[td-11818] merge 3.0

include/common/tmsg.h

@@ -71,6 +71,9 @@ typedef enum _mgmt_table {
   TSDB_MGMT_TABLE_TABLE,
   TSDB_MGMT_TABLE_DNODE,
   TSDB_MGMT_TABLE_MNODE,
+  TSDB_MGMT_TABLE_QNODE,
+  TSDB_MGMT_TABLE_SNODE,
+  TSDB_MGMT_TABLE_BNODE,
   TSDB_MGMT_TABLE_VGROUP,
   TSDB_MGMT_TABLE_STB,
   TSDB_MGMT_TABLE_MODULE,
@@ -219,6 +222,26 @@ typedef struct {
   char     data[];
 } SMDCreateTableMsg;
 
+// typedef struct {
+//  int32_t len;  // one create table message
+//  char    tableName[TSDB_TABLE_FNAME_LEN];
+//  int16_t numOfColumns;
+//  int16_t sqlLen;  // the length of SQL, it starts after schema , sql is a null-terminated string
+//  int8_t  igExists;
+//  int8_t  rspMeta;
+//  int8_t  reserved[16];
+//  char    schema[];
+//} SCreateTableMsg;
+
+typedef struct {
+  char    tableName[TSDB_TABLE_FNAME_LEN];
+  int16_t numOfColumns;
+  int16_t numOfTags;
+  int8_t  igExists;
+  int8_t  rspMeta;
+  char    schema[];
+} SCreateCTableMsg;
+
 typedef struct {
   char    name[TSDB_TABLE_FNAME_LEN];
   int8_t  igExists;
@@ -242,19 +265,7 @@ typedef struct {
   SMsgHead head;
   char     name[TSDB_TABLE_FNAME_LEN];
   uint64_t suid;
-  int32_t  sverson;
-  uint32_t ttl;
-  uint32_t keep;
-  int32_t  numOfTags;
-  int32_t  numOfColumns;
-  SSchema  pSchema[];
-} SCreateStbInternalMsg;
-
-typedef struct {
-  SMsgHead head;
-  char     name[TSDB_TABLE_FNAME_LEN];
-  uint64_t suid;
-} SDropStbInternalMsg;
+} SVDropStbReq;
 
 typedef struct {
   SMsgHead head;
@@ -299,6 +310,18 @@ typedef struct SEpSet {
   char     fqdn[TSDB_MAX_REPLICA][TSDB_FQDN_LEN];
 } SEpSet;
 
+static FORCE_INLINE int taosEncodeSEpSet(void** buf, const SEpSet* pEp) {
+  if(buf == NULL) return sizeof(SEpSet);
+  memcpy(buf, pEp, sizeof(SEpSet));
+  //TODO: endian conversion
+  return sizeof(SEpSet);
+}
+
+static FORCE_INLINE void* taosDecodeSEpSet(void* buf, SEpSet* pEpSet) {
+  memcpy(pEpSet, buf, sizeof(SEpSet));
+  return buf;
+}
+
 typedef struct {
   int32_t acctId;
   int64_t clusterId;
@@ -642,8 +665,6 @@ typedef struct {
 typedef struct {
   int32_t dnodeId;
   int64_t clusterId;
-  int8_t  dropped;
-  char    reserved[7];
 } SDnodeCfg;
 
 typedef struct {
@@ -822,29 +843,25 @@ typedef struct {
 
 typedef struct {
   int32_t dnodeId;
-} SCreateMnodeMsg, SDropMnodeMsg;
+} SMCreateMnodeMsg, SMDropMnodeMsg, SDDropMnodeMsg;
 
 typedef struct {
   int32_t  dnodeId;
   int8_t   replica;
   SReplica replicas[TSDB_MAX_REPLICA];
-} SCreateMnodeInMsg, SAlterMnodeInMsg;
+} SDCreateMnodeMsg, SDAlterMnodeMsg;
 
 typedef struct {
   int32_t dnodeId;
-} SDropMnodeInMsg;
+} SMCreateQnodeMsg, SMDropQnodeMsg, SDCreateQnodeMsg, SDDropQnodeMsg;
 
 typedef struct {
   int32_t dnodeId;
-} SCreateQnodeInMsg, SDropQnodeInMsg;
+} SMCreateSnodeMsg, SMDropSnodeMsg, SDCreateSnodeMsg, SDDropSnodeMsg;
 
 typedef struct {
   int32_t dnodeId;
-} SCreateSnodeInMsg, SDropSnodeInMsg;
-
-typedef struct {
-  int32_t dnodeId;
-} SCreateBnodeInMsg, SDropBnodeInMsg;
+} SMCreateBnodeMsg, SMDropBnodeMsg, SDCreateBnodeMsg, SDDropBnodeMsg;
 
 typedef struct {
   int32_t dnodeId;
@@ -1060,24 +1077,92 @@ typedef struct STaskDropRsp {
 typedef struct {
   int8_t  igExists;
   char*   name;
-  char*  phyPlan;
+  char*   physicalPlan;
+  char*   logicalPlan;
 } SCMCreateTopicReq;
 
 static FORCE_INLINE int tSerializeSCMCreateTopicReq(void** buf, const SCMCreateTopicReq* pReq) {
   int tlen = 0;
   tlen += taosEncodeString(buf, pReq->name);
   tlen += taosEncodeFixedI8(buf, pReq->igExists);
-  tlen += taosEncodeString(buf, pReq->phyPlan);
+  tlen += taosEncodeString(buf, pReq->physicalPlan);
+  tlen += taosEncodeString(buf, pReq->logicalPlan);
   return tlen;
 }
 
 static FORCE_INLINE void* tDeserializeSCMCreateTopicReq(void* buf, SCMCreateTopicReq* pReq) {
   buf = taosDecodeFixedI8(buf, &(pReq->igExists));
   buf = taosDecodeString(buf, &(pReq->name));
-  buf = taosDecodeString(buf, &(pReq->phyPlan));
+  buf = taosDecodeString(buf, &(pReq->physicalPlan));
+  buf = taosDecodeString(buf, &(pReq->logicalPlan));
   return buf;
 }
 
+typedef struct {
+  int64_t topicId;
+} SCMCreateTopicRsp;
+
+static FORCE_INLINE int tSerializeSCMCreateTopicRsp(void** buf, const SCMCreateTopicRsp* pRsp) {
+  int tlen = 0;
+  tlen += taosEncodeFixedI64(buf, pRsp->topicId);
+  return tlen;
+}
+
+static FORCE_INLINE void* tDeserializeSCMCreateTopicRsp(void* buf, SCMCreateTopicRsp* pRsp) {
+  buf = taosDecodeFixedI64(buf, &pRsp->topicId);
+  return buf;
+}
+
+typedef struct {
+  char* topicName;
+  char* consumerGroup;
+  int64_t consumerId;
+} SCMSubscribeReq;
+
+static FORCE_INLINE int tSerializeSCMSubscribeReq(void** buf, const SCMSubscribeReq* pReq) {
+  int tlen = 0;
+  tlen += taosEncodeString(buf, pReq->topicName);
+  tlen += taosEncodeString(buf, pReq->consumerGroup);
+  tlen += taosEncodeFixedI64(buf, pReq->consumerId);
+  return tlen;
+}
+
+static FORCE_INLINE void* tDeserializeSCMSubscribeReq(void* buf, SCMSubscribeReq* pReq) {
+  buf = taosDecodeString(buf, &pReq->topicName);
+  buf = taosDecodeString(buf, &pReq->consumerGroup);
+  buf = taosDecodeFixedI64(buf, &pReq->consumerId);
+  return buf;
+}
+
+typedef struct {
+  int32_t vgId;
+  SEpSet pEpSet;
+} SCMSubscribeRsp;
+
+static FORCE_INLINE int tSerializeSCMSubscribeRsp(void** buf, const SCMSubscribeRsp* pRsp) {
+  int tlen = 0;
+  tlen += taosEncodeFixedI32(buf, pRsp->vgId);
+  tlen += taosEncodeSEpSet(buf, &pRsp->pEpSet);
+  return tlen;
+}
+
+static FORCE_INLINE void* tDeserializeSCMSubscribeRsp(void* buf, SCMSubscribeRsp* pRsp) {
+  buf = taosDecodeFixedI32(buf, &pRsp->vgId);
+  buf = taosDecodeSEpSet(buf, &pRsp->pEpSet);
+  return buf;
+}
+
+typedef struct {
+  int64_t topicId;
+  int64_t consumerId;
+  int64_t consumerGroupId;
+  int64_t offset;
+} SMVSubscribeReq;
+
+typedef struct {
+  int64_t newOffset;
+} SMVSubscribeRsp;
+
 typedef struct {
   char    name[TSDB_TOPIC_FNAME_LEN];
   int8_t  igExists;
@@ -1107,13 +1192,13 @@ typedef struct {
   char*    executor;
   int32_t  sqlLen;
   char*    sql;
-} SCreateTopicInternalMsg;
+} SDCreateTopicMsg;
 
 typedef struct {
   SMsgHead head;
   char     name[TSDB_TABLE_FNAME_LEN];
   uint64_t tuid;
-} SDropTopicInternalMsg;
+} SDDropTopicMsg;
 
 typedef struct SVCreateTbReq {
   uint64_t ver;  // use a general definition

include/common/tmsgdef.h

@@ -70,6 +70,15 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_DND_CREATE_MNODE, "dnode-create-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_ALTER_MNODE, "dnode-alter-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_DROP_MNODE, "dnode-drop-mnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_CREATE_QNODE, "dnode-create-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_ALTER_QNODE, "dnode-alter-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_DROP_QNODE, "dnode-drop-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_CREATE_SNODE, "dnode-create-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_ALTER_SNODE, "dnode-alter-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_DROP_SNODE, "dnode-drop-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_CREATE_BNODE, "dnode-create-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_ALTER_BNODE, "dnode-alter-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_DROP_BNODE, "dnode-drop-bnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_CREATE_VNODE, "dnode-create-vnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_ALTER_VNODE, "dnode-alter-vnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_DROP_VNODE, "dnode-drop-vnode", NULL, NULL)
@@ -90,9 +99,20 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_USER, "mnode-drop-user", NULL, NULL) 
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_DNODE, "mnode-create-dnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CONFIG_DNODE, "mnode-config-dnode", NULL, NULL) 
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_DNODE, "mnode-alter-dnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_DNODE, "mnode-drop-dnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_MNODE, "mnode-create-mnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_MNODE, "mnode-alter-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_MNODE, "mnode-drop-mnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_QNODE, "mnode-create-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_QNODE, "mnode-alter-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_DROP_QNODE, "mnode-drop-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_SNODE, "mnode-create-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_SNODE, "mnode-alter-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_DROP_SNODE, "mnode-drop-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_BNODE, "mnode-create-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_BNODE, "mnode-alter-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_DROP_BNODE, "mnode-drop-bnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_DB, "mnode-create-db", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_DB, "mnode-drop-db", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_USE_DB, "mnode-use-db", NULL, NULL)
@@ -116,9 +136,10 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_TRANS, "mnode-trans", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_GRANT, "mnode-grant", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_AUTH, "mnode-auth", NULL, NULL)
-  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_TOPIC, "mnode-create-topic", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_TOPIC, "mnode-create-topic", SCMCreateTopicReq, SCMCreateTopicRsp)
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_TOPIC, "mnode-alter-topic", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_TOPIC, "mnode-drop-topic", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_SUBSCRIBE, "mnode-subscribe", SCMSubscribeReq, SCMSubscribeRsp)
 
   // Requests handled by VNODE
   TD_NEW_MSG_SEG(TDMT_VND_MSG)
@@ -149,6 +170,9 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_SHOW_TABLES, "vnode-show-tables", SVShowTablesReq, SVShowTablesRsp)
   TD_DEF_MSG_TYPE(TDMT_VND_SHOW_TABLES_FETCH, "vnode-show-tables-fetch", SVShowTablesFetchReq, SVShowTablesFetchRsp)
 
+  TD_DEF_MSG_TYPE(TDMT_VND_SUBSCRIBE, "vnode-subscribe", SMVSubscribeReq, SMVSubscribeRsp)
+
+
   // Requests handled by QNODE
   TD_NEW_MSG_SEG(TDMT_QND_MSG)
 

include/dnode/mnode/sdb/sdb.h

@@ -25,7 +25,7 @@ extern "C" {
 #define SDB_GET_INT64(pData, pRow, dataPos, val)   \
   {                                                \
     if (sdbGetRawInt64(pRaw, dataPos, val) != 0) { \
-      sdbFreeRow(pRow);                            \
+      tfree(pRow);                                 \
       return NULL;                                 \
     }                                              \
     dataPos += sizeof(int64_t);                    \
@@ -34,7 +34,7 @@ extern "C" {
 #define SDB_GET_INT32(pData, pRow, dataPos, val)   \
   {                                                \
     if (sdbGetRawInt32(pRaw, dataPos, val) != 0) { \
-      sdbFreeRow(pRow);                            \
+      tfree(pRow);                                 \
       return NULL;                                 \
     }                                              \
     dataPos += sizeof(int32_t);                    \
@@ -43,7 +43,7 @@ extern "C" {
 #define SDB_GET_INT16(pData, pRow, dataPos, val)   \
   {                                                \
     if (sdbGetRawInt16(pRaw, dataPos, val) != 0) { \
-      sdbFreeRow(pRow);                            \
+      tfree(pRow);                                 \
       return NULL;                                 \
     }                                              \
     dataPos += sizeof(int16_t);                    \
@@ -52,7 +52,7 @@ extern "C" {
 #define SDB_GET_INT8(pData, pRow, dataPos, val)   \
   {                                               \
     if (sdbGetRawInt8(pRaw, dataPos, val) != 0) { \
-      sdbFreeRow(pRow);                           \
+      tfree(pRow);                                \
       return NULL;                                \
     }                                             \
     dataPos += sizeof(int8_t);                    \
@@ -61,7 +61,7 @@ extern "C" {
 #define SDB_GET_BINARY(pRaw, pRow, dataPos, val, valLen)    \
   {                                                         \
     if (sdbGetRawBinary(pRaw, dataPos, val, valLen) != 0) { \
-      sdbFreeRow(pRow);                                     \
+      tfree(pRow);                                          \
       return NULL;                                          \
     }                                                       \
     dataPos += valLen;                                      \
@@ -71,7 +71,7 @@ extern "C" {
   {                                                         \
     char val[valLen] = {0};                                 \
     if (sdbGetRawBinary(pRaw, dataPos, val, valLen) != 0) { \
-      sdbFreeRow(pRow);                                     \
+      tfree(pRow);                                          \
       return NULL;                                          \
     }                                                       \
     dataPos += valLen;                                      \
@@ -157,16 +157,21 @@ typedef enum {
   SDB_TRANS = 1,
   SDB_CLUSTER = 2,
   SDB_MNODE = 3,
-  SDB_DNODE = 4,
-  SDB_USER = 5,
-  SDB_AUTH = 6,
-  SDB_ACCT = 7,
-  SDB_TOPIC = 8,
-  SDB_VGROUP = 9,
-  SDB_STB = 10,
-  SDB_DB = 11,
-  SDB_FUNC = 12,
-  SDB_MAX = 13
+  SDB_QNODE = 4,
+  SDB_SNODE = 5,
+  SDB_BNODE = 6,
+  SDB_DNODE = 7,
+  SDB_USER = 8,
+  SDB_AUTH = 9,
+  SDB_ACCT = 10,
+  SDB_CONSUMER = 11,
+  SDB_CGROUP = 12,
+  SDB_TOPIC = 13,
+  SDB_VGROUP = 14,
+  SDB_STB = 15,
+  SDB_DB = 16,
+  SDB_FUNC = 17,
+  SDB_MAX = 18
 } ESdbType;
 
 typedef struct SSdb SSdb;
@@ -325,7 +330,7 @@ int32_t  sdbGetRawSoftVer(SSdbRaw *pRaw, int8_t *sver);
 int32_t  sdbGetRawTotalSize(SSdbRaw *pRaw);
 
 SSdbRow *sdbAllocRow(int32_t objSize);
-void     sdbFreeRow(SSdbRow *pRow);
+void     sdbFreeRow(SSdb *pSdb, SSdbRow *pRow);
 void    *sdbGetRowObj(SSdbRow *pRow);
 
 #ifdef __cplusplus

include/dnode/vnode/vnode.h

@@ -162,16 +162,6 @@ int vnodeProcessQueryReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
  */
 int vnodeProcessFetchReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
 
-/**
- * @brief Process a consume message.
- *
- * @param pVnode The vnode object.
- * @param pMsg The request message
- * @param pRsp The response message
- * @return int 0 for success, -1 for failure
- */
-int vnodeProcessConsumeReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
-
 /* ------------------------ SVnodeCfg ------------------------ */
 /**
  * @brief Initialize VNODE options.

include/libs/catalog/catalog.h

@@ -62,6 +62,18 @@ int32_t catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle)
 
 int32_t catalogGetDBVgroupVersion(struct SCatalog* pCatalog, const char* dbName, int32_t* version);
 
+/**
+ * Get a DB's all vgroup info.
+ * @param pCatalog (input, got with catalogGetHandle)
+ * @param pRpc (input, rpc object)
+ * @param pMgmtEps (input, mnode EPs)
+ * @param pDBName (input, full db name)
+ * @param forceUpdate (input, force update db vgroup info from mnode) 
+ * @param pVgroupList (output, vgroup info list, element is SVgroupInfo, NEED to simply free the array by caller)
+ * @return error code
+ */
+int32_t catalogGetDBVgroup(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, int32_t forceUpdate, SArray** pVgroupList);
+
 int32_t catalogUpdateDBVgroup(struct SCatalog* pCatalog, const char* dbName, SDBVgroupInfo* dbInfo);
 
 /**

include/libs/qcom/query.h

@@ -75,6 +75,7 @@ typedef struct STableMeta {
 } STableMeta;
 
 typedef struct SDBVgroupInfo {
+  int32_t   lock;
   int32_t   vgVersion;  
   int8_t    hashMethod;
   SHashObj *vgInfo;  //key:vgId, value:SVgroupInfo

include/util/taoserror.h

@@ -172,10 +172,15 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_INVALID_DNODE_EP          TAOS_DEF_ERROR_CODE(0, 0x0347)
 #define TSDB_CODE_MND_INVALID_DNODE_ID          TAOS_DEF_ERROR_CODE(0, 0x0348)
 
-// mnode-mnode
+// mnode-node
 #define TSDB_CODE_MND_MNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0350)
 #define TSDB_CODE_MND_MNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0351)
-#define TSDB_CODE_MND_TOO_MANY_MNODES           TAOS_DEF_ERROR_CODE(0, 0x0352)
+#define TSDB_CODE_MND_QNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0352)
+#define TSDB_CODE_MND_QNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0353)
+#define TSDB_CODE_MND_SNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0354)
+#define TSDB_CODE_MND_SNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0355)
+#define TSDB_CODE_MND_BNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0356)
+#define TSDB_CODE_MND_BNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0357)
 
 // mnode-acct
 #define TSDB_CODE_MND_ACCT_ALREADY_EXIST        TAOS_DEF_ERROR_CODE(0, 0x0360)

include/util/thash.h

@@ -144,6 +144,16 @@ void *taosHashGet(SHashObj *pHashObj, const void *key, size_t keyLen);
  */
 void *taosHashGetClone(SHashObj *pHashObj, const void *key, size_t keyLen, void* destBuf);
 
+/**
+ * Clone the result to interval allocated buffer
+ * @param pHashObj
+ * @param key
+ * @param keyLen
+ * @param destBuf
+ * @return
+ */
+void* taosHashGetCloneExt(SHashObj *pHashObj, const void *key, size_t keyLen, void (*fp)(void *), void** d, size_t *sz);
+
 /**
  * remove item with the specified key
  * @param pHashObj
@@ -200,6 +210,26 @@ void  taosHashCancelIterate(SHashObj *pHashObj, void *p);
  */
 int32_t taosHashGetKey(void *data, void** key, size_t* keyLen);
 
+/**
+ * return the payload data with the specified key(reference number added)
+ *
+ * @param pHashObj
+ * @param key
+ * @param keyLen
+ * @return
+ */
+void* taosHashAcquire(SHashObj *pHashObj, const void *key, size_t keyLen);
+
+/**
+ * release the prevous acquired obj
+ *
+ * @param pHashObj
+ * @param data
+ * @return
+ */
+void taosHashRelease(SHashObj *pHashObj, void *p);
+
+
 #ifdef __cplusplus
 }
 #endif

source/client/src/clientImpl.c

@@ -210,6 +210,9 @@ TAOS_RES *tmq_create_topic(TAOS* taos, const char* name, const char* sql, int sq
   terrno = TSDB_CODE_SUCCESS;
 
   CHECK_CODE_GOTO(buildRequest(pTscObj, sql, sqlLen, &pRequest), _return);
+
+//temporary disabled until planner ready
+#if 0
   CHECK_CODE_GOTO(parseSql(pRequest, &pQuery), _return);
   //TODO: check sql valid
 
@@ -219,15 +222,24 @@ TAOS_RES *tmq_create_topic(TAOS* taos, const char* name, const char* sql, int sq
   if(dagStr == NULL) {
     //TODO
   }
+#endif
 
   SCMCreateTopicReq req = {
     .name = (char*)name,
     .igExists = 0,
-    .phyPlan = dagStr,
+    /*.physicalPlan = dagStr,*/
+    .physicalPlan = (char*)sql,
+    .logicalPlan = "",
   };
 
-  void* buf = NULL;
-  int tlen = tSerializeSCMCreateTopicReq(&buf, &req);
+  int tlen = tSerializeSCMCreateTopicReq(NULL, &req);
+  void* buf = malloc(tlen);
+  if(buf == NULL) {
+    goto _return;
+  }
+  void* abuf = buf;
+  tSerializeSCMCreateTopicReq(&abuf, &req);
+  /*printf("formatted: %s\n", dagStr);*/
 
   pRequest->body.requestMsg = (SDataBuf){ .pData = buf, .len = tlen };
 
@@ -239,8 +251,6 @@ TAOS_RES *tmq_create_topic(TAOS* taos, const char* name, const char* sql, int sq
 
   tsem_wait(&pRequest->body.rspSem);
 
-  destroySendMsgInfo(body);
-
 _return:
   qDestroyQuery(pQuery);
   qDestroyQueryDag(pDag); 

source/client/test/clientTests.cpp

@@ -49,6 +49,7 @@ int main(int argc, char** argv) {
 
 TEST(testCase, driverInit_Test) { taos_init(); }
 
+#if 0
 TEST(testCase, connect_Test) {
   TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);
   assert(pConn != NULL);
@@ -398,6 +399,7 @@ TEST(testCase, drop_stable_Test) {
   taos_free_result(pRes);
   taos_close(pConn);
 }
+#endif
 
 //TEST(testCase, create_topic_Test) {
 //  TAOS* pConn = taos_connect("localhost", "root", "taosdata", NULL, 0);

source/dnode/mgmt/daemon/src/daemon.c

@@ -139,7 +139,7 @@ void dmnWaitSignal() {
 void dmnInitOption(SDnodeOpt *pOption) {
   pOption->sver = 30000000; //3.0.0.0
   pOption->numOfCores = tsNumOfCores;
-  pOption->numOfSupportVnodes = 1;
+  pOption->numOfSupportVnodes = 16;
   pOption->numOfCommitThreads = 1;
   pOption->statusInterval = tsStatusInterval;
   pOption->numOfThreadsPerCore = tsNumOfThreadsPerCore;

source/dnode/mgmt/impl/src/dndBnode.c

@@ -256,7 +256,7 @@ static int32_t dndDropBnode(SDnode *pDnode) {
 }
 
 int32_t dndProcessCreateBnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SCreateBnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDCreateBnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {
@@ -268,7 +268,7 @@ int32_t dndProcessCreateBnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
 }
 
 int32_t dndProcessDropBnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SDropBnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDDropBnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {

source/dnode/mgmt/impl/src/dndDnode.c

@@ -15,9 +15,12 @@
 
 #define _DEFAULT_SOURCE
 #include "dndDnode.h"
+#include "dndBnode.h"
+#include "dndMnode.h"
+#include "dndQnode.h"
+#include "dndSnode.h"
 #include "dndTransport.h"
 #include "dndVnodes.h"
-#include "dndMnode.h"
 
 static int32_t dndInitMgmtWorker(SDnode *pDnode);
 static void    dndCleanupMgmtWorker(SDnode *pDnode);
@@ -393,13 +396,11 @@ void dndSendStatusMsg(SDnode *pDnode) {
 
 static void dndUpdateDnodeCfg(SDnode *pDnode, SDnodeCfg *pCfg) {
   SDnodeMgmt *pMgmt = &pDnode->dmgmt;
-  if (pMgmt->dnodeId == 0 || pMgmt->dropped != pCfg->dropped) {
-    dInfo("set dnodeId:%d clusterId:% " PRId64 " dropped:%d", pCfg->dnodeId, pCfg->clusterId, pCfg->dropped);
-
+  if (pMgmt->dnodeId == 0) {
+    dInfo("set dnodeId:%d clusterId:% " PRId64, pCfg->dnodeId, pCfg->clusterId);
     taosWLockLatch(&pMgmt->latch);
     pMgmt->dnodeId = pCfg->dnodeId;
     pMgmt->clusterId = pCfg->clusterId;
-    pMgmt->dropped = pCfg->dropped;
     dndWriteDnodes(pDnode);
     taosWUnLockLatch(&pMgmt->latch);
   }
@@ -430,6 +431,11 @@ static void dndProcessStatusRsp(SDnode *pDnode, SRpcMsg *pMsg) {
 
   if (pMsg->code != TSDB_CODE_SUCCESS) {
     pMgmt->statusSent = 0;
+    if (pMsg->code == TSDB_CODE_MND_DNODE_NOT_EXIST && !pMgmt->dropped && pMgmt->dnodeId > 0) {
+      dInfo("dnode:%d, set to dropped since not exist in mnode", pMgmt->dnodeId);
+      pMgmt->dropped = 1;
+      dndWriteDnodes(pDnode);
+    }
     return;
   }
 
@@ -439,11 +445,6 @@ static void dndProcessStatusRsp(SDnode *pDnode, SRpcMsg *pMsg) {
   pCfg->clusterId = htobe64(pCfg->clusterId);
   dndUpdateDnodeCfg(pDnode, pCfg);
 
-  if (pCfg->dropped) {
-    pMgmt->statusSent = 0;
-    return;
-  }
-
   SDnodeEps *pDnodeEps = &pRsp->dnodeEps;
   pDnodeEps->num = htonl(pDnodeEps->num);
   for (int32_t i = 0; i < pDnodeEps->num; ++i) {
@@ -487,7 +488,7 @@ static void *dnodeThreadRoutine(void *param) {
     pthread_testcancel();
     taosMsleep(ms);
 
-    if (dndGetStat(pDnode) == DND_STAT_RUNNING && !pMgmt->statusSent) {
+    if (dndGetStat(pDnode) == DND_STAT_RUNNING && !pMgmt->statusSent && !pMgmt->dropped) {
       dndSendStatusMsg(pDnode);
     }
   }
@@ -522,6 +523,11 @@ int32_t dndInitDnode(SDnode *pDnode) {
     return -1;
   }
 
+  if (pMgmt->dropped) {
+    dError("dnode will not start for its already dropped");
+    return -1;
+  }
+
   if (dndInitMgmtWorker(pDnode) != 0) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
@@ -645,6 +651,24 @@ static void dndProcessMgmtQueue(SDnode *pDnode, SRpcMsg *pMsg) {
     case TDMT_DND_DROP_MNODE:
       code = dndProcessDropMnodeReq(pDnode, pMsg);
       break;
+    case TDMT_DND_CREATE_QNODE:
+      code = dndProcessCreateQnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_DROP_QNODE:
+      code = dndProcessDropQnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_CREATE_SNODE:
+      code = dndProcessCreateSnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_DROP_SNODE:
+      code = dndProcessDropSnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_CREATE_BNODE:
+      code = dndProcessCreateBnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_DROP_BNODE:
+      code = dndProcessDropBnodeReq(pDnode, pMsg);
+      break;
     case TDMT_DND_CONFIG_DNODE:
       code = dndProcessConfigDnodeReq(pDnode, pMsg);
       break;

source/dnode/mgmt/impl/src/dndMnode.c

@@ -299,7 +299,7 @@ static void dndBuildMnodeOpenOption(SDnode *pDnode, SMnodeOpt *pOption) {
   memcpy(&pOption->replicas, pMgmt->replicas, sizeof(SReplica) * TSDB_MAX_REPLICA);
 }
 
-static int32_t dndBuildMnodeOptionFromMsg(SDnode *pDnode, SMnodeOpt *pOption, SCreateMnodeInMsg *pMsg) {
+static int32_t dndBuildMnodeOptionFromMsg(SDnode *pDnode, SMnodeOpt *pOption, SDCreateMnodeMsg *pMsg) {
   dndInitMnodeOption(pDnode, pOption);
   pOption->dnodeId = dndGetDnodeId(pDnode);
   pOption->clusterId = dndGetClusterId(pDnode);
@@ -417,8 +417,8 @@ static int32_t dndDropMnode(SDnode *pDnode) {
   return 0;
 }
 
-static SCreateMnodeInMsg *dndParseCreateMnodeMsg(SRpcMsg *pRpcMsg) {
-  SCreateMnodeInMsg *pMsg = pRpcMsg->pCont;
+static SDCreateMnodeMsg *dndParseCreateMnodeMsg(SRpcMsg *pRpcMsg) {
+  SDCreateMnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
   for (int32_t i = 0; i < pMsg->replica; ++i) {
     pMsg->replicas[i].id = htonl(pMsg->replicas[i].id);
@@ -429,7 +429,7 @@ static SCreateMnodeInMsg *dndParseCreateMnodeMsg(SRpcMsg *pRpcMsg) {
 }
 
 int32_t dndProcessCreateMnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SCreateMnodeInMsg *pMsg = dndParseCreateMnodeMsg(pRpcMsg);
+  SDCreateMnodeMsg *pMsg = dndParseCreateMnodeMsg(pRpcMsg);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {
     terrno = TSDB_CODE_DND_MNODE_ID_INVALID;
@@ -445,7 +445,7 @@ int32_t dndProcessCreateMnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
 }
 
 int32_t dndProcessAlterMnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SAlterMnodeInMsg *pMsg = dndParseCreateMnodeMsg(pRpcMsg);
+  SDAlterMnodeMsg *pMsg = dndParseCreateMnodeMsg(pRpcMsg);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {
     terrno = TSDB_CODE_DND_MNODE_ID_INVALID;
@@ -465,7 +465,7 @@ int32_t dndProcessAlterMnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
 }
 
 int32_t dndProcessDropMnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SDropMnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDDropMnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {

source/dnode/mgmt/impl/src/dndQnode.c

@@ -261,7 +261,7 @@ static int32_t dndDropQnode(SDnode *pDnode) {
 }
 
 int32_t dndProcessCreateQnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SCreateQnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDCreateQnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {
@@ -273,7 +273,7 @@ int32_t dndProcessCreateQnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
 }
 
 int32_t dndProcessDropQnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SDropQnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDDropQnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {

source/dnode/mgmt/impl/src/dndSnode.c

@@ -256,7 +256,7 @@ static int32_t dndDropSnode(SDnode *pDnode) {
 }
 
 int32_t dndProcessCreateSnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SCreateSnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDCreateSnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {
@@ -268,7 +268,7 @@ int32_t dndProcessCreateSnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
 }
 
 int32_t dndProcessDropSnodeReq(SDnode *pDnode, SRpcMsg *pRpcMsg) {
-  SDropSnodeInMsg *pMsg = pRpcMsg->pCont;
+  SDDropSnodeMsg *pMsg = pRpcMsg->pCont;
   pMsg->dnodeId = htonl(pMsg->dnodeId);
 
   if (pMsg->dnodeId != dndGetDnodeId(pDnode)) {

source/dnode/mgmt/impl/src/dndTransport.c

@@ -37,6 +37,18 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_ALTER_MNODE_RSP)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_MNODE)] = dndProcessMgmtMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_MNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_QNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_QNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_QNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_QNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_SNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_SNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_SNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_SNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_BNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_BNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_BNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_BNODE_RSP)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_VNODE)] = dndProcessMgmtMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_VNODE_RSP)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_ALTER_VNODE)] = dndProcessMgmtMsg;
@@ -66,6 +78,12 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_DNODE)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_MNODE)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_MNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_QNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_QNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_SNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_SNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_BNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_BNODE)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_DB)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_DB)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_USE_DB)] = dndProcessMnodeWriteMsg;
@@ -367,6 +385,7 @@ void dndCleanupTrans(SDnode *pDnode) {
 
 void dndSendMsgToDnode(SDnode *pDnode, SEpSet *pEpSet, SRpcMsg *pMsg) {
   STransMgmt *pMgmt = &pDnode->tmgmt;
+  if (pMgmt->clientRpc == NULL) return;
   rpcSendRequest(pMgmt->clientRpc, pEpSet, pMsg, NULL);
 }
 

source/dnode/mgmt/impl/test/CMakeLists.txt

@@ -8,6 +8,9 @@ add_subdirectory(db)
 add_subdirectory(dnode)
 # add_subdirectory(func)
 add_subdirectory(mnode)
+add_subdirectory(qnode)
+add_subdirectory(snode)
+add_subdirectory(bnode)
 add_subdirectory(profile)
 add_subdirectory(show)
 add_subdirectory(stb)

source/dnode/mgmt/impl/test/bnode/CMakeLists.txt

@@ -0,0 +1,11 @@
+aux_source_directory(. STEST_SRC)
+add_executable(dnode_test_bnode ${STEST_SRC})
+target_link_libraries(
+    dnode_test_bnode
+    PUBLIC sut
+)
+
+add_test(
+    NAME dnode_test_bnode
+    COMMAND dnode_test_bnode
+)

source/dnode/mgmt/impl/test/bnode/bnode.cpp

@@ -0,0 +1,154 @@
+/**
+ * @file dnode.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief DNODE module dnode-msg tests
+ * @version 0.1
+ * @date 2021-12-15
+ *
+ * @copyright Copyright (c) 2021
+ *
+ */
+
+#include "base.h"
+
+class DndTestBnode : public ::testing::Test {
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+ public:
+  static void SetUpTestSuite() {
+    test.Init("/tmp/dnode_test_bnode1", 9068);
+    const char* fqdn = "localhost";
+    const char* firstEp = "localhost:9068";
+
+    server2.Start("/tmp/dnode_test_bnode2", fqdn, 9069, firstEp);
+    taosMsleep(300);
+  }
+
+  static void TearDownTestSuite() {
+    server2.Stop();
+    test.Cleanup();
+  }
+
+  static Testbase   test;
+  static TestServer server2;
+};
+
+Testbase   DndTestBnode::test;
+TestServer DndTestBnode::server2;
+
+TEST_F(DndTestBnode, 01_ShowBnode) {
+  test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+  CHECK_META("show bnodes", 3);
+
+  CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+  CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+  CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+  test.SendShowRetrieveMsg();
+  EXPECT_EQ(test.GetShowRows(), 0);
+}
+
+TEST_F(DndTestBnode, 02_Create_Bnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateBnodeMsg);
+
+    SMCreateBnodeMsg* pReq = (SMCreateBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(1);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+    CHECK_META("show bnodes", 3);
+
+    CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+    CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+    CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9068", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
+
+TEST_F(DndTestBnode, 03_Create_Bnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateBnodeMsg);
+
+    SMCreateBnodeMsg* pReq = (SMCreateBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, TSDB_CODE_MND_DNODE_NOT_EXIST);
+  }
+}
+
+TEST_F(DndTestBnode, 04_Create_Bnode) {
+  {
+    // create dnode
+    int32_t contLen = sizeof(SCreateDnodeMsg);
+
+    SCreateDnodeMsg* pReq = (SCreateDnodeMsg*)rpcMallocCont(contLen);
+    strcpy(pReq->fqdn, "localhost");
+    pReq->port = htonl(9069);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_DNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    taosMsleep(1300);
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_DNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+  }
+
+  {
+    // create bnode
+    int32_t contLen = sizeof(SMCreateBnodeMsg);
+
+    SMCreateBnodeMsg* pReq = (SMCreateBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+
+    CheckInt16(1);
+    CheckInt16(2);
+    CheckBinary("localhost:9068", TSDB_EP_LEN);
+    CheckBinary("localhost:9069", TSDB_EP_LEN);
+    CheckTimestamp();
+    CheckTimestamp();
+  }
+
+  {
+    // drop bnode
+    int32_t contLen = sizeof(SMDropBnodeMsg);
+
+    SMDropBnodeMsg* pReq = (SMDropBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9068", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}

source/dnode/mgmt/impl/test/dnode/dnode.cpp

@@ -68,7 +68,7 @@ TEST_F(DndTestDnode, 01_ShowDnode) {
   CheckInt16(1);
   CheckBinary("localhost:9041", TSDB_EP_LEN);
   CheckInt16(0);
-  CheckInt16(1);
+  CheckInt16(16);
   CheckBinary("ready", 10);
   CheckTimestamp();
   CheckBinary("", 24);
@@ -112,8 +112,8 @@ TEST_F(DndTestDnode, 03_Create_Drop_Restart_Dnode) {
   CheckBinary("localhost:9042", TSDB_EP_LEN);
   CheckInt16(0);
   CheckInt16(0);
-  CheckInt16(1);
-  CheckInt16(1);
+  CheckInt16(16);
+  CheckInt16(16);
   CheckBinary("ready", 10);
   CheckBinary("ready", 10);
   CheckTimestamp();
@@ -140,7 +140,7 @@ TEST_F(DndTestDnode, 03_Create_Drop_Restart_Dnode) {
   CheckInt16(1);
   CheckBinary("localhost:9041", TSDB_EP_LEN);
   CheckInt16(0);
-  CheckInt16(1);
+  CheckInt16(16);
   CheckBinary("ready", 10);
   CheckTimestamp();
   CheckBinary("", 24);
@@ -199,10 +199,10 @@ TEST_F(DndTestDnode, 03_Create_Drop_Restart_Dnode) {
   CheckInt16(0);
   CheckInt16(0);
   CheckInt16(0);
-  CheckInt16(1);
-  CheckInt16(1);
-  CheckInt16(1);
-  CheckInt16(1);
+  CheckInt16(16);
+  CheckInt16(16);
+  CheckInt16(16);
+  CheckInt16(16);
   CheckBinary("ready", 10);
   CheckBinary("ready", 10);
   CheckBinary("ready", 10);
@@ -242,10 +242,10 @@ TEST_F(DndTestDnode, 03_Create_Drop_Restart_Dnode) {
   CheckInt16(0);
   CheckInt16(0);
   CheckInt16(0);
-  CheckInt16(1);
-  CheckInt16(1);
-  CheckInt16(1);
-  CheckInt16(1);
+  CheckInt16(16);
+  CheckInt16(16);
+  CheckInt16(16);
+  CheckInt16(16);
   CheckBinary("ready", 10);
   CheckBinary("ready", 10);
   CheckBinary("ready", 10);

source/dnode/mgmt/impl/test/mnode/mnode.cpp

@@ -72,9 +72,9 @@ TEST_F(DndTestMnode, 01_ShowDnode) {
 
 TEST_F(DndTestMnode, 02_Create_Mnode_Invalid_Id) {
   {
-    int32_t contLen = sizeof(SCreateMnodeMsg);
+    int32_t contLen = sizeof(SMCreateMnodeMsg);
 
-    SCreateMnodeMsg* pReq = (SCreateMnodeMsg*)rpcMallocCont(contLen);
+    SMCreateMnodeMsg* pReq = (SMCreateMnodeMsg*)rpcMallocCont(contLen);
     pReq->dnodeId = htonl(1);
 
     SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_MNODE, pReq, contLen);
@@ -85,9 +85,9 @@ TEST_F(DndTestMnode, 02_Create_Mnode_Invalid_Id) {
 
 TEST_F(DndTestMnode, 03_Create_Mnode_Invalid_Id) {
   {
-    int32_t contLen = sizeof(SCreateMnodeMsg);
+    int32_t contLen = sizeof(SMCreateMnodeMsg);
 
-    SCreateMnodeMsg* pReq = (SCreateMnodeMsg*)rpcMallocCont(contLen);
+    SMCreateMnodeMsg* pReq = (SMCreateMnodeMsg*)rpcMallocCont(contLen);
     pReq->dnodeId = htonl(2);
 
     SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_MNODE, pReq, contLen);
@@ -117,9 +117,9 @@ TEST_F(DndTestMnode, 04_Create_Mnode) {
 
   {
     // create mnode
-    int32_t contLen = sizeof(SCreateMnodeMsg);
+    int32_t contLen = sizeof(SMCreateMnodeMsg);
 
-    SCreateMnodeMsg* pReq = (SCreateMnodeMsg*)rpcMallocCont(contLen);
+    SMCreateMnodeMsg* pReq = (SMCreateMnodeMsg*)rpcMallocCont(contLen);
     pReq->dnodeId = htonl(2);
 
     SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_MNODE, pReq, contLen);
@@ -144,9 +144,9 @@ TEST_F(DndTestMnode, 04_Create_Mnode) {
 
   {
     // drop mnode
-    int32_t contLen = sizeof(SDropMnodeMsg);
+    int32_t contLen = sizeof(SMDropMnodeMsg);
 
-    SDropMnodeMsg* pReq = (SDropMnodeMsg*)rpcMallocCont(contLen);
+    SMDropMnodeMsg* pReq = (SMDropMnodeMsg*)rpcMallocCont(contLen);
     pReq->dnodeId = htonl(2);
 
     SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_MNODE, pReq, contLen);

source/dnode/mgmt/impl/test/qnode/CMakeLists.txt

@@ -0,0 +1,11 @@
+aux_source_directory(. QTEST_SRC)
+add_executable(dnode_test_qnode ${QTEST_SRC})
+target_link_libraries(
+    dnode_test_qnode
+    PUBLIC sut
+)
+
+add_test(
+    NAME dnode_test_qnode
+    COMMAND dnode_test_qnode
+)

source/dnode/mgmt/impl/test/qnode/qnode.cpp

@@ -0,0 +1,154 @@
+/**
+ * @file dnode.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief DNODE module dnode-msg tests
+ * @version 0.1
+ * @date 2021-12-15
+ *
+ * @copyright Copyright (c) 2021
+ *
+ */
+
+#include "base.h"
+
+class DndTestQnode : public ::testing::Test {
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+ public:
+  static void SetUpTestSuite() {
+    test.Init("/tmp/dnode_test_qnode1", 9064);
+    const char* fqdn = "localhost";
+    const char* firstEp = "localhost:9064";
+
+    server2.Start("/tmp/dnode_test_qnode2", fqdn, 9065, firstEp);
+    taosMsleep(300);
+  }
+
+  static void TearDownTestSuite() {
+    server2.Stop();
+    test.Cleanup();
+  }
+
+  static Testbase   test;
+  static TestServer server2;
+};
+
+Testbase   DndTestQnode::test;
+TestServer DndTestQnode::server2;
+
+TEST_F(DndTestQnode, 01_ShowQnode) {
+  test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+  CHECK_META("show qnodes", 3);
+
+  CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+  CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+  CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+  test.SendShowRetrieveMsg();
+  EXPECT_EQ(test.GetShowRows(), 0);
+}
+
+TEST_F(DndTestQnode, 02_Create_Qnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateQnodeMsg);
+
+    SMCreateQnodeMsg* pReq = (SMCreateQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(1);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+    CHECK_META("show qnodes", 3);
+
+    CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+    CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+    CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9064", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
+
+TEST_F(DndTestQnode, 03_Create_Qnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateQnodeMsg);
+
+    SMCreateQnodeMsg* pReq = (SMCreateQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, TSDB_CODE_MND_DNODE_NOT_EXIST);
+  }
+}
+
+TEST_F(DndTestQnode, 04_Create_Qnode) {
+  {
+    // create dnode
+    int32_t contLen = sizeof(SCreateDnodeMsg);
+
+    SCreateDnodeMsg* pReq = (SCreateDnodeMsg*)rpcMallocCont(contLen);
+    strcpy(pReq->fqdn, "localhost");
+    pReq->port = htonl(9065);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_DNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    taosMsleep(1300);
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_DNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+  }
+
+  {
+    // create qnode
+    int32_t contLen = sizeof(SMCreateQnodeMsg);
+
+    SMCreateQnodeMsg* pReq = (SMCreateQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+
+    CheckInt16(1);
+    CheckInt16(2);
+    CheckBinary("localhost:9064", TSDB_EP_LEN);
+    CheckBinary("localhost:9065", TSDB_EP_LEN);
+    CheckTimestamp();
+    CheckTimestamp();
+  }
+
+  {
+    // drop qnode
+    int32_t contLen = sizeof(SMDropQnodeMsg);
+
+    SMDropQnodeMsg* pReq = (SMDropQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9064", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}

source/dnode/mgmt/impl/test/snode/CMakeLists.txt

@@ -0,0 +1,11 @@
+aux_source_directory(. STEST_SRC)
+add_executable(dnode_test_snode ${STEST_SRC})
+target_link_libraries(
+    dnode_test_snode
+    PUBLIC sut
+)
+
+add_test(
+    NAME dnode_test_snode
+    COMMAND dnode_test_snode
+)

source/dnode/mgmt/impl/test/snode/snode.cpp

@@ -0,0 +1,154 @@
+/**
+ * @file dnode.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief DNODE module dnode-msg tests
+ * @version 0.1
+ * @date 2021-12-15
+ *
+ * @copyright Copyright (c) 2021
+ *
+ */
+
+#include "base.h"
+
+class DndTestSnode : public ::testing::Test {
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+ public:
+  static void SetUpTestSuite() {
+    test.Init("/tmp/dnode_test_snode1", 9066);
+    const char* fqdn = "localhost";
+    const char* firstEp = "localhost:9066";
+
+    server2.Start("/tmp/dnode_test_snode2", fqdn, 9067, firstEp);
+    taosMsleep(300);
+  }
+
+  static void TearDownTestSuite() {
+    server2.Stop();
+    test.Cleanup();
+  }
+
+  static Testbase   test;
+  static TestServer server2;
+};
+
+Testbase   DndTestSnode::test;
+TestServer DndTestSnode::server2;
+
+TEST_F(DndTestSnode, 01_ShowSnode) {
+  test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+  CHECK_META("show snodes", 3);
+
+  CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+  CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+  CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+  test.SendShowRetrieveMsg();
+  EXPECT_EQ(test.GetShowRows(), 0);
+}
+
+TEST_F(DndTestSnode, 02_Create_Snode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateSnodeMsg);
+
+    SMCreateSnodeMsg* pReq = (SMCreateSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(1);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+    CHECK_META("show snodes", 3);
+
+    CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+    CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+    CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9066", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
+
+TEST_F(DndTestSnode, 03_Create_Snode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateSnodeMsg);
+
+    SMCreateSnodeMsg* pReq = (SMCreateSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, TSDB_CODE_MND_DNODE_NOT_EXIST);
+  }
+}
+
+TEST_F(DndTestSnode, 04_Create_Snode) {
+  {
+    // create dnode
+    int32_t contLen = sizeof(SCreateDnodeMsg);
+
+    SCreateDnodeMsg* pReq = (SCreateDnodeMsg*)rpcMallocCont(contLen);
+    strcpy(pReq->fqdn, "localhost");
+    pReq->port = htonl(9067);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_DNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    taosMsleep(1300);
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_DNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+  }
+
+  {
+    // create snode
+    int32_t contLen = sizeof(SMCreateSnodeMsg);
+
+    SMCreateSnodeMsg* pReq = (SMCreateSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+
+    CheckInt16(1);
+    CheckInt16(2);
+    CheckBinary("localhost:9066", TSDB_EP_LEN);
+    CheckBinary("localhost:9067", TSDB_EP_LEN);
+    CheckTimestamp();
+    CheckTimestamp();
+  }
+
+  {
+    // drop snode
+    int32_t contLen = sizeof(SMDropSnodeMsg);
+
+    SMDropSnodeMsg* pReq = (SMDropSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9066", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}

source/dnode/mgmt/impl/test/stb/stb.cpp

@@ -146,8 +146,8 @@ TEST_F(DndTestStb, 01_Create_Show_Meta_Drop_Restart_Stb) {
       pSchema->bytes = htonl(pSchema->bytes);
     }
 
-    EXPECT_STREQ(pRsp->tbFname, "");
-    EXPECT_STREQ(pRsp->stbFname, "1.d1.stb");
+    EXPECT_STREQ(pRsp->tbFname, "1.d1.stb");
+    EXPECT_STREQ(pRsp->stbFname, "");
     EXPECT_EQ(pRsp->numOfColumns, 2);
     EXPECT_EQ(pRsp->numOfTags, 3);
     EXPECT_EQ(pRsp->precision, TSDB_TIME_PRECISION_MILLI);

source/dnode/mgmt/impl/test/sut/src/server.cpp

@@ -26,7 +26,7 @@ SDnodeOpt TestServer::BuildOption(const char* path, const char* fqdn, uint16_t p
   SDnodeOpt option = {0};
   option.sver = 1;
   option.numOfCores = 1;
-  option.numOfSupportVnodes = 1;
+  option.numOfSupportVnodes = 16;
   option.numOfCommitThreads = 1;
   option.statusInterval = 1;
   option.numOfThreadsPerCore = 1;

source/dnode/mnode/impl/inc/mndBnode.h

@@ -13,8 +13,8 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_MND_BALANCE_H_
-#define _TD_MND_BALANCE_H_
+#ifndef _TD_MND_BNODE_H_
+#define _TD_MND_BNODE_H_
 
 #include "mndInt.h"
 
@@ -22,11 +22,11 @@
 extern "C" {
 #endif
 
-int32_t mndInitBalance(SMnode *pMnode);
-void    mndCleanupBalance(SMnode *pMnode);
+int32_t mndInitBnode(SMnode *pMnode);
+void    mndCleanupBnode(SMnode *pMnode);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif /*_TD_MND_BALANCE_H_*/
+#endif /*_TD_MND_BNODE_H_*/

source/dnode/mnode/impl/inc/mndConsumer.h

@@ -0,0 +1,38 @@
+/*
+ * 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 _TD_MND_CONSUMER_H_
+#define _TD_MND_CONSUMER_H_
+
+#include "mndInt.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int32_t mndInitConsumer(SMnode *pMnode);
+void    mndCleanupConsumer(SMnode *pMnode);
+
+SConsumerObj *mndAcquireConsumer(SMnode *pMnode, int32_t consumerId);
+void          mndReleaseConsumer(SMnode *pMnode, SConsumerObj *pConsumer);
+
+SCGroupObj *mndAcquireCGroup(SMnode *pMnode, char *consumerGroup);
+void        mndReleaseCGroup(SMnode *pMnode, SCGroupObj *pCGroup);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_MND_CONSUMER_H_*/

source/dnode/mnode/impl/inc/mndDef.h

@@ -124,12 +124,8 @@ typedef struct {
   int64_t    rebootTime;
   int64_t    lastAccessTime;
   int32_t    accessTimes;
-  int16_t    numOfMnodes;
   int16_t    numOfVnodes;
-  int16_t    numOfQnodes;
-  int16_t    numOfSupportMnodes;
   int16_t    numOfSupportVnodes;
-  int16_t    numOfSupportQnodes;
   int16_t    numOfCores;
   EDndStatus status;
   EDndReason offlineReason;
@@ -148,6 +144,27 @@ typedef struct {
   SDnodeObj *pDnode;
 } SMnodeObj;
 
+typedef struct {
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  SDnodeObj *pDnode;
+} SQnodeObj;
+
+typedef struct {
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  SDnodeObj *pDnode;
+} SSnodeObj;
+
+typedef struct {
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  SDnodeObj *pDnode;
+} SBnodeObj;
+
 typedef struct {
   int32_t maxUsers;
   int32_t maxDbs;
@@ -305,8 +322,34 @@ typedef struct {
   void*    executor;
   int32_t  sqlLen;
   char*    sql;
+  char*    logicalPlan;
+  char*    physicalPlan;
 } STopicObj;
 
+typedef struct {
+  char name[TSDB_TOPIC_FNAME_LEN];
+  char db[TSDB_DB_FNAME_LEN];
+  int64_t createTime;
+  int64_t updateTime;
+  uint64_t uid;
+  //uint64_t dbUid;
+  int32_t version;
+  SRWLatch lock;
+
+} SConsumerObj;
+
+typedef struct {
+  char name[TSDB_TOPIC_FNAME_LEN];
+  char db[TSDB_DB_FNAME_LEN];
+  int64_t createTime;
+  int64_t updateTime;
+  uint64_t uid;
+  //uint64_t dbUid;
+  int32_t version;
+  SRWLatch lock;
+
+} SCGroupObj;
+
 typedef struct SMnodeMsg {
   char    user[TSDB_USER_LEN];
   char    db[TSDB_DB_FNAME_LEN];

source/dnode/mnode/impl/inc/mndInt.h

@@ -19,6 +19,7 @@
 #include "mndDef.h"
 #include "sdb.h"
 #include "tcache.h"
+#include "tep.h"
 #include "tqueue.h"
 #include "ttime.h"
 

source/dnode/mnode/impl/inc/mndQnode.h

@@ -13,9 +13,20 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#define _DEFAULT_SOURCE
-#include "os.h"
+#ifndef _TD_MND_QNODE_H_
+#define _TD_MND_QNODE_H_
+
 #include "mndInt.h"
 
-int32_t mndInitBalance(SMnode *pMnode) { return 0; }
-void    mndCleanupBalance(SMnode *pMnode) {}
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int32_t mndInitQnode(SMnode *pMnode);
+void    mndCleanupQnode(SMnode *pMnode);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_MND_QNODE_H_*/

source/dnode/mnode/impl/inc/mndSnode.h

@@ -0,0 +1,32 @@
+/*
+ * 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 _TD_MND_SNODE_H_
+#define _TD_MND_SNODE_H_
+
+#include "mndInt.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int32_t mndInitSnode(SMnode *pMnode);
+void    mndCleanupSnode(SMnode *pMnode);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_MND_SNODE_H_*/

source/dnode/mnode/impl/inc/mndVgroup.h

@@ -29,6 +29,7 @@ void     mndReleaseVgroup(SMnode *pMnode, SVgObj *pVgroup);
 SSdbRaw *mndVgroupActionEncode(SVgObj *pVgroup);
 int32_t  mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups);
 SEpSet   mndGetVgroupEpset(SMnode *pMnode, SVgObj *pVgroup);
+int32_t  mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId);
 
 SCreateVnodeMsg *mndBuildCreateVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup);
 SDropVnodeMsg   *mndBuildDropVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup);

source/dnode/mnode/impl/src/mndBnode.c

@@ -0,0 +1,446 @@
+/*
+ * 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 "mndBnode.h"
+#include "mndDnode.h"
+#include "mndShow.h"
+#include "mndTrans.h"
+
+#define TSDB_BNODE_VER_NUMBER 1
+#define TSDB_BNODE_RESERVE_SIZE 64
+
+static SSdbRaw *mndBnodeActionEncode(SBnodeObj *pObj);
+static SSdbRow *mndBnodeActionDecode(SSdbRaw *pRaw);
+static int32_t  mndBnodeActionInsert(SSdb *pSdb, SBnodeObj *pObj);
+static int32_t  mndBnodeActionDelete(SSdb *pSdb, SBnodeObj *pObj);
+static int32_t  mndBnodeActionUpdate(SSdb *pSdb, SBnodeObj *pOldBnode, SBnodeObj *pNewBnode);
+static int32_t  mndProcessCreateBnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropBnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessCreateBnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropBnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndGetBnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveBnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextBnode(SMnode *pMnode, void *pIter);
+
+int32_t mndInitBnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_BNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)mndBnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndBnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndBnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndBnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndBnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_BNODE, mndProcessCreateBnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_BNODE, mndProcessDropBnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_DND_CREATE_BNODE_RSP, mndProcessCreateBnodeRsp);
+  mndSetMsgHandle(pMnode, TDMT_DND_DROP_BNODE_RSP, mndProcessDropBnodeRsp);
+
+  mndAddShowMetaHandle(pMnode, TSDB_MGMT_TABLE_BNODE, mndGetBnodeMeta);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_BNODE, mndRetrieveBnodes);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_BNODE, mndCancelGetNextBnode);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupBnode(SMnode *pMnode) {}
+
+static SBnodeObj *mndAcquireBnode(SMnode *pMnode, int32_t snodeId) {
+  SSdb      *pSdb = pMnode->pSdb;
+  SBnodeObj *pObj = sdbAcquire(pSdb, SDB_BNODE, &snodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_BNODE_NOT_EXIST;
+  }
+  return pObj;
+}
+
+static void mndReleaseBnode(SMnode *pMnode, SBnodeObj *pObj) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pObj);
+}
+
+static SSdbRaw *mndBnodeActionEncode(SBnodeObj *pObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_BNODE, TSDB_BNODE_VER_NUMBER, sizeof(SBnodeObj) + TSDB_BNODE_RESERVE_SIZE);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_BNODE_RESERVE_SIZE)
+
+  return pRaw;
+}
+
+static SSdbRow *mndBnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != TSDB_BNODE_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode snode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SBnodeObj));
+  SBnodeObj *pObj = sdbGetRowObj(pRow);
+  if (pObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_BNODE_RESERVE_SIZE)
+
+  return pRow;
+}
+
+static int32_t mndBnodeActionInsert(SSdb *pSdb, SBnodeObj *pObj) {
+  mTrace("snode:%d, perform insert action", pObj->id);
+  pObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pObj->id);
+  if (pObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("snode:%d, failed to perform insert action since %s", pObj->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndBnodeActionDelete(SSdb *pSdb, SBnodeObj *pObj) {
+  mTrace("snode:%d, perform delete action", pObj->id);
+  if (pObj->pDnode != NULL) {
+    sdbRelease(pSdb, pObj->pDnode);
+    pObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndBnodeActionUpdate(SSdb *pSdb, SBnodeObj *pOldBnode, SBnodeObj *pNewBnode) {
+  mTrace("snode:%d, perform update action", pOldBnode->id);
+  pOldBnode->updateTime = pNewBnode->updateTime;
+  return 0;
+}
+
+static int32_t mndSetCreateBnodeRedoLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndBnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateBnodeCommitLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndBnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateBnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SBnodeObj *pObj) {
+  SDCreateBnodeMsg *pMsg = malloc(sizeof(SDCreateBnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDCreateBnodeMsg);
+  action.msgType = TDMT_DND_CREATE_BNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndCreateBnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateBnodeMsg *pCreate) {
+  SBnodeObj snodeObj = {0};
+  snodeObj.id = pDnode->id;
+  snodeObj.createdTime = taosGetTimestampMs();
+  snodeObj.updateTime = snodeObj.createdTime;
+
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+  mDebug("trans:%d, used to create snode:%d", pTrans->id, pCreate->dnodeId);
+
+  if (mndSetCreateBnodeRedoLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  if (mndSetCreateBnodeCommitLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  if (mndSetCreateBnodeRedoActions(pTrans, pDnode, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  code = 0;
+
+CREATE_BNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessCreateBnodeReq(SMnodeMsg *pMsg) {
+  SMnode           *pMnode = pMsg->pMnode;
+  SMCreateBnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  pCreate->dnodeId = htonl(pCreate->dnodeId);
+
+  mDebug("snode:%d, start to create", pCreate->dnodeId);
+
+  SBnodeObj *pObj = mndAcquireBnode(pMnode, pCreate->dnodeId);
+  if (pObj != NULL) {
+    mError("snode:%d, snode already exist", pObj->id);
+    mndReleaseBnode(pMnode, pObj);
+    return -1;
+  }
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pCreate->dnodeId);
+  if (pDnode == NULL) {
+    mError("snode:%d, dnode not exist", pCreate->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndCreateBnode(pMnode, pMsg, pDnode, pCreate);
+  mndReleaseDnode(pMnode, pDnode);
+
+  if (code != 0) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndSetDropBnodeRedoLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndBnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropBnodeCommitLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndBnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropBnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SBnodeObj *pObj) {
+  SDDropBnodeMsg *pMsg = malloc(sizeof(SDDropBnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDDropBnodeMsg);
+  action.msgType = TDMT_DND_DROP_BNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndDropBnode(SMnode *pMnode, SMnodeMsg *pMsg, SBnodeObj *pObj) {
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to drop since %s", pObj->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  mDebug("trans:%d, used to drop snode:%d", pTrans->id, pObj->id);
+
+  if (mndSetDropBnodeRedoLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  if (mndSetDropBnodeCommitLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  if (mndSetDropBnodeRedoActions(pTrans, pObj->pDnode, pObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  code = 0;
+
+DROP_BNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessDropBnodeReq(SMnodeMsg *pMsg) {
+  SMnode         *pMnode = pMsg->pMnode;
+  SMDropBnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  pDrop->dnodeId = htonl(pDrop->dnodeId);
+
+  mDebug("snode:%d, start to drop", pDrop->dnodeId);
+
+  if (pDrop->dnodeId <= 0) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    mError("snode:%d, failed to drop since %s", pDrop->dnodeId, terrstr());
+    return -1;
+  }
+
+  SBnodeObj *pObj = mndAcquireBnode(pMnode, pDrop->dnodeId);
+  if (pObj == NULL) {
+    mError("snode:%d, not exist", pDrop->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndDropBnode(pMnode, pMsg, pObj);
+  if (code != 0) {
+    mError("snode:%d, failed to drop since %s", pMnode->dnodeId, terrstr());
+    return -1;
+  }
+
+  sdbRelease(pMnode->pSdb, pMnode);
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndProcessCreateBnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessDropBnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndGetBnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = 2;
+  pSchema[cols].type = TSDB_DATA_TYPE_SMALLINT;
+  strcpy(pSchema[cols].name, "id");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_EP_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "endpoint");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  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 = sdbGetSize(pSdb, SDB_BNODE);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveBnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode    *pMnode = pMsg->pMnode;
+  SSdb      *pSdb = pMnode->pSdb;
+  int32_t    numOfRows = 0;
+  int32_t    cols = 0;
+  SBnodeObj *pObj = NULL;
+  char      *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_BNODE, pShow->pIter, (void **)&pObj);
+    if (pShow->pIter == NULL) break;
+
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int16_t *)pWrite = pObj->id;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pObj->pDnode->ep, pShow->bytes[cols]);
+
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pObj->createdTime;
+    cols++;
+
+    numOfRows++;
+    sdbRelease(pSdb, pObj);
+  }
+
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  pShow->numOfReads += numOfRows;
+
+  return numOfRows;
+}
+
+static void mndCancelGetNextBnode(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

source/dnode/mnode/impl/src/mndConsumer.c

@@ -0,0 +1,373 @@
+/*
+ * 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 "mndConsumer.h"
+#include "mndDb.h"
+#include "mndDnode.h"
+#include "mndMnode.h"
+#include "mndShow.h"
+#include "mndStb.h"
+#include "mndTopic.h"
+#include "mndTrans.h"
+#include "mndUser.h"
+#include "mndVgroup.h"
+#include "tname.h"
+
+#define MND_CONSUMER_VER_NUMBER 1
+#define MND_CONSUMER_RESERVE_SIZE 64
+
+static SSdbRaw *mndConsumerActionEncode(SConsumerObj *pConsumer);
+static SSdbRow *mndConsumerActionDecode(SSdbRaw *pRaw);
+static int32_t  mndConsumerActionInsert(SSdb *pSdb, SConsumerObj *pConsumer);
+static int32_t  mndConsumerActionDelete(SSdb *pSdb, SConsumerObj *pConsumer);
+static int32_t  mndConsumerActionUpdate(SSdb *pSdb, SConsumerObj *pConsumer, SConsumerObj *pNewConsumer);
+static int32_t  mndProcessCreateConsumerMsg(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropConsumerMsg(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropConsumerInRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessConsumerMetaMsg(SMnodeMsg *pMsg);
+static int32_t  mndGetConsumerMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveConsumer(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextConsumer(SMnode *pMnode, void *pIter);
+
+static int32_t mndProcessSubscribeReq(SMnodeMsg *pMsg);
+static int32_t mndProcessSubscribeRsp(SMnodeMsg *pMsg);
+static int32_t mndProcessSubscribeInternalReq(SMnodeMsg *pMsg);
+static int32_t mndProcessSubscribeInternalRsp(SMnodeMsg *pMsg);
+
+int32_t mndInitConsumer(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_CONSUMER,
+                     .keyType = SDB_KEY_BINARY,
+                     .encodeFp = (SdbEncodeFp)mndConsumerActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndConsumerActionDecode,
+                     .insertFp = (SdbInsertFp)mndConsumerActionInsert,
+                     .updateFp = (SdbUpdateFp)mndConsumerActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndConsumerActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_SUBSCRIBE, mndProcessSubscribeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_SUBSCRIBE_RSP, mndProcessSubscribeRsp);
+  mndSetMsgHandle(pMnode, TDMT_VND_SUBSCRIBE, mndProcessSubscribeInternalReq);
+  mndSetMsgHandle(pMnode, TDMT_VND_SUBSCRIBE_RSP, mndProcessSubscribeInternalRsp);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupConsumer(SMnode *pMnode) {}
+
+static SSdbRaw *mndConsumerActionEncode(SConsumerObj *pConsumer) {
+  int32_t  size = sizeof(SConsumerObj) + MND_CONSUMER_RESERVE_SIZE;
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_CONSUMER, MND_CONSUMER_VER_NUMBER, size);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_BINARY(pRaw, dataPos, pConsumer->name, TSDB_TABLE_FNAME_LEN);
+  SDB_SET_BINARY(pRaw, dataPos, pConsumer->db, TSDB_DB_FNAME_LEN);
+  SDB_SET_INT64(pRaw, dataPos, pConsumer->createTime);
+  SDB_SET_INT64(pRaw, dataPos, pConsumer->updateTime);
+  SDB_SET_INT64(pRaw, dataPos, pConsumer->uid);
+  /*SDB_SET_INT64(pRaw, dataPos, pConsumer->dbUid);*/
+  SDB_SET_INT32(pRaw, dataPos, pConsumer->version);
+
+  SDB_SET_RESERVE(pRaw, dataPos, MND_CONSUMER_RESERVE_SIZE);
+  SDB_SET_DATALEN(pRaw, dataPos);
+
+  return pRaw;
+}
+
+static SSdbRow *mndConsumerActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != MND_CONSUMER_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode consumer since %s", terrstr());
+    return NULL;
+  }
+
+  int32_t       size = sizeof(SConsumerObj) + TSDB_MAX_COLUMNS * sizeof(SSchema);
+  SSdbRow      *pRow = sdbAllocRow(size);
+  SConsumerObj *pConsumer = sdbGetRowObj(pRow);
+  if (pConsumer == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pConsumer->name, TSDB_TABLE_FNAME_LEN);
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pConsumer->db, TSDB_DB_FNAME_LEN);
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pConsumer->createTime);
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pConsumer->updateTime);
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pConsumer->uid);
+  /*SDB_GET_INT64(pRaw, pRow, dataPos, &pConsumer->dbUid);*/
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pConsumer->version);
+
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, MND_CONSUMER_RESERVE_SIZE);
+
+  return pRow;
+}
+
+static int32_t mndConsumerActionInsert(SSdb *pSdb, SConsumerObj *pConsumer) {
+  mTrace("consumer:%s, perform insert action", pConsumer->name);
+  return 0;
+}
+
+static int32_t mndConsumerActionDelete(SSdb *pSdb, SConsumerObj *pConsumer) {
+  mTrace("consumer:%s, perform delete action", pConsumer->name);
+  return 0;
+}
+
+static int32_t mndConsumerActionUpdate(SSdb *pSdb, SConsumerObj *pOldConsumer, SConsumerObj *pNewConsumer) {
+  mTrace("consumer:%s, perform update action", pOldConsumer->name);
+  atomic_exchange_32(&pOldConsumer->updateTime, pNewConsumer->updateTime);
+  atomic_exchange_32(&pOldConsumer->version, pNewConsumer->version);
+
+  taosWLockLatch(&pOldConsumer->lock);
+
+  // TODO handle update
+
+  taosWUnLockLatch(&pOldConsumer->lock);
+  return 0;
+}
+
+SConsumerObj *mndAcquireConsumer(SMnode *pMnode, int32_t consumerId) {
+  SSdb         *pSdb = pMnode->pSdb;
+  SConsumerObj *pConsumer = sdbAcquire(pSdb, SDB_CONSUMER, &consumerId);
+  if (pConsumer == NULL) {
+    /*terrno = TSDB_CODE_MND_CONSUMER_NOT_EXIST;*/
+  }
+  return pConsumer;
+}
+
+void mndReleaseConsumer(SMnode *pMnode, SConsumerObj *pConsumer) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pConsumer);
+}
+
+static int32_t mndProcessSubscribeReq(SMnodeMsg *pMsg) {
+  SMnode          *pMnode = pMsg->pMnode;
+  char            *msgStr = pMsg->rpcMsg.pCont;
+  SCMSubscribeReq *pSubscribe;
+  tDeserializeSCMSubscribeReq(msgStr, pSubscribe);
+  // add consumerGroupId -> list<consumerId> to sdb
+  // add consumerId -> list<consumer> to sdb
+  // add consumer -> list<consumerId> to sdb
+  return 0;
+}
+
+static int32_t mndProcessSubscribeRsp(SMnodeMsg *pMsg) { return 0; }
+
+static int32_t mndProcessSubscribeInternalReq(SMnodeMsg *pMsg) { return 0; }
+
+static int32_t mndProcessSubscribeInternalRsp(SMnodeMsg *pMsg) { return 0; }
+
+static int32_t mndProcessDropConsumerInRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessConsumerMetaMsg(SMnodeMsg *pMsg) {
+  SMnode        *pMnode = pMsg->pMnode;
+  STableInfoMsg *pInfo = pMsg->rpcMsg.pCont;
+
+  mDebug("consumer:%s, start to retrieve meta", pInfo->tableFname);
+
+#if 0
+  SDbObj *pDb = mndAcquireDbByConsumer(pMnode, pInfo->tableFname);
+  if (pDb == NULL) {
+    terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
+    mError("consumer:%s, failed to retrieve meta since %s", pInfo->tableFname, terrstr());
+    return -1;
+  }
+
+  SConsumerObj *pConsumer = mndAcquireConsumer(pMnode, pInfo->tableFname);
+  if (pConsumer == NULL) {
+    mndReleaseDb(pMnode, pDb);
+    terrno = TSDB_CODE_MND_INVALID_CONSUMER;
+    mError("consumer:%s, failed to get meta since %s", pInfo->tableFname, terrstr());
+    return -1;
+  }
+
+  taosRLockLatch(&pConsumer->lock);
+  int32_t totalCols = pConsumer->numOfColumns + pConsumer->numOfTags;
+  int32_t contLen = sizeof(STableMetaMsg) + totalCols * sizeof(SSchema);
+
+  STableMetaMsg *pMeta = rpcMallocCont(contLen);
+  if (pMeta == NULL) {
+    taosRUnLockLatch(&pConsumer->lock);
+    mndReleaseDb(pMnode, pDb);
+    mndReleaseConsumer(pMnode, pConsumer);
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    mError("consumer:%s, failed to get meta since %s", pInfo->tableFname, terrstr());
+    return -1;
+  }
+
+  memcpy(pMeta->consumerFname, pConsumer->name, TSDB_TABLE_FNAME_LEN);
+  pMeta->numOfTags = htonl(pConsumer->numOfTags);
+  pMeta->numOfColumns = htonl(pConsumer->numOfColumns);
+  pMeta->precision = pDb->cfg.precision;
+  pMeta->tableType = TSDB_SUPER_TABLE;
+  pMeta->update = pDb->cfg.update;
+  pMeta->sversion = htonl(pConsumer->version);
+  pMeta->tuid = htonl(pConsumer->uid);
+
+  for (int32_t i = 0; i < totalCols; ++i) {
+    SSchema *pSchema = &pMeta->pSchema[i];
+    SSchema *pSrcSchema = &pConsumer->pSchema[i];
+    memcpy(pSchema->name, pSrcSchema->name, TSDB_COL_NAME_LEN);
+    pSchema->type = pSrcSchema->type;
+    pSchema->colId = htonl(pSrcSchema->colId);
+    pSchema->bytes = htonl(pSrcSchema->bytes);
+  }
+  taosRUnLockLatch(&pConsumer->lock);
+  mndReleaseDb(pMnode, pDb);
+  mndReleaseConsumer(pMnode, pConsumer);
+
+  pMsg->pCont = pMeta;
+  pMsg->contLen = contLen;
+
+  mDebug("consumer:%s, meta is retrieved, cols:%d tags:%d", pInfo->tableFname, pConsumer->numOfColumns, pConsumer->numOfTags);
+#endif
+  return 0;
+}
+
+static int32_t mndGetNumOfConsumers(SMnode *pMnode, char *dbName, int32_t *pNumOfConsumers) {
+  SSdb *pSdb = pMnode->pSdb;
+
+  SDbObj *pDb = mndAcquireDb(pMnode, dbName);
+  if (pDb == NULL) {
+    terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
+    return -1;
+  }
+
+  int32_t numOfConsumers = 0;
+  void   *pIter = NULL;
+  while (1) {
+    SConsumerObj *pConsumer = NULL;
+    pIter = sdbFetch(pSdb, SDB_CONSUMER, pIter, (void **)&pConsumer);
+    if (pIter == NULL) break;
+
+    if (strcmp(pConsumer->db, dbName) == 0) {
+      numOfConsumers++;
+    }
+
+    sdbRelease(pSdb, pConsumer);
+  }
+
+  *pNumOfConsumers = numOfConsumers;
+  return 0;
+}
+
+static int32_t mndGetConsumerMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  if (mndGetNumOfConsumers(pMnode, pShow->db, &pShow->numOfRows) != 0) {
+    return -1;
+  }
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = TSDB_TABLE_NAME_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "name");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 4;
+  pSchema[cols].type = TSDB_DATA_TYPE_INT;
+  strcpy(pSchema[cols].name, "columns");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 4;
+  pSchema[cols].type = TSDB_DATA_TYPE_INT;
+  strcpy(pSchema[cols].name, "tags");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  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 = sdbGetSize(pSdb, SDB_CONSUMER);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveConsumer(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode       *pMnode = pMsg->pMnode;
+  SSdb         *pSdb = pMnode->pSdb;
+  int32_t       numOfRows = 0;
+  SConsumerObj *pConsumer = NULL;
+  int32_t       cols = 0;
+  char         *pWrite;
+  char          prefix[64] = {0};
+
+  tstrncpy(prefix, pShow->db, 64);
+  strcat(prefix, TS_PATH_DELIMITER);
+  int32_t prefixLen = (int32_t)strlen(prefix);
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_CONSUMER, pShow->pIter, (void **)&pConsumer);
+    if (pShow->pIter == NULL) break;
+
+    if (strncmp(pConsumer->name, prefix, prefixLen) != 0) {
+      sdbRelease(pSdb, pConsumer);
+      continue;
+    }
+
+    cols = 0;
+
+    char consumerName[TSDB_TABLE_NAME_LEN] = {0};
+    tstrncpy(consumerName, pConsumer->name + prefixLen, TSDB_TABLE_NAME_LEN);
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_TO_VARSTR(pWrite, consumerName);
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pConsumer->createTime;
+    cols++;
+
+    /*pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;*/
+    /**(int32_t *)pWrite = pConsumer->numOfColumns;*/
+    /*cols++;*/
+
+    /*pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;*/
+    /**(int32_t *)pWrite = pConsumer->numOfTags;*/
+    /*cols++;*/
+
+    numOfRows++;
+    sdbRelease(pSdb, pConsumer);
+  }
+
+  pShow->numOfReads += numOfRows;
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  return numOfRows;
+}
+
+static void mndCancelGetNextConsumer(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

source/dnode/mnode/impl/src/mndDnode.c

@@ -18,8 +18,7 @@
 #include "mndMnode.h"
 #include "mndShow.h"
 #include "mndTrans.h"
-#include "tep.h"
-#include "ttime.h"
+#include "mndVgroup.h"
 
 #define TSDB_DNODE_VER_NUMBER 1
 #define TSDB_DNODE_RESERVE_SIZE 64
@@ -370,7 +369,6 @@ static int32_t mndProcessStatusMsg(SMnodeMsg *pMsg) {
   }
 
   pRsp->dnodeCfg.dnodeId = htonl(pDnode->id);
-  pRsp->dnodeCfg.dropped = 0;
   pRsp->dnodeCfg.clusterId = htobe64(pMnode->clusterId);
   mndGetDnodeData(pMnode, &pRsp->dnodeEps, numOfEps);
 
@@ -700,7 +698,7 @@ static int32_t mndRetrieveDnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, i
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
-    *(int16_t *)pWrite = pDnode->numOfVnodes;
+    *(int16_t *)pWrite = mndGetVnodesNum(pMnode, pDnode->id);
     cols++;
 
     pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;

source/dnode/mnode/impl/src/mndMnode.c

@@ -251,7 +251,7 @@ static int32_t mndSetCreateMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDno
   void   *pIter = NULL;
   int32_t numOfReplicas = 0;
 
-  SCreateMnodeInMsg createMsg = {0};
+  SDCreateMnodeMsg createMsg = {0};
   while (1) {
     SMnodeObj *pMObj = NULL;
     pIter = sdbFetch(pSdb, SDB_MNODE, pIter, (void **)&pMObj);
@@ -281,18 +281,18 @@ static int32_t mndSetCreateMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDno
 
     STransAction action = {0};
 
-    SAlterMnodeInMsg *pMsg = malloc(sizeof(SAlterMnodeInMsg));
+    SDAlterMnodeMsg *pMsg = malloc(sizeof(SDAlterMnodeMsg));
     if (pMsg == NULL) {
       sdbCancelFetch(pSdb, pIter);
       sdbRelease(pSdb, pMObj);
       return -1;
     }
-    memcpy(pMsg, &createMsg, sizeof(SAlterMnodeInMsg));
+    memcpy(pMsg, &createMsg, sizeof(SDAlterMnodeMsg));
 
     pMsg->dnodeId = htonl(pMObj->id);
     action.epSet = mndGetDnodeEpset(pMObj->pDnode);
     action.pCont = pMsg;
-    action.contLen = sizeof(SAlterMnodeInMsg);
+    action.contLen = sizeof(SDAlterMnodeMsg);
     action.msgType = TDMT_DND_ALTER_MNODE;
 
     if (mndTransAppendRedoAction(pTrans, &action) != 0) {
@@ -309,14 +309,14 @@ static int32_t mndSetCreateMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDno
     STransAction action = {0};
     action.epSet = mndGetDnodeEpset(pDnode);
 
-    SCreateMnodeInMsg *pMsg = malloc(sizeof(SCreateMnodeInMsg));
+    SDCreateMnodeMsg *pMsg = malloc(sizeof(SDCreateMnodeMsg));
     if (pMsg == NULL) return -1;
-    memcpy(pMsg, &createMsg, sizeof(SAlterMnodeInMsg));
+    memcpy(pMsg, &createMsg, sizeof(SDAlterMnodeMsg));
     pMsg->dnodeId = htonl(pObj->id);
 
     action.epSet = mndGetDnodeEpset(pDnode);
     action.pCont = pMsg;
-    action.contLen = sizeof(SCreateMnodeInMsg);
+    action.contLen = sizeof(SDCreateMnodeMsg);
     action.msgType = TDMT_DND_CREATE_MNODE;
     if (mndTransAppendRedoAction(pTrans, &action) != 0) {
       free(pMsg);
@@ -327,9 +327,9 @@ static int32_t mndSetCreateMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDno
   return 0;
 }
 
-static int32_t mndCreateMnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SCreateMnodeMsg *pCreate) {
+static int32_t mndCreateMnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateMnodeMsg *pCreate) {
   SMnodeObj mnodeObj = {0};
-  mnodeObj.id = sdbGetMaxId(pMnode->pSdb, SDB_MNODE);
+  mnodeObj.id = pDnode->id;
   mnodeObj.createdTime = taosGetTimestampMs();
   mnodeObj.updateTime = mnodeObj.createdTime;
 
@@ -370,7 +370,7 @@ CREATE_MNODE_OVER:
 
 static int32_t mndProcessCreateMnodeReq(SMnodeMsg *pMsg) {
   SMnode          *pMnode = pMsg->pMnode;
-  SCreateMnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+  SMCreateMnodeMsg *pCreate = pMsg->rpcMsg.pCont;
 
   pCreate->dnodeId = htonl(pCreate->dnodeId);
 
@@ -423,7 +423,7 @@ static int32_t mndSetDropMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDnode
   void   *pIter = NULL;
   int32_t numOfReplicas = 0;
 
-  SAlterMnodeInMsg alterMsg = {0};
+  SDAlterMnodeMsg alterMsg = {0};
   while (1) {
     SMnodeObj *pMObj = NULL;
     pIter = sdbFetch(pSdb, SDB_MNODE, pIter, (void **)&pMObj);
@@ -449,18 +449,18 @@ static int32_t mndSetDropMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDnode
     if (pMObj->id != pObj->id) {
       STransAction action = {0};
 
-      SAlterMnodeInMsg *pMsg = malloc(sizeof(SAlterMnodeInMsg));
+      SDAlterMnodeMsg *pMsg = malloc(sizeof(SDAlterMnodeMsg));
       if (pMsg == NULL) {
         sdbCancelFetch(pSdb, pIter);
         sdbRelease(pSdb, pMObj);
         return -1;
       }
-      memcpy(pMsg, &alterMsg, sizeof(SAlterMnodeInMsg));
+      memcpy(pMsg, &alterMsg, sizeof(SDAlterMnodeMsg));
 
       pMsg->dnodeId = htonl(pMObj->id);
       action.epSet = mndGetDnodeEpset(pMObj->pDnode);
       action.pCont = pMsg;
-      action.contLen = sizeof(SAlterMnodeInMsg);
+      action.contLen = sizeof(SDAlterMnodeMsg);
       action.msgType = TDMT_DND_ALTER_MNODE;
 
       if (mndTransAppendRedoAction(pTrans, &action) != 0) {
@@ -478,7 +478,7 @@ static int32_t mndSetDropMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDnode
     STransAction action = {0};
     action.epSet = mndGetDnodeEpset(pDnode);
 
-    SDropMnodeInMsg *pMsg = malloc(sizeof(SDropMnodeInMsg));
+    SDDropMnodeMsg *pMsg = malloc(sizeof(SDDropMnodeMsg));
     if (pMsg == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
@@ -487,7 +487,7 @@ static int32_t mndSetDropMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDnode
 
     action.epSet = mndGetDnodeEpset(pDnode);
     action.pCont = pMsg;
-    action.contLen = sizeof(SDropMnodeInMsg);
+    action.contLen = sizeof(SDDropMnodeMsg);
     action.msgType = TDMT_DND_DROP_MNODE;
     if (mndTransAppendRedoAction(pTrans, &action) != 0) {
       free(pMsg);
@@ -537,7 +537,7 @@ DROP_MNODE_OVER:
 
 static int32_t mndProcessDropMnodeReq(SMnodeMsg *pMsg) {
   SMnode        *pMnode = pMsg->pMnode;
-  SDropMnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  SMDropMnodeMsg *pDrop = pMsg->rpcMsg.pCont;
   pDrop->dnodeId = htonl(pDrop->dnodeId);
 
   mDebug("mnode:%d, start to drop", pDrop->dnodeId);
@@ -562,7 +562,7 @@ static int32_t mndProcessDropMnodeReq(SMnodeMsg *pMsg) {
     return -1;
   }
 
-  sdbRelease(pMnode->pSdb, pMnode);
+  sdbRelease(pMnode->pSdb, pObj);
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 

source/dnode/mnode/impl/src/mndProfile.c

@@ -118,17 +118,17 @@ static SConnObj *mndCreateConn(SMnode *pMnode, SRpcConnInfo *pInfo, int32_t pid,
   SConnObj *pConn = taosCachePut(pMgmt->cache, &connId, sizeof(int32_t), &connObj, sizeof(connObj), keepTime * 1000);
   if (pConn == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
-    mError("conn:%d, data:%p failed to put into cache since %s, user:%s", connId, pConn, pInfo->user, terrstr());
+    mError("conn:%d, failed to put into cache since %s, user:%s", connId, pInfo->user, terrstr());
     return NULL;
   } else {
-    mTrace("conn:%d, data:%p created, user:%s", pConn->id, pConn, pInfo->user);
+    mTrace("conn:%d, is created, data:%p user:%s", pConn->id, pConn, pInfo->user);
     return pConn;
   }
 }
 
 static void mndFreeConn(SConnObj *pConn) {
   tfree(pConn->pQueries);
-  mTrace("conn:%d, data:%p destroyed", pConn->id, pConn);
+  mTrace("conn:%d, is destroyed, data:%p", pConn->id, pConn);
 }
 
 static SConnObj *mndAcquireConn(SMnode *pMnode, int32_t connId) {
@@ -143,13 +143,13 @@ static SConnObj *mndAcquireConn(SMnode *pMnode, int32_t connId) {
   int32_t keepTime = pMnode->cfg.shellActivityTimer * 3;
   pConn->lastAccessTimeMs = keepTime * 1000 + (uint64_t)taosGetTimestampMs();
 
-  mTrace("conn:%d, data:%p acquired from cache", pConn->id, pConn);
+  mTrace("conn:%d, acquired from cache, data:%p", pConn->id, pConn);
   return pConn;
 }
 
 static void mndReleaseConn(SMnode *pMnode, SConnObj *pConn) {
   if (pConn == NULL) return;
-  mTrace("conn:%d, data:%p released from cache", pConn->id, pConn);
+  mTrace("conn:%d, released from cache, data:%p", pConn->id, pConn);
 
   SProfileMgmt *pMgmt = &pMnode->profileMgmt;
   taosCacheRelease(pMgmt->cache, (void **)&pConn, false);

source/dnode/mnode/impl/src/mndQnode.c

@@ -0,0 +1,446 @@
+/*
+ * 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 "mndQnode.h"
+#include "mndDnode.h"
+#include "mndShow.h"
+#include "mndTrans.h"
+
+#define TSDB_QNODE_VER_NUMBER 1
+#define TSDB_QNODE_RESERVE_SIZE 64
+
+static SSdbRaw *mndQnodeActionEncode(SQnodeObj *pObj);
+static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw);
+static int32_t  mndQnodeActionInsert(SSdb *pSdb, SQnodeObj *pObj);
+static int32_t  mndQnodeActionDelete(SSdb *pSdb, SQnodeObj *pObj);
+static int32_t  mndQnodeActionUpdate(SSdb *pSdb, SQnodeObj *pOldQnode, SQnodeObj *pNewQnode);
+static int32_t  mndProcessCreateQnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropQnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessCreateQnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropQnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndGetQnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveQnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextQnode(SMnode *pMnode, void *pIter);
+
+int32_t mndInitQnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_QNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)mndQnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndQnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndQnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndQnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndQnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_QNODE, mndProcessCreateQnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_QNODE, mndProcessDropQnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_DND_CREATE_QNODE_RSP, mndProcessCreateQnodeRsp);
+  mndSetMsgHandle(pMnode, TDMT_DND_DROP_QNODE_RSP, mndProcessDropQnodeRsp);
+
+  mndAddShowMetaHandle(pMnode, TSDB_MGMT_TABLE_QNODE, mndGetQnodeMeta);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_QNODE, mndRetrieveQnodes);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_QNODE, mndCancelGetNextQnode);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupQnode(SMnode *pMnode) {}
+
+static SQnodeObj *mndAcquireQnode(SMnode *pMnode, int32_t qnodeId) {
+  SSdb      *pSdb = pMnode->pSdb;
+  SQnodeObj *pObj = sdbAcquire(pSdb, SDB_QNODE, &qnodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_QNODE_NOT_EXIST;
+  }
+  return pObj;
+}
+
+static void mndReleaseQnode(SMnode *pMnode, SQnodeObj *pObj) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pObj);
+}
+
+static SSdbRaw *mndQnodeActionEncode(SQnodeObj *pObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_QNODE, TSDB_QNODE_VER_NUMBER, sizeof(SQnodeObj) + TSDB_QNODE_RESERVE_SIZE);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_QNODE_RESERVE_SIZE)
+
+  return pRaw;
+}
+
+static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != TSDB_QNODE_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode qnode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SQnodeObj));
+  SQnodeObj *pObj = sdbGetRowObj(pRow);
+  if (pObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_QNODE_RESERVE_SIZE)
+
+  return pRow;
+}
+
+static int32_t mndQnodeActionInsert(SSdb *pSdb, SQnodeObj *pObj) {
+  mTrace("qnode:%d, perform insert action", pObj->id);
+  pObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pObj->id);
+  if (pObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("qnode:%d, failed to perform insert action since %s", pObj->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndQnodeActionDelete(SSdb *pSdb, SQnodeObj *pObj) {
+  mTrace("qnode:%d, perform delete action", pObj->id);
+  if (pObj->pDnode != NULL) {
+    sdbRelease(pSdb, pObj->pDnode);
+    pObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndQnodeActionUpdate(SSdb *pSdb, SQnodeObj *pOldQnode, SQnodeObj *pNewQnode) {
+  mTrace("qnode:%d, perform update action", pOldQnode->id);
+  pOldQnode->updateTime = pNewQnode->updateTime;
+  return 0;
+}
+
+static int32_t mndSetCreateQnodeRedoLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndQnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateQnodeCommitLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndQnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateQnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SQnodeObj *pObj) {
+  SDCreateQnodeMsg *pMsg = malloc(sizeof(SDCreateQnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDCreateQnodeMsg);
+  action.msgType = TDMT_DND_CREATE_QNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndCreateQnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateQnodeMsg *pCreate) {
+  SQnodeObj qnodeObj = {0};
+  qnodeObj.id = pDnode->id;
+  qnodeObj.createdTime = taosGetTimestampMs();
+  qnodeObj.updateTime = qnodeObj.createdTime;
+
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("qnode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+  mDebug("trans:%d, used to create qnode:%d", pTrans->id, pCreate->dnodeId);
+
+  if (mndSetCreateQnodeRedoLogs(pTrans, &qnodeObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  if (mndSetCreateQnodeCommitLogs(pTrans, &qnodeObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  if (mndSetCreateQnodeRedoActions(pTrans, pDnode, &qnodeObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  code = 0;
+
+CREATE_QNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessCreateQnodeReq(SMnodeMsg *pMsg) {
+  SMnode           *pMnode = pMsg->pMnode;
+  SMCreateQnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  pCreate->dnodeId = htonl(pCreate->dnodeId);
+
+  mDebug("qnode:%d, start to create", pCreate->dnodeId);
+
+  SQnodeObj *pObj = mndAcquireQnode(pMnode, pCreate->dnodeId);
+  if (pObj != NULL) {
+    mError("qnode:%d, qnode already exist", pObj->id);
+    mndReleaseQnode(pMnode, pObj);
+    return -1;
+  }
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pCreate->dnodeId);
+  if (pDnode == NULL) {
+    mError("qnode:%d, dnode not exist", pCreate->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndCreateQnode(pMnode, pMsg, pDnode, pCreate);
+  mndReleaseDnode(pMnode, pDnode);
+
+  if (code != 0) {
+    mError("qnode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndSetDropQnodeRedoLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndQnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropQnodeCommitLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndQnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropQnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SQnodeObj *pObj) {
+  SDDropQnodeMsg *pMsg = malloc(sizeof(SDDropQnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDDropQnodeMsg);
+  action.msgType = TDMT_DND_DROP_QNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndDropQnode(SMnode *pMnode, SMnodeMsg *pMsg, SQnodeObj *pObj) {
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("qnode:%d, failed to drop since %s", pObj->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  mDebug("trans:%d, used to drop qnode:%d", pTrans->id, pObj->id);
+
+  if (mndSetDropQnodeRedoLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  if (mndSetDropQnodeCommitLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  if (mndSetDropQnodeRedoActions(pTrans, pObj->pDnode, pObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  code = 0;
+
+DROP_QNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessDropQnodeReq(SMnodeMsg *pMsg) {
+  SMnode         *pMnode = pMsg->pMnode;
+  SMDropQnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  pDrop->dnodeId = htonl(pDrop->dnodeId);
+
+  mDebug("qnode:%d, start to drop", pDrop->dnodeId);
+
+  if (pDrop->dnodeId <= 0) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    mError("qnode:%d, failed to drop since %s", pDrop->dnodeId, terrstr());
+    return -1;
+  }
+
+  SQnodeObj *pObj = mndAcquireQnode(pMnode, pDrop->dnodeId);
+  if (pObj == NULL) {
+    mError("qnode:%d, not exist", pDrop->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndDropQnode(pMnode, pMsg, pObj);
+  if (code != 0) {
+    mError("qnode:%d, failed to drop since %s", pMnode->dnodeId, terrstr());
+    return -1;
+  }
+
+  sdbRelease(pMnode->pSdb, pMnode);
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndProcessCreateQnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessDropQnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndGetQnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = 2;
+  pSchema[cols].type = TSDB_DATA_TYPE_SMALLINT;
+  strcpy(pSchema[cols].name, "id");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_EP_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "endpoint");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  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 = sdbGetSize(pSdb, SDB_QNODE);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveQnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode    *pMnode = pMsg->pMnode;
+  SSdb      *pSdb = pMnode->pSdb;
+  int32_t    numOfRows = 0;
+  int32_t    cols = 0;
+  SQnodeObj *pObj = NULL;
+  char      *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_QNODE, pShow->pIter, (void **)&pObj);
+    if (pShow->pIter == NULL) break;
+
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int16_t *)pWrite = pObj->id;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pObj->pDnode->ep, pShow->bytes[cols]);
+
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pObj->createdTime;
+    cols++;
+
+    numOfRows++;
+    sdbRelease(pSdb, pObj);
+  }
+
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  pShow->numOfReads += numOfRows;
+
+  return numOfRows;
+}
+
+static void mndCancelGetNextQnode(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

source/dnode/mnode/impl/src/mndShow.c

@@ -56,31 +56,24 @@ static SShowObj *mndCreateShowObj(SMnode *pMnode, SShowMsg *pMsg) {
   if (showId == 0) atomic_add_fetch_32(&pMgmt->showId, 1);
 
   int32_t  size = sizeof(SShowObj) + pMsg->payloadLen;
-  SShowObj *pShow = calloc(1, size);
-  if (pShow != NULL) {
-    pShow->id = showId;
-    pShow->pMnode = pMnode;
-    pShow->type = pMsg->type;
-    pShow->payloadLen = pMsg->payloadLen;
-    memcpy(pShow->db, pMsg->db, TSDB_DB_FNAME_LEN);
-    memcpy(pShow->payload, pMsg->payload, pMsg->payloadLen);
-  } else {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    mError("failed to process show-meta msg:%s since %s", mndShowStr(pMsg->type), terrstr());
-    return NULL;
-  }
+  SShowObj showObj = {0};
+  showObj.id = showId;
+  showObj.pMnode = pMnode;
+  showObj.type = pMsg->type;
+  showObj.payloadLen = pMsg->payloadLen;
+  memcpy(showObj.db, pMsg->db, TSDB_DB_FNAME_LEN);
+  memcpy(showObj.payload, pMsg->payload, pMsg->payloadLen);
 
   int32_t   keepTime = pMnode->cfg.shellActivityTimer * 6 * 1000;
-  SShowObj *pShowRet = taosCachePut(pMgmt->cache, &showId, sizeof(int32_t), pShow, size, keepTime);
-  free(pShow);
-  if (pShowRet == NULL) {
+  SShowObj *pShow = taosCachePut(pMgmt->cache, &showId, sizeof(int32_t), &showObj, size, keepTime);
+  if (pShow == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     mError("show:%d, failed to put into cache since %s", showId, terrstr());
     return NULL;
-  } else {
-    mTrace("show:%d, data:%p created", showId, pShowRet);
-    return pShowRet;
   }
+
+  mTrace("show:%d, is created, data:%p", showId, pShow);
+  return pShow;
 }
 
 static void mndFreeShowObj(SShowObj *pShow) {
@@ -94,7 +87,7 @@ static void mndFreeShowObj(SShowObj *pShow) {
     }
   }
 
-  mTrace("show:%d, data:%p destroyed", pShow->id, pShow);
+  mTrace("show:%d, is destroyed, data:%p", pShow->id, pShow);
 }
 
 static SShowObj *mndAcquireShowObj(SMnode *pMnode, int32_t showId) {
@@ -106,13 +99,13 @@ static SShowObj *mndAcquireShowObj(SMnode *pMnode, int32_t showId) {
     return NULL;
   }
 
-  mTrace("show:%d, data:%p acquired from cache", pShow->id, pShow);
+  mTrace("show:%d, acquired from cache, data:%p", pShow->id, pShow);
   return pShow;
 }
 
 static void mndReleaseShowObj(SShowObj *pShow, bool forceRemove) {
   if (pShow == NULL) return;
-  mTrace("show:%d, data:%p released from cache, force:%d", pShow->id, pShow, forceRemove);
+  mTrace("show:%d, released from cache, data:%p force:%d", pShow->id, pShow, forceRemove);
 
   // A bug in tcache.c
   forceRemove = 0;
@@ -158,8 +151,8 @@ static int32_t mndProcessShowMsg(SMnodeMsg *pMnodeMsg) {
   }
 
   int32_t code = (*metaFp)(pMnodeMsg, pShow, &pRsp->tableMeta);
-  mDebug("show:%d, data:%p get meta finished, numOfRows:%d cols:%d type:%s result:%s", pShow->id, pShow,
-         pShow->numOfRows, pShow->numOfColumns, mndShowStr(type), tstrerror(code));
+  mDebug("show:%d, get meta finished, numOfRows:%d cols:%d type:%s result:%s", pShow->id, pShow->numOfRows,
+         pShow->numOfColumns, mndShowStr(type), tstrerror(code));
 
   if (code == TSDB_CODE_SUCCESS) {
     pMnodeMsg->contLen = sizeof(SShowRsp) + sizeof(SSchema) * pShow->numOfColumns;
@@ -195,16 +188,15 @@ static int32_t mndProcessRetrieveMsg(SMnodeMsg *pMnodeMsg) {
   if (retrieveFp == NULL) {
     mndReleaseShowObj(pShow, false);
     terrno = TSDB_CODE_MSG_NOT_PROCESSED;
-    mError("show:%d, data:%p failed to retrieve data since %s", pShow->id, pShow, terrstr());
+    mError("show:%d, failed to retrieve data since %s", pShow->id, terrstr());
     return -1;
   }
 
-  mDebug("show:%d, data:%p start retrieve data, numOfReads:%d numOfRows:%d type:%s", pShow->id, pShow,
-         pShow->numOfReads, pShow->numOfRows, mndShowStr(pShow->type));
+  mDebug("show:%d, start retrieve data, numOfReads:%d numOfRows:%d type:%s", pShow->id, pShow->numOfReads,
+         pShow->numOfRows, mndShowStr(pShow->type));
 
   if (mndCheckRetrieveFinished(pShow)) {
-    mDebug("show:%d, data:%p read finished, numOfReads:%d numOfRows:%d", pShow->id, pShow, pShow->numOfReads,
-           pShow->numOfRows);
+    mDebug("show:%d, read finished, numOfReads:%d numOfRows:%d", pShow->id, pShow->numOfReads, pShow->numOfRows);
     pShow->numOfReads = pShow->numOfRows;
   }
 
@@ -227,7 +219,7 @@ static int32_t mndProcessRetrieveMsg(SMnodeMsg *pMnodeMsg) {
   if (pRsp == NULL) {
     mndReleaseShowObj(pShow, false);
     terrno = TSDB_CODE_OUT_OF_MEMORY;
-    mError("show:%d, data:%p failed to retrieve data since %s", pShow->id, pShow, terrstr());
+    mError("show:%d, failed to retrieve data since %s", pShow->id, terrstr());
     return -1;
   }
 
@@ -236,7 +228,7 @@ static int32_t mndProcessRetrieveMsg(SMnodeMsg *pMnodeMsg) {
     rowsRead = (*retrieveFp)(pMnodeMsg, pShow, pRsp->data, rowsToRead);
   }
 
-  mDebug("show:%d, data:%p stop retrieve data, rowsRead:%d rowsToRead:%d", pShow->id, pShow, rowsRead, rowsToRead);
+  mDebug("show:%d, stop retrieve data, rowsRead:%d rowsToRead:%d", pShow->id, rowsRead, rowsToRead);
 
   pRsp->numOfRows = htonl(rowsRead);
   pRsp->precision = TSDB_TIME_PRECISION_MILLI;  // millisecond time precision
@@ -246,10 +238,10 @@ static int32_t mndProcessRetrieveMsg(SMnodeMsg *pMnodeMsg) {
 
   if (rowsRead == 0 || rowsToRead == 0 || (rowsRead == rowsToRead && pShow->numOfRows == pShow->numOfReads)) {
     pRsp->completed = 1;
-    mDebug("show:%d, data:%p retrieve completed", pShow->id, pShow);
+    mDebug("show:%d, retrieve completed", pShow->id);
     mndReleaseShowObj(pShow, true);
   } else {
-    mDebug("show:%d, data:%p retrieve not completed yet", pShow->id, pShow);
+    mDebug("show:%d, retrieve not completed yet", pShow->id);
     mndReleaseShowObj(pShow, false);
   }
 
@@ -270,6 +262,12 @@ char *mndShowStr(int32_t showType) {
       return "show dnodes";
     case TSDB_MGMT_TABLE_MNODE:
       return "show mnodes";
+    case TSDB_MGMT_TABLE_QNODE:
+      return "show qnodes";
+    case TSDB_MGMT_TABLE_SNODE:
+      return "show snodes";
+    case TSDB_MGMT_TABLE_BNODE:
+      return "show bnodes";
     case TSDB_MGMT_TABLE_VGROUP:
       return "show vgroups";
     case TSDB_MGMT_TABLE_STB:

source/dnode/mnode/impl/src/mndSnode.c

@@ -0,0 +1,446 @@
+/*
+ * 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 "mndSnode.h"
+#include "mndDnode.h"
+#include "mndShow.h"
+#include "mndTrans.h"
+
+#define TSDB_SNODE_VER_NUMBER 1
+#define TSDB_SNODE_RESERVE_SIZE 64
+
+static SSdbRaw *mndSnodeActionEncode(SSnodeObj *pObj);
+static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw);
+static int32_t  mndSnodeActionInsert(SSdb *pSdb, SSnodeObj *pObj);
+static int32_t  mndSnodeActionDelete(SSdb *pSdb, SSnodeObj *pObj);
+static int32_t  mndSnodeActionUpdate(SSdb *pSdb, SSnodeObj *pOldSnode, SSnodeObj *pNewSnode);
+static int32_t  mndProcessCreateSnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropSnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessCreateSnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropSnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndGetSnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveSnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextSnode(SMnode *pMnode, void *pIter);
+
+int32_t mndInitSnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_SNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)mndSnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndSnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndSnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndSnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndSnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_SNODE, mndProcessCreateSnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_SNODE, mndProcessDropSnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_DND_CREATE_SNODE_RSP, mndProcessCreateSnodeRsp);
+  mndSetMsgHandle(pMnode, TDMT_DND_DROP_SNODE_RSP, mndProcessDropSnodeRsp);
+
+  mndAddShowMetaHandle(pMnode, TSDB_MGMT_TABLE_SNODE, mndGetSnodeMeta);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_SNODE, mndRetrieveSnodes);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_SNODE, mndCancelGetNextSnode);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupSnode(SMnode *pMnode) {}
+
+static SSnodeObj *mndAcquireSnode(SMnode *pMnode, int32_t snodeId) {
+  SSdb      *pSdb = pMnode->pSdb;
+  SSnodeObj *pObj = sdbAcquire(pSdb, SDB_SNODE, &snodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_SNODE_NOT_EXIST;
+  }
+  return pObj;
+}
+
+static void mndReleaseSnode(SMnode *pMnode, SSnodeObj *pObj) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pObj);
+}
+
+static SSdbRaw *mndSnodeActionEncode(SSnodeObj *pObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_SNODE, TSDB_SNODE_VER_NUMBER, sizeof(SSnodeObj) + TSDB_SNODE_RESERVE_SIZE);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_SNODE_RESERVE_SIZE)
+
+  return pRaw;
+}
+
+static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != TSDB_SNODE_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode snode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SSnodeObj));
+  SSnodeObj *pObj = sdbGetRowObj(pRow);
+  if (pObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_SNODE_RESERVE_SIZE)
+
+  return pRow;
+}
+
+static int32_t mndSnodeActionInsert(SSdb *pSdb, SSnodeObj *pObj) {
+  mTrace("snode:%d, perform insert action", pObj->id);
+  pObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pObj->id);
+  if (pObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("snode:%d, failed to perform insert action since %s", pObj->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndSnodeActionDelete(SSdb *pSdb, SSnodeObj *pObj) {
+  mTrace("snode:%d, perform delete action", pObj->id);
+  if (pObj->pDnode != NULL) {
+    sdbRelease(pSdb, pObj->pDnode);
+    pObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndSnodeActionUpdate(SSdb *pSdb, SSnodeObj *pOldSnode, SSnodeObj *pNewSnode) {
+  mTrace("snode:%d, perform update action", pOldSnode->id);
+  pOldSnode->updateTime = pNewSnode->updateTime;
+  return 0;
+}
+
+static int32_t mndSetCreateSnodeRedoLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndSnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateSnodeCommitLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndSnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateSnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SSnodeObj *pObj) {
+  SDCreateSnodeMsg *pMsg = malloc(sizeof(SDCreateSnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDCreateSnodeMsg);
+  action.msgType = TDMT_DND_CREATE_SNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndCreateSnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateSnodeMsg *pCreate) {
+  SSnodeObj snodeObj = {0};
+  snodeObj.id = pDnode->id;
+  snodeObj.createdTime = taosGetTimestampMs();
+  snodeObj.updateTime = snodeObj.createdTime;
+
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+  mDebug("trans:%d, used to create snode:%d", pTrans->id, pCreate->dnodeId);
+
+  if (mndSetCreateSnodeRedoLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  if (mndSetCreateSnodeCommitLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  if (mndSetCreateSnodeRedoActions(pTrans, pDnode, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  code = 0;
+
+CREATE_SNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessCreateSnodeReq(SMnodeMsg *pMsg) {
+  SMnode           *pMnode = pMsg->pMnode;
+  SMCreateSnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  pCreate->dnodeId = htonl(pCreate->dnodeId);
+
+  mDebug("snode:%d, start to create", pCreate->dnodeId);
+
+  SSnodeObj *pObj = mndAcquireSnode(pMnode, pCreate->dnodeId);
+  if (pObj != NULL) {
+    mError("snode:%d, snode already exist", pObj->id);
+    mndReleaseSnode(pMnode, pObj);
+    return -1;
+  }
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pCreate->dnodeId);
+  if (pDnode == NULL) {
+    mError("snode:%d, dnode not exist", pCreate->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndCreateSnode(pMnode, pMsg, pDnode, pCreate);
+  mndReleaseDnode(pMnode, pDnode);
+
+  if (code != 0) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndSetDropSnodeRedoLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndSnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropSnodeCommitLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndSnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropSnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SSnodeObj *pObj) {
+  SDDropSnodeMsg *pMsg = malloc(sizeof(SDDropSnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDDropSnodeMsg);
+  action.msgType = TDMT_DND_DROP_SNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndDropSnode(SMnode *pMnode, SMnodeMsg *pMsg, SSnodeObj *pObj) {
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to drop since %s", pObj->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  mDebug("trans:%d, used to drop snode:%d", pTrans->id, pObj->id);
+
+  if (mndSetDropSnodeRedoLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  if (mndSetDropSnodeCommitLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  if (mndSetDropSnodeRedoActions(pTrans, pObj->pDnode, pObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  code = 0;
+
+DROP_SNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessDropSnodeReq(SMnodeMsg *pMsg) {
+  SMnode         *pMnode = pMsg->pMnode;
+  SMDropSnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  pDrop->dnodeId = htonl(pDrop->dnodeId);
+
+  mDebug("snode:%d, start to drop", pDrop->dnodeId);
+
+  if (pDrop->dnodeId <= 0) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    mError("snode:%d, failed to drop since %s", pDrop->dnodeId, terrstr());
+    return -1;
+  }
+
+  SSnodeObj *pObj = mndAcquireSnode(pMnode, pDrop->dnodeId);
+  if (pObj == NULL) {
+    mError("snode:%d, not exist", pDrop->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndDropSnode(pMnode, pMsg, pObj);
+  if (code != 0) {
+    mError("snode:%d, failed to drop since %s", pMnode->dnodeId, terrstr());
+    return -1;
+  }
+
+  sdbRelease(pMnode->pSdb, pMnode);
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndProcessCreateSnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessDropSnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndGetSnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = 2;
+  pSchema[cols].type = TSDB_DATA_TYPE_SMALLINT;
+  strcpy(pSchema[cols].name, "id");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_EP_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "endpoint");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  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 = sdbGetSize(pSdb, SDB_SNODE);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveSnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode    *pMnode = pMsg->pMnode;
+  SSdb      *pSdb = pMnode->pSdb;
+  int32_t    numOfRows = 0;
+  int32_t    cols = 0;
+  SSnodeObj *pObj = NULL;
+  char      *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_SNODE, pShow->pIter, (void **)&pObj);
+    if (pShow->pIter == NULL) break;
+
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int16_t *)pWrite = pObj->id;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pObj->pDnode->ep, pShow->bytes[cols]);
+
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pObj->createdTime;
+    cols++;
+
+    numOfRows++;
+    sdbRelease(pSdb, pObj);
+  }
+
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  pShow->numOfReads += numOfRows;
+
+  return numOfRows;
+}
+
+static void mndCancelGetNextSnode(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

source/dnode/mnode/impl/src/mndStb.c

@@ -201,7 +201,6 @@ static SDbObj *mndAcquireDbByStb(SMnode *pMnode, char *stbName) {
 }
 
 static void *mndBuildCreateStbMsg(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb, int *pContLen) {
-#if 1
   SVCreateTbReq req;
   void *        buf;
   int           bsize;
@@ -235,43 +234,12 @@ static void *mndBuildCreateStbMsg(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb
 
   *pContLen = sizeof(SMsgHead) + bsize;
   return buf;
-
-#else
-  int32_t totalCols = pStb->numOfTags + pStb->numOfColumns;
-  int32_t contLen = totalCols * sizeof(SSchema) + sizeof(SCreateStbInternalMsg);
-
-  SCreateStbInternalMsg *pCreate = calloc(1, contLen);
-  if (pCreate == NULL) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  }
-
-  pCreate->head.contLen = htonl(contLen);
-  pCreate->head.vgId = htonl(pVgroup->vgId);
-  memcpy(pCreate->name, pStb->name, TSDB_TABLE_FNAME_LEN);
-  pCreate->suid = htobe64(pStb->uid);
-  pCreate->sverson = htonl(pStb->version);
-  pCreate->ttl = 0;
-  pCreate->keep = 0;
-  pCreate->numOfTags = htonl(pStb->numOfTags);
-  pCreate->numOfColumns = htonl(pStb->numOfColumns);
-
-  memcpy(pCreate->pSchema, pStb->pSchema, totalCols * sizeof(SSchema));
-  for (int32_t t = 0; t < totalCols; ++t) {
-    SSchema *pSchema = &pCreate->pSchema[t];
-    pSchema->bytes = htonl(pSchema->bytes);
-    pSchema->colId = htonl(pSchema->colId);
-  }
-
-  *pContLen = contLen;
-  return pCreate;
-#endif
 }
 
-static SDropStbInternalMsg *mndBuildDropStbMsg(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb) {
-  int32_t contLen = sizeof(SDropStbInternalMsg);
+static SVDropStbReq *mndBuildDropStbMsg(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb) {
+  int32_t contLen = sizeof(SVDropStbReq);
 
-  SDropStbInternalMsg *pDrop = calloc(1, contLen);
+  SVDropStbReq *pDrop = calloc(1, contLen);
   if (pDrop == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
@@ -402,7 +370,7 @@ static int32_t mndSetCreateStbUndoActions(SMnode *pMnode, STrans *pTrans, SDbObj
     if (pIter == NULL) break;
     if (pVgroup->dbUid != pDb->uid) continue;
 
-    SDropStbInternalMsg *pMsg = mndBuildDropStbMsg(pMnode, pVgroup, pStb);
+    SVDropStbReq *pMsg = mndBuildDropStbMsg(pMnode, pVgroup, pStb);
     if (pMsg == NULL) {
       sdbCancelFetch(pSdb, pIter);
       sdbRelease(pSdb, pVgroup);
@@ -413,7 +381,7 @@ static int32_t mndSetCreateStbUndoActions(SMnode *pMnode, STrans *pTrans, SDbObj
     STransAction action = {0};
     action.epSet = mndGetVgroupEpset(pMnode, pVgroup);
     action.pCont = pMsg;
-    action.contLen = sizeof(SDropStbInternalMsg);
+    action.contLen = sizeof(SVDropStbReq);
     action.msgType = TDMT_VND_DROP_STB;
     if (mndTransAppendUndoAction(pTrans, &action) != 0) {
       free(pMsg);

source/dnode/mnode/impl/src/mndTopic.c

@@ -33,10 +33,7 @@ static int32_t  mndTopicActionInsert(SSdb *pSdb, STopicObj *pTopic);
 static int32_t  mndTopicActionDelete(SSdb *pSdb, STopicObj *pTopic);
 static int32_t  mndTopicActionUpdate(SSdb *pSdb, STopicObj *pTopic, STopicObj *pNewTopic);
 static int32_t  mndProcessCreateTopicMsg(SMnodeMsg *pMsg);
-static int32_t  mndProcessAlterTopicMsg(SMnodeMsg *pMsg);
 static int32_t  mndProcessDropTopicMsg(SMnodeMsg *pMsg);
-static int32_t  mndProcessCreateTopicInRsp(SMnodeMsg *pMsg);
-static int32_t  mndProcessAlterTopicInRsp(SMnodeMsg *pMsg);
 static int32_t  mndProcessDropTopicInRsp(SMnodeMsg *pMsg);
 static int32_t  mndProcessTopicMetaMsg(SMnodeMsg *pMsg);
 static int32_t  mndGetTopicMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
@@ -53,19 +50,8 @@ int32_t mndInitTopic(SMnode *pMnode) {
                      .deleteFp = (SdbDeleteFp)mndTopicActionDelete};
 
   mndSetMsgHandle(pMnode, TDMT_MND_CREATE_TOPIC, mndProcessCreateTopicMsg);
-  mndSetMsgHandle(pMnode, TDMT_MND_ALTER_TOPIC, mndProcessAlterTopicMsg);
   mndSetMsgHandle(pMnode, TDMT_MND_DROP_TOPIC, mndProcessDropTopicMsg);
-  mndSetMsgHandle(pMnode, TDMT_VND_CREATE_TOPIC_RSP, mndProcessCreateTopicInRsp);
-  mndSetMsgHandle(pMnode, TDMT_VND_ALTER_TOPIC_RSP, mndProcessAlterTopicInRsp);
   mndSetMsgHandle(pMnode, TDMT_VND_DROP_TOPIC_RSP, mndProcessDropTopicInRsp);
-  mndSetMsgHandle(pMnode, TDMT_VND_TABLE_META, mndProcessTopicMetaMsg);
-
-  /*mndAddShowMetaHandle(pMnode, TSDB_MGMT_TOPIC, mndGetTopicMeta);*/
-  /*mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TOPIC, mndRetrieveTopic);*/
-  /*mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TOPIC, mndCancelGetNextTopic);*/
-
-  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_TOPIC, mndProcessCreateTopicMsg);
-  mndSetMsgHandle(pMnode, TDMT_VND_CREATE_TOPIC_RSP, mndProcessCreateTopicInRsp);
 
   return sdbSetTable(pMnode->pSdb, table);
 }
@@ -145,24 +131,9 @@ static int32_t mndTopicActionUpdate(SSdb *pSdb, STopicObj *pOldTopic, STopicObj
   atomic_exchange_32(&pOldTopic->version, pNewTopic->version);
 
   taosWLockLatch(&pOldTopic->lock);
-#if 0
   
-  pOldTopic->numOfColumns = pNewTopic->numOfColumns;
-  pOldTopic->numOfTags = pNewTopic->numOfTags;
-  int32_t totalCols = pNewTopic->numOfTags + pNewTopic->numOfColumns;
-  int32_t totalSize = totalCols * sizeof(SSchema);
+  //TODO handle update
 
-  if (pOldTopic->numOfTags + pOldTopic->numOfColumns < totalCols) {
-    void *pSchema = malloc(totalSize);
-    if (pSchema != NULL) {
-      free(pOldTopic->pSchema);
-      pOldTopic->pSchema = pSchema;
-    }
-  }
-
-  memcpy(pOldTopic->pSchema, pNewTopic->pSchema, totalSize);
-
-#endif
   taosWUnLockLatch(&pOldTopic->lock);
   return 0;
 }
@@ -191,45 +162,10 @@ static SDbObj *mndAcquireDbByTopic(SMnode *pMnode, char *topicName) {
   return mndAcquireDb(pMnode, db);
 }
 
-static SCreateTopicInternalMsg *mndBuildCreateTopicMsg(SMnode *pMnode, SVgObj *pVgroup, STopicObj *pTopic) {
-  int32_t totalCols = 0;
-  int32_t contLen = sizeof(SCreateTopicInternalMsg) + pTopic->execLen + pTopic->sqlLen;
+static SDDropTopicMsg *mndBuildDropTopicMsg(SMnode *pMnode, SVgObj *pVgroup, STopicObj *pTopic) {
+  int32_t contLen = sizeof(SDDropTopicMsg);
 
-  SCreateTopicInternalMsg *pCreate = calloc(1, contLen);
-  if (pCreate == NULL) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  }
-
-  pCreate->head.contLen = htonl(contLen);
-  pCreate->head.vgId = htonl(pVgroup->vgId);
-  memcpy(pCreate->name, pTopic->name, TSDB_TABLE_FNAME_LEN);
-  pCreate->tuid = htobe64(pTopic->uid);
-  pCreate->sverson = htonl(pTopic->version);
-
-  pCreate->sql = malloc(pTopic->sqlLen);
-  if (pCreate->sql == NULL) {
-    free(pCreate);
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  }
-  memcpy(pCreate->sql, pTopic->sql, pTopic->sqlLen);
-
-  pCreate->executor = malloc(pTopic->execLen);
-  if (pCreate->executor == NULL) {
-    free(pCreate);
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  }
-  memcpy(pCreate->executor, pTopic->executor, pTopic->execLen);
-
-  return pCreate;
-}
-
-static SDropTopicInternalMsg *mndBuildDropTopicMsg(SMnode *pMnode, SVgObj *pVgroup, STopicObj *pTopic) {
-  int32_t contLen = sizeof(SDropTopicInternalMsg);
-
-  SDropTopicInternalMsg *pDrop = calloc(1, contLen);
+  SDDropTopicMsg *pDrop = calloc(1, contLen);
   if (pDrop == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
@@ -243,109 +179,12 @@ static SDropTopicInternalMsg *mndBuildDropTopicMsg(SMnode *pMnode, SVgObj *pVgro
   return pDrop;
 }
 
-static int32_t mndCheckCreateTopicMsg(SCreateTopicMsg *pCreate) {
+static int32_t mndCheckCreateTopicMsg(SCMCreateTopicReq *pCreate) {
   // deserialize and other stuff
   return 0;
 }
 
-static int32_t mndSetCreateTopicRedoLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, STopicObj *pTopic) {
-  SSdbRaw *pRedoRaw = mndTopicActionEncode(pTopic);
-  if (pRedoRaw == NULL) return -1;
-  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
-  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
-
-  return 0;
-}
-
-static int32_t mndSetCreateTopicUndoLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, STopicObj *pTopic) {
-  SSdbRaw *pUndoRaw = mndTopicActionEncode(pTopic);
-  if (pUndoRaw == NULL) return -1;
-  if (mndTransAppendUndolog(pTrans, pUndoRaw) != 0) return -1;
-  if (sdbSetRawStatus(pUndoRaw, SDB_STATUS_DROPPED) != 0) return -1;
-
-  return 0;
-}
-
-static int32_t mndSetCreateTopicCommitLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, STopicObj *pTopic) {
-  SSdbRaw *pCommitRaw = mndTopicActionEncode(pTopic);
-  if (pCommitRaw == NULL) return -1;
-  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
-  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
-
-  return 0;
-}
-
-static int32_t mndSetCreateTopicRedoActions(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, STopicObj *pTopic) {
-  SSdb   *pSdb = pMnode->pSdb;
-  SVgObj *pVgroup = NULL;
-  void   *pIter = NULL;
-
-  while (1) {
-    pIter = sdbFetch(pSdb, SDB_VGROUP, pIter, (void **)&pVgroup);
-    if (pIter == NULL) break;
-    if (pVgroup->dbUid != pDb->uid) continue;
-
-    SCreateTopicInternalMsg *pMsg = mndBuildCreateTopicMsg(pMnode, pVgroup, pTopic);
-    if (pMsg == NULL) {
-      sdbCancelFetch(pSdb, pIter);
-      sdbRelease(pSdb, pVgroup);
-      terrno = TSDB_CODE_OUT_OF_MEMORY;
-      return -1;
-    }
-
-    STransAction action = {0};
-    action.epSet = mndGetVgroupEpset(pMnode, pVgroup);
-    action.pCont = pMsg;
-    action.contLen = htonl(pMsg->head.contLen);
-    action.msgType = TDMT_VND_CREATE_TOPIC;
-    if (mndTransAppendRedoAction(pTrans, &action) != 0) {
-      free(pMsg);
-      sdbCancelFetch(pSdb, pIter);
-      sdbRelease(pSdb, pVgroup);
-      return -1;
-    }
-    sdbRelease(pSdb, pVgroup);
-  }
-
-  return 0;
-}
-
-static int32_t mndSetCreateTopicUndoActions(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, STopicObj *pTopic) {
-  SSdb   *pSdb = pMnode->pSdb;
-  SVgObj *pVgroup = NULL;
-  void   *pIter = NULL;
-
-  while (1) {
-    pIter = sdbFetch(pSdb, SDB_VGROUP, pIter, (void **)&pVgroup);
-    if (pIter == NULL) break;
-    if (pVgroup->dbUid != pDb->uid) continue;
-
-    SDropTopicInternalMsg *pMsg = mndBuildDropTopicMsg(pMnode, pVgroup, pTopic);
-    if (pMsg == NULL) {
-      sdbCancelFetch(pSdb, pIter);
-      sdbRelease(pSdb, pVgroup);
-      terrno = TSDB_CODE_OUT_OF_MEMORY;
-      return -1;
-    }
-
-    STransAction action = {0};
-    action.epSet = mndGetVgroupEpset(pMnode, pVgroup);
-    action.pCont = pMsg;
-    action.contLen = sizeof(SDropTopicInternalMsg);
-    action.msgType = TDMT_VND_DROP_TOPIC;
-    if (mndTransAppendUndoAction(pTrans, &action) != 0) {
-      free(pMsg);
-      sdbCancelFetch(pSdb, pIter);
-      sdbRelease(pSdb, pVgroup);
-      return -1;
-    }
-    sdbRelease(pSdb, pVgroup);
-  }
-
-  return 0;
-}
-
-static int32_t mndCreateTopic(SMnode *pMnode, SMnodeMsg *pMsg, SCreateTopicMsg *pCreate, SDbObj *pDb) {
+static int32_t mndCreateTopic(SMnode *pMnode, SMnodeMsg *pMsg, SCMCreateTopicReq *pCreate, SDbObj *pDb) {
   STopicObj topicObj = {0};
   tstrncpy(topicObj.name, pCreate->name, TSDB_TABLE_FNAME_LEN);
   tstrncpy(topicObj.db, pDb->name, TSDB_DB_FNAME_LEN);
@@ -355,66 +194,17 @@ static int32_t mndCreateTopic(SMnode *pMnode, SMnodeMsg *pMsg, SCreateTopicMsg *
   topicObj.dbUid = pDb->uid;
   topicObj.version = 1;
 
-#if 0
-  int32_t totalCols = topicObj.numOfColumns + topicObj.numOfTags;
-  int32_t totalSize = totalCols * sizeof(SSchema);
-  topicObj.sql = malloc(totalSize);
-  if (topicObj.sql == NULL) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return -1;
-  }
-  memcpy(topicObj.sql, pCreate->sql, totalSize);
-#endif
-
-  int32_t code = 0;
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, &pMsg->rpcMsg);
-  if (pTrans == NULL) {
-    mError("topic:%s, failed to create since %s", pCreate->name, terrstr());
-    return -1;
-  }
-  mDebug("trans:%d, used to create topic:%s", pTrans->id, pCreate->name);
-
-  if (mndSetCreateTopicRedoLogs(pMnode, pTrans, pDb, &topicObj) != 0) {
-    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
-    goto CREATE_TOPIC_OVER;
-  }
-
-  if (mndSetCreateTopicUndoLogs(pMnode, pTrans, pDb, &topicObj) != 0) {
-    mError("trans:%d, failed to set undo log since %s", pTrans->id, terrstr());
-    goto CREATE_TOPIC_OVER;
-  }
-
-  if (mndSetCreateTopicCommitLogs(pMnode, pTrans, pDb, &topicObj) != 0) {
-    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
-    goto CREATE_TOPIC_OVER;
-  }
-
-  if (mndSetCreateTopicRedoActions(pMnode, pTrans, pDb, &topicObj) != 0) {
-    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
-    goto CREATE_TOPIC_OVER;
-  }
-
-  if (mndSetCreateTopicUndoActions(pMnode, pTrans, pDb, &topicObj) != 0) {
-    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
-    goto CREATE_TOPIC_OVER;
-  }
-
-  if (mndTransPrepare(pMnode, pTrans) != 0) {
-    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
-  }
-
-  code = 0;
-
-CREATE_TOPIC_OVER:
-  mndTransDrop(pTrans);
-  return code;
+  SSdbRaw *pTopicRaw = mndTopicActionEncode(&topicObj);
+  if (pTopicRaw == NULL) return -1;
+  if (sdbSetRawStatus(pTopicRaw, SDB_STATUS_READY) != 0) return -1;
+  return sdbWrite(pMnode->pSdb, pTopicRaw);
 }
 
 static int32_t mndProcessCreateTopicMsg(SMnodeMsg *pMsg) {
   SMnode        *pMnode = pMsg->pMnode;
-  SCreateTopicMsg *pCreate = pMsg->rpcMsg.pCont;
+  char *msgStr = pMsg->rpcMsg.pCont;
+  SCMCreateTopicReq* pCreate;
+  tDeserializeSCMCreateTopicReq(msgStr, pCreate);
 
   mDebug("topic:%s, start to create", pCreate->name);
 
@@ -436,15 +226,6 @@ static int32_t mndProcessCreateTopicMsg(SMnodeMsg *pMsg) {
     }
   }
 
-  // topic should have different name with stb
-  SStbObj *pStb = mndAcquireStb(pMnode, pCreate->name);
-  if (pStb != NULL) {
-    sdbRelease(pMnode->pSdb, pStb);
-    terrno = TSDB_CODE_MND_NAME_CONFLICT_WITH_STB;
-    mError("topic:%s, failed to create since %s", pCreate->name, terrstr());
-    return -1;
-  }
-
   SDbObj *pDb = mndAcquireDbByTopic(pMnode, pCreate->name);
   if (pDb == NULL) {
     terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
@@ -464,144 +245,7 @@ static int32_t mndProcessCreateTopicMsg(SMnodeMsg *pMsg) {
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
-static int32_t mndCheckAlterTopicMsg(SAlterTopicMsg *pAlter) {
-  SSchema *pSchema = &pAlter->schema;
-  pSchema->colId = htonl(pSchema->colId);
-  pSchema->bytes = htonl(pSchema->bytes);
-
-  if (pSchema->type <= 0) {
-    terrno = TSDB_CODE_MND_INVALID_TOPIC_OPTION;
-    return -1;
-  }
-  if (pSchema->colId < 0 || pSchema->colId >= (TSDB_MAX_COLUMNS + TSDB_MAX_TAGS)) {
-    terrno = TSDB_CODE_MND_INVALID_TOPIC_OPTION;
-    return -1;
-  }
-  if (pSchema->bytes <= 0) {
-    terrno = TSDB_CODE_MND_INVALID_TOPIC_OPTION;
-    return -1;
-  }
-  if (pSchema->name[0] == 0) {
-    terrno = TSDB_CODE_MND_INVALID_TOPIC_OPTION;
-    return -1;
-  }
-
-  return 0;
-}
-
-static int32_t mndUpdateTopic(SMnode *pMnode, SMnodeMsg *pMsg, STopicObj *pOldTopic, STopicObj *pNewTopic) { return 0; }
-
-static int32_t mndProcessAlterTopicMsg(SMnodeMsg *pMsg) {
-  SMnode         *pMnode = pMsg->pMnode;
-  SAlterTopicMsg *pAlter = pMsg->rpcMsg.pCont;
-
-  mDebug("topic:%s, start to alter", pAlter->name);
-
-  if (mndCheckAlterTopicMsg(pAlter) != 0) {
-    mError("topic:%s, failed to alter since %s", pAlter->name, terrstr());
-    return -1;
-  }
-
-  STopicObj *pTopic = mndAcquireTopic(pMnode, pAlter->name);
-  if (pTopic == NULL) {
-    terrno = TSDB_CODE_MND_TOPIC_NOT_EXIST;
-    mError("topic:%s, failed to alter since %s", pAlter->name, terrstr());
-    return -1;
-  }
-
-  STopicObj topicObj = {0};
-  memcpy(&topicObj, pTopic, sizeof(STopicObj));
-
-  int32_t code = mndUpdateTopic(pMnode, pMsg, pTopic, &topicObj);
-  mndReleaseTopic(pMnode, pTopic);
-
-  if (code != 0) {
-    mError("topic:%s, failed to alter since %s", pAlter->name, tstrerror(code));
-    return code;
-  }
-
-  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
-}
-
-static int32_t mndProcessAlterTopicInRsp(SMnodeMsg *pMsg) {
-  mndTransProcessRsp(pMsg);
-  return 0;
-}
-
-static int32_t mndSetDropTopicRedoLogs(SMnode *pMnode, STrans *pTrans, STopicObj *pTopic) {
-  SSdbRaw *pRedoRaw = mndTopicActionEncode(pTopic);
-  if (pRedoRaw == NULL) return -1;
-  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
-  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
-
-  return 0;
-}
-
-static int32_t mndSetDropTopicUndoLogs(SMnode *pMnode, STrans *pTrans, STopicObj *pTopic) {
-  SSdbRaw *pUndoRaw = mndTopicActionEncode(pTopic);
-  if (pUndoRaw == NULL) return -1;
-  if (mndTransAppendUndolog(pTrans, pUndoRaw) != 0) return -1;
-  if (sdbSetRawStatus(pUndoRaw, SDB_STATUS_READY) != 0) return -1;
-
-  return 0;
-}
-
-static int32_t mndSetDropTopicCommitLogs(SMnode *pMnode, STrans *pTrans, STopicObj *pTopic) {
-  SSdbRaw *pCommitRaw = mndTopicActionEncode(pTopic);
-  if (pCommitRaw == NULL) return -1;
-  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
-  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
-
-  return 0;
-}
-
-static int32_t mndSetDropTopicRedoActions(SMnode *pMnode, STrans *pTrans, STopicObj *pTopic) { return 0; }
-
-static int32_t mndSetDropTopicUndoActions(SMnode *pMnode, STrans *pTrans, STopicObj *pTopic) { return 0; }
-
 static int32_t mndDropTopic(SMnode *pMnode, SMnodeMsg *pMsg, STopicObj *pTopic) {
-  int32_t code = -1;
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, &pMsg->rpcMsg);
-  if (pTrans == NULL) {
-    mError("topic:%s, failed to drop since %s", pTopic->name, terrstr());
-    return -1;
-  }
-  mDebug("trans:%d, used to drop topic:%s", pTrans->id, pTopic->name);
-
-  if (mndSetDropTopicRedoLogs(pMnode, pTrans, pTopic) != 0) {
-    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
-    goto DROP_TOPIC_OVER;
-  }
-
-  if (mndSetDropTopicUndoLogs(pMnode, pTrans, pTopic) != 0) {
-    mError("trans:%d, failed to set undo log since %s", pTrans->id, terrstr());
-    goto DROP_TOPIC_OVER;
-  }
-
-  if (mndSetDropTopicCommitLogs(pMnode, pTrans, pTopic) != 0) {
-    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
-    goto DROP_TOPIC_OVER;
-  }
-
-  if (mndSetDropTopicRedoActions(pMnode, pTrans, pTopic) != 0) {
-    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
-    goto DROP_TOPIC_OVER;
-  }
-
-  if (mndSetDropTopicUndoActions(pMnode, pTrans, pTopic) != 0) {
-    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
-    goto DROP_TOPIC_OVER;
-  }
-
-  if (mndTransPrepare(pMnode, pTrans) != 0) {
-    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
-    goto DROP_TOPIC_OVER;
-  }
-
-  code = 0;
-
-DROP_TOPIC_OVER:
-  mndTransDrop(pTrans);
   return 0;
 }
 
@@ -705,11 +349,6 @@ static int32_t mndProcessTopicMetaMsg(SMnodeMsg *pMsg) {
   return 0;
 }
 
-static int32_t mndProcessCreateTopicInRsp(SMnodeMsg *pMsg) {
-  mndTransProcessRsp(pMsg);
-  return 0;
-}
-
 static int32_t mndGetNumOfTopics(SMnode *pMnode, char *dbName, int32_t *pNumOfTopics) {
   SSdb *pSdb = pMnode->pSdb;
 

source/dnode/mnode/impl/src/mndTrans.c

@@ -294,18 +294,18 @@ TRANS_DECODE_OVER:
     return NULL;
   }
 
-  mTrace("trans:%d, decode from raw:%p", pTrans->id, pRaw);
+  mTrace("trans:%d, decode from raw:%p, data:%p", pTrans->id, pRaw, pTrans);
   return pRow;
 }
 
 static int32_t mndTransActionInsert(SSdb *pSdb, STrans *pTrans) {
   pTrans->stage = TRN_STAGE_PREPARE;
-  mTrace("trans:%d, perform insert action", pTrans->id);
+  mTrace("trans:%d, perform insert action, data:%p", pTrans->id, pTrans);
   return 0;
 }
 
 static int32_t mndTransActionDelete(SSdb *pSdb, STrans *pTrans) {
-  mTrace("trans:%d, perform delete action", pTrans->id);
+  mTrace("trans:%d, perform delete action, data:%p", pTrans->id, pTrans);
 
   mndTransDropLogs(pTrans->redoLogs);
   mndTransDropLogs(pTrans->undoLogs);
@@ -317,7 +317,7 @@ static int32_t mndTransActionDelete(SSdb *pSdb, STrans *pTrans) {
 }
 
 static int32_t mndTransActionUpdate(SSdb *pSdb, STrans *pOldTrans, STrans *pNewTrans) {
-  mTrace("trans:%d, perform update action", pOldTrans->id);
+  mTrace("trans:%d, perform update action, data:%p", pOldTrans->id, pOldTrans);
   pOldTrans->stage = pNewTrans->stage;
   return 0;
 }
@@ -362,14 +362,14 @@ STrans *mndTransCreate(SMnode *pMnode, ETrnPolicy policy, SRpcMsg *pMsg) {
     return NULL;
   }
 
-  mDebug("trans:%d, is created", pTrans->id);
+  mDebug("trans:%d, is created, data:%p", pTrans->id, pTrans);
   return pTrans;
 }
 
 static void mndTransDropLogs(SArray *pArray) {
   for (int32_t i = 0; i < pArray->size; ++i) {
     SSdbRaw *pRaw = taosArrayGetP(pArray, i);
-    tfree(pRaw);
+    sdbFreeRaw(pRaw);
   }
 
   taosArrayDestroy(pArray);
@@ -391,7 +391,7 @@ void mndTransDrop(STrans *pTrans) {
   mndTransDropActions(pTrans->redoActions);
   mndTransDropActions(pTrans->undoActions);
 
-  // mDebug("trans:%d, is dropped, data:%p", pTrans->id, pTrans);
+  mDebug("trans:%d, is dropped, data:%p", pTrans->id, pTrans);
   tfree(pTrans);
 }
 
@@ -442,7 +442,7 @@ static int32_t mndTransSync(SMnode *pMnode, STrans *pTrans) {
   }
   sdbSetRawStatus(pRaw, SDB_STATUS_READY);
 
-  mTrace("trans:%d, sync to other nodes", pTrans->id);
+  mDebug("trans:%d, sync to other nodes", pTrans->id);
   int32_t code = mndSyncPropose(pMnode, pRaw);
   if (code != 0) {
     mError("trans:%d, failed to sync since %s", pTrans->id, terrstr());
@@ -450,7 +450,7 @@ static int32_t mndTransSync(SMnode *pMnode, STrans *pTrans) {
     return -1;
   }
 
-  mTrace("trans:%d, sync finished", pTrans->id);
+  mDebug("trans:%d, sync finished", pTrans->id);
 
   code = sdbWrite(pMnode->pSdb, pRaw);
   if (code != 0) {

source/dnode/mnode/impl/src/mndVgroup.c

@@ -86,7 +86,6 @@ SSdbRaw *mndVgroupActionEncode(SVgObj *pVgroup) {
   for (int8_t i = 0; i < pVgroup->replica; ++i) {
     SVnodeGid *pVgid = &pVgroup->vnodeGid[i];
     SDB_SET_INT32(pRaw, dataPos, pVgid->dnodeId)
-    SDB_SET_INT8(pRaw, dataPos, pVgid->role)
   }
   SDB_SET_RESERVE(pRaw, dataPos, TSDB_VGROUP_RESERVE_SIZE)
   SDB_SET_DATALEN(pRaw, dataPos);
@@ -121,7 +120,9 @@ SSdbRow *mndVgroupActionDecode(SSdbRaw *pRaw) {
   for (int8_t i = 0; i < pVgroup->replica; ++i) {
     SVnodeGid *pVgid = &pVgroup->vnodeGid[i];
     SDB_GET_INT32(pRaw, pRow, dataPos, &pVgid->dnodeId)
-    SDB_GET_INT8(pRaw, pRow, dataPos, (int8_t *)&pVgid->role)
+    if (pVgroup->replica == 1) {
+      pVgid->role = TAOS_SYNC_STATE_LEADER;
+    }
   }
   SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_VGROUP_RESERVE_SIZE)
 
@@ -237,44 +238,95 @@ SDropVnodeMsg *mndBuildDropVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *p
   return pDrop;
 }
 
-static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup) {
+static SArray *mndBuildDnodesArray(SMnode *pMnode) {
   SSdb   *pSdb = pMnode->pSdb;
-  int32_t allocedVnodes = 0;
-  void   *pIter = NULL;
+  int32_t numOfDnodes = mndGetDnodeSize(pMnode);
+  SArray *pArray = taosArrayInit(numOfDnodes, sizeof(SDnodeObj));
+  if (pArray == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
 
-  while (allocedVnodes < pVgroup->replica) {
+  void *pIter = NULL;
+  while (1) {
     SDnodeObj *pDnode = NULL;
     pIter = sdbFetch(pSdb, SDB_DNODE, pIter, (void **)&pDnode);
     if (pIter == NULL) break;
 
-    // todo
-    if (mndIsDnodeInReadyStatus(pMnode, pDnode)) {
-      SVnodeGid *pVgid = &pVgroup->vnodeGid[allocedVnodes];
+    int32_t numOfVnodes = mndGetVnodesNum(pMnode, pDnode->id);
+
+    bool isMnode = mndIsMnode(pMnode, pDnode->id);
+    if (isMnode) {
+      pDnode->numOfVnodes++;
+    }
+
+    bool isReady = mndIsDnodeInReadyStatus(pMnode, pDnode);
+    if (isReady) {
+      taosArrayPush(pArray, pDnode);
+    }
+
+    mDebug("dnode:%d, numOfVnodes:%d numOfSupportVnodes:%d isMnode:%d ready:%d", pDnode->id, numOfVnodes,
+           pDnode->numOfSupportVnodes, isMnode, isReady);
+    sdbRelease(pSdb, pDnode);
+  }
+
+  return pArray;
+}
+
+static int32_t mndCompareDnodeVnodes(SDnodeObj *pDnode1, SDnodeObj *pDnode2) {
+  float d1Score = (float)pDnode1->numOfVnodes / pDnode1->numOfSupportVnodes;
+  float d2Score = (float)pDnode2->numOfVnodes / pDnode2->numOfSupportVnodes;
+  return d1Score > d2Score ? 1 : 0;
+}
+
+static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {
+  SSdb   *pSdb = pMnode->pSdb;
+  int32_t allocedVnodes = 0;
+  void   *pIter = NULL;
+
+  taosArraySort(pArray, (__compar_fn_t)mndCompareDnodeVnodes);
+
+  for (int32_t v = 0; v < pVgroup->replica; ++v) {
+    SVnodeGid *pVgid = &pVgroup->vnodeGid[v];
+    SDnodeObj *pDnode = taosArrayGet(pArray, v);
+    if (pDnode == NULL || pDnode->numOfVnodes > pDnode->numOfSupportVnodes) {
+      terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
+      return -1;
+    }
+
     pVgid->dnodeId = pDnode->id;
     if (pVgroup->replica == 1) {
       pVgid->role = TAOS_SYNC_STATE_LEADER;
     } else {
       pVgid->role = TAOS_SYNC_STATE_FOLLOWER;
     }
-      allocedVnodes++;
-    }
-    sdbRelease(pSdb, pDnode);
+
+    mDebug("db:%s, vgId:%d, vindex:%d dnodeId:%d is alloced", pVgroup->dbName, pVgroup->vgId, v, pVgid->dnodeId);
+    pDnode->numOfVnodes++;
   }
 
-  if (allocedVnodes != pVgroup->replica) {
-    terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
-    return -1;
-  }
   return 0;
 }
 
 int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
-  SVgObj *pVgroups = calloc(pDb->cfg.numOfVgroups, sizeof(SVgObj));
+  int32_t code = -1;
+  SArray *pArray = NULL;
+  SVgObj *pVgroups = NULL;
+
+  pVgroups = calloc(pDb->cfg.numOfVgroups, sizeof(SVgObj));
   if (pVgroups == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return -1;
+    goto ALLOC_VGROUP_OVER;
   }
 
+  pArray = mndBuildDnodesArray(pMnode);
+  if (pArray == NULL) {
+    goto ALLOC_VGROUP_OVER;
+  }
+
+  mDebug("db:%s, total %d dnodes used to create %d vgroups (%d vnodes)", pDb->name, (int32_t)taosArrayGetSize(pArray),
+         pDb->cfg.numOfVgroups, pDb->cfg.numOfVgroups * pDb->cfg.replications);
+
   int32_t  allocedVgroups = 0;
   int32_t  maxVgId = sdbGetMaxId(pMnode->pSdb, SDB_VGROUP);
   uint32_t hashMin = 0;
@@ -298,17 +350,23 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
     pVgroup->dbUid = pDb->uid;
     pVgroup->replica = pDb->cfg.replications;
 
-    if (mndGetAvailableDnode(pMnode, pVgroup) != 0) {
+    if (mndGetAvailableDnode(pMnode, pVgroup, pArray) != 0) {
       terrno = TSDB_CODE_MND_NO_ENOUGH_DNODES;
-      free(pVgroups);
-      return -1;
+      goto ALLOC_VGROUP_OVER;
     }
 
     allocedVgroups++;
   }
 
   *ppVgroups = pVgroups;
-  return 0;
+  code = 0;
+
+  mDebug("db:%s, %d vgroups is alloced, replica:%d", pDb->name, pDb->cfg.numOfVgroups, pDb->cfg.replications);
+
+ALLOC_VGROUP_OVER:
+  if (code != 0) free(pVgroups);
+  taosArrayDestroy(pArray);
+  return code;
 }
 
 SEpSet mndGetVgroupEpset(SMnode *pMnode, SVgObj *pVgroup) {
@@ -348,6 +406,7 @@ static int32_t mndProcessDropVnodeRsp(SMnodeMsg *pMsg) {
 }
 
 static int32_t mndProcessSyncVnodeRsp(SMnodeMsg *pMsg) { return 0; }
+
 static int32_t mndProcessCompactVnodeRsp(SMnodeMsg *pMsg) { return 0; }
 
 static int32_t mndGetVgroupMaxReplica(SMnode *pMnode, char *dbName, int8_t *pReplica, int32_t *pNumOfVgroups) {
@@ -478,7 +537,7 @@ static void mndCancelGetNextVgroup(SMnode *pMnode, void *pIter) {
   sdbCancelFetch(pSdb, pIter);
 }
 
-static int32_t mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId) {
+int32_t mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId) {
   SSdb   *pSdb = pMnode->pSdb;
   int32_t numOfVnodes = 0;
   void   *pIter = NULL;

source/dnode/mnode/impl/src/mnode.c

@@ -16,14 +16,16 @@
 #define _DEFAULT_SOURCE
 #include "mndAcct.h"
 #include "mndAuth.h"
-#include "mndBalance.h"
+#include "mndBnode.h"
 #include "mndCluster.h"
 #include "mndDb.h"
 #include "mndDnode.h"
 #include "mndFunc.h"
 #include "mndMnode.h"
 #include "mndProfile.h"
+#include "mndQnode.h"
 #include "mndShow.h"
+#include "mndSnode.h"
 #include "mndStb.h"
 #include "mndSync.h"
 #include "mndTelem.h"
@@ -146,15 +148,18 @@ static int32_t mndInitSteps(SMnode *pMnode) {
   if (mndAllocStep(pMnode, "mnode-sdb", mndInitSdb, mndCleanupSdb) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-trans", mndInitTrans, mndCleanupTrans) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-cluster", mndInitCluster, mndCleanupCluster) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-dnode", mndInitDnode, mndCleanupDnode) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-mnode", mndInitMnode, mndCleanupMnode) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-acct", mndInitAcct, mndCleanupAcct) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-auth", mndInitAuth, mndCleanupAuth) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-qnode", mndInitQnode, mndCleanupQnode) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-qnode", mndInitSnode, mndCleanupSnode) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-qnode", mndInitBnode, mndCleanupBnode) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-dnode", mndInitDnode, mndCleanupDnode) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-user", mndInitUser, mndCleanupUser) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-db", mndInitDb, mndCleanupDb) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-auth", mndInitAuth, mndCleanupAuth) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-acct", mndInitAcct, mndCleanupAcct) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-topic", mndInitTopic, mndCleanupTopic) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-vgroup", mndInitVgroup, mndCleanupVgroup) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-stb", mndInitStb, mndCleanupStb) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-topic", mndInitTopic, mndCleanupTopic) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-db", mndInitDb, mndCleanupDb) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-func", mndInitFunc, mndCleanupFunc) != 0) return -1;
   if (pMnode->clusterId <= 0) {
     if (mndAllocStep(pMnode, "mnode-sdb-deploy", mndDeploySdb, NULL) != 0) return -1;
@@ -162,7 +167,6 @@ static int32_t mndInitSteps(SMnode *pMnode) {
     if (mndAllocStep(pMnode, "mnode-sdb-read", mndReadSdb, NULL) != 0) return -1;
   }
   if (mndAllocStep(pMnode, "mnode-timer", mndInitTimer, NULL) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-balance", mndInitBalance, mndCleanupBalance) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-profile", mndInitProfile, mndCleanupProfile) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-show", mndInitShow, mndCleanupShow) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-sync", mndInitSync, mndCleanupSync) != 0) return -1;

source/dnode/mnode/sdb/inc/sdbInt.h

@@ -72,6 +72,7 @@ typedef struct SSdb {
 } SSdb;
 
 int32_t sdbWriteFile(SSdb *pSdb);
+void    sdbPrintOper(SSdb *pSdb, SSdbRow *pRow, const char *oper);
 
 #ifdef __cplusplus
 }

source/dnode/mnode/sdb/src/sdb.c

@@ -80,16 +80,12 @@ void sdbCleanup(SSdb *pSdb) {
     SHashObj *hash = pSdb->hashObjs[i];
     if (hash == NULL) continue;
 
-    SdbDeleteFp deleteFp = pSdb->deleteFps[i];
     SSdbRow **ppRow = taosHashIterate(hash, NULL);
     while (ppRow != NULL) {
       SSdbRow *pRow = *ppRow;
       if (pRow == NULL) continue;
 
-      if (deleteFp != NULL) {
-        (*deleteFp)(pSdb, pRow->pObj);
-      }
-      sdbFreeRow(pRow);
+      sdbFreeRow(pSdb, pRow);
       ppRow = taosHashIterate(hash, ppRow);
     }
   }

source/dnode/mnode/sdb/src/sdbFile.c

@@ -151,7 +151,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
         if (taosWriteFile(fd, pRaw, writeLen) != writeLen) {
           code = TAOS_SYSTEM_ERROR(terrno);
           taosHashCancelIterate(hash, ppRow);
-          free(pRaw);
+          sdbFreeRaw(pRaw);
           break;
         }
 
@@ -159,7 +159,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
         if (taosWriteFile(fd, &cksum, sizeof(int32_t)) != sizeof(int32_t)) {
           code = TAOS_SYSTEM_ERROR(terrno);
           taosHashCancelIterate(hash, ppRow);
-          free(pRaw);
+          sdbFreeRaw(pRaw);
           break;
         }
       } else {
@@ -168,7 +168,7 @@ int32_t sdbWriteFile(SSdb *pSdb) {
         break;
       }
 
-      free(pRaw);
+      sdbFreeRaw(pRaw);
       ppRow = taosHashIterate(hash, ppRow);
     }
     taosWUnLockLatch(pLock);

source/dnode/mnode/sdb/src/sdbHash.c

@@ -16,6 +16,57 @@
 #define _DEFAULT_SOURCE
 #include "sdbInt.h"
 
+static const char *sdbTableName(ESdbType type) {
+  switch (type) {
+    case SDB_TRANS:
+      return "trans";
+    case SDB_CLUSTER:
+      return "cluster";
+    case SDB_MNODE:
+      return "mnode";
+    case SDB_QNODE:
+      return "qnode";
+    case SDB_SNODE:
+      return "snode";
+    case SDB_BNODE:
+      return "bnode";
+    case SDB_DNODE:
+      return "dnode";
+    case SDB_USER:
+      return "user";
+    case SDB_AUTH:
+      return "auth";
+    case SDB_ACCT:
+      return "acct";
+    case SDB_TOPIC:
+      return "topic";
+    case SDB_VGROUP:
+      return "vgId";
+    case SDB_STB:
+      return "stb";
+    case SDB_DB:
+      return "db";
+    case SDB_FUNC:
+      return "func";
+    default:
+      return "undefine";
+  }
+}
+
+void sdbPrintOper(SSdb *pSdb, SSdbRow *pRow, const char *oper) {
+  EKeyType keyType = pSdb->keyTypes[pRow->type];
+
+  if (keyType == SDB_KEY_BINARY) {
+    mTrace("%s:%s, refCount:%d oper:%s", sdbTableName(pRow->type), (char *)pRow->pObj, pRow->refCount, oper);
+  } else if (keyType == SDB_KEY_INT32) {
+    mTrace("%s:%d, refCount:%d oper:%s", sdbTableName(pRow->type), *(int32_t *)pRow->pObj, pRow->refCount, oper);
+  } else if (keyType == SDB_KEY_INT64) {
+    mTrace("%s:%" PRId64 ", refCount:%d oper:%s", sdbTableName(pRow->type), *(int64_t *)pRow->pObj, pRow->refCount,
+           oper);
+  } else {
+  }
+}
+
 static SHashObj *sdbGetHash(SSdb *pSdb, int32_t type) {
   if (type >= SDB_MAX || type <= SDB_START) {
     terrno = TSDB_CODE_SDB_INVALID_TABLE_TYPE;
@@ -55,17 +106,18 @@ static int32_t sdbInsertRow(SSdb *pSdb, SHashObj *hash, SSdbRaw *pRaw, SSdbRow *
   SSdbRow *pOldRow = taosHashGet(hash, pRow->pObj, keySize);
   if (pOldRow != NULL) {
     taosWUnLockLatch(pLock);
-    sdbFreeRow(pRow);
+    sdbFreeRow(pSdb, pRow);
     terrno = TSDB_CODE_SDB_OBJ_ALREADY_THERE;
     return terrno;
   }
 
-  pRow->refCount = 1;
+  pRow->refCount = 0;
   pRow->status = pRaw->status;
+  sdbPrintOper(pSdb, pRow, "insertRow");
 
   if (taosHashPut(hash, pRow->pObj, keySize, &pRow, sizeof(void *)) != 0) {
     taosWUnLockLatch(pLock);
-    sdbFreeRow(pRow);
+    sdbFreeRow(pSdb, pRow);
     terrno = TSDB_CODE_SDB_OBJ_ALREADY_THERE;
     return terrno;
   }
@@ -83,7 +135,7 @@ static int32_t sdbInsertRow(SSdb *pSdb, SHashObj *hash, SSdbRaw *pRaw, SSdbRow *
       taosWLockLatch(pLock);
       taosHashRemove(hash, pRow->pObj, keySize);
       taosWUnLockLatch(pLock);
-      sdbFreeRow(pRow);
+      sdbFreeRow(pSdb, pRow);
       terrno = code;
       return terrno;
     }
@@ -113,7 +165,7 @@ static int32_t sdbUpdateRow(SSdb *pSdb, SHashObj *hash, SSdbRaw *pRaw, SSdbRow *
     code = (*updateFp)(pSdb, pOldRow->pObj, pNewRow->pObj);
   }
 
-  sdbFreeRow(pNewRow);
+  sdbFreeRow(pSdb, pNewRow);
   return code;
 }
 
@@ -123,14 +175,10 @@ static int32_t sdbDeleteRow(SSdb *pSdb, SHashObj *hash, SSdbRaw *pRaw, SSdbRow *
   SRWLatch *pLock = &pSdb->locks[pRow->type];
   taosWLockLatch(pLock);
 
-  // remove attached object such as trans
-  SdbDeleteFp deleteFp = pSdb->deleteFps[pRow->type];
-  if (deleteFp != NULL) (*deleteFp)(pSdb, pRow->pObj);
-
   SSdbRow **ppOldRow = taosHashGet(hash, pRow->pObj, keySize);
   if (ppOldRow == NULL || *ppOldRow == NULL) {
     taosWUnLockLatch(pLock);
-    sdbFreeRow(pRow);
+    sdbFreeRow(pSdb, pRow);
     terrno = TSDB_CODE_SDB_OBJ_NOT_THERE;
     return terrno;
   }
@@ -140,8 +188,8 @@ static int32_t sdbDeleteRow(SSdb *pSdb, SHashObj *hash, SSdbRaw *pRaw, SSdbRow *
   taosHashRemove(hash, pOldRow->pObj, keySize);
   taosWUnLockLatch(pLock);
 
-  sdbRelease(pSdb, pOldRow->pObj);
-  sdbFreeRow(pRow);
+  // sdbRelease(pSdb, pOldRow->pObj);
+  sdbFreeRow(pSdb, pRow);
   return code;
 }
 
@@ -206,6 +254,7 @@ void *sdbAcquire(SSdb *pSdb, ESdbType type, void *pKey) {
     case SDB_STATUS_UPDATING:
       atomic_add_fetch_32(&pRow->refCount, 1);
       pRet = pRow->pObj;
+      sdbPrintOper(pSdb, pRow, "acquireRow");
       break;
     case SDB_STATUS_CREATING:
       terrno = TSDB_CODE_SDB_OBJ_CREATING;
@@ -232,13 +281,9 @@ void sdbRelease(SSdb *pSdb, void *pObj) {
   taosRLockLatch(pLock);
 
   int32_t ref = atomic_sub_fetch_32(&pRow->refCount, 1);
+  sdbPrintOper(pSdb, pRow, "releaseRow");
   if (ref <= 0 && pRow->status == SDB_STATUS_DROPPED) {
-    SdbDeleteFp deleteFp = pSdb->deleteFps[pRow->type];
-    if (deleteFp != NULL) {
-      (*deleteFp)(pSdb, pRow->pObj);
-    }
-
-    sdbFreeRow(pRow);
+    sdbFreeRow(pSdb, pRow);
   }
 
   taosRUnLockLatch(pLock);
@@ -255,9 +300,9 @@ void *sdbFetch(SSdb *pSdb, ESdbType type, void *pIter, void **ppObj) {
 
   if (pIter != NULL) {
     SSdbRow *pLastRow = *(SSdbRow **)pIter;
-    int32_t  ref = atomic_sub_fetch_32(&pLastRow->refCount, 1);
+    int32_t  ref = atomic_load_32(&pLastRow->refCount);
     if (ref <= 0 && pLastRow->status == SDB_STATUS_DROPPED) {
-      sdbFreeRow(pLastRow);
+      sdbFreeRow(pSdb, pLastRow);
     }
   }
 
@@ -270,6 +315,7 @@ void *sdbFetch(SSdb *pSdb, ESdbType type, void *pIter, void **ppObj) {
     }
 
     atomic_add_fetch_32(&pRow->refCount, 1);
+    sdbPrintOper(pSdb, pRow, "fetchRow");
     *ppObj = pRow->pObj;
     break;
   }

source/dnode/mnode/sdb/src/sdbRaw.c

@@ -27,12 +27,12 @@ SSdbRaw *sdbAllocRaw(ESdbType type, int8_t sver, int32_t dataLen) {
   pRaw->sver = sver;
   pRaw->dataLen = dataLen;
 
-  // mTrace("raw:%p, is created, len:%d", pRaw, dataLen);
+  mTrace("raw:%p, is created, len:%d", pRaw, dataLen);
   return pRaw;
 }
 
 void sdbFreeRaw(SSdbRaw *pRaw) {
-  // mTrace("raw:%p, is freed", pRaw);
+  mTrace("raw:%p, is freed", pRaw);
   free(pRaw);
 }
 

source/dnode/mnode/sdb/src/sdbRow.c

@@ -35,4 +35,13 @@ void *sdbGetRowObj(SSdbRow *pRow) {
   return pRow->pObj;
 }
 
-void sdbFreeRow(SSdbRow *pRow) { tfree(pRow); }
+void sdbFreeRow(SSdb *pSdb, SSdbRow *pRow) {
+  // remove attached object such as trans
+  SdbDeleteFp deleteFp = pSdb->deleteFps[pRow->type];
+  if (deleteFp != NULL) {
+    (*deleteFp)(pSdb, pRow->pObj);
+  }
+
+  sdbPrintOper(pSdb, pRow, "freeRow");
+  tfree(pRow);
+}

source/dnode/qnode/src/qnode.c

@@ -24,12 +24,8 @@ void qndClose(SQnode *pQnode) { free(pQnode); }
 
 int32_t qndGetLoad(SQnode *pQnode, SQnodeLoad *pLoad) { return 0; }
 
-int32_t qndProcessQueryReq(SQnode *pQnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+int32_t qndProcessMsg(SQnode *pQnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
   *pRsp = NULL;
   return 0;
 }
 
-int32_t qndProcessFetchReq(SQnode *pQnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
-  *pRsp = NULL;
-  return 0;
-}

source/dnode/snode/src/snode.c

@@ -24,7 +24,7 @@ void sndClose(SSnode *pSnode) { free(pSnode); }
 
 int32_t sndGetLoad(SSnode *pSnode, SSnodeLoad *pLoad) { return 0; }
 
-int32_t sndProcessWriteMsg(SSnode *pSnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+int32_t sndProcessMsg(SSnode *pSnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
   *pRsp = NULL;
   return 0;
 }

source/dnode/vnode/impl/src/vnodeInt.c

@@ -28,8 +28,3 @@ int vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
   vInfo("sync message is processed");
   return 0;
 }
-
-int vnodeProcessConsumeReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
-  vInfo("consume message is processed");
-  return 0;
-}

source/libs/catalog/inc/catalogInt.h

@@ -31,6 +31,11 @@ extern "C" {
 
 #define CTG_DEFAULT_INVALID_VERSION (-1)
 
+enum {
+  CTG_READ = 1,
+  CTG_WRITE,
+};
+
 typedef struct SVgroupListCache {
   int32_t vgroupVersion;
   SHashObj *cache;        // key:vgId, value:SVgroupInfo
@@ -41,6 +46,7 @@ typedef struct SDBVgroupCache {
 } SDBVgroupCache;
 
 typedef struct STableMetaCache {
+  SRWLatch  stableLock;
   SHashObj *cache;           //key:fulltablename, value:STableMeta
   SHashObj *stableCache;     //key:suid, value:STableMeta*
 } STableMetaCache;
@@ -71,6 +77,31 @@ typedef uint32_t (*tableNameHashFp)(const char *, uint32_t);
 #define CTG_ERR_LRET(c,...) do { int32_t _code = c; if (_code != TSDB_CODE_SUCCESS) { ctgError(__VA_ARGS__); terrno = _code; return _code; } } while (0)
 #define CTG_ERR_JRET(c) do { code = c; if (code != TSDB_CODE_SUCCESS) { terrno = code; goto _return; } } while (0)
 
+#define CTG_LOCK(type, _lock) do {   \
+  if (CTG_READ == (type)) {          \
+    if ((*(_lock)) < 0) assert(0);  \
+    taosRLockLatch(_lock);           \
+    ctgDebug("CTG RLOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
+  } else {                                                \
+    if ((*(_lock)) < 0) assert(0);                          \
+    taosWLockLatch(_lock);                                \
+    ctgDebug("CTG WLOCK%p, %s:%d", (_lock), __FILE__, __LINE__);  \
+  }                                                       \
+} while (0)
+
+#define CTG_UNLOCK(type, _lock) do {                       \
+  if (CTG_READ == (type)) {                                \
+    if ((*(_lock)) <= 0) assert(0);                         \
+    taosRUnLockLatch(_lock);                              \
+    ctgDebug("CTG RULOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
+  } else {                                                \
+    if ((*(_lock)) <= 0) assert(0);                         \
+    taosWUnLockLatch(_lock);                              \
+    ctgDebug("CTG WULOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
+  }                                                       \
+} while (0)
+
+
 #ifdef __cplusplus
 }
 #endif

source/libs/catalog/src/catalog.c

@@ -20,24 +20,28 @@
 
 SCatalogMgmt ctgMgmt = {0};
 
-int32_t ctgGetDBVgroupFromCache(struct SCatalog* pCatalog, const char *dbName, SDBVgroupInfo *dbInfo, int32_t *exist) {
+int32_t ctgGetDBVgroupFromCache(struct SCatalog* pCatalog, const char *dbName, SDBVgroupInfo **dbInfo, bool *inCache) {
   if (NULL == pCatalog->dbCache.cache) {
-    *exist = 0;
+    *inCache = false;
     return TSDB_CODE_SUCCESS;
   }
 
-  SDBVgroupInfo *info = taosHashGet(pCatalog->dbCache.cache, dbName, strlen(dbName));
+  SDBVgroupInfo *info = taosHashAcquire(pCatalog->dbCache.cache, dbName, strlen(dbName));
 
   if (NULL == info) {
-    *exist = 0;
+    *inCache = false;
     return TSDB_CODE_SUCCESS;
   }
 
-  if (dbInfo) {
-    *dbInfo = *info;
+  CTG_LOCK(CTG_READ, &info->lock);
+  if (NULL == info->vgInfo) {
+    CTG_UNLOCK(CTG_READ, &info->lock);
+    *inCache = false;
+    return TSDB_CODE_SUCCESS;
   }
 
-  *exist = 1;
+  *dbInfo = info;
+  *inCache = true;
   
   return TSDB_CODE_SUCCESS;
 }
@@ -80,46 +84,51 @@ int32_t ctgGetTableMetaFromCache(struct SCatalog* pCatalog, const SName* pTableN
   char tbFullName[TSDB_TABLE_FNAME_LEN];
   tNameExtractFullName(pTableName, tbFullName);
 
-  STableMeta *tbMeta = taosHashGet(pCatalog->tableCache.cache, tbFullName, strlen(tbFullName));
+  *pTableMeta = NULL;
 
-  if (NULL == tbMeta) {
+  size_t sz = 0;
+  STableMeta *tbMeta = taosHashGetCloneExt(pCatalog->tableCache.cache, tbFullName, strlen(tbFullName), NULL, (void **)pTableMeta, &sz);
+
+  if (NULL == *pTableMeta) {
     *exist = 0;
     return TSDB_CODE_SUCCESS;
   }
 
-  if (tbMeta->tableType == TSDB_CHILD_TABLE) {
+  *exist = 1;
+
+  if (tbMeta->tableType != TSDB_CHILD_TABLE) {
+    return TSDB_CODE_SUCCESS;
+  }
+  
+  CTG_LOCK(CTG_READ, &pCatalog->tableCache.stableLock);
+  
   STableMeta **stbMeta = taosHashGet(pCatalog->tableCache.stableCache, &tbMeta->suid, sizeof(tbMeta->suid));
   if (NULL == stbMeta || NULL == *stbMeta) {
+    CTG_UNLOCK(CTG_READ, &pCatalog->tableCache.stableLock);
+    qError("no stable:%"PRIx64 " meta in cache", tbMeta->suid);
+    tfree(*pTableMeta);
     *exist = 0;
     return TSDB_CODE_SUCCESS;
   }
 
   if ((*stbMeta)->suid != tbMeta->suid) {    
+    CTG_UNLOCK(CTG_READ, &pCatalog->tableCache.stableLock);
+    tfree(*pTableMeta);
     ctgError("stable cache error, expected suid:%"PRId64 ",actual suid:%"PRId64, tbMeta->suid, (*stbMeta)->suid);
     CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR);
   }
 
   int32_t metaSize = sizeof(STableMeta) + ((*stbMeta)->tableInfo.numOfTags + (*stbMeta)->tableInfo.numOfColumns) * sizeof(SSchema);
-    *pTableMeta = calloc(1, metaSize);
+  *pTableMeta = realloc(*pTableMeta, metaSize);
   if (NULL == *pTableMeta) {    
+    CTG_UNLOCK(CTG_READ, &pCatalog->tableCache.stableLock);
     ctgError("calloc size[%d] failed", metaSize);
     CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
   }
 
-    memcpy(*pTableMeta, tbMeta, sizeof(SCTableMeta));
   memcpy(&(*pTableMeta)->sversion, &(*stbMeta)->sversion, metaSize - sizeof(SCTableMeta));
-  } else {
-    int32_t metaSize = sizeof(STableMeta) + (tbMeta->tableInfo.numOfTags + tbMeta->tableInfo.numOfColumns) * sizeof(SSchema);
-    *pTableMeta = calloc(1, metaSize);
-    if (NULL == *pTableMeta) {
-      ctgError("calloc size[%d] failed", metaSize);
-      CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
-    }
 
-    memcpy(*pTableMeta, tbMeta, metaSize);
-  }
-
-  *exist = 1;
+  CTG_UNLOCK(CTG_READ, &pCatalog->tableCache.stableLock);
   
   return TSDB_CODE_SUCCESS;
 }
@@ -223,9 +232,11 @@ int32_t ctgGetHashFunction(int8_t hashMethod, tableNameHashFp *fp) {
 int32_t ctgGetVgInfoFromDB(struct SCatalog *pCatalog, void *pRpc, const SEpSet *pMgmtEps, SDBVgroupInfo *dbInfo, SArray** vgroupList) {
   SHashObj *vgroupHash = NULL;
   SVgroupInfo *vgInfo = NULL;
+  SArray *vgList = NULL;
+  int32_t code = 0;
 
-  *vgroupList = taosArrayInit(taosHashGetSize(dbInfo->vgInfo), sizeof(SVgroupInfo));
-  if (NULL == *vgroupList) {
+  vgList = taosArrayInit(taosHashGetSize(dbInfo->vgInfo), sizeof(SVgroupInfo));
+  if (NULL == vgList) {
     ctgError("taosArrayInit failed");
     CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);    
   }
@@ -234,19 +245,34 @@ int32_t ctgGetVgInfoFromDB(struct SCatalog *pCatalog, void *pRpc, const SEpSet *
   while (pIter) {
     vgInfo = pIter;
 
-    if (NULL == taosArrayPush(*vgroupList, vgInfo)) {
+    if (NULL == taosArrayPush(vgList, vgInfo)) {
       ctgError("taosArrayPush failed");
-      CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
     }
     
     pIter = taosHashIterate(dbInfo->vgInfo, pIter);
     vgInfo = NULL;
   }
 
+  *vgroupList = vgList;
+  vgList = NULL;
+
   return TSDB_CODE_SUCCESS;
+
+_return:
+
+  if (vgList) {
+    taosArrayDestroy(vgList);
+  }
+
+  CTG_RET(code);
 }
 
 int32_t ctgGetVgInfoFromHashValue(SDBVgroupInfo *dbInfo, const SName *pTableName, SVgroupInfo *pVgroup) {
+  int32_t code = 0;
+  
+  CTG_LOCK(CTG_READ, &dbInfo->lock);
+  
   int32_t vgNum = taosHashGetSize(dbInfo->vgInfo);
   char db[TSDB_DB_FNAME_LEN] = {0};
   tNameGetFullDbName(pTableName, db);
@@ -259,7 +285,7 @@ int32_t ctgGetVgInfoFromHashValue(SDBVgroupInfo *dbInfo, const SName *pTableName
   tableNameHashFp fp = NULL;
   SVgroupInfo *vgInfo = NULL;
 
-  CTG_ERR_RET(ctgGetHashFunction(dbInfo->hashMethod, &fp));
+  CTG_ERR_JRET(ctgGetHashFunction(dbInfo->hashMethod, &fp));
 
   char tbFullName[TSDB_TABLE_FNAME_LEN];
   tNameExtractFullName(pTableName, tbFullName);
@@ -279,12 +305,16 @@ int32_t ctgGetVgInfoFromHashValue(SDBVgroupInfo *dbInfo, const SName *pTableName
 
   if (NULL == vgInfo) {
     ctgError("no hash range found for hashvalue[%u]", hashValue);
-    CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR);
+    CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);
   }
 
   *pVgroup = *vgInfo;
 
-  return TSDB_CODE_SUCCESS;
+_return:
+
+  CTG_UNLOCK(CTG_READ, &dbInfo->lock);
+  
+  CTG_RET(TSDB_CODE_SUCCESS);
 }
 
 int32_t ctgGetTableMetaImpl(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const SName* pTableName, bool forceUpdate, STableMeta** pTableMeta) {
@@ -316,21 +346,23 @@ int32_t ctgGetTableMetaImpl(struct SCatalog* pCatalog, void *pRpc, const SEpSet*
 
 
 int32_t ctgUpdateTableMetaCache(struct SCatalog *pCatalog, STableMetaOutput *output) {
+  int32_t code = 0;
+  
   if (output->metaNum != 1 && output->metaNum != 2) {
     ctgError("invalid table meta number[%d] got from meta rsp", output->metaNum);
-    CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR);
+    CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);
   }
 
   if (NULL == output->tbMeta) {
     ctgError("no valid table meta got from meta rsp");
-    CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR);
+    CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);
   }
 
   if (NULL == pCatalog->tableCache.cache) {
     pCatalog->tableCache.cache = taosHashInit(ctgMgmt.cfg.maxTblCacheNum, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
     if (NULL == pCatalog->tableCache.cache) {
       ctgError("init hash[%d] for tablemeta cache failed", ctgMgmt.cfg.maxTblCacheNum);
-      CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
     }
   }
 
@@ -338,50 +370,59 @@ int32_t ctgUpdateTableMetaCache(struct SCatalog *pCatalog, STableMetaOutput *out
     pCatalog->tableCache.stableCache = taosHashInit(ctgMgmt.cfg.maxTblCacheNum, taosGetDefaultHashFunction(TSDB_DATA_TYPE_UBIGINT), true, HASH_ENTRY_LOCK);
     if (NULL == pCatalog->tableCache.stableCache) {
       ctgError("init hash[%d] for stablemeta cache failed", ctgMgmt.cfg.maxTblCacheNum);
-      CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
     }
   }
 
   if (output->metaNum == 2) {
     if (taosHashPut(pCatalog->tableCache.cache, output->ctbFname, strlen(output->ctbFname), &output->ctbMeta, sizeof(output->ctbMeta)) != 0) {
       ctgError("push ctable[%s] to table cache failed", output->ctbFname);
-      CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
     }
 
     if (TSDB_SUPER_TABLE != output->tbMeta->tableType) {
       ctgError("table type[%d] error, expected:%d", output->tbMeta->tableType, TSDB_SUPER_TABLE);
-      CTG_ERR_RET(TSDB_CODE_CTG_INTERNAL_ERROR);
+      CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);
     }    
   }
 
   int32_t tbSize = sizeof(*output->tbMeta) + sizeof(SSchema) * (output->tbMeta->tableInfo.numOfColumns + output->tbMeta->tableInfo.numOfTags);
-  if (taosHashPut(pCatalog->tableCache.cache, output->tbFname, strlen(output->tbFname), output->tbMeta, tbSize) != 0) {
-    ctgError("push table[%s] to table cache failed", output->tbFname);
-    CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
-  }
 
   if (TSDB_SUPER_TABLE == output->tbMeta->tableType) {
-    if (taosHashPut(pCatalog->tableCache.stableCache, &output->tbMeta->suid, sizeof(output->tbMeta->suid), &output->tbMeta, POINTER_BYTES) != 0) {
+    CTG_LOCK(CTG_WRITE, &pCatalog->tableCache.stableLock);
+    if (taosHashPut(pCatalog->tableCache.cache, output->tbFname, strlen(output->tbFname), output->tbMeta, tbSize) != 0) {
+      CTG_UNLOCK(CTG_WRITE, &pCatalog->tableCache.stableLock);
+      ctgError("push table[%s] to table cache failed", output->tbFname);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
+    }
+
+    STableMeta *tbMeta = taosHashGet(pCatalog->tableCache.cache, output->tbFname, strlen(output->tbFname));
+    if (taosHashPut(pCatalog->tableCache.stableCache, &output->tbMeta->suid, sizeof(output->tbMeta->suid), &tbMeta, POINTER_BYTES) != 0) {
+      CTG_UNLOCK(CTG_WRITE, &pCatalog->tableCache.stableLock);
       ctgError("push suid[%"PRIu64"] to stable cache failed", output->tbMeta->suid);
-      CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
+    }
+    CTG_UNLOCK(CTG_WRITE, &pCatalog->tableCache.stableLock);
+  } else {
+    if (taosHashPut(pCatalog->tableCache.cache, output->tbFname, strlen(output->tbFname), output->tbMeta, tbSize) != 0) {
+      ctgError("push table[%s] to table cache failed", output->tbFname);
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
     }
   }
 
-  return TSDB_CODE_SUCCESS;
+_return:
+  tfree(output->tbMeta);
+  
+  CTG_RET(code);
 }
 
 
-int32_t ctgGetDBVgroup(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* dbName, int32_t forceUpdate, SDBVgroupInfo* dbInfo) {
-  if (NULL == pCatalog || NULL == dbName || NULL == pRpc || NULL == pMgmtEps) {
-    CTG_ERR_RET(TSDB_CODE_CTG_INVALID_INPUT);
-  }
-
-  int32_t exist = 0;
-
+int32_t ctgGetDBVgroup(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* dbName, int32_t forceUpdate, SDBVgroupInfo** dbInfo) {
+  bool inCache = false;
   if (0 == forceUpdate) {
-    CTG_ERR_RET(ctgGetDBVgroupFromCache(pCatalog, dbName, dbInfo, &exist));
+    CTG_ERR_RET(ctgGetDBVgroupFromCache(pCatalog, dbName, dbInfo, &inCache));
 
-    if (exist) {
+    if (inCache) {
       return TSDB_CODE_SUCCESS;
     }
   }
@@ -397,9 +438,7 @@ int32_t ctgGetDBVgroup(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgm
 
   CTG_ERR_RET(catalogUpdateDBVgroup(pCatalog, dbName, &DbOut.dbVgroup));
 
-  if (dbInfo) {
-    *dbInfo = DbOut.dbVgroup;
-  }
+  CTG_ERR_RET(ctgGetDBVgroupFromCache(pCatalog, dbName, dbInfo, &inCache));
 
   return TSDB_CODE_SUCCESS;
 }
@@ -479,17 +518,68 @@ int32_t catalogGetDBVgroupVersion(struct SCatalog* pCatalog, const char* dbName,
     return TSDB_CODE_SUCCESS;
   }
 
-  SDBVgroupInfo * dbInfo = taosHashGet(pCatalog->dbCache.cache, dbName, strlen(dbName));
+  SDBVgroupInfo * dbInfo = taosHashAcquire(pCatalog->dbCache.cache, dbName, strlen(dbName));
   if (NULL == dbInfo) {
     *version = CTG_DEFAULT_INVALID_VERSION;
     return TSDB_CODE_SUCCESS;
   }
 
   *version = dbInfo->vgVersion;
+  taosHashRelease(pCatalog->dbCache.cache, dbInfo);
 
   return TSDB_CODE_SUCCESS;
 }
 
+int32_t catalogGetDBVgroup(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* dbName, int32_t forceUpdate, SArray** vgroupList) {
+  if (NULL == pCatalog || NULL == dbName || NULL == pRpc || NULL == pMgmtEps || NULL == vgroupList) {
+    CTG_ERR_RET(TSDB_CODE_CTG_INVALID_INPUT);
+  }
+
+  SDBVgroupInfo* db = NULL;
+  int32_t code = 0;
+  SVgroupInfo *vgInfo = NULL;
+  SArray *vgList = NULL;
+  
+  CTG_ERR_JRET(ctgGetDBVgroup(pCatalog, pRpc, pMgmtEps, dbName, forceUpdate, &db));
+
+  vgList = taosArrayInit(taosHashGetSize(db->vgInfo), sizeof(SVgroupInfo));
+  if (NULL == vgList) {
+    ctgError("taosArrayInit failed");
+    CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);    
+  }
+
+  void *pIter = taosHashIterate(db->vgInfo, NULL);
+  while (pIter) {
+    vgInfo = pIter;
+
+    if (NULL == taosArrayPush(vgList, vgInfo)) {
+      ctgError("taosArrayPush failed");
+      CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);
+    }
+    
+    pIter = taosHashIterate(db->vgInfo, pIter);
+    vgInfo = NULL;
+  }
+
+  *vgroupList = vgList;
+  vgList = NULL;
+
+_return:
+
+  if (db) {
+    CTG_UNLOCK(CTG_READ, &db->lock);
+    taosHashRelease(pCatalog->dbCache.cache, db);
+  }
+
+  if (vgList) {
+    taosArrayDestroy(vgList);
+    vgList = NULL;
+  }
+
+  CTG_RET(code);  
+}
+
+
 int32_t catalogUpdateDBVgroup(struct SCatalog* pCatalog, const char* dbName, SDBVgroupInfo* dbInfo) {
   if (NULL == pCatalog || NULL == dbName || NULL == dbInfo) {
     CTG_ERR_RET(TSDB_CODE_CTG_INVALID_INPUT);
@@ -497,13 +587,17 @@ int32_t catalogUpdateDBVgroup(struct SCatalog* pCatalog, const char* dbName, SDB
 
   if (dbInfo->vgVersion < 0) {
     if (pCatalog->dbCache.cache) {
-      SDBVgroupInfo *oldInfo = taosHashGet(pCatalog->dbCache.cache, dbName, strlen(dbName));
-      if (oldInfo && oldInfo->vgInfo) {
+      SDBVgroupInfo *oldInfo = taosHashAcquire(pCatalog->dbCache.cache, dbName, strlen(dbName));
+      if (oldInfo) {
+        CTG_LOCK(CTG_WRITE, &oldInfo->lock);
+        if (oldInfo->vgInfo) {
           taosHashCleanup(oldInfo->vgInfo);
           oldInfo->vgInfo = NULL;
         }
+        CTG_UNLOCK(CTG_WRITE, &oldInfo->lock);
 
-      taosHashRemove(pCatalog->dbCache.cache, dbName, strlen(dbName));
+        taosHashRelease(pCatalog->dbCache.cache, oldInfo);
+      }
     }
     
     ctgWarn("remove db [%s] from cache", dbName);
@@ -517,11 +611,17 @@ int32_t catalogUpdateDBVgroup(struct SCatalog* pCatalog, const char* dbName, SDB
       CTG_ERR_RET(TSDB_CODE_CTG_MEM_ERROR);
     }
   } else {
-    SDBVgroupInfo *oldInfo = taosHashGet(pCatalog->dbCache.cache, dbName, strlen(dbName));
-    if (oldInfo && oldInfo->vgInfo) {
+    SDBVgroupInfo *oldInfo = taosHashAcquire(pCatalog->dbCache.cache, dbName, strlen(dbName));
+    if (oldInfo) {
+      CTG_LOCK(CTG_WRITE, &oldInfo->lock);
+      if (oldInfo->vgInfo) {
         taosHashCleanup(oldInfo->vgInfo);
         oldInfo->vgInfo = NULL;
       }
+      CTG_UNLOCK(CTG_WRITE, &oldInfo->lock);
+    
+      taosHashRelease(pCatalog->dbCache.cache, oldInfo);
+    }
   }
 
   if (taosHashPut(pCatalog->dbCache.cache, dbName, strlen(dbName), dbInfo, sizeof(*dbInfo)) != 0) {
@@ -573,7 +673,10 @@ int32_t catalogGetTableDistVgroup(struct SCatalog* pCatalog, void *pRpc, const S
   STableMeta *tbMeta = NULL;
   int32_t code = 0;
   SVgroupInfo vgroupInfo = {0};
-  SDBVgroupInfo dbVgroup = {0};
+  SDBVgroupInfo* dbVgroup = NULL;
+  SArray *vgList = NULL;
+
+  *pVgroupList = NULL;
   
   CTG_ERR_JRET(catalogGetTableMeta(pCatalog, pRpc, pMgmtEps, pTableName, &tbMeta));
 
@@ -582,38 +685,48 @@ int32_t catalogGetTableDistVgroup(struct SCatalog* pCatalog, void *pRpc, const S
   CTG_ERR_JRET(ctgGetDBVgroup(pCatalog, pRpc, pMgmtEps, db, false, &dbVgroup));
 
   if (tbMeta->tableType == TSDB_SUPER_TABLE) {
-    CTG_ERR_JRET(ctgGetVgInfoFromDB(pCatalog, pRpc, pMgmtEps, &dbVgroup, pVgroupList));
+    CTG_ERR_JRET(ctgGetVgInfoFromDB(pCatalog, pRpc, pMgmtEps, dbVgroup, pVgroupList));
   } else {
     int32_t vgId = tbMeta->vgId;
-    if (NULL == taosHashGetClone(dbVgroup.vgInfo, &vgId, sizeof(vgId), &vgroupInfo)) {
+    if (NULL == taosHashGetClone(dbVgroup->vgInfo, &vgId, sizeof(vgId), &vgroupInfo)) {
       ctgError("vgId[%d] not found in vgroup list", vgId);
       CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);    
     }
 
-    *pVgroupList = taosArrayInit(1, sizeof(SVgroupInfo));
-    if (NULL == *pVgroupList) {
+    vgList = taosArrayInit(1, sizeof(SVgroupInfo));
+    if (NULL == vgList) {
       ctgError("taosArrayInit failed");
       CTG_ERR_JRET(TSDB_CODE_CTG_MEM_ERROR);    
     }
 
-    if (NULL == taosArrayPush(*pVgroupList, &vgroupInfo)) {
+    if (NULL == taosArrayPush(vgList, &vgroupInfo)) {
       ctgError("push vgroupInfo to array failed");
       CTG_ERR_JRET(TSDB_CODE_CTG_INTERNAL_ERROR);
     }
-  }
 
-  tfree(tbMeta);
-  return TSDB_CODE_SUCCESS;
+    *pVgroupList = vgList;
+    vgList = NULL;
+  }
 
 _return:
   tfree(tbMeta);
-  taosArrayDestroy(*pVgroupList);
+
+  if (dbVgroup) {
+    CTG_UNLOCK(CTG_READ, &dbVgroup->lock);
+    taosHashRelease(pCatalog->dbCache.cache, dbVgroup);
+  }
+
+  if (vgList) {
+    taosArrayDestroy(vgList);
+    vgList = NULL;
+  }
+  
   CTG_RET(code);
 }
 
 
 int32_t catalogGetTableHashVgroup(struct SCatalog *pCatalog, void *pTransporter, const SEpSet *pMgmtEps, const SName *pTableName, SVgroupInfo *pVgroup) {
-  SDBVgroupInfo dbInfo = {0};
+  SDBVgroupInfo* dbInfo = NULL;
   int32_t code = 0;
 
   char db[TSDB_DB_FNAME_LEN] = {0};
@@ -621,12 +734,14 @@ int32_t catalogGetTableHashVgroup(struct SCatalog *pCatalog, void *pTransporter,
 
   CTG_ERR_RET(ctgGetDBVgroup(pCatalog, pTransporter, pMgmtEps, db, false, &dbInfo));
 
-  if (dbInfo.vgVersion < 0 || NULL == dbInfo.vgInfo) {
-    ctgError("db[%s] vgroup cache invalid, vgroup version:%d, vgInfo:%p", db, dbInfo.vgVersion, dbInfo.vgInfo);
-    CTG_ERR_RET(TSDB_CODE_TSC_DB_NOT_SELECTED);
-  }
+  CTG_ERR_JRET(ctgGetVgInfoFromHashValue(dbInfo, pTableName, pVgroup));
 
-  CTG_ERR_RET(ctgGetVgInfoFromHashValue(&dbInfo, pTableName, pVgroup));
+_return:
+
+  if (dbInfo) {
+    CTG_UNLOCK(CTG_READ, &dbInfo->lock);  
+    taosHashRelease(pCatalog->dbCache.cache, dbInfo);
+  }
 
   CTG_RET(code);
 }

source/libs/index/inc/index_fst_counting_writer.h

@@ -37,6 +37,7 @@ typedef struct WriterCtx {
     struct {
       int  fd;
       bool readOnly;
+      char buf[256];
     } file;
     struct {
       int32_t capa;
@@ -53,7 +54,7 @@ static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t off
 static int writeCtxDoFlush(WriterCtx* ctx);
 
 WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int32_t capacity);
-void       writerCtxDestroy(WriterCtx* w);
+void       writerCtxDestroy(WriterCtx* w, bool remove);
 
 typedef uint32_t CheckSummer;
 

source/libs/index/inc/index_tfile.h

@@ -107,6 +107,7 @@ void         tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader*
 
 TFileReader* tfileGetReaderByCol(IndexTFile* tf, char* colName);
 
+TFileReader* tfileReaderOpen(char* path, uint64_t suid, int32_t version, const char* colName);
 TFileReader* tfileReaderCreate(WriterCtx* ctx);
 void         tfileReaderDestroy(TFileReader* reader);
 int          tfileReaderSearch(TFileReader* reader, SIndexTermQuery* query, SArray* result);
@@ -114,10 +115,10 @@ void         tfileReaderRef(TFileReader* reader);
 void         tfileReaderUnRef(TFileReader* reader);
 
 TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int32_t version, const char* colName, uint8_t type);
-void         tfileWriteClose(TFileWriter* tw);
+void         tfileWriterClose(TFileWriter* tw);
 TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header);
 void         tfileWriterDestroy(TFileWriter* tw);
-int          tfileWriterPut(TFileWriter* tw, void* data);
+int          tfileWriterPut(TFileWriter* tw, void* data, bool order);
 int          tfileWriterFinish(TFileWriter* tw);
 
 //

source/libs/index/src/index.c

@@ -25,7 +25,7 @@
 #endif
 
 #define INDEX_NUM_OF_THREADS 4
-#define INDEX_QUEUE_SIZE 4
+#define INDEX_QUEUE_SIZE 200
 
 void* indexQhandle = NULL;
 
@@ -52,18 +52,17 @@ typedef struct SIdxColInfo {
 
 static pthread_once_t isInit = PTHREAD_ONCE_INIT;
 // static void           indexInit();
-
 static int indexTermSearch(SIndex* sIdx, SIndexTermQuery* term, SArray** result);
 
 static void indexInterResultsDestroy(SArray* results);
 static int  indexMergeFinalResults(SArray* interResults, EIndexOperatorType oType, SArray* finalResult);
 
+static int indexGenTFile(SIndex* index, IndexCache* cache, SArray* batch);
+
 int indexOpen(SIndexOpts* opts, const char* path, SIndex** index) {
   // pthread_once(&isInit, indexInit);
   SIndex* sIdx = calloc(1, sizeof(SIndex));
-  if (sIdx == NULL) {
-    return -1;
-  }
+  if (sIdx == NULL) { return -1; }
 
 #ifdef USE_LUCENE
   index_t* index = index_open(path);
@@ -99,9 +98,7 @@ void indexClose(SIndex* sIdx) {
   void* iter = taosHashIterate(sIdx->colObj, NULL);
   while (iter) {
     IndexCache** pCache = iter;
-    if (*pCache) {
-      indexCacheUnRef(*pCache);
-    }
+    if (*pCache) { indexCacheUnRef(*pCache); }
     iter = taosHashIterate(sIdx->colObj, iter);
   }
   taosHashCleanup(sIdx->colObj);
@@ -147,9 +144,7 @@ int indexPut(SIndex* index, SIndexMultiTerm* fVals, uint64_t uid) {
     IndexCache** cache = taosHashGet(index->colObj, p->colName, p->nColName);
     assert(*cache != NULL);
     int ret = indexCachePut(*cache, p, uid);
-    if (ret != 0) {
-      return ret;
-    }
+    if (ret != 0) { return ret; }
   }
 
 #endif
@@ -179,9 +174,7 @@ int indexSearch(SIndex* index, SIndexMultiTermQuery* multiQuerys, SArray* result
   int  tsz = 0;
   index_multi_search(index->index, (const char**)fields, (const char**)keys, types, nQuery, opera, &tResult, &tsz);
 
-  for (int i = 0; i < tsz; i++) {
-    taosArrayPush(result, &tResult[i]);
-  }
+  for (int i = 0; i < tsz; i++) { taosArrayPush(result, &tResult[i]); }
 
   for (int i = 0; i < nQuery; i++) {
     free(fields[i]);
@@ -238,9 +231,7 @@ void indexOptsDestroy(SIndexOpts* opts) {
  */
 SIndexMultiTermQuery* indexMultiTermQueryCreate(EIndexOperatorType opera) {
   SIndexMultiTermQuery* p = (SIndexMultiTermQuery*)malloc(sizeof(SIndexMultiTermQuery));
-  if (p == NULL) {
-    return NULL;
-  }
+  if (p == NULL) { return NULL; }
   p->opera = opera;
   p->query = taosArrayInit(4, sizeof(SIndexTermQuery));
   return p;
@@ -262,9 +253,7 @@ int indexMultiTermQueryAdd(SIndexMultiTermQuery* pQuery, SIndexTerm* term, EInde
 SIndexTerm* indexTermCreate(int64_t suid, SIndexOperOnColumn oper, uint8_t colType, const char* colName,
                             int32_t nColName, const char* colVal, int32_t nColVal) {
   SIndexTerm* t = (SIndexTerm*)calloc(1, (sizeof(SIndexTerm)));
-  if (t == NULL) {
-    return NULL;
-  }
+  if (t == NULL) { return NULL; }
 
   t->suid = suid;
   t->operType = oper;
@@ -317,7 +306,7 @@ static int indexTermSearch(SIndex* sIdx, SIndexTermQuery* query, SArray** result
 
   *result = taosArrayInit(4, sizeof(uint64_t));
   // TODO: iterator mem and tidex
-  STermValueType s;
+  STermValueType s = kTypeValue;
   if (0 == indexCacheSearch(cache, query, *result, &s)) {
     if (s == kTypeDeletion) {
       indexInfo("col: %s already drop by other opera", term->colName);
@@ -336,9 +325,7 @@ static int indexTermSearch(SIndex* sIdx, SIndexTermQuery* query, SArray** result
   return 0;
 }
 static void indexInterResultsDestroy(SArray* results) {
-  if (results == NULL) {
-    return;
-  }
+  if (results == NULL) { return; }
 
   size_t sz = taosArrayGetSize(results);
   for (size_t i = 0; i < sz; i++) {
@@ -368,23 +355,43 @@ static int indexMergeFinalResults(SArray* interResults, EIndexOperatorType oType
   return 0;
 }
 
-int indexFlushCacheTFile(SIndex* sIdx, void* cache) {
-  if (sIdx == NULL) {
-    return -1;
+static void indexMergeSameKey(SArray* result, TFileValue* tv) {
+  int32_t sz = result ? taosArrayGetSize(result) : 0;
+  if (sz > 0) {
+    // TODO(yihao): remove duplicate tableid
+    TFileValue* lv = taosArrayGetP(result, sz - 1);
+    if (strcmp(lv->colVal, tv->colVal) == 0) {
+      taosArrayAddAll(lv->tableId, tv->tableId);
+      tfileValueDestroy(tv);
+    } else {
+      taosArrayPush(result, &tv);
     }
+  } else {
+    taosArrayPush(result, &tv);
+  }
+}
+static void indexDestroyTempResult(SArray* result) {
+  int32_t sz = result ? taosArrayGetSize(result) : 0;
+  for (size_t i = 0; i < sz; i++) {
+    TFileValue* tv = taosArrayGetP(result, i);
+    tfileValueDestroy(tv);
+  }
+  taosArrayDestroy(result);
+}
+int indexFlushCacheTFile(SIndex* sIdx, void* cache) {
+  if (sIdx == NULL) { return -1; }
   indexWarn("suid %" PRIu64 " merge cache into tindex", sIdx->suid);
 
   IndexCache*  pCache = (IndexCache*)cache;
   TFileReader* pReader = tfileGetReaderByCol(sIdx->tindex, pCache->colName);
-
   // handle flush
   Iterate* cacheIter = indexCacheIteratorCreate(pCache);
   Iterate* tfileIter = tfileIteratorCreate(pReader);
 
   SArray* result = taosArrayInit(1024, sizeof(void*));
 
-  bool cn = cacheIter->next(cacheIter);
-  bool tn = tfileIter->next(tfileIter);
+  bool cn = cacheIter ? cacheIter->next(cacheIter) : false;
+  bool tn = tfileIter ? tfileIter->next(tfileIter) : false;
   while (cn == true && tn == true) {
     IterateValue* cv = cacheIter->getValue(cacheIter);
     IterateValue* tv = tfileIter->getValue(tfileIter);
@@ -395,7 +402,7 @@ int indexFlushCacheTFile(SIndex* sIdx, void* cache) {
       TFileValue* tfv = tfileValueCreate(cv->colVal);
       taosArrayAddAll(tfv->tableId, cv->val);
       taosArrayAddAll(tfv->tableId, tv->val);
-      taosArrayPush(result, &tfv);
+      indexMergeSameKey(result, tfv);
 
       cn = cacheIter->next(cacheIter);
       tn = tfileIter->next(tfileIter);
@@ -403,13 +410,15 @@ int indexFlushCacheTFile(SIndex* sIdx, void* cache) {
     } else if (comp < 0) {
       TFileValue* tfv = tfileValueCreate(cv->colVal);
       taosArrayAddAll(tfv->tableId, cv->val);
-      taosArrayPush(result, &tfv);
+
+      indexMergeSameKey(result, tfv);
       // copy to final Result;
       cn = cacheIter->next(cacheIter);
     } else {
       TFileValue* tfv = tfileValueCreate(tv->colVal);
-      taosArrayPush(result, &tfv);
       taosArrayAddAll(tfv->tableId, tv->val);
+
+      indexMergeSameKey(result, tfv);
       // copy to final result
       tn = tfileIter->next(tfileIter);
     }
@@ -418,33 +427,24 @@ int indexFlushCacheTFile(SIndex* sIdx, void* cache) {
     IterateValue* cv = cacheIter->getValue(cacheIter);
     TFileValue*   tfv = tfileValueCreate(cv->colVal);
     taosArrayAddAll(tfv->tableId, cv->val);
-    taosArrayPush(result, &tfv);
+    indexMergeSameKey(result, tfv);
     cn = cacheIter->next(cacheIter);
   }
   while (tn == true) {
     IterateValue* tv = tfileIter->getValue(tfileIter);
     TFileValue*   tfv = tfileValueCreate(tv->colVal);
+    if (tv->val == NULL) {
+      // HO
+      printf("NO....");
+    }
     taosArrayAddAll(tfv->tableId, tv->val);
-    taosArrayPush(result, &tfv);
+    indexMergeSameKey(result, tfv);
     tn = tfileIter->next(tfileIter);
   }
-
-  int32_t version = CACHE_VERSION(pCache);
-  uint8_t colType = pCache->type;
-
-  TFileWriter* tw = tfileWriterOpen(sIdx->path, sIdx->suid, version, pCache->colName, colType);
-  if (tw == NULL) {
-    indexError("faile to open file to write");
-  } else {
-    int ret = tfileWriterPut(tw, result);
-    if (ret != 0) {
-      indexError("faile to write into tindex ");
-    }
-  }
-  // not free later, just put int table cache
+  int ret = indexGenTFile(sIdx, pCache, result);
+  indexDestroyTempResult(result);
   indexCacheDestroyImm(pCache);
 
-  tfileWriteClose(tw);
   indexCacheIteratorDestroy(cacheIter);
   tfileIteratorDestroy(tfileIter);
 
@@ -455,9 +455,45 @@ int indexFlushCacheTFile(SIndex* sIdx, void* cache) {
 void iterateValueDestroy(IterateValue* value, bool destroy) {
   if (destroy) {
     taosArrayDestroy(value->val);
+    value->val = NULL;
   } else {
-    taosArrayClear(value->val);
+    if (value->val != NULL) { taosArrayClear(value->val); }
   }
-  free(value->colVal);
+  // free(value->colVal);
   value->colVal = NULL;
 }
+static int indexGenTFile(SIndex* sIdx, IndexCache* cache, SArray* batch) {
+  int32_t version = CACHE_VERSION(cache);
+  uint8_t colType = cache->type;
+
+  TFileWriter* tw = tfileWriterOpen(sIdx->path, sIdx->suid, version, cache->colName, colType);
+  if (tw == NULL) {
+    indexError("failed to open file to write");
+    return -1;
+  }
+
+  int ret = tfileWriterPut(tw, batch, true);
+  if (ret != 0) {
+    indexError("failed to write into tindex ");
+    goto END;
+  }
+  tfileWriterClose(tw);
+
+  TFileReader* reader = tfileReaderOpen(sIdx->path, sIdx->suid, version, cache->colName);
+
+  char          buf[128] = {0};
+  TFileHeader*  header = &reader->header;
+  TFileCacheKey key = {.suid = header->suid,
+                       .colName = header->colName,
+                       .nColName = strlen(header->colName),
+                       .colType = header->colType};
+  pthread_mutex_lock(&sIdx->mtx);
+
+  IndexTFile* ifile = (IndexTFile*)sIdx->tindex;
+  tfileCachePut(ifile->cache, &key, reader);
+
+  pthread_mutex_unlock(&sIdx->mtx);
+  return ret;
+END:
+  tfileWriterClose(tw);
+}

source/libs/index/src/index_cache.c

@@ -20,7 +20,7 @@
 
 #define MAX_INDEX_KEY_LEN 256  // test only, change later
 
-#define MEM_TERM_LIMIT 1000000
+#define MEM_TERM_LIMIT 200
 // ref index_cache.h:22
 //#define CACHE_KEY_LEN(p) \
 //  (sizeof(int32_t) + sizeof(uint16_t) + sizeof(p->colType) + sizeof(p->nColVal) + p->nColVal + sizeof(uint64_t) +
@@ -66,6 +66,7 @@ void indexCacheDebug(IndexCache* cache) {
   indexMemRef(tbl);
   pthread_mutex_unlock(&cache->mtx);
 
+  {
     SSkipList*         slt = tbl->mem;
     SSkipListIterator* iter = tSkipListCreateIter(slt);
     while (tSkipListIterNext(iter)) {
@@ -79,6 +80,29 @@ void indexCacheDebug(IndexCache* cache) {
     tSkipListDestroyIter(iter);
 
     indexMemUnRef(tbl);
+  }
+
+  {
+    pthread_mutex_lock(&cache->mtx);
+    tbl = cache->imm;
+    indexMemRef(tbl);
+    pthread_mutex_unlock(&cache->mtx);
+    if (tbl != NULL) {
+      SSkipList*         slt = tbl->mem;
+      SSkipListIterator* iter = tSkipListCreateIter(slt);
+      while (tSkipListIterNext(iter)) {
+        SSkipListNode* node = tSkipListIterGet(iter);
+        CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
+        if (ct != NULL) {
+          // TODO, add more debug info
+          indexInfo("{colVal: %s, version: %d} \t", ct->colVal, ct->version);
+        }
+      }
+      tSkipListDestroyIter(iter);
+    }
+
+    indexMemUnRef(tbl);
+  }
 }
 
 void indexCacheDestroySkiplist(SSkipList* slt) {
@@ -86,25 +110,26 @@ void indexCacheDestroySkiplist(SSkipList* slt) {
   while (tSkipListIterNext(iter)) {
     SSkipListNode* node = tSkipListIterGet(iter);
     CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
-    if (ct != NULL) {
-    }
+    if (ct != NULL) {}
   }
   tSkipListDestroyIter(iter);
   tSkipListDestroy(slt);
 }
 void indexCacheDestroyImm(IndexCache* cache) {
+  if (cache == NULL) { return; }
+
   MemTable* tbl = NULL;
   pthread_mutex_lock(&cache->mtx);
   tbl = cache->imm;
   cache->imm = NULL;  // or throw int bg thread
   pthread_mutex_unlock(&cache->mtx);
+
+  indexMemUnRef(tbl);
   indexMemUnRef(tbl);
 }
 void indexCacheDestroy(void* cache) {
   IndexCache* pCache = cache;
-  if (pCache == NULL) {
-    return;
-  }
+  if (pCache == NULL) { return; }
   indexMemUnRef(pCache->mem);
   indexMemUnRef(pCache->imm);
   free(pCache->colName);
@@ -114,9 +139,11 @@ void indexCacheDestroy(void* cache) {
 
 Iterate* indexCacheIteratorCreate(IndexCache* cache) {
   Iterate* iiter = calloc(1, sizeof(Iterate));
-  if (iiter == NULL) {
-    return NULL;
-  }
+  if (iiter == NULL) { return NULL; }
+
+  pthread_mutex_lock(&cache->mtx);
+
+  indexMemRef(cache->imm);
 
   MemTable* tbl = cache->imm;
   iiter->val.val = taosArrayInit(1, sizeof(uint64_t));
@@ -124,12 +151,12 @@ Iterate* indexCacheIteratorCreate(IndexCache* cache) {
   iiter->next = indexCacheIteratorNext;
   iiter->getValue = indexCacheIteratorGetValue;
 
+  pthread_mutex_unlock(&cache->mtx);
+
   return iiter;
 }
 void indexCacheIteratorDestroy(Iterate* iter) {
-  if (iter == NULL) {
-    return;
-  }
+  if (iter == NULL) { return; }
   tSkipListDestroyIter(iter->iter);
   iterateValueDestroy(&iter->val, true);
   free(iter);
@@ -155,6 +182,7 @@ static void indexCacheMakeRoomForWrite(IndexCache* cache) {
       taosMsleep(50);
       pthread_mutex_lock(&cache->mtx);
     } else {
+      indexCacheRef(cache);
       cache->imm = cache->mem;
       cache->mem = indexInternalCacheCreate(cache->type);
       cache->nTerm = 1;
@@ -166,17 +194,13 @@ static void indexCacheMakeRoomForWrite(IndexCache* cache) {
 }
 
 int indexCachePut(void* cache, SIndexTerm* term, uint64_t uid) {
-  if (cache == NULL) {
-    return -1;
-  }
+  if (cache == NULL) { return -1; }
 
   IndexCache* pCache = cache;
   indexCacheRef(pCache);
   // encode data
   CacheTerm* ct = calloc(1, sizeof(CacheTerm));
-  if (cache == NULL) {
-    return -1;
-  }
+  if (cache == NULL) { return -1; }
   // set up key
   ct->colType = term->colType;
   ct->colVal = (char*)calloc(1, sizeof(char) * (term->nColVal + 1));
@@ -205,32 +229,11 @@ int indexCacheDel(void* cache, const char* fieldValue, int32_t fvlen, uint64_t u
   IndexCache* pCache = cache;
   return 0;
 }
-int indexCacheSearch(void* cache, SIndexTermQuery* query, SArray* result, STermValueType* s) {
-  if (cache == NULL) {
-    return -1;
-  }
-  IndexCache*     pCache = cache;
-  SIndexTerm*     term = query->term;
-  EIndexQueryType qtype = query->qType;
-
-  MemTable *mem = NULL, *imm = NULL;
-  pthread_mutex_lock(&pCache->mtx);
-  mem = pCache->mem;
-  imm = pCache->imm;
-  indexMemRef(mem);
-  indexMemRef(imm);
-  pthread_mutex_unlock(&pCache->mtx);
-
-  CacheTerm* ct = calloc(1, sizeof(CacheTerm));
-  if (ct == NULL) {
-    return -1;
-  }
-  ct->colVal = calloc(1, sizeof(char) * (term->nColVal + 1));
-  memcpy(ct->colVal, term->colVal, term->nColVal);
-  ct->version = atomic_load_32(&pCache->version);
 
+static int indexQueryMem(MemTable* mem, CacheTerm* ct, EIndexQueryType qtype, SArray* result, STermValueType* s) {
+  if (mem == NULL) { return 0; }
   char* key = getIndexKey(ct);
-  // TODO handle multi situation later, and refactor
+
   SSkipListIterator* iter = tSkipListCreateIterFromVal(mem->mem, key, TSDB_DATA_TYPE_BINARY, TSDB_ORDER_ASC);
   while (tSkipListIterNext(iter)) {
     SSkipListNode* node = tSkipListIterGet(iter);
@@ -251,51 +254,56 @@ int indexCacheSearch(void* cache, SIndexTermQuery* query, SArray* result, STermV
     }
   }
   tSkipListDestroyIter(iter);
-  cacheTermDestroy(ct);
-  // int32_t keyLen = CACHE_KEY_LEN(term);
-  // char*   buf = calloc(1, keyLen);
-  if (qtype == QUERY_TERM) {
-    //
-  } else if (qtype == QUERY_PREFIX) {
-    //
-  } else if (qtype == QUERY_SUFFIX) {
-    //
-  } else if (qtype == QUERY_REGEX) {
-    //
+  return 0;
+}
+int indexCacheSearch(void* cache, SIndexTermQuery* query, SArray* result, STermValueType* s) {
+  if (cache == NULL) { return -1; }
+  IndexCache* pCache = cache;
+
+  MemTable *mem = NULL, *imm = NULL;
+  pthread_mutex_lock(&pCache->mtx);
+  mem = pCache->mem;
+  imm = pCache->imm;
+  indexMemRef(mem);
+  indexMemRef(imm);
+  pthread_mutex_unlock(&pCache->mtx);
+
+  SIndexTerm*     term = query->term;
+  EIndexQueryType qtype = query->qType;
+  CacheTerm       ct = {.colVal = term->colVal, .version = atomic_load_32(&pCache->version)};
+  indexCacheDebug(pCache);
+
+  int ret = indexQueryMem(mem, &ct, qtype, result, s);
+  if (ret == 0 && *s != kTypeDeletion) {
+    // continue search in imm
+    ret = indexQueryMem(imm, &ct, qtype, result, s);
   }
+  // cacheTermDestroy(ct);
+
   indexMemUnRef(mem);
   indexMemUnRef(imm);
-  return 0;
+
+  return ret;
 }
 
 void indexCacheRef(IndexCache* cache) {
-  if (cache == NULL) {
-    return;
-  }
+  if (cache == NULL) { return; }
   int ref = T_REF_INC(cache);
   UNUSED(ref);
 }
 void indexCacheUnRef(IndexCache* cache) {
-  if (cache == NULL) {
-    return;
-  }
+  if (cache == NULL) { return; }
   int ref = T_REF_DEC(cache);
-  if (ref == 0) {
-    indexCacheDestroy(cache);
-  }
+  if (ref == 0) { indexCacheDestroy(cache); }
 }
 
 void indexMemRef(MemTable* tbl) {
-  if (tbl == NULL) {
-    return;
-  }
+  if (tbl == NULL) { return; }
   int ref = T_REF_INC(tbl);
   UNUSED(ref);
 }
 void indexMemUnRef(MemTable* tbl) {
-  if (tbl == NULL) {
-    return;
-  }
+  if (tbl == NULL) { return; }
   int ref = T_REF_DEC(tbl);
   if (ref == 0) {
     SSkipList* slt = tbl->mem;
@@ -305,9 +313,7 @@ void indexMemUnRef(MemTable* tbl) {
 }
 
 static void cacheTermDestroy(CacheTerm* ct) {
-  if (ct == NULL) {
-    return;
-  }
+  if (ct == NULL) { return; }
   free(ct->colVal);
   free(ct);
 }
@@ -322,9 +328,7 @@ static int32_t compareKey(const void* l, const void* r) {
 
   // compare colVal
   int32_t cmp = strcmp(lt->colVal, rt->colVal);
-  if (cmp == 0) {
-    return rt->version - lt->version;
-  }
+  if (cmp == 0) { return rt->version - lt->version; }
   return cmp;
 }
 
@@ -344,9 +348,7 @@ static void doMergeWork(SSchedMsg* msg) {
 }
 static bool indexCacheIteratorNext(Iterate* itera) {
   SSkipListIterator* iter = itera->iter;
-  if (iter == NULL) {
-    return false;
-  }
+  if (iter == NULL) { return false; }
   IterateValue* iv = &itera->val;
   iterateValueDestroy(iv, false);
 
@@ -356,7 +358,8 @@ static bool indexCacheIteratorNext(Iterate* itera) {
     CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
 
     iv->type = ct->operaType;
-    iv->colVal = ct->colVal;
+    iv->colVal = calloc(1, strlen(ct->colVal) + 1);
+    memcpy(iv->colVal, ct->colVal, strlen(ct->colVal));
 
     taosArrayPush(iv->val, &ct->uid);
   }

source/libs/index/src/index_fst_automation.c

@@ -17,7 +17,9 @@
 
 StartWithStateValue* startWithStateValueCreate(StartWithStateKind kind, ValueType ty, void* val) {
   StartWithStateValue* nsv = calloc(1, sizeof(StartWithStateValue));
-  if (nsv == NULL) { return NULL; }
+  if (nsv == NULL) {
+    return NULL;
+  }
 
   nsv->kind = kind;
   nsv->type = ty;
@@ -35,7 +37,9 @@ StartWithStateValue* startWithStateValueCreate(StartWithStateKind kind, ValueTyp
 }
 void startWithStateValueDestroy(void* val) {
   StartWithStateValue* sv = (StartWithStateValue*)val;
-  if (sv == NULL) { return; }
+  if (sv == NULL) {
+    return;
+  }
 
   if (sv->type == FST_INT) {
     //
@@ -48,7 +52,9 @@ void startWithStateValueDestroy(void* val) {
 }
 StartWithStateValue* startWithStateValueDump(StartWithStateValue* sv) {
   StartWithStateValue* nsv = calloc(1, sizeof(StartWithStateValue));
-  if (nsv == NULL) { return NULL; }
+  if (nsv == NULL) {
+    return NULL;
+  }
 
   nsv->kind = sv->kind;
   nsv->type = sv->type;
@@ -65,24 +71,12 @@ StartWithStateValue* startWithStateValueDump(StartWithStateValue* sv) {
 }
 
 // iterate fst
-static void* alwaysMatchStart(AutomationCtx* ctx) {
-  return NULL;
-}
-static bool alwaysMatchIsMatch(AutomationCtx* ctx, void* state) {
-  return true;
-}
-static bool alwaysMatchCanMatch(AutomationCtx* ctx, void* state) {
-  return true;
-}
-static bool alwaysMatchWillAlwaysMatch(AutomationCtx* ctx, void* state) {
-  return true;
-}
-static void* alwaysMatchAccpet(AutomationCtx* ctx, void* state, uint8_t byte) {
-  return NULL;
-}
-static void* alwaysMatchAccpetEof(AutomationCtx* ctx, void* state) {
-  return NULL;
-}
+static void* alwaysMatchStart(AutomationCtx* ctx) { return NULL; }
+static bool  alwaysMatchIsMatch(AutomationCtx* ctx, void* state) { return true; }
+static bool  alwaysMatchCanMatch(AutomationCtx* ctx, void* state) { return true; }
+static bool  alwaysMatchWillAlwaysMatch(AutomationCtx* ctx, void* state) { return true; }
+static void* alwaysMatchAccpet(AutomationCtx* ctx, void* state, uint8_t byte) { return NULL; }
+static void* alwaysMatchAccpetEof(AutomationCtx* ctx, void* state) { return NULL; }
 // prefix query, impl later
 
 static void* prefixStart(AutomationCtx* ctx) {
@@ -97,17 +91,20 @@ static bool prefixCanMatch(AutomationCtx* ctx, void* sv) {
   StartWithStateValue* ssv = (StartWithStateValue*)sv;
   return ssv->val >= 0;
 }
-static bool prefixWillAlwaysMatch(AutomationCtx* ctx, void* state) {
-  return true;
-}
+static bool  prefixWillAlwaysMatch(AutomationCtx* ctx, void* state) { return true; }
 static void* prefixAccept(AutomationCtx* ctx, void* state, uint8_t byte) {
   StartWithStateValue* ssv = (StartWithStateValue*)state;
-  if (ssv == NULL || ctx == NULL) { return NULL; }
+  if (ssv == NULL || ctx == NULL) {
+    return NULL;
+  }
 
   char* data = ctx->data;
-  if (ssv->kind == Done) { return startWithStateValueCreate(Done, FST_INT, &ssv->val); }
+  if (ssv->kind == Done) {
+    return startWithStateValueCreate(Done, FST_INT, &ssv->val);
+  }
   if ((strlen(data) > ssv->val) && data[ssv->val] == byte) {
     int val = ssv->val + 1;
+
     StartWithStateValue* nsv = startWithStateValueCreate(Running, FST_INT, &val);
     if (prefixIsMatch(ctx, nsv)) {
       nsv->kind = Done;
@@ -118,35 +115,22 @@ static void* prefixAccept(AutomationCtx* ctx, void* state, uint8_t byte) {
   }
   return NULL;
 }
-static void* prefixAcceptEof(AutomationCtx* ctx, void* state) {
-  return NULL;
-}
+static void* prefixAcceptEof(AutomationCtx* ctx, void* state) { return NULL; }
 
 // pattern query, impl later
 
-static void* patternStart(AutomationCtx* ctx) {
-  return NULL;
-}
-static bool patternIsMatch(AutomationCtx* ctx, void* data) {
-  return true;
-}
-static bool patternCanMatch(AutomationCtx* ctx, void* data) {
-  return true;
-}
-static bool patternWillAlwaysMatch(AutomationCtx* ctx, void* state) {
-  return true;
-}
+static void* patternStart(AutomationCtx* ctx) { return NULL; }
+static bool  patternIsMatch(AutomationCtx* ctx, void* data) { return true; }
+static bool  patternCanMatch(AutomationCtx* ctx, void* data) { return true; }
+static bool  patternWillAlwaysMatch(AutomationCtx* ctx, void* state) { return true; }
 
-static void* patternAccept(AutomationCtx* ctx, void* state, uint8_t byte) {
-  return NULL;
-}
+static void* patternAccept(AutomationCtx* ctx, void* state, uint8_t byte) { return NULL; }
 
-static void* patternAcceptEof(AutomationCtx* ctx, void* state) {
-  return NULL;
-}
+static void* patternAcceptEof(AutomationCtx* ctx, void* state) { return NULL; }
 
 AutomationFunc automFuncs[] = {
-    {alwaysMatchStart, alwaysMatchIsMatch, alwaysMatchCanMatch, alwaysMatchWillAlwaysMatch, alwaysMatchAccpet, alwaysMatchAccpetEof},
+    {alwaysMatchStart, alwaysMatchIsMatch, alwaysMatchCanMatch, alwaysMatchWillAlwaysMatch, alwaysMatchAccpet,
+     alwaysMatchAccpetEof},
     {prefixStart, prefixIsMatch, prefixCanMatch, prefixWillAlwaysMatch, prefixAccept, prefixAcceptEof},
     {patternStart, patternIsMatch, patternCanMatch, patternWillAlwaysMatch, patternAccept, patternAcceptEof}
     // add more search type
@@ -154,7 +138,9 @@ AutomationFunc automFuncs[] = {
 
 AutomationCtx* automCtxCreate(void* data, AutomationType atype) {
   AutomationCtx* ctx = calloc(1, sizeof(AutomationCtx));
-  if (ctx == NULL) { return NULL; }
+  if (ctx == NULL) {
+    return NULL;
+  }
 
   StartWithStateValue* sv = NULL;
   if (atype == AUTOMATION_ALWAYS) {
@@ -170,11 +156,14 @@ AutomationCtx* automCtxCreate(void* data, AutomationType atype) {
     // add more search type
   }
 
+  char* dst = NULL;
+  if (data != NULL) {
     char*  src = (char*)data;
     size_t len = strlen(src);
-  char*  dst = (char*)malloc(len * sizeof(char) + 1);
+    dst = (char*)malloc(len * sizeof(char) + 1);
     memcpy(dst, src, len);
     dst[len] = 0;
+  }
 
   ctx->data = dst;
   ctx->type = atype;

source/libs/index/src/index_fst_counting_writer.c

@@ -52,7 +52,7 @@ static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t off
 }
 static int writeCtxDoFlush(WriterCtx* ctx) {
   if (ctx->type == TFile) {
-    // tfFsync(ctx->fd);
+    tfFsync(ctx->file.fd);
     // tfFlush(ctx->file.fd);
   } else {
     // do nothing
@@ -73,6 +73,7 @@ WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int
     } else {
       ctx->file.fd = tfOpenReadWrite(path);
     }
+    memcpy(ctx->file.buf, path, strlen(path));
     if (ctx->file.fd < 0) {
       indexError("open file error %d", errno);
       goto END;
@@ -95,11 +96,15 @@ END:
   free(ctx);
   return NULL;
 }
-void writerCtxDestroy(WriterCtx* ctx) {
+void writerCtxDestroy(WriterCtx* ctx, bool remove) {
   if (ctx->type == TMemory) {
     free(ctx->mem.buf);
   } else {
     tfClose(ctx->file.fd);
+    if (remove) {
+      indexError("rm file %s", ctx->file.buf);
+      unlink(ctx->file.buf);
+    }
   }
   free(ctx);
 }
@@ -138,9 +143,7 @@ int fstCountingWriterRead(FstCountingWriter* write, uint8_t* buf, uint32_t len)
   return nRead;
 }
 
-uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write) {
-  return 0;
-}
+uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write) { return 0; }
 int      fstCountingWriterFlush(FstCountingWriter* write) {
   WriterCtx* ctx = write->wrt;
   ctx->flush(ctx);

source/libs/index/src/index_tfile.c

@@ -32,6 +32,7 @@ typedef struct TFileFstIter {
 
 #define TF_TABLE_TATOAL_SIZE(sz) (sizeof(sz) + sz * sizeof(uint64_t))
 
+static int  tfileUidCompare(const void* a, const void* b);
 static int  tfileStrCompare(const void* a, const void* b);
 static int  tfileValueCompare(const void* a, const void* b, const void* param);
 static void tfileSerialTableIdsToBuf(char* buf, SArray* tableIds);
@@ -52,11 +53,16 @@ static int  tfileParseFileName(const char* filename, uint64_t* suid, int* colId,
 static void tfileGenFileName(char* filename, uint64_t suid, int colId, int version);
 static void tfileSerialCacheKey(TFileCacheKey* key, char* buf);
 
+static TFileReader* tfileReaderCreateImpl(WriterCtx* ctx) {
+  TFileReader* reader = tfileReaderCreate(ctx);
+  tfileReaderRef(reader);
+  // tfileSerialCacheKey(&key, buf);
+  return reader;
+}
+
 TFileCache* tfileCacheCreate(const char* path) {
   TFileCache* tcache = calloc(1, sizeof(TFileCache));
-  if (tcache == NULL) {
-    return NULL;
-  }
+  if (tcache == NULL) { return NULL; }
 
   tcache->tableCache = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   tcache->capacity = 64;
@@ -75,22 +81,19 @@ TFileCache* tfileCacheCreate(const char* path) {
       continue;
     }
 
-    WriterCtx* wc = writerCtxCreate(TFile, file, true, 1024 * 64);
+    WriterCtx* wc = writerCtxCreate(TFile, file, true, 1024 * 1024 * 64);
     if (wc == NULL) {
       indexError("failed to open index:%s", file);
       goto End;
     }
 
-    TFileReader* reader = tfileReaderCreate(wc);
-    tfileReaderRef(reader);
-    // loader fst and validate it
+    char          buf[128] = {0};
+    TFileReader*  reader = tfileReaderCreateImpl(wc);
     TFileHeader*  header = &reader->header;
     TFileCacheKey key = {.suid = header->suid,
                          .colName = header->colName,
                          .nColName = strlen(header->colName),
                          .colType = header->colType};
-
-    char buf[128] = {0};
     tfileSerialCacheKey(&key, buf);
     taosHashPut(tcache->tableCache, buf, strlen(buf), &reader, sizeof(void*));
   }
@@ -102,9 +105,7 @@ End:
   return NULL;
 }
 void tfileCacheDestroy(TFileCache* tcache) {
-  if (tcache == NULL) {
-    return;
-  }
+  if (tcache == NULL) { return; }
 
   // free table cache
   TFileReader** reader = taosHashIterate(tcache->tableCache, NULL);
@@ -125,9 +126,7 @@ TFileReader* tfileCacheGet(TFileCache* tcache, TFileCacheKey* key) {
   tfileSerialCacheKey(key, buf);
 
   TFileReader** reader = taosHashGet(tcache->tableCache, buf, strlen(buf));
-  if (reader == NULL) {
-    return NULL;
-  }
+  if (reader == NULL) { return NULL; }
   tfileReaderRef(*reader);
 
   return *reader;
@@ -137,7 +136,7 @@ void tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader* reader)
   tfileSerialCacheKey(key, buf);
   // remove last version index reader
   TFileReader** p = taosHashGet(tcache->tableCache, buf, strlen(buf));
-  if (*p != NULL) {
+  if (p != NULL) {
     TFileReader* oldReader = *p;
     taosHashRemove(tcache->tableCache, buf, strlen(buf));
     tfileReaderUnRef(oldReader);
@@ -149,9 +148,7 @@ void tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader* reader)
 }
 TFileReader* tfileReaderCreate(WriterCtx* ctx) {
   TFileReader* reader = calloc(1, sizeof(TFileReader));
-  if (reader == NULL) {
-    return NULL;
-  }
+  if (reader == NULL) { return NULL; }
 
   // T_REF_INC(reader);
   reader->ctx = ctx;
@@ -172,12 +169,10 @@ TFileReader* tfileReaderCreate(WriterCtx* ctx) {
   return reader;
 }
 void tfileReaderDestroy(TFileReader* reader) {
-  if (reader == NULL) {
-    return;
-  }
+  if (reader == NULL) { return; }
   // T_REF_INC(reader);
   fstDestroy(reader->fst);
-  writerCtxDestroy(reader->ctx);
+  writerCtxDestroy(reader->ctx, true);
   free(reader);
 }
 
@@ -216,7 +211,8 @@ TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int32_t version, const c
 
   char fullname[256] = {0};
   snprintf(fullname, sizeof(fullname), "%s/%s", path, filename);
-  WriterCtx* wcx = writerCtxCreate(TFile, fullname, true, 1024 * 1024);
+  WriterCtx* wcx = writerCtxCreate(TFile, fullname, false, 1024 * 1024 * 64);
+  if (wcx == NULL) { return NULL; }
 
   TFileHeader tfh = {0};
   tfh.suid = suid;
@@ -226,6 +222,21 @@ TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int32_t version, const c
 
   return tfileWriterCreate(wcx, &tfh);
 }
+TFileReader* tfileReaderOpen(char* path, uint64_t suid, int32_t version, const char* colName) {
+  char    filename[128] = {0};
+  int32_t coldId = 1;
+  tfileGenFileName(filename, suid, coldId, version);
+
+  char fullname[256] = {0};
+  snprintf(fullname, sizeof(fullname), "%s/%s", path, filename);
+  WriterCtx* wc = writerCtxCreate(TFile, fullname, true, 1024 * 1024 * 1024);
+  if (wc == NULL) { return NULL; }
+
+  TFileReader* reader = tfileReaderCreateImpl(wc);
+  return reader;
+
+  // tfileSerialCacheKey(&key, buf);
+}
 TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header) {
   // char pathBuf[128] = {0};
   // sprintf(pathBuf, "%s/% " PRIu64 "-%d-%d.tindex", path, suid, colId, version);
@@ -249,10 +260,10 @@ TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header) {
   return tw;
 }
 
-int tfileWriterPut(TFileWriter* tw, void* data) {
+int tfileWriterPut(TFileWriter* tw, void* data, bool order) {
   // sort by coltype and write to tindex
+  if (order == false) {
     __compar_fn_t fn;
-
     int8_t        colType = tw->header.colType;
     if (colType == TSDB_DATA_TYPE_BINARY || colType == TSDB_DATA_TYPE_NCHAR) {
       fn = tfileStrCompare;
@@ -260,17 +271,18 @@ int tfileWriterPut(TFileWriter* tw, void* data) {
       fn = getComparFunc(colType, 0);
     }
     taosArraySortPWithExt((SArray*)(data), tfileValueCompare, &fn);
+  }
 
-  int32_t bufLimit = 4096, offset = 0;
-  char*   buf = calloc(1, sizeof(char) * bufLimit);
-  char*   p = buf;
+  int32_t bufLimit = 64 * 4096, offset = 0;
+  // char*   buf = calloc(1, sizeof(char) * bufLimit);
+  // char*   p = buf;
   int32_t sz = taosArrayGetSize((SArray*)data);
   int32_t fstOffset = tw->offset;
 
   // ugly code, refactor later
   for (size_t i = 0; i < sz; i++) {
     TFileValue* v = taosArrayGetP((SArray*)data, i);
-
+    // taosArrayRemoveDuplicate(v->tablId, tfileUidCompare, NULL);
     int32_t tbsz = taosArrayGetSize(v->tableId);
     fstOffset += TF_TABLE_TATOAL_SIZE(tbsz);
   }
@@ -282,30 +294,39 @@ int tfileWriterPut(TFileWriter* tw, void* data) {
     int32_t tbsz = taosArrayGetSize(v->tableId);
     // check buf has enough space or not
     int32_t ttsz = TF_TABLE_TATOAL_SIZE(tbsz);
-    if (offset + ttsz > bufLimit) {
-      // batch write
-      tw->ctx->write(tw->ctx, buf, offset);
-      offset = 0;
-      memset(buf, 0, bufLimit);
-      p = buf;
-    }
 
+    // if (offset + ttsz >= bufLimit) {
+    //  // batch write
+    //  indexInfo("offset: %d, ttsz: %d", offset, ttsz);
+    //  // std::cout << "offset: " << offset << std::endl;
+    //  // std::cout << "ttsz:" << ttsz < < < std::endl;
+    //  tw->ctx->write(tw->ctx, buf, offset);
+    //  offset = 0;
+    //  memset(buf, 0, bufLimit);
+    //  p = buf;
+    //}
+    // if (ttsz >= bufLimit) {
+    //}
+    char* buf = calloc(1, ttsz * sizeof(char));
+    char* p = buf;
     tfileSerialTableIdsToBuf(p, v->tableId);
-    offset += ttsz;
-    p = buf + offset;
+    tw->ctx->write(tw->ctx, buf, ttsz);
+    // offset += ttsz;
+    // p = buf + offset;
     // set up value offset
     v->offset = tw->offset;
     tw->offset += ttsz;
+    free(buf);
   }
-  if (offset != 0) {
+  // if (offset != 0) {
   // write reversed data in buf to tindex
-    tw->ctx->write(tw->ctx, buf, offset);
-  }
-  tfree(buf);
+  // tw->ctx->write(tw->ctx, buf, offset);
+  //}
+  // tfree(buf);
 
   tw->fb = fstBuilderCreate(tw->ctx, 0);
   if (tw->fb == NULL) {
-    tfileWriterDestroy(tw);
+    tfileWriterClose(tw);
     return -1;
   }
   // write fst
@@ -321,27 +342,20 @@ int tfileWriterPut(TFileWriter* tw, void* data) {
   tw->fb = NULL;
   return 0;
 }
-void tfileWriteClose(TFileWriter* tw) {
-  if (tw == NULL) {
-    return;
-  }
-  writerCtxDestroy(tw->ctx);
+void tfileWriterClose(TFileWriter* tw) {
+  if (tw == NULL) { return; }
+  writerCtxDestroy(tw->ctx, false);
   free(tw);
 }
 void tfileWriterDestroy(TFileWriter* tw) {
-  if (tw == NULL) {
-    return;
-  }
-
-  writerCtxDestroy(tw->ctx);
+  if (tw == NULL) { return; }
+  writerCtxDestroy(tw->ctx, false);
   free(tw);
 }
 
 IndexTFile* indexTFileCreate(const char* path) {
   IndexTFile* tfile = calloc(1, sizeof(IndexTFile));
-  if (tfile == NULL) {
-    return NULL;
-  }
+  if (tfile == NULL) { return NULL; }
 
   tfile->cache = tfileCacheCreate(path);
   return tfile;
@@ -350,18 +364,14 @@ void IndexTFileDestroy(IndexTFile* tfile) { free(tfile); }
 
 int indexTFileSearch(void* tfile, SIndexTermQuery* query, SArray* result) {
   int ret = -1;
-  if (tfile == NULL) {
-    return ret;
-  }
+  if (tfile == NULL) { return ret; }
   IndexTFile* pTfile = (IndexTFile*)tfile;
 
   SIndexTerm*   term = query->term;
   TFileCacheKey key = {
       .suid = term->suid, .colType = term->colType, .colName = term->colName, .nColName = term->nColName};
   TFileReader* reader = tfileCacheGet(pTfile->cache, &key);
-  if (reader == NULL) {
-    return 0;
-  }
+  if (reader == NULL) { return 0; }
 
   return tfileReaderSearch(reader, query, result);
 }
@@ -381,9 +391,7 @@ static bool tfileIteratorNext(Iterate* iiter) {
 
   TFileFstIter*          tIter = iiter->iter;
   StreamWithStateResult* rt = streamWithStateNextWith(tIter->st, NULL);
-  if (rt == NULL) {
-    return false;
-  }
+  if (rt == NULL) { return false; }
 
   int32_t sz = 0;
   char*   ch = (char*)fstSliceData(&rt->data, &sz);
@@ -391,12 +399,9 @@ static bool tfileIteratorNext(Iterate* iiter) {
   memcpy(colVal, ch, sz);
 
   offset = (uint64_t)(rt->out.out);
-
   swsResultDestroy(rt);
   // set up iterate value
-  if (tfileReaderLoadTableIds(tIter->rdr, offset, iv->val) != 0) {
-    return false;
-  }
+  if (tfileReaderLoadTableIds(tIter->rdr, offset, iv->val) != 0) { return false; }
 
   iv->colVal = colVal;
 
@@ -407,9 +412,8 @@ static IterateValue* tifileIterateGetValue(Iterate* iter) { return &iter->val; }
 
 static TFileFstIter* tfileFstIteratorCreate(TFileReader* reader) {
   TFileFstIter* tIter = calloc(1, sizeof(Iterate));
-  if (tIter == NULL) {
-    return NULL;
-  }
+  if (tIter == NULL) { return NULL; }
+
   tIter->ctx = automCtxCreate(NULL, AUTOMATION_ALWAYS);
   tIter->fb = fstSearch(reader->fst, tIter->ctx);
   tIter->st = streamBuilderIntoStream(tIter->fb);
@@ -418,21 +422,22 @@ static TFileFstIter* tfileFstIteratorCreate(TFileReader* reader) {
 }
 
 Iterate* tfileIteratorCreate(TFileReader* reader) {
-  Iterate* iter = calloc(1, sizeof(Iterate));
+  if (reader == NULL) { return NULL; }
 
+  Iterate* iter = calloc(1, sizeof(Iterate));
   iter->iter = tfileFstIteratorCreate(reader);
   if (iter->iter == NULL) {
+    free(iter);
     return NULL;
   }
-
   iter->next = tfileIteratorNext;
   iter->getValue = tifileIterateGetValue;
+  iter->val.val = taosArrayInit(1, sizeof(uint64_t));
   return iter;
 }
 void tfileIteratorDestroy(Iterate* iter) {
-  if (iter == NULL) {
-    return;
-  }
+  if (iter == NULL) { return; }
+
   IterateValue* iv = &iter->val;
   iterateValueDestroy(iv, true);
 
@@ -445,18 +450,19 @@ void tfileIteratorDestroy(Iterate* iter) {
 }
 
 TFileReader* tfileGetReaderByCol(IndexTFile* tf, char* colName) {
-  if (tf == NULL) {
-    return NULL;
-  }
+  if (tf == NULL) { return NULL; }
   TFileCacheKey key = {.suid = 0, .colType = TSDB_DATA_TYPE_BINARY, .colName = colName, .nColName = strlen(colName)};
   return tfileCacheGet(tf->cache, &key);
 }
 
+static int tfileUidCompare(const void* a, const void* b) {
+  uint64_t l = *(uint64_t*)a;
+  uint64_t r = *(uint64_t*)b;
+  return l - r;
+}
 static int tfileStrCompare(const void* a, const void* b) {
   int ret = strcmp((char*)a, (char*)b);
-  if (ret == 0) {
-    return ret;
-  }
+  if (ret == 0) { return ret; }
   return ret < 0 ? -1 : 1;
 }
 
@@ -471,17 +477,13 @@ static int tfileValueCompare(const void* a, const void* b, const void* param) {
 
 TFileValue* tfileValueCreate(char* val) {
   TFileValue* tf = calloc(1, sizeof(TFileValue));
-  if (tf == NULL) {
-    return NULL;
-  }
-
+  if (tf == NULL) { return NULL; }
+  tf->colVal = val;
   tf->tableId = taosArrayInit(32, sizeof(uint64_t));
   return tf;
 }
 int tfileValuePush(TFileValue* tf, uint64_t val) {
-  if (tf == NULL) {
-    return -1;
-  }
+  if (tf == NULL) { return -1; }
   taosArrayPush(tf->tableId, &val);
   return 0;
 }
@@ -501,9 +503,7 @@ static void tfileSerialTableIdsToBuf(char* buf, SArray* ids) {
 static int tfileWriteFstOffset(TFileWriter* tw, int32_t offset) {
   int32_t fstOffset = offset + sizeof(tw->header.fstOffset);
   tw->header.fstOffset = fstOffset;
-  if (sizeof(fstOffset) != tw->ctx->write(tw->ctx, (char*)&fstOffset, sizeof(fstOffset))) {
-    return -1;
-  }
+  if (sizeof(fstOffset) != tw->ctx->write(tw->ctx, (char*)&fstOffset, sizeof(fstOffset))) { return -1; }
   tw->offset += sizeof(fstOffset);
   return 0;
 }
@@ -514,9 +514,7 @@ static int tfileWriteHeader(TFileWriter* writer) {
   memcpy(buf, (char*)header, sizeof(buf));
 
   int nwrite = writer->ctx->write(writer->ctx, buf, sizeof(buf));
-  if (sizeof(buf) != nwrite) {
-    return -1;
-  }
+  if (sizeof(buf) != nwrite) { return -1; }
   writer->offset = nwrite;
   return 0;
 }
@@ -547,12 +545,10 @@ static int tfileReaderLoadHeader(TFileReader* reader) {
 }
 static int tfileReaderLoadFst(TFileReader* reader) {
   // current load fst into memory, refactor it later
-  static int FST_MAX_SIZE = 16 * 1024;
+  static int FST_MAX_SIZE = 64 * 1024;
 
   char* buf = calloc(1, sizeof(char) * FST_MAX_SIZE);
-  if (buf == NULL) {
-    return -1;
-  }
+  if (buf == NULL) { return -1; }
 
   WriterCtx* ctx = reader->ctx;
   int32_t    nread = ctx->readFrom(ctx, buf, FST_MAX_SIZE, reader->header.fstOffset);
@@ -575,36 +571,33 @@ static int tfileReaderLoadTableIds(TFileReader* reader, int32_t offset, SArray*
 
   int32_t total = sizeof(uint64_t) * nid;
   char*   buf = calloc(1, total);
-  if (buf == NULL) {
-    return -1;
-  }
+  if (buf == NULL) { return -1; }
 
   nread = ctx->read(ctx, buf, total);
   assert(total == nread);
 
-  for (int32_t i = 0; i < nid; i++) {
-    taosArrayPush(result, (uint64_t*)buf + i);
-  }
+  for (int32_t i = 0; i < nid; i++) { taosArrayPush(result, (uint64_t*)buf + i); }
   free(buf);
   return 0;
 }
 void tfileReaderRef(TFileReader* reader) {
+  if (reader == NULL) { return; }
   int ref = T_REF_INC(reader);
   UNUSED(ref);
 }
 
 void tfileReaderUnRef(TFileReader* reader) {
+  if (reader == NULL) { return; }
   int ref = T_REF_DEC(reader);
   if (ref == 0) {
+    // do nothing
     tfileReaderDestroy(reader);
   }
 }
 
 static int tfileGetFileList(const char* path, SArray* result) {
   DIR* dir = opendir(path);
-  if (NULL == dir) {
-    return -1;
-  }
+  if (NULL == dir) { return -1; }
 
   struct dirent* entry;
   while ((entry = readdir(dir)) != NULL) {
@@ -632,9 +625,7 @@ static int tfileCompare(const void* a, const void* b) {
   size_t bLen = strlen(bName);
 
   int ret = strncmp(aName, bName, aLen > bLen ? aLen : bLen);
-  if (ret == 0) {
-    return ret;
-  }
+  if (ret == 0) { return ret; }
   return ret < 0 ? -1 : 1;
 }
 // tfile name suid-colId-version.tindex

source/libs/index/test/fstTest.cc

@@ -12,9 +12,7 @@
 #include "tskiplist.h"
 #include "tutil.h"
 
-void* callback(void* s) {
-  return s;
-}
+void* callback(void* s) { return s; }
 
 static std::string fileName = "/tmp/tindex.tindex";
 class FstWriter {
@@ -34,7 +32,7 @@ class FstWriter {
     fstBuilderFinish(_b);
     fstBuilderDestroy(_b);
 
-    writerCtxDestroy(_wc);
+    writerCtxDestroy(_wc, false);
   }
 
  private:
@@ -102,7 +100,7 @@ class FstReadMemory {
     fstCountingWriterDestroy(_w);
     fstDestroy(_fst);
     fstSliceDestroy(&_s);
-    writerCtxDestroy(_wc);
+    writerCtxDestroy(_wc, false);
   }
 
  private:

source/libs/index/test/indexTests.cc

@@ -57,7 +57,7 @@ class FstWriter {
     fstBuilderFinish(_b);
     fstBuilderDestroy(_b);
 
-    writerCtxDestroy(_wc);
+    writerCtxDestroy(_wc, false);
   }
 
  private:
@@ -76,9 +76,7 @@ class FstReadMemory {
   bool init() {
     char* buf = (char*)calloc(1, sizeof(char) * _size);
     int   nRead = fstCountingWriterRead(_w, (uint8_t*)buf, _size);
-    if (nRead <= 0) {
-      return false;
-    }
+    if (nRead <= 0) { return false; }
     _size = nRead;
     _s = fstSliceCreate((uint8_t*)buf, _size);
     _fst = fstCreate(&_s);
@@ -104,9 +102,7 @@ class FstReadMemory {
     StreamWithState*       st = streamBuilderIntoStream(sb);
     StreamWithStateResult* rt = NULL;
 
-    while ((rt = streamWithStateNextWith(st, NULL)) != NULL) {
-      result.push_back((uint64_t)(rt->out.out));
-    }
+    while ((rt = streamWithStateNextWith(st, NULL)) != NULL) { result.push_back((uint64_t)(rt->out.out)); }
     return true;
   }
   bool SearchWithTimeCostUs(AutomationCtx* ctx, std::vector<uint64_t>& result) {
@@ -120,7 +116,7 @@ class FstReadMemory {
     fstCountingWriterDestroy(_w);
     fstDestroy(_fst);
     fstSliceDestroy(&_s);
-    writerCtxDestroy(_wc);
+    writerCtxDestroy(_wc, true);
   }
 
  private:
@@ -182,9 +178,7 @@ void checkFstPerf() {
   delete fw;
 
   FstReadMemory* m = new FstReadMemory(1024 * 64);
-  if (m->init()) {
-    printf("success to init fst read");
-  }
+  if (m->init()) { printf("success to init fst read"); }
   Performance_fstReadRecords(m);
   delete m;
 }
@@ -346,10 +340,8 @@ class TFileObj {
       tfileReaderDestroy(reader_);
       reader_ = NULL;
     }
-    if (writer_ == NULL) {
-      InitWriter();
-    }
-    return tfileWriterPut(writer_, tv);
+    if (writer_ == NULL) { InitWriter(); }
+    return tfileWriterPut(writer_, tv, false);
   }
   bool InitWriter() {
     TFileHeader header;
@@ -388,12 +380,8 @@ class TFileObj {
     return tfileReaderSearch(reader_, query, result);
   }
   ~TFileObj() {
-    if (writer_) {
-      tfileWriterDestroy(writer_);
-    }
-    if (reader_) {
-      tfileReaderDestroy(reader_);
-    }
+    if (writer_) { tfileWriterDestroy(writer_); }
+    if (reader_) { tfileReaderDestroy(reader_); }
   }
 
  private:
@@ -439,7 +427,7 @@ static TFileValue* genTFileValue(const char* val) {
   memcpy(tv->colVal, val, vlen);
 
   tv->tableId = (SArray*)taosArrayInit(1, sizeof(uint64_t));
-  for (size_t i = 0; i < 10; i++) {
+  for (size_t i = 0; i < 200; i++) {
     uint64_t v = i;
     taosArrayPush(tv->tableId, &v);
   }
@@ -452,22 +440,17 @@ static void destroyTFileValue(void* val) {
   free(tv);
 }
 TEST_F(IndexTFileEnv, test_tfile_write) {
-  TFileValue* v1 = genTFileValue("c");
-  TFileValue* v2 = genTFileValue("ab");
-  TFileValue* v3 = genTFileValue("b");
-  TFileValue* v4 = genTFileValue("d");
+  TFileValue* v1 = genTFileValue("ab");
 
   SArray* data = (SArray*)taosArrayInit(4, sizeof(void*));
 
   taosArrayPush(data, &v1);
-  taosArrayPush(data, &v2);
-  taosArrayPush(data, &v3);
-  taosArrayPush(data, &v4);
+  // taosArrayPush(data, &v2);
+  // taosArrayPush(data, &v3);
+  // taosArrayPush(data, &v4);
 
   fObj->Put(data);
-  for (size_t i = 0; i < taosArrayGetSize(data); i++) {
-    destroyTFileValue(taosArrayGetP(data, i));
-  }
+  for (size_t i = 0; i < taosArrayGetSize(data); i++) { destroyTFileValue(taosArrayGetP(data, i)); }
   taosArrayDestroy(data);
 
   std::string     colName("voltage");
@@ -478,7 +461,7 @@ TEST_F(IndexTFileEnv, test_tfile_write) {
 
   SArray* result = (SArray*)taosArrayInit(1, sizeof(uint64_t));
   fObj->Get(&query, result);
-  assert(taosArrayGetSize(result) == 10);
+  assert(taosArrayGetSize(result) == 200);
   indexTermDestroy(term);
 
   // tfileWriterDestroy(twrite);
@@ -569,7 +552,8 @@ TEST_F(IndexCacheEnv, cache_test) {
                                        colVal.c_str(), colVal.size());
     coj->Put(term, colId, version++, suid++);
   }
-
+  coj->Debug();
+  std::cout << "--------first----------" << std::endl;
   {
     std::string colVal("v3");
     SIndexTerm* term = indexTermCreate(0, ADD_VALUE, TSDB_DATA_TYPE_BINARY, colName.c_str(), colName.size(),
@@ -582,6 +566,8 @@ TEST_F(IndexCacheEnv, cache_test) {
                                        colVal.c_str(), colVal.size());
     coj->Put(term, othColId, version++, suid++);
   }
+  coj->Debug();
+  std::cout << "--------second----------" << std::endl;
   {
     std::string colVal("v4");
     for (size_t i = 0; i < 10; i++) {
@@ -602,7 +588,7 @@ TEST_F(IndexCacheEnv, cache_test) {
     STermValueType  valType;
 
     coj->Get(&query, colId, 10000, ret, &valType);
-    // std::cout << "size : " << taosArrayGetSize(ret) << std::endl;
+    std::cout << "size : " << taosArrayGetSize(ret) << std::endl;
     assert(taosArrayGetSize(ret) == 4);
   }
   {
@@ -635,6 +621,20 @@ class IndexObj {
     }
     return ret;
   }
+  int WriteMillonData(const std::string& colName, const std::string& colVal = "Hello world",
+                      size_t numOfTable = 100 * 10000) {
+    SIndexTerm*      term = indexTermCreate(0, ADD_VALUE, TSDB_DATA_TYPE_BINARY, colName.c_str(), colName.size(),
+                                       colVal.c_str(), colVal.size());
+    SIndexMultiTerm* terms = indexMultiTermCreate();
+    indexMultiTermAdd(terms, term);
+    for (size_t i = 0; i < numOfTable; i++) {
+      int ret = Put(terms, i);
+      assert(ret == 0);
+    }
+    indexMultiTermDestroy(terms);
+    return numOfTable;
+  }
+
   int Put(SIndexMultiTerm* fvs, uint64_t uid) {
     numOfWrite += taosArrayGetSize(fvs);
     return indexPut(idx, fvs, uid);
@@ -645,14 +645,25 @@ class IndexObj {
     return indexSearch(idx, multiQ, result);
   }
 
+  int SearchOne(const std::string& colName, const std::string& colVal) {
+    SIndexMultiTermQuery* mq = indexMultiTermQueryCreate(MUST);
+    SIndexTerm*           term = indexTermCreate(0, ADD_VALUE, TSDB_DATA_TYPE_BINARY, colName.c_str(), colName.size(),
+                                       colVal.c_str(), colVal.size());
+    indexMultiTermQueryAdd(mq, term, QUERY_TERM);
+
+    SArray* result = (SArray*)taosArrayInit(1, sizeof(uint64_t));
+    if (Search(mq, result) == 0) { std::cout << "search one successfully" << std::endl; }
+    return taosArrayGetSize(result);
+    // assert(taosArrayGetSize(result) == targetSize);
+  }
   void Debug() {
     std::cout << "numOfWrite:" << numOfWrite << std::endl;
     std::cout << "numOfRead:" << numOfRead << std::endl;
   }
 
   ~IndexObj() {
-    indexClose(idx);
     indexCleanUp();
+    indexClose(idx);
   }
 
  private:
@@ -665,12 +676,13 @@ class IndexObj {
 class IndexEnv2 : public ::testing::Test {
  protected:
   virtual void SetUp() {
+    tfInit();
     index = new IndexObj();
     //
   }
   virtual void TearDown() {
-    // r
     delete index;
+    tfCleanup();
   }
   IndexObj* index;
 };
@@ -681,9 +693,9 @@ TEST_F(IndexEnv2, testIndexOpen) {
     exit(1);
   }
 
-  int targetSize = 100;
+  int targetSize = 200;
   {
-    std::string colName("tag1"), colVal("Hello world");
+    std::string colName("tag1"), colVal("Hello");
 
     SIndexTerm*      term = indexTermCreate(0, ADD_VALUE, TSDB_DATA_TYPE_BINARY, colName.c_str(), colName.size(),
                                        colVal.c_str(), colVal.size());
@@ -697,8 +709,8 @@ TEST_F(IndexEnv2, testIndexOpen) {
     indexMultiTermDestroy(terms);
   }
   {
-    size_t      size = 100;
-    std::string colName("tag1"), colVal("hello world");
+    size_t      size = 200;
+    std::string colName("tag1"), colVal("hello");
 
     SIndexTerm*      term = indexTermCreate(0, ADD_VALUE, TSDB_DATA_TYPE_BINARY, colName.c_str(), colName.size(),
                                        colVal.c_str(), colVal.size());
@@ -713,7 +725,7 @@ TEST_F(IndexEnv2, testIndexOpen) {
   }
 
   {
-    std::string colName("tag1"), colVal("Hello world");
+    std::string colName("tag1"), colVal("Hello");
 
     SIndexMultiTermQuery* mq = indexMultiTermQueryCreate(MUST);
     SIndexTerm*           term = indexTermCreate(0, ADD_VALUE, TSDB_DATA_TYPE_BINARY, colName.c_str(), colName.size(),
@@ -722,27 +734,39 @@ TEST_F(IndexEnv2, testIndexOpen) {
 
     SArray* result = (SArray*)taosArrayInit(1, sizeof(uint64_t));
     index->Search(mq, result);
-    assert(taosArrayGetSize(result) == targetSize);
-  }
-}
-TEST_F(IndexEnv2, testIndex_CachePut) {
-  std::string path = "/tmp";
-  if (index->Init(path) != 0) {
+    std::cout << "target size: " << taosArrayGetSize(result) << std::endl;
+    // assert(taosArrayGetSize(result) == targetSize);
   }
 }
 
-TEST_F(IndexEnv2, testIndexr_TFilePut) {
+TEST_F(IndexEnv2, testIndex_TrigeFlush) {
   std::string path = "/tmp";
-  if (index->Init(path) != 0) {
-  }
+  if (index->Init(path) != 0) {}
+  int numOfTable = 100 * 10000;
+  index->WriteMillonData("tag1", "Hello world", numOfTable);
+  int target = index->SearchOne("tag1", "Hello world");
+  assert(numOfTable == target);
 }
-TEST_F(IndexEnv2, testIndex_CacheSearch) {
+TEST_F(IndexEnv2, testIndex_serarch_cache_and_tfile) {
   std::string path = "/tmp";
-  if (index->Init(path) != 0) {
-  }
+  if (index->Init(path) != 0) {}
 }
-TEST_F(IndexEnv2, testIndex_TFileSearch) {
+TEST_F(IndexEnv2, testIndex_multi_thread_write) {
   std::string path = "/tmp";
-  if (index->Init(path) != 0) {
-  }
+  if (index->Init(path) != 0) {}
 }
+TEST_F(IndexEnv2, testIndex_multi_thread_read) {
+  std::string path = "/tmp";
+  if (index->Init(path) != 0) {}
+}
+
+TEST_F(IndexEnv2, testIndex_restart) {
+  std::string path = "/tmp";
+  if (index->Init(path) != 0) {}
+}
+
+TEST_F(IndexEnv2, testIndex_performance) {
+  std::string path = "/tmp";
+  if (index->Init(path) != 0) {}
+}
+TEST_F(IndexEnv2, testIndexMultiTag) {}

source/libs/qworker/inc/qworkerInt.h

@@ -106,11 +106,11 @@ typedef struct SQWorkerMgmt {
   if (QW_READ == (type)) {          \
     if ((*(_lock)) < 0) assert(0);    \
     taosRLockLatch(_lock);          \
-    qDebug("RLOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
+    qDebug("QW RLOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
   } else {                                                \
     if ((*(_lock)) < 0) assert(0);                          \
     taosWLockLatch(_lock);                                \
-    qDebug("WLOCK%p, %s:%d", (_lock), __FILE__, __LINE__);  \
+    qDebug("QW WLOCK%p, %s:%d", (_lock), __FILE__, __LINE__);  \
   }                                                       \
 } while (0)
 
@@ -118,11 +118,11 @@ typedef struct SQWorkerMgmt {
   if (QW_READ == (type)) {                                \
     if ((*(_lock)) <= 0) assert(0);                         \
     taosRUnLockLatch(_lock);                              \
-    qDebug("RULOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
+    qDebug("QW RULOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
   } else {                                                \
     if ((*(_lock)) <= 0) assert(0);                         \
     taosWUnLockLatch(_lock);                              \
-    qDebug("WULOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
+    qDebug("QW WULOCK%p, %s:%d", (_lock), __FILE__, __LINE__); \
   }                                                       \
 } while (0)
 

source/util/CMakeLists.txt

@@ -13,3 +13,5 @@ target_link_libraries(
     PUBLIC lz4_static
     PUBLIC api
 )
+
+ADD_SUBDIRECTORY(test)

source/util/src/terror.c

@@ -182,10 +182,15 @@ TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_DNODE_CFG,        "Invalid dnode cfg")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_DNODE_EP,         "Invalid dnode end point")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_DNODE_ID,         "Invalid dnode id")
 
-// mnode-mnode
+// mnode-node
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_MNODE_ALREADY_EXIST,      "Mnode already exists")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_MNODE_NOT_EXIST,          "Mnode not there")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_TOO_MANY_MNODES,          "Too many mnodes")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_QNODE_ALREADY_EXIST,      "Qnode already exists")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_QNODE_NOT_EXIST,          "Qnode not there")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_SNODE_ALREADY_EXIST,      "Snode already exists")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_SNODE_NOT_EXIST,          "Snode not there")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_BNODE_ALREADY_EXIST,      "Bnode already exists")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_BNODE_NOT_EXIST,          "Bnode not there")
 
 // mnode-acct
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_ACCT_ALREADY_EXIST,       "Account already exists")

source/util/src/thash.c

@@ -362,7 +362,7 @@ void* taosHashGetCloneExt(SHashObj *pHashObj, const void *key, size_t keyLen, vo
   return data;
 }
 
-void* taosHashGetClone(SHashObj *pHashObj, const void *key, size_t keyLen, void* d) {
+void* taosHashGetCloneImpl(SHashObj *pHashObj, const void *key, size_t keyLen, void* d, bool acquire) {
   if (taosHashTableEmpty(pHashObj) || keyLen == 0 || key == NULL) {
     return NULL;
   }
@@ -404,6 +404,10 @@ void* taosHashGetClone(SHashObj *pHashObj, const void *key, size_t keyLen, void*
       memcpy(d, GET_HASH_NODE_DATA(pNode), pNode->dataLen);
     }
 
+    if (acquire) {
+      pNode->count++;
+    }
+
     data = GET_HASH_NODE_DATA(pNode);
   }
 
@@ -415,6 +419,15 @@ void* taosHashGetClone(SHashObj *pHashObj, const void *key, size_t keyLen, void*
   return data;
 }
 
+void* taosHashGetClone(SHashObj *pHashObj, const void *key, size_t keyLen, void* d) {
+  return taosHashGetCloneImpl(pHashObj, key, keyLen, d, false);
+}
+
+void* taosHashAcquire(SHashObj *pHashObj, const void *key, size_t keyLen) {
+  return taosHashGetCloneImpl(pHashObj, key, keyLen, NULL, true);
+}
+
+
 int32_t taosHashRemove(SHashObj *pHashObj, const void *key, size_t keyLen/*, void *data, size_t dsize*/) {
   if (pHashObj == NULL || taosHashTableEmpty(pHashObj)) {
     return -1;
@@ -919,3 +932,9 @@ void taosHashCancelIterate(SHashObj *pHashObj, void *p) {
 
   __rd_unlock(&pHashObj->lock, pHashObj->type);
 }
+
+void taosHashRelease(SHashObj *pHashObj, void *p) {
+  taosHashCancelIterate(pHashObj, p);
+}
+
+

source/util/test/CMakeLists.txt

@@ -13,17 +13,22 @@ IF (HEADER_GTEST_INCLUDE_DIR AND (LIB_GTEST_STATIC_DIR OR LIB_GTEST_SHARED_DIR))
 
     LIST(REMOVE_ITEM SOURCE_LIST ${CMAKE_CURRENT_SOURCE_DIR}/trefTest.c)
     ADD_EXECUTABLE(utilTest ${SOURCE_LIST})
-    TARGET_LINK_LIBRARIES(utilTest tutil common os gtest pthread gcov)
+    TARGET_LINK_LIBRARIES(utilTest util common os gtest pthread gcov)
+
+    LIST(REMOVE_ITEM SOURCE_LIST ${CMAKE_CURRENT_SOURCE_DIR}/cacheTest.cpp)
+    LIST(APPEND SOURCE_LIST ${CMAKE_CURRENT_SOURCE_DIR}/hashTest.cpp)
+    ADD_EXECUTABLE(hashTest ${SOURCE_LIST})
+    TARGET_LINK_LIBRARIES(hashTest util common os gtest pthread gcov)
     
     LIST(APPEND BIN_SRC ${CMAKE_CURRENT_SOURCE_DIR}/trefTest.c)
     ADD_EXECUTABLE(trefTest ${BIN_SRC})
-    TARGET_LINK_LIBRARIES(trefTest common tutil)
+    TARGET_LINK_LIBRARIES(trefTest common util)
 
 ENDIF()
 
 #IF (TD_LINUX)
 #    ADD_EXECUTABLE(trefTest ./trefTest.c)
-#    TARGET_LINK_LIBRARIES(trefTest tutil common)
+#    TARGET_LINK_LIBRARIES(trefTest util common)
 #ENDIF ()
 
 INCLUDE_DIRECTORIES(${TD_COMMUNITY_DIR}/src/util/inc)

source/util/test/hashTest.cpp

@@ -4,10 +4,15 @@
 #include <taosdef.h>
 #include <iostream>
 
-#include "hash.h"
+#include "thash.h"
 #include "taos.h"
 
 namespace {
+
+typedef struct TESTSTRUCT {
+    char *p;
+}TESTSTRUCT;
+
 // the simple test code for basic operations
 void simpleTest() {
   SHashObj* hashTable = (SHashObj*) taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false, HASH_ENTRY_LOCK);
@@ -141,6 +146,52 @@ void invalidOperationTest() {
 
 }
 
+void acquireRleaseTest() {
+  SHashObj* hashTable = (SHashObj*) taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_ENTRY_LOCK);
+  ASSERT_EQ(taosHashGetSize(hashTable), 0);
+
+  int32_t key = 2;
+  int32_t code = 0;
+  int32_t num = 0;
+  TESTSTRUCT data = {0};
+  char *str1 = "abcdefg";
+  char *str2 = "aaaaaaa";
+  char *str3 = "123456789";
+
+  data.p = (char *)malloc(10);
+  strcpy(data.p, str1);
+
+  code = taosHashPut(hashTable, &key, sizeof(key), &data, sizeof(data));
+  ASSERT_EQ(code, 0);
+
+  TESTSTRUCT* pdata = (TESTSTRUCT*)taosHashAcquire(hashTable, &key, sizeof(key));
+  ASSERT_TRUE(pdata != nullptr);
+  ASSERT_TRUE(strcmp(pdata->p, str1) == 0);
+  
+  code = taosHashRemove(hashTable, &key, sizeof(key));
+  ASSERT_EQ(code, 0);
+  ASSERT_TRUE(strcmp(pdata->p, str1) == 0);
+
+  num = taosHashGetSize(hashTable);
+  ASSERT_EQ(num, 1);
+  
+  strcpy(pdata->p, str3);
+
+  data.p = (char *)malloc(10);
+  strcpy(data.p, str2);
+  code = taosHashPut(hashTable, &key, sizeof(key), &data, sizeof(data));
+  ASSERT_EQ(code, 0);
+  num = taosHashGetSize(hashTable);
+  ASSERT_EQ(num, 2);
+
+  printf("%s,expect:%s", pdata->p, str3);
+  ASSERT_TRUE(strcmp(pdata->p, str3) == 0);
+  
+  taosHashRelease(hashTable, pdata);
+  num = taosHashGetSize(hashTable);
+  ASSERT_EQ(num, 1);
+}
+
 }
 
 int main(int argc, char** argv) {
@@ -153,4 +204,5 @@ TEST(testCase, hashTest) {
   stringKeyTest();
   noLockPerformanceTest();
   multithreadsTest();
+  acquireRleaseTest();
 }

tests/script/general/db/basic1.sim

@@ -59,9 +59,29 @@ if $data03 != 0 then
 endi
 
 print =============== show vgroups
-sql use d4
+sql show databases
 
-if $rows == 0 then 
+if $rows != 1 then 
+  return -1
+endi
+
+sql_error use d1
+
+sql use d4
+sql show vgroups
+
+if $rows != 2 then
+  return -1
+endi
+
+print =============== show dnodes
+sql show dnodes
+
+if $data00 != 1 then
+  return -1
+endi
+
+if $data02 != 2 then
   return -1
 endi
 

tests/script/general/table/basic1.sim

@@ -23,7 +23,7 @@ endi
 
 print $data00 $data01 $data02
 
-sql create table st2 (ts timestamp, i float) tags (j bigint)
+sql create table st2 (ts timestamp, i float) tags (j int)
 sql show stables
 if $rows != 2 then 
   return -1
@@ -39,15 +39,14 @@ if $rows != 1 then
   return -1
 endi
 
-print -->
 print $data00 $data01 $data02
 print $data10 $data11 $data12
 
-return
-
 print =============== create child table
 sql create table c1 using st tags(1)
 sql create table c2 using st tags(2)
+
+return 
 sql show tables
 if $rows != 2 then 
   return -1

tests/script/sh/exec.sh

@@ -17,7 +17,7 @@ OS_TYPE=`$UNAME_BIN`
 NODE_NAME=
 EXEC_OPTON=
 CLEAR_OPTION="false"
-while getopts "n:s:u:x:ct" arg
+while getopts "n:s:u:x:cv" arg
 do
   case $arg in
     n)
@@ -29,7 +29,7 @@ do
     c)
       CLEAR_OPTION="clear"
       ;;
-    t)
+    v)
       SHELL_OPTION="true"
       ;;
     u)

tests/script/tmp/dnodes.sim

@@ -2,20 +2,19 @@ system sh/stop_dnodes.sh
 
 
 ############## config parameter  #####################
-$node1 = 192.168.101.174 
+$node1 = 192.168.0.201 
 $node2 = 192.168.0.202
-$node2 = 192.168.0.203
-$node3 = 192.168.0.204
+$node3 = 192.168.0.203
+$node4 = 192.168.0.204
 
-$first = 1
-$num = 5
 $self = $node1
+$num = 25
 
 ###############  deploy firstEp  #####################
 
 $firstEp = $node1 . :7100
 $firstPort = 7100
-if $first == 1 then 
+if $self == $node1 then 
   system sh/deploy.sh -n dnode1 -i 1
   system sh/cfg.sh -n dnode1 -c firstEp -v $firstEp
   system sh/cfg.sh -n dnode1 -c secondEp -v $firstEp
@@ -28,7 +27,7 @@ if $first == 1 then
   $i = 0
   while $i < $num
     $port = $i * 100
-    $port = $port + 8000
+    $port = $port + 8100
     $i = $i + 1
     sql create dnode $node1 port $port
   endw
@@ -36,7 +35,7 @@ if $first == 1 then
   $i = 0
   while $i < $num
     $port = $i * 100
-    $port = $port + 8000
+    $port = $port + 8100
     $i = $i + 1
     sql create dnode $node2 port $port
   endw
@@ -44,7 +43,7 @@ if $first == 1 then
   $i = 0
   while $i < $num
     $port = $i * 100
-    $port = $port + 8000
+    $port = $port + 8100
     $i = $i + 1
     sql create dnode $node3 port $port
   endw
@@ -52,7 +51,7 @@ if $first == 1 then
   $i = 0
   while $i < $num
     $port = $i * 100
-    $port = $port + 8000
+    $port = $port + 8100
     $i = $i + 1
     sql create dnode $node4 port $port
   endw
@@ -64,7 +63,7 @@ $i = 0
 while $i < $num
   $index = $i + 80
   $port = $i * 100
-  $port = $port + 8000
+  $port = $port + 8100
   $dnodename = dnode . $index
   $i = $i + 1 
 
@@ -74,5 +73,5 @@ while $i < $num
   system sh/cfg.sh -n $dnodename -c fqdn -v $self
   system sh/cfg.sh -n $dnodename -c serverPort -v $port
 
-  #system sh/exec.sh -n $dnodename -s start
+  system sh/exec.sh -n $dnodename -s start
 endw