Merge branch '3.0' into feature/vnode

include/common/tmsg.h

@@ -114,6 +114,9 @@ typedef enum _mgmt_table {
   TSDB_MGMT_TABLE_TABLE,
   TSDB_MGMT_TABLE_DNODE,
   TSDB_MGMT_TABLE_MNODE,
+  TSDB_MGMT_TABLE_QNODE,
+  TSDB_MGMT_TABLE_SNODE,
+  TSDB_MGMT_TABLE_BNODE,
   TSDB_MGMT_TABLE_VGROUP,
   TSDB_MGMT_TABLE_STB,
   TSDB_MGMT_TABLE_MODULE,
@@ -909,15 +912,15 @@ typedef struct {
 
 typedef struct {
   int32_t dnodeId;
-} SCreateQnodeInMsg, SDropQnodeInMsg;
+} SMCreateQnodeMsg, SMDropQnodeMsg, SDCreateQnodeMsg, SDDropQnodeMsg;
 
 typedef struct {
   int32_t dnodeId;
-} SCreateSnodeInMsg, SDropSnodeInMsg;
+} SMCreateSnodeMsg, SMDropSnodeMsg, SDCreateSnodeMsg, SDDropSnodeMsg;
 
 typedef struct {
   int32_t dnodeId;
-} SCreateBnodeInMsg, SDropBnodeInMsg;
+} SMCreateBnodeMsg, SMDropBnodeMsg, SDCreateBnodeMsg, SDDropBnodeMsg;
 
 typedef struct {
   int32_t dnodeId;

include/common/tmsgdef.h

@@ -70,6 +70,15 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_DND_CREATE_MNODE, "dnode-create-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_ALTER_MNODE, "dnode-alter-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_DROP_MNODE, "dnode-drop-mnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_CREATE_QNODE, "dnode-create-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_ALTER_QNODE, "dnode-alter-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_DROP_QNODE, "dnode-drop-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_CREATE_SNODE, "dnode-create-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_ALTER_SNODE, "dnode-alter-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_DROP_SNODE, "dnode-drop-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_CREATE_BNODE, "dnode-create-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_ALTER_BNODE, "dnode-alter-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_DND_DROP_BNODE, "dnode-drop-bnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_CREATE_VNODE, "dnode-create-vnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_ALTER_VNODE, "dnode-alter-vnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_DND_DROP_VNODE, "dnode-drop-vnode", NULL, NULL)
@@ -90,9 +99,20 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_USER, "mnode-drop-user", NULL, NULL) 
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_DNODE, "mnode-create-dnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CONFIG_DNODE, "mnode-config-dnode", NULL, NULL) 
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_DNODE, "mnode-alter-dnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_DNODE, "mnode-drop-dnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_MNODE, "mnode-create-mnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_MNODE, "mnode-alter-mnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_MNODE, "mnode-drop-mnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_QNODE, "mnode-create-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_QNODE, "mnode-alter-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_DROP_QNODE, "mnode-drop-qnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_SNODE, "mnode-create-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_SNODE, "mnode-alter-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_DROP_SNODE, "mnode-drop-snode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_CREATE_BNODE, "mnode-create-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_ALTER_BNODE, "mnode-alter-bnode", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_DROP_BNODE, "mnode-drop-bnode", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_DB, "mnode-create-db", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_DB, "mnode-drop-db", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_USE_DB, "mnode-use-db", NULL, NULL)

include/dnode/mnode/sdb/sdb.h

@@ -157,18 +157,21 @@ typedef enum {
   SDB_TRANS = 1,
   SDB_CLUSTER = 2,
   SDB_MNODE = 3,
-  SDB_DNODE = 4,
-  SDB_USER = 5,
-  SDB_AUTH = 6,
-  SDB_ACCT = 7,
-  SDB_CONSUMER = 8,
-  SDB_CGROUP = 9,
-  SDB_TOPIC = 10,
-  SDB_VGROUP = 11,
-  SDB_STB = 12,
-  SDB_DB = 13,
-  SDB_FUNC = 14,
-  SDB_MAX = 15
+  SDB_QNODE = 4,
+  SDB_SNODE = 5,
+  SDB_BNODE = 6,
+  SDB_DNODE = 7,
+  SDB_USER = 8,
+  SDB_AUTH = 9,
+  SDB_ACCT = 10,
+  SDB_CONSUMER = 11,
+  SDB_CGROUP = 12,
+  SDB_TOPIC = 13,
+  SDB_VGROUP = 14,
+  SDB_STB = 15,
+  SDB_DB = 16,
+  SDB_FUNC = 17,
+  SDB_MAX = 18
 } ESdbType;
 
 typedef struct SSdb SSdb;

include/util/taoserror.h

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

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

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

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

@@ -15,9 +15,12 @@
 
 #define _DEFAULT_SOURCE
 #include "dndDnode.h"
+#include "dndBnode.h"
+#include "dndMnode.h"
+#include "dndQnode.h"
+#include "dndSnode.h"
 #include "dndTransport.h"
 #include "dndVnodes.h"
-#include "dndMnode.h"
 
 static int32_t dndInitMgmtWorker(SDnode *pDnode);
 static void    dndCleanupMgmtWorker(SDnode *pDnode);
@@ -648,6 +651,24 @@ static void dndProcessMgmtQueue(SDnode *pDnode, SRpcMsg *pMsg) {
     case TDMT_DND_DROP_MNODE:
       code = dndProcessDropMnodeReq(pDnode, pMsg);
       break;
+    case TDMT_DND_CREATE_QNODE:
+      code = dndProcessCreateQnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_DROP_QNODE:
+      code = dndProcessDropQnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_CREATE_SNODE:
+      code = dndProcessCreateSnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_DROP_SNODE:
+      code = dndProcessDropSnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_CREATE_BNODE:
+      code = dndProcessCreateBnodeReq(pDnode, pMsg);
+      break;
+    case TDMT_DND_DROP_BNODE:
+      code = dndProcessDropBnodeReq(pDnode, pMsg);
+      break;
     case TDMT_DND_CONFIG_DNODE:
       code = dndProcessConfigDnodeReq(pDnode, pMsg);
       break;

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

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

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

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

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

@@ -37,6 +37,18 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_ALTER_MNODE_RSP)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_MNODE)] = dndProcessMgmtMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_MNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_QNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_QNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_QNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_QNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_SNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_SNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_SNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_SNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_BNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_BNODE_RSP)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_BNODE)] = dndProcessMgmtMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_DND_DROP_BNODE_RSP)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_VNODE)] = dndProcessMgmtMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_CREATE_VNODE_RSP)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_DND_ALTER_VNODE)] = dndProcessMgmtMsg;
@@ -66,6 +78,12 @@ static void dndInitMsgFp(STransMgmt *pMgmt) {
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_DNODE)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_MNODE)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_MNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_QNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_QNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_SNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_SNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_BNODE)] = dndProcessMnodeWriteMsg;
+  pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_BNODE)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_CREATE_DB)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_DROP_DB)] = dndProcessMnodeWriteMsg;
   pMgmt->msgFp[TMSG_INDEX(TDMT_MND_USE_DB)] = dndProcessMnodeWriteMsg;

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

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

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

+aux_source_directory(. STEST_SRC)
+add_executable(dnode_test_bnode ${STEST_SRC})
+target_link_libraries(
+    dnode_test_bnode
+    PUBLIC sut
+)
+
+add_test(
+    NAME dnode_test_bnode
+    COMMAND dnode_test_bnode
+)

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

+/**
+ * @file dnode.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief DNODE module dnode-msg tests
+ * @version 0.1
+ * @date 2021-12-15
+ *
+ * @copyright Copyright (c) 2021
+ *
+ */
+
+#include "base.h"
+
+class DndTestBnode : public ::testing::Test {
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+ public:
+  static void SetUpTestSuite() {
+    test.Init("/tmp/dnode_test_bnode1", 9068);
+    const char* fqdn = "localhost";
+    const char* firstEp = "localhost:9068";
+
+    server2.Start("/tmp/dnode_test_bnode2", fqdn, 9069, firstEp);
+    taosMsleep(300);
+  }
+
+  static void TearDownTestSuite() {
+    server2.Stop();
+    test.Cleanup();
+  }
+
+  static Testbase   test;
+  static TestServer server2;
+};
+
+Testbase   DndTestBnode::test;
+TestServer DndTestBnode::server2;
+
+TEST_F(DndTestBnode, 01_ShowBnode) {
+  test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+  CHECK_META("show bnodes", 3);
+
+  CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+  CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+  CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+  test.SendShowRetrieveMsg();
+  EXPECT_EQ(test.GetShowRows(), 0);
+}
+
+TEST_F(DndTestBnode, 02_Create_Bnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateBnodeMsg);
+
+    SMCreateBnodeMsg* pReq = (SMCreateBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(1);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+    CHECK_META("show bnodes", 3);
+
+    CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+    CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+    CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9068", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
+
+TEST_F(DndTestBnode, 03_Create_Bnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateBnodeMsg);
+
+    SMCreateBnodeMsg* pReq = (SMCreateBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, TSDB_CODE_MND_DNODE_NOT_EXIST);
+  }
+}
+
+TEST_F(DndTestBnode, 04_Create_Bnode) {
+  {
+    // create dnode
+    int32_t contLen = sizeof(SCreateDnodeMsg);
+
+    SCreateDnodeMsg* pReq = (SCreateDnodeMsg*)rpcMallocCont(contLen);
+    strcpy(pReq->fqdn, "localhost");
+    pReq->port = htonl(9069);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_DNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    taosMsleep(1300);
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_DNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+  }
+
+  {
+    // create bnode
+    int32_t contLen = sizeof(SMCreateBnodeMsg);
+
+    SMCreateBnodeMsg* pReq = (SMCreateBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+
+    CheckInt16(1);
+    CheckInt16(2);
+    CheckBinary("localhost:9068", TSDB_EP_LEN);
+    CheckBinary("localhost:9069", TSDB_EP_LEN);
+    CheckTimestamp();
+    CheckTimestamp();
+  }
+
+  {
+    // drop bnode
+    int32_t contLen = sizeof(SMDropBnodeMsg);
+
+    SMDropBnodeMsg* pReq = (SMDropBnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_BNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_BNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9068", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
\ No newline at end of file

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

+aux_source_directory(. QTEST_SRC)
+add_executable(dnode_test_qnode ${QTEST_SRC})
+target_link_libraries(
+    dnode_test_qnode
+    PUBLIC sut
+)
+
+add_test(
+    NAME dnode_test_qnode
+    COMMAND dnode_test_qnode
+)

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

+/**
+ * @file dnode.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief DNODE module dnode-msg tests
+ * @version 0.1
+ * @date 2021-12-15
+ *
+ * @copyright Copyright (c) 2021
+ *
+ */
+
+#include "base.h"
+
+class DndTestQnode : public ::testing::Test {
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+ public:
+  static void SetUpTestSuite() {
+    test.Init("/tmp/dnode_test_qnode1", 9064);
+    const char* fqdn = "localhost";
+    const char* firstEp = "localhost:9064";
+
+    server2.Start("/tmp/dnode_test_qnode2", fqdn, 9065, firstEp);
+    taosMsleep(300);
+  }
+
+  static void TearDownTestSuite() {
+    server2.Stop();
+    test.Cleanup();
+  }
+
+  static Testbase   test;
+  static TestServer server2;
+};
+
+Testbase   DndTestQnode::test;
+TestServer DndTestQnode::server2;
+
+TEST_F(DndTestQnode, 01_ShowQnode) {
+  test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+  CHECK_META("show qnodes", 3);
+
+  CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+  CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+  CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+  test.SendShowRetrieveMsg();
+  EXPECT_EQ(test.GetShowRows(), 0);
+}
+
+TEST_F(DndTestQnode, 02_Create_Qnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateQnodeMsg);
+
+    SMCreateQnodeMsg* pReq = (SMCreateQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(1);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+    CHECK_META("show qnodes", 3);
+
+    CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+    CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+    CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9064", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
+
+TEST_F(DndTestQnode, 03_Create_Qnode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateQnodeMsg);
+
+    SMCreateQnodeMsg* pReq = (SMCreateQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, TSDB_CODE_MND_DNODE_NOT_EXIST);
+  }
+}
+
+TEST_F(DndTestQnode, 04_Create_Qnode) {
+  {
+    // create dnode
+    int32_t contLen = sizeof(SCreateDnodeMsg);
+
+    SCreateDnodeMsg* pReq = (SCreateDnodeMsg*)rpcMallocCont(contLen);
+    strcpy(pReq->fqdn, "localhost");
+    pReq->port = htonl(9065);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_DNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    taosMsleep(1300);
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_DNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+  }
+
+  {
+    // create qnode
+    int32_t contLen = sizeof(SMCreateQnodeMsg);
+
+    SMCreateQnodeMsg* pReq = (SMCreateQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+
+    CheckInt16(1);
+    CheckInt16(2);
+    CheckBinary("localhost:9064", TSDB_EP_LEN);
+    CheckBinary("localhost:9065", TSDB_EP_LEN);
+    CheckTimestamp();
+    CheckTimestamp();
+  }
+
+  {
+    // drop qnode
+    int32_t contLen = sizeof(SMDropQnodeMsg);
+
+    SMDropQnodeMsg* pReq = (SMDropQnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_QNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_QNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9064", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
\ No newline at end of file

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

+aux_source_directory(. STEST_SRC)
+add_executable(dnode_test_snode ${STEST_SRC})
+target_link_libraries(
+    dnode_test_snode
+    PUBLIC sut
+)
+
+add_test(
+    NAME dnode_test_snode
+    COMMAND dnode_test_snode
+)

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

+/**
+ * @file dnode.cpp
+ * @author slguan (slguan@taosdata.com)
+ * @brief DNODE module dnode-msg tests
+ * @version 0.1
+ * @date 2021-12-15
+ *
+ * @copyright Copyright (c) 2021
+ *
+ */
+
+#include "base.h"
+
+class DndTestSnode : public ::testing::Test {
+ public:
+  void SetUp() override {}
+  void TearDown() override {}
+
+ public:
+  static void SetUpTestSuite() {
+    test.Init("/tmp/dnode_test_snode1", 9066);
+    const char* fqdn = "localhost";
+    const char* firstEp = "localhost:9066";
+
+    server2.Start("/tmp/dnode_test_snode2", fqdn, 9067, firstEp);
+    taosMsleep(300);
+  }
+
+  static void TearDownTestSuite() {
+    server2.Stop();
+    test.Cleanup();
+  }
+
+  static Testbase   test;
+  static TestServer server2;
+};
+
+Testbase   DndTestSnode::test;
+TestServer DndTestSnode::server2;
+
+TEST_F(DndTestSnode, 01_ShowSnode) {
+  test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+  CHECK_META("show snodes", 3);
+
+  CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+  CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+  CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+  test.SendShowRetrieveMsg();
+  EXPECT_EQ(test.GetShowRows(), 0);
+}
+
+TEST_F(DndTestSnode, 02_Create_Snode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateSnodeMsg);
+
+    SMCreateSnodeMsg* pReq = (SMCreateSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(1);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+    CHECK_META("show snodes", 3);
+
+    CHECK_SCHEMA(0, TSDB_DATA_TYPE_SMALLINT, 2, "id");
+    CHECK_SCHEMA(1, TSDB_DATA_TYPE_BINARY, TSDB_EP_LEN + VARSTR_HEADER_SIZE, "endpoint");
+    CHECK_SCHEMA(2, TSDB_DATA_TYPE_TIMESTAMP, 8, "create_time");
+
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9066", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
+
+TEST_F(DndTestSnode, 03_Create_Snode_Invalid_Id) {
+  {
+    int32_t contLen = sizeof(SMCreateSnodeMsg);
+
+    SMCreateSnodeMsg* pReq = (SMCreateSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, TSDB_CODE_MND_DNODE_NOT_EXIST);
+  }
+}
+
+TEST_F(DndTestSnode, 04_Create_Snode) {
+  {
+    // create dnode
+    int32_t contLen = sizeof(SCreateDnodeMsg);
+
+    SCreateDnodeMsg* pReq = (SCreateDnodeMsg*)rpcMallocCont(contLen);
+    strcpy(pReq->fqdn, "localhost");
+    pReq->port = htonl(9067);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_DNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    taosMsleep(1300);
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_DNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+  }
+
+  {
+    // create snode
+    int32_t contLen = sizeof(SMCreateSnodeMsg);
+
+    SMCreateSnodeMsg* pReq = (SMCreateSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_CREATE_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 2);
+
+    CheckInt16(1);
+    CheckInt16(2);
+    CheckBinary("localhost:9066", TSDB_EP_LEN);
+    CheckBinary("localhost:9067", TSDB_EP_LEN);
+    CheckTimestamp();
+    CheckTimestamp();
+  }
+
+  {
+    // drop snode
+    int32_t contLen = sizeof(SMDropSnodeMsg);
+
+    SMDropSnodeMsg* pReq = (SMDropSnodeMsg*)rpcMallocCont(contLen);
+    pReq->dnodeId = htonl(2);
+
+    SRpcMsg* pMsg = test.SendMsg(TDMT_MND_DROP_SNODE, pReq, contLen);
+    ASSERT_NE(pMsg, nullptr);
+    ASSERT_EQ(pMsg->code, 0);
+
+    test.SendShowMetaMsg(TSDB_MGMT_TABLE_SNODE, "");
+    test.SendShowRetrieveMsg();
+    EXPECT_EQ(test.GetShowRows(), 1);
+
+    CheckInt16(1);
+    CheckBinary("localhost:9066", TSDB_EP_LEN);
+    CheckTimestamp();
+  }
+}
\ No newline at end of file

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

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

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

@@ -144,6 +144,27 @@ typedef struct {
   SDnodeObj *pDnode;
 } SMnodeObj;
 
+typedef struct {
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  SDnodeObj *pDnode;
+} SQnodeObj;
+
+typedef struct {
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  SDnodeObj *pDnode;
+} SSnodeObj;
+
+typedef struct {
+  int32_t    id;
+  int64_t    createdTime;
+  int64_t    updateTime;
+  SDnodeObj *pDnode;
+} SBnodeObj;
+
 typedef struct {
   int32_t maxUsers;
   int32_t maxDbs;

source/dnode/mnode/impl/src/mndBalance.c → source/dnode/mnode/impl/inc/mndQnode.h

@@ -13,9 +13,20 @@
  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#define _DEFAULT_SOURCE
-#include "os.h"
+#ifndef _TD_MND_QNODE_H_
+#define _TD_MND_QNODE_H_
+
 #include "mndInt.h"
 
-int32_t mndInitBalance(SMnode *pMnode) { return 0; }
-void    mndCleanupBalance(SMnode *pMnode) {}
\ No newline at end of file
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+int32_t mndInitQnode(SMnode *pMnode);
+void    mndCleanupQnode(SMnode *pMnode);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_MND_QNODE_H_*/

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

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

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "mndBnode.h"
+#include "mndDnode.h"
+#include "mndShow.h"
+#include "mndTrans.h"
+
+#define TSDB_BNODE_VER_NUMBER 1
+#define TSDB_BNODE_RESERVE_SIZE 64
+
+static SSdbRaw *mndBnodeActionEncode(SBnodeObj *pObj);
+static SSdbRow *mndBnodeActionDecode(SSdbRaw *pRaw);
+static int32_t  mndBnodeActionInsert(SSdb *pSdb, SBnodeObj *pObj);
+static int32_t  mndBnodeActionDelete(SSdb *pSdb, SBnodeObj *pObj);
+static int32_t  mndBnodeActionUpdate(SSdb *pSdb, SBnodeObj *pOldBnode, SBnodeObj *pNewBnode);
+static int32_t  mndProcessCreateBnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropBnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessCreateBnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropBnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndGetBnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveBnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextBnode(SMnode *pMnode, void *pIter);
+
+int32_t mndInitBnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_BNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)mndBnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndBnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndBnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndBnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndBnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_BNODE, mndProcessCreateBnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_BNODE, mndProcessDropBnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_DND_CREATE_BNODE_RSP, mndProcessCreateBnodeRsp);
+  mndSetMsgHandle(pMnode, TDMT_DND_DROP_BNODE_RSP, mndProcessDropBnodeRsp);
+
+  mndAddShowMetaHandle(pMnode, TSDB_MGMT_TABLE_BNODE, mndGetBnodeMeta);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_BNODE, mndRetrieveBnodes);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_BNODE, mndCancelGetNextBnode);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupBnode(SMnode *pMnode) {}
+
+static SBnodeObj *mndAcquireBnode(SMnode *pMnode, int32_t snodeId) {
+  SSdb      *pSdb = pMnode->pSdb;
+  SBnodeObj *pObj = sdbAcquire(pSdb, SDB_BNODE, &snodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_BNODE_NOT_EXIST;
+  }
+  return pObj;
+}
+
+static void mndReleaseBnode(SMnode *pMnode, SBnodeObj *pObj) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pObj);
+}
+
+static SSdbRaw *mndBnodeActionEncode(SBnodeObj *pObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_BNODE, TSDB_BNODE_VER_NUMBER, sizeof(SBnodeObj) + TSDB_BNODE_RESERVE_SIZE);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_BNODE_RESERVE_SIZE)
+
+  return pRaw;
+}
+
+static SSdbRow *mndBnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != TSDB_BNODE_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode snode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SBnodeObj));
+  SBnodeObj *pObj = sdbGetRowObj(pRow);
+  if (pObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_BNODE_RESERVE_SIZE)
+
+  return pRow;
+}
+
+static int32_t mndBnodeActionInsert(SSdb *pSdb, SBnodeObj *pObj) {
+  mTrace("snode:%d, perform insert action", pObj->id);
+  pObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pObj->id);
+  if (pObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("snode:%d, failed to perform insert action since %s", pObj->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndBnodeActionDelete(SSdb *pSdb, SBnodeObj *pObj) {
+  mTrace("snode:%d, perform delete action", pObj->id);
+  if (pObj->pDnode != NULL) {
+    sdbRelease(pSdb, pObj->pDnode);
+    pObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndBnodeActionUpdate(SSdb *pSdb, SBnodeObj *pOldBnode, SBnodeObj *pNewBnode) {
+  mTrace("snode:%d, perform update action", pOldBnode->id);
+  pOldBnode->updateTime = pNewBnode->updateTime;
+  return 0;
+}
+
+static int32_t mndSetCreateBnodeRedoLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndBnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateBnodeCommitLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndBnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateBnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SBnodeObj *pObj) {
+  SDCreateBnodeMsg *pMsg = malloc(sizeof(SDCreateBnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDCreateBnodeMsg);
+  action.msgType = TDMT_DND_CREATE_BNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndCreateBnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateBnodeMsg *pCreate) {
+  SBnodeObj snodeObj = {0};
+  snodeObj.id = pDnode->id;
+  snodeObj.createdTime = taosGetTimestampMs();
+  snodeObj.updateTime = snodeObj.createdTime;
+
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+  mDebug("trans:%d, used to create snode:%d", pTrans->id, pCreate->dnodeId);
+
+  if (mndSetCreateBnodeRedoLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  if (mndSetCreateBnodeCommitLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  if (mndSetCreateBnodeRedoActions(pTrans, pDnode, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto CREATE_BNODE_OVER;
+  }
+
+  code = 0;
+
+CREATE_BNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessCreateBnodeReq(SMnodeMsg *pMsg) {
+  SMnode           *pMnode = pMsg->pMnode;
+  SMCreateBnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  pCreate->dnodeId = htonl(pCreate->dnodeId);
+
+  mDebug("snode:%d, start to create", pCreate->dnodeId);
+
+  SBnodeObj *pObj = mndAcquireBnode(pMnode, pCreate->dnodeId);
+  if (pObj != NULL) {
+    mError("snode:%d, snode already exist", pObj->id);
+    mndReleaseBnode(pMnode, pObj);
+    return -1;
+  }
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pCreate->dnodeId);
+  if (pDnode == NULL) {
+    mError("snode:%d, dnode not exist", pCreate->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndCreateBnode(pMnode, pMsg, pDnode, pCreate);
+  mndReleaseDnode(pMnode, pDnode);
+
+  if (code != 0) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndSetDropBnodeRedoLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndBnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropBnodeCommitLogs(STrans *pTrans, SBnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndBnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropBnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SBnodeObj *pObj) {
+  SDDropBnodeMsg *pMsg = malloc(sizeof(SDDropBnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDDropBnodeMsg);
+  action.msgType = TDMT_DND_DROP_BNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndDropBnode(SMnode *pMnode, SMnodeMsg *pMsg, SBnodeObj *pObj) {
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to drop since %s", pObj->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  mDebug("trans:%d, used to drop snode:%d", pTrans->id, pObj->id);
+
+  if (mndSetDropBnodeRedoLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  if (mndSetDropBnodeCommitLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  if (mndSetDropBnodeRedoActions(pTrans, pObj->pDnode, pObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto DROP_BNODE_OVER;
+  }
+
+  code = 0;
+
+DROP_BNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessDropBnodeReq(SMnodeMsg *pMsg) {
+  SMnode         *pMnode = pMsg->pMnode;
+  SMDropBnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  pDrop->dnodeId = htonl(pDrop->dnodeId);
+
+  mDebug("snode:%d, start to drop", pDrop->dnodeId);
+
+  if (pDrop->dnodeId <= 0) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    mError("snode:%d, failed to drop since %s", pDrop->dnodeId, terrstr());
+    return -1;
+  }
+
+  SBnodeObj *pObj = mndAcquireBnode(pMnode, pDrop->dnodeId);
+  if (pObj == NULL) {
+    mError("snode:%d, not exist", pDrop->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndDropBnode(pMnode, pMsg, pObj);
+  if (code != 0) {
+    mError("snode:%d, failed to drop since %s", pMnode->dnodeId, terrstr());
+    return -1;
+  }
+
+  sdbRelease(pMnode->pSdb, pMnode);
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndProcessCreateBnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessDropBnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndGetBnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = 2;
+  pSchema[cols].type = TSDB_DATA_TYPE_SMALLINT;
+  strcpy(pSchema[cols].name, "id");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_EP_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "endpoint");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  pShow->numOfColumns = cols;
+
+  pShow->offset[0] = 0;
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
+
+  pShow->numOfRows = sdbGetSize(pSdb, SDB_BNODE);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveBnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode    *pMnode = pMsg->pMnode;
+  SSdb      *pSdb = pMnode->pSdb;
+  int32_t    numOfRows = 0;
+  int32_t    cols = 0;
+  SBnodeObj *pObj = NULL;
+  char      *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_BNODE, pShow->pIter, (void **)&pObj);
+    if (pShow->pIter == NULL) break;
+
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int16_t *)pWrite = pObj->id;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pObj->pDnode->ep, pShow->bytes[cols]);
+
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pObj->createdTime;
+    cols++;
+
+    numOfRows++;
+    sdbRelease(pSdb, pObj);
+  }
+
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  pShow->numOfReads += numOfRows;
+
+  return numOfRows;
+}
+
+static void mndCancelGetNextBnode(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

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

@@ -329,7 +329,7 @@ static int32_t mndSetCreateMnodeRedoActions(SMnode *pMnode, STrans *pTrans, SDno
 
 static int32_t mndCreateMnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SCreateMnodeMsg *pCreate) {
   SMnodeObj mnodeObj = {0};
-  mnodeObj.id = sdbGetMaxId(pMnode->pSdb, SDB_MNODE);
+  mnodeObj.id = pDnode->id;
   mnodeObj.createdTime = taosGetTimestampMs();
   mnodeObj.updateTime = mnodeObj.createdTime;
 
@@ -562,7 +562,7 @@ static int32_t mndProcessDropMnodeReq(SMnodeMsg *pMsg) {
     return -1;
   }
 
-  sdbRelease(pMnode->pSdb, pMnode);
+  sdbRelease(pMnode->pSdb, pObj);
   return TSDB_CODE_MND_ACTION_IN_PROGRESS;
 }
 

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "mndQnode.h"
+#include "mndDnode.h"
+#include "mndShow.h"
+#include "mndTrans.h"
+
+#define TSDB_QNODE_VER_NUMBER 1
+#define TSDB_QNODE_RESERVE_SIZE 64
+
+static SSdbRaw *mndQnodeActionEncode(SQnodeObj *pObj);
+static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw);
+static int32_t  mndQnodeActionInsert(SSdb *pSdb, SQnodeObj *pObj);
+static int32_t  mndQnodeActionDelete(SSdb *pSdb, SQnodeObj *pObj);
+static int32_t  mndQnodeActionUpdate(SSdb *pSdb, SQnodeObj *pOldQnode, SQnodeObj *pNewQnode);
+static int32_t  mndProcessCreateQnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropQnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessCreateQnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropQnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndGetQnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveQnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextQnode(SMnode *pMnode, void *pIter);
+
+int32_t mndInitQnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_QNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)mndQnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndQnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndQnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndQnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndQnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_QNODE, mndProcessCreateQnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_QNODE, mndProcessDropQnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_DND_CREATE_QNODE_RSP, mndProcessCreateQnodeRsp);
+  mndSetMsgHandle(pMnode, TDMT_DND_DROP_QNODE_RSP, mndProcessDropQnodeRsp);
+
+  mndAddShowMetaHandle(pMnode, TSDB_MGMT_TABLE_QNODE, mndGetQnodeMeta);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_QNODE, mndRetrieveQnodes);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_QNODE, mndCancelGetNextQnode);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupQnode(SMnode *pMnode) {}
+
+static SQnodeObj *mndAcquireQnode(SMnode *pMnode, int32_t qnodeId) {
+  SSdb      *pSdb = pMnode->pSdb;
+  SQnodeObj *pObj = sdbAcquire(pSdb, SDB_QNODE, &qnodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_QNODE_NOT_EXIST;
+  }
+  return pObj;
+}
+
+static void mndReleaseQnode(SMnode *pMnode, SQnodeObj *pObj) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pObj);
+}
+
+static SSdbRaw *mndQnodeActionEncode(SQnodeObj *pObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_QNODE, TSDB_QNODE_VER_NUMBER, sizeof(SQnodeObj) + TSDB_QNODE_RESERVE_SIZE);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_QNODE_RESERVE_SIZE)
+
+  return pRaw;
+}
+
+static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != TSDB_QNODE_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode qnode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SQnodeObj));
+  SQnodeObj *pObj = sdbGetRowObj(pRow);
+  if (pObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_QNODE_RESERVE_SIZE)
+
+  return pRow;
+}
+
+static int32_t mndQnodeActionInsert(SSdb *pSdb, SQnodeObj *pObj) {
+  mTrace("qnode:%d, perform insert action", pObj->id);
+  pObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pObj->id);
+  if (pObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("qnode:%d, failed to perform insert action since %s", pObj->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndQnodeActionDelete(SSdb *pSdb, SQnodeObj *pObj) {
+  mTrace("qnode:%d, perform delete action", pObj->id);
+  if (pObj->pDnode != NULL) {
+    sdbRelease(pSdb, pObj->pDnode);
+    pObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndQnodeActionUpdate(SSdb *pSdb, SQnodeObj *pOldQnode, SQnodeObj *pNewQnode) {
+  mTrace("qnode:%d, perform update action", pOldQnode->id);
+  pOldQnode->updateTime = pNewQnode->updateTime;
+  return 0;
+}
+
+static int32_t mndSetCreateQnodeRedoLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndQnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateQnodeCommitLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndQnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateQnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SQnodeObj *pObj) {
+  SDCreateQnodeMsg *pMsg = malloc(sizeof(SDCreateQnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDCreateQnodeMsg);
+  action.msgType = TDMT_DND_CREATE_QNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndCreateQnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateQnodeMsg *pCreate) {
+  SQnodeObj qnodeObj = {0};
+  qnodeObj.id = pDnode->id;
+  qnodeObj.createdTime = taosGetTimestampMs();
+  qnodeObj.updateTime = qnodeObj.createdTime;
+
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("qnode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+  mDebug("trans:%d, used to create qnode:%d", pTrans->id, pCreate->dnodeId);
+
+  if (mndSetCreateQnodeRedoLogs(pTrans, &qnodeObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  if (mndSetCreateQnodeCommitLogs(pTrans, &qnodeObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  if (mndSetCreateQnodeRedoActions(pTrans, pDnode, &qnodeObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto CREATE_QNODE_OVER;
+  }
+
+  code = 0;
+
+CREATE_QNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessCreateQnodeReq(SMnodeMsg *pMsg) {
+  SMnode           *pMnode = pMsg->pMnode;
+  SMCreateQnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  pCreate->dnodeId = htonl(pCreate->dnodeId);
+
+  mDebug("qnode:%d, start to create", pCreate->dnodeId);
+
+  SQnodeObj *pObj = mndAcquireQnode(pMnode, pCreate->dnodeId);
+  if (pObj != NULL) {
+    mError("qnode:%d, qnode already exist", pObj->id);
+    mndReleaseQnode(pMnode, pObj);
+    return -1;
+  }
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pCreate->dnodeId);
+  if (pDnode == NULL) {
+    mError("qnode:%d, dnode not exist", pCreate->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndCreateQnode(pMnode, pMsg, pDnode, pCreate);
+  mndReleaseDnode(pMnode, pDnode);
+
+  if (code != 0) {
+    mError("qnode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndSetDropQnodeRedoLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndQnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropQnodeCommitLogs(STrans *pTrans, SQnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndQnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropQnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SQnodeObj *pObj) {
+  SDDropQnodeMsg *pMsg = malloc(sizeof(SDDropQnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDDropQnodeMsg);
+  action.msgType = TDMT_DND_DROP_QNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndDropQnode(SMnode *pMnode, SMnodeMsg *pMsg, SQnodeObj *pObj) {
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("qnode:%d, failed to drop since %s", pObj->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  mDebug("trans:%d, used to drop qnode:%d", pTrans->id, pObj->id);
+
+  if (mndSetDropQnodeRedoLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  if (mndSetDropQnodeCommitLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  if (mndSetDropQnodeRedoActions(pTrans, pObj->pDnode, pObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto DROP_QNODE_OVER;
+  }
+
+  code = 0;
+
+DROP_QNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessDropQnodeReq(SMnodeMsg *pMsg) {
+  SMnode         *pMnode = pMsg->pMnode;
+  SMDropQnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  pDrop->dnodeId = htonl(pDrop->dnodeId);
+
+  mDebug("qnode:%d, start to drop", pDrop->dnodeId);
+
+  if (pDrop->dnodeId <= 0) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    mError("qnode:%d, failed to drop since %s", pDrop->dnodeId, terrstr());
+    return -1;
+  }
+
+  SQnodeObj *pObj = mndAcquireQnode(pMnode, pDrop->dnodeId);
+  if (pObj == NULL) {
+    mError("qnode:%d, not exist", pDrop->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndDropQnode(pMnode, pMsg, pObj);
+  if (code != 0) {
+    mError("qnode:%d, failed to drop since %s", pMnode->dnodeId, terrstr());
+    return -1;
+  }
+
+  sdbRelease(pMnode->pSdb, pMnode);
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndProcessCreateQnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessDropQnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndGetQnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = 2;
+  pSchema[cols].type = TSDB_DATA_TYPE_SMALLINT;
+  strcpy(pSchema[cols].name, "id");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_EP_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "endpoint");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  pShow->numOfColumns = cols;
+
+  pShow->offset[0] = 0;
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
+
+  pShow->numOfRows = sdbGetSize(pSdb, SDB_QNODE);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveQnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode    *pMnode = pMsg->pMnode;
+  SSdb      *pSdb = pMnode->pSdb;
+  int32_t    numOfRows = 0;
+  int32_t    cols = 0;
+  SQnodeObj *pObj = NULL;
+  char      *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_QNODE, pShow->pIter, (void **)&pObj);
+    if (pShow->pIter == NULL) break;
+
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int16_t *)pWrite = pObj->id;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pObj->pDnode->ep, pShow->bytes[cols]);
+
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pObj->createdTime;
+    cols++;
+
+    numOfRows++;
+    sdbRelease(pSdb, pObj);
+  }
+
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  pShow->numOfReads += numOfRows;
+
+  return numOfRows;
+}
+
+static void mndCancelGetNextQnode(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

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

@@ -262,6 +262,12 @@ char *mndShowStr(int32_t showType) {
       return "show dnodes";
     case TSDB_MGMT_TABLE_MNODE:
       return "show mnodes";
+    case TSDB_MGMT_TABLE_QNODE:
+      return "show qnodes";
+    case TSDB_MGMT_TABLE_SNODE:
+      return "show snodes";
+    case TSDB_MGMT_TABLE_BNODE:
+      return "show bnodes";
     case TSDB_MGMT_TABLE_VGROUP:
       return "show vgroups";
     case TSDB_MGMT_TABLE_STB:

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

+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#define _DEFAULT_SOURCE
+#include "mndSnode.h"
+#include "mndDnode.h"
+#include "mndShow.h"
+#include "mndTrans.h"
+
+#define TSDB_SNODE_VER_NUMBER 1
+#define TSDB_SNODE_RESERVE_SIZE 64
+
+static SSdbRaw *mndSnodeActionEncode(SSnodeObj *pObj);
+static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw);
+static int32_t  mndSnodeActionInsert(SSdb *pSdb, SSnodeObj *pObj);
+static int32_t  mndSnodeActionDelete(SSdb *pSdb, SSnodeObj *pObj);
+static int32_t  mndSnodeActionUpdate(SSdb *pSdb, SSnodeObj *pOldSnode, SSnodeObj *pNewSnode);
+static int32_t  mndProcessCreateSnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropSnodeReq(SMnodeMsg *pMsg);
+static int32_t  mndProcessCreateSnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndProcessDropSnodeRsp(SMnodeMsg *pMsg);
+static int32_t  mndGetSnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta);
+static int32_t  mndRetrieveSnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows);
+static void     mndCancelGetNextSnode(SMnode *pMnode, void *pIter);
+
+int32_t mndInitSnode(SMnode *pMnode) {
+  SSdbTable table = {.sdbType = SDB_SNODE,
+                     .keyType = SDB_KEY_INT32,
+                     .encodeFp = (SdbEncodeFp)mndSnodeActionEncode,
+                     .decodeFp = (SdbDecodeFp)mndSnodeActionDecode,
+                     .insertFp = (SdbInsertFp)mndSnodeActionInsert,
+                     .updateFp = (SdbUpdateFp)mndSnodeActionUpdate,
+                     .deleteFp = (SdbDeleteFp)mndSnodeActionDelete};
+
+  mndSetMsgHandle(pMnode, TDMT_MND_CREATE_SNODE, mndProcessCreateSnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_MND_DROP_SNODE, mndProcessDropSnodeReq);
+  mndSetMsgHandle(pMnode, TDMT_DND_CREATE_SNODE_RSP, mndProcessCreateSnodeRsp);
+  mndSetMsgHandle(pMnode, TDMT_DND_DROP_SNODE_RSP, mndProcessDropSnodeRsp);
+
+  mndAddShowMetaHandle(pMnode, TSDB_MGMT_TABLE_SNODE, mndGetSnodeMeta);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_SNODE, mndRetrieveSnodes);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_SNODE, mndCancelGetNextSnode);
+
+  return sdbSetTable(pMnode->pSdb, table);
+}
+
+void mndCleanupSnode(SMnode *pMnode) {}
+
+static SSnodeObj *mndAcquireSnode(SMnode *pMnode, int32_t snodeId) {
+  SSdb      *pSdb = pMnode->pSdb;
+  SSnodeObj *pObj = sdbAcquire(pSdb, SDB_SNODE, &snodeId);
+  if (pObj == NULL) {
+    terrno = TSDB_CODE_MND_SNODE_NOT_EXIST;
+  }
+  return pObj;
+}
+
+static void mndReleaseSnode(SMnode *pMnode, SSnodeObj *pObj) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbRelease(pSdb, pObj);
+}
+
+static SSdbRaw *mndSnodeActionEncode(SSnodeObj *pObj) {
+  SSdbRaw *pRaw = sdbAllocRaw(SDB_SNODE, TSDB_SNODE_VER_NUMBER, sizeof(SSnodeObj) + TSDB_SNODE_RESERVE_SIZE);
+  if (pRaw == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_SET_INT32(pRaw, dataPos, pObj->id);
+  SDB_SET_INT64(pRaw, dataPos, pObj->createdTime)
+  SDB_SET_INT64(pRaw, dataPos, pObj->updateTime)
+  SDB_SET_RESERVE(pRaw, dataPos, TSDB_SNODE_RESERVE_SIZE)
+
+  return pRaw;
+}
+
+static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw) {
+  int8_t sver = 0;
+  if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
+
+  if (sver != TSDB_SNODE_VER_NUMBER) {
+    terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
+    mError("failed to decode snode since %s", terrstr());
+    return NULL;
+  }
+
+  SSdbRow   *pRow = sdbAllocRow(sizeof(SSnodeObj));
+  SSnodeObj *pObj = sdbGetRowObj(pRow);
+  if (pObj == NULL) return NULL;
+
+  int32_t dataPos = 0;
+  SDB_GET_INT32(pRaw, pRow, dataPos, &pObj->id)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->createdTime)
+  SDB_GET_INT64(pRaw, pRow, dataPos, &pObj->updateTime)
+  SDB_GET_RESERVE(pRaw, pRow, dataPos, TSDB_SNODE_RESERVE_SIZE)
+
+  return pRow;
+}
+
+static int32_t mndSnodeActionInsert(SSdb *pSdb, SSnodeObj *pObj) {
+  mTrace("snode:%d, perform insert action", pObj->id);
+  pObj->pDnode = sdbAcquire(pSdb, SDB_DNODE, &pObj->id);
+  if (pObj->pDnode == NULL) {
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    mError("snode:%d, failed to perform insert action since %s", pObj->id, terrstr());
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndSnodeActionDelete(SSdb *pSdb, SSnodeObj *pObj) {
+  mTrace("snode:%d, perform delete action", pObj->id);
+  if (pObj->pDnode != NULL) {
+    sdbRelease(pSdb, pObj->pDnode);
+    pObj->pDnode = NULL;
+  }
+
+  return 0;
+}
+
+static int32_t mndSnodeActionUpdate(SSdb *pSdb, SSnodeObj *pOldSnode, SSnodeObj *pNewSnode) {
+  mTrace("snode:%d, perform update action", pOldSnode->id);
+  pOldSnode->updateTime = pNewSnode->updateTime;
+  return 0;
+}
+
+static int32_t mndSetCreateSnodeRedoLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndSnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_CREATING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateSnodeCommitLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndSnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_READY) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetCreateSnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SSnodeObj *pObj) {
+  SDCreateSnodeMsg *pMsg = malloc(sizeof(SDCreateSnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDCreateSnodeMsg);
+  action.msgType = TDMT_DND_CREATE_SNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndCreateSnode(SMnode *pMnode, SMnodeMsg *pMsg, SDnodeObj *pDnode, SMCreateSnodeMsg *pCreate) {
+  SSnodeObj snodeObj = {0};
+  snodeObj.id = pDnode->id;
+  snodeObj.createdTime = taosGetTimestampMs();
+  snodeObj.updateTime = snodeObj.createdTime;
+
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+  mDebug("trans:%d, used to create snode:%d", pTrans->id, pCreate->dnodeId);
+
+  if (mndSetCreateSnodeRedoLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  if (mndSetCreateSnodeCommitLogs(pTrans, &snodeObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  if (mndSetCreateSnodeRedoActions(pTrans, pDnode, &snodeObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto CREATE_SNODE_OVER;
+  }
+
+  code = 0;
+
+CREATE_SNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessCreateSnodeReq(SMnodeMsg *pMsg) {
+  SMnode           *pMnode = pMsg->pMnode;
+  SMCreateSnodeMsg *pCreate = pMsg->rpcMsg.pCont;
+
+  pCreate->dnodeId = htonl(pCreate->dnodeId);
+
+  mDebug("snode:%d, start to create", pCreate->dnodeId);
+
+  SSnodeObj *pObj = mndAcquireSnode(pMnode, pCreate->dnodeId);
+  if (pObj != NULL) {
+    mError("snode:%d, snode already exist", pObj->id);
+    mndReleaseSnode(pMnode, pObj);
+    return -1;
+  }
+
+  SDnodeObj *pDnode = mndAcquireDnode(pMnode, pCreate->dnodeId);
+  if (pDnode == NULL) {
+    mError("snode:%d, dnode not exist", pCreate->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndCreateSnode(pMnode, pMsg, pDnode, pCreate);
+  mndReleaseDnode(pMnode, pDnode);
+
+  if (code != 0) {
+    mError("snode:%d, failed to create since %s", pCreate->dnodeId, terrstr());
+    return -1;
+  }
+
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndSetDropSnodeRedoLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pRedoRaw = mndSnodeActionEncode(pObj);
+  if (pRedoRaw == NULL) return -1;
+  if (mndTransAppendRedolog(pTrans, pRedoRaw) != 0) return -1;
+  if (sdbSetRawStatus(pRedoRaw, SDB_STATUS_DROPPING) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropSnodeCommitLogs(STrans *pTrans, SSnodeObj *pObj) {
+  SSdbRaw *pCommitRaw = mndSnodeActionEncode(pObj);
+  if (pCommitRaw == NULL) return -1;
+  if (mndTransAppendCommitlog(pTrans, pCommitRaw) != 0) return -1;
+  if (sdbSetRawStatus(pCommitRaw, SDB_STATUS_DROPPED) != 0) return -1;
+  return 0;
+}
+
+static int32_t mndSetDropSnodeRedoActions(STrans *pTrans, SDnodeObj *pDnode, SSnodeObj *pObj) {
+  SDDropSnodeMsg *pMsg = malloc(sizeof(SDDropSnodeMsg));
+  if (pMsg == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+  pMsg->dnodeId = htonl(pDnode->id);
+
+  STransAction action = {0};
+  action.epSet = mndGetDnodeEpset(pDnode);
+  action.pCont = pMsg;
+  action.contLen = sizeof(SDDropSnodeMsg);
+  action.msgType = TDMT_DND_DROP_SNODE;
+
+  if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+    free(pMsg);
+    return -1;
+  }
+
+  return 0;
+}
+
+static int32_t mndDropSnode(SMnode *pMnode, SMnodeMsg *pMsg, SSnodeObj *pObj) {
+  int32_t code = -1;
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, &pMsg->rpcMsg);
+  if (pTrans == NULL) {
+    mError("snode:%d, failed to drop since %s", pObj->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  mDebug("trans:%d, used to drop snode:%d", pTrans->id, pObj->id);
+
+  if (mndSetDropSnodeRedoLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set redo log since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  if (mndSetDropSnodeCommitLogs(pTrans, pObj) != 0) {
+    mError("trans:%d, failed to set commit log since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  if (mndSetDropSnodeRedoActions(pTrans, pObj->pDnode, pObj) != 0) {
+    mError("trans:%d, failed to set redo actions since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
+    goto DROP_SNODE_OVER;
+  }
+
+  code = 0;
+
+DROP_SNODE_OVER:
+  mndTransDrop(pTrans);
+  return code;
+}
+
+static int32_t mndProcessDropSnodeReq(SMnodeMsg *pMsg) {
+  SMnode         *pMnode = pMsg->pMnode;
+  SMDropSnodeMsg *pDrop = pMsg->rpcMsg.pCont;
+  pDrop->dnodeId = htonl(pDrop->dnodeId);
+
+  mDebug("snode:%d, start to drop", pDrop->dnodeId);
+
+  if (pDrop->dnodeId <= 0) {
+    terrno = TSDB_CODE_SDB_APP_ERROR;
+    mError("snode:%d, failed to drop since %s", pDrop->dnodeId, terrstr());
+    return -1;
+  }
+
+  SSnodeObj *pObj = mndAcquireSnode(pMnode, pDrop->dnodeId);
+  if (pObj == NULL) {
+    mError("snode:%d, not exist", pDrop->dnodeId);
+    terrno = TSDB_CODE_MND_DNODE_NOT_EXIST;
+    return -1;
+  }
+
+  int32_t code = mndDropSnode(pMnode, pMsg, pObj);
+  if (code != 0) {
+    mError("snode:%d, failed to drop since %s", pMnode->dnodeId, terrstr());
+    return -1;
+  }
+
+  sdbRelease(pMnode->pSdb, pMnode);
+  return TSDB_CODE_MND_ACTION_IN_PROGRESS;
+}
+
+static int32_t mndProcessCreateSnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndProcessDropSnodeRsp(SMnodeMsg *pMsg) {
+  mndTransProcessRsp(pMsg);
+  return 0;
+}
+
+static int32_t mndGetSnodeMeta(SMnodeMsg *pMsg, SShowObj *pShow, STableMetaMsg *pMeta) {
+  SMnode *pMnode = pMsg->pMnode;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  int32_t  cols = 0;
+  SSchema *pSchema = pMeta->pSchema;
+
+  pShow->bytes[cols] = 2;
+  pSchema[cols].type = TSDB_DATA_TYPE_SMALLINT;
+  strcpy(pSchema[cols].name, "id");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = TSDB_EP_LEN + VARSTR_HEADER_SIZE;
+  pSchema[cols].type = TSDB_DATA_TYPE_BINARY;
+  strcpy(pSchema[cols].name, "endpoint");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pShow->bytes[cols] = 8;
+  pSchema[cols].type = TSDB_DATA_TYPE_TIMESTAMP;
+  strcpy(pSchema[cols].name, "create_time");
+  pSchema[cols].bytes = htonl(pShow->bytes[cols]);
+  cols++;
+
+  pMeta->numOfColumns = htonl(cols);
+  pShow->numOfColumns = cols;
+
+  pShow->offset[0] = 0;
+  for (int32_t i = 1; i < cols; ++i) {
+    pShow->offset[i] = pShow->offset[i - 1] + pShow->bytes[i - 1];
+  }
+
+  pShow->numOfRows = sdbGetSize(pSdb, SDB_SNODE);
+  pShow->rowSize = pShow->offset[cols - 1] + pShow->bytes[cols - 1];
+  strcpy(pMeta->tbFname, mndShowStr(pShow->type));
+
+  return 0;
+}
+
+static int32_t mndRetrieveSnodes(SMnodeMsg *pMsg, SShowObj *pShow, char *data, int32_t rows) {
+  SMnode    *pMnode = pMsg->pMnode;
+  SSdb      *pSdb = pMnode->pSdb;
+  int32_t    numOfRows = 0;
+  int32_t    cols = 0;
+  SSnodeObj *pObj = NULL;
+  char      *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_SNODE, pShow->pIter, (void **)&pObj);
+    if (pShow->pIter == NULL) break;
+
+    cols = 0;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int16_t *)pWrite = pObj->id;
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    STR_WITH_MAXSIZE_TO_VARSTR(pWrite, pObj->pDnode->ep, pShow->bytes[cols]);
+
+    cols++;
+
+    pWrite = data + pShow->offset[cols] * rows + pShow->bytes[cols] * numOfRows;
+    *(int64_t *)pWrite = pObj->createdTime;
+    cols++;
+
+    numOfRows++;
+    sdbRelease(pSdb, pObj);
+  }
+
+  mndVacuumResult(data, pShow->numOfColumns, numOfRows, rows, pShow);
+  pShow->numOfReads += numOfRows;
+
+  return numOfRows;
+}
+
+static void mndCancelGetNextSnode(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}

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

@@ -273,7 +273,7 @@ static SArray *mndBuildDnodesArray(SMnode *pMnode) {
 static int32_t mndCompareDnodeVnodes(SDnodeObj *pDnode1, SDnodeObj *pDnode2) {
   float d1Score = (float)pDnode1->numOfVnodes / pDnode1->numOfSupportVnodes;
   float d2Score = (float)pDnode2->numOfVnodes / pDnode2->numOfSupportVnodes;
-  return d1Score > d2Score ? 0 : 1;
+  return d1Score > d2Score ? 1 : 0;
 }
 
 static int32_t mndGetAvailableDnode(SMnode *pMnode, SVgObj *pVgroup, SArray *pArray) {

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

@@ -16,14 +16,16 @@
 #define _DEFAULT_SOURCE
 #include "mndAcct.h"
 #include "mndAuth.h"
-#include "mndBalance.h"
+#include "mndBnode.h"
 #include "mndCluster.h"
 #include "mndDb.h"
 #include "mndDnode.h"
 #include "mndFunc.h"
 #include "mndMnode.h"
 #include "mndProfile.h"
+#include "mndQnode.h"
 #include "mndShow.h"
+#include "mndSnode.h"
 #include "mndStb.h"
 #include "mndSync.h"
 #include "mndTelem.h"
@@ -147,6 +149,9 @@ static int32_t mndInitSteps(SMnode *pMnode) {
   if (mndAllocStep(pMnode, "mnode-trans", mndInitTrans, mndCleanupTrans) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-cluster", mndInitCluster, mndCleanupCluster) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-mnode", mndInitMnode, mndCleanupMnode) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-qnode", mndInitQnode, mndCleanupQnode) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-qnode", mndInitSnode, mndCleanupSnode) != 0) return -1;
+  if (mndAllocStep(pMnode, "mnode-qnode", mndInitBnode, mndCleanupBnode) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-dnode", mndInitDnode, mndCleanupDnode) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-user", mndInitUser, mndCleanupUser) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-auth", mndInitAuth, mndCleanupAuth) != 0) return -1;
@@ -162,7 +167,6 @@ static int32_t mndInitSteps(SMnode *pMnode) {
     if (mndAllocStep(pMnode, "mnode-sdb-read", mndReadSdb, NULL) != 0) return -1;
   }
   if (mndAllocStep(pMnode, "mnode-timer", mndInitTimer, NULL) != 0) return -1;
-  if (mndAllocStep(pMnode, "mnode-balance", mndInitBalance, mndCleanupBalance) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-profile", mndInitProfile, mndCleanupProfile) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-show", mndInitShow, mndCleanupShow) != 0) return -1;
   if (mndAllocStep(pMnode, "mnode-sync", mndInitSync, mndCleanupSync) != 0) return -1;

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

@@ -24,6 +24,12 @@ static const char *sdbTableName(ESdbType type) {
       return "cluster";
     case SDB_MNODE:
       return "mnode";
+    case SDB_QNODE:
+      return "qnode";
+    case SDB_SNODE:
+      return "snode";
+    case SDB_BNODE:
+      return "bnode";
     case SDB_DNODE:
       return "dnode";
     case SDB_USER:
@@ -55,7 +61,8 @@ void sdbPrintOper(SSdb *pSdb, SSdbRow *pRow, const char *oper) {
   } else if (keyType == SDB_KEY_INT32) {
     mTrace("%s:%d, refCount:%d oper:%s", sdbTableName(pRow->type), *(int32_t *)pRow->pObj, pRow->refCount, oper);
   } else if (keyType == SDB_KEY_INT64) {
-    mTrace("%s:%" PRId64 ", refCount:%d oper:%s", sdbTableName(pRow->type), *(int64_t *)pRow->pObj, pRow->refCount, oper);
+    mTrace("%s:%" PRId64 ", refCount:%d oper:%s", sdbTableName(pRow->type), *(int64_t *)pRow->pObj, pRow->refCount,
+           oper);
   } else {
   }
 }

source/dnode/qnode/src/qnode.c

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

source/dnode/snode/src/snode.c

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

source/libs/index/inc/index_fst_counting_writer.h

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

source/libs/index/inc/index_tfile.h

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

source/libs/index/src/index.c

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

source/libs/index/src/index_cache.c

@@ -20,7 +20,7 @@
 
 #define MAX_INDEX_KEY_LEN 256  // test only, change later
 
-#define MEM_TERM_LIMIT 1000000
+#define MEM_TERM_LIMIT 200
 // ref index_cache.h:22
 //#define CACHE_KEY_LEN(p) \
 //  (sizeof(int32_t) + sizeof(uint16_t) + sizeof(p->colType) + sizeof(p->nColVal) + p->nColVal + sizeof(uint64_t) +
@@ -66,19 +66,43 @@ void indexCacheDebug(IndexCache* cache) {
   indexMemRef(tbl);
   pthread_mutex_unlock(&cache->mtx);
 
-  SSkipList*         slt = tbl->mem;
-  SSkipListIterator* iter = tSkipListCreateIter(slt);
-  while (tSkipListIterNext(iter)) {
-    SSkipListNode* node = tSkipListIterGet(iter);
-    CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
-    if (ct != NULL) {
-      // TODO, add more debug info
-      indexInfo("{colVal: %s, version: %d} \t", ct->colVal, ct->version);
+  {
+    SSkipList*         slt = tbl->mem;
+    SSkipListIterator* iter = tSkipListCreateIter(slt);
+    while (tSkipListIterNext(iter)) {
+      SSkipListNode* node = tSkipListIterGet(iter);
+      CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
+      if (ct != NULL) {
+        // TODO, add more debug info
+        indexInfo("{colVal: %s, version: %d} \t", ct->colVal, ct->version);
+      }
     }
+    tSkipListDestroyIter(iter);
+
+    indexMemUnRef(tbl);
   }
-  tSkipListDestroyIter(iter);
 
-  indexMemUnRef(tbl);
+  {
+    pthread_mutex_lock(&cache->mtx);
+    tbl = cache->imm;
+    indexMemRef(tbl);
+    pthread_mutex_unlock(&cache->mtx);
+    if (tbl != NULL) {
+      SSkipList*         slt = tbl->mem;
+      SSkipListIterator* iter = tSkipListCreateIter(slt);
+      while (tSkipListIterNext(iter)) {
+        SSkipListNode* node = tSkipListIterGet(iter);
+        CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
+        if (ct != NULL) {
+          // TODO, add more debug info
+          indexInfo("{colVal: %s, version: %d} \t", ct->colVal, ct->version);
+        }
+      }
+      tSkipListDestroyIter(iter);
+    }
+
+    indexMemUnRef(tbl);
+  }
 }
 
 void indexCacheDestroySkiplist(SSkipList* slt) {
@@ -86,25 +110,26 @@ void indexCacheDestroySkiplist(SSkipList* slt) {
   while (tSkipListIterNext(iter)) {
     SSkipListNode* node = tSkipListIterGet(iter);
     CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
-    if (ct != NULL) {
-    }
+    if (ct != NULL) {}
   }
   tSkipListDestroyIter(iter);
   tSkipListDestroy(slt);
 }
 void indexCacheDestroyImm(IndexCache* cache) {
+  if (cache == NULL) { return; }
+
   MemTable* tbl = NULL;
   pthread_mutex_lock(&cache->mtx);
   tbl = cache->imm;
   cache->imm = NULL;  // or throw int bg thread
   pthread_mutex_unlock(&cache->mtx);
+
+  indexMemUnRef(tbl);
   indexMemUnRef(tbl);
 }
 void indexCacheDestroy(void* cache) {
   IndexCache* pCache = cache;
-  if (pCache == NULL) {
-    return;
-  }
+  if (pCache == NULL) { return; }
   indexMemUnRef(pCache->mem);
   indexMemUnRef(pCache->imm);
   free(pCache->colName);
@@ -114,9 +139,11 @@ void indexCacheDestroy(void* cache) {
 
 Iterate* indexCacheIteratorCreate(IndexCache* cache) {
   Iterate* iiter = calloc(1, sizeof(Iterate));
-  if (iiter == NULL) {
-    return NULL;
-  }
+  if (iiter == NULL) { return NULL; }
+
+  pthread_mutex_lock(&cache->mtx);
+
+  indexMemRef(cache->imm);
 
   MemTable* tbl = cache->imm;
   iiter->val.val = taosArrayInit(1, sizeof(uint64_t));
@@ -124,12 +151,12 @@ Iterate* indexCacheIteratorCreate(IndexCache* cache) {
   iiter->next = indexCacheIteratorNext;
   iiter->getValue = indexCacheIteratorGetValue;
 
+  pthread_mutex_unlock(&cache->mtx);
+
   return iiter;
 }
 void indexCacheIteratorDestroy(Iterate* iter) {
-  if (iter == NULL) {
-    return;
-  }
+  if (iter == NULL) { return; }
   tSkipListDestroyIter(iter->iter);
   iterateValueDestroy(&iter->val, true);
   free(iter);
@@ -155,6 +182,7 @@ static void indexCacheMakeRoomForWrite(IndexCache* cache) {
       taosMsleep(50);
       pthread_mutex_lock(&cache->mtx);
     } else {
+      indexCacheRef(cache);
       cache->imm = cache->mem;
       cache->mem = indexInternalCacheCreate(cache->type);
       cache->nTerm = 1;
@@ -166,17 +194,13 @@ static void indexCacheMakeRoomForWrite(IndexCache* cache) {
 }
 
 int indexCachePut(void* cache, SIndexTerm* term, uint64_t uid) {
-  if (cache == NULL) {
-    return -1;
-  }
+  if (cache == NULL) { return -1; }
 
   IndexCache* pCache = cache;
   indexCacheRef(pCache);
   // encode data
   CacheTerm* ct = calloc(1, sizeof(CacheTerm));
-  if (cache == NULL) {
-    return -1;
-  }
+  if (cache == NULL) { return -1; }
   // set up key
   ct->colType = term->colType;
   ct->colVal = (char*)calloc(1, sizeof(char) * (term->nColVal + 1));
@@ -205,32 +229,11 @@ int indexCacheDel(void* cache, const char* fieldValue, int32_t fvlen, uint64_t u
   IndexCache* pCache = cache;
   return 0;
 }
-int indexCacheSearch(void* cache, SIndexTermQuery* query, SArray* result, STermValueType* s) {
-  if (cache == NULL) {
-    return -1;
-  }
-  IndexCache*     pCache = cache;
-  SIndexTerm*     term = query->term;
-  EIndexQueryType qtype = query->qType;
-
-  MemTable *mem = NULL, *imm = NULL;
-  pthread_mutex_lock(&pCache->mtx);
-  mem = pCache->mem;
-  imm = pCache->imm;
-  indexMemRef(mem);
-  indexMemRef(imm);
-  pthread_mutex_unlock(&pCache->mtx);
-
-  CacheTerm* ct = calloc(1, sizeof(CacheTerm));
-  if (ct == NULL) {
-    return -1;
-  }
-  ct->colVal = calloc(1, sizeof(char) * (term->nColVal + 1));
-  memcpy(ct->colVal, term->colVal, term->nColVal);
-  ct->version = atomic_load_32(&pCache->version);
 
+static int indexQueryMem(MemTable* mem, CacheTerm* ct, EIndexQueryType qtype, SArray* result, STermValueType* s) {
+  if (mem == NULL) { return 0; }
   char* key = getIndexKey(ct);
-  // TODO handle multi situation later, and refactor
+
   SSkipListIterator* iter = tSkipListCreateIterFromVal(mem->mem, key, TSDB_DATA_TYPE_BINARY, TSDB_ORDER_ASC);
   while (tSkipListIterNext(iter)) {
     SSkipListNode* node = tSkipListIterGet(iter);
@@ -251,51 +254,56 @@ int indexCacheSearch(void* cache, SIndexTermQuery* query, SArray* result, STermV
     }
   }
   tSkipListDestroyIter(iter);
-  cacheTermDestroy(ct);
-  // int32_t keyLen = CACHE_KEY_LEN(term);
-  // char*   buf = calloc(1, keyLen);
-  if (qtype == QUERY_TERM) {
-    //
-  } else if (qtype == QUERY_PREFIX) {
-    //
-  } else if (qtype == QUERY_SUFFIX) {
-    //
-  } else if (qtype == QUERY_REGEX) {
-    //
+  return 0;
+}
+int indexCacheSearch(void* cache, SIndexTermQuery* query, SArray* result, STermValueType* s) {
+  if (cache == NULL) { return -1; }
+  IndexCache* pCache = cache;
+
+  MemTable *mem = NULL, *imm = NULL;
+  pthread_mutex_lock(&pCache->mtx);
+  mem = pCache->mem;
+  imm = pCache->imm;
+  indexMemRef(mem);
+  indexMemRef(imm);
+  pthread_mutex_unlock(&pCache->mtx);
+
+  SIndexTerm*     term = query->term;
+  EIndexQueryType qtype = query->qType;
+  CacheTerm       ct = {.colVal = term->colVal, .version = atomic_load_32(&pCache->version)};
+  indexCacheDebug(pCache);
+
+  int ret = indexQueryMem(mem, &ct, qtype, result, s);
+  if (ret == 0 && *s != kTypeDeletion) {
+    // continue search in imm
+    ret = indexQueryMem(imm, &ct, qtype, result, s);
   }
+  // cacheTermDestroy(ct);
+
   indexMemUnRef(mem);
   indexMemUnRef(imm);
-  return 0;
+
+  return ret;
 }
 
 void indexCacheRef(IndexCache* cache) {
-  if (cache == NULL) {
-    return;
-  }
+  if (cache == NULL) { return; }
   int ref = T_REF_INC(cache);
   UNUSED(ref);
 }
 void indexCacheUnRef(IndexCache* cache) {
-  if (cache == NULL) {
-    return;
-  }
+  if (cache == NULL) { return; }
   int ref = T_REF_DEC(cache);
-  if (ref == 0) {
-    indexCacheDestroy(cache);
-  }
+  if (ref == 0) { indexCacheDestroy(cache); }
 }
 
 void indexMemRef(MemTable* tbl) {
-  if (tbl == NULL) {
-    return;
-  }
+  if (tbl == NULL) { return; }
   int ref = T_REF_INC(tbl);
   UNUSED(ref);
 }
 void indexMemUnRef(MemTable* tbl) {
-  if (tbl == NULL) {
-    return;
-  }
+  if (tbl == NULL) { return; }
   int ref = T_REF_DEC(tbl);
   if (ref == 0) {
     SSkipList* slt = tbl->mem;
@@ -305,9 +313,7 @@ void indexMemUnRef(MemTable* tbl) {
 }
 
 static void cacheTermDestroy(CacheTerm* ct) {
-  if (ct == NULL) {
-    return;
-  }
+  if (ct == NULL) { return; }
   free(ct->colVal);
   free(ct);
 }
@@ -322,9 +328,7 @@ static int32_t compareKey(const void* l, const void* r) {
 
   // compare colVal
   int32_t cmp = strcmp(lt->colVal, rt->colVal);
-  if (cmp == 0) {
-    return rt->version - lt->version;
-  }
+  if (cmp == 0) { return rt->version - lt->version; }
   return cmp;
 }
 
@@ -344,9 +348,7 @@ static void doMergeWork(SSchedMsg* msg) {
 }
 static bool indexCacheIteratorNext(Iterate* itera) {
   SSkipListIterator* iter = itera->iter;
-  if (iter == NULL) {
-    return false;
-  }
+  if (iter == NULL) { return false; }
   IterateValue* iv = &itera->val;
   iterateValueDestroy(iv, false);
 
@@ -356,7 +358,8 @@ static bool indexCacheIteratorNext(Iterate* itera) {
     CacheTerm*     ct = (CacheTerm*)SL_GET_NODE_DATA(node);
 
     iv->type = ct->operaType;
-    iv->colVal = ct->colVal;
+    iv->colVal = calloc(1, strlen(ct->colVal) + 1);
+    memcpy(iv->colVal, ct->colVal, strlen(ct->colVal));
 
     taosArrayPush(iv->val, &ct->uid);
   }

source/libs/index/src/index_fst_automation.c

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

source/libs/index/src/index_fst_counting_writer.c

@@ -52,7 +52,7 @@ static int writeCtxDoReadFrom(WriterCtx* ctx, uint8_t* buf, int len, int32_t off
 }
 static int writeCtxDoFlush(WriterCtx* ctx) {
   if (ctx->type == TFile) {
-    // tfFsync(ctx->fd);
+    tfFsync(ctx->file.fd);
     // tfFlush(ctx->file.fd);
   } else {
     // do nothing
@@ -73,6 +73,7 @@ WriterCtx* writerCtxCreate(WriterType type, const char* path, bool readOnly, int
     } else {
       ctx->file.fd = tfOpenReadWrite(path);
     }
+    memcpy(ctx->file.buf, path, strlen(path));
     if (ctx->file.fd < 0) {
       indexError("open file error %d", errno);
       goto END;
@@ -95,11 +96,15 @@ END:
   free(ctx);
   return NULL;
 }
-void writerCtxDestroy(WriterCtx* ctx) {
+void writerCtxDestroy(WriterCtx* ctx, bool remove) {
   if (ctx->type == TMemory) {
     free(ctx->mem.buf);
   } else {
     tfClose(ctx->file.fd);
+    if (remove) {
+      indexError("rm file %s", ctx->file.buf);
+      unlink(ctx->file.buf);
+    }
   }
   free(ctx);
 }
@@ -138,10 +143,8 @@ int fstCountingWriterRead(FstCountingWriter* write, uint8_t* buf, uint32_t len)
   return nRead;
 }
 
-uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write) {
-  return 0;
-}
-int fstCountingWriterFlush(FstCountingWriter* write) {
+uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter* write) { return 0; }
+int      fstCountingWriterFlush(FstCountingWriter* write) {
   WriterCtx* ctx = write->wrt;
   ctx->flush(ctx);
   // write->wtr->flush

source/libs/index/src/index_tfile.c

@@ -32,6 +32,7 @@ typedef struct TFileFstIter {
 
 #define TF_TABLE_TATOAL_SIZE(sz) (sizeof(sz) + sz * sizeof(uint64_t))
 
+static int  tfileUidCompare(const void* a, const void* b);
 static int  tfileStrCompare(const void* a, const void* b);
 static int  tfileValueCompare(const void* a, const void* b, const void* param);
 static void tfileSerialTableIdsToBuf(char* buf, SArray* tableIds);
@@ -52,11 +53,16 @@ static int  tfileParseFileName(const char* filename, uint64_t* suid, int* colId,
 static void tfileGenFileName(char* filename, uint64_t suid, int colId, int version);
 static void tfileSerialCacheKey(TFileCacheKey* key, char* buf);
 
+static TFileReader* tfileReaderCreateImpl(WriterCtx* ctx) {
+  TFileReader* reader = tfileReaderCreate(ctx);
+  tfileReaderRef(reader);
+  // tfileSerialCacheKey(&key, buf);
+  return reader;
+}
+
 TFileCache* tfileCacheCreate(const char* path) {
   TFileCache* tcache = calloc(1, sizeof(TFileCache));
-  if (tcache == NULL) {
-    return NULL;
-  }
+  if (tcache == NULL) { return NULL; }
 
   tcache->tableCache = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   tcache->capacity = 64;
@@ -75,22 +81,19 @@ TFileCache* tfileCacheCreate(const char* path) {
       continue;
     }
 
-    WriterCtx* wc = writerCtxCreate(TFile, file, true, 1024 * 64);
+    WriterCtx* wc = writerCtxCreate(TFile, file, true, 1024 * 1024 * 64);
     if (wc == NULL) {
       indexError("failed to open index:%s", file);
       goto End;
     }
 
-    TFileReader* reader = tfileReaderCreate(wc);
-    tfileReaderRef(reader);
-    // loader fst and validate it
+    char          buf[128] = {0};
+    TFileReader*  reader = tfileReaderCreateImpl(wc);
     TFileHeader*  header = &reader->header;
     TFileCacheKey key = {.suid = header->suid,
                          .colName = header->colName,
                          .nColName = strlen(header->colName),
                          .colType = header->colType};
-
-    char buf[128] = {0};
     tfileSerialCacheKey(&key, buf);
     taosHashPut(tcache->tableCache, buf, strlen(buf), &reader, sizeof(void*));
   }
@@ -102,9 +105,7 @@ End:
   return NULL;
 }
 void tfileCacheDestroy(TFileCache* tcache) {
-  if (tcache == NULL) {
-    return;
-  }
+  if (tcache == NULL) { return; }
 
   // free table cache
   TFileReader** reader = taosHashIterate(tcache->tableCache, NULL);
@@ -125,9 +126,7 @@ TFileReader* tfileCacheGet(TFileCache* tcache, TFileCacheKey* key) {
   tfileSerialCacheKey(key, buf);
 
   TFileReader** reader = taosHashGet(tcache->tableCache, buf, strlen(buf));
-  if (reader == NULL) {
-    return NULL;
-  }
+  if (reader == NULL) { return NULL; }
   tfileReaderRef(*reader);
 
   return *reader;
@@ -137,7 +136,7 @@ void tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader* reader)
   tfileSerialCacheKey(key, buf);
   // remove last version index reader
   TFileReader** p = taosHashGet(tcache->tableCache, buf, strlen(buf));
-  if (*p != NULL) {
+  if (p != NULL) {
     TFileReader* oldReader = *p;
     taosHashRemove(tcache->tableCache, buf, strlen(buf));
     tfileReaderUnRef(oldReader);
@@ -149,9 +148,7 @@ void tfileCachePut(TFileCache* tcache, TFileCacheKey* key, TFileReader* reader)
 }
 TFileReader* tfileReaderCreate(WriterCtx* ctx) {
   TFileReader* reader = calloc(1, sizeof(TFileReader));
-  if (reader == NULL) {
-    return NULL;
-  }
+  if (reader == NULL) { return NULL; }
 
   // T_REF_INC(reader);
   reader->ctx = ctx;
@@ -172,12 +169,10 @@ TFileReader* tfileReaderCreate(WriterCtx* ctx) {
   return reader;
 }
 void tfileReaderDestroy(TFileReader* reader) {
-  if (reader == NULL) {
-    return;
-  }
+  if (reader == NULL) { return; }
   // T_REF_INC(reader);
   fstDestroy(reader->fst);
-  writerCtxDestroy(reader->ctx);
+  writerCtxDestroy(reader->ctx, true);
   free(reader);
 }
 
@@ -216,7 +211,8 @@ TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int32_t version, const c
 
   char fullname[256] = {0};
   snprintf(fullname, sizeof(fullname), "%s/%s", path, filename);
-  WriterCtx* wcx = writerCtxCreate(TFile, fullname, true, 1024 * 1024);
+  WriterCtx* wcx = writerCtxCreate(TFile, fullname, false, 1024 * 1024 * 64);
+  if (wcx == NULL) { return NULL; }
 
   TFileHeader tfh = {0};
   tfh.suid = suid;
@@ -226,6 +222,21 @@ TFileWriter* tfileWriterOpen(char* path, uint64_t suid, int32_t version, const c
 
   return tfileWriterCreate(wcx, &tfh);
 }
+TFileReader* tfileReaderOpen(char* path, uint64_t suid, int32_t version, const char* colName) {
+  char    filename[128] = {0};
+  int32_t coldId = 1;
+  tfileGenFileName(filename, suid, coldId, version);
+
+  char fullname[256] = {0};
+  snprintf(fullname, sizeof(fullname), "%s/%s", path, filename);
+  WriterCtx* wc = writerCtxCreate(TFile, fullname, true, 1024 * 1024 * 1024);
+  if (wc == NULL) { return NULL; }
+
+  TFileReader* reader = tfileReaderCreateImpl(wc);
+  return reader;
+
+  // tfileSerialCacheKey(&key, buf);
+}
 TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header) {
   // char pathBuf[128] = {0};
   // sprintf(pathBuf, "%s/% " PRIu64 "-%d-%d.tindex", path, suid, colId, version);
@@ -249,28 +260,29 @@ TFileWriter* tfileWriterCreate(WriterCtx* ctx, TFileHeader* header) {
   return tw;
 }
 
-int tfileWriterPut(TFileWriter* tw, void* data) {
+int tfileWriterPut(TFileWriter* tw, void* data, bool order) {
   // sort by coltype and write to tindex
-  __compar_fn_t fn;
-
-  int8_t colType = tw->header.colType;
-  if (colType == TSDB_DATA_TYPE_BINARY || colType == TSDB_DATA_TYPE_NCHAR) {
-    fn = tfileStrCompare;
-  } else {
-    fn = getComparFunc(colType, 0);
+  if (order == false) {
+    __compar_fn_t fn;
+    int8_t        colType = tw->header.colType;
+    if (colType == TSDB_DATA_TYPE_BINARY || colType == TSDB_DATA_TYPE_NCHAR) {
+      fn = tfileStrCompare;
+    } else {
+      fn = getComparFunc(colType, 0);
+    }
+    taosArraySortPWithExt((SArray*)(data), tfileValueCompare, &fn);
   }
-  taosArraySortPWithExt((SArray*)(data), tfileValueCompare, &fn);
 
-  int32_t bufLimit = 4096, offset = 0;
-  char*   buf = calloc(1, sizeof(char) * bufLimit);
-  char*   p = buf;
+  int32_t bufLimit = 64 * 4096, offset = 0;
+  // char*   buf = calloc(1, sizeof(char) * bufLimit);
+  // char*   p = buf;
   int32_t sz = taosArrayGetSize((SArray*)data);
   int32_t fstOffset = tw->offset;
 
   // ugly code, refactor later
   for (size_t i = 0; i < sz; i++) {
     TFileValue* v = taosArrayGetP((SArray*)data, i);
-
+    // taosArrayRemoveDuplicate(v->tablId, tfileUidCompare, NULL);
     int32_t tbsz = taosArrayGetSize(v->tableId);
     fstOffset += TF_TABLE_TATOAL_SIZE(tbsz);
   }
@@ -282,30 +294,39 @@ int tfileWriterPut(TFileWriter* tw, void* data) {
     int32_t tbsz = taosArrayGetSize(v->tableId);
     // check buf has enough space or not
     int32_t ttsz = TF_TABLE_TATOAL_SIZE(tbsz);
-    if (offset + ttsz > bufLimit) {
-      // batch write
-      tw->ctx->write(tw->ctx, buf, offset);
-      offset = 0;
-      memset(buf, 0, bufLimit);
-      p = buf;
-    }
 
+    // if (offset + ttsz >= bufLimit) {
+    //  // batch write
+    //  indexInfo("offset: %d, ttsz: %d", offset, ttsz);
+    //  // std::cout << "offset: " << offset << std::endl;
+    //  // std::cout << "ttsz:" << ttsz < < < std::endl;
+    //  tw->ctx->write(tw->ctx, buf, offset);
+    //  offset = 0;
+    //  memset(buf, 0, bufLimit);
+    //  p = buf;
+    //}
+    // if (ttsz >= bufLimit) {
+    //}
+    char* buf = calloc(1, ttsz * sizeof(char));
+    char* p = buf;
     tfileSerialTableIdsToBuf(p, v->tableId);
-    offset += ttsz;
-    p = buf + offset;
+    tw->ctx->write(tw->ctx, buf, ttsz);
+    // offset += ttsz;
+    // p = buf + offset;
     // set up value offset
     v->offset = tw->offset;
     tw->offset += ttsz;
+    free(buf);
   }
-  if (offset != 0) {
-    // write reversed data in buf to tindex
-    tw->ctx->write(tw->ctx, buf, offset);
-  }
-  tfree(buf);
+  // if (offset != 0) {
+  // write reversed data in buf to tindex
+  // tw->ctx->write(tw->ctx, buf, offset);
+  //}
+  // tfree(buf);
 
   tw->fb = fstBuilderCreate(tw->ctx, 0);
   if (tw->fb == NULL) {
-    tfileWriterDestroy(tw);
+    tfileWriterClose(tw);
     return -1;
   }
   // write fst
@@ -321,27 +342,20 @@ int tfileWriterPut(TFileWriter* tw, void* data) {
   tw->fb = NULL;
   return 0;
 }
-void tfileWriteClose(TFileWriter* tw) {
-  if (tw == NULL) {
-    return;
-  }
-  writerCtxDestroy(tw->ctx);
+void tfileWriterClose(TFileWriter* tw) {
+  if (tw == NULL) { return; }
+  writerCtxDestroy(tw->ctx, false);
   free(tw);
 }
 void tfileWriterDestroy(TFileWriter* tw) {
-  if (tw == NULL) {
-    return;
-  }
-
-  writerCtxDestroy(tw->ctx);
+  if (tw == NULL) { return; }
+  writerCtxDestroy(tw->ctx, false);
   free(tw);
 }
 
 IndexTFile* indexTFileCreate(const char* path) {
   IndexTFile* tfile = calloc(1, sizeof(IndexTFile));
-  if (tfile == NULL) {
-    return NULL;
-  }
+  if (tfile == NULL) { return NULL; }
 
   tfile->cache = tfileCacheCreate(path);
   return tfile;
@@ -350,18 +364,14 @@ void IndexTFileDestroy(IndexTFile* tfile) { free(tfile); }
 
 int indexTFileSearch(void* tfile, SIndexTermQuery* query, SArray* result) {
   int ret = -1;
-  if (tfile == NULL) {
-    return ret;
-  }
+  if (tfile == NULL) { return ret; }
   IndexTFile* pTfile = (IndexTFile*)tfile;
 
   SIndexTerm*   term = query->term;
   TFileCacheKey key = {
       .suid = term->suid, .colType = term->colType, .colName = term->colName, .nColName = term->nColName};
   TFileReader* reader = tfileCacheGet(pTfile->cache, &key);
-  if (reader == NULL) {
-    return 0;
-  }
+  if (reader == NULL) { return 0; }
 
   return tfileReaderSearch(reader, query, result);
 }
@@ -381,9 +391,7 @@ static bool tfileIteratorNext(Iterate* iiter) {
 
   TFileFstIter*          tIter = iiter->iter;
   StreamWithStateResult* rt = streamWithStateNextWith(tIter->st, NULL);
-  if (rt == NULL) {
-    return false;
-  }
+  if (rt == NULL) { return false; }
 
   int32_t sz = 0;
   char*   ch = (char*)fstSliceData(&rt->data, &sz);
@@ -391,12 +399,9 @@ static bool tfileIteratorNext(Iterate* iiter) {
   memcpy(colVal, ch, sz);
 
   offset = (uint64_t)(rt->out.out);
-
   swsResultDestroy(rt);
   // set up iterate value
-  if (tfileReaderLoadTableIds(tIter->rdr, offset, iv->val) != 0) {
-    return false;
-  }
+  if (tfileReaderLoadTableIds(tIter->rdr, offset, iv->val) != 0) { return false; }
 
   iv->colVal = colVal;
 
@@ -407,9 +412,8 @@ static IterateValue* tifileIterateGetValue(Iterate* iter) { return &iter->val; }
 
 static TFileFstIter* tfileFstIteratorCreate(TFileReader* reader) {
   TFileFstIter* tIter = calloc(1, sizeof(Iterate));
-  if (tIter == NULL) {
-    return NULL;
-  }
+  if (tIter == NULL) { return NULL; }
+
   tIter->ctx = automCtxCreate(NULL, AUTOMATION_ALWAYS);
   tIter->fb = fstSearch(reader->fst, tIter->ctx);
   tIter->st = streamBuilderIntoStream(tIter->fb);
@@ -418,21 +422,22 @@ static TFileFstIter* tfileFstIteratorCreate(TFileReader* reader) {
 }
 
 Iterate* tfileIteratorCreate(TFileReader* reader) {
-  Iterate* iter = calloc(1, sizeof(Iterate));
+  if (reader == NULL) { return NULL; }
 
+  Iterate* iter = calloc(1, sizeof(Iterate));
   iter->iter = tfileFstIteratorCreate(reader);
   if (iter->iter == NULL) {
+    free(iter);
     return NULL;
   }
-
   iter->next = tfileIteratorNext;
   iter->getValue = tifileIterateGetValue;
+  iter->val.val = taosArrayInit(1, sizeof(uint64_t));
   return iter;
 }
 void tfileIteratorDestroy(Iterate* iter) {
-  if (iter == NULL) {
-    return;
-  }
+  if (iter == NULL) { return; }
+
   IterateValue* iv = &iter->val;
   iterateValueDestroy(iv, true);
 
@@ -445,18 +450,19 @@ void tfileIteratorDestroy(Iterate* iter) {
 }
 
 TFileReader* tfileGetReaderByCol(IndexTFile* tf, char* colName) {
-  if (tf == NULL) {
-    return NULL;
-  }
+  if (tf == NULL) { return NULL; }
   TFileCacheKey key = {.suid = 0, .colType = TSDB_DATA_TYPE_BINARY, .colName = colName, .nColName = strlen(colName)};
   return tfileCacheGet(tf->cache, &key);
 }
 
+static int tfileUidCompare(const void* a, const void* b) {
+  uint64_t l = *(uint64_t*)a;
+  uint64_t r = *(uint64_t*)b;
+  return l - r;
+}
 static int tfileStrCompare(const void* a, const void* b) {
   int ret = strcmp((char*)a, (char*)b);
-  if (ret == 0) {
-    return ret;
-  }
+  if (ret == 0) { return ret; }
   return ret < 0 ? -1 : 1;
 }
 
@@ -471,17 +477,13 @@ static int tfileValueCompare(const void* a, const void* b, const void* param) {
 
 TFileValue* tfileValueCreate(char* val) {
   TFileValue* tf = calloc(1, sizeof(TFileValue));
-  if (tf == NULL) {
-    return NULL;
-  }
-
+  if (tf == NULL) { return NULL; }
+  tf->colVal = val;
   tf->tableId = taosArrayInit(32, sizeof(uint64_t));
   return tf;
 }
 int tfileValuePush(TFileValue* tf, uint64_t val) {
-  if (tf == NULL) {
-    return -1;
-  }
+  if (tf == NULL) { return -1; }
   taosArrayPush(tf->tableId, &val);
   return 0;
 }
@@ -501,9 +503,7 @@ static void tfileSerialTableIdsToBuf(char* buf, SArray* ids) {
 static int tfileWriteFstOffset(TFileWriter* tw, int32_t offset) {
   int32_t fstOffset = offset + sizeof(tw->header.fstOffset);
   tw->header.fstOffset = fstOffset;
-  if (sizeof(fstOffset) != tw->ctx->write(tw->ctx, (char*)&fstOffset, sizeof(fstOffset))) {
-    return -1;
-  }
+  if (sizeof(fstOffset) != tw->ctx->write(tw->ctx, (char*)&fstOffset, sizeof(fstOffset))) { return -1; }
   tw->offset += sizeof(fstOffset);
   return 0;
 }
@@ -514,9 +514,7 @@ static int tfileWriteHeader(TFileWriter* writer) {
   memcpy(buf, (char*)header, sizeof(buf));
 
   int nwrite = writer->ctx->write(writer->ctx, buf, sizeof(buf));
-  if (sizeof(buf) != nwrite) {
-    return -1;
-  }
+  if (sizeof(buf) != nwrite) { return -1; }
   writer->offset = nwrite;
   return 0;
 }
@@ -547,12 +545,10 @@ static int tfileReaderLoadHeader(TFileReader* reader) {
 }
 static int tfileReaderLoadFst(TFileReader* reader) {
   // current load fst into memory, refactor it later
-  static int FST_MAX_SIZE = 16 * 1024;
+  static int FST_MAX_SIZE = 64 * 1024;
 
   char* buf = calloc(1, sizeof(char) * FST_MAX_SIZE);
-  if (buf == NULL) {
-    return -1;
-  }
+  if (buf == NULL) { return -1; }
 
   WriterCtx* ctx = reader->ctx;
   int32_t    nread = ctx->readFrom(ctx, buf, FST_MAX_SIZE, reader->header.fstOffset);
@@ -575,36 +571,33 @@ static int tfileReaderLoadTableIds(TFileReader* reader, int32_t offset, SArray*
 
   int32_t total = sizeof(uint64_t) * nid;
   char*   buf = calloc(1, total);
-  if (buf == NULL) {
-    return -1;
-  }
+  if (buf == NULL) { return -1; }
 
   nread = ctx->read(ctx, buf, total);
   assert(total == nread);
 
-  for (int32_t i = 0; i < nid; i++) {
-    taosArrayPush(result, (uint64_t*)buf + i);
-  }
+  for (int32_t i = 0; i < nid; i++) { taosArrayPush(result, (uint64_t*)buf + i); }
   free(buf);
   return 0;
 }
 void tfileReaderRef(TFileReader* reader) {
+  if (reader == NULL) { return; }
   int ref = T_REF_INC(reader);
   UNUSED(ref);
 }
 
 void tfileReaderUnRef(TFileReader* reader) {
+  if (reader == NULL) { return; }
   int ref = T_REF_DEC(reader);
   if (ref == 0) {
+    // do nothing
     tfileReaderDestroy(reader);
   }
 }
 
 static int tfileGetFileList(const char* path, SArray* result) {
   DIR* dir = opendir(path);
-  if (NULL == dir) {
-    return -1;
-  }
+  if (NULL == dir) { return -1; }
 
   struct dirent* entry;
   while ((entry = readdir(dir)) != NULL) {
@@ -632,9 +625,7 @@ static int tfileCompare(const void* a, const void* b) {
   size_t bLen = strlen(bName);
 
   int ret = strncmp(aName, bName, aLen > bLen ? aLen : bLen);
-  if (ret == 0) {
-    return ret;
-  }
+  if (ret == 0) { return ret; }
   return ret < 0 ? -1 : 1;
 }
 // tfile name suid-colId-version.tindex

source/libs/index/test/fstTest.cc

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

source/libs/index/test/indexTests.cc

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

source/util/src/terror.c

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