Merge remote-tracking branch 'origin/3.0' into feature/3.0_liaohj

Jenkinsfile

@@ -106,6 +106,10 @@ pipeline {
               abortPreviousBuilds()
             }
           pre_test()
+          sh'''
+          cd ${WKC}/tests
+          ./test-all.sh b1fq
+          '''
           }
       }
   //     stage('Parallel test stage') {

cmake/stub_CMakeLists.txt.in

+
+# stub
+ExternalProject_Add(stub
+        GIT_REPOSITORY https://github.com/coolxv/cpp-stub.git
+        GIT_SUBMODULES "src"
+        SOURCE_DIR "${CMAKE_CONTRIB_DIR}/cpp-stub"
+        BINARY_DIR "${CMAKE_CONTRIB_DIR}/cpp-stub/src"
+        CONFIGURE_COMMAND ""
+        BUILD_COMMAND ""
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+)

contrib/CMakeLists.txt

@@ -12,6 +12,7 @@ configure_file("${CMAKE_SUPPORT_DIR}/deps_CMakeLists.txt.in" ${CONTRIB_TMP_FILE}
 # googletest
 if(${BUILD_TEST})
     cat("${CMAKE_SUPPORT_DIR}/gtest_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    cat("${CMAKE_SUPPORT_DIR}/stub_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
 endif(${BUILD_TEST})
 
 # lz4
@@ -79,6 +80,11 @@ execute_process(COMMAND "${CMAKE_COMMAND}" --build .
 # googletest
 if(${BUILD_TEST})
     add_subdirectory(googletest)
+    target_include_directories(
+        gtest
+        PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/cpp-stub/src>
+        PUBLIC $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/cpp-stub/src_linux>
+    )
 endif(${BUILD_TEST})
 
 # cJson

include/common/taosmsg.h

@@ -50,9 +50,8 @@ TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_CONSUME, "mq-consume" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_QUERY, "mq-query" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_CONNECT, "mq-connect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_DISCONNECT, "mq-disconnect" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_SET, "mq-set" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_SET_CUR, "mq-set-cur" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_RSP_READY, "rsp-ready" )
-
 // message from client to mnode
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CONNECT, "connect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_ACCT, "create-acct" )	
@@ -95,22 +94,22 @@ TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_NETWORK_TEST, "nettest" )
 // message from vnode to dnode
 
 // message from mnode to vnode
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_STB_IN, "create-stb-in" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_ALTER_STB_IN, "alter-stb-in" )	
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_STB_IN, "drop-stb-in" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_STB_IN, "create-stb-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_ALTER_STB_IN, "alter-stb-internal" )	
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_STB_IN, "drop-stb-internal" )
 // message from mnode to mnode
 // message from mnode to qnode
 // message from mnode to dnode
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_VNODE_IN, "create-vnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_ALTER_VNODE_IN, "alter-vnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_VNODE_IN, "drop-vnode" )	
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_AUTH_VNODE_IN, "auth-vnode" )	
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_SYNC_VNODE_IN, "sync-vnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_COMPACT_VNODE_IN, "compact-vnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_MNODE_IN, "create-mnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_ALTER_MNODE_IN, "alter-mnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_MNODE_IN, "drop-mnode" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CONFIG_DNODE_IN, "config-dnode" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_VNODE_IN, "create-vnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_ALTER_VNODE_IN, "alter-vnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_VNODE_IN, "drop-vnode-internal" )	
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_AUTH_VNODE_IN, "auth-vnode-internal" )	
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_SYNC_VNODE_IN, "sync-vnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_COMPACT_VNODE_IN, "compact-vnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_MNODE_IN, "create-mnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_ALTER_MNODE_IN, "alter-mnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_DROP_MNODE_IN, "drop-mnode-internal" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CONFIG_DNODE_IN, "config-dnode-internal" )
 
 // message from qnode to vnode
 // message from qnode to mnode
@@ -225,6 +224,7 @@ typedef struct SBuildUseDBInput {
   int32_t vgVersion;
 } SBuildUseDBInput;
 
+
 #pragma pack(push, 1)
 
 // null-terminated string instead of char array to avoid too many memory consumption in case of more than 1M tableMeta
@@ -290,37 +290,6 @@ typedef struct SSchema {
   char    name[TSDB_COL_NAME_LEN];
 } SSchema;
 
-typedef struct {
-  int32_t  contLen;
-  int32_t  vgId;
-  int8_t   tableType;
-  int16_t  numOfColumns;
-  int16_t  numOfTags;
-  int32_t  tid;
-  int32_t  sversion;
-  int32_t  tversion;
-  int32_t  tagDataLen;
-  int32_t  sqlDataLen;
-  uint64_t uid;
-  uint64_t superTableUid;
-  uint64_t createdTime;
-  char     tableFname[TSDB_TABLE_FNAME_LEN];
-  char     stbFname[TSDB_TABLE_FNAME_LEN];
-  char     data[];
-} SMDCreateTableMsg;
-
-typedef struct {
-  int32_t len;  // one create table message
-  char    tableName[TSDB_TABLE_FNAME_LEN];
-  int8_t  igExists;
-  int8_t  getMeta;
-  int16_t numOfTags;
-  int16_t numOfColumns;
-  int16_t sqlLen;  // the length of SQL, it starts after schema , sql is a null-terminated string
-  int8_t  reserved[16];
-  char    schema[];
-} SCreateTableMsg;
-
 typedef struct {
   char    name[TSDB_TABLE_FNAME_LEN];
   int8_t  igExists;
@@ -341,16 +310,50 @@ typedef struct {
 } SAlterStbMsg;
 
 typedef struct {
-  char    tableFname[TSDB_TABLE_FNAME_LEN];
-  char    db[TSDB_FULL_DB_NAME_LEN];
-  int16_t type;      /* operation type   */
-  int16_t numOfCols; /* number of schema */
-  int32_t tagValLen;
-  SSchema schema[];
-  // tagVal is padded after schema
-  // char    tagVal[];
+  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;
+
+typedef struct {
+  SMsgHead head;
+  char     name[TSDB_TABLE_FNAME_LEN];
+  char     stbFname[TSDB_TABLE_FNAME_LEN];
+  int8_t   tableType;
+  uint64_t suid;
+  int32_t  sversion;
+  int32_t  numOfTags;
+  int32_t  numOfColumns;
+  int32_t  tagDataLen;
+  char     data[];
+} SCreateTableMsg;
+
+typedef struct {
+  SMsgHead head;
+  char     name[TSDB_TABLE_FNAME_LEN];
+  int8_t   type;      /* operation type   */
+  int32_t  numOfCols; /* number of schema */
+  int32_t  numOfTags;
+  char     data[];
 } SAlterTableMsg;
 
+typedef struct {
+  SMsgHead head;
+  char     name[TSDB_TABLE_FNAME_LEN];
+} SDropTableMsg;
+
 typedef struct {
   SMsgHead head;
   int64_t  uid;
@@ -545,8 +548,8 @@ typedef struct {
   int32_t     sqlstrLen;      // sql query string
   int32_t     prevResultLen;  // previous result length
   int32_t     numOfOperator;
-  int32_t     tableScanOperator;  // table scan operator. -1 means no scan operator
-  int32_t     udfNum;             // number of udf function
+  int32_t     tableScanOperator;// table scan operator. -1 means no scan operator
+  int32_t     udfNum;           // number of udf function
   int32_t     udfContentOffset;
   int32_t     udfContentLen;
   SColumnInfo tableCols[];
@@ -591,8 +594,8 @@ typedef struct {
   int32_t daysToKeep0;
   int32_t daysToKeep1;
   int32_t daysToKeep2;
-  int32_t minRowsPerFileBlock;
-  int32_t maxRowsPerFileBlock;
+  int32_t minRows;
+  int32_t maxRows;
   int32_t commitTime;
   int32_t fsyncPeriod;
   int8_t  walLevel;
@@ -706,7 +709,7 @@ typedef struct {
   SVnodeLoad data[];
 } SVnodeLoads;
 
-typedef struct SStatusMsg {
+typedef struct {
   int32_t     sver;
   int32_t     dnodeId;
   int32_t     clusterId;
@@ -756,6 +759,7 @@ typedef struct {
   int32_t  dnodeId;
   char     db[TSDB_FULL_DB_NAME_LEN];
   uint64_t dbUid;
+  int32_t  vgVersion;
   int32_t  cacheBlockSize;
   int32_t  totalBlocks;
   int32_t  daysPerFile;
@@ -1005,27 +1009,35 @@ typedef struct {
 
 // mq related
 typedef struct {
+
 } SMqConnectReq;
 
 typedef struct {
+
 } SMqConnectRsp;
 
 typedef struct {
+
 } SMqDisconnectReq;
 
 typedef struct {
+
 } SMqDisconnectRsp;
 
 typedef struct {
+
 } SMqAckReq;
 
 typedef struct {
+
 } SMqAckRsp;
 
 typedef struct {
+
 } SMqResetReq;
 
 typedef struct {
+
 } SMqResetRsp;
 // mq related end
 

include/dnode/vnode/tq/tq.h

@@ -16,86 +16,76 @@
 #ifndef _TD_TQ_H_
 #define _TD_TQ_H_
 
+#include "common.h"
 #include "mallocator.h"
 #include "os.h"
+#include "taoserror.h"
+#include "taosmsg.h"
+#include "tlist.h"
 #include "tutil.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef struct TmqMsgHead {
+typedef struct STqMsgHead {
   int32_t protoVer;
   int32_t msgType;
   int64_t cgId;
   int64_t clientId;
-} TmqMsgHead;
+} STqMsgHead;
 
-typedef struct TmqOneAck {
+typedef struct STqOneAck {
   int64_t topicId;
   int64_t consumeOffset;
-} TmqOneAck;
+} STqOneAck;
 
-typedef struct TmqAcks {
+typedef struct STqAcks {
   int32_t ackNum;
   // should be sorted
-  TmqOneAck acks[];
-} TmqAcks;
+  STqOneAck acks[];
+} STqAcks;
 
-// TODO: put msgs into common
-typedef struct TmqConnectReq {
-  TmqMsgHead head;
-  TmqAcks    acks;
-} TmqConnectReq;
-
-typedef struct TmqConnectRsp {
-  TmqMsgHead head;
-  int8_t     status;
-} TmqConnectRsp;
-
-typedef struct TmqDisconnectReq {
-  TmqMsgHead head;
-} TmqDiscconectReq;
-
-typedef struct TmqDisconnectRsp {
-  TmqMsgHead head;
-  int8_t     status;
-} TmqDisconnectRsp;
+typedef struct STqSetCurReq {
+  STqMsgHead head;
+  int64_t    topicId;
+  int64_t    offset;
+} STqSetCurReq;
 
 typedef struct STqConsumeReq {
-  TmqMsgHead head;
-  TmqAcks    acks;
+  STqMsgHead head;
+  STqAcks    acks;
 } STqConsumeReq;
 
-typedef struct TmqMsgContent {
+typedef struct STqMsgContent {
   int64_t topicId;
   int64_t msgLen;
   char    msg[];
-} TmqMsgContent;
+} STqMsgContent;
 
 typedef struct STqConsumeRsp {
-  TmqMsgHead    head;
+  STqMsgHead    head;
   int64_t       bodySize;
-  TmqMsgContent msgs[];
+  STqMsgContent msgs[];
 } STqConsumeRsp;
 
-typedef struct TmqSubscribeReq {
-  TmqMsgHead head;
+typedef struct STqSubscribeReq {
+  STqMsgHead head;
   int32_t    topicNum;
   int64_t    topic[];
-} TmqSubscribeReq;
+} STqSubscribeReq;
 
-typedef struct tmqSubscribeRsp {
-  TmqMsgHead head;
+typedef struct STqSubscribeRsp {
+  STqMsgHead head;
   int64_t    vgId;
   char       ep[TSDB_EP_LEN];  // TSDB_EP_LEN
-} TmqSubscribeRsp;
+} STqSubscribeRsp;
 
-typedef struct TmqHeartbeatReq {
-} TmqHeartbeatReq;
+typedef struct STqHeartbeatReq {
+} STqHeartbeatReq;
 
-typedef struct TmqHeartbeatRsp {
-} TmqHeartbeatRsp;
+typedef struct STqHeartbeatRsp {
+} STqHeartbeatRsp;
 
 typedef struct STqTopicVhandle {
   int64_t topicId;
@@ -108,48 +98,54 @@ typedef struct STqTopicVhandle {
 
 #define TQ_BUFFER_SIZE 8
 
+typedef struct STqExec {
+  void* runtimeEnv;
+  SSDataBlock* (*exec)(void* runtimeEnv);
+  void* (*assign)(void* runtimeEnv, SSubmitBlk* inputData);
+  void (*clear)(void* runtimeEnv);
+  char* (*serialize)(struct STqExec*);
+  struct STqExec* (*deserialize)(char*);
+} STqExec;
+
 typedef struct STqBufferItem {
   int64_t offset;
   // executors are identical but not concurrent
   // so there must be a copy in each item
-  void*   executor;
-  int64_t size;
-  void*   content;
-} STqBufferItem;
+  STqExec* executor;
+  int32_t  status;
+  int64_t  size;
+  void*    content;
+} STqMsgItem;
 
-typedef struct STqBufferHandle {
+typedef struct STqTopic {
   // char* topic; //c style, end with '\0'
   // int64_t cgId;
   // void* ahandle;
-  int64_t       nextConsumeOffset;
-  int64_t       floatingCursor;
-  int64_t       topicId;
-  int32_t       head;
-  int32_t       tail;
-  STqBufferItem buffer[TQ_BUFFER_SIZE];
-} STqBufferHandle;
+  int64_t    nextConsumeOffset;
+  int64_t    floatingCursor;
+  int64_t    topicId;
+  int32_t    head;
+  int32_t    tail;
+  STqMsgItem buffer[TQ_BUFFER_SIZE];
+} STqTopic;
 
 typedef struct STqListHandle {
-  STqBufferHandle       bufHandle;
+  STqTopic              topic;
   struct STqListHandle* next;
-} STqListHandle;
-
-typedef struct STqGroupHandle {
-  int64_t        cId;
-  int64_t        cgId;
-  void*          ahandle;
-  int32_t        topicNum;
-  STqListHandle* head;
-} STqGroupHandle;
-
-typedef struct STqQueryExec {
-  void*          src;
-  STqBufferItem* dest;
-  void*          executor;
-} STqQueryExec;
+} STqList;
+
+typedef struct STqGroup {
+  int64_t  clientId;
+  int64_t  cgId;
+  void*    ahandle;
+  int32_t  topicNum;
+  STqList* head;
+  SList*   topicList;  // SList<STqTopic>
+  void*    returnMsg;  // SVReadMsg
+} STqGroup;
 
 typedef struct STqQueryMsg {
-  STqQueryExec*       exec;
+  STqMsgItem*         item;
   struct STqQueryMsg* next;
 } STqQueryMsg;
 
@@ -209,15 +205,15 @@ typedef void (*FTqDelete)(void*);
 #define TQ_DUP_INTXN_REWRITE 0
 #define TQ_DUP_INTXN_REJECT 2
 
-static inline bool TqUpdateAppend(int32_t tqConfigFlag) { return tqConfigFlag & TQ_UPDATE_APPEND; }
+static inline bool tqUpdateAppend(int32_t tqConfigFlag) { return tqConfigFlag & TQ_UPDATE_APPEND; }
 
-static inline bool TqDupIntxnReject(int32_t tqConfigFlag) { return tqConfigFlag & TQ_DUP_INTXN_REJECT; }
+static inline bool tqDupIntxnReject(int32_t tqConfigFlag) { return tqConfigFlag & TQ_DUP_INTXN_REJECT; }
 
 static const int8_t TQ_CONST_DELETE = TQ_ACTION_CONST;
 
 #define TQ_DELETE_TOKEN (void*)&TQ_CONST_DELETE
 
-typedef struct TqMetaHandle {
+typedef struct STqMetaHandle {
   int64_t key;
   int64_t offset;
   int64_t serializedSize;
@@ -225,23 +221,25 @@ typedef struct TqMetaHandle {
   void*   valueInTxn;
 } STqMetaHandle;
 
-typedef struct TqMetaList {
-  STqMetaHandle      handle;
-  struct TqMetaList* next;
-  // struct TqMetaList* inTxnPrev;
-  // struct TqMetaList* inTxnNext;
-  struct TqMetaList* unpersistPrev;
-  struct TqMetaList* unpersistNext;
+typedef struct STqMetaList {
+  STqMetaHandle       handle;
+  struct STqMetaList* next;
+  // struct STqMetaList* inTxnPrev;
+  // struct STqMetaList* inTxnNext;
+  struct STqMetaList* unpersistPrev;
+  struct STqMetaList* unpersistNext;
 } STqMetaList;
 
-typedef struct TqMetaStore {
+typedef struct STqMetaStore {
   STqMetaList* bucket[TQ_BUCKET_SIZE];
   // a table head
   STqMetaList* unpersistHead;
+
   // TODO:temporaral use, to be replaced by unified tfile
   int fileFd;
   // TODO:temporaral use, to be replaced by unified tfile
-  int            idxFd;
+  int idxFd;
+
   char*          dirPath;
   int32_t        tqConfigFlag;
   FTqSerialize   pSerializer;
@@ -250,8 +248,8 @@ typedef struct TqMetaStore {
 } STqMetaStore;
 
 typedef struct STQ {
-  // the collection of group handle
-  // the handle of kvstore
+  // the collection of groups
+  // the handle of meta kvstore
   char*         path;
   STqCfg*       tqConfig;
   STqLogReader* tqLogReader;
@@ -266,23 +264,25 @@ void tqClose(STQ*);
 // void* will be replace by a msg type
 int tqPushMsg(STQ*, void* msg, int64_t version);
 int tqCommit(STQ*);
-int tqSetCursor(STQ*, void* msg);
-
 int tqConsume(STQ*, STqConsumeReq*);
 
-STqGroupHandle* tqGetGroupHandle(STQ*, int64_t cId);
+int tqSetCursor(STQ*, STqSetCurReq* pMsg);
+int tqBufferSetOffset(STqTopic*, int64_t offset);
+
+STqTopic* tqFindTopic(STqGroup*, int64_t topicId);
+
+STqGroup* tqGetGroup(STQ*, int64_t clientId);
+
+STqGroup* tqOpenGroup(STQ*, int64_t topicId, int64_t cgId, int64_t cId);
+int       tqCloseGroup(STQ*, int64_t topicId, int64_t cgId, int64_t cId);
+int       tqRegisterContext(STqGroup*, void* ahandle);
+int       tqSendLaunchQuery(STqMsgItem*, int64_t offset);
 
-STqGroupHandle* tqOpenTCGroup(STQ*, int64_t topicId, int64_t cgId, int64_t cId);
-int             tqCloseTCGroup(STQ*, int64_t topicId, int64_t cgId, int64_t cId);
-int             tqMoveOffsetToNext(STqGroupHandle*);
-int             tqResetOffset(STQ*, int64_t topicId, int64_t cgId, int64_t offset);
-int             tqRegisterContext(STqGroupHandle*, void* ahandle);
-int             tqLaunchQuery(STqGroupHandle*);
-int             tqSendLaunchQuery(STqGroupHandle*);
+int tqSerializeGroup(const STqGroup*, STqSerializedHead**);
 
-int tqSerializeGroupHandle(const STqGroupHandle* gHandle, STqSerializedHead** ppHead);
+const void* tqDeserializeGroup(const STqSerializedHead*, STqGroup**);
 
-const void* tqDeserializeGroupHandle(const STqSerializedHead* pHead, STqGroupHandle** gHandle);
+static int tqQueryExecuting(int32_t status) { return status; }
 
 #ifdef __cplusplus
 }

include/dnode/vnode/vnode.h

@@ -132,6 +132,36 @@ int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
  */
 int vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
 
+/**
+ * @brief Process a query 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 vnodeProcessQueryReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp);
+
+/**
+ * @brief Process a fetch 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 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.
@@ -170,73 +200,52 @@ typedef struct {
   char info[];
 } SVnodeRsp;
 
-#define VNODE_INIT_CREATE_STB_REQ(NAME, TTL, KEEP, SUID, PSCHEMA, PTAGSCHEMA) \
-  { .ver = 0, .ctReq = META_INIT_STB_CFG(NAME, TTL, KEEP, SUID, PSCHEMA, PTAGSCHEMA) }
+static FORCE_INLINE void vnodeSetCreateStbReq(SVnodeReq *pReq, char *name, uint32_t ttl, uint32_t keep, tb_uid_t suid,
+                                              STSchema *pSchema, STSchema *pTagSchema) {
+  pReq->ver = 0;
+
+  pReq->ctReq.name = name;
+  pReq->ctReq.ttl = ttl;
+  pReq->ctReq.keep = keep;
+  pReq->ctReq.type = META_SUPER_TABLE;
+  pReq->ctReq.stbCfg.suid = suid;
+  pReq->ctReq.stbCfg.pSchema = pSchema;
+  pReq->ctReq.stbCfg.pTagSchema = pTagSchema;
+}
+
+static FORCE_INLINE void vnodeSetCreateCtbReq(SVnodeReq *pReq, char *name, uint32_t ttl, uint32_t keep, tb_uid_t suid,
+                                              SKVRow pTag) {
+  pReq->ver = 0;
 
-#define VNODE_INIT_CREATE_CTB_REQ(NAME, TTL, KEEP, SUID, PTAG) \
-  { .ver = 0, .ctReq = META_INIT_CTB_CFG(NAME, TTL, KEEP, SUID, PTAG) }
+  pReq->ctReq.name = name;
+  pReq->ctReq.ttl = ttl;
+  pReq->ctReq.keep = keep;
+  pReq->ctReq.type = META_CHILD_TABLE;
+  pReq->ctReq.ctbCfg.suid = suid;
+  pReq->ctReq.ctbCfg.pTag = pTag;
+}
 
-#define VNODE_INIT_CREATE_NTB_REQ(NAME, TTL, KEEP, SUID, PSCHEMA) \
-  { .ver = 0, .ctReq = META_INIT_NTB_CFG(NAME, TTL, KEEP, SUID, PSCHEMA) }
+static FORCE_INLINE void vnodeSetCreateNtbReq(SVnodeReq *pReq, char *name, uint32_t ttl, uint32_t keep,
+                                              STSchema *pSchema) {
+  pReq->ver = 0;
+
+  pReq->ctReq.name = name;
+  pReq->ctReq.ttl = ttl;
+  pReq->ctReq.keep = keep;
+  pReq->ctReq.type = META_NORMAL_TABLE;
+  pReq->ctReq.ntbCfg.pSchema = pSchema;
+}
 
 int   vnodeBuildReq(void **buf, const SVnodeReq *pReq, uint8_t type);
 void *vnodeParseReq(void *buf, SVnodeReq *pReq, uint8_t type);
 
 /* ------------------------ FOR COMPILE ------------------------ */
 
-#if 1
-
-#include "taosmsg.h"
-#include "trpc.h"
-
-// typedef struct {
-//   char     db[TSDB_FULL_DB_NAME_LEN];
-//   int32_t  cacheBlockSize;  // MB
-//   int32_t  totalBlocks;
-//   int32_t  daysPerFile;
-//   int32_t  daysToKeep0;
-//   int32_t  daysToKeep1;
-//   int32_t  daysToKeep2;
-//   int32_t  minRowsPerFileBlock;
-//   int32_t  maxRowsPerFileBlock;
-//   int8_t   precision;  // time resolution
-//   int8_t   compression;
-//   int8_t   cacheLastRow;
-//   int8_t   update;
-//   int8_t   quorum;
-//   int8_t   replica;
-//   int8_t   selfIndex;
-//   int8_t   walLevel;
-//   int32_t  fsyncPeriod;  // millisecond
-//   SReplica replicas[TSDB_MAX_REPLICA];
-// } SVnodeCfg;
-
-typedef enum {
-  VN_MSG_TYPE_WRITE = 1,
-  VN_MSG_TYPE_APPLY,
-  VN_MSG_TYPE_SYNC,
-  VN_MSG_TYPE_QUERY,
-  VN_MSG_TYPE_FETCH
-} EVnMsgType;
-
-typedef struct {
-  int32_t curNum;
-  int32_t allocNum;
-  SRpcMsg rpcMsg[];
-} SVnodeMsg;
-
 int32_t vnodeAlter(SVnode *pVnode, const SVnodeCfg *pCfg);
 int32_t vnodeCompact(SVnode *pVnode);
 int32_t vnodeSync(SVnode *pVnode);
 int32_t vnodeGetLoad(SVnode *pVnode, SVnodeLoad *pLoad);
 
-SVnodeMsg *vnodeInitMsg(int32_t msgNum);
-int32_t    vnodeAppendMsg(SVnodeMsg *pMsg, SRpcMsg *pRpcMsg);
-void       vnodeCleanupMsg(SVnodeMsg *pMsg);
-void       vnodeProcessMsg(SVnode *pVnode, SVnodeMsg *pMsg, EVnMsgType msgType);
-
-#endif
-
 #ifdef __cplusplus
 }
 #endif

include/libs/index/index.h

@@ -23,64 +23,63 @@
 extern "C" {
 #endif
 
-typedef struct SIndex SIndex;
-typedef struct SIndexTerm SIndexTerm;
-typedef struct SIndexOpts SIndexOpts;
+typedef struct SIndex               SIndex;
+typedef struct SIndexTerm           SIndexTerm;
+typedef struct SIndexOpts           SIndexOpts;
 typedef struct SIndexMultiTermQuery SIndexMultiTermQuery;
 typedef struct SArray               SIndexMultiTerm;
 
-typedef enum  {
-   ADD_VALUE,    // add index colume value   
-   DEL_VALUE,    // delete index column value  
-   UPDATE_VALUE, // update index column value 
-   ADD_INDEX,    // add index on specify column
-   DROP_INDEX,   // drop existed index 
-   DROP_SATBLE   // drop stable 
+typedef enum {
+  ADD_VALUE,     // add index colume value
+  DEL_VALUE,     // delete index column value
+  UPDATE_VALUE,  // update index column value
+  ADD_INDEX,     // add index on specify column
+  DROP_INDEX,    // drop existed index
+  DROP_SATBLE    // drop stable
 } SIndexOperOnColumn;
 
-typedef enum  { MUST = 0, SHOULD = 1, NOT = 2 } EIndexOperatorType;
-typedef enum  { QUERY_TERM = 0, QUERY_PREFIX = 1, QUERY_SUFFIX = 2,QUERY_REGEX = 3} EIndexQueryType;
+typedef enum { MUST = 0, SHOULD = 1, NOT = 2 } EIndexOperatorType;
+typedef enum { QUERY_TERM = 0, QUERY_PREFIX = 1, QUERY_SUFFIX = 2, QUERY_REGEX = 3 } EIndexQueryType;
 /*
- * @param: oper 
+ * @param: oper
  *
-*/
+ */
 SIndexMultiTermQuery *indexMultiTermQueryCreate(EIndexOperatorType oper);
-void            indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery);
-int             indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType type);
-/* 
- * @param:    
+void                  indexMultiTermQueryDestroy(SIndexMultiTermQuery *pQuery);
+int                   indexMultiTermQueryAdd(SIndexMultiTermQuery *pQuery, SIndexTerm *term, EIndexQueryType type);
+/*
+ * @param:
  * @param:
  */
-int   indexOpen(SIndexOpts *opt, const char *path, SIndex **index);
-void  indexClose(SIndex *index);
-int   indexPut(SIndex *index,    SIndexMultiTerm *terms, uint64_t uid);
-int   indexDelete(SIndex *index, SIndexMultiTermQuery *query); 
-int   indexSearch(SIndex *index, SIndexMultiTermQuery *query, SArray *result);
-int   indexRebuild(SIndex *index, SIndexOpts *opt);
+int  indexOpen(SIndexOpts *opt, const char *path, SIndex **index);
+void indexClose(SIndex *index);
+int  indexPut(SIndex *index, SIndexMultiTerm *terms, uint64_t uid);
+int  indexDelete(SIndex *index, SIndexMultiTermQuery *query);
+int  indexSearch(SIndex *index, SIndexMultiTermQuery *query, SArray *result);
+int  indexRebuild(SIndex *index, SIndexOpts *opt);
 /*
  * @param
  * @param
  */
-SIndexMultiTerm *indexMultiTermCreate(); 
-int indexMultiTermAdd(SIndexMultiTerm  *terms, SIndexTerm *term);
-void indexMultiTermDestroy(SIndexMultiTerm *terms);
+SIndexMultiTerm *indexMultiTermCreate();
+int              indexMultiTermAdd(SIndexMultiTerm *terms, SIndexTerm *term);
+void             indexMultiTermDestroy(SIndexMultiTerm *terms);
 /*
- * @param: 
+ * @param:
  * @param:
  */
 SIndexOpts *indexOptsCreate();
-void       indexOptsDestroy(SIndexOpts *opts);
+void        indexOptsDestroy(SIndexOpts *opts);
 
 /*
  * @param:
  * @param:
  */
 
-SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn operType, uint8_t colType, 
-                            const char *colName, int32_t nColName, const char *colVal, int32_t nColVal);
+SIndexTerm *indexTermCreate(int64_t suid, SIndexOperOnColumn operType, uint8_t colType, const char *colName,
+    int32_t nColName, const char *colVal, int32_t nColVal);
 void        indexTermDestroy(SIndexTerm *p);
 
-
 #ifdef __cplusplus
 }
 #endif

include/util/taoserror.h

@@ -230,6 +230,10 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_INVALID_FUNC_CODE         TAOS_DEF_ERROR_CODE(0, 0x03C5)
 #define TSDB_CODE_MND_INVALID_FUNC_BUFSIZE      TAOS_DEF_ERROR_CODE(0, 0x03C6)
 
+// mnode-trans
+#define TSDB_CODE_MND_TRANS_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x03D0)
+#define TSDB_CODE_MND_TRANS_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x03D1)
+
 // dnode
 #define TSDB_CODE_DND_ACTION_IN_PROGRESS        TAOS_DEF_ERROR_CODE(0, 0x0400)
 #define TSDB_CODE_DND_EXITING                   TAOS_DEF_ERROR_CODE(0, 0x0401)
@@ -339,6 +343,20 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_SYN_INVALID_MSGLEN            TAOS_DEF_ERROR_CODE(0, 0x0909)  //"Invalid msg length")
 #define TSDB_CODE_SYN_INVALID_MSGTYPE           TAOS_DEF_ERROR_CODE(0, 0x090A)  //"Invalid msg type")
 
+// tq
+#define TSDB_CODE_TQ_INVALID_CONFIG             TAOS_DEF_ERROR_CODE(0, 0x0A00)  //"Invalid configuration")
+#define TSDB_CODE_TQ_INIT_FAILED                TAOS_DEF_ERROR_CODE(0, 0x0A01)  //"Tq init failed")
+#define TSDB_CODE_TQ_NO_DISKSPACE               TAOS_DEF_ERROR_CODE(0, 0x0A02)  //"No diskspace for tq")
+#define TSDB_CODE_TQ_NO_DISK_PERMISSIONS        TAOS_DEF_ERROR_CODE(0, 0x0A03)  //"No permission for disk files")
+#define TSDB_CODE_TQ_FILE_CORRUPTED             TAOS_DEF_ERROR_CODE(0, 0x0A04)  //"Data file(s) corrupted")
+#define TSDB_CODE_TQ_OUT_OF_MEMORY              TAOS_DEF_ERROR_CODE(0, 0x0A05)  //"Out of memory")
+#define TSDB_CODE_TQ_FILE_ALREADY_EXISTS        TAOS_DEF_ERROR_CODE(0, 0x0A06)  //"File already exists")
+#define TSDB_CODE_TQ_FAILED_TO_CREATE_DIR       TAOS_DEF_ERROR_CODE(0, 0x0A07)  //"Failed to create dir")
+#define TSDB_CODE_TQ_META_NO_SUCH_KEY           TAOS_DEF_ERROR_CODE(0, 0x0A08)  //"Target key not found")
+#define TSDB_CODE_TQ_META_KEY_NOT_IN_TXN        TAOS_DEF_ERROR_CODE(0, 0x0A09)  //"Target key not in transaction")
+#define TSDB_CODE_TQ_META_KEY_DUP_IN_TXN        TAOS_DEF_ERROR_CODE(0, 0x0A0A)  //"Target key duplicated in transaction")
+#define TSDB_CODE_TQ_GROUP_NOT_SET              TAOS_DEF_ERROR_CODE(0, 0x0A0B)  //"Group of corresponding client is not set by mnode")
+
 // wal
 #define TSDB_CODE_WAL_APP_ERROR                 TAOS_DEF_ERROR_CODE(0, 0x1000)  //"Unexpected generic error in wal")
 #define TSDB_CODE_WAL_FILE_CORRUPTED            TAOS_DEF_ERROR_CODE(0, 0x1001)  //"WAL file is corrupted")

include/util/tlog.h

@@ -42,11 +42,11 @@ extern int32_t qDebugFlag;
 extern int32_t wDebugFlag;
 extern int32_t sDebugFlag;
 extern int32_t tsdbDebugFlag;
+extern int32_t tqDebugFlag;
 extern int32_t cqDebugFlag;
 extern int32_t debugFlag;
 extern int32_t ctgDebugFlag;
 
-
 #define DEBUG_FATAL 1U
 #define DEBUG_ERROR DEBUG_FATAL
 #define DEBUG_WARN 2U

include/util/tthread.h

@@ -24,8 +24,8 @@ extern "C" {
 #include "tdef.h"
 
 // create new thread
-pthread_t* taosCreateThread( void *(*__start_routine) (void *), void* param);
-// destory thread 
+pthread_t* taosCreateThread(void* (*__start_routine)(void*), void* param);
+// destory thread
 bool taosDestoryThread(pthread_t* pthread);
 // thread running return true
 bool taosThreadRunning(pthread_t* pthread);

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

@@ -31,7 +31,7 @@ static struct {
 } global = {0};
 
 void dmnSigintHandle(int signum, void *info, void *ctx) {
-  uError("singal:%d is received", signum);
+  uInfo("singal:%d is received", signum);
   global.stop = true;
 }
 

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

@@ -44,7 +44,7 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TSDB_MSG_TYPE_MQ_CONSUME] = dndProcessVnodeQueryMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_MQ_CONNECT] = dndProcessVnodeWriteMsg;
   pMgmt->msgFp[TSDB_MSG_TYPE_MQ_DISCONNECT] = dndProcessVnodeWriteMsg;
-  pMgmt->msgFp[TSDB_MSG_TYPE_MQ_SET] = dndProcessVnodeWriteMsg;
+  pMgmt->msgFp[TSDB_MSG_TYPE_MQ_SET_CUR] = dndProcessVnodeWriteMsg;
 
   // msg from client to mnode
   pMgmt->msgFp[TSDB_MSG_TYPE_CONNECT] = dndProcessMnodeReadMsg;

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

@@ -176,6 +176,12 @@ SDnode *dndInit(SDnodeOpt *pOption) {
     return NULL;
   }
 
+  if (vnodeInit(1) != 0) {
+    dError("failed to init vnode env");
+    dndCleanup(pDnode);
+    return NULL;
+  }
+
   if (dndInitDnode(pDnode) != 0) {
     dError("failed to init dnode");
     dndCleanup(pDnode);
@@ -222,8 +228,10 @@ void dndCleanup(SDnode *pDnode) {
   dndCleanupMnode(pDnode);
   dndCleanupVnodes(pDnode);
   dndCleanupDnode(pDnode);
+  vnodeClear();
   walCleanUp();
   rpcCleanup();
+
   dndCleanupEnv(pDnode);
   free(pDnode);
   dInfo("TDengine is cleaned up successfully");

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

@@ -210,8 +210,8 @@ TEST_F(DndTestDb, 02_Create_Alter_Drop_Db) {
     pReq->daysToKeep0 = htonl(3650);
     pReq->daysToKeep1 = htonl(3650);
     pReq->daysToKeep2 = htonl(3650);
-    pReq->minRowsPerFileBlock = htonl(100);
-    pReq->maxRowsPerFileBlock = htonl(4096);
+    pReq->minRows = htonl(100);
+    pReq->maxRows = htonl(4096);
     pReq->commitTime = htonl(3600);
     pReq->fsyncPeriod = htonl(3000);
     pReq->walLevel = 1;
@@ -375,8 +375,8 @@ TEST_F(DndTestDb, 03_Create_Use_Restart_Use_Db) {
     pReq->daysToKeep0 = htonl(3650);
     pReq->daysToKeep1 = htonl(3650);
     pReq->daysToKeep2 = htonl(3650);
-    pReq->minRowsPerFileBlock = htonl(100);
-    pReq->maxRowsPerFileBlock = htonl(4096);
+    pReq->minRows = htonl(100);
+    pReq->maxRows = htonl(4096);
     pReq->commitTime = htonl(3600);
     pReq->fsyncPeriod = htonl(3000);
     pReq->walLevel = 1;

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

@@ -187,8 +187,8 @@ TEST_F(DndTestStb, 01_Create_Show_Meta_Drop_Restart_Stb) {
     pReq->daysToKeep0 = htonl(3650);
     pReq->daysToKeep1 = htonl(3650);
     pReq->daysToKeep2 = htonl(3650);
-    pReq->minRowsPerFileBlock = htonl(100);
-    pReq->maxRowsPerFileBlock = htonl(4096);
+    pReq->minRows = htonl(100);
+    pReq->maxRows = htonl(4096);
     pReq->commitTime = htonl(3600);
     pReq->fsyncPeriod = htonl(3000);
     pReq->walLevel = 1;

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

@@ -16,7 +16,7 @@
 #include "deploy.h"
 
 void initLog(const char* path) {
-  dDebugFlag = 143;
+  dDebugFlag = 207;
   vDebugFlag = 0;
   mDebugFlag = 207;
   cDebugFlag = 0;

source/dnode/mgmt/impl/test/sut/deploy.h

@@ -46,3 +46,4 @@ void     stopServer(SServer* pServer);
 SClient* createClient(const char* user, const char* pass, const char* fqdn, uint16_t port);
 void     dropClient(SClient* pClient);
 void     sendMsg(SClient* pClient, SRpcMsg* pMsg);
+

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

@@ -176,49 +176,112 @@ SServer* DndTestVgroup::pServer;
 SClient* DndTestVgroup::pClient;
 int32_t  DndTestVgroup::connId;
 
-
 TEST_F(DndTestVgroup, 01_Create_Restart_Drop_Vnode) {
   {
-    SCreateVnodeMsg* pReq = (SCreateVnodeMsg*)rpcMallocCont(sizeof(SCreateVnodeMsg));
-    pReq->vgId = htonl(2);
-    pReq->dnodeId = htonl(1);
-    strcpy(pReq->db, "1.d1");
-    pReq->dbUid = htobe64(9527);
-    pReq->cacheBlockSize = htonl(16);
-    pReq->totalBlocks = htonl(10);
-    pReq->daysPerFile = htonl(10);
-    pReq->daysToKeep0 = htonl(3650);
-    pReq->daysToKeep1 = htonl(3650);
-    pReq->daysToKeep2 = htonl(3650);
-    pReq->minRows = htonl(100);
-    pReq->minRows = htonl(4096);
-    pReq->commitTime = htonl(3600);
-    pReq->fsyncPeriod = htonl(3000);
-    pReq->walLevel = 1;
-    pReq->precision = 0;
-    pReq->compression = 2;
-    pReq->replica = 1;
-    pReq->quorum = 1;
-    pReq->update = 0;
-    pReq->cacheLastRow = 0;
-    pReq->selfIndex = 0;
-    for (int r = 0; r < pReq->replica; ++r) {
-      SReplica* pReplica = &pReq->replicas[r];
-      pReplica->id = htonl(1);
-      pReplica->port = htons(9150);
+    for (int i = 0; i < 3; ++i) {
+      SCreateVnodeMsg* pReq = (SCreateVnodeMsg*)rpcMallocCont(sizeof(SCreateVnodeMsg));
+      pReq->vgId = htonl(2);
+      pReq->dnodeId = htonl(1);
+      strcpy(pReq->db, "1.d1");
+      pReq->dbUid = htobe64(9527);
+      pReq->vgVersion = htonl(1);
+      pReq->cacheBlockSize = htonl(16);
+      pReq->totalBlocks = htonl(10);
+      pReq->daysPerFile = htonl(10);
+      pReq->daysToKeep0 = htonl(3650);
+      pReq->daysToKeep1 = htonl(3650);
+      pReq->daysToKeep2 = htonl(3650);
+      pReq->minRows = htonl(100);
+      pReq->minRows = htonl(4096);
+      pReq->commitTime = htonl(3600);
+      pReq->fsyncPeriod = htonl(3000);
+      pReq->walLevel = 1;
+      pReq->precision = 0;
+      pReq->compression = 2;
+      pReq->replica = 1;
+      pReq->quorum = 1;
+      pReq->update = 0;
+      pReq->cacheLastRow = 0;
+      pReq->selfIndex = 0;
+      for (int r = 0; r < pReq->replica; ++r) {
+        SReplica* pReplica = &pReq->replicas[r];
+        pReplica->id = htonl(1);
+        pReplica->port = htons(9150);
+      }
+
+      SRpcMsg rpcMsg = {0};
+      rpcMsg.pCont = pReq;
+      rpcMsg.contLen = sizeof(SCreateVnodeMsg);
+      rpcMsg.msgType = TSDB_MSG_TYPE_CREATE_VNODE_IN;
+
+      sendMsg(pClient, &rpcMsg);
+      SRpcMsg* pMsg = pClient->pRsp;
+      ASSERT_NE(pMsg, nullptr);
+      ASSERT_EQ(pMsg->code, 0);
     }
+  }
 
-    SRpcMsg rpcMsg = {0};
-    rpcMsg.pCont = pReq;
-    rpcMsg.contLen = sizeof(SCreateVnodeMsg);
-    rpcMsg.msgType = TSDB_MSG_TYPE_CREATE_VNODE_IN;
-
-    sendMsg(pClient, &rpcMsg);
-    SRpcMsg* pMsg = pClient->pRsp;
-    ASSERT_NE(pMsg, nullptr);
-    ASSERT_EQ(pMsg->code, 0);
-    taosMsleep(1000000);
+  {
+    for (int i = 0; i < 3; ++i) {
+      SAlterVnodeMsg* pReq = (SAlterVnodeMsg*)rpcMallocCont(sizeof(SAlterVnodeMsg));
+      pReq->vgId = htonl(2);
+      pReq->dnodeId = htonl(1);
+      strcpy(pReq->db, "1.d1");
+      pReq->dbUid = htobe64(9527);
+      pReq->vgVersion = htonl(2);
+      pReq->cacheBlockSize = htonl(16);
+      pReq->totalBlocks = htonl(10);
+      pReq->daysPerFile = htonl(10);
+      pReq->daysToKeep0 = htonl(3650);
+      pReq->daysToKeep1 = htonl(3650);
+      pReq->daysToKeep2 = htonl(3650);
+      pReq->minRows = htonl(100);
+      pReq->minRows = htonl(4096);
+      pReq->commitTime = htonl(3600);
+      pReq->fsyncPeriod = htonl(3000);
+      pReq->walLevel = 1;
+      pReq->precision = 0;
+      pReq->compression = 2;
+      pReq->replica = 1;
+      pReq->quorum = 1;
+      pReq->update = 0;
+      pReq->cacheLastRow = 0;
+      pReq->selfIndex = 0;
+      for (int r = 0; r < pReq->replica; ++r) {
+        SReplica* pReplica = &pReq->replicas[r];
+        pReplica->id = htonl(1);
+        pReplica->port = htons(9150);
+      }
+
+      SRpcMsg rpcMsg = {0};
+      rpcMsg.pCont = pReq;
+      rpcMsg.contLen = sizeof(SAlterVnodeMsg);
+      rpcMsg.msgType = TSDB_MSG_TYPE_ALTER_VNODE_IN;
+
+      sendMsg(pClient, &rpcMsg);
+      SRpcMsg* pMsg = pClient->pRsp;
+      ASSERT_NE(pMsg, nullptr);
+      ASSERT_EQ(pMsg->code, 0);
+    }
   }
 
+  {
+    for (int i = 0; i < 3; ++i) {
+      SDropVnodeMsg* pReq = (SDropVnodeMsg*)rpcMallocCont(sizeof(SDropVnodeMsg));
+      pReq->vgId = htonl(2);
+      pReq->dnodeId = htonl(1);
+      strcpy(pReq->db, "1.d1");
+      pReq->dbUid = htobe64(9527);
+
+      SRpcMsg rpcMsg = {0};
+      rpcMsg.pCont = pReq;
+      rpcMsg.contLen = sizeof(SDropVnodeMsg);
+      rpcMsg.msgType = TSDB_MSG_TYPE_DROP_VNODE_IN;
+
+      sendMsg(pClient, &rpcMsg);
+      SRpcMsg* pMsg = pClient->pRsp;
+      ASSERT_NE(pMsg, nullptr);
+      ASSERT_EQ(pMsg->code, 0);
+    }
+  }
 }
-

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

@@ -61,15 +61,14 @@ typedef enum {
 } EAuthOp;
 
 typedef enum {
-  TRN_STAGE_PREPARE = 1,
-  TRN_STAGE_EXECUTE = 2,
+  TRN_STAGE_PREPARE = 0,
+  TRN_STAGE_EXECUTE = 1,
+  TRN_STAGE_ROLLBACK = 2,
   TRN_STAGE_COMMIT = 3,
-  TRN_STAGE_ROLLBACK = 4,
-  TRN_STAGE_RETRY = 5,
-  TRN_STAGE_OVER = 6,
+  TRN_STAGE_OVER = 4,
 } ETrnStage;
 
-typedef enum { TRN_POLICY_ROLLBACK = 1, TRN_POLICY_RETRY = 2 } ETrnPolicy;
+typedef enum { TRN_POLICY_ROLLBACK = 0, TRN_POLICY_RETRY = 1 } ETrnPolicy;
 
 typedef enum {
   DND_STATUS_OFFLINE = 0,

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

@@ -70,6 +70,7 @@ typedef struct SMnode {
   tmr_h             timer;
   char             *path;
   SMnodeCfg         cfg;
+  int64_t           checkTime;
   SSdb             *pSdb;
   SDnode           *pDnode;
   SArray           *pSteps;

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

@@ -22,6 +22,16 @@
 extern "C" {
 #endif
 
+typedef struct {
+  SEpSet  epSet;
+  int8_t  msgType;
+  int8_t  msgSent;
+  int8_t  msgReceived;
+  int32_t errCode;
+  int32_t contLen;
+  void   *pCont;
+} STransAction;
+
 int32_t mndInitTrans(SMnode *pMnode);
 void    mndCleanupTrans(SMnode *pMnode);
 
@@ -30,12 +40,15 @@ void    mndTransDrop(STrans *pTrans);
 int32_t mndTransAppendRedolog(STrans *pTrans, SSdbRaw *pRaw);
 int32_t mndTransAppendUndolog(STrans *pTrans, SSdbRaw *pRaw);
 int32_t mndTransAppendCommitlog(STrans *pTrans, SSdbRaw *pRaw);
-int32_t mndTransAppendRedoAction(STrans *pTrans, SEpSet *pEpSet, int8_t msgType, int32_t contLen, void *pCont);
-int32_t mndTransAppendUndoAction(STrans *pTrans, SEpSet *pEpSet, int8_t msgType, int32_t contLen, void *pCont);
+int32_t mndTransAppendRedoAction(STrans *pTrans, STransAction *pAction);
+int32_t mndTransAppendUndoAction(STrans *pTrans, STransAction *pAction);
+
 int32_t mndTransPrepare(SMnode *pMnode, STrans *pTrans);
 void    mndTransApply(SMnode *pMnode, SSdbRaw *pRaw, STransMsg *pMsg, int32_t code);
-char   *mndTransStageStr(ETrnStage stage);
-char   *mndTransPolicyStr(ETrnPolicy policy);
+void    mndTransHandleActionRsp(SMnodeMsg *pMsg);
+
+char *mndTransStageStr(ETrnStage stage);
+char *mndTransPolicyStr(ETrnPolicy policy);
 
 #ifdef __cplusplus
 }

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

@@ -26,8 +26,9 @@ int32_t  mndInitVgroup(SMnode *pMnode);
 void     mndCleanupVgroup(SMnode *pMnode);
 SVgObj  *mndAcquireVgroup(SMnode *pMnode, int32_t vgId);
 void     mndReleaseVgroup(SMnode *pMnode, SVgObj *pVgroup);
-int32_t  mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups);
 SSdbRaw *mndVgroupActionEncode(SVgObj *pVgroup);
+int32_t  mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups);
+SEpSet   mndGetVgroupEpset(SMnode *pMnode, SVgObj *pVgroup);
 
 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/mndAcct.c

@@ -17,7 +17,8 @@
 #include "mndAcct.h"
 #include "mndShow.h"
 
-#define SDB_ACCT_VER 1
+#define TSDB_ACCT_VER_NUMBER 1
+#define TSDB_ACCT_RESERVE_SIZE 64
 
 static int32_t  mndCreateDefaultAcct(SMnode *pMnode);
 static SSdbRaw *mndAcctActionEncode(SAcctObj *pAcct);
@@ -70,7 +71,7 @@ static int32_t mndCreateDefaultAcct(SMnode *pMnode) {
 }
 
 static SSdbRaw *mndAcctActionEncode(SAcctObj *pAcct) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_ACCT, SDB_ACCT_VER, sizeof(SAcctObj));
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_ACCT, TSDB_ACCT_VER_NUMBER, sizeof(SAcctObj) + TSDB_ACCT_RESERVE_SIZE);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -85,6 +86,7 @@ static SSdbRaw *mndAcctActionEncode(SAcctObj *pAcct) {
   SDB_SET_INT32(pRaw, dataPos, pAcct->cfg.maxStreams)
   SDB_SET_INT64(pRaw, dataPos, pAcct->cfg.maxStorage)
   SDB_SET_INT32(pRaw, dataPos, pAcct->cfg.accessState)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_ACCT_RESERVE_SIZE)
   SDB_SET_DATALEN(pRaw, dataPos);
 
   return pRaw;
@@ -94,7 +96,7 @@ static SSdbRow *mndAcctActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != SDB_ACCT_VER) {
+  if (sver != TSDB_ACCT_VER_NUMBER) {
     mError("failed to decode acct since %s", terrstr());
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     return NULL;
@@ -116,13 +118,13 @@ static SSdbRow *mndAcctActionDecode(SSdbRaw *pRaw) {
   SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->cfg.maxStreams)
   SDB_GET_INT64(pRaw, pRow, dataPos, &pAcct->cfg.maxStorage)
   SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->cfg.accessState)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_ACCT_RESERVE_SIZE)
 
   return pRow;
 }
 
 static int32_t mndAcctActionInsert(SSdb *pSdb, SAcctObj *pAcct) {
   mTrace("acct:%s, perform insert action", pAcct->acct);
-  memset(&pAcct->info, 0, sizeof(SAcctInfo));
   return 0;
 }
 
@@ -134,12 +136,9 @@ static int32_t mndAcctActionDelete(SSdb *pSdb, SAcctObj *pAcct) {
 static int32_t mndAcctActionUpdate(SSdb *pSdb, SAcctObj *pOldAcct, SAcctObj *pNewAcct) {
   mTrace("acct:%s, perform update action", pOldAcct->acct);
 
-  memcpy(pOldAcct->acct, pNewAcct->acct, TSDB_USER_LEN);
-  pOldAcct->createdTime = pNewAcct->createdTime;
   pOldAcct->updateTime = pNewAcct->updateTime;
-  pOldAcct->acctId = pNewAcct->acctId;
   pOldAcct->status = pNewAcct->status;
-  pOldAcct->cfg = pNewAcct->cfg;
+  memcpy(&pOldAcct->cfg, &pNewAcct->cfg, sizeof(SAcctInfo));
   return 0;
 }
 

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

@@ -18,7 +18,8 @@
 #include "mndShow.h"
 #include "mndTrans.h"
 
-#define SDB_CLUSTER_VER 1
+#define TSDB_CLUSTER_VER_NUMBE 1
+#define TSDB_CLUSTER_RESERVE_SIZE 64
 
 static SSdbRaw *mndClusterActionEncode(SClusterObj *pCluster);
 static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw);
@@ -62,7 +63,7 @@ int32_t mndGetClusterName(SMnode *pMnode, char *clusterName, int32_t len) {
 }
 
 static SSdbRaw *mndClusterActionEncode(SClusterObj *pCluster) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_CLUSTER, SDB_CLUSTER_VER, sizeof(SClusterObj));
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_CLUSTER, TSDB_CLUSTER_VER_NUMBE, sizeof(SClusterObj) + TSDB_CLUSTER_RESERVE_SIZE);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -70,6 +71,7 @@ static SSdbRaw *mndClusterActionEncode(SClusterObj *pCluster) {
   SDB_SET_INT64(pRaw, dataPos, pCluster->createdTime)
   SDB_SET_INT64(pRaw, dataPos, pCluster->updateTime)
   SDB_SET_BINARY(pRaw, dataPos, pCluster->name, TSDB_CLUSTER_ID_LEN)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_CLUSTER_RESERVE_SIZE)
 
   return pRaw;
 }
@@ -78,7 +80,7 @@ static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != SDB_CLUSTER_VER) {
+  if (sver != TSDB_CLUSTER_VER_NUMBE) {
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     mError("failed to decode cluster since %s", terrstr());
     return NULL;
@@ -93,6 +95,7 @@ static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw) {
   SDB_GET_INT64(pRaw, pRow, dataPos, &pCluster->createdTime)
   SDB_GET_INT64(pRaw, pRow, dataPos, &pCluster->updateTime)
   SDB_GET_BINARY(pRaw, pRow, dataPos, pCluster->name, TSDB_CLUSTER_ID_LEN)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_CLUSTER_RESERVE_SIZE)
 
   return pRow;
 }
@@ -169,6 +172,7 @@ static int32_t mndGetClusterMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg
 
   pShow->numOfRows = 1;
   pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
 
   return 0;
 }

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

@@ -18,10 +18,10 @@
 #include "mndMnode.h"
 #include "mndShow.h"
 #include "mndTrans.h"
-#include "ttime.h"
 #include "tep.h"
+#include "ttime.h"
 
-#define TSDB_DNODE_VER 1
+#define TSDB_DNODE_VER_NUMBER 1
 #define TSDB_DNODE_RESERVE_SIZE 64
 #define TSDB_CONFIG_OPTION_LEN 16
 #define TSDB_CONIIG_VALUE_LEN 48
@@ -101,14 +101,14 @@ static int32_t mndCreateDefaultDnode(SMnode *pMnode) {
 
   SSdbRaw *pRaw = mndDnodeActionEncode(&dnodeObj);
   if (pRaw == NULL) return -1;
-  sdbSetRawStatus(pRaw, SDB_STATUS_READY);
+  if (sdbSetRawStatus(pRaw, SDB_STATUS_READY) != 0) return -1;
 
   mDebug("dnode:%d, will be created while deploy sdb", dnodeObj.id);
   return sdbWrite(pMnode->pSdb, pRaw);
 }
 
 static SSdbRaw *mndDnodeActionEncode(SDnodeObj *pDnode) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_DNODE, TSDB_DNODE_VER, sizeof(SDnodeObj) + TSDB_DNODE_RESERVE_SIZE);
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_DNODE, TSDB_DNODE_VER_NUMBER, sizeof(SDnodeObj) + TSDB_DNODE_RESERVE_SIZE);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -127,7 +127,7 @@ static SSdbRow *mndDnodeActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != TSDB_DNODE_VER) {
+  if (sver != TSDB_DNODE_VER_NUMBER) {
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     mError("failed to decode dnode since %s", terrstr());
     return NULL;
@@ -150,21 +150,8 @@ static SSdbRow *mndDnodeActionDecode(SSdbRaw *pRaw) {
 
 static int32_t mndDnodeActionInsert(SSdb *pSdb, SDnodeObj *pDnode) {
   mTrace("dnode:%d, perform insert action", pDnode->id);
-
-  pDnode->rebootTime = 0;
-  pDnode->lastAccessTime = 0;
-  pDnode->accessTimes = 0;
-  pDnode->numOfMnodes = 0;
-  pDnode->numOfVnodes = 0;
-  pDnode->numOfQnodes = 0;
-  pDnode->numOfSupportMnodes = 0;
-  pDnode->numOfSupportVnodes = 0;
-  pDnode->numOfSupportQnodes = 0;
-  pDnode->numOfCores = 0;
-  pDnode->status = DND_STATUS_OFFLINE;
   pDnode->offlineReason = DND_REASON_STATUS_NOT_RECEIVED;
   snprintf(pDnode->ep, TSDB_EP_LEN, "%s:%u", pDnode->fqdn, pDnode->port);
-
   return 0;
 }
 
@@ -225,7 +212,7 @@ int32_t mndGetDnodeSize(SMnode *pMnode) {
 bool mndIsDnodeInReadyStatus(SMnode *pMnode, SDnodeObj *pDnode) {
   int64_t ms = taosGetTimestampMs();
   int64_t interval = ABS(pDnode->lastAccessTime - ms);
-  if (interval > 3000 * pMnode->cfg.statusInterval) {
+  if (interval > 3500 * pMnode->cfg.statusInterval) {
     return false;
   }
   return true;
@@ -267,12 +254,9 @@ static int32_t mndCheckClusterCfgPara(SMnode *pMnode, const SClusterCfg *pCfg) {
     return DND_REASON_STATUS_INTERVAL_NOT_MATCH;
   }
 
-  int64_t checkTime = 0;
-  char    timestr[32] = "1970-01-01 00:00:00.00";
-  (void)taosParseTime(timestr, &checkTime, (int32_t)strlen(timestr), TSDB_TIME_PRECISION_MILLI, 0);
-  if ((0 != strcasecmp(pCfg->timezone, pMnode->cfg.timezone)) && (checkTime != pCfg->checkTime)) {
+  if ((0 != strcasecmp(pCfg->timezone, pMnode->cfg.timezone)) && (pMnode->checkTime != pCfg->checkTime)) {
     mError("timezone [%s - %s] [%" PRId64 " - %" PRId64 "] cfg inconsistent", pCfg->timezone, pMnode->cfg.timezone,
-           pCfg->checkTime, checkTime);
+           pCfg->checkTime, pMnode->checkTime);
     return DND_REASON_TIME_ZONE_NOT_MATCH;
   }
 

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

@@ -19,7 +19,8 @@
 #include "mndShow.h"
 #include "mndTrans.h"
 
-#define SDB_MNODE_VER 1
+#define TSDB_MNODE_VER_NUMBER 1
+#define TSDB_MNODE_RESERVE_SIZE 64
 
 static int32_t  mndCreateDefaultMnode(SMnode *pMnode);
 static SSdbRaw *mndMnodeActionEncode(SMnodeObj *pMnodeObj);
@@ -60,8 +61,12 @@ int32_t mndInitMnode(SMnode *pMnode) {
 void mndCleanupMnode(SMnode *pMnode) {}
 
 static SMnodeObj *mndAcquireMnode(SMnode *pMnode, int32_t mnodeId) {
-  SSdb *pSdb = pMnode->pSdb;
-  return sdbAcquire(pSdb, SDB_MNODE, &mnodeId);
+  SSdb      *pSdb = pMnode->pSdb;
+  SMnodeObj *pObj = sdbAcquire(pSdb, SDB_MNODE, &mnodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_MNODE_NOT_EXIST;
+  }
+  return pObj;
 }
 
 static void mndReleaseMnode(SMnode *pMnode, SMnodeObj *pMnodeObj) {
@@ -97,13 +102,14 @@ static int32_t mndCreateDefaultMnode(SMnode *pMnode) {
 }
 
 static SSdbRaw *mndMnodeActionEncode(SMnodeObj *pMnodeObj) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_MNODE, SDB_MNODE_VER, sizeof(SMnodeObj));
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_MNODE, TSDB_MNODE_VER_NUMBER, sizeof(SMnodeObj) + TSDB_MNODE_RESERVE_SIZE);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
   SDB_SET_INT32(pRaw, dataPos, pMnodeObj->id);
   SDB_SET_INT64(pRaw, dataPos, pMnodeObj->createdTime)
   SDB_SET_INT64(pRaw, dataPos, pMnodeObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_MNODE_RESERVE_SIZE)
 
   return pRaw;
 }
@@ -112,7 +118,7 @@ static SSdbRow *mndMnodeActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != SDB_MNODE_VER) {
+  if (sver != TSDB_MNODE_VER_NUMBER) {
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     mError("failed to decode mnode since %s", terrstr());
     return NULL;
@@ -126,6 +132,7 @@ static SSdbRow *mndMnodeActionDecode(SSdbRaw *pRaw) {
   SDB_GET_INT32(pRaw, pRow, dataPos, &pMnodeObj->id)
   SDB_GET_INT64(pRaw, pRow, dataPos, &pMnodeObj->createdTime)
   SDB_GET_INT64(pRaw, pRow, dataPos, &pMnodeObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_MNODE_RESERVE_SIZE)
 
   return pRow;
 }

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

@@ -21,9 +21,10 @@
 #include "mndShow.h"
 #include "mndTrans.h"
 #include "mndUser.h"
+#include "mndVgroup.h"
 #include "tname.h"
 
-#define TSDB_STB_VER_NUM 1
+#define TSDB_STB_VER_NUMBER 1
 #define TSDB_STB_RESERVE_SIZE 64
 
 static SSdbRaw *mndStbActionEncode(SStbObj *pStb);
@@ -70,7 +71,7 @@ void mndCleanupStb(SMnode *pMnode) {}
 
 static SSdbRaw *mndStbActionEncode(SStbObj *pStb) {
   int32_t  size = sizeof(SStbObj) + (pStb->numOfColumns + pStb->numOfTags) * sizeof(SSchema) + TSDB_STB_RESERVE_SIZE;
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_STB, TSDB_STB_VER_NUM, size);
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_STB, TSDB_STB_VER_NUMBER, size);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -103,7 +104,7 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != TSDB_STB_VER_NUM) {
+  if (sver != TSDB_STB_VER_NUMBER) {
     mError("failed to decode stable since %s", terrstr());
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     return NULL;
@@ -176,8 +177,12 @@ static int32_t mndStbActionUpdate(SSdb *pSdb, SStbObj *pOldStb, SStbObj *pNewStb
 }
 
 SStbObj *mndAcquireStb(SMnode *pMnode, char *stbName) {
-  SSdb *pSdb = pMnode->pSdb;
-  return sdbAcquire(pSdb, SDB_STB, stbName);
+  SSdb    *pSdb = pMnode->pSdb;
+  SStbObj *pStb = sdbAcquire(pSdb, SDB_STB, stbName);
+  if (pStb == NULL) {
+    terrno = TSDB_CODE_MND_STB_NOT_EXIST;
+  }
+  return pStb;
 }
 
 void mndReleaseStb(SMnode *pMnode, SStbObj *pStb) {
@@ -195,7 +200,54 @@ static SDbObj *mndAcquireDbByStb(SMnode *pMnode, char *stbName) {
   return mndAcquireDb(pMnode, db);
 }
 
-static int32_t mndCheckStbMsg(SCreateStbMsg *pCreate) {
+static SCreateStbInternalMsg *mndBuildCreateStbMsg(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb) {
+  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);
+  }
+
+  return pCreate;
+}
+
+static SDropStbInternalMsg *mndBuildDropStbMsg(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pStb) {
+  int32_t contLen = sizeof(SDropStbInternalMsg);
+
+  SDropStbInternalMsg *pDrop = calloc(1, contLen);
+  if (pDrop == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  pDrop->head.contLen = htonl(contLen);
+  pDrop->head.vgId = htonl(pVgroup->vgId);
+  memcpy(pDrop->name, pStb->name, TSDB_TABLE_FNAME_LEN);
+  pDrop->suid = htobe64(pStb->uid);
+
+  return pDrop;
+}
+
+static int32_t mndCheckCreateStbMsg(SCreateStbMsg *pCreate) {
   pCreate->numOfColumns = htonl(pCreate->numOfColumns);
   pCreate->numOfTags = htonl(pCreate->numOfTags);
   int32_t totalCols = pCreate->numOfColumns + pCreate->numOfTags;
@@ -244,6 +296,103 @@ static int32_t mndCheckStbMsg(SCreateStbMsg *pCreate) {
   return 0;
 }
 
+static int32_t mndSetCreateStbRedoLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb) {
+  SSdbRaw *pRedoRaw = mndStbActionEncode(pStb);
+  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 mndSetCreateStbUndoLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb) {
+  SSdbRaw *pUndoRaw = mndStbActionEncode(pStb);
+  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 mndSetCreateStbCommitLogs(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb) {
+  SSdbRaw *pCommitRaw = mndStbActionEncode(pStb);
+  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 mndSetCreateStbRedoActions(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb) {
+  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;
+
+    SCreateStbInternalMsg *pMsg = mndBuildCreateStbMsg(pMnode, pVgroup, pStb);
+    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 = TSDB_MSG_TYPE_CREATE_STB_IN;
+    if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+      free(pMsg);
+      sdbCancelFetch(pSdb, pIter);
+      sdbRelease(pSdb, pVgroup);
+      return -1;
+    }
+    sdbRelease(pSdb, pVgroup);
+  }
+
+  return 0;
+}
+
+static int32_t mndSetCreateStbUndoActions(SMnode *pMnode, STrans *pTrans, SDbObj *pDb, SStbObj *pStb) {
+  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;
+
+    SDropStbInternalMsg *pMsg = mndBuildDropStbMsg(pMnode, pVgroup, pStb);
+    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(SDropStbInternalMsg);
+    action.msgType = TSDB_MSG_TYPE_DROP_STB_IN;
+    if (mndTransAppendUndoAction(pTrans, &action) != 0) {
+      free(pMsg);
+      sdbCancelFetch(pSdb, pIter);
+      sdbRelease(pSdb, pVgroup);
+      return -1;
+    }
+    sdbRelease(pSdb, pVgroup);
+  }
+
+  return 0;
+}
+
 static int32_t mndCreateStb(SMnode *pMnode, SMnodeMsg *pMsg, SCreateStbMsg *pCreate, SDbObj *pDb) {
   SStbObj stbObj = {0};
   tstrncpy(stbObj.name, pCreate->name, TSDB_TABLE_FNAME_LEN);
@@ -265,6 +414,7 @@ static int32_t mndCreateStb(SMnode *pMnode, SMnodeMsg *pMsg, SCreateStbMsg *pCre
   }
   memcpy(stbObj.pSchema, pCreate->pSchema, totalSize);
 
+  int32_t code = 0;
   STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, pMsg->rpcMsg.handle);
   if (pTrans == NULL) {
     mError("stb:%s, failed to create since %s", pCreate->name, terrstr());
@@ -272,29 +422,30 @@ static int32_t mndCreateStb(SMnode *pMnode, SMnodeMsg *pMsg, SCreateStbMsg *pCre
   }
   mDebug("trans:%d, used to create stb:%s", pTrans->id, pCreate->name);
 
-  SSdbRaw *pRedoRaw = mndStbActionEncode(&stbObj);
-  if (pRedoRaw == NULL || mndTransAppendRedolog(pTrans, pRedoRaw) != 0) {
-    mError("trans:%d, failed to append redo log since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+  if (mndSetCreateStbRedoLogs(pMnode, pTrans, pDb, &stbObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_STB_OVER;
   }
-  sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING);
 
-  SSdbRaw *pUndoRaw = mndStbActionEncode(&stbObj);
-  if (pUndoRaw == NULL || mndTransAppendUndolog(pTrans, pUndoRaw) != 0) {
-    mError("trans:%d, failed to append undo log since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+  if (mndSetCreateStbUndoLogs(pMnode, pTrans, pDb, &stbObj) != 0) {
+    mError("trans:%d, failed to set undo log since %s", pTrans->id, terrstr());
+    goto CREATE_STB_OVER;
   }
-  sdbSetRawStatus(pUndoRaw, SDB_STATUS_DROPPED);
 
-  SSdbRaw *pCommitRaw = mndStbActionEncode(&stbObj);
-  if (pCommitRaw == NULL || mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) {
-    mError("trans:%d, failed to append commit log since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+  if (mndSetCreateStbCommitLogs(pMnode, pTrans, pDb, &stbObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_STB_OVER;
+  }
+
+  if (mndSetCreateStbRedoActions(pMnode, pTrans, pDb, &stbObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_STB_OVER;
+  }
+
+  if (mndSetCreateStbUndoActions(pMnode, pTrans, pDb, &stbObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_STB_OVER;
   }
-  sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY);
 
   if (mndTransPrepare(pMnode, pTrans) != 0) {
     mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
@@ -302,8 +453,11 @@ static int32_t mndCreateStb(SMnode *pMnode, SMnodeMsg *pMsg, SCreateStbMsg *pCre
     return -1;
   }
 
+  code = 0;
+
+CREATE_STB_OVER:
   mndTransDrop(pTrans);
-  return 0;
+  return code;
 }
 
 static int32_t mndProcessCreateStbMsg(SMnodeMsg *pMsg) {
@@ -312,7 +466,7 @@ static int32_t mndProcessCreateStbMsg(SMnodeMsg *pMsg) {
 
   mDebug("stb:%s, start to create", pCreate->name);
 
-  if (mndCheckStbMsg(pCreate) != 0) {
+  if (mndCheckCreateStbMsg(pCreate) != 0) {
     mError("stb:%s, failed to create since %s", pCreate->name, terrstr());
     return -1;
   }
@@ -349,7 +503,10 @@ static int32_t mndProcessCreateStbMsg(SMnodeMsg *pMsg) {
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
-static int32_t mndProcessCreateStbInRsp(SMnodeMsg *pMsg) { return 0; }
+static int32_t mndProcessCreateStbInRsp(SMnodeMsg *pMsg) {
+  mndTransHandleActionRsp(pMsg);
+  return 0;
+}
 
 static int32_t mndCheckAlterStbMsg(SAlterStbMsg *pAlter) {
   SSchema *pSchema = &pAlter->schema;
@@ -410,9 +567,44 @@ static int32_t mndProcessAlterStbMsg(SMnodeMsg *pMsg) {
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
-static int32_t mndProcessAlterStbInRsp(SMnodeMsg *pMsg) { return 0; }
+static int32_t mndProcessAlterStbInRsp(SMnodeMsg *pMsg) {
+  mndTransHandleActionRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndSetDropStbRedoLogs(SMnode *pMnode, STrans *pTrans, SStbObj *pStb) {
+  SSdbRaw *pRedoRaw = mndStbActionEncode(pStb);
+  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 mndSetDropStbUndoLogs(SMnode *pMnode, STrans *pTrans, SStbObj *pStb) {
+  SSdbRaw *pUndoRaw = mndStbActionEncode(pStb);
+  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 mndSetDropStbCommitLogs(SMnode *pMnode, STrans *pTrans, SStbObj *pStb) {
+  SSdbRaw *pCommitRaw = mndStbActionEncode(pStb);
+  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 mndSetDropStbRedoActions(SMnode *pMnode, STrans *pTrans, SStbObj *pStb) { return 0; }
+
+static int32_t mndSetDropStbUndoActions(SMnode *pMnode, STrans *pTrans, SStbObj *pStb) { return 0; }
 
 static int32_t mndDropStb(SMnode *pMnode, SMnodeMsg *pMsg, SStbObj *pStb) {
+  int32_t code = -1;
   STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, pMsg->rpcMsg.handle);
   if (pTrans == NULL) {
     mError("stb:%s, failed to drop since %s", pStb->name, terrstr());
@@ -420,36 +612,39 @@ static int32_t mndDropStb(SMnode *pMnode, SMnodeMsg *pMsg, SStbObj *pStb) {
   }
   mDebug("trans:%d, used to drop stb:%s", pTrans->id, pStb->name);
 
-  SSdbRaw *pRedoRaw = mndStbActionEncode(pStb);
-  if (pRedoRaw == NULL || mndTransAppendRedolog(pTrans, pRedoRaw) != 0) {
-    mError("trans:%d, failed to append redo log since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+  if (mndSetDropStbRedoLogs(pMnode, pTrans, pStb) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_STB_OVER;
   }
-  sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING);
 
-  SSdbRaw *pUndoRaw = mndStbActionEncode(pStb);
-  if (pUndoRaw == NULL || mndTransAppendUndolog(pTrans, pUndoRaw) != 0) {
-    mError("trans:%d, failed to append undo log since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+  if (mndSetDropStbUndoLogs(pMnode, pTrans, pStb) != 0) {
+    mError("trans:%d, failed to set undo log since %s", pTrans->id, terrstr());
+    goto DROP_STB_OVER;
   }
-  sdbSetRawStatus(pUndoRaw, SDB_STATUS_READY);
 
-  SSdbRaw *pCommitRaw = mndStbActionEncode(pStb);
-  if (pCommitRaw == NULL || mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) {
-    mError("trans:%d, failed to append commit log since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+  if (mndSetDropStbCommitLogs(pMnode, pTrans, pStb) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_STB_OVER;
+  }
+
+  if (mndSetDropStbRedoActions(pMnode, pTrans, pStb) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_STB_OVER;
+  }
+
+  if (mndSetDropStbUndoActions(pMnode, pTrans, pStb) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_STB_OVER;
   }
-  sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED);
 
   if (mndTransPrepare(pMnode, pTrans) != 0) {
     mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
-    mndTransDrop(pTrans);
-    return -1;
+    goto DROP_STB_OVER;
   }
 
+  code = 0;
+
+DROP_STB_OVER:
   mndTransDrop(pTrans);
   return 0;
 }
@@ -484,7 +679,10 @@ static int32_t mndProcessDropStbMsg(SMnodeMsg *pMsg) {
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 
-static int32_t mndProcessDropStbInRsp(SMnodeMsg *pMsg) { return 0; }
+static int32_t mndProcessDropStbInRsp(SMnodeMsg *pMsg) {
+  mndTransHandleActionRsp(pMsg);
+  return 0;
+}
 
 static int32_t mndProcessStbMetaMsg(SMnodeMsg *pMsg) {
   SMnode        *pMnode = pMsg->pMnode;

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

@@ -17,17 +17,10 @@
 #include "mndTrans.h"
 #include "mndSync.h"
 
-#define TSDB_TRANS_VER 1
+#define TSDB_TRANS_VER_NUMBER 1
 #define TSDB_TRN_ARRAY_SIZE 8
 #define TSDB_TRN_RESERVE_SIZE 64
 
-typedef struct {
-  SEpSet  epSet;
-  int8_t  msgType;
-  int32_t contLen;
-  void   *pCont;
-} STransAction;
-
 static SSdbRaw *mndTransActionEncode(STrans *pTrans);
 static SSdbRow *mndTransActionDecode(SSdbRaw *pRaw);
 static int32_t  mndTransActionInsert(SSdb *pSdb, STrans *pTrans);
@@ -37,11 +30,11 @@ static int32_t  mndTransActionDelete(SSdb *pSdb, STrans *pTrans);
 static void    mndTransSetRpcHandle(STrans *pTrans, void *rpcHandle);
 static void    mndTransSendRpcRsp(STrans *pTrans, int32_t code);
 static int32_t mndTransAppendLog(SArray *pArray, SSdbRaw *pRaw);
-static int32_t mndTransAppendAction(SArray *pArray, SEpSet *pEpSet, int8_t msgType, int32_t contLen, void *pCont);
+static int32_t mndTransAppendAction(SArray *pArray, STransAction *pAction);
 static void    mndTransDropLogs(SArray *pArray);
 static void    mndTransDropActions(SArray *pArray);
 static int32_t mndTransExecuteLogs(SMnode *pMnode, SArray *pArray);
-static int32_t mndTransExecuteActions(SMnode *pMnode, SArray *pArray);
+static int32_t mndTransExecuteActions(SMnode *pMnode, STrans *pTrans, SArray *pArray);
 static int32_t mndTransExecuteRedoLogs(SMnode *pMnode, STrans *pTrans);
 static int32_t mndTransExecuteUndoLogs(SMnode *pMnode, STrans *pTrans);
 static int32_t mndTransExecuteCommitLogs(SMnode *pMnode, STrans *pTrans);
@@ -100,7 +93,7 @@ static SSdbRaw *mndTransActionEncode(STrans *pTrans) {
     rawDataLen += (sizeof(STransAction) + pAction->contLen);
   }
 
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_TRANS, TSDB_TRANS_VER, rawDataLen);
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_TRANS, TSDB_TRANS_VER_NUMBER, rawDataLen);
   if (pRaw == NULL) {
     mError("trans:%d, failed to alloc raw since %s", pTrans->id, terrstr());
     return NULL;
@@ -167,7 +160,7 @@ static SSdbRow *mndTransActionDecode(SSdbRaw *pRaw) {
     return NULL;
   }
 
-  if (sver != TSDB_TRANS_VER) {
+  if (sver != TSDB_TRANS_VER_NUMBER) {
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     mError("failed to get check soft ver from raw:%p since %s", pRaw, terrstr());
     return NULL;
@@ -324,8 +317,12 @@ static int32_t mndTransActionUpdate(SSdb *pSdb, STrans *pOldTrans, STrans *pNewT
 }
 
 STrans *mndAcquireTrans(SMnode *pMnode, int32_t transId) {
-  SSdb *pSdb = pMnode->pSdb;
-  return sdbAcquire(pSdb, SDB_TRANS, &transId);
+  SSdb   *pSdb = pMnode->pSdb;
+  STrans *pTrans = sdbAcquire(pSdb, SDB_TRANS, &transId);
+  if (pTrans == NULL) {
+    terrno = TSDB_CODE_MND_TRANS_NOT_EXIST;
+  }
+  return pTrans;
 }
 
 void mndReleaseTrans(SMnode *pMnode, STrans *pTrans) {
@@ -343,10 +340,8 @@ char *mndTransStageStr(ETrnStage stage) {
       return "commit";
     case TRN_STAGE_ROLLBACK:
       return "rollback";
-    case TRN_STAGE_RETRY:
-      return "retry";
     case TRN_STAGE_OVER:
-      return "stop";
+      return "over";
     default:
       return "undefined";
   }
@@ -388,7 +383,7 @@ STrans *mndTransCreate(SMnode *pMnode, ETrnPolicy policy, void *rpcHandle) {
     return NULL;
   }
 
-  mDebug("trans:%d, data:%p is created", pTrans->id, pTrans);
+  mDebug("trans:%d, is created", pTrans->id);
   return pTrans;
 }
 
@@ -417,7 +412,7 @@ void mndTransDrop(STrans *pTrans) {
   mndTransDropActions(pTrans->redoActions);
   mndTransDropActions(pTrans->undoActions);
 
-  mDebug("trans:%d, data:%p is dropped", pTrans->id, pTrans);
+  // mDebug("trans:%d, is dropped, data:%p", pTrans->id, pTrans);
   tfree(pTrans);
 }
 
@@ -459,10 +454,8 @@ int32_t mndTransAppendCommitlog(STrans *pTrans, SSdbRaw *pRaw) {
   return code;
 }
 
-static int32_t mndTransAppendAction(SArray *pArray, SEpSet *pEpSet, int8_t msgType, int32_t contLen, void *pCont) {
-  STransAction action = {.epSet = *pEpSet, .msgType = msgType, .contLen = contLen, .pCont = pCont};
-
-  void *ptr = taosArrayPush(pArray, &action);
+static int32_t mndTransAppendAction(SArray *pArray, STransAction *pAction) {
+  void *ptr = taosArrayPush(pArray, pAction);
   if (ptr == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return -1;
@@ -471,15 +464,15 @@ static int32_t mndTransAppendAction(SArray *pArray, SEpSet *pEpSet, int8_t msgTy
   return 0;
 }
 
-int32_t mndTransAppendRedoAction(STrans *pTrans, SEpSet *pEpSet, int8_t msgType, int32_t contLen, void *pCont) {
-  int32_t code = mndTransAppendAction(pTrans->redoActions, pEpSet, msgType, contLen, pCont);
-  mTrace("trans:%d, msg:%s len:%d append to redo actions", pTrans->id, taosMsg[msgType], contLen);
+int32_t mndTransAppendRedoAction(STrans *pTrans, STransAction *pAction) {
+  int32_t code = mndTransAppendAction(pTrans->redoActions, pAction);
+  mTrace("trans:%d, msg:%s append to redo actions, code:0x%x", pTrans->id, taosMsg[pAction->msgType], code);
   return code;
 }
 
-int32_t mndTransAppendUndoAction(STrans *pTrans, SEpSet *pEpSet, int8_t msgType, int32_t contLen, void *pCont) {
-  int32_t code = mndTransAppendAction(pTrans->undoActions, pEpSet, msgType, contLen, pCont);
-  mTrace("trans:%d, msg:%s len:%d append to undo actions", pTrans->id, taosMsg[msgType], contLen);
+int32_t mndTransAppendUndoAction(STrans *pTrans, STransAction *pAction) {
+  int32_t code = mndTransAppendAction(pTrans->undoActions, pAction);
+  mTrace("trans:%d, msg:%s append to undo actions, code:0x%x", pTrans->id, taosMsg[pAction->msgType], code);
   return code;
 }
 
@@ -493,7 +486,7 @@ int32_t mndTransPrepare(SMnode *pMnode, STrans *pTrans) {
   }
   sdbSetRawStatus(pRaw, SDB_STATUS_READY);
 
-  mTrace("trans:%d, start sync", pTrans->id);
+  mTrace("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());
@@ -515,7 +508,6 @@ int32_t mndTransPrepare(SMnode *pMnode, STrans *pTrans) {
     return -1;
   }
 
-  mDebug("trans:%d, prepare finished", pNewTrans->id);
   pNewTrans->rpcHandle = pTrans->rpcHandle;
   mndTransExecute(pMnode, pNewTrans);
   mndReleaseTrans(pMnode, pNewTrans);
@@ -533,7 +525,7 @@ int32_t mndTransCommit(SMnode *pMnode, STrans *pTrans) {
   sdbSetRawStatus(pRaw, SDB_STATUS_DROPPED);
 
   if (taosArrayGetSize(pTrans->commitLogs) != 0) {
-    mTrace("trans:%d, start sync", pTrans->id);
+    mTrace("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());
@@ -563,7 +555,7 @@ int32_t mndTransRollback(SMnode *pMnode, STrans *pTrans) {
   }
   sdbSetRawStatus(pRaw, SDB_STATUS_DROPPED);
 
-  mTrace("trans:%d, start sync", pTrans->id);
+  mTrace("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());
@@ -596,6 +588,50 @@ void mndTransApply(SMnode *pMnode, SSdbRaw *pRaw, STransMsg *pMsg, int32_t code)
   // todo
 }
 
+void mndTransHandleActionRsp(SMnodeMsg *pMsg) {
+  SMnode *pMnode = pMsg->pMnode;
+  int64_t sig = (int64_t)(pMsg->rpcMsg.ahandle);
+  int32_t transId = (int32_t)(sig >> 32);
+  int32_t action = (int32_t)((sig << 32) >> 32);
+
+  STrans *pTrans = mndAcquireTrans(pMnode, transId);
+  if (pTrans == NULL) {
+    mError("trans:%d, failed to get transId from vnode rsp since %s", transId, terrstr());
+    goto HANDLE_ACTION_RSP_OVER;
+  }
+
+  SArray *pArray = NULL;
+  if (pTrans->stage == TRN_STAGE_EXECUTE) {
+    pArray = pTrans->redoActions;
+  } else if (pTrans->stage == TRN_STAGE_ROLLBACK) {
+    pArray = pTrans->undoActions;
+  } else {
+  }
+
+  if (pArray == NULL) {
+    mError("trans:%d, invalid trans stage:%s", transId, mndTransStageStr(pTrans->stage));
+    goto HANDLE_ACTION_RSP_OVER;
+  }
+
+  int32_t actionNum = taosArrayGetSize(pTrans->redoActions);
+  if (action < 0 || action > actionNum) {
+    mError("trans:%d, invalid action:%d", transId, action);
+    goto HANDLE_ACTION_RSP_OVER;
+  }
+
+  STransAction *pAction = taosArrayGet(pArray, action);
+  if (pAction != NULL) {
+    pAction->msgReceived = 1;
+    pAction->errCode = pMsg->code;
+  }
+
+  mDebug("trans:%d, action:%d response is received, code:0x%x", transId, action, pMsg->code);
+  mndTransExecute(pMnode, pTrans);
+
+HANDLE_ACTION_RSP_OVER:
+  mndReleaseTrans(pMnode, pTrans);
+}
+
 static int32_t mndTransExecuteLogs(SMnode *pMnode, SArray *pArray) {
   SSdb   *pSdb = pMnode->pSdb;
   int32_t arraySize = taosArrayGetSize(pArray);
@@ -618,7 +654,7 @@ static int32_t mndTransExecuteRedoLogs(SMnode *pMnode, STrans *pTrans) {
     if (code != 0) {
       mError("trans:%d, failed to execute redo logs since %s", pTrans->id, terrstr())
     } else {
-      mTrace("trans:%d, execute redo logs finished", pTrans->id)
+      mDebug("trans:%d, execute redo logs finished", pTrans->id)
     }
   }
 
@@ -632,7 +668,7 @@ static int32_t mndTransExecuteUndoLogs(SMnode *pMnode, STrans *pTrans) {
     if (code != 0) {
       mError("trans:%d, failed to execute undo logs since %s", pTrans->id, terrstr())
     } else {
-      mTrace("trans:%d, execute undo logs finished", pTrans->id)
+      mDebug("trans:%d, execute undo logs finished", pTrans->id)
     }
   }
 
@@ -646,47 +682,70 @@ static int32_t mndTransExecuteCommitLogs(SMnode *pMnode, STrans *pTrans) {
     if (code != 0) {
       mError("trans:%d, failed to execute commit logs since %s", pTrans->id, terrstr())
     } else {
-      mTrace("trans:%d, execute commit logs finished", pTrans->id)
+      mDebug("trans:%d, execute commit logs finished", pTrans->id)
     }
   }
 
   return code;
 }
 
-static int32_t mndTransExecuteActions(SMnode *pMnode, SArray *pArray) {
-#if 0
-  int32_t arraySize = taosArrayGetSize(pArray);
-  for (int32_t i = 0; i < arraySize; ++i) {
-    STransAction *pAction = taosArrayGet(pArray, i);
+static int32_t mndTransExecuteActions(SMnode *pMnode, STrans *pTrans, SArray *pArray) {
+  int32_t numOfActions = taosArrayGetSize(pArray);
+  if (numOfActions == 0) return 0;
+
+  for (int32_t action = 0; action < numOfActions; ++action) {
+    STransAction *pAction = taosArrayGet(pArray, action);
+    if (pAction == NULL) continue;
+    if (pAction->msgSent) continue;
 
-    SRpcMsg rpcMsg = {.msgType = pAction->msgType, .contLen = pAction->contLen};
+    int64_t signature = pTrans->id;
+    signature = (signature << 32);
+    signature += action;
+
+    SRpcMsg rpcMsg = {.msgType = pAction->msgType, .contLen = pAction->contLen, .ahandle = (void *)signature};
     rpcMsg.pCont = rpcMallocCont(pAction->contLen);
     if (rpcMsg.pCont == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
     }
     memcpy(rpcMsg.pCont, pAction->pCont, pAction->contLen);
+
+    pAction->msgSent = 1;
+    pAction->msgReceived = 0;
+    pAction->errCode = 0;
+
+    mDebug("trans:%d, action:%d is sent", pTrans->id, action);
     mndSendMsgToDnode(pMnode, &pAction->epSet, &rpcMsg);
   }
 
-  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
-#else
-  return 0;
-#endif
+  int32_t numOfReceivedMsgs = 0;
+  int32_t errorCode = 0;
+  for (int32_t action = 0; action < numOfActions; ++action) {
+    STransAction *pAction = taosArrayGet(pArray, action);
+    if (pAction == NULL) continue;
+    if (pAction->msgSent && pAction->msgReceived) {
+      numOfReceivedMsgs++;
+      if (pAction->errCode != 0) {
+        errorCode = pAction->errCode;
+      }
+    }
+  }
+
+  if (numOfReceivedMsgs == numOfActions) {
+    mDebug("trans:%d, all %d actions executed, code:0x%x", pTrans->id, numOfActions, errorCode);
+    terrno = errorCode;
+    return errorCode;
+  } else {
+    return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+  }
 }
 
 static int32_t mndTransExecuteRedoActions(SMnode *pMnode, STrans *pTrans) {
-  if (taosArrayGetSize(pTrans->redoActions) <= 0) return 0;
-
-  mTrace("trans:%d, start to execute redo actions", pTrans->id);
-  return mndTransExecuteActions(pMnode, pTrans->redoActions);
+  return mndTransExecuteActions(pMnode, pTrans, pTrans->redoActions);
 }
 
 static int32_t mndTransExecuteUndoActions(SMnode *pMnode, STrans *pTrans) {
-  if (taosArrayGetSize(pTrans->undoActions) <= 0) return 0;
-
-  mTrace("trans:%d, start to execute undo actions", pTrans->id);
-  return mndTransExecuteActions(pMnode, pTrans->undoActions);
+  return mndTransExecuteActions(pMnode, pTrans, pTrans->undoActions);
 }
 
 static int32_t mndTransPerformPrepareStage(SMnode *pMnode, STrans *pTrans) {
@@ -694,7 +753,7 @@ static int32_t mndTransPerformPrepareStage(SMnode *pMnode, STrans *pTrans) {
 
   if (code == 0) {
     pTrans->stage = TRN_STAGE_EXECUTE;
-    mTrace("trans:%d, stage from prepare to execute", pTrans->id);
+    mDebug("trans:%d, stage from prepare to execute", pTrans->id);
   } else {
     pTrans->stage = TRN_STAGE_ROLLBACK;
     mError("trans:%d, stage from prepare to rollback since %s", pTrans->id, terrstr());
@@ -708,17 +767,17 @@ static int32_t mndTransPerformExecuteStage(SMnode *pMnode, STrans *pTrans) {
 
   if (code == 0) {
     pTrans->stage = TRN_STAGE_COMMIT;
-    mTrace("trans:%d, stage from execute to commit", pTrans->id);
+    mDebug("trans:%d, stage from execute to commit", pTrans->id);
   } else if (code == TSDB_CODE_MND_ACTION_IN_PROGRESS) {
-    mTrace("trans:%d, stage keep on execute since %s", pTrans->id, terrstr(code));
+    mDebug("trans:%d, stage keep on execute since %s", pTrans->id, tstrerror(code));
     return code;
   } else {
     if (pTrans->policy == TRN_POLICY_ROLLBACK) {
       pTrans->stage = TRN_STAGE_ROLLBACK;
       mError("trans:%d, stage from execute to rollback since %s", pTrans->id, terrstr());
     } else {
-      pTrans->stage = TRN_STAGE_RETRY;
-      mError("trans:%d, stage from execute to retry since %s", pTrans->id, terrstr());
+      pTrans->stage = TRN_STAGE_EXECUTE;
+      mError("trans:%d, stage keep on execute since %s", pTrans->id, terrstr());
     }
   }
 
@@ -726,29 +785,16 @@ static int32_t mndTransPerformExecuteStage(SMnode *pMnode, STrans *pTrans) {
 }
 
 static int32_t mndTransPerformCommitStage(SMnode *pMnode, STrans *pTrans) {
-  int32_t code = mndTransExecuteCommitLogs(pMnode, pTrans);
-
-  if (code == 0) {
-    pTrans->stage = TRN_STAGE_OVER;
-    mTrace("trans:%d, commit stage finished", pTrans->id);
-  } else {
-    if (pTrans->policy == TRN_POLICY_ROLLBACK) {
-      pTrans->stage = TRN_STAGE_ROLLBACK;
-      mError("trans:%d, stage from commit to rollback since %s", pTrans->id, terrstr());
-    } else {
-      pTrans->stage = TRN_STAGE_RETRY;
-      mError("trans:%d, stage from commit to retry since %s", pTrans->id, terrstr());
-    }
-  }
-
-  return code;
+  mndTransExecuteCommitLogs(pMnode, pTrans);
+  pTrans->stage = TRN_STAGE_OVER;
+  return 0;
 }
 
 static int32_t mndTransPerformRollbackStage(SMnode *pMnode, STrans *pTrans) {
   int32_t code = mndTransExecuteUndoActions(pMnode, pTrans);
 
   if (code == 0) {
-    mTrace("trans:%d, rollbacked", pTrans->id);
+    mDebug("trans:%d, rollbacked", pTrans->id);
   } else {
     pTrans->stage = TRN_STAGE_ROLLBACK;
     mError("trans:%d, stage keep on rollback since %s", pTrans->id, terrstr());
@@ -757,20 +803,6 @@ static int32_t mndTransPerformRollbackStage(SMnode *pMnode, STrans *pTrans) {
   return code;
 }
 
-static int32_t mndTransPerformRetryStage(SMnode *pMnode, STrans *pTrans) {
-  int32_t code = mndTransExecuteRedoActions(pMnode, pTrans);
-
-  if (code == 0) {
-    pTrans->stage = TRN_STAGE_COMMIT;
-    mTrace("trans:%d, stage from retry to commit", pTrans->id);
-  } else {
-    pTrans->stage = TRN_STAGE_RETRY;
-    mError("trans:%d, stage keep on retry since %s", pTrans->id, terrstr());
-  }
-
-  return code;
-}
-
 static void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
   int32_t code = 0;
 
@@ -785,7 +817,7 @@ static void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
       case TRN_STAGE_COMMIT:
         code = mndTransCommit(pMnode, pTrans);
         if (code == 0) {
-          code = mndTransPerformCommitStage(pMnode, pTrans);
+          mndTransPerformCommitStage(pMnode, pTrans);
         }
         break;
       case TRN_STAGE_ROLLBACK:
@@ -794,9 +826,6 @@ static void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
           code = mndTransRollback(pMnode, pTrans);
         }
         break;
-      case TRN_STAGE_RETRY:
-        code = mndTransPerformRetryStage(pMnode, pTrans);
-        break;
       default:
         mndTransSendRpcRsp(pTrans, 0);
         return;

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

@@ -20,7 +20,7 @@
 #include "mndTrans.h"
 #include "tkey.h"
 
-#define TSDB_USER_VER 1
+#define TSDB_USER_VER_NUMBER 1
 #define TSDB_USER_RESERVE_SIZE 64
 
 static int32_t  mndCreateDefaultUsers(SMnode *pMnode);
@@ -94,7 +94,7 @@ static int32_t mndCreateDefaultUsers(SMnode *pMnode) {
 }
 
 static SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, TSDB_USER_VER, sizeof(SUserObj) + TSDB_USER_RESERVE_SIZE);
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, TSDB_USER_VER_NUMBER, sizeof(SUserObj) + TSDB_USER_RESERVE_SIZE);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -114,7 +114,7 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != TSDB_USER_VER) {
+  if (sver != TSDB_USER_VER_NUMBER) {
     mError("failed to decode user since %s", terrstr());
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     return NULL;
@@ -175,8 +175,12 @@ static int32_t mndUserActionUpdate(SSdb *pSdb, SUserObj *pOldUser, SUserObj *pNe
 }
 
 SUserObj *mndAcquireUser(SMnode *pMnode, char *userName) {
-  SSdb *pSdb = pMnode->pSdb;
-  return sdbAcquire(pSdb, SDB_USER, userName);
+  SSdb     *pSdb = pMnode->pSdb;
+  SUserObj *pUser = sdbAcquire(pSdb, SDB_USER, userName);
+  if (pUser == NULL) {
+    terrno = TSDB_CODE_MND_DB_NOT_EXIST;
+  }
+  return pUser;
 }
 
 void mndReleaseUser(SMnode *pMnode, SUserObj *pUser) {

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

@@ -21,7 +21,7 @@
 #include "mndShow.h"
 #include "mndTrans.h"
 
-#define TSDB_VGROUP_VER_NUM 1
+#define TSDB_VGROUP_VER_NUMBER 1
 #define TSDB_VGROUP_RESERVE_SIZE 64
 
 static SSdbRow *mndVgroupActionDecode(SSdbRaw *pRaw);
@@ -70,7 +70,7 @@ int32_t mndInitVgroup(SMnode *pMnode) {
 void mndCleanupVgroup(SMnode *pMnode) {}
 
 SSdbRaw *mndVgroupActionEncode(SVgObj *pVgroup) {
-  SSdbRaw *pRaw = sdbAllocRaw(SDB_VGROUP, TSDB_VGROUP_VER_NUM, sizeof(SVgObj) + TSDB_VGROUP_RESERVE_SIZE);
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_VGROUP, TSDB_VGROUP_VER_NUMBER, sizeof(SVgObj) + TSDB_VGROUP_RESERVE_SIZE);
   if (pRaw == NULL) return NULL;
 
   int32_t dataPos = 0;
@@ -98,7 +98,7 @@ SSdbRow *mndVgroupActionDecode(SSdbRaw *pRaw) {
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
 
-  if (sver != TSDB_VGROUP_VER_NUM) {
+  if (sver != TSDB_VGROUP_VER_NUMBER) {
     mError("failed to decode vgroup since %s", terrstr());
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     return NULL;
@@ -142,14 +142,20 @@ static int32_t mndVgroupActionUpdate(SSdb *pSdb, SVgObj *pOldVgroup, SVgObj *pNe
   mTrace("vgId:%d, perform update action", pOldVgroup->vgId);
   pOldVgroup->updateTime = pNewVgroup->updateTime;
   pOldVgroup->version = pNewVgroup->version;
+  pOldVgroup->hashBegin = pNewVgroup->hashBegin;
+  pOldVgroup->hashEnd = pNewVgroup->hashEnd;
   pOldVgroup->replica = pNewVgroup->replica;
   memcpy(pOldVgroup->vnodeGid, pNewVgroup->vnodeGid, TSDB_MAX_REPLICA * sizeof(SVnodeGid));
   return 0;
 }
 
 SVgObj *mndAcquireVgroup(SMnode *pMnode, int32_t vgId) {
-  SSdb *pSdb = pMnode->pSdb;
-  return sdbAcquire(pSdb, SDB_VGROUP, &vgId);
+  SSdb   *pSdb = pMnode->pSdb;
+  SVgObj *pVgroup = sdbAcquire(pSdb, SDB_VGROUP, &vgId);
+  if (pVgroup == NULL) {
+    terrno = TSDB_CODE_MND_VGROUP_NOT_EXIST;
+  }
+  return pVgroup;
 }
 
 void mndReleaseVgroup(SMnode *pMnode, SVgObj *pVgroup) {
@@ -158,16 +164,17 @@ void mndReleaseVgroup(SMnode *pMnode, SVgObj *pVgroup) {
 }
 
 SCreateVnodeMsg *mndBuildCreateVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup) {
-  SCreateVnodeMsg *pCreate = malloc(sizeof(SCreateVnodeMsg));
+  SCreateVnodeMsg *pCreate = calloc(1, sizeof(SCreateVnodeMsg));
   if (pCreate == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
   }
 
-  pCreate->dnodeId = htonl(pDnode->id);
   pCreate->vgId = htonl(pVgroup->vgId);
+  pCreate->dnodeId = htonl(pDnode->id);
   memcpy(pCreate->db, pDb->name, TSDB_FULL_DB_NAME_LEN);
   pCreate->dbUid = htobe64(pDb->uid);
+  pCreate->vgVersion = htonl(pVgroup->version);
   pCreate->cacheBlockSize = htonl(pDb->cfg.cacheBlockSize);
   pCreate->totalBlocks = htonl(pDb->cfg.totalBlocks);
   pCreate->daysPerFile = htonl(pDb->cfg.daysPerFile);
@@ -193,7 +200,6 @@ SCreateVnodeMsg *mndBuildCreateVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbOb
     SDnodeObj *pVgidDnode = mndAcquireDnode(pMnode, pVgid->dnodeId);
     if (pVgidDnode == NULL) {
       free(pCreate);
-      terrno = TSDB_CODE_MND_APP_ERROR;
       return NULL;
     }
 
@@ -217,7 +223,7 @@ SCreateVnodeMsg *mndBuildCreateVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbOb
 }
 
 SDropVnodeMsg *mndBuildDropVnodeMsg(SMnode *pMnode, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup) {
-  SDropVnodeMsg *pDrop = malloc(sizeof(SDropVnodeMsg));
+  SDropVnodeMsg *pDrop = calloc(1, sizeof(SDropVnodeMsg));
   if (pDrop == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
@@ -269,7 +275,7 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
     return -1;
   }
 
-  int32_t  alloceVgroups = 0;
+  int32_t  allocedVgroups = 0;
   int32_t  maxVgId = sdbGetMaxId(pMnode->pSdb, SDB_VGROUP);
   uint32_t hashMin = 0;
   uint32_t hashMax = UINT32_MAX;
@@ -281,7 +287,6 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
     pVgroup->createdTime = taosGetTimestampMs();
     pVgroup->updateTime = pVgroups->createdTime;
     pVgroup->version = 1;
-    pVgroup->dbUid = pDb->uid;
     pVgroup->hashBegin = hashMin + hashInterval * v;
     if (v == pDb->cfg.numOfVgroups - 1) {
       pVgroup->hashEnd = hashMax;
@@ -290,6 +295,7 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
     }
 
     memcpy(pVgroup->dbName, pDb->name, TSDB_FULL_DB_NAME_LEN);
+    pVgroup->dbUid = pDb->uid;
     pVgroup->replica = pDb->cfg.replications;
 
     if (mndGetAvailableDnode(pMnode, pVgroup) != 0) {
@@ -298,22 +304,46 @@ int32_t mndAllocVgroup(SMnode *pMnode, SDbObj *pDb, SVgObj **ppVgroups) {
       return -1;
     }
 
-    alloceVgroups++;
+    allocedVgroups++;
   }
 
   *ppVgroups = pVgroups;
   return 0;
 }
 
-static int32_t mndProcessCreateVnodeRsp(SMnodeMsg *pMsg) { return 0; }
+SEpSet mndGetVgroupEpset(SMnode *pMnode, SVgObj *pVgroup) {
+  SEpSet epset = {0};
+
+  for (int32_t v = 0; v < pVgroup->replica; ++v) {
+    SVnodeGid *pVgid = &pVgroup->vnodeGid[v];
+    SDnodeObj *pDnode = mndAcquireDnode(pMnode, pVgid->dnodeId);
+    if (pDnode == NULL) continue;
+
+    if (pVgid->role == TAOS_SYNC_STATE_LEADER) {
+      epset.inUse = epset.numOfEps;
+    }
+
+    epset.port[epset.numOfEps] = pDnode->port;
+    memcpy(&epset.fqdn[epset.numOfEps], pDnode->fqdn, TSDB_FQDN_LEN);
+    epset.numOfEps++;
+    mndReleaseDnode(pMnode, pDnode);
+  }
+
+  return epset;
+}
+
+static int32_t mndProcessCreateVnodeRsp(SMnodeMsg *pMsg) {
+  mndTransHandleActionRsp(pMsg);
+  return 0;
+}
+
 static int32_t mndProcessAlterVnodeRsp(SMnodeMsg *pMsg) { return 0; }
 static int32_t mndProcessDropVnodeRsp(SMnodeMsg *pMsg) { return 0; }
 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) {
-  SSdb *pSdb = pMnode->pSdb;
-
+  SSdb   *pSdb = pMnode->pSdb;
   SDbObj *pDb = mndAcquireDb(pMnode, dbName);
   if (pDb == NULL) {
     terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
@@ -329,7 +359,7 @@ static int32_t mndGetVgroupMaxReplica(SMnode *pMnode, char *dbName, int8_t *pRep
     pIter = sdbFetch(pSdb, SDB_VGROUP, pIter, (void **)&pVgroup);
     if (pIter == NULL) break;
 
-    if (strcmp(pVgroup->dbName, dbName) == 0) {
+    if (pVgroup->dbUid == pDb->uid) {
       replica = MAX(replica, pVgroup->replica);
       numOfVgroups++;
     }
@@ -441,11 +471,25 @@ static void mndCancelGetNextVgroup(SMnode *pMnode, void *pIter) {
 }
 
 static int32_t mndGetVnodesNum(SMnode *pMnode, int32_t dnodeId) {
-  if (dnodeId == 0) {
-    return 0;
+  SSdb   *pSdb = pMnode->pSdb;
+  int32_t numOfVnodes = 0;
+  void   *pIter = NULL;
+
+  while (1) {
+    SVgObj *pVgroup = NULL;
+    pIter = sdbFetch(pSdb, SDB_VGROUP, pIter, (void **)&pVgroup);
+    if (pIter == NULL) break;
+
+    for (int32_t v = 0; v < pVgroup->replica; ++v) {
+      if (pVgroup->vnodeGid[v].dnodeId == dnodeId) {
+        numOfVnodes++;
+      }
+    }
+
+    sdbRelease(pSdb, pVgroup);
   }
 
-  return 0;
+  return numOfVnodes;
 }
 
 static int32_t mndGetVnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {

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

@@ -225,7 +225,7 @@ static int32_t mndSetOptions(SMnode *pMnode, const SMnodeOpt *pOption) {
   }
 
   return 0;
-}
+} 
 
 SMnode *mndOpen(const char *path, const SMnodeOpt *pOption) {
   mDebug("start to open mnode in %s", path);
@@ -237,6 +237,9 @@ SMnode *mndOpen(const char *path, const SMnodeOpt *pOption) {
     return NULL;
   }
 
+  char timestr[24] = "1970-01-01 00:00:00.00";
+  (void)taosParseTime(timestr, &pMnode->checkTime, (int32_t)strlen(timestr), TSDB_TIME_PRECISION_MILLI, 0);
+
   pMnode->pSteps = taosArrayInit(24, sizeof(SMnodeStep));
   if (pMnode->pSteps == NULL) {
     free(pMnode);

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

@@ -240,6 +240,8 @@ void sdbRelease(SSdb *pSdb, void *pObj) {
 }
 
 void *sdbFetch(SSdb *pSdb, ESdbType type, void *pIter, void **ppObj) {
+  *ppObj = NULL;
+
   SHashObj *hash = sdbGetHash(pSdb, type);
   if (hash == NULL) return NULL;
 

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/vnode/impl/CMakeLists.txt

@@ -19,5 +19,5 @@ target_link_libraries(
 
 # test
 if(${BUILD_TEST})
-   #add_subdirectory(test)
+   add_subdirectory(test)
 endif(${BUILD_TEST})
\ No newline at end of file

source/dnode/vnode/impl/inc/vnodeDef.h

@@ -71,6 +71,7 @@ struct SVnode {
   SWal*       pWal;
   SVnodeSync* pSync;
   SVnodeFS*   pFs;
+  tsem_t      canCommit;
 };
 
 int vnodeScheduleTask(SVnodeTask* task);

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

@@ -38,9 +38,10 @@ int vnodeCommit(void *arg) {
 
   metaCommit(pVnode->pMeta);
   tqCommit(pVnode->pTq);
-  tsdbCommit(pVnode->pTq);
+  tsdbCommit(pVnode->pTsdb);
 
   vnodeBufPoolRecycle(pVnode);
+  tsem_post(&(pVnode->canCommit));
   // TODO
   return 0;
 }

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

@@ -15,53 +15,31 @@
 
 #define _DEFAULT_SOURCE
 #include "vnodeInt.h"
-#include "tqueue.h"
 
 int32_t vnodeAlter(SVnode *pVnode, const SVnodeCfg *pCfg) { return 0; }
-SVnode *vnodeCreate(int32_t vgId, const char *path, const SVnodeCfg *pCfg) { return NULL; }
-void    vnodeDrop(SVnode *pVnode) {}
+
 int32_t vnodeCompact(SVnode *pVnode) { return 0; }
+
 int32_t vnodeSync(SVnode *pVnode) { return 0; }
 
 int32_t vnodeGetLoad(SVnode *pVnode, SVnodeLoad *pLoad) { return 0; }
 
-SVnodeMsg *vnodeInitMsg(int32_t msgNum) {
-  SVnodeMsg *pMsg = taosAllocateQitem(msgNum * sizeof(SRpcMsg *) + sizeof(SVnodeMsg));
-  if (pMsg == NULL) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  } else {
-    pMsg->allocNum = msgNum;
-    return pMsg;
-  }
+int vnodeProcessQueryReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+  vInfo("query message is processed");
+  return 0;
 }
 
-int32_t vnodeAppendMsg(SVnodeMsg *pMsg, SRpcMsg *pRpcMsg) {
-  if (pMsg->curNum >= pMsg->allocNum) {
-    return TSDB_CODE_OUT_OF_MEMORY;
-  }
-
-  pMsg->rpcMsg[pMsg->curNum++] = *pRpcMsg;
+int vnodeProcessFetchReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+  vInfo("fetch message is processed");
+  return 0;
 }
 
-void vnodeCleanupMsg(SVnodeMsg *pMsg) {
-  for (int32_t i = 0; i < pMsg->curNum; ++i) {
-    rpcFreeCont(pMsg->rpcMsg[i].pCont);
-  }
-  taosFreeQitem(pMsg);
+int vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+  vInfo("sync message is processed");
+  return 0;
 }
 
-void vnodeProcessMsg(SVnode *pVnode, SVnodeMsg *pMsg, EVnMsgType msgType) {
-  switch (msgType) {
-    case VN_MSG_TYPE_WRITE:
-      break;
-    case VN_MSG_TYPE_APPLY:
-      break;
-    case VN_MSG_TYPE_SYNC:
-      break;
-    case VN_MSG_TYPE_QUERY:
-      break;
-    case VN_MSG_TYPE_FETCH:
-      break;
-  }
-}
+int vnodeProcessConsumeReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+  vInfo("consume message is processed");
+  return 0;
+}
\ No newline at end of file

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

@@ -74,11 +74,14 @@ static SVnode *vnodeNew(const char *path, const SVnodeCfg *pVnodeCfg) {
   pVnode->path = strdup(path);
   vnodeOptionsCopy(&(pVnode->config), pVnodeCfg);
 
+  tsem_init(&(pVnode->canCommit), 0, 1);
+
   return pVnode;
 }
 
 static void vnodeFree(SVnode *pVnode) {
   if (pVnode) {
+    tsem_destroy(&(pVnode->canCommit));
     tfree(pVnode->path);
     free(pVnode);
   }

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

@@ -16,29 +16,25 @@
 #include "vnodeDef.h"
 
 int vnodeProcessNoWalWMsgs(SVnode *pVnode, SRpcMsg *pMsg) {
-  SVnodeReq *pVnodeReq;
-
   switch (pMsg->msgType) {
-    case TSDB_MSG_TYPE_MQ_SET:
+    case TSDB_MSG_TYPE_MQ_SET_CUR:
       if (tqSetCursor(pVnode->pTq, pMsg->pCont) < 0) {
         // TODO: handle error
       }
       break;
   }
-
-  void *pBuf = pMsg->pCont;
   return 0;
 }
 
 int vnodeProcessWMsgs(SVnode *pVnode, SArray *pMsgs) {
-  SRpcMsg   *pMsg;
+  SRpcMsg *  pMsg;
   SVnodeReq *pVnodeReq;
 
   for (int i = 0; i < taosArrayGetSize(pMsgs); i++) {
     pMsg = *(SRpcMsg **)taosArrayGet(pMsgs, i);
 
     // ser request version
-    void   *pBuf = pMsg->pCont;
+    void *  pBuf = pMsg->pCont;
     int64_t ver = pVnode->state.processed++;
     taosEncodeFixedU64(&pBuf, ver);
 
@@ -49,68 +45,63 @@ int vnodeProcessWMsgs(SVnode *pVnode, SArray *pMsgs) {
 
   walFsync(pVnode->pWal, false);
 
-  // Apply each request now
-  for (int i = 0; i < taosArrayGetSize(pMsgs); i++) {
-    pMsg = *(SRpcMsg **)taosArrayGet(pMsgs, i);
-    SVnodeReq vReq;
+  // TODO: Integrate RAFT module here
 
-    // Apply the request
-    {
-      void *ptr = vnodeMalloc(pVnode, pMsg->contLen);
-      if (ptr == NULL) {
-        // TODO: handle error
-      }
+  return 0;
+}
+
+int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
+  SVnodeReq vReq;
+  void *    ptr = vnodeMalloc(pVnode, pMsg->contLen);
+  if (ptr == NULL) {
+    // TODO: handle error
+  }
 
-      // TODO: copy here need to be extended
-      memcpy(ptr, pMsg->pCont, pMsg->contLen);
+  // TODO: copy here need to be extended
+  memcpy(ptr, pMsg->pCont, pMsg->contLen);
 
-      // todo: change the interface here
-      uint64_t ver;
-      taosDecodeFixedU64(pMsg->pCont, &ver);
-      if (tqPushMsg(pVnode->pTq, ptr, ver) < 0) {
+  // todo: change the interface here
+  uint64_t ver;
+  taosDecodeFixedU64(pMsg->pCont, &ver);
+  if (tqPushMsg(pVnode->pTq, ptr, ver) < 0) {
+    // TODO: handle error
+  }
+
+  vnodeParseReq(pMsg->pCont, &vReq, pMsg->msgType);
+
+  switch (pMsg->msgType) {
+    case TSDB_MSG_TYPE_CREATE_STB_IN:
+    case TSDB_MSG_TYPE_CREATE_TABLE:
+      if (metaCreateTable(pVnode->pMeta, &(vReq.ctReq)) < 0) {
         // TODO: handle error
       }
 
-      vnodeParseReq(pMsg->pCont, &vReq, pMsg->msgType);
-
-      switch (pMsg->msgType) {
-        case TSDB_MSG_TYPE_CREATE_TABLE:
-          if (metaCreateTable(pVnode->pMeta, &(vReq.ctReq)) < 0) {
-            // TODO: handle error
-          }
-
-          // TODO: maybe need to clear the requst struct
-          break;
-        case TSDB_MSG_TYPE_DROP_TABLE:
-          if (metaDropTable(pVnode->pMeta, vReq.dtReq.uid) < 0) {
-            // TODO: handle error
-          }
-          break;
-        case TSDB_MSG_TYPE_SUBMIT:
-          if (tsdbInsertData(pVnode->pTsdb, (SSubmitMsg *)ptr) < 0) {
-            // TODO: handle error
-          }
-          break;
-        default:
-          break;
+      // TODO: maybe need to clear the requst struct
+      break;
+    case TSDB_MSG_TYPE_DROP_STB_IN:
+    case TSDB_MSG_TYPE_DROP_TABLE:
+      if (metaDropTable(pVnode->pMeta, vReq.dtReq.uid) < 0) {
+        // TODO: handle error
       }
-
-      pVnode->state.applied = ver;
-    }
-
-    // Check if it needs to commit
-    if (vnodeShouldCommit(pVnode)) {
-      if (vnodeAsyncCommit(pVnode) < 0) {
+      break;
+    case TSDB_MSG_TYPE_SUBMIT:
+      if (tsdbInsertData(pVnode->pTsdb, (SSubmitMsg *)ptr) < 0) {
         // TODO: handle error
       }
-    }
+      break;
+    default:
+      break;
   }
 
-  return 0;
-}
+  pVnode->state.applied = ver;
 
-int vnodeApplyWMsg(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
-  // TODO
+  // Check if it needs to commit
+  if (vnodeShouldCommit(pVnode)) {
+    tsem_wait(&(pVnode->canCommit));
+    if (vnodeAsyncCommit(pVnode) < 0) {
+      // TODO: handle error
+    }
+  }
   return 0;
 }
 

source/dnode/vnode/impl/test/vnodeApiTests.cpp

@@ -81,7 +81,8 @@ static void vtBuildCreateStbReq(tb_uid_t suid, char *tbname, SRpcMsg **ppMsg) {
   pSchema = vtCreateBasicSchema();
   pTagSchema = vtCreateBasicTagSchema();
 
-  SVnodeReq vCreateSTbReq = VNODE_INIT_CREATE_STB_REQ(tbname, UINT32_MAX, UINT32_MAX, suid, pSchema, pTagSchema);
+  SVnodeReq vCreateSTbReq;
+  vnodeSetCreateStbReq(&vCreateSTbReq, tbname, UINT32_MAX, UINT32_MAX, suid, pSchema, pTagSchema);
 
   zs = vnodeBuildReq(NULL, &vCreateSTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
   pMsg = (SRpcMsg *)malloc(sizeof(SRpcMsg) + zs);
@@ -104,7 +105,8 @@ static void vtBuildCreateCtbReq(tb_uid_t suid, char *tbname, SRpcMsg **ppMsg) {
   int      tz;
   SKVRow   pTag = vtCreateBasicTag();
 
-  SVnodeReq vCreateCTbReq = VNODE_INIT_CREATE_CTB_REQ(tbname, UINT32_MAX, UINT32_MAX, suid, pTag);
+  SVnodeReq vCreateCTbReq;
+  vnodeSetCreateCtbReq(&vCreateCTbReq, tbname, UINT32_MAX, UINT32_MAX, suid, pTag);
 
   tz = vnodeBuildReq(NULL, &vCreateCTbReq, TSDB_MSG_TYPE_CREATE_TABLE);
   pMsg = (SRpcMsg *)malloc(sizeof(SRpcMsg) + tz);
@@ -166,6 +168,21 @@ static void vtClearMsgBatch(SArray *pMsgArr) {
   taosArrayClear(pMsgArr);
 }
 
+static void vtProcessAndApplyReqs(SVnode *pVnode, SArray *pMsgArr) {
+  int      rcode;
+  SRpcMsg *pReq;
+  SRpcMsg *pRsp;
+
+  rcode = vnodeProcessWMsgs(pVnode, pMsgArr);
+  GTEST_ASSERT_EQ(rcode, 0);
+
+  for (size_t i = 0; i < taosArrayGetSize(pMsgArr); i++) {
+    pReq = *(SRpcMsg **)taosArrayGet(pMsgArr, i);
+    rcode = vnodeApplyWMsg(pVnode, pReq, NULL);
+    GTEST_ASSERT_EQ(rcode, 0);
+  }
+}
+
 TEST(vnodeApiTest, vnode_simple_create_table_test) {
   tb_uid_t suid = 1638166374163;
   SRpcMsg *pMsg;
@@ -189,8 +206,7 @@ TEST(vnodeApiTest, vnode_simple_create_table_test) {
   sprintf(tbname, "st");
   vtBuildCreateStbReq(suid, tbname, &pMsg);
   taosArrayPush(pMsgArr, &pMsg);
-  rcode = vnodeProcessWMsgs(pVnode, pMsgArr);
-  ASSERT_EQ(rcode, 0);
+  vtProcessAndApplyReqs(pVnode, pMsgArr);
   vtClearMsgBatch(pMsgArr);
 
   // CREATE A LOT OF CHILD TABLES
@@ -203,8 +219,7 @@ TEST(vnodeApiTest, vnode_simple_create_table_test) {
     }
 
     // Process request batch
-    rcode = vnodeProcessWMsgs(pVnode, pMsgArr);
-    ASSERT_EQ(rcode, 0);
+    vtProcessAndApplyReqs(pVnode, pMsgArr);
 
     // Clear request batch
     vtClearMsgBatch(pMsgArr);
@@ -242,16 +257,14 @@ TEST(vnodeApiTest, vnode_simple_insert_test) {
   sprintf(tbname, "st");
   vtBuildCreateStbReq(suid, tbname, &pMsg);
   taosArrayPush(pMsgArr, &pMsg);
-  rcode = vnodeProcessWMsgs(pVnode, pMsgArr);
-  GTEST_ASSERT_EQ(rcode, 0);
+  vtProcessAndApplyReqs(pVnode, pMsgArr);
   vtClearMsgBatch(pMsgArr);
 
   // 2. CREATE A CHILD TABLE
   sprintf(tbname, "t0");
   vtBuildCreateCtbReq(suid, tbname, &pMsg);
   taosArrayPush(pMsgArr, &pMsg);
-  rcode = vnodeProcessWMsgs(pVnode, pMsgArr);
-  GTEST_ASSERT_EQ(rcode, 0);
+  vtProcessAndApplyReqs(pVnode, pMsgArr);
   vtClearMsgBatch(pMsgArr);
 
   // 3. WRITE A LOT OF TIME-SERIES DATA
@@ -260,8 +273,7 @@ TEST(vnodeApiTest, vnode_simple_insert_test) {
       vtBuildSubmitReq(&pMsg);
       taosArrayPush(pMsgArr, &pMsg);
     }
-    rcode = vnodeProcessWMsgs(pVnode, pMsgArr);
-    GTEST_ASSERT_EQ(rcode, 0);
+    vtProcessAndApplyReqs(pVnode, pMsgArr);
     vtClearMsgBatch(pMsgArr);
   }
 

source/dnode/vnode/tq/CMakeLists.txt

@@ -11,6 +11,7 @@ target_link_libraries(
     PUBLIC wal
     PUBLIC os
     PUBLIC util
+    PUBLIC common
 )
 
 if(${BUILD_TEST})

source/dnode/vnode/tq/inc/tqInt.h

@@ -17,11 +17,20 @@
 #define _TD_TQ_INT_H_
 
 #include "tq.h"
-
+#include "tlog.h"
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+extern int32_t tqDebugFlag;
+
+#define tqFatal(...) { if (tqDebugFlag & DEBUG_FATAL) { taosPrintLog("TQ  FATAL ", 255, __VA_ARGS__); }}
+#define tqError(...) { if (tqDebugFlag & DEBUG_ERROR) { taosPrintLog("TQ  ERROR ", 255, __VA_ARGS__); }}
+#define tqWarn(...)  { if (tqDebugFlag & DEBUG_WARN)  { taosPrintLog("TQ  WARN ", 255, __VA_ARGS__); }}
+#define tqInfo(...)  { if (tqDebugFlag & DEBUG_INFO)  { taosPrintLog("TQ  ", 255, __VA_ARGS__); }}
+#define tqDebug(...) { if (tqDebugFlag & DEBUG_DEBUG) { taosPrintLog("TQ  ", tqDebugFlag, __VA_ARGS__); }}
+#define tqTrace(...) { if (tqDebugFlag & DEBUG_TRACE) { taosPrintLog("TQ  ", tqDebugFlag, __VA_ARGS__); }}
+
 // create persistent storage for meta info such as consuming offset
 // return value > 0: cgId
 // return value <= 0: error code

source/dnode/vnode/tq/inc/tqMetaStore.h

@@ -17,7 +17,7 @@
 #define _TQ_META_STORE_H_
 
 #include "os.h"
-#include "tq.h"
+#include "tqInt.h"
 
 #ifdef __cplusplus
 extern "C" {

source/dnode/vnode/tq/src/tq.c

@@ -24,89 +24,82 @@
 // handle management message
 //
 
-int tqGetgHandleSSize(const STqGroupHandle* gHandle);
-int tqBufHandleSSize();
-int tqBufItemSSize();
+int tqGroupSSize(const STqGroup* pGroup);
+int tqTopicSSize();
+int tqItemSSize();
 
-STqGroupHandle* tqFindHandle(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
-  STqGroupHandle* gHandle;
-  return NULL;
-}
+void* tqSerializeListHandle(STqList* listHandle, void* ptr);
+void* tqSerializeTopic(STqTopic* pTopic, void* ptr);
+void* tqSerializeItem(STqMsgItem* pItem, void* ptr);
 
-void* tqSerializeListHandle(STqListHandle* listHandle, void* ptr);
-void* tqSerializeBufHandle(STqBufferHandle* bufHandle, void* ptr);
-void* tqSerializeBufItem(STqBufferItem* bufItem, void* ptr);
-
-const void* tqDeserializeBufHandle(const void* pBytes, STqBufferHandle* bufHandle);
-const void* tqDeserializeBufItem(const void* pBytes, STqBufferItem* bufItem);
+const void* tqDeserializeTopic(const void* pBytes, STqTopic* pTopic);
+const void* tqDeserializeItem(const void* pBytes, STqMsgItem* pItem);
 
 STQ* tqOpen(const char* path, STqCfg* tqConfig, STqLogReader* tqLogReader, SMemAllocatorFactory* allocFac) {
   STQ* pTq = malloc(sizeof(STQ));
   if (pTq == NULL) {
-    // TODO: memory error
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return NULL;
   }
   pTq->path = strdup(path);
   pTq->tqConfig = tqConfig;
   pTq->tqLogReader = tqLogReader;
+#if 0
   pTq->tqMemRef.pAlloctorFactory = allocFac;
-  // pTq->tqMemRef.pAllocator = allocFac->create(allocFac);
+  pTq->tqMemRef.pAllocator = allocFac->create(allocFac);
   if (pTq->tqMemRef.pAllocator == NULL) {
-    // TODO
+    // TODO: error code of buffer pool
   }
-  pTq->tqMeta =
-      tqStoreOpen(path, (FTqSerialize)tqSerializeGroupHandle, (FTqDeserialize)tqDeserializeGroupHandle, free, 0);
+#endif
+  pTq->tqMeta = tqStoreOpen(path, (FTqSerialize)tqSerializeGroup, (FTqDeserialize)tqDeserializeGroup, free, 0);
   if (pTq->tqMeta == NULL) {
     // TODO: free STQ
     return NULL;
   }
   return pTq;
 }
-
-void tqClose(STQ*pTq) {
+void tqClose(STQ* pTq) {
   // TODO
 }
 
-static int tqProtoCheck(TmqMsgHead *pMsg) {
-  return pMsg->protoVer == 0;
-}
+static int tqProtoCheck(STqMsgHead* pMsg) { return pMsg->protoVer == 0; }
 
-static int tqAckOneTopic(STqBufferHandle* bHandle, TmqOneAck* pAck, STqQueryMsg** ppQuery) {
+static int tqAckOneTopic(STqTopic* pTopic, STqOneAck* pAck, STqQueryMsg** ppQuery) {
   // clean old item and move forward
   int32_t consumeOffset = pAck->consumeOffset;
   int     idx = consumeOffset % TQ_BUFFER_SIZE;
-  ASSERT(bHandle->buffer[idx].content && bHandle->buffer[idx].executor);
-  tfree(bHandle->buffer[idx].content);
+  ASSERT(pTopic->buffer[idx].content && pTopic->buffer[idx].executor);
+  tfree(pTopic->buffer[idx].content);
   if (1 /* TODO: need to launch new query */) {
     STqQueryMsg* pNewQuery = malloc(sizeof(STqQueryMsg));
     if (pNewQuery == NULL) {
-      // TODO: memory insufficient
+      terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
       return -1;
     }
     // TODO: lock executor
-    pNewQuery->exec->executor = bHandle->buffer[idx].executor;
     // TODO: read from wal and assign to src
-    pNewQuery->exec->src = 0;
-    pNewQuery->exec->dest = &bHandle->buffer[idx];
-    pNewQuery->next = *ppQuery;
-    *ppQuery = pNewQuery;
+    /*pNewQuery->exec->executor = pTopic->buffer[idx].executor;*/
+    /*pNewQuery->exec->src = 0;*/
+    /*pNewQuery->exec->dest = &pTopic->buffer[idx];*/
+    /*pNewQuery->next = *ppQuery;*/
+    /**ppQuery = pNewQuery;*/
   }
   return 0;
 }
 
-static int tqAck(STqGroupHandle* gHandle, TmqAcks* pAcks) {
+static int tqAck(STqGroup* pGroup, STqAcks* pAcks) {
   int32_t    ackNum = pAcks->ackNum;
-  TmqOneAck* acks = pAcks->acks;
+  STqOneAck* acks = pAcks->acks;
   // double ptr for acks and list
-  int            i = 0;
-  STqListHandle* node = gHandle->head;
-  int            ackCnt = 0;
-  STqQueryMsg*   pQuery = NULL;
+  int          i = 0;
+  STqList*     node = pGroup->head;
+  int          ackCnt = 0;
+  STqQueryMsg* pQuery = NULL;
   while (i < ackNum && node->next) {
-    if (acks[i].topicId == node->next->bufHandle.topicId) {
+    if (acks[i].topicId == node->next->topic.topicId) {
       ackCnt++;
-      tqAckOneTopic(&node->next->bufHandle, &acks[i], &pQuery);
-    } else if (acks[i].topicId < node->next->bufHandle.topicId) {
+      tqAckOneTopic(&node->next->topic, &acks[i], &pQuery);
+    } else if (acks[i].topicId < node->next->topic.topicId) {
       i++;
     } else {
       node = node->next;
@@ -118,52 +111,56 @@ static int tqAck(STqGroupHandle* gHandle, TmqAcks* pAcks) {
   return ackCnt;
 }
 
-static int tqCommitTCGroup(STqGroupHandle* handle) {
+static int tqCommitGroup(STqGroup* pGroup) {
   // persist modification into disk
   return 0;
 }
 
-int tqCreateTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId, STqGroupHandle** handle) {
+int tqCreateGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId, STqGroup** ppGroup) {
   // create in disk
-  STqGroupHandle* gHandle = (STqGroupHandle*)malloc(sizeof(STqGroupHandle));
-  if (gHandle == NULL) {
+  STqGroup* pGroup = (STqGroup*)malloc(sizeof(STqGroup));
+  if (pGroup == NULL) {
     // TODO
     return -1;
   }
-  memset(gHandle, 0, sizeof(STqGroupHandle));
+  *ppGroup = pGroup;
+  memset(pGroup, 0, sizeof(STqGroup));
 
   return 0;
 }
 
-STqGroupHandle* tqOpenTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
-  STqGroupHandle* gHandle = tqHandleGet(pTq->tqMeta, cId);
-  if (gHandle == NULL) {
-    int code = tqCreateTCGroup(pTq, topicId, cgId, cId, &gHandle);
-    if (code != 0) {
+STqGroup* tqOpenGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+  STqGroup* pGroup = tqHandleGet(pTq->tqMeta, cId);
+  if (pGroup == NULL) {
+    int code = tqCreateGroup(pTq, topicId, cgId, cId, &pGroup);
+    if (code < 0) {
       // TODO
       return NULL;
     }
+    tqHandleMovePut(pTq->tqMeta, cId, pGroup);
   }
+  ASSERT(pGroup);
 
-  // create
-  // open
-  return gHandle;
+  return pGroup;
 }
 
-int tqCloseTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) { return 0; }
+int tqCloseGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+  // TODO
+  return 0;
+}
 
-int tqDropTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+int tqDropGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
   // delete from disk
   return 0;
 }
 
-static int tqFetch(STqGroupHandle* gHandle, void** msg) {
-  STqListHandle* head = gHandle->head;
-  STqListHandle* node = head;
-  int            totSize = 0;
+static int tqFetch(STqGroup* pGroup, void** msg) {
+  STqList* head = pGroup->head;
+  STqList* node = head;
+  int      totSize = 0;
   // TODO: make it a macro
   int            sizeLimit = 4 * 1024;
-  TmqMsgContent* buffer = malloc(sizeLimit);
+  STqMsgContent* buffer = malloc(sizeLimit);
   if (buffer == NULL) {
     // TODO:memory insufficient
     return -1;
@@ -172,25 +169,25 @@ static int tqFetch(STqGroupHandle* gHandle, void** msg) {
   // until all topic iterated or msgs over sizeLimit
   while (node->next) {
     node = node->next;
-    STqBufferHandle* bufHandle = &node->bufHandle;
-    int              idx = bufHandle->nextConsumeOffset % TQ_BUFFER_SIZE;
-    if (bufHandle->buffer[idx].content != NULL && bufHandle->buffer[idx].offset == bufHandle->nextConsumeOffset) {
-      totSize += bufHandle->buffer[idx].size;
+    STqTopic* topicHandle = &node->topic;
+    int       idx = topicHandle->nextConsumeOffset % TQ_BUFFER_SIZE;
+    if (topicHandle->buffer[idx].content != NULL && topicHandle->buffer[idx].offset == topicHandle->nextConsumeOffset) {
+      totSize += topicHandle->buffer[idx].size;
       if (totSize > sizeLimit) {
         void* ptr = realloc(buffer, totSize);
         if (ptr == NULL) {
-          totSize -= bufHandle->buffer[idx].size;
+          totSize -= topicHandle->buffer[idx].size;
           // TODO:memory insufficient
           // return msgs already copied
           break;
         }
       }
-      *((int64_t*)buffer) = bufHandle->topicId;
+      *((int64_t*)buffer) = topicHandle->topicId;
       buffer = POINTER_SHIFT(buffer, sizeof(int64_t));
-      *((int64_t*)buffer) = bufHandle->buffer[idx].size;
+      *((int64_t*)buffer) = topicHandle->buffer[idx].size;
       buffer = POINTER_SHIFT(buffer, sizeof(int64_t));
-      memcpy(buffer, bufHandle->buffer[idx].content, bufHandle->buffer[idx].size);
-      buffer = POINTER_SHIFT(buffer, bufHandle->buffer[idx].size);
+      memcpy(buffer, topicHandle->buffer[idx].content, topicHandle->buffer[idx].size);
+      buffer = POINTER_SHIFT(buffer, topicHandle->buffer[idx].size);
       if (totSize > sizeLimit) {
         break;
       }
@@ -199,11 +196,19 @@ static int tqFetch(STqGroupHandle* gHandle, void** msg) {
   return totSize;
 }
 
-STqGroupHandle* tqGetGroupHandle(STQ* pTq, int64_t cId) { return NULL; }
-
-int tqLaunchQuery(STqGroupHandle* gHandle) { return 0; }
+STqGroup* tqGetGroup(STQ* pTq, int64_t clientId) { return tqHandleGet(pTq->tqMeta, clientId); }
 
-int tqSendLaunchQuery(STqGroupHandle* gHandle) { return 0; }
+int tqSendLaunchQuery(STqMsgItem* bufItem, int64_t offset) {
+  if (tqQueryExecuting(bufItem->status)) {
+    return 0;
+  }
+  bufItem->status = 1;
+  // load data from wal or buffer pool
+  // put into exec
+  // send exec into non blocking queue
+  // when query finished, put into buffer pool
+  return 0;
+}
 
 /*int tqMoveOffsetToNext(TqGroupHandle* gHandle) {*/
 /*return 0;*/
@@ -220,23 +225,96 @@ int tqCommit(STQ* pTq) {
   return 0;
 }
 
-int tqSetCursor(STQ* pTq, void* msg) {
+int tqBufferSetOffset(STqTopic* pTopic, int64_t offset) {
+  int code;
+  memset(pTopic->buffer, 0, sizeof(pTopic->buffer));
+  // launch query
+  for (int i = offset; i < offset + TQ_BUFFER_SIZE; i++) {
+    int pos = i % TQ_BUFFER_SIZE;
+    code = tqSendLaunchQuery(&pTopic->buffer[pos], offset);
+    if (code < 0) {
+      // TODO: error handling
+    }
+  }
+  // set offset
+  pTopic->nextConsumeOffset = offset;
+  pTopic->floatingCursor = offset;
+  return 0;
+}
+
+STqTopic* tqFindTopic(STqGroup* pGroup, int64_t topicId) {
+  // TODO
+  return NULL;
+}
+
+int tqSetCursor(STQ* pTq, STqSetCurReq* pMsg) {
+  int       code;
+  int64_t   clientId = pMsg->head.clientId;
+  int64_t   topicId = pMsg->topicId;
+  int64_t   offset = pMsg->offset;
+  STqGroup* gHandle = tqGetGroup(pTq, clientId);
+  if (gHandle == NULL) {
+    // client not connect
+    return -1;
+  }
+  STqTopic* topicHandle = tqFindTopic(gHandle, topicId);
+  if (topicHandle == NULL) {
+    return -1;
+  }
+  if (pMsg->offset == topicHandle->nextConsumeOffset) {
+    return 0;
+  }
+  // TODO: check log last version
+
+  code = tqBufferSetOffset(topicHandle, offset);
+  if (code < 0) {
+    // set error code
+    return -1;
+  }
+
+  return 0;
+}
+
+int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
+  int64_t   clientId = pMsg->head.clientId;
+  STqGroup* pGroup = tqGetGroup(pTq, clientId);
+  if (pGroup == NULL) {
+    terrno = TSDB_CODE_TQ_GROUP_NOT_SET;
+    return -1;
+  }
+
+  STqConsumeRsp* pRsp = (STqConsumeRsp*)pMsg;
+  int            numOfMsgs = tqFetch(pGroup, (void**)&pRsp->msgs);
+  if (numOfMsgs < 0) {
+    return -1;
+  }
+  if (numOfMsgs == 0) {
+    // most recent data has been fetched
+
+    // enable timer for blocking wait
+    // once new data written during wait time
+    // launch query and response
+  }
+
+  // fetched a num of msgs, rpc response
+
   return 0;
 }
 
+#if 0
 int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
-  if (!tqProtoCheck((TmqMsgHead*)pMsg)) {
+  if (!tqProtoCheck((STqMsgHead*)pMsg)) {
     // proto version invalid
     return -1;
   }
-  int64_t         clientId = pMsg->head.clientId;
-  STqGroupHandle* gHandle = tqGetGroupHandle(pTq, clientId);
-  if (gHandle == NULL) {
+  int64_t   clientId = pMsg->head.clientId;
+  STqGroup* pGroup = tqGetGroup(pTq, clientId);
+  if (pGroup == NULL) {
     // client not connect
     return -1;
   }
   if (pMsg->acks.ackNum != 0) {
-    if (tqAck(gHandle, &pMsg->acks) != 0) {
+    if (tqAck(pGroup, &pMsg->acks) != 0) {
       // ack not success
       return -1;
     }
@@ -244,22 +322,23 @@ int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
 
   STqConsumeRsp* pRsp = (STqConsumeRsp*)pMsg;
 
-  if (tqFetch(gHandle, (void**)&pRsp->msgs) <= 0) {
+  if (tqFetch(pGroup, (void**)&pRsp->msgs) <= 0) {
     // fetch error
     return -1;
   }
 
   // judge and launch new query
-  if (tqLaunchQuery(gHandle)) {
-    // launch query error
-    return -1;
-  }
+  /*if (tqSendLaunchQuery(gHandle)) {*/
+  // launch query error
+  /*return -1;*/
+  /*}*/
   return 0;
 }
+#endif
 
-int tqSerializeGroupHandle(const STqGroupHandle* gHandle, STqSerializedHead** ppHead) {
+int tqSerializeGroup(const STqGroup* pGroup, STqSerializedHead** ppHead) {
   // calculate size
-  int sz = tqGetgHandleSSize(gHandle) + sizeof(STqSerializedHead);
+  int sz = tqGroupSSize(pGroup) + sizeof(STqSerializedHead);
   if (sz > (*ppHead)->ssize) {
     void* tmpPtr = realloc(*ppHead, sz);
     if (tmpPtr == NULL) {
@@ -272,53 +351,53 @@ int tqSerializeGroupHandle(const STqGroupHandle* gHandle, STqSerializedHead** pp
   }
   void* ptr = (*ppHead)->content;
   // do serialization
-  *(int64_t*)ptr = gHandle->cId;
+  *(int64_t*)ptr = pGroup->clientId;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int64_t*)ptr = gHandle->cgId;
+  *(int64_t*)ptr = pGroup->cgId;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int32_t*)ptr = gHandle->topicNum;
+  *(int32_t*)ptr = pGroup->topicNum;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
-  if (gHandle->topicNum > 0) {
-    tqSerializeListHandle(gHandle->head, ptr);
+  if (pGroup->topicNum > 0) {
+    tqSerializeListHandle(pGroup->head, ptr);
   }
   return 0;
 }
 
-void* tqSerializeListHandle(STqListHandle* listHandle, void* ptr) {
-  STqListHandle* node = listHandle;
+void* tqSerializeListHandle(STqList* listHandle, void* ptr) {
+  STqList* node = listHandle;
   ASSERT(node != NULL);
   while (node) {
-    ptr = tqSerializeBufHandle(&node->bufHandle, ptr);
+    ptr = tqSerializeTopic(&node->topic, ptr);
     node = node->next;
   }
   return ptr;
 }
 
-void* tqSerializeBufHandle(STqBufferHandle* bufHandle, void* ptr) {
-  *(int64_t*)ptr = bufHandle->nextConsumeOffset;
+void* tqSerializeTopic(STqTopic* pTopic, void* ptr) {
+  *(int64_t*)ptr = pTopic->nextConsumeOffset;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int64_t*)ptr = bufHandle->topicId;
+  *(int64_t*)ptr = pTopic->topicId;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int32_t*)ptr = bufHandle->head;
+  *(int32_t*)ptr = pTopic->head;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
-  *(int32_t*)ptr = bufHandle->tail;
+  *(int32_t*)ptr = pTopic->tail;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
   for (int i = 0; i < TQ_BUFFER_SIZE; i++) {
-    ptr = tqSerializeBufItem(&bufHandle->buffer[i], ptr);
+    ptr = tqSerializeItem(&pTopic->buffer[i], ptr);
   }
   return ptr;
 }
 
-void* tqSerializeBufItem(STqBufferItem* bufItem, void* ptr) {
+void* tqSerializeItem(STqMsgItem* bufItem, void* ptr) {
   // TODO: do we need serialize this?
   // mainly for executor
   return ptr;
 }
 
-const void* tqDeserializeGroupHandle(const STqSerializedHead* pHead, STqGroupHandle** ppGHandle) {
-  STqGroupHandle* gHandle = *ppGHandle;
-  const void*     ptr = pHead->content;
-  gHandle->cId = *(int64_t*)ptr;
+const void* tqDeserializeGroup(const STqSerializedHead* pHead, STqGroup** ppGroup) {
+  STqGroup*   gHandle = *ppGroup;
+  const void* ptr = pHead->content;
+  gHandle->clientId = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
   gHandle->cgId = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
@@ -326,63 +405,63 @@ const void* tqDeserializeGroupHandle(const STqSerializedHead* pHead, STqGroupHan
   gHandle->topicNum = *(int32_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
   gHandle->head = NULL;
-  STqListHandle* node = gHandle->head;
+  STqList* node = gHandle->head;
   for (int i = 0; i < gHandle->topicNum; i++) {
     if (gHandle->head == NULL) {
-      if ((node = malloc(sizeof(STqListHandle))) == NULL) {
+      if ((node = malloc(sizeof(STqList))) == NULL) {
         // TODO: error
         return NULL;
       }
       node->next = NULL;
-      ptr = tqDeserializeBufHandle(ptr, &node->bufHandle);
+      ptr = tqDeserializeTopic(ptr, &node->topic);
       gHandle->head = node;
     } else {
-      node->next = malloc(sizeof(STqListHandle));
+      node->next = malloc(sizeof(STqList));
       if (node->next == NULL) {
         // TODO: error
         return NULL;
       }
       node->next->next = NULL;
-      ptr = tqDeserializeBufHandle(ptr, &node->next->bufHandle);
+      ptr = tqDeserializeTopic(ptr, &node->next->topic);
       node = node->next;
     }
   }
   return ptr;
 }
 
-const void* tqDeserializeBufHandle(const void* pBytes, STqBufferHandle* bufHandle) {
+const void* tqDeserializeTopic(const void* pBytes, STqTopic* topic) {
   const void* ptr = pBytes;
-  bufHandle->nextConsumeOffset = *(int64_t*)ptr;
+  topic->nextConsumeOffset = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  bufHandle->topicId = *(int64_t*)ptr;
+  topic->topicId = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  bufHandle->head = *(int32_t*)ptr;
+  topic->head = *(int32_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
-  bufHandle->tail = *(int32_t*)ptr;
+  topic->tail = *(int32_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
   for (int i = 0; i < TQ_BUFFER_SIZE; i++) {
-    ptr = tqDeserializeBufItem(ptr, &bufHandle->buffer[i]);
+    ptr = tqDeserializeItem(ptr, &topic->buffer[i]);
   }
   return ptr;
 }
 
-const void* tqDeserializeBufItem(const void* pBytes, STqBufferItem* bufItem) { return pBytes; }
+const void* tqDeserializeItem(const void* pBytes, STqMsgItem* bufItem) { return pBytes; }
 
 // TODO: make this a macro
-int tqGetgHandleSSize(const STqGroupHandle* gHandle) {
+int tqGroupSSize(const STqGroup* gHandle) {
   return sizeof(int64_t) * 2  // cId + cgId
          + sizeof(int32_t)    // topicNum
-         + gHandle->topicNum * tqBufHandleSSize();
+         + gHandle->topicNum * tqTopicSSize();
 }
 
 // TODO: make this a macro
-int tqBufHandleSSize() {
+int tqTopicSSize() {
   return sizeof(int64_t) * 2    // nextConsumeOffset + topicId
          + sizeof(int32_t) * 2  // head + tail
-         + TQ_BUFFER_SIZE * tqBufItemSSize();
+         + TQ_BUFFER_SIZE * tqItemSSize();
 }
 
-int tqBufItemSSize() {
+int tqItemSSize() {
   // TODO: do this need serialization?
   // mainly for executor
   return 0;

source/dnode/vnode/tq/src/tqMetaStore.c

@@ -56,6 +56,7 @@ static inline int tqReadLastPage(int fd, STqIdxPageBuf* pBuf) {
   int offset = tqSeekLastPage(fd);
   int nBytes;
   if ((nBytes = read(fd, pBuf, TQ_PAGE_SIZE)) == -1) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
     return -1;
   }
   if (nBytes == 0) {
@@ -71,7 +72,7 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
                           int32_t tqConfigFlag) {
   STqMetaStore* pMeta = malloc(sizeof(STqMetaStore));
   if (pMeta == NULL) {
-    // close
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return NULL;
   }
   memset(pMeta, 0, sizeof(STqMetaStore));
@@ -79,8 +80,9 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
   // concat data file name and index file name
   size_t pathLen = strlen(path);
   pMeta->dirPath = malloc(pathLen + 1);
-  if (pMeta->dirPath != NULL) {
-    // TODO: memory insufficient
+  if (pMeta->dirPath == NULL) {
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
+    return NULL;
   }
   strcpy(pMeta->dirPath, path);
 
@@ -88,13 +90,14 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
 
   strcpy(name, path);
   if (taosDirExist(name) != 0 && taosMkDir(name) != 0) {
-    ASSERT(false);
+    terrno = TSDB_CODE_TQ_FAILED_TO_CREATE_DIR;
+    tqError("failed to create dir:%s since %s ", name, terrstr());
   }
   strcat(name, "/" TQ_IDX_NAME);
   int idxFd = open(name, O_RDWR | O_CREAT, 0755);
   if (idxFd < 0) {
-    ASSERT(false);
-    // close file
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    tqError("failed to open file:%s since %s ", name, terrstr());
     // free memory
     return NULL;
   }
@@ -102,9 +105,7 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
   pMeta->idxFd = idxFd;
   pMeta->unpersistHead = malloc(sizeof(STqMetaList));
   if (pMeta->unpersistHead == NULL) {
-    ASSERT(false);
-    // close file
-    // free memory
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return NULL;
   }
   memset(pMeta->unpersistHead, 0, sizeof(STqMetaList));
@@ -114,7 +115,8 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
   strcat(name, "/" TQ_META_NAME);
   int fileFd = open(name, O_RDWR | O_CREAT, 0755);
   if (fileFd < 0) {
-    ASSERT(false);
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    tqError("failed to open file:%s since %s", name, terrstr());
     return NULL;
   }
 
@@ -129,7 +131,7 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
   STqIdxPageBuf      idxBuf;
   STqSerializedHead* serializedObj = malloc(TQ_PAGE_SIZE);
   if (serializedObj == NULL) {
-    // TODO:memory insufficient
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
   }
   int  idxRead;
   int  allocated = TQ_PAGE_SIZE;
@@ -137,14 +139,16 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
   while ((idxRead = read(idxFd, &idxBuf, TQ_PAGE_SIZE))) {
     if (idxRead == -1) {
       // TODO: handle error
-      ASSERT(false);
+      terrno = TAOS_SYSTEM_ERROR(errno);
+      tqError("failed to read tq index file since %s", terrstr());
     }
     ASSERT(idxBuf.head.writeOffset == idxRead);
     // loop read every entry
     for (int i = 0; i < idxBuf.head.writeOffset - TQ_IDX_PAGE_HEAD_SIZE; i += TQ_IDX_SIZE) {
       STqMetaList* pNode = malloc(sizeof(STqMetaList));
       if (pNode == NULL) {
-        // TODO: free memory and return error
+        terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
+        // TODO: free memory
       }
       memset(pNode, 0, sizeof(STqMetaList));
       memcpy(&pNode->handle, &idxBuf.buffer[i], TQ_IDX_SIZE);
@@ -153,7 +157,8 @@ STqMetaStore* tqStoreOpen(const char* path, FTqSerialize serializer, FTqDeserial
       if (allocated < pNode->handle.serializedSize) {
         void* ptr = realloc(serializedObj, pNode->handle.serializedSize);
         if (ptr == NULL) {
-          // TODO: memory insufficient
+          terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
+          // TODO: free memory
         }
         serializedObj = ptr;
         allocated = pNode->handle.serializedSize;
@@ -292,7 +297,7 @@ int32_t tqStorePersist(STqMetaStore* pMeta) {
   STqMetaList*       pNode = pHead->unpersistNext;
   STqSerializedHead* pSHead = malloc(sizeof(STqSerializedHead));
   if (pSHead == NULL) {
-    // TODO: memory error
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return -1;
   }
   pSHead->ver = TQ_SVER;
@@ -403,7 +408,6 @@ static int32_t tqHandlePutCommitted(STqMetaStore* pMeta, int64_t key, void* valu
   STqMetaList* pNode = pMeta->bucket[bucketKey];
   while (pNode) {
     if (pNode->handle.key == key) {
-      // TODO: think about thread safety
       if (pNode->handle.valueInUse && pNode->handle.valueInUse != TQ_DELETE_TOKEN) {
         pMeta->pDeleter(pNode->handle.valueInUse);
       }
@@ -416,7 +420,7 @@ static int32_t tqHandlePutCommitted(STqMetaStore* pMeta, int64_t key, void* valu
   }
   STqMetaList* pNewNode = malloc(sizeof(STqMetaList));
   if (pNewNode == NULL) {
-    // TODO: memory error
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return -1;
   }
   memset(pNewNode, 0, sizeof(STqMetaList));
@@ -470,10 +474,10 @@ static inline int32_t tqHandlePutImpl(STqMetaStore* pMeta, int64_t key, void* va
   STqMetaList* pNode = pMeta->bucket[bucketKey];
   while (pNode) {
     if (pNode->handle.key == key) {
-      // TODO: think about thread safety
       if (pNode->handle.valueInTxn) {
-        if (TqDupIntxnReject(pMeta->tqConfigFlag)) {
-          return -2;
+        if (tqDupIntxnReject(pMeta->tqConfigFlag)) {
+          terrno = TSDB_CODE_TQ_META_KEY_DUP_IN_TXN;
+          return -1;
         }
         if (pNode->handle.valueInTxn != TQ_DELETE_TOKEN) {
           pMeta->pDeleter(pNode->handle.valueInTxn);
@@ -488,7 +492,7 @@ static inline int32_t tqHandlePutImpl(STqMetaStore* pMeta, int64_t key, void* va
   }
   STqMetaList* pNewNode = malloc(sizeof(STqMetaList));
   if (pNewNode == NULL) {
-    // TODO: memory error
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return -1;
   }
   memset(pNewNode, 0, sizeof(STqMetaList));
@@ -505,7 +509,7 @@ int32_t tqHandleMovePut(STqMetaStore* pMeta, int64_t key, void* value) { return
 int32_t tqHandleCopyPut(STqMetaStore* pMeta, int64_t key, void* value, size_t vsize) {
   void* vmem = malloc(vsize);
   if (vmem == NULL) {
-    // TODO: memory error
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return -1;
   }
   memcpy(vmem, value, vsize);
@@ -535,6 +539,7 @@ int32_t tqHandleCommit(STqMetaStore* pMeta, int64_t key) {
   while (pNode) {
     if (pNode->handle.key == key) {
       if (pNode->handle.valueInTxn == NULL) {
+        terrno = TSDB_CODE_TQ_META_KEY_NOT_IN_TXN;
         return -1;
       }
       if (pNode->handle.valueInUse && pNode->handle.valueInUse != TQ_DELETE_TOKEN) {
@@ -548,7 +553,8 @@ int32_t tqHandleCommit(STqMetaStore* pMeta, int64_t key) {
       pNode = pNode->next;
     }
   }
-  return -2;
+  terrno = TSDB_CODE_TQ_META_NO_SUCH_KEY;
+  return -1;
 }
 
 int32_t tqHandleAbort(STqMetaStore* pMeta, int64_t key) {
@@ -564,12 +570,14 @@ int32_t tqHandleAbort(STqMetaStore* pMeta, int64_t key) {
         tqLinkUnpersist(pMeta, pNode);
         return 0;
       }
+      terrno = TSDB_CODE_TQ_META_KEY_NOT_IN_TXN;
       return -1;
     } else {
       pNode = pNode->next;
     }
   }
-  return -2;
+  terrno = TSDB_CODE_TQ_META_NO_SUCH_KEY;
+  return -1;
 }
 
 int32_t tqHandleDel(STqMetaStore* pMeta, int64_t key) {
@@ -588,7 +596,7 @@ int32_t tqHandleDel(STqMetaStore* pMeta, int64_t key) {
       pNode = pNode->next;
     }
   }
-  // no such key
+  terrno = TSDB_CODE_TQ_META_NO_SUCH_KEY;
   return -1;
 }
 

source/dnode/vnode/tsdb/CMakeLists.txt

 aux_source_directory(src TSDB_SRC)
-add_library(tsdb ${TSDB_SRC})
+if(0)
+    add_library(tsdb ${TSDB_SRC})
+else(0)
+    add_library(tsdb "")
+    target_sources(tsdb
+        PRIVATE
+        "src/tsdbCommit.c"
+        "src/tsdbMain.c"
+        "src/tsdbMemTable.c"
+        "src/tsdbOptions.c"
+        "src/tsdbWrite.c"
+    )
+endif(0)
+
 target_include_directories(
     tsdb
     PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/vnode/tsdb"
     PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/inc"
 )
+
 target_link_libraries(
     tsdb
     PUBLIC os

src/tsdb/inc/tsdbCommit.h → source/dnode/vnode/tsdb/inc/tsdbCommit.h

@@ -16,6 +16,7 @@
 #ifndef _TD_TSDB_COMMIT_H_
 #define _TD_TSDB_COMMIT_H_
 
+#if 0
 typedef struct {
   int   minFid;
   int   midFid;
@@ -53,5 +54,6 @@ static FORCE_INLINE int tsdbGetFidLevel(int fid, SRtn *pRtn) {
     return -1;
   }
 }
+#endif
 
 #endif /* _TD_TSDB_COMMIT_H_ */
\ No newline at end of file

src/tsdb/inc/tsdbCompact.h → source/dnode/vnode/tsdb/inc/tsdbCompact.h

@@ -19,8 +19,12 @@
 extern "C" {
 #endif
 
+#if 0
+
 void *tsdbCompactImpl(STsdbRepo *pRepo);
 
+#endif
+
 #ifdef __cplusplus
 }
 #endif

src/tsdb/inc/tsdbFS.h → source/dnode/vnode/tsdb/inc/tsdbFS.h

@@ -16,6 +16,8 @@
 #ifndef _TD_TSDB_FS_H_
 #define _TD_TSDB_FS_H_
 
+#if 0
+
 #define TSDB_FS_VERSION 0
 
 // ================== TSDB global config
@@ -113,4 +115,6 @@ static FORCE_INLINE int tsdbUnLockFS(STsdbFS* pFs) {
   return 0;
 }
 
+#endif
+
 #endif /* _TD_TSDB_FS_H_ */