Merge branch '3.0' into fix/TD-20802

cmake/taostools_CMakeLists.txt.in

@@ -2,7 +2,7 @@
 # taos-tools
 ExternalProject_Add(taos-tools
         GIT_REPOSITORY https://github.com/taosdata/taos-tools.git
-        GIT_TAG d5df76d
+        GIT_TAG 823fae5
         SOURCE_DIR "${TD_SOURCE_DIR}/tools/taos-tools"
         BINARY_DIR ""
         #BUILD_IN_SOURCE TRUE

docs/zh/27-train-faq/01-faq.md

@@ -201,3 +201,45 @@ TDengine 中时间戳的时区总是由客户端进行处理，而与服务端
 OOM 是操作系统的保护机制，当操作系统内存(包括 SWAP )不足时，会杀掉某些进程，从而保证操作系统的稳定运行。通常内存不足主要是如下两个原因导致，一是剩余内存小于 vm.min_free_kbytes ；二是程序请求的内存大于剩余内存。还有一种情况是内存充足但程序占用了特殊的内存地址，也会触发 OOM 。
 
 TDengine 会预先为每个 VNode 分配好内存，每个 Database 的 VNode 个数受 建库时的vgroups参数影响，每个 VNode 占用的内存大小受 buffer参数 影响。要防止 OOM，需要在项目建设之初合理规划内存，并合理设置 SWAP ，除此之外查询过量的数据也有可能导致内存暴涨，这取决于具体的查询语句。TDengine 企业版对内存管理做了优化，采用了新的内存分配器，对稳定性有更高要求的用户可以考虑选择企业版。
+
+### 19. 在macOS上遇到Too many open files怎么办？
+
+taosd日志文件报错Too many open file，是由于taosd打开文件数超过系统设置的上限所致。
+解决方案如下：
+1. 新建文件 /Library/LaunchDaemons/limit.maxfiles.plist，写入以下内容(以下示例将limit和maxfiles改为10万，可按需修改)：
+```
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN"
+"http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+<key>Label</key>
+  <string>limit.maxfiles</string>
+<key>ProgramArguments</key>
+<array>
+  <string>launchctl</string>
+  <string>limit</string>
+  <string>maxfiles</string>
+  <string>100000</string>
+  <string>100000</string>
+</array>
+<key>RunAtLoad</key>
+  <true/>
+<key>ServiceIPC</key>
+  <false/>
+</dict>
+</plist>
+```
+2. 修改文件权限
+```
+sudo chown root:wheel /Library/LaunchDaemons/limit.maxfiles.plist
+sudo chmod 644 /Library/LaunchDaemons/limit.maxfiles.plist
+```
+3. 加载 plist 文件 (或重启系统后生效。launchd在启动时会自动加载该目录的 plist)
+```
+sudo launchctl load -w /Library/LaunchDaemons/limit.maxfiles.plist
+```
+4.确认更改后的限制
+```
+launchctl limit maxfiles
+```

include/common/systable.h

@@ -47,6 +47,7 @@ extern "C" {
 #define TSDB_INS_TABLE_TOPICS            "ins_topics"
 #define TSDB_INS_TABLE_STREAMS           "ins_streams"
 #define TSDB_INS_TABLE_STREAM_TASKS      "ins_stream_tasks"
+#define TSDB_INS_TABLE_USER_PRIVILEGES   "ins_user_privileges"
 
 #define TSDB_PERFORMANCE_SCHEMA_DB   "performance_schema"
 #define TSDB_PERFS_TABLE_SMAS        "perf_smas"

include/common/tglobal.h

@@ -137,6 +137,7 @@ extern int64_t tsWalFsyncDataSizeLimit;
 // internal
 extern int32_t tsTransPullupInterval;
 extern int32_t tsMqRebalanceInterval;
+extern int32_t tsStreamCheckpointTickInterval;
 extern int32_t tsTtlUnit;
 extern int32_t tsTtlPushInterval;
 extern int32_t tsGrantHBInterval;

include/common/tmsg.h

@@ -120,6 +120,7 @@ typedef enum _mgmt_table {
   TSDB_MGMT_TABLE_VNODES,
   TSDB_MGMT_TABLE_APPS,
   TSDB_MGMT_TABLE_STREAM_TASKS,
+  TSDB_MGMT_TABLE_PRIVILEGES,
   TSDB_MGMT_TABLE_MAX,
 } EShowType;
 
@@ -151,6 +152,8 @@ typedef enum _mgmt_table {
 #define TSDB_ALTER_USER_REMOVE_ALL_DB          0x8
 #define TSDB_ALTER_USER_ENABLE                 0x9
 #define TSDB_ALTER_USER_SYSINFO                0xA
+#define TSDB_ALTER_USER_ADD_SUBSCRIBE_TOPIC    0xB
+#define TSDB_ALTER_USER_REMOVE_SUBSCRIBE_TOPIC 0xC
 
 #define TSDB_ALTER_USER_PRIVILEGES 0x2
 
@@ -620,7 +623,7 @@ typedef struct {
   int8_t enable;
   char   user[TSDB_USER_LEN];
   char   pass[TSDB_USET_PASSWORD_LEN];
-  char   dbname[TSDB_DB_FNAME_LEN];
+  char   objname[TSDB_DB_FNAME_LEN];  // db or topic
 } SAlterUserReq;
 
 int32_t tSerializeSAlterUserReq(void* buf, int32_t bufLen, SAlterUserReq* pReq);
@@ -1059,6 +1062,7 @@ typedef struct {
   int32_t vgId;
   int8_t  syncState;
   int8_t  syncRestore;
+  int8_t  syncCanRead;
   int64_t cacheUsage;
   int64_t numOfTables;
   int64_t numOfTimeSeries;
@@ -1144,6 +1148,13 @@ typedef struct {
 int32_t tSerializeSMTimerMsg(void* buf, int32_t bufLen, SMTimerReq* pReq);
 int32_t tDeserializeSMTimerMsg(void* buf, int32_t bufLen, SMTimerReq* pReq);
 
+typedef struct {
+  int64_t tick;
+} SMStreamTickReq;
+
+int32_t tSerializeSMStreamTickMsg(void* buf, int32_t bufLen, SMStreamTickReq* pReq);
+int32_t tDeserializeSMStreamTickMsg(void* buf, int32_t bufLen, SMStreamTickReq* pReq);
+
 typedef struct {
   int32_t  id;
   uint16_t port;                 // node sync Port
@@ -1744,6 +1755,8 @@ typedef struct {
   int64_t watermark;
   int32_t numOfTags;
   SArray* pTags;  // array of SField
+  // 3.0.20
+  int64_t checkpointFreq;  // ms
 } SCMCreateStreamReq;
 
 typedef struct {
@@ -1943,6 +1956,12 @@ typedef struct {
   SHashObj* rebSubHash;  // SHashObj<key, SMqRebSubscribe>
 } SMqDoRebalanceMsg;
 
+typedef struct {
+  int64_t streamId;
+  int64_t checkpointId;
+  char    streamName[TSDB_STREAM_FNAME_LEN];
+} SMStreamDoCheckpointMsg;
+
 typedef struct {
   int64_t status;
 } SMVSubscribeRsp;

include/common/tmsgdef.h

@@ -172,6 +172,8 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_MND_SERVER_VERSION, "server-version", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_UPTIME_TIMER, "uptime-timer", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_TMQ_LOST_CONSUMER_CLEAR, "lost-consumer-clear", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_STREAM_CHECKPOINT_TIMER, "stream-checkpoint-tmr", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_MND_STREAM_BEGIN_CHECKPOINT, "stream-begin-checkpoint", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_MND_MAX_MSG, "mnd-max", NULL, NULL)
 
   TD_NEW_MSG_SEG(TDMT_VND_MSG)
@@ -241,8 +243,11 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DISPATCH, "stream-task-dispatch", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_STREAM_UNUSED1, "stream-unused1", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_STREAM_RETRIEVE, "stream-retrieve", NULL, NULL)
-  TD_DEF_MSG_TYPE(TDMT_STREAM_RECOVER_FINISH, "vnode-stream-finish", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_STREAM_RECOVER_FINISH, "stream-recover-finish", NULL, NULL)
-  TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_CHECK, "vnode-stream-task-check", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_CHECK, "stream-task-check", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_CHECKPOINT, "stream-checkpoint", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_REPORT_CHECKPOINT, "stream-report-checkpoint", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_RESTORE_CHECKPOINT, "stream-restore-checkpoint", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_STREAM_MAX_MSG, "stream-max", NULL, NULL)
 
   TD_NEW_MSG_SEG(TDMT_MON_MSG)
@@ -282,6 +287,7 @@ enum {
   TD_DEF_MSG_TYPE(TDMT_VND_STREAM_TRIGGER, "vnode-stream-trigger", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_STREAM_RECOVER_NONBLOCKING_STAGE, "vnode-stream-recover1", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_STREAM_RECOVER_BLOCKING_STAGE, "vnode-stream-recover2", NULL, NULL)
+  TD_DEF_MSG_TYPE(TDMT_VND_STREAM_CHECK_POINT_SOURCE, "vnode-stream-checkpoint-source", NULL, NULL)
   TD_DEF_MSG_TYPE(TDMT_VND_STREAM_MAX_MSG, "vnd-stream-max", NULL, NULL)
 
   TD_NEW_MSG_SEG(TDMT_VND_TMQ_MSG)

include/common/ttokendef.h

@@ -58,282 +58,284 @@
 #define TK_TO                   40
 #define TK_REVOKE               41
 #define TK_FROM                 42
-#define TK_NK_COMMA             43
+#define TK_SUBSCRIBE            43
-#define TK_READ                 44
+#define TK_NK_COMMA             44
-#define TK_WRITE                45
+#define TK_READ                 45
-#define TK_NK_DOT               46
+#define TK_WRITE                46
-#define TK_DNODE                47
+#define TK_NK_DOT               47
-#define TK_PORT                 48
+#define TK_DNODE                48
-#define TK_DNODES               49
+#define TK_PORT                 49
-#define TK_NK_IPTOKEN           50
+#define TK_DNODES               50
-#define TK_FORCE                51
+#define TK_NK_IPTOKEN           51
-#define TK_LOCAL                52
+#define TK_FORCE                52
-#define TK_QNODE                53
+#define TK_LOCAL                53
-#define TK_BNODE                54
+#define TK_QNODE                54
-#define TK_SNODE                55
+#define TK_BNODE                55
-#define TK_MNODE                56
+#define TK_SNODE                56
-#define TK_DATABASE             57
+#define TK_MNODE                57
-#define TK_USE                  58
+#define TK_DATABASE             58
-#define TK_FLUSH                59
+#define TK_USE                  59
-#define TK_TRIM                 60
+#define TK_FLUSH                60
-#define TK_IF                   61
+#define TK_TRIM                 61
-#define TK_NOT                  62
+#define TK_IF                   62
-#define TK_EXISTS               63
+#define TK_NOT                  63
-#define TK_BUFFER               64
+#define TK_EXISTS               64
-#define TK_CACHEMODEL           65
+#define TK_BUFFER               65
-#define TK_CACHESIZE            66
+#define TK_CACHEMODEL           66
-#define TK_COMP                 67
+#define TK_CACHESIZE            67
-#define TK_DURATION             68
+#define TK_COMP                 68
-#define TK_NK_VARIABLE          69
+#define TK_DURATION             69
-#define TK_MAXROWS              70
+#define TK_NK_VARIABLE          70
-#define TK_MINROWS              71
+#define TK_MAXROWS              71
-#define TK_KEEP                 72
+#define TK_MINROWS              72
-#define TK_PAGES                73
+#define TK_KEEP                 73
-#define TK_PAGESIZE             74
+#define TK_PAGES                74
-#define TK_TSDB_PAGESIZE        75
+#define TK_PAGESIZE             75
-#define TK_PRECISION            76
+#define TK_TSDB_PAGESIZE        76
-#define TK_REPLICA              77
+#define TK_PRECISION            77
-#define TK_STRICT               78
+#define TK_REPLICA              78
-#define TK_VGROUPS              79
+#define TK_STRICT               79
-#define TK_SINGLE_STABLE        80
+#define TK_VGROUPS              80
-#define TK_RETENTIONS           81
+#define TK_SINGLE_STABLE        81
-#define TK_SCHEMALESS           82
+#define TK_RETENTIONS           82
-#define TK_WAL_LEVEL            83
+#define TK_SCHEMALESS           83
-#define TK_WAL_FSYNC_PERIOD     84
+#define TK_WAL_LEVEL            84
-#define TK_WAL_RETENTION_PERIOD 85
+#define TK_WAL_FSYNC_PERIOD     85
-#define TK_WAL_RETENTION_SIZE   86
+#define TK_WAL_RETENTION_PERIOD 86
-#define TK_WAL_ROLL_PERIOD      87
+#define TK_WAL_RETENTION_SIZE   87
-#define TK_WAL_SEGMENT_SIZE     88
+#define TK_WAL_ROLL_PERIOD      88
-#define TK_STT_TRIGGER          89
+#define TK_WAL_SEGMENT_SIZE     89
-#define TK_TABLE_PREFIX         90
+#define TK_STT_TRIGGER          90
-#define TK_TABLE_SUFFIX         91
+#define TK_TABLE_PREFIX         91
-#define TK_NK_COLON             92
+#define TK_TABLE_SUFFIX         92
-#define TK_MAX_SPEED            93
+#define TK_NK_COLON             93
-#define TK_TABLE                94
+#define TK_MAX_SPEED            94
-#define TK_NK_LP                95
+#define TK_TABLE                95
-#define TK_NK_RP                96
+#define TK_NK_LP                96
-#define TK_STABLE               97
+#define TK_NK_RP                97
-#define TK_ADD                  98
+#define TK_STABLE               98
-#define TK_COLUMN               99
+#define TK_ADD                  99
-#define TK_MODIFY               100
+#define TK_COLUMN               100
-#define TK_RENAME               101
+#define TK_MODIFY               101
-#define TK_TAG                  102
+#define TK_RENAME               102
-#define TK_SET                  103
+#define TK_TAG                  103
-#define TK_NK_EQ                104
+#define TK_SET                  104
-#define TK_USING                105
+#define TK_NK_EQ                105
-#define TK_TAGS                 106
+#define TK_USING                106
-#define TK_COMMENT              107
+#define TK_TAGS                 107
-#define TK_BOOL                 108
+#define TK_COMMENT              108
-#define TK_TINYINT              109
+#define TK_BOOL                 109
-#define TK_SMALLINT             110
+#define TK_TINYINT              110
-#define TK_INT                  111
+#define TK_SMALLINT             111
-#define TK_INTEGER              112
+#define TK_INT                  112
-#define TK_BIGINT               113
+#define TK_INTEGER              113
-#define TK_FLOAT                114
+#define TK_BIGINT               114
-#define TK_DOUBLE               115
+#define TK_FLOAT                115
-#define TK_BINARY               116
+#define TK_DOUBLE               116
-#define TK_TIMESTAMP            117
+#define TK_BINARY               117
-#define TK_NCHAR                118
+#define TK_TIMESTAMP            118
-#define TK_UNSIGNED             119
+#define TK_NCHAR                119
-#define TK_JSON                 120
+#define TK_UNSIGNED             120
-#define TK_VARCHAR              121
+#define TK_JSON                 121
-#define TK_MEDIUMBLOB           122
+#define TK_VARCHAR              122
-#define TK_BLOB                 123
+#define TK_MEDIUMBLOB           123
-#define TK_VARBINARY            124
+#define TK_BLOB                 124
-#define TK_DECIMAL              125
+#define TK_VARBINARY            125
-#define TK_MAX_DELAY            126
+#define TK_DECIMAL              126
-#define TK_WATERMARK            127
+#define TK_MAX_DELAY            127
-#define TK_ROLLUP               128
+#define TK_WATERMARK            128
-#define TK_TTL                  129
+#define TK_ROLLUP               129
-#define TK_SMA                  130
+#define TK_TTL                  130
-#define TK_FIRST                131
+#define TK_SMA                  131
-#define TK_LAST                 132
+#define TK_FIRST                132
-#define TK_SHOW                 133
+#define TK_LAST                 133
-#define TK_DATABASES            134
+#define TK_SHOW                 134
-#define TK_TABLES               135
+#define TK_PRIVILEGES           135
-#define TK_STABLES              136
+#define TK_DATABASES            136
-#define TK_MNODES               137
+#define TK_TABLES               137
-#define TK_QNODES               138
+#define TK_STABLES              138
-#define TK_FUNCTIONS            139
+#define TK_MNODES               139
-#define TK_INDEXES              140
+#define TK_QNODES               140
-#define TK_ACCOUNTS             141
+#define TK_FUNCTIONS            141
-#define TK_APPS                 142
+#define TK_INDEXES              142
-#define TK_CONNECTIONS          143
+#define TK_ACCOUNTS             143
-#define TK_LICENCES             144
+#define TK_APPS                 144
-#define TK_GRANTS               145
+#define TK_CONNECTIONS          145
-#define TK_QUERIES              146
+#define TK_LICENCES             146
-#define TK_SCORES               147
+#define TK_GRANTS               147
-#define TK_TOPICS               148
+#define TK_QUERIES              148
-#define TK_VARIABLES            149
+#define TK_SCORES               149
-#define TK_CLUSTER              150
+#define TK_TOPICS               150
-#define TK_BNODES               151
+#define TK_VARIABLES            151
-#define TK_SNODES               152
+#define TK_CLUSTER              152
-#define TK_TRANSACTIONS         153
+#define TK_BNODES               153
-#define TK_DISTRIBUTED          154
+#define TK_SNODES               154
-#define TK_CONSUMERS            155
+#define TK_TRANSACTIONS         155
-#define TK_SUBSCRIPTIONS        156
+#define TK_DISTRIBUTED          156
-#define TK_VNODES               157
+#define TK_CONSUMERS            157
-#define TK_LIKE                 158
+#define TK_SUBSCRIPTIONS        158
-#define TK_TBNAME               159
+#define TK_VNODES               159
-#define TK_QTAGS                160
+#define TK_LIKE                 160
-#define TK_AS                   161
+#define TK_TBNAME               161
-#define TK_INDEX                162
+#define TK_QTAGS                162
-#define TK_FUNCTION             163
+#define TK_AS                   163
-#define TK_INTERVAL             164
+#define TK_INDEX                164
-#define TK_TOPIC                165
+#define TK_FUNCTION             165
-#define TK_WITH                 166
+#define TK_INTERVAL             166
-#define TK_META                 167
+#define TK_TOPIC                167
-#define TK_CONSUMER             168
+#define TK_WITH                 168
-#define TK_GROUP                169
+#define TK_META                 169
-#define TK_DESC                 170
+#define TK_CONSUMER             170
-#define TK_DESCRIBE             171
+#define TK_GROUP                171
-#define TK_RESET                172
+#define TK_DESC                 172
-#define TK_QUERY                173
+#define TK_DESCRIBE             173
-#define TK_CACHE                174
+#define TK_RESET                174
-#define TK_EXPLAIN              175
+#define TK_QUERY                175
-#define TK_ANALYZE              176
+#define TK_CACHE                176
-#define TK_VERBOSE              177
+#define TK_EXPLAIN              177
-#define TK_NK_BOOL              178
+#define TK_ANALYZE              178
-#define TK_RATIO                179
+#define TK_VERBOSE              179
-#define TK_NK_FLOAT             180
+#define TK_NK_BOOL              180
-#define TK_OUTPUTTYPE           181
+#define TK_RATIO                181
-#define TK_AGGREGATE            182
+#define TK_NK_FLOAT             182
-#define TK_BUFSIZE              183
+#define TK_OUTPUTTYPE           183
-#define TK_STREAM               184
+#define TK_AGGREGATE            184
-#define TK_INTO                 185
+#define TK_BUFSIZE              185
-#define TK_TRIGGER              186
+#define TK_STREAM               186
-#define TK_AT_ONCE              187
+#define TK_INTO                 187
-#define TK_WINDOW_CLOSE         188
+#define TK_TRIGGER              188
-#define TK_IGNORE               189
+#define TK_AT_ONCE              189
-#define TK_EXPIRED              190
+#define TK_WINDOW_CLOSE         190
-#define TK_FILL_HISTORY         191
+#define TK_IGNORE               191
-#define TK_SUBTABLE             192
+#define TK_EXPIRED              192
-#define TK_KILL                 193
+#define TK_FILL_HISTORY         193
-#define TK_CONNECTION           194
+#define TK_SUBTABLE             194
-#define TK_TRANSACTION          195
+#define TK_KILL                 195
-#define TK_BALANCE              196
+#define TK_CONNECTION           196
-#define TK_VGROUP               197
+#define TK_TRANSACTION          197
-#define TK_MERGE                198
+#define TK_BALANCE              198
-#define TK_REDISTRIBUTE         199
+#define TK_VGROUP               199
-#define TK_SPLIT                200
+#define TK_MERGE                200
-#define TK_DELETE               201
+#define TK_REDISTRIBUTE         201
-#define TK_INSERT               202
+#define TK_SPLIT                202
-#define TK_NULL                 203
+#define TK_DELETE               203
-#define TK_NK_QUESTION          204
+#define TK_INSERT               204
-#define TK_NK_ARROW             205
+#define TK_NULL                 205
-#define TK_ROWTS                206
+#define TK_NK_QUESTION          206
-#define TK_QSTART               207
+#define TK_NK_ARROW             207
-#define TK_QEND                 208
+#define TK_ROWTS                208
-#define TK_QDURATION            209
+#define TK_QSTART               209
-#define TK_WSTART               210
+#define TK_QEND                 210
-#define TK_WEND                 211
+#define TK_QDURATION            211
-#define TK_WDURATION            212
+#define TK_WSTART               212
-#define TK_IROWTS               213
+#define TK_WEND                 213
-#define TK_CAST                 214
+#define TK_WDURATION            214
-#define TK_NOW                  215
+#define TK_IROWTS               215
-#define TK_TODAY                216
+#define TK_CAST                 216
-#define TK_TIMEZONE             217
+#define TK_NOW                  217
-#define TK_CLIENT_VERSION       218
+#define TK_TODAY                218
-#define TK_SERVER_VERSION       219
+#define TK_TIMEZONE             219
-#define TK_SERVER_STATUS        220
+#define TK_CLIENT_VERSION       220
-#define TK_CURRENT_USER         221
+#define TK_SERVER_VERSION       221
-#define TK_COUNT                222
+#define TK_SERVER_STATUS        222
-#define TK_LAST_ROW             223
+#define TK_CURRENT_USER         223
-#define TK_CASE                 224
+#define TK_COUNT                224
-#define TK_END                  225
+#define TK_LAST_ROW             225
-#define TK_WHEN                 226
+#define TK_CASE                 226
-#define TK_THEN                 227
+#define TK_END                  227
-#define TK_ELSE                 228
+#define TK_WHEN                 228
-#define TK_BETWEEN              229
+#define TK_THEN                 229
-#define TK_IS                   230
+#define TK_ELSE                 230
-#define TK_NK_LT                231
+#define TK_BETWEEN              231
-#define TK_NK_GT                232
+#define TK_IS                   232
-#define TK_NK_LE                233
+#define TK_NK_LT                233
-#define TK_NK_GE                234
+#define TK_NK_GT                234
-#define TK_NK_NE                235
+#define TK_NK_LE                235
-#define TK_MATCH                236
+#define TK_NK_GE                236
-#define TK_NMATCH               237
+#define TK_NK_NE                237
-#define TK_CONTAINS             238
+#define TK_MATCH                238
-#define TK_IN                   239
+#define TK_NMATCH               239
-#define TK_JOIN                 240
+#define TK_CONTAINS             240
-#define TK_INNER                241
+#define TK_IN                   241
-#define TK_SELECT               242
+#define TK_JOIN                 242
-#define TK_DISTINCT             243
+#define TK_INNER                243
-#define TK_WHERE                244
+#define TK_SELECT               244
-#define TK_PARTITION            245
+#define TK_DISTINCT             245
-#define TK_BY                   246
+#define TK_WHERE                246
-#define TK_SESSION              247
+#define TK_PARTITION            247
-#define TK_STATE_WINDOW         248
+#define TK_BY                   248
-#define TK_SLIDING              249
+#define TK_SESSION              249
-#define TK_FILL                 250
+#define TK_STATE_WINDOW         250
-#define TK_VALUE                251
+#define TK_SLIDING              251
-#define TK_NONE                 252
+#define TK_FILL                 252
-#define TK_PREV                 253
+#define TK_VALUE                253
-#define TK_LINEAR               254
+#define TK_NONE                 254
-#define TK_NEXT                 255
+#define TK_PREV                 255
-#define TK_HAVING               256
+#define TK_LINEAR               256
-#define TK_RANGE                257
+#define TK_NEXT                 257
-#define TK_EVERY                258
+#define TK_HAVING               258
-#define TK_ORDER                259
+#define TK_RANGE                259
-#define TK_SLIMIT               260
+#define TK_EVERY                260
-#define TK_SOFFSET              261
+#define TK_ORDER                261
-#define TK_LIMIT                262
+#define TK_SLIMIT               262
-#define TK_OFFSET               263
+#define TK_SOFFSET              263
-#define TK_ASC                  264
+#define TK_LIMIT                264
-#define TK_NULLS                265
+#define TK_OFFSET               265
-#define TK_ABORT                266
+#define TK_ASC                  266
-#define TK_AFTER                267
+#define TK_NULLS                267
-#define TK_ATTACH               268
+#define TK_ABORT                268
-#define TK_BEFORE               269
+#define TK_AFTER                269
-#define TK_BEGIN                270
+#define TK_ATTACH               270
-#define TK_BITAND               271
+#define TK_BEFORE               271
-#define TK_BITNOT               272
+#define TK_BEGIN                272
-#define TK_BITOR                273
+#define TK_BITAND               273
-#define TK_BLOCKS               274
+#define TK_BITNOT               274
-#define TK_CHANGE               275
+#define TK_BITOR                275
-#define TK_COMMA                276
+#define TK_BLOCKS               276
-#define TK_COMPACT              277
+#define TK_CHANGE               277
-#define TK_CONCAT               278
+#define TK_COMMA                278
-#define TK_CONFLICT             279
+#define TK_COMPACT              279
-#define TK_COPY                 280
+#define TK_CONCAT               280
-#define TK_DEFERRED             281
+#define TK_CONFLICT             281
-#define TK_DELIMITERS           282
+#define TK_COPY                 282
-#define TK_DETACH               283
+#define TK_DEFERRED             283
-#define TK_DIVIDE               284
+#define TK_DELIMITERS           284
-#define TK_DOT                  285
+#define TK_DETACH               285
-#define TK_EACH                 286
+#define TK_DIVIDE               286
-#define TK_FAIL                 287
+#define TK_DOT                  287
-#define TK_FILE                 288
+#define TK_EACH                 288
-#define TK_FOR                  289
+#define TK_FAIL                 289
-#define TK_GLOB                 290
+#define TK_FILE                 290
-#define TK_ID                   291
+#define TK_FOR                  291
-#define TK_IMMEDIATE            292
+#define TK_GLOB                 292
-#define TK_IMPORT               293
+#define TK_ID                   293
-#define TK_INITIALLY            294
+#define TK_IMMEDIATE            294
-#define TK_INSTEAD              295
+#define TK_IMPORT               295
-#define TK_ISNULL               296
+#define TK_INITIALLY            296
-#define TK_KEY                  297
+#define TK_INSTEAD              297
-#define TK_MODULES              298
+#define TK_ISNULL               298
-#define TK_NK_BITNOT            299
+#define TK_KEY                  299
-#define TK_NK_SEMI              300
+#define TK_MODULES              300
-#define TK_NOTNULL              301
+#define TK_NK_BITNOT            301
-#define TK_OF                   302
+#define TK_NK_SEMI              302
-#define TK_PLUS                 303
+#define TK_NOTNULL              303
-#define TK_PRIVILEGE            304
+#define TK_OF                   304
-#define TK_RAISE                305
+#define TK_PLUS                 305
-#define TK_REPLACE              306
+#define TK_PRIVILEGE            306
-#define TK_RESTRICT             307
+#define TK_RAISE                307
-#define TK_ROW                  308
+#define TK_REPLACE              308
-#define TK_SEMI                 309
+#define TK_RESTRICT             309
-#define TK_STAR                 310
+#define TK_ROW                  310
-#define TK_STATEMENT            311
+#define TK_SEMI                 311
-#define TK_STRING               312
+#define TK_STAR                 312
-#define TK_TIMES                313
+#define TK_STATEMENT            313
-#define TK_UPDATE               314
+#define TK_STRING               314
-#define TK_VALUES               315
+#define TK_TIMES                315
-#define TK_VARIABLE             316
+#define TK_UPDATE               316
-#define TK_VIEW                 317
+#define TK_VALUES               317
-#define TK_WAL                  318
+#define TK_VARIABLE             318
+#define TK_VIEW                 319
+#define TK_WAL                  320
 
 #define TK_NK_SPACE   600
 #define TK_NK_COMMENT 601

include/libs/nodes/cmdnodes.h

@@ -45,6 +45,7 @@ extern "C" {
 #define PRIVILEGE_TYPE_ALL       PRIVILEGE_TYPE_MASK(0)
 #define PRIVILEGE_TYPE_READ      PRIVILEGE_TYPE_MASK(1)
 #define PRIVILEGE_TYPE_WRITE     PRIVILEGE_TYPE_MASK(2)
+#define PRIVILEGE_TYPE_SUBSCRIBE PRIVILEGE_TYPE_MASK(3)
 
 #define PRIVILEGE_TYPE_TEST_MASK(val, mask) (((val) & (mask)) != 0)
 
@@ -423,7 +424,7 @@ typedef struct SDropFunctionStmt {
 typedef struct SGrantStmt {
   ENodeType type;
   char      userName[TSDB_USER_LEN];
-  char      dbName[TSDB_DB_NAME_LEN];
+  char      objName[TSDB_DB_NAME_LEN];  // db or topic
   int64_t   privileges;
 } SGrantStmt;
 

include/libs/nodes/nodes.h

@@ -187,6 +187,7 @@ typedef enum ENodeType {
   QUERY_NODE_SHOW_TRANSACTIONS_STMT,
   QUERY_NODE_SHOW_SUBSCRIPTIONS_STMT,
   QUERY_NODE_SHOW_VNODES_STMT,
+  QUERY_NODE_SHOW_USER_PRIVILEGES_STMT,
   QUERY_NODE_SHOW_CREATE_DATABASE_STMT,
   QUERY_NODE_SHOW_CREATE_TABLE_STMT,
   QUERY_NODE_SHOW_CREATE_STABLE_STMT,

include/libs/stream/tstream.h

@@ -275,31 +275,6 @@ typedef struct {
   SEpSet  epSet;
 } SStreamChildEpInfo;
 
-typedef struct {
-  int32_t srcNodeId;
-  int32_t srcChildId;
-  int64_t stateSaveVer;
-  int64_t stateProcessedVer;
-} SStreamCheckpointInfo;
-
-typedef struct {
-  int64_t streamId;
-  int64_t checkTs;
-  int32_t checkpointId;  // incremental
-  int32_t taskId;
-  SArray* checkpointVer;  // SArray<SStreamCheckpointInfo>
-} SStreamMultiVgCheckpointInfo;
-
-typedef struct {
-  int32_t taskId;
-  int32_t checkpointId;  // incremental
-} SStreamCheckpointKey;
-
-typedef struct {
-  int32_t taskId;
-  SArray* checkpointVer;
-} SStreamRecoveringState;
-
 typedef struct SStreamTask {
   int64_t streamId;
   int32_t taskId;
@@ -364,6 +339,10 @@ typedef struct SStreamTask {
   int64_t checkReqId;
   SArray* checkReqIds;  // shuffle
   int32_t refCnt;
+
+  int64_t checkpointingId;
+  int32_t checkpointAlignCnt;
+
 } SStreamTask;
 
 int32_t tEncodeStreamEpInfo(SEncoder* pEncoder, const SStreamChildEpInfo* pInfo);
@@ -509,6 +488,60 @@ typedef struct {
 int32_t tEncodeSStreamRecoverFinishReq(SEncoder* pEncoder, const SStreamRecoverFinishReq* pReq);
 int32_t tDecodeSStreamRecoverFinishReq(SDecoder* pDecoder, SStreamRecoverFinishReq* pReq);
 
+typedef struct {
+  int64_t streamId;
+  int64_t checkpointId;
+  int32_t taskId;
+  int32_t nodeId;
+  int64_t expireTime;
+} SStreamCheckpointSourceReq;
+
+typedef struct {
+  int64_t streamId;
+  int64_t checkpointId;
+  int32_t taskId;
+  int32_t nodeId;
+  int64_t expireTime;
+} SStreamCheckpointSourceRsp;
+
+int32_t tEncodeSStreamCheckpointSourceReq(SEncoder* pEncoder, const SStreamCheckpointSourceReq* pReq);
+int32_t tDecodeSStreamCheckpointSourceReq(SDecoder* pDecoder, SStreamCheckpointSourceReq* pReq);
+
+int32_t tEncodeSStreamCheckpointSourceRsp(SEncoder* pEncoder, const SStreamCheckpointSourceRsp* pRsp);
+int32_t tDecodeSStreamCheckpointSourceRsp(SDecoder* pDecoder, SStreamCheckpointSourceRsp* pRsp);
+
+typedef struct {
+  SMsgHead msgHead;
+  int64_t  streamId;
+  int64_t  checkpointId;
+  int32_t  downstreamTaskId;
+  int32_t  downstreamNodeId;
+  int32_t  upstreamTaskId;
+  int32_t  upstreamNodeId;
+  int32_t  childId;
+  int64_t  expireTime;
+  int8_t   taskLevel;
+} SStreamCheckpointReq;
+
+typedef struct {
+  SMsgHead msgHead;
+  int64_t  streamId;
+  int64_t  checkpointId;
+  int32_t  downstreamTaskId;
+  int32_t  downstreamNodeId;
+  int32_t  upstreamTaskId;
+  int32_t  upstreamNodeId;
+  int32_t  childId;
+  int64_t  expireTime;
+  int8_t   taskLevel;
+} SStreamCheckpointRsp;
+
+int32_t tEncodeSStreamCheckpointReq(SEncoder* pEncoder, const SStreamCheckpointReq* pReq);
+int32_t tDecodeSStreamCheckpointReq(SDecoder* pDecoder, SStreamCheckpointReq* pReq);
+
+int32_t tEncodeSStreamCheckpointRsp(SEncoder* pEncoder, const SStreamCheckpointRsp* pRsp);
+int32_t tDecodeSStreamCheckpointRsp(SDecoder* pDecoder, SStreamCheckpointRsp* pRsp);
+
 typedef struct {
   int64_t streamId;
   int32_t downstreamTaskId;
@@ -598,18 +631,22 @@ void         streamMetaClose(SStreamMeta* streamMeta);
 
 int32_t      streamMetaAddTask(SStreamMeta* pMeta, int64_t ver, SStreamTask* pTask);
 int32_t      streamMetaAddSerializedTask(SStreamMeta* pMeta, int64_t startVer, char* msg, int32_t msgLen);
-int32_t      streamMetaRemoveTask(SStreamMeta* pMeta, int32_t taskId);
 SStreamTask* streamMetaGetTask(SStreamMeta* pMeta, int32_t taskId);
 
 SStreamTask* streamMetaAcquireTask(SStreamMeta* pMeta, int32_t taskId);
 void         streamMetaReleaseTask(SStreamMeta* pMeta, SStreamTask* pTask);
-void         streamMetaRemoveTask1(SStreamMeta* pMeta, int32_t taskId);
+void         streamMetaRemoveTask(SStreamMeta* pMeta, int32_t taskId);
 
 int32_t streamMetaBegin(SStreamMeta* pMeta);
 int32_t streamMetaCommit(SStreamMeta* pMeta);
 int32_t streamMetaRollBack(SStreamMeta* pMeta);
 int32_t streamLoadTasks(SStreamMeta* pMeta);
 
+// checkpoint
+int32_t streamProcessCheckpointSourceReq(SStreamMeta* pMeta, SStreamTask* pTask, SStreamCheckpointSourceReq* pReq);
+int32_t streamProcessCheckpointReq(SStreamMeta* pMeta, SStreamTask* pTask, SStreamCheckpointReq* pReq);
+int32_t streamProcessCheckpointRsp(SStreamMeta* pMeta, SStreamTask* pTask, SStreamCheckpointRsp* pRsp);
+
 #ifdef __cplusplus
 }
 #endif

include/libs/sync/sync.h

@@ -40,6 +40,8 @@ extern "C" {
 #define SNAPSHOT_MAX_CLOCK_SKEW_MS   1000 * 10
 #define SNAPSHOT_WAIT_MS             1000 * 30
 
+#define SYNC_MAX_RETRY_BACKOFF         5
+#define SYNC_LOG_REPL_RETRY_WAIT_MS    100
 #define SYNC_APPEND_ENTRIES_TIMEOUT_MS 10000
 #define SYNC_HEART_TIMEOUT_MS          1000 * 8
 
@@ -49,7 +51,7 @@ extern "C" {
 #define SYNC_MAX_BATCH_SIZE 1
 #define SYNC_INDEX_BEGIN    0
 #define SYNC_INDEX_INVALID  -1
-#define SYNC_TERM_INVALID   0xFFFFFFFFFFFFFFFF
+#define SYNC_TERM_INVALID   -1  // 0xFFFFFFFFFFFFFFFF
 
 typedef enum {
   SYNC_STRATEGY_NO_SNAPSHOT = 0,
@@ -60,7 +62,7 @@ typedef enum {
 typedef uint64_t SyncNodeId;
 typedef int32_t  SyncGroupId;
 typedef int64_t  SyncIndex;
-typedef uint64_t SyncTerm;
+typedef int64_t  SyncTerm;
 
 typedef struct SSyncNode      SSyncNode;
 typedef struct SWal           SWal;
@@ -136,13 +138,13 @@ typedef struct SSnapshotMeta {
 typedef struct SSyncFSM {
   void* data;
 
-  void (*FpCommitCb)(const struct SSyncFSM* pFsm, const SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
+  void (*FpCommitCb)(const struct SSyncFSM* pFsm, SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
-  void (*FpPreCommitCb)(const struct SSyncFSM* pFsm, const SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
+  void (*FpPreCommitCb)(const struct SSyncFSM* pFsm, SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
-  void (*FpRollBackCb)(const struct SSyncFSM* pFsm, const SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
+  void (*FpRollBackCb)(const struct SSyncFSM* pFsm, SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
 
   void (*FpRestoreFinishCb)(const struct SSyncFSM* pFsm);
-  void (*FpReConfigCb)(const struct SSyncFSM* pFsm, const SRpcMsg* pMsg, const SReConfigCbMeta* pMeta);
+  void (*FpReConfigCb)(const struct SSyncFSM* pFsm, SRpcMsg* pMsg, const SReConfigCbMeta* pMeta);
-  void (*FpLeaderTransferCb)(const struct SSyncFSM* pFsm, const SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
+  void (*FpLeaderTransferCb)(const struct SSyncFSM* pFsm, SRpcMsg* pMsg, const SFsmCbMeta* pMeta);
   bool (*FpApplyQueueEmptyCb)(const struct SSyncFSM* pFsm);
   int32_t (*FpApplyQueueItems)(const struct SSyncFSM* pFsm);
 
@@ -211,15 +213,20 @@ typedef struct SSyncInfo {
   int32_t (*syncEqCtrlMsg)(const SMsgCb* msgcb, SRpcMsg* pMsg);
 } SSyncInfo;
 
+// if state == leader
+//     if restored, display "leader"
+//     if !restored && canRead, display "leader*"
+//     if !restored && !canRead, display "leader**"
 typedef struct SSyncState {
   ESyncState state;
   bool       restored;
+  bool       canRead;
 } SSyncState;
 
 int32_t syncInit();
 void    syncCleanUp();
 int64_t syncOpen(SSyncInfo* pSyncInfo);
-void    syncStart(int64_t rid);
+int32_t syncStart(int64_t rid);
 void    syncStop(int64_t rid);
 void    syncPreStop(int64_t rid);
 int32_t syncPropose(int64_t rid, SRpcMsg* pMsg, bool isWeak);

include/libs/wal/wal.h

@@ -170,7 +170,7 @@ int32_t walWriteWithSyncInfo(SWal *, int64_t index, tmsg_t msgType, SWalSyncInfo
 
 // Assign version automatically and return to caller,
 // -1 will be returned for failed writes
-int64_t walAppendLog(SWal *, tmsg_t msgType, SWalSyncInfo syncMeta, const void *body, int32_t bodyLen);
+int64_t walAppendLog(SWal *, int64_t index, tmsg_t msgType, SWalSyncInfo syncMeta, const void *body, int32_t bodyLen);
 
 void walFsync(SWal *, bool force);
 

include/os/osTime.h

@@ -35,6 +35,7 @@ extern "C" {
 #ifdef WINDOWS
 
 #define CLOCK_REALTIME 0
+#define CLOCK_MONOTONIC 0
 
 #define MILLISECOND_PER_SECOND (1000i64)
 #else
@@ -82,6 +83,13 @@ static FORCE_INLINE int64_t taosGetTimestampNs() {
   return (int64_t)systemTime.tv_sec * 1000000000LL + (int64_t)systemTime.tv_nsec;
 }
 
+//@return timestamp of monotonic clock in millisecond
+static FORCE_INLINE int64_t taosGetMonoTimestampMs() {
+  struct timespec systemTime = {0};
+  taosClockGetTime(CLOCK_MONOTONIC, &systemTime);
+  return (int64_t)systemTime.tv_sec * 1000LL + (int64_t)systemTime.tv_nsec / 1000000;
+}
+
 char      *taosStrpTime(const char *buf, const char *fmt, struct tm *tm);
 struct tm *taosLocalTime(const time_t *timep, struct tm *result);
 struct tm *taosLocalTimeNolock(struct tm *result, const time_t *timep, int dst);

include/util/tdef.h

@@ -281,6 +281,7 @@ typedef enum ELogicConditionType {
 #define TSDB_DNODE_ROLE_VNODE 2
 
 #define TSDB_MAX_REPLICA 5
+#define TSDB_SYNC_LOG_BUFFER_SIZE 4096
 
 #define TSDB_TBNAME_COLUMN_INDEX     (-1)
 #define TSDB_MULTI_TABLEMETA_MAX_NUM 100000  // maximum batch size allowed to load table meta

source/client/src/clientSml.c

@@ -139,6 +139,8 @@ typedef struct {
   int32_t numOfSTables;
   int32_t numOfCTables;
   int32_t numOfCreateSTables;
+  int32_t numOfAlterColSTables;
+  int32_t numOfAlterTagSTables;
 
   int64_t parseTime;
   int64_t schemaTime;
@@ -512,6 +514,7 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
           goto end;
         }
 
+        info->cost.numOfAlterTagSTables++;
         taosMemoryFreeClear(pTableMeta);
         code = catalogRefreshTableMeta(info->pCatalog, &conn, &pName, -1);
         if (code != TSDB_CODE_SUCCESS) {
@@ -559,6 +562,8 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
           goto end;
         }
 
+        info->cost.numOfAlterColSTables++;
+        taosMemoryFreeClear(pTableMeta);
         code = catalogRefreshTableMeta(info->pCatalog, &conn, &pName, -1);
         if (code != TSDB_CODE_SUCCESS) {
           goto end;
@@ -820,11 +825,6 @@ static int8_t smlGetTsTypeByPrecision(int8_t precision) {
 }
 
 static int64_t smlParseInfluxTime(SSmlHandle *info, const char *data, int32_t len) {
-  void *tmp = taosMemoryCalloc(1, len + 1);
-  memcpy(tmp, data, len);
-  uDebug("SML:0x%" PRIx64 " smlParseInfluxTime tslen:%d, ts:%s", info->id, len, (char*)tmp);
-  taosMemoryFree(tmp);
-
   if (len == 0 || (len == 1 && data[0] == '0')) {
     return taosGetTimestampNs();
   }
@@ -877,7 +877,10 @@ static int32_t smlParseTS(SSmlHandle *info, const char *data, int32_t len, SArra
   }
   uDebug("SML:0x%" PRIx64 " smlParseTS:%" PRId64, info->id, ts);
 
-  if (ts == -1) return TSDB_CODE_INVALID_TIMESTAMP;
+  if (ts <= 0) {
+    uError("SML:0x%" PRIx64 " smlParseTS error:%" PRId64, info->id, ts);
+    return TSDB_CODE_INVALID_TIMESTAMP;
+  }
 
   // add ts to
   SSmlKv *kv = (SSmlKv *)taosMemoryCalloc(sizeof(SSmlKv), 1);
@@ -2076,10 +2079,7 @@ static int32_t smlParseJSONString(SSmlHandle *info, cJSON *root, SSmlTableInfo *
 
 static int32_t smlParseInfluxLine(SSmlHandle *info, const char *sql, const int len) {
   SSmlLineInfo elements = {0};
-  void *tmp = taosMemoryCalloc(1, len + 1);
+  uDebug("SML:0x%" PRIx64 " smlParseInfluxLine raw:%d, len:%d, sql:%s", info->id, info->isRawLine, len, (info->isRawLine ? "rawdata" : sql));
-  memcpy(tmp, sql, len);
-  uDebug("SML:0x%" PRIx64 " smlParseInfluxLine raw:%d, len:%d, sql:%s", info->id, info->isRawLine, len, (info->isRawLine ? (char*)tmp : sql));
-  taosMemoryFree(tmp);
 
   int ret = smlParseInfluxString(sql, sql + len, &elements, &info->msgBuf);
   if (ret != TSDB_CODE_SUCCESS) {
@@ -2372,11 +2372,12 @@ static int32_t smlInsertData(SSmlHandle *info) {
 
 static void smlPrintStatisticInfo(SSmlHandle *info) {
   uError("SML:0x%" PRIx64
-         " smlInsertLines result, code:%d,lineNum:%d,stable num:%d,ctable num:%d,create stable num:%d \
+         " smlInsertLines result, code:%d,lineNum:%d,stable num:%d,ctable num:%d,create stable num:%d,alter stable tag num:%d,alter stable col num:%d \
         parse cost:%" PRId64 ",schema cost:%" PRId64 ",bind cost:%" PRId64 ",rpc cost:%" PRId64 ",total cost:%" PRId64
          "",
          info->id, info->cost.code, info->cost.lineNum, info->cost.numOfSTables, info->cost.numOfCTables,
-         info->cost.numOfCreateSTables, info->cost.schemaTime - info->cost.parseTime,
+         info->cost.numOfCreateSTables, info->cost.numOfAlterTagSTables, info->cost.numOfAlterColSTables,
+         info->cost.schemaTime - info->cost.parseTime,
          info->cost.insertBindTime - info->cost.schemaTime, info->cost.insertRpcTime - info->cost.insertBindTime,
          info->cost.endTime - info->cost.insertRpcTime, info->cost.endTime - info->cost.parseTime);
 }

source/common/src/systable.c

@@ -273,6 +273,12 @@ static const SSysDbTableSchema vnodesSchema[] = {
     {.name = "dnode_ep", .bytes = TSDB_EP_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = true},
 };
 
+static const SSysDbTableSchema userUserPrivilegesSchema[] = {
+    {.name = "user_name", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
+    {.name = "privilege", .bytes = 10 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
+    {.name = "object_name", .bytes = TSDB_DB_NAME_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
+};
+
 static const SSysTableMeta infosMeta[] = {
     {TSDB_INS_TABLE_DNODES, dnodesSchema, tListLen(dnodesSchema), true},
     {TSDB_INS_TABLE_MNODES, mnodesSchema, tListLen(mnodesSchema), true},
@@ -297,6 +303,7 @@ static const SSysTableMeta infosMeta[] = {
     {TSDB_INS_TABLE_STREAMS, streamSchema, tListLen(streamSchema), false},
     {TSDB_INS_TABLE_STREAM_TASKS, streamTaskSchema, tListLen(streamTaskSchema), false},
     {TSDB_INS_TABLE_VNODES, vnodesSchema, tListLen(vnodesSchema), true},
+    {TSDB_INS_TABLE_USER_PRIVILEGES, userUserPrivilegesSchema, tListLen(userUserPrivilegesSchema), false},
 };
 
 static const SSysDbTableSchema connectionsSchema[] = {

source/common/src/tglobal.c

@@ -167,6 +167,7 @@ int64_t tsWalFsyncDataSizeLimit = (100 * 1024 * 1024L);
 // internal
 int32_t tsTransPullupInterval = 2;
 int32_t tsMqRebalanceInterval = 2;
+int32_t tsStreamCheckpointTickInterval = 1;
 int32_t tsTtlUnit = 86400;
 int32_t tsTtlPushInterval = 86400;
 int32_t tsGrantHBInterval = 60;

source/common/src/tmsg.c

@@ -992,15 +992,20 @@ int32_t tSerializeSStatusReq(void *buf, int32_t bufLen, SStatusReq *pReq) {
   if (tEncodeI32(&encoder, vlen) < 0) return -1;
   for (int32_t i = 0; i < vlen; ++i) {
     SVnodeLoad *pload = taosArrayGet(pReq->pVloads, i);
+    int64_t reserved = 0;
     if (tEncodeI32(&encoder, pload->vgId) < 0) return -1;
     if (tEncodeI8(&encoder, pload->syncState) < 0) return -1;
     if (tEncodeI8(&encoder, pload->syncRestore) < 0) return -1;
+    if (tEncodeI8(&encoder, pload->syncCanRead) < 0) return -1;
     if (tEncodeI64(&encoder, pload->cacheUsage) < 0) return -1;
     if (tEncodeI64(&encoder, pload->numOfTables) < 0) return -1;
     if (tEncodeI64(&encoder, pload->numOfTimeSeries) < 0) return -1;
     if (tEncodeI64(&encoder, pload->totalStorage) < 0) return -1;
     if (tEncodeI64(&encoder, pload->compStorage) < 0) return -1;
     if (tEncodeI64(&encoder, pload->pointsWritten) < 0) return -1;
+    if (tEncodeI64(&encoder, reserved) < 0) return -1;
+    if (tEncodeI64(&encoder, reserved) < 0) return -1;
+    if (tEncodeI64(&encoder, reserved) < 0) return -1;
   }
 
   // mnode loads
@@ -1065,15 +1070,20 @@ int32_t tDeserializeSStatusReq(void *buf, int32_t bufLen, SStatusReq *pReq) {
 
   for (int32_t i = 0; i < vlen; ++i) {
     SVnodeLoad vload = {0};
+    int64_t    reserved = 0;
     if (tDecodeI32(&decoder, &vload.vgId) < 0) return -1;
     if (tDecodeI8(&decoder, &vload.syncState) < 0) return -1;
     if (tDecodeI8(&decoder, &vload.syncRestore) < 0) return -1;
+    if (tDecodeI8(&decoder, &vload.syncCanRead) < 0) return -1;
     if (tDecodeI64(&decoder, &vload.cacheUsage) < 0) return -1;
     if (tDecodeI64(&decoder, &vload.numOfTables) < 0) return -1;
     if (tDecodeI64(&decoder, &vload.numOfTimeSeries) < 0) return -1;
     if (tDecodeI64(&decoder, &vload.totalStorage) < 0) return -1;
     if (tDecodeI64(&decoder, &vload.compStorage) < 0) return -1;
     if (tDecodeI64(&decoder, &vload.pointsWritten) < 0) return -1;
+    if (tDecodeI64(&decoder, &reserved) < 0) return -1;
+    if (tDecodeI64(&decoder, &reserved) < 0) return -1;
+    if (tDecodeI64(&decoder, &reserved) < 0) return -1;
     if (taosArrayPush(pReq->pVloads, &vload) == NULL) {
       terrno = TSDB_CODE_OUT_OF_MEMORY;
       return -1;
@@ -1288,7 +1298,7 @@ int32_t tSerializeSAlterUserReq(void *buf, int32_t bufLen, SAlterUserReq *pReq)
   if (tEncodeI8(&encoder, pReq->enable) < 0) return -1;
   if (tEncodeCStr(&encoder, pReq->user) < 0) return -1;
   if (tEncodeCStr(&encoder, pReq->pass) < 0) return -1;
-  if (tEncodeCStr(&encoder, pReq->dbname) < 0) return -1;
+  if (tEncodeCStr(&encoder, pReq->objname) < 0) return -1;
   tEndEncode(&encoder);
 
   int32_t tlen = encoder.pos;
@@ -1307,7 +1317,7 @@ int32_t tDeserializeSAlterUserReq(void *buf, int32_t bufLen, SAlterUserReq *pReq
   if (tDecodeI8(&decoder, &pReq->enable) < 0) return -1;
   if (tDecodeCStrTo(&decoder, pReq->user) < 0) return -1;
   if (tDecodeCStrTo(&decoder, pReq->pass) < 0) return -1;
-  if (tDecodeCStrTo(&decoder, pReq->dbname) < 0) return -1;
+  if (tDecodeCStrTo(&decoder, pReq->objname) < 0) return -1;
   tEndDecode(&decoder);
 
   tDecoderClear(&decoder);
@@ -3738,6 +3748,31 @@ int32_t tDeserializeSMTimerMsg(void *buf, int32_t bufLen, SMTimerReq *pReq) {
   return 0;
 }
 
+int32_t tSerializeSMStreamTickMsg(void *buf, int32_t bufLen, SMStreamTickReq *pReq) {
+  SEncoder encoder = {0};
+  tEncoderInit(&encoder, buf, bufLen);
+
+  if (tStartEncode(&encoder) < 0) return -1;
+  if (tEncodeI64(&encoder, pReq->tick) < 0) return -1;
+  tEndEncode(&encoder);
+
+  int32_t tlen = encoder.pos;
+  tEncoderClear(&encoder);
+  return tlen;
+}
+
+int32_t tDeserializeSMStreamTickMsg(void *buf, int32_t bufLen, SMStreamTickReq *pReq) {
+  SDecoder decoder = {0};
+  tDecoderInit(&decoder, buf, bufLen);
+
+  if (tStartDecode(&decoder) < 0) return -1;
+  if (tDecodeI64(&decoder, &pReq->tick) < 0) return -1;
+  tEndDecode(&decoder);
+
+  tDecoderClear(&decoder);
+  return 0;
+}
+
 int32_t tEncodeSReplica(SEncoder *pEncoder, SReplica *pReplica) {
   if (tEncodeI32(pEncoder, pReplica->id) < 0) return -1;
   if (tEncodeU16(pEncoder, pReplica->port) < 0) return -1;

source/dnode/mgmt/mgmt_mnode/src/mmHandle.c

@@ -196,7 +196,7 @@ SArray *mmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_PRE_SNAPSHOT, mmPutMsgToSyncQueue, 1) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_PRE_SNAPSHOT_REPLY, mmPutMsgToSyncQueue, 1) == NULL) goto _OVER;
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_HEARTBEAT, mmPutMsgToSyncCtrlQueue, 1) == NULL) goto _OVER;
-  if (dmSetMgmtHandle(pArray, TDMT_SYNC_HEARTBEAT_REPLY, mmPutMsgToSyncQueue, 1) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_SYNC_HEARTBEAT_REPLY, mmPutMsgToSyncCtrlQueue, 1) == NULL) goto _OVER;
 
   code = 0;
 

source/dnode/mgmt/mgmt_mnode/src/mmWorker.c

@@ -162,11 +162,13 @@ int32_t mmPutMsgToQueue(SMnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc) {
   SRpcMsg *pMsg = taosAllocateQitem(sizeof(SRpcMsg), RPC_QITEM);
   if (pMsg == NULL) return -1;
   memcpy(pMsg, pRpc, sizeof(SRpcMsg));
+  pRpc->pCont = NULL;
 
   dTrace("msg:%p, is created and will put into %s queue, type:%s", pMsg, pWorker->name, TMSG_INFO(pRpc->msgType));
   int32_t code = mmPutMsgToWorker(pMgmt, pWorker, pMsg);
   if (code != 0) {
     dTrace("msg:%p, is freed", pMsg);
+    rpcFreeCont(pMsg->pCont);
     taosFreeQitem(pMsg);
   }
   return code;

source/dnode/mgmt/mgmt_qnode/src/qmWorker.c

@@ -61,6 +61,7 @@ int32_t qmPutRpcMsgToQueue(SQnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc) {
   SRpcMsg *pMsg = taosAllocateQitem(sizeof(SRpcMsg), RPC_QITEM);
   if (pMsg == NULL) return -1;
   memcpy(pMsg, pRpc, sizeof(SRpcMsg));
+  pRpc->pCont = NULL;
 
   switch (qtype) {
     case QUERY_QUEUE:
@@ -74,6 +75,7 @@ int32_t qmPutRpcMsgToQueue(SQnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc) {
       return 0;
     default:
       terrno = TSDB_CODE_INVALID_PARA;
+      rpcFreeCont(pMsg->pCont);
       taosFreeQitem(pMsg);
       return -1;
   }

source/dnode/mgmt/mgmt_snode/src/smWorker.c

@@ -151,6 +151,7 @@ int32_t smPutMsgToQueue(SSnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc) {
   pHead->contLen = htonl(pHead->contLen);
   pHead->vgId = SNODE_HANDLE;
   memcpy(pMsg, pRpc, sizeof(SRpcMsg));
+  pRpc->pCont = NULL;
 
   switch (qtype) {
     case STREAM_QUEUE:

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

@@ -464,7 +464,7 @@ SArray *vmGetMsgHandles() {
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_PRE_SNAPSHOT_REPLY, vmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
 
   if (dmSetMgmtHandle(pArray, TDMT_SYNC_HEARTBEAT, vmPutMsgToSyncCtrlQueue, 0) == NULL) goto _OVER;
-  if (dmSetMgmtHandle(pArray, TDMT_SYNC_HEARTBEAT_REPLY, vmPutMsgToSyncQueue, 0) == NULL) goto _OVER;
+  if (dmSetMgmtHandle(pArray, TDMT_SYNC_HEARTBEAT_REPLY, vmPutMsgToSyncCtrlQueue, 0) == NULL) goto _OVER;
 
   code = 0;
 

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

@@ -246,12 +246,12 @@ int32_t vmPutRpcMsgToQueue(SVnodeMgmt *pMgmt, EQueueType qtype, SRpcMsg *pRpc) {
   pHead->contLen = htonl(pHead->contLen);
   pHead->vgId = htonl(pHead->vgId);
   memcpy(pMsg, pRpc, sizeof(SRpcMsg));
+  pRpc->pCont = NULL;
 
   int32_t code = vmPutMsgToQueue(pMgmt, pMsg, qtype);
   if (code != 0) {
     dTrace("msg:%p, is freed", pMsg);
     rpcFreeCont(pMsg->pCont);
-    pRpc->pCont = NULL;
     taosFreeQitem(pMsg);
   }
 

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

@@ -71,6 +71,9 @@ typedef enum {
   MND_OPER_READ_DB,
   MND_OPER_READ_OR_WRITE_DB,
   MND_OPER_SHOW_VARIBALES,
+  MND_OPER_SUBSCRIBE,
+  MND_OPER_CREATE_TOPIC,
+  MND_OPER_DROP_TOPIC,
 } EOperType;
 
 typedef enum {
@@ -273,6 +276,7 @@ typedef struct {
   int32_t   authVersion;
   SHashObj* readDbs;
   SHashObj* writeDbs;
+  SHashObj* topics;
   SRWLatch  lock;
 } SUserObj;
 
@@ -328,6 +332,7 @@ typedef struct {
   int32_t    dnodeId;
   ESyncState syncState;
   bool       syncRestore;
+  bool       syncCanRead;
 } SVnodeGid;
 
 typedef struct {
@@ -635,10 +640,14 @@ typedef struct {
   SArray*        tasks;  // SArray<SArray<SStreamTask>>
   SSchemaWrapper outputSchema;
   SSchemaWrapper tagSchema;
+
+  // 3.0.20
+  int64_t checkpointFreq;  // ms
+  int64_t currentTick;     // do not serialize
 } SStreamObj;
 
 int32_t tEncodeSStreamObj(SEncoder* pEncoder, const SStreamObj* pObj);
-int32_t tDecodeSStreamObj(SDecoder* pDecoder, SStreamObj* pObj);
+int32_t tDecodeSStreamObj(SDecoder* pDecoder, SStreamObj* pObj, int32_t sver);
 void    tFreeStreamObj(SStreamObj* pObj);
 
 typedef struct {
@@ -648,15 +657,6 @@ typedef struct {
   SArray* childInfo;  // SArray<SStreamChildEpInfo>
 } SStreamCheckpointObj;
 
-#if 0
-typedef struct {
-  int64_t uid;
-  int64_t streamId;
-  int8_t  status;
-  int8_t  stage;
-} SStreamRecoverObj;
-#endif
-
 #ifdef __cplusplus
 }
 #endif

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

@@ -28,6 +28,8 @@ void    mndCleanupPrivilege(SMnode *pMnode);
 int32_t mndCheckOperPrivilege(SMnode *pMnode, const char *user, EOperType operType);
 int32_t mndCheckDbPrivilege(SMnode *pMnode, const char *user, EOperType operType, SDbObj *pDb);
 int32_t mndCheckDbPrivilegeByName(SMnode *pMnode, const char *user, EOperType operType, const char *dbname);
+int32_t mndCheckTopicPrivilege(SMnode *pMnode, const char *user, EOperType operType, SMqTopicObj *pTopic);
+int32_t mndCheckTopicPrivilegeByName(SMnode *pMnode, const char *user, EOperType operType, const char *topicName);
 int32_t mndCheckShowPrivilege(SMnode *pMnode, const char *user, EShowType showType, const char *dbname);
 int32_t mndCheckAlterUserPrivilege(SUserObj *pOperUser, SUserObj *pUser, SAlterUserReq *pAlter);
 int32_t mndSetUserAuthRsp(SMnode *pMnode, SUserObj *pUser, SGetUserAuthRsp *pRsp);

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

@@ -25,7 +25,7 @@ extern "C" {
 int32_t mndInitTopic(SMnode *pMnode);
 void    mndCleanupTopic(SMnode *pMnode);
 
-SMqTopicObj *mndAcquireTopic(SMnode *pMnode, char *topicName);
+SMqTopicObj *mndAcquireTopic(SMnode *pMnode, const char *topicName);
 void         mndReleaseTopic(SMnode *pMnode, SMqTopicObj *pTopic);
 
 SSdbRaw *mndTopicActionEncode(SMqTopicObj *pTopic);

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

@@ -147,6 +147,8 @@ _OVER:
 
 static SSdbRow *mndAcctActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SAcctObj *pAcct = NULL;
+  SSdbRow  *pRow = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -156,10 +158,10 @@ static SSdbRow *mndAcctActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SAcctObj));
+  pRow = sdbAllocRow(sizeof(SAcctObj));
   if (pRow == NULL) goto _OVER;
 
-  SAcctObj *pAcct = sdbGetRowObj(pRow);
+  pAcct = sdbGetRowObj(pRow);
   if (pAcct == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -186,7 +188,7 @@ static SSdbRow *mndAcctActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("acct:%s, failed to decode from raw:%p since %s", pAcct->acct, pRaw, terrstr());
+    mError("acct:%s, failed to decode from raw:%p since %s", pAcct == NULL ? "null" : pAcct->acct, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -157,6 +157,8 @@ _OVER:
 
 static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SClusterObj *pCluster = NULL;
+  SSdbRow *pRow = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -166,10 +168,10 @@ static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SClusterObj));
+  pRow = sdbAllocRow(sizeof(SClusterObj));
   if (pRow == NULL) goto _OVER;
 
-  SClusterObj *pCluster = sdbGetRowObj(pRow);
+  pCluster = sdbGetRowObj(pRow);
   if (pCluster == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -184,7 +186,8 @@ static SSdbRow *mndClusterActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("cluster:%" PRId64 ", failed to decode from raw:%p since %s", pCluster->id, pRaw, terrstr());
+    mError("cluster:%" PRId64 ", failed to decode from raw:%p since %s", pCluster == NULL ? 0 : pCluster->id, pRaw,
+           terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -712,6 +712,8 @@ CM_ENCODE_OVER:
 
 SSdbRow *mndConsumerActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow        *pRow = NULL;
+  SMqConsumerObj *pConsumer = NULL;
   void           *buf = NULL;
 
   int8_t sver = 0;
@@ -722,10 +724,10 @@ SSdbRow *mndConsumerActionDecode(SSdbRaw *pRaw) {
     goto CM_DECODE_OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SMqConsumerObj));
+  pRow = sdbAllocRow(sizeof(SMqConsumerObj));
   if (pRow == NULL) goto CM_DECODE_OVER;
 
-  SMqConsumerObj *pConsumer = sdbGetRowObj(pRow);
+  pConsumer = sdbGetRowObj(pRow);
   if (pConsumer == NULL) goto CM_DECODE_OVER;
 
   int32_t dataPos = 0;
@@ -745,7 +747,8 @@ SSdbRow *mndConsumerActionDecode(SSdbRaw *pRaw) {
 CM_DECODE_OVER:
   taosMemoryFreeClear(buf);
   if (terrno != TSDB_CODE_SUCCESS) {
-    mError("consumer:%" PRId64 ", failed to decode from raw:%p since %s", pConsumer->consumerId, pRaw, terrstr());
+    mError("consumer:%" PRId64 ", failed to decode from raw:%p since %s", pConsumer == NULL ? 0 : pConsumer->consumerId,
+           pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -147,6 +147,8 @@ _OVER:
 
 static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow *pRow = NULL;
+  SDbObj  *pDb = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -156,10 +158,10 @@ static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SDbObj));
+  pRow = sdbAllocRow(sizeof(SDbObj));
   if (pRow == NULL) goto _OVER;
 
-  SDbObj *pDb = sdbGetRowObj(pRow);
+  pDb = sdbGetRowObj(pRow);
   if (pDb == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -232,7 +234,7 @@ static SSdbRow *mndDbActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("db:%s, failed to decode from raw:%p since %s", pDb->name, pRaw, terrstr());
+    mError("db:%s, failed to decode from raw:%p since %s", pDb == NULL ? "null" : pDb->name, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -76,11 +76,14 @@ int32_t tEncodeSStreamObj(SEncoder *pEncoder, const SStreamObj *pObj) {
 
   if (tEncodeSSchemaWrapper(pEncoder, &pObj->outputSchema) < 0) return -1;
 
+  // 3.0.20
+  if (tEncodeI64(pEncoder, pObj->checkpointFreq) < 0) return -1;
+
   tEndEncode(pEncoder);
   return pEncoder->pos;
 }
 
-int32_t tDecodeSStreamObj(SDecoder *pDecoder, SStreamObj *pObj) {
+int32_t tDecodeSStreamObj(SDecoder *pDecoder, SStreamObj *pObj, int32_t sver) {
   if (tStartDecode(pDecoder) < 0) return -1;
   if (tDecodeCStrTo(pDecoder, pObj->name) < 0) return -1;
 
@@ -139,6 +142,10 @@ int32_t tDecodeSStreamObj(SDecoder *pDecoder, SStreamObj *pObj) {
 
   if (tDecodeSSchemaWrapper(pDecoder, &pObj->outputSchema) < 0) return -1;
 
+  // 3.0.20
+  if (sver >= 2) {
+    if (tDecodeI64(pDecoder, &pObj->checkpointFreq) < 0) return -1;
+  }
   tEndDecode(pDecoder);
   return 0;
 }

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

@@ -154,8 +154,9 @@ _OVER:
 }
 
 static SSdbRow *mndDnodeActionDecode(SSdbRaw *pRaw) {
-  SSdbRow *pRow = NULL;
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow   *pRow = NULL;
+  SDnodeObj *pDnode = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -166,7 +167,8 @@ static SSdbRow *mndDnodeActionDecode(SSdbRaw *pRaw) {
 
   pRow = sdbAllocRow(sizeof(SDnodeObj));
   if (pRow == NULL) goto _OVER;
-  SDnodeObj *pDnode = sdbGetRowObj(pRow);
+
+  pDnode = sdbGetRowObj(pRow);
   if (pDnode == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -181,7 +183,7 @@ static SSdbRow *mndDnodeActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("dnode:%d, failed to decode from raw:%p since %s", pDnode->id, pRaw, terrstr());
+    mError("dnode:%d, failed to decode from raw:%p since %s", pDnode == NULL ? 0 : pDnode->id, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }
@@ -375,14 +377,18 @@ static int32_t mndProcessStatusReq(SRpcMsg *pReq) {
       }
       bool roleChanged = false;
       for (int32_t vg = 0; vg < pVgroup->replica; ++vg) {
-        if (pVgroup->vnodeGid[vg].dnodeId == statusReq.dnodeId) {
+        SVnodeGid *pGid = &pVgroup->vnodeGid[vg];
-          if (pVgroup->vnodeGid[vg].syncState != pVload->syncState ||
+        if (pGid->dnodeId == statusReq.dnodeId) {
-              pVgroup->vnodeGid[vg].syncRestore != pVload->syncRestore) {
+          if (pGid->syncState != pVload->syncState || pGid->syncRestore != pVload->syncRestore ||
-            mInfo("vgId:%d, state changed by status msg, old state:%s restored:%d new state:%s restored:%d",
+              pGid->syncCanRead != pVload->syncCanRead) {
-                  pVgroup->vgId, syncStr(pVgroup->vnodeGid[vg].syncState), pVgroup->vnodeGid[vg].syncRestore,
+            mInfo(
-                  syncStr(pVload->syncState), pVload->syncRestore);
+                "vgId:%d, state changed by status msg, old state:%s restored:%d canRead:%d new state:%s restored:%d "
-            pVgroup->vnodeGid[vg].syncState = pVload->syncState;
+                "canRead:%d",
-            pVgroup->vnodeGid[vg].syncRestore = pVload->syncRestore;
+                pVgroup->vgId, syncStr(pGid->syncState), pGid->syncRestore, pGid->syncCanRead,
+                syncStr(pVload->syncState), pVload->syncRestore, pVload->syncCanRead);
+            pGid->syncState = pVload->syncState;
+            pGid->syncRestore = pVload->syncRestore;
+            pGid->syncCanRead = pVload->syncCanRead;
             roleChanged = true;
           }
           break;

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

@@ -101,6 +101,8 @@ _OVER:
 
 static SSdbRow *mndFuncActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow  *pRow = NULL;
+  SFuncObj *pFunc = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -110,10 +112,10 @@ static SSdbRow *mndFuncActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SFuncObj));
+  pRow = sdbAllocRow(sizeof(SFuncObj));
   if (pRow == NULL) goto _OVER;
 
-  SFuncObj *pFunc = sdbGetRowObj(pRow);
+  pFunc = sdbGetRowObj(pRow);
   if (pFunc == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -148,7 +150,7 @@ static SSdbRow *mndFuncActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("func:%s, failed to decode from raw:%p since %s", pFunc->name, pRaw, terrstr());
+    mError("func:%s, failed to decode from raw:%p since %s", pFunc == NULL ? "null" : pFunc->name, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -85,6 +85,21 @@ static void *mndBuildTimerMsg(int32_t *pContLen) {
   return pReq;
 }
 
+static void *mndBuildCheckpointTickMsg(int32_t *pContLen, int64_t sec) {
+  SMStreamTickReq timerReq = {
+      .tick = sec,
+  };
+
+  int32_t contLen = tSerializeSMStreamTickMsg(NULL, 0, &timerReq);
+  if (contLen <= 0) return NULL;
+  void *pReq = rpcMallocCont(contLen);
+  if (pReq == NULL) return NULL;
+
+  tSerializeSMStreamTickMsg(pReq, contLen, &timerReq);
+  *pContLen = contLen;
+  return pReq;
+}
+
 static void mndPullupTrans(SMnode *pMnode) {
   int32_t contLen = 0;
   void   *pReq = mndBuildTimerMsg(&contLen);
@@ -105,7 +120,24 @@ static void mndCalMqRebalance(SMnode *pMnode) {
   int32_t contLen = 0;
   void   *pReq = mndBuildTimerMsg(&contLen);
   if (pReq != NULL) {
-    SRpcMsg rpcMsg = {.msgType = TDMT_MND_TMQ_TIMER, .pCont = pReq, .contLen = contLen};
+    SRpcMsg rpcMsg = {
+        .msgType = TDMT_MND_TMQ_TIMER,
+        .pCont = pReq,
+        .contLen = contLen,
+    };
+    tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
+  }
+}
+
+static void mndStreamCheckpointTick(SMnode *pMnode, int64_t sec) {
+  int32_t contLen = 0;
+  void   *pReq = mndBuildCheckpointTickMsg(&contLen, sec);
+  if (pReq != NULL) {
+    SRpcMsg rpcMsg = {
+        .msgType = TDMT_MND_STREAM_CHECKPOINT_TIMER,
+        .pCont = pReq,
+        .contLen = contLen,
+    };
     tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
   }
 }
@@ -150,12 +182,16 @@ static void mndSetVgroupOffline(SMnode *pMnode, int32_t dnodeId, int64_t curMs)
 
     bool roleChanged = false;
     for (int32_t vg = 0; vg < pVgroup->replica; ++vg) {
-      if (pVgroup->vnodeGid[vg].dnodeId == dnodeId) {
+      SVnodeGid *pGid = &pVgroup->vnodeGid[vg];
-        if (pVgroup->vnodeGid[vg].syncState != TAOS_SYNC_STATE_OFFLINE) {
+      if (pGid->dnodeId == dnodeId) {
-          mInfo("vgId:%d, state changed by offline check, old state:%s restored:%d new state:error restored:0",
+        if (pGid->syncState != TAOS_SYNC_STATE_OFFLINE) {
-                pVgroup->vgId, syncStr(pVgroup->vnodeGid[vg].syncState), pVgroup->vnodeGid[vg].syncRestore);
+          mInfo(
-          pVgroup->vnodeGid[vg].syncState = TAOS_SYNC_STATE_OFFLINE;
+              "vgId:%d, state changed by offline check, old state:%s restored:%d canRead:%d new state:error restored:0 "
-          pVgroup->vnodeGid[vg].syncRestore = 0;
+              "canRead:0",
+              pVgroup->vgId, syncStr(pGid->syncState), pGid->syncRestore, pGid->syncCanRead);
+          pGid->syncState = TAOS_SYNC_STATE_OFFLINE;
+          pGid->syncRestore = 0;
+          pGid->syncCanRead = 0;
           roleChanged = true;
         }
         break;
@@ -220,6 +256,12 @@ static void *mndThreadFp(void *param) {
       mndCalMqRebalance(pMnode);
     }
 
+#if 0
+    if (sec % tsStreamCheckpointTickInterval == 0) {
+      mndStreamCheckpointTick(pMnode, sec);
+    }
+#endif
+
     if (sec % tsTelemInterval == (TMIN(60, (tsTelemInterval - 1)))) {
       mndPullupTelem(pMnode);
     }
@@ -491,7 +533,7 @@ void mndPreClose(SMnode *pMnode) {
   if (pMnode != NULL) {
     syncLeaderTransfer(pMnode->syncMgmt.sync);
     syncPreStop(pMnode->syncMgmt.sync);
-
+#if 0
     while (syncSnapshotRecving(pMnode->syncMgmt.sync)) {
       mInfo("vgId:1, snapshot is recving");
       taosMsleep(300);
@@ -500,6 +542,7 @@ void mndPreClose(SMnode *pMnode) {
       mInfo("vgId:1, snapshot is sending");
       taosMsleep(300);
     }
+#endif
   }
 }
 

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

@@ -142,6 +142,8 @@ _OVER:
 
 static SSdbRow *mndMnodeActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow   *pRow = NULL;
+  SMnodeObj *pObj = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) return NULL;
@@ -151,10 +153,10 @@ static SSdbRow *mndMnodeActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SMnodeObj));
+  pRow = sdbAllocRow(sizeof(SMnodeObj));
   if (pRow == NULL) goto _OVER;
 
-  SMnodeObj *pObj = sdbGetRowObj(pRow);
+  pObj = sdbGetRowObj(pRow);
   if (pObj == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -167,7 +169,7 @@ static SSdbRow *mndMnodeActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("mnode:%d, failed to decode from raw:%p since %s", pObj->id, pRaw, terrstr());
+    mError("mnode:%d, failed to decode from raw:%p since %s", pObj == NULL ? 0 : pObj->id, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -28,6 +28,10 @@ int32_t mndCheckDbPrivilege(SMnode *pMnode, const char *user, EOperType operType
 int32_t mndCheckDbPrivilegeByName(SMnode *pMnode, const char *user, EOperType operType, const char *dbname) {
   return 0;
 }
+int32_t mndCheckTopicPrivilege(SMnode *pMnode, const char *user, EOperType operType, SMqTopicObj *pTopic) { return 0; }
+int32_t mndCheckTopicPrivilegeByName(SMnode *pMnode, const char *user, EOperType operType, const char *topicName) {
+  return 0;
+}
 int32_t mndSetUserAuthRsp(SMnode *pMnode, SUserObj *pUser, SGetUserAuthRsp *pRsp) {
   memcpy(pRsp->user, pUser->user, TSDB_USER_LEN);
   pRsp->superAuth = 1;

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

@@ -100,6 +100,8 @@ _OVER:
 
 static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow   *pRow = NULL;
+  SQnodeObj *pObj = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -109,10 +111,10 @@ static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SQnodeObj));
+  pRow = sdbAllocRow(sizeof(SQnodeObj));
   if (pRow == NULL) goto _OVER;
 
-  SQnodeObj *pObj = sdbGetRowObj(pRow);
+  pObj = sdbGetRowObj(pRow);
   if (pObj == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -125,7 +127,7 @@ static SSdbRow *mndQnodeActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("qnode:%d, failed to decode from raw:%p since %s", pObj->id, pRaw, terrstr());
+    mError("qnode:%d, failed to decode from raw:%p since %s", pObj == NULL ? 0 : pObj->id, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -132,6 +132,8 @@ _OVER:
 
 static SSdbRow *mndSmaActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow *pRow = NULL;
+  SSmaObj *pSma = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -141,10 +143,10 @@ static SSdbRow *mndSmaActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SSmaObj));
+  pRow = sdbAllocRow(sizeof(SSmaObj));
   if (pRow == NULL) goto _OVER;
 
-  SSmaObj *pSma = sdbGetRowObj(pRow);
+  pSma = sdbGetRowObj(pRow);
   if (pSma == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -200,7 +202,7 @@ static SSdbRow *mndSmaActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("sma:%s, failed to decode from raw:%p since %s", pSma->name, pRaw, terrstr());
+    mError("sma:%s, failed to decode from raw:%p since %s", pSma == NULL ? "null" : pSma->name, pRaw, terrstr());
     taosMemoryFreeClear(pSma->expr);
     taosMemoryFreeClear(pSma->tagsFilter);
     taosMemoryFreeClear(pSma->sql);

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

@@ -105,6 +105,8 @@ _OVER:
 
 static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow   *pRow = NULL;
+  SSnodeObj *pObj = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -114,10 +116,10 @@ static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SSnodeObj));
+  pRow = sdbAllocRow(sizeof(SSnodeObj));
   if (pRow == NULL) goto _OVER;
 
-  SSnodeObj *pObj = sdbGetRowObj(pRow);
+  pObj = sdbGetRowObj(pRow);
   if (pObj == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -130,7 +132,7 @@ static SSdbRow *mndSnodeActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("snode:%d, failed to decode from raw:%p since %s", pObj->id, pRaw, terrstr());
+    mError("snode:%d, failed to decode from raw:%p since %s", pObj == NULL ? 0 : pObj->id, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -162,6 +162,8 @@ _OVER:
 
 static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow *pRow = NULL;
+  SStbObj *pStb = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -171,10 +173,10 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SStbObj));
+  pRow = sdbAllocRow(sizeof(SStbObj));
   if (pRow == NULL) goto _OVER;
 
-  SStbObj *pStb = sdbGetRowObj(pRow);
+  pStb = sdbGetRowObj(pRow);
   if (pStb == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -254,10 +256,12 @@ static SSdbRow *mndStbActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("stb:%s, failed to decode from raw:%p since %s", pStb->name, pRaw, terrstr());
+    mError("stb:%s, failed to decode from raw:%p since %s", pStb == NULL ? "null" : pStb->name, pRaw, terrstr());
+    if (pStb != NULL) {
       taosMemoryFreeClear(pStb->pColumns);
       taosMemoryFreeClear(pStb->pTags);
       taosMemoryFreeClear(pStb->comment);
+    }
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -28,7 +28,7 @@
 #include "parser.h"
 #include "tname.h"
 
-#define MND_STREAM_VER_NUMBER   1
+#define MND_STREAM_VER_NUMBER   2
 #define MND_STREAM_RESERVE_SIZE 64
 
 static int32_t mndStreamActionInsert(SSdb *pSdb, SStreamObj *pStream);
@@ -36,6 +36,8 @@ static int32_t mndStreamActionDelete(SSdb *pSdb, SStreamObj *pStream);
 static int32_t mndStreamActionUpdate(SSdb *pSdb, SStreamObj *pStream, SStreamObj *pNewStream);
 static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq);
 static int32_t mndProcessDropStreamReq(SRpcMsg *pReq);
+static int32_t mndProcessStreamCheckpointTmr(SRpcMsg *pReq);
+static int32_t mndProcessStreamDoCheckpoint(SRpcMsg *pReq);
 /*static int32_t mndProcessRecoverStreamReq(SRpcMsg *pReq);*/
 static int32_t mndProcessStreamMetaReq(SRpcMsg *pReq);
 static int32_t mndGetStreamMeta(SRpcMsg *pReq, SShowObj *pShow, STableMetaRsp *pMeta);
@@ -62,6 +64,10 @@ int32_t mndInitStream(SMnode *pMnode) {
   mndSetMsgHandle(pMnode, TDMT_STREAM_TASK_DEPLOY_RSP, mndTransProcessRsp);
   mndSetMsgHandle(pMnode, TDMT_STREAM_TASK_DROP_RSP, mndTransProcessRsp);
 
+  mndSetMsgHandle(pMnode, TDMT_MND_STREAM_CHECKPOINT_TIMER, mndProcessStreamCheckpointTmr);
+  mndSetMsgHandle(pMnode, TDMT_MND_STREAM_BEGIN_CHECKPOINT, mndProcessStreamDoCheckpoint);
+  mndSetMsgHandle(pMnode, TDMT_STREAM_TASK_REPORT_CHECKPOINT, mndTransProcessRsp);
+
   mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_STREAMS, mndRetrieveStream);
   mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_STREAMS, mndCancelGetNextStream);
   mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_STREAM_TASKS, mndRetrieveStreamTask);
@@ -120,21 +126,22 @@ STREAM_ENCODE_OVER:
 
 SSdbRow *mndStreamActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow    *pRow = NULL;
+  SStreamObj *pStream = NULL;
   void       *buf = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto STREAM_DECODE_OVER;
 
-  if (sver != MND_STREAM_VER_NUMBER) {
+  if (sver != 1 && sver != 2) {
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     goto STREAM_DECODE_OVER;
   }
 
-  int32_t  size = sizeof(SStreamObj);
+  pRow = sdbAllocRow(sizeof(SStreamObj));
-  SSdbRow *pRow = sdbAllocRow(size);
   if (pRow == NULL) goto STREAM_DECODE_OVER;
 
-  SStreamObj *pStream = sdbGetRowObj(pRow);
+  pStream = sdbGetRowObj(pRow);
   if (pStream == NULL) goto STREAM_DECODE_OVER;
 
   int32_t tlen;
@@ -146,7 +153,7 @@ SSdbRow *mndStreamActionDecode(SSdbRaw *pRaw) {
 
   SDecoder decoder;
   tDecoderInit(&decoder, buf, tlen + 1);
-  if (tDecodeSStreamObj(&decoder, pStream) < 0) {
+  if (tDecodeSStreamObj(&decoder, pStream, sver) < 0) {
     tDecoderClear(&decoder);
     goto STREAM_DECODE_OVER;
   }
@@ -157,7 +164,7 @@ SSdbRow *mndStreamActionDecode(SSdbRaw *pRaw) {
 STREAM_DECODE_OVER:
   taosMemoryFreeClear(buf);
   if (terrno != TSDB_CODE_SUCCESS) {
-    mError("stream:%s, failed to decode from raw:%p since %s", pStream->name, pRaw, terrstr());
+    mError("stream:%s, failed to decode from raw:%p since %s", pStream == NULL ? "null" : pStream->name, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }
@@ -679,6 +686,164 @@ _OVER:
   tFreeStreamObj(&streamObj);
   return code;
 }
+static int32_t mndProcessStreamCheckpointTmr(SRpcMsg *pReq) {
+  SMnode     *pMnode = pReq->info.node;
+  SSdb       *pSdb = pMnode->pSdb;
+  void       *pIter = NULL;
+  SStreamObj *pStream = NULL;
+
+  // iterate all stream obj
+  while (1) {
+    pIter = sdbFetch(pSdb, SDB_STREAM, pIter, (void **)&pStream);
+    if (pIter == NULL) break;
+    // incr tick
+    int64_t currentTick = atomic_add_fetch_64(&pStream->currentTick, 1);
+    // if >= checkpointFreq, build msg TDMT_MND_STREAM_BEGIN_CHECKPOINT, put into write q
+    if (currentTick >= pStream->checkpointFreq) {
+      atomic_store_64(&pStream->currentTick, 0);
+      SMStreamDoCheckpointMsg *pMsg = rpcMallocCont(sizeof(SMStreamDoCheckpointMsg));
+
+      pMsg->streamId = pStream->uid;
+      pMsg->checkpointId = tGenIdPI64();
+      memcpy(pMsg->streamName, pStream->name, TSDB_STREAM_FNAME_LEN);
+
+      SRpcMsg rpcMsg = {
+          .msgType = TDMT_MND_STREAM_BEGIN_CHECKPOINT,
+          .pCont = pMsg,
+          .contLen = sizeof(SMStreamDoCheckpointMsg),
+      };
+
+      tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
+    }
+  }
+
+  return 0;
+}
+
+static int32_t mndBuildStreamCheckpointSourceReq(void **pBuf, int32_t *pLen, const SStreamTask *pTask,
+                                                 SMStreamDoCheckpointMsg *pMsg) {
+  SStreamCheckpointSourceReq req = {0};
+  req.checkpointId = pMsg->checkpointId;
+  req.nodeId = pTask->nodeId;
+  req.expireTime = -1;
+  req.streamId = pTask->streamId;
+  req.taskId = pTask->taskId;
+
+  int32_t code;
+  int32_t blen;
+
+  tEncodeSize(tEncodeSStreamCheckpointSourceReq, &req, blen, code);
+  if (code < 0) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  int32_t tlen = sizeof(SMsgHead) + blen;
+
+  void *buf = taosMemoryMalloc(tlen);
+  if (buf == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  void    *abuf = POINTER_SHIFT(buf, sizeof(SMsgHead));
+  SEncoder encoder;
+  tEncoderInit(&encoder, abuf, tlen);
+  tEncodeSStreamCheckpointSourceReq(&encoder, &req);
+
+  SMsgHead *pMsgHead = (SMsgHead *)buf;
+  pMsgHead->contLen = htonl(tlen);
+  pMsgHead->vgId = htonl(pTask->nodeId);
+
+  tEncoderClear(&encoder);
+
+  *pBuf = buf;
+  *pLen = tlen;
+
+  return 0;
+}
+
+static int32_t mndProcessStreamDoCheckpoint(SRpcMsg *pReq) {
+  SMnode *pMnode = pReq->info.node;
+  SSdb   *pSdb = pMnode->pSdb;
+
+  SMStreamDoCheckpointMsg *pMsg = (SMStreamDoCheckpointMsg *)pReq->pCont;
+
+  SStreamObj *pStream = mndAcquireStream(pMnode, pMsg->streamName);
+
+  if (pStream == NULL || pStream->uid != pMsg->streamId) {
+    mError("start checkpointing failed since stream %s not found", pMsg->streamName);
+    return -1;
+  }
+
+  // build new transaction:
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_DB_INSIDE, pReq, "stream-checkpoint");
+  if (pTrans == NULL) return -1;
+  mndTransSetDbName(pTrans, pStream->sourceDb, pStream->targetDb);
+  taosRLockLatch(&pStream->lock);
+  // 1. redo action: broadcast checkpoint source msg for all source vg
+  int32_t totLevel = taosArrayGetSize(pStream->tasks);
+  for (int32_t i = 0; i < totLevel; i++) {
+    SArray      *pLevel = taosArrayGetP(pStream->tasks, i);
+    SStreamTask *pTask = taosArrayGetP(pLevel, 0);
+    if (pTask->taskLevel == TASK_LEVEL__SOURCE) {
+      int32_t sz = taosArrayGetSize(pLevel);
+      for (int32_t j = 0; j < sz; j++) {
+        SStreamTask *pTask = taosArrayGetP(pLevel, j);
+        ASSERT(pTask->nodeId > 0);
+        SVgObj *pVgObj = mndAcquireVgroup(pMnode, pTask->nodeId);
+        if (pVgObj == NULL) {
+          ASSERT(0);
+          taosRUnLockLatch(&pStream->lock);
+          mndReleaseStream(pMnode, pStream);
+          mndTransDrop(pTrans);
+          return -1;
+        }
+
+        void   *buf;
+        int32_t tlen;
+        if (mndBuildStreamCheckpointSourceReq(&buf, &tlen, pTask, pMsg) < 0) {
+          taosRUnLockLatch(&pStream->lock);
+          mndReleaseStream(pMnode, pStream);
+          mndTransDrop(pTrans);
+          return -1;
+        }
+
+        STransAction action = {0};
+        action.epSet = mndGetVgroupEpset(pMnode, pVgObj);
+        action.pCont = buf;
+        action.contLen = tlen;
+        action.msgType = TDMT_VND_STREAM_CHECK_POINT_SOURCE;
+
+        mndReleaseVgroup(pMnode, pVgObj);
+
+        if (mndTransAppendRedoAction(pTrans, &action) != 0) {
+          taosMemoryFree(buf);
+          taosRUnLockLatch(&pStream->lock);
+          mndReleaseStream(pMnode, pStream);
+          mndTransDrop(pTrans);
+          return -1;
+        }
+      }
+    }
+  }
+  // 2. reset tick
+  atomic_store_64(&pStream->currentTick, 0);
+  // 3. commit log: stream checkpoint info
+  taosRUnLockLatch(&pStream->lock);
+
+  if (mndTransPrepare(pMnode, pTrans) != 0) {
+    mError("failed to prepare trans rebalance since %s", terrstr());
+    mndTransDrop(pTrans);
+    mndReleaseStream(pMnode, pStream);
+    return -1;
+  }
+
+  mndReleaseStream(pMnode, pStream);
+  mndTransDrop(pTrans);
+
+  return 0;
+}
 
 static int32_t mndProcessDropStreamReq(SRpcMsg *pReq) {
   SMnode     *pMnode = pReq->info.node;
@@ -747,71 +912,6 @@ static int32_t mndProcessDropStreamReq(SRpcMsg *pReq) {
   return TSDB_CODE_ACTION_IN_PROGRESS;
 }
 
-#if 0
-static int32_t mndProcessRecoverStreamReq(SRpcMsg *pReq) {
-  SMnode     *pMnode = pReq->info.node;
-  SStreamObj *pStream = NULL;
-  /*SDbObj     *pDb = NULL;*/
-  /*SUserObj   *pUser = NULL;*/
-
-  SMRecoverStreamReq recoverReq = {0};
-  if (tDeserializeSMRecoverStreamReq(pReq->pCont, pReq->contLen, &recoverReq) < 0) {
-    ASSERT(0);
-    terrno = TSDB_CODE_INVALID_MSG;
-    return -1;
-  }
-
-  pStream = mndAcquireStream(pMnode, recoverReq.name);
-
-  if (pStream == NULL) {
-    if (recoverReq.igNotExists) {
-      mInfo("stream:%s, not exist, ignore not exist is set", recoverReq.name);
-      sdbRelease(pMnode->pSdb, pStream);
-      return 0;
-    } else {
-      terrno = TSDB_CODE_MND_STREAM_NOT_EXIST;
-      return -1;
-    }
-  }
-
-  if (mndCheckDbPrivilegeByName(pMnode, pReq->info.conn.user, MND_OPER_WRITE_DB, pStream->targetDb) != 0) {
-    return -1;
-  }
-
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_NOTHING, pReq);
-  if (pTrans == NULL) {
-    mError("stream:%s, failed to recover since %s", recoverReq.name, terrstr());
-    sdbRelease(pMnode->pSdb, pStream);
-    return -1;
-  }
-  mInfo("trans:%d, used to drop stream:%s", pTrans->id, recoverReq.name);
-
-  // broadcast to recover all tasks
-  if (mndRecoverStreamTasks(pMnode, pTrans, pStream) < 0) {
-    mError("stream:%s, failed to recover task since %s", recoverReq.name, terrstr());
-    sdbRelease(pMnode->pSdb, pStream);
-    return -1;
-  }
-
-  // update stream status
-  if (mndSetStreamRecover(pMnode, pTrans, pStream) < 0) {
-    sdbRelease(pMnode->pSdb, pStream);
-    return -1;
-  }
-
-  if (mndTransPrepare(pMnode, pTrans) != 0) {
-    mError("trans:%d, failed to prepare recover stream trans since %s", pTrans->id, terrstr());
-    sdbRelease(pMnode->pSdb, pStream);
-    mndTransDrop(pTrans);
-    return -1;
-  }
-
-  sdbRelease(pMnode->pSdb, pStream);
-
-  return TSDB_CODE_ACTION_IN_PROGRESS;
-}
-#endif
-
 int32_t mndDropStreamByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb) {
   SSdb *pSdb = pMnode->pSdb;
   void *pIter = NULL;
@@ -846,13 +946,6 @@ int32_t mndDropStreamByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb) {
       }
     }
 
-#if 0
-    if (mndSetDropOffsetStreamLogs(pMnode, pTrans, pStream) < 0) {
-      sdbRelease(pSdb, pStream);
-      goto END;
-    }
-#endif
-
     sdbRelease(pSdb, pStream);
   }
 

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

@@ -440,9 +440,9 @@ static int32_t mndDoRebalance(SMnode *pMnode, const SMqRebInputObj *pInput, SMqR
 }
 
 static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOutputObj *pOutput) {
-  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_CONFLICT_DB_INSIDE, pMsg, "persist-reb");
+  STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_CONFLICT_DB_INSIDE, pMsg, "tmq-reb");
-  mndTransSetDbName(pTrans, pOutput->pSub->dbName, NULL);
   if (pTrans == NULL) return -1;
+  mndTransSetDbName(pTrans, pOutput->pSub->dbName, NULL);
 
   // make txn:
   // 1. redo action: action to all vg
@@ -523,28 +523,6 @@ static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOu
     tDeleteSMqConsumerObj(pConsumerNew);
     taosMemoryFree(pConsumerNew);
   }
-#if 0
-  if (consumerNum) {
-    char topic[TSDB_TOPIC_FNAME_LEN];
-    char cgroup[TSDB_CGROUP_LEN];
-    mndSplitSubscribeKey(pOutput->pSub->key, topic, cgroup, true);
-    SMqTopicObj *pTopic = mndAcquireTopic(pMnode, topic);
-    if (pTopic) {
-      // TODO make topic complete
-      SMqTopicObj topicObj = {0};
-      memcpy(&topicObj, pTopic, sizeof(SMqTopicObj));
-      topicObj.refConsumerCnt = pTopic->refConsumerCnt - consumerNum;
-      // TODO is that correct?
-      pTopic->refConsumerCnt = topicObj.refConsumerCnt;
-      mInfo("subscribe topic %s unref %d consumer cgroup %s, refcnt %d", pTopic->name, consumerNum, cgroup,
-            topicObj.refConsumerCnt);
-      if (mndSetTopicCommitLogs(pMnode, pTrans, &topicObj) != 0) {
-        ASSERT(0);
-        goto REB_FAIL;
-      }
-    }
-  }
-#endif
 
   // 4. TODO commit log: modification log
 
@@ -743,6 +721,8 @@ SUB_ENCODE_OVER:
 
 static SSdbRow *mndSubActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow         *pRow = NULL;
+  SMqSubscribeObj *pSub = NULL;
   void            *buf = NULL;
 
   int8_t sver = 0;
@@ -753,11 +733,10 @@ static SSdbRow *mndSubActionDecode(SSdbRaw *pRaw) {
     goto SUB_DECODE_OVER;
   }
 
-  int32_t  size = sizeof(SMqSubscribeObj);
+  pRow = sdbAllocRow(sizeof(SMqSubscribeObj));
-  SSdbRow *pRow = sdbAllocRow(size);
   if (pRow == NULL) goto SUB_DECODE_OVER;
 
-  SMqSubscribeObj *pSub = sdbGetRowObj(pRow);
+  pSub = sdbGetRowObj(pRow);
   if (pSub == NULL) goto SUB_DECODE_OVER;
 
   int32_t dataPos = 0;
@@ -777,7 +756,7 @@ static SSdbRow *mndSubActionDecode(SSdbRaw *pRaw) {
 SUB_DECODE_OVER:
   taosMemoryFreeClear(buf);
   if (terrno != TSDB_CODE_SUCCESS) {
-    mError("subscribe:%s, failed to decode from raw:%p since %s", pSub->key, pRaw, terrstr());
+    mError("subscribe:%s, failed to decode from raw:%p since %s", pSub == NULL ? "null" : pSub->key, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }

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

@@ -72,15 +72,11 @@ static int32_t mndSyncSendMsg(const SEpSet *pEpSet, SRpcMsg *pMsg) {
   return code;
 }
 
-void mndSyncCommitMsg(const SSyncFSM *pFsm, const SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
+void mndProcessWriteMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
   SMnode    *pMnode = pFsm->data;
   SSyncMgmt *pMgmt = &pMnode->syncMgmt;
   SSdbRaw   *pRaw = pMsg->pCont;
 
-  // delete msg handle
-  SRpcMsg rpcMsg = {0};
-  rpcMsg.info = pMsg->info;
-
   int32_t transId = sdbGetIdFromRaw(pMnode->pSdb, pRaw);
   pMgmt->errCode = pMeta->code;
   mInfo("trans:%d, is proposed, saved:%d code:0x%x, apply index:%" PRId64 " term:%" PRIu64 " config:%" PRId64
@@ -120,6 +116,12 @@ void mndSyncCommitMsg(const SSyncFSM *pFsm, const SRpcMsg *pMsg, const SFsmCbMet
   }
 }
 
+void mndSyncCommitMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
+  mndProcessWriteMsg(pFsm, pMsg, pMeta);
+  rpcFreeCont(pMsg->pCont);
+  pMsg->pCont = NULL;
+}
+
 int32_t mndSyncGetSnapshot(const SSyncFSM *pFsm, SSnapshot *pSnapshot, void *pReaderParam, void **ppReader) {
   mInfo("start to read snapshot from sdb in atomic way");
   SMnode *pMnode = pFsm->data;
@@ -361,7 +363,10 @@ int32_t mndSyncPropose(SMnode *pMnode, SSdbRaw *pRaw, int32_t transId) {
 
 void mndSyncStart(SMnode *pMnode) {
   SSyncMgmt *pMgmt = &pMnode->syncMgmt;
-  syncStart(pMgmt->sync);
+  if (syncStart(pMgmt->sync) < 0) {
+    mError("vgId:1, failed to start sync, id:%" PRId64, pMgmt->sync);
+    return;
+  }
   mInfo("vgId:1, sync started, id:%" PRId64, pMgmt->sync);
 }
 

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

@@ -152,8 +152,10 @@ TOPIC_ENCODE_OVER:
 
 SSdbRow *mndTopicActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
-
+  SSdbRow     *pRow = NULL;
+  SMqTopicObj *pTopic = NULL;
   void        *buf = NULL;
+
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto TOPIC_DECODE_OVER;
 
@@ -162,11 +164,10 @@ SSdbRow *mndTopicActionDecode(SSdbRaw *pRaw) {
     goto TOPIC_DECODE_OVER;
   }
 
-  int32_t  size = sizeof(SMqTopicObj);
+  pRow = sdbAllocRow(sizeof(SMqTopicObj));
-  SSdbRow *pRow = sdbAllocRow(size);
   if (pRow == NULL) goto TOPIC_DECODE_OVER;
 
-  SMqTopicObj *pTopic = sdbGetRowObj(pRow);
+  pTopic = sdbGetRowObj(pRow);
   if (pTopic == NULL) goto TOPIC_DECODE_OVER;
 
   int32_t len;
@@ -251,7 +252,7 @@ SSdbRow *mndTopicActionDecode(SSdbRaw *pRaw) {
 TOPIC_DECODE_OVER:
   taosMemoryFreeClear(buf);
   if (terrno != TSDB_CODE_SUCCESS) {
-    mError("topic:%s, failed to decode from raw:%p since %s", pTopic->name, pRaw, terrstr());
+    mError("topic:%s, failed to decode from raw:%p since %s", pTopic == NULL ? "null" : pTopic->name, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }
@@ -288,7 +289,7 @@ static int32_t mndTopicActionUpdate(SSdb *pSdb, SMqTopicObj *pOldTopic, SMqTopic
   return 0;
 }
 
-SMqTopicObj *mndAcquireTopic(SMnode *pMnode, char *topicName) {
+SMqTopicObj *mndAcquireTopic(SMnode *pMnode, const char *topicName) {
   SSdb        *pSdb = pMnode->pSdb;
   SMqTopicObj *pTopic = sdbAcquire(pSdb, SDB_TOPIC, topicName);
   if (pTopic == NULL && terrno == TSDB_CODE_SDB_OBJ_NOT_THERE) {
@@ -573,7 +574,7 @@ static int32_t mndProcessCreateTopicReq(SRpcMsg *pReq) {
     goto _OVER;
   }
 
-  if (mndCheckDbPrivilege(pMnode, pReq->info.conn.user, MND_OPER_READ_DB, pDb) != 0) {
+  if (mndCheckOperPrivilege(pMnode, pReq->info.conn.user, MND_OPER_CREATE_TOPIC) != 0) {
     goto _OVER;
   }
 
@@ -633,6 +634,11 @@ static int32_t mndProcessDropTopicReq(SRpcMsg *pReq) {
     }
   }
 
+  if (mndCheckOperPrivilege(pMnode, pReq->info.conn.user, MND_OPER_DROP_TOPIC) != 0) {
+    mndReleaseTopic(pMnode, pTopic);
+    return -1;
+  }
+
   void           *pIter = NULL;
   SMqConsumerObj *pConsumer;
   while (1) {

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

@@ -18,10 +18,11 @@
 #include "mndDb.h"
 #include "mndPrivilege.h"
 #include "mndShow.h"
+#include "mndTopic.h"
 #include "mndTrans.h"
 #include "tbase64.h"
 
-#define USER_VER_NUMBER   1
+#define USER_VER_NUMBER   2
 #define USER_RESERVE_SIZE 64
 
 static int32_t  mndCreateDefaultUsers(SMnode *pMnode);
@@ -36,6 +37,8 @@ static int32_t  mndProcessDropUserReq(SRpcMsg *pReq);
 static int32_t  mndProcessGetUserAuthReq(SRpcMsg *pReq);
 static int32_t  mndRetrieveUsers(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
 static void     mndCancelGetNextUser(SMnode *pMnode, void *pIter);
+static int32_t  mndRetrievePrivileges(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows);
+static void     mndCancelGetNextPrivileges(SMnode *pMnode, void *pIter);
 
 int32_t mndInitUser(SMnode *pMnode) {
   SSdbTable table = {
@@ -56,6 +59,8 @@ int32_t mndInitUser(SMnode *pMnode) {
 
   mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_USER, mndRetrieveUsers);
   mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_USER, mndCancelGetNextUser);
+  mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_PRIVILEGES, mndRetrievePrivileges);
+  mndAddShowFreeIterHandle(pMnode, TSDB_MGMT_TABLE_PRIVILEGES, mndCancelGetNextPrivileges);
   return sdbSetTable(pMnode->pSdb, table);
 }
 
@@ -119,7 +124,9 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
 
   int32_t numOfReadDbs = taosHashGetSize(pUser->readDbs);
   int32_t numOfWriteDbs = taosHashGetSize(pUser->writeDbs);
-  int32_t size = sizeof(SUserObj) + USER_RESERVE_SIZE + (numOfReadDbs + numOfWriteDbs) * TSDB_DB_FNAME_LEN;
+  int32_t numOfTopics = taosHashGetSize(pUser->topics);
+  int32_t size = sizeof(SUserObj) + USER_RESERVE_SIZE + (numOfReadDbs + numOfWriteDbs) * TSDB_DB_FNAME_LEN +
+                 numOfTopics * TSDB_TOPIC_FNAME_LEN;
 
   SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, USER_VER_NUMBER, size);
   if (pRaw == NULL) goto _OVER;
@@ -137,6 +144,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
   SDB_SET_INT32(pRaw, dataPos, pUser->authVersion, _OVER)
   SDB_SET_INT32(pRaw, dataPos, numOfReadDbs, _OVER)
   SDB_SET_INT32(pRaw, dataPos, numOfWriteDbs, _OVER)
+  SDB_SET_INT32(pRaw, dataPos, numOfTopics, _OVER)
 
   char *db = taosHashIterate(pUser->readDbs, NULL);
   while (db != NULL) {
@@ -150,6 +158,12 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
     db = taosHashIterate(pUser->writeDbs, db);
   }
 
+  char *topic = taosHashIterate(pUser->topics, NULL);
+  while (topic != NULL) {
+    SDB_SET_BINARY(pRaw, dataPos, topic, TSDB_TOPIC_FNAME_LEN, _OVER);
+    db = taosHashIterate(pUser->topics, topic);
+  }
+
   SDB_SET_RESERVE(pRaw, dataPos, USER_RESERVE_SIZE, _OVER)
   SDB_SET_DATALEN(pRaw, dataPos, _OVER)
 
@@ -168,19 +182,21 @@ _OVER:
 
 static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow  *pRow = NULL;
+  SUserObj *pUser = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
 
-  if (sver != USER_VER_NUMBER) {
+  if (sver != 1 && sver != 2) {
     terrno = TSDB_CODE_SDB_INVALID_DATA_VER;
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SUserObj));
+  pRow = sdbAllocRow(sizeof(SUserObj));
   if (pRow == NULL) goto _OVER;
 
-  SUserObj *pUser = sdbGetRowObj(pRow);
+  pUser = sdbGetRowObj(pRow);
   if (pUser == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -197,12 +213,18 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
 
   int32_t numOfReadDbs = 0;
   int32_t numOfWriteDbs = 0;
+  int32_t numOfTopics = 0;
   SDB_GET_INT32(pRaw, dataPos, &numOfReadDbs, _OVER)
   SDB_GET_INT32(pRaw, dataPos, &numOfWriteDbs, _OVER)
+  if (sver >= 2) {
+    SDB_GET_INT32(pRaw, dataPos, &numOfTopics, _OVER)
+  }
+
   pUser->readDbs = taosHashInit(numOfReadDbs, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
   pUser->writeDbs =
       taosHashInit(numOfWriteDbs, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
-  if (pUser->readDbs == NULL || pUser->writeDbs == NULL) goto _OVER;
+  pUser->topics = taosHashInit(numOfTopics, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), true, HASH_ENTRY_LOCK);
+  if (pUser->readDbs == NULL || pUser->writeDbs == NULL || pUser->topics == NULL) goto _OVER;
 
   for (int32_t i = 0; i < numOfReadDbs; ++i) {
     char db[TSDB_DB_FNAME_LEN] = {0};
@@ -218,6 +240,15 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
     taosHashPut(pUser->writeDbs, db, len, db, TSDB_DB_FNAME_LEN);
   }
 
+  if (sver >= 2) {
+    for (int32_t i = 0; i < numOfTopics; ++i) {
+      char topic[TSDB_TOPIC_FNAME_LEN] = {0};
+      SDB_GET_BINARY(pRaw, dataPos, topic, TSDB_TOPIC_FNAME_LEN, _OVER)
+      int32_t len = strlen(topic) + 1;
+      taosHashPut(pUser->topics, topic, len, topic, TSDB_TOPIC_FNAME_LEN);
+    }
+  }
+
   SDB_GET_RESERVE(pRaw, dataPos, USER_RESERVE_SIZE, _OVER)
   taosInitRWLatch(&pUser->lock);
 
@@ -225,9 +256,12 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("user:%s, failed to decode from raw:%p since %s", pUser->user, pRaw, terrstr());
+    mError("user:%s, failed to decode from raw:%p since %s", pUser == NULL ? "null" : pUser->user, pRaw, terrstr());
+    if (pUser != NULL) {
       taosHashCleanup(pUser->readDbs);
       taosHashCleanup(pUser->writeDbs);
+      taosHashCleanup(pUser->topics);
+    }
     taosMemoryFreeClear(pRow);
     return NULL;
   }
@@ -255,8 +289,10 @@ static int32_t mndUserActionDelete(SSdb *pSdb, SUserObj *pUser) {
   mTrace("user:%s, perform delete action, row:%p", pUser->user, pUser);
   taosHashCleanup(pUser->readDbs);
   taosHashCleanup(pUser->writeDbs);
+  taosHashCleanup(pUser->topics);
   pUser->readDbs = NULL;
   pUser->writeDbs = NULL;
+  pUser->topics = NULL;
   return 0;
 }
 
@@ -270,6 +306,7 @@ static int32_t mndUserActionUpdate(SSdb *pSdb, SUserObj *pOld, SUserObj *pNew) {
   memcpy(pOld->pass, pNew->pass, TSDB_PASSWORD_LEN);
   TSWAP(pOld->readDbs, pNew->readDbs);
   TSWAP(pOld->writeDbs, pNew->writeDbs);
+  TSWAP(pOld->topics, pNew->topics);
   taosWUnLockLatch(&pOld->lock);
 
   return 0;
@@ -482,9 +519,10 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
   taosRLockLatch(&pUser->lock);
   newUser.readDbs = mndDupDbHash(pUser->readDbs);
   newUser.writeDbs = mndDupDbHash(pUser->writeDbs);
+  newUser.topics = mndDupDbHash(pUser->topics);
   taosRUnLockLatch(&pUser->lock);
 
-  if (newUser.readDbs == NULL || newUser.writeDbs == NULL) {
+  if (newUser.readDbs == NULL || newUser.writeDbs == NULL || newUser.topics == NULL) {
     goto _OVER;
   }
 
@@ -507,14 +545,14 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
   }
 
   if (alterReq.alterType == TSDB_ALTER_USER_ADD_READ_DB || alterReq.alterType == TSDB_ALTER_USER_ADD_ALL_DB) {
-    if (strcmp(alterReq.dbname, "1.*") != 0) {
+    if (strcmp(alterReq.objname, "1.*") != 0) {
-      int32_t len = strlen(alterReq.dbname) + 1;
+      int32_t len = strlen(alterReq.objname) + 1;
-      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.dbname);
+      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.objname);
       if (pDb == NULL) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
-      if (taosHashPut(newUser.readDbs, alterReq.dbname, len, alterReq.dbname, TSDB_DB_FNAME_LEN) != 0) {
+      if (taosHashPut(newUser.readDbs, alterReq.objname, len, alterReq.objname, TSDB_DB_FNAME_LEN) != 0) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
@@ -531,14 +569,14 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
   }
 
   if (alterReq.alterType == TSDB_ALTER_USER_ADD_WRITE_DB || alterReq.alterType == TSDB_ALTER_USER_ADD_ALL_DB) {
-    if (strcmp(alterReq.dbname, "1.*") != 0) {
+    if (strcmp(alterReq.objname, "1.*") != 0) {
-      int32_t len = strlen(alterReq.dbname) + 1;
+      int32_t len = strlen(alterReq.objname) + 1;
-      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.dbname);
+      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.objname);
       if (pDb == NULL) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
-      if (taosHashPut(newUser.writeDbs, alterReq.dbname, len, alterReq.dbname, TSDB_DB_FNAME_LEN) != 0) {
+      if (taosHashPut(newUser.writeDbs, alterReq.objname, len, alterReq.objname, TSDB_DB_FNAME_LEN) != 0) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
@@ -555,33 +593,53 @@ static int32_t mndProcessAlterUserReq(SRpcMsg *pReq) {
   }
 
   if (alterReq.alterType == TSDB_ALTER_USER_REMOVE_READ_DB || alterReq.alterType == TSDB_ALTER_USER_REMOVE_ALL_DB) {
-    if (strcmp(alterReq.dbname, "1.*") != 0) {
+    if (strcmp(alterReq.objname, "1.*") != 0) {
-      int32_t len = strlen(alterReq.dbname) + 1;
+      int32_t len = strlen(alterReq.objname) + 1;
-      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.dbname);
+      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.objname);
       if (pDb == NULL) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
-      taosHashRemove(newUser.readDbs, alterReq.dbname, len);
+      taosHashRemove(newUser.readDbs, alterReq.objname, len);
     } else {
       taosHashClear(newUser.readDbs);
     }
   }
 
   if (alterReq.alterType == TSDB_ALTER_USER_REMOVE_WRITE_DB || alterReq.alterType == TSDB_ALTER_USER_REMOVE_ALL_DB) {
-    if (strcmp(alterReq.dbname, "1.*") != 0) {
+    if (strcmp(alterReq.objname, "1.*") != 0) {
-      int32_t len = strlen(alterReq.dbname) + 1;
+      int32_t len = strlen(alterReq.objname) + 1;
-      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.dbname);
+      SDbObj *pDb = mndAcquireDb(pMnode, alterReq.objname);
       if (pDb == NULL) {
         mndReleaseDb(pMnode, pDb);
         goto _OVER;
       }
-      taosHashRemove(newUser.writeDbs, alterReq.dbname, len);
+      taosHashRemove(newUser.writeDbs, alterReq.objname, len);
     } else {
       taosHashClear(newUser.writeDbs);
     }
   }
 
+  if (alterReq.alterType == TSDB_ALTER_USER_ADD_SUBSCRIBE_TOPIC) {
+    int32_t      len = strlen(alterReq.objname) + 1;
+    SMqTopicObj *pTopic = mndAcquireTopic(pMnode, alterReq.objname);
+    if (pTopic == NULL) {
+      mndReleaseTopic(pMnode, pTopic);
+      goto _OVER;
+    }
+    taosHashPut(newUser.topics, pTopic->name, len, pTopic->name, TSDB_TOPIC_FNAME_LEN);
+  }
+
+  if (alterReq.alterType == TSDB_ALTER_USER_REMOVE_SUBSCRIBE_TOPIC) {
+    int32_t      len = strlen(alterReq.objname) + 1;
+    SMqTopicObj *pTopic = mndAcquireTopic(pMnode, alterReq.objname);
+    if (pTopic == NULL) {
+      mndReleaseTopic(pMnode, pTopic);
+      goto _OVER;
+    }
+    taosHashRemove(newUser.topics, alterReq.objname, len);
+  }
+
   code = mndAlterUser(pMnode, pUser, &newUser, pReq);
   if (code == 0) code = TSDB_CODE_ACTION_IN_PROGRESS;
 
@@ -594,6 +652,7 @@ _OVER:
   mndReleaseUser(pMnode, pUser);
   taosHashCleanup(newUser.writeDbs);
   taosHashCleanup(newUser.readDbs);
+  taosHashCleanup(newUser.topics);
 
   return code;
 }
@@ -756,6 +815,108 @@ static void mndCancelGetNextUser(SMnode *pMnode, void *pIter) {
   sdbCancelFetch(pSdb, pIter);
 }
 
+static int32_t mndRetrievePrivileges(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBlock, int32_t rows) {
+  SMnode   *pMnode = pReq->info.node;
+  SSdb     *pSdb = pMnode->pSdb;
+  int32_t   numOfRows = 0;
+  SUserObj *pUser = NULL;
+  int32_t   cols = 0;
+  char     *pWrite;
+
+  while (numOfRows < rows) {
+    pShow->pIter = sdbFetch(pSdb, SDB_USER, pShow->pIter, (void **)&pUser);
+    if (pShow->pIter == NULL) break;
+
+    int32_t numOfReadDbs = taosHashGetSize(pUser->readDbs);
+    int32_t numOfWriteDbs = taosHashGetSize(pUser->writeDbs);
+    int32_t numOfTopics = taosHashGetSize(pUser->topics);
+    if (numOfRows + numOfReadDbs + numOfWriteDbs + numOfTopics >= rows) break;
+
+    char *db = taosHashIterate(pUser->readDbs, NULL);
+    while (db != NULL) {
+      cols = 0;
+      SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      char             userName[TSDB_USER_LEN + VARSTR_HEADER_SIZE] = {0};
+      STR_WITH_MAXSIZE_TO_VARSTR(userName, pUser->user, pShow->pMeta->pSchemas[cols].bytes);
+      colDataAppend(pColInfo, numOfRows, (const char *)userName, false);
+
+      char privilege[20] = {0};
+      STR_WITH_MAXSIZE_TO_VARSTR(privilege, "read", pShow->pMeta->pSchemas[cols].bytes);
+      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      colDataAppend(pColInfo, numOfRows, (const char *)privilege, false);
+
+      SName name = {0};
+      char  objName[TSDB_DB_NAME_LEN + VARSTR_HEADER_SIZE] = {0};
+      tNameFromString(&name, db, T_NAME_ACCT | T_NAME_DB);
+      tNameGetDbName(&name, varDataVal(objName));
+      varDataSetLen(objName, strlen(varDataVal(objName)));
+      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      colDataAppend(pColInfo, numOfRows, (const char *)objName, false);
+
+      numOfRows++;
+      db = taosHashIterate(pUser->readDbs, db);
+    }
+
+    db = taosHashIterate(pUser->writeDbs, NULL);
+    while (db != NULL) {
+      cols = 0;
+      SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      char             userName[TSDB_USER_LEN + VARSTR_HEADER_SIZE] = {0};
+      STR_WITH_MAXSIZE_TO_VARSTR(userName, pUser->user, pShow->pMeta->pSchemas[cols].bytes);
+      colDataAppend(pColInfo, numOfRows, (const char *)userName, false);
+
+      char privilege[20] = {0};
+      STR_WITH_MAXSIZE_TO_VARSTR(privilege, "write", pShow->pMeta->pSchemas[cols].bytes);
+      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      colDataAppend(pColInfo, numOfRows, (const char *)privilege, false);
+
+      SName name = {0};
+      char  objName[TSDB_DB_NAME_LEN + VARSTR_HEADER_SIZE] = {0};
+      tNameFromString(&name, db, T_NAME_ACCT | T_NAME_DB);
+      tNameGetDbName(&name, varDataVal(objName));
+      varDataSetLen(objName, strlen(varDataVal(objName)));
+      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      colDataAppend(pColInfo, numOfRows, (const char *)objName, false);
+
+      numOfRows++;
+      db = taosHashIterate(pUser->writeDbs, db);
+    }
+
+    char *topic = taosHashIterate(pUser->topics, NULL);
+    while (topic != NULL) {
+      cols = 0;
+      SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      char             userName[TSDB_USER_LEN + VARSTR_HEADER_SIZE] = {0};
+      STR_WITH_MAXSIZE_TO_VARSTR(userName, pUser->user, pShow->pMeta->pSchemas[cols].bytes);
+      colDataAppend(pColInfo, numOfRows, (const char *)userName, false);
+
+      char privilege[20] = {0};
+      STR_WITH_MAXSIZE_TO_VARSTR(privilege, "subscribe", pShow->pMeta->pSchemas[cols].bytes);
+      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      colDataAppend(pColInfo, numOfRows, (const char *)privilege, false);
+
+      char topicName[TSDB_TOPIC_NAME_LEN + VARSTR_HEADER_SIZE + 5] = {0};
+      tstrncpy(varDataVal(topicName), mndGetDbStr(topic), TSDB_TOPIC_NAME_LEN - 2);
+      varDataSetLen(topicName, strlen(varDataVal(topicName)));
+      pColInfo = taosArrayGet(pBlock->pDataBlock, cols++);
+      colDataAppend(pColInfo, numOfRows, (const char *)topicName, false);
+
+      numOfRows++;
+      db = taosHashIterate(pUser->topics, topic);
+    }
+
+    sdbRelease(pSdb, pUser);
+  }
+
+  pShow->numOfRows += numOfRows;
+  return numOfRows;
+}
+
+static void mndCancelGetNextPrivileges(SMnode *pMnode, void *pIter) {
+  SSdb *pSdb = pMnode->pSdb;
+  sdbCancelFetch(pSdb, pIter);
+}
+
 int32_t mndValidateUserAuthInfo(SMnode *pMnode, SUserAuthVersion *pUsers, int32_t numOfUses, void **ppRsp,
                                 int32_t *pRspLen) {
   SUserAuthBatchRsp batchRsp = {0};

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

@@ -113,6 +113,8 @@ _OVER:
 
 SSdbRow *mndVgroupActionDecode(SSdbRaw *pRaw) {
   terrno = TSDB_CODE_OUT_OF_MEMORY;
+  SSdbRow *pRow = NULL;
+  SVgObj  *pVgroup = NULL;
 
   int8_t sver = 0;
   if (sdbGetRawSoftVer(pRaw, &sver) != 0) goto _OVER;
@@ -122,10 +124,10 @@ SSdbRow *mndVgroupActionDecode(SSdbRaw *pRaw) {
     goto _OVER;
   }
 
-  SSdbRow *pRow = sdbAllocRow(sizeof(SVgObj));
+  pRow = sdbAllocRow(sizeof(SVgObj));
   if (pRow == NULL) goto _OVER;
 
-  SVgObj *pVgroup = sdbGetRowObj(pRow);
+  pVgroup = sdbGetRowObj(pRow);
   if (pVgroup == NULL) goto _OVER;
 
   int32_t dataPos = 0;
@@ -152,7 +154,7 @@ SSdbRow *mndVgroupActionDecode(SSdbRaw *pRaw) {
 
 _OVER:
   if (terrno != 0) {
-    mError("vgId:%d, failed to decode from raw:%p since %s", pVgroup->vgId, pRaw, terrstr());
+    mError("vgId:%d, failed to decode from raw:%p since %s", pVgroup == NULL ? 0 : pVgroup->vgId, pRaw, terrstr());
     taosMemoryFreeClear(pRow);
     return NULL;
   }
@@ -186,6 +188,7 @@ static int32_t mndVgroupActionUpdate(SSdb *pSdb, SVgObj *pOld, SVgObj *pNew) {
       if (pNewGid->dnodeId == pOldGid->dnodeId) {
         pNewGid->syncState = pOldGid->syncState;
         pNewGid->syncRestore = pOldGid->syncRestore;
+        pNewGid->syncCanRead = pOldGid->syncCanRead;
       }
     }
   }
@@ -696,8 +699,16 @@ static int32_t mndRetrieveVgroups(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *p
         if (!exist) {
           strcpy(role, "dropping");
         } else if (online) {
-          bool show = (pVgroup->vnodeGid[i].syncState == TAOS_SYNC_STATE_LEADER && !pVgroup->vnodeGid[i].syncRestore);
+          char *star = "";
-          snprintf(role, sizeof(role), "%s%s", syncStr(pVgroup->vnodeGid[i].syncState), show ? "*" : "");
+          if (pVgroup->vnodeGid[i].syncState == TAOS_SYNC_STATE_LEADER) {
+            if (!pVgroup->vnodeGid[i].syncRestore && !pVgroup->vnodeGid[i].syncCanRead) {
+              star = "**";
+            } else if (!pVgroup->vnodeGid[i].syncRestore && pVgroup->vnodeGid[i].syncCanRead) {
+              star = "*";
+            } else {
+            }
+          }
+          snprintf(role, sizeof(role), "%s%s", syncStr(pVgroup->vnodeGid[i].syncState), star);
         } else {
         }
         STR_WITH_MAXSIZE_TO_VARSTR(buf1, role, pShow->pMeta->pSchemas[cols].bytes);

source/dnode/snode/src/snode.c

@@ -168,7 +168,7 @@ int32_t sndProcessTaskDeployReq(SSnode *pSnode, char *msg, int32_t msgLen) {
 
 int32_t sndProcessTaskDropReq(SSnode *pSnode, char *msg, int32_t msgLen) {
   SVDropStreamTaskReq *pReq = (SVDropStreamTaskReq *)msg;
-  streamMetaRemoveTask1(pSnode->pMeta, pReq->taskId);
+  streamMetaRemoveTask(pSnode->pMeta, pReq->taskId);
   return 0;
 }
 

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

@@ -97,7 +97,7 @@ bool    vnodeShouldRollback(SVnode* pVnode);
 
 // vnodeSync.c
 int32_t vnodeSyncOpen(SVnode* pVnode, char* path);
-void    vnodeSyncStart(SVnode* pVnode);
+int32_t vnodeSyncStart(SVnode* pVnode);
 void    vnodeSyncPreClose(SVnode* pVnode);
 void    vnodeSyncClose(SVnode* pVnode);
 void    vnodeRedirectRpcMsg(SVnode* pVnode, SRpcMsg* pMsg);

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

@@ -1425,7 +1425,7 @@ int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg) {
 
 int32_t tqProcessTaskDropReq(STQ* pTq, int64_t version, char* msg, int32_t msgLen) {
   SVDropStreamTaskReq* pReq = (SVDropStreamTaskReq*)msg;
-  streamMetaRemoveTask1(pTq->pStreamMeta, pReq->taskId);
+  streamMetaRemoveTask(pTq->pStreamMeta, pReq->taskId);
   return 0;
 }
 

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

@@ -811,7 +811,7 @@ static int32_t tsdbStartCommit(STsdb *pTsdb, SCommitter *pCommitter) {
   int32_t lino = 0;
 
   memset(pCommitter, 0, sizeof(*pCommitter));
-  ASSERT(pTsdb->mem && pTsdb->imem == NULL);
+  ASSERT(pTsdb->mem && pTsdb->imem == NULL && "last tsdb commit incomplete");
 
   taosThreadRwlockWrlock(&pTsdb->rwLock);
   pTsdb->imem = pTsdb->mem;
@@ -1261,7 +1261,7 @@ static int32_t tsdbCommitMergeBlock(SCommitter *pCommitter, SDataBlk *pDataBlk)
         }
       }
     } else {
-      ASSERT(0);
+      ASSERT(0 && "dup rows not allowed");
     }
 
     if (pBDataW->nRow >= pCommitter->maxRow) {

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

@@ -728,7 +728,7 @@ int32_t tRowMergerAdd(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) {
         taosArraySet(pMerger->pArray, iCol, pColVal);
       }
     } else {
-      ASSERT(0);
+      ASSERT(0 && "dup versions not allowed");
     }
   }
 
@@ -902,7 +902,6 @@ int32_t tsdbBuildDeleteSkyline(SArray *aDelData, int32_t sidx, int32_t eidx, SAr
       code = TSDB_CODE_OUT_OF_MEMORY;
       goto _clear;
     }
-
     midx = (sidx + eidx) / 2;
 
     code = tsdbBuildDeleteSkyline(aDelData, sidx, midx, aSkyline1);

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

@@ -176,6 +176,9 @@ void vnodeBufPoolRef(SVBufPool *pPool) {
 }
 
 void vnodeBufPoolUnRef(SVBufPool *pPool) {
+  if (pPool == NULL) {
+    return;
+  }
   int32_t nRef = atomic_sub_fetch_32(&pPool->nRef, 1);
   if (nRef == 0) {
     SVnode *pVnode = pPool->pVnode;

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

@@ -209,8 +209,8 @@ int vnodeCommit(SVnode *pVnode) {
   SVnodeInfo info = {0};
   char       dir[TSDB_FILENAME_LEN];
 
-  vInfo("vgId:%d, start to commit, commit ID:%" PRId64 " version:%" PRId64, TD_VID(pVnode), pVnode->state.commitID,
+  vInfo("vgId:%d, start to commit, commit ID:%" PRId64 " version:%" PRId64 " term: %" PRId64, TD_VID(pVnode),
-        pVnode->state.applied);
+        pVnode->state.commitID, pVnode->state.applied, pVnode->state.applyTerm);
 
   // persist wal before starting
   if (walPersist(pVnode->pWal) < 0) {

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

@@ -144,9 +144,9 @@ SVnode *vnodeOpen(const char *path, STfs *pTfs, SMsgCb msgCb) {
   pVnode->config = info.config;
   pVnode->state.committed = info.state.committed;
   pVnode->state.commitTerm = info.state.commitTerm;
-  pVnode->state.applied = info.state.committed;
   pVnode->state.commitID = info.state.commitID;
-  pVnode->state.commitTerm = info.state.commitTerm;
+  pVnode->state.applied = info.state.committed;
+  pVnode->state.applyTerm = info.state.commitTerm;
   pVnode->pTfs = pTfs;
   pVnode->msgCb = msgCb;
   taosThreadMutexInit(&pVnode->lock, NULL);
@@ -269,10 +269,7 @@ void vnodeClose(SVnode *pVnode) {
 }
 
 // start the sync timer after the queue is ready
-int32_t vnodeStart(SVnode *pVnode) {
+int32_t vnodeStart(SVnode *pVnode) { return vnodeSyncStart(pVnode); }
-  vnodeSyncStart(pVnode);
-  return 0;
-}
 
 void vnodeStop(SVnode *pVnode) {}
 

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

@@ -380,6 +380,7 @@ int32_t vnodeGetLoad(SVnode *pVnode, SVnodeLoad *pLoad) {
   pLoad->vgId = TD_VID(pVnode);
   pLoad->syncState = state.state;
   pLoad->syncRestore = state.restored;
+  pLoad->syncCanRead = state.canRead;
   pLoad->cacheUsage = tsdbCacheGetUsage(pVnode);
   pLoad->numOfTables = metaGetTbNum(pVnode->pMeta);
   pLoad->numOfTimeSeries = metaGetTimeSeriesNum(pVnode->pMeta);

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

@@ -178,8 +178,16 @@ int32_t vnodeProcessWriteMsg(SVnode *pVnode, SRpcMsg *pMsg, int64_t version, SRp
     return -1;
   }
 
+  if (version <= pVnode->state.applied) {
+    vError("vgId:%d, duplicate write request. version: %" PRId64 ", applied: %" PRId64 "", TD_VID(pVnode), version,
+           pVnode->state.applied);
+    pRsp->info.handle = NULL;
+    return -1;
+  }
+
   vDebug("vgId:%d, start to process write request %s, index:%" PRId64, TD_VID(pVnode), TMSG_INFO(pMsg->msgType),
          version);
+  ASSERT(pVnode->state.applyTerm <= pMsg->info.conn.applyTerm);
 
   pVnode->state.applied = version;
   pVnode->state.applyTerm = pMsg->info.conn.applyTerm;

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

@@ -295,36 +295,31 @@ static int32_t vnodeSyncGetSnapshot(const SSyncFSM *pFsm, SSnapshot *pSnapshot)
   return 0;
 }
 
-static void vnodeSyncApplyMsg(const SSyncFSM *pFsm, const SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
+static void vnodeSyncApplyMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
   SVnode *pVnode = pFsm->data;
-
+  pMsg->info.conn.applyIndex = pMeta->index;
-  SRpcMsg rpcMsg = {.msgType = pMsg->msgType, .contLen = pMsg->contLen};
+  pMsg->info.conn.applyTerm = pMeta->term;
-  rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen);
-  memcpy(rpcMsg.pCont, pMsg->pCont, pMsg->contLen);
-  rpcMsg.info = pMsg->info;
-  rpcMsg.info.conn.applyIndex = pMeta->index;
-  rpcMsg.info.conn.applyTerm = pMeta->term;
 
   const STraceId *trace = &pMsg->info.traceId;
   vGTrace("vgId:%d, commit-cb is excuted, fsm:%p, index:%" PRId64 ", term:%" PRIu64 ", msg-index:%" PRId64
           ", weak:%d, code:%d, state:%d %s, type:%s",
-          pVnode->config.vgId, pFsm, pMeta->index, pMeta->term, rpcMsg.info.conn.applyIndex, pMeta->isWeak, pMeta->code,
+          pVnode->config.vgId, pFsm, pMeta->index, pMeta->term, pMsg->info.conn.applyIndex, pMeta->isWeak, pMeta->code,
           pMeta->state, syncStr(pMeta->state), TMSG_INFO(pMsg->msgType));
 
-  tmsgPutToQueue(&pVnode->msgCb, APPLY_QUEUE, &rpcMsg);
+  tmsgPutToQueue(&pVnode->msgCb, APPLY_QUEUE, pMsg);
 }
 
-static void vnodeSyncCommitMsg(const SSyncFSM *pFsm, const SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
+static void vnodeSyncCommitMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
   vnodeSyncApplyMsg(pFsm, pMsg, pMeta);
 }
 
-static void vnodeSyncPreCommitMsg(const SSyncFSM *pFsm, const SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
+static void vnodeSyncPreCommitMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
   if (pMeta->isWeak == 1) {
     vnodeSyncApplyMsg(pFsm, pMsg, pMeta);
   }
 }
 
-static void vnodeSyncRollBackMsg(const SSyncFSM *pFsm, const SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
+static void vnodeSyncRollBackMsg(const SSyncFSM *pFsm, SRpcMsg *pMsg, const SFsmCbMeta *pMeta) {
   SVnode *pVnode = pFsm->data;
   vTrace("vgId:%d, rollback-cb is excuted, fsm:%p, index:%" PRId64 ", weak:%d, code:%d, state:%d %s, type:%s",
          pVnode->config.vgId, pFsm, pMeta->index, pMeta->isWeak, pMeta->code, pMeta->state, syncStr(pMeta->state),
@@ -404,7 +399,7 @@ static void vnodeRestoreFinish(const SSyncFSM *pFsm) {
   walApplyVer(pVnode->pWal, pVnode->state.applied);
 
   pVnode->restored = true;
-  vDebug("vgId:%d, sync restore finished", pVnode->config.vgId);
+  vInfo("vgId:%d, sync restore finished", pVnode->config.vgId);
 }
 
 static void vnodeBecomeFollower(const SSyncFSM *pFsm) {
@@ -506,16 +501,20 @@ int32_t vnodeSyncOpen(SVnode *pVnode, char *path) {
   return 0;
 }
 
-void vnodeSyncStart(SVnode *pVnode) {
+int32_t vnodeSyncStart(SVnode *pVnode) {
   vInfo("vgId:%d, start sync", pVnode->config.vgId);
-  syncStart(pVnode->sync);
+  if (syncStart(pVnode->sync) < 0) {
+    vError("vgId:%d, failed to start sync subsystem since %s", pVnode->config.vgId, terrstr());
+    return -1;
+  }
+  return 0;
 }
 
 void vnodeSyncPreClose(SVnode *pVnode) {
   vInfo("vgId:%d, pre close sync", pVnode->config.vgId);
   syncLeaderTransfer(pVnode->sync);
   syncPreStop(pVnode->sync);
-
+#if 0
   while (syncSnapshotRecving(pVnode->sync)) {
     vInfo("vgId:%d, snapshot is recving", pVnode->config.vgId);
     taosMsleep(300);
@@ -524,7 +523,7 @@ void vnodeSyncPreClose(SVnode *pVnode) {
     vInfo("vgId:%d, snapshot is sending", pVnode->config.vgId);
     taosMsleep(300);
   }
-
+#endif
   taosThreadMutexLock(&pVnode->lock);
   if (pVnode->blocked) {
     vInfo("vgId:%d, post block after close sync", pVnode->config.vgId);
@@ -553,12 +552,12 @@ bool vnodeIsLeader(SVnode *pVnode) {
     } else {
       terrno = TSDB_CODE_APP_NOT_READY;
     }
-    vDebug("vgId:%d, vnode not ready, state:%s, restore:%d", pVnode->config.vgId, syncStr(state.state), state.restored);
+    vInfo("vgId:%d, vnode not ready, state:%s, restore:%d", pVnode->config.vgId, syncStr(state.state), state.restored);
     return false;
   }
 
   if (!pVnode->restored) {
-    vDebug("vgId:%d, vnode not restored", pVnode->config.vgId);
+    vInfo("vgId:%d, vnode not restored", pVnode->config.vgId);
     terrno = TSDB_CODE_APP_NOT_READY;
     return false;
   }

source/libs/executor/inc/executorimpl.h

@@ -536,6 +536,7 @@ typedef struct SStreamIntervalOperatorInfo {
   SArray*            pChildren;
   SStreamState*      pState;
   SWinKey            delKey;
+  uint64_t           numOfDatapack;
 } SStreamIntervalOperatorInfo;
 
 typedef struct SDataGroupInfo {

source/libs/executor/src/filloperator.c

@@ -1146,7 +1146,7 @@ static void doDeleteFillResult(SOperatorInfo* pOperator) {
       if (delTs > nextKey.ts) {
         break;
       }
-      endTs = delTs;
+
       SWinKey delKey = {.groupId = delGroupId, .ts = delTs};
       if (delTs == nextKey.ts) {
         code = streamStateCurNext(pOperator->pTaskInfo->streamInfo.pState, pCur);
@@ -1159,7 +1159,7 @@ static void doDeleteFillResult(SOperatorInfo* pOperator) {
           streamStateFreeCur(pCur);
           pCur = streamStateGetAndCheckCur(pOperator->pTaskInfo->streamInfo.pState, &nextKey);
         }
-        endTs = TMAX(ts, nextKey.ts - 1);
+        endTs = TMAX(delTs, nextKey.ts - 1);
         if (code != TSDB_CODE_SUCCESS) {
           break;
         }

source/libs/executor/src/scanoperator.c

@@ -1113,7 +1113,7 @@ static STimeWindow getSlidingWindow(TSKEY* startTsCol, TSKEY* endTsCol, uint64_t
     if (hasGroup) {
       (*pRowIndex) += 1;
     } else {
-      while ((groupId == gpIdCol[(*pRowIndex)] && startTsCol[*pRowIndex] < endWin.ekey)) {
+      while ((groupId == gpIdCol[(*pRowIndex)] && startTsCol[*pRowIndex] <= endWin.ekey)) {
         (*pRowIndex) += 1;
         if ((*pRowIndex) == pDataBlockInfo->rows) {
           break;

source/libs/executor/src/timewindowoperator.c

@@ -2516,9 +2516,11 @@ static SSDataBlock* doStreamFinalIntervalAgg(SOperatorInfo* pOperator) {
     SSDataBlock* pBlock = downstream->fpSet.getNextFn(downstream);
     if (pBlock == NULL) {
       pOperator->status = OP_RES_TO_RETURN;
-      qDebug("%s return data", IS_FINAL_OP(pInfo) ? "interval final" : "interval semi");
+      qDebug("===stream===return data:%s. recv datablock num:%" PRIu64 , IS_FINAL_OP(pInfo) ? "interval final" : "interval semi", pInfo->numOfDatapack);
+      pInfo->numOfDatapack = 0;
       break;
     }
+    pInfo->numOfDatapack++;
     printDataBlock(pBlock, IS_FINAL_OP(pInfo) ? "interval final recv" : "interval semi recv");
 
     ASSERT(pBlock->info.type != STREAM_INVERT);
@@ -2734,6 +2736,7 @@ SOperatorInfo* createStreamFinalIntervalOperatorInfo(SOperatorInfo* downstream,
   pInfo->pDelWins = taosArrayInit(4, sizeof(SWinKey));
   pInfo->delKey.ts = INT64_MAX;
   pInfo->delKey.groupId = 0;
+  pInfo->numOfDatapack = 0;
 
   pOperator->operatorType = pPhyNode->type;
   pOperator->blocking = true;
@@ -4700,8 +4703,11 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
   while (1) {
     SSDataBlock* pBlock = downstream->fpSet.getNextFn(downstream);
     if (pBlock == NULL) {
+      qDebug("===stream===return data:single interval. recv datablock num:%" PRIu64, pInfo->numOfDatapack);
+      pInfo->numOfDatapack = 0;
       break;
     }
+    pInfo->numOfDatapack++;
     printDataBlock(pBlock, "single interval recv");
 
     if (pBlock->info.type == STREAM_DELETE_DATA || pBlock->info.type == STREAM_DELETE_RESULT ||
@@ -4851,6 +4857,7 @@ SOperatorInfo* createStreamIntervalOperatorInfo(SOperatorInfo* downstream, SPhys
   pInfo->pChildren = NULL;
   pInfo->delKey.ts = INT64_MAX;
   pInfo->delKey.groupId = 0;
+  pInfo->numOfDatapack = 0;
 
   setOperatorInfo(pOperator, "StreamIntervalOperator", QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL, true, OP_NOT_OPENED,
                   pInfo, pTaskInfo);

source/libs/nodes/src/nodesUtilFuncs.c

@@ -424,6 +424,7 @@ SNode* nodesMakeNode(ENodeType type) {
     case QUERY_NODE_SHOW_TRANSACTIONS_STMT:
     case QUERY_NODE_SHOW_SUBSCRIPTIONS_STMT:
     case QUERY_NODE_SHOW_TAGS_STMT:
+    case QUERY_NODE_SHOW_USER_PRIVILEGES_STMT:
       return makeNode(type, sizeof(SShowStmt));
     case QUERY_NODE_SHOW_TABLE_TAGS_STMT:
       return makeNode(type, sizeof(SShowTableTagsStmt));
@@ -943,7 +944,8 @@ void nodesDestroyNode(SNode* pNode) {
     case QUERY_NODE_SHOW_LOCAL_VARIABLES_STMT:
     case QUERY_NODE_SHOW_TRANSACTIONS_STMT:
     case QUERY_NODE_SHOW_SUBSCRIPTIONS_STMT:
-    case QUERY_NODE_SHOW_TAGS_STMT: {
+    case QUERY_NODE_SHOW_TAGS_STMT:
+    case QUERY_NODE_SHOW_USER_PRIVILEGES_STMT: {
       SShowStmt* pStmt = (SShowStmt*)pNode;
       nodesDestroyNode(pStmt->pDbName);
       nodesDestroyNode(pStmt->pTbName);

source/libs/parser/inc/sql.y

@@ -101,6 +101,7 @@ cmd ::= REVOKE privileges(A) ON priv_level(B) FROM user_name(C).
 %destructor privileges                                                            { }
 privileges(A) ::= ALL.                                                            { A = PRIVILEGE_TYPE_ALL; }
 privileges(A) ::= priv_type_list(B).                                              { A = B; }
+privileges(A) ::= SUBSCRIBE.                                                      { A = PRIVILEGE_TYPE_SUBSCRIBE; }
 
 %type priv_type_list                                                              { int64_t }
 %destructor priv_type_list                                                        { }
@@ -116,6 +117,7 @@ priv_type(A) ::= WRITE.
 %destructor priv_level                                                            { }
 priv_level(A) ::= NK_STAR(B) NK_DOT NK_STAR.                                      { A = B; }
 priv_level(A) ::= db_name(B) NK_DOT NK_STAR.                                      { A = B; }
+priv_level(A) ::= topic_name(B).                                                  { A = B; }
 
 /************************************************ create/drop/alter dnode *********************************************/
 cmd ::= CREATE DNODE dnode_endpoint(A).                                           { pCxt->pRootNode = createCreateDnodeStmt(pCxt, &A, NULL); }
@@ -393,6 +395,7 @@ col_name(A) ::= column_name(B).
 /************************************************ show ****************************************************************/
 cmd ::= SHOW DNODES.                                                              { pCxt->pRootNode = createShowStmt(pCxt, QUERY_NODE_SHOW_DNODES_STMT); }
 cmd ::= SHOW USERS.                                                               { pCxt->pRootNode = createShowStmt(pCxt, QUERY_NODE_SHOW_USERS_STMT); }
+cmd ::= SHOW USER PRIVILEGES.                                                     { pCxt->pRootNode = createShowStmt(pCxt, QUERY_NODE_SHOW_USER_PRIVILEGES_STMT); }
 cmd ::= SHOW DATABASES.                                                           { pCxt->pRootNode = createShowStmt(pCxt, QUERY_NODE_SHOW_DATABASES_STMT); }
 cmd ::= SHOW db_name_cond_opt(A) TABLES like_pattern_opt(B).                      { pCxt->pRootNode = createShowStmtWithCond(pCxt, QUERY_NODE_SHOW_TABLES_STMT, A, B, OP_TYPE_LIKE); }
 cmd ::= SHOW db_name_cond_opt(A) STABLES like_pattern_opt(B).                     { pCxt->pRootNode = createShowStmtWithCond(pCxt, QUERY_NODE_SHOW_STABLES_STMT, A, B, OP_TYPE_LIKE); }

source/libs/parser/src/parAstCreater.c

@@ -1815,7 +1815,7 @@ SNode* createGrantStmt(SAstCreateContext* pCxt, int64_t privileges, SToken* pDbN
   SGrantStmt* pStmt = (SGrantStmt*)nodesMakeNode(QUERY_NODE_GRANT_STMT);
   CHECK_OUT_OF_MEM(pStmt);
   pStmt->privileges = privileges;
-  COPY_STRING_FORM_ID_TOKEN(pStmt->dbName, pDbName);
+  COPY_STRING_FORM_ID_TOKEN(pStmt->objName, pDbName);
   COPY_STRING_FORM_ID_TOKEN(pStmt->userName, pUserName);
   return (SNode*)pStmt;
 }
@@ -1828,7 +1828,7 @@ SNode* createRevokeStmt(SAstCreateContext* pCxt, int64_t privileges, SToken* pDb
   SRevokeStmt* pStmt = (SRevokeStmt*)nodesMakeNode(QUERY_NODE_REVOKE_STMT);
   CHECK_OUT_OF_MEM(pStmt);
   pStmt->privileges = privileges;
-  COPY_STRING_FORM_ID_TOKEN(pStmt->dbName, pDbName);
+  COPY_STRING_FORM_ID_TOKEN(pStmt->objName, pDbName);
   COPY_STRING_FORM_ID_TOKEN(pStmt->userName, pUserName);
   return (SNode*)pStmt;
 }

source/libs/parser/src/parAstParser.c

@@ -504,6 +504,11 @@ static int32_t collectMetaKeyFromShowVnodes(SCollectMetaKeyCxt* pCxt, SShowVnode
                                  pCxt->pMetaCache);
 }
 
+static int32_t collectMetaKeyFromShowUserPrivileges(SCollectMetaKeyCxt* pCxt, SShowStmt* pStmt) {
+  return reserveTableMetaInCache(pCxt->pParseCxt->acctId, TSDB_INFORMATION_SCHEMA_DB, TSDB_INS_TABLE_USER_PRIVILEGES,
+                                 pCxt->pMetaCache);
+}
+
 static int32_t collectMetaKeyFromShowCreateDatabase(SCollectMetaKeyCxt* pCxt, SShowCreateDatabaseStmt* pStmt) {
   return reserveDbCfgInCache(pCxt->pParseCxt->acctId, pStmt->dbName, pCxt->pMetaCache);
 }
@@ -648,6 +653,8 @@ static int32_t collectMetaKeyFromQuery(SCollectMetaKeyCxt* pCxt, SNode* pStmt) {
       return collectMetaKeyFromShowDnodeVariables(pCxt, (SShowDnodeVariablesStmt*)pStmt);
     case QUERY_NODE_SHOW_VNODES_STMT:
       return collectMetaKeyFromShowVnodes(pCxt, (SShowVnodesStmt*)pStmt);
+    case QUERY_NODE_SHOW_USER_PRIVILEGES_STMT:
+      return collectMetaKeyFromShowUserPrivileges(pCxt, (SShowStmt*)pStmt);
     case QUERY_NODE_SHOW_CREATE_DATABASE_STMT:
       return collectMetaKeyFromShowCreateDatabase(pCxt, (SShowCreateDatabaseStmt*)pStmt);
     case QUERY_NODE_SHOW_CREATE_TABLE_STMT:

source/libs/parser/src/parTokenizer.c

@@ -161,6 +161,7 @@ static SKeyword keywordTable[] = {
     {"PPS",                  TK_PPS},
     {"PRECISION",            TK_PRECISION},
     {"PREV",                 TK_PREV},
+    {"PRIVILEGES",           TK_PRIVILEGES},
     {"QNODE",                TK_QNODE},
     {"QNODES",               TK_QNODES},
     {"QTIME",                TK_QTIME},
@@ -202,6 +203,7 @@ static SKeyword keywordTable[] = {
     {"STREAM",               TK_STREAM},
     {"STREAMS",              TK_STREAMS},
     {"STRICT",               TK_STRICT},
+    {"SUBSCRIBE",            TK_SUBSCRIBE},
     {"SUBSCRIPTIONS",        TK_SUBSCRIPTIONS},
     {"SUBTABLE",             TK_SUBTABLE},
     {"SYSINFO",              TK_SYSINFO},

source/libs/parser/src/parTranslater.c

@@ -251,6 +251,12 @@ static const SSysTableShowAdapter sysTableShowAdapter[] = {
     .numOfShowCols = 1,
     .pShowCols = {"*"}
   },
+  { .showType = QUERY_NODE_SHOW_USER_PRIVILEGES_STMT,
+    .pDbName = TSDB_INFORMATION_SCHEMA_DB,
+    .pTableName = TSDB_INS_TABLE_USER_PRIVILEGES,
+    .numOfShowCols = 1,
+    .pShowCols = {"*"}
+  },
 };
 // clang-format on
 
@@ -5031,7 +5037,7 @@ static int32_t translateAlterUser(STranslateContext* pCxt, SAlterUserStmt* pStmt
   alterReq.sysInfo = pStmt->sysinfo;
   snprintf(alterReq.pass, sizeof(alterReq.pass), "%s", pStmt->password);
   if (NULL != pCxt->pParseCxt->db) {
-    snprintf(alterReq.dbname, sizeof(alterReq.dbname), "%s", pCxt->pParseCxt->db);
+    snprintf(alterReq.objname, sizeof(alterReq.objname), "%s", pCxt->pParseCxt->db);
   }
 
   return buildCmdMsg(pCxt, TDMT_MND_ALTER_USER, (FSerializeFunc)tSerializeSAlterUserReq, &alterReq);
@@ -5710,9 +5716,11 @@ static int32_t translateGrant(STranslateContext* pCxt, SGrantStmt* pStmt) {
     req.alterType = TSDB_ALTER_USER_ADD_READ_DB;
   } else if (PRIVILEGE_TYPE_TEST_MASK(pStmt->privileges, PRIVILEGE_TYPE_WRITE)) {
     req.alterType = TSDB_ALTER_USER_ADD_WRITE_DB;
+  } else if (PRIVILEGE_TYPE_TEST_MASK(pStmt->privileges, PRIVILEGE_TYPE_SUBSCRIBE)) {
+    req.alterType = TSDB_ALTER_USER_ADD_SUBSCRIBE_TOPIC;
   }
   strcpy(req.user, pStmt->userName);
-  sprintf(req.dbname, "%d.%s", pCxt->pParseCxt->acctId, pStmt->dbName);
+  sprintf(req.objname, "%d.%s", pCxt->pParseCxt->acctId, pStmt->objName);
   return buildCmdMsg(pCxt, TDMT_MND_ALTER_USER, (FSerializeFunc)tSerializeSAlterUserReq, &req);
 }
 
@@ -5726,9 +5734,11 @@ static int32_t translateRevoke(STranslateContext* pCxt, SRevokeStmt* pStmt) {
     req.alterType = TSDB_ALTER_USER_REMOVE_READ_DB;
   } else if (PRIVILEGE_TYPE_TEST_MASK(pStmt->privileges, PRIVILEGE_TYPE_WRITE)) {
     req.alterType = TSDB_ALTER_USER_REMOVE_WRITE_DB;
+  } else if (PRIVILEGE_TYPE_TEST_MASK(pStmt->privileges, PRIVILEGE_TYPE_SUBSCRIBE)) {
+    req.alterType = TSDB_ALTER_USER_REMOVE_SUBSCRIBE_TOPIC;
   }
   strcpy(req.user, pStmt->userName);
-  sprintf(req.dbname, "%d.%s", pCxt->pParseCxt->acctId, pStmt->dbName);
+  sprintf(req.objname, "%d.%s", pCxt->pParseCxt->acctId, pStmt->objName);
   return buildCmdMsg(pCxt, TDMT_MND_ALTER_USER, (FSerializeFunc)tSerializeSAlterUserReq, &req);
 }
 
@@ -7504,6 +7514,7 @@ static int32_t rewriteQuery(STranslateContext* pCxt, SQuery* pQuery) {
     case QUERY_NODE_SHOW_CONSUMERS_STMT:
     case QUERY_NODE_SHOW_SUBSCRIPTIONS_STMT:
     case QUERY_NODE_SHOW_TAGS_STMT:
+    case QUERY_NODE_SHOW_USER_PRIVILEGES_STMT:
       code = rewriteShow(pCxt, pQuery);
       break;
     case QUERY_NODE_SHOW_VGROUPS_STMT:

source/libs/parser/test/mockCatalog.cpp

@@ -101,6 +101,10 @@ void generateInformationSchema(MockCatalogService* mcs) {
       .addColumn("table_name", TSDB_DATA_TYPE_BINARY, TSDB_TABLE_NAME_LEN)
       .addColumn("db_name", TSDB_DATA_TYPE_BINARY, TSDB_DB_NAME_LEN)
       .done();
+  mcs->createTableBuilder(TSDB_INFORMATION_SCHEMA_DB, TSDB_INS_TABLE_USER_PRIVILEGES, TSDB_SYSTEM_TABLE, 2)
+      .addColumn("user_name", TSDB_DATA_TYPE_BINARY, TSDB_USER_LEN)
+      .addColumn("privilege", TSDB_DATA_TYPE_BINARY, 10)
+      .done();
 }
 
 void generatePerformanceSchema(MockCatalogService* mcs) {
@@ -248,8 +252,8 @@ int32_t __catalogGetTableDistVgInfo(SCatalog* pCtg, SRequestConnInfo* pConn, con
   return g_mockCatalogService->catalogGetTableDistVgInfo(pTableName, pVgList);
 }
 
-int32_t __catalogGetDBVgVersion(SCatalog* pCtg, const char* dbFName, int32_t* version, int64_t* dbId,
+int32_t __catalogGetDBVgVersion(SCatalog* pCtg, const char* dbFName, int32_t* version, int64_t* dbId, int32_t* tableNum,
-                                int32_t* tableNum, int64_t* stateTs) {
+                                int64_t* stateTs) {
   return 0;
 }
 

source/libs/parser/test/parAlterToBalanceTest.cpp

@@ -613,7 +613,7 @@ TEST_F(ParserInitialATest, alterUser) {
     if (nullptr != pPass) {
       strcpy(expect.pass, pPass);
     }
-    strcpy(expect.dbname, "test");
+    strcpy(expect.objname, "test");
   };
 
   setCheckDdlFunc([&](const SQuery* pQuery, ParserStage stage) {
@@ -627,7 +627,7 @@ TEST_F(ParserInitialATest, alterUser) {
     ASSERT_EQ(req.enable, expect.enable);
     ASSERT_EQ(std::string(req.user), std::string(expect.user));
     ASSERT_EQ(std::string(req.pass), std::string(expect.pass));
-    ASSERT_EQ(std::string(req.dbname), std::string(expect.dbname));
+    ASSERT_EQ(std::string(req.objname), std::string(expect.objname));
   });
 
   setAlterUserReq("wxy", TSDB_ALTER_USER_PASSWD, "123456");

source/libs/parser/test/parExplainToSyncdbTest.cpp

@@ -34,10 +34,38 @@ TEST_F(ParserExplainToSyncdbTest, explain) {
 TEST_F(ParserExplainToSyncdbTest, grant) {
   useDb("root", "test");
 
+  SAlterUserReq expect = {0};
+
+  auto setAlterUserReq = [&](int8_t alterType, const string& user, const string& obj) {
+    expect.alterType = alterType;
+    snprintf(expect.user, sizeof(expect.user), "%s", user.c_str());
+    snprintf(expect.objname, sizeof(expect.objname), "%s", obj.c_str());
+  };
+
+  setCheckDdlFunc([&](const SQuery* pQuery, ParserStage stage) {
+    ASSERT_EQ(nodeType(pQuery->pRoot), QUERY_NODE_GRANT_STMT);
+    ASSERT_EQ(pQuery->pCmdMsg->msgType, TDMT_MND_ALTER_USER);
+    SAlterUserReq req = {0};
+    ASSERT_EQ(tDeserializeSAlterUserReq(pQuery->pCmdMsg->pMsg, pQuery->pCmdMsg->msgLen, &req), TSDB_CODE_SUCCESS);
+    ASSERT_EQ(req.alterType, expect.alterType);
+    ASSERT_EQ(string(req.user), string(expect.user));
+    ASSERT_EQ(string(req.objname), string(expect.objname));
+  });
+
+  setAlterUserReq(TSDB_ALTER_USER_ADD_ALL_DB, "wxy", "0.test");
   run("GRANT ALL ON test.* TO wxy");
+
+  setAlterUserReq(TSDB_ALTER_USER_ADD_READ_DB, "wxy", "0.test");
   run("GRANT READ ON test.* TO wxy");
+
+  setAlterUserReq(TSDB_ALTER_USER_ADD_WRITE_DB, "wxy", "0.test");
   run("GRANT WRITE ON test.* TO wxy");
+
+  setAlterUserReq(TSDB_ALTER_USER_ADD_ALL_DB, "wxy", "0.test");
   run("GRANT READ, WRITE ON test.* TO wxy");
+
+  setAlterUserReq(TSDB_ALTER_USER_ADD_SUBSCRIBE_TOPIC, "wxy", "0.tp1");
+  run("GRANT SUBSCRIBE ON tp1 TO wxy");
 }
 
 TEST_F(ParserExplainToSyncdbTest, insert) {

source/libs/parser/test/parShowToUse.cpp

@@ -213,7 +213,13 @@ TEST_F(ParserShowToUseTest, showTags) {
 TEST_F(ParserShowToUseTest, showUsers) {
   useDb("root", "test");
 
-  run("SHOW users");
+  run("SHOW USERS");
+}
+
+TEST_F(ParserShowToUseTest, showUserPrivileges) {
+  useDb("root", "test");
+
+  run("SHOW USER PRIVILEGES");
 }
 
 TEST_F(ParserShowToUseTest, showVariables) {

source/libs/stream/inc/streamInc.h

@@ -17,7 +17,6 @@
 #define _STREAM_INC_H_
 
 #include "executor.h"
-#include "tref.h"
 #include "tstream.h"
 
 #ifdef __cplusplus
@@ -26,7 +25,6 @@ extern "C" {
 
 typedef struct {
   int8_t inited;
-  int32_t refPool;
   void*  timer;
 } SStreamGlobalEnv;
 

source/libs/stream/src/streamCheckpoint.c

@@ -0,0 +1,194 @@
+/*
+ * 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 "streamInc.h"
+
+int32_t tEncodeSStreamCheckpointSourceReq(SEncoder* pEncoder, const SStreamCheckpointSourceReq* pReq) {
+  if (tStartEncode(pEncoder) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->streamId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->checkpointId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pReq->taskId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pReq->nodeId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->expireTime) < 0) return -1;
+  tEndEncode(pEncoder);
+  return pEncoder->pos;
+}
+
+int32_t tDecodeSStreamCheckpointSourceReq(SDecoder* pDecoder, SStreamCheckpointSourceReq* pReq) {
+  if (tStartDecode(pDecoder) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pReq->streamId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pReq->checkpointId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->taskId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->nodeId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pReq->expireTime) < 0) return -1;
+  tEndDecode(pDecoder);
+  return 0;
+}
+
+int32_t tEncodeSStreamCheckpointSourceRsp(SEncoder* pEncoder, const SStreamCheckpointSourceRsp* pRsp) {
+  if (tStartEncode(pEncoder) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->streamId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->checkpointId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pRsp->taskId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pRsp->nodeId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->expireTime) < 0) return -1;
+  tEndEncode(pEncoder);
+  return pEncoder->pos;
+}
+
+int32_t tDecodeSStreamCheckpointSourceRsp(SDecoder* pDecoder, SStreamCheckpointSourceRsp* pRsp) {
+  if (tStartDecode(pDecoder) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pRsp->streamId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pRsp->checkpointId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->taskId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->nodeId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pRsp->expireTime) < 0) return -1;
+  tEndDecode(pDecoder);
+  return 0;
+}
+
+int32_t tEncodeSStreamCheckpointReq(SEncoder* pEncoder, const SStreamCheckpointReq* pReq) {
+  if (tStartEncode(pEncoder) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->streamId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->checkpointId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pReq->downstreamTaskId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pReq->downstreamNodeId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->upstreamTaskId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->upstreamNodeId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pReq->childId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pReq->expireTime) < 0) return -1;
+  if (tEncodeI8(pEncoder, pReq->taskLevel) < 0) return -1;
+  tEndEncode(pEncoder);
+  return pEncoder->pos;
+}
+
+int32_t tDecodeSStreamCheckpointReq(SDecoder* pDecoder, SStreamCheckpointReq* pReq) {
+  if (tStartDecode(pDecoder) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pReq->streamId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pReq->checkpointId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->downstreamTaskId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->downstreamNodeId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->upstreamTaskId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->upstreamNodeId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pReq->childId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pReq->expireTime) < 0) return -1;
+  if (tDecodeI8(pDecoder, &pReq->taskLevel) < 0) return -1;
+  tEndDecode(pDecoder);
+  return 0;
+}
+
+int32_t tEncodeSStreamCheckpointRsp(SEncoder* pEncoder, const SStreamCheckpointRsp* pRsp) {
+  if (tStartEncode(pEncoder) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->streamId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->checkpointId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pRsp->downstreamTaskId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pRsp->downstreamNodeId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->upstreamTaskId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->upstreamNodeId) < 0) return -1;
+  if (tEncodeI32(pEncoder, pRsp->childId) < 0) return -1;
+  if (tEncodeI64(pEncoder, pRsp->expireTime) < 0) return -1;
+  if (tEncodeI8(pEncoder, pRsp->taskLevel) < 0) return -1;
+  tEndEncode(pEncoder);
+  return pEncoder->pos;
+}
+
+int32_t tDecodeSStreamCheckpointRsp(SDecoder* pDecoder, SStreamCheckpointRsp* pRsp) {
+  if (tStartDecode(pDecoder) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pRsp->streamId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pRsp->checkpointId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->downstreamTaskId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->downstreamNodeId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->upstreamTaskId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->upstreamNodeId) < 0) return -1;
+  if (tDecodeI32(pDecoder, &pRsp->childId) < 0) return -1;
+  if (tDecodeI64(pDecoder, &pRsp->expireTime) < 0) return -1;
+  if (tDecodeI8(pDecoder, &pRsp->taskLevel) < 0) return -1;
+  tEndDecode(pDecoder);
+  return 0;
+}
+
+static int32_t streamAlignCheckpoint(SStreamTask* pTask, int64_t checkpointId, int32_t childId) {
+  if (pTask->checkpointingId == 0) {
+    pTask->checkpointingId = checkpointId;
+    pTask->checkpointAlignCnt = taosArrayGetSize(pTask->childEpInfo);
+  }
+
+  ASSERT(pTask->checkpointingId == checkpointId);
+
+  return atomic_sub_fetch_32(&pTask->checkpointAlignCnt, 1);
+}
+
+static int32_t streamDoCheckpoint(SStreamMeta* pMeta, SStreamTask* pTask, int64_t checkpointId) {
+  // commit tdb state
+  streamStateCommit(pTask->pState);
+  // commit non-tdb state
+  // copy and save new state
+  // report to mnode
+  // send checkpoint req to downstream
+  return 0;
+}
+
+static int32_t streamDoSourceCheckpoint(SStreamMeta* pMeta, SStreamTask* pTask, int64_t checkpointId) {
+  // ref wal
+  // set status checkpointing
+  // do checkpoint
+  return 0;
+}
+int32_t streamProcessCheckpointSourceReq(SStreamMeta* pMeta, SStreamTask* pTask, SStreamCheckpointSourceReq* pReq) {
+  int32_t code;
+  int64_t checkpointId = pReq->checkpointId;
+
+  code = streamDoSourceCheckpoint(pMeta, pTask, checkpointId);
+  if (code < 0) {
+    // rsp error
+    return -1;
+  }
+
+  return 0;
+}
+
+int32_t streamProcessCheckpointReq(SStreamMeta* pMeta, SStreamTask* pTask, SStreamCheckpointReq* pReq) {
+  int32_t code;
+  int64_t checkpointId = pReq->checkpointId;
+  int32_t childId = pReq->childId;
+
+  if (taosArrayGetSize(pTask->childEpInfo) > 0) {
+    code = streamAlignCheckpoint(pTask, checkpointId, childId);
+    if (code > 0) {
+      return 0;
+    }
+    if (code < 0) {
+      ASSERT(0);
+      return -1;
+    }
+  }
+
+  code = streamDoCheckpoint(pMeta, pTask, checkpointId);
+  if (code < 0) {
+    // rsp error
+    return -1;
+  }
+
+  // send rsp to all children
+
+  return 0;
+}
+
+int32_t streamProcessCheckpointRsp(SStreamMeta* pMeta, SStreamTask* pTask, SStreamCheckpointRsp* pRsp) {
+  // recover step2, scan from wal
+  // unref wal
+  // set status normal
+  return 0;
+}

source/libs/stream/src/streamMeta.c

@@ -202,7 +202,7 @@ void streamMetaReleaseTask(SStreamMeta* pMeta, SStreamTask* pTask) {
   }
 }
 
-void streamMetaRemoveTask1(SStreamMeta* pMeta, int32_t taskId) {
+void streamMetaRemoveTask(SStreamMeta* pMeta, int32_t taskId) {
   SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasks, &taskId, sizeof(int32_t));
   if (ppTask) {
     SStreamTask* pTask = *ppTask;
@@ -219,35 +219,6 @@ void streamMetaRemoveTask1(SStreamMeta* pMeta, int32_t taskId) {
   }
 }
 
-int32_t streamMetaRemoveTask(SStreamMeta* pMeta, int32_t taskId) {
-  SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasks, &taskId, sizeof(int32_t));
-  if (ppTask) {
-    SStreamTask* pTask = *ppTask;
-    taosHashRemove(pMeta->pTasks, &taskId, sizeof(int32_t));
-    atomic_store_8(&pTask->taskStatus, TASK_STATUS__DROPPING);
-
-    if (tdbTbDelete(pMeta->pTaskDb, &taskId, sizeof(int32_t), &pMeta->txn) < 0) {
-      /*return -1;*/
-    }
-
-    if (pTask->triggerParam != 0) {
-      taosTmrStop(pTask->timer);
-    }
-
-    while (1) {
-      int8_t schedStatus =
-          atomic_val_compare_exchange_8(&pTask->schedStatus, TASK_SCHED_STATUS__INACTIVE, TASK_SCHED_STATUS__DROPPING);
-      if (schedStatus != TASK_SCHED_STATUS__ACTIVE) {
-        tFreeSStreamTask(pTask);
-        break;
-      }
-      taosMsleep(10);
-    }
-  }
-
-  return 0;
-}
-
 int32_t streamMetaBegin(SStreamMeta* pMeta) {
   if (tdbTxnOpen(&pMeta->txn, 0, tdbDefaultMalloc, tdbDefaultFree, NULL, TDB_TXN_WRITE | TDB_TXN_READ_UNCOMMITTED) <
       0) {

source/libs/stream/src/streamRecover.c

@@ -325,46 +325,3 @@ int32_t tDecodeSStreamRecoverFinishReq(SDecoder* pDecoder, SStreamRecoverFinishR
   tEndDecode(pDecoder);
   return 0;
 }
-
-int32_t tEncodeSStreamCheckpointInfo(SEncoder* pEncoder, const SStreamCheckpointInfo* pCheckpoint) {
-  if (tEncodeI32(pEncoder, pCheckpoint->srcNodeId) < 0) return -1;
-  if (tEncodeI32(pEncoder, pCheckpoint->srcChildId) < 0) return -1;
-  if (tEncodeI64(pEncoder, pCheckpoint->stateProcessedVer) < 0) return -1;
-  return 0;
-}
-
-int32_t tDecodeSStreamCheckpointInfo(SDecoder* pDecoder, SStreamCheckpointInfo* pCheckpoint) {
-  if (tDecodeI32(pDecoder, &pCheckpoint->srcNodeId) < 0) return -1;
-  if (tDecodeI32(pDecoder, &pCheckpoint->srcChildId) < 0) return -1;
-  if (tDecodeI64(pDecoder, &pCheckpoint->stateProcessedVer) < 0) return -1;
-  return 0;
-}
-
-int32_t tEncodeSStreamMultiVgCheckpointInfo(SEncoder* pEncoder, const SStreamMultiVgCheckpointInfo* pCheckpoint) {
-  if (tEncodeI64(pEncoder, pCheckpoint->streamId) < 0) return -1;
-  if (tEncodeI64(pEncoder, pCheckpoint->checkTs) < 0) return -1;
-  if (tEncodeI32(pEncoder, pCheckpoint->checkpointId) < 0) return -1;
-  if (tEncodeI32(pEncoder, pCheckpoint->taskId) < 0) return -1;
-  int32_t sz = taosArrayGetSize(pCheckpoint->checkpointVer);
-  if (tEncodeI32(pEncoder, sz) < 0) return -1;
-  for (int32_t i = 0; i < sz; i++) {
-    SStreamCheckpointInfo* pOneVgCkpoint = taosArrayGet(pCheckpoint->checkpointVer, i);
-    if (tEncodeSStreamCheckpointInfo(pEncoder, pOneVgCkpoint) < 0) return -1;
-  }
-  return 0;
-}
-
-int32_t tDecodeSStreamMultiVgCheckpointInfo(SDecoder* pDecoder, SStreamMultiVgCheckpointInfo* pCheckpoint) {
-  if (tDecodeI64(pDecoder, &pCheckpoint->streamId) < 0) return -1;
-  if (tDecodeI64(pDecoder, &pCheckpoint->checkTs) < 0) return -1;
-  if (tDecodeI32(pDecoder, &pCheckpoint->checkpointId) < 0) return -1;
-  if (tDecodeI32(pDecoder, &pCheckpoint->taskId) < 0) return -1;
-  int32_t sz;
-  if (tDecodeI32(pDecoder, &sz) < 0) return -1;
-  for (int32_t i = 0; i < sz; i++) {
-    SStreamCheckpointInfo oneVgCheckpoint;
-    if (tDecodeSStreamCheckpointInfo(pDecoder, &oneVgCheckpoint) < 0) return -1;
-    taosArrayPush(pCheckpoint->checkpointVer, &oneVgCheckpoint);
-  }
-  return 0;
-}

source/libs/sync/inc/syncCommit.h

@@ -47,8 +47,13 @@ extern "C" {
 //
 void syncOneReplicaAdvance(SSyncNode* pSyncNode);
 void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode);
+
 bool syncAgreeIndex(SSyncNode* pSyncNode, SRaftId* pRaftId, SyncIndex index);
 bool syncAgree(SSyncNode* pSyncNode, SyncIndex index);
+bool syncNodeAgreedUpon(SSyncNode* pNode, SyncIndex index);
+
+int64_t syncNodeUpdateCommitIndex(SSyncNode* ths, SyncIndex commitIndex);
+int64_t syncNodeCheckCommitIndex(SSyncNode* ths, SyncIndex indexLikely);
 
 #ifdef __cplusplus
 }

source/libs/sync/inc/syncInt.h

@@ -50,6 +50,8 @@ typedef struct SyncPreSnapshotReply   SyncPreSnapshotReply;
 typedef struct SyncHeartbeatReply     SyncHeartbeatReply;
 typedef struct SyncHeartbeat          SyncHeartbeat;
 typedef struct SyncPreSnapshot        SyncPreSnapshot;
+typedef struct SSyncLogBuffer         SSyncLogBuffer;
+typedef struct SSyncLogReplMgr        SSyncLogReplMgr;
 
 typedef struct SRaftId {
   SyncNodeId  addr;
@@ -97,6 +99,7 @@ typedef struct SSyncNode {
   char        configPath[TSDB_FILENAME_LEN * 2];
 
   // sync io
+  SSyncLogBuffer* pLogBuf;
   SWal*         pWal;
   const SMsgCb* msgcb;
   int32_t (*syncSendMSg)(const SEpSet* pEpSet, SRpcMsg* pMsg);
@@ -179,6 +182,9 @@ typedef struct SSyncNode {
   SSyncSnapshotSender*   senders[TSDB_MAX_REPLICA];
   SSyncSnapshotReceiver* pNewNodeReceiver;
 
+  // log replication mgr
+  SSyncLogReplMgr* logReplMgrs[TSDB_MAX_REPLICA];
+
   SPeerState peerStates[TSDB_MAX_REPLICA];
 
   // is config changing
@@ -205,11 +211,12 @@ typedef struct SSyncNode {
 
 // open/close --------------
 SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo);
-void       syncNodeStart(SSyncNode* pSyncNode);
+int32_t    syncNodeStart(SSyncNode* pSyncNode);
-void       syncNodeStartStandBy(SSyncNode* pSyncNode);
+int32_t    syncNodeStartStandBy(SSyncNode* pSyncNode);
 void       syncNodeClose(SSyncNode* pSyncNode);
 void       syncNodePreClose(SSyncNode* pSyncNode);
 int32_t    syncNodePropose(SSyncNode* pSyncNode, SRpcMsg* pMsg, bool isWeak);
+int32_t    syncNodeRestore(SSyncNode* pSyncNode);
 void       syncHbTimerDataFree(SSyncHbTimerData* pData);
 
 // on message ---------------------
@@ -244,6 +251,7 @@ int32_t   syncCacheEntry(SSyncLogStore* pLogStore, SSyncRaftEntry* pEntry, LRUHa
 bool      syncNodeHeartbeatReplyTimeout(SSyncNode* pSyncNode);
 bool      syncNodeSnapshotSending(SSyncNode* pSyncNode);
 bool      syncNodeSnapshotRecving(SSyncNode* pSyncNode);
+bool      syncNodeIsReadyForRead(SSyncNode* pSyncNode);
 
 // raft state change --------------
 void syncNodeUpdateTerm(SSyncNode* pSyncNode, SyncTerm term);
@@ -260,6 +268,9 @@ void syncNodeCandidate2Follower(SSyncNode* pSyncNode);
 void syncNodeVoteForTerm(SSyncNode* pSyncNode, SyncTerm term, SRaftId* pRaftId);
 void syncNodeVoteForSelf(SSyncNode* pSyncNode);
 
+// log replication
+SSyncLogReplMgr* syncNodeGetLogReplMgr(SSyncNode* pNode, SRaftId* pDestId);
+
 // snapshot --------------
 bool    syncNodeHasSnapshot(SSyncNode* pSyncNode);
 void    syncNodeMaybeUpdateCommitBySnapshot(SSyncNode* pSyncNode);

source/libs/sync/inc/syncMessage.h

@@ -106,7 +106,7 @@ typedef struct SyncAppendEntriesReply {
   SRaftId  destId;
   // private data
   SyncTerm  term;
-  SyncTerm  privateTerm;
+  SyncTerm  lastMatchTerm;
   bool      success;
   SyncIndex matchIndex;
   SyncIndex lastSendIndex;
@@ -246,6 +246,8 @@ int32_t syncBuildRequestVote(SRpcMsg* pMsg, int32_t vgId);
 int32_t syncBuildRequestVoteReply(SRpcMsg* pMsg, int32_t vgId);
 int32_t syncBuildAppendEntries(SRpcMsg* pMsg, int32_t dataLen, int32_t vgId);
 int32_t syncBuildAppendEntriesReply(SRpcMsg* pMsg, int32_t vgId);
+int32_t syncBuildAppendEntriesFromRaftLog(SSyncNode* pNode, SSyncRaftEntry* pEntry, SyncTerm prevLogTerm,
+                                          SRpcMsg* pRpcMsg);
 int32_t syncBuildHeartbeat(SRpcMsg* pMsg, int32_t vgId);
 int32_t syncBuildHeartbeatReply(SRpcMsg* pMsg, int32_t vgId);
 int32_t syncBuildPreSnapshot(SRpcMsg* pMsg, int32_t vgId);

source/libs/sync/inc/syncPipeline.h

@@ -0,0 +1,116 @@
+/*
+ * Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
+ *
+ * This program is free software: you can use, redistribute, and/or modify
+ * it under the terms of the GNU Affero General Public License, version 3
+ * or later ("AGPL"), as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * You should have received a copy of the GNU Affero General Public License
+ * along with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef _TD_LIBS_SYNC_PIPELINE_H
+#define _TD_LIBS_SYNC_PIPELINE_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "syncInt.h"
+
+typedef struct SSyncReplInfo {
+  bool    barrier;
+  bool    acked;
+  int64_t timeMs;
+  int64_t term;
+} SSyncReplInfo;
+
+typedef struct SSyncLogReplMgr {
+  SSyncReplInfo states[TSDB_SYNC_LOG_BUFFER_SIZE];
+  int64_t       startIndex;
+  int64_t       matchIndex;
+  int64_t       endIndex;
+  int64_t       size;
+  bool          restored;
+  int64_t       peerStartTime;
+  int32_t       retryBackoff;
+  int32_t       peerId;
+} SSyncLogReplMgr;
+
+typedef struct SSyncLogBufEntry {
+  SSyncRaftEntry* pItem;
+  SyncIndex       prevLogIndex;
+  SyncTerm        prevLogTerm;
+} SSyncLogBufEntry;
+
+typedef struct SSyncLogBuffer {
+  SSyncLogBufEntry entries[TSDB_SYNC_LOG_BUFFER_SIZE];
+  int64_t          startIndex;
+  int64_t          commitIndex;
+  int64_t          matchIndex;
+  int64_t          endIndex;
+  int64_t          size;
+  TdThreadMutex    mutex;
+  TdThreadMutexAttr attr;
+} SSyncLogBuffer;
+
+// SSyncLogRepMgr
+SSyncLogReplMgr* syncLogReplMgrCreate();
+void             syncLogReplMgrDestroy(SSyncLogReplMgr* pMgr);
+int32_t          syncLogReplMgrReset(SSyncLogReplMgr* pMgr);
+
+int32_t          syncNodeLogReplMgrInit(SSyncNode* pNode);
+void             syncNodeLogReplMgrDestroy(SSyncNode* pNode);
+
+// access
+static FORCE_INLINE int64_t syncLogGetRetryBackoffTimeMs(SSyncLogReplMgr* pMgr) {
+  return (1 << pMgr->retryBackoff) * SYNC_LOG_REPL_RETRY_WAIT_MS;
+}
+
+static FORCE_INLINE int32_t syncLogGetNextRetryBackoff(SSyncLogReplMgr* pMgr) {
+  return TMIN(pMgr->retryBackoff + 1, SYNC_MAX_RETRY_BACKOFF);
+}
+
+SyncTerm syncLogReplMgrGetPrevLogTerm(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncIndex index);
+
+int32_t  syncLogReplMgrReplicateOnce(SSyncLogReplMgr* pMgr, SSyncNode* pNode);
+int32_t  syncLogBufferReplicateOneTo(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncIndex index, SyncTerm* pTerm,
+                                     SRaftId* pDestId, bool* pBarrier);
+int32_t  syncLogReplMgrReplicateAttemptedOnce(SSyncLogReplMgr* pMgr, SSyncNode* pNode);
+int32_t  syncLogReplMgrReplicateProbeOnce(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncIndex index);
+
+int32_t  syncLogReplMgrProcessReply(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncAppendEntriesReply* pMsg);
+int32_t syncLogReplMgrProcessReplyInRecoveryMode(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncAppendEntriesReply* pMsg);
+int32_t  syncLogReplMgrProcessReplyInNormalMode(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncAppendEntriesReply* pMsg);
+
+int32_t syncLogReplMgrProcessHeartbeatReply(SSyncLogReplMgr* pMgr, SSyncNode* pNode, SyncHeartbeatReply* pMsg);
+int32_t  syncLogReplMgrRetryOnNeed(SSyncLogReplMgr* pMgr, SSyncNode* pNode);
+
+// SSyncLogBuffer
+SSyncLogBuffer* syncLogBufferCreate();
+void            syncLogBufferDestroy(SSyncLogBuffer* pBuf);
+int32_t         syncLogBufferInit(SSyncLogBuffer* pBuf, SSyncNode* pNode);
+int32_t         syncLogBufferReInit(SSyncLogBuffer* pBuf, SSyncNode* pNode);
+
+// access
+int64_t syncLogBufferGetEndIndex(SSyncLogBuffer* pBuf);
+int32_t syncLogBufferAppend(SSyncLogBuffer* pBuf, SSyncNode* pNode, SSyncRaftEntry* pEntry);
+int32_t syncLogBufferAccept(SSyncLogBuffer* pBuf, SSyncNode* pNode, SSyncRaftEntry* pEntry, SyncTerm prevTerm);
+int64_t syncLogBufferProceed(SSyncLogBuffer* pBuf, SSyncNode* pNode, SyncTerm* pMatchTerm);
+int32_t syncLogBufferCommit(SSyncLogBuffer* pBuf, SSyncNode* pNode, int64_t commitIndex);
+int32_t syncLogBufferReset(SSyncLogBuffer* pBuf, SSyncNode* pNode);
+
+// private
+SSyncRaftEntry* syncLogBufferGetOneEntry(SSyncLogBuffer* pBuf, SSyncNode* pNode, SyncIndex index, bool* pInBuf);
+int32_t         syncLogBufferValidate(SSyncLogBuffer* pBuf);
+int32_t         syncLogBufferRollback(SSyncLogBuffer* pBuf, SyncIndex toIndex);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /*_TD_LIBS_SYNC_PIPELINE_H*/

source/libs/sync/inc/syncRaftEntry.h

@@ -42,9 +42,13 @@ SSyncRaftEntry* syncEntryBuildFromClientRequest(const SyncClientRequest* pMsg, S
 SSyncRaftEntry* syncEntryBuildFromRpcMsg(const SRpcMsg* pMsg, SyncTerm term, SyncIndex index);
 SSyncRaftEntry* syncEntryBuildFromAppendEntries(const SyncAppendEntries* pMsg);
 SSyncRaftEntry* syncEntryBuildNoop(SyncTerm term, SyncIndex index, int32_t vgId);
-void            syncEntryDestory(SSyncRaftEntry* pEntry);
+void            syncEntryDestroy(SSyncRaftEntry* pEntry);
 void            syncEntry2OriginalRpc(const SSyncRaftEntry* pEntry, SRpcMsg* pRpcMsg);  // step 7
 
+static FORCE_INLINE bool syncLogIsReplicationBarrier(SSyncRaftEntry* pEntry) {
+  return pEntry->originalRpcType == TDMT_SYNC_NOOP;
+}
+
 typedef struct SRaftEntryHashCache {
   SHashObj*     pEntryHash;
   int32_t       maxCount;

source/libs/sync/inc/syncReplication.h

@@ -52,6 +52,10 @@ int32_t syncNodeSendHeartbeat(SSyncNode* pSyncNode, const SRaftId* pDestId, SRpc
 
 int32_t syncNodeReplicate(SSyncNode* pSyncNode);
 int32_t syncNodeReplicateOne(SSyncNode* pSyncNode, SRaftId* pDestId, bool snapshot);
+int32_t syncNodeReplicateWithoutLock(SSyncNode* pNode);
+
+int32_t syncNodeSendAppendEntries(SSyncNode* pNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg);
+int32_t syncNodeMaybeSendAppendEntries(SSyncNode* pNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg);
 
 #ifdef __cplusplus
 }

source/libs/sync/inc/syncSnapshot.h

@@ -87,6 +87,8 @@ void                   snapshotReceiverForceStop(SSyncSnapshotReceiver *pReceive
 int32_t syncNodeOnSnapshot(SSyncNode *ths, const SRpcMsg *pMsg);
 int32_t syncNodeOnSnapshotReply(SSyncNode *ths, const SRpcMsg *pMsg);
 
+SyncIndex syncNodeGetSnapshotConfigIndex(SSyncNode *pSyncNode, SyncIndex snapshotLastApplyIndex);
+
 // start
 
 #ifdef __cplusplus

source/libs/sync/inc/syncUtil.h

@@ -94,11 +94,12 @@ void syncLogRecvLocalCmd(SSyncNode* pSyncNode, const SyncLocalCmd* pMsg, const c
 void syncLogSendAppendEntriesReply(SSyncNode* pSyncNode, const SyncAppendEntriesReply* pMsg, const char* s);
 void syncLogRecvAppendEntriesReply(SSyncNode* pSyncNode, const SyncAppendEntriesReply* pMsg, const char* s);
 
-void syncLogSendHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, bool printX, int64_t timerElapsed, int64_t execTime);
+void syncLogSendHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, bool printX, int64_t timerElapsed,
-void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64_t timeDiff);
+                          int64_t execTime);
+void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64_t timeDiff, const char* s);
 
 void syncLogSendHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* pMsg, const char* s);
-void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* pMsg, int64_t timeDiff);
+void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* pMsg, int64_t timeDiff, const char* s);
 
 void syncLogSendSyncPreSnapshot(SSyncNode* pSyncNode, const SyncPreSnapshot* pMsg, const char* s);
 void syncLogRecvSyncPreSnapshot(SSyncNode* pSyncNode, const SyncPreSnapshot* pMsg, const char* s);

source/libs/sync/src/syncAppendEntries.c

@@ -15,10 +15,13 @@
 
 #define _DEFAULT_SOURCE
 #include "syncAppendEntries.h"
+#include "syncPipeline.h"
 #include "syncMessage.h"
 #include "syncRaftLog.h"
 #include "syncRaftStore.h"
+#include "syncReplication.h"
 #include "syncUtil.h"
+#include "syncCommit.h"
 
 // TLA+ Spec
 // HandleAppendEntriesRequest(i, j, m) ==
@@ -124,9 +127,113 @@ int32_t syncNodeFollowerCommit(SSyncNode* ths, SyncIndex newCommitIndex) {
   return 0;
 }
 
+SSyncRaftEntry* syncLogAppendEntriesToRaftEntry(const SyncAppendEntries* pMsg) {
+  SSyncRaftEntry* pEntry = taosMemoryMalloc(pMsg->dataLen);
+  if (pEntry == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return NULL;
+  }
+  (void)memcpy(pEntry, pMsg->data, pMsg->dataLen);
+  ASSERT(pEntry->bytes == pMsg->dataLen);
+  return pEntry;
+}
+
 int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
   SyncAppendEntries* pMsg = pRpcMsg->pCont;
   SRpcMsg            rpcRsp = {0};
+  bool               accepted = false;
+  // if already drop replica, do not process
+  if (!syncNodeInRaftGroup(ths, &(pMsg->srcId))) {
+    syncLogRecvAppendEntries(ths, pMsg, "not in my config");
+    goto _IGNORE;
+  }
+
+  int32_t code = syncBuildAppendEntriesReply(&rpcRsp, ths->vgId);
+  if (code != 0) {
+    syncLogRecvAppendEntries(ths, pMsg, "build rsp error");
+    goto _IGNORE;
+  }
+
+  SyncAppendEntriesReply* pReply = rpcRsp.pCont;
+  // prepare response msg
+  pReply->srcId = ths->myRaftId;
+  pReply->destId = pMsg->srcId;
+  pReply->term = ths->pRaftStore->currentTerm;
+  pReply->success = false;
+  pReply->matchIndex = SYNC_INDEX_INVALID;
+  pReply->lastSendIndex = pMsg->prevLogIndex + 1;
+  pReply->startTime = ths->startTime;
+
+  if (pMsg->term < ths->pRaftStore->currentTerm) {
+    goto _SEND_RESPONSE;
+  }
+
+  if (pMsg->term > ths->pRaftStore->currentTerm) {
+    pReply->term = pMsg->term;
+  }
+
+  syncNodeStepDown(ths, pMsg->term);
+  syncNodeResetElectTimer(ths);
+
+  if (pMsg->dataLen < (int32_t)sizeof(SSyncRaftEntry)) {
+    sError("vgId:%d, incomplete append entries received. prev index:%" PRId64 ", term:%" PRId64 ", datalen:%d",
+           ths->vgId, pMsg->prevLogIndex, pMsg->prevLogTerm, pMsg->dataLen);
+    goto _IGNORE;
+  }
+
+  SSyncRaftEntry* pEntry = syncLogAppendEntriesToRaftEntry(pMsg);
+
+  if (pEntry == NULL) {
+    sError("vgId:%d, failed to get raft entry from append entries since %s", ths->vgId, terrstr());
+    goto _IGNORE;
+  }
+
+  if (pMsg->prevLogIndex + 1 != pEntry->index || pEntry->term < 0) {
+    sError("vgId:%d, invalid previous log index in msg. index:%" PRId64 ",  term:%" PRId64 ", prevLogIndex:%" PRId64
+           ", prevLogTerm:%" PRId64,
+           ths->vgId, pEntry->index, pEntry->term, pMsg->prevLogIndex, pMsg->prevLogTerm);
+    goto _IGNORE;
+  }
+
+  sTrace("vgId:%d, recv append entries msg. index:%" PRId64 ", term:%" PRId64 ", preLogIndex:%" PRId64
+         ", prevLogTerm:%" PRId64 " commitIndex:%" PRId64 "",
+         pMsg->vgId, pMsg->prevLogIndex + 1, pMsg->term, pMsg->prevLogIndex, pMsg->prevLogTerm, pMsg->commitIndex);
+
+  // accept
+  if (syncLogBufferAccept(ths->pLogBuf, ths, pEntry, pMsg->prevLogTerm) < 0) {
+    goto _SEND_RESPONSE;
+  }
+  accepted = true;
+
+_SEND_RESPONSE:
+  pReply->matchIndex = syncLogBufferProceed(ths->pLogBuf, ths, &pReply->lastMatchTerm);
+  bool matched = (pReply->matchIndex >= pReply->lastSendIndex);
+  if (accepted && matched) {
+    pReply->success = true;
+    // update commit index only after matching
+    (void)syncNodeUpdateCommitIndex(ths, pMsg->commitIndex);
+  }
+
+  // ack, i.e. send response
+  (void)syncNodeSendMsgById(&pReply->destId, ths, &rpcRsp);
+
+  // commit index, i.e. leader notice me
+  if (syncLogBufferCommit(ths->pLogBuf, ths, ths->commitIndex) < 0) {
+    sError("vgId:%d, failed to commit raft fsm log since %s.", ths->vgId, terrstr());
+    goto _out;
+  }
+
+_out:
+  return 0;
+
+_IGNORE:
+  rpcFreeCont(rpcRsp.pCont);
+  return 0;
+}
+
+int32_t syncNodeOnAppendEntriesOld(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
+  SyncAppendEntries* pMsg = pRpcMsg->pCont;
+  SRpcMsg            rpcRsp = {0};
 
   // if already drop replica, do not process
   if (!syncNodeInRaftGroup(ths, &(pMsg->srcId))) {
@@ -233,18 +340,20 @@ int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
           if (hLocal) {
             taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false);
           } else {
-            syncEntryDestory(pLocalEntry);
+            syncEntryDestroy(pLocalEntry);
           }
 
           if (hAppend) {
             taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false);
           } else {
-            syncEntryDestory(pAppendEntry);
+            syncEntryDestroy(pAppendEntry);
           }
 
           goto _IGNORE;
         }
 
+        ASSERT(pAppendEntry->index == appendIndex);
+
         // append
         code = ths->pLogStore->syncLogAppendEntry(ths->pLogStore, pAppendEntry);
         if (code != 0) {
@@ -255,13 +364,13 @@ int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
           if (hLocal) {
             taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false);
           } else {
-            syncEntryDestory(pLocalEntry);
+            syncEntryDestroy(pLocalEntry);
           }
 
           if (hAppend) {
             taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false);
           } else {
-            syncEntryDestory(pAppendEntry);
+            syncEntryDestroy(pAppendEntry);
           }
 
           goto _IGNORE;
@@ -281,8 +390,8 @@ int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
           snprintf(logBuf, sizeof(logBuf), "ignore, log not exist, truncate error, append-index:%" PRId64, appendIndex);
           syncLogRecvAppendEntries(ths, pMsg, logBuf);
 
-          syncEntryDestory(pLocalEntry);
+          syncEntryDestroy(pLocalEntry);
-          syncEntryDestory(pAppendEntry);
+          syncEntryDestroy(pAppendEntry);
           goto _IGNORE;
         }
 
@@ -296,13 +405,13 @@ int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
           if (hLocal) {
             taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false);
           } else {
-            syncEntryDestory(pLocalEntry);
+            syncEntryDestroy(pLocalEntry);
           }
 
           if (hAppend) {
             taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false);
           } else {
-            syncEntryDestory(pAppendEntry);
+            syncEntryDestroy(pAppendEntry);
           }
 
           goto _IGNORE;
@@ -320,13 +429,13 @@ int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
         if (hLocal) {
           taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false);
         } else {
-          syncEntryDestory(pLocalEntry);
+          syncEntryDestroy(pLocalEntry);
         }
 
         if (hAppend) {
           taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false);
         } else {
-          syncEntryDestory(pAppendEntry);
+          syncEntryDestroy(pAppendEntry);
         }
 
         goto _IGNORE;
@@ -339,13 +448,13 @@ int32_t syncNodeOnAppendEntries(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
     if (hLocal) {
       taosLRUCacheRelease(ths->pLogStore->pCache, hLocal, false);
     } else {
-      syncEntryDestory(pLocalEntry);
+      syncEntryDestroy(pLocalEntry);
     }
 
     if (hAppend) {
       taosLRUCacheRelease(ths->pLogStore->pCache, hAppend, false);
     } else {
-      syncEntryDestory(pAppendEntry);
+      syncEntryDestroy(pAppendEntry);
     }
 
   } else {

source/libs/sync/src/syncAppendEntriesReply.c

@@ -17,7 +17,9 @@
 #include "syncAppendEntriesReply.h"
 #include "syncCommit.h"
 #include "syncIndexMgr.h"
+#include "syncPipeline.h"
 #include "syncMessage.h"
+#include "syncRaftEntry.h"
 #include "syncRaftStore.h"
 #include "syncReplication.h"
 #include "syncSnapshot.h"
@@ -38,8 +40,58 @@
 //
 
 int32_t syncNodeOnAppendEntriesReply(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
-  int32_t                 ret = 0;
   SyncAppendEntriesReply* pMsg = pRpcMsg->pCont;
+  int32_t ret = 0;
+
+  // if already drop replica, do not process
+  if (!syncNodeInRaftGroup(ths, &(pMsg->srcId))) {
+    syncLogRecvAppendEntriesReply(ths, pMsg, "not in my config");
+    return 0;
+  }
+
+  // drop stale response
+  if (pMsg->term < ths->pRaftStore->currentTerm) {
+    syncLogRecvAppendEntriesReply(ths, pMsg, "drop stale response");
+    return 0;
+  }
+
+  if (ths->state == TAOS_SYNC_STATE_LEADER) {
+    if (pMsg->term > ths->pRaftStore->currentTerm) {
+      syncLogRecvAppendEntriesReply(ths, pMsg, "error term");
+      syncNodeStepDown(ths, pMsg->term);
+      return -1;
+    }
+
+    ASSERT(pMsg->term == ths->pRaftStore->currentTerm);
+
+    sTrace("vgId:%d received append entries reply. srcId:0x%016" PRIx64 ",  term:%" PRId64 ", matchIndex:%" PRId64 "",
+           pMsg->vgId, pMsg->srcId.addr, pMsg->term, pMsg->matchIndex);
+
+    if (pMsg->success) {
+      SyncIndex oldMatchIndex = syncIndexMgrGetIndex(ths->pMatchIndex, &(pMsg->srcId));
+      if (pMsg->matchIndex > oldMatchIndex) {
+        syncIndexMgrSetIndex(ths->pMatchIndex, &(pMsg->srcId), pMsg->matchIndex);
+      }
+
+      // commit if needed
+      SyncIndex indexLikely = TMIN(pMsg->matchIndex, ths->pLogBuf->matchIndex);
+      SyncIndex commitIndex = syncNodeCheckCommitIndex(ths, indexLikely);
+      (void)syncLogBufferCommit(ths->pLogBuf, ths, commitIndex);
+    }
+
+    // replicate log
+    SSyncLogReplMgr* pMgr = syncNodeGetLogReplMgr(ths, &pMsg->srcId);
+    if (pMgr == NULL) {
+      sError("vgId:%d, failed to get log repl mgr for src addr: 0x%016" PRIx64, ths->vgId, pMsg->srcId.addr);
+      return -1;
+    }
+    (void)syncLogReplMgrProcessReply(pMgr, ths, pMsg);
+  }
+  return 0;
+}
+
+int32_t syncNodeOnAppendEntriesReplyOld(SSyncNode* ths, SyncAppendEntriesReply* pMsg) {
+  int32_t ret = 0;
 
   // if already drop replica, do not process
   if (!syncNodeInRaftGroup(ths, &(pMsg->srcId))) {

source/libs/sync/src/syncCommit.c

@@ -84,6 +84,7 @@ void syncOneReplicaAdvance(SSyncNode* pSyncNode) {
 }
 
 void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) {
+  ASSERT(false && "deprecated");
   if (pSyncNode == NULL) {
     sError("pSyncNode is NULL");
     return;
@@ -138,7 +139,7 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) {
         if (h) {
           taosLRUCacheRelease(pCache, h, false);
         } else {
-          syncEntryDestory(pEntry);
+          syncEntryDestroy(pEntry);
         }
 
         break;
@@ -150,7 +151,7 @@ void syncMaybeAdvanceCommitIndex(SSyncNode* pSyncNode) {
       if (h) {
         taosLRUCacheRelease(pCache, h, false);
       } else {
-        syncEntryDestory(pEntry);
+        syncEntryDestroy(pEntry);
       }
     }
   }
@@ -286,15 +287,48 @@ bool syncAgree(SSyncNode* pSyncNode, SyncIndex index) {
 }
 */
 
-bool syncAgree(SSyncNode* pSyncNode, SyncIndex index) {
+bool syncNodeAgreedUpon(SSyncNode* pNode, SyncIndex index) {
+  int            count = 0;
+  SSyncIndexMgr* pMatches = pNode->pMatchIndex;
+  ASSERT(pNode->replicaNum == pMatches->replicaNum);
+
+  for (int i = 0; i < pNode->replicaNum; i++) {
+    SyncIndex matchIndex = pMatches->index[i];
+    if (matchIndex >= index) {
+      count++;
+    }
+  }
+
+  return count >= pNode->quorum;
+}
+
+bool syncAgree(SSyncNode* pNode, SyncIndex index) {
   int agreeCount = 0;
-  for (int i = 0; i < pSyncNode->replicaNum; ++i) {
+  for (int i = 0; i < pNode->replicaNum; ++i) {
-    if (syncAgreeIndex(pSyncNode, &(pSyncNode->replicasId[i]), index)) {
+    if (syncAgreeIndex(pNode, &(pNode->replicasId[i]), index)) {
       ++agreeCount;
     }
-    if (agreeCount >= pSyncNode->quorum) {
+    if (agreeCount >= pNode->quorum) {
       return true;
     }
   }
   return false;
 }
+
+int64_t syncNodeUpdateCommitIndex(SSyncNode* ths, SyncIndex commitIndex) {
+  SyncIndex lastVer = ths->pLogStore->syncLogLastIndex(ths->pLogStore);
+  commitIndex = TMAX(commitIndex, ths->commitIndex);
+  ths->commitIndex = TMIN(commitIndex, lastVer);
+  ths->pLogStore->syncLogUpdateCommitIndex(ths->pLogStore, ths->commitIndex);
+  return ths->commitIndex;
+}
+
+int64_t syncNodeCheckCommitIndex(SSyncNode* ths, SyncIndex indexLikely) {
+  if (indexLikely > ths->commitIndex && syncNodeAgreedUpon(ths, indexLikely)) {
+    SyncIndex commitIndex = indexLikely;
+    syncNodeUpdateCommitIndex(ths, commitIndex);
+    sTrace("vgId:%d, agreed upon. role:%d, term:%" PRId64 ", index: %" PRId64 "", ths->vgId, ths->state,
+           ths->pRaftStore->currentTerm, commitIndex);
+  }
+  return ths->commitIndex;
+}

source/libs/sync/src/syncEnv.c

@@ -100,7 +100,9 @@ SSyncNode *syncNodeAcquire(int64_t rid) {
   return pNode;
 }
 
-void syncNodeRelease(SSyncNode *pNode) { taosReleaseRef(gNodeRefId, pNode->rid); }
+void syncNodeRelease(SSyncNode *pNode) {
+  if (pNode) taosReleaseRef(gNodeRefId, pNode->rid);
+}
 
 int64_t syncHbTimerDataAdd(SSyncHbTimerData *pData) {
   pData->rid = taosAddRef(gHbDataRefId, pData);

source/libs/sync/src/syncIndexMgr.c

@@ -81,6 +81,15 @@ void syncIndexMgrSetIndex(SSyncIndexMgr *pSyncIndexMgr, const SRaftId *pRaftId,
          index);
 }
 
+SSyncLogReplMgr *syncNodeGetLogReplMgr(SSyncNode *pNode, SRaftId *pDestId) {
+  for (int i = 0; i < pNode->replicaNum; i++) {
+    if (syncUtilSameId(&(pNode->replicasId[i]), pDestId)) {
+      return pNode->logReplMgrs[i];
+    }
+  }
+  return NULL;
+}
+
 SyncIndex syncIndexMgrGetIndex(SSyncIndexMgr *pSyncIndexMgr, const SRaftId *pRaftId) {
   if (pSyncIndexMgr == NULL) {
     return SYNC_INDEX_INVALID;

source/libs/sync/src/syncMain.c

@@ -22,6 +22,7 @@
 #include "syncEnv.h"
 #include "syncIndexMgr.h"
 #include "syncInt.h"
+#include "syncPipeline.h"
 #include "syncMessage.h"
 #include "syncRaftCfg.h"
 #include "syncRaftLog.h"
@@ -34,6 +35,7 @@
 #include "syncTimeout.h"
 #include "syncUtil.h"
 #include "syncVoteMgr.h"
+#include "tref.h"
 
 static void    syncNodeEqPingTimer(void* param, void* tmrId);
 static void    syncNodeEqElectTimer(void* param, void* tmrId);
@@ -56,7 +58,6 @@ static int32_t syncNodeLeaderTransferTo(SSyncNode* pSyncNode, SNodeInfo newLeade
 static int32_t syncDoLeaderTransfer(SSyncNode* ths, SRpcMsg* pRpcMsg, SSyncRaftEntry* pEntry);
 
 static ESyncStrategy syncNodeStrategy(SSyncNode* pSyncNode);
-static SyncIndex     syncNodeGetSnapshotConfigIndex(SSyncNode* pSyncNode, SyncIndex snapshotLastApplyIndex);
 
 int64_t syncOpen(SSyncInfo* pSyncInfo) {
   SSyncNode* pSyncNode = syncNodeOpen(pSyncInfo);
@@ -80,12 +81,29 @@ int64_t syncOpen(SSyncInfo* pSyncInfo) {
   return pSyncNode->rid;
 }
 
-void syncStart(int64_t rid) {
+int32_t syncStart(int64_t rid) {
   SSyncNode* pSyncNode = syncNodeAcquire(rid);
-  if (pSyncNode != NULL) {
+  if (pSyncNode == NULL) {
-    syncNodeStart(pSyncNode);
+    sError("failed to acquire rid: %" PRId64 " of tsNodeReftId for pSyncNode", rid);
-    syncNodeRelease(pSyncNode);
+    return -1;
   }
+
+  if (syncNodeRestore(pSyncNode) < 0) {
+    sError("vgId:%d, failed to restore sync log buffer since %s", pSyncNode->vgId, terrstr());
+    goto _err;
+  }
+
+  if (syncNodeStart(pSyncNode) < 0) {
+    sError("vgId:%d, failed to start sync node since %s", pSyncNode->vgId, terrstr());
+    goto _err;
+  }
+
+  syncNodeRelease(pSyncNode);
+  return 0;
+
+_err:
+  syncNodeRelease(pSyncNode);
+  return -1;
 }
 
 void syncStop(int64_t rid) {
@@ -381,15 +399,13 @@ int32_t syncStepDown(int64_t rid, SyncTerm newTerm) {
   return 0;
 }
 
-bool syncIsReadyForRead(int64_t rid) {
+bool syncNodeIsReadyForRead(SSyncNode* pSyncNode) {
-  SSyncNode* pSyncNode = syncNodeAcquire(rid);
   if (pSyncNode == NULL) {
     sError("sync ready for read error");
     return false;
   }
 
   if (pSyncNode->state == TAOS_SYNC_STATE_LEADER && pSyncNode->restoreFinish) {
-    syncNodeRelease(pSyncNode);
     return true;
   }
 
@@ -428,7 +444,7 @@ bool syncIsReadyForRead(int64_t rid) {
           if (h) {
             taosLRUCacheRelease(pCache, h, false);
           } else {
-            syncEntryDestory(pEntry);
+            syncEntryDestroy(pEntry);
           }
         }
       }
@@ -443,6 +459,18 @@ bool syncIsReadyForRead(int64_t rid) {
     }
   }
 
+  return ready;
+}
+
+bool syncIsReadyForRead(int64_t rid) {
+  SSyncNode* pSyncNode = syncNodeAcquire(rid);
+  if (pSyncNode == NULL) {
+    sError("sync ready for read error");
+    return false;
+  }
+
+  bool ready = syncNodeIsReadyForRead(pSyncNode);
+
   syncNodeRelease(pSyncNode);
   return ready;
 }
@@ -521,6 +549,7 @@ SSyncState syncGetState(int64_t rid) {
   if (pSyncNode != NULL) {
     state.state = pSyncNode->state;
     state.restored = pSyncNode->restoreFinish;
+    state.canRead = syncNodeIsReadyForRead(pSyncNode);
     syncNodeRelease(pSyncNode);
   }
 
@@ -543,7 +572,7 @@ int32_t syncGetSnapshotByIndex(int64_t rid, SyncIndex index, SSnapshot* pSnapsho
   int32_t         code = pSyncNode->pLogStore->syncLogGetEntry(pSyncNode->pLogStore, index, &pEntry);
   if (code != 0) {
     if (pEntry != NULL) {
-      syncEntryDestory(pEntry);
+      syncEntryDestroy(pEntry);
     }
     syncNodeRelease(pSyncNode);
     return -1;
@@ -555,7 +584,7 @@ int32_t syncGetSnapshotByIndex(int64_t rid, SyncIndex index, SSnapshot* pSnapsho
   pSnapshot->lastApplyTerm = pEntry->term;
   pSnapshot->lastConfigIndex = syncNodeGetSnapshotConfigIndex(pSyncNode, index);
 
-  syncEntryDestory(pEntry);
+  syncEntryDestroy(pEntry);
   syncNodeRelease(pSyncNode);
   return 0;
 }
@@ -626,7 +655,7 @@ void syncGetRetryEpSet(int64_t rid, SEpSet* pEpSet) {
     tstrncpy(pEp->fqdn, pSyncNode->pRaftCfg->cfg.nodeInfo[i].nodeFqdn, TSDB_FQDN_LEN);
     pEp->port = (pSyncNode->pRaftCfg->cfg.nodeInfo)[i].nodePort;
     pEpSet->numOfEps++;
-    sInfo("vgId:%d, sync get retry epset, index:%d %s:%d", pSyncNode->vgId, i, pEp->fqdn, pEp->port);
+    sDebug("vgId:%d, sync get retry epset, index:%d %s:%d", pSyncNode->vgId, i, pEp->fqdn, pEp->port);
   }
   if (pEpSet->numOfEps > 0) {
     pEpSet->inUse = (pSyncNode->pRaftCfg->cfg.myIndex + 1) % pEpSet->numOfEps;
@@ -756,6 +785,29 @@ static int32_t syncHbTimerStop(SSyncNode* pSyncNode, SSyncTimer* pSyncTimer) {
   return ret;
 }
 
+int32_t syncNodeLogStoreRestoreOnNeed(SSyncNode* pNode) {
+  ASSERT(pNode->pLogStore != NULL && "log store not created");
+  ASSERT(pNode->pFsm != NULL && "pFsm not registered");
+  ASSERT(pNode->pFsm->FpGetSnapshotInfo != NULL && "FpGetSnapshotInfo not registered");
+  SSnapshot snapshot;
+  if (pNode->pFsm->FpGetSnapshotInfo(pNode->pFsm, &snapshot) < 0) {
+    sError("vgId:%d, failed to get snapshot info since %s", pNode->vgId, terrstr());
+    return -1;
+  }
+  SyncIndex commitIndex = snapshot.lastApplyIndex;
+  SyncIndex firstVer = pNode->pLogStore->syncLogBeginIndex(pNode->pLogStore);
+  SyncIndex lastVer = pNode->pLogStore->syncLogLastIndex(pNode->pLogStore);
+  if (lastVer < commitIndex || firstVer > commitIndex + 1) {
+    if (pNode->pLogStore->syncLogRestoreFromSnapshot(pNode->pLogStore, commitIndex)) {
+      sError("vgId:%d, failed to restore log store from snapshot since %s. lastVer: %" PRId64 ", snapshotVer: %" PRId64,
+             pNode->vgId, terrstr(), lastVer, commitIndex);
+      return -1;
+    }
+  }
+  return 0;
+}
+
+// open/close --------------
 SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) {
   SSyncNode* pSyncNode = taosMemoryCalloc(1, sizeof(SSyncNode));
   if (pSyncNode == NULL) {
@@ -831,6 +883,13 @@ SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) {
   pSyncNode->syncEqMsg = pSyncInfo->syncEqMsg;
   pSyncNode->syncEqCtrlMsg = pSyncInfo->syncEqCtrlMsg;
 
+  // create raft log ring buffer
+  pSyncNode->pLogBuf = syncLogBufferCreate();
+  if (pSyncNode->pLogBuf == NULL) {
+    sError("failed to init sync log buffer since %s. vgId:%d", terrstr(), pSyncNode->vgId);
+    goto _error;
+  }
+
   // init raft config
   pSyncNode->pRaftCfg = raftCfgOpen(pSyncNode->configPath);
   if (pSyncNode->pRaftCfg == NULL) {
@@ -956,6 +1015,9 @@ SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) {
   }
   pSyncNode->commitIndex = commitIndex;
 
+  if (syncNodeLogStoreRestoreOnNeed(pSyncNode) < 0) {
+    goto _error;
+  }
   // timer ms init
   pSyncNode->pingBaseLine = PING_TIMER_MS;
   pSyncNode->electBaseLine = ELECT_TIMER_MS_MIN;
@@ -1013,9 +1075,13 @@ SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) {
   // is config changing
   pSyncNode->changing = false;
 
+  // replication mgr
+  syncNodeLogReplMgrInit(pSyncNode);
+
   // peer state
   syncNodePeerStateInit(pSyncNode);
 
+  //
   // min match index
   pSyncNode->minMatchIndex = SYNC_INDEX_INVALID;
 
@@ -1031,6 +1097,12 @@ SSyncNode* syncNodeOpen(SSyncInfo* pSyncInfo) {
   // snapshotting
   atomic_store_64(&pSyncNode->snapshottingIndex, SYNC_INDEX_INVALID);
 
+  // init log buffer
+  if (syncLogBufferInit(pSyncNode->pLogBuf, pSyncNode) < 0) {
+    sError("vgId:%d, failed to init sync log buffer since %s", pSyncNode->vgId, terrstr());
+    goto _error;
+  }
+
   pSyncNode->isStart = true;
   pSyncNode->electNum = 0;
   pSyncNode->becomeLeaderNum = 0;
@@ -1064,7 +1136,49 @@ void syncNodeMaybeUpdateCommitBySnapshot(SSyncNode* pSyncNode) {
   }
 }
 
-void syncNodeStart(SSyncNode* pSyncNode) {
+int32_t syncNodeRestore(SSyncNode* pSyncNode) {
+  ASSERT(pSyncNode->pLogStore != NULL && "log store not created");
+  ASSERT(pSyncNode->pLogBuf != NULL && "ring log buffer not created");
+
+  SyncIndex lastVer = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore);
+  SyncIndex commitIndex = pSyncNode->pLogStore->syncLogCommitIndex(pSyncNode->pLogStore);
+  SyncIndex endIndex = pSyncNode->pLogBuf->endIndex;
+  if (lastVer != -1 && endIndex != lastVer + 1) {
+    terrno = TSDB_CODE_WAL_LOG_INCOMPLETE;
+    sError("vgId:%d, failed to restore sync node since %s. expected lastLogIndex: %" PRId64 ", lastVer: %" PRId64 "",
+           pSyncNode->vgId, terrstr(), endIndex - 1, lastVer);
+    return -1;
+  }
+
+  ASSERT(endIndex == lastVer + 1);
+  commitIndex = TMAX(pSyncNode->commitIndex, commitIndex);
+
+  if (syncLogBufferCommit(pSyncNode->pLogBuf, pSyncNode, commitIndex) < 0) {
+    return -1;
+  }
+
+  return 0;
+}
+
+int32_t syncNodeStart(SSyncNode* pSyncNode) {
+  // start raft
+  if (pSyncNode->replicaNum == 1) {
+    raftStoreNextTerm(pSyncNode->pRaftStore);
+    syncNodeBecomeLeader(pSyncNode, "one replica start");
+
+    // Raft 3.6.2 Committing entries from previous terms
+    syncNodeAppendNoop(pSyncNode);
+  } else {
+    syncNodeBecomeFollower(pSyncNode, "first start");
+  }
+
+  int32_t ret = 0;
+  ret = syncNodeStartPingTimer(pSyncNode);
+  ASSERT(ret == 0);
+  return ret;
+}
+
+void syncNodeStartOld(SSyncNode* pSyncNode) {
   // start raft
   if (pSyncNode->replicaNum == 1) {
     raftStoreNextTerm(pSyncNode->pRaftStore);
@@ -1083,7 +1197,7 @@ void syncNodeStart(SSyncNode* pSyncNode) {
   ASSERT(ret == 0);
 }
 
-void syncNodeStartStandBy(SSyncNode* pSyncNode) {
+int32_t syncNodeStartStandBy(SSyncNode* pSyncNode) {
   // state change
   pSyncNode->state = TAOS_SYNC_STATE_FOLLOWER;
   syncNodeStopHeartbeatTimer(pSyncNode);
@@ -1096,6 +1210,7 @@ void syncNodeStartStandBy(SSyncNode* pSyncNode) {
   ret = 0;
   ret = syncNodeStartPingTimer(pSyncNode);
   ASSERT(ret == 0);
+  return ret;
 }
 
 void syncNodePreClose(SSyncNode* pSyncNode) {
@@ -1116,6 +1231,7 @@ void syncNodeClose(SSyncNode* pSyncNode) {
   ASSERT(ret == 0);
   pSyncNode->pRaftStore = NULL;
 
+  syncNodeLogReplMgrDestroy(pSyncNode);
   syncRespMgrDestroy(pSyncNode->pSyncRespMgr);
   pSyncNode->pSyncRespMgr = NULL;
   voteGrantedDestroy(pSyncNode->pVotesGranted);
@@ -1128,6 +1244,8 @@ void syncNodeClose(SSyncNode* pSyncNode) {
   pSyncNode->pMatchIndex = NULL;
   logStoreDestory(pSyncNode->pLogStore);
   pSyncNode->pLogStore = NULL;
+  syncLogBufferDestroy(pSyncNode->pLogBuf);
+  pSyncNode->pLogBuf = NULL;
   raftCfgClose(pSyncNode->pRaftCfg);
   pSyncNode->pRaftCfg = NULL;
 
@@ -1529,7 +1647,7 @@ void syncNodeDoConfigChange(SSyncNode* pSyncNode, SSyncCfg* pNewConfig, SyncInde
 
       // Raft 3.6.2 Committing entries from previous terms
       syncNodeAppendNoop(pSyncNode);
-      syncMaybeAdvanceCommitIndex(pSyncNode);
+      // syncMaybeAdvanceCommitIndex(pSyncNode);
 
     } else {
       syncNodeBecomeFollower(pSyncNode, tmpbuf);
@@ -1617,6 +1735,9 @@ void syncNodeBecomeFollower(SSyncNode* pSyncNode, const char* debugStr) {
   // min match index
   pSyncNode->minMatchIndex = SYNC_INDEX_INVALID;
 
+  // reset log buffer
+  syncLogBufferReset(pSyncNode->pLogBuf, pSyncNode);
+
   // trace log
   sNTrace(pSyncNode, "become follower %s", debugStr);
 }
@@ -1712,6 +1833,9 @@ void syncNodeBecomeLeader(SSyncNode* pSyncNode, const char* debugStr) {
   // min match index
   pSyncNode->minMatchIndex = SYNC_INDEX_INVALID;
 
+  // reset log buffer
+  syncLogBufferReset(pSyncNode->pLogBuf, pSyncNode);
+
   // trace log
   sNInfo(pSyncNode, "become leader %s", debugStr);
 }
@@ -1723,6 +1847,22 @@ void syncNodeCandidate2Leader(SSyncNode* pSyncNode) {
 
   sNTrace(pSyncNode, "state change syncNodeCandidate2Leader");
 
+  int32_t ret = syncNodeAppendNoop(pSyncNode);
+  if (ret < 0) {
+    sError("vgId:%d, failed to append noop entry since %s", pSyncNode->vgId, terrstr());
+  }
+
+  SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore);
+  ASSERT(lastIndex >= 0);
+  sInfo("vgId:%d, become leader. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64 "",
+        pSyncNode->vgId, pSyncNode->pRaftStore->currentTerm, pSyncNode->commitIndex, lastIndex);
+}
+
+void syncNodeCandidate2LeaderOld(SSyncNode* pSyncNode) {
+  ASSERT(pSyncNode->state == TAOS_SYNC_STATE_CANDIDATE);
+  ASSERT(voteGrantedMajority(pSyncNode->pVotesGranted));
+  syncNodeBecomeLeader(pSyncNode, "candidate to leader");
+
   // Raft 3.6.2 Committing entries from previous terms
   syncNodeAppendNoop(pSyncNode);
   syncMaybeAdvanceCommitIndex(pSyncNode);
@@ -1746,18 +1886,30 @@ int32_t syncNodePeerStateInit(SSyncNode* pSyncNode) {
 void syncNodeFollower2Candidate(SSyncNode* pSyncNode) {
   ASSERT(pSyncNode->state == TAOS_SYNC_STATE_FOLLOWER);
   pSyncNode->state = TAOS_SYNC_STATE_CANDIDATE;
+  SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore);
+  sInfo("vgId:%d, become candidate from follower. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64,
+        pSyncNode->vgId, pSyncNode->pRaftStore->currentTerm, pSyncNode->commitIndex, lastIndex);
+
   sNTrace(pSyncNode, "follower to candidate");
 }
 
 void syncNodeLeader2Follower(SSyncNode* pSyncNode) {
   ASSERT(pSyncNode->state == TAOS_SYNC_STATE_LEADER);
   syncNodeBecomeFollower(pSyncNode, "leader to follower");
+  SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore);
+  sInfo("vgId:%d, become follower from leader. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64,
+        pSyncNode->vgId, pSyncNode->pRaftStore->currentTerm, pSyncNode->commitIndex, lastIndex);
+
   sNTrace(pSyncNode, "leader to follower");
 }
 
 void syncNodeCandidate2Follower(SSyncNode* pSyncNode) {
   ASSERT(pSyncNode->state == TAOS_SYNC_STATE_CANDIDATE);
   syncNodeBecomeFollower(pSyncNode, "candidate to follower");
+  SyncIndex lastIndex = pSyncNode->pLogStore->syncLogLastIndex(pSyncNode->pLogStore);
+  sInfo("vgId:%d, become follower from candidate. term: %" PRId64 ", commit index: %" PRId64 ", last index: %" PRId64,
+        pSyncNode->vgId, pSyncNode->pRaftStore->currentTerm, pSyncNode->commitIndex, lastIndex);
+
   sNTrace(pSyncNode, "candidate to follower");
 }
 
@@ -1911,7 +2063,7 @@ SyncTerm syncNodeGetPreTerm(SSyncNode* pSyncNode, SyncIndex index) {
     if (h) {
       taosLRUCacheRelease(pCache, h, false);
     } else {
-      syncEntryDestory(pPreEntry);
+      syncEntryDestroy(pPreEntry);
     }
 
     return preTerm;
@@ -2153,7 +2305,7 @@ static int32_t syncNodeEqNoop(SSyncNode* pNode) {
 
   SRpcMsg rpcMsg = {0};
   int32_t code = syncBuildClientRequestFromNoopEntry(&rpcMsg, pEntry, pNode->vgId);
-  syncEntryDestory(pEntry);
+  syncEntryDestroy(pEntry);
 
   sNTrace(pNode, "propose msg, type:noop");
   code = (*pNode->syncEqMsg)(pNode->msgcb, &rpcMsg);
@@ -2181,6 +2333,37 @@ int32_t syncCacheEntry(SSyncLogStore* pLogStore, SSyncRaftEntry* pEntry, LRUHand
   return code;
 }
 
+int32_t syncNodeAppend(SSyncNode* ths, SSyncRaftEntry* pEntry) {
+  // append to log buffer
+  if (syncLogBufferAppend(ths->pLogBuf, ths, pEntry) < 0) {
+    sError("vgId:%d, failed to enqueue sync log buffer. index:%" PRId64 "", ths->vgId, pEntry->index);
+    return -1;
+  }
+
+  // proceed match index, with replicating on needed
+  SyncIndex matchIndex = syncLogBufferProceed(ths->pLogBuf, ths, NULL);
+
+  sTrace("vgId:%d, append raft entry. index: %" PRId64 ", term: %" PRId64 " pBuf: [%" PRId64 " %" PRId64 " %" PRId64
+         ", %" PRId64 ")",
+         ths->vgId, pEntry->index, pEntry->term, ths->pLogBuf->startIndex, ths->pLogBuf->commitIndex,
+         ths->pLogBuf->matchIndex, ths->pLogBuf->endIndex);
+
+  // multi replica
+  if (ths->replicaNum > 1) {
+    return 0;
+  }
+
+  // single replica
+  (void)syncNodeUpdateCommitIndex(ths, matchIndex);
+
+  if (syncLogBufferCommit(ths->pLogBuf, ths, ths->commitIndex) < 0) {
+    sError("vgId:%d, failed to commit until commitIndex:%" PRId64 "", ths->vgId, ths->commitIndex);
+    return -1;
+  }
+
+  return 0;
+}
+
 bool syncNodeHeartbeatReplyTimeout(SSyncNode* pSyncNode) {
   if (pSyncNode->replicaNum == 1) {
     return false;
@@ -2224,6 +2407,20 @@ bool syncNodeSnapshotRecving(SSyncNode* pSyncNode) {
 }
 
 static int32_t syncNodeAppendNoop(SSyncNode* ths) {
+  SyncIndex index = syncLogBufferGetEndIndex(ths->pLogBuf);
+  SyncTerm  term = ths->pRaftStore->currentTerm;
+
+  SSyncRaftEntry* pEntry = syncEntryBuildNoop(term, index, ths->vgId);
+  if (pEntry == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  int32_t ret = syncNodeAppend(ths, pEntry);
+  return 0;
+}
+
+static int32_t syncNodeAppendNoopOld(SSyncNode* ths) {
   int32_t ret = 0;
 
   SyncIndex       index = ths->pLogStore->syncLogWriteIndex(ths->pLogStore);
@@ -2246,7 +2443,7 @@ static int32_t syncNodeAppendNoop(SSyncNode* ths) {
   if (h) {
     taosLRUCacheRelease(ths->pLogStore->pCache, h, false);
   } else {
-    syncEntryDestory(pEntry);
+    syncEntryDestroy(pEntry);
   }
 
   return ret;
@@ -2255,9 +2452,13 @@ static int32_t syncNodeAppendNoop(SSyncNode* ths) {
 int32_t syncNodeOnHeartbeat(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
   SyncHeartbeat* pMsg = pRpcMsg->pCont;
 
+  const STraceId* trace = &pRpcMsg->info.traceId;
+  char            tbuf[40] = {0};
+  TRACE_TO_STR(trace, tbuf);
+
   int64_t tsMs = taosGetTimestampMs();
   int64_t timeDiff = tsMs - pMsg->timeStamp;
-  syncLogRecvHeartbeat(ths, pMsg, timeDiff);
+  syncLogRecvHeartbeat(ths, pMsg, timeDiff, tbuf);
 
   SRpcMsg rpcMsg = {0};
   (void)syncBuildHeartbeatReply(&rpcMsg, ths->vgId);
@@ -2267,6 +2468,7 @@ int32_t syncNodeOnHeartbeat(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
   pMsgReply->srcId = ths->myRaftId;
   pMsgReply->term = ths->pRaftStore->currentTerm;
   pMsgReply->privateTerm = 8864;  // magic number
+  pMsgReply->startTime = ths->startTime;
   pMsgReply->timeStamp = tsMs;
 
   if (pMsg->term == ths->pRaftStore->currentTerm && ths->state != TAOS_SYNC_STATE_LEADER) {
@@ -2330,11 +2532,35 @@ int32_t syncNodeOnHeartbeat(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
 }
 
 int32_t syncNodeOnHeartbeatReply(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
+  const STraceId* trace = &pRpcMsg->info.traceId;
+  char            tbuf[40] = {0};
+  TRACE_TO_STR(trace, tbuf);
+
   SyncHeartbeatReply* pMsg = pRpcMsg->pCont;
+  SSyncLogReplMgr*    pMgr = syncNodeGetLogReplMgr(ths, &pMsg->srcId);
+  if (pMgr == NULL) {
+    sError("vgId:%d, failed to get log repl mgr for the peer at addr 0x016%" PRIx64 "", ths->vgId, pMsg->srcId.addr);
+    return -1;
+  }
+
+  int64_t tsMs = taosGetTimestampMs();
+  syncLogRecvHeartbeatReply(ths, pMsg, tsMs - pMsg->timeStamp, tbuf);
+
+  syncIndexMgrSetRecvTime(ths->pMatchIndex, &pMsg->srcId, tsMs);
+
+  return syncLogReplMgrProcessHeartbeatReply(pMgr, ths, pMsg);
+}
+
+int32_t syncNodeOnHeartbeatReplyOld(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
+  SyncHeartbeatReply* pMsg = pRpcMsg->pCont;
+
+  const STraceId* trace = &pRpcMsg->info.traceId;
+  char            tbuf[40] = {0};
+  TRACE_TO_STR(trace, tbuf);
 
   int64_t tsMs = taosGetTimestampMs();
   int64_t timeDiff = tsMs - pMsg->timeStamp;
-  syncLogRecvHeartbeatReply(ths, pMsg, timeDiff);
+  syncLogRecvHeartbeatReply(ths, pMsg, timeDiff, tbuf);
 
   // update last reply time, make decision whether the other node is alive or not
   syncIndexMgrSetRecvTime(ths->pMatchIndex, &pMsg->srcId, tsMs);
@@ -2345,6 +2571,26 @@ int32_t syncNodeOnLocalCmd(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
   SyncLocalCmd* pMsg = pRpcMsg->pCont;
   syncLogRecvLocalCmd(ths, pMsg, "");
 
+  if (pMsg->cmd == SYNC_LOCAL_CMD_STEP_DOWN) {
+    syncNodeStepDown(ths, pMsg->sdNewTerm);
+
+  } else if (pMsg->cmd == SYNC_LOCAL_CMD_FOLLOWER_CMT) {
+    (void)syncNodeUpdateCommitIndex(ths, pMsg->fcIndex);
+    if (syncLogBufferCommit(ths->pLogBuf, ths, ths->commitIndex) < 0) {
+      sError("vgId:%d, failed to commit raft log since %s. commit index: %" PRId64 "", ths->vgId, terrstr(),
+             ths->commitIndex);
+    }
+  } else {
+    sError("error local cmd");
+  }
+
+  return 0;
+}
+
+int32_t syncNodeOnLocalCmdOld(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
+  SyncLocalCmd* pMsg = pRpcMsg->pCont;
+  syncLogRecvLocalCmd(ths, pMsg, "");
+
   if (pMsg->cmd == SYNC_LOCAL_CMD_STEP_DOWN) {
     syncNodeStepDown(ths, pMsg->sdNewTerm);
 
@@ -2372,6 +2618,31 @@ int32_t syncNodeOnLocalCmd(SSyncNode* ths, const SRpcMsg* pRpcMsg) {
 int32_t syncNodeOnClientRequest(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIndex) {
   sNTrace(ths, "on client request");
 
+  int32_t code = 0;
+
+  SyncIndex       index = syncLogBufferGetEndIndex(ths->pLogBuf);
+  SyncTerm        term = ths->pRaftStore->currentTerm;
+  SSyncRaftEntry* pEntry = NULL;
+  if (pMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) {
+    pEntry = syncEntryBuildFromClientRequest(pMsg->pCont, term, index);
+  } else {
+    pEntry = syncEntryBuildFromRpcMsg(pMsg, term, index);
+  }
+
+  if (ths->state == TAOS_SYNC_STATE_LEADER) {
+    if (pRetIndex) {
+      (*pRetIndex) = index;
+    }
+
+    return syncNodeAppend(ths, pEntry);
+  }
+
+  return -1;
+}
+
+int32_t syncNodeOnClientRequestOld(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIndex) {
+  sNTrace(ths, "on client request");
+
   int32_t ret = 0;
   int32_t code = 0;
 
@@ -2395,7 +2666,7 @@ int32_t syncNodeOnClientRequest(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIn
         if (h) {
           taosLRUCacheRelease(ths->pLogStore->pCache, h, false);
         } else {
-          syncEntryDestory(pEntry);
+          syncEntryDestroy(pEntry);
         }
 
         return -1;
@@ -2418,7 +2689,7 @@ int32_t syncNodeOnClientRequest(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIn
         if (h) {
           taosLRUCacheRelease(ths->pLogStore->pCache, h, false);
         } else {
-          syncEntryDestory(pEntry);
+          syncEntryDestroy(pEntry);
         }
 
         return -1;
@@ -2453,7 +2724,7 @@ int32_t syncNodeOnClientRequest(SSyncNode* ths, SRpcMsg* pMsg, SyncIndex* pRetIn
   if (h) {
     taosLRUCacheRelease(ths->pLogStore->pCache, h, false);
   } else {
-    syncEntryDestory(pEntry);
+    syncEntryDestroy(pEntry);
   }
 
   return ret;
@@ -2563,6 +2834,7 @@ bool syncNodeIsOptimizedOneReplica(SSyncNode* ths, SRpcMsg* pMsg) {
 }
 
 int32_t syncNodeDoCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, uint64_t flag) {
+  ASSERT(false);
   if (beginIndex > endIndex) {
     return 0;
   }
@@ -2676,7 +2948,7 @@ int32_t syncNodeDoCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endInde
         if (h) {
           taosLRUCacheRelease(pCache, h, false);
         } else {
-          syncEntryDestory(pEntry);
+          syncEntryDestroy(pEntry);
         }
       }
     }

source/libs/sync/src/syncMessage.c

@@ -16,6 +16,7 @@
 #define _DEFAULT_SOURCE
 #include "syncMessage.h"
 #include "syncRaftEntry.h"
+#include "syncRaftStore.h"
 
 int32_t syncBuildTimeout(SRpcMsg* pMsg, ESyncTimeoutType timeoutType, uint64_t logicClock, int32_t timerMS,
                          SSyncNode* pNode) {
@@ -153,6 +154,34 @@ int32_t syncBuildAppendEntriesReply(SRpcMsg* pMsg, int32_t vgId) {
   return 0;
 }
 
+int32_t syncBuildAppendEntriesFromRaftLog(SSyncNode* pNode, SSyncRaftEntry* pEntry, SyncTerm prevLogTerm,
+                                          SRpcMsg* pRpcMsg) {
+  uint32_t dataLen = pEntry->bytes;
+  uint32_t bytes = sizeof(SyncAppendEntries) + dataLen;
+  pRpcMsg->contLen = bytes;
+  pRpcMsg->pCont = rpcMallocCont(pRpcMsg->contLen);
+  if (pRpcMsg->pCont == NULL) {
+    terrno = TSDB_CODE_OUT_OF_MEMORY;
+    return -1;
+  }
+
+  SyncAppendEntries* pMsg = pRpcMsg->pCont;
+  pMsg->bytes = pRpcMsg->contLen;
+  pMsg->msgType = pRpcMsg->msgType = TDMT_SYNC_APPEND_ENTRIES;
+  pMsg->dataLen = dataLen;
+
+  (void)memcpy(pMsg->data, pEntry, dataLen);
+
+  pMsg->prevLogIndex = pEntry->index - 1;
+  pMsg->prevLogTerm = prevLogTerm;
+  pMsg->vgId = pNode->vgId;
+  pMsg->srcId = pNode->myRaftId;
+  pMsg->term = pNode->pRaftStore->currentTerm;
+  pMsg->commitIndex = pNode->commitIndex;
+  pMsg->privateTerm = 0;
+  return 0;
+}
+
 int32_t syncBuildHeartbeat(SRpcMsg* pMsg, int32_t vgId) {
   int32_t bytes = sizeof(SyncHeartbeat);
   pMsg->pCont = rpcMallocCont(bytes);

source/libs/sync/src/syncRaftCfg.c

@@ -395,4 +395,3 @@ int32_t raftCfgFromStr(const char *s, SRaftCfg *pRaftCfg) {
   cJSON_Delete(pRoot);
   return 0;
 }
-

source/libs/sync/src/syncRaftEntry.c

@@ -20,7 +20,7 @@
 
 SSyncRaftEntry* syncEntryBuild(int32_t dataLen) {
   int32_t         bytes = sizeof(SSyncRaftEntry) + dataLen;
-  SSyncRaftEntry* pEntry = taosMemoryMalloc(bytes);
+  SSyncRaftEntry* pEntry = taosMemoryCalloc(1, bytes);
   if (pEntry == NULL) {
     terrno = TSDB_CODE_OUT_OF_MEMORY;
     return NULL;
@@ -89,7 +89,7 @@ SSyncRaftEntry* syncEntryBuildNoop(SyncTerm term, SyncIndex index, int32_t vgId)
   return pEntry;
 }
 
-void syncEntryDestory(SSyncRaftEntry* pEntry) {
+void syncEntryDestroy(SSyncRaftEntry* pEntry) {
   if (pEntry != NULL) {
     sTrace("free entry: %p", pEntry);
     taosMemoryFree(pEntry);
@@ -265,7 +265,7 @@ static int cmpFn(const void* p1, const void* p2) { return memcmp(p1, p2, sizeof(
 
 static void freeRaftEntry(void* param) {
   SSyncRaftEntry* pEntry = (SSyncRaftEntry*)param;
-  syncEntryDestory(pEntry);
+  syncEntryDestroy(pEntry);
 }
 
 SRaftEntryCache* raftEntryCacheCreate(SSyncNode* pSyncNode, int32_t maxCount) {
@@ -395,7 +395,7 @@ int32_t raftEntryCacheClear(struct SRaftEntryCache* pCache, int32_t count) {
       SSkipListNode* pNode = tSkipListIterGet(pIter);
       ASSERT(pNode != NULL);
       SSyncRaftEntry* pEntry = (SSyncRaftEntry*)SL_GET_NODE_DATA(pNode);
-      syncEntryDestory(pEntry);
+      syncEntryDestroy(pEntry);
       ++returnCnt;
     }
     tSkipListDestroyIter(pIter);
@@ -424,7 +424,7 @@ int32_t raftEntryCacheClear(struct SRaftEntryCache* pCache, int32_t count) {
       ++returnCnt;
       SSyncRaftEntry* pEntry = (SSyncRaftEntry*)SL_GET_NODE_DATA(pNode);
 
-      // syncEntryDestory(pEntry);
+      // syncEntryDestroy(pEntry);
       taosRemoveRef(pCache->refMgr, pEntry->rid);
     }
     tSkipListDestroyIter(pIter);

source/libs/sync/src/syncRaftLog.c

@@ -201,9 +201,8 @@ static int32_t raftLogAppendEntry(struct SSyncLogStore* pLogStore, SSyncRaftEntr
   syncMeta.isWeek = pEntry->isWeak;
   syncMeta.seqNum = pEntry->seqNum;
   syncMeta.term = pEntry->term;
-
   int64_t tsWriteBegin = taosGetTimestampNs();
-  index = walAppendLog(pWal, pEntry->originalRpcType, syncMeta, pEntry->data, pEntry->dataLen);
+  index = walAppendLog(pWal, pEntry->index, pEntry->originalRpcType, syncMeta, pEntry->data, pEntry->dataLen);
   int64_t tsWriteEnd = taosGetTimestampNs();
   int64_t tsElapsed = tsWriteEnd - tsWriteBegin;
 
@@ -217,7 +216,8 @@ static int32_t raftLogAppendEntry(struct SSyncLogStore* pLogStore, SSyncRaftEntr
             pEntry->index, err, err, errStr, sysErr, sysErrStr);
     return -1;
   }
-  pEntry->index = index;
+
+  ASSERT(pEntry->index == index);
 
   sNTrace(pData->pSyncNode, "write index:%" PRId64 ", type:%s, origin type:%s, elapsed:%" PRId64, pEntry->index,
           TMSG_INFO(pEntry->msgType), TMSG_INFO(pEntry->originalRpcType), tsElapsed);
@@ -344,8 +344,6 @@ static int32_t raftLogTruncate(struct SSyncLogStore* pLogStore, SyncIndex fromIn
     const char* sysErrStr = strerror(errno);
     sError("vgId:%d, wal truncate error, from-index:%" PRId64 ", err:%d %X, msg:%s, syserr:%d, sysmsg:%s",
            pData->pSyncNode->vgId, fromIndex, err, err, errStr, sysErr, sysErrStr);
-
-    // ASSERT(0);
   }
 
   // event log
@@ -397,8 +395,6 @@ int32_t raftLogUpdateCommitIndex(SSyncLogStore* pLogStore, SyncIndex index) {
     const char* sysErrStr = strerror(errno);
     sError("vgId:%d, wal update commit index error, index:%" PRId64 ", err:%d %X, msg:%s, syserr:%d, sysmsg:%s",
            pData->pSyncNode->vgId, index, err, err, errStr, sysErr, sysErrStr);
-
-    ASSERT(0);
     return -1;
   }
   return 0;

source/libs/sync/src/syncReplication.c

@@ -16,13 +16,11 @@
 #define _DEFAULT_SOURCE
 #include "syncReplication.h"
 #include "syncIndexMgr.h"
+#include "syncPipeline.h"
 #include "syncRaftEntry.h"
 #include "syncRaftStore.h"
 #include "syncUtil.h"
 
-static int32_t syncNodeSendAppendEntries(SSyncNode* pNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg);
-static int32_t syncNodeMaybeSendAppendEntries(SSyncNode* pNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg);
-
 // TLA+ Spec
 // AppendEntries(i, j) ==
 //    /\ i /= j
@@ -48,7 +46,10 @@ static int32_t syncNodeMaybeSendAppendEntries(SSyncNode* pNode, const SRaftId* d
 //                mdest          |-> j])
 //    /\ UNCHANGED <<serverVars, candidateVars, leaderVars, logVars>>
 
+int32_t syncNodeMaybeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg);
+
 int32_t syncNodeReplicateOne(SSyncNode* pSyncNode, SRaftId* pDestId, bool snapshot) {
+  ASSERT(false && "deprecated");
   // next index
   SyncIndex nextIndex = syncIndexMgrGetIndex(pSyncNode->pNextIndex, pDestId);
 
@@ -117,7 +118,7 @@ int32_t syncNodeReplicateOne(SSyncNode* pSyncNode, SRaftId* pDestId, bool snapsh
   if (h) {
     taosLRUCacheRelease(pCache, h, false);
   } else {
-    syncEntryDestory(pEntry);
+    syncEntryDestroy(pEntry);
   }
 
   // prepare msg
@@ -136,7 +137,29 @@ int32_t syncNodeReplicateOne(SSyncNode* pSyncNode, SRaftId* pDestId, bool snapsh
   return 0;
 }
 
-int32_t syncNodeReplicate(SSyncNode* pSyncNode) {
+int32_t syncNodeReplicate(SSyncNode* pNode) {
+  SSyncLogBuffer* pBuf = pNode->pLogBuf;
+  taosThreadMutexLock(&pBuf->mutex);
+  int32_t ret = syncNodeReplicateWithoutLock(pNode);
+  taosThreadMutexUnlock(&pBuf->mutex);
+  return ret;
+}
+
+int32_t syncNodeReplicateWithoutLock(SSyncNode* pNode) {
+  if (pNode->state != TAOS_SYNC_STATE_LEADER || pNode->replicaNum == 1) {
+    return -1;
+  }
+  for (int32_t i = 0; i < pNode->replicaNum; i++) {
+    if (syncUtilSameId(&pNode->replicasId[i], &pNode->myRaftId)) {
+      continue;
+    }
+    SSyncLogReplMgr* pMgr = pNode->logReplMgrs[i];
+    (void)syncLogReplMgrReplicateOnce(pMgr, pNode);
+  }
+  return 0;
+}
+
+int32_t syncNodeReplicateOld(SSyncNode* pSyncNode) {
   if (pSyncNode->state != TAOS_SYNC_STATE_LEADER) {
     return -1;
   }
@@ -159,6 +182,14 @@ int32_t syncNodeReplicate(SSyncNode* pSyncNode) {
 }
 
 int32_t syncNodeSendAppendEntries(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) {
+  SyncAppendEntries* pMsg = pRpcMsg->pCont;
+  int32_t ret = 0;
+  pMsg->destId = *destRaftId;
+  syncNodeSendMsgById(destRaftId, pSyncNode, pRpcMsg);
+  return 0;
+}
+
+int32_t syncNodeSendAppendEntriesOld(SSyncNode* pSyncNode, const SRaftId* destRaftId, SRpcMsg* pRpcMsg) {
   int32_t            ret = 0;
   SyncAppendEntries* pMsg = pRpcMsg->pCont;
   if (pMsg == NULL) {

source/libs/sync/src/syncSnapshot.c

@@ -16,6 +16,7 @@
 #define _DEFAULT_SOURCE
 #include "syncSnapshot.h"
 #include "syncIndexMgr.h"
+#include "syncPipeline.h"
 #include "syncRaftCfg.h"
 #include "syncRaftLog.h"
 #include "syncRaftStore.h"
@@ -273,6 +274,11 @@ int32_t syncNodeStartSnapshot(SSyncNode *pSyncNode, SRaftId *pDestId) {
     return 1;
   }
 
+  char     host[64];
+  uint16_t port;
+  syncUtilU642Addr(pDestId->addr, host, sizeof(host), &port);
+  sInfo("vgId:%d, start snapshot for peer: %s:%d", pSyncNode->vgId, host, port);
+
   code = snapshotSenderStart(pSender);
   if (code != 0) {
     sNError(pSyncNode, "snapshot sender start error");
@@ -372,7 +378,10 @@ int32_t snapshotReceiverStartWriter(SSyncSnapshotReceiver *pReceiver, SyncSnapsh
 }
 
 int32_t snapshotReceiverStart(SSyncSnapshotReceiver *pReceiver, SyncSnapshotSend *pPreMsg) {
-  ASSERT(!snapshotReceiverIsStart(pReceiver));
+  if (snapshotReceiverIsStart(pReceiver)) {
+    sWarn("vgId:%d, snapshot receiver has started.", pReceiver->pSyncNode->vgId);
+    return 0;
+  }
 
   pReceiver->start = true;
   pReceiver->ack = SYNC_SNAPSHOT_SEQ_PRE_SNAPSHOT;
@@ -742,6 +751,7 @@ int32_t syncNodeOnSnapshot(SSyncNode *pSyncNode, const SRpcMsg *pRpcMsg) {
 
       } else if (pMsg->seq == SYNC_SNAPSHOT_SEQ_END) {
         syncNodeOnSnapshotEnd(pSyncNode, pMsg);
+        (void)syncLogBufferReInit(pSyncNode->pLogBuf, pSyncNode);
 
       } else if (pMsg->seq == SYNC_SNAPSHOT_SEQ_FORCE_CLOSE) {
         // force close, no response
@@ -873,11 +883,10 @@ int32_t syncNodeOnSnapshotReply(SSyncNode *pSyncNode, const SRpcMsg *pRpcMsg) {
       // receive ack is finish, close sender
       if (pMsg->ack == SYNC_SNAPSHOT_SEQ_END) {
         snapshotSenderStop(pSender, true);
-
+        SSyncLogReplMgr* pMgr = syncNodeGetLogReplMgr(pSyncNode, &pMsg->srcId);
-        // update next-index
+        if (pMgr) {
-        syncIndexMgrSetIndex(pSyncNode->pNextIndex, &(pMsg->srcId), pMsg->lastIndex + 1);
+            syncLogReplMgrReset(pMgr);
-        syncNodeReplicateOne(pSyncNode, &(pMsg->srcId), false);
+        }
-
         return 0;
       }
 

source/libs/sync/src/syncUtil.c

@@ -278,20 +278,21 @@ void syncPrintNodeLog(const char* flags, ELogLevel level, int32_t dflag, SSyncNo
   terrno = errCode;
 
   if (pNode != NULL && pNode->pRaftCfg != NULL) {
-    taosPrintLog(
+    taosPrintLog(flags, level, dflag,
-        flags, level, dflag,
                  "vgId:%d, sync %s "
                  "%s"
-        ", tm:%" PRIu64 ", cmt:%" PRId64 ", fst:%" PRId64 ", lst:%" PRId64 ", min:%" PRId64 ", snap:%" PRId64
+                 ", term:%" PRIu64 ", commit-index:%" PRId64 ", first-ver:%" PRId64 ", last-ver:%" PRId64
-        ", snap-tm:%" PRIu64
+                 ", min:%" PRId64 ", snap:%" PRId64 ", snap-term:%" PRIu64
-        ", elt-num:%d, bl-num:%d, cc-num:%d, hit:%d, mis:%d, hb-slow:%d, hbr-slow:%d, aq:%d, snaping:%" PRId64
+                 ", elect-times:%d, as-leader-times:%d, cfg-ch-times:%d, hit:%d, mis:%d, hb-slow:%d, hbr-slow:%d, "
-        ", r-num:%d, lcfg:%" PRId64 ", chging:%d, rsto:%d, dquorum:%d, elt:%" PRId64 ", hb:%" PRId64 ", %s, %s, %s, %s",
+                 "aq-items:%d, snaping:%" PRId64 ", replicas:%d, last-cfg:%" PRId64
-        pNode->vgId, syncStr(pNode->state), eventLog, currentTerm, pNode->commitIndex, logBeginIndex, logLastIndex,
+                 ", chging:%d, restore:%d, quorum:%d, elect-lc-timer:%" PRId64 ", hb:%" PRId64 ", %s, %s, %s, %s",
-        pNode->minMatchIndex, snapshot.lastApplyIndex, snapshot.lastApplyTerm, pNode->electNum, pNode->becomeLeaderNum,
+                 pNode->vgId, syncStr(pNode->state), eventLog, currentTerm, pNode->commitIndex, logBeginIndex,
-        pNode->configChangeNum, cacheHit, cacheMiss, pNode->hbSlowNum, pNode->hbrSlowNum, aqItems,
+                 logLastIndex, pNode->minMatchIndex, snapshot.lastApplyIndex, snapshot.lastApplyTerm, pNode->electNum,
-        pNode->snapshottingIndex, pNode->replicaNum, pNode->pRaftCfg->lastConfigIndex, pNode->changing,
+                 pNode->becomeLeaderNum, pNode->configChangeNum, cacheHit, cacheMiss, pNode->hbSlowNum,
-        pNode->restoreFinish, quorum, pNode->electTimerLogicClock, pNode->heartbeatTimerLogicClockUser, peerStr, cfgStr,
+                 pNode->hbrSlowNum, aqItems, pNode->snapshottingIndex, pNode->replicaNum,
-        hbTimeStr, hbrTimeStr);
+                 pNode->pRaftCfg->lastConfigIndex, pNode->changing, pNode->restoreFinish, quorum,
+                 pNode->electTimerLogicClock, pNode->heartbeatTimerLogicClockUser, peerStr, cfgStr, hbTimeStr,
+                 hbrTimeStr);
   }
 }
 
@@ -427,7 +428,7 @@ void syncLogSendAppendEntriesReply(SSyncNode* pSyncNode, const SyncAppendEntries
   sNTrace(pSyncNode,
           "send sync-append-entries-reply to %s:%d, {term:%" PRId64 ", pterm:%" PRId64
           ", success:%d, lsend-index:%" PRId64 ", match:%" PRId64 "}, %s",
-          host, port, pMsg->term, pMsg->privateTerm, pMsg->success, pMsg->lastSendIndex, pMsg->matchIndex, s);
+          host, port, pMsg->term, pMsg->lastMatchTerm, pMsg->success, pMsg->lastSendIndex, pMsg->matchIndex, s);
 }
 
 void syncLogRecvAppendEntriesReply(SSyncNode* pSyncNode, const SyncAppendEntriesReply* pMsg, const char* s) {
@@ -440,7 +441,7 @@ void syncLogRecvAppendEntriesReply(SSyncNode* pSyncNode, const SyncAppendEntries
   sNTrace(pSyncNode,
           "recv sync-append-entries-reply from %s:%d {term:%" PRId64 ", pterm:%" PRId64
           ", success:%d, lsend-index:%" PRId64 ", match:%" PRId64 "}, %s",
-          host, port, pMsg->term, pMsg->privateTerm, pMsg->success, pMsg->lastSendIndex, pMsg->matchIndex, s);
+          host, port, pMsg->term, pMsg->lastMatchTerm, pMsg->success, pMsg->lastSendIndex, pMsg->matchIndex, s);
 }
 
 void syncLogSendHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, bool printX, int64_t timerElapsed,
@@ -464,7 +465,7 @@ void syncLogSendHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, bool
   }
 }
 
-void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64_t timeDiff) {
+void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64_t timeDiff, const char* s) {
   if (timeDiff > SYNC_HEARTBEAT_SLOW_MS) {
     pSyncNode->hbSlowNum++;
 
@@ -473,8 +474,8 @@ void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64
     syncUtilU642Addr(pMsg->srcId.addr, host, sizeof(host), &port);
     sNInfo(pSyncNode,
            "recv sync-heartbeat from %s:%d slow {term:%" PRId64 ", cmt:%" PRId64 ", min-match:%" PRId64 ", ts:%" PRId64
-           "}, net elapsed:%" PRId64,
+           "}, %s, net elapsed:%" PRId64,
-           host, port, pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, timeDiff);
+           host, port, pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, s, timeDiff);
   }
 
   if (!(sDebugFlag & DEBUG_TRACE)) return;
@@ -484,8 +485,8 @@ void syncLogRecvHeartbeat(SSyncNode* pSyncNode, const SyncHeartbeat* pMsg, int64
   syncUtilU642Addr(pMsg->srcId.addr, host, sizeof(host), &port);
   sNTrace(pSyncNode,
           "recv sync-heartbeat from %s:%d {term:%" PRId64 ", cmt:%" PRId64 ", min-match:%" PRId64 ", ts:%" PRId64
-          "}, net elapsed:%" PRId64,
+          "}, %s, net elapsed:%" PRId64,
-          host, port, pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, timeDiff);
+          host, port, pMsg->term, pMsg->commitIndex, pMsg->minMatchIndex, pMsg->timeStamp, s, timeDiff);
 }
 
 void syncLogSendHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* pMsg, const char* s) {
@@ -499,7 +500,7 @@ void syncLogSendHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* p
           pMsg->term, pMsg->timeStamp, s);
 }
 
-void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* pMsg, int64_t timeDiff) {
+void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* pMsg, int64_t timeDiff, const char* s) {
   if (timeDiff > SYNC_HEARTBEAT_REPLY_SLOW_MS) {
     pSyncNode->hbrSlowNum++;
 
@@ -507,8 +508,8 @@ void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* p
     uint16_t port;
     syncUtilU642Addr(pMsg->srcId.addr, host, sizeof(host), &port);
     sNTrace(pSyncNode,
-            "recv sync-heartbeat-reply from %s:%d slow {term:%" PRId64 ", ts:%" PRId64 "}, net elapsed:%" PRId64, host,
+            "recv sync-heartbeat-reply from %s:%d slow {term:%" PRId64 ", ts:%" PRId64 "}, %s, net elapsed:%" PRId64,
-            port, pMsg->term, pMsg->timeStamp, timeDiff);
+            host, port, pMsg->term, pMsg->timeStamp, s, timeDiff);
   }
 
   if (!(sDebugFlag & DEBUG_TRACE)) return;
@@ -516,8 +517,9 @@ void syncLogRecvHeartbeatReply(SSyncNode* pSyncNode, const SyncHeartbeatReply* p
   char     host[64];
   uint16_t port;
   syncUtilU642Addr(pMsg->srcId.addr, host, sizeof(host), &port);
-  sNTrace(pSyncNode, "recv sync-heartbeat-reply from %s:%d {term:%" PRId64 ", ts:%" PRId64 "}, net elapsed:%" PRId64,
+  sNTrace(pSyncNode,
-          host, port, pMsg->term, pMsg->timeStamp, timeDiff);
+          "recv sync-heartbeat-reply from %s:%d {term:%" PRId64 ", ts:%" PRId64 "}, %s, net elapsed:%" PRId64, host,
+          port, pMsg->term, pMsg->timeStamp, s, timeDiff);
 }
 
 void syncLogSendSyncPreSnapshot(SSyncNode* pSyncNode, const SyncPreSnapshot* pMsg, const char* s) {

source/libs/sync/test/syncAppendEntriesBatchTest.cpp

@@ -1,5 +1,5 @@
-#include "syncTest.h"
 #include "syncBatch.h"
+#include "syncTest.h"
 
 void logTest() {
   sTrace("--- sync log test: trace");