAOS_RES *res);
 DLL_EXPORT int taos_result_precision(TAOS_RES *res);  // get the time precision of result
 DLL_EXPORT void taos_free_result(TAOS_RES *res);

src/inc/taosdef.h

@@ -377,6 +377,7 @@ void tsDataSwap(void *pLeft, void *pRight, int32_t type, int32_t size);
 #define TSDB_ORDER_ASC   1
 #define TSDB_ORDER_DESC  2
 
+#define TSDB_DEFAULT_CLUSTER_HASH_SIZE  1
 #define TSDB_DEFAULT_MNODES_HASH_SIZE   5
 #define TSDB_DEFAULT_DNODES_HASH_SIZE   10
 #define TSDB_DEFAULT_ACCOUNTS_HASH_SIZE 10

src/inc/taoserror.h

@@ -120,12 +120,18 @@ TAOS_DEFINE_ERROR(TSDB_CODE_MND_SDB_OBJ_NOT_THERE,        0, 0x0323, "sdb object
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_SDB_INVAID_META_ROW,      0, 0x0324, "sdb invalid meta row")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_SDB_INVAID_KEY_TYPE,      0, 0x0325, "sdb invalid key type")
 
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_ALREADY_EXIST,      0, 0x0330, "mnode dnode already exist")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_NOT_EXIST,          0, 0x0331, "mnode dnode not exist")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_VGROUP_NOT_EXIST,         0, 0x0332, "mnode vgroup not exist")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_NO_REMOVE_MASTER,         0, 0x0333, "mnode cant not remove master")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_NO_ENOUGH_DNODES,         0, 0x0334, "mnode no enough dnodes")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_CLUSTER_CFG_INCONSISTENT, 0, 0x0335, "mnode cluster cfg inconsistent")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_ALREADY_EXIST,      0, 0x0330, "dnode already exist")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_NOT_EXIST,          0, 0x0331, "dnode not exist")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_VGROUP_NOT_EXIST,         0, 0x0332, "vgroup not exist")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_NO_REMOVE_MASTER,         0, 0x0333, "cant not remove master")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_NO_ENOUGH_DNODES,         0, 0x0334, "no enough dnodes")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_CLUSTER_CFG_INCONSISTENT, 0, 0x0335, "cluster cfg inconsistent")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_DNODE_CFG_OPTION, 0, 0x0336, "invalid dnode cfg option")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_BALANCE_ENABLED,          0, 0x0337, "balance already enabled")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_VGROUP_NOT_IN_DNODE,      0, 0x0338, "vgroup not in dnode")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_VGROUP_ALREADY_IN_DNODE,  0, 0x0339, "vgroup already in dnode")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_DNODE_NOT_FREE,           0, 0x033A, "dnode not avaliable")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_CLUSTER_ID,       0, 0x033B, "cluster id not match")
 
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_ACCT_ALREADY_EXIST,       0, 0x0340, "mnode accounts already exist")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_ACCT,             0, 0x0341, "mnode invalid account")
@@ -227,6 +233,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_GRANT_CPU_LIMITED,            0, 0x080B, "grant cpu
 
 // sync
 TAOS_DEFINE_ERROR(TSDB_CODE_SYN_INVALID_CONFIG,           0, 0x0900, "sync invalid configuration")
+TAOS_DEFINE_ERROR(TSDB_CODE_SYN_NOT_ENABLED,              0, 0x0901, "sync module not enabled")
 
 // wal
 TAOS_DEFINE_ERROR(TSDB_CODE_WAL_APP_ERROR,                0, 0x1000, "wal app error")

src/inc/taosmsg.h

@@ -139,6 +139,7 @@ enum _mgmt_table {
   TSDB_MGMT_TABLE_GRANTS,
   TSDB_MGMT_TABLE_VNODES,
   TSDB_MGMT_TABLE_STREAMTABLES,
+  TSDB_MGMT_TABLE_CLUSTER,
   TSDB_MGMT_TABLE_MAX,
 };
 
@@ -149,6 +150,7 @@ enum _mgmt_table {
 
 #define TSDB_ALTER_TABLE_ADD_COLUMN        5
 #define TSDB_ALTER_TABLE_DROP_COLUMN       6
+#define TSDB_ALTER_TABLE_CHANGE_COLUMN     7
 
 #define TSDB_FILL_NONE      0
 #define TSDB_FILL_NULL      1
@@ -176,7 +178,7 @@ extern char *taosMsg[];
 typedef struct {
   char     fqdn[TSDB_FQDN_LEN];
   uint16_t port;
-} SIpAddr;
+} SEpAddr;
 
 typedef struct {
   int32_t numOfVnodes;
@@ -306,7 +308,7 @@ typedef struct {
   int8_t    reserved1;
   int8_t    reserved2;
   int32_t   connId;
-  SRpcIpSet ipList;
+  SRpcEpSet epSet;
 } SCMConnectRsp;
 
 typedef struct {
@@ -545,6 +547,7 @@ typedef struct {
 
 typedef struct {
   int32_t  dnodeId;
+  int32_t  clusterId;
   uint32_t moduleStatus;
   uint32_t numOfVnodes;
   uint32_t reserved;
@@ -585,6 +588,7 @@ typedef struct {
   uint16_t    openVnodes;
   uint16_t    numOfCores;
   float       diskAvailable;  // GB
+  int32_t     clusterId;
   uint8_t     alternativeRole;
   uint8_t     reserve2[15];
   SClusterCfg clusterCfg;
@@ -648,8 +652,8 @@ typedef struct SCMSTableVgroupMsg {
 
 typedef struct {
   int32_t   vgId;
-  int8_t    numOfIps;
-  SIpAddr   ipAddr[TSDB_MAX_REPLICA];
+  int8_t    numOfEps;
+  SEpAddr   epAddr[TSDB_MAX_REPLICA];
 } SCMVgroupInfo;
 
 typedef struct {
@@ -753,7 +757,7 @@ typedef struct {
   uint32_t  onlineDnodes;
   uint32_t  connId;
   int8_t    killConnection;
-  SRpcIpSet ipList;
+  SRpcEpSet epSet;
 } SCMHeartBeatRsp;
 
 typedef struct {

src/inc/tbalance.h

@@ -29,6 +29,7 @@ void    balanceAsyncNotify();
 void    balanceSyncNotify();
 void    balanceReset();
 int32_t balanceAllocVnodes(struct SVgObj *pVgroup);
+int32_t balanceAlterDnode(struct SDnodeObj *pDnode, int32_t vnodeId, int32_t dnodeId);
 int32_t balanceDropDnode(struct SDnodeObj *pDnode);
 
 #ifdef __cplusplus

src/inc/trpc.h

@@ -28,12 +28,17 @@ extern "C" {
 
 extern int tsRpcHeadSize;
 
-typedef struct SRpcIpSet {
+typedef struct SRpcEpSet {
   int8_t    inUse; 
-  int8_t    numOfIps;
+  int8_t    numOfEps;
   uint16_t  port[TSDB_MAX_REPLICA];
   char      fqdn[TSDB_MAX_REPLICA][TSDB_FQDN_LEN];
-} SRpcIpSet;
+} SRpcEpSet;
+
+typedef struct SRpcCorEpSet {
+  int32_t version; 
+  SRpcEpSet epSet; 
+} SRpcCorEpSet;
 
 typedef struct SRpcConnInfo {
   uint32_t  clientIp;
@@ -67,7 +72,7 @@ typedef struct SRpcInit {
   char *ckey;         // ciphering key
 
   // call back to process incoming msg, code shall be ignored by server app
-  void (*cfp)(SRpcMsg *, SRpcIpSet *);  
+  void (*cfp)(SRpcMsg *, SRpcEpSet *);  
 
   // call back to retrieve the client auth info, for server app only 
   int  (*afp)(char *tableId, char *spi, char *encrypt, char *secret, char *ckey);
@@ -78,11 +83,11 @@ void  rpcClose(void *);
 void *rpcMallocCont(int contLen);
 void  rpcFreeCont(void *pCont);
 void *rpcReallocCont(void *ptr, int contLen);
-void  rpcSendRequest(void *thandle, const SRpcIpSet *pIpSet, SRpcMsg *pMsg);
+void  rpcSendRequest(void *thandle, const SRpcEpSet *pEpSet, SRpcMsg *pMsg);
 void  rpcSendResponse(const SRpcMsg *pMsg);
-void  rpcSendRedirectRsp(void *pConn, const SRpcIpSet *pIpSet); 
+void  rpcSendRedirectRsp(void *pConn, const SRpcEpSet *pEpSet); 
 int   rpcGetConnInfo(void *thandle, SRpcConnInfo *pInfo);
-void  rpcSendRecv(void *shandle, SRpcIpSet *pIpSet, SRpcMsg *pReq, SRpcMsg *pRsp);
+void  rpcSendRecv(void *shandle, SRpcEpSet *pEpSet, SRpcMsg *pReq, SRpcMsg *pRsp);
 int   rpcReportProgress(void *pConn, char *pCont, int contLen);
 void  rpcCancelRequest(void *pContext);
 

src/inc/tsdb.h

@@ -53,14 +53,12 @@ typedef struct {
   int32_t tsdbId;
   int32_t cacheBlockSize;
   int32_t totalBlocks;
-  int32_t maxTables;    // maximum number of tables this repository can have
   int32_t daysPerFile;  // day per file sharding policy
   int32_t keep;         // day of data to keep
   int32_t keep1;
   int32_t keep2;
   int32_t minRowsPerFileBlock;  // minimum rows per file block
   int32_t maxRowsPerFileBlock;  // maximum rows per file block
-  int32_t commitTime;
   int8_t  precision;
   int8_t  compression;
 } STsdbCfg;

src/kit/taosdemo/taosdemo.c

@@ -361,7 +361,7 @@ int main(int argc, char *argv[]) {
   arguments.num_of_DPT = 100000;
   arguments.num_of_RPR = 1000;
   arguments.use_metric = true;
-  arguments.insert_only = true;
+  arguments.insert_only = false;
   // end change
 
   argp_parse(&argp, argc, argv, 0, 0, &arguments);
@@ -954,13 +954,13 @@ void *readMetric(void *sarg) {
 
     for (int i = 1; i <= m; i++) {
       if (i == 1) {
-        sprintf(tempS, "index = %d", i);
+        sprintf(tempS, "areaid = %d", i);
       } else {
-        sprintf(tempS, " or index = %d ", i);
+        sprintf(tempS, " or areaid = %d ", i);
       }
       strcat(condition, tempS);
 
-      sprintf(command, "select %s from m1 where %s", aggreFunc[j], condition);
+      sprintf(command, "select %s from meters where %s", aggreFunc[j], condition);
 
       printf("Where condition: %s\n", condition);
       fprintf(fp, "%s\n", command);

src/kit/taosmigrate/taosmigrate.c

@@ -210,10 +210,10 @@ int32_t main(int32_t argc, char *argv[]) {
   (void)snprintf(mnodeWal, TSDB_FILENAME_LEN*2, "%s/mnode/wal/wal0", arguments.dataDir);
   walModWalFile(mnodeWal);
 
-  // 2. modfiy dnode config: mnodeIpList.json
-  char dnodeIpList[TSDB_FILENAME_LEN*2] = {0};
-  (void)snprintf(dnodeIpList, TSDB_FILENAME_LEN*2, "%s/dnode/mnodeIpList.json", arguments.dataDir);
-  modDnodeIpList(dnodeIpList);
+  // 2. modfiy dnode config: mnodeEpSet.json
+  char dnodeEpSet[TSDB_FILENAME_LEN*2] = {0};
+  (void)snprintf(dnodeEpSet, TSDB_FILENAME_LEN*2, "%s/dnode/mnodeEpSet.json", arguments.dataDir);
+  modDnodeEpSet(dnodeEpSet);
 
   // 3. modify vnode config: config.json
   char vnodeDir[TSDB_FILENAME_LEN*2] = {0};

src/kit/taosmigrate/taosmigrate.h

@@ -71,7 +71,7 @@ int tSystemShell(const char * cmd);
 void taosMvFile(char* destFile, char *srcFile) ;
 void walModWalFile(char* walfile);
 SdnodeIfo* getDnodeInfo(int32_t      dnodeId);
-void modDnodeIpList(char* dnodeIpList);
+void modDnodeEpSet(char* dnodeEpSet);
 void modAllVnode(char *vnodeDir);
 
 #endif

src/kit/taosmigrate/taosmigrateDnodeCfg.c

@@ -23,10 +23,10 @@
 
 static SDMMnodeInfos tsDnodeIpInfos = {0};
 
-static bool dnodeReadMnodeInfos(char* dnodeIpList) {
-  FILE *fp = fopen(dnodeIpList, "r");
+static bool dnodeReadMnodeInfos(char* dnodeEpSet) {
+  FILE *fp = fopen(dnodeEpSet, "r");
   if (!fp) {
-    printf("failed to read mnodeIpList.json, file not exist\n");
+    printf("failed to read mnodeEpSet.json, file not exist\n");
     return false;
   }
 
@@ -37,40 +37,40 @@ static bool dnodeReadMnodeInfos(char* dnodeIpList) {
   if (len <= 0) {
     free(content);
     fclose(fp);
-    printf("failed to read mnodeIpList.json, content is null\n");
+    printf("failed to read mnodeEpSet.json, content is null\n");
     return false;
   }
 
   content[len] = 0;
   cJSON* root = cJSON_Parse(content);
   if (root == NULL) {
-    printf("failed to read mnodeIpList.json, invalid json format\n");
+    printf("failed to read mnodeEpSet.json, invalid json format\n");
     goto PARSE_OVER;
   }
 
   cJSON* inUse = cJSON_GetObjectItem(root, "inUse");
   if (!inUse || inUse->type != cJSON_Number) {
-    printf("failed to read mnodeIpList.json, inUse not found\n");
+    printf("failed to read mnodeEpSet.json, inUse not found\n");
     goto PARSE_OVER;
   }
   tsDnodeIpInfos.inUse = inUse->valueint;
 
   cJSON* nodeNum = cJSON_GetObjectItem(root, "nodeNum");
   if (!nodeNum || nodeNum->type != cJSON_Number) {
-    printf("failed to read mnodeIpList.json, nodeNum not found\n");
+    printf("failed to read mnodeEpSet.json, nodeNum not found\n");
     goto PARSE_OVER;
   }
   tsDnodeIpInfos.nodeNum = nodeNum->valueint;
 
   cJSON* nodeInfos = cJSON_GetObjectItem(root, "nodeInfos");
   if (!nodeInfos || nodeInfos->type != cJSON_Array) {
-    printf("failed to read mnodeIpList.json, nodeInfos not found\n");
+    printf("failed to read mnodeEpSet.json, nodeInfos not found\n");
     goto PARSE_OVER;
   }
 
   int size = cJSON_GetArraySize(nodeInfos);
   if (size != tsDnodeIpInfos.nodeNum) {
-    printf("failed to read mnodeIpList.json, nodeInfos size not matched\n");
+    printf("failed to read mnodeEpSet.json, nodeInfos size not matched\n");
     goto PARSE_OVER;
   }
 
@@ -80,14 +80,14 @@ static bool dnodeReadMnodeInfos(char* dnodeIpList) {
 
     cJSON *nodeId = cJSON_GetObjectItem(nodeInfo, "nodeId");
     if (!nodeId || nodeId->type != cJSON_Number) {
-      printf("failed to read mnodeIpList.json, nodeId not found\n");
+      printf("failed to read mnodeEpSet.json, nodeId not found\n");
       goto PARSE_OVER;
     }
     tsDnodeIpInfos.nodeInfos[i].nodeId = nodeId->valueint;
 
     cJSON *nodeEp = cJSON_GetObjectItem(nodeInfo, "nodeEp");
     if (!nodeEp || nodeEp->type != cJSON_String || nodeEp->valuestring == NULL) {
-      printf("failed to read mnodeIpList.json, nodeName not found\n");
+      printf("failed to read mnodeEpSet.json, nodeName not found\n");
       goto PARSE_OVER;
     }
     strncpy(tsDnodeIpInfos.nodeInfos[i].nodeEp, nodeEp->valuestring, TSDB_EP_LEN);
@@ -102,7 +102,7 @@ static bool dnodeReadMnodeInfos(char* dnodeIpList) {
 
   ret = true;
 
-  //printf("read mnode iplist successed, numOfIps:%d inUse:%d\n", tsDnodeIpInfos.nodeNum, tsDnodeIpInfos.inUse);
+  //printf("read mnode epSet successed, numOfEps:%d inUse:%d\n", tsDnodeIpInfos.nodeNum, tsDnodeIpInfos.inUse);
   //for (int32_t i = 0; i < tsDnodeIpInfos.nodeNum; i++) {
   //  printf("mnode:%d, %s\n", tsDnodeIpInfos.nodeInfos[i].nodeId, tsDnodeIpInfos.nodeInfos[i].nodeEp);
   //}
@@ -115,8 +115,8 @@ PARSE_OVER:
 }
 
 
-static void dnodeSaveMnodeInfos(char* dnodeIpList) {
-  FILE *fp = fopen(dnodeIpList, "w");
+static void dnodeSaveMnodeInfos(char* dnodeEpSet) {
+  FILE *fp = fopen(dnodeEpSet, "w");
   if (!fp) return;
 
   int32_t len = 0;
@@ -143,13 +143,13 @@ static void dnodeSaveMnodeInfos(char* dnodeIpList) {
   fclose(fp);
   free(content);
   
-  printf("mod mnode iplist successed\n");
+  printf("mod mnode epSet successed\n");
 }
 
-void modDnodeIpList(char* dnodeIpList)
+void modDnodeEpSet(char* dnodeEpSet)
 {
-  (void)dnodeReadMnodeInfos(dnodeIpList);
-  dnodeSaveMnodeInfos(dnodeIpList);
+  (void)dnodeReadMnodeInfos(dnodeEpSet);
+  dnodeSaveMnodeInfos(dnodeEpSet);
   return;
 }
 

src/kit/taosmigrate/taosmigrateVnodeCfg.c

@@ -37,14 +37,12 @@ static int32_t saveVnodeCfg(SVnodeObj *pVnode, char* cfgFile)
   len += snprintf(content + len, maxLen - len, "  \"cfgVersion\": %d,\n", pVnode->cfgVersion);
   len += snprintf(content + len, maxLen - len, "  \"cacheBlockSize\": %d,\n", pVnode->tsdbCfg.cacheBlockSize);
   len += snprintf(content + len, maxLen - len, "  \"totalBlocks\": %d,\n", pVnode->tsdbCfg.totalBlocks);
-  len += snprintf(content + len, maxLen - len, "  \"maxTables\": %d,\n", pVnode->tsdbCfg.maxTables);
   len += snprintf(content + len, maxLen - len, "  \"daysPerFile\": %d,\n", pVnode->tsdbCfg.daysPerFile);
   len += snprintf(content + len, maxLen - len, "  \"daysToKeep\": %d,\n", pVnode->tsdbCfg.keep);
   len += snprintf(content + len, maxLen - len, "  \"daysToKeep1\": %d,\n", pVnode->tsdbCfg.keep1);
   len += snprintf(content + len, maxLen - len, "  \"daysToKeep2\": %d,\n", pVnode->tsdbCfg.keep2);
   len += snprintf(content + len, maxLen - len, "  \"minRowsPerFileBlock\": %d,\n", pVnode->tsdbCfg.minRowsPerFileBlock);
   len += snprintf(content + len, maxLen - len, "  \"maxRowsPerFileBlock\": %d,\n", pVnode->tsdbCfg.maxRowsPerFileBlock);
-  len += snprintf(content + len, maxLen - len, "  \"commitTime\": %d,\n", pVnode->tsdbCfg.commitTime);
   len += snprintf(content + len, maxLen - len, "  \"precision\": %d,\n", pVnode->tsdbCfg.precision);
   len += snprintf(content + len, maxLen - len, "  \"compression\": %d,\n", pVnode->tsdbCfg.compression);
   len += snprintf(content + len, maxLen - len, "  \"walLevel\": %d,\n", pVnode->walCfg.walLevel);
@@ -136,12 +134,12 @@ static int32_t readVnodeCfg(SVnodeObj *pVnode, char* cfgFile)
   }
   pVnode->tsdbCfg.totalBlocks = totalBlocks->valueint;
 
-  cJSON *maxTables = cJSON_GetObjectItem(root, "maxTables");
-  if (!maxTables || maxTables->type != cJSON_Number) {
-    printf("vgId:%d, failed to read vnode cfg, maxTables not found\n", pVnode->vgId);
-    goto PARSE_OVER;
-  }
-  pVnode->tsdbCfg.maxTables = maxTables->valueint;
+  // cJSON *maxTables = cJSON_GetObjectItem(root, "maxTables");
+  // if (!maxTables || maxTables->type != cJSON_Number) {
+  //   printf("vgId:%d, failed to read vnode cfg, maxTables not found\n", pVnode->vgId);
+  //   goto PARSE_OVER;
+  // }
+  // pVnode->tsdbCfg.maxTables = maxTables->valueint;
 
   cJSON *daysPerFile = cJSON_GetObjectItem(root, "daysPerFile");
   if (!daysPerFile || daysPerFile->type != cJSON_Number) {
@@ -185,12 +183,12 @@ static int32_t readVnodeCfg(SVnodeObj *pVnode, char* cfgFile)
   }
   pVnode->tsdbCfg.maxRowsPerFileBlock = maxRowsPerFileBlock->valueint;
 
-  cJSON *commitTime = cJSON_GetObjectItem(root, "commitTime");
-  if (!commitTime || commitTime->type != cJSON_Number) {
-    printf("vgId:%d, failed to read vnode cfg, commitTime not found\n", pVnode->vgId);
-    goto PARSE_OVER;
-  }
-  pVnode->tsdbCfg.commitTime = (int8_t)commitTime->valueint;
+  // cJSON *commitTime = cJSON_GetObjectItem(root, "commitTime");
+  // if (!commitTime || commitTime->type != cJSON_Number) {
+  //   printf("vgId:%d, failed to read vnode cfg, commitTime not found\n", pVnode->vgId);
+  //   goto PARSE_OVER;
+  // }
+  // pVnode->tsdbCfg.commitTime = (int8_t)commitTime->valueint;
 
   cJSON *precision = cJSON_GetObjectItem(root, "precision");
   if (!precision || precision->type != cJSON_Number) {

src/mnode/inc/mnodeCluster.h

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef TDENGINE_MNODE_CLUSTER_H
+#define TDENGINE_MNODE_CLUSTER_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct SClusterObj;
+
+int32_t mnodeInitCluster();
+void    mnodeCleanupCluster();
+int32_t mnodeGetClusterId();
+void    mnodeUpdateClusterId();
+void *  mnodeGetCluster(int32_t clusterId);
+void *  mnodeGetNextCluster(void *pIter, struct SClusterObj **pCluster);
+void    mnodeIncClusterRef(struct SClusterObj *pCluster);
+void    mnodeDecClusterRef(struct SClusterObj *pCluster);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

src/mnode/inc/mnodeDef.h

@@ -36,6 +36,14 @@ struct define notes:
 3. The fields behind the updataEnd field can be changed;
 */
 
+typedef struct SClusterObj {
+  int32_t clusterId;
+  int64_t createdTime;
+  int8_t  reserved[36];
+  int8_t  updateEnd[4];
+  int32_t refCount;
+} SClusterObj;
+
 typedef struct SDnodeObj {
   int32_t    dnodeId;
   int32_t    openVnodes;
@@ -50,8 +58,8 @@ typedef struct SDnodeObj {
   int8_t     alternativeRole;  // from dnode status msg, 0-any, 1-mgmt, 2-dnode
   int8_t     status;           // set in balance function
   int8_t     isMgmt;
-  int8_t     reserve1[14];  
-  int8_t     updateEnd[1];
+  int8_t     reserve1[11];  
+  int8_t     updateEnd[4];
   int32_t    refCount;
   uint32_t   moduleStatus;
   uint32_t   lastReboot;       // time stamp for last reboot
@@ -68,8 +76,8 @@ typedef struct SMnodeObj {
   int32_t    mnodeId;
   int8_t     reserved0[4];
   int64_t    createdTime;
-  int8_t     reserved1[7];
-  int8_t     updateEnd[1];
+  int8_t     reserved1[4];
+  int8_t     updateEnd[4];
   int32_t    refCount;
   int8_t     role;
   int8_t     reserved2[3];
@@ -90,8 +98,7 @@ typedef struct SSuperTableObj {
   int32_t    tversion;
   int32_t    numOfColumns;
   int32_t    numOfTags;
-  int8_t     reserved1[3];
-  int8_t     updateEnd[1];
+  int8_t     updateEnd[4];
   int32_t    refCount;
   int32_t    numOfTables;
   SSchema *  schema;
@@ -111,8 +118,7 @@ typedef struct {
   int32_t    sid;
   int32_t    vgId;
   int32_t    sqlLen;
-  int8_t     updateEnd[1];
-  int8_t     reserved1[1]; 
+  int8_t     updateEnd[4];
   int32_t    refCount;
   char*      sql;          //used by normal table
   SSchema*   schema;       //used by normal table
@@ -138,8 +144,8 @@ typedef struct SVgObj {
   int8_t         status;
   int8_t         reserved0[4];
   SVnodeGid      vnodeGid[TSDB_MAX_REPLICA];
-  int8_t         reserved1[7];
-  int8_t         updateEnd[1];
+  int8_t         reserved1[4];
+  int8_t         updateEnd[4];
   int32_t        refCount;
   int32_t        numOfTables;
   int64_t        totalStorage;
@@ -176,8 +182,8 @@ typedef struct SDbObj {
   int32_t cfgVersion;
   SDbCfg  cfg;
   int8_t  status;
-  int8_t  reserved1[14];
-  int8_t  updateEnd[1];
+  int8_t  reserved1[11];
+  int8_t  updateEnd[4];
   int32_t refCount;
   int32_t numOfVgroups;
   int32_t numOfTables;
@@ -196,8 +202,8 @@ typedef struct SUserObj {
   int64_t           createdTime;
   int8_t            superAuth;
   int8_t            writeAuth;
-  int8_t            reserved[13];
-  int8_t            updateEnd[1];
+  int8_t            reserved[10];
+  int8_t            updateEnd[4];
   int32_t           refCount;
   struct SAcctObj * pAcct;
 } SUserObj;
@@ -228,11 +234,11 @@ typedef struct SAcctObj {
   int64_t   createdTime;
   int32_t   acctId;
   int8_t    status;
-  int8_t    reserved0[10];
-  int8_t    updateEnd[1];
-  SAcctInfo acctInfo;
+  int8_t    reserved0[7];
+  int8_t    updateEnd[4];
   int32_t   refCount;
   int8_t    reserved1[4];
+  SAcctInfo acctInfo;
   pthread_mutex_t  mutex;
 } SAcctObj;
 

src/mnode/inc/mnodeMnode.h

@@ -42,12 +42,12 @@ void    mnodeIncMnodeRef(struct SMnodeObj *pMnode);
 void    mnodeDecMnodeRef(struct SMnodeObj *pMnode);
 
 char *  mnodeGetMnodeRoleStr();
-void    mnodeGetMnodeIpSetForPeer(SRpcIpSet *ipSet);
-void    mnodeGetMnodeIpSetForShell(SRpcIpSet *ipSet);
+void    mnodeGetMnodeEpSetForPeer(SRpcEpSet *epSet);
+void    mnodeGetMnodeEpSetForShell(SRpcEpSet *epSet);
 char*   mnodeGetMnodeMasterEp();
 
 void    mnodeGetMnodeInfos(void *mnodes);
-void    mnodeUpdateMnodeIpSet();
+void    mnodeUpdateMnodeEpSet();
 
 #ifdef __cplusplus
 }

src/mnode/inc/mnodeSdb.h

@@ -23,15 +23,16 @@ extern "C" {
 struct SMnodeMsg;
 
 typedef enum {
-  SDB_TABLE_DNODE   = 0,
-  SDB_TABLE_MNODE   = 1,
-  SDB_TABLE_ACCOUNT = 2,
-  SDB_TABLE_USER    = 3,
-  SDB_TABLE_DB      = 4,
-  SDB_TABLE_VGROUP  = 5,
-  SDB_TABLE_STABLE  = 6,
-  SDB_TABLE_CTABLE  = 7,
-  SDB_TABLE_MAX     = 8
+  SDB_TABLE_CLUSTER = 0,
+  SDB_TABLE_DNODE   = 1,
+  SDB_TABLE_MNODE   = 2,
+  SDB_TABLE_ACCOUNT = 3,
+  SDB_TABLE_USER    = 4,
+  SDB_TABLE_DB      = 5,
+  SDB_TABLE_VGROUP  = 6,
+  SDB_TABLE_STABLE  = 7,
+  SDB_TABLE_CTABLE  = 8,
+  SDB_TABLE_MAX     = 9
 } ESdbTable;
 
 typedef enum {

src/mnode/inc/mnodeVgroup.h

@@ -44,12 +44,12 @@ int32_t mnodeGetAvailableVgroup(struct SMnodeMsg *pMsg, SVgObj **pVgroup, int32_
 
 void    mnodeAddTableIntoVgroup(SVgObj *pVgroup, SChildTableObj *pTable);
 void    mnodeRemoveTableFromVgroup(SVgObj *pVgroup, SChildTableObj *pTable);
-void    mnodeSendDropVnodeMsg(int32_t vgId, SRpcIpSet *ipSet, void *ahandle);
+void    mnodeSendDropVnodeMsg(int32_t vgId, SRpcEpSet *epSet, void *ahandle);
 void    mnodeSendCreateVgroupMsg(SVgObj *pVgroup, void *ahandle);
 void    mnodeSendAlterVgroupMsg(SVgObj *pVgroup);
 
-SRpcIpSet mnodeGetIpSetFromVgroup(SVgObj *pVgroup);
-SRpcIpSet mnodeGetIpSetFromIp(char *ep);
+SRpcEpSet mnodeGetEpSetFromVgroup(SVgObj *pVgroup);
+SRpcEpSet mnodeGetEpSetFromIp(char *ep);
 
 #ifdef __cplusplus
 }

src/mnode/src/mnodeAcct.c

@@ -64,7 +64,7 @@ static int32_t mnodeAcctActionUpdate(SSdbOper *pOper) {
   return TSDB_CODE_SUCCESS;
 }
 
-static int32_t mnodeActionActionEncode(SSdbOper *pOper) {
+static int32_t mnodeAcctActionEncode(SSdbOper *pOper) {
   SAcctObj *pAcct = pOper->pObj;
   memcpy(pOper->rowData, pAcct, tsAcctUpdateSize);
   pOper->rowSize = tsAcctUpdateSize;
@@ -109,7 +109,7 @@ int32_t mnodeInitAccts() {
     .insertFp     = mnodeAcctActionInsert,
     .deleteFp     = mnodeAcctActionDelete,
     .updateFp     = mnodeAcctActionUpdate,
-    .encodeFp     = mnodeActionActionEncode,
+    .encodeFp     = mnodeAcctActionEncode,
     .decodeFp     = mnodeAcctActionDecode,
     .destroyFp    = mnodeAcctActionDestroy,
     .restoredFp   = mnodeAcctActionRestored

src/mnode/src/mnodeBalance.c

@@ -28,6 +28,7 @@ void    balanceCleanUp() {}
 void    balanceAsyncNotify() {}
 void    balanceSyncNotify() {}
 void    balanceReset() {}
+int32_t balanceAlterDnode(struct SDnodeObj *pDnode, int32_t vnodeId, int32_t dnodeId) { return TSDB_CODE_SYN_NOT_ENABLED; }
 
 int32_t balanceAllocVnodes(SVgObj *pVgroup) {
   void *     pIter = NULL;

src/mnode/src/mnodeCluster.c

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "os.h"
+#include "taoserror.h"
+#include "ttime.h"
+#include "dnode.h"
+#include "mnodeDef.h"
+#include "mnodeInt.h"
+#include "mnodeCluster.h"
+#include "mnodeSdb.h"
+#include "mnodeShow.h"
+#include "tglobal.h"
+
+static void *  tsClusterSdb = NULL;
+static int32_t tsClusterUpdateSize;
+static int32_t tsClusterId;
+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 int32_t mnodeClusterActionDestroy(SSdbOper *pOper) {
+  tfree(pOper->pObj);
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t mnodeClusterActionInsert(SSdbOper *pOper) {
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t mnodeClusterActionDelete(SSdbOper *pOper) {
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t mnodeClusterActionUpdate(SSdbOper *pOper) {
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t mnodeClusterActionEncode(SSdbOper *pOper) {
+  SClusterObj *pCluster = pOper->pObj;
+  memcpy(pOper->rowData, pCluster, tsClusterUpdateSize);
+  pOper->rowSize = tsClusterUpdateSize;
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t mnodeClusterActionDecode(SSdbOper *pOper) {
+  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;
+  return TSDB_CODE_SUCCESS;
+}
+
+static int32_t mnodeClusterActionRestored() {
+  int32_t numOfRows = sdbGetNumOfRows(tsClusterSdb);
+  if (numOfRows <= 0 && dnodeIsFirstDeploy()) {
+    mInfo("dnode first deploy, create cluster");
+    int32_t code = mnodeCreateCluster();
+    if (code != TSDB_CODE_SUCCESS) {
+      mError("failed to create cluster, reason:%s", tstrerror(code));
+      return code;
+    }
+  }
+
+  mnodeUpdateClusterId();
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t mnodeInitCluster() {
+  SClusterObj tObj;
+  tsClusterUpdateSize = (int8_t *)tObj.updateEnd - (int8_t *)&tObj;
+
+  SSdbTableDesc tableDesc = {
+    .tableId      = SDB_TABLE_CLUSTER,
+    .tableName    = "cluster",
+    .hashSessions = TSDB_DEFAULT_CLUSTER_HASH_SIZE,
+    .maxRowSize   = tsClusterUpdateSize,
+    .refCountPos  = (int8_t *)(&tObj.refCount) - (int8_t *)&tObj,
+    .keyType      = SDB_KEY_INT,
+    .insertFp     = mnodeClusterActionInsert,
+    .deleteFp     = mnodeClusterActionDelete,
+    .updateFp     = mnodeClusterActionUpdate,
+    .encodeFp     = mnodeClusterActionEncode,
+    .decodeFp     = mnodeClusterActionDecode,
+    .destroyFp    = mnodeClusterActionDestroy,
+    .restoredFp   = mnodeClusterActionRestored
+  };
+
+  tsClusterSdb = sdbOpenTable(&tableDesc);
+  if (tsClusterSdb == NULL) {
+    mError("table:%s, failed to create hash", tableDesc.tableName);
+    return -1;
+  }
+
+  mnodeAddShowMetaHandle(TSDB_MGMT_TABLE_CLUSTER, mnodeGetClusterMeta);
+  mnodeAddShowRetrieveHandle(TSDB_MGMT_TABLE_CLUSTER, mnodeRetrieveClusters);
+
+  mDebug("table:%s, hash is created", tableDesc.tableName);
+  return TSDB_CODE_SUCCESS;
+}
+
+void mnodeCleanupCluster() {
+  sdbCloseTable(tsClusterSdb);
+  tsClusterSdb = NULL;
+}
+
+void *mnodeGetCluster(int32_t clusterId) {
+  return sdbGetRow(tsClusterSdb, &clusterId);
+}
+
+void *mnodeGetNextCluster(void *pIter, SClusterObj **pCluster) {
+  return sdbFetchRow(tsClusterSdb, pIter, (void **)pCluster); 
+}
+
+void mnodeIncClusterRef(SClusterObj *pCluster) {
+  sdbIncRef(tsClusterSdb, pCluster);
+}
+
+void mnodeDecClusterRef(SClusterObj *pCluster) {
+  sdbDecRef(tsClusterSdb, pCluster);
+}
+
+static int32_t mnodeCreateCluster() {
+  int32_t numOfClusters = sdbGetNumOfRows(tsClusterSdb);
+  if (numOfClusters != 0) return TSDB_CODE_SUCCESS;
+
+  SClusterObj *pCluster = malloc(sizeof(SClusterObj));
+  memset(pCluster, 0, sizeof(SClusterObj));
+  pCluster->createdTime = taosGetTimestampMs();
+  pCluster->clusterId = labs((pCluster->createdTime >> 32) & (pCluster->createdTime)) | (*(int32_t*)tsFirst);
+  
+  SSdbOper oper = {
+    .type = SDB_OPER_GLOBAL,
+    .table = tsClusterSdb,
+    .pObj = pCluster,
+  };
+
+  return sdbInsertRow(&oper);
+}
+
+int32_t mnodeGetClusterId() {
+  return tsClusterId;
+}
+
+void mnodeUpdateClusterId() {
+  SClusterObj *pCluster = NULL;
+  mnodeGetNextCluster(NULL, &pCluster);
+  if (pCluster != NULL) {
+    tsClusterId = pCluster->clusterId;
+    mnodeDecClusterRef(pCluster);
+    mInfo("cluster id is %d", tsClusterId);
+  } else {
+    //assert(false);
+  }
+}
+
+
+static int32_t mnodeGetClusterMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
+  int32_t cols = 0;
+  SSchema *pSchema = pMeta->schema;
+
+  pShow->bytes[cols] = 4;
+  pSchema[cols].type = TSDB_DATA_TYPE_INT;
+  strcpy(pSchema[cols].name, "clusterId");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htons(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htons(cols);
+  strcpy(pMeta->tableId, "show cluster");
+  pShow->numOfColumns = cols;
+
+  pShow->offset[0] = 0;
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
+
+  pShow->numOfRows = 1;
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+
+  return 0;
+}
+
+static int32_t mnodeRetrieveClusters(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
+  int32_t numOfRows = 0;
+  int32_t cols = 0;
+  char *  pWrite;
+  SClusterObj *pCluster = NULL;
+
+  while (numOfRows < rows) {
+    pShow->pIter = mnodeGetNextCluster(pShow->pIter, &pCluster);
+    if (pCluster == NULL) break;
+    
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int32_t *) pWrite = pCluster->clusterId;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int32_t *) pWrite = pCluster->createdTime;
+    cols++;
+
+    mnodeDecClusterRef(pCluster);
+    numOfRows++;
+  }
+
+  pShow->numOfReads += numOfRows;
+  return numOfRows;
+}

src/mnode/src/mnodeDb.c

@@ -67,8 +67,11 @@ static int32_t mnodeDbActionInsert(SSdbOper *pOper) {
   SAcctObj *pAcct = mnodeGetAcct(pDb->acct);
 
   pthread_mutex_init(&pDb->mutex, NULL);
+  pthread_mutex_lock(&pDb->mutex);
   pDb->vgListSize = VG_LIST_SIZE;
   pDb->vgList = calloc(pDb->vgListSize, sizeof(SVgObj *));
+  pthread_mutex_unlock(&pDb->mutex);
+
   pDb->numOfVgroups = 0;
   pDb->numOfTables = 0;
   pDb->numOfSuperTables = 0;
@@ -395,8 +398,8 @@ static int32_t mnodeCreateDb(SAcctObj *pAcct, SCMCreateDbMsg *pCreate, void *pMs
 
   code = sdbInsertRow(&oper);
   if (code != TSDB_CODE_SUCCESS) {
-    mnodeDestroyDb(pDb);
     mLInfo("db:%s, failed to create, reason:%s", pDb->name, tstrerror(code));
+    mnodeDestroyDb(pDb);
     return code;
   } else {
     return TSDB_CODE_MND_ACTION_IN_PROGRESS;
@@ -605,7 +608,9 @@ static int32_t mnodeGetDbMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn
 
 static char *mnodeGetDbStr(char *src) {
   char *pos = strstr(src, TS_PATH_DELIMITER);
-  return ++pos;
+  if (pos != NULL) ++pos;
+
+  return pos;
 }
 
 static int32_t mnodeRetrieveDbs(SShowObj *pShow, char *data, int32_t rows, void *pConn) {
@@ -622,10 +627,13 @@ static int32_t mnodeRetrieveDbs(SShowObj *pShow, char *data, int32_t rows, void
 
     cols = 0;
 
-    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;  
     char* name = mnodeGetDbStr(pDb->name);
-    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, name, pShow->bytes[cols]);
+    if (name != NULL) {
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, name, pShow->bytes[cols]);
+    } else {
+      STR_TO_VARSTR(pWrite, "NULL");
+    }
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;

src/mnode/src/mnodeDnode.c

@@ -37,6 +37,7 @@
 #include "mnodeVgroup.h"
 #include "mnodeWrite.h"
 #include "mnodePeer.h"
+#include "mnodeCluster.h"
 
 int32_t tsAccessSquence = 0;
 static void   *tsDnodeSdb = NULL;
@@ -277,45 +278,55 @@ static int32_t mnodeProcessCfgDnodeMsg(SMnodeMsg *pMsg) {
   SCMCfgDnodeMsg *pCmCfgDnode = pMsg->rpcMsg.pCont;
   if (pCmCfgDnode->ep[0] == 0) {
     tstrncpy(pCmCfgDnode->ep, tsLocalEp, TSDB_EP_LEN);
-  } 
+  }
 
-  int32_t dnodeId = 0;
-  char* pos = strchr(pCmCfgDnode->ep, ':');
-  if (NULL == pos) {
-    dnodeId = strtol(pCmCfgDnode->ep, NULL, 10);
+  SDnodeObj *pDnode = mnodeGetDnodeByEp(pCmCfgDnode->ep);
+  if (pDnode == NULL) {
+    int32_t dnodeId = strtol(pCmCfgDnode->ep, NULL, 10);
     if (dnodeId <= 0 || dnodeId > 65536) {
-      mError("failed to cfg dnode, invalid dnodeId:%s", pCmCfgDnode->ep);
+      mError("failed to cfg dnode, invalid dnodeEp:%s", pCmCfgDnode->ep);
       return TSDB_CODE_MND_DNODE_NOT_EXIST;
     }
-  }
 
-  SRpcIpSet ipSet = mnodeGetIpSetFromIp(pCmCfgDnode->ep);
-  if (dnodeId != 0) {
-    SDnodeObj *pDnode = mnodeGetDnode(dnodeId);
+    pDnode = mnodeGetDnode(dnodeId);
     if (pDnode == NULL) {
       mError("failed to cfg dnode, invalid dnodeId:%d", dnodeId);
       return TSDB_CODE_MND_DNODE_NOT_EXIST;
     }
-    ipSet = mnodeGetIpSetFromIp(pDnode->dnodeEp);
-    mnodeDecDnodeRef(pDnode);
   }
 
-  SMDCfgDnodeMsg *pMdCfgDnode = rpcMallocCont(sizeof(SMDCfgDnodeMsg));
-  strcpy(pMdCfgDnode->ep, pCmCfgDnode->ep);
-  strcpy(pMdCfgDnode->config, pCmCfgDnode->config);
+  SRpcEpSet epSet = mnodeGetEpSetFromIp(pDnode->dnodeEp);
 
-  SRpcMsg rpcMdCfgDnodeMsg = {
-    .ahandle = 0,
-    .code = 0,
-    .msgType = TSDB_MSG_TYPE_MD_CONFIG_DNODE,
-    .pCont = pMdCfgDnode,
-    .contLen = sizeof(SMDCfgDnodeMsg)
-  };
-
-  mInfo("dnode:%s, is configured by %s", pCmCfgDnode->ep, pMsg->pUser->user);
-  dnodeSendMsgToDnode(&ipSet, &rpcMdCfgDnodeMsg);
+  if (strncasecmp(pCmCfgDnode->config, "balance", 7) == 0) {
+    int32_t vnodeId = 0;
+    int32_t dnodeId = 0;
+    bool parseOk = taosCheckBalanceCfgOptions(pCmCfgDnode->config + 8, &vnodeId, &dnodeId);
+    if (!parseOk) {
+      mnodeDecDnodeRef(pDnode);
+      return TSDB_CODE_MND_INVALID_DNODE_CFG_OPTION;
+    }
 
-  return TSDB_CODE_SUCCESS;
+    int32_t code = balanceAlterDnode(pDnode, vnodeId, dnodeId);
+    mnodeDecDnodeRef(pDnode);
+    return code;
+  } else {
+    SMDCfgDnodeMsg *pMdCfgDnode = rpcMallocCont(sizeof(SMDCfgDnodeMsg));
+    strcpy(pMdCfgDnode->ep, pCmCfgDnode->ep);
+    strcpy(pMdCfgDnode->config, pCmCfgDnode->config);
+
+    SRpcMsg rpcMdCfgDnodeMsg = {
+      .ahandle = 0,
+      .code = 0,
+      .msgType = TSDB_MSG_TYPE_MD_CONFIG_DNODE,
+      .pCont = pMdCfgDnode,
+      .contLen = sizeof(SMDCfgDnodeMsg)
+    };
+
+    mInfo("dnode:%s, is configured by %s", pCmCfgDnode->ep, pMsg->pUser->user);
+    dnodeSendMsgToDnode(&epSet, &rpcMdCfgDnodeMsg);
+    mnodeDecDnodeRef(pDnode);
+    return TSDB_CODE_SUCCESS;
+  }
 }
 
 static void mnodeProcessCfgDnodeMsgRsp(SRpcMsg *rpcMsg) {
@@ -345,6 +356,7 @@ static int32_t mnodeProcessDnodeStatusMsg(SMnodeMsg *pMsg) {
   pStatus->moduleStatus = htonl(pStatus->moduleStatus);
   pStatus->lastReboot   = htonl(pStatus->lastReboot);
   pStatus->numOfCores   = htons(pStatus->numOfCores);
+  pStatus->clusterId    = htonl(pStatus->clusterId);
   
   uint32_t version = htonl(pStatus->version);
   if (version != tsVersion) {
@@ -372,13 +384,19 @@ static int32_t mnodeProcessDnodeStatusMsg(SMnodeMsg *pMsg) {
   pDnode->diskAvailable    = pStatus->diskAvailable;
   pDnode->alternativeRole  = pStatus->alternativeRole;
   pDnode->moduleStatus     = pStatus->moduleStatus;
-  
+
   if (pStatus->dnodeId == 0) {
-    mDebug("dnode:%d %s, first access", pDnode->dnodeId, pDnode->dnodeEp);
+    mDebug("dnode:%d %s, first access, set clusterId %d", pDnode->dnodeId, pDnode->dnodeEp, mnodeGetClusterId());
   } else {
-    mTrace("dnode:%d, status received, access times %d", pDnode->dnodeId, pDnode->lastAccess);
+    if (pStatus->clusterId != mnodeGetClusterId()) {
+      mError("dnode:%d, input clusterId %d not match with exist %d", pDnode->dnodeId, pStatus->clusterId,
+             mnodeGetClusterId());
+      return TSDB_CODE_MND_INVALID_CLUSTER_ID;
+    } else {
+      mTrace("dnode:%d, status received, access times %d", pDnode->dnodeId, pDnode->lastAccess);
+    }
   }
- 
+
   int32_t openVnodes = htons(pStatus->openVnodes);
   int32_t contLen = sizeof(SDMStatusRsp) + openVnodes * sizeof(SDMVgroupAccess);
   SDMStatusRsp *pRsp = rpcMallocCont(contLen);
@@ -390,6 +408,7 @@ static int32_t mnodeProcessDnodeStatusMsg(SMnodeMsg *pMsg) {
   pRsp->dnodeCfg.dnodeId = htonl(pDnode->dnodeId);
   pRsp->dnodeCfg.moduleStatus = htonl((int32_t)pDnode->isMgmt);
   pRsp->dnodeCfg.numOfVnodes = htonl(openVnodes);
+  pRsp->dnodeCfg.clusterId = htonl(mnodeGetClusterId());
   SDMVgroupAccess *pAccess = (SDMVgroupAccess *)((char *)pRsp + sizeof(SDMStatusRsp));
 
   for (int32_t j = 0; j < openVnodes; ++j) {
@@ -399,9 +418,9 @@ static int32_t mnodeProcessDnodeStatusMsg(SMnodeMsg *pMsg) {
 
     SVgObj *pVgroup = mnodeGetVgroup(pVload->vgId);
     if (pVgroup == NULL) {
-      SRpcIpSet ipSet = mnodeGetIpSetFromIp(pDnode->dnodeEp);
+      SRpcEpSet epSet = mnodeGetEpSetFromIp(pDnode->dnodeEp);
       mInfo("dnode:%d, vgId:%d not exist in mnode, drop it", pDnode->dnodeId, pVload->vgId);
-      mnodeSendDropVnodeMsg(pVload->vgId, &ipSet, NULL);
+      mnodeSendDropVnodeMsg(pVload->vgId, &epSet, NULL);
     } else {
       mnodeUpdateVgroupStatus(pVgroup, pDnode, pVload);
       pAccess->vgId = htonl(pVload->vgId);

src/mnode/src/mnodeMain.c

@@ -32,6 +32,7 @@
 #include "mnodeVgroup.h"
 #include "mnodeUser.h"
 #include "mnodeTable.h"
+#include "mnodeCluster.h"
 #include "mnodeShow.h"
 #include "mnodeProfile.h"
 
@@ -46,6 +47,7 @@ static bool tsMgmtIsRunning = false;
 
 static const SMnodeComponent tsMnodeComponents[] = {
   {"profile", mnodeInitProfile, mnodeCleanupProfile},
+  {"cluster", mnodeInitCluster, mnodeCleanupCluster},
   {"accts",   mnodeInitAccts,   mnodeCleanupAccts},
   {"users",   mnodeInitUsers,   mnodeCleanupUsers},
   {"dnodes",  mnodeInitDnodes,  mnodeCleanupDnodes},

src/mnode/src/mnodeMnode.c

@@ -35,8 +35,8 @@
 
 static void *        tsMnodeSdb = NULL;
 static int32_t       tsMnodeUpdateSize = 0;
-static SRpcIpSet     tsMnodeIpSetForShell;
-static SRpcIpSet     tsMnodeIpSetForPeer;
+static SRpcEpSet     tsMnodeEpSetForShell;
+static SRpcEpSet     tsMnodeEpSetForPeer;
 static SDMMnodeInfos tsMnodeInfos;
 static int32_t mnodeGetMnodeMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn);
 static int32_t mnodeRetrieveMnodes(SShowObj *pShow, char *data, int32_t rows, void *pConn);
@@ -123,7 +123,7 @@ static int32_t mnodeMnodeActionRestored() {
     sdbFreeIter(pIter);
   }
 
-  mnodeUpdateMnodeIpSet();
+  mnodeUpdateMnodeEpSet();
 
   return TSDB_CODE_SUCCESS;
 }
@@ -204,13 +204,13 @@ char *mnodeGetMnodeRoleStr(int32_t role) {
   }
 }
 
-void mnodeUpdateMnodeIpSet() {
-  mInfo("update mnodes ipset, numOfIps:%d ", mnodeGetMnodesNum());
+void mnodeUpdateMnodeEpSet() {
+  mInfo("update mnodes epSet, numOfEps:%d ", mnodeGetMnodesNum());
 
   mnodeMnodeWrLock();
 
-  memset(&tsMnodeIpSetForShell, 0, sizeof(SRpcIpSet));
-  memset(&tsMnodeIpSetForPeer, 0, sizeof(SRpcIpSet));
+  memset(&tsMnodeEpSetForShell, 0, sizeof(SRpcEpSet));
+  memset(&tsMnodeEpSetForPeer, 0, sizeof(SRpcEpSet));
   memset(&tsMnodeInfos, 0, sizeof(SDMMnodeInfos));
 
   int32_t index = 0;
@@ -222,20 +222,20 @@ void mnodeUpdateMnodeIpSet() {
 
     SDnodeObj *pDnode = mnodeGetDnode(pMnode->mnodeId);
     if (pDnode != NULL) {
-      strcpy(tsMnodeIpSetForShell.fqdn[index], pDnode->dnodeFqdn);
-      tsMnodeIpSetForShell.port[index] = htons(pDnode->dnodePort);
-      mDebug("mnode:%d, for shell fqdn:%s %d", pDnode->dnodeId, tsMnodeIpSetForShell.fqdn[index], htons(tsMnodeIpSetForShell.port[index]));      
+      strcpy(tsMnodeEpSetForShell.fqdn[index], pDnode->dnodeFqdn);
+      tsMnodeEpSetForShell.port[index] = htons(pDnode->dnodePort);
+      mDebug("mnode:%d, for shell fqdn:%s %d", pDnode->dnodeId, tsMnodeEpSetForShell.fqdn[index], htons(tsMnodeEpSetForShell.port[index]));      
 
-      strcpy(tsMnodeIpSetForPeer.fqdn[index], pDnode->dnodeFqdn);
-      tsMnodeIpSetForPeer.port[index] = htons(pDnode->dnodePort + TSDB_PORT_DNODEDNODE);
-      mDebug("mnode:%d, for peer fqdn:%s %d", pDnode->dnodeId, tsMnodeIpSetForPeer.fqdn[index], htons(tsMnodeIpSetForPeer.port[index]));
+      strcpy(tsMnodeEpSetForPeer.fqdn[index], pDnode->dnodeFqdn);
+      tsMnodeEpSetForPeer.port[index] = htons(pDnode->dnodePort + TSDB_PORT_DNODEDNODE);
+      mDebug("mnode:%d, for peer fqdn:%s %d", pDnode->dnodeId, tsMnodeEpSetForPeer.fqdn[index], htons(tsMnodeEpSetForPeer.port[index]));
 
       tsMnodeInfos.nodeInfos[index].nodeId = htonl(pMnode->mnodeId);
       strcpy(tsMnodeInfos.nodeInfos[index].nodeEp, pDnode->dnodeEp);
 
       if (pMnode->role == TAOS_SYNC_ROLE_MASTER) {
-        tsMnodeIpSetForShell.inUse = index;
-        tsMnodeIpSetForPeer.inUse = index;
+        tsMnodeEpSetForShell.inUse = index;
+        tsMnodeEpSetForPeer.inUse = index;
         tsMnodeInfos.inUse = index;
       }
 
@@ -248,23 +248,23 @@ void mnodeUpdateMnodeIpSet() {
   }
 
   tsMnodeInfos.nodeNum = index;
-  tsMnodeIpSetForShell.numOfIps = index;
-  tsMnodeIpSetForPeer.numOfIps = index;
+  tsMnodeEpSetForShell.numOfEps = index;
+  tsMnodeEpSetForPeer.numOfEps = index;
 
   sdbFreeIter(pIter);
 
   mnodeMnodeUnLock();
 }
 
-void mnodeGetMnodeIpSetForPeer(SRpcIpSet *ipSet) {
+void mnodeGetMnodeEpSetForPeer(SRpcEpSet *epSet) {
   mnodeMnodeRdLock();
-  *ipSet = tsMnodeIpSetForPeer;
+  *epSet = tsMnodeEpSetForPeer;
   mnodeMnodeUnLock();
 }
 
-void mnodeGetMnodeIpSetForShell(SRpcIpSet *ipSet) {
+void mnodeGetMnodeEpSetForShell(SRpcEpSet *epSet) {
   mnodeMnodeRdLock();
-  *ipSet = tsMnodeIpSetForShell;
+  *epSet = tsMnodeEpSetForShell;
   mnodeMnodeUnLock();
 }
 
@@ -295,7 +295,7 @@ int32_t mnodeAddMnode(int32_t dnodeId) {
     code = TSDB_CODE_MND_SDB_ERROR;
   }
 
-  mnodeUpdateMnodeIpSet();
+  mnodeUpdateMnodeEpSet();
 
   return code;
 }
@@ -308,7 +308,7 @@ void mnodeDropMnodeLocal(int32_t dnodeId) {
     mnodeDecMnodeRef(pMnode);
   }
 
-  mnodeUpdateMnodeIpSet();
+  mnodeUpdateMnodeEpSet();
 }
 
 int32_t mnodeDropMnode(int32_t dnodeId) {
@@ -330,7 +330,7 @@ int32_t mnodeDropMnode(int32_t dnodeId) {
 
   sdbDecRef(tsMnodeSdb, pMnode);
 
-  mnodeUpdateMnodeIpSet();
+  mnodeUpdateMnodeEpSet();
 
   return code;
 }

src/mnode/src/mnodePeer.c

@@ -53,14 +53,14 @@ int32_t mnodeProcessPeerReq(SMnodeMsg *pMsg) {
 
   if (!sdbIsMaster()) {
     SMnodeRsp *rpcRsp = &pMsg->rpcRsp;
-    SRpcIpSet *ipSet = rpcMallocCont(sizeof(SRpcIpSet));
-    mnodeGetMnodeIpSetForPeer(ipSet);
-    rpcRsp->rsp = ipSet;
-    rpcRsp->len = sizeof(SRpcIpSet);
+    SRpcEpSet *epSet = rpcMallocCont(sizeof(SRpcEpSet));
+    mnodeGetMnodeEpSetForPeer(epSet);
+    rpcRsp->rsp = epSet;
+    rpcRsp->len = sizeof(SRpcEpSet);
 
-    mDebug("%p, msg:%s in mpeer queue, will be redireced inUse:%d", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], ipSet->inUse);
-    for (int32_t i = 0; i < ipSet->numOfIps; ++i) {
-      mDebug("mnode index:%d ip:%s:%d", i, ipSet->fqdn[i], htons(ipSet->port[i]));
+    mDebug("%p, msg:%s in mpeer queue, will be redireced inUse:%d", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], epSet->inUse);
+    for (int32_t i = 0; i < epSet->numOfEps; ++i) {
+      mDebug("mnode index:%d ep:%s:%d", i, epSet->fqdn[i], htons(epSet->port[i]));
     }
 
     return TSDB_CODE_RPC_REDIRECT;

src/mnode/src/mnodeRead.c

@@ -49,14 +49,14 @@ int32_t mnodeProcessRead(SMnodeMsg *pMsg) {
 
   if (!sdbIsMaster()) {
     SMnodeRsp *rpcRsp = &pMsg->rpcRsp;
-    SRpcIpSet *ipSet = rpcMallocCont(sizeof(SRpcIpSet));
-    mnodeGetMnodeIpSetForShell(ipSet);
-    rpcRsp->rsp = ipSet;
-    rpcRsp->len = sizeof(SRpcIpSet);
+    SRpcEpSet *epSet = rpcMallocCont(sizeof(SRpcEpSet));
+    mnodeGetMnodeEpSetForShell(epSet);
+    rpcRsp->rsp = epSet;
+    rpcRsp->len = sizeof(SRpcEpSet);
 
-    mDebug("%p, msg:%s in mread queue, will be redireced, inUse:%d", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], ipSet->inUse);
-    for (int32_t i = 0; i < ipSet->numOfIps; ++i) {
-      mDebug("mnode index:%d ip:%s:%d", i, ipSet->fqdn[i], htons(ipSet->port[i]));
+    mDebug("%p, msg:%s in mread queue, will be redireced, inUse:%d", pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType], epSet->inUse);
+    for (int32_t i = 0; i < epSet->numOfEps; ++i) {
+      mDebug("mnode index:%d ep:%s:%d", i, epSet->fqdn[i], htons(epSet->port[i]));
     }
 
     return TSDB_CODE_RPC_REDIRECT;

src/mnode/src/mnodeSdb.c

@@ -29,6 +29,7 @@
 #include "mnodeInt.h"
 #include "mnodeMnode.h"
 #include "mnodeDnode.h"
+#include "mnodeCluster.h"
 #include "mnodeSdb.h"
 
 #define SDB_TABLE_LEN 12
@@ -77,11 +78,6 @@ typedef struct {
   pthread_mutex_t mutex;
 } SSdbObject;
 
-typedef struct {
-  int32_t rowSize;
-  void *  row;
-} SSdbRow;
-
 typedef struct {
   pthread_t thread;
   int32_t   workerId;
@@ -219,7 +215,8 @@ void sdbUpdateMnodeRoles() {
     }
   }
 
-  mnodeUpdateMnodeIpSet();
+  mnodeUpdateClusterId();
+  mnodeUpdateMnodeEpSet();
 }
 
 static uint32_t sdbGetFileInfo(void *ahandle, char *name, uint32_t *index, uint32_t eindex, int32_t *size, uint64_t *fversion) {
@@ -411,7 +408,7 @@ void sdbDecRef(void *handle, void *pObj) {
   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), *pRefCount);
 
-  int8_t *updateEnd = pObj + pTable->refCountPos - 1;
+  int32_t *updateEnd = pObj + pTable->refCountPos - 4;
   if (refCount <= 0 && *updateEnd) {
     sdbTrace("table:%s, record:%p:%s:%d is destroyed", pTable->tableName, pObj, sdbGetKeyStrFromObj(pTable, pObj), *pRefCount);
     SSdbOper oper = {.pObj = pObj};
@@ -419,32 +416,28 @@ void sdbDecRef(void *handle, void *pObj) {
   }
 }
 
-static SSdbRow *sdbGetRowMeta(SSdbTable *pTable, void *key) {
+static void *sdbGetRowMeta(SSdbTable *pTable, void *key) {
   if (pTable == NULL) return NULL;
 
   int32_t keySize = sizeof(int32_t);
   if (pTable->keyType == SDB_KEY_STRING || pTable->keyType == SDB_KEY_VAR_STRING) {
     keySize = strlen((char *)key);
   }
-  
-  return taosHashGet(pTable->iHandle, key, keySize);
+
+  void **ppRow = (void **)taosHashGet(pTable->iHandle, key, keySize);
+  if (ppRow == NULL) return NULL;
+  return *ppRow;
 }
 
-static SSdbRow *sdbGetRowMetaFromObj(SSdbTable *pTable, void *key) {
+static void *sdbGetRowMetaFromObj(SSdbTable *pTable, void *key) {
   return sdbGetRowMeta(pTable, sdbGetObjKey(pTable, key));
 }
 
 void *sdbGetRow(void *handle, void *key) {
-  SSdbTable *pTable = (SSdbTable *)handle;
-  int32_t keySize = sizeof(int32_t);
-  if (pTable->keyType == SDB_KEY_STRING || pTable->keyType == SDB_KEY_VAR_STRING) {
-    keySize = strlen((char *)key);
-  }
-  
-  SSdbRow *pMeta = taosHashGet(pTable->iHandle, key, keySize);
-  if (pMeta) {
-    sdbIncRef(pTable, pMeta->row);
-    return pMeta->row;
+  void *pRow = sdbGetRowMeta(handle, key);
+  if (pRow) {
+    sdbIncRef(handle, pRow);
+    return pRow;
   } else {
     return NULL;
   }
@@ -455,10 +448,6 @@ static void *sdbGetRowFromObj(SSdbTable *pTable, void *key) {
 }
 
 static int32_t sdbInsertHash(SSdbTable *pTable, SSdbOper *pOper) {
-  SSdbRow rowMeta;
-  rowMeta.rowSize = pOper->rowSize;
-  rowMeta.row = pOper->pObj;
-
   void *  key = sdbGetObjKey(pTable, pOper->pObj);
   int32_t keySize = sizeof(int32_t);
 
@@ -466,7 +455,7 @@ static int32_t sdbInsertHash(SSdbTable *pTable, SSdbOper *pOper) {
     keySize = strlen((char *)key);
   }
 
-  taosHashPut(pTable->iHandle, key, keySize, &rowMeta, sizeof(SSdbRow));
+  taosHashPut(pTable->iHandle, key, keySize, &pOper->pObj, sizeof(int64_t));
 
   sdbIncRef(pTable, pOper->pObj);
   atomic_add_fetch_32(&pTable->numOfRows, 1);
@@ -485,6 +474,14 @@ static int32_t sdbInsertHash(SSdbTable *pTable, SSdbOper *pOper) {
 }
 
 static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbOper *pOper) {
+  int32_t *updateEnd = pOper->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));
+    return TSDB_CODE_MND_SDB_OBJ_NOT_THERE;
+  }
+
   (*pTable->deleteFp)(pOper);
   
   void *  key = sdbGetObjKey(pTable, pOper->pObj);
@@ -499,8 +496,6 @@ static int32_t sdbDeleteHash(SSdbTable *pTable, SSdbOper *pOper) {
   sdbDebug("table:%s, delete record:%s from hash, numOfRows:%" PRId64 ", msg:%p", pTable->tableName,
            sdbGetKeyStrFromObj(pTable, pOper->pObj), pTable->numOfRows, pOper->pMsg);
 
-  int8_t *updateEnd = pOper->pObj + pTable->refCountPos - 1;
-  *updateEnd = 1;
   sdbDecRef(pTable, pOper->pObj);
 
   return TSDB_CODE_SUCCESS;
@@ -586,17 +581,17 @@ static int sdbWrite(void *param, void *data, int type) {
     code = (*pTable->decodeFp)(&oper);
     return sdbInsertHash(pTable, &oper);
   } else if (action == SDB_ACTION_DELETE) {
-    SSdbRow *rowMeta = sdbGetRowMeta(pTable, pHead->cont);
-    if (rowMeta == NULL || rowMeta->row == NULL) {
+    void *pRow = sdbGetRowMeta(pTable, pHead->cont);
+    if (pRow == NULL) {
       sdbError("table:%s, failed to get object:%s from wal while dispose delete action", pTable->tableName,
                pHead->cont);
       return TSDB_CODE_SUCCESS;
     }
-    SSdbOper oper = {.table = pTable, .pObj = rowMeta->row};
+    SSdbOper oper = {.table = pTable, .pObj = pRow};
     return sdbDeleteHash(pTable, &oper);
   } else if (action == SDB_ACTION_UPDATE) {
-    SSdbRow *rowMeta = sdbGetRowMeta(pTable, pHead->cont);
-    if (rowMeta == NULL || rowMeta->row == NULL) {
+    void *pRow = sdbGetRowMeta(pTable, pHead->cont);
+    if (pRow == NULL) {
       sdbError("table:%s, failed to get object:%s from wal while dispose update action", pTable->tableName,
                pHead->cont);
       return TSDB_CODE_SUCCESS;
@@ -667,26 +662,21 @@ bool sdbCheckRowDeleted(void *pTableInput, void *pRow) {
   SSdbTable *pTable = pTableInput;
   if (pTable == NULL) return false;
 
-  int8_t *updateEnd = pRow + pTable->refCountPos - 1;
-  return (*updateEnd == 1);
+  int32_t *updateEnd = pRow + pTable->refCountPos - 4;
+  return atomic_val_compare_exchange_32(updateEnd, 1, 1) == 1;
+  // return (*updateEnd == 1);
 }
 
 int32_t sdbDeleteRow(SSdbOper *pOper) {
   SSdbTable *pTable = (SSdbTable *)pOper->table;
   if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE;
 
-  SSdbRow *pMeta = sdbGetRowMetaFromObj(pTable, pOper->pObj);
-  if (pMeta == NULL) {
+  void *pRow = sdbGetRowMetaFromObj(pTable, pOper->pObj);
+  if (pRow == NULL) {
     sdbDebug("table:%s, record is not there, delete failed", pTable->tableName);
     return TSDB_CODE_MND_SDB_OBJ_NOT_THERE;
   }
 
-  void *pMetaRow = pMeta->row;
-  if (pMetaRow == NULL) {
-    sdbError("table:%s, record meta is null", pTable->tableName);
-    return TSDB_CODE_MND_SDB_INVAID_META_ROW;
-  }
-
   sdbIncRef(pTable, pOper->pObj);
 
   int32_t code = sdbDeleteHash(pTable, pOper);
@@ -728,18 +718,12 @@ int32_t sdbUpdateRow(SSdbOper *pOper) {
   SSdbTable *pTable = (SSdbTable *)pOper->table;
   if (pTable == NULL) return TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE;
 
-  SSdbRow *pMeta = sdbGetRowMetaFromObj(pTable, pOper->pObj);
-  if (pMeta == NULL) {
+  void *pRow = sdbGetRowMetaFromObj(pTable, pOper->pObj);
+  if (pRow == NULL) {
     sdbDebug("table:%s, record is not there, update failed", pTable->tableName);
     return TSDB_CODE_MND_SDB_OBJ_NOT_THERE;
   }
 
-  void *pMetaRow = pMeta->row;
-  if (pMetaRow == NULL) {
-    sdbError("table:%s, record meta is null", pTable->tableName);
-    return TSDB_CODE_MND_SDB_INVAID_META_ROW;
-  }
-
   int32_t code = sdbUpdateHash(pTable, pOper);
   if (code != TSDB_CODE_SUCCESS) {
     sdbError("table:%s, failed to update hash", pTable->tableName);
@@ -789,14 +773,14 @@ void *sdbFetchRow(void *handle, void *pNode, void **ppRow) {
     return NULL;
   }
 
-  SSdbRow *pMeta = taosHashIterGet(pIter);
-  if (pMeta == NULL) {
+  void **ppMetaRow = taosHashIterGet(pIter);
+  if (ppMetaRow == NULL) {
     taosHashDestroyIter(pIter);
     return NULL;
   }
 
-  *ppRow = pMeta->row;
-  sdbIncRef(handle, pMeta->row);
+  *ppRow = *ppMetaRow;
+  sdbIncRef(handle, *ppMetaRow);
 
   return pIter;
 }
@@ -846,11 +830,11 @@ void sdbCloseTable(void *handle) {
 
   SHashMutableIterator *pIter = taosHashCreateIter(pTable->iHandle);
   while (taosHashIterNext(pIter)) {
-    SSdbRow *pMeta = taosHashIterGet(pIter);
-    if (pMeta == NULL) continue;
+    void **ppRow = taosHashIterGet(pIter);
+    if (ppRow == NULL) continue;
 
     SSdbOper oper = {
-      .pObj = pMeta->row,
+      .pObj = *ppRow,
       .table = pTable,
     };
 

src/mnode/src/mnodeShow.c

@@ -103,6 +103,8 @@ static char *mnodeGetShowType(int32_t showType) {
     case TSDB_MGMT_TABLE_SCORES:  return "show scores";
     case TSDB_MGMT_TABLE_GRANTS:  return "show grants";
     case TSDB_MGMT_TABLE_VNODES:  return "show vnodes";
+    case TSDB_MGMT_TABLE_CLUSTER: return "show clusters";
+    case TSDB_MGMT_TABLE_STREAMTABLES : return "show streamtables";
     default:                      return "undefined";
   }
 }
@@ -236,7 +238,7 @@ static int32_t mnodeProcessHeartBeatMsg(SMnodeMsg *pMsg) {
   }
 
   SCMHeartBeatMsg *pHBMsg = pMsg->rpcMsg.pCont;
-  SRpcConnInfo connInfo;
+  SRpcConnInfo connInfo = {0};
   rpcGetConnInfo(pMsg->rpcMsg.handle, &connInfo);
     
   int32_t connId = htonl(pHBMsg->connId);
@@ -270,7 +272,7 @@ static int32_t mnodeProcessHeartBeatMsg(SMnodeMsg *pMsg) {
 
   pHBRsp->onlineDnodes = htonl(mnodeGetOnlinDnodesNum());
   pHBRsp->totalDnodes = htonl(mnodeGetDnodesNum());
-  mnodeGetMnodeIpSetForShell(&pHBRsp->ipList);
+  mnodeGetMnodeEpSetForShell(&pHBRsp->epSet);
 
   pMsg->rpcRsp.rsp = pHBRsp;
   pMsg->rpcRsp.len = sizeof(SCMHeartBeatRsp);
@@ -284,7 +286,7 @@ static int32_t mnodeProcessConnectMsg(SMnodeMsg *pMsg) {
   SCMConnectRsp *pConnectRsp = NULL;
   int32_t code = TSDB_CODE_SUCCESS;
 
-  SRpcConnInfo connInfo;
+  SRpcConnInfo connInfo = {0};
   if (rpcGetConnInfo(pMsg->rpcMsg.handle, &connInfo) != 0) {
     mError("thandle:%p is already released while process connect msg", pMsg->rpcMsg.handle);
     code = TSDB_CODE_MND_INVALID_CONNECTION;
@@ -335,7 +337,7 @@ static int32_t mnodeProcessConnectMsg(SMnodeMsg *pMsg) {
   pConnectRsp->writeAuth = pUser->writeAuth;
   pConnectRsp->superAuth = pUser->superAuth;
   
-  mnodeGetMnodeIpSetForShell(&pConnectRsp->ipList);
+  mnodeGetMnodeEpSetForShell(&pConnectRsp->epSet);
 
 connect_over:
   if (code != TSDB_CODE_SUCCESS) {

src/mnode/src/mnodeTable.c

@@ -72,7 +72,7 @@ static void    mnodeProcessCreateChildTableRsp(SRpcMsg *rpcMsg);
 static int32_t mnodeProcessDropTableMsg(SMnodeMsg *mnodeMsg);
 static int32_t mnodeProcessDropSuperTableMsg(SMnodeMsg *pMsg);
 static void    mnodeProcessDropSuperTableRsp(SRpcMsg *rpcMsg);
-static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn);
+static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg);
 static void    mnodeProcessDropChildTableRsp(SRpcMsg *rpcMsg);
 
 static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *mnodeMsg);
@@ -107,42 +107,41 @@ static int32_t mnodeChildTableActionInsert(SSdbOper *pOper) {
   SVgObj *pVgroup = mnodeGetVgroup(pTable->vgId);
   if (pVgroup == NULL) {
     mError("ctable:%s, not in vgId:%d", pTable->info.tableId, pTable->vgId);
-    return TSDB_CODE_MND_VGROUP_NOT_EXIST;
   }
-  mnodeDecVgroupRef(pVgroup);
 
-  SDbObj *pDb = mnodeGetDb(pVgroup->dbName);
-  if (pDb == NULL) {
-    mError("ctable:%s, vgId:%d not in db:%s", pTable->info.tableId, pVgroup->vgId, pVgroup->dbName);
-    return TSDB_CODE_MND_INVALID_DB;
-  }
-  
-  if (pDb->status != TSDB_DB_STATUS_READY) {
-    mError("db:%s, status:%d, in dropping", pDb->name, pDb->status);
-    return TSDB_CODE_MND_DB_IN_DROPPING;
+  SDbObj *pDb = NULL;
+  if (pVgroup != NULL) {
+    pDb = mnodeGetDb(pVgroup->dbName);
+    if (pDb == NULL) {
+      mError("ctable:%s, vgId:%d not in db:%s", pTable->info.tableId, pVgroup->vgId, pVgroup->dbName);
+    }
   }
-  mnodeDecDbRef(pDb);
 
-  SAcctObj *pAcct = mnodeGetAcct(pDb->acct);
-  if (pAcct == NULL) {
-    mError("ctable:%s, acct:%s not exists", pTable->info.tableId, pDb->acct);
-    return TSDB_CODE_MND_INVALID_ACCT;
+  SAcctObj *pAcct = NULL;
+  if (pDb != NULL) {
+    pAcct = mnodeGetAcct(pDb->acct);
+    if (pAcct == NULL) {
+      mError("ctable:%s, acct:%s not exists", pTable->info.tableId, pDb->acct);
+    }
   }
-  mnodeDecAcctRef(pAcct);
 
   if (pTable->info.type == TSDB_CHILD_TABLE) {
     // add ref
     pTable->superTable = mnodeGetSuperTableByUid(pTable->suid);
     mnodeAddTableIntoStable(pTable->superTable, pTable);
     grantAdd(TSDB_GRANT_TIMESERIES, pTable->superTable->numOfColumns - 1);
-    pAcct->acctInfo.numOfTimeSeries += (pTable->superTable->numOfColumns - 1);
+    if (pAcct) pAcct->acctInfo.numOfTimeSeries += (pTable->superTable->numOfColumns - 1);
   } else {
     grantAdd(TSDB_GRANT_TIMESERIES, pTable->numOfColumns - 1);
-    pAcct->acctInfo.numOfTimeSeries += (pTable->numOfColumns - 1);
+    if (pAcct) pAcct->acctInfo.numOfTimeSeries += (pTable->numOfColumns - 1);
   }
 
-  mnodeAddTableIntoDb(pDb);
-  mnodeAddTableIntoVgroup(pVgroup, pTable);
+  if (pDb) mnodeAddTableIntoDb(pDb);
+  if (pVgroup) mnodeAddTableIntoVgroup(pVgroup, pTable);
+
+  mnodeDecVgroupRef(pVgroup);
+  mnodeDecDbRef(pDb);
+  mnodeDecAcctRef(pAcct);
 
   return TSDB_CODE_SUCCESS;
 }
@@ -760,7 +759,7 @@ static int32_t mnodeProcessDropTableMsg(SMnodeMsg *pMsg) {
     SChildTableObj *pCTable = (SChildTableObj *)pMsg->pTable;
     mInfo("app:%p:%p, table:%s, start to drop ctable, vgId:%d sid:%d uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg,
           pDrop->tableId, pCTable->vgId, pCTable->sid, pCTable->uid);
-    return mnodeProcessDropChildTableMsg(pMsg, true);
+    return mnodeProcessDropChildTableMsg(pMsg);
   }
 }
 
@@ -883,7 +882,7 @@ static int32_t mnodeProcessCreateSuperTableMsg(SMnodeMsg *pMsg) {
 static int32_t mnodeDropSuperTableCb(SMnodeMsg *pMsg, int32_t code) {
   SSuperTableObj *pTable = (SSuperTableObj *)pMsg->pTable;
   if (code != TSDB_CODE_SUCCESS) {
-    mError("app:%p:%p, table:%s, failed to drop, sdb error", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId);
+    mError("app:%p:%p, stable:%s, failed to drop, sdb error", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId);
   } else {
     mLInfo("app:%p:%p, stable:%s, is dropped from sdb", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId);
   }
@@ -910,9 +909,9 @@ static int32_t mnodeProcessDropSuperTableMsg(SMnodeMsg *pMsg) {
 
       mInfo("app:%p:%p, stable:%s, send drop stable msg to vgId:%d", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId,
              pVgroup->vgId);
-      SRpcIpSet ipSet = mnodeGetIpSetFromVgroup(pVgroup);
+      SRpcEpSet epSet = mnodeGetEpSetFromVgroup(pVgroup);
       SRpcMsg   rpcMsg = {.pCont = pDrop, .contLen = sizeof(SMDDropSTableMsg), .msgType = TSDB_MSG_TYPE_MD_DROP_STABLE};
-      dnodeSendMsgToDnode(&ipSet, &rpcMsg);
+      dnodeSendMsgToDnode(&epSet, &rpcMsg);
       mnodeDecVgroupRef(pVgroup);
     }
     taosHashDestroyIter(pIter);
@@ -1224,6 +1223,55 @@ static int32_t mnodeDropSuperTableColumn(SMnodeMsg *pMsg, char *colName) {
   return code;
 }
 
+static int32_t mnodeChangeSuperTableColumnCb(SMnodeMsg *pMsg, int32_t code) {
+  SSuperTableObj *pStable = (SSuperTableObj *)pMsg->pTable;
+  mLInfo("app:%p:%p, stable %s, change column result:%s", pMsg->rpcMsg.ahandle, pMsg, pStable->info.tableId,
+         tstrerror(code));
+  return code;
+}
+
+static int32_t mnodeChangeSuperTableColumn(SMnodeMsg *pMsg, char *oldName, char *newName) {
+  SSuperTableObj *pStable = (SSuperTableObj *)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,
+           pStable->info.tableId, oldName, newName);
+    return TSDB_CODE_MND_FIELD_NOT_EXIST;
+  }
+
+  // int32_t  rowSize = 0;
+  uint32_t len = strlen(newName);
+  if (len >= TSDB_COL_NAME_LEN) {
+    return TSDB_CODE_MND_COL_NAME_TOO_LONG;
+  }
+
+  if (mnodeFindSuperTableColumnIndex(pStable, newName) >= 0) {
+    return TSDB_CODE_MND_FIELD_ALREAY_EXIST;
+  }
+  
+  // update
+  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,
+         oldName, newName);
+
+  SSdbOper oper = {
+    .type = SDB_OPER_GLOBAL,
+    .table = tsSuperTableSdb,
+    .pObj = pStable,
+    .pMsg = pMsg,
+    .cb = mnodeChangeSuperTableColumnCb
+  };
+
+  int32_t code = sdbUpdateRow(&oper);
+  if (code == TSDB_CODE_SUCCESS) {
+    code = TSDB_CODE_MND_ACTION_IN_PROGRESS;
+  }
+
+  return code;
+}
+
 // show super tables
 static int32_t mnodeGetShowSuperTableMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *pConn) {
   SDbObj *pDb = mnodeGetDb(pShow->db);
@@ -1406,6 +1454,9 @@ static int32_t mnodeSetSchemaFromSuperTable(SSchema *pSchema, SSuperTableObj *pT
 static int32_t mnodeGetSuperTableMeta(SMnodeMsg *pMsg) {
   SSuperTableObj *pTable = (SSuperTableObj *)pMsg->pTable;
   STableMetaMsg *pMeta   = rpcMallocCont(sizeof(STableMetaMsg) + sizeof(SSchema) * (TSDB_MAX_TAGS + TSDB_MAX_COLUMNS + 16));
+  if (pMeta == NULL) {
+    return TSDB_CODE_MND_OUT_OF_MEMORY;
+  }
   pMeta->uid          = htobe64(pTable->uid);
   pMeta->sversion     = htons(pTable->sversion);
   pMeta->tversion     = htons(pTable->tversion);
@@ -1484,10 +1535,10 @@ static int32_t mnodeProcessSuperTableVgroupMsg(SMnodeMsg *pMsg) {
           SDnodeObj *pDnode = pVgroup->vnodeGid[vn].pDnode;
           if (pDnode == NULL) break;
 
-          tstrncpy(pVgroupInfo->vgroups[vgSize].ipAddr[vn].fqdn, pDnode->dnodeFqdn, TSDB_FQDN_LEN);
-          pVgroupInfo->vgroups[vgSize].ipAddr[vn].port = htons(pDnode->dnodePort);
+          tstrncpy(pVgroupInfo->vgroups[vgSize].epAddr[vn].fqdn, pDnode->dnodeFqdn, TSDB_FQDN_LEN);
+          pVgroupInfo->vgroups[vgSize].epAddr[vn].port = htons(pDnode->dnodePort);
 
-          pVgroupInfo->vgroups[vgSize].numOfIps++;
+          pVgroupInfo->vgroups[vgSize].numOfEps++;
         }
 
         vgSize++;
@@ -1615,7 +1666,7 @@ static int32_t mnodeDoCreateChildTableCb(SMnodeMsg *pMsg, int32_t code) {
     return terrno;
   }
 
-  SRpcIpSet ipSet = mnodeGetIpSetFromVgroup(pMsg->pVgroup);
+  SRpcEpSet epSet = mnodeGetEpSetFromVgroup(pMsg->pVgroup);
   SRpcMsg rpcMsg = {
       .ahandle = pMsg,
       .pCont   = pMDCreate,
@@ -1624,7 +1675,7 @@ static int32_t mnodeDoCreateChildTableCb(SMnodeMsg *pMsg, int32_t code) {
       .msgType = TSDB_MSG_TYPE_MD_CREATE_TABLE
   };
 
-  dnodeSendMsgToDnode(&ipSet, &rpcMsg);
+  dnodeSendMsgToDnode(&epSet, &rpcMsg);
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
@@ -1766,18 +1817,13 @@ static int32_t mnodeProcessCreateChildTableMsg(SMnodeMsg *pMsg) {
   }
 }
 
-static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) {
+static int32_t mnodeSendDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) {
   SChildTableObj *pTable = (SChildTableObj *)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,
-           pTable->info.tableId);
-    return TSDB_CODE_MND_APP_ERROR;
-  }
+  mLInfo("app:%p:%p, ctable:%s, is dropped from sdb", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId);
 
   SMDDropTableMsg *pDrop = rpcMallocCont(sizeof(SMDDropTableMsg));
   if (pDrop == NULL) {
-    mError("app:%p:%p, table:%s, failed to drop ctable, no enough memory", pMsg->rpcMsg.ahandle, pMsg,
+    mError("app:%p:%p, ctable:%s, failed to drop ctable, no enough memory", pMsg->rpcMsg.ahandle, pMsg,
            pTable->info.tableId);
     return TSDB_CODE_MND_OUT_OF_MEMORY;
   }
@@ -1788,9 +1834,9 @@ static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) {
   pDrop->sid     = htonl(pTable->sid);
   pDrop->uid     = htobe64(pTable->uid);
 
-  SRpcIpSet ipSet = mnodeGetIpSetFromVgroup(pMsg->pVgroup);
+  SRpcEpSet epSet = mnodeGetEpSetFromVgroup(pMsg->pVgroup);
 
-  mInfo("app:%p:%p, table:%s, send drop ctable msg, vgId:%d sid:%d uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg,
+  mInfo("app:%p:%p, ctable:%s, send drop ctable msg, vgId:%d sid:%d uid:%" PRIu64, pMsg->rpcMsg.ahandle, pMsg,
         pDrop->tableId, pTable->vgId, pTable->sid, pTable->uid);
 
   SRpcMsg rpcMsg = {
@@ -1803,11 +1849,45 @@ static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg, bool needReturn) {
 
   if (!needReturn) rpcMsg.ahandle = NULL;
 
-  dnodeSendMsgToDnode(&ipSet, &rpcMsg);
+  dnodeSendMsgToDnode(&epSet, &rpcMsg);
 
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
+static int32_t mnodeDropChildTableCb(SMnodeMsg *pMsg, int32_t code) {
+  if (code != TSDB_CODE_SUCCESS) {
+    SChildTableObj *pTable = (SChildTableObj *)pMsg->pTable;
+    mError("app:%p:%p, ctable:%s, failed to drop, sdb error", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId);
+    return code;
+  } 
+
+  return mnodeSendDropChildTableMsg(pMsg, true);
+}
+
+static int32_t mnodeProcessDropChildTableMsg(SMnodeMsg *pMsg) {
+  SChildTableObj *pTable = (SChildTableObj *)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,
+           pTable->info.tableId);
+    return TSDB_CODE_MND_APP_ERROR;
+  }
+
+  SSdbOper oper = {
+    .type  = SDB_OPER_GLOBAL,
+    .table = tsChildTableSdb,
+    .pObj  = pTable,
+    .pMsg  = pMsg,
+    .cb    = mnodeDropChildTableCb
+  };
+
+  int32_t code = sdbDeleteRow(&oper);
+  if (code == TSDB_CODE_SUCCESS) {
+    return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+  }
+  return code;
+}
+
 static int32_t mnodeFindNormalTableColumnIndex(SChildTableObj *pTable, char *colName) {
   SSchema *schema = (SSchema *) pTable->schema;
   for (int32_t col = 0; col < pTable->numOfColumns; col++) {
@@ -1842,7 +1922,7 @@ static int32_t mnodeAlterNormalTableColumnCb(SMnodeMsg *pMsg, int32_t code) {
     }
   }
 
-  SRpcIpSet ipSet = mnodeGetIpSetFromVgroup(pMsg->pVgroup);
+  SRpcEpSet epSet = mnodeGetEpSetFromVgroup(pMsg->pVgroup);
   SRpcMsg rpcMsg = {
       .ahandle = pMsg,
       .pCont   = pMDCreate,
@@ -1854,7 +1934,7 @@ static int32_t mnodeAlterNormalTableColumnCb(SMnodeMsg *pMsg, int32_t code) {
   mDebug("app:%p:%p, ctable %s, send alter column msg to vgId:%d", pMsg->rpcMsg.ahandle, pMsg, pTable->info.tableId,
          pMsg->pVgroup->vgId);
 
-  dnodeSendMsgToDnode(&ipSet, &rpcMsg);
+  dnodeSendMsgToDnode(&epSet, &rpcMsg);
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
@@ -1949,6 +2029,48 @@ static int32_t mnodeDropNormalTableColumn(SMnodeMsg *pMsg, char *colName) {
   return code;
 }
 
+static int32_t mnodeChangeNormalTableColumn(SMnodeMsg *pMsg, char *oldName, char *newName) {
+  SChildTableObj *pTable = (SChildTableObj *)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,
+           pTable->info.tableId, oldName, newName);
+    return TSDB_CODE_MND_FIELD_NOT_EXIST;
+  }
+
+  // int32_t  rowSize = 0;
+  uint32_t len = strlen(newName);
+  if (len >= TSDB_COL_NAME_LEN) {
+    return TSDB_CODE_MND_COL_NAME_TOO_LONG;
+  }
+
+  if (mnodeFindNormalTableColumnIndex(pTable, newName) >= 0) {
+    return TSDB_CODE_MND_FIELD_ALREAY_EXIST;
+  }
+  
+  // update
+  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,
+         oldName, newName);
+
+  SSdbOper oper = {
+    .type = SDB_OPER_GLOBAL,
+    .table = tsChildTableSdb,
+    .pObj = pTable,
+    .pMsg = pMsg,
+    .cb = mnodeAlterNormalTableColumnCb
+  };
+
+  int32_t code = sdbUpdateRow(&oper);
+  if (code == TSDB_CODE_SUCCESS) {
+    code = TSDB_CODE_MND_ACTION_IN_PROGRESS;
+  }
+
+  return code;
+}
+
 static int32_t mnodeSetSchemaFromNormalTable(SSchema *pSchema, SChildTableObj *pTable) {
   int32_t numOfCols = pTable->numOfColumns;
   for (int32_t i = 0; i < numOfCols; ++i) {
@@ -1996,9 +2118,9 @@ static int32_t mnodeDoGetChildTableMeta(SMnodeMsg *pMsg, STableMetaMsg *pMeta) {
   for (int32_t i = 0; i < pMsg->pVgroup->numOfVnodes; ++i) {
     SDnodeObj *pDnode = mnodeGetDnode(pMsg->pVgroup->vnodeGid[i].dnodeId);
     if (pDnode == NULL) break;
-    strcpy(pMeta->vgroup.ipAddr[i].fqdn, pDnode->dnodeFqdn);
-    pMeta->vgroup.ipAddr[i].port = htons(pDnode->dnodePort + TSDB_PORT_DNODESHELL);
-    pMeta->vgroup.numOfIps++;
+    strcpy(pMeta->vgroup.epAddr[i].fqdn, pDnode->dnodeFqdn);
+    pMeta->vgroup.epAddr[i].port = htons(pDnode->dnodePort + TSDB_PORT_DNODESHELL);
+    pMeta->vgroup.numOfEps++;
     mnodeDecDnodeRef(pDnode);
   }
   pMeta->vgroup.vgId = htonl(pMsg->pVgroup->vgId);
@@ -2221,19 +2343,6 @@ static void mnodeProcessDropChildTableRsp(SRpcMsg *rpcMsg) {
     return;
   }
 
-  SSdbOper oper = {
-    .type = SDB_OPER_GLOBAL,
-    .table = tsChildTableSdb,
-    .pObj = pTable
-  };
-
-  int32_t code = sdbDeleteRow(&oper);
-  if (code != TSDB_CODE_SUCCESS) {
-    mError("app:%p:%p, table:%s, update ctables sdb error", mnodeMsg->rpcMsg.ahandle, mnodeMsg, pTable->info.tableId);
-    dnodeSendRpcMnodeWriteRsp(mnodeMsg, TSDB_CODE_MND_SDB_ERROR);
-    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);
@@ -2260,7 +2369,7 @@ static void mnodeProcessCreateChildTableRsp(SRpcMsg *rpcMsg) {
   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->sid, pTable->uid);
-    mnodeProcessDropChildTableMsg(mnodeMsg, false);
+    mnodeSendDropChildTableMsg(mnodeMsg, false);
     rpcMsg->code = TSDB_CODE_SUCCESS;
   }
 
@@ -2581,6 +2690,8 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) {
       code = mnodeAddSuperTableColumn(pMsg, pAlter->schema, 1);
     } else if (pAlter->type == TSDB_ALTER_TABLE_DROP_COLUMN) {
       code = mnodeDropSuperTableColumn(pMsg, pAlter->schema[0].name);
+    } else if (pAlter->type == TSDB_ALTER_TABLE_CHANGE_COLUMN) {
+      code = mnodeChangeSuperTableColumn(pMsg, pAlter->schema[0].name, pAlter->schema[1].name);
     } else {
     }
   } else {
@@ -2591,6 +2702,8 @@ static int32_t mnodeProcessAlterTableMsg(SMnodeMsg *pMsg) {
       code = mnodeAddNormalTableColumn(pMsg, pAlter->schema, 1);
     } else if (pAlter->type == TSDB_ALTER_TABLE_DROP_COLUMN) {
       code = mnodeDropNormalTableColumn(pMsg, pAlter->schema[0].name);
+    } else if (pAlter->type == TSDB_ALTER_TABLE_CHANGE_COLUMN) {
+      code = mnodeChangeNormalTableColumn(pMsg, pAlter->schema[0].name, pAlter->schema[1].name);
     } else {
     }
   }

src/mnode/src/mnodeUser.c

@@ -358,7 +358,7 @@ static int32_t mnodeRetrieveUsers(SShowObj *pShow, char *data, int32_t rows, voi
 }
 
 SUserObj *mnodeGetUserFromConn(void *pConn) {
-  SRpcConnInfo connInfo;
+  SRpcConnInfo connInfo = {0};
   if (rpcGetConnInfo(pConn, &connInfo) == 0) {
     return mnodeGetUser(connInfo.user);
   } else {

src/mnode/src/mnodeVgroup.c

@@ -317,9 +317,9 @@ void mnodeUpdateVgroupStatus(SVgObj *pVgroup, SDnodeObj *pDnode, SVnodeLoad *pVl
   }
 
   if (!dnodeExist) {
-    SRpcIpSet ipSet = mnodeGetIpSetFromIp(pDnode->dnodeEp);
+    SRpcEpSet epSet = mnodeGetEpSetFromIp(pDnode->dnodeEp);
     mError("vgId:%d, dnode:%d not exist in mnode, drop it", pVload->vgId, pDnode->dnodeId);
-    mnodeSendDropVnodeMsg(pVload->vgId, &ipSet, NULL);
+    mnodeSendDropVnodeMsg(pVload->vgId, &epSet, NULL);
     return;
   }
 
@@ -434,15 +434,22 @@ int32_t mnodeGetAvailableVgroup(SMnodeMsg *pMsg, SVgObj **ppVgroup, int32_t *pSi
   }
 
   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("app:%p:%p, db:%s, try to create a new vgroup, numOfVgroups:%d maxVgroupsPerDb:%d", pMsg->rpcMsg.ahandle,
+           pMsg, pDb->name, pDb->numOfVgroups, maxVgroupsPerDb);
     pthread_mutex_unlock(&pDb->mutex);
     int32_t code = mnodeCreateVgroup(pMsg);
-    if (code == TSDB_CODE_MND_ACTION_IN_PROGRESS) return code;
+    if (code == TSDB_CODE_MND_ACTION_IN_PROGRESS) {
+      return code;
+    } else {
+      pthread_mutex_lock(&pDb->mutex);
+    }
   }
 
   SVgObj *pVgroup = pDb->vgList[0];
-  if (pVgroup == NULL) return TSDB_CODE_MND_NO_ENOUGH_DNODES;
+  if (pVgroup == NULL) {
+    pthread_mutex_unlock(&pDb->mutex);
+    return TSDB_CODE_MND_NO_ENOUGH_DNODES;
+  }
 
   int32_t code = mnodeAllocVgroupIdPool(pVgroup);
   if (code != TSDB_CODE_SUCCESS) {
@@ -483,7 +490,7 @@ static int32_t mnodeCreateVgroupCb(SMnodeMsg *pMsg, int32_t code) {
   } else {
     pVgroup->status = TAOS_VG_STATUS_READY;
     SSdbOper desc = {.type = SDB_OPER_GLOBAL, .pObj = pVgroup, .table = tsVgroupSdb};
-    sdbUpdateRow(&desc);
+    (void)sdbUpdateRow(&desc);
   }
 
   mInfo("app:%p:%p, vgId:%d, is created in mnode, db:%s replica:%d", pMsg->rpcMsg.ahandle, pMsg, pVgroup->vgId,
@@ -585,7 +592,7 @@ static int32_t mnodeGetVgroupMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *p
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
   cols++;
 
-  pShow->bytes[cols] = 12 + VARSTR_HEADER_SIZE;
+  pShow->bytes[cols] = 8 + VARSTR_HEADER_SIZE;
   pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
   strcpy(pSchema[cols].name, "status");
   pSchema[cols].bytes = htons(pShow->bytes[cols]);
@@ -612,12 +619,6 @@ static int32_t mnodeGetVgroupMeta(STableMetaMsg *pMeta, SShowObj *pShow, void *p
     pSchema[cols].bytes = htons(pShow->bytes[cols]);
     cols++;
 
-    pShow->bytes[cols] = 40 + VARSTR_HEADER_SIZE;
-    pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
-    strcpy(pSchema[cols].name, "end_point");
-    pSchema[cols].bytes = htons(pShow->bytes[cols]);
-    cols++;
-
     pShow->bytes[cols] = 9 + VARSTR_HEADER_SIZE;
     pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
     strcpy(pSchema[cols].name, "vstatus");
@@ -709,27 +710,15 @@ static int32_t mnodeRetrieveVgroups(SShowObj *pShow, char *data, int32_t rows, v
       *(int16_t *) pWrite = pVgroup->vnodeGid[i].dnodeId;
       cols++;
 
-      SDnodeObj *pDnode = pVgroup->vnodeGid[i].pDnode;
-
+      SDnodeObj * pDnode = pVgroup->vnodeGid[i].pDnode;
+      const char *role = "NULL";
       if (pDnode != NULL) {
-        pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-        STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pDnode->dnodeEp, pShow->bytes[cols]);
-        cols++;
-
-        pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-        char *role = mnodeGetMnodeRoleStr(pVgroup->vnodeGid[i].role);
-        STR_WITH_MAXSIZE_TO_VARSTR(pWrite, role, pShow->bytes[cols]);
-        cols++;
-      } else {
-        pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-        const char *src = "NULL";
-        STR_WITH_SIZE_TO_VARSTR(pWrite, src, strlen(src));
-        cols++;
-        
-        pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-        STR_WITH_SIZE_TO_VARSTR(pWrite, src, strlen(src));
-        cols++;
+        role = mnodeGetMnodeRoleStr(pVgroup->vnodeGid[i].role);
       }
+
+      pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+      STR_WITH_MAXSIZE_TO_VARSTR(pWrite, role, pShow->bytes[cols]);
+      cols++;
     }
 
     mnodeDecVgroupRef(pVgroup);
@@ -809,29 +798,29 @@ static SMDCreateVnodeMsg *mnodeBuildVnodeMsg(SVgObj *pVgroup) {
   return pVnode;
 }
 
-SRpcIpSet mnodeGetIpSetFromVgroup(SVgObj *pVgroup) {
-  SRpcIpSet ipSet = {
-    .numOfIps = pVgroup->numOfVnodes,
+SRpcEpSet mnodeGetEpSetFromVgroup(SVgObj *pVgroup) {
+  SRpcEpSet epSet = {
+    .numOfEps = pVgroup->numOfVnodes,
     .inUse = 0,
   };
   for (int i = 0; i < pVgroup->numOfVnodes; ++i) {
-    strcpy(ipSet.fqdn[i], pVgroup->vnodeGid[i].pDnode->dnodeFqdn);
-    ipSet.port[i] = pVgroup->vnodeGid[i].pDnode->dnodePort + TSDB_PORT_DNODEDNODE;
+    strcpy(epSet.fqdn[i], pVgroup->vnodeGid[i].pDnode->dnodeFqdn);
+    epSet.port[i] = pVgroup->vnodeGid[i].pDnode->dnodePort + TSDB_PORT_DNODEDNODE;
   }
-  return ipSet;
+  return epSet;
 }
 
-SRpcIpSet mnodeGetIpSetFromIp(char *ep) {
-  SRpcIpSet ipSet;
+SRpcEpSet mnodeGetEpSetFromIp(char *ep) {
+  SRpcEpSet epSet;
 
-  ipSet.numOfIps = 1;
-  ipSet.inUse = 0;
-  taosGetFqdnPortFromEp(ep, ipSet.fqdn[0], &ipSet.port[0]);
-  ipSet.port[0] += TSDB_PORT_DNODEDNODE;
-  return ipSet;
+  epSet.numOfEps = 1;
+  epSet.inUse = 0;
+  taosGetFqdnPortFromEp(ep, epSet.fqdn[0], &epSet.port[0]);
+  epSet.port[0] += TSDB_PORT_DNODEDNODE;
+  return epSet;
 }
 
-static void mnodeSendAlterVnodeMsg(SVgObj *pVgroup, SRpcIpSet *ipSet) {
+static void mnodeSendAlterVnodeMsg(SVgObj *pVgroup, SRpcEpSet *epSet) {
   SMDAlterVnodeMsg *pAlter = mnodeBuildVnodeMsg(pVgroup);
   SRpcMsg rpcMsg = {
     .ahandle = NULL,
@@ -840,21 +829,21 @@ static void mnodeSendAlterVnodeMsg(SVgObj *pVgroup, SRpcIpSet *ipSet) {
     .code    = 0,
     .msgType = TSDB_MSG_TYPE_MD_ALTER_VNODE
   };
-  dnodeSendMsgToDnode(ipSet, &rpcMsg);
+  dnodeSendMsgToDnode(epSet, &rpcMsg);
 }
 
 void mnodeSendAlterVgroupMsg(SVgObj *pVgroup) {
   mDebug("vgId:%d, send alter all vnodes msg, numOfVnodes:%d db:%s", pVgroup->vgId, pVgroup->numOfVnodes,
          pVgroup->dbName);
   for (int32_t i = 0; i < pVgroup->numOfVnodes; ++i) {
-    SRpcIpSet ipSet = mnodeGetIpSetFromIp(pVgroup->vnodeGid[i].pDnode->dnodeEp);
+    SRpcEpSet epSet = mnodeGetEpSetFromIp(pVgroup->vnodeGid[i].pDnode->dnodeEp);
     mDebug("vgId:%d, index:%d, send alter vnode msg to dnode %s", pVgroup->vgId, i,
            pVgroup->vnodeGid[i].pDnode->dnodeEp);
-    mnodeSendAlterVnodeMsg(pVgroup, &ipSet);
+    mnodeSendAlterVnodeMsg(pVgroup, &epSet);
   }
 }
 
-static void mnodeSendCreateVnodeMsg(SVgObj *pVgroup, SRpcIpSet *ipSet, void *ahandle) {
+static void mnodeSendCreateVnodeMsg(SVgObj *pVgroup, SRpcEpSet *epSet, void *ahandle) {
   SMDCreateVnodeMsg *pCreate = mnodeBuildVnodeMsg(pVgroup);
   SRpcMsg rpcMsg = {
     .ahandle = ahandle,
@@ -863,17 +852,17 @@ static void mnodeSendCreateVnodeMsg(SVgObj *pVgroup, SRpcIpSet *ipSet, void *aha
     .code    = 0,
     .msgType = TSDB_MSG_TYPE_MD_CREATE_VNODE
   };
-  dnodeSendMsgToDnode(ipSet, &rpcMsg);
+  dnodeSendMsgToDnode(epSet, &rpcMsg);
 }
 
 void mnodeSendCreateVgroupMsg(SVgObj *pVgroup, void *ahandle) {
   mDebug("vgId:%d, send create all vnodes msg, numOfVnodes:%d db:%s", pVgroup->vgId, pVgroup->numOfVnodes,
          pVgroup->dbName);
   for (int32_t i = 0; i < pVgroup->numOfVnodes; ++i) {
-    SRpcIpSet ipSet = mnodeGetIpSetFromIp(pVgroup->vnodeGid[i].pDnode->dnodeEp);
+    SRpcEpSet epSet = mnodeGetEpSetFromIp(pVgroup->vnodeGid[i].pDnode->dnodeEp);
     mDebug("vgId:%d, index:%d, send create vnode msg to dnode %s, ahandle:%p", pVgroup->vgId,
            i, pVgroup->vnodeGid[i].pDnode->dnodeEp, ahandle);
-    mnodeSendCreateVnodeMsg(pVgroup, &ipSet, ahandle);
+    mnodeSendCreateVnodeMsg(pVgroup, &epSet, ahandle);
   }
 }
 
@@ -926,7 +915,7 @@ static SMDDropVnodeMsg *mnodeBuildDropVnodeMsg(int32_t vgId) {
   return pDrop;
 }
 
-void mnodeSendDropVnodeMsg(int32_t vgId, SRpcIpSet *ipSet, void *ahandle) {
+void mnodeSendDropVnodeMsg(int32_t vgId, SRpcEpSet *epSet, void *ahandle) {
   SMDDropVnodeMsg *pDrop = mnodeBuildDropVnodeMsg(vgId);
   SRpcMsg rpcMsg = {
       .ahandle = ahandle,
@@ -935,16 +924,16 @@ void mnodeSendDropVnodeMsg(int32_t vgId, SRpcIpSet *ipSet, void *ahandle) {
       .code    = 0,
       .msgType = TSDB_MSG_TYPE_MD_DROP_VNODE
   };
-  dnodeSendMsgToDnode(ipSet, &rpcMsg);
+  dnodeSendMsgToDnode(epSet, &rpcMsg);
 }
 
 static void mnodeSendDropVgroupMsg(SVgObj *pVgroup, void *ahandle) {
   pVgroup->status = TAOS_VG_STATUS_DROPPING; // deleting
   mDebug("vgId:%d, send drop all vnodes msg, ahandle:%p", pVgroup->vgId, ahandle);
   for (int32_t i = 0; i < pVgroup->numOfVnodes; ++i) {
-    SRpcIpSet ipSet = mnodeGetIpSetFromIp(pVgroup->vnodeGid[i].pDnode->dnodeEp);
+    SRpcEpSet epSet = mnodeGetEpSetFromIp(pVgroup->vnodeGid[i].pDnode->dnodeEp);
     mDebug("vgId:%d, send drop vnode msg to dnode:%d, ahandle:%p", pVgroup->vgId, pVgroup->vnodeGid[i].dnodeId, ahandle);
-    mnodeSendDropVnodeMsg(pVgroup->vgId, &ipSet, ahandle);
+    mnodeSendDropVnodeMsg(pVgroup->vgId, &epSet, ahandle);
   }
 }
 
@@ -998,8 +987,8 @@ static int32_t mnodeProcessVnodeCfgMsg(SMnodeMsg *pMsg) {
   }
 
   mDebug("vgId:%d, send create vnode msg to dnode %s for vnode cfg msg", pVgroup->vgId, pDnode->dnodeEp);
-  SRpcIpSet ipSet = mnodeGetIpSetFromIp(pDnode->dnodeEp);
-  mnodeSendCreateVnodeMsg(pVgroup, &ipSet, NULL);
+  SRpcEpSet epSet = mnodeGetEpSetFromIp(pDnode->dnodeEp);
+  mnodeSendCreateVnodeMsg(pVgroup, &epSet, NULL);
 
   mnodeDecDnodeRef(pDnode);
   mnodeDecVgroupRef(pVgroup);

src/mnode/src/mnodeWrite.c

@@ -49,16 +49,16 @@ int32_t mnodeProcessWrite(SMnodeMsg *pMsg) {
 
   if (!sdbIsMaster()) {
     SMnodeRsp *rpcRsp = &pMsg->rpcRsp;
-    SRpcIpSet *ipSet = rpcMallocCont(sizeof(SRpcIpSet));
-    mnodeGetMnodeIpSetForShell(ipSet);
-    rpcRsp->rsp = ipSet;
-    rpcRsp->len = sizeof(SRpcIpSet);
+    SRpcEpSet *epSet = rpcMallocCont(sizeof(SRpcEpSet));
+    mnodeGetMnodeEpSetForShell(epSet);
+    rpcRsp->rsp = epSet;
+    rpcRsp->len = sizeof(SRpcEpSet);
 
     mDebug("app:%p:%p, msg:%s will be redireced inUse:%d", pMsg->rpcMsg.ahandle, pMsg, taosMsg[pMsg->rpcMsg.msgType],
-           ipSet->inUse);
-    for (int32_t i = 0; i < ipSet->numOfIps; ++i) {
-      mDebug("app:%p:%p, mnode index:%d ip:%s:%d", pMsg->rpcMsg.ahandle, pMsg, i, ipSet->fqdn[i],
-             htons(ipSet->port[i]));
+           epSet->inUse);
+    for (int32_t i = 0; i < epSet->numOfEps; ++i) {
+      mDebug("app:%p:%p, mnode index:%d ep:%s:%d", pMsg->rpcMsg.ahandle, pMsg, i, epSet->fqdn[i],
+             htons(epSet->port[i]));
     }
 
     return TSDB_CODE_RPC_REDIRECT;

src/plugins/http/src/gcJson.c

@@ -121,6 +121,10 @@ bool gcBuildQueryJson(HttpContext *pContext, HttpSqlCmd *cmd, TAOS_RES *result,
 
   for (int k = 0; k < numOfRows; ++k) {
     TAOS_ROW row = taos_fetch_row(result);
+    if (row == NULL) {
+      cmd->numOfRows--;
+      continue;
+    }
     int32_t* length = taos_fetch_lengths(result);
 
     // for group by

src/plugins/http/src/httpContext.c

@@ -108,7 +108,7 @@ HttpContext *httpCreateContext(int32_t fd) {
   pContext->lastAccessTime = taosGetTimestampSec();
   pContext->state = HTTP_CONTEXT_STATE_READY;
 
-  HttpContext **ppContext = taosCachePut(tsHttpServer.contextCache, &pContext, sizeof(void *), &pContext, sizeof(void *), 3);
+  HttpContext **ppContext = taosCachePut(tsHttpServer.contextCache, &pContext, sizeof(int64_t), &pContext, sizeof(int64_t), 3);
   pContext->ppContext = ppContext;
   httpDebug("context:%p, fd:%d, is created, data:%p", pContext, fd, ppContext);
 

src/plugins/http/src/restJson.c

@@ -94,6 +94,10 @@ bool restBuildSqlJson(HttpContext *pContext, HttpSqlCmd *cmd, TAOS_RES *result,
 
   for (int k = 0; k < numOfRows; ++k) {
     TAOS_ROW row = taos_fetch_row(result);
+    if (row == NULL) {
+      cmd->numOfRows--;
+      continue;
+    }
     int32_t* length = taos_fetch_lengths(result);
 
     // data row array begin

src/plugins/monitor/src/monitorMain.c

@@ -27,7 +27,6 @@
 #include "dnode.h"
 #include "monitor.h"
 
-
 #define monitorFatal(...) { if (monitorDebugFlag & DEBUG_FATAL) { taosPrintLog("MON FATAL ", 255, __VA_ARGS__); }}
 #define monitorError(...) { if (monitorDebugFlag & DEBUG_ERROR) { taosPrintLog("MON ERROR ", 255, __VA_ARGS__); }}
 #define monitorWarn(...)  { if (monitorDebugFlag & DEBUG_WARN)  { taosPrintLog("MON WARN ", 255, __VA_ARGS__); }}
@@ -78,6 +77,7 @@ static void monitorStartTimer();
 static void monitorSaveSystemInfo();
 extern int32_t (*monitorStartSystemFp)();
 extern void (*monitorStopSystemFp)();
+extern void (*monitorExecuteSQLFp)(char *sql);
 
 static void monitorCheckDiskUsage(void *para, void *unused) {
   taosGetDisk();
@@ -207,6 +207,7 @@ static void monitorInitDatabase() {
     taos_query_a(tsMonitorConn.conn, tsMonitorConn.sql, monitorInitDatabaseCb, NULL);
   } else {
     tsMonitorConn.state = MONITOR_STATE_INITIALIZED;
+    monitorExecuteSQLFp = monitorExecuteSQL;
     monitorInfo("monitor service init success");
 
     monitorStartTimer();
@@ -230,6 +231,7 @@ static void monitorInitDatabaseCb(void *param, TAOS_RES *result, int32_t code) {
 
 void monitorStopSystem() {
   monitorInfo("monitor module is stopped");
+  monitorExecuteSQLFp = NULL;
   tsMonitorConn.state = MONITOR_STATE_STOPPED;
   if (tsMonitorConn.initTimer != NULL) {
     taosTmrStopA(&(tsMonitorConn.initTimer));
@@ -248,33 +250,13 @@ static void monitorStartTimer() {
   taosTmrReset(monitorSaveSystemInfo, tsMonitorInterval * 1000, NULL, tscTmr, &tsMonitorConn.timer);
 }
 
-static void dnodeMontiorInsertAcctCallback(void *param, TAOS_RES *result, int32_t code) {
-  if (code < 0) {
-    monitorError("monitor:%p, save account info failed, code:%s", tsMonitorConn.conn, tstrerror(code));
-  } else if (code == 0) {
-    monitorError("monitor:%p, save account info failed, affect rows:%d", tsMonitorConn.conn, code);
-  } else {
-    monitorDebug("monitor:%p, save account info success, code:%s", tsMonitorConn.conn, tstrerror(code));
-  }
-}
-
-static void dnodeMontiorInsertSysCallback(void *param, TAOS_RES *result, int32_t code) {
+static void dnodeMontiorLogCallback(void *param, TAOS_RES *result, int32_t code) {
   if (code < 0) {
-    monitorError("monitor:%p, save system info failed, code:%s %s", tsMonitorConn.conn, tstrerror(code), tsMonitorConn.sql);
+    monitorError("monitor:%p, save %s failed, reason:%s", tsMonitorConn.conn, (char *)param, tstrerror(code));
   } else if (code == 0) {
-    monitorError("monitor:%p, save system info failed, affect rows:%d %s", tsMonitorConn.conn, code, tsMonitorConn.sql);
+    monitorError("monitor:%p, save %s failed, affect rows:%d", tsMonitorConn.conn, (char *)param, code);
   } else {
-    monitorDebug("monitor:%p, save system info success, code:%s %s", tsMonitorConn.conn, tstrerror(code), tsMonitorConn.sql);
-  }
-}
-
-static void dnodeMontiorInsertLogCallback(void *param, TAOS_RES *result, int32_t code) {
-  if (code < 0) {
-    monitorError("monitor:%p, save log failed, code:%s", tsMonitorConn.conn, tstrerror(code));
-  } else if (code == 0) {
-    monitorError("monitor:%p, save log failed, affect rows:%d", tsMonitorConn.conn, code);
-  } else {
-    monitorDebug("monitor:%p, save log info success, code:%s", tsMonitorConn.conn, tstrerror(code));
+    monitorDebug("monitor:%p, save %s info success, reason:%s", tsMonitorConn.conn, (char *)param, tstrerror(code));
   }
 }
 
@@ -359,7 +341,7 @@ static void monitorSaveSystemInfo() {
   pos += monitorBuildReqSql(sql + pos);
 
   monitorDebug("monitor:%p, save system info, sql:%s", tsMonitorConn.conn, sql);
-  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorInsertSysCallback, "log");
+  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorLogCallback, "sys");
 
   if (tsMonitorConn.timer != NULL && tsMonitorConn.state != MONITOR_STATE_STOPPED) {
     monitorStartTimer();
@@ -397,7 +379,7 @@ void monitorSaveAcctLog(SAcctMonitorObj *pMon) {
           pMon->accessState);
 
   monitorDebug("monitor:%p, save account info, sql %s", tsMonitorConn.conn, sql);
-  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorInsertAcctCallback, "account");
+  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorLogCallback, "account");
 }
 
 void monitorSaveLog(int32_t level, const char *const format, ...) {
@@ -421,14 +403,11 @@ void monitorSaveLog(int32_t level, const char *const format, ...) {
   sql[len++] = 0;
 
   monitorDebug("monitor:%p, save log, sql: %s", tsMonitorConn.conn, sql);
-  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorInsertLogCallback, "log");
+  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorLogCallback, "log");
 }
 
 void monitorExecuteSQL(char *sql) {
   if (tsMonitorConn.state != MONITOR_STATE_INITIALIZED) return;
-
   monitorDebug("monitor:%p, execute sql: %s", tsMonitorConn.conn, sql);
-  
-  // bug while insert binary
-  // taos_query_a(tsMonitorConn.conn, sql, NULL, NULL);
+  taos_query_a(tsMonitorConn.conn, sql, dnodeMontiorLogCallback, "sql");
 }

src/plugins/mqtt/src/mqttSystem.c

@@ -64,7 +64,7 @@ int32_t mqttInitSystem() {
   }
 
   char* _begin_hostname = strstr(url, recntStatus.hostname);
-  if (strstr(_begin_hostname, ":") != NULL) {
+  if (_begin_hostname != NULL && strstr(_begin_hostname, ":") != NULL) {
     recntStatus.port = strbetween(_begin_hostname, ":", "/");
   } else {
     recntStatus.port = strbetween("'1883'", "'", "'");

src/query/inc/qast.h → src/query/inc/qAst.h

@@ -45,7 +45,6 @@ typedef void (*__do_filter_suppl_fn_t)(void *, void *);
  *
  */
 typedef struct tQueryInfo {
-  int32_t       colIndex; // index of column in schema
   uint8_t       optr;     // expression operator
   SSchema       sch;      // schema of tags
   char*         q;

src/query/inc/qExecutor.h

@@ -18,16 +18,16 @@
 #include "os.h"
 
 #include "hash.h"
-#include "qfill.h"
-#include "qresultBuf.h"
+#include "qFill.h"
+#include "qResultbuf.h"
+#include "qTsbuf.h"
 #include "qsqlparser.h"
-#include "qtsbuf.h"
+#include "query.h"
 #include "taosdef.h"
 #include "tarray.h"
 #include "tlockfree.h"
 #include "tsdb.h"
 #include "tsqlfunction.h"
-#include "query.h"
 
 struct SColumnFilterElem;
 typedef bool (*__filter_func_t)(struct SColumnFilterElem* pFilter, char* val1, char* val2);
@@ -42,7 +42,7 @@ typedef struct SSqlGroupbyExpr {
 } SSqlGroupbyExpr;
 
 typedef struct SPosInfo {
-  int16_t pageId;
+  int32_t pageId;
   int16_t rowId;
 } SPosInfo;
 
@@ -158,7 +158,7 @@ typedef struct SQueryRuntimeEnv {
   SResultInfo*         resultInfo;       // todo refactor to merge with SWindowResInfo
   SQuery*              pQuery;
   SQLFunctionCtx*      pCtx;
-  int16_t              numOfRowsPerPage;
+  int32_t              numOfRowsPerPage;
   int16_t              offset[TSDB_MAX_COLUMNS];
   uint16_t             scanFlag;         // denotes reversed scan of data or not
   SFillInfo*           pFillInfo;

src/query/inc/qextbuffer.h → src/query/inc/qExtbuffer.h

@@ -28,7 +28,7 @@ extern "C" {
 #include "tdataformat.h"
 #include "talgo.h"
 
-#define DEFAULT_PAGE_SIZE (1024L*64)  // 16k larger than the SHistoInfo
+#define DEFAULT_PAGE_SIZE (1024L*4)  // 16k larger than the SHistoInfo
 #define MAX_TMPFILE_PATH_LENGTH PATH_MAX
 #define INITIAL_ALLOCATION_BUFFER_SIZE 64
 

src/query/inc/qfill.h → src/query/inc/qFill.h

@@ -21,8 +21,8 @@ extern "C" {
 #endif
 
 #include "os.h"
+#include "qExtbuffer.h"
 #include "taosdef.h"
-#include "qextbuffer.h"
 
 typedef struct {
   STColumn col;             // column info

src/query/inc/qpercentile.h → src/query/inc/qPercentile.h

@@ -16,7 +16,7 @@
 #ifndef TDENGINE_QPERCENTILE_H
 #define TDENGINE_QPERCENTILE_H
 
-#include "qextbuffer.h"
+#include "qExtbuffer.h"
 
 typedef struct MinMaxEntry {
   union {

src/query/inc/qresultBuf.h → src/query/inc/qResultbuf.h

@@ -20,9 +20,9 @@
 extern "C" {
 #endif
 
-#include "os.h"
-#include "qextbuffer.h"
 #include "hash.h"
+#include "os.h"
+#include "qExtbuffer.h"
 
 typedef struct SArray* SIDList;
 
@@ -33,14 +33,20 @@ typedef struct SDiskbasedResultBuf {
   int32_t   fd;                  // data file fd
   int32_t   allocateId;          // allocated page id
   int32_t   incStep;             // minimum allocated pages
-  char*     pBuf;                // mmap buffer pointer
+  void*     pBuf;                // mmap buffer pointer
   char*     path;                // file path
-  
+  int32_t   pageSize;            // current used page size
+  int32_t   inMemPages;          // numOfPages that are allocated in memory
   SHashObj* idsTable;            // id hash table
   SIDList   list;                // for each id, there is a page id list
+
+  void*     iBuf;                // inmemory buf
+  void*     handle;              // for debug purpose
+  void*     emptyDummyIdList;    // dummy id list
 } SDiskbasedResultBuf;
 
-#define DEFAULT_INTERN_BUF_PAGE_SIZE (8192L*5)
+#define DEFAULT_INTERN_BUF_PAGE_SIZE (1024L)
+#define DEFAULT_INMEM_BUF_PAGES       10
 
 /**
  * create disk-based result buffer
@@ -49,7 +55,8 @@ typedef struct SDiskbasedResultBuf {
  * @param rowSize
  * @return
  */
-int32_t createDiskbasedResultBuffer(SDiskbasedResultBuf** pResultBuf, int32_t size, int32_t rowSize, void* handle);
+int32_t createDiskbasedResultBuffer(SDiskbasedResultBuf** pResultBuf, int32_t numOfPages, int32_t rowSize, int32_t pagesize,
+    int32_t inMemPages, void* handle);
 
 /**
  *
@@ -81,8 +88,13 @@ SIDList getDataBufPagesIdList(SDiskbasedResultBuf* pResultBuf, int32_t groupId);
  * @param id
  * @return
  */
-#define GET_RES_BUF_PAGE_BY_ID(buf, id)  ((tFilePage*)((buf)->pBuf + DEFAULT_INTERN_BUF_PAGE_SIZE*(id)))
-
+static FORCE_INLINE tFilePage* getResBufPage(SDiskbasedResultBuf* pResultBuf, int32_t id) {
+  if (id < pResultBuf->inMemPages) {
+    return (tFilePage*) ((char*) pResultBuf->iBuf + id * pResultBuf->pageSize);
+  } else {
+    return (tFilePage*) ((char*) pResultBuf->pBuf + (id - pResultBuf->inMemPages) * pResultBuf->pageSize);
+  }
+}
 /**
  * get the total buffer size in the format of disk file
  * @param pResultBuf

src/query/inc/qUtil.h

@@ -49,7 +49,7 @@ static FORCE_INLINE char *getPosInResultPage(SQueryRuntimeEnv *pRuntimeEnv, int3
   assert(pResult != NULL && pRuntimeEnv != NULL);
 
   SQuery    *pQuery = pRuntimeEnv->pQuery;
-  tFilePage *page = GET_RES_BUF_PAGE_BY_ID(pRuntimeEnv->pResultBuf, pResult->pos.pageId);
+  tFilePage *page = getResBufPage(pRuntimeEnv->pResultBuf, pResult->pos.pageId);
   int32_t realRowId = pResult->pos.rowId * GET_ROW_PARAM_FOR_MULTIOUTPUT(pQuery, pRuntimeEnv->topBotQuery, pRuntimeEnv->stableQuery);
 
   return ((char *)page->data) + pRuntimeEnv->offset[columnIndex] * pRuntimeEnv->numOfRowsPerPage +
@@ -59,6 +59,4 @@ static FORCE_INLINE char *getPosInResultPage(SQueryRuntimeEnv *pRuntimeEnv, int3
 __filter_func_t *getRangeFilterFuncArray(int32_t type);
 __filter_func_t *getValueFilterFuncArray(int32_t type);
 
-bool supportPrefilter(int32_t type);
-
 #endif  // TDENGINE_QUERYUTIL_H

src/query/inc/tsqlfunction.h

@@ -255,7 +255,15 @@ extern int32_t functionCompatList[]; // compatible check array list
 
 bool topbot_datablock_filter(SQLFunctionCtx *pCtx, int32_t functionId, const char *minval, const char *maxval);
 
-void resetResultInfo(SResultInfo *pResInfo);
+/**
+ * the numOfRes should be kept, since it may be used later
+ * and allow the ResultInfo to be re initialized
+ */
+#define RESET_RESULT_INFO(_r)  \
+  do {                         \
+    (_r)->initialized = false; \
+  } while (0)
+
 void setResultInfoBuf(SResultInfo *pResInfo, int32_t size, bool superTable, char* buf);
 
 static FORCE_INLINE void initResultInfo(SResultInfo *pResInfo) {

src/query/src/qast.c → src/query/src/qAst.c

@@ -16,17 +16,17 @@
 
 #include "os.h"
 
-#include "tname.h"
-#include "qast.h"
-#include "tsdb.h"
 #include "exception.h"
+#include "qAst.h"
+#include "qSyntaxtreefunction.h"
 #include "qsqlparser.h"
-#include "qsyntaxtreefunction.h"
 #include "taosdef.h"
 #include "taosmsg.h"
 #include "tarray.h"
 #include "tbuffer.h"
 #include "tcompare.h"
+#include "tname.h"
+#include "tsdb.h"
 #include "tskiplist.h"
 #include "tsqlfunction.h"
 #include "tstoken.h"
@@ -678,7 +678,7 @@ static void tQueryIndexlessColumn(SSkipList* pSkipList, tQueryInfo* pQueryInfo,
 
     tstr *name = (tstr*) tsdbGetTableName(*(void**) pData);
     // todo speed up by using hash
-    if (pQueryInfo->colIndex == TSDB_TBNAME_COLUMN_INDEX) {
+    if (pQueryInfo->sch.colId == TSDB_TBNAME_COLUMN_INDEX) {
       if (pQueryInfo->optr == TSDB_RELATION_IN) {
         addToResult = pQueryInfo->compare(name, pQueryInfo->q);
       } else if (pQueryInfo->optr == TSDB_RELATION_LIKE) {
@@ -716,7 +716,7 @@ void tExprTreeTraverse(tExprNode *pExpr, SSkipList *pSkipList, SArray *result, S
     }
 
     tQueryInfo *pQueryInfo = pExpr->_node.info;
-    if (pQueryInfo->colIndex == 0 && pQueryInfo->optr != TSDB_RELATION_LIKE) {
+    if (pQueryInfo->sch.colId == PRIMARYKEY_TIMESTAMP_COL_INDEX && pQueryInfo->optr != TSDB_RELATION_LIKE) {
       tQueryIndexColumn(pSkipList, pQueryInfo, result);
     } else {
       tQueryIndexlessColumn(pSkipList, pQueryInfo, result, param->nodeFilterFn);

src/query/src/qextbuffer.c → src/query/src/qExtbuffer.c

@@ -12,16 +12,15 @@
  * You should have received a copy of the GNU Affero General Public License
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
+#include "qExtbuffer.h"
 #include "os.h"
-#include "tulog.h"
-#include "qextbuffer.h"
+#include "queryLog.h"
 #include "taos.h"
 #include "taosdef.h"
 #include "taosmsg.h"
 #include "tsqlfunction.h"
-#include "ttime.h"
+#include "tulog.h"
 #include "tutil.h"
-#include "queryLog.h"
 
 #define COLMODEL_GET_VAL(data, schema, allrow, rowId, colId) \
   (data + (schema)->pFields[colId].offset * (allrow) + (rowId) * (schema)->pFields[colId].field.bytes)

src/query/src/qfill.c → src/query/src/qFill.c

@@ -13,9 +13,9 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
+#include "qFill.h"
 #include "os.h"
-#include "qfill.h"
-#include "qextbuffer.h"
+#include "qExtbuffer.h"
 #include "taosdef.h"
 #include "taosmsg.h"
 #include "tsqlfunction.h"

src/query/src/qFilterFunc.c → src/query/src/qFilterfunc.c

@@ -554,5 +554,3 @@ __filter_func_t* getValueFilterFuncArray(int32_t type) {
     default: return NULL;
   }
 }
-
-bool supportPrefilter(int32_t type) { return type != TSDB_DATA_TYPE_BINARY && type != TSDB_DATA_TYPE_NCHAR; }

src/query/src/qhistogram.c → src/query/src/qHistogram.c

@@ -14,7 +14,7 @@
  */
 #include "os.h"
 
-#include "qhistogram.h"
+#include "qHistogram.h"
 #include "taosdef.h"
 #include "taosmsg.h"
 #include "tlosertree.h"

src/query/src/qpercentile.c → src/query/src/qPercentile.c

@@ -13,12 +13,12 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
+#include "qPercentile.h"
 #include "os.h"
-#include "tulog.h"
-#include "qpercentile.h"
+#include "queryLog.h"
 #include "taosdef.h"
 #include "taosmsg.h"
-#include "queryLog.h"
+#include "tulog.h"
 
 tExtMemBuffer *releaseBucketsExceptFor(tMemBucket *pMemBucket, int16_t segIdx, int16_t slotIdx) {
   tExtMemBuffer *pBuffer = NULL;

src/query/src/qresultBuf.c → src/query/src/qResultbuf.c

-#include "qresultBuf.h"
+#include "qResultbuf.h"
 #include "hash.h"
-#include "qextbuffer.h"
-#include "taoserror.h"
+#include "qExtbuffer.h"
 #include "queryLog.h"
+#include "taoserror.h"
+
+int32_t createDiskbasedResultBuffer(SDiskbasedResultBuf** pResultBuf, int32_t numOfPages, int32_t rowSize,
+    int32_t pagesize, int32_t inMemPages, void* handle) {
 
-int32_t createDiskbasedResultBuffer(SDiskbasedResultBuf** pResultBuf, int32_t size, int32_t rowSize, void* handle) {
   *pResultBuf = calloc(1, sizeof(SDiskbasedResultBuf));
   SDiskbasedResultBuf* pResBuf = *pResultBuf;
   if (pResBuf == NULL) {
     return TSDB_CODE_COM_OUT_OF_MEMORY;  
   }
-  
-  pResBuf->numOfRowsPerPage = (DEFAULT_INTERN_BUF_PAGE_SIZE - sizeof(tFilePage)) / rowSize;
-  pResBuf->numOfPages = size;
 
-  pResBuf->totalBufSize = pResBuf->numOfPages * DEFAULT_INTERN_BUF_PAGE_SIZE;
+  pResBuf->pageSize   = pagesize;
+  pResBuf->numOfPages = inMemPages;  // all pages are in buffer in the first place
+  pResBuf->inMemPages = inMemPages;
+  assert(inMemPages <= numOfPages);
+
+  pResBuf->numOfRowsPerPage = (pagesize - sizeof(tFilePage)) / rowSize;
+
+  pResBuf->totalBufSize = pResBuf->numOfPages * pagesize;
   pResBuf->incStep = 4;
+  pResBuf->allocateId = -1;
+
+  pResBuf->iBuf = calloc(pResBuf->inMemPages, pResBuf->pageSize);
 
   // init id hash table
-  pResBuf->idsTable = taosHashInit(size, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false);
-  pResBuf->list = taosArrayInit(size, POINTER_BYTES);
+  pResBuf->idsTable = taosHashInit(numOfPages, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false);
+  pResBuf->list = taosArrayInit(numOfPages, POINTER_BYTES);
 
-  char path[4096] = {0};
+  char path[PATH_MAX] = {0};
   getTmpfilePath("tsdb_qbuf", path);
   pResBuf->path = strdup(path);
 
-  pResBuf->fd = open(pResBuf->path, O_CREAT | O_RDWR, 0666);
-  if (!FD_VALID(pResBuf->fd)) {
-    qError("failed to create tmp file: %s on disk. %s", pResBuf->path, strerror(errno));
+  pResBuf->fd = FD_INITIALIZER;
+  pResBuf->pBuf = NULL;
+  pResBuf->emptyDummyIdList = taosArrayInit(1, sizeof(int32_t));
+
+  qDebug("QInfo:%p create resBuf for output, page size:%d, initial pages:%d, %" PRId64 "bytes", handle,
+      pResBuf->pageSize, pResBuf->numOfPages, pResBuf->totalBufSize);
+  
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t getNumOfResultBufGroupId(SDiskbasedResultBuf* pResultBuf) { return taosHashGetSize(pResultBuf->idsTable); }
+
+int32_t getResBufSize(SDiskbasedResultBuf* pResultBuf) { return pResultBuf->totalBufSize; }
+
+#define NUM_OF_PAGES_ON_DISK(_r) ((_r)->numOfPages - (_r)->inMemPages)
+#define FILE_SIZE_ON_DISK(_r) (NUM_OF_PAGES_ON_DISK(_r) * (_r)->pageSize)
+
+static int32_t createDiskResidesBuf(SDiskbasedResultBuf* pResultBuf) {
+  pResultBuf->fd = open(pResultBuf->path, O_CREAT | O_RDWR | O_TRUNC, 0666);
+  if (!FD_VALID(pResultBuf->fd)) {
+    qError("failed to create tmp file: %s on disk. %s", pResultBuf->path, strerror(errno));
     return TAOS_SYSTEM_ERROR(errno);
   }
 
-  int32_t ret = ftruncate(pResBuf->fd, pResBuf->numOfPages * DEFAULT_INTERN_BUF_PAGE_SIZE);
+  assert(pResultBuf->numOfPages == pResultBuf->inMemPages);
+  pResultBuf->numOfPages += pResultBuf->incStep;
+
+  int32_t ret = ftruncate(pResultBuf->fd, NUM_OF_PAGES_ON_DISK(pResultBuf) * pResultBuf->pageSize);
   if (ret != TSDB_CODE_SUCCESS) {
-    qError("failed to create tmp file: %s on disk. %s", pResBuf->path, strerror(errno));
+    qError("failed to create tmp file: %s on disk. %s", pResultBuf->path, strerror(errno));
     return TAOS_SYSTEM_ERROR(errno);
   }
 
-  pResBuf->pBuf = mmap(NULL, pResBuf->totalBufSize, PROT_READ | PROT_WRITE, MAP_SHARED, pResBuf->fd, 0);
-  if (pResBuf->pBuf == MAP_FAILED) {
-    qError("QInfo:%p failed to map temp file: %s. %s", handle, pResBuf->path, strerror(errno));
+  pResultBuf->pBuf = mmap(NULL, FILE_SIZE_ON_DISK(pResultBuf), PROT_READ | PROT_WRITE, MAP_SHARED, pResultBuf->fd, 0);
+  if (pResultBuf->pBuf == MAP_FAILED) {
+    qError("QInfo:%p failed to map temp file: %s. %s", pResultBuf->handle, pResultBuf->path, strerror(errno));
     return TAOS_SYSTEM_ERROR(errno);
   }
 
-  qDebug("QInfo:%p create tmp file for output result:%s, %" PRId64 "bytes", handle, pResBuf->path,
-      pResBuf->totalBufSize);
-  
+  pResultBuf->totalBufSize = pResultBuf->numOfPages * pResultBuf->pageSize;
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t getNumOfResultBufGroupId(SDiskbasedResultBuf* pResultBuf) { return taosHashGetSize(pResultBuf->idsTable); }
+static int32_t extendDiskFileSize(SDiskbasedResultBuf* pResultBuf, int32_t incNumOfPages) {
+  assert(pResultBuf->numOfPages * pResultBuf->pageSize == pResultBuf->totalBufSize);
+  int32_t ret = TSDB_CODE_SUCCESS;
 
-int32_t getResBufSize(SDiskbasedResultBuf* pResultBuf) { return pResultBuf->totalBufSize; }
+  if (pResultBuf->pBuf == NULL) {
+    assert(pResultBuf->fd == FD_INITIALIZER);
 
-static int32_t extendDiskFileSize(SDiskbasedResultBuf* pResultBuf, int32_t numOfPages) {
-  assert(pResultBuf->numOfPages * DEFAULT_INTERN_BUF_PAGE_SIZE == pResultBuf->totalBufSize);
-
-  int32_t ret = munmap(pResultBuf->pBuf, pResultBuf->totalBufSize);
-  pResultBuf->numOfPages += numOfPages;
-
-  /*
-   * disk-based output buffer is exhausted, try to extend the disk-based buffer, the available disk space may
-   * be insufficient
-   */
-  ret = ftruncate(pResultBuf->fd, pResultBuf->numOfPages * DEFAULT_INTERN_BUF_PAGE_SIZE);
-  if (ret != 0) {
-    //    dError("QInfo:%p failed to create intermediate result output file:%s. %s", pQInfo, pSupporter->extBufFile,
-    //           strerror(errno));
-    return TSDB_CODE_QRY_NO_DISKSPACE;
-  }
+    if ((ret = createDiskResidesBuf(pResultBuf)) != TSDB_CODE_SUCCESS) {
+      return ret;
+    }
+  } else {
+    ret = munmap(pResultBuf->pBuf, FILE_SIZE_ON_DISK(pResultBuf));
+    pResultBuf->numOfPages += incNumOfPages;
+
+    /*
+     * disk-based output buffer is exhausted, try to extend the disk-based buffer, the available disk space may
+     * be insufficient
+     */
+    ret = ftruncate(pResultBuf->fd, NUM_OF_PAGES_ON_DISK(pResultBuf) * pResultBuf->pageSize);
+    if (ret != TSDB_CODE_SUCCESS) {
+      //    dError("QInfo:%p failed to create intermediate result output file:%s. %s", pQInfo, pSupporter->extBufFile,
+      //           strerror(errno));
+      return TSDB_CODE_QRY_NO_DISKSPACE;
+    }
 
-  pResultBuf->totalBufSize = pResultBuf->numOfPages * DEFAULT_INTERN_BUF_PAGE_SIZE;
-  pResultBuf->pBuf = mmap(NULL, pResultBuf->totalBufSize, PROT_READ | PROT_WRITE, MAP_SHARED, pResultBuf->fd, 0);
+    pResultBuf->totalBufSize = pResultBuf->numOfPages * pResultBuf->pageSize;
+    pResultBuf->pBuf = mmap(NULL, FILE_SIZE_ON_DISK(pResultBuf), PROT_READ | PROT_WRITE, MAP_SHARED, pResultBuf->fd, 0);
 
-  if (pResultBuf->pBuf == MAP_FAILED) {
-    //    dError("QInfo:%p failed to map temp file: %s. %s", pQInfo, pSupporter->extBufFile, strerror(errno));
-    return TSDB_CODE_QRY_OUT_OF_MEMORY;
+    if (pResultBuf->pBuf == MAP_FAILED) {
+      //    dError("QInfo:%p failed to map temp file: %s. %s", pQInfo, pSupporter->extBufFile, strerror(errno));
+      return TSDB_CODE_QRY_OUT_OF_MEMORY;
+    }
   }
 
   return TSDB_CODE_SUCCESS;
 }
 
-static FORCE_INLINE bool noMoreAvailablePages(SDiskbasedResultBuf* pResultBuf) {
-  return (pResultBuf->allocateId == pResultBuf->numOfPages - 1);
-}
+#define NO_AVAILABLE_PAGES(_b) ((_b)->allocateId == (_b)->numOfPages - 1)
 
 static FORCE_INLINE int32_t getGroupIndex(SDiskbasedResultBuf* pResultBuf, int32_t groupId) {
   assert(pResultBuf != NULL);
@@ -121,20 +152,19 @@ static void registerPageId(SDiskbasedResultBuf* pResultBuf, int32_t groupId, int
 }
 
 tFilePage* getNewDataBuf(SDiskbasedResultBuf* pResultBuf, int32_t groupId, int32_t* pageId) {
-  if (noMoreAvailablePages(pResultBuf)) {
+  if (NO_AVAILABLE_PAGES(pResultBuf)) {
     if (extendDiskFileSize(pResultBuf, pResultBuf->incStep) != TSDB_CODE_SUCCESS) {
       return NULL;
     }
   }
 
   // register new id in this group
-  *pageId = (pResultBuf->allocateId++);
+  *pageId = (++pResultBuf->allocateId);
   registerPageId(pResultBuf, groupId, *pageId);
 
-  tFilePage* page = GET_RES_BUF_PAGE_BY_ID(pResultBuf, *pageId);
-  
   // clear memory for the new page
-  memset(page, 0, DEFAULT_INTERN_BUF_PAGE_SIZE);
+  tFilePage* page = getResBufPage(pResultBuf, *pageId);
+  memset(page, 0, pResultBuf->pageSize);
   
   return page;
 }
@@ -144,7 +174,7 @@ int32_t getNumOfRowsPerPage(SDiskbasedResultBuf* pResultBuf) { return pResultBuf
 SIDList getDataBufPagesIdList(SDiskbasedResultBuf* pResultBuf, int32_t groupId) {
   int32_t slot = getGroupIndex(pResultBuf, groupId);
   if (slot < 0) {
-    return taosArrayInit(1, sizeof(int32_t));
+    return pResultBuf->emptyDummyIdList;
   } else {
     return taosArrayGetP(pResultBuf->list, slot);
   }
@@ -156,13 +186,18 @@ void destroyResultBuf(SDiskbasedResultBuf* pResultBuf, void* handle) {
   }
 
   if (FD_VALID(pResultBuf->fd)) {
+    qDebug("QInfo:%p disk-based output buffer closed, total:%" PRId64 " bytes, file created:%s, file size:%d", handle,
+        pResultBuf->totalBufSize, pResultBuf->path, FILE_SIZE_ON_DISK(pResultBuf));
+
     close(pResultBuf->fd);
+    munmap(pResultBuf->pBuf, FILE_SIZE_ON_DISK(pResultBuf));
+    pResultBuf->pBuf = NULL;
+  } else {
+    qDebug("QInfo:%p disk-based output buffer closed, total:%" PRId64 " bytes, no file created", handle,
+           pResultBuf->totalBufSize);
   }
 
-  qDebug("QInfo:%p disk-based output buffer closed, %" PRId64 " bytes, file:%s", handle, pResultBuf->totalBufSize, pResultBuf->path);
-  munmap(pResultBuf->pBuf, pResultBuf->totalBufSize);
   unlink(pResultBuf->path);
-  
   tfree(pResultBuf->path);
 
   size_t size = taosArrayGetSize(pResultBuf->list);
@@ -172,8 +207,10 @@ void destroyResultBuf(SDiskbasedResultBuf* pResultBuf, void* handle) {
   }
 
   taosArrayDestroy(pResultBuf->list);
+  taosArrayDestroy(pResultBuf->emptyDummyIdList);
   taosHashCleanup(pResultBuf->idsTable);
-  
+
+  tfree(pResultBuf->iBuf);
   tfree(pResultBuf);
 }
 

src/query/src/qsyntaxtreefunction.c → src/query/src/qSyntaxtreefunction.c

@@ -15,7 +15,7 @@
 
 #include "os.h"
 
-#include "qsyntaxtreefunction.h"
+#include "qSyntaxtreefunction.h"
 #include "taosdef.h"
 #include "tutil.h"
 

src/query/src/qtsbuf.c → src/query/src/qTsbuf.c

-#include "qtsbuf.h"
+#include "qTsbuf.h"
+#include "taoserror.h"
 #include "tscompression.h"
 #include "tutil.h"
-#include "taoserror.h"
 
 static int32_t getDataStartOffset();
 static void TSBufUpdateVnodeInfo(STSBuf* pTSBuf, int32_t index, STSVnodeBlockInfo* pBlockInfo);

src/query/src/qUtil.c

@@ -26,12 +26,10 @@ int32_t getOutputInterResultBufSize(SQuery* pQuery) {
   int32_t size = 0;
 
   for (int32_t i = 0; i < pQuery->numOfOutput; ++i) {
-    assert(pQuery->pSelectExpr[i].interBytes <= DEFAULT_INTERN_BUF_PAGE_SIZE);
     size += pQuery->pSelectExpr[i].interBytes;
   }
 
   assert(size > 0);
-
   return size;
 }
 
@@ -43,6 +41,9 @@ int32_t initWindowResInfo(SWindowResInfo *pWindowResInfo, SQueryRuntimeEnv *pRun
   pWindowResInfo->type = type;
   _hash_fn_t fn = taosGetDefaultHashFunction(type);
   pWindowResInfo->hashList = taosHashInit(threshold, fn, false);
+  if (pWindowResInfo->hashList == NULL) {
+    return TSDB_CODE_QRY_OUT_OF_MEMORY;
+  }
   
   pWindowResInfo->curIndex = -1;
   pWindowResInfo->size     = 0;
@@ -243,7 +244,7 @@ void clearTimeWindowResBuf(SQueryRuntimeEnv *pRuntimeEnv, SWindowResult *pWindow
     size_t size = pRuntimeEnv->pQuery->pSelectExpr[i].bytes;
     memset(s, 0, size);
     
-    resetResultInfo(pResultInfo);
+    RESET_RESULT_INFO(pResultInfo);
   }
   
   pWindowRes->numOfRows = 0;

src/query/tests/CMakeLists.txt

@@ -11,5 +11,5 @@ IF (HEADER_GTEST_INCLUDE_DIR AND LIB_GTEST_STATIC_DIR)
     AUX_SOURCE_DIRECTORY(${CMAKE_CURRENT_SOURCE_DIR} SOURCE_LIST)
 
     ADD_EXECUTABLE(queryTest ${SOURCE_LIST})
-    TARGET_LINK_LIBRARIES(queryTest taos query gtest pthread)
-ENDIF()
+    TARGET_LINK_LIBRARIES(queryTest taos query gtest pthread gcov)
+ENDIF()
\ No newline at end of file

src/query/tests/astTest.cpp

@@ -3,8 +3,8 @@
 #include <cassert>
 #include <iostream>
 
+#include "qAst.h"
 #include "taosmsg.h"
-#include "qast.h"
 #include "tsdb.h"
 #include "tskiplist.h"
 

src/query/tests/histogramTest.cpp

@@ -9,7 +9,7 @@
 #include "tstoken.h"
 #include "tutil.h"
 
-#include "qhistogram.h"
+#include "qHistogram.h"
 
 /* test validate the names for table/database */
 TEST(testCase, histogram_binary_search) {

src/query/tests/resultBufferTest.cpp

@@ -2,15 +2,15 @@
 #include <cassert>
 #include <iostream>
 
+#include "qResultbuf.h"
 #include "taos.h"
-#include "qresultBuf.h"
 #include "tsdb.h"
 
 namespace {
 // simple test
 void simpleTest() {
   SDiskbasedResultBuf* pResultBuf = NULL;
-  int32_t ret = createDiskbasedResultBuffer(&pResultBuf, 1000, 64, NULL);
+  int32_t ret = createDiskbasedResultBuffer(&pResultBuf, 1000, 64, 1024, 4, NULL);
   
   int32_t pageId = 0;
   int32_t groupId = 0;
@@ -22,8 +22,7 @@ void simpleTest() {
   ASSERT_EQ(getResBufSize(pResultBuf), 1000*16384L);
   
   SIDList list = getDataBufPagesIdList(pResultBuf, groupId);
-  ASSERT_EQ(list.size, 1);
-  
+  ASSERT_EQ(taosArrayGetSize(list), 1);
   ASSERT_EQ(getNumOfResultBufGroupId(pResultBuf), 1);
   
   destroyResultBuf(pResultBuf, NULL);

src/query/tests/tsBufTest.cpp

@@ -5,10 +5,10 @@
 #include "taos.h"
 #include "tsdb.h"
 
+#include "qTsbuf.h"
 #include "tstoken.h"
 #include "ttime.h"
 #include "tutil.h"
-#include "qtsbuf.h"
 
 namespace {
 /**

src/rpc/src/rpcMain.c

@@ -55,7 +55,7 @@ typedef struct {
   char     secret[TSDB_KEY_LEN]; // secret for the link
   char     ckey[TSDB_KEY_LEN];   // ciphering key 
 
-  void   (*cfp)(SRpcMsg *, SRpcIpSet *);
+  void   (*cfp)(SRpcMsg *, SRpcEpSet *);
   int    (*afp)(char *user, char *spi, char *encrypt, char *secret, char *ckey); 
 
   int32_t   refCount;
@@ -71,7 +71,7 @@ typedef struct {
 
 typedef struct {
   SRpcInfo *pRpc;       // associated SRpcInfo
-  SRpcIpSet ipSet;      // ip list provided by app
+  SRpcEpSet epSet;      // ip list provided by app
   void     *ahandle;    // handle provided by app
   void     *signature;  // for validation
   struct SRpcConn *pConn; // pConn allocated
@@ -80,12 +80,12 @@ typedef struct {
   int32_t   contLen;    // content length
   int32_t   code;       // error code
   int16_t   numOfTry;   // number of try for different servers
-  int8_t    oldInUse;   // server IP inUse passed by app
+  int8_t    oldInUse;   // server EP inUse passed by app
   int8_t    redirect;   // flag to indicate redirect
   int8_t    connType;   // connection type
   SRpcMsg  *pRsp;       // for synchronous API
   tsem_t   *pSem;       // for synchronous API
-  SRpcIpSet *pSet;      // for synchronous API 
+  SRpcEpSet *pSet;      // for synchronous API 
   char      msg[0];     // RpcHead starts from here
 } SRpcReqContext;
 
@@ -355,7 +355,7 @@ void *rpcReallocCont(void *ptr, int contLen) {
   return start + sizeof(SRpcReqContext) + sizeof(SRpcHead);
 }
 
-void rpcSendRequest(void *shandle, const SRpcIpSet *pIpSet, SRpcMsg *pMsg) {
+void rpcSendRequest(void *shandle, const SRpcEpSet *pEpSet, SRpcMsg *pMsg) {
   SRpcInfo       *pRpc = (SRpcInfo *)shandle;
   SRpcReqContext *pContext;
 
@@ -364,11 +364,11 @@ void rpcSendRequest(void *shandle, const SRpcIpSet *pIpSet, SRpcMsg *pMsg) {
   pContext->ahandle = pMsg->ahandle;
   pContext->signature = pContext;
   pContext->pRpc = (SRpcInfo *)shandle;
-  pContext->ipSet = *pIpSet;
+  pContext->epSet = *pEpSet;
   pContext->contLen = contLen;
   pContext->pCont = pMsg->pCont;
   pContext->msgType = pMsg->msgType;
-  pContext->oldInUse = pIpSet->inUse;
+  pContext->oldInUse = pEpSet->inUse;
 
   pContext->connType = RPC_CONN_UDPC; 
   if (contLen > tsRpcMaxUdpSize) pContext->connType = RPC_CONN_TCPC;
@@ -446,7 +446,10 @@ void rpcSendResponse(const SRpcMsg *pRsp) {
   // set the idle timer to monitor the activity
   taosTmrReset(rpcProcessIdleTimer, pRpc->idleTime, pConn, pRpc->tmrCtrl, &pConn->pIdleTimer);
   rpcSendMsgToPeer(pConn, msg, msgLen);
-  pConn->secured = 1; // connection shall be secured
+
+  // if not set to secured, set it expcet NOT_READY case, since client wont treat it as secured
+  if (pConn->secured == 0 && pMsg->code != TSDB_CODE_RPC_NOT_READY)
+    pConn->secured = 1; // connection shall be secured
 
   if (pConn->pReqMsg) rpcFreeCont(pConn->pReqMsg);
   pConn->pReqMsg = NULL;
@@ -458,15 +461,15 @@ void rpcSendResponse(const SRpcMsg *pRsp) {
   return;
 }
 
-void rpcSendRedirectRsp(void *thandle, const SRpcIpSet *pIpSet) {
+void rpcSendRedirectRsp(void *thandle, const SRpcEpSet *pEpSet) {
   SRpcMsg  rpcMsg; 
   memset(&rpcMsg, 0, sizeof(rpcMsg));
   
-  rpcMsg.contLen = sizeof(SRpcIpSet);
+  rpcMsg.contLen = sizeof(SRpcEpSet);
   rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen);
   if (rpcMsg.pCont == NULL) return;
 
-  memcpy(rpcMsg.pCont, pIpSet, sizeof(SRpcIpSet));
+  memcpy(rpcMsg.pCont, pEpSet, sizeof(SRpcEpSet));
 
   rpcMsg.code = TSDB_CODE_RPC_REDIRECT;
   rpcMsg.handle = thandle;
@@ -488,7 +491,7 @@ int rpcGetConnInfo(void *thandle, SRpcConnInfo *pInfo) {
   return 0;
 }
 
-void rpcSendRecv(void *shandle, SRpcIpSet *pIpSet, SRpcMsg *pMsg, SRpcMsg *pRsp) {
+void rpcSendRecv(void *shandle, SRpcEpSet *pEpSet, SRpcMsg *pMsg, SRpcMsg *pRsp) {
   SRpcReqContext *pContext;
   pContext = (SRpcReqContext *) (pMsg->pCont-sizeof(SRpcHead)-sizeof(SRpcReqContext));
 
@@ -498,9 +501,9 @@ void rpcSendRecv(void *shandle, SRpcIpSet *pIpSet, SRpcMsg *pMsg, SRpcMsg *pRsp)
   tsem_init(&sem, 0, 0);
   pContext->pSem = &sem;
   pContext->pRsp = pRsp;
-  pContext->pSet = pIpSet;
+  pContext->pSet = pEpSet;
 
-  rpcSendRequest(shandle, pIpSet, pMsg);
+  rpcSendRequest(shandle, pEpSet, pMsg);
 
   tsem_wait(&sem);
   tsem_destroy(&sem);
@@ -657,7 +660,7 @@ static SRpcConn *rpcAllocateClientConn(SRpcInfo *pRpc) {
     pConn->spi = pRpc->spi;
     pConn->encrypt = pRpc->encrypt;
     if (pConn->spi) memcpy(pConn->secret, pRpc->secret, TSDB_KEY_LEN);
-    tDebug("%s %p client connection is allocated", pRpc->label, pConn);
+    tDebug("%s %p client connection is allocated, uid:0x%x", pRpc->label, pConn, pConn->linkUid);
   }
 
   return pConn;
@@ -718,7 +721,7 @@ static SRpcConn *rpcAllocateServerConn(SRpcInfo *pRpc, SRecvInfo *pRecv) {
     }
   
     taosHashPut(pRpc->hash, hashstr, size, (char *)&pConn, POINTER_BYTES);
-    tDebug("%s %p server connection is allocated", pRpc->label, pConn);
+    tDebug("%s %p server connection is allocated, uid:0x%x", pRpc->label, pConn, pConn->linkUid);
   }
 
   return pConn;
@@ -755,11 +758,11 @@ static SRpcConn *rpcGetConnObj(SRpcInfo *pRpc, int sid, SRecvInfo *pRecv) {
 static SRpcConn *rpcSetupConnToServer(SRpcReqContext *pContext) {
   SRpcConn   *pConn;
   SRpcInfo   *pRpc = pContext->pRpc;
-  SRpcIpSet  *pIpSet = &pContext->ipSet;
+  SRpcEpSet  *pEpSet = &pContext->epSet;
 
-  pConn = rpcGetConnFromCache(pRpc->pCache, pIpSet->fqdn[pIpSet->inUse], pIpSet->port[pIpSet->inUse], pContext->connType);
+  pConn = rpcGetConnFromCache(pRpc->pCache, pEpSet->fqdn[pEpSet->inUse], pEpSet->port[pEpSet->inUse], pContext->connType);
   if ( pConn == NULL || pConn->user[0] == 0) {
-    pConn = rpcOpenConn(pRpc, pIpSet->fqdn[pIpSet->inUse], pIpSet->port[pIpSet->inUse], pContext->connType);
+    pConn = rpcOpenConn(pRpc, pEpSet->fqdn[pEpSet->inUse], pEpSet->port[pEpSet->inUse], pContext->connType);
   } 
 
   if (pConn) {
@@ -845,6 +848,16 @@ static int rpcProcessRspHead(SRpcConn *pConn, SRpcHead *pHead) {
     return TSDB_CODE_RPC_ALREADY_PROCESSED;
   }
 
+  if (pHead->code == TSDB_CODE_RPC_MISMATCHED_LINK_ID) {
+    tDebug("%s, mismatched linkUid, link shall be restarted", pConn->info);
+    pConn->secured = 0;
+    ((SRpcHead *)pConn->pReqMsg)->destId = 0;
+    rpcSendMsgToPeer(pConn, pConn->pReqMsg, pConn->reqMsgLen);      
+    if (pConn->connType != RPC_CONN_TCPC)
+      pConn->pTimer = taosTmrStart(rpcProcessRetryTimer, tsRpcTimer, pConn, pRpc->tmrCtrl);
+    return TSDB_CODE_RPC_ALREADY_PROCESSED;
+  }
+
   if (pHead->code == TSDB_CODE_RPC_ACTION_IN_PROGRESS) {
     if (pConn->tretry <= tsRpcMaxRetry) {
       tDebug("%s, peer is still processing the transaction, retry:%d", pConn->info, pConn->tretry);
@@ -1020,16 +1033,16 @@ static void rpcNotifyClient(SRpcReqContext *pContext, SRpcMsg *pMsg) {
   pContext->pConn = NULL;
   if (pContext->pRsp) { 
     // for synchronous API
-    memcpy(pContext->pSet, &pContext->ipSet, sizeof(SRpcIpSet));
+    memcpy(pContext->pSet, &pContext->epSet, sizeof(SRpcEpSet));
     memcpy(pContext->pRsp, pMsg, sizeof(SRpcMsg));
     tsem_post(pContext->pSem);
   } else {
     // for asynchronous API 
-    SRpcIpSet *pIpSet = NULL;
-    if (pContext->ipSet.inUse != pContext->oldInUse || pContext->redirect) 
-      pIpSet = &pContext->ipSet;  
+    SRpcEpSet *pEpSet = NULL;
+    if (pContext->epSet.inUse != pContext->oldInUse || pContext->redirect) 
+      pEpSet = &pContext->epSet;  
 
-    (*pRpc->cfp)(pMsg, pIpSet);  
+    (*pRpc->cfp)(pMsg, pEpSet);  
   }
 
   // free the request message
@@ -1070,9 +1083,9 @@ static void rpcProcessIncomingMsg(SRpcConn *pConn, SRpcHead *pHead) {
     pConn->pContext = NULL;
     pConn->pReqMsg = NULL;
 
-    // for UDP, port may be changed by server, the port in ipSet shall be used for cache
+    // for UDP, port may be changed by server, the port in epSet shall be used for cache
     if (pHead->code != TSDB_CODE_RPC_TOO_SLOW) {
-      rpcAddConnIntoCache(pRpc->pCache, pConn, pConn->peerFqdn, pContext->ipSet.port[pContext->ipSet.inUse], pConn->connType);    
+      rpcAddConnIntoCache(pRpc->pCache, pConn, pConn->peerFqdn, pContext->epSet.port[pContext->epSet.inUse], pConn->connType);    
     } else {
       rpcCloseConn(pConn);
     }
@@ -1087,10 +1100,10 @@ static void rpcProcessIncomingMsg(SRpcConn *pConn, SRpcHead *pHead) {
 
     if (pHead->code == TSDB_CODE_RPC_REDIRECT) {
       pContext->numOfTry = 0;
-      memcpy(&pContext->ipSet, pHead->content, sizeof(pContext->ipSet));
-      tDebug("%s, redirect is received, numOfIps:%d", pConn->info, pContext->ipSet.numOfIps);
-      for (int i=0; i<pContext->ipSet.numOfIps; ++i) 
-        pContext->ipSet.port[i] = htons(pContext->ipSet.port[i]);
+      memcpy(&pContext->epSet, pHead->content, sizeof(pContext->epSet));
+      tDebug("%s, redirect is received, numOfEps:%d", pConn->info, pContext->epSet.numOfEps);
+      for (int i=0; i<pContext->epSet.numOfEps; ++i) 
+        pContext->epSet.port[i] = htons(pContext->epSet.port[i]);
       rpcSendReqToServer(pRpc, pContext);
       rpcFreeCont(rpcMsg.pCont);
     } else if (pHead->code == TSDB_CODE_RPC_NOT_READY) {
@@ -1269,7 +1282,7 @@ static void rpcProcessConnError(void *param, void *id) {
   
   tDebug("%s %p, connection error happens", pRpc->label, pContext->ahandle);
 
-  if (pContext->numOfTry >= pContext->ipSet.numOfIps) {
+  if (pContext->numOfTry >= pContext->epSet.numOfEps) {
     rpcMsg.msgType = pContext->msgType+1;
     rpcMsg.ahandle = pContext->ahandle;
     rpcMsg.code = pContext->code;
@@ -1279,8 +1292,8 @@ static void rpcProcessConnError(void *param, void *id) {
     rpcNotifyClient(pContext, &rpcMsg);
   } else {
     // move to next IP 
-    pContext->ipSet.inUse++;
-    pContext->ipSet.inUse = pContext->ipSet.inUse % pContext->ipSet.numOfIps;
+    pContext->epSet.inUse++;
+    pContext->epSet.inUse = pContext->epSet.inUse % pContext->epSet.numOfEps;
     rpcSendReqToServer(pRpc, pContext);
   }
 }