Repository navigation
[Bug] (planner) SlotDescriptor set as 'is not null' at right hand in left outer join #9925
Description
Search before asking
- I had searched in the issues and found no similar issues.
Version
- master
- dev-1.0.1
What's Wrong?
When plan left outer join, we need to set all SlotDescriptor at right hand to nullable. But under some case, it is not happen anymore.
Because of this error of planner, BE produce wrong result or generate a core dump when execute this query plan.
core dump as below:
#0 0x00007f1477f26bb5 in ?? () from /usr/lib/x86_64-linux-gnu/libc-2.31.so
#1 0x0000562e62f8c61d in google::LogMessage::Fail () at src/logging.cc:1650
#2 0x0000562e62f8eb59 in google::LogMessage::SendToLog (this=0x7f1428072020) at src/logging.cc:1607
#3 0x0000562e62f8c186 in google::LogMessage::Flush (this=0x7f1428072020) at src/logging.cc:1476
#4 0x0000562e62f8f1c9 in google::LogMessageFatal::~LogMessageFatal (this=<optimized out>, __in_chrg=<optimized out>) at src/logging.cc:2226
#5 0x0000562e5d2e27b9 in doris::MemPool::allocate<false> (alignment=8, size=-4702111234474983746, this=0x61400086c0f8) at /root/incubator-doris/be/src/runtime/mem_pool.h:248
#6 doris::MemPool::allocate (this=0x61400086c0f8, size=-4702111234474983746) at /root/incubator-doris/be/src/runtime/mem_pool.h:111
#7 0x0000562e5dd3b6d5 in doris::Tuple::deep_copy (this=0x6250001f20b0, dst=0x625004261010, desc=..., pool=0x61400086c0f8, convert_ptrs=false) at /root/incubator-doris/be/src/runtime/tuple.cpp:79
#8 0x0000562e5dd3b2c0 in doris::Tuple::deep_copy (this=0x6250001f20b0, desc=..., pool=0x61400086c0f8, convert_ptrs=false) at /root/incubator-doris/be/src/runtime/tuple.cpp:65
#9 0x0000562e5dcf6861 in doris::Tuple::deep_copy (this=0x6250001f20b0, desc=..., pool=0x61400086c0f8) at /root/incubator-doris/be/src/runtime/tuple.h:77
#10 0x0000562e5dcf6ede in doris::TupleRow::deep_copy (this=0x62d00cd96400, dst=0x62d003eee400, descs=std::vector of length 1, capacity 1 = {...}, pool=0x61400086c0f8, reuse_tuple_mem=false) at /root/incubator-doris/be/src/runtime/tuple_row.h:61
#11 0x0000562e5dcf1dd0 in doris::RowBatch::deep_copy_to (this=0x61400078f840, dst=0x61400086c040) at /root/incubator-doris/be/src/runtime/row_batch.cpp:561
#12 0x0000562e5dd7a318 in doris::DataStreamRecvr::SenderQueue::add_batch (this=0x6140007a2240, batch=0x61400078f840, use_move=false) at /root/incubator-doris/be/src/runtime/data_stream_recvr.cc:278
#13 0x0000562e5dd7efd0 in doris::DataStreamRecvr::add_batch (this=0x612000912340, batch=0x61400078f840, sender_id=3, use_move=false) at /root/incubator-doris/be/src/runtime/data_stream_recvr.cc:483
#14 0x0000562e5f583902 in doris::DataStreamSender::Channel::send_local_batch (this=0x615000154300, batch=0x61400078f840, use_move=false) at /root/incubator-doris/be/src/runtime/data_stream_sender.cpp:225
#15 0x0000562e5f58a16c in doris::DataStreamSender::send (this=0x616001931780, state=0x61d0003f7080, batch=0x61400078f840) at /root/incubator-doris/be/src/runtime/data_stream_sender.cpp:455
#16 0x0000562e5dcd29aa in doris::PlanFragmentExecutor::open_internal (this=0x61a000439ef0) at /root/incubator-doris/be/src/runtime/plan_fragment_executor.cpp:404
#17 0x0000562e5dccf4ec in doris::PlanFragmentExecutor::open (this=0x61a000439ef0) at /root/incubator-doris/be/src/runtime/plan_fragment_executor.cpp:257
#18 0x0000562e5dacb2d0 in doris::FragmentExecState::execute (this=0x61a000439e80) at /root/incubator-doris/be/src/runtime/fragment_mgr.cpp:231
#19 0x0000562e5dad2fbc in doris::FragmentMgr::_exec_actual(std::shared_ptr<doris::FragmentExecState>, std::function<void (doris::PlanFragmentExecutor*)>) (this=0x614000004640, exec_state=std::shared_ptr<doris::FragmentExecState> (use count 3, weak count 0) = {...}, cb=...) at /root/incubator-doris/be/src/runtime/fragment_mgr.cpp:464
What You Expected?
all SlotDescriptor at right hand to nullable. The query can be executed normally
How to Reproduce?
create table t1(
k1 int not null,
v1 int not null
)
distributed by hash(k1)
properties(
'replication_num' = '1'
);
create table t2(
k1 int not null,
c1 varchar(255) not null
)
distributed by hash(k1)
properties('replication_num' = '1');
insert into t1 values(1, 1), (2, 2), (3, 3), (4, 4);
insert into t2 values(1, '1'), (2, '2');
explain verbose
select * from
(select * from t1) a
left outer join
(select * from t2) b
on a.k1 = b.k1
order by a.v1; Anything Else?
No response
Are you willing to submit PR?
- Yes I am willing to submit a PR!
Code of Conduct
- I agree to follow this project's Code of Conduct
Activity