xxq250
/
homework-jianmu
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

Merge branch '3.0' into fix/TD-27455

This commit is contained in:
WANG MINGMING 2023-11-30 09:37:23 +08:00 committed by GitHub
parent 4f168b4b2d 130614c938
commit 9b72d890c8
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
56 changed files with 4387 additions and 4279 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Jenkinsfile2
View File

@ -355,7 +355,7 @@ pipeline {
}
parallel {
stage('check docs') {
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_52 || worker03 || slave215 || slave217 || slave219 || Mac_catalina "}
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_50 || slave1_52 || slave1_59 || slave1_63 || worker03 || slave215 || slave217 || slave219 || Mac_catalina "}
steps {
check_docs()
}
@ -401,7 +401,7 @@ pipeline {
}
}
stage('linux test') {
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_52 || worker03 || slave215 || slave217 || slave219 "}
agent{label " slave1_47 || slave1_48 || slave1_49 || slave1_50 || slave1_52 || slave1_59 || slave1_63 || worker03 || slave215 || slave217 || slave219 "}
options { skipDefaultCheckout() }
when {
changeRequest()

17
cmake/cmake.define
View File

@ -1,5 +1,5 @@
cmake_minimum_required(VERSION 3.0)
set(CMAKE_VERBOSE_MAKEFILE FALSE)
set(CMAKE_VERBOSE_MAKEFILE TRUE)
set(TD_BUILD_TAOSA_INTERNAL FALSE)
#set output directory
@ -159,6 +159,7 @@ ELSE ()
CHECK_C_COMPILER_FLAG("-mavx2" COMPILER_SUPPORT_AVX2)
CHECK_C_COMPILER_FLAG("-mavx512f" COMPILER_SUPPORT_AVX512F)
CHECK_C_COMPILER_FLAG("-mavx512vbmi" COMPILER_SUPPORT_AVX512BMI)
CHECK_C_COMPILER_FLAG("-mavx512vl" COMPILER_SUPPORT_AVX512VL)
IF (COMPILER_SUPPORT_SSE42)
SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -msse4.2")
@ -166,11 +167,11 @@ ELSE ()
ENDIF()
IF ("${SIMD_SUPPORT}" MATCHES "true")
IF (COMPILER_SUPPORT_FMA)
IF (COMPILER_SUPPORT_FMA)
SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mfma")
SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mfma")
ENDIF()
IF (COMPILER_SUPPORT_AVX)
ENDIF()
IF (COMPILER_SUPPORT_AVX)
SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mavx")
SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mavx")
ENDIF()
@ -183,7 +184,13 @@ ELSE ()
IF (COMPILER_SUPPORT_AVX512F AND COMPILER_SUPPORT_AVX512BMI)
SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mavx512f -mavx512vbmi")
SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mavx512f -mavx512vbmi")
MESSAGE(STATUS "avx512 supported by gcc")
MESSAGE(STATUS "avx512f/avx512bmi supported by compiler")
ENDIF()
IF (COMPILER_SUPPORT_AVX512VL)
SET(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mavx512vl")
SET(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mavx512vl")
MESSAGE(STATUS "avx512vl supported by compiler")
ENDIF()
ENDIF()

3
include/common/tmsg.h
View File

@ -414,7 +414,8 @@ typedef enum ENodeType {
QUERY_NODE_PHYSICAL_PLAN_STREAM_EVENT,
QUERY_NODE_PHYSICAL_PLAN_HASH_JOIN,
QUERY_NODE_PHYSICAL_PLAN_GROUP_CACHE,
QUERY_NODE_PHYSICAL_PLAN_DYN_QUERY_CTRL
QUERY_NODE_PHYSICAL_PLAN_DYN_QUERY_CTRL,
QUERY_NODE_END
} ENodeType;
typedef struct {

22
include/common/tmsgdef.h
View File

@ -67,7 +67,7 @@ enum {
enum { // WARN: new msg should be appended to segment tail
#endif
TD_NEW_MSG_SEG(TDMT_DND_MSG)
TD_NEW_MSG_SEG(TDMT_DND_MSG) // 0<<8
TD_DEF_MSG_TYPE(TDMT_DND_CREATE_MNODE, "dnode-create-mnode", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_DND_DROP_MNODE, "dnode-drop-mnode", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_DND_CREATE_QNODE, "dnode-create-qnode", NULL, NULL)
@ -87,7 +87,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_DND_ALTER_VNODE_TYPE, "dnode-alter-vnode-type", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_DND_CHECK_VNODE_LEARNER_CATCHUP, "dnode-check-vnode-learner-catchup", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_MND_MSG)
TD_NEW_MSG_SEG(TDMT_MND_MSG) // 1<<8
TD_DEF_MSG_TYPE(TDMT_MND_CONNECT, "connect", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_CREATE_ACCT, "create-acct", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_ALTER_ACCT, "alter-acct", NULL, NULL)
@ -194,7 +194,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_MND_VIEW_META, "view-meta", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_MND_MAX_MSG, "mnd-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_VND_MSG)
TD_NEW_MSG_SEG(TDMT_VND_MSG) // 2<<8
TD_DEF_MSG_TYPE(TDMT_VND_SUBMIT, "submit", SSubmitReq, SSubmitRsp)
TD_DEF_MSG_TYPE(TDMT_VND_CREATE_TABLE, "create-table", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_ALTER_TABLE, "alter-table", NULL, NULL)
@ -243,7 +243,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_VND_DISABLE_WRITE, "vnode-disable-write", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_MAX_MSG, "vnd-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_SCH_MSG)
TD_NEW_MSG_SEG(TDMT_SCH_MSG) // 3<<8
TD_DEF_MSG_TYPE(TDMT_SCH_QUERY, "query", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SCH_MERGE_QUERY, "merge-query", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SCH_QUERY_CONTINUE, "query-continue", NULL, NULL)
@ -258,13 +258,13 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_SCH_MAX_MSG, "sch-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_STREAM_MSG)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DEPLOY, "stream-task-deploy", SStreamTaskDeployReq, SStreamTaskDeployRsp)
TD_NEW_MSG_SEG(TDMT_STREAM_MSG) //4 << 8
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DEPLOY, "stream-task-deploy", SStreamTaskDeployReq, SStreamTaskDeployRsp) //1025 1026
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_DROP, "stream-task-drop", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_RUN, "stream-task-run", NULL, NULL)
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_RETRIEVE, "stream-retrieve", NULL, NULL) //1035 1036
TD_DEF_MSG_TYPE(TDMT_STREAM_TASK_CHECKPOINT_READY, "stream-checkpoint-ready", 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)
@ -274,10 +274,10 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_STREAM_HTASK_DROP, "stream-htask-drop", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_STREAM_MAX_MSG, "stream-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_MON_MSG)
TD_NEW_MSG_SEG(TDMT_MON_MSG) //5 << 8
TD_DEF_MSG_TYPE(TDMT_MON_MAX_MSG, "monitor-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_SYNC_MSG)
TD_NEW_MSG_SEG(TDMT_SYNC_MSG) //6 << 8
TD_DEF_MSG_TYPE(TDMT_SYNC_TIMEOUT, "sync-timer", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SYNC_TIMEOUT_ELECTION, "sync-elect", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SYNC_PING_REPLY, "sync-ping-reply", NULL, NULL) // no longer used
@ -308,7 +308,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_SYNC_MAX_MSG, "sync-max", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_SYNC_FORCE_FOLLOWER, "sync-force-become-follower", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_VND_STREAM_MSG)
TD_NEW_MSG_SEG(TDMT_VND_STREAM_MSG) //7 << 8
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_SCAN_HISTORY, "vnode-stream-scan-history", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_SCAN_HISTORY_FINISH, "vnode-stream-scan-history-finish", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_CHECK_POINT_SOURCE, "vnode-stream-checkpoint-source", NULL, NULL)
@ -317,7 +317,7 @@ enum { // WARN: new msg should be appended to segment tail
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_TASK_CHECK, "vnode-stream-task-check", NULL, NULL)
TD_DEF_MSG_TYPE(TDMT_VND_STREAM_MAX_MSG, "vnd-stream-max", NULL, NULL)
TD_NEW_MSG_SEG(TDMT_VND_TMQ_MSG)
TD_NEW_MSG_SEG(TDMT_VND_TMQ_MSG) //8 << 8
TD_DEF_MSG_TYPE(TDMT_VND_TMQ_SUBSCRIBE, "vnode-tmq-subscribe", SMqRebVgReq, SMqRebVgRsp)
TD_DEF_MSG_TYPE(TDMT_VND_TMQ_DELETE_SUB, "vnode-tmq-delete-sub", SMqVDeleteReq, SMqVDeleteRsp)
TD_DEF_MSG_TYPE(TDMT_VND_TMQ_COMMIT_OFFSET, "vnode-tmq-commit-offset", STqOffset, STqOffset)

18
include/dnode/vnode/stream.h
View File

@ -1,18 +0,0 @@
//
// Created by mingming wanng on 2023/11/15.
//
#ifndef TDENGINE_STREAM_H
#define TDENGINE_STREAM_H
#define STREAM_EXEC_EXTRACT_DATA_IN_WAL_ID (-1)
#define STREAM_EXEC_START_ALL_TASKS_ID (-2)
#define STREAM_EXEC_RESTART_ALL_TASKS_ID (-3)
typedef struct STaskUpdateEntry {
int64_t streamId;
int32_t taskId;
int32_t transId;
} STaskUpdateEntry;
#endif // TDENGINE_STREAM_H

36
include/dnode/vnode/tqCommon.h Normal file
View File

@ -0,0 +1,36 @@
/*
* 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 TDENGINE_TQ_COMMON_H
#define TDENGINE_TQ_COMMON_H
// message process
int32_t tqStreamTaskStartAsync(SStreamMeta* pMeta, SMsgCb* cb, bool restart);
int32_t tqStreamTaskProcessUpdateReq(SStreamMeta* pMeta, SMsgCb* cb, SRpcMsg* pMsg, bool restored);
int32_t tqStreamTaskProcessDispatchReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessDispatchRsp(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessRetrieveReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessScanHistoryFinishReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessScanHistoryFinishRsp(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessCheckReq(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessCheckRsp(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader);
int32_t tqStreamTaskProcessCheckpointReadyMsg(SStreamMeta* pMeta, SRpcMsg* pMsg);
int32_t tqStreamTaskProcessDeployReq(SStreamMeta* pMeta, int64_t sversion, char* msg, int32_t msgLen, bool isLeader, bool restored);
int32_t tqStreamTaskProcessDropReq(SStreamMeta* pMeta, char* msg, int32_t msgLen);
int32_t tqStreamTaskProcessRunReq(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader);
int32_t startStreamTasks(SStreamMeta* pMeta);
int32_t resetStreamTaskStatus(SStreamMeta* pMeta);
#endif // TDENGINE_TQ_COMMON_H

1
include/libs/nodes/nodes.h
View File

@ -110,6 +110,7 @@ int64_t nodesMakeAllocatorWeakRef(int64_t allocatorId);
int64_t nodesReleaseAllocatorWeakRef(int64_t allocatorId);
void nodesDestroyAllocator(int64_t allocatorId);
int32_t getNodeSize(ENodeType type);
SNode* nodesMakeNode(ENodeType type);
void nodesDestroyNode(SNode* pNode);
void nodesFree(void* p);

6
include/libs/stream/tstream.h
View File

@ -50,6 +50,10 @@ extern "C" {
(_t)->hTaskInfo.id.streamId = 0; \
} while (0)
#define STREAM_EXEC_EXTRACT_DATA_IN_WAL_ID (-1)
#define STREAM_EXEC_START_ALL_TASKS_ID (-2)
#define STREAM_EXEC_RESTART_ALL_TASKS_ID (-3)
typedef struct SStreamTask SStreamTask;
typedef struct SStreamQueue SStreamQueue;
typedef struct SStreamTaskSM SStreamTaskSM;
@ -664,6 +668,8 @@ typedef struct STaskStatusEntry {
int32_t relatedHTask; // has related fill-history task
int64_t activeCheckpointId; // current active checkpoint id
bool checkpointFailed; // denote if the checkpoint is failed or not
bool inputQChanging; // inputQ is changing or not
int64_t inputQUnchangeCounter;
double inputQUsed; // in MiB
double inputRate;
double sinkQuota; // existed quota size for sink task

1
include/os/osFile.h
View File

@ -65,6 +65,7 @@ typedef struct TdFile *TdFilePtr;
#define TD_FILE_STREAM 0x0100 // Only support taosFprintfFile, taosGetLineFile, taosEOFFile
#define TD_FILE_WRITE_THROUGH 0x0200
#define TD_FILE_CLOEXEC 0x0400
TdFilePtr taosOpenFile(const char *path, int32_t tdFileOptions);
TdFilePtr taosCreateFile(const char *path, int32_t tdFileOptions);

2
include/util/tcompression.h
View File

@ -139,6 +139,8 @@ int32_t getWordLength(char type);
int32_t tsDecompressIntImpl_Hw(const char *const input, const int32_t nelements, char *const output, const char type);
int32_t tsDecompressFloatImplAvx512(const char *const input, const int32_t nelements, char *const output);
int32_t tsDecompressFloatImplAvx2(const char *const input, const int32_t nelements, char *const output);
int32_t tsDecompressTimestampAvx512(const char* const input, const int32_t nelements, char *const output, bool bigEndian);
int32_t tsDecompressTimestampAvx2(const char* const input, const int32_t nelements, char *const output, bool bigEndian);
/*************************************************************************
* STREAM COMPRESSION

1
packaging/tools/set_core.sh
View File

@ -38,3 +38,4 @@ source /etc/profile
${csudo}mkdir -p ${corePath} ||:
${csudo}sysctl -w kernel.core_pattern=${corePath}/core-%e-%p ||:
${csudo}echo "${corePath}/core-%e-%p" | ${csudo}tee /proc/sys/kernel/core_pattern ||:
${csudo}echo "kernel.core_pattern = ${corePath}/core_%e-%p" >> /etc/sysctl.conf ||:

1
source/client/src/clientMain.c
View File

@ -876,6 +876,7 @@ int taos_get_current_db(TAOS *taos, char *database, int len, int *required) {
code = 0;
}
taosThreadMutexUnlock(&pTscObj->mutex);
releaseTscObj(*(int64_t *)taos);
return code;
}

102
source/client/test/clientTests.cpp
View File

@ -126,9 +126,9 @@ void queryCallback(void* param, void* res, int32_t code) {
taos_fetch_raw_block_a(res, fetchCallback, param);
}
void createNewTable(TAOS* pConn, int32_t index) {
void createNewTable(TAOS* pConn, int32_t index, int32_t numOfRows, int64_t startTs, const char* pVarchar) {
char str[1024] = {0};
sprintf(str, "create table tu%d using st2 tags(%d)", index, index);
sprintf(str, "create table if not exists tu%d using st2 tags(%d)", index, index);
TAOS_RES* pRes = taos_query(pConn, str);
if (taos_errno(pRes) != 0) {
@ -136,22 +136,43 @@ void createNewTable(TAOS* pConn, int32_t index) {
}
taos_free_result(pRes);
for (int32_t i = 0; i < 10000; i += 20) {
char sql[1024] = {0};
sprintf(sql,
"insert into tu%d values(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
"(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
"(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
"(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)",
index, i, i, i + 1, i + 1, i + 2, i + 2, i + 3, i + 3, i + 4, i + 4, i + 5, i + 5, i + 6, i + 6, i + 7,
i + 7, i + 8, i + 8, i + 9, i + 9, i + 10, i + 10, i + 11, i + 11, i + 12, i + 12, i + 13, i + 13, i + 14,
i + 14, i + 15, i + 15, i + 16, i + 16, i + 17, i + 17, i + 18, i + 18, i + 19, i + 19);
TAOS_RES* p = taos_query(pConn, sql);
if (taos_errno(p) != 0) {
printf("failed to insert data, reason:%s\n", taos_errstr(p));
}
if (startTs == 0) {
for (int32_t i = 0; i < numOfRows; i += 20) {
char sql[1024] = {0};
sprintf(sql,
"insert into tu%d values(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
"(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
"(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)"
"(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)(now+%da, %d)",
index, i, i, i + 1, i + 1, i + 2, i + 2, i + 3, i + 3, i + 4, i + 4, i + 5, i + 5, i + 6, i + 6, i + 7,
i + 7, i + 8, i + 8, i + 9, i + 9, i + 10, i + 10, i + 11, i + 11, i + 12, i + 12, i + 13, i + 13, i + 14,
i + 14, i + 15, i + 15, i + 16, i + 16, i + 17, i + 17, i + 18, i + 18, i + 19, i + 19);
TAOS_RES* p = taos_query(pConn, sql);
if (taos_errno(p) != 0) {
printf("failed to insert data, reason:%s\n", taos_errstr(p));
}
taos_free_result(p);
taos_free_result(p);
}
} else {
for (int32_t i = 0; i < numOfRows; i += 20) {
char sql[1024*50] = {0};
sprintf(sql,
"insert into tu%d values(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, "
"%d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, "
"'%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')(%ld, %d, '%s')",
index, startTs, i, pVarchar, startTs + 1, i + 1, pVarchar, startTs + 2, i + 2, pVarchar, startTs + 3, i + 3, pVarchar, startTs + 4, i + 4,
pVarchar, startTs + 5, i + 5, pVarchar, startTs + 6, i + 6, pVarchar, startTs + 7, i + 7, pVarchar, startTs + 8, i + 8, pVarchar, startTs + 9, i + 9,
pVarchar, startTs + 10, i + 10, pVarchar, startTs + 11, i + 11, pVarchar, startTs + 12, i + 12, pVarchar, startTs + 13, i + 13, pVarchar, startTs + 14,
i + 14, pVarchar, startTs + 15, i + 15, pVarchar, startTs + 16, i + 16, pVarchar, startTs + 17, i + 17, pVarchar, startTs + 18, i + 18,
pVarchar, startTs + 19, i + 19, pVarchar);
TAOS_RES* p = taos_query(pConn, sql);
if (taos_errno(p) != 0) {
printf("failed to insert data, reason:%s\n", taos_errstr(p));
}
taos_free_result(p);
}
}
}
@ -808,14 +829,7 @@ TEST(clientCase, projection_query_tables) {
TAOS_RES* pRes = taos_query(pConn, "use abc1");
taos_free_result(pRes);
pRes = taos_query(pConn, "create stable st1 (ts timestamp, k int) tags(a int)");
if (taos_errno(pRes) != 0) {
printf("failed to create table tu, reason:%s\n", taos_errstr(pRes));
}
taos_free_result(pRes);
pRes = taos_query(pConn, "create stable st2 (ts timestamp, k int) tags(a int)");
pRes = taos_query(pConn, "create stable st2 (ts timestamp, k int, f varchar(4096)) tags(a int)");
if (taos_errno(pRes) != 0) {
printf("failed to create table tu, reason:%s\n", taos_errstr(pRes));
}
@ -828,28 +842,32 @@ TEST(clientCase, projection_query_tables) {
taos_free_result(pRes);
int64_t start = 1685959190000;
const char* pstr =
"abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefgh"
"ijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnop"
"qrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwx"
"yzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdef"
"ghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz!@#$%^&&*&^^%$#@!qQWERTYUIOPASDFGHJKL:"
"QWERTYUIOP{}";
int32_t code = -1;
for(int32_t i = 0; i < 1000000; ++i) {
char t[512] = {0};
for(int32_t i = 0; i < 10000; ++i) {
char str[1024] = {0};
sprintf(str, "create table if not exists tu%d using st2 tags(%d)", i, i);
sprintf(t, "insert into t1 values(now, %d)", i);
while(1) {
void* p = taos_query(pConn, t);
code = taos_errno(p);
taos_free_result(p);
if (code != 0) {
printf("insert data error, retry\n");
} else {
break;
}
TAOS_RES* px = taos_query(pConn, str);
if (taos_errno(px) != 0) {
printf("failed to create table tu, reason:%s\n", taos_errstr(pRes));
}
taos_free_result(px);
}
for(int32_t j = 0; j < 5000; ++j) {
start += 20;
for (int32_t i = 0; i < 10000; ++i) {
createNewTable(pConn, i, 20, start, pstr);
}
}
for (int32_t i = 0; i < 1; ++i) {
printf("create table :%d\n", i);
createNewTable(pConn, i);
}
//
// pRes = taos_query(pConn, "select * from tu");
// if (taos_errno(pRes) != 0) {

2
source/common/src/systable.c
View File

@ -140,7 +140,7 @@ static const SSysDbTableSchema userStbsSchema[] = {
{.name = "columns", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = false},
{.name = "tags", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = false},
{.name = "last_update", .bytes = 8, .type = TSDB_DATA_TYPE_TIMESTAMP, .sysInfo = false},
{.name = "table_comment", .bytes = TSDB_TB_COMMENT_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "table_comment", .bytes = TSDB_TB_COMMENT_LEN - 1 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "watermark", .bytes = 64 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "max_delay", .bytes = 64 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
{.name = "rollup", .bytes = 128 + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},

6
source/dnode/mgmt/mgmt_snode/src/smWorker.c
View File

@ -193,10 +193,6 @@ int32_t smPutNodeMsgToStreamQueue(SSnodeMgmt *pMgmt, SRpcMsg *pMsg) {
SSingleWorker *pWorker = &pMgmt->streamWorker;
dTrace("msg:%p, put into worker %s", pMsg, pWorker->name);
if (pMsg->msgType == TDMT_STREAM_TASK_DISPATCH) {
sndEnqueueStreamDispatch(pMgmt->pSnode, pMsg);
} else {
taosWriteQitem(pWorker->queue, pMsg);
}
taosWriteQitem(pWorker->queue, pMsg);
return 0;
}

6
source/dnode/mgmt/mgmt_vnode/src/vmWorker.c
View File

@ -206,11 +206,7 @@ static int32_t vmPutMsgToQueue(SVnodeMgmt *pMgmt, SRpcMsg *pMsg, EQueueType qtyp
break;
case STREAM_QUEUE:
dGTrace("vgId:%d, msg:%p put into vnode-stream queue", pVnode->vgId, pMsg);
if (pMsg->msgType == TDMT_STREAM_TASK_DISPATCH) {
vnodeEnqueueStreamMsg(pVnode->pImpl, pMsg);
} else {
taosWriteQitem(pVnode->pStreamQ, pMsg);
}
taosWriteQitem(pVnode->pStreamQ, pMsg);
break;
case FETCH_QUEUE:
dGTrace("vgId:%d, msg:%p put into vnode-fetch queue", pVnode->vgId, pMsg);

2
source/dnode/mnode/impl/inc/mndStream.h
View File

@ -52,7 +52,7 @@ int32_t mndDropStreamByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb);
int32_t mndPersistStream(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream);
int32_t mndStreamRegisterTrans(STrans* pTrans, const char* pName, const char* pSrcDb, const char* pDstDb);
bool streamTransConflictOtherTrans(SMnode *pMnode, const char *pSrcDb, const char *pDstDb);
bool streamTransConflictOtherTrans(SMnode *pMnode, const char *pSrcDb, const char *pDstDb, bool lock);
// for sma
// TODO refactor

53
source/dnode/mnode/impl/src/mndStream.c
View File

@ -742,14 +742,14 @@ static int32_t checkForNumOfStreams(SMnode *pMnode, SStreamObj *pStreamObj) { /
}
static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) {
SMnode * pMnode = pReq->info.node;
int32_t code = -1;
SStreamObj * pStream = NULL;
SDbObj * pDb = NULL;
SCMCreateStreamReq createStreamReq = {0};
SStreamObj streamObj = {0};
char* sql = NULL;
SMnode *pMnode = pReq->info.node;
int32_t code = -1;
SStreamObj *pStream = NULL;
SStreamObj streamObj = {0};
char *sql = NULL;
int32_t sqlLen = 0;
SCMCreateStreamReq createStreamReq = {0};
if (tDeserializeSCMCreateStreamReq(pReq->pCont, pReq->contLen, &createStreamReq) != 0) {
terrno = TSDB_CODE_INVALID_MSG;
goto _OVER;
@ -780,8 +780,7 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) {
goto _OVER;
}
int32_t sqlLen = 0;
if(createStreamReq.sql != NULL){
if (createStreamReq.sql != NULL) {
sqlLen = strlen(createStreamReq.sql);
sql = taosMemoryMalloc(sqlLen + 1);
memset(sql, 0, sqlLen + 1);
@ -868,14 +867,13 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) {
// reuse this function for stream
if (sql != NULL && sqlLen > 0) {
auditRecord(pReq, pMnode->clusterId, "createStream", dbname.dbname, name.dbname, sql,
sqlLen);
}
else{
auditRecord(pReq, pMnode->clusterId, "createStream", dbname.dbname, name.dbname, sql, sqlLen);
} else {
char detail[1000] = {0};
sprintf(detail, "dbname:%s, stream name:%s", dbname.dbname, name.dbname);
auditRecord(pReq, pMnode->clusterId, "createStream", dbname.dbname, name.dbname, detail, strlen(detail));
}
_OVER:
if (code != 0 && code != TSDB_CODE_ACTION_IN_PROGRESS) {
mError("stream:%s, failed to create since %s", createStreamReq.name, terrstr());
@ -884,7 +882,7 @@ _OVER:
mndReleaseStream(pMnode, pStream);
tFreeSCMCreateStreamReq(&createStreamReq);
tFreeStreamObj(&streamObj);
if(sql != NULL){
if (sql != NULL) {
taosMemoryFreeClear(sql);
}
return code;
@ -1335,7 +1333,7 @@ static int32_t mndProcessDropStreamReq(SRpcMsg *pReq) {
}
// check if it is conflict with other trans in both sourceDb and targetDb.
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb);
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb, true);
if (conflict) {
sdbRelease(pMnode->pSdb, pStream);
tFreeMDropStreamReq(&dropReq);
@ -1818,7 +1816,7 @@ static int32_t mndProcessPauseStreamReq(SRpcMsg *pReq) {
}
// check if it is conflict with other trans in both sourceDb and targetDb.
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb);
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb, true);
if (conflict) {
sdbRelease(pMnode->pSdb, pStream);
return -1;
@ -1953,7 +1951,7 @@ static int32_t mndProcessResumeStreamReq(SRpcMsg *pReq) {
}
// check if it is conflict with other trans in both sourceDb and targetDb.
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb);
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb, true);
if (conflict) {
sdbRelease(pMnode->pSdb, pStream);
return -1;
@ -2741,7 +2739,7 @@ static int32_t mndResetStatusFromCheckpoint(SMnode *pMnode, int32_t transId) {
break;
}
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb);
bool conflict = streamTransConflictOtherTrans(pMnode, pStream->sourceDb, pStream->targetDb, false);
if (conflict) {
mError("stream:%s other trans exists in DB:%s & %s failed to start reset-status trans",
pStream->name, pStream->sourceDb, pStream->targetDb);
@ -2777,7 +2775,7 @@ static SStreamTask* mndGetStreamTask(STaskId* pId, SStreamObj* pStream) {
static bool needDropRelatedFillhistoryTask(STaskStatusEntry *pTaskEntry, SStreamExecInfo *pExecNode) {
if (pTaskEntry->status == TASK_STATUS__STREAM_SCAN_HISTORY && pTaskEntry->statusLastDuration >= 10) {
if (fabs(pTaskEntry->inputQUsed) <= DBL_EPSILON) {
if (!pTaskEntry->inputQChanging && pTaskEntry->inputQUnchangeCounter > 10) {
int32_t numOfReady = 0;
int32_t numOfTotal = 0;
for (int32_t k = 0; k < taosArrayGetSize(pExecNode->pTaskList); ++k) {
@ -2920,6 +2918,7 @@ int32_t mndProcessStreamHb(SRpcMsg *pReq) {
bool snodeChanged = false;
for (int32_t i = 0; i < req.numOfTasks; ++i) {
STaskStatusEntry *p = taosArrayGet(req.pTaskStatus, i);
STaskStatusEntry *pTaskEntry = taosHashGet(execInfo.pTaskMap, &p->id, sizeof(p->id));
if (pTaskEntry == NULL) {
mError("s-task:0x%" PRIx64 " not found in mnode task list", p->id.taskId);
@ -2928,8 +2927,22 @@ int32_t mndProcessStreamHb(SRpcMsg *pReq) {
if (pTaskEntry->stage != p->stage && pTaskEntry->stage != -1) {
updateStageInfo(pTaskEntry, p->stage);
if(pTaskEntry->nodeId == SNODE_HANDLE) snodeChanged = true;
if(pTaskEntry->nodeId == SNODE_HANDLE) {
snodeChanged = true;
}
} else {
// task is idle for more than 50 sec.
if (fabs(pTaskEntry->inputQUsed - p->inputQUsed) <= DBL_EPSILON) {
if (!pTaskEntry->inputQChanging) {
pTaskEntry->inputQUnchangeCounter++;
} else {
pTaskEntry->inputQChanging = false;
}
} else {
pTaskEntry->inputQChanging = true;
pTaskEntry->inputQUnchangeCounter = 0;
}
streamTaskStatusCopy(pTaskEntry, p);
if (p->activeCheckpointId != 0) {
if (activeCheckpointId != 0) {

41
source/dnode/mnode/impl/src/mndStreamTrans.c
View File

@ -35,17 +35,15 @@ int32_t mndStreamRegisterTrans(STrans* pTrans, const char* pName, const char* pS
}
int32_t clearFinishedTrans(SMnode* pMnode) {
SArray* pList = taosArrayInit(4, sizeof(SKeyInfo));
size_t keyLen = 0;
SArray* pList = taosArrayInit(4, sizeof(SKeyInfo));
void* pIter = NULL;
taosThreadMutexLock(&execInfo.lock);
void* pIter = NULL;
while ((pIter = taosHashIterate(execInfo.transMgmt.pDBTrans, pIter)) != NULL) {
SStreamTransInfo *pEntry = (SStreamTransInfo *)pIter;
STrans* pTrans = mndAcquireTrans(pMnode, pEntry->transId);
SStreamTransInfo* pEntry = (SStreamTransInfo*)pIter;
// let's clear the finished trans
STrans* pTrans = mndAcquireTrans(pMnode, pEntry->transId);
if (pTrans == NULL) {
void* pKey = taosHashGetKey(pEntry, &keyLen);
// key is the name of src/dst db name
@ -60,44 +58,55 @@ int32_t clearFinishedTrans(SMnode* pMnode) {
}
size_t num = taosArrayGetSize(pList);
for(int32_t i = 0; i < num; ++i) {
for (int32_t i = 0; i < num; ++i) {
SKeyInfo* pKey = taosArrayGet(pList, i);
taosHashRemove(execInfo.transMgmt.pDBTrans, pKey->pKey, pKey->keyLen);
}
mDebug("clear %d finished stream-trans, remained:%d", (int32_t) num, taosHashGetSize(execInfo.transMgmt.pDBTrans));
taosThreadMutexUnlock(&execInfo.lock);
mDebug("clear %d finished stream-trans, remained:%d", (int32_t)num, taosHashGetSize(execInfo.transMgmt.pDBTrans));
terrno = TSDB_CODE_SUCCESS;
taosArrayDestroy(pList);
return 0;
}
bool streamTransConflictOtherTrans(SMnode* pMnode, const char* pSrcDb, const char* pDstDb) {
clearFinishedTrans(pMnode);
bool streamTransConflictOtherTrans(SMnode* pMnode, const char* pSrcDb, const char* pDstDb, bool lock) {
if (lock) {
taosThreadMutexLock(&execInfo.lock);
}
taosThreadMutexLock(&execInfo.lock);
int32_t num = taosHashGetSize(execInfo.transMgmt.pDBTrans);
if (num <= 0) {
taosThreadMutexUnlock(&execInfo.lock);
if (lock) {
taosThreadMutexUnlock(&execInfo.lock);
}
return false;
}
clearFinishedTrans(pMnode);
SStreamTransInfo *pEntry = taosHashGet(execInfo.transMgmt.pDBTrans, pSrcDb, strlen(pSrcDb));
if (pEntry != NULL) {
taosThreadMutexUnlock(&execInfo.lock);
if (lock) {
taosThreadMutexUnlock(&execInfo.lock);
}
mWarn("conflict with other transId:%d in Db:%s, trans:%s", pEntry->transId, pSrcDb, pEntry->name);
return true;
}
pEntry = taosHashGet(execInfo.transMgmt.pDBTrans, pDstDb, strlen(pDstDb));
if (pEntry != NULL) {
taosThreadMutexUnlock(&execInfo.lock);
if (lock) {
taosThreadMutexUnlock(&execInfo.lock);
}
mWarn("conflict with other transId:%d in Db:%s, trans:%s", pEntry->transId, pSrcDb, pEntry->name);
return true;
}
taosThreadMutexUnlock(&execInfo.lock);
if (lock) {
taosThreadMutexUnlock(&execInfo.lock);
}
return false;
}

31
source/dnode/mnode/impl/src/mndUser.c
View File

@ -708,9 +708,8 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
int32_t numOfAlterViews = taosHashGetSize(pUser->alterViews);
int32_t numOfTopics = taosHashGetSize(pUser->topics);
int32_t numOfUseDbs = taosHashGetSize(pUser->useDbs);
int32_t size = sizeof(SUserObj) + USER_RESERVE_SIZE +
(numOfReadDbs + numOfWriteDbs + numOfUseDbs) * TSDB_DB_FNAME_LEN + numOfTopics * TSDB_TOPIC_FNAME_LEN +
ipWhiteReserve;
int32_t size = sizeof(SUserObj) + USER_RESERVE_SIZE + (numOfReadDbs + numOfWriteDbs) * TSDB_DB_FNAME_LEN +
numOfTopics * TSDB_TOPIC_FNAME_LEN + ipWhiteReserve;
char *stb = taosHashIterate(pUser->readTbs, NULL);
while (stb != NULL) {
@ -720,7 +719,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->readTbs, stb);
@ -734,7 +733,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->writeTbs, stb);
@ -748,7 +747,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->alterTbs, stb);
@ -762,7 +761,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->readViews, stb);
@ -776,7 +775,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->writeViews, stb);
@ -790,11 +789,21 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
size += keyLen;
size_t valueLen = 0;
valueLen = strlen(stb);
valueLen = strlen(stb) + 1;
size += sizeof(int32_t);
size += valueLen;
stb = taosHashIterate(pUser->alterViews, stb);
}
}
int32_t *useDb = taosHashIterate(pUser->useDbs, NULL);
while (useDb != NULL) {
size_t keyLen = 0;
void *key = taosHashGetKey(useDb, &keyLen);
size += sizeof(int32_t);
size += keyLen;
size += sizeof(int32_t);
useDb = taosHashIterate(pUser->useDbs, useDb);
}
SSdbRaw *pRaw = sdbAllocRaw(SDB_USER, USER_VER_NUMBER, size);
if (pRaw == NULL) goto _OVER;
@ -925,7 +934,7 @@ SSdbRaw *mndUserActionEncode(SUserObj *pUser) {
stb = taosHashIterate(pUser->alterViews, stb);
}
int32_t *useDb = taosHashIterate(pUser->useDbs, NULL);
useDb = taosHashIterate(pUser->useDbs, NULL);
while (useDb != NULL) {
size_t keyLen = 0;
void *key = taosHashGetKey(useDb, &keyLen);

1
source/dnode/snode/CMakeLists.txt
View File

@ -17,4 +17,5 @@ target_link_libraries(
PRIVATE stream
PRIVATE wal
PRIVATE index
PRIVATE tqCommon
)

785
source/dnode/snode/src/snode.c
View File

@ -16,9 +16,8 @@
#include "rsync.h"
#include "executor.h"
#include "sndInt.h"
#include "tstream.h"
#include "tqCommon.h"
#include "tuuid.h"
#include "stream.h"
#define sndError(...) \
do { \
@ -41,45 +40,6 @@
} \
} while (0)
void sndEnqueueStreamDispatch(SSnode *pSnode, SRpcMsg *pMsg) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamDispatchReq req;
SDecoder decoder;
tDecoderInit(&decoder, msgBody, msgLen);
if (tDecodeStreamDispatchReq(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
goto FAIL;
}
tDecoderClear(&decoder);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = { .info = pMsg->info, .code = 0 };
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pSnode->pMeta, pTask);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return;
} else {
tDeleteStreamDispatchReq(&req);
return;
}
FAIL:
if (pMsg->info.handle == NULL) return;
SRpcMsg rsp = { .code = code, .info = pMsg->info};
tmsgSendRsp(&rsp);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
}
int32_t sndExpandTask(SSnode *pSnode, SStreamTask *pTask, int64_t nextProcessVer) {
ASSERT(pTask->info.taskLevel == TASK_LEVEL__AGG && taosArrayGetSize(pTask->upstreamInfo.pList) != 0);
int32_t code = streamTaskInit(pTask, pSnode->pMeta, &pSnode->msgCb, nextProcessVer);
@ -130,159 +90,29 @@ int32_t sndExpandTask(SSnode *pSnode, SStreamTask *pTask, int64_t nextProcessVer
pTask->chkInfo.nextProcessVer = pTask->chkInfo.checkpointVer + 1;
sndInfo("s-task:%s restore from the checkpointId:%" PRId64 " ver:%" PRId64 " nextProcessVer:%" PRId64, pTask->id.idStr,
pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer);
} else {
if (pTask->chkInfo.nextProcessVer == -1) {
pTask->chkInfo.nextProcessVer = 0;
}
}
char* p = NULL;
streamTaskGetStatus(pTask, &p);
sndInfo("snode:%d expand stream task, s-task:%s, checkpointId:%" PRId64 " checkpointVer:%" PRId64
" nextProcessVer:%" PRId64 " child id:%d, level:%d, status:%s fill-history:%d, trigger:%" PRId64 " ms",
SNODE_HANDLE, pTask->id.idStr, pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer,
pTask->info.selfChildId, pTask->info.taskLevel, p, pTask->info.fillHistory, pTask->info.triggerParam);
if (pTask->info.fillHistory) {
sndInfo("vgId:%d expand stream task, s-task:%s, checkpointId:%" PRId64 " checkpointVer:%" PRId64
" nextProcessVer:%" PRId64
" child id:%d, level:%d, status:%s fill-history:%d, related stream task:0x%x trigger:%" PRId64 " ms",
SNODE_HANDLE, pTask->id.idStr, pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer,
pTask->info.selfChildId, pTask->info.taskLevel, p, pTask->info.fillHistory,
(int32_t)pTask->streamTaskId.taskId, pTask->info.triggerParam);
} else {
sndInfo("vgId:%d expand stream task, s-task:%s, checkpointId:%" PRId64 " checkpointVer:%" PRId64
" nextProcessVer:%" PRId64
" child id:%d, level:%d, status:%s fill-history:%d, related fill-task:0x%x trigger:%" PRId64 " ms",
SNODE_HANDLE, pTask->id.idStr, pChkInfo->checkpointId, pChkInfo->checkpointVer, pChkInfo->nextProcessVer,
pTask->info.selfChildId, pTask->info.taskLevel, p, pTask->info.fillHistory,
(int32_t)pTask->hTaskInfo.id.taskId, pTask->info.triggerParam);
}
return 0;
}
int32_t sndStartStreamTasks(SSnode* pSnode) {
int32_t code = TSDB_CODE_SUCCESS;
int32_t vgId = SNODE_HANDLE;
SStreamMeta* pMeta = pSnode->pMeta;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
sndDebug("vgId:%d start to check all %d stream task(s) downstream status", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
streamMetaWLock(pMeta);
pTaskList = taosArrayDup(pMeta->pTaskList, NULL);
taosHashClear(pMeta->startInfo.pReadyTaskSet);
taosHashClear(pMeta->startInfo.pFailedTaskSet);
pMeta->startInfo.startTs = taosGetTimestampMs();
streamMetaWUnLock(pMeta);
// broadcast the check downstream tasks msg
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pTaskId->streamId, pTaskId->taskId);
if (pTask == NULL) {
continue;
}
// fill-history task can only be launched by related stream tasks.
if (pTask->info.fillHistory == 1) {
streamMetaReleaseTask(pMeta, pTask);
continue;
}
if (pTask->status.downstreamReady == 1) {
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
sndDebug("s-task:%s downstream ready, no need to check downstream, check only related fill-history task",
pTask->id.idStr);
streamLaunchFillHistoryTask(pTask);
}
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, pTask->execInfo.start, true);
streamMetaReleaseTask(pMeta, pTask);
continue;
}
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
int32_t ret = streamTaskHandleEvent(pTask->status.pSM, event);
if (ret != TSDB_CODE_SUCCESS) {
code = ret;
}
streamMetaReleaseTask(pMeta, pTask);
}
taosArrayDestroy(pTaskList);
return code;
}
int32_t sndResetStreamTaskStatus(SSnode* pSnode) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
sndDebug("vgId:%d reset all %d stream task(s) status to be uninit", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pMeta->pTaskList, i);
STaskId id = {.streamId = pTaskId->streamId, .taskId = pTaskId->taskId};
SStreamTask** pTask = taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
streamTaskResetStatus(*pTask);
}
return 0;
}
int32_t sndRestartStreamTasks(SSnode* pSnode) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = pMeta->vgId;
int32_t code = 0;
int64_t st = taosGetTimestampMs();
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
sndDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
terrno = 0;
sndInfo("vgId:%d tasks are all updated and stopped, restart all tasks, triggered by transId:%d", vgId,
pMeta->updateInfo.transId);
while (streamMetaTaskInTimer(pMeta)) {
sndDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
code = streamMetaReopen(pMeta);
if (code != TSDB_CODE_SUCCESS) {
sndError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
streamMetaInitBackend(pMeta);
int64_t el = taosGetTimestampMs() - st;
sndInfo("vgId:%d close&reload state elapsed time:%.3fs", vgId, el/1000.);
code = streamMetaLoadAllTasks(pMeta);
if (code != TSDB_CODE_SUCCESS) {
sndError("vgId:%d failed to load stream tasks, code:%s", vgId, tstrerror(terrno));
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
sndInfo("vgId:%d restart all stream tasks after all tasks being updated", vgId);
sndResetStreamTaskStatus(pSnode);
streamMetaWUnLock(pMeta);
sndStartStreamTasks(pSnode);
code = terrno;
return code;
}
SSnode *sndOpen(const char *path, const SSnodeOpt *pOption) {
SSnode *pSnode = taosMemoryCalloc(1, sizeof(SSnode));
if (pSnode == NULL) {
@ -318,8 +148,8 @@ FAIL:
}
int32_t sndInit(SSnode * pSnode) {
sndResetStreamTaskStatus(pSnode);
sndStartStreamTasks(pSnode);
resetStreamTaskStatus(pSnode->pMeta);
startStreamTasks(pSnode->pMeta);
return 0;
}
@ -332,578 +162,30 @@ void sndClose(SSnode *pSnode) {
taosMemoryFree(pSnode);
}
int32_t sndGetLoad(SSnode *pSnode, SSnodeLoad *pLoad) { return 0; }
int32_t sndStartStreamTaskAsync(SSnode* pSnode, bool restart) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
sndDebug("vgId:%d no stream tasks existed to run", vgId);
return 0;
}
SStreamTaskRunReq* pRunReq = rpcMallocCont(sizeof(SStreamTaskRunReq));
if (pRunReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
sndError("vgId:%d failed to create msg to start wal scanning to launch stream tasks, code:%s", vgId, terrstr());
return -1;
}
sndDebug("vgId:%d start all %d stream task(s) async", vgId, numOfTasks);
pRunReq->head.vgId = vgId;
pRunReq->streamId = 0;
pRunReq->taskId = restart? STREAM_EXEC_RESTART_ALL_TASKS_ID:STREAM_EXEC_START_ALL_TASKS_ID;
SRpcMsg msg = {.msgType = TDMT_STREAM_TASK_RUN, .pCont = pRunReq, .contLen = sizeof(SStreamTaskRunReq)};
tmsgPutToQueue(&pSnode->msgCb, STREAM_QUEUE, &msg);
return 0;
}
int32_t sndProcessTaskDeployReq(SSnode *pSnode, char *msg, int32_t msgLen) {
int32_t code;
// 1.deserialize msg and build task
SStreamTask *pTask = taosMemoryCalloc(1, sizeof(SStreamTask));
if (pTask == NULL) {
return -1;
}
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t *)msg, msgLen);
code = tDecodeStreamTask(&decoder, pTask);
if (code < 0) {
tDecoderClear(&decoder);
taosMemoryFree(pTask);
return -1;
}
tDecoderClear(&decoder);
ASSERT(pTask->info.taskLevel == TASK_LEVEL__AGG);
// 2.save task
streamMetaWLock(pSnode->pMeta);
bool added = false;
code = streamMetaRegisterTask(pSnode->pMeta, -1, pTask, &added);
if (code < 0) {
streamMetaWUnLock(pSnode->pMeta);
return -1;
}
int32_t numOfTasks = streamMetaGetNumOfTasks(pSnode->pMeta);
streamMetaWUnLock(pSnode->pMeta);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
sndDebug("snode:%d s-task:%s is deployed on snode and add into meta, status:%s, numOfTasks:%d", SNODE_HANDLE,
pTask->id.idStr, p, numOfTasks);
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
streamTaskHandleEvent(pTask->status.pSM, event);
return 0;
}
int32_t sndProcessTaskDropReq(SSnode *pSnode, char *msg, int32_t msgLen) {
SVDropStreamTaskReq *pReq = (SVDropStreamTaskReq *)msg;
sndDebug("snode:%d receive msg to drop stream task:0x%x", pSnode->pMeta->vgId, pReq->taskId);
streamMetaUnregisterTask(pSnode->pMeta, pReq->streamId, pReq->taskId);
// commit the update
streamMetaWLock(pSnode->pMeta);
int32_t numOfTasks = streamMetaGetNumOfTasks(pSnode->pMeta);
sndDebug("vgId:%d task:0x%x dropped, remain tasks:%d", pSnode->pMeta->vgId, pReq->taskId, numOfTasks);
if (streamMetaCommit(pSnode->pMeta) < 0) {
// persist to disk
}
streamMetaWUnLock(pSnode->pMeta);
return 0;
}
int32_t sndProcessTaskRunReq(SSnode *pSnode, SRpcMsg *pMsg) {
SStreamTaskRunReq *pReq = pMsg->pCont;
int32_t taskId = pReq->taskId;
if (taskId == STREAM_EXEC_START_ALL_TASKS_ID) {
sndStartStreamTasks(pSnode);
return 0;
} else if (taskId == STREAM_EXEC_RESTART_ALL_TASKS_ID) {
sndRestartStreamTasks(pSnode);
return 0;
}
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, pReq->streamId, pReq->taskId);
if (pTask) {
streamExecTask(pTask);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
} else {
return -1;
}
}
int32_t sndProcessTaskDispatchReq(SSnode *pSnode, SRpcMsg *pMsg, bool exec) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamDispatchReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t *)msgBody, msgLen);
tDecodeStreamDispatchReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
} else {
tDeleteStreamDispatchReq(&req);
return -1;
}
}
int32_t sndProcessTaskRetrieveReq(SSnode *pSnode, SRpcMsg *pMsg) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamRetrieveReq req;
SDecoder decoder;
tDecoderInit(&decoder, msgBody, msgLen);
tDecodeStreamRetrieveReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.dstTaskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessRetrieveReq(pTask, &req, &rsp);
streamMetaReleaseTask(pSnode->pMeta, pTask);
tDeleteStreamRetrieveReq(&req);
return 0;
} else {
return -1;
}
}
int32_t sndProcessTaskDispatchRsp(SSnode *pSnode, SRpcMsg *pMsg) {
SStreamDispatchRsp *pRsp = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
pRsp->upstreamTaskId = htonl(pRsp->upstreamTaskId);
pRsp->streamId = htobe64(pRsp->streamId);
pRsp->downstreamTaskId = htonl(pRsp->downstreamTaskId);
pRsp->downstreamNodeId = htonl(pRsp->downstreamNodeId);
pRsp->stage = htobe64(pRsp->stage);
pRsp->msgId = htonl(pRsp->msgId);
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, pRsp->streamId, pRsp->upstreamTaskId);
if (pTask) {
streamProcessDispatchRsp(pTask, pRsp, pMsg->code);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
} else {
return -1;
}
}
int32_t sndProcessTaskRetrieveRsp(SSnode *pSnode, SRpcMsg *pMsg) {
//
return 0;
}
int32_t sndProcessTaskScanHistoryFinishReq(SSnode *pSnode, SRpcMsg *pMsg) {
char *msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamScanHistoryFinishReq req;
SDecoder decoder;
tDecoderInit(&decoder, msg, msgLen);
tDecodeStreamScanHistoryFinishReq(&decoder, &req);
tDecoderClear(&decoder);
// find task
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.downstreamTaskId);
if (pTask == NULL) {
return -1;
}
// do process request
if (streamProcessScanHistoryFinishReq(pTask, &req, &pMsg->info) < 0) {
streamMetaReleaseTask(pSnode->pMeta, pTask);
return -1;
}
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
}
int32_t sndProcessTaskScanHistoryFinishRsp(SSnode *pSnode, SRpcMsg *pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamCompleteHistoryMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeCompleteHistoryDataMsg(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
sndError("vgId:%d process scan history finish rsp, failed to find task:0x%x, it may be destroyed",
pSnode->pMeta->vgId, req.upstreamTaskId);
return -1;
}
int32_t remain = atomic_sub_fetch_32(&pTask->notReadyTasks, 1);
if (remain > 0) {
sndDebug("s-task:%s scan-history finish rsp received from downstream task:0x%x, unfinished remain:%d",
pTask->id.idStr, req.downstreamId, remain);
} else {
sndDebug(
"s-task:%s scan-history finish rsp received from downstream task:0x%x, all downstream tasks rsp scan-history "
"completed msg",
pTask->id.idStr, req.downstreamId);
streamProcessScanHistoryFinishRsp(pTask);
}
streamMetaReleaseTask(pSnode->pMeta, pTask);
return 0;
}
// downstream task has complete the stream task checkpoint procedure, let's start the handle the rsp by execute task
int32_t sndProcessTaskCheckpointReadyMsg(SSnode *pSnode, SRpcMsg* pMsg) {
SStreamMeta* pMeta = pSnode->pMeta;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamCheckpointReadyMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamCheckpointReadyMsg(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
sndError("vgId:%d failed to find s-task:0x%x, it may have been destroyed already", pMeta->vgId, req.downstreamTaskId);
return code;
}
sndDebug("snode vgId:%d s-task:%s received the checkpoint ready msg from task:0x%x (vgId:%d), handle it", pMeta->vgId,
pTask->id.idStr, req.downstreamTaskId, req.downstreamNodeId);
streamProcessCheckpointReadyMsg(pTask);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t sndProcessStreamTaskCheckReq(SSnode *pSnode, SRpcMsg *pMsg) {
char *msgStr = pMsg->pCont;
char *msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamTaskCheckReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t *)msgBody, msgLen);
tDecodeStreamTaskCheckReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t taskId = req.downstreamTaskId;
SStreamTaskCheckRsp rsp = {
.reqId = req.reqId,
.streamId = req.streamId,
.childId = req.childId,
.downstreamNodeId = req.downstreamNodeId,
.downstreamTaskId = req.downstreamTaskId,
.upstreamNodeId = req.upstreamNodeId,
.upstreamTaskId = req.upstreamTaskId,
};
SStreamTask *pTask = streamMetaAcquireTask(pSnode->pMeta, req.streamId, taskId);
if (pTask != NULL) {
rsp.status = streamTaskCheckStatus(pTask, req.upstreamTaskId, req.upstreamNodeId, req.stage, &rsp.oldStage);
streamMetaReleaseTask(pSnode->pMeta, pTask);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
sndDebug("s-task:%s status:%s, recv task check req(reqId:0x%" PRIx64 ") task:0x%x (vgId:%d), ready:%d",
pTask->id.idStr, p, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
} else {
rsp.status = TASK_DOWNSTREAM_NOT_READY;
sndDebug("recv task check(taskId:0x%x not built yet) req(reqId:0x%" PRIx64 ") from task:0x%x (vgId:%d), rsp status %d",
taskId, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
}
SEncoder encoder;
int32_t code;
int32_t len;
tEncodeSize(tEncodeStreamTaskCheckRsp, &rsp, len, code);
if (code < 0) {
sndError("vgId:%d failed to encode task check rsp, task:0x%x", pSnode->pMeta->vgId, taskId);
return -1;
}
void *buf = rpcMallocCont(sizeof(SMsgHead) + len);
((SMsgHead *)buf)->vgId = htonl(req.upstreamNodeId);
void *abuf = POINTER_SHIFT(buf, sizeof(SMsgHead));
tEncoderInit(&encoder, (uint8_t *)abuf, len);
tEncodeStreamTaskCheckRsp(&encoder, &rsp);
tEncoderClear(&encoder);
SRpcMsg rspMsg = {.code = 0, .pCont = buf, .contLen = sizeof(SMsgHead) + len, .info = pMsg->info};
tmsgSendRsp(&rspMsg);
return 0;
}
int32_t sndProcessStreamTaskCheckRsp(SSnode* pSnode, SRpcMsg* pMsg) {
char* pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code;
SStreamTaskCheckRsp rsp;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)pReq, len);
code = tDecodeStreamTaskCheckRsp(&decoder, &rsp);
if (code < 0) {
tDecoderClear(&decoder);
return -1;
}
tDecoderClear(&decoder);
sndDebug("tq task:0x%x (vgId:%d) recv check rsp(reqId:0x%" PRIx64 ") from 0x%x (vgId:%d) status %d",
rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.reqId, rsp.downstreamTaskId, rsp.downstreamNodeId, rsp.status);
SStreamTask* pTask = streamMetaAcquireTask(pSnode->pMeta, rsp.streamId, rsp.upstreamTaskId);
if (pTask == NULL) {
sndError("tq failed to locate the stream task:0x%x (vgId:%d), it may have been destroyed", rsp.upstreamTaskId,
pSnode->pMeta->vgId);
return -1;
}
code = streamProcessCheckRsp(pTask, &rsp);
streamMetaReleaseTask(pSnode->pMeta, pTask);
return code;
}
int32_t sndProcessTaskUpdateReq(SSnode* pSnode, SRpcMsg* pMsg) {
SStreamMeta* pMeta = pSnode->pMeta;
int32_t vgId = SNODE_HANDLE;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
SRpcMsg rsp = {.info = pMsg->info, .code = TSDB_CODE_SUCCESS};
SStreamTaskNodeUpdateMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamTaskUpdateMsg(&decoder, &req) < 0) {
rsp.code = TSDB_CODE_MSG_DECODE_ERROR;
sndError("vgId:%d failed to decode task update msg, code:%s", vgId, tstrerror(rsp.code));
tDecoderClear(&decoder);
return rsp.code;
}
tDecoderClear(&decoder);
// update the nodeEpset when it exists
streamMetaWLock(pMeta);
// the task epset may be updated again and again, when replaying the WAL, the task may be in stop status.
STaskId id = {.streamId = req.streamId, .taskId = req.taskId};
SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
if (ppTask == NULL || *ppTask == NULL) {
sndError("vgId:%d failed to acquire task:0x%x when handling update, it may have been dropped already", pMeta->vgId,
req.taskId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
SStreamTask* pTask = *ppTask;
if (pMeta->updateInfo.transId != req.transId) {
pMeta->updateInfo.transId = req.transId;
sndInfo("s-task:%s receive new trans to update nodeEp msg from mnode, transId:%d", pTask->id.idStr, req.transId);
// info needs to be kept till the new trans to update the nodeEp arrived.
taosHashClear(pMeta->updateInfo.pTasks);
} else {
sndDebug("s-task:%s recv trans to update nodeEp from mnode, transId:%d", pTask->id.idStr, req.transId);
}
STaskUpdateEntry entry = {.streamId = req.streamId, .taskId = req.taskId, .transId = req.transId};
void* exist = taosHashGet(pMeta->updateInfo.pTasks, &entry, sizeof(STaskUpdateEntry));
if (exist != NULL) {
sndDebug("s-task:%s (vgId:%d) already update in trans:%d, discard the nodeEp update msg", pTask->id.idStr, vgId,
req.transId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
streamMetaWUnLock(pMeta);
// the following two functions should not be executed within the scope of meta lock to avoid deadlock
streamTaskUpdateEpsetInfo(pTask, req.pNodeList);
streamTaskResetStatus(pTask);
// continue after lock the meta again
streamMetaWLock(pMeta);
SStreamTask** ppHTask = NULL;
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
ppHTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &pTask->hTaskInfo.id, sizeof(pTask->hTaskInfo.id));
if (ppHTask == NULL || *ppHTask == NULL) {
sndError("vgId:%d failed to acquire fill-history task:0x%x when handling update, it may have been dropped already",
pMeta->vgId, req.taskId);
CLEAR_RELATED_FILLHISTORY_TASK(pTask);
} else {
sndDebug("s-task:%s fill-history task update nodeEp along with stream task", (*ppHTask)->id.idStr);
streamTaskUpdateEpsetInfo(*ppHTask, req.pNodeList);
}
}
{
streamMetaSaveTask(pMeta, pTask);
if (ppHTask != NULL) {
streamMetaSaveTask(pMeta, *ppHTask);
}
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
}
streamTaskStop(pTask);
// keep the already handled info
taosHashPut(pMeta->updateInfo.pTasks, &entry, sizeof(entry), NULL, 0);
if (ppHTask != NULL) {
streamTaskStop(*ppHTask);
sndDebug("s-task:%s task nodeEp update completed, streamTask and related fill-history task closed", pTask->id.idStr);
taosHashPut(pMeta->updateInfo.pTasks, &(*ppHTask)->id, sizeof(pTask->id), NULL, 0);
} else {
sndDebug("s-task:%s task nodeEp update completed, streamTask closed", pTask->id.idStr);
}
rsp.code = 0;
// possibly only handle the stream task.
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
int32_t updateTasks = taosHashGetSize(pMeta->updateInfo.pTasks);
pMeta->startInfo.tasksWillRestart = 1;
if (updateTasks < numOfTasks) {
sndDebug("vgId:%d closed tasks:%d, unclosed:%d, all tasks will be started when nodeEp update completed", vgId,
updateTasks, (numOfTasks - updateTasks));
streamMetaWUnLock(pMeta);
} else {
sndDebug("vgId:%d all %d task(s) nodeEp updated and closed", vgId, numOfTasks);
#if 1
sndStartStreamTaskAsync(pSnode, true);
streamMetaWUnLock(pMeta);
#else
streamMetaWUnLock(pMeta);
// For debug purpose.
// the following procedure consume many CPU resource, result in the re-election of leader
// with high probability. So we employ it as a test case for the stream processing framework, with
// checkpoint/restart/nodeUpdate etc.
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
int32_t code = streamMetaReopen(pMeta);
if (code != 0) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (streamMetaLoadAllTasks(pTq->pStreamMeta) < 0) {
tqError("vgId:%d failed to load stream tasks", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d start all stream tasks after all being updated", vgId);
tqResetStreamTaskStatus(pTq);
tqStartStreamTaskAsync(pTq, false);
} else {
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
streamMetaWUnLock(pMeta);
#endif
}
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
int32_t sndProcessStreamMsg(SSnode *pSnode, SRpcMsg *pMsg) {
switch (pMsg->msgType) {
case TDMT_STREAM_TASK_RUN:
return sndProcessTaskRunReq(pSnode, pMsg);
return tqStreamTaskProcessRunReq(pSnode->pMeta, pMsg, true);
case TDMT_STREAM_TASK_DISPATCH:
return sndProcessTaskDispatchReq(pSnode, pMsg, true);
return tqStreamTaskProcessDispatchReq(pSnode->pMeta, pMsg);
case TDMT_STREAM_TASK_DISPATCH_RSP:
return sndProcessTaskDispatchRsp(pSnode, pMsg);
return tqStreamTaskProcessDispatchRsp(pSnode->pMeta, pMsg);
case TDMT_STREAM_RETRIEVE:
return sndProcessTaskRetrieveReq(pSnode, pMsg);
case TDMT_STREAM_RETRIEVE_RSP:
return sndProcessTaskRetrieveRsp(pSnode, pMsg);
return tqStreamTaskProcessRetrieveReq(pSnode->pMeta, pMsg);
case TDMT_STREAM_RETRIEVE_RSP: // 1036
break;
case TDMT_VND_STREAM_SCAN_HISTORY_FINISH:
return sndProcessTaskScanHistoryFinishReq(pSnode, pMsg);
return tqStreamTaskProcessScanHistoryFinishReq(pSnode->pMeta, pMsg);
case TDMT_VND_STREAM_SCAN_HISTORY_FINISH_RSP:
return sndProcessTaskScanHistoryFinishRsp(pSnode, pMsg);
return tqStreamTaskProcessScanHistoryFinishRsp(pSnode->pMeta, pMsg);
case TDMT_VND_STREAM_TASK_CHECK:
return sndProcessStreamTaskCheckReq(pSnode, pMsg);
return tqStreamTaskProcessCheckReq(pSnode->pMeta, pMsg);
case TDMT_VND_STREAM_TASK_CHECK_RSP:
return sndProcessStreamTaskCheckRsp(pSnode, pMsg);
return tqStreamTaskProcessCheckRsp(pSnode->pMeta, pMsg, true);
case TDMT_STREAM_TASK_CHECKPOINT_READY:
return sndProcessTaskCheckpointReadyMsg(pSnode, pMsg);
return tqStreamTaskProcessCheckpointReadyMsg(pSnode->pMeta, pMsg);
default:
sndError("invalid snode msg:%d", pMsg->msgType);
ASSERT(0);
}
return 0;
@ -914,14 +196,13 @@ int32_t sndProcessWriteMsg(SSnode *pSnode, SRpcMsg *pMsg, SRpcMsg *pRsp) {
case TDMT_STREAM_TASK_DEPLOY: {
void *pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
return sndProcessTaskDeployReq(pSnode, pReq, len);
return tqStreamTaskProcessDeployReq(pSnode->pMeta, -1, pReq, len, true, true);
}
case TDMT_STREAM_TASK_DROP:
return sndProcessTaskDropReq(pSnode, pMsg->pCont, pMsg->contLen);
return tqStreamTaskProcessDropReq(pSnode->pMeta, pMsg->pCont, pMsg->contLen);
case TDMT_VND_STREAM_TASK_UPDATE:
sndProcessTaskUpdateReq(pSnode, pMsg);
break;
return tqStreamTaskProcessUpdateReq(pSnode->pMeta, &pSnode->msgCb, pMsg, true);
default:
ASSERT(0);
}

9
source/dnode/vnode/CMakeLists.txt
View File

@ -1,4 +1,5 @@
# vnode
add_subdirectory(src/tqCommon)
add_library(vnode STATIC "")
set(
VNODE_SOURCE_FILES
@ -117,6 +118,7 @@ if (${BUILD_CONTRIB})
PUBLIC "inc"
PUBLIC "src/inc"
PUBLIC "${TD_SOURCE_DIR}/include/libs/scalar"
PUBLIC "${TD_SOURCE_DIR}/include/dnode/vnode"
PUBLIC "${TD_SOURCE_DIR}/contrib/rocksdb/include"
)
else()
@ -125,6 +127,7 @@ else()
PUBLIC "inc"
PUBLIC "src/inc"
PUBLIC "${TD_SOURCE_DIR}/include/libs/scalar"
PUBLIC "${TD_SOURCE_DIR}/include/dnode/vnode"
)
if (${TD_LINUX})
target_include_directories(
@ -138,11 +141,6 @@ else()
endif()
endif()
target_include_directories(
vnode
PUBLIC "${TD_SOURCE_DIR}/include/dnode/vnode"
)
target_link_libraries(
vnode
PUBLIC os
@ -163,6 +161,7 @@ target_link_libraries(
PUBLIC transport
PUBLIC stream
PUBLIC index
PUBLIC tqCommon
)
IF (TD_GRANT)

2
source/dnode/vnode/inc/vnode.h
View File

@ -253,8 +253,6 @@ bool tqNextDataBlockFilterOut(STqReader *pReader, SHashObj *filterOutUids);
int32_t tqRetrieveDataBlock(STqReader *pReader, SSDataBlock **pRes, const char *idstr);
int32_t tqRetrieveTaosxBlock(STqReader *pReader, SArray *blocks, SArray *schemas, SSubmitTbData **pSubmitTbDataRet);
int32_t vnodeEnqueueStreamMsg(SVnode *pVnode, SRpcMsg *pMsg);
// sma
int32_t smaGetTSmaDays(SVnodeCfg *pCfg, void *pCont, uint32_t contLen, int32_t *days);

1
source/dnode/vnode/src/inc/tq.h
View File

@ -153,7 +153,6 @@ char* tqOffsetBuildFName(const char* path, int32_t fVer);
int32_t tqOffsetRestoreFromFile(STqOffsetStore* pStore, const char* fname);
// tqStream
int32_t tqResetStreamTaskStatus(STQ* pTq);
int32_t tqStopStreamTasks(STQ* pTq);
// tq util

7
source/dnode/vnode/src/inc/vnodeInt.h
View File

@ -93,7 +93,11 @@ typedef struct SQueryNode SQueryNode;
#define VNODE_RSMA2_DIR "rsma2"
#define VNODE_TQ_STREAM "stream"
#if SUSPEND_RESUME_TEST // only for test purpose
#define VNODE_BUFPOOL_SEGMENTS 1
#else
#define VNODE_BUFPOOL_SEGMENTS 3
#endif
#define VND_INFO_FNAME "vnode.json"
#define VND_INFO_FNAME_TMP "vnode_tmp.json"
@ -234,7 +238,6 @@ int32_t tqProcessTaskUpdateReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskResetReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDropHTask(STQ* pTq, SRpcMsg* pMsg);
int32_t tqStartStreamTaskAsync(STQ* pTq, bool restart);
int32_t tqRestartStreamTasks(STQ* pTq);
int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t ver);
int32_t tqScanWal(STQ* pTq);
@ -263,7 +266,7 @@ int32_t tqProcessTaskResumeReq(STQ* pTq, int64_t version, char* msg, int32_t msg
int32_t tqProcessTaskCheckReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskCheckRsp(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg, bool exec);
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskRetrieveReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskRetrieveRsp(STQ* pTq, SRpcMsg* pMsg);

661
source/dnode/vnode/src/tq/tq.c
View File

@ -14,8 +14,8 @@
*/
#include "tq.h"
#include "stream.h"
#include "vnd.h"
#include "tqCommon.h"
typedef struct {
int8_t inited;
@ -891,175 +891,15 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t nextProcessVer) {
}
int32_t tqProcessTaskCheckReq(STQ* pTq, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamMeta* pMeta = pTq->pStreamMeta;
SStreamTaskCheckReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamTaskCheckReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t taskId = req.downstreamTaskId;
SStreamTaskCheckRsp rsp = {
.reqId = req.reqId,
.streamId = req.streamId,
.childId = req.childId,
.downstreamNodeId = req.downstreamNodeId,
.downstreamTaskId = req.downstreamTaskId,
.upstreamNodeId = req.upstreamNodeId,
.upstreamTaskId = req.upstreamTaskId,
};
// only the leader node handle the check request
if (pMeta->role == NODE_ROLE_FOLLOWER) {
tqError(
"s-task:0x%x invalid check msg from upstream:0x%x(vgId:%d), vgId:%d is follower, not handle check status msg",
taskId, req.upstreamTaskId, req.upstreamNodeId, pMeta->vgId);
rsp.status = TASK_DOWNSTREAM_NOT_LEADER;
} else {
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, taskId);
if (pTask != NULL) {
rsp.status = streamTaskCheckStatus(pTask, req.upstreamTaskId, req.upstreamNodeId, req.stage, &rsp.oldStage);
streamMetaReleaseTask(pMeta, pTask);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
tqDebug("s-task:%s status:%s, stage:%" PRId64 " recv task check req(reqId:0x%" PRIx64
") task:0x%x (vgId:%d), check_status:%d",
pTask->id.idStr, p, rsp.oldStage, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
} else {
rsp.status = TASK_DOWNSTREAM_NOT_READY;
tqDebug("tq recv task check(taskId:0x%" PRIx64 "-0x%x not built yet) req(reqId:0x%" PRIx64
") from task:0x%x (vgId:%d), rsp check_status %d",
req.streamId, taskId, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
}
}
return streamSendCheckRsp(pMeta, &req, &rsp, &pMsg->info, taskId);
return tqStreamTaskProcessCheckReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskCheckRsp(STQ* pTq, SRpcMsg* pMsg) {
char* pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t vgId = pTq->pStreamMeta->vgId;
int32_t code;
SStreamTaskCheckRsp rsp;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)pReq, len);
code = tDecodeStreamTaskCheckRsp(&decoder, &rsp);
if (code < 0) {
terrno = TSDB_CODE_INVALID_MSG;
tDecoderClear(&decoder);
tqError("vgId:%d failed to parse check rsp msg, code:%s", vgId, tstrerror(terrno));
return -1;
}
tDecoderClear(&decoder);
tqDebug("tq task:0x%x (vgId:%d) recv check rsp(reqId:0x%" PRIx64 ") from 0x%x (vgId:%d) status %d",
rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.reqId, rsp.downstreamTaskId, rsp.downstreamNodeId, rsp.status);
if (!vnodeIsRoleLeader(pTq->pVnode)) {
tqError("vgId:%d not leader, task:0x%x not handle the check rsp, downstream:0x%x (vgId:%d)", vgId,
rsp.upstreamTaskId, rsp.downstreamTaskId, rsp.downstreamNodeId);
return code;
}
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, rsp.streamId, rsp.upstreamTaskId);
if (pTask == NULL) {
tqError("tq failed to locate the stream task:0x%" PRIx64 "-0x%x (vgId:%d), it may have been destroyed or stopped",
rsp.streamId, rsp.upstreamTaskId, pTq->pStreamMeta->vgId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return -1;
}
code = streamProcessCheckRsp(pTask, &rsp);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return code;
return tqStreamTaskProcessCheckRsp(pTq->pStreamMeta, pMsg, vnodeIsRoleLeader(pTq->pVnode));
}
int32_t tqProcessTaskDeployReq(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen) {
int32_t code = 0;
int32_t vgId = TD_VID(pTq->pVnode);
if (tsDisableStream) {
tqInfo("vgId:%d stream disabled, not deploy stream tasks", vgId);
return code;
}
tqDebug("vgId:%d receive new stream task deploy msg, start to build stream task", vgId);
// 1.deserialize msg and build task
int32_t size = sizeof(SStreamTask);
SStreamTask* pTask = taosMemoryCalloc(1, size);
if (pTask == NULL) {
tqError("vgId:%d failed to create stream task due to out of memory, alloc size:%d", vgId, size);
return TSDB_CODE_OUT_OF_MEMORY;
}
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
code = tDecodeStreamTask(&decoder, pTask);
tDecoderClear(&decoder);
if (code != TSDB_CODE_SUCCESS) {
taosMemoryFree(pTask);
return TSDB_CODE_INVALID_MSG;
}
SStreamMeta* pStreamMeta = pTq->pStreamMeta;
// 2.save task, use the latest commit version as the initial start version of stream task.
int32_t taskId = pTask->id.taskId;
int64_t streamId = pTask->id.streamId;
bool added = false;
streamMetaWLock(pStreamMeta);
code = streamMetaRegisterTask(pStreamMeta, sversion, pTask, &added);
int32_t numOfTasks = streamMetaGetNumOfTasks(pStreamMeta);
streamMetaWUnLock(pStreamMeta);
if (code < 0) {
tqError("failed to add s-task:0x%x into vgId:%d meta, total:%d, code:%s", vgId, taskId, numOfTasks,
tstrerror(code));
tFreeStreamTask(pTask);
return code;
}
// added into meta store, pTask cannot be reference since it may have been destroyed by other threads already now if
// it is added into the meta store
if (added) {
// only handled in the leader node
if (vnodeIsRoleLeader(pTq->pVnode)) {
tqDebug("vgId:%d s-task:0x%x is deployed and add into meta, numOfTasks:%d", vgId, taskId, numOfTasks);
SStreamTask* p = streamMetaAcquireTask(pStreamMeta, streamId, taskId);
bool restored = pTq->pVnode->restored;
if (p != NULL && restored && p->info.fillHistory == 0) {
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(p)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
streamTaskHandleEvent(p->status.pSM, event);
} else if (!restored) {
tqWarn("s-task:%s not launched since vnode(vgId:%d) not ready", p->id.idStr, vgId);
}
if (p != NULL) {
streamMetaReleaseTask(pStreamMeta, p);
}
} else {
tqDebug("vgId:%d not leader, not launch stream task s-task:0x%x", vgId, taskId);
}
} else {
tqWarn("vgId:%d failed to add s-task:0x%x, since already exists in meta store", vgId, taskId);
tFreeStreamTask(pTask);
}
return code;
return tqStreamTaskProcessDeployReq(pTq->pStreamMeta, sversion, msg, msgLen, vnodeIsRoleLeader(pTq->pVnode), pTq->pVnode->restored);
}
static void doStartFillhistoryStep2(SStreamTask* pTask, SStreamTask* pStreamTask, STQ* pTq) {
@ -1241,190 +1081,39 @@ int32_t tqProcessTaskScanHistory(STQ* pTq, SRpcMsg* pMsg) {
// only the agg tasks and the sink tasks will receive this message from upstream tasks
int32_t tqProcessTaskScanHistoryFinishReq(STQ* pTq, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamScanHistoryFinishReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeStreamScanHistoryFinishReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.downstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish msg, failed to find task:0x%x, it may be destroyed",
pTq->pStreamMeta->vgId, req.downstreamTaskId);
return -1;
}
tqDebug("s-task:%s receive scan-history finish msg from task:0x%x", pTask->id.idStr, req.upstreamTaskId);
int32_t code = streamProcessScanHistoryFinishReq(pTask, &req, &pMsg->info);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return code;
return tqStreamTaskProcessScanHistoryFinishReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskScanHistoryFinishRsp(STQ* pTq, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamCompleteHistoryMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeCompleteHistoryDataMsg(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish rsp, failed to find task:0x%x, it may be destroyed",
pTq->pStreamMeta->vgId, req.upstreamTaskId);
return -1;
}
int32_t remain = atomic_sub_fetch_32(&pTask->notReadyTasks, 1);
if (remain > 0) {
tqDebug("s-task:%s scan-history finish rsp received from downstream task:0x%x, unfinished remain:%d",
pTask->id.idStr, req.downstreamId, remain);
} else {
tqDebug(
"s-task:%s scan-history finish rsp received from downstream task:0x%x, all downstream tasks rsp scan-history "
"completed msg",
pTask->id.idStr, req.downstreamId);
streamProcessScanHistoryFinishRsp(pTask);
}
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return 0;
return tqStreamTaskProcessScanHistoryFinishRsp(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg) {
SStreamTaskRunReq* pReq = pMsg->pCont;
int32_t taskId = pReq->taskId;
int32_t vgId = TD_VID(pTq->pVnode);
if (taskId == STREAM_EXEC_EXTRACT_DATA_IN_WAL_ID) { // all tasks are extracted submit data from the wal
tqScanWal(pTq);
return 0;
} else if (taskId == STREAM_EXEC_START_ALL_TASKS_ID) {
tqStartStreamTasks(pTq);
return 0;
} else if (taskId == STREAM_EXEC_RESTART_ALL_TASKS_ID) {
tqRestartStreamTasks(pTq);
return 0;
}
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, pReq->streamId, taskId);
if (pTask != NULL) { // even in halt status, the data in inputQ must be processed
char* p = NULL;
if (streamTaskReadyToRun(pTask, &p)) {
tqDebug("vgId:%d s-task:%s start to process block from inputQ, next checked ver:%" PRId64, vgId, pTask->id.idStr,
pTask->chkInfo.nextProcessVer);
streamExecTask(pTask);
} else {
int8_t status = streamTaskSetSchedStatusInactive(pTask);
tqDebug("vgId:%d s-task:%s ignore run req since not in ready state, status:%s, sched-status:%d", vgId,
pTask->id.idStr, p, status);
}
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
int32_t code = tqStreamTaskProcessRunReq(pTq->pStreamMeta, pMsg, vnodeIsRoleLeader(pTq->pVnode));
if(code == 0 && taskId > 0){
tqScanWalAsync(pTq, false);
return 0;
} else { // NOTE: pTask->status.schedStatus is not updated since it is not be handled by the run exec.
// todo add one function to handle this
tqError("vgId:%d failed to found s-task, taskId:0x%x may have been dropped", vgId, taskId);
return -1;
}
return code;
}
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg, bool exec) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamDispatchReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamDispatchReq(&decoder, &req);
tDecoderClear(&decoder);
tqDebug("s-task:0x%x recv dispatch msg from 0x%x(vgId:%d)", req.taskId, req.upstreamTaskId, req.upstreamNodeId);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return 0;
} else {
tqError("vgId:%d failed to find task:0x%x to handle the dispatch req, it may have been destroyed already",
pTq->pStreamMeta->vgId, req.taskId);
tDeleteStreamDispatchReq(&req);
return -1;
}
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg) {
return tqStreamTaskProcessDispatchReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskDispatchRsp(STQ* pTq, SRpcMsg* pMsg) {
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t vgId = pTq->pStreamMeta->vgId;
pRsp->upstreamTaskId = htonl(pRsp->upstreamTaskId);
pRsp->streamId = htobe64(pRsp->streamId);
pRsp->downstreamTaskId = htonl(pRsp->downstreamTaskId);
pRsp->downstreamNodeId = htonl(pRsp->downstreamNodeId);
pRsp->stage = htobe64(pRsp->stage);
pRsp->msgId = htonl(pRsp->msgId);
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, pRsp->streamId, pRsp->upstreamTaskId);
if (pTask) {
streamProcessDispatchRsp(pTask, pRsp, pMsg->code);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return TSDB_CODE_SUCCESS;
} else {
tqDebug("vgId:%d failed to handle the dispatch rsp, since find task:0x%x failed", vgId, pRsp->upstreamTaskId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return terrno;
}
return tqStreamTaskProcessDispatchRsp(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskDropReq(STQ* pTq, char* msg, int32_t msgLen) {
SVDropStreamTaskReq* pReq = (SVDropStreamTaskReq*)msg;
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
tqDebug("vgId:%d receive msg to drop s-task:0x%x", vgId, pReq->taskId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, pReq->taskId);
if (pTask != NULL) {
// drop the related fill-history task firstly
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
STaskId* pHTaskId = &pTask->hTaskInfo.id;
streamMetaUnregisterTask(pMeta, pHTaskId->streamId, pHTaskId->taskId);
tqDebug("vgId:%d drop fill-history task:0x%x dropped firstly", vgId, (int32_t)pHTaskId->taskId);
}
streamMetaReleaseTask(pMeta, pTask);
}
// drop the stream task now
streamMetaUnregisterTask(pMeta, pReq->streamId, pReq->taskId);
// commit the update
streamMetaWLock(pMeta);
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
tqDebug("vgId:%d task:0x%x dropped, remain tasks:%d", vgId, pReq->taskId, numOfTasks);
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
streamMetaWUnLock(pMeta);
return 0;
return tqStreamTaskProcessDropReq(pTq->pStreamMeta, msg, msgLen);
}
int32_t tqProcessTaskPauseReq(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen) {
@ -1533,30 +1222,7 @@ int32_t tqProcessTaskResumeReq(STQ* pTq, int64_t sversion, char* msg, int32_t ms
}
int32_t tqProcessTaskRetrieveReq(STQ* pTq, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SDecoder decoder;
SStreamRetrieveReq req;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamRetrieveReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t vgId = pTq->pStreamMeta->vgId;
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, req.dstTaskId);
if (pTask == NULL) {
tqError("vgId:%d process retrieve req, failed to acquire task:0x%x, it may have been dropped already", vgId,
req.dstTaskId);
return -1;
}
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessRetrieveReq(pTask, &req, &rsp);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
tDeleteStreamRetrieveReq(&req);
return 0;
return tqStreamTaskProcessRetrieveReq(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskRetrieveRsp(STQ* pTq, SRpcMsg* pMsg) {
@ -1564,89 +1230,6 @@ int32_t tqProcessTaskRetrieveRsp(STQ* pTq, SRpcMsg* pMsg) {
return 0;
}
// todo refactor.
int32_t vnodeEnqueueStreamMsg(SVnode* pVnode, SRpcMsg* pMsg) {
STQ* pTq = pVnode->pTq;
int32_t vgId = pVnode->config.vgId;
SMsgHead* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamDispatchReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
if (tDecodeStreamDispatchReq(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
goto FAIL;
}
tDecoderClear(&decoder);
int32_t taskId = req.taskId;
tqDebug("vgId:%d receive dispatch msg to s-task:0x%" PRIx64 "-0x%x", vgId, req.streamId, taskId);
// for test purpose
// if (req.type == STREAM_INPUT__CHECKPOINT_TRIGGER) {
// code = TSDB_CODE_STREAM_TASK_NOT_EXIST;
// goto FAIL;
// }
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, req.streamId, taskId);
if (pTask != NULL) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessDispatchMsg(pTask, &req, &rsp);
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return 0;
} else {
tDeleteStreamDispatchReq(&req);
}
code = TSDB_CODE_STREAM_TASK_NOT_EXIST;
FAIL:
if (pMsg->info.handle == NULL) {
tqError("s-task:0x%x vgId:%d msg handle is null, abort enqueue dispatch msg", vgId, taskId);
return -1;
}
SMsgHead* pRspHead = rpcMallocCont(sizeof(SMsgHead) + sizeof(SStreamDispatchRsp));
if (pRspHead == NULL) {
SRpcMsg rsp = {.code = TSDB_CODE_OUT_OF_MEMORY, .info = pMsg->info};
tqError("s-task:0x%x send dispatch error rsp, code:%s", taskId, tstrerror(code));
tmsgSendRsp(&rsp);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return -1;
}
pRspHead->vgId = htonl(req.upstreamNodeId);
ASSERT(pRspHead->vgId != 0);
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pRspHead, sizeof(SMsgHead));
pRsp->streamId = htobe64(req.streamId);
pRsp->upstreamTaskId = htonl(req.upstreamTaskId);
pRsp->upstreamNodeId = htonl(req.upstreamNodeId);
pRsp->downstreamNodeId = htonl(pVnode->config.vgId);
pRsp->downstreamTaskId = htonl(req.taskId);
pRsp->msgId = htonl(req.msgId);
pRsp->stage = htobe64(req.stage);
pRsp->inputStatus = TASK_OUTPUT_STATUS__NORMAL;
int32_t len = sizeof(SMsgHead) + sizeof(SStreamDispatchRsp);
SRpcMsg rsp = {.code = code, .info = pMsg->info, .contLen = len, .pCont = pRspHead};
tqError("s-task:0x%x send dispatch error rsp, code:%s", taskId, tstrerror(code));
tmsgSendRsp(&rsp);
rpcFreeCont(pMsg->pCont);
taosFreeQitem(pMsg);
return -1;
}
int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp) {
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
@ -1776,220 +1359,11 @@ int32_t tqProcessTaskCheckPointSourceReq(STQ* pTq, SRpcMsg* pMsg, SRpcMsg* pRsp)
// downstream task has complete the stream task checkpoint procedure, let's start the handle the rsp by execute task
int32_t tqProcessTaskCheckpointReadyMsg(STQ* pTq, SRpcMsg* pMsg) {
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamCheckpointReadyMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamCheckpointReadyMsg(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d failed to find s-task:0x%x, it may have been destroyed already", vgId, req.downstreamTaskId);
return code;
}
tqDebug("vgId:%d s-task:%s received the checkpoint ready msg from task:0x%x (vgId:%d), handle it", vgId,
pTask->id.idStr, req.downstreamTaskId, req.downstreamNodeId);
streamProcessCheckpointReadyMsg(pTask);
streamMetaReleaseTask(pMeta, pTask);
return code;
return tqStreamTaskProcessCheckpointReadyMsg(pTq->pStreamMeta, pMsg);
}
int32_t tqProcessTaskUpdateReq(STQ* pTq, SRpcMsg* pMsg) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = TD_VID(pTq->pVnode);
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
SRpcMsg rsp = {.info = pMsg->info, .code = TSDB_CODE_SUCCESS};
SStreamTaskNodeUpdateMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamTaskUpdateMsg(&decoder, &req) < 0) {
rsp.code = TSDB_CODE_MSG_DECODE_ERROR;
tqError("vgId:%d failed to decode task update msg, code:%s", vgId, tstrerror(rsp.code));
tDecoderClear(&decoder);
return rsp.code;
}
tDecoderClear(&decoder);
// update the nodeEpset when it exists
streamMetaWLock(pMeta);
// the task epset may be updated again and again, when replaying the WAL, the task may be in stop status.
STaskId id = {.streamId = req.streamId, .taskId = req.taskId};
SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
if (ppTask == NULL || *ppTask == NULL) {
tqError("vgId:%d failed to acquire task:0x%x when handling update, it may have been dropped already", pMeta->vgId,
req.taskId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
SStreamTask* pTask = *ppTask;
if (pMeta->updateInfo.transId != req.transId) {
pMeta->updateInfo.transId = req.transId;
tqInfo("s-task:%s receive new trans to update nodeEp msg from mnode, transId:%d", pTask->id.idStr, req.transId);
// info needs to be kept till the new trans to update the nodeEp arrived.
taosHashClear(pMeta->updateInfo.pTasks);
} else {
tqDebug("s-task:%s recv trans to update nodeEp from mnode, transId:%d", pTask->id.idStr, req.transId);
}
STaskUpdateEntry entry = {.streamId = req.streamId, .taskId = req.taskId, .transId = req.transId};
void* exist = taosHashGet(pMeta->updateInfo.pTasks, &entry, sizeof(STaskUpdateEntry));
if (exist != NULL) {
tqDebug("s-task:%s (vgId:%d) already update in trans:%d, discard the nodeEp update msg", pTask->id.idStr, vgId,
req.transId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
streamMetaWUnLock(pMeta);
// the following two functions should not be executed within the scope of meta lock to avoid deadlock
streamTaskUpdateEpsetInfo(pTask, req.pNodeList);
streamTaskResetStatus(pTask);
// continue after lock the meta again
streamMetaWLock(pMeta);
SStreamTask** ppHTask = NULL;
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
ppHTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &pTask->hTaskInfo.id, sizeof(pTask->hTaskInfo.id));
if (ppHTask == NULL || *ppHTask == NULL) {
tqError("vgId:%d failed to acquire fill-history task:0x%x when handling update, it may have been dropped already",
pMeta->vgId, req.taskId);
CLEAR_RELATED_FILLHISTORY_TASK(pTask);
} else {
tqDebug("s-task:%s fill-history task update nodeEp along with stream task", (*ppHTask)->id.idStr);
streamTaskUpdateEpsetInfo(*ppHTask, req.pNodeList);
}
}
{
streamMetaSaveTask(pMeta, pTask);
if (ppHTask != NULL) {
streamMetaSaveTask(pMeta, *ppHTask);
}
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
}
streamTaskStop(pTask);
// keep the already handled info
taosHashPut(pMeta->updateInfo.pTasks, &entry, sizeof(entry), NULL, 0);
if (ppHTask != NULL) {
streamTaskStop(*ppHTask);
tqDebug("s-task:%s task nodeEp update completed, streamTask and related fill-history task closed", pTask->id.idStr);
taosHashPut(pMeta->updateInfo.pTasks, &(*ppHTask)->id, sizeof(pTask->id), NULL, 0);
} else {
tqDebug("s-task:%s task nodeEp update completed, streamTask closed", pTask->id.idStr);
}
rsp.code = 0;
// possibly only handle the stream task.
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
int32_t updateTasks = taosHashGetSize(pMeta->updateInfo.pTasks);
pMeta->startInfo.tasksWillRestart = 1;
if (updateTasks < numOfTasks) {
tqDebug("vgId:%d closed tasks:%d, unclosed:%d, all tasks will be started when nodeEp update completed", vgId,
updateTasks, (numOfTasks - updateTasks));
streamMetaWUnLock(pMeta);
} else {
if (!pTq->pVnode->restored) {
tqDebug("vgId:%d vnode restore not completed, not restart the tasks, clear the start after nodeUpdate flag",
vgId);
pMeta->startInfo.tasksWillRestart = 0;
streamMetaWUnLock(pMeta);
} else {
tqDebug("vgId:%d all %d task(s) nodeEp updated and closed", vgId, numOfTasks);
#if 1
tqStartStreamTaskAsync(pTq, true);
streamMetaWUnLock(pMeta);
#else
streamMetaWUnLock(pMeta);
// For debug purpose.
// the following procedure consume many CPU resource, result in the re-election of leader
// with high probability. So we employ it as a test case for the stream processing framework, with
// checkpoint/restart/nodeUpdate etc.
while (1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
int32_t code = streamMetaReopen(pMeta);
if (code != 0) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
streamMetaInitBackend(pMeta);
if (streamMetaLoadAllTasks(pTq->pStreamMeta) < 0) {
tqError("vgId:%d failed to load stream tasks", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d start all stream tasks after all being updated", vgId);
tqResetStreamTaskStatus(pTq);
tqStartStreamTaskAsync(pTq, false);
} else {
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
streamMetaWUnLock(pMeta);
#endif
}
}
taosArrayDestroy(req.pNodeList);
return rsp.code;
return tqStreamTaskProcessUpdateReq(pTq->pStreamMeta, &pTq->pVnode->msgCb, pMsg, pTq->pVnode->restored);
}
int32_t tqProcessTaskResetReq(STQ* pTq, SRpcMsg* pMsg) {
@ -2041,6 +1415,9 @@ int32_t tqProcessTaskDropHTask(STQ* pTq, SRpcMsg* pMsg) {
SStreamTaskId id = {.streamId = pTask->hTaskInfo.id.streamId, .taskId = pTask->hTaskInfo.id.taskId};
streamBuildAndSendDropTaskMsg(pTask->pMsgCb, pMeta->vgId, &id);
// clear the scheduler status
streamTaskSetSchedStatusInactive(pTask);
tqDebug("s-task:%s set scheduler status:%d after drop fill-history task", pTask->id.idStr, pTask->status.schedStatus);
streamMetaReleaseTask(pMeta, pTask);
return TSDB_CODE_SUCCESS;
}

170
source/dnode/vnode/src/tq/tqStreamTask.c
View File

@ -15,7 +15,6 @@
#include "tq.h"
#include "vnd.h"
#include "stream.h"
#define MAX_REPEAT_SCAN_THRESHOLD 3
#define SCAN_WAL_IDLE_DURATION 100
@ -61,154 +60,6 @@ int32_t tqScanWal(STQ* pTq) {
return 0;
}
int32_t tqStartStreamTasks(STQ* pTq) {
int32_t code = TSDB_CODE_SUCCESS;
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d start to check all %d stream task(s) downstream status", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
streamMetaWLock(pMeta);
pTaskList = taosArrayDup(pMeta->pTaskList, NULL);
taosHashClear(pMeta->startInfo.pReadyTaskSet);
taosHashClear(pMeta->startInfo.pFailedTaskSet);
pMeta->startInfo.startTs = taosGetTimestampMs();
streamMetaWUnLock(pMeta);
// broadcast the check downstream tasks msg
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pTaskId->streamId, pTaskId->taskId);
if (pTask == NULL) {
continue;
}
// fill-history task can only be launched by related stream tasks.
if (pTask->info.fillHistory == 1) {
streamMetaReleaseTask(pMeta, pTask);
continue;
}
if (pTask->status.downstreamReady == 1) {
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
tqDebug("s-task:%s downstream ready, no need to check downstream, check only related fill-history task",
pTask->id.idStr);
streamLaunchFillHistoryTask(pTask);
}
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, pTask->execInfo.start, true);
streamMetaReleaseTask(pMeta, pTask);
continue;
}
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
int32_t ret = streamTaskHandleEvent(pTask->status.pSM, event);
if (ret != TSDB_CODE_SUCCESS) {
code = ret;
}
streamMetaReleaseTask(pMeta, pTask);
}
taosArrayDestroy(pTaskList);
return code;
}
int32_t tqRestartStreamTasks(STQ* pTq) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = pMeta->vgId;
int32_t code = 0;
int64_t st = taosGetTimestampMs();
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
terrno = 0;
tqInfo("vgId:%d tasks are all updated and stopped, restart all tasks, triggered by transId:%d", vgId,
pMeta->updateInfo.transId);
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
code = streamMetaReopen(pMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
streamMetaInitBackend(pMeta);
int64_t el = taosGetTimestampMs() - st;
tqInfo("vgId:%d close&reload state elapsed time:%.3fs", vgId, el/1000.);
code = streamMetaLoadAllTasks(pTq->pStreamMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to load stream tasks, code:%s", vgId, tstrerror(terrno));
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d restart all stream tasks after all tasks being updated", vgId);
tqResetStreamTaskStatus(pTq);
streamMetaWUnLock(pMeta);
tqStartStreamTasks(pTq);
} else {
streamMetaResetStartInfo(&pMeta->startInfo);
streamMetaWUnLock(pMeta);
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
code = terrno;
return code;
}
int32_t tqStartStreamTaskAsync(STQ* pTq, bool restart) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
tqDebug("vgId:%d no stream tasks existed to run", vgId);
return 0;
}
SStreamTaskRunReq* pRunReq = rpcMallocCont(sizeof(SStreamTaskRunReq));
if (pRunReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("vgId:%d failed to create msg to start wal scanning to launch stream tasks, code:%s", vgId, terrstr());
return -1;
}
tqDebug("vgId:%d start all %d stream task(s) async", vgId, numOfTasks);
pRunReq->head.vgId = vgId;
pRunReq->streamId = 0;
pRunReq->taskId = restart? STREAM_EXEC_RESTART_ALL_TASKS_ID:STREAM_EXEC_START_ALL_TASKS_ID;
SRpcMsg msg = {.msgType = TDMT_STREAM_TASK_RUN, .pCont = pRunReq, .contLen = sizeof(SStreamTaskRunReq)};
tmsgPutToQueue(&pTq->pVnode->msgCb, STREAM_QUEUE, &msg);
return 0;
}
int32_t tqScanWalAsync(STQ* pTq, bool ckPause) {
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
@ -302,27 +153,6 @@ int32_t tqStopStreamTasks(STQ* pTq) {
return 0;
}
int32_t tqResetStreamTaskStatus(STQ* pTq) {
SStreamMeta* pMeta = pTq->pStreamMeta;
int32_t vgId = TD_VID(pTq->pVnode);
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d reset all %d stream task(s) status to be uninit", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pMeta->pTaskList, i);
STaskId id = {.streamId = pTaskId->streamId, .taskId = pTaskId->taskId};
SStreamTask** pTask = taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
streamTaskResetStatus(*pTask);
}
return 0;
}
int32_t setWalReaderStartOffset(SStreamTask* pTask, int32_t vgId) {
// seek the stored version and extract data from WAL
int64_t firstVer = walReaderGetValidFirstVer(pTask->exec.pWalReader);

20
source/dnode/vnode/src/tqCommon/CMakeLists.txt Normal file
View File

@ -0,0 +1,20 @@
aux_source_directory(. TQ_SOURCE_FILES)
add_library(tqCommon STATIC ${TQ_SOURCE_FILES})
target_include_directories(
tqCommon
PUBLIC "../inc"
PUBLIC "../../inc"
)
target_link_libraries(
tqCommon
PRIVATE stream
PRIVATE common
PRIVATE transport
PRIVATE executor
PRIVATE index
PRIVATE qcom
PRIVATE qworker
PRIVATE sync
PRIVATE tfs
)

809
source/dnode/vnode/src/tqCommon/tqCommon.c Normal file
View File

@ -0,0 +1,809 @@
/*
* 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 "tstream.h"
#include "tmsgcb.h"
#include "tq.h"
typedef struct STaskUpdateEntry {
int64_t streamId;
int32_t taskId;
int32_t transId;
} STaskUpdateEntry;
int32_t tqStreamTaskStartAsync(SStreamMeta* pMeta, SMsgCb* cb, bool restart) {
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
tqDebug("vgId:%d no stream tasks existed to run", vgId);
return 0;
}
SStreamTaskRunReq* pRunReq = rpcMallocCont(sizeof(SStreamTaskRunReq));
if (pRunReq == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("vgId:%d failed to create msg to start wal scanning to launch stream tasks, code:%s", vgId, terrstr());
return -1;
}
tqDebug("vgId:%d start all %d stream task(s) async", vgId, numOfTasks);
pRunReq->head.vgId = vgId;
pRunReq->streamId = 0;
pRunReq->taskId = restart? STREAM_EXEC_RESTART_ALL_TASKS_ID:STREAM_EXEC_START_ALL_TASKS_ID;
SRpcMsg msg = {.msgType = TDMT_STREAM_TASK_RUN, .pCont = pRunReq, .contLen = sizeof(SStreamTaskRunReq)};
tmsgPutToQueue(cb, STREAM_QUEUE, &msg);
return 0;
}
int32_t tqStreamTaskProcessUpdateReq(SStreamMeta* pMeta, SMsgCb* cb, SRpcMsg* pMsg, bool restored) {
int32_t vgId = pMeta->vgId;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
SRpcMsg rsp = {.info = pMsg->info, .code = TSDB_CODE_SUCCESS};
SStreamTaskNodeUpdateMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamTaskUpdateMsg(&decoder, &req) < 0) {
rsp.code = TSDB_CODE_MSG_DECODE_ERROR;
tqError("vgId:%d failed to decode task update msg, code:%s", vgId, tstrerror(rsp.code));
tDecoderClear(&decoder);
return rsp.code;
}
tDecoderClear(&decoder);
// update the nodeEpset when it exists
streamMetaWLock(pMeta);
// the task epset may be updated again and again, when replaying the WAL, the task may be in stop status.
STaskId id = {.streamId = req.streamId, .taskId = req.taskId};
SStreamTask** ppTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
if (ppTask == NULL || *ppTask == NULL) {
tqError("vgId:%d failed to acquire task:0x%x when handling update, it may have been dropped already", pMeta->vgId,
req.taskId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
SStreamTask* pTask = *ppTask;
if (pMeta->updateInfo.transId != req.transId) {
pMeta->updateInfo.transId = req.transId;
tqInfo("s-task:%s receive new trans to update nodeEp msg from mnode, transId:%d", pTask->id.idStr, req.transId);
// info needs to be kept till the new trans to update the nodeEp arrived.
taosHashClear(pMeta->updateInfo.pTasks);
} else {
tqDebug("s-task:%s recv trans to update nodeEp from mnode, transId:%d", pTask->id.idStr, req.transId);
}
STaskUpdateEntry entry = {.streamId = req.streamId, .taskId = req.taskId, .transId = req.transId};
void* exist = taosHashGet(pMeta->updateInfo.pTasks, &entry, sizeof(STaskUpdateEntry));
if (exist != NULL) {
tqDebug("s-task:%s (vgId:%d) already update in trans:%d, discard the nodeEp update msg", pTask->id.idStr, vgId,
req.transId);
rsp.code = TSDB_CODE_SUCCESS;
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
streamMetaWUnLock(pMeta);
// the following two functions should not be executed within the scope of meta lock to avoid deadlock
streamTaskUpdateEpsetInfo(pTask, req.pNodeList);
streamTaskResetStatus(pTask);
// continue after lock the meta again
streamMetaWLock(pMeta);
SStreamTask** ppHTask = NULL;
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
ppHTask = (SStreamTask**)taosHashGet(pMeta->pTasksMap, &pTask->hTaskInfo.id, sizeof(pTask->hTaskInfo.id));
if (ppHTask == NULL || *ppHTask == NULL) {
tqError("vgId:%d failed to acquire fill-history task:0x%x when handling update, it may have been dropped already",
pMeta->vgId, req.taskId);
CLEAR_RELATED_FILLHISTORY_TASK(pTask);
} else {
tqDebug("s-task:%s fill-history task update nodeEp along with stream task", (*ppHTask)->id.idStr);
streamTaskUpdateEpsetInfo(*ppHTask, req.pNodeList);
}
}
{
streamMetaSaveTask(pMeta, pTask);
if (ppHTask != NULL) {
streamMetaSaveTask(pMeta, *ppHTask);
}
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
}
streamTaskStop(pTask);
// keep the already handled info
taosHashPut(pMeta->updateInfo.pTasks, &entry, sizeof(entry), NULL, 0);
if (ppHTask != NULL) {
streamTaskStop(*ppHTask);
tqDebug("s-task:%s task nodeEp update completed, streamTask and related fill-history task closed", pTask->id.idStr);
taosHashPut(pMeta->updateInfo.pTasks, &(*ppHTask)->id, sizeof(pTask->id), NULL, 0);
} else {
tqDebug("s-task:%s task nodeEp update completed, streamTask closed", pTask->id.idStr);
}
rsp.code = 0;
// possibly only handle the stream task.
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
int32_t updateTasks = taosHashGetSize(pMeta->updateInfo.pTasks);
pMeta->startInfo.tasksWillRestart = 1;
if (updateTasks < numOfTasks) {
tqDebug("vgId:%d closed tasks:%d, unclosed:%d, all tasks will be started when nodeEp update completed", vgId,
updateTasks, (numOfTasks - updateTasks));
streamMetaWUnLock(pMeta);
} else {
if (!restored) {
tqDebug("vgId:%d vnode restore not completed, not restart the tasks, clear the start after nodeUpdate flag", vgId);
pMeta->startInfo.tasksWillRestart = 0;
streamMetaWUnLock(pMeta);
} else {
tqDebug("vgId:%d all %d task(s) nodeEp updated and closed", vgId, numOfTasks);
#if 1
tqStreamTaskStartAsync(pMeta, cb, true);
streamMetaWUnLock(pMeta);
#else
streamMetaWUnLock(pMeta);
// For debug purpose.
// the following procedure consume many CPU resource, result in the re-election of leader
// with high probability. So we employ it as a test case for the stream processing framework, with
// checkpoint/restart/nodeUpdate etc.
while (1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
int32_t code = streamMetaReopen(pMeta);
if (code != 0) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
streamMetaInitBackend(pMeta);
if (streamMetaLoadAllTasks(pTq->pStreamMeta) < 0) {
tqError("vgId:%d failed to load stream tasks", vgId);
streamMetaWUnLock(pMeta);
taosArrayDestroy(req.pNodeList);
return -1;
}
if (vnodeIsRoleLeader(pTq->pVnode) && !tsDisableStream) {
tqInfo("vgId:%d start all stream tasks after all being updated", vgId);
resetStreamTaskStatus(pTq->pStreamMeta);
tqStartStreamTaskAsync(pTq, false);
} else {
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
streamMetaWUnLock(pMeta);
#endif
}
}
taosArrayDestroy(req.pNodeList);
return rsp.code;
}
int32_t tqStreamTaskProcessDispatchReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamDispatchReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
if (tDecodeStreamDispatchReq(&decoder, &req) < 0) {
tDecoderClear(&decoder);
return TSDB_CODE_MSG_DECODE_ERROR;
}
tDecoderClear(&decoder);
tqDebug("s-task:0x%x recv dispatch msg from 0x%x(vgId:%d)", req.taskId, req.upstreamTaskId, req.upstreamNodeId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.taskId);
if (pTask) {
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
if (streamProcessDispatchMsg(pTask, &req, &rsp) != 0){
return -1;
}
tDeleteStreamDispatchReq(&req);
streamMetaReleaseTask(pMeta, pTask);
return 0;
} else {
tqError("vgId:%d failed to find task:0x%x to handle the dispatch req, it may have been destroyed already",
pMeta->vgId, req.taskId);
SMsgHead* pRspHead = rpcMallocCont(sizeof(SMsgHead) + sizeof(SStreamDispatchRsp));
if (pRspHead == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("s-task:0x%x send dispatch error rsp, out of memory", req.taskId);
return -1;
}
pRspHead->vgId = htonl(req.upstreamNodeId);
ASSERT(pRspHead->vgId != 0);
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pRspHead, sizeof(SMsgHead));
pRsp->streamId = htobe64(req.streamId);
pRsp->upstreamTaskId = htonl(req.upstreamTaskId);
pRsp->upstreamNodeId = htonl(req.upstreamNodeId);
pRsp->downstreamNodeId = htonl(pMeta->vgId);
pRsp->downstreamTaskId = htonl(req.taskId);
pRsp->msgId = htonl(req.msgId);
pRsp->stage = htobe64(req.stage);
pRsp->inputStatus = TASK_OUTPUT_STATUS__NORMAL;
int32_t len = sizeof(SMsgHead) + sizeof(SStreamDispatchRsp);
SRpcMsg rsp = {.code = TSDB_CODE_STREAM_TASK_NOT_EXIST, .info = pMsg->info, .contLen = len, .pCont = pRspHead};
tqError("s-task:0x%x send dispatch error rsp, no task", req.taskId);
tmsgSendRsp(&rsp);
tDeleteStreamDispatchReq(&req);
return 0;
}
}
int32_t tqStreamTaskProcessDispatchRsp(SStreamMeta* pMeta, SRpcMsg* pMsg) {
SStreamDispatchRsp* pRsp = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t vgId = pMeta->vgId;
pRsp->upstreamTaskId = htonl(pRsp->upstreamTaskId);
pRsp->streamId = htobe64(pRsp->streamId);
pRsp->downstreamTaskId = htonl(pRsp->downstreamTaskId);
pRsp->downstreamNodeId = htonl(pRsp->downstreamNodeId);
pRsp->stage = htobe64(pRsp->stage);
pRsp->msgId = htonl(pRsp->msgId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pRsp->streamId, pRsp->upstreamTaskId);
if (pTask) {
streamProcessDispatchRsp(pTask, pRsp, pMsg->code);
streamMetaReleaseTask(pMeta, pTask);
return TSDB_CODE_SUCCESS;
} else {
tqDebug("vgId:%d failed to handle the dispatch rsp, since find task:0x%x failed", vgId, pRsp->upstreamTaskId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return terrno;
}
}
int32_t tqStreamTaskProcessRetrieveReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SDecoder decoder;
SStreamRetrieveReq req;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamRetrieveReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.dstTaskId);
if (pTask == NULL) {
tqError("vgId:%d process retrieve req, failed to acquire task:0x%x, it may have been dropped already", pMeta->vgId,
req.dstTaskId);
return -1;
}
SRpcMsg rsp = {.info = pMsg->info, .code = 0};
streamProcessRetrieveReq(pTask, &req, &rsp);
streamMetaReleaseTask(pMeta, pTask);
tDeleteStreamRetrieveReq(&req);
return 0;
}
int32_t tqStreamTaskProcessScanHistoryFinishReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamScanHistoryFinishReq req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeStreamScanHistoryFinishReq(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.downstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish msg, failed to find task:0x%x, it may be destroyed",
pMeta->vgId, req.downstreamTaskId);
return -1;
}
tqDebug("s-task:%s receive scan-history finish msg from task:0x%x", pTask->id.idStr, req.upstreamTaskId);
int32_t code = streamProcessScanHistoryFinishReq(pTask, &req, &pMsg->info);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t tqStreamTaskProcessScanHistoryFinishRsp(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
// deserialize
SStreamCompleteHistoryMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
tDecodeCompleteHistoryDataMsg(&decoder, &req);
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d process scan history finish rsp, failed to find task:0x%x, it may be destroyed",
pMeta->vgId, req.upstreamTaskId);
return -1;
}
int32_t remain = atomic_sub_fetch_32(&pTask->notReadyTasks, 1);
if (remain > 0) {
tqDebug("s-task:%s scan-history finish rsp received from downstream task:0x%x, unfinished remain:%d",
pTask->id.idStr, req.downstreamId, remain);
} else {
tqDebug(
"s-task:%s scan-history finish rsp received from downstream task:0x%x, all downstream tasks rsp scan-history "
"completed msg",
pTask->id.idStr, req.downstreamId);
streamProcessScanHistoryFinishRsp(pTask);
}
streamMetaReleaseTask(pMeta, pTask);
return 0;
}
int32_t tqStreamTaskProcessCheckReq(SStreamMeta* pMeta, SRpcMsg* pMsg) {
char* msgStr = pMsg->pCont;
char* msgBody = POINTER_SHIFT(msgStr, sizeof(SMsgHead));
int32_t msgLen = pMsg->contLen - sizeof(SMsgHead);
SStreamTaskCheckReq req;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msgBody, msgLen);
tDecodeStreamTaskCheckReq(&decoder, &req);
tDecoderClear(&decoder);
int32_t taskId = req.downstreamTaskId;
SStreamTaskCheckRsp rsp = {
.reqId = req.reqId,
.streamId = req.streamId,
.childId = req.childId,
.downstreamNodeId = req.downstreamNodeId,
.downstreamTaskId = req.downstreamTaskId,
.upstreamNodeId = req.upstreamNodeId,
.upstreamTaskId = req.upstreamTaskId,
};
// only the leader node handle the check request
if (pMeta->role == NODE_ROLE_FOLLOWER) {
tqError("s-task:0x%x invalid check msg from upstream:0x%x(vgId:%d), vgId:%d is follower, not handle check status msg",
taskId, req.upstreamTaskId, req.upstreamNodeId, pMeta->vgId);
rsp.status = TASK_DOWNSTREAM_NOT_LEADER;
} else {
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, taskId);
if (pTask != NULL) {
rsp.status = streamTaskCheckStatus(pTask, req.upstreamTaskId, req.upstreamNodeId, req.stage, &rsp.oldStage);
streamMetaReleaseTask(pMeta, pTask);
char* p = NULL;
streamTaskGetStatus(pTask, &p);
tqDebug("s-task:%s status:%s, stage:%"PRId64" recv task check req(reqId:0x%" PRIx64 ") task:0x%x (vgId:%d), check_status:%d",
pTask->id.idStr, p, rsp.oldStage, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
} else {
rsp.status = TASK_DOWNSTREAM_NOT_READY;
tqDebug("tq recv task check(taskId:0x%" PRIx64 "-0x%x not built yet) req(reqId:0x%" PRIx64
") from task:0x%x (vgId:%d), rsp check_status %d",
req.streamId, taskId, rsp.reqId, rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.status);
}
}
return streamSendCheckRsp(pMeta, &req, &rsp, &pMsg->info, taskId);
}
int32_t tqStreamTaskProcessCheckRsp(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader) {
char* pReq = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t vgId = pMeta->vgId;
int32_t code;
SStreamTaskCheckRsp rsp;
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)pReq, len);
code = tDecodeStreamTaskCheckRsp(&decoder, &rsp);
if (code < 0) {
terrno = TSDB_CODE_INVALID_MSG;
tDecoderClear(&decoder);
tqError("vgId:%d failed to parse check rsp msg, code:%s", vgId, tstrerror(terrno));
return -1;
}
tDecoderClear(&decoder);
tqDebug("tq task:0x%x (vgId:%d) recv check rsp(reqId:0x%" PRIx64 ") from 0x%x (vgId:%d) status %d",
rsp.upstreamTaskId, rsp.upstreamNodeId, rsp.reqId, rsp.downstreamTaskId, rsp.downstreamNodeId, rsp.status);
if (!isLeader) {
tqError("vgId:%d not leader, task:0x%x not handle the check rsp, downstream:0x%x (vgId:%d)", vgId,
rsp.upstreamTaskId, rsp.downstreamTaskId, rsp.downstreamNodeId);
return code;
}
SStreamTask* pTask = streamMetaAcquireTask(pMeta, rsp.streamId, rsp.upstreamTaskId);
if (pTask == NULL) {
tqError("tq failed to locate the stream task:0x%" PRIx64 "-0x%x (vgId:%d), it may have been destroyed or stopped",
rsp.streamId, rsp.upstreamTaskId, vgId);
terrno = TSDB_CODE_STREAM_TASK_NOT_EXIST;
return -1;
}
code = streamProcessCheckRsp(pTask, &rsp);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t tqStreamTaskProcessCheckpointReadyMsg(SStreamMeta* pMeta, SRpcMsg* pMsg) {
int32_t vgId = pMeta->vgId;
char* msg = POINTER_SHIFT(pMsg->pCont, sizeof(SMsgHead));
int32_t len = pMsg->contLen - sizeof(SMsgHead);
int32_t code = 0;
SStreamCheckpointReadyMsg req = {0};
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, len);
if (tDecodeStreamCheckpointReadyMsg(&decoder, &req) < 0) {
code = TSDB_CODE_MSG_DECODE_ERROR;
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, req.streamId, req.upstreamTaskId);
if (pTask == NULL) {
tqError("vgId:%d failed to find s-task:0x%x, it may have been destroyed already", vgId, req.downstreamTaskId);
return code;
}
tqDebug("vgId:%d s-task:%s received the checkpoint ready msg from task:0x%x (vgId:%d), handle it", vgId,
pTask->id.idStr, req.downstreamTaskId, req.downstreamNodeId);
streamProcessCheckpointReadyMsg(pTask);
streamMetaReleaseTask(pMeta, pTask);
return code;
}
int32_t tqStreamTaskProcessDeployReq(SStreamMeta* pMeta, int64_t sversion, char* msg, int32_t msgLen, bool isLeader, bool restored) {
int32_t code = 0;
int32_t vgId = pMeta->vgId;
if (tsDisableStream) {
tqInfo("vgId:%d stream disabled, not deploy stream tasks", vgId);
return code;
}
tqDebug("vgId:%d receive new stream task deploy msg, start to build stream task", vgId);
// 1.deserialize msg and build task
int32_t size = sizeof(SStreamTask);
SStreamTask* pTask = taosMemoryCalloc(1, size);
if (pTask == NULL) {
tqError("vgId:%d failed to create stream task due to out of memory, alloc size:%d", vgId, size);
return TSDB_CODE_OUT_OF_MEMORY;
}
SDecoder decoder;
tDecoderInit(&decoder, (uint8_t*)msg, msgLen);
code = tDecodeStreamTask(&decoder, pTask);
tDecoderClear(&decoder);
if (code != TSDB_CODE_SUCCESS) {
taosMemoryFree(pTask);
return TSDB_CODE_INVALID_MSG;
}
// 2.save task, use the latest commit version as the initial start version of stream task.
int32_t taskId = pTask->id.taskId;
int64_t streamId = pTask->id.streamId;
bool added = false;
streamMetaWLock(pMeta);
code = streamMetaRegisterTask(pMeta, sversion, pTask, &added);
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
streamMetaWUnLock(pMeta);
if (code < 0) {
tqError("failed to add s-task:0x%x into vgId:%d meta, total:%d, code:%s", vgId, taskId, numOfTasks, tstrerror(code));
tFreeStreamTask(pTask);
return code;
}
// added into meta store, pTask cannot be reference since it may have been destroyed by other threads already now if
// it is added into the meta store
if (added) {
// only handled in the leader node
if (isLeader) {
tqDebug("vgId:%d s-task:0x%x is deployed and add into meta, numOfTasks:%d", vgId, taskId, numOfTasks);
SStreamTask* p = streamMetaAcquireTask(pMeta, streamId, taskId);
if (p != NULL && restored && p->info.fillHistory == 0) {
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(p)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
streamTaskHandleEvent(p->status.pSM, event);
} else if (!restored) {
tqWarn("s-task:%s not launched since vnode(vgId:%d) not ready", p->id.idStr, vgId);
}
if (p != NULL) {
streamMetaReleaseTask(pMeta, p);
}
} else {
tqDebug("vgId:%d not leader, not launch stream task s-task:0x%x", vgId, taskId);
}
} else {
tqWarn("vgId:%d failed to add s-task:0x%x, since already exists in meta store", vgId, taskId);
tFreeStreamTask(pTask);
}
return code;
}
int32_t tqStreamTaskProcessDropReq(SStreamMeta* pMeta, char* msg, int32_t msgLen) {
SVDropStreamTaskReq* pReq = (SVDropStreamTaskReq*)msg;
int32_t vgId = pMeta->vgId;
tqDebug("vgId:%d receive msg to drop s-task:0x%x", vgId, pReq->taskId);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, pReq->taskId);
if (pTask != NULL) {
// drop the related fill-history task firstly
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
STaskId* pHTaskId = &pTask->hTaskInfo.id;
streamMetaUnregisterTask(pMeta, pHTaskId->streamId, pHTaskId->taskId);
tqDebug("vgId:%d drop fill-history task:0x%x dropped firstly", vgId, (int32_t)pHTaskId->taskId);
}
streamMetaReleaseTask(pMeta, pTask);
}
// drop the stream task now
streamMetaUnregisterTask(pMeta, pReq->streamId, pReq->taskId);
// commit the update
streamMetaWLock(pMeta);
int32_t numOfTasks = streamMetaGetNumOfTasks(pMeta);
tqDebug("vgId:%d task:0x%x dropped, remain tasks:%d", vgId, pReq->taskId, numOfTasks);
if (streamMetaCommit(pMeta) < 0) {
// persist to disk
}
streamMetaWUnLock(pMeta);
return 0;
}
int32_t startStreamTasks(SStreamMeta* pMeta) {
int32_t code = TSDB_CODE_SUCCESS;
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d start to check all %d stream task(s) downstream status", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
streamMetaWLock(pMeta);
pTaskList = taosArrayDup(pMeta->pTaskList, NULL);
taosHashClear(pMeta->startInfo.pReadyTaskSet);
taosHashClear(pMeta->startInfo.pFailedTaskSet);
pMeta->startInfo.startTs = taosGetTimestampMs();
streamMetaWUnLock(pMeta);
// broadcast the check downstream tasks msg
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pTaskId->streamId, pTaskId->taskId);
if (pTask == NULL) {
continue;
}
// fill-history task can only be launched by related stream tasks.
if (pTask->info.fillHistory == 1) {
streamMetaReleaseTask(pMeta, pTask);
continue;
}
if (pTask->status.downstreamReady == 1) {
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
tqDebug("s-task:%s downstream ready, no need to check downstream, check only related fill-history task",
pTask->id.idStr);
streamLaunchFillHistoryTask(pTask);
}
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, pTask->execInfo.start, true);
streamMetaReleaseTask(pMeta, pTask);
continue;
}
EStreamTaskEvent event = (HAS_RELATED_FILLHISTORY_TASK(pTask)) ? TASK_EVENT_INIT_STREAM_SCANHIST : TASK_EVENT_INIT;
int32_t ret = streamTaskHandleEvent(pTask->status.pSM, event);
if (ret != TSDB_CODE_SUCCESS) {
code = ret;
}
streamMetaReleaseTask(pMeta, pTask);
}
taosArrayDestroy(pTaskList);
return code;
}
int32_t resetStreamTaskStatus(SStreamMeta* pMeta) {
int32_t vgId = pMeta->vgId;
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
tqDebug("vgId:%d reset all %d stream task(s) status to be uninit", vgId, numOfTasks);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
for (int32_t i = 0; i < numOfTasks; ++i) {
SStreamTaskId* pTaskId = taosArrayGet(pMeta->pTaskList, i);
STaskId id = {.streamId = pTaskId->streamId, .taskId = pTaskId->taskId};
SStreamTask** pTask = taosHashGet(pMeta->pTasksMap, &id, sizeof(id));
streamTaskResetStatus(*pTask);
}
return 0;
}
static int32_t restartStreamTasks(SStreamMeta* pMeta, bool isLeader) {
int32_t vgId = pMeta->vgId;
int32_t code = 0;
int64_t st = taosGetTimestampMs();
while(1) {
int32_t startVal = atomic_val_compare_exchange_32(&pMeta->startInfo.taskStarting, 0, 1);
if (startVal == 0) {
break;
}
tqDebug("vgId:%d in start stream tasks procedure, wait for 500ms and recheck", vgId);
taosMsleep(500);
}
terrno = 0;
tqInfo("vgId:%d tasks are all updated and stopped, restart all tasks, triggered by transId:%d", vgId,
pMeta->updateInfo.transId);
while (streamMetaTaskInTimer(pMeta)) {
tqDebug("vgId:%d some tasks in timer, wait for 100ms and recheck", pMeta->vgId);
taosMsleep(100);
}
streamMetaWLock(pMeta);
code = streamMetaReopen(pMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to reopen stream meta", vgId);
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
streamMetaInitBackend(pMeta);
int64_t el = taosGetTimestampMs() - st;
tqInfo("vgId:%d close&reload state elapsed time:%.3fs", vgId, el/1000.);
code = streamMetaLoadAllTasks(pMeta);
if (code != TSDB_CODE_SUCCESS) {
tqError("vgId:%d failed to load stream tasks, code:%s", vgId, tstrerror(terrno));
streamMetaWUnLock(pMeta);
code = terrno;
return code;
}
if (isLeader && !tsDisableStream) {
tqInfo("vgId:%d restart all stream tasks after all tasks being updated", vgId);
resetStreamTaskStatus(pMeta);
streamMetaWUnLock(pMeta);
startStreamTasks(pMeta);
} else {
streamMetaResetStartInfo(&pMeta->startInfo);
streamMetaWUnLock(pMeta);
tqInfo("vgId:%d, follower node not start stream tasks", vgId);
}
code = terrno;
return code;
}
int32_t tqStreamTaskProcessRunReq(SStreamMeta* pMeta, SRpcMsg* pMsg, bool isLeader) {
SStreamTaskRunReq* pReq = pMsg->pCont;
int32_t taskId = pReq->taskId;
int32_t vgId = pMeta->vgId;
if (taskId == STREAM_EXEC_START_ALL_TASKS_ID) {
startStreamTasks(pMeta);
return 0;
} else if (taskId == STREAM_EXEC_RESTART_ALL_TASKS_ID) {
restartStreamTasks(pMeta, isLeader);
return 0;
}
SStreamTask* pTask = streamMetaAcquireTask(pMeta, pReq->streamId, taskId);
if (pTask != NULL) { // even in halt status, the data in inputQ must be processed
char* p = NULL;
if (streamTaskReadyToRun(pTask, &p)) {
tqDebug("vgId:%d s-task:%s start to process block from inputQ, next checked ver:%" PRId64, vgId, pTask->id.idStr,
pTask->chkInfo.nextProcessVer);
streamExecTask(pTask);
} else {
int8_t status = streamTaskSetSchedStatusInactive(pTask);
tqDebug("vgId:%d s-task:%s ignore run req since not in ready state, status:%s, sched-status:%d", vgId,
pTask->id.idStr, p, status);
}
streamMetaReleaseTask(pMeta, pTask);
return 0;
} else { // NOTE: pTask->status.schedStatus is not updated since it is not be handled by the run exec.
// todo add one function to handle this
tqError("vgId:%d failed to found s-task, taskId:0x%x may have been dropped", vgId, taskId);
return -1;
}
}

167
source/dnode/vnode/src/tsdb/tsdbRead2.c
View File

@ -48,7 +48,7 @@ static int32_t doMergeMemIMemRows(TSDBROW* pRow, TSDBROW* piRow, STableBlockScan
static int32_t mergeRowsInFileBlocks(SBlockData* pBlockData, STableBlockScanInfo* pBlockScanInfo, int64_t key,
STsdbReader* pReader);
static int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, int32_t order, SCostSummary* pCost);
static int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, int32_t order, SReadCostSummary* pCost);
static STsdb* getTsdbByRetentions(SVnode* pVnode, SQueryTableDataCond* pCond, SRetention* retentions, const char* idstr,
int8_t* pLevel);
static SVersionRange getQueryVerRange(SVnode* pVnode, SQueryTableDataCond* pCond, int8_t level);
@ -58,6 +58,7 @@ static TSDBKEY getCurrentKeyInBuf(STableBlockScanInfo* pScanInfo, STsdbRea
static bool hasDataInFileBlock(const SBlockData* pBlockData, const SFileBlockDumpInfo* pDumpInfo);
static void initBlockDumpInfo(STsdbReader* pReader, SDataBlockIter* pBlockIter);
static int32_t getInitialDelIndex(const SArray* pDelSkyline, int32_t order);
static void updateComposedBlockInfo(STsdbReader* pReader, double el, STableBlockScanInfo* pBlockScanInfo);
static bool outOfTimeWindow(int64_t ts, STimeWindow* pWindow) { return (ts > pWindow->ekey) || (ts < pWindow->skey); }
@ -168,7 +169,7 @@ static int32_t filesetIteratorNext(SFilesetIter* pIter, STsdbReader* pReader, bo
return TSDB_CODE_SUCCESS;
}
SCostSummary* pCost = &pReader->cost;
SReadCostSummary* pCost = &pReader->cost;
pIter->pLastBlockReader->uid = 0;
tMergeTreeClose(&pIter->pLastBlockReader->mergeTree);
@ -291,11 +292,7 @@ static SSDataBlock* createResBlock(SQueryTableDataCond* pCond, int32_t capacity)
}
static int32_t tsdbInitReaderLock(STsdbReader* pReader) {
int32_t code = -1;
qTrace("tsdb/read: %p, pre-init read mutex: %p, code: %d", pReader, &pReader->readerMutex, code);
code = taosThreadMutexInit(&pReader->readerMutex, NULL);
int32_t code = taosThreadMutexInit(&pReader->readerMutex, NULL);
qTrace("tsdb/read: %p, post-init read mutex: %p, code: %d", pReader, &pReader->readerMutex, code);
return code;
@ -324,22 +321,14 @@ static int32_t tsdbAcquireReader(STsdbReader* pReader) {
}
static int32_t tsdbTryAcquireReader(STsdbReader* pReader) {
int32_t code = -1;
qTrace("tsdb/read: %p, pre-trytake read mutex: %p, code: %d", pReader, &pReader->readerMutex, code);
code = taosThreadMutexTryLock(&pReader->readerMutex);
int32_t code = taosThreadMutexTryLock(&pReader->readerMutex);
qTrace("tsdb/read: %p, post-trytake read mutex: %p, code: %d", pReader, &pReader->readerMutex, code);
return code;
}
static int32_t tsdbReleaseReader(STsdbReader* pReader) {
int32_t code = -1;
qTrace("tsdb/read: %p, pre-untake read mutex: %p, code: %d", pReader, &pReader->readerMutex, code);
code = taosThreadMutexUnlock(&pReader->readerMutex);
int32_t code = taosThreadMutexUnlock(&pReader->readerMutex);
qTrace("tsdb/read: %p, post-untake read mutex: %p, code: %d", pReader, &pReader->readerMutex, code);
return code;
@ -432,6 +421,7 @@ static int32_t tsdbReaderCreate(SVnode* pVnode, SQueryTableDataCond* pCond, void
}
tsdbInitReaderLock(pReader);
tsem_init(&pReader->resumeAfterSuspend, 0, 0);
*ppReader = pReader;
return code;
@ -1015,8 +1005,8 @@ static int32_t copyBlockDataToSDataBlock(STsdbReader* pReader) {
// check if current block are all handled
if (pDumpInfo->rowIndex >= 0 && pDumpInfo->rowIndex < pRecord->numRow) {
int64_t ts = pBlockData->aTSKEY[pDumpInfo->rowIndex];
if (outOfTimeWindow(ts,
&pReader->info.window)) { // the remain data has out of query time window, ignore current block
if (outOfTimeWindow(ts, &pReader->info.window)) {
// the remain data has out of query time window, ignore current block
setBlockAllDumped(pDumpInfo, ts, pReader->info.order);
}
} else {
@ -1123,16 +1113,12 @@ static bool getNeighborBlockOfSameTable(SDataBlockIter* pBlockIter, SFileDataBlo
}
int32_t step = asc ? 1 : -1;
// *nextIndex = pBlockInfo->tbBlockIdx + step;
// *pBlockIndex = *(SBlockIndex*)taosArrayGet(pTableBlockScanInfo->pBlockList, *nextIndex);
STableDataBlockIdx* pTableDataBlockIdx =
taosArrayGet(pTableBlockScanInfo->pBlockIdxList, pBlockInfo->tbBlockIdx + step);
SFileDataBlockInfo* p = taosArrayGet(pBlockIter->blockList, pTableDataBlockIdx->globalIndex);
memcpy(pRecord, &p->record, sizeof(SBrinRecord));
*nextIndex = pBlockInfo->tbBlockIdx + step;
// tMapDataGetItemByIdx(&pTableBlockScanInfo->mapData, pIndex->ordinalIndex, pBlock, tGetDataBlk);
return true;
}
@ -1376,23 +1362,19 @@ static int32_t buildDataBlockFromBuf(STsdbReader* pReader, STableBlockScanInfo*
return TSDB_CODE_SUCCESS;
}
SSDataBlock* pBlock = pReader->resBlockInfo.pResBlock;
int64_t st = taosGetTimestampUs();
SSDataBlock* pBlock = pReader->resBlockInfo.pResBlock;
int32_t code = buildDataBlockFromBufImpl(pBlockScanInfo, endKey, pReader->resBlockInfo.capacity, pReader);
blockDataUpdateTsWindow(pBlock, pReader->suppInfo.slotId[0]);
pBlock->info.id.uid = pBlockScanInfo->uid;
double el = (taosGetTimestampUs() - st) / 1000.0;
updateComposedBlockInfo(pReader, el, pBlockScanInfo);
setComposedBlockFlag(pReader, true);
double elapsedTime = (taosGetTimestampUs() - st) / 1000.0;
tsdbDebug("%p build data block from cache completed, elapsed time:%.2f ms, numOfRows:%" PRId64 ", brange:%" PRId64
" - %" PRId64 ", uid:%" PRIu64 ", %s",
pReader, elapsedTime, pBlock->info.rows, pBlock->info.window.skey, pBlock->info.window.ekey,
pReader, el, pBlock->info.rows, pBlock->info.window.skey, pBlock->info.window.ekey,
pBlockScanInfo->uid, pReader->idStr);
pReader->cost.buildmemBlock += elapsedTime;
pReader->cost.buildmemBlock += el;
return code;
}
@ -2293,13 +2275,12 @@ static int32_t loadNeighborIfOverlap(SFileDataBlockInfo* pBlockInfo, STableBlock
return code;
}
static void updateComposedBlockInfo(STsdbReader* pReader, double el, STableBlockScanInfo* pBlockScanInfo) {
void updateComposedBlockInfo(STsdbReader* pReader, double el, STableBlockScanInfo* pBlockScanInfo) {
SSDataBlock* pResBlock = pReader->resBlockInfo.pResBlock;
pResBlock->info.id.uid = (pBlockScanInfo != NULL) ? pBlockScanInfo->uid : 0;
pResBlock->info.dataLoad = 1;
blockDataUpdateTsWindow(pResBlock, pReader->suppInfo.slotId[0]);
setComposedBlockFlag(pReader, true);
pReader->cost.composedBlocks += 1;
@ -2356,7 +2337,6 @@ static int32_t buildComposedDataBlock(STsdbReader* pReader) {
pBlockScanInfo = *pReader->status.pTableIter;
if (pReader->pIgnoreTables &&
taosHashGet(*pReader->pIgnoreTables, &pBlockScanInfo->uid, sizeof(pBlockScanInfo->uid))) {
// setBlockAllDumped(pDumpInfo, pBlock->maxKey.ts, pReader->info.order);
return code;
}
}
@ -2436,7 +2416,7 @@ int32_t getInitialDelIndex(const SArray* pDelSkyline, int32_t order) {
return ASCENDING_TRAVERSE(order) ? 0 : taosArrayGetSize(pDelSkyline) - 1;
}
int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, int32_t order, SCostSummary* pCost) {
int32_t initDelSkylineIterator(STableBlockScanInfo* pBlockScanInfo, int32_t order, SReadCostSummary* pCost) {
int32_t code = 0;
int32_t newDelDataInFile = taosArrayGetSize(pBlockScanInfo->pFileDelData);
if (newDelDataInFile == 0 &&
@ -2935,6 +2915,8 @@ static int32_t buildBlockFromBufferSequentially(STsdbReader* pReader) {
SReaderStatus* pStatus = &pReader->status;
STableUidList* pUidList = &pStatus->uidList;
tsdbDebug("seq load data blocks from cache, %s", pReader->idStr);
while (1) {
if (pReader->code != TSDB_CODE_SUCCESS) {
tsdbWarn("tsdb reader is stopped ASAP, code:%s, %s", strerror(pReader->code), pReader->idStr);
@ -3043,6 +3025,8 @@ static ERetrieveType doReadDataFromLastFiles(STsdbReader* pReader) {
SSDataBlock* pResBlock = pReader->resBlockInfo.pResBlock;
SDataBlockIter* pBlockIter = &pReader->status.blockIter;
tsdbDebug("seq load data blocks from stt files %s", pReader->idStr);
while (1) {
terrno = 0;
@ -3774,7 +3758,6 @@ int32_t buildDataBlockFromBufImpl(STableBlockScanInfo* pBlockScanInfo, int64_t e
int32_t code = TSDB_CODE_SUCCESS;
do {
// SRow* pTSRow = NULL;
TSDBROW row = {.type = -1};
bool freeTSRow = false;
tsdbGetNextRowInMem(pBlockScanInfo, pReader, &row, endKey, &freeTSRow);
@ -3783,6 +3766,7 @@ int32_t buildDataBlockFromBufImpl(STableBlockScanInfo* pBlockScanInfo, int64_t e
}
if (row.type == TSDBROW_ROW_FMT) {
int64_t ts = row.pTSRow->ts;;
code = doAppendRowFromTSRow(pBlock, pReader, row.pTSRow, pBlockScanInfo);
if (freeTSRow) {
@ -3792,13 +3776,17 @@ int32_t buildDataBlockFromBufImpl(STableBlockScanInfo* pBlockScanInfo, int64_t e
if (code) {
return code;
}
pBlockScanInfo->lastProcKey = ts;
} else {
code = doAppendRowFromFileBlock(pBlock, pReader, row.pBlockData, row.iRow);
if (code) {
break;
}
pBlockScanInfo->lastProcKey = row.pBlockData->aTSKEY[row.iRow];
}
// no data in buffer, return immediately
if (!(pBlockScanInfo->iter.hasVal || pBlockScanInfo->iiter.hasVal)) {
break;
@ -4107,7 +4095,7 @@ void tsdbReaderClose2(STsdbReader* pReader) {
tsdbDataFileReaderClose(&pReader->pFileReader);
}
SCostSummary* pCost = &pReader->cost;
SReadCostSummary* pCost = &pReader->cost;
SFilesetIter* pFilesetIter = &pReader->status.fileIter;
if (pFilesetIter->pLastBlockReader != NULL) {
SLastBlockReader* pLReader = pFilesetIter->pLastBlockReader;
@ -4122,6 +4110,7 @@ void tsdbReaderClose2(STsdbReader* pReader) {
tsdbUntakeReadSnap2(pReader, pReader->pReadSnap, true);
pReader->pReadSnap = NULL;
tsem_destroy(&pReader->resumeAfterSuspend);
tsdbReleaseReader(pReader);
tsdbUninitReaderLock(pReader);
@ -4154,6 +4143,8 @@ int32_t tsdbReaderSuspend2(STsdbReader* pReader) {
SReaderStatus* pStatus = &pReader->status;
STableBlockScanInfo* pBlockScanInfo = NULL;
pReader->status.suspendInvoked = true; // record the suspend status
if (pStatus->loadFromFile) {
SFileDataBlockInfo* pBlockInfo = getCurrentBlockInfo(&pReader->status.blockIter);
if (pBlockInfo != NULL) {
@ -4167,84 +4158,34 @@ int32_t tsdbReaderSuspend2(STsdbReader* pReader) {
tsdbDataFileReaderClose(&pReader->pFileReader);
SCostSummary* pCost = &pReader->cost;
SReadCostSummary* pCost = &pReader->cost;
pReader->status.pLDataIterArray = destroySttBlockReader(pReader->status.pLDataIterArray, &pCost->sttCost);
pReader->status.pLDataIterArray = taosArrayInit(4, POINTER_BYTES);
// resetDataBlockScanInfo excluding lastKey
STableBlockScanInfo** p = NULL;
int32_t iter = 0;
while ((p = tSimpleHashIterate(pStatus->pTableMap, p, &iter)) != NULL) {
STableBlockScanInfo* pInfo = *(STableBlockScanInfo**)p;
pInfo->iterInit = false;
pInfo->iter.hasVal = false;
pInfo->iiter.hasVal = false;
if (pInfo->iter.iter != NULL) {
pInfo->iter.iter = tsdbTbDataIterDestroy(pInfo->iter.iter);
}
if (pInfo->iiter.iter != NULL) {
pInfo->iiter.iter = tsdbTbDataIterDestroy(pInfo->iiter.iter);
}
pInfo->delSkyline = taosArrayDestroy(pInfo->delSkyline);
pInfo->pFileDelData = taosArrayDestroy(pInfo->pFileDelData);
}
} else {
// resetDataBlockScanInfo excluding lastKey
STableBlockScanInfo** p = NULL;
int32_t iter = 0;
while ((p = tSimpleHashIterate(pStatus->pTableMap, p, &iter)) != NULL) {
STableBlockScanInfo* pInfo = *(STableBlockScanInfo**)p;
pInfo->iterInit = false;
pInfo->iter.hasVal = false;
pInfo->iiter.hasVal = false;
if (pInfo->iter.iter != NULL) {
pInfo->iter.iter = tsdbTbDataIterDestroy(pInfo->iter.iter);
}
if (pInfo->iiter.iter != NULL) {
pInfo->iiter.iter = tsdbTbDataIterDestroy(pInfo->iiter.iter);
}
pInfo->delSkyline = taosArrayDestroy(pInfo->delSkyline);
}
pBlockScanInfo = pStatus->pTableIter == NULL ? NULL : *pStatus->pTableIter;
if (pBlockScanInfo) {
// save lastKey to restore memory iterator
STimeWindow w = pReader->resBlockInfo.pResBlock->info.window;
pBlockScanInfo->lastProcKey = ASCENDING_TRAVERSE(pReader->info.order) ? w.ekey : w.skey;
// reset current current table's data block scan info,
pBlockScanInfo->iterInit = false;
pBlockScanInfo->iter.hasVal = false;
pBlockScanInfo->iiter.hasVal = false;
if (pBlockScanInfo->iter.iter != NULL) {
pBlockScanInfo->iter.iter = tsdbTbDataIterDestroy(pBlockScanInfo->iter.iter);
}
if (pBlockScanInfo->iiter.iter != NULL) {
pBlockScanInfo->iiter.iter = tsdbTbDataIterDestroy(pBlockScanInfo->iiter.iter);
}
pBlockScanInfo->pBlockList = taosArrayDestroy(pBlockScanInfo->pBlockList);
pBlockScanInfo->pBlockIdxList = taosArrayDestroy(pBlockScanInfo->pBlockIdxList);
// TODO: keep skyline for reuse
pBlockScanInfo->delSkyline = taosArrayDestroy(pBlockScanInfo->delSkyline);
}
}
// resetDataBlockScanInfo excluding lastKey
STableBlockScanInfo** p = NULL;
int32_t step = ASCENDING_TRAVERSE(pReader->info.order)? 1:-1;
int32_t iter = 0;
while ((p = tSimpleHashIterate(pStatus->pTableMap, p, &iter)) != NULL) {
STableBlockScanInfo* pInfo = *(STableBlockScanInfo**)p;
clearBlockScanInfo(pInfo);
pInfo->sttKeyInfo.nextProcKey = pInfo->lastProcKey + step;
}
pStatus->uidList.currentIndex = 0;
initReaderStatus(pStatus);
tsdbUntakeReadSnap2(pReader, pReader->pReadSnap, false);
pReader->pReadSnap = NULL;
pReader->flag = READER_STATUS_SUSPEND;
#if SUSPEND_RESUME_TEST
tsem_post(&pReader->resumeAfterSuspend);
#endif
tsdbDebug("reader: %p suspended uid %" PRIu64 " in this query %s", pReader, pBlockScanInfo ? pBlockScanInfo->uid : 0,
pReader->idStr);
return code;
@ -4399,6 +4340,16 @@ int32_t tsdbNextDataBlock2(STsdbReader* pReader, bool* hasNext) {
SReaderStatus* pStatus = &pReader->status;
// NOTE: the following codes is used to perform test for suspend/resume for tsdbReader when it blocks the commit
// the data should be ingested in round-robin and all the child tables should be createted before ingesting data
// the version range of query will be used to identify the correctness of suspend/resume functions.
// this function will blocked before loading the SECOND block from vnode-buffer, and restart itself from sst-files
#if SUSPEND_RESUME_TEST
if (!pReader->status.suspendInvoked && !pReader->status.loadFromFile) {
tsem_wait(&pReader->resumeAfterSuspend);
}
#endif
code = tsdbAcquireReader(pReader);
qTrace("tsdb/read: %p, take read mutex, code: %d", pReader, code);

1
source/dnode/vnode/src/tsdb/tsdbReadUtil.c
View File

@ -210,6 +210,7 @@ void clearBlockScanInfo(STableBlockScanInfo* p) {
p->iterInit = false;
p->iter.hasVal = false;
p->iiter.hasVal = false;
p->sttKeyInfo.status = STT_FILE_READER_UNINIT;
if (p->iter.iter != NULL) {
p->iter.iter = tsdbTbDataIterDestroy(p->iter.iter);

14
source/dnode/vnode/src/tsdb/tsdbReadUtil.h
View File

@ -96,7 +96,7 @@ typedef struct SResultBlockInfo {
int64_t capacity;
} SResultBlockInfo;
typedef struct SCostSummary {
typedef struct SReadCostSummary {
int64_t numOfBlocks;
double blockLoadTime;
double buildmemBlock;
@ -110,7 +110,7 @@ typedef struct SCostSummary {
double createScanInfoList;
double createSkylineIterTime;
double initLastBlockReader;
} SCostSummary;
} SReadCostSummary;
typedef struct STableUidList {
uint64_t* tableUidList; // access table uid list in uid ascending order list
@ -122,12 +122,6 @@ typedef struct {
int32_t numOfSttFiles;
} SBlockNumber;
typedef struct SBlockIndex {
int32_t ordinalIndex;
int64_t inFileOffset;
STimeWindow window; // todo replace it with overlap flag.
} SBlockIndex;
typedef struct SBlockOrderWrapper {
int64_t uid;
int64_t offset;
@ -192,6 +186,7 @@ typedef struct SFileBlockDumpInfo {
} SFileBlockDumpInfo;
typedef struct SReaderStatus {
bool suspendInvoked;
bool loadFromFile; // check file stage
bool composedDataBlock; // the returned data block is a composed block or not
SSHashObj* pTableMap; // SHash<STableBlockScanInfo>
@ -220,7 +215,8 @@ struct STsdbReader {
int32_t type; // query type: 1. retrieve all data blocks, 2. retrieve direct prev|next rows
SBlockLoadSuppInfo suppInfo;
STsdbReadSnap* pReadSnap;
SCostSummary cost;
tsem_t resumeAfterSuspend;
SReadCostSummary cost;
SHashObj** pIgnoreTables;
SSHashObj* pSchemaMap; // keep the retrieved schema info, to avoid the overhead by repeatly load schema
SDataFileReader* pFileReader; // the file reader

3
source/dnode/vnode/src/vnd/vnodeSvr.c
View File

@ -463,7 +463,6 @@ int32_t vnodePreProcessWriteMsg(SVnode *pVnode, SRpcMsg *pMsg) {
break;
}
_exit:
if (code) {
vError("vgId:%d, failed to preprocess write request since %s, msg type:%s", TD_VID(pVnode), tstrerror(code),
TMSG_INFO(pMsg->msgType));
@ -767,7 +766,7 @@ int32_t vnodeProcessStreamMsg(SVnode *pVnode, SRpcMsg *pMsg, SQueueInfo *pInfo)
case TDMT_STREAM_TASK_RUN:
return tqProcessTaskRunReq(pVnode->pTq, pMsg);
case TDMT_STREAM_TASK_DISPATCH:
return tqProcessTaskDispatchReq(pVnode->pTq, pMsg, true);
return tqProcessTaskDispatchReq(pVnode->pTq, pMsg);
case TDMT_STREAM_TASK_DISPATCH_RSP:
return tqProcessTaskDispatchRsp(pVnode->pTq, pMsg);
case TDMT_VND_STREAM_TASK_CHECK:

5
source/dnode/vnode/src/vnd/vnodeSync.c
View File

@ -18,6 +18,7 @@
#include "sync.h"
#include "tsdb.h"
#include "vnd.h"
#include "tqCommon.h"
#define BATCH_ENABLE 0
@ -569,8 +570,8 @@ static void vnodeRestoreFinish(const SSyncFSM *pFsm, const SyncIndex commitIdx)
vInfo("vgId:%d, sync restore finished, not launch stream tasks, since stream tasks are disabled", vgId);
} else {
vInfo("vgId:%d sync restore finished, start to launch stream tasks", pVnode->config.vgId);
tqResetStreamTaskStatus(pVnode->pTq);
tqStartStreamTaskAsync(pVnode->pTq, false);
resetStreamTaskStatus(pVnode->pTq->pStreamMeta);
tqStreamTaskStartAsync(pMeta, &pVnode->msgCb, false);
}
} else {
vInfo("vgId:%d, sync restore finished, not launch stream tasks since not leader", vgId);

11
source/libs/executor/src/streameventwindowoperator.c
View File

@ -610,6 +610,13 @@ void streamEventReloadState(SOperatorInfo* pOperator) {
compactEventWindow(pOperator, &curInfo, pInfo->pSeUpdated, pInfo->pSeDeleted, false);
qDebug("===stream=== reload state. save result %" PRId64 ", %" PRIu64, curInfo.winInfo.sessionWin.win.skey,
curInfo.winInfo.sessionWin.groupId);
if (IS_VALID_SESSION_WIN(curInfo.winInfo)) {
saveSessionOutputBuf(pAggSup, &curInfo.winInfo);
}
if (!curInfo.pWinFlag->endFlag) {
continue;
}
if (pInfo->twAggSup.calTrigger == STREAM_TRIGGER_AT_ONCE) {
saveResult(curInfo.winInfo, pInfo->pSeUpdated);
@ -621,10 +628,6 @@ void streamEventReloadState(SOperatorInfo* pOperator) {
getSessionHashKey(&curInfo.winInfo.sessionWin, &key);
tSimpleHashPut(pAggSup->pResultRows, &key, sizeof(SSessionKey), &curInfo.winInfo, sizeof(SResultWindowInfo));
}
if (IS_VALID_SESSION_WIN(curInfo.winInfo)) {
saveSessionOutputBuf(pAggSup, &curInfo.winInfo);
}
}
taosMemoryFree(pBuf);

3273
source/libs/nodes/src/nodesCodeFuncs.c
View File

File diff suppressed because it is too large Load Diff

1239
source/libs/nodes/src/nodesUtilFuncs.c
View File

File diff suppressed because it is too large Load Diff

1
source/libs/stream/src/stream.c
View File

@ -276,6 +276,7 @@ int32_t streamProcessDispatchMsg(SStreamTask* pTask, SStreamDispatchReq* pReq, S
int32_t code = buildDispatchRsp(pTask, pReq, status, &pRsp->pCont);
if (code != TSDB_CODE_SUCCESS) {
stError("s-task:%s failed to build dispatch rsp, msgId:%d, code:%s", id, pReq->msgId, tstrerror(code));
terrno = code;
return code;
}

24
source/libs/stream/src/streamExec.c
View File

@ -252,14 +252,15 @@ static void streamScanHistoryDataImpl(SStreamTask* pTask, SArray* pRes, int32_t*
SScanhistoryDataInfo streamScanHistoryData(SStreamTask* pTask, int64_t st) {
ASSERT(pTask->info.taskLevel == TASK_LEVEL__SOURCE);
void* exec = pTask->exec.pExecutor;
bool finished = false;
void* exec = pTask->exec.pExecutor;
bool finished = false;
const char* id = pTask->id.idStr;
qSetStreamOpOpen(exec);
while (1) {
if (streamTaskShouldPause(pTask)) {
stDebug("s-task:%s paused from the scan-history task", pTask->id.idStr);
stDebug("s-task:%s paused from the scan-history task", id);
// quit from step1, not continue to handle the step2
return (SScanhistoryDataInfo){TASK_SCANHISTORY_QUIT, 0};
}
@ -267,8 +268,7 @@ SScanhistoryDataInfo streamScanHistoryData(SStreamTask* pTask, int64_t st) {
SArray* pRes = taosArrayInit(0, sizeof(SSDataBlock));
if (pRes == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
stError("s-task:%s scan-history prepare result block failed, code:%s, retry later", pTask->id.idStr,
tstrerror(terrno));
stError("s-task:%s scan-history prepare result block failed, code:%s, retry later", id, tstrerror(terrno));
continue;
}
@ -281,12 +281,12 @@ SScanhistoryDataInfo streamScanHistoryData(SStreamTask* pTask, int64_t st) {
}
// dispatch the generated results
int32_t code = handleResultBlocks(pTask, pRes, size);
/*int32_t code = */handleResultBlocks(pTask, pRes, size);
int64_t el = taosGetTimestampMs() - st;
// downstream task input queue is full, try in 5sec
if (pTask->inputq.status == TASK_INPUT_STATUS__BLOCKED) {
if (pTask->inputq.status == TASK_INPUT_STATUS__BLOCKED && (pTask->info.fillHistory == 1)) {
return (SScanhistoryDataInfo){TASK_SCANHISTORY_REXEC, 5000};
}
@ -294,9 +294,9 @@ SScanhistoryDataInfo streamScanHistoryData(SStreamTask* pTask, int64_t st) {
return (SScanhistoryDataInfo){TASK_SCANHISTORY_CONT, 0};
}
if (el >= STREAM_SCAN_HISTORY_TIMESLICE) {
stDebug("s-task:%s fill-history:%d time slice exhausted, elapsed time:%.2fs, retry in 100ms",
pTask->id.idStr, pTask->info.fillHistory, el / 1000.0);
if (el >= STREAM_SCAN_HISTORY_TIMESLICE && (pTask->info.fillHistory == 1)) {
stDebug("s-task:%s fill-history:%d time slice exhausted, elapsed time:%.2fs, retry in 100ms", id,
pTask->info.fillHistory, el / 1000.0);
return (SScanhistoryDataInfo){TASK_SCANHISTORY_REXEC, 100};
}
}
@ -543,7 +543,7 @@ int32_t streamProcessTranstateBlock(SStreamTask* pTask, SStreamDataBlock* pBlock
* todo: the batch of blocks should be tuned dynamic, according to the total elapsed time of each batch of blocks, the
* appropriate batch of blocks should be handled in 5 to 10 sec.
*/
int32_t streamExecForAll(SStreamTask* pTask) {
int32_t doStreamExecTask(SStreamTask* pTask) {
const char* id = pTask->id.idStr;
// merge multiple input data if possible in the input queue.
@ -654,7 +654,7 @@ int32_t streamExecTask(SStreamTask* pTask) {
int8_t schedStatus = streamTaskSetSchedStatusActive(pTask);
if (schedStatus == TASK_SCHED_STATUS__WAITING) {
while (1) {
int32_t code = streamExecForAll(pTask);
int32_t code = doStreamExecTask(pTask);
if (code < 0) { // todo this status should be removed
atomic_store_8(&pTask->status.schedStatus, TASK_SCHED_STATUS__FAILED);
return -1;

17
source/libs/stream/src/streamStart.c
View File

@ -409,7 +409,6 @@ static void addIntoNodeUpdateList(SStreamTask* pTask, int32_t nodeId) {
int32_t streamProcessCheckRsp(SStreamTask* pTask, const SStreamTaskCheckRsp* pRsp) {
ASSERT(pTask->id.taskId == pRsp->upstreamTaskId);
const char* id = pTask->id.idStr;
int32_t vgId = pTask->pMeta->vgId;
if (streamTaskShouldStop(pTask)) {
stDebug("s-task:%s should stop, do not do check downstream again", id);
@ -461,16 +460,25 @@ int32_t streamProcessCheckRsp(SStreamTask* pTask, const SStreamTaskCheckRsp* pRs
"s-task:%s vgId:%d self vnode-transfer/leader-change/restart detected, old stage:%"PRId64", current stage:%"PRId64", "
"not check wait for downstream task nodeUpdate, and all tasks restart",
id, pRsp->upstreamNodeId, pRsp->oldStage, pTask->pMeta->stage);
addIntoNodeUpdateList(pTask, pRsp->upstreamNodeId);
} else {
stError(
"s-task:%s downstream taskId:0x%x (vgId:%d) not leader, self dispatch epset needs to be updated, not check "
"downstream again, nodeUpdate needed",
id, pRsp->downstreamTaskId, pRsp->downstreamNodeId);
addIntoNodeUpdateList(pTask, pRsp->downstreamNodeId);
}
addIntoNodeUpdateList(pTask, pRsp->downstreamNodeId);
streamMetaUpdateTaskDownstreamStatus(pTask, pTask->execInfo.init, taosGetTimestampMs(), false);
// automatically set the related fill-history task to be failed.
if (HAS_RELATED_FILLHISTORY_TASK(pTask)) {
STaskId* pId = &pTask->hTaskInfo.id;
SStreamTask* pHTask = streamMetaAcquireTask(pTask->pMeta, pId->streamId, pId->taskId);
streamMetaUpdateTaskDownstreamStatus(pHTask, pHTask->execInfo.init, taosGetTimestampMs(), false);
streamMetaReleaseTask(pTask->pMeta, pHTask);
}
} else { // TASK_DOWNSTREAM_NOT_READY, let's retry in 100ms
STaskRecheckInfo* pInfo = createRecheckInfo(pTask, pRsp);
@ -1072,8 +1080,9 @@ int32_t streamMetaUpdateTaskDownstreamStatus(SStreamTask* pTask, int64_t startTs
taosHashPut(pDst, &id, sizeof(id), &initTs, sizeof(STaskInitTs));
int32_t numOfTotal = streamMetaGetNumOfTasks(pMeta);
int32_t numOfRecv = taosHashGetSize(pStartInfo->pReadyTaskSet) + taosHashGetSize(pStartInfo->pFailedTaskSet);
if (taosHashGetSize(pStartInfo->pReadyTaskSet) + taosHashGetSize(pStartInfo->pFailedTaskSet) == numOfTotal) {
if (numOfRecv == numOfTotal) {
pStartInfo->readyTs = taosGetTimestampMs();
pStartInfo->elapsedTime = (pStartInfo->startTs != 0) ? pStartInfo->readyTs - pStartInfo->startTs : 0;
@ -1087,6 +1096,8 @@ int32_t streamMetaUpdateTaskDownstreamStatus(SStreamTask* pTask, int64_t startTs
displayStatusInfo(pMeta, pStartInfo->pFailedTaskSet, false);
streamMetaResetStartInfo(pStartInfo);
} else {
stDebug("vgId:%d recv check down results:%d, total:%d", pMeta->vgId, numOfRecv, numOfTotal);
}
streamMetaWUnLock(pMeta);

3
source/libs/stream/src/streamTask.c
View File

@ -782,4 +782,5 @@ void streamTaskStatusCopy(STaskStatusEntry* pDst, const STaskStatusEntry* pSrc)
pDst->sinkDataSize = pSrc->sinkDataSize;
pDst->activeCheckpointId = pSrc->activeCheckpointId;
pDst->checkpointFailed = pSrc->checkpointFailed;
}
}

4
source/libs/transport/src/transCli.c
View File

@ -957,8 +957,8 @@ static void cliSendCb(uv_write_t* req, int status) {
SCliMsg* pMsg = !transQueueEmpty(&pConn->cliMsgs) ? transQueueGet(&pConn->cliMsgs, 0) : NULL;
if (pMsg != NULL) {
int64_t cost = taosGetTimestampUs() - pMsg->st;
if (cost > 1000 * 20) {
tWarn("%s conn %p send cost:%dus, send exception", CONN_GET_INST_LABEL(pConn), pConn, (int)cost);
if (cost > 1000 * 50) {
tTrace("%s conn %p send cost:%dus ", CONN_GET_INST_LABEL(pConn), pConn, (int)cost);
}
}

1
source/os/src/osFile.c
View File

@ -648,6 +648,7 @@ int taosOpenFileNotStream(const char *path, int32_t tdFileOptions) {
access |= (tdFileOptions & TD_FILE_TEXT) ? O_TEXT : 0;
access |= (tdFileOptions & TD_FILE_EXCL) ? O_EXCL : 0;
access |= (tdFileOptions & TD_FILE_CLOEXEC) ? O_CLOEXEC : 0;
int fd = open(path, access, S_IRWXU | S_IRWXG | S_IRWXO);
return fd;
}

105
source/util/src/tcompression.c
View File

@ -538,66 +538,71 @@ int32_t tsDecompressTimestampImp(const char *const input, const int32_t nelement
memcpy(output, input + 1, nelements * longBytes);
return nelements * longBytes;
} else if (input[0] == 1) { // Decompress
int64_t *ostream = (int64_t *)output;
if (tsSIMDEnable && tsAVX512Enable) {
tsDecompressTimestampAvx512(input, nelements, output, false);
} else if (tsSIMDEnable && tsAVX2Enable) {
tsDecompressTimestampAvx2(input, nelements, output, false);
} else {
int64_t *ostream = (int64_t *)output;
int32_t ipos = 1, opos = 0;
int8_t nbytes = 0;
int64_t prev_value = 0;
int64_t prev_delta = 0;
int64_t delta_of_delta = 0;
int32_t ipos = 1, opos = 0;
int8_t nbytes = 0;
int64_t prev_value = 0;
int64_t prev_delta = 0;
int64_t delta_of_delta = 0;
while (1) {
uint8_t flags = input[ipos++];
// Decode dd1
uint64_t dd1 = 0;
nbytes = flags & INT8MASK(4);
if (nbytes == 0) {
delta_of_delta = 0;
} else {
if (is_bigendian()) {
memcpy(((char *)(&dd1)) + longBytes - nbytes, input + ipos, nbytes);
while (1) {
uint8_t flags = input[ipos++];
// Decode dd1
uint64_t dd1 = 0;
nbytes = flags & INT8MASK(4);
if (nbytes == 0) {
delta_of_delta = 0;
} else {
memcpy(&dd1, input + ipos, nbytes);
if (is_bigendian()) {
memcpy(((char *)(&dd1)) + longBytes - nbytes, input + ipos, nbytes);
} else {
memcpy(&dd1, input + ipos, nbytes);
}
delta_of_delta = ZIGZAG_DECODE(int64_t, dd1);
}
delta_of_delta = ZIGZAG_DECODE(int64_t, dd1);
}
ipos += nbytes;
if (opos == 0) {
prev_value = delta_of_delta;
prev_delta = 0;
ostream[opos++] = delta_of_delta;
} else {
ipos += nbytes;
if (opos == 0) {
prev_value = delta_of_delta;
prev_delta = 0;
ostream[opos++] = delta_of_delta;
} else {
prev_delta = delta_of_delta + prev_delta;
prev_value = prev_value + prev_delta;
ostream[opos++] = prev_value;
}
if (opos == nelements) return nelements * longBytes;
// Decode dd2
uint64_t dd2 = 0;
nbytes = (flags >> 4) & INT8MASK(4);
if (nbytes == 0) {
delta_of_delta = 0;
} else {
if (is_bigendian()) {
memcpy(((char *)(&dd2)) + longBytes - nbytes, input + ipos, nbytes);
} else {
memcpy(&dd2, input + ipos, nbytes);
}
// zigzag_decoding
delta_of_delta = ZIGZAG_DECODE(int64_t, dd2);
}
ipos += nbytes;
prev_delta = delta_of_delta + prev_delta;
prev_value = prev_value + prev_delta;
ostream[opos++] = prev_value;
if (opos == nelements) return nelements * longBytes;
}
if (opos == nelements) return nelements * longBytes;
// Decode dd2
uint64_t dd2 = 0;
nbytes = (flags >> 4) & INT8MASK(4);
if (nbytes == 0) {
delta_of_delta = 0;
} else {
if (is_bigendian()) {
memcpy(((char *)(&dd2)) + longBytes - nbytes, input + ipos, nbytes);
} else {
memcpy(&dd2, input + ipos, nbytes);
}
// zigzag_decoding
delta_of_delta = ZIGZAG_DECODE(int64_t, dd2);
}
ipos += nbytes;
prev_delta = delta_of_delta + prev_delta;
prev_value = prev_value + prev_delta;
ostream[opos++] = prev_value;
if (opos == nelements) return nelements * longBytes;
}
} else {
ASSERT(0);
return -1;
}
return nelements * longBytes;
}
/* --------------------------------------------Double Compression ---------------------------------------------- */

300
source/util/src/tdecompress.c
View File

@ -53,11 +53,8 @@ int32_t tsDecompressIntImpl_Hw(const char *const input, const int32_t nelements,
int64_t prevValue = 0;
#if __AVX2__
while (1) {
if (_pos == nelements) break;
uint64_t w = 0;
memcpy(&w, ip, LONG_BYTES);
while (_pos < nelements) {
uint64_t w = *(uint64_t*) ip;
char selector = (char)(w & INT64MASK(4)); // selector = 4
char bit = bit_per_integer[(int32_t)selector]; // bit = 3
@ -114,7 +111,7 @@ int32_t tsDecompressIntImpl_Hw(const char *const input, const int32_t nelements,
__m256i signmask = _mm256_and_si256(_mm256_set1_epi64x(1), zigzagVal);
signmask = _mm256_sub_epi64(_mm256_setzero_si256(), signmask);
// get the four zigzag values here
// get four zigzag values here
__m256i delta = _mm256_xor_si256(_mm256_srli_epi64(zigzagVal, 1), signmask);
// calculate the cumulative sum (prefix sum) for each number
@ -250,73 +247,264 @@ int32_t tsDecompressFloatImplAvx2(const char *const input, const int32_t nelemen
return 0;
}
int32_t tsDecompressTimestampAvx2(const char* const input, const int32_t nelements, char *const output, bool bigEndian) {
int32_t tsDecompressTimestampAvx2(const char *const input, const int32_t nelements, char *const output,
bool bigEndian) {
#if 0
int64_t *ostream = (int64_t *)output;
int32_t ipos = 1, opos = 0;
int8_t nbytes = 0;
int64_t prevValue = 0;
int64_t prevDelta = 0;
int64_t deltaOfDelta = 0;
int32_t longBytes = LONG_BYTES;
__m128i prevVal = _mm_setzero_si128();
__m128i prevDelta = _mm_setzero_si128();
#if __AVX2__
int32_t batch = nelements >> 1;
int32_t remainder = nelements & 0x01;
__mmask16 mask2[16] = {0, 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff};
int32_t batch = nelements >> 2;
int32_t remainder = nelements & 0x1;
while (1) {
int32_t i = 0;
if (batch > 1) {
// first loop
uint8_t flags = input[ipos++];
// Decode dd1
uint64_t dd1 = 0;
nbytes = flags & INT8MASK(4);
int8_t nbytes1 = flags & INT8MASK(4); // range of nbytes starts from 0 to 7
int8_t nbytes2 = (flags >> 4) & INT8MASK(4);
__m128i data1;
if (nbytes1 == 0) {
data1 = _mm_setzero_si128();
} else {
memcpy(&data1, (const void*) (input + ipos), nbytes1);
}
__m128i data2;
if (nbytes2 == 0) {
data2 = _mm_setzero_si128();
} else {
memcpy(&data2, (const void*) (input + ipos + nbytes1), nbytes2);
}
data2 = _mm_broadcastq_epi64(data2);
__m128i zzVal = _mm_blend_epi32(data2, data1, 0x03);
// ZIGZAG_DECODE(T, v) (((v) >> 1) ^ -((T)((v)&1)))
__m128i signmask = _mm_and_si128(_mm_set1_epi64x(1), zzVal);
signmask = _mm_sub_epi64(_mm_setzero_si128(), signmask);
// get two zigzag values here
__m128i deltaOfDelta = _mm_xor_si128(_mm_srli_epi64(zzVal, 1), signmask);
__m128i deltaCurrent = _mm_add_epi64(deltaOfDelta, prevDelta);
deltaCurrent = _mm_add_epi64(_mm_slli_si128(deltaCurrent, 8), deltaCurrent);
__m128i val = _mm_add_epi64(deltaCurrent, prevVal);
_mm_storeu_si128((__m128i *)&ostream[opos], val);
// keep the previous value
prevVal = _mm_shuffle_epi32 (val, 0xEE);
// keep the previous delta of delta, for the first item
prevDelta = _mm_shuffle_epi32(deltaOfDelta, 0xEE);
opos += 2;
ipos += nbytes1 + nbytes2;
i += 1;
}
// the remain
for(; i < batch; ++i) {
uint8_t flags = input[ipos++];
int8_t nbytes1 = flags & INT8MASK(4); // range of nbytes starts from 0 to 7
int8_t nbytes2 = (flags >> 4) & INT8MASK(4);
// __m128i data1 = _mm_maskz_loadu_epi8(mask2[nbytes1], (const void*)(input + ipos));
// __m128i data2 = _mm_maskz_loadu_epi8(mask2[nbytes2], (const void*)(input + ipos + nbytes1));
__m128i data1;
if (nbytes1 == 0) {
data1 = _mm_setzero_si128();
} else {
int64_t dd = 0;
memcpy(&dd, (const void*) (input + ipos), nbytes1);
data1 = _mm_loadu_si64(&dd);
}
__m128i data2;
if (nbytes2 == 0) {
data2 = _mm_setzero_si128();
} else {
int64_t dd = 0;
memcpy(&dd, (const void*) (input + ipos + nbytes1), nbytes2);
data2 = _mm_loadu_si64(&dd);
}
data2 = _mm_broadcastq_epi64(data2);
__m128i zzVal = _mm_blend_epi32(data2, data1, 0x03);
// ZIGZAG_DECODE(T, v) (((v) >> 1) ^ -((T)((v)&1)))
__m128i signmask = _mm_and_si128(_mm_set1_epi64x(1), zzVal);
signmask = _mm_sub_epi64(_mm_setzero_si128(), signmask);
// get two zigzag values here
__m128i deltaOfDelta = _mm_xor_si128(_mm_srli_epi64(zzVal, 1), signmask);
__m128i deltaCurrent = _mm_add_epi64(deltaOfDelta, prevDelta);
deltaCurrent = _mm_add_epi64(_mm_slli_si128(deltaCurrent, 8), deltaCurrent);
__m128i val = _mm_add_epi64(deltaCurrent, prevVal);
_mm_storeu_si128((__m128i *)&ostream[opos], val);
// keep the previous value
prevVal = _mm_shuffle_epi32 (val, 0xEE);
// keep the previous delta of delta
__m128i delta = _mm_add_epi64(_mm_slli_si128(deltaOfDelta, 8), deltaOfDelta);
prevDelta = _mm_shuffle_epi32(_mm_add_epi64(delta, prevDelta), 0xEE);
opos += 2;
ipos += nbytes1 + nbytes2;
}
if (remainder > 0) {
uint64_t dd = 0;
uint8_t flags = input[ipos++];
int32_t nbytes = flags & INT8MASK(4);
int64_t deltaOfDelta = 0;
if (nbytes == 0) {
deltaOfDelta = 0;
} else {
if (bigEndian) {
memcpy(((char *)(&dd1)) + longBytes - nbytes, input + ipos, nbytes);
} else {
memcpy(&dd1, input + ipos, nbytes);
}
deltaOfDelta = ZIGZAG_DECODE(int64_t, dd1);
// if (is_bigendian()) {
// memcpy(((char *)(&dd1)) + longBytes - nbytes, input + ipos, nbytes);
// } else {
memcpy(&dd, input + ipos, nbytes);
// }
deltaOfDelta = ZIGZAG_DECODE(int64_t, dd);
}
ipos += nbytes;
prevDelta += deltaOfDelta;
prevValue += prevDelta;
ostream[opos++] = prevValue;
if (opos == nelements) {
return nelements * longBytes;
}
// Decode dd2
uint64_t dd2 = 0;
nbytes = (flags >> 4) & INT8MASK(4);
if (nbytes == 0) {
deltaOfDelta = 0;
if (opos == 0) {
ostream[opos++] = deltaOfDelta;
} else {
if (bigEndian) {
memcpy(((char *)(&dd2)) + longBytes - nbytes, input + ipos, nbytes);
} else {
memcpy(&dd2, input + ipos, nbytes);
}
// zigzag_decoding
deltaOfDelta = ZIGZAG_DECODE(int64_t, dd2);
}
ipos += nbytes;
prevDelta += deltaOfDelta;
prevValue += prevDelta;
ostream[opos++] = prevValue;
if (opos == nelements) {
return nelements * longBytes;
int64_t prevDeltaX = deltaOfDelta + prevDelta[1];
ostream[opos++] = prevVal[1] + prevDeltaX;
}
}
#endif
#endif
return 0;
}
int32_t tsDecompressTimestampAvx512(const char *const input, const int32_t nelements, char *const output,
bool UNUSED_PARAM(bigEndian)) {
int64_t *ostream = (int64_t *)output;
int32_t ipos = 1, opos = 0;
#if __AVX512VL__
__m128i prevVal = _mm_setzero_si128();
__m128i prevDelta = _mm_setzero_si128();
int32_t numOfBatch = nelements >> 1;
int32_t remainder = nelements & 0x01;
__mmask16 mask2[16] = {0, 0x0001, 0x0003, 0x0007, 0x000f, 0x001f, 0x003f, 0x007f, 0x00ff};
int32_t i = 0;
if (numOfBatch > 1) {
// first loop
uint8_t flags = input[ipos++];
int8_t nbytes1 = flags & INT8MASK(4); // range of nbytes starts from 0 to 7
int8_t nbytes2 = (flags >> 4) & INT8MASK(4);
__m128i data1 = _mm_maskz_loadu_epi8(mask2[nbytes1], (const void*)(input + ipos));
__m128i data2 = _mm_maskz_loadu_epi8(mask2[nbytes2], (const void*)(input + ipos + nbytes1));
data2 = _mm_broadcastq_epi64(data2);
__m128i zzVal = _mm_blend_epi32(data2, data1, 0x03);
// ZIGZAG_DECODE(T, v) (((v) >> 1) ^ -((T)((v)&1)))
__m128i signmask = _mm_and_si128(_mm_set1_epi64x(1), zzVal);
signmask = _mm_sub_epi64(_mm_setzero_si128(), signmask);
// get two zigzag values here
__m128i deltaOfDelta = _mm_xor_si128(_mm_srli_epi64(zzVal, 1), signmask);
__m128i deltaCurrent = _mm_add_epi64(deltaOfDelta, prevDelta);
deltaCurrent = _mm_add_epi64(_mm_slli_si128(deltaCurrent, 8), deltaCurrent);
__m128i val = _mm_add_epi64(deltaCurrent, prevVal);
_mm_storeu_si128((__m128i *)&ostream[opos], val);
// keep the previous value
prevVal = _mm_shuffle_epi32 (val, 0xEE);
// keep the previous delta of delta, for the first item
prevDelta = _mm_shuffle_epi32(deltaOfDelta, 0xEE);
opos += 2;
ipos += nbytes1 + nbytes2;
i += 1;
}
// the remain
for(; i < numOfBatch; ++i) {
uint8_t flags = input[ipos++];
int8_t nbytes1 = flags & INT8MASK(4); // range of nbytes starts from 0 to 7
int8_t nbytes2 = (flags >> 4) & INT8MASK(4);
__m128i data1 = _mm_maskz_loadu_epi8(mask2[nbytes1], (const void*)(input + ipos));
__m128i data2 = _mm_maskz_loadu_epi8(mask2[nbytes2], (const void*)(input + ipos + nbytes1));
data2 = _mm_broadcastq_epi64(data2);
__m128i zzVal = _mm_blend_epi32(data2, data1, 0x03);
// ZIGZAG_DECODE(T, v) (((v) >> 1) ^ -((T)((v)&1)))
__m128i signmask = _mm_and_si128(_mm_set1_epi64x(1), zzVal);
signmask = _mm_sub_epi64(_mm_setzero_si128(), signmask);
// get two zigzag values here
__m128i deltaOfDelta = _mm_xor_si128(_mm_srli_epi64(zzVal, 1), signmask);
__m128i deltaCurrent = _mm_add_epi64(deltaOfDelta, prevDelta);
deltaCurrent = _mm_add_epi64(_mm_slli_si128(deltaCurrent, 8), deltaCurrent);
__m128i val = _mm_add_epi64(deltaCurrent, prevVal);
_mm_storeu_si128((__m128i *)&ostream[opos], val);
// keep the previous value
prevVal = _mm_shuffle_epi32 (val, 0xEE);
// keep the previous delta of delta
__m128i delta = _mm_add_epi64(_mm_slli_si128(deltaOfDelta, 8), deltaOfDelta);
prevDelta = _mm_shuffle_epi32(_mm_add_epi64(delta, prevDelta), 0xEE);
opos += 2;
ipos += nbytes1 + nbytes2;
}
if (remainder > 0) {
uint64_t dd = 0;
uint8_t flags = input[ipos++];
int32_t nbytes = flags & INT8MASK(4);
int64_t deltaOfDelta = 0;
if (nbytes == 0) {
deltaOfDelta = 0;
} else {
memcpy(&dd, input + ipos, nbytes);
deltaOfDelta = ZIGZAG_DECODE(int64_t, dd);
}
ipos += nbytes;
if (opos == 0) {
ostream[opos++] = deltaOfDelta;
} else {
int64_t prevDeltaX = deltaOfDelta + prevDelta[1];
ostream[opos++] = prevVal[1] + prevDeltaX;
}
}
#endif
return 0;
}

94
source/util/test/decompressTest.cpp Normal file
View File

@ -0,0 +1,94 @@
#include <gtest/gtest.h>
#include <stdlib.h>
#include <tcompression.h>
#include <random>
namespace {} // namespace
TEST(utilTest, decompress_test) {
int64_t tsList[10] = {1700000000, 1700000100, 1700000200, 1700000300, 1700000400,
1700000500, 1700000600, 1700000700, 1700000800, 1700000900};
char* pOutput[10 * sizeof(int64_t)] = {0};
int32_t len = tsCompressTimestamp(tsList, sizeof(tsList), sizeof(tsList) / sizeof(tsList[0]), pOutput, 10, ONE_STAGE_COMP, NULL, 0);
char* decompOutput[10 * 8] = {0};
tsDecompressTimestamp(pOutput, len, 10, decompOutput, sizeof(int64_t)*10, ONE_STAGE_COMP, NULL, 0);
for(int32_t i = 0; i < 10; ++i) {
std::cout<< ((int64_t*)decompOutput)[i] << std::endl;
}
memset(decompOutput, 0, 10*8);
tsDecompressTimestampAvx512(reinterpret_cast<const char* const>(pOutput), 10,
reinterpret_cast<char* const>(decompOutput), false);
for(int32_t i = 0; i < 10; ++i) {
std::cout<<((int64_t*)decompOutput)[i] << std::endl;
}
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
int64_t tsList1[7] = {1700000000, 1700000000, 1700000000, 1700000000, 1700000000, 1700000000, 1700000900};
int32_t len1 = tsCompressTimestamp(tsList1, sizeof(tsList1), sizeof(tsList1) / sizeof(tsList1[0]), pOutput, 7, ONE_STAGE_COMP, NULL, 0);
memset(decompOutput, 0, 10*8);
tsDecompressTimestampAvx512(reinterpret_cast<const char* const>(pOutput), 7,
reinterpret_cast<char* const>(decompOutput), false);
for(int32_t i = 0; i < 7; ++i) {
std::cout<<((int64_t*)decompOutput)[i] << std::endl;
}
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
int64_t tsList2[1] = {1700000000};
int32_t len2 = tsCompressTimestamp(tsList2, sizeof(tsList2), sizeof(tsList2) / sizeof(tsList2[0]), pOutput, 1, ONE_STAGE_COMP, NULL, 0);
memset(decompOutput, 0, 10*8);
tsDecompressTimestampAvx512(reinterpret_cast<const char* const>(pOutput), 1,
reinterpret_cast<char* const>(decompOutput), false);
for(int32_t i = 0; i < 1; ++i) {
std::cout<<((int64_t*)decompOutput)[i] << std::endl;
}
}
TEST(utilTest, decompress_perf_test) {
int32_t num = 10000;
int64_t* pList = static_cast<int64_t*>(taosMemoryCalloc(num, sizeof(int64_t)));
int64_t iniVal = 1700000000;
uint32_t v = 100;
for(int32_t i = 0; i < num; ++i) {
iniVal += taosRandR(&v)%10;
pList[i] = iniVal;
}
char* px = static_cast<char*>(taosMemoryMalloc(num * sizeof(int64_t)));
int32_t len = tsCompressTimestamp(pList, num * sizeof(int64_t), num, px, num, ONE_STAGE_COMP, NULL, 0);
char* pOutput = static_cast<char*>(taosMemoryMalloc(num * sizeof(int64_t)));
int64_t st = taosGetTimestampUs();
for(int32_t k = 0; k < 10000; ++k) {
tsDecompressTimestamp(px, len, num, pOutput, sizeof(int64_t) * num, ONE_STAGE_COMP, NULL, 0);
}
int64_t el1 = taosGetTimestampUs() - st;
std::cout << "soft decompress elapsed time:" << el1 << " us" << std::endl;
memset(pOutput, 0, num * sizeof(int64_t));
st = taosGetTimestampUs();
for(int32_t k = 0; k < 10000; ++k) {
tsDecompressTimestampAvx512(px, num, pOutput, false);
}
int64_t el2 = taosGetTimestampUs() - st;
std::cout << "SIMD decompress elapsed time:" << el2 << " us" << std::endl;
taosMemoryFree(pList);
taosMemoryFree(pOutput);
taosMemoryFree(px);
}

3
tests/develop-test/win-test-file
View File

@ -1,4 +1,7 @@
python3 ./test.py -f 2-query/table_count_scan.py
python3 ./test.py -f 2-query/pseudo_column.py
python3 ./test.py -f 2-query/ts-range.py
python3 ./test.py -f 2-query/tag_scan.py
python3 ./test.py -f 2-query/show_create_db.py
python3 ./test.py -f 5-taos-tools/taosbenchmark/auto_create_table_json.py
python3 ./test.py -f 5-taos-tools/taosbenchmark/custom_col_tag.py

1
tests/parallel_test/cases.task
View File

@ -198,6 +198,7 @@
#,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeTransform-db.py -N 6 -n 3
,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select.py -N 2 -n 1
,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select-duplicatedata.py -N 3 -n 3
,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select-duplicatedata-false.py -N 3 -n 3
,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select.py -N 3 -n 3
,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeSplit-stb.py -N 3 -n 3
,,y,system-test,./pytest.sh python3 test.py -f 7-tmq/tmqVnodeSplit-column.py -N 3 -n 3

60
tests/script/win-test-file
View File

@ -19,19 +19,18 @@
./test.sh -f tsim/dnode/balance3.sim
./test.sh -f tsim/vnode/replica3_many.sim
./test.sh -f tsim/stable/metrics_idx.sim
./test.sh -f tsim/db/alter_replica_13.sim
./test.sh -f tsim/sync/3Replica1VgElect.sim
./test.sh -f tsim/sync/3Replica5VgElect.sim
./test.sh -f tsim/valgrind/checkError6.sim
./test.sh -f tsim/user/basic.sim
./test.sh -f tsim/user/password.sim
./test.sh -f tsim/user/whitelist.sim
./test.sh -f tsim/user/privilege_db.sim
./test.sh -f tsim/user/privilege_sysinfo.sim
./test.sh -f tsim/user/privilege_topic.sim
./test.sh -f tsim/user/privilege_table.sim
./test.sh -f tsim/user/privilege_create_db.sim
./test.sh -f tsim/db/alter_option.sim
./test.sh -f tsim/db/alter_replica_31.sim
./test.sh -f tsim/db/basic1.sim
./test.sh -f tsim/db/basic2.sim
./test.sh -f tsim/db/basic3.sim
@ -97,6 +96,7 @@
./test.sh -f tsim/insert/delete0.sim
./test.sh -f tsim/insert/update1_sort_merge.sim
./test.sh -f tsim/insert/update2.sim
./test.sh -f tsim/insert/insert_stb.sim
./test.sh -f tsim/parser/alter__for_community_version.sim
./test.sh -f tsim/parser/alter_column.sim
./test.sh -f tsim/parser/alter_stable.sim
@ -186,6 +186,7 @@
./test.sh -f tsim/query/session.sim
./test.sh -f tsim/query/join_interval.sim
./test.sh -f tsim/query/join_pk.sim
./test.sh -f tsim/query/count_spread.sim
./test.sh -f tsim/query/unionall_as_table.sim
./test.sh -f tsim/query/multi_order_by.sim
./test.sh -f tsim/query/sys_tbname.sim
@ -197,10 +198,14 @@
./test.sh -f tsim/query/emptyTsRange_scl.sim
./test.sh -f tsim/query/partitionby.sim
./test.sh -f tsim/query/tableCount.sim
./test.sh -f tsim/query/show_db_table_kind.sim
./test.sh -f tsim/query/bi_star_table.sim
./test.sh -f tsim/query/bi_tag_scan.sim
./test.sh -f tsim/query/tag_scan.sim
./test.sh -f tsim/query/nullColSma.sim
./test.sh -f tsim/query/bug3398.sim
./test.sh -f tsim/query/explain_tsorder.sim
./test.sh -f tsim/query/apercentile.sim
./test.sh -f tsim/qnode/basic1.sim
./test.sh -f tsim/snode/basic1.sim
./test.sh -f tsim/mnode/basic1.sim
@ -237,52 +242,6 @@
./test.sh -f tsim/table/table.sim
./test.sh -f tsim/table/tinyint.sim
./test.sh -f tsim/table/vgroup.sim
./test.sh -f tsim/stream/basic0.sim -g
./test.sh -f tsim/stream/basic1.sim
./test.sh -f tsim/stream/basic2.sim
./test.sh -f tsim/stream/basic3.sim
./test.sh -f tsim/stream/basic4.sim
./test.sh -f tsim/stream/checkpointInterval0.sim
./test.sh -f tsim/stream/checkStreamSTable1.sim
./test.sh -f tsim/stream/checkStreamSTable.sim
./test.sh -f tsim/stream/deleteInterval.sim
./test.sh -f tsim/stream/deleteSession.sim
./test.sh -f tsim/stream/deleteState.sim
./test.sh -f tsim/stream/distributeInterval0.sim
./test.sh -f tsim/stream/distributeIntervalRetrive0.sim
./test.sh -f tsim/stream/distributeSession0.sim
./test.sh -f tsim/stream/drop_stream.sim
./test.sh -f tsim/stream/fillHistoryBasic1.sim
./test.sh -f tsim/stream/fillHistoryBasic2.sim
./test.sh -f tsim/stream/fillHistoryBasic3.sim
./test.sh -f tsim/stream/fillIntervalDelete0.sim
./test.sh -f tsim/stream/fillIntervalDelete1.sim
./test.sh -f tsim/stream/fillIntervalLinear.sim
./test.sh -f tsim/stream/fillIntervalPartitionBy.sim
./test.sh -f tsim/stream/fillIntervalPrevNext1.sim
./test.sh -f tsim/stream/fillIntervalPrevNext.sim
./test.sh -f tsim/stream/fillIntervalRange.sim
./test.sh -f tsim/stream/fillIntervalValue.sim
./test.sh -f tsim/stream/ignoreCheckUpdate.sim
./test.sh -f tsim/stream/ignoreExpiredData.sim
./test.sh -f tsim/stream/partitionby1.sim
./test.sh -f tsim/stream/partitionbyColumnInterval.sim
./test.sh -f tsim/stream/partitionbyColumnSession.sim
./test.sh -f tsim/stream/partitionbyColumnState.sim
./test.sh -f tsim/stream/partitionby.sim
./test.sh -f tsim/stream/pauseAndResume.sim
./test.sh -f tsim/stream/schedSnode.sim
./test.sh -f tsim/stream/session0.sim
./test.sh -f tsim/stream/session1.sim
./test.sh -f tsim/stream/sliding.sim
./test.sh -f tsim/stream/state0.sim
./test.sh -f tsim/stream/state1.sim
./test.sh -f tsim/stream/triggerInterval0.sim
./test.sh -f tsim/stream/triggerSession0.sim
./test.sh -f tsim/stream/udTableAndTag0.sim
./test.sh -f tsim/stream/udTableAndTag1.sim
./test.sh -f tsim/stream/udTableAndTag2.sim
./test.sh -f tsim/stream/windowClose.sim
./test.sh -f tsim/trans/lossdata1.sim
./test.sh -f tsim/tmq/basic1.sim
./test.sh -f tsim/tmq/basic2.sim
@ -431,3 +390,8 @@
./test.sh -f tsim/tag/drop_tag.sim
./test.sh -f tsim/tag/tbNameIn.sim
./test.sh -f tmp/monitor.sim
./test.sh -f tsim/tagindex/add_index.sim
./test.sh -f tsim/tagindex/sma_and_tag_index.sim
./test.sh -f tsim/view/view.sim
./test.sh -f tsim/query/cache_last.sim
./test.sh -f tsim/query/const.sim

19
tests/system-test/0-others/user_privilege.py
View File

@ -27,6 +27,7 @@ class TDTestCase:
tdSql.init(conn.cursor())
self.setsql = TDSetSql()
self.stbname = 'stb'
self.user_name = 'test'
self.binary_length = 20 # the length of binary for column_dict
self.nchar_length = 20 # the length of nchar for column_dict
self.dbnames = ['db1', 'db2']
@ -54,12 +55,12 @@ class TDTestCase:
]
self.tbnum = 4
self.stbnum_grant = 200
def create_user(self):
user_name = 'test'
tdSql.execute(f'create user {user_name} pass "test"')
tdSql.execute(f'grant read on {self.dbnames[0]}.{self.stbname} with t2 = "Beijing" to {user_name}')
tdSql.execute(f'grant write on {self.dbnames[1]}.{self.stbname} with t1 = 2 to {user_name}')
tdSql.execute(f'create user {self.user_name} pass "test"')
tdSql.execute(f'grant read on {self.dbnames[0]}.{self.stbname} with t2 = "Beijing" to {self.user_name}')
tdSql.execute(f'grant write on {self.dbnames[1]}.{self.stbname} with t1 = 2 to {self.user_name}')
def prepare_data(self):
for db in self.dbnames:
@ -70,6 +71,8 @@ class TDTestCase:
tdSql.execute(f'create table {self.stbname}_{i} using {self.stbname} tags({self.tag_list[i]})')
for j in self.values_list:
tdSql.execute(f'insert into {self.stbname}_{i} values({j})')
for i in range(self.stbnum_grant):
tdSql.execute(f'create table {self.stbname}_grant_{i} (ts timestamp, c0 int) tags(t0 int)')
def user_read_privilege_check(self, dbname):
testconn = taos.connect(user='test', password='test')
@ -128,12 +131,20 @@ class TDTestCase:
tdLog.exit(f"{caller.filename}({caller.lineno}) failed: sql:{sql}, expect error not occured")
pass
def user_privilege_grant_check(self):
for db in self.dbnames:
tdSql.execute(f"use {db}")
for i in range(self.stbnum_grant):
tdSql.execute(f'grant read on {db}.{self.stbname}_grant_{i} to {self.user_name}')
tdSql.execute(f'grant write on {db}.{self.stbname}_grant_{i} to {self.user_name}')
def run(self):
self.prepare_data()
self.create_user()
self.user_read_privilege_check(self.dbnames[0])
self.user_write_privilege_check(self.dbnames[1])
self.user_privilege_error_check()
self.user_privilege_grant_check()
def stop(self):
tdSql.close()

6
tests/system-test/7-tmq/tmqCommon.py
View File

@ -75,7 +75,7 @@ class TMQCom:
if tdSql.getRows() == expectRows:
break
else:
time.sleep(5)
time.sleep(0.5)
for i in range(expectRows):
tdLog.info ("consume id: %d, consume msgs: %d, consume rows: %d"%(tdSql.getData(i , 1), tdSql.getData(i , 2), tdSql.getData(i , 3)))
@ -156,7 +156,7 @@ class TMQCom:
tdLog.info("row: %d"%(actRows))
if (actRows >= rows):
loopFlag = 0
time.sleep(0.02)
time.sleep(0.5)
return
def getStartCommitNotifyFromTmqsim(self,cdbName='cdb',rows=1):
@ -167,7 +167,7 @@ class TMQCom:
tdLog.info("row: %d"%(actRows))
if (actRows >= rows):
loopFlag = 0
time.sleep(0.02)
time.sleep(0.5)
return
def create_database(self,tsql, dbName,dropFlag=1,vgroups=4,replica=1):

218
tests/system-test/7-tmq/tmqVnodeSplit-stb-select-duplicatedata-false.py Normal file
View File

@ -0,0 +1,218 @@
import taos
import sys
import time
import socket
import os
import threading
import math
from util.log import *
from util.sql import *
from util.cases import *
from util.dnodes import *
from util.common import *
from util.cluster import *
sys.path.append("./7-tmq")
from tmqCommon import *
sys.path.append("./6-cluster")
from clusterCommonCreate import *
from clusterCommonCheck import clusterComCheck
class TDTestCase:
def __init__(self):
self.vgroups = 1
self.ctbNum = 10
self.rowsPerTbl = 10000
def init(self, conn, logSql, replicaVar=1):
self.replicaVar = int(replicaVar)
tdLog.debug(f"start to excute {__file__}")
tdSql.init(conn.cursor(), False)
def getDataPath(self):
selfPath = tdCom.getBuildPath()
return selfPath + '/../sim/dnode%d/data/vnode/vnode%d/wal/*';
def prepareTestEnv(self):
tdLog.printNoPrefix("======== prepare test env include database, stable, ctables, and insert data: ")
paraDict = {'dbName': 'dbt',
'dropFlag': 1,
'event': '',
'vgroups': 1,
'stbName': 'stb',
'colPrefix': 'c',
'tagPrefix': 't',
'colSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1},{'type': 'TIMESTAMP', 'count':1}],
'tagSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1}],
'ctbPrefix': 'ctb',
'ctbStartIdx': 0,
'ctbNum': 10,
'rowsPerTbl': 10000,
'batchNum': 10,
'startTs': 1640966400000, # 2022-01-01 00:00:00.000
'pollDelay': 60,
'showMsg': 1,
'showRow': 1,
'snapshot': 0}
paraDict['vgroups'] = self.vgroups
paraDict['ctbNum'] = self.ctbNum
paraDict['rowsPerTbl'] = self.rowsPerTbl
tdCom.drop_all_db()
tmqCom.initConsumerTable()
tdCom.create_database(tdSql, paraDict["dbName"],paraDict["dropFlag"], wal_retention_period=36000,vgroups=paraDict["vgroups"],replica=self.replicaVar)
tdLog.info("create stb")
tmqCom.create_stable(tdSql, dbName=paraDict["dbName"],stbName=paraDict["stbName"])
return
def restartAndRemoveWal(self, deleteWal):
tdDnodes = cluster.dnodes
tdSql.query("select * from information_schema.ins_vnodes")
for result in tdSql.queryResult:
if result[2] == 'dbt':
tdLog.debug("dnode is %d"%(result[0]))
dnodeId = result[0]
vnodeId = result[1]
tdDnodes[dnodeId - 1].stoptaosd()
time.sleep(1)
dataPath = self.getDataPath()
dataPath = dataPath%(dnodeId,vnodeId)
tdLog.debug("dataPath:%s"%dataPath)
if deleteWal:
if os.system('rm -rf ' + dataPath) != 0:
tdLog.exit("rm error")
tdDnodes[dnodeId - 1].starttaosd()
time.sleep(1)
break
tdLog.debug("restart dnode ok")
def splitVgroups(self):
tdSql.query("select * from information_schema.ins_vnodes")
vnodeId = 0
for result in tdSql.queryResult:
if result[2] == 'dbt':
vnodeId = result[1]
tdLog.debug("vnode is %d"%(vnodeId))
break
splitSql = "split vgroup %d" %(vnodeId)
tdLog.debug("splitSql:%s"%(splitSql))
tdSql.query(splitSql)
tdLog.debug("splitSql ok")
def tmqCase1(self, deleteWal=False):
tdLog.printNoPrefix("======== test case 1: ")
paraDict = {'dbName': 'dbt',
'dropFlag': 1,
'event': '',
'vgroups': 1,
'stbName': 'stb',
'colPrefix': 'c',
'tagPrefix': 't',
'colSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1},{'type': 'TIMESTAMP', 'count':1}],
'tagSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1}],
'ctbPrefix': 'ctb1',
'ctbStartIdx': 0,
'ctbNum': 10,
'rowsPerTbl': 10000,
'batchNum': 10,
'startTs': 1640966400000, # 2022-01-01 00:00:00.000
'pollDelay': 120,
'showMsg': 1,
'showRow': 1,
'snapshot': 0}
paraDict['vgroups'] = self.vgroups
paraDict['ctbNum'] = self.ctbNum
paraDict['rowsPerTbl'] = self.rowsPerTbl
topicNameList = ['topic1']
# expectRowsList = []
tmqCom.initConsumerTable()
tdLog.info("create topics from stb with filter")
queryString = "select * from %s.%s"%(paraDict['dbName'], paraDict['stbName'])
# sqlString = "create topic %s as stable %s" %(topicNameList[0], paraDict['stbName'])
sqlString = "create topic %s as %s" %(topicNameList[0], queryString)
tdLog.info("create topic sql: %s"%sqlString)
tdSql.execute(sqlString)
# tdSql.query(queryString)
# expectRowsList.append(tdSql.getRows())
# init consume info, and start tmq_sim, then check consume result
tdLog.info("insert consume info to consume processor")
consumerId = 0
expectrowcnt = paraDict["rowsPerTbl"] * paraDict["ctbNum"] * 2
topicList = topicNameList[0]
ifcheckdata = 1
ifManualCommit = 1
keyList = 'group.id:cgrp1, enable.auto.commit:true, auto.commit.interval.ms:200, auto.offset.reset:earliest'
tmqCom.insertConsumerInfo(consumerId, expectrowcnt,topicList,keyList,ifcheckdata,ifManualCommit)
tdLog.info("start consume processor")
tmqCom.startTmqSimProcess(pollDelay=paraDict['pollDelay'],dbName=paraDict["dbName"],showMsg=paraDict['showMsg'], showRow=paraDict['showRow'],snapshot=paraDict['snapshot'])
tdLog.info("wait the consume result")
tdLog.info("create ctb1")
tmqCom.create_ctable(tdSql, dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict['ctbPrefix'],
ctbNum=paraDict["ctbNum"],ctbStartIdx=paraDict['ctbStartIdx'])
tdLog.info("create ctb2")
paraDict['ctbPrefix'] = "ctb2"
tmqCom.create_ctable(tdSql, dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict['ctbPrefix'],
ctbNum=paraDict["ctbNum"],ctbStartIdx=paraDict['ctbStartIdx'])
tdLog.info("insert ctb1 data")
pInsertThread = tmqCom.asyncInsertDataByInterlace(paraDict)
tmqCom.getStartConsumeNotifyFromTmqsim()
tmqCom.getStartCommitNotifyFromTmqsim()
#restart dnode & remove wal
self.restartAndRemoveWal(deleteWal)
# split vgroup
self.splitVgroups()
tdLog.info("insert ctb2 data")
pInsertThread1 = tmqCom.asyncInsertDataByInterlace(paraDict)
pInsertThread.join()
pInsertThread1.join()
expectRows = 1
resultList = tmqCom.selectConsumeResult(expectRows)
if expectrowcnt / 2 >= resultList[0]:
tdLog.info("expect consume rows: %d, act consume rows: %d"%(expectrowcnt / 2, resultList[0]))
tdLog.exit("%d tmq consume rows error!"%consumerId)
# tmqCom.checkFileContent(consumerId, queryString)
time.sleep(2)
for i in range(len(topicNameList)):
tdSql.query("drop topic %s"%topicNameList[i])
if deleteWal == True:
clusterComCheck.check_vgroups_status(vgroup_numbers=2,db_replica=self.replicaVar,db_name="dbt",count_number=240)
tdLog.printNoPrefix("======== test case 1 end ...... ")
def run(self):
self.prepareTestEnv()
self.tmqCase1(False)
def stop(self):
tdSql.close()
tdLog.success(f"{__file__} successfully executed")
event = threading.Event()
tdCases.addLinux(__file__, TDTestCase())
tdCases.addWindows(__file__, TDTestCase())

2
tests/system-test/7-tmq/tmqVnodeSplit-stb-select-duplicatedata.py
View File

@ -207,8 +207,6 @@ class TDTestCase:
def run(self):
self.prepareTestEnv()
self.tmqCase1(True)
self.prepareTestEnv()
self.tmqCase1(False)
def stop(self):
tdSql.close()

196
tests/system-test/7-tmq/tmq_per.py Normal file
View File

@ -0,0 +1,196 @@
import taos
import sys
import time
import socket
import os
import threading
import math
from util.log import *
from util.sql import *
from util.cases import *
from util.dnodes import *
from util.common import *
from util.cluster import *
sys.path.append("./7-tmq")
from tmqCommon import *
from util.cluster import *
sys.path.append("./6-cluster")
from clusterCommonCreate import *
from clusterCommonCheck import clusterComCheck
class TDTestCase:
def __init__(self):
self.vgroups = 1
self.ctbNum = 10000
self.rowsPerTbl = 10000
def init(self, conn, logSql, replicaVar=1):
self.replicaVar = int(replicaVar)
tdLog.debug(f"start to excute {__file__}")
tdSql.init(conn.cursor(), False)
def getDataPath(self):
selfPath = tdCom.getBuildPath()
return selfPath + '/../sim/dnode%d/data/vnode/vnode%d/wal/*';
def prepareTestEnv(self):
tdLog.printNoPrefix("======== prepare test env include database, stable, ctables, and insert data: ")
paraDict = {'dbName': 'dbt',
'dropFlag': 1,
'event': '',
'vgroups': 1,
'stbName': 'stb',
'colPrefix': 'c',
'tagPrefix': 't',
'colSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1},{'type': 'TIMESTAMP', 'count':1}],
'tagSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1}],
'ctbPrefix': 'ctb',
'ctbStartIdx': 0,
'ctbNum': 10,
'rowsPerTbl': 10000,
'batchNum': 10,
'startTs': 1640966400000, # 2022-01-01 00:00:00.000
'pollDelay': 60,
'showMsg': 1,
'showRow': 1,
'snapshot': 0}
paraDict['vgroups'] = self.vgroups
paraDict['ctbNum'] = self.ctbNum
paraDict['rowsPerTbl'] = self.rowsPerTbl
tdCom.drop_all_db()
tmqCom.initConsumerTable()
tdCom.create_database(tdSql, paraDict["dbName"],paraDict["dropFlag"], wal_retention_period=36000,vgroups=paraDict["vgroups"],replica=self.replicaVar)
tdLog.info("create stb")
tmqCom.create_stable(tdSql, dbName=paraDict["dbName"],stbName=paraDict["stbName"])
return
def restartAndRemoveWal(self, deleteWal):
tdDnodes = cluster.dnodes
tdSql.query("select * from information_schema.ins_vnodes")
for result in tdSql.queryResult:
if result[2] == 'dbt':
tdLog.debug("dnode is %d"%(result[0]))
dnodeId = result[0]
vnodeId = result[1]
tdDnodes[dnodeId - 1].stoptaosd()
time.sleep(1)
dataPath = self.getDataPath()
dataPath = dataPath%(dnodeId,vnodeId)
tdLog.debug("dataPath:%s"%dataPath)
if deleteWal:
if os.system('rm -rf ' + dataPath) != 0:
tdLog.exit("rm error")
tdDnodes[dnodeId - 1].starttaosd()
time.sleep(1)
break
tdLog.debug("restart dnode ok")
def splitVgroups(self):
tdSql.query("select * from information_schema.ins_vnodes")
vnodeId = 0
for result in tdSql.queryResult:
if result[2] == 'dbt':
vnodeId = result[1]
tdLog.debug("vnode is %d"%(vnodeId))
break
splitSql = "split vgroup %d" %(vnodeId)
tdLog.debug("splitSql:%s"%(splitSql))
tdSql.query(splitSql)
tdLog.debug("splitSql ok")
def tmqCase1(self, deleteWal=False):
tdLog.printNoPrefix("======== test case 1: ")
paraDict = {'dbName': 'dbt',
'dropFlag': 1,
'event': '',
'vgroups': 1,
'stbName': 'stb',
'colPrefix': 'c',
'tagPrefix': 't',
'colSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1},{'type': 'TIMESTAMP', 'count':1}],
'tagSchema': [{'type': 'INT', 'count':1},{'type': 'BIGINT', 'count':1},{'type': 'DOUBLE', 'count':1},{'type': 'BINARY', 'len':32, 'count':1},{'type': 'NCHAR', 'len':32, 'count':1}],
'ctbPrefix': 'ctb1',
'ctbStartIdx': 0,
'ctbNum': 10000,
'rowsPerTbl': 10000,
'batchNum': 10000,
'startTs': 1640966400000, # 2022-01-01 00:00:00.000
'pollDelay': 5,
'showMsg': 1,
'showRow': 1,
'snapshot': 0}
paraDict['vgroups'] = self.vgroups
paraDict['ctbNum'] = self.ctbNum
print
paraDict['rowsPerTbl'] = self.rowsPerTbl
topicNameList = ['topic1']
# expectRowsList = []
tmqCom.initConsumerTable()
tdLog.info("create topics from stb ")
queryString = "stable %s.%s"%(paraDict['dbName'], paraDict['stbName'])
# sqlString = "create topic %s as stable %s" %(topicNameList[0], paraDict['stbName'])
sqlString = "create topic %s as %s" %(topicNameList[0], queryString)
tdLog.info("create topic sql: %s"%sqlString)
tdSql.execute(sqlString)
# tdSql.query(queryString)
# expectRowsList.append(tdSql.getRows())
# init consume info, and start tmq_sim, then check consume result
tdLog.info("insert consume info to consume processor")
consumerId = 0
expectrowcnt = paraDict["rowsPerTbl"] * paraDict["ctbNum"]
topicList = topicNameList[0]
ifcheckdata = 1
ifManualCommit = 1
keyList = 'group.id:cgrp1, enable.auto.commit:true, auto.commit.interval.ms:200, auto.offset.reset:earliest'
tmqCom.insertConsumerInfo(consumerId, expectrowcnt,topicList,keyList,ifcheckdata,ifManualCommit)
tdLog.info("create ctb1")
tmqCom.create_ctable(tdSql, dbName=paraDict["dbName"],stbName=paraDict["stbName"],ctbPrefix=paraDict['ctbPrefix'],
ctbNum=paraDict["ctbNum"],ctbStartIdx=paraDict['ctbStartIdx'])
tdLog.info("insert ctb1 data")
pInsertThread = tmqCom.asyncInsertDataByInterlace(paraDict)
pInsertThread.join()
tdLog.info("start consume processor")
tmqCom.startTmqSimProcess(pollDelay=paraDict['pollDelay'],dbName=paraDict["dbName"],showMsg=paraDict['showMsg'], showRow=paraDict['showRow'],snapshot=paraDict['snapshot'])
tdLog.info("wait the consume result")
tmqCom.getStartConsumeNotifyFromTmqsim()
tmqCom.getStartCommitNotifyFromTmqsim()
expectRows = 1
tdLog.info("expectRows:%d"%expectRows)
resultList = tmqCom.selectConsumeResult(expectRows)
# for i in range(len(topicNameList)):
# tdSql.query("drop topic %s"%topicNameList[i])
if deleteWal == True:
clusterComCheck.check_vgroups_status(vgroup_numbers=1,db_replica=self.replicaVar,db_name="dbt",count_number=240)
tdLog.printNoPrefix("======== test case 1 end ...... ")
def run(self):
self.prepareTestEnv()
self.tmqCase1(True)
def stop(self):
tdSql.close()
tdLog.success(f"{__file__} successfully executed")
event = threading.Event()
tdCases.addLinux(__file__, TDTestCase())
tdCases.addWindows(__file__, TDTestCase())

79
tests/system-test/win-test-file
View File

@ -1,3 +1,12 @@
python3 ./test.py -f 2-query/tbname_vgroup.py
python3 ./test.py -f 2-query/stbJoin.py
python3 ./test.py -f 2-query/stbJoin.py -Q 2
python3 ./test.py -f 2-query/stbJoin.py -Q 3
python3 ./test.py -f 2-query/stbJoin.py -Q 4
python3 ./test.py -f 2-query/hint.py
python3 ./test.py -f 2-query/hint.py -Q 2
python3 ./test.py -f 2-query/hint.py -Q 3
python3 ./test.py -f 2-query/hint.py -Q 4
python3 ./test.py -f 2-query/nestedQuery.py
python3 ./test.py -f 2-query/nestedQuery_str.py
python3 ./test.py -f 2-query/nestedQuery_math.py
@ -18,7 +27,30 @@ python3 ./test.py -f 2-query/nestedQuery_math.py -Q 4
python3 ./test.py -f 2-query/nestedQuery_time.py -Q 4
python3 ./test.py -f 2-query/nestedQuery_26.py -Q 4
python3 ./test.py -f 2-query/interval_limit_opt.py -Q 4
python3 ./test.py -f 2-query/interval_unit.py
python3 ./test.py -f 2-query/interval_unit.py -Q 2
python3 ./test.py -f 2-query/interval_unit.py -Q 3
python3 ./test.py -f 2-query/interval_unit.py -Q 4
python3 ./test.py -f 2-query/partition_by_col.py -Q 4
python3 ./test.py -f 2-query/partition_by_col.py -Q 3
python3 ./test.py -f 2-query/partition_by_col.py -Q 2
python3 ./test.py -f 2-query/partition_by_col.py
python3 ./test.py -f 2-query/partition_by_col_agg.py
python3 ./test.py -f 2-query/partition_by_col_agg.py -Q 2
python3 ./test.py -f 2-query/partition_by_col_agg.py -Q 3
python3 ./test.py -f 2-query/partition_by_col_agg.py -Q 4
python3 ./test.py -f 2-query/interval_limit_opt_2.py -Q 4
python3 ./test.py -f 2-query/interval_limit_opt_2.py -Q 3
python3 ./test.py -f 2-query/interval_limit_opt_2.py -Q 2
python3 ./test.py -f 2-query/interval_limit_opt_2.py
python3 ./test.py -f 2-query/func_to_char_timestamp.py
python3 ./test.py -f 2-query/func_to_char_timestamp.py -Q 2
python3 ./test.py -f 2-query/func_to_char_timestamp.py -Q 3
python3 ./test.py -f 2-query/func_to_char_timestamp.py -Q 4
python3 ./test.py -f 2-query/last_cache_scan.py
python3 ./test.py -f 2-query/last_cache_scan.py -Q 2
python3 ./test.py -f 2-query/last_cache_scan.py -Q 3
python3 ./test.py -f 2-query/last_cache_scan.py -Q 4
python3 ./test.py -f 7-tmq/tmqShow.py
python3 ./test.py -f 7-tmq/tmqDropStb.py
python3 ./test.py -f 7-tmq/subscribeStb0.py
@ -29,11 +61,13 @@ python3 ./test.py -f 7-tmq/subscribeDb0.py -N 3 -n 3
python3 ./test.py -f 7-tmq/ins_topics_test.py
python3 ./test.py -f 7-tmq/tmqMaxTopic.py
python3 ./test.py -f 7-tmq/tmqParamsTest.py
python3 ./test.py -f 7-tmq/tmqParamsTest.py -R
python3 ./test.py -f 7-tmq/tmqClientConsLog.py
python3 ./test.py -f 7-tmq/tmqMaxGroupIds.py
python3 ./test.py -f 7-tmq/tmqConsumeDiscontinuousData.py
python3 ./test.py -f 7-tmq/tmqOffset.py
python3 ./test.py -f 7-tmq/tmqDropConsumer.py
python3 ./test.py -f 1-insert/insert_stb.py
python3 ./test.py -f 1-insert/delete_stable.py
python3 ./test.py -f 2-query/out_of_order.py -Q 3
python3 ./test.py -f 2-query/out_of_order.py
@ -61,10 +95,14 @@ python3 ./test.py -f 2-query/slimit.py -R
python3 ./test.py -f 2-query/slimit.py -Q 2
python3 ./test.py -f 2-query/slimit.py -Q 3
python3 ./test.py -f 2-query/slimit.py -Q 4
python3 ./test.py -f 3-enterprise/restore/restoreDnode.py -N 5 -M 3
python3 ./test.py -f 3-enterprise/restore/restoreVnode.py -N 5 -M 3
python3 ./test.py -f 3-enterprise/restore/restoreMnode.py -N 5 -M 3
python3 ./test.py -f 3-enterprise/restore/restoreQnode.py -N 5 -M 3
python3 ./test.py -f 2-query/ts-4233.py
python3 ./test.py -f 2-query/ts-4233.py -Q 2
python3 ./test.py -f 2-query/ts-4233.py -Q 3
python3 ./test.py -f 2-query/ts-4233.py -Q 4
python3 ./test.py -f 3-enterprise/restore/restoreDnode.py -N 5 -M 3 -i False
python3 ./test.py -f 3-enterprise/restore/restoreVnode.py -N 5 -M 3 -i False
python3 ./test.py -f 3-enterprise/restore/restoreMnode.py -N 5 -M 3 -i False
python3 ./test.py -f 3-enterprise/restore/restoreQnode.py -N 5 -M 3 -i False
python3 ./test.py -f 7-tmq/create_wrong_topic.py
python3 ./test.py -f 7-tmq/dropDbR3ConflictTransaction.py -N 3
python3 ./test.py -f 7-tmq/basic5.py
@ -112,16 +150,30 @@ python3 ./test.py -f 7-tmq/stbTagFilter-1ctb.py
python3 ./test.py -f 7-tmq/dataFromTsdbNWal.py
python3 ./test.py -f 7-tmq/dataFromTsdbNWal-multiCtb.py
python3 ./test.py -f 7-tmq/tmq_taosx.py
python3 ./test.py -f 7-tmq/tmq_replay.py
python3 ./test.py -f 7-tmq/tmqSeekAndCommit.py
python3 ./test.py -f 7-tmq/tmq_offset.py
python3 ./test.py -f 7-tmq/tmqDataPrecisionUnit.py
python3 ./test.py -f 7-tmq/raw_block_interface_test.py
python3 ./test.py -f 7-tmq/stbTagFilter-multiCtb.py
python3 ./test.py -f 7-tmq/tmqSubscribeStb-r3.py -N 5
python3 ./test.py -f 7-tmq/tmq3mnodeSwitch.py -N 6 -M 3
python3 ./test.py -f 7-tmq/tmq3mnodeSwitch.py -N 6 -M 3 -n 3
python3 ./test.py -f 7-tmq/tmq3mnodeSwitch.py -N 6 -M 3 -i True
python3 ./test.py -f 7-tmq/tmq3mnodeSwitch.py -N 6 -M 3 -n 3 -i True
python3 test.py -f 7-tmq/tmqVnodeTransform-stb.py -N 2 -n 1
python3 test.py -f 7-tmq/tmqVnodeTransform-stb.py -N 6 -n 3
python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select.py -N 2 -n 1
python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select-duplicatedata.py -N 3 -n 3
python3 test.py -f 7-tmq/tmqVnodeSplit-stb-select.py -N 3 -n 3
python3 test.py -f 7-tmq/tmqVnodeSplit-stb.py -N 3 -n 3
python3 test.py -f 7-tmq/tmqVnodeSplit-column.py -N 3 -n 3
python3 test.py -f 7-tmq/tmqVnodeSplit-db.py -N 3 -n 3
python3 test.py -f 7-tmq/tmqVnodeReplicate.py -M 3 -N 3 -n 3
python3 ./test.py -f 99-TDcase/TD-19201.py
python3 ./test.py -f 99-TDcase/TD-21561.py
python3 ./test.py -f 99-TDcase/TS-3404.py
python3 ./test.py -f 99-TDcase/TS-3581.py
python3 ./test.py -f 99-TDcase/TS-3311.py
python3 ./test.py -f 99-TDcase/TS-3821.py
python3 ./test.py -f 0-others/balance_vgroups_r1.py -N 6
python3 ./test.py -f 0-others/taosShell.py
python3 ./test.py -f 0-others/taosShellError.py
@ -139,10 +191,19 @@ python3 ./test.py -f 0-others/user_privilege_show.py
python3 ./test.py -f 0-others/user_privilege_all.py
python3 ./test.py -f 0-others/fsync.py
python3 ./test.py -f 0-others/multilevel.py
python3 ./test.py -f 0-others/ttl.py
python3 ./test.py -f 0-others/ttlChangeOnWrite.py
python3 ./test.py -f 0-others/compress_tsz1.py
python3 ./test.py -f 0-others/compress_tsz2.py
python3 ./test.py -f 0-others/view/non_marterial_view/test_view.py
python3 ./test.py -f 0-others/compatibility.py
python3 ./test.py -f 0-others/tag_index_basic.py
python3 ./test.py -N 3 -f 0-others/walRetention.py
python3 ./test.py -f 0-others/splitVGroupRep1.py -N 3
python3 ./test.py -f 0-others/splitVGroupRep3.py -N 3
python3 ./test.py -f 0-others/timeRangeWise.py -N 3
python3 ./test.py -f 0-others/delete_check.py
python3 ./test.py -f 0-others/test_hot_refresh_configurations.py
python3 ./test.py -f 1-insert/alter_database.py
python3 ./test.py -f 1-insert/alter_replica.py -N 3
python3 ./test.py -f 1-insert/influxdb_line_taosc_insert.py
@ -190,6 +251,7 @@ python3 ./test.py -f 1-insert/rowlength64k_4.py -Q 3
python3 ./test.py -f 1-insert/rowlength64k_4.py -Q 4
python3 ./test.py -f 1-insert/precisionUS.py
python3 ./test.py -f 1-insert/precisionNS.py
python3 ./test.py -f 1-insert/test_ts4219.py
python3 ./test.py -f 0-others/show.py
python3 ./test.py -f 0-others/show_tag_index.py
python3 ./test.py -f 0-others/information_schema.py
@ -325,6 +387,8 @@ python3 ./test.py -f 2-query/smaTest.py
python3 ./test.py -f 2-query/smaTest.py -R
python3 ./test.py -f 0-others/sma_index.py
python3 ./test.py -f 2-query/sml_TS-3724.py
python3 ./test.py -f 2-query/sml-TD19291.py
python3 ./test.py -f 2-query/varbinary.py
python3 ./test.py -f 2-query/sml.py
python3 ./test.py -f 2-query/sml.py -R
python3 ./test.py -f 2-query/spread.py
@ -426,7 +490,6 @@ python3 ./test.py -f 6-cluster/5dnode3mnodeRestartDnodeInsertData.py -N 6 -M 3
python3 ./test.py -f 6-cluster/5dnode3mnodeRestartDnodeInsertData.py -N 6 -M 3 -n 3
python3 ./test.py -f 6-cluster/5dnode3mnodeRestartDnodeInsertDataAsync.py -N 6 -M 3
python3 ./test.py -f 6-cluster/manually-test/6dnode3mnodeInsertLessDataAlterRep3to1to3.py -N 6 -M 3
python3 ./test.py -f 6-cluster/5dnode3mnodeRoll.py -N 3 -C 1
python3 ./test.py -f 6-cluster/5dnode3mnodeAdd1Ddnoe.py -N 7 -M 3 -C 6
python3 ./test.py -f 6-cluster/5dnode3mnodeAdd1Ddnoe.py -N 7 -M 3 -C 6 -n 3
python3 ./test.py -f 6-cluster/5dnode3mnodeRecreateMnode.py -N 6 -M 3
@ -731,7 +794,9 @@ python3 ./test.py -f 2-query/out_of_order.py -R
python3 ./test.py -f 2-query/blockSMA.py -Q 4
python3 ./test.py -f 2-query/projectionDesc.py -Q 4
python3 ./test.py -f 2-query/odbc.py
python3 ./test.py -f 2-query/fill_with_group.py
python3 ./test.py -f 99-TDcase/TD-21561.py -Q 4
python3 ./test.py -f 99-TDcase/TD-20582.py
python3 ./test.py -f 5-taos-tools/taosbenchmark/insertMix.py -N 3
python3 ./test.py -f 5-taos-tools/taosbenchmark/stt.py -N 3
python3 ./test.py -f eco-system/meta/database/keep_time_offset.py