Merge remote-tracking branch 'origin/3.0' into feature/dnode3

Jenkinsfile

@@ -42,7 +42,6 @@ def pre_test(){
     killall -9 gdb || echo "no gdb running"
     killall -9 python3.8 || echo "no python program running"
     cd ${WKC}
-    git reset --hard HEAD~10 >/dev/null
     '''
     script {
       if (env.CHANGE_TARGET == 'master') {
@@ -75,11 +74,9 @@ def pre_test(){
     git pull >/dev/null
     git fetch origin +refs/pull/${CHANGE_ID}/merge
     git checkout -qf FETCH_HEAD
-    git clean -dfx
-    
-    git clean -dfx
     export TZ=Asia/Harbin
     date
+    rm -rf debug
     mkdir debug
     cd debug
     cmake .. > /dev/null

cmake/cmake.options

@@ -44,6 +44,18 @@ option(
     OFF
 )
 
+option(
+    BUILD_WITH_UV
+    "If build with libuv" 
+    OFF
+)
+
+option(
+    BUILD_WITH_CRAFT
+    "If build with canonical-raft"
+    OFF
+)
+
 option(
     BUILD_DEPENDENCY_TESTS 
     "If build dependency tests" 

cmake/craft_CMakeLists.txt.in

@@ -0,0 +1,14 @@
+
+# canonical-raft
+ExternalProject_Add(craft
+        GIT_REPOSITORY https://github.com/canonical/raft.git
+        GIT_TAG v0.11.2
+        SOURCE_DIR "${CMAKE_CONTRIB_DIR}/craft"
+        BINARY_DIR "${CMAKE_CONTRIB_DIR}/craft"
+        #BUILD_IN_SOURCE TRUE
+        # https://answers.ros.org/question/333125/how-to-include-external-automakeautoconf-projects-into-ament_cmake/
+        CONFIGURE_COMMAND COMMAND autoreconf -i COMMAND ./configure --enable-example
+        BUILD_COMMAND "$(MAKE)"
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+)

cmake/libuv_CMakeLists.txt.in

@@ -0,0 +1,12 @@
+
+# libuv
+ExternalProject_Add(libuv
+        GIT_REPOSITORY https://github.com/libuv/libuv.git
+        GIT_TAG v1.42.0
+        SOURCE_DIR "${CMAKE_CONTRIB_DIR}/libuv"
+        BINARY_DIR ""
+        CONFIGURE_COMMAND ""
+        BUILD_COMMAND ""
+        INSTALL_COMMAND ""
+        TEST_COMMAND ""
+        )

contrib/CMakeLists.txt

@@ -34,10 +34,21 @@ if(${BUILD_WITH_ROCKSDB})
     add_definitions(-DUSE_ROCKSDB)
 endif(${BUILD_WITH_ROCKSDB})
 
+# canonical-raft
+if(${BUILD_WITH_CRAFT})
+    cat("${CMAKE_SUPPORT_DIR}/craft_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+    SET(BUILD_WITH_UV ON CACHE BOOL "craft need libuv" FORCE)
+endif(${BUILD_WITH_CRAFT})
+
+#libuv
+if(${BUILD_WITH_UV})
+    cat("${CMAKE_SUPPORT_DIR}/libuv_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
+endif(${BUILD_WITH_UV})
+
 # bdb
 if(${BUILD_WITH_BDB})
     cat("${CMAKE_SUPPORT_DIR}/bdb_CMakeLists.txt.in" ${CONTRIB_TMP_FILE})
-endif(${BUILD_WITH_DBD})
+endif(${BUILD_WITH_BDB})
 
 # sqlite
 if(${BUILD_WITH_SQLITE})
@@ -144,6 +155,23 @@ if(${BUILD_WITH_NURAFT})
     add_subdirectory(nuraft)
 endif(${BUILD_WITH_NURAFT})
 
+# CRAFT
+if(${BUILD_WITH_CRAFT})
+    add_library(craft STATIC IMPORTED GLOBAL)
+    set_target_properties(craft PROPERTIES
+        IMPORTED_LOCATION "${CMAKE_CURRENT_SOURCE_DIR}/craft/.libs/libraft.a"
+        INTERFACE_INCLUDE_DIRECTORIES "${CMAKE_CURRENT_SOURCE_DIR}/craft/include"
+    )
+    # target_link_libraries(craft
+    #     INTERFACE pthread
+    # )
+endif(${BUILD_WITH_CRAFT})
+
+# LIBUV
+if(${BUILD_WITH_UV})
+    add_subdirectory(libuv)
+endif(${BUILD_WITH_UV})
+
 # BDB
 if(${BUILD_WITH_BDB})
     add_library(bdb STATIC IMPORTED GLOBAL)

contrib/test/CMakeLists.txt

@@ -15,4 +15,8 @@ if(${BUILD_WITH_SQLITE})
     add_subdirectory(sqlite)
 endif(${BUILD_WITH_SQLITE})
 
+if(${BUILD_WITH_CRAFT})
+    add_subdirectory(craft)
+endif(${BUILD_WITH_CRAFT})
+
 add_subdirectory(tdev)

contrib/test/craft/CMakeLists.txt

@@ -0,0 +1,2 @@
+add_executable(simulate_vnode "simulate_vnode.c")
+target_link_libraries(simulate_vnode PUBLIC craft lz4 uv_a)

contrib/test/craft/clear.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+
+rm -rf 127.0.0.1*

contrib/test/craft/help.txt

@@ -0,0 +1,15 @@
+
+
+make craft:
+
+sudo apt-get install libuv1-dev liblz4-dev
+autoreconf -i
+./configure --enable-example
+make
+
+
+start:
+
+./simulate_vnode 10000 10001 10002
+./simulate_vnode 10001 10000 10002
+./simulate_vnode 10002 10000 10001

contrib/test/craft/simulate_vnode.c

@@ -0,0 +1,252 @@
+#include <stdio.h>
+#include <raft.h>
+#include <raft/uv.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <string.h>
+#include <time.h>
+#include <stdlib.h>
+#include <getopt.h>
+
+const char* exe_name;
+
+// simulate ------------------------
+typedef struct SVnode {
+    int vid;
+} SVnode;
+
+
+#define VNODE_COUNT 10
+SVnode vnodes[VNODE_COUNT];
+
+int vnodeApplyWMsg(SVnode *pVnode, char *pMsg, void **pRsp) {
+	printf("put value to tsdb, vid:%d msg:%s \n", pVnode->vid, pMsg);
+	return 0;
+}
+
+int applyCB(struct raft_fsm *fsm,
+            const struct raft_buffer *buf,
+            void **result) {
+	char *msg = (char*)buf->base;
+	//printf("%s \n", msg);
+
+	// parse msg
+	char* context;
+    char* token = strtok_r(msg, ":", &context);
+	int vid = atoi(token);
+
+	token = strtok_r(NULL, ":", &context);
+	char *value = token;
+
+	SVnode* tmp_vnodes = (SVnode*)(fsm->data);
+	vnodeApplyWMsg(&tmp_vnodes[vid], value, NULL);
+
+    return 0;
+}
+
+// Config ------------------------
+#define HOST_LEN 32
+#define MAX_PEERS 10
+typedef struct Address {
+	char host[HOST_LEN];
+	uint32_t port;
+} Address;
+
+uint64_t raftId(Address *addr) {
+	// test in a single machine, port is unique
+	// if in multi machines, use host and port
+	return addr->port;
+}
+
+typedef struct Config {
+	Address me;
+	Address peers[MAX_PEERS];
+    int peer_count;
+} Config;
+
+Config gConf;
+
+void printConf(Config *c) {
+	printf("me: %s:%u \n", c->me.host, c->me.port);
+	for (int i = 0; i < c->peer_count; ++i) {
+		printf("peer%d: %s:%u \n", i, c->peers[i].host, c->peers[i].port);
+	}
+}
+
+// RaftServer ------------------------
+typedef struct RaftServer {
+	struct uv_loop_s loop;
+	struct raft_uv_transport transport;
+	struct raft_io io;
+	struct raft_fsm fsm;
+	struct raft raft;
+	struct raft_configuration conf;
+} RaftServer;
+
+RaftServer gRaftServer;
+
+static void* startRaftServer(void *param) {
+	//RaftServer* rs = (RaftServer*)param;
+	RaftServer* rs = &gRaftServer;
+    raft_start(&rs->raft);
+    uv_run(&rs->loop, UV_RUN_DEFAULT);
+}
+
+static const char* state2String(unsigned short state) {
+    if (state == RAFT_UNAVAILABLE) {
+        return "RAFT_UNAVAILABLE";
+
+    } else if (state == RAFT_FOLLOWER) {
+        return "RAFT_FOLLOWER";
+
+    } else if (state == RAFT_CANDIDATE) {
+        return "RAFT_CANDIDATE";
+
+    } else if (state == RAFT_LEADER) {
+        return "RAFT_LEADER";
+
+    }
+    return "UNKNOWN_RAFT_STATE";
+}
+
+static void printRaftState(struct raft *r) {
+	printf("\n");
+	printf("my_id: %llu \n", r->id);
+	printf("address: %s \n", r->address);
+	printf("current_term: %llu \n", r->current_term);
+	printf("voted_for: %llu \n", r->voted_for);
+	printf("role: %s \n", state2String(r->state));
+	printf("commit_index: %llu \n", r->commit_index);
+	printf("last_applied: %llu \n", r->last_applied);
+	printf("last_stored: %llu \n", r->last_stored);
+	printf("\n");
+}
+
+// console -----------------------------------------
+#define PROPOSE_VALUE_LEN 128
+static void proposeValue(struct raft *r) {
+    struct raft_buffer buf;
+
+    // need free
+    buf.len = PROPOSE_VALUE_LEN;
+    buf.base = raft_malloc(buf.len);
+
+    // mock ts value
+    int vid = rand() % VNODE_COUNT;
+    snprintf(buf.base, buf.len, "%d:value_%ld", vid, time(NULL));
+
+	printf("propose value: %s \n", (char*)buf.base);
+
+    // need free
+    struct raft_apply *req = raft_malloc(sizeof(struct raft_apply));
+    raft_apply(r, req, &buf, 1, NULL);
+}
+
+static void* console(void *param) {
+    while (1) {
+        // notice! memory buffer overflow!
+        char buf[128];
+		memset(buf, 0, sizeof(buf));
+		fgets(buf, 128, stdin);
+		if (strlen(buf) == 1) {
+			continue;
+		}	
+		buf[strlen(buf)-1] = '\0';
+
+        // do not use strcmp
+        if (strcmp(buf, "state") == 0) {
+            printRaftState(&gRaftServer.raft);
+
+        } else if (strcmp(buf, "put") == 0) {
+            proposeValue(&gRaftServer.raft);
+
+        } else {
+            printf("unknown command: [%s], support command: state, put \n", buf);
+        }
+    }
+}
+
+// -----------------------------------------
+void usage() {
+	printf("\n");
+	printf("%s my_port peer1_port peer2_port ... \n", exe_name);
+	printf("\n");
+}
+
+int main(int argc, char **argv) {
+	srand(time(NULL));
+
+	exe_name = argv[0];
+	if (argc < 2) {
+        usage();
+        exit(-1);
+    }
+
+	// read conf from argv
+	strncpy(gConf.me.host, "127.0.0.1", HOST_LEN);
+	sscanf(argv[1], "%u", &gConf.me.port);
+
+	gConf.peer_count = 0;
+	for (int i = 2; i < argc; ++i) {
+		strncpy(gConf.peers[gConf.peer_count].host, "127.0.0.1", HOST_LEN);
+		sscanf(argv[i], "%u", &gConf.peers[gConf.peer_count].port);
+		gConf.peer_count++;
+	}
+	printConf(&gConf);
+
+	// mkdir
+	char dir[128];
+	snprintf(dir, sizeof(dir), "./%s_%u", gConf.me.host, gConf.me.port);
+
+	char cmd[128];
+	snprintf(cmd, sizeof(cmd), "rm -rf ./%s", dir);
+	system(cmd);
+	snprintf(cmd, sizeof(cmd), "mkdir -p ./%s", dir);
+	system(cmd);
+
+    // init io
+    uv_loop_init(&gRaftServer.loop);
+    raft_uv_tcp_init(&gRaftServer.transport, &gRaftServer.loop);
+    raft_uv_init(&gRaftServer.io, &gRaftServer.loop, dir, &gRaftServer.transport);
+
+	// init fsm
+	gRaftServer.fsm.apply = applyCB;
+	gRaftServer.fsm.data = vnodes;
+	for (int i = 0; i < VNODE_COUNT; ++i) {
+		vnodes[i].vid = i;
+	}
+
+    // init raft instance with io and fsm
+	char address_buf[128];
+	snprintf(address_buf, sizeof(address_buf), "%s:%u", gConf.me.host, gConf.me.port);
+	
+	// test in a single machine, port is unique
+	uint64_t raft_id = raftId(&gConf.me);
+    raft_init(&gRaftServer.raft, &gRaftServer.io, &gRaftServer.fsm, raft_id, address_buf);
+    //raft_init(&gRaftServer.raft, &gRaftServer.io, &gRaftServer.fsm, 11, "127.0.0.1:9000");
+
+    // init cluster configuration
+	struct raft_configuration conf;
+    raft_configuration_init(&conf);
+    raft_configuration_add(&conf, raftId(&gConf.me), address_buf, RAFT_VOTER);
+	for (int i = 0; i < gConf.peer_count; ++i) {
+		char address_buf[128];
+		snprintf(address_buf, sizeof(address_buf), "%s:%u", gConf.peers[i].host, gConf.peers[i].port);
+		raft_configuration_add(&conf, raftId(&gConf.peers[i]), address_buf, RAFT_VOTER);
+	}
+    raft_bootstrap(&gRaftServer.raft, &conf);
+
+    // start raft server and loop
+    pthread_t tid;
+    pthread_create(&tid, NULL, startRaftServer, &gRaftServer);
+
+    // simulate console
+	pthread_t tid2;
+	pthread_create(&tid2, NULL, console, NULL);
+
+	while (1) {
+		sleep(10);
+	}
+
+	return 0;
+}

include/common/taosmsg.h

@@ -50,7 +50,8 @@ TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_CONSUME, "mq-consume" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_QUERY, "mq-query" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_CONNECT, "mq-connect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_DISCONNECT, "mq-disconnect" )
-TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_SET, "mq-set" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_MQ_SET_CUR, "mq-set-cur" )
+TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_RSP_READY, "rsp-ready" )
 // message from client to mnode
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CONNECT, "connect" )
 TAOS_DEFINE_MESSAGE_TYPE( TSDB_MSG_TYPE_CREATE_ACCT, "create-acct" )	
@@ -223,6 +224,7 @@ typedef struct SBuildUseDBInput {
   int32_t vgVersion;
 } SBuildUseDBInput;
 
+
 #pragma pack(push, 1)
 
 // null-terminated string instead of char array to avoid too many memory consumption in case of more than 1M tableMeta
@@ -232,9 +234,9 @@ typedef struct {
 } SEpAddrMsg;
 
 typedef struct {
-  char    *fqdn;
+  char     fqdn[TSDB_FQDN_LEN];
   uint16_t port;
-} SEpAddr1;
+} SEpAddr;
 
 typedef struct {
   int32_t numOfVnodes;
@@ -341,7 +343,7 @@ typedef struct {
 typedef struct {
   char    tableFname[TSDB_TABLE_FNAME_LEN];
   char    db[TSDB_FULL_DB_NAME_LEN];
-  int16_t type;      /* operation type   */
+  int16_t type; /* operation type   */
   int16_t numOfCols; /* number of schema */
   int32_t tagValLen;
   SSchema schema[];
@@ -543,8 +545,8 @@ typedef struct {
   int32_t     sqlstrLen;      // sql query string
   int32_t     prevResultLen;  // previous result length
   int32_t     numOfOperator;
-  int32_t     tableScanOperator;  // table scan operator. -1 means no scan operator
+  int32_t     tableScanOperator;// table scan operator. -1 means no scan operator
-  int32_t     udfNum;             // number of udf function
+  int32_t     udfNum;           // number of udf function
   int32_t     udfContentOffset;
   int32_t     udfContentLen;
   SColumnInfo tableCols[];
@@ -1004,27 +1006,35 @@ typedef struct {
 
 // mq related
 typedef struct {
+
 } SMqConnectReq;
 
 typedef struct {
+
 } SMqConnectRsp;
 
 typedef struct {
+
 } SMqDisconnectReq;
 
 typedef struct {
+
 } SMqDisconnectRsp;
 
 typedef struct {
+
 } SMqAckReq;
 
 typedef struct {
+
 } SMqAckRsp;
 
 typedef struct {
+
 } SMqResetReq;
 
 typedef struct {
+
 } SMqResetRsp;
 // mq related end
 
@@ -1080,6 +1090,24 @@ typedef struct {
   /* data */
 } SUpdateTagValRsp;
 
+typedef struct SSchedulerQueryMsg {
+  uint64_t  queryId;
+  uint64_t  taskId;
+  uint32_t  contentLen;
+  char      msg[];
+} SSchedulerQueryMsg;
+
+typedef struct SSchedulerReadyMsg {
+  uint64_t  queryId;
+  uint64_t  taskId;
+} SSchedulerReadyMsg;
+
+typedef struct SSchedulerFetchMsg {
+  uint64_t  queryId;
+  uint64_t  taskId;
+} SSchedulerFetchMsg;
+
+
 #pragma pack(pop)
 
 #ifdef __cplusplus

include/common/tmsgtype.h

@@ -35,7 +35,7 @@ enum {
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_SELECT, "select" )
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_FETCH, "fetch" )
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_INSERT, "insert" )
-  TSDB_DEFINE_SQL_TYPE( TSDB_SQL_UPDATE_TAGS_VAL, "update-tag-val" )
+  TSDB_DEFINE_SQL_TYPE( TSDB_SQL_UPDATE_TAG_VAL, "update-tag-val" )
 
   // the SQL below is for mgmt node
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_MGMT, "mgmt" )
@@ -54,7 +54,7 @@ enum {
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_ALTER_TABLE, "alter-table" )
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_ALTER_DB, "alter-db" )
  
-  TSDB_DEFINE_SQL_TYPE(TSDB_SQL_SYNC_DB_REPLICA, "sync db-replica")
+  TSDB_DEFINE_SQL_TYPE( TSDB_SQL_SYNC_DB_REPLICA, "sync db-replica")
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_CREATE_MNODE, "create-mnode" )
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_DROP_MNODE, "drop-mnode" )
   TSDB_DEFINE_SQL_TYPE( TSDB_SQL_CREATE_DNODE, "create-dnode" )

include/dnode/vnode/tq/tq.h

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

include/libs/catalog/catalog.h

@@ -137,7 +137,7 @@ int32_t catalogGetTableHashVgroup(struct SCatalog* pCatalog, void *pRpc, const S
 int32_t catalogGetAllMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const SCatalogReq* pReq, SMetaData* pRsp);
 
 
-int32_t catalogGetQnodeList(struct SCatalog* pCatalog, const SEpSet* pMgmtEps, SEpSet* pQnodeEpSet);
+int32_t catalogGetQnodeList(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, SEpSet* pQnodeEpSet);
 
 
 

include/libs/parser/parser.h

@@ -209,6 +209,7 @@ typedef struct SSourceParam {
 
 SExprInfo* createExprInfo(STableMetaInfo* pTableMetaInfo, const char* funcName, SSourceParam* pSource, SSchema* pResSchema, int16_t interSize);
 int32_t copyExprInfoList(SArray* dst, const SArray* src, uint64_t uid, bool deepcopy);
+int32_t copyAllExprInfo(SArray* dst, const SArray* src, bool deepcopy);
 int32_t getExprFunctionLevel(SQueryStmtInfo* pQueryInfo);
 
 STableMetaInfo* getMetaInfo(SQueryStmtInfo* pQueryInfo, int32_t tableIndex);

include/libs/planner/planner.h

@@ -141,8 +141,8 @@ int32_t qCreateQueryDag(const struct SQueryStmtInfo* pQueryInfo, struct SEpSet*
 // Set datasource of this subplan, multiple calls may be made to a subplan.
 // @subplan subplan to be schedule
 // @templateId templateId of a group of datasource subplans of this @subplan
-// @eps Execution location of this group of datasource subplans, is an array of SEpAddr structures 
+// @ep one execution location of this group of datasource subplans 
-int32_t qSetSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SArray* eps);
+int32_t qSetSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SEpAddr* ep);
 
 int32_t qExplainQuery(const struct SQueryStmtInfo* pQueryInfo, struct SEpSet* pQnode, char** str);
 

include/libs/scheduler/scheduler.h

@@ -21,9 +21,10 @@ extern "C" {
 #endif
 
 #include "planner.h"
+#include "catalog.h"
 
 typedef struct SSchedulerCfg {
-
+  int32_t clusterType;
 } SSchedulerCfg;
 
 typedef struct SQueryProfileSummary {
@@ -49,15 +50,17 @@ typedef struct SQueryProfileSummary {
   uint64_t resultSize;   // generated result size in Kb.
 } SQueryProfileSummary;
 
+int32_t schedulerInit(SSchedulerCfg *cfg);
+
 /**
  * Process the query job, generated according to the query physical plan.
  * This is a synchronized API, and is also thread-safety.
- * @param pJob
+ * @param qnodeList  Qnode address list, element is SEpAddr
  * @return
  */
-int32_t scheduleQueryJob(SQueryDag* pDag, void** pJob);
+int32_t scheduleExecJob(void *transport, SArray *qnodeList, SQueryDag* pDag, void** pJob);
 
-int32_t scheduleFetchRows(void *pJob, void *data);
+int32_t scheduleFetchRows(void *pJob, void **data);
 
 
 /**
@@ -69,6 +72,8 @@ int32_t scheduleCancelJob(void *pJob);
 
 void scheduleFreeJob(void *pJob);
 
+void schedulerDestroy(void);
+
 #ifdef __cplusplus
 }
 #endif

include/util/taoserror.h

@@ -339,6 +339,20 @@ int32_t* taosGetErrno();
 #define TSDB_CODE_SYN_INVALID_MSGLEN            TAOS_DEF_ERROR_CODE(0, 0x0909)  //"Invalid msg length")
 #define TSDB_CODE_SYN_INVALID_MSGTYPE           TAOS_DEF_ERROR_CODE(0, 0x090A)  //"Invalid msg type")
 
+// tq
+#define TSDB_CODE_TQ_INVALID_CONFIG             TAOS_DEF_ERROR_CODE(0, 0x0A00)  //"Invalid configuration")
+#define TSDB_CODE_TQ_INIT_FAILED                TAOS_DEF_ERROR_CODE(0, 0x0A01)  //"Tq init failed")
+#define TSDB_CODE_TQ_NO_DISKSPACE               TAOS_DEF_ERROR_CODE(0, 0x0A02)  //"No diskspace for tq")
+#define TSDB_CODE_TQ_NO_DISK_PERMISSIONS        TAOS_DEF_ERROR_CODE(0, 0x0A03)  //"No permission for disk files")
+#define TSDB_CODE_TQ_FILE_CORRUPTED             TAOS_DEF_ERROR_CODE(0, 0x0A04)  //"Data file(s) corrupted")
+#define TSDB_CODE_TQ_OUT_OF_MEMORY              TAOS_DEF_ERROR_CODE(0, 0x0A05)  //"Out of memory")
+#define TSDB_CODE_TQ_FILE_ALREADY_EXISTS        TAOS_DEF_ERROR_CODE(0, 0x0A06)  //"File already exists")
+#define TSDB_CODE_TQ_FAILED_TO_CREATE_DIR       TAOS_DEF_ERROR_CODE(0, 0x0A07)  //"Failed to create dir")
+#define TSDB_CODE_TQ_META_NO_SUCH_KEY           TAOS_DEF_ERROR_CODE(0, 0x0A08)  //"Target key not found")
+#define TSDB_CODE_TQ_META_KEY_NOT_IN_TXN        TAOS_DEF_ERROR_CODE(0, 0x0A09)  //"Target key not in transaction")
+#define TSDB_CODE_TQ_META_KEY_DUP_IN_TXN        TAOS_DEF_ERROR_CODE(0, 0x0A0A)  //"Target key duplicated in transaction")
+#define TSDB_CODE_TQ_GROUP_NOT_SET              TAOS_DEF_ERROR_CODE(0, 0x0A0B)  //"Group of corresponding client is not set by mnode")
+
 // wal
 #define TSDB_CODE_WAL_APP_ERROR                 TAOS_DEF_ERROR_CODE(0, 0x1000)  //"Unexpected generic error in wal")
 #define TSDB_CODE_WAL_FILE_CORRUPTED            TAOS_DEF_ERROR_CODE(0, 0x1001)  //"WAL file is corrupted")
@@ -493,6 +507,7 @@ int32_t* taosGetErrno();
 
 //scheduler
 #define TSDB_CODE_SCH_STATUS_ERROR              TAOS_DEF_ERROR_CODE(0, 0x2501)  //scheduler status error
+#define TSDB_CODE_SCH_INTERNAL_ERROR            TAOS_DEF_ERROR_CODE(0, 0x2502)  //scheduler internal error
 
 
 

include/util/tlog.h

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

source/client/src/clientImpl.c

@@ -153,7 +153,7 @@ TAOS_RES *taos_query_l(TAOS *taos, const char *sql, int sqlLen) {
     void*   output = NULL;
     int32_t outputLen = 0;
     code = qParseQuerySql(pRequest->sqlstr, sqlLen, pRequest->requestId, &type, &output, &outputLen, pRequest->msgBuf, ERROR_MSG_BUF_DEFAULT_SIZE);
-    if (type == TSDB_SQL_CREATE_USER || type == TSDB_SQL_SHOW || type == TSDB_SQL_DROP_USER || type == TSDB_SQL_CREATE_DB) {
+    if (type == TSDB_SQL_CREATE_USER || type == TSDB_SQL_SHOW || type == TSDB_SQL_DROP_USER || type == TSDB_SQL_DROP_ACCT || type == TSDB_SQL_CREATE_DB || type == TSDB_SQL_CREATE_ACCT) {
       pRequest->type = type;
       pRequest->body.param = output;
       pRequest->body.paramLen = outputLen;

source/client/src/clientMsgHandler.c

@@ -407,7 +407,7 @@ void tscProcessMsgFromServer(SRpcMsg *rpcMsg, SRpcEpSet *pEpSet) {
   }
 
   // single table query error need to be handled here.
-  if ((cmd == TSDB_SQL_SELECT || cmd == TSDB_SQL_UPDATE_TAGS_VAL) &&
+  if ((cmd == TSDB_SQL_SELECT || cmd == TSDB_SQL_UPDATE_TAG_VAL) &&
       (((rpcMsg->code == TSDB_CODE_TDB_INVALID_TABLE_ID || rpcMsg->code == TSDB_CODE_VND_INVALID_VGROUP_ID)) ||
        rpcMsg->code == TSDB_CODE_RPC_NETWORK_UNAVAIL || rpcMsg->code == TSDB_CODE_APP_NOT_READY)) {
 
@@ -3176,6 +3176,15 @@ int32_t doBuildMsgSupp(SRequestObj *pRequest, SRequestMsgBody* pMsgBody) {
     case TSDB_SQL_CREATE_USER:
       pMsgBody->msgType = TSDB_MSG_TYPE_CREATE_USER;
       break;
+    case TSDB_SQL_DROP_USER:
+      pMsgBody->msgType = TSDB_MSG_TYPE_DROP_USER;
+      break;
+    case TSDB_SQL_CREATE_ACCT:
+      pMsgBody->msgType = TSDB_MSG_TYPE_CREATE_ACCT;
+      break;
+    case TSDB_SQL_DROP_ACCT:
+      pMsgBody->msgType = TSDB_MSG_TYPE_DROP_ACCT;
+      break;
     case TSDB_SQL_CREATE_DB: {
       pMsgBody->msgType = TSDB_MSG_TYPE_CREATE_DB;
 
@@ -3328,7 +3337,7 @@ void initMsgHandleFp() {
   tscBuildMsg[TSDB_SQL_DROP_DNODE] = tscBuildDropDnodeMsg;
   tscBuildMsg[TSDB_SQL_CFG_DNODE] = tscBuildCfgDnodeMsg;
   tscBuildMsg[TSDB_SQL_ALTER_TABLE] = tscBuildAlterTableMsg;
-  tscBuildMsg[TSDB_SQL_UPDATE_TAGS_VAL] = tscBuildUpdateTagMsg;
+  tscBuildMsg[TSDB_SQL_UPDATE_TAG_VAL] = tscBuildUpdateTagMsg;
   tscBuildMsg[TSDB_SQL_ALTER_DB] = tscAlterDbMsg;
   tscBuildMsg[TSDB_SQL_COMPACT_VNODE] = tscBuildCompactMsg;
 
@@ -3383,6 +3392,10 @@ void initMsgHandleFp() {
   handleRequestRspFp[TSDB_SQL_CONNECT] = processConnectRsp;
 
   buildRequestMsgFp[TSDB_SQL_CREATE_USER]  = doBuildMsgSupp;
+  buildRequestMsgFp[TSDB_SQL_DROP_USER]    = doBuildMsgSupp;
+
+  buildRequestMsgFp[TSDB_SQL_CREATE_ACCT]  = doBuildMsgSupp;
+  buildRequestMsgFp[TSDB_SQL_DROP_ACCT]    = doBuildMsgSupp;
 
   buildRequestMsgFp[TSDB_SQL_SHOW]         = doBuildMsgSupp;
   handleRequestRspFp[TSDB_SQL_SHOW]        = processShowRsp;

source/client/src/tscEnv.c

@@ -189,7 +189,6 @@ static void doDestroyRequest(void* p) {
   tfree(pRequest->pInfo);
 
   if (pRequest->body.pResInfo != NULL) {
-    tfree(pRequest->body.pResInfo->pData);
     tfree(pRequest->body.pResInfo->pMsg);
     tfree(pRequest->body.pResInfo);
   }

source/client/test/clientTests.cpp

@@ -40,13 +40,13 @@ TEST(testCase, driverInit_Test) {
 
 TEST(testCase, connect_Test) {
   TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
-  assert(pConn != NULL);
+//  assert(pConn != NULL);
   taos_close(pConn);
 }
 
 TEST(testCase, create_user_Test) {
   TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
-  assert(pConn != NULL);
+//  assert(pConn != NULL);
 
   TAOS_RES* pRes = taos_query(pConn, "create user abc pass 'abc'");
   if (taos_errno(pRes) != TSDB_CODE_SUCCESS) {
@@ -57,22 +57,35 @@ TEST(testCase, create_user_Test) {
   taos_close(pConn);
 }
 
-//TEST(testCase, drop_user_Test) {
+TEST(testCase, create_account_Test) {
-//  TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
+  TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
-//  assert(pConn != NULL);
+  assert(pConn != NULL);
-//
+
-//  TAOS_RES* pRes = taos_query(pConn, "drop user abc");
+  TAOS_RES* pRes = taos_query(pConn, "create account aabc pass 'abc'");
-//  if (taos_errno(pRes) != TSDB_CODE_SUCCESS) {
+  if (taos_errno(pRes) != TSDB_CODE_SUCCESS) {
-//    printf("failed to create user, reason:%s\n", taos_errstr(pRes));
+    printf("failed to create user, reason:%s\n", taos_errstr(pRes));
-//  }
+  }
-//
+
-//  taos_free_result(pRes);
+  taos_free_result(pRes);
-//  taos_close(pConn);
+  taos_close(pConn);
-//}
+}
+
+TEST(testCase, drop_account_Test) {
+  TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
+  assert(pConn != NULL);
+
+  TAOS_RES* pRes = taos_query(pConn, "drop account aabc");
+  if (taos_errno(pRes) != TSDB_CODE_SUCCESS) {
+    printf("failed to create user, reason:%s\n", taos_errstr(pRes));
+  }
+
+  taos_free_result(pRes);
+  taos_close(pConn);
+}
 
 TEST(testCase, show_user_Test) {
   TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
-  assert(pConn != NULL);
+//  assert(pConn != NULL);
 
   TAOS_RES* pRes = taos_query(pConn, "show users");
   TAOS_ROW pRow = NULL;
@@ -89,10 +102,23 @@ TEST(testCase, show_user_Test) {
   taos_close(pConn);
 }
 
-TEST(testCase, show_db_Test) {
+TEST(testCase, drop_user_Test) {
   TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
   assert(pConn != NULL);
 
+  TAOS_RES* pRes = taos_query(pConn, "drop user abc");
+  if (taos_errno(pRes) != TSDB_CODE_SUCCESS) {
+    printf("failed to create user, reason:%s\n", taos_errstr(pRes));
+  }
+
+  taos_free_result(pRes);
+  taos_close(pConn);
+}
+
+TEST(testCase, show_db_Test) {
+  TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
+//  assert(pConn != NULL);
+
   TAOS_RES* pRes = taos_query(pConn, "show databases");
   TAOS_ROW pRow = NULL;
 
@@ -112,7 +138,7 @@ TEST(testCase, create_db_Test) {
   TAOS* pConn = taos_connect("ubuntu", "root", "taosdata", NULL, 0);
   assert(pConn != NULL);
 
-  TAOS_RES* pRes = taos_query(pConn, "create database abc");
+  TAOS_RES* pRes = taos_query(pConn, "create database abc1");
 
   TAOS_FIELD* pFields = taos_fetch_fields(pRes);
   ASSERT_TRUE(pFields == NULL);

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

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

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

@@ -16,17 +16,13 @@
 #include "vnodeDef.h"
 
 int vnodeProcessNoWalWMsgs(SVnode *pVnode, SRpcMsg *pMsg) {
-  SVnodeReq *pVnodeReq;
-
   switch (pMsg->msgType) {
-    case TSDB_MSG_TYPE_MQ_SET:
+    case TSDB_MSG_TYPE_MQ_SET_CUR:
       if (tqSetCursor(pVnode->pTq, pMsg->pCont) < 0) {
         // TODO: handle error
       }
       break;
   }
-
-  void *pBuf = pMsg->pCont;
   return 0;
 }
 

source/dnode/vnode/tq/CMakeLists.txt

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

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

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

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

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

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

@@ -24,89 +24,80 @@
 // handle management message
 //
 
-int tqGetgHandleSSize(const STqGroupHandle* gHandle);
+int tqGroupSSize(const STqGroup* pGroup);
-int tqBufHandleSSize();
+int tqTopicSSize();
-int tqBufItemSSize();
+int tqItemSSize();
 
-STqGroupHandle* tqFindHandle(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+void* tqSerializeListHandle(STqList* listHandle, void* ptr);
-  STqGroupHandle* gHandle;
+void* tqSerializeTopic(STqTopic* pTopic, void* ptr);
-  return NULL;
+void* tqSerializeItem(STqMsgItem* pItem, void* ptr);
-}
 
-void* tqSerializeListHandle(STqListHandle* listHandle, void* ptr);
+const void* tqDeserializeTopic(const void* pBytes, STqTopic* pTopic);
-void* tqSerializeBufHandle(STqBufferHandle* bufHandle, void* ptr);
+const void* tqDeserializeItem(const void* pBytes, STqMsgItem* pItem);
-void* tqSerializeBufItem(STqBufferItem* bufItem, void* ptr);
-
-const void* tqDeserializeBufHandle(const void* pBytes, STqBufferHandle* bufHandle);
-const void* tqDeserializeBufItem(const void* pBytes, STqBufferItem* bufItem);
 
 STQ* tqOpen(const char* path, STqCfg* tqConfig, STqLogReader* tqLogReader, SMemAllocatorFactory* allocFac) {
   STQ* pTq = malloc(sizeof(STQ));
   if (pTq == NULL) {
-    // TODO: memory error
+    terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
     return NULL;
   }
   pTq->path = strdup(path);
   pTq->tqConfig = tqConfig;
   pTq->tqLogReader = tqLogReader;
   pTq->tqMemRef.pAlloctorFactory = allocFac;
-  // pTq->tqMemRef.pAllocator = allocFac->create(allocFac);
+  pTq->tqMemRef.pAllocator = allocFac->create(allocFac);
   if (pTq->tqMemRef.pAllocator == NULL) {
-    // TODO
+    // TODO: error code of buffer pool
   }
-  pTq->tqMeta =
+  pTq->tqMeta = tqStoreOpen(path, (FTqSerialize)tqSerializeGroup, (FTqDeserialize)tqDeserializeGroup, free, 0);
-      tqStoreOpen(path, (FTqSerialize)tqSerializeGroupHandle, (FTqDeserialize)tqDeserializeGroupHandle, free, 0);
   if (pTq->tqMeta == NULL) {
     // TODO: free STQ
     return NULL;
   }
   return pTq;
 }
-
+void tqClose(STQ* pTq) {
-void tqClose(STQ*pTq) {
   // TODO
 }
 
-static int tqProtoCheck(TmqMsgHead *pMsg) {
+static int tqProtoCheck(STqMsgHead* pMsg) { return pMsg->protoVer == 0; }
-  return pMsg->protoVer == 0;
-}
 
-static int tqAckOneTopic(STqBufferHandle* bHandle, TmqOneAck* pAck, STqQueryMsg** ppQuery) {
+static int tqAckOneTopic(STqTopic* pTopic, STqOneAck* pAck, STqQueryMsg** ppQuery) {
   // clean old item and move forward
   int32_t consumeOffset = pAck->consumeOffset;
   int     idx = consumeOffset % TQ_BUFFER_SIZE;
-  ASSERT(bHandle->buffer[idx].content && bHandle->buffer[idx].executor);
+  ASSERT(pTopic->buffer[idx].content && pTopic->buffer[idx].executor);
-  tfree(bHandle->buffer[idx].content);
+  tfree(pTopic->buffer[idx].content);
   if (1 /* TODO: need to launch new query */) {
     STqQueryMsg* pNewQuery = malloc(sizeof(STqQueryMsg));
     if (pNewQuery == NULL) {
-      // TODO: memory insufficient
+      terrno = TSDB_CODE_TQ_OUT_OF_MEMORY;
       return -1;
     }
     // TODO: lock executor
-    pNewQuery->exec->executor = bHandle->buffer[idx].executor;
     // TODO: read from wal and assign to src
-    pNewQuery->exec->src = 0;
+    /*pNewQuery->exec->executor = pTopic->buffer[idx].executor;*/
-    pNewQuery->exec->dest = &bHandle->buffer[idx];
+    /*pNewQuery->exec->src = 0;*/
-    pNewQuery->next = *ppQuery;
+    /*pNewQuery->exec->dest = &pTopic->buffer[idx];*/
-    *ppQuery = pNewQuery;
+    /*pNewQuery->next = *ppQuery;*/
+    /**ppQuery = pNewQuery;*/
   }
   return 0;
 }
 
-static int tqAck(STqGroupHandle* gHandle, TmqAcks* pAcks) {
+static int tqAck(STqGroup* pGroup, STqAcks* pAcks) {
   int32_t    ackNum = pAcks->ackNum;
-  TmqOneAck* acks = pAcks->acks;
+  STqOneAck* acks = pAcks->acks;
   // double ptr for acks and list
-  int            i = 0;
+  int          i = 0;
-  STqListHandle* node = gHandle->head;
+  STqList*     node = pGroup->head;
-  int            ackCnt = 0;
+  int          ackCnt = 0;
-  STqQueryMsg*   pQuery = NULL;
+  STqQueryMsg* pQuery = NULL;
   while (i < ackNum && node->next) {
-    if (acks[i].topicId == node->next->bufHandle.topicId) {
+    if (acks[i].topicId == node->next->topic.topicId) {
       ackCnt++;
-      tqAckOneTopic(&node->next->bufHandle, &acks[i], &pQuery);
+      tqAckOneTopic(&node->next->topic, &acks[i], &pQuery);
-    } else if (acks[i].topicId < node->next->bufHandle.topicId) {
+    } else if (acks[i].topicId < node->next->topic.topicId) {
       i++;
     } else {
       node = node->next;
@@ -118,52 +109,56 @@ static int tqAck(STqGroupHandle* gHandle, TmqAcks* pAcks) {
   return ackCnt;
 }
 
-static int tqCommitTCGroup(STqGroupHandle* handle) {
+static int tqCommitGroup(STqGroup* pGroup) {
   // persist modification into disk
   return 0;
 }
 
-int tqCreateTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId, STqGroupHandle** handle) {
+int tqCreateGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId, STqGroup** ppGroup) {
   // create in disk
-  STqGroupHandle* gHandle = (STqGroupHandle*)malloc(sizeof(STqGroupHandle));
+  STqGroup* pGroup = (STqGroup*)malloc(sizeof(STqGroup));
-  if (gHandle == NULL) {
+  if (pGroup == NULL) {
     // TODO
     return -1;
   }
-  memset(gHandle, 0, sizeof(STqGroupHandle));
+  *ppGroup = pGroup;
+  memset(pGroup, 0, sizeof(STqGroup));
 
   return 0;
 }
 
-STqGroupHandle* tqOpenTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+STqGroup* tqOpenGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
-  STqGroupHandle* gHandle = tqHandleGet(pTq->tqMeta, cId);
+  STqGroup* pGroup = tqHandleGet(pTq->tqMeta, cId);
-  if (gHandle == NULL) {
+  if (pGroup == NULL) {
-    int code = tqCreateTCGroup(pTq, topicId, cgId, cId, &gHandle);
+    int code = tqCreateGroup(pTq, topicId, cgId, cId, &pGroup);
-    if (code != 0) {
+    if (code < 0) {
       // TODO
       return NULL;
     }
+    tqHandleMovePut(pTq->tqMeta, cId, pGroup);
   }
+  ASSERT(pGroup);
 
-  // create
+  return pGroup;
-  // open
-  return gHandle;
 }
 
-int tqCloseTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) { return 0; }
+int tqCloseGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+  // TODO
+  return 0;
+}
 
-int tqDropTCGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
+int tqDropGroup(STQ* pTq, int64_t topicId, int64_t cgId, int64_t cId) {
   // delete from disk
   return 0;
 }
 
-static int tqFetch(STqGroupHandle* gHandle, void** msg) {
+static int tqFetch(STqGroup* pGroup, void** msg) {
-  STqListHandle* head = gHandle->head;
+  STqList* head = pGroup->head;
-  STqListHandle* node = head;
+  STqList* node = head;
-  int            totSize = 0;
+  int      totSize = 0;
   // TODO: make it a macro
   int            sizeLimit = 4 * 1024;
-  TmqMsgContent* buffer = malloc(sizeLimit);
+  STqMsgContent* buffer = malloc(sizeLimit);
   if (buffer == NULL) {
     // TODO:memory insufficient
     return -1;
@@ -172,25 +167,25 @@ static int tqFetch(STqGroupHandle* gHandle, void** msg) {
   // until all topic iterated or msgs over sizeLimit
   while (node->next) {
     node = node->next;
-    STqBufferHandle* bufHandle = &node->bufHandle;
+    STqTopic* topicHandle = &node->topic;
-    int              idx = bufHandle->nextConsumeOffset % TQ_BUFFER_SIZE;
+    int       idx = topicHandle->nextConsumeOffset % TQ_BUFFER_SIZE;
-    if (bufHandle->buffer[idx].content != NULL && bufHandle->buffer[idx].offset == bufHandle->nextConsumeOffset) {
+    if (topicHandle->buffer[idx].content != NULL && topicHandle->buffer[idx].offset == topicHandle->nextConsumeOffset) {
-      totSize += bufHandle->buffer[idx].size;
+      totSize += topicHandle->buffer[idx].size;
       if (totSize > sizeLimit) {
         void* ptr = realloc(buffer, totSize);
         if (ptr == NULL) {
-          totSize -= bufHandle->buffer[idx].size;
+          totSize -= topicHandle->buffer[idx].size;
           // TODO:memory insufficient
           // return msgs already copied
           break;
         }
       }
-      *((int64_t*)buffer) = bufHandle->topicId;
+      *((int64_t*)buffer) = topicHandle->topicId;
       buffer = POINTER_SHIFT(buffer, sizeof(int64_t));
-      *((int64_t*)buffer) = bufHandle->buffer[idx].size;
+      *((int64_t*)buffer) = topicHandle->buffer[idx].size;
       buffer = POINTER_SHIFT(buffer, sizeof(int64_t));
-      memcpy(buffer, bufHandle->buffer[idx].content, bufHandle->buffer[idx].size);
+      memcpy(buffer, topicHandle->buffer[idx].content, topicHandle->buffer[idx].size);
-      buffer = POINTER_SHIFT(buffer, bufHandle->buffer[idx].size);
+      buffer = POINTER_SHIFT(buffer, topicHandle->buffer[idx].size);
       if (totSize > sizeLimit) {
         break;
       }
@@ -199,11 +194,19 @@ static int tqFetch(STqGroupHandle* gHandle, void** msg) {
   return totSize;
 }
 
-STqGroupHandle* tqGetGroupHandle(STQ* pTq, int64_t cId) { return NULL; }
+STqGroup* tqGetGroup(STQ* pTq, int64_t clientId) { return tqHandleGet(pTq->tqMeta, clientId); }
 
-int tqLaunchQuery(STqGroupHandle* gHandle) { return 0; }
+int tqSendLaunchQuery(STqMsgItem* bufItem, int64_t offset) {
-
+  if (tqQueryExecuting(bufItem->status)) {
-int tqSendLaunchQuery(STqGroupHandle* gHandle) { return 0; }
+    return 0;
+  }
+  bufItem->status = 1;
+  // load data from wal or buffer pool
+  // put into exec
+  // send exec into non blocking queue
+  // when query finished, put into buffer pool
+  return 0;
+}
 
 /*int tqMoveOffsetToNext(TqGroupHandle* gHandle) {*/
 /*return 0;*/
@@ -220,23 +223,96 @@ int tqCommit(STQ* pTq) {
   return 0;
 }
 
-int tqSetCursor(STQ* pTq, void* msg) {
+int tqBufferSetOffset(STqTopic* pTopic, int64_t offset) {
+  int code;
+  memset(pTopic->buffer, 0, sizeof(pTopic->buffer));
+  // launch query
+  for (int i = offset; i < offset + TQ_BUFFER_SIZE; i++) {
+    int pos = i % TQ_BUFFER_SIZE;
+    code = tqSendLaunchQuery(&pTopic->buffer[pos], offset);
+    if (code < 0) {
+      // TODO: error handling
+    }
+  }
+  // set offset
+  pTopic->nextConsumeOffset = offset;
+  pTopic->floatingCursor = offset;
   return 0;
 }
 
-int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
+STqTopic* tqFindTopic(STqGroup* pGroup, int64_t topicId) {
-  if (!tqProtoCheck((TmqMsgHead*)pMsg)) {
+  // TODO
-    // proto version invalid
+  return NULL;
-    return -1;
+}
-  }
+
-  int64_t         clientId = pMsg->head.clientId;
+int tqSetCursor(STQ* pTq, STqSetCurReq* pMsg) {
-  STqGroupHandle* gHandle = tqGetGroupHandle(pTq, clientId);
+  int       code;
+  int64_t   clientId = pMsg->head.clientId;
+  int64_t   topicId = pMsg->topicId;
+  int64_t   offset = pMsg->offset;
+  STqGroup* gHandle = tqGetGroup(pTq, clientId);
   if (gHandle == NULL) {
     // client not connect
     return -1;
   }
+  STqTopic* topicHandle = tqFindTopic(gHandle, topicId);
+  if (topicHandle == NULL) {
+    return -1;
+  }
+  if (pMsg->offset == topicHandle->nextConsumeOffset) {
+    return 0;
+  }
+  // TODO: check log last version
+
+  code = tqBufferSetOffset(topicHandle, offset);
+  if (code < 0) {
+    // set error code
+    return -1;
+  }
+
+  return 0;
+}
+
+int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
+  int64_t   clientId = pMsg->head.clientId;
+  STqGroup* pGroup = tqGetGroup(pTq, clientId);
+  if (pGroup == NULL) {
+    terrno = TSDB_CODE_TQ_GROUP_NOT_SET;
+    return -1;
+  }
+
+  STqConsumeRsp* pRsp = (STqConsumeRsp*)pMsg;
+  int            numOfMsgs = tqFetch(pGroup, (void**)&pRsp->msgs);
+  if (numOfMsgs < 0) {
+    return -1;
+  }
+  if (numOfMsgs == 0) {
+    // most recent data has been fetched
+
+    // enable timer for blocking wait
+    // once new data written during wait time
+    // launch query and response
+  }
+
+  // fetched a num of msgs, rpc response
+
+  return 0;
+}
+
+#if 0
+int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
+  if (!tqProtoCheck((STqMsgHead*)pMsg)) {
+    // proto version invalid
+    return -1;
+  }
+  int64_t   clientId = pMsg->head.clientId;
+  STqGroup* pGroup = tqGetGroup(pTq, clientId);
+  if (pGroup == NULL) {
+    // client not connect
+    return -1;
+  }
   if (pMsg->acks.ackNum != 0) {
-    if (tqAck(gHandle, &pMsg->acks) != 0) {
+    if (tqAck(pGroup, &pMsg->acks) != 0) {
       // ack not success
       return -1;
     }
@@ -244,22 +320,23 @@ int tqConsume(STQ* pTq, STqConsumeReq* pMsg) {
 
   STqConsumeRsp* pRsp = (STqConsumeRsp*)pMsg;
 
-  if (tqFetch(gHandle, (void**)&pRsp->msgs) <= 0) {
+  if (tqFetch(pGroup, (void**)&pRsp->msgs) <= 0) {
     // fetch error
     return -1;
   }
 
   // judge and launch new query
-  if (tqLaunchQuery(gHandle)) {
+  /*if (tqSendLaunchQuery(gHandle)) {*/
-    // launch query error
+  // launch query error
-    return -1;
+  /*return -1;*/
-  }
+  /*}*/
   return 0;
 }
+#endif
 
-int tqSerializeGroupHandle(const STqGroupHandle* gHandle, STqSerializedHead** ppHead) {
+int tqSerializeGroup(const STqGroup* pGroup, STqSerializedHead** ppHead) {
   // calculate size
-  int sz = tqGetgHandleSSize(gHandle) + sizeof(STqSerializedHead);
+  int sz = tqGroupSSize(pGroup) + sizeof(STqSerializedHead);
   if (sz > (*ppHead)->ssize) {
     void* tmpPtr = realloc(*ppHead, sz);
     if (tmpPtr == NULL) {
@@ -272,53 +349,53 @@ int tqSerializeGroupHandle(const STqGroupHandle* gHandle, STqSerializedHead** pp
   }
   void* ptr = (*ppHead)->content;
   // do serialization
-  *(int64_t*)ptr = gHandle->cId;
+  *(int64_t*)ptr = pGroup->clientId;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int64_t*)ptr = gHandle->cgId;
+  *(int64_t*)ptr = pGroup->cgId;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int32_t*)ptr = gHandle->topicNum;
+  *(int32_t*)ptr = pGroup->topicNum;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
-  if (gHandle->topicNum > 0) {
+  if (pGroup->topicNum > 0) {
-    tqSerializeListHandle(gHandle->head, ptr);
+    tqSerializeListHandle(pGroup->head, ptr);
   }
   return 0;
 }
 
-void* tqSerializeListHandle(STqListHandle* listHandle, void* ptr) {
+void* tqSerializeListHandle(STqList* listHandle, void* ptr) {
-  STqListHandle* node = listHandle;
+  STqList* node = listHandle;
   ASSERT(node != NULL);
   while (node) {
-    ptr = tqSerializeBufHandle(&node->bufHandle, ptr);
+    ptr = tqSerializeTopic(&node->topic, ptr);
     node = node->next;
   }
   return ptr;
 }
 
-void* tqSerializeBufHandle(STqBufferHandle* bufHandle, void* ptr) {
+void* tqSerializeTopic(STqTopic* pTopic, void* ptr) {
-  *(int64_t*)ptr = bufHandle->nextConsumeOffset;
+  *(int64_t*)ptr = pTopic->nextConsumeOffset;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int64_t*)ptr = bufHandle->topicId;
+  *(int64_t*)ptr = pTopic->topicId;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  *(int32_t*)ptr = bufHandle->head;
+  *(int32_t*)ptr = pTopic->head;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
-  *(int32_t*)ptr = bufHandle->tail;
+  *(int32_t*)ptr = pTopic->tail;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
   for (int i = 0; i < TQ_BUFFER_SIZE; i++) {
-    ptr = tqSerializeBufItem(&bufHandle->buffer[i], ptr);
+    ptr = tqSerializeItem(&pTopic->buffer[i], ptr);
   }
   return ptr;
 }
 
-void* tqSerializeBufItem(STqBufferItem* bufItem, void* ptr) {
+void* tqSerializeItem(STqMsgItem* bufItem, void* ptr) {
   // TODO: do we need serialize this?
   // mainly for executor
   return ptr;
 }
 
-const void* tqDeserializeGroupHandle(const STqSerializedHead* pHead, STqGroupHandle** ppGHandle) {
+const void* tqDeserializeGroup(const STqSerializedHead* pHead, STqGroup** ppGroup) {
-  STqGroupHandle* gHandle = *ppGHandle;
+  STqGroup*   gHandle = *ppGroup;
-  const void*     ptr = pHead->content;
+  const void* ptr = pHead->content;
-  gHandle->cId = *(int64_t*)ptr;
+  gHandle->clientId = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
   gHandle->cgId = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
@@ -326,63 +403,63 @@ const void* tqDeserializeGroupHandle(const STqSerializedHead* pHead, STqGroupHan
   gHandle->topicNum = *(int32_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
   gHandle->head = NULL;
-  STqListHandle* node = gHandle->head;
+  STqList* node = gHandle->head;
   for (int i = 0; i < gHandle->topicNum; i++) {
     if (gHandle->head == NULL) {
-      if ((node = malloc(sizeof(STqListHandle))) == NULL) {
+      if ((node = malloc(sizeof(STqList))) == NULL) {
         // TODO: error
         return NULL;
       }
       node->next = NULL;
-      ptr = tqDeserializeBufHandle(ptr, &node->bufHandle);
+      ptr = tqDeserializeTopic(ptr, &node->topic);
       gHandle->head = node;
     } else {
-      node->next = malloc(sizeof(STqListHandle));
+      node->next = malloc(sizeof(STqList));
       if (node->next == NULL) {
         // TODO: error
         return NULL;
       }
       node->next->next = NULL;
-      ptr = tqDeserializeBufHandle(ptr, &node->next->bufHandle);
+      ptr = tqDeserializeTopic(ptr, &node->next->topic);
       node = node->next;
     }
   }
   return ptr;
 }
 
-const void* tqDeserializeBufHandle(const void* pBytes, STqBufferHandle* bufHandle) {
+const void* tqDeserializeTopic(const void* pBytes, STqTopic* topic) {
   const void* ptr = pBytes;
-  bufHandle->nextConsumeOffset = *(int64_t*)ptr;
+  topic->nextConsumeOffset = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  bufHandle->topicId = *(int64_t*)ptr;
+  topic->topicId = *(int64_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int64_t));
-  bufHandle->head = *(int32_t*)ptr;
+  topic->head = *(int32_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
-  bufHandle->tail = *(int32_t*)ptr;
+  topic->tail = *(int32_t*)ptr;
   ptr = POINTER_SHIFT(ptr, sizeof(int32_t));
   for (int i = 0; i < TQ_BUFFER_SIZE; i++) {
-    ptr = tqDeserializeBufItem(ptr, &bufHandle->buffer[i]);
+    ptr = tqDeserializeItem(ptr, &topic->buffer[i]);
   }
   return ptr;
 }
 
-const void* tqDeserializeBufItem(const void* pBytes, STqBufferItem* bufItem) { return pBytes; }
+const void* tqDeserializeItem(const void* pBytes, STqMsgItem* bufItem) { return pBytes; }
 
 // TODO: make this a macro
-int tqGetgHandleSSize(const STqGroupHandle* gHandle) {
+int tqGroupSSize(const STqGroup* gHandle) {
   return sizeof(int64_t) * 2  // cId + cgId
          + sizeof(int32_t)    // topicNum
-         + gHandle->topicNum * tqBufHandleSSize();
+         + gHandle->topicNum * tqTopicSSize();
 }
 
 // TODO: make this a macro
-int tqBufHandleSSize() {
+int tqTopicSSize() {
   return sizeof(int64_t) * 2    // nextConsumeOffset + topicId
          + sizeof(int32_t) * 2  // head + tail
-         + TQ_BUFFER_SIZE * tqBufItemSSize();
+         + TQ_BUFFER_SIZE * tqItemSSize();
 }
 
-int tqBufItemSSize() {
+int tqItemSSize() {
   // TODO: do this need serialization?
   // mainly for executor
   return 0;

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

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

source/dnode/vnode/tq/test/tqMetaTest.cpp

@@ -10,8 +10,8 @@ struct Foo {
 };
 
 int FooSerializer(const void* pObj, STqSerializedHead** ppHead) {
-  Foo* foo = (Foo*) pObj;
+  Foo* foo = (Foo*)pObj;
-  if((*ppHead) == NULL || (*ppHead)->ssize < sizeof(STqSerializedHead) + sizeof(int32_t)) {
+  if ((*ppHead) == NULL || (*ppHead)->ssize < sizeof(STqSerializedHead) + sizeof(int32_t)) {
     *ppHead = (STqSerializedHead*)realloc(*ppHead, sizeof(STqSerializedHead) + sizeof(int32_t));
     (*ppHead)->ssize = sizeof(STqSerializedHead) + sizeof(int32_t);
   }
@@ -20,36 +20,28 @@ int FooSerializer(const void* pObj, STqSerializedHead** ppHead) {
 }
 
 const void* FooDeserializer(const STqSerializedHead* pHead, void** ppObj) {
-  if(*ppObj == NULL) {
+  if (*ppObj == NULL) {
     *ppObj = realloc(*ppObj, sizeof(int32_t));
   }
   Foo* pFoo = *(Foo**)ppObj;
-  pFoo->a = *(int32_t*)pHead->content; 
+  pFoo->a = *(int32_t*)pHead->content;
   return NULL;
 }
 
-void FooDeleter(void* pObj) {
+void FooDeleter(void* pObj) { free(pObj); }
-  free(pObj); 
-}
 
 class TqMetaUpdateAppendTest : public ::testing::Test {
-  protected:
+ protected:
+  void SetUp() override {
+    taosRemoveDir(pathName);
+    pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
+    ASSERT(pMeta);
+  }
 
-    void SetUp() override {
+  void TearDown() override { tqStoreClose(pMeta); }
-      taosRemoveDir(pathName);
-      pMeta = tqStoreOpen(pathName,
-          FooSerializer, FooDeserializer, FooDeleter,
-          TQ_UPDATE_APPEND
-        );
-      ASSERT(pMeta);
-    }
 
-    void TearDown() override {
+  STqMetaStore* pMeta;
-      tqStoreClose(pMeta);
+  const char*   pathName = "/tmp/tq_test";
-    }
-
-    TqMetaStore* pMeta;
-    const char* pathName = "/tmp/tq_test";
 };
 
 TEST_F(TqMetaUpdateAppendTest, copyPutTest) {
@@ -57,11 +49,11 @@ TEST_F(TqMetaUpdateAppendTest, copyPutTest) {
   foo.a = 3;
   tqHandleCopyPut(pMeta, 1, &foo, sizeof(Foo));
 
-  Foo* pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  Foo* pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 
   tqHandleCommit(pMeta, 1);
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo->a, 3);
 }
 
@@ -78,10 +70,7 @@ TEST_F(TqMetaUpdateAppendTest, persistTest) {
   EXPECT_EQ(pBar == NULL, true);
 
   tqStoreClose(pMeta);
-  pMeta = tqStoreOpen(pathName,
+  pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
-      FooSerializer, FooDeserializer, FooDeleter,
-      TQ_UPDATE_APPEND
-    );
   ASSERT(pMeta);
 
   pBar = (Foo*)tqHandleGet(pMeta, 1);
@@ -97,7 +86,7 @@ TEST_F(TqMetaUpdateAppendTest, uncommittedTest) {
   pFoo->a = 3;
   tqHandleMovePut(pMeta, 1, pFoo);
 
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 }
 
@@ -106,11 +95,11 @@ TEST_F(TqMetaUpdateAppendTest, abortTest) {
   pFoo->a = 3;
   tqHandleMovePut(pMeta, 1, pFoo);
 
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 
   tqHandleAbort(pMeta, 1);
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 }
 
@@ -119,32 +108,29 @@ TEST_F(TqMetaUpdateAppendTest, deleteTest) {
   pFoo->a = 3;
   tqHandleMovePut(pMeta, 1, pFoo);
 
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 
   tqHandleCommit(pMeta, 1);
 
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   ASSERT_EQ(pFoo != NULL, true);
   EXPECT_EQ(pFoo->a, 3);
 
   tqHandleDel(pMeta, 1);
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   ASSERT_EQ(pFoo != NULL, true);
   EXPECT_EQ(pFoo->a, 3);
 
   tqHandleCommit(pMeta, 1);
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 
   tqStoreClose(pMeta);
-  pMeta = tqStoreOpen(pathName,
+  pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
-      FooSerializer, FooDeserializer, FooDeleter,
-      TQ_UPDATE_APPEND
-    );
   ASSERT(pMeta);
 
-  pFoo = (Foo*) tqHandleGet(pMeta, 1);
+  pFoo = (Foo*)tqHandleGet(pMeta, 1);
   EXPECT_EQ(pFoo == NULL, true);
 }
 
@@ -162,10 +148,7 @@ TEST_F(TqMetaUpdateAppendTest, intxnPersist) {
   EXPECT_EQ(pFoo1->a, 3);
 
   tqStoreClose(pMeta);
-  pMeta = tqStoreOpen(pathName,
+  pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
-      FooSerializer, FooDeserializer, FooDeleter,
-      TQ_UPDATE_APPEND
-    );
   ASSERT(pMeta);
 
   pFoo1 = (Foo*)tqHandleGet(pMeta, 1);
@@ -177,10 +160,7 @@ TEST_F(TqMetaUpdateAppendTest, intxnPersist) {
   EXPECT_EQ(pFoo1->a, 4);
 
   tqStoreClose(pMeta);
-  pMeta = tqStoreOpen(pathName,
+  pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
-      FooSerializer, FooDeserializer, FooDeleter,
-      TQ_UPDATE_APPEND
-    );
   ASSERT(pMeta);
 
   pFoo1 = (Foo*)tqHandleGet(pMeta, 1);
@@ -190,13 +170,13 @@ TEST_F(TqMetaUpdateAppendTest, intxnPersist) {
 TEST_F(TqMetaUpdateAppendTest, multiplePage) {
   srand(0);
   std::vector<int> v;
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     v.push_back(rand());
     Foo foo;
     foo.a = v[i];
     tqHandleCopyPut(pMeta, i, &foo, sizeof(Foo));
   }
-  for(int i = 0; i < 500; i++) {
+  for (int i = 0; i < 500; i++) {
     tqHandleCommit(pMeta, i);
     Foo* pFoo = (Foo*)tqHandleGet(pMeta, i);
     ASSERT_EQ(pFoo != NULL, true) << " at idx " << i << "\n";
@@ -204,38 +184,34 @@ TEST_F(TqMetaUpdateAppendTest, multiplePage) {
   }
 
   tqStoreClose(pMeta);
-  pMeta = tqStoreOpen(pathName,
+  pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
-      FooSerializer, FooDeserializer, FooDeleter,
-      TQ_UPDATE_APPEND
-    );
   ASSERT(pMeta);
-  
+
-  for(int i = 500; i < 1000; i++) {
+  for (int i = 500; i < 1000; i++) {
     tqHandleCommit(pMeta, i);
     Foo* pFoo = (Foo*)tqHandleGet(pMeta, i);
     ASSERT_EQ(pFoo != NULL, true) << " at idx " << i << "\n";
     EXPECT_EQ(pFoo->a, v[i]);
   }
 
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     Foo* pFoo = (Foo*)tqHandleGet(pMeta, i);
     ASSERT_EQ(pFoo != NULL, true) << " at idx " << i << "\n";
     EXPECT_EQ(pFoo->a, v[i]);
   }
-
 }
 
 TEST_F(TqMetaUpdateAppendTest, multipleRewrite) {
   srand(0);
   std::vector<int> v;
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     v.push_back(rand());
     Foo foo;
     foo.a = v[i];
     tqHandleCopyPut(pMeta, i, &foo, sizeof(Foo));
   }
 
-  for(int i = 0; i < 500; i++) {
+  for (int i = 0; i < 500; i++) {
     tqHandleCommit(pMeta, i);
     v[i] = rand();
     Foo foo;
@@ -243,25 +219,22 @@ TEST_F(TqMetaUpdateAppendTest, multipleRewrite) {
     tqHandleCopyPut(pMeta, i, &foo, sizeof(Foo));
   }
 
-  for(int i = 500; i < 1000; i++) {
+  for (int i = 500; i < 1000; i++) {
     v[i] = rand();
     Foo foo;
     foo.a = v[i];
     tqHandleCopyPut(pMeta, i, &foo, sizeof(Foo));
   }
 
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     tqHandleCommit(pMeta, i);
   }
 
   tqStoreClose(pMeta);
-  pMeta = tqStoreOpen(pathName,
+  pMeta = tqStoreOpen(pathName, FooSerializer, FooDeserializer, FooDeleter, TQ_UPDATE_APPEND);
-      FooSerializer, FooDeserializer, FooDeleter,
-      TQ_UPDATE_APPEND
-    );
   ASSERT(pMeta);
-  
+
-  for(int i = 500; i < 1000; i++) {
+  for (int i = 500; i < 1000; i++) {
     v[i] = rand();
     Foo foo;
     foo.a = v[i];
@@ -269,40 +242,38 @@ TEST_F(TqMetaUpdateAppendTest, multipleRewrite) {
     tqHandleCommit(pMeta, i);
   }
 
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     Foo* pFoo = (Foo*)tqHandleGet(pMeta, i);
     ASSERT_EQ(pFoo != NULL, true) << " at idx " << i << "\n";
     EXPECT_EQ(pFoo->a, v[i]);
   }
-
 }
 
 TEST_F(TqMetaUpdateAppendTest, dupCommit) {
   srand(0);
   std::vector<int> v;
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     v.push_back(rand());
     Foo foo;
     foo.a = v[i];
     tqHandleCopyPut(pMeta, i, &foo, sizeof(Foo));
   }
 
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     int ret = tqHandleCommit(pMeta, i);
     EXPECT_EQ(ret, 0);
     ret = tqHandleCommit(pMeta, i);
     EXPECT_EQ(ret, -1);
   }
 
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     int ret = tqHandleCommit(pMeta, i);
     EXPECT_EQ(ret, -1);
   }
 
-  for(int i = 0; i < 1000; i++) {
+  for (int i = 0; i < 1000; i++) {
     Foo* pFoo = (Foo*)tqHandleGet(pMeta, i);
     ASSERT_EQ(pFoo != NULL, true) << " at idx " << i << "\n";
     EXPECT_EQ(pFoo->a, v[i]);
   }
-
 }

source/libs/catalog/src/catalog.c

@@ -634,6 +634,16 @@ _return:
   CTG_RET(code);
 }
 
+int32_t catalogGetQnodeList(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, SEpSet* pQnodeEpSet) {
+  if (NULL == pCatalog || NULL == pRpc  || NULL == pMgmtEps || NULL == pQnodeEpSet) {
+    CTG_ERR_RET(TSDB_CODE_CTG_INVALID_INPUT);
+  }
+
+
+  return TSDB_CODE_SUCCESS;
+}
+
+
 void catalogDestroy(void) {
   if (ctgMgmt.pCluster) {
     taosHashCleanup(ctgMgmt.pCluster); //TBD

source/libs/index/inc/indexInt.h

@@ -76,7 +76,7 @@ struct SIndexMultiTermQuery {
 
 // field and key;
 typedef struct SIndexTerm {
-  int64_t           suid; 
+  int64_t            suid; 
   SIndexOperOnColumn operType; // oper type, add/del/update
   uint8_t            colType;  // term data type, str/interger/json
   char    *colName;

source/libs/index/inc/index_fst_counting_writer.h

@@ -34,8 +34,14 @@ typedef struct WriterCtx {
   int (*flush)(struct WriterCtx *ctx);
   WriterType type; 
   union {
-    int fd;
+    struct {
-    void *mem;
+      int fd; 
+      bool readOnly;
+    } file; 
+    struct {
+      int32_t capa;
+      char *buf;
+    } mem;
   };
   int32_t offset;
   int32_t limit;
@@ -45,7 +51,7 @@ static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len);
 static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len);
 static int writeCtxDoFlush(WriterCtx *ctx);
 
-WriterCtx* writerCtxCreate(WriterType type, bool readOnly);
+WriterCtx* writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity);
 void writerCtxDestroy(WriterCtx *w);
 
 typedef uint32_t CheckSummer;
@@ -66,7 +72,7 @@ int fstCountingWriterFlush(FstCountingWriter *write);
 
 uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write);
 
-FstCountingWriter *fstCountingWriterCreate(void *wtr, bool readOnly);
+FstCountingWriter *fstCountingWriterCreate(void *wtr);
 void fstCountingWriterDestroy(FstCountingWriter *w); 
 
 

source/libs/index/inc/index_tfile.h

@@ -18,23 +18,81 @@
 #include "index.h"
 #include "indexInt.h"
 #include "tlockfree.h"
-#include "tskiplist.h"
+#include "index_tfile.h"
+#include "index_fst_counting_writer.h"
+#include "index_fst.h"
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-typedef struct IndexTFile {
+  
+
+typedef struct TFileCacheKey {
+  uint64_t   suid;
+  uint8_t    colType;
+  int32_t    version;
+  const char *colName;
+  int32_t    nColName;
+} TFileCacheKey; 
+
+
+// table cache
+// refactor to LRU cache later
+typedef struct TFileCache {
+  SHashObj *tableCache;        
+  int16_t  capacity;
+  // add more param  
+} TFileCache;
+
+
+typedef struct TFileWriter {
+  FstBuilder *fb;
+  WriterCtx  *wc; 
+} TFileWriter;
+
+typedef struct TFileReader {
   T_REF_DECLARE() 
+  Fst *fst;
+  
+} TFileReader; 
+
+typedef struct IndexTFile {
+  char *path;
+  TFileReader *tb;
+  TFileWriter *tw;      
 } IndexTFile;
 
+typedef struct TFileWriterOpt {
+  uint64_t suid;
+  int8_t   colType;
+  char     *colName;  
+  int32_t  nColName; 
+  int32_t  version;
+} TFileWriterOpt; 
 
+typedef struct TFileReaderOpt {
+  uint64_t suid; 
+  char     *colName;
+  int32_t  nColName; 
+  
+} TFileReaderOpt;
+
+// tfile cache 
+TFileCache *tfileCacheCreate();
+void tfileCacheDestroy(TFileCache *tcache);
+TFileReader* tfileCacheGet(TFileCache *tcache, TFileCacheKey *key);
+void tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader);
+  
+TFileWriter *tfileWriterCreate(const char *suid, const char *colName);
 
 IndexTFile *indexTFileCreate();
 
+int indexTFilePut(void *tfile, SIndexTerm *term,  uint64_t uid); 
 
 int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result);
 
+
 #ifdef __cplusplus
 }
 

source/libs/index/inc/index_util.h

@@ -0,0 +1,53 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+#ifndef __INDEX_UTIL_H__
+#define __INDEX_UTIL_H__
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define SERIALIZE_MEM_TO_BUF(buf, key, mem) \
+  do { \
+   memcpy((void *)buf, (void *)(&key->mem), sizeof(key->mem)); \
+   buf += sizeof(key->mem);  \
+  } while (0)
+
+#define SERIALIZE_STR_MEM_TO_BUF(buf, key, mem, len) \
+  do { \
+    memcpy((void *)buf, (void *)key->mem, len); \
+    buf += len; \
+  } while (0)
+
+#define SERIALIZE_VAR_TO_BUF(buf, var, type) \
+  do { \
+     type c = var; \
+     assert(sizeof(var) == sizeof(type));\
+     memcpy((void *)buf, (void *)&c, sizeof(c)); \
+     buf += sizeof(c); \
+  } while (0)
+
+#define SERIALIZE_STR_VAR_TO_BUF(buf, var, len) \
+  do { \
+      memcpy((void *)buf, (void *)var, len); \
+      buf += len;\
+  } while (0)
+
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

source/libs/index/src/index_cache.c

@@ -15,6 +15,7 @@
 
 #include "index_cache.h"
 #include "tcompare.h"
+#include "index_util.h"
 
 #define MAX_INDEX_KEY_LEN 256// test only, change later
 
@@ -110,35 +111,22 @@ int indexCachePut(void *cache, SIndexTerm *term, int16_t colId, int32_t version,
   if (cache == NULL) { return -1;} 
 
   IndexCache *pCache = cache;
-
   // encode data
   int32_t total = CACHE_KEY_LEN(term); 
-
   char *buf = calloc(1, total); 
   char *p   = buf;
 
-  memcpy(p, &total, sizeof(total)); 
+  SERIALIZE_VAR_TO_BUF(p, total,int32_t);
-  p += sizeof(total);
+  SERIALIZE_VAR_TO_BUF(p, colId, int16_t);
 
-  memcpy(p, &colId, sizeof(colId));   
+  SERIALIZE_MEM_TO_BUF(p, term, colType);
-  p += sizeof(colId);
+  SERIALIZE_MEM_TO_BUF(p, term, nColVal); 
-
+  SERIALIZE_STR_MEM_TO_BUF(p, term, colVal, term->nColVal); 
-  memcpy(p, &term->colType, sizeof(term->colType));
-  p += sizeof(term->colType);
   
-  memcpy(p, &term->nColVal, sizeof(term->nColVal));
+  SERIALIZE_VAR_TO_BUF(p, version, int32_t);
-  p += sizeof(term->nColVal);
+  SERIALIZE_VAR_TO_BUF(p, uid, uint64_t); 
-  memcpy(p, term->colVal, term->nColVal); 
-  p += term->nColVal;
 
-  memcpy(p, &version, sizeof(version));
+  SERIALIZE_MEM_TO_BUF(p, term, operType);
-  p += sizeof(version);
-
-  memcpy(p, &uid, sizeof(uid));  
-  p += sizeof(uid);
-
-  memcpy(p, &term->operType, sizeof(term->operType));
-  p += sizeof(term->operType); 
 
   tSkipListPut(pCache->skiplist, (void *)buf);  
   return 0;

source/libs/index/src/index_fst.c

@@ -779,7 +779,7 @@ FstBuilder *fstBuilderCreate(void *w, FstType ty) {
   if (NULL == b) { return b; }
 
    
-  b->wrt        = fstCountingWriterCreate(w, false);
+  b->wrt        = fstCountingWriterCreate(w);
   b->unfinished = fstUnFinishedNodesCreate();   
   b->registry   = fstRegistryCreate(10000, 2) ;
   b->last       = fstSliceCreate(NULL, 0);

source/libs/index/src/index_fst_counting_writer.c

@@ -23,9 +23,9 @@ static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
   }
 
   if (ctx->type == TFile) {
-    assert(len == tfWrite(ctx->fd, buf, len));
+    assert(len == tfWrite(ctx->file.fd, buf, len));
   } else {
-    memcpy(ctx->mem + ctx->offset, buf, len);
+    memcpy(ctx->mem.buf+ ctx->offset, buf, len);
   } 
   ctx->offset += len;
   return len;
@@ -33,9 +33,9 @@ static int writeCtxDoWrite(WriterCtx *ctx, uint8_t *buf, int len) {
 static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
   int nRead = 0; 
   if (ctx->type == TFile) {
-    nRead = tfRead(ctx->fd, buf, len);
+    nRead = tfRead(ctx->file.fd, buf, len);
   } else {
-    memcpy(buf, ctx->mem + ctx->offset, len);
+    memcpy(buf, ctx->mem.buf + ctx->offset, len);
   }
   ctx->offset += nRead;
 
@@ -44,63 +44,64 @@ static int writeCtxDoRead(WriterCtx *ctx, uint8_t *buf, int len) {
 static int writeCtxDoFlush(WriterCtx *ctx) {
   if (ctx->type == TFile) {
     //tfFsync(ctx->fd);
-    //tfFlush(ctx->fd);
+    //tfFlush(ctx->file.fd);
   } else {
     // do nothing
   }
   return 1;
 }
 
-WriterCtx* writerCtxCreate(WriterType type, bool readOnly) {     
+WriterCtx* writerCtxCreate(WriterType type, const char *path, bool readOnly, int32_t capacity) {     
   WriterCtx *ctx = calloc(1, sizeof(WriterCtx));
   if (ctx == NULL) { return NULL; }
 
   ctx->type = type;
   if (ctx->type == TFile) {
-    tfInit();
     // ugly code, refactor later
+    ctx->file.readOnly = readOnly;
     if (readOnly == false) {
-      ctx->fd = tfOpenCreateWriteAppend(tmpFile);  
+      ctx->file.fd = tfOpenCreateWriteAppend(tmpFile);  
     } else {
-      ctx->fd = tfOpenReadWrite(tmpFile);
+      ctx->file.fd = tfOpenReadWrite(tmpFile);
     } 
-    if (ctx->fd < 0) {
+    if (ctx->file.fd < 0) {
       indexError("open file error %d", errno);       
     }
   } else if (ctx->type == TMemory) {
-    ctx->mem = calloc(1, DefaultMem * sizeof(uint8_t));
+    ctx->mem.buf  = calloc(1, sizeof(char) * capacity); 
+    ctx->mem.capa = capacity; 
   } 
   ctx->write = writeCtxDoWrite;
   ctx->read  = writeCtxDoRead;
   ctx->flush = writeCtxDoFlush;
 
   ctx->offset = 0;
-  ctx->limit  = DefaultMem;
+  ctx->limit  = capacity;
 
   return ctx;
 }
 void writerCtxDestroy(WriterCtx *ctx) {
   if (ctx->type == TMemory) {
-    free(ctx->mem);
+    free(ctx->mem.buf);
   } else {
-    tfClose(ctx->fd);    
+    tfClose(ctx->file.fd);    
-    tfCleanup();
   }
   free(ctx);
 }
 
 
-FstCountingWriter *fstCountingWriterCreate(void *wrt, bool readOnly) {
+FstCountingWriter *fstCountingWriterCreate(void *wrt) {
   FstCountingWriter *cw = calloc(1, sizeof(FstCountingWriter)); 
   if (cw == NULL) { return NULL; }
   
-  cw->wrt = (void *)(writerCtxCreate(TFile, readOnly)); 
+  cw->wrt = wrt;
+  //(void *)(writerCtxCreate(TFile, readOnly)); 
   return cw; 
 }
 void fstCountingWriterDestroy(FstCountingWriter *cw) {
   // free wrt object: close fd or free mem 
   fstCountingWriterFlush(cw);
-  writerCtxDestroy((WriterCtx *)(cw->wrt));
+  //writerCtxDestroy((WriterCtx *)(cw->wrt));
   free(cw);
 }
 
@@ -124,6 +125,7 @@ int fstCountingWriterRead(FstCountingWriter *write, uint8_t *buf, uint32_t len)
 } 
 
 uint32_t fstCountingWriterMaskedCheckSum(FstCountingWriter *write) {
+  
   return 0;
 }
 int fstCountingWriterFlush(FstCountingWriter *write) {

source/libs/index/src/index_tfile.c

@@ -14,6 +14,50 @@
  */
 
 #include "index_tfile.h"
+#include "index_fst.h"
+#include "index_util.h"
+
+
+
+static void tfileSerialCacheKey(TFileCacheKey *key, char *buf) {
+  SERIALIZE_MEM_TO_BUF(buf, key, suid);
+  SERIALIZE_VAR_TO_BUF(buf, '_', char); 
+  SERIALIZE_MEM_TO_BUF(buf, key, colType);
+  SERIALIZE_VAR_TO_BUF(buf, '_', char); 
+  SERIALIZE_MEM_TO_BUF(buf, key, version);
+  SERIALIZE_VAR_TO_BUF(buf, '_', char); 
+  SERIALIZE_STR_MEM_TO_BUF(buf, key, colName, key->nColName);
+}
+
+TFileCache *tfileCacheCreate() {
+  TFileCache *tcache = calloc(1, sizeof(TFileCache)); 
+  if (tcache == NULL) { return NULL; }
+  
+  tcache->tableCache = taosHashInit(8, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK); 
+  tcache->capacity   = 64;
+  return tcache;
+}
+void tfileCacheDestroy(TFileCache *tcache) {
+  
+  free(tcache);    
+   
+}
+
+TFileReader *tfileCacheGet(TFileCache *tcache, TFileCacheKey *key) {
+  char buf[128] = {0}; 
+  tfileSerialCacheKey(key, buf);
+  TFileReader *reader = taosHashGet(tcache->tableCache, buf, strlen(buf)); 
+  return reader; 
+}
+void tfileCachePut(TFileCache *tcache, TFileCacheKey *key, TFileReader *reader) {
+  char buf[128] = {0};
+  tfileSerialCacheKey(key, buf);
+  taosHashPut(tcache->tableCache, buf, strlen(buf), &reader, sizeof(void *));     
+  return;
+} 
+
+
+
 
 IndexTFile *indexTFileCreate() {
   IndexTFile *tfile = calloc(1, sizeof(IndexTFile));   
@@ -22,10 +66,24 @@ IndexTFile *indexTFileCreate() {
 void IndexTFileDestroy(IndexTFile *tfile) {
   free(tfile); 
 }
+
+
 int indexTFileSearch(void *tfile, SIndexTermQuery *query, SArray *result) {
   IndexTFile *ptfile = (IndexTFile *)tfile;
+     
   return 0;
 }
+int indexTFilePut(void *tfile, SIndexTerm *term,  uint64_t uid) {
+  TFileWriterOpt wOpt = {.suid    = term->suid,
+                         .colType = term->colType,
+                         .colName = term->colName,
+                         .nColName= term->nColName,
+                         .version = 1}; 
+
+  
+   
+  return 0; 
+}
 
 
 

source/libs/index/test/indexTests.cc

@@ -45,7 +45,7 @@ class FstWriter {
 class FstReadMemory {
   public:
    FstReadMemory(size_t size) {
-     _w    = fstCountingWriterCreate(NULL, true); 
+     _w    = fstCountingWriterCreate(NULL); 
      _size = size; 
      memset((void *)&_s, 0, sizeof(_s));
    }

source/libs/parser/inc/astToMsg.h

@@ -4,7 +4,8 @@
 #include "parserInt.h"
 #include "taosmsg.h"
 
-SCreateUserMsg* buildUserManipulationMsg(SSqlInfo* pInfo, int64_t id, char* msgBuf, int32_t msgLen);
+SCreateUserMsg* buildUserManipulationMsg(SSqlInfo* pInfo, int32_t* outputLen, int64_t id, char* msgBuf, int32_t msgLen);
+SDropUserMsg* buildDropUserMsg(SSqlInfo* pInfo, int32_t* outputLen, int64_t id, char* msgBuf, int32_t msgLen);
 SShowMsg* buildShowMsg(SShowInfo* pShowInfo, int64_t id, char* msgBuf, int32_t msgLen);
 SCreateDbMsg* buildCreateDbMsg(SCreateDbInfo* pCreateDbInfo, char* msgBuf, int32_t msgLen);
 

source/libs/parser/inc/queryInfoUtil.h

@@ -36,7 +36,6 @@ void       addExprInfo(SArray* pExprList, int32_t index, SExprInfo* pExprInfo, i
 void       updateExprInfo(SExprInfo* pExprInfo, int16_t functionId, int32_t colId, int16_t srcColumnIndex, int16_t resType, int16_t resSize);
 
 SExprInfo* getExprInfo(SQueryStmtInfo* pQueryInfo, int32_t index);
-int32_t    copyAllExprInfo(SArray* dst, const SArray* src, bool deepcopy);
 
 void       addExprInfoParam(SSqlExpr* pExpr, char* argument, int32_t type, int32_t bytes);
 

source/libs/parser/src/astToMsg.c

@@ -1,7 +1,7 @@
 #include "parserInt.h"
 #include "parserUtil.h"
 
-SCreateUserMsg* buildUserManipulationMsg(SSqlInfo* pInfo, int64_t id, char* msgBuf, int32_t msgLen) {
+SCreateUserMsg* buildUserManipulationMsg(SSqlInfo* pInfo, int32_t* outputLen, int64_t id, char* msgBuf, int32_t msgLen) {
   SCreateUserMsg* pMsg = (SCreateUserMsg*)calloc(1, sizeof(SCreateUserMsg));
   if (pMsg == NULL) {
     //    tscError("0x%" PRIx64 " failed to malloc for query msg", id);
@@ -20,6 +20,24 @@ SCreateUserMsg* buildUserManipulationMsg(SSqlInfo* pInfo, int64_t id, char* msgB
     strncpy(pMsg->pass, pUser->passwd.z, pUser->passwd.n);
   }
 
+  *outputLen = sizeof(SUserInfo);
+  return pMsg;
+}
+
+SDropUserMsg* buildDropUserMsg(SSqlInfo* pInfo, int32_t *msgLen, int64_t id, char* msgBuf, int32_t msgBufLen) {
+  SToken* pName = taosArrayGet(pInfo->pMiscInfo->a, 0);
+  if (pName->n >= TSDB_USER_LEN) {
+    return NULL;
+  }
+
+
+  SDropUserMsg* pMsg = calloc(1, sizeof(SDropUserMsg));
+  if (pMsg == NULL) {
+    return NULL;
+  }
+
+  strncpy(pMsg->user, pName->z, pName->n);
+  *msgLen = sizeof(SDropUserMsg);
   return pMsg;
 }
 
@@ -89,7 +107,7 @@ static int32_t setTimePrecision(SCreateDbMsg* pMsg, const SCreateDbInfo* pCreate
 
   pMsg->precision = TSDB_TIME_PRECISION_MILLI;  // millisecond by default
 
-  SToken* pToken = &pCreateDbInfo->precision;
+  SToken* pToken = (SToken*) &pCreateDbInfo->precision;
   if (pToken->n > 0) {
     pToken->n = strdequote(pToken->z);
 
@@ -141,7 +159,6 @@ int32_t setDbOptions(SCreateDbMsg* pCreateDbMsg, const SCreateDbInfo* pCreateDbS
 
   // todo configurable
   pCreateDbMsg->numOfVgroups = htonl(2);
-
   return TSDB_CODE_SUCCESS;
 }
 

source/libs/parser/src/astValidate.c

@@ -4224,7 +4224,13 @@ int32_t qParserValidateDclSqlNode(SSqlInfo* pInfo, int64_t id, void** output, in
         }
       }
 
-      *output = buildUserManipulationMsg(pInfo, id, msgBuf, msgBufLen);
+      *output = buildUserManipulationMsg(pInfo, outputLen, id, msgBuf, msgBufLen);
+      break;
+    }
+
+    case TSDB_SQL_DROP_ACCT:
+    case TSDB_SQL_DROP_USER: {
+      *output = buildDropUserMsg(pInfo, outputLen, id, msgBuf, msgBufLen);
       break;
     }
     

source/libs/parser/src/queryInfoUtil.c

@@ -1,10 +1,7 @@
-#include "queryInfoUtil.h"
-#include <function.h>
-#include "astGenerator.h"
-#include "function.h"
 #include "os.h"
+#include "queryInfoUtil.h"
+#include "function.h"
 #include "parser.h"
-#include "parserInt.h"
 #include "parserUtil.h"
 
 static struct SSchema _s = {

source/libs/parser/test/insertTest.cpp

@@ -57,17 +57,20 @@ void *__wrap_malloc(size_t c) {
 //   [...];
 class InsertTest : public Test {
 protected:
-  void setDatabase(const string& db) {
+  void setDatabase(const string& acctId, const string& db) {
+    acctId_ = acctId;
     db_ = db;
   }
 
   void bind(const char* sql) {
     reset();
-    cxt_.sqlLen = strlen(sql);
+    cxt_.pAcctId = acctId_.c_str();
+    cxt_.pDbname = db_.c_str();
     strcpy(sqlBuf_, sql);
+    cxt_.sqlLen = strlen(sql);
     sqlBuf_[cxt_.sqlLen] = '\0';
     cxt_.pSql = sqlBuf_;
-    cxt_.pDbname = db_.c_str();
+
   }
 
   int32_t run() {
@@ -95,6 +98,7 @@ private:
     res_ = nullptr;
   }
 
+  string acctId_;
   string db_;
   char errMagBuf_[max_err_len];
   char sqlBuf_[max_sql_len];
@@ -105,7 +109,7 @@ private:
 
 // INSERT INTO tb_name VALUES (field1_value, ...)
 TEST_F(InsertTest, simpleTest) {
-  setDatabase("test");
+  setDatabase("root", "test");
 
   bind("insert into t1 values (now, 1, \"wxy\")");
   ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
@@ -116,7 +120,7 @@ TEST_F(InsertTest, simpleTest) {
 }
 
 TEST_F(InsertTest, toleranceTest) {
-  setDatabase("test");
+  setDatabase("root", "test");
 
   bind("insert into");
   ASSERT_NE(run(), TSDB_CODE_SUCCESS);

source/libs/parser/test/mockCatalog.cpp

@@ -20,34 +20,42 @@
 namespace {
 
 void generateTestT1(MockCatalogService* mcs) {
-  ITableBuilder& builder = mcs->createTableBuilder("test", "t1", TSDB_NORMAL_TABLE, 3)
+  ITableBuilder& builder = mcs->createTableBuilder("root.test", "t1", TSDB_NORMAL_TABLE, 3)
       .setPrecision(TSDB_TIME_PRECISION_MILLI).setVgid(1).addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
       .addColumn("c1", TSDB_DATA_TYPE_INT).addColumn("c2", TSDB_DATA_TYPE_BINARY, 10);
   builder.done();
 }
 
 void generateTestST1(MockCatalogService* mcs) {
-  ITableBuilder& builder = mcs->createTableBuilder("test", "st1", TSDB_SUPER_TABLE, 3, 2)
+  ITableBuilder& builder = mcs->createTableBuilder("root.test", "st1", TSDB_SUPER_TABLE, 3, 2)
       .setPrecision(TSDB_TIME_PRECISION_MILLI).addColumn("ts", TSDB_DATA_TYPE_TIMESTAMP)
       .addTag("tag1", TSDB_DATA_TYPE_INT).addTag("tag2", TSDB_DATA_TYPE_BINARY, 10)
       .addColumn("c1", TSDB_DATA_TYPE_INT).addColumn("c2", TSDB_DATA_TYPE_BINARY, 10);
   builder.done();
-  mcs->createSubTable("test", "st1", "st1s1", 1);
+  mcs->createSubTable("root.test", "st1", "st1s1", 1);
-  mcs->createSubTable("test", "st1", "st1s2", 2);
+  mcs->createSubTable("root.test", "st1", "st1s2", 2);
 }
 
 }
 
-void generateMetaData(MockCatalogService* mcs) {
+int32_t catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle) {
-  generateTestT1(mcs);
+  return mockCatalogService->catalogGetHandle(clusterId, catalogHandle);
-  generateTestST1(mcs);
-  mcs->showTables();
 }
 
-struct SCatalog* getCatalogHandle(const SEpSet* pMgmtEps) {
+int32_t catalogGetTableMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, const char* pTableName, STableMeta** pTableMeta) {
-  return mockCatalogService->getCatalogHandle(pMgmtEps);
+  return mockCatalogService->catalogGetTableMeta(pCatalog, pRpc, pMgmtEps, pDBName, pTableName, pTableMeta);
 }
 
-int32_t catalogGetMetaData(struct SCatalog* pCatalog, const SCatalogReq* pMetaReq, SMetaData* pMetaData) {
+void initMetaDataEnv() {
-  return mockCatalogService->catalogGetMetaData(pCatalog, pMetaReq, pMetaData);
+  mockCatalogService.reset(new MockCatalogService());
+}
+
+void generateMetaData() {
+  generateTestT1(mockCatalogService.get());
+  generateTestST1(mockCatalogService.get());
+  mockCatalogService->showTables();
+}
+
+void destroyMetaDataEnv() {
+  mockCatalogService.reset();
 }

source/libs/parser/test/mockCatalog.h

@@ -18,10 +18,12 @@
 
 #include "mockCatalogService.h"
 
-void generateMetaData(MockCatalogService* mcs);
+void initMetaDataEnv();
+void generateMetaData();
+void destroyMetaDataEnv();
 
 // mock
-struct SCatalog* getCatalogHandle(const SEpSet* pMgmtEps);
+int32_t catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle);
-int32_t catalogGetMetaData(struct SCatalog* pCatalog, const SCatalogReq* pMetaReq, SMetaData* pMetaData);
+int32_t catalogGetTableMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, const char* pTableName, STableMeta** pTableMeta);
 
 #endif  // MOCK_CATALOG_H

source/libs/parser/test/mockCatalogService.cpp

@@ -27,27 +27,28 @@ std::unique_ptr<MockCatalogService> mockCatalogService;
 class TableBuilder : public ITableBuilder {
 public:
   virtual TableBuilder& addColumn(const std::string& name, int8_t type, int32_t bytes) {
-    assert(colIndex_ < meta_->tableInfo.numOfTags + meta_->tableInfo.numOfColumns);
+    assert(colId_ < schema()->tableInfo.numOfTags + schema()->tableInfo.numOfColumns);
-    SSchema* col = meta_->schema + colIndex_;
+    SSchema* col = schema()->schema + colId_;
     col->type = type;
-    col->colId = colIndex_++;
+    col->colId = colId_++;
     col->bytes = bytes;
     strcpy(col->name, name.c_str());
     return *this;
   }
 
   virtual TableBuilder& setVgid(int16_t vgid) {
-    meta_->vgId = vgid;
+    schema()->vgId = vgid;
+    meta_->vgs.emplace_back(SVgroupInfo{.vgId = vgid, .hashBegin = 0, .hashEnd = 0, .inUse = 0, .numOfEps = 3, .epAddr = {{"dnode_1", 6030}, {"dnode_2", 6030}, {"dnode_3", 6030}}});
     return *this;
   }
 
   virtual TableBuilder& setPrecision(uint8_t precision) {
-    meta_->tableInfo.precision = precision;
+    schema()->tableInfo.precision = precision;
     return *this;
   }
 
   virtual void done() {
-    meta_->tableInfo.rowSize = rowsize_;
+    schema()->tableInfo.rowSize = rowsize_;
   }
 
 private:
@@ -64,61 +65,62 @@ private:
     return std::unique_ptr<TableBuilder>(new TableBuilder(meta));
   }
 
-  TableBuilder(STableMeta* meta) : colIndex_(0), rowsize_(0), meta_(meta) {
+  TableBuilder(STableMeta* schemaMeta) : colId_(0), rowsize_(0), meta_(new MockTableMeta()) {
+    meta_->schema.reset(schemaMeta);
   }
 
-  STableMeta* table() {
+  std::shared_ptr<STableMeta> schema() {
+    return meta_->schema;
+  }
+
+  std::shared_ptr<MockTableMeta> table() {
     return meta_;
   }
 
-  int32_t colIndex_;
+  int32_t colId_;
   int32_t rowsize_;
-  STableMeta* meta_;
+  std::shared_ptr<MockTableMeta> meta_;
 };
 
 class MockCatalogServiceImpl {
 public:
   static const int32_t numOfDataTypes = sizeof(tDataTypes) / sizeof(tDataTypes[0]);
 
-  MockCatalogServiceImpl() {
+  MockCatalogServiceImpl() : id_(1) {
   }
 
-  struct SCatalog* getCatalogHandle(const SEpSet* pMgmtEps) const {
+  int32_t catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle) const {
-    return (struct SCatalog*)0x01;
+    return 0;
   }
 
-  int32_t catalogGetMetaData(struct SCatalog* pCatalog, const SCatalogReq* pMetaReq, SMetaData* pMetaData) const {
+  int32_t catalogGetTableMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, const char* pTableName, STableMeta** pTableMeta) const {
-    assert(nullptr != pMetaReq && 1 == taosArrayGetSize(pMetaReq->pTableName));
-    SName* fullName = (SName*)taosArrayGet(pMetaReq->pTableName, 0);
     std::unique_ptr<STableMeta> table;
-    int32_t code = copyTableMeta(fullName->dbname, fullName->tname, &table);
+    int32_t code = copyTableSchemaMeta(pDBName, pTableName, &table);
     if (TSDB_CODE_SUCCESS != code) {
       return code;
     }
-    std::unique_ptr<SArray> tables((SArray*)taosArrayInit(1, sizeof(STableMeta*)));
+    *pTableMeta = table.release();
-    if (!tables) {
-      return TSDB_CODE_TSC_OUT_OF_MEMORY;
-    }
-    STableMeta* elem = table.release();
-    taosArrayPush(tables.get(), &elem);
-    pMetaData->pTableMeta = tables.release();
     return TSDB_CODE_SUCCESS;
   }
 
   TableBuilder& createTableBuilder(const std::string& db, const std::string& tbname, int8_t tableType, int32_t numOfColumns, int32_t numOfTags) {
     builder_ = TableBuilder::createTableBuilder(tableType, numOfColumns, numOfTags);
-    meta_[db][tbname].reset(builder_->table());
+    meta_[db][tbname] = builder_->table();
-    meta_[db][tbname]->uid = id_++;
+    meta_[db][tbname]->schema->uid = id_++;
     return *(builder_.get());
   }
 
   void createSubTable(const std::string& db, const std::string& stbname, const std::string& tbname, int16_t vgid) {
     std::unique_ptr<STableMeta> table;
-    if (TSDB_CODE_SUCCESS != copyTableMeta(db, stbname, &table)) {
+    if (TSDB_CODE_SUCCESS != copyTableSchemaMeta(db, stbname, &table)) {
-      throw std::runtime_error("copyTableMeta failed");
+      throw std::runtime_error("copyTableSchemaMeta failed");
     }
-    meta_[db][tbname].reset(table.release());
+    meta_[db][tbname].reset(new MockTableMeta());
-    meta_[db][tbname]->uid = id_++;
+    meta_[db][tbname]->schema.reset(table.release());
+    meta_[db][tbname]->schema->uid = id_++;
+    meta_[db][tbname]->vgs.emplace_back((SVgroupInfo){.vgId = vgid, .hashBegin = 0, .hashEnd = 0, .inUse = 0, .numOfEps = 3, .epAddr = {{"dnode_1", 6030}, {"dnode_2", 6030}, {"dnode_3", 6030}}});
+    // super table
+    meta_[db][stbname]->vgs.emplace_back((SVgroupInfo){.vgId = vgid, .hashBegin = 0, .hashEnd = 0, .inUse = 0, .numOfEps = 3, .epAddr = {{"dnode_1", 6030}, {"dnode_2", 6030}, {"dnode_3", 6030}}});
   }
 
   void showTables() const {
@@ -148,29 +150,43 @@ public:
       std::cout << SH("Table") << SH("Type") << SH("Precision") << IH("Vgid") << std::endl;
       std::cout << SL(3, 1) << std::endl;
       for (const auto& table : db.second) {
-        std::cout << SF(table.first) << SF(ttToString(table.second->tableType)) << SF(pToString(table.second->tableInfo.precision)) << IF(table.second->vgId) << std::endl;
+        const auto& schema = table.second->schema;
+        std::cout << SF(table.first) << SF(ttToString(schema->tableType)) << SF(pToString(schema->tableInfo.precision)) << IF(schema->vgId) << std::endl;
       }
       std::cout << std::endl;
     }
 
     for (const auto& db : meta_) {
       for (const auto& table : db.second) {
+        const auto& schema = table.second->schema;
         std::cout << "Table:" << table.first << std::endl;
         std::cout << SH("Field") << SH("Type") << SH("DataType") << IH("Bytes") << std::endl;
         std::cout << SL(3, 1) << std::endl;
-        int16_t numOfTags = table.second->tableInfo.numOfTags;
+        int16_t numOfTags = schema->tableInfo.numOfTags;
-        int16_t numOfFields = numOfTags + table.second->tableInfo.numOfColumns;
+        int16_t numOfFields = numOfTags + schema->tableInfo.numOfColumns;
         for (int16_t i = 0; i < numOfFields; ++i) {
-          const SSchema* schema = table.second->schema + i;
+          const SSchema* col = schema->schema + i;
-          std::cout << SF(std::string(schema->name)) << SH(ftToString(i, numOfTags)) << SH(dtToString(schema->type)) << IF(schema->bytes) << std::endl;
+          std::cout << SF(std::string(col->name)) << SH(ftToString(i, numOfTags)) << SH(dtToString(col->type)) << IF(col->bytes) << std::endl;
         }
         std::cout << std::endl;
       }
     }
   }
 
+  std::shared_ptr<MockTableMeta> getTableMeta(const std::string& db, const std::string& tbname) const {
+    DbMetaCache::const_iterator it = meta_.find(db);
+    if (meta_.end() == it) {
+      return std::shared_ptr<MockTableMeta>();
+    }
+    TableMetaCache::const_iterator tit = it->second.find(tbname);
+    if (it->second.end() == tit) {
+      return std::shared_ptr<MockTableMeta>();
+    }
+    return tit->second;
+  }
+
 private:
-  typedef std::map<std::string, std::shared_ptr<STableMeta> > TableMetaCache;
+  typedef std::map<std::string, std::shared_ptr<MockTableMeta>> TableMetaCache;
   typedef std::map<std::string, TableMetaCache> DbMetaCache;
 
   std::string ttToString(int8_t tableType) const {
@@ -207,20 +223,13 @@ private:
     return (0 == colid ? "column" : (colid <= numOfTags ? "tag" : "column"));
   }
 
-  std::shared_ptr<STableMeta> getTableMeta(const std::string& db, const std::string& tbname) const {
+  std::shared_ptr<STableMeta> getTableSchemaMeta(const std::string& db, const std::string& tbname) const {
-    DbMetaCache::const_iterator it = meta_.find(db);
+    std::shared_ptr<MockTableMeta> table = getTableMeta(db, tbname);
-    if (meta_.end() == it) {
+    return table ? table->schema : std::shared_ptr<STableMeta>();
-      return std::shared_ptr<STableMeta>();
-    }
-    TableMetaCache::const_iterator tit = it->second.find(tbname);
-    if (it->second.end() == tit) {
-      return std::shared_ptr<STableMeta>();
-    }
-    return tit->second;
   }
 
-  int32_t copyTableMeta(const std::string& db, const std::string& tbname, std::unique_ptr<STableMeta>* dst) const {
+  int32_t copyTableSchemaMeta(const std::string& db, const std::string& tbname, std::unique_ptr<STableMeta>* dst) const {
-    std::shared_ptr<STableMeta> src = getTableMeta(db, tbname);
+    std::shared_ptr<STableMeta> src = getTableSchemaMeta(db, tbname);
     if (!src) {
       return TSDB_CODE_TSC_INVALID_TABLE_NAME;
     }
@@ -244,14 +253,6 @@ MockCatalogService::MockCatalogService() : impl_(new MockCatalogServiceImpl()) {
 MockCatalogService::~MockCatalogService() {
 }
 
-struct SCatalog* MockCatalogService::getCatalogHandle(const SEpSet* pMgmtEps) const {
-  return impl_->getCatalogHandle(pMgmtEps);
-}
-
-int32_t MockCatalogService::catalogGetMetaData(struct SCatalog* pCatalog, const SCatalogReq* pMetaReq, SMetaData* pMetaData) const {
-  return impl_->catalogGetMetaData(pCatalog, pMetaReq, pMetaData);
-}
-
 ITableBuilder& MockCatalogService::createTableBuilder(const std::string& db, const std::string& tbname, int8_t tableType, int32_t numOfColumns, int32_t numOfTags) {
   return impl_->createTableBuilder(db, tbname, tableType, numOfColumns, numOfTags);
 }
@@ -262,4 +263,16 @@ void MockCatalogService::createSubTable(const std::string& db, const std::string
 
 void MockCatalogService::showTables() const {
   impl_->showTables();
-}
+}
+
+std::shared_ptr<MockTableMeta> MockCatalogService::getTableMeta(const std::string& db, const std::string& tbname) const {
+  return impl_->getTableMeta(db, tbname);
+}
+
+int32_t MockCatalogService::catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle) const {
+  return impl_->catalogGetHandle(clusterId, catalogHandle);
+}
+
+int32_t MockCatalogService::catalogGetTableMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, const char* pTableName, STableMeta** pTableMeta) const {
+  return impl_->catalogGetTableMeta(pCatalog, pRpc, pMgmtEps, pDBName, pTableName, pTableMeta);
+}

source/libs/parser/test/mockCatalogService.h

@@ -18,6 +18,7 @@
 
 #include <memory>
 #include <string>
+#include <vector>
 
 #include "catalog.h"
 
@@ -41,19 +42,24 @@ public:
   virtual void done() = 0;
 };
 
-class MockCatalogServiceImpl;
+struct MockTableMeta {
+  std::shared_ptr<STableMeta> schema;
+  std::vector<SVgroupInfo> vgs;
+};
 
+class MockCatalogServiceImpl;
 class MockCatalogService {
 public:
-  static const int32_t numOfDataTypes = sizeof(tDataTypes) / sizeof(tDataTypes[0]);
-
   MockCatalogService();
   ~MockCatalogService();
-  struct SCatalog* getCatalogHandle(const SEpSet* pMgmtEps) const;
-  int32_t catalogGetMetaData(struct SCatalog* pCatalog, const SCatalogReq* pMetaReq, SMetaData* pMetaData) const;
   ITableBuilder& createTableBuilder(const std::string& db, const std::string& tbname, int8_t tableType, int32_t numOfColumns, int32_t numOfTags = 0);
   void createSubTable(const std::string& db, const std::string& stbname, const std::string& tbname, int16_t vgid);
   void showTables() const;
+  std::shared_ptr<MockTableMeta> getTableMeta(const std::string& db, const std::string& tbname) const;
+
+  // mock interface
+  int32_t catalogGetHandle(const char *clusterId, struct SCatalog** catalogHandle) const;
+  int32_t catalogGetTableMeta(struct SCatalog* pCatalog, void *pRpc, const SEpSet* pMgmtEps, const char* pDBName, const char* pTableName, STableMeta** pTableMeta) const;
 
 private:
   std::unique_ptr<MockCatalogServiceImpl> impl_;

source/libs/parser/test/parserTestMain.cpp

@@ -22,12 +22,12 @@
 class ParserEnv : public testing::Environment {
 public:
   virtual void SetUp() {
-    mockCatalogService.reset(new MockCatalogService());
+    initMetaDataEnv();
-    generateMetaData(mockCatalogService.get());
+    generateMetaData();
   }
 
   virtual void TearDown() {
-    mockCatalogService.reset();
+    destroyMetaDataEnv();
   }
 
   ParserEnv() {}

source/libs/planner/inc/plannerInt.h

@@ -100,7 +100,7 @@ int32_t queryPlanToString(struct SQueryPlanNode* pQueryNode, char** str);
 int32_t queryPlanToSql(struct SQueryPlanNode* pQueryNode, char** sql);
 
 int32_t createDag(SQueryPlanNode* pQueryNode, struct SCatalog* pCatalog, SQueryDag** pDag);
-int32_t setSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SArray* eps);
+int32_t setSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SEpAddr* ep);
 int32_t subPlanToString(const SSubplan *pPhyNode, char** str);
 int32_t stringToSubplan(const char* str, SSubplan** subplan);
 
@@ -117,6 +117,7 @@ void destroyQueryPlan(struct SQueryPlanNode* pQueryNode);
  */
 void* destroyQueryPhyPlan(struct SPhyNode* pQueryPhyNode);
 
+const char* opTypeToOpName(int32_t type);
 int32_t opNameToOpType(const char* name);
 
 #ifdef __cplusplus

source/libs/planner/src/physicalPlan.c

@@ -33,6 +33,10 @@ static const char* gOpName[] = {
 #undef INCLUDE_AS_NAME
 };
 
+const char* opTypeToOpName(int32_t type) {
+  return gOpName[type];
+}
+
 int32_t opNameToOpType(const char* name) {
   for (int32_t i = 1; i < sizeof(gOpName) / sizeof(gOpName[0]); ++i) {
     if (strcmp(name, gOpName[i])) {
@@ -42,17 +46,43 @@ int32_t opNameToOpType(const char* name) {
   return OP_Unknown;
 }
 
-static void toDataBlockSchema(SQueryPlanNode* pPlanNode, SDataBlockSchema* dataBlockSchema) {
+static bool toDataBlockSchema(SQueryPlanNode* pPlanNode, SDataBlockSchema* dataBlockSchema) {
-  SWAP(dataBlockSchema->pSchema, pPlanNode->pSchema, SSchema*);
   dataBlockSchema->numOfCols = pPlanNode->numOfCols;
+  dataBlockSchema->pSchema = malloc(sizeof(SSlotSchema) * pPlanNode->numOfCols);
+  if (NULL == dataBlockSchema->pSchema) {
+    return false;
+  }
+  memcpy(dataBlockSchema->pSchema, pPlanNode->pSchema, sizeof(SSlotSchema) * pPlanNode->numOfCols);
+  return true;
+}
+
+static bool cloneExprArray(SArray** dst, SArray* src) {
+  if (NULL == src) {
+    return true;
+  }
+  size_t size = taosArrayGetSize(src);
+  if (0 == size) {
+    return true;
+  }
+  *dst = taosArrayInit(size, POINTER_BYTES);
+  if (NULL == *dst) {
+    return false;
+  }
+  return (TSDB_CODE_SUCCESS == copyAllExprInfo(*dst, src, true) ? true : false);
 }
 
 static SPhyNode* initPhyNode(SQueryPlanNode* pPlanNode, int32_t type, int32_t size) {
   SPhyNode* node = (SPhyNode*)calloc(1, size);
+  if (NULL == node) {
+    return NULL;
+  }
   node->info.type = type;
-  node->info.name = gOpName[type];
+  node->info.name = opTypeToOpName(type);
-  SWAP(node->pTargets, pPlanNode->pExpr, SArray*);
+  if (!cloneExprArray(&node->pTargets, pPlanNode->pExpr) || !toDataBlockSchema(pPlanNode, &(node->targetSchema))) {
-  toDataBlockSchema(pPlanNode, &(node->targetSchema));
+    free(node);
+    return NULL;
+  }
+  return node;
 }
 
 static SPhyNode* initScanNode(SQueryPlanNode* pPlanNode, SQueryTableInfo* pTable, int32_t type, int32_t size) {
@@ -203,9 +233,6 @@ static void createSubplanByLevel(SPlanContext* pCxt, SQueryPlanNode* pRoot) {
   SSubplan* subplan = initSubplan(pCxt, QUERY_TYPE_MERGE);
   ++(pCxt->nextId.templateId);
   subplan->pNode = createPhyNode(pCxt, pRoot);
-  SArray* l0 = taosArrayInit(TARRAY_MIN_SIZE, POINTER_BYTES);
-  taosArrayPush(l0, &subplan);
-  taosArrayPush(pCxt->pDag->pSubplans, &l0);
   // todo deal subquery
 }
 
@@ -225,6 +252,6 @@ int32_t createDag(SQueryPlanNode* pQueryNode, struct SCatalog* pCatalog, SQueryD
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t setSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SArray* eps) {
+int32_t setSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SEpAddr* ep) {
   //todo
 }

source/libs/planner/src/planner.c

@@ -46,8 +46,8 @@ int32_t qCreateQueryDag(const struct SQueryStmtInfo* pQueryInfo, struct SEpSet*
   return TSDB_CODE_SUCCESS;
 }
 
-int32_t qSetSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SArray* eps) {
+int32_t qSetSubplanExecutionNode(SSubplan* subplan, uint64_t templateId, SEpAddr* ep) {
-  return setSubplanExecutionNode(subplan, templateId, eps);
+  return setSubplanExecutionNode(subplan, templateId, ep);
 }
 
 int32_t qSubPlanToString(const SSubplan *subplan, char** str) {

source/libs/planner/test/CMakeLists.txt

@@ -5,7 +5,12 @@ MESSAGE(STATUS "build planner unit test")
 SET(CMAKE_CXX_STANDARD 11)
 AUX_SOURCE_DIRECTORY(${CMAKE_CURRENT_SOURCE_DIR} SOURCE_LIST)
 
-ADD_EXECUTABLE(plannerTest ${SOURCE_LIST})
+ADD_EXECUTABLE(plannerTest
+        ${SOURCE_LIST}
+        "${SOURCE_LIST}/../../../parser/test/mockCatalog.cpp"
+        "${SOURCE_LIST}/../../../parser/test/mockCatalogService.cpp"
+)
+
 TARGET_LINK_LIBRARIES(
         plannerTest
         PUBLIC os util common planner parser catalog transport gtest function qcom
@@ -15,4 +20,5 @@ TARGET_INCLUDE_DIRECTORIES(
         plannerTest
         PUBLIC "${CMAKE_SOURCE_DIR}/include/libs/planner/"
         PRIVATE "${CMAKE_SOURCE_DIR}/source/libs/planner/inc"
+        PRIVATE "${CMAKE_SOURCE_DIR}/source/libs/parser/test"
 )

source/libs/planner/test/phyPlanTests.cpp

@@ -0,0 +1,131 @@
+/*
+ * 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 <gtest/gtest.h>
+
+#include "plannerInt.h"
+#include "mockCatalogService.h"
+
+using namespace std;
+using namespace testing;
+
+void* myCalloc(size_t nmemb, size_t size) {
+  if (void* p = calloc(nmemb, size)) {
+    return p;
+  }
+  throw bad_alloc();
+}
+
+class PhyPlanTest : public Test {
+protected:
+  void pushScan(const string& db, const string& table, int32_t scanOp) {
+    shared_ptr<MockTableMeta> meta = mockCatalogService->getTableMeta(db, table);
+    EXPECT_TRUE(meta);
+// typedef struct SQueryPlanNode {
+//   SArray             *pExpr;        // the query functions or sql aggregations
+//   int32_t             numOfExpr;  // number of result columns, which is also the number of pExprs
+// } SQueryPlanNode;
+    unique_ptr<SQueryPlanNode> scan((SQueryPlanNode*)calloc(1, sizeof(SQueryPlanNode)));
+    scan->info.type = scanOp;
+    scan->numOfCols = meta->schema->tableInfo.numOfColumns;
+    scan->pSchema = (SSchema*)myCalloc(1, sizeof(SSchema) * scan->numOfCols);
+    memcpy(scan->pSchema, meta->schema->schema, sizeof(SSchema) * scan->numOfCols);
+    //todo 'pExpr' 'numOfExpr'
+    scan->pExtInfo = createScanExtInfo(meta);
+    pushNode(scan.release());
+  }
+
+  int32_t run() {
+    SQueryDag* dag = nullptr;
+    int32_t code = createDag(logicPlan_.get(), nullptr, &dag);
+    dag_.reset(dag);
+    return code;
+  }
+
+  void explain() {
+    size_t level = taosArrayGetSize(dag_->pSubplans);
+    for (size_t i = 0; i < level; ++i) {
+      std::cout << "level " << i  << ":" << std::endl;
+      const SArray* subplans = (const SArray*)taosArrayGetP(dag_->pSubplans, i);
+      size_t num = taosArrayGetSize(subplans);
+      for (size_t j = 0; j < num; ++j) {
+        std::cout << "no " << j << ":" << std::endl;
+        char* str = nullptr;
+        ASSERT_EQ (TSDB_CODE_SUCCESS, qSubPlanToString((const SSubplan*)taosArrayGetP(subplans, j), &str));
+        std::cout << str << std::endl;
+        free(str);
+      }
+    }
+  }
+
+  SQueryDag* reslut() {
+    return dag_.get();
+  }
+
+private:
+  void pushNode(SQueryPlanNode* node) {
+    if (logicPlan_) {
+      // todo
+    } else {
+      logicPlan_.reset(node);
+    }
+  }
+
+  void copySchemaMeta(STableMeta** dst, const STableMeta* src) {
+    int32_t size = sizeof(STableMeta) + sizeof(SSchema) * (src->tableInfo.numOfTags + src->tableInfo.numOfColumns);
+    *dst = (STableMeta*)myCalloc(1, size);
+    memcpy(*dst, src, size);
+  }
+
+  void copyStorageMeta(SVgroupsInfo** dst, const std::vector<SVgroupInfo>& src) {
+    *dst = (SVgroupsInfo*)myCalloc(1, sizeof(SVgroupsInfo) + sizeof(SVgroupMsg) * src.size());
+    (*dst)->numOfVgroups = src.size();
+    for (int32_t i = 0; i < src.size(); ++i) {
+      (*dst)->vgroups[i].vgId = src[i].vgId;
+      (*dst)->vgroups[i].numOfEps = src[i].numOfEps;
+      memcpy((*dst)->vgroups[i].epAddr, src[i].epAddr, src[i].numOfEps);
+    }
+  }
+
+  SQueryTableInfo* createScanExtInfo(shared_ptr<MockTableMeta>& meta) {
+    SQueryTableInfo* info = (SQueryTableInfo*)myCalloc(1, sizeof(SQueryTableInfo));
+    info->pMeta = (STableMetaInfo*)myCalloc(1, sizeof(STableMetaInfo));
+    copySchemaMeta(&info->pMeta->pTableMeta, meta->schema.get());
+    copyStorageMeta(&info->pMeta->vgroupList, meta->vgs);
+    return info;
+  }
+
+  shared_ptr<MockTableMeta> meta_;
+  unique_ptr<SQueryPlanNode> logicPlan_;
+  unique_ptr<SQueryDag> dag_;
+};
+
+// select * from table
+TEST_F(PhyPlanTest, tableScanTest) {
+  pushScan("root.test", "t1", QNODE_TABLESCAN);
+  ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
+  explain();
+  SQueryDag* dag = reslut();
+  // todo check
+}
+
+// select * from supertable
+TEST_F(PhyPlanTest, superTableScanTest) {
+  pushScan("root.test", "st1", QNODE_TABLESCAN);
+  ASSERT_EQ(run(), TSDB_CODE_SUCCESS);
+  explain();
+  SQueryDag* dag = reslut();
+  // todo check
+}

source/libs/planner/test/plannerTests.cpp

@@ -20,6 +20,7 @@
 
 #include "taos.h"
 #include "parser.h"
+#include "mockCatalog.h"
 
 #pragma GCC diagnostic ignored "-Wwrite-strings"
 
@@ -27,9 +28,25 @@
 #pragma GCC diagnostic ignored "-Wunused-variable"
 #pragma GCC diagnostic ignored "-Wsign-compare"
 
-int main(int argc, char** argv) {
+class ParserEnv : public testing::Environment {
-  testing::InitGoogleTest(&argc, argv);
+public:
-  return RUN_ALL_TESTS();
+  virtual void SetUp() {
+    initMetaDataEnv();
+    generateMetaData();
+  }
+
+  virtual void TearDown() {
+    destroyMetaDataEnv();
+  }
+
+  ParserEnv() {}
+  virtual ~ParserEnv() {}
+};
+
+int main(int argc, char* argv[]) {
+	testing::AddGlobalTestEnvironment(new ParserEnv());
+	testing::InitGoogleTest(&argc, argv);
+	return RUN_ALL_TESTS();
 }
 
 TEST(testCase, planner_test) {

source/libs/qcom/src/querymsg.c

@@ -294,7 +294,7 @@ void msgInit() {
   tscBuildMsg[TSDB_SQL_DROP_DNODE] = tscBuildDropDnodeMsg;
   tscBuildMsg[TSDB_SQL_CFG_DNODE] = tscBuildCfgDnodeMsg;
   tscBuildMsg[TSDB_SQL_ALTER_TABLE] = tscBuildAlterTableMsg;
-  tscBuildMsg[TSDB_SQL_UPDATE_TAGS_VAL] = tscBuildUpdateTagMsg;
+  tscBuildMsg[TSDB_SQL_UPDATE_TAG_VAL] = tscBuildUpdateTagMsg;
   tscBuildMsg[TSDB_SQL_ALTER_DB] = tscAlterDbMsg;
   tscBuildMsg[TSDB_SQL_COMPACT_VNODE] = tscBuildCompactMsg;  
 

source/libs/scheduler/CMakeLists.txt

@@ -9,5 +9,7 @@ target_include_directories(
 
 target_link_libraries(
         scheduler
-        PRIVATE os util planner qcom common
+        PRIVATE os util planner qcom common catalog transport
-)
+)
+
+ADD_SUBDIRECTORY(test)

source/libs/scheduler/inc/schedulerInt.h

@@ -27,6 +27,9 @@ extern "C" {
 #include "thash.h"
 
 #define SCHEDULE_DEFAULT_JOB_NUMBER 1000
+#define SCHEDULE_DEFAULT_TASK_NUMBER 1000
+
+#define SCH_MAX_CONDIDATE_EP_NUM TSDB_MAX_REPLICA
 
 enum {
   SCH_STATUS_NOT_START = 1,
@@ -39,22 +42,27 @@ enum {
 
 typedef struct SSchedulerMgmt {
   uint64_t  taskId;
+  SSchedulerCfg cfg;
   SHashObj *Jobs;  // key: queryId, value: SQueryJob*
 } SSchedulerMgmt;
 
 typedef struct SQueryTask {
-  uint64_t             taskId;  // task id
+  uint64_t             taskId;     // task id
-  char                *msg;     // operator tree
+  SSubplan            *plan;       // subplan
-  int8_t               status;  // task status
+  char                *msg;        // operator tree
-  SQueryProfileSummary summary; // task execution summary
+  int8_t               status;     // task status
+  SEpAddr              execAddr;   // task actual executed node address
+  SQueryProfileSummary summary;    // task execution summary
+  int32_t              childReady; // child task ready number
+  SArray              *children;   // the datasource tasks,from which to fetch the result, element is SQueryTask*
+  SArray              *parents;    // the data destination tasks, get data from current task, element is SQueryTask*
 } SQueryTask;
 
 typedef struct SQueryLevel {
+  int32_t level;
   int8_t  status;
   int32_t taskNum;
-  
   SArray *subTasks;  // Element is SQueryTask
-  SArray *subPlans;  // Element is SSubplan
 } SQueryLevel;
 
 typedef struct SQueryJob {
@@ -63,13 +71,28 @@ typedef struct SQueryJob {
   int32_t   levelIdx;
   int8_t    status;
   SQueryProfileSummary summary;
-  
+  SEpSet    dataSrcEps;
-  SArray  *levels;    // Element is SQueryLevel, starting from 0.
+  SEpAddr   resEp;
-  SArray  *subPlans;  // Element is SArray*, and nested element is SSubplan. The execution level of subplan, starting from 0.
+  void            *transport;
+  SArray          *qnodeList;
+  tsem_t           rspSem;
+  int32_t          userFetch;
+  int32_t          remoteFetch;
+  void            *res;
+
+  SHashObj *execTasks; // executing tasks, key:taskid, value:SQueryTask*
+  SHashObj *succTasks; // succeed tasks, key:taskid, value:SQueryTask*
+    
+  SArray   *levels;    // Element is SQueryLevel, starting from 0.
+  SArray   *subPlans;  // Element is SArray*, and nested element is SSubplan. The execution level of subplan, starting from 0.
 } SQueryJob;
 
+#define SCH_HAS_QNODE_IN_CLUSTER(type) (false) //TODO CLUSTER TYPE
+#define SCH_TASK_READY_TO_LUNCH(task) ((task)->childReady >= taosArrayGetSize((task)->children))   // MAY NEED TO ENHANCE
+#define SCH_IS_DATA_SRC_TASK(task) (task->plan->type == QUERY_TYPE_SCAN)
 
 #define SCH_JOB_ERR_LOG(param, ...) qError("QID:%"PRIx64 param, job->queryId, __VA_ARGS__)
+#define SCH_TASK_ERR_LOG(param, ...) qError("QID:%"PRIx64",TID:%"PRIx64 param, job->queryId, task->taskId, __VA_ARGS__)
 
 #define SCH_ERR_RET(c) do { int32_t _code = c; if (_code != TSDB_CODE_SUCCESS) { terrno = _code; return _code; } } while (0)
 #define SCH_RET(c) do { int32_t _code = c; if (_code != TSDB_CODE_SUCCESS) { terrno = _code; } return _code; } while (0)
@@ -77,6 +100,8 @@ typedef struct SQueryJob {
 #define SCH_ERR_JRET(c) do { code = c; if (code != TSDB_CODE_SUCCESS) { terrno = code; goto _return; } } while (0)
 
 
+extern int32_t schLaunchTask(SQueryJob *job, SQueryTask *task);
+
 #ifdef __cplusplus
 }
 #endif

source/libs/scheduler/src/scheduler.c

@@ -16,6 +16,7 @@
 #include "schedulerInt.h"
 #include "taosmsg.h"
 #include "query.h"
+#include "catalog.h"
 
 SSchedulerMgmt schMgmt = {0};
 
@@ -50,23 +51,109 @@ int32_t schBuildAndSendRequest(void *pRpc, const SEpSet* pMgmtEps, __taos_async_
 */  
 }
 
+int32_t schBuildTaskRalation(SQueryJob *job, SHashObj *planToTask) {
+  for (int32_t i = 0; i < job->levelNum; ++i) {
+    SQueryLevel *level = taosArrayGet(job->levels, i);
+    
+    for (int32_t m = 0; m < level->taskNum; ++m) {
+      SQueryTask *task = taosArrayGet(level->subTasks, m);
+      SSubplan *plan = task->plan;
+      int32_t childNum = plan->pChildern ? (int32_t)taosArrayGetSize(plan->pChildern) : 0;
+      int32_t parentNum = plan->pParents ? (int32_t)taosArrayGetSize(plan->pParents) : 0;
+
+      if (childNum > 0) {
+        task->children = taosArrayInit(childNum, POINTER_BYTES);
+        if (NULL == task->children) {
+          qError("taosArrayInit %d failed", childNum);
+          SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+        }
+      }
+
+      for (int32_t n = 0; n < childNum; ++n) {
+        SSubplan *child = taosArrayGet(plan->pChildern, n);
+        SQueryTask *childTask = taosHashGet(planToTask, &child, POINTER_BYTES);
+        if (childTask) {
+          qError("subplan relationship error, level:%d, taskIdx:%d, childIdx:%d", i, m, n);
+          SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
+        }
+
+        if (NULL == taosArrayPush(task->children, &childTask)) {
+          qError("taosArrayPush failed");
+          SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+        }
+      }
+
+      if (parentNum > 0) {
+        task->parents = taosArrayInit(parentNum, POINTER_BYTES);
+        if (NULL == task->parents) {
+          qError("taosArrayInit %d failed", parentNum);
+          SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+        }
+      }
+
+      for (int32_t n = 0; n < parentNum; ++n) {
+        SSubplan *parent = taosArrayGet(plan->pParents, n);
+        SQueryTask *parentTask = taosHashGet(planToTask, &parent, POINTER_BYTES);
+        if (parentTask) {
+          qError("subplan relationship error, level:%d, taskIdx:%d, childIdx:%d", i, m, n);
+          SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
+        }
+
+        if (NULL == taosArrayPush(task->parents, &parentTask)) {
+          qError("taosArrayPush failed");
+          SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+        }
+      }      
+    }
+  }
+
+  SQueryLevel *level = taosArrayGet(job->levels, 0);
+  if (level->taskNum > 1) {
+    qError("invalid plan info, level 0, taskNum:%d", level->taskNum);
+    SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
+  }
+
+  SQueryTask *task = taosArrayGet(level->subTasks, 0);
+  if (task->parents && taosArrayGetSize(task->parents) > 0) {
+    qError("invalid plan info, level 0, parentNum:%d", (int32_t)taosArrayGetSize(task->parents));
+    SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
+  }
+
+
+  return TSDB_CODE_SUCCESS;
+}
+
+
 int32_t schValidateAndBuildJob(SQueryDag *dag, SQueryJob *job) {
   int32_t code = 0;
+
+  job->queryId = dag->queryId;
+  
+  if (dag->numOfSubplans <= 0) {
+    qError("invalid subplan num:%d", dag->numOfSubplans);
+    SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT);
+  }
+  
   int32_t levelNum = (int32_t)taosArrayGetSize(dag->pSubplans);
   if (levelNum <= 0) {
     qError("invalid level num:%d", levelNum);
     SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT);
   }
 
+  SHashObj *planToTask = taosHashInit(SCHEDULE_DEFAULT_TASK_NUMBER, taosGetDefaultHashFunction(POINTER_BYTES == sizeof(int64_t) ? TSDB_DATA_TYPE_BIGINT : TSDB_DATA_TYPE_INT), false, HASH_NO_LOCK);
+  if (NULL == planToTask) {
+    qError("taosHashInit %d failed", SCHEDULE_DEFAULT_TASK_NUMBER);
+    SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+  }
+  
   job->levels = taosArrayInit(levelNum, sizeof(SQueryLevel));
   if (NULL == job->levels) {
     qError("taosArrayInit %d failed", levelNum);
-    SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+    SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
   }
 
   job->levelNum = levelNum;
   job->levelIdx = levelNum - 1;
-  job->status = SCH_STATUS_NOT_START;
 
   job->subPlans = dag->pSubplans;
 
@@ -77,32 +164,45 @@ int32_t schValidateAndBuildJob(SQueryDag *dag, SQueryJob *job) {
   level.status = SCH_STATUS_NOT_START;
 
   for (int32_t i = 0; i < levelNum; ++i) {
+    level.level = i;
     levelPlans = taosArrayGetP(dag->pSubplans, i);
     if (NULL == levelPlans) {
       qError("no level plans for level %d", i);
-      SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT);
+      SCH_ERR_JRET(TSDB_CODE_QRY_INVALID_INPUT);
     }
 
     levelPlanNum = (int32_t)taosArrayGetSize(levelPlans);
     if (levelPlanNum <= 0) {
       qError("invalid level plans number:%d, level:%d", levelPlanNum, i);
-      SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT);
+      SCH_ERR_JRET(TSDB_CODE_QRY_INVALID_INPUT);
     }
 
     level.taskNum = levelPlanNum;
-    level.subPlans = levelPlans;
     
     level.subTasks = taosArrayInit(levelPlanNum, sizeof(SQueryTask));
     if (NULL == level.subTasks) {
       qError("taosArrayInit %d failed", levelPlanNum);
-      SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
     }
     
     for (int32_t n = 0; n < levelPlanNum; ++n) {
-      SQueryTask *task = taosArrayGet(level.subTasks, n);
+      SSubplan *plan = taosArrayGet(levelPlans, n);
+      SQueryTask task = {0};
       
-      task->taskId = atomic_add_fetch_64(&schMgmt.taskId, 1);
+      task.taskId = atomic_add_fetch_64(&schMgmt.taskId, 1);
-      task->status = SCH_STATUS_NOT_START;
+      task.plan = plan;
+      task.status = SCH_STATUS_NOT_START;
+
+      void *p = taosArrayPush(level.subTasks, &task);
+      if (NULL == p) {
+        qError("taosArrayPush failed");
+        SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      }
+      
+      if (0 != taosHashPut(planToTask, &plan, POINTER_BYTES, &p, POINTER_BYTES)) {
+        qError("taosHashPut failed");
+        SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      }
     }
 
     if (NULL == taosArrayPush(job->levels, &level)) {
@@ -111,6 +211,12 @@ int32_t schValidateAndBuildJob(SQueryDag *dag, SQueryJob *job) {
     }
   }
 
+  SCH_ERR_JRET(schBuildTaskRalation(job, planToTask));
+
+  if (planToTask) {
+    taosHashCleanup(planToTask);
+  }
+  
   return TSDB_CODE_SUCCESS;
 
 _return:
@@ -118,20 +224,331 @@ _return:
     taosArrayDestroy(level.subTasks);
   }
 
+  if (planToTask) {
+    taosHashCleanup(planToTask);
+  }
+
   SCH_RET(code);
 }
 
-
+int32_t schSetTaskExecEpSet(SQueryJob *job, SEpSet *epSet) {  
-int32_t schJobExecute(SQueryJob *job) {
+  if (epSet->numOfEps >= SCH_MAX_CONDIDATE_EP_NUM) {
-  switch (job->status) {
+    return TSDB_CODE_SUCCESS;
-    case SCH_STATUS_NOT_START:
-
-      break;
-      
-    default:
-      SCH_JOB_ERR_LOG("invalid job status:%d", job->status);
-      SCH_ERR_RET(TSDB_CODE_SCH_STATUS_ERROR);
   }
+
+  int32_t qnodeNum = taosArrayGetSize(job->qnodeList);
+  
+  for (int32_t i = 0; i < qnodeNum && epSet->numOfEps < tListLen(epSet->port); ++i) {
+    SEpAddr *addr = taosArrayGet(job->qnodeList, i);
+    
+    strncpy(epSet->fqdn[epSet->numOfEps], addr->fqdn, sizeof(addr->fqdn));
+    epSet->port[epSet->numOfEps] = addr->port;
+    
+    ++epSet->numOfEps;
+  }
+
+  for (int32_t i = 0; i < job->dataSrcEps.numOfEps && epSet->numOfEps < tListLen(epSet->port); ++i) {
+    strncpy(epSet->fqdn[epSet->numOfEps], job->dataSrcEps.fqdn[i], sizeof(job->dataSrcEps.fqdn[i]));
+    epSet->port[epSet->numOfEps] = job->dataSrcEps.port[i];
+    
+    ++epSet->numOfEps;
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+
+int32_t schPushTaskToExecList(SQueryJob *job, SQueryTask *task) {
+  if (0 != taosHashPut(job->execTasks, &task->taskId, sizeof(task->taskId), &task, POINTER_BYTES)) {
+    qError("taosHashPut failed");
+    SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schMoveTaskToSuccList(SQueryJob *job, SQueryTask *task, bool *moved) {
+  if (0 != taosHashRemove(job->execTasks, &task->taskId, sizeof(task->taskId))) {
+    qWarn("remove task[%"PRIx64"] from execTasks failed", task->taskId);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  if (0 != taosHashPut(job->execTasks, &task->taskId, sizeof(task->taskId), &task, POINTER_BYTES)) {
+    qError("taosHashPut failed");
+    SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+  }
+
+  *moved = true;
+  
+  return TSDB_CODE_SUCCESS;
+}
+
+
+int32_t schAsyncSendMsg(SQueryJob *job, SQueryTask *task, int32_t msgType) {
+  int32_t msgSize = 0;
+  void *msg = NULL;
+  
+  switch (msgType) {
+    case TSDB_MSG_TYPE_QUERY: {
+      if (NULL == task->msg) {
+        qError("query msg is NULL");
+        SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
+      }
+
+      int32_t len = strlen(task->msg);
+      msgSize = sizeof(SSchedulerQueryMsg) + len;
+      msg = calloc(1, msgSize);
+      if (NULL == msg) {
+        qError("calloc %d failed", msgSize);
+        SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      }
+
+      SSchedulerQueryMsg *pMsg = msg;
+      pMsg->queryId = job->queryId;
+      pMsg->taskId = task->taskId;
+      pMsg->contentLen = len;
+      memcpy(pMsg->msg, task->msg, len);
+      break;
+    }
+    case TSDB_MSG_TYPE_RSP_READY: {
+      msgSize = sizeof(SSchedulerReadyMsg);
+      msg = calloc(1, msgSize);
+      if (NULL == msg) {
+        qError("calloc %d failed", msgSize);
+        SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      }
+
+      SSchedulerReadyMsg *pMsg = msg;
+      pMsg->queryId = job->queryId;
+      pMsg->taskId = task->taskId;      
+      break;
+    }
+    case TSDB_MSG_TYPE_FETCH: {
+      msgSize = sizeof(SSchedulerFetchMsg);
+      msg = calloc(1, msgSize);
+      if (NULL == msg) {
+        qError("calloc %d failed", msgSize);
+        SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+      }
+    
+      SSchedulerFetchMsg *pMsg = msg;
+      pMsg->queryId = job->queryId;
+      pMsg->taskId = task->taskId;      
+      break;
+    }
+    default:
+      qError("unknown msg type:%d", msgType);
+      break;
+  }
+
+  //TODO SEND MSG
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schTaskCheckAndSetRetry(SQueryJob *job, SQueryTask *task, int32_t errCode, bool *needRetry) {
+  // TODO set retry or not based on task type/errCode/retry times/job status/available eps...
+  // TODO if needRetry, set task retry info
+
+  *needRetry = false;
+
+  return TSDB_CODE_SUCCESS;
+}
+
+
+int32_t schFetchFromRemote(SQueryJob *job) {
+  int32_t code = 0;
+  
+  if (atomic_val_compare_exchange_32(&job->remoteFetch, 0, 1) != 0) {
+    qInfo("prior fetching not finished");
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SCH_ERR_JRET(schAsyncSendMsg(job, NULL, TSDB_MSG_TYPE_FETCH));
+
+  return TSDB_CODE_SUCCESS;
+  
+_return:
+  atomic_val_compare_exchange_32(&job->remoteFetch, 1, 0);
+
+  return code;
+}
+
+
+int32_t schProcessOnJobSuccess(SQueryJob *job) {
+  job->status = SCH_STATUS_SUCCEED;
+  
+  if (job->userFetch) {
+    SCH_ERR_RET(schFetchFromRemote(job));
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schProcessOnJobFailure(SQueryJob *job) {
+  job->status = SCH_STATUS_FAILED;
+
+  atomic_val_compare_exchange_32(&job->remoteFetch, 1, 0);
+
+  if (job->userFetch) {
+    tsem_post(&job->rspSem);
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schProcessOnDataFetched(SQueryJob *job) {
+  atomic_val_compare_exchange_32(&job->remoteFetch, 1, 0);
+
+  tsem_post(&job->rspSem);
+}
+
+
+int32_t schProcessOnTaskSuccess(SQueryJob *job, SQueryTask *task) {
+  bool moved = false;
+  
+  SCH_ERR_RET(schMoveTaskToSuccList(job, task, &moved));
+  if (!moved) {
+    SCH_TASK_ERR_LOG("task may already moved, status:%d", task->status);
+    return TSDB_CODE_SUCCESS;
+  }
+
+  task->status = SCH_STATUS_SUCCEED;
+  
+  int32_t parentNum = (int32_t)taosArrayGetSize(task->parents);
+  if (parentNum == 0) {
+    if (task->plan->level != 0) {
+      qError("level error");
+      SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT);
+    }
+
+    strncpy(job->resEp.fqdn, task->execAddr.fqdn, sizeof(job->resEp.fqdn));
+    job->resEp.port = task->execAddr.port;
+
+    SCH_ERR_RET(schProcessOnJobSuccess(job));
+
+    return TSDB_CODE_SUCCESS;
+  }
+
+  if (SCH_IS_DATA_SRC_TASK(task) && job->dataSrcEps.numOfEps < SCH_MAX_CONDIDATE_EP_NUM) {
+    strncpy(job->dataSrcEps.fqdn[job->dataSrcEps.numOfEps], task->execAddr.fqdn, sizeof(task->execAddr.fqdn));
+    job->dataSrcEps.port[job->dataSrcEps.numOfEps] = task->execAddr.port;
+
+    ++job->dataSrcEps.numOfEps;
+  }
+
+  for (int32_t i = 0; i < parentNum; ++i) {
+    SQueryTask *par = taosArrayGet(task->parents, i);
+
+    ++par->childReady;
+
+    SCH_ERR_RET(qSetSubplanExecutionNode(par->plan, task->plan->id.templateId, &task->execAddr));
+    
+    if (SCH_TASK_READY_TO_LUNCH(par)) {
+      SCH_ERR_RET(schLaunchTask(job, task));
+    }
+  }
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schProcessOnTaskFailure(SQueryJob *job, SQueryTask *task, int32_t errCode) {
+  bool needRetry = false;
+  SCH_ERR_RET(schTaskCheckAndSetRetry(job, task, errCode, &needRetry));
+  
+  if (!needRetry) {
+    SCH_TASK_ERR_LOG("task failed[%x], no more retry", errCode);
+    
+    job->status = SCH_STATUS_FAILED;
+    SCH_ERR_RET(schProcessOnJobFailure(job));
+
+    return TSDB_CODE_SUCCESS;
+  }
+
+  SCH_ERR_RET(schLaunchTask(job, task));
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schHandleRspMsg(SQueryJob *job, SQueryTask *task, int32_t msgType, int32_t rspCode) {
+  int32_t code = 0;
+  
+  switch (msgType) {
+    case TSDB_MSG_TYPE_QUERY:
+      if (rspCode != TSDB_CODE_SUCCESS) {
+        SCH_ERR_JRET(schProcessOnTaskFailure(job, task, rspCode));
+      } else {
+        code = schAsyncSendMsg(job, task, TSDB_MSG_TYPE_RSP_READY);
+        if (code) {
+          goto _task_error;
+        }
+      }
+      break;
+    case TSDB_MSG_TYPE_RSP_READY:
+      if (rspCode != TSDB_CODE_SUCCESS) {
+        SCH_ERR_JRET(schProcessOnTaskFailure(job, task, rspCode));
+      } else {
+        code = schProcessOnTaskSuccess(job, task);
+        if (code) {
+          goto _task_error;
+        }        
+      }
+      break;
+    case TSDB_MSG_TYPE_FETCH:
+      SCH_ERR_JRET(rspCode);
+      SCH_ERR_JRET(schProcessOnDataFetched(job));
+      break;
+    default:
+      qError("unknown msg type:%d received", msgType);
+      return TSDB_CODE_QRY_INVALID_INPUT;
+  }
+
+  return TSDB_CODE_SUCCESS;
+
+_task_error:
+  SCH_ERR_JRET(schProcessOnTaskFailure(job, task, code));
+  return TSDB_CODE_SUCCESS;
+
+_return:
+  code = schProcessOnJobFailure(job);
+  return code;
+}
+
+
+
+
+int32_t schLaunchTask(SQueryJob *job, SQueryTask *task) {
+  SSubplan *plan = task->plan;
+  
+  SCH_ERR_RET(qSubPlanToString(plan, &task->msg));
+  if (plan->execEpSet.numOfEps <= 0) {
+    SCH_ERR_RET(schSetTaskExecEpSet(job, &plan->execEpSet));
+  }
+
+  if (plan->execEpSet.numOfEps <= 0) {
+    SCH_TASK_ERR_LOG("invalid execEpSet num:%d", plan->execEpSet.numOfEps);
+    SCH_ERR_RET(TSDB_CODE_SCH_INTERNAL_ERROR);
+  }
+  
+  SCH_ERR_RET(schAsyncSendMsg(job, task, TSDB_MSG_TYPE_QUERY));
+
+  SCH_ERR_RET(schPushTaskToExecList(job, task));
+
+  task->status = SCH_STATUS_EXECUTING;
+
+  return TSDB_CODE_SUCCESS;
+}
+
+int32_t schLaunchJob(SQueryJob *job) {
+  SQueryLevel *level = taosArrayGet(job->levels, job->levelIdx);
+  for (int32_t i = 0; i < level->taskNum; ++i) {
+    SQueryTask *task = taosArrayGet(level->subTasks, i);
+    SCH_ERR_RET(schLaunchTask(job, task));
+  }
+
+  job->status = SCH_STATUS_EXECUTING;
+  
+  return TSDB_CODE_SUCCESS;
 }
 
 
@@ -141,27 +558,59 @@ int32_t schedulerInit(SSchedulerCfg *cfg) {
     SCH_ERR_LRET(TSDB_CODE_QRY_OUT_OF_MEMORY, "init %d schduler jobs failed", SCHEDULE_DEFAULT_JOB_NUMBER);
   }
 
+  if (cfg) {
+    schMgmt.cfg = *cfg;
+  }
+
   return TSDB_CODE_SUCCESS;
 }
 
 
-int32_t scheduleQueryJob(SQueryDag* pDag, void** pJob) {
+int32_t scheduleExecJob(void *transport, SArray *qnodeList, SQueryDag* pDag, void** pJob) {
-  if (NULL == pDag || NULL == pDag->pSubplans || NULL == pJob) {
+  if (NULL == transport || NULL == transport ||NULL == pDag || NULL == pDag->pSubplans || NULL == pJob) {
     SCH_ERR_RET(TSDB_CODE_QRY_INVALID_INPUT);
   }
 
+  if (taosArrayGetSize(qnodeList) <= 0) {
+    qInfo("qnodeList is empty");
+  }
+
   int32_t code = 0;
   SQueryJob *job = calloc(1, sizeof(SQueryJob));
   if (NULL == job) {
     SCH_ERR_RET(TSDB_CODE_QRY_OUT_OF_MEMORY);
   }
 
+  job->transport = transport;
+  job->qnodeList = qnodeList;
+
   SCH_ERR_JRET(schValidateAndBuildJob(pDag, job));
 
-  SCH_ERR_JRET(schJobExecute(job));
+  job->execTasks = taosHashInit(pDag->numOfSubplans, taosGetDefaultHashFunction(TSDB_DATA_TYPE_UBIGINT), false, HASH_ENTRY_LOCK);
+  if (NULL == job->execTasks) {
+    qError("taosHashInit %d failed", pDag->numOfSubplans);
+    SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+  }
+
+  job->succTasks = taosHashInit(pDag->numOfSubplans, taosGetDefaultHashFunction(TSDB_DATA_TYPE_UBIGINT), false, HASH_ENTRY_LOCK);
+  if (NULL == job->succTasks) {
+    qError("taosHashInit %d failed", pDag->numOfSubplans);
+    SCH_ERR_JRET(TSDB_CODE_QRY_OUT_OF_MEMORY);
+  }
+
+  tsem_init(&job->rspSem, 0, 0);
+  
+  if (0 != taosHashPut(schMgmt.Jobs, &job->queryId, sizeof(job->queryId), &job, POINTER_BYTES)) {
+    qError("taosHashPut queryId:%"PRIx64" failed", job->queryId);
+    SCH_ERR_JRET(TSDB_CODE_SCH_INTERNAL_ERROR);
+  }
+
+  job->status = SCH_STATUS_NOT_START;
+  
+  SCH_ERR_JRET(schLaunchJob(job));
 
   *(SQueryJob **)pJob = job;
-  
+
   return TSDB_CODE_SUCCESS;
 
 _return:
@@ -172,17 +621,64 @@ _return:
   SCH_RET(code);
 }
 
-int32_t scheduleFetchRows(void *pJob, void *data);
+int32_t scheduleFetchRows(void *pJob, void **data) {
+  if (NULL == pJob || NULL == data) {
+    return TSDB_CODE_QRY_INVALID_INPUT;
+  }
 
-int32_t scheduleCancelJob(void *pJob);
+  SQueryJob *job = pJob;
+  int32_t code = 0;
+
+  if (atomic_val_compare_exchange_32(&job->userFetch, 0, 1) != 0) {
+    qError("prior fetching not finished");
+    return TSDB_CODE_QRY_APP_ERROR;
+  }
+
+  if (job->status == SCH_STATUS_SUCCEED) {
+    SCH_ERR_JRET(schFetchFromRemote(job));
+  }
+
+  tsem_wait(&job->rspSem);
+
+  *data = job->res;
+  job->res = NULL;
+
+_return:
+  atomic_val_compare_exchange_32(&job->userFetch, 1, 0);
+
+  return code;
+}
+
+int32_t scheduleCancelJob(void *pJob) {
+  //TODO
+
+  return TSDB_CODE_SUCCESS;
+}
 
 void scheduleFreeJob(void *pJob) {
+  if (NULL == pJob) {
+    return;
+  }
 
+  SQueryJob *job = pJob;
+
+  if (job->status > 0) {
+    if (0 != taosHashRemove(schMgmt.Jobs, &job->queryId, sizeof(job->queryId))) {
+      qError("remove job:%"PRIx64"from mgmt failed", job->queryId); // maybe already freed
+      return;
+    }
+
+    if (job->status == SCH_STATUS_EXECUTING) {
+      scheduleCancelJob(pJob);
+    }
+  }
+  
+  //TODO free job
 }
 
 void schedulerDestroy(void) {
   if (schMgmt.Jobs) {
-    taosHashCleanup(schMgmt.Jobs); //TBD
+    taosHashCleanup(schMgmt.Jobs); //TODO
     schMgmt.Jobs = NULL;
   }
 }

source/libs/scheduler/test/CMakeLists.txt

@@ -0,0 +1,18 @@
+
+MESSAGE(STATUS "build scheduler unit test")
+
+# GoogleTest requires at least C++11
+SET(CMAKE_CXX_STANDARD 11)
+AUX_SOURCE_DIRECTORY(${CMAKE_CURRENT_SOURCE_DIR} SOURCE_LIST)
+
+ADD_EXECUTABLE(schedulerTest ${SOURCE_LIST})
+TARGET_LINK_LIBRARIES(
+        schedulerTest
+        PUBLIC os util common catalog transport gtest qcom taos planner scheduler
+)
+
+TARGET_INCLUDE_DIRECTORIES(
+        schedulerTest
+        PUBLIC "${CMAKE_SOURCE_DIR}/include/libs/scheduler/"
+        PRIVATE "${CMAKE_SOURCE_DIR}/source/libs/scheduler/inc"
+)

source/libs/scheduler/test/schedulerTests.cpp

@@ -0,0 +1,104 @@
+/*
+ * 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 <gtest/gtest.h>
+#include <tglobal.h>
+#include <iostream>
+#pragma GCC diagnostic ignored "-Wwrite-strings"
+
+#pragma GCC diagnostic ignored "-Wunused-function"
+#pragma GCC diagnostic ignored "-Wunused-variable"
+#pragma GCC diagnostic ignored "-Wsign-compare"
+#include "os.h"
+
+#include "taos.h"
+#include "tdef.h"
+#include "tvariant.h"
+#include "catalog.h"
+#include "scheduler.h"
+#include "tep.h"
+#include "trpc.h"
+
+namespace {
+void mockBuildDag(SQueryDag *dag) {
+  uint64_t qId = 0x111111111111;
+  
+  dag->queryId = qId;
+  dag->numOfSubplans = 2;
+  dag->pSubplans = taosArrayInit(dag->numOfSubplans, POINTER_BYTES);
+  SArray *scan = taosArrayInit(1, sizeof(SSubplan));
+  SArray *merge = taosArrayInit(1, sizeof(SSubplan));
+  
+  SSubplan scanPlan = {0};
+  SSubplan mergePlan = {0};
+
+  scanPlan.id.queryId = qId;
+  scanPlan.id.templateId = 0x2222222222;
+  scanPlan.id.subplanId = 0x3333333333;
+  scanPlan.type = QUERY_TYPE_SCAN;
+  scanPlan.level = 1;
+  scanPlan.execEpSet.numOfEps = 1;
+  scanPlan.pChildern = NULL;
+  scanPlan.pParents = taosArrayInit(1, POINTER_BYTES);
+
+  mergePlan.id.queryId = qId;
+  mergePlan.id.templateId = 0x4444444444;
+  mergePlan.id.subplanId = 0x5555555555;
+  mergePlan.type = QUERY_TYPE_MERGE;
+  mergePlan.level = 0;
+  mergePlan.execEpSet.numOfEps = 1;
+  mergePlan.pChildern = taosArrayInit(1, POINTER_BYTES);
+  mergePlan.pParents = NULL;
+
+  SSubplan *mergePointer = (SSubplan *)taosArrayPush(merge, &mergePlan);
+  SSubplan *scanPointer = (SSubplan *)taosArrayPush(scan, &scanPlan);
+
+  taosArrayPush(mergePointer->pChildern, &scanPointer);
+  taosArrayPush(scanPointer->pParents, &mergePointer);
+
+  taosArrayPush(dag->pSubplans, &merge);  
+  taosArrayPush(dag->pSubplans, &scan);
+}
+
+}
+
+TEST(testCase, normalCase) {
+  void *mockPointer = (void *)0x1;
+  char *clusterId = "cluster1";
+  char *dbname = "1.db1";
+  char *tablename = "table1";
+  SVgroupInfo vgInfo = {0};
+  void *pJob = NULL;
+  SQueryDag dag = {0};
+  SArray *qnodeList = taosArrayInit(1, sizeof(SEpAddr));
+  
+  int32_t code = schedulerInit(NULL);
+  ASSERT_EQ(code, 0);
+
+  mockBuildDag(&dag);
+  
+  code = scheduleExecJob(mockPointer, qnodeList, &dag, &pJob);
+  ASSERT_EQ(code, 0);
+}
+
+
+int main(int argc, char** argv) {
+  testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}
+
+
+
+

source/util/src/terror.c

@@ -499,6 +499,7 @@ TAOS_DEFINE_ERROR(TSDB_CODE_CTG_SYS_ERROR,                "catalog system error"
 
 //scheduler
 TAOS_DEFINE_ERROR(TSDB_CODE_SCH_STATUS_ERROR,             "scheduler status error")
+TAOS_DEFINE_ERROR(TSDB_CODE_SCH_INTERNAL_ERROR,           "scheduler internal error")
 
 
 #ifdef TAOS_ERROR_C

source/util/src/tlog.c

@@ -14,21 +14,21 @@
  */
 
 #define _DEFAULT_SOURCE
-#include "os.h"
-#include "ulog.h"
 #include "tlog.h"
+#include "os.h"
 #include "tnote.h"
 #include "tutil.h"
+#include "ulog.h"
 #include "zlib.h"
-  
+
-#define MAX_LOGLINE_SIZE              (1000)
+#define MAX_LOGLINE_SIZE (1000)
-#define MAX_LOGLINE_BUFFER_SIZE       (MAX_LOGLINE_SIZE + 10)
+#define MAX_LOGLINE_BUFFER_SIZE (MAX_LOGLINE_SIZE + 10)
-#define MAX_LOGLINE_CONTENT_SIZE      (MAX_LOGLINE_SIZE - 100)
+#define MAX_LOGLINE_CONTENT_SIZE (MAX_LOGLINE_SIZE - 100)
-#define MAX_LOGLINE_DUMP_SIZE         (65 * 1024)
+#define MAX_LOGLINE_DUMP_SIZE (65 * 1024)
-#define MAX_LOGLINE_DUMP_BUFFER_SIZE  (MAX_LOGLINE_DUMP_SIZE + 10)
+#define MAX_LOGLINE_DUMP_BUFFER_SIZE (MAX_LOGLINE_DUMP_SIZE + 10)
 #define MAX_LOGLINE_DUMP_CONTENT_SIZE (MAX_LOGLINE_DUMP_SIZE - 100)
 
-#define LOG_FILE_NAME_LEN          300
+#define LOG_FILE_NAME_LEN 300
 #define TSDB_DEFAULT_LOG_BUF_SIZE (20 * 1024 * 1024)  // 20MB
 
 #define DEFAULT_LOG_INTERVAL 25
@@ -38,13 +38,13 @@
 #define LOG_MAX_WAIT_MSEC 1000
 
 #define LOG_BUF_BUFFER(x) ((x)->buffer)
-#define LOG_BUF_START(x)  ((x)->buffStart)
+#define LOG_BUF_START(x) ((x)->buffStart)
-#define LOG_BUF_END(x)    ((x)->buffEnd)
+#define LOG_BUF_END(x) ((x)->buffEnd)
-#define LOG_BUF_SIZE(x)   ((x)->buffSize)
+#define LOG_BUF_SIZE(x) ((x)->buffSize)
-#define LOG_BUF_MUTEX(x)  ((x)->buffMutex)
+#define LOG_BUF_MUTEX(x) ((x)->buffMutex)
 
 typedef struct {
-  char *          buffer;
+  char           *buffer;
   int32_t         buffStart;
   int32_t         buffEnd;
   int32_t         buffSize;
@@ -57,18 +57,18 @@ typedef struct {
 } SLogBuff;
 
 typedef struct {
-  int32_t fileNum;
+  int32_t         fileNum;
-  int32_t maxLines;
+  int32_t         maxLines;
-  int32_t lines;
+  int32_t         lines;
-  int32_t flag;
+  int32_t         flag;
-  int32_t openInProgress;
+  int32_t         openInProgress;
-  pid_t   pid;
+  pid_t           pid;
-  char    logName[LOG_FILE_NAME_LEN];
+  char            logName[LOG_FILE_NAME_LEN];
-  SLogBuff *      logHandle;
+  SLogBuff       *logHandle;
   pthread_mutex_t logMutex;
 } SLogObj;
 
-int8_t  tscEmbeddedInUtil = 0;
+int8_t tscEmbeddedInUtil = 0;
 
 int32_t tsLogKeepDays = 0;
 int8_t  tsAsyncLog = 1;
@@ -93,19 +93,19 @@ int32_t debugFlag = 0;
 int32_t sDebugFlag = 135;
 int32_t wDebugFlag = 135;
 int32_t tsdbDebugFlag = 131;
+int32_t tqDebugFlag = 131;
 int32_t cqDebugFlag = 131;
 int32_t fsDebugFlag = 135;
 int32_t ctgDebugFlag = 131;
 
-
 int64_t dbgEmptyW = 0;
 int64_t dbgWN = 0;
 int64_t dbgSmallWN = 0;
 int64_t dbgBigWN = 0;
 int64_t dbgWSize = 0;
 
-static SLogObj   tsLogObj = { .fileNum = 1 };
+static SLogObj   tsLogObj = {.fileNum = 1};
-static void *    taosAsyncOutputLog(void *param);
+static void     *taosAsyncOutputLog(void *param);
 static int32_t   taosPushLogBuffer(SLogBuff *tLogBuff, char *msg, int32_t msgLen);
 static SLogBuff *taosLogBuffNew(int32_t bufSize);
 static void      taosCloseLogByFd(int32_t oldFd);
@@ -139,8 +139,8 @@ static void taosStopLog() {
 
 void taosCloseLog() {
   taosStopLog();
-  //tsem_post(&(tsLogObj.logHandle->buffNotEmpty));
+  // tsem_post(&(tsLogObj.logHandle->buffNotEmpty));
-  taosMsleep(MAX_LOG_INTERVAL/1000);
+  taosMsleep(MAX_LOG_INTERVAL / 1000);
   if (taosCheckPthreadValid(tsLogObj.logHandle->asyncThread)) {
     pthread_join(tsLogObj.logHandle->asyncThread, NULL);
   }
@@ -217,7 +217,7 @@ static void *taosThreadToOpenNewFile(void *param) {
   tsLogObj.lines = 0;
   tsLogObj.openInProgress = 0;
   taosCloseLogByFd(oldFd);
-  
+
   uInfo("   new log file:%d is opened", tsLogObj.flag);
   uInfo("==================================");
   taosPrintCfg();
@@ -308,9 +308,9 @@ static void taosGetLogFileName(char *fn) {
 static int32_t taosOpenLogFile(char *fn, int32_t maxLines, int32_t maxFileNum) {
 #ifdef WINDOWS
   /*
-  * always set maxFileNum to 1
+   * always set maxFileNum to 1
-  * means client log filename is unique in windows
+   * means client log filename is unique in windows
-  */
+   */
   maxFileNum = 1;
 #endif
 
@@ -381,13 +381,14 @@ static int32_t taosOpenLogFile(char *fn, int32_t maxLines, int32_t maxFileNum) {
 
 void taosPrintLog(const char *flags, int32_t dflag, const char *format, ...) {
   if (tsTotalLogDirGB != 0 && tsAvailLogDirGB < tsMinimalLogDirGB) {
-    printf("server disk:%s space remain %.3f GB, total %.1f GB, stop print log.\n", tsLogDir, tsAvailLogDirGB, tsTotalLogDirGB);
+    printf("server disk:%s space remain %.3f GB, total %.1f GB, stop print log.\n", tsLogDir, tsAvailLogDirGB,
+           tsTotalLogDirGB);
     fflush(stdout);
     return;
   }
 
   va_list        argpointer;
-  char           buffer[MAX_LOGLINE_BUFFER_SIZE] = { 0 };
+  char           buffer[MAX_LOGLINE_BUFFER_SIZE] = {0};
   int32_t        len;
   struct tm      Tm, *ptm;
   struct timeval timeSecs;
@@ -434,20 +435,20 @@ void taosPrintLog(const char *flags, int32_t dflag, const char *format, ...) {
     }
   }
 
-  if (dflag & DEBUG_SCREEN)
+  if (dflag & DEBUG_SCREEN) taosWriteFile(1, buffer, (uint32_t)len);
-      taosWriteFile(1, buffer, (uint32_t)len);
   if (dflag == 255) nInfo(buffer, len);
 }
 
 void taosDumpData(unsigned char *msg, int32_t len) {
   if (tsTotalLogDirGB != 0 && tsAvailLogDirGB < tsMinimalLogDirGB) {
-    printf("server disk:%s space remain %.3f GB, total %.1f GB, stop dump log.\n", tsLogDir, tsAvailLogDirGB, tsTotalLogDirGB);
+    printf("server disk:%s space remain %.3f GB, total %.1f GB, stop dump log.\n", tsLogDir, tsAvailLogDirGB,
+           tsTotalLogDirGB);
     fflush(stdout);
     return;
   }
 
-  char temp[256];
+  char    temp[256];
-  int32_t  i, pos = 0, c = 0;
+  int32_t i, pos = 0, c = 0;
 
   for (i = 0; i < len; ++i) {
     sprintf(temp + pos, "%02x ", msg[i]);
@@ -468,7 +469,8 @@ void taosDumpData(unsigned char *msg, int32_t len) {
 
 void taosPrintLongString(const char *flags, int32_t dflag, const char *format, ...) {
   if (tsTotalLogDirGB != 0 && tsAvailLogDirGB < tsMinimalLogDirGB) {
-    printf("server disk:%s space remain %.3f GB, total %.1f GB, stop write log.\n", tsLogDir, tsAvailLogDirGB, tsTotalLogDirGB);
+    printf("server disk:%s space remain %.3f GB, total %.1f GB, stop write log.\n", tsLogDir, tsAvailLogDirGB,
+           tsTotalLogDirGB);
     fflush(stdout);
     return;
   }
@@ -503,7 +505,7 @@ void taosPrintLongString(const char *flags, int32_t dflag, const char *format, .
     } else {
       taosWriteFile(tsLogObj.logHandle->fd, buffer, len);
     }
-    
+
     if (tsLogObj.maxLines > 0) {
       atomic_add_fetch_32(&tsLogObj.lines, 1);
 
@@ -542,7 +544,7 @@ static SLogBuff *taosLogBuffNew(int32_t bufSize) {
   tLogBuff->stop = 0;
 
   if (pthread_mutex_init(&LOG_BUF_MUTEX(tLogBuff), NULL) < 0) goto _err;
-  //tsem_init(&(tLogBuff->buffNotEmpty), 0, 0);
+  // tsem_init(&(tLogBuff->buffNotEmpty), 0, 0);
 
   return tLogBuff;
 
@@ -576,12 +578,12 @@ static void taosCopyLogBuffer(SLogBuff *tLogBuff, int32_t start, int32_t end, ch
 }
 
 static int32_t taosPushLogBuffer(SLogBuff *tLogBuff, char *msg, int32_t msgLen) {
-  int32_t start = 0;
+  int32_t        start = 0;
-  int32_t end = 0;
+  int32_t        end = 0;
-  int32_t remainSize = 0;
+  int32_t        remainSize = 0;
   static int64_t lostLine = 0;
-  char tmpBuf[40] = {0};
+  char           tmpBuf[40] = {0};
-  int32_t tmpBufLen = 0;
+  int32_t        tmpBufLen = 0;
 
   if (tLogBuff == NULL || tLogBuff->stop) return -1;
 
@@ -592,7 +594,7 @@ static int32_t taosPushLogBuffer(SLogBuff *tLogBuff, char *msg, int32_t msgLen)
   remainSize = (start > end) ? (start - end - 1) : (start + LOG_BUF_SIZE(tLogBuff) - end - 1);
 
   if (lostLine > 0) {
-    sprintf(tmpBuf, "...Lost %"PRId64" lines here...\n", lostLine);
+    sprintf(tmpBuf, "...Lost %" PRId64 " lines here...\n", lostLine);
     tmpBufLen = (int32_t)strlen(tmpBuf);
   }
 
@@ -610,7 +612,7 @@ static int32_t taosPushLogBuffer(SLogBuff *tLogBuff, char *msg, int32_t msgLen)
 
   taosCopyLogBuffer(tLogBuff, LOG_BUF_START(tLogBuff), LOG_BUF_END(tLogBuff), msg, msgLen);
 
-  //int32_t w = atomic_sub_fetch_32(&waitLock, 1);
+  // int32_t w = atomic_sub_fetch_32(&waitLock, 1);
   /*
   if (w <= 0 || ((remainSize - msgLen - tmpBufLen) < (LOG_BUF_SIZE(tLogBuff) * 4 /5))) {
     tsem_post(&(tLogBuff->buffNotEmpty));
@@ -622,7 +624,6 @@ static int32_t taosPushLogBuffer(SLogBuff *tLogBuff, char *msg, int32_t msgLen)
 
   pthread_mutex_unlock(&LOG_BUF_MUTEX(tLogBuff));
 
-
   return 0;
 }
 
@@ -634,9 +635,9 @@ static int32_t taosGetLogRemainSize(SLogBuff *tLogBuff, int32_t start, int32_t e
 
 static void taosWriteLog(SLogBuff *tLogBuff) {
   static int32_t lastDuration = 0;
-  int32_t remainChecked = 0;
+  int32_t        remainChecked = 0;
-  int32_t start, end, pollSize;
+  int32_t        start, end, pollSize;
-  
+
   do {
     if (remainChecked == 0) {
       start = LOG_BUF_START(tLogBuff);
@@ -662,24 +663,24 @@ static void taosWriteLog(SLogBuff *tLogBuff) {
     if (start < end) {
       taosWriteFile(tLogBuff->fd, LOG_BUF_BUFFER(tLogBuff) + start, pollSize);
     } else {
-        int32_t tsize = LOG_BUF_SIZE(tLogBuff) - start;
+      int32_t tsize = LOG_BUF_SIZE(tLogBuff) - start;
-        taosWriteFile(tLogBuff->fd, LOG_BUF_BUFFER(tLogBuff) + start, tsize);
+      taosWriteFile(tLogBuff->fd, LOG_BUF_BUFFER(tLogBuff) + start, tsize);
 
-        taosWriteFile(tLogBuff->fd, LOG_BUF_BUFFER(tLogBuff), end);
+      taosWriteFile(tLogBuff->fd, LOG_BUF_BUFFER(tLogBuff), end);
     }
 
     dbgWN++;
-    dbgWSize+=pollSize;
+    dbgWSize += pollSize;
-    
+
     if (pollSize < tLogBuff->minBuffSize) {
       dbgSmallWN++;
       if (writeInterval < MAX_LOG_INTERVAL) {
         writeInterval += LOG_INTERVAL_STEP;
       }
-    } else if (pollSize >  LOG_BUF_SIZE(tLogBuff)/3) {
+    } else if (pollSize > LOG_BUF_SIZE(tLogBuff) / 3) {
       dbgBigWN++;
       writeInterval = MIN_LOG_INTERVAL;
-    } else if (pollSize > LOG_BUF_SIZE(tLogBuff)/4) {
+    } else if (pollSize > LOG_BUF_SIZE(tLogBuff) / 4) {
       if (writeInterval > MIN_LOG_INTERVAL) {
         writeInterval -= LOG_INTERVAL_STEP;
       }
@@ -698,13 +699,13 @@ static void taosWriteLog(SLogBuff *tLogBuff) {
     writeInterval = MIN_LOG_INTERVAL;
 
     remainChecked = 1;
-  }while (1);
+  } while (1);
 }
 
 static void *taosAsyncOutputLog(void *param) {
   SLogBuff *tLogBuff = (SLogBuff *)param;
   setThreadName("log");
-  
+
   while (1) {
     taosMsleep(writeInterval);
 
@@ -721,8 +722,8 @@ int32_t taosCompressFile(char *srcFileName, char *destFileName) {
   int32_t compressSize = 163840;
   int32_t ret = 0;
   int32_t len = 0;
-  char *  data = malloc(compressSize);
+  char   *data = malloc(compressSize);
-  FILE *  srcFp = NULL;
+  FILE   *srcFp = NULL;
   gzFile  dstFp = NULL;
 
   srcFp = fopen(srcFileName, "r");