[MCOL-5645] Primproc Crash | libthreadpool.so | boost10shared_ptr Created: 2024-01-25  Updated: 2024-02-01

Status: Open
Project: MariaDB ColumnStore
Component/s: None
Affects Version/s: 23.02.7
Fix Version/s: None

Type: Bug Priority: Major
Reporter: Allen Herrera Assignee: Allen Herrera
Resolution: Unresolved Votes: 0
Labels: None


 Description   

Customer running daily agg workload ran into this crash trace, about 1/3rd into the test.

Trace

Date/time: 2024-01-25 14:40:01
Signal: 6
/usr/bin/PrimProc(+0xb91b6)[0x5653888971b6]
/lib64/libpthread.so.0(+0x12cf0)[0x7fa23e193cf0]
/lib64/libc.so.6(gsignal+0x10f)[0x7fa23cbaeaff]
/lib64/libc.so.6(abort+0x127)[0x7fa23cb81ea5]
/lib64/libc.so.6(+0x21d79)[0x7fa23cb81d79]
/lib64/libc.so.6(+0x47456)[0x7fa23cba7456]
/lib64/libthreadpool.so(_ZN14error_handling12sendErrorMsgEtjjN5boost10shared_ptrIN11messageqcpp8IOSocketEEE+0x97)[0x7fa23df2f547]
/lib64/libthreadpool.so(+0x1836f)[0x7fa23df1a36f]
/usr/bin/PrimProc(+0xbac97)[0x565388898c97]
/lib64/libpthread.so.0(+0x81ca)[0x7fa23e1891ca]
/lib64/libc.so.6(clone+0x43)[0x7fa23cb99e73]

Analyze Results:

[root@ip-172-31-42-182 rocky]# bash crash.sh "/lib64/libthreadpool.so(_ZN14error_handling12sendErrorMsgEtjjN5boost10shared_ptrIN11messageqcpp8IOSocketEEE+0x97)[0x7fa23df2f547]"
File Options:
/usr/lib/debug/usr/lib64/libthreadpool.so-10.6.14_9_23.02.7-1.el8.x86_64.debug
/usr/lib64/libthreadpool.so
Target file: /usr/lib/debug/usr/lib64/libthreadpool.so-10.6.14_9_23.02.7-1.el8.x86_64.debug
Function: _ZN14error_handling12sendErrorMsgEtjjN5boost10shared_ptrIN11messageqcpp8IOSocketEEE
Offset: 0x97
nm /usr/lib/debug/usr/lib64/libthreadpool.so-10.6.14_9_23.02.7-1.el8.x86_64.debug | grep -i _ZN14error_handling12sendErrorMsgEtjjN5boost10shared_ptrIN11messageqcpp8IOSocketEEE
000000000002d4b0 T _ZN14error_handling12sendErrorMsgEtjjN5boost10shared_ptrIN11messageqcpp8IOSocketEEE
0000000000017f47 t _ZN14error_handling12sendErrorMsgEtjjN5boost10shared_ptrIN11messageqcpp8IOSocketEEE.cold
Adding: 000000000002d4b0 +  0x97
Total Offset: 0x2d547
Final: addr2line -e /usr/lib/debug/usr/lib64/libthreadpool.so-10.6.14_9_23.02.7-1.el8.x86_64.debug 0x2d547
================= RESULTS =================
addr2line: Dwarf Error: Offset (738551022) greater than or equal to .debug_str size (11695905).
addr2line: Dwarf Error: Offset (2080725742) greater than or equal to .debug_str size (11695905).
/usr/src/debug/MariaDB-/src_0/storage/columnstore/columnstore/.boost/boost-lib/include/boost/smart_ptr/detail/shared_count.hpp:432

 
    ~shared_count() /*BOOST_SP_NOEXCEPT*/
    {
        if( pi_ != 0 ) pi_->release();                        <------ 432
#if defined(BOOST_SP_ENABLE_DEBUG_HOOKS)
        id_ = 0;
#endif
    }
 
    shared_count(shared_count const & r) BOOST_SP_NOEXCEPT: pi_(r.pi_)
#if defined(BOOST_SP_ENABLE_DEBUG_HOOKS)
        , id_(shared_count_id)
#endif
    {
        if( pi_ != 0 ) pi_->add_ref_copy();
    }

Could be related or not to https://jira.mariadb.org/browse/MCOL-5636


Generated at Thu Feb 08 02:59:24 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.