Merge branch '3.0' into feature/tq

CMakeLists.txt

@@ -50,6 +50,11 @@ if(${BUILD_WITH_LUCENE})
     cat("${CMAKE_SUPPORT_DIR}/lucene_CMakeLists.txt.in" ${DEPS_TMP_FILE})
 endif(${BUILD_WITH_LUCENE})
 
+## NuRaft
+if(${BUILD_WITH_NURAFT})
+    cat("${CMAKE_SUPPORT_DIR}/nuraft_CMakeLists.txt.in" ${DEPS_TMP_FILE})
+endif(${BUILD_WITH_NURAFT})
+
 ## download dependencies
 configure_file(${DEPS_TMP_FILE} "${CMAKE_SOURCE_DIR}/deps/deps-download/CMakeLists.txt")
 execute_process(COMMAND "${CMAKE_COMMAND}" -G "${CMAKE_GENERATOR}" .
@@ -69,4 +74,7 @@ target_include_directories(api INTERFACE "include/client")
 # src
 add_subdirectory(source)
 
+# docs 
+add_subdirectory(docs)
+
 # tests (TODO)

Jenkinsfile

@@ -142,7 +142,7 @@ pipeline {
           }
       }
   //     stage('Parallel test stage') {
-  //       //only build pr
+  //     skip defaultCheckout 
   //       options { skipDefaultCheckout() } 
   //       when {
   //         allOf{

cmake/cmake.options

@@ -16,7 +16,7 @@ option(
 option(
     BUILD_WITH_ROCKSDB 
     "If build with rocksdb" 
-    OFF
+    ON
 )
 
 option(
@@ -25,8 +25,20 @@ option(
     OFF
 )
 
+option(
+    BUILD_WITH_NURAFT
+    "If build with NuRaft" 
+    OFF
+)
+
 option(
     BUILD_DEPENDENCY_TESTS 
     "If build dependency tests" 
     OFF
+)
+
+option(
+    BUILD_DOCS
+    "If use doxygen build documents" 
+    ON
 )

cmake/nuraft_CMakeLists.txt.in

@@ -0,0 +1,12 @@
+
+# NuRaft
+ExternalProject_Add(NuRaft
+        GIT_REPOSITORY https://github.com/eBay/NuRaft.git
+        GIT_TAG v1.3.0
+        SOURCE_DIR "${CMAKE_SOURCE_DIR}/deps/nuraft"
+        BINARY_DIR "${CMAKE_SOURCE_DIR}/deps/nuraft"
+        CONFIGURE_COMMAND "./prepare.sh"
+        BUILD_COMMAND ""
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+        )

deps/CMakeLists.txt

@@ -67,6 +67,12 @@ if(${BUILD_WITH_LUCENE})
     add_subdirectory(lucene)
 endif(${BUILD_WITH_LUCENE})
 
+# NuRaft
+if(${BUILD_WITH_NURAFT})
+    add_subdirectory(nuraft)
+endif(${BUILD_WITH_NURAFT})
+
+
 # ================================================================================================
 # DEPENDENCY TEST
 # ================================================================================================

docs/CMakeLists.txt

@@ -0,0 +1,23 @@
+# Generate API documentation
+## https://vicrucann.github.io/tutorials/quick-cmake-doxygen/
+if(${BUILD_DOCS})
+    find_package(Doxygen)
+    if (DOXYGEN_FOUND)
+        # Build the doc
+        set(DOXYGEN_IN ${CMAKE_SOURCE_DIR}/docs/Doxyfile.in)
+        set(DOXYGEN_OUT ${CMAKE_BINARY_DIR}/Doxyfile)
+
+        configure_file(${DOXYGEN_IN} ${DOXYGEN_OUT} @ONLY)
+        message("Doxygen build start")
+
+        add_custom_target(
+            tdengine_doxygen ALL
+            COMMAND ${DOXYGEN_EXECUTABLE} ${DOXYGEN_OUT}
+            WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}
+            COMMENT "Generating API doxumentation with Doxygen"
+            VERBATIM
+        )
+    else(DOXYGEN_FOUND)
+        message("Doxygen need to be installed to generate the doxygen documentation")
+    endif(DOXYGEN_FOUND)
+endif(${BUILD_DOCS})

include/common/common.h

@@ -19,7 +19,7 @@
 #include "taosdef.h"
 #include "taosmsg.h"
 #include "tarray.h"
-
+#include "tvariant.h"
 //typedef struct STimeWindow {
 //  TSKEY skey;
 //  TSKEY ekey;
@@ -66,4 +66,59 @@ typedef struct SColumnInfoData {
   char       *pData;    // the corresponding block data in memory
 } SColumnInfoData;
 
+//======================================================================================================================
+// the following structure shared by parser and executor
+typedef struct SColumn {
+  uint64_t     uid;
+  char         name[TSDB_COL_NAME_LEN];
+  int8_t       flag;      // column type: normal column, tag, or user-input column (integer/float/string)
+  SColumnInfo  info;
+} SColumn;
+
+typedef struct SLimit {
+  int64_t   limit;
+  int64_t   offset;
+} SLimit;
+
+typedef struct SOrder {
+  uint32_t  order;
+  int32_t   orderColId;
+} SOrder;
+
+typedef struct SGroupbyExpr {
+  SArray*  columnInfo;  // SArray<SColIndex>, group by columns information
+  bool     groupbyTag;  // group by tag or column
+} SGroupbyExpr;
+
+// the structure for sql function in select clause
+typedef struct SSqlExpr {
+  char      token[TSDB_COL_NAME_LEN];      // original token
+  SSchema   resSchema;
+
+  int32_t   numOfCols;
+  SColumn*  pColumns;       // data columns that are required by query
+  int32_t   interBytes;     // inter result buffer size
+  int16_t   numOfParams;    // argument value of each function
+  SVariant  param[3];       // parameters are not more than 3
+} SSqlExpr;
+
+typedef struct SExprInfo {
+  struct SSqlExpr    base;
+  struct tExprNode  *pExpr;
+} SExprInfo;
+
+typedef struct SStateWindow {
+  SColumn col;
+} SStateWindow;
+
+typedef struct SSessionWindow {
+  int64_t gap;             // gap between two session window(in microseconds)
+  SColumn col;
+} SSessionWindow;
+
+#define QUERY_ASC_FORWARD_STEP   1
+#define QUERY_DESC_FORWARD_STEP -1
+
+#define GET_FORWARD_DIRECTION_FACTOR(ord) (((ord) == TSDB_ORDER_ASC) ? QUERY_ASC_FORWARD_STEP : QUERY_DESC_FORWARD_STEP)
+
 #endif  // TDENGINE_COMMON_H

include/common/taosmsg.h

@@ -204,14 +204,14 @@ enum _mgmt_table {
 #define TSDB_COL_NORMAL             0x0u    // the normal column of the table
 #define TSDB_COL_TAG                0x1u    // the tag column type
 #define TSDB_COL_UDC                0x2u    // the user specified normal string column, it is a dummy column
-#define TSDB_COL_NULL               0x4u    // the column filter NULL or not
+#define TSDB_COL_TMP                0x4u    // internal column generated by the previous operators
+#define TSDB_COL_NULL               0x8u    // the column filter NULL or not
 
 #define TSDB_COL_IS_TAG(f)          (((f&(~(TSDB_COL_NULL)))&TSDB_COL_TAG) != 0)
 #define TSDB_COL_IS_NORMAL_COL(f)   ((f&(~(TSDB_COL_NULL))) == TSDB_COL_NORMAL)
 #define TSDB_COL_IS_UD_COL(f)       ((f&(~(TSDB_COL_NULL))) == TSDB_COL_UDC)
 #define TSDB_COL_REQ_NULL(f)        (((f)&TSDB_COL_NULL) != 0)
 
-
 extern char *taosMsg[];
 
 #pragma pack(push, 1)
@@ -399,8 +399,6 @@ typedef struct {
 typedef struct {
   char   user[TSDB_USER_LEN];
   char   pass[TSDB_KEY_LEN];
-  int8_t privilege;
-  int8_t flag;
 } SCreateUserMsg, SAlterUserMsg;
 
 typedef struct {
@@ -491,11 +489,6 @@ typedef struct SInterval {
   int64_t offset;
 } SInterval;
 
-typedef struct SSessionWindow {
-  int64_t gap;             // gap between two session window(in microseconds)
-  int32_t primaryColId;    // primary timestamp column
-} SSessionWindow;
-
 typedef struct {
   SMsgHead    head;
   char        version[TSDB_VERSION_LEN];
@@ -520,7 +513,7 @@ typedef struct {
   int16_t     orderColId;
   int16_t     numOfCols;        // the number of columns will be load from vnode
   SInterval   interval;
-  SSessionWindow sw;            // session window
+//  SSessionWindow sw;            // session window
   uint16_t    tagCondLen;       // tag length in current query
   uint16_t    colCondLen;       // column length in current query
   int16_t     numOfGroupCols;   // num of group by columns

include/common/tname.h

@@ -16,6 +16,8 @@
 #ifndef TDENGINE_TNAME_H
 #define TDENGINE_TNAME_H
 
+#include "taosmsg.h"
+
 #define TSDB_DB_NAME_T     1
 #define TSDB_TABLE_NAME_T  2
 
@@ -52,6 +54,8 @@ int32_t tNameFromString(SName* dst, const char* str, uint32_t type);
 
 int32_t tNameSetAcctId(SName* dst, const char* acct);
 
+SSchema* tGetTbnameColumnSchema();
+
 #if 0
 int32_t tNameSetDbName(SName* dst, const char* acct, SToken* dbToken);
 #endif

include/common/trequest.h

@@ -0,0 +1,47 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_TREQUEST_H_
+#define _TD_TREQUEST_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* ------------------------ TYPES EXPOSED ------------------------ */
+typedef struct SRequest      SRequest;
+typedef struct SReqBatch     SReqBatch;
+typedef struct SReqBatchIter SReqBatchIter;
+
+// SRequest
+
+// SReqBatch
+
+// SReqBatchIter
+void            tdInitRBIter(SReqBatchIter *pIter, SReqBatch *pReqBatch);
+const SRequest *tdRBIterNext(SReqBatchIter *pIter);
+void            tdClearRBIter(SReqBatchIter *pIter);
+
+/* ------------------------ TYPES DEFINITION ------------------------ */
+struct SReqBatchIter {
+  int        iReq;
+  SReqBatch *pReqBatch;
+};
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_TREQUEST_H_*/

include/common/ttszip.h

@@ -21,7 +21,7 @@ extern "C" {
 #endif
 
 #include "os.h"
-#include "taosdef.h"
+#include "tdef.h"
 #include "tvariant.h"
 
 #define MEM_BUF_SIZE (1 << 20)

include/dnode/mnode/mnode.h

@@ -55,17 +55,16 @@ typedef struct {
 int32_t mnodeInit(SMnodePara para);
 void    mnodeCleanup();
 
-int32_t mnodeDeploy(char *path, SMnodeCfg *pCfg);
+int32_t mnodeDeploy(SMnodeCfg *pCfg);
-void    mnodeUnDeploy(char *path);
+void    mnodeUnDeploy();
-int32_t mnodeStart(char *path, SMnodeCfg *pCfg);
+int32_t mnodeStart(SMnodeCfg *pCfg);
 int32_t mnodeAlter(SMnodeCfg *pCfg);
 void    mnodeStop();
 
 int32_t mnodeGetLoad(SMnodeLoad *pLoad);
 int32_t mnodeRetriveAuth(char *user, char *spi, char *encrypt, char *secret, char *ckey);
 
-SMnodeMsg *mnodeInitMsg(int32_t msgNum);
+SMnodeMsg *mnodeInitMsg(SRpcMsg *pRpcMsg);
-int32_t    mnodeAppendMsg(SMnodeMsg *pMsg, SRpcMsg *pRpcMsg);
 void       mnodeCleanupMsg(SMnodeMsg *pMsg);
 void       mnodeProcessMsg(SMnodeMsg *pMsg, EMnMsgType msgType);
 

include/dnode/mnode/sdb/sdb.h

@@ -0,0 +1,187 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_SDB_H_
+#define _TD_SDB_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define SDB_GET_INT64(pData, pRow, dataPos, val)   \
+  {                                                \
+    if (sdbGetRawInt64(pRaw, dataPos, val) != 0) { \
+      sdbFreeRow(pRow);                            \
+      return NULL;                                 \
+    }                                              \
+    dataPos += sizeof(int64_t);                    \
+  }
+
+#define SDB_GET_INT32(pData, pRow, dataPos, val)   \
+  {                                                \
+    if (sdbGetRawInt32(pRaw, dataPos, val) != 0) { \
+      sdbFreeRow(pRow);                            \
+      return NULL;                                 \
+    }                                              \
+    dataPos += sizeof(int32_t);                    \
+  }
+
+#define SDB_GET_INT8(pData, pRow, dataPos, val)   \
+  {                                               \
+    if (sdbGetRawInt8(pRaw, dataPos, val) != 0) { \
+      sdbFreeRow(pRow);                           \
+      return NULL;                                \
+    }                                             \
+    dataPos += sizeof(int8_t);                    \
+  }
+
+#define SDB_GET_BINARY(pRaw, pRow, dataPos, val, valLen)    \
+  {                                                         \
+    if (sdbGetRawBinary(pRaw, dataPos, val, valLen) != 0) { \
+      sdbFreeRow(pRow);                                     \
+      return NULL;                                          \
+    }                                                       \
+    dataPos += valLen;                                      \
+  }
+
+#define SDB_SET_INT64(pData, dataPos, val)         \
+  {                                                \
+    if (sdbSetRawInt64(pRaw, dataPos, val) != 0) { \
+      sdbFreeRaw(pRaw);                               \
+      return NULL;                                 \
+    };                                             \
+    dataPos += sizeof(int64_t);                    \
+  }
+
+#define SDB_SET_INT32(pData, dataPos, val)         \
+  {                                                \
+    if (sdbSetRawInt32(pRaw, dataPos, val) != 0) { \
+      sdbFreeRaw(pRaw);                               \
+      return NULL;                                 \
+    };                                             \
+    dataPos += sizeof(int32_t);                    \
+  }
+
+#define SDB_SET_INT8(pData, dataPos, val)         \
+  {                                               \
+    if (sdbSetRawInt8(pRaw, dataPos, val) != 0) { \
+      sdbFreeRaw(pRaw);                              \
+      return NULL;                                \
+    };                                            \
+    dataPos += sizeof(int8_t);                    \
+  }
+
+#define SDB_SET_BINARY(pRaw, dataPos, val, valLen)          \
+  {                                                         \
+    if (sdbSetRawBinary(pRaw, dataPos, val, valLen) != 0) { \
+      sdbFreeRaw(pRaw);                                        \
+      return NULL;                                          \
+    };                                                      \
+    dataPos += valLen;                                      \
+  }
+
+#define SDB_SET_DATALEN(pRaw, dataLen)          \
+  {                                             \
+    if (sdbSetRawDataLen(pRaw, dataLen) != 0) { \
+      sdbFreeRaw(pRaw);                         \
+      return NULL;                              \
+    };                                          \
+  }
+
+typedef struct SSdbRaw SSdbRaw;
+typedef struct SSdbRow SSdbRow;
+typedef enum { SDB_KEY_BINARY = 1, SDB_KEY_INT32 = 2, SDB_KEY_INT64 = 3 } EKeyType;
+typedef enum {
+  SDB_STATUS_CREATING = 1,
+  SDB_STATUS_READY = 2,
+  SDB_STATUS_DROPPING = 3,
+  SDB_STATUS_DROPPED = 4
+} ESdbStatus;
+
+typedef enum {
+  SDB_START = 0,
+  SDB_TRANS = 1,
+  SDB_CLUSTER = 2,
+  SDB_DNODE = 3,
+  SDB_MNODE = 4,
+  SDB_USER = 5,
+  SDB_AUTH = 6,
+  SDB_ACCT = 7,
+  SDB_DB = 8,
+  SDB_VGROUP = 9,
+  SDB_STABLE = 10,
+  SDB_FUNC = 11,
+  SDB_MAX = 12
+} ESdbType;
+
+typedef int32_t (*SdbInsertFp)(void *pObj);
+typedef int32_t (*SdbUpdateFp)(void *pSrcObj, void *pDstObj);
+typedef int32_t (*SdbDeleteFp)(void *pObj);
+typedef int32_t (*SdbDeployFp)();
+typedef SSdbRow *(*SdbDecodeFp)(SSdbRaw *pRaw);
+typedef SSdbRaw *(*SdbEncodeFp)(void *pObj);
+
+typedef struct {
+  ESdbType    sdbType;
+  EKeyType    keyType;
+  SdbDeployFp deployFp;
+  SdbEncodeFp encodeFp;
+  SdbDecodeFp decodeFp;
+  SdbInsertFp insertFp;
+  SdbUpdateFp updateFp;
+  SdbDeleteFp deleteFp;
+} SSdbTable;
+
+int32_t sdbInit();
+void    sdbCleanup();
+void    sdbSetTable(SSdbTable table);
+
+int32_t sdbOpen();
+void    sdbClose();
+int32_t sdbWrite(SSdbRaw *pRaw);
+
+int32_t sdbDeploy();
+void    sdbUnDeploy();
+
+void   *sdbAcquire(ESdbType sdb, void *pKey);
+void    sdbRelease(void *pObj);
+void   *sdbFetch(ESdbType sdb, void *pIter, void **ppObj);
+void    sdbCancelFetch(void *pIter);
+int32_t sdbGetSize(ESdbType sdb);
+
+SSdbRaw *sdbAllocRaw(ESdbType sdb, int8_t sver, int32_t dataLen);
+void     sdbFreeRaw(SSdbRaw *pRaw);
+int32_t  sdbSetRawInt8(SSdbRaw *pRaw, int32_t dataPos, int8_t val);
+int32_t  sdbSetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t val);
+int32_t  sdbSetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t val);
+int32_t  sdbSetRawBinary(SSdbRaw *pRaw, int32_t dataPos, const char *pVal, int32_t valLen);
+int32_t  sdbSetRawDataLen(SSdbRaw *pRaw, int32_t dataLen);
+int32_t  sdbSetRawStatus(SSdbRaw *pRaw, ESdbStatus status);
+int32_t  sdbGetRawInt8(SSdbRaw *pRaw, int32_t dataPos, int8_t *val);
+int32_t  sdbGetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t *val);
+int32_t  sdbGetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t *val);
+int32_t  sdbGetRawBinary(SSdbRaw *pRaw, int32_t dataPos, char *pVal, int32_t valLen);
+int32_t  sdbGetRawSoftVer(SSdbRaw *pRaw, int8_t *sver);
+int32_t  sdbGetRawTotalSize(SSdbRaw *pRaw);
+
+SSdbRow *sdbAllocRow(int32_t objSize);
+void     sdbFreeRow(SSdbRow *pRow);
+void    *sdbGetRowObj(SSdbRow *pRow);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_SDB_H_*/

include/dnode/mnode/transaction/trn.h

@@ -0,0 +1,49 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_TRANSACTION_H_
+#define _TD_TRANSACTION_H_
+
+#include "sdb.h"
+#include "taosmsg.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct STrans STrans;
+typedef enum { TRN_POLICY_ROLLBACK = 1, TRN_POLICY_RETRY = 2 } ETrnPolicy;
+
+int32_t trnInit();
+void    trnCleanup();
+
+STrans *trnCreate(ETrnPolicy);
+void    trnDrop(STrans *pTrans);
+void    trnSetRpcHandle(STrans *pTrans, void *rpcHandle);
+int32_t trnAppendRedoLog(STrans *pTrans, SSdbRaw *pRaw);
+int32_t trnAppendUndoLog(STrans *pTrans, SSdbRaw *pRaw);
+int32_t trnAppendCommitLog(STrans *pTrans, SSdbRaw *pRaw);
+int32_t trnAppendRedoAction(STrans *pTrans, SEpSet *, void *pMsg);
+int32_t trnAppendUndoAction(STrans *pTrans, SEpSet *, void *pMsg);
+
+int32_t trnPrepare(STrans *pTrans, int32_t (*syncfp)(SSdbRaw *pRaw, void *pData));
+int32_t trnApply(SSdbRaw *pRaw, void *pData, int32_t code);
+int32_t trnExecute(int32_t tranId);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_TRANSACTION_H_*/

include/dnode/vnode/meta/impl/metaImpl.h

@@ -0,0 +1,75 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_META_IMPL_H_
+#define _TD_META_IMPL_H_
+
+#include "os.h"
+
+#include "taosmsg.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+typedef uint64_t tb_uid_t;
+
+/* ------------------------ SMetaOptions ------------------------ */
+struct SMetaOptions {
+  size_t lruCacheSize;  // LRU cache size
+};
+
+/* ------------------------ STbOptions ------------------------ */
+#define META_NORMAL_TABLE ((uint8_t)1)
+#define META_SUPER_TABLE ((uint8_t)2)
+#define META_CHILD_TABLE ((uint8_t)3)
+
+typedef struct {
+} SSMAOptions;
+
+// super table options
+typedef struct {
+  tb_uid_t  uid;
+  STSchema* pSchema;
+  STSchema* pTagSchema;
+} SSTbOptions;
+
+// child table options
+typedef struct {
+  tb_uid_t suid;
+  SKVRow   tags;
+} SCTbOptions;
+
+// normal table options
+typedef struct {
+  STSchema* pSchame;
+} SNTbOptions;
+
+struct STbOptions {
+  uint8_t     type;
+  char*       name;
+  uint32_t    ttl;   // time to live in (SECONDS)
+  SSMAOptions bsma;  // Block-wise sma
+  union {
+    SSTbOptions stbOptions;
+    SNTbOptions ntbOptions;
+    SCTbOptions ctbOptions;
+  };
+};
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_META_IMPL_H_*/

include/dnode/vnode/meta/meta.h

@@ -0,0 +1,54 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_META_H_
+#define _TD_META_H_
+
+#include "impl/metaImpl.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Types exported
+typedef struct SMeta        SMeta;
+typedef struct SMetaOptions SMetaOptions;
+typedef struct STbOptions   STbOptions;
+
+// SMeta operations
+SMeta *metaOpen(const char *path, const SMetaOptions *);
+void   metaClose(SMeta *);
+void   metaRemove(const char *path);
+int    metaCreateTable(SMeta *pMeta, const STbOptions *);
+int    metaDropTable(SMeta *pMeta, tb_uid_t uid);
+int    metaCommit(SMeta *);
+
+// Options
+void metaOptionsInit(SMetaOptions *);
+void metaOptionsClear(SMetaOptions *);
+
+// STableOpts
+#define META_TABLE_OPTS_DECLARE(name) STableOpts name = {0}
+void metaNormalTableOptsInit(STbOptions *, const char *name, const STSchema *pSchema);
+void metaSuperTableOptsInit(STbOptions *, const char *name, tb_uid_t uid, const STSchema *pSchema,
+                            const STSchema *pTagSchema);
+void metaChildTableOptsInit(STbOptions *, const char *name, tb_uid_t suid, const SKVRow tags);
+void metaTableOptsClear(STbOptions *);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_META_H_*/

include/dnode/vnode/tsdb/impl/tsdbImpl.h

@@ -23,6 +23,7 @@ extern "C" {
 #endif
 
 struct STsdbOptions {
+  size_t lruCacheSize;
   /* TODO */
 };
 

include/dnode/vnode/vnode.h

@@ -17,14 +17,80 @@
 #define _TD_VNODE_H_
 
 #include "os.h"
-#include "taosmsg.h"
+#include "trequest.h"
-#include "trpc.h"
+
+#include "meta.h"
+#include "tq.h"
+#include "tsdb.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef struct SVnode SVnode;
+/* ------------------------ TYPES EXPOSED ------------------------ */
+typedef struct SVnode        SVnode;
+typedef struct SVnodeOptions SVnodeOptions;
+
+/* ------------------------ SVnode ------------------------ */
+SVnode *vnodeOpen(const char *path, const SVnodeOptions *pVnodeOptions);
+void    vnodeClose(SVnode *pVnode);
+void    vnodeDestroy(const char *path);
+int     vnodeProcessWriteReqs(SVnode *pVnode, SReqBatch *pReqBatch);
+int     vnodeApplyWriteRequest(SVnode *pVnode, const SRequest *pRequest);
+int     vnodeProcessReadReq(SVnode *pVnode, SRequest *pReq);
+int     vnodeProcessSyncReq(SVnode *pVnode, SRequest *pReq);
+
+/* ------------------------ SVnodeOptions ------------------------ */
+void vnodeOptionsInit(SVnodeOptions *);
+void vnodeOptionsClear(SVnodeOptions *);
+
+/* ------------------------ STRUCT DEFINITIONS ------------------------ */
+struct SVnodeOptions {
+  /**
+   * @brief write buffer size in BYTES
+   *
+   */
+  uint64_t wsize;
+
+  /**
+   * @brief time to live of tables in this vnode
+   * in SECONDS
+   *
+   */
+  uint32_t ttl;
+
+  /**
+   * @brief if time-series requests eventual consistency
+   *
+   */
+  bool isWeak;
+
+  /**
+   * @brief if the allocator is heap allcator or arena allocator
+   *
+   */
+  bool isHeapAllocator;
+
+  /**
+   * @brief TSDB options
+   *
+   */
+  STsdbOptions tsdbOptions;
+
+  /**
+   * @brief META options
+   *
+   */
+  SMetaOptions metaOptions;
+  // STqOptions   tqOptions; // TODO
+};
+
+/* ------------------------ FOR COMPILE ------------------------ */
+
+#if 1
+
+#include "taosmsg.h"
+#include "trpc.h"
 
 typedef struct {
   char     db[TSDB_FULL_DB_NAME_LEN];
@@ -71,8 +137,6 @@ typedef struct {
 int32_t vnodeInit(SVnodePara);
 void    vnodeCleanup();
 
-SVnode *vnodeOpen(int32_t vgId, const char *path);
-void    vnodeClose(SVnode *pVnode);
 int32_t vnodeAlter(SVnode *pVnode, const SVnodeCfg *pCfg);
 SVnode *vnodeCreate(int32_t vgId, const char *path, const SVnodeCfg *pCfg);
 void    vnodeDrop(SVnode *pVnode);
@@ -86,6 +150,8 @@ 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/function/function.h

@@ -26,8 +26,8 @@ extern "C" {
 
 #define MAX_INTERVAL_TIME_WINDOW 1000000  // maximum allowed time windows in final results
 
-#define FUNCTION_SCALAR       1
+#define FUNCTION_TYPE_SCALAR       1
-#define FUNCTION_AGG          2
+#define FUNCTION_TYPE_AGG          2
 
 #define TOP_BOTTOM_QUERY_LIMIT    100
 #define FUNCTIONS_NAME_MAX_LENGTH 16
@@ -78,13 +78,30 @@ extern "C" {
 #define FUNCTION_MODE         36
 #define FUNCTION_SAMPLE       37
 
+#define FUNCTION_COV          38
+
+// determine the real data need to calculated the result
+enum {
+  BLK_DATA_NO_NEEDED     = 0x0,
+  BLK_DATA_STATIS_NEEDED = 0x1,
+  BLK_DATA_ALL_NEEDED    = 0x3,
+  BLK_DATA_DISCARD       = 0x4,   // discard current data block since it is not qualified for filter
+};
+
+enum {
+  MASTER_SCAN   = 0x0u,
+  REVERSE_SCAN  = 0x1u,
+  REPEAT_SCAN   = 0x2u,  //repeat scan belongs to the master scan
+  MERGE_STAGE   = 0x20u,
+};
+
 typedef struct SPoint1 {
   int64_t   key;
   union{double  val; char* ptr;};
 } SPoint1;
 
 struct SQLFunctionCtx;
-struct SResultRowCellInfo;
+struct SResultRowEntryInfo;
 
 //for selectivity query, the corresponding tag value is assigned if the data is qualified
 typedef struct SExtTagsInfo {
@@ -93,6 +110,23 @@ typedef struct SExtTagsInfo {
   struct SQLFunctionCtx **pTagCtxList;
 } SExtTagsInfo;
 
+typedef struct SResultDataInfo {
+  int16_t type;
+  int16_t bytes;
+  int32_t intermediateBytes;
+} SResultDataInfo;
+
+#define GET_RES_INFO(ctx) ((ctx)->resultInfo)
+
+typedef struct SFunctionFpSet {
+  bool (*init)(struct SQLFunctionCtx *pCtx, struct SResultRowEntryInfo* pResultCellInfo);  // setup the execute environment
+  void (*addInput)(struct SQLFunctionCtx *pCtx);
+
+  // finalizer must be called after all exec has been executed to generated final result.
+  void (*finalize)(struct SQLFunctionCtx *pCtx);
+  void (*combine)(struct SQLFunctionCtx *pCtx);
+} SFunctionFpSet;
+
 // sql function runtime context
 typedef struct SQLFunctionCtx {
   int32_t      size;      // number of rows
@@ -101,9 +135,7 @@ typedef struct SQLFunctionCtx {
   int16_t      inputType;
   int16_t      inputBytes;
 
-  int16_t      outputType;
+  SResultDataInfo resDataInfo;
-  int16_t      outputBytes;   // size of results, determined by function and input column data type
-  int32_t      interBufBytes; // internal buffer size
   bool         hasNull;       // null value exist in current block
   bool         requireNull;   // require null in some function
   bool         stableQuery;
@@ -117,18 +149,21 @@ typedef struct SQLFunctionCtx {
   void        *ptsOutputBuf;  // corresponding output buffer for timestamp of each result, e.g., top/bottom*/
   SVariant     tag;
 
-  bool         isSmaSet;
+  bool        isAggSet;
-  SColumnDataAgg sma;
+  SColumnDataAgg agg;
-  struct  SResultRowCellInfo *resultInfo;
+  struct  SResultRowEntryInfo *resultInfo;
   SExtTagsInfo tagInfo;
   SPoint1      start;
   SPoint1      end;
+
+  SFunctionFpSet* fpSet;
 } SQLFunctionCtx;
 
 enum {
   TEXPR_NODE_DUMMY     = 0x0,
   TEXPR_BINARYEXPR_NODE= 0x1,
   TEXPR_UNARYEXPR_NODE = 0x2,
+  TEXPR_FUNCTION_NODE  = 0x3,
   TEXPR_COL_NODE       = 0x4,
   TEXPR_VALUE_NODE     = 0x8,
 };
@@ -137,10 +172,7 @@ typedef struct tExprNode {
   uint8_t nodeType;
   union {
     struct {
-      union {
+      int32_t           optr;   // binary operator
-        int32_t         optr;   // binary operator
-        int32_t         functionId;// unary operator
-      };
       void             *info;   // support filter operation on this expression only available for leaf node
       struct tExprNode *pLeft;  // left child pointer
       struct tExprNode *pRight; // right child pointer
@@ -148,44 +180,52 @@ typedef struct tExprNode {
 
     SSchema            *pSchema;// column node
     struct SVariant    *pVal;   // value node
+
+    struct {// function node
+      char              functionName[FUNCTIONS_NAME_MAX_LENGTH];
+//      int32_t           functionId;
+      int32_t           num;
+
+      // Note that the attribute of pChild is not the parameter of function, it is the columns that involved in the
+      // calculation instead.
+      // E.g., Cov(col1, col2), the column information, w.r.t. the col1 and col2, is kept in pChild nodes.
+      //  The concat function, concat(col1, col2), is a binary scalar
+      //  operator and is kept in the attribute of _node.
+      struct tExprNode **pChild;
+    } _function;
   };
 } tExprNode;
 
+//TODO create?
 void exprTreeToBinary(SBufferWriter* bw, tExprNode* pExprTree);
 void tExprTreeDestroy(tExprNode *pNode, void (*fp)(void *));
 
 typedef struct SAggFunctionInfo {
-  char     name[FUNCTIONS_NAME_MAX_LENGTH];
+  char      name[FUNCTIONS_NAME_MAX_LENGTH];
-  int8_t   type;         // Scalar function or aggregation function
+  int8_t    type;         // Scalar function or aggregation function
-  uint8_t  functionId;   // Function Id
+  uint32_t  functionId;   // Function Id
-  int8_t   sFunctionId;  // Transfer function for super table query
+  int8_t    sFunctionId;  // Transfer function for super table query
-  uint16_t status;
+  uint16_t  status;
 
-  bool (*init)(SQLFunctionCtx *pCtx, struct SResultRowCellInfo* pResultCellInfo);  // setup the execute environment
+  bool (*init)(SQLFunctionCtx *pCtx, struct SResultRowEntryInfo* pResultCellInfo);  // setup the execute environment
-  void (*exec)(SQLFunctionCtx *pCtx);
+  void (*addInput)(SQLFunctionCtx *pCtx);
 
   // finalizer must be called after all exec has been executed to generated final result.
-  void (*xFinalize)(SQLFunctionCtx *pCtx);
+  void (*finalize)(SQLFunctionCtx *pCtx);
-  void (*mergeFunc)(SQLFunctionCtx *pCtx);
+  void (*combine)(SQLFunctionCtx *pCtx);
 
   int32_t (*dataReqFunc)(SQLFunctionCtx *pCtx, STimeWindow* w, int32_t colId);
 } SAggFunctionInfo;
 
+struct SScalarFuncParam;
+
 typedef struct SScalarFunctionInfo {
-  char     name[FUNCTIONS_NAME_MAX_LENGTH];
+  char      name[FUNCTIONS_NAME_MAX_LENGTH];
-  int8_t   type;              // scalar function or aggregation function
+  int8_t    type;              // scalar function or aggregation function
-  uint8_t  functionId;        // index of scalar function
+  uint32_t  functionId;        // index of scalar function
-
+  void     (*process)(struct SScalarFuncParam* pOutput, size_t numOfInput, const struct SScalarFuncParam *pInput);
-  bool    (*init)(SQLFunctionCtx *pCtx, struct SResultRowCellInfo* pResultCellInfo);  // setup the execute environment
-  void    (*exec)(SQLFunctionCtx *pCtx);
 } SScalarFunctionInfo;
 
-typedef struct SResultDataInfo {
-  int16_t type;
-  int16_t bytes;
-  int32_t intermediateBytes;
-} SResultDataInfo;
-
 typedef struct SMultiFunctionsDesc {
   bool stableQuery;
   bool groupbyColumn;
@@ -195,11 +235,12 @@ typedef struct SMultiFunctionsDesc {
   bool hasFilter;
   bool onlyTagQuery;
   bool orderProjectQuery;
-  bool stateWindow;
   bool globalMerge;
   bool multigroupResult;
   bool blockDistribution;
+  bool stateWindow;
   bool timewindow;
+  bool sessionWindow;
   bool topbotQuery;
   bool interpQuery;
   bool distinct;
@@ -215,16 +256,61 @@ int32_t getResultDataInfo(int32_t dataType, int32_t dataBytes, int32_t functionI
  * @param len
  * @return
  */
-int32_t qIsBuiltinFunction(const char* name, int32_t len);
+int32_t qIsBuiltinFunction(const char* name, int32_t len, bool* scalarFunction);
 
 bool qIsValidUdf(SArray* pUdfInfo, const char* name, int32_t len, int32_t* functionId);
 
-const char* qGetFunctionName(int32_t functionId);
+bool qIsAggregateFunction(const char* functionName);
+
+tExprNode* exprTreeFromBinary(const void* data, size_t size);
 
 void extractFunctionDesc(SArray* pFunctionIdList, SMultiFunctionsDesc* pDesc);
 
 tExprNode* exprdup(tExprNode* pTree);
 
+void resetResultRowEntryResult(SQLFunctionCtx* pCtx, int32_t num);
+void cleanupResultRowEntry(struct SResultRowEntryInfo* pCell);
+int32_t getNumOfResult(SQLFunctionCtx* pCtx, int32_t num);
+bool isRowEntryCompleted(struct SResultRowEntryInfo* pEntry);
+bool isRowEntryInitialized(struct SResultRowEntryInfo* pEntry);
+
+struct SScalarFunctionSupport* createScalarFuncSupport(int32_t num);
+void destroyScalarFuncSupport(struct SScalarFunctionSupport* pSupport, int32_t num);
+struct SScalarFunctionSupport* getScalarFuncSupport(struct SScalarFunctionSupport* pSupport, int32_t index);
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+// fill api
+struct SFillInfo;
+struct SFillColInfo;
+
+typedef struct SPoint {
+  int64_t key;
+  void *  val;
+} SPoint;
+
+void taosFillSetStartInfo(struct SFillInfo* pFillInfo, int32_t numOfRows, TSKEY endKey);
+void taosResetFillInfo(struct SFillInfo* pFillInfo, TSKEY startTimestamp);
+void taosFillSetInputDataBlock(struct SFillInfo* pFillInfo, const struct SSDataBlock* pInput);
+struct SFillColInfo* createFillColInfo(SExprInfo* pExpr, int32_t numOfOutput, const int64_t* fillVal);
+bool taosFillHasMoreResults(struct SFillInfo* pFillInfo);
+
+struct SFillInfo* taosCreateFillInfo(int32_t order, TSKEY skey, int32_t numOfTags, int32_t capacity, int32_t numOfCols,
+                              int64_t slidingTime, int8_t slidingUnit, int8_t precision, int32_t fillType,
+                              struct SFillColInfo* pFillCol, void* handle);
+
+void* taosDestroyFillInfo(struct SFillInfo *pFillInfo);
+int64_t taosFillResultDataBlock(struct SFillInfo* pFillInfo, void** output, int32_t capacity);
+int64_t getFillInfoStart(struct SFillInfo *pFillInfo);
+
+int32_t taosGetLinearInterpolationVal(SPoint* point, int32_t outputType, SPoint* point1, SPoint* point2, int32_t inputType);
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
+// udf api
+struct SUdfInfo;
+
+void qAddUdfInfo(uint64_t id, struct SUdfInfo* pUdfInfo);
+void qRemoveUdfInfo(uint64_t id, struct SUdfInfo* pUdfInfo);
+
 #ifdef __cplusplus
 }
 #endif

include/libs/parser/parser.h

@@ -26,50 +26,6 @@ extern "C" {
 #include "tvariant.h"
 #include "function.h"
 
-typedef struct SColumn {
-  uint64_t     tableUid;
-  int32_t      columnIndex;
-  SColumnInfo  info;
-} SColumn;
-
-// the structure for sql function in select clause
-typedef struct SSqlExpr {
-  char      token[TSDB_COL_NAME_LEN];      // original token
-  SSchema   resSchema;
-  SColIndex colInfo;
-  uint64_t  uid;            // table uid, todo refactor use the pointer
-  int32_t   interBytes;     // inter result buffer size
-  int16_t   numOfParams;    // argument value of each function
-  SVariant  param[3];       // parameters are not more than 3
-} SSqlExpr;
-
-typedef struct SExprInfo {
-  SSqlExpr           base;
-  struct tExprNode  *pExpr;
-} SExprInfo;
-
-//typedef struct SInterval {
-//  int32_t  tz;            // query client timezone
-//  char     intervalUnit;
-//  char     slidingUnit;
-//  char     offsetUnit;
-//  int64_t  interval;
-//  int64_t  sliding;
-//  int64_t  offset;
-//} SInterval;
-//
-//typedef struct SSessionWindow {
-//  int64_t  gap;             // gap between two session window(in microseconds)
-//  int32_t  primaryColId;    // primary timestamp column
-//} SSessionWindow;
-
-typedef struct SGroupbyExpr {
-  int16_t  tableIndex;
-  SArray*  columnInfo;  // SArray<SColIndex>, group by columns information
-  int16_t  orderIndex;  // order by column index
-  int16_t  orderType;   // order by type: asc/desc
-} SGroupbyExpr;
-
 typedef struct SField {
   char     name[TSDB_COL_NAME_LEN];
   uint8_t  type;
@@ -82,16 +38,6 @@ typedef struct SFieldInfo {
   SArray     *internalField; // SArray<SInternalField>
 } SFieldInfo;
 
-typedef struct SLimit {
-  int64_t   limit;
-  int64_t   offset;
-} SLimit;
-
-typedef struct SOrder {
-  uint32_t  order;
-  int32_t   orderColId;
-} SOrder;
-
 typedef struct SCond {
   uint64_t uid;
   int32_t  len;  // length of tag query condition data
@@ -120,12 +66,6 @@ typedef struct STagCond {
 typedef struct STableMetaInfo {
   STableMeta    *pTableMeta;      // table meta, cached in client side and acquired by name
   SVgroupsInfo  *vgroupList;
-
-  /*
-   * 1. keep the vgroup index during the multi-vnode super table projection query
-   * 2. keep the vgroup index for multi-vnode insertion
-   */
-  int32_t       vgroupIndex;
   SName         name;
   char          aliasName[TSDB_TABLE_NAME_LEN];    // alias name of table specified in query sql
   SArray       *tagColList;                        // SArray<SColumn*>, involved tag columns
@@ -137,11 +77,11 @@ typedef struct SQueryStmtInfo {
   STimeWindow      window;        // the whole query time window
   SInterval        interval;      // tumble time window
   SSessionWindow   sessionWindow; // session time window
+  SStateWindow     stateWindow;   // state window query
   SGroupbyExpr     groupbyExpr;   // groupby tags info
   SArray *         colList;       // SArray<SColumn*>
   SFieldInfo       fieldsInfo;
-  SArray *         exprList;      // SArray<SExprInfo*>
+  SArray**         exprList;      // SArray<SExprInfo*>
-  SArray *         exprList1;     // final exprlist in case of arithmetic expression exists
   SLimit           limit;
   SLimit           slimit;
   STagCond         tagCond;
@@ -172,6 +112,7 @@ typedef struct SQueryStmtInfo {
   struct SQueryStmtInfo *pDownstream;
   int32_t            havingFieldNum;
   SMultiFunctionsDesc     info;
+  int32_t            exprListLevelIndex;
 } SQueryStmtInfo;
 
 typedef struct SColumnIndex {
@@ -225,14 +166,23 @@ void assignExprInfo(SExprInfo* dst, const SExprInfo* src);
 void columnListCopy(SArray* dst, const SArray* src, uint64_t uid);
 void columnListDestroy(SArray* pColumnList);
 
-void dropAllExprInfo(SArray* pExprInfo);
+void dropAllExprInfo(SArray** pExprInfo, int32_t numOfLevel);
-SExprInfo* createExprInfo(STableMetaInfo* pTableMetaInfo, int16_t functionId, SColumnIndex* pColIndex, struct tExprNode* pParamExpr, SSchema* pResSchema, int16_t interSize);
+
+typedef struct SSourceParam {
+  SArray            *pExprNodeList; //Array<struct tExprNode*>
+  SArray            *pColumnList;   //Array<struct SColumn>
+  int32_t    num;
+} SSourceParam;
+
+SExprInfo* createExprInfo(STableMetaInfo* pTableMetaInfo, const char* funcName, SSourceParam* pSource, SSchema* pResSchema, int16_t interSize);
 int32_t copyExprInfoList(SArray* dst, const SArray* src, uint64_t uid, bool deepcopy);
 
 STableMetaInfo* getMetaInfo(SQueryStmtInfo* pQueryInfo, int32_t tableIndex);
 SSchema *getOneColumnSchema(const STableMeta* pTableMeta, int32_t colIndex);
+SSchema createSchema(uint8_t type, int16_t bytes, int16_t colId, const char* name);
 
 int32_t getNewResColId();
+void addIntoSourceParam(SSourceParam* pSourceParam, tExprNode* pNode, SColumn* pColumn);
 
 #ifdef __cplusplus
 }

include/os/os.h

@@ -45,6 +45,8 @@ extern "C" {
 #include <float.h>
 #include <math.h>
 #include <sys/stat.h>
+#include <sys/types.h>
+#include <unistd.h>
 
 #include "osAtomic.h"
 #include "osDef.h"

include/os/osDir.h

@@ -22,7 +22,7 @@ extern "C" {
 
 void taosRemoveDir(const char *dirname);
 bool taosDirExist(char *dirname);
-bool taosMkDir(char *dirname);
+bool taosMkDir(const char *dirname);
 void taosRemoveOldFiles(char *dirname, int32_t keepDays);
 bool taosExpandDir(char *dirname, char *outname, int32_t maxlen);
 bool taosRealPath(char *dirname, int32_t maxlen);

include/server/vnode/meta/meta.h

@@ -1,99 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef _TD_META_H_
-#define _TD_META_H_
-
-#include "taosmsg.h"
-
-#include "os.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* ------------------------ APIs Exposed ------------------------ */
-
-// Types exported
-typedef uint64_t                tb_uid_t;
-typedef struct SMeta            SMeta;
-typedef struct SMetaOpts        SMetaOpts;
-typedef struct SMetaQueryHandle SMetaQueryHandle;
-typedef struct SMetaQueryOpts   SMetaQueryOpts;
-typedef struct STableOpts       STableOpts;
-
-// SMeta operations
-int    metaCreate(const char *path);
-void   metaDestroy(const char *path);
-SMeta *metaOpen(SMetaOpts *);
-void   metaClose(SMeta *);
-int    metaCreateTable(SMeta *, const STableOpts *);
-int    metaDropTable(SMeta *, uint64_t tuid_t);
-int    metaAlterTable(SMeta *, void *);
-int    metaCommit(SMeta *);
-
-// Options
-SMetaOpts *metaOptionsCreate();
-void       metaOptionsDestroy(SMetaOpts *);
-void       metaOptionsSetCache(SMetaOpts *, size_t capacity);
-
-// SMetaQueryHandle
-SMetaQueryHandle *metaQueryHandleCreate(SMetaQueryOpts *);
-void              metaQueryHandleDestroy(SMetaQueryHandle *);
-
-// SMetaQueryOpts
-SMetaQueryOpts *metaQueryOptionsCreate();
-void            metaQueryOptionsDestroy(SMetaQueryOpts *);
-
-// STableOpts
-#define META_TABLE_OPTS_DECLARE(name) STableOpts name = {0}
-void metaNormalTableOptsInit(STableOpts *, const char *name, const STSchema *pSchema);
-void metaSuperTableOptsInit(STableOpts *, const char *name, tb_uid_t uid, const STSchema *pSchema,
-                            const STSchema *pTagSchema);
-void metaChildTableOptsInit(STableOpts *, const char *name, tb_uid_t suid, const SKVRow tags);
-void metaTableOptsClear(STableOpts *);
-
-/* ------------------------ Impl should hidden ------------------------ */
-typedef enum { META_INIT_TABLE = 0, META_SUPER_TABLE = 1, META_CHILD_TABLE = 2, META_NORMAL_TABLE = 3 } EMetaTableT;
-typedef struct SSuperTableOpts {
-  tb_uid_t  uid;
-  STSchema *pSchema;     // (ts timestamp, a int)
-  STSchema *pTagSchema;  // (tag1 binary(10), tag2 int)
-} SSuperTableOpts;
-
-typedef struct SChildTableOpts {
-  tb_uid_t suid;  // super table uid
-  SKVRow   tags;  // tag value of the child table
-} SChildTableOpts;
-
-typedef struct SNormalTableOpts {
-  STSchema *pSchema;
-} SNormalTableOpts;
-
-struct STableOpts {
-  int8_t type;
-  char * name;
-  union {
-    SSuperTableOpts  superOpts;
-    SChildTableOpts  childOpts;
-    SNormalTableOpts normalOpts;
-  };
-};
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_META_H_*/

include/util/amalloc.h

@@ -1,52 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef _TD_AMALLOC_H_
-#define _TD_AMALLOC_H_
-
-#include "os.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#define AMALLOC_APIS                                  \
-  void *(*malloc)(void *, size_t size);               \
-  void *(*calloc)(void *, size_t nmemb, size_t size); \
-  void *(*realloc)(void *, size_t size);              \
-  void (*free)(void *ptr);
-
-// Interfaces to implement
-typedef struct {
-  AMALLOC_APIS
-} SMemAllocatorIf;
-
-typedef struct {
-  void *impl;
-  AMALLOC_APIS
-} SMemAllocator;
-
-#define amalloc(allocator, size) ((allocator) ? (*((allocator)->malloc))((allocator)->impl, (size)) : malloc(size))
-#define acalloc(allocator, nmemb, size) \
-  ((allocator) ? (*((allocator)->calloc))((allocator)->impl, (nmemb), (size)) : calloc((nmemb), (size)))
-#define arealloc(allocator, ptr, size) \
-  ((allocator) ? (*((allocator)->realloc))((allocator)->impl, (ptr), (size)) : realloc((ptr), (size)))
-#define afree(allocator, ptr, size) ((allocator) ? (*((allocator)->free))((allocator)->impl, (ptr), (size)) : free(ptr))
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_AMALLOC_H_*/

include/util/mallocator.h

@@ -0,0 +1,63 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_MALLOCATOR_H_
+#define _TD_MALLOCATOR_H_
+
+#include "os.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct SMemAllocator SMemAllocator;
+
+#define MALLOCATOR_APIS                                        \
+  void *(*malloc)(SMemAllocator *, size_t size);               \
+  void *(*calloc)(SMemAllocator *, size_t nmemb, size_t size); \
+  void *(*realloc)(SMemAllocator *, void *ptr, size_t size);   \
+  void (*free)(SMemAllocator *, void *ptr);                    \
+  size_t (*usage)(SMemAllocator *);
+
+// Interfaces to implement
+typedef struct {
+  MALLOCATOR_APIS
+} SMemAllocatorIf;
+
+struct SMemAllocator {
+  void * impl;
+  size_t usize;
+  MALLOCATOR_APIS
+};
+
+// heap allocator
+SMemAllocator *tdCreateHeapAllocator();
+void           tdDestroyHeapAllocator(SMemAllocator *pMemAllocator);
+
+// arena allocator
+SMemAllocator *tdCreateArenaAllocator(size_t size);
+void           tdDestroyArenaAllocator(SMemAllocator *);
+
+#define mMalloc(pMemAllocator, size) (*(pMemAllocator->malloc))(pMemAllocator, size)
+#define mCalloc(pMemAllocator, nmemb, size) (*(pMemAllocator->calloc))(pMemAllocator, nmemb, size)
+#define mRealloc(pMemAllocator, ptr, size) (*(pMemAllocator->realloc))(pMemAllocator, ptr, size)
+#define mFree(pMemAllocator, ptr) (*(pMemAllocator->free))(pMemAllocator, ptr)
+#define mUsage(pMemAllocator) (*(pMemAllocator->usage))(pMemAllocator)
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_MALLOCATOR_H_*/

include/util/taoserror.h

@@ -27,6 +27,7 @@ extern "C" {
 #define TAOS_FAILED(err)                    ((err) < 0)
 
 const char* tstrerror(int32_t err);
+const char* terrstr();
 
 int32_t* taosGetErrno();
 #define terrno                              (*taosGetErrno())
@@ -59,17 +60,20 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_RPC_INVALID_VERSION           TAOS_DEF_ERROR_CODE(0, 0x0016)  //"Invalid app version")
 
 //common & util
-#define TSDB_CODE_COM_OPS_NOT_SUPPORT           TAOS_DEF_ERROR_CODE(0, 0x0100)  //"Operation not supported")
+#define TSDB_CODE_OPS_NOT_SUPPORT               TAOS_DEF_ERROR_CODE(0, 0x0100)
-#define TSDB_CODE_COM_MEMORY_CORRUPTED          TAOS_DEF_ERROR_CODE(0, 0x0101)  //"Memory corrupted")
+#define TSDB_CODE_OUT_OF_MEMORY                 TAOS_DEF_ERROR_CODE(0, 0x0101)
-#define TSDB_CODE_COM_OUT_OF_MEMORY             TAOS_DEF_ERROR_CODE(0, 0x0102)  //"Out of memory")
+#define TSDB_CODE_OUT_OF_RANGE                  TAOS_DEF_ERROR_CODE(0, 0x0102)
-#define TSDB_CODE_COM_INVALID_CFG_MSG           TAOS_DEF_ERROR_CODE(0, 0x0103)  //"Invalid config message")
+#define TSDB_CODE_INVALID_PTR                   TAOS_DEF_ERROR_CODE(0, 0x0103)
-#define TSDB_CODE_COM_FILE_CORRUPTED            TAOS_DEF_ERROR_CODE(0, 0x0104)  //"Data file corrupted")
+#define TSDB_CODE_MEMORY_CORRUPTED              TAOS_DEF_ERROR_CODE(0, 0x0104)
-#define TSDB_CODE_REF_NO_MEMORY                 TAOS_DEF_ERROR_CODE(0, 0x0105)  //"Ref out of memory")
+#define TSDB_CODE_FILE_CORRUPTED                TAOS_DEF_ERROR_CODE(0, 0x0106)
-#define TSDB_CODE_REF_FULL                      TAOS_DEF_ERROR_CODE(0, 0x0106)  //"too many Ref Objs")
+#define TSDB_CODE_CHECKSUM_ERROR                TAOS_DEF_ERROR_CODE(0, 0x0107)
-#define TSDB_CODE_REF_ID_REMOVED                TAOS_DEF_ERROR_CODE(0, 0x0107)  //"Ref ID is removed")
+#define TSDB_CODE_INVALID_MSG                   TAOS_DEF_ERROR_CODE(0, 0x0108)
-#define TSDB_CODE_REF_INVALID_ID                TAOS_DEF_ERROR_CODE(0, 0x0108)  //"Invalid Ref ID")
+#define TSDB_CODE_REF_NO_MEMORY                 TAOS_DEF_ERROR_CODE(0, 0x0109)
-#define TSDB_CODE_REF_ALREADY_EXIST             TAOS_DEF_ERROR_CODE(0, 0x0109)  //"Ref is already there")
+#define TSDB_CODE_REF_FULL                      TAOS_DEF_ERROR_CODE(0, 0x010A)
-#define TSDB_CODE_REF_NOT_EXIST                 TAOS_DEF_ERROR_CODE(0, 0x010A)  //"Ref is not there")
+#define TSDB_CODE_REF_ID_REMOVED                TAOS_DEF_ERROR_CODE(0, 0x010B)
+#define TSDB_CODE_REF_INVALID_ID                TAOS_DEF_ERROR_CODE(0, 0x010C)
+#define TSDB_CODE_REF_ALREADY_EXIST             TAOS_DEF_ERROR_CODE(0, 0x010D)
+#define TSDB_CODE_REF_NOT_EXIST                 TAOS_DEF_ERROR_CODE(0, 0x010E)
 
 //client
 #define TSDB_CODE_TSC_INVALID_OPERATION         TAOS_DEF_ERROR_CODE(0, 0x0200)  //"Invalid Operation")
@@ -120,7 +124,6 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_INVALID_MSG_LEN           TAOS_DEF_ERROR_CODE(0, 0x0307)  //"Invalid message length")
 #define TSDB_CODE_MND_INVALID_MSG_TYPE          TAOS_DEF_ERROR_CODE(0, 0x0308)  //"Invalid message type")
 #define TSDB_CODE_MND_TOO_MANY_SHELL_CONNS      TAOS_DEF_ERROR_CODE(0, 0x0309)  //"Too many connections")
-#define TSDB_CODE_MND_OUT_OF_MEMORY             TAOS_DEF_ERROR_CODE(0, 0x030A)  //"Out of memory in mnode")
 #define TSDB_CODE_MND_INVALID_SHOWOBJ           TAOS_DEF_ERROR_CODE(0, 0x030B)  //"Data expired")
 #define TSDB_CODE_MND_INVALID_QUERY_ID          TAOS_DEF_ERROR_CODE(0, 0x030C)  //"Invalid query id")
 #define TSDB_CODE_MND_INVALID_STREAM_ID         TAOS_DEF_ERROR_CODE(0, 0x030D)  //"Invalid stream id")
@@ -131,12 +134,18 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_FAILED_TO_CREATE_DIR      TAOS_DEF_ERROR_CODE(0, 0x0313)  //"failed to create mnode dir")
 #define TSDB_CODE_MND_FAILED_TO_INIT_STEP       TAOS_DEF_ERROR_CODE(0, 0x0314)  //"failed to init components")
 
-#define TSDB_CODE_MND_SDB_OBJ_ALREADY_THERE     TAOS_DEF_ERROR_CODE(0, 0x0320)  //"Object already there")
+#define TSDB_CODE_SDB_APP_ERROR                 TAOS_DEF_ERROR_CODE(0, 0x0320)
-#define TSDB_CODE_MND_SDB_ERROR                 TAOS_DEF_ERROR_CODE(0, 0x0321)  //"Unexpected generic error in sdb")
+#define TSDB_CODE_SDB_OBJ_ALREADY_THERE         TAOS_DEF_ERROR_CODE(0, 0x0321)
-#define TSDB_CODE_MND_SDB_INVALID_TABLE_TYPE    TAOS_DEF_ERROR_CODE(0, 0x0322)  //"Invalid table type")
+#define TSDB_CODE_SDB_OBJ_NOT_THERE             TAOS_DEF_ERROR_CODE(0, 0x0322)
-#define TSDB_CODE_MND_SDB_OBJ_NOT_THERE         TAOS_DEF_ERROR_CODE(0, 0x0323)  //"Object not there")
+#define TSDB_CODE_SDB_OBJ_CREATING              TAOS_DEF_ERROR_CODE(0, 0x0323)
-#define TSDB_CODE_MND_SDB_INVAID_META_ROW       TAOS_DEF_ERROR_CODE(0, 0x0324)  //"Invalid meta row")
+#define TSDB_CODE_SDB_OBJ_DROPPING              TAOS_DEF_ERROR_CODE(0, 0x0324)
-#define TSDB_CODE_MND_SDB_INVAID_KEY_TYPE       TAOS_DEF_ERROR_CODE(0, 0x0325)  //"Invalid key type")
+#define TSDB_CODE_SDB_INVALID_TABLE_TYPE        TAOS_DEF_ERROR_CODE(0, 0x0325)
+#define TSDB_CODE_SDB_INVALID_KEY_TYPE          TAOS_DEF_ERROR_CODE(0, 0x0326)
+#define TSDB_CODE_SDB_INVALID_ACTION_TYPE       TAOS_DEF_ERROR_CODE(0, 0x0327)
+#define TSDB_CODE_SDB_INVALID_STATUS_TYPE       TAOS_DEF_ERROR_CODE(0, 0x0328)
+#define TSDB_CODE_SDB_INVALID_DATA_VER          TAOS_DEF_ERROR_CODE(0, 0x0329)
+#define TSDB_CODE_SDB_INVALID_DATA_LEN          TAOS_DEF_ERROR_CODE(0, 0x032A)
+#define TSDB_CODE_SDB_INVALID_DATA_CONTENT      TAOS_DEF_ERROR_CODE(0, 0x032B)
 
 #define TSDB_CODE_MND_DNODE_ALREADY_EXIST       TAOS_DEF_ERROR_CODE(0, 0x0330)  //"DNode already exists")
 #define TSDB_CODE_MND_DNODE_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x0331)  //"DNode does not exist")
@@ -155,12 +164,12 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_DNODE_EP_NOT_CONFIGURED   TAOS_DEF_ERROR_CODE(0, 0x033E)  //"Dnode Ep not configured")
 
 #define TSDB_CODE_MND_ACCT_ALREADY_EXIST        TAOS_DEF_ERROR_CODE(0, 0x0340)  //"Account already exists")
-#define TSDB_CODE_MND_INVALID_ACCT              TAOS_DEF_ERROR_CODE(0, 0x0341)  //"Invalid account")
+#define TSDB_CODE_MND_ACCT_NOT_EXIST            TAOS_DEF_ERROR_CODE(0, 0x0341)  //"Invalid account")
 #define TSDB_CODE_MND_INVALID_ACCT_OPTION       TAOS_DEF_ERROR_CODE(0, 0x0342)  //"Invalid account options")
 #define TSDB_CODE_MND_ACCT_EXPIRED              TAOS_DEF_ERROR_CODE(0, 0x0343)  //"Account authorization has expired")
 
 #define TSDB_CODE_MND_USER_ALREADY_EXIST        TAOS_DEF_ERROR_CODE(0, 0x0350)  //"User already exists")
-#define TSDB_CODE_MND_INVALID_USER              TAOS_DEF_ERROR_CODE(0, 0x0351)  //"Invalid user")
+#define TSDB_CODE_MND_USER_NOT_EXIST            TAOS_DEF_ERROR_CODE(0, 0x0351)  //"Invalid user")
 #define TSDB_CODE_MND_INVALID_USER_FORMAT       TAOS_DEF_ERROR_CODE(0, 0x0352)  //"Invalid user format")
 #define TSDB_CODE_MND_INVALID_PASS_FORMAT       TAOS_DEF_ERROR_CODE(0, 0x0353)  //"Invalid password format")
 #define TSDB_CODE_MND_NO_USER_FROM_CONN         TAOS_DEF_ERROR_CODE(0, 0x0354)  //"Can not get user from conn")

include/util/tcompression.h

@@ -20,7 +20,7 @@
 extern "C" {
 #endif
 
-#include "tdef.h"
+#include "taos.h"
 #include "tutil.h"
 
 #define COMP_OVERFLOW_BYTES 2

include/util/tdef.h

@@ -134,6 +134,15 @@ do { \
 #define TSDB_BINARY_OP_REMAINDER  4004
 #define TSDB_BINARY_OP_CONCAT     4005
 
+#define FUNCTION_CEIL        4500
+#define FUNCTION_FLOOR       4501
+#define FUNCTION_ABS         4502
+#define FUNCTION_ROUND       4503
+
+#define FUNCTION_LENGTH      4800
+#define FUNCTION_CONCAT      4801
+#define FUNCTION_LTRIM       4802
+#define FUNCTION_RTRIM       4803
 
 #define IS_RELATION_OPTR(op) (((op) >= TSDB_RELATION_LESS) && ((op) < TSDB_RELATION_IN))
 #define IS_ARITHMETIC_OPTR(op) (((op) >= TSDB_BINARY_OP_ADD) && ((op) <= TSDB_BINARY_OP_REMAINDER))
@@ -311,10 +320,6 @@ do { \
 #define TSDB_QUERY_TYPE_NON_TYPE               0x00u     // none type
 #define TSDB_QUERY_TYPE_FREE_RESOURCE          0x01u     // free qhandle at vnode
 
-#define TSDB_UDF_TYPE_SCALAR       1
-#define TSDB_UDF_TYPE_AGGREGATE    2
-
-
 /*
  * 1. ordinary sub query for select * from super_table
  * 2. all sqlobj generated by createSubqueryObj with this flag
@@ -367,21 +372,6 @@ do { \
 #define TSDB_MAX_DISKS_PER_TIER 16
 #define TSDB_MAX_DISKS           (TSDB_MAX_TIERS * TSDB_MAX_DISKS_PER_TIER)
 
-#define TSDB_DATA_TYPE_NULL       0     // 1 bytes
-#define TSDB_DATA_TYPE_BOOL       1     // 1 bytes
-#define TSDB_DATA_TYPE_TINYINT    2     // 1 byte
-#define TSDB_DATA_TYPE_SMALLINT   3     // 2 bytes
-#define TSDB_DATA_TYPE_INT        4     // 4 bytes
-#define TSDB_DATA_TYPE_BIGINT     5     // 8 bytes
-#define TSDB_DATA_TYPE_FLOAT      6     // 4 bytes
-#define TSDB_DATA_TYPE_DOUBLE     7     // 8 bytes
-#define TSDB_DATA_TYPE_BINARY     8     // string
-#define TSDB_DATA_TYPE_TIMESTAMP  9     // 8 bytes
-#define TSDB_DATA_TYPE_NCHAR      10    // unicode string
-#define TSDB_DATA_TYPE_UTINYINT   11    // 1 byte
-#define TSDB_DATA_TYPE_USMALLINT  12    // 2 bytes
-#define TSDB_DATA_TYPE_UINT       13    // 4 bytes
-#define TSDB_DATA_TYPE_UBIGINT    14    // 8 bytes
 
 enum { TRANS_STAT_INIT = 0, TRANS_STAT_EXECUTING, TRANS_STAT_EXECUTED, TRANS_STAT_ROLLBACKING, TRANS_STAT_ROLLBACKED };
 enum { TRANS_OPER_INIT = 0, TRANS_OPER_EXECUTE, TRANS_OPER_ROLLBACK };

include/util/tskiplist.h

@@ -21,7 +21,7 @@ extern "C" {
 #endif
 
 #include "os.h"
-//#include "tdef.h"
+#include "taos.h"
 #include "tarray.h"
 #include "tfunctional.h"
 

source/common/src/tglobal.c

@@ -40,7 +40,7 @@ uint16_t tsArbitratorPort = 6042;
 int32_t  tsStatusInterval = 1;  // second
 int32_t  tsNumOfMnodes = 1;
 int8_t   tsEnableVnodeBak = 1;
-int8_t   tsEnableTelemetryReporting = 1;
+int8_t   tsEnableTelemetryReporting = 0;
 int8_t   tsArbOnline = 0;
 int64_t  tsArbOnlineTimestamp = TSDB_ARB_DUMMY_TIME;
 char     tsEmail[TSDB_FQDN_LEN] = {0};

source/dnode/mgmt/CMakeLists.txt

@@ -11,6 +11,6 @@ target_link_libraries(
 )
 target_include_directories(
     taosd
-    PUBLIC "${CMAKE_SOURCE_DIR}/include/server/dnode"
+    PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode"
     private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
 )

source/dnode/mgmt/src/dnodeDnode.c

@@ -302,6 +302,13 @@ PRASE_DNODE_OVER:
     return -1;
   }
 
+  if (tsDnode.dnodeEps == NULL) {
+    tsDnode.dnodeEps = calloc(1, sizeof(SDnodeEps) + sizeof(SDnodeEp));
+    tsDnode.dnodeEps->dnodeNum = 1;
+    tsDnode.dnodeEps->dnodeEps[0].dnodePort = tsServerPort;
+    tstrncpy(tsDnode.dnodeEps->dnodeEps[0].dnodeFqdn, tsLocalFqdn, TSDB_FQDN_LEN);
+  }
+
   dnodeResetDnodes(tsDnode.dnodeEps);
 
   terrno = 0;

source/dnode/mgmt/src/dnodeInt.c

@@ -135,7 +135,7 @@ static int32_t dnodeInitMain() {
 
   dnodeInitDir();
 
-  return -1;
+  return 0;
 }
 
 static void dnodeCleanupMain() {
@@ -145,14 +145,14 @@ static void dnodeCleanupMain() {
 }
 
 int32_t dnodeInit() {
-  SSteps *steps = taosStepInit(24, dnodeReportStartup);
+  SSteps *steps = taosStepInit(10, dnodeReportStartup);
   if (steps == NULL) return -1;
 
   taosStepAdd(steps, "dnode-main", dnodeInitMain, dnodeCleanupMain);
   taosStepAdd(steps, "dnode-rpc", rpcInit, rpcCleanup);
   taosStepAdd(steps, "dnode-tfs", NULL, NULL);
   taosStepAdd(steps, "dnode-wal", walInit, walCleanUp);
-  taosStepAdd(steps, "dnode-sync", syncInit, syncCleanUp);
+  //taosStepAdd(steps, "dnode-sync", syncInit, syncCleanUp);
   taosStepAdd(steps, "dnode-dnode", dnodeInitDnode, dnodeCleanupDnode);
   taosStepAdd(steps, "dnode-vnodes", dnodeInitVnodes, dnodeCleanupVnodes);
   taosStepAdd(steps, "dnode-mnode", dnodeInitMnode, dnodeCleanupMnode);

source/dnode/mgmt/src/dnodeMnode.c

@@ -136,8 +136,8 @@ static int32_t dnodeWriteMnodeFile() {
   char   *content = calloc(1, maxLen + 1);
 
   len += snprintf(content + len, maxLen - len, "{\n");
-  len += snprintf(content + len, maxLen - len, "  \"deployed\": \"%d\",\n", tsMnode.dropped);
+  len += snprintf(content + len, maxLen - len, "  \"deployed\": \"%d\",\n", tsMnode.deployed);
-  len += snprintf(content + len, maxLen - len, "  \"dropped\": \"%d\",\n", tsMnode.dropped);
+  len += snprintf(content + len, maxLen - len, "  \"dropped\": \"%d\"\n", tsMnode.dropped);
   len += snprintf(content + len, maxLen - len, "}\n");
 
   fwrite(content, 1, len, fp);
@@ -180,7 +180,7 @@ static int32_t dnodeStartMnode() {
   tsMnode.deployed = 1;
   taosWUnLockLatch(&tsMnode.latch);
 
-  return code;
+  return mnodeStart(NULL);
 }
 
 static void dnodeStopMnode() {
@@ -212,14 +212,14 @@ static int32_t dnodeUnDeployMnode() {
   }
 
   dnodeStopMnode();
-  mnodeUnDeploy(tsMnodeDir);
+  mnodeUnDeploy();
   dnodeWriteMnodeFile();
 
   return code;
 }
 
 static int32_t dnodeDeployMnode(SMnodeCfg *pCfg) {
-  int32_t code = mnodeDeploy(tsMnodeDir, pCfg);
+  int32_t code = mnodeDeploy(pCfg);
   if (code != 0) {
     dError("failed to deploy mnode since %s", tstrerror(code));
     return code;
@@ -335,12 +335,9 @@ static int32_t dnodeWriteMnodeMsgToQueue(taos_queue pQueue, SRpcMsg *pRpcMsg) {
   if (pQueue == NULL) {
     code = TSDB_CODE_DND_MSG_NOT_PROCESSED;
   } else {
-    SMnodeMsg *pMsg = mnodeInitMsg(1);
+    SMnodeMsg *pMsg = mnodeInitMsg(pRpcMsg);
     if (pMsg == NULL) {
-      code = TSDB_CODE_DND_OUT_OF_MEMORY;
+      code = terrno;
-    } else {
-      mnodeAppendMsg(pMsg, pRpcMsg);
-      code = taosWriteQitem(pQueue, pMsg);
     }
   }
 
@@ -539,7 +536,7 @@ static int32_t dnodeOpenMnode() {
     SMnodeCfg cfg = {.replica = 1};
     cfg.replicas[0].port = tsServerPort;
     tstrncpy(cfg.replicas[0].fqdn, tsLocalFqdn, TSDB_FQDN_LEN);
-    return dnodeDeployMnode(&cfg);
+    code = dnodeDeployMnode(&cfg);
   } else {
     dInfo("start to open mnode");
     return dnodeStartMnode();

source/dnode/mgmt/src/dnodeVnodes.c

@@ -376,7 +376,7 @@ static void *dnodeOpenVnodeFunc(void *param) {
 
     char path[PATH_MAX + 20] = {0};
     snprintf(path, sizeof(path),"%s/vnode%d", tsVnodeDir, pVnode->vgId);
-    SVnode *pImpl = vnodeOpen(pVnode->vgId, path);
+    SVnode *pImpl = vnodeOpen(path, NULL);
     if (pImpl == NULL) {
       dError("vgId:%d, failed to open vnode by thread:%d", pVnode->vgId, pThread->threadIndex);
       pThread->failed++;

source/dnode/mnode/CMakeLists.txt

@@ -1,12 +1,3 @@
-aux_source_directory(src MNODE_SRC)
+add_subdirectory(impl)
-add_library(mnode ${MNODE_SRC})
+add_subdirectory(sdb)
-target_include_directories(
+add_subdirectory(transaction)
-    mnode
-    PUBLIC "${CMAKE_SOURCE_DIR}/include/server/mnode"
-    private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
-)
-target_link_libraries(
-    mnode
-    PUBLIC transport
-    PUBLIC cjson
-)

source/dnode/mnode/impl/CMakeLists.txt

@@ -0,0 +1,14 @@
+aux_source_directory(src MNODE_SRC)
+add_library(mnode ${MNODE_SRC})
+target_include_directories(
+    mnode
+    PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/mnode"
+    private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
+)
+target_link_libraries(
+    mnode
+    PRIVATE sdb
+    PRIVATE transaction
+    PUBLIC transport
+    PUBLIC cjson
+)

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

@@ -57,27 +57,6 @@ typedef struct SVgObj SVgObj;
 typedef struct SSTableObj SSTableObj;
 typedef struct SFuncObj SFuncObj; 
 typedef struct SOperObj SOperObj;
-typedef struct SMnMsg SMnMsg;
-
-typedef enum {
-  MN_SDB_START = 0,
-  MN_SDB_CLUSTER = 1,
-  MN_SDB_DNODE = 2,
-  MN_SDB_MNODE = 3,
-  MN_SDB_ACCT = 4,
-  MN_SDB_AUTH = 5,
-  MN_SDB_USER = 6,
-  MN_SDB_DB = 7,
-  MN_SDB_VGROUP = 8,
-  MN_SDB_STABLE = 9,
-  MN_SDB_FUNC = 10,
-  MN_SDB_OPER = 11,
-  MN_SDB_MAX = 12
-} EMnSdb;
-
-typedef enum { MN_OP_START = 0, MN_OP_INSERT = 1, MN_OP_UPDATE = 2, MN_OP_DELETE = 3, MN_OP_MAX = 4 } EMnOp;
-
-typedef enum { MN_KEY_START = 0, MN_KEY_BINARY = 1, MN_KEY_INT32 = 2, MN_KEY_INT64 = 3, MN_KEY_MAX } EMnKey;
 
 typedef enum {
   MN_AUTH_ACCT_START = 0,
@@ -97,16 +76,9 @@ typedef enum {
   MN_AUTH_MAX
 } EMnAuthOp;
 
-typedef enum { MN_SDB_STAT_AVAIL = 0, MN_SDB_STAT_DROPPED = 1 } EMnSdbStat;
 
-typedef struct {
-  int8_t type;
-  int8_t status;
-  int8_t align[6];
-} SdbHead;
 
 typedef struct SClusterObj {
-  SdbHead head;
   int64_t id;
   char    uid[TSDB_CLUSTER_ID_LEN];
   int64_t createdTime;
@@ -114,7 +86,6 @@ typedef struct SClusterObj {
 } SClusterObj;
 
 typedef struct SDnodeObj {
-  SdbHead  head;
   int32_t  id;
   int32_t  vnodes;
   int64_t  createdTime;
@@ -131,7 +102,6 @@ typedef struct SDnodeObj {
 } SDnodeObj;
 
 typedef struct SMnodeObj {
-  SdbHead    head;
   int32_t    id;
   int8_t     status;
   int8_t     role;
@@ -147,8 +117,8 @@ typedef struct {
   int32_t maxDbs;
   int32_t maxTimeSeries;
   int32_t maxStreams;
-  int64_t maxStorage;    // In unit of GB
+  int64_t maxStorage;   // In unit of GB
-  int8_t  accessState;  // Configured only by command
+  int32_t accessState;  // Configured only by command
 } SAcctCfg;
 
 typedef struct {
@@ -161,18 +131,16 @@ typedef struct {
 } SAcctInfo;
 
 typedef struct SAcctObj {
-  SdbHead   head;
   char      acct[TSDB_USER_LEN];
   int64_t   createdTime;
   int64_t   updateTime;
   int32_t   acctId;
-  int8_t    status;
+  int32_t   status;
   SAcctCfg  cfg;
   SAcctInfo info;
 } SAcctObj;
 
 typedef struct SUserObj {
-  SdbHead   head;
   char      user[TSDB_USER_LEN];
   char      pass[TSDB_KEY_LEN];
   char      acct[TSDB_USER_LEN];
@@ -207,7 +175,6 @@ typedef struct {
 } SDbCfg;
 
 typedef struct SDbObj {
-  SdbHead   head;
   char      name[TSDB_FULL_DB_NAME_LEN];
   char      acct[TSDB_USER_LEN];
   int64_t   createdTime;
@@ -251,7 +218,6 @@ typedef struct SVgObj {
 } SVgObj;
 
 typedef struct SSTableObj {
-  SdbHead    head;
   char       tableId[TSDB_TABLE_NAME_LEN];
   uint64_t   uid;
   int64_t    createdTime;
@@ -262,7 +228,6 @@ typedef struct SSTableObj {
 } SSTableObj;
 
 typedef struct SFuncObj {
-  SdbHead head;
   char    name[TSDB_FUNC_NAME_LEN];
   char    path[128];
   int32_t contLen;
@@ -299,8 +264,9 @@ typedef struct {
   void   *rsp;
 } SMnRsp;
 
-typedef struct SMnMsg {
+typedef struct SMnodeMsg {
-  void (*fp)(SMnMsg *pMsg, int32_t code);
+  void (*fp)(SMnodeMsg *pMsg, int32_t code);
+  SRpcConnInfo conn;
   SUserObj *pUser;
   int16_t   received;
   int16_t   successed;
@@ -311,7 +277,7 @@ typedef struct SMnMsg {
   SMnRsp    rpcRsp;
   SRpcMsg   rpcMsg;
   char      pCont[];
-} SMnReq;
+} SMnodeMsg;
 
 #ifdef __cplusplus
 }

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

@@ -17,22 +17,25 @@
 #define _TD_MNODE_INT_H_
 
 #include "mnodeDef.h"
+#include "sdb.h"
+#include "trn.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef enum { MN_STATUS_UNINIT = 0, MN_STATUS_INIT = 1, MN_STATUS_READY = 2, MN_STATUS_CLOSING = 3 } EMnStatus;
+typedef void (*MnodeRpcFp)(SMnodeMsg *pMsg);
 
-tmr_h     mnodeGetTimer();
+tmr_h   mnodeGetTimer();
-int32_t   mnodeGetDnodeId();
+int32_t mnodeGetDnodeId();
-int64_t   mnodeGetClusterId();
+int64_t mnodeGetClusterId();
-EMnStatus mnodeGetStatus();
 
 void mnodeSendMsgToDnode(struct SEpSet *epSet, struct SRpcMsg *rpcMsg);
 void mnodeSendMsgToMnode(struct SRpcMsg *rpcMsg);
 void mnodeSendRedirectMsg(struct SRpcMsg *rpcMsg, bool forShell);
 
+void mnodeSetMsgFp(int32_t msgType, MnodeRpcFp fp);
+
 #ifdef __cplusplus
 }
 #endif

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

@@ -24,6 +24,7 @@ extern "C" {
 
 int32_t mnodeInitSync();
 void    mnodeCleanUpSync();
+int32_t mnodeSyncPropose(SSdbRaw *pRaw, void *pData);
 
 bool mnodeIsMaster();
 

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

@@ -0,0 +1,393 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "os.h"
+#include "tglobal.h"
+#include "tstep.h"
+#include "tqueue.h"
+#include "mnodeAcct.h"
+#include "mnodeAuth.h"
+#include "mnodeBalance.h"
+#include "mnodeCluster.h"
+#include "mnodeDb.h"
+#include "mnodeDnode.h"
+#include "mnodeFunc.h"
+#include "mnodeMnode.h"
+#include "mnodeOper.h"
+#include "mnodeProfile.h"
+#include "mnodeShow.h"
+#include "mnodeStable.h"
+#include "mnodeSync.h"
+#include "mnodeTelem.h"
+#include "mnodeUser.h"
+#include "mnodeVgroup.h"
+
+static struct {
+  int32_t    dnodeId;
+  int64_t    clusterId;
+  tmr_h      timer;
+  SSteps    *pInitSteps;
+  SSteps    *pStartSteps;
+  SMnodePara para;
+  MnodeRpcFp msgFp[TSDB_MSG_TYPE_MAX];
+} tsMint;
+
+int32_t mnodeGetDnodeId() { return tsMint.para.dnodeId; }
+
+int64_t mnodeGetClusterId() { return tsMint.para.clusterId; }
+
+void mnodeSendMsgToDnode(struct SEpSet *epSet, struct SRpcMsg *rpcMsg) { (*tsMint.para.SendMsgToDnode)(epSet, rpcMsg); }
+
+void mnodeSendMsgToMnode(struct SRpcMsg *rpcMsg) { return (*tsMint.para.SendMsgToMnode)(rpcMsg); }
+
+void mnodeSendRedirectMsg(struct SRpcMsg *rpcMsg, bool forShell) { (*tsMint.para.SendRedirectMsg)(rpcMsg, forShell); }
+
+static int32_t mnodeInitTimer() {
+  if (tsMint.timer == NULL) {
+    tsMint.timer = taosTmrInit(tsMaxShellConns, 200, 3600000, "MND");
+  }
+
+  if (tsMint.timer == NULL) {
+    return -1;
+  }
+
+  return 0;
+}
+
+static void mnodeCleanupTimer() {
+  if (tsMint.timer != NULL) {
+    taosTmrCleanUp(tsMint.timer);
+    tsMint.timer = NULL;
+  }
+}
+
+tmr_h mnodeGetTimer() { return tsMint.timer; }
+
+static int32_t mnodeSetPara(SMnodePara para) {
+  tsMint.para = para;
+
+  if (tsMint.para.SendMsgToDnode == NULL) {
+    terrno = TSDB_CODE_MND_APP_ERROR;
+    return -1;
+  }
+
+  if (tsMint.para.SendMsgToMnode == NULL) {
+    terrno = TSDB_CODE_MND_APP_ERROR;
+    return -1;
+  }
+
+  if (tsMint.para.SendRedirectMsg == NULL) {
+    terrno = TSDB_CODE_MND_APP_ERROR;
+    return -1;
+  }
+
+  if (tsMint.para.PutMsgIntoApplyQueue == NULL) {
+    terrno = TSDB_CODE_MND_APP_ERROR;
+    return -1;
+  }
+
+  if (tsMint.para.dnodeId < 0) {
+    terrno = TSDB_CODE_MND_APP_ERROR;
+    return -1;
+  }
+
+  if (tsMint.para.clusterId < 0) {
+    terrno = TSDB_CODE_MND_APP_ERROR;
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mnodeAllocInitSteps() {
+  struct SSteps *steps = taosStepInit(16, NULL);
+  if (steps == NULL) return -1;
+
+  if (taosStepAdd(steps, "mnode-trans", trnInit, trnCleanup) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-cluster", mnodeInitCluster, mnodeCleanupCluster) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-dnode", mnodeInitDnode, mnodeCleanupDnode) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-mnode", mnodeInitMnode, mnodeCleanupMnode) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-acct", mnodeInitAcct, mnodeCleanupAcct) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-auth", mnodeInitAuth, mnodeCleanupAuth) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-user", mnodeInitUser, mnodeCleanupUser) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-db", mnodeInitDb, mnodeCleanupDb) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-vgroup", mnodeInitVgroup, mnodeCleanupVgroup) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-stable", mnodeInitStable, mnodeCleanupStable) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-func", mnodeInitFunc, mnodeCleanupFunc) != 0) return -1;
+  if (taosStepAdd(steps, "mnode-sdb", sdbInit, sdbCleanup) != 0) return -1;
+
+  tsMint.pInitSteps = steps;
+  return 0;
+}
+
+static int32_t mnodeAllocStartSteps() {
+  struct SSteps *steps = taosStepInit(8, NULL);
+  if (steps == NULL) return -1;
+
+  taosStepAdd(steps, "mnode-timer", mnodeInitTimer, NULL);
+  taosStepAdd(steps, "mnode-sdb-file", sdbOpen, sdbClose);
+  taosStepAdd(steps, "mnode-balance", mnodeInitBalance, mnodeCleanupBalance);
+  taosStepAdd(steps, "mnode-profile", mnodeInitProfile, mnodeCleanupProfile);
+  taosStepAdd(steps, "mnode-show", mnodeInitShow, mnodeCleanUpShow);
+  taosStepAdd(steps, "mnode-sync", mnodeInitSync, mnodeCleanUpSync);
+  taosStepAdd(steps, "mnode-telem", mnodeInitTelem, mnodeCleanupTelem);
+  taosStepAdd(steps, "mnode-timer", NULL, mnodeCleanupTimer);
+
+  tsMint.pStartSteps = steps;
+  return 0;
+}
+
+int32_t mnodeInit(SMnodePara para) {
+  if (mnodeSetPara(para) != 0) {
+    mError("failed to init mnode para since %s", terrstr());
+    return -1;
+  }
+
+  if (mnodeAllocInitSteps() != 0) {
+    mError("failed to alloc init steps since %s", terrstr());
+    return -1;
+  }
+
+  if (mnodeAllocStartSteps() != 0) {
+    mError("failed to alloc start steps since %s", terrstr());
+    return -1;
+  }
+
+  return taosStepExec(tsMint.pInitSteps);
+}
+
+void mnodeCleanup() { taosStepCleanup(tsMint.pInitSteps); }
+
+int32_t mnodeDeploy(SMnodeCfg *pCfg) {
+  if (tsMint.para.dnodeId <= 0 && tsMint.para.clusterId <= 0) {
+    if (sdbDeploy() != 0) {
+      mError("failed to deploy sdb since %s", terrstr());
+      return -1;
+    }
+  }
+
+  mDebug("mnode is deployed");
+  return 0;
+}
+
+void mnodeUnDeploy() { sdbUnDeploy(); }
+
+int32_t mnodeStart(SMnodeCfg *pCfg) { return taosStepExec(tsMint.pStartSteps); }
+
+int32_t mnodeAlter(SMnodeCfg *pCfg) { return 0; }
+
+void mnodeStop() { taosStepCleanup(tsMint.pStartSteps); }
+
+int32_t mnodeGetLoad(SMnodeLoad *pLoad) { return 0; }
+
+SMnodeMsg *mnodeInitMsg(SRpcMsg *pRpcMsg) {
+  SMnodeMsg *pMsg = taosAllocateQitem(sizeof(SMnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  if (rpcGetConnInfo(pRpcMsg->handle, &pMsg->conn) != 0) {
+    mnodeCleanupMsg(pMsg);
+    mError("can not get user from conn:%p", pMsg->rpcMsg.handle);
+    terrno = TSDB_CODE_MND_NO_USER_FROM_CONN;
+    return NULL;
+  }
+
+  pMsg->rpcMsg = *pRpcMsg;
+  pMsg->createdTime = taosGetTimestampSec();
+
+  return pMsg;
+}
+
+void mnodeCleanupMsg(SMnodeMsg *pMsg) {
+  if (pMsg->pUser != NULL) {
+    sdbRelease(pMsg->pUser);
+  }
+
+  taosFreeQitem(pMsg);
+}
+
+static void mnodeProcessRpcMsg(SMnodeMsg *pMsg) {
+  int32_t msgType = pMsg->rpcMsg.msgType;
+
+  if (tsMint.msgFp[msgType] == NULL) {
+  }
+
+  (*tsMint.msgFp[msgType])(pMsg);
+}
+
+void mnodeSetMsgFp(int32_t msgType, MnodeRpcFp fp) {
+  if (msgType > 0 || msgType < TSDB_MSG_TYPE_MAX) {
+    tsMint.msgFp[msgType] = fp;
+  }
+}
+
+void mnodeProcessMsg(SMnodeMsg *pMsg, EMnMsgType msgType) {
+  if (!mnodeIsMaster()) {
+    mnodeSendRedirectMsg(&pMsg->rpcMsg, true);
+    mnodeCleanupMsg(pMsg);
+    return;
+  }
+
+  switch (msgType) {
+    case MN_MSG_TYPE_READ:
+    case MN_MSG_TYPE_WRITE:
+    case MN_MSG_TYPE_SYNC:
+      mnodeProcessRpcMsg(pMsg);
+      break;
+    case MN_MSG_TYPE_APPLY:
+      break;
+    default:
+      break;
+  }
+}
+
+#if 0
+
+static void mnodeProcessWriteReq(SMnodeMsg *pMsg, void *unused) {
+  int32_t msgType = pMsg->rpcMsg.msgType;
+  void   *ahandle = pMsg->rpcMsg.ahandle;
+  int32_t code = 0;
+
+  if (pMsg->rpcMsg.pCont == NULL) {
+    mError("msg:%p, app:%p type:%s content is null", pMsg, ahandle, taosMsg[msgType]);
+    code = TSDB_CODE_MND_INVALID_MSG_LEN;
+    goto PROCESS_WRITE_REQ_END;
+  }
+
+  if (!mnodeIsMaster()) {
+    SMnRsp *rpcRsp = &pMsg->rpcRsp;
+    SEpSet *epSet = rpcMallocCont(sizeof(SEpSet));
+    mnodeGetMnodeEpSetForShell(epSet, true);
+    rpcRsp->rsp = epSet;
+    rpcRsp->len = sizeof(SEpSet);
+
+    mDebug("msg:%p, app:%p type:%s in write queue, is redirected, numOfEps:%d inUse:%d", pMsg, ahandle,
+           taosMsg[msgType], epSet->numOfEps, epSet->inUse);
+
+    code = TSDB_CODE_RPC_REDIRECT;
+    goto PROCESS_WRITE_REQ_END;
+  }
+
+  if (tsMworker.writeMsgFp[msgType] == NULL) {
+    mError("msg:%p, app:%p type:%s not processed", pMsg, ahandle, taosMsg[msgType]);
+    code = TSDB_CODE_MND_MSG_NOT_PROCESSED;
+    goto PROCESS_WRITE_REQ_END;
+  }
+
+  code = (*tsMworker.writeMsgFp[msgType])(pMsg);
+
+PROCESS_WRITE_REQ_END:
+  mnodeSendRsp(pMsg, code);
+}
+
+static void mnodeProcessReadReq(SMnodeMsg *pMsg, void *unused) {
+  int32_t msgType = pMsg->rpcMsg.msgType;
+  void   *ahandle = pMsg->rpcMsg.ahandle;
+  int32_t code = 0;
+
+  if (pMsg->rpcMsg.pCont == NULL) {
+    mError("msg:%p, app:%p type:%s in mread queue, content is null", pMsg, ahandle, taosMsg[msgType]);
+    code = TSDB_CODE_MND_INVALID_MSG_LEN;
+    goto PROCESS_READ_REQ_END;
+  }
+
+  if (!mnodeIsMaster()) {
+    SMnRsp *rpcRsp = &pMsg->rpcRsp;
+    SEpSet *epSet = rpcMallocCont(sizeof(SEpSet));
+    if (!epSet) {
+      code = TSDB_CODE_OUT_OF_MEMORY;
+      goto PROCESS_READ_REQ_END;
+    }
+    mnodeGetMnodeEpSetForShell(epSet, true);
+    rpcRsp->rsp = epSet;
+    rpcRsp->len = sizeof(SEpSet);
+
+    mDebug("msg:%p, app:%p type:%s in mread queue is redirected, numOfEps:%d inUse:%d", pMsg, ahandle, taosMsg[msgType],
+           epSet->numOfEps, epSet->inUse);
+    code = TSDB_CODE_RPC_REDIRECT;
+    goto PROCESS_READ_REQ_END;
+  }
+
+  if (tsMworker.readMsgFp[msgType] == NULL) {
+    mError("msg:%p, app:%p type:%s in mread queue, not processed", pMsg, ahandle, taosMsg[msgType]);
+    code = TSDB_CODE_MND_MSG_NOT_PROCESSED;
+    goto PROCESS_READ_REQ_END;
+  }
+
+  mTrace("msg:%p, app:%p type:%s will be processed in mread queue", pMsg, ahandle, taosMsg[msgType]);
+  code = (*tsMworker.readMsgFp[msgType])(pMsg);
+
+PROCESS_READ_REQ_END:
+  mnodeSendRsp(pMsg, code);
+}
+
+static void mnodeProcessPeerReq(SMnodeMsg *pMsg, void *unused) {
+  int32_t msgType = pMsg->rpcMsg.msgType;
+  void   *ahandle = pMsg->rpcMsg.ahandle;
+  int32_t code = 0;
+
+  if (pMsg->rpcMsg.pCont == NULL) {
+    mError("msg:%p, ahandle:%p type:%s in mpeer queue, content is null", pMsg, ahandle, taosMsg[msgType]);
+    code = TSDB_CODE_MND_INVALID_MSG_LEN;
+    goto PROCESS_PEER_REQ_END;
+  }
+
+  if (!mnodeIsMaster()) {
+    SMnRsp *rpcRsp = &pMsg->rpcRsp;
+    SEpSet *epSet = rpcMallocCont(sizeof(SEpSet));
+    mnodeGetMnodeEpSetForPeer(epSet, true);
+    rpcRsp->rsp = epSet;
+    rpcRsp->len = sizeof(SEpSet);
+
+    mDebug("msg:%p, ahandle:%p type:%s in mpeer queue is redirected, numOfEps:%d inUse:%d", pMsg, ahandle,
+           taosMsg[msgType], epSet->numOfEps, epSet->inUse);
+
+    code = TSDB_CODE_RPC_REDIRECT;
+    goto PROCESS_PEER_REQ_END;
+  }
+
+  if (tsMworker.peerReqFp[msgType] == NULL) {
+    mError("msg:%p, ahandle:%p type:%s in mpeer queue, not processed", pMsg, ahandle, taosMsg[msgType]);
+    code = TSDB_CODE_MND_MSG_NOT_PROCESSED;
+    goto PROCESS_PEER_REQ_END;
+  }
+
+  code = (*tsMworker.peerReqFp[msgType])(pMsg);
+
+PROCESS_PEER_REQ_END:
+  mnodeSendRsp(pMsg, code);
+}
+
+static void mnodeProcessPeerRsp(SMnodeMsg *pMsg, void *unused) {
+  int32_t  msgType = pMsg->rpcMsg.msgType;
+  SRpcMsg *pRpcMsg = &pMsg->rpcMsg;
+
+  if (!mnodeIsMaster()) {
+    mError("msg:%p, ahandle:%p type:%s not processed for not master", pRpcMsg, pRpcMsg->ahandle, taosMsg[msgType]);
+    mnodeCleanupMsg2(pMsg);
+  }
+
+  if (tsMworker.peerRspFp[msgType]) {
+    (*tsMworker.peerRspFp[msgType])(pRpcMsg);
+  } else {
+    mError("msg:%p, ahandle:%p type:%s is not processed", pRpcMsg, pRpcMsg->ahandle, taosMsg[msgType]);
+  }
+
+  mnodeCleanupMsg2(pMsg);
+}
+#endif

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

@@ -0,0 +1,118 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "mnodeInt.h"
+
+#define SDB_ACCT_VER 1
+
+static SSdbRaw *mnodeAcctActionEncode(SAcctObj *pAcct) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_ACCT, SDB_ACCT_VER, sizeof(SAcctObj));
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_BINARY(pRaw, dataPos, pAcct->acct, TSDB_USER_LEN)
+  SDB_SET_INT64(pRaw, dataPos, pAcct->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pAcct->updateTime)
+  SDB_SET_INT32(pRaw, dataPos, pAcct->acctId)
+  SDB_SET_INT32(pRaw, dataPos, pAcct->status)
+  SDB_SET_INT32(pRaw, dataPos, pAcct->cfg.maxUsers)
+  SDB_SET_INT32(pRaw, dataPos, pAcct->cfg.maxDbs)
+  SDB_SET_INT32(pRaw, dataPos, pAcct->cfg.maxTimeSeries)
+  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_DATALEN(pRaw, dataPos);
+
+  return pRaw;
+}
+
+static SSdbRow *mnodeAcctActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != SDB_ACCT_VER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    return NULL;
+  }
+
+  SSdbRow *pRow = sdbAllocRow(sizeof(SAcctObj));
+  SAcctObj *pAcct = sdbGetRowObj(pRow);
+  if (pAcct == NULL) return NULL;
+  
+  int32_t dataPos = 0;
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pAcct->acct, TSDB_USER_LEN)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pAcct->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pAcct->updateTime)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->acctId)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->status)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->cfg.maxUsers)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->cfg.maxDbs)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pAcct->cfg.maxTimeSeries)
+  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)
+
+  return pRow;
+}
+
+static int32_t mnodeAcctActionInsert(SAcctObj *pAcct) { return 0; }
+
+static int32_t mnodeAcctActionDelete(SAcctObj *pAcct) { return 0; }
+
+static int32_t mnodeAcctActionUpdate(SAcctObj *pSrcAcct, SAcctObj *pDstAcct) {
+  SAcctObj tObj;
+  int32_t  len = (int32_t)((int8_t *)&tObj.info - (int8_t *)&tObj);
+  memcpy(pDstAcct, pSrcAcct, len);
+  return 0;
+}
+
+static int32_t mnodeCreateDefaultAcct() {
+  int32_t code = 0;
+
+  SAcctObj acctObj = {0};
+  tstrncpy(acctObj.acct, TSDB_DEFAULT_USER, TSDB_USER_LEN);
+  acctObj.createdTime = taosGetTimestampMs();
+  acctObj.updateTime = acctObj.createdTime;
+  acctObj.acctId = 1;
+  acctObj.cfg = (SAcctCfg){.maxUsers = 1024,
+                           .maxDbs = 1024,
+                           .maxTimeSeries = INT32_MAX,
+                           .maxStreams = 8092,
+                           .maxStorage = INT64_MAX,
+                           .accessState = TSDB_VN_ALL_ACCCESS};
+
+  SSdbRaw *pRaw = mnodeAcctActionEncode(&acctObj);
+  if (pRaw == NULL) return -1;
+  sdbSetRawStatus(pRaw, SDB_STATUS_READY);
+
+  return sdbWrite(pRaw);
+}
+
+int32_t mnodeInitAcct() {
+  SSdbTable table = {.sdbType = SDB_ACCT,
+                     .keyType = SDB_KEY_BINARY,
+                     .deployFp = (SdbDeployFp)mnodeCreateDefaultAcct,
+                     .encodeFp = (SdbEncodeFp)mnodeAcctActionEncode,
+                     .decodeFp = (SdbDecodeFp)mnodeAcctActionDecode,
+                     .insertFp = (SdbInsertFp)mnodeAcctActionInsert,
+                     .updateFp = (SdbUpdateFp)mnodeAcctActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mnodeAcctActionDelete};
+  sdbSetTable(table);
+
+  return 0;
+}
+
+void mnodeCleanupAcct() {}

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

@@ -20,4 +20,10 @@
 int32_t mnodeInitSync() { return 0; }
 void    mnodeCleanUpSync() {}
 
+int32_t mnodeSyncPropose(SSdbRaw *pRaw, void *pData) {
+  trnApply(pData, pData, 0);
+  free(pData);
+  return 0;
+}
+
 bool mnodeIsMaster() { return true; }

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

@@ -17,6 +17,7 @@
 #include "mnodeTelem.h"
 #include "tbuffer.h"
 #include "tglobal.h"
+#include "mnodeSync.h"
 
 #define TELEMETRY_SERVER "telemetry.taosdata.com"
 #define TELEMETRY_PORT 80
@@ -255,7 +256,7 @@ static void* mnodeTelemThreadFp(void* param) {
     if (r == 0) break;
     if (r != ETIMEDOUT) continue;
 
-    if (mnodeGetStatus() == MN_STATUS_READY) {
+    if (mnodeIsMaster()) {
       mnodeSendTelemetryReport();
     }
     end.tv_sec += REPORT_INTERVAL;

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

@@ -0,0 +1,234 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "mnodeSync.h"
+#include "os.h"
+#include "tglobal.h"
+#include "tkey.h"
+
+#define SDB_USER_VER 1
+
+static SSdbRaw *mnodeUserActionEncode(SUserObj *pUser) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, SDB_USER_VER, sizeof(SAcctObj));
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_BINARY(pRaw, dataPos, pUser->user, TSDB_USER_LEN)
+  SDB_SET_BINARY(pRaw, dataPos, pUser->pass, TSDB_KEY_LEN)
+  SDB_SET_BINARY(pRaw, dataPos, pUser->acct, TSDB_USER_LEN)
+  SDB_SET_INT64(pRaw, dataPos, pUser->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pUser->updateTime)
+  SDB_SET_INT8(pRaw, dataPos, pUser->rootAuth)
+  SDB_SET_DATALEN(pRaw, dataPos);
+
+  return pRaw;
+}
+
+static SSdbRow *mnodeUserActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != SDB_USER_VER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    return NULL;
+  }
+
+  SSdbRow  *pRow = sdbAllocRow(sizeof(SUserObj));
+  SUserObj *pUser = sdbGetRowObj(pRow);
+  if (pUser == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pUser->user, TSDB_USER_LEN)
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pUser->pass, TSDB_KEY_LEN)
+  SDB_GET_BINARY(pRaw, pRow, dataPos, pUser->acct, TSDB_USER_LEN)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pUser->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pUser->updateTime)
+  SDB_GET_INT8(pRaw, pRow, dataPos, &pUser->rootAuth)
+
+  return pRow;
+}
+
+static int32_t mnodeUserActionInsert(SUserObj *pUser) {
+  pUser->prohibitDbHash = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
+  if (pUser->prohibitDbHash == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  pUser->pAcct = sdbAcquire(SDB_ACCT, pUser->acct);
+  if (pUser->pAcct == NULL) {
+    terrno = TSDB_CODE_MND_ACCT_NOT_EXIST;
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mnodeUserActionDelete(SUserObj *pUser) {
+  if (pUser->prohibitDbHash) {
+    taosHashCleanup(pUser->prohibitDbHash);
+    pUser->prohibitDbHash = NULL;
+  }
+
+  if (pUser->acct != NULL) {
+    sdbRelease(pUser->pAcct);
+    pUser->pAcct = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mnodeUserActionUpdate(SUserObj *pSrcUser, SUserObj *pDstUser) {
+  SUserObj tObj;
+  int32_t  len = (int32_t)((int8_t *)tObj.prohibitDbHash - (int8_t *)&tObj);
+  memcpy(pDstUser, pSrcUser, len);
+  return 0;
+}
+
+static int32_t mnodeCreateDefaultUser(char *acct, char *user, char *pass) {
+  SUserObj userObj = {0};
+  tstrncpy(userObj.user, user, TSDB_USER_LEN);
+  tstrncpy(userObj.acct, acct, TSDB_USER_LEN);
+  taosEncryptPass((uint8_t *)pass, strlen(pass), userObj.pass);
+  userObj.createdTime = taosGetTimestampMs();
+  userObj.updateTime = userObj.createdTime;
+
+  if (strcmp(user, TSDB_DEFAULT_USER) == 0) {
+    userObj.rootAuth = 1;
+  }
+
+  SSdbRaw *pRaw = mnodeUserActionEncode(&userObj);
+  if (pRaw == NULL) return -1;
+  sdbSetRawStatus(pRaw, SDB_STATUS_READY);
+
+  return sdbWrite(pRaw);
+}
+
+static int32_t mnodeCreateDefaultUsers() {
+  if (mnodeCreateDefaultUser(TSDB_DEFAULT_USER, TSDB_DEFAULT_USER, TSDB_DEFAULT_PASS) != 0) {
+    return -1;
+  }
+
+  if (mnodeCreateDefaultUser(TSDB_DEFAULT_USER, "monitor", tsInternalPass) != 0) {
+    return -1;
+  }
+
+  if (mnodeCreateDefaultUser(TSDB_DEFAULT_USER, "_" TSDB_DEFAULT_USER, tsInternalPass) != 0) {
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mnodeCreateUser(char *acct, char *user, char *pass, SMnodeMsg *pMsg) {
+  SUserObj userObj = {0};
+  tstrncpy(userObj.user, user, TSDB_USER_LEN);
+  tstrncpy(userObj.acct, acct, TSDB_USER_LEN);
+  taosEncryptPass((uint8_t *)pass, strlen(pass), userObj.pass);
+  userObj.createdTime = taosGetTimestampMs();
+  userObj.updateTime = userObj.createdTime;
+  userObj.rootAuth = 0;
+
+  STrans *pTrans = trnCreate(TRN_POLICY_ROLLBACK);
+  if (pTrans == NULL) return -1;
+  trnSetRpcHandle(pTrans, pMsg->rpcMsg.handle);
+
+  SSdbRaw *pRedoRaw = mnodeUserActionEncode(&userObj);
+  if (pRedoRaw == NULL || trnAppendRedoLog(pTrans, pRedoRaw) != 0) {
+    trnDrop(pTrans);
+    return -1;
+  }
+  sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING);
+
+  SSdbRaw *pUndoRaw = mnodeUserActionEncode(&userObj);
+  if (pUndoRaw == NULL || trnAppendUndoLog(pTrans, pUndoRaw) != 0) {
+    trnDrop(pTrans);
+    return -1;
+  }
+  sdbSetRawStatus(pUndoRaw, SDB_STATUS_DROPPED);
+
+  SSdbRaw *pCommitRaw = mnodeUserActionEncode(&userObj);
+  if (pCommitRaw == NULL || trnAppendCommitLog(pTrans, pCommitRaw) != 0) {
+    trnDrop(pTrans);
+    return -1;
+  }
+  sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY);
+
+  if (trnPrepare(pTrans, mnodeSyncPropose) != 0) {
+    trnDrop(pTrans);
+    return -1;
+  }
+
+  trnDrop(pTrans);
+  return 0;
+}
+
+static int32_t mnodeProcessCreateUserMsg(SMnodeMsg *pMsg) {
+  SCreateUserMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  if (pCreate->user[0] == 0) {
+    terrno = TSDB_CODE_MND_INVALID_USER_FORMAT;
+    mError("user:%s, failed to create since %s", pCreate->user, terrstr());
+    return -1;
+  }
+
+  if (pCreate->pass[0] == 0) {
+    terrno = TSDB_CODE_MND_INVALID_PASS_FORMAT;
+    mError("user:%s, failed to create since %s", pCreate->user, terrstr());
+    return -1;
+  }
+
+  SUserObj *pUser = sdbAcquire(SDB_USER, pCreate->user);
+  if (pUser != NULL) {
+    sdbRelease(pUser);
+    terrno = TSDB_CODE_MND_USER_ALREADY_EXIST;
+    mError("user:%s, failed to create since %s", pCreate->user, terrstr());
+    return -1;
+  }
+
+  SUserObj *pOperUser = sdbAcquire(SDB_USER, pMsg->conn.user);
+  if (pOperUser == NULL) {
+    terrno = TSDB_CODE_MND_NO_USER_FROM_CONN;
+    mError("user:%s, failed to create since %s", pCreate->user, terrstr());
+    return -1;
+  }
+
+  int32_t code = mnodeCreateUser(pOperUser->acct, pCreate->user, pCreate->pass, pMsg);
+  sdbRelease(pOperUser);
+
+  if (code != 0) {
+    mError("user:%s, failed to create since %s", pCreate->user, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+int32_t mnodeInitUser() {
+  SSdbTable table = {.sdbType = SDB_USER,
+                     .keyType = SDB_KEY_BINARY,
+                     .deployFp = (SdbDeployFp)mnodeCreateDefaultUsers,
+                     .encodeFp = (SdbEncodeFp)mnodeUserActionEncode,
+                     .decodeFp = (SdbDecodeFp)mnodeUserActionDecode,
+                     .insertFp = (SdbInsertFp)mnodeUserActionInsert,
+                     .updateFp = (SdbUpdateFp)mnodeUserActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mnodeUserActionDelete};
+  sdbSetTable(table);
+
+  return 0;
+}
+
+void mnodeCleanupUser() {}

source/dnode/mnode/inc/mnodeSdb.h

@@ -1,52 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#ifndef _TD_MNODE_SDB_H_
-#define _TD_MNODE_SDB_H_
-
-#include "mnodeInt.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-typedef void (*SdbDeployFp)();
-typedef void *(*SdbDecodeFp)(cJSON *root);
-typedef int32_t (*SdbEncodeFp)(void *pHead, char *buf, int32_t maxLen);
-
-int32_t sdbInit();
-void    sdbCleanup();
-
-int32_t sdbRead();
-int32_t sdbCommit();
-
-int32_t sdbDeploy();
-void    sdbUnDeploy();
-
-void   *sdbInsertRow(EMnSdb sdb, void *pObj);
-void    sdbDeleteRow(EMnSdb sdb, void *pHead);
-void   *sdbUpdateRow(EMnSdb sdb, void *pHead);
-void   *sdbGetRow(EMnSdb sdb, void *pKey);
-void   *sdbFetchRow(EMnSdb sdb, void *pIter);
-void    sdbCancelFetch(EMnSdb sdb, void *pIter);
-int32_t sdbGetCount(EMnSdb sdb);
-
-void sdbSetFp(EMnSdb, EMnKey, SdbDeployFp, SdbEncodeFp, SdbDecodeFp, int32_t dataSize);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /*_TD_MNODE_INT_H_*/

source/dnode/mnode/sdb/CMakeLists.txt

@@ -0,0 +1,13 @@
+aux_source_directory(src MNODE_SRC)
+add_library(sdb ${MNODE_SRC})
+target_include_directories(
+    sdb
+    PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/mnode/sdb"
+    private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
+)
+target_link_libraries(
+    sdb
+    PRIVATE os
+    PRIVATE common
+    PRIVATE util
+)

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

@@ -0,0 +1,80 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_SDB_INT_H_
+#define _TD_SDB_INT_H_
+
+#include "os.h"
+#include "sdb.h"
+#include "taosmsg.h"
+#include "thash.h"
+#include "tlockfree.h"
+#include "tlog.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define mFatal(...) { if (mDebugFlag & DEBUG_FATAL) { taosPrintLog("MND FATAL ", 255, __VA_ARGS__); }}
+#define mError(...) { if (mDebugFlag & DEBUG_ERROR) { taosPrintLog("MND ERROR ", 255, __VA_ARGS__); }}
+#define mWarn(...)  { if (mDebugFlag & DEBUG_WARN)  { taosPrintLog("MND WARN ", 255, __VA_ARGS__); }}
+#define mInfo(...)  { if (mDebugFlag & DEBUG_INFO)  { taosPrintLog("MND ", 255, __VA_ARGS__); }}
+#define mDebug(...) { if (mDebugFlag & DEBUG_DEBUG) { taosPrintLog("MND ", mDebugFlag, __VA_ARGS__); }}
+#define mTrace(...) { if (mDebugFlag & DEBUG_TRACE) { taosPrintLog("MND ", mDebugFlag, __VA_ARGS__); }}
+
+#define SDB_MAX_SIZE (32 * 1024)
+
+typedef struct SSdbRaw {
+  int8_t  sdb;
+  int8_t  sver;
+  int8_t  status;
+  int8_t  reserved;
+  int32_t dataLen;
+  char    pData[];
+} SSdbRaw;
+
+typedef struct SSdbRow {
+  ESdbType   sdb;
+  ESdbStatus status;
+  int32_t    refCount;
+  char       pObj[];
+} SSdbRow;
+
+typedef struct {
+  char       *currDir;
+  char       *syncDir;
+  char       *tmpDir;
+  int64_t     lastCommitVer;
+  int64_t     curVer;
+  EKeyType    keyTypes[SDB_MAX];
+  SHashObj   *hashObjs[SDB_MAX];
+  SRWLatch    locks[SDB_MAX];
+  SdbInsertFp insertFps[SDB_MAX];
+  SdbUpdateFp updateFps[SDB_MAX];
+  SdbDeleteFp deleteFps[SDB_MAX];
+  SdbDeployFp deployFps[SDB_MAX];
+  SdbEncodeFp encodeFps[SDB_MAX];
+  SdbDecodeFp decodeFps[SDB_MAX];
+} SSdbMgr;
+
+extern SSdbMgr tsSdb;
+
+int32_t sdbWriteImp(SSdbRaw *pRaw);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_SDB_INT_H_*/

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

@@ -0,0 +1,91 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "sdbInt.h"
+#include "tglobal.h"
+
+SSdbMgr tsSdb = {0};
+
+int32_t sdbInit() {
+  char path[PATH_MAX + 100];
+
+  snprintf(path, PATH_MAX + 100, "%s%scur%s", tsMnodeDir, TD_DIRSEP, TD_DIRSEP);
+  tsSdb.currDir = strdup(path);
+
+  snprintf(path, PATH_MAX + 100, "%s%ssync%s", tsMnodeDir, TD_DIRSEP, TD_DIRSEP);
+  tsSdb.syncDir = strdup(path);
+
+  snprintf(path, PATH_MAX + 100, "%s%stmp%s", tsMnodeDir, TD_DIRSEP, TD_DIRSEP);
+  tsSdb.tmpDir = strdup(path);
+
+  if (tsSdb.currDir == NULL || tsSdb.currDir == NULL || tsSdb.currDir == NULL) {
+    return TSDB_CODE_OUT_OF_MEMORY;
+  }
+
+  for (int32_t i = 0; i < SDB_MAX; ++i) {
+    int32_t type;
+    if (tsSdb.keyTypes[i] == SDB_KEY_INT32) {
+      type = TSDB_DATA_TYPE_INT;
+    } else if (tsSdb.keyTypes[i] == SDB_KEY_INT64) {
+      type = TSDB_DATA_TYPE_BIGINT;
+    } else {
+      type = TSDB_DATA_TYPE_BINARY;
+    }
+
+    SHashObj *hash = taosHashInit(64, taosGetDefaultHashFunction(type), true, HASH_NO_LOCK);
+    if (hash == NULL) {
+      return TSDB_CODE_OUT_OF_MEMORY;
+    }
+
+    tsSdb.hashObjs[i] = hash;
+    taosInitRWLatch(&tsSdb.locks[i]);
+  }
+
+  return 0;
+}
+
+void sdbCleanup() {
+  if (tsSdb.currDir != NULL) {
+    tfree(tsSdb.currDir);
+  }
+
+  if (tsSdb.syncDir != NULL) {
+    tfree(tsSdb.syncDir);
+  }
+
+  if (tsSdb.tmpDir != NULL) {
+    tfree(tsSdb.tmpDir);
+  }
+
+  for (int32_t i = 0; i < SDB_MAX; ++i) {
+    SHashObj *hash = tsSdb.hashObjs[i];
+    if (hash != NULL) {
+      taosHashCleanup(hash);
+    }
+    tsSdb.hashObjs[i] = NULL;
+  }
+}
+
+void sdbSetTable(SSdbTable table) {
+  ESdbType sdb = table.sdbType;
+  tsSdb.keyTypes[sdb] = table.keyType;
+  tsSdb.insertFps[sdb] = table.insertFp;
+  tsSdb.updateFps[sdb] = table.updateFp;
+  tsSdb.deleteFps[sdb] = table.deleteFp;
+  tsSdb.deployFps[sdb] = table.deployFp;
+  tsSdb.encodeFps[sdb] = table.encodeFp;
+  tsSdb.decodeFps[sdb] = table.decodeFp;
+}

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

@@ -0,0 +1,258 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "sdbInt.h"
+#include "tglobal.h"
+#include "tchecksum.h"
+
+static int32_t sdbCreateDir() {
+  mDebug("start to create mnode at %s", tsMnodeDir);
+  
+  if (!taosMkDir(tsSdb.currDir)) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    mError("failed to create dir:%s since %s", tsSdb.currDir, terrstr());
+    return -1;
+  }
+
+  if (!taosMkDir(tsSdb.syncDir)) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    mError("failed to create dir:%s since %s", tsSdb.syncDir, terrstr());
+    return -1;
+  }
+
+  if (!taosMkDir(tsSdb.tmpDir)) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    mError("failed to create dir:%s since %s", tsSdb.tmpDir, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t sdbRunDeployFp() {
+  mDebug("start to run deploy functions");
+
+  for (int32_t i = SDB_START; i < SDB_MAX; ++i) {
+    SdbDeployFp fp = tsSdb.deployFps[i];
+    if (fp == NULL) continue;
+    if ((*fp)() != 0) {
+      mError("failed to deploy sdb:%d since %s", i, terrstr());
+      return -1;
+    }
+  }
+
+  return 0;
+}
+
+static int32_t sdbReadDataFile() {
+  SSdbRaw *pRaw = malloc(SDB_MAX_SIZE);
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  char file[PATH_MAX] = {0};
+  snprintf(file, sizeof(file), "%ssdb.data", tsSdb.currDir);
+  FileFd fd = taosOpenFileRead(file);
+  if (fd <= 0) {
+    free(pRaw);
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    mError("failed to open file:%s for read since %s", file, terrstr());
+    return -1;
+  }
+
+  int64_t offset = 0;
+  int32_t code = 0;
+  int32_t readLen = 0;
+  int64_t ret = 0;
+
+  while (1) {
+    readLen = sizeof(SSdbRaw);
+    ret = taosReadFile(fd, pRaw, readLen);
+    if (ret == 0) break;
+
+    if (ret < 0) {
+      code = TAOS_SYSTEM_ERROR(errno);
+      mError("failed to read file:%s since %s", file, tstrerror(code));
+      break;
+    }
+
+    if (ret != readLen) {
+      code = TSDB_CODE_FILE_CORRUPTED;
+      mError("failed to read file:%s since %s", file, tstrerror(code));
+      break;
+    }
+
+    readLen = pRaw->dataLen + sizeof(int32_t);
+    ret = taosReadFile(fd, pRaw->pData, readLen);
+    if (ret < 0) {
+      code = TAOS_SYSTEM_ERROR(errno);
+      mError("failed to read file:%s since %s", file, tstrerror(code));
+      break;
+    }
+
+    if (ret != readLen) {
+      code = TSDB_CODE_FILE_CORRUPTED;
+      mError("failed to read file:%s since %s", file, tstrerror(code));
+      break;
+    }
+
+    int32_t totalLen = sizeof(SSdbRaw) + pRaw->dataLen + sizeof(int32_t);
+    if (!taosCheckChecksumWhole((const uint8_t *)pRaw, totalLen) != 0) {
+      code = TSDB_CODE_CHECKSUM_ERROR;
+      mError("failed to read file:%s since %s", file, tstrerror(code));
+      break;
+    }
+
+    code = sdbWriteImp(pRaw);
+    if (code != 0) {
+      mError("failed to read file:%s since %s", file, terrstr());
+      goto PARSE_SDB_DATA_ERROR;
+    }
+  }
+
+  code = 0;
+
+PARSE_SDB_DATA_ERROR:
+  taosCloseFile(fd);
+  sdbFreeRaw(pRaw);
+  terrno = code;
+  return code;
+}
+
+static int32_t sdbWriteDataFile() {
+  char tmpfile[PATH_MAX] = {0};
+  snprintf(tmpfile, sizeof(tmpfile), "%ssdb.data", tsSdb.tmpDir);
+
+  FileFd fd = taosOpenFileCreateWrite(tmpfile);
+  if (fd <= 0) {
+    terrno = TAOS_SYSTEM_ERROR(errno);
+    mError("failed to open file:%s for write since %s", tmpfile, terrstr());
+    return -1;
+  }
+
+  int32_t code = 0;
+
+  for (int32_t i = SDB_MAX - 1; i > SDB_START; --i) {
+    SdbEncodeFp encodeFp = tsSdb.encodeFps[i];
+    if (encodeFp == NULL) continue;
+
+    SHashObj *hash = tsSdb.hashObjs[i];
+    SRWLatch *pLock = &tsSdb.locks[i];
+    taosWLockLatch(pLock);
+
+    SSdbRow **ppRow = taosHashIterate(hash, NULL);
+    while (ppRow != NULL) {
+      SSdbRow *pRow = *ppRow;
+      if (pRow == NULL || pRow->status != SDB_STATUS_READY) {
+        ppRow = taosHashIterate(hash, ppRow);
+        continue;
+      }
+
+      SSdbRaw *pRaw = (*encodeFp)(pRow->pObj);
+      if (pRaw != NULL) {
+        pRaw->status = pRow->status;
+        int32_t writeLen = sizeof(SSdbRaw) + pRaw->dataLen;
+        if (taosWriteFile(fd, pRaw, writeLen) != writeLen) {
+          code = TAOS_SYSTEM_ERROR(terrno);
+          taosHashCancelIterate(hash, ppRow);
+          break;
+        }
+
+        int32_t cksum = taosCalcChecksum(0, (const uint8_t *)pRaw, sizeof(SSdbRaw) + pRaw->dataLen);
+        if (taosWriteFile(fd, &cksum, sizeof(int32_t)) != sizeof(int32_t)) {
+          code = TAOS_SYSTEM_ERROR(terrno);
+          taosHashCancelIterate(hash, ppRow);
+          break;
+        }
+      } else {
+        code = TSDB_CODE_SDB_APP_ERROR;
+        taosHashCancelIterate(hash, ppRow);
+        break;
+      }
+
+      ppRow = taosHashIterate(hash, ppRow);
+    }
+    taosWUnLockLatch(pLock);
+  }
+
+  if (code == 0) {
+    code = taosFsyncFile(fd);
+  }
+
+  taosCloseFile(fd);
+
+  if (code == 0) {
+    char curfile[PATH_MAX] = {0};
+    snprintf(curfile, sizeof(curfile), "%ssdb.data", tsSdb.currDir);
+    code = taosRenameFile(tmpfile, curfile);
+  }
+
+  if (code != 0) {
+    terrno = code;
+    mError("failed to write sdb file since %s", terrstr());
+  } else {
+    mDebug("write sdb file successfully");
+  }
+
+  return code;
+}
+
+int32_t sdbOpen() {
+  mDebug("start to read mnode file");
+
+  if (sdbReadDataFile() != 0) {
+    return -1;
+  }
+
+  return 0;
+}
+
+void sdbClose() {
+  if (tsSdb.curVer != tsSdb.lastCommitVer) {
+    mDebug("start to write mnode file");
+    sdbWriteDataFile();
+  }
+
+  for (int32_t i = 0; i < SDB_MAX; ++i) {
+    SHashObj *hash = tsSdb.hashObjs[i];
+    if (hash != NULL) {
+      taosHashClear(hash);
+    }
+  }
+}
+
+int32_t sdbDeploy() {
+  if (sdbCreateDir() != 0) {
+    return -1;
+  }
+
+  if (sdbRunDeployFp() != 0) {
+    return -1;
+  }
+
+  if (sdbWriteDataFile() != 0) {
+    return -1;
+  }
+
+  sdbClose();
+  return 0;
+}
+
+void sdbUnDeploy() {
+  mDebug("start to undeploy mnode");
+  taosRemoveDir(tsMnodeDir);
+}

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

@@ -0,0 +1,280 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "sdbInt.h"
+#include "tglobal.h"
+
+static SHashObj *sdbGetHash(int32_t sdb) {
+  if (sdb >= SDB_MAX || sdb <= SDB_START) {
+    terrno = TSDB_CODE_SDB_INVALID_TABLE_TYPE;
+    return NULL;
+  }
+
+  SHashObj *hash = tsSdb.hashObjs[sdb];
+  if (hash == NULL) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    return NULL;
+  }
+
+  return hash;
+}
+
+static int32_t sdbGetkeySize(ESdbType sdb, void *pKey) {
+  int32_t  keySize;
+  EKeyType keyType = tsSdb.keyTypes[sdb];
+
+  if (keyType == SDB_KEY_INT32) {
+    keySize = sizeof(int32_t);
+  } else if (keyType == SDB_KEY_BINARY) {
+    keySize = strlen(pKey) + 1;
+  } else {
+    keySize = sizeof(int64_t);
+  }
+
+  return keySize;
+}
+
+static int32_t sdbInsertRow(SHashObj *hash, SSdbRaw *pRaw, SSdbRow *pRow, int32_t keySize) {
+  SRWLatch *pLock = &tsSdb.locks[pRow->sdb];
+  taosWLockLatch(pLock);
+
+  SSdbRow *pDstRow = taosHashGet(hash, pRow->pObj, keySize);
+  if (pDstRow != NULL) {
+    terrno = TSDB_CODE_SDB_OBJ_ALREADY_THERE;
+    taosWUnLockLatch(pLock);
+    sdbFreeRow(pRow);
+    return -1;
+  }
+
+  pRow->refCount = 1;
+  pRow->status = pRaw->status;
+
+  if (taosHashPut(hash, pRow->pObj, keySize, &pRow, sizeof(void *)) != 0) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    taosWUnLockLatch(pLock);
+    sdbFreeRow(pRow);
+    return -1;
+  }
+
+  taosWUnLockLatch(pLock);
+
+  SdbInsertFp insertFp = tsSdb.insertFps[pRow->sdb];
+  if (insertFp != NULL) {
+    if ((*insertFp)(pRow->pObj) != 0) {
+      taosWLockLatch(pLock);
+      taosHashRemove(hash, pRow->pObj, keySize);
+      taosWUnLockLatch(pLock);
+      sdbFreeRow(pRow);
+      return -1;
+    }
+  }
+
+  return 0;
+}
+
+static int32_t sdbUpdateRow(SHashObj *hash, SSdbRaw *pRaw, SSdbRow *pRow, int32_t keySize) {
+  SRWLatch *pLock = &tsSdb.locks[pRow->sdb];
+  taosRLockLatch(pLock);
+
+  SSdbRow **ppDstRow = taosHashGet(hash, pRow->pObj, keySize);
+  if (ppDstRow == NULL || *ppDstRow == NULL) {
+    taosRUnLockLatch(pLock);
+    return sdbInsertRow(hash, pRaw, pRow, keySize);
+  }
+  SSdbRow *pDstRow = *ppDstRow;
+
+  pRow->status = pRaw->status;
+  taosRUnLockLatch(pLock);
+
+  SdbUpdateFp updateFp = tsSdb.updateFps[pRow->sdb];
+  if (updateFp != NULL) {
+    (*updateFp)(pRow->pObj, pDstRow->pObj);
+  }
+
+  sdbFreeRow(pRow);
+  return 0;
+}
+
+static int32_t sdbDeleteRow(SHashObj *hash, SSdbRaw *pRaw, SSdbRow *pRow, int32_t keySize) {
+  SRWLatch *pLock = &tsSdb.locks[pRow->sdb];
+  taosWLockLatch(pLock);
+
+  SSdbRow **ppDstRow = taosHashGet(hash, pRow->pObj, keySize);
+  if (ppDstRow == NULL || *ppDstRow == NULL) {
+    terrno = TSDB_CODE_SDB_OBJ_NOT_THERE;
+    taosWUnLockLatch(pLock);
+    sdbFreeRow(pRow);
+    return -1;
+  }
+  SSdbRow *pDstRow = *ppDstRow;
+
+  pDstRow->status = pRaw->status;
+  taosHashRemove(hash, pDstRow->pObj, keySize);
+  taosWUnLockLatch(pLock);
+
+  SdbDeleteFp deleteFp = tsSdb.deleteFps[pDstRow->sdb];
+  if (deleteFp != NULL) {
+    (void)(*deleteFp)(pDstRow->pObj);
+  }
+
+  sdbRelease(pDstRow->pObj);
+  sdbFreeRow(pRow);
+  return 0;
+}
+
+int32_t sdbWriteImp(SSdbRaw *pRaw) {
+  SHashObj *hash = sdbGetHash(pRaw->sdb);
+  if (hash == NULL) return -1;
+
+  SdbDecodeFp decodeFp = tsSdb.decodeFps[pRaw->sdb];
+  SSdbRow    *pRow = (*decodeFp)(pRaw);
+  if (pRow == NULL) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_CONTENT;
+    return -1;
+  }
+
+  pRow->sdb = pRaw->sdb;
+
+  int32_t keySize = sdbGetkeySize(pRow->sdb, pRow->pObj);
+  int32_t code = -1;
+
+  switch (pRaw->status) {
+    case SDB_STATUS_CREATING:
+      code = sdbInsertRow(hash, pRaw, pRow, keySize);
+      break;
+    case SDB_STATUS_READY:
+    case SDB_STATUS_DROPPING:
+      code = sdbUpdateRow(hash, pRaw, pRow, keySize);
+      break;
+    case SDB_STATUS_DROPPED:
+      code = sdbDeleteRow(hash, pRaw, pRow, keySize);
+      break;
+    default:
+      terrno = TSDB_CODE_SDB_INVALID_ACTION_TYPE;
+      break;
+  }
+
+  return code;
+}
+
+int32_t sdbWrite(SSdbRaw *pRaw) {
+  int32_t code = sdbWriteImp(pRaw);
+  sdbFreeRaw(pRaw);
+  return code;
+}
+
+void *sdbAcquire(ESdbType sdb, void *pKey) {
+  SHashObj *hash = sdbGetHash(sdb);
+  if (hash == NULL) return NULL;
+
+  void   *pRet = NULL;
+  int32_t keySize = sdbGetkeySize(sdb, pKey);
+
+  SRWLatch *pLock = &tsSdb.locks[sdb];
+  taosRLockLatch(pLock);
+
+  SSdbRow **ppRow = taosHashGet(hash, pKey, keySize);
+  if (ppRow == NULL || *ppRow == NULL) {
+    terrno = TSDB_CODE_SDB_OBJ_NOT_THERE;
+    taosRUnLockLatch(pLock);
+    return NULL;
+  }
+
+  SSdbRow *pRow = *ppRow;
+  switch (pRow->status) {
+    case SDB_STATUS_READY:
+      atomic_add_fetch_32(&pRow->refCount, 1);
+      pRet = pRow->pObj;
+      break;
+    case SDB_STATUS_CREATING:
+      terrno = TSDB_CODE_SDB_OBJ_CREATING;
+      break;
+    case SDB_STATUS_DROPPING:
+      terrno = TSDB_CODE_SDB_OBJ_DROPPING;
+      break;
+    default:
+      terrno = TSDB_CODE_SDB_APP_ERROR;
+      break;
+  }
+
+  taosRUnLockLatch(pLock);
+  return pRet;
+}
+
+void sdbRelease(void *pObj) {
+  if (pObj == NULL) return;
+
+  SSdbRow *pRow = (SSdbRow *)((char *)pObj - sizeof(SSdbRow));
+  if (pRow->sdb >= SDB_MAX || pRow->sdb <= SDB_START) return;
+
+  SRWLatch *pLock = &tsSdb.locks[pRow->sdb];
+  taosRLockLatch(pLock);
+
+  int32_t ref = atomic_sub_fetch_32(&pRow->refCount, 1);
+  if (ref <= 0 && pRow->status == SDB_STATUS_DROPPED) {
+    sdbFreeRow(pRow);
+  }
+
+  taosRUnLockLatch(pLock);
+}
+
+void *sdbFetch(ESdbType sdb, void *pIter, void **ppObj) {
+  SHashObj *hash = sdbGetHash(sdb);
+  if (hash == NULL) return NULL;
+
+  SRWLatch *pLock = &tsSdb.locks[sdb];
+  taosRLockLatch(pLock);
+
+  SSdbRow **ppRow = taosHashIterate(hash, ppRow);
+  while (ppRow != NULL) {
+    SSdbRow *pRow = *ppRow;
+    if (pRow == NULL || pRow->status != SDB_STATUS_READY) {
+      ppRow = taosHashIterate(hash, ppRow);
+      continue;
+    }
+
+    atomic_add_fetch_32(&pRow->refCount, 1);
+    *ppObj = pRow->pObj;
+    break;
+  }
+  taosRUnLockLatch(pLock);
+
+  return ppRow;
+}
+
+void sdbCancelFetch(void *pIter) {
+  if (pIter == NULL) return;
+  SSdbRow  *pRow = *(SSdbRow **)pIter;
+  SHashObj *hash = sdbGetHash(pRow->sdb);
+  if (hash == NULL) return;
+
+  SRWLatch *pLock = &tsSdb.locks[pRow->sdb];
+  taosRLockLatch(pLock);
+  taosHashCancelIterate(hash, pIter);
+  taosRUnLockLatch(pLock);
+}
+
+int32_t sdbGetSize(ESdbType sdb) {
+  SHashObj *hash = sdbGetHash(sdb);
+  if (hash == NULL) return 0;
+
+  SRWLatch *pLock = &tsSdb.locks[sdb];
+  taosRLockLatch(pLock);
+  int32_t size = taosHashGetSize(hash);
+  taosRUnLockLatch(pLock);
+
+  return size;
+}

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

@@ -0,0 +1,196 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "sdbInt.h"
+
+SSdbRaw *sdbAllocRaw(ESdbType sdb, int8_t sver, int32_t dataLen) {
+  SSdbRaw *pRaw = calloc(1, dataLen + sizeof(SSdbRaw));
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  pRaw->sdb = sdb;
+  pRaw->sver = sver;
+  pRaw->dataLen = dataLen;
+  return pRaw;
+}
+
+void sdbFreeRaw(SSdbRaw *pRaw) { free(pRaw); }
+
+int32_t sdbSetRawInt8(SSdbRaw *pRaw, int32_t dataPos, int8_t val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int8_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *(int8_t *)(pRaw->pData + dataPos) = val;
+  return 0;
+}
+
+int32_t sdbSetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int32_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *(int32_t *)(pRaw->pData + dataPos) = val;
+  return 0;
+}
+
+int32_t sdbSetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int64_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *(int64_t *)(pRaw->pData + dataPos) = val;
+  return 0;
+}
+
+int32_t sdbSetRawBinary(SSdbRaw *pRaw, int32_t dataPos, const char *pVal, int32_t valLen) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + valLen > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  memcpy(pRaw->pData + dataPos, pVal, valLen);
+  return 0;
+}
+
+int32_t sdbSetRawDataLen(SSdbRaw *pRaw, int32_t dataLen) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataLen > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  pRaw->dataLen = dataLen;
+  return 0;
+}
+
+int32_t sdbSetRawStatus(SSdbRaw *pRaw, ESdbStatus status) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  pRaw->status = status;
+  return 0;
+}
+
+int32_t sdbGetRawInt8(SSdbRaw *pRaw, int32_t dataPos, int8_t *val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int8_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *val = *(int8_t *)(pRaw->pData + dataPos);
+  return 0;
+}
+
+int32_t sdbGetRawInt32(SSdbRaw *pRaw, int32_t dataPos, int32_t *val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int32_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *val = *(int32_t *)(pRaw->pData + dataPos);
+  return 0;
+}
+
+int32_t sdbGetRawInt64(SSdbRaw *pRaw, int32_t dataPos, int64_t *val) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + sizeof(int64_t) > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  *val = *(int64_t *)(pRaw->pData + dataPos);
+  return 0;
+}
+
+int32_t sdbGetRawBinary(SSdbRaw *pRaw, int32_t dataPos, char *pVal, int32_t valLen) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  if (dataPos + valLen > pRaw->dataLen) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_LEN;
+    return -1;
+  }
+
+  memcpy(pVal, pRaw->pData + dataPos, valLen);
+  return 0;
+}
+
+int32_t sdbGetRawSoftVer(SSdbRaw *pRaw, int8_t *sver) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  *sver = pRaw->sver;
+  return 0;
+}
+
+int32_t sdbGetRawTotalSize(SSdbRaw *pRaw) {
+  if (pRaw == NULL) {
+    terrno = TSDB_CODE_INVALID_PTR;
+    return -1;
+  }
+
+  return sizeof(SSdbRaw) + pRaw->dataLen;
+}

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

@@ -0,0 +1,38 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "sdbInt.h"
+
+SSdbRow *sdbAllocRow(int32_t objSize) {
+  SSdbRow *pRow = calloc(1, objSize + sizeof(SSdbRow));
+  if (pRow == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  return pRow;
+}
+
+void *sdbGetRowObj(SSdbRow *pRow) {
+  if (pRow == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  return pRow->pObj;
+}
+
+void sdbFreeRow(SSdbRow *pRow) { free(pRow); }

source/dnode/mnode/src/mnodeAcct.c

@@ -1,148 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "os.h"
-#include "mnodeSdb.h"
-
-static void mnodeCreateDefaultAcct() {
-  int32_t code = TSDB_CODE_SUCCESS;
-
-  SAcctObj acctObj = {0};
-  tstrncpy(acctObj.acct, TSDB_DEFAULT_USER, TSDB_USER_LEN);
-  acctObj.cfg = (SAcctCfg){.maxUsers = 128,
-                           .maxDbs = 128,
-                           .maxTimeSeries = INT32_MAX,
-                           .maxStreams = 1000,
-                           .maxStorage = INT64_MAX,
-                           .accessState = TSDB_VN_ALL_ACCCESS};
-  acctObj.acctId = 1;
-  acctObj.createdTime = taosGetTimestampMs();
-  acctObj.updateTime = taosGetTimestampMs();
-
-  sdbInsertRow(MN_SDB_ACCT, &acctObj);
-}
-
-int32_t mnodeEncodeAcct(SAcctObj *pAcct, char *buf, int32_t maxLen) {
-  int32_t len = 0;
-
-  len += snprintf(buf + len, maxLen - len, "{\"type\":%d, ", MN_SDB_ACCT);
-  len += snprintf(buf + len, maxLen - len, "\"acct\":\"%s\", ", pAcct->acct);
-  len += snprintf(buf + len, maxLen - len, "\"acctId\":\"%d\", ", pAcct->acctId);
-  len += snprintf(buf + len, maxLen - len, "\"maxUsers\":\"%d\", ", pAcct->cfg.maxUsers);
-  len += snprintf(buf + len, maxLen - len, "\"maxDbs\":\"%d\", ", pAcct->cfg.maxDbs);
-  len += snprintf(buf + len, maxLen - len, "\"maxTimeSeries\":\"%d\", ", pAcct->cfg.maxTimeSeries);
-  len += snprintf(buf + len, maxLen - len, "\"maxStreams\":\"%d\", ", pAcct->cfg.maxStreams);
-  len += snprintf(buf + len, maxLen - len, "\"maxStorage\":\"%" PRIu64 "\", ", pAcct->cfg.maxStorage);
-  len += snprintf(buf + len, maxLen - len, "\"accessState\":\"%d\", ", pAcct->cfg.accessState);
-  len += snprintf(buf + len, maxLen - len, "\"createdTime\":\"%" PRIu64 "\", ", pAcct->createdTime);
-  len += snprintf(buf + len, maxLen - len, "\"updateTime\":\"%" PRIu64 "\"}\n", pAcct->updateTime);
-
-  return len;
-}
-
-SAcctObj *mnodeDecodeAcct(cJSON *root) {
-  int32_t   code = -1;
-  SAcctObj *pAcct = calloc(1, sizeof(SAcctObj));
-
-  cJSON *acct = cJSON_GetObjectItem(root, "acct");
-  if (!acct || acct->type != cJSON_String) {
-    mError("failed to parse acct since acct not found");
-    goto DECODE_ACCT_OVER;
-  }
-  tstrncpy(pAcct->acct, acct->valuestring, TSDB_USER_LEN);
-
-  cJSON *acctId = cJSON_GetObjectItem(root, "acctId");
-  if (!acctId || acctId->type != cJSON_String) {
-    mError("acct:%s, failed to parse since acctId not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->acctId = atol(acctId->valuestring);
-
-  cJSON *maxUsers = cJSON_GetObjectItem(root, "maxUsers");
-  if (!maxUsers || maxUsers->type != cJSON_String) {
-    mError("acct:%s, failed to parse since maxUsers not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->cfg.maxUsers = atol(maxUsers->valuestring);
-
-  cJSON *maxDbs = cJSON_GetObjectItem(root, "maxDbs");
-  if (!maxDbs || maxDbs->type != cJSON_String) {
-    mError("acct:%s, failed to parse since maxDbs not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->cfg.maxDbs = atol(maxDbs->valuestring);
-
-  cJSON *maxTimeSeries = cJSON_GetObjectItem(root, "maxTimeSeries");
-  if (!maxTimeSeries || maxTimeSeries->type != cJSON_String) {
-    mError("acct:%s, failed to parse since maxTimeSeries not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->cfg.maxTimeSeries = atol(maxTimeSeries->valuestring);
-
-  cJSON *maxStreams = cJSON_GetObjectItem(root, "maxStreams");
-  if (!maxStreams || maxStreams->type != cJSON_String) {
-    mError("acct:%s, failed to parse since maxStreams not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->cfg.maxStreams = atol(maxStreams->valuestring);
-
-  cJSON *maxStorage = cJSON_GetObjectItem(root, "maxStorage");
-  if (!maxStorage || maxStorage->type != cJSON_String) {
-    mError("acct:%s, failed to parse since maxStorage not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->cfg.maxStorage = atoll(maxStorage->valuestring);
-
-  cJSON *accessState = cJSON_GetObjectItem(root, "accessState");
-  if (!accessState || accessState->type != cJSON_String) {
-    mError("acct:%s, failed to parse since accessState not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->cfg.accessState = atol(accessState->valuestring);
-
-  cJSON *createdTime = cJSON_GetObjectItem(root, "createdTime");
-  if (!createdTime || createdTime->type != cJSON_String) {
-    mError("acct:%s, failed to parse since createdTime not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->createdTime = atol(createdTime->valuestring);
-
-  cJSON *updateTime = cJSON_GetObjectItem(root, "updateTime");
-  if (!updateTime || updateTime->type != cJSON_String) {
-    mError("acct:%s, failed to parse since updateTime not found", pAcct->acct);
-    goto DECODE_ACCT_OVER;
-  }
-  pAcct->updateTime = atol(updateTime->valuestring);
-
-  code = 0;
-  mTrace("acct:%s, parse success", pAcct->acct);
-
-DECODE_ACCT_OVER:
-  if (code != 0) {
-    free(pAcct);
-    pAcct = NULL;
-  }
-  return pAcct;
-}
-
-int32_t mnodeInitAcct() {
-  sdbSetFp(MN_SDB_ACCT, MN_KEY_BINARY, mnodeCreateDefaultAcct, (SdbEncodeFp)mnodeEncodeAcct,
-           (SdbDecodeFp)(mnodeDecodeAcct), sizeof(SAcctObj));
-
-  return 0;
-}
-
-void mnodeCleanupAcct() {}

source/dnode/mnode/src/mnodeSdb.c

@@ -1,395 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "os.h"
-#include "thash.h"
-#include "tglobal.h"
-#include "cJSON.h"
-#include "mnodeSdb.h"
-
-static struct {
-  char        currDir[PATH_MAX];
-  char        backDir[PATH_MAX];
-  char        tmpDir[PATH_MAX];
-  int64_t     version;
-  EMnKey      hashKey[MN_SDB_MAX];
-  int32_t     dataSize[MN_SDB_MAX];
-  SHashObj   *hashObj[MN_SDB_MAX];
-  SdbDeployFp deployFp[MN_SDB_MAX];
-  SdbEncodeFp encodeFp[MN_SDB_MAX];
-  SdbDecodeFp decodeFp[MN_SDB_MAX];
-} tsSdb = {0};
-
-static int32_t sdbCreateDir() {
-  if (!taosMkDir(tsSdb.currDir)) {
-    mError("failed to create dir:%s", tsSdb.currDir);
-    return TAOS_SYSTEM_ERROR(errno);
-  }
-
-  if (!taosMkDir(tsSdb.backDir)) {
-    mError("failed to create dir:%s", tsSdb.backDir);
-    return -1;
-  }
-
-  if (!taosMkDir(tsSdb.tmpDir)) {
-    mError("failed to create dir:%s", tsSdb.tmpDir);
-    return -1;
-  }
-
-  return 0;
-}
-
-static int32_t sdbRunDeployFp() {
-  for (int32_t i = MN_SDB_START; i < MN_SDB_MAX; ++i) {
-    SdbDeployFp fp = tsSdb.deployFp[i];
-    if (fp) {
-      (*fp)();
-    }
-  }
-
-  return 0;
-}
-
-static int32_t sdbReadVersion(cJSON *root) {
-  cJSON *ver = cJSON_GetObjectItem(root, "version");
-  if (!ver || ver->type != cJSON_String) {
-    mError("failed to parse version since version not found");
-    return -1;
-  }
-
-  tsSdb.version = (int64_t)atoll(ver->valuestring);
-  mTrace("parse version success, version:%" PRIu64, tsSdb.version);
-
-  return 0;
-}
-
-static void sdbWriteVersion(FileFd fd) {
-  char    content[128];
-  int32_t len =
-      snprintf(content, sizeof(content), "{\"type\":0, \"version\":\"%" PRIu64 "\", \"updateTime\":\"%" PRIu64 "\"}\n",
-               tsSdb.version, taosGetTimestampMs());
-  taosWriteFile(fd, content, len);
-}
-
-static int32_t sdbReadDataFile() {
-  ssize_t _bytes = 0;
-  size_t  len = 4096;
-  char   *line = calloc(1, len);
-  int32_t code = -1;
-  FILE   *fp = NULL;
-  cJSON  *root = NULL;
-
-  char file[PATH_MAX + 20];
-  snprintf(file, sizeof(file), "%ssdb.data", tsSdb.currDir);
-  fp = fopen(file, "r");
-  if (!fp) {
-    mDebug("failed to open file:%s for read since %s", file, strerror(errno));
-    goto PARSE_SDB_DATA_ERROR;
-  }
-
-  while (!feof(fp)) {
-    memset(line, 0, len);
-    _bytes = tgetline(&line, &len, fp);
-    if (_bytes < 0) {
-      break;
-    }
-
-    line[len - 1] = 0;
-    if (len <= 10) continue;
-
-    root = cJSON_Parse(line);
-    if (root == NULL) {
-      mError("failed to parse since invalid json format, %s", line);
-      goto PARSE_SDB_DATA_ERROR;
-    }
-
-    cJSON *type = cJSON_GetObjectItem(root, "type");
-    if (!type || type->type != cJSON_Number) {
-      mError("failed to parse since invalid type not found, %s", line);
-      goto PARSE_SDB_DATA_ERROR;
-    }
-
-    if (type->valueint >= MN_SDB_MAX || type->valueint < MN_SDB_START) {
-      mError("failed to parse since invalid type, %s", line);
-      goto PARSE_SDB_DATA_ERROR;
-    }
-
-    if (type->valueint == MN_SDB_START) {
-      if (sdbReadVersion(root) != 0) {
-        mError("failed to parse version, %s", line);
-        goto PARSE_SDB_DATA_ERROR;
-      }
-      cJSON_Delete(root);
-      root = NULL;
-      continue;
-    }
-
-    SdbDecodeFp decodeFp = tsSdb.decodeFp[type->valueint];
-    SdbHead    *pHead = (*decodeFp)(root);
-    if (pHead == NULL) {
-      mError("failed to parse since decode error, %s", line);
-      goto PARSE_SDB_DATA_ERROR;
-    }
-
-    pHead->type = type->valueint;
-    pHead->status = MN_SDB_STAT_AVAIL;
-
-    sdbInsertRow(pHead->type, pHead);
-    free(pHead);
-    cJSON_Delete(root);
-    root = NULL;
-  }
-
-  code = 0;
-
-PARSE_SDB_DATA_ERROR:
-  if (line) free(line);
-  if (fp) fclose(fp);
-  if (root) cJSON_Delete(root);
-
-  return code;
-}
-
-static int32_t sdbWriteDataFile() {
-  char file[PATH_MAX + 20] = {0};
-  snprintf(file, sizeof(file), "%ssdb.data", tsSdb.currDir);
-  FileFd fd = taosOpenFileCreateWrite(file);
-  if (fd <= 0) {
-    mError("failed to open file:%s for write since %s", file, strerror(errno));
-    return -1;
-  }
-
-  int32_t len;
-  int32_t maxLen = 10240;
-  char   *buf = malloc(maxLen);
-
-  for (int32_t i = MN_SDB_START; i < MN_SDB_MAX; ++i) {
-    SHashObj *hash = tsSdb.hashObj[i];
-    if (!hash) continue;
-
-    SdbEncodeFp encodeFp = tsSdb.encodeFp[i];
-    if (!encodeFp) continue;
-
-    SdbHead *pHead = taosHashIterate(hash, NULL);
-    while (pHead != NULL) {
-      len = (*encodeFp)(pHead, buf, maxLen);
-      if (len >= 0) {
-        taosWriteFile(fd, buf, len);
-      }
-
-      pHead = taosHashIterate(hash, pHead);
-    }
-  }
-
-  sdbWriteVersion(fd);
-  taosFsyncFile(fd);
-  taosCloseFile(fd);
-
-  mInfo("write file:%s successfully", file);
-  return 0;
-}
-
-int32_t sdbCommit() {
-  int32_t code = sdbWriteDataFile();
-  if (code != 0) {
-    return code;
-  }
-
-  return 0;
-}
-
-int32_t sdbRead() {
-  int32_t code = sdbReadDataFile();
-  if (code != 0) {
-    return code;
-  }
-
-  mInfo("read sdb file successfully");
-  return -1;
-}
-
-int32_t sdbDeploy() {
-  if (sdbCreateDir() != 0) {
-    return -1;
-  }
-
-  if (sdbRunDeployFp() != 0) {
-    return -1;
-  }
-
-  if (sdbCommit() != 0) {
-    return -1;
-  }
-
-  // if (!taosMkDir())
-  // if (pMinfos == NULL) {  // first deploy
-  //   tsMint.dnodeId = 1;
-  //   bool getuid = taosGetSystemUid(tsMint.clusterId);
-  //   if (!getuid) {
-  //     strcpy(tsMint.clusterId, "tdengine3.0");
-  //     mError("deploy new mnode but failed to get uid, set to default val %s", tsMint.clusterId);
-  //   } else {
-  //     mDebug("deploy new mnode and uid is %s", tsMint.clusterId);
-  //   }
-  // } else {  // todo
-  // }
-
-  // if (mkdir(tsMnodeDir, 0755) != 0 && errno != EEXIST) {
-  //   mError("failed to init mnode dir:%s, reason:%s", tsMnodeDir, strerror(errno));
-  //   return -1;
-  // }
-  return 0;
-}
-
-void sdbUnDeploy() {}
-
-int32_t sdbInit() {
-  snprintf(tsSdb.currDir, PATH_MAX, "%s%scurrent%s", tsMnodeDir, TD_DIRSEP, TD_DIRSEP);
-  snprintf(tsSdb.backDir, PATH_MAX, "%s%sbackup%s", tsMnodeDir, TD_DIRSEP, TD_DIRSEP);
-  snprintf(tsSdb.tmpDir, PATH_MAX, "%s%stmp%s", tsMnodeDir, TD_DIRSEP, TD_DIRSEP);
-
-  for (int32_t i = 0; i < MN_SDB_MAX; ++i) {
-    int32_t type;
-    if (tsSdb.hashKey[i] == MN_KEY_INT32) {
-      type = TSDB_DATA_TYPE_INT;
-    } else if (tsSdb.hashKey[i] == MN_KEY_INT64) {
-      type = TSDB_DATA_TYPE_BIGINT;
-    } else {
-      type = TSDB_DATA_TYPE_BINARY;
-    }
-
-    SHashObj *hash = taosHashInit(128, taosGetDefaultHashFunction(type), true, HASH_NO_LOCK);
-    if (hash == NULL) {
-      return -1;
-    }
-
-    tsSdb.hashObj[i] = hash;
-  }
-
-  return 0;
-}
-
-void sdbCleanup() {
-  for (int32_t i = 0; i < MN_SDB_MAX; ++i) {
-    SHashObj *hash = tsSdb.hashObj[i];
-    if (hash != NULL) {
-      taosHashCleanup(hash);
-    }
-    tsSdb.hashObj[i] = NULL;
-  }
-}
-
-void sdbSetFp(EMnSdb sdb, EMnKey keyType, SdbDeployFp deployFp, SdbEncodeFp encodeFp, SdbDecodeFp decodeFp,
-              int32_t dataSize) {
-  tsSdb.deployFp[sdb] = deployFp;
-  tsSdb.encodeFp[sdb] = encodeFp;
-  tsSdb.decodeFp[sdb] = decodeFp;
-  tsSdb.dataSize[sdb] = dataSize;
-  tsSdb.hashKey[sdb] = keyType;
-}
-
-static SHashObj *sdbGetHash(int32_t sdb) {
-  if (sdb >= MN_SDB_MAX || sdb <= MN_SDB_START) {
-    return NULL;
-  }
-
-  SHashObj *hash = tsSdb.hashObj[sdb];
-  if (hash == NULL) {
-    return NULL;
-  }
-
-  return hash;
-}
-
-void *sdbInsertRow(EMnSdb sdb, void *p) {
-  SdbHead *pHead = p;
-  pHead->type = sdb;
-  pHead->status = MN_SDB_STAT_AVAIL;
-
-  char   *pKey = (char *)pHead + sizeof(pHead);
-  int32_t keySize;
-  EMnKey  keyType = tsSdb.hashKey[pHead->type];
-  int32_t dataSize = tsSdb.dataSize[pHead->type];
-
-  SHashObj *hash = sdbGetHash(pHead->type);
-  if (hash == NULL) {
-    return NULL;
-  }
-
-  if (keyType == MN_KEY_INT32) {
-    keySize = sizeof(int32_t);
-  } else if (keyType == MN_KEY_BINARY) {
-    keySize = strlen(pKey) + 1;
-  } else {
-    keySize = sizeof(int64_t);
-  }
-
-  taosHashPut(hash, pKey, keySize, pHead, dataSize);
-  return taosHashGet(hash, pKey, keySize);
-}
-
-void sdbDeleteRow(EMnSdb sdb, void *p) {
-  SdbHead *pHead = p;
-  pHead->status = MN_SDB_STAT_DROPPED;
-}
-
-void *sdbUpdateRow(EMnSdb sdb, void *pHead) { return sdbInsertRow(sdb, pHead); }
-
-void *sdbGetRow(EMnSdb sdb, void *pKey) {
-  SHashObj *hash = sdbGetHash(sdb);
-  if (hash == NULL) {
-    return NULL;
-  }
-
-  int32_t keySize;
-  EMnKey  keyType = tsSdb.hashKey[sdb];
-
-  if (keyType == MN_KEY_INT32) {
-    keySize = sizeof(int32_t);
-  } else if (keyType == MN_KEY_BINARY) {
-    keySize = strlen(pKey) + 1;
-  } else {
-    keySize = sizeof(int64_t);
-  }
-
-  return taosHashGet(hash, pKey, keySize);
-}
-
-void *sdbFetchRow(EMnSdb sdb, void *pIter) {
-  SHashObj *hash = sdbGetHash(sdb);
-  if (hash == NULL) {
-    return NULL;
-  }
-
-  return taosHashIterate(hash, pIter);
-}
-
-void sdbCancelFetch(EMnSdb sdb, void *pIter) {
-  SHashObj *hash = sdbGetHash(sdb);
-  if (hash == NULL) {
-    return;
-  }
-
-  taosHashCancelIterate(hash, pIter);
-}
-
-int32_t sdbGetCount(EMnSdb sdb) {
-  SHashObj *hash = sdbGetHash(sdb);
-  if (hash == NULL) {
-    return 0;
-  }
-  return taosHashGetSize(hash);
-}

source/dnode/mnode/src/mnodeUser.c

@@ -1,130 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "os.h"
-#include "tkey.h"
-#include "tglobal.h"
-#include "mnodeSdb.h"
-
-static int32_t mnodeCreateDefaultUser(char *acct, char *user, char *pass) {
-  int32_t code = TSDB_CODE_SUCCESS;
-
-  SUserObj userObj = {0};
-  tstrncpy(userObj.user, user, TSDB_USER_LEN);
-  tstrncpy(userObj.acct, acct, TSDB_USER_LEN);
-  taosEncryptPass((uint8_t *)pass, strlen(pass), userObj.pass);
-  userObj.createdTime = taosGetTimestampMs();
-  userObj.updateTime = taosGetTimestampMs();
-
-  if (strcmp(user, TSDB_DEFAULT_USER) == 0) {
-    userObj.rootAuth = 1;
-  }
-
-  sdbInsertRow(MN_SDB_USER, &userObj);
-}
-
-static void mnodeCreateDefaultUsers() {
-  mnodeCreateDefaultUser(TSDB_DEFAULT_USER, TSDB_DEFAULT_USER, TSDB_DEFAULT_PASS);
-  mnodeCreateDefaultUser(TSDB_DEFAULT_USER, "monitor", tsInternalPass);
-  mnodeCreateDefaultUser(TSDB_DEFAULT_USER, "_" TSDB_DEFAULT_USER, tsInternalPass);
-}
-
-int32_t mnodeEncodeUser(SUserObj *pUser, char *buf, int32_t maxLen) {
-  int32_t len = 0;
-  char   *base64 = base64_encode((const unsigned char *)pUser->pass, TSDB_KEY_LEN);
-
-  len += snprintf(buf + len, maxLen - len, "{\"type\":%d, ", MN_SDB_USER);
-  len += snprintf(buf + len, maxLen - len, "\"user\":\"%s\", ", pUser->user);
-  len += snprintf(buf + len, maxLen - len, "\"auth\":\"%24s\", ", base64);
-  len += snprintf(buf + len, maxLen - len, "\"acct\":\"%s\", ", pUser->acct);
-  len += snprintf(buf + len, maxLen - len, "\"createdTime\":\"%" PRIu64 "\", ", pUser->createdTime);
-  len += snprintf(buf + len, maxLen - len, "\"updateTime\":\"%" PRIu64 "\"}\n", pUser->updateTime);
-
-  free(base64);
-  return len;
-}
-
-SUserObj *mnodeDecodeUser(cJSON *root) {
-  int32_t   code = -1;
-  SUserObj *pUser = calloc(1, sizeof(SUserObj));
-
-  cJSON *user = cJSON_GetObjectItem(root, "user");
-  if (!user || user->type != cJSON_String) {
-    mError("failed to parse user since user not found");
-    goto DECODE_USER_OVER;
-  }
-  tstrncpy(pUser->user, user->valuestring, TSDB_USER_LEN);
-
-  if (strcmp(pUser->user, TSDB_DEFAULT_USER) == 0) {
-    pUser->rootAuth = 1;
-  }
-
-  cJSON *pass = cJSON_GetObjectItem(root, "auth");
-  if (!pass || pass->type != cJSON_String) {
-    mError("user:%s, failed to parse since auth not found", pUser->user);
-    goto DECODE_USER_OVER;
-  }
-
-  int32_t outlen = 0;
-  char   *base64 = (char *)base64_decode(pass->valuestring, strlen(pass->valuestring), &outlen);
-  if (outlen != TSDB_KEY_LEN) {
-    mError("user:%s, failed to parse since invalid auth format", pUser->user);
-    free(base64);
-    goto DECODE_USER_OVER;
-  } else {
-    memcpy(pUser->pass, base64, outlen);
-    free(base64);
-  }
-
-  cJSON *acct = cJSON_GetObjectItem(root, "acct");
-  if (!acct || acct->type != cJSON_String) {
-    mError("user:%s, failed to parse since acct not found", pUser->user);
-    goto DECODE_USER_OVER;
-  }
-  tstrncpy(pUser->acct, acct->valuestring, TSDB_USER_LEN);
-
-  cJSON *createdTime = cJSON_GetObjectItem(root, "createdTime");
-  if (!createdTime || createdTime->type != cJSON_String) {
-    mError("user:%s, failed to parse since createdTime not found", pUser->user);
-    goto DECODE_USER_OVER;
-  }
-  pUser->createdTime = atol(createdTime->valuestring);
-
-  cJSON *updateTime = cJSON_GetObjectItem(root, "updateTime");
-  if (!updateTime || updateTime->type != cJSON_String) {
-    mError("user:%s, failed to parse since updateTime not found", pUser->user);
-    goto DECODE_USER_OVER;
-  }
-  pUser->updateTime = atol(updateTime->valuestring);
-
-  code = 0;
-  mTrace("user:%s, parse success", pUser->user);
-
-DECODE_USER_OVER:
-  if (code != 0) {
-    free(pUser);
-    pUser = NULL;
-  }
-  return pUser;
-}
-
-int32_t mnodeInitUser() {
-  sdbSetFp(MN_SDB_USER, MN_KEY_BINARY, mnodeCreateDefaultUsers, (SdbEncodeFp)mnodeEncodeUser,
-           (SdbDecodeFp)(mnodeDecodeUser), sizeof(SUserObj));
-  return 0;
-}
-
-void mnodeCleanupUser() {}

source/dnode/mnode/src/mnodeWorker.c

@@ -1,494 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "os.h"
-#include "tworker.h"
-#include "tglobal.h"
-#include "mnodeMnode.h"
-#include "mnodeSdb.h"
-#include "mnodeShow.h"
-#include "mnodeSync.h"
-#include "mnodeWorker.h"
-
-static struct {
-  SWorkerPool read;
-  SWorkerPool write;
-  SWorkerPool peerReq;
-  SWorkerPool peerRsp;
-  taos_queue        readQ;
-  taos_queue        writeQ;
-  taos_queue        peerReqQ;
-  taos_queue        peerRspQ;
-  int32_t (*writeMsgFp[TSDB_MSG_TYPE_MAX])(SMnMsg *);
-  int32_t (*readMsgFp[TSDB_MSG_TYPE_MAX])(SMnMsg *);
-  int32_t (*peerReqFp[TSDB_MSG_TYPE_MAX])(SMnMsg *);
-  void (*peerRspFp[TSDB_MSG_TYPE_MAX])(SRpcMsg *);
-  void (*msgFp[TSDB_MSG_TYPE_MAX])(SRpcMsg *pMsg);
-} tsMworker = {0};
-
-static SMnMsg *mnodeInitMsg2(SRpcMsg *pRpcMsg) {
-  int32_t    size = sizeof(SMnMsg) + pRpcMsg->contLen;
-  SMnMsg *pMsg = taosAllocateQitem(size);
-
-  pMsg->rpcMsg = *pRpcMsg;
-  pMsg->rpcMsg.pCont = pMsg->pCont;
-  pMsg->createdTime = taosGetTimestampSec();
-  memcpy(pMsg->pCont, pRpcMsg->pCont, pRpcMsg->contLen);
-
-  SRpcConnInfo connInfo = {0};
-  if (rpcGetConnInfo(pMsg->rpcMsg.handle, &connInfo) == 0) {
-    pMsg->pUser = sdbGetRow(MN_SDB_USER, connInfo.user);
-  }
-
-  if (pMsg->pUser == NULL) {
-    mError("can not get user from conn:%p", pMsg->rpcMsg.handle);
-    taosFreeQitem(pMsg);
-    return NULL;
-  }
-
-  return pMsg;
-}
-
-static void mnodeCleanupMsg2(SMnMsg *pMsg) {
-  if (pMsg == NULL) return;
-  if (pMsg->rpcMsg.pCont != pMsg->pCont) {
-    tfree(pMsg->rpcMsg.pCont);
-  }
-
-  taosFreeQitem(pMsg);
-}
-
-static void mnodeDispatchToWriteQueue(SRpcMsg *pRpcMsg) {
-  if (mnodeGetStatus() != MN_STATUS_READY || tsMworker.writeQ == NULL) {
-    mnodeSendRedirectMsg(pRpcMsg, true);
-  } else {
-    SMnMsg *pMsg = mnodeInitMsg2(pRpcMsg);
-    if (pMsg == NULL) {
-      SRpcMsg rpcRsp = {.handle = pRpcMsg->handle, .code = TSDB_CODE_MND_INVALID_USER};
-      rpcSendResponse(&rpcRsp);
-    } else {
-      mTrace("msg:%p, app:%p type:%s is put into wqueue", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]);
-      taosWriteQitem(tsMworker.writeQ, pMsg);
-    }
-  }
-
-  rpcFreeCont(pRpcMsg->pCont);
-}
-
-void mnodeReDispatchToWriteQueue(SMnMsg *pMsg) {
-  if (mnodeGetStatus() != MN_STATUS_READY || tsMworker.writeQ == NULL) {
-    mnodeSendRedirectMsg(&pMsg->rpcMsg, true);
-    mnodeCleanupMsg2(pMsg);
-  } else {
-    taosWriteQitem(tsMworker.writeQ, pMsg);
-  }
-}
-
-static void mnodeDispatchToReadQueue(SRpcMsg *pRpcMsg) {
-  if (mnodeGetStatus() != MN_STATUS_READY || tsMworker.readQ == NULL) {
-    mnodeSendRedirectMsg(pRpcMsg, true);
-  } else {
-    SMnMsg *pMsg = mnodeInitMsg2(pRpcMsg);
-    if (pMsg == NULL) {
-      SRpcMsg rpcRsp = {.handle = pRpcMsg->handle, .code = TSDB_CODE_MND_INVALID_USER};
-      rpcSendResponse(&rpcRsp);
-    } else {
-      mTrace("msg:%p, app:%p type:%s is put into rqueue", pMsg, pMsg->rpcMsg.ahandle, taosMsg[pMsg->rpcMsg.msgType]);
-      taosWriteQitem(tsMworker.readQ, pMsg);
-    }
-  }
-
-  rpcFreeCont(pRpcMsg->pCont);
-}
-
-static void mnodeDispatchToPeerQueue(SRpcMsg *pRpcMsg) {
-  if (mnodeGetStatus() != MN_STATUS_READY || tsMworker.peerReqQ == NULL) {
-    mnodeSendRedirectMsg(pRpcMsg, false);
-  } else {
-    SMnMsg *pMsg = mnodeInitMsg2(pRpcMsg);
-    if (pMsg == NULL) {
-      SRpcMsg rpcRsp = {.handle = pRpcMsg->handle, .code = TSDB_CODE_MND_INVALID_USER};
-      rpcSendResponse(&rpcRsp);
-    } else {
-      mTrace("msg:%p, app:%p type:%s is put into peer req queue", pMsg, pMsg->rpcMsg.ahandle,
-             taosMsg[pMsg->rpcMsg.msgType]);
-      taosWriteQitem(tsMworker.peerReqQ, pMsg);
-    }
-  }
-
-  rpcFreeCont(pRpcMsg->pCont);
-}
-
-void mnodeDispatchToPeerRspQueue(SRpcMsg *pRpcMsg) {
-  SMnMsg *pMsg = mnodeInitMsg2(pRpcMsg);
-  if (pMsg == NULL) {
-    SRpcMsg rpcRsp = {.handle = pRpcMsg->handle, .code = TSDB_CODE_MND_INVALID_USER};
-    rpcSendResponse(&rpcRsp);
-  } else {
-    mTrace("msg:%p, app:%p type:%s is put into peer rsp queue", pMsg, pMsg->rpcMsg.ahandle,
-           taosMsg[pMsg->rpcMsg.msgType]);
-    taosWriteQitem(tsMworker.peerRspQ, pMsg);
-  }
-
-  // rpcFreeCont(pRpcMsg->pCont);
-}
-
-void mnodeSendRsp(SMnMsg *pMsg, int32_t code) {
-  if (pMsg == NULL) return;
-  if (code == TSDB_CODE_MND_ACTION_IN_PROGRESS) return;
-  if (code == TSDB_CODE_MND_ACTION_NEED_REPROCESSED) {
-    mnodeReDispatchToWriteQueue(pMsg);
-    return;
-  }
-
-  SRpcMsg rpcRsp = {
-      .handle = pMsg->rpcMsg.handle,
-      .pCont = pMsg->rpcRsp.rsp,
-      .contLen = pMsg->rpcRsp.len,
-      .code = code,
-  };
-
-  rpcSendResponse(&rpcRsp);
-  mnodeCleanupMsg2(pMsg);
-}
-
-static void mnodeInitMsgFp() {
-//   // peer req
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DM_CONFIG_TABLE] = mnodeDispatchToPeerQueue;
-//   tsMworker.peerReqFp[TSDB_MSG_TYPE_DM_CONFIG_TABLE] = mnodeProcessTableCfgMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DM_CONFIG_VNODE] = mnodeDispatchToPeerQueue;
-//   tsMworker.peerReqFp[TSDB_MSG_TYPE_DM_CONFIG_VNODE] = mnodeProcessVnodeCfgMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_AUTH] = mnodeDispatchToPeerQueue;
-//   tsMworker.peerReqFp[TSDB_MSG_TYPE_AUTH] = mnodeProcessAuthMsg;
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_GRANT] = mnodeDispatchToPeerQueue;
-//   // tsMworker.peerReqFp[TSDB_MSG_TYPE_GRANT] = grantProcessMsgInMgmt;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_STATUS] = mnodeDispatchToPeerQueue;
-//   tsMworker.peerReqFp[TSDB_MSG_TYPE_STATUS] = mnodeProcessDnodeStatusMsg;
-
-//   // peer rsp
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_CONFIG_DNODE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_CONFIG_DNODE_RSP] = mnodeProcessCfgDnodeMsgRsp;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_DROP_STABLE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_DROP_STABLE_RSP] = mnodeProcessDropSuperTableRsp;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_CREATE_TABLE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_CREATE_TABLE_RSP] = mnodeProcessCreateChildTableRsp;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_DROP_TABLE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_DROP_TABLE_RSP] = mnodeProcessDropChildTableRsp;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_ALTER_TABLE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_ALTER_TABLE_RSP] = mnodeProcessAlterTableRsp;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_CREATE_VNODE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_CREATE_VNODE_RSP] = mnodeProcessCreateVnodeRsp;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_ALTER_VNODE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_ALTER_VNODE_RSP] = mnodeProcessAlterVnodeRsp;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_COMPACT_VNODE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_COMPACT_VNODE_RSP] = mnodeProcessCompactVnodeRsp;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_MD_DROP_VNODE_RSP] = mnodeDispatchToPeerRspQueue;
-//   tsMworker.peerRspFp[TSDB_MSG_TYPE_MD_DROP_VNODE_RSP] = mnodeProcessDropVnodeRsp;
-
-//   // read msg
-//   tsMworker.msgFp[TSDB_MSG_TYPE_HEARTBEAT] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_HEARTBEAT] = mnodeProcessHeartBeatMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CONNECT] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_CONNECT] = mnodeProcessConnectMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_USE_DB] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_USE_DB] = mnodeProcessUseMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_TABLE_META] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_TABLE_META] = mnodeProcessTableMetaMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_TABLES_META] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_TABLES_META] = mnodeProcessMultiTableMetaMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_STABLE_VGROUP] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_STABLE_VGROUP] = mnodeProcessSuperTableVgroupMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_SHOW] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_SHOW] = mnodeProcessShowMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_SHOW_RETRIEVE] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_SHOW_RETRIEVE] = mnodeProcessRetrieveMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_RETRIEVE_FUNC] = mnodeDispatchToReadQueue;
-//   tsMworker.readMsgFp[TSDB_MSG_TYPE_RETRIEVE_FUNC] = mnodeProcessRetrieveFuncReq;
-
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_ACCT] = mnodeDispatchToWriteQueue;
-//   // tsMworker.readMsgFp[TSDB_MSG_TYPE_CREATE_ACCT] = acctProcessCreateAcctMsg;
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_ALTER_ACCT] = mnodeDispatchToWriteQueue;
-//   // tsMworker.readMsgFp[TSDB_MSG_TYPE_ALTER_ACCT] = acctProcessDropAcctMsg;
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_DROP_ACCT] = mnodeDispatchToWriteQueue;
-//   // tsMworker.readMsgFp[TSDB_MSG_TYPE_DROP_ACCT] = acctProcessAlterAcctMsg;
-
-//   // write msg
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_USER] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_CREATE_USER] = mnodeProcessCreateUserMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_ALTER_USER] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_ALTER_USER] = mnodeProcessAlterUserMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DROP_USER] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_DROP_USER] = mnodeProcessDropUserMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_DNODE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_CREATE_DNODE] = mnodeProcessCreateDnodeMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DROP_DNODE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_DROP_DNODE] = mnodeProcessDropDnodeMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CONFIG_DNODE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_CONFIG_DNODE] = mnodeProcessCfgDnodeMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_DB] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_CREATE_DB] = mnodeProcessCreateDbMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_ALTER_DB] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_ALTER_DB] = mnodeProcessAlterDbMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DROP_DB] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_DROP_DB] = mnodeProcessDropDbMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_SYNC_DB] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_SYNC_DB] = mnodeProcessSyncDbMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_COMPACT_VNODE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_COMPACT_VNODE] = mnodeProcessCompactMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_FUNCTION] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_CREATE_FUNCTION] = mnodeProcessCreateFuncMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DROP_FUNCTION] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_DROP_FUNCTION] = mnodeProcessDropFuncMsg;
-
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_TP] = mnodeDispatchToWriteQueue;
-//   // tsMworker.readMsgFp[TSDB_MSG_TYPE_CREATE_TP] = tpProcessCreateTpMsg;
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_DROP_TP] = mnodeDispatchToWriteQueue;
-//   // tsMworker.readMsgFp[TSDB_MSG_TYPE_DROP_TP] = tpProcessAlterTpMsg;
-//   // tsMworker.msgFp[TSDB_MSG_TYPE_ALTER_TP] = mnodeDispatchToWriteQueue;
-//   // tsMworker.readMsgFp[TSDB_MSG_TYPE_ALTER_TP] = tpProcessDropTpMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_CREATE_TABLE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_CREATE_TABLE] = mnodeProcessCreateTableMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_DROP_TABLE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_DROP_TABLE] = mnodeProcessDropTableMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_ALTER_TABLE] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_ALTER_TABLE] = mnodeProcessAlterTableMsg;
-
-//   tsMworker.msgFp[TSDB_MSG_TYPE_ALTER_STREAM] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_ALTER_STREAM] = NULL;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_KILL_QUERY] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_KILL_QUERY] = mnodeProcessKillQueryMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_KILL_STREAM] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_KILL_STREAM] = mnodeProcessKillStreamMsg;
-//   tsMworker.msgFp[TSDB_MSG_TYPE_KILL_CONN] = mnodeDispatchToWriteQueue;
-//   tsMworker.writeMsgFp[TSDB_MSG_TYPE_KILL_CONN] = mnodeProcessKillConnectionMsg;
-}
-
-static void mnodeProcessWriteReq(SMnMsg *pMsg, void *unused) {
-  int32_t msgType = pMsg->rpcMsg.msgType;
-  void   *ahandle = pMsg->rpcMsg.ahandle;
-  int32_t code = 0;
-
-  if (pMsg->rpcMsg.pCont == NULL) {
-    mError("msg:%p, app:%p type:%s content is null", pMsg, ahandle, taosMsg[msgType]);
-    code = TSDB_CODE_MND_INVALID_MSG_LEN;
-    goto PROCESS_WRITE_REQ_END;
-  }
-
-  if (!mnodeIsMaster()) {
-    SMnRsp *rpcRsp = &pMsg->rpcRsp;
-    SEpSet *epSet = rpcMallocCont(sizeof(SEpSet));
-    mnodeGetMnodeEpSetForShell(epSet, true);
-    rpcRsp->rsp = epSet;
-    rpcRsp->len = sizeof(SEpSet);
-
-    mDebug("msg:%p, app:%p type:%s in write queue, is redirected, numOfEps:%d inUse:%d", pMsg, ahandle,
-           taosMsg[msgType], epSet->numOfEps, epSet->inUse);
-
-    code = TSDB_CODE_RPC_REDIRECT;
-    goto PROCESS_WRITE_REQ_END;
-  }
-
-  if (tsMworker.writeMsgFp[msgType] == NULL) {
-    mError("msg:%p, app:%p type:%s not processed", pMsg, ahandle, taosMsg[msgType]);
-    code = TSDB_CODE_MND_MSG_NOT_PROCESSED;
-    goto PROCESS_WRITE_REQ_END;
-  }
-
-  code = (*tsMworker.writeMsgFp[msgType])(pMsg);
-
-PROCESS_WRITE_REQ_END:
-  mnodeSendRsp(pMsg, code);
-}
-
-static void mnodeProcessReadReq(SMnMsg *pMsg, void *unused) {
-  int32_t msgType = pMsg->rpcMsg.msgType;
-  void   *ahandle = pMsg->rpcMsg.ahandle;
-  int32_t code = 0;
-
-  if (pMsg->rpcMsg.pCont == NULL) {
-    mError("msg:%p, app:%p type:%s in mread queue, content is null", pMsg, ahandle, taosMsg[msgType]);
-    code = TSDB_CODE_MND_INVALID_MSG_LEN;
-    goto PROCESS_READ_REQ_END;
-  }
-
-  if (!mnodeIsMaster()) {
-    SMnRsp *rpcRsp = &pMsg->rpcRsp;
-    SEpSet *epSet = rpcMallocCont(sizeof(SEpSet));
-    if (!epSet) {
-      code = TSDB_CODE_MND_OUT_OF_MEMORY;
-      goto PROCESS_READ_REQ_END;
-    }
-    mnodeGetMnodeEpSetForShell(epSet, true);
-    rpcRsp->rsp = epSet;
-    rpcRsp->len = sizeof(SEpSet);
-
-    mDebug("msg:%p, app:%p type:%s in mread queue is redirected, numOfEps:%d inUse:%d", pMsg, ahandle, taosMsg[msgType],
-           epSet->numOfEps, epSet->inUse);
-    code = TSDB_CODE_RPC_REDIRECT;
-    goto PROCESS_READ_REQ_END;
-  }
-
-  if (tsMworker.readMsgFp[msgType] == NULL) {
-    mError("msg:%p, app:%p type:%s in mread queue, not processed", pMsg, ahandle, taosMsg[msgType]);
-    code = TSDB_CODE_MND_MSG_NOT_PROCESSED;
-    goto PROCESS_READ_REQ_END;
-  }
-
-  mTrace("msg:%p, app:%p type:%s will be processed in mread queue", pMsg, ahandle, taosMsg[msgType]);
-  code = (*tsMworker.readMsgFp[msgType])(pMsg);
-
-PROCESS_READ_REQ_END:
-  mnodeSendRsp(pMsg, code);
-}
-
-static void mnodeProcessPeerReq(SMnMsg *pMsg, void *unused) {
-  int32_t msgType = pMsg->rpcMsg.msgType;
-  void   *ahandle = pMsg->rpcMsg.ahandle;
-  int32_t code = 0;
-
-  if (pMsg->rpcMsg.pCont == NULL) {
-    mError("msg:%p, ahandle:%p type:%s in mpeer queue, content is null", pMsg, ahandle, taosMsg[msgType]);
-    code = TSDB_CODE_MND_INVALID_MSG_LEN;
-    goto PROCESS_PEER_REQ_END;
-  }
-
-  if (!mnodeIsMaster()) {
-    SMnRsp *rpcRsp = &pMsg->rpcRsp;
-    SEpSet *epSet = rpcMallocCont(sizeof(SEpSet));
-    mnodeGetMnodeEpSetForPeer(epSet, true);
-    rpcRsp->rsp = epSet;
-    rpcRsp->len = sizeof(SEpSet);
-
-    mDebug("msg:%p, ahandle:%p type:%s in mpeer queue is redirected, numOfEps:%d inUse:%d", pMsg, ahandle,
-           taosMsg[msgType], epSet->numOfEps, epSet->inUse);
-
-    code = TSDB_CODE_RPC_REDIRECT;
-    goto PROCESS_PEER_REQ_END;
-  }
-
-  if (tsMworker.peerReqFp[msgType] == NULL) {
-    mError("msg:%p, ahandle:%p type:%s in mpeer queue, not processed", pMsg, ahandle, taosMsg[msgType]);
-    code = TSDB_CODE_MND_MSG_NOT_PROCESSED;
-    goto PROCESS_PEER_REQ_END;
-  }
-
-  code = (*tsMworker.peerReqFp[msgType])(pMsg);
-
-PROCESS_PEER_REQ_END:
-  mnodeSendRsp(pMsg, code);
-}
-
-static void mnodeProcessPeerRsp(SMnMsg *pMsg, void *unused) {
-  int32_t  msgType = pMsg->rpcMsg.msgType;
-  SRpcMsg *pRpcMsg = &pMsg->rpcMsg;
-
-  if (!mnodeIsMaster()) {
-    mError("msg:%p, ahandle:%p type:%s not processed for not master", pRpcMsg, pRpcMsg->ahandle, taosMsg[msgType]);
-    mnodeCleanupMsg2(pMsg);
-  }
-
-  if (tsMworker.peerRspFp[msgType]) {
-    (*tsMworker.peerRspFp[msgType])(pRpcMsg);
-  } else {
-    mError("msg:%p, ahandle:%p type:%s is not processed", pRpcMsg, pRpcMsg->ahandle, taosMsg[msgType]);
-  }
-
-  mnodeCleanupMsg2(pMsg);
-}
-
-int32_t mnodeInitWorker() {
-  mnodeInitMsgFp();
-
-  SWorkerPool *pPool = &tsMworker.write;
-  pPool->name = "mnode-write";
-  pPool->min = 1;
-  pPool->max = 1;
-  if (tWorkerInit(pPool) != 0) {
-    return TSDB_CODE_MND_OUT_OF_MEMORY;
-  } else {
-    tsMworker.writeQ = tWorkerAllocQueue(pPool, NULL, (FProcessItem)mnodeProcessWriteReq);
-  }
-
-  pPool = &tsMworker.read;
-  pPool->name = "mnode-read";
-  pPool->min = 2;
-  pPool->max = (int32_t)(tsNumOfCores * tsNumOfThreadsPerCore / 2);
-  pPool->max = MAX(2, pPool->max);
-  pPool->max = MIN(4, pPool->max);
-  if (tWorkerInit(pPool) != 0) {
-    return TSDB_CODE_MND_OUT_OF_MEMORY;
-  } else {
-    tsMworker.readQ = tWorkerAllocQueue(pPool, NULL, (FProcessItem)mnodeProcessReadReq);
-  }
-
-  pPool = &tsMworker.peerReq;
-  pPool->name = "mnode-peer-req";
-  pPool->min = 1;
-  pPool->max = 1;
-  if (tWorkerInit(pPool) != 0) {
-    return TSDB_CODE_MND_OUT_OF_MEMORY;
-  } else {
-    tsMworker.peerReqQ = tWorkerAllocQueue(pPool, NULL, (FProcessItem)mnodeProcessPeerReq);
-  }
-
-  pPool = &tsMworker.peerRsp;
-  pPool->name = "mnode-peer-rsp";
-  pPool->min = 1;
-  pPool->max = 1;
-  if (tWorkerInit(pPool) != 0) {
-    return TSDB_CODE_MND_OUT_OF_MEMORY;
-  } else {
-    tsMworker.peerRspQ = tWorkerAllocQueue(pPool, NULL, (FProcessItem)mnodeProcessPeerRsp);
-  }
-
-  mInfo("mnode worker is initialized");
-  return 0;
-}
-
-void mnodeCleanupWorker() {
-  tWorkerFreeQueue(&tsMworker.write, tsMworker.writeQ);
-  tWorkerCleanup(&tsMworker.write);
-  tsMworker.writeQ = NULL;
-
-  tWorkerFreeQueue(&tsMworker.read, tsMworker.readQ);
-  tWorkerCleanup(&tsMworker.read);
-  tsMworker.readQ = NULL;
-
-  tWorkerFreeQueue(&tsMworker.peerReq, tsMworker.peerReqQ);
-  tWorkerCleanup(&tsMworker.peerReq);
-  tsMworker.peerReqQ = NULL;
-
-  tWorkerFreeQueue(&tsMworker.peerRsp, tsMworker.peerRspQ);
-  tWorkerCleanup(&tsMworker.peerRsp);
-  tsMworker.peerRspQ = NULL;
-
-  mInfo("mnode worker is closed");
-}
-
-SMnodeMsg *mnodeInitMsg(int32_t msgNum) { return NULL; }
-
-int32_t mnodeAppendMsg(SMnodeMsg *pMsg, SRpcMsg *pRpcMsg) { return 0; }
-
-void mnodeCleanupMsg(SMnodeMsg *pMsg) {}
-void mnodeProcessMsg(SMnodeMsg *pMsg, EMnMsgType msgType) {}

source/dnode/mnode/src/mondeInt.c

@@ -1,250 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#define _DEFAULT_SOURCE
-#include "os.h"
-#include "tglobal.h"
-#include "tstep.h"
-#include "mnodeAcct.h"
-#include "mnodeAuth.h"
-#include "mnodeBalance.h"
-#include "mnodeCluster.h"
-#include "mnodeDb.h"
-#include "mnodeDnode.h"
-#include "mnodeFunc.h"
-#include "mnodeMnode.h"
-#include "mnodeOper.h"
-#include "mnodeProfile.h"
-#include "mnodeSdb.h"
-#include "mnodeShow.h"
-#include "mnodeStable.h"
-#include "mnodeSync.h"
-#include "mnodeUser.h"
-#include "mnodeVgroup.h"
-#include "mnodeWorker.h"
-#include "mnodeTelem.h"
-
-static struct {
-  int32_t    state;
-  int32_t    dnodeId;
-  int64_t    clusterId;
-  tmr_h      timer;
-  SSteps    *steps1;
-  SSteps    *steps2;
-  SMnodePara para;
-} tsMint;
-
-tmr_h mnodeGetTimer() { return tsMint.timer; }
-
-int32_t mnodeGetDnodeId() { return tsMint.para.dnodeId; }
-
-int64_t mnodeGetClusterId() { return tsMint.para.clusterId; }
-
-EMnStatus mnodeGetStatus() { return tsMint.state; }
-
-void mnodeSendMsgToDnode(struct SEpSet *epSet, struct SRpcMsg *rpcMsg) {
-  (*tsMint.para.SendMsgToDnode)(epSet, rpcMsg);
-}
-
-void mnodeSendMsgToMnode(struct SRpcMsg *rpcMsg) { return (*tsMint.para.SendMsgToMnode)(rpcMsg); }
-
-void mnodeSendRedirectMsg(struct SRpcMsg *rpcMsg, bool forShell) { (*tsMint.para.SendRedirectMsg)(rpcMsg, forShell); }
-
-int32_t mnodeGetLoad(SMnodeLoad *pLoad) { return 0; }
-
-static int32_t mnodeSetPara(SMnodePara para) {
-  tsMint.para = para;
-
-  if (tsMint.para.SendMsgToDnode == NULL) return -1;
-  if (tsMint.para.SendMsgToMnode == NULL) return -1;
-  if (tsMint.para.SendRedirectMsg == NULL) return -1;
-  if (tsMint.para.PutMsgIntoApplyQueue == NULL) return -1;
-  if (tsMint.para.dnodeId < 0) return -1;
-  if (tsMint.para.clusterId < 0) return -1;
-
-  return 0;
-}
-
-static int32_t mnodeInitTimer() {
-  if (tsMint.timer == NULL) {
-    tsMint.timer = taosTmrInit(tsMaxShellConns, 200, 3600000, "MND");
-  }
-
-  return 0;
-}
-
-static void mnodeCleanupTimer() {
-  if (tsMint.timer != NULL) {
-    taosTmrCleanUp(tsMint.timer);
-    tsMint.timer = NULL;
-  }
-}
-
-static int32_t mnodeInitStep1() {
-  struct SSteps *steps = taosStepInit(16, NULL);
-  if (steps == NULL) return -1;
-
-  taosStepAdd(steps, "mnode-sdb", sdbInit, sdbCleanup);
-  taosStepAdd(steps, "mnode-cluster", mnodeInitCluster, mnodeCleanupCluster);
-  taosStepAdd(steps, "mnode-dnode", mnodeInitDnode, mnodeCleanupDnode);
-  taosStepAdd(steps, "mnode-mnode", mnodeInitMnode, mnodeCleanupMnode);
-  taosStepAdd(steps, "mnode-acct", mnodeInitAcct, mnodeCleanupAcct);
-  taosStepAdd(steps, "mnode-auth", mnodeInitAuth, mnodeCleanupAuth);
-  taosStepAdd(steps, "mnode-user", mnodeInitUser, mnodeCleanupUser);
-  taosStepAdd(steps, "mnode-db", mnodeInitDb, mnodeCleanupDb);
-  taosStepAdd(steps, "mnode-vgroup", mnodeInitVgroup, mnodeCleanupVgroup);
-  taosStepAdd(steps, "mnode-stable", mnodeInitStable, mnodeCleanupStable);
-  taosStepAdd(steps, "mnode-func", mnodeInitFunc, mnodeCleanupFunc);
-  taosStepAdd(steps, "mnode-oper", mnodeInitOper, mnodeCleanupOper);
-
-  tsMint.steps1 = steps;
-  return taosStepExec(tsMint.steps1);
-}
-
-static int32_t mnodeInitStep2() {
-  struct SSteps *steps = taosStepInit(12, NULL);
-  if (steps == NULL) return -1;
-
-  taosStepAdd(steps, "mnode-timer", mnodeInitTimer, NULL);
-  taosStepAdd(steps, "mnode-worker", mnodeInitWorker, NULL);
-  taosStepAdd(steps, "mnode-balance", mnodeInitBalance, mnodeCleanupBalance);
-  taosStepAdd(steps, "mnode-profile", mnodeInitProfile, mnodeCleanupProfile);
-  taosStepAdd(steps, "mnode-show", mnodeInitShow, mnodeCleanUpShow);
-  taosStepAdd(steps, "mnode-sync", mnodeInitSync, mnodeCleanUpSync);
-  taosStepAdd(steps, "mnode-worker", NULL, mnodeCleanupWorker);
-  taosStepAdd(steps, "mnode-telem", mnodeInitTelem, mnodeCleanupTelem);
-  taosStepAdd(steps, "mnode-timer", NULL, mnodeCleanupTimer);
-
-  tsMint.steps2 = steps;
-  return taosStepExec(tsMint.steps2);
-}
-
-static void mnodeCleanupStep1() { taosStepCleanup(tsMint.steps1); }
-
-static void mnodeCleanupStep2() { taosStepCleanup(tsMint.steps2); }
-
-static bool mnodeNeedDeploy() {
-  if (tsMint.para.dnodeId > 0) return false;
-  if (tsMint.para.clusterId > 0) return false;
-  if (strcmp(tsFirst, tsLocalEp) != 0) return false;
-  return true;
-}
-
-int32_t mnodeDeploy(char *path, SMnodeCfg *pCfg) {
-  if (tsMint.state != MN_STATUS_UNINIT) {
-    mError("failed to deploy mnode since its deployed");
-    return 0;
-  } else {
-    tsMint.state = MN_STATUS_INIT;
-  }
-
-  if (tsMint.para.dnodeId <= 0 || tsMint.para.clusterId <= 0) {
-    mError("failed to deploy mnode since cluster not ready");
-    return TSDB_CODE_MND_NOT_READY;
-  }
-
-  mInfo("starting to deploy mnode");
-
-  int32_t code = mnodeInitStep1();
-  if (code != 0) {
-    mError("failed to deploy mnode since init step1 error");
-    tsMint.state = MN_STATUS_UNINIT;
-    return TSDB_CODE_MND_SDB_ERROR;
-  }
-
-  code = mnodeInitStep2();
-  if (code != 0) {
-    mnodeCleanupStep1();
-    mError("failed to deploy mnode since init step2 error");
-    tsMint.state = MN_STATUS_UNINIT;
-    return TSDB_CODE_MND_SDB_ERROR;
-  }
-
-  mDebug("mnode is deployed and waiting for raft to confirm");
-  tsMint.state = MN_STATUS_READY;
-  return 0;
-}
-
-void mnodeUnDeploy(char *path) {
-  sdbUnDeploy();
-  mnodeCleanup();
-}
-
-int32_t mnodeInit(SMnodePara para) {
-  mDebugFlag = 207;
-  if (tsMint.state != MN_STATUS_UNINIT) {
-    return 0;
-  } else {
-    tsMint.state = MN_STATUS_INIT;
-  }
-
-  mInfo("starting to initialize mnode ...");
-
-  int32_t code = mnodeSetPara(para);
-  if (code != 0) {
-    tsMint.state = MN_STATUS_UNINIT;
-    return code;
-  }
-
-  code = mnodeInitStep1();
-  if (code != 0) {
-    tsMint.state = MN_STATUS_UNINIT;
-    return -1;
-  }
-
-  code = sdbRead();
-  if (code != 0) {
-    if (mnodeNeedDeploy()) {
-      code = sdbDeploy();
-      if (code != 0) {
-        mnodeCleanupStep1();
-        tsMint.state = MN_STATUS_UNINIT;
-        return -1;
-      }
-    } else {
-      mnodeCleanupStep1();
-      tsMint.state = MN_STATUS_UNINIT;
-      return -1;
-    }
-  }
-
-  code = mnodeInitStep2();
-  if (code != 0) {
-    mnodeCleanupStep1();
-    tsMint.state = MN_STATUS_UNINIT;
-    return -1;
-  }
-
-  tsMint.state = MN_STATUS_READY;
-  mInfo("mnode is initialized successfully");
-  return 0;
-}
-
-void mnodeCleanup() {
-  if (tsMint.state != MN_STATUS_UNINIT && tsMint.state != MN_STATUS_CLOSING) {
-    mInfo("starting to clean up mnode");
-    tsMint.state = MN_STATUS_CLOSING;
-
-    mnodeCleanupStep2();
-    mnodeCleanupStep1();
-
-    tsMint.state = MN_STATUS_UNINIT;
-    mInfo("mnode is cleaned up");
-  }
-}
-
-int32_t mnodeStart(char *path, SMnodeCfg *pCfg) { return 0; }
-int32_t mnodeAlter(SMnodeCfg *pCfg) { return 0; }
-void    mnodeStop() {}

source/dnode/mnode/transaction/CMakeLists.txt

@@ -0,0 +1,15 @@
+aux_source_directory(src MNODE_SRC)
+add_library(transaction ${MNODE_SRC})
+target_include_directories(
+    transaction
+    PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/mnode/transaction"
+    private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
+)
+target_link_libraries(
+    transaction
+    PRIVATE os
+    PRIVATE common
+    PRIVATE util
+    PRIVATE sdb
+    PRIVATE transport
+)

source/dnode/mnode/transaction/inc/trnInt.h

@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_TRANSACTION_INT_H_
+#define _TD_TRANSACTION_INT_H_
+
+#include "os.h"
+#include "trn.h"
+#include "tglobal.h"
+#include "tarray.h"
+#include "tlog.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define mFatal(...) { if (mDebugFlag & DEBUG_FATAL) { taosPrintLog("MND FATAL ", 255, __VA_ARGS__); }}
+#define mError(...) { if (mDebugFlag & DEBUG_ERROR) { taosPrintLog("MND ERROR ", 255, __VA_ARGS__); }}
+#define mWarn(...)  { if (mDebugFlag & DEBUG_WARN)  { taosPrintLog("MND WARN ", 255, __VA_ARGS__); }}
+#define mInfo(...)  { if (mDebugFlag & DEBUG_INFO)  { taosPrintLog("MND ", 255, __VA_ARGS__); }}
+#define mDebug(...) { if (mDebugFlag & DEBUG_DEBUG) { taosPrintLog("MND ", mDebugFlag, __VA_ARGS__); }}
+#define mTrace(...) { if (mDebugFlag & DEBUG_TRACE) { taosPrintLog("MND ", mDebugFlag, __VA_ARGS__); }}
+
+#define TRN_VER 1
+#define TRN_DEFAULT_ARRAY_SIZE 8
+
+typedef enum {
+  TRN_STAGE_PREPARE = 1,
+  TRN_STAGE_EXECUTE = 2,
+  TRN_STAGE_COMMIT = 3,
+  TRN_STAGE_ROLLBACK = 4,
+  TRN_STAGE_RETRY = 5
+} ETrnStage;
+
+typedef struct STrans {
+  int32_t id;
+  int8_t  stage;
+  int8_t  policy;
+  void   *rpcHandle;
+  SArray *redoLogs;
+  SArray *undoLogs;
+  SArray *commitLogs;
+  SArray *redoActions;
+  SArray *undoActions;
+} STrans;
+
+SSdbRaw *trnActionEncode(STrans *pTrans);
+STrans  *trnActionDecode(SSdbRaw *pRaw);
+int32_t  trnActionInsert(STrans *pTrans);
+int32_t  trnActionDelete(STrans *pTrans);
+int32_t  trnActionUpdate(STrans *pSrcTrans, STrans *pDstTrans);
+int32_t  trnGenerateTransId();
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_TRANSACTION_INT_H_*/

source/dnode/mnode/transaction/src/trn.c

@@ -0,0 +1,233 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "trnInt.h"
+
+#define SDB_TRANS_VER 1
+
+SSdbRaw *trnActionEncode(STrans *pTrans) {
+  int32_t rawDataLen = 10 * sizeof(int32_t);
+  int32_t redoLogNum = taosArrayGetSize(pTrans->redoLogs);
+  int32_t undoLogNum = taosArrayGetSize(pTrans->undoLogs);
+  int32_t commitLogNum = taosArrayGetSize(pTrans->commitLogs);
+  int32_t redoActionNum = taosArrayGetSize(pTrans->redoActions);
+  int32_t undoActionNum = taosArrayGetSize(pTrans->undoActions);
+
+  for (int32_t index = 0; index < redoLogNum; ++index) {
+    SSdbRaw *pRaw = taosArrayGet(pTrans->redoLogs, index);
+    rawDataLen += sdbGetRawTotalSize(pRaw);
+  }
+
+  for (int32_t index = 0; index < undoLogNum; ++index) {
+    SSdbRaw *pRaw = taosArrayGet(pTrans->undoLogs, index);
+    rawDataLen += sdbGetRawTotalSize(pRaw);
+  }
+
+  for (int32_t index = 0; index < commitLogNum; ++index) {
+    SSdbRaw *pRaw = taosArrayGet(pTrans->commitLogs, index);
+    rawDataLen += sdbGetRawTotalSize(pRaw);
+  }
+
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_TRANS, SDB_TRANS_VER, rawDataLen);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pData, dataPos, pTrans->id)
+  SDB_SET_INT8(pData, dataPos, pTrans->stage)
+  SDB_SET_INT8(pData, dataPos, pTrans->policy)
+  SDB_SET_INT32(pData, dataPos, redoLogNum)
+  SDB_SET_INT32(pData, dataPos, undoLogNum)
+  SDB_SET_INT32(pData, dataPos, commitLogNum)
+  SDB_SET_INT32(pData, dataPos, redoActionNum)
+  SDB_SET_INT32(pData, dataPos, undoActionNum)
+  SDB_SET_DATALEN(pRaw, dataPos);
+
+  return pRaw;
+}
+
+STrans *trnActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != SDB_TRANS_VER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    return NULL;
+  }
+
+  SSdbRow *pRow = sdbAllocRow(sizeof(STrans));
+  STrans  *pTrans = sdbGetRowObj(pRow);
+  if (pTrans == NULL) return NULL;
+
+  int32_t  redoLogNum = 0;
+  int32_t  undoLogNum = 0;
+  int32_t  commitLogNum = 0;
+  int32_t  redoActionNum = 0;
+  int32_t  undoActionNum = 0;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pTrans->id)
+  SDB_GET_INT8(pRaw, pRow, dataPos, &pTrans->stage)
+  SDB_GET_INT8(pRaw, pRow, dataPos, &pTrans->policy)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &redoLogNum)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &undoLogNum)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &commitLogNum)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &redoActionNum)
+  SDB_GET_INT32(pRaw, pRow, dataPos, &undoActionNum)
+
+  for (int32_t index = 0; index < redoLogNum; ++index) {
+    int32_t dataLen = 0;
+    SDB_GET_INT32(pRaw, pRow, dataPos, &dataLen)
+
+    char *pData = malloc(dataLen);
+    SDB_GET_BINARY(pRaw, pRow, dataPos, pData, dataLen);
+    void *ret = taosArrayPush(pTrans->redoLogs, pData);
+    if (ret == NULL) {
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
+      break;
+    }
+  }
+
+  // if (code != 0) {
+  //   trnDrop(pTrans);
+  //   terrno = code;
+  //   return NULL;
+  // }
+
+  return pTrans;
+}
+
+int32_t trnActionInsert(STrans *pTrans) {
+  SArray *pArray = pTrans->redoLogs;
+  int32_t arraySize = taosArrayGetSize(pArray);
+
+  for (int32_t index = 0; index < arraySize; ++index) {
+    SSdbRaw *pRaw = taosArrayGetP(pArray, index);
+    int32_t  code = sdbWrite(pRaw);
+    if (code != 0) {
+      return code;
+    }
+  }
+
+  return 0;
+}
+
+int32_t trnActionDelete(STrans *pTrans) {
+  SArray *pArray = pTrans->redoLogs;
+  int32_t arraySize = taosArrayGetSize(pArray);
+
+  for (int32_t index = 0; index < arraySize; ++index) {
+    SSdbRaw *pRaw = taosArrayGetP(pArray, index);
+    int32_t  code = sdbWrite(pRaw);
+    if (code != 0) {
+      return code;
+    }
+  }
+
+  return 0;
+}
+
+int32_t trnActionUpdate(STrans *pSrcTrans, STrans *pDstTrans) { return 0; }
+
+int32_t trnGenerateTransId() { return 1; }
+
+STrans *trnCreate(ETrnPolicy policy) {
+  STrans *pTrans = calloc(1, sizeof(STrans));
+  if (pTrans == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  pTrans->id = trnGenerateTransId();
+  pTrans->stage = TRN_STAGE_PREPARE;
+  pTrans->policy = policy;
+  pTrans->redoLogs = taosArrayInit(TRN_DEFAULT_ARRAY_SIZE, sizeof(void *));
+  pTrans->undoLogs = taosArrayInit(TRN_DEFAULT_ARRAY_SIZE, sizeof(void *));
+  pTrans->commitLogs = taosArrayInit(TRN_DEFAULT_ARRAY_SIZE, sizeof(void *));
+  pTrans->redoActions = taosArrayInit(TRN_DEFAULT_ARRAY_SIZE, sizeof(void *));
+  pTrans->undoActions = taosArrayInit(TRN_DEFAULT_ARRAY_SIZE, sizeof(void *));
+
+  if (pTrans->redoLogs == NULL || pTrans->undoLogs == NULL || pTrans->commitLogs == NULL ||
+      pTrans->redoActions == NULL || pTrans->undoActions == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+
+  return pTrans;
+}
+
+static void trnDropArray(SArray *pArray) {
+  for (int32_t index = 0; index < pArray->size; ++index) {
+    SSdbRaw *pRaw = taosArrayGetP(pArray, index);
+    tfree(pRaw);
+  }
+
+  taosArrayDestroy(pArray);
+}
+
+void trnDrop(STrans *pTrans) {
+  trnDropArray(pTrans->redoLogs);
+  trnDropArray(pTrans->undoLogs);
+  trnDropArray(pTrans->commitLogs);
+  trnDropArray(pTrans->redoActions);
+  trnDropArray(pTrans->undoActions);
+  tfree(pTrans);
+}
+
+void trnSetRpcHandle(STrans *pTrans, void *rpcHandle) { pTrans->rpcHandle = rpcHandle; }
+
+static int32_t trnAppendArray(SArray *pArray, SSdbRaw *pRaw) {
+  if (pArray == NULL || pRaw == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  void *ptr = taosArrayPush(pArray, &pRaw);
+  if (ptr == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  return 0;
+}
+
+int32_t trnAppendRedoLog(STrans *pTrans, SSdbRaw *pRaw) { return trnAppendArray(pTrans->redoLogs, pRaw); }
+
+int32_t trnAppendUndoLog(STrans *pTrans, SSdbRaw *pRaw) { return trnAppendArray(pTrans->undoLogs, pRaw); }
+
+int32_t trnAppendCommitLog(STrans *pTrans, SSdbRaw *pRaw) { return trnAppendArray(pTrans->commitLogs, pRaw); }
+
+int32_t trnAppendRedoAction(STrans *pTrans, SEpSet *pEpSet, void *pMsg) {
+  return trnAppendArray(pTrans->redoActions, pMsg);
+}
+
+int32_t trnAppendUndoAction(STrans *pTrans, SEpSet *pEpSet, void *pMsg) {
+  return trnAppendArray(pTrans->undoActions, pMsg);
+}
+
+int32_t trnInit() {
+  SSdbTable table = {.sdbType = SDB_TRANS,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)trnActionEncode,
+                     .decodeFp = (SdbDecodeFp)trnActionDecode,
+                     .insertFp = (SdbInsertFp)trnActionInsert,
+                     .updateFp = (SdbUpdateFp)trnActionUpdate,
+                     .deleteFp = (SdbDeleteFp)trnActionDelete};
+  sdbSetTable(table);
+
+  return 0;
+}
+
+void trnCleanup() {}

source/dnode/mnode/transaction/src/trnExec.c

@@ -0,0 +1,189 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "trnInt.h"
+#include "trpc.h"
+
+int32_t trnPrepare(STrans *pTrans, int32_t (*syncfp)(SSdbRaw *pRaw, void *pData)) {
+  if (syncfp == NULL) return -1;
+
+  SSdbRaw *pRaw = trnActionEncode(pTrans);
+  if (pRaw == NULL) {
+    mError("tranId:%d, failed to decode trans since %s", pTrans->id, terrstr());
+    return -1;
+  }
+
+  if (sdbWrite(pRaw) != 0) {
+    mError("tranId:%d, failed to write trans since %s", pTrans->id, terrstr());
+    return -1;
+  }
+
+  if ((*syncfp)(pRaw, pTrans->rpcHandle) != 0) {
+    mError("tranId:%d, failed to sync trans since %s", pTrans->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static void trnSendRpcRsp(void *rpcHandle, int32_t code) {
+  if (rpcHandle != NULL) {
+    SRpcMsg rspMsg = {.handle = rpcHandle, .code = terrno};
+    rpcSendResponse(&rspMsg);
+  }
+}
+
+int32_t trnApply(SSdbRaw *pRaw, void *pData, int32_t code) {
+  if (code != 0) {
+    trnSendRpcRsp(pData, terrno);
+    return 0;
+  }
+
+  if (sdbWrite(pData) != 0) {
+    code = terrno;
+    trnSendRpcRsp(pData, code);
+    terrno = code;
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t trnExecuteArray(SArray *pArray) {
+  for (int32_t index = 0; index < pArray->size; ++index) {
+    SSdbRaw *pRaw = taosArrayGetP(pArray, index);
+    if (sdbWrite(pRaw) != 0) {
+      return -1;
+    }
+  }
+
+  return 0;
+}
+
+static int32_t trnExecuteRedoLogs(STrans *pTrans) { return trnExecuteArray(pTrans->redoLogs); }
+
+static int32_t trnExecuteUndoLogs(STrans *pTrans) { return trnExecuteArray(pTrans->undoLogs); }
+
+static int32_t trnExecuteCommitLogs(STrans *pTrans) { return trnExecuteArray(pTrans->commitLogs); }
+
+static int32_t trnExecuteRedoActions(STrans *pTrans) { return trnExecuteArray(pTrans->redoActions); }
+
+static int32_t trnExecuteUndoActions(STrans *pTrans) { return trnExecuteArray(pTrans->undoActions); }
+
+static int32_t trnPerformPrepareStage(STrans *pTrans) {
+  if (trnExecuteRedoLogs(pTrans) == 0) {
+    pTrans->stage = TRN_STAGE_EXECUTE;
+    return 0;
+  } else {
+    pTrans->stage = TRN_STAGE_ROLLBACK;
+    return -1;
+  }
+}
+
+static int32_t trnPerformExecuteStage(STrans *pTrans) {
+  int32_t code = trnExecuteRedoActions(pTrans);
+
+  if (code == 0) {
+    pTrans->stage = TRN_STAGE_COMMIT;
+    return 0;
+  } else if (code == TSDB_CODE_MND_ACTION_IN_PROGRESS) {
+    return -1;
+  } else {
+    if (pTrans->policy == TRN_POLICY_RETRY) {
+      pTrans->stage = TRN_STAGE_RETRY;
+    } else {
+      pTrans->stage = TRN_STAGE_ROLLBACK;
+    }
+    return 0;
+  }
+}
+
+static int32_t trnPerformCommitStage(STrans *pTrans) {
+  if (trnExecuteCommitLogs(pTrans) == 0) {
+    pTrans->stage = TRN_STAGE_EXECUTE;
+    return 0;
+  } else {
+    pTrans->stage = TRN_STAGE_ROLLBACK;
+    return -1;
+  }
+}
+
+static int32_t trnPerformRollbackStage(STrans *pTrans) {
+  if (trnExecuteCommitLogs(pTrans) == 0) {
+    pTrans->stage = TRN_STAGE_EXECUTE;
+    return 0;
+  } else {
+    pTrans->stage = TRN_STAGE_ROLLBACK;
+    return -1;
+  }
+}
+
+static int32_t trnPerformRetryStage(STrans *pTrans) {
+  if (trnExecuteCommitLogs(pTrans) == 0) {
+    pTrans->stage = TRN_STAGE_EXECUTE;
+    return 0;
+  } else {
+    pTrans->stage = TRN_STAGE_ROLLBACK;
+    return -1;
+  }
+}
+
+int32_t trnExecute(int32_t tranId) {
+  int32_t code = 0;
+
+  STrans *pTrans = sdbAcquire(SDB_TRANS, &tranId);
+  if (pTrans == NULL) {
+    return -1;
+  }
+
+  if (pTrans->stage == TRN_STAGE_PREPARE) {
+    if (trnPerformPrepareStage(pTrans) != 0) {
+      sdbRelease(pTrans);
+      return -1;
+    }
+  }
+
+  if (pTrans->stage == TRN_STAGE_EXECUTE) {
+    if (trnPerformExecuteStage(pTrans) != 0) {
+      sdbRelease(pTrans);
+      return -1;
+    }
+  }
+
+  if (pTrans->stage == TRN_STAGE_COMMIT) {
+    if (trnPerformCommitStage(pTrans) != 0) {
+      sdbRelease(pTrans);
+      return -1;
+    }
+  }
+
+  if (pTrans->stage == TRN_STAGE_ROLLBACK) {
+    if (trnPerformRollbackStage(pTrans) != 0) {
+      sdbRelease(pTrans);
+      return -1;
+    }
+  }
+
+  if (pTrans->stage == TRN_STAGE_RETRY) {
+    if (trnPerformRetryStage(pTrans) != 0) {
+      sdbRelease(pTrans);
+      return -1;
+    }
+  }
+
+  sdbRelease(pTrans);
+  return 0;
+}

source/dnode/qnode/CMakeLists.txt

@@ -2,6 +2,6 @@ aux_source_directory(src QNODE_SRC)
 add_library(qnode ${QNODE_SRC})
 target_include_directories(
     qnode
-    PUBLIC "${CMAKE_SOURCE_DIR}/include/server/qnode"
+    PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/qnode"
     private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
 )

source/dnode/vnode/impl/CMakeLists.txt

@@ -2,11 +2,12 @@ aux_source_directory(src VNODE_SRC)
 add_library(vnode STATIC ${VNODE_SRC})
 target_include_directories(
     vnode
-    PUBLIC "${CMAKE_SOURCE_DIR}/include/server/vnode"
+    PUBLIC "${CMAKE_SOURCE_DIR}/include/dnode/vnode"
     private "${CMAKE_CURRENT_SOURCE_DIR}/inc"
 )
 target_link_libraries(
     vnode
+    PUBLIC os
     PUBLIC transport
     PUBLIC meta
     PUBLIC tq

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

@@ -13,20 +13,27 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef _TD_COMMON_QARITHMETICOPERATOR_H_
+#ifndef _TD_VNODE_ALLOCATOR_POOL_H_
-#define _TD_COMMON_QARITHMETICOPERATOR_H_
+#define _TD_VNODE_ALLOCATOR_POOL_H_
+
+#include "vnode.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef void (*_arithmetic_operator_fn_t)(void *left, int32_t numLeft, int32_t leftType, void *right, int32_t numRight,
+typedef struct {
-                                          int32_t rightType, void *output, int32_t order);
+  int            nexta;
+  int            enda;
+  SMemAllocator *free[3];
+  SMemAllocator *used[3];
+} SVAllocatorPool;
 
-_arithmetic_operator_fn_t getArithmeticOperatorFn(int32_t arithmeticOptr);
+int  vnodeOpenAllocatorPool(SVnode *pVnode);
+void vnodeCloseAllocatorPool(SVnode *pVnode);
 
 #ifdef __cplusplus
 }
 #endif
 
-#endif  /*_TD_COMMON_QARITHMETICOPERATOR_H_*/
+#endif /*_TD_VNODE_ALLOCATOR_POOL_H_*/

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

@@ -16,13 +16,14 @@
 #ifndef _TD_VNODE_COMMIT_H_
 #define _TD_VNODE_COMMIT_H_
 
-#include "vnodeInt.h"
+#include "vnode.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-int vnodeAsyncCommit(SVnode *pVnode);
+bool vnodeShouldCommit(SVnode *pVnode);
+int  vnodeAsyncCommit(SVnode *pVnode);
 
 #ifdef __cplusplus
 }

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

@@ -0,0 +1,52 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_VNODE_DEF_H_
+#define _TD_VNODE_DEF_H_
+
+#include "mallocator.h"
+#include "sync.h"
+#include "tlockfree.h"
+
+#include "vnode.h"
+#include "vnodeAllocatorPool.h"
+#include "vnodeCommit.h"
+#include "vnodeFileSystem.h"
+#include "vnodeOptions.h"
+#include "vnodeStateMgr.h"
+#include "vnodeSync.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+struct SVnode {
+  char*            path;
+  SVnodeOptions    options;
+  SVState          state;
+  SVAllocatorPool* pool;
+  SMemAllocator*   inuse;
+  SMeta*           pMeta;
+  STsdb*           pTsdb;
+  STQ*             pTq;
+  SVnodeSync*      pSync;
+  SVnodeFS*        pFs;
+};
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_VNODE_DEF_H_*/