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

Merge branch 'feature/TD-13041' of github.com:taosdata/TDengine into feature/TD-13041

This commit is contained in:
wangmm0220 2022-06-27 21:21:22 +08:00
parent b6b4455689 8e5099c44c
commit e2f4a2a03c
170 changed files with 7032 additions and 2177 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

59
Jenkinsfile2
View File

@ -4,11 +4,6 @@ import jenkins.model.CauseOfInterruption
node {
}
win_test_stage = 0
linux_ready = 0
linux_node_ip = ""
linux_node_pass = ""
def abortPreviousBuilds() {
def currentJobName = env.JOB_NAME
def currentBuildNumber = env.BUILD_NUMBER.toInteger()
@ -84,6 +79,7 @@ def pre_test(){
git pull >/dev/null
git log -5
echo "`date "+%Y%m%d-%H%M%S"` ${JOB_NAME}:${BRANCH_NAME}:${BUILD_ID}:${CHANGE_TARGET}" >>${WKDIR}/jenkins.log
echo "CHANGE_BRANCH:${CHANGE_BRANCH}" >>${WKDIR}/jenkins.log
echo "community log: `git log -5`" >>${WKDIR}/jenkins.log
git fetch origin +refs/pull/${CHANGE_ID}/merge
git checkout -qf FETCH_HEAD
@ -100,6 +96,7 @@ def pre_test(){
git pull >/dev/null
git log -5
echo "`date "+%Y%m%d-%H%M%S"` ${JOB_NAME}:${BRANCH_NAME}:${BUILD_ID}:${CHANGE_TARGET}" >>${WKDIR}/jenkins.log
echo "CHANGE_BRANCH:${CHANGE_BRANCH}" >>${WKDIR}/jenkins.log
echo "tdinternal log: `git log -5`" >>${WKDIR}/jenkins.log
git fetch origin +refs/pull/${CHANGE_ID}/merge
git checkout -qf FETCH_HEAD
@ -289,7 +286,6 @@ def run_win_ctest() {
'''
}
def run_win_test() {
echo "LINUX NODE: ${linux_node_ip} - ${linux_node_pass}"
bat '''
echo "windows test ..."
cd %WIN_CONNECTOR_ROOT%
@ -298,9 +294,8 @@ def run_win_test() {
ls -l C:\\Windows\\System32\\taos.dll
time /t
cd %WIN_SYSTEM_TEST_ROOT%
echo "node: ''' + linux_node_ip + ''':''' + linux_node_pass + '''"
echo "testing ..."
test-all.bat "{\\\"host\\\":\\\"''' + linux_node_ip + '''\\\",\\\"port\\\":22,\\\"user\\\":\\\"root\\\",\\\"password\\\":\\\"''' + linux_node_pass + '''\\\",\\\"path\\\":\\\"/var/lib/jenkins/workspace/TDinternal\\\"}"
test-all.bat ci
time /t
'''
}
@ -331,17 +326,9 @@ pipeline {
pre_test_win()
pre_test_build_win()
run_win_ctest()
script {
while(linux_ready == 0) {
sleep(8)
}
}
run_win_test()
}
}
script {
win_test_stage = 1
}
}
}
stage('linux test') {
@ -352,15 +339,12 @@ pipeline {
}
steps {
script {
linux_node_ip = sh (
script: 'jq .ip /home/node_info.json | sed "s/\\\"//g"',
def linux_node_ip = sh (
script: 'ip addr|grep 192|grep -v virbr|awk "{print \\\$2}"|sed "s/\\/.*//"',
returnStdout: true
).trim()
linux_node_pass = sh (
script: 'jq .password /home/node_info.json | sed "s/\\\"//g" |sed "s/\\!/^^^^^^^^\\!/g"',
returnStdout: true
).trim()
echo "${linux_node_ip}:${linux_node_pass}"
echo "${linux_node_ip}"
echo "${WKDIR}/restore.sh -p ${BRANCH_NAME} -n ${BUILD_ID} -c {container name}"
}
catchError(buildResult: 'FAILURE', stageResult: 'FAILURE') {
timeout(time: 120, unit: 'MINUTES'){
@ -418,38 +402,9 @@ pipeline {
cd ${WKC}/packaging
./release.sh -v cluster -n 3.0.0.100 -s static
'''
sh '''
echo "install ..."
cd ${WKC}/release
tar xzf TDengine-enterprise-server-3.0.0.100-Linux-x64.tar.gz
cd TDengine-enterprise-server-3.0.0.100
service taosd stop || :
rm -rf /var/lib/taos
./install.sh -e no
'''
sh '''
echo "checking ..."
which taos
which taosd
rm -rf ${WK}/debug
mv ${WKC}/debug ${WK}/
'''
sh '''
echo "install taospy ..."
cd ${WKPY}
pip3 install .
'''
}
}
}
script {
linux_ready = 1
}
script {
while(win_test_stage == 0){
sleep(12)
}
}
}
}
}

2
docs/en/21-tdinternal/01-arch.md
View File

@ -22,7 +22,7 @@ A complete TDengine system runs on one or more physical nodes. Logically, it inc
**Virtual node (vnode)**: To better support data sharding, load balancing and prevent data from overheating or skewing, data nodes are virtualized into multiple virtual nodes (vnode, V2, V3, V4, etc. in the figure). Each vnode is a relatively independent work unit, which is the basic unit of time-series data storage and has independent running threads, memory space and persistent storage path. A vnode contains a certain number of tables (data collection points). When a new table is created, the system checks whether a new vnode needs to be created. The number of vnodes that can be created on a data node depends on the capacity of the hardware of the physical node where the data node is located. A vnode belongs to only one DB, but a DB can have multiple vnodes. In addition to the stored time-series data, a vnode also stores the schema and tag values of the included tables. A virtual node is uniquely identified in the system by the EP of the data node and the VGroup ID to which it belongs and is created and managed by the management node.
**Management node (mnode)**: A virtual logical unit responsible for monitoring and maintaining the running status of all data nodes and load balancing among nodes (M in the figure). At the same time, the management node is also responsible for the storage and management of metadata (including users, databases, tables, static tags, etc.), so it is also called Meta Node. Multiple (up to 5) mnodes can be configured in a TDengine cluster, and they are automatically constructed into a virtual management node group (M0, M1, M2 in the figure). The leader/follower mechanism is adopted for the mnode group and the data synchronization is carried out in a strongly consistent way. Any data update operation can only be executed on the leader. The creation of mnode cluster is completed automatically by the system without manual intervention. There is at most one mnode on each dnode, which is uniquely identified by the EP of the data node to which it belongs. Each dnode automatically obtains the EP of the dnode where all mnodes in the whole cluster are located, through internal messaging interaction.
**Management node (mnode)**: A virtual logical unit responsible for monitoring and maintaining the running status of all data nodes and load balancing among nodes (M in the figure). At the same time, the management node is also responsible for the storage and management of metadata (including users, databases, tables, static tags, etc.), so it is also called Meta Node. Multiple (up to 3) mnodes can be configured in a TDengine cluster, and they are automatically constructed into a virtual management node group (M0, M1, M2 in the figure). The leader/follower mechanism is adopted for the mnode group and the data synchronization is carried out in a strongly consistent way. Any data update operation can only be executed on the leader. The creation of mnode cluster is completed automatically by the system without manual intervention. There is at most one mnode on each dnode, which is uniquely identified by the EP of the data node to which it belongs. Each dnode automatically obtains the EP of the dnode where all mnodes in the whole cluster are located, through internal messaging interaction.
**Virtual node group (VGroup)**: Vnodes on different data nodes can form a virtual node group to ensure the high availability of the system. The virtual node group is managed in a leader/follower mechanism. Write operations can only be performed on the leader vnode, and then replicated to follower vnodes, thus ensuring that one single replica of data is copied on multiple physical nodes. The number of virtual nodes in a vgroup equals the number of data replicas. If the number of replicas of a DB is N, the system must have at least N data nodes. The number of replicas can be specified by the parameter `“replica”` when creating a DB, and the default is 1. Using the multi-replication feature of TDengine, the same high data reliability can be achieved without the need for expensive storage devices such as disk arrays. Virtual node groups are created and managed by the management node, and the management node assigns a system unique ID, aka VGroup ID. If two virtual nodes have the same vnode group ID, it means that they belong to the same group and the data is backed up to each other. The number of virtual nodes in a virtual node group can be dynamically changed, allowing only one, that is, no data replication. VGroup ID is never changed. Even if a virtual node group is deleted, its ID will not be reused.

2
docs/zh/21-tdinternal/01-arch.md
View File

@ -23,7 +23,7 @@ TDengine 分布式架构的逻辑结构图如下：
**虚拟节点（vnode）：** 为更好的支持数据分片、负载均衡，防止数据过热或倾斜，数据节点被虚拟化成多个虚拟节点（vnode，图中 V2，V3，V4 等）。每个 vnode 都是一个相对独立的工作单元，是时序数据存储的基本单元，具有独立的运行线程、内存空间与持久化存储的路径。一个 vnode 包含一定数量的表（数据采集点）。当创建一张新表时，系统会检查是否需要创建新的 vnode。一个数据节点上能创建的 vnode 的数量取决于该数据节点所在物理节点的硬件资源。一个 vnode 只属于一个 DB，但一个 DB 可以有多个 vnode。一个 vnode 除存储的时序数据外，也保存有所包含的表的 schema、标签值等。一个虚拟节点由所属的数据节点的 EP，以及所属的 VGroup ID 在系统内唯一标识，由管理节点创建并管理。
**管理节点（mnode）：** 一个虚拟的逻辑单元，负责所有数据节点运行状态的监控和维护，以及节点之间的负载均衡（图中 M）。同时，管理节点也负责元数据（包括用户、数据库、表、静态标签等）的存储和管理，因此也称为 Meta Node。TDengine 集群中可配置多个（开源版最多不超过 3 个）mnode，它们自动构建成为一个虚拟管理节点组（图中 M0，M1，M2）。mnode 间采用 master/slave 的机制进行管理，而且采取强一致方式进行数据同步，任何数据更新操作只能在 Master 上进行。mnode 集群的创建由系统自动完成，无需人工干预。每个 dnode 上至多有一个 mnode，由所属的数据节点的 EP 来唯一标识。每个 dnode 通过内部消息交互自动获取整个集群中所有 mnode 所在的 dnode 的 EP。
**管理节点（mnode）：** 一个虚拟的逻辑单元，负责所有数据节点运行状态的监控和维护，以及节点之间的负载均衡（图中 M）。同时，管理节点也负责元数据（包括用户、数据库、表、静态标签等）的存储和管理，因此也称为 Meta Node。TDengine 集群中可配置多个（最多不超过 3 个）mnode，它们自动构建成为一个虚拟管理节点组（图中 M0，M1，M2）。mnode 间采用 master/slave 的机制进行管理，而且采取强一致方式进行数据同步，任何数据更新操作只能在 Master 上进行。mnode 集群的创建由系统自动完成，无需人工干预。每个 dnode 上至多有一个 mnode，由所属的数据节点的 EP 来唯一标识。每个 dnode 通过内部消息交互自动获取整个集群中所有 mnode 所在的 dnode 的 EP。
**虚拟节点组（VGroup）：** 不同数据节点上的 vnode 可以组成一个虚拟节点组（vgroup）来保证系统的高可靠。虚拟节点组内采取 master/slave 的方式进行管理。写操作只能在 master vnode 上进行，系统采用异步复制的方式将数据同步到 slave vnode，这样确保了一份数据在多个物理节点上有拷贝。一个 vgroup 里虚拟节点个数就是数据的副本数。如果一个 DB 的副本数为 N，系统必须有至少 N 数据节点。副本数在创建 DB 时通过参数 replica 可以指定，缺省为 1。使用 TDengine 的多副本特性，可以不再需要昂贵的磁盘阵列等存储设备，就可以获得同样的数据高可靠性。虚拟节点组由管理节点创建、管理，并且由管理节点分配一个系统唯一的 ID，VGroup ID。如果两个虚拟节点的 VGroup ID 相同，说明他们属于同一个组，数据互为备份。虚拟节点组里虚拟节点的个数是可以动态改变的，容许只有一个，也就是没有数据复制。VGroup ID 是永远不变的，即使一个虚拟节点组被删除，它的 ID 也不会被收回重复利用。

1
include/client/taos.h
View File

@ -128,6 +128,7 @@ typedef struct setConfRet {
DLL_EXPORT void taos_cleanup(void);
DLL_EXPORT int taos_options(TSDB_OPTION option, const void *arg, ...);
DLL_EXPORT setConfRet taos_set_config(const char *config);
DLL_EXPORT int taos_init(void);
DLL_EXPORT TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port);
DLL_EXPORT TAOS *taos_connect_auth(const char *ip, const char *user, const char *auth, const char *db, uint16_t port);

6
include/common/tcommon.h
View File

@ -42,11 +42,13 @@ enum {
typedef enum EStreamType {
STREAM_NORMAL = 1,
STREAM_INVERT,
STREAM_REPROCESS,
STREAM_CLEAR,
STREAM_INVALID,
STREAM_GET_ALL,
STREAM_DELETE,
STREAM_RETRIEVE,
STREAM_PUSH_DATA,
STREAM_PUSH_EMPTY,
} EStreamType;
typedef struct {
@ -71,7 +73,7 @@ typedef struct SColumnDataAgg {
typedef struct SDataBlockInfo {
STimeWindow window;
int32_t rows; // todo hide this attribute
int32_t rows; // todo hide this attribute
int32_t rowSize;
uint64_t uid; // the uid of table, from which current data block comes
uint16_t blockId; // block id, generated by physical planner

5
include/common/tdatablock.h
View File

@ -225,6 +225,7 @@ int32_t blockDataTrimFirstNRows(SSDataBlock* pBlock, size_t n);
int32_t blockDataKeepFirstNRows(SSDataBlock* pBlock, size_t n);
int32_t assignOneDataBlock(SSDataBlock* dst, const SSDataBlock* src);
int32_t copyDataBlock(SSDataBlock* dst, const SSDataBlock* src);
SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData);
SSDataBlock* createDataBlock();
int32_t blockDataAppendColInfo(SSDataBlock* pBlock, SColumnInfoData* pColInfoData);
@ -236,7 +237,9 @@ void blockCompressEncode(const SSDataBlock* pBlock, char* data, int32_t*
int8_t needCompress);
const char* blockCompressDecode(SSDataBlock* pBlock, int32_t numOfCols, int32_t numOfRows, const char* pData);
void blockDebugShowData(const SArray* dataBlocks, const char* flag);
void blockDebugShowDataBlocks(const SArray* dataBlocks, const char* flag);
// for debug
char* dumpBlockData(SSDataBlock* pDataBlock, const char* flag, char** dumpBuf);
int32_t buildSubmitReqFromDataBlock(SSubmitReq** pReq, const SArray* pDataBlocks, STSchema* pTSchema, int32_t vgId,
tb_uid_t suid);

1
include/common/tdataformat.h
View File

@ -78,7 +78,6 @@ int32_t tEncodeTag(SEncoder *pEncoder, const STag *pTag);
int32_t tDecodeTag(SDecoder *pDecoder, STag **ppTag);
int32_t tTagToValArray(const STag *pTag, SArray **ppArray);
void debugPrintSTag(STag *pTag, const char *tag, int32_t ln); // TODO: remove
void debugCheckTags(STag *pTag); // TODO: remove
// STRUCT =================
struct STColumn {

2
include/common/tmsg.h
View File

@ -1886,7 +1886,7 @@ typedef struct SVCreateStbReq {
int8_t rollup;
SSchemaWrapper schemaRow;
SSchemaWrapper schemaTag;
SRSmaParam pRSmaParam;
SRSmaParam rsmaParam;
} SVCreateStbReq;
int tEncodeSVCreateStbReq(SEncoder* pCoder, const SVCreateStbReq* pReq);

14
include/libs/nodes/plannodes.h
View File

@ -24,6 +24,8 @@ extern "C" {
#include "querynodes.h"
#include "tname.h"
#define SLOT_NAME_LEN TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN
typedef struct SLogicNode {
ENodeType type;
SNodeList* pTargets; // SColumnNode
@ -74,8 +76,8 @@ typedef struct SScanLogicNode {
int16_t tsColId;
double filesFactor;
SArray* pSmaIndexes;
SNodeList* pPartTags;
bool partSort;
SNodeList* pGroupTags;
bool groupSort;
} SScanLogicNode;
typedef struct SJoinLogicNode {
@ -100,6 +102,7 @@ typedef struct SProjectLogicNode {
typedef struct SIndefRowsFuncLogicNode {
SLogicNode node;
SNodeList* pFuncs;
bool isTailFunc;
} SIndefRowsFuncLogicNode;
typedef struct SInterpFuncLogicNode {
@ -138,6 +141,7 @@ typedef struct SMergeLogicNode {
SNodeList* pInputs;
int32_t numOfChannels;
int32_t srcGroupId;
bool groupSort;
} SMergeLogicNode;
typedef enum EWindowType { WINDOW_TYPE_INTERVAL = 1, WINDOW_TYPE_SESSION, WINDOW_TYPE_STATE } EWindowType;
@ -184,6 +188,7 @@ typedef struct SFillLogicNode {
typedef struct SSortLogicNode {
SLogicNode node;
SNodeList* pSortKeys;
bool groupSort;
} SSortLogicNode;
typedef struct SPartitionLogicNode {
@ -230,6 +235,7 @@ typedef struct SSlotDescNode {
bool reserve;
bool output;
bool tag;
char name[SLOT_NAME_LEN];
} SSlotDescNode;
typedef struct SDataBlockDescNode {
@ -279,7 +285,8 @@ typedef struct STableScanPhysiNode {
double ratio;
int32_t dataRequired;
SNodeList* pDynamicScanFuncs;
SNodeList* pPartitionTags;
SNodeList* pGroupTags;
bool groupSort;
int64_t interval;
int64_t offset;
int64_t sliding;
@ -353,6 +360,7 @@ typedef struct SMergePhysiNode {
SNodeList* pTargets;
int32_t numOfChannels;
int32_t srcGroupId;
bool groupSort;
} SMergePhysiNode;
typedef struct SWinodwPhysiNode {

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

@ -259,6 +259,7 @@ typedef struct SSelectStmt {
bool hasTailFunc;
bool hasInterpFunc;
bool hasLastRowFunc;
bool groupSort;
} SSelectStmt;
typedef enum ESetOperatorType { SET_OP_TYPE_UNION_ALL = 1, SET_OP_TYPE_UNION } ESetOperatorType;

40
include/libs/qcom/query.h
View File

@ -16,6 +16,7 @@
#ifndef _TD_QUERY_H_
#define _TD_QUERY_H_
// clang-foramt off
#ifdef __cplusplus
extern "C" {
#endif
@ -71,7 +72,7 @@ typedef struct SIndexMeta {
} SIndexMeta;
typedef struct STbVerInfo {
char tbFName[TSDB_TABLE_FNAME_LEN];
char tbFName[TSDB_TABLE_FNAME_LEN];
int32_t sversion;
int32_t tversion;
} STbVerInfo;
@ -141,7 +142,7 @@ typedef struct SDataBuf {
typedef struct STargetInfo {
ETargetType type;
char* dbFName; // used to update db's vgroup epset
char* dbFName; // used to update db's vgroup epset
int32_t vgId;
} STargetInfo;
@ -149,15 +150,15 @@ typedef int32_t (*__async_send_cb_fn_t)(void* param, const SDataBuf* pMsg, int32
typedef int32_t (*__async_exec_fn_t)(void* param);
typedef struct SRequestConnInfo {
void* pTrans;
uint64_t requestId;
int64_t requestObjRefId;
SEpSet mgmtEps;
void* pTrans;
uint64_t requestId;
int64_t requestObjRefId;
SEpSet mgmtEps;
} SRequestConnInfo;
typedef struct SMsgSendInfo {
__async_send_cb_fn_t fp; // async callback function
STargetInfo target; // for update epset
__async_send_cb_fn_t fp; // async callback function
STargetInfo target; // for update epset
void* param;
uint64_t requestId;
uint64_t requestObjRefId;
@ -206,13 +207,15 @@ int32_t queryCreateTableMetaFromMsg(STableMetaRsp* msg, bool isSuperTable, STabl
char* jobTaskStatusStr(int32_t status);
SSchema createSchema(int8_t type, int32_t bytes, col_id_t colId, const char* name);
void destroyQueryExecRes(SQueryExecRes* pRes);
int32_t dataConverToStr(char *str, int type, void *buf, int32_t bufSize, int32_t *len);
char* parseTagDatatoJson(void* p);
void destroyQueryExecRes(SQueryExecRes* pRes);
int32_t dataConverToStr(char* str, int type, void* buf, int32_t bufSize, int32_t* len);
char* parseTagDatatoJson(void* p);
int32_t cloneTableMeta(STableMeta* pSrc, STableMeta** pDst);
int32_t cloneDbVgInfo(SDBVgInfo* pSrc, SDBVgInfo** pDst);
extern int32_t (*queryBuildMsg[TDMT_MAX])(void *input, char **msg, int32_t msgSize, int32_t *msgLen, void*(*mallocFp)(int32_t));
extern int32_t (*queryBuildMsg[TDMT_MAX])(void* input, char** msg, int32_t msgSize, int32_t* msgLen,
void* (*mallocFp)(int32_t));
extern int32_t (*queryProcessMsgRsp[TDMT_MAX])(void* output, char* msg, int32_t msgSize);
#define SET_META_TYPE_NULL(t) (t) = META_TYPE_NULL_TABLE
@ -223,7 +226,7 @@ extern int32_t (*queryProcessMsgRsp[TDMT_MAX])(void* output, char* msg, int32_t
#define NEED_CLIENT_RM_TBLMETA_ERROR(_code) \
((_code) == TSDB_CODE_PAR_TABLE_NOT_EXIST || (_code) == TSDB_CODE_VND_TB_NOT_EXIST || \
(_code) == TSDB_CODE_PAR_INVALID_COLUMNS_NUM || (_code) == TSDB_CODE_PAR_INVALID_COLUMN || \
(_code) == TSDB_CODE_PAR_TAGS_NOT_MATCHED || (_code) == TSDB_CODE_PAR_VALUE_TOO_LONG || \
(_code) == TSDB_CODE_PAR_TAGS_NOT_MATCHED || (_code) == TSDB_CODE_PAR_VALUE_TOO_LONG || \
(_code) == TSDB_CODE_PAR_INVALID_DROP_COL || ((_code) == TSDB_CODE_TDB_INVALID_TABLE_ID))
#define NEED_CLIENT_REFRESH_VG_ERROR(_code) \
((_code) == TSDB_CODE_VND_HASH_MISMATCH || (_code) == TSDB_CODE_VND_INVALID_VGROUP_ID)
@ -231,11 +234,13 @@ extern int32_t (*queryProcessMsgRsp[TDMT_MAX])(void* output, char* msg, int32_t
#define NEED_CLIENT_HANDLE_ERROR(_code) \
(NEED_CLIENT_RM_TBLMETA_ERROR(_code) || NEED_CLIENT_REFRESH_VG_ERROR(_code) || \
NEED_CLIENT_REFRESH_TBLMETA_ERROR(_code))
#define NEED_CLIENT_RM_TBLMETA_REQ(_type) ((_type) == TDMT_VND_CREATE_TABLE || (_type) == TDMT_VND_CREATE_STB \
|| (_type) == TDMT_VND_DROP_TABLE || (_type) == TDMT_VND_DROP_STB)
#define NEED_CLIENT_RM_TBLMETA_REQ(_type) \
((_type) == TDMT_VND_CREATE_TABLE || (_type) == TDMT_VND_CREATE_STB || (_type) == TDMT_VND_DROP_TABLE || \
(_type) == TDMT_VND_DROP_STB)
#define NEED_SCHEDULER_RETRY_ERROR(_code) \
((_code) == TSDB_CODE_RPC_REDIRECT || (_code) == TSDB_CODE_RPC_NETWORK_UNAVAIL || (_code) == TSDB_CODE_SCH_TIMEOUT_ERROR)
#define NEED_SCHEDULER_RETRY_ERROR(_code) \
((_code) == TSDB_CODE_RPC_REDIRECT || (_code) == TSDB_CODE_RPC_NETWORK_UNAVAIL || \
(_code) == TSDB_CODE_SCH_TIMEOUT_ERROR)
#define REQUEST_TOTAL_EXEC_TIMES 2
@ -312,3 +317,4 @@ extern int32_t (*queryProcessMsgRsp[TDMT_MAX])(void* output, char* msg, int32_t
#endif
#endif /*_TD_QUERY_H_*/
// clang-foramt on

3
include/libs/stream/tstream.h
View File

@ -58,6 +58,7 @@ enum {
enum {
STREAM_INPUT__DATA_SUBMIT = 1,
STREAM_INPUT__DATA_BLOCK,
STREAM_INPUT__DATA_RETRIEVE,
STREAM_INPUT__TRIGGER,
STREAM_INPUT__CHECKPOINT,
STREAM_INPUT__DROP,
@ -318,7 +319,7 @@ static FORCE_INLINE int32_t streamTaskInput(SStreamTask* pTask, SStreamQueueItem
return -1;
}
taosWriteQitem(pTask->inputQueue->queue, pSubmitClone);
} else if (pItem->type == STREAM_INPUT__DATA_BLOCK) {
} else if (pItem->type == STREAM_INPUT__DATA_BLOCK || pItem->type == STREAM_INPUT__DATA_RETRIEVE) {
taosWriteQitem(pTask->inputQueue->queue, pItem);
} else if (pItem->type == STREAM_INPUT__CHECKPOINT) {
taosWriteQitem(pTask->inputQueue->queue, pItem);

3
include/libs/stream/tstreamUpdate.h
View File

@ -32,12 +32,15 @@ typedef struct SUpdateInfo {
int64_t interval;
int64_t watermark;
TSKEY minTS;
SScalableBf* pCloseWinSBF;
} SUpdateInfo;
SUpdateInfo *updateInfoInitP(SInterval* pInterval, int64_t watermark);
SUpdateInfo *updateInfoInit(int64_t interval, int32_t precision, int64_t watermark);
bool updateInfoIsUpdated(SUpdateInfo *pInfo, tb_uid_t tableId, TSKEY ts);
void updateInfoDestroy(SUpdateInfo *pInfo);
void updateInfoAddCloseWindowSBF(SUpdateInfo *pInfo);
void updateInfoDestoryColseWinSBF(SUpdateInfo *pInfo);
#ifdef __cplusplus
}

4
include/libs/sync/sync.h
View File

@ -26,9 +26,9 @@ extern "C" {
extern bool gRaftDetailLog;
#define SYNC_INDEX_BEGIN 0
#define SYNC_INDEX_BEGIN 0
#define SYNC_INDEX_INVALID -1
#define SYNC_TERM_INVALID 0xFFFFFFFFFFFFFFFF
#define SYNC_TERM_INVALID 0xFFFFFFFFFFFFFFFF
typedef uint64_t SyncNodeId;
typedef int32_t SyncGroupId;

4
include/libs/sync/syncTools.h
View File

@ -524,7 +524,7 @@ void syncReconfigFinishLog2(char* s, const SyncReconfigFinish* pMsg);
int32_t syncNodeOnPingCb(SSyncNode* ths, SyncPing* pMsg);
int32_t syncNodeOnPingReplyCb(SSyncNode* ths, SyncPingReply* pMsg);
int32_t syncNodeOnTimeoutCb(SSyncNode* ths, SyncTimeout* pMsg);
int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg);
int32_t syncNodeOnClientRequestCb(SSyncNode* ths, SyncClientRequest* pMsg, SyncIndex* pRetIndex);
int32_t syncNodeOnRequestVoteCb(SSyncNode* ths, SyncRequestVote* pMsg);
int32_t syncNodeOnRequestVoteReplyCb(SSyncNode* ths, SyncRequestVoteReply* pMsg);
int32_t syncNodeOnAppendEntriesCb(SSyncNode* ths, SyncAppendEntries* pMsg);
@ -541,7 +541,7 @@ int32_t syncNodeOnSnapshotRspCb(SSyncNode* ths, SyncSnapshotRsp* pMsg);
// -----------------------------------------
typedef int32_t (*FpOnPingCb)(SSyncNode* ths, SyncPing* pMsg);
typedef int32_t (*FpOnPingReplyCb)(SSyncNode* ths, SyncPingReply* pMsg);
typedef int32_t (*FpOnClientRequestCb)(SSyncNode* ths, SyncClientRequest* pMsg);
typedef int32_t (*FpOnClientRequestCb)(SSyncNode* ths, SyncClientRequest* pMsg, SyncIndex* pRetIndex);
typedef int32_t (*FpOnRequestVoteCb)(SSyncNode* ths, SyncRequestVote* pMsg);
typedef int32_t (*FpOnRequestVoteReplyCb)(SSyncNode* ths, SyncRequestVoteReply* pMsg);
typedef int32_t (*FpOnAppendEntriesCb)(SSyncNode* ths, SyncAppendEntries* pMsg);

6
include/libs/transport/trpc.h
View File

@ -34,10 +34,8 @@ extern int32_t tsRpcHeadSize;
typedef struct {
uint32_t clientIp;
uint16_t clientPort;
union {
char user[TSDB_USER_LEN];
int64_t applyIndex;
};
int64_t applyIndex;
char user[TSDB_USER_LEN];
} SRpcConnInfo;
typedef struct SRpcHandleInfo {

5
include/os/osSocket.h
View File

@ -157,7 +157,10 @@ int32_t taosNonblockwrite(TdSocketPtr pSocket, char *ptr, int32_t nbytes);
int64_t taosCopyFds(TdSocketPtr pSrcSocket, TdSocketPtr pDestSocket, int64_t len);
void taosWinSocketInit();
int taosCreateSocketWithTimeOutOpt(uint32_t conn_timeout_sec);
/*
* set timeout(ms)
*/
int32_t taosCreateSocketWithTimeout(uint32_t timeout);
TdSocketPtr taosOpenUdpSocket(uint32_t localIp, uint16_t localPort);
TdSocketPtr taosOpenTcpClientSocket(uint32_t ip, uint16_t port, uint32_t localIp);

8
include/util/ttrace.h
View File

@ -45,9 +45,11 @@ typedef struct STraceId {
#define TRACE_GET_MSGID(traceId) (traceId)->msgId
#define TRACE_TO_STR(traceId, buf) \
do { \
sprintf(buf, "0x%" PRIx64 ":0x%" PRIx64 "", traceId->rootId, traceId->msgId); \
#define TRACE_TO_STR(traceId, buf) \
do { \
int64_t rootId = (traceId) != NULL ? (traceId)->rootId : 0; \
int64_t msgId = (traceId) != NULL ? (traceId)->msgId : 0; \
sprintf(buf, "0x%" PRIx64 ":0x%" PRIx64 "", rootId, msgId); \
} while (0)
#ifdef __cplusplus

34
source/client/src/clientEnv.c
View File

@ -13,11 +13,11 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "os.h"
#include "catalog.h"
#include "functionMgt.h"
#include "clientInt.h"
#include "clientLog.h"
#include "functionMgt.h"
#include "os.h"
#include "query.h"
#include "scheduler.h"
#include "tcache.h"
@ -38,7 +38,7 @@ static TdThreadOnce tscinit = PTHREAD_ONCE_INIT;
volatile int32_t tscInitRes = 0;
static void registerRequest(SRequestObj *pRequest) {
STscObj *pTscObj = acquireTscObj(*(int64_t*)pRequest->pTscObj->id);
STscObj *pTscObj = acquireTscObj(*(int64_t *)pRequest->pTscObj->id);
assert(pTscObj != NULL);
@ -54,14 +54,14 @@ static void registerRequest(SRequestObj *pRequest) {
int32_t currentInst = atomic_add_fetch_64((int64_t *)&pSummary->currentRequests, 1);
tscDebug("0x%" PRIx64 " new Request from connObj:0x%" PRIx64
", current:%d, app current:%d, total:%d, reqId:0x%" PRIx64,
pRequest->self, *(int64_t*)pRequest->pTscObj->id, num, currentInst, total, pRequest->requestId);
pRequest->self, *(int64_t *)pRequest->pTscObj->id, num, currentInst, total, pRequest->requestId);
}
}
static void deregisterRequest(SRequestObj *pRequest) {
assert(pRequest != NULL);
STscObj *pTscObj = pRequest->pTscObj;
STscObj * pTscObj = pRequest->pTscObj;
SAppClusterSummary *pActivity = &pTscObj->pAppInfo->summary;
int32_t currentInst = atomic_sub_fetch_64((int64_t *)&pActivity->currentRequests, 1);
@ -70,8 +70,8 @@ static void deregisterRequest(SRequestObj *pRequest) {
int64_t duration = taosGetTimestampUs() - pRequest->metric.start;
tscDebug("0x%" PRIx64 " free Request from connObj: 0x%" PRIx64 ", reqId:0x%" PRIx64 " elapsed:%" PRIu64
" ms, current:%d, app current:%d",
pRequest->self, *(int64_t*)pTscObj->id, pRequest->requestId, duration / 1000, num, currentInst);
releaseTscObj(*(int64_t*)pTscObj->id);
pRequest->self, *(int64_t *)pTscObj->id, pRequest->requestId, duration / 1000, num, currentInst);
releaseTscObj(*(int64_t *)pTscObj->id);
}
// todo close the transporter properly
@ -80,12 +80,13 @@ void closeTransporter(STscObj *pTscObj) {
return;
}
tscDebug("free transporter:%p in connObj: 0x%" PRIx64, pTscObj->pAppInfo->pTransporter, *(int64_t*)pTscObj->id);
tscDebug("free transporter:%p in connObj: 0x%" PRIx64, pTscObj->pAppInfo->pTransporter, *(int64_t *)pTscObj->id);
rpcClose(pTscObj->pAppInfo->pTransporter);
}
static bool clientRpcRfp(int32_t code) {
if (code == TSDB_CODE_RPC_REDIRECT) {
if (code == TSDB_CODE_RPC_REDIRECT || code == TSDB_CODE_RPC_NETWORK_UNAVAIL || code == TSDB_CODE_NODE_NOT_DEPLOYED ||
code == TSDB_CODE_SYN_NOT_LEADER || code == TSDB_CODE_APP_NOT_READY) {
return true;
} else {
return false;
@ -128,16 +129,17 @@ void closeAllRequests(SHashObj *pRequests) {
void destroyTscObj(void *pObj) {
STscObj *pTscObj = pObj;
SClientHbKey connKey = {.tscRid = *(int64_t*)pTscObj->id, .connType = pTscObj->connType};
SClientHbKey connKey = {.tscRid = *(int64_t *)pTscObj->id, .connType = pTscObj->connType};
hbDeregisterConn(pTscObj->pAppInfo->pAppHbMgr, connKey);
int64_t connNum = atomic_sub_fetch_64(&pTscObj->pAppInfo->numOfConns, 1);
closeAllRequests(pTscObj->pRequests);
schedulerStopQueryHb(pTscObj->pAppInfo->pTransporter);
if (0 == connNum) {
// TODO
//closeTransporter(pTscObj);
// TODO
// closeTransporter(pTscObj);
}
tscDebug("connObj 0x%" PRIx64 " destroyed, totalConn:%" PRId64, *(int64_t*)pTscObj->id, pTscObj->pAppInfo->numOfConns);
tscDebug("connObj 0x%" PRIx64 " destroyed, totalConn:%" PRId64, *(int64_t *)pTscObj->id,
pTscObj->pAppInfo->numOfConns);
taosThreadMutexDestroy(&pTscObj->mutex);
taosMemoryFreeClear(pTscObj);
}
@ -167,10 +169,10 @@ void *createTscObj(const char *user, const char *auth, const char *db, int32_t c
taosThreadMutexInit(&pObj->mutex, NULL);
pObj->id = taosMemoryMalloc(sizeof(int64_t));
*(int64_t*)pObj->id = taosAddRef(clientConnRefPool, pObj);
*(int64_t *)pObj->id = taosAddRef(clientConnRefPool, pObj);
pObj->schemalessType = 1;
tscDebug("connObj created, 0x%" PRIx64, *(int64_t*)pObj->id);
tscDebug("connObj created, 0x%" PRIx64, *(int64_t *)pObj->id);
return pObj;
}
@ -325,7 +327,7 @@ int taos_options_imp(TSDB_OPTION option, const char *str) {
return 0;
}
SConfig *pCfg = taosGetCfg();
SConfig * pCfg = taosGetCfg();
SConfigItem *pItem = NULL;
switch (option) {

44
source/client/src/clientMain.c
View File

@ -81,6 +81,19 @@ void taos_cleanup(void) {
taosCloseLog();
}
static setConfRet taos_set_config_imp(const char *config){
setConfRet ret = {SET_CONF_RET_SUCC, {0}};
// TODO: need re-implementation
return ret;
}
setConfRet taos_set_config(const char *config){
// TODO pthread_mutex_lock(&setConfMutex);
setConfRet ret = taos_set_config_imp(config);
// pthread_mutex_unlock(&setConfMutex);
return ret;
}
TAOS *taos_connect(const char *ip, const char *user, const char *pass, const char *db, uint16_t port) {
tscDebug("try to connect to %s:%u, user:%s db:%s", ip, port, user, db);
if (user == NULL) {
@ -199,10 +212,7 @@ TAOS_FIELD *taos_fetch_fields(TAOS_RES *res) {
return pResInfo->userFields;
}
TAOS_RES *taos_query(TAOS *taos, const char *sql) {
return taosQueryImpl(taos, sql, false);
}
TAOS_RES *taos_query(TAOS *taos, const char *sql) { return taosQueryImpl(taos, sql, false); }
TAOS_ROW taos_fetch_row(TAOS_RES *res) {
if (res == NULL) {
@ -593,11 +603,11 @@ int *taos_get_column_data_offset(TAOS_RES *res, int columnIndex) {
return pResInfo->pCol[columnIndex].offset;
}
int taos_validate_sql(TAOS *taos, const char *sql) {
TAOS_RES* pObj = taosQueryImpl(taos, sql, true);
int taos_validate_sql(TAOS *taos, const char *sql) {
TAOS_RES *pObj = taosQueryImpl(taos, sql, true);
int code = taos_errno(pObj);
taos_free_result(pObj);
return code;
}
@ -884,10 +894,10 @@ void taos_unsubscribe(TAOS_SUB *tsub, int keepProgress) {
int taos_load_table_info(TAOS *taos, const char *tableNameList) {
const int32_t MAX_TABLE_NAME_LENGTH = 12 * 1024 * 1024; // 12MB list
int32_t code = 0;
SRequestObj *pRequest = NULL;
SCatalogReq catalogReq = {0};
int32_t code = 0;
SRequestObj *pRequest = NULL;
SCatalogReq catalogReq = {0};
if (NULL == tableNameList) {
return TSDB_CODE_SUCCESS;
}
@ -911,26 +921,25 @@ int taos_load_table_info(TAOS *taos, const char *tableNameList) {
goto _return;
}
SCatalog* pCtg = NULL;
SCatalog *pCtg = NULL;
code = catalogGetHandle(pTscObj->pAppInfo->clusterId, &pCtg);
if (code != TSDB_CODE_SUCCESS) {
goto _return;
}
char* sql = "taos_load_table_info";
char *sql = "taos_load_table_info";
code = buildRequest(pTscObj, sql, strlen(sql), &pRequest);
if (code != TSDB_CODE_SUCCESS) {
terrno = code;
goto _return;
}
SSyncQueryParam param = {0};
tsem_init(&param.sem, 0, 0);
param.pRequest = pRequest;
SRequestConnInfo conn = {.pTrans = pTscObj->pAppInfo->pTransporter,
.requestId = pRequest->requestId,
.requestObjRefId = pRequest->self};
SRequestConnInfo conn = {
.pTrans = pTscObj->pAppInfo->pTransporter, .requestId = pRequest->requestId, .requestObjRefId = pRequest->self};
conn.mgmtEps = getEpSet_s(&pTscObj->pAppInfo->mgmtEp);
@ -951,7 +960,6 @@ _return:
return code;
}
TAOS_STMT *taos_stmt_init(TAOS *taos) {
STscObj *pObj = acquireTscObj(*(int64_t *)taos);
if (NULL == pObj) {

83
source/common/src/tdatablock.c
View File

@ -1170,7 +1170,7 @@ int32_t colInfoDataEnsureCapacity(SColumnInfoData* pColumn, uint32_t numOfRows)
int32_t blockDataEnsureCapacity(SSDataBlock* pDataBlock, uint32_t numOfRows) {
int32_t code = 0;
ASSERT(numOfRows > 0);
// ASSERT(numOfRows > 0);
if (numOfRows == 0) {
return TSDB_CODE_SUCCESS;
@ -1236,6 +1236,32 @@ int32_t assignOneDataBlock(SSDataBlock* dst, const SSDataBlock* src) {
return 0;
}
int32_t copyDataBlock(SSDataBlock* dst, const SSDataBlock* src) {
ASSERT(src != NULL && dst != NULL);
blockDataCleanup(dst);
int32_t code = blockDataEnsureCapacity(dst, src->info.rows);
if (code != TSDB_CODE_SUCCESS) {
terrno = code;
return code;
}
size_t numOfCols = taosArrayGetSize(src->pDataBlock);
for (int32_t i = 0; i < numOfCols; ++i) {
SColumnInfoData* pDst = taosArrayGet(dst->pDataBlock, i);
SColumnInfoData* pSrc = taosArrayGet(src->pDataBlock, i);
if (pSrc->pData == NULL) {
continue;
}
colDataAssign(pDst, pSrc, src->info.rows, &src->info);
}
dst->info.rows = src->info.rows;
dst->info.window = src->info.window;
return TSDB_CODE_SUCCESS;
}
SSDataBlock* createOneDataBlock(const SSDataBlock* pDataBlock, bool copyData) {
if (pDataBlock == NULL) {
return NULL;
@ -1613,7 +1639,7 @@ static char* formatTimestamp(char* buf, int64_t val, int precision) {
return buf;
}
void blockDebugShowData(const SArray* dataBlocks, const char* flag) {
void blockDebugShowDataBlocks(const SArray* dataBlocks, const char* flag) {
char pBuf[128] = {0};
int32_t sz = taosArrayGetSize(dataBlocks);
for (int32_t i = 0; i < sz; i++) {
@ -1621,7 +1647,7 @@ void blockDebugShowData(const SArray* dataBlocks, const char* flag) {
size_t numOfCols = taosArrayGetSize(pDataBlock->pDataBlock);
int32_t rows = pDataBlock->info.rows;
printf("%s |block type %d |child id %d|\n", flag, (int32_t)pDataBlock->info.type, pDataBlock->info.childId);
printf("%s |block type %d |child id %d|group id %zX\n", flag, (int32_t)pDataBlock->info.type, pDataBlock->info.childId, pDataBlock->info.groupId);
for (int32_t j = 0; j < rows; j++) {
printf("%s |", flag);
for (int32_t k = 0; k < numOfCols; k++) {
@ -1661,6 +1687,57 @@ void blockDebugShowData(const SArray* dataBlocks, const char* flag) {
}
}
// for debug
char* dumpBlockData(SSDataBlock* pDataBlock, const char* flag, char** pDataBuf) {
int32_t size = 2048;
*pDataBuf = taosMemoryCalloc(size, 1);
char* dumpBuf = *pDataBuf;
char pBuf[128] = {0};
int32_t colNum = taosArrayGetSize(pDataBlock->pDataBlock);
int32_t rows = pDataBlock->info.rows;
int32_t len = 0;
len += snprintf(dumpBuf + len, size - len, "\n%s |block type %d |child id %d|\n", flag,
(int32_t)pDataBlock->info.type, pDataBlock->info.childId);
for (int32_t j = 0; j < rows; j++) {
len += snprintf(dumpBuf + len, size - len, "%s |", flag);
for (int32_t k = 0; k < colNum; k++) {
SColumnInfoData* pColInfoData = taosArrayGet(pDataBlock->pDataBlock, k);
void* var = POINTER_SHIFT(pColInfoData->pData, j * pColInfoData->info.bytes);
if (colDataIsNull(pColInfoData, rows, j, NULL)) {
len += snprintf(dumpBuf + len, size - len, " %15s |", "NULL");
continue;
}
switch (pColInfoData->info.type) {
case TSDB_DATA_TYPE_TIMESTAMP:
formatTimestamp(pBuf, *(uint64_t*)var, TSDB_TIME_PRECISION_MILLI);
len += snprintf(dumpBuf + len, size - len, " %25s |", pBuf);
break;
case TSDB_DATA_TYPE_INT:
len += snprintf(dumpBuf + len, size - len, " %15d |", *(int32_t*)var);
break;
case TSDB_DATA_TYPE_UINT:
len += snprintf(dumpBuf + len, size - len, " %15u |", *(uint32_t*)var);
break;
case TSDB_DATA_TYPE_BIGINT:
len += snprintf(dumpBuf + len, size - len, " %15ld |", *(int64_t*)var);
break;
case TSDB_DATA_TYPE_UBIGINT:
len += snprintf(dumpBuf + len, size - len, " %15lu |", *(uint64_t*)var);
break;
case TSDB_DATA_TYPE_FLOAT:
len += snprintf(dumpBuf + len, size - len, " %15f |", *(float*)var);
break;
case TSDB_DATA_TYPE_DOUBLE:
len += snprintf(dumpBuf + len, size - len, " %15lf |", *(double*)var);
break;
}
}
len += snprintf(dumpBuf + len, size - len, "\n");
}
len += snprintf(dumpBuf + len, size - len, "%s |end\n", flag);
return dumpBuf;
}
/**
* @brief TODO: Assume that the final generated result it less than 3M
*

16
source/common/src/tdataformat.c
View File

@ -862,21 +862,6 @@ void debugPrintSTag(STag *pTag, const char *tag, int32_t ln) {
printf("\n");
}
void debugCheckTags(STag *pTag) {
switch (pTag->flags) {
case 0x0:
case 0x20:
case 0x40:
case 0x60:
break;
default:
ASSERT(0);
}
ASSERT(pTag->nTag <= 128 && pTag->nTag >= 0);
ASSERT(pTag->ver <= 512 && pTag->ver >= 0); // temp condition for pTag->ver
}
static int32_t tPutTagVal(uint8_t *p, STagVal *pTagVal, int8_t isJson) {
int32_t n = 0;
@ -999,7 +984,6 @@ int32_t tTagNew(SArray *pArray, int32_t version, int8_t isJson, STag **ppTag) {
debugPrintSTag(*ppTag, __func__, __LINE__);
#endif
debugCheckTags(*ppTag); // TODO: remove this line after debug
return code;
_err:

14
source/common/src/tmsg.c
View File

@ -3696,7 +3696,7 @@ int32_t tDeserializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *
if (tDecodeI8(&decoder, &pReq->isTsma) < 0) return -1;
if (pReq->isTsma) {
if (tDecodeBinaryAlloc(&decoder, &pReq->pTsma, NULL) < 0) return -1;
if (tDecodeBinary(&decoder, (uint8_t **)&pReq->pTsma, NULL) < 0) return -1;
}
tEndDecode(&decoder);
@ -3707,9 +3707,6 @@ int32_t tDeserializeSCreateVnodeReq(void *buf, int32_t bufLen, SCreateVnodeReq *
int32_t tFreeSCreateVnodeReq(SCreateVnodeReq *pReq) {
taosArrayDestroy(pReq->pRetensions);
pReq->pRetensions = NULL;
if (pReq->isTsma) {
taosMemoryFreeClear(pReq->pTsma);
}
return 0;
}
@ -4747,9 +4744,8 @@ int32_t tDecodeSRSmaParam(SDecoder *pCoder, SRSmaParam *pRSmaParam) {
if (tDecodeI64v(pCoder, &pRSmaParam->watermark[i]) < 0) return -1;
if (tDecodeI32v(pCoder, &pRSmaParam->qmsgLen[i]) < 0) return -1;
if (pRSmaParam->qmsgLen[i] > 0) {
uint64_t len;
if (tDecodeBinaryAlloc(pCoder, (void **)&pRSmaParam->qmsg[i], &len) < 0)
return -1; // qmsgLen contains len of '\0'
tDecoderMalloc(pCoder, pRSmaParam->qmsgLen[i]);
if (tDecodeBinary(pCoder, (uint8_t **)&pRSmaParam->qmsg[i], NULL) < 0) return -1; // qmsgLen contains len of '\0'
} else {
pRSmaParam->qmsg[i] = NULL;
}
@ -4767,7 +4763,7 @@ int tEncodeSVCreateStbReq(SEncoder *pCoder, const SVCreateStbReq *pReq) {
if (tEncodeSSchemaWrapper(pCoder, &pReq->schemaRow) < 0) return -1;
if (tEncodeSSchemaWrapper(pCoder, &pReq->schemaTag) < 0) return -1;
if (pReq->rollup) {
if (tEncodeSRSmaParam(pCoder, &pReq->pRSmaParam) < 0) return -1;
if (tEncodeSRSmaParam(pCoder, &pReq->rsmaParam) < 0) return -1;
}
tEndEncode(pCoder);
@ -4783,7 +4779,7 @@ int tDecodeSVCreateStbReq(SDecoder *pCoder, SVCreateStbReq *pReq) {
if (tDecodeSSchemaWrapper(pCoder, &pReq->schemaRow) < 0) return -1;
if (tDecodeSSchemaWrapper(pCoder, &pReq->schemaTag) < 0) return -1;
if (pReq->rollup) {
if (tDecodeSRSmaParam(pCoder, &pReq->pRSmaParam) < 0) return -1;
if (tDecodeSRSmaParam(pCoder, &pReq->rsmaParam) < 0) return -1;
}
tEndDecode(pCoder);

12
source/dnode/mgmt/mgmt_dnode/src/dmHandle.c
View File

@ -247,11 +247,9 @@ int32_t dmAppendVariablesToBlock(SSDataBlock* pBlock, int32_t dnodeId) {
return TSDB_CODE_SUCCESS;
}
int32_t dmProcessRetrieve(SDnodeMgmt *pMgmt, SRpcMsg *pMsg) {
int32_t size = 0;
int32_t rowsRead = 0;
int32_t size = 0;
int32_t rowsRead = 0;
SRetrieveTableReq retrieveReq = {0};
if (tDeserializeSRetrieveTableReq(pMsg->pCont, pMsg->contLen, &retrieveReq) != 0) {
@ -259,6 +257,11 @@ int32_t dmProcessRetrieve(SDnodeMgmt *pMgmt, SRpcMsg *pMsg) {
return -1;
}
if (strcmp(retrieveReq.user, TSDB_DEFAULT_USER) != 0) {
terrno = TSDB_CODE_MND_NO_RIGHTS;
return -1;
}
if (strcasecmp(retrieveReq.tb, TSDB_INS_TABLE_DNODE_VARIABLES)) {
terrno = TSDB_CODE_INVALID_MSG;
return -1;
@ -309,7 +312,6 @@ int32_t dmProcessRetrieve(SDnodeMgmt *pMgmt, SRpcMsg *pMsg) {
return TSDB_CODE_SUCCESS;
}
SArray *dmGetMsgHandles() {
int32_t code = -1;
SArray *pArray = taosArrayInit(16, sizeof(SMgmtHandle));

17
source/dnode/mgmt/node_mgmt/src/dmTransport.c
View File

@ -70,9 +70,9 @@ int32_t dmProcessNodeMsg(SMgmtWrapper *pWrapper, SRpcMsg *pMsg) {
}
static void dmProcessRpcMsg(SDnode *pDnode, SRpcMsg *pRpc, SEpSet *pEpSet) {
SDnodeTrans *pTrans = &pDnode->trans;
SDnodeTrans * pTrans = &pDnode->trans;
int32_t code = -1;
SRpcMsg *pMsg = NULL;
SRpcMsg * pMsg = NULL;
SMgmtWrapper *pWrapper = NULL;
SDnodeHandle *pHandle = &pTrans->msgHandles[TMSG_INDEX(pRpc->msgType)];
@ -194,11 +194,11 @@ int32_t dmInitMsgHandle(SDnode *pDnode) {
for (EDndNodeType ntype = DNODE; ntype < NODE_END; ++ntype) {
SMgmtWrapper *pWrapper = &pDnode->wrappers[ntype];
SArray *pArray = (*pWrapper->func.getHandlesFp)();
SArray * pArray = (*pWrapper->func.getHandlesFp)();
if (pArray == NULL) return -1;
for (int32_t i = 0; i < taosArrayGetSize(pArray); ++i) {
SMgmtHandle *pMgmt = taosArrayGet(pArray, i);
SMgmtHandle * pMgmt = taosArrayGet(pArray, i);
SDnodeHandle *pHandle = &pTrans->msgHandles[TMSG_INDEX(pMgmt->msgType)];
if (pMgmt->needCheckVgId) {
pHandle->needCheckVgId = pMgmt->needCheckVgId;
@ -248,7 +248,14 @@ static inline void dmReleaseHandle(SRpcHandleInfo *pHandle, int8_t type) {
}
}
static bool rpcRfp(int32_t code) { return code == TSDB_CODE_RPC_REDIRECT; }
static bool rpcRfp(int32_t code) {
if (code == TSDB_CODE_RPC_REDIRECT || code == TSDB_CODE_RPC_NETWORK_UNAVAIL || code == TSDB_CODE_NODE_NOT_DEPLOYED ||
code == TSDB_CODE_SYN_NOT_LEADER || code == TSDB_CODE_APP_NOT_READY) {
return true;
} else {
return false;
}
}
int32_t dmInitClient(SDnode *pDnode) {
SDnodeTrans *pTrans = &pDnode->trans;

5
source/dnode/mnode/impl/inc/mndPrivilege.h
View File

@ -57,6 +57,8 @@ typedef enum {
MND_OPER_USE_DB,
MND_OPER_WRITE_DB,
MND_OPER_READ_DB,
MND_OPER_READ_OR_WRITE_DB,
MND_OPER_SHOW_VARIBALES,
} EOperType;
int32_t mndInitPrivilege(SMnode *pMnode);
@ -64,7 +66,8 @@ void mndCleanupPrivilege(SMnode *pMnode);
int32_t mndCheckOperPrivilege(SMnode *pMnode, const char *user, EOperType operType);
int32_t mndCheckDbPrivilege(SMnode *pMnode, const char *user, EOperType operType, SDbObj *pDb);
int32_t mndCheckShowPrivilege(SMnode *pMnode, const char *user, int32_t showType);
int32_t mndCheckDbPrivilegeByName(SMnode *pMnode, const char *user, EOperType operType, const char *dbname);
int32_t mndCheckShowPrivilege(SMnode *pMnode, const char *user, EShowType showType, const char *dbname);
int32_t mndCheckAlterUserPrivilege(SUserObj *pOperUser, SUserObj *pUser, SAlterUserReq *pAlter);
#ifdef __cplusplus

4
source/dnode/mnode/impl/inc/mndStream.h
View File

@ -33,6 +33,10 @@ SSdbRow *mndStreamActionDecode(SSdbRaw *pRaw);
int32_t mndDropStreamByDb(SMnode *pMnode, STrans *pTrans, SDbObj *pDb);
int32_t mndPersistStream(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream);
// for sma
// TODO refactor
int32_t mndDropStreamTasks(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream);
int32_t mndPersistDropStreamLog(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream);
#ifdef __cplusplus
}

4
source/dnode/mnode/impl/inc/mndTrans.h
View File

@ -39,8 +39,10 @@ typedef struct {
int32_t id;
int32_t errCode;
int32_t acceptableCode;
ETrnStage stage;
int32_t retryCode;
ETrnAct actionType;
ETrnStage stage;
int8_t reserved;
int8_t rawWritten;
int8_t msgSent;
int8_t msgReceived;

4
source/dnode/mnode/impl/src/mndConsumer.c
View File

@ -431,6 +431,10 @@ static int32_t mndProcessSubscribeReq(SRpcMsg *pMsg) {
goto SUBSCRIBE_OVER;
}
if (mndCheckDbPrivilegeByName(pMnode, pMsg->info.conn.user, MND_OPER_READ_DB, pTopic->db) != 0) {
goto SUBSCRIBE_OVER;
}
#if 0
// ref topic to prevent drop
// TODO make topic complete

27
source/dnode/mnode/impl/src/mndDb.c
View File

@ -15,9 +15,9 @@
#define _DEFAULT_SOURCE
#include "mndDb.h"
#include "mndPrivilege.h"
#include "mndDnode.h"
#include "mndOffset.h"
#include "mndPrivilege.h"
#include "mndShow.h"
#include "mndSma.h"
#include "mndStb.h"
@ -1336,7 +1336,7 @@ char *buildRetension(SArray *pRetension) {
}
static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, int32_t rows, int64_t numOfTables,
bool sysDb, ESdbStatus objStatus) {
bool sysDb, ESdbStatus objStatus, bool sysinfo) {
int32_t cols = 0;
int32_t bytes = pShow->pMeta->pSchemas[cols].bytes;
@ -1354,7 +1354,7 @@ static void dumpDbInfoData(SSDataBlock *pBlock, SDbObj *pDb, SShowObj *pShow, in
char statusB[24] = {0};
STR_WITH_SIZE_TO_VARSTR(statusB, status, strlen(status));
if (sysDb) {
if (sysDb || !sysinfo) {
for (int32_t i = 0; i < pShow->numOfColumns; ++i) {
SColumnInfoData *pColInfo = taosArrayGet(pBlock->pDataBlock, i);
if (i == 0) {
@ -1528,17 +1528,21 @@ static int32_t mndRetrieveDbs(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBloc
SDbObj *pDb = NULL;
ESdbStatus objStatus = 0;
SUserObj *pUser = mndAcquireUser(pMnode, pReq->info.conn.user);
if (pUser == NULL) return 0;
bool sysinfo = pUser->sysInfo;
// Append the information_schema database into the result.
if (!pShow->sysDbRsp) {
SDbObj infoschemaDb = {0};
setInformationSchemaDbCfg(&infoschemaDb);
dumpDbInfoData(pBlock, &infoschemaDb, pShow, numOfRows, 14, true, 0);
dumpDbInfoData(pBlock, &infoschemaDb, pShow, numOfRows, 14, true, 0, 1);
numOfRows += 1;
SDbObj perfschemaDb = {0};
setPerfSchemaDbCfg(&perfschemaDb);
dumpDbInfoData(pBlock, &perfschemaDb, pShow, numOfRows, 3, true, 0);
dumpDbInfoData(pBlock, &perfschemaDb, pShow, numOfRows, 3, true, 0, 1);
numOfRows += 1;
pShow->sysDbRsp = true;
@ -1550,16 +1554,19 @@ static int32_t mndRetrieveDbs(SRpcMsg *pReq, SShowObj *pShow, SSDataBlock *pBloc
break;
}
int32_t numOfTables = 0;
sdbTraverse(pSdb, SDB_VGROUP, mndGetTablesOfDbFp, &numOfTables, NULL, NULL);
if (mndCheckDbPrivilege(pMnode, pReq->info.conn.user, MND_OPER_READ_OR_WRITE_DB, pDb) == 0) {
int32_t numOfTables = 0;
sdbTraverse(pSdb, SDB_VGROUP, mndGetTablesOfDbFp, &numOfTables, NULL, NULL);
dumpDbInfoData(pBlock, pDb, pShow, numOfRows, numOfTables, false, objStatus, sysinfo);
numOfRows++;
}
dumpDbInfoData(pBlock, pDb, pShow, numOfRows, numOfTables, false, objStatus);
numOfRows++;
sdbRelease(pSdb, pDb);
}
pShow->numOfRows += numOfRows;
mndReleaseUser(pMnode, pUser);
return numOfRows;
}

6
source/dnode/mnode/impl/src/mndDnode.c
View File

@ -558,7 +558,11 @@ _OVER:
static int32_t mndProcessShowVariablesReq(SRpcMsg *pReq) {
SShowVariablesRsp rsp = {0};
int32_t code = -1;
int32_t code = -1;
if (mndCheckOperPrivilege(pReq->info.node, pReq->info.conn.user, MND_OPER_SHOW_VARIBALES) != 0) {
goto _OVER;
}
rsp.variables = taosArrayInit(4, sizeof(SVariablesInfo));
if (NULL == rsp.variables) {

4
source/dnode/mnode/impl/src/mndMain.c
View File

@ -442,7 +442,7 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
syncPingReplyDestroy(pSyncMsg);
} else if (pMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) {
SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pMsg);
code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg);
code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL);
syncClientRequestDestroy(pSyncMsg);
} else if (pMsg->msgType == TDMT_SYNC_REQUEST_VOTE) {
SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pMsg);
@ -491,7 +491,7 @@ int32_t mndProcessSyncMsg(SRpcMsg *pMsg) {
syncPingReplyDestroy(pSyncMsg);
} else if (pMsg->msgType == TDMT_SYNC_CLIENT_REQUEST) {
SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pMsg);
code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg);
code = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL);
syncClientRequestDestroy(pSyncMsg);
} else if (pMsg->msgType == TDMT_SYNC_REQUEST_VOTE) {
SyncRequestVote *pSyncMsg = syncRequestVoteFromRpcMsg2(pMsg);

55
source/dnode/mnode/impl/src/mndPrivilege.c
View File

@ -15,6 +15,7 @@
#define _DEFAULT_SOURCE
#include "mndPrivilege.h"
#include "mndDb.h"
#include "mndUser.h"
int32_t mndInitPrivilege(SMnode *pMnode) { return 0; }
@ -45,6 +46,7 @@ int32_t mndCheckOperPrivilege(SMnode *pMnode, const char *user, EOperType operTy
case MND_OPER_CONNECT:
case MND_OPER_CREATE_FUNC:
case MND_OPER_DROP_FUNC:
case MND_OPER_SHOW_VARIBALES:
break;
default:
terrno = TSDB_CODE_MND_NO_RIGHTS;
@ -79,7 +81,7 @@ int32_t mndCheckAlterUserPrivilege(SUserObj *pOperUser, SUserObj *pUser, SAlterU
return -1;
}
int32_t mndCheckShowPrivilege(SMnode *pMnode, const char *user, int32_t showType) {
int32_t mndCheckShowPrivilege(SMnode *pMnode, const char *user, EShowType showType, const char *dbname) {
int32_t code = 0;
SUserObj *pUser = mndAcquireUser(pMnode, user);
@ -98,14 +100,34 @@ int32_t mndCheckShowPrivilege(SMnode *pMnode, const char *user, int32_t showType
goto _OVER;
}
if (!pUser->sysInfo) {
terrno = TSDB_CODE_MND_NO_RIGHTS;
code = -1;
if (pUser->sysInfo) {
goto _OVER;
}
terrno = TSDB_CODE_MND_NO_RIGHTS;
code = -1;
switch (showType) {
case TSDB_MGMT_TABLE_DB:
case TSDB_MGMT_TABLE_STB:
case TSDB_MGMT_TABLE_INDEX:
case TSDB_MGMT_TABLE_STREAMS:
case TSDB_MGMT_TABLE_CONSUMERS:
case TSDB_MGMT_TABLE_TOPICS:
case TSDB_MGMT_TABLE_SUBSCRIPTIONS:
case TSDB_MGMT_TABLE_FUNC:
case TSDB_MGMT_TABLE_QUERIES:
case TSDB_MGMT_TABLE_CONNS:
case TSDB_MGMT_TABLE_APPS:
case TSDB_MGMT_TABLE_TRANS:
code = 0;
break;
default:
terrno = TSDB_CODE_MND_NO_RIGHTS;
code = -1;
goto _OVER;
}
if (showType == TSDB_MGMT_TABLE_STB || showType == TSDB_MGMT_TABLE_VGROUP || showType == TSDB_MGMT_TABLE_INDEX) {
code = mndCheckDbPrivilegeByName(pMnode, user, MND_OPER_READ_OR_WRITE_DB, dbname);
}
_OVER:
mndReleaseUser(pMnode, pUser);
@ -133,19 +155,11 @@ int32_t mndCheckDbPrivilege(SMnode *pMnode, const char *user, EOperType operType
if (pUser->sysInfo) goto _OVER;
}
if (operType == MND_OPER_ALTER_DB) {
if (operType == MND_OPER_ALTER_DB || operType == MND_OPER_DROP_DB || operType == MND_OPER_COMPACT_DB) {
if (strcmp(pUser->user, pDb->createUser) == 0 && pUser->sysInfo) goto _OVER;
}
if (operType == MND_OPER_DROP_DB) {
if (strcmp(pUser->user, pDb->createUser) == 0 && pUser->sysInfo) goto _OVER;
}
if (operType == MND_OPER_COMPACT_DB) {
if (strcmp(pUser->user, pDb->createUser) == 0 && pUser->sysInfo) goto _OVER;
}
if (operType == MND_OPER_USE_DB) {
if (operType == MND_OPER_USE_DB || operType == MND_OPER_READ_OR_WRITE_DB) {
if (strcmp(pUser->user, pDb->createUser) == 0) goto _OVER;
if (taosHashGet(pUser->readDbs, pDb->name, strlen(pDb->name) + 1) != NULL) goto _OVER;
if (taosHashGet(pUser->writeDbs, pDb->name, strlen(pDb->name) + 1) != NULL) goto _OVER;
@ -168,3 +182,12 @@ _OVER:
mndReleaseUser(pMnode, pUser);
return code;
}
int32_t mndCheckDbPrivilegeByName(SMnode *pMnode, const char *user, EOperType operType, const char *dbname) {
SDbObj *pDb = mndAcquireDb(pMnode, dbname);
if (pDb == NULL) return -1;
int32_t code = mndCheckDbPrivilege(pMnode, user, operType, pDb);
mndReleaseDb(pMnode, pDb);
return code;
}

4
source/dnode/mnode/impl/src/mndProfile.c
View File

@ -254,6 +254,10 @@ static int32_t mndProcessConnectReq(SRpcMsg *pReq) {
terrstr());
goto _OVER;
}
if (mndCheckDbPrivilege(pMnode, pReq->info.conn.user, MND_OPER_READ_OR_WRITE_DB, pDb) != 0) {
goto _OVER;
}
}
pConn = mndCreateConn(pMnode, pReq->info.conn.user, connReq.connType, pReq->info.conn.clientIp,

10
source/dnode/mnode/impl/src/mndShow.c
View File

@ -231,8 +231,14 @@ static int32_t mndProcessRetrieveSysTableReq(SRpcMsg *pReq) {
}
mDebug("show:0x%" PRIx64 ", start retrieve data, type:%d", pShow->id, pShow->type);
// if (mndCheckShowPrivilege(pMnode, pReq->info.conn.user, pShow->type) != 0) return -1;
if (retrieveReq.user[0] != 0) {
memcpy(pReq->info.conn.user, retrieveReq.user, TSDB_USER_LEN);
} else {
memcpy(pReq->info.conn.user, TSDB_DEFAULT_USER, strlen(TSDB_DEFAULT_USER) + 1);
}
if (mndCheckShowPrivilege(pMnode, pReq->info.conn.user, pShow->type, retrieveReq.db) != 0) {
return -1;
}
int32_t numOfCols = pShow->pMeta->numOfColumns;
SSDataBlock *pBlock = taosMemoryCalloc(1, sizeof(SSDataBlock));

19
source/dnode/mnode/impl/src/mndSma.c
View File

@ -15,11 +15,11 @@
#define _DEFAULT_SOURCE
#include "mndSma.h"
#include "mndPrivilege.h"
#include "mndDb.h"
#include "mndDnode.h"
#include "mndInfoSchema.h"
#include "mndMnode.h"
#include "mndPrivilege.h"
#include "mndScheduler.h"
#include "mndShow.h"
#include "mndStb.h"
@ -857,6 +857,23 @@ static int32_t mndDropSma(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SSmaObj *p
mDebug("trans:%d, used to drop sma:%s", pTrans->id, pSma->name);
mndTransSetDbName(pTrans, pDb->name, NULL);
SStreamObj *pStream = mndAcquireStream(pMnode, pSma->name);
if (pStream == NULL || pStream->smaId != pSma->uid) {
sdbRelease(pMnode->pSdb, pStream);
goto _OVER;
} else {
if (mndDropStreamTasks(pMnode, pTrans, pStream) < 0) {
mError("stream:%s, failed to drop task since %s", pStream->name, terrstr());
sdbRelease(pMnode->pSdb, pStream);
goto _OVER;
}
// drop stream
if (mndPersistDropStreamLog(pMnode, pTrans, pStream) < 0) {
sdbRelease(pMnode->pSdb, pStream);
goto _OVER;
}
}
if (mndSetDropSmaRedoLogs(pMnode, pTrans, pSma) != 0) goto _OVER;
if (mndSetDropSmaVgroupRedoLogs(pMnode, pTrans, pVgroup) != 0) goto _OVER;
if (mndSetDropSmaCommitLogs(pMnode, pTrans, pSma) != 0) goto _OVER;

76
source/dnode/mnode/impl/src/mndStb.c
View File

@ -15,12 +15,12 @@
#define _DEFAULT_SOURCE
#include "mndStb.h"
#include "mndPrivilege.h"
#include "mndDb.h"
#include "mndDnode.h"
#include "mndInfoSchema.h"
#include "mndMnode.h"
#include "mndPerfSchema.h"
#include "mndPrivilege.h"
#include "mndScheduler.h"
#include "mndShow.h"
#include "mndSma.h"
@ -107,7 +107,7 @@ SSdbRaw *mndStbActionEncode(SStbObj *pStb) {
int32_t funcNum = taosArrayGetSize(pStb->pFuncs);
SDB_SET_INT32(pRaw, dataPos, funcNum, _OVER)
for (int32_t i = 0; i < funcNum; ++i) {
char* func = taosArrayGet(pStb->pFuncs, i);
char *func = taosArrayGet(pStb->pFuncs, i);
SDB_SET_BINARY(pRaw, dataPos, func, TSDB_FUNC_NAME_LEN, _OVER)
}
@ -427,17 +427,17 @@ static void *mndBuildVCreateStbReq(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pSt
req.schemaTag.pSchema = pStb->pTags;
if (req.rollup) {
req.pRSmaParam.maxdelay[0] = pStb->maxdelay[0];
req.pRSmaParam.maxdelay[1] = pStb->maxdelay[1];
req.rsmaParam.maxdelay[0] = pStb->maxdelay[0];
req.rsmaParam.maxdelay[1] = pStb->maxdelay[1];
if (pStb->ast1Len > 0) {
if (mndConvertRsmaTask(&req.pRSmaParam.qmsg[0], &req.pRSmaParam.qmsgLen[0], pStb->pAst1, pStb->uid,
STREAM_TRIGGER_WINDOW_CLOSE, req.pRSmaParam.watermark[0]) < 0) {
if (mndConvertRsmaTask(&req.rsmaParam.qmsg[0], &req.rsmaParam.qmsgLen[0], pStb->pAst1, pStb->uid,
STREAM_TRIGGER_WINDOW_CLOSE, req.rsmaParam.watermark[0]) < 0) {
goto _err;
}
}
if (pStb->ast2Len > 0) {
if (mndConvertRsmaTask(&req.pRSmaParam.qmsg[1], &req.pRSmaParam.qmsgLen[1], pStb->pAst2, pStb->uid,
STREAM_TRIGGER_WINDOW_CLOSE, req.pRSmaParam.watermark[1]) < 0) {
if (mndConvertRsmaTask(&req.rsmaParam.qmsg[1], &req.rsmaParam.qmsgLen[1], pStb->pAst2, pStb->uid,
STREAM_TRIGGER_WINDOW_CLOSE, req.rsmaParam.watermark[1]) < 0) {
goto _err;
}
}
@ -470,12 +470,12 @@ static void *mndBuildVCreateStbReq(SMnode *pMnode, SVgObj *pVgroup, SStbObj *pSt
tEncoderClear(&encoder);
*pContLen = contLen;
taosMemoryFreeClear(req.pRSmaParam.qmsg[0]);
taosMemoryFreeClear(req.pRSmaParam.qmsg[1]);
taosMemoryFreeClear(req.rsmaParam.qmsg[0]);
taosMemoryFreeClear(req.rsmaParam.qmsg[1]);
return pHead;
_err:
taosMemoryFreeClear(req.pRSmaParam.qmsg[0]);
taosMemoryFreeClear(req.pRSmaParam.qmsg[1]);
taosMemoryFreeClear(req.rsmaParam.qmsg[0]);
taosMemoryFreeClear(req.rsmaParam.qmsg[1]);
return NULL;
}
@ -708,7 +708,7 @@ int32_t mndBuildStbFromReq(SMnode *pMnode, SStbObj *pDst, SMCreateStbReq *pCreat
pDst->commentLen = pCreate->commentLen;
pDst->pFuncs = pCreate->pFuncs;
pCreate->pFuncs = NULL;
if (pDst->commentLen > 0) {
pDst->comment = taosMemoryCalloc(pDst->commentLen + 1, 1);
if (pDst->comment == NULL) {
@ -1391,12 +1391,11 @@ static int32_t mndBuildStbCfgImp(SDbObj *pDb, SStbObj *pStb, const char *tbName,
if (pStb->pFuncs) {
pRsp->pFuncs = taosArrayDup(pStb->pFuncs);
}
taosRUnLockLatch(&pStb->lock);
return 0;
}
static int32_t mndBuildStbSchema(SMnode *pMnode, const char *dbFName, const char *tbName, STableMetaRsp *pRsp,
int32_t *smaVer) {
char tbFName[TSDB_TABLE_FNAME_LEN] = {0};
@ -1426,31 +1425,29 @@ static int32_t mndBuildStbSchema(SMnode *pMnode, const char *dbFName, const char
}
static int32_t mndBuildStbCfg(SMnode *pMnode, const char *dbFName, const char *tbName, STableCfgRsp *pRsp) {
char tbFName[TSDB_TABLE_FNAME_LEN] = {0};
snprintf(tbFName, sizeof(tbFName), "%s.%s", dbFName, tbName);
char tbFName[TSDB_TABLE_FNAME_LEN] = {0};
snprintf(tbFName, sizeof(tbFName), "%s.%s", dbFName, tbName);
SDbObj *pDb = mndAcquireDb(pMnode, dbFName);
if (pDb == NULL) {
terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
return -1;
}
SDbObj *pDb = mndAcquireDb(pMnode, dbFName);
if (pDb == NULL) {
terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
return -1;
}
SStbObj *pStb = mndAcquireStb(pMnode, tbFName);
if (pStb == NULL) {
mndReleaseDb(pMnode, pDb);
terrno = TSDB_CODE_PAR_TABLE_NOT_EXIST;
return -1;
}
SStbObj *pStb = mndAcquireStb(pMnode, tbFName);
if (pStb == NULL) {
mndReleaseDb(pMnode, pDb);
terrno = TSDB_CODE_PAR_TABLE_NOT_EXIST;
return -1;
}
int32_t code = mndBuildStbCfgImp(pDb, pStb, tbName, pRsp);
mndReleaseDb(pMnode, pDb);
mndReleaseStb(pMnode, pStb);
return code;
int32_t code = mndBuildStbCfgImp(pDb, pStb, tbName, pRsp);
mndReleaseDb(pMnode, pDb);
mndReleaseStb(pMnode, pStb);
return code;
}
static int32_t mndBuildSMAlterStbRsp(SDbObj *pDb, const SMAlterStbReq *pAlter, SStbObj *pObj, void **pCont,
int32_t *pLen) {
int32_t ret;
@ -1811,10 +1808,10 @@ _OVER:
}
static int32_t mndProcessTableCfgReq(SRpcMsg *pReq) {
SMnode *pMnode = pReq->info.node;
int32_t code = -1;
STableCfgReq cfgReq = {0};
STableCfgRsp cfgRsp = {0};
SMnode *pMnode = pReq->info.node;
int32_t code = -1;
STableCfgReq cfgReq = {0};
STableCfgRsp cfgRsp = {0};
if (tDeserializeSTableCfgReq(pReq->pCont, pReq->contLen, &cfgReq) != 0) {
terrno = TSDB_CODE_INVALID_MSG;
@ -1866,7 +1863,6 @@ _OVER:
return code;
}
int32_t mndValidateStbInfo(SMnode *pMnode, SSTableVersion *pStbVersions, int32_t numOfStbs, void **ppRsp,
int32_t *pRspLen) {
SSTbHbRsp hbRsp = {0};

39
source/dnode/mnode/impl/src/mndStream.c
View File

@ -14,10 +14,10 @@
*/
#include "mndStream.h"
#include "mndPrivilege.h"
#include "mndDb.h"
#include "mndDnode.h"
#include "mndMnode.h"
#include "mndPrivilege.h"
#include "mndScheduler.h"
#include "mndShow.h"
#include "mndStb.h"
@ -437,10 +437,6 @@ static int32_t mndCreateStbForStream(SMnode *pMnode, STrans *pTrans, const SStre
goto _OVER;
}
if (mndCheckDbPrivilege(pMnode, user, MND_OPER_WRITE_DB, pDb) != 0) {
goto _OVER;
}
int32_t numOfStbs = -1;
if (mndGetNumOfStbs(pMnode, pDb->name, &numOfStbs) != 0) {
goto _OVER;
@ -494,7 +490,7 @@ static int32_t mndPersistTaskDropReq(STrans *pTrans, SStreamTask *pTask) {
return 0;
}
static int32_t mndDropStreamTasks(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream) {
int32_t mndDropStreamTasks(SMnode *pMnode, STrans *pTrans, SStreamObj *pStream) {
int32_t lv = taosArrayGetSize(pStream->tasks);
for (int32_t i = 0; i < lv; i++) {
SArray *pTasks = taosArrayGetP(pStream->tasks, i);
@ -542,19 +538,6 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) {
goto _OVER;
}
// TODO check read auth for source and write auth for target
#if 0
pDb = mndAcquireDb(pMnode, createStreamReq.sourceDB);
if (pDb == NULL) {
terrno = TSDB_CODE_MND_DB_NOT_SELECTED;
goto _OVER;
}
if (mndCheckDbPrivilege(pMnode, pReq->info.conn.user, MND_OPER_WRITE_DB, pDb) != 0) {
goto _OVER;
}
#endif
// build stream obj from request
SStreamObj streamObj = {0};
if (mndBuildStreamObjFromCreateReq(pMnode, &streamObj, &createStreamReq) < 0) {
@ -592,6 +575,16 @@ static int32_t mndProcessCreateStreamReq(SRpcMsg *pReq) {
goto _OVER;
}
if (mndCheckDbPrivilegeByName(pMnode, pReq->info.conn.user, MND_OPER_READ_DB, streamObj.sourceDb) != 0) {
mndTransDrop(pTrans);
goto _OVER;
}
if (mndCheckDbPrivilegeByName(pMnode, pReq->info.conn.user, MND_OPER_WRITE_DB, streamObj.targetDb) != 0) {
mndTransDrop(pTrans);
goto _OVER;
}
// execute creation
if (mndTransPrepare(pMnode, pTrans) != 0) {
mError("trans:%d, failed to prepare since %s", pTrans->id, terrstr());
@ -641,13 +634,9 @@ static int32_t mndProcessDropStreamReq(SRpcMsg *pReq) {
}
}
#if 0
// todo check auth
pUser = mndAcquireUser(pMnode, pReq->info.conn.user);
if (pUser == NULL) {
goto DROP_STREAM_OVER;
if (mndCheckDbPrivilegeByName(pMnode, pReq->info.conn.user, MND_OPER_WRITE_DB, pStream->targetDb) != 0) {
return -1;
}
#endif
STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_RETRY, TRN_CONFLICT_NOTHING, pReq);
if (pTrans == NULL) {

12
source/dnode/mnode/impl/src/mndTopic.c
View File

@ -14,12 +14,12 @@
*/
#include "mndTopic.h"
#include "mndPrivilege.h"
#include "mndConsumer.h"
#include "mndDb.h"
#include "mndDnode.h"
#include "mndMnode.h"
#include "mndOffset.h"
#include "mndPrivilege.h"
#include "mndShow.h"
#include "mndStb.h"
#include "mndSubscribe.h"
@ -401,6 +401,10 @@ static int32_t mndCreateTopic(SMnode *pMnode, SRpcMsg *pReq, SCMCreateTopicReq *
}
} else if (pCreate->subType == TOPIC_SUB_TYPE__TABLE) {
SStbObj *pStb = mndAcquireStb(pMnode, pCreate->subStbName);
if (pStb == NULL) {
terrno = TSDB_CODE_MND_STB_NOT_EXIST;
return -1;
}
topicObj.stbUid = pStb->uid;
}
/*} else if (pCreate->subType == TOPIC_SUB_TYPE__DB) {*/
@ -480,7 +484,7 @@ static int32_t mndProcessCreateTopicReq(SRpcMsg *pReq) {
goto _OVER;
}
if (mndCheckDbPrivilege(pMnode, pReq->info.conn.user, MND_OPER_WRITE_DB, pDb) != 0) {
if (mndCheckDbPrivilege(pMnode, pReq->info.conn.user, MND_OPER_READ_DB, pDb) != 0) {
goto _OVER;
}
@ -571,6 +575,10 @@ static int32_t mndProcessDropTopicReq(SRpcMsg *pReq) {
}
#endif
if (mndCheckDbPrivilegeByName(pMnode, pReq->info.conn.user, MND_OPER_READ_DB, pTopic->db) != 0) {
return -1;
}
STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_CONFLICT_DB_INSIDE, pReq);
mndTransSetDbName(pTrans, pTopic->db, NULL);
if (pTrans == NULL) {

90
source/dnode/mnode/impl/src/mndTrans.c
View File

@ -15,15 +15,15 @@
#define _DEFAULT_SOURCE
#include "mndTrans.h"
#include "mndPrivilege.h"
#include "mndConsumer.h"
#include "mndDb.h"
#include "mndPrivilege.h"
#include "mndShow.h"
#include "mndSync.h"
#include "mndUser.h"
#define TRANS_VER_NUMBER 1
#define TRANS_ARRAY_SIZE 8
#define TRANS_VER_NUMBER 1
#define TRANS_ARRAY_SIZE 8
#define TRANS_RESERVE_SIZE 64
static SSdbRaw *mndTransActionEncode(STrans *pTrans);
@ -55,7 +55,7 @@ static bool mndTransPerfromFinishedStage(SMnode *pMnode, STrans *pTrans);
static bool mndCannotExecuteTransAction(SMnode *pMnode) { return !pMnode->deploy && !mndIsMaster(pMnode); }
static void mndTransSendRpcRsp(SMnode *pMnode, STrans *pTrans);
static int32_t mndProcessTransReq(SRpcMsg *pReq);
static int32_t mndProcessTransTimer(SRpcMsg *pReq);
static int32_t mndProcessTtl(SRpcMsg *pReq);
static int32_t mndProcessKillTransReq(SRpcMsg *pReq);
@ -73,7 +73,7 @@ int32_t mndInitTrans(SMnode *pMnode) {
.deleteFp = (SdbDeleteFp)mndTransActionDelete,
};
mndSetMsgHandle(pMnode, TDMT_MND_TRANS_TIMER, mndProcessTransReq);
mndSetMsgHandle(pMnode, TDMT_MND_TRANS_TIMER, mndProcessTransTimer);
mndSetMsgHandle(pMnode, TDMT_MND_KILL_TRANS, mndProcessKillTransReq);
mndAddShowRetrieveHandle(pMnode, TSDB_MGMT_TABLE_TRANS, mndRetrieveTrans);
@ -139,8 +139,10 @@ static SSdbRaw *mndTransActionEncode(STrans *pTrans) {
SDB_SET_INT32(pRaw, dataPos, pAction->id, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->errCode, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->acceptableCode, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->retryCode, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->actionType, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->stage, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->reserved, _OVER)
if (pAction->actionType == TRANS_ACTION_RAW) {
int32_t len = sdbGetRawTotalSize(pAction->pRaw);
SDB_SET_INT8(pRaw, dataPos, pAction->rawWritten, _OVER)
@ -163,8 +165,10 @@ static SSdbRaw *mndTransActionEncode(STrans *pTrans) {
SDB_SET_INT32(pRaw, dataPos, pAction->id, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->errCode, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->acceptableCode, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->retryCode, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->actionType, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->stage, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->reserved, _OVER)
if (pAction->actionType == TRANS_ACTION_RAW) {
int32_t len = sdbGetRawTotalSize(pAction->pRaw);
SDB_SET_INT8(pRaw, dataPos, pAction->rawWritten, _OVER)
@ -187,8 +191,10 @@ static SSdbRaw *mndTransActionEncode(STrans *pTrans) {
SDB_SET_INT32(pRaw, dataPos, pAction->id, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->errCode, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->acceptableCode, _OVER)
SDB_SET_INT32(pRaw, dataPos, pAction->retryCode, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->actionType, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->stage, _OVER)
SDB_SET_INT8(pRaw, dataPos, pAction->reserved, _OVER)
if (pAction->actionType == TRANS_ACTION_RAW) {
int32_t len = sdbGetRawTotalSize(pAction->pRaw);
SDB_SET_INT8(pRaw, dataPos, pAction->rawWritten, _OVER)
@ -291,10 +297,12 @@ static SSdbRow *mndTransActionDecode(SSdbRaw *pRaw) {
SDB_GET_INT32(pRaw, dataPos, &action.id, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.errCode, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.acceptableCode, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.retryCode, _OVER)
SDB_GET_INT8(pRaw, dataPos, &actionType, _OVER)
action.actionType = actionType;
SDB_GET_INT8(pRaw, dataPos, &stage, _OVER)
action.stage = stage;
SDB_GET_INT8(pRaw, dataPos, &action.reserved, _OVER)
if (action.actionType == TRANS_ACTION_RAW) {
SDB_GET_INT8(pRaw, dataPos, &action.rawWritten, _OVER)
SDB_GET_INT32(pRaw, dataPos, &dataLen, _OVER)
@ -324,10 +332,12 @@ static SSdbRow *mndTransActionDecode(SSdbRaw *pRaw) {
SDB_GET_INT32(pRaw, dataPos, &action.id, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.errCode, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.acceptableCode, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.retryCode, _OVER)
SDB_GET_INT8(pRaw, dataPos, &actionType, _OVER)
action.actionType = actionType;
SDB_GET_INT8(pRaw, dataPos, &stage, _OVER)
action.stage = stage;
SDB_GET_INT8(pRaw, dataPos, &action.reserved, _OVER)
if (action.actionType == TRANS_ACTION_RAW) {
SDB_GET_INT8(pRaw, dataPos, &action.rawWritten, _OVER)
SDB_GET_INT32(pRaw, dataPos, &dataLen, _OVER)
@ -357,10 +367,12 @@ static SSdbRow *mndTransActionDecode(SSdbRaw *pRaw) {
SDB_GET_INT32(pRaw, dataPos, &action.id, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.errCode, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.acceptableCode, _OVER)
SDB_GET_INT32(pRaw, dataPos, &action.retryCode, _OVER)
SDB_GET_INT8(pRaw, dataPos, &actionType, _OVER)
action.actionType = actionType;
SDB_GET_INT8(pRaw, dataPos, &stage, _OVER)
action.stage = stage;
SDB_GET_INT8(pRaw, dataPos, &action.reserved, _OVER)
if (action.actionType) {
SDB_GET_INT8(pRaw, dataPos, &action.rawWritten, _OVER)
SDB_GET_INT32(pRaw, dataPos, &dataLen, _OVER)
@ -463,15 +475,25 @@ static int32_t mndTransActionInsert(SSdb *pSdb, STrans *pTrans) {
if (fp) {
(*fp)(pSdb->pMnode, pTrans->param, pTrans->paramLen);
}
pTrans->startFunc = 0;
}
return 0;
}
static void mndTransDropData(STrans *pTrans) {
mndTransDropActions(pTrans->redoActions);
mndTransDropActions(pTrans->undoActions);
mndTransDropActions(pTrans->commitActions);
if (pTrans->redoActions != NULL) {
mndTransDropActions(pTrans->redoActions);
pTrans->redoActions = NULL;
}
if (pTrans->undoActions != NULL) {
mndTransDropActions(pTrans->undoActions);
pTrans->undoActions = NULL;
}
if (pTrans->commitActions != NULL) {
mndTransDropActions(pTrans->commitActions);
pTrans->commitActions = NULL;
}
if (pTrans->rpcRsp != NULL) {
taosMemoryFree(pTrans->rpcRsp);
pTrans->rpcRsp = NULL;
@ -492,6 +514,7 @@ static int32_t mndTransActionDelete(SSdb *pSdb, STrans *pTrans, bool callFunc) {
if (fp) {
(*fp)(pSdb->pMnode, pTrans->param, pTrans->paramLen);
}
pTrans->stopFunc = 0;
}
mndTransDropData(pTrans);
@ -805,7 +828,7 @@ static void mndTransSendRpcRsp(SMnode *pMnode, STrans *pTrans) {
sendRsp = true;
}
} else {
if (pTrans->stage == TRN_STAGE_REDO_ACTION && pTrans->failedTimes > 3) {
if (pTrans->stage == TRN_STAGE_REDO_ACTION && pTrans->failedTimes > 6) {
if (code == 0) code = TSDB_CODE_MND_TRANS_UNKNOW_ERROR;
sendRsp = true;
}
@ -875,8 +898,8 @@ int32_t mndTransProcessRsp(SRpcMsg *pRsp) {
pAction->errCode = pRsp->code;
}
mDebug("trans:%d, %s:%d response is received, code:0x%x, accept:0x%x", transId, mndTransStr(pAction->stage), action,
pRsp->code, pAction->acceptableCode);
mDebug("trans:%d, %s:%d response is received, code:0x%x, accept:0x%x retry:0x%x", transId,
mndTransStr(pAction->stage), action, pRsp->code, pAction->acceptableCode, pAction->retryCode);
mndTransExecute(pMnode, pTrans);
_OVER:
@ -884,6 +907,21 @@ _OVER:
return 0;
}
static void mndTransResetAction(SMnode *pMnode, STrans *pTrans, STransAction *pAction) {
pAction->rawWritten = 0;
pAction->msgSent = 0;
pAction->msgReceived = 0;
if (pAction->errCode == TSDB_CODE_RPC_REDIRECT || pAction->errCode == TSDB_CODE_SYN_NEW_CONFIG_ERROR ||
pAction->errCode == TSDB_CODE_SYN_INTERNAL_ERROR || pAction->errCode == TSDB_CODE_SYN_NOT_LEADER) {
pAction->epSet.inUse = (pAction->epSet.inUse + 1) % pAction->epSet.numOfEps;
mDebug("trans:%d, %s:%d execute status is reset and set epset inuse:%d", pTrans->id, mndTransStr(pAction->stage),
pAction->id, pAction->epSet.inUse);
} else {
mDebug("trans:%d, %s:%d execute status is reset", pTrans->id, mndTransStr(pAction->stage), pAction->id);
}
pAction->errCode = 0;
}
static void mndTransResetActions(SMnode *pMnode, STrans *pTrans, SArray *pArray) {
int32_t numOfActions = taosArrayGetSize(pArray);
@ -894,18 +932,7 @@ static void mndTransResetActions(SMnode *pMnode, STrans *pTrans, SArray *pArray)
continue;
if (pAction->rawWritten && (pAction->errCode == 0 || pAction->errCode == pAction->acceptableCode)) continue;
pAction->rawWritten = 0;
pAction->msgSent = 0;
pAction->msgReceived = 0;
if (pAction->errCode == TSDB_CODE_RPC_REDIRECT || pAction->errCode == TSDB_CODE_SYN_NEW_CONFIG_ERROR ||
pAction->errCode == TSDB_CODE_SYN_INTERNAL_ERROR || pAction->errCode == TSDB_CODE_SYN_NOT_LEADER) {
pAction->epSet.inUse = (pAction->epSet.inUse + 1) % pAction->epSet.numOfEps;
mDebug("trans:%d, %s:%d execute status is reset and set epset inuse:%d", pTrans->id, mndTransStr(pAction->stage),
action, pAction->epSet.inUse);
} else {
mDebug("trans:%d, %s:%d execute status is reset", pTrans->id, mndTransStr(pAction->stage), action);
}
pAction->errCode = 0;
mndTransResetAction(pMnode, pTrans, pAction);
}
}
@ -1112,9 +1139,9 @@ static int32_t mndTransExecuteRedoActionsSerial(SMnode *pMnode, STrans *pTrans)
if (pAction->msgReceived) {
if (pAction->errCode != 0 && pAction->errCode != pAction->acceptableCode) {
code = pAction->errCode;
pAction->msgSent = 0;
pAction->msgReceived = 0;
mDebug("trans:%d, %s:%d execute status is reset", pTrans->id, mndTransStr(pAction->stage), action);
mndTransResetAction(pMnode, pTrans, pAction);
} else {
mDebug("trans:%d, %s:%d execute successfully", pTrans->id, mndTransStr(pAction->stage), action);
}
} else {
code = TSDB_CODE_ACTION_IN_PROGRESS;
@ -1123,6 +1150,8 @@ static int32_t mndTransExecuteRedoActionsSerial(SMnode *pMnode, STrans *pTrans)
if (pAction->rawWritten) {
if (pAction->errCode != 0 && pAction->errCode != pAction->acceptableCode) {
code = pAction->errCode;
} else {
mDebug("trans:%d, %s:%d write successfully", pTrans->id, mndTransStr(pAction->stage), action);
}
}
}
@ -1156,9 +1185,16 @@ static int32_t mndTransExecuteRedoActionsSerial(SMnode *pMnode, STrans *pTrans)
} else if (code == TSDB_CODE_ACTION_IN_PROGRESS) {
mDebug("trans:%d, %s:%d is in progress and wait it finish", pTrans->id, mndTransStr(pAction->stage), pAction->id);
break;
} else if (code == pAction->retryCode) {
mDebug("trans:%d, %s:%d receive code:0x%x and retry", pTrans->id, mndTransStr(pAction->stage), pAction->id, code);
taosMsleep(300);
action--;
continue;
} else {
terrno = code;
pTrans->code = code;
mDebug("trans:%d, %s:%d receive code:0x%x and wait another schedule, failedTimes:%d", pTrans->id,
mndTransStr(pAction->stage), pAction->id, code, pTrans->failedTimes);
break;
}
}
@ -1343,7 +1379,7 @@ void mndTransExecute(SMnode *pMnode, STrans *pTrans) {
mndTransSendRpcRsp(pMnode, pTrans);
}
static int32_t mndProcessTransReq(SRpcMsg *pReq) {
static int32_t mndProcessTransTimer(SRpcMsg *pReq) {
mndTransPullup(pReq->info.node);
return 0;
}

12
source/dnode/mnode/impl/src/mndVgroup.c
View File

@ -15,10 +15,10 @@
#define _DEFAULT_SOURCE
#include "mndVgroup.h"
#include "mndPrivilege.h"
#include "mndDb.h"
#include "mndDnode.h"
#include "mndMnode.h"
#include "mndPrivilege.h"
#include "mndShow.h"
#include "mndTrans.h"
#include "mndUser.h"
@ -896,6 +896,8 @@ int32_t mndAddAlterVnodeConfirmAction(SMnode *pMnode, STrans *pTrans, SDbObj *pD
action.pCont = pHead;
action.contLen = contLen;
action.msgType = TDMT_VND_ALTER_CONFIRM;
// incorrect redirect result will cause this erro
action.retryCode = TSDB_CODE_VND_INVALID_VGROUP_ID;
if (mndTransAppendRedoAction(pTrans, &action) != 0) {
taosMemoryFree(pHead);
@ -942,6 +944,8 @@ static int32_t mndAddSetVnodeStandByAction(SMnode *pMnode, STrans *pTrans, SDbOb
action.contLen = contLen;
action.msgType = TDMT_SYNC_SET_VNODE_STANDBY;
action.acceptableCode = TSDB_CODE_NODE_NOT_DEPLOYED;
// Keep retrying until the target vnode is not the leader
action.retryCode = TSDB_CODE_SYN_IS_LEADER;
if (isRedo) {
if (mndTransAppendRedoAction(pTrans, &action) != 0) {
@ -1229,7 +1233,8 @@ static int32_t mndProcessRedistributeVgroupMsg(SRpcMsg *pReq) {
}
if (req.dnodeId1 == pVgroup->vnodeGid[0].dnodeId) {
terrno = TSDB_CODE_MND_VGROUP_UN_CHANGED;
// terrno = TSDB_CODE_MND_VGROUP_UN_CHANGED;
code = 0;
goto _OVER;
}
@ -1351,7 +1356,8 @@ static int32_t mndProcessRedistributeVgroupMsg(SRpcMsg *pReq) {
}
if (pNew1 == NULL && pOld1 == NULL && pNew2 == NULL && pOld2 == NULL && pNew3 == NULL && pOld3 == NULL) {
terrno = TSDB_CODE_MND_VGROUP_UN_CHANGED;
// terrno = TSDB_CODE_MND_VGROUP_UN_CHANGED;
code = 0;
goto _OVER;
}

2
source/dnode/snode/src/snode.c
View File

@ -257,7 +257,7 @@ int32_t sndProcessSMsg(SSnode *pSnode, SRpcMsg *pMsg) {
case TDMT_STREAM_TASK_RECOVER_RSP:
return sndProcessTaskRecoverRsp(pSnode, pMsg);
case TDMT_STREAM_RETRIEVE_RSP:
return sndProcessTaskRecoverRsp(pSnode, pMsg);
return sndProcessTaskRetrieveRsp(pSnode, pMsg);
default:
ASSERT(0);
}

34
source/dnode/vnode/inc/vnode.h
View File

@ -116,18 +116,19 @@ typedef void *tsdbReaderT;
#define BLOCK_LOAD_TABLE_SEQ_ORDER 2
#define BLOCK_LOAD_TABLE_RR_ORDER 3
int32_t tsdbSetTableList(tsdbReaderT reader, SArray* tableList);
tsdbReaderT tsdbReaderOpen(SVnode *pVnode, SQueryTableDataCond *pCond, SArray *tableList, uint64_t qId,
uint64_t taskId);
tsdbReaderT tsdbQueryCacheLast(SVnode *pVnode, SQueryTableDataCond *pCond, STableListInfo *groupList, uint64_t qId,
void *pMemRef);
int32_t tsdbGetFileBlocksDistInfo(tsdbReaderT *pReader, STableBlockDistInfo *pTableBlockInfo);
bool isTsdbCacheLastRow(tsdbReaderT *pReader);
int32_t tsdbGetAllTableList(SMeta *pMeta, uint64_t uid, SArray *list);
int32_t tsdbGetCtbIdList(SMeta *pMeta, int64_t suid, SArray *list);
void *tsdbGetIdx(SMeta *pMeta);
void *tsdbGetIvtIdx(SMeta *pMeta);
int64_t tsdbGetNumOfRowsInMemTable(tsdbReaderT *pHandle);
int32_t tsdbSetTableList(tsdbReaderT reader, SArray *tableList);
tsdbReaderT tsdbReaderOpen(SVnode *pVnode, SQueryTableDataCond *pCond, SArray *tableList, uint64_t qId,
uint64_t taskId);
tsdbReaderT tsdbQueryCacheLast(SVnode *pVnode, SQueryTableDataCond *pCond, STableListInfo *groupList, uint64_t qId,
void *pMemRef);
int32_t tsdbGetFileBlocksDistInfo(tsdbReaderT *pReader, STableBlockDistInfo *pTableBlockInfo);
bool isTsdbCacheLastRow(tsdbReaderT *pReader);
int32_t tsdbGetAllTableList(SMeta *pMeta, uint64_t uid, SArray *list);
int32_t tsdbGetCtbIdList(SMeta *pMeta, int64_t suid, SArray *list);
int32_t tsdbGetStbIdList(SMeta *pMeta, int64_t suid, SArray *list);
void *tsdbGetIdx(SMeta *pMeta);
void *tsdbGetIvtIdx(SMeta *pMeta);
int64_t tsdbGetNumOfRowsInMemTable(tsdbReaderT *pHandle);
bool tsdbNextDataBlock(tsdbReaderT pTsdbReadHandle);
void tsdbRetrieveDataBlockInfo(tsdbReaderT *pTsdbReadHandle, SDataBlockInfo *pBlockInfo);
@ -150,8 +151,7 @@ int32_t tqReadHandleRemoveTbUidList(STqReadHandle *pHandle, const SArray *tbUidL
int32_t tqReadHandleSetMsg(STqReadHandle *pHandle, SSubmitReq *pMsg, int64_t ver);
bool tqNextDataBlock(STqReadHandle *pHandle);
bool tqNextDataBlockFilterOut(STqReadHandle *pHandle, SHashObj *filterOutUids);
int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReadHandle *pHandle, uint64_t *pGroupId, uint64_t *pUid,
int32_t *pNumOfRows);
int32_t tqRetrieveDataBlock(SSDataBlock *pBlock, STqReadHandle *pHandle);
// sma
int32_t smaGetTSmaDays(SVnodeCfg *pCfg, void *pCont, uint32_t contLen, int32_t *days);
@ -196,17 +196,23 @@ struct SVnodeCfg {
typedef struct {
TSKEY lastKey;
uint64_t uid;
uint64_t groupId;
} STableKeyInfo;
#define TABLE_ROLLUP_ON ((int8_t)0x1)
#define TABLE_IS_ROLLUP(FLG) (((FLG) & (TABLE_ROLLUP_ON)) != 0)
#define TABLE_SET_ROLLUP(FLG) ((FLG) |= TABLE_ROLLUP_ON)
struct SMetaEntry {
int64_t version;
int8_t type;
int8_t flags; // TODO: need refactor?
tb_uid_t uid;
char *name;
union {
struct {
SSchemaWrapper schemaRow;
SSchemaWrapper schemaTag;
SRSmaParam rsmaParam;
} stbEntry;
struct {
int64_t ctime;

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

@ -69,6 +69,7 @@ struct SMeta {
TTB* pUidIdx;
TTB* pNameIdx;
TTB* pCtbIdx;
TTB* pSuidIdx;
// ivt idx and idx
void* pTagIvtIdx;
TTB* pTagIdx;

30
source/dnode/vnode/src/inc/sma.h
View File

@ -60,8 +60,9 @@ struct SRSmaStat {
SSma *pSma;
void *tmrHandle;
tmr_h tmrId;
int8_t tmrStat;
int32_t tmrSeconds;
int8_t triggerStat;
int8_t runningStat;
SHashObj *rsmaInfoHash; // key: stbUid, value: SRSmaInfo;
};
@ -72,13 +73,21 @@ struct SSmaStat {
};
T_REF_DECLARE()
};
#define SMA_TSMA_STAT(s) (&(s)->tsmaStat)
#define SMA_RSMA_STAT(s) (&(s)->rsmaStat)
#define SMA_RSMA_INFO_HASH(s) ((s)->rsmaStat.rsmaInfoHash)
#define SMA_RSMA_TMR_HANDLE(s) ((s)->rsmaStat.tmrHandle)
#define SMA_RSMA_TMR_STAT(s) ((s)->rsmaStat.tmrStat)
#define RSMA_INFO_HASH(r) ((r)->rsmaInfoHash)
#define SMA_TSMA_STAT(s) (&(s)->tsmaStat)
#define SMA_RSMA_STAT(s) (&(s)->rsmaStat)
#define RSMA_INFO_HASH(r) ((r)->rsmaInfoHash)
#define RSMA_TMR_ID(r) ((r)->tmrId)
#define RSMA_TMR_HANDLE(r) ((r)->tmrHandle)
#define RSMA_TRIGGER_STAT(r) (&(r)->triggerStat)
#define RSMA_RUNNING_STAT(r) (&(r)->runningStat)
enum {
TASK_TRIGGER_STAT_INIT = 0,
TASK_TRIGGER_STAT_ACTIVE = 1,
TASK_TRIGGER_STAT_INACTIVE = 2,
TASK_TRIGGER_STAT_CANCELLED = 3,
TASK_TRIGGER_STAT_FINISHED = 4,
};
void tdDestroySmaEnv(SSmaEnv *pSmaEnv);
void *tdFreeSmaEnv(SSmaEnv *pSmaEnv);
@ -174,6 +183,8 @@ int32_t tdProcessTSmaCreateImpl(SSma *pSma, int64_t version, const char *pMsg);
int32_t tdProcessTSmaInsertImpl(SSma *pSma, int64_t indexUid, const char *msg);
int32_t tdProcessTSmaGetDaysImpl(SVnodeCfg *pCfg, void *pCont, uint32_t contLen, int32_t *days);
// smaFileUtil ================
typedef struct STFInfo STFInfo;
typedef struct STFile STFile;
@ -181,7 +192,7 @@ struct STFInfo {
uint32_t magic;
uint32_t ftype;
uint32_t fver;
uint64_t fsize;
int64_t fsize;
};
struct STFile {
@ -199,11 +210,8 @@ struct STFile {
#define TD_FILE_OPENED(tf) (TD_FILE_PFILE(tf) != NULL)
#define TD_FILE_CLOSED(tf) (!TD_FILE_OPENED(tf))
#define TD_FILE_SET_CLOSED(f) (TD_FILE_PFILE(f) = NULL)
#define TD_FILE_STATE(tf) ((tf)->state)
#define TD_FILE_SET_STATE(tf, s) ((tf)->state = (s))
#define TD_FILE_DID(tf) (TD_FILE_F(tf)->did)
#define TD_FILE_IS_OK(tf) (TD_FILE_STATE(tf) == TD_FILE_STATE_OK)
#define TD_FILE_IS_BAD(tf) (TD_FILE_STATE(tf) == TD_FILE_STATE_BAD)
int32_t tdInitTFile(STFile *pTFile, STfs *pTfs, const char *fname);
int32_t tdCreateTFile(STFile *pTFile, STfs *pTfs, bool updateHeader, int8_t fType);

6
source/dnode/vnode/src/inc/vnodeInt.h
View File

@ -76,6 +76,7 @@ void vnodeFree(void* p);
// meta
typedef struct SMCtbCursor SMCtbCursor;
typedef struct SMStbCursor SMStbCursor;
typedef struct STbUidStore STbUidStore;
int metaOpen(SVnode* pVnode, SMeta** ppMeta);
@ -97,6 +98,9 @@ int metaGetTbNum(SMeta* pMeta);
SMCtbCursor* metaOpenCtbCursor(SMeta* pMeta, tb_uid_t uid);
void metaCloseCtbCursor(SMCtbCursor* pCtbCur);
tb_uid_t metaCtbCursorNext(SMCtbCursor* pCtbCur);
SMStbCursor* metaOpenStbCursor(SMeta* pMeta, tb_uid_t uid);
void metaCloseStbCursor(SMStbCursor* pStbCur);
tb_uid_t metaStbCursorNext(SMStbCursor* pStbCur);
STSma* metaGetSmaInfoByIndex(SMeta* pMeta, int64_t indexUid);
STSmaWrapper* metaGetSmaInfoByTable(SMeta* pMeta, tb_uid_t uid, bool deepCopy);
SArray* metaGetSmaIdsByTable(SMeta* pMeta, tb_uid_t uid);
@ -158,6 +162,8 @@ SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pSchema, bool
// sma
int32_t smaOpen(SVnode* pVnode);
int32_t smaClose(SSma* pSma);
int32_t smaCloseEnv(SSma* pSma);
int32_t smaCloseEx(SSma* pSma);
int32_t tdProcessTSmaCreate(SSma* pSma, int64_t version, const char* msg);
int32_t tdProcessTSmaInsert(SSma* pSma, int64_t indexUid, const char* msg);

18
source/dnode/vnode/src/meta/metaEntry.c
View File

@ -24,22 +24,25 @@ int metaEncodeEntry(SEncoder *pCoder, const SMetaEntry *pME) {
if (tEncodeCStr(pCoder, pME->name) < 0) return -1;
if (pME->type == TSDB_SUPER_TABLE) {
if (tEncodeI8(pCoder, pME->flags) < 0) return -1; // TODO: need refactor?
if (tEncodeSSchemaWrapper(pCoder, &pME->stbEntry.schemaRow) < 0) return -1;
if (tEncodeSSchemaWrapper(pCoder, &pME->stbEntry.schemaTag) < 0) return -1;
if (TABLE_IS_ROLLUP(pME->flags)) {
if (tEncodeSRSmaParam(pCoder, &pME->stbEntry.rsmaParam) < 0) return -1;
}
} else if (pME->type == TSDB_CHILD_TABLE) {
if (tEncodeI64(pCoder, pME->ctbEntry.ctime) < 0) return -1;
if (tEncodeI32(pCoder, pME->ctbEntry.ttlDays) < 0) return -1;
if (tEncodeI32(pCoder, pME->ctbEntry.commentLen) < 0) return -1;
if (tEncodeI32v(pCoder, pME->ctbEntry.commentLen) < 0) return -1;
if (pME->ctbEntry.commentLen > 0){
if (tEncodeCStr(pCoder, pME->ctbEntry.comment) < 0) return -1;
}
if (tEncodeI64(pCoder, pME->ctbEntry.suid) < 0) return -1;
debugCheckTags((STag*)pME->ctbEntry.pTags); // TODO: remove after debug
if (tEncodeTag(pCoder, (const STag *)pME->ctbEntry.pTags) < 0) return -1;
} else if (pME->type == TSDB_NORMAL_TABLE) {
if (tEncodeI64(pCoder, pME->ntbEntry.ctime) < 0) return -1;
if (tEncodeI32(pCoder, pME->ntbEntry.ttlDays) < 0) return -1;
if (tEncodeI32(pCoder, pME->ntbEntry.commentLen) < 0) return -1;
if (tEncodeI32v(pCoder, pME->ntbEntry.commentLen) < 0) return -1;
if (pME->ntbEntry.commentLen > 0){
if (tEncodeCStr(pCoder, pME->ntbEntry.comment) < 0) return -1;
}
@ -64,23 +67,26 @@ int metaDecodeEntry(SDecoder *pCoder, SMetaEntry *pME) {
if (tDecodeCStr(pCoder, &pME->name) < 0) return -1;
if (pME->type == TSDB_SUPER_TABLE) {
if (tDecodeI8(pCoder, &pME->flags) < 0) return -1; // TODO: need refactor?
if (tDecodeSSchemaWrapperEx(pCoder, &pME->stbEntry.schemaRow) < 0) return -1;
if (tDecodeSSchemaWrapperEx(pCoder, &pME->stbEntry.schemaTag) < 0) return -1;
if (TABLE_IS_ROLLUP(pME->flags)) {
if (tDecodeSRSmaParam(pCoder, &pME->stbEntry.rsmaParam) < 0) return -1;
}
} else if (pME->type == TSDB_CHILD_TABLE) {
if (tDecodeI64(pCoder, &pME->ctbEntry.ctime) < 0) return -1;
if (tDecodeI32(pCoder, &pME->ctbEntry.ttlDays) < 0) return -1;
if (tDecodeI32(pCoder, &pME->ctbEntry.commentLen) < 0) return -1;
if (tDecodeI32v(pCoder, &pME->ctbEntry.commentLen) < 0) return -1;
if (pME->ctbEntry.commentLen > 0){
if (tDecodeCStr(pCoder, &pME->ctbEntry.comment) < 0)
return -1;
}
if (tDecodeI64(pCoder, &pME->ctbEntry.suid) < 0) return -1;
if (tDecodeTag(pCoder, (STag **)&pME->ctbEntry.pTags) < 0) return -1; // (TODO)
debugCheckTags((STag*)pME->ctbEntry.pTags); // TODO: remove after debug
} else if (pME->type == TSDB_NORMAL_TABLE) {
if (tDecodeI64(pCoder, &pME->ntbEntry.ctime) < 0) return -1;
if (tDecodeI32(pCoder, &pME->ntbEntry.ttlDays) < 0) return -1;
if (tDecodeI32(pCoder, &pME->ntbEntry.commentLen) < 0) return -1;
if (tDecodeI32v(pCoder, &pME->ntbEntry.commentLen) < 0) return -1;
if (pME->ntbEntry.commentLen > 0){
if (tDecodeCStr(pCoder, &pME->ntbEntry.comment) < 0) return -1;
}

9
source/dnode/vnode/src/meta/metaOpen.c
View File

@ -92,6 +92,13 @@ int metaOpen(SVnode *pVnode, SMeta **ppMeta) {
goto _err;
}
// open pSuidIdx
ret = tdbTbOpen("suid.idx", sizeof(tb_uid_t), 0, uidIdxKeyCmpr, pMeta->pEnv, &pMeta->pSuidIdx);
if (ret < 0) {
metaError("vgId:%d, failed to open meta super table index since %s", TD_VID(pVnode), tstrerror(terrno));
goto _err;
}
// open pTagIdx
// TODO(yihaoDeng), refactor later
char indexFullPath[128] = {0};
@ -141,6 +148,7 @@ _err:
if (pMeta->pTagIvtIdx) indexClose(pMeta->pTagIvtIdx);
if (pMeta->pTagIdx) tdbTbClose(pMeta->pTagIdx);
if (pMeta->pCtbIdx) tdbTbClose(pMeta->pCtbIdx);
if (pMeta->pSuidIdx) tdbTbClose(pMeta->pSuidIdx);
if (pMeta->pNameIdx) tdbTbClose(pMeta->pNameIdx);
if (pMeta->pUidIdx) tdbTbClose(pMeta->pUidIdx);
if (pMeta->pSkmDb) tdbTbClose(pMeta->pSkmDb);
@ -162,6 +170,7 @@ int metaClose(SMeta *pMeta) {
if (pMeta->pTagIdx) tdbTbClose(pMeta->pTagIdx);
#endif
if (pMeta->pCtbIdx) tdbTbClose(pMeta->pCtbIdx);
if (pMeta->pSuidIdx) tdbTbClose(pMeta->pSuidIdx);
if (pMeta->pNameIdx) tdbTbClose(pMeta->pNameIdx);
if (pMeta->pUidIdx) tdbTbClose(pMeta->pUidIdx);
if (pMeta->pSkmDb) tdbTbClose(pMeta->pSkmDb);

66
source/dnode/vnode/src/meta/metaQuery.c
View File

@ -325,6 +325,72 @@ tb_uid_t metaCtbCursorNext(SMCtbCursor *pCtbCur) {
return pCtbIdxKey->uid;
}
struct SMStbCursor {
SMeta *pMeta;
TBC *pCur;
tb_uid_t suid;
void *pKey;
void *pVal;
int kLen;
int vLen;
};
SMStbCursor *metaOpenStbCursor(SMeta *pMeta, tb_uid_t suid) {
SMStbCursor *pStbCur = NULL;
int ret = 0;
int c = 0;
pStbCur = (SMStbCursor *)taosMemoryCalloc(1, sizeof(*pStbCur));
if (pStbCur == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
}
pStbCur->pMeta = pMeta;
pStbCur->suid = suid;
metaRLock(pMeta);
ret = tdbTbcOpen(pMeta->pSuidIdx, &pStbCur->pCur, NULL);
if (ret < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
metaULock(pMeta);
taosMemoryFree(pStbCur);
return NULL;
}
// move to the suid
tdbTbcMoveTo(pStbCur->pCur, &suid, sizeof(suid), &c);
if (c > 0) {
tdbTbcMoveToNext(pStbCur->pCur);
}
return pStbCur;
}
void metaCloseStbCursor(SMStbCursor *pStbCur) {
if (pStbCur) {
if (pStbCur->pMeta) metaULock(pStbCur->pMeta);
if (pStbCur->pCur) {
tdbTbcClose(pStbCur->pCur);
tdbFree(pStbCur->pKey);
tdbFree(pStbCur->pVal);
}
taosMemoryFree(pStbCur);
}
}
tb_uid_t metaStbCursorNext(SMStbCursor *pStbCur) {
int ret;
ret = tdbTbcNext(pStbCur->pCur, &pStbCur->pKey, &pStbCur->kLen, &pStbCur->pVal, &pStbCur->vLen);
if (ret < 0) {
return 0;
}
return *(tb_uid_t*)pStbCur->pKey;
}
STSchema *metaGetTbTSchema(SMeta *pMeta, tb_uid_t uid, int32_t sver) {
// SMetaReader mr = {0};
STSchema * pTSchema = NULL;

18
source/dnode/vnode/src/meta/metaTable.c
View File

@ -23,6 +23,7 @@ static int metaUpdateNameIdx(SMeta *pMeta, const SMetaEntry *pME);
static int metaUpdateTtlIdx(SMeta *pMeta, const SMetaEntry *pME);
static int metaSaveToSkmDb(SMeta *pMeta, const SMetaEntry *pME);
static int metaUpdateCtbIdx(SMeta *pMeta, const SMetaEntry *pME);
static int metaUpdateSuidIdx(SMeta *pMeta, const SMetaEntry *pME);
static int metaUpdateTagIdx(SMeta *pMeta, const SMetaEntry *pCtbEntry);
static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type);
@ -138,6 +139,10 @@ int metaCreateSTable(SMeta *pMeta, int64_t version, SVCreateStbReq *pReq) {
me.name = pReq->name;
me.stbEntry.schemaRow = pReq->schemaRow;
me.stbEntry.schemaTag = pReq->schemaTag;
if (pReq->rollup) {
TABLE_SET_ROLLUP(me.flags);
me.stbEntry.rsmaParam = pReq->rsmaParam;
}
if (metaHandleEntry(pMeta, &me) < 0) goto _err;
@ -209,6 +214,7 @@ _drop_super_table:
&pMeta->txn);
tdbTbDelete(pMeta->pNameIdx, pReq->name, strlen(pReq->name) + 1, &pMeta->txn);
tdbTbDelete(pMeta->pUidIdx, &pReq->suid, sizeof(tb_uid_t), &pMeta->txn);
tdbTbDelete(pMeta->pSuidIdx, &pReq->suid, sizeof(tb_uid_t), &pMeta->txn);
metaULock(pMeta);
@ -388,7 +394,7 @@ int metaTtlDropTable(SMeta *pMeta, int64_t ttl, SArray *tbUids) {
}
static void metaBuildTtlIdxKey(STtlIdxKey *ttlKey, const SMetaEntry *pME){
int32_t ttlDays;
int64_t ttlDays;
int64_t ctime;
if (pME->type == TSDB_CHILD_TABLE) {
ctime = pME->ctbEntry.ctime;
@ -436,11 +442,13 @@ static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type) {
tdbTbDelete(pMeta->pUidIdx, &uid, sizeof(uid), &pMeta->txn);
if(e.type != TSDB_SUPER_TABLE) metaDeleteTtlIdx(pMeta, &e);
if (e.type == TSDB_CHILD_TABLE) {
tdbTbDelete(pMeta->pCtbIdx, &(SCtbIdxKey){.suid = e.ctbEntry.suid, .uid = uid}, sizeof(SCtbIdxKey), &pMeta->txn);
} else if (e.type == TSDB_NORMAL_TABLE) {
// drop schema.db (todo)
} else if (e.type == TSDB_SUPER_TABLE) {
tdbTbDelete(pMeta->pSuidIdx, &e.uid, sizeof(tb_uid_t), &pMeta->txn);
// drop schema.db (todo)
}
@ -911,6 +919,10 @@ static int metaUpdateUidIdx(SMeta *pMeta, const SMetaEntry *pME) {
return tdbTbInsert(pMeta->pUidIdx, &pME->uid, sizeof(tb_uid_t), &pME->version, sizeof(int64_t), &pMeta->txn);
}
static int metaUpdateSuidIdx(SMeta *pMeta, const SMetaEntry *pME) {
return tdbTbInsert(pMeta->pSuidIdx, &pME->uid, sizeof(tb_uid_t), NULL, 0, &pMeta->txn);
}
static int metaUpdateNameIdx(SMeta *pMeta, const SMetaEntry *pME) {
return tdbTbInsert(pMeta->pNameIdx, pME->name, strlen(pME->name) + 1, &pME->uid, sizeof(tb_uid_t), &pMeta->txn);
}
@ -1081,6 +1093,10 @@ static int metaHandleEntry(SMeta *pMeta, const SMetaEntry *pME) {
} else {
// update schema.db
if (metaSaveToSkmDb(pMeta, pME) < 0) goto _err;
if (pME->type == TSDB_SUPER_TABLE) {
if (metaUpdateSuidIdx(pMeta, pME) < 0) goto _err;
}
}
if (pME->type != TSDB_SUPER_TABLE) {

136
source/dnode/vnode/src/sma/smaEnv.c
View File

@ -126,22 +126,21 @@ static int32_t tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType, const SSma *pS
}
if (smaType == TSDB_SMA_TYPE_ROLLUP) {
SMA_RSMA_STAT(*pSmaStat)->pSma = (SSma*)pSma;
SRSmaStat *pRSmaStat = (SRSmaStat *)(*pSmaStat);
pRSmaStat->pSma = (SSma *)pSma;
// init timer
SMA_RSMA_TMR_HANDLE(*pSmaStat) = taosTmrInit(10000, 100, 10000, "RSMA_G");
if (!SMA_RSMA_TMR_HANDLE(*pSmaStat)) {
RSMA_TMR_HANDLE(pRSmaStat) = taosTmrInit(10000, 100, 10000, "RSMA");
if (!RSMA_TMR_HANDLE(pRSmaStat)) {
taosMemoryFreeClear(*pSmaStat);
return TSDB_CODE_FAILED;
}
atomic_store_8(&SMA_RSMA_TMR_STAT(*pSmaStat), TASK_TRIGGER_STATUS__ACTIVE);
// init hash
SMA_RSMA_INFO_HASH(*pSmaStat) = taosHashInit(
RSMA_INFO_HASH(pRSmaStat) = taosHashInit(
RSMA_TASK_INFO_HASH_SLOT, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, HASH_ENTRY_LOCK);
if (!SMA_RSMA_INFO_HASH(*pSmaStat)) {
if (SMA_RSMA_TMR_HANDLE(*pSmaStat)) {
taosTmrCleanUp(SMA_RSMA_TMR_HANDLE(*pSmaStat));
if (!RSMA_INFO_HASH(pRSmaStat)) {
if (RSMA_TMR_HANDLE(pRSmaStat)) {
taosTmrCleanUp(RSMA_TMR_HANDLE(pRSmaStat));
}
taosMemoryFreeClear(*pSmaStat);
return TSDB_CODE_FAILED;
@ -155,12 +154,80 @@ static int32_t tdInitSmaStat(SSmaStat **pSmaStat, int8_t smaType, const SSma *pS
return TSDB_CODE_SUCCESS;
}
static void *tdFreeTSmaStat(STSmaStat *pStat) {
static void tdDestroyTSmaStat(STSmaStat *pStat) {
if (pStat) {
smaDebug("destroy tsma stat");
tDestroyTSma(pStat->pTSma);
taosMemoryFreeClear(pStat->pTSma);
taosMemoryFreeClear(pStat);
taosMemoryFreeClear(pStat->pTSchema);
}
}
static void *tdFreeTSmaStat(STSmaStat *pStat) {
tdDestroyTSmaStat(pStat);
taosMemoryFreeClear(pStat);
return NULL;
}
static void tdDestroyRSmaStat(SRSmaStat *pStat) {
if (pStat) {
smaDebug("vgId:%d destroy rsma stat", SMA_VID(pStat->pSma));
// step 1: set persistence task cancelled
atomic_store_8(RSMA_TRIGGER_STAT(pStat), TASK_TRIGGER_STAT_CANCELLED);
// step 2: stop the persistence timer
taosTmrStopA(&RSMA_TMR_ID(pStat));
// step 3: wait the persistence thread to finish
int32_t nLoops = 0;
if (atomic_load_8(RSMA_RUNNING_STAT(pStat)) == 1) {
while (1) {
if (atomic_load_8(RSMA_TRIGGER_STAT(pStat)) == TASK_TRIGGER_STAT_FINISHED) {
break;
} else {
smaDebug("not destroyed since rsma stat in %" PRIi8, atomic_load_8(RSMA_TRIGGER_STAT(pStat)));
}
++nLoops;
if (nLoops > 1000) {
sched_yield();
nLoops = 0;
}
}
}
// step 4: destroy the rsma info and associated fetch tasks
// TODO: use taosHashSetFreeFp when taosHashSetFreeFp is ready.
void *infoHash = taosHashIterate(RSMA_INFO_HASH(pStat), NULL);
while (infoHash) {
SRSmaInfo *pSmaInfo = *(SRSmaInfo **)infoHash;
tdFreeRSmaInfo(pSmaInfo);
infoHash = taosHashIterate(RSMA_INFO_HASH(pStat), infoHash);
}
taosHashCleanup(RSMA_INFO_HASH(pStat));
// step 5: wait all triggered fetch tasks finished
nLoops = 0;
while (1) {
if (T_REF_VAL_GET((SSmaStat *)pStat) == 0) {
break;
}
++nLoops;
if (nLoops > 1000) {
sched_yield();
nLoops = 0;
}
}
// step 6: cleanup the timer handle
if (RSMA_TMR_HANDLE(pStat)) {
taosTmrCleanUp(RSMA_TMR_HANDLE(pStat));
}
}
}
static void *tdFreeRSmaStat(SRSmaStat *pStat) {
tdDestroyRSmaStat(pStat);
taosMemoryFreeClear(pStat);
return NULL;
}
@ -179,19 +246,9 @@ void *tdFreeSmaState(SSmaStat *pSmaStat, int8_t smaType) {
int32_t tdDestroySmaState(SSmaStat *pSmaStat, int8_t smaType) {
if (pSmaStat) {
if (smaType == TSDB_SMA_TYPE_TIME_RANGE) {
tdFreeTSmaStat(&pSmaStat->tsmaStat);
tdDestroyTSmaStat(SMA_TSMA_STAT(pSmaStat));
} else if (smaType == TSDB_SMA_TYPE_ROLLUP) {
if (SMA_RSMA_TMR_HANDLE(pSmaStat)) {
taosTmrCleanUp(SMA_RSMA_TMR_HANDLE(pSmaStat));
}
// TODO: use taosHashSetFreeFp when taosHashSetFreeFp is ready.
void *infoHash = taosHashIterate(SMA_RSMA_INFO_HASH(pSmaStat), NULL);
while (infoHash) {
SRSmaInfo *pInfoHash = *(SRSmaInfo **)infoHash;
tdFreeRSmaInfo(pInfoHash);
infoHash = taosHashIterate(SMA_RSMA_INFO_HASH(pSmaStat), infoHash);
}
taosHashCleanup(SMA_RSMA_INFO_HASH(pSmaStat));
tdDestroyRSmaStat(SMA_RSMA_STAT(pSmaStat));
} else {
ASSERT(0);
}
@ -259,35 +316,4 @@ int32_t tdCheckAndInitSmaEnv(SSma *pSma, int8_t smaType) {
tdUnLockSma(pSma);
return TSDB_CODE_SUCCESS;
};
int32_t smaTimerInit(void **timer, int8_t *initFlag, const char *label) {
int8_t old;
while (1) {
old = atomic_val_compare_exchange_8(initFlag, 0, 2);
if (old != 2) break;
}
if (old == 0) {
*timer = taosTmrInit(10000, 100, 10000, label);
if (!(*timer)) {
atomic_store_8(initFlag, 0);
return -1;
}
atomic_store_8(initFlag, 1);
}
return 0;
}
void smaTimerCleanUp(void *timer, int8_t *initFlag) {
int8_t old;
while (1) {
old = atomic_val_compare_exchange_8(initFlag, 1, 2);
if (old != 2) break;
}
if (old == 1) {
taosTmrCleanUp(timer);
atomic_store_8(initFlag, 0);
}
}
};

70
source/dnode/vnode/src/sma/smaOpen.c
View File

@ -18,6 +18,7 @@
static int32_t smaEvalDays(SRetention *r, int8_t precision);
static int32_t smaSetKeepCfg(STsdbKeepCfg *pKeepCfg, STsdbCfg *pCfg, int type);
static int32_t rsmaRestore(SSma *pSma);
#define SMA_SET_KEEP_CFG(l) \
do { \
@ -100,6 +101,9 @@ int32_t smaOpen(SVnode *pVnode) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return -1;
}
pVnode->pSma = pSma;
pSma->pVnode = pVnode;
taosThreadMutexInit(&pSma->mutex, NULL);
pSma->locked = false;
@ -117,37 +121,83 @@ int32_t smaOpen(SVnode *pVnode) {
ASSERT(0);
}
}
// restore the rsma
if (rsmaRestore(pSma) < 0) {
goto _err;
}
}
pVnode->pSma = pSma;
return 0;
_err:
taosMemoryFreeClear(pSma);
return -1;
}
int32_t smaClose(SSma *pSma) {
int32_t smaCloseEnv(SSma *pSma) {
if (pSma) {
SMA_TSMA_ENV(pSma) = tdFreeSmaEnv(SMA_TSMA_ENV(pSma));
SMA_RSMA_ENV(pSma) = tdFreeSmaEnv(SMA_RSMA_ENV(pSma));
}
return 0;
}
int32_t smaCloseEx(SSma *pSma) {
if (pSma) {
taosThreadMutexDestroy(&pSma->mutex);
if SMA_RSMA_TSDB0 (pSma) tsdbClose(&SMA_RSMA_TSDB0(pSma));
if SMA_RSMA_TSDB1 (pSma) tsdbClose(&SMA_RSMA_TSDB1(pSma));
if SMA_RSMA_TSDB2 (pSma) tsdbClose(&SMA_RSMA_TSDB2(pSma));
// SMA_TSMA_ENV(pSma) = tdFreeSmaEnv(SMA_TSMA_ENV(pSma));
// SMA_RSMA_ENV(pSma) = tdFreeSmaEnv(SMA_RSMA_ENV(pSma));
taosMemoryFreeClear(pSma);
}
return 0;
}
int32_t smaClose(SSma *pSma) {
smaCloseEnv(pSma);
smaCloseEx(pSma);
return 0;
}
/**
* @brief rsma env restore
*
* @param pSma
* @return int32_t
*
* @param pSma
* @return int32_t
*/
int32_t smaRestore(SSma *pSma) {
if (!pSma) return 0;
static int32_t rsmaRestore(SSma *pSma) {
ASSERT(VND_IS_RSMA(pSma->pVnode));
// iterate all stables to restore the rsma env
SArray *suidList = taosArrayInit(1, sizeof(tb_uid_t));
if (tsdbGetStbIdList(SMA_META(pSma), 0, suidList) < 0) {
smaError("failed to restore rsma since get stb id list error: %s", terrstr());
return TSDB_CODE_FAILED;
}
SMetaReader mr = {0};
metaReaderInit(&mr, SMA_META(pSma), 0);
for (int32_t i = 0; i < taosArrayGetSize(suidList); ++i) {
tb_uid_t suid = *(tb_uid_t *)taosArrayGet(suidList, i);
smaDebug("suid [%d] is %" PRIi64, i, suid);
if (metaGetTableEntryByUid(&mr, suid) < 0) {
metaReaderClear(&mr);
taosArrayDestroy(suidList);
smaError("failed to get table meta for %" PRIi64 " since %s", suid, terrstr());
return TSDB_CODE_FAILED;
}
ASSERT(mr.me.type == TSDB_SUPER_TABLE);
if (TABLE_IS_ROLLUP(mr.me.flags)) {
SRSmaParam *param = &mr.me.stbEntry.rsmaParam;
for (int i = 0; i < 2; ++i) {
smaDebug("vgId: %d table:%" PRIi64 " maxdelay[%d]:%" PRIi64 " watermark[%d]:%" PRIi64, TD_VID(pSma->pVnode),
suid, i, param->maxdelay[i], i, param->watermark[i]);
}
}
}
metaReaderClear(&mr);
taosArrayDestroy(suidList);
return TSDB_CODE_SUCCESS;
}

372
source/dnode/vnode/src/sma/smaRollup.c
View File

@ -14,8 +14,8 @@
*/
#include "sma.h"
#include "tstream.h"
#define RSMA_QTASK_PERSIST_MS 7200000
typedef enum { TD_QTASK_TMP_FILE = 0, TD_QTASK_CUR_FILE } TD_QTASK_FILE_T;
static const char *tdQTaskInfoFname[] = {"qtaskinfo.t", "qtaskinfo"};
@ -31,13 +31,13 @@ static void tdRSmaPersistTrigger(void *param, void *tmrId);
struct SRSmaInfoItem {
SRSmaInfo *pRsmaInfo;
void *taskInfo; // qTaskInfo_t
void *tmrHandle;
tmr_h tmrId;
int8_t level;
int8_t tmrInitFlag;
int8_t triggerStatus; // TASK_TRIGGER_STATUS__IN_ACTIVE/TASK_TRIGGER_STATUS__ACTIVE
int8_t triggerStat;
int32_t maxDelay;
};
struct SRSmaInfo {
STSchema *pTSchema;
SSma *pSma;
@ -45,11 +45,14 @@ struct SRSmaInfo {
SRSmaInfoItem items[TSDB_RETENTION_L2];
};
static FORCE_INLINE void tdFreeTaskHandle(qTaskInfo_t *taskHandle) {
static FORCE_INLINE void tdFreeTaskHandle(qTaskInfo_t *taskHandle, int32_t vgId, int32_t level) {
// Note: free/kill may in RC
qTaskInfo_t otaskHandle = atomic_load_ptr(taskHandle);
if (otaskHandle && atomic_val_compare_exchange_ptr(taskHandle, otaskHandle, NULL)) {
smaDebug("vgId:%d, free qTaskInfo_t %p of level %d", vgId, otaskHandle, level);
qDestroyTask(otaskHandle);
} else {
smaDebug("vgId:%d, not free qTaskInfo_t %p of level %d", vgId, otaskHandle, level);
}
}
@ -58,14 +61,19 @@ void *tdFreeRSmaInfo(SRSmaInfo *pInfo) {
for (int32_t i = 0; i < TSDB_RETENTION_L2; ++i) {
SRSmaInfoItem *pItem = &pInfo->items[i];
if (pItem->taskInfo) {
tdFreeTaskHandle(pItem->taskInfo);
}
if (pItem->tmrHandle) {
taosTmrCleanUp(pItem->tmrHandle);
smaDebug("vgId:%d, stb %" PRIi64 " stop fetch-timer %p level %d", SMA_VID(pInfo->pSma), pInfo->suid,
pItem->tmrId, i + 1);
taosTmrStopA(&pItem->tmrId);
tdFreeTaskHandle(&pItem->taskInfo, SMA_VID(pInfo->pSma), i + 1);
} else {
smaDebug("vgId:%d, stb %" PRIi64 " no need to destroy rsma info level %d since empty taskInfo",
SMA_VID(pInfo->pSma), pInfo->suid, i + 1);
}
}
taosMemoryFree(pInfo->pTSchema);
taosMemoryFree(pInfo);
} else {
smaDebug("vgId:%d, stb %" PRIi64 " no need to destroy rsma info since empty", SMA_VID(pInfo->pSma), pInfo->suid);
}
return NULL;
@ -81,9 +89,9 @@ static FORCE_INLINE int32_t tdUidStoreInit(STbUidStore **pStore) {
return TSDB_CODE_SUCCESS;
}
static FORCE_INLINE int32_t tdUpdateTbUidListImpl(SSma *pSma, tb_uid_t *suid, SArray *tbUids) {
static int32_t tdUpdateTbUidListImpl(SSma *pSma, tb_uid_t *suid, SArray *tbUids) {
SSmaEnv *pEnv = SMA_RSMA_ENV(pSma);
SSmaStat *pStat = SMA_ENV_STAT(pEnv);
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SRSmaInfo *pRSmaInfo = NULL;
if (!suid || !tbUids) {
@ -92,27 +100,31 @@ static FORCE_INLINE int32_t tdUpdateTbUidListImpl(SSma *pSma, tb_uid_t *suid, SA
return TSDB_CODE_FAILED;
}
pRSmaInfo = taosHashGet(SMA_RSMA_INFO_HASH(pStat), suid, sizeof(tb_uid_t));
pRSmaInfo = taosHashGet(RSMA_INFO_HASH(pStat), suid, sizeof(tb_uid_t));
if (!pRSmaInfo || !(pRSmaInfo = *(SRSmaInfo **)pRSmaInfo)) {
smaError("vgId:%d, failed to get rsma info for uid:%" PRIi64, SMA_VID(pSma), *suid);
terrno = TSDB_CODE_RSMA_INVALID_STAT;
return TSDB_CODE_FAILED;
}
if (pRSmaInfo->items[0].taskInfo && (qUpdateQualifiedTableId(pRSmaInfo->items[0].taskInfo, tbUids, true) < 0)) {
smaError("vgId:%d, update tbUidList failed for uid:%" PRIi64 " since %s", SMA_VID(pSma), *suid, terrstr(terrno));
return TSDB_CODE_FAILED;
} else {
smaDebug("vgId:%d, update tbUidList succeed for qTaskInfo:%p with suid:%" PRIi64 ", uid:%" PRIi64, SMA_VID(pSma),
pRSmaInfo->items[0].taskInfo, *suid, *(int64_t *)taosArrayGet(tbUids, 0));
if (pRSmaInfo->items[0].taskInfo) {
if ((qUpdateQualifiedTableId(pRSmaInfo->items[0].taskInfo, tbUids, true) < 0)) {
smaError("vgId:%d, update tbUidList failed for uid:%" PRIi64 " since %s", SMA_VID(pSma), *suid, terrstr(terrno));
return TSDB_CODE_FAILED;
} else {
smaDebug("vgId:%d, update tbUidList succeed for qTaskInfo:%p with suid:%" PRIi64 ", uid:%" PRIi64, SMA_VID(pSma),
pRSmaInfo->items[0].taskInfo, *suid, *(int64_t *)taosArrayGet(tbUids, 0));
}
}
if (pRSmaInfo->items[1].taskInfo && (qUpdateQualifiedTableId(pRSmaInfo->items[1].taskInfo, tbUids, true) < 0)) {
smaError("vgId:%d, update tbUidList failed for uid:%" PRIi64 " since %s", SMA_VID(pSma), *suid, terrstr(terrno));
return TSDB_CODE_FAILED;
} else {
smaDebug("vgId:%d, update tbUidList succeed for qTaskInfo:%p with suid:%" PRIi64 ", uid:%" PRIi64, SMA_VID(pSma),
pRSmaInfo->items[1].taskInfo, *suid, *(int64_t *)taosArrayGet(tbUids, 0));
if (pRSmaInfo->items[1].taskInfo) {
if ((qUpdateQualifiedTableId(pRSmaInfo->items[1].taskInfo, tbUids, true) < 0)) {
smaError("vgId:%d, update tbUidList failed for uid:%" PRIi64 " since %s", SMA_VID(pSma), *suid, terrstr(terrno));
return TSDB_CODE_FAILED;
} else {
smaDebug("vgId:%d, update tbUidList succeed for qTaskInfo:%p with suid:%" PRIi64 ", uid:%" PRIi64, SMA_VID(pSma),
pRSmaInfo->items[1].taskInfo, *suid, *(int64_t *)taosArrayGet(tbUids, 0));
}
}
return TSDB_CODE_SUCCESS;
@ -159,9 +171,9 @@ int32_t tdFetchTbUidList(SSma *pSma, STbUidStore **ppStore, tb_uid_t suid, tb_ui
return TSDB_CODE_SUCCESS;
}
SSmaStat *pStat = SMA_ENV_STAT(pEnv);
SHashObj *infoHash = NULL;
if (!pStat || !(infoHash = SMA_RSMA_INFO_HASH(pStat))) {
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SHashObj *infoHash = NULL;
if (!pStat || !(infoHash = RSMA_INFO_HASH(pStat))) {
terrno = TSDB_CODE_RSMA_INVALID_STAT;
return TSDB_CODE_FAILED;
}
@ -195,11 +207,11 @@ static int32_t tdSetRSmaInfoItemParams(SSma *pSma, SRSmaParam *param, SRSmaInfo
if (param->qmsg[idx]) {
SRSmaInfoItem *pItem = &(pRSmaInfo->items[idx]);
pItem->pRsmaInfo = pRSmaInfo;
pItem->taskInfo = qCreateStreamExecTaskInfo(param->qmsg[0], pReadHandle);
pItem->taskInfo = qCreateStreamExecTaskInfo(param->qmsg[idx], pReadHandle);
if (!pItem->taskInfo) {
goto _err;
}
pItem->triggerStatus = TASK_TRIGGER_STATUS__IN_ACTIVE;
pItem->triggerStat = TASK_TRIGGER_STAT_INACTIVE;
if (param->maxdelay[idx] < TSDB_MIN_ROLLUP_MAX_DELAY) {
int64_t msInterval =
convertTimeFromPrecisionToUnit(pRetention[idx + 1].freq, pTsdbCfg->precision, TIME_UNIT_MILLISECOND);
@ -211,10 +223,6 @@ static int32_t tdSetRSmaInfoItemParams(SSma *pSma, SRSmaParam *param, SRSmaInfo
pItem->maxDelay = TSDB_MAX_ROLLUP_MAX_DELAY;
}
pItem->level = (idx == 0 ? TSDB_RETENTION_L1 : TSDB_RETENTION_L2);
pItem->tmrHandle = taosTmrInit(10000, 100, 10000, "RSMA");
if (!pItem->tmrHandle) {
goto _err;
}
}
return TSDB_CODE_SUCCESS;
_err:
@ -238,7 +246,7 @@ int32_t tdProcessRSmaCreate(SVnode *pVnode, SVCreateStbReq *pReq) {
SMeta *pMeta = pVnode->pMeta;
SMsgCb *pMsgCb = &pVnode->msgCb;
SRSmaParam *param = &pReq->pRSmaParam;
SRSmaParam *param = &pReq->rsmaParam;
if ((param->qmsgLen[0] == 0) && (param->qmsgLen[1] == 0)) {
smaWarn("vgId:%d, no qmsg1/qmsg2 for rollup stable %s %" PRIi64, SMA_VID(pSma), pReq->name, pReq->suid);
@ -251,10 +259,10 @@ int32_t tdProcessRSmaCreate(SVnode *pVnode, SVCreateStbReq *pReq) {
}
SSmaEnv *pEnv = SMA_RSMA_ENV(pSma);
SSmaStat *pStat = SMA_ENV_STAT(pEnv);
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SRSmaInfo *pRSmaInfo = NULL;
pRSmaInfo = taosHashGet(SMA_RSMA_INFO_HASH(pStat), &pReq->suid, sizeof(tb_uid_t));
pRSmaInfo = taosHashGet(RSMA_INFO_HASH(pStat), &pReq->suid, sizeof(tb_uid_t));
if (pRSmaInfo) {
ASSERT(0); // TODO: free original pRSmaInfo is exists abnormally
smaWarn("vgId:%d, rsma info already exists for stb: %s, %" PRIi64, SMA_VID(pSma), pReq->name, pReq->suid);
@ -293,16 +301,23 @@ int32_t tdProcessRSmaCreate(SVnode *pVnode, SVCreateStbReq *pReq) {
if (tdSetRSmaInfoItemParams(pSma, param, pRSmaInfo, &handle, 0) < 0) {
goto _err;
}
if (tdSetRSmaInfoItemParams(pSma, param, pRSmaInfo, &handle, 1) < 0) {
goto _err;
}
if (taosHashPut(SMA_RSMA_INFO_HASH(pStat), &pReq->suid, sizeof(tb_uid_t), &pRSmaInfo, sizeof(pRSmaInfo)) < 0) {
if (taosHashPut(RSMA_INFO_HASH(pStat), &pReq->suid, sizeof(tb_uid_t), &pRSmaInfo, sizeof(pRSmaInfo)) < 0) {
goto _err;
} else {
smaDebug("vgId:%d, register rsma info succeed for suid:%" PRIi64, SMA_VID(pSma), pReq->suid);
}
// start the persist timer
if (TASK_TRIGGER_STAT_INIT ==
atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pStat), TASK_TRIGGER_STAT_INIT, TASK_TRIGGER_STAT_ACTIVE)) {
taosTmrStart(tdRSmaPersistTrigger, RSMA_QTASK_PERSIST_MS, pStat, RSMA_TMR_HANDLE(pStat));
}
return TSDB_CODE_SUCCESS;
_err:
tdFreeRSmaInfo(pRSmaInfo);
@ -432,6 +447,16 @@ static int32_t tdFetchSubmitReqSuids(SSubmitReq *pMsg, STbUidStore *pStore) {
return 0;
}
static void tdDestroySDataBlockArray(SArray *pArray) {
#if 0
for (int32_t i = 0; i < taosArrayGetSize(pArray); ++i) {
SSDataBlock *pDataBlock = taosArrayGet(pArray, i);
blockDestroyInner(pDataBlock);
}
#endif
taosArrayDestroy(pArray);
}
static int32_t tdFetchAndSubmitRSmaResult(SRSmaInfoItem *pItem, int8_t blkType) {
SArray *pResult = NULL;
SRSmaInfo *pRSmaInfo = pItem->pRsmaInfo;
@ -462,10 +487,11 @@ static int32_t tdFetchAndSubmitRSmaResult(SRSmaInfoItem *pItem, int8_t blkType)
#if 1
char flag[10] = {0};
snprintf(flag, 10, "level %" PRIi8, pItem->level);
blockDebugShowData(pResult, flag);
blockDebugShowDataBlocks(pResult, flag);
#endif
STsdb *sinkTsdb = (pItem->level == TSDB_RETENTION_L1 ? pSma->pRSmaTsdb1 : pSma->pRSmaTsdb2);
SSubmitReq *pReq = NULL;
// TODO: the schema update should be handled
if (buildSubmitReqFromDataBlock(&pReq, pResult, pRSmaInfo->pTSchema, SMA_VID(pSma), pRSmaInfo->suid) < 0) {
goto _err;
}
@ -476,18 +502,16 @@ static int32_t tdFetchAndSubmitRSmaResult(SRSmaInfoItem *pItem, int8_t blkType)
}
taosMemoryFreeClear(pReq);
} else if (terrno == 0) {
smaDebug("vgId:%d, no rsma %" PRIi8 " data fetched yet", SMA_VID(pSma), pItem->level);
} else {
smaDebug("vgId:%d, no rsma % " PRIi8 " data generated since %s", SMA_VID(pSma), pItem->level, tstrerror(terrno));
smaDebug("vgId:%d, no rsma %" PRIi8 " data fetched since %s", SMA_VID(pSma), pItem->level, tstrerror(terrno));
}
if (blkType == STREAM_DATA_TYPE_SUBMIT_BLOCK) {
atomic_store_8(&pItem->triggerStatus, TASK_TRIGGER_STATUS__ACTIVE);
}
taosArrayDestroy(pResult);
tdDestroySDataBlockArray(pResult);
return TSDB_CODE_SUCCESS;
_err:
taosArrayDestroy(pResult);
tdDestroySDataBlockArray(pResult);
return TSDB_CODE_FAILED;
}
@ -499,26 +523,38 @@ _err:
*/
static void tdRSmaFetchTrigger(void *param, void *tmrId) {
SRSmaInfoItem *pItem = param;
SSma *pSma = pItem->pRsmaInfo->pSma;
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT((SSmaEnv *)pSma->pRSmaEnv);
if (atomic_load_8(&pItem->triggerStatus) == TASK_TRIGGER_STATUS__ACTIVE) {
smaWarn("%s:%d THREAD:%" PRIi64 " level %" PRIi8 " status is active for tb suid:%" PRIi64, __func__, __LINE__,
taosGetSelfPthreadId(), pItem->level, pItem->pRsmaInfo->suid);
SSDataBlock dataBlock = {.info.type = STREAM_GET_ALL};
atomic_store_8(&pItem->triggerStatus, TASK_TRIGGER_STATUS__IN_ACTIVE);
qSetStreamInput(pItem->taskInfo, &dataBlock, STREAM_DATA_TYPE_SSDATA_BLOCK, false);
tdFetchAndSubmitRSmaResult(pItem, STREAM_DATA_TYPE_SSDATA_BLOCK);
} else {
smaWarn("%s:%d THREAD:%" PRIi64 " level %" PRIi8 " status is inactive for tb suid:%" PRIi64, __func__, __LINE__,
taosGetSelfPthreadId(), pItem->level, pItem->pRsmaInfo->suid);
int8_t rsmaTriggerStat = atomic_load_8(RSMA_TRIGGER_STAT(pStat));
if (rsmaTriggerStat == TASK_TRIGGER_STAT_CANCELLED || rsmaTriggerStat == TASK_TRIGGER_STAT_FINISHED) {
smaDebug("vgId:%d, level %" PRIi8 " not fetch since stat is cancelled for table suid:%" PRIi64, SMA_VID(pSma),
pItem->level, pItem->pRsmaInfo->suid);
return;
}
// taosTmrReset(tdRSmaFetchTrigger, pItem->maxDelay, pItem, pItem->tmrHandle, &pItem->tmrId);
int8_t fetchTriggerStat =
atomic_val_compare_exchange_8(&pItem->triggerStat, TASK_TRIGGER_STAT_ACTIVE, TASK_TRIGGER_STAT_INACTIVE);
if (fetchTriggerStat == TASK_TRIGGER_STAT_ACTIVE) {
smaDebug("vgId:%d, level %" PRIi8 " stat is active for table suid:%" PRIi64, SMA_VID(pSma), pItem->level,
pItem->pRsmaInfo->suid);
tdRefSmaStat(pSma, (SSmaStat *)pStat);
SSDataBlock dataBlock = {.info.type = STREAM_GET_ALL};
qSetStreamInput(pItem->taskInfo, &dataBlock, STREAM_DATA_TYPE_SSDATA_BLOCK, false);
tdFetchAndSubmitRSmaResult(pItem, STREAM_DATA_TYPE_SSDATA_BLOCK);
tdUnRefSmaStat(pSma, (SSmaStat *)pStat);
} else {
smaDebug("vgId:%d, level %" PRIi8 " stat is inactive for table suid:%" PRIi64, SMA_VID(pSma), pItem->level,
pItem->pRsmaInfo->suid);
}
}
static FORCE_INLINE int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int32_t inputType, SRSmaInfoItem *pItem,
tb_uid_t suid, int8_t level) {
static int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int32_t inputType, SRSmaInfoItem *pItem, tb_uid_t suid,
int8_t level) {
if (!pItem || !pItem->taskInfo) {
smaDebug("vgId:%d, no qTaskInfo to execute rsma %" PRIi8 " task for suid:%" PRIu64, SMA_VID(pSma), level, suid);
return TSDB_CODE_SUCCESS;
@ -533,14 +569,17 @@ static FORCE_INLINE int32_t tdExecuteRSmaImpl(SSma *pSma, const void *pMsg, int3
}
tdFetchAndSubmitRSmaResult(pItem, STREAM_DATA_TYPE_SUBMIT_BLOCK);
atomic_store_8(&pItem->triggerStatus, TASK_TRIGGER_STATUS__ACTIVE);
smaWarn("%s:%d THREAD:%" PRIi64 " process rsma insert", __func__, __LINE__, taosGetSelfPthreadId());
atomic_store_8(&pItem->triggerStat, TASK_TRIGGER_STAT_ACTIVE);
smaDebug("vgId:%d, process rsma insert", SMA_VID(pSma));
SSmaEnv *pEnv = SMA_RSMA_ENV(pSma);
SRSmaStat *pStat = SMA_RSMA_STAT(pEnv->pStat);
taosTmrStart(tdRSmaPersistTrigger, 5000, pStat, pStat->tmrHandle);
taosTmrReset(tdRSmaFetchTrigger, pItem->maxDelay, pItem, pItem->tmrHandle, &pItem->tmrId);
if (pStat->tmrHandle) {
taosTmrReset(tdRSmaFetchTrigger, pItem->maxDelay, pItem, pStat->tmrHandle, &pItem->tmrId);
} else {
ASSERT(0);
}
return TSDB_CODE_SUCCESS;
}
@ -552,10 +591,10 @@ static int32_t tdExecuteRSma(SSma *pSma, const void *pMsg, int32_t inputType, tb
return TSDB_CODE_SUCCESS;
}
SSmaStat *pStat = SMA_ENV_STAT(pEnv);
SRSmaStat *pStat = (SRSmaStat *)SMA_ENV_STAT(pEnv);
SRSmaInfo *pRSmaInfo = NULL;
pRSmaInfo = taosHashGet(SMA_RSMA_INFO_HASH(pStat), &suid, sizeof(tb_uid_t));
pRSmaInfo = taosHashGet(RSMA_INFO_HASH(pStat), &suid, sizeof(tb_uid_t));
if (!pRSmaInfo || !(pRSmaInfo = *(SRSmaInfo **)pRSmaInfo)) {
smaDebug("vgId:%d, return as no rsma info for suid:%" PRIu64, SMA_VID(pSma), suid);
@ -608,7 +647,7 @@ int32_t tdProcessRSmaSubmit(SSma *pSma, void *pMsg, int32_t inputType) {
return TSDB_CODE_SUCCESS;
}
void tdRSmaQTaskGetFName(int32_t vid, int8_t ftype, char* outputName) {
void tdRSmaQTaskGetFName(int32_t vid, int8_t ftype, char *outputName) {
tdGetVndFileName(vid, "rsma", tdQTaskInfoFname[ftype], outputName);
}
@ -618,6 +657,11 @@ static void *tdRSmaPersistExec(void *param) {
SSma *pSma = pRSmaStat->pSma;
STfs *pTfs = pSma->pVnode->pTfs;
int64_t toffset = 0;
bool isFileCreated = false;
if (TASK_TRIGGER_STAT_CANCELLED == atomic_load_8(RSMA_TRIGGER_STAT(pRSmaStat))) {
goto _end;
}
void *infoHash = taosHashIterate(RSMA_INFO_HASH(pRSmaStat), NULL);
if (!infoHash) {
@ -625,67 +669,142 @@ static void *tdRSmaPersistExec(void *param) {
}
STFile tFile = {0};
int32_t vid = 2;
char qTaskInfoFName[TSDB_FILENAME_LEN];
tdRSmaQTaskGetFName(vid, TD_QTASK_TMP_FILE, qTaskInfoFName);
tdInitTFile(&tFile, pTfs, qTaskInfoFName);
tdCreateTFile(&tFile, pTfs, true, -1);
int32_t vid = SMA_VID(pSma);
while (infoHash) {
SRSmaInfo *pRSmaInfo = *(SRSmaInfo **)infoHash;
char *pOutput = NULL;
int32_t len = 0;
if (qSerializeTaskStatus(pRSmaInfo->items[0].taskInfo, &pOutput, &len) < 0) {
smaError("serialize rsma task for table %" PRIi64 " failed since %s", pRSmaInfo->items[0].pRsmaInfo->suid,
terrstr(terrno));
} else {
smaWarn("serialize rsma task for table %" PRIi64 " success and len is %d", pRSmaInfo->items[0].pRsmaInfo->suid,
len);
}
tdAppendTFile(&tFile, &len, sizeof(len), &toffset);
tdAppendTFile(&tFile, pOutput, len, &toffset);
taosMemoryFree(pOutput);
#if 0
smaDebug("table %" PRIi64 " sleep 15s start ...", pRSmaInfo->items[0].pRsmaInfo->suid);
for (int32_t i = 15; i > 0; --i) {
taosSsleep(1);
smaDebug("table %" PRIi64 " countdown %d", pRSmaInfo->items[0].pRsmaInfo->suid, i);
}
smaDebug("table %" PRIi64 " sleep 15s end ...", pRSmaInfo->items[0].pRsmaInfo->suid);
#endif
for (int32_t i = 0; i < TSDB_RETENTION_L2; ++i) {
qTaskInfo_t taskInfo = pRSmaInfo->items[i].taskInfo;
if (!taskInfo) {
smaDebug("vgId:%d, table %" PRIi64 " level %d qTaskInfo is NULL", vid, pRSmaInfo->suid, i + 1);
continue;
}
char *pOutput = NULL;
int32_t len = 0;
int8_t type = 0;
if (qSerializeTaskStatus(taskInfo, &pOutput, &len) < 0) {
smaError("vgId:%d, table %" PRIi64 " level %d serialize rsma task failed since %s", vid, pRSmaInfo->suid, i + 1,
terrstr(terrno));
goto _err;
} else {
if (!pOutput) {
smaDebug("vgId:%d, table %" PRIi64
" level %d serialize rsma task success but no output(len %d) and no need to persist",
vid, pRSmaInfo->suid, i + 1, len);
continue;
} else if (len <= 0) {
smaDebug("vgId:%d, table %" PRIi64 " level %d serialize rsma task success with len %d and no need to persist",
vid, pRSmaInfo->suid, i + 1, len);
taosMemoryFree(pOutput);
}
smaDebug("vgId:%d, table %" PRIi64 " level %d serialize rsma task success with len %d and need persist", vid,
pRSmaInfo->suid, i + 1, len);
#if 1
if (qDeserializeTaskStatus(taskInfo, pOutput, len) < 0) {
smaError("vgId:%d, table %" PRIi64 "level %d deserialize rsma task failed since %s", vid, pRSmaInfo->suid,
i + 1, terrstr(terrno));
} else {
smaDebug("vgId:%d, table %" PRIi64 " level %d deserialize rsma task success", vid, pRSmaInfo->suid, i + 1);
}
#endif
}
if (!isFileCreated) {
char qTaskInfoFName[TSDB_FILENAME_LEN];
tdRSmaQTaskGetFName(vid, TD_QTASK_TMP_FILE, qTaskInfoFName);
tdInitTFile(&tFile, pTfs, qTaskInfoFName);
tdCreateTFile(&tFile, pTfs, true, -1);
isFileCreated = true;
}
len += (sizeof(len) + sizeof(pRSmaInfo->suid));
tdAppendTFile(&tFile, &len, sizeof(len), &toffset);
tdAppendTFile(&tFile, &pRSmaInfo->suid, sizeof(pRSmaInfo->suid), &toffset);
tdAppendTFile(&tFile, pOutput, len, &toffset);
taosMemoryFree(pOutput);
}
infoHash = taosHashIterate(RSMA_INFO_HASH(pRSmaStat), infoHash);
}
_end:
_normal:
if (isFileCreated) {
if (tdUpdateTFileHeader(&tFile) < 0) {
smaError("vgId:%d, failed to update tfile %s header since %s", vid, TD_FILE_FULL_NAME(&tFile), tstrerror(terrno));
tdCloseTFile(&tFile);
tdRemoveTFile(&tFile);
goto _err;
} else {
smaDebug("vgId:%d, succeed to update tfile %s header", vid, TD_FILE_FULL_NAME(&tFile));
}
if (tdUpdateTFileHeader(&tFile) < 0) {
smaError("vgId:%d, failed to update tfile %s header since %s", vid, TD_FILE_FULL_NAME(&tFile), tstrerror(terrno));
tdCloseTFile(&tFile);
tdRemoveTFile(&tFile);
return NULL;
char newFName[TSDB_FILENAME_LEN];
strncpy(newFName, TD_FILE_FULL_NAME(&tFile), TSDB_FILENAME_LEN);
char *pos = strstr(newFName, tdQTaskInfoFname[TD_QTASK_TMP_FILE]);
strncpy(pos, tdQTaskInfoFname[TD_QTASK_CUR_FILE], TSDB_FILENAME_LEN - POINTER_DISTANCE(pos, newFName));
if (taosRenameFile(TD_FILE_FULL_NAME(&tFile), newFName) != 0) {
smaError("vgId:%d, failed to rename %s to %s", vid, TD_FILE_FULL_NAME(&tFile), newFName);
goto _err;
} else {
smaDebug("vgId:%d, succeed to rename %s to %s", vid, TD_FILE_FULL_NAME(&tFile), newFName);
}
}
tdCloseTFile(&tFile);
char newFName[TSDB_FILENAME_LEN];
strncpy(newFName, TD_FILE_FULL_NAME(&tFile), TSDB_FILENAME_LEN);
char *pos = strstr(newFName, tdQTaskInfoFname[TD_QTASK_TMP_FILE]);
strncpy(pos, tdQTaskInfoFname[TD_QTASK_CUR_FILE], TSDB_FILENAME_LEN - POINTER_DISTANCE(pos, newFName));
taosRenameFile(TD_FILE_FULL_NAME(&tFile), newFName);
atomic_store_8(&pRSmaStat->tmrStat, TASK_TRIGGER_STATUS__ACTIVE);
return NULL;
goto _end;
_err:
atomic_store_8(&pRSmaStat->tmrStat, TASK_TRIGGER_STATUS__ACTIVE);
// remove the .tmp file
if (isFileCreated) {
tdRemoveTFile(&tFile);
}
_end:
if (TASK_TRIGGER_STAT_INACTIVE == atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat),
TASK_TRIGGER_STAT_INACTIVE,
TASK_TRIGGER_STAT_ACTIVE)) {
smaDebug("vgId:%d, persist task is active again", vid);
} else if (TASK_TRIGGER_STAT_CANCELLED == atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat),
TASK_TRIGGER_STAT_CANCELLED,
TASK_TRIGGER_STAT_FINISHED)) {
smaDebug("vgId:%d, persist task is cancelled", vid);
} else {
smaWarn("vgId:%d, persist task in abnormal stat %" PRIi8, vid, atomic_load_8(RSMA_TRIGGER_STAT(pRSmaStat)));
ASSERT(0);
}
atomic_store_8(RSMA_RUNNING_STAT(pRSmaStat), 0);
taosThreadExit(NULL);
return NULL;
}
static void tdRSmaPersistTask(SRSmaStat *pRSmaStat) {
smaWarn("%s:%d entry ", __func__, __LINE__);
TdThread threadId;
TdThreadAttr thAttr;
taosThreadAttrInit(&thAttr);
taosThreadAttrSetDetachState(&thAttr, PTHREAD_CREATE_DETACHED);
TdThread tid;
if (taosThreadCreate(&threadId, &thAttr, tdRSmaPersistExec, pRSmaStat) != 0) {
smaError("failed to create thread to persist rsma qTaskInfo since %s", strerror(errno));
if (taosThreadCreate(&tid, &thAttr, tdRSmaPersistExec, pRSmaStat) != 0) {
if (TASK_TRIGGER_STAT_INACTIVE == atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat),
TASK_TRIGGER_STAT_INACTIVE,
TASK_TRIGGER_STAT_ACTIVE)) {
smaDebug("persist task is active again");
} else if (TASK_TRIGGER_STAT_CANCELLED == atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat),
TASK_TRIGGER_STAT_CANCELLED,
TASK_TRIGGER_STAT_FINISHED)) {
smaDebug(" persist task is cancelled and set finished");
} else {
smaWarn("persist task in abnormal stat %" PRIi8, atomic_load_8(RSMA_TRIGGER_STAT(pRSmaStat)));
ASSERT(0);
}
atomic_store_8(RSMA_RUNNING_STAT(pRSmaStat), 0);
}
taosThreadAttrDestroy(&thAttr);
smaWarn("%s:%d end ", __func__, __LINE__);
}
/**
@ -696,17 +815,34 @@ static void tdRSmaPersistTask(SRSmaStat *pRSmaStat) {
*/
static void tdRSmaPersistTrigger(void *param, void *tmrId) {
SRSmaStat *pRSmaStat = param;
if (atomic_load_8(&pRSmaStat->tmrStat) == TASK_TRIGGER_STATUS__ACTIVE) {
smaWarn("%s:%d THREAD:%" PRIi64 " rsma persistence start since active", __func__, __LINE__, taosGetSelfPthreadId());
atomic_store_8(&pRSmaStat->tmrStat, TASK_TRIGGER_STATUS__IN_ACTIVE);
// execution
tdRSmaPersistTask(pRSmaStat);
} else {
smaWarn("%s:%d THREAD:%" PRIi64 " rsma persistence not start since inactive", __func__, __LINE__,
taosGetSelfPthreadId());
int8_t tmrStat =
atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_ACTIVE, TASK_TRIGGER_STAT_INACTIVE);
switch (tmrStat) {
case TASK_TRIGGER_STAT_ACTIVE: {
atomic_store_8(RSMA_RUNNING_STAT(pRSmaStat), 1);
if (TASK_TRIGGER_STAT_CANCELLED != atomic_val_compare_exchange_8(RSMA_TRIGGER_STAT(pRSmaStat),
TASK_TRIGGER_STAT_CANCELLED,
TASK_TRIGGER_STAT_FINISHED)) {
smaDebug("rsma persistence start since active");
tdRSmaPersistTask(pRSmaStat);
taosTmrReset(tdRSmaPersistTrigger, RSMA_QTASK_PERSIST_MS, pRSmaStat, pRSmaStat->tmrHandle, &pRSmaStat->tmrId);
} else {
atomic_store_8(RSMA_RUNNING_STAT(pRSmaStat), 0);
}
} break;
case TASK_TRIGGER_STAT_CANCELLED: {
atomic_store_8(RSMA_TRIGGER_STAT(pRSmaStat), TASK_TRIGGER_STAT_FINISHED);
smaDebug("rsma persistence not start since cancelled and finished");
} break;
case TASK_TRIGGER_STAT_INACTIVE: {
smaDebug("rsma persistence not start since inactive");
} break;
case TASK_TRIGGER_STAT_INIT: {
smaDebug("rsma persistence not start since init");
} break;
default: {
smaWarn("rsma persistence not start since unknown stat %" PRIi8, tmrStat);
ASSERT(0);
} break;
}
taosTmrReset(tdRSmaPersistTrigger, 3600000, pRSmaStat, pRSmaStat->tmrHandle, &pRSmaStat->tmrId);
}

11
source/dnode/vnode/src/sma/smaUtil.c
View File

@ -15,6 +15,8 @@
#include "sma.h"
// smaFileUtil ================
#define TD_FILE_HEAD_SIZE 512
#define TD_FILE_STATE_OK 0
@ -32,7 +34,7 @@ static int32_t tdEncodeTFInfo(void **buf, STFInfo *pInfo) {
tlen += taosEncodeFixedU32(buf, pInfo->magic);
tlen += taosEncodeFixedU32(buf, pInfo->ftype);
tlen += taosEncodeFixedU32(buf, pInfo->fver);
tlen += taosEncodeFixedU64(buf, pInfo->fsize);
tlen += taosEncodeFixedI64(buf, pInfo->fsize);
return tlen;
}
@ -41,7 +43,7 @@ static void *tdDecodeTFInfo(void *buf, STFInfo *pInfo) {
buf = taosDecodeFixedU32(buf, &(pInfo->magic));
buf = taosDecodeFixedU32(buf, &(pInfo->ftype));
buf = taosDecodeFixedU32(buf, &(pInfo->fver));
buf = taosDecodeFixedU64(buf, &(pInfo->fsize));
buf = taosDecodeFixedI64(buf, &(pInfo->fsize));
return buf;
}
@ -235,4 +237,7 @@ int32_t tdCreateTFile(STFile *pTFile, STfs *pTfs, bool updateHeader, int8_t fTyp
return 0;
}
int32_t tdRemoveTFile(STFile *pTFile) { return tfsRemoveFile(TD_FILE_F(pTFile)); }
int32_t tdRemoveTFile(STFile *pTFile) { return tfsRemoveFile(TD_FILE_F(pTFile)); }
// smaXXXUtil ================
// ...

4
source/dnode/vnode/src/tq/tqExec.c
View File

@ -112,7 +112,7 @@ int32_t tqDataExec(STQ* pTq, STqExecHandle* pExec, SSubmitReq* pReq, SMqDataBlkR
tqReadHandleSetMsg(pReader, pReq, 0);
while (tqNextDataBlock(pReader)) {
SSDataBlock block = {0};
if (tqRetrieveDataBlock(&block, pReader, &block.info.groupId, &block.info.uid, &block.info.rows) < 0) {
if (tqRetrieveDataBlock(&block, pReader) < 0) {
if (terrno == TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND) continue;
ASSERT(0);
}
@ -129,7 +129,7 @@ int32_t tqDataExec(STQ* pTq, STqExecHandle* pExec, SSubmitReq* pReq, SMqDataBlkR
tqReadHandleSetMsg(pReader, pReq, 0);
while (tqNextDataBlockFilterOut(pReader, pExec->execDb.pFilterOutTbUid)) {
SSDataBlock block = {0};
if (tqRetrieveDataBlock(&block, pReader, &block.info.groupId, &block.info.uid, &block.info.rows) < 0) {
if (tqRetrieveDataBlock(&block, pReader) < 0) {
if (terrno == TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND) continue;
ASSERT(0);
}

22
source/dnode/vnode/src/tq/tqRead.c
View File

@ -109,11 +109,15 @@ int32_t tqReadHandleSetMsg(STqReadHandle* pReadHandle, SSubmitReq* pMsg, int64_t
}
bool tqNextDataBlock(STqReadHandle* pHandle) {
if (pHandle->pMsg == NULL) return false;
while (1) {
if (tGetSubmitMsgNext(&pHandle->msgIter, &pHandle->pBlock) < 0) {
return false;
}
if (pHandle->pBlock == NULL) return false;
if (pHandle->pBlock == NULL) {
pHandle->pMsg = NULL;
return false;
}
if (pHandle->tbIdHash == NULL) {
return true;
@ -142,10 +146,7 @@ bool tqNextDataBlockFilterOut(STqReadHandle* pHandle, SHashObj* filterOutUids) {
return false;
}
int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReadHandle* pHandle, uint64_t* pGroupId, uint64_t* pUid,
int32_t* pNumOfRows) {
*pUid = 0;
int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReadHandle* pHandle) {
// TODO: cache multiple schema
int32_t sversion = htonl(pHandle->pBlock->sversion);
if (pHandle->cachedSchemaSuid == 0 || pHandle->cachedSchemaVer != sversion ||
@ -176,7 +177,6 @@ int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReadHandle* pHandle, uint64_
STSchema* pTschema = pHandle->pSchema;
SSchemaWrapper* pSchemaWrapper = pHandle->pSchemaWrapper;
*pNumOfRows = pHandle->msgIter.numOfRows;
int32_t colNumNeed = taosArrayGetSize(pHandle->pColIdList);
if (colNumNeed == 0) {
@ -217,22 +217,22 @@ int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReadHandle* pHandle, uint64_
}
}
if (blockDataEnsureCapacity(pBlock, *pNumOfRows) < 0) {
if (blockDataEnsureCapacity(pBlock, pHandle->msgIter.numOfRows) < 0) {
goto FAIL;
}
int32_t colActual = blockDataGetNumOfCols(pBlock);
// TODO in stream shuffle case, fetch groupId
*pGroupId = 0;
STSRowIter iter = {0};
tdSTSRowIterInit(&iter, pTschema);
STSRow* row;
int32_t curRow = 0;
tInitSubmitBlkIter(&pHandle->msgIter, pHandle->pBlock, &pHandle->blkIter);
*pUid = pHandle->msgIter.uid; // set the uid of table for submit block
pBlock->info.groupId = 0;
pBlock->info.uid = pHandle->msgIter.uid; // set the uid of table for submit block
pBlock->info.rows = pHandle->msgIter.numOfRows;
while ((row = tGetSubmitBlkNext(&pHandle->blkIter)) != NULL) {
tdSTSRowIterReset(&iter, row);

133
source/dnode/vnode/src/tq/tqSink.c
View File

@ -18,57 +18,87 @@
SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pTSchema, bool createTb, int64_t suid,
const char* stbFullName, int32_t vgId) {
SSubmitReq* ret = NULL;
SArray* schemaReqs = NULL;
SArray* schemaReqSz = NULL;
SArray* tagArray = taosArrayInit(1, sizeof(STagVal));
if (!tagArray) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
}
// cal size
int32_t cap = sizeof(SSubmitReq);
int32_t sz = taosArrayGetSize(pBlocks);
for (int32_t i = 0; i < sz; i++) {
SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
int32_t rows = pDataBlock->info.rows;
// TODO min
int32_t rowSize = pDataBlock->info.rowSize;
int32_t maxLen = TD_ROW_MAX_BYTES_FROM_SCHEMA(pTSchema);
int32_t schemaLen = 0;
if (createTb) {
SVCreateTbReq createTbReq = {0};
char* cname = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
createTbReq.name = cname;
createTbReq.flags = 0;
createTbReq.type = TSDB_CHILD_TABLE;
createTbReq.ctb.suid = suid;
STagVal tagVal = {
.cid = taosArrayGetSize(pDataBlock->pDataBlock) + 1,
.type = TSDB_DATA_TYPE_UBIGINT,
.i64 = (int64_t)pDataBlock->info.groupId,
if (createTb) {
schemaReqs = taosArrayInit(sz, sizeof(void*));
schemaReqSz = taosArrayInit(sz, sizeof(int32_t));
for (int32_t i = 0; i < sz; i++) {
SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
STagVal tagVal = {
.cid = taosArrayGetSize(pDataBlock->pDataBlock) + 1,
.type = TSDB_DATA_TYPE_UBIGINT,
.i64 = (int64_t)pDataBlock->info.groupId,
};
STag* pTag = NULL;
taosArrayClear(tagArray);
taosArrayPush(tagArray, &tagVal);
tTagNew(tagArray, 1, false, &pTag);
if (pTag == NULL) {
tdDestroySVCreateTbReq(&createTbReq);
terrno = TSDB_CODE_OUT_OF_MEMORY;
taosArrayDestroy(tagArray);
return NULL;
}
SVCreateTbReq createTbReq = {0};
createTbReq.name = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
createTbReq.flags = 0;
createTbReq.type = TSDB_CHILD_TABLE;
createTbReq.ctb.suid = suid;
createTbReq.ctb.pTag = (uint8_t*)pTag;
int32_t code;
int32_t schemaLen;
tEncodeSize(tEncodeSVCreateTbReq, &createTbReq, schemaLen, code);
tdDestroySVCreateTbReq(&createTbReq);
if (code < 0) {
tdDestroySVCreateTbReq(&createTbReq);
taosArrayDestroy(tagArray);
taosMemoryFreeClear(ret);
return NULL;
}
}
void* schemaStr = taosMemoryMalloc(schemaLen);
if (schemaStr == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
}
taosArrayPush(schemaReqs, &schemaStr);
taosArrayPush(schemaReqSz, &schemaLen);
SEncoder encoder = {0};
tEncoderInit(&encoder, schemaStr, schemaLen);
code = tEncodeSVCreateTbReq(&encoder, &createTbReq);
if (code < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
}
tEncoderClear(&encoder);
tdDestroySVCreateTbReq(&createTbReq);
}
}
taosArrayDestroy(tagArray);
// cal size
int32_t cap = sizeof(SSubmitReq);
for (int32_t i = 0; i < sz; i++) {
SSDataBlock* pDataBlock = taosArrayGet(pBlocks, i);
int32_t rows = pDataBlock->info.rows;
// TODO min
int32_t rowSize = pDataBlock->info.rowSize;
int32_t maxLen = TD_ROW_MAX_BYTES_FROM_SCHEMA(pTSchema);
int32_t schemaLen = 0;
if (createTb) {
schemaLen = *(int32_t*)taosArrayGet(schemaReqSz, i);
}
cap += sizeof(SSubmitBlk) + schemaLen + rows * maxLen;
}
@ -99,55 +129,13 @@ SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pTSchema, boo
int32_t schemaLen = 0;
if (createTb) {
SVCreateTbReq createTbReq = {0};
char* cname = buildCtbNameByGroupId(stbFullName, pDataBlock->info.groupId);
createTbReq.name = cname;
createTbReq.flags = 0;
createTbReq.type = TSDB_CHILD_TABLE;
createTbReq.ctb.suid = suid;
STagVal tagVal = {
.cid = taosArrayGetSize(pDataBlock->pDataBlock) + 1,
.type = TSDB_DATA_TYPE_UBIGINT,
.i64 = (int64_t)pDataBlock->info.groupId,
};
taosArrayClear(tagArray);
taosArrayPush(tagArray, &tagVal);
STag* pTag = NULL;
tTagNew(tagArray, 1, false, &pTag);
if (pTag == NULL) {
tdDestroySVCreateTbReq(&createTbReq);
taosArrayDestroy(tagArray);
taosMemoryFreeClear(ret);
return NULL;
}
createTbReq.ctb.pTag = (uint8_t*)pTag;
int32_t code;
tEncodeSize(tEncodeSVCreateTbReq, &createTbReq, schemaLen, code);
if (code < 0) {
tdDestroySVCreateTbReq(&createTbReq);
taosArrayDestroy(tagArray);
taosMemoryFreeClear(ret);
return NULL;
}
SEncoder encoder = {0};
tEncoderInit(&encoder, blkSchema, schemaLen);
code = tEncodeSVCreateTbReq(&encoder, &createTbReq);
tEncoderClear(&encoder);
tdDestroySVCreateTbReq(&createTbReq);
if (code < 0) {
taosArrayDestroy(tagArray);
taosMemoryFreeClear(ret);
return NULL;
}
schemaLen = *(int32_t*)taosArrayGet(schemaReqSz, i);
void* schemaStr = taosArrayGetP(schemaReqs, i);
memcpy(blkSchema, schemaStr, schemaLen);
}
blkHead->schemaLen = htonl(schemaLen);
STSRow* rowData = POINTER_SHIFT(blkSchema, schemaLen);
for (int32_t j = 0; j < rows; j++) {
SRowBuilder rb = {0};
tdSRowInit(&rb, pTSchema->version);
@ -175,7 +163,10 @@ SSubmitReq* tdBlockToSubmit(const SArray* pBlocks, const STSchema* pTSchema, boo
}
ret->length = htonl(ret->length);
taosArrayDestroy(tagArray);
if (schemaReqs) taosArrayDestroyP(schemaReqs, taosMemoryFree);
taosArrayDestroy(schemaReqSz);
return ret;
}

29
source/dnode/vnode/src/tsdb/tsdbRead.c
View File

@ -2852,7 +2852,7 @@ int32_t tsdbGetAllTableList(SMeta* pMeta, uint64_t uid, SArray* list) {
break;
}
STableKeyInfo info = {.lastKey = TSKEY_INITIAL_VAL, uid = id};
STableKeyInfo info = {.lastKey = TSKEY_INITIAL_VAL, uid = id, .groupId = 0};
taosArrayPush(list, &info);
}
@ -2876,6 +2876,33 @@ int32_t tsdbGetCtbIdList(SMeta* pMeta, int64_t suid, SArray* list) {
return TSDB_CODE_SUCCESS;
}
/**
* @brief Get all suids since suid
*
* @param pMeta
* @param suid return all suids in one vnode if suid is 0
* @param list
* @return int32_t
*/
int32_t tsdbGetStbIdList(SMeta* pMeta, int64_t suid, SArray* list) {
SMStbCursor* pCur = metaOpenStbCursor(pMeta, suid);
if(!pCur) {
return TSDB_CODE_FAILED;
}
while (1) {
tb_uid_t id = metaStbCursorNext(pCur);
if (id == 0) {
break;
}
taosArrayPush(list, &id);
}
metaCloseStbCursor(pCur);
return TSDB_CODE_SUCCESS;
}
static void destroyHelper(void* param) {
if (param == NULL) {
return;

7
source/dnode/vnode/src/vnd/vnodeCommit.c
View File

@ -72,7 +72,12 @@ int vnodeBegin(SVnode *pVnode) {
return 0;
}
int vnodeShouldCommit(SVnode *pVnode) { return pVnode->inUse->size > pVnode->config.szBuf / 3; }
int vnodeShouldCommit(SVnode *pVnode) {
if (pVnode->inUse) {
return pVnode->inUse->size > pVnode->config.szBuf / 3;
}
return false;
}
int vnodeSaveInfo(const char *dir, const SVnodeInfo *pInfo) {
char fname[TSDB_FILENAME_LEN];

6
source/dnode/vnode/src/vnd/vnodeOpen.c
View File

@ -152,12 +152,13 @@ SVnode *vnodeOpen(const char *path, STfs *pTfs, SMsgCb msgCb) {
return pVnode;
_err:
if (pVnode->pSma) smaClose(pVnode->pSma);
if (pVnode->pQuery) vnodeQueryClose(pVnode);
if (pVnode->pTq) tqClose(pVnode->pTq);
if (pVnode->pWal) walClose(pVnode->pWal);
if (pVnode->pTsdb) tsdbClose(&pVnode->pTsdb);
if (pVnode->pMeta) metaClose(pVnode->pMeta);
if (pVnode->pSma) smaClose(pVnode->pSma);
tsem_destroy(&(pVnode->canCommit));
taosMemoryFree(pVnode);
@ -166,13 +167,14 @@ _err:
void vnodeClose(SVnode *pVnode) {
if (pVnode) {
smaCloseEnv(pVnode->pSma);
vnodeCommit(pVnode);
vnodeSyncClose(pVnode);
vnodeQueryClose(pVnode);
walClose(pVnode->pWal);
tqClose(pVnode->pTq);
if (pVnode->pTsdb) tsdbClose(&pVnode->pTsdb);
smaClose(pVnode->pSma);
smaCloseEx(pVnode->pSma);
metaClose(pVnode->pMeta);
vnodeCloseBufPool(pVnode);
// destroy handle

2
source/dnode/vnode/src/vnd/vnodeSvr.c
View File

@ -299,7 +299,7 @@ int32_t vnodeProcessWriteMsg(SVnode *pVnode, int64_t version, SRpcMsg *pMsg, SRp
void smaHandleRes(void *pVnode, int64_t smaId, const SArray *data) {
// TODO
blockDebugShowData(data, __func__);
blockDebugShowDataBlocks(data, __func__);
tdProcessTSmaInsert(((SVnode *)pVnode)->pSma, smaId, (const char *)data);
}

69
source/dnode/vnode/src/vnd/vnodeSync.c
View File

@ -137,6 +137,25 @@ void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
vError("vgId:%d, failed to pre-process msg:%p since %s", vgId, pMsg, terrstr());
} else {
code = syncPropose(pVnode->sync, pMsg, vnodeIsMsgWeak(pMsg->msgType));
if (code == 1) {
do {
static int32_t cnt = 0;
if (cnt++ % 1000 == 1) {
vInfo("vgId:%d, msg:%p apply right now, apply index:%ld, msgtype:%s,%d", vgId, pMsg,
pMsg->info.conn.applyIndex, TMSG_INFO(pMsg->msgType), pMsg->msgType);
}
} while (0);
SRpcMsg rsp = {.code = pMsg->code, .info = pMsg->info};
if (vnodeProcessWriteReq(pVnode, pMsg, pMsg->info.conn.applyIndex, &rsp) < 0) {
rsp.code = terrno;
vInfo("vgId:%d, msg:%p failed to apply right now since %s", vgId, pMsg, terrstr());
}
if (rsp.info.handle != NULL) {
tmsgSendRsp(&rsp);
}
}
}
}
@ -163,10 +182,12 @@ void vnodeProposeMsg(SQueueInfo *pInfo, STaosQall *qall, int32_t numOfMsgs) {
SRpcMsg rsp = {.code = TSDB_CODE_RPC_REDIRECT, .info = pMsg->info};
tmsgSendRedirectRsp(&rsp, &newEpSet);
} else {
if (terrno != 0) code = terrno;
vError("vgId:%d, msg:%p failed to propose since %s, code:0x%x", vgId, pMsg, tstrerror(code), code);
SRpcMsg rsp = {.code = code, .info = pMsg->info};
tmsgSendRsp(&rsp);
if (code != 1) {
if (terrno != 0) code = terrno;
vError("vgId:%d, msg:%p failed to propose since %s, code:0x%x", vgId, pMsg, tstrerror(code), code);
SRpcMsg rsp = {.code = code, .info = pMsg->info};
tmsgSendRsp(&rsp);
}
}
vGTrace("vgId:%d, msg:%p is freed, code:0x%x", vgId, pMsg, code);
@ -260,7 +281,7 @@ int32_t vnodeProcessSyncReq(SVnode *pVnode, SRpcMsg *pMsg, SRpcMsg **pRsp) {
SyncClientRequest *pSyncMsg = syncClientRequestFromRpcMsg2(pRpcMsg);
assert(pSyncMsg != NULL);
ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg);
ret = syncNodeOnClientRequestCb(pSyncNode, pSyncMsg, NULL);
syncClientRequestDestroy(pSyncMsg);
} else if (pRpcMsg->msgType == TDMT_SYNC_REQUEST_VOTE) {
@ -359,34 +380,18 @@ static void vnodeSyncCommitMsg(SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta c
SyncIndex beginIndex = SYNC_INDEX_INVALID;
char logBuf[256] = {0};
if (pFsm->FpGetSnapshotInfo != NULL) {
(*pFsm->FpGetSnapshotInfo)(pFsm, &snapshot);
beginIndex = snapshot.lastApplyIndex;
}
snprintf(logBuf, sizeof(logBuf),
"==callback== ==CommitCb== execute, pFsm:%p, index:%ld, isWeak:%d, code:%d, state:%d %s, beginIndex :%ld\n",
pFsm, cbMeta.index, cbMeta.isWeak, cbMeta.code, cbMeta.state, syncUtilState2String(cbMeta.state),
beginIndex);
syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg);
if (cbMeta.index > beginIndex) {
snprintf(
logBuf, sizeof(logBuf),
"==callback== ==CommitCb== execute, pFsm:%p, index:%ld, isWeak:%d, code:%d, state:%d %s, beginIndex :%ld\n",
pFsm, cbMeta.index, cbMeta.isWeak, cbMeta.code, cbMeta.state, syncUtilState2String(cbMeta.state), beginIndex);
syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg);
SRpcMsg rpcMsg = {.msgType = pMsg->msgType, .contLen = pMsg->contLen};
rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen);
memcpy(rpcMsg.pCont, pMsg->pCont, pMsg->contLen);
syncGetAndDelRespRpc(pVnode->sync, cbMeta.seqNum, &rpcMsg.info);
rpcMsg.info.conn.applyIndex = cbMeta.index;
tmsgPutToQueue(&pVnode->msgCb, APPLY_QUEUE, &rpcMsg);
} else {
char logBuf[256] = {0};
snprintf(logBuf, sizeof(logBuf),
"==callback== ==CommitCb== do not execute, pFsm:%p, index:%ld, isWeak:%d, code:%d, state:%d %s, "
"beginIndex :%ld\n",
pFsm, cbMeta.index, cbMeta.isWeak, cbMeta.code, cbMeta.state, syncUtilState2String(cbMeta.state),
beginIndex);
syncRpcMsgLog2(logBuf, (SRpcMsg *)pMsg);
}
SRpcMsg rpcMsg = {.msgType = pMsg->msgType, .contLen = pMsg->contLen};
rpcMsg.pCont = rpcMallocCont(rpcMsg.contLen);
memcpy(rpcMsg.pCont, pMsg->pCont, pMsg->contLen);
syncGetAndDelRespRpc(pVnode->sync, cbMeta.seqNum, &rpcMsg.info);
rpcMsg.info.conn.applyIndex = cbMeta.index;
tmsgPutToQueue(&pVnode->msgCb, APPLY_QUEUE, &rpcMsg);
}
static void vnodeSyncPreCommitMsg(SSyncFSM *pFsm, const SRpcMsg *pMsg, SFsmCbMeta cbMeta) {

23
source/libs/executor/inc/executorimpl.h
View File

@ -293,6 +293,7 @@ typedef enum EStreamScanMode {
STREAM_SCAN_FROM_RES,
STREAM_SCAN_FROM_UPDATERES,
STREAM_SCAN_FROM_DATAREADER,
STREAM_SCAN_FROM_DATAREADER_RETRIEVE,
} EStreamScanMode;
typedef struct SCatchSupporter {
@ -348,7 +349,9 @@ typedef struct SStreamBlockScanInfo {
SArray* childIds;
SessionWindowSupporter sessionSup;
bool assignBlockUid; // assign block uid to groupId, temporarily used for generating rollup SMA.
int32_t scanWinIndex;
int32_t scanWinIndex; // for state operator
int32_t pullDataResIndex;
SSDataBlock* pPullDataRes; // pull data SSDataBlock
} SStreamBlockScanInfo;
typedef struct SSysTableScanInfo {
@ -400,7 +403,7 @@ typedef struct SIntervalAggOperatorInfo {
// SOptrBasicInfo should be first, SAggSupporter should be second for stream encode
SOptrBasicInfo binfo; // basic info
SAggSupporter aggSup; // aggregate supporter
SExprSupp scalarSupp; // supporter for perform scalar function
SGroupResInfo groupResInfo; // multiple results build supporter
SInterval interval; // interval info
int32_t primaryTsIndex; // primary time stamp slot id from result of downstream operator.
@ -427,8 +430,13 @@ typedef struct SStreamFinalIntervalOperatorInfo {
STimeWindowAggSupp twAggSup;
SArray* pChildren;
SSDataBlock* pUpdateRes;
bool returnUpdate;
SPhysiNode* pPhyNode; // create new child
bool isFinal;
SHashObj* pPullDataMap;
SArray* pPullWins; // SPullWindowInfo
int32_t pullIndex;
SSDataBlock* pPullDataRes;
} SStreamFinalIntervalOperatorInfo;
typedef struct SAggOperatorInfo {
@ -722,15 +730,12 @@ SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SExprInfo*
SOperatorInfo* createIndefinitOutputOperatorInfo(SOperatorInfo* downstream, SPhysiNode *pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createProjectOperatorInfo(SOperatorInfo* downstream, SProjectPhysiNode* pProjPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createSortOperatorInfo(SOperatorInfo* downstream, SSortPhysiNode* pSortPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMultiwaySortMergeOperatorInfo(SOperatorInfo** downStreams, int32_t numStreams, SSDataBlock* pInputBlock,
SSDataBlock* pResBlock, SArray* pSortInfo, SArray* pColMatchColInfo,
SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMultiwayMergeOperatorInfo(SOperatorInfo** dowStreams, size_t numStreams, SMergePhysiNode* pMergePhysiNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createSortedMergeOperatorInfo(SOperatorInfo** downstream, int32_t numOfDownstream, SExprInfo* pExprInfo, int32_t num, SArray* pSortInfo, SArray* pGroupInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols,
SSDataBlock* pResBlock, SInterval* pInterval, int32_t primaryTsSlotId,
STimeWindowAggSupp *pTwAggSupp, SExecTaskInfo* pTaskInfo, bool isStream);
STimeWindowAggSupp* pTwAggSupp, SIntervalPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, bool isStream);
SOperatorInfo* createMergeIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols,
SSDataBlock* pResBlock, SInterval* pInterval, int32_t primaryTsSlotId,
@ -800,9 +805,10 @@ int32_t getMaximumIdleDurationSec();
* ops: root operator
* data: *data save the result of encode, need to be freed by caller
* length: *length save the length of *data
* nOptrWithVal: *nOptrWithVal save the number of optr with value
* return: result code, 0 means success
*/
int32_t encodeOperator(SOperatorInfo* ops, char** data, int32_t *length);
int32_t encodeOperator(SOperatorInfo* ops, char** data, int32_t *length, int32_t *nOptrWithVal);
/*
* ops: root operator, created by caller
@ -851,6 +857,7 @@ SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanN
void copyUpdateDataBlock(SSDataBlock* pDest, SSDataBlock* pSource, int32_t tsColIndex);
int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle, SNodeList* groupKey);
SSDataBlock* createPullDataBlock();
#ifdef __cplusplus
}

7
source/libs/executor/inc/tsort.h
View File

@ -136,6 +136,13 @@ bool tsortIsNullVal(STupleHandle* pVHandle, int32_t colId);
*/
void* tsortGetValue(STupleHandle* pVHandle, int32_t colId);
/**
*
* @param pVHandle
* @return
*/
uint64_t tsortGetGroupId(STupleHandle* pVHandle);
/**
*
* @param pSortHandle

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

@ -159,7 +159,12 @@ int32_t getNumOfTotalRes(SGroupResInfo* pGroupResInfo) {
}
SArray* createSortInfo(SNodeList* pNodeList) {
size_t numOfCols = LIST_LENGTH(pNodeList);
size_t numOfCols = 0;
if (pNodeList != NULL) {
numOfCols = LIST_LENGTH(pNodeList);
} else {
numOfCols = 0;
}
SArray* pList = taosArrayInit(numOfCols, sizeof(SBlockOrderInfo));
if (pList == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
@ -316,7 +321,7 @@ int32_t getTableList(void* metaHandle, SScanPhysiNode* pScanNode, STableListInfo
}
for (int i = 0; i < taosArrayGetSize(res); i++) {
STableKeyInfo info = {.lastKey = TSKEY_INITIAL_VAL, .uid = *(uint64_t*)taosArrayGet(res, i)};
STableKeyInfo info = {.lastKey = TSKEY_INITIAL_VAL, .uid = *(uint64_t*)taosArrayGet(res, i), .groupId = 0};
taosArrayPush(pListInfo->pTableList, &info);
}
taosArrayDestroy(res);
@ -344,7 +349,7 @@ int32_t getTableList(void* metaHandle, SScanPhysiNode* pScanNode, STableListInfo
}
}
}else { // Create one table group.
STableKeyInfo info = {.lastKey = 0, .uid = tableUid};
STableKeyInfo info = {.lastKey = 0, .uid = tableUid, .groupId = 0};
taosArrayPush(pListInfo->pTableList, &info);
}
pListInfo->pGroupList = taosArrayInit(4, POINTER_BYTES);

8
source/libs/executor/src/executorMain.c
View File

@ -222,7 +222,13 @@ int32_t qSerializeTaskStatus(qTaskInfo_t tinfo, char** pOutput, int32_t* len) {
return TSDB_CODE_INVALID_PARA;
}
return encodeOperator(pTaskInfo->pRoot, pOutput, len);
int32_t nOptrWithVal = 0;
int32_t code = encodeOperator(pTaskInfo->pRoot, pOutput, len, &nOptrWithVal);
if ((code == TSDB_CODE_SUCCESS) && (nOptrWithVal = 0)) {
taosMemoryFreeClear(*pOutput);
*len = 0;
}
return code;
}
int32_t qDeserializeTaskStatus(qTaskInfo_t tinfo, const char* pInput, int32_t len) {

40
source/libs/executor/src/executorimpl.c
View File

@ -2928,6 +2928,13 @@ int32_t aggEncodeResultRow(SOperatorInfo* pOperator, char** result, int32_t* len
int32_t totalSize =
sizeof(int32_t) + sizeof(int32_t) + size * (sizeof(int32_t) + keyLen + sizeof(int32_t) + pSup->resultRowSize);
// no result
if (getTotalBufSize(pSup->pResultBuf) == 0) {
*result = NULL;
*length = 0;
return TSDB_CODE_SUCCESS;
}
*result = (char*)taosMemoryCalloc(1, totalSize);
if (*result == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
@ -4034,6 +4041,7 @@ int32_t generateGroupIdMap(STableListInfo* pTableListInfo, SReadHandle* pHandle,
int32_t len = (int32_t)(pStart - (char*)keyBuf);
uint64_t groupId = calcGroupId(keyBuf, len);
taosHashPut(pTableListInfo->map, &(info->uid), sizeof(uint64_t), &groupId, sizeof(uint64_t));
info->groupId = groupId;
groupNum++;
nodesClearList(groupNew);
@ -4137,7 +4145,7 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
return NULL;
}
} else { // Create one table group.
STableKeyInfo info = {.lastKey = 0, .uid = pBlockNode->uid};
STableKeyInfo info = {.lastKey = 0, .uid = pBlockNode->uid, .groupId = 0};
taosArrayPush(pTableListInfo->pTableList, &info);
}
@ -4227,8 +4235,8 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
int32_t tsSlotId = ((SColumnNode*)pIntervalPhyNode->window.pTspk)->slotId;
bool isStream = (QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL == type);
pOptr =
createIntervalOperatorInfo(ops[0], pExprInfo, num, pResBlock, &interval, tsSlotId, &as, pTaskInfo, isStream);
pOptr = createIntervalOperatorInfo(ops[0], pExprInfo, num, pResBlock, &interval, tsSlotId, &as, pIntervalPhyNode,
pTaskInfo, isStream);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL == type) {
SMergeAlignedIntervalPhysiNode* pIntervalPhyNode = (SMergeAlignedIntervalPhysiNode*)pPhyNode;
@ -4272,17 +4280,7 @@ SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo
pOptr = createGroupSortOperatorInfo(ops[0], (SGroupSortPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE == type) {
SMergePhysiNode* pMergePhyNode = (SMergePhysiNode*)pPhyNode;
SDataBlockDescNode* pDescNode = pPhyNode->pOutputDataBlockDesc;
SSDataBlock* pResBlock = createResDataBlock(pDescNode);
SArray* sortInfo = createSortInfo(pMergePhyNode->pMergeKeys);
int32_t numOfOutputCols = 0;
SArray* pColList =
extractColMatchInfo(pMergePhyNode->pTargets, pDescNode, &numOfOutputCols, COL_MATCH_FROM_SLOT_ID);
SPhysiNode* pChildNode = (SPhysiNode*)nodesListGetNode(pPhyNode->pChildren, 0);
SSDataBlock* pInputDataBlock = createResDataBlock(pChildNode->pOutputDataBlockDesc);
pOptr = createMultiwaySortMergeOperatorInfo(ops, size, pInputDataBlock, pResBlock, sortInfo, pColList, pTaskInfo);
pOptr = createMultiwayMergeOperatorInfo(ops, size, pMergePhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION == type) {
SSessionWinodwPhysiNode* pSessionNode = (SSessionWinodwPhysiNode*)pPhyNode;
@ -4475,12 +4473,12 @@ int32_t rebuildReader(SOperatorInfo* pOperator, SSubplan* plan, SReadHandle* pHa
return 0;
}
int32_t encodeOperator(SOperatorInfo* ops, char** result, int32_t* length) {
int32_t encodeOperator(SOperatorInfo* ops, char** result, int32_t* length, int32_t* nOptrWithVal) {
int32_t code = TDB_CODE_SUCCESS;
char* pCurrent = NULL;
int32_t currLength = 0;
if (ops->fpSet.encodeResultRow) {
if (result == NULL || length == NULL) {
if (result == NULL || length == NULL || nOptrWithVal == NULL) {
return TSDB_CODE_TSC_INVALID_INPUT;
}
code = ops->fpSet.encodeResultRow(ops, &pCurrent, &currLength);
@ -4491,8 +4489,15 @@ int32_t encodeOperator(SOperatorInfo* ops, char** result, int32_t* length) {
*result = NULL;
}
return code;
} else if (currLength == 0) {
ASSERT(!pCurrent);
goto _downstream;
}
++(*nOptrWithVal);
ASSERT(currLength >= 0);
if (*result == NULL) {
*result = (char*)taosMemoryCalloc(1, currLength + sizeof(int32_t));
if (*result == NULL) {
@ -4518,8 +4523,9 @@ int32_t encodeOperator(SOperatorInfo* ops, char** result, int32_t* length) {
*length = *(int32_t*)(*result);
}
_downstream:
for (int32_t i = 0; i < ops->numOfDownstream; ++i) {
code = encodeOperator(ops->pDownstream[i], result, length);
code = encodeOperator(ops->pDownstream[i], result, length, nOptrWithVal);
if (code != TDB_CODE_SUCCESS) {
return code;
}

258
source/libs/executor/src/scanoperator.c
View File

@ -448,7 +448,7 @@ static SSDataBlock* doTableScanGroup(SOperatorInfo* pOperator) {
qDebug("%s start to repeat ascending order scan data blocks due to query func required", GET_TASKID(pTaskInfo));
for (int32_t i = 0; i < pTableScanInfo->cond.numOfTWindows; ++i) {
STimeWindow* pWin = &pTableScanInfo->cond.twindows[i];
qDebug("%s\t qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey);
qDebug("%s qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey);
}
// do prepare for the next round table scan operation
tsdbResetReadHandle(pTableScanInfo->dataReader, &pTableScanInfo->cond, 0);
@ -467,7 +467,7 @@ static SSDataBlock* doTableScanGroup(SOperatorInfo* pOperator) {
qDebug("%s start to descending order scan data blocks due to query func required", GET_TASKID(pTaskInfo));
for (int32_t i = 0; i < pTableScanInfo->cond.numOfTWindows; ++i) {
STimeWindow* pWin = &pTableScanInfo->cond.twindows[i];
qDebug("%s\t qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey);
qDebug("%s qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey);
}
while (pTableScanInfo->scanTimes < total) {
@ -492,7 +492,7 @@ static SSDataBlock* doTableScanGroup(SOperatorInfo* pOperator) {
GET_TASKID(pTaskInfo));
for (int32_t i = 0; i < pTableScanInfo->cond.numOfTWindows; ++i) {
STimeWindow* pWin = &pTableScanInfo->cond.twindows[i];
qDebug("%s\t qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey);
qDebug("%s qrange:%" PRId64 "-%" PRId64, GET_TASKID(pTaskInfo), pWin->skey, pWin->ekey);
}
tsdbResetReadHandle(pTableScanInfo->dataReader, &pTableScanInfo->cond, 0);
pTableScanInfo->curTWinIdx = 0;
@ -507,20 +507,21 @@ static SSDataBlock* doTableScan(SOperatorInfo* pOperator) {
STableScanInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
if(pInfo->currentGroupId == -1){
if (pInfo->currentGroupId == -1) {
pInfo->currentGroupId++;
if (pInfo->currentGroupId >= taosArrayGetSize(pTaskInfo->tableqinfoList.pGroupList)) {
setTaskStatus(pTaskInfo, TASK_COMPLETED);
return NULL;
}
SArray *tableList = taosArrayGetP(pTaskInfo->tableqinfoList.pGroupList, pInfo->currentGroupId);
SArray* tableList = taosArrayGetP(pTaskInfo->tableqinfoList.pGroupList, pInfo->currentGroupId);
tsdbCleanupReadHandle(pInfo->dataReader);
tsdbReaderT* pReader = tsdbReaderOpen(pInfo->readHandle.vnode, &pInfo->cond, tableList, pInfo->queryId, pInfo->taskId);
tsdbReaderT* pReader =
tsdbReaderOpen(pInfo->readHandle.vnode, &pInfo->cond, tableList, pInfo->queryId, pInfo->taskId);
pInfo->dataReader = pReader;
}
SSDataBlock* result = doTableScanGroup(pOperator);
if(result){
if (result) {
return result;
}
@ -530,7 +531,7 @@ static SSDataBlock* doTableScan(SOperatorInfo* pOperator) {
return NULL;
}
SArray *tableList = taosArrayGetP(pTaskInfo->tableqinfoList.pGroupList, pInfo->currentGroupId);
SArray* tableList = taosArrayGetP(pTaskInfo->tableqinfoList.pGroupList, pInfo->currentGroupId);
tsdbSetTableList(pInfo->dataReader, tableList);
tsdbResetReadHandle(pInfo->dataReader, &pInfo->cond, 0);
@ -538,7 +539,7 @@ static SSDataBlock* doTableScan(SOperatorInfo* pOperator) {
pInfo->scanTimes = 0;
result = doTableScanGroup(pOperator);
if(result){
if (result) {
return result;
}
@ -776,15 +777,14 @@ static bool isStateWindow(SStreamBlockScanInfo* pInfo) {
return pInfo->sessionSup.parentType == QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE;
}
static bool prepareDataScan(SStreamBlockScanInfo* pInfo) {
SSDataBlock* pSDB = pInfo->pUpdateRes;
STimeWindow win = {
.skey = INT64_MIN,
.ekey = INT64_MAX,
static bool prepareDataScan(SStreamBlockScanInfo* pInfo, SSDataBlock* pSDB, int32_t tsColIndex, int32_t* pRowIndex) {
STimeWindow win = {
.skey = INT64_MIN,
.ekey = INT64_MAX,
};
bool needRead = false;
if (!isStateWindow(pInfo) && pInfo->updateResIndex < pSDB->info.rows) {
SColumnInfoData* pColDataInfo = taosArrayGet(pSDB->pDataBlock, pInfo->primaryTsIndex);
if (!isStateWindow(pInfo) && (*pRowIndex) < pSDB->info.rows) {
SColumnInfoData* pColDataInfo = taosArrayGet(pSDB->pDataBlock, tsColIndex);
TSKEY* tsCols = (TSKEY*)pColDataInfo->pData;
SResultRowInfo dumyInfo;
dumyInfo.cur.pageId = -1;
@ -793,15 +793,14 @@ static bool prepareDataScan(SStreamBlockScanInfo* pInfo) {
int64_t gap = pInfo->sessionSup.gap;
int32_t winIndex = 0;
SResultWindowInfo* pCurWin =
getSessionTimeWindow(pAggSup, tsCols[pInfo->updateResIndex], INT64_MIN, pSDB->info.groupId, gap, &winIndex);
getSessionTimeWindow(pAggSup, tsCols[(*pRowIndex)], INT64_MIN, pSDB->info.groupId, gap, &winIndex);
win = pCurWin->win;
pInfo->updateResIndex +=
updateSessionWindowInfo(pCurWin, tsCols, NULL, pSDB->info.rows, pInfo->updateResIndex, gap, NULL);
(*pRowIndex) += updateSessionWindowInfo(pCurWin, tsCols, NULL, pSDB->info.rows, (*pRowIndex), gap, NULL);
} else {
win = getActiveTimeWindow(NULL, &dumyInfo, tsCols[pInfo->updateResIndex], &pInfo->interval,
pInfo->interval.precision, NULL);
pInfo->updateResIndex += getNumOfRowsInTimeWindow(&pSDB->info, tsCols, pInfo->updateResIndex, win.ekey,
binarySearchForKey, NULL, TSDB_ORDER_ASC);
win =
getActiveTimeWindow(NULL, &dumyInfo, tsCols[(*pRowIndex)], &pInfo->interval, pInfo->interval.precision, NULL);
(*pRowIndex) += getNumOfRowsInTimeWindow(&pSDB->info, tsCols, (*pRowIndex), win.ekey, binarySearchForKey, NULL,
TSDB_ORDER_ASC);
}
needRead = true;
} else if (isStateWindow(pInfo)) {
@ -822,7 +821,10 @@ static bool prepareDataScan(SStreamBlockScanInfo* pInfo) {
STableScanInfo* pTableScanInfo = pInfo->pSnapshotReadOp->info;
pTableScanInfo->cond.twindows[0] = win;
pTableScanInfo->curTWinIdx = 0;
// tsdbResetReadHandle(pTableScanInfo->dataReader, &pTableScanInfo->cond, 0);
// tsdbResetReadHandle(pTableScanInfo->dataReader, &pTableScanInfo->cond, 0);
// if (!pTableScanInfo->dataReader) {
// return false;
// }
pTableScanInfo->scanTimes = 0;
pTableScanInfo->currentGroupId = -1;
return true;
@ -862,12 +864,12 @@ static uint64_t getGroupId(SOperatorInfo* pOperator, SSDataBlock* pBlock, int32_
*/
}
static SSDataBlock* doDataScan(SStreamBlockScanInfo* pInfo) {
static SSDataBlock* doDataScan(SStreamBlockScanInfo* pInfo, SSDataBlock* pSDB, int32_t tsColIndex, int32_t* pRowIndex) {
while (1) {
SSDataBlock* pResult = NULL;
pResult = doTableScan(pInfo->pSnapshotReadOp);
if (pResult == NULL) {
if (prepareDataScan(pInfo)) {
if (prepareDataScan(pInfo, pSDB, tsColIndex, pRowIndex)) {
// scan next window data
pResult = doTableScan(pInfo->pSnapshotReadOp);
}
@ -916,7 +918,7 @@ static void setUpdateData(SStreamBlockScanInfo* pInfo, SSDataBlock* pBlock, SSDa
pUpdateBlock->info.rows = i;
pInfo->tsArrayIndex += i;
pUpdateBlock->info.groupId = pInfo->groupId;
pUpdateBlock->info.type = STREAM_REPROCESS;
pUpdateBlock->info.type = STREAM_CLEAR;
blockDataUpdateTsWindow(pUpdateBlock, 0);
}
// all rows have same group id
@ -970,6 +972,14 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
int32_t current = pInfo->validBlockIndex++;
SSDataBlock* pBlock = taosArrayGetP(pInfo->pBlockLists, current);
blockDataUpdateTsWindow(pBlock, 0);
if (pBlock->info.type == STREAM_RETRIEVE) {
pInfo->blockType = STREAM_DATA_TYPE_SUBMIT_BLOCK;
pInfo->scanMode = STREAM_SCAN_FROM_DATAREADER_RETRIEVE;
copyDataBlock(pInfo->pPullDataRes, pBlock);
pInfo->pullDataResIndex = 0;
prepareDataScan(pInfo, pInfo->pPullDataRes, 0, &pInfo->pullDataResIndex);
updateInfoAddCloseWindowSBF(pInfo->pUpdateInfo);
}
return pBlock;
} else if (pInfo->blockType == STREAM_DATA_TYPE_SUBMIT_BLOCK) {
if (pInfo->scanMode == STREAM_SCAN_FROM_RES) {
@ -979,28 +989,39 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
} else if (pInfo->scanMode == STREAM_SCAN_FROM_UPDATERES) {
pInfo->scanMode = STREAM_SCAN_FROM_DATAREADER;
if (!isStateWindow(pInfo)) {
prepareDataScan(pInfo);
prepareDataScan(pInfo, pInfo->pUpdateRes, pInfo->primaryTsIndex, &pInfo->updateResIndex);
}
return pInfo->pUpdateRes;
} else if (pInfo->scanMode == STREAM_SCAN_FROM_DATAREADER_RETRIEVE) {
SSDataBlock* pSDB = doDataScan(pInfo, pInfo->pPullDataRes, 0, &pInfo->pullDataResIndex);
if (pSDB != NULL) {
getUpdateDataBlock(pInfo, true, pSDB, NULL);
pSDB->info.type = STREAM_PUSH_DATA;
return pSDB;
}
pInfo->scanMode = STREAM_SCAN_FROM_DATAREADER;
} else {
if (isStateWindow(pInfo) && taosArrayGetSize(pInfo->sessionSup.pStreamAggSup->pScanWindow) > 0) {
pInfo->scanMode = STREAM_SCAN_FROM_DATAREADER;
pInfo->updateResIndex = pInfo->pUpdateRes->info.rows;
prepareDataScan(pInfo);
prepareDataScan(pInfo, pInfo->pUpdateRes, pInfo->primaryTsIndex, &pInfo->updateResIndex);
}
if (pInfo->scanMode == STREAM_SCAN_FROM_DATAREADER) {
SSDataBlock* pSDB = doDataScan(pInfo);
SSDataBlock* pSDB = doDataScan(pInfo, pInfo->pUpdateRes, pInfo->primaryTsIndex, &pInfo->updateResIndex);
if (pSDB == NULL) {
setUpdateData(pInfo, pInfo->pRes, pInfo->pUpdateRes);
if (pInfo->pUpdateRes->info.rows > 0) {
if (!isStateWindow(pInfo)) {
prepareDataScan(pInfo);
// Todo(liuyao) mybe can delete this.
bool test = prepareDataScan(pInfo, pInfo->pUpdateRes, pInfo->primaryTsIndex, &pInfo->updateResIndex);
ASSERT(test == false);
}
return pInfo->pUpdateRes;
} else {
pInfo->scanMode = STREAM_SCAN_FROM_READERHANDLE;
}
} else {
pSDB->info.type = STREAM_NORMAL;
getUpdateDataBlock(pInfo, true, pSDB, NULL);
return pSDB;
}
@ -1012,12 +1033,13 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
while (tqNextDataBlock(pInfo->streamBlockReader)) {
SSDataBlock block = {0};
uint64_t groupId = 0;
uint64_t uid = 0;
int32_t numOfRows = 0;
// todo refactor
int32_t code = tqRetrieveDataBlock(&block, pInfo->streamBlockReader, &groupId, &uid, &numOfRows);
int32_t code = tqRetrieveDataBlock(&block, pInfo->streamBlockReader);
uint64_t groupId = block.info.groupId;
uint64_t uid = block.info.uid;
int32_t numOfRows = block.info.rows;
if (code != TSDB_CODE_SUCCESS || numOfRows == 0) {
pTaskInfo->code = code;
@ -1070,10 +1092,12 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
taosArrayDestroy(block.pDataBlock);
if (pInfo->pRes->pDataBlock == NULL) {
// TODO add log
updateInfoDestoryColseWinSBF(pInfo->pUpdateInfo);
pOperator->status = OP_EXEC_DONE;
pTaskInfo->code = terrno;
return NULL;
}
// currently only the tbname pseudo column
if (pInfo->numOfPseudoExpr > 0) {
addTagPseudoColumnData(&pInfo->readHandle, pInfo->pPseudoExpr, pInfo->numOfPseudoExpr, pInfo->pRes);
@ -1091,12 +1115,13 @@ static SSDataBlock* doStreamBlockScan(SOperatorInfo* pOperator) {
pOperator->resultInfo.totalRows += pBlockInfo->rows;
if (pBlockInfo->rows == 0) {
updateInfoDestoryColseWinSBF(pInfo->pUpdateInfo);
pOperator->status = OP_EXEC_DONE;
} else if (pInfo->pUpdateInfo) {
pInfo->tsArrayIndex = 0;
getUpdateDataBlock(pInfo, true, pInfo->pRes, pInfo->pUpdateRes);
if (pInfo->pUpdateRes->info.rows > 0) {
if (pInfo->pUpdateRes->info.type == STREAM_REPROCESS) {
if (pInfo->pUpdateRes->info.type == STREAM_CLEAR) {
pInfo->updateResIndex = 0;
pInfo->scanMode = STREAM_SCAN_FROM_UPDATERES;
} else if (pInfo->pUpdateRes->info.type == STREAM_INVERT) {
@ -1130,9 +1155,9 @@ static SArray* extractTableIdList(const STableListInfo* pTableGroupInfo) {
return tableIdList;
}
SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle,
STableScanPhysiNode* pTableScanNode, SExecTaskInfo* pTaskInfo,
STimeWindowAggSupp* pTwSup, uint64_t queryId, uint64_t taskId) {
SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle, STableScanPhysiNode* pTableScanNode,
SExecTaskInfo* pTaskInfo, STimeWindowAggSupp* pTwSup, uint64_t queryId,
uint64_t taskId) {
SStreamBlockScanInfo* pInfo = taosMemoryCalloc(1, sizeof(SStreamBlockScanInfo));
SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
@ -1209,6 +1234,7 @@ SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle,
pInfo->scanMode = STREAM_SCAN_FROM_READERHANDLE;
pInfo->sessionSup = (SessionWindowSupporter){.pStreamAggSup = NULL, .gap = -1};
pInfo->groupId = 0;
pInfo->pPullDataRes = createPullDataBlock();
pOperator->name = "StreamBlockScanOperator";
pOperator->operatorType = QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN;
@ -1718,14 +1744,14 @@ SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSystemTableScan
SSDataBlock* pResBlock = createResDataBlock(pDescNode);
int32_t num = 0;
SArray* colList = extractColMatchInfo(pScanNode->pScanCols, pDescNode, &num, COL_MATCH_FROM_COL_ID);
SArray* colList = extractColMatchInfo(pScanNode->pScanCols, pDescNode, &num, COL_MATCH_FROM_COL_ID);
pInfo->accountId = pScanPhyNode->accountId;
pInfo->pUser = taosMemoryStrDup((void*) pUser);
pInfo->accountId = pScanPhyNode->accountId;
pInfo->pUser = taosMemoryStrDup((void*)pUser);
pInfo->showRewrite = pScanPhyNode->showRewrite;
pInfo->pRes = pResBlock;
pInfo->pCondition = pScanNode->node.pConditions;
pInfo->scanCols = colList;
pInfo->pRes = pResBlock;
pInfo->pCondition = pScanNode->node.pConditions;
pInfo->scanCols = colList;
initResultSizeInfo(pOperator, 4096);
@ -1741,13 +1767,13 @@ SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSystemTableScan
pInfo->readHandle = *(SReadHandle*)readHandle;
}
pOperator->name = "SysTableScanOperator";
pOperator->name = "SysTableScanOperator";
pOperator->operatorType = QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN;
pOperator->blocking = false;
pOperator->status = OP_NOT_OPENED;
pOperator->info = pInfo;
pOperator->blocking = false;
pOperator->status = OP_NOT_OPENED;
pOperator->info = pInfo;
pOperator->exprSupp.numOfExprs = taosArrayGetSize(pResBlock->pDataBlock);
pOperator->pTaskInfo = pTaskInfo;
pOperator->pTaskInfo = pTaskInfo;
pOperator->fpSet =
createOperatorFpSet(operatorDummyOpenFn, doSysTableScan, NULL, NULL, destroySysScanOperator, NULL, NULL, NULL);
@ -1934,11 +1960,11 @@ SOperatorInfo* createTagScanOperatorInfo(SReadHandle* pReadHandle, STagScanPhysi
goto _error;
}
pInfo->pTableList = pTableListInfo;
pInfo->pColMatchInfo = colList;
pInfo->pRes = createResDataBlock(pDescNode);
pInfo->readHandle = *pReadHandle;
pInfo->curPos = 0;
pInfo->pTableList = pTableListInfo;
pInfo->pColMatchInfo = colList;
pInfo->pRes = createResDataBlock(pDescNode);
pInfo->readHandle = *pReadHandle;
pInfo->curPos = 0;
pOperator->name = "TagScanOperator";
pOperator->operatorType = QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN;
@ -1965,7 +1991,10 @@ _error:
typedef struct STableMergeScanInfo {
STableListInfo* tableListInfo;
int32_t currentGroupId;
int32_t tableStartIndex;
int32_t tableEndIndex;
bool hasGroupId;
uint64_t groupId;
SArray* dataReaders; // array of tsdbReaderT*
SReadHandle readHandle;
@ -2022,7 +2051,7 @@ int32_t createScanTableListInfo(STableScanPhysiNode* pTableScanNode, SReadHandle
qDebug("no table qualified for query, TID:0x%" PRIx64 ", QID:0x%" PRIx64, taskId, queryId);
return TSDB_CODE_SUCCESS;
}
code = generateGroupIdMap(pTableListInfo, pHandle, pTableScanNode->pPartitionTags);
code = generateGroupIdMap(pTableListInfo, pHandle, pTableScanNode->pGroupTags);
if (code != TSDB_CODE_SUCCESS) {
return code;
}
@ -2030,6 +2059,22 @@ int32_t createScanTableListInfo(STableScanPhysiNode* pTableScanNode, SReadHandle
return TSDB_CODE_SUCCESS;
}
int32_t createMultipleDataReaders(SQueryTableDataCond* pQueryCond, SReadHandle* pHandle, STableListInfo* pTableListInfo,
int32_t tableStartIdx, int32_t tableEndIdx, SArray* arrayReader, uint64_t queryId,
uint64_t taskId) {
for (int32_t i = tableStartIdx; i <= tableEndIdx; ++i) {
SArray* subTableList = taosArrayInit(1, sizeof(STableKeyInfo));
taosArrayPush(subTableList, taosArrayGet(pTableListInfo->pTableList, i));
tsdbReaderT* pReader = tsdbReaderOpen(pHandle->vnode, pQueryCond, subTableList, queryId, taskId);
taosArrayPush(arrayReader, &pReader);
taosArrayDestroy(subTableList);
}
return TSDB_CODE_SUCCESS;
}
// todo refactor
static int32_t loadDataBlockFromOneTable(SOperatorInfo* pOperator, STableMergeScanInfo* pTableScanInfo,
int32_t readerIdx, SSDataBlock* pBlock, uint32_t* status) {
@ -2216,34 +2261,32 @@ SArray* generateSortByTsInfo(int32_t order) {
return pList;
}
static int32_t createMultipleDataReaders(SQueryTableDataCond* pQueryCond, SReadHandle* pHandle, SArray* tableList, SArray* arrayReader, uint64_t queryId,
uint64_t taskId) {
for (int32_t i = 0; i < taosArrayGetSize(tableList); ++i) {
SArray* tmp = taosArrayInit(1, sizeof(STableKeyInfo));
taosArrayPush(tmp, taosArrayGet(tableList, i));
tsdbReaderT* pReader = tsdbReaderOpen(pHandle->vnode, pQueryCond, tmp, queryId, taskId);
taosArrayPush(arrayReader, &pReader);
taosArrayDestroy(tmp);
}
return TSDB_CODE_SUCCESS;
}
int32_t startGroupTableMergeScan(SOperatorInfo* pOperator) {
STableMergeScanInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
SArray* tableList = taosArrayGetP(pInfo->tableListInfo->pGroupList, pInfo->currentGroupId);
{
size_t tableListSize = taosArrayGetSize(pInfo->tableListInfo->pTableList);
int32_t i = pInfo->tableStartIndex + 1;
for (; i < tableListSize; ++i) {
STableKeyInfo* tableKeyInfo = taosArrayGet(pInfo->tableListInfo->pTableList, i);
if (tableKeyInfo->groupId != pInfo->groupId) {
break;
}
}
pInfo->tableEndIndex = i - 1;
}
createMultipleDataReaders(&pInfo->cond, &pInfo->readHandle, tableList,
int32_t tableStartIdx = pInfo->tableStartIndex;
int32_t tableEndIdx = pInfo->tableEndIndex;
STableListInfo* tableListInfo = pInfo->tableListInfo;
createMultipleDataReaders(&pInfo->cond, &pInfo->readHandle, tableListInfo, tableStartIdx, tableEndIdx,
pInfo->dataReaders, pInfo->queryId, pInfo->taskId);
// todo the total available buffer should be determined by total capacity of buffer of this task.
// the additional one is reserved for merge result
int32_t tableLen = taosArrayGetSize(tableList);
pInfo->sortBufSize = pInfo->bufPageSize * ((tableLen==0?1:tableLen) + 1);
pInfo->sortBufSize = pInfo->bufPageSize * (tableEndIdx - tableStartIdx + 1 + 1);
int32_t numOfBufPage = pInfo->sortBufSize / pInfo->bufPageSize;
pInfo->pSortHandle = tsortCreateSortHandle(pInfo->pSortInfo, SORT_MULTISOURCE_MERGE, pInfo->bufPageSize, numOfBufPage,
pInfo->pSortInputBlock, pTaskInfo->id.str);
@ -2330,43 +2373,38 @@ SSDataBlock* doTableMergeScan(SOperatorInfo* pOperator) {
if (code != TSDB_CODE_SUCCESS) {
longjmp(pTaskInfo->env, code);
}
size_t tableListSize = taosArrayGetSize(pInfo->tableListInfo->pTableList);
if (!pInfo->hasGroupId) {
pInfo->hasGroupId = true;
if (pInfo->currentGroupId == -1) {
pInfo->currentGroupId++;
if (pInfo->currentGroupId >= taosArrayGetSize(pInfo->tableListInfo->pGroupList)) {
if (tableListSize == 0) {
doSetOperatorCompleted(pOperator);
return NULL;
}
pInfo->tableStartIndex = 0;
pInfo->groupId = ((STableKeyInfo*)taosArrayGet(pInfo->tableListInfo->pTableList, pInfo->tableStartIndex))->groupId;
startGroupTableMergeScan(pOperator);
}
SSDataBlock* pBlock = getSortedTableMergeScanBlockData(pInfo->pSortHandle, pOperator->resultInfo.capacity, pOperator);
if (pBlock != NULL) {
uint64_t* groupId = taosHashGet(pInfo->tableListInfo->map, &(pBlock->info.uid), sizeof(uint64_t));
if(groupId) pBlock->info.groupId = *groupId;
pOperator->resultInfo.totalRows += pBlock->info.rows;
return pBlock;
SSDataBlock* pBlock = NULL;
while (pInfo->tableStartIndex < tableListSize) {
pBlock = getSortedTableMergeScanBlockData(pInfo->pSortHandle, pOperator->resultInfo.capacity, pOperator);
if (pBlock != NULL) {
pBlock->info.groupId = pInfo->groupId;
pOperator->resultInfo.totalRows += pBlock->info.rows;
return pBlock;
} else {
stopGroupTableMergeScan(pOperator);
if (pInfo->tableEndIndex >= tableListSize - 1) {
doSetOperatorCompleted(pOperator);
break;
}
pInfo->tableStartIndex = pInfo->tableEndIndex + 1;
pInfo->groupId =
((STableKeyInfo*)taosArrayGet(pInfo->tableListInfo->pTableList, pInfo->tableStartIndex))->groupId;
startGroupTableMergeScan(pOperator);
}
}
stopGroupTableMergeScan(pOperator);
pInfo->currentGroupId++;
if (pInfo->currentGroupId >= taosArrayGetSize(pInfo->tableListInfo->pGroupList)) {
doSetOperatorCompleted(pOperator);
return NULL;
}
startGroupTableMergeScan(pOperator);
pBlock = getSortedTableMergeScanBlockData(pInfo->pSortHandle, pOperator->resultInfo.capacity, pOperator);
if (pBlock != NULL) {
uint64_t* groupId = taosHashGet(pInfo->tableListInfo->map, &(pBlock->info.uid), sizeof(uint64_t));
if(groupId) pBlock->info.groupId = *groupId;
pOperator->resultInfo.totalRows += pBlock->info.rows;
return pBlock;
}
doSetOperatorCompleted(pOperator);
return pBlock;
}
@ -2403,6 +2441,12 @@ int32_t getTableMergeScanExplainExecInfo(SOperatorInfo* pOptr, void** pOptrExpla
return TSDB_CODE_SUCCESS;
}
int32_t compareTableKeyInfoByGid(const void* p1, const void* p2) {
const STableKeyInfo* info1 = p1;
const STableKeyInfo* info2 = p2;
return info1->groupId - info2->groupId;
}
SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanNode, STableListInfo* pTableListInfo,
SReadHandle* readHandle, SExecTaskInfo* pTaskInfo, uint64_t queryId,
uint64_t taskId) {
@ -2411,6 +2455,9 @@ SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanN
if (pInfo == NULL || pOperator == NULL) {
goto _error;
}
if (pTableScanNode->pGroupTags) {
taosArraySort(pTableListInfo->pTableList, compareTableKeyInfoByGid);
}
SDataBlockDescNode* pDescNode = pTableScanNode->scan.node.pOutputDataBlockDesc;
@ -2443,7 +2490,6 @@ SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanN
pInfo->dataReaders = taosArrayInit(64, POINTER_BYTES);
pInfo->queryId = queryId;
pInfo->taskId = taskId;
pInfo->currentGroupId = -1;
pInfo->sortSourceParams = taosArrayInit(64, sizeof(STableMergeScanSortSourceParam));

117
source/libs/executor/src/sortoperator.c
View File

@ -142,7 +142,8 @@ SSDataBlock* getSortedBlockData(SSortHandle* pHandle, SSDataBlock* pDataBlock, i
SSDataBlock* loadNextDataBlock(void* param) {
SOperatorInfo* pOperator = (SOperatorInfo*)param;
return pOperator->fpSet.getNextFn(pOperator);
SSDataBlock* pBlock = pOperator->fpSet.getNextFn(pOperator);
return pBlock;
}
// todo refactor: merged with fetch fp
@ -442,7 +443,7 @@ void destroyGroupSortOperatorInfo(void* param, int32_t numOfOutput) {
SOperatorInfo* createGroupSortOperatorInfo(SOperatorInfo* downstream, SGroupSortPhysiNode* pSortPhyNode,
SExecTaskInfo* pTaskInfo) {
SGroupSortOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SGroupSortOperatorInfo));
SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
if (pInfo == NULL || pOperator == NULL /* || rowSize > 100 * 1024 * 1024*/) {
goto _error;
}
@ -474,8 +475,8 @@ SOperatorInfo* createGroupSortOperatorInfo(SOperatorInfo* downstream, SGroupSort
pOperator->exprSupp.numOfExprs = numOfCols;
pOperator->pTaskInfo = pTaskInfo;
pOperator->fpSet = createOperatorFpSet(operatorDummyOpenFn, doGroupSort, NULL, NULL, destroyGroupSortOperatorInfo, NULL,
NULL, getGroupSortExplainExecInfo);
pOperator->fpSet = createOperatorFpSet(operatorDummyOpenFn, doGroupSort, NULL, NULL, destroyGroupSortOperatorInfo,
NULL, NULL, getGroupSortExplainExecInfo);
int32_t code = appendDownstream(pOperator, &downstream, 1);
if (code != TSDB_CODE_SUCCESS) {
@ -493,7 +494,7 @@ _error:
//=====================================================================================
// Multiway Sort Merge operator
typedef struct SMultiwaySortMergeOperatorInfo {
typedef struct SMultiwayMergeOperatorInfo {
SOptrBasicInfo binfo;
int32_t bufPageSize;
@ -503,14 +504,17 @@ typedef struct SMultiwaySortMergeOperatorInfo {
SSortHandle* pSortHandle;
SArray* pColMatchInfo; // for index map from table scan output
SSDataBlock* pInputBlock;
int64_t startTs; // sort start time
uint64_t groupId;
} SMultiwaySortMergeOperatorInfo;
SSDataBlock* pInputBlock;
int64_t startTs; // sort start time
bool groupSort;
bool hasGroupId;
uint64_t groupId;
STupleHandle* prefetchedTuple;
} SMultiwayMergeOperatorInfo;
int32_t doOpenMultiwaySortMergeOperator(SOperatorInfo* pOperator) {
SMultiwaySortMergeOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
int32_t doOpenMultiwayMergeOperator(SOperatorInfo* pOperator) {
SMultiwayMergeOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
if (OPTR_IS_OPENED(pOperator)) {
return TSDB_CODE_SUCCESS;
@ -524,8 +528,9 @@ int32_t doOpenMultiwaySortMergeOperator(SOperatorInfo* pOperator) {
pInfo->pInputBlock, pTaskInfo->id.str);
tsortSetFetchRawDataFp(pInfo->pSortHandle, loadNextDataBlock, NULL, NULL);
tsortSetCompareGroupId(pInfo->pSortHandle, true);
tsortSetCompareGroupId(pInfo->pSortHandle, pInfo->groupSort);
for (int32_t i = 0; i < pOperator->numOfDownstream; ++i) {
SSortSource* ps = taosMemoryCalloc(1, sizeof(SSortSource));
ps->param = pOperator->pDownstream[i];
@ -547,8 +552,8 @@ int32_t doOpenMultiwaySortMergeOperator(SOperatorInfo* pOperator) {
SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pDataBlock, int32_t capacity,
SArray* pColMatchInfo, SOperatorInfo* pOperator) {
SMultiwaySortMergeOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
SMultiwayMergeOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
blockDataCleanup(pDataBlock);
@ -560,18 +565,47 @@ SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pData
blockDataEnsureCapacity(p, capacity);
while (1) {
STupleHandle* pTupleHandle = tsortNextTuple(pHandle);
STupleHandle* pTupleHandle = NULL;
if (pInfo->groupSort) {
if (pInfo->prefetchedTuple == NULL) {
pTupleHandle = tsortNextTuple(pHandle);
} else {
pTupleHandle = pInfo->prefetchedTuple;
pInfo->groupId = tsortGetGroupId(pTupleHandle);
pInfo->prefetchedTuple = NULL;
}
}
else {
pTupleHandle = tsortNextTuple(pHandle);
pInfo->groupId = 0;
}
if (pTupleHandle == NULL) {
break;
}
appendOneRowToDataBlock(p, pTupleHandle);
if (pInfo->groupSort)
{
uint64_t tupleGroupId = tsortGetGroupId(pTupleHandle);
if (!pInfo->hasGroupId) {
pInfo->groupId = tupleGroupId;
pInfo->hasGroupId = true;
appendOneRowToDataBlock(p, pTupleHandle);
} else if (pInfo->groupId == tupleGroupId) {
appendOneRowToDataBlock(p, pTupleHandle);
} else {
pInfo->prefetchedTuple = pTupleHandle;
break;
}
} else {
appendOneRowToDataBlock(p, pTupleHandle);
}
if (p->info.rows >= capacity) {
break;
}
}
if (p->info.rows > 0) {// todo extract method
if (p->info.rows > 0) { // todo extract method
blockDataEnsureCapacity(pDataBlock, p->info.rows);
int32_t numOfCols = taosArrayGetSize(pColMatchInfo);
for (int32_t i = 0; i < numOfCols; ++i) {
@ -593,13 +627,13 @@ SSDataBlock* getMultiwaySortedBlockData(SSortHandle* pHandle, SSDataBlock* pData
return (pDataBlock->info.rows > 0) ? pDataBlock : NULL;
}
SSDataBlock* doMultiwaySortMerge(SOperatorInfo* pOperator) {
SSDataBlock* doMultiwayMerge(SOperatorInfo* pOperator) {
if (pOperator->status == OP_EXEC_DONE) {
return NULL;
}
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
SMultiwaySortMergeOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
SMultiwayMergeOperatorInfo* pInfo = pOperator->info;
int32_t code = pOperator->fpSet._openFn(pOperator);
if (code != TSDB_CODE_SUCCESS) {
@ -608,7 +642,6 @@ SSDataBlock* doMultiwaySortMerge(SOperatorInfo* pOperator) {
SSDataBlock* pBlock = getMultiwaySortedBlockData(pInfo->pSortHandle, pInfo->binfo.pRes,
pOperator->resultInfo.capacity, pInfo->pColMatchInfo, pOperator);
if (pBlock != NULL) {
pOperator->resultInfo.totalRows += pBlock->info.rows;
} else {
@ -617,8 +650,8 @@ SSDataBlock* doMultiwaySortMerge(SOperatorInfo* pOperator) {
return pBlock;
}
void destroyMultiwaySortMergeOperatorInfo(void* param, int32_t numOfOutput) {
SMultiwaySortMergeOperatorInfo* pInfo = (SMultiwaySortMergeOperatorInfo*)param;
void destroyMultiwayMergeOperatorInfo(void* param, int32_t numOfOutput) {
SMultiwayMergeOperatorInfo* pInfo = (SMultiwayMergeOperatorInfo*)param;
pInfo->binfo.pRes = blockDataDestroy(pInfo->binfo.pRes);
pInfo->pInputBlock = blockDataDestroy(pInfo->pInputBlock);
@ -626,11 +659,11 @@ void destroyMultiwaySortMergeOperatorInfo(void* param, int32_t numOfOutput) {
taosArrayDestroy(pInfo->pColMatchInfo);
}
int32_t getMultiwaySortMergeExplainExecInfo(SOperatorInfo* pOptr, void** pOptrExplain, uint32_t* len) {
int32_t getMultiwayMergeExplainExecInfo(SOperatorInfo* pOptr, void** pOptrExplain, uint32_t* len) {
ASSERT(pOptr != NULL);
SSortExecInfo* pInfo = taosMemoryCalloc(1, sizeof(SSortExecInfo));
SMultiwaySortMergeOperatorInfo* pOperatorInfo = (SMultiwaySortMergeOperatorInfo*)pOptr->info;
SMultiwayMergeOperatorInfo* pOperatorInfo = (SMultiwayMergeOperatorInfo*)pOptr->info;
*pInfo = tsortGetSortExecInfo(pOperatorInfo->pSortHandle);
*pOptrExplain = pInfo;
@ -638,24 +671,35 @@ int32_t getMultiwaySortMergeExplainExecInfo(SOperatorInfo* pOptr, void** pOptrEx
return TSDB_CODE_SUCCESS;
}
SOperatorInfo* createMultiwaySortMergeOperatorInfo(SOperatorInfo** downStreams, int32_t numStreams,
SSDataBlock* pInputBlock, SSDataBlock* pResBlock, SArray* pSortInfo,
SArray* pColMatchColInfo, SExecTaskInfo* pTaskInfo) {
SMultiwaySortMergeOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SMultiwaySortMergeOperatorInfo));
SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
int32_t rowSize = pResBlock->info.rowSize;
SOperatorInfo* createMultiwayMergeOperatorInfo(SOperatorInfo** downStreams, size_t numStreams,
SMergePhysiNode* pMergePhyNode, SExecTaskInfo* pTaskInfo) {
SPhysiNode* pPhyNode = (SPhysiNode*)pMergePhyNode;
SMultiwayMergeOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SMultiwayMergeOperatorInfo));
SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
SDataBlockDescNode* pDescNode = pPhyNode->pOutputDataBlockDesc;
SSDataBlock* pResBlock = createResDataBlock(pDescNode);
int32_t rowSize = pResBlock->info.rowSize;
if (pInfo == NULL || pOperator == NULL || rowSize > 100 * 1024 * 1024) {
goto _error;
}
SArray* pSortInfo = createSortInfo(pMergePhyNode->pMergeKeys);
int32_t numOfOutputCols = 0;
SArray* pColMatchColInfo =
extractColMatchInfo(pMergePhyNode->pTargets, pDescNode, &numOfOutputCols, COL_MATCH_FROM_SLOT_ID);
SPhysiNode* pChildNode = (SPhysiNode*)nodesListGetNode(pPhyNode->pChildren, 0);
SSDataBlock* pInputBlock = createResDataBlock(pChildNode->pOutputDataBlockDesc);
initResultSizeInfo(pOperator, 1024);
pInfo->groupSort = pMergePhyNode->groupSort;
pInfo->binfo.pRes = pResBlock;
pInfo->pSortInfo = pSortInfo;
pInfo->pColMatchInfo = pColMatchColInfo;
pInfo->pInputBlock = pInputBlock;
pOperator->name = "MultiwaySortMerge";
pOperator->name = "MultiwayMerge";
pOperator->operatorType = QUERY_NODE_PHYSICAL_PLAN_MERGE;
pOperator->blocking = false;
pOperator->status = OP_NOT_OPENED;
@ -667,9 +711,8 @@ SOperatorInfo* createMultiwaySortMergeOperatorInfo(SOperatorInfo** downStreams,
// one additional is reserved for merged result.
pInfo->sortBufSize = pInfo->bufPageSize * (numStreams + 1);
pOperator->fpSet =
createOperatorFpSet(doOpenMultiwaySortMergeOperator, doMultiwaySortMerge, NULL, NULL,
destroyMultiwaySortMergeOperatorInfo, NULL, NULL, getMultiwaySortMergeExplainExecInfo);
pOperator->fpSet = createOperatorFpSet(doOpenMultiwayMergeOperator, doMultiwayMerge, NULL, NULL,
destroyMultiwayMergeOperatorInfo, NULL, NULL, getMultiwayMergeExplainExecInfo);
int32_t code = appendDownstream(pOperator, downStreams, numStreams);
if (code != TSDB_CODE_SUCCESS) {

465
source/libs/executor/src/timewindowoperator.c
View File

@ -15,6 +15,7 @@
#include "executorimpl.h"
#include "function.h"
#include "functionMgt.h"
#include "tcompare.h"
#include "tdatablock.h"
#include "tfill.h"
#include "ttime.h"
@ -26,6 +27,16 @@ typedef enum SResultTsInterpType {
#define IS_FINAL_OP(op) ((op)->isFinal)
typedef struct SWinRes {
TSKEY ts;
uint64_t groupId;
} SWinRes;
typedef struct SPullWindowInfo {
STimeWindow window;
uint64_t groupId;
} SPullWindowInfo;
static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator);
static int64_t* extractTsCol(SSDataBlock* pBlock, const SIntervalAggOperatorInfo* pInfo);
@ -684,11 +695,13 @@ static void doInterpUnclosedTimeWindow(SOperatorInfo* pOperatorInfo, int32_t num
}
void printDataBlock(SSDataBlock* pBlock, const char* flag) {
if (pBlock == NULL) return;
SArray* blocks = taosArrayInit(1, sizeof(SSDataBlock));
taosArrayPush(blocks, pBlock);
blockDebugShowData(blocks, flag);
taosArrayDestroy(blocks);
if (pBlock == NULL){
qDebug("======printDataBlock Block is Null");
return;
}
char *pBuf = NULL;
qDebug("%s", dumpBlockData(pBlock, flag, &pBuf));
taosMemoryFree(pBuf);
}
typedef int64_t (*__get_value_fn_t)(void* data, int32_t index);
@ -956,6 +969,12 @@ static int32_t doOpenIntervalAgg(SOperatorInfo* pOperator) {
getTableScanInfo(pOperator, &pInfo->order, &scanFlag);
if (pInfo->scalarSupp.pExprInfo != NULL) {
SExprSupp* pExprSup =& pInfo->scalarSupp;
projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx,
pExprSup->numOfExprs, NULL);
}
// the pDataBlock are always the same one, no need to call this again
setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->order, scanFlag, true);
hashIntervalAgg(pOperator, &pInfo->binfo.resultRowInfo, pBlock, scanFlag, NULL);
@ -1217,30 +1236,40 @@ void doClearWindowImpl(SResultRowPosition* p1, SDiskbasedBuf* pResultBuf, SExprS
}
}
void doClearWindow(SAggSupporter* pAggSup, SExprSupp* pSup, char* pData, int16_t bytes, uint64_t groupId,
bool doClearWindow(SAggSupporter* pAggSup, SExprSupp* pSup, char* pData, int16_t bytes, uint64_t groupId,
int32_t numOfOutput) {
SET_RES_WINDOW_KEY(pAggSup->keyBuf, pData, bytes, groupId);
SResultRowPosition* p1 =
(SResultRowPosition*)taosHashGet(pAggSup->pResultRowHashTable, pAggSup->keyBuf, GET_RES_WINDOW_KEY_LEN(bytes));
if (!p1) {
// window has been closed
return;
return false;
}
doClearWindowImpl(p1, pAggSup->pResultBuf, pSup, numOfOutput);
return true;
}
static void doClearWindows(SAggSupporter* pAggSup, SExprSupp* pSup1, SInterval* pInterval, int32_t tsIndex,
int32_t numOfOutput, SSDataBlock* pBlock, SArray* pUpWins) {
SColumnInfoData* pColDataInfo = taosArrayGet(pBlock->pDataBlock, tsIndex);
TSKEY* tsCols = (TSKEY*)pColDataInfo->pData;
SColumnInfoData* pTsCol = taosArrayGet(pBlock->pDataBlock, tsIndex);
TSKEY* tsCols = (TSKEY*)pTsCol->pData;
uint64_t* pGpDatas = NULL;
if (pBlock->info.type == STREAM_RETRIEVE) {
SColumnInfoData* pGpCol = taosArrayGet(pBlock->pDataBlock, 2);
pGpDatas = (uint64_t*)pGpCol->pData;
}
int32_t step = 0;
for (int32_t i = 0; i < pBlock->info.rows; i += step) {
SResultRowInfo dumyInfo;
dumyInfo.cur.pageId = -1;
STimeWindow win = getActiveTimeWindow(NULL, &dumyInfo, tsCols[i], pInterval, pInterval->precision, NULL);
step = getNumOfRowsInTimeWindow(&pBlock->info, tsCols, i, win.ekey, binarySearchForKey, NULL, TSDB_ORDER_ASC);
doClearWindow(pAggSup, pSup1, (char*)&win.skey, sizeof(TKEY), pBlock->info.groupId, numOfOutput);
if (pUpWins) {
uint64_t groupId = pBlock->info.groupId;
if (pGpDatas) {
groupId = pGpDatas[i];
}
bool res = doClearWindow(pAggSup, pSup1, (char*)&win.skey, sizeof(TKEY), groupId, numOfOutput);
if (pUpWins && res) {
taosArrayPush(pUpWins, &win);
}
}
@ -1268,8 +1297,8 @@ bool isCloseWindow(STimeWindow* pWin, STimeWindowAggSupp* pSup) {
return pSup->maxTs != INT64_MIN && pWin->ekey < pSup->maxTs - pSup->waterMark;
}
static int32_t closeIntervalWindow(SHashObj* pHashMap, STimeWindowAggSupp* pSup, SInterval* pInterval,
SArray* closeWins) {
static int32_t closeIntervalWindow(SHashObj* pHashMap, STimeWindowAggSupp* pSup,
SInterval* pInterval, SHashObj* pPullDataMap, SArray* closeWins) {
void* pIte = NULL;
size_t keyLen = 0;
while ((pIte = taosHashIterate(pHashMap, pIte)) != NULL) {
@ -1280,10 +1309,20 @@ static int32_t closeIntervalWindow(SHashObj* pHashMap, STimeWindowAggSupp* pSup,
SResultRowInfo dumyInfo;
dumyInfo.cur.pageId = -1;
STimeWindow win = getActiveTimeWindow(NULL, &dumyInfo, ts, pInterval, pInterval->precision, NULL);
SWinRes winRe = {.ts = win.skey, .groupId = groupId,};
void* chIds = taosHashGet(pPullDataMap, &winRe, sizeof(SWinRes));
if (isCloseWindow(&win, pSup)) {
char keyBuf[GET_RES_WINDOW_KEY_LEN(sizeof(TSKEY))];
SET_RES_WINDOW_KEY(keyBuf, &ts, sizeof(TSKEY), groupId);
taosHashRemove(pHashMap, keyBuf, keyLen);
if (chIds && pPullDataMap) {
SArray* chAy = *(SArray**) chIds;
int32_t size = taosArrayGetSize(chAy);
qInfo("======window %ld wait child size:%d", win.skey ,size);
for (int32_t i = 0; i < size; i++) {
qInfo("======window %ld wait chid id:%d", win.skey ,*(int32_t*)taosArrayGet(chAy, i));
}
continue;
} else if (pPullDataMap) {
qInfo("======close window %ld", win.skey);
}
SResultRowPosition* pPos = (SResultRowPosition*)pIte;
if (pSup->calTrigger == STREAM_TRIGGER_WINDOW_CLOSE) {
int32_t code = saveResult(ts, pPos->pageId, pPos->offset, groupId, closeWins);
@ -1291,11 +1330,25 @@ static int32_t closeIntervalWindow(SHashObj* pHashMap, STimeWindowAggSupp* pSup,
return code;
}
}
char keyBuf[GET_RES_WINDOW_KEY_LEN(sizeof(TSKEY))];
SET_RES_WINDOW_KEY(keyBuf, &ts, sizeof(TSKEY), groupId);
taosHashRemove(pHashMap, keyBuf, keyLen);
}
}
return TSDB_CODE_SUCCESS;
}
static void closeChildIntervalWindow(SArray* pChildren, TSKEY maxTs) {
int32_t size = taosArrayGetSize(pChildren);
for (int32_t i = 0; i < size; i++) {
SOperatorInfo* pChildOp = taosArrayGetP(pChildren, i);
SStreamFinalIntervalOperatorInfo* pChInfo = pChildOp->info;
ASSERT(pChInfo->twAggSup.calTrigger == STREAM_TRIGGER_AT_ONCE);
pChInfo->twAggSup.maxTs = TMAX(pChInfo->twAggSup.maxTs, maxTs);
closeIntervalWindow(pChInfo->aggSup.pResultRowHashTable, &pChInfo->twAggSup, &pChInfo->interval, NULL, NULL);
}
}
static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
SIntervalAggOperatorInfo* pInfo = pOperator->info;
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
@ -1324,7 +1377,7 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
break;
}
if (pBlock->info.type == STREAM_REPROCESS) {
if (pBlock->info.type == STREAM_CLEAR) {
doClearWindows(&pInfo->aggSup, &pOperator->exprSupp, &pInfo->interval, 0, pOperator->exprSupp.numOfExprs, pBlock,
NULL);
qDebug("%s clear existed time window results for updates checked", GET_TASKID(pTaskInfo));
@ -1334,6 +1387,11 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
continue;
}
if (pInfo->scalarSupp.pExprInfo != NULL) {
SExprSupp* pExprSup = &pInfo->scalarSupp;
projectApplyFunctions(pExprSup->pExprInfo, pBlock, pBlock, pExprSup->pCtx, pExprSup->numOfExprs, NULL);
}
// The timewindow that overlaps the timestamps of the input pBlock need to be recalculated and return to the
// caller. Note that all the time window are not close till now.
// the pDataBlock are always the same one, no need to call this again
@ -1345,7 +1403,7 @@ static SSDataBlock* doStreamIntervalAgg(SOperatorInfo* pOperator) {
pInfo->twAggSup.maxTs = TMAX(pInfo->twAggSup.maxTs, pBlock->info.window.ekey);
hashIntervalAgg(pOperator, &pInfo->binfo.resultRowInfo, pBlock, MAIN_SCAN, pUpdated);
}
closeIntervalWindow(pInfo->aggSup.pResultRowHashTable, &pInfo->twAggSup, &pInfo->interval, pUpdated);
closeIntervalWindow(pInfo->aggSup.pResultRowHashTable, &pInfo->twAggSup, &pInfo->interval, NULL, pUpdated);
finalizeUpdatedResult(pOperator->exprSupp.numOfExprs, pInfo->aggSup.pResultBuf, pUpdated, pSup->rowEntryInfoOffset);
initMultiResInfoFromArrayList(&pInfo->groupResInfo, pUpdated);
@ -1373,6 +1431,11 @@ void destroyStreamFinalIntervalOperatorInfo(void* param, int32_t numOfOutput) {
SStreamFinalIntervalOperatorInfo* pInfo = (SStreamFinalIntervalOperatorInfo*)param;
cleanupBasicInfo(&pInfo->binfo);
cleanupAggSup(&pInfo->aggSup);
//it should be empty.
taosHashCleanup(pInfo->pPullDataMap);
taosArrayDestroy(pInfo->pPullWins);
blockDataDestroy(pInfo->pPullDataRes);
if (pInfo->pChildren) {
int32_t size = taosArrayGetSize(pInfo->pChildren);
for (int32_t i = 0; i < size; i++) {
@ -1446,18 +1509,27 @@ void increaseTs(SqlFunctionCtx* pCtx) {
SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SExprInfo* pExprInfo, int32_t numOfCols,
SSDataBlock* pResBlock, SInterval* pInterval, int32_t primaryTsSlotId,
STimeWindowAggSupp* pTwAggSupp, SExecTaskInfo* pTaskInfo, bool isStream) {
STimeWindowAggSupp* pTwAggSupp, SIntervalPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, bool isStream) {
SIntervalAggOperatorInfo* pInfo = taosMemoryCalloc(1, sizeof(SIntervalAggOperatorInfo));
SOperatorInfo* pOperator = taosMemoryCalloc(1, sizeof(SOperatorInfo));
if (pInfo == NULL || pOperator == NULL) {
goto _error;
}
pInfo->win = pTaskInfo->window;
pInfo->order = TSDB_ORDER_ASC;
pInfo->interval = *pInterval;
pInfo->win = pTaskInfo->window;
pInfo->order = TSDB_ORDER_ASC;
pInfo->interval = *pInterval;
pInfo->execModel = pTaskInfo->execModel;
pInfo->twAggSup = *pTwAggSupp;
pInfo->twAggSup = *pTwAggSupp;
if (pPhyNode->window.pExprs != NULL) {
int32_t numOfScalar = 0;
SExprInfo* pScalarExprInfo = createExprInfo(pPhyNode->window.pExprs, NULL, &numOfScalar);
int32_t code = initExprSupp(&pInfo->scalarSupp, pScalarExprInfo, numOfScalar);
if (code != TSDB_CODE_SUCCESS) {
goto _error;
}
}
pInfo->primaryTsIndex = primaryTsSlotId;
@ -2164,6 +2236,24 @@ bool isDeletedWindow(STimeWindow* pWin, uint64_t groupId, SAggSupporter* pSup) {
return p1 == NULL;
}
int32_t getNexWindowPos(SInterval* pInterval, SDataBlockInfo* pBlockInfo, TSKEY* tsCols,
int32_t startPos, TSKEY eKey, STimeWindow* pNextWin) {
int32_t forwardRows = getNumOfRowsInTimeWindow(pBlockInfo, tsCols, startPos,
eKey, binarySearchForKey, NULL, TSDB_ORDER_ASC);
int32_t prevEndPos = forwardRows - 1 + startPos;
return getNextQualifiedWindow(pInterval, pNextWin, pBlockInfo, tsCols, prevEndPos, TSDB_ORDER_ASC);
}
void addPullWindow(SHashObj* pMap, SWinRes* pWinRes, int32_t size) {
SArray* childIds = taosArrayInit(8, sizeof(int32_t));
for (int32_t i = 0; i < size; i++) {
taosArrayPush(childIds, &i);
}
taosHashPut(pMap, pWinRes, sizeof(SWinRes), &childIds, sizeof(void*));
}
static int32_t getChildIndex(SSDataBlock* pBlock) { return pBlock->info.childId; }
static void doHashInterval(SOperatorInfo* pOperatorInfo, SSDataBlock* pSDataBlock, uint64_t tableGroupId,
SArray* pUpdated) {
SStreamFinalIntervalOperatorInfo* pInfo = (SStreamFinalIntervalOperatorInfo*)pOperatorInfo->info;
@ -2177,35 +2267,59 @@ static void doHashInterval(SOperatorInfo* pOperatorInfo, SSDataBlock* pSDataBloc
SResultRow* pResult = NULL;
int32_t forwardRows = 0;
if (pSDataBlock->pDataBlock != NULL) {
SColumnInfoData* pColDataInfo = taosArrayGet(pSDataBlock->pDataBlock, pInfo->primaryTsIndex);
tsCols = (int64_t*)pColDataInfo->pData;
} else {
return;
}
ASSERT(pSDataBlock->pDataBlock != NULL);
SColumnInfoData* pColDataInfo = taosArrayGet(pSDataBlock->pDataBlock, pInfo->primaryTsIndex);
tsCols = (int64_t*)pColDataInfo->pData;
int32_t startPos = ascScan ? 0 : (pSDataBlock->info.rows - 1);
TSKEY ts = getStartTsKey(&pSDataBlock->info.window, tsCols);
STimeWindow nextWin = getActiveTimeWindow(pInfo->aggSup.pResultBuf, pResultRowInfo, ts, &pInfo->interval,
pInfo->interval.precision, NULL);
while (1) {
if (IS_FINAL_OP(pInfo) && isCloseWindow(&nextWin, &pInfo->twAggSup) &&
isDeletedWindow(&nextWin, tableGroupId, &pInfo->aggSup)) {
SArray* pUpWins = taosArrayInit(8, sizeof(STimeWindow));
taosArrayPush(pUpWins, &nextWin);
rebuildIntervalWindow(pInfo, pSup, pUpWins, pInfo->binfo.pRes->info.groupId, pSup->numOfExprs,
pOperatorInfo->pTaskInfo);
taosArrayDestroy(pUpWins);
if (IS_FINAL_OP(pInfo) && isCloseWindow(&nextWin, &pInfo->twAggSup) && pInfo->pChildren) {
bool ignore = true;
SWinRes winRes = {.ts = nextWin.skey, .groupId = tableGroupId,};
void* chIds = taosHashGet(pInfo->pPullDataMap, &winRes, sizeof(SWinRes));
if (isDeletedWindow(&nextWin, tableGroupId, &pInfo->aggSup) && !chIds) {
SPullWindowInfo pull = {.window = nextWin, .groupId = tableGroupId};
// add pull data request
taosArrayPush(pInfo->pPullWins, &pull);
addPullWindow(pInfo->pPullDataMap, &winRes, taosArrayGetSize(pInfo->pChildren));
} else {
int32_t index = -1;
SArray* chArray = NULL;
if (chIds) {
chArray = *(void**) chIds;
int32_t chId = getChildIndex(pSDataBlock);
index = taosArraySearchIdx(chArray, &chId, compareInt32Val, TD_EQ);
}
if (index != -1 && pSDataBlock->info.type == STREAM_PUSH_DATA) {
taosArrayRemove(chArray, index);
if (taosArrayGetSize(chArray) == 0) {
// pull data is over
taosHashRemove(pInfo->pPullDataMap, &winRes, sizeof(SWinRes));
}
}
if ( index == -1 || pSDataBlock->info.type == STREAM_PUSH_DATA) {
ignore = false;
}
}
if (ignore) {
startPos = getNexWindowPos(&pInfo->interval, &pSDataBlock->info, tsCols, startPos, nextWin.ekey, &nextWin);
if (startPos < 0) {
break;
}
continue;
}
}
int32_t code = setTimeWindowOutputBuf(pResultRowInfo, &nextWin, true, &pResult, tableGroupId, pSup->pCtx,
numOfOutput, pSup->rowEntryInfoOffset, &pInfo->aggSup, pTaskInfo);
if (code != TSDB_CODE_SUCCESS || pResult == NULL) {
longjmp(pTaskInfo->env, TSDB_CODE_QRY_OUT_OF_MEMORY);
}
SResKeyPos* pos = taosMemoryMalloc(sizeof(SResKeyPos) + sizeof(uint64_t));
pos->groupId = tableGroupId;
pos->pos = (SResultRowPosition){.pageId = pResult->pageId, .offset = pResult->offset};
*(int64_t*)pos->key = pResult->win.skey;
forwardRows = getNumOfRowsInTimeWindow(&pSDataBlock->info, tsCols, startPos, nextWin.ekey, binarySearchForKey, NULL,
TSDB_ORDER_ASC);
if (pInfo->twAggSup.calTrigger == STREAM_TRIGGER_AT_ONCE && pUpdated) {
@ -2230,14 +2344,17 @@ static void clearStreamIntervalOperator(SStreamFinalIntervalOperatorInfo* pInfo)
initResultRowInfo(&pInfo->binfo.resultRowInfo);
}
static void clearUpdateDataBlock(SSDataBlock* pBlock) {
static void clearSpecialDataBlock(SSDataBlock* pBlock) {
if (pBlock->info.rows <= 0) {
return;
}
blockDataCleanup(pBlock);
}
void copyUpdateDataBlock(SSDataBlock* pDest, SSDataBlock* pSource, int32_t tsColIndex) {
// ASSERT(pDest->info.capacity >= pSource->info.rows);
blockDataEnsureCapacity(pDest, pSource->info.rows);
clearUpdateDataBlock(pDest);
clearSpecialDataBlock(pDest);
SColumnInfoData* pDestCol = taosArrayGet(pDest->pDataBlock, 0);
SColumnInfoData* pSourceCol = taosArrayGet(pSource->pDataBlock, tsColIndex);
@ -2254,7 +2371,63 @@ void copyUpdateDataBlock(SSDataBlock* pDest, SSDataBlock* pSource, int32_t tsCol
blockDataUpdateTsWindow(pDest, 0);
}
static int32_t getChildIndex(SSDataBlock* pBlock) { return pBlock->info.childId; }
static bool needBreak(SStreamFinalIntervalOperatorInfo* pInfo) {
int32_t size = taosArrayGetSize(pInfo->pPullWins);
if (pInfo->pullIndex < size) {
return true;
}
return false;
}
static void doBuildPullDataBlock(SArray* array, int32_t* pIndex, SSDataBlock* pBlock) {
clearSpecialDataBlock(pBlock);
int32_t size = taosArrayGetSize(array);
if (size - (*pIndex) == 0) {
return;
}
blockDataEnsureCapacity(pBlock, size - (*pIndex) );
ASSERT(3 <= taosArrayGetSize(pBlock->pDataBlock));
for (; (*pIndex) < size; (*pIndex)++) {
SPullWindowInfo* pWin = taosArrayGet(array, (*pIndex) );
SColumnInfoData* pStartTs = (SColumnInfoData*) taosArrayGet(pBlock->pDataBlock, 0);
colDataAppend(pStartTs, pBlock->info.rows, (const char*)&pWin->window.skey, false);
SColumnInfoData* pEndTs = (SColumnInfoData*) taosArrayGet(pBlock->pDataBlock, 1);
colDataAppend(pEndTs, pBlock->info.rows, (const char*)&pWin->window.ekey, false);
SColumnInfoData* pGroupId = (SColumnInfoData*) taosArrayGet(pBlock->pDataBlock, 2);
colDataAppend(pGroupId, pBlock->info.rows, (const char*)&pWin->groupId, false);
pBlock->info.rows++;
}
if ((*pIndex) == size) {
*pIndex = 0;
taosArrayClear(array);
}
blockDataUpdateTsWindow(pBlock, 0);
}
void processPushEmpty(SSDataBlock* pBlock, SHashObj* pMap) {
SColumnInfoData* pStartCol = taosArrayGet(pBlock->pDataBlock, 0);
TSKEY* tsData = (TSKEY*)pStartCol->pData;
SColumnInfoData* pGroupCol = taosArrayGet(pBlock->pDataBlock, 2);
uint64_t* groupIdData = (uint64_t*)pGroupCol->pData;
int32_t chId = getChildIndex(pBlock);
for (int32_t i = 0; i < pBlock->info.rows; i++) {
SWinRes winRes = {.ts = tsData[i], .groupId = groupIdData[i]};
void* chIds = taosHashGet(pMap, &winRes, sizeof(SWinRes));
if (chIds) {
SArray* chArray = *(SArray**) chIds;
int32_t index = taosArraySearchIdx(chArray, &chId, compareInt32Val, TD_EQ);
if (index != -1) {
taosArrayRemove(chArray, index);
if (taosArrayGetSize(chArray) == 0) {
// pull data is over
taosHashRemove(pMap, &winRes, sizeof(SWinRes));
}
}
}
}
}
static SSDataBlock* doStreamFinalIntervalAgg(SOperatorInfo* pOperator) {
SStreamFinalIntervalOperatorInfo* pInfo = pOperator->info;
@ -2270,35 +2443,57 @@ static SSDataBlock* doStreamFinalIntervalAgg(SOperatorInfo* pOperator) {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
if (pInfo->binfo.pRes->info.rows == 0) {
pOperator->status = OP_EXEC_DONE;
if (IS_FINAL_OP(pInfo) || pInfo->pUpdateRes->info.rows == 0) {
if (!IS_FINAL_OP(pInfo)) {
// semi interval operator clear disk buffer
clearStreamIntervalOperator(pInfo);
}
return NULL;
if (!IS_FINAL_OP(pInfo)) {
// semi interval operator clear disk buffer
clearStreamIntervalOperator(pInfo);
}
return NULL;
}
printDataBlock(pInfo->binfo.pRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
return pInfo->binfo.pRes;
} else {
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
if (pInfo->binfo.pRes->info.rows != 0) {
printDataBlock(pInfo->binfo.pRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
return pInfo->binfo.pRes;
}
if (pInfo->pUpdateRes->info.rows != 0 && pInfo->returnUpdate) {
pInfo->returnUpdate = false;
ASSERT(!IS_FINAL_OP(pInfo));
printDataBlock(pInfo->pUpdateRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
// process the rest of the data
pOperator->status = OP_OPENED;
return pInfo->pUpdateRes;
}
return pInfo->binfo.pRes;
doBuildPullDataBlock(pInfo->pPullWins, &pInfo->pullIndex, pInfo->pPullDataRes);
if (pInfo->pPullDataRes->info.rows != 0) {
// process the rest of the data
ASSERT(IS_FINAL_OP(pInfo));
printDataBlock(pInfo->pPullDataRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
return pInfo->pPullDataRes;
}
}
while (1) {
SSDataBlock* pBlock = downstream->fpSet.getNextFn(downstream);
if (pBlock == NULL) {
clearUpdateDataBlock(pInfo->pUpdateRes);
clearSpecialDataBlock(pInfo->pUpdateRes);
pOperator->status = OP_RES_TO_RETURN;
qInfo("Stream Final Interval return data");
break;
}
printDataBlock(pBlock, IS_FINAL_OP(pInfo) ? "interval Final recv" : "interval Semi recv");
maxTs = TMAX(maxTs, pBlock->info.window.ekey);
if (pBlock->info.type == STREAM_REPROCESS) {
if (pBlock->info.type == STREAM_NORMAL || pBlock->info.type == STREAM_PUSH_DATA || pBlock->info.type == STREAM_INVALID) {
pInfo->binfo.pRes->info.type = pBlock->info.type;
} else if (pBlock->info.type == STREAM_CLEAR) {
SArray* pUpWins = taosArrayInit(8, sizeof(STimeWindow));
doClearWindows(&pInfo->aggSup, pSup, &pInfo->interval, pInfo->primaryTsIndex, pOperator->exprSupp.numOfExprs,
pBlock, pUpWins);
if (IS_FINAL_OP(pInfo)) {
int32_t childIndex = getChildIndex(pBlock);
SOperatorInfo* pChildOp = taosArrayGetP(pInfo->pChildren, childIndex);
SIntervalAggOperatorInfo* pChildInfo = pChildOp->info;
SStreamFinalIntervalOperatorInfo* pChildInfo = pChildOp->info;
SExprSupp* pChildSup = &pChildOp->exprSupp;
doClearWindows(&pChildInfo->aggSup, pChildSup, &pChildInfo->interval, pChildInfo->primaryTsIndex,
@ -2310,11 +2505,25 @@ static SSDataBlock* doStreamFinalIntervalAgg(SOperatorInfo* pOperator) {
}
removeResults(pUpWins, pUpdated);
copyUpdateDataBlock(pInfo->pUpdateRes, pBlock, pInfo->primaryTsIndex);
pInfo->returnUpdate = true;
taosArrayDestroy(pUpWins);
break;
} else if (pBlock->info.type == STREAM_GET_ALL && IS_FINAL_OP(pInfo)) {
getAllIntervalWindow(pInfo->aggSup.pResultRowHashTable, pUpdated);
continue;
} else if (pBlock->info.type == STREAM_RETRIEVE && !IS_FINAL_OP(pInfo)) {
SArray* pUpWins = taosArrayInit(8, sizeof(STimeWindow));
doClearWindows(&pInfo->aggSup, pSup, &pInfo->interval, 0, pOperator->exprSupp.numOfExprs,
pBlock, pUpWins);
removeResults(pUpWins, pUpdated);
taosArrayDestroy(pUpWins);
if (taosArrayGetSize(pUpdated) > 0) {
break;
}
continue;
} else if (pBlock->info.type == STREAM_PUSH_EMPTY && IS_FINAL_OP(pInfo)) {
processPushEmpty(pBlock, pInfo->pPullDataMap);
continue;
}
setInputDataBlock(pOperator, pSup->pCtx, pBlock, pInfo->order, MAIN_SCAN, true);
@ -2334,31 +2543,70 @@ static SSDataBlock* doStreamFinalIntervalAgg(SOperatorInfo* pOperator) {
SStreamFinalIntervalOperatorInfo* pChInfo = pChildOp->info;
setInputDataBlock(pChildOp, pChildOp->exprSupp.pCtx, pBlock, pChInfo->order, MAIN_SCAN, true);
doHashInterval(pChildOp, pBlock, pBlock->info.groupId, NULL);
pChInfo->twAggSup.maxTs = TMAX(pChInfo->twAggSup.maxTs, pBlock->info.window.ekey);
if (needBreak(pInfo)) {
break;
}
}
maxTs = TMAX(maxTs, pBlock->info.window.ekey);
}
pInfo->twAggSup.maxTs = TMAX(pInfo->twAggSup.maxTs, maxTs);
if (IS_FINAL_OP(pInfo)) {
closeIntervalWindow(pInfo->aggSup.pResultRowHashTable, &pInfo->twAggSup, &pInfo->interval, pUpdated);
closeIntervalWindow(pInfo->aggSup.pResultRowHashTable, &pInfo->twAggSup,
&pInfo->interval, pInfo->pPullDataMap, pUpdated);
closeChildIntervalWindow(pInfo->pChildren, pInfo->twAggSup.maxTs);
}
finalizeUpdatedResult(pOperator->exprSupp.numOfExprs, pInfo->aggSup.pResultBuf, pUpdated, pSup->rowEntryInfoOffset);
initMultiResInfoFromArrayList(&pInfo->groupResInfo, pUpdated);
blockDataEnsureCapacity(pInfo->binfo.pRes, pOperator->resultInfo.capacity);
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->aggSup.pResultBuf);
pOperator->status = OP_RES_TO_RETURN;
if (pInfo->binfo.pRes->info.rows == 0) {
pOperator->status = OP_EXEC_DONE;
if (pInfo->pUpdateRes->info.rows == 0) {
return NULL;
}
if (pInfo->binfo.pRes->info.rows != 0) {
printDataBlock(pInfo->binfo.pRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
return pInfo->binfo.pRes;
}
if (pInfo->pUpdateRes->info.rows != 0 && pInfo->returnUpdate) {
pInfo->returnUpdate = false;
ASSERT(!IS_FINAL_OP(pInfo));
printDataBlock(pInfo->pUpdateRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
// process the rest of the data
pOperator->status = OP_OPENED;
return pInfo->pUpdateRes;
}
return pInfo->binfo.pRes;
doBuildPullDataBlock(pInfo->pPullWins, &pInfo->pullIndex, pInfo->pPullDataRes);
if (pInfo->pPullDataRes->info.rows != 0) {
// process the rest of the data
ASSERT(IS_FINAL_OP(pInfo));
printDataBlock(pInfo->pPullDataRes, IS_FINAL_OP(pInfo) ? "interval Final" : "interval Semi");
return pInfo->pPullDataRes;
}
// ASSERT(false);
return NULL;
}
SSDataBlock* createPullDataBlock() {
SSDataBlock* pBlock = taosMemoryCalloc(1, sizeof(SSDataBlock));
pBlock->info.hasVarCol = false;
pBlock->info.groupId = 0;
pBlock->info.rows = 0;
pBlock->info.type = STREAM_RETRIEVE;
pBlock->info.rowSize = sizeof(TSKEY) + sizeof(TSKEY) + sizeof(uint64_t);
pBlock->pDataBlock = taosArrayInit(3, sizeof(SColumnInfoData));
SColumnInfoData infoData = {0};
infoData.info.type = TSDB_DATA_TYPE_TIMESTAMP;
infoData.info.bytes = sizeof(TSKEY);
// window start ts
taosArrayPush(pBlock->pDataBlock, &infoData);
// window end ts
taosArrayPush(pBlock->pDataBlock, &infoData);
infoData.info.type = TSDB_DATA_TYPE_UBIGINT;
infoData.info.bytes = sizeof(uint64_t);
taosArrayPush(pBlock->pDataBlock, &infoData);
return pBlock;
}
SOperatorInfo* createStreamFinalIntervalOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode,
@ -2412,23 +2660,30 @@ SOperatorInfo* createStreamFinalIntervalOperatorInfo(SOperatorInfo* downstream,
goto _error;
}
}
// semi interval operator does not catch result
pInfo->pUpdateRes = createResDataBlock(pPhyNode->pOutputDataBlockDesc);
pInfo->pUpdateRes->info.type = STREAM_REPROCESS;
pInfo->pUpdateRes->info.type = STREAM_CLEAR;
blockDataEnsureCapacity(pInfo->pUpdateRes, 128);
pInfo->returnUpdate = false;
pInfo->pPhyNode = (SPhysiNode*)nodesCloneNode((SNode*)pPhyNode);
if (pPhyNode->type == QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_INTERVAL) {
pInfo->isFinal = true;
pOperator->name = "StreamFinalIntervalOperator";
} else {
// semi interval operator does not catch result
pInfo->isFinal = false;
pOperator->name = "StreamSemiIntervalOperator";
}
if (!IS_FINAL_OP(pInfo)) {
if (!IS_FINAL_OP(pInfo) || numOfChild == 0) {
pInfo->twAggSup.calTrigger = STREAM_TRIGGER_AT_ONCE;
}
pInfo->pPullWins = taosArrayInit(8, sizeof(SPullWindowInfo));
pInfo->pullIndex = 0;
_hash_fn_t hashFn = taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY);
pInfo->pPullDataMap = taosHashInit(64, hashFn, false, HASH_NO_LOCK);
pInfo->pPullDataRes = createPullDataBlock();
pOperator->operatorType = pPhyNode->type;
pOperator->blocking = true;
@ -2811,11 +3066,6 @@ void compactTimeWindow(SStreamSessionAggOperatorInfo* pInfo, int32_t startIndex,
}
}
typedef struct SWinRes {
TSKEY ts;
uint64_t groupId;
} SWinRes;
static void doStreamSessionAggImpl(SOperatorInfo* pOperator, SSDataBlock* pSDataBlock, SHashObj* pStUpdated,
SHashObj* pStDeleted, bool hasEndTs) {
SExecTaskInfo* pTaskInfo = pOperator->pTaskInfo;
@ -3027,14 +3277,14 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) {
} else if (pOperator->status == OP_RES_TO_RETURN) {
doBuildDeleteDataBlock(pInfo->pStDeleted, pInfo->pDelRes, &pInfo->pDelIterator);
if (pInfo->pDelRes->info.rows > 0) {
/*printDataBlock(pInfo->pDelRes, "session del");*/
printDataBlock(pInfo->pDelRes, IS_FINAL_OP(pInfo)? "Final Session" : "Single Session");
return pInfo->pDelRes;
}
doBuildResultDatablock(pOperator, pBInfo, &pInfo->groupResInfo, pInfo->streamAggSup.pResultBuf);
if (pBInfo->pRes->info.rows == 0 || !hasDataInGroupInfo(&pInfo->groupResInfo)) {
doSetOperatorCompleted(pOperator);
}
/*printDataBlock(pBInfo->pRes, "session insert");*/
printDataBlock(pBInfo->pRes, IS_FINAL_OP(pInfo)? "Final Session" : "Single Session");
return pBInfo->pRes->info.rows == 0 ? NULL : pBInfo->pRes;
}
@ -3048,7 +3298,7 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) {
break;
}
if (pBlock->info.type == STREAM_REPROCESS) {
if (pBlock->info.type == STREAM_CLEAR) {
SArray* pWins = taosArrayInit(16, sizeof(SResultWindowInfo));
doClearSessionWindows(&pInfo->streamAggSup, &pOperator->exprSupp, pBlock, 0, pOperator->exprSupp.numOfExprs,
pInfo->gap, pWins);
@ -3102,11 +3352,11 @@ static SSDataBlock* doStreamSessionAgg(SOperatorInfo* pOperator) {
blockDataEnsureCapacity(pInfo->binfo.pRes, pOperator->resultInfo.capacity);
doBuildDeleteDataBlock(pInfo->pStDeleted, pInfo->pDelRes, &pInfo->pDelIterator);
if (pInfo->pDelRes->info.rows > 0) {
/*printDataBlock(pInfo->pDelRes, "session del");*/
printDataBlock(pInfo->pDelRes, IS_FINAL_OP(pInfo)? "Final Session" : "Single Session");
return pInfo->pDelRes;
}
doBuildResultDatablock(pOperator, &pInfo->binfo, &pInfo->groupResInfo, pInfo->streamAggSup.pResultBuf);
/*printDataBlock(pBInfo->pRes, "session insert");*/
printDataBlock(pBInfo->pRes, IS_FINAL_OP(pInfo)? "Final Session" : "Single Session");
return pBInfo->pRes->info.rows == 0 ? NULL : pBInfo->pRes;
}
@ -3169,11 +3419,11 @@ static SSDataBlock* doStreamSessionSemiAgg(SOperatorInfo* pOperator) {
while (1) {
SSDataBlock* pBlock = downstream->fpSet.getNextFn(downstream);
if (pBlock == NULL) {
clearUpdateDataBlock(pInfo->pUpdateRes);
clearSpecialDataBlock(pInfo->pUpdateRes);
break;
}
if (pBlock->info.type == STREAM_REPROCESS) {
if (pBlock->info.type == STREAM_CLEAR) {
SArray* pWins = taosArrayInit(16, sizeof(SResultWindowInfo));
doClearSessionWindows(&pInfo->streamAggSup, pSup, pBlock, 0, pSup->numOfExprs, pInfo->gap, pWins);
removeSessionResults(pStUpdated, pWins);
@ -3236,7 +3486,7 @@ SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream
} else {
pInfo->isFinal = false;
pInfo->pUpdateRes = createResDataBlock(pPhyNode->pOutputDataBlockDesc);
pInfo->pUpdateRes->info.type = STREAM_REPROCESS;
pInfo->pUpdateRes->info.type = STREAM_CLEAR;
blockDataEnsureCapacity(pInfo->pUpdateRes, 128);
pOperator->name = "StreamSessionSemiAggOperator";
pOperator->fpSet =
@ -3554,7 +3804,7 @@ static SSDataBlock* doStreamStateAgg(SOperatorInfo* pOperator) {
break;
}
if (pBlock->info.type == STREAM_REPROCESS) {
if (pBlock->info.type == STREAM_CLEAR) {
doClearStateWindows(&pInfo->streamAggSup, pBlock, pInfo->primaryTsIndex, &pInfo->stateCol, pInfo->stateCol.slotId,
pSeUpdated, pInfo->pSeDeleted);
continue;
@ -3900,18 +4150,22 @@ _error:
// merge interval operator
typedef struct SMergeIntervalAggOperatorInfo {
SIntervalAggOperatorInfo intervalAggOperatorInfo;
SHashObj* groupIntervalHash;
void* groupIntervalIter;
SList* groupIntervals;
SListIter groupIntervalsIter;
bool hasGroupId;
uint64_t groupId;
SSDataBlock* prefetchedBlock;
bool inputBlocksFinished;
} SMergeIntervalAggOperatorInfo;
typedef struct SGroupTimeWindow {
uint64_t groupId;
STimeWindow window;
} SGroupTimeWindow;
void destroyMergeIntervalOperatorInfo(void* param, int32_t numOfOutput) {
SMergeIntervalAggOperatorInfo* miaInfo = (SMergeIntervalAggOperatorInfo*)param;
taosHashCleanup(miaInfo->groupIntervalHash);
tdListFree(miaInfo->groupIntervals);
destroyIntervalOperatorInfo(&miaInfo->intervalAggOperatorInfo, numOfOutput);
}
@ -3940,15 +4194,22 @@ static int32_t outputPrevIntervalResult(SOperatorInfo* pOperatorInfo, uint64_t t
bool ascScan = (iaInfo->order == TSDB_ORDER_ASC);
SExprSupp* pExprSup = &pOperatorInfo->exprSupp;
STimeWindow* prevWin = taosHashGet(miaInfo->groupIntervalHash, &tableGroupId, sizeof(tableGroupId));
if (prevWin == NULL) {
taosHashPut(miaInfo->groupIntervalHash, &tableGroupId, sizeof(tableGroupId), newWin, sizeof(STimeWindow));
return 0;
}
SGroupTimeWindow groupTimeWindow = {.groupId = tableGroupId, .window = *newWin};
tdListAppend(miaInfo->groupIntervals, &groupTimeWindow);
if ((ascScan && newWin->skey > prevWin->skey || (!ascScan) && newWin->skey < prevWin->skey)) {
finalizeWindowResult(pOperatorInfo, tableGroupId, prevWin, pResultBlock);
taosHashPut(miaInfo->groupIntervalHash, &tableGroupId, sizeof(tableGroupId), newWin, sizeof(STimeWindow));
SListIter iter = {0};
tdListInitIter(miaInfo->groupIntervals, &iter, TD_LIST_FORWARD);
SListNode* listNode = NULL;
while ((listNode = tdListNext(&iter)) != NULL) {
SGroupTimeWindow* prevGrpWin = (SGroupTimeWindow*)listNode->data;
if (prevGrpWin->groupId != tableGroupId ) {
continue;
}
STimeWindow* prevWin = &prevGrpWin->window;
if ((ascScan && newWin->skey > prevWin->ekey || (!ascScan) && newWin->skey < prevWin->ekey)) {
finalizeWindowResult(pOperatorInfo, tableGroupId, prevWin, pResultBlock);
tdListPopNode(miaInfo->groupIntervals, listNode);
}
}
return 0;
@ -4075,6 +4336,7 @@ static SSDataBlock* doMergeIntervalAgg(SOperatorInfo* pOperator) {
}
if (pBlock == NULL) {
tdListInitIter(miaInfo->groupIntervals, &miaInfo->groupIntervalsIter, TD_LIST_FORWARD);
miaInfo->inputBlocksFinished = true;
break;
}
@ -4100,14 +4362,12 @@ static SSDataBlock* doMergeIntervalAgg(SOperatorInfo* pOperator) {
}
if (miaInfo->inputBlocksFinished) {
void* win = taosHashIterate(miaInfo->groupIntervalHash, miaInfo->groupIntervalIter);
if (win != NULL) {
miaInfo->groupIntervalIter = win;
SListNode* listNode = tdListNext(&miaInfo->groupIntervalsIter);
size_t len = 0;
uint64_t* pTableGroupId = taosHashGetKey(win, &len);
finalizeWindowResult(pOperator, *pTableGroupId, win, pRes);
pRes->info.groupId = *pTableGroupId;
if (listNode != NULL) {
SGroupTimeWindow* grpWin = (SGroupTimeWindow*)(listNode->data);
finalizeWindowResult(pOperator, grpWin->groupId, &grpWin->window, pRes);
pRes->info.groupId = grpWin->groupId;
}
}
@ -4129,8 +4389,7 @@ SOperatorInfo* createMergeIntervalOperatorInfo(SOperatorInfo* downstream, SExprI
goto _error;
}
miaInfo->groupIntervalHash = taosHashInit(128, taosGetDefaultHashFunction(TSDB_DATA_TYPE_UBIGINT), true, HASH_NO_LOCK);
miaInfo->groupIntervalIter = NULL;
miaInfo->groupIntervals = tdListNew(sizeof(SGroupTimeWindow));
SIntervalAggOperatorInfo* iaInfo = &miaInfo->intervalAggOperatorInfo;

4
source/libs/executor/src/tsort.c
View File

@ -758,6 +758,10 @@ void* tsortGetValue(STupleHandle* pVHandle, int32_t colIndex) {
}
}
uint64_t tsortGetGroupId(STupleHandle* pVHandle) {
return pVHandle->pBlock->info.groupId;
}
SSortExecInfo tsortGetSortExecInfo(SSortHandle* pHandle) {
SSortExecInfo info = {0};

21
source/libs/function/src/builtins.c
View File

@ -1220,18 +1220,19 @@ static int32_t translateSubstr(SFunctionNode* pFunc, char* pErrBuf, int32_t len)
static int32_t translateCast(SFunctionNode* pFunc, char* pErrBuf, int32_t len) {
// The number of parameters has been limited by the syntax definition
uint8_t para1Type = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
//uint8_t para1Type = ((SExprNode*)nodesListGetNode(pFunc->pParameterList, 0))->resType.type;
// The function return type has been set during syntax parsing
uint8_t para2Type = pFunc->node.resType.type;
if (para2Type != TSDB_DATA_TYPE_BIGINT && para2Type != TSDB_DATA_TYPE_UBIGINT &&
para2Type != TSDB_DATA_TYPE_VARCHAR && para2Type != TSDB_DATA_TYPE_NCHAR &&
para2Type != TSDB_DATA_TYPE_TIMESTAMP) {
return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
}
if ((para2Type == TSDB_DATA_TYPE_TIMESTAMP && IS_VAR_DATA_TYPE(para1Type)) ||
(para2Type == TSDB_DATA_TYPE_BINARY && para1Type == TSDB_DATA_TYPE_NCHAR)) {
return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
}
//if (para2Type != TSDB_DATA_TYPE_BIGINT && para2Type != TSDB_DATA_TYPE_UBIGINT &&
// para2Type != TSDB_DATA_TYPE_VARCHAR && para2Type != TSDB_DATA_TYPE_NCHAR &&
// para2Type != TSDB_DATA_TYPE_TIMESTAMP) {
// return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
//}
//if ((para2Type == TSDB_DATA_TYPE_TIMESTAMP && IS_VAR_DATA_TYPE(para1Type)) ||
// (para2Type == TSDB_DATA_TYPE_BINARY && para1Type == TSDB_DATA_TYPE_NCHAR)) {
// return invaildFuncParaTypeErrMsg(pErrBuf, len, pFunc->functionName);
//}
int32_t para2Bytes = pFunc->node.resType.bytes;
if (IS_VAR_DATA_TYPE(para2Type)) {

40
source/libs/function/src/udfd.c
View File

@ -12,6 +12,8 @@
* 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/>.
*/
// clang-format off
#include "uv.h"
#include "os.h"
#include "fnLog.h"
@ -25,6 +27,7 @@
#include "tglobal.h"
#include "tmsg.h"
#include "trpc.h"
// clang-foramt on
typedef struct SUdfdContext {
uv_loop_t * loop;
@ -103,12 +106,12 @@ typedef struct SUdfdRpcSendRecvInfo {
uv_sem_t resultSem;
} SUdfdRpcSendRecvInfo;
static void udfdProcessRpcRsp(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet);
static void udfdProcessRpcRsp(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet);
static int32_t udfdFillUdfInfoFromMNode(void *clientRpc, char *udfName, SUdf *udf);
static int32_t udfdConnectToMnode();
static int32_t udfdLoadUdf(char *udfName, SUdf *udf);
static bool udfdRpcRfp(int32_t code);
static int initEpSetFromCfg(const char *firstEp, const char *secondEp, SCorEpSet *pEpSet);
static bool udfdRpcRfp(int32_t code);
static int initEpSetFromCfg(const char *firstEp, const char *secondEp, SCorEpSet *pEpSet);
static int32_t udfdOpenClientRpc();
static int32_t udfdCloseClientRpc();
@ -126,19 +129,19 @@ static void udfdUvHandleError(SUdfdUvConn *conn) { uv_close((uv_handle_t *)conn-
static void udfdPipeRead(uv_stream_t *client, ssize_t nread, const uv_buf_t *buf);
static void udfdOnNewConnection(uv_stream_t *server, int status);
static void udfdIntrSignalHandler(uv_signal_t *handle, int signum);
static void udfdIntrSignalHandler(uv_signal_t *handle, int signum);
static int32_t removeListeningPipe();
static void udfdPrintVersion();
static void udfdPrintVersion();
static int32_t udfdParseArgs(int32_t argc, char *argv[]);
static int32_t udfdInitLog();
static void udfdCtrlAllocBufCb(uv_handle_t *handle, size_t suggested_size, uv_buf_t *buf);
static void udfdCtrlReadCb(uv_stream_t *q, ssize_t nread, const uv_buf_t *buf);
static void udfdCtrlAllocBufCb(uv_handle_t *handle, size_t suggested_size, uv_buf_t *buf);
static void udfdCtrlReadCb(uv_stream_t *q, ssize_t nread, const uv_buf_t *buf);
static int32_t udfdUvInit();
static void udfdCloseWalkCb(uv_handle_t *handle, void *arg);
static void udfdCloseWalkCb(uv_handle_t *handle, void *arg);
static int32_t udfdRun();
static void udfdConnectMnodeThreadFunc(void* args);
static void udfdConnectMnodeThreadFunc(void *args);
void udfdProcessRequest(uv_work_t *req) {
SUvUdfWork *uvUdf = (SUvUdfWork *)(req->data);
@ -401,11 +404,11 @@ void udfdProcessRpcRsp(void *parent, SRpcMsg *pMsg, SEpSet *pEpSet) {
udf->bufSize = pFuncInfo->bufSize;
char path[PATH_MAX] = {0};
#ifdef WINDOWS
#ifdef WINDOWS
snprintf(path, sizeof(path), "%s%s.dll", TD_TMP_DIR_PATH, pFuncInfo->name);
#else
#else
snprintf(path, sizeof(path), "%s/lib%s.so", TD_TMP_DIR_PATH, pFuncInfo->name);
#endif
#endif
TdFilePtr file =
taosOpenFile(path, TD_FILE_CREATE | TD_FILE_WRITE | TD_FILE_READ | TD_FILE_TRUNC | TD_FILE_AUTO_DEL);
if (file == NULL) {
@ -544,7 +547,8 @@ int32_t udfdLoadUdf(char *udfName, SUdf *udf) {
return 0;
}
static bool udfdRpcRfp(int32_t code) {
if (code == TSDB_CODE_RPC_REDIRECT) {
if (code == TSDB_CODE_RPC_REDIRECT || code == TSDB_CODE_RPC_NETWORK_UNAVAIL || code == TSDB_CODE_NODE_NOT_DEPLOYED ||
code == TSDB_CODE_SYN_NOT_LEADER || code == TSDB_CODE_APP_NOT_READY) {
return true;
} else {
return false;
@ -652,8 +656,7 @@ void udfdAllocBuffer(uv_handle_t *handle, size_t suggestedSize, uv_buf_t *buf) {
buf->base = ctx->inputBuf;
buf->len = ctx->inputCap;
} else {
fnError("udfd can not allocate enough memory")
buf->base = NULL;
fnError("udfd can not allocate enough memory") buf->base = NULL;
buf->len = 0;
}
} else {
@ -664,8 +667,7 @@ void udfdAllocBuffer(uv_handle_t *handle, size_t suggestedSize, uv_buf_t *buf) {
buf->base = ctx->inputBuf + ctx->inputLen;
buf->len = ctx->inputCap - ctx->inputLen;
} else {
fnError("udfd can not allocate enough memory")
buf->base = NULL;
fnError("udfd can not allocate enough memory") buf->base = NULL;
buf->len = 0;
}
}
@ -881,7 +883,7 @@ static int32_t udfdRun() {
return 0;
}
void udfdConnectMnodeThreadFunc(void* args) {
void udfdConnectMnodeThreadFunc(void *args) {
int32_t retryMnodeTimes = 0;
int32_t code = 0;
while (retryMnodeTimes++ <= TSDB_MAX_REPLICA) {
@ -939,7 +941,7 @@ int main(int argc, char *argv[]) {
uv_thread_create(&mnodeConnectThread, udfdConnectMnodeThreadFunc, NULL);
udfdRun();
removeListeningPipe();
udfdCloseClientRpc();

6
source/libs/nodes/src/nodesCloneFuncs.c
View File

@ -351,7 +351,7 @@ static int32_t logicScanCopy(const SScanLogicNode* pSrc, SScanLogicNode* pDst) {
COPY_SCALAR_FIELD(watermark);
COPY_SCALAR_FIELD(tsColId);
COPY_SCALAR_FIELD(filesFactor);
CLONE_NODE_LIST_FIELD(pPartTags);
CLONE_NODE_LIST_FIELD(pGroupTags);
return TSDB_CODE_SUCCESS;
}
@ -401,6 +401,7 @@ static int32_t logicMergeCopy(const SMergeLogicNode* pSrc, SMergeLogicNode* pDst
CLONE_NODE_LIST_FIELD(pInputs);
COPY_SCALAR_FIELD(numOfChannels);
COPY_SCALAR_FIELD(srcGroupId);
COPY_SCALAR_FIELD(groupSort);
return TSDB_CODE_SUCCESS;
}
@ -436,6 +437,7 @@ static int32_t logicFillCopy(const SFillLogicNode* pSrc, SFillLogicNode* pDst) {
static int32_t logicSortCopy(const SSortLogicNode* pSrc, SSortLogicNode* pDst) {
COPY_BASE_OBJECT_FIELD(node, logicNodeCopy);
CLONE_NODE_LIST_FIELD(pSortKeys);
COPY_SCALAR_FIELD(groupSort);
return TSDB_CODE_SUCCESS;
}
@ -500,7 +502,7 @@ static int32_t physiTableScanCopy(const STableScanPhysiNode* pSrc, STableScanPhy
COPY_SCALAR_FIELD(ratio);
COPY_SCALAR_FIELD(dataRequired);
CLONE_NODE_LIST_FIELD(pDynamicScanFuncs);
CLONE_NODE_LIST_FIELD(pPartitionTags);
CLONE_NODE_LIST_FIELD(pGroupTags);
COPY_SCALAR_FIELD(interval);
COPY_SCALAR_FIELD(offset);
COPY_SCALAR_FIELD(sliding);

37
source/libs/nodes/src/nodesCodeFuncs.c
View File

@ -234,6 +234,8 @@ const char* nodesNodeName(ENodeType type) {
return "PhysiMerge";
case QUERY_NODE_PHYSICAL_PLAN_SORT:
return "PhysiSort";
case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT:
return "PhysiGroupSort";
case QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL:
return "PhysiHashInterval";
case QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL:
@ -539,7 +541,7 @@ static const char* jkScanLogicPlanScanPseudoCols = "ScanPseudoCols";
static const char* jkScanLogicPlanTableId = "TableId";
static const char* jkScanLogicPlanTableType = "TableType";
static const char* jkScanLogicPlanTagCond = "TagCond";
static const char* jkScanLogicPlanPartTags = "PartTags";
static const char* jkScanLogicPlanGroupTags = "GroupTags";
static int32_t logicScanNodeToJson(const void* pObj, SJson* pJson) {
const SScanLogicNode* pNode = (const SScanLogicNode*)pObj;
@ -561,7 +563,7 @@ static int32_t logicScanNodeToJson(const void* pObj, SJson* pJson) {
code = tjsonAddObject(pJson, jkScanLogicPlanTagCond, nodeToJson, pNode->pTagCond);
}
if (TSDB_CODE_SUCCESS == code) {
code = nodeListToJson(pJson, jkScanLogicPlanPartTags, pNode->pPartTags);
code = nodeListToJson(pJson, jkScanLogicPlanGroupTags, pNode->pGroupTags);
}
return code;
@ -588,7 +590,7 @@ static int32_t jsonToLogicScanNode(const SJson* pJson, void* pObj) {
code = jsonToNodeObject(pJson, jkScanLogicPlanTagCond, &pNode->pTagCond);
}
if (TSDB_CODE_SUCCESS == code) {
code = jsonToNodeList(pJson, jkScanLogicPlanPartTags, &pNode->pPartTags);
code = jsonToNodeList(pJson, jkScanLogicPlanGroupTags, &pNode->pGroupTags);
}
return code;
@ -1430,7 +1432,8 @@ static const char* jkTableScanPhysiPlanTriggerType = "triggerType";
static const char* jkTableScanPhysiPlanWatermark = "watermark";
static const char* jkTableScanPhysiPlanTsColId = "tsColId";
static const char* jkTableScanPhysiPlanFilesFactor = "FilesFactor";
static const char* jkTableScanPhysiPlanPartitionTags = "PartitionTags";
static const char* jkTableScanPhysiPlanGroupTags = "GroupTags";
static const char* jkTableScanPhysiPlanGroupSort = "GroupSort";
static int32_t physiTableScanNodeToJson(const void* pObj, SJson* pJson) {
const STableScanPhysiNode* pNode = (const STableScanPhysiNode*)pObj;
@ -1485,7 +1488,10 @@ static int32_t physiTableScanNodeToJson(const void* pObj, SJson* pJson) {
code = tjsonAddDoubleToObject(pJson, jkTableScanPhysiPlanFilesFactor, pNode->filesFactor);
}
if (TSDB_CODE_SUCCESS == code) {
code = nodeListToJson(pJson, jkTableScanPhysiPlanPartitionTags, pNode->pPartitionTags);
code = nodeListToJson(pJson, jkTableScanPhysiPlanGroupTags, pNode->pGroupTags);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddBoolToObject(pJson, jkTableScanPhysiPlanGroupSort, pNode->groupSort);
}
return code;
@ -1544,7 +1550,10 @@ static int32_t jsonToPhysiTableScanNode(const SJson* pJson, void* pObj) {
code = tjsonGetDoubleValue(pJson, jkTableScanPhysiPlanFilesFactor, &pNode->filesFactor);
}
if (TSDB_CODE_SUCCESS == code) {
code = jsonToNodeList(pJson, jkTableScanPhysiPlanPartitionTags, &pNode->pPartitionTags);
code = jsonToNodeList(pJson, jkTableScanPhysiPlanGroupTags, &pNode->pGroupTags);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetBoolValue(pJson, jkTableScanPhysiPlanGroupSort, &pNode->groupSort);
}
return code;
@ -1725,6 +1734,7 @@ static const char* jkMergePhysiPlanMergeKeys = "MergeKeys";
static const char* jkMergePhysiPlanTargets = "Targets";
static const char* jkMergePhysiPlanNumOfChannels = "NumOfChannels";
static const char* jkMergePhysiPlanSrcGroupId = "SrcGroupId";
static const char* jkMergePhysiPlanGroupSort = "GroupSort";
static int32_t physiMergeNodeToJson(const void* pObj, SJson* pJson) {
const SMergePhysiNode* pNode = (const SMergePhysiNode*)pObj;
@ -1742,6 +1752,9 @@ static int32_t physiMergeNodeToJson(const void* pObj, SJson* pJson) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddIntegerToObject(pJson, jkMergePhysiPlanSrcGroupId, pNode->srcGroupId);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddBoolToObject(pJson, jkMergePhysiPlanGroupSort, pNode->groupSort);
}
return code;
}
@ -1762,6 +1775,9 @@ static int32_t jsonToPhysiMergeNode(const SJson* pJson, void* pObj) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetIntValue(pJson, jkMergePhysiPlanSrcGroupId, &pNode->srcGroupId);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetBoolValue(pJson, jkMergePhysiPlanGroupSort, &pNode->groupSort);
}
return code;
}
@ -3369,6 +3385,7 @@ static const char* jkSlotDescSlotId = "SlotId";
static const char* jkSlotDescDataType = "DataType";
static const char* jkSlotDescReserve = "Reserve";
static const char* jkSlotDescOutput = "Output";
static const char* jkSlotDescName = "Name";
static int32_t slotDescNodeToJson(const void* pObj, SJson* pJson) {
const SSlotDescNode* pNode = (const SSlotDescNode*)pObj;
@ -3383,6 +3400,9 @@ static int32_t slotDescNodeToJson(const void* pObj, SJson* pJson) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddBoolToObject(pJson, jkSlotDescOutput, pNode->output);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonAddStringToObject(pJson, jkSlotDescName, pNode->name);
}
return code;
}
@ -3400,6 +3420,9 @@ static int32_t jsonToSlotDescNode(const SJson* pJson, void* pObj) {
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetBoolValue(pJson, jkSlotDescOutput, &pNode->output);
}
if (TSDB_CODE_SUCCESS == code) {
code = tjsonGetStringValue(pJson, jkSlotDescName, pNode->name);
}
return code;
}
@ -4137,6 +4160,7 @@ static int32_t specificNodeToJson(const void* pObj, SJson* pJson) {
case QUERY_NODE_PHYSICAL_PLAN_MERGE:
return physiMergeNodeToJson(pObj, pJson);
case QUERY_NODE_PHYSICAL_PLAN_SORT:
case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT:
return physiSortNodeToJson(pObj, pJson);
case QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL:
case QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL:
@ -4280,6 +4304,7 @@ static int32_t jsonToSpecificNode(const SJson* pJson, void* pObj) {
case QUERY_NODE_PHYSICAL_PLAN_MERGE:
return jsonToPhysiMergeNode(pJson, pObj);
case QUERY_NODE_PHYSICAL_PLAN_SORT:
case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT:
return jsonToPhysiSortNode(pJson, pObj);
case QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL:
case QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL:

3
source/libs/nodes/src/nodesTraverseFuncs.c
View File

@ -500,7 +500,8 @@ static EDealRes dispatchPhysiPlan(SNode* pNode, ETraversalOrder order, FNodeWalk
}
break;
}
case QUERY_NODE_PHYSICAL_PLAN_SORT: {
case QUERY_NODE_PHYSICAL_PLAN_SORT:
case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT: {
SSortPhysiNode* pSort = (SSortPhysiNode*)pNode;
res = walkPhysiNode((SPhysiNode*)pNode, order, walker, pContext);
if (DEAL_RES_ERROR != res && DEAL_RES_END != res) {

9
source/libs/nodes/src/nodesUtilFuncs.c
View File

@ -288,6 +288,8 @@ SNode* nodesMakeNode(ENodeType type) {
return makeNode(type, sizeof(SMergePhysiNode));
case QUERY_NODE_PHYSICAL_PLAN_SORT:
return makeNode(type, sizeof(SSortPhysiNode));
case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT:
return makeNode(type, sizeof(SGroupSortPhysiNode));
case QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL:
return makeNode(type, sizeof(SIntervalPhysiNode));
case QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL:
@ -709,7 +711,7 @@ void nodesDestroyNode(SNode* pNode) {
nodesDestroyNode(pLogicNode->pTagCond);
nodesDestroyNode(pLogicNode->pTagIndexCond);
taosArrayDestroy(pLogicNode->pSmaIndexes);
nodesDestroyList(pLogicNode->pPartTags);
nodesDestroyList(pLogicNode->pGroupTags);
break;
}
case QUERY_NODE_LOGIC_PLAN_JOIN: {
@ -813,7 +815,7 @@ void nodesDestroyNode(SNode* pNode) {
STableScanPhysiNode* pPhyNode = (STableScanPhysiNode*)pNode;
destroyScanPhysiNode((SScanPhysiNode*)pNode);
nodesDestroyList(pPhyNode->pDynamicScanFuncs);
nodesDestroyList(pPhyNode->pPartitionTags);
nodesDestroyList(pPhyNode->pGroupTags);
break;
}
case QUERY_NODE_PHYSICAL_PLAN_PROJECT: {
@ -850,7 +852,8 @@ void nodesDestroyNode(SNode* pNode) {
nodesDestroyList(pPhyNode->pTargets);
break;
}
case QUERY_NODE_PHYSICAL_PLAN_SORT: {
case QUERY_NODE_PHYSICAL_PLAN_SORT:
case QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT: {
SSortPhysiNode* pPhyNode = (SSortPhysiNode*)pNode;
destroyPhysiNode((SPhysiNode*)pPhyNode);
nodesDestroyList(pPhyNode->pExprs);

90
source/libs/parser/src/parTranslater.c
View File

@ -2456,90 +2456,6 @@ static int32_t rewriteUniqueStmt(STranslateContext* pCxt, SSelectStmt* pSelect)
return cxt.pTranslateCxt->errCode;
}
typedef struct SRwriteTailCxt {
STranslateContext* pTranslateCxt;
int64_t limit;
int64_t offset;
} SRwriteTailCxt;
static EDealRes rewriteTailFunc(SNode** pNode, void* pContext) {
SRwriteTailCxt* pCxt = pContext;
if (QUERY_NODE_FUNCTION == nodeType(*pNode)) {
SFunctionNode* pFunc = (SFunctionNode*)*pNode;
if (FUNCTION_TYPE_TAIL == pFunc->funcType) {
pCxt->limit = ((SValueNode*)nodesListGetNode(pFunc->pParameterList, 1))->datum.i;
if (3 == LIST_LENGTH(pFunc->pParameterList)) {
pCxt->offset = ((SValueNode*)nodesListGetNode(pFunc->pParameterList, 2))->datum.i;
}
SNode* pExpr = nodesListGetNode(pFunc->pParameterList, 0);
strcpy(((SExprNode*)pExpr)->aliasName, ((SExprNode*)*pNode)->aliasName);
NODES_CLEAR_LIST(pFunc->pParameterList);
nodesDestroyNode(*pNode);
*pNode = pExpr;
return DEAL_RES_IGNORE_CHILD;
}
}
return DEAL_RES_CONTINUE;
}
static int32_t createLimieNode(SRwriteTailCxt* pCxt, SLimitNode** pOutput) {
*pOutput = (SLimitNode*)nodesMakeNode(QUERY_NODE_LIMIT);
if (NULL == *pOutput) {
return TSDB_CODE_OUT_OF_MEMORY;
}
(*pOutput)->limit = pCxt->limit;
(*pOutput)->offset = pCxt->offset;
return TSDB_CODE_SUCCESS;
}
static SNode* createOrderByExpr(STranslateContext* pCxt) {
SOrderByExprNode* pOrder = (SOrderByExprNode*)nodesMakeNode(QUERY_NODE_ORDER_BY_EXPR);
if (NULL == pOrder) {
return NULL;
}
pCxt->errCode = createPrimaryKeyCol(pCxt, &pOrder->pExpr);
if (TSDB_CODE_SUCCESS != pCxt->errCode) {
nodesDestroyNode((SNode*)pOrder);
return NULL;
}
pOrder->order = ORDER_DESC;
pOrder->nullOrder = NULL_ORDER_FIRST;
return (SNode*)pOrder;
}
/* case 1:
* in: select tail(expr, k, f) from t where_clause
* out: select expr from t where_clause order by _rowts desc limit k offset f
*
* case 2:
* in: select tail(expr, k, f) from t where_clause partition_by_clause
* out: select expr from t where_clause partition_by_clause sort by _rowts desc limit k offset f
*
* case 3:
* in: select tail(expr, k, f) from t where_clause order_by_clause limit_clause
* out: select expr from (
* select expr from t where_clause order by _rowts desc limit k offset f
* ) order_by_clause limit_clause
*
* case 4:
* in: select tail(expr, k, f) from t where_clause partition_by_clause limit_clause
* out:
*/
static int32_t rewriteTailStmt(STranslateContext* pCxt, SSelectStmt* pSelect) {
if (!pSelect->hasTailFunc) {
return TSDB_CODE_SUCCESS;
}
SRwriteTailCxt cxt = {.pTranslateCxt = pCxt, .limit = -1, .offset = -1};
nodesRewriteExprs(pSelect->pProjectionList, rewriteTailFunc, &cxt);
int32_t code = nodesListMakeStrictAppend(&pSelect->pOrderByList, createOrderByExpr(pCxt));
if (TSDB_CODE_SUCCESS == code) {
code = createLimieNode(&cxt, &pSelect->pLimit);
}
pSelect->hasIndefiniteRowsFunc = false;
return code;
}
typedef struct SReplaceOrderByAliasCxt {
STranslateContext* pTranslateCxt;
SNodeList* pProjectionList;
@ -2616,9 +2532,6 @@ static int32_t translateSelectFrom(STranslateContext* pCxt, SSelectStmt* pSelect
if (TSDB_CODE_SUCCESS == code) {
code = rewriteUniqueStmt(pCxt, pSelect);
}
// if (TSDB_CODE_SUCCESS == code) {
// code = rewriteTailStmt(pCxt, pSelect);
// }
if (TSDB_CODE_SUCCESS == code) {
code = rewriteTimelineFunc(pCxt, pSelect);
}
@ -5023,6 +4936,7 @@ typedef struct SVgroupCreateTableBatch {
static void destroyCreateTbReq(SVCreateTbReq* pReq) {
taosMemoryFreeClear(pReq->name);
taosMemoryFreeClear(pReq->comment);
taosMemoryFreeClear(pReq->ntb.schemaRow.pSchema);
}
@ -5040,6 +4954,7 @@ static int32_t buildNormalTableBatchReq(int32_t acctId, const SCreateTableStmt*
if (pStmt->pOptions->commentNull == false) {
req.comment = strdup(pStmt->pOptions->comment);
if (NULL == req.comment) {
destroyCreateTbReq(&req);
return TSDB_CODE_OUT_OF_MEMORY;
}
req.commentLen = strlen(pStmt->pOptions->comment);
@ -5111,6 +5026,7 @@ static void destroyCreateTbReqBatch(SVgroupCreateTableBatch* pTbBatch) {
for (int32_t i = 0; i < size; ++i) {
SVCreateTbReq* pTableReq = taosArrayGet(pTbBatch->req.pArray, i);
taosMemoryFreeClear(pTableReq->name);
taosMemoryFreeClear(pTableReq->comment);
if (pTableReq->type == TSDB_NORMAL_TABLE) {
taosMemoryFreeClear(pTableReq->ntb.schemaRow.pSchema);

5
source/libs/parser/src/parUtil.c
View File

@ -337,7 +337,11 @@ int32_t trimString(const char* src, int32_t len, char* dst, int32_t dlen) {
static bool isValidateTag(char* input) {
if (!input) return false;
for (size_t i = 0; i < strlen(input); ++i) {
#ifdef WINDOWS
if (input[i] < 0x20 || input[i] > 0x7E) return false;
#else
if (isprint(input[i]) == 0) return false;
#endif
}
return true;
}
@ -377,6 +381,7 @@ int32_t parseJsontoTagData(const char* json, SArray* pTagVals, STag** ppTag, SMs
char* jsonKey = item->string;
if (!isValidateTag(jsonKey)) {
fprintf(stdout,"%s(%d) %s %08" PRId64 "\n", __FILE__, __LINE__,__func__,taosGetSelfPthreadId());fflush(stdout);
retCode = buildSyntaxErrMsg(pMsgBuf, "json key not validate", jsonKey);
goto end;
}

4
source/libs/planner/src/planLogicCreater.c
View File

@ -507,6 +507,8 @@ static int32_t createIndefRowsFuncLogicNode(SLogicPlanContext* pCxt, SSelectStmt
return TSDB_CODE_OUT_OF_MEMORY;
}
pIdfRowsFunc->isTailFunc = pSelect->hasTailFunc;
// indefinite rows functions and _select_values functions
int32_t code = nodesCollectFuncs(pSelect, SQL_CLAUSE_SELECT, fmIsVectorFunc, &pIdfRowsFunc->pFuncs);
if (TSDB_CODE_SUCCESS == code) {
@ -733,6 +735,8 @@ static int32_t createSortLogicNode(SLogicPlanContext* pCxt, SSelectStmt* pSelect
return TSDB_CODE_OUT_OF_MEMORY;
}
pSort->groupSort = pSelect->groupSort;
int32_t code = nodesCollectColumns(pSelect, SQL_CLAUSE_ORDER_BY, NULL, COLLECT_COL_TYPE_ALL, &pSort->node.pTargets);
if (TSDB_CODE_SUCCESS == code && NULL == pSort->node.pTargets) {
code = nodesListMakeStrictAppend(&pSort->node.pTargets,

577
source/libs/planner/src/planOptimizer.c
View File

File diff suppressed because it is too large Load Diff

40
source/libs/planner/src/planPhysiCreater.c
View File

@ -41,8 +41,12 @@ typedef struct SPhysiPlanContext {
static int32_t getSlotKey(SNode* pNode, const char* pStmtName, char* pKey) {
if (QUERY_NODE_COLUMN == nodeType(pNode)) {
SColumnNode* pCol = (SColumnNode*)pNode;
if (NULL != pStmtName && '\0' != pStmtName[0]) {
return sprintf(pKey, "%s.%s", pStmtName, pCol->node.aliasName);
if (NULL != pStmtName) {
if ('\0' != pStmtName[0]) {
return sprintf(pKey, "%s.%s", pStmtName, pCol->node.aliasName);
} else {
return sprintf(pKey, "%s", pCol->node.aliasName);
}
}
if ('\0' == pCol->tableAlias[0]) {
return sprintf(pKey, "%s", pCol->colName);
@ -56,11 +60,13 @@ static int32_t getSlotKey(SNode* pNode, const char* pStmtName, char* pKey) {
return sprintf(pKey, "%s", ((SExprNode*)pNode)->aliasName);
}
static SNode* createSlotDesc(SPhysiPlanContext* pCxt, const SNode* pNode, int16_t slotId, bool output, bool reserve) {
static SNode* createSlotDesc(SPhysiPlanContext* pCxt, const char* pName, const SNode* pNode, int16_t slotId,
bool output, bool reserve) {
SSlotDescNode* pSlot = (SSlotDescNode*)nodesMakeNode(QUERY_NODE_SLOT_DESC);
if (NULL == pSlot) {
return NULL;
}
strcpy(pSlot->name, pName);
pSlot->slotId = slotId;
pSlot->dataType = ((SExprNode*)pNode)->resType;
pSlot->reserve = reserve;
@ -99,10 +105,8 @@ static int32_t putSlotToHashImpl(int16_t dataBlockId, int16_t slotId, const char
return taosHashPut(pHash, pName, len, &index, sizeof(SSlotIndex));
}
static int32_t putSlotToHash(int16_t dataBlockId, int16_t slotId, SNode* pNode, SHashObj* pHash) {
char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN];
int32_t len = getSlotKey(pNode, NULL, name);
return putSlotToHashImpl(dataBlockId, slotId, name, len, pHash);
static int32_t putSlotToHash(const char* pName, int16_t dataBlockId, int16_t slotId, SNode* pNode, SHashObj* pHash) {
return putSlotToHashImpl(dataBlockId, slotId, pName, strlen(pName), pHash);
}
static int32_t createDataBlockDescHash(SPhysiPlanContext* pCxt, int32_t capacity, int16_t dataBlockId,
@ -131,9 +135,11 @@ static int32_t buildDataBlockSlots(SPhysiPlanContext* pCxt, SNodeList* pList, SD
int16_t slotId = 0;
SNode* pNode = NULL;
FOREACH(pNode, pList) {
code = nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, pNode, slotId, true, false));
char name[TSDB_TABLE_NAME_LEN + TSDB_COL_NAME_LEN];
getSlotKey(pNode, NULL, name);
code = nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, name, pNode, slotId, true, false));
if (TSDB_CODE_SUCCESS == code) {
code = putSlotToHash(pDataBlockDesc->dataBlockId, slotId, pNode, pHash);
code = putSlotToHash(name, pDataBlockDesc->dataBlockId, slotId, pNode, pHash);
}
if (TSDB_CODE_SUCCESS == code) {
pDataBlockDesc->totalRowSize += ((SExprNode*)pNode)->resType.bytes;
@ -196,7 +202,8 @@ static int32_t addDataBlockSlotsImpl(SPhysiPlanContext* pCxt, SNodeList* pList,
int32_t len = getSlotKey(pExpr, pStmtName, name);
SSlotIndex* pIndex = taosHashGet(pHash, name, len);
if (NULL == pIndex) {
code = nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, pExpr, nextSlotId, output, reserve));
code =
nodesListStrictAppend(pDataBlockDesc->pSlots, createSlotDesc(pCxt, name, pExpr, nextSlotId, output, reserve));
if (TSDB_CODE_SUCCESS == code) {
code = putSlotToHashImpl(pDataBlockDesc->dataBlockId, nextSlotId, name, len, pHash);
}
@ -513,12 +520,13 @@ static int32_t createTableScanPhysiNode(SPhysiPlanContext* pCxt, SSubplan* pSubp
tNameGetFullDbName(&pScanLogicNode->tableName, pSubplan->dbFName);
pTableScan->dataRequired = pScanLogicNode->dataRequired;
pTableScan->pDynamicScanFuncs = nodesCloneList(pScanLogicNode->pDynamicScanFuncs);
pTableScan->pPartitionTags = nodesCloneList(pScanLogicNode->pPartTags);
pTableScan->pGroupTags = nodesCloneList(pScanLogicNode->pGroupTags);
if ((NULL != pScanLogicNode->pDynamicScanFuncs && NULL == pTableScan->pDynamicScanFuncs) ||
(NULL != pScanLogicNode->pPartTags && NULL == pTableScan->pPartitionTags)) {
(NULL != pScanLogicNode->pGroupTags && NULL == pTableScan->pGroupTags)) {
nodesDestroyNode((SNode*)pTableScan);
return TSDB_CODE_OUT_OF_MEMORY;
}
pTableScan->groupSort = pScanLogicNode->groupSort;
pTableScan->interval = pScanLogicNode->interval;
pTableScan->offset = pScanLogicNode->offset;
pTableScan->sliding = pScanLogicNode->sliding;
@ -1170,8 +1178,9 @@ static int32_t createWindowPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildr
static int32_t createSortPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren, SSortLogicNode* pSortLogicNode,
SPhysiNode** pPhyNode) {
SSortPhysiNode* pSort =
(SSortPhysiNode*)makePhysiNode(pCxt, (SLogicNode*)pSortLogicNode, QUERY_NODE_PHYSICAL_PLAN_SORT);
SSortPhysiNode* pSort = (SSortPhysiNode*)makePhysiNode(
pCxt, (SLogicNode*)pSortLogicNode,
pSortLogicNode->groupSort ? QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT : QUERY_NODE_PHYSICAL_PLAN_SORT);
if (NULL == pSort) {
return TSDB_CODE_OUT_OF_MEMORY;
}
@ -1185,7 +1194,7 @@ static int32_t createSortPhysiNode(SPhysiPlanContext* pCxt, SNodeList* pChildren
if (TSDB_CODE_SUCCESS == code && NULL != pPrecalcExprs) {
code = setListSlotId(pCxt, pChildTupe->dataBlockId, -1, pPrecalcExprs, &pSort->pExprs);
if (TSDB_CODE_SUCCESS == code) {
code = addDataBlockSlots(pCxt, pSort->pExprs, pChildTupe);
code = pushdownDataBlockSlots(pCxt, pSort->pExprs, pChildTupe);
}
}
@ -1322,6 +1331,7 @@ static int32_t createMergePhysiNode(SPhysiPlanContext* pCxt, SMergeLogicNode* pM
pMerge->numOfChannels = pMergeLogicNode->numOfChannels;
pMerge->srcGroupId = pMergeLogicNode->srcGroupId;
pMerge->groupSort = pMergeLogicNode->groupSort;
int32_t code = addDataBlockSlots(pCxt, pMergeLogicNode->pInputs, pMerge->node.pOutputDataBlockDesc);

48
source/libs/planner/src/planSpliter.c
View File

@ -362,7 +362,7 @@ static int32_t stbSplGetNumOfVgroups(SLogicNode* pNode) {
}
static int32_t stbSplCreateMergeNode(SSplitContext* pCxt, SLogicSubplan* pSubplan, SLogicNode* pSplitNode,
SNodeList* pMergeKeys, SLogicNode* pPartChild) {
SNodeList* pMergeKeys, SLogicNode* pPartChild, bool groupSort) {
SMergeLogicNode* pMerge = (SMergeLogicNode*)nodesMakeNode(QUERY_NODE_LOGIC_PLAN_MERGE);
if (NULL == pMerge) {
return TSDB_CODE_OUT_OF_MEMORY;
@ -371,6 +371,7 @@ static int32_t stbSplCreateMergeNode(SSplitContext* pCxt, SLogicSubplan* pSubpla
pMerge->srcGroupId = pCxt->groupId;
pMerge->node.precision = pPartChild->precision;
pMerge->pMergeKeys = pMergeKeys;
pMerge->groupSort = groupSort;
int32_t code = TSDB_CODE_SUCCESS;
pMerge->pInputs = nodesCloneList(pPartChild->pTargets);
@ -430,7 +431,7 @@ static int32_t stbSplSplitIntervalForBatch(SSplitContext* pCxt, SStableSplitInfo
SNodeList* pMergeKeys = NULL;
code = stbSplCreateMergeKeysByPrimaryKey(((SWindowLogicNode*)pInfo->pSplitNode)->pTspk, &pMergeKeys);
if (TSDB_CODE_SUCCESS == code) {
code = stbSplCreateMergeNode(pCxt, NULL, pInfo->pSplitNode, pMergeKeys, pPartWindow);
code = stbSplCreateMergeNode(pCxt, NULL, pInfo->pSplitNode, pMergeKeys, pPartWindow, false);
}
if (TSDB_CODE_SUCCESS != code) {
nodesDestroyList(pMergeKeys);
@ -497,12 +498,16 @@ static int32_t stbSplSplitSessionForStream(SSplitContext* pCxt, SStableSplitInfo
return code;
}
static void splSetTableScanType(SLogicNode* pNode, EScanType scanType) {
static void stbSplSetTableMergeScan(SLogicNode* pNode) {
if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pNode)) {
((SScanLogicNode*)pNode)->scanType = scanType;
SScanLogicNode* pScan = (SScanLogicNode*)pNode;
pScan->scanType = SCAN_TYPE_TABLE_MERGE;
if (NULL != pScan->pGroupTags) {
pScan->groupSort = true;
}
} else {
if (1 == LIST_LENGTH(pNode->pChildren)) {
splSetTableScanType((SLogicNode*)nodesListGetNode(pNode->pChildren, 0), scanType);
stbSplSetTableMergeScan((SLogicNode*)nodesListGetNode(pNode->pChildren, 0));
}
}
}
@ -515,7 +520,7 @@ static int32_t stbSplSplitSessionOrStateForBatch(SSplitContext* pCxt, SStableSpl
int32_t code = stbSplCreateMergeKeysByPrimaryKey(((SWindowLogicNode*)pWindow)->pTspk, &pMergeKeys);
if (TSDB_CODE_SUCCESS == code) {
code = stbSplCreateMergeNode(pCxt, pInfo->pSubplan, pChild, pMergeKeys, (SLogicNode*)pChild);
code = stbSplCreateMergeNode(pCxt, pInfo->pSubplan, pChild, pMergeKeys, (SLogicNode*)pChild, true);
}
if (TSDB_CODE_SUCCESS == code) {
@ -524,13 +529,10 @@ static int32_t stbSplSplitSessionOrStateForBatch(SSplitContext* pCxt, SStableSpl
}
if (TSDB_CODE_SUCCESS == code) {
splSetTableScanType(pChild, SCAN_TYPE_TABLE_MERGE);
++(pCxt->groupId);
}
if (TSDB_CODE_SUCCESS == code) {
stbSplSetTableMergeScan(pChild);
pInfo->pSubplan->subplanType = SUBPLAN_TYPE_MERGE;
SPLIT_FLAG_SET_MASK(pInfo->pSubplan->splitFlag, SPLIT_FLAG_STABLE_SPLIT);
++(pCxt->groupId);
} else {
nodesDestroyList(pMergeKeys);
}
@ -560,7 +562,7 @@ static int32_t stbSplSplitState(SSplitContext* pCxt, SStableSplitInfo* pInfo) {
static SNodeList* stbSplGetPartKeys(SLogicNode* pNode) {
if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pNode)) {
return ((SScanLogicNode*)pNode)->pPartTags;
return ((SScanLogicNode*)pNode)->pGroupTags;
} else if (QUERY_NODE_LOGIC_PLAN_PARTITION == nodeType(pNode)) {
return ((SPartitionLogicNode*)pNode)->pPartitionKeys;
} else {
@ -775,6 +777,7 @@ static int32_t stbSplCreatePartSortNode(SSortLogicNode* pSort, SLogicNode** pOut
pPartSort->node.pChildren = pChildren;
splSetParent((SLogicNode*)pPartSort);
pPartSort->pSortKeys = pSortKeys;
pPartSort->groupSort = pSort->groupSort;
code = stbSplCreateMergeKeys(pPartSort->pSortKeys, pPartSort->node.pTargets, &pMergeKeys);
}
@ -789,12 +792,29 @@ static int32_t stbSplCreatePartSortNode(SSortLogicNode* pSort, SLogicNode** pOut
return code;
}
static void stbSplSetScanPartSort(SLogicNode* pNode) {
if (QUERY_NODE_LOGIC_PLAN_SCAN == nodeType(pNode)) {
SScanLogicNode* pScan = (SScanLogicNode*)pNode;
if (NULL != pScan->pGroupTags) {
pScan->groupSort = true;
}
} else {
if (1 == LIST_LENGTH(pNode->pChildren)) {
stbSplSetScanPartSort((SLogicNode*)nodesListGetNode(pNode->pChildren, 0));
}
}
}
static int32_t stbSplSplitSortNode(SSplitContext* pCxt, SStableSplitInfo* pInfo) {
SLogicNode* pPartSort = NULL;
SNodeList* pMergeKeys = NULL;
bool groupSort = ((SSortLogicNode*)pInfo->pSplitNode)->groupSort;
int32_t code = stbSplCreatePartSortNode((SSortLogicNode*)pInfo->pSplitNode, &pPartSort, &pMergeKeys);
if (TSDB_CODE_SUCCESS == code) {
code = stbSplCreateMergeNode(pCxt, pInfo->pSubplan, pInfo->pSplitNode, pMergeKeys, pPartSort);
code = stbSplCreateMergeNode(pCxt, pInfo->pSubplan, pInfo->pSplitNode, pMergeKeys, pPartSort, groupSort);
}
if (TSDB_CODE_SUCCESS == code && groupSort) {
stbSplSetScanPartSort(pPartSort);
}
if (TSDB_CODE_SUCCESS == code) {
code = nodesListMakeStrictAppend(&pInfo->pSubplan->pChildren,
@ -829,7 +849,7 @@ static int32_t stbSplSplitScanNodeForJoin(SSplitContext* pCxt, SLogicSubplan* pS
SNodeList* pMergeKeys = NULL;
int32_t code = stbSplCreateMergeKeysByPrimaryKey(stbSplFindPrimaryKeyFromScan(pScan), &pMergeKeys);
if (TSDB_CODE_SUCCESS == code) {
code = stbSplCreateMergeNode(pCxt, pSubplan, (SLogicNode*)pScan, pMergeKeys, (SLogicNode*)pScan);
code = stbSplCreateMergeNode(pCxt, pSubplan, (SLogicNode*)pScan, pMergeKeys, (SLogicNode*)pScan, false);
}
if (TSDB_CODE_SUCCESS == code) {
code = nodesListMakeStrictAppend(&pSubplan->pChildren,

8
source/libs/planner/test/planBasicTest.cpp
View File

@ -67,6 +67,14 @@ TEST_F(PlanBasicTest, tailFunc) {
run("SELECT TAIL(c1, 10) FROM t1");
run("SELECT TAIL(c2 + 10, 10, 80) FROM t1 WHERE c1 > 10");
run("SELECT TAIL(c2 + 10, 10, 80) FROM t1 WHERE c1 > 10 PARTITION BY c1");
run("SELECT TAIL(c2 + 10, 10, 80) FROM t1 WHERE c1 > 10 ORDER BY 1");
run("SELECT TAIL(c2 + 10, 10, 80) FROM t1 WHERE c1 > 10 LIMIT 5");
run("SELECT TAIL(c2 + 10, 10, 80) FROM t1 WHERE c1 > 10 PARTITION BY c1 LIMIT 5");
}
TEST_F(PlanBasicTest, interpFunc) {

4
source/libs/planner/test/planTestMain.cpp
View File

@ -76,6 +76,7 @@ static void parseArg(int argc, char* argv[]) {
static struct option long_options[] = {
{"dump", optional_argument, NULL, 'd'},
{"skipSql", required_argument, NULL, 's'},
{"limitSql", required_argument, NULL, 'i'},
{"log", required_argument, NULL, 'l'},
{0, 0, 0, 0}
};
@ -88,6 +89,9 @@ static void parseArg(int argc, char* argv[]) {
case 's':
setSkipSqlNum(optarg);
break;
case 'i':
setLimitSqlNum(optarg);
break;
case 'l':
setLogLevel(optarg);
break;

30
source/libs/planner/test/planTestUtil.cpp
View File

@ -51,6 +51,7 @@ enum DumpModule {
DumpModule g_dumpModule = DUMP_MODULE_NOTHING;
int32_t g_skipSql = 0;
int32_t g_limitSql = 0;
int32_t g_logLevel = 131;
void setDumpModule(const char* pModule) {
@ -76,28 +77,33 @@ void setDumpModule(const char* pModule) {
}
void setSkipSqlNum(const char* pNum) { g_skipSql = stoi(pNum); }
void setLimitSqlNum(const char* pNum) { g_limitSql = stoi(pNum); }
void setLogLevel(const char* pLogLevel) { g_logLevel = stoi(pLogLevel); }
int32_t getLogLevel() { return g_logLevel; }
class PlannerTestBaseImpl {
public:
PlannerTestBaseImpl() : sqlNo_(0) {}
PlannerTestBaseImpl() : sqlNo_(0), sqlNum_(0) {}
void useDb(const string& user, const string& db) {
caseEnv_.acctId_ = 0;
caseEnv_.user_ = user;
caseEnv_.db_ = db;
caseEnv_.nsql_ = g_skipSql;
caseEnv_.numOfSkipSql_ = g_skipSql;
caseEnv_.numOfLimitSql_ = g_limitSql;
}
void run(const string& sql) {
++sqlNo_;
if (caseEnv_.nsql_ > 0) {
--(caseEnv_.nsql_);
if (caseEnv_.numOfSkipSql_ > 0) {
--(caseEnv_.numOfSkipSql_);
return;
}
if (caseEnv_.numOfLimitSql_ > 0 && caseEnv_.numOfLimitSql_ == sqlNum_) {
return;
}
++sqlNum_;
reset();
try {
@ -134,7 +140,7 @@ class PlannerTestBaseImpl {
}
void prepare(const string& sql) {
if (caseEnv_.nsql_ > 0) {
if (caseEnv_.numOfSkipSql_ > 0) {
return;
}
@ -148,7 +154,7 @@ class PlannerTestBaseImpl {
}
void bindParams(TAOS_MULTI_BIND* pParams, int32_t colIdx) {
if (caseEnv_.nsql_ > 0) {
if (caseEnv_.numOfSkipSql_ > 0) {
return;
}
@ -161,8 +167,8 @@ class PlannerTestBaseImpl {
}
void exec() {
if (caseEnv_.nsql_ > 0) {
--(caseEnv_.nsql_);
if (caseEnv_.numOfSkipSql_ > 0) {
--(caseEnv_.numOfSkipSql_);
return;
}
@ -197,9 +203,10 @@ class PlannerTestBaseImpl {
int32_t acctId_;
string user_;
string db_;
int32_t nsql_;
int32_t numOfSkipSql_;
int32_t numOfLimitSql_;
caseEnv() : nsql_(0) {}
caseEnv() : numOfSkipSql_(0) {}
};
struct stmtEnv {
@ -401,6 +408,7 @@ class PlannerTestBaseImpl {
stmtEnv stmtEnv_;
stmtRes res_;
int32_t sqlNo_;
int32_t sqlNum_;
};
PlannerTestBase::PlannerTestBase() : impl_(new PlannerTestBaseImpl()) {}

1
source/libs/planner/test/planTestUtil.h
View File

@ -43,6 +43,7 @@ class PlannerTestBase : public testing::Test {
extern void setDumpModule(const char* pModule);
extern void setSkipSqlNum(const char* pNum);
extern void setLimitSqlNum(const char* pNum);
extern void setLogLevel(const char* pLogLevel);
extern int32_t getLogLevel();

44
source/libs/qcom/src/queryUtil.c
View File

@ -19,6 +19,7 @@
#include "tmsg.h"
#include "trpc.h"
#include "tsched.h"
// clang-format off
#include "cJSON.h"
#define VALIDNUMOFCOLS(x) ((x) >= TSDB_MIN_COLUMNS && (x) <= TSDB_MAX_COLUMNS)
@ -147,13 +148,15 @@ int32_t asyncSendMsgToServerExt(void* pTransporter, SEpSet* epSet, int64_t* pTra
}
memcpy(pMsg, pInfo->msgInfo.pData, pInfo->msgInfo.len);
SRpcMsg rpcMsg = {.msgType = pInfo->msgType,
.pCont = pMsg,
.contLen = pInfo->msgInfo.len,
.info.ahandle = (void*)pInfo,
.info.handle = pInfo->msgInfo.handle,
.info.persistHandle = persistHandle,
.code = 0};
SRpcMsg rpcMsg = {
.msgType = pInfo->msgType,
.pCont = pMsg,
.contLen = pInfo->msgInfo.len,
.info.ahandle = (void*)pInfo,
.info.handle = pInfo->msgInfo.handle,
.info.persistHandle = persistHandle,
.code = 0
};
assert(pInfo->fp != NULL);
TRACE_SET_ROOTID(&rpcMsg.info.traceId, pInfo->requestId);
rpcSendRequestWithCtx(pTransporter, epSet, &rpcMsg, pTransporterId, rpcCtx);
@ -221,8 +224,9 @@ void destroyQueryExecRes(SQueryExecRes* pRes) {
qError("invalid exec result for request type %d", pRes->msgType);
}
}
// clang-format on
int32_t dataConverToStr(char *str, int type, void *buf, int32_t bufSize, int32_t *len) {
int32_t dataConverToStr(char* str, int type, void* buf, int32_t bufSize, int32_t* len) {
int32_t n = 0;
switch (type) {
@ -262,7 +266,7 @@ int32_t dataConverToStr(char *str, int type, void *buf, int32_t bufSize, int32_t
case TSDB_DATA_TYPE_BINARY:
case TSDB_DATA_TYPE_NCHAR:
if (bufSize < 0) {
// tscError("invalid buf size");
// tscError("invalid buf size");
return TSDB_CODE_TSC_INVALID_VALUE;
}
@ -289,7 +293,7 @@ int32_t dataConverToStr(char *str, int type, void *buf, int32_t bufSize, int32_t
break;
default:
// tscError("unsupported type:%d", type);
// tscError("unsupported type:%d", type);
return TSDB_CODE_TSC_INVALID_VALUE;
}
@ -332,7 +336,7 @@ char* parseTagDatatoJson(void* p) {
int32_t length = taosUcs4ToMbs((TdUcs4*)pTagVal->pData, pTagVal->nData, tagJsonValue);
if (length < 0) {
qError("charset:%s to %s. val:%s convert json value failed.", DEFAULT_UNICODE_ENCODEC, tsCharset,
pTagVal->pData);
pTagVal->pData);
taosMemoryFree(tagJsonValue);
goto end;
}
@ -372,7 +376,6 @@ end:
return string;
}
int32_t cloneTableMeta(STableMeta* pSrc, STableMeta** pDst) {
if (NULL == pSrc) {
*pDst = NULL;
@ -393,37 +396,36 @@ int32_t cloneDbVgInfo(SDBVgInfo* pSrc, SDBVgInfo** pDst) {
*pDst = NULL;
return TSDB_CODE_SUCCESS;
}
*pDst = taosMemoryMalloc(sizeof(*pSrc));
if (NULL == *pDst) {
return TSDB_CODE_TSC_OUT_OF_MEMORY;
}
memcpy(*pDst, pSrc, sizeof(*pSrc));
if (pSrc->vgHash) {
(*pDst)->vgHash = taosHashInit(taosHashGetSize(pSrc->vgHash), taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true, HASH_ENTRY_LOCK);
(*pDst)->vgHash = taosHashInit(taosHashGetSize(pSrc->vgHash), taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), true,
HASH_ENTRY_LOCK);
if (NULL == (*pDst)->vgHash) {
return TSDB_CODE_TSC_OUT_OF_MEMORY;
}
SVgroupInfo* vgInfo = NULL;
void *pIter = taosHashIterate(pSrc->vgHash, NULL);
void* pIter = taosHashIterate(pSrc->vgHash, NULL);
while (pIter) {
vgInfo = pIter;
int32_t* vgId = taosHashGetKey(pIter, NULL);
if (0 != taosHashPut((*pDst)->vgHash, vgId, sizeof(*vgId), vgInfo, sizeof(*vgInfo))) {
qError("taosHashPut failed, vgId:%d", vgInfo->vgId);
taosHashCancelIterate(pSrc->vgHash, pIter);
taosHashCancelIterate(pSrc->vgHash, pIter);
taosHashCleanup((*pDst)->vgHash);
taosMemoryFreeClear(*pDst);
return TSDB_CODE_CTG_MEM_ERROR;
}
pIter = taosHashIterate(pSrc->vgHash, pIter);
}
}
return TSDB_CODE_SUCCESS;
}

178
source/libs/scalar/src/sclfunc.c
View File

@ -735,6 +735,60 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
char *input = colDataGetData(pInput[0].columnData, i);
switch(outputType) {
case TSDB_DATA_TYPE_TINYINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(int8_t *)output = taosStr2Int8(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(int8_t *)output = taosStr2Int8(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(int8_t *)output, int8_t, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_SMALLINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(int16_t *)output = taosStr2Int16(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(int16_t *)output = taosStr2Int16(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(int16_t *)output, int16_t, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_INT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(int32_t *)output = taosStr2Int32(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(int32_t *)output = taosStr2Int32(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(int32_t *)output, int32_t, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_BIGINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(int64_t *)output = taosStr2Int64(varDataVal(input), NULL, 10);
@ -753,6 +807,60 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
}
break;
}
case TSDB_DATA_TYPE_UTINYINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(uint8_t *)output = taosStr2UInt8(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(uint8_t *)output = taosStr2UInt8(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(uint8_t *)output, uint8_t, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_USMALLINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(uint16_t *)output = taosStr2UInt16(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(uint16_t *)output = taosStr2UInt16(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(uint16_t *)output, uint16_t, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_UINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(uint32_t *)output = taosStr2UInt32(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(uint32_t *)output = taosStr2UInt32(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(uint32_t *)output, uint32_t, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_UBIGINT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(uint64_t *)output = taosStr2UInt64(varDataVal(input), NULL, 10);
@ -771,10 +879,64 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
}
break;
}
case TSDB_DATA_TYPE_FLOAT: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(float *)output = taosStr2Float(varDataVal(input), NULL);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(float *)output = taosStr2Float(newBuf, NULL);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(float *)output, float, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_DOUBLE: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(double *)output = taosStr2Double(varDataVal(input), NULL);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(double *)output = taosStr2Double(newBuf, NULL);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(double *)output, double, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_BOOL: {
if (inputType == TSDB_DATA_TYPE_BINARY) {
*(bool *)output = taosStr2Int8(varDataVal(input), NULL, 10);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
newBuf[len] = 0;
*(bool *)output = taosStr2Int8(newBuf, NULL, 10);
taosMemoryFree(newBuf);
} else {
GET_TYPED_DATA(*(bool *)output, bool, inputType, input);
}
break;
}
case TSDB_DATA_TYPE_TIMESTAMP: {
if (inputType == TSDB_DATA_TYPE_BINARY || inputType == TSDB_DATA_TYPE_NCHAR) {
//not support
return TSDB_CODE_FAILED;
//convert to 0
*(int64_t *)output = 0;
} else {
GET_TYPED_DATA(*(int64_t *)output, int64_t, inputType, input);
}
@ -789,8 +951,16 @@ int32_t castFunction(SScalarParam *pInput, int32_t inputNum, SScalarParam *pOutp
len = sprintf(varDataVal(output), "%.*s", len, varDataVal(input));
varDataSetLen(output, len);
} else if (inputType == TSDB_DATA_TYPE_NCHAR) {
//not support
return TSDB_CODE_FAILED;
char *newBuf = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(input), varDataLen(input), newBuf);
if (len < 0) {
taosMemoryFree(newBuf);
return TSDB_CODE_FAILED;
}
len = TMIN(len, outputLen - VARSTR_HEADER_SIZE);
memcpy(varDataVal(output), newBuf, len);
varDataSetLen(output, len);
taosMemoryFree(newBuf);
} else {
char tmp[400] = {0};
NUM_TO_STRING(inputType, input, sizeof(tmp), tmp);

2
source/libs/scalar/src/sclvector.c
View File

@ -477,7 +477,7 @@ double getVectorDoubleValue_JSON(void *src, int32_t index){
}
void* ncharTobinary(void *buf){ // todo need to remove , if tobinary is nchar
int32_t inputLen = varDataLen(buf);
int32_t inputLen = varDataTLen(buf);
void* t = taosMemoryCalloc(1, inputLen);
int32_t len = taosUcs4ToMbs((TdUcs4 *)varDataVal(buf), varDataLen(buf), varDataVal(t));

6
source/libs/stream/src/stream.c
View File

@ -111,7 +111,7 @@ int32_t streamTaskEnqueue(SStreamTask* pTask, SStreamDispatchReq* pReq, SRpcMsg*
// enqueue
if (pData != NULL) {
pData->type = STREAM_DATA_TYPE_SSDATA_BLOCK;
pData->type = STREAM_INPUT__DATA_BLOCK;
pData->srcVgId = pReq->dataSrcVgId;
// decode
/*pData->blocks = pReq->data;*/
@ -146,7 +146,7 @@ int32_t streamTaskEnqueueRetrieve(SStreamTask* pTask, SStreamRetrieveReq* pReq,
// enqueue
if (pData != NULL) {
pData->type = STREAM_DATA_TYPE_SSDATA_BLOCK;
pData->type = STREAM_INPUT__DATA_RETRIEVE;
pData->srcVgId = 0;
// decode
/*pData->blocks = pReq->data;*/
@ -170,7 +170,7 @@ int32_t streamTaskEnqueueRetrieve(SStreamTask* pTask, SStreamRetrieveReq* pReq,
pCont->rspToTaskId = pReq->srcTaskId;
pCont->rspFromTaskId = pReq->dstTaskId;
pRsp->pCont = buf;
pRsp->contLen = sizeof(SMsgHead) + sizeof(SStreamDispatchRsp);
pRsp->contLen = sizeof(SMsgHead) + sizeof(SStreamRetrieveRsp);
tmsgSendRsp(pRsp);
return status == TASK_INPUT_STATUS__NORMAL ? 0 : -1;
}

5
source/libs/stream/src/streamDispatch.c
View File

@ -115,6 +115,7 @@ int32_t streamBroadcastToChildren(SStreamTask* pTask, const SSDataBlock* pBlock)
.srcNodeId = pTask->nodeId,
.srcTaskId = pTask->taskId,
.pRetrieve = pRetrieve,
.retrieveLen = dataStrLen,
};
int32_t sz = taosArrayGetSize(pTask->childEpInfo);
@ -146,7 +147,7 @@ int32_t streamBroadcastToChildren(SStreamTask* pTask, const SSDataBlock* pBlock)
.code = 0,
.msgType = TDMT_STREAM_RETRIEVE,
.pCont = buf,
.contLen = len,
.contLen = sizeof(SMsgHead) + len,
};
if (tmsgSendReq(&pEpInfo->epSet, &rpcMsg) < 0) {
@ -300,7 +301,7 @@ int32_t streamDispatch(SStreamTask* pTask, SMsgCb* pMsgCb) {
atomic_store_8(&pTask->outputStatus, TASK_OUTPUT_STATUS__NORMAL);
return 0;
}
ASSERT(pBlock->type == STREAM_DATA_TYPE_SSDATA_BLOCK);
ASSERT(pBlock->type == STREAM_INPUT__DATA_BLOCK);
qInfo("stream continue dispatching: task %d", pTask->taskId);

21
source/libs/stream/src/streamExec.c
View File

@ -17,6 +17,7 @@
static int32_t streamTaskExecImpl(SStreamTask* pTask, void* data, SArray* pRes) {
void* exec = pTask->exec.executor;
bool hasData = false;
// set input
SStreamQueueItem* pItem = (SStreamQueueItem*)data;
@ -27,7 +28,7 @@ static int32_t streamTaskExecImpl(SStreamTask* pTask, void* data, SArray* pRes)
ASSERT(pTask->isDataScan);
SStreamDataSubmit* pSubmit = (SStreamDataSubmit*)data;
qSetStreamInput(exec, pSubmit->data, STREAM_DATA_TYPE_SUBMIT_BLOCK, false);
} else if (pItem->type == STREAM_INPUT__DATA_BLOCK) {
} else if (pItem->type == STREAM_INPUT__DATA_BLOCK || pItem->type == STREAM_INPUT__DATA_RETRIEVE) {
SStreamDataBlock* pBlock = (SStreamDataBlock*)data;
SArray* blocks = pBlock->blocks;
qSetMultiStreamInput(exec, blocks->pData, blocks->size, STREAM_DATA_TYPE_SSDATA_BLOCK, false);
@ -43,7 +44,21 @@ static int32_t streamTaskExecImpl(SStreamTask* pTask, void* data, SArray* pRes)
if (qExecTask(exec, &output, &ts) < 0) {
ASSERT(false);
}
if (output == NULL) break;
if (output == NULL) {
if (pItem->type == STREAM_INPUT__DATA_RETRIEVE) {
//SSDataBlock block = {0};
//block.info.type = STREAM_PUSH_EMPTY;
//block.info.childId = pTask->selfChildId;
SStreamDataBlock* pRetrieveBlock = (SStreamDataBlock*)data;
ASSERT(taosArrayGetSize(pRetrieveBlock->blocks) == 1);
SSDataBlock* pBlock = createOneDataBlock(taosArrayGet(pRetrieveBlock->blocks, 0), true);
pBlock->info.type = STREAM_PUSH_EMPTY;
pBlock->info.childId = pTask->selfChildId;
taosArrayPush(pRes, pBlock);
}
break;
}
hasData = true;
if (output->info.type == STREAM_RETRIEVE) {
if (streamBroadcastToChildren(pTask, output) < 0) {
@ -99,7 +114,7 @@ static SArray* streamExecForQall(SStreamTask* pTask, SArray* pRes) {
if (type == STREAM_INPUT__TRIGGER) {
blockDataDestroy(((SStreamTrigger*)data)->pBlock);
taosFreeQitem(data);
} else if (type == STREAM_INPUT__DATA_BLOCK) {
} else if (type == STREAM_INPUT__DATA_BLOCK || type == STREAM_INPUT__DATA_RETRIEVE) {
taosArrayDestroyEx(((SStreamDataBlock*)data)->blocks, (FDelete)tDeleteSSDataBlock);
taosFreeQitem(data);
} else if (type == STREAM_INPUT__DATA_SUBMIT) {

20
source/libs/stream/src/streamUpdate.c
View File

@ -119,6 +119,7 @@ SUpdateInfo *updateInfoInit(int64_t interval, int32_t precision, int64_t waterma
taosArrayPush(pInfo->pTsBuckets, &dumy);
}
pInfo->numBuckets = DEFAULT_BUCKET_SIZE;
pInfo->pCloseWinSBF = NULL;
return pInfo;
}
@ -154,6 +155,9 @@ bool updateInfoIsUpdated(SUpdateInfo *pInfo, tb_uid_t tableId, TSKEY ts) {
TSKEY maxTs = *(TSKEY *)taosArrayGet(pInfo->pTsBuckets, index);
if (ts < maxTs - pInfo->watermark) {
// this window has been closed.
if (pInfo->pCloseWinSBF) {
return tScalableBfPut(pInfo->pCloseWinSBF, &ts, sizeof(TSKEY));
}
return true;
}
@ -193,3 +197,19 @@ void updateInfoDestroy(SUpdateInfo *pInfo) {
taosArrayDestroy(pInfo->pTsSBFs);
taosMemoryFree(pInfo);
}
void updateInfoAddCloseWindowSBF(SUpdateInfo *pInfo) {
if (pInfo->pCloseWinSBF) {
return;
}
int64_t rows = adjustExpEntries(pInfo->interval * ROWS_PER_MILLISECOND);
pInfo->pCloseWinSBF = tScalableBfInit(rows, DEFAULT_FALSE_POSITIVE);
}
void updateInfoDestoryColseWinSBF(SUpdateInfo *pInfo) {
if (!pInfo || !pInfo->pCloseWinSBF) {
return;
}
tScalableBfDestroy(pInfo->pCloseWinSBF);
pInfo->pCloseWinSBF = NULL;
}

12
source/libs/sync/inc/syncEnv.h
View File

@ -28,13 +28,13 @@ extern "C" {
#include "trpc.h"
#include "ttimer.h"
#define TIMER_MAX_MS 0x7FFFFFFF
#define ENV_TICK_TIMER_MS 1000
#define PING_TIMER_MS 1000
#define ELECT_TIMER_MS_MIN 1300
#define ELECT_TIMER_MS_MAX (ELECT_TIMER_MS_MIN * 2)
#define TIMER_MAX_MS 0x7FFFFFFF
#define ENV_TICK_TIMER_MS 1000
#define PING_TIMER_MS 1000
#define ELECT_TIMER_MS_MIN 1300
#define ELECT_TIMER_MS_MAX (ELECT_TIMER_MS_MIN * 2)
#define ELECT_TIMER_MS_RANGE (ELECT_TIMER_MS_MAX - ELECT_TIMER_MS_MIN)
#define HEARTBEAT_TIMER_MS 900
#define HEARTBEAT_TIMER_MS 900
#define EMPTY_RAFT_ID ((SRaftId){.addr = 0, .vgId = 0})

2
source/libs/sync/inc/syncIO.h
View File

@ -50,7 +50,7 @@ typedef struct SSyncIO {
void *pSyncNode;
int32_t (*FpOnSyncPing)(SSyncNode *pSyncNode, SyncPing *pMsg);
int32_t (*FpOnSyncPingReply)(SSyncNode *pSyncNode, SyncPingReply *pMsg);
int32_t (*FpOnSyncClientRequest)(SSyncNode *pSyncNode, SyncClientRequest *pMsg);
int32_t (*FpOnSyncClientRequest)(SSyncNode *pSyncNode, SyncClientRequest *pMsg, SyncIndex *pRetIndex);
int32_t (*FpOnSyncRequestVote)(SSyncNode *pSyncNode, SyncRequestVote *pMsg);
int32_t (*FpOnSyncRequestVoteReply)(SSyncNode *pSyncNode, SyncRequestVoteReply *pMsg);
int32_t (*FpOnSyncAppendEntries)(SSyncNode *pSyncNode, SyncAppendEntries *pMsg);

3
source/libs/sync/inc/syncInt.h
View File

@ -169,7 +169,7 @@ SSyncNode* syncNodeOpen(const SSyncInfo* pSyncInfo);
void syncNodeStart(SSyncNode* pSyncNode);
void syncNodeStartStandBy(SSyncNode* pSyncNode);
void syncNodeClose(SSyncNode* pSyncNode);
int32_t syncNodePropose(SSyncNode* pSyncNode, const SRpcMsg* pMsg, bool isWeak);
int32_t syncNodePropose(SSyncNode* pSyncNode, SRpcMsg* pMsg, bool isWeak);
// option
bool syncNodeSnapshotEnable(SSyncNode* pSyncNode);
@ -233,6 +233,7 @@ SyncIndex syncNodeGetPreIndex(SSyncNode* pSyncNode, SyncIndex index);
SyncTerm syncNodeGetPreTerm(SSyncNode* pSyncNode, SyncIndex index);
int32_t syncNodeGetPreIndexTerm(SSyncNode* pSyncNode, SyncIndex index, SyncIndex* pPreIndex, SyncTerm* pPreTerm);
bool syncNodeIsOptimizedOneReplica(SSyncNode* ths, SRpcMsg* pMsg);
int32_t syncNodeCommit(SSyncNode* ths, SyncIndex beginIndex, SyncIndex endIndex, uint64_t flag);
int32_t syncNodeUpdateNewConfigIndex(SSyncNode* ths, SSyncCfg* pNewCfg);

10
source/libs/sync/inc/syncRaftCfg.h
View File

@ -49,14 +49,14 @@ int32_t raftCfgClose(SRaftCfg *pRaftCfg);
int32_t raftCfgPersist(SRaftCfg *pRaftCfg);
int32_t raftCfgAddConfigIndex(SRaftCfg *pRaftCfg, SyncIndex configIndex);
cJSON *syncCfg2Json(SSyncCfg *pSyncCfg);
char *syncCfg2Str(SSyncCfg *pSyncCfg);
char *syncCfg2SimpleStr(SSyncCfg *pSyncCfg);
cJSON * syncCfg2Json(SSyncCfg *pSyncCfg);
char * syncCfg2Str(SSyncCfg *pSyncCfg);
char * syncCfg2SimpleStr(SSyncCfg *pSyncCfg);
int32_t syncCfgFromJson(const cJSON *pRoot, SSyncCfg *pSyncCfg);
int32_t syncCfgFromStr(const char *s, SSyncCfg *pSyncCfg);
cJSON *raftCfg2Json(SRaftCfg *pRaftCfg);
char *raftCfg2Str(SRaftCfg *pRaftCfg);
cJSON * raftCfg2Json(SRaftCfg *pRaftCfg);
char * raftCfg2Str(SRaftCfg *pRaftCfg);
int32_t raftCfgFromJson(const cJSON *pRoot, SRaftCfg *pRaftCfg);
int32_t raftCfgFromStr(const char *s, SRaftCfg *pRaftCfg);

25
source/libs/sync/inc/syncRaftEntry.h
View File

@ -56,6 +56,31 @@ void syncEntryPrint2(char* s, const SSyncRaftEntry* pObj);
void syncEntryLog(const SSyncRaftEntry* pObj);
void syncEntryLog2(char* s, const SSyncRaftEntry* pObj);
//-----------------------------------
typedef struct SRaftEntryCache {
SHashObj* pEntryHash;
int32_t maxCount;
int32_t currentCount;
TdThreadMutex mutex;
SSyncNode* pSyncNode;
} SRaftEntryCache;
SRaftEntryCache* raftCacheCreate(SSyncNode* pSyncNode, int32_t maxCount);
void raftCacheDestroy(SRaftEntryCache* pCache);
int32_t raftCachePutEntry(struct SRaftEntryCache* pCache, SSyncRaftEntry* pEntry);
int32_t raftCacheGetEntry(struct SRaftEntryCache* pCache, SyncIndex index, SSyncRaftEntry** ppEntry);
int32_t raftCacheGetEntryP(struct SRaftEntryCache* pCache, SyncIndex index, SSyncRaftEntry** ppEntry);
int32_t raftCacheDelEntry(struct SRaftEntryCache* pCache, SyncIndex index);
int32_t raftCacheGetAndDel(struct SRaftEntryCache* pCache, SyncIndex index, SSyncRaftEntry** ppEntry);
int32_t raftCacheClear(struct SRaftEntryCache* pCache);
cJSON* raftCache2Json(SRaftEntryCache* pObj);
char* raftCache2Str(SRaftEntryCache* pObj);
void raftCachePrint(SRaftEntryCache* pObj);
void raftCachePrint2(char* s, SRaftEntryCache* pObj);
void raftCacheLog(SRaftEntryCache* pObj);
void raftCacheLog2(char* s, SRaftEntryCache* pObj);
#ifdef __cplusplus
}
#endif

7
source/libs/sync/inc/syncRaftLog.h
View File

@ -29,9 +29,12 @@ extern "C" {
#include "wal.h"
typedef struct SSyncLogStoreData {
SSyncNode* pSyncNode;
SWal* pWal;
SSyncNode* pSyncNode;
SWal* pWal;
TdThreadMutex mutex;
SWalReadHandle* pWalHandle;
// SyncIndex beginIndex; // valid begin index, default 0, may be set beginIndex > 0
} SSyncLogStoreData;

Some files were not shown because too many files have changed in this diff Show More