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

Merge branch 'main' into test/TS-3105

This commit is contained in:
Ping Xiao 2023-05-06 13:57:05 +08:00
parent 7b73130bd4 ca790277b9
commit 6b35a4ae40
159 changed files with 5556 additions and 3859 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
cmake/cmake.version
View File

@ -2,7 +2,7 @@
IF (DEFINED VERNUMBER)
SET(TD_VER_NUMBER ${VERNUMBER})
ELSE ()
SET(TD_VER_NUMBER "3.0.4.0")
SET(TD_VER_NUMBER "3.0.4.1")
ENDIF ()
IF (DEFINED VERCOMPATIBLE)

2
cmake/taosadapter_CMakeLists.txt.in
View File

@ -2,7 +2,7 @@
# taosadapter
ExternalProject_Add(taosadapter
GIT_REPOSITORY https://github.com/taosdata/taosadapter.git
GIT_TAG e02ddb2
GIT_TAG ae8d51c
SOURCE_DIR "${TD_SOURCE_DIR}/tools/taosadapter"
BINARY_DIR ""
#BUILD_IN_SOURCE TRUE

2
include/common/systable.h
View File

@ -12,7 +12,7 @@
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TDENGINE_SYSTABLE_H
#define TDENGINE_SYSTABLE_H

8
include/common/tcommon.h
View File

@ -212,14 +212,6 @@ enum {
FETCH_TYPE__NONE,
};
typedef struct {
int8_t fetchType;
union {
SSDataBlock data;
void* meta;
};
} SFetchRet;
typedef struct SVarColAttr {
int32_t* offset; // start position for each entry in the list
uint32_t length; // used buffer size that contain the valid data

11
include/common/tmsg.h
View File

@ -415,7 +415,7 @@ static FORCE_INLINE SSchemaWrapper* tCloneSSchemaWrapper(const SSchemaWrapper* p
return pSW;
}
static FORCE_INLINE void tDeleteSSchemaWrapper(SSchemaWrapper* pSchemaWrapper) {
static FORCE_INLINE void tDeleteSchemaWrapper(SSchemaWrapper* pSchemaWrapper) {
if (pSchemaWrapper) {
taosMemoryFree(pSchemaWrapper->pSchema);
taosMemoryFree(pSchemaWrapper);
@ -689,6 +689,7 @@ typedef struct {
int32_t tSerializeSAlterUserReq(void* buf, int32_t bufLen, SAlterUserReq* pReq);
int32_t tDeserializeSAlterUserReq(void* buf, int32_t bufLen, SAlterUserReq* pReq);
void tFreeSAlterUserReq(SAlterUserReq* pReq);
typedef struct {
char user[TSDB_USER_LEN];
@ -3420,10 +3421,10 @@ typedef struct {
char data[]; // SSubmitReq2
} SSubmitReq2Msg;
int32_t tEncodeSSubmitReq2(SEncoder* pCoder, const SSubmitReq2* pReq);
int32_t tDecodeSSubmitReq2(SDecoder* pCoder, SSubmitReq2* pReq);
void tDestroySSubmitTbData(SSubmitTbData* pTbData, int32_t flag);
void tDestroySSubmitReq2(SSubmitReq2* pReq, int32_t flag);
int32_t tEncodeSubmitReq(SEncoder* pCoder, const SSubmitReq2* pReq);
int32_t tDecodeSubmitReq(SDecoder* pCoder, SSubmitReq2* pReq);
void tDestroySubmitTbData(SSubmitTbData* pTbData, int32_t flag);
void tDestroySubmitReq(SSubmitReq2* pReq, int32_t flag);
typedef struct {
int32_t affectedRows;

8
include/libs/catalog/catalog.h
View File

@ -82,6 +82,7 @@ typedef struct SCatalogReq {
SArray* pUser; // element is SUserAuthInfo
SArray* pTableIndex; // element is SNAME
SArray* pTableCfg; // element is SNAME
SArray* pTableTag; // element is SNAME
bool qNodeRequired; // valid qnode
bool dNodeRequired; // valid dnode
bool svrVerRequired;
@ -105,6 +106,7 @@ typedef struct SMetaData {
SArray* pUser; // pRes = SUserAuthRes*
SArray* pQnodeList; // pRes = SArray<SQueryNodeLoad>*
SArray* pTableCfg; // pRes = STableCfg*
SArray* pTableTag; // pRes = SArray<STagVal>*
SArray* pDnodeList; // pRes = SArray<SEpSet>*
SMetaRes* pSvrVer; // pRes = char*
} SMetaData;
@ -122,8 +124,8 @@ typedef struct SSTableVersion {
char stbName[TSDB_TABLE_NAME_LEN];
uint64_t dbId;
uint64_t suid;
int16_t sversion;
int16_t tversion;
int32_t sversion;
int32_t tversion;
int32_t smaVer;
} SSTableVersion;
@ -312,6 +314,8 @@ int32_t catalogGetIndexMeta(SCatalog* pCtg, SRequestConnInfo* pConn, const char*
int32_t catalogGetTableIndex(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName, SArray** pRes);
int32_t catalogGetTableTag(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName, SArray** pRes);
int32_t catalogRefreshGetTableCfg(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName, STableCfg** pCfg);
int32_t catalogUpdateTableIndex(SCatalog* pCtg, STableIndexRsp* pRsp);

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

@ -192,8 +192,6 @@ SArray* qGetQueriedTableListInfo(qTaskInfo_t tinfo);
int32_t qStreamPrepareScan(qTaskInfo_t tinfo, STqOffsetVal* pOffset, int8_t subType);
int32_t qStreamSetScanMemData(qTaskInfo_t tinfo, SPackedData submit);
void qStreamSetOpen(qTaskInfo_t tinfo);
void qStreamExtractOffset(qTaskInfo_t tinfo, STqOffsetVal* pOffset);
@ -208,8 +206,6 @@ void* qExtractReaderFromStreamScanner(void* scanner);
int32_t qExtractStreamScanner(qTaskInfo_t tinfo, void** scanner);
int32_t qStreamInput(qTaskInfo_t tinfo, void* pItem);
int32_t qStreamSetParamForRecover(qTaskInfo_t tinfo);
int32_t qStreamSourceRecoverStep1(qTaskInfo_t tinfo, int64_t ver);
int32_t qStreamSourceRecoverStep2(qTaskInfo_t tinfo, int64_t ver);

2
include/libs/nodes/querynodes.h
View File

@ -379,6 +379,8 @@ typedef struct SVnodeModifyOpStmt {
SName usingTableName;
const char* pBoundCols;
struct STableMeta* pTableMeta;
SNode* pTagCond;
SArray* pTableTag;
SHashObj* pVgroupsHashObj;
SHashObj* pTableBlockHashObj; // SHashObj<tuid, STableDataCxt*>
SHashObj* pSubTableHashObj;

4
include/libs/qcom/query.h
View File

@ -116,8 +116,8 @@ typedef struct STableMeta {
// if the table is TSDB_CHILD_TABLE, the following information is acquired from the corresponding super table meta
// info
int16_t sversion;
int16_t tversion;
int32_t sversion;
int32_t tversion;
STableComInfo tableInfo;
SSchema schema[];
} STableMeta;

7
include/libs/stream/tstream.h
View File

@ -50,7 +50,6 @@ enum {
TASK_STATUS__RECOVER_PREPARE,
TASK_STATUS__RECOVER1,
TASK_STATUS__RECOVER2,
TASK_STATUS__RESTORE, // only available for source task to replay WAL from the checkpoint
};
enum {
@ -341,12 +340,13 @@ typedef struct SStreamMeta {
TTB* pTaskDb;
TTB* pCheckpointDb;
SHashObj* pTasks;
SArray* pTaskList; // SArray<task_id*>
void* ahandle;
TXN* txn;
FTaskExpand* expandFunc;
int32_t vgId;
SRWLatch lock;
int32_t walScan;
int32_t walScanCounter;
} SStreamMeta;
int32_t tEncodeStreamEpInfo(SEncoder* pEncoder, const SStreamChildEpInfo* pInfo);
@ -545,8 +545,9 @@ int32_t streamScanExec(SStreamTask* pTask, int32_t batchSz);
// recover and fill history
int32_t streamTaskCheckDownstream(SStreamTask* pTask, int64_t version);
int32_t streamTaskLaunchRecover(SStreamTask* pTask, int64_t version);
int32_t streamProcessTaskCheckReq(SStreamTask* pTask, const SStreamTaskCheckReq* pReq);
int32_t streamTaskCheckStatus(SStreamTask* pTask);
int32_t streamProcessTaskCheckRsp(SStreamTask* pTask, const SStreamTaskCheckRsp* pRsp, int64_t version);
// common
int32_t streamSetParamForRecover(SStreamTask* pTask);
int32_t streamRestoreParam(SStreamTask* pTask);

2
include/libs/wal/wal.h
View File

@ -147,8 +147,6 @@ typedef struct SWalReader {
int64_t curFileFirstVer;
int64_t curVersion;
int64_t capacity;
// int8_t curInvalid;
// int8_t curStopped;
TdThreadMutex mutex;
SWalFilterCond cond;
// TODO remove it

1
include/util/taoserror.h
View File

@ -261,6 +261,7 @@ int32_t* taosGetErrno();
// #define TSDB_CODE_MND_INVALID_STABLE_NAME TAOS_DEF_ERROR_CODE(0, 0x036D) // 2.x
#define TSDB_CODE_MND_INVALID_STB_OPTION TAOS_DEF_ERROR_CODE(0, 0x036E)
#define TSDB_CODE_MND_INVALID_ROW_BYTES TAOS_DEF_ERROR_CODE(0, 0x036F)
#define TSDB_CODE_MND_FIELD_VALUE_OVERFLOW TAOS_DEF_ERROR_CODE(0, 0x0370)
// mnode-func

37
packaging/deb/DEBIAN/preinst
View File

@ -26,6 +26,38 @@ if pidof taosd &> /dev/null; then
sleep 1
fi
# Stop adapter service if running
if pidof taosadapter &> /dev/null; then
if pidof systemd &> /dev/null; then
${csudo}systemctl stop taosadapter || :
elif $(which service &> /dev/null); then
${csudo}service taosadapter stop || :
else
pid=$(ps -ef | grep "taosadapter" | grep -v "grep" | awk '{print $2}')
if [ -n "$pid" ]; then
${csudo}kill -9 $pid || :
fi
fi
echo "Stop taosadapter service success!"
sleep 1
fi
# Stop keeper service if running
if pidof taoskeeper &> /dev/null; then
if pidof systemd &> /dev/null; then
${csudo}systemctl stop taoskeeper || :
elif $(which service &> /dev/null); then
${csudo}service taoskeeper stop || :
else
pid=$(ps -ef | grep "taoskeeper" | grep -v "grep" | awk '{print $2}')
if [ -n "$pid" ]; then
${csudo}kill -9 $pid || :
fi
fi
echo "Stop taoskeeper service success!"
sleep 1
fi
# if taos.cfg already softlink, remove it
cfg_install_dir="/etc/taos"
install_main_dir="/usr/local/taos"
@ -41,6 +73,11 @@ if [ -f "${install_main_dir}/taosadapter.service" ]; then
${csudo}rm -f ${install_main_dir}/cfg/taosadapter.service || :
fi
if [ -f "${install_main_dir}/taoskeeper.toml" ]; then
${csudo}rm -f ${install_main_dir}/cfg/taoskeeper.toml || :
fi
# there can not libtaos.so*, otherwise ln -s error
${csudo}rm -f ${install_main_dir}/driver/libtaos.* || :
[ -f ${install_main_dir}/driver/libtaosws.so ] && ${csudo}rm -f ${install_main_dir}/driver/libtaosws.so || :

1
packaging/deb/DEBIAN/prerm
View File

@ -32,6 +32,7 @@ else
${csudo}rm -f ${bin_link_dir}/udfd || :
${csudo}rm -f ${bin_link_dir}/taosadapter || :
${csudo}rm -f ${bin_link_dir}/taosdemo || :
${csudo}rm -f ${bin_link_dir}/taoskeeper || :
${csudo}rm -f ${cfg_link_dir}/* || :
${csudo}rm -f ${inc_link_dir}/taos.h || :
${csudo}rm -f ${inc_link_dir}/taosdef.h || :

26
packaging/deb/makedeb.sh
View File

@ -44,8 +44,31 @@ mkdir -p ${pkg_dir}${install_home_path}/include
#mkdir -p ${pkg_dir}${install_home_path}/init.d
mkdir -p ${pkg_dir}${install_home_path}/script
# download taoskeeper and build
if [ "$cpuType" = "x64" ] || [ "$cpuType" = "x86_64" ] || [ "$cpuType" = "amd64" ]; then
arch=amd64
elif [ "$cpuType" = "x32" ] || [ "$cpuType" = "i386" ] || [ "$cpuType" = "i686" ]; then
arch=386
elif [ "$cpuType" = "arm" ] || [ "$cpuType" = "aarch32" ]; then
arch=arm
elif [ "$cpuType" = "arm64" ] || [ "$cpuType" = "aarch64" ]; then
arch=arm64
else
arch=$cpuType
fi
echo "${top_dir}/../enterprise/packaging/build_taoskeeper.sh -r ${arch} -e taoskeeper"
echo "$top_dir=${top_dir}"
taoskeeper_binary=`${top_dir}/../enterprise/packaging/build_taoskeeper.sh -r $arch -e taoskeeper`
echo "taoskeeper_binary: ${taoskeeper_binary}"
# copy config files
cp $(dirname ${taoskeeper_binary})/config/taoskeeper.toml ${pkg_dir}${install_home_path}/cfg
cp $(dirname ${taoskeeper_binary})/taoskeeper.service ${pkg_dir}${install_home_path}/cfg
cp ${compile_dir}/../packaging/cfg/taos.cfg ${pkg_dir}${install_home_path}/cfg
cp ${compile_dir}/../packaging/cfg/taosd.service ${pkg_dir}${install_home_path}/cfg
if [ -f "${compile_dir}/test/cfg/taosadapter.toml" ]; then
cp ${compile_dir}/test/cfg/taosadapter.toml ${pkg_dir}${install_home_path}/cfg || :
fi
@ -53,6 +76,7 @@ if [ -f "${compile_dir}/test/cfg/taosadapter.service" ]; then
cp ${compile_dir}/test/cfg/taosadapter.service ${pkg_dir}${install_home_path}/cfg || :
fi
cp ${taoskeeper_binary} ${pkg_dir}${install_home_path}/bin
#cp ${compile_dir}/../packaging/deb/taosd ${pkg_dir}${install_home_path}/init.d
cp ${compile_dir}/../packaging/tools/post.sh ${pkg_dir}${install_home_path}/script
cp ${compile_dir}/../packaging/tools/preun.sh ${pkg_dir}${install_home_path}/script
@ -143,6 +167,7 @@ else
exit 1
fi
rm -rf ${pkg_dir}/build-taoskeeper
# make deb package
dpkg -b ${pkg_dir} $debname
echo "make deb package success!"
@ -150,4 +175,5 @@ echo "make deb package success!"
cp ${pkg_dir}/*.deb ${output_dir}
# clean temp dir
rm -rf ${pkg_dir}

24
packaging/rpm/makerpm.sh
View File

@ -35,14 +35,16 @@ function cp_rpm_package() {
local cur_dir
cd $1
cur_dir=$(pwd)
echo "cp_rpm_package cd: ${cur_dir}"
for dirlist in "$(ls ${cur_dir})"; do
if test -d ${dirlist}; then
cd ${dirlist}
echo 'cp_rpm_package ${cur_dir}/${dirlist}'
cp_rpm_package ${cur_dir}/${dirlist}
cd ..
fi
if test -e ${dirlist}; then
echo "${cur_dir}/${dirlist} ${output_dir}/TDengine-${tdengine_ver}.rpm"
cp ${cur_dir}/${dirlist} ${output_dir}/TDengine-${tdengine_ver}.rpm
fi
done
@ -54,6 +56,25 @@ fi
${csudo}mkdir -p ${pkg_dir}
cd ${pkg_dir}
# download taoskeeper and build
if [ "$cpuType" = "x64" ] || [ "$cpuType" = "x86_64" ] || [ "$cpuType" = "amd64" ]; then
arch=amd64
elif [ "$cpuType" = "x32" ] || [ "$cpuType" = "i386" ] || [ "$cpuType" = "i686" ]; then
arch=386
elif [ "$cpuType" = "arm" ] || [ "$cpuType" = "aarch32" ]; then
arch=arm
elif [ "$cpuType" = "arm64" ] || [ "$cpuType" = "aarch64" ]; then
arch=arm64
else
arch=$cpuType
fi
cd ${top_dir}
echo "${top_dir}/../enterprise/packaging/build_taoskeeper.sh -r ${arch} -e taoskeeper"
taoskeeper_binary=`${top_dir}/../enterprise/packaging/build_taoskeeper.sh -r $arch -e taoskeeper`
echo "taoskeeper_binary: ${taoskeeper_binary}"
cd ${package_dir}
${csudo}mkdir -p BUILD BUILDROOT RPMS SOURCES SPECS SRPMS
${csudo}rpmbuild --define="_version ${tdengine_ver}" --define="_topdir ${pkg_dir}" --define="_compiledir ${compile_dir}" -bb ${spec_file}
@ -85,3 +106,4 @@ mv ${output_dir}/TDengine-${tdengine_ver}.rpm ${output_dir}/${rpmname}
cd ..
${csudo}rm -rf ${pkg_dir}
rm -rf ${top_dir}/build-taoskeeper

21
packaging/rpm/tdengine.spec
View File

@ -3,6 +3,7 @@
%define cfg_install_dir /etc/taos
%define __strip /bin/true
%global __python /usr/bin/python3
%global _build_id_links none
Name: tdengine
Version: %{_version}
@ -62,6 +63,15 @@ fi
if [ -f %{_compiledir}/test/cfg/taosadapter.service ]; then
cp %{_compiledir}/test/cfg/taosadapter.service %{buildroot}%{homepath}/cfg
fi
if [ -f %{_compiledir}/../build-taoskeeper/config/taoskeeper.toml ]; then
cp %{_compiledir}/../build-taoskeeper/config/taoskeeper.toml %{buildroot}%{homepath}/cfg ||:
fi
if [ -f %{_compiledir}/../build-taoskeeper/taoskeeper.service ]; then
cp %{_compiledir}/../build-taoskeeper/taoskeeper.service %{buildroot}%{homepath}/cfg ||:
fi
#cp %{_compiledir}/../packaging/rpm/taosd %{buildroot}%{homepath}/init.d
cp %{_compiledir}/../packaging/tools/post.sh %{buildroot}%{homepath}/script
cp %{_compiledir}/../packaging/tools/preun.sh %{buildroot}%{homepath}/script
@ -73,8 +83,12 @@ cp %{_compiledir}/build/bin/taosd %{buildroot}%{homepath}/bin
cp %{_compiledir}/build/bin/udfd %{buildroot}%{homepath}/bin
cp %{_compiledir}/build/bin/taosBenchmark %{buildroot}%{homepath}/bin
if [ -f %{_compiledir}/../build-taoskeeper/taoskeeper ]; then
cp %{_compiledir}/../build-taoskeeper/taoskeeper %{buildroot}%{homepath}/bin
fi
if [ -f %{_compiledir}/build/bin/taosadapter ]; then
cp %{_compiledir}/build/bin/taosadapter %{buildroot}%{homepath}/bin ||:
cp %{_compiledir}/build/bin/taosadapter %{buildroot}%{homepath}/bin
fi
cp %{_compiledir}/build/lib/${libfile} %{buildroot}%{homepath}/driver
[ -f %{_compiledir}/build/lib/${wslibfile} ] && cp %{_compiledir}/build/lib/${wslibfile} %{buildroot}%{homepath}/driver ||:
@ -119,7 +133,9 @@ if [ -f %{_compiledir}/build/bin/jemalloc-config ]; then
cp %{_compiledir}/build/lib/pkgconfig/jemalloc.pc %{buildroot}%{homepath}/jemalloc/lib/pkgconfig
fi
fi
ls -al %{buildroot}%{homepath}/bin
tree -L 5
echo "==============================copying files done"
#Scripts executed before installation
%pre
if [ -f /var/lib/taos/dnode/dnodeCfg.json ]; then
@ -196,6 +212,7 @@ if [ $1 -eq 0 ];then
${csudo}rm -f ${bin_link_dir}/taosd || :
${csudo}rm -f ${bin_link_dir}/udfd || :
${csudo}rm -f ${bin_link_dir}/taosadapter || :
${csudo}rm -f ${bin_link_dir}/taoskeeper || :
${csudo}rm -f ${cfg_link_dir}/* || :
${csudo}rm -f ${inc_link_dir}/taos.h || :
${csudo}rm -f ${inc_link_dir}/taosdef.h || :

2
packaging/tools/makepkg.sh
View File

@ -341,7 +341,7 @@ if [ "$verMode" == "cluster" ] || [ "$verMode" == "cloud" ]; then
tmp_pwd=`pwd`
cd ${install_dir}/connector
if [ ! -d taos-connector-jdbc ];then
git clone -b main --depth=1 https://github.com/taosdata/taos-connector-jdbc.git ||:
git clone -b 3.1.0 --depth=1 https://github.com/taosdata/taos-connector-jdbc.git ||:
fi
cd taos-connector-jdbc
mvn clean package -Dmaven.test.skip=true

39
packaging/tools/post.sh
View File

@ -436,7 +436,7 @@ function local_fqdn_check() {
function install_taosadapter_config() {
if [ ! -f "${cfg_install_dir}/taosadapter.toml" ]; then
[ ! -d %{cfg_install_dir} ] &&
[ ! -d ${cfg_install_dir} ] &&
${csudo}${csudo}mkdir -p ${cfg_install_dir}
[ -f ${cfg_dir}/taosadapter.toml ] && ${csudo}cp ${cfg_dir}/taosadapter.toml ${cfg_install_dir}
[ -f ${cfg_install_dir}/taosadapter.toml ] &&
@ -451,19 +451,26 @@ function install_taosadapter_config() {
}
function install_taoskeeper_config() {
if [ ! -f "${cfg_install_dir}/keeper.toml" ]; then
[ ! -d %{cfg_install_dir} ] &&
${csudo}${csudo}mkdir -p ${cfg_install_dir}
[ -f ${cfg_dir}/keeper.toml ] && ${csudo}cp ${cfg_dir}/keeper.toml ${cfg_install_dir}
[ -f ${cfg_install_dir}/keeper.toml ] &&
${csudo}chmod 644 ${cfg_install_dir}/keeper.toml
# if new environment without taoskeeper
if [[ ! -f "${cfg_install_dir}/keeper.toml" ]] && [[ ! -f "${cfg_install_dir}/taoskeeper.toml" ]]; then
[ ! -d ${cfg_install_dir} ] && ${csudo}${csudo}mkdir -p ${cfg_install_dir}
[ -f ${cfg_dir}/taoskeeper.toml ] && ${csudo}cp ${cfg_dir}/taoskeeper.toml ${cfg_install_dir}
[ -f ${cfg_install_dir}/taoskeeper.toml ] &&
${csudo}chmod 644 ${cfg_install_dir}/taoskeeper.toml
fi
# if old machine with taoskeeper.toml file
if [ -f ${cfg_install_dir}/taoskeeper.toml ]; then
${csudo}mv ${cfg_dir}/taoskeeper.toml ${cfg_dir}/taoskeeper.toml.new
fi
[ -f ${cfg_dir}/keeper.toml ] &&
${csudo}mv ${cfg_dir}/keeper.toml ${cfg_dir}/keeper.toml.new
if [ -f ${cfg_install_dir}/keeper.toml ]; then
echo "The file keeper.toml will be renamed to taoskeeper.toml"
${csudo}mv ${cfg_install_dir}/keeper.toml ${cfg_install_dir}/taoskeeper.toml
${csudo}mv ${cfg_dir}/taoskeeper.toml ${cfg_dir}/taoskeeper.toml.new
fi
[ -f ${cfg_install_dir}/keeper.toml ] &&
${csudo}ln -s ${cfg_install_dir}/keeper.toml ${cfg_dir}
[ -f ${cfg_install_dir}/taoskeeper.toml ] &&
${csudo}ln -s ${cfg_install_dir}/taoskeeper.toml ${cfg_dir}
}
function install_config() {
@ -655,6 +662,15 @@ function install_taosadapter_service() {
fi
}
function install_taoskeeper_service() {
if ((${service_mod}==0)); then
[ -f ${script_dir}/../cfg/taoskeeper.service ] &&\
${csudo}cp ${script_dir}/../cfg/taoskeeper.service \
${service_config_dir}/ || :
${csudo}systemctl daemon-reload
fi
}
function install_service() {
log_print "start install service"
if [ "$osType" != "Darwin" ]; then
@ -732,6 +748,7 @@ function install_TDengine() {
install_taosadapter_config
install_taoskeeper_config
install_taosadapter_service
install_taoskeeper_service
install_service
install_app

17
packaging/tools/preun.sh
View File

@ -17,7 +17,7 @@ cfg_link_dir="/usr/local/taos/cfg"
service_config_dir="/etc/systemd/system"
taos_service_name="taosd"
taoskeeper_service_name="taoskeeper"
csudo=""
if command -v sudo > /dev/null; then
csudo="sudo "
@ -57,6 +57,13 @@ function kill_taosd() {
fi
}
function kill_taoskeeper() {
pid=$(ps -ef | grep "taoskeeper" | grep -v "grep" | awk '{print $2}')
if [ -n "$pid" ]; then
${csudo}kill -9 $pid || :
fi
}
function clean_service_on_systemd() {
taosadapter_service_config="${service_config_dir}/taosadapter.service"
if systemctl is-active --quiet taosadapter; then
@ -76,6 +83,12 @@ function clean_service_on_systemd() {
[ -f ${taosadapter_service_config} ] && ${csudo}rm -f ${taosadapter_service_config}
taoskeeper_service_config="${service_config_dir}/${taoskeeper_service_name}.service"
if systemctl is-active --quiet ${taoskeeper_service_name}; then
echo "TDengine taoskeeper is running, stopping it..."
${csudo}systemctl stop ${taoskeeper_service_name} &> /dev/null || echo &> /dev/null
fi
[ -f ${taoskeeper_service_config} ] && ${csudo}rm -f ${taoskeeper_service_config}
}
function clean_service_on_sysvinit() {
@ -111,6 +124,7 @@ function clean_service() {
# must manual stop taosd
kill_taosadapter
kill_taosd
kill_taoskeeper
fi
}
@ -124,6 +138,7 @@ ${csudo}rm -f ${bin_link_dir}/taosadapter || :
${csudo}rm -f ${bin_link_dir}/taosBenchmark || :
${csudo}rm -f ${bin_link_dir}/taosdemo || :
${csudo}rm -f ${bin_link_dir}/set_core || :
${csudo}rm -f ${bin_link_dir}/taoskeeper || :
${csudo}rm -f ${cfg_link_dir}/*.new || :
${csudo}rm -f ${inc_link_dir}/taos.h || :
${csudo}rm -f ${inc_link_dir}/taosdef.h || :

40
source/client/src/clientMain.c
View File

@ -108,7 +108,7 @@ TAOS *taos_connect(const char *ip, const char *user, const char *pass, const cha
if (pass == NULL) {
pass = TSDB_DEFAULT_PASS;
}
STscObj *pObj = taos_connect_internal(ip, user, pass, NULL, db, port, CONN_TYPE__QUERY);
if (pObj) {
int64_t *rid = taosMemoryCalloc(1, sizeof(int64_t));
@ -191,7 +191,7 @@ void taos_free_result(TAOS_RES *res) {
taosArrayDestroyP(pRsp->rsp.blockData, taosMemoryFree);
taosArrayDestroy(pRsp->rsp.blockDataLen);
taosArrayDestroyP(pRsp->rsp.blockTbName, taosMemoryFree);
taosArrayDestroyP(pRsp->rsp.blockSchema, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pRsp->rsp.blockSchema, (FDelete)tDeleteSchemaWrapper);
// taosx
taosArrayDestroy(pRsp->rsp.createTableLen);
taosArrayDestroyP(pRsp->rsp.createTableReq, taosMemoryFree);
@ -204,7 +204,7 @@ void taos_free_result(TAOS_RES *res) {
taosArrayDestroyP(pRsp->rsp.blockData, taosMemoryFree);
taosArrayDestroy(pRsp->rsp.blockDataLen);
taosArrayDestroyP(pRsp->rsp.blockTbName, taosMemoryFree);
taosArrayDestroyP(pRsp->rsp.blockSchema, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pRsp->rsp.blockSchema, (FDelete)tDeleteSchemaWrapper);
pRsp->resInfo.pRspMsg = NULL;
doFreeReqResultInfo(&pRsp->resInfo);
taosMemoryFree(pRsp);
@ -359,11 +359,11 @@ int taos_print_row(char *str, TAOS_ROW row, TAOS_FIELD *fields, int num_fields)
case TSDB_DATA_TYPE_NCHAR: {
int32_t charLen = varDataLen((char *)row[i] - VARSTR_HEADER_SIZE);
if (fields[i].type == TSDB_DATA_TYPE_BINARY) {
if(ASSERT(charLen <= fields[i].bytes && charLen >= 0)){
if (ASSERT(charLen <= fields[i].bytes && charLen >= 0)) {
tscError("taos_print_row error binary. charLen:%d, fields[i].bytes:%d", charLen, fields[i].bytes);
}
} else {
if(ASSERT(charLen <= fields[i].bytes * TSDB_NCHAR_SIZE && charLen >= 0)){
if (ASSERT(charLen <= fields[i].bytes * TSDB_NCHAR_SIZE && charLen >= 0)) {
tscError("taos_print_row error. charLen:%d, fields[i].bytes:%d", charLen, fields[i].bytes);
}
}
@ -705,16 +705,16 @@ int taos_get_current_db(TAOS *taos, char *database, int len, int *required) {
int code = TSDB_CODE_SUCCESS;
taosThreadMutexLock(&pTscObj->mutex);
if(database == NULL || len <= 0){
if(required != NULL) *required = strlen(pTscObj->db) + 1;
if (database == NULL || len <= 0) {
if (required != NULL) *required = strlen(pTscObj->db) + 1;
terrno = TSDB_CODE_INVALID_PARA;
code = -1;
}else if(len < strlen(pTscObj->db) + 1){
} else if (len < strlen(pTscObj->db) + 1) {
tstrncpy(database, pTscObj->db, len);
if(required) *required = strlen(pTscObj->db) + 1;
if (required) *required = strlen(pTscObj->db) + 1;
terrno = TSDB_CODE_INVALID_PARA;
code = -1;
}else{
} else {
strcpy(database, pTscObj->db);
code = 0;
}
@ -741,6 +741,7 @@ static void destoryCatalogReq(SCatalogReq *pCatalogReq) {
taosArrayDestroy(pCatalogReq->pUser);
taosArrayDestroy(pCatalogReq->pTableIndex);
taosArrayDestroy(pCatalogReq->pTableCfg);
taosArrayDestroy(pCatalogReq->pTableTag);
taosMemoryFree(pCatalogReq);
}
@ -975,8 +976,10 @@ void doAsyncQuery(SRequestObj *pRequest, bool updateMetaForce) {
if (TSDB_CODE_SUCCESS == code) {
pRequest->stmtType = pRequest->pQuery->pRoot->type;
phaseAsyncQuery(pWrapper);
} else {
code = phaseAsyncQuery(pWrapper);
}
if (TSDB_CODE_SUCCESS != code) {
tscError("0x%" PRIx64 " error happens, code:%d - %s, reqId:0x%" PRIx64, pRequest->self, code, tstrerror(code),
pRequest->requestId);
destorySqlCallbackWrapper(pWrapper);
@ -1042,11 +1045,11 @@ static void fetchCallback(void *pResult, void *param, int32_t code) {
}
void taos_fetch_rows_a(TAOS_RES *res, __taos_async_fn_t fp, void *param) {
if(ASSERT(res != NULL && fp != NULL)){
if (ASSERT(res != NULL && fp != NULL)) {
tscError("taos_fetch_rows_a invalid paras");
return;
}
if(ASSERT(TD_RES_QUERY(res))){
if (ASSERT(TD_RES_QUERY(res))) {
tscError("taos_fetch_rows_a res is NULL");
return;
}
@ -1092,11 +1095,11 @@ void taos_fetch_rows_a(TAOS_RES *res, __taos_async_fn_t fp, void *param) {
}
void taos_fetch_raw_block_a(TAOS_RES *res, __taos_async_fn_t fp, void *param) {
if(ASSERT(res != NULL && fp != NULL)){
if (ASSERT(res != NULL && fp != NULL)) {
tscError("taos_fetch_rows_a invalid paras");
return;
}
if(ASSERT(TD_RES_QUERY(res))){
if (ASSERT(TD_RES_QUERY(res))) {
tscError("taos_fetch_rows_a res is NULL");
return;
}
@ -1111,11 +1114,11 @@ void taos_fetch_raw_block_a(TAOS_RES *res, __taos_async_fn_t fp, void *param) {
}
const void *taos_get_raw_block(TAOS_RES *res) {
if(ASSERT(res != NULL)){
if (ASSERT(res != NULL)) {
tscError("taos_fetch_rows_a invalid paras");
return NULL;
}
if(ASSERT(TD_RES_QUERY(res))){
if (ASSERT(TD_RES_QUERY(res))) {
tscError("taos_fetch_rows_a res is NULL");
return NULL;
}
@ -1273,7 +1276,6 @@ _return:
return code;
}
int taos_load_table_info(TAOS *taos, const char *tableNameList) {
if (NULL == taos) {
terrno = TSDB_CODE_TSC_DISCONNECTED;

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

@ -92,7 +92,6 @@ int32_t processConnectRsp(void* param, SDataBuf* pMsg, int32_t code) {
goto End;
}
/*assert(connectRsp.epSet.numOfEps > 0);*/
if (connectRsp.epSet.numOfEps == 0) {
setErrno(pRequest, TSDB_CODE_APP_ERROR);
tsem_post(&pRequest->body.rspSem);

56
source/client/src/clientSml.c
View File

@ -534,7 +534,7 @@ static int32_t smlGenerateSchemaAction(SSchema *colField, SHashObj *colHash, SSm
uint16_t *index = colHash ? (uint16_t *)taosHashGet(colHash, kv->key, kv->keyLen) : NULL;
if (index) {
if (colField[*index].type != kv->type) {
uError("SML:0x%" PRIx64 " point type and db type mismatch. point type: %d, db type: %d, key: %s", info->id, colField[*index].type, kv->type, kv->key);
uError("SML:0x%" PRIx64 " point type and db type mismatch. db type: %d, point type: %d, key: %s", info->id, colField[*index].type, kv->type, kv->key);
return TSDB_CODE_SML_INVALID_DATA;
}
@ -558,10 +558,15 @@ static int32_t smlGenerateSchemaAction(SSchema *colField, SHashObj *colHash, SSm
return 0;
}
#define BOUNDARY 1024
static int32_t smlFindNearestPowerOf2(int32_t length, uint8_t type) {
int32_t result = 1;
while (result <= length) {
result *= 2;
if (length >= BOUNDARY){
result = length;
}else{
while (result <= length) {
result *= 2;
}
}
if (type == TSDB_DATA_TYPE_BINARY && result > TSDB_MAX_BINARY_LEN - VARSTR_HEADER_SIZE) {
result = TSDB_MAX_BINARY_LEN - VARSTR_HEADER_SIZE;
@ -649,6 +654,17 @@ static int32_t smlBuildFieldsList(SSmlHandle *info, SSchema *schemaField, SHashO
field->bytes = getBytes(kv->type, kv->length);
}
}
int32_t maxLen = isTag ? TSDB_MAX_TAGS_LEN : TSDB_MAX_BYTES_PER_ROW;
int32_t len = 0;
for (int j = 0; j < taosArrayGetSize(results); ++j) {
SField *field = taosArrayGet(results, j);
len += field->bytes;
}
if(len > maxLen){
return isTag ? TSDB_CODE_PAR_INVALID_TAGS_LENGTH : TSDB_CODE_PAR_INVALID_ROW_LENGTH;
}
return TSDB_CODE_SUCCESS;
}
@ -781,11 +797,15 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
code = smlBuildFieldsList(info, NULL, NULL, sTableData->tags, pTags, 0, true);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlBuildFieldsList tag1 failed. %s", info->id, pName.tname);
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
code = smlBuildFieldsList(info, NULL, NULL, sTableData->cols, pColumns, 0, false);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlBuildFieldsList col1 failed. %s", info->id, pName.tname);
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
code = smlSendMetaMsg(info, &pName, pColumns, pTags, NULL, SCHEMA_ACTION_CREATE_STABLE);
@ -837,6 +857,23 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
pTableMeta->tableInfo.numOfColumns, true);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlBuildFieldsList tag2 failed. %s", info->id, pName.tname);
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
if (taosArrayGetSize(pTags) + pTableMeta->tableInfo.numOfColumns > TSDB_MAX_COLUMNS) {
uError("SML:0x%" PRIx64 " too many columns than 4096", info->id);
code = TSDB_CODE_PAR_TOO_MANY_COLUMNS;
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
if (taosArrayGetSize(pTags) > TSDB_MAX_TAGS) {
uError("SML:0x%" PRIx64 " too many tags than 128", info->id);
code = TSDB_CODE_PAR_INVALID_TAGS_NUM;
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
@ -891,6 +928,16 @@ static int32_t smlModifyDBSchemas(SSmlHandle *info) {
pTableMeta->tableInfo.numOfColumns, false);
if (code != TSDB_CODE_SUCCESS) {
uError("SML:0x%" PRIx64 " smlBuildFieldsList col2 failed. %s", info->id, pName.tname);
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
if (taosArrayGetSize(pColumns) + pTableMeta->tableInfo.numOfTags > TSDB_MAX_COLUMNS) {
uError("SML:0x%" PRIx64 " too many columns than 4096", info->id);
code = TSDB_CODE_PAR_TOO_MANY_COLUMNS;
taosArrayDestroy(pColumns);
taosArrayDestroy(pTags);
goto end;
}
@ -1514,7 +1561,8 @@ static int smlProcess(SSmlHandle *info, char *lines[], char *rawLine, char *rawL
do {
code = smlModifyDBSchemas(info);
if (code == 0 || code == TSDB_CODE_SML_INVALID_DATA) break;
if (code == 0 || code == TSDB_CODE_SML_INVALID_DATA || code == TSDB_CODE_PAR_TOO_MANY_COLUMNS
|| code == TSDB_CODE_PAR_INVALID_TAGS_NUM) break;
taosMsleep(100);
uInfo("SML:0x%" PRIx64 " smlModifyDBSchemas retry code:%s, times:%d", info->id, tstrerror(code), retryNum);
} while (retryNum++ < taosHashGetSize(info->superTables) * MAX_RETRY_TIMES);

2
source/client/src/clientSmlJson.c
View File

@ -575,7 +575,7 @@ static int32_t smlConvertJSONString(SSmlKv *pVal, char *typeStr, cJSON *value) {
uError("OTD:invalid type(%s) for JSON String", typeStr);
return TSDB_CODE_TSC_INVALID_JSON_TYPE;
}
pVal->length = (int16_t)strlen(value->valuestring);
pVal->length = strlen(value->valuestring);
if (pVal->type == TSDB_DATA_TYPE_BINARY && pVal->length > TSDB_MAX_BINARY_LEN - VARSTR_HEADER_SIZE) {
return TSDB_CODE_PAR_INVALID_VAR_COLUMN_LEN;

4
source/client/src/clientStmt.c
View File

@ -325,7 +325,7 @@ int32_t stmtCleanExecInfo(STscStmt* pStmt, bool keepTable, bool deepClean) {
taosHashCleanup(pStmt->exec.pBlockHash);
pStmt->exec.pBlockHash = NULL;
tDestroySSubmitTbData(pStmt->exec.pCurrTbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(pStmt->exec.pCurrTbData, TSDB_MSG_FLG_ENCODE);
taosMemoryFreeClear(pStmt->exec.pCurrTbData);
STMT_ERR_RET(stmtCleanBindInfo(pStmt));
@ -895,7 +895,7 @@ int stmtExec(TAOS_STMT* stmt) {
if (STMT_TYPE_QUERY == pStmt->sql.type) {
launchQueryImpl(pStmt->exec.pRequest, pStmt->sql.pQuery, true, NULL);
} else {
tDestroySSubmitTbData(pStmt->exec.pCurrTbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(pStmt->exec.pCurrTbData, TSDB_MSG_FLG_ENCODE);
taosMemoryFreeClear(pStmt->exec.pCurrTbData);
STMT_ERR_RET(qCloneCurrentTbData(pStmt->exec.pCurrBlock, &pStmt->exec.pCurrTbData));

32
source/client/src/clientTmq.c
View File

@ -864,7 +864,7 @@ static void* tmqFreeRspWrapper(SMqRspWrapper* rspWrapper) {
taosArrayDestroyP(pRsp->dataRsp.blockData, taosMemoryFree);
taosArrayDestroy(pRsp->dataRsp.blockDataLen);
taosArrayDestroyP(pRsp->dataRsp.blockTbName, taosMemoryFree);
taosArrayDestroyP(pRsp->dataRsp.blockSchema, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pRsp->dataRsp.blockSchema, (FDelete)tDeleteSchemaWrapper);
} else if (rspWrapper->tmqRspType == TMQ_MSG_TYPE__POLL_META_RSP) {
SMqPollRspWrapper* pRsp = (SMqPollRspWrapper*)rspWrapper;
taosMemoryFreeClear(pRsp->pEpset);
@ -877,7 +877,7 @@ static void* tmqFreeRspWrapper(SMqRspWrapper* rspWrapper) {
taosArrayDestroyP(pRsp->taosxRsp.blockData, taosMemoryFree);
taosArrayDestroy(pRsp->taosxRsp.blockDataLen);
taosArrayDestroyP(pRsp->taosxRsp.blockTbName, taosMemoryFree);
taosArrayDestroyP(pRsp->taosxRsp.blockSchema, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pRsp->taosxRsp.blockSchema, (FDelete)tDeleteSchemaWrapper);
// taosx
taosArrayDestroy(pRsp->taosxRsp.createTableLen);
taosArrayDestroyP(pRsp->taosxRsp.createTableReq, taosMemoryFree);
@ -1377,7 +1377,7 @@ static void initClientTopicFromRsp(SMqClientTopic* pTopic, SMqSubTopicEp* pTopic
tstrncpy(pTopic->topicName, pTopicEp->topic, TSDB_TOPIC_FNAME_LEN);
tstrncpy(pTopic->db, pTopicEp->db, TSDB_DB_FNAME_LEN);
tscDebug("consumer:0x%" PRIx64 ", update topic:%s, numOfVgs:%d", tmq->consumerId, pTopic->topicName, vgNumGet);
tscDebug("consumer:0x%" PRIx64 ", update topic:%s, new numOfVgs:%d", tmq->consumerId, pTopic->topicName, vgNumGet);
pTopic->vgs = taosArrayInit(vgNumGet, sizeof(SMqClientVg));
for (int32_t j = 0; j < vgNumGet; j++) {
@ -1447,14 +1447,14 @@ static bool doUpdateLocalEp(tmq_t* tmq, int32_t epoch, const SMqAskEpRsp* pRsp)
SMqClientTopic* pTopicCur = taosArrayGet(tmq->clientTopics, i);
if (pTopicCur->vgs) {
int32_t vgNumCur = taosArrayGetSize(pTopicCur->vgs);
tscDebug("consumer:0x%" PRIx64 ", new vg num: %d", tmq->consumerId, vgNumCur);
tscDebug("consumer:0x%" PRIx64 ", current vg num: %d", tmq->consumerId, vgNumCur);
for (int32_t j = 0; j < vgNumCur; j++) {
SMqClientVg* pVgCur = taosArrayGet(pTopicCur->vgs, j);
makeTopicVgroupKey(vgKey, pTopicCur->topicName, pVgCur->vgId);
char buf[80];
tFormatOffset(buf, 80, &pVgCur->currentOffset);
tscDebug("consumer:0x%" PRIx64 ", epoch:%d vgId:%d vgKey:%s, offset:%s", tmq->consumerId, epoch, pVgCur->vgId,
tscDebug("consumer:0x%" PRIx64 ", doUpdateLocalEp current vg, epoch:%d vgId:%d vgKey:%s, offset:%s", tmq->consumerId, tmq->epoch, pVgCur->vgId,
vgKey, buf);
SVgroupSaveInfo info = {.offset = pVgCur->currentOffset, .numOfRows = pVgCur->numOfRows};
@ -1664,11 +1664,7 @@ static int32_t doTmqPollImpl(tmq_t* pTmq, SMqClientTopic* pTopic, SMqClientVg* p
return handleErrorBeforePoll(pVg, pTmq);
}
sendInfo->msgInfo = (SDataBuf){
.pData = msg,
.len = msgSize,
.handle = NULL,
};
sendInfo->msgInfo = (SDataBuf){ .pData = msg, .len = msgSize, .handle = NULL };
sendInfo->requestId = req.reqId;
sendInfo->requestObjRefId = 0;
@ -1794,8 +1790,9 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
pVg->epSet = *pollRspWrapper->pEpset;
}
// update the local offset value only for the returned values.
pVg->currentOffset = pDataRsp->rspOffset;
if(pDataRsp->rspOffset.type != 0){ // if offset is validate
pVg->currentOffset = pDataRsp->rspOffset; // update the local offset value only for the returned values.
}
atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
char buf[80];
@ -1805,12 +1802,13 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
" total:%" PRId64 " reqId:0x%" PRIx64,
tmq->consumerId, pVg->vgId, buf, pVg->numOfRows, tmq->totalRows, pollRspWrapper->reqId);
pRspWrapper = tmqFreeRspWrapper(pRspWrapper);
pVg->emptyBlockReceiveTs = taosGetTimestampMs();
taosFreeQitem(pollRspWrapper);
} else { // build rsp
int64_t numOfRows = 0;
SMqRspObj* pRsp = tmqBuildRspFromWrapper(pollRspWrapper, pVg, &numOfRows);
tmq->totalRows += numOfRows;
pVg->emptyBlockReceiveTs = 0;
tscDebug("consumer:0x%" PRIx64 " process poll rsp, vgId:%d, offset:%s, blocks:%d, rows:%" PRId64
" vg total:%" PRId64 " total:%" PRId64 ", reqId:0x%" PRIx64,
tmq->consumerId, pVg->vgId, buf, pDataRsp->blockNum, numOfRows, pVg->numOfRows, tmq->totalRows,
@ -1832,7 +1830,9 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
if (pollRspWrapper->metaRsp.head.epoch == consumerEpoch) {
SMqClientVg* pVg = pollRspWrapper->vgHandle;
pVg->currentOffset = pollRspWrapper->metaRsp.rspOffset;
if(pollRspWrapper->metaRsp.rspOffset.type != 0){ // if offset is validate
pVg->currentOffset = pollRspWrapper->metaRsp.rspOffset;
}
atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
// build rsp
SMqMetaRspObj* pRsp = tmqBuildMetaRspFromWrapper(pollRspWrapper);
@ -1850,7 +1850,9 @@ static void* tmqHandleAllRsp(tmq_t* tmq, int64_t timeout, bool pollIfReset) {
if (pollRspWrapper->taosxRsp.head.epoch == consumerEpoch) {
SMqClientVg* pVg = pollRspWrapper->vgHandle;
pVg->currentOffset = pollRspWrapper->taosxRsp.rspOffset;
if(pollRspWrapper->taosxRsp.rspOffset.type != 0){ // if offset is validate
pVg->currentOffset = pollRspWrapper->taosxRsp.rspOffset;
}
atomic_store_32(&pVg->vgStatus, TMQ_VG_STATUS__IDLE);
if (pollRspWrapper->taosxRsp.blockNum == 0) {

41
source/client/test/clientTests.cpp
View File

@ -165,7 +165,7 @@ void* queryThread(void* arg) {
int32_t numOfThreads = 1;
void tmq_commit_cb_print(tmq_t* pTmq, int32_t code, void* param) {
printf("auto commit success, code:%d\n\n\n\n", code);
// printf("auto commit success, code:%d\n", code);
}
void* doConsumeData(void* param) {
@ -1059,13 +1059,18 @@ TEST(clientCase, sub_tb_test) {
ASSERT_NE(pConn, nullptr);
tmq_conf_t* conf = tmq_conf_new();
int32_t ts = taosGetTimestampMs()%INT32_MAX;
char consumerGroupid[128] = {0};
sprintf(consumerGroupid, "group_id_%d", ts);
tmq_conf_set(conf, "enable.auto.commit", "true");
tmq_conf_set(conf, "auto.commit.interval.ms", "1000");
tmq_conf_set(conf, "group.id", "cgrpName45");
tmq_conf_set(conf, "auto.commit.interval.ms", "2000");
tmq_conf_set(conf, "group.id", consumerGroupid);
tmq_conf_set(conf, "td.connect.user", "root");
tmq_conf_set(conf, "td.connect.pass", "taosdata");
tmq_conf_set(conf, "auto.offset.reset", "earliest");
tmq_conf_set(conf, "experimental.snapshot.enable", "true");
tmq_conf_set(conf, "experimental.snapshot.enable", "false");
tmq_conf_set(conf, "msg.with.table.name", "true");
tmq_conf_set_auto_commit_cb(conf, tmq_commit_cb_print, NULL);
@ -1074,7 +1079,7 @@ TEST(clientCase, sub_tb_test) {
// 创建订阅 topics 列表
tmq_list_t* topicList = tmq_list_new();
tmq_list_append(topicList, "topic_t2");
tmq_list_append(topicList, "topic_t1");
// 启动订阅
tmq_subscribe(tmq, topicList);
@ -1086,22 +1091,22 @@ TEST(clientCase, sub_tb_test) {
int32_t precision = 0;
int32_t totalRows = 0;
int32_t msgCnt = 0;
int32_t timeout = 25000;
int32_t timeout = 2500000;
int32_t count = 0;
while (1) {
TAOS_RES* pRes = tmq_consumer_poll(tmq, timeout);
if (pRes) {
char buf[1024];
char buf[128];
const char* topicName = tmq_get_topic_name(pRes);
const char* dbName = tmq_get_db_name(pRes);
int32_t vgroupId = tmq_get_vgroup_id(pRes);
printf("topic: %s\n", topicName);
printf("db: %s\n", dbName);
printf("vgroup id: %d\n", vgroupId);
// const char* dbName = tmq_get_db_name(pRes);
// int32_t vgroupId = tmq_get_vgroup_id(pRes);
//
// printf("topic: %s\n", topicName);
// printf("db: %s\n", dbName);
// printf("vgroup id: %d\n", vgroupId);
while (1) {
TAOS_ROW row = taos_fetch_row(pRes);
@ -1111,16 +1116,14 @@ TEST(clientCase, sub_tb_test) {
fields = taos_fetch_fields(pRes);
numOfFields = taos_field_count(pRes);
precision = taos_result_precision(pRes);
taos_print_row(buf, row, fields, numOfFields);
totalRows += 1;
printf("precision: %d, row content: %s\n", precision, buf);
// if (totalRows % 100000 == 0) {
taos_print_row(buf, row, fields, numOfFields);
printf("row content: %s\n", buf);
// }
}
taos_free_result(pRes);
// if ((++count) > 1) {
// break;
// }
} else {
break;
}

4
source/common/src/tdatablock.c
View File

@ -2374,7 +2374,7 @@ int32_t buildSubmitReqFromDataBlock(SSubmitReq2** ppReq, const SSDataBlock* pDat
}
SRow* pRow = NULL;
if ((terrno = tRowBuild(pVals, pTSchema, &pRow)) < 0) {
tDestroySSubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
goto _end;
}
ASSERT(pRow);
@ -2388,7 +2388,7 @@ _end:
if (terrno != 0) {
*ppReq = NULL;
if (pReq) {
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFreeClear(pReq);
}

4
source/common/src/tdataformat.c
View File

@ -1509,7 +1509,9 @@ void tTagSetCid(const STag *pTag, int16_t iTag, int16_t cid) {
// STSchema ========================================
STSchema *tBuildTSchema(SSchema *aSchema, int32_t numOfCols, int32_t version) {
STSchema *pTSchema = taosMemoryCalloc(1, sizeof(STSchema) + sizeof(STColumn) * numOfCols);
if (pTSchema == NULL) return NULL;
if (pTSchema == NULL) {
return NULL;
}
pTSchema->numOfCols = numOfCols;
pTSchema->version = version;

37
source/common/src/tmsg.c
View File

@ -1409,6 +1409,8 @@ int32_t tDeserializeSAlterUserReq(void *buf, int32_t bufLen, SAlterUserReq *pReq
return 0;
}
void tFreeSAlterUserReq(SAlterUserReq *pReq) { taosMemoryFreeClear(pReq->tagCond); }
int32_t tSerializeSGetUserAuthReq(void *buf, int32_t bufLen, SGetUserAuthReq *pReq) {
SEncoder encoder = {0};
tEncoderInit(&encoder, buf, bufLen);
@ -1635,6 +1637,7 @@ int32_t tDeserializeSGetUserAuthRspImpl(SDecoder *pDecoder, SGetUserAuthRsp *pRs
int32_t ref = 0;
if (tDecodeI32(pDecoder, &ref) < 0) return -1;
taosHashPut(pRsp->useDbs, key, strlen(key), &ref, sizeof(ref));
taosMemoryFree(key);
}
}
@ -1831,7 +1834,6 @@ int32_t tSerializeSCreateFuncReq(void *buf, int32_t bufLen, SCreateFuncReq *pReq
if (tEncodeCStr(&encoder, pReq->pComment) < 0) return -1;
}
if (tEncodeI8(&encoder, pReq->orReplace) < 0) return -1;
tEndEncode(&encoder);
@ -1876,7 +1878,6 @@ int32_t tDeserializeSCreateFuncReq(void *buf, int32_t bufLen, SCreateFuncReq *pR
if (tDecodeCStrTo(&decoder, pReq->pComment) < 0) return -1;
}
if (!tDecodeIsEnd(&decoder)) {
if (tDecodeI8(&decoder, &pReq->orReplace) < 0) return -1;
} else {
@ -2053,12 +2054,12 @@ int32_t tDeserializeSRetrieveFuncRsp(void *buf, int32_t bufLen, SRetrieveFuncRsp
if (pRsp->pFuncExtraInfos == NULL) return -1;
if (tDecodeIsEnd(&decoder)) {
for (int32_t i = 0; i < pRsp->numOfFuncs; ++i) {
SFuncExtraInfo extraInfo = { 0 };
SFuncExtraInfo extraInfo = {0};
taosArrayPush(pRsp->pFuncExtraInfos, &extraInfo);
}
} else {
for (int32_t i = 0; i < pRsp->numOfFuncs; ++i) {
SFuncExtraInfo extraInfo = { 0 };
SFuncExtraInfo extraInfo = {0};
if (tDecodeI32(&decoder, &extraInfo.funcVersion) < 0) return -1;
if (tDecodeI64(&decoder, &extraInfo.funcCreatedTime) < 0) return -1;
taosArrayPush(pRsp->pFuncExtraInfos, &extraInfo);
@ -5328,9 +5329,9 @@ int32_t tSerializeSMqPollReq(void *buf, int32_t bufLen, SMqPollReq *pReq) {
int32_t tDeserializeSMqPollReq(void *buf, int32_t bufLen, SMqPollReq *pReq) {
int32_t headLen = sizeof(SMsgHead);
SMsgHead *pHead = buf;
pHead->vgId = pReq->head.vgId;
pHead->contLen = pReq->head.contLen;
// SMsgHead *pHead = buf;
// pHead->vgId = pReq->head.vgId;
// pHead->contLen = pReq->head.contLen;
SDecoder decoder = {0};
tDecoderInit(&decoder, (char *)buf + headLen, bufLen - headLen);
@ -6838,10 +6839,8 @@ int32_t tEncodeSTqOffsetVal(SEncoder *pEncoder, const STqOffsetVal *pOffsetVal)
if (tEncodeI64(pEncoder, pOffsetVal->ts) < 0) return -1;
} else if (pOffsetVal->type == TMQ_OFFSET__LOG) {
if (tEncodeI64(pEncoder, pOffsetVal->version) < 0) return -1;
} else if (pOffsetVal->type < 0) {
// do nothing
} else {
ASSERT(0);
// do nothing
}
return 0;
}
@ -6853,10 +6852,8 @@ int32_t tDecodeSTqOffsetVal(SDecoder *pDecoder, STqOffsetVal *pOffsetVal) {
if (tDecodeI64(pDecoder, &pOffsetVal->ts) < 0) return -1;
} else if (pOffsetVal->type == TMQ_OFFSET__LOG) {
if (tDecodeI64(pDecoder, &pOffsetVal->version) < 0) return -1;
} else if (pOffsetVal->type < 0) {
// do nothing
} else {
ASSERT(0);
// do nothing
}
return 0;
}
@ -7061,7 +7058,7 @@ void tDeleteSMqDataRsp(SMqDataRsp *pRsp) {
pRsp->blockDataLen = taosArrayDestroy(pRsp->blockDataLen);
taosArrayDestroyP(pRsp->blockData, (FDelete)taosMemoryFree);
pRsp->blockData = NULL;
taosArrayDestroyP(pRsp->blockSchema, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pRsp->blockSchema, (FDelete)tDeleteSchemaWrapper);
pRsp->blockSchema = NULL;
taosArrayDestroyP(pRsp->blockTbName, (FDelete)taosMemoryFree);
pRsp->blockTbName = NULL;
@ -7162,7 +7159,7 @@ void tDeleteSTaosxRsp(STaosxRsp *pRsp) {
pRsp->blockDataLen = NULL;
taosArrayDestroyP(pRsp->blockData, (FDelete)taosMemoryFree);
pRsp->blockData = NULL;
taosArrayDestroyP(pRsp->blockSchema, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pRsp->blockSchema, (FDelete)tDeleteSchemaWrapper);
pRsp->blockSchema = NULL;
taosArrayDestroyP(pRsp->blockTbName, (FDelete)taosMemoryFree);
pRsp->blockTbName = NULL;
@ -7335,7 +7332,7 @@ _exit:
return 0;
}
int32_t tEncodeSSubmitReq2(SEncoder *pCoder, const SSubmitReq2 *pReq) {
int32_t tEncodeSubmitReq(SEncoder *pCoder, const SSubmitReq2 *pReq) {
if (tStartEncode(pCoder) < 0) return -1;
if (tEncodeU64v(pCoder, taosArrayGetSize(pReq->aSubmitTbData)) < 0) return -1;
@ -7347,7 +7344,7 @@ int32_t tEncodeSSubmitReq2(SEncoder *pCoder, const SSubmitReq2 *pReq) {
return 0;
}
int32_t tDecodeSSubmitReq2(SDecoder *pCoder, SSubmitReq2 *pReq) {
int32_t tDecodeSubmitReq(SDecoder *pCoder, SSubmitReq2 *pReq) {
int32_t code = 0;
memset(pReq, 0, sizeof(*pReq));
@ -7390,7 +7387,7 @@ _exit:
return code;
}
void tDestroySSubmitTbData(SSubmitTbData *pTbData, int32_t flag) {
void tDestroySubmitTbData(SSubmitTbData *pTbData, int32_t flag) {
if (NULL == pTbData) {
return;
}
@ -7436,14 +7433,14 @@ void tDestroySSubmitTbData(SSubmitTbData *pTbData, int32_t flag) {
}
}
void tDestroySSubmitReq2(SSubmitReq2 *pReq, int32_t flag) {
void tDestroySubmitReq(SSubmitReq2 *pReq, int32_t flag) {
if (pReq->aSubmitTbData == NULL) return;
int32_t nSubmitTbData = TARRAY_SIZE(pReq->aSubmitTbData);
SSubmitTbData *aSubmitTbData = (SSubmitTbData *)TARRAY_DATA(pReq->aSubmitTbData);
for (int32_t i = 0; i < nSubmitTbData; i++) {
tDestroySSubmitTbData(&aSubmitTbData[i], flag);
tDestroySubmitTbData(&aSubmitTbData[i], flag);
}
taosArrayDestroy(pReq->aSubmitTbData);
pReq->aSubmitTbData = NULL;

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

@ -122,10 +122,8 @@ int32_t tNameLen(const SName* name) {
int32_t len2 = (int32_t)strlen(name->tname);
if (name->type == TSDB_DB_NAME_T) {
ASSERT(len2 == 0);
return len + len1 + TSDB_NAME_DELIMITER_LEN;
} else {
ASSERT(len2 > 0);
return len + len1 + len2 + TSDB_NAME_DELIMITER_LEN * 2;
}
}

1
source/common/src/ttszip.c
View File

@ -982,7 +982,6 @@ void tsBufSetCursor(STSBuf* pTSBuf, STSCursor* pCur) {
return;
}
// assert(pCur->vgroupIndex != -1 && pCur->tsIndex >= 0 && pCur->blockIndex >= 0);
if (pCur->vgroupIndex != -1) {
tsBufGetBlock(pTSBuf, pCur->vgroupIndex, pCur->blockIndex);
}

2
source/dnode/mgmt/mgmt_vnode/src/vmHandle.c
View File

@ -519,7 +519,7 @@ SArray *vmGetMsgHandles() {
if (dmSetMgmtHandle(pArray, TDMT_VND_TMQ_COMMIT_OFFSET, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_TMQ_ADD_CHECKINFO, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_TMQ_DEL_CHECKINFO, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_TMQ_CONSUME, vmPutMsgToFetchQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_TMQ_CONSUME, vmPutMsgToQueryQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_DELETE, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_BATCH_DEL, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;
if (dmSetMgmtHandle(pArray, TDMT_VND_COMMIT, vmPutMsgToWriteQueue, 0) == NULL) goto _OVER;

12
source/dnode/mgmt/mgmt_vnode/src/vmWorker.c
View File

@ -113,8 +113,16 @@ static void vmProcessFetchQueue(SQueueInfo *pInfo, STaosQall *qall, int32_t numO
int32_t code = vnodeProcessFetchMsg(pVnode->pImpl, pMsg, pInfo);
if (code != 0) {
if (terrno != 0) code = terrno;
dGError("vnodeProcessFetchMsg vgId:%d, msg:%p failed to fetch since %s", pVnode->vgId, pMsg, terrstr());
if (terrno != 0) {
code = terrno;
}
if (code == TSDB_CODE_WAL_LOG_NOT_EXIST) {
dGDebug("vnodeProcessFetchMsg vgId:%d, msg:%p failed to fetch since %s", pVnode->vgId, pMsg, terrstr());
} else {
dGError("vnodeProcessFetchMsg vgId:%d, msg:%p failed to fetch since %s", pVnode->vgId, pMsg, terrstr());
}
vmSendRsp(pMsg, code);
}

3
source/dnode/mnode/impl/inc/mndConsumer.h
View File

@ -23,13 +23,12 @@ extern "C" {
#endif
enum {
MQ_CONSUMER_STATUS__MODIFY = 1,
MQ_CONSUMER_STATUS_REBALANCE = 1,
// MQ_CONSUMER_STATUS__MODIFY_IN_REB, // this value is not used anymore
MQ_CONSUMER_STATUS__READY,
MQ_CONSUMER_STATUS__LOST,
// MQ_CONSUMER_STATUS__LOST_IN_REB, // this value is not used anymore
MQ_CONSUMER_STATUS__LOST_REBD,
MQ_CONSUMER_STATUS__REMOVED,
};
int32_t mndInitConsumer(SMnode *pMnode);

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

@ -142,7 +142,7 @@ typedef enum {
CONSUMER_UPDATE__REMOVE,
CONSUMER_UPDATE__LOST,
CONSUMER_UPDATE__RECOVER,
CONSUMER_UPDATE__MODIFY, // subscribe req need change consume topic
CONSUMER_UPDATE__REBALANCE, // subscribe req need change consume topic
} ECsmUpdateType;
typedef struct {

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

@ -50,6 +50,8 @@ void *mndBuildCreateVnodeReq(SMnode *, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *p
void *mndBuildDropVnodeReq(SMnode *, SDnodeObj *pDnode, SDbObj *pDb, SVgObj *pVgroup, int32_t *pContLen);
bool mndVgroupInDb(SVgObj *pVgroup, int64_t dbUid);
int32_t mndSplitVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SVgObj *pVgroup);
#ifdef __cplusplus
}
#endif

136
source/dnode/mnode/impl/src/mndConsumer.c
View File

@ -192,15 +192,18 @@ FAIL:
return -1;
}
// todo check the clear process
static int32_t mndProcessConsumerClearMsg(SRpcMsg *pMsg) {
SMnode *pMnode = pMsg->info.node;
SMqConsumerClearMsg *pClearMsg = pMsg->pCont;
SMqConsumerObj *pConsumer = mndAcquireConsumer(pMnode, pClearMsg->consumerId);
SMqConsumerObj *pConsumer = mndAcquireConsumer(pMnode, pClearMsg->consumerId);
if (pConsumer == NULL) {
mError("consumer:0x%"PRIx64" failed to be found to clear it", pClearMsg->consumerId);
return 0;
}
mInfo("receive consumer clear msg, consumer id %" PRId64 ", status %s", pClearMsg->consumerId,
mInfo("consumer:0x%" PRIx64 " needs to be cleared, status %s", pClearMsg->consumerId,
mndConsumerStatusName(pConsumer->status));
if (pConsumer->status != MQ_CONSUMER_STATUS__LOST_REBD) {
@ -215,6 +218,8 @@ static int32_t mndProcessConsumerClearMsg(SRpcMsg *pMsg) {
STrans *pTrans = mndTransCreate(pMnode, TRN_POLICY_ROLLBACK, TRN_CONFLICT_NOTHING, pMsg, "clear-csm");
if (pTrans == NULL) goto FAIL;
// this is the drop action, not the update action
if (mndSetConsumerDropLogs(pMnode, pTrans, pConsumerNew) != 0) goto FAIL;
if (mndTransPrepare(pMnode, pTrans) != 0) goto FAIL;
@ -299,28 +304,36 @@ static int32_t mndProcessMqTimerMsg(SRpcMsg *pMsg) {
if (status == MQ_CONSUMER_STATUS__READY) {
if (hbStatus > MND_CONSUMER_LOST_HB_CNT) {
SMqConsumerLostMsg *pLostMsg = rpcMallocCont(sizeof(SMqConsumerLostMsg));
if (pLostMsg == NULL) {
mError("consumer:0x%"PRIx64" failed to transfer consumer status to lost due to out of memory. alloc size:%d",
pConsumer->consumerId, (int32_t)sizeof(SMqConsumerLostMsg));
continue;
}
pLostMsg->consumerId = pConsumer->consumerId;
SRpcMsg rpcMsg = {
.msgType = TDMT_MND_TMQ_CONSUMER_LOST,
.pCont = pLostMsg,
.contLen = sizeof(SMqConsumerLostMsg),
};
.msgType = TDMT_MND_TMQ_CONSUMER_LOST, .pCont = pLostMsg, .contLen = sizeof(SMqConsumerLostMsg)};
mDebug("consumer:0x%"PRIx64" hb not received beyond threshold %d, set to lost", pConsumer->consumerId,
MND_CONSUMER_LOST_HB_CNT);
tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
}
} else if (status == MQ_CONSUMER_STATUS__LOST_REBD) {
// if the client is lost longer than one day, clear it. Otherwise, do nothing about the lost consumers.
if (hbStatus > MND_CONSUMER_LOST_CLEAR_THRESHOLD) {
SMqConsumerClearMsg *pClearMsg = rpcMallocCont(sizeof(SMqConsumerClearMsg));
if (pClearMsg == NULL) {
mError("consumer:0x%"PRIx64" failed to clear consumer due to out of memory. alloc size:%d",
pConsumer->consumerId, (int32_t)sizeof(SMqConsumerClearMsg));
continue;
}
pClearMsg->consumerId = pConsumer->consumerId;
SRpcMsg rpcMsg = {
.msgType = TDMT_MND_TMQ_LOST_CONSUMER_CLEAR,
.pCont = pClearMsg,
.contLen = sizeof(SMqConsumerClearMsg),
};
.msgType = TDMT_MND_TMQ_LOST_CONSUMER_CLEAR, .pCont = pClearMsg, .contLen = sizeof(SMqConsumerClearMsg)};
mDebug("consumer:0x%" PRIx64 " lost beyond threshold %d, clear it", pConsumer->consumerId,
MND_CONSUMER_LOST_CLEAR_THRESHOLD);
tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &rpcMsg);
}
} else if (status == MQ_CONSUMER_STATUS__LOST) {
@ -334,7 +347,7 @@ static int32_t mndProcessMqTimerMsg(SRpcMsg *pMsg) {
taosArrayPush(pRebSub->removedConsumers, &pConsumer->consumerId);
}
taosRUnLockLatch(&pConsumer->lock);
} else {
} else { // MQ_CONSUMER_STATUS_REBALANCE
taosRLockLatch(&pConsumer->lock);
int32_t newTopicNum = taosArrayGetSize(pConsumer->rebNewTopics);
@ -449,7 +462,6 @@ static int32_t mndProcessAskEpReq(SRpcMsg *pMsg) {
// 1. check consumer status
int32_t status = atomic_load_32(&pConsumer->status);
#if 1
if (status == MQ_CONSUMER_STATUS__LOST_REBD) {
mInfo("try to recover consumer:0x%" PRIx64, consumerId);
SMqConsumerRecoverMsg *pRecoverMsg = rpcMallocCont(sizeof(SMqConsumerRecoverMsg));
@ -463,7 +475,6 @@ static int32_t mndProcessAskEpReq(SRpcMsg *pMsg) {
tmsgPutToQueue(&pMnode->msgCb, WRITE_QUEUE, &pRpcMsg);
}
#endif
if (status != MQ_CONSUMER_STATUS__READY) {
mInfo("consumer:0x%" PRIx64 " not ready, status: %s", consumerId, mndConsumerStatusName(status));
@ -660,7 +671,7 @@ int32_t mndProcessSubscribeReq(SRpcMsg *pMsg) {
tstrncpy(pConsumerNew->clientId, subscribe.clientId, tListLen(pConsumerNew->clientId));
// set the update type
pConsumerNew->updateType = CONSUMER_UPDATE__MODIFY;
pConsumerNew->updateType = CONSUMER_UPDATE__REBALANCE;
taosArrayDestroy(pConsumerNew->assignedTopics);
pConsumerNew->assignedTopics = taosArrayDup(pTopicList, topicNameDup);
@ -673,7 +684,6 @@ int32_t mndProcessSubscribeReq(SRpcMsg *pMsg) {
if (mndTransPrepare(pMnode, pTrans) != 0) goto _over;
} else {
/*taosRLockLatch(&pExistedConsumer->lock);*/
int32_t status = atomic_load_32(&pExistedConsumer->status);
mInfo("receive subscribe request from existed consumer:0x%" PRIx64
@ -691,7 +701,7 @@ int32_t mndProcessSubscribeReq(SRpcMsg *pMsg) {
}
// set the update type
pConsumerNew->updateType = CONSUMER_UPDATE__MODIFY;
pConsumerNew->updateType = CONSUMER_UPDATE__REBALANCE;
taosArrayDestroy(pConsumerNew->assignedTopics);
pConsumerNew->assignedTopics = taosArrayDup(pTopicList, topicNameDup);
@ -870,9 +880,10 @@ static void updateConsumerStatus(SMqConsumerObj *pConsumer) {
int32_t status = pConsumer->status;
if (taosArrayGetSize(pConsumer->rebNewTopics) == 0 && taosArrayGetSize(pConsumer->rebRemovedTopics) == 0) {
if (status == MQ_CONSUMER_STATUS__MODIFY) {
if (status == MQ_CONSUMER_STATUS_REBALANCE) {
pConsumer->status = MQ_CONSUMER_STATUS__READY;
} else if (status == MQ_CONSUMER_STATUS__LOST) {
ASSERT(taosArrayGetSize(pConsumer->currentTopics) == 0);
pConsumer->status = MQ_CONSUMER_STATUS__LOST_REBD;
}
}
@ -881,7 +892,7 @@ static void updateConsumerStatus(SMqConsumerObj *pConsumer) {
// remove from new topic
static void removeFromNewTopicList(SMqConsumerObj *pConsumer, const char *pTopic) {
int32_t size = taosArrayGetSize(pConsumer->rebNewTopics);
for (int32_t i = 0; i < taosArrayGetSize(pConsumer->rebNewTopics); i++) {
for (int32_t i = 0; i < size; i++) {
char *p = taosArrayGetP(pConsumer->rebNewTopics, i);
if (strcmp(pTopic, p) == 0) {
taosArrayRemove(pConsumer->rebNewTopics, i);
@ -902,32 +913,57 @@ static void removeFromRemoveTopicList(SMqConsumerObj *pConsumer, const char *pTo
if (strcmp(pTopic, p) == 0) {
taosArrayRemove(pConsumer->rebRemovedTopics, i);
taosMemoryFree(p);
mDebug("consumer:0x%" PRIx64 " remove topic:%s in the removed topic list, remain removedTopics:%d",
pConsumer->consumerId, pTopic, (int)taosArrayGetSize(pConsumer->rebRemovedTopics));
break;
}
}
}
static void removeFromCurrentTopicList(SMqConsumerObj *pConsumer, const char *pTopic) {
int32_t sz = taosArrayGetSize(pConsumer->currentTopics);
for (int32_t i = 0; i < sz; i++) {
char *topic = taosArrayGetP(pConsumer->currentTopics, i);
if (strcmp(pTopic, topic) == 0) {
taosArrayRemove(pConsumer->currentTopics, i);
taosMemoryFree(topic);
mDebug("consumer:0x%" PRIx64 " remove topic:%s in the current topic list, remain currentTopics:%d",
pConsumer->consumerId, pTopic, (int)taosArrayGetSize(pConsumer->currentTopics));
break;
}
}
}
static bool existInCurrentTopicList(const SMqConsumerObj* pConsumer, const char* pTopic) {
bool existing = false;
int32_t size = taosArrayGetSize(pConsumer->currentTopics);
for (int32_t i = 0; i < size; i++) {
char *topic = taosArrayGetP(pConsumer->currentTopics, i);
if (strcmp(topic, pTopic) == 0) {
existing = true;
break;
}
}
return existing;
}
static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer, SMqConsumerObj *pNewConsumer) {
mDebug("consumer:0x%" PRIx64 " perform update action, update type:%d, subscribe-time:%" PRId64 ", uptime:%" PRId64,
pOldConsumer->consumerId, pNewConsumer->updateType, pOldConsumer->subscribeTime, pOldConsumer->upTime);
taosWLockLatch(&pOldConsumer->lock);
if (pNewConsumer->updateType == CONSUMER_UPDATE__MODIFY) {
SArray *tmp = pOldConsumer->rebNewTopics;
pOldConsumer->rebNewTopics = pNewConsumer->rebNewTopics;
pNewConsumer->rebNewTopics = tmp;
tmp = pOldConsumer->rebRemovedTopics;
pOldConsumer->rebRemovedTopics = pNewConsumer->rebRemovedTopics;
pNewConsumer->rebRemovedTopics = tmp;
tmp = pOldConsumer->assignedTopics;
pOldConsumer->assignedTopics = pNewConsumer->assignedTopics;
pNewConsumer->assignedTopics = tmp;
if (pNewConsumer->updateType == CONSUMER_UPDATE__REBALANCE) {
TSWAP(pOldConsumer->rebNewTopics, pNewConsumer->rebNewTopics);
TSWAP(pOldConsumer->rebRemovedTopics, pNewConsumer->rebRemovedTopics);
TSWAP(pOldConsumer->assignedTopics, pNewConsumer->assignedTopics);
pOldConsumer->subscribeTime = pNewConsumer->upTime;
pOldConsumer->status = MQ_CONSUMER_STATUS__MODIFY;
pOldConsumer->status = MQ_CONSUMER_STATUS_REBALANCE;
} else if (pNewConsumer->updateType == CONSUMER_UPDATE__LOST) {
int32_t sz = taosArrayGetSize(pOldConsumer->currentTopics);
for (int32_t i = 0; i < sz; i++) {
@ -937,10 +973,10 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
pOldConsumer->rebalanceTime = pNewConsumer->upTime;
int32_t status = pOldConsumer->status;
int32_t prevStatus = pOldConsumer->status;
pOldConsumer->status = MQ_CONSUMER_STATUS__LOST;
mDebug("consumer:0x%" PRIx64 " state %s -> %s, reb-time:%" PRId64 ", reb-removed-topics:%d",
pOldConsumer->consumerId, mndConsumerStatusName(status), mndConsumerStatusName(pOldConsumer->status),
pOldConsumer->consumerId, mndConsumerStatusName(prevStatus), mndConsumerStatusName(pOldConsumer->status),
pOldConsumer->rebalanceTime, (int)taosArrayGetSize(pOldConsumer->rebRemovedTopics));
} else if (pNewConsumer->updateType == CONSUMER_UPDATE__RECOVER) {
int32_t sz = taosArrayGetSize(pOldConsumer->assignedTopics);
@ -950,8 +986,7 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
}
pOldConsumer->rebalanceTime = pNewConsumer->upTime;
pOldConsumer->status = MQ_CONSUMER_STATUS__MODIFY;
pOldConsumer->status = MQ_CONSUMER_STATUS_REBALANCE;
} else if (pNewConsumer->updateType == CONSUMER_UPDATE__TOUCH) {
atomic_add_fetch_32(&pOldConsumer->epoch, 1);
@ -960,24 +995,16 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
} else if (pNewConsumer->updateType == CONSUMER_UPDATE__ADD) {
char *pNewTopic = taosStrdup(taosArrayGetP(pNewConsumer->rebNewTopics, 0));
// not exist in current topic
bool existing = false;
int32_t numOfExistedTopics = taosArrayGetSize(pOldConsumer->currentTopics);
for (int32_t i = 0; i < numOfExistedTopics; i++) {
char *topic = taosArrayGetP(pOldConsumer->currentTopics, i);
if (strcmp(topic, pNewTopic) == 0) {
existing = true;
}
}
// check if exist in current topic
removeFromNewTopicList(pOldConsumer, pNewTopic);
// add to current topic
if (!existing) {
bool existing = existInCurrentTopicList(pOldConsumer, pNewTopic);
if (existing) {
taosMemoryFree(pNewTopic);
} else { // added into current topic list
taosArrayPush(pOldConsumer->currentTopics, &pNewTopic);
taosArraySort(pOldConsumer->currentTopics, taosArrayCompareString);
} else {
taosMemoryFree(pNewTopic);
}
// set status
@ -1002,16 +1029,7 @@ static int32_t mndConsumerActionUpdate(SSdb *pSdb, SMqConsumerObj *pOldConsumer,
removeFromRemoveTopicList(pOldConsumer, removedTopic);
// remove from current topic
int32_t i = 0;
int32_t sz = taosArrayGetSize(pOldConsumer->currentTopics);
for (i = 0; i < sz; i++) {
char *topic = taosArrayGetP(pOldConsumer->currentTopics, i);
if (strcmp(removedTopic, topic) == 0) {
taosArrayRemove(pOldConsumer->currentTopics, i);
taosMemoryFree(topic);
break;
}
}
removeFromCurrentTopicList(pOldConsumer, removedTopic);
// set status
int32_t status = pOldConsumer->status;
@ -1160,7 +1178,7 @@ static const char *mndConsumerStatusName(int status) {
case MQ_CONSUMER_STATUS__LOST:
case MQ_CONSUMER_STATUS__LOST_REBD:
return "lost";
case MQ_CONSUMER_STATUS__MODIFY:
case MQ_CONSUMER_STATUS_REBALANCE:
return "rebalancing";
default:
return "unknown";

2
source/dnode/mnode/impl/src/mndDef.c
View File

@ -225,7 +225,7 @@ SMqConsumerObj *tNewSMqConsumerObj(int64_t consumerId, char cgroup[TSDB_CGROUP_L
memcpy(pConsumer->cgroup, cgroup, TSDB_CGROUP_LEN);
pConsumer->epoch = 0;
pConsumer->status = MQ_CONSUMER_STATUS__MODIFY;
pConsumer->status = MQ_CONSUMER_STATUS_REBALANCE;
pConsumer->hbStatus = 0;
taosInitRWLatch(&pConsumer->lock);

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

@ -124,11 +124,7 @@ static void mndCalMqRebalance(SMnode *pMnode) {
int32_t contLen = 0;
void *pReq = mndBuildTimerMsg(&contLen);
if (pReq != NULL) {
SRpcMsg rpcMsg = {
.msgType = TDMT_MND_TMQ_TIMER,
.pCont = pReq,
.contLen = contLen,
};
SRpcMsg rpcMsg = { .msgType = TDMT_MND_TMQ_TIMER, .pCont = pReq, .contLen = contLen };
tmsgPutToQueue(&pMnode->msgCb, READ_QUEUE, &rpcMsg);
}
}

23
source/dnode/mnode/impl/src/mndStb.c
View File

@ -797,6 +797,11 @@ int32_t mndBuildStbFromReq(SMnode *pMnode, SStbObj *pDst, SMCreateStbReq *pCreat
return -1;
}
if(pDst->nextColId < 0 || pDst->nextColId >= 0x7fff - pDst->numOfColumns - pDst->numOfTags){
terrno = TSDB_CODE_MND_FIELD_VALUE_OVERFLOW;
return -1;
}
for (int32_t i = 0; i < pDst->numOfColumns; ++i) {
SField *pField = taosArrayGet(pCreate->pColumns, i);
SSchema *pSchema = &pDst->pColumns[i];
@ -927,6 +932,11 @@ static int32_t mndBuildStbFromAlter(SStbObj *pStb, SStbObj *pDst, SMCreateStbReq
return -1;
}
if(pDst->nextColId < 0 && pDst->nextColId >= 0x7fff - pDst->numOfColumns - pDst->numOfTags){
terrno = TSDB_CODE_MND_FIELD_VALUE_OVERFLOW;
return -1;
}
for (int32_t i = 0; i < pDst->numOfColumns; ++i) {
SField *pField = taosArrayGet(createReq->pColumns, i);
SSchema *pSchema = &pDst->pColumns[i];
@ -1154,6 +1164,11 @@ static int32_t mndAddSuperTableTag(const SStbObj *pOld, SStbObj *pNew, SArray *p
return -1;
}
if(pNew->nextColId < 0 && pNew->nextColId >= 0x7fff - ntags){
terrno = TSDB_CODE_MND_FIELD_VALUE_OVERFLOW;
return -1;
}
for (int32_t i = 0; i < ntags; i++) {
SField *pField = taosArrayGet(pFields, i);
if (mndFindSuperTableColumnIndex(pOld, pField->name) >= 0) {
@ -1461,6 +1476,11 @@ static int32_t mndAddSuperTableColumn(const SStbObj *pOld, SStbObj *pNew, SArray
return -1;
}
if(pNew->nextColId < 0 && pNew->nextColId >= 0x7fff - ncols){
terrno = TSDB_CODE_MND_FIELD_VALUE_OVERFLOW;
return -1;
}
for (int32_t i = 0; i < ncols; i++) {
SField *pField = taosArrayGet(pFields, i);
if (mndFindSuperTableColumnIndex(pOld, pField->name) >= 0) {
@ -2524,6 +2544,9 @@ int32_t mndValidateStbInfo(SMnode *pMnode, SSTableVersion *pStbVersions, int32_t
if (mndBuildStbSchema(pMnode, pStbVersion->dbFName, pStbVersion->stbName, &metaRsp, &smaVer) != 0) {
metaRsp.numOfColumns = -1;
metaRsp.suid = pStbVersion->suid;
tstrncpy(metaRsp.dbFName, pStbVersion->dbFName, sizeof(metaRsp.dbFName));
tstrncpy(metaRsp.tbName, pStbVersion->stbName, sizeof(metaRsp.tbName));
tstrncpy(metaRsp.stbName, pStbVersion->stbName, sizeof(metaRsp.stbName));
taosArrayPush(hbRsp.pMetaRsp, &metaRsp);
continue;
}

64
source/dnode/mnode/impl/src/mndSubscribe.c
View File

@ -197,7 +197,7 @@ static SMqRebInfo *mndGetOrCreateRebSub(SHashObj *pHash, const char *key) {
return pRebSub;
}
static void doRemoveExistedConsumers(SMqRebOutputObj *pOutput, SHashObj *pHash, const SMqRebInputObj *pInput) {
static void doRemoveLostConsumers(SMqRebOutputObj *pOutput, SHashObj *pHash, const SMqRebInputObj *pInput) {
int32_t numOfRemoved = taosArrayGetSize(pInput->pRebInfo->removedConsumers);
const char *pSubKey = pOutput->pSub->key;
@ -213,13 +213,9 @@ static void doRemoveExistedConsumers(SMqRebOutputObj *pOutput, SHashObj *pHash,
int32_t consumerVgNum = taosArrayGetSize(pConsumerEp->vgs);
for (int32_t j = 0; j < consumerVgNum; j++) {
SMqVgEp *pVgEp = taosArrayGetP(pConsumerEp->vgs, j);
SMqRebOutputVg outputVg = {
.oldConsumerId = consumerId,
.newConsumerId = -1,
.pVgEp = pVgEp,
};
SMqVgEp *pVgEp = taosArrayGetP(pConsumerEp->vgs, j);
SMqRebOutputVg outputVg = {.oldConsumerId = consumerId, .newConsumerId = -1, .pVgEp = pVgEp};
taosHashPut(pHash, &pVgEp->vgId, sizeof(int32_t), &outputVg, sizeof(SMqRebOutputVg));
mInfo("sub:%s mq re-balance remove vgId:%d from consumer:%" PRIx64, pSubKey, pVgEp->vgId, consumerId);
}
@ -343,7 +339,7 @@ static int32_t mndDoRebalance(SMnode *pMnode, const SMqRebInputObj *pInput, SMqR
SHashObj *pHash = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_INT), false, HASH_NO_LOCK);
// 2. check and get actual removed consumers, put their vg into pHash
doRemoveExistedConsumers(pOutput, pHash, pInput);
doRemoveLostConsumers(pOutput, pHash, pInput);
// 3. if previously no consumer, there are vgs not assigned, put these vg into pHash
addUnassignedVgroups(pOutput, pHash);
@ -484,14 +480,16 @@ static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOu
for (int32_t i = 0; i < vgNum; i++) {
SMqRebOutputVg *pRebVg = taosArrayGet(rebVgs, i);
if (mndPersistSubChangeVgReq(pMnode, pTrans, pOutput->pSub, pRebVg) < 0) {
goto REB_FAIL;
mndTransDrop(pTrans);
return -1;
}
}
// 2. redo log: subscribe and vg assignment
// subscribe
if (mndSetSubCommitLogs(pMnode, pTrans, pOutput->pSub) != 0) {
goto REB_FAIL;
mndTransDrop(pTrans);
return -1;
}
// 3. commit log: consumer to update status and epoch
@ -506,11 +504,15 @@ static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOu
if (mndSetConsumerCommitLogs(pMnode, pTrans, pConsumerNew) != 0) {
tDeleteSMqConsumerObj(pConsumerNew);
taosMemoryFree(pConsumerNew);
goto REB_FAIL;
mndTransDrop(pTrans);
return -1;
}
tDeleteSMqConsumerObj(pConsumerNew);
taosMemoryFree(pConsumerNew);
}
// 3.2 set new consumer
consumerNum = taosArrayGetSize(pOutput->newConsumers);
for (int32_t i = 0; i < consumerNum; i++) {
@ -527,8 +529,11 @@ static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOu
if (mndSetConsumerCommitLogs(pMnode, pTrans, pConsumerNew) != 0) {
tDeleteSMqConsumerObj(pConsumerNew);
taosMemoryFree(pConsumerNew);
goto REB_FAIL;
mndTransDrop(pTrans);
return -1;
}
tDeleteSMqConsumerObj(pConsumerNew);
taosMemoryFree(pConsumerNew);
}
@ -549,8 +554,11 @@ static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOu
if (mndSetConsumerCommitLogs(pMnode, pTrans, pConsumerNew) != 0) {
tDeleteSMqConsumerObj(pConsumerNew);
taosMemoryFree(pConsumerNew);
goto REB_FAIL;
mndTransDrop(pTrans);
return -1;
}
tDeleteSMqConsumerObj(pConsumerNew);
taosMemoryFree(pConsumerNew);
}
@ -563,15 +571,12 @@ static int32_t mndPersistRebResult(SMnode *pMnode, SRpcMsg *pMsg, const SMqRebOu
// 6. execution
if (mndTransPrepare(pMnode, pTrans) != 0) {
mError("failed to prepare trans rebalance since %s", terrstr());
goto REB_FAIL;
mndTransDrop(pTrans);
return -1;
}
mndTransDrop(pTrans);
return 0;
REB_FAIL:
mndTransDrop(pTrans);
return -1;
}
static int32_t mndProcessRebalanceReq(SRpcMsg *pMsg) {
@ -584,16 +589,11 @@ static int32_t mndProcessRebalanceReq(SRpcMsg *pMsg) {
// here we only handle one topic rebalance requirement to ensure the atomic execution of this transaction.
while (1) {
// if (rebalanceOnce) {
// break;
// }
pIter = taosHashIterate(pReq->rebSubHash, pIter);
if (pIter == NULL) {
break;
}
// todo handle the malloc failure
SMqRebInputObj rebInput = {0};
SMqRebOutputObj rebOutput = {0};
rebOutput.newConsumers = taosArrayInit(0, sizeof(int64_t));
@ -601,6 +601,20 @@ static int32_t mndProcessRebalanceReq(SRpcMsg *pMsg) {
rebOutput.modifyConsumers = taosArrayInit(0, sizeof(int64_t));
rebOutput.rebVgs = taosArrayInit(0, sizeof(SMqRebOutputVg));
if (rebOutput.newConsumers == NULL || rebOutput.removedConsumers == NULL || rebOutput.modifyConsumers == NULL ||
rebOutput.rebVgs == NULL) {
taosArrayDestroy(rebOutput.newConsumers);
taosArrayDestroy(rebOutput.removedConsumers);
taosArrayDestroy(rebOutput.modifyConsumers);
taosArrayDestroy(rebOutput.rebVgs);
terrno = TSDB_CODE_OUT_OF_MEMORY;
mInfo("mq re-balance failed, due to out of memory");
taosHashCleanup(pReq->rebSubHash);
mndRebEnd();
return -1;
}
SMqRebInfo *pRebInfo = (SMqRebInfo *)pIter;
SMqSubscribeObj *pSub = mndAcquireSubscribeByKey(pMnode, pRebInfo->key);
@ -640,6 +654,7 @@ static int32_t mndProcessRebalanceReq(SRpcMsg *pMsg) {
rebInput.oldConsumerNum = taosHashGetSize(pSub->consumerHash);
rebOutput.pSub = tCloneSubscribeObj(pSub);
taosRUnLockLatch(&pSub->lock);
mInfo("sub topic:%s has %d consumers sub till now", pRebInfo->key, rebInput.oldConsumerNum);
mndReleaseSubscribe(pMnode, pSub);
}
@ -661,9 +676,6 @@ static int32_t mndProcessRebalanceReq(SRpcMsg *pMsg) {
taosArrayDestroy(rebOutput.rebVgs);
tDeleteSubscribeObj(rebOutput.pSub);
taosMemoryFree(rebOutput.pSub);
// taosSsleep(100);
// rebalanceOnce = true;
}
// reset flag

2
source/dnode/mnode/impl/src/mndUser.c
View File

@ -390,6 +390,7 @@ static SSdbRow *mndUserActionDecode(SSdbRaw *pRaw) {
SDB_GET_INT32(pRaw, dataPos, &ref, _OVER);
taosHashPut(pUser->useDbs, key, keyLen, &ref, sizeof(ref));
taosMemoryFree(key);
}
}
@ -956,6 +957,7 @@ _OVER:
mError("user:%s, failed to alter since %s", alterReq.user, terrstr());
}
tFreeSAlterUserReq(&alterReq);
mndReleaseUser(pMnode, pOperUser);
mndReleaseUser(pMnode, pUser);
mndUserFreeObj(&newUser);

2
source/dnode/mnode/impl/src/mndVgroup.c
View File

@ -2006,7 +2006,7 @@ static int32_t mndAddAdjustVnodeHashRangeAction(SMnode *pMnode, STrans *pTrans,
return 0;
}
static int32_t mndSplitVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SVgObj *pVgroup) {
int32_t mndSplitVgroup(SMnode *pMnode, SRpcMsg *pReq, SDbObj *pDb, SVgObj *pVgroup) {
int32_t code = -1;
STrans *pTrans = NULL;
SSdbRaw *pRaw = NULL;

4
source/dnode/snode/src/snode.c
View File

@ -153,11 +153,15 @@ int32_t sndProcessTaskDeployReq(SSnode *pSnode, char *msg, int32_t msgLen) {
ASSERT(pTask->taskLevel == TASK_LEVEL__AGG);
// 2.save task
taosWLockLatch(&pSnode->pMeta->lock);
code = streamMetaAddDeployedTask(pSnode->pMeta, -1, pTask);
if (code < 0) {
taosWUnLockLatch(&pSnode->pMeta->lock);
return -1;
}
taosWUnLockLatch(&pSnode->pMeta->lock);
// 3.go through recover steps to fill history
if (pTask->fillHistory) {
streamSetParamForRecover(pTask);

53
source/dnode/vnode/inc/vnode.h
View File

@ -162,7 +162,6 @@ int32_t metaTbCursorPrev(SMTbCursor *pTbCur, ETableType jumpTableType);
#endif
// tsdb
// typedef struct STsdb STsdb;
typedef struct STsdbReader STsdbReader;
#define TSDB_DEFAULT_STT_FILE 8
@ -176,11 +175,8 @@ typedef struct STsdbReader STsdbReader;
#define CACHESCAN_RETRIEVE_LAST_ROW 0x4
#define CACHESCAN_RETRIEVE_LAST 0x8
int32_t tsdbSetTableList(STsdbReader *pReader, const void *pTableList, int32_t num);
int32_t tsdbReaderOpen(SVnode *pVnode, SQueryTableDataCond *pCond, void *pTableList, int32_t numOfTables,
SSDataBlock *pResBlock, STsdbReader **ppReader, const char *idstr, bool countOnly);
void tsdbReaderSetId(STsdbReader* pReader, const char* idstr);
int32_t tsdbReaderOpen(SVnode *pVnode, SQueryTableDataCond *pCond, void *pTableList, int32_t numOfTables,
SSDataBlock *pResBlock, STsdbReader **ppReader, const char *idstr, bool countOnly);
void tsdbReaderClose(STsdbReader *pReader);
int32_t tsdbNextDataBlock(STsdbReader *pReader, bool *hasNext);
int32_t tsdbRetrieveDatablockSMA(STsdbReader *pReader, SSDataBlock *pDataBlock, bool *allHave);
@ -191,7 +187,10 @@ int32_t tsdbGetFileBlocksDistInfo(STsdbReader *pReader, STableBlockDistInfo
int64_t tsdbGetNumOfRowsInMemTable(STsdbReader *pHandle);
void *tsdbGetIdx(SMeta *pMeta);
void *tsdbGetIvtIdx(SMeta *pMeta);
uint64_t getReaderMaxVersion(STsdbReader *pReader);
uint64_t tsdbGetReaderMaxVersion(STsdbReader *pReader);
int32_t tsdbSetTableList(STsdbReader *pReader, const void *pTableList, int32_t num);
void tsdbReaderSetId(STsdbReader *pReader, const char *idstr);
void tsdbReaderSetCloseFlag(STsdbReader *pReader);
int32_t tsdbCacherowsReaderOpen(void *pVnode, int32_t type, void *pTableIdList, int32_t numOfTables, int32_t numOfCols,
uint64_t suid, void **pReader, const char *idstr);
@ -232,26 +231,22 @@ typedef struct SSnapContext {
} SSnapContext;
typedef struct STqReader {
SPackedData msg2;
SSubmitReq2 submit;
int32_t nextBlk;
int64_t lastBlkUid;
SWalReader *pWalReader;
SMeta *pVnodeMeta;
SHashObj *tbIdHash;
SArray *pColIdList; // SArray<int16_t>
SPackedData msg;
SSubmitReq2 submit;
int32_t nextBlk;
int64_t lastBlkUid;
SWalReader *pWalReader;
SMeta *pVnodeMeta;
SHashObj *tbIdHash;
SArray *pColIdList; // SArray<int16_t>
int32_t cachedSchemaVer;
int64_t cachedSchemaSuid;
int64_t cachedSchemaUid;
SSchemaWrapper *pSchemaWrapper;
STSchema *pSchema;
SSDataBlock *pResBlock;
} STqReader;
STqReader *tqOpenReader(SVnode *pVnode);
STqReader *tqReaderOpen(SVnode *pVnode);
void tqCloseReader(STqReader *);
void tqReaderSetColIdList(STqReader *pReader, SArray *pColIdList);
@ -260,17 +255,15 @@ int32_t tqReaderAddTbUidList(STqReader *pReader, const SArray *pTableUidList);
int32_t tqReaderRemoveTbUidList(STqReader *pReader, const SArray *tbUidList);
int32_t tqSeekVer(STqReader *pReader, int64_t ver, const char *id);
void tqNextBlock(STqReader *pReader, SFetchRet *ret);
int32_t tqNextBlock(STqReader *pReader, SSDataBlock* pBlock);
int32_t tqNextBlockInWal(STqReader* pReader);
int32_t extractSubmitMsgFromWal(SWalReader *pReader, SPackedData *pPackedData);
int32_t tqReaderSetSubmitMsg(STqReader *pReader, void *msgStr, int32_t msgLen, int64_t ver);
// int32_t tqReaderSetDataMsg(STqReader *pReader, const SSubmitReq *pMsg, int64_t ver);
bool tqNextDataBlock(STqReader *pReader);
bool tqNextDataBlockFilterOut2(STqReader *pReader, SHashObj *filterOutUids);
int32_t tqRetrieveDataBlock2(SSDataBlock *pBlock, STqReader *pReader, SSubmitTbData **pSubmitTbDataRet);
int32_t tqRetrieveTaosxBlock2(STqReader *pReader, SArray *blocks, SArray *schemas, SSubmitTbData **pSubmitTbDataRet);
// int32_t tqRetrieveDataBlock(SSDataBlock *pBlock, STqReader *pReader);
// int32_t tqRetrieveTaosxBlock(STqReader *pReader, SArray *blocks, SArray *schemas);
bool tqNextBlockImpl(STqReader *pReader);
bool tqNextDataBlockFilterOut(STqReader *pReader, SHashObj *filterOutUids);
int32_t tqRetrieveDataBlock(SSDataBlock *pBlock, STqReader *pReader, SSubmitTbData **pSubmitTbDataRet);
int32_t tqRetrieveTaosxBlock(STqReader *pReader, SArray *blocks, SArray *schemas, SSubmitTbData **pSubmitTbDataRet);
int32_t vnodeEnqueueStreamMsg(SVnode *pVnode, SRpcMsg *pMsg);

5
source/dnode/vnode/src/inc/tq.h
View File

@ -100,6 +100,7 @@ typedef struct {
SWalRef* pRef;
STqPushHandle pushHandle; // push
STqExecHandle execHandle; // exec
SRpcMsg* msg;
} STqHandle;
typedef struct {
@ -113,7 +114,7 @@ struct STQ {
char* path;
int64_t walLogLastVer;
SRWLatch lock;
SHashObj* pPushMgr; // consumerId -> STqPushEntry
SHashObj* pPushMgr; // subKey -> STqHandle
SHashObj* pHandle; // subKey -> STqHandle
SHashObj* pCheckInfo; // topic -> SAlterCheckInfo
STqOffsetStore* pOffsetStore;
@ -146,7 +147,7 @@ int32_t tqFetchLog(STQ* pTq, STqHandle* pHandle, int64_t* fetchOffset, SWalCkHea
int32_t tqTaosxScanLog(STQ* pTq, STqHandle* pHandle, SPackedData submit, STaosxRsp* pRsp, int32_t* totalRows);
int32_t tqAddBlockDataToRsp(const SSDataBlock* pBlock, SMqDataRsp* pRsp, int32_t numOfCols, int8_t precision);
int32_t tqSendDataRsp(STQ* pTq, const SRpcMsg* pMsg, const SMqPollReq* pReq, const SMqDataRsp* pRsp, int32_t type);
int32_t tqPushDataRsp(STQ* pTq, STqPushEntry* pPushEntry);
int32_t tqPushDataRsp(STQ* pTq, STqHandle* pHandle);
// tqMeta
int32_t tqMetaOpen(STQ* pTq);

42
source/dnode/vnode/src/inc/tsdb.h
View File

@ -16,6 +16,7 @@
#ifndef _TD_VNODE_TSDB_H_
#define _TD_VNODE_TSDB_H_
#include "tsimplehash.h"
#include "vnodeInt.h"
#ifdef __cplusplus
@ -122,14 +123,14 @@ int32_t tsdbRowCmprFn(const void *p1, const void *p2);
int32_t tsdbRowIterOpen(STSDBRowIter *pIter, TSDBROW *pRow, STSchema *pTSchema);
void tsdbRowClose(STSDBRowIter *pIter);
SColVal *tsdbRowIterNext(STSDBRowIter *pIter);
// SRowMerger
int32_t tsdbRowMergerInit(SRowMerger *pMerger, STSchema *pResTSchema, TSDBROW *pRow, STSchema *pTSchema);
int32_t tsdbRowMergerAdd(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema);
// int32_t tsdbRowMergerInit(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema);
void tsdbRowMergerClear(SRowMerger *pMerger);
// int32_t tsdbRowMerge(SRowMerger *pMerger, TSDBROW *pRow);
// SRowMerger
int32_t tsdbRowMergerInit(SRowMerger *pMerger, STSchema *pSchema);
int32_t tsdbRowMergerAdd(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema);
int32_t tsdbRowMergerGetRow(SRowMerger *pMerger, SRow **ppRow);
void tsdbRowMergerClear(SRowMerger *pMerger);
void tsdbRowMergerCleanup(SRowMerger *pMerger);
// TABLEID
int32_t tTABLEIDCmprFn(const void *p1, const void *p2);
// TSDBKEY
@ -224,7 +225,7 @@ int32_t tsdbTbDataIterCreate(STbData *pTbData, TSDBKEY *pFrom, int8_t backward,
void *tsdbTbDataIterDestroy(STbDataIter *pIter);
void tsdbTbDataIterOpen(STbData *pTbData, TSDBKEY *pFrom, int8_t backward, STbDataIter *pIter);
bool tsdbTbDataIterNext(STbDataIter *pIter);
void tsdbMemTableCountRows(SMemTable *pMemTable, SHashObj *pTableMap, int64_t *rowsNum);
void tsdbMemTableCountRows(SMemTable *pMemTable, SSHashObj *pTableMap, int64_t *rowsNum);
// STbData
int32_t tsdbGetNRowsInTbData(STbData *pTbData);
@ -322,8 +323,9 @@ int32_t tGnrtDiskData(SDiskDataBuilder *pBuilder, const SDiskData **ppDiskData,
#define TSDB_STT_FILE_DATA_ITER 2
#define TSDB_TOMB_FILE_DATA_ITER 3
#define TSDB_FILTER_FLAG_BY_VERSION 0x1
#define TSDB_FILTER_FLAG_BY_TABLEID 0x2
#define TSDB_FILTER_FLAG_BY_VERSION 0x1
#define TSDB_FILTER_FLAG_BY_TABLEID 0x2
#define TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE 0x4
#define TSDB_RBTN_TO_DATA_ITER(pNode) ((STsdbDataIter2 *)(((char *)pNode) - offsetof(STsdbDataIter2, rbtn)))
/* open */
@ -705,7 +707,6 @@ typedef struct SSttBlockLoadInfo {
typedef struct SMergeTree {
int8_t backward;
SRBTree rbt;
SArray *pIterList;
SLDataIter *pIter;
bool destroyLoadInfo;
SSttBlockLoadInfo *pLoadInfo;
@ -751,13 +752,29 @@ struct SDiskDataBuilder {
SBlkInfo bi;
};
typedef struct SLDataIter {
SRBTreeNode node;
SSttBlk *pSttBlk;
SDataFReader *pReader;
int32_t iStt;
int8_t backward;
int32_t iSttBlk;
int32_t iRow;
SRowInfo rInfo;
uint64_t uid;
STimeWindow timeWindow;
SVersionRange verRange;
SSttBlockLoadInfo *pBlockLoadInfo;
bool ignoreEarlierTs;
} SLDataIter;
#define tMergeTreeGetRow(_t) (&((_t)->pIter->rInfo.row))
int32_t tMergeTreeOpen(SMergeTree *pMTree, int8_t backward, SDataFReader *pFReader, uint64_t suid, uint64_t uid,
STimeWindow *pTimeWindow, SVersionRange *pVerRange, SSttBlockLoadInfo *pBlockLoadInfo,
bool destroyLoadInfo, const char *idStr, bool strictTimeRange);
bool destroyLoadInfo, const char *idStr, bool strictTimeRange, SLDataIter* pLDataIter);
void tMergeTreeAddIter(SMergeTree *pMTree, SLDataIter *pIter);
bool tMergeTreeNext(SMergeTree *pMTree);
bool tMergeTreeIgnoreEarlierTs(SMergeTree *pMTree);
TSDBROW tMergeTreeGetRow(SMergeTree *pMTree);
void tMergeTreeClose(SMergeTree *pMTree);
SSttBlockLoadInfo *tCreateLastBlockLoadInfo(STSchema *pSchema, int16_t *colList, int32_t numOfCols, int32_t numOfStt);
@ -782,6 +799,7 @@ typedef struct SCacheRowsReader {
STableKeyInfo *pTableList; // table id list
int32_t numOfTables;
SSttBlockLoadInfo *pLoadInfo;
SLDataIter *pDataIter;
STsdbReadSnap *pReadSnap;
SDataFReader *pDataFReader;
SDataFReader *pDataFReaderLast;

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

@ -193,9 +193,8 @@ STQ* tqOpen(const char* path, SVnode* pVnode);
void tqNotifyClose(STQ*);
void tqClose(STQ*);
int tqPushMsg(STQ*, void* msg, int32_t msgLen, tmsg_t msgType, int64_t ver);
int tqRegisterPushHandle(STQ* pTq, void* pHandle, const SMqPollReq* pRequest, SRpcMsg* pRpcMsg, SMqDataRsp* pDataRsp,
int32_t type);
int tqUnregisterPushHandle(STQ* pTq, const char* pKey, int32_t keyLen, uint64_t consumerId, bool rspConsumer);
int tqRegisterPushHandle(STQ* pTq, void* handle, SRpcMsg* pMsg);
int tqUnregisterPushHandle(STQ* pTq, void* pHandle);
int tqStartStreamTasks(STQ* pTq); // restore all stream tasks after vnode launching completed.
int tqCommit(STQ*);
@ -214,6 +213,7 @@ int32_t tqProcessTaskDropReq(STQ* pTq, int64_t version, char* msg, int32_t msgLe
int32_t tqProcessStreamTaskCheckReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessStreamTaskCheckRsp(STQ* pTq, int64_t version, char* msg, int32_t msgLen);
int32_t tqProcessSubmitReq(STQ* pTq, SPackedData submit);
int32_t tqProcessSubmitReqForSubscribe(STQ* pTq);
int32_t tqProcessDelReq(STQ* pTq, void* pReq, int32_t len, int64_t ver);
int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg);
int32_t tqProcessTaskDispatchReq(STQ* pTq, SRpcMsg* pMsg, bool exec);

1
source/dnode/vnode/src/meta/metaQuery.c
View File

@ -639,7 +639,6 @@ tb_uid_t metaStbCursorNext(SMStbCursor *pStbCur) {
STSchema *metaGetTbTSchema(SMeta *pMeta, tb_uid_t uid, int32_t sver, int lock) {
STSchema *pTSchema = NULL;
SSchemaWrapper *pSW = NULL;
SSchema *pSchema = NULL;
pSW = metaGetTableSchema(pMeta, uid, sver, lock);
if (!pSW) return NULL;

4
source/dnode/vnode/src/meta/metaSnapshot.c
View File

@ -217,8 +217,8 @@ typedef struct STableInfoForChildTable {
static void destroySTableInfoForChildTable(void* data) {
STableInfoForChildTable* pData = (STableInfoForChildTable*)data;
taosMemoryFree(pData->tableName);
tDeleteSSchemaWrapper(pData->schemaRow);
tDeleteSSchemaWrapper(pData->tagRow);
tDeleteSchemaWrapper(pData->schemaRow);
tDeleteSchemaWrapper(pData->tagRow);
}
static void MoveToSnapShotVersion(SSnapContext* ctx) {

7
source/dnode/vnode/src/meta/metaTable.c
View File

@ -673,8 +673,8 @@ int metaDropIndexFromSTable(SMeta *pMeta, int64_t version, SDropIndexReq *pReq)
metaUpdateUidIdx(pMeta, &nStbEntry);
metaULock(pMeta);
tDeleteSSchemaWrapper(tag);
tDeleteSSchemaWrapper(row);
tDeleteSchemaWrapper(tag);
tDeleteSchemaWrapper(row);
if (oStbEntry.pBuf) taosMemoryFree(oStbEntry.pBuf);
tDecoderClear(&dc);
@ -936,8 +936,7 @@ static int metaDropTableByUid(SMeta *pMeta, tb_uid_t uid, int *type) {
int tLen = 0;
if (tdbTbGet(pMeta->pUidIdx, &e.ctbEntry.suid, sizeof(tb_uid_t), &tData, &tLen) == 0) {
version = ((SUidIdxVal *)tData)[0].version;
STbDbKey tbDbKey = {.uid = e.ctbEntry.suid, .version = version};
STbDbKey tbDbKey = {.uid = e.ctbEntry.suid, .version = ((SUidIdxVal *)tData)[0].version};
if (tdbTbGet(pMeta->pTbDb, &tbDbKey, sizeof(tbDbKey), &tData, &tLen) == 0) {
SDecoder tdc = {0};
SMetaEntry stbEntry = {0};

4
source/dnode/vnode/src/sma/smaRollup.c
View File

@ -684,7 +684,7 @@ static int32_t tdRSmaExecAndSubmitResult(SSma *pSma, qTaskInfo_t taskInfo, SRSma
}
if (pReq && tdProcessSubmitReq(sinkTsdb, output->info.version, pReq) < 0) {
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFree(pReq);
smaError("vgId:%d, process submit req for rsma suid:%" PRIu64 ", uid:%" PRIu64 " level %" PRIi8
" failed since %s",
@ -696,7 +696,7 @@ static int32_t tdRSmaExecAndSubmitResult(SSma *pSma, qTaskInfo_t taskInfo, SRSma
SMA_VID(pSma), suid, output->info.id.groupId, pItem->level, output->info.version);
if (pReq) {
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFree(pReq);
}
}

8
source/dnode/vnode/src/sma/smaTimeRange.c
View File

@ -299,7 +299,7 @@ int32_t smaBlockToSubmit(SVnode *pVnode, const SArray *pBlocks, const STSchema *
}
SRow *pRow = NULL;
if ((terrno = tRowBuild(pVals, (STSchema *)pTSchema, &pRow)) < 0) {
tDestroySSubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
goto _end;
}
taosArrayPush(tbData.aRowP, &pRow);
@ -309,7 +309,7 @@ int32_t smaBlockToSubmit(SVnode *pVnode, const SArray *pBlocks, const STSchema *
}
// encode
tEncodeSize(tEncodeSSubmitReq2, pReq, len, terrno);
tEncodeSize(tEncodeSubmitReq, pReq, len, terrno);
if (TSDB_CODE_SUCCESS == terrno) {
SEncoder encoder;
len += sizeof(SSubmitReq2Msg);
@ -321,7 +321,7 @@ int32_t smaBlockToSubmit(SVnode *pVnode, const SArray *pBlocks, const STSchema *
((SSubmitReq2Msg *)pBuf)->header.contLen = htonl(len);
((SSubmitReq2Msg *)pBuf)->version = htobe64(1);
tEncoderInit(&encoder, POINTER_SHIFT(pBuf, sizeof(SSubmitReq2Msg)), len - sizeof(SSubmitReq2Msg));
if (tEncodeSSubmitReq2(&encoder, pReq) < 0) {
if (tEncodeSubmitReq(&encoder, pReq) < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
/*vError("failed to encode submit req since %s", terrstr());*/
}
@ -332,7 +332,7 @@ _end:
taosArrayDestroy(tagArray);
taosArrayDestroy(pVals);
if (pReq) {
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFree(pReq);
}

144
source/dnode/vnode/src/tq/tq.c
View File

@ -71,18 +71,11 @@ static void destroyTqHandle(void* data) {
walCloseReader(pData->pWalReader);
tqCloseReader(pData->execHandle.pTqReader);
}
}
static void tqPushEntryFree(void* data) {
STqPushEntry* p = *(void**)data;
if (p->pDataRsp->head.mqMsgType == TMQ_MSG_TYPE__POLL_RSP) {
tDeleteSMqDataRsp(p->pDataRsp);
} else if (p->pDataRsp->head.mqMsgType == TMQ_MSG_TYPE__TAOSX_RSP) {
tDeleteSTaosxRsp((STaosxRsp*)p->pDataRsp);
if(pData->msg != NULL) {
rpcFreeCont(pData->msg->pCont);
taosMemoryFree(pData->msg);
pData->msg = NULL;
}
taosMemoryFree(p->pDataRsp);
taosMemoryFree(p);
}
static bool tqOffsetLessOrEqual(const STqOffset* pLeft, const STqOffset* pRight) {
@ -105,14 +98,18 @@ STQ* tqOpen(const char* path, SVnode* pVnode) {
taosHashSetFreeFp(pTq->pHandle, destroyTqHandle);
taosInitRWLatch(&pTq->lock);
pTq->pPushMgr = taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), true, HASH_NO_LOCK);
taosHashSetFreeFp(pTq->pPushMgr, tqPushEntryFree);
pTq->pPushMgr = taosHashInit(64, MurmurHash3_32, false, HASH_NO_LOCK);
pTq->pCheckInfo = taosHashInit(64, MurmurHash3_32, true, HASH_ENTRY_LOCK);
taosHashSetFreeFp(pTq->pCheckInfo, (FDelete)tDeleteSTqCheckInfo);
tqInitialize(pTq);
return pTq;
int32_t code = tqInitialize(pTq);
if (code != TSDB_CODE_SUCCESS) {
tqClose(pTq);
return NULL;
} else {
return pTq;
}
}
int32_t tqInitialize(STQ* pTq) {
@ -228,17 +225,19 @@ static int32_t doSendDataRsp(const SRpcHandleInfo* pRpcHandleInfo, const SMqData
return 0;
}
int32_t tqPushDataRsp(STQ* pTq, STqPushEntry* pPushEntry) {
SMqDataRsp* pRsp = pPushEntry->pDataRsp;
SMqRspHead* pHeader = &pPushEntry->pDataRsp->head;
doSendDataRsp(&pPushEntry->info, pRsp, pHeader->epoch, pHeader->consumerId, pHeader->mqMsgType);
int32_t tqPushDataRsp(STQ* pTq, STqHandle* pHandle) {
SMqDataRsp dataRsp = {0};
dataRsp.head.consumerId = pHandle->consumerId;
dataRsp.head.epoch = pHandle->epoch;
dataRsp.head.mqMsgType = TMQ_MSG_TYPE__POLL_RSP;
doSendDataRsp(&pHandle->msg->info, &dataRsp, pHandle->epoch, pHandle->consumerId, TMQ_MSG_TYPE__POLL_RSP);
char buf1[80] = {0};
char buf2[80] = {0};
tFormatOffset(buf1, tListLen(buf1), &pRsp->reqOffset);
tFormatOffset(buf2, tListLen(buf2), &pRsp->rspOffset);
tFormatOffset(buf1, tListLen(buf1), &dataRsp.reqOffset);
tFormatOffset(buf2, tListLen(buf2), &dataRsp.rspOffset);
tqDebug("vgId:%d, from consumer:0x%" PRIx64 " (epoch %d) push rsp, block num: %d, req:%s, rsp:%s",
TD_VID(pTq->pVnode), pRsp->head.consumerId, pRsp->head.epoch, pRsp->blockNum, buf1, buf2);
TD_VID(pTq->pVnode), dataRsp.head.consumerId, dataRsp.head.epoch, dataRsp.blockNum, buf1, buf2);
return 0;
}
@ -382,13 +381,13 @@ int32_t tqProcessDeleteSubReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
SMqVDeleteReq* pReq = (SMqVDeleteReq*)msg;
tqDebug("vgId:%d, tq process delete sub req %s", pTq->pVnode->config.vgId, pReq->subKey);
taosWLockLatch(&pTq->lock);
int32_t code = taosHashRemove(pTq->pPushMgr, pReq->subKey, strlen(pReq->subKey));
if (code != 0) {
tqDebug("vgId:%d, tq remove push handle %s", pTq->pVnode->config.vgId, pReq->subKey);
}
taosWUnLockLatch(&pTq->lock);
int32_t code = 0;
// taosWLockLatch(&pTq->lock);
// int32_t code = taosHashRemove(pTq->pPushMgr, pReq->subKey, strlen(pReq->subKey));
// if (code != 0) {
// tqDebug("vgId:%d, tq remove push handle %s", pTq->pVnode->config.vgId, pReq->subKey);
// }
// taosWUnLockLatch(&pTq->lock);
STqHandle* pHandle = taosHashGet(pTq->pHandle, pReq->subKey, strlen(pReq->subKey));
if (pHandle) {
@ -504,7 +503,7 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
pHandle->execHandle.pTqReader = qExtractReaderFromStreamScanner(scanner);
} else if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__DB) {
pHandle->pWalReader = walOpenReader(pVnode->pWal, NULL);
pHandle->execHandle.pTqReader = tqOpenReader(pVnode);
pHandle->execHandle.pTqReader = tqReaderOpen(pVnode);
pHandle->execHandle.execDb.pFilterOutTbUid =
taosHashInit(64, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BIGINT), false, HASH_NO_LOCK);
@ -523,7 +522,7 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
int64_t tbUid = *(int64_t*)taosArrayGet(tbUidList, i);
tqDebug("vgId:%d, idx %d, uid:%" PRId64, vgId, i, tbUid);
}
pHandle->execHandle.pTqReader = tqOpenReader(pVnode);
pHandle->execHandle.pTqReader = tqReaderOpen(pVnode);
tqReaderSetTbUidList(pHandle->execHandle.pTqReader, tbUidList);
taosArrayDestroy(tbUidList);
@ -557,13 +556,12 @@ int32_t tqProcessSubscribeReq(STQ* pTq, int64_t sversion, char* msg, int32_t msg
}
taosWLockLatch(&pTq->lock);
atomic_store_32(&pHandle->epoch, -1);
atomic_store_32(&pHandle->epoch, 0);
// remove if it has been register in the push manager, and return one empty block to consumer
tqUnregisterPushHandle(pTq, req.subKey, (int32_t)strlen(req.subKey), pHandle->consumerId, true);
tqUnregisterPushHandle(pTq, pHandle);
atomic_store_64(&pHandle->consumerId, req.newConsumerId);
atomic_add_fetch_32(&pHandle->epoch, 1);
if (pHandle->execHandle.subType == TOPIC_SUB_TYPE__COLUMN) {
qStreamCloseTsdbReader(pTaskInfo);
@ -601,11 +599,7 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t ver) {
pTask->chkInfo.currentVer = ver;
// expand executor
if (pTask->fillHistory) {
pTask->status.taskStatus = TASK_STATUS__WAIT_DOWNSTREAM;
} else {
pTask->status.taskStatus = TASK_STATUS__RESTORE;
}
pTask->status.taskStatus = (pTask->fillHistory)? TASK_STATUS__WAIT_DOWNSTREAM:TASK_STATUS__NORMAL;
if (pTask->taskLevel == TASK_LEVEL__SOURCE) {
pTask->pState = streamStateOpen(pTq->pStreamMeta->path, pTask, false, -1, -1);
@ -664,6 +658,7 @@ int32_t tqExpandTask(STQ* pTq, SStreamTask* pTask, int64_t ver) {
}
streamSetupTrigger(pTask);
tqInfo("vgId:%d expand stream task, s-task:%s, checkpoint ver:%" PRId64 " child id:%d, level:%d", vgId, pTask->id.idStr,
pTask->chkInfo.version, pTask->selfChildId, pTask->taskLevel);
@ -693,8 +688,9 @@ int32_t tqProcessStreamTaskCheckReq(STQ* pTq, SRpcMsg* pMsg) {
};
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, taskId);
if (pTask) {
rsp.status = (atomic_load_8(&pTask->status.taskStatus) == TASK_STATUS__NORMAL) ? 1 : 0;
rsp.status = streamTaskCheckStatus(pTask);
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
tqDebug("tq recv task check req(reqId:0x%" PRIx64
@ -785,13 +781,17 @@ int32_t tqProcessTaskDeployReq(STQ* pTq, int64_t sversion, char* msg, int32_t ms
tDecoderClear(&decoder);
// 2.save task, use the newest commit version as the initial start version of stream task.
taosWLockLatch(&pTq->pStreamMeta->lock);
code = streamMetaAddDeployedTask(pTq->pStreamMeta, sversion, pTask);
if (code < 0) {
tqError("vgId:%d failed to add s-task:%s, total:%d", TD_VID(pTq->pVnode), pTask->id.idStr,
streamMetaGetNumOfTasks(pTq->pStreamMeta));
taosWUnLockLatch(&pTq->pStreamMeta->lock);
return -1;
}
taosWUnLockLatch(&pTq->pStreamMeta->lock);
// 3.go through recover steps to fill history
if (pTask->fillHistory) {
streamTaskCheckDownstream(pTask, sversion);
@ -821,13 +821,18 @@ int32_t tqProcessTaskRecover1Req(STQ* pTq, SRpcMsg* pMsg) {
}
// do recovery step 1
streamSourceRecoverScanStep1(pTask);
tqDebug("s-task:%s start recover step 1 scan", pTask->id.idStr);
int64_t st = taosGetTimestampMs();
streamSourceRecoverScanStep1(pTask);
if (atomic_load_8(&pTask->status.taskStatus) == TASK_STATUS__DROPPING) {
streamMetaReleaseTask(pTq->pStreamMeta, pTask);
return 0;
}
double el = (taosGetTimestampMs() - st) / 1000.0;
tqDebug("s-task:%s recover step 1 ended, elapsed time:%.2fs", pTask->id.idStr, el);
// build msg to launch next step
SStreamRecoverStep2Req req;
code = streamBuildSourceRecover2Req(pTask, &req);
@ -853,20 +858,17 @@ int32_t tqProcessTaskRecover1Req(STQ* pTq, SRpcMsg* pMsg) {
memcpy(serializedReq, &req, len);
// dispatch msg
tqDebug("s-task:%s start recover block stage", pTask->id.idStr);
SRpcMsg rpcMsg = {
.code = 0,
.contLen = len,
.msgType = TDMT_VND_STREAM_RECOVER_BLOCKING_STAGE,
.pCont = serializedReq,
};
.code = 0, .contLen = len, .msgType = TDMT_VND_STREAM_RECOVER_BLOCKING_STAGE, .pCont = serializedReq};
tmsgPutToQueue(&pTq->pVnode->msgCb, WRITE_QUEUE, &rpcMsg);
return 0;
}
int32_t tqProcessTaskRecover2Req(STQ* pTq, int64_t sversion, char* msg, int32_t msgLen) {
int32_t code;
int32_t code = 0;
SStreamRecoverStep2Req* pReq = (SStreamRecoverStep2Req*)msg;
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, pReq->taskId);
if (pTask == NULL) {
@ -1067,6 +1069,34 @@ int32_t tqProcessDelReq(STQ* pTq, void* pReq, int32_t len, int64_t ver) {
return 0;
}
int32_t tqProcessSubmitReqForSubscribe(STQ* pTq) {
int32_t vgId = TD_VID(pTq->pVnode);
taosWLockLatch(&pTq->lock);
if(taosHashGetSize(pTq->pPushMgr) > 0){
void *pIter = taosHashIterate(pTq->pPushMgr, NULL);
while(pIter){
STqHandle* pHandle = *(STqHandle**)pIter;
tqDebug("vgId:%d start set submit for pHandle:%p, consume id:0x%"PRIx64, vgId, pHandle, pHandle->consumerId);
if(ASSERT(pHandle->msg != NULL)){
tqError("pHandle->msg should not be null");
break;
}else{
SRpcMsg msg = {.msgType = TDMT_VND_TMQ_CONSUME, .pCont = pHandle->msg->pCont, .contLen = pHandle->msg->contLen, .info = pHandle->msg->info};
tmsgPutToQueue(&pTq->pVnode->msgCb, QUERY_QUEUE, &msg);
taosMemoryFree(pHandle->msg);
pHandle->msg = NULL;
}
pIter = taosHashIterate(pTq->pPushMgr, pIter);
}
taosHashClear(pTq->pPushMgr);
}
// unlock
taosWUnLockLatch(&pTq->lock);
return 0;
}
int32_t tqProcessSubmitReq(STQ* pTq, SPackedData submit) {
#if 0
void* pIter = NULL;
@ -1145,9 +1175,6 @@ int32_t tqProcessTaskRunReq(STQ* pTq, SRpcMsg* pMsg) {
SStreamTask* pTask = streamMetaAcquireTask(pTq->pStreamMeta, taskId);
if (pTask != NULL) {
if (pTask->status.taskStatus == TASK_STATUS__NORMAL) {
tqDebug("vgId:%d s-task:%s start to process run req", vgId, pTask->id.idStr);
streamProcessRunReq(pTask);
} else if (pTask->status.taskStatus == TASK_STATUS__RESTORE) {
tqDebug("vgId:%d s-task:%s start to process block from wal, last chk point:%" PRId64, vgId,
pTask->id.idStr, pTask->chkInfo.version);
streamProcessRunReq(pTask);
@ -1300,21 +1327,22 @@ FAIL:
int32_t tqCheckLogInWal(STQ* pTq, int64_t sversion) { return sversion <= pTq->walLogLastVer; }
int32_t tqStartStreamTasks(STQ* pTq) {
int32_t vgId = TD_VID(pTq->pVnode);
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
taosWLockLatch(&pMeta->lock);
int32_t numOfTasks = taosHashGetSize(pTq->pStreamMeta->pTasks);
int32_t numOfTasks = taosArrayGetSize(pMeta->pTaskList);
if (numOfTasks == 0) {
tqInfo("vgId:%d no stream tasks exists", vgId);
taosWUnLockLatch(&pTq->pStreamMeta->lock);
return 0;
}
pMeta->walScan += 1;
pMeta->walScanCounter += 1;
if (pMeta->walScan > 1) {
tqDebug("vgId:%d wal read task has been launched, remain scan times:%d", vgId, pMeta->walScan);
if (pMeta->walScanCounter > 1) {
tqDebug("vgId:%d wal read task has been launched, remain scan times:%d", vgId, pMeta->walScanCounter);
taosWUnLockLatch(&pTq->pStreamMeta->lock);
return 0;
}

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

@ -328,7 +328,7 @@ int32_t tqMetaRestoreHandle(STQ* pTq) {
}
} else if (handle.execHandle.subType == TOPIC_SUB_TYPE__DB) {
handle.pWalReader = walOpenReader(pTq->pVnode->pWal, NULL);
handle.execHandle.pTqReader = tqOpenReader(pTq->pVnode);
handle.execHandle.pTqReader = tqReaderOpen(pTq->pVnode);
buildSnapContext(reader.meta, reader.version, 0, handle.execHandle.subType, handle.fetchMeta,
(SSnapContext**)(&reader.sContext));
@ -343,7 +343,7 @@ int32_t tqMetaRestoreHandle(STQ* pTq) {
int64_t tbUid = *(int64_t*)taosArrayGet(tbUidList, i);
tqDebug("vgId:%d, idx %d, uid:%" PRId64, vgId, i, tbUid);
}
handle.execHandle.pTqReader = tqOpenReader(pTq->pVnode);
handle.execHandle.pTqReader = tqReaderOpen(pTq->pVnode);
tqReaderSetTbUidList(handle.execHandle.pTqReader, tbUidList);
taosArrayDestroy(tbUidList);

103
source/dnode/vnode/src/tq/tqOffset.c
View File

@ -31,57 +31,67 @@ char* tqOffsetBuildFName(const char* path, int32_t fVer) {
int32_t tqOffsetRestoreFromFile(STqOffsetStore* pStore, const char* fname) {
TdFilePtr pFile = taosOpenFile(fname, TD_FILE_READ);
if (pFile != NULL) {
STqOffsetHead head = {0};
int64_t code;
if (pFile == NULL) {
return TSDB_CODE_SUCCESS;
}
while (1) {
if ((code = taosReadFile(pFile, &head, sizeof(STqOffsetHead))) != sizeof(STqOffsetHead)) {
if (code == 0) {
break;
} else {
return -1;
}
}
int32_t size = htonl(head.size);
void* memBuf = taosMemoryCalloc(1, size);
if (memBuf == NULL) {
int32_t vgId = TD_VID(pStore->pTq->pVnode);
int64_t code = 0;
STqOffsetHead head = {0};
while (1) {
if ((code = taosReadFile(pFile, &head, sizeof(STqOffsetHead))) != sizeof(STqOffsetHead)) {
if (code == 0) {
break;
} else {
return -1;
}
if ((code = taosReadFile(pFile, memBuf, size)) != size) {
taosMemoryFree(memBuf);
return -1;
}
STqOffset offset;
SDecoder decoder;
tDecoderInit(&decoder, memBuf, size);
if (tDecodeSTqOffset(&decoder, &offset) < 0) {
taosMemoryFree(memBuf);
tDecoderClear(&decoder);
return -1;
}
tDecoderClear(&decoder);
if (taosHashPut(pStore->pHash, offset.subKey, strlen(offset.subKey), &offset, sizeof(STqOffset)) < 0) {
return -1;
}
if (offset.val.type == TMQ_OFFSET__LOG) {
STqHandle* pHandle = taosHashGet(pStore->pTq->pHandle, offset.subKey, strlen(offset.subKey));
if (pHandle) {
if (walRefVer(pHandle->pRef, offset.val.version) < 0) {
tqError("vgId: %d, tq handle %s ref ver %" PRId64 "error", pStore->pTq->pVnode->config.vgId,
pHandle->subKey, offset.val.version);
}
}
}
taosMemoryFree(memBuf);
}
taosCloseFile(&pFile);
int32_t size = htonl(head.size);
void* pMemBuf = taosMemoryCalloc(1, size);
if (pMemBuf == NULL) {
tqError("vgId:%d failed to restore offset from file, since out of memory, malloc size:%d", vgId, size);
terrno = TSDB_CODE_OUT_OF_MEMORY;
return -1;
}
if ((code = taosReadFile(pFile, pMemBuf, size)) != size) {
taosMemoryFree(pMemBuf);
return -1;
}
STqOffset offset;
SDecoder decoder;
tDecoderInit(&decoder, pMemBuf, size);
if (tDecodeSTqOffset(&decoder, &offset) < 0) {
taosMemoryFree(pMemBuf);
tDecoderClear(&decoder);
return code;
}
tDecoderClear(&decoder);
if (taosHashPut(pStore->pHash, offset.subKey, strlen(offset.subKey), &offset, sizeof(STqOffset)) < 0) {
return -1;
}
// todo remove this
if (offset.val.type == TMQ_OFFSET__LOG) {
STqHandle* pHandle = taosHashGet(pStore->pTq->pHandle, offset.subKey, strlen(offset.subKey));
if (pHandle) {
if (walRefVer(pHandle->pRef, offset.val.version) < 0) {
// tqError("vgId: %d, tq handle %s ref ver %" PRId64 "error", pStore->pTq->pVnode->config.vgId, pHandle->subKey,
// offset.val.version);
}
}
}
taosMemoryFree(pMemBuf);
}
return 0;
taosCloseFile(&pFile);
return TSDB_CODE_SUCCESS;
}
STqOffsetStore* tqOffsetOpen(STQ* pTq) {
@ -89,6 +99,7 @@ STqOffsetStore* tqOffsetOpen(STQ* pTq) {
if (pStore == NULL) {
return NULL;
}
pStore->pTq = pTq;
pStore->needCommit = 0;
pTq->pOffsetStore = pStore;
@ -98,12 +109,14 @@ STqOffsetStore* tqOffsetOpen(STQ* pTq) {
taosMemoryFree(pStore);
return NULL;
}
char* fname = tqOffsetBuildFName(pStore->pTq->path, 0);
if (tqOffsetRestoreFromFile(pStore, fname) < 0) {
taosMemoryFree(fname);
taosMemoryFree(pStore);
return NULL;
}
taosMemoryFree(fname);
return pStore;
}

275
source/dnode/vnode/src/tq/tqPush.c
View File

@ -206,150 +206,75 @@ int32_t tqPushMsgNew(STQ* pTq, void* msg, int32_t msgLen, tmsg_t msgType, int64_
}
#endif
typedef struct {
void* pKey;
int64_t keyLen;
} SItem;
static void recordPushedEntry(SArray* cachedKey, void* pIter);
static void doRemovePushedEntry(SArray* pCachedKeys, STQ* pTq);
static void freeItem(void* param) {
SItem* p = (SItem*)param;
taosMemoryFree(p->pKey);
}
static void doPushDataForEntry(void* pIter, STqExecHandle* pExec, STQ* pTq, int64_t ver, int32_t vgId, char* pData,
int32_t dataLen, SArray* pCachedKey) {
STqPushEntry* pPushEntry = *(STqPushEntry**)pIter;
SMqDataRsp* pRsp = pPushEntry->pDataRsp;
if (pRsp->reqOffset.version >= ver) {
tqDebug("vgId:%d, push entry req version %" PRId64 ", while push version %" PRId64 ", skip", vgId,
pRsp->reqOffset.version, ver);
return;
}
qTaskInfo_t pTaskInfo = pExec->task;
// prepare scan mem data
SPackedData submit = {.msgStr = pData, .msgLen = dataLen, .ver = ver};
if (qStreamSetScanMemData(pTaskInfo, submit) != 0) {
return;
}
qStreamSetOpen(pTaskInfo);
// here start to scan submit block to extract the subscribed data
int32_t totalRows = 0;
while (1) {
SSDataBlock* pDataBlock = NULL;
uint64_t ts = 0;
if (qExecTask(pTaskInfo, &pDataBlock, &ts) < 0) {
tqDebug("vgId:%d, tq exec error since %s", vgId, terrstr());
}
if (pDataBlock == NULL) {
break;
}
tqAddBlockDataToRsp(pDataBlock, pRsp, pExec->numOfCols, pTq->pVnode->config.tsdbCfg.precision);
pRsp->blockNum++;
totalRows += pDataBlock->info.rows;
}
tqDebug("vgId:%d, tq handle push, subkey:%s, block num:%d, rows:%d", vgId, pPushEntry->subKey, pRsp->blockNum,
totalRows);
if (pRsp->blockNum > 0) {
tqOffsetResetToLog(&pRsp->rspOffset, ver);
tqPushDataRsp(pTq, pPushEntry);
recordPushedEntry(pCachedKey, pIter);
}
}
int32_t tqPushMsg(STQ* pTq, void* msg, int32_t msgLen, tmsg_t msgType, int64_t ver) {
void* pReq = POINTER_SHIFT(msg, sizeof(SSubmitReq2Msg));
int32_t len = msgLen - sizeof(SSubmitReq2Msg);
int32_t vgId = TD_VID(pTq->pVnode);
// void* pReq = POINTER_SHIFT(msg, sizeof(SSubmitReq2Msg));
// int32_t len = msgLen - sizeof(SSubmitReq2Msg);
// int32_t vgId = TD_VID(pTq->pVnode);
if (msgType == TDMT_VND_SUBMIT) {
tqProcessSubmitReqForSubscribe(pTq);
// lock push mgr to avoid potential msg lost
taosWLockLatch(&pTq->lock);
int32_t numOfRegisteredPush = taosHashGetSize(pTq->pPushMgr);
if (numOfRegisteredPush > 0) {
tqDebug("vgId:%d tq push msg version:%" PRId64 " type:%s, head:%p, body:%p len:%d, numOfPushed consumers:%d",
vgId, ver, TMSG_INFO(msgType), msg, pReq, len, numOfRegisteredPush);
void* data = taosMemoryMalloc(len);
if (data == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("failed to copy data for stream since out of memory, vgId:%d", vgId);
taosWUnLockLatch(&pTq->lock);
return -1;
}
memcpy(data, pReq, len);
SArray* cachedKey = taosArrayInit(0, sizeof(SItem));
void* pIter = NULL;
while (1) {
pIter = taosHashIterate(pTq->pPushMgr, pIter);
if (pIter == NULL) {
break;
}
STqPushEntry* pPushEntry = *(STqPushEntry**)pIter;
STqHandle* pHandle = taosHashGet(pTq->pHandle, pPushEntry->subKey, strlen(pPushEntry->subKey));
if (pHandle == NULL) {
tqDebug("vgId:%d, failed to find handle %s in pushing data to consumer, ignore", pTq->pVnode->config.vgId,
pPushEntry->subKey);
continue;
}
STqExecHandle* pExec = &pHandle->execHandle;
doPushDataForEntry(pIter, pExec, pTq, ver, vgId, data, len, cachedKey);
}
doRemovePushedEntry(cachedKey, pTq);
taosArrayDestroyEx(cachedKey, freeItem);
taosMemoryFree(data);
}
// unlock
taosWUnLockLatch(&pTq->lock);
// taosWLockLatch(&pTq->lock);
//
// int32_t numOfRegisteredPush = taosHashGetSize(pTq->pPushMgr);
// if (numOfRegisteredPush > 0) {
// tqDebug("vgId:%d tq push msg version:%" PRId64 " type:%s, head:%p, body:%p len:%d, numOfPushed consumers:%d",
// vgId, ver, TMSG_INFO(msgType), msg, pReq, len, numOfRegisteredPush);
//
// void* data = taosMemoryMalloc(len);
// if (data == NULL) {
// terrno = TSDB_CODE_OUT_OF_MEMORY;
// tqError("failed to copy data for stream since out of memory, vgId:%d", vgId);
// taosWUnLockLatch(&pTq->lock);
// return -1;
// }
//
// memcpy(data, pReq, len);
//
// SArray* cachedKey = taosArrayInit(0, sizeof(SItem));
// void* pIter = NULL;
//
// while (1) {
// pIter = taosHashIterate(pTq->pPushMgr, pIter);
// if (pIter == NULL) {
// break;
// }
//
// STqPushEntry* pPushEntry = *(STqPushEntry**)pIter;
//
// STqHandle* pHandle = taosHashGet(pTq->pHandle, pPushEntry->subKey, strlen(pPushEntry->subKey));
// if (pHandle == NULL) {
// tqDebug("vgId:%d, failed to find handle %s in pushing data to consumer, ignore", pTq->pVnode->config.vgId,
// pPushEntry->subKey);
// continue;
// }
//
// STqExecHandle* pExec = &pHandle->execHandle;
// doPushDataForEntry(pIter, pExec, pTq, ver, vgId, data, len, cachedKey);
// }
//
// doRemovePushedEntry(cachedKey, pTq);
// taosArrayDestroyEx(cachedKey, freeItem);
// taosMemoryFree(data);
// }
//
// // unlock
// taosWUnLockLatch(&pTq->lock);
}
tqDebug("handle submit, restore:%d, size:%d", pTq->pVnode->restored, (int)taosHashGetSize(pTq->pStreamMeta->pTasks));
int32_t numOfTasks = streamMetaGetNumOfTasks(pTq->pStreamMeta);
tqDebug("handle submit, restore:%d, size:%d", pTq->pVnode->restored, numOfTasks);
// push data for stream processing:
// 1. the vnode has already been restored.
// 2. the vnode should be the leader.
// 3. the stream is not suspended yet.
if (!tsDisableStream && vnodeIsRoleLeader(pTq->pVnode) && pTq->pVnode->restored) {
if (taosHashGetSize(pTq->pStreamMeta->pTasks) == 0) {
if (numOfTasks == 0) {
return 0;
}
if (msgType == TDMT_VND_SUBMIT) {
#if 0
void* data = taosMemoryMalloc(len);
if (data == NULL) {
// todo: for all stream in this vnode, keep this offset in the offset files, and wait for a moment, and then retry
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("vgId:%d, failed to copy submit data for stream processing, since out of memory", vgId);
return -1;
}
memcpy(data, pReq, len);
SPackedData submit = {.msgStr = data, .msgLen = len, .ver = ver};
tqDebug("vgId:%d tq copy submit msg:%p len:%d ver:%" PRId64 " from %p for stream", vgId, data, len, ver, pReq);
tqProcessSubmitReq(pTq, submit);
#endif
SPackedData submit = {0};
tqProcessSubmitReq(pTq, submit);
}
@ -362,83 +287,39 @@ int32_t tqPushMsg(STQ* pTq, void* msg, int32_t msgLen, tmsg_t msgType, int64_t v
return 0;
}
int32_t tqRegisterPushHandle(STQ* pTq, void* pHandle, const SMqPollReq* pRequest, SRpcMsg* pRpcMsg, SMqDataRsp* pDataRsp,
int32_t type) {
uint64_t consumerId = pRequest->consumerId;
int32_t tqRegisterPushHandle(STQ* pTq, void* handle, SRpcMsg* pMsg) {
int32_t vgId = TD_VID(pTq->pVnode);
STqHandle* pTqHandle = pHandle;
STqPushEntry* pPushEntry = taosMemoryCalloc(1, sizeof(STqPushEntry));
if (pPushEntry == NULL) {
tqDebug("tmq poll: consumer:0x%" PRIx64 ", vgId:%d failed to malloc, size:%d", consumerId, vgId,
(int32_t)sizeof(STqPushEntry));
terrno = TSDB_CODE_OUT_OF_MEMORY;
return -1;
STqHandle* pHandle = (STqHandle*) handle;
if(pHandle->msg == NULL){
pHandle->msg = taosMemoryCalloc(1, sizeof(SRpcMsg));
memcpy(pHandle->msg, pMsg, sizeof(SRpcMsg));
pHandle->msg->pCont = rpcMallocCont(pMsg->contLen);
}else{
void *tmp = pHandle->msg->pCont;
memcpy(pHandle->msg, pMsg, sizeof(SRpcMsg));
pHandle->msg->pCont = tmp;
}
pPushEntry->info = pRpcMsg->info;
memcpy(pPushEntry->subKey, pTqHandle->subKey, TSDB_SUBSCRIBE_KEY_LEN);
if (type == TMQ_MSG_TYPE__TAOSX_RSP) {
pPushEntry->pDataRsp = taosMemoryCalloc(1, sizeof(STaosxRsp));
memcpy(pPushEntry->pDataRsp, pDataRsp, sizeof(STaosxRsp));
} else if (type == TMQ_MSG_TYPE__POLL_RSP) {
pPushEntry->pDataRsp = taosMemoryCalloc(1, sizeof(SMqDataRsp));
memcpy(pPushEntry->pDataRsp, pDataRsp, sizeof(SMqDataRsp));
}
SMqRspHead* pHead = &pPushEntry->pDataRsp->head;
pHead->consumerId = consumerId;
pHead->epoch = pRequest->epoch;
pHead->mqMsgType = type;
taosHashPut(pTq->pPushMgr, pTqHandle->subKey, strlen(pTqHandle->subKey), &pPushEntry, sizeof(void*));
tqDebug("tmq poll: consumer:0x%" PRIx64 ", subkey %s offset:%" PRId64 ", vgId:%d save handle to push mgr, total:%d",
consumerId, pTqHandle->subKey, pDataRsp->reqOffset.version, vgId, taosHashGetSize(pTq->pPushMgr));
memcpy(pHandle->msg->pCont, pMsg->pCont, pMsg->contLen);
pHandle->msg->contLen = pMsg->contLen;
int32_t ret = taosHashPut(pTq->pPushMgr, pHandle->subKey, strlen(pHandle->subKey), &pHandle, POINTER_BYTES);
tqDebug("vgId:%d data is over, ret:%d, consumerId:0x%" PRIx64", register to pHandle:%p, pCont:%p, len:%d", vgId, ret, pHandle->consumerId, pHandle, pHandle->msg->pCont, pHandle->msg->contLen);
return 0;
}
int32_t tqUnregisterPushHandle(STQ* pTq, const char* pKey, int32_t keyLen, uint64_t consumerId, bool rspConsumer) {
int32_t vgId = TD_VID(pTq->pVnode);
STqPushEntry** pEntry = taosHashGet(pTq->pPushMgr, pKey, keyLen);
int32_t tqUnregisterPushHandle(STQ* pTq, void *handle) {
STqHandle *pHandle = (STqHandle*)handle;
int32_t vgId = TD_VID(pTq->pVnode);
if (pEntry != NULL) {
uint64_t cId = (*pEntry)->pDataRsp->head.consumerId;
ASSERT(consumerId == cId);
int32_t ret = taosHashRemove(pTq->pPushMgr, pHandle->subKey, strlen(pHandle->subKey));
tqError("vgId:%d remove pHandle:%p,ret:%d consumer Id:0x%" PRIx64, vgId, pHandle, ret, pHandle->consumerId);
if(pHandle->msg != NULL) {
tqPushDataRsp(pTq, pHandle);
tqDebug("tmq poll: consumer:0x%" PRIx64 ", subkey %s vgId:%d remove from push mgr, remains:%d", consumerId,
(*pEntry)->subKey, vgId, taosHashGetSize(pTq->pPushMgr) - 1);
if (rspConsumer) { // rsp the old consumer with empty block.
tqPushDataRsp(pTq, *pEntry);
}
taosHashRemove(pTq->pPushMgr, pKey, keyLen);
rpcFreeCont(pHandle->msg->pCont);
taosMemoryFree(pHandle->msg);
pHandle->msg = NULL;
}
return 0;
}
void recordPushedEntry(SArray* cachedKey, void* pIter) {
size_t kLen = 0;
void* key = taosHashGetKey(pIter, &kLen);
SItem item = {.pKey = strndup(key, kLen), .keyLen = kLen};
taosArrayPush(cachedKey, &item);
}
void doRemovePushedEntry(SArray* pCachedKeys, STQ* pTq) {
int32_t vgId = TD_VID(pTq->pVnode);
int32_t numOfKeys = (int32_t)taosArrayGetSize(pCachedKeys);
for (int32_t i = 0; i < numOfKeys; i++) {
SItem* pItem = taosArrayGet(pCachedKeys, i);
if (taosHashRemove(pTq->pPushMgr, pItem->pKey, pItem->keyLen) != 0) {
tqError("vgId:%d, tq push hash remove key error, key: %s", vgId, (char*)pItem->pKey);
}
}
if (numOfKeys > 0) {
tqDebug("vgId:%d, pushed %d items and remain:%d", vgId, numOfKeys, (int32_t)taosHashGetSize(pTq->pPushMgr));
}
}

448
source/dnode/vnode/src/tq/tqRead.c
View File

@ -13,6 +13,7 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "tmsg.h"
#include "tq.h"
bool isValValidForTable(STqHandle* pHandle, SWalCont* pHead) {
@ -249,7 +250,7 @@ END:
return code;
}
STqReader* tqOpenReader(SVnode* pVnode) {
STqReader* tqReaderOpen(SVnode* pVnode) {
STqReader* pReader = taosMemoryCalloc(1, sizeof(STqReader));
if (pReader == NULL) {
return NULL;
@ -265,9 +266,9 @@ STqReader* tqOpenReader(SVnode* pVnode) {
pReader->pColIdList = NULL;
pReader->cachedSchemaVer = 0;
pReader->cachedSchemaSuid = 0;
pReader->pSchema = NULL;
pReader->pSchemaWrapper = NULL;
pReader->tbIdHash = NULL;
pReader->pResBlock = createDataBlock();
return pReader;
}
@ -276,19 +277,19 @@ void tqCloseReader(STqReader* pReader) {
if (pReader->pWalReader) {
walCloseReader(pReader->pWalReader);
}
// free cached schema
if (pReader->pSchema) {
taosMemoryFree(pReader->pSchema);
}
if (pReader->pSchemaWrapper) {
tDeleteSSchemaWrapper(pReader->pSchemaWrapper);
tDeleteSchemaWrapper(pReader->pSchemaWrapper);
}
if (pReader->pColIdList) {
taosArrayDestroy(pReader->pColIdList);
}
// free hash
blockDataDestroy(pReader->pResBlock);
taosHashCleanup(pReader->tbIdHash);
tDestroySSubmitReq2(&pReader->submit, TSDB_MSG_FLG_DECODE);
tDestroySubmitReq(&pReader->submit, TSDB_MSG_FLG_DECODE);
taosMemoryFree(pReader);
}
@ -322,12 +323,87 @@ int32_t extractSubmitMsgFromWal(SWalReader* pReader, SPackedData* pPackedData) {
return 0;
}
void tqNextBlock(STqReader* pReader, SFetchRet* ret) {
// todo ignore the error in wal?
int32_t tqNextBlockInWal(STqReader* pReader) {
SWalReader* pWalReader = pReader->pWalReader;
while(1) {
SArray* pBlockList = pReader->submit.aSubmitTbData;
if (pBlockList == NULL || pReader->nextBlk >= taosArrayGetSize(pBlockList)) {
// try next message in wal file
if (walNextValidMsg(pWalReader) < 0) {
return FETCH_TYPE__NONE;
}
void* pBody = POINTER_SHIFT(pWalReader->pHead->head.body, sizeof(SSubmitReq2Msg));
int32_t bodyLen = pWalReader->pHead->head.bodyLen - sizeof(SSubmitReq2Msg);
int64_t ver = pWalReader->pHead->head.version;
SDecoder decoder = {0};
tDecoderInit(&decoder, pBody, bodyLen);
{
int32_t nSubmitTbData = taosArrayGetSize(pReader->submit.aSubmitTbData);
for (int32_t i = 0; i < nSubmitTbData; i++) {
SSubmitTbData* pData = taosArrayGet(pReader->submit.aSubmitTbData, i);
if (pData->pCreateTbReq != NULL) {
taosArrayDestroy(pData->pCreateTbReq->ctb.tagName);
taosMemoryFreeClear(pData->pCreateTbReq);
}
pData->aRowP = taosArrayDestroy(pData->aRowP);
}
pReader->submit.aSubmitTbData = taosArrayDestroy(pReader->submit.aSubmitTbData);
}
if (tDecodeSubmitReq(&decoder, &pReader->submit) < 0) {
tDecoderClear(&decoder);
tqError("decode wal file error, msgLen:%d, ver:%"PRId64, bodyLen, ver);
return FETCH_TYPE__NONE;
}
tDecoderClear(&decoder);
pReader->nextBlk = 0;
}
size_t numOfBlocks = taosArrayGetSize(pReader->submit.aSubmitTbData);
while (pReader->nextBlk < numOfBlocks) {
tqDebug("tq reader next data block %p, %d %" PRId64 " %d", pReader->msg.msgStr, pReader->msg.msgLen,
pReader->msg.ver, pReader->nextBlk);
SSubmitTbData* pSubmitTbData = taosArrayGet(pReader->submit.aSubmitTbData, pReader->nextBlk);
if (pReader->tbIdHash == NULL) {
int32_t code = tqRetrieveDataBlock(pReader->pResBlock, pReader, NULL);
if (code == TSDB_CODE_SUCCESS && pReader->pResBlock->info.rows > 0) {
return FETCH_TYPE__DATA;
}
}
void* ret = taosHashGet(pReader->tbIdHash, &pSubmitTbData->uid, sizeof(int64_t));
if (ret != NULL) {
tqDebug("tq reader return submit block, uid:%"PRId64", ver:%"PRId64, pSubmitTbData->uid, pReader->msg.ver);
int32_t code = tqRetrieveDataBlock(pReader->pResBlock, pReader, NULL);
if (code == TSDB_CODE_SUCCESS && pReader->pResBlock->info.rows > 0) {
return FETCH_TYPE__DATA;
}
} else {
pReader->nextBlk += 1;
tqDebug("tq reader discard submit block, uid:%"PRId64", continue", pSubmitTbData->uid);
}
}
tDestroySubmitReq(&pReader->submit, TSDB_MSG_FLG_DECODE);
pReader->msg.msgStr = NULL;
}
}
int32_t tqNextBlock(STqReader* pReader, SSDataBlock* pBlock) {
while (1) {
if (pReader->msg2.msgStr == NULL) {
if (pReader->msg.msgStr == NULL) {
if (walNextValidMsg(pReader->pWalReader) < 0) {
ret->fetchType = FETCH_TYPE__NONE;
return;
return FETCH_TYPE__NONE;
}
void* pBody = POINTER_SHIFT(pReader->pWalReader->pHead->head.body, sizeof(SSubmitReq2Msg));
@ -337,45 +413,45 @@ void tqNextBlock(STqReader* pReader, SFetchRet* ret) {
tqReaderSetSubmitMsg(pReader, pBody, bodyLen, ver);
}
while (tqNextDataBlock(pReader)) {
memset(&ret->data, 0, sizeof(SSDataBlock));
int32_t code = tqRetrieveDataBlock2(&ret->data, pReader, NULL);
if (code != 0 || ret->data.info.rows == 0) {
while (tqNextBlockImpl(pReader)) {
int32_t code = tqRetrieveDataBlock(pReader->pResBlock, pReader, NULL);
if (code != TSDB_CODE_SUCCESS || pBlock->info.rows == 0) {
continue;
}
ret->fetchType = FETCH_TYPE__DATA;
return;
return FETCH_TYPE__DATA;
}
}
}
int32_t tqReaderSetSubmitMsg(STqReader* pReader, void* msgStr, int32_t msgLen, int64_t ver) {
pReader->msg2.msgStr = msgStr;
pReader->msg2.msgLen = msgLen;
pReader->msg2.ver = ver;
pReader->msg.msgStr = msgStr;
pReader->msg.msgLen = msgLen;
pReader->msg.ver = ver;
tqDebug("tq reader set msg %p %d", msgStr, msgLen);
SDecoder decoder;
tDecoderInit(&decoder, pReader->msg2.msgStr, pReader->msg2.msgLen);
if (tDecodeSSubmitReq2(&decoder, &pReader->submit) < 0) {
tDecoderInit(&decoder, pReader->msg.msgStr, pReader->msg.msgLen);
if (tDecodeSubmitReq(&decoder, &pReader->submit) < 0) {
tDecoderClear(&decoder);
tqError("DecodeSSubmitReq2 error, msgLen:%d, ver:%"PRId64, msgLen, ver);
return -1;
}
tDecoderClear(&decoder);
return 0;
}
bool tqNextDataBlock(STqReader* pReader) {
if (pReader->msg2.msgStr == NULL) {
bool tqNextBlockImpl(STqReader* pReader) {
if (pReader->msg.msgStr == NULL) {
return false;
}
int32_t blockSz = taosArrayGetSize(pReader->submit.aSubmitTbData);
while (pReader->nextBlk < blockSz) {
tqDebug("tq reader next data block %p, %d %" PRId64 " %d", pReader->msg2.msgStr, pReader->msg2.msgLen,
pReader->msg2.ver, pReader->nextBlk);
tqDebug("tq reader next data block %p, %d %" PRId64 " %d", pReader->msg.msgStr, pReader->msg.msgLen,
pReader->msg.ver, pReader->nextBlk);
SSubmitTbData* pSubmitTbData = taosArrayGet(pReader->submit.aSubmitTbData, pReader->nextBlk);
if (pReader->tbIdHash == NULL) {
@ -384,24 +460,24 @@ bool tqNextDataBlock(STqReader* pReader) {
void* ret = taosHashGet(pReader->tbIdHash, &pSubmitTbData->uid, sizeof(int64_t));
if (ret != NULL) {
tqDebug("tq reader block found, ver:%"PRId64", uid:%"PRId64, pReader->msg2.ver, pSubmitTbData->uid);
tqDebug("tq reader block found, ver:%"PRId64", uid:%"PRId64, pReader->msg.ver, pSubmitTbData->uid);
return true;
} else {
tqDebug("tq reader discard block, uid:%"PRId64", continue", pSubmitTbData->uid);
tqDebug("tq reader discard submit block, uid:%"PRId64", continue", pSubmitTbData->uid);
}
pReader->nextBlk++;
}
tDestroySSubmitReq2(&pReader->submit, TSDB_MSG_FLG_DECODE);
tDestroySubmitReq(&pReader->submit, TSDB_MSG_FLG_DECODE);
pReader->nextBlk = 0;
pReader->msg2.msgStr = NULL;
pReader->msg.msgStr = NULL;
return false;
}
bool tqNextDataBlockFilterOut2(STqReader* pReader, SHashObj* filterOutUids) {
if (pReader->msg2.msgStr == NULL) return false;
bool tqNextDataBlockFilterOut(STqReader* pReader, SHashObj* filterOutUids) {
if (pReader->msg.msgStr == NULL) return false;
int32_t blockSz = taosArrayGetSize(pReader->submit.aSubmitTbData);
while (pReader->nextBlk < blockSz) {
@ -415,9 +491,9 @@ bool tqNextDataBlockFilterOut2(STqReader* pReader, SHashObj* filterOutUids) {
pReader->nextBlk++;
}
tDestroySSubmitReq2(&pReader->submit, TSDB_MSG_FLG_DECODE);
tDestroySubmitReq(&pReader->submit, TSDB_MSG_FLG_DECODE);
pReader->nextBlk = 0;
pReader->msg2.msgStr = NULL;
pReader->msg.msgStr = NULL;
return false;
}
@ -451,132 +527,183 @@ int32_t tqMaskBlock(SSchemaWrapper* pDst, SSDataBlock* pBlock, const SSchemaWrap
return 0;
}
int32_t tqRetrieveDataBlock2(SSDataBlock* pBlock, STqReader* pReader, SSubmitTbData** pSubmitTbDataRet) {
tqDebug("tq reader retrieve data block %p, index:%d", pReader->msg2.msgStr, pReader->nextBlk);
SSubmitTbData* pSubmitTbData = taosArrayGet(pReader->submit.aSubmitTbData, pReader->nextBlk);
pReader->nextBlk++;
int32_t tqRetrieveDataBlock(SSDataBlock* pBlock, STqReader* pReader, SSubmitTbData** pSubmitTbDataRet) {
tqDebug("tq reader retrieve data block %p, index:%d", pReader->msg.msgStr, pReader->nextBlk);
SSubmitTbData* pSubmitTbData = taosArrayGet(pReader->submit.aSubmitTbData, pReader->nextBlk++);
if (pSubmitTbDataRet) {
*pSubmitTbDataRet = pSubmitTbData;
}
blockDataCleanup(pBlock);
int32_t sversion = pSubmitTbData->sver;
int64_t suid = pSubmitTbData->suid;
int64_t uid = pSubmitTbData->uid;
pReader->lastBlkUid = uid;
pBlock->info.id.uid = uid;
pBlock->info.version = pReader->msg2.ver;
pBlock->info.version = pReader->msg.ver;
if (pReader->cachedSchemaSuid == 0 || pReader->cachedSchemaVer != sversion || pReader->cachedSchemaSuid != suid) {
taosMemoryFree(pReader->pSchema);
pReader->pSchema = metaGetTbTSchema(pReader->pVnodeMeta, uid, sversion, 1);
if (pReader->pSchema == NULL) {
tqWarn("vgId:%d, cannot found tsschema for table: uid:%" PRId64 " (suid:%" PRId64
"), version %d, possibly dropped table",
pReader->pWalReader->pWal->cfg.vgId, uid, suid, sversion);
pReader->cachedSchemaSuid = 0;
terrno = TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND;
return -1;
}
if ((suid != 0 && pReader->cachedSchemaSuid != suid) || (suid == 0 && pReader->cachedSchemaUid != uid) || (pReader->cachedSchemaVer != sversion)) {
tDeleteSchemaWrapper(pReader->pSchemaWrapper);
tDeleteSSchemaWrapper(pReader->pSchemaWrapper);
pReader->pSchemaWrapper = metaGetTableSchema(pReader->pVnodeMeta, uid, sversion, 1);
if (pReader->pSchemaWrapper == NULL) {
tqWarn("vgId:%d, cannot found schema wrapper for table: suid:%" PRId64 ", version %d, possibly dropped table",
pReader->pWalReader->pWal->cfg.vgId, uid, pReader->cachedSchemaVer);
tqWarn("vgId:%d, cannot found schema wrapper for table: suid:%" PRId64 ", uid:%" PRId64 "version %d, possibly dropped table",
pReader->pWalReader->pWal->cfg.vgId, suid, uid, pReader->cachedSchemaVer);
pReader->cachedSchemaSuid = 0;
terrno = TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND;
return -1;
}
STSchema* pTschema = pReader->pSchema;
pReader->cachedSchemaUid = uid;
pReader->cachedSchemaSuid = suid;
pReader->cachedSchemaVer = sversion;
SSchemaWrapper* pSchemaWrapper = pReader->pSchemaWrapper;
if (blockDataGetNumOfCols(pBlock) > 0) {
blockDataDestroy(pReader->pResBlock);
pReader->pResBlock = createDataBlock();
pBlock = pReader->pResBlock;
int32_t colNumNeed = taosArrayGetSize(pReader->pColIdList);
pBlock->info.id.uid = uid;
pBlock->info.version = pReader->msg.ver;
}
if (colNumNeed == 0) {
int32_t colMeta = 0;
while (colMeta < pSchemaWrapper->nCols) {
SSchema* pColSchema = &pSchemaWrapper->pSchema[colMeta];
int32_t numOfCols = taosArrayGetSize(pReader->pColIdList);
if (numOfCols == 0) { // all columns are required
for (int32_t i = 0; i < pSchemaWrapper->nCols; ++i) {
SSchema* pColSchema = &pSchemaWrapper->pSchema[i];
SColumnInfoData colInfo = createColumnInfoData(pColSchema->type, pColSchema->bytes, pColSchema->colId);
int32_t code = blockDataAppendColInfo(pBlock, &colInfo);
int32_t code = blockDataAppendColInfo(pBlock, &colInfo);
if (code != TSDB_CODE_SUCCESS) {
goto FAIL;
blockDataFreeRes(pBlock);
return -1;
}
colMeta++;
}
} else {
if (colNumNeed > pSchemaWrapper->nCols) {
colNumNeed = pSchemaWrapper->nCols;
if (numOfCols > pSchemaWrapper->nCols) {
numOfCols = pSchemaWrapper->nCols;
}
int32_t colMeta = 0;
int32_t colNeed = 0;
while (colMeta < pSchemaWrapper->nCols && colNeed < colNumNeed) {
SSchema* pColSchema = &pSchemaWrapper->pSchema[colMeta];
int32_t i = 0;
int32_t j = 0;
while (i < pSchemaWrapper->nCols && j < numOfCols) {
SSchema* pColSchema = &pSchemaWrapper->pSchema[i];
col_id_t colIdSchema = pColSchema->colId;
col_id_t colIdNeed = *(col_id_t*)taosArrayGet(pReader->pColIdList, colNeed);
col_id_t colIdNeed = *(col_id_t*)taosArrayGet(pReader->pColIdList, j);
if (colIdSchema < colIdNeed) {
colMeta++;
i++;
} else if (colIdSchema > colIdNeed) {
colNeed++;
j++;
} else {
SColumnInfoData colInfo = createColumnInfoData(pColSchema->type, pColSchema->bytes, pColSchema->colId);
int32_t code = blockDataAppendColInfo(pBlock, &colInfo);
if (code != TSDB_CODE_SUCCESS) {
goto FAIL;
}
colMeta++;
colNeed++;
i++;
j++;
}
}
}
}
int32_t numOfRows = 0;
int32_t numOfRows = 0;
if (pSubmitTbData->flags & SUBMIT_REQ_COLUMN_DATA_FORMAT) {
SColData* pCol = taosArrayGet(pSubmitTbData->aCol, 0);
numOfRows = pCol->nVal;
} else {
numOfRows = taosArrayGetSize(pSubmitTbData->aRowP);
}
if (pSubmitTbData->flags & SUBMIT_REQ_COLUMN_DATA_FORMAT) {
SArray* pCols = pSubmitTbData->aCol;
SColData* pCol = taosArrayGet(pCols, 0);
numOfRows = pCol->nVal;
} else {
SArray* pRows = pSubmitTbData->aRowP;
numOfRows = taosArrayGetSize(pRows);
if (blockDataEnsureCapacity(pBlock, numOfRows) < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
goto FAIL;
}
pBlock->info.rows = numOfRows;
int32_t colActual = blockDataGetNumOfCols(pBlock);
// convert and scan one block
if (pSubmitTbData->flags & SUBMIT_REQ_COLUMN_DATA_FORMAT) {
SArray* pCols = pSubmitTbData->aCol;
int32_t numOfCols = taosArrayGetSize(pCols);
int32_t targetIdx = 0;
int32_t sourceIdx = 0;
while (targetIdx < colActual) {
if (sourceIdx >= numOfCols) {
tqError("tqRetrieveDataBlock sourceIdx:%d >= numOfCols:%d", sourceIdx, numOfCols);
goto FAIL;
}
SColData* pCol = taosArrayGet(pCols, sourceIdx);
SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, targetIdx);
SColVal colVal;
if (pCol->nVal != numOfRows) {
tqError("tqRetrieveDataBlock pCol->nVal:%d != numOfRows:%d", pCol->nVal, numOfRows);
goto FAIL;
}
if (pCol->cid < pColData->info.colId) {
sourceIdx++;
} else if (pCol->cid == pColData->info.colId) {
for (int32_t i = 0; i < pCol->nVal; i++) {
tColDataGetValue(pCol, i, &colVal);
if (IS_STR_DATA_TYPE(colVal.type)) {
if (colVal.value.pData != NULL) {
char val[65535 + 2] = {0};
memcpy(varDataVal(val), colVal.value.pData, colVal.value.nData);
varDataSetLen(val, colVal.value.nData);
if (colDataAppend(pColData, i, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
goto FAIL;
}
} else {
colDataSetNULL(pColData, i);
}
} else {
if (colDataAppend(pColData, i, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
goto FAIL;
}
}
}
sourceIdx++;
targetIdx++;
} else {
for (int32_t i = 0; i < pCol->nVal; i++) {
colDataSetNULL(pColData, i);
}
targetIdx++;
}
}
} else {
SArray* pRows = pSubmitTbData->aRowP;
SSchemaWrapper* pWrapper = pReader->pSchemaWrapper;
STSchema* pTSchema = tBuildTSchema(pWrapper->pSchema, pWrapper->nCols, pWrapper->version);
if (blockDataEnsureCapacity(pBlock, numOfRows) < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
goto FAIL;
}
pBlock->info.rows = numOfRows;
int32_t colActual = blockDataGetNumOfCols(pBlock);
// convert and scan one block
if (pSubmitTbData->flags & SUBMIT_REQ_COLUMN_DATA_FORMAT) {
SArray* pCols = pSubmitTbData->aCol;
int32_t numOfCols = taosArrayGetSize(pCols);
int32_t targetIdx = 0;
for (int32_t i = 0; i < numOfRows; i++) {
SRow* pRow = taosArrayGetP(pRows, i);
int32_t sourceIdx = 0;
while (targetIdx < colActual) {
if(sourceIdx >= numOfCols){
tqError("tqRetrieveDataBlock2 sourceIdx:%d >= numOfCols:%d", sourceIdx, numOfCols);
goto FAIL;
}
SColData* pCol = taosArrayGet(pCols, sourceIdx);
SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, targetIdx);
SColVal colVal;
if(pCol->nVal != numOfRows){
tqError("tqRetrieveDataBlock2 pCol->nVal:%d != numOfRows:%d", pCol->nVal, numOfRows);
goto FAIL;
}
for (int32_t j = 0; j < colActual; j++) {
SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, j);
while (1) {
SColVal colVal;
tqDebug("start to extract column id:%d, index:%d", pColData->info.colId, sourceIdx);
if (pCol->cid < pColData->info.colId) {
sourceIdx++;
} else if (pCol->cid == pColData->info.colId) {
for (int32_t i = 0; i < pCol->nVal; i++) {
tColDataGetValue(pCol, i, &colVal);
tRowGet(pRow, pTSchema, sourceIdx, &colVal);
if (colVal.cid < pColData->info.colId) {
tqDebug("colIndex:%d column id:%d in row, ignore, the required colId:%d, total cols in schema:%d",
sourceIdx, colVal.cid, pColData->info.colId, pTSchema->numOfCols);
sourceIdx++;
continue;
} else if (colVal.cid == pColData->info.colId) {
if (IS_STR_DATA_TYPE(colVal.type)) {
if (colVal.value.pData != NULL) {
char val[65535 + 2] = {0};
@ -593,59 +720,18 @@ int32_t tqRetrieveDataBlock2(SSDataBlock* pBlock, STqReader* pReader, SSubmitTbD
goto FAIL;
}
}
}
sourceIdx++;
targetIdx++;
} else {
for (int32_t i = 0; i < pCol->nVal; i++) {
sourceIdx++;
break;
} else {
colDataSetNULL(pColData, i);
}
targetIdx++;
}
}
} else {
SArray* pRows = pSubmitTbData->aRowP;
for (int32_t i = 0; i < numOfRows; i++) {
SRow* pRow = taosArrayGetP(pRows, i);
int32_t sourceIdx = 0;
for (int32_t j = 0; j < colActual; j++) {
SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, j);
while (1) {
SColVal colVal;
tRowGet(pRow, pTschema, sourceIdx, &colVal);
if (colVal.cid < pColData->info.colId) {
sourceIdx++;
continue;
} else if (colVal.cid == pColData->info.colId) {
if (IS_STR_DATA_TYPE(colVal.type)) {
if (colVal.value.pData != NULL) {
char val[65535 + 2] = {0};
memcpy(varDataVal(val), colVal.value.pData, colVal.value.nData);
varDataSetLen(val, colVal.value.nData);
if (colDataAppend(pColData, i, val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
goto FAIL;
}
} else {
colDataSetNULL(pColData, i);
}
} else {
if (colDataAppend(pColData, i, (void*)&colVal.value.val, !COL_VAL_IS_VALUE(&colVal)) < 0) {
goto FAIL;
}
}
sourceIdx++;
break;
} else {
colDataSetNULL(pColData, i);
break;
}
break;
}
}
}
}
taosMemoryFreeClear(pTSchema);
}
return 0;
@ -655,8 +741,8 @@ FAIL:
return -1;
}
int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schemas, SSubmitTbData** pSubmitTbDataRet) {
tqDebug("tq reader retrieve data block %p, %d", pReader->msg2.msgStr, pReader->nextBlk);
int32_t tqRetrieveTaosxBlock(STqReader* pReader, SArray* blocks, SArray* schemas, SSubmitTbData** pSubmitTbDataRet) {
tqDebug("tq reader retrieve data block %p, %d", pReader->msg.msgStr, pReader->nextBlk);
SSubmitTbData* pSubmitTbData = taosArrayGet(pReader->submit.aSubmitTbData, pReader->nextBlk);
pReader->nextBlk++;
@ -667,18 +753,7 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
int64_t uid = pSubmitTbData->uid;
pReader->lastBlkUid = uid;
taosMemoryFree(pReader->pSchema);
pReader->pSchema = metaGetTbTSchema(pReader->pVnodeMeta, uid, sversion, 1);
if (pReader->pSchema == NULL) {
tqWarn("vgId:%d, cannot found tsschema for table: uid:%" PRId64 " (suid:%" PRId64
"), version %d, possibly dropped table",
pReader->pWalReader->pWal->cfg.vgId, uid, suid, sversion);
pReader->cachedSchemaSuid = 0;
terrno = TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND;
return -1;
}
tDeleteSSchemaWrapper(pReader->pSchemaWrapper);
tDeleteSchemaWrapper(pReader->pSchemaWrapper);
pReader->pSchemaWrapper = metaGetTableSchema(pReader->pVnodeMeta, uid, sversion, 1);
if (pReader->pSchemaWrapper == NULL) {
tqWarn("vgId:%d, cannot found schema wrapper for table: suid:%" PRId64 ", version %d, possibly dropped table",
@ -688,7 +763,6 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
return -1;
}
STSchema* pTschema = pReader->pSchema;
SSchemaWrapper* pSchemaWrapper = pReader->pSchemaWrapper;
int32_t numOfRows = 0;
@ -745,18 +819,18 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
if (tqMaskBlock(pSW, &block, pSchemaWrapper, assigned) < 0) {
blockDataFreeRes(&block);
tDeleteSSchemaWrapper(pSW);
tDeleteSchemaWrapper(pSW);
goto FAIL;
}
tqDebug("vgId:%d, build new block, col %d", pReader->pWalReader->pWal->cfg.vgId,
(int32_t)taosArrayGetSize(block.pDataBlock));
block.info.id.uid = uid;
block.info.version = pReader->msg2.ver;
block.info.version = pReader->msg.ver;
if (blockDataEnsureCapacity(&block, numOfRows - curRow) < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
blockDataFreeRes(&block);
tDeleteSSchemaWrapper(pSW);
tDeleteSchemaWrapper(pSW);
goto FAIL;
}
taosArrayPush(blocks, &block);
@ -805,14 +879,17 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
curRow++;
}
} else {
SSchemaWrapper* pWrapper = pReader->pSchemaWrapper;
STSchema* pTSchema = tBuildTSchema(pWrapper->pSchema, pWrapper->nCols, pWrapper->version);
SArray* pRows = pSubmitTbData->aRowP;
for (int32_t i = 0; i < numOfRows; i++) {
SRow* pRow = taosArrayGetP(pRows, i);
bool buildNew = false;
for (int32_t j = 0; j < pTschema->numOfCols; j++) {
for (int32_t j = 0; j < pTSchema->numOfCols; j++) {
SColVal colVal;
tRowGet(pRow, pTschema, j, &colVal);
tRowGet(pRow, pTSchema, j, &colVal);
if (curRow == 0) {
assigned[j] = !COL_VAL_IS_NONE(&colVal);
buildNew = true;
@ -841,18 +918,18 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
if (tqMaskBlock(pSW, &block, pSchemaWrapper, assigned) < 0) {
blockDataFreeRes(&block);
tDeleteSSchemaWrapper(pSW);
tDeleteSchemaWrapper(pSW);
goto FAIL;
}
tqDebug("vgId:%d, build new block, col %d", pReader->pWalReader->pWal->cfg.vgId,
(int32_t)taosArrayGetSize(block.pDataBlock));
block.info.id.uid = uid;
block.info.version = pReader->msg2.ver;
block.info.version = pReader->msg.ver;
if (blockDataEnsureCapacity(&block, numOfRows - curRow) < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
blockDataFreeRes(&block);
tDeleteSSchemaWrapper(pSW);
tDeleteSchemaWrapper(pSW);
goto FAIL;
}
taosArrayPush(blocks, &block);
@ -870,7 +947,7 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
while (targetIdx < colActual) {
SColumnInfoData* pColData = taosArrayGet(pBlock->pDataBlock, targetIdx);
SColVal colVal;
tRowGet(pRow, pTschema, sourceIdx, &colVal);
tRowGet(pRow, pTSchema, sourceIdx, &colVal);
if (colVal.cid < pColData->info.colId) {
sourceIdx++;
@ -897,6 +974,8 @@ int32_t tqRetrieveTaosxBlock2(STqReader* pReader, SArray* blocks, SArray* schema
}
curRow++;
}
taosMemoryFreeClear(pTSchema);
}
SSDataBlock* pLastBlock = taosArrayGetLast(blocks);
@ -1041,6 +1120,5 @@ int32_t tqUpdateTbUidList(STQ* pTq, const SArray* tbUidList, bool isAdd) {
}
taosWUnLockLatch(&pTq->pStreamMeta->lock);
return 0;
}

56
source/dnode/vnode/src/tq/tqRestore.c
View File

@ -18,16 +18,15 @@
static int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle);
// this function should be executed by stream threads.
// there is a case that the WAL increases more fast than the restore procedure, and this restore procedure
// will not stop eventually.
// extract submit block from WAL, and add them into the input queue for the sources tasks.
int32_t tqStreamTasksScanWal(STQ* pTq) {
int32_t vgId = TD_VID(pTq->pVnode);
int32_t vgId = TD_VID(pTq->pVnode);
SStreamMeta* pMeta = pTq->pStreamMeta;
int64_t st = taosGetTimestampMs();
int64_t st = taosGetTimestampMs();
while (1) {
int32_t scan = pMeta->walScan;
tqDebug("vgId:%d continue check if data in wal are available, scan:%d", vgId, scan);
int32_t scan = pMeta->walScanCounter;
tqDebug("vgId:%d continue check if data in wal are available, walScanCounter:%d", vgId, scan);
// check all restore tasks
bool shouldIdle = true;
@ -37,12 +36,13 @@ int32_t tqStreamTasksScanWal(STQ* pTq) {
if (shouldIdle) {
taosWLockLatch(&pMeta->lock);
pMeta->walScan -= 1;
times = pMeta->walScan;
ASSERT(pMeta->walScan >= 0);
pMeta->walScanCounter -= 1;
times = pMeta->walScanCounter;
if (pMeta->walScan <= 0) {
ASSERT(pMeta->walScanCounter >= 0);
if (pMeta->walScanCounter <= 0) {
taosWUnLockLatch(&pMeta->lock);
break;
}
@ -57,42 +57,28 @@ int32_t tqStreamTasksScanWal(STQ* pTq) {
return 0;
}
static SArray* extractTaskIdList(SStreamMeta* pStreamMeta, int32_t numOfTasks) {
SArray* pTaskIdList = taosArrayInit(numOfTasks, sizeof(int32_t));
void* pIter = NULL;
taosWLockLatch(&pStreamMeta->lock);
while(1) {
pIter = taosHashIterate(pStreamMeta->pTasks, pIter);
if (pIter == NULL) {
break;
}
SStreamTask* pTask = *(SStreamTask**)pIter;
taosArrayPush(pTaskIdList, &pTask->id.taskId);
}
taosWUnLockLatch(&pStreamMeta->lock);
return pTaskIdList;
}
int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle) {
*pScanIdle = true;
bool noNewDataInWal = true;
int32_t vgId = pStreamMeta->vgId;
int32_t numOfTasks = taosHashGetSize(pStreamMeta->pTasks);
int32_t numOfTasks = taosArrayGetSize(pStreamMeta->pTaskList);
if (numOfTasks == 0) {
return TSDB_CODE_SUCCESS;
}
SArray* pTaskList = NULL;
taosWLockLatch(&pStreamMeta->lock);
pTaskList = taosArrayDup(pStreamMeta->pTaskList, NULL);
taosWUnLockLatch(&pStreamMeta->lock);
tqDebug("vgId:%d start to check wal to extract new submit block for %d tasks", vgId, numOfTasks);
SArray* pTaskIdList = extractTaskIdList(pStreamMeta, numOfTasks);
// update the new task number
numOfTasks = taosArrayGetSize(pTaskIdList);
numOfTasks = taosArrayGetSize(pTaskList);
for (int32_t i = 0; i < numOfTasks; ++i) {
int32_t* pTaskId = taosArrayGet(pTaskIdList, i);
int32_t* pTaskId = taosArrayGet(pTaskList, i);
SStreamTask* pTask = streamMetaAcquireTask(pStreamMeta, *pTaskId);
if (pTask == NULL) {
continue;
@ -107,7 +93,7 @@ int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle) {
if (streamTaskShouldStop(&pTask->status) || status == TASK_STATUS__RECOVER_PREPARE ||
status == TASK_STATUS__WAIT_DOWNSTREAM) {
tqDebug("s-task:%s skip push data, not ready for processing, status %d", pTask->id.idStr, status);
tqDebug("s-task:%s not ready for new submit block from wal, status:%d", pTask->id.idStr, status);
streamMetaReleaseTask(pStreamMeta, pTask);
continue;
}
@ -166,7 +152,7 @@ int32_t createStreamRunReq(SStreamMeta* pStreamMeta, bool* pScanIdle) {
*pScanIdle = true;
}
taosArrayDestroy(pTaskIdList);
taosArrayDestroy(pTaskList);
return 0;
}

15
source/dnode/vnode/src/tq/tqScan.c
View File

@ -74,7 +74,6 @@ int32_t tqScanData(STQ* pTq, const STqHandle* pHandle, SMqDataRsp* pRsp, STqOffs
qTaskInfo_t task = pExec->task;
if (qStreamPrepareScan(task, pOffset, pHandle->execHandle.subType) < 0) {
tqError("prepare scan failed, return");
return -1;
}
@ -119,7 +118,6 @@ int32_t tqScanTaosx(STQ* pTq, const STqHandle* pHandle, STaosxRsp* pRsp, SMqMeta
qTaskInfo_t task = pExec->task;
if (qStreamPrepareScan(task, pOffset, pHandle->execHandle.subType) < 0) {
tqDebug("tqScanTaosx prepare scan failed, return");
return -1;
}
@ -132,6 +130,7 @@ int32_t tqScanTaosx(STQ* pTq, const STqHandle* pHandle, STaosxRsp* pRsp, SMqMeta
tqError("vgId:%d, task exec error since %s", pTq->pVnode->config.vgId, terrstr());
return -1;
}
tqDebug("tmqsnap task execute end, get %p", pDataBlock);
if (pDataBlock != NULL && pDataBlock->info.rows > 0) {
@ -205,18 +204,18 @@ int32_t tqTaosxScanLog(STQ* pTq, STqHandle* pHandle, SPackedData submit, STaosxR
if (pExec->subType == TOPIC_SUB_TYPE__TABLE) {
STqReader* pReader = pExec->pTqReader;
tqReaderSetSubmitMsg(pReader, submit.msgStr, submit.msgLen, submit.ver);
while (tqNextDataBlock(pReader)) {
while (tqNextBlockImpl(pReader)) {
taosArrayClear(pBlocks);
taosArrayClear(pSchemas);
SSubmitTbData* pSubmitTbDataRet = NULL;
if (tqRetrieveTaosxBlock2(pReader, pBlocks, pSchemas, &pSubmitTbDataRet) < 0) {
if (tqRetrieveTaosxBlock(pReader, pBlocks, pSchemas, &pSubmitTbDataRet) < 0) {
if (terrno == TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND) continue;
}
if (pRsp->withTbName) {
int64_t uid = pExec->pTqReader->lastBlkUid;
if (tqAddTbNameToRsp(pTq, uid, pRsp, taosArrayGetSize(pBlocks)) < 0) {
taosArrayDestroyEx(pBlocks, (FDelete)blockDataFreeRes);
taosArrayDestroyP(pSchemas, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pSchemas, (FDelete)tDeleteSchemaWrapper);
pBlocks = taosArrayInit(0, sizeof(SSDataBlock));
pSchemas = taosArrayInit(0, sizeof(void*));
continue;
@ -264,18 +263,18 @@ int32_t tqTaosxScanLog(STQ* pTq, STqHandle* pHandle, SPackedData submit, STaosxR
} else if (pExec->subType == TOPIC_SUB_TYPE__DB) {
STqReader* pReader = pExec->pTqReader;
tqReaderSetSubmitMsg(pReader, submit.msgStr, submit.msgLen, submit.ver);
while (tqNextDataBlockFilterOut2(pReader, pExec->execDb.pFilterOutTbUid)) {
while (tqNextDataBlockFilterOut(pReader, pExec->execDb.pFilterOutTbUid)) {
taosArrayClear(pBlocks);
taosArrayClear(pSchemas);
SSubmitTbData* pSubmitTbDataRet = NULL;
if (tqRetrieveTaosxBlock2(pReader, pBlocks, pSchemas, &pSubmitTbDataRet) < 0) {
if (tqRetrieveTaosxBlock(pReader, pBlocks, pSchemas, &pSubmitTbDataRet) < 0) {
if (terrno == TSDB_CODE_TQ_TABLE_SCHEMA_NOT_FOUND) continue;
}
if (pRsp->withTbName) {
int64_t uid = pExec->pTqReader->lastBlkUid;
if (tqAddTbNameToRsp(pTq, uid, pRsp, taosArrayGetSize(pBlocks)) < 0) {
taosArrayDestroyEx(pBlocks, (FDelete)blockDataFreeRes);
taosArrayDestroyP(pSchemas, (FDelete)tDeleteSSchemaWrapper);
taosArrayDestroyP(pSchemas, (FDelete)tDeleteSchemaWrapper);
pBlocks = taosArrayInit(0, sizeof(SSDataBlock));
pSchemas = taosArrayInit(0, sizeof(void*));
continue;

14
source/dnode/vnode/src/tq/tqSink.c
View File

@ -672,7 +672,7 @@ void tqSinkToTablePipeline2(SStreamTask* pTask, void* vnode, int64_t ver, void*
}
SRow* pRow = NULL;
if ((terrno = tRowBuild(pVals, (STSchema*)pTSchema, &pRow)) < 0) {
tDestroySSubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
goto _end;
}
ASSERT(pRow);
@ -681,7 +681,7 @@ void tqSinkToTablePipeline2(SStreamTask* pTask, void* vnode, int64_t ver, void*
SSubmitReq2 submitReq = {0};
if (!(submitReq.aSubmitTbData = taosArrayInit(1, sizeof(SSubmitTbData)))) {
tDestroySSubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
goto _end;
}
@ -690,28 +690,28 @@ void tqSinkToTablePipeline2(SStreamTask* pTask, void* vnode, int64_t ver, void*
// encode
int32_t len;
int32_t code;
tEncodeSize(tEncodeSSubmitReq2, &submitReq, len, code);
tEncodeSize(tEncodeSubmitReq, &submitReq, len, code);
SEncoder encoder;
len += sizeof(SSubmitReq2Msg);
pBuf = rpcMallocCont(len);
if (NULL == pBuf) {
tDestroySSubmitReq2(&submitReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(&submitReq, TSDB_MSG_FLG_ENCODE);
goto _end;
}
((SSubmitReq2Msg*)pBuf)->header.vgId = TD_VID(pVnode);
((SSubmitReq2Msg*)pBuf)->header.contLen = htonl(len);
((SSubmitReq2Msg*)pBuf)->version = htobe64(1);
tEncoderInit(&encoder, POINTER_SHIFT(pBuf, sizeof(SSubmitReq2Msg)), len - sizeof(SSubmitReq2Msg));
if (tEncodeSSubmitReq2(&encoder, &submitReq) < 0) {
if (tEncodeSubmitReq(&encoder, &submitReq) < 0) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
tqError("failed to encode submit req since %s", terrstr());
tEncoderClear(&encoder);
rpcFreeCont(pBuf);
tDestroySSubmitReq2(&submitReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(&submitReq, TSDB_MSG_FLG_ENCODE);
continue;
}
tEncoderClear(&encoder);
tDestroySSubmitReq2(&submitReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(&submitReq, TSDB_MSG_FLG_ENCODE);
SRpcMsg msg = {
.msgType = TDMT_VND_SUBMIT,

18
source/dnode/vnode/src/tq/tqUtil.c
View File

@ -169,20 +169,20 @@ static int32_t extractDataAndRspForNormalSubscribe(STQ* pTq, STqHandle* pHandle,
SMqDataRsp dataRsp = {0};
tqInitDataRsp(&dataRsp, pRequest, pHandle->execHandle.subType);
// lock
taosWLockLatch(&pTq->lock);
qSetTaskId(pHandle->execHandle.task, consumerId, pRequest->reqId);
int code = tqScanData(pTq, pHandle, &dataRsp, pOffset);
if(code != 0) {
goto end;
}
// till now, all data has been transferred to consumer, new data needs to push client once arrived.
// till now, all data has been transferred to consumer, new data needs to push client once arrived.
if (dataRsp.blockNum == 0 && dataRsp.reqOffset.type == TMQ_OFFSET__LOG &&
dataRsp.reqOffset.version == dataRsp.rspOffset.version && pHandle->consumerId == pRequest->consumerId) {
code = tqRegisterPushHandle(pTq, pHandle, pRequest, pMsg, &dataRsp, TMQ_MSG_TYPE__POLL_RSP);
// lock
taosWLockLatch(&pTq->lock);
code = tqRegisterPushHandle(pTq, pHandle, pMsg);
taosWUnLockLatch(&pTq->lock);
tDeleteSMqDataRsp(&dataRsp);
return code;
}
@ -197,7 +197,7 @@ static int32_t extractDataAndRspForNormalSubscribe(STQ* pTq, STqHandle* pHandle,
tFormatOffset(buf, 80, &dataRsp.rspOffset);
tqDebug("tmq poll: consumer:0x%" PRIx64 ", subkey %s, vgId:%d, rsp block:%d, rsp offset type:%s, reqId:0x%" PRIx64 " code:%d",
consumerId, pHandle->subKey, vgId, dataRsp.blockNum, buf, pRequest->reqId, code);
taosWUnLockLatch(&pTq->lock);
// taosWUnLockLatch(&pTq->lock);
tDeleteSMqDataRsp(&dataRsp);
}
return code;
@ -211,6 +211,12 @@ static int32_t extractDataAndRspForDbStbSubscribe(STQ* pTq, STqHandle* pHandle,
SMqMetaRsp metaRsp = {0};
STaosxRsp taosxRsp = {0};
tqInitTaosxRsp(&taosxRsp, pRequest);
qTaskInfo_t task = pHandle->execHandle.task;
if(qTaskIsExecuting(task)){
code = tqSendDataRsp(pTq, pMsg, pRequest, (SMqDataRsp*)&taosxRsp, TMQ_MSG_TYPE__TAOSX_RSP);
tDeleteSTaosxRsp(&taosxRsp);
return code;
}
if (offset->type != TMQ_OFFSET__LOG) {
if (tqScanTaosx(pTq, pHandle, &taosxRsp, &metaRsp, offset) < 0) {

12
source/dnode/vnode/src/tsdb/tsdbCache.c
View File

@ -598,6 +598,7 @@ typedef struct {
SMergeTree mergeTree;
SMergeTree *pMergeTree;
SSttBlockLoadInfo *pLoadInfo;
SLDataIter* pDataIter;
int64_t lastTs;
} SFSLastNextRowIter;
@ -645,7 +646,7 @@ static int32_t getNextRowFromFSLast(void *iter, TSDBROW **ppRow, bool *pIgnoreEa
}
tMergeTreeOpen(&state->mergeTree, 1, *state->pDataFReader, state->suid, state->uid,
&(STimeWindow){.skey = state->lastTs, .ekey = TSKEY_MAX},
&(SVersionRange){.minVer = 0, .maxVer = UINT64_MAX}, state->pLoadInfo, false, NULL, true);
&(SVersionRange){.minVer = 0, .maxVer = UINT64_MAX}, state->pLoadInfo, false, NULL, true, state->pDataIter);
state->pMergeTree = &state->mergeTree;
state->state = SFSLASTNEXTROW_BLOCKROW;
}
@ -667,7 +668,7 @@ static int32_t getNextRowFromFSLast(void *iter, TSDBROW **ppRow, bool *pIgnoreEa
state->state = SFSLASTNEXTROW_FILESET;
goto _next_fileset;
}
state->row = tMergeTreeGetRow(&state->mergeTree);
state->row = *tMergeTreeGetRow(&state->mergeTree);
*ppRow = &state->row;
if (TSDBROW_TS(&state->row) <= state->lastTs) {
@ -1211,7 +1212,7 @@ typedef struct {
} CacheNextRowIter;
static int32_t nextRowIterOpen(CacheNextRowIter *pIter, tb_uid_t uid, STsdb *pTsdb, STSchema *pTSchema, tb_uid_t suid,
SSttBlockLoadInfo *pLoadInfo, STsdbReadSnap *pReadSnap, SDataFReader **pDataFReader,
SSttBlockLoadInfo *pLoadInfo, SLDataIter* pLDataIter, STsdbReadSnap *pReadSnap, SDataFReader **pDataFReader,
SDataFReader **pDataFReaderLast, int64_t lastTs) {
int code = 0;
@ -1274,6 +1275,7 @@ static int32_t nextRowIterOpen(CacheNextRowIter *pIter, tb_uid_t uid, STsdb *pTs
pIter->fsLastState.pLoadInfo = pLoadInfo;
pIter->fsLastState.pDataFReader = pDataFReaderLast;
pIter->fsLastState.lastTs = lastTs;
pIter->fsLastState.pDataIter = pLDataIter;
pIter->fsState.state = SFSNEXTROW_FS;
pIter->fsState.pTsdb = pTsdb;
@ -1465,7 +1467,7 @@ static int32_t mergeLastRow(tb_uid_t uid, STsdb *pTsdb, bool *dup, SArray **ppCo
TSKEY lastRowTs = TSKEY_MAX;
CacheNextRowIter iter = {0};
nextRowIterOpen(&iter, uid, pTsdb, pTSchema, pr->suid, pr->pLoadInfo, pr->pReadSnap, &pr->pDataFReader,
nextRowIterOpen(&iter, uid, pTsdb, pTSchema, pr->suid, pr->pLoadInfo, pr->pDataIter, pr->pReadSnap, &pr->pDataFReader,
&pr->pDataFReaderLast, pr->lastTs);
do {
@ -1622,7 +1624,7 @@ static int32_t mergeLast(tb_uid_t uid, STsdb *pTsdb, SArray **ppLastArray, SCach
TSKEY lastRowTs = TSKEY_MAX;
CacheNextRowIter iter = {0};
nextRowIterOpen(&iter, uid, pTsdb, pTSchema, pr->suid, pr->pLoadInfo, pr->pReadSnap, &pr->pDataFReader,
nextRowIterOpen(&iter, uid, pTsdb, pTSchema, pr->suid, pr->pLoadInfo, pr->pDataIter, pr->pReadSnap, &pr->pDataFReader,
&pr->pDataFReaderLast, pr->lastTs);
do {

11
source/dnode/vnode/src/tsdb/tsdbCacheRead.c
View File

@ -187,13 +187,21 @@ int32_t tsdbCacherowsReaderOpen(void* pVnode, int32_t type, void* pTableIdList,
}
}
int32_t numOfStt = ((SVnode*)pVnode)->config.sttTrigger;
SVnodeCfg* pCfg = &((SVnode*)pVnode)->config;
int32_t numOfStt = pCfg->sttTrigger;
p->pLoadInfo = tCreateLastBlockLoadInfo(p->pSchema, NULL, 0, numOfStt);
if (p->pLoadInfo == NULL) {
tsdbCacherowsReaderClose(p);
return TSDB_CODE_OUT_OF_MEMORY;
}
p->pDataIter = taosMemoryCalloc(pCfg->sttTrigger, sizeof(SLDataIter));
if (p->pDataIter == NULL) {
tsdbCacherowsReaderClose(p);
return TSDB_CODE_OUT_OF_MEMORY;
}
p->idstr = taosStrdup(idstr);
taosThreadMutexInit(&p->readerMutex, NULL);
@ -215,6 +223,7 @@ void* tsdbCacherowsReaderClose(void* pReader) {
taosMemoryFree(p->pSchema);
}
taosMemoryFreeClear(p->pDataIter);
taosMemoryFree(p->pCurrSchema);
destroyLastBlockLoadInfo(p->pLoadInfo);

61
source/dnode/vnode/src/tsdb/tsdbDataIter.c
View File

@ -14,6 +14,7 @@
*/
#include "tsdb.h"
#include "vnodeInt.h"
// STsdbDataIter2
/* open */
@ -202,13 +203,6 @@ static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo*
for (;;) {
while (pIter->dIter.iRow < pIter->dIter.bData.nRow) {
if (pFilterInfo) {
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_TABLEID) {
if (pFilterInfo->tbid.uid == pIter->dIter.bData.uid) {
pIter->dIter.iRow = pIter->dIter.bData.nRow;
continue;
}
}
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) {
if (pIter->dIter.bData.aVersion[pIter->dIter.iRow] < pFilterInfo->sver ||
pIter->dIter.bData.aVersion[pIter->dIter.iRow] > pFilterInfo->ever) {
@ -232,13 +226,6 @@ static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo*
// filter
if (pFilterInfo) {
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_TABLEID) {
if (tTABLEIDCmprFn(&pFilterInfo->tbid, &pIter->rowInfo) == 0) {
pIter->dIter.iDataBlk = pIter->dIter.mDataBlk.nItem;
continue;
}
}
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) {
if (pFilterInfo->sver > dataBlk.maxVer || pFilterInfo->ever < dataBlk.minVer) {
pIter->dIter.iDataBlk++;
@ -262,13 +249,23 @@ static int32_t tsdbDataFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo*
if (pIter->dIter.iBlockIdx < taosArrayGetSize(pIter->dIter.aBlockIdx)) {
SBlockIdx* pBlockIdx = taosArrayGet(pIter->dIter.aBlockIdx, pIter->dIter.iBlockIdx);
if (pFilterInfo && (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_TABLEID)) {
int32_t c = tTABLEIDCmprFn(pBlockIdx, &pFilterInfo->tbid);
if (c == 0) {
pIter->dIter.iBlockIdx++;
continue;
} else if (c < 0) {
ASSERT(0);
if (pFilterInfo) {
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_TABLEID) {
int32_t c = tTABLEIDCmprFn(pBlockIdx, &pFilterInfo->tbid);
if (c == 0) {
pIter->dIter.iBlockIdx++;
continue;
} else if (c < 0) {
ASSERT(0);
}
}
if (pFilterInfo->flag & TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE) {
SMetaInfo info;
if (metaGetInfo(pIter->dIter.pReader->pTsdb->pVnode->pMeta, pBlockIdx->uid, &info, NULL)) {
pIter->dIter.iBlockIdx++;
continue;
}
}
}
@ -304,14 +301,24 @@ static int32_t tsdbSttFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo* p
for (;;) {
while (pIter->sIter.iRow < pIter->sIter.bData.nRow) {
if (pFilterInfo) {
int64_t uid = pIter->sIter.bData.uid ? pIter->sIter.bData.uid : pIter->sIter.bData.aUid[pIter->sIter.iRow];
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_TABLEID) {
int64_t uid = pIter->sIter.bData.uid ? pIter->sIter.bData.uid : pIter->sIter.bData.aUid[pIter->sIter.iRow];
if (pFilterInfo->tbid.uid == uid) {
pIter->sIter.iRow++;
continue;
}
}
if (pFilterInfo->flag & TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE) {
if (pIter->rowInfo.uid != uid) {
SMetaInfo info;
if (metaGetInfo(pIter->sIter.pReader->pTsdb->pVnode->pMeta, uid, &info, NULL)) {
pIter->sIter.iRow++;
continue;
}
}
}
if (pFilterInfo->flag & TSDB_FILTER_FLAG_BY_VERSION) {
if (pFilterInfo->sver > pIter->sIter.bData.aVersion[pIter->sIter.iRow] ||
pFilterInfo->ever < pIter->sIter.bData.aVersion[pIter->sIter.iRow]) {
@ -395,6 +402,16 @@ static int32_t tsdbTombFileDataIterNext(STsdbDataIter2* pIter, STsdbFilterInfo*
if (pIter->tIter.iDelIdx < taosArrayGetSize(pIter->tIter.aDelIdx)) {
SDelIdx* pDelIdx = taosArrayGet(pIter->tIter.aDelIdx, pIter->tIter.iDelIdx);
if (pFilterInfo) {
if (pFilterInfo->flag & TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE) {
SMetaInfo info;
if (metaGetInfo(pIter->dIter.pReader->pTsdb->pVnode->pMeta, pDelIdx->uid, &info, NULL)) {
pIter->tIter.iDelIdx++;
continue;
}
}
}
code = tsdbReadDelData(pIter->tIter.pReader, pDelIdx, pIter->tIter.aDelData);
TSDB_CHECK_CODE(code, lino, _exit);

5
source/dnode/vnode/src/tsdb/tsdbMemTable.c
View File

@ -13,6 +13,7 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include <util/tsimplehash.h>
#include "tsdb.h"
#define MEM_MIN_HASH 1024
@ -298,12 +299,12 @@ int64_t tsdbCountTbDataRows(STbData *pTbData) {
return rowsNum;
}
void tsdbMemTableCountRows(SMemTable *pMemTable, SHashObj* pTableMap, int64_t *rowsNum) {
void tsdbMemTableCountRows(SMemTable *pMemTable, SSHashObj* pTableMap, int64_t *rowsNum) {
taosRLockLatch(&pMemTable->latch);
for (int32_t i = 0; i < pMemTable->nBucket; ++i) {
STbData *pTbData = pMemTable->aBucket[i];
while (pTbData) {
void* p = taosHashGet(pTableMap, &pTbData->uid, sizeof(pTbData->uid));
void* p = tSimpleHashGet(pTableMap, &pTbData->uid, sizeof(pTbData->uid));
if (p == NULL) {
pTbData = pTbData->next;
continue;

105
source/dnode/vnode/src/tsdb/tsdbMergeTree.c
View File

@ -16,22 +16,6 @@
#include "tsdb.h"
// SLDataIter =================================================
struct SLDataIter {
SRBTreeNode node;
SSttBlk *pSttBlk;
SDataFReader *pReader;
int32_t iStt;
int8_t backward;
int32_t iSttBlk;
int32_t iRow;
SRowInfo rInfo;
uint64_t uid;
STimeWindow timeWindow;
SVersionRange verRange;
SSttBlockLoadInfo *pBlockLoadInfo;
bool ignoreEarlierTs;
};
SSttBlockLoadInfo *tCreateLastBlockLoadInfo(STSchema *pSchema, int16_t *colList, int32_t numOfCols,
int32_t numOfSttTrigger) {
SSttBlockLoadInfo *pLoadInfo = taosMemoryCalloc(numOfSttTrigger, sizeof(SSttBlockLoadInfo));
@ -268,25 +252,21 @@ static int32_t binarySearchForStartRowIndex(uint64_t *uidList, int32_t num, uint
}
}
int32_t tLDataIterOpen(struct SLDataIter **pIter, SDataFReader *pReader, int32_t iStt, int8_t backward, uint64_t suid,
int32_t tLDataIterOpen(struct SLDataIter *pIter, SDataFReader *pReader, int32_t iStt, int8_t backward, uint64_t suid,
uint64_t uid, STimeWindow *pTimeWindow, SVersionRange *pRange, SSttBlockLoadInfo *pBlockLoadInfo,
const char *idStr, bool strictTimeRange) {
int32_t code = TSDB_CODE_SUCCESS;
*pIter = taosMemoryCalloc(1, sizeof(SLDataIter));
if (*pIter == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
}
pIter->uid = uid;
pIter->pReader = pReader;
pIter->iStt = iStt;
pIter->backward = backward;
pIter->verRange.minVer = pRange->minVer;
pIter->verRange.maxVer = pRange->maxVer;
pIter->timeWindow.skey = pTimeWindow->skey;
pIter->timeWindow.ekey = pTimeWindow->ekey;
(*pIter)->uid = uid;
(*pIter)->pReader = pReader;
(*pIter)->iStt = iStt;
(*pIter)->backward = backward;
(*pIter)->verRange = *pRange;
(*pIter)->timeWindow = *pTimeWindow;
(*pIter)->pBlockLoadInfo = pBlockLoadInfo;
pIter->pBlockLoadInfo = pBlockLoadInfo;
if (!pBlockLoadInfo->sttBlockLoaded) {
int64_t st = taosGetTimestampUs();
@ -294,7 +274,7 @@ int32_t tLDataIterOpen(struct SLDataIter **pIter, SDataFReader *pReader, int32_t
code = tsdbReadSttBlk(pReader, iStt, pBlockLoadInfo->aSttBlk);
if (code) {
goto _exit;
return code;
}
// only apply to the child tables, ordinary tables will not incur this filter procedure.
@ -310,7 +290,7 @@ int32_t tLDataIterOpen(struct SLDataIter **pIter, SDataFReader *pReader, int32_t
// no qualified stt block existed
taosArrayClear(pBlockLoadInfo->aSttBlk);
(*pIter)->iSttBlk = -1;
pIter->iSttBlk = -1;
double el = (taosGetTimestampUs() - st) / 1000.0;
tsdbDebug("load the last file info completed, elapsed time:%.2fms, %s", el, idStr);
return code;
@ -343,31 +323,27 @@ int32_t tLDataIterOpen(struct SLDataIter **pIter, SDataFReader *pReader, int32_t
size_t size = taosArrayGetSize(pBlockLoadInfo->aSttBlk);
// find the start block
(*pIter)->iSttBlk = binarySearchForStartBlock(pBlockLoadInfo->aSttBlk->pData, size, uid, backward);
if ((*pIter)->iSttBlk != -1) {
(*pIter)->pSttBlk = taosArrayGet(pBlockLoadInfo->aSttBlk, (*pIter)->iSttBlk);
(*pIter)->iRow = ((*pIter)->backward) ? (*pIter)->pSttBlk->nRow : -1;
pIter->iSttBlk = binarySearchForStartBlock(pBlockLoadInfo->aSttBlk->pData, size, uid, backward);
if (pIter->iSttBlk != -1) {
pIter->pSttBlk = taosArrayGet(pBlockLoadInfo->aSttBlk, pIter->iSttBlk);
pIter->iRow = (pIter->backward) ? pIter->pSttBlk->nRow : -1;
if ((!backward) && ((strictTimeRange && (*pIter)->pSttBlk->minKey >= (*pIter)->timeWindow.ekey) ||
(!strictTimeRange && (*pIter)->pSttBlk->minKey > (*pIter)->timeWindow.ekey))) {
(*pIter)->pSttBlk = NULL;
if ((!backward) && ((strictTimeRange && pIter->pSttBlk->minKey >= pIter->timeWindow.ekey) ||
(!strictTimeRange && pIter->pSttBlk->minKey > pIter->timeWindow.ekey))) {
pIter->pSttBlk = NULL;
}
if (backward && ((strictTimeRange && (*pIter)->pSttBlk->maxKey <= (*pIter)->timeWindow.skey) ||
(!strictTimeRange && (*pIter)->pSttBlk->maxKey < (*pIter)->timeWindow.skey))) {
(*pIter)->pSttBlk = NULL;
(*pIter)->ignoreEarlierTs = true;
if (backward && ((strictTimeRange && pIter->pSttBlk->maxKey <= pIter->timeWindow.skey) ||
(!strictTimeRange && pIter->pSttBlk->maxKey < pIter->timeWindow.skey))) {
pIter->pSttBlk = NULL;
pIter->ignoreEarlierTs = true;
}
}
return code;
_exit:
taosMemoryFree(*pIter);
return code;
}
void tLDataIterClose(SLDataIter *pIter) { taosMemoryFree(pIter); }
void tLDataIterClose(SLDataIter *pIter) { /*taosMemoryFree(pIter); */}
void tLDataIterNextBlock(SLDataIter *pIter, const char *idStr) {
int32_t step = pIter->backward ? -1 : 1;
@ -594,43 +570,38 @@ static FORCE_INLINE int32_t tLDataIterDescCmprFn(const SRBTreeNode *p1, const SR
int32_t tMergeTreeOpen(SMergeTree *pMTree, int8_t backward, SDataFReader *pFReader, uint64_t suid, uint64_t uid,
STimeWindow *pTimeWindow, SVersionRange *pVerRange, SSttBlockLoadInfo *pBlockLoadInfo,
bool destroyLoadInfo, const char *idStr, bool strictTimeRange) {
bool destroyLoadInfo, const char *idStr, bool strictTimeRange, SLDataIter* pLDataIter) {
int32_t code = TSDB_CODE_SUCCESS;
pMTree->backward = backward;
pMTree->pIter = NULL;
pMTree->pIterList = taosArrayInit(4, POINTER_BYTES);
if (pMTree->pIterList == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
}
pMTree->idStr = idStr;
if (!pMTree->backward) { // asc
tRBTreeCreate(&pMTree->rbt, tLDataIterCmprFn);
} else { // desc
tRBTreeCreate(&pMTree->rbt, tLDataIterDescCmprFn);
}
int32_t code = TSDB_CODE_SUCCESS;
pMTree->pLoadInfo = pBlockLoadInfo;
pMTree->destroyLoadInfo = destroyLoadInfo;
pMTree->ignoreEarlierTs = false;
for (int32_t i = 0; i < pFReader->pSet->nSttF; ++i) { // open all last file
struct SLDataIter *pIter = NULL;
code = tLDataIterOpen(&pIter, pFReader, i, pMTree->backward, suid, uid, pTimeWindow, pVerRange,
memset(&pLDataIter[i], 0, sizeof(SLDataIter));
code = tLDataIterOpen(&pLDataIter[i], pFReader, i, pMTree->backward, suid, uid, pTimeWindow, pVerRange,
&pMTree->pLoadInfo[i], pMTree->idStr, strictTimeRange);
if (code != TSDB_CODE_SUCCESS) {
goto _end;
}
bool hasVal = tLDataIterNextRow(pIter, pMTree->idStr);
bool hasVal = tLDataIterNextRow(&pLDataIter[i], pMTree->idStr);
if (hasVal) {
taosArrayPush(pMTree->pIterList, &pIter);
tMergeTreeAddIter(pMTree, pIter);
tMergeTreeAddIter(pMTree, &pLDataIter[i]);
} else {
if (!pMTree->ignoreEarlierTs) {
pMTree->ignoreEarlierTs = pIter->ignoreEarlierTs;
pMTree->ignoreEarlierTs = pLDataIter[i].ignoreEarlierTs;
}
tLDataIterClose(pIter);
}
}
@ -678,18 +649,8 @@ bool tMergeTreeNext(SMergeTree *pMTree) {
return pMTree->pIter != NULL;
}
TSDBROW tMergeTreeGetRow(SMergeTree *pMTree) { return pMTree->pIter->rInfo.row; }
void tMergeTreeClose(SMergeTree *pMTree) {
size_t size = taosArrayGetSize(pMTree->pIterList);
for (int32_t i = 0; i < size; ++i) {
SLDataIter *pIter = taosArrayGetP(pMTree->pIterList, i);
tLDataIterClose(pIter);
}
pMTree->pIterList = taosArrayDestroy(pMTree->pIterList);
pMTree->pIter = NULL;
if (pMTree->destroyLoadInfo) {
pMTree->pLoadInfo = destroyLastBlockLoadInfo(pMTree->pLoadInfo);
pMTree->destroyLoadInfo = false;

1006
source/dnode/vnode/src/tsdb/tsdbRead.c
View File

File diff suppressed because it is too large Load Diff

26
source/dnode/vnode/src/tsdb/tsdbSnapshot.c
View File

@ -70,10 +70,11 @@ static int32_t tsdbSnapReadFileDataStart(STsdbSnapReader* pReader) {
if (pReader->pIter) {
// iter to next with filter info (sver, ever)
code = tsdbDataIterNext2(pReader->pIter,
&(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, // flag
.sver = pReader->sver,
.ever = pReader->ever});
code = tsdbDataIterNext2(
pReader->pIter,
&(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION | TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE, // flag
.sver = pReader->sver,
.ever = pReader->ever});
TSDB_CHECK_CODE(code, lino, _exit);
if (pReader->pIter->rowInfo.suid || pReader->pIter->rowInfo.uid) {
@ -94,10 +95,11 @@ static int32_t tsdbSnapReadFileDataStart(STsdbSnapReader* pReader) {
if (pReader->pIter) {
// iter to valid row
code = tsdbDataIterNext2(pReader->pIter,
&(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, // flag
.sver = pReader->sver,
.ever = pReader->ever});
code = tsdbDataIterNext2(
pReader->pIter,
&(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION | TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE, // flag
.sver = pReader->sver,
.ever = pReader->ever});
TSDB_CHECK_CODE(code, lino, _exit);
if (pReader->pIter->rowInfo.suid || pReader->pIter->rowInfo.uid) {
@ -139,7 +141,8 @@ static int32_t tsdbSnapReadNextRow(STsdbSnapReader* pReader, SRowInfo** ppRowInf
int32_t lino = 0;
if (pReader->pIter) {
code = tsdbDataIterNext2(pReader->pIter, &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, // flag
code = tsdbDataIterNext2(pReader->pIter, &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION |
TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE, // flag
.sver = pReader->sver,
.ever = pReader->ever});
TSDB_CHECK_CODE(code, lino, _exit);
@ -346,8 +349,9 @@ static int32_t tsdbSnapReadNextTombData(STsdbSnapReader* pReader, SDelInfo** ppD
int32_t lino = 0;
code = tsdbDataIterNext2(
pReader->pTIter,
&(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION, .sver = pReader->sver, .ever = pReader->ever});
pReader->pTIter, &(STsdbFilterInfo){.flag = TSDB_FILTER_FLAG_BY_VERSION | TSDB_FILTER_FLAG_IGNORE_DROPPED_TABLE,
.sver = pReader->sver,
.ever = pReader->ever});
TSDB_CHECK_CODE(code, lino, _exit);
if (ppDelInfo) {

389
source/dnode/vnode/src/tsdb/tsdbUtil.c
View File

@ -637,201 +637,157 @@ SColVal *tsdbRowIterNext(STSDBRowIter *pIter) {
}
// SRowMerger ======================================================
int32_t tsdbRowMergerInit(SRowMerger *pMerger, STSchema *pResTSchema, TSDBROW *pRow, STSchema *pTSchema) {
int32_t code = 0;
TSDBKEY key = TSDBROW_KEY(pRow);
SColVal *pColVal = &(SColVal){0};
STColumn *pTColumn;
int32_t iCol, jCol = 0;
if (NULL == pResTSchema) {
pResTSchema = pTSchema;
}
pMerger->pTSchema = pResTSchema;
pMerger->version = key.version;
pMerger->pArray = taosArrayInit(pResTSchema->numOfCols, sizeof(SColVal));
if (pMerger->pArray == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
}
// ts
pTColumn = &pTSchema->columns[jCol++];
ASSERT(pTColumn->type == TSDB_DATA_TYPE_TIMESTAMP);
*pColVal = COL_VAL_VALUE(pTColumn->colId, pTColumn->type, (SValue){.val = key.ts});
if (taosArrayPush(pMerger->pArray, pColVal) == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
}
// other
for (iCol = 1; jCol < pTSchema->numOfCols && iCol < pResTSchema->numOfCols; ++iCol) {
pTColumn = &pResTSchema->columns[iCol];
if (pTSchema->columns[jCol].colId < pTColumn->colId) {
++jCol;
--iCol;
continue;
} else if (pTSchema->columns[jCol].colId > pTColumn->colId) {
taosArrayPush(pMerger->pArray, &COL_VAL_NONE(pTColumn->colId, pTColumn->type));
continue;
}
tsdbRowGetColVal(pRow, pTSchema, jCol++, pColVal);
if ((!COL_VAL_IS_NONE(pColVal)) && (!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) {
uint8_t *pVal = pColVal->value.pData;
pColVal->value.pData = NULL;
code = tRealloc(&pColVal->value.pData, pColVal->value.nData);
if (code) goto _exit;
if (pColVal->value.nData) {
memcpy(pColVal->value.pData, pVal, pColVal->value.nData);
}
}
if (taosArrayPush(pMerger->pArray, pColVal) == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
}
}
for (; iCol < pResTSchema->numOfCols; ++iCol) {
pTColumn = &pResTSchema->columns[iCol];
taosArrayPush(pMerger->pArray, &COL_VAL_NONE(pTColumn->colId, pTColumn->type));
}
_exit:
return code;
}
int32_t tsdbRowMergerAdd(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) {
int32_t code = 0;
TSDBKEY key = TSDBROW_KEY(pRow);
SColVal *pColVal = &(SColVal){0};
SColVal * pColVal = &(SColVal){0};
STColumn *pTColumn;
int32_t iCol, jCol = 1;
if (NULL == pTSchema) {
pTSchema = pMerger->pTSchema;
}
ASSERT(((SColVal *)pMerger->pArray->pData)->value.val == key.ts);
for (iCol = 1; iCol < pMerger->pTSchema->numOfCols && jCol < pTSchema->numOfCols; ++iCol) {
pTColumn = &pMerger->pTSchema->columns[iCol];
if (pTSchema->columns[jCol].colId < pTColumn->colId) {
++jCol;
--iCol;
continue;
} else if (pTSchema->columns[jCol].colId > pTColumn->colId) {
continue;
}
if (taosArrayGetSize(pMerger->pArray) == 0) {
// ts
jCol = 0;
pTColumn = &pTSchema->columns[jCol++];
tsdbRowGetColVal(pRow, pTSchema, jCol++, pColVal);
if (key.version > pMerger->version) {
if (!COL_VAL_IS_NONE(pColVal)) {
if (IS_VAR_DATA_TYPE(pColVal->type)) {
SColVal *pTColVal = taosArrayGet(pMerger->pArray, iCol);
if (!COL_VAL_IS_NULL(pColVal)) {
code = tRealloc(&pTColVal->value.pData, pColVal->value.nData);
if (code) return code;
pTColVal->value.nData = pColVal->value.nData;
if (pTColVal->value.nData) {
memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData);
}
pTColVal->flag = 0;
} else {
tFree(pTColVal->value.pData);
taosArraySet(pMerger->pArray, iCol, pColVal);
}
} else {
taosArraySet(pMerger->pArray, iCol, pColVal);
}
}
} else if (key.version < pMerger->version) {
SColVal *tColVal = (SColVal *)taosArrayGet(pMerger->pArray, iCol);
if (COL_VAL_IS_NONE(tColVal) && !COL_VAL_IS_NONE(pColVal)) {
if ((!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) {
code = tRealloc(&tColVal->value.pData, pColVal->value.nData);
if (code) return code;
tColVal->value.nData = pColVal->value.nData;
if (pColVal->value.nData) {
memcpy(tColVal->value.pData, pColVal->value.pData, pColVal->value.nData);
}
tColVal->flag = 0;
} else {
taosArraySet(pMerger->pArray, iCol, pColVal);
}
}
} else {
ASSERT(0 && "dup versions not allowed");
}
}
pMerger->version = key.version;
return code;
}
/*
int32_t tsdbRowMergerInit(SRowMerger *pMerger, TSDBROW *pRow, STSchema *pTSchema) {
int32_t code = 0;
TSDBKEY key = TSDBROW_KEY(pRow);
SColVal *pColVal = &(SColVal){0};
STColumn *pTColumn;
pMerger->pTSchema = pTSchema;
pMerger->version = key.version;
pMerger->pArray = taosArrayInit(pTSchema->numOfCols, sizeof(SColVal));
if (pMerger->pArray == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
}
// ts
pTColumn = &pTSchema->columns[0];
ASSERT(pTColumn->type == TSDB_DATA_TYPE_TIMESTAMP);
*pColVal = COL_VAL_VALUE(pTColumn->colId, pTColumn->type, (SValue){.val = key.ts});
if (taosArrayPush(pMerger->pArray, pColVal) == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
}
// other
for (int16_t iCol = 1; iCol < pTSchema->numOfCols; iCol++) {
tsdbRowGetColVal(pRow, pTSchema, iCol, pColVal);
if ((!COL_VAL_IS_NONE(pColVal)) && (!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) {
uint8_t *pVal = pColVal->value.pData;
pColVal->value.pData = NULL;
code = tRealloc(&pColVal->value.pData, pColVal->value.nData);
if (code) goto _exit;
if (pColVal->value.nData) {
memcpy(pColVal->value.pData, pVal, pColVal->value.nData);
}
}
ASSERT(pTColumn->type == TSDB_DATA_TYPE_TIMESTAMP);
*pColVal = COL_VAL_VALUE(pTColumn->colId, pTColumn->type, (SValue){.val = key.ts});
if (taosArrayPush(pMerger->pArray, pColVal) == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
goto _exit;
return code;
// goto _exit;
}
// other
for (iCol = 1; jCol < pTSchema->numOfCols && iCol < pMerger->pTSchema->numOfCols; ++iCol) {
pTColumn = &pMerger->pTSchema->columns[iCol];
if (pTSchema->columns[jCol].colId < pTColumn->colId) {
++jCol;
--iCol;
continue;
} else if (pTSchema->columns[jCol].colId > pTColumn->colId) {
taosArrayPush(pMerger->pArray, &COL_VAL_NONE(pTColumn->colId, pTColumn->type));
continue;
}
tsdbRowGetColVal(pRow, pTSchema, jCol++, pColVal);
if ((!COL_VAL_IS_NONE(pColVal)) && (!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) {
uint8_t *pVal = pColVal->value.pData;
pColVal->value.pData = NULL;
code = tRealloc(&pColVal->value.pData, pColVal->value.nData);
if (code) {
return TSDB_CODE_OUT_OF_MEMORY;
}
if (pColVal->value.nData) {
memcpy(pColVal->value.pData, pVal, pColVal->value.nData);
}
}
if (taosArrayPush(pMerger->pArray, pColVal) == NULL) {
code = TSDB_CODE_OUT_OF_MEMORY;
return code;
}
}
for (; iCol < pMerger->pTSchema->numOfCols; ++iCol) {
pTColumn = &pMerger->pTSchema->columns[iCol];
taosArrayPush(pMerger->pArray, &COL_VAL_NONE(pTColumn->colId, pTColumn->type));
}
pMerger->version = key.version;
return 0;
} else {
ASSERT(((SColVal *)pMerger->pArray->pData)->value.val == key.ts);
for (iCol = 1; iCol < pMerger->pTSchema->numOfCols && jCol < pTSchema->numOfCols; ++iCol) {
pTColumn = &pMerger->pTSchema->columns[iCol];
if (pTSchema->columns[jCol].colId < pTColumn->colId) {
++jCol;
--iCol;
continue;
} else if (pTSchema->columns[jCol].colId > pTColumn->colId) {
continue;
}
tsdbRowGetColVal(pRow, pTSchema, jCol++, pColVal);
if (key.version > pMerger->version) {
if (!COL_VAL_IS_NONE(pColVal)) {
if (IS_VAR_DATA_TYPE(pColVal->type)) {
SColVal *pTColVal = taosArrayGet(pMerger->pArray, iCol);
if (!COL_VAL_IS_NULL(pColVal)) {
code = tRealloc(&pTColVal->value.pData, pColVal->value.nData);
if (code) return code;
pTColVal->value.nData = pColVal->value.nData;
if (pTColVal->value.nData) {
memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData);
}
pTColVal->flag = 0;
} else {
tFree(pTColVal->value.pData);
taosArraySet(pMerger->pArray, iCol, pColVal);
}
} else {
taosArraySet(pMerger->pArray, iCol, pColVal);
}
}
} else if (key.version < pMerger->version) {
SColVal *tColVal = (SColVal *)taosArrayGet(pMerger->pArray, iCol);
if (COL_VAL_IS_NONE(tColVal) && !COL_VAL_IS_NONE(pColVal)) {
if ((!COL_VAL_IS_NULL(pColVal)) && IS_VAR_DATA_TYPE(pColVal->type)) {
code = tRealloc(&tColVal->value.pData, pColVal->value.nData);
if (code) return code;
tColVal->value.nData = pColVal->value.nData;
if (pColVal->value.nData) {
memcpy(tColVal->value.pData, pColVal->value.pData, pColVal->value.nData);
}
tColVal->flag = 0;
} else {
taosArraySet(pMerger->pArray, iCol, pColVal);
}
}
} else {
ASSERT(0 && "dup versions not allowed");
}
}
pMerger->version = key.version;
return code;
}
}
int32_t tsdbRowMergerInit(SRowMerger* pMerger, STSchema *pSchema) {
pMerger->pTSchema = pSchema;
pMerger->pArray = taosArrayInit(pSchema->numOfCols, sizeof(SColVal));
if (pMerger->pArray == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
} else {
return TSDB_CODE_SUCCESS;
}
}
void tsdbRowMergerClear(SRowMerger* pMerger) {
for (int32_t iCol = 1; iCol < pMerger->pTSchema->numOfCols; iCol++) {
SColVal *pTColVal = taosArrayGet(pMerger->pArray, iCol);
if (IS_VAR_DATA_TYPE(pTColVal->type)) {
tFree(pTColVal->value.pData);
}
}
_exit:
return code;
taosArrayClear(pMerger->pArray);
}
*/
void tsdbRowMergerClear(SRowMerger *pMerger) {
for (int32_t iCol = 1; iCol < pMerger->pTSchema->numOfCols; iCol++) {
void tsdbRowMergerCleanup(SRowMerger* pMerger) {
int32_t numOfCols = taosArrayGetSize(pMerger->pArray);
for (int32_t iCol = 1; iCol < numOfCols; iCol++) {
SColVal *pTColVal = taosArrayGet(pMerger->pArray, iCol);
if (IS_VAR_DATA_TYPE(pTColVal->type)) {
tFree(pTColVal->value.pData);
@ -840,72 +796,7 @@ void tsdbRowMergerClear(SRowMerger *pMerger) {
taosArrayDestroy(pMerger->pArray);
}
/*
int32_t tsdbRowMerge(SRowMerger *pMerger, TSDBROW *pRow) {
int32_t code = 0;
TSDBKEY key = TSDBROW_KEY(pRow);
SColVal *pColVal = &(SColVal){0};
ASSERT(((SColVal *)pMerger->pArray->pData)->value.val == key.ts);
for (int32_t iCol = 1; iCol < pMerger->pTSchema->numOfCols; iCol++) {
tsdbRowGetColVal(pRow, pMerger->pTSchema, iCol, pColVal);
if (key.version > pMerger->version) {
if (!COL_VAL_IS_NONE(pColVal)) {
if (IS_VAR_DATA_TYPE(pColVal->type)) {
SColVal *pTColVal = taosArrayGet(pMerger->pArray, iCol);
if (!COL_VAL_IS_NULL(pColVal)) {
code = tRealloc(&pTColVal->value.pData, pColVal->value.nData);
if (code) goto _exit;
pTColVal->value.nData = pColVal->value.nData;
if (pTColVal->value.nData) {
memcpy(pTColVal->value.pData, pColVal->value.pData, pTColVal->value.nData);
}
pTColVal->flag = 0;
} else {
tFree(pTColVal->value.pData);
pTColVal->value.pData = NULL;
taosArraySet(pMerger->pArray, iCol, pColVal);
}
} else {
taosArraySet(pMerger->pArray, iCol, pColVal);
}
}
} else if (key.version < pMerger->version) {
SColVal *tColVal = (SColVal *)taosArrayGet(pMerger->pArray, iCol);
if (COL_VAL_IS_NONE(tColVal) && !COL_VAL_IS_NONE(pColVal)) {
if (IS_VAR_DATA_TYPE(pColVal->type)) {
if (!COL_VAL_IS_NULL(pColVal)) {
code = tRealloc(&tColVal->value.pData, pColVal->value.nData);
if (code) goto _exit;
tColVal->value.nData = pColVal->value.nData;
if (tColVal->value.nData) {
memcpy(tColVal->value.pData, pColVal->value.pData, tColVal->value.nData);
}
tColVal->flag = 0;
} else {
tFree(tColVal->value.pData);
tColVal->value.pData = NULL;
taosArraySet(pMerger->pArray, iCol, pColVal);
}
} else {
taosArraySet(pMerger->pArray, iCol, pColVal);
}
}
} else {
ASSERT(0);
}
}
pMerger->version = key.version;
_exit:
return code;
}
*/
int32_t tsdbRowMergerGetRow(SRowMerger *pMerger, SRow **ppRow) {
return tRowBuild(pMerger->pArray, pMerger->pTSchema, ppRow);
}
@ -1160,8 +1051,6 @@ int32_t tBlockDataCreate(SBlockData *pBlockData) {
pBlockData->aTSKEY = NULL;
pBlockData->nColData = 0;
pBlockData->aColData = NULL;
_exit:
return code;
}
@ -1218,9 +1107,10 @@ int32_t tBlockDataInit(SBlockData *pBlockData, TABLEID *pId, STSchema *pTSchema,
int32_t iColumn = 1;
STColumn *pTColumn = &pTSchema->columns[iColumn];
for (int32_t iCid = 0; iCid < nCid; iCid++) {
if (ASSERTS(pTColumn != NULL, "invalid input param")) {
code = TSDB_CODE_INVALID_PARA;
goto _exit;
// aCid array (from taos client catalog) contains columns that does not exist in the pTSchema. the pTSchema is newer
if (pTColumn == NULL) {
continue;
}
while (pTColumn->colId < aCid[iCid]) {
@ -1229,9 +1119,8 @@ int32_t tBlockDataInit(SBlockData *pBlockData, TABLEID *pId, STSchema *pTSchema,
pTColumn = &pTSchema->columns[iColumn];
}
if (ASSERTS(pTColumn->colId == aCid[iCid], "invalid input param")) {
code = TSDB_CODE_INVALID_PARA;
goto _exit;
if (pTColumn->colId != aCid[iCid]) {
continue;
}
tColDataInit(&pBlockData->aColData[iCid], pTColumn->colId, pTColumn->type,

32
source/dnode/vnode/src/vnd/vnodeSvr.c
View File

@ -448,7 +448,7 @@ int32_t vnodeProcessWriteMsg(SVnode *pVnode, SRpcMsg *pMsg, int64_t version, SRp
walApplyVer(pVnode->pWal, version);
if (tqPushMsg(pVnode->pTq, pMsg->pCont, pMsg->contLen, pMsg->msgType, version) < 0) {
/*vInfo("vgId:%d, push msg end", pVnode->config.vgId);*/
// /*vInfo("vgId:%d, push msg end", pVnode->config.vgId);*/
vError("vgId:%d, failed to push msg to TQ since %s", TD_VID(pVnode), tstrerror(terrno));
return -1;
}
@ -487,11 +487,16 @@ int32_t vnodePreprocessQueryMsg(SVnode *pVnode, SRpcMsg *pMsg) {
int32_t vnodeProcessQueryMsg(SVnode *pVnode, SRpcMsg *pMsg) {
vTrace("message in vnode query queue is processing");
if ((pMsg->msgType == TDMT_SCH_QUERY) && !syncIsReadyForRead(pVnode->sync)) {
if ((pMsg->msgType == TDMT_SCH_QUERY || pMsg->msgType == TDMT_VND_TMQ_CONSUME) && !syncIsReadyForRead(pVnode->sync)) {
vnodeRedirectRpcMsg(pVnode, pMsg, terrno);
return 0;
}
if (pMsg->msgType == TDMT_VND_TMQ_CONSUME && !pVnode->restored) {
vnodeRedirectRpcMsg(pVnode, pMsg, TSDB_CODE_SYN_RESTORING);
return 0;
}
SReadHandle handle = {.meta = pVnode->pMeta, .config = &pVnode->config, .vnode = pVnode, .pMsgCb = &pVnode->msgCb};
switch (pMsg->msgType) {
case TDMT_SCH_QUERY:
@ -499,6 +504,8 @@ int32_t vnodeProcessQueryMsg(SVnode *pVnode, SRpcMsg *pMsg) {
return qWorkerProcessQueryMsg(&handle, pVnode->pQuery, pMsg, 0);
case TDMT_SCH_QUERY_CONTINUE:
return qWorkerProcessCQueryMsg(&handle, pVnode->pQuery, pMsg, 0);
case TDMT_VND_TMQ_CONSUME:
return tqProcessPollReq(pVnode->pTq, pMsg);
default:
vError("unknown msg type:%d in query queue", pMsg->msgType);
return TSDB_CODE_APP_ERROR;
@ -508,17 +515,12 @@ int32_t vnodeProcessQueryMsg(SVnode *pVnode, SRpcMsg *pMsg) {
int32_t vnodeProcessFetchMsg(SVnode *pVnode, SRpcMsg *pMsg, SQueueInfo *pInfo) {
vTrace("vgId:%d, msg:%p in fetch queue is processing", pVnode->config.vgId, pMsg);
if ((pMsg->msgType == TDMT_SCH_FETCH || pMsg->msgType == TDMT_VND_TABLE_META || pMsg->msgType == TDMT_VND_TABLE_CFG ||
pMsg->msgType == TDMT_VND_BATCH_META || pMsg->msgType == TDMT_VND_TMQ_CONSUME) &&
pMsg->msgType == TDMT_VND_BATCH_META) &&
!syncIsReadyForRead(pVnode->sync)) {
vnodeRedirectRpcMsg(pVnode, pMsg, terrno);
return 0;
}
if (pMsg->msgType == TDMT_VND_TMQ_CONSUME && !pVnode->restored) {
vnodeRedirectRpcMsg(pVnode, pMsg, TSDB_CODE_SYN_RESTORING);
return 0;
}
switch (pMsg->msgType) {
case TDMT_SCH_FETCH:
case TDMT_SCH_MERGE_FETCH:
@ -537,8 +539,6 @@ int32_t vnodeProcessFetchMsg(SVnode *pVnode, SRpcMsg *pMsg, SQueueInfo *pInfo) {
return vnodeGetTableCfg(pVnode, pMsg, true);
case TDMT_VND_BATCH_META:
return vnodeGetBatchMeta(pVnode, pMsg);
case TDMT_VND_TMQ_CONSUME:
return tqProcessPollReq(pVnode->pTq, pMsg);
case TDMT_STREAM_TASK_RUN:
return tqProcessTaskRunReq(pVnode->pTq, pMsg);
case TDMT_STREAM_TASK_DISPATCH:
@ -1007,7 +1007,7 @@ static int32_t vnodeResetTableCxt(SMeta *pMeta, SSubmitReqConvertCxt *pCxt) {
}
tdSTSRowIterInit(&pCxt->rowIter, pCxt->pTbSchema);
tDestroySSubmitTbData(pCxt->pTbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(pCxt->pTbData, TSDB_MSG_FLG_ENCODE);
if (NULL == pCxt->pTbData) {
pCxt->pTbData = taosMemoryCalloc(1, sizeof(SSubmitTbData));
if (NULL == pCxt->pTbData) {
@ -1039,7 +1039,7 @@ static int32_t vnodeResetTableCxt(SMeta *pMeta, SSubmitReqConvertCxt *pCxt) {
static void vnodeDestroySubmitReqConvertCxt(SSubmitReqConvertCxt *pCxt) {
taosMemoryFreeClear(pCxt->pTbSchema);
tDestroySSubmitTbData(pCxt->pTbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(pCxt->pTbData, TSDB_MSG_FLG_ENCODE);
taosMemoryFreeClear(pCxt->pTbData);
taosArrayDestroy(pCxt->pColValues);
}
@ -1149,7 +1149,7 @@ static int32_t vnodeRebuildSubmitReqMsg(SSubmitReq2 *pSubmitReq, void **ppMsg) {
int32_t code = TSDB_CODE_SUCCESS;
char *pMsg = NULL;
uint32_t msglen = 0;
tEncodeSize(tEncodeSSubmitReq2, pSubmitReq, msglen, code);
tEncodeSize(tEncodeSubmitReq, pSubmitReq, msglen, code);
if (TSDB_CODE_SUCCESS == code) {
pMsg = taosMemoryMalloc(msglen);
if (NULL == pMsg) {
@ -1159,7 +1159,7 @@ static int32_t vnodeRebuildSubmitReqMsg(SSubmitReq2 *pSubmitReq, void **ppMsg) {
if (TSDB_CODE_SUCCESS == code) {
SEncoder encoder;
tEncoderInit(&encoder, pMsg, msglen);
code = tEncodeSSubmitReq2(&encoder, pSubmitReq);
code = tEncodeSubmitReq(&encoder, pSubmitReq);
tEncoderClear(&encoder);
}
if (TSDB_CODE_SUCCESS == code) {
@ -1199,7 +1199,7 @@ static int32_t vnodeProcessSubmitReq(SVnode *pVnode, int64_t version, void *pReq
len -= sizeof(SSubmitReq2Msg);
SDecoder dc = {0};
tDecoderInit(&dc, pReq, len);
if (tDecodeSSubmitReq2(&dc, pSubmitReq) < 0) {
if (tDecodeSubmitReq(&dc, pSubmitReq) < 0) {
code = TSDB_CODE_INVALID_MSG;
goto _exit;
}
@ -1388,7 +1388,7 @@ _exit:
// clear
taosArrayDestroy(newTbUids);
tDestroySSubmitReq2(pSubmitReq, 0 == pMsg->version ? TSDB_MSG_FLG_CMPT : TSDB_MSG_FLG_DECODE);
tDestroySubmitReq(pSubmitReq, 0 == pMsg->version ? TSDB_MSG_FLG_CMPT : TSDB_MSG_FLG_DECODE);
tDestroySSubmitRsp2(pSubmitRsp, TSDB_MSG_FLG_ENCODE);
if (code) terrno = code;

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

@ -555,7 +555,12 @@ static void vnodeRestoreFinish(const SSyncFSM *pFsm, const SyncIndex commitIdx)
vInfo("vgId:%d, sync restore finished, start to restore stream tasks by replay wal", pVnode->config.vgId);
// start to restore all stream tasks
tqStartStreamTasks(pVnode->pTq);
if (tsDisableStream) {
vInfo("vgId:%d, not restore stream tasks, since disabled", pVnode->config.vgId);
} else {
vInfo("vgId:%d start to restore stream tasks", pVnode->config.vgId);
tqStartStreamTasks(pVnode->pTq);
}
}
static void vnodeBecomeFollower(const SSyncFSM *pFsm) {

19
source/libs/catalog/inc/catalogInt.h
View File

@ -58,6 +58,7 @@ typedef enum {
CTG_CI_OTHERTABLE_META,
CTG_CI_TBL_SMA,
CTG_CI_TBL_CFG,
CTG_CI_TBL_TAG,
CTG_CI_INDEX_INFO,
CTG_CI_USER,
CTG_CI_UDF,
@ -110,6 +111,7 @@ typedef enum {
CTG_TASK_GET_SVR_VER,
CTG_TASK_GET_TB_META_BATCH,
CTG_TASK_GET_TB_HASH_BATCH,
CTG_TASK_GET_TB_TAG,
} CTG_TASK_TYPE;
typedef enum {
@ -152,6 +154,11 @@ typedef struct SCtgTbCacheInfo {
int32_t tbType;
} SCtgTbCacheInfo;
typedef struct SCtgTbMetaParam {
SName* pName;
int32_t flag;
} SCtgTbMetaParam;
typedef struct SCtgTbMetaCtx {
SCtgTbCacheInfo tbInfo;
int32_t vgId;
@ -186,6 +193,11 @@ typedef struct SCtgTbCfgCtx {
SVgroupInfo* pVgInfo;
} SCtgTbCfgCtx;
typedef struct SCtgTbTagCtx {
SName* pName;
SVgroupInfo* pVgInfo;
} SCtgTbTagCtx;
typedef struct SCtgDbVgCtx {
char dbFName[TSDB_DB_FNAME_LEN];
} SCtgDbVgCtx;
@ -304,6 +316,7 @@ typedef struct SCtgJob {
catalogCallback userFp;
int32_t tbMetaNum;
int32_t tbHashNum;
int32_t tbTagNum;
int32_t dbVgNum;
int32_t udfNum;
int32_t qnodeNum;
@ -346,6 +359,7 @@ typedef struct SCtgSubRes {
struct SCtgTask {
CTG_TASK_TYPE type;
bool subTask;
int32_t taskId;
SCtgJob* pJob;
void* taskCtx;
@ -623,6 +637,7 @@ typedef struct SCtgCacheItemInfo {
#define CTG_FLAG_SYS_DB 0x8
#define CTG_FLAG_FORCE_UPDATE 0x10
#define CTG_FLAG_ONLY_CACHE 0x20
#define CTG_FLAG_SYNC_OP 0x40
#define CTG_FLAG_SET(_flag, _v) ((_flag) |= (_v))
@ -925,6 +940,10 @@ void ctgReleaseVgMetaToCache(SCatalog* pCtg, SCtgDBCache* dbCache, SCtgTbCach
void ctgReleaseTbMetaToCache(SCatalog* pCtg, SCtgDBCache* dbCache, SCtgTbCache* pCache);
void ctgGetGlobalCacheStat(SCtgCacheStat* pStat);
int32_t ctgChkSetAuthRes(SCatalog* pCtg, SCtgAuthReq* req, SCtgAuthRsp* res);
int32_t ctgGetTbMeta(SCatalog* pCtg, SRequestConnInfo* pConn, SCtgTbMetaCtx* ctx, STableMeta** pTableMeta);
int32_t ctgGetCachedStbNameFromSuid(SCatalog* pCtg, char* dbFName, uint64_t suid, char **stbName);
int32_t ctgGetTbTagCb(SCtgTask* pTask);
int32_t ctgGetUserCb(SCtgTask* pTask);
extern SCatalogMgmt gCtgMgmt;
extern SCtgDebug gCTGDebug;

59
source/libs/catalog/src/catalog.c
View File

@ -208,7 +208,7 @@ int32_t ctgGetTbMeta(SCatalog* pCtg, SRequestConnInfo* pConn, SCtgTbMetaCtx* ctx
}
while (true) {
CTG_ERR_JRET(ctgRefreshTbMeta(pCtg, pConn, ctx, &output, false));
CTG_ERR_JRET(ctgRefreshTbMeta(pCtg, pConn, ctx, &output, ctx->flag & CTG_FLAG_SYNC_OP));
if (CTG_IS_META_TABLE(output->metaType)) {
*pTableMeta = output->tbMeta;
@ -429,6 +429,48 @@ int32_t ctgGetTbCfg(SCatalog* pCtg, SRequestConnInfo* pConn, SName* pTableName,
CTG_RET(TSDB_CODE_SUCCESS);
}
int32_t ctgGetTbTag(SCatalog* pCtg, SRequestConnInfo* pConn, SName* pTableName, SArray** pRes) {
SVgroupInfo vgroupInfo = {0};
STableCfg* pCfg = NULL;
int32_t code = 0;
CTG_ERR_RET(ctgGetTbHashVgroup(pCtg, pConn, pTableName, &vgroupInfo, NULL));
CTG_ERR_RET(ctgGetTableCfgFromVnode(pCtg, pConn, pTableName, &vgroupInfo, &pCfg, NULL));
if (NULL == pCfg->pTags || pCfg->tagsLen <= 0) {
ctgError("invalid tag in tbCfg rsp, pTags:%p, len:%d", pCfg->pTags, pCfg->tagsLen);
CTG_ERR_JRET(TSDB_CODE_INVALID_MSG);
}
SArray* pTagVals = NULL;
STag* pTag = (STag*)pCfg->pTags;
if (tTagIsJson(pTag)) {
pTagVals = taosArrayInit(1, sizeof(STagVal));
if (NULL == pTagVals) {
CTG_ERR_JRET(TSDB_CODE_OUT_OF_MEMORY);
}
char* pJson = parseTagDatatoJson(pTag);
STagVal tagVal;
tagVal.cid = 0;
tagVal.type = TSDB_DATA_TYPE_JSON;
tagVal.pData = pJson;
tagVal.nData = strlen(pJson);
taosArrayPush(pTagVals, &tagVal);
} else {
CTG_ERR_JRET(tTagToValArray((const STag*)pCfg->pTags, &pTagVals));
}
*pRes = pTagVals;
_return:
tFreeSTableCfgRsp((STableCfgRsp*)pCfg);
CTG_RET(code);
}
int32_t ctgGetTbDistVgInfo(SCatalog* pCtg, SRequestConnInfo* pConn, SName* pTableName, SArray** pVgList) {
STableMeta* tbMeta = NULL;
int32_t code = 0;
@ -1414,6 +1456,21 @@ _return:
CTG_API_LEAVE(code);
}
int32_t catalogGetTableTag(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName, SArray** pRes) {
CTG_API_ENTER();
if (NULL == pCtg || NULL == pConn || NULL == pTableName || NULL == pRes) {
CTG_API_LEAVE(TSDB_CODE_CTG_INVALID_INPUT);
}
int32_t code = 0;
CTG_ERR_JRET(ctgGetTbTag(pCtg, pConn, (SName*)pTableName, pRes));
_return:
CTG_API_LEAVE(code);
}
int32_t catalogRefreshGetTableCfg(SCatalog* pCtg, SRequestConnInfo* pConn, const SName* pTableName, STableCfg** pCfg) {
CTG_API_ENTER();

278
source/libs/catalog/src/ctgAsync.c
View File

@ -21,7 +21,8 @@
#include "trpc.h"
int32_t ctgInitGetTbMetaTask(SCtgJob* pJob, int32_t taskIdx, void* param) {
SName* name = (SName*)param;
SCtgTbMetaParam* pParam = (SCtgTbMetaParam*)param;
SName* name = pParam->pName;
SCtgTask task = {0};
task.type = CTG_TASK_GET_TB_META;
@ -41,7 +42,7 @@ int32_t ctgInitGetTbMetaTask(SCtgJob* pJob, int32_t taskIdx, void* param) {
}
memcpy(ctx->pName, name, sizeof(*name));
ctx->flag = CTG_FLAG_UNKNOWN_STB;
ctx->flag = pParam->flag | CTG_FLAG_UNKNOWN_STB;
taosArrayPush(pJob->pTasks, &task);
@ -386,6 +387,37 @@ int32_t ctgInitGetTbCfgTask(SCtgJob* pJob, int32_t taskIdx, void* param) {
return TSDB_CODE_SUCCESS;
}
int32_t ctgInitGetTbTagTask(SCtgJob* pJob, int32_t taskIdx, void* param) {
SName* name = (SName*)param;
SCtgTask task = {0};
task.type = CTG_TASK_GET_TB_TAG;
task.taskId = taskIdx;
task.pJob = pJob;
task.taskCtx = taosMemoryCalloc(1, sizeof(SCtgTbTagCtx));
if (NULL == task.taskCtx) {
CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
}
SCtgTbTagCtx* ctx = task.taskCtx;
ctx->pName = taosMemoryMalloc(sizeof(*name));
if (NULL == ctx->pName) {
taosMemoryFree(task.taskCtx);
CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
}
memcpy(ctx->pName, name, sizeof(*name));
taosArrayPush(pJob->pTasks, &task);
qDebug("QID:0x%" PRIx64 " the %dth task type %s initialized, tbName:%s", pJob->queryId, taskIdx,
ctgTaskTypeStr(task.type), name->tname);
return TSDB_CODE_SUCCESS;
}
int32_t ctgHandleForceUpdate(SCatalog* pCtg, int32_t taskNum, SCtgJob* pJob, const SCatalogReq* pReq) {
SHashObj* pDb = taosHashInit(taskNum, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
SHashObj* pTb = taosHashInit(taskNum, taosGetDefaultHashFunction(TSDB_DATA_TYPE_BINARY), false, HASH_NO_LOCK);
@ -437,6 +469,15 @@ int32_t ctgHandleForceUpdate(SCatalog* pCtg, int32_t taskNum, SCtgJob* pJob, con
char dbFName[TSDB_DB_FNAME_LEN];
tNameGetFullDbName(name, dbFName);
taosHashPut(pDb, dbFName, strlen(dbFName), dbFName, TSDB_DB_FNAME_LEN);
taosHashPut(pTb, name, sizeof(SName), name, sizeof(SName));
}
for (int32_t i = 0; i < pJob->tbTagNum; ++i) {
SName* name = taosArrayGet(pReq->pTableTag, i);
char dbFName[TSDB_DB_FNAME_LEN];
tNameGetFullDbName(name, dbFName);
taosHashPut(pDb, dbFName, strlen(dbFName), dbFName, TSDB_DB_FNAME_LEN);
taosHashPut(pTb, name, sizeof(SName), name, sizeof(SName));
}
char* dbFName = taosHashIterate(pDb, NULL);
@ -505,9 +546,10 @@ int32_t ctgInitJob(SCatalog* pCtg, SRequestConnInfo* pConn, SCtgJob** job, const
int32_t dbInfoNum = (int32_t)taosArrayGetSize(pReq->pDbInfo);
int32_t tbIndexNum = (int32_t)taosArrayGetSize(pReq->pTableIndex);
int32_t tbCfgNum = (int32_t)taosArrayGetSize(pReq->pTableCfg);
int32_t tbTagNum = (int32_t)taosArrayGetSize(pReq->pTableTag);
int32_t taskNum = tbMetaNum + dbVgNum + udfNum + tbHashNum + qnodeNum + dnodeNum + svrVerNum + dbCfgNum + indexNum +
userNum + dbInfoNum + tbIndexNum + tbCfgNum;
userNum + dbInfoNum + tbIndexNum + tbCfgNum + tbTagNum;
*job = taosMemoryCalloc(1, sizeof(SCtgJob));
if (NULL == *job) {
@ -537,6 +579,7 @@ int32_t ctgInitJob(SCatalog* pCtg, SRequestConnInfo* pConn, SCtgJob** job, const
pJob->tbIndexNum = tbIndexNum;
pJob->tbCfgNum = tbCfgNum;
pJob->svrVerNum = svrVerNum;
pJob->tbTagNum = tbTagNum;
#if CTG_BATCH_FETCH
pJob->pBatchs =
@ -604,6 +647,12 @@ int32_t ctgInitJob(SCatalog* pCtg, SRequestConnInfo* pConn, SCtgJob** job, const
CTG_ERR_JRET(ctgInitTask(pJob, CTG_TASK_GET_TB_CFG, name, NULL));
}
for (int32_t i = 0; i < tbTagNum; ++i) {
SName* name = taosArrayGet(pReq->pTableTag, i);
CTG_ERR_JRET(ctgInitTask(pJob, CTG_TASK_GET_TB_TAG, name, NULL));
}
for (int32_t i = 0; i < indexNum; ++i) {
char* indexName = taosArrayGet(pReq->pIndex, i);
CTG_ERR_JRET(ctgInitTask(pJob, CTG_TASK_GET_INDEX_INFO, indexName, NULL));
@ -650,6 +699,10 @@ _return:
}
int32_t ctgDumpTbMetaRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pTableMeta) {
pJob->jobRes.pTableMeta = taosArrayInit(pJob->tbMetaNum, sizeof(SMetaRes));
@ -665,6 +718,10 @@ int32_t ctgDumpTbMetaRes(SCtgTask* pTask) {
}
int32_t ctgDumpTbMetasRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
pJob->jobRes.pTableMeta = pTask->res;
@ -673,6 +730,10 @@ int32_t ctgDumpTbMetasRes(SCtgTask* pTask) {
}
int32_t ctgDumpDbVgRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pDbVgroup) {
pJob->jobRes.pDbVgroup = taosArrayInit(pJob->dbVgNum, sizeof(SMetaRes));
@ -688,6 +749,10 @@ int32_t ctgDumpDbVgRes(SCtgTask* pTask) {
}
int32_t ctgDumpTbHashRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pTableHash) {
pJob->jobRes.pTableHash = taosArrayInit(pJob->tbHashNum, sizeof(SMetaRes));
@ -703,6 +768,10 @@ int32_t ctgDumpTbHashRes(SCtgTask* pTask) {
}
int32_t ctgDumpTbHashsRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
pJob->jobRes.pTableHash = pTask->res;
@ -711,9 +780,17 @@ int32_t ctgDumpTbHashsRes(SCtgTask* pTask) {
}
int32_t ctgDumpTbIndexRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pTableIndex) {
pJob->jobRes.pTableIndex = taosArrayInit(pJob->tbIndexNum, sizeof(SMetaRes));
SArray* pRes = taosArrayInit(pJob->tbIndexNum, sizeof(SMetaRes));
if (atomic_val_compare_exchange_ptr(&pJob->jobRes.pTableIndex, NULL, pRes)) {
taosArrayDestroy(pRes);
}
if (NULL == pJob->jobRes.pTableIndex) {
CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
}
@ -726,9 +803,17 @@ int32_t ctgDumpTbIndexRes(SCtgTask* pTask) {
}
int32_t ctgDumpTbCfgRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pTableCfg) {
pJob->jobRes.pTableCfg = taosArrayInit(pJob->tbCfgNum, sizeof(SMetaRes));
SArray* pRes = taosArrayInit(pJob->tbCfgNum, sizeof(SMetaRes));
if (atomic_val_compare_exchange_ptr(&pJob->jobRes.pTableCfg, NULL, pRes)) {
taosArrayDestroy(pRes);
}
if (NULL == pJob->jobRes.pTableCfg) {
CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
}
@ -740,7 +825,35 @@ int32_t ctgDumpTbCfgRes(SCtgTask* pTask) {
return TSDB_CODE_SUCCESS;
}
int32_t ctgDumpTbTagRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pTableTag) {
SArray* pRes = taosArrayInit(pJob->tbTagNum, sizeof(SMetaRes));
if (atomic_val_compare_exchange_ptr(&pJob->jobRes.pTableTag, NULL, pRes)) {
taosArrayDestroy(pRes);
}
if (NULL == pJob->jobRes.pTableTag) {
CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
}
}
SMetaRes res = {.code = pTask->code, .pRes = pTask->res};
taosArrayPush(pJob->jobRes.pTableTag, &res);
return TSDB_CODE_SUCCESS;
}
int32_t ctgDumpIndexRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pIndex) {
pJob->jobRes.pIndex = taosArrayInit(pJob->indexNum, sizeof(SMetaRes));
@ -756,6 +869,10 @@ int32_t ctgDumpIndexRes(SCtgTask* pTask) {
}
int32_t ctgDumpQnodeRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pQnodeList) {
pJob->jobRes.pQnodeList = taosArrayInit(1, sizeof(SMetaRes));
@ -771,6 +888,10 @@ int32_t ctgDumpQnodeRes(SCtgTask* pTask) {
}
int32_t ctgDumpDnodeRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pDnodeList) {
pJob->jobRes.pDnodeList = taosArrayInit(1, sizeof(SMetaRes));
@ -786,6 +907,10 @@ int32_t ctgDumpDnodeRes(SCtgTask* pTask) {
}
int32_t ctgDumpDbCfgRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pDbCfg) {
pJob->jobRes.pDbCfg = taosArrayInit(pJob->dbCfgNum, sizeof(SMetaRes));
@ -801,6 +926,10 @@ int32_t ctgDumpDbCfgRes(SCtgTask* pTask) {
}
int32_t ctgDumpDbInfoRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pDbInfo) {
pJob->jobRes.pDbInfo = taosArrayInit(pJob->dbInfoNum, sizeof(SMetaRes));
@ -816,6 +945,10 @@ int32_t ctgDumpDbInfoRes(SCtgTask* pTask) {
}
int32_t ctgDumpUdfRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pUdfList) {
pJob->jobRes.pUdfList = taosArrayInit(pJob->udfNum, sizeof(SMetaRes));
@ -831,6 +964,10 @@ int32_t ctgDumpUdfRes(SCtgTask* pTask) {
}
int32_t ctgDumpUserRes(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pUser) {
pJob->jobRes.pUser = taosArrayInit(pJob->userNum, sizeof(SMetaRes));
@ -846,6 +983,10 @@ int32_t ctgDumpUserRes(SCtgTask* pTask) {
}
int32_t ctgDumpSvrVer(SCtgTask* pTask) {
if (pTask->subTask) {
return TSDB_CODE_SUCCESS;
}
SCtgJob* pJob = pTask->pJob;
if (NULL == pJob->jobRes.pSvrVer) {
pJob->jobRes.pSvrVer = taosMemoryCalloc(1, sizeof(SMetaRes));
@ -1075,7 +1216,7 @@ int32_t ctgHandleGetTbMetaRsp(SCtgTaskReq* tReq, int32_t reqType, const SDataBuf
STableMetaOutput* pOut = (STableMetaOutput*)pMsgCtx->out;
ctgUpdateTbMetaToCache(pCtg, pOut, false);
ctgUpdateTbMetaToCache(pCtg, pOut, flag & CTG_FLAG_SYNC_OP);
if (CTG_IS_META_BOTH(pOut->metaType)) {
memcpy(pOut->tbMeta, &pOut->ctbMeta, sizeof(pOut->ctbMeta));
@ -1473,6 +1614,49 @@ _return:
CTG_RET(code);
}
int32_t ctgHandleGetTbTagRsp(SCtgTaskReq* tReq, int32_t reqType, const SDataBuf* pMsg, int32_t rspCode) {
int32_t code = 0;
SCtgTask* pTask = tReq->pTask;
SCatalog* pCtg = pTask->pJob->pCtg;
CTG_ERR_JRET(ctgProcessRspMsg(&pTask->msgCtx.out, reqType, pMsg->pData, pMsg->len, rspCode, pTask->msgCtx.target));
STableCfgRsp* pRsp = (STableCfgRsp*)pTask->msgCtx.out;
if (NULL == pRsp->pTags || pRsp->tagsLen <= 0) {
ctgError("invalid tag in tbCfg rsp, pTags:%p, len:%d", pRsp->pTags, pRsp->tagsLen);
CTG_ERR_JRET(TSDB_CODE_INVALID_MSG);
}
SArray* pTagVals = NULL;
STag* pTag = (STag*)pRsp->pTags;
if (tTagIsJson(pTag)) {
pTagVals = taosArrayInit(1, sizeof(STagVal));
if (NULL == pTagVals) {
CTG_ERR_JRET(TSDB_CODE_OUT_OF_MEMORY);
}
char* pJson = parseTagDatatoJson(pTag);
STagVal tagVal;
tagVal.cid = 0;
tagVal.type = TSDB_DATA_TYPE_JSON;
tagVal.pData = pJson;
tagVal.nData = strlen(pJson);
taosArrayPush(pTagVals, &tagVal);
} else {
CTG_ERR_JRET(tTagToValArray((const STag*)pRsp->pTags, &pTagVals));
}
pTask->res = pTagVals;
_return:
ctgHandleTaskEnd(pTask, code);
CTG_RET(code);
}
int32_t ctgHandleGetDbCfgRsp(SCtgTaskReq* tReq, int32_t reqType, const SDataBuf* pMsg, int32_t rspCode) {
int32_t code = 0;
SCtgTask* pTask = tReq->pTask;
@ -1905,7 +2089,10 @@ int32_t ctgLaunchGetTbCfgTask(SCtgTask* pTask) {
if (pCtx->tbType <= 0) {
CTG_ERR_JRET(ctgReadTbTypeFromCache(pCtg, dbFName, pCtx->pName->tname, &pCtx->tbType));
if (pCtx->tbType <= 0) {
CTG_ERR_JRET(ctgLaunchSubTask(pTask, CTG_TASK_GET_TB_META, ctgGetTbCfgCb, pCtx->pName));
SCtgTbMetaParam param;
param.pName = pCtx->pName;
param.flag = 0;
CTG_ERR_JRET(ctgLaunchSubTask(pTask, CTG_TASK_GET_TB_META, ctgGetTbCfgCb, &param));
return TSDB_CODE_SUCCESS;
}
}
@ -1935,6 +2122,45 @@ _return:
CTG_RET(code);
}
int32_t ctgLaunchGetTbTagTask(SCtgTask* pTask) {
int32_t code = 0;
SCatalog* pCtg = pTask->pJob->pCtg;
SRequestConnInfo* pConn = &pTask->pJob->conn;
SCtgTbTagCtx* pCtx = (SCtgTbTagCtx*)pTask->taskCtx;
SArray* pRes = NULL;
char dbFName[TSDB_DB_FNAME_LEN];
tNameGetFullDbName(pCtx->pName, dbFName);
SCtgJob* pJob = pTask->pJob;
SCtgMsgCtx* pMsgCtx = CTG_GET_TASK_MSGCTX(pTask, -1);
if (NULL == pMsgCtx->pBatchs) {
pMsgCtx->pBatchs = pJob->pBatchs;
}
if (NULL == pCtx->pVgInfo) {
CTG_ERR_JRET(ctgGetTbHashVgroupFromCache(pCtg, pCtx->pName, &pCtx->pVgInfo));
if (NULL == pCtx->pVgInfo) {
CTG_ERR_JRET(ctgLaunchSubTask(pTask, CTG_TASK_GET_DB_VGROUP, ctgGetTbTagCb, dbFName));
return TSDB_CODE_SUCCESS;
}
}
CTG_CACHE_NHIT_INC(CTG_CI_TBL_TAG, 1);
CTG_ERR_JRET(ctgGetTableCfgFromVnode(pCtg, pConn, pCtx->pName, pCtx->pVgInfo, NULL, pTask));
return TSDB_CODE_SUCCESS;
_return:
if (CTG_TASK_LAUNCHED == pTask->status) {
ctgHandleTaskEnd(pTask, code);
}
CTG_RET(code);
}
int32_t ctgLaunchGetQnodeTask(SCtgTask* pTask) {
SCatalog* pCtg = pTask->pJob->pCtg;
SRequestConnInfo* pConn = &pTask->pJob->conn;
@ -2077,6 +2303,8 @@ int32_t ctgLaunchGetUserTask(SCtgTask* pTask) {
if (inCache) {
pTask->res = rsp.pRawRes;
ctgTaskDebug("Final res got, pass:%d, pCond:%p", rsp.pRawRes->pass, rsp.pRawRes->pCond);
CTG_ERR_RET(ctgHandleTaskEnd(pTask, 0));
return TSDB_CODE_SUCCESS;
}
@ -2084,7 +2312,10 @@ int32_t ctgLaunchGetUserTask(SCtgTask* pTask) {
taosMemoryFreeClear(rsp.pRawRes);
if (rsp.metaNotExists) {
CTG_ERR_RET(ctgLaunchSubTask(pTask, CTG_TASK_GET_TB_META, ctgGetTbCfgCb, &pCtx->user.tbName));
SCtgTbMetaParam param;
param.pName = &pCtx->user.tbName;
param.flag = CTG_FLAG_SYNC_OP;
CTG_ERR_RET(ctgLaunchSubTask(pTask, CTG_TASK_GET_TB_META, ctgGetUserCb, &param));
} else {
CTG_ERR_RET(ctgGetUserDbAuthFromMnode(pCtg, pConn, pCtx->user.user, NULL, pTask));
}
@ -2138,6 +2369,27 @@ _return:
CTG_RET(ctgHandleTaskEnd(pTask, pTask->subRes.code));
}
int32_t ctgGetTbTagCb(SCtgTask* pTask) {
int32_t code = 0;
CTG_ERR_JRET(pTask->subRes.code);
SCtgTbTagCtx* pCtx = (SCtgTbTagCtx*)pTask->taskCtx;
SDBVgInfo* pDb = (SDBVgInfo*)pTask->subRes.res;
if (NULL == pCtx->pVgInfo) {
pCtx->pVgInfo = taosMemoryCalloc(1, sizeof(SVgroupInfo));
CTG_ERR_JRET(ctgGetVgInfoFromHashValue(pTask->pJob->pCtg, pDb, pCtx->pName, pCtx->pVgInfo));
}
CTG_RET(ctgLaunchGetTbTagTask(pTask));
_return:
CTG_RET(ctgHandleTaskEnd(pTask, pTask->subRes.code));
}
int32_t ctgGetUserCb(SCtgTask* pTask) {
int32_t code = 0;
@ -2162,8 +2414,12 @@ int32_t ctgCompDbVgTasks(SCtgTask* pTask, void* param, bool* equal) {
int32_t ctgCompTbMetaTasks(SCtgTask* pTask, void* param, bool* equal) {
SCtgTbMetaCtx* ctx = pTask->taskCtx;
SCtgTbMetaParam* pParam = (SCtgTbMetaParam*)param;
*equal = tNameTbNameEqual(ctx->pName, (SName*)param);
*equal = tNameTbNameEqual(ctx->pName, (SName*)pParam->pName);
if (*equal) {
ctx->flag |= pParam->flag;
}
return TSDB_CODE_SUCCESS;
}
@ -2197,6 +2453,7 @@ SCtgAsyncFps gCtgAsyncFps[] = {
{ctgInitGetSvrVerTask, ctgLaunchGetSvrVerTask, ctgHandleGetSvrVerRsp, ctgDumpSvrVer, NULL, NULL},
{ctgInitGetTbMetasTask, ctgLaunchGetTbMetasTask, ctgHandleGetTbMetasRsp, ctgDumpTbMetasRes, NULL, NULL},
{ctgInitGetTbHashsTask, ctgLaunchGetTbHashsTask, ctgHandleGetTbHashsRsp, ctgDumpTbHashsRes, NULL, NULL},
{ctgInitGetTbTagTask, ctgLaunchGetTbTagTask, ctgHandleGetTbTagRsp, ctgDumpTbTagRes, NULL, NULL},
};
int32_t ctgMakeAsyncRes(SCtgJob* pJob) {
@ -2284,6 +2541,9 @@ int32_t ctgLaunchSubTask(SCtgTask* pTask, CTG_TASK_TYPE type, ctgSubTaskCbFp fp,
}
SCtgTask* pSub = taosArrayGet(pJob->pTasks, subTaskId);
if (newTask) {
pSub->subTask = true;
}
CTG_ERR_RET(ctgSetSubTaskCb(pSub, pTask));

31
source/libs/catalog/src/ctgCache.c
View File

@ -703,7 +703,31 @@ _return:
CTG_RET(code);
}
int32_t ctgGetCachedStbNameFromSuid(SCatalog* pCtg, char* dbFName, uint64_t suid, char **stbName) {
*stbName = NULL;
SCtgDBCache *dbCache = NULL;
ctgAcquireDBCache(pCtg, dbFName, &dbCache);
if (NULL == dbCache) {
ctgDebug("db %s not in cache", dbFName);
return TSDB_CODE_SUCCESS;
}
char *stb = taosHashAcquire(dbCache->stbCache, &suid, sizeof(suid));
if (NULL == stb) {
ctgDebug("stb 0x%" PRIx64 " not in cache, dbFName:%s", suid, dbFName);
return TSDB_CODE_SUCCESS;
}
*stbName = taosStrdup(stb);
taosHashRelease(dbCache->stbCache, stb);
return TSDB_CODE_SUCCESS;
}
int32_t ctgChkAuthFromCache(SCatalog *pCtg, SUserAuthInfo *pReq, bool *inCache, SCtgAuthRsp *pRes) {
int32_t code = 0;
if (IS_SYS_DBNAME(pReq->tbName.dbname)) {
*inCache = true;
pRes->pRawRes->pass = true;
@ -728,7 +752,7 @@ int32_t ctgChkAuthFromCache(SCatalog *pCtg, SUserAuthInfo *pReq, bool *inCache,
CTG_LOCK(CTG_READ, &pUser->lock);
memcpy(&req.authInfo, &pUser->userAuth, sizeof(pUser->userAuth));
int32_t code = ctgChkSetAuthRes(pCtg, &req, pRes);
code = ctgChkSetAuthRes(pCtg, &req, pRes);
CTG_UNLOCK(CTG_READ, &pUser->lock);
CTG_ERR_JRET(code);
@ -742,8 +766,9 @@ _return:
*inCache = false;
CTG_CACHE_NHIT_INC(CTG_CI_USER, 1);
ctgDebug("Get user from cache failed, user:%s, metaNotExists:%d, code:%d", pReq->user, pRes->metaNotExists, code);
return TSDB_CODE_SUCCESS;
return code;
}
void ctgDequeue(SCtgCacheOperation **op) {
@ -1057,7 +1082,7 @@ int32_t ctgUpdateUserEnqueue(SCatalog *pCtg, SGetUserAuthRsp *pAuth, bool syncOp
if (NULL == msg) {
ctgError("malloc %d failed", (int32_t)sizeof(SCtgUpdateUserMsg));
taosMemoryFree(op);
CTG_ERR_RET(TSDB_CODE_OUT_OF_MEMORY);
CTG_ERR_JRET(TSDB_CODE_OUT_OF_MEMORY);
}
msg->pCtg = pCtg;

148
source/libs/catalog/src/ctgUtil.c
View File

@ -170,6 +170,9 @@ void ctgFreeSMetaData(SMetaData* pData) {
taosArrayDestroy(pData->pTableCfg);
pData->pTableCfg = NULL;
taosArrayDestroy(pData->pTableTag);
pData->pTableTag = NULL;
taosMemoryFreeClear(pData->pSvrVer);
}
@ -486,6 +489,18 @@ void ctgFreeBatchHash(void* hash) {
taosMemoryFreeClear(pRes->pRes);
}
void ctgFreeJsonTagVal(void* val) {
if (NULL == val) {
return;
}
STagVal* pVal = (STagVal*)val;
if (TSDB_DATA_TYPE_JSON == pVal->type) {
taosMemoryFree(pVal->pData);
}
}
void ctgFreeTaskRes(CTG_TASK_TYPE type, void** pRes) {
switch (type) {
case CTG_TASK_GET_QNODE:
@ -516,16 +531,32 @@ void ctgFreeTaskRes(CTG_TASK_TYPE type, void** pRes) {
}
break;
}
case CTG_TASK_GET_USER: {
if (*pRes) {
SUserAuthRes* pAuth = (SUserAuthRes*)*pRes;
nodesDestroyNode(pAuth->pCond);
taosMemoryFreeClear(*pRes);
}
break;
}
case CTG_TASK_GET_TB_HASH:
case CTG_TASK_GET_DB_INFO:
case CTG_TASK_GET_INDEX_INFO:
case CTG_TASK_GET_UDF:
case CTG_TASK_GET_USER:
case CTG_TASK_GET_SVR_VER:
case CTG_TASK_GET_TB_META: {
taosMemoryFreeClear(*pRes);
break;
}
case CTG_TASK_GET_TB_TAG: {
if (1 == taosArrayGetSize(*pRes)) {
taosArrayDestroyEx(*pRes, ctgFreeJsonTagVal);
} else {
taosArrayDestroy(*pRes);
}
*pRes = NULL;
break;
}
case CTG_TASK_GET_TB_META_BATCH: {
SArray* pArray = (SArray*)*pRes;
int32_t num = taosArrayGetSize(pArray);
@ -679,6 +710,13 @@ void ctgFreeTaskCtx(SCtgTask* pTask) {
taosMemoryFreeClear(pTask->taskCtx);
break;
}
case CTG_TASK_GET_TB_TAG: {
SCtgTbTagCtx* taskCtx = (SCtgTbTagCtx*)pTask->taskCtx;
taosMemoryFreeClear(taskCtx->pName);
taosMemoryFreeClear(taskCtx->pVgInfo);
taosMemoryFreeClear(taskCtx);
break;
}
case CTG_TASK_GET_DB_VGROUP:
case CTG_TASK_GET_DB_CFG:
case CTG_TASK_GET_DB_INFO:
@ -1336,57 +1374,75 @@ int32_t ctgChkSetTbAuthRes(SCatalog* pCtg, SCtgAuthReq* req, SCtgAuthRsp* res) {
STableMeta* pMeta = NULL;
SGetUserAuthRsp* pInfo = &req->authInfo;
SHashObj* pTbs = (AUTH_TYPE_READ == req->singleType) ? pInfo->readTbs : pInfo->writeTbs;
char* stbName = NULL;
char tbFullName[TSDB_TABLE_FNAME_LEN];
tNameExtractFullName(&req->pRawReq->tbName, tbFullName);
char* pCond = taosHashGet(pTbs, tbFullName, strlen(tbFullName));
if (pCond) {
if (strlen(pCond) > 1) {
CTG_ERR_RET(nodesStringToNode(pCond, &res->pRawRes->pCond));
char tbFName[TSDB_TABLE_FNAME_LEN];
char dbFName[TSDB_DB_FNAME_LEN];
tNameExtractFullName(&req->pRawReq->tbName, tbFName);
tNameGetFullDbName(&req->pRawReq->tbName, dbFName);
while (true) {
taosMemoryFreeClear(pMeta);
char* pCond = taosHashGet(pTbs, tbFName, strlen(tbFName));
if (pCond) {
if (strlen(pCond) > 1) {
CTG_ERR_JRET(nodesStringToNode(pCond, &res->pRawRes->pCond));
}
res->pRawRes->pass = true;
goto _return;
}
res->pRawRes->pass = true;
return TSDB_CODE_SUCCESS;
if (stbName) {
res->pRawRes->pass = false;
goto _return;
}
CTG_ERR_JRET(catalogGetCachedTableMeta(pCtg, &req->pRawReq->tbName, &pMeta));
if (NULL == pMeta) {
if (req->onlyCache) {
res->metaNotExists = true;
ctgDebug("db %s tb %s meta not in cache for auth", req->pRawReq->tbName.dbname, req->pRawReq->tbName.tname);
goto _return;
}
SCtgTbMetaCtx ctx = {0};
ctx.pName = (SName*)&req->pRawReq->tbName;
ctx.flag = CTG_FLAG_UNKNOWN_STB | CTG_FLAG_SYNC_OP;
CTG_ERR_JRET(ctgGetTbMeta(pCtg, req->pConn, &ctx, &pMeta));
}
if (TSDB_SUPER_TABLE == pMeta->tableType || TSDB_NORMAL_TABLE == pMeta->tableType) {
res->pRawRes->pass = false;
goto _return;
}
if (TSDB_CHILD_TABLE == pMeta->tableType) {
CTG_ERR_JRET(ctgGetCachedStbNameFromSuid(pCtg, dbFName, pMeta->suid, &stbName));
if (NULL == stbName) {
if (req->onlyCache) {
res->metaNotExists = true;
ctgDebug("suid %" PRIu64 " name not in cache for auth", pMeta->suid);
goto _return;
}
continue;
}
sprintf(tbFName, "%s.%s", dbFName, stbName);
continue;
}
ctgError("Invalid table type %d for %s", pMeta->tableType, tbFName);
CTG_ERR_JRET(TSDB_CODE_INVALID_PARA);
}
res->pRawRes->pass = false;
// CTG_ERR_RET(catalogGetCachedTableMeta(pCtg, &req->pRawReq->tbName, &pMeta));
// if (NULL == pMeta) {
// if (req->onlyCache) {
// res->metaNotExists = true;
// ctgDebug("db %s tb %s meta not in cache for auth", req->pRawReq->tbName.dbname, req->pRawReq->tbName.tname);
// return TSDB_CODE_SUCCESS;
// }
// CTG_ERR_RET(catalogGetTableMeta(pCtg, req->pConn, &req->pRawReq->tbName, &pMeta));
// }
// if (TSDB_SUPER_TABLE == pMeta->tableType || TSDB_NORMAL_TABLE == pMeta->tableType) {
// res->pRawRes->pass = false;
// goto _return;
// }
// if (TSDB_CHILD_TABLE == pMeta->tableType) {
// res->pRawRes->pass = true;
// /*
// char stbName[TSDB_TABLE_NAME_LEN] = {0};
// CTG_ERR_JRET(ctgGetCachedStbNameFromSuid(pCtg, pMeta->suid, stbName));
// if (0 == stbName[0]) {
// if (req->onlyCache) {
// res->notExists = true;
// return TSDB_CODE_SUCCESS;
// }
// CTG_ERR_RET(catalogRefreshTableMeta(pCtg, req->pConn, &req->pRawReq->tbName, 0));
// }
// */
// }
_return:
taosMemoryFree(pMeta);
taosMemoryFree(stbName);
CTG_RET(code);
}
@ -1423,7 +1479,7 @@ int32_t ctgChkSetAuthRes(SCatalog* pCtg, SCtgAuthReq* req, SCtgAuthRsp* res) {
if (pInfo->readTbs && taosHashGetSize(pInfo->readTbs) > 0) {
req->singleType = AUTH_TYPE_READ;
CTG_ERR_RET(ctgChkSetTbAuthRes(pCtg, req, res));
if (pRes->pass) {
if (pRes->pass || res->metaNotExists) {
return TSDB_CODE_SUCCESS;
}
}
@ -1439,7 +1495,7 @@ int32_t ctgChkSetAuthRes(SCatalog* pCtg, SCtgAuthReq* req, SCtgAuthRsp* res) {
if (pInfo->writeTbs && taosHashGetSize(pInfo->writeTbs) > 0) {
req->singleType = AUTH_TYPE_WRITE;
CTG_ERR_RET(ctgChkSetTbAuthRes(pCtg, req, res));
if (pRes->pass) {
if (pRes->pass || res->metaNotExists) {
return TSDB_CODE_SUCCESS;
}
}

2
source/libs/executor/inc/executil.h
View File

@ -39,8 +39,6 @@
#define GET_RES_WINDOW_KEY_LEN(_l) ((_l) + sizeof(uint64_t))
#define GET_TASKID(_t) (((SExecTaskInfo*)(_t))->id.str)
typedef struct SGroupResInfo {
int32_t index;
SArray* pRows; // SArray<SResKeyPos>

644
source/libs/executor/inc/executorInt.h
View File

@ -12,14 +12,75 @@
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef _TD_EXECUTOR_INT_H
#define _TD_EXECUTOR_INT_H
#ifndef TDENGINE_EXECUTORINT_H
#define TDENGINE_EXECUTORINT_H
#ifdef __cplusplus
extern "C" {
#endif
#include "os.h"
#include "tcommon.h"
#include "tlosertree.h"
#include "tsort.h"
#include "ttszip.h"
#include "tvariant.h"
#include "dataSinkMgt.h"
#include "executil.h"
#include "executor.h"
#include "planner.h"
#include "scalar.h"
#include "taosdef.h"
#include "tarray.h"
#include "tfill.h"
#include "thash.h"
#include "tlockfree.h"
#include "tmsg.h"
#include "tpagedbuf.h"
#include "tstream.h"
#include "tstreamUpdate.h"
#include "vnode.h"
typedef int32_t (*__block_search_fn_t)(char* data, int32_t num, int64_t key, int32_t order);
#define IS_VALID_SESSION_WIN(winInfo) ((winInfo).sessionWin.win.skey > 0)
#define SET_SESSION_WIN_INVALID(winInfo) ((winInfo).sessionWin.win.skey = INT64_MIN)
#define IS_INVALID_SESSION_WIN_KEY(winKey) ((winKey).win.skey <= 0)
#define SET_SESSION_WIN_KEY_INVALID(pWinKey) ((pWinKey)->win.skey = INT64_MIN)
/**
* If the number of generated results is greater than this value,
* query query will be halt and return results to client immediate.
*/
typedef struct SResultInfo { // TODO refactor
int64_t totalRows; // total generated result size in rows
int64_t totalBytes; // total results in bytes.
int32_t capacity; // capacity of current result output buffer
int32_t threshold; // result size threshold in rows.
} SResultInfo;
typedef struct STableQueryInfo {
TSKEY lastKey; // last check ts, todo remove it later
SResultRowPosition pos; // current active time window
} STableQueryInfo;
typedef struct SLimit {
int64_t limit;
int64_t offset;
} SLimit;
typedef struct STableScanAnalyzeInfo SFileBlockLoadRecorder;
enum {
STREAM_RECOVER_STEP__NONE = 0,
STREAM_RECOVER_STEP__PREPARE1,
STREAM_RECOVER_STEP__PREPARE2,
STREAM_RECOVER_STEP__SCAN1,
STREAM_RECOVER_STEP__SCAN2,
};
extern int32_t exchangeObjRefPool;
typedef struct {
@ -29,9 +90,584 @@ typedef struct {
int32_t bytes;
} SGroupKeys, SStateKeys;
typedef struct {
char* tablename;
char* dbname;
int32_t tversion;
SSchemaWrapper* sw;
SSchemaWrapper* qsw;
} SSchemaInfo;
typedef struct SExchangeOpStopInfo {
int32_t operatorType;
int64_t refId;
} SExchangeOpStopInfo;
typedef struct SExprSupp {
SExprInfo* pExprInfo;
int32_t numOfExprs; // the number of scalar expression in group operator
SqlFunctionCtx* pCtx;
int32_t* rowEntryInfoOffset; // offset value for each row result cell info
SFilterInfo* pFilterInfo;
} SExprSupp;
typedef enum {
EX_SOURCE_DATA_NOT_READY = 0x1,
EX_SOURCE_DATA_READY = 0x2,
EX_SOURCE_DATA_EXHAUSTED = 0x3,
} EX_SOURCE_STATUS;
#define COL_MATCH_FROM_COL_ID 0x1
#define COL_MATCH_FROM_SLOT_ID 0x2
typedef struct SLoadRemoteDataInfo {
uint64_t totalSize; // total load bytes from remote
uint64_t totalRows; // total number of rows
uint64_t totalElapsed; // total elapsed time
} SLoadRemoteDataInfo;
typedef struct SLimitInfo {
SLimit limit;
SLimit slimit;
uint64_t currentGroupId;
int64_t remainGroupOffset;
int64_t numOfOutputGroups;
int64_t remainOffset;
int64_t numOfOutputRows;
} SLimitInfo;
typedef struct SExchangeInfo {
SArray* pSources;
SArray* pSourceDataInfo;
tsem_t ready;
void* pTransporter;
// SArray<SSDataBlock*>, result block list, used to keep the multi-block that
// passed by downstream operator
SArray* pResultBlockList;
SArray* pRecycledBlocks; // build a pool for small data block to avoid to repeatly create and then destroy.
SSDataBlock* pDummyBlock; // dummy block, not keep data
bool seqLoadData; // sequential load data or not, false by default
int32_t current;
SLoadRemoteDataInfo loadInfo;
uint64_t self;
SLimitInfo limitInfo;
int64_t openedTs; // start exec time stamp, todo: move to SLoadRemoteDataInfo
} SExchangeInfo;
typedef struct SScanInfo {
int32_t numOfAsc;
int32_t numOfDesc;
} SScanInfo;
typedef struct SSampleExecInfo {
double sampleRatio; // data block sample ratio, 1 by default
uint32_t seed; // random seed value
} SSampleExecInfo;
enum {
TABLE_SCAN__TABLE_ORDER = 1,
TABLE_SCAN__BLOCK_ORDER = 2,
};
typedef struct SAggSupporter {
SSHashObj* pResultRowHashTable; // quick locate the window object for each result
char* keyBuf; // window key buffer
SDiskbasedBuf* pResultBuf; // query result buffer based on blocked-wised disk file
int32_t resultRowSize; // the result buffer size for each result row, with the meta data size for each row
int32_t currentPageId; // current write page id
} SAggSupporter;
typedef struct {
// if the upstream is an interval operator, the interval info is also kept here to get the time window to check if
// current data block needs to be loaded.
SInterval interval;
SAggSupporter* pAggSup;
SExprSupp* pExprSup; // expr supporter of aggregate operator
} SAggOptrPushDownInfo;
typedef struct STableMetaCacheInfo {
SLRUCache* pTableMetaEntryCache; // 100 by default
uint64_t metaFetch;
uint64_t cacheHit;
} STableMetaCacheInfo;
typedef struct STableScanBase {
STsdbReader* dataReader;
SFileBlockLoadRecorder readRecorder;
SQueryTableDataCond cond;
SAggOptrPushDownInfo pdInfo;
SColMatchInfo matchInfo;
SReadHandle readHandle;
SExprSupp pseudoSup;
STableMetaCacheInfo metaCache;
int32_t scanFlag; // table scan flag to denote if it is a repeat/reverse/main scan
int32_t dataBlockLoadFlag;
SLimitInfo limitInfo;
// there are more than one table list exists in one task, if only one vnode exists.
STableListInfo* pTableListInfo;
} STableScanBase;
typedef struct STableScanInfo {
STableScanBase base;
SScanInfo scanInfo;
int32_t scanTimes;
SSDataBlock* pResBlock;
SSampleExecInfo sample; // sample execution info
int32_t currentGroupId;
int32_t currentTable;
int8_t scanMode;
int8_t assignBlockUid;
bool hasGroupByTag;
bool countOnly;
} STableScanInfo;
typedef struct STableMergeScanInfo {
int32_t tableStartIndex;
int32_t tableEndIndex;
bool hasGroupId;
uint64_t groupId;
SArray* queryConds; // array of queryTableDataCond
STableScanBase base;
int32_t bufPageSize;
uint32_t sortBufSize; // max buffer size for in-memory sort
SArray* pSortInfo;
SSortHandle* pSortHandle;
SSDataBlock* pSortInputBlock;
int64_t startTs; // sort start time
SArray* sortSourceParams;
SLimitInfo limitInfo;
int64_t numOfRows;
SScanInfo scanInfo;
int32_t scanTimes;
SSDataBlock* pResBlock;
SSampleExecInfo sample; // sample execution info
SSortExecInfo sortExecInfo;
} STableMergeScanInfo;
typedef struct STagScanInfo {
SColumnInfo* pCols;
SSDataBlock* pRes;
SColMatchInfo matchInfo;
int32_t curPos;
SReadHandle readHandle;
STableListInfo* pTableListInfo;
} STagScanInfo;
typedef enum EStreamScanMode {
STREAM_SCAN_FROM_READERHANDLE = 1,
STREAM_SCAN_FROM_RES,
STREAM_SCAN_FROM_UPDATERES,
STREAM_SCAN_FROM_DELETE_DATA,
STREAM_SCAN_FROM_DATAREADER_RETRIEVE,
STREAM_SCAN_FROM_DATAREADER_RANGE,
} EStreamScanMode;
enum {
PROJECT_RETRIEVE_CONTINUE = 0x1,
PROJECT_RETRIEVE_DONE = 0x2,
};
typedef struct SStreamAggSupporter {
int32_t resultRowSize; // the result buffer size for each result row, with the meta data size for each row
SSDataBlock* pScanBlock;
SStreamState* pState;
int64_t gap; // stream session window gap
SqlFunctionCtx* pDummyCtx; // for combine
SSHashObj* pResultRows;
int32_t stateKeySize;
int16_t stateKeyType;
SDiskbasedBuf* pResultBuf;
} SStreamAggSupporter;
typedef struct SWindowSupporter {
SStreamAggSupporter* pStreamAggSup;
int64_t gap;
uint16_t parentType;
SAggSupporter* pIntervalAggSup;
} SWindowSupporter;
typedef struct SPartitionBySupporter {
SArray* pGroupCols; // group by columns, SArray<SColumn>
SArray* pGroupColVals; // current group column values, SArray<SGroupKeys>
char* keyBuf; // group by keys for hash
bool needCalc; // partition by column
} SPartitionBySupporter;
typedef struct SPartitionDataInfo {
uint64_t groupId;
char* tbname;
SArray* tags;
SArray* rowIds;
} SPartitionDataInfo;
typedef struct STimeWindowAggSupp {
int8_t calTrigger;
int8_t calTriggerSaved;
int64_t deleteMark;
int64_t deleteMarkSaved;
int64_t waterMark;
TSKEY maxTs;
TSKEY minTs;
SColumnInfoData timeWindowData; // query time window info for scalar function execution.
} STimeWindowAggSupp;
typedef struct SStreamScanInfo {
SExprInfo* pPseudoExpr;
int32_t numOfPseudoExpr;
SExprSupp tbnameCalSup;
SExprSupp tagCalSup;
int32_t primaryTsIndex; // primary time stamp slot id
SReadHandle readHandle;
SInterval interval; // if the upstream is an interval operator, the interval info is also kept here.
SColMatchInfo matchInfo;
SArray* pBlockLists; // multiple SSDatablock.
SSDataBlock* pRes; // result SSDataBlock
SSDataBlock* pUpdateRes; // update SSDataBlock
int32_t updateResIndex;
int32_t blockType; // current block type
int32_t validBlockIndex; // Is current data has returned?
uint64_t numOfExec; // execution times
STqReader* tqReader;
uint64_t groupId;
SUpdateInfo* pUpdateInfo;
EStreamScanMode scanMode;
struct SOperatorInfo* pStreamScanOp;
struct SOperatorInfo* pTableScanOp;
SArray* childIds;
SWindowSupporter windowSup;
SPartitionBySupporter partitionSup;
SExprSupp* pPartScalarSup;
bool assignBlockUid; // assign block uid to groupId, temporarily used for generating rollup SMA.
int32_t scanWinIndex; // for state operator
int32_t pullDataResIndex;
SSDataBlock* pPullDataRes; // pull data SSDataBlock
SSDataBlock* pDeleteDataRes; // delete data SSDataBlock
int32_t deleteDataIndex;
STimeWindow updateWin;
STimeWindowAggSupp twAggSup;
SSDataBlock* pUpdateDataRes;
// status for tmq
SNodeList* pGroupTags;
SNode* pTagCond;
SNode* pTagIndexCond;
// recover
int32_t blockRecoverContiCnt;
int32_t blockRecoverTotCnt;
SSDataBlock* pRecoverRes;
SSDataBlock* pCreateTbRes;
int8_t igCheckUpdate;
int8_t igExpired;
} SStreamScanInfo;
typedef struct {
SVnode* vnode;
SSDataBlock pRes; // result SSDataBlock
STsdbReader* dataReader;
SSnapContext* sContext;
STableListInfo* pTableListInfo;
} SStreamRawScanInfo;
typedef struct STableCountScanSupp {
int16_t dbNameSlotId;
int16_t stbNameSlotId;
int16_t tbCountSlotId;
bool groupByDbName;
bool groupByStbName;
char dbNameFilter[TSDB_DB_NAME_LEN];
char stbNameFilter[TSDB_TABLE_NAME_LEN];
} STableCountScanSupp;
typedef struct SOptrBasicInfo {
SResultRowInfo resultRowInfo;
SSDataBlock* pRes;
bool mergeResultBlock;
} SOptrBasicInfo;
typedef struct SIntervalAggOperatorInfo {
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.
STimeWindow win; // query time range
bool timeWindowInterpo; // interpolation needed or not
SArray* pInterpCols; // interpolation columns
int32_t resultTsOrder; // result timestamp order
int32_t inputOrder; // input data ts order
EOPTR_EXEC_MODEL execModel; // operator execution model [batch model|stream model]
STimeWindowAggSupp twAggSup;
SArray* pPrevValues; // SArray<SGroupKeys> used to keep the previous not null value for interpolation.
} SIntervalAggOperatorInfo;
typedef struct SMergeAlignedIntervalAggOperatorInfo {
SIntervalAggOperatorInfo* intervalAggOperatorInfo;
uint64_t groupId; // current groupId
int64_t curTs; // current ts
SSDataBlock* prefetchedBlock;
SResultRow* pResultRow;
} SMergeAlignedIntervalAggOperatorInfo;
typedef struct SStreamIntervalOperatorInfo {
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.
STimeWindowAggSupp twAggSup;
bool invertible;
bool ignoreExpiredData;
bool ignoreExpiredDataSaved;
SArray* pDelWins; // SWinRes
int32_t delIndex;
SSDataBlock* pDelRes;
SPhysiNode* pPhyNode; // create new child
SHashObj* pPullDataMap;
SArray* pPullWins; // SPullWindowInfo
int32_t pullIndex;
SSDataBlock* pPullDataRes;
bool isFinal;
SArray* pChildren;
SStreamState* pState;
SWinKey delKey;
uint64_t numOfDatapack;
SArray* pUpdated;
SSHashObj* pUpdatedMap;
int64_t dataVersion;
} SStreamIntervalOperatorInfo;
typedef struct SDataGroupInfo {
uint64_t groupId;
int64_t numOfRows;
SArray* pPageList;
} SDataGroupInfo;
typedef struct SWindowRowsSup {
STimeWindow win;
TSKEY prevTs;
int32_t startRowIndex;
int32_t numOfRows;
uint64_t groupId;
} SWindowRowsSup;
typedef struct SResultWindowInfo {
void* pOutputBuf;
SSessionKey sessionWin;
bool isOutput;
} SResultWindowInfo;
typedef struct SStreamSessionAggOperatorInfo {
SOptrBasicInfo binfo;
SStreamAggSupporter streamAggSup;
SExprSupp scalarSupp; // supporter for perform scalar function
SGroupResInfo groupResInfo;
int32_t primaryTsIndex; // primary timestamp slot id
int32_t endTsIndex; // window end timestamp slot id
int32_t order; // current SSDataBlock scan order
STimeWindowAggSupp twAggSup;
SSDataBlock* pWinBlock; // window result
SSDataBlock* pDelRes; // delete result
SSDataBlock* pUpdateRes; // update window
bool returnUpdate;
SSHashObj* pStDeleted;
void* pDelIterator;
SArray* pChildren; // cache for children's result; final stream operator
SPhysiNode* pPhyNode; // create new child
bool isFinal;
bool ignoreExpiredData;
bool ignoreExpiredDataSaved;
SArray* pUpdated;
SSHashObj* pStUpdated;
int64_t dataVersion;
} SStreamSessionAggOperatorInfo;
typedef struct SStreamStateAggOperatorInfo {
SOptrBasicInfo binfo;
SStreamAggSupporter streamAggSup;
SExprSupp scalarSupp; // supporter for perform scalar function
SGroupResInfo groupResInfo;
int32_t primaryTsIndex; // primary timestamp slot id
STimeWindowAggSupp twAggSup;
SColumn stateCol;
SSDataBlock* pDelRes;
SSHashObj* pSeDeleted;
void* pDelIterator;
SArray* pChildren; // cache for children's result;
bool ignoreExpiredData;
bool ignoreExpiredDataSaved;
SArray* pUpdated;
SSHashObj* pSeUpdated;
int64_t dataVersion;
} SStreamStateAggOperatorInfo;
typedef struct SStreamPartitionOperatorInfo {
SOptrBasicInfo binfo;
SPartitionBySupporter partitionSup;
SExprSupp scalarSup;
SExprSupp tbnameCalSup;
SExprSupp tagCalSup;
SHashObj* pPartitions;
void* parIte;
void* pTbNameIte;
SSDataBlock* pInputDataBlock;
int32_t tsColIndex;
SSDataBlock* pDelRes;
SSDataBlock* pCreateTbRes;
} SStreamPartitionOperatorInfo;
typedef struct SStreamFillSupporter {
int32_t type; // fill type
SInterval interval;
SResultRowData prev;
SResultRowData cur;
SResultRowData next;
SResultRowData nextNext;
SFillColInfo* pAllColInfo; // fill exprs and not fill exprs
SExprSupp notFillExprSup;
int32_t numOfAllCols; // number of all exprs, including the tags columns
int32_t numOfFillCols;
int32_t numOfNotFillCols;
int32_t rowSize;
SSHashObj* pResMap;
bool hasDelete;
} SStreamFillSupporter;
typedef struct SStreamFillOperatorInfo {
SStreamFillSupporter* pFillSup;
SSDataBlock* pRes;
SSDataBlock* pSrcBlock;
int32_t srcRowIndex;
SSDataBlock* pSrcDelBlock;
int32_t srcDelRowIndex;
SSDataBlock* pDelRes;
SColMatchInfo matchInfo;
int32_t primaryTsCol;
int32_t primarySrcSlotId;
SStreamFillInfo* pFillInfo;
} SStreamFillOperatorInfo;
#define OPTR_IS_OPENED(_optr) (((_optr)->status & OP_OPENED) == OP_OPENED)
#define OPTR_SET_OPENED(_optr) ((_optr)->status |= OP_OPENED)
SSchemaWrapper* extractQueriedColumnSchema(SScanPhysiNode* pScanNode);
int32_t initQueriedTableSchemaInfo(SReadHandle* pHandle, SScanPhysiNode* pScanNode, const char* dbName, SExecTaskInfo* pTaskInfo);
void cleanupQueriedTableScanInfo(SSchemaInfo* pSchemaInfo);
void initBasicInfo(SOptrBasicInfo* pInfo, SSDataBlock* pBlock);
void cleanupBasicInfo(SOptrBasicInfo* pInfo);
int32_t initExprSupp(SExprSupp* pSup, SExprInfo* pExprInfo, int32_t numOfExpr);
void cleanupExprSupp(SExprSupp* pSup);
void destroyExprInfo(SExprInfo* pExpr, int32_t numOfExprs);
int32_t initAggSup(SExprSupp* pSup, SAggSupporter* pAggSup, SExprInfo* pExprInfo, int32_t numOfCols, size_t keyBufSize,
const char* pkey, void* pState);
void cleanupAggSup(SAggSupporter* pAggSup);
void initResultSizeInfo(SResultInfo* pResultInfo, int32_t numOfRows);
void doBuildStreamResBlock(struct SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SGroupResInfo* pGroupResInfo,
SDiskbasedBuf* pBuf);
void doBuildResultDatablock(struct SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SGroupResInfo* pGroupResInfo,
SDiskbasedBuf* pBuf);
bool hasLimitOffsetInfo(SLimitInfo* pLimitInfo);
bool hasSlimitOffsetInfo(SLimitInfo* pLimitInfo);
void initLimitInfo(const SNode* pLimit, const SNode* pSLimit, SLimitInfo* pLimitInfo);
void resetLimitInfoForNextGroup(SLimitInfo* pLimitInfo);
bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo);
void applyAggFunctionOnPartialTuples(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, SColumnInfoData* pTimeWindowData,
int32_t offset, int32_t forwardStep, int32_t numOfTotal, int32_t numOfOutput);
int32_t extractDataBlockFromFetchRsp(SSDataBlock* pRes, char* pData, SArray* pColList, char** pNextStart);
void updateLoadRemoteInfo(SLoadRemoteDataInfo* pInfo, int64_t numOfRows, int32_t dataLen, int64_t startTs,
struct SOperatorInfo* pOperator);
STimeWindow getFirstQualifiedTimeWindow(int64_t ts, STimeWindow* pWindow, SInterval* pInterval, int32_t order);
int32_t getBufferPgSize(int32_t rowSize, uint32_t* defaultPgsz, uint32_t* defaultBufsz);
extern void doDestroyExchangeOperatorInfo(void* param);
void doFilter(SSDataBlock* pBlock, SFilterInfo* pFilterInfo, SColMatchInfo* pColMatchInfo);
int32_t addTagPseudoColumnData(SReadHandle* pHandle, const SExprInfo* pExpr, int32_t numOfExpr, SSDataBlock* pBlock,
int32_t rows, const char* idStr, STableMetaCacheInfo* pCache);
void appendOneRowToDataBlock(SSDataBlock* pBlock, STupleHandle* pTupleHandle);
void setTbNameColData(const SSDataBlock* pBlock, SColumnInfoData* pColInfoData, int32_t functionId, const char* name);
void setResultRowInitCtx(SResultRow* pResult, SqlFunctionCtx* pCtx, int32_t numOfOutput, int32_t* rowEntryInfoOffset);
void clearResultRowInitFlag(SqlFunctionCtx* pCtx, int32_t numOfOutput);
SResultRow* doSetResultOutBufByKey(SDiskbasedBuf* pResultBuf, SResultRowInfo* pResultRowInfo, char* pData,
int16_t bytes, bool masterscan, uint64_t groupId, SExecTaskInfo* pTaskInfo,
bool isIntervalQuery, SAggSupporter* pSup, bool keepGroup);
int32_t projectApplyFunctions(SExprInfo* pExpr, SSDataBlock* pResult, SSDataBlock* pSrcBlock, SqlFunctionCtx* pCtx,
int32_t numOfOutput, SArray* pPseudoList);
void setInputDataBlock(SExprSupp* pExprSupp, SSDataBlock* pBlock, int32_t order, int32_t scanFlag, bool createDummyCol);
int32_t checkForQueryBuf(size_t numOfTables);
int32_t createDataSinkParam(SDataSinkNode* pNode, void** pParam, SExecTaskInfo* pTask, SReadHandle* readHandle);
STimeWindow getActiveTimeWindow(SDiskbasedBuf* pBuf, SResultRowInfo* pResultRowInfo, int64_t ts, SInterval* pInterval,
int32_t order);
int32_t getNumOfRowsInTimeWindow(SDataBlockInfo* pDataBlockInfo, TSKEY* pPrimaryColumn, int32_t startPos, TSKEY ekey,
__block_search_fn_t searchFn, STableQueryInfo* item, int32_t order);
int32_t binarySearchForKey(char* pValue, int num, TSKEY key, int order);
SResultRow* getNewResultRow(SDiskbasedBuf* pResultBuf, int32_t* currentPageId, int32_t interBufSize);
void getCurSessionWindow(SStreamAggSupporter* pAggSup, TSKEY startTs, TSKEY endTs, uint64_t groupId, SSessionKey* pKey);
bool isInTimeWindow(STimeWindow* pWin, TSKEY ts, int64_t gap);
bool functionNeedToExecute(SqlFunctionCtx* pCtx);
bool isOverdue(TSKEY ts, STimeWindowAggSupp* pSup);
bool isCloseWindow(STimeWindow* pWin, STimeWindowAggSupp* pSup);
bool isDeletedStreamWindow(STimeWindow* pWin, uint64_t groupId, SStreamState* pState, STimeWindowAggSupp* pTwSup);
void appendOneRowToStreamSpecialBlock(SSDataBlock* pBlock, TSKEY* pStartTs, TSKEY* pEndTs, uint64_t* pUid,
uint64_t* pGp, void* pTbName);
uint64_t calGroupIdByData(SPartitionBySupporter* pParSup, SExprSupp* pExprSup, SSDataBlock* pBlock, int32_t rowId);
int32_t finalizeResultRows(SDiskbasedBuf* pBuf, SResultRowPosition* resultRowPosition, SExprSupp* pSup,
SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo);
bool groupbyTbname(SNodeList* pGroupList);
int32_t buildDataBlockFromGroupRes(struct SOperatorInfo* pOperator, SStreamState* pState, SSDataBlock* pBlock, SExprSupp* pSup,
SGroupResInfo* pGroupResInfo);
int32_t saveSessionDiscBuf(SStreamState* pState, SSessionKey* key, void* buf, int32_t size);
int32_t buildSessionResultDataBlock(struct SOperatorInfo* pOperator, SStreamState* pState, SSDataBlock* pBlock,
SExprSupp* pSup, SGroupResInfo* pGroupResInfo);
int32_t setOutputBuf(SStreamState* pState, STimeWindow* win, SResultRow** pResult, int64_t tableGroupId,
SqlFunctionCtx* pCtx, int32_t numOfOutput, int32_t* rowEntryInfoOffset, SAggSupporter* pAggSup);
int32_t releaseOutputBuf(SStreamState* pState, SWinKey* pKey, SResultRow* pResult);
int32_t saveOutputBuf(SStreamState* pState, SWinKey* pKey, SResultRow* pResult, int32_t resSize);
void getNextIntervalWindow(SInterval* pInterval, STimeWindow* tw, int32_t order);
int32_t getForwardStepsInBlock(int32_t numOfRows, __block_search_fn_t searchFn, TSKEY ekey, int32_t pos, int32_t order,
int64_t* pData);
void appendCreateTableRow(SStreamState* pState, SExprSupp* pTableSup, SExprSupp* pTagSup, uint64_t groupId,
SSDataBlock* pSrcBlock, int32_t rowId, SSDataBlock* pDestBlock);
SSDataBlock* buildCreateTableBlock(SExprSupp* tbName, SExprSupp* tag);
SExprInfo* createExpr(SNodeList* pNodeList, int32_t* numOfExprs);
void copyResultrowToDataBlock(SExprInfo* pExprInfo, int32_t numOfExprs, SResultRow* pRow, SqlFunctionCtx* pCtx,
SSDataBlock* pBlock, const int32_t* rowEntryOffset, SExecTaskInfo* pTaskInfo);
void doUpdateNumOfRows(SqlFunctionCtx* pCtx, SResultRow* pRow, int32_t numOfExprs, const int32_t* rowEntryOffset);
void doClearBufferedBlocks(SStreamScanInfo* pInfo);
uint64_t calcGroupId(char* pData, int32_t len);
#ifdef __cplusplus
}
#endif
#endif /*_TD_EXECUTOR_INT_H*/
#endif // TDENGINE_EXECUTORINT_H

885
source/libs/executor/inc/executorimpl.h
View File

@ -1,885 +0,0 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TDENGINE_EXECUTORIMPL_H
#define TDENGINE_EXECUTORIMPL_H
#ifdef __cplusplus
extern "C" {
#endif
#include "os.h"
#include "tcommon.h"
#include "tlosertree.h"
#include "tsort.h"
#include "ttszip.h"
#include "tvariant.h"
#include "dataSinkMgt.h"
#include "executil.h"
#include "executor.h"
#include "planner.h"
#include "scalar.h"
#include "taosdef.h"
#include "tarray.h"
#include "tfill.h"
#include "thash.h"
#include "tlockfree.h"
#include "tmsg.h"
#include "tpagedbuf.h"
#include "tstream.h"
#include "tstreamUpdate.h"
#include "executorInt.h"
#include "vnode.h"
typedef int32_t (*__block_search_fn_t)(char* data, int32_t num, int64_t key, int32_t order);
#define IS_VALID_SESSION_WIN(winInfo) ((winInfo).sessionWin.win.skey > 0)
#define SET_SESSION_WIN_INVALID(winInfo) ((winInfo).sessionWin.win.skey = INT64_MIN)
#define IS_INVALID_SESSION_WIN_KEY(winKey) ((winKey).win.skey <= 0)
#define SET_SESSION_WIN_KEY_INVALID(pWinKey) ((pWinKey)->win.skey = INT64_MIN)
enum {
// when this task starts to execute, this status will set
TASK_NOT_COMPLETED = 0x1u,
/* Task is over
* 1. this status is used in one row result query process, e.g., count/sum/first/last/ avg...etc.
* 2. when all data within queried time window, it is also denoted as query_completed
*/
TASK_COMPLETED = 0x2u,
};
/**
* If the number of generated results is greater than this value,
* query query will be halt and return results to client immediate.
*/
typedef struct SResultInfo { // TODO refactor
int64_t totalRows; // total generated result size in rows
int64_t totalBytes; // total results in bytes.
int32_t capacity; // capacity of current result output buffer
int32_t threshold; // result size threshold in rows.
} SResultInfo;
typedef struct STableQueryInfo {
TSKEY lastKey; // last check ts, todo remove it later
SResultRowPosition pos; // current active time window
} STableQueryInfo;
typedef struct SLimit {
int64_t limit;
int64_t offset;
} SLimit;
typedef struct STableScanAnalyzeInfo SFileBlockLoadRecorder;
typedef struct STaskCostInfo {
int64_t created;
int64_t start;
uint64_t elapsedTime;
double extractListTime;
double groupIdMapTime;
SFileBlockLoadRecorder* pRecoder;
} STaskCostInfo;
typedef struct SOperatorCostInfo {
double openCost;
double totalCost;
} SOperatorCostInfo;
struct SOperatorInfo;
typedef int32_t (*__optr_encode_fn_t)(struct SOperatorInfo* pOperator, char** result, int32_t* length);
typedef int32_t (*__optr_decode_fn_t)(struct SOperatorInfo* pOperator, char* result);
typedef int32_t (*__optr_open_fn_t)(struct SOperatorInfo* pOptr);
typedef SSDataBlock* (*__optr_fn_t)(struct SOperatorInfo* pOptr);
typedef void (*__optr_close_fn_t)(void* param);
typedef int32_t (*__optr_explain_fn_t)(struct SOperatorInfo* pOptr, void** pOptrExplain, uint32_t* len);
typedef int32_t (*__optr_reqBuf_fn_t)(struct SOperatorInfo* pOptr);
typedef struct STaskIdInfo {
uint64_t queryId; // this is also a request id
uint64_t subplanId;
uint64_t templateId;
char* str;
int32_t vgId;
} STaskIdInfo;
enum {
STREAM_RECOVER_STEP__NONE = 0,
STREAM_RECOVER_STEP__PREPARE1,
STREAM_RECOVER_STEP__PREPARE2,
STREAM_RECOVER_STEP__SCAN1,
STREAM_RECOVER_STEP__SCAN2,
};
typedef struct {
STqOffsetVal currentOffset; // for tmq
SMqMetaRsp metaRsp; // for tmq fetching meta
int64_t snapshotVer;
SPackedData submit;
SSchemaWrapper* schema;
char tbName[TSDB_TABLE_NAME_LEN];
int8_t recoverStep;
int8_t recoverScanFinished;
SQueryTableDataCond tableCond;
int64_t fillHistoryVer1;
int64_t fillHistoryVer2;
SStreamState* pState;
int64_t dataVersion;
int64_t checkPointId;
} SStreamTaskInfo;
typedef struct {
char* tablename;
char* dbname;
int32_t tversion;
SSchemaWrapper* sw;
SSchemaWrapper* qsw;
} SSchemaInfo;
typedef struct SExchangeOpStopInfo {
int32_t operatorType;
int64_t refId;
} SExchangeOpStopInfo;
typedef struct STaskStopInfo {
SRWLatch lock;
SArray* pStopInfo;
} STaskStopInfo;
struct SExecTaskInfo {
STaskIdInfo id;
uint32_t status;
STimeWindow window;
STaskCostInfo cost;
int64_t owner; // if it is in execution
int32_t code;
int32_t qbufQuota; // total available buffer (in KB) during execution query
int64_t version; // used for stream to record wal version, why not move to sschemainfo
SStreamTaskInfo streamInfo;
SSchemaInfo schemaInfo;
const char* sql; // query sql string
jmp_buf env; // jump to this position when error happens.
EOPTR_EXEC_MODEL execModel; // operator execution model [batch model|stream model]
SSubplan* pSubplan;
struct SOperatorInfo* pRoot;
SLocalFetch localFetch;
SArray* pResultBlockList; // result block list
STaskStopInfo stopInfo;
SRWLatch lock; // secure the access of STableListInfo
};
enum {
OP_NOT_OPENED = 0x0,
OP_OPENED = 0x1,
OP_RES_TO_RETURN = 0x5,
OP_EXEC_DONE = 0x9,
};
typedef struct SOperatorFpSet {
__optr_open_fn_t _openFn; // DO NOT invoke this function directly
__optr_fn_t getNextFn;
__optr_fn_t cleanupFn; // call this function to release the allocated resources ASAP
__optr_close_fn_t closeFn;
__optr_reqBuf_fn_t reqBufFn; // total used buffer for blocking operator
__optr_encode_fn_t encodeResultRow;
__optr_decode_fn_t decodeResultRow;
__optr_explain_fn_t getExplainFn;
} SOperatorFpSet;
typedef struct SExprSupp {
SExprInfo* pExprInfo;
int32_t numOfExprs; // the number of scalar expression in group operator
SqlFunctionCtx* pCtx;
int32_t* rowEntryInfoOffset; // offset value for each row result cell info
SFilterInfo* pFilterInfo;
} SExprSupp;
typedef struct SOperatorInfo {
uint16_t operatorType;
int16_t resultDataBlockId;
bool blocking; // block operator or not
uint8_t status; // denote if current operator is completed
char* name; // name, for debug purpose
void* info; // extension attribution
SExprSupp exprSupp;
SExecTaskInfo* pTaskInfo;
SOperatorCostInfo cost;
SResultInfo resultInfo;
struct SOperatorInfo** pDownstream; // downstram pointer list
int32_t numOfDownstream; // number of downstream. The value is always ONE expect for join operator
SOperatorFpSet fpSet;
} SOperatorInfo;
typedef enum {
EX_SOURCE_DATA_NOT_READY = 0x1,
EX_SOURCE_DATA_READY = 0x2,
EX_SOURCE_DATA_EXHAUSTED = 0x3,
} EX_SOURCE_STATUS;
#define COL_MATCH_FROM_COL_ID 0x1
#define COL_MATCH_FROM_SLOT_ID 0x2
typedef struct SLoadRemoteDataInfo {
uint64_t totalSize; // total load bytes from remote
uint64_t totalRows; // total number of rows
uint64_t totalElapsed; // total elapsed time
} SLoadRemoteDataInfo;
typedef struct SLimitInfo {
SLimit limit;
SLimit slimit;
uint64_t currentGroupId;
int64_t remainGroupOffset;
int64_t numOfOutputGroups;
int64_t remainOffset;
int64_t numOfOutputRows;
} SLimitInfo;
typedef struct SExchangeInfo {
SArray* pSources;
SArray* pSourceDataInfo;
tsem_t ready;
void* pTransporter;
// SArray<SSDataBlock*>, result block list, used to keep the multi-block that
// passed by downstream operator
SArray* pResultBlockList;
SArray* pRecycledBlocks; // build a pool for small data block to avoid to repeatly create and then destroy.
SSDataBlock* pDummyBlock; // dummy block, not keep data
bool seqLoadData; // sequential load data or not, false by default
int32_t current;
SLoadRemoteDataInfo loadInfo;
uint64_t self;
SLimitInfo limitInfo;
int64_t openedTs; // start exec time stamp, todo: move to SLoadRemoteDataInfo
} SExchangeInfo;
typedef struct SScanInfo {
int32_t numOfAsc;
int32_t numOfDesc;
} SScanInfo;
typedef struct SSampleExecInfo {
double sampleRatio; // data block sample ratio, 1 by default
uint32_t seed; // random seed value
} SSampleExecInfo;
enum {
TABLE_SCAN__TABLE_ORDER = 1,
TABLE_SCAN__BLOCK_ORDER = 2,
};
typedef struct SAggSupporter {
SSHashObj* pResultRowHashTable; // quick locate the window object for each result
char* keyBuf; // window key buffer
SDiskbasedBuf* pResultBuf; // query result buffer based on blocked-wised disk file
int32_t resultRowSize; // the result buffer size for each result row, with the meta data size for each row
int32_t currentPageId; // current write page id
} SAggSupporter;
typedef struct {
// if the upstream is an interval operator, the interval info is also kept here to get the time window to check if
// current data block needs to be loaded.
SInterval interval;
SAggSupporter* pAggSup;
SExprSupp* pExprSup; // expr supporter of aggregate operator
} SAggOptrPushDownInfo;
typedef struct STableMetaCacheInfo {
SLRUCache* pTableMetaEntryCache; // 100 by default
uint64_t metaFetch;
uint64_t cacheHit;
} STableMetaCacheInfo;
typedef struct STableScanBase {
STsdbReader* dataReader;
SFileBlockLoadRecorder readRecorder;
SQueryTableDataCond cond;
SAggOptrPushDownInfo pdInfo;
SColMatchInfo matchInfo;
SReadHandle readHandle;
SExprSupp pseudoSup;
STableMetaCacheInfo metaCache;
int32_t scanFlag; // table scan flag to denote if it is a repeat/reverse/main scan
int32_t dataBlockLoadFlag;
SLimitInfo limitInfo;
// there are more than one table list exists in one task, if only one vnode exists.
STableListInfo* pTableListInfo;
} STableScanBase;
typedef struct STableScanInfo {
STableScanBase base;
SScanInfo scanInfo;
int32_t scanTimes;
SSDataBlock* pResBlock;
SSampleExecInfo sample; // sample execution info
int32_t currentGroupId;
int32_t currentTable;
int8_t scanMode;
int8_t assignBlockUid;
bool hasGroupByTag;
bool countOnly;
} STableScanInfo;
typedef struct STableMergeScanInfo {
int32_t tableStartIndex;
int32_t tableEndIndex;
bool hasGroupId;
uint64_t groupId;
SArray* queryConds; // array of queryTableDataCond
STableScanBase base;
int32_t bufPageSize;
uint32_t sortBufSize; // max buffer size for in-memory sort
SArray* pSortInfo;
SSortHandle* pSortHandle;
SSDataBlock* pSortInputBlock;
int64_t startTs; // sort start time
SArray* sortSourceParams;
SLimitInfo limitInfo;
int64_t numOfRows;
SScanInfo scanInfo;
int32_t scanTimes;
SSDataBlock* pResBlock;
SSampleExecInfo sample; // sample execution info
SSortExecInfo sortExecInfo;
} STableMergeScanInfo;
typedef struct STagScanInfo {
SColumnInfo* pCols;
SSDataBlock* pRes;
SColMatchInfo matchInfo;
int32_t curPos;
SReadHandle readHandle;
STableListInfo* pTableListInfo;
} STagScanInfo;
typedef enum EStreamScanMode {
STREAM_SCAN_FROM_READERHANDLE = 1,
STREAM_SCAN_FROM_RES,
STREAM_SCAN_FROM_UPDATERES,
STREAM_SCAN_FROM_DELETE_DATA,
STREAM_SCAN_FROM_DATAREADER_RETRIEVE,
STREAM_SCAN_FROM_DATAREADER_RANGE,
} EStreamScanMode;
enum {
PROJECT_RETRIEVE_CONTINUE = 0x1,
PROJECT_RETRIEVE_DONE = 0x2,
};
typedef struct SStreamAggSupporter {
int32_t resultRowSize; // the result buffer size for each result row, with the meta data size for each row
SSDataBlock* pScanBlock;
SStreamState* pState;
int64_t gap; // stream session window gap
SqlFunctionCtx* pDummyCtx; // for combine
SSHashObj* pResultRows;
int32_t stateKeySize;
int16_t stateKeyType;
SDiskbasedBuf* pResultBuf;
} SStreamAggSupporter;
typedef struct SWindowSupporter {
SStreamAggSupporter* pStreamAggSup;
int64_t gap;
uint16_t parentType;
SAggSupporter* pIntervalAggSup;
} SWindowSupporter;
typedef struct SPartitionBySupporter {
SArray* pGroupCols; // group by columns, SArray<SColumn>
SArray* pGroupColVals; // current group column values, SArray<SGroupKeys>
char* keyBuf; // group by keys for hash
bool needCalc; // partition by column
} SPartitionBySupporter;
typedef struct SPartitionDataInfo {
uint64_t groupId;
char* tbname;
SArray* tags;
SArray* rowIds;
} SPartitionDataInfo;
typedef struct STimeWindowAggSupp {
int8_t calTrigger;
int8_t calTriggerSaved;
int64_t deleteMark;
int64_t deleteMarkSaved;
int64_t waterMark;
TSKEY maxTs;
TSKEY minTs;
SColumnInfoData timeWindowData; // query time window info for scalar function execution.
} STimeWindowAggSupp;
typedef struct SStreamScanInfo {
SExprInfo* pPseudoExpr;
int32_t numOfPseudoExpr;
SExprSupp tbnameCalSup;
SExprSupp tagCalSup;
int32_t primaryTsIndex; // primary time stamp slot id
SReadHandle readHandle;
SInterval interval; // if the upstream is an interval operator, the interval info is also kept here.
SColMatchInfo matchInfo;
SArray* pBlockLists; // multiple SSDatablock.
SSDataBlock* pRes; // result SSDataBlock
SSDataBlock* pUpdateRes; // update SSDataBlock
int32_t updateResIndex;
int32_t blockType; // current block type
int32_t validBlockIndex; // Is current data has returned?
uint64_t numOfExec; // execution times
STqReader* tqReader;
uint64_t groupId;
SUpdateInfo* pUpdateInfo;
EStreamScanMode scanMode;
SOperatorInfo* pStreamScanOp;
SOperatorInfo* pTableScanOp;
SArray* childIds;
SWindowSupporter windowSup;
SPartitionBySupporter partitionSup;
SExprSupp* pPartScalarSup;
bool assignBlockUid; // assign block uid to groupId, temporarily used for generating rollup SMA.
int32_t scanWinIndex; // for state operator
int32_t pullDataResIndex;
SSDataBlock* pPullDataRes; // pull data SSDataBlock
SSDataBlock* pDeleteDataRes; // delete data SSDataBlock
int32_t deleteDataIndex;
STimeWindow updateWin;
STimeWindowAggSupp twAggSup;
SSDataBlock* pUpdateDataRes;
// status for tmq
SNodeList* pGroupTags;
SNode* pTagCond;
SNode* pTagIndexCond;
// recover
int32_t blockRecoverContiCnt;
int32_t blockRecoverTotCnt;
SSDataBlock* pRecoverRes;
SSDataBlock* pCreateTbRes;
int8_t igCheckUpdate;
int8_t igExpired;
} SStreamScanInfo;
typedef struct {
SVnode* vnode;
SSDataBlock pRes; // result SSDataBlock
STsdbReader* dataReader;
SSnapContext* sContext;
STableListInfo* pTableListInfo;
} SStreamRawScanInfo;
typedef struct STableCountScanSupp {
int16_t dbNameSlotId;
int16_t stbNameSlotId;
int16_t tbCountSlotId;
bool groupByDbName;
bool groupByStbName;
char dbNameFilter[TSDB_DB_NAME_LEN];
char stbNameFilter[TSDB_TABLE_NAME_LEN];
} STableCountScanSupp;
typedef struct SOptrBasicInfo {
SResultRowInfo resultRowInfo;
SSDataBlock* pRes;
bool mergeResultBlock;
} SOptrBasicInfo;
typedef struct SIntervalAggOperatorInfo {
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.
STimeWindow win; // query time range
bool timeWindowInterpo; // interpolation needed or not
SArray* pInterpCols; // interpolation columns
int32_t resultTsOrder; // result timestamp order
int32_t inputOrder; // input data ts order
EOPTR_EXEC_MODEL execModel; // operator execution model [batch model|stream model]
STimeWindowAggSupp twAggSup;
SArray* pPrevValues; // SArray<SGroupKeys> used to keep the previous not null value for interpolation.
} SIntervalAggOperatorInfo;
typedef struct SMergeAlignedIntervalAggOperatorInfo {
SIntervalAggOperatorInfo* intervalAggOperatorInfo;
uint64_t groupId; // current groupId
int64_t curTs; // current ts
SSDataBlock* prefetchedBlock;
SResultRow* pResultRow;
} SMergeAlignedIntervalAggOperatorInfo;
typedef struct SStreamIntervalOperatorInfo {
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.
STimeWindowAggSupp twAggSup;
bool invertible;
bool ignoreExpiredData;
bool ignoreExpiredDataSaved;
SArray* pDelWins; // SWinRes
int32_t delIndex;
SSDataBlock* pDelRes;
SPhysiNode* pPhyNode; // create new child
SHashObj* pPullDataMap;
SArray* pPullWins; // SPullWindowInfo
int32_t pullIndex;
SSDataBlock* pPullDataRes;
bool isFinal;
SArray* pChildren;
SStreamState* pState;
SWinKey delKey;
uint64_t numOfDatapack;
SArray* pUpdated;
SSHashObj* pUpdatedMap;
int64_t dataVersion;
} SStreamIntervalOperatorInfo;
typedef struct SDataGroupInfo {
uint64_t groupId;
int64_t numOfRows;
SArray* pPageList;
} SDataGroupInfo;
typedef struct SWindowRowsSup {
STimeWindow win;
TSKEY prevTs;
int32_t startRowIndex;
int32_t numOfRows;
uint64_t groupId;
} SWindowRowsSup;
typedef struct SResultWindowInfo {
void* pOutputBuf;
SSessionKey sessionWin;
bool isOutput;
} SResultWindowInfo;
typedef struct SStateWindowInfo {
SResultWindowInfo winInfo;
SStateKeys* pStateKey;
} SStateWindowInfo;
typedef struct SStreamSessionAggOperatorInfo {
SOptrBasicInfo binfo;
SStreamAggSupporter streamAggSup;
SExprSupp scalarSupp; // supporter for perform scalar function
SGroupResInfo groupResInfo;
int32_t primaryTsIndex; // primary timestamp slot id
int32_t endTsIndex; // window end timestamp slot id
int32_t order; // current SSDataBlock scan order
STimeWindowAggSupp twAggSup;
SSDataBlock* pWinBlock; // window result
SSDataBlock* pDelRes; // delete result
SSDataBlock* pUpdateRes; // update window
bool returnUpdate;
SSHashObj* pStDeleted;
void* pDelIterator;
SArray* pChildren; // cache for children's result; final stream operator
SPhysiNode* pPhyNode; // create new child
bool isFinal;
bool ignoreExpiredData;
bool ignoreExpiredDataSaved;
SArray* pUpdated;
SSHashObj* pStUpdated;
int64_t dataVersion;
} SStreamSessionAggOperatorInfo;
typedef struct SStreamStateAggOperatorInfo {
SOptrBasicInfo binfo;
SStreamAggSupporter streamAggSup;
SExprSupp scalarSupp; // supporter for perform scalar function
SGroupResInfo groupResInfo;
int32_t primaryTsIndex; // primary timestamp slot id
STimeWindowAggSupp twAggSup;
SColumn stateCol;
SSDataBlock* pDelRes;
SSHashObj* pSeDeleted;
void* pDelIterator;
SArray* pChildren; // cache for children's result;
bool ignoreExpiredData;
bool ignoreExpiredDataSaved;
SArray* pUpdated;
SSHashObj* pSeUpdated;
int64_t dataVersion;
} SStreamStateAggOperatorInfo;
typedef struct SStreamPartitionOperatorInfo {
SOptrBasicInfo binfo;
SPartitionBySupporter partitionSup;
SExprSupp scalarSup;
SExprSupp tbnameCalSup;
SExprSupp tagCalSup;
SHashObj* pPartitions;
void* parIte;
void* pTbNameIte;
SSDataBlock* pInputDataBlock;
int32_t tsColIndex;
SSDataBlock* pDelRes;
SSDataBlock* pCreateTbRes;
} SStreamPartitionOperatorInfo;
typedef struct SStreamFillSupporter {
int32_t type; // fill type
SInterval interval;
SResultRowData prev;
SResultRowData cur;
SResultRowData next;
SResultRowData nextNext;
SFillColInfo* pAllColInfo; // fill exprs and not fill exprs
SExprSupp notFillExprSup;
int32_t numOfAllCols; // number of all exprs, including the tags columns
int32_t numOfFillCols;
int32_t numOfNotFillCols;
int32_t rowSize;
SSHashObj* pResMap;
bool hasDelete;
} SStreamFillSupporter;
typedef struct SStreamFillOperatorInfo {
SStreamFillSupporter* pFillSup;
SSDataBlock* pRes;
SSDataBlock* pSrcBlock;
int32_t srcRowIndex;
SSDataBlock* pSrcDelBlock;
int32_t srcDelRowIndex;
SSDataBlock* pDelRes;
SColMatchInfo matchInfo;
int32_t primaryTsCol;
int32_t primarySrcSlotId;
SStreamFillInfo* pFillInfo;
} SStreamFillOperatorInfo;
#define OPTR_IS_OPENED(_optr) (((_optr)->status & OP_OPENED) == OP_OPENED)
#define OPTR_SET_OPENED(_optr) ((_optr)->status |= OP_OPENED)
SExecTaskInfo* doCreateExecTaskInfo(uint64_t queryId, uint64_t taskId, int32_t vgId, EOPTR_EXEC_MODEL model,
char* dbFName);
SOperatorFpSet createOperatorFpSet(__optr_open_fn_t openFn, __optr_fn_t nextFn, __optr_fn_t cleanup,
__optr_close_fn_t closeFn, __optr_reqBuf_fn_t reqBufFn, __optr_explain_fn_t explain);
int32_t optrDummyOpenFn(SOperatorInfo* pOperator);
int32_t appendDownstream(SOperatorInfo* p, SOperatorInfo** pDownstream, int32_t num);
void setOperatorCompleted(SOperatorInfo* pOperator);
void setOperatorInfo(SOperatorInfo* pOperator, const char* name, int32_t type, bool blocking, int32_t status,
void* pInfo, SExecTaskInfo* pTaskInfo);
void destroyOperatorInfo(SOperatorInfo* pOperator);
int32_t optrDefaultBufFn(SOperatorInfo* pOperator);
void initBasicInfo(SOptrBasicInfo* pInfo, SSDataBlock* pBlock);
void cleanupBasicInfo(SOptrBasicInfo* pInfo);
int32_t initExprSupp(SExprSupp* pSup, SExprInfo* pExprInfo, int32_t numOfExpr);
void cleanupExprSupp(SExprSupp* pSup);
void destroyExprInfo(SExprInfo* pExpr, int32_t numOfExprs);
int32_t initAggSup(SExprSupp* pSup, SAggSupporter* pAggSup, SExprInfo* pExprInfo, int32_t numOfCols, size_t keyBufSize,
const char* pkey, void* pState);
void cleanupAggSup(SAggSupporter* pAggSup);
void initResultSizeInfo(SResultInfo* pResultInfo, int32_t numOfRows);
void doBuildStreamResBlock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SGroupResInfo* pGroupResInfo,
SDiskbasedBuf* pBuf);
void doBuildResultDatablock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SGroupResInfo* pGroupResInfo,
SDiskbasedBuf* pBuf);
bool hasLimitOffsetInfo(SLimitInfo* pLimitInfo);
bool hasSlimitOffsetInfo(SLimitInfo* pLimitInfo);
void initLimitInfo(const SNode* pLimit, const SNode* pSLimit, SLimitInfo* pLimitInfo);
void resetLimitInfoForNextGroup(SLimitInfo* pLimitInfo);
bool applyLimitOffset(SLimitInfo* pLimitInfo, SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo);
void applyAggFunctionOnPartialTuples(SExecTaskInfo* taskInfo, SqlFunctionCtx* pCtx, SColumnInfoData* pTimeWindowData,
int32_t offset, int32_t forwardStep, int32_t numOfTotal, int32_t numOfOutput);
int32_t extractDataBlockFromFetchRsp(SSDataBlock* pRes, char* pData, SArray* pColList, char** pNextStart);
void updateLoadRemoteInfo(SLoadRemoteDataInfo* pInfo, int64_t numOfRows, int32_t dataLen, int64_t startTs,
SOperatorInfo* pOperator);
STimeWindow getFirstQualifiedTimeWindow(int64_t ts, STimeWindow* pWindow, SInterval* pInterval, int32_t order);
SOperatorInfo* extractOperatorInTree(SOperatorInfo* pOperator, int32_t type, const char* id);
int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t* order, int32_t* scanFlag, bool inheritUsOrder);
int32_t getBufferPgSize(int32_t rowSize, uint32_t* defaultPgsz, uint32_t* defaultBufsz);
extern void doDestroyExchangeOperatorInfo(void* param);
void doFilter(SSDataBlock* pBlock, SFilterInfo* pFilterInfo, SColMatchInfo* pColMatchInfo);
int32_t addTagPseudoColumnData(SReadHandle* pHandle, const SExprInfo* pExpr, int32_t numOfExpr, SSDataBlock* pBlock,
int32_t rows, const char* idStr, STableMetaCacheInfo* pCache);
void appendOneRowToDataBlock(SSDataBlock* pBlock, STupleHandle* pTupleHandle);
void setTbNameColData(const SSDataBlock* pBlock, SColumnInfoData* pColInfoData, int32_t functionId, const char* name);
void setResultRowInitCtx(SResultRow* pResult, SqlFunctionCtx* pCtx, int32_t numOfOutput, int32_t* rowEntryInfoOffset);
void clearResultRowInitFlag(SqlFunctionCtx* pCtx, int32_t numOfOutput);
SResultRow* doSetResultOutBufByKey(SDiskbasedBuf* pResultBuf, SResultRowInfo* pResultRowInfo, char* pData,
int16_t bytes, bool masterscan, uint64_t groupId, SExecTaskInfo* pTaskInfo,
bool isIntervalQuery, SAggSupporter* pSup, bool keepGroup);
// operator creater functions
// clang-format off
SOperatorInfo* createExchangeOperatorInfo(void* pTransporter, SExchangePhysiNode* pExNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode, SReadHandle* pHandle, STableListInfo* pTableList, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanNode, SReadHandle* readHandle, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTagScanOperatorInfo(SReadHandle* pReadHandle, STagScanPhysiNode* pPhyNode, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSystemTableScanPhysiNode* pScanPhyNode, const char* pUser, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTableCountScanOperatorInfo(SReadHandle* handle, STableCountScanPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SAggPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createIndefinitOutputOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createProjectOperatorInfo(SOperatorInfo* downstream, SProjectPhysiNode* pProjPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createSortOperatorInfo(SOperatorInfo* downstream, SSortPhysiNode* pSortNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMultiwayMergeOperatorInfo(SOperatorInfo** dowStreams, size_t numStreams, SMergePhysiNode* pMergePhysiNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createCacherowsScanOperator(SLastRowScanPhysiNode* pTableScanNode, SReadHandle* readHandle, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SIntervalPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMergeIntervalOperatorInfo(SOperatorInfo* downstream, SMergeIntervalPhysiNode* pIntervalPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMergeAlignedIntervalOperatorInfo(SOperatorInfo* downstream, SMergeAlignedIntervalPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamFinalIntervalOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, int32_t numOfChild);
SOperatorInfo* createSessionAggOperatorInfo(SOperatorInfo* downstream, SSessionWinodwPhysiNode* pSessionNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createGroupOperatorInfo(SOperatorInfo* downstream, SAggPhysiNode* pAggNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createDataBlockInfoScanOperator(SReadHandle* readHandle, SBlockDistScanPhysiNode* pBlockScanNode, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle, STableScanPhysiNode* pTableScanNode, SNode* pTagCond, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createRawScanOperatorInfo(SReadHandle* pHandle, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createFillOperatorInfo(SOperatorInfo* downstream, SFillPhysiNode* pPhyFillNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStatewindowOperatorInfo(SOperatorInfo* downstream, SStateWinodwPhysiNode* pStateNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartitionPhysiNode* pPartNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamPartitionOperatorInfo(SOperatorInfo* downstream, SStreamPartitionPhysiNode* pPartNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTimeSliceOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMergeJoinOperatorInfo(SOperatorInfo** pDownstream, int32_t numOfDownstream, SSortMergeJoinPhysiNode* pJoinNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, int32_t numOfChild);
SOperatorInfo* createStreamIntervalOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamStateAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamFillOperatorInfo(SOperatorInfo* downstream, SStreamFillPhysiNode* pPhyFillNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createGroupSortOperatorInfo(SOperatorInfo* downstream, SGroupSortPhysiNode* pSortPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createEventwindowOperatorInfo(SOperatorInfo* downstream, SPhysiNode* physiNode, SExecTaskInfo* pTaskInfo);
// clang-format on
int32_t projectApplyFunctions(SExprInfo* pExpr, SSDataBlock* pResult, SSDataBlock* pSrcBlock, SqlFunctionCtx* pCtx,
int32_t numOfOutput, SArray* pPseudoList);
void setInputDataBlock(SExprSupp* pExprSupp, SSDataBlock* pBlock, int32_t order, int32_t scanFlag, bool createDummyCol);
int32_t checkForQueryBuf(size_t numOfTables);
bool isTaskKilled(SExecTaskInfo* pTaskInfo);
void setTaskKilled(SExecTaskInfo* pTaskInfo, int32_t rspCode);
void doDestroyTask(SExecTaskInfo* pTaskInfo);
void setTaskStatus(SExecTaskInfo* pTaskInfo, int8_t status);
void buildTaskId(uint64_t taskId, uint64_t queryId, char* dst);
SArray* getTableListInfo(const SExecTaskInfo* pTaskInfo);
int32_t createExecTaskInfo(SSubplan* pPlan, SExecTaskInfo** pTaskInfo, SReadHandle* pHandle, uint64_t taskId,
int32_t vgId, char* sql, EOPTR_EXEC_MODEL model);
int32_t createDataSinkParam(SDataSinkNode* pNode, void** pParam, SExecTaskInfo* pTask, SReadHandle* readHandle);
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SArray* pExecInfoList);
STimeWindow getActiveTimeWindow(SDiskbasedBuf* pBuf, SResultRowInfo* pResultRowInfo, int64_t ts, SInterval* pInterval,
int32_t order);
int32_t getNumOfRowsInTimeWindow(SDataBlockInfo* pDataBlockInfo, TSKEY* pPrimaryColumn, int32_t startPos, TSKEY ekey,
__block_search_fn_t searchFn, STableQueryInfo* item, int32_t order);
int32_t binarySearchForKey(char* pValue, int num, TSKEY key, int order);
SResultRow* getNewResultRow(SDiskbasedBuf* pResultBuf, int32_t* currentPageId, int32_t interBufSize);
void getCurSessionWindow(SStreamAggSupporter* pAggSup, TSKEY startTs, TSKEY endTs, uint64_t groupId, SSessionKey* pKey);
bool isInTimeWindow(STimeWindow* pWin, TSKEY ts, int64_t gap);
bool functionNeedToExecute(SqlFunctionCtx* pCtx);
bool isOverdue(TSKEY ts, STimeWindowAggSupp* pSup);
bool isCloseWindow(STimeWindow* pWin, STimeWindowAggSupp* pSup);
bool isDeletedStreamWindow(STimeWindow* pWin, uint64_t groupId, SStreamState* pState, STimeWindowAggSupp* pTwSup);
void appendOneRowToStreamSpecialBlock(SSDataBlock* pBlock, TSKEY* pStartTs, TSKEY* pEndTs, uint64_t* pUid,
uint64_t* pGp, void* pTbName);
uint64_t calGroupIdByData(SPartitionBySupporter* pParSup, SExprSupp* pExprSup, SSDataBlock* pBlock, int32_t rowId);
int32_t finalizeResultRows(SDiskbasedBuf* pBuf, SResultRowPosition* resultRowPosition, SExprSupp* pSup,
SSDataBlock* pBlock, SExecTaskInfo* pTaskInfo);
bool groupbyTbname(SNodeList* pGroupList);
int32_t buildDataBlockFromGroupRes(SOperatorInfo* pOperator, SStreamState* pState, SSDataBlock* pBlock, SExprSupp* pSup,
SGroupResInfo* pGroupResInfo);
int32_t saveSessionDiscBuf(SStreamState* pState, SSessionKey* key, void* buf, int32_t size);
int32_t buildSessionResultDataBlock(SOperatorInfo* pOperator, SStreamState* pState, SSDataBlock* pBlock,
SExprSupp* pSup, SGroupResInfo* pGroupResInfo);
int32_t setOutputBuf(SStreamState* pState, STimeWindow* win, SResultRow** pResult, int64_t tableGroupId,
SqlFunctionCtx* pCtx, int32_t numOfOutput, int32_t* rowEntryInfoOffset, SAggSupporter* pAggSup);
int32_t releaseOutputBuf(SStreamState* pState, SWinKey* pKey, SResultRow* pResult);
int32_t saveOutputBuf(SStreamState* pState, SWinKey* pKey, SResultRow* pResult, int32_t resSize);
void getNextIntervalWindow(SInterval* pInterval, STimeWindow* tw, int32_t order);
int32_t qAppendTaskStopInfo(SExecTaskInfo* pTaskInfo, SExchangeOpStopInfo* pInfo);
int32_t getForwardStepsInBlock(int32_t numOfRows, __block_search_fn_t searchFn, TSKEY ekey, int32_t pos, int32_t order,
int64_t* pData);
void appendCreateTableRow(SStreamState* pState, SExprSupp* pTableSup, SExprSupp* pTagSup, uint64_t groupId,
SSDataBlock* pSrcBlock, int32_t rowId, SSDataBlock* pDestBlock);
SSDataBlock* buildCreateTableBlock(SExprSupp* tbName, SExprSupp* tag);
SExprInfo* createExpr(SNodeList* pNodeList, int32_t* numOfExprs);
void copyResultrowToDataBlock(SExprInfo* pExprInfo, int32_t numOfExprs, SResultRow* pRow, SqlFunctionCtx* pCtx,
SSDataBlock* pBlock, const int32_t* rowEntryOffset, SExecTaskInfo* pTaskInfo);
void doUpdateNumOfRows(SqlFunctionCtx* pCtx, SResultRow* pRow, int32_t numOfExprs, const int32_t* rowEntryOffset);
void doClearBufferedBlocks(SStreamScanInfo* pInfo);
#ifdef __cplusplus
}
#endif
#endif // TDENGINE_EXECUTORIMPL_H

166
source/libs/executor/inc/operator.h Normal file
View File

@ -0,0 +1,166 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TDENGINE_OPERATOR_H
#define TDENGINE_OPERATOR_H
#ifdef __cplusplus
extern "C" {
#endif
typedef struct SOperatorCostInfo {
double openCost;
double totalCost;
} SOperatorCostInfo;
struct SOperatorInfo;
typedef int32_t (*__optr_encode_fn_t)(struct SOperatorInfo* pOperator, char** result, int32_t* length);
typedef int32_t (*__optr_decode_fn_t)(struct SOperatorInfo* pOperator, char* result);
typedef int32_t (*__optr_open_fn_t)(struct SOperatorInfo* pOptr);
typedef SSDataBlock* (*__optr_fn_t)(struct SOperatorInfo* pOptr);
typedef void (*__optr_close_fn_t)(void* param);
typedef int32_t (*__optr_explain_fn_t)(struct SOperatorInfo* pOptr, void** pOptrExplain, uint32_t* len);
typedef int32_t (*__optr_reqBuf_fn_t)(struct SOperatorInfo* pOptr);
typedef struct SOperatorFpSet {
__optr_open_fn_t _openFn; // DO NOT invoke this function directly
__optr_fn_t getNextFn;
__optr_fn_t cleanupFn; // call this function to release the allocated resources ASAP
__optr_close_fn_t closeFn;
__optr_reqBuf_fn_t reqBufFn; // total used buffer for blocking operator
__optr_encode_fn_t encodeResultRow;
__optr_decode_fn_t decodeResultRow;
__optr_explain_fn_t getExplainFn;
} SOperatorFpSet;
enum {
OP_NOT_OPENED = 0x0,
OP_OPENED = 0x1,
OP_RES_TO_RETURN = 0x5,
OP_EXEC_DONE = 0x9,
};
typedef struct SOperatorInfo {
uint16_t operatorType;
int16_t resultDataBlockId;
bool blocking; // block operator or not
uint8_t status; // denote if current operator is completed
char* name; // name, for debug purpose
void* info; // extension attribution
SExprSupp exprSupp;
SExecTaskInfo* pTaskInfo;
SOperatorCostInfo cost;
SResultInfo resultInfo;
struct SOperatorInfo** pDownstream; // downstram pointer list
int32_t numOfDownstream; // number of downstream. The value is always ONE expect for join operator
SOperatorFpSet fpSet;
} SOperatorInfo;
// operator creater functions
// clang-format off
SOperatorInfo* createExchangeOperatorInfo(void* pTransporter, SExchangePhysiNode* pExNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTableScanOperatorInfo(STableScanPhysiNode* pTableScanNode, SReadHandle* pHandle, STableListInfo* pTableList, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTableMergeScanOperatorInfo(STableScanPhysiNode* pTableScanNode, SReadHandle* readHandle, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTagScanOperatorInfo(SReadHandle* pReadHandle, STagScanPhysiNode* pPhyNode, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createSysTableScanOperatorInfo(void* readHandle, SSystemTableScanPhysiNode* pScanPhyNode, const char* pUser, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTableCountScanOperatorInfo(SReadHandle* handle, STableCountScanPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createAggregateOperatorInfo(SOperatorInfo* downstream, SAggPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createIndefinitOutputOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createProjectOperatorInfo(SOperatorInfo* downstream, SProjectPhysiNode* pProjPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createSortOperatorInfo(SOperatorInfo* downstream, SSortPhysiNode* pSortNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMultiwayMergeOperatorInfo(SOperatorInfo** dowStreams, size_t numStreams, SMergePhysiNode* pMergePhysiNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createCacherowsScanOperator(SLastRowScanPhysiNode* pTableScanNode, SReadHandle* readHandle, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createIntervalOperatorInfo(SOperatorInfo* downstream, SIntervalPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMergeIntervalOperatorInfo(SOperatorInfo* downstream, SMergeIntervalPhysiNode* pIntervalPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMergeAlignedIntervalOperatorInfo(SOperatorInfo* downstream, SMergeAlignedIntervalPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamFinalIntervalOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, int32_t numOfChild);
SOperatorInfo* createSessionAggOperatorInfo(SOperatorInfo* downstream, SSessionWinodwPhysiNode* pSessionNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createGroupOperatorInfo(SOperatorInfo* downstream, SAggPhysiNode* pAggNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createDataBlockInfoScanOperator(SReadHandle* readHandle, SBlockDistScanPhysiNode* pBlockScanNode, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle, STableScanPhysiNode* pTableScanNode, SNode* pTagCond, STableListInfo* pTableListInfo, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createRawScanOperatorInfo(SReadHandle* pHandle, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createFillOperatorInfo(SOperatorInfo* downstream, SFillPhysiNode* pPhyFillNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStatewindowOperatorInfo(SOperatorInfo* downstream, SStateWinodwPhysiNode* pStateNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createPartitionOperatorInfo(SOperatorInfo* downstream, SPartitionPhysiNode* pPartNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamPartitionOperatorInfo(SOperatorInfo* downstream, SStreamPartitionPhysiNode* pPartNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createTimeSliceOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createMergeJoinOperatorInfo(SOperatorInfo** pDownstream, int32_t numOfDownstream, SSortMergeJoinPhysiNode* pJoinNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamSessionAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamFinalSessionAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, int32_t numOfChild);
SOperatorInfo* createStreamIntervalOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamStateAggOperatorInfo(SOperatorInfo* downstream, SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createStreamFillOperatorInfo(SOperatorInfo* downstream, SStreamFillPhysiNode* pPhyFillNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createGroupSortOperatorInfo(SOperatorInfo* downstream, SGroupSortPhysiNode* pSortPhyNode, SExecTaskInfo* pTaskInfo);
SOperatorInfo* createEventwindowOperatorInfo(SOperatorInfo* downstream, SPhysiNode* physiNode, SExecTaskInfo* pTaskInfo);
// clang-format on
SOperatorFpSet createOperatorFpSet(__optr_open_fn_t openFn, __optr_fn_t nextFn, __optr_fn_t cleanup,
__optr_close_fn_t closeFn, __optr_reqBuf_fn_t reqBufFn, __optr_explain_fn_t explain);
int32_t optrDummyOpenFn(SOperatorInfo* pOperator);
int32_t appendDownstream(SOperatorInfo* p, SOperatorInfo** pDownstream, int32_t num);
void setOperatorCompleted(SOperatorInfo* pOperator);
void setOperatorInfo(SOperatorInfo* pOperator, const char* name, int32_t type, bool blocking, int32_t status,
void* pInfo, SExecTaskInfo* pTaskInfo);
int32_t optrDefaultBufFn(SOperatorInfo* pOperator);
SOperatorInfo* createOperator(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, SReadHandle* pHandle, SNode* pTagCond,
SNode* pTagIndexCond, const char* pUser, const char* dbname);
void destroyOperator(SOperatorInfo* pOperator);
SOperatorInfo* extractOperatorInTree(SOperatorInfo* pOperator, int32_t type, const char* id);
int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t* order, int32_t* scanFlag, bool inheritUsOrder);
int32_t stopTableScanOperator(SOperatorInfo* pOperator, const char* pIdStr);
int32_t getOperatorExplainExecInfo(struct SOperatorInfo* operatorInfo, SArray* pExecInfoList);
#ifdef __cplusplus
}
#endif
#endif // TDENGINE_OPERATOR_H

112
source/libs/executor/inc/querytask.h Normal file
View File

@ -0,0 +1,112 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef TDENGINE_QUERYTASK_H
#define TDENGINE_QUERYTASK_H
#ifdef __cplusplus
extern "C" {
#endif
#define GET_TASKID(_t) (((SExecTaskInfo*)(_t))->id.str)
enum {
// when this task starts to execute, this status will set
TASK_NOT_COMPLETED = 0x1u,
/* Task is over
* 1. this status is used in one row result query process, e.g., count/sum/first/last/ avg...etc.
* 2. when all data within queried time window, it is also denoted as query_completed
*/
TASK_COMPLETED = 0x2u,
};
typedef struct STaskIdInfo {
uint64_t queryId; // this is also a request id
uint64_t subplanId;
uint64_t templateId;
char* str;
int32_t vgId;
} STaskIdInfo;
typedef struct STaskCostInfo {
int64_t created;
int64_t start;
uint64_t elapsedTime;
double extractListTime;
double groupIdMapTime;
SFileBlockLoadRecorder* pRecoder;
} STaskCostInfo;
typedef struct STaskStopInfo {
SRWLatch lock;
SArray* pStopInfo;
} STaskStopInfo;
typedef struct {
STqOffsetVal currentOffset; // for tmq
SMqMetaRsp metaRsp; // for tmq fetching meta
int64_t snapshotVer;
SPackedData submit; // todo remove it
SSchemaWrapper* schema;
char tbName[TSDB_TABLE_NAME_LEN]; // this is the current scan table: todo refactor
int8_t recoverStep;
int8_t recoverScanFinished;
SQueryTableDataCond tableCond;
int64_t fillHistoryVer1;
int64_t fillHistoryVer2;
SStreamState* pState;
int64_t dataVersion;
int64_t checkPointId;
} SStreamTaskInfo;
struct SExecTaskInfo {
STaskIdInfo id;
uint32_t status;
STimeWindow window;
STaskCostInfo cost;
int64_t owner; // if it is in execution
int32_t code;
int32_t qbufQuota; // total available buffer (in KB) during execution query
int64_t version; // used for stream to record wal version, why not move to sschemainfo
SStreamTaskInfo streamInfo;
SSchemaInfo schemaInfo;
const char* sql; // query sql string
jmp_buf env; // jump to this position when error happens.
EOPTR_EXEC_MODEL execModel; // operator execution model [batch model|stream model]
SSubplan* pSubplan;
struct SOperatorInfo* pRoot;
SLocalFetch localFetch;
SArray* pResultBlockList; // result block list
STaskStopInfo stopInfo;
SRWLatch lock; // secure the access of STableListInfo
};
void buildTaskId(uint64_t taskId, uint64_t queryId, char* dst);
SExecTaskInfo* doCreateTask(uint64_t queryId, uint64_t taskId, int32_t vgId, EOPTR_EXEC_MODEL model);
void doDestroyTask(SExecTaskInfo* pTaskInfo);
bool isTaskKilled(SExecTaskInfo* pTaskInfo);
void setTaskKilled(SExecTaskInfo* pTaskInfo, int32_t rspCode);
void setTaskStatus(SExecTaskInfo* pTaskInfo, int8_t status);
int32_t createExecTaskInfo(SSubplan* pPlan, SExecTaskInfo** pTaskInfo, SReadHandle* pHandle, uint64_t taskId,
int32_t vgId, char* sql, EOPTR_EXEC_MODEL model);
int32_t qAppendTaskStopInfo(SExecTaskInfo* pTaskInfo, SExchangeOpStopInfo* pInfo);
SArray* getTableListInfo(const SExecTaskInfo* pTaskInfo);
#ifdef __cplusplus
}
#endif
#endif // TDENGINE_QUERYTASK_H

15
source/libs/executor/src/aggregateoperator.c
View File

@ -15,24 +15,21 @@
#include "filter.h"
#include "function.h"
#include "functionMgt.h"
#include "os.h"
#include "querynodes.h"
#include "tfill.h"
#include "tname.h"
#include "executorInt.h"
#include "index.h"
#include "operator.h"
#include "query.h"
#include "querytask.h"
#include "tcompare.h"
#include "tdatablock.h"
#include "tglobal.h"
#include "tmsg.h"
#include "ttime.h"
#include "executorimpl.h"
#include "index.h"
#include "query.h"
#include "tcompare.h"
#include "thash.h"
#include "ttypes.h"
#include "vnode.h"
typedef struct {
bool hasAgg;

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

@ -20,7 +20,9 @@
#include "tdatablock.h"
#include "tmsg.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "operator.h"
#include "querytask.h"
#include "tcompare.h"
#include "thash.h"
#include "ttypes.h"

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

@ -15,7 +15,7 @@
#include "dataSinkInt.h"
#include "dataSinkMgt.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "planner.h"
#include "tcompression.h"
#include "tdatablock.h"
@ -179,7 +179,7 @@ static void getDataLength(SDataSinkHandle* pHandle, int64_t* pLen, bool* pQueryE
static int32_t getDataBlock(SDataSinkHandle* pHandle, SOutputData* pOutput) {
SDataDeleterHandle* pDeleter = (SDataDeleterHandle*)pHandle;
if (NULL == pDeleter->nextOutput.pData) {
assert(pDeleter->queryEnd);
ASSERT(pDeleter->queryEnd);
pOutput->useconds = pDeleter->useconds;
pOutput->precision = pDeleter->pSchema->precision;
pOutput->bufStatus = DS_BUF_EMPTY;

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

@ -15,7 +15,7 @@
#include "dataSinkInt.h"
#include "dataSinkMgt.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "planner.h"
#include "tcompression.h"
#include "tdatablock.h"
@ -181,7 +181,7 @@ static void getDataLength(SDataSinkHandle* pHandle, int64_t* pLen, bool* pQueryE
static int32_t getDataBlock(SDataSinkHandle* pHandle, SOutputData* pOutput) {
SDataDispatchHandle* pDispatcher = (SDataDispatchHandle*)pHandle;
if (NULL == pDispatcher->nextOutput.pData) {
assert(pDispatcher->queryEnd);
ASSERT(pDispatcher->queryEnd);
pOutput->useconds = pDispatcher->useconds;
pOutput->precision = pDispatcher->pSchema->precision;
pOutput->bufStatus = DS_BUF_EMPTY;

14
source/libs/executor/src/dataInserter.c
View File

@ -15,7 +15,7 @@
#include "dataSinkInt.h"
#include "dataSinkMgt.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "planner.h"
#include "tcompression.h"
#include "tdatablock.h"
@ -126,7 +126,7 @@ static int32_t submitReqToMsg(int32_t vgId, SSubmitReq2* pReq, void** pData, int
int32_t code = TSDB_CODE_SUCCESS;
int32_t len = 0;
void* pBuf = NULL;
tEncodeSize(tEncodeSSubmitReq2, pReq, len, code);
tEncodeSize(tEncodeSubmitReq, pReq, len, code);
if (TSDB_CODE_SUCCESS == code) {
SEncoder encoder;
len += sizeof(SSubmitReq2Msg);
@ -138,7 +138,7 @@ static int32_t submitReqToMsg(int32_t vgId, SSubmitReq2* pReq, void** pData, int
((SSubmitReq2Msg*)pBuf)->header.contLen = htonl(len);
((SSubmitReq2Msg*)pBuf)->version = htobe64(1);
tEncoderInit(&encoder, POINTER_SHIFT(pBuf, sizeof(SSubmitReq2Msg)), len - sizeof(SSubmitReq2Msg));
code = tEncodeSSubmitReq2(&encoder, pReq);
code = tEncodeSubmitReq(&encoder, pReq);
tEncoderClear(&encoder);
}
@ -281,7 +281,7 @@ int32_t buildSubmitReqFromBlock(SDataInserterHandle* pInserter, SSubmitReq2** pp
SRow* pRow = NULL;
if ((terrno = tRowBuild(pVals, pTSchema, &pRow)) < 0) {
tDestroySSubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
tDestroySubmitTbData(&tbData, TSDB_MSG_FLG_ENCODE);
goto _end;
}
taosArrayPush(tbData.aRowP, &pRow);
@ -301,7 +301,7 @@ _end:
if (terrno != 0) {
*ppReq = NULL;
if (pReq) {
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFree(pReq);
}
return terrno;
@ -326,7 +326,7 @@ int32_t dataBlocksToSubmitReq(SDataInserterHandle* pInserter, void** pMsg, int32
code = buildSubmitReqFromBlock(pInserter, &pReq, pDataBlock, pTSchema, uid, vgId, suid);
if (code) {
if (pReq) {
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFree(pReq);
}
@ -335,7 +335,7 @@ int32_t dataBlocksToSubmitReq(SDataInserterHandle* pInserter, void** pMsg, int32
}
code = submitReqToMsg(vgId, pReq, pMsg, msgLen);
tDestroySSubmitReq2(pReq, TSDB_MSG_FLG_ENCODE);
tDestroySubmitReq(pReq, TSDB_MSG_FLG_ENCODE);
taosMemoryFree(pReq);
return code;

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

@ -13,10 +13,12 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "executorimpl.h"
#include "executorInt.h"
#include "filter.h"
#include "function.h"
#include "functionMgt.h"
#include "operator.h"
#include "querytask.h"
#include "tcommon.h"
#include "tcompare.h"
#include "tdatablock.h"

14
source/libs/executor/src/exchangeoperator.c
View File

@ -13,17 +13,19 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "executorInt.h"
#include "filter.h"
#include "function.h"
#include "index.h"
#include "operator.h"
#include "os.h"
#include "query.h"
#include "querytask.h"
#include "tdatablock.h"
#include "thash.h"
#include "tmsg.h"
#include "tname.h"
#include "tref.h"
#include "tdatablock.h"
#include "tmsg.h"
#include "executorimpl.h"
#include "index.h"
#include "query.h"
#include "thash.h"
typedef struct SFetchRspHandleWrapper {
uint32_t exchangeId;

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

@ -24,7 +24,8 @@
#include "ttime.h"
#include "executil.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "querytask.h"
#include "tcompression.h"
typedef struct STableListIdInfo {
@ -88,7 +89,6 @@ void resetResultRow(SResultRow* pResultRow, size_t entrySize) {
// TODO refactor: use macro
SResultRowEntryInfo* getResultEntryInfo(const SResultRow* pRow, int32_t index, const int32_t* offset) {
assert(index >= 0 && offset != NULL);
return (SResultRowEntryInfo*)((char*)pRow->pEntryInfo + offset[index]);
}

74
source/libs/executor/src/executor.c
View File

@ -14,9 +14,10 @@
*/
#include "executor.h"
#include <vnode.h>
#include "executorimpl.h"
#include "executorInt.h"
#include "operator.h"
#include "planner.h"
#include "querytask.h"
#include "tdatablock.h"
#include "tref.h"
#include "tudf.h"
@ -138,7 +139,6 @@ static int32_t doSetStreamBlock(SOperatorInfo* pOperator, void* input, size_t nu
}
pInfo->blockType = STREAM_INPUT__DATA_SUBMIT;
} else if (type == STREAM_INPUT__DATA_SUBMIT) {
ASSERT(numOfBlocks == 1);
taosArrayPush(pInfo->pBlockLists, input);
pInfo->blockType = STREAM_INPUT__DATA_SUBMIT;
} else if (type == STREAM_INPUT__DATA_BLOCK) {
@ -249,7 +249,7 @@ int32_t qSetSMAInput(qTaskInfo_t tinfo, const void* pBlocks, size_t numOfBlocks,
qTaskInfo_t qCreateQueueExecTaskInfo(void* msg, SReadHandle* pReaderHandle, int32_t vgId, int32_t* numOfCols,
uint64_t id) {
if (msg == NULL) { // create raw scan
SExecTaskInfo* pTaskInfo = doCreateExecTaskInfo(0, id, vgId, OPTR_EXEC_MODEL_QUEUE, "");
SExecTaskInfo* pTaskInfo = doCreateTask(0, id, vgId, OPTR_EXEC_MODEL_QUEUE);
if (NULL == pTaskInfo) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
@ -312,7 +312,6 @@ qTaskInfo_t qCreateStreamExecTaskInfo(void* msg, SReadHandle* readers, int32_t v
qTaskInfo_t pTaskInfo = NULL;
code = qCreateExecTask(readers, vgId, 0, pPlan, &pTaskInfo, NULL, NULL, OPTR_EXEC_MODEL_STREAM);
if (code != TSDB_CODE_SUCCESS) {
nodesDestroyNode((SNode*)pPlan);
qDestroyTask(pTaskInfo);
terrno = code;
return NULL;
@ -717,8 +716,6 @@ void qRemoveTaskStopInfo(SExecTaskInfo* pTaskInfo, SExchangeOpStopInfo* pInfo) {
taosArrayRemove(pTaskInfo->stopInfo.pStopInfo, idx);
}
taosWUnLockLatch(&pTaskInfo->stopInfo.lock);
return;
}
void qStopTaskOperators(SExecTaskInfo* pTaskInfo) {
@ -802,7 +799,11 @@ void qDestroyTask(qTaskInfo_t qTaskHandle) {
return;
}
qDebug("%s execTask completed, numOfRows:%" PRId64, GET_TASKID(pTaskInfo), pTaskInfo->pRoot->resultInfo.totalRows);
if (pTaskInfo->pRoot != NULL) {
qDebug("%s execTask completed, numOfRows:%" PRId64, GET_TASKID(pTaskInfo), pTaskInfo->pRoot->resultInfo.totalRows);
} else {
qDebug("%s execTask completed", GET_TASKID(pTaskInfo));
}
printTaskExecCostInLog(pTaskInfo); // print the query cost summary
doDestroyTask(pTaskInfo);
@ -855,15 +856,6 @@ int32_t qExtractStreamScanner(qTaskInfo_t tinfo, void** scanner) {
}
}
#if 0
int32_t qStreamInput(qTaskInfo_t tinfo, void* pItem) {
SExecTaskInfo* pTaskInfo = (SExecTaskInfo*)tinfo;
ASSERT(pTaskInfo->execModel == OPTR_EXEC_MODEL_STREAM);
taosWriteQitem(pTaskInfo->streamInfo.inputQueue->queue, pItem);
return 0;
}
#endif
int32_t qStreamSourceRecoverStep1(qTaskInfo_t tinfo, int64_t ver) {
SExecTaskInfo* pTaskInfo = (SExecTaskInfo*)tinfo;
ASSERT(pTaskInfo->execModel == OPTR_EXEC_MODEL_STREAM);
@ -898,8 +890,7 @@ int32_t qStreamSetParamForRecover(qTaskInfo_t tinfo) {
SStreamIntervalOperatorInfo* pInfo = pOperator->info;
ASSERT(pInfo->twAggSup.calTrigger == STREAM_TRIGGER_AT_ONCE ||
pInfo->twAggSup.calTrigger == STREAM_TRIGGER_WINDOW_CLOSE);
ASSERT(pInfo->twAggSup.calTriggerSaved == 0);
ASSERT(pInfo->twAggSup.deleteMarkSaved == 0);
ASSERT(pInfo->twAggSup.calTriggerSaved == 0 && pInfo->twAggSup.deleteMarkSaved == 0);
qInfo("save stream param for interval: %d, %" PRId64, pInfo->twAggSup.calTrigger, pInfo->twAggSup.deleteMark);
@ -915,9 +906,8 @@ int32_t qStreamSetParamForRecover(qTaskInfo_t tinfo) {
SStreamSessionAggOperatorInfo* pInfo = pOperator->info;
ASSERT(pInfo->twAggSup.calTrigger == STREAM_TRIGGER_AT_ONCE ||
pInfo->twAggSup.calTrigger == STREAM_TRIGGER_WINDOW_CLOSE);
ASSERT(pInfo->twAggSup.calTriggerSaved == 0);
ASSERT(pInfo->twAggSup.deleteMarkSaved == 0);
ASSERT(pInfo->twAggSup.calTriggerSaved == 0 && pInfo->twAggSup.deleteMarkSaved == 0);
qInfo("save stream param for session: %d, %" PRId64, pInfo->twAggSup.calTrigger, pInfo->twAggSup.deleteMark);
pInfo->twAggSup.calTriggerSaved = pInfo->twAggSup.calTrigger;
@ -930,8 +920,7 @@ int32_t qStreamSetParamForRecover(qTaskInfo_t tinfo) {
SStreamStateAggOperatorInfo* pInfo = pOperator->info;
ASSERT(pInfo->twAggSup.calTrigger == STREAM_TRIGGER_AT_ONCE ||
pInfo->twAggSup.calTrigger == STREAM_TRIGGER_WINDOW_CLOSE);
ASSERT(pInfo->twAggSup.calTriggerSaved == 0);
ASSERT(pInfo->twAggSup.deleteMarkSaved == 0);
ASSERT(pInfo->twAggSup.calTriggerSaved == 0 && pInfo->twAggSup.deleteMarkSaved == 0);
qInfo("save stream param for state: %d, %" PRId64, pInfo->twAggSup.calTrigger, pInfo->twAggSup.deleteMark);
@ -992,7 +981,6 @@ int32_t qStreamRestoreParam(qTaskInfo_t tinfo) {
if (pOperator->numOfDownstream != 1 || pOperator->pDownstream[0] == NULL) {
if (pOperator->numOfDownstream > 1) {
qError("unexpected stream, multiple downstream");
/*ASSERT(0);*/
return -1;
}
return 0;
@ -1064,19 +1052,6 @@ int32_t initQueryTableDataCondForTmq(SQueryTableDataCond* pCond, SSnapContext* s
return TSDB_CODE_SUCCESS;
}
int32_t qStreamSetScanMemData(qTaskInfo_t tinfo, SPackedData submit) {
SExecTaskInfo* pTaskInfo = (SExecTaskInfo*)tinfo;
if ((pTaskInfo->execModel != OPTR_EXEC_MODEL_QUEUE) || (pTaskInfo->streamInfo.submit.msgStr != NULL)) {
qError("qStreamSetScanMemData err:%d,%p", pTaskInfo->execModel, pTaskInfo->streamInfo.submit.msgStr);
terrno = TSDB_CODE_PAR_INTERNAL_ERROR;
return -1;
}
qDebug("set the submit block for future scan");
pTaskInfo->streamInfo.submit = submit;
return 0;
}
void qStreamSetOpen(qTaskInfo_t tinfo) {
SExecTaskInfo* pTaskInfo = (SExecTaskInfo*)tinfo;
SOperatorInfo* pOperator = pTaskInfo->pRoot;
@ -1098,6 +1073,7 @@ int32_t qStreamPrepareScan(qTaskInfo_t tinfo, STqOffsetVal* pOffset, int8_t subT
if (pOperator == NULL) {
return -1;
}
SStreamScanInfo* pInfo = pOperator->info;
STableScanInfo* pScanInfo = pInfo->pTableScanOp->info;
STableScanBase* pScanBaseInfo = &pScanInfo->base;
@ -1233,7 +1209,7 @@ int32_t qStreamPrepareScan(qTaskInfo_t tinfo, STqOffsetVal* pOffset, int8_t subT
cleanupQueryTableDataCond(&pTaskInfo->streamInfo.tableCond);
strcpy(pTaskInfo->streamInfo.tbName, mtInfo.tbName);
tDeleteSSchemaWrapper(pTaskInfo->streamInfo.schema);
tDeleteSchemaWrapper(pTaskInfo->streamInfo.schema);
pTaskInfo->streamInfo.schema = mtInfo.schema;
qDebug("tmqsnap qStreamPrepareScan snapshot data uid:%" PRId64 " ts %" PRId64 " %s", mtInfo.uid, pOffset->ts, id);
@ -1303,3 +1279,25 @@ SArray* qGetQueriedTableListInfo(qTaskInfo_t tinfo) {
taosArrayDestroy(plist);
return pUidList;
}
static void extractTableList(SArray* pList, const SOperatorInfo* pOperator) {
if (pOperator->operatorType == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN) {
SStreamScanInfo* pScanInfo = pOperator->info;
STableScanInfo* pTableScanInfo = pScanInfo->pTableScanOp->info;
taosArrayPush(pList, &pTableScanInfo->base.pTableListInfo);
} else if (pOperator->operatorType == QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
STableScanInfo* pScanInfo = pOperator->info;
taosArrayPush(pList, &pScanInfo->base.pTableListInfo);
} else {
if (pOperator->pDownstream != NULL && pOperator->pDownstream[0] != NULL) {
extractTableList(pList, pOperator->pDownstream[0]);
}
}
}
SArray* getTableListInfo(const SExecTaskInfo* pTaskInfo) {
SArray* pArray = taosArrayInit(0, POINTER_BYTES);
SOperatorInfo* pOperator = pTaskInfo->pRoot;
extractTableList(pArray, pOperator);
return pArray;
}

735
source/libs/executor/src/executorimpl.c → source/libs/executor/src/executorInt.c
View File

@ -22,13 +22,14 @@
#include "tname.h"
#include "tdatablock.h"
#include "tglobal.h"
#include "tmsg.h"
#include "ttime.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "index.h"
#include "operator.h"
#include "query.h"
#include "querytask.h"
#include "tcompare.h"
#include "thash.h"
#include "ttypes.h"
@ -71,12 +72,8 @@ static UNUSED_FUNC void* u_realloc(void* p, size_t __size) {
#define realloc u_realloc
#endif
#define CLEAR_QUERY_STATUS(q, st) ((q)->status &= (~(st)))
static void setBlockSMAInfo(SqlFunctionCtx* pCtx, SExprInfo* pExpr, SSDataBlock* pBlock);
static void releaseQueryBuf(size_t numOfTables);
static void initCtxOutputBuffer(SqlFunctionCtx* pCtx, int32_t size);
static void doApplyScalarCalculation(SOperatorInfo* pOperator, SSDataBlock* pBlock, int32_t order, int32_t scanFlag);
@ -86,44 +83,6 @@ static int32_t doSetInputDataBlock(SExprSupp* pExprSup, SSDataBlock* pBlock, int
bool createDummyCol);
static int32_t doCopyToSDataBlock(SExecTaskInfo* pTaskInfo, SSDataBlock* pBlock, SExprSupp* pSup, SDiskbasedBuf* pBuf,
SGroupResInfo* pGroupResInfo);
static SSchemaWrapper* extractQueriedColumnSchema(SScanPhysiNode* pScanNode);
void setOperatorCompleted(SOperatorInfo* pOperator) {
pOperator->status = OP_EXEC_DONE;
pOperator->cost.totalCost = (taosGetTimestampUs() - pOperator->pTaskInfo->cost.start) / 1000.0;
setTaskStatus(pOperator->pTaskInfo, TASK_COMPLETED);
}
void setOperatorInfo(SOperatorInfo* pOperator, const char* name, int32_t type, bool blocking, int32_t status,
void* pInfo, SExecTaskInfo* pTaskInfo) {
pOperator->name = (char*)name;
pOperator->operatorType = type;
pOperator->blocking = blocking;
pOperator->status = status;
pOperator->info = pInfo;
pOperator->pTaskInfo = pTaskInfo;
}
int32_t optrDummyOpenFn(SOperatorInfo* pOperator) {
OPTR_SET_OPENED(pOperator);
pOperator->cost.openCost = 0;
return TSDB_CODE_SUCCESS;
}
SOperatorFpSet createOperatorFpSet(__optr_open_fn_t openFn, __optr_fn_t nextFn, __optr_fn_t cleanup,
__optr_close_fn_t closeFn, __optr_reqBuf_fn_t reqBufFn,
__optr_explain_fn_t explain) {
SOperatorFpSet fpSet = {
._openFn = openFn,
.getNextFn = nextFn,
.cleanupFn = cleanup,
.closeFn = closeFn,
.reqBufFn = reqBufFn,
.getExplainFn = explain,
};
return fpSet;
}
SResultRow* getNewResultRow(SDiskbasedBuf* pResultBuf, int32_t* currentPageId, int32_t interBufSize) {
SFilePage* pData = NULL;
@ -482,10 +441,6 @@ void setBlockSMAInfo(SqlFunctionCtx* pCtx, SExprInfo* pExprInfo, SSDataBlock* pB
}
}
bool isTaskKilled(SExecTaskInfo* pTaskInfo) { return (0 != pTaskInfo->code); }
void setTaskKilled(SExecTaskInfo* pTaskInfo, int32_t rspCode) { pTaskInfo->code = rspCode; }
/////////////////////////////////////////////////////////////////////////////////////////////
STimeWindow getAlignQueryTimeWindow(SInterval* pInterval, int32_t precision, int64_t key) {
STimeWindow win = {0};
@ -503,16 +458,6 @@ STimeWindow getAlignQueryTimeWindow(SInterval* pInterval, int32_t precision, int
return win;
}
void setTaskStatus(SExecTaskInfo* pTaskInfo, int8_t status) {
if (status == TASK_NOT_COMPLETED) {
pTaskInfo->status = status;
} else {
// QUERY_NOT_COMPLETED is not compatible with any other status, so clear its position first
CLEAR_QUERY_STATUS(pTaskInfo, TASK_NOT_COMPLETED);
pTaskInfo->status |= status;
}
}
void setResultRowInitCtx(SResultRow* pResult, SqlFunctionCtx* pCtx, int32_t numOfOutput, int32_t* rowEntryInfoOffset) {
bool init = false;
for (int32_t i = 0; i < numOfOutput; ++i) {
@ -949,72 +894,6 @@ void doBuildResultDatablock(SOperatorInfo* pOperator, SOptrBasicInfo* pbInfo, SG
}
}
int32_t appendDownstream(SOperatorInfo* p, SOperatorInfo** pDownstream, int32_t num) {
p->pDownstream = taosMemoryCalloc(1, num * POINTER_BYTES);
if (p->pDownstream == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
}
memcpy(p->pDownstream, pDownstream, num * POINTER_BYTES);
p->numOfDownstream = num;
return TSDB_CODE_SUCCESS;
}
int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t* order, int32_t* scanFlag, bool inheritUsOrder) {
// todo add more information about exchange operation
int32_t type = pOperator->operatorType;
if (type == QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN || type == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN ||
type == QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN || type == QUERY_NODE_PHYSICAL_PLAN_BLOCK_DIST_SCAN ||
type == QUERY_NODE_PHYSICAL_PLAN_LAST_ROW_SCAN || type == QUERY_NODE_PHYSICAL_PLAN_TABLE_COUNT_SCAN) {
*order = TSDB_ORDER_ASC;
*scanFlag = MAIN_SCAN;
return TSDB_CODE_SUCCESS;
} else if (type == QUERY_NODE_PHYSICAL_PLAN_EXCHANGE) {
if (!inheritUsOrder) {
*order = TSDB_ORDER_ASC;
}
*scanFlag = MAIN_SCAN;
return TSDB_CODE_SUCCESS;
} else if (type == QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
STableScanInfo* pTableScanInfo = pOperator->info;
*order = pTableScanInfo->base.cond.order;
*scanFlag = pTableScanInfo->base.scanFlag;
return TSDB_CODE_SUCCESS;
} else if (type == QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN) {
STableMergeScanInfo* pTableScanInfo = pOperator->info;
*order = pTableScanInfo->base.cond.order;
*scanFlag = pTableScanInfo->base.scanFlag;
return TSDB_CODE_SUCCESS;
} else {
if (pOperator->pDownstream == NULL || pOperator->pDownstream[0] == NULL) {
return TSDB_CODE_INVALID_PARA;
} else {
return getTableScanInfo(pOperator->pDownstream[0], order, scanFlag, inheritUsOrder);
}
}
}
// QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN
SOperatorInfo* extractOperatorInTree(SOperatorInfo* pOperator, int32_t type, const char* id) {
if (pOperator == NULL) {
qError("invalid operator, failed to find tableScanOperator %s", id);
terrno = TSDB_CODE_PAR_INTERNAL_ERROR;
return NULL;
}
if (pOperator->operatorType == type) {
return pOperator;
} else {
if (pOperator->pDownstream == NULL || pOperator->pDownstream[0] == NULL) {
qError("invalid operator, failed to find tableScanOperator %s", id);
terrno = TSDB_CODE_PAR_INTERNAL_ERROR;
return NULL;
}
return extractOperatorInTree(pOperator->pDownstream[0], type, id);
}
}
void destroyExprInfo(SExprInfo* pExpr, int32_t numOfExprs) {
for (int32_t i = 0; i < numOfExprs; ++i) {
SExprInfo* pExprInfo = &pExpr[i];
@ -1031,37 +910,6 @@ void destroyExprInfo(SExprInfo* pExpr, int32_t numOfExprs) {
}
}
void destroyOperatorInfo(SOperatorInfo* pOperator) {
if (pOperator == NULL) {
return;
}
if (pOperator->fpSet.closeFn != NULL) {
pOperator->fpSet.closeFn(pOperator->info);
}
if (pOperator->pDownstream != NULL) {
for (int32_t i = 0; i < pOperator->numOfDownstream; ++i) {
destroyOperatorInfo(pOperator->pDownstream[i]);
}
taosMemoryFreeClear(pOperator->pDownstream);
pOperator->numOfDownstream = 0;
}
cleanupExprSupp(&pOperator->exprSupp);
taosMemoryFreeClear(pOperator);
}
// each operator should be set their own function to return total cost buffer
int32_t optrDefaultBufFn(SOperatorInfo* pOperator) {
if (pOperator->blocking) {
return -1;
} else {
return 0;
}
}
int32_t getBufferPgSize(int32_t rowSize, uint32_t* defaultPgsz, uint32_t* defaultBufsz) {
*defaultPgsz = 4096;
while (*defaultPgsz < rowSize * 4) {
@ -1151,136 +999,6 @@ void cleanupExprSupp(SExprSupp* pSupp) {
void cleanupBasicInfo(SOptrBasicInfo* pInfo) { pInfo->pRes = blockDataDestroy(pInfo->pRes); }
void buildTaskId(uint64_t taskId, uint64_t queryId, char* dst) {
char* p = dst;
int32_t offset = 6;
memcpy(p, "TID:0x", offset);
offset += tintToHex(taskId, &p[offset]);
memcpy(&p[offset], " QID:0x", 7);
offset += 7;
offset += tintToHex(queryId, &p[offset]);
p[offset] = 0;
}
SExecTaskInfo* doCreateExecTaskInfo(uint64_t queryId, uint64_t taskId, int32_t vgId, EOPTR_EXEC_MODEL model,
char* dbFName) {
SExecTaskInfo* pTaskInfo = taosMemoryCalloc(1, sizeof(SExecTaskInfo));
if (pTaskInfo == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
}
setTaskStatus(pTaskInfo, TASK_NOT_COMPLETED);
pTaskInfo->cost.created = taosGetTimestampUs();
pTaskInfo->schemaInfo.dbname = taosStrdup(dbFName);
pTaskInfo->execModel = model;
pTaskInfo->stopInfo.pStopInfo = taosArrayInit(4, sizeof(SExchangeOpStopInfo));
pTaskInfo->pResultBlockList = taosArrayInit(128, POINTER_BYTES);
taosInitRWLatch(&pTaskInfo->lock);
pTaskInfo->id.vgId = vgId;
pTaskInfo->id.queryId = queryId;
pTaskInfo->id.str = taosMemoryMalloc(64);
buildTaskId(taskId, queryId, pTaskInfo->id.str);
return pTaskInfo;
}
int32_t extractTableSchemaInfo(SReadHandle* pHandle, SScanPhysiNode* pScanNode, SExecTaskInfo* pTaskInfo) {
SMetaReader mr = {0};
if (pHandle == NULL) {
terrno = TSDB_CODE_INVALID_PARA;
return terrno;
}
metaReaderInit(&mr, pHandle->meta, 0);
int32_t code = metaGetTableEntryByUidCache(&mr, pScanNode->uid);
if (code != TSDB_CODE_SUCCESS) {
qError("failed to get the table meta, uid:0x%" PRIx64 ", suid:0x%" PRIx64 ", %s", pScanNode->uid, pScanNode->suid,
GET_TASKID(pTaskInfo));
metaReaderClear(&mr);
return terrno;
}
SSchemaInfo* pSchemaInfo = &pTaskInfo->schemaInfo;
pSchemaInfo->tablename = taosStrdup(mr.me.name);
if (mr.me.type == TSDB_SUPER_TABLE) {
pSchemaInfo->sw = tCloneSSchemaWrapper(&mr.me.stbEntry.schemaRow);
pSchemaInfo->tversion = mr.me.stbEntry.schemaTag.version;
} else if (mr.me.type == TSDB_CHILD_TABLE) {
tDecoderClear(&mr.coder);
tb_uid_t suid = mr.me.ctbEntry.suid;
code = metaGetTableEntryByUidCache(&mr, suid);
if (code != TSDB_CODE_SUCCESS) {
metaReaderClear(&mr);
return terrno;
}
pSchemaInfo->sw = tCloneSSchemaWrapper(&mr.me.stbEntry.schemaRow);
pSchemaInfo->tversion = mr.me.stbEntry.schemaTag.version;
} else {
pSchemaInfo->sw = tCloneSSchemaWrapper(&mr.me.ntbEntry.schemaRow);
}
metaReaderClear(&mr);
pSchemaInfo->qsw = extractQueriedColumnSchema(pScanNode);
return TSDB_CODE_SUCCESS;
}
SSchemaWrapper* extractQueriedColumnSchema(SScanPhysiNode* pScanNode) {
int32_t numOfCols = LIST_LENGTH(pScanNode->pScanCols);
int32_t numOfTags = LIST_LENGTH(pScanNode->pScanPseudoCols);
SSchemaWrapper* pqSw = taosMemoryCalloc(1, sizeof(SSchemaWrapper));
pqSw->pSchema = taosMemoryCalloc(numOfCols + numOfTags, sizeof(SSchema));
for (int32_t i = 0; i < numOfCols; ++i) {
STargetNode* pNode = (STargetNode*)nodesListGetNode(pScanNode->pScanCols, i);
SColumnNode* pColNode = (SColumnNode*)pNode->pExpr;
SSchema* pSchema = &pqSw->pSchema[pqSw->nCols++];
pSchema->colId = pColNode->colId;
pSchema->type = pColNode->node.resType.type;
pSchema->bytes = pColNode->node.resType.bytes;
tstrncpy(pSchema->name, pColNode->colName, tListLen(pSchema->name));
}
// this the tags and pseudo function columns, we only keep the tag columns
for (int32_t i = 0; i < numOfTags; ++i) {
STargetNode* pNode = (STargetNode*)nodesListGetNode(pScanNode->pScanPseudoCols, i);
int32_t type = nodeType(pNode->pExpr);
if (type == QUERY_NODE_COLUMN) {
SColumnNode* pColNode = (SColumnNode*)pNode->pExpr;
SSchema* pSchema = &pqSw->pSchema[pqSw->nCols++];
pSchema->colId = pColNode->colId;
pSchema->type = pColNode->node.resType.type;
pSchema->bytes = pColNode->node.resType.bytes;
tstrncpy(pSchema->name, pColNode->colName, tListLen(pSchema->name));
}
}
return pqSw;
}
static void cleanupTableSchemaInfo(SSchemaInfo* pSchemaInfo) {
taosMemoryFreeClear(pSchemaInfo->dbname);
taosMemoryFreeClear(pSchemaInfo->tablename);
tDeleteSSchemaWrapper(pSchemaInfo->sw);
tDeleteSSchemaWrapper(pSchemaInfo->qsw);
}
static void cleanupStreamInfo(SStreamTaskInfo* pStreamInfo) { tDeleteSSchemaWrapper(pStreamInfo->schema); }
bool groupbyTbname(SNodeList* pGroupList) {
bool bytbname = false;
if (LIST_LENGTH(pGroupList) == 1) {
@ -1294,306 +1012,6 @@ bool groupbyTbname(SNodeList* pGroupList) {
return bytbname;
}
SOperatorInfo* createOperatorTree(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, SReadHandle* pHandle, SNode* pTagCond,
SNode* pTagIndexCond, const char* pUser) {
int32_t type = nodeType(pPhyNode);
const char* idstr = GET_TASKID(pTaskInfo);
if (pPhyNode->pChildren == NULL || LIST_LENGTH(pPhyNode->pChildren) == 0) {
SOperatorInfo* pOperator = NULL;
if (QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN == type) {
STableScanPhysiNode* pTableScanNode = (STableScanPhysiNode*)pPhyNode;
// NOTE: this is an patch to fix the physical plan
// TODO remove it later
if (pTableScanNode->scan.node.pLimit != NULL) {
pTableScanNode->groupSort = true;
}
STableListInfo* pTableListInfo = tableListCreate();
int32_t code =
createScanTableListInfo(&pTableScanNode->scan, pTableScanNode->pGroupTags, pTableScanNode->groupSort, pHandle,
pTableListInfo, pTagCond, pTagIndexCond, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
qError("failed to createScanTableListInfo, code:%s, %s", tstrerror(code), idstr);
return NULL;
}
code = extractTableSchemaInfo(pHandle, &pTableScanNode->scan, pTaskInfo);
if (code) {
pTaskInfo->code = terrno;
tableListDestroy(pTableListInfo);
return NULL;
}
pOperator = createTableScanOperatorInfo(pTableScanNode, pHandle, pTableListInfo, pTaskInfo);
if (NULL == pOperator) {
pTaskInfo->code = terrno;
return NULL;
}
STableScanInfo* pScanInfo = pOperator->info;
pTaskInfo->cost.pRecoder = &pScanInfo->base.readRecorder;
} else if (QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN == type) {
STableMergeScanPhysiNode* pTableScanNode = (STableMergeScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
int32_t code = createScanTableListInfo(&pTableScanNode->scan, pTableScanNode->pGroupTags, true, pHandle,
pTableListInfo, pTagCond, pTagIndexCond, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
qError("failed to createScanTableListInfo, code: %s", tstrerror(code));
return NULL;
}
code = extractTableSchemaInfo(pHandle, &pTableScanNode->scan, pTaskInfo);
if (code) {
pTaskInfo->code = terrno;
tableListDestroy(pTableListInfo);
return NULL;
}
pOperator = createTableMergeScanOperatorInfo(pTableScanNode, pHandle, pTableListInfo, pTaskInfo);
if (NULL == pOperator) {
pTaskInfo->code = terrno;
tableListDestroy(pTableListInfo);
return NULL;
}
STableScanInfo* pScanInfo = pOperator->info;
pTaskInfo->cost.pRecoder = &pScanInfo->base.readRecorder;
} else if (QUERY_NODE_PHYSICAL_PLAN_EXCHANGE == type) {
pOperator = createExchangeOperatorInfo(pHandle ? pHandle->pMsgCb->clientRpc : NULL, (SExchangePhysiNode*)pPhyNode,
pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN == type) {
STableScanPhysiNode* pTableScanNode = (STableScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
if (pHandle->vnode) {
int32_t code =
createScanTableListInfo(&pTableScanNode->scan, pTableScanNode->pGroupTags, pTableScanNode->groupSort,
pHandle, pTableListInfo, pTagCond, pTagIndexCond, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
qError("failed to createScanTableListInfo, code: %s", tstrerror(code));
return NULL;
}
}
pTaskInfo->schemaInfo.qsw = extractQueriedColumnSchema(&pTableScanNode->scan);
pOperator = createStreamScanOperatorInfo(pHandle, pTableScanNode, pTagCond, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN == type) {
SSystemTableScanPhysiNode* pSysScanPhyNode = (SSystemTableScanPhysiNode*)pPhyNode;
pOperator = createSysTableScanOperatorInfo(pHandle, pSysScanPhyNode, pUser, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_TABLE_COUNT_SCAN == type) {
STableCountScanPhysiNode* pTblCountScanNode = (STableCountScanPhysiNode*)pPhyNode;
pOperator = createTableCountScanOperatorInfo(pHandle, pTblCountScanNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN == type) {
STagScanPhysiNode* pScanPhyNode = (STagScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
int32_t code = createScanTableListInfo(pScanPhyNode, NULL, false, pHandle, pTableListInfo, pTagCond,
pTagIndexCond, pTaskInfo);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = code;
qError("failed to getTableList, code: %s", tstrerror(code));
return NULL;
}
pOperator = createTagScanOperatorInfo(pHandle, pScanPhyNode, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_BLOCK_DIST_SCAN == type) {
SBlockDistScanPhysiNode* pBlockNode = (SBlockDistScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
if (pBlockNode->tableType == TSDB_SUPER_TABLE) {
SArray* pList = taosArrayInit(4, sizeof(STableKeyInfo));
int32_t code = vnodeGetAllTableList(pHandle->vnode, pBlockNode->uid, pList);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = terrno;
return NULL;
}
size_t num = taosArrayGetSize(pList);
for (int32_t i = 0; i < num; ++i) {
STableKeyInfo* p = taosArrayGet(pList, i);
tableListAddTableInfo(pTableListInfo, p->uid, 0);
}
taosArrayDestroy(pList);
} else { // Create group with only one table
tableListAddTableInfo(pTableListInfo, pBlockNode->uid, 0);
}
pOperator = createDataBlockInfoScanOperator(pHandle, pBlockNode, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_LAST_ROW_SCAN == type) {
SLastRowScanPhysiNode* pScanNode = (SLastRowScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
int32_t code = createScanTableListInfo(&pScanNode->scan, pScanNode->pGroupTags, true, pHandle, pTableListInfo,
pTagCond, pTagIndexCond, pTaskInfo);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = code;
return NULL;
}
code = extractTableSchemaInfo(pHandle, &pScanNode->scan, pTaskInfo);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = code;
return NULL;
}
pOperator = createCacherowsScanOperator(pScanNode, pHandle, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_PROJECT == type) {
pOperator = createProjectOperatorInfo(NULL, (SProjectPhysiNode*)pPhyNode, pTaskInfo);
} else {
terrno = TSDB_CODE_INVALID_PARA;
return NULL;
}
if (pOperator != NULL) { // todo moved away
pOperator->resultDataBlockId = pPhyNode->pOutputDataBlockDesc->dataBlockId;
}
return pOperator;
}
size_t size = LIST_LENGTH(pPhyNode->pChildren);
SOperatorInfo** ops = taosMemoryCalloc(size, POINTER_BYTES);
if (ops == NULL) {
return NULL;
}
for (int32_t i = 0; i < size; ++i) {
SPhysiNode* pChildNode = (SPhysiNode*)nodesListGetNode(pPhyNode->pChildren, i);
ops[i] = createOperatorTree(pChildNode, pTaskInfo, pHandle, pTagCond, pTagIndexCond, pUser);
if (ops[i] == NULL) {
taosMemoryFree(ops);
return NULL;
}
}
SOperatorInfo* pOptr = NULL;
if (QUERY_NODE_PHYSICAL_PLAN_PROJECT == type) {
pOptr = createProjectOperatorInfo(ops[0], (SProjectPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_HASH_AGG == type) {
SAggPhysiNode* pAggNode = (SAggPhysiNode*)pPhyNode;
if (pAggNode->pGroupKeys != NULL) {
pOptr = createGroupOperatorInfo(ops[0], pAggNode, pTaskInfo);
} else {
pOptr = createAggregateOperatorInfo(ops[0], pAggNode, pTaskInfo);
}
} else if (QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL == type) {
SIntervalPhysiNode* pIntervalPhyNode = (SIntervalPhysiNode*)pPhyNode;
pOptr = createIntervalOperatorInfo(ops[0], pIntervalPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL == type) {
pOptr = createStreamIntervalOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL == type) {
SMergeAlignedIntervalPhysiNode* pIntervalPhyNode = (SMergeAlignedIntervalPhysiNode*)pPhyNode;
pOptr = createMergeAlignedIntervalOperatorInfo(ops[0], pIntervalPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_INTERVAL == type) {
SMergeIntervalPhysiNode* pIntervalPhyNode = (SMergeIntervalPhysiNode*)pPhyNode;
pOptr = createMergeIntervalOperatorInfo(ops[0], pIntervalPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_INTERVAL == type) {
int32_t children = 0;
pOptr = createStreamFinalIntervalOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_INTERVAL == type) {
int32_t children = pHandle->numOfVgroups;
pOptr = createStreamFinalIntervalOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_SORT == type) {
pOptr = createSortOperatorInfo(ops[0], (SSortPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT == type) {
pOptr = createGroupSortOperatorInfo(ops[0], (SGroupSortPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE == type) {
SMergePhysiNode* pMergePhyNode = (SMergePhysiNode*)pPhyNode;
pOptr = createMultiwayMergeOperatorInfo(ops, size, pMergePhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION == type) {
SSessionWinodwPhysiNode* pSessionNode = (SSessionWinodwPhysiNode*)pPhyNode;
pOptr = createSessionAggOperatorInfo(ops[0], pSessionNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION == type) {
pOptr = createStreamSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION == type) {
int32_t children = 0;
pOptr = createStreamFinalSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION == type) {
int32_t children = pHandle->numOfVgroups;
pOptr = createStreamFinalSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_PARTITION == type) {
pOptr = createPartitionOperatorInfo(ops[0], (SPartitionPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_PARTITION == type) {
pOptr = createStreamPartitionOperatorInfo(ops[0], (SStreamPartitionPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE == type) {
SStateWinodwPhysiNode* pStateNode = (SStateWinodwPhysiNode*)pPhyNode;
pOptr = createStatewindowOperatorInfo(ops[0], pStateNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE == type) {
pOptr = createStreamStateAggOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_JOIN == type) {
pOptr = createMergeJoinOperatorInfo(ops, size, (SSortMergeJoinPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_FILL == type) {
pOptr = createFillOperatorInfo(ops[0], (SFillPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FILL == type) {
pOptr = createStreamFillOperatorInfo(ops[0], (SStreamFillPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_INDEF_ROWS_FUNC == type) {
pOptr = createIndefinitOutputOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_INTERP_FUNC == type) {
pOptr = createTimeSliceOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_EVENT == type) {
pOptr = createEventwindowOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else {
terrno = TSDB_CODE_INVALID_PARA;
taosMemoryFree(ops);
return NULL;
}
taosMemoryFree(ops);
if (pOptr) {
pOptr->resultDataBlockId = pPhyNode->pOutputDataBlockDesc->dataBlockId;
}
return pOptr;
}
static int32_t extractTbscanInStreamOpTree(SOperatorInfo* pOperator, STableScanInfo** ppInfo) {
if (pOperator->operatorType != QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN) {
if (pOperator->numOfDownstream == 0) {
qError("failed to find stream scan operator");
return TSDB_CODE_APP_ERROR;
}
if (pOperator->numOfDownstream > 1) {
qError("join not supported for stream block scan");
return TSDB_CODE_APP_ERROR;
}
return extractTbscanInStreamOpTree(pOperator->pDownstream[0], ppInfo);
} else {
SStreamScanInfo* pInfo = pOperator->info;
*ppInfo = pInfo->pTableScanOp->info;
return 0;
}
}
int32_t extractTableScanNode(SPhysiNode* pNode, STableScanPhysiNode** ppNode) {
if (pNode->pChildren == NULL || LIST_LENGTH(pNode->pChildren) == 0) {
if (QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN == pNode->type) {
*ppNode = (STableScanPhysiNode*)pNode;
return 0;
} else {
terrno = TSDB_CODE_APP_ERROR;
return -1;
}
} else {
if (LIST_LENGTH(pNode->pChildren) != 1) {
terrno = TSDB_CODE_APP_ERROR;
return -1;
}
SPhysiNode* pChildNode = (SPhysiNode*)nodesListGetNode(pNode->pChildren, 0);
return extractTableScanNode(pChildNode, ppNode);
}
return -1;
}
int32_t createDataSinkParam(SDataSinkNode* pNode, void** pParam, SExecTaskInfo* pTask, SReadHandle* readHandle) {
switch (pNode->type) {
case QUERY_NODE_PHYSICAL_PLAN_QUERY_INSERT: {
@ -1641,131 +1059,6 @@ int32_t createDataSinkParam(SDataSinkNode* pNode, void** pParam, SExecTaskInfo*
return TSDB_CODE_SUCCESS;
}
int32_t createExecTaskInfo(SSubplan* pPlan, SExecTaskInfo** pTaskInfo, SReadHandle* pHandle, uint64_t taskId,
int32_t vgId, char* sql, EOPTR_EXEC_MODEL model) {
*pTaskInfo = doCreateExecTaskInfo(pPlan->id.queryId, taskId, vgId, model, pPlan->dbFName);
if (*pTaskInfo == NULL) {
goto _complete;
}
if (pHandle) {
if (pHandle->pStateBackend) {
(*pTaskInfo)->streamInfo.pState = pHandle->pStateBackend;
}
}
(*pTaskInfo)->sql = sql;
sql = NULL;
(*pTaskInfo)->pSubplan = pPlan;
(*pTaskInfo)->pRoot =
createOperatorTree(pPlan->pNode, *pTaskInfo, pHandle, pPlan->pTagCond, pPlan->pTagIndexCond, pPlan->user);
if (NULL == (*pTaskInfo)->pRoot) {
terrno = (*pTaskInfo)->code;
goto _complete;
}
return TSDB_CODE_SUCCESS;
_complete:
taosMemoryFree(sql);
doDestroyTask(*pTaskInfo);
return terrno;
}
static void freeBlock(void* pParam) {
SSDataBlock* pBlock = *(SSDataBlock**)pParam;
blockDataDestroy(pBlock);
}
void doDestroyTask(SExecTaskInfo* pTaskInfo) {
qDebug("%s execTask is freed", GET_TASKID(pTaskInfo));
destroyOperatorInfo(pTaskInfo->pRoot);
cleanupTableSchemaInfo(&pTaskInfo->schemaInfo);
cleanupStreamInfo(&pTaskInfo->streamInfo);
if (!pTaskInfo->localFetch.localExec) {
nodesDestroyNode((SNode*)pTaskInfo->pSubplan);
}
taosArrayDestroyEx(pTaskInfo->pResultBlockList, freeBlock);
taosArrayDestroy(pTaskInfo->stopInfo.pStopInfo);
taosMemoryFreeClear(pTaskInfo->sql);
taosMemoryFreeClear(pTaskInfo->id.str);
taosMemoryFreeClear(pTaskInfo);
}
static int64_t getQuerySupportBufSize(size_t numOfTables) {
size_t s1 = sizeof(STableQueryInfo);
// size_t s3 = sizeof(STableCheckInfo); buffer consumption in tsdb
return (int64_t)(s1 * 1.5 * numOfTables);
}
int32_t checkForQueryBuf(size_t numOfTables) {
int64_t t = getQuerySupportBufSize(numOfTables);
if (tsQueryBufferSizeBytes < 0) {
return TSDB_CODE_SUCCESS;
} else if (tsQueryBufferSizeBytes > 0) {
while (1) {
int64_t s = tsQueryBufferSizeBytes;
int64_t remain = s - t;
if (remain >= 0) {
if (atomic_val_compare_exchange_64(&tsQueryBufferSizeBytes, s, remain) == s) {
return TSDB_CODE_SUCCESS;
}
} else {
return TSDB_CODE_QRY_NOT_ENOUGH_BUFFER;
}
}
}
// disable query processing if the value of tsQueryBufferSize is zero.
return TSDB_CODE_QRY_NOT_ENOUGH_BUFFER;
}
void releaseQueryBuf(size_t numOfTables) {
if (tsQueryBufferSizeBytes < 0) {
return;
}
int64_t t = getQuerySupportBufSize(numOfTables);
// restore value is not enough buffer available
atomic_add_fetch_64(&tsQueryBufferSizeBytes, t);
}
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SArray* pExecInfoList) {
SExplainExecInfo execInfo = {0};
SExplainExecInfo* pExplainInfo = taosArrayPush(pExecInfoList, &execInfo);
pExplainInfo->numOfRows = operatorInfo->resultInfo.totalRows;
pExplainInfo->startupCost = operatorInfo->cost.openCost;
pExplainInfo->totalCost = operatorInfo->cost.totalCost;
pExplainInfo->verboseLen = 0;
pExplainInfo->verboseInfo = NULL;
if (operatorInfo->fpSet.getExplainFn) {
int32_t code =
operatorInfo->fpSet.getExplainFn(operatorInfo, &pExplainInfo->verboseInfo, &pExplainInfo->verboseLen);
if (code) {
qError("%s operator getExplainFn failed, code:%s", GET_TASKID(operatorInfo->pTaskInfo), tstrerror(code));
return code;
}
}
int32_t code = 0;
for (int32_t i = 0; i < operatorInfo->numOfDownstream; ++i) {
code = getOperatorExplainExecInfo(operatorInfo->pDownstream[i], pExecInfoList);
if (code != TSDB_CODE_SUCCESS) {
// taosMemoryFreeClear(*pRes);
return TSDB_CODE_OUT_OF_MEMORY;
}
}
return TSDB_CODE_SUCCESS;
}
int32_t setOutputBuf(SStreamState* pState, STimeWindow* win, SResultRow** pResult, int64_t tableGroupId,
SqlFunctionCtx* pCtx, int32_t numOfOutput, int32_t* rowEntryInfoOffset, SAggSupporter* pAggSup) {
SWinKey key = {
@ -2007,25 +1300,3 @@ void qStreamCloseTsdbReader(void* task) {
}
}
}
static void extractTableList(SArray* pList, const SOperatorInfo* pOperator) {
if (pOperator->operatorType == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN) {
SStreamScanInfo* pScanInfo = pOperator->info;
STableScanInfo* pTableScanInfo = pScanInfo->pTableScanOp->info;
taosArrayPush(pList, &pTableScanInfo->base.pTableListInfo);
} else if (pOperator->operatorType == QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
STableScanInfo* pScanInfo = pOperator->info;
taosArrayPush(pList, &pScanInfo->base.pTableListInfo);
} else {
if (pOperator->pDownstream != NULL && pOperator->pDownstream[0] != NULL) {
extractTableList(pList, pOperator->pDownstream[0]);
}
}
}
SArray* getTableListInfo(const SExecTaskInfo* pTaskInfo) {
SArray* pArray = taosArrayInit(0, POINTER_BYTES);
SOperatorInfo* pOperator = pTaskInfo->pRoot;
extractTableList(pArray, pOperator);
return pArray;
}

6
source/libs/executor/src/filloperator.c
View File

@ -20,16 +20,18 @@
#include "tmsg.h"
#include "ttypes.h"
#include "executorimpl.h"
#include "executorInt.h"
#include "tcommon.h"
#include "thash.h"
#include "ttime.h"
#include "executorInt.h"
#include "function.h"
#include "querynodes.h"
#include "tdatablock.h"
#include "tfill.h"
#include "operator.h"
#include "querytask.h"
#define FILL_POS_INVALID 0
#define FILL_POS_START 1

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

@ -22,7 +22,8 @@
#include "tmsg.h"
#include "executorInt.h"
#include "executorimpl.h"
#include "operator.h"
#include "querytask.h"
#include "tcompare.h"
#include "thash.h"
#include "ttypes.h"
@ -926,7 +927,6 @@ int32_t setGroupResultOutputBuf(SOperatorInfo* pOperator, SOptrBasicInfo* binfo,
SResultRow* pResultRow =
doSetResultOutBufByKey(pBuf, pResultRowInfo, (char*)pData, bytes, true, groupId, pTaskInfo, false, pAggSup, false);
assert(pResultRow != NULL);
setResultRowInitCtx(pResultRow, pCtx, numOfCols, pOperator->exprSupp.rowEntryInfoOffset);
return TSDB_CODE_SUCCESS;

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

@ -13,11 +13,13 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "executorInt.h"
#include "filter.h"
#include "executorimpl.h"
#include "function.h"
#include "operator.h"
#include "os.h"
#include "querynodes.h"
#include "querytask.h"
#include "tcompare.h"
#include "tdatablock.h"
#include "thash.h"

578
source/libs/executor/src/operator.c Normal file
View File

@ -0,0 +1,578 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "filter.h"
#include "function.h"
#include "os.h"
#include "tname.h"
#include "tglobal.h"
#include "executorInt.h"
#include "index.h"
#include "operator.h"
#include "query.h"
#include "querytask.h"
#include "vnode.h"
SOperatorFpSet createOperatorFpSet(__optr_open_fn_t openFn, __optr_fn_t nextFn, __optr_fn_t cleanup,
__optr_close_fn_t closeFn, __optr_reqBuf_fn_t reqBufFn,
__optr_explain_fn_t explain) {
SOperatorFpSet fpSet = {
._openFn = openFn,
.getNextFn = nextFn,
.cleanupFn = cleanup,
.closeFn = closeFn,
.reqBufFn = reqBufFn,
.getExplainFn = explain,
};
return fpSet;
}
int32_t optrDummyOpenFn(SOperatorInfo* pOperator) {
OPTR_SET_OPENED(pOperator);
pOperator->cost.openCost = 0;
return TSDB_CODE_SUCCESS;
}
int32_t appendDownstream(SOperatorInfo* p, SOperatorInfo** pDownstream, int32_t num) {
p->pDownstream = taosMemoryCalloc(1, num * POINTER_BYTES);
if (p->pDownstream == NULL) {
return TSDB_CODE_OUT_OF_MEMORY;
}
memcpy(p->pDownstream, pDownstream, num * POINTER_BYTES);
p->numOfDownstream = num;
return TSDB_CODE_SUCCESS;
}
void setOperatorCompleted(SOperatorInfo* pOperator) {
pOperator->status = OP_EXEC_DONE;
pOperator->cost.totalCost = (taosGetTimestampUs() - pOperator->pTaskInfo->cost.start) / 1000.0;
setTaskStatus(pOperator->pTaskInfo, TASK_COMPLETED);
}
void setOperatorInfo(SOperatorInfo* pOperator, const char* name, int32_t type, bool blocking, int32_t status,
void* pInfo, SExecTaskInfo* pTaskInfo) {
pOperator->name = (char*)name;
pOperator->operatorType = type;
pOperator->blocking = blocking;
pOperator->status = status;
pOperator->info = pInfo;
pOperator->pTaskInfo = pTaskInfo;
}
// each operator should be set their own function to return total cost buffer
int32_t optrDefaultBufFn(SOperatorInfo* pOperator) {
if (pOperator->blocking) {
return -1;
} else {
return 0;
}
}
static int64_t getQuerySupportBufSize(size_t numOfTables) {
size_t s1 = sizeof(STableQueryInfo);
// size_t s3 = sizeof(STableCheckInfo); buffer consumption in tsdb
return (int64_t)(s1 * 1.5 * numOfTables);
}
int32_t checkForQueryBuf(size_t numOfTables) {
int64_t t = getQuerySupportBufSize(numOfTables);
if (tsQueryBufferSizeBytes < 0) {
return TSDB_CODE_SUCCESS;
} else if (tsQueryBufferSizeBytes > 0) {
while (1) {
int64_t s = tsQueryBufferSizeBytes;
int64_t remain = s - t;
if (remain >= 0) {
if (atomic_val_compare_exchange_64(&tsQueryBufferSizeBytes, s, remain) == s) {
return TSDB_CODE_SUCCESS;
}
} else {
return TSDB_CODE_QRY_NOT_ENOUGH_BUFFER;
}
}
}
// disable query processing if the value of tsQueryBufferSize is zero.
return TSDB_CODE_QRY_NOT_ENOUGH_BUFFER;
}
void releaseQueryBuf(size_t numOfTables) {
if (tsQueryBufferSizeBytes < 0) {
return;
}
int64_t t = getQuerySupportBufSize(numOfTables);
// restore value is not enough buffer available
atomic_add_fetch_64(&tsQueryBufferSizeBytes, t);
}
typedef enum {
OPTR_FN_RET_CONTINUE = 0x1,
OPTR_FN_RET_ABORT = 0x2,
} ERetType;
typedef struct STraverParam {
void* pRet;
int32_t code;
void* pParam;
} STraverParam;
// iterate the operator tree helper
typedef ERetType (*optr_fn_t)(SOperatorInfo *pOperator, STraverParam *pParam, const char* pIdstr);
void traverseOperatorTree(SOperatorInfo* pOperator, optr_fn_t fn, STraverParam* pParam, const char* id) {
if (pOperator == NULL) {
return;
}
ERetType ret = fn(pOperator, pParam, id);
if (ret == OPTR_FN_RET_ABORT || pParam->code != TSDB_CODE_SUCCESS) {
return;
}
for (int32_t i = 0; i < pOperator->numOfDownstream; ++i) {
traverseOperatorTree(pOperator->pDownstream[i], fn, pParam, id);
if (pParam->code != 0) {
break;
}
}
}
ERetType extractOperatorInfo(SOperatorInfo* pOperator, STraverParam* pParam, const char* pIdStr) {
STraverParam* p = pParam;
if (pOperator->operatorType == *(int32_t*)p->pParam) {
p->pRet = pOperator;
return OPTR_FN_RET_ABORT;
} else {
return OPTR_FN_RET_CONTINUE;
}
}
// QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN
SOperatorInfo* extractOperatorInTree(SOperatorInfo* pOperator, int32_t type, const char* id) {
if (pOperator == NULL) {
qError("invalid operator, failed to find tableScanOperator %s", id);
terrno = TSDB_CODE_PAR_INTERNAL_ERROR;
return NULL;
}
STraverParam p = {.pParam = &type, .pRet = NULL};
traverseOperatorTree(pOperator, extractOperatorInfo, &p, id);
if (p.code != 0) {
terrno = p.code;
return NULL;
} else {
return p.pRet;
}
}
typedef struct SExtScanInfo {
int32_t order;
int32_t scanFlag;
int32_t inheritUsOrder;
} SExtScanInfo;
static ERetType extractScanInfo(SOperatorInfo* pOperator, STraverParam* pParam, const char* pIdStr) {
int32_t type = pOperator->operatorType;
SExtScanInfo* pInfo = pParam->pParam;
if (type == QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN || type == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN ||
type == QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN || type == QUERY_NODE_PHYSICAL_PLAN_BLOCK_DIST_SCAN ||
type == QUERY_NODE_PHYSICAL_PLAN_LAST_ROW_SCAN || type == QUERY_NODE_PHYSICAL_PLAN_TABLE_COUNT_SCAN) {
pInfo->order = TSDB_ORDER_ASC;
pInfo->scanFlag= MAIN_SCAN;
return OPTR_FN_RET_ABORT;
} else if (type == QUERY_NODE_PHYSICAL_PLAN_EXCHANGE) {
if (!pInfo->inheritUsOrder) {
pInfo->order = TSDB_ORDER_ASC;
}
pInfo->scanFlag= MAIN_SCAN;
return OPTR_FN_RET_ABORT;
} else if (type == QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
STableScanInfo* pTableScanInfo = pOperator->info;
pInfo->order = pTableScanInfo->base.cond.order;
pInfo->scanFlag= pTableScanInfo->base.scanFlag;
return OPTR_FN_RET_ABORT;
} else if (type == QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN) {
STableMergeScanInfo* pTableScanInfo = pOperator->info;
pInfo->order = pTableScanInfo->base.cond.order;
pInfo->scanFlag= pTableScanInfo->base.scanFlag;
return OPTR_FN_RET_ABORT;
} else {
return OPTR_FN_RET_CONTINUE;
}
}
int32_t getTableScanInfo(SOperatorInfo* pOperator, int32_t* order, int32_t* scanFlag, bool inheritUsOrder) {
SExtScanInfo info = {.inheritUsOrder = inheritUsOrder, .order = *order};
STraverParam p = {.pParam = &info};
traverseOperatorTree(pOperator, extractScanInfo, &p, NULL);
*order = info.order;
*scanFlag = info.scanFlag;
ASSERT(*order == TSDB_ORDER_ASC || *order == TSDB_ORDER_DESC);
return p.code;
}
static ERetType doStopDataReader(SOperatorInfo* pOperator, STraverParam* pParam, const char* pIdStr) {
if (pOperator->operatorType == QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN) {
STableScanInfo* pInfo = pOperator->info;
if (pInfo->base.dataReader != NULL) {
tsdbReaderSetCloseFlag(pInfo->base.dataReader);
}
return OPTR_FN_RET_ABORT;
} else if (pOperator->operatorType == QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN) {
SStreamScanInfo* pInfo = pOperator->info;
if (pInfo->pTableScanOp != NULL) {
STableScanInfo* pTableScanInfo = pInfo->pTableScanOp->info;
if (pTableScanInfo != NULL && pTableScanInfo->base.dataReader != NULL) {
tsdbReaderSetCloseFlag(pTableScanInfo->base.dataReader);
}
}
return OPTR_FN_RET_ABORT;
}
return OPTR_FN_RET_CONTINUE;
}
int32_t stopTableScanOperator(SOperatorInfo* pOperator, const char* pIdStr) {
STraverParam p = {0};
traverseOperatorTree(pOperator, doStopDataReader, &p, pIdStr);
return p.code;
}
SOperatorInfo* createOperator(SPhysiNode* pPhyNode, SExecTaskInfo* pTaskInfo, SReadHandle* pHandle, SNode* pTagCond,
SNode* pTagIndexCond, const char* pUser, const char* dbname) {
int32_t type = nodeType(pPhyNode);
const char* idstr = GET_TASKID(pTaskInfo);
if (pPhyNode->pChildren == NULL || LIST_LENGTH(pPhyNode->pChildren) == 0) {
SOperatorInfo* pOperator = NULL;
if (QUERY_NODE_PHYSICAL_PLAN_TABLE_SCAN == type) {
STableScanPhysiNode* pTableScanNode = (STableScanPhysiNode*)pPhyNode;
// NOTE: this is an patch to fix the physical plan
// TODO remove it later
if (pTableScanNode->scan.node.pLimit != NULL) {
pTableScanNode->groupSort = true;
}
STableListInfo* pTableListInfo = tableListCreate();
int32_t code =
createScanTableListInfo(&pTableScanNode->scan, pTableScanNode->pGroupTags, pTableScanNode->groupSort, pHandle,
pTableListInfo, pTagCond, pTagIndexCond, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
qError("failed to createScanTableListInfo, code:%s, %s", tstrerror(code), idstr);
return NULL;
}
code = initQueriedTableSchemaInfo(pHandle, &pTableScanNode->scan, dbname, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
return NULL;
}
pOperator = createTableScanOperatorInfo(pTableScanNode, pHandle, pTableListInfo, pTaskInfo);
if (NULL == pOperator) {
pTaskInfo->code = terrno;
tableListDestroy(pTableListInfo);
return NULL;
}
STableScanInfo* pScanInfo = pOperator->info;
pTaskInfo->cost.pRecoder = &pScanInfo->base.readRecorder;
} else if (QUERY_NODE_PHYSICAL_PLAN_TABLE_MERGE_SCAN == type) {
STableMergeScanPhysiNode* pTableScanNode = (STableMergeScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
int32_t code = createScanTableListInfo(&pTableScanNode->scan, pTableScanNode->pGroupTags, true, pHandle,
pTableListInfo, pTagCond, pTagIndexCond, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
qError("failed to createScanTableListInfo, code: %s", tstrerror(code));
return NULL;
}
code = initQueriedTableSchemaInfo(pHandle, &pTableScanNode->scan, dbname, pTaskInfo);
if (code) {
pTaskInfo->code = terrno;
tableListDestroy(pTableListInfo);
return NULL;
}
pOperator = createTableMergeScanOperatorInfo(pTableScanNode, pHandle, pTableListInfo, pTaskInfo);
if (NULL == pOperator) {
pTaskInfo->code = terrno;
tableListDestroy(pTableListInfo);
return NULL;
}
STableScanInfo* pScanInfo = pOperator->info;
pTaskInfo->cost.pRecoder = &pScanInfo->base.readRecorder;
} else if (QUERY_NODE_PHYSICAL_PLAN_EXCHANGE == type) {
pOperator = createExchangeOperatorInfo(pHandle ? pHandle->pMsgCb->clientRpc : NULL, (SExchangePhysiNode*)pPhyNode,
pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SCAN == type) {
STableScanPhysiNode* pTableScanNode = (STableScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
if (pHandle->vnode) {
int32_t code =
createScanTableListInfo(&pTableScanNode->scan, pTableScanNode->pGroupTags, pTableScanNode->groupSort,
pHandle, pTableListInfo, pTagCond, pTagIndexCond, pTaskInfo);
if (code) {
pTaskInfo->code = code;
tableListDestroy(pTableListInfo);
qError("failed to createScanTableListInfo, code: %s", tstrerror(code));
return NULL;
}
}
pTaskInfo->schemaInfo.qsw = extractQueriedColumnSchema(&pTableScanNode->scan);
pOperator = createStreamScanOperatorInfo(pHandle, pTableScanNode, pTagCond, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_SYSTABLE_SCAN == type) {
SSystemTableScanPhysiNode* pSysScanPhyNode = (SSystemTableScanPhysiNode*)pPhyNode;
pOperator = createSysTableScanOperatorInfo(pHandle, pSysScanPhyNode, pUser, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_TABLE_COUNT_SCAN == type) {
STableCountScanPhysiNode* pTblCountScanNode = (STableCountScanPhysiNode*)pPhyNode;
pOperator = createTableCountScanOperatorInfo(pHandle, pTblCountScanNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_TAG_SCAN == type) {
STagScanPhysiNode* pScanPhyNode = (STagScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
int32_t code = createScanTableListInfo(pScanPhyNode, NULL, false, pHandle, pTableListInfo, pTagCond,
pTagIndexCond, pTaskInfo);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = code;
qError("failed to getTableList, code: %s", tstrerror(code));
return NULL;
}
pOperator = createTagScanOperatorInfo(pHandle, pScanPhyNode, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_BLOCK_DIST_SCAN == type) {
SBlockDistScanPhysiNode* pBlockNode = (SBlockDistScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
if (pBlockNode->tableType == TSDB_SUPER_TABLE) {
SArray* pList = taosArrayInit(4, sizeof(STableKeyInfo));
int32_t code = vnodeGetAllTableList(pHandle->vnode, pBlockNode->uid, pList);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = terrno;
return NULL;
}
size_t num = taosArrayGetSize(pList);
for (int32_t i = 0; i < num; ++i) {
STableKeyInfo* p = taosArrayGet(pList, i);
tableListAddTableInfo(pTableListInfo, p->uid, 0);
}
taosArrayDestroy(pList);
} else { // Create group with only one table
tableListAddTableInfo(pTableListInfo, pBlockNode->uid, 0);
}
pOperator = createDataBlockInfoScanOperator(pHandle, pBlockNode, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_LAST_ROW_SCAN == type) {
SLastRowScanPhysiNode* pScanNode = (SLastRowScanPhysiNode*)pPhyNode;
STableListInfo* pTableListInfo = tableListCreate();
int32_t code = createScanTableListInfo(&pScanNode->scan, pScanNode->pGroupTags, true, pHandle, pTableListInfo,
pTagCond, pTagIndexCond, pTaskInfo);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = code;
return NULL;
}
code = initQueriedTableSchemaInfo(pHandle, &pScanNode->scan, dbname, pTaskInfo);
if (code != TSDB_CODE_SUCCESS) {
pTaskInfo->code = code;
return NULL;
}
pOperator = createCacherowsScanOperator(pScanNode, pHandle, pTableListInfo, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_PROJECT == type) {
pOperator = createProjectOperatorInfo(NULL, (SProjectPhysiNode*)pPhyNode, pTaskInfo);
} else {
terrno = TSDB_CODE_INVALID_PARA;
return NULL;
}
if (pOperator != NULL) { // todo moved away
pOperator->resultDataBlockId = pPhyNode->pOutputDataBlockDesc->dataBlockId;
}
return pOperator;
}
size_t size = LIST_LENGTH(pPhyNode->pChildren);
SOperatorInfo** ops = taosMemoryCalloc(size, POINTER_BYTES);
if (ops == NULL) {
return NULL;
}
for (int32_t i = 0; i < size; ++i) {
SPhysiNode* pChildNode = (SPhysiNode*)nodesListGetNode(pPhyNode->pChildren, i);
ops[i] = createOperator(pChildNode, pTaskInfo, pHandle, pTagCond, pTagIndexCond, pUser, dbname);
if (ops[i] == NULL) {
taosMemoryFree(ops);
return NULL;
}
}
SOperatorInfo* pOptr = NULL;
if (QUERY_NODE_PHYSICAL_PLAN_PROJECT == type) {
pOptr = createProjectOperatorInfo(ops[0], (SProjectPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_HASH_AGG == type) {
SAggPhysiNode* pAggNode = (SAggPhysiNode*)pPhyNode;
if (pAggNode->pGroupKeys != NULL) {
pOptr = createGroupOperatorInfo(ops[0], pAggNode, pTaskInfo);
} else {
pOptr = createAggregateOperatorInfo(ops[0], pAggNode, pTaskInfo);
}
} else if (QUERY_NODE_PHYSICAL_PLAN_HASH_INTERVAL == type) {
SIntervalPhysiNode* pIntervalPhyNode = (SIntervalPhysiNode*)pPhyNode;
pOptr = createIntervalOperatorInfo(ops[0], pIntervalPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_INTERVAL == type) {
pOptr = createStreamIntervalOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_ALIGNED_INTERVAL == type) {
SMergeAlignedIntervalPhysiNode* pIntervalPhyNode = (SMergeAlignedIntervalPhysiNode*)pPhyNode;
pOptr = createMergeAlignedIntervalOperatorInfo(ops[0], pIntervalPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_INTERVAL == type) {
SMergeIntervalPhysiNode* pIntervalPhyNode = (SMergeIntervalPhysiNode*)pPhyNode;
pOptr = createMergeIntervalOperatorInfo(ops[0], pIntervalPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_INTERVAL == type) {
int32_t children = 0;
pOptr = createStreamFinalIntervalOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_INTERVAL == type) {
int32_t children = pHandle->numOfVgroups;
pOptr = createStreamFinalIntervalOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_SORT == type) {
pOptr = createSortOperatorInfo(ops[0], (SSortPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_GROUP_SORT == type) {
pOptr = createGroupSortOperatorInfo(ops[0], (SGroupSortPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE == type) {
SMergePhysiNode* pMergePhyNode = (SMergePhysiNode*)pPhyNode;
pOptr = createMultiwayMergeOperatorInfo(ops, size, pMergePhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_SESSION == type) {
SSessionWinodwPhysiNode* pSessionNode = (SSessionWinodwPhysiNode*)pPhyNode;
pOptr = createSessionAggOperatorInfo(ops[0], pSessionNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SESSION == type) {
pOptr = createStreamSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_SEMI_SESSION == type) {
int32_t children = 0;
pOptr = createStreamFinalSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FINAL_SESSION == type) {
int32_t children = pHandle->numOfVgroups;
pOptr = createStreamFinalSessionAggOperatorInfo(ops[0], pPhyNode, pTaskInfo, children);
} else if (QUERY_NODE_PHYSICAL_PLAN_PARTITION == type) {
pOptr = createPartitionOperatorInfo(ops[0], (SPartitionPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_PARTITION == type) {
pOptr = createStreamPartitionOperatorInfo(ops[0], (SStreamPartitionPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_STATE == type) {
SStateWinodwPhysiNode* pStateNode = (SStateWinodwPhysiNode*)pPhyNode;
pOptr = createStatewindowOperatorInfo(ops[0], pStateNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_STATE == type) {
pOptr = createStreamStateAggOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_JOIN == type) {
pOptr = createMergeJoinOperatorInfo(ops, size, (SSortMergeJoinPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_FILL == type) {
pOptr = createFillOperatorInfo(ops[0], (SFillPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_STREAM_FILL == type) {
pOptr = createStreamFillOperatorInfo(ops[0], (SStreamFillPhysiNode*)pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_INDEF_ROWS_FUNC == type) {
pOptr = createIndefinitOutputOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_INTERP_FUNC == type) {
pOptr = createTimeSliceOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else if (QUERY_NODE_PHYSICAL_PLAN_MERGE_EVENT == type) {
pOptr = createEventwindowOperatorInfo(ops[0], pPhyNode, pTaskInfo);
} else {
terrno = TSDB_CODE_INVALID_PARA;
taosMemoryFree(ops);
return NULL;
}
taosMemoryFree(ops);
if (pOptr) {
pOptr->resultDataBlockId = pPhyNode->pOutputDataBlockDesc->dataBlockId;
}
return pOptr;
}
void destroyOperator(SOperatorInfo* pOperator) {
if (pOperator == NULL) {
return;
}
if (pOperator->fpSet.closeFn != NULL) {
pOperator->fpSet.closeFn(pOperator->info);
}
if (pOperator->pDownstream != NULL) {
for (int32_t i = 0; i < pOperator->numOfDownstream; ++i) {
destroyOperator(pOperator->pDownstream[i]);
}
taosMemoryFreeClear(pOperator->pDownstream);
pOperator->numOfDownstream = 0;
}
cleanupExprSupp(&pOperator->exprSupp);
taosMemoryFreeClear(pOperator);
}
int32_t getOperatorExplainExecInfo(SOperatorInfo* operatorInfo, SArray* pExecInfoList) {
SExplainExecInfo execInfo = {0};
SExplainExecInfo* pExplainInfo = taosArrayPush(pExecInfoList, &execInfo);
pExplainInfo->numOfRows = operatorInfo->resultInfo.totalRows;
pExplainInfo->startupCost = operatorInfo->cost.openCost;
pExplainInfo->totalCost = operatorInfo->cost.totalCost;
pExplainInfo->verboseLen = 0;
pExplainInfo->verboseInfo = NULL;
if (operatorInfo->fpSet.getExplainFn) {
int32_t code =
operatorInfo->fpSet.getExplainFn(operatorInfo, &pExplainInfo->verboseInfo, &pExplainInfo->verboseLen);
if (code) {
qError("%s operator getExplainFn failed, code:%s", GET_TASKID(operatorInfo->pTaskInfo), tstrerror(code));
return code;
}
}
int32_t code = 0;
for (int32_t i = 0; i < operatorInfo->numOfDownstream; ++i) {
code = getOperatorExplainExecInfo(operatorInfo->pDownstream[i], pExecInfoList);
if (code != TSDB_CODE_SUCCESS) {
// taosMemoryFreeClear(*pRes);
return TSDB_CODE_OUT_OF_MEMORY;
}
}
return TSDB_CODE_SUCCESS;
}

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

@ -13,9 +13,11 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "executorimpl.h"
#include "executorInt.h"
#include "filter.h"
#include "functionMgt.h"
#include "operator.h"
#include "querytask.h"
typedef struct SProjectOperatorInfo {
SOptrBasicInfo binfo;

239
source/libs/executor/src/querytask.c Normal file
View File

@ -0,0 +1,239 @@
/*
* Copyright (c) 2019 TAOS Data, Inc. <jhtao@taosdata.com>
*
* This program is free software: you can use, redistribute, and/or modify
* it under the terms of the GNU Affero General Public License, version 3
* or later ("AGPL"), as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE.
*
* You should have received a copy of the GNU Affero General Public License
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "filter.h"
#include "function.h"
#include "functionMgt.h"
#include "os.h"
#include "querynodes.h"
#include "tfill.h"
#include "tname.h"
#include "tdatablock.h"
#include "tmsg.h"
#include "executorInt.h"
#include "index.h"
#include "operator.h"
#include "query.h"
#include "querytask.h"
#include "thash.h"
#include "ttypes.h"
#include "vnode.h"
#define CLEAR_QUERY_STATUS(q, st) ((q)->status &= (~(st)))
SExecTaskInfo* doCreateTask(uint64_t queryId, uint64_t taskId, int32_t vgId, EOPTR_EXEC_MODEL model) {
SExecTaskInfo* pTaskInfo = taosMemoryCalloc(1, sizeof(SExecTaskInfo));
if (pTaskInfo == NULL) {
terrno = TSDB_CODE_OUT_OF_MEMORY;
return NULL;
}
setTaskStatus(pTaskInfo, TASK_NOT_COMPLETED);
pTaskInfo->cost.created = taosGetTimestampUs();
pTaskInfo->execModel = model;
pTaskInfo->stopInfo.pStopInfo = taosArrayInit(4, sizeof(SExchangeOpStopInfo));
pTaskInfo->pResultBlockList = taosArrayInit(128, POINTER_BYTES);
taosInitRWLatch(&pTaskInfo->lock);
pTaskInfo->id.vgId = vgId;
pTaskInfo->id.queryId = queryId;
pTaskInfo->id.str = taosMemoryMalloc(64);
buildTaskId(taskId, queryId, pTaskInfo->id.str);
return pTaskInfo;
}
bool isTaskKilled(SExecTaskInfo* pTaskInfo) { return (0 != pTaskInfo->code); }
void setTaskKilled(SExecTaskInfo* pTaskInfo, int32_t rspCode) {
pTaskInfo->code = rspCode;
stopTableScanOperator(pTaskInfo->pRoot, pTaskInfo->id.str);
}
void setTaskStatus(SExecTaskInfo* pTaskInfo, int8_t status) {
if (status == TASK_NOT_COMPLETED) {
pTaskInfo->status = status;
} else {
// QUERY_NOT_COMPLETED is not compatible with any other status, so clear its position first
CLEAR_QUERY_STATUS(pTaskInfo, TASK_NOT_COMPLETED);
pTaskInfo->status |= status;
}
}
int32_t createExecTaskInfo(SSubplan* pPlan, SExecTaskInfo** pTaskInfo, SReadHandle* pHandle, uint64_t taskId,
int32_t vgId, char* sql, EOPTR_EXEC_MODEL model) {
*pTaskInfo = doCreateTask(pPlan->id.queryId, taskId, vgId, model);
if (*pTaskInfo == NULL) {
taosMemoryFree(sql);
return terrno;
}
if (pHandle) {
if (pHandle->pStateBackend) {
(*pTaskInfo)->streamInfo.pState = pHandle->pStateBackend;
}
}
TSWAP((*pTaskInfo)->sql, sql);
(*pTaskInfo)->pSubplan = pPlan;
(*pTaskInfo)->pRoot = createOperator(pPlan->pNode, *pTaskInfo, pHandle, pPlan->pTagCond, pPlan->pTagIndexCond,
pPlan->user, pPlan->dbFName);
if (NULL == (*pTaskInfo)->pRoot) {
int32_t code = (*pTaskInfo)->code;
doDestroyTask(*pTaskInfo);
(*pTaskInfo) = NULL;
return code;
} else {
return TSDB_CODE_SUCCESS;
}
}
void cleanupQueriedTableScanInfo(SSchemaInfo* pSchemaInfo) {
taosMemoryFreeClear(pSchemaInfo->dbname);
taosMemoryFreeClear(pSchemaInfo->tablename);
tDeleteSchemaWrapper(pSchemaInfo->sw);
tDeleteSchemaWrapper(pSchemaInfo->qsw);
}
int32_t initQueriedTableSchemaInfo(SReadHandle* pHandle, SScanPhysiNode* pScanNode, const char* dbName, SExecTaskInfo* pTaskInfo) {
SMetaReader mr = {0};
if (pHandle == NULL) {
terrno = TSDB_CODE_INVALID_PARA;
return terrno;
}
metaReaderInit(&mr, pHandle->meta, 0);
int32_t code = metaGetTableEntryByUidCache(&mr, pScanNode->uid);
if (code != TSDB_CODE_SUCCESS) {
qError("failed to get the table meta, uid:0x%" PRIx64 ", suid:0x%" PRIx64 ", %s", pScanNode->uid, pScanNode->suid,
GET_TASKID(pTaskInfo));
metaReaderClear(&mr);
return terrno;
}
SSchemaInfo* pSchemaInfo = &pTaskInfo->schemaInfo;
pSchemaInfo->tablename = taosStrdup(mr.me.name);
pSchemaInfo->dbname = taosStrdup(dbName);
if (mr.me.type == TSDB_SUPER_TABLE) {
pSchemaInfo->sw = tCloneSSchemaWrapper(&mr.me.stbEntry.schemaRow);
pSchemaInfo->tversion = mr.me.stbEntry.schemaTag.version;
} else if (mr.me.type == TSDB_CHILD_TABLE) {
tDecoderClear(&mr.coder);
tb_uid_t suid = mr.me.ctbEntry.suid;
code = metaGetTableEntryByUidCache(&mr, suid);
if (code != TSDB_CODE_SUCCESS) {
metaReaderClear(&mr);
return terrno;
}
pSchemaInfo->sw = tCloneSSchemaWrapper(&mr.me.stbEntry.schemaRow);
pSchemaInfo->tversion = mr.me.stbEntry.schemaTag.version;
} else {
pSchemaInfo->sw = tCloneSSchemaWrapper(&mr.me.ntbEntry.schemaRow);
}
metaReaderClear(&mr);
pSchemaInfo->qsw = extractQueriedColumnSchema(pScanNode);
return TSDB_CODE_SUCCESS;
}
SSchemaWrapper* extractQueriedColumnSchema(SScanPhysiNode* pScanNode) {
int32_t numOfCols = LIST_LENGTH(pScanNode->pScanCols);
int32_t numOfTags = LIST_LENGTH(pScanNode->pScanPseudoCols);
SSchemaWrapper* pqSw = taosMemoryCalloc(1, sizeof(SSchemaWrapper));
pqSw->pSchema = taosMemoryCalloc(numOfCols + numOfTags, sizeof(SSchema));
for (int32_t i = 0; i < numOfCols; ++i) {
STargetNode* pNode = (STargetNode*)nodesListGetNode(pScanNode->pScanCols, i);
SColumnNode* pColNode = (SColumnNode*)pNode->pExpr;
SSchema* pSchema = &pqSw->pSchema[pqSw->nCols++];
pSchema->colId = pColNode->colId;
pSchema->type = pColNode->node.resType.type;
pSchema->bytes = pColNode->node.resType.bytes;
tstrncpy(pSchema->name, pColNode->colName, tListLen(pSchema->name));
}
// this the tags and pseudo function columns, we only keep the tag columns
for (int32_t i = 0; i < numOfTags; ++i) {
STargetNode* pNode = (STargetNode*)nodesListGetNode(pScanNode->pScanPseudoCols, i);
int32_t type = nodeType(pNode->pExpr);
if (type == QUERY_NODE_COLUMN) {
SColumnNode* pColNode = (SColumnNode*)pNode->pExpr;
SSchema* pSchema = &pqSw->pSchema[pqSw->nCols++];
pSchema->colId = pColNode->colId;
pSchema->type = pColNode->node.resType.type;
pSchema->bytes = pColNode->node.resType.bytes;
tstrncpy(pSchema->name, pColNode->colName, tListLen(pSchema->name));
}
}
return pqSw;
}
static void cleanupStreamInfo(SStreamTaskInfo* pStreamInfo) { tDeleteSchemaWrapper(pStreamInfo->schema); }
static void freeBlock(void* pParam) {
SSDataBlock* pBlock = *(SSDataBlock**)pParam;
blockDataDestroy(pBlock);
}
void doDestroyTask(SExecTaskInfo* pTaskInfo) {
qDebug("%s execTask is freed", GET_TASKID(pTaskInfo));
destroyOperator(pTaskInfo->pRoot);
pTaskInfo->pRoot = NULL;
cleanupQueriedTableScanInfo(&pTaskInfo->schemaInfo);
cleanupStreamInfo(&pTaskInfo->streamInfo);
if (!pTaskInfo->localFetch.localExec) {
nodesDestroyNode((SNode*)pTaskInfo->pSubplan);
pTaskInfo->pSubplan = NULL;
}
taosArrayDestroyEx(pTaskInfo->pResultBlockList, freeBlock);
taosArrayDestroy(pTaskInfo->stopInfo.pStopInfo);
taosMemoryFreeClear(pTaskInfo->sql);
taosMemoryFreeClear(pTaskInfo->id.str);
taosMemoryFreeClear(pTaskInfo);
}
void buildTaskId(uint64_t taskId, uint64_t queryId, char* dst) {
char* p = dst;
int32_t offset = 6;
memcpy(p, "TID:0x", offset);
offset += tintToHex(taskId, &p[offset]);
memcpy(&p[offset], " QID:0x", 7);
offset += 7;
offset += tintToHex(queryId, &p[offset]);
p[offset] = 0;
}

65
source/libs/executor/src/scanoperator.c
View File

@ -13,7 +13,7 @@
* along with this program. If not, see <http://www.gnu.org/licenses/>.
*/
#include "executorimpl.h"
#include "executorInt.h"
#include "filter.h"
#include "function.h"
#include "functionMgt.h"
@ -30,6 +30,8 @@
#include "tcompare.h"
#include "thash.h"
#include "ttypes.h"
#include "operator.h"
#include "querytask.h"
int32_t scanDebug = 0;
@ -156,7 +158,7 @@ static bool overlapWithTimeWindow(SInterval* pInterval, SDataBlockInfo* pBlockIn
break;
}
assert(w.skey < pBlockInfo->window.skey);
ASSERT(w.skey < pBlockInfo->window.skey);
if (pBlockInfo->window.skey <= TMIN(w.ekey, pBlockInfo->window.ekey)) {
return true;
}
@ -1621,7 +1623,7 @@ static int32_t setBlockIntoRes(SStreamScanInfo* pInfo, const SSDataBlock* pBlock
pInfo->pRes->info.dataLoad = 1;
blockDataUpdateTsWindow(pInfo->pRes, pInfo->primaryTsIndex);
blockDataFreeRes((SSDataBlock*)pBlock);
// blockDataFreeRes((SSDataBlock*)pBlock);
calBlockTbName(pInfo, pInfo->pRes);
return 0;
@ -1635,7 +1637,7 @@ static SSDataBlock* doQueueScan(SOperatorInfo* pOperator) {
qDebug("start to exec queue scan, %s", id);
if (pTaskInfo->streamInfo.submit.msgStr != NULL) {
if (pInfo->tqReader->msg2.msgStr == NULL) {
if (pInfo->tqReader->msg.msgStr == NULL) {
SPackedData submit = pTaskInfo->streamInfo.submit;
if (tqReaderSetSubmitMsg(pInfo->tqReader, submit.msgStr, submit.msgLen, submit.ver) < 0) {
qError("submit msg messed up when initing stream submit block %p", submit.msgStr);
@ -1646,22 +1648,20 @@ static SSDataBlock* doQueueScan(SOperatorInfo* pOperator) {
blockDataCleanup(pInfo->pRes);
SDataBlockInfo* pBlockInfo = &pInfo->pRes->info;
while (tqNextDataBlock(pInfo->tqReader)) {
SSDataBlock block = {0};
int32_t code = tqRetrieveDataBlock2(&block, pInfo->tqReader, NULL);
if (code != TSDB_CODE_SUCCESS || block.info.rows == 0) {
while (tqNextBlockImpl(pInfo->tqReader)) {
int32_t code = tqRetrieveDataBlock(pInfo->tqReader->pResBlock, pInfo->tqReader, NULL);
if (code != TSDB_CODE_SUCCESS || pInfo->tqReader->pResBlock->info.rows == 0) {
continue;
}
setBlockIntoRes(pInfo, &block, true);
setBlockIntoRes(pInfo, pInfo->tqReader->pResBlock, true);
if (pBlockInfo->rows > 0) {
return pInfo->pRes;
}
}
pInfo->tqReader->msg2 = (SPackedData){0};
pInfo->tqReader->msg = (SPackedData){0};
pTaskInfo->streamInfo.submit = (SPackedData){0};
return NULL;
}
@ -1687,23 +1687,21 @@ static SSDataBlock* doQueueScan(SOperatorInfo* pOperator) {
if (pTaskInfo->streamInfo.currentOffset.type == TMQ_OFFSET__LOG) {
while (1) {
SFetchRet ret = {0};
tqNextBlock(pInfo->tqReader, &ret);
tqOffsetResetToLog(
&pTaskInfo->streamInfo.currentOffset,
pInfo->tqReader->pWalReader->curVersion - 1); // curVersion move to next, so currentOffset = curVersion - 1
int32_t type = tqNextBlockInWal(pInfo->tqReader);
SSDataBlock* pRes = pInfo->tqReader->pResBlock;
if (ret.fetchType == FETCH_TYPE__DATA) {
qDebug("doQueueScan get data from log %" PRId64 " rows, version:%" PRId64, ret.data.info.rows,
// curVersion move to next, so currentOffset = curVersion - 1
tqOffsetResetToLog(&pTaskInfo->streamInfo.currentOffset, pInfo->tqReader->pWalReader->curVersion - 1);
if (type == FETCH_TYPE__DATA) {
qDebug("doQueueScan get data from log %" PRId64 " rows, version:%" PRId64, pRes->info.rows,
pTaskInfo->streamInfo.currentOffset.version);
blockDataCleanup(pInfo->pRes);
setBlockIntoRes(pInfo, &ret.data, true);
setBlockIntoRes(pInfo, pRes, true);
if (pInfo->pRes->info.rows > 0) {
qDebug("doQueueScan get data from log %" PRId64 " rows, return, version:%" PRId64, pInfo->pRes->info.rows,
pTaskInfo->streamInfo.currentOffset.version);
return pInfo->pRes;
}
} else if (ret.fetchType == FETCH_TYPE__NONE) {
} else if (type == FETCH_TYPE__NONE) {
qDebug("doQueueScan get none from log, return, version:%" PRId64, pTaskInfo->streamInfo.currentOffset.version);
return NULL;
}
@ -1855,7 +1853,7 @@ static SSDataBlock* doStreamScan(SOperatorInfo* pOperator) {
SSDataBlock* pSDB = doRangeScan(pInfo, pInfo->pUpdateRes, pInfo->primaryTsIndex, &pInfo->updateResIndex);
if (pSDB) {
STableScanInfo* pTableScanInfo = pInfo->pTableScanOp->info;
uint64_t version = getReaderMaxVersion(pTableScanInfo->base.dataReader);
uint64_t version = tsdbGetReaderMaxVersion(pTableScanInfo->base.dataReader);
updateInfoSetScanRange(pInfo->pUpdateInfo, &pTableScanInfo->base.cond.twindows, pInfo->groupId, version);
pSDB->info.type = pInfo->scanMode == STREAM_SCAN_FROM_DATAREADER_RANGE ? STREAM_NORMAL : STREAM_PULL_DATA;
checkUpdateData(pInfo, true, pSDB, false);
@ -2021,7 +2019,7 @@ FETCH_NEXT_BLOCK:
SSDataBlock* pSDB = doRangeScan(pInfo, pInfo->pUpdateRes, pInfo->primaryTsIndex, &pInfo->updateResIndex);
if (pSDB) {
STableScanInfo* pTableScanInfo = pInfo->pTableScanOp->info;
uint64_t version = getReaderMaxVersion(pTableScanInfo->base.dataReader);
uint64_t version = tsdbGetReaderMaxVersion(pTableScanInfo->base.dataReader);
updateInfoSetScanRange(pInfo->pUpdateInfo, &pTableScanInfo->base.cond.twindows, pInfo->groupId, version);
pSDB->info.type = pInfo->scanMode == STREAM_SCAN_FROM_DATAREADER_RANGE ? STREAM_NORMAL : STREAM_PULL_DATA;
checkUpdateData(pInfo, true, pSDB, false);
@ -2053,7 +2051,7 @@ FETCH_NEXT_BLOCK:
NEXT_SUBMIT_BLK:
while (1) {
if (pInfo->tqReader->msg2.msgStr == NULL) {
if (pInfo->tqReader->msg.msgStr == NULL) {
if (pInfo->validBlockIndex >= totBlockNum) {
updateInfoDestoryColseWinSBF(pInfo->pUpdateInfo);
doClearBufferedBlocks(pInfo);
@ -2072,16 +2070,13 @@ FETCH_NEXT_BLOCK:
blockDataCleanup(pInfo->pRes);
while (tqNextDataBlock(pInfo->tqReader)) {
SSDataBlock block = {0};
int32_t code = tqRetrieveDataBlock2(&block, pInfo->tqReader, NULL);
if (code != TSDB_CODE_SUCCESS || block.info.rows == 0) {
while (tqNextBlockImpl(pInfo->tqReader)) {
int32_t code = tqRetrieveDataBlock(pInfo->tqReader->pResBlock, pInfo->tqReader, NULL);
if (code != TSDB_CODE_SUCCESS || pInfo->tqReader->pResBlock->info.rows == 0) {
continue;
}
setBlockIntoRes(pInfo, &block, false);
setBlockIntoRes(pInfo, pInfo->tqReader->pResBlock, false);
if (updateInfoIgnore(pInfo->pUpdateInfo, &pInfo->pRes->info.window, pInfo->pRes->info.id.groupId,
pInfo->pRes->info.version)) {
@ -2190,7 +2185,7 @@ static SSDataBlock* doRawScan(SOperatorInfo* pOperator) {
qDebug("tmqsnap change get data uid:%" PRId64 "", mtInfo.uid);
}
qStreamPrepareScan(pTaskInfo, &offset, pInfo->sContext->subType);
tDeleteSSchemaWrapper(mtInfo.schema);
tDeleteSchemaWrapper(mtInfo.schema);
return NULL;
} else if (pTaskInfo->streamInfo.currentOffset.type == TMQ_OFFSET__SNAPSHOT_META) {
SSnapContext* sContext = pInfo->sContext;
@ -2305,7 +2300,7 @@ static void destroyStreamScanOperatorInfo(void* param) {
SStreamScanInfo* pStreamScan = (SStreamScanInfo*)param;
if (pStreamScan->pTableScanOp && pStreamScan->pTableScanOp->info) {
destroyOperatorInfo(pStreamScan->pTableScanOp);
destroyOperator(pStreamScan->pTableScanOp);
}
if (pStreamScan->tqReader) {
@ -2427,7 +2422,7 @@ SOperatorInfo* createStreamScanOperatorInfo(SReadHandle* pHandle, STableScanPhys
if (pHandle->initTqReader) {
ASSERT(pHandle->tqReader == NULL);
pInfo->tqReader = tqOpenReader(pHandle->vnode);
pInfo->tqReader = tqReaderOpen(pHandle->vnode);
ASSERT(pInfo->tqReader);
} else {
ASSERT(pHandle->tqReader);

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