Skip to content

[Bug] BE coredump when create new tablet #5090

@acelyc111

Description

@acelyc111

Describe the bug
I found a coredump of BE when create new tablet:

Core was generated by `/home/work/app/doris/zjyprc-newretail/be/package/be/lib/palo_be'.
Program terminated with signal 11, Segmentation fault.
#0  0x0000000000de6fda in construct<doris::TabletInfo, doris::TabletInfo const&> (this=<optimized out>, __p=<optimized out>) at /usr/include/c++/7.3.0/ext/new_allocator.h:111
111	/usr/include/c++/7.3.0/ext/new_allocator.h: 没有那个文件或目录.
Missing separate debuginfos, use: debuginfo-install glibc-2.17-157.el7_3.1.x86_64 libgcc-4.8.5-28.el7_5.1.x86_64 zlib-1.2.7-17.el7.x86_64
(gdb) bt
#0  0x0000000000de6fda in construct<doris::TabletInfo, doris::TabletInfo const&> (this=<optimized out>, __p=<optimized out>) at /usr/include/c++/7.3.0/ext/new_allocator.h:111
#1  construct<doris::TabletInfo, doris::TabletInfo const&> (__a=..., __p=<optimized out>) at /usr/include/c++/7.3.0/bits/alloc_traits.h:475
#2  _M_construct_node<doris::TabletInfo const&> (this=<optimized out>, __node=<optimized out>) at /usr/include/c++/7.3.0/bits/stl_tree.h:626
#3  _M_create_node<doris::TabletInfo const&> (this=<optimized out>) at /usr/include/c++/7.3.0/bits/stl_tree.h:643
#4  operator()<const doris::TabletInfo&> (this=0x7f338a724a08, __arg=...) at /usr/include/c++/7.3.0/bits/stl_tree.h:556
#5  _M_clone_node<std::_Rb_tree<doris::TabletInfo, doris::TabletInfo, std::_Identity<doris::TabletInfo>, std::less<doris::TabletInfo>, std::allocator<doris::TabletInfo> >::_Alloc_node> (this=0x7f338a724a40, __node_gen=..., __x=0x3330346436346135)
    at /usr/include/c++/7.3.0/bits/stl_tree.h:666
#6  std::_Rb_tree<doris::TabletInfo, doris::TabletInfo, std::_Identity<doris::TabletInfo>, std::less<doris::TabletInfo>, std::allocator<doris::TabletInfo> >::_M_copy<std::_Rb_tree<doris::TabletInfo, doris::TabletInfo, std::_Identity<doris::TabletInfo>, std::less<doris::TabletInfo>, std::allocator<doris::TabletInfo> >::_Alloc_node> (this=this@entry=0x7f338a724a40, __x=0x3330346436346135, __p=0x2840ceec0, __p@entry=0x7f338a724a48, __node_gen=...) at /usr/include/c++/7.3.0/bits/stl_tree.h:1830
#7  0x0000000000ddecd1 in _M_copy<std::_Rb_tree<doris::TabletInfo, doris::TabletInfo, std::_Identity<doris::TabletInfo>, std::less<doris::TabletInfo>, std::allocator<doris::TabletInfo> >::_Alloc_node> (__gen=..., __x=..., this=0x7f338a724a40)
    at /usr/include/c++/7.3.0/bits/stl_tree.h:875
#8  _M_copy (__x=..., this=0x7f338a724a40) at /usr/include/c++/7.3.0/bits/stl_tree.h:886
#9  _Rb_tree (__x=..., this=0x7f338a724a40) at /usr/include/c++/7.3.0/bits/stl_tree.h:924
#10 set (this=0x7f338a724a40) at /usr/include/c++/7.3.0/bits/stl_set.h:213
#11 tablet_set (this=0x5f22d80) at /builds/olap/doris/be/src/olap/data_dir.h:126
#12 doris::StorageEngine::get_stores_for_create_tablet (this=this@entry=0x5939c00, storage_medium=<optimized out>) at /builds/olap/doris/be/src/olap/storage_engine.cpp:423
#13 0x0000000000ddefa6 in doris::StorageEngine::create_tablet (this=0x5939c00, request=...) at /builds/olap/doris/be/src/olap/storage_engine.cpp:814
#14 0x00000000013f088f in doris::TaskWorkerPool::_create_tablet_worker_thread_callback (arg_this=0x5944a20) at /builds/olap/doris/be/src/agent/task_worker_pool.cpp:338
#15 0x00007f33f62dfdc5 in start_thread () from /lib64/libpthread.so.0
#16 0x00007f33f65eb73d in clone () from /lib64/libc.so.6
(gdb)

It was cause by DataDir::tablet_set, it's out of lock:

std::set<TabletInfo> tablet_set() { return _tablet_set; }

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions