Merge branch '3.0' of github.com:taosdata/TDengine into test/chr/TD-14699

include/common/tmsg.h

@@ -438,6 +438,7 @@ typedef struct {
 
 int32_t tSerializeSGetUserAuthRsp(void* buf, int32_t bufLen, SGetUserAuthRsp* pRsp);
 int32_t tDeserializeSGetUserAuthRsp(void* buf, int32_t bufLen, SGetUserAuthRsp* pRsp);
+void    tFreeSGetUserAuthRsp(SGetUserAuthRsp* pRsp);
 
 typedef struct {
   int16_t colId;     // column id
@@ -1332,7 +1333,7 @@ int32_t tDeserializeSCMCreateTopicRsp(void* buf, int32_t bufLen, SCMCreateTopicR
 
 typedef struct {
   int64_t consumerId;
-} SMqConsumerLostMsg;
+} SMqConsumerLostMsg, SMqConsumerRecoverMsg;
 
 typedef struct {
   int64_t consumerId;

include/common/tmsgdef.h

@@ -145,10 +145,11 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_TOPIC, "mnode-alter-topic", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_DROP_TOPIC, "mnode-drop-topic", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_SUBSCRIBE, "mnode-subscribe", SCMSubscribeReq, SCMSubscribeRsp)
-  TD_DEF_MSG_TYPE(TDMT_MND_MQ_ASK_EP, "mnode-mq-ask-ep", SMqAskEpReq, SMqAskEpReq)
+  TD_DEF_MSG_TYPE(TDMT_MND_MQ_ASK_EP, "mnode-mq-ask-ep", SMqAskEpReq, SMqAskEpRsp)
-  TD_DEF_MSG_TYPE(TDMT_MND_MQ_TIMER, "mnode-mq-tmr", SMTimerReq, SMTimerReq)
+  TD_DEF_MSG_TYPE(TDMT_MND_MQ_TIMER, "mnode-mq-tmr", SMTimerReq, NULL)
-  TD_DEF_MSG_TYPE(TDMT_MND_MQ_CONSUMER_LOST, "mnode-mq-consumer-lost", SMTimerReq, SMTimerReq)
+  TD_DEF_MSG_TYPE(TDMT_MND_MQ_CONSUMER_LOST, "mnode-mq-consumer-lost", SMqConsumerLostMsg, NULL)
-  TD_DEF_MSG_TYPE(TDMT_MND_MQ_DO_REBALANCE, "mnode-mq-do-rebalance", SMqDoRebalanceMsg, SMqDoRebalanceMsg)
+  TD_DEF_MSG_TYPE(TDMT_MND_MQ_CONSUMER_RECOVER, "mnode-mq-consumer-recover", SMqConsumerRecoverMsg, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_MQ_DO_REBALANCE, "mnode-mq-do-rebalance", SMqDoRebalanceMsg, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_MQ_COMMIT_OFFSET, "mnode-mq-commit-offset", SMqCMCommitOffsetReq, SMqCMCommitOffsetRsp)
   TD_DEF_MSG_TYPE(TDMT_MND_CREATE_STREAM, "mnode-create-stream", SCMCreateStreamReq, SCMCreateStreamRsp)
   TD_DEF_MSG_TYPE(TDMT_MND_ALTER_STREAM, "mnode-alter-stream", NULL, NULL)

include/libs/executor/executor.h

@@ -32,6 +32,7 @@ typedef struct SReadHandle {
   void* reader;
   void* meta;
   void* config;
+  void* vnode;
 } SReadHandle;
 
 #define STREAM_DATA_TYPE_SUBMIT_BLOCK 0x1

include/os/osSocket.h

@@ -19,53 +19,53 @@
 // If the error is in a third-party library, place this header file under the third-party library header file.
 // When you want to use this feature, you should find or add the same function in the following section.
 #ifndef ALLOW_FORBID_FUNC
-  #define socket SOCKET_FUNC_TAOS_FORBID
+#define socket       SOCKET_FUNC_TAOS_FORBID
-  #define bind   BIND_FUNC_TAOS_FORBID
+#define bind         BIND_FUNC_TAOS_FORBID
-  #define listen LISTEN_FUNC_TAOS_FORBID
+#define listen       LISTEN_FUNC_TAOS_FORBID
-  #define accept ACCEPT_FUNC_TAOS_FORBID
+#define accept       ACCEPT_FUNC_TAOS_FORBID
-  #define epoll_create EPOLL_CREATE_FUNC_TAOS_FORBID
+#define epoll_create EPOLL_CREATE_FUNC_TAOS_FORBID
-  #define epoll_ctl EPOLL_CTL_FUNC_TAOS_FORBID
+#define epoll_ctl    EPOLL_CTL_FUNC_TAOS_FORBID
-  #define epoll_wait EPOLL_WAIT_FUNC_TAOS_FORBID
+#define epoll_wait   EPOLL_WAIT_FUNC_TAOS_FORBID
-  #define inet_addr INET_ADDR_FUNC_TAOS_FORBID
+#define inet_addr    INET_ADDR_FUNC_TAOS_FORBID
-  #define inet_ntoa INET_NTOA_FUNC_TAOS_FORBID
+#define inet_ntoa    INET_NTOA_FUNC_TAOS_FORBID
 #endif
 
 #if defined(WINDOWS)
-  #if BYTE_ORDER == LITTLE_ENDIAN
+#if BYTE_ORDER == LITTLE_ENDIAN
-    #include <stdlib.h>
+#include <stdlib.h>
-    #define htobe16(x) _byteswap_ushort(x)
+#define htobe16(x) _byteswap_ushort(x)
-    #define htole16(x) (x)
+#define htole16(x) (x)
-    #define be16toh(x) _byteswap_ushort(x)
+#define be16toh(x) _byteswap_ushort(x)
-    #define le16toh(x) (x)
+#define le16toh(x) (x)
 
-    #define htobe32(x) _byteswap_ulong(x)
+#define htobe32(x) _byteswap_ulong(x)
-    #define htole32(x) (x)
+#define htole32(x) (x)
-    #define be32toh(x) _byteswap_ulong(x)
+#define be32toh(x) _byteswap_ulong(x)
-    #define le32toh(x) (x)
+#define le32toh(x) (x)
 
-    #define htobe64(x) _byteswap_uint64(x)
+#define htobe64(x) _byteswap_uint64(x)
-    #define htole64(x) (x)
+#define htole64(x) (x)
-    #define be64toh(x) _byteswap_uint64(x)
+#define be64toh(x) _byteswap_uint64(x)
-    #define le64toh(x) (x)
+#define le64toh(x) (x)
-  #else
+#else
-    #error byte order not supported
+#error byte order not supported
-  #endif
+#endif
 
-  #define __BYTE_ORDER    BYTE_ORDER
+#define __BYTE_ORDER    BYTE_ORDER
-  #define __BIG_ENDIAN    BIG_ENDIAN
+#define __BIG_ENDIAN    BIG_ENDIAN
-  #define __LITTLE_ENDIAN LITTLE_ENDIAN
+#define __LITTLE_ENDIAN LITTLE_ENDIAN
-  #define __PDP_ENDIAN    PDP_ENDIAN
+#define __PDP_ENDIAN    PDP_ENDIAN
 
 #else
-  #include <netinet/in.h>
+#include <netinet/in.h>
-  #include <sys/socket.h>
+#include <sys/socket.h>
 
-  #if defined(_TD_DARWIN_64)
+#if defined(_TD_DARWIN_64)
-    #include <osEok.h>
+#include <osEok.h>
-  #else
+#else
-    #include <netinet/in.h>
+#include <netinet/in.h>
-    #include <sys/epoll.h>
+#include <sys/epoll.h>
-  #endif
+#endif
 #endif
 
 #ifdef __cplusplus
@@ -73,24 +73,24 @@ extern "C" {
 #endif
 
 #if defined(WINDOWS)
-  typedef int socklen_t;
+typedef int socklen_t;
-  #define TAOS_EPOLL_WAIT_TIME 100
+#define TAOS_EPOLL_WAIT_TIME 100
-  typedef SOCKET eventfd_t;
+typedef SOCKET eventfd_t;
-  #define eventfd(a, b) -1
+#define eventfd(a, b)      -1
-  #define EpollClose(pollFd) epoll_close(pollFd)
+#define EpollClose(pollFd) epoll_close(pollFd)
-  #ifndef EPOLLWAKEUP
+#ifndef EPOLLWAKEUP
-    #define EPOLLWAKEUP (1u << 29)
+#define EPOLLWAKEUP (1u << 29)
-  #endif
+#endif
 #elif defined(_TD_DARWIN_64)
-  #define TAOS_EPOLL_WAIT_TIME 500
+#define TAOS_EPOLL_WAIT_TIME 500
-  typedef int32_t SOCKET;
+typedef int32_t SOCKET;
-  typedef SOCKET EpollFd;
+typedef SOCKET  EpollFd;
-  #define EpollClose(pollFd) epoll_close(pollFd)
+#define EpollClose(pollFd)   epoll_close(pollFd)
 #else
-  #define TAOS_EPOLL_WAIT_TIME 500
+#define TAOS_EPOLL_WAIT_TIME 500
-  typedef int32_t SOCKET;
+typedef int32_t SOCKET;
-  typedef SOCKET EpollFd;
+typedef SOCKET  EpollFd;
-  #define EpollClose(pollFd) taosCloseSocket(pollFd)
+#define EpollClose(pollFd)   taosCloseSocket(pollFd)
 #endif
 
 #if defined(_TD_DARWIN_64)
@@ -128,16 +128,17 @@ typedef struct TdSocket {
 #endif
   int      refId;
   SocketFd fd;
-} *TdSocketPtr, TdSocket;
+} * TdSocketPtr, TdSocket;
 
 typedef struct TdSocketServer *TdSocketServerPtr;
-typedef struct TdSocket *TdSocketPtr;
+typedef struct TdSocket *      TdSocketPtr;
-typedef struct TdEpoll *TdEpollPtr;
+typedef struct TdEpoll *       TdEpollPtr;
 
-int32_t taosSendto(TdSocketPtr pSocket, void * msg, int len, unsigned int flags, const struct sockaddr * to, int tolen);
+int32_t taosSendto(TdSocketPtr pSocket, void *msg, int len, unsigned int flags, const struct sockaddr *to, int tolen);
 int32_t taosWriteSocket(TdSocketPtr pSocket, void *msg, int len);
 int32_t taosReadSocket(TdSocketPtr pSocket, void *msg, int len);
-int32_t taosReadFromSocket(TdSocketPtr pSocket, void *buf, int32_t len, int32_t flags, struct sockaddr *destAddr, int *addrLen);
+int32_t taosReadFromSocket(TdSocketPtr pSocket, void *buf, int32_t len, int32_t flags, struct sockaddr *destAddr,
+                           int *addrLen);
 int32_t taosCloseSocketNoCheck1(SocketFd fd);
 int32_t taosCloseSocket(TdSocketPtr *ppSocket);
 int32_t taosCloseSocketServer(TdSocketServerPtr *ppSocketServer);
@@ -156,13 +157,15 @@ int32_t taosNonblockwrite(TdSocketPtr pSocket, char *ptr, int32_t nbytes);
 int64_t taosCopyFds(TdSocketPtr pSrcSocket, TdSocketPtr pDestSocket, int64_t len);
 void    taosWinSocketInit();
 
+int taosCreateSocketWithTimeOutOpt(uint32_t conn_timeout_sec);
+
 TdSocketPtr       taosOpenUdpSocket(uint32_t localIp, uint16_t localPort);
 TdSocketPtr       taosOpenTcpClientSocket(uint32_t ip, uint16_t port, uint32_t localIp);
 TdSocketServerPtr taosOpenTcpServerSocket(uint32_t ip, uint16_t port);
 int32_t           taosKeepTcpAlive(TdSocketPtr pSocket);
 TdSocketPtr       taosAcceptTcpConnectSocket(TdSocketServerPtr pServerSocket, struct sockaddr *destAddr, int *addrLen);
 
-int32_t taosGetSocketName(TdSocketPtr pSocket,struct sockaddr *destAddr, int *addrLen);
+int32_t taosGetSocketName(TdSocketPtr pSocket, struct sockaddr *destAddr, int *addrLen);
 
 void        taosBlockSIGPIPE();
 uint32_t    taosGetIpv4FromFqdn(const char *);

include/util/taoserror.h

@@ -268,14 +268,13 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_MND_TOPIC_NOT_EXIST           TAOS_DEF_ERROR_CODE(0, 0x03E1)
 #define TSDB_CODE_MND_TOO_MANY_TOPICS           TAOS_DEF_ERROR_CODE(0, 0x03E2)
 #define TSDB_CODE_MND_INVALID_TOPIC             TAOS_DEF_ERROR_CODE(0, 0x03E3)
-#define TSDB_CODE_MND_INVALID_TOPIC_OPTION      TAOS_DEF_ERROR_CODE(0, 0x03E4)
+#define TSDB_CODE_MND_INVALID_TOPIC_QUERY       TAOS_DEF_ERROR_CODE(0, 0x03E4)
-#define TSDB_CODE_MND_TOPIC_OPTION_UNCHNAGED    TAOS_DEF_ERROR_CODE(0, 0x03E5)
+#define TSDB_CODE_MND_INVALID_TOPIC_OPTION      TAOS_DEF_ERROR_CODE(0, 0x03E5)
-#define TSDB_CODE_MND_NAME_CONFLICT_WITH_STB    TAOS_DEF_ERROR_CODE(0, 0x03E6)
+#define TSDB_CODE_MND_CONSUMER_NOT_EXIST        TAOS_DEF_ERROR_CODE(0, 0x03E6)
-#define TSDB_CODE_MND_CONSUMER_NOT_EXIST        TAOS_DEF_ERROR_CODE(0, 0x03E7)
+#define TSDB_CODE_MND_TOPIC_OPTION_UNCHNAGED    TAOS_DEF_ERROR_CODE(0, 0x03E7)
-#define TSDB_CODE_MND_UNSUPPORTED_TOPIC         TAOS_DEF_ERROR_CODE(0, 0x03E8)
+#define TSDB_CODE_MND_SUBSCRIBE_NOT_EXIST       TAOS_DEF_ERROR_CODE(0, 0x03E8)
-#define TSDB_CODE_MND_SUBSCRIBE_NOT_EXIST       TAOS_DEF_ERROR_CODE(0, 0x03E9)
+#define TSDB_CODE_MND_OFFSET_NOT_EXIST          TAOS_DEF_ERROR_CODE(0, 0x03E9)
-#define TSDB_CODE_MND_OFFSET_NOT_EXIST          TAOS_DEF_ERROR_CODE(0, 0x03EA)
+#define TSDB_CODE_MND_CONSUMER_NOT_READY        TAOS_DEF_ERROR_CODE(0, 0x03EA)
-#define TSDB_CODE_MND_CONSUMER_NOT_READY        TAOS_DEF_ERROR_CODE(0, 0x03EB)
 
 // mnode-stream
 #define TSDB_CODE_MND_STREAM_ALREADY_EXIST      TAOS_DEF_ERROR_CODE(0, 0x03F0)

include/util/tarray.h

@@ -205,7 +205,6 @@ SArray* taosArrayDup(const SArray* pSrc);
  */
 SArray* taosArrayDeepCopy(const SArray* pSrc, FCopy deepCopy);
 
-
 /**
  * clear the array (remove all element)
  * @param pArray
@@ -272,6 +271,8 @@ void taosArraySortPWithExt(SArray* pArray, __ext_compar_fn_t fn, const void* par
 int32_t taosEncodeArray(void** buf, const SArray* pArray, FEncode encode);
 void*   taosDecodeArray(const void* buf, SArray** pArray, FDecode decode, int32_t dataSz);
 
+char* taosShowStrArray(const SArray* pArray);
+
 #ifdef __cplusplus
 }
 #endif

include/util/tdef.h

@@ -131,6 +131,7 @@ extern const int32_t TYPE_BYTES[15];
 #define TSDB_PERFS_TABLE_TOPICS        "topics"
 #define TSDB_PERFS_TABLE_CONSUMERS     "consumers"
 #define TSDB_PERFS_TABLE_SUBSCRIPTIONS "subscriptions"
+#define TSDB_PERFS_TABLE_OFFSETS       "offsets"
 
 #define TSDB_INDEX_TYPE_SMA      "SMA"
 #define TSDB_INDEX_TYPE_FULLTEXT "FULLTEXT"

source/client/src/clientImpl.c

@@ -245,7 +245,6 @@ void setResSchemaInfo(SReqResultInfo* pResInfo, const SSchema* pSchema, int32_t
   ASSERT(pSchema != NULL && numOfCols > 0);
 
   pResInfo->numOfCols = numOfCols;
-  // TODO handle memory leak
   if (pResInfo->fields != NULL) {
     taosMemoryFree(pResInfo->fields);
   }

source/client/src/tmq.c

@@ -666,7 +666,6 @@ tmq_resp_err_t tmq_subscribe(tmq_t* tmq, const tmq_list_t* topic_list) {
   code = param.rspErr;
   if (code != 0) goto FAIL;
 
-  // TODO: add max retry cnt
   while (TSDB_CODE_MND_CONSUMER_NOT_READY == tmqAskEp(tmq, false)) {
     tscDebug("not ready, retry");
     taosMsleep(500);
@@ -683,7 +682,7 @@ tmq_resp_err_t tmq_subscribe(tmq_t* tmq, const tmq_list_t* topic_list) {
   code = 0;
 FAIL:
   if (req.topicNames != NULL) taosArrayDestroyP(req.topicNames, taosMemoryFree);
-  if (code != 0) {
+  if (code != 0 && buf) {
     taosMemoryFree(buf);
   }
   return code;
@@ -1265,6 +1264,7 @@ TAOS_RES* tmq_consumer_poll(tmq_t* tmq, int64_t wait_time) {
     return (TAOS_RES*)rspObj;
   }
 
+  // in no topic status also need process delayed task
   if (atomic_load_8(&tmq->status) == TMQ_CONSUMER_STATUS__INIT) {
     return NULL;
   }
@@ -1285,6 +1285,9 @@ TAOS_RES* tmq_consumer_poll(tmq_t* tmq, int64_t wait_time) {
         return NULL;
       }
       tsem_timewait(&tmq->rspSem, leftTime * 1000);
+    } else {
+      // use tsem_timewait instead of tsem_wait to avoid unexpected stuck
+      tsem_timewait(&tmq->rspSem, 500 * 1000);
     }
   }
 }

source/common/src/tmsg.c

@@ -1302,6 +1302,11 @@ int32_t tDeserializeSGetUserAuthRsp(void *buf, int32_t bufLen, SGetUserAuthRsp *
   return 0;
 }
 
+void tFreeSGetUserAuthRsp(SGetUserAuthRsp *pRsp) {
+  taosHashCleanup(pRsp->readDbs);
+  taosHashCleanup(pRsp->writeDbs);
+}
+
 int32_t tSerializeSCreateDropMQSBNodeReq(void *buf, int32_t bufLen, SMCreateQnodeReq *pReq) {
   SCoder encoder = {0};
   tCoderInit(&encoder, TD_LITTLE_ENDIAN, buf, bufLen, TD_ENCODER);

source/dnode/mgmt/implement/src/dmEps.c

@@ -120,7 +120,7 @@ int32_t dmReadEps(SDnode *pDnode) {
       goto PRASE_DNODE_OVER;
     }
 
-    dnodeEp.id = dnodeId->valueint;
+    dnodeEp.id = did->valueint;
 
     cJSON *dnodeFqdn = cJSON_GetObjectItem(node, "fqdn");
     if (!dnodeFqdn || dnodeFqdn->type != cJSON_String || dnodeFqdn->valuestring == NULL) {
@@ -156,11 +156,6 @@ PRASE_DNODE_OVER:
   if (root != NULL) cJSON_Delete(root);
   if (pFile != NULL) taosCloseFile(&pFile);
 
-  if (dmIsEpChanged(pDnode, pDnode->data.dnodeId, pDnode->data.localEp)) {
-    dError("localEp %s different with %s and need reconfigured", pDnode->data.localEp, file);
-    return -1;
-  }
-
   if (taosArrayGetSize(pDnode->data.dnodeEps) == 0) {
     SDnodeEp dnodeEp = {0};
     dnodeEp.isMnode = 1;
@@ -170,6 +165,11 @@ PRASE_DNODE_OVER:
 
   dmResetEps(pDnode, pDnode->data.dnodeEps);
 
+  if (dmIsEpChanged(pDnode, pDnode->data.dnodeId, pDnode->data.localEp)) {
+    dError("localEp %s different with %s and need reconfigured", pDnode->data.localEp, file);
+    return -1;
+  }
+
   terrno = code;
   return code;
 }
@@ -291,13 +291,17 @@ static void dmPrintEps(SDnode *pDnode) {
 
 static bool dmIsEpChanged(SDnode *pDnode, int32_t dnodeId, const char *ep) {
   bool changed = false;
+  if (dnodeId == 0) return changed;
   taosRLockLatch(&pDnode->data.latch);
 
   SDnodeEp *pDnodeEp = taosHashGet(pDnode->data.dnodeHash, &dnodeId, sizeof(int32_t));
   if (pDnodeEp != NULL) {
-    char epstr[TSDB_EP_LEN + 1];
+    char epstr[TSDB_EP_LEN + 1] = {0};
     snprintf(epstr, TSDB_EP_LEN, "%s:%u", pDnodeEp->ep.fqdn, pDnodeEp->ep.port);
-    changed = strcmp(ep, epstr) != 0;
+    changed = (strcmp(ep, epstr) != 0);
+    if (changed) {
+      dError("dnode:%d, localEp %s different from %s", dnodeId, ep, epstr);
+    }
   }
 
   taosRUnLockLatch(&pDnode->data.latch);

source/dnode/mgmt/mgmt_vnode/src/vmHandle.c

@@ -114,7 +114,6 @@ static void vmGenerateVnodeCfg(SCreateVnodeReq *pCreate, SVnodeCfg *pCfg) {
   pCfg->tsdbCfg.keep2 = 3650;  // pCreate->daysToKeep0;
   pCfg->tsdbCfg.keep0 = 3650;  // pCreate->daysToKeep2;
   pCfg->tsdbCfg.keep1 = 3650;  // pCreate->daysToKeep0;
-  pCfg->tsdbCfg.lruCacheSize = pCreate->cacheBlockSize;
   pCfg->tsdbCfg.retentions = pCreate->pRetensions;
   pCfg->walCfg.vgId = pCreate->vgId;
   pCfg->hashBegin = pCreate->hashBegin;

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

@@ -23,10 +23,8 @@ extern "C" {
 #endif
 
 enum {
-  // MQ_CONSUMER_STATUS__INIT = 1,
   MQ_CONSUMER_STATUS__MODIFY = 1,
   MQ_CONSUMER_STATUS__MODIFY_IN_REB,
-  // MQ_CONSUMER_STATUS__IDLE,
   MQ_CONSUMER_STATUS__READY,
   MQ_CONSUMER_STATUS__LOST,
   MQ_CONSUMER_STATUS__LOST_IN_REB,

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

@@ -89,6 +89,7 @@ typedef enum {
   TRN_TYPE_DROP_STREAM = 1020,
   TRN_TYPE_ALTER_STREAM = 1021,
   TRN_TYPE_CONSUMER_LOST = 1022,
+  TRN_TYPE_CONSUMER_RECOVER = 1023,
   TRN_TYPE_BASIC_SCOPE_END,
   TRN_TYPE_GLOBAL_SCOPE = 2000,
   TRN_TYPE_CREATE_DNODE = 2001,
@@ -463,12 +464,14 @@ enum {
   CONSUMER_UPDATE__ADD,
   CONSUMER_UPDATE__REMOVE,
   CONSUMER_UPDATE__LOST,
+  CONSUMER_UPDATE__RECOVER,
   CONSUMER_UPDATE__MODIFY,
 };
 
 typedef struct {
   int64_t consumerId;
   char    cgroup[TSDB_CGROUP_LEN];
+  char    appId[TSDB_CGROUP_LEN];
   int8_t  updateType;  // used only for update
   int32_t epoch;
   int32_t status;
@@ -479,6 +482,17 @@ typedef struct {
   SArray*  currentTopics;     // SArray<char*>
   SArray*  rebNewTopics;      // SArray<char*>
   SArray*  rebRemovedTopics;  // SArray<char*>
+
+  // subscribed by user
+  SArray* assignedTopics;  // SArray<char*>
+
+  // data for display
+  int32_t pid;
+  SEpSet  ep;
+  int64_t upTime;
+  int64_t subscribeTime;
+  int64_t rebalanceTime;
+
 } SMqConsumerObj;
 
 SMqConsumerObj* tNewSMqConsumerObj(int64_t consumerId, char cgroup[TSDB_CGROUP_LEN]);

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

@@ -37,6 +37,8 @@
 
 static int8_t mqInRebFlag = 0;
 
+static const char *mndConsumerStatusName(int status);
+
 static int32_t mndConsumerActionInsert(SSdb *pSdb, SMqConsumerObj *pConsumer);
 static int32_t mndConsumerActionDelete(SSdb *pSdb, SMqConsumerObj *pConsumer);
 static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pConsumer, SMqConsumerObj *pNewConsumer);
@@ -48,6 +50,7 @@ static int32_t mndProcessSubscribeReq(SNodeMsg *pMsg);
 static int32_t mndProcessAskEpReq(SNodeMsg *pMsg);
 static int32_t mndProcessMqTimerMsg(SNodeMsg *pMsg);
 static int32_t mndProcessConsumerLostMsg(SNodeMsg *pMsg);
+static int32_t mndProcessConsumerRecoverMsg(SNodeMsg *pMsg);
 
 int32_t mndInitConsumer(SMnode *pMnode) {
   SSdbTable table = {.sdbType = SDB_CONSUMER,
@@ -62,6 +65,11 @@ int32_t mndInitConsumer(SMnode *pMnode) {
   mndSetMsgHandle(pMnode, TDMT_MND_MQ_ASK_EP, mndProcessAskEpReq);
   mndSetMsgHandle(pMnode, TDMT_MND_MQ_TIMER, mndProcessMqTimerMsg);
   mndSetMsgHandle(pMnode, TDMT_MND_MQ_CONSUMER_LOST, mndProcessConsumerLostMsg);
+  mndSetMsgHandle(pMnode, TDMT_MND_MQ_CONSUMER_RECOVER, mndProcessConsumerRecoverMsg);
+
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_CONSUMERS, mndRetrieveConsumer);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_CONSUMERS, mndCancelGetNextConsumer);
+
   return sdbSetTable(pMnode->pSdb, table);
 }
 
@@ -91,6 +99,30 @@ FAIL:
   return -1;
 }
 
+static int32_t mndProcessConsumerRecoverMsg(SNodeMsg *pMsg) {
+  SMnode                *pMnode = pMsg->pNode;
+  SMqConsumerRecoverMsg *pRecoverMsg = pMsg->rpcMsg.pCont;
+  SMqConsumerObj        *pConsumer = mndAcquireConsumer(pMnode, pRecoverMsg->consumerId);
+  ASSERT(pConsumer);
+
+  SMqConsumerObj *pConsumerNew = tNewSMqConsumerObj(pConsumer->consumerId, pConsumer->cgroup);
+  pConsumerNew->updateType = CONSUMER_UPDATE__RECOVER;
+
+  mndReleaseConsumer(pMnode, pConsumer);
+
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_TYPE_CONSUMER_RECOVER, &pMsg->rpcMsg);
+  if (pTrans == NULL) goto FAIL;
+  if (mndSetConsumerCommitLogs(pMnode, pTrans, pConsumerNew) != 0) goto FAIL;
+  if (mndTransPrepare(pMnode, pTrans) != 0) goto FAIL;
+
+  mndTransDrop(pTrans);
+  return 0;
+FAIL:
+  tDeleteSMqConsumerObj(pConsumerNew);
+  mndTransDrop(pTrans);
+  return -1;
+}
+
 static SMqRebSubscribe *mndGetOrCreateRebSub(SHashObj *pHash, const char *key) {
   SMqRebSubscribe *pRebSub = taosHashGet(pHash, key, strlen(key) + 1);
   if (pRebSub == NULL) {
@@ -216,8 +248,15 @@ static int32_t mndProcessAskEpReq(SNodeMsg *pMsg) {
   // 1. check consumer status
   int32_t status = atomic_load_32(&pConsumer->status);
 
-  if (status == MQ_CONSUMER_STATUS__LOST) {
+  if (status == MQ_CONSUMER_STATUS__LOST_REBD) {
-    // TODO: recover consumer
+    SMqConsumerRecoverMsg *pRecoverMsg = rpcMallocCont(sizeof(SMqConsumerRecoverMsg));
+
+    pRecoverMsg->consumerId = consumerId;
+    SRpcMsg *pRpcMsg = taosMemoryCalloc(1, sizeof(SRpcMsg));
+    pRpcMsg->msgType = TDMT_MND_MQ_CONSUMER_RECOVER;
+    pRpcMsg->pCont = pRecoverMsg;
+    pRpcMsg->contLen = sizeof(SMqConsumerRecoverMsg);
+    tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, pRpcMsg);
   }
 
   if (status != MQ_CONSUMER_STATUS__READY) {
@@ -366,10 +405,14 @@ static int32_t mndProcessSubscribeReq(SNodeMsg *pMsg) {
   if (pConsumerOld == NULL) {
     pConsumerNew = tNewSMqConsumerObj(consumerId, cgroup);
     pConsumerNew->updateType = CONSUMER_UPDATE__MODIFY;
-    /*pConsumerNew->waitingRebTopics = newSub;*/
     pConsumerNew->rebNewTopics = newSub;
     subscribe.topicNames = NULL;
 
+    for (int32_t i = 0; i < newTopicNum; i++) {
+      char *newTopicCopy = strdup(taosArrayGetP(newSub, i));
+      taosArrayPush(pConsumerNew->assignedTopics, &newTopicCopy);
+    }
+
     STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_TYPE_SUBSCRIBE, &pMsg->rpcMsg);
     if (pTrans == NULL) goto SUBSCRIBE_OVER;
     if (mndSetConsumerCommitLogs(pMnode, pTrans, pConsumerNew) != 0) goto SUBSCRIBE_OVER;
@@ -389,7 +432,11 @@ static int32_t mndProcessSubscribeReq(SNodeMsg *pMsg) {
       goto SUBSCRIBE_OVER;
     }
     pConsumerNew->updateType = CONSUMER_UPDATE__MODIFY;
-    /*pConsumerOld->waitingRebTopics = newSub;*/
+
+    for (int32_t i = 0; i < newTopicNum; i++) {
+      char *newTopicCopy = strdup(taosArrayGetP(newSub, i));
+      taosArrayPush(pConsumerNew->assignedTopics, &newTopicCopy);
+    }
 
     int32_t oldTopicNum = 0;
     if (pConsumerOld->currentTopics) {
@@ -532,6 +579,7 @@ CM_DECODE_OVER:
 
 static int32_t mndConsumerActionInsert(SSdb *pSdb, SMqConsumerObj *pConsumer) {
   mTrace("consumer:%" PRId64 ", perform insert action", pConsumer->consumerId);
+  pConsumer->subscribeTime = pConsumer->upTime;
   return 0;
 }
 
@@ -557,17 +605,45 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
     pOldConsumer->rebRemovedTopics = pNewConsumer->rebRemovedTopics;
     pNewConsumer->rebRemovedTopics = tmp;
 
+    tmp = pOldConsumer->assignedTopics;
+    pOldConsumer->assignedTopics = pNewConsumer->assignedTopics;
+    pNewConsumer->assignedTopics = tmp;
+
+    pOldConsumer->subscribeTime = pNewConsumer->upTime;
+
     pOldConsumer->status = MQ_CONSUMER_STATUS__MODIFY;
   } else if (pNewConsumer->updateType == CONSUMER_UPDATE__LOST) {
+    ASSERT(taosArrayGetSize(pOldConsumer->rebNewTopics) == 0);
+    ASSERT(taosArrayGetSize(pOldConsumer->rebRemovedTopics) == 0);
+
     int32_t sz = taosArrayGetSize(pOldConsumer->currentTopics);
-    pOldConsumer->rebRemovedTopics = taosArrayInit(sz, sizeof(void *));
+    /*pOldConsumer->rebRemovedTopics = taosArrayInit(sz, sizeof(void *));*/
     for (int32_t i = 0; i < sz; i++) {
       char *topic = strdup(taosArrayGetP(pOldConsumer->currentTopics, i));
-      taosArrayPush(pNewConsumer->rebRemovedTopics, &topic);
+      taosArrayPush(pOldConsumer->rebRemovedTopics, &topic);
     }
+
+    pOldConsumer->rebalanceTime = pNewConsumer->upTime;
+
     pOldConsumer->status = MQ_CONSUMER_STATUS__LOST;
+  } else if (pNewConsumer->updateType == CONSUMER_UPDATE__RECOVER) {
+    ASSERT(taosArrayGetSize(pOldConsumer->currentTopics) == 0);
+    ASSERT(taosArrayGetSize(pOldConsumer->rebNewTopics) == 0);
+
+    int32_t sz = taosArrayGetSize(pOldConsumer->assignedTopics);
+    for (int32_t i = 0; i < sz; i++) {
+      char *topic = strdup(taosArrayGetP(pOldConsumer->assignedTopics, i));
+      taosArrayPush(pOldConsumer->rebNewTopics, &topic);
+    }
+
+    pOldConsumer->rebalanceTime = pNewConsumer->upTime;
+
+    pOldConsumer->status = MQ_CONSUMER_STATUS__MODIFY;
   } else if (pNewConsumer->updateType == CONSUMER_UPDATE__TOUCH) {
     atomic_add_fetch_32(&pOldConsumer->epoch, 1);
+
+    pOldConsumer->rebalanceTime = pNewConsumer->upTime;
+
   } else if (pNewConsumer->updateType == CONSUMER_UPDATE__ADD) {
     ASSERT(taosArrayGetSize(pNewConsumer->rebNewTopics) == 1);
     ASSERT(taosArrayGetSize(pNewConsumer->rebRemovedTopics) == 0);
@@ -612,6 +688,9 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
         pOldConsumer->status = MQ_CONSUMER_STATUS__LOST_IN_REB;
       }
     }
+
+    pOldConsumer->rebalanceTime = pNewConsumer->upTime;
+
     atomic_add_fetch_32(&pOldConsumer->epoch, 1);
   } else if (pNewConsumer->updateType == CONSUMER_UPDATE__REMOVE) {
     ASSERT(taosArrayGetSize(pNewConsumer->rebNewTopics) == 0);
@@ -668,6 +747,9 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
         pOldConsumer->status = MQ_CONSUMER_STATUS__LOST_IN_REB;
       }
     }
+
+    pOldConsumer->rebalanceTime = pNewConsumer->upTime;
+
     atomic_add_fetch_32(&pOldConsumer->epoch, 1);
   }
 
@@ -688,3 +770,104 @@ void mndReleaseConsumer(SMnode *pMnode, SMqConsumerObj *pConsumer) {
   SSdb *pSdb = pMnode->pSdb;
   sdbRelease(pSdb, pConsumer);
 }
+
+static int32_t mndRetrieveConsumer(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rowsCapacity) {
+  SMnode         *pMnode = pReq->pNode;
+  SSdb           *pSdb = pMnode->pSdb;
+  int32_t         numOfRows = 0;
+  SMqConsumerObj *pConsumer = NULL;
+
+  while (numOfRows < rowsCapacity) {
+    pShow->pIter = sdbFetch(pSdb, SDB_CONSUMER, pShow->pIter, (void **)&pConsumer);
+    if (pShow->pIter == NULL) break;
+
+    SColumnInfoData *pColInfo;
+    int32_t          cols = 0;
+
+    taosRLockLatch(&pConsumer->lock);
+
+    // consumer id
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pConsumer->consumerId, false);
+
+    // group id
+    char groupId[TSDB_CGROUP_LEN + VARSTR_HEADER_SIZE] = {0};
+    tstrncpy(varDataVal(groupId), pConsumer->cgroup, TSDB_CGROUP_LEN);
+    varDataSetLen(groupId, strlen(varDataVal(groupId)));
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)groupId, false);
+
+    // app id
+    char appId[TSDB_CGROUP_LEN + VARSTR_HEADER_SIZE] = {0};
+    tstrncpy(varDataVal(appId), pConsumer->appId, TSDB_CGROUP_LEN);
+    varDataSetLen(appId, strlen(varDataVal(appId)));
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)appId, false);
+
+    // status
+    char status[20 + VARSTR_HEADER_SIZE] = {0};
+    tstrncpy(varDataVal(status), mndConsumerStatusName(pConsumer->status), 20);
+    varDataSetLen(status, strlen(varDataVal(status)));
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)status, false);
+
+    // subscribed topics
+    char  topics[TSDB_SHOW_LIST_LEN + VARSTR_HEADER_SIZE] = {0};
+    char *showStr = taosShowStrArray(pConsumer->assignedTopics);
+    tstrncpy(varDataVal(topics), showStr, TSDB_SHOW_LIST_LEN);
+    taosMemoryFree(showStr);
+    varDataSetLen(topics, strlen(varDataVal(topics)));
+
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)topics, false);
+
+    // pid
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pConsumer->pid, true);
+
+    // end point
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pConsumer->ep, true);
+
+    // up time
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pConsumer->upTime, false);
+
+    // subscribe time
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pConsumer->subscribeTime, false);
+
+    // rebalance time
+    pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+    colDataAppend(pColInfo, numOfRows, (const char *)&pConsumer->rebalanceTime, pConsumer->rebalanceTime == 0);
+
+    taosRUnLockLatch(&pConsumer->lock);
+    sdbRelease(pSdb, pConsumer);
+
+    numOfRows++;
+  }
+
+  pShow->numOfRows += numOfRows;
+  return numOfRows;
+}
+
+static void mndCancelGetNextConsumer(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}
+
+static const char *mndConsumerStatusName(int status) {
+  switch (status) {
+    case MQ_CONSUMER_STATUS__READY:
+      return "ready";
+    case MQ_CONSUMER_STATUS__LOST:
+    case MQ_CONSUMER_STATUS__LOST_REBD:
+    case MQ_CONSUMER_STATUS__LOST_IN_REB:
+      return "lost";
+    case MQ_CONSUMER_STATUS__MODIFY:
+    case MQ_CONSUMER_STATUS__MODIFY_IN_REB:
+      return "rebalancing";
+    default:
+      return "unknown";
+  }
+}

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

@@ -34,15 +34,20 @@ SMqConsumerObj *tNewSMqConsumerObj(int64_t consumerId, char cgroup[TSDB_CGROUP_L
   pConsumer->currentTopics = taosArrayInit(0, sizeof(void *));
   pConsumer->rebNewTopics = taosArrayInit(0, sizeof(void *));
   pConsumer->rebRemovedTopics = taosArrayInit(0, sizeof(void *));
+  pConsumer->assignedTopics = taosArrayInit(0, sizeof(void *));
 
-  if (pConsumer->currentTopics == NULL || pConsumer->rebNewTopics == NULL || pConsumer->rebRemovedTopics == NULL) {
+  if (pConsumer->currentTopics == NULL || pConsumer->rebNewTopics == NULL || pConsumer->rebRemovedTopics == NULL ||
+      pConsumer->assignedTopics == NULL) {
     taosArrayDestroy(pConsumer->currentTopics);
     taosArrayDestroy(pConsumer->rebNewTopics);
     taosArrayDestroy(pConsumer->rebRemovedTopics);
+    taosArrayDestroy(pConsumer->assignedTopics);
     taosMemoryFree(pConsumer);
     return NULL;
   }
 
+  pConsumer->upTime = taosGetTimestampMs();
+
   return pConsumer;
 }
 
@@ -56,6 +61,9 @@ void tDeleteSMqConsumerObj(SMqConsumerObj *pConsumer) {
   if (pConsumer->rebRemovedTopics) {
     taosArrayDestroyP(pConsumer->rebRemovedTopics, (FDelete)taosMemoryFree);
   }
+  if (pConsumer->assignedTopics) {
+    taosArrayDestroyP(pConsumer->assignedTopics, (FDelete)taosMemoryFree);
+  }
 }
 
 int32_t tEncodeSMqConsumerObj(void **buf, const SMqConsumerObj *pConsumer) {
@@ -67,6 +75,12 @@ int32_t tEncodeSMqConsumerObj(void **buf, const SMqConsumerObj *pConsumer) {
   tlen += taosEncodeFixedI32(buf, pConsumer->epoch);
   tlen += taosEncodeFixedI32(buf, pConsumer->status);
 
+  tlen += taosEncodeFixedI32(buf, pConsumer->pid);
+  tlen += taosEncodeSEpSet(buf, &pConsumer->ep);
+  tlen += taosEncodeFixedI64(buf, pConsumer->upTime);
+  tlen += taosEncodeFixedI64(buf, pConsumer->subscribeTime);
+  tlen += taosEncodeFixedI64(buf, pConsumer->rebalanceTime);
+
   // current topics
   if (pConsumer->currentTopics) {
     sz = taosArrayGetSize(pConsumer->currentTopics);
@@ -103,6 +117,18 @@ int32_t tEncodeSMqConsumerObj(void **buf, const SMqConsumerObj *pConsumer) {
     tlen += taosEncodeFixedI32(buf, 0);
   }
 
+  // lost topics
+  if (pConsumer->assignedTopics) {
+    sz = taosArrayGetSize(pConsumer->assignedTopics);
+    tlen += taosEncodeFixedI32(buf, sz);
+    for (int32_t i = 0; i < sz; i++) {
+      char *topic = taosArrayGetP(pConsumer->assignedTopics, i);
+      tlen += taosEncodeString(buf, topic);
+    }
+  } else {
+    tlen += taosEncodeFixedI32(buf, 0);
+  }
+
   return tlen;
 }
 
@@ -114,6 +140,12 @@ void *tDecodeSMqConsumerObj(const void *buf, SMqConsumerObj *pConsumer) {
   buf = taosDecodeFixedI32(buf, &pConsumer->epoch);
   buf = taosDecodeFixedI32(buf, &pConsumer->status);
 
+  buf = taosDecodeFixedI32(buf, &pConsumer->pid);
+  buf = taosDecodeSEpSet(buf, &pConsumer->ep);
+  buf = taosDecodeFixedI64(buf, &pConsumer->upTime);
+  buf = taosDecodeFixedI64(buf, &pConsumer->subscribeTime);
+  buf = taosDecodeFixedI64(buf, &pConsumer->rebalanceTime);
+
   // current topics
   buf = taosDecodeFixedI32(buf, &sz);
   pConsumer->currentTopics = taosArrayInit(sz, sizeof(void *));
@@ -141,6 +173,15 @@ void *tDecodeSMqConsumerObj(const void *buf, SMqConsumerObj *pConsumer) {
     taosArrayPush(pConsumer->rebRemovedTopics, &topic);
   }
 
+  // reb removed topics
+  buf = taosDecodeFixedI32(buf, &sz);
+  pConsumer->assignedTopics = taosArrayInit(sz, sizeof(void *));
+  for (int32_t i = 0; i < sz; i++) {
+    char *topic;
+    buf = taosDecodeString(buf, &topic);
+    taosArrayPush(pConsumer->assignedTopics, &topic);
+  }
+
   return (void *)buf;
 }
 
@@ -329,6 +370,7 @@ int32_t tEncodeSMqSubActionLogEntry(void **buf, const SMqSubActionLogEntry *pEnt
   tlen += taosEncodeArray(buf, pEntry->consumers, (FEncode)tEncodeSMqSubActionLogEntry);
   return tlen;
 }
+
 void *tDecodeSMqSubActionLogEntry(const void *buf, SMqSubActionLogEntry *pEntry) {
   buf = taosDecodeFixedI32(buf, &pEntry->epoch);
   buf = taosDecodeArray(buf, &pEntry->consumers, (FDecode)tDecodeSMqSubActionLogEntry, sizeof(SMqSubActionLogEntry));

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

@@ -142,6 +142,7 @@ static const SInfosTableSchema userTblsSchema[] = {
     {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "ttl", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "table_comment", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "type", .bytes = 20 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
 };
 
 static const SInfosTableSchema userTblDistSchema[] = {
@@ -164,7 +165,6 @@ static const SInfosTableSchema userUsersSchema[] = {
     {.name = "name", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "privilege", .bytes = 10 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
-    {.name = "account", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR},
 };
 
 static const SInfosTableSchema grantsSchema[] = {
@@ -199,23 +199,6 @@ static const SInfosTableSchema vgroupsSchema[] = {
     {.name = "file_size", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
 };
 
-static const SInfosTableSchema consumerSchema[] = {
-    {.name = "client_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "pid", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-    {.name = "status", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-    // ep
-    // up time
-    // topics
-};
-
-static const SInfosTableSchema subscribeSchema[] = {
-    {.name = "topic_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
-    {.name = "client_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
-};
-
 static const SInfosTableSchema smaSchema[] = {
     {.name = "sma_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR},
     {.name = "create_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
@@ -282,8 +265,6 @@ static const SInfosTableMeta infosMeta[] = {
     {TSDB_INS_TABLE_USER_USERS, userUsersSchema, tListLen(userUsersSchema)},
     {TSDB_INS_TABLE_LICENCES, grantsSchema, tListLen(grantsSchema)},
     {TSDB_INS_TABLE_VGROUPS, vgroupsSchema, tListLen(vgroupsSchema)},
-    {TSDB_INS_TABLE_CONSUMERS, consumerSchema, tListLen(consumerSchema)},
-    {TSDB_INS_TABLE_SUBSCRIBES, subscribeSchema, tListLen(subscribeSchema)},
     {TSDB_INS_TABLE_TRANS, transSchema, tListLen(transSchema)},
     {TSDB_INS_TABLE_SMAS, smaSchema, tListLen(smaSchema)},
     {TSDB_INS_TABLE_CONFIGS, configSchema, tListLen(configSchema)},

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

@@ -49,13 +49,15 @@ static const SPerfsTableSchema topicSchema[] = {
 
 static const SPerfsTableSchema consumerSchema[] = {
     {.name = "consumer_id", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
-    {.name = "app_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
     {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
-    {.name = "status", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
+    {.name = "app_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "status", .bytes = 20 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
     {.name = "topics", .bytes = TSDB_SHOW_LIST_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
     {.name = "pid", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "end_point", .bytes = TSDB_EP_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_BINARY},
     {.name = "up_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "subscribe_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
+    {.name = "rebalance_time", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP},
 };
 
 static const SPerfsTableSchema subscriptionSchema[] = {
@@ -63,6 +65,12 @@ static const SPerfsTableSchema subscriptionSchema[] = {
     {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
     {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "consumer_id", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
+};
+
+static const SPerfsTableSchema offsetSchema[] = {
+    {.name = "topic_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "group_id", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_BINARY},
+    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT},
     {.name = "committed_offset", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
     {.name = "current_offset", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
     {.name = "skip_log_cnt", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT},
@@ -74,6 +82,7 @@ static const SPerfsTableMeta perfsMeta[] = {
     {TSDB_PERFS_TABLE_TOPICS, topicSchema, tListLen(topicSchema)},
     {TSDB_PERFS_TABLE_CONSUMERS, consumerSchema, tListLen(consumerSchema)},
     {TSDB_PERFS_TABLE_SUBSCRIPTIONS, subscriptionSchema, tListLen(subscriptionSchema)},
+    {TSDB_PERFS_TABLE_OFFSETS, offsetSchema, tListLen(offsetSchema)},
 };
 
 // connection/application/

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

@@ -196,7 +196,7 @@ static int32_t mndProcessConnectReq(SNodeMsg *pReq) {
     goto CONN_OVER;
   }
   if (0 != strncmp(connReq.passwd, pUser->pass, TSDB_PASSWORD_LEN - 1)) {
-    mError("user:%s, failed to auth while acquire user\n %s \r\n %s", pReq->user, connReq.passwd, pUser->pass);
+    mError("user:%s, failed to auth while acquire user, input:%s saved:%s", pReq->user, connReq.passwd, pUser->pass);
     code = TSDB_CODE_RPC_AUTH_FAILURE;
     goto CONN_OVER;
   }

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

@@ -489,7 +489,7 @@ int32_t mndSchedInitSubEp(SMnode* pMnode, const SMqTopicObj* pTopic, SMqSubscrib
     int32_t levelNum = LIST_LENGTH(pPlan->pSubplans);
     if (levelNum != 1) {
       qDestroyQueryPlan(pPlan);
-      terrno = TSDB_CODE_MND_UNSUPPORTED_TOPIC;
+      terrno = TSDB_CODE_MND_INVALID_TOPIC_QUERY;
       return -1;
     }
 
@@ -498,7 +498,7 @@ int32_t mndSchedInitSubEp(SMnode* pMnode, const SMqTopicObj* pTopic, SMqSubscrib
     int32_t opNum = LIST_LENGTH(inner->pNodeList);
     if (opNum != 1) {
       qDestroyQueryPlan(pPlan);
-      terrno = TSDB_CODE_MND_UNSUPPORTED_TOPIC;
+      terrno = TSDB_CODE_MND_INVALID_TOPIC_QUERY;
       return -1;
     }
     plan = nodesListGetNode(inner->pNodeList, 0);

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

@@ -261,7 +261,7 @@ static SDDropTopicReq *mndBuildDropTopicMsg(SMnode *pMnode, SVgObj *pVgroup, SMq
 
 static int32_t mndCheckCreateTopicReq(SCMCreateTopicReq *pCreate) {
   if (pCreate->name[0] == 0 || pCreate->sql == NULL || pCreate->sql[0] == 0 || pCreate->subscribeDbName[0] == 0) {
-    terrno = TSDB_CODE_MND_INVALID_TOPIC_OPTION;
+    terrno = TSDB_CODE_MND_INVALID_TOPIC;
     return -1;
   }
 

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

@@ -165,8 +165,9 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
   int32_t numOfWriteDbs = 0;
   SDB_GET_INT32(pRaw, dataPos, &numOfReadDbs, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &numOfWriteDbs, _OVER)
-  pUser->readDbs = taosHashInit(numOfReadDbs, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, true);
+  pUser->readDbs = taosHashInit(numOfReadDbs, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
-  pUser->writeDbs = taosHashInit(numOfWriteDbs, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, true);
+  pUser->writeDbs =
+      taosHashInit(numOfWriteDbs, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   if (pUser->readDbs == NULL || pUser->writeDbs == NULL) goto _OVER;
 
   for (int32_t i = 0; i < numOfReadDbs; ++i) {
@@ -340,13 +341,13 @@ _OVER:
   return code;
 }
 
-static int32_t mndUpdateUser(SMnode *pMnode, SUserObj *pOld, SUserObj *pNew, SNodeMsg *pReq) {
+static int32_t mndAlterUser(SMnode *pMnode, SUserObj *pOld, SUserObj *pNew, SNodeMsg *pReq) {
   STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_TYPE_ALTER_USER, &pReq->rpcMsg);
   if (pTrans == NULL) {
-    mError("user:%s, failed to update since %s", pOld->user, terrstr());
+    mError("user:%s, failed to alter since %s", pOld->user, terrstr());
     return -1;
   }
-  mDebug("trans:%d, used to update user:%s", pTrans->id, pOld->user);
+  mDebug("trans:%d, used to alter user:%s", pTrans->id, pOld->user);
 
   SSdbRaw *pRedoRaw = mndUserActionEncode(pNew);
   if (pRedoRaw == NULL || mndTransAppendRedolog(pTrans, pRedoRaw) != 0) {
@@ -367,7 +368,8 @@ static int32_t mndUpdateUser(SMnode *pMnode, SUserObj *pOld, SUserObj *pNew, SNo
 }
 
 static SHashObj *mndDupDbHash(SHashObj *pOld) {
-  SHashObj *pNew = taosHashInit(4, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, true);
+  SHashObj *pNew =
+      taosHashInit(taosHashGetSize(pOld), taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   if (pNew == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
@@ -378,8 +380,8 @@ static SHashObj *mndDupDbHash(SHashObj *pOld) {
     int32_t len = strlen(db) + 1;
     if (taosHashPut(pNew, db, len, db, TSDB_DB_FNAME_LEN) != 0) {
       taosHashCancelIterate(pOld, db);
-      terrno = TSDB_CODE_OUT_OF_MEMORY;
       taosHashCleanup(pNew);
+      terrno = TSDB_CODE_OUT_OF_MEMORY;
       return NULL;
     }
     db = taosHashIterate(pOld, db);
@@ -439,7 +441,7 @@ static int32_t mndProcessAlterUserReq(SNodeMsg *pReq) {
   if (alterReq.alterType == TSDB_ALTER_USER_PASSWD) {
     char pass[TSDB_PASSWORD_LEN + 1] = {0};
     taosEncryptPass_c((uint8_t *)alterReq.pass, strlen(alterReq.pass), pass);
-    memcpy(pUser->pass, pass, TSDB_PASSWORD_LEN);
+    memcpy(newUser.pass, pass, TSDB_PASSWORD_LEN);
   } else if (alterReq.alterType == TSDB_ALTER_USER_SUPERUSER) {
     newUser.superUser = alterReq.superUser;
   } else if (alterReq.alterType == TSDB_ALTER_USER_ADD_READ_DB) {
@@ -485,7 +487,7 @@ static int32_t mndProcessAlterUserReq(SNodeMsg *pReq) {
     goto _OVER;
   }
 
-  code = mndUpdateUser(pMnode, pUser, &newUser, pReq);
+  code = mndAlterUser(pMnode, pUser, &newUser, pReq);
   if (code == 0) code = TSDB_CODE_MND_ACTION_IN_PROGRESS;
 
 _OVER:
@@ -632,8 +634,7 @@ static int32_t mndProcessGetUserAuthReq(SNodeMsg *pReq) {
 
 _OVER:
   mndReleaseUser(pMnode, pUser);
-  taosHashCleanup(authRsp.readDbs);
+  tFreeSGetUserAuthRsp(&authRsp);
-  taosHashCleanup(authRsp.writeDbs);
 
   return code;
 }
@@ -670,11 +671,6 @@ static int32_t mndRetrieveUsers(SNodeMsg *pReq, SShowObj *pShow, SSDataBlock *pB
     pColInfo = taosArrayGet(pBlock->pDataBlock, cols);
     colDataAppend(pColInfo, numOfRows, (const char *)&pUser->createdTime, false);
 
-    cols++;
-    pColInfo = taosArrayGet(pBlock->pDataBlock, cols);
-    STR_WITH_MAXSIZE_TO_VARSTR(name, pUser->acct, pShow->bytes[cols]);
-    colDataAppend(pColInfo, numOfRows, (const char *)name, false);
-
     numOfRows++;
     sdbRelease(pSdb, pUser);
   }

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

@@ -43,7 +43,7 @@
 #include "mndUser.h"
 #include "mndVgroup.h"
 
-#define MQ_TIMER_MS    3000
+#define MQ_TIMER_MS    2000
 #define TRNAS_TIMER_MS 6000
 
 static void *mndBuildTimerMsg(int32_t *pContLen) {
@@ -418,7 +418,6 @@ int64_t mndGenerateUid(char *name, int32_t len) {
   } while (true);
 }
 
-
 int32_t mndGetMonitorInfo(SMnode *pMnode, SMonClusterInfo *pClusterInfo, SMonVgroupInfo *pVgroupInfo,
                           SMonGrantInfo *pGrantInfo) {
   if (!mndIsMaster(pMnode)) return -1;

source/dnode/vnode/CMakeLists.txt

@@ -27,10 +27,9 @@ target_sources(
     "src/tsdb/tsdbTDBImpl.c"
     "src/tsdb/tsdbCommit.c"
     "src/tsdb/tsdbCommit2.c"
-    "src/tsdb/tsdbCompact.c"
     "src/tsdb/tsdbFile.c"
     "src/tsdb/tsdbFS.c"
-    "src/tsdb/tsdbMain.c"
+    "src/tsdb/tsdbOpen.c"
     "src/tsdb/tsdbMemTable.c"
     "src/tsdb/tsdbRead.c"
     "src/tsdb/tsdbReadImpl.c"

source/dnode/vnode/inc/vnode.h

@@ -68,6 +68,7 @@ void    vnodeStop(SVnode *pVnode);
 
 int64_t vnodeGetSyncHandle(SVnode *pVnode);
 void    vnodeGetSnapshot(SVnode *pVnode, SSnapshot *pSnapshot);
+void    vnodeGetInfo(SVnode *pVnode, const char **dbname, int32_t *vgId);
 
 // meta
 typedef struct SMeta       SMeta;  // todo: remove
@@ -76,6 +77,7 @@ typedef struct SMetaEntry  SMetaEntry;
 
 void metaReaderInit(SMetaReader *pReader, SMeta *pMeta, int32_t flags);
 void metaReaderClear(SMetaReader *pReader);
+int  metaGetTableEntryByUid(SMetaReader *pReader, tb_uid_t uid);
 int  metaReadNext(SMetaReader *pReader);
 
 #if 1  // refact APIs below (TODO)
@@ -133,10 +135,6 @@ int32_t tqRetrieveDataBlock(SArray **ppCols, STqReadHandle *pHandle, uint64_t *p
 // need to reposition
 
 // structs
-struct SMetaCfg {
-  uint64_t lruSize;
-};
-
 struct STsdbCfg {
   int8_t  precision;
   int8_t  update;
@@ -145,10 +143,9 @@ struct STsdbCfg {
   int32_t days;
   int32_t minRows;
   int32_t maxRows;
-  int32_t  keep2;
   int32_t keep0;
   int32_t keep1;
-  uint64_t lruCacheSize;
+  int32_t keep2;
   SArray *retentions;
 };
 
@@ -160,8 +157,6 @@ struct SVnodeCfg {
   int32_t  szCache;
   uint64_t szBuf;
   bool     isHeap;
-  uint32_t ttl;
-  uint32_t keep;
   int8_t   streamMode;
   bool     isWeak;
   STsdbCfg tsdbCfg;

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

@@ -46,12 +46,45 @@ int  tsdbLoadDataFromCache(STable *pTable, SSkipListIterator *pIter, TSKEY maxKe
 
 // tsdbCommit ================
 
-#if 1
+// tsdbFS ================
+typedef struct STsdbFS STsdbFS;
 
-typedef struct SSmaStat SSmaStat;
+// tsdbSma ================
 typedef struct SSmaEnv  SSmaEnv;
 typedef struct SSmaEnvs SSmaEnvs;
 
+// structs
+typedef struct {
+  int   minFid;
+  int   midFid;
+  int   maxFid;
+  TSKEY minKey;
+} SRtn;
+
+struct SSmaEnvs {
+  int16_t  nTSma;
+  int16_t  nRSma;
+  SSmaEnv *pTSmaEnv;
+  SSmaEnv *pRSmaEnv;
+};
+
+struct STsdb {
+  char          *path;
+  SVnode        *pVnode;
+  bool           repoLocked;
+  TdThreadMutex  mutex;
+  STsdbCfg       config;
+  STsdbMemTable *mem;
+  STsdbMemTable *imem;
+  SRtn           rtn;
+  STsdbFS       *fs;
+  SSmaEnvs       smaEnvs;
+};
+
+#if 1  // ======================================
+
+typedef struct SSmaStat SSmaStat;
+
 struct STable {
   uint64_t  tid;
   uint64_t  uid;
@@ -97,13 +130,6 @@ typedef struct {
   uint8_t   state;
 } SDFile;
 
-struct SSmaEnvs {
-  int16_t  nTSma;
-  int16_t  nRSma;
-  SSmaEnv *pTSmaEnv;
-  SSmaEnv *pRSmaEnv;
-};
-
 typedef struct {
   int      fid;
   int8_t   state;  // -128~127
@@ -112,13 +138,6 @@ typedef struct {
   SDFile   files[TSDB_FILE_MAX];
 } SDFileSet;
 
-typedef struct {
-  int   minFid;
-  int   midFid;
-  int   maxFid;
-  TSKEY minKey;
-} SRtn;
-
 struct STbData {
   tb_uid_t   uid;
   TSKEY      keyMin;
@@ -155,7 +174,7 @@ typedef struct {
   SArray     *sf;    // sma data file array    v2f1900.index_name_1
 } SFSStatus;
 
-typedef struct {
+struct STsdbFS {
   TdThreadRwlock lock;
 
   SFSStatus *cstatus;        // current status
@@ -163,23 +182,9 @@ typedef struct {
   SHashObj  *metaCacheComp;  // meta cache for compact
   bool       intxn;
   SFSStatus *nstatus;  // new status
-} STsdbFS;
-
-struct STsdb {
-  char          *path;
-  SVnode        *pVnode;
-  int32_t        vgId;
-  bool           repoLocked;
-  TdThreadMutex  mutex;
-  STsdbCfg       config;
-  STsdbMemTable *mem;
-  STsdbMemTable *imem;
-  SRtn           rtn;
-  STsdbFS       *fs;
-  SSmaEnvs       smaEnvs;
 };
 
-#define REPO_ID(r)        ((r)->vgId)
+#define REPO_ID(r)        TD_VID((r)->pVnode)
 #define REPO_CFG(r)       (&(r)->config)
 #define REPO_FS(r)        ((r)->fs)
 #define REPO_META(r)      ((r)->pVnode->pMeta)
@@ -507,12 +512,6 @@ static FORCE_INLINE void *taosTZfree(void *ptr) {
 
 // tsdbCommit
 
-typedef struct {
-  uint64_t uid;
-  int64_t  offset;
-  int64_t  size;
-} SKVRecord;
-
 void tsdbGetRtnSnap(STsdb *pRepo, SRtn *pRtn);
 
 static FORCE_INLINE int TSDB_KEY_FID(TSKEY key, int32_t days, int8_t precision) {
@@ -891,66 +890,6 @@ static FORCE_INLINE int tsdbUnLockFS(STsdbFS *pFs) {
   return 0;
 }
 
-// tsdbSma
-// #define TSDB_SMA_TEST // remove after test finished
-
-// struct SSmaEnv {
-//   TdThreadRwlock lock;
-//   SDiskID        did;
-//   TDBEnv         dbEnv;  // TODO: If it's better to put it in smaIndex level?
-//   char          *path;   // relative path
-//   SSmaStat      *pStat;
-// };
-
-// #define SMA_ENV_LOCK(env)       ((env)->lock)
-// #define SMA_ENV_DID(env)        ((env)->did)
-// #define SMA_ENV_ENV(env)        ((env)->dbEnv)
-// #define SMA_ENV_PATH(env)       ((env)->path)
-// #define SMA_ENV_STAT(env)       ((env)->pStat)
-// #define SMA_ENV_STAT_ITEMS(env) ((env)->pStat->smaStatItems)
-
-// void  tsdbDestroySmaEnv(SSmaEnv *pSmaEnv);
-// void *tsdbFreeSmaEnv(SSmaEnv *pSmaEnv);
-// #if 0
-// int32_t tsdbGetTSmaStatus(STsdb *pTsdb, STSma *param, void *result);
-// int32_t tsdbRemoveTSmaData(STsdb *pTsdb, STSma *param, STimeWindow *pWin);
-// #endif
-
-// // internal func
-// static FORCE_INLINE int32_t tsdbEncodeTSmaKey(int64_t groupId, TSKEY tsKey, void **pData) {
-//   int32_t len = 0;
-//   len += taosEncodeFixedI64(pData, tsKey);
-//   len += taosEncodeFixedI64(pData, groupId);
-//   return len;
-// }
-
-// static FORCE_INLINE int32_t tsdbRLockSma(SSmaEnv *pEnv) {
-//   int code = taosThreadRwlockRdlock(&(pEnv->lock));
-//   if (code != 0) {
-//     terrno = TAOS_SYSTEM_ERROR(code);
-//     return -1;
-//   }
-//   return 0;
-// }
-
-// static FORCE_INLINE int32_t tsdbWLockSma(SSmaEnv *pEnv) {
-//   int code = taosThreadRwlockWrlock(&(pEnv->lock));
-//   if (code != 0) {
-//     terrno = TAOS_SYSTEM_ERROR(code);
-//     return -1;
-//   }
-//   return 0;
-// }
-
-// static FORCE_INLINE int32_t tsdbUnLockSma(SSmaEnv *pEnv) {
-//   int code = taosThreadRwlockUnlock(&(pEnv->lock));
-//   if (code != 0) {
-//     terrno = TAOS_SYSTEM_ERROR(code);
-//     return -1;
-//   }
-//   return 0;
-// }
-
 typedef struct SSmaKey SSmaKey;
 
 struct SSmaKey {

source/dnode/vnode/src/inc/vnodeInt.h

@@ -75,7 +75,6 @@ int             metaCreateSTable(SMeta* pMeta, int64_t version, SVCreateStbReq*
 int             metaCreateTable(SMeta* pMeta, int64_t version, SVCreateTbReq* pReq);
 SSchemaWrapper* metaGetTableSchema(SMeta* pMeta, tb_uid_t uid, int32_t sver, bool isinline);
 STSchema*       metaGetTbTSchema(SMeta* pMeta, tb_uid_t uid, int32_t sver);
-int             metaGetTableEntryByUid(SMetaReader* pReader, tb_uid_t uid);
 int             metaGetTableEntryByName(SMetaReader* pReader, const char* name);
 int             metaGetTbNum(SMeta* pMeta);
 SMCtbCursor*    metaOpenCtbCursor(SMeta* pMeta, tb_uid_t uid);

source/dnode/vnode/src/meta/metaTable.c

@@ -88,7 +88,7 @@ int metaCreateTable(SMeta *pMeta, int64_t version, SVCreateTbReq *pReq) {
 
   // preprocess req
   pReq->uid = tGenIdPI64();
-  pReq->ctime = taosGetTimestampSec();
+  pReq->ctime = taosGetTimestampMs();
 
   // validate req
   metaReaderInit(&mr, pMeta, 0);

source/dnode/vnode/src/tsdb/tsdbCommit.c

@@ -752,334 +752,7 @@ int tsdbWriteBlockIdx(SDFile *pHeadf, SArray *pIdxA, void **ppBuf) {
   return 0;
 }
 
-// // =================== Commit Meta Data
+// =================== Commit Time-Series Data
-// static int tsdbInitCommitMetaFile(STsdbRepo *pRepo, SMFile* pMf, bool open) {
-//   STsdbFS *  pfs = REPO_FS(pRepo);
-//   SMFile *   pOMFile = pfs->cstatus->pmf;
-//   SDiskID    did;
-
-//   // Create/Open a meta file or open the existing file
-//   if (pOMFile == NULL) {
-//     // Create a new meta file
-//     did.level = TFS_PRIMARY_LEVEL;
-//     did.id = TFS_PRIMARY_ID;
-//     tsdbInitMFile(pMf, did, REPO_ID(pRepo), FS_TXN_VERSION(REPO_FS(pRepo)));
-
-//     if (open && tsdbCreateMFile(pMf, true) < 0) {
-//       tsdbError("vgId:%d failed to create META file since %s", REPO_ID(pRepo), tstrerror(terrno));
-//       return -1;
-//     }
-
-//     tsdbInfo("vgId:%d meta file %s is created to commit", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMf));
-//   } else {
-//     tsdbInitMFileEx(pMf, pOMFile);
-//     if (open && tsdbOpenMFile(pMf, O_WRONLY) < 0) {
-//       tsdbError("vgId:%d failed to open META file since %s", REPO_ID(pRepo), tstrerror(terrno));
-//       return -1;
-//     }
-//   }
-
-//   return 0;
-// }
-
-// static int tsdbCommitMeta(STsdbRepo *pRepo) {
-//   STsdbFS *  pfs = REPO_FS(pRepo);
-//   SMemTable *pMem = pRepo->imem;
-//   SMFile *   pOMFile = pfs->cstatus->pmf;
-//   SMFile     mf;
-//   SActObj *  pAct = NULL;
-//   SActCont * pCont = NULL;
-//   SListNode *pNode = NULL;
-
-//   ASSERT(pOMFile != NULL || listNEles(pMem->actList) > 0);
-
-//   if (listNEles(pMem->actList) <= 0) {
-//     // no meta data to commit, just keep the old meta file
-//     tsdbUpdateMFile(pfs, pOMFile);
-//     if (tsTsdbMetaCompactRatio > 0) {
-//       if (tsdbInitCommitMetaFile(pRepo, &mf, false) < 0) {
-//         return -1;
-//       }
-//       int ret = tsdbCompactMetaFile(pRepo, pfs, &mf);
-//       if (ret < 0) tsdbError("compact meta file error");
-
-//       return ret;
-//     }
-//     return 0;
-//   } else {
-//     if (tsdbInitCommitMetaFile(pRepo, &mf, true) < 0) {
-//       return -1;
-//     }
-//   }
-
-//   // Loop to write
-//   while ((pNode = tdListPopHead(pMem->actList)) != NULL) {
-//     pAct = (SActObj *)pNode->data;
-//     if (pAct->act == TSDB_UPDATE_META) {
-//       pCont = (SActCont *)POINTER_SHIFT(pAct, sizeof(SActObj));
-//       if (tsdbUpdateMetaRecord(pfs, &mf, pAct->uid, (void *)(pCont->cont), pCont->len, false) < 0) {
-//         tsdbError("vgId:%d failed to update META record, uid %" PRIu64 " since %s", REPO_ID(pRepo), pAct->uid,
-//                   tstrerror(terrno));
-//         tsdbCloseMFile(&mf);
-//         (void)tsdbApplyMFileChange(&mf, pOMFile);
-//         // TODO: need to reload metaCache
-//         return -1;
-//       }
-//     } else if (pAct->act == TSDB_DROP_META) {
-//       if (tsdbDropMetaRecord(pfs, &mf, pAct->uid) < 0) {
-//         tsdbError("vgId:%d failed to drop META record, uid %" PRIu64 " since %s", REPO_ID(pRepo), pAct->uid,
-//                   tstrerror(terrno));
-//         tsdbCloseMFile(&mf);
-//         tsdbApplyMFileChange(&mf, pOMFile);
-//         // TODO: need to reload metaCache
-//         return -1;
-//       }
-//     } else {
-//       ASSERT(false);
-//     }
-//   }
-
-//   if (tsdbUpdateMFileHeader(&mf) < 0) {
-//     tsdbError("vgId:%d failed to update META file header since %s, revert it", REPO_ID(pRepo), tstrerror(terrno));
-//     tsdbApplyMFileChange(&mf, pOMFile);
-//     // TODO: need to reload metaCache
-//     return -1;
-//   }
-
-//   TSDB_FILE_FSYNC(&mf);
-//   tsdbCloseMFile(&mf);
-//   tsdbUpdateMFile(pfs, &mf);
-
-//   if (tsTsdbMetaCompactRatio > 0 && tsdbCompactMetaFile(pRepo, pfs, &mf) < 0) {
-//     tsdbError("compact meta file error");
-//   }
-
-//   return 0;
-// }
-
-// int tsdbEncodeKVRecord(void **buf, SKVRecord *pRecord) {
-//   int tlen = 0;
-//   tlen += taosEncodeFixedU64(buf, pRecord->uid);
-//   tlen += taosEncodeFixedI64(buf, pRecord->offset);
-//   tlen += taosEncodeFixedI64(buf, pRecord->size);
-
-//   return tlen;
-// }
-
-// void *tsdbDecodeKVRecord(void *buf, SKVRecord *pRecord) {
-//   buf = taosDecodeFixedU64(buf, &(pRecord->uid));
-//   buf = taosDecodeFixedI64(buf, &(pRecord->offset));
-//   buf = taosDecodeFixedI64(buf, &(pRecord->size));
-
-//   return buf;
-// }
-
-// static int tsdbUpdateMetaRecord(STsdbFS *pfs, SMFile *pMFile, uint64_t uid, void *cont, int contLen, bool compact) {
-//   char      buf[64] = "\0";
-//   void *    pBuf = buf;
-//   SKVRecord rInfo;
-//   int64_t   offset;
-
-//   // Seek to end of meta file
-//   offset = tsdbSeekMFile(pMFile, 0, SEEK_END);
-//   if (offset < 0) {
-//     return -1;
-//   }
-
-//   rInfo.offset = offset;
-//   rInfo.uid = uid;
-//   rInfo.size = contLen;
-
-//   int tlen = tsdbEncodeKVRecord((void **)(&pBuf), &rInfo);
-//   if (tsdbAppendMFile(pMFile, buf, tlen, NULL) < tlen) {
-//     return -1;
-//   }
-
-//   if (tsdbAppendMFile(pMFile, cont, contLen, NULL) < contLen) {
-//     return -1;
-//   }
-
-//   tsdbUpdateMFileMagic(pMFile, POINTER_SHIFT(cont, contLen - sizeof(TSCKSUM)));
-
-//   SHashObj* cache = compact ? pfs->metaCacheComp : pfs->metaCache;
-
-//   pMFile->info.nRecords++;
-
-//   SKVRecord *pRecord = taosHashGet(cache, (void *)&uid, sizeof(uid));
-//   if (pRecord != NULL) {
-//     pMFile->info.tombSize += (pRecord->size + sizeof(SKVRecord));
-//   } else {
-//     pMFile->info.nRecords++;
-//   }
-//   taosHashPut(cache, (void *)(&uid), sizeof(uid), (void *)(&rInfo), sizeof(rInfo));
-
-//   return 0;
-// }
-
-// static int tsdbDropMetaRecord(STsdbFS *pfs, SMFile *pMFile, uint64_t uid) {
-//   SKVRecord rInfo = {0};
-//   char      buf[128] = "\0";
-
-//   SKVRecord *pRecord = taosHashGet(pfs->metaCache, (void *)(&uid), sizeof(uid));
-//   if (pRecord == NULL) {
-//     tsdbError("failed to drop META record with key %" PRIu64 " since not find", uid);
-//     return -1;
-//   }
-
-//   rInfo.offset = -pRecord->offset;
-//   rInfo.uid = pRecord->uid;
-//   rInfo.size = pRecord->size;
-
-//   void *pBuf = buf;
-//   tsdbEncodeKVRecord(&pBuf, &rInfo);
-
-//   if (tsdbAppendMFile(pMFile, buf, sizeof(SKVRecord), NULL) < 0) {
-//     return -1;
-//   }
-
-//   pMFile->info.magic = taosCalcChecksum(pMFile->info.magic, (uint8_t *)buf, sizeof(SKVRecord));
-//   pMFile->info.nDels++;
-//   pMFile->info.nRecords--;
-//   pMFile->info.tombSize += (rInfo.size + sizeof(SKVRecord) * 2);
-
-//   taosHashRemove(pfs->metaCache, (void *)(&uid), sizeof(uid));
-//   return 0;
-// }
-
-// static int tsdbCompactMetaFile(STsdbRepo *pRepo, STsdbFS *pfs, SMFile *pMFile) {
-//   float delPercent = (float)(pMFile->info.nDels) / (float)(pMFile->info.nRecords);
-//   float tombPercent = (float)(pMFile->info.tombSize) / (float)(pMFile->info.size);
-//   float compactRatio = (float)(tsTsdbMetaCompactRatio)/100;
-
-//   if (delPercent < compactRatio && tombPercent < compactRatio) {
-//     return 0;
-//   }
-
-//   if (tsdbOpenMFile(pMFile, O_RDONLY) < 0) {
-//     tsdbError("open meta file %s compact fail", pMFile->f.rname);
-//     return -1;
-//   }
-
-//   tsdbInfo("begin compact tsdb meta file, ratio:%d, nDels:%" PRId64 ",nRecords:%" PRId64 ",tombSize:%" PRId64
-//   ",size:%" PRId64,
-//     tsTsdbMetaCompactRatio, pMFile->info.nDels,pMFile->info.nRecords,pMFile->info.tombSize,pMFile->info.size);
-
-//   SMFile mf;
-//   SDiskID did;
-
-//   // first create tmp meta file
-//   did.level = TFS_PRIMARY_LEVEL;
-//   did.id = TFS_PRIMARY_ID;
-//   tsdbInitMFile(&mf, did, REPO_ID(pRepo), FS_TXN_VERSION(REPO_FS(pRepo)) + 1);
-
-//   if (tsdbCreateMFile(&mf, true) < 0) {
-//     tsdbError("vgId:%d failed to create META file since %s", REPO_ID(pRepo), tstrerror(terrno));
-//     return -1;
-//   }
-
-//   tsdbInfo("vgId:%d meta file %s is created to compact meta data", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(&mf));
-
-//   // second iterator metaCache
-//   int code = -1;
-//   int64_t maxBufSize = 1024;
-//   SKVRecord *pRecord;
-//   void *pBuf = NULL;
-
-//   pBuf = taosMemoryMalloc((size_t)maxBufSize);
-//   if (pBuf == NULL) {
-//     goto _err;
-//   }
-
-//   // init Comp
-//   assert(pfs->metaCacheComp == NULL);
-//   pfs->metaCacheComp = taosHashInit(4096, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, HASH_NO_LOCK);
-//   if (pfs->metaCacheComp == NULL) {
-//     goto _err;
-//   }
-
-//   pRecord = taosHashIterate(pfs->metaCache, NULL);
-//   while (pRecord) {
-//     if (tsdbSeekMFile(pMFile, pRecord->offset + sizeof(SKVRecord), SEEK_SET) < 0) {
-//       tsdbError("vgId:%d failed to seek file %s since %s", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile),
-//                 tstrerror(terrno));
-//       goto _err;
-//     }
-//     if (pRecord->size > maxBufSize) {
-//       maxBufSize = pRecord->size;
-//       void* tmp = taosMemoryRealloc(pBuf, (size_t)maxBufSize);
-//       if (tmp == NULL) {
-//         goto _err;
-//       }
-//       pBuf = tmp;
-//     }
-//     int nread = (int)tsdbReadMFile(pMFile, pBuf, pRecord->size);
-//     if (nread < 0) {
-//       tsdbError("vgId:%d failed to read file %s since %s", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile),
-//         tstrerror(terrno));
-//       goto _err;
-//     }
-
-//     if (nread < pRecord->size) {
-//       tsdbError("vgId:%d failed to read file %s since file corrupted, expected read:%" PRId64 " actual read:%d",
-//                 REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile), pRecord->size, nread);
-//       goto _err;
-//     }
-
-//     if (tsdbUpdateMetaRecord(pfs, &mf, pRecord->uid, pBuf, (int)pRecord->size, true) < 0) {
-//       tsdbError("vgId:%d failed to update META record, uid %" PRIu64 " since %s", REPO_ID(pRepo), pRecord->uid,
-//                 tstrerror(terrno));
-//       goto _err;
-//     }
-
-//     pRecord = taosHashIterate(pfs->metaCache, pRecord);
-//   }
-//   code = 0;
-
-// _err:
-//   if (code == 0) TSDB_FILE_FSYNC(&mf);
-//   tsdbCloseMFile(&mf);
-//   tsdbCloseMFile(pMFile);
-
-//   if (code == 0) {
-//     // rename meta.tmp -> meta
-//     tsdbInfo("vgId:%d meta file rename %s -> %s", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(&mf),
-//     TSDB_FILE_FULL_NAME(pMFile)); taosRename(mf.f.aname,pMFile->f.aname); tstrncpy(mf.f.aname, pMFile->f.aname,
-//     TSDB_FILENAME_LEN); tstrncpy(mf.f.rname, pMFile->f.rname, TSDB_FILENAME_LEN);
-//     // update current meta file info
-//     pfs->nstatus->pmf = NULL;
-//     tsdbUpdateMFile(pfs, &mf);
-
-//     taosHashCleanup(pfs->metaCache);
-//     pfs->metaCache = pfs->metaCacheComp;
-//     pfs->metaCacheComp = NULL;
-//   } else {
-//     // remove meta.tmp file
-//     taosRemoveFile(mf.f.aname);
-//     taosHashCleanup(pfs->metaCacheComp);
-//     pfs->metaCacheComp = NULL;
-//   }
-
-//   taosMemoryFreeClear(pBuf);
-
-//   ASSERT(mf.info.nDels == 0);
-//   ASSERT(mf.info.tombSize == 0);
-
-//   tsdbInfo("end compact tsdb meta file,code:%d,nRecords:%" PRId64 ",size:%" PRId64,
-//     code,mf.info.nRecords,mf.info.size);
-//   return code;
-// }
-
-// // =================== Commit Time-Series Data
-// #if 0
-// static bool tsdbHasDataToCommit(SCommitIter *iters, int nIters, TSKEY minKey, TSKEY maxKey) {
-//   for (int i = 0; i < nIters; i++) {
-//     TSKEY nextKey = tsdbNextIterKey((iters + i)->pIter);
-//     if (nextKey != TSDB_DATA_TIMESTAMP_NULL && (nextKey >= minKey && nextKey <= maxKey)) return true;
-//   }
-//   return false;
-// }
-// #endif
-
 static int tsdbCommitToTable(SCommitH *pCommith, int tid) {
   SCommitIter *pIter = pCommith->iters + tid;
   TSKEY        nextKey = tsdbNextIterKey(pIter->pIter);

source/dnode/vnode/src/tsdb/tsdbCompact.c

@@ -1,533 +0,0 @@
-/*
- * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
- *
- * This program is free software: you can use, redistribute, and/or modify
- * it under the terms of the GNU Affero General Public License, version 3
- * or later ("AGPL"), as published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.
- *
- * You should have received a copy of the GNU Affero General Public License
- * along with this program. If not, see <http://www.gnu.org/licenses/>.
- */
-#if 0
-#include "tsdb.h"
-
-typedef struct {
-  STable *    pTable;
-  SBlockIdx * pBlkIdx;
-  SBlockIdx   bindex;
-  SBlockInfo *pInfo;
-} STableCompactH;
-
-typedef struct {
-  SRtn       rtn;
-  SFSIter    fsIter;
-  SArray *   tbArray;  // table array to cache table obj and block indexes
-  SReadH     readh;
-  SDFileSet  wSet;
-  SArray *   aBlkIdx;
-  SArray *   aSupBlk;
-  SDataCols *pDataCols;
-} SCompactH;
-
-#define TSDB_COMPACT_WSET(pComph)      (&((pComph)->wSet))
-#define TSDB_COMPACT_REPO(pComph)      TSDB_READ_REPO(&((pComph)->readh))
-#define TSDB_COMPACT_HEAD_FILE(pComph) TSDB_DFILE_IN_SET(TSDB_COMPACT_WSET(pComph), TSDB_FILE_HEAD)
-#define TSDB_COMPACT_DATA_FILE(pComph) TSDB_DFILE_IN_SET(TSDB_COMPACT_WSET(pComph), TSDB_FILE_DATA)
-#define TSDB_COMPACT_LAST_FILE(pComph) TSDB_DFILE_IN_SET(TSDB_COMPACT_WSET(pComph), TSDB_FILE_LAST)
-#define TSDB_COMPACT_SMAD_FILE(pComph) TSDB_DFILE_IN_SET(TSDB_COMPACT_WSET(pComph), TSDB_FILE_SMAD)
-#define TSDB_COMPACT_SMAL_FILE(pComph) TSDB_DFILE_IN_SET(TSDB_COMPACT_WSET(pComph), TSDB_FILE_SMAL)
-#define TSDB_COMPACT_BUF(pComph)       TSDB_READ_BUF(&((pComph)->readh))
-#define TSDB_COMPACT_COMP_BUF(pComph)  TSDB_READ_COMP_BUF(&((pComph)->readh))
-
-static int  tsdbAsyncCompact(STsdbRepo *pRepo);
-static void tsdbStartCompact(STsdbRepo *pRepo);
-static void tsdbEndCompact(STsdbRepo *pRepo, int eno);
-static int  tsdbCompactMeta(STsdbRepo *pRepo);
-static int  tsdbCompactTSData(STsdbRepo *pRepo);
-static int  tsdbCompactFSet(SCompactH *pComph, SDFileSet *pSet);
-static bool tsdbShouldCompact(SCompactH *pComph);
-static int  tsdbInitCompactH(SCompactH *pComph, STsdbRepo *pRepo);
-static void tsdbDestroyCompactH(SCompactH *pComph);
-static int  tsdbInitCompTbArray(SCompactH *pComph);
-static void tsdbDestroyCompTbArray(SCompactH *pComph);
-static int  tsdbCacheFSetIndex(SCompactH *pComph);
-static int  tsdbCompactFSetInit(SCompactH *pComph, SDFileSet *pSet);
-static void tsdbCompactFSetEnd(SCompactH *pComph);
-static int  tsdbCompactFSetImpl(SCompactH *pComph);
-static int  tsdbWriteBlockToRightFile(SCompactH *pComph, STable *pTable, SDataCols *pDataCols, void **ppBuf,
-                                      void **ppCBuf);
-
-enum { TSDB_NO_COMPACT, TSDB_IN_COMPACT, TSDB_WAITING_COMPACT};
-int tsdbCompact(STsdbRepo *pRepo) { return tsdbAsyncCompact(pRepo); }
-
-void *tsdbCompactImpl(STsdbRepo *pRepo) {
-  // Check if there are files in TSDB FS to compact
-  if (REPO_FS(pRepo)->cstatus->pmf == NULL) {
-    tsdbInfo("vgId:%d no file to compact in FS", REPO_ID(pRepo));
-    return NULL;
-  }
-
-  tsdbStartCompact(pRepo);
-
-  if (tsdbCompactMeta(pRepo) < 0) {
-    tsdbError("vgId:%d failed to compact META data since %s", REPO_ID(pRepo), tstrerror(terrno));
-    goto _err;
-  }
-
-  if (tsdbCompactTSData(pRepo) < 0) {
-    tsdbError("vgId:%d failed to compact TS data since %s", REPO_ID(pRepo), tstrerror(terrno));
-    goto _err;
-  }
-
-  tsdbEndCompact(pRepo, TSDB_CODE_SUCCESS);
-  return NULL;
-
-_err:
-  pRepo->code = terrno;
-  tsdbEndCompact(pRepo, terrno);
-  return NULL;
-}
-
-static int tsdbAsyncCompact(STsdbRepo *pRepo) {
-  if (pRepo->compactState != TSDB_NO_COMPACT) {
-    tsdbInfo("vgId:%d not compact tsdb again ", REPO_ID(pRepo));
-    return 0; 
-  } 
-  pRepo->compactState = TSDB_WAITING_COMPACT;   
-  tsem_wait(&(pRepo->readyToCommit));
-  return tsdbScheduleCommit(pRepo, COMPACT_REQ);
-}
-
-static void tsdbStartCompact(STsdbRepo *pRepo) {
-  assert(pRepo->compactState != TSDB_IN_COMPACT);
-  tsdbInfo("vgId:%d start to compact!", REPO_ID(pRepo));
-  tsdbStartFSTxn(pRepo, 0, 0);
-  pRepo->code = TSDB_CODE_SUCCESS;
-  pRepo->compactState = TSDB_IN_COMPACT;
-}
-
-static void tsdbEndCompact(STsdbRepo *pRepo, int eno) {
-  if (eno != TSDB_CODE_SUCCESS) {
-    tsdbEndFSTxnWithError(REPO_FS(pRepo));
-  } else {
-    tsdbEndFSTxn(pRepo);
-  }
-  pRepo->compactState = TSDB_NO_COMPACT;
-  tsdbInfo("vgId:%d compact over, %s", REPO_ID(pRepo), (eno == TSDB_CODE_SUCCESS) ? "succeed" : "failed");
-  tsem_post(&(pRepo->readyToCommit));
-}
-
-static int tsdbCompactMeta(STsdbRepo *pRepo) {
-  STsdbFS *pfs = REPO_FS(pRepo);
-  tsdbUpdateMFile(pfs, pfs->cstatus->pmf);
-  return 0;
-}
-
-  static int tsdbCompactTSData(STsdbRepo *pRepo) {
-    SCompactH  compactH;
-    SDFileSet *pSet = NULL;
-
-    tsdbDebug("vgId:%d start to compact TS data", REPO_ID(pRepo));
-
-    // If no file, just return 0;
-    if (taosArrayGetSize(REPO_FS(pRepo)->cstatus->df) <= 0) {
-      tsdbDebug("vgId:%d no TS data file to compact, compact over", REPO_ID(pRepo));
-      return 0;
-    }
-
-    if (tsdbInitCompactH(&compactH, pRepo) < 0) {
-      return -1;
-    }
-
-    while ((pSet = tsdbFSIterNext(&(compactH.fsIter)))) {
-      // Remove those expired files
-      if (pSet->fid < compactH.rtn.minFid) {
-        tsdbInfo("vgId:%d FSET %d on level %d disk id %d expires, remove it", REPO_ID(pRepo), pSet->fid,
-                TSDB_FSET_LEVEL(pSet), TSDB_FSET_ID(pSet));
-        continue;
-      }
-
-      if (TSDB_FSET_LEVEL(pSet) == TFS_MAX_LEVEL) {
-        tsdbDebug("vgId:%d FSET %d on level %d, should not compact", REPO_ID(pRepo), pSet->fid, TFS_MAX_LEVEL);
-        tsdbUpdateDFileSet(REPO_FS(pRepo), pSet);
-        continue;
-      }
-
-      if (tsdbCompactFSet(&compactH, pSet) < 0) {
-        tsdbDestroyCompactH(&compactH);
-        tsdbError("vgId:%d failed to compact FSET %d since %s", REPO_ID(pRepo), pSet->fid, tstrerror(terrno));
-        return -1;
-      }
-    }
-
-    tsdbDestroyCompactH(&compactH);
-    tsdbDebug("vgId:%d compact TS data over", REPO_ID(pRepo));
-    return 0;
-  }
-
-  static int tsdbCompactFSet(SCompactH *pComph, SDFileSet *pSet) {
-    STsdbRepo *pRepo = TSDB_COMPACT_REPO(pComph);
-    SDiskID    did;
-
-    tsdbDebug("vgId:%d start to compact FSET %d on level %d id %d", REPO_ID(pRepo), pSet->fid, TSDB_FSET_LEVEL(pSet),
-              TSDB_FSET_ID(pSet));
-
-    if (tsdbCompactFSetInit(pComph, pSet) < 0) {
-      return -1;
-    }
-
-    if (!tsdbShouldCompact(pComph)) {
-      tsdbDebug("vgId:%d no need to compact FSET %d", REPO_ID(pRepo), pSet->fid);
-      if (tsdbApplyRtnOnFSet(TSDB_COMPACT_REPO(pComph), pSet, &(pComph->rtn)) < 0) {
-        tsdbCompactFSetEnd(pComph);
-        return -1;
-      }
-    } else {
-      // Create new fset as compacted fset
-      if (tfsAllocDisk(pRepo->pTfs, tsdbGetFidLevel(pSet->fid, &(pComph->rtn)), &did) < 0) {
-        terrno = TSDB_CODE_TDB_NO_AVAIL_DISK;
-        tsdbError("vgId:%d failed to compact FSET %d since %s", REPO_ID(pRepo), pSet->fid, tstrerror(terrno));
-        tsdbCompactFSetEnd(pComph);
-        return -1;
-      }
-
-      tsdbInitDFileSet(TSDB_COMPACT_WSET(pComph), did, REPO_ID(pRepo), TSDB_FSET_FID(pSet),
-                      FS_TXN_VERSION(REPO_FS(pRepo)));
-      if (tsdbCreateDFileSet(TSDB_COMPACT_WSET(pComph), true) < 0) {
-        tsdbError("vgId:%d failed to compact FSET %d since %s", REPO_ID(pRepo), pSet->fid, tstrerror(terrno));
-        tsdbCompactFSetEnd(pComph);
-        return -1;
-      }
-
-      if (tsdbCompactFSetImpl(pComph) < 0) {
-        tsdbCloseDFileSet(TSDB_COMPACT_WSET(pComph));
-        tsdbRemoveDFileSet(TSDB_COMPACT_WSET(pComph));
-        tsdbCompactFSetEnd(pComph);
-        return -1;
-      }
-
-      tsdbCloseDFileSet(TSDB_COMPACT_WSET(pComph));
-      tsdbUpdateDFileSet(REPO_FS(pRepo), TSDB_COMPACT_WSET(pComph));
-      tsdbDebug("vgId:%d FSET %d compact over", REPO_ID(pRepo), pSet->fid);
-    }
-
-    tsdbCompactFSetEnd(pComph);
-    return 0;
-  }
-
-  static bool tsdbShouldCompact(SCompactH *pComph) {
-    STsdbRepo *     pRepo = TSDB_COMPACT_REPO(pComph);
-    STsdbCfg *      pCfg = REPO_CFG(pRepo);
-    SReadH *        pReadh = &(pComph->readh);
-    STableCompactH *pTh;
-    SBlock *        pBlock;
-    int             defaultRows = TSDB_DEFAULT_BLOCK_ROWS(pCfg->maxRowsPerFileBlock);
-    SDFile *        pDataF = TSDB_READ_DATA_FILE(pReadh);
-    SDFile *        pLastF = TSDB_READ_LAST_FILE(pReadh);
-
-    int     tblocks = 0;       // total blocks
-    int     nSubBlocks = 0;    // # of blocks with sub-blocks
-    int     nSmallBlocks = 0;  // # of blocks with rows < defaultRows
-    int64_t tsize = 0;
-
-    for (size_t i = 0; i < taosArrayGetSize(pComph->tbArray); i++) {
-      pTh = (STableCompactH *)taosArrayGet(pComph->tbArray, i);
-
-      if (pTh->pTable == NULL || pTh->pBlkIdx == NULL) continue;
-
-      for (size_t bidx = 0; bidx < pTh->pBlkIdx->numOfBlocks; bidx++) {
-        tblocks++;
-        pBlock = pTh->pInfo->blocks + bidx;
-
-        if (pBlock->numOfRows < defaultRows) {
-          nSmallBlocks++;
-        }
-
-        if (pBlock->numOfSubBlocks > 1) {
-          nSubBlocks++;
-          for (int k = 0; k < pBlock->numOfSubBlocks; k++) {
-            SBlock *iBlock = ((SBlock *)POINTER_SHIFT(pTh->pInfo, pBlock->offset)) + k;
-            tsize = tsize + iBlock->len;
-          }
-        } else if (pBlock->numOfSubBlocks == 1) {
-          tsize += pBlock->len;
-        } else {
-          ASSERT(0);
-        }
-      }
-    }
-
-    return (((nSubBlocks * 1.0 / tblocks) > 0.33) || ((nSmallBlocks * 1.0 / tblocks) > 0.33) ||
-            (tsize * 1.0 / (pDataF->info.size + pLastF->info.size - 2 * TSDB_FILE_HEAD_SIZE) < 0.85));
-  }
-
-  static int tsdbInitCompactH(SCompactH *pComph, STsdbRepo *pRepo) {
-    STsdbCfg *pCfg = REPO_CFG(pRepo);
-
-    memset(pComph, 0, sizeof(*pComph));
-
-    TSDB_FSET_SET_CLOSED(TSDB_COMPACT_WSET(pComph));
-
-    tsdbGetRtnSnap(pRepo, &(pComph->rtn));
-    tsdbFSIterInit(&(pComph->fsIter), REPO_FS(pRepo), TSDB_FS_ITER_FORWARD);
-
-    if (tsdbInitReadH(&(pComph->readh), pRepo) < 0) {
-      return -1;
-    }
-
-    if (tsdbInitCompTbArray(pComph) < 0) {
-      tsdbDestroyCompactH(pComph);
-      return -1;
-    }
-
-    pComph->aBlkIdx = taosArrayInit(1024, sizeof(SBlockIdx));
-    if (pComph->aBlkIdx == NULL) {
-      terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-      tsdbDestroyCompactH(pComph);
-      return -1;
-    }
-
-    pComph->aSupBlk = taosArrayInit(1024, sizeof(SBlock));
-    if (pComph->aSupBlk == NULL) {
-      terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-      tsdbDestroyCompactH(pComph);
-      return -1;
-    }
-
-    pComph->pDataCols = tdNewDataCols(0, pCfg->maxRowsPerFileBlock);
-    if (pComph->pDataCols == NULL) {
-      terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-      tsdbDestroyCompactH(pComph);
-      return -1;
-    }
-
-    return 0;
-  }
-
-  static void tsdbDestroyCompactH(SCompactH *pComph) {
-    pComph->pDataCols = tdFreeDataCols(pComph->pDataCols);
-    pComph->aSupBlk = taosArrayDestroy(pComph->aSupBlk);
-    pComph->aBlkIdx = taosArrayDestroy(pComph->aBlkIdx);
-    tsdbDestroyCompTbArray(pComph);
-    tsdbDestroyReadH(&(pComph->readh));
-    tsdbCloseDFileSet(TSDB_COMPACT_WSET(pComph));
-  }
-
-  static int tsdbInitCompTbArray(SCompactH *pComph) {  // Init pComp->tbArray
-    STsdbRepo *pRepo = TSDB_COMPACT_REPO(pComph);
-    STsdbMeta *pMeta = pRepo->tsdbMeta;
-
-    if (tsdbRLockRepoMeta(pRepo) < 0) return -1;
-
-    pComph->tbArray = taosArrayInit(pMeta->maxTables, sizeof(STableCompactH));
-    if (pComph->tbArray == NULL) {
-      terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-      tsdbUnlockRepoMeta(pRepo);
-      return -1;
-    }
-
-    // Note here must start from 0
-    for (int i = 0; i < pMeta->maxTables; i++) {
-      STableCompactH ch = {0};
-      if (pMeta->tables[i] != NULL) {
-        tsdbRefTable(pMeta->tables[i]);
-        ch.pTable = pMeta->tables[i];
-      }
-
-      if (taosArrayPush(pComph->tbArray, &ch) == NULL) {
-        terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-        tsdbUnlockRepoMeta(pRepo);
-        return -1;
-      }
-    }
-
-    if (tsdbUnlockRepoMeta(pRepo) < 0) return -1;
-    return 0;
-  }
-
-  static void tsdbDestroyCompTbArray(SCompactH *pComph) {
-    STableCompactH *pTh;
-
-    if (pComph->tbArray == NULL) return;
-
-    for (size_t i = 0; i < taosArrayGetSize(pComph->tbArray); i++) {
-      pTh = (STableCompactH *)taosArrayGet(pComph->tbArray, i);
-      if (pTh->pTable) {
-        tsdbUnRefTable(pTh->pTable);
-      }
-
-      pTh->pInfo = taosTZfree(pTh->pInfo);
-    }
-
-    pComph->tbArray = taosArrayDestroy(pComph->tbArray);
-  }
-
-  static int tsdbCacheFSetIndex(SCompactH *pComph) {
-    SReadH *pReadH = &(pComph->readh);
-
-    if (tsdbLoadBlockIdx(pReadH) < 0) {
-      return -1;
-    }
-
-    for (int tid = 1; tid < taosArrayGetSize(pComph->tbArray); tid++) {
-      STableCompactH *pTh = (STableCompactH *)taosArrayGet(pComph->tbArray, tid);
-      pTh->pBlkIdx = NULL;
-
-      if (pTh->pTable == NULL) continue;
-      if (tsdbSetReadTable(pReadH, pTh->pTable) < 0) {
-        return -1;
-      }
-
-      if (pReadH->pBlkIdx == NULL) continue;
-      pTh->bindex = *(pReadH->pBlkIdx);
-      pTh->pBlkIdx = &(pTh->bindex);
-
-      if (tsdbMakeRoom((void **)(&(pTh->pInfo)), pTh->pBlkIdx->len) < 0) {
-        return -1;
-      }
-
-      if (tsdbLoadBlockInfo(pReadH, (void *)(pTh->pInfo)) < 0) {
-        return -1;
-      }
-    }
-
-    return 0;
-  }
-
-  static int tsdbCompactFSetInit(SCompactH *pComph, SDFileSet *pSet) {
-    taosArrayClear(pComph->aBlkIdx);
-    taosArrayClear(pComph->aSupBlk);
-
-    if (tsdbSetAndOpenReadFSet(&(pComph->readh), pSet) < 0) {
-      return -1;
-    }
-
-    if (tsdbCacheFSetIndex(pComph) < 0) {
-      tsdbCloseAndUnsetFSet(&(pComph->readh));
-      return -1;
-    }
-
-    return 0;
-  }
-
-  static void tsdbCompactFSetEnd(SCompactH *pComph) { tsdbCloseAndUnsetFSet(&(pComph->readh)); }
-
-  static int tsdbCompactFSetImpl(SCompactH *pComph) {
-    STsdbRepo *pRepo = TSDB_COMPACT_REPO(pComph);
-    STsdbCfg * pCfg = REPO_CFG(pRepo);
-    SReadH *   pReadh = &(pComph->readh);
-    SBlockIdx  blkIdx;
-    void **    ppBuf = &(TSDB_COMPACT_BUF(pComph));
-    void **    ppCBuf = &(TSDB_COMPACT_COMP_BUF(pComph));
-    int        defaultRows = TSDB_DEFAULT_BLOCK_ROWS(pCfg->maxRowsPerFileBlock);
-
-    taosArrayClear(pComph->aBlkIdx);
-
-    for (int tid = 1; tid < taosArrayGetSize(pComph->tbArray); tid++) {
-      STableCompactH *pTh = (STableCompactH *)taosArrayGet(pComph->tbArray, tid);
-      STSchema *      pSchema;
-
-      if (pTh->pTable == NULL || pTh->pBlkIdx == NULL) continue;
-
-      pSchema = tsdbGetTableSchemaImpl(pTh->pTable, true, true, -1);
-      taosArrayClear(pComph->aSupBlk);
-      if ((tdInitDataCols(pComph->pDataCols, pSchema) < 0) || (tdInitDataCols(pReadh->pDCols[0], pSchema) < 0) ||
-          (tdInitDataCols(pReadh->pDCols[1], pSchema) < 0)) {
-        terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-        return -1;
-      }
-      tdFreeSchema(pSchema);
-
-      // Loop to compact each block data
-      for (int i = 0; i < pTh->pBlkIdx->numOfBlocks; i++) {
-        SBlock *pBlock = pTh->pInfo->blocks + i;
-
-        // Load the block data
-        if (tsdbLoadBlockData(pReadh, pBlock, pTh->pInfo) < 0) {
-          return -1;
-        }
-
-        // Merge pComph->pDataCols and pReadh->pDCols[0] and write data to file
-        if (pComph->pDataCols->numOfRows == 0 && pBlock->numOfRows >= defaultRows) {
-          if (tsdbWriteBlockToRightFile(pComph, pTh->pTable, pReadh->pDCols[0], ppBuf, ppCBuf) < 0) {
-            return -1;
-          }
-        } else {
-          int ridx = 0;
-
-          while (true) {
-            if (pReadh->pDCols[0]->numOfRows - ridx == 0) break;
-            int rowsToMerge = TMIN(pReadh->pDCols[0]->numOfRows - ridx, defaultRows - pComph->pDataCols->numOfRows);
-
-            tdMergeDataCols(pComph->pDataCols, pReadh->pDCols[0], rowsToMerge, &ridx, pCfg->update != TD_ROW_PARTIAL_UPDATE);
-
-            if (pComph->pDataCols->numOfRows < defaultRows) {
-              break;
-            }
-
-            if (tsdbWriteBlockToRightFile(pComph, pTh->pTable, pComph->pDataCols, ppBuf, ppCBuf) < 0) {
-              return -1;
-            }
-            tdResetDataCols(pComph->pDataCols);
-          }
-        }
-      }
-
-      if (pComph->pDataCols->numOfRows > 0 &&
-          tsdbWriteBlockToRightFile(pComph, pTh->pTable, pComph->pDataCols, ppBuf, ppCBuf) < 0) {
-        return -1;
-      }
-
-      if (tsdbWriteBlockInfoImpl(TSDB_COMPACT_HEAD_FILE(pComph), pTh->pTable, pComph->aSupBlk, NULL, ppBuf, &blkIdx) <
-          0) {
-        return -1;
-      }
-
-      if ((blkIdx.numOfBlocks > 0) && (taosArrayPush(pComph->aBlkIdx, (void *)(&blkIdx)) == NULL)) {
-        terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-        return -1;
-      }
-    }
-
-    if (tsdbWriteBlockIdx(TSDB_COMPACT_HEAD_FILE(pComph), pComph->aBlkIdx, ppBuf) < 0) {
-      return -1;
-    }
-
-    return 0;
-  }
-
-  static int tsdbWriteBlockToRightFile(SCompactH *pComph, STable *pTable, SDataCols *pDataCols, void **ppBuf,
-                                      void **ppCBuf) {
-    STsdbRepo *pRepo = TSDB_COMPACT_REPO(pComph);
-    STsdbCfg * pCfg = REPO_CFG(pRepo);
-    SDFile *   pDFile;
-    bool       isLast;
-    SBlock     block;
-
-    ASSERT(pDataCols->numOfRows > 0);
-
-    if (pDataCols->numOfRows < pCfg->minRowsPerFileBlock) {
-      pDFile = TSDB_COMPACT_LAST_FILE(pComph);
-      isLast = true;
-    } else {
-      pDFile = TSDB_COMPACT_DATA_FILE(pComph);
-      isLast = false;
-    }
-
-    if (tsdbWriteBlockImpl(pRepo, pTable, pDFile, pDataCols, &block, isLast, true, ppBuf, ppCBuf) < 0) {
-      return -1;
-    }
-
-    if (taosArrayPush(pComph->aSupBlk, (void *)(&block)) == NULL) {
-      terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-      return -1;
-    }
-
-    return 0;
-}
-
-#endif

source/dnode/vnode/src/tsdb/tsdbFS.c

@@ -246,62 +246,6 @@ void *tsdbFreeFS(STsdbFS *pfs) {
   return NULL;
 }
 
-// static int tsdbProcessExpiredFS(STsdb *pRepo) {
-//   tsdbStartFSTxn(pRepo, 0, 0);
-//   // if (tsdbCreateMeta(pRepo) < 0) {
-//   //   tsdbError("vgId:%d failed to create meta since %s", REPO_ID(pRepo), tstrerror(terrno));
-//   //   return -1;
-//   // }
-
-//   if (tsdbApplyRtn(pRepo) < 0) {
-//     tsdbEndFSTxnWithError(REPO_FS(pRepo));
-//     tsdbError("vgId:%d failed to apply rtn since %s", REPO_ID(pRepo), tstrerror(terrno));
-//     return -1;
-//   }
-//   if (tsdbEndFSTxn(pRepo) < 0) {
-//     tsdbError("vgId:%d failed to end fs txn since %s", REPO_ID(pRepo), tstrerror(terrno));
-//     return -1;
-//   }
-//   return 0;
-// }
-
-// static int tsdbCreateMeta(STsdb *pRepo) {
-//   STsdbFS *pfs = REPO_FS(pRepo);
-//   SMFile * pOMFile = pfs->cstatus->pmf;
-//   SMFile   mf;
-//   SDiskID  did;
-
-//   if (pOMFile != NULL) {
-//     // keep the old meta file
-//     tsdbUpdateMFile(pfs, pOMFile);
-//     return 0;
-//   }
-
-//   // Create a new meta file
-//   did.level = TFS_PRIMARY_LEVEL;
-//   did.id = TFS_PRIMARY_ID;
-//   tsdbInitMFile(&mf, did, REPO_ID(pRepo), FS_TXN_VERSION(REPO_FS(pRepo)));
-
-//   if (tsdbCreateMFile(&mf, true) < 0) {
-//     tsdbError("vgId:%d failed to create META file since %s", REPO_ID(pRepo), tstrerror(terrno));
-//     return -1;
-//   }
-
-//   tsdbInfo("vgId:%d meta file %s is created", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(&mf));
-
-//   if (tsdbUpdateMFileHeader(&mf) < 0) {
-//     tsdbError("vgId:%d failed to update META file header since %s, revert it", REPO_ID(pRepo), tstrerror(terrno));
-//     tsdbApplyMFileChange(&mf, pOMFile);
-//     return -1;
-//   }
-
-//   TSDB_FILE_FSYNC(&mf);
-//   tsdbCloseMFile(&mf);
-//   tsdbUpdateMFile(pfs, &mf);
-
-//   return 0;
-// }
-
 int tsdbOpenFS(STsdb *pRepo) {
   STsdbFS *pfs = REPO_FS(pRepo);
   char     current[TSDB_FILENAME_LEN] = "\0";
@@ -644,7 +588,7 @@ static int tsdbComparFidFSet(const void *arg1, const void *arg2) {
 }
 
 static void tsdbGetTxnFname(STsdb *pRepo, TSDB_TXN_FILE_T ftype, char fname[]) {
-  snprintf(fname, TSDB_FILENAME_LEN, "%s/vnode/vnode%d/tsdb/%s", tfsGetPrimaryPath(REPO_TFS(pRepo)), pRepo->vgId,
+  snprintf(fname, TSDB_FILENAME_LEN, "%s/vnode/vnode%d/tsdb/%s", tfsGetPrimaryPath(REPO_TFS(pRepo)), REPO_ID(pRepo),
            tsdbTxnFname[ftype]);
 }
 
@@ -769,142 +713,6 @@ static int tsdbScanAndTryFixFS(STsdb *pRepo) {
   return 0;
 }
 
-// int tsdbLoadMetaCache(STsdb *pRepo, bool recoverMeta) {
-//   char      tbuf[128];
-//   STsdbFS * pfs = REPO_FS(pRepo);
-//   SMFile    mf;
-//   SMFile *  pMFile = &mf;
-//   void *    pBuf = NULL;
-//   SKVRecord rInfo;
-//   int64_t   maxBufSize = 0;
-//   SMFInfo   minfo;
-
-//   taosHashClear(pfs->metaCache);
-
-//   // No meta file, just return
-//   if (pfs->cstatus->pmf == NULL) return 0;
-
-//   mf = pfs->cstatus->mf;
-//   // Load cache first
-//   if (tsdbOpenMFile(pMFile, O_RDONLY) < 0) {
-//     return -1;
-//   }
-
-//   if (tsdbLoadMFileHeader(pMFile, &minfo) < 0) {
-//     tsdbCloseMFile(pMFile);
-//     return -1;
-//   }
-
-//   while (true) {
-//     int64_t tsize = tsdbReadMFile(pMFile, tbuf, sizeof(SKVRecord));
-//     if (tsize == 0) break;
-
-//     if (tsize < 0) {
-//       tsdbError("vgId:%d failed to read META file since %s", REPO_ID(pRepo), tstrerror(terrno));
-//       return -1;
-//     }
-
-//     if (tsize < sizeof(SKVRecord)) {
-//       tsdbError("vgId:%d failed to read %" PRIzu " bytes from file %s", REPO_ID(pRepo), sizeof(SKVRecord),
-//                 TSDB_FILE_FULL_NAME(pMFile));
-//       terrno = TSDB_CODE_TDB_FILE_CORRUPTED;
-//       tsdbCloseMFile(pMFile);
-//       return -1;
-//     }
-
-//     void *ptr = tsdbDecodeKVRecord(tbuf, &rInfo);
-//     ASSERT(POINTER_DISTANCE(ptr, tbuf) == sizeof(SKVRecord));
-//     // ASSERT((rInfo.offset > 0) ? (pStore->info.size == rInfo.offset) : true);
-
-//     if (rInfo.offset < 0) {
-//       taosHashRemove(pfs->metaCache, (void *)(&rInfo.uid), sizeof(rInfo.uid));
-// #if 0
-//       pStore->info.size += sizeof(SKVRecord);
-//       pStore->info.nRecords--;
-//       pStore->info.nDels++;
-//       pStore->info.tombSize += (rInfo.size + sizeof(SKVRecord) * 2);
-// #endif
-//     } else {
-//       ASSERT(rInfo.offset > 0 && rInfo.size > 0);
-//       if (taosHashPut(pfs->metaCache, (void *)(&rInfo.uid), sizeof(rInfo.uid), &rInfo, sizeof(rInfo)) < 0) {
-//         tsdbError("vgId:%d failed to load meta cache from file %s since OOM", REPO_ID(pRepo),
-//                   TSDB_FILE_FULL_NAME(pMFile));
-//         terrno = TSDB_CODE_COM_OUT_OF_MEMORY;
-//         tsdbCloseMFile(pMFile);
-//         return -1;
-//       }
-
-//       maxBufSize = TMAX(maxBufSize, rInfo.size);
-
-//       if (tsdbSeekMFile(pMFile, rInfo.size, SEEK_CUR) < 0) {
-//         tsdbError("vgId:%d failed to lseek file %s since %s", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile),
-//                   tstrerror(terrno));
-//         tsdbCloseMFile(pMFile);
-//         return -1;
-//       }
-
-// #if 0
-//       pStore->info.size += (sizeof(SKVRecord) + rInfo.size);
-//       pStore->info.nRecords++;
-// #endif
-//     }
-//   }
-
-//   if (recoverMeta) {
-//     pBuf = taosMemoryMalloc((size_t)maxBufSize);
-//     if (pBuf == NULL) {
-//       terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-//       tsdbCloseMFile(pMFile);
-//       return -1;
-//     }
-
-//     SKVRecord *pRecord = taosHashIterate(pfs->metaCache, NULL);
-//     while (pRecord) {
-//       if (tsdbSeekMFile(pMFile, pRecord->offset + sizeof(SKVRecord), SEEK_SET) < 0) {
-//         tsdbError("vgId:%d failed to seek file %s since %s", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile),
-//                   tstrerror(terrno));
-//         taosMemoryFreeClear(pBuf);
-//         tsdbCloseMFile(pMFile);
-//         return -1;
-//       }
-
-//       int nread = (int)tsdbReadMFile(pMFile, pBuf, pRecord->size);
-//       if (nread < 0) {
-//         tsdbError("vgId:%d failed to read file %s since %s", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile),
-//                   tstrerror(terrno));
-//         taosMemoryFreeClear(pBuf);
-//         tsdbCloseMFile(pMFile);
-//         return -1;
-//       }
-
-//       if (nread < pRecord->size) {
-//         tsdbError("vgId:%d failed to read file %s since file corrupted, expected read:%" PRId64 " actual read:%d",
-//                   REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile), pRecord->size, nread);
-//         terrno = TSDB_CODE_TDB_FILE_CORRUPTED;
-//         taosMemoryFreeClear(pBuf);
-//         tsdbCloseMFile(pMFile);
-//         return -1;
-//       }
-
-//       if (tsdbRestoreTable(pRepo, pBuf, (int)pRecord->size) < 0) {
-//         tsdbError("vgId:%d failed to restore table, uid %" PRId64 ", since %s" PRIu64, REPO_ID(pRepo), pRecord->uid,
-//                   tstrerror(terrno));
-//         taosMemoryFreeClear(pBuf);
-//         tsdbCloseMFile(pMFile);
-//         return -1;
-//       }
-
-//       pRecord = taosHashIterate(pfs->metaCache, pRecord);
-//     }
-
-//     tsdbOrgMeta(pRepo);
-//   }
-
-//   tsdbCloseMFile(pMFile);
-//   taosMemoryFreeClear(pBuf);
-//   return 0;
-// }
-
 static int tsdbScanRootDir(STsdb *pRepo) {
   char            rootDir[TSDB_FILENAME_LEN];
   char            bname[TSDB_FILENAME_LEN];
@@ -983,127 +791,6 @@ static bool tsdbIsTFileInFS(STsdbFS *pfs, const STfsFile *pf) {
   return false;
 }
 
-// static int tsdbRestoreMeta(STsdb *pRepo) {
-//   char         rootDir[TSDB_FILENAME_LEN];
-//   char         bname[TSDB_FILENAME_LEN];
-//   STfsDir *       tdir = NULL;
-//   const STfsFile *pf = NULL;
-//   const char * pattern = "^meta(-ver[0-9]+)?$";
-//   regex_t      regex;
-//   STsdbFS *    pfs = REPO_FS(pRepo);
-
-//   regcomp(&regex, pattern, REG_EXTENDED);
-
-//   tsdbInfo("vgId:%d try to restore meta", REPO_ID(pRepo));
-
-//   tsdbGetRootDir(REPO_ID(pRepo), rootDir);
-
-//   tdir = tfsOpendir(rootDir);
-//   if (tdir == NULL) {
-//     tsdbError("vgId:%d failed to open dir %s since %s", REPO_ID(pRepo), rootDir, tstrerror(terrno));
-//     regfree(&regex);
-//     return -1;
-//   }
-
-//   while ((pf = tfsReaddir(tdir))) {
-//     tfsBasename(pf, bname);
-
-//     if (strcmp(bname, "data") == 0) {
-//       // Skip the data/ directory
-//       continue;
-//     }
-
-//     if (strcmp(bname, tsdbTxnFname[TSDB_TXN_TEMP_FILE]) == 0) {
-//       // Skip current.t file
-//       tsdbInfo("vgId:%d file %s exists, remove it", REPO_ID(pRepo), pf->aname);
-//       (void)tfsremove(pf);
-//       continue;
-//     }
-
-//     int code = regexec(&regex, bname, 0, NULL, 0);
-//     if (code == 0) {
-//       // Match
-//       if (pfs->cstatus->pmf != NULL) {
-//         tsdbError("vgId:%d failed to restore meta since two file exists, file1 %s and file2 %s", REPO_ID(pRepo),
-//                   TSDB_FILE_FULL_NAME(pfs->cstatus->pmf), pf->aname);
-//         terrno = TSDB_CODE_TDB_FILE_CORRUPTED;
-//         tfsClosedir(tdir);
-//         regfree(&regex);
-//         return -1;
-//       } else {
-//         uint32_t _version = 0;
-//         if (strcmp(bname, "meta") != 0) {
-//           sscanf(bname, "meta-ver%" PRIu32, &_version);
-//           pfs->cstatus->meta.version = _version;
-//         }
-
-//         pfs->cstatus->pmf = &(pfs->cstatus->mf);
-//         pfs->cstatus->pmf->f = *pf;
-//         TSDB_FILE_SET_CLOSED(pfs->cstatus->pmf);
-
-//         if (tsdbOpenMFile(pfs->cstatus->pmf, O_RDONLY) < 0) {
-//           tsdbError("vgId:%d failed to restore meta since %s", REPO_ID(pRepo), tstrerror(terrno));
-//           tfsClosedir(tdir);
-//           regfree(&regex);
-//           return -1;
-//         }
-
-//         if (tsdbLoadMFileHeader(pfs->cstatus->pmf, &(pfs->cstatus->pmf->info)) < 0) {
-//           tsdbError("vgId:%d failed to restore meta since %s", REPO_ID(pRepo), tstrerror(terrno));
-//           tsdbCloseMFile(pfs->cstatus->pmf);
-//           tfsClosedir(tdir);
-//           regfree(&regex);
-//           return -1;
-//         }
-
-//         if (tsdbForceKeepFile) {
-//           struct stat tfstat;
-
-//           // Get real file size
-//           if (fstat(pfs->cstatus->pmf->fd, &tfstat) < 0) {
-//             terrno = TAOS_SYSTEM_ERROR(errno);
-//             tsdbCloseMFile(pfs->cstatus->pmf);
-//             tfsClosedir(tdir);
-//             regfree(&regex);
-//             return -1;
-//           }
-
-//           if (pfs->cstatus->pmf->info.size != tfstat.st_size) {
-//             int64_t tfsize = pfs->cstatus->pmf->info.size;
-//             pfs->cstatus->pmf->info.size = tfstat.st_size;
-//             tsdbInfo("vgId:%d file %s header size is changed from %" PRId64 " to %" PRId64, REPO_ID(pRepo),
-//                      TSDB_FILE_FULL_NAME(pfs->cstatus->pmf), tfsize, pfs->cstatus->pmf->info.size);
-//           }
-//         }
-
-//         tsdbCloseMFile(pfs->cstatus->pmf);
-//       }
-//     } else if (code == REG_NOMATCH) {
-//       // Not match
-//       tsdbInfo("vgId:%d invalid file %s exists, remove it", REPO_ID(pRepo), pf->aname);
-//       tfsremove(pf);
-//       continue;
-//     } else {
-//       // Has other error
-//       tsdbError("vgId:%d failed to restore meta file while run regexec since %s", REPO_ID(pRepo), strerror(code));
-//       terrno = TAOS_SYSTEM_ERROR(code);
-//       tfsClosedir(tdir);
-//       regfree(&regex);
-//       return -1;
-//     }
-//   }
-
-//   if (pfs->cstatus->pmf) {
-//     tsdbInfo("vgId:%d meta file %s is restored", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pfs->cstatus->pmf));
-//   } else {
-//     tsdbInfo("vgId:%d no meta file is restored", REPO_ID(pRepo));
-//   }
-
-//   tfsClosedir(tdir);
-//   regfree(&regex);
-//   return 0;
-// }
-
 static int tsdbRestoreDFileSet(STsdb *pRepo) {
   char            dataDir[TSDB_FILENAME_LEN];
   char            bname[TSDB_FILENAME_LEN];

source/dnode/vnode/src/tsdb/tsdbFile.c

@@ -33,271 +33,6 @@ static int   tsdbEncodeDFInfo(void **buf, SDFInfo *pInfo);
 static void *tsdbDecodeDFInfo(void *buf, SDFInfo *pInfo);
 static int   tsdbRollBackDFile(SDFile *pDFile);
 
-#if 0
-// ============== SMFile
-void tsdbInitMFile(SMFile *pMFile, SDiskID did, int vid, uint32_t ver) {
-  char fname[TSDB_FILENAME_LEN];
-
-  TSDB_FILE_SET_STATE(pMFile, TSDB_FILE_STATE_OK);
-
-  memset(&(pMFile->info), 0, sizeof(pMFile->info));
-  pMFile->info.magic = TSDB_FILE_INIT_MAGIC;
-
-  tsdbGetFilename(vid, 0, ver, TSDB_FILE_META, fname);
-  tfsInitFile(TSDB_FILE_F(pMFile), did.level, did.id, fname);
-}
-
-void tsdbInitMFileEx(SMFile *pMFile, const SMFile *pOMFile) {
-  *pMFile = *pOMFile;
-  TSDB_FILE_SET_CLOSED(pMFile);
-}
-
-int tsdbEncodeSMFile(void **buf, SMFile *pMFile) {
-  int tlen = 0;
-
-  tlen += tsdbEncodeMFInfo(buf, &(pMFile->info));
-  tlen += tfsEncodeFile(buf, &(pMFile->f));
-
-  return tlen;
-}
-
-void *tsdbDecodeSMFile(void *buf, SMFile *pMFile) {
-  buf = tsdbDecodeMFInfo(buf, &(pMFile->info));
-  buf = tfsDecodeFile(buf, &(pMFile->f));
-  TSDB_FILE_SET_CLOSED(pMFile);
-
-  return buf;
-}
-
-int tsdbEncodeSMFileEx(void **buf, SMFile *pMFile) {
-  int tlen = 0;
-
-  tlen += tsdbEncodeMFInfo(buf, &(pMFile->info));
-  tlen += taosEncodeString(buf, TSDB_FILE_FULL_NAME(pMFile));
-
-  return tlen;
-}
-
-void *tsdbDecodeSMFileEx(void *buf, SMFile *pMFile) {
-  char *aname;
-  buf = tsdbDecodeMFInfo(buf, &(pMFile->info));
-  buf = taosDecodeString(buf, &aname);
-  strncpy(TSDB_FILE_FULL_NAME(pMFile), aname, TSDB_FILENAME_LEN);
-  TSDB_FILE_SET_CLOSED(pMFile);
-
-  taosMemoryFreeClear(aname);
-
-  return buf;
-}
-
-int tsdbApplyMFileChange(SMFile *from, SMFile *to) {
-  if (from == NULL && to == NULL) return 0;
-
-  if (from != NULL) {
-    if (to == NULL) {
-      return tsdbRemoveMFile(from);
-    } else {
-      if (tfsIsSameFile(TSDB_FILE_F(from), TSDB_FILE_F(to))) {
-        if (from->info.size > to->info.size) {
-          tsdbRollBackMFile(to);
-        }
-      } else {
-        return tsdbRemoveMFile(from);
-      }
-    }
-  }
-
-  return 0;
-}
-
-int tsdbCreateMFile(SMFile *pMFile, bool updateHeader) {
-  ASSERT(pMFile->info.size == 0 && pMFile->info.magic == TSDB_FILE_INIT_MAGIC);
-
-  pMFile->fd = open(TSDB_FILE_FULL_NAME(pMFile), O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0755);
-  if (pMFile->fd < 0) {
-    if (errno == ENOENT) {
-      // Try to create directory recursively
-      char *s = strdup(TFILE_REL_NAME(&(pMFile->f)));
-      if (tfsMkdirRecurAt(dirname(s), TSDB_FILE_LEVEL(pMFile), TSDB_FILE_ID(pMFile)) < 0) {
-        taosMemoryFreeClear(s);
-        return -1;
-      }
-      taosMemoryFreeClear(s);
-
-      pMFile->fd = open(TSDB_FILE_FULL_NAME(pMFile), O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0755);
-      if (pMFile->fd < 0) {
-        terrno = TAOS_SYSTEM_ERROR(errno);
-        return -1;
-      }
-    } else {
-      terrno = TAOS_SYSTEM_ERROR(errno);
-      return -1;
-    }
-  }
-
-  if (!updateHeader) {
-    return 0;
-  }
-
-  pMFile->info.size += TSDB_FILE_HEAD_SIZE;
-
-  if (tsdbUpdateMFileHeader(pMFile) < 0) {
-    tsdbCloseMFile(pMFile);
-    tsdbRemoveMFile(pMFile);
-    return -1;
-  }
-
-  return 0;
-}
-
-int tsdbUpdateMFileHeader(SMFile *pMFile) {
-  char buf[TSDB_FILE_HEAD_SIZE] = "\0";
-
-  if (tsdbSeekMFile(pMFile, 0, SEEK_SET) < 0) {
-    return -1;
-  }
-
-  void *ptr = buf;
-  tsdbEncodeMFInfo(&ptr, TSDB_FILE_INFO(pMFile));
-
-  taosCalcChecksumAppend(0, (uint8_t *)buf, TSDB_FILE_HEAD_SIZE);
-  if (tsdbWriteMFile(pMFile, buf, TSDB_FILE_HEAD_SIZE) < 0) {
-    return -1;
-  }
-
-  return 0;
-}
-
-int tsdbLoadMFileHeader(SMFile *pMFile, SMFInfo *pInfo) {
-  char buf[TSDB_FILE_HEAD_SIZE] = "\0";
-
-  ASSERT(TSDB_FILE_OPENED(pMFile));
-
-  if (tsdbSeekMFile(pMFile, 0, SEEK_SET) < 0) {
-    return -1;
-  }
-
-  if (tsdbReadMFile(pMFile, buf, TSDB_FILE_HEAD_SIZE) < 0) {
-    return -1;
-  }
-
-  if (!taosCheckChecksumWhole((uint8_t *)buf, TSDB_FILE_HEAD_SIZE)) {
-    terrno = TSDB_CODE_TDB_FILE_CORRUPTED;
-    return -1;
-  }
-
-  tsdbDecodeMFInfo(buf, pInfo);
-  return 0;
-}
-
-int tsdbScanAndTryFixMFile(STsdb *pRepo) {
-  SMFile *    pMFile = pRepo->fs->cstatus->pmf;
-  struct stat mfstat;
-  SMFile      mf;
-
-  if (pMFile == NULL) {
-    // No meta file, no need to scan
-    return 0;
-  }
-
-  tsdbInitMFileEx(&mf, pMFile);
-
-  if (access(TSDB_FILE_FULL_NAME(pMFile), F_OK) != 0) {
-    tsdbError("vgId:%d meta file %s not exit, report to upper layer to fix it", REPO_ID(pRepo),
-              TSDB_FILE_FULL_NAME(pMFile));
-    pRepo->state |= TSDB_STATE_BAD_META;
-    TSDB_FILE_SET_STATE(pMFile, TSDB_FILE_STATE_BAD);
-    return 0;
-  }
-
-  if (stat(TSDB_FILE_FULL_NAME(&mf), &mfstat) < 0) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    return -1;
-  }
-
-  if (pMFile->info.size < mfstat.st_size) {
-    if (tsdbOpenMFile(&mf, O_WRONLY) < 0) {
-      return -1;
-    }
-
-    if (taosFtruncate(mf.fd, mf.info.size) < 0) {
-      terrno = TAOS_SYSTEM_ERROR(errno);
-      tsdbCloseMFile(&mf);
-      return -1;
-    }
-
-    if (tsdbUpdateMFileHeader(&mf) < 0) {
-      tsdbCloseMFile(&mf);
-      return -1;
-    }
-
-    tsdbCloseMFile(&mf);
-    tsdbInfo("vgId:%d file %s is truncated from %" PRId64 " to %" PRId64, REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile),
-             mfstat.st_size, pMFile->info.size);
-  } else if (pMFile->info.size > mfstat.st_size) {
-    tsdbError("vgId:%d meta file %s has wrong size %" PRId64 " expected %" PRId64 ", report to upper layer to fix it",
-              REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile), mfstat.st_size, pMFile->info.size);
-    pRepo->state |= TSDB_STATE_BAD_META;
-    TSDB_FILE_SET_STATE(pMFile, TSDB_FILE_STATE_BAD);
-    terrno = TSDB_CODE_TDB_FILE_CORRUPTED;
-    return 0;
-  } else {
-    tsdbDebug("vgId:%d meta file %s passes the scan", REPO_ID(pRepo), TSDB_FILE_FULL_NAME(pMFile));
-  }
-
-  return 0;
-}
-
-int tsdbEncodeMFInfo(void **buf, SMFInfo *pInfo) {
-  int tlen = 0;
-
-  tlen += taosEncodeVariantI64(buf, pInfo->size);
-  tlen += taosEncodeVariantI64(buf, pInfo->tombSize);
-  tlen += taosEncodeVariantI64(buf, pInfo->nRecords);
-  tlen += taosEncodeVariantI64(buf, pInfo->nDels);
-  tlen += taosEncodeFixedU32(buf, pInfo->magic);
-
-  return tlen;
-}
-
-void *tsdbDecodeMFInfo(void *buf, SMFInfo *pInfo) {
-  buf = taosDecodeVariantI64(buf, &(pInfo->size));
-  buf = taosDecodeVariantI64(buf, &(pInfo->tombSize));
-  buf = taosDecodeVariantI64(buf, &(pInfo->nRecords));
-  buf = taosDecodeVariantI64(buf, &(pInfo->nDels));
-  buf = taosDecodeFixedU32(buf, &(pInfo->magic));
-
-  return buf;
-}
-
-static int tsdbRollBackMFile(SMFile *pMFile) {
-  SMFile mf;
-
-  tsdbInitMFileEx(&mf, pMFile);
-
-  if (tsdbOpenMFile(&mf, O_WRONLY) < 0) {
-    return -1;
-  }
-
-  if (taosFtruncate(TSDB_FILE_FD(&mf), pMFile->info.size) < 0) {
-    terrno = TAOS_SYSTEM_ERROR(errno);
-    tsdbCloseMFile(&mf);
-    return -1;
-  }
-
-  if (tsdbUpdateMFileHeader(&mf) < 0) {
-    tsdbCloseMFile(&mf);
-    return -1;
-  }
-
-  TSDB_FILE_FSYNC(&mf);
-
-  tsdbCloseMFile(&mf);
-  return 0;
-}
-
-#endif
-
 // ============== Operations on SDFile
 void tsdbInitDFile(STsdb *pRepo, SDFile *pDFile, SDiskID did, int fid, uint32_t ver, TSDB_FILE_T ftype) {
   char fname[TSDB_FILENAME_LEN];
@@ -310,7 +45,7 @@ void tsdbInitDFile(STsdb *pRepo, SDFile *pDFile, SDiskID did, int fid, uint32_t
   pDFile->info.magic = TSDB_FILE_INIT_MAGIC;
   pDFile->info.fver = tsdbGetDFSVersion(ftype);
 
-  tsdbGetFilename(pRepo->vgId, fid, ver, ftype, fname);
+  tsdbGetFilename(REPO_ID(pRepo), fid, ver, ftype, fname);
   tfsInitFile(REPO_TFS(pRepo), &(pDFile->f), did, fname);
 }
 

source/dnode/vnode/src/tsdb/tsdbMemTable.c

@@ -285,21 +285,6 @@ static STbData *tsdbNewTbData(tb_uid_t uid) {
   pTbData->keyMax = TSKEY_MIN;
   pTbData->nrows = 0;
 
-  // uint8_t skipListCreateFlags;
-  // if (pCfg->update == TD_ROW_DISCARD_UPDATE)
-  //   skipListCreateFlags = SL_DISCARD_DUP_KEY;
-  // else
-  //   skipListCreateFlags = SL_UPDATE_DUP_KEY;
-
-  // pTableData->pData =
-  //     tSkipListCreate(TSDB_DATA_SKIPLIST_LEVEL, TSDB_DATA_TYPE_TIMESTAMP, TYPE_BYTES[TSDB_DATA_TYPE_TIMESTAMP],
-  //                     tkeyComparFn, skipListCreateFlags, tsdbGetTsTupleKey);
-  // if (pTableData->pData == NULL) {
-  //   terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-  //   taosMemoryFree(pTableData);
-  //   return NULL;
-  // }
-
   pTbData->pData = tSkipListCreate(5, TSDB_DATA_TYPE_TIMESTAMP, sizeof(int64_t), tkeyComparFn, SL_DISCARD_DUP_KEY,
                                    tsdbGetTsTupleKey);
   if (pTbData->pData == NULL) {
@@ -351,608 +336,3 @@ static int tsdbAppendTableRowToCols(STable *pTable, SDataCols *pCols, STSchema *
 
   return 0;
 }
-
-/* ------------------------ REFACTORING ------------------------ */
-#if 0
-int tsdbInsertDataToMemTable(STsdbMemTable *pMemTable, SSubmitReq *pMsg) {
-  SMemAllocator *pMA = pMemTable->pMA;
-  STbData *      pTbData = (STbData *)TD_MA_MALLOC(pMA, sizeof(*pTbData));
-  if (pTbData == NULL) {
-    // TODO
-  }
-
-  TD_SLIST_PUSH(&(pMemTable->list), pTbData);
-
-  return 0;
-}
-
-#include "tdataformat.h"
-#include "tfunctional.h"
-#include "tsdbRowMergeBuf.h"
-#include "tsdbint.h"
-#include "tskiplist.h"
-
-#define TSDB_DATA_SKIPLIST_LEVEL 5
-#define TSDB_MAX_INSERT_BATCH    512
-
-typedef struct {
-  int32_t  totalLen;
-  int32_t  len;
-  STSRow*  row;
-} SSubmitBlkIter;
-
-typedef struct {
-  int32_t totalLen;
-  int32_t len;
-  void *  pMsg;
-} SSubmitMsgIter;
-
-static SMemTable *  tsdbNewMemTable(STsdbRepo *pRepo);
-static void         tsdbFreeMemTable(SMemTable *pMemTable);
-static STableData*  tsdbNewTableData(STsdbCfg *pCfg, STable *pTable);
-static void         tsdbFreeTableData(STableData *pTableData);
-static int          tsdbAdjustMemMaxTables(SMemTable *pMemTable, int maxTables);
-static int          tsdbAppendTableRowToCols(STable *pTable, SDataCols *pCols, STSchema **ppSchema, STSRow* row);
-static int          tsdbInitSubmitBlkIter(SSubmitBlk *pBlock, SSubmitBlkIter *pIter);
-static STSRow*      tsdbGetSubmitBlkNext(SSubmitBlkIter *pIter);
-static int          tsdbInsertDataToTable(STsdbRepo *pRepo, SSubmitBlk *pBlock, int32_t *affectedrows);
-static int          tsdbInitSubmitMsgIter(SSubmitReq *pMsg, SSubmitMsgIter *pIter);
-static int          tsdbGetSubmitMsgNext(SSubmitMsgIter *pIter, SSubmitBlk **pPBlock);
-static int          tsdbCheckTableSchema(STsdbRepo *pRepo, SSubmitBlk *pBlock, STable *pTable);
-static int          tsdbUpdateTableLatestInfo(STsdbRepo *pRepo, STable *pTable, STSRow* row);
-
-static FORCE_INLINE int tsdbCheckRowRange(STsdbRepo *pRepo, STable *pTable, STSRow* row, TSKEY minKey, TSKEY maxKey,
-                                          TSKEY now);
-
-
-// ---------------- INTERNAL FUNCTIONS ----------------
-int tsdbRefMemTable(STsdbRepo *pRepo, SMemTable *pMemTable) {
-  if (pMemTable == NULL) return 0;
-  int ref = T_REF_INC(pMemTable);
-	tsdbDebug("vgId:%d ref memtable %p ref %d", REPO_ID(pRepo), pMemTable, ref);
-  return 0;
-}
-
-// Need to lock the repository
-int tsdbUnRefMemTable(STsdbRepo *pRepo, SMemTable *pMemTable) {
-  if (pMemTable == NULL) return 0;
-
-	int ref = T_REF_DEC(pMemTable);
-	tsdbDebug("vgId:%d unref memtable %p ref %d", REPO_ID(pRepo), pMemTable, ref);
-  if (ref == 0) {
-    STsdbBufPool *pBufPool = pRepo->pPool;
-
-    SListNode *pNode = NULL;
-    bool addNew = false;
-    if (tsdbLockRepo(pRepo) < 0) return -1;
-    while ((pNode = tdListPopHead(pMemTable->bufBlockList)) != NULL) {
-      if (pBufPool->nRecycleBlocks > 0) {
-        tsdbRecycleBufferBlock(pBufPool, pNode, false);
-        pBufPool->nRecycleBlocks -= 1;
-      } else {
-        if(pBufPool->nElasticBlocks > 0 && listNEles(pBufPool->bufBlockList) > 2) {
-          tsdbRecycleBufferBlock(pBufPool, pNode, true);
-        } else {
-          tdListAppendNode(pBufPool->bufBlockList, pNode);
-          addNew = true;
-        }
-      }      
-    }
-    if (addNew) {
-      int code = taosThreadCondSignal(&pBufPool->poolNotEmpty);
-      if (code != 0) {
-        if (tsdbUnlockRepo(pRepo) < 0) return -1;
-        tsdbError("vgId:%d failed to signal pool not empty since %s", REPO_ID(pRepo), strerror(code));
-        terrno = TAOS_SYSTEM_ERROR(code);
-        return -1;
-      }
-    }
-
-    if (tsdbUnlockRepo(pRepo) < 0) return -1;
-
-    for (int i = 0; i < pMemTable->maxTables; i++) {
-      if (pMemTable->tData[i] != NULL) {
-        tsdbFreeTableData(pMemTable->tData[i]);
-      }
-    }
-
-    tdListDiscard(pMemTable->actList);
-    tdListDiscard(pMemTable->bufBlockList);
-    tsdbFreeMemTable(pMemTable);
-  }
-  return 0;
-}
-
-int tsdbTakeMemSnapshot(STsdbRepo *pRepo, SMemSnapshot *pSnapshot, SArray *pATable) {
-  memset(pSnapshot, 0, sizeof(*pSnapshot));
-
-  if (tsdbLockRepo(pRepo) < 0) return -1;
-
-  pSnapshot->omem = pRepo->mem;
-  pSnapshot->imem = pRepo->imem;
-  tsdbRefMemTable(pRepo, pRepo->mem);
-  tsdbRefMemTable(pRepo, pRepo->imem);
-
-  if (tsdbUnlockRepo(pRepo) < 0) return -1;
-
-  if (pSnapshot->omem) {
-    taosRLockLatch(&(pSnapshot->omem->latch));
-
-    pSnapshot->mem = &(pSnapshot->mtable);
-
-    pSnapshot->mem->tData = (STableData **)taosMemoryCalloc(pSnapshot->omem->maxTables, sizeof(STableData *));
-    if (pSnapshot->mem->tData == NULL) {
-      terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-      taosRUnLockLatch(&(pSnapshot->omem->latch));
-      tsdbUnRefMemTable(pRepo, pSnapshot->omem);
-      tsdbUnRefMemTable(pRepo, pSnapshot->imem);
-      pSnapshot->mem = NULL;
-      pSnapshot->imem = NULL;
-      pSnapshot->omem = NULL;
-      return -1;
-    }
-
-    pSnapshot->mem->keyFirst = pSnapshot->omem->keyFirst;
-    pSnapshot->mem->keyLast = pSnapshot->omem->keyLast;
-    pSnapshot->mem->numOfRows = pSnapshot->omem->numOfRows;
-    pSnapshot->mem->maxTables = pSnapshot->omem->maxTables;
-
-    for (size_t i = 0; i < taosArrayGetSize(pATable); i++) {
-      STable *    pTable = *(STable **)taosArrayGet(pATable, i);
-      int32_t     tid = TABLE_TID(pTable);
-      STableData *pTableData = (tid < pSnapshot->omem->maxTables) ? pSnapshot->omem->tData[tid] : NULL;
-
-      if ((pTableData == NULL) || (TABLE_UID(pTable) != pTableData->uid)) continue;
-
-      pSnapshot->mem->tData[tid] = pTableData;
-      T_REF_INC(pTableData);
-    }
-
-    taosRUnLockLatch(&(pSnapshot->omem->latch));
-  }
-
-  tsdbDebug("vgId:%d take memory snapshot, pMem %p pIMem %p", REPO_ID(pRepo), pSnapshot->omem, pSnapshot->imem);
-  return 0;
-}
-
-void tsdbUnTakeMemSnapShot(STsdbRepo *pRepo, SMemSnapshot *pSnapshot) {
-  tsdbDebug("vgId:%d untake memory snapshot, pMem %p pIMem %p", REPO_ID(pRepo), pSnapshot->omem, pSnapshot->imem);
-
-  if (pSnapshot->mem) {
-    ASSERT(pSnapshot->omem != NULL);
-
-    for (size_t i = 0; i < pSnapshot->mem->maxTables; i++) {
-      STableData *pTableData = pSnapshot->mem->tData[i];
-      if (pTableData) {
-        tsdbFreeTableData(pTableData);
-      }
-    }
-    taosMemoryFreeClear(pSnapshot->mem->tData);
-
-    tsdbUnRefMemTable(pRepo, pSnapshot->omem);
-  }
-
-  tsdbUnRefMemTable(pRepo, pSnapshot->imem);
-
-  pSnapshot->mem = NULL;
-  pSnapshot->imem = NULL;
-  pSnapshot->omem = NULL;
-}
-
-int tsdbSyncCommitConfig(STsdbRepo* pRepo) {
-  ASSERT(pRepo->config_changed == true);
-  tsem_wait(&(pRepo->readyToCommit));
-
-  if (pRepo->code != TSDB_CODE_SUCCESS) {
-    tsdbWarn("vgId:%d try to commit config when TSDB not in good state: %s", REPO_ID(pRepo), tstrerror(terrno));
-  }
-
-  if (tsdbLockRepo(pRepo) < 0) return -1;
-  tsdbScheduleCommit(pRepo, COMMIT_CONFIG_REQ);
-  if (tsdbUnlockRepo(pRepo) < 0) return -1;
-
-  tsem_wait(&(pRepo->readyToCommit));
-  tsem_post(&(pRepo->readyToCommit));
-
-  if (pRepo->code != TSDB_CODE_SUCCESS) {
-    terrno = pRepo->code;
-    return -1;
-  }
-
-  terrno = TSDB_CODE_SUCCESS;
-  return 0;
-}
-
-/**
- * This is an important function to load data or try to load data from memory skiplist iterator.
- * 
- * This function load memory data until:
- * 1. iterator ends
- * 2. data key exceeds maxKey
- * 3. rowsIncreased = rowsInserted - rowsDeleteSucceed >= maxRowsToRead
- * 4. operations in pCols not exceeds its max capacity if pCols is given
- * 
- * The function tries to procceed AS MUCH AS POSSIBLE.
- */
-int tsdbLoadDataFromCache(STable *pTable, SSkipListIterator *pIter, TSKEY maxKey, int maxRowsToRead, SDataCols *pCols,
-                          TKEY *filterKeys, int nFilterKeys, bool keepDup, SMergeInfo *pMergeInfo) {
-  ASSERT(maxRowsToRead > 0 && nFilterKeys >= 0);
-  if (pIter == NULL) return 0;
-  STSchema * pSchema = NULL;
-  TSKEY      rowKey = 0;
-  TSKEY      fKey = 0;
-  bool       isRowDel = false;
-  int        filterIter = 0;
-  STSRow*    row = NULL;
-  SMergeInfo mInfo;
-
-  if (pMergeInfo == NULL) pMergeInfo = &mInfo;
-
-  memset(pMergeInfo, 0, sizeof(*pMergeInfo));
-  pMergeInfo->keyFirst = INT64_MAX;
-  pMergeInfo->keyLast = INT64_MIN;
-  if (pCols) tdResetDataCols(pCols);
-
-  row = tsdbNextIterRow(pIter);
-  if (row == NULL || TD_ROW_KEY(row) > maxKey) {
-    rowKey = INT64_MAX;
-    isRowDel = false;
-  } else {
-    rowKey = TD_ROW_KEY(row);
-    isRowDel = memRowDeleted(row);
-  }
-
-  if (filterIter >= nFilterKeys) {
-    fKey = INT64_MAX;
-  } else {
-    fKey = tdGetKey(filterKeys[filterIter]);
-  }
-
-  while (true) {
-    if (fKey == INT64_MAX && rowKey == INT64_MAX) break;
-
-    if (fKey < rowKey) {
-      pMergeInfo->keyFirst = TMIN(pMergeInfo->keyFirst, fKey);
-      pMergeInfo->keyLast = TMAX(pMergeInfo->keyLast, fKey);
-
-      filterIter++;
-      if (filterIter >= nFilterKeys) {
-        fKey = INT64_MAX;
-      } else {
-        fKey = tdGetKey(filterKeys[filterIter]);
-      }
-    } else if (fKey > rowKey) {
-      if (isRowDel) {
-        pMergeInfo->rowsDeleteFailed++;
-      } else {
-        if (pMergeInfo->rowsInserted - pMergeInfo->rowsDeleteSucceed >= maxRowsToRead) break;
-        if (pCols && pMergeInfo->nOperations >= pCols->maxPoints) break;
-        pMergeInfo->rowsInserted++;
-        pMergeInfo->nOperations++;
-        pMergeInfo->keyFirst = TMIN(pMergeInfo->keyFirst, rowKey);
-        pMergeInfo->keyLast = TMAX(pMergeInfo->keyLast, rowKey);
-        tsdbAppendTableRowToCols(pTable, pCols, &pSchema, row);
-      }
-
-      tSkipListIterNext(pIter);
-      row = tsdbNextIterRow(pIter);
-      if (row == NULL || TD_ROW_KEY(row) > maxKey) {
-        rowKey = INT64_MAX;
-        isRowDel = false;
-      } else {
-        rowKey = TD_ROW_KEY(row);
-        isRowDel = memRowDeleted(row);
-      }
-    } else {
-      if (isRowDel) {
-        ASSERT(!keepDup);
-        if (pCols && pMergeInfo->nOperations >= pCols->maxPoints) break;
-        pMergeInfo->rowsDeleteSucceed++;
-        pMergeInfo->nOperations++;
-        tsdbAppendTableRowToCols(pTable, pCols, &pSchema, row);
-      } else {
-        if (keepDup) {
-          if (pCols && pMergeInfo->nOperations >= pCols->maxPoints) break;
-          pMergeInfo->rowsUpdated++;
-          pMergeInfo->nOperations++;
-          pMergeInfo->keyFirst = TMIN(pMergeInfo->keyFirst, rowKey);
-          pMergeInfo->keyLast = TMAX(pMergeInfo->keyLast, rowKey);
-          tsdbAppendTableRowToCols(pTable, pCols, &pSchema, row);
-        } else {
-          pMergeInfo->keyFirst = TMIN(pMergeInfo->keyFirst, fKey);
-          pMergeInfo->keyLast = TMAX(pMergeInfo->keyLast, fKey);
-        }
-      }
-
-      tSkipListIterNext(pIter);
-      row = tsdbNextIterRow(pIter);
-      if (row == NULL || TD_ROW_KEY(row) > maxKey) {
-        rowKey = INT64_MAX;
-        isRowDel = false;
-      } else {
-        rowKey = TD_ROW_KEY(row);
-        isRowDel = memRowDeleted(row);
-      }
-
-      filterIter++;
-      if (filterIter >= nFilterKeys) {
-        fKey = INT64_MAX;
-      } else {
-        fKey = tdGetKey(filterKeys[filterIter]);
-      }
-    }
-  }
-
-  return 0;
-}
-
-// ---------------- LOCAL FUNCTIONS ----------------
-
-static FORCE_INLINE int tsdbCheckRowRange(STsdbRepo *pRepo, STable *pTable, STSRow* row, TSKEY minKey, TSKEY maxKey,
-                                          TSKEY now) {
-  TSKEY rowKey = TD_ROW_KEY(row);
-  if (rowKey < minKey || rowKey > maxKey) {
-    tsdbError("vgId:%d table %s tid %d uid %" PRIu64 " timestamp is out of range! now %" PRId64 " minKey %" PRId64
-              " maxKey %" PRId64 " row key %" PRId64,
-              REPO_ID(pRepo), TABLE_CHAR_NAME(pTable), TABLE_TID(pTable), TABLE_UID(pTable), now, minKey, maxKey,
-              rowKey);
-    terrno = TSDB_CODE_TDB_TIMESTAMP_OUT_OF_RANGE;
-    return -1;
-  }
-
-  return 0;
-}
-
-
-//row1 has higher priority
-static STSRow* tsdbInsertDupKeyMerge(STSRow* row1, STSRow* row2, STsdbRepo* pRepo,
-                                     STSchema **ppSchema1, STSchema **ppSchema2,
-                                     STable* pTable, int32_t* pPoints, STSRow** pLastRow) {
-
-  //for compatiblity, duplicate key inserted when update=0 should be also calculated as affected rows!
-  if(row1 == NULL && row2 == NULL && pRepo->config.update == TD_ROW_DISCARD_UPDATE) {
-    (*pPoints)++;
-    return NULL;
-  }
-
-  tsdbTrace("vgId:%d a row is %s table %s tid %d uid %" PRIu64 " key %" PRIu64, REPO_ID(pRepo),
-            "updated in", TABLE_CHAR_NAME(pTable), TABLE_TID(pTable), TABLE_UID(pTable),
-            TD_ROW_KEY(row1));
-
-  if(row2 == NULL || pRepo->config.update != TD_ROW_PARTIAL_UPDATE) {
-    void* pMem = tsdbAllocBytes(pRepo, TD_ROW_LEN(row1));
-    if(pMem == NULL) return NULL;
-    memRowCpy(pMem, row1);
-    (*pPoints)++;
-    *pLastRow = pMem;
-    return pMem;
-  }
-
-  STSchema *pSchema1 = *ppSchema1;
-  STSchema *pSchema2 = *ppSchema2;
-  SMergeBuf * pBuf = &pRepo->mergeBuf;
-  int dv1 = memRowVersion(row1);
-  int dv2 = memRowVersion(row2);
-  if(pSchema1 == NULL || schemaVersion(pSchema1) != dv1) {
-    if(pSchema2 != NULL && schemaVersion(pSchema2) == dv1) {
-      *ppSchema1 = pSchema2;
-    } else {
-      *ppSchema1 = tsdbGetTableSchemaImpl(pTable, false, false, memRowVersion(row1), (int8_t)memRowType(row1));
-    }
-    pSchema1 = *ppSchema1;
-  }
-
-  if(pSchema2 == NULL || schemaVersion(pSchema2) != dv2) {
-    if(schemaVersion(pSchema1) == dv2) {
-      pSchema2 = pSchema1;
-    } else {
-      *ppSchema2 = tsdbGetTableSchemaImpl(pTable, false, false, memRowVersion(row2), (int8_t)memRowType(row2));
-      pSchema2 = *ppSchema2;
-    }
-  }
-
-  STSRow* tmp = tsdbMergeTwoRows(pBuf, row1, row2, pSchema1, pSchema2);
-
-  void* pMem = tsdbAllocBytes(pRepo, TD_ROW_LEN(tmp));
-  if(pMem == NULL) return NULL;
-  memRowCpy(pMem, tmp);
-
-  (*pPoints)++;
-  *pLastRow = pMem;
-  return pMem;
-}
-
-static void* tsdbInsertDupKeyMergePacked(void** args) {
-  return tsdbInsertDupKeyMerge(args[0], args[1], args[2], (STSchema**)&args[3], (STSchema**)&args[4], args[5], args[6], args[7]);
-}
-
-static void tsdbSetupSkipListHookFns(SSkipList* pSkipList, STsdbRepo *pRepo, STable *pTable, int32_t* pPoints, STSRow** pLastRow) {
-
-  if(pSkipList->insertHandleFn == NULL) {
-    tGenericSavedFunc *dupHandleSavedFunc = genericSavedFuncInit((GenericVaFunc)&tsdbInsertDupKeyMergePacked, 9);
-    dupHandleSavedFunc->args[2] = pRepo;
-    dupHandleSavedFunc->args[3] = NULL;
-    dupHandleSavedFunc->args[4] = NULL;
-    dupHandleSavedFunc->args[5] = pTable;
-    pSkipList->insertHandleFn = dupHandleSavedFunc;
-  }
-  pSkipList->insertHandleFn->args[6] = pPoints;
-  pSkipList->insertHandleFn->args[7] = pLastRow;
-}
-
-static int tsdbCheckTableSchema(STsdbRepo *pRepo, SSubmitBlk *pBlock, STable *pTable) {
-  ASSERT(pTable != NULL);
-
-  STSchema *pSchema = tsdbGetTableSchemaImpl(pTable, false, false, -1, -1);
-  int       sversion = schemaVersion(pSchema);
-
-  if (pBlock->sversion == sversion) {
-    return 0;
-  } else {
-    if (TABLE_TYPE(pTable) == TSDB_STREAM_TABLE) {  // stream table is not allowed to change schema
-      terrno = TSDB_CODE_TDB_IVD_TB_SCHEMA_VERSION;
-      return -1;
-    }
-  }
-
-  if (pBlock->sversion > sversion) {  // may need to update table schema
-    if (pBlock->schemaLen > 0) {
-      tsdbDebug(
-          "vgId:%d table %s tid %d uid %" PRIu64 " schema version %d is out of data, client version %d, update...",
-          REPO_ID(pRepo), TABLE_CHAR_NAME(pTable), TABLE_TID(pTable), TABLE_UID(pTable), sversion, pBlock->sversion);
-      ASSERT(pBlock->schemaLen % sizeof(STColumn) == 0);
-      int       numOfCols = pBlock->schemaLen / sizeof(STColumn);
-      STColumn *pTCol = (STColumn *)pBlock->data;
-
-      STSchemaBuilder schemaBuilder = {0};
-      if (tdInitTSchemaBuilder(&schemaBuilder, pBlock->sversion) < 0) {
-        terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-        tsdbError("vgId:%d failed to update schema of table %s since %s", REPO_ID(pRepo), TABLE_CHAR_NAME(pTable),
-                  tstrerror(terrno));
-        return -1;
-      }
-
-      for (int i = 0; i < numOfCols; i++) {
-        if (tdAddColToSchema(&schemaBuilder, pTCol[i].type, htons(pTCol[i].colId), htons(pTCol[i].bytes)) < 0) {
-          terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-          tsdbError("vgId:%d failed to update schema of table %s since %s", REPO_ID(pRepo), TABLE_CHAR_NAME(pTable),
-                    tstrerror(terrno));
-          tdDestroyTSchemaBuilder(&schemaBuilder);
-          return -1;
-        }
-      }
-
-      STSchema *pNSchema = tdGetSchemaFromBuilder(&schemaBuilder);
-      if (pNSchema == NULL) {
-        terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-        tdDestroyTSchemaBuilder(&schemaBuilder);
-        return -1;
-      }
-
-      tdDestroyTSchemaBuilder(&schemaBuilder);
-      tsdbUpdateTableSchema(pRepo, pTable, pNSchema, true);
-    } else {
-      tsdbDebug(
-          "vgId:%d table %s tid %d uid %" PRIu64 " schema version %d is out of data, client version %d, reconfigure...",
-          REPO_ID(pRepo), TABLE_CHAR_NAME(pTable), TABLE_TID(pTable), TABLE_UID(pTable), sversion, pBlock->sversion);
-      terrno = TSDB_CODE_TDB_TABLE_RECONFIGURE;
-      return -1;
-    }
-  } else {
-    ASSERT(pBlock->sversion >= 0);
-    if (tsdbGetTableSchemaImpl(pTable, false, false, pBlock->sversion, -1) == NULL) {
-      tsdbError("vgId:%d invalid submit schema version %d to table %s tid %d from client", REPO_ID(pRepo),
-                pBlock->sversion, TABLE_CHAR_NAME(pTable), TABLE_TID(pTable));
-      terrno = TSDB_CODE_TDB_IVD_TB_SCHEMA_VERSION;
-      return -1;
-    }
-  }
-
-  return 0;
-}
-
-static void updateTableLatestColumn(STsdbRepo *pRepo, STable *pTable, STSRow* row) {
-  tsdbDebug("vgId:%d updateTableLatestColumn, %s row version:%d", REPO_ID(pRepo), pTable->name->data,
-            memRowVersion(row));
-
-  STSchema* pSchema = tsdbGetTableLatestSchema(pTable);
-  if (tsdbUpdateLastColSchema(pTable, pSchema) < 0) {
-    return;
-  }
-
-  pSchema = tsdbGetTableSchemaByVersion(pTable, memRowVersion(row), (int8_t)memRowType(row));
-  if (pSchema == NULL) {
-    return;
-  }
-
-  SDataCol *pLatestCols = pTable->lastCols;
-  int32_t kvIdx = 0;
-
-  for (int16_t j = 0; j < schemaNCols(pSchema); j++) {
-    STColumn *pTCol = schemaColAt(pSchema, j);
-    // ignore not exist colId
-    int16_t idx = tsdbGetLastColumnsIndexByColId(pTable, pTCol->colId);
-    if (idx == -1) {
-      continue;
-    }
-
-    void *value = NULL;
-
-    value = tdGetMemRowDataOfColEx(row, pTCol->colId, (int8_t)pTCol->type,
-                                   TD_DATA_ROW_HEAD_SIZE + pSchema->columns[j].offset, &kvIdx);
-
-    if ((value == NULL) || isNull(value, pTCol->type)) {
-      continue;
-    }
-    // lock
-    TSDB_WLOCK_TABLE(pTable); 
-    SDataCol *pDataCol = &(pLatestCols[idx]);
-    if (pDataCol->pData == NULL) {
-      pDataCol->pData = taosMemoryMalloc(pTCol->bytes);
-      pDataCol->bytes = pTCol->bytes;
-    } else if (pDataCol->bytes < pTCol->bytes) {
-      pDataCol->pData = taosMemoryRealloc(pDataCol->pData, pTCol->bytes);
-      pDataCol->bytes = pTCol->bytes;
-    }
-    // the actual value size
-    uint16_t bytes = IS_VAR_DATA_TYPE(pTCol->type) ? varDataTLen(value) : pTCol->bytes;
-    // the actual data size CANNOT larger than column size
-    assert(pTCol->bytes >= bytes);
-    memcpy(pDataCol->pData, value, bytes);
-    //tsdbInfo("updateTableLatestColumn vgId:%d cache column %d for %d,%s", REPO_ID(pRepo), j, pDataCol->bytes, (char*)pDataCol->pData);
-    pDataCol->ts = TD_ROW_KEY(row);
-    // unlock
-    TSDB_WUNLOCK_TABLE(pTable); 
-  }
-}
-
-static int tsdbUpdateTableLatestInfo(STsdbRepo *pRepo, STable *pTable, STSRow* row) {
-  STsdbCfg *pCfg = &pRepo->config;
-
-  // if cacheLastRow config has been reset, free the lastRow
-  if (!pCfg->cacheLastRow && pTable->lastRow != NULL) {
-    STSRow* cachedLastRow = pTable->lastRow;
-    TSDB_WLOCK_TABLE(pTable);
-    pTable->lastRow = NULL;
-    TSDB_WUNLOCK_TABLE(pTable);
-    taosTZfree(cachedLastRow);
-  }
-
-  if (tsdbGetTableLastKeyImpl(pTable) <= TD_ROW_KEY(row)) {
-    if (CACHE_LAST_ROW(pCfg) || pTable->lastRow != NULL) {
-      STSRow* nrow = pTable->lastRow;
-      if (taosTSizeof(nrow) < TD_ROW_LEN(row)) {
-        STSRow* orow = nrow;
-        nrow = taosTMalloc(TD_ROW_LEN(row));
-        if (nrow == NULL) {
-          terrno = TSDB_CODE_TDB_OUT_OF_MEMORY;
-          return -1;
-        }
-
-        memRowCpy(nrow, row);
-        TSDB_WLOCK_TABLE(pTable);
-        pTable->lastKey = TD_ROW_KEY(row);
-        pTable->lastRow = nrow;
-        TSDB_WUNLOCK_TABLE(pTable);
-        taosTZfree(orow);
-      } else {
-        TSDB_WLOCK_TABLE(pTable);
-        pTable->lastKey = TD_ROW_KEY(row);
-        memRowCpy(nrow, row);
-        TSDB_WUNLOCK_TABLE(pTable);
-      }
-    } else {
-      pTable->lastKey = TD_ROW_KEY(row);
-    }
-
-    if (CACHE_LAST_NULL_COLUMN(pCfg)) {
-      updateTableLatestColumn(pRepo, pTable, row);
-    }
-  }
-
-  pTable->cacheLastConfigVersion = pRepo->cacheLastConfigVersion;
-
-  return 0;
-}
-
-#endif

source/dnode/vnode/src/tsdb/tsdbOpen.c

@@ -0,0 +1,89 @@
+/*
+ * 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/>.
+ */
+
+#include "tsdb.h"
+
+int tsdbOpen(SVnode *pVnode, STsdb **ppTsdb) {
+  STsdb *pTsdb = NULL;
+  int    slen = 0;
+
+  *ppTsdb = NULL;
+  slen = strlen(tfsGetPrimaryPath(pVnode->pTfs)) + strlen(pVnode->path) + strlen(VNODE_TSDB_DIR) + 3;
+
+  // create handle
+  pTsdb = (STsdb *)taosMemoryCalloc(1, sizeof(*pTsdb) + slen);
+  if (pTsdb == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  pTsdb->path = (char *)&pTsdb[1];
+  sprintf(pTsdb->path, "%s%s%s%s%s", tfsGetPrimaryPath(pVnode->pTfs), TD_DIRSEP, pVnode->path, TD_DIRSEP,
+          VNODE_TSDB_DIR);
+  pTsdb->pVnode = pVnode;
+  pTsdb->repoLocked = false;
+  tdbMutexInit(&pTsdb->mutex, NULL);
+  pTsdb->config = pVnode->config.tsdbCfg;
+  pTsdb->fs = tsdbNewFS(&pTsdb->config);
+
+  // create dir (TODO: use tfsMkdir)
+  taosMkDir(pTsdb->path);
+
+  // open tsdb
+  if (tsdbOpenFS(pTsdb) < 0) {
+    goto _err;
+  }
+
+  tsdbDebug("vgId: %d tsdb is opened", TD_VID(pVnode));
+
+  *ppTsdb = pTsdb;
+  return 0;
+
+_err:
+  taosMemoryFree(pTsdb);
+  return -1;
+}
+
+int tsdbClose(STsdb *pTsdb) {
+  if (pTsdb) {
+    tsdbCloseFS(pTsdb);
+    tsdbFreeFS(pTsdb->fs);
+    taosMemoryFree(pTsdb);
+  }
+  return 0;
+}
+
+int tsdbLockRepo(STsdb *pTsdb) {
+  int code = taosThreadMutexLock(&pTsdb->mutex);
+  if (code != 0) {
+    tsdbError("vgId:%d failed to lock tsdb since %s", REPO_ID(pTsdb), strerror(errno));
+    terrno = TAOS_SYSTEM_ERROR(code);
+    return -1;
+  }
+  pTsdb->repoLocked = true;
+  return 0;
+}
+
+int tsdbUnlockRepo(STsdb *pTsdb) {
+  ASSERT(IS_REPO_LOCKED(pTsdb));
+  pTsdb->repoLocked = false;
+  int code = taosThreadMutexUnlock(&pTsdb->mutex);
+  if (code != 0) {
+    tsdbError("vgId:%d failed to unlock tsdb since %s", REPO_ID(pTsdb), strerror(errno));
+    terrno = TAOS_SYSTEM_ERROR(code);
+    return -1;
+  }
+  return 0;
+}

source/dnode/vnode/src/vnd/vnodeCfg.c

@@ -23,8 +23,6 @@ const SVnodeCfg vnodeCfgDefault = {
     .szCache = 256,
     .szBuf = 96 * 1024 * 1024,
     .isHeap = false,
-    .ttl = 0,
-    .keep = 0,
     .streamMode = 0,
     .isWeak = 0,
     .tsdbCfg = {.precision = TSDB_TIME_PRECISION_MILLI,
@@ -58,8 +56,6 @@ int vnodeEncodeConfig(const void *pObj, SJson *pJson) {
   if (tjsonAddIntegerToObject(pJson, "szCache", pCfg->szCache) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "szBuf", pCfg->szBuf) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "isHeap", pCfg->isHeap) < 0) return -1;
-  if (tjsonAddIntegerToObject(pJson, "ttl", pCfg->ttl) < 0) return -1;
-  if (tjsonAddIntegerToObject(pJson, "keep", pCfg->keep) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "streamMode", pCfg->streamMode) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "isWeak", pCfg->isWeak) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "precision", pCfg->tsdbCfg.precision) < 0) return -1;
@@ -72,7 +68,6 @@ int vnodeEncodeConfig(const void *pObj, SJson *pJson) {
   if (tjsonAddIntegerToObject(pJson, "keep0", pCfg->tsdbCfg.keep0) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "keep1", pCfg->tsdbCfg.keep1) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "keep2", pCfg->tsdbCfg.keep2) < 0) return -1;
-  if (tjsonAddIntegerToObject(pJson, "lruCacheSize", pCfg->tsdbCfg.lruCacheSize) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "wal.vgId", pCfg->walCfg.vgId) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "wal.fsyncPeriod", pCfg->walCfg.fsyncPeriod) < 0) return -1;
   if (tjsonAddIntegerToObject(pJson, "wal.retentionPeriod", pCfg->walCfg.retentionPeriod) < 0) return -1;
@@ -109,8 +104,6 @@ int vnodeDecodeConfig(const SJson *pJson, void *pObj) {
   if (tjsonGetNumberValue(pJson, "szCache", pCfg->szCache) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "szBuf", pCfg->szBuf) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "isHeap", pCfg->isHeap) < 0) return -1;
-  if (tjsonGetNumberValue(pJson, "ttl", pCfg->ttl) < 0) return -1;
-  if (tjsonGetNumberValue(pJson, "keep", pCfg->keep) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "streamMode", pCfg->streamMode) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "isWeak", pCfg->isWeak) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "precision", pCfg->tsdbCfg.precision) < 0) return -1;
@@ -123,7 +116,6 @@ int vnodeDecodeConfig(const SJson *pJson, void *pObj) {
   if (tjsonGetNumberValue(pJson, "keep0", pCfg->tsdbCfg.keep0) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "keep1", pCfg->tsdbCfg.keep1) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "keep2", pCfg->tsdbCfg.keep2) < 0) return -1;
-  if (tjsonGetNumberValue(pJson, "lruCacheSize", pCfg->tsdbCfg.lruCacheSize) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "wal.vgId", pCfg->walCfg.vgId) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "wal.fsyncPeriod", pCfg->walCfg.fsyncPeriod) < 0) return -1;
   if (tjsonGetNumberValue(pJson, "wal.retentionPeriod", pCfg->walCfg.retentionPeriod) < 0) return -1;

source/dnode/vnode/src/vnd/vnodeQuery.c

@@ -138,3 +138,13 @@ int32_t vnodeGetLoad(SVnode *pVnode, SVnodeLoad *pLoad) {
   pLoad->numOfBatchInsertSuccessReqs = 4;
   return 0;
 }
+
+void vnodeGetInfo(SVnode *pVnode, const char **dbname, int32_t *vgId) {
+  if (dbname) {
+    *dbname = pVnode->config.dbname;
+  }
+
+  if (vgId) {
+    *vgId = TD_VID(pVnode);
+  }
+}

source/dnode/vnode/src/vnd/vnodeSvr.c

@@ -142,7 +142,7 @@ _err:
 
 int vnodeProcessQueryMsg(SVnode *pVnode, SRpcMsg *pMsg) {
   vTrace("message in vnode query queue is processing");
-  SReadHandle handle = {.reader = pVnode->pTsdb, .meta = pVnode->pMeta, .config = &pVnode->config};
+  SReadHandle handle = {.reader = pVnode->pTsdb, .meta = pVnode->pMeta, .config = &pVnode->config, .vnode = pVnode};
 
   switch (pMsg->msgType) {
     case TDMT_VND_QUERY:

source/libs/executor/inc/executorimpl.h

@@ -380,7 +380,7 @@ typedef struct SStreamBlockScanInfo {
 typedef struct SSysTableScanInfo {
   union {
     void* pTransporter;
-    void* readHandle;
+    SReadHandle readHandle;
   };
 
   SRetrieveMetaTableRsp* pRsp;

source/libs/executor/src/executorimpl.c

@@ -3738,7 +3738,32 @@ static int32_t doSendFetchDataRequest(SExchangeInfo* pExchangeInfo, SExecTaskInf
   return TSDB_CODE_SUCCESS;
 }
 
-// TODO if only one or two columns required, how to extract data?
+// NOTE: sources columns are more than the destination SSDatablock columns.
+static void relocateColumnData(SSDataBlock* pBlock, const SArray* pColMatchInfo, SArray* pCols) {
+  size_t numOfSrcCols = taosArrayGetSize(pCols);
+  ASSERT(numOfSrcCols >= pBlock->info.numOfCols);
+
+  int32_t i = 0, j = 0;
+  while(i < numOfSrcCols && j < taosArrayGetSize(pColMatchInfo)) {
+    SColumnInfoData* p = taosArrayGet(pCols, i);
+    SColMatchInfo*   pmInfo = taosArrayGet(pColMatchInfo, j);
+    if (!pmInfo->output) {
+      j++;
+      continue;
+    }
+
+    if (p->info.colId == pmInfo->colId) {
+      taosArraySet(pBlock->pDataBlock, pmInfo->targetSlotId, p);
+      i++;
+      j++;
+    } else if (p->info.colId < pmInfo->colId) {
+      i++;
+    } else {
+      ASSERT(0);
+    }
+  }
+}
+
 int32_t setSDataBlockFromFetchRsp(SSDataBlock* pRes, SLoadRemoteDataInfo* pLoadInfo, int32_t numOfRows, char* pData,
                                   int32_t compLen, int32_t numOfOutput, int64_t startTs, uint64_t* total,
                                   SArray* pColList) {
@@ -3756,7 +3781,7 @@ int32_t setSDataBlockFromFetchRsp(SSDataBlock* pRes, SLoadRemoteDataInfo* pLoadI
     char* pStart = pData + sizeof(int32_t) * numOfOutput;
     for (int32_t i = 0; i < numOfOutput; ++i) {
       colLen[i] = htonl(colLen[i]);
-      ASSERT(colLen[i] > 0);
+      ASSERT(colLen[i] >= 0);
 
       SColumnInfoData* pColInfoData = taosArrayGet(pRes->pDataBlock, i);
       if (IS_VAR_DATA_TYPE(pColInfoData->info.type)) {
@@ -3766,13 +3791,18 @@ int32_t setSDataBlockFromFetchRsp(SSDataBlock* pRes, SLoadRemoteDataInfo* pLoadI
         memcpy(pColInfoData->varmeta.offset, pStart, sizeof(int32_t) * numOfRows);
         pStart += sizeof(int32_t) * numOfRows;
 
+        if (colLen[i] > 0) {
           pColInfoData->pData = taosMemoryMalloc(colLen[i]);
+        }
       } else {
         memcpy(pColInfoData->nullbitmap, pStart, BitmapLen(numOfRows));
         pStart += BitmapLen(numOfRows);
       }
 
+      if (colLen[i] > 0) {
         memcpy(pColInfoData->pData, pStart, colLen[i]);
+      }
+
       //TODO setting this flag to true temporarily so aggregate function on stable will
       //examine NULL value for non-primary key column
       pColInfoData->hasNull = true;
@@ -3785,6 +3815,7 @@ int32_t setSDataBlockFromFetchRsp(SSDataBlock* pRes, SLoadRemoteDataInfo* pLoadI
     int32_t numOfCols = htonl(*(int32_t*)pStart);
     pStart += sizeof(int32_t);
 
+    // todo refactor:extract method
     SSysTableSchema* pSchema = (SSysTableSchema*)pStart;
     for (int32_t i = 0; i < numOfCols; ++i) {
       SSysTableSchema* p = (SSysTableSchema*)pStart;
@@ -3839,19 +3870,7 @@ int32_t setSDataBlockFromFetchRsp(SSDataBlock* pRes, SLoadRemoteDataInfo* pLoadI
     }
 
     // data from mnode
-    for (int32_t i = 0; i < numOfCols; ++i) {
+    relocateColumnData(pRes,  pColList, block.pDataBlock);
-      SColumnInfoData* pSrc = taosArrayGet(block.pDataBlock, i);
-
-      for (int32_t j = 0; j < numOfOutput; ++j) {
-        int16_t colIndex = *(int16_t*)taosArrayGet(pColList, j);
-
-        if (colIndex - 1 == i) {
-          SColumnInfoData* pColInfoData = taosArrayGet(pRes->pDataBlock, j);
-          colDataAssign(pColInfoData, pSrc, numOfRows);
-          break;
-        }
-      }
-    }
   }
 
   pRes->info.rows = numOfRows;
@@ -6423,7 +6442,6 @@ static tsdbReaderT doCreateDataReader(STableScanPhysiNode* pTableScanNode, SRead
 static int32_t doCreateTableGroup(void* metaHandle, int32_t tableType, uint64_t tableUid, STableGroupInfo* pGroupInfo,
                                   uint64_t queryId, uint64_t taskId);
 static SArray* extractTableIdList(const STableGroupInfo* pTableGroupInfo);
-static SArray* extractScanColumnId(SNodeList* pNodeList);
 static SArray* extractColumnInfo(SNodeList* pNodeList);
 static SArray* extractColMatchInfo(SNodeList* pNodeList, SDataBlockDescNode* pOutputNodeList, int32_t* numOfOutputCols);
 
@@ -6494,10 +6512,11 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
       SScanPhysiNode* pScanNode = &pSysScanPhyNode->scan;
 
       SSDataBlock* pResBlock = createResDataBlock(pScanNode->node.pOutputDataBlockDesc);
-      SArray* colList = extractScanColumnId(pScanNode->pScanCols);
 
+      int32_t numOfOutputCols = 0;
+      SArray* colList = extractColMatchInfo(pScanNode->pScanCols, pScanNode->node.pOutputDataBlockDesc, &numOfOutputCols);
       SOperatorInfo* pOperator = createSysTableScanOperatorInfo(
-          pHandle->meta, pResBlock, &pScanNode->tableName, pScanNode->node.pConditions, pSysScanPhyNode->mgmtEpSet,
+          pHandle, pResBlock, &pScanNode->tableName, pScanNode->node.pConditions, pSysScanPhyNode->mgmtEpSet,
           colList, pTaskInfo, pSysScanPhyNode->showRewrite, pSysScanPhyNode->accountId);
       return pOperator;
     } else {
@@ -6658,28 +6677,6 @@ static  int32_t initQueryTableDataCond(SQueryTableDataCond* pCond, const STableS
   return TSDB_CODE_SUCCESS;
 }
 
-SArray* extractScanColumnId(SNodeList* pNodeList) {
-  size_t  numOfCols = LIST_LENGTH(pNodeList);
-  SArray* pList = taosArrayInit(numOfCols, sizeof(int16_t));
-  if (pList == NULL) {
-    terrno = TSDB_CODE_OUT_OF_MEMORY;
-    return NULL;
-  }
-
-  for (int32_t i = 0; i < numOfCols; ++i) {
-    for (int32_t j = 0; j < numOfCols; ++j) {
-      STargetNode* pNode = (STargetNode*)nodesListGetNode(pNodeList, j);
-      if (pNode->slotId == i) {
-        SColumnNode* pColNode = (SColumnNode*)pNode->pExpr;
-        taosArrayPush(pList, &pColNode->colId);
-        break;
-      }
-    }
-  }
-
-  return pList;
-}
-
 SArray* extractColumnInfo(SNodeList* pNodeList) {
   size_t  numOfCols = LIST_LENGTH(pNodeList);
   SArray* pList = taosArrayInit(numOfCols, sizeof(SColumn));
@@ -6815,9 +6812,9 @@ SArray* extractColMatchInfo(SNodeList* pNodeList, SDataBlockDescNode* pOutputNod
     SColumnNode* pColNode = (SColumnNode*)pNode->pExpr;
 
     SColMatchInfo c = {0};
+    c.output = true;
     c.colId = pColNode->colId;
     c.targetSlotId = pNode->slotId;
-    c.output = true;
     taosArrayPush(pList, &c);
   }
 
@@ -6825,8 +6822,10 @@ SArray* extractColMatchInfo(SNodeList* pNodeList, SDataBlockDescNode* pOutputNod
   int32_t num = LIST_LENGTH(pOutputNodeList->pSlots);
   for (int32_t i = 0; i < num; ++i) {
     SSlotDescNode* pNode = (SSlotDescNode*)nodesListGetNode(pOutputNodeList->pSlots, i);
+
     // todo: add reserve flag check
-    if (pNode->slotId >= numOfCols) {  // it is a column reserved for the arithmetic expression calculation
+    // it is a column reserved for the arithmetic expression calculation
+    if (pNode->slotId >= numOfCols) {
       (*numOfOutputCols) += 1;
       continue;
     }

source/libs/executor/src/scanoperator.c

@@ -811,7 +811,7 @@ static SSDataBlock* doSysTableScan(SOperatorInfo* pOperator, bool* newgroup) {
   const char* name = tNameGetTableName(&pInfo->name);
   if (strncasecmp(name, TSDB_INS_TABLE_USER_TABLES, TSDB_TABLE_FNAME_LEN) == 0) {
     if (pInfo->pCur == NULL) {
-      pInfo->pCur = metaOpenTbCursor(pInfo->readHandle);
+      pInfo->pCur = metaOpenTbCursor(pInfo->readHandle.meta);
     }
 
     blockDataCleanup(pInfo->pRes);
@@ -819,33 +819,98 @@ static SSDataBlock* doSysTableScan(SOperatorInfo* pOperator, bool* newgroup) {
     int32_t          tableNameSlotId = 1;
     SColumnInfoData* pTableNameCol = taosArrayGet(pInfo->pRes->pDataBlock, tableNameSlotId);
 
-    char*   tb = NULL;
     int32_t numOfRows = 0;
 
+    const char* db = NULL;
+    int32_t vgId = 0;
+    vnodeGetInfo(pInfo->readHandle.vnode, &db, &vgId);
+
+    SName sn = {0};
+    char dbname[TSDB_DB_FNAME_LEN + VARSTR_HEADER_SIZE] = {0};
+    tNameFromString(&sn, db, T_NAME_ACCT|T_NAME_DB);
+
+    tNameGetDbName(&sn, varDataVal(dbname));
+    varDataSetLen(dbname, strlen(varDataVal(dbname)));
+
     char n[TSDB_TABLE_NAME_LEN] = {0};
     while (metaTbCursorNext(pInfo->pCur) == 0) {
       STR_TO_VARSTR(n, pInfo->pCur->mr.me.name);
       colDataAppend(pTableNameCol, numOfRows, n, false);
-      numOfRows += 1;
+
-      if (numOfRows >= pInfo->capacity) {
+      int32_t tableType = pInfo->pCur->mr.me.type;
+
+      // database name
+      SColumnInfoData* pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 0);
+      colDataAppend(pColInfoData, numOfRows, dbname, false);
+
+      // vgId
+      pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 6);
+      colDataAppend(pColInfoData, numOfRows, (char*) &vgId, false);
+
+      // table comment
+      // todo: set the correct comment
+      pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 8);
+      colDataAppendNULL(pColInfoData, numOfRows);
+
+      char str[256] = {0};
+      if (tableType == TSDB_CHILD_TABLE) {
+        SMetaReader mr = {0};
+        metaReaderInit(&mr, pInfo->readHandle.meta, 0);
+        metaGetTableEntryByUid(&mr, pInfo->pCur->mr.me.ctbEntry.suid);
+
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 3);
+        colDataAppend(pColInfoData, numOfRows,  (char*) &mr.me.stbEntry.schema.nCols, false);
+
+        // create time
+        int64_t ts = pInfo->pCur->mr.me.ctbEntry.ctime;
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 2);
+        colDataAppend(pColInfoData, numOfRows, (char*) &ts, false);
+
+        // super table name
+        STR_TO_VARSTR(str, mr.me.name);
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 4);
+        colDataAppend(pColInfoData, numOfRows, str, false);
+        metaReaderClear(&mr);
+
+        // uid
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 5);
+        colDataAppend(pColInfoData, numOfRows, (char*) &pInfo->pCur->mr.me.uid, false);
+
+        // ttl
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 7);
+        colDataAppend(pColInfoData, numOfRows, (char*) &pInfo->pCur->mr.me.ctbEntry.ttlDays, false);
+
+        STR_TO_VARSTR(str, "CHILD_TABLE");
+      } else if (tableType == TSDB_NORMAL_TABLE) {
+        // create time
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 2);
+        colDataAppend(pColInfoData, numOfRows, (char*) &pInfo->pCur->mr.me.ntbEntry.ctime, false);
+
+        // number of columns
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 3);
+        colDataAppend(pColInfoData, numOfRows,  (char*) &pInfo->pCur->mr.me.ntbEntry.schema.nCols, false);
+
+        // super table name
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 4);
+        colDataAppendNULL(pColInfoData, numOfRows);
+
+        // uid
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 5);
+        colDataAppend(pColInfoData, numOfRows, (char*) &pInfo->pCur->mr.me.uid, false);
+
+        // ttl
+        pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 7);
+        colDataAppend(pColInfoData, numOfRows, (char*) &pInfo->pCur->mr.me.ntbEntry.ttlDays, false);
+
+        STR_TO_VARSTR(str, "NORMAL_TABLE");
+      }
+
+      pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, 9);
+      colDataAppend(pColInfoData, numOfRows, str, false);
+
+      if (++numOfRows >= pInfo->capacity) {
         break;
       }
-
-      for (int32_t i = 0; i < pInfo->pRes->info.numOfCols; ++i) {
-        if (i == tableNameSlotId) {
-          continue;
-        }
-
-        SColumnInfoData* pColInfoData = taosArrayGet(pInfo->pRes->pDataBlock, i);
-        int64_t          tmp = 0;
-        char             t[10] = {0};
-        STR_TO_VARSTR(t, "_");  // TODO
-        if (IS_VAR_DATA_TYPE(pColInfoData->info.type)) {
-          colDataAppend(pColInfoData, numOfRows, t, false);
-        } else {
-          colDataAppend(pColInfoData, numOfRows, (char*)&tmp, false);
-        }
-      }
     }
 
     pInfo->loadInfo.totalRows += numOfRows;
@@ -923,7 +988,7 @@ static SSDataBlock* doSysTableScan(SOperatorInfo* pOperator, bool* newgroup) {
   }
 }
 
-SOperatorInfo* createSysTableScanOperatorInfo(void* pSysTableReadHandle, SSDataBlock* pResBlock, const SName* pName,
+SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSDataBlock* pResBlock, const SName* pName,
                                               SNode* pCondition, SEpSet epset, SArray* colList,
                                               SExecTaskInfo* pTaskInfo, bool showRewrite, int32_t accountId) {
   SSysTableScanInfo* pInfo = taosMemoryCalloc(1, sizeof(SSysTableScanInfo));
@@ -945,7 +1010,7 @@ SOperatorInfo* createSysTableScanOperatorInfo(void* pSysTableReadHandle, SSDataB
   tNameAssign(&pInfo->name, pName);
   const char* name = tNameGetTableName(&pInfo->name);
   if (strncasecmp(name, TSDB_INS_TABLE_USER_TABLES, TSDB_TABLE_FNAME_LEN) == 0) {
-    pInfo->readHandle = pSysTableReadHandle;
+    pInfo->readHandle = *(SReadHandle*) readHandle;
     blockDataEnsureCapacity(pInfo->pRes, pInfo->capacity);
   } else {
     tsem_init(&pInfo->ready, 0, 0);

source/libs/transport/inc/transComm.h

@@ -21,6 +21,7 @@ extern "C" {
 #include <uv.h>
 #include "lz4.h"
 #include "os.h"
+#include "osSocket.h"
 #include "rpcCache.h"
 #include "rpcHead.h"
 #include "rpcLog.h"
@@ -105,6 +106,7 @@ typedef void* queue[2];
 
 #define TRANS_RETRY_COUNT_LIMIT 10  // retry count limit
 #define TRANS_RETRY_INTERVAL    5   // ms retry interval
+#define TRANS_CONN_TIMEOUT      3   // connect timeout
 
 typedef struct {
   SRpcInfo* pRpc;     // associated SRpcInfo

source/libs/transport/src/transCli.c

@@ -103,6 +103,10 @@ static void      cliDestroyConn(SCliConn* pConn, bool clear /*clear tcp handle o
 static void      cliDestroy(uv_handle_t* handle);
 static void      cliSend(SCliConn* pConn);
 
+/*
+ * set TCP connection timeout per-socket level
+ */
+static int cliCreateSocket();
 // process data read from server, add decompress etc later
 static void cliHandleResp(SCliConn* conn);
 // handle except about conn
@@ -729,9 +733,15 @@ void cliHandleReq(SCliMsg* pMsg, SCliThrdObj* pThrd) {
     if (ret) {
       tError("%s cli conn %p failed to set conn option, errmsg %s", pTransInst->label, conn, uv_err_name(ret));
     }
+    int fd = taosCreateSocketWithTimeOutOpt(TRANS_CONN_TIMEOUT);
+    if (fd == -1) {
+      tTrace("%s cli conn %p failed to create socket", pTransInst->label, conn);
+      cliHandleExcept(conn);
+      return;
+    }
+    uv_tcp_open((uv_tcp_t*)conn->stream, fd);
 
     struct sockaddr_in addr;
-
     addr.sin_family = AF_INET;
     addr.sin_addr.s_addr = taosGetIpv4FromFqdn(conn->ip);
     addr.sin_port = (uint16_t)htons((uint16_t)conn->port);

source/os/src/osSocket.c

@@ -37,20 +37,24 @@
 #include <unistd.h>
 
 #if defined(DARWIN)
-  #include <dispatch/dispatch.h>
+#include <dispatch/dispatch.h>
-  #include "osEok.h"
+#include "osEok.h"
 #else
-  #include <sys/epoll.h>
+#include <sys/epoll.h>
 #endif
 #endif
 
+#ifndef INVALID_SOCKET
+#define INVALID_SOCKET -1
+#endif
+
 typedef struct TdSocketServer {
 #if SOCKET_WITH_LOCK
   TdThreadRwlock rwlock;
 #endif
   int      refId;
   SocketFd fd;
-} *TdSocketServerPtr, TdSocketServer;
+} * TdSocketServerPtr, TdSocketServer;
 
 typedef struct TdEpoll {
 #if SOCKET_WITH_LOCK
@@ -58,7 +62,7 @@ typedef struct TdEpoll {
 #endif
   int     refId;
   EpollFd fd;
-} *TdEpollPtr, TdEpoll;
+} * TdEpollPtr, TdEpoll;
 
 int32_t taosSendto(TdSocketPtr pSocket, void *buf, int len, unsigned int flags, const struct sockaddr *dest_addr,
                    int addrlen) {
@@ -94,7 +98,8 @@ int32_t taosReadSocket(TdSocketPtr pSocket, void *buf, int len) {
 #endif
 }
 
-int32_t taosReadFromSocket(TdSocketPtr pSocket, void *buf, int32_t len, int32_t flags, struct sockaddr *destAddr, int *addrLen) {
+int32_t taosReadFromSocket(TdSocketPtr pSocket, void *buf, int32_t len, int32_t flags, struct sockaddr *destAddr,
+                           int *addrLen) {
   if (pSocket == NULL || pSocket->fd < 0) {
     return -1;
   }
@@ -318,7 +323,7 @@ int32_t taosWriteMsg(TdSocketPtr pSocket, void *buf, int32_t nbytes) {
     return -1;
   }
   int32_t nleft, nwritten;
-  char   *ptr = (char *)buf;
+  char *  ptr = (char *)buf;
 
   nleft = nbytes;
 
@@ -347,7 +352,7 @@ int32_t taosReadMsg(TdSocketPtr pSocket, void *buf, int32_t nbytes) {
     return -1;
   }
   int32_t nleft, nread;
-  char   *ptr = (char *)buf;
+  char *  ptr = (char *)buf;
 
   nleft = nbytes;
 
@@ -689,8 +694,7 @@ TdSocketServerPtr taosOpenTcpServerSocket(uint32_t ip, uint16_t port) {
   return (TdSocketServerPtr)pSocket;
 }
 
-TdSocketPtr taosAcceptTcpConnectSocket(TdSocketServerPtr pServerSocket, struct sockaddr *destAddr,
+TdSocketPtr taosAcceptTcpConnectSocket(TdSocketServerPtr pServerSocket, struct sockaddr *destAddr, int *addrLen) {
-                                       int *addrLen) {
   if (pServerSocket == NULL || pServerSocket->fd < 0) {
     return NULL;
   }
@@ -771,7 +775,7 @@ uint32_t taosGetIpv4FromFqdn(const char *fqdn) {
 
   int32_t ret = getaddrinfo(fqdn, NULL, &hints, &result);
   if (result) {
-    struct sockaddr    *sa = result->ai_addr;
+    struct sockaddr *   sa = result->ai_addr;
     struct sockaddr_in *si = (struct sockaddr_in *)sa;
     struct in_addr      ia = si->sin_addr;
     uint32_t            ip = ia.s_addr;
@@ -887,7 +891,6 @@ int32_t taosGetSocketName(TdSocketPtr pSocket, struct sockaddr *destAddr, int *a
   return getsockname(pSocket->fd, destAddr, addrLen);
 }
 
-
 TdEpollPtr taosCreateEpoll(int32_t size) {
   EpollFd fd = -1;
 #ifdef WINDOWS
@@ -939,3 +942,28 @@ int32_t taosCloseEpoll(TdEpollPtr *ppEpoll) {
   taosMemoryFree(*ppEpoll);
   return code;
 }
+/*
+ * Set TCP connection timeout per-socket level.
+ * ref [https://github.com/libuv/help/issues/54]
+ */
+int taosCreateSocketWithTimeOutOpt(uint32_t conn_timeout_sec) {
+#if defined(WINDOWS)
+  SOCKET fd;
+#else
+  int      fd;
+#endif
+  if ((fd = socket(PF_INET, SOCK_STREAM, IPPROTO_TCP)) == INVALID_SOCKET) {
+    return -1;
+  }
+#if defined(WINDOWS)
+  if (0 != setsockopt(fd, IPPROTO_TCP, TCP_MAXRT, (char *)&conn_timeout_sec, sizeof(conn_timeout_sec))) {
+    return -1;
+  }
+#else  // Linux like systems
+  uint32_t conn_timeout_ms = conn_timeout_sec * 1000;
+  if (0 != setsockopt(fd, IPPROTO_TCP, TCP_USER_TIMEOUT, (char *)&conn_timeout_ms, sizeof(conn_timeout_ms))) {
+    return -1;
+  }
+#endif
+  return (int)fd;
+}

source/util/src/tarray.c

@@ -482,3 +482,22 @@ void taosArraySortPWithExt(SArray* pArray, __ext_compar_fn_t fn, const void* par
   taosArrayGetSize(pArray) > 8 ? taosArrayQuickSort(pArray, fn, param) : taosArrayInsertSort(pArray, fn, param);
 }
 // TODO(yihaoDeng) add order array<type>
+//
+
+char* taosShowStrArray(const SArray* pArray) {
+  int32_t sz = pArray->size;
+  int32_t tlen = 0;
+  for (int32_t i = 0; i < sz; i++) {
+    tlen += strlen(taosArrayGetP(pArray, i)) + 1;
+  }
+  char* res = taosMemoryCalloc(1, tlen);
+  char* buf = res;
+  for (int32_t i = 0; i < sz; i++) {
+    char*   str = taosArrayGetP(pArray, i);
+    int32_t len = strlen(str);
+    memcpy(buf, str, len);
+    buf += len;
+    if (i != sz - 1) *buf = ',';
+  }
+  return res;
+}

source/util/src/terror.c

@@ -271,9 +271,14 @@ TAOS_DEFINE_ERROR(TSDB_CODE_MND_TRANS_INVALID_STAGE,      "Invalid stage to kill
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_TRANS_CANT_PARALLEL,      "Invalid stage to kill")
 
 // mnode-mq
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_UNSUPPORTED_TOPIC,        "Topic with aggregation is unsupported")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_TOPIC_ALREADY_EXIST,      "Topic already exists")
-TAOS_DEFINE_ERROR(TSDB_CODE_MND_CONSUMER_NOT_READY,       "Consumer waiting for rebalance")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_TOPIC_NOT_EXIST,          "Topic not exist")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_TOO_MANY_TOPICS,          "Too many Topics")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_TOPIC,            "Invalid topic")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_TOPIC_QUERY,      "Topic with invalid query")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_INVALID_TOPIC_OPTION,     "Topic with invalid option")
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_CONSUMER_NOT_EXIST,       "Consumer not exist")
+TAOS_DEFINE_ERROR(TSDB_CODE_MND_CONSUMER_NOT_READY,       "Consumer waiting for rebalance")
 
 // mnode-sma
 TAOS_DEFINE_ERROR(TSDB_CODE_MND_SMA_ALREADY_EXIST,        "SMA already exists")

tests/script/general/user/authority.sim

@@ -1,66 +0,0 @@
-system sh/stop_dnodes.sh
-system sh/deploy.sh -n dnode1 -i 1
-system sh/exec.sh -n dnode1 -s start
-sql connect
-sleep 2000
-
-print ============= step1
-sql create user read pass 'taosdata'
-sql create user write pass 'taosdata'
-sql show users
-if $rows != 5 then
-  return -1
-endi
-
-print ============= step2
-sql close
-sql connect read
-sleep 2000
-
-sql create database dread
-sql show databases
-if $rows != 1 then
-  return -1
-endi
-
-print ============= step3
-sql close
-sql connect write
-sleep 2000
-
-sql create database dwrite
-sql show databases
-if $rows != 2 then
-  return -1
-endi
-
-print ============ step4
-sql close
-sql connect
-sleep 2000
-
-sql show databases
-if $row != 2 then
-  return -1
-endi
-
-print ============ step5
-sql close
-sql connect read
-sleep 2000
-sql drop database dread
-sql drop database dwrite
-
-
-sql close
-sql connect
-sql show databases
-if $rows != 0 then
-  return -1
-endi
-
-sql close
-sql connect
-sleep 2000
-
-system sh/exec.sh -n dnode1 -s stop  -x SIGINT

tests/script/general/user/user_create.sim

@@ -1,84 +0,0 @@
-system sh/stop_dnodes.sh
-
-system sh/deploy.sh -n dnode1 -i 1
-system sh/cfg.sh -n dnode1 -c wallevel -v 0
-system sh/exec.sh -n dnode1 -s start
-sql connect
-
-print =============== step1
-sql show users
-if $rows != 3 then 
-  return -1
-endi
-
-sql create user read PASS 'pass123'
-sql create user read PASS 'pass123' -x step1
-	return -1
-step1:
-
-sql show users
-if $rows != 4 then 
-  return -1
-endi
-
-sql alter user read PASS 'taosdata'
-
-print =============== step2
-sql close
-sql connect read
-sleep 2000
-
-sql alter user read PASS 'taosdata'
-
-print =============== step3
-sql drop user read -x step31
-  return -1
-step31:
-sql drop user _root -x step32
-  return -1
-step32:
-sql drop user monitor -x step33
-  return -1
-step33:
-
-print =============== step4
-sql close
-sql connect
-sleep 2000
-
-sql alter user read privilege read
-sql show users
-print $data1_read
-if $data1_read != readable then 
-  return -1
-endi
-
-sql_error alter user read privilege super
-sql show users
-print $data1_read
-if $data1_read != readable then
-  return -1
-endi
-
-sql alter user read privilege write
-sql show users
-print $data1_read
-if $data1_read != writable then 
-  return -1
-endi
-
-sql alter user read privilege 1 -x step43
-  return -1
-step43:
-
-sql drop user _root -x step41
-  return -1
-step41:
-
-sql drop user monitor -x step42
-  return -1
-step42:
-
-sql drop user read
-
-system sh/exec.sh -n dnode1 -s stop -x SIGINT

tests/script/jenkins/basic.txt

@@ -3,6 +3,9 @@
 
 # ---- user
 ./test.sh -f tsim/user/basic1.sim
+./test.sh -f tsim/user/pass_alter.sim
+./test.sh -f tsim/user/pass_len.sim
+./test.sh -f tsim/user/user_len.sim
 
 # ---- db
 ./test.sh -f tsim/db/create_all_options.sim

tests/script/tsim/user/pass_alter.sim

@@ -1,8 +1,6 @@
 system sh/stop_dnodes.sh
 system sh/deploy.sh -n dnode1 -i 1
-system sh/cfg.sh -n dnode1 -c wallevel -v 0
 system sh/exec.sh -n dnode1 -s start
-sleep 2000
 sql connect
 
 print ============= step1
@@ -13,7 +11,7 @@ sql alter user read pass 'taosdata'
 sql alter user write pass 'taosdata'
 
 sql show users
-if $rows != 5 then 
+if $rows != 3 then 
   return -1
 endi
 
@@ -27,11 +25,11 @@ sql alter user write pass 'taosdata1' -x step2
   return -1
 step2:
 
-sql_error create user read pass 'taosdata1'
+sql_error create user read1 pass 'taosdata1'
-sql_error create user write pass 'taosdata1'
+sql_error create user write1 pass 'taosdata1'
 
 sql show users
-if $rows != 5 then 
+if $rows != 3 then 
   return -1
 endi
 
@@ -41,27 +39,27 @@ sleep 2500
 print user write login
 sql connect write
 
-sql_error create user read pass 'taosdata1'
+sql_error create user read2 pass 'taosdata1'
-sql_error create user write pass 'taosdata1'
+sql_error create user write2 pass 'taosdata1'
 sql alter user write pass 'taosdata'
 sql alter user read pass 'taosdata' -x step3
   return -1
 step3:
 
 sql show users
-if $rows != 5 then 
+if $rows != 3 then 
   return -1
 endi
 
 print ============= step4
 sql close 
 sleep 2500
-print root write login
+print user root login
 sql connect
 sql create user oroot pass 'taosdata'
 
 sql show users
-if $rows != 6 then 
+if $rows != 4 then 
   return -1
 endi
 

tests/script/tsim/user/pass_len.sim

@@ -1,7 +1,5 @@
 system sh/stop_dnodes.sh
-
 system sh/deploy.sh -n dnode1 -i 1
-system sh/cfg.sh -n dnode1 -c wallevel -v 0
 system sh/exec.sh -n dnode1 -s start
 sql connect
 
@@ -50,15 +48,16 @@ step3:
 
 sql create user $user PASS 'abc0123456789'
 sql show users
-if $rows != 3 then 
+if $rows != 4 then 
   return -1
 endi
 
 print =============== step4
 $i = 3
 $user = $userPrefix . $i
-sql create user $user PASS 'abcd012345678901234567891234567890' -x step4
+sql create user $user PASS 'abcd012345678901234567891234567890abcd012345678901234567891234567890abcd012345678901234567891234567890abcd012345678901234567891234567890123' -x step4
 	return -1
+
 step4:
 sql show users
 if $rows != 4 then 

tests/script/tsim/user/user_len.sim

@@ -1,11 +1,6 @@
 system sh/stop_dnodes.sh
-
-
 system sh/deploy.sh -n dnode1 -i 1
-system sh/cfg.sh -n dnode1 -c wallevel -v 0
 system sh/exec.sh -n dnode1 -s start
-
-sleep 2000
 sql connect
 
 $i = 0
@@ -24,7 +19,7 @@ sql create user PASS '123' -x step1
 step1:
 
 sql show users
-if $rows != 3 then 
+if $rows != 1 then 
   return -1
 endi
 
@@ -33,13 +28,13 @@ sql drop user a -x step2
 step2:
 sql create user a PASS '123'
 sql show users
-if $rows != 4 then 
+if $rows != 2 then 
   return -1
 endi
 
 sql drop user a
 sql show users
-if $rows != 3 then 
+if $rows != 1 then 
   return -1
 endi
 
@@ -49,13 +44,13 @@ step3:
 
 sql create user abc01234567890123456789 PASS '123'
 sql show users
-if $rows != 4 then 
+if $rows != 2 then 
   return -1
 endi
 
 sql drop user abc01234567890123456789
 sql show users
-if $rows != 3 then 
+if $rows != 1 then 
   return -1
 endi
 
@@ -64,7 +59,7 @@ sql create user abcd0123456789012345678901234567890111 PASS '123' -x step4
 	return -1
 step4:
 sql show users
-if $rows != 3 then 
+if $rows != 1 then 
   return -1
 endi
 
@@ -77,13 +72,13 @@ step61:
 
 sql create user a123 PASS '123'
 sql show users
-if $rows != 4 then 
+if $rows != 2 then 
   return -1
 endi
 
 sql drop user a123
 sql show users
-if $rows != 3 then 
+if $rows != 1 then 
   return -1
 endi