pyibbc에서 FreeTDS ODBC를 사용할 때 glibc가 손상된 이중 연결 목록을 찾았습니다.

Question

우리 프로덕션 환경에서 pyodbc 4.0.1을 실행 중이었고 갑자기 메모리 오류가 발생하기 시작했습니다. 그때까지 모든 것이 잘 작동했기 때문에 왜 시작되었는지 확신 할 수 없습니다. 서버 구성 측면에서 알려진 문제점이 있습니까?

DEV 및 UAT에서 Pyodbc 버전을 4.0.1에서 4.0.16으로 업그레이드했습니다. 불행히도 문제는 여전히 존재하며 재현 가능합니다 (아래 추적 참조).

[...] 
*** glibc detected *** python: corrupted double-linked list: 0x0000000002644650 *** 
======= Backtrace: ========= 
/lib64/libc.so.6(+0x75f4e)[0x7f751c639f4e] 
/lib64/libc.so.6(+0x79066)[0x7f751c63d066] 
/usr/lib64/libtdsodbc.so.0(+0x1e2fd)[0x7f74fd2aa2fd] 
/usr/lib64/libtdsodbc.so.0(+0x2095e)[0x7f74fd2ac95e] 
/usr/lib64/libtdsodbc.so.0(+0x27de5)[0x7f74fd2b3de5] 
/usr/lib64/libtdsodbc.so.0(+0x3729f)[0x7f74fd2c329f] 
/usr/lib64/libtdsodbc.so.0(+0x11e15)[0x7f74fd29de15] 
/usr/lib64/libtdsodbc.so.0(SQLGetTypeInfo+0x187)[0x7f74fd2a0547] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/lib/libodbc.so.2(SQLGetTypeInfo+0x1e3)[0x7f75033c8c93] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/lib/python3.5/site-packages/pyodbc.cpython-35m-x86_64-linux-gnu.so(_Z17GetConnectionInfoP7_objectP10Connection+0x42f)[0x7f75036297af] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/lib/python3.5/site-packages/pyodbc.cpython-35m-x86_64-linux-gnu.so(_Z14Connection_NewP7_objectbblbS0_R6Object+0x6a9)[0x7f7503621249] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/lib/python3.5/site-packages/pyodbc.cpython-35m-x86_64-linux-gnu.so(+0xb174)[0x7f7503622174] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyCFunction_Call+0x131)[0x7f751d4c4621] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyEval_EvalFrameEx+0x9017)[0x7f751d54bc37] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyEval_EvalFrameEx+0x9546)[0x7f751d54c166] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyEval_EvalFrameEx+0x9546)[0x7f751d54c166] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyEval_EvalFrameEx+0x9546)[0x7f751d54c166] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(+0x144b49)[0x7f751d54cb49] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyEval_EvalCodeEx+0x48)[0x7f751d54ccd8] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyEval_EvalCode+0x3b)[0x7f751d54cd1b] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyRun_FileExFlags+0x130)[0x7f751d572020] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(PyRun_SimpleFileExFlags+0x173)[0x7f751d573623] 
/opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/../lib/libpython3.5m.so.1.0(Py_Main+0xca7)[0x7f751d58e8c7] 
python(main+0x15d)[0x400add] 
/lib64/libc.so.6(__libc_start_main+0xfd)[0x7f751c5e2d5d] 
python[0x4008b9] 
======= Memory map: ======== 
00400000-00401000 r-xp 00000000 fd:05 11011012       /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/python3.5 
00601000-00602000 rw-p 00001000 fd:05 11011012       /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/envs/ed/bin/python3.5 
01241000-02873000 rw-p 00000000 00:00 0         [heap] 
7f74f8000000-7f74f8021000 rw-p 00000000 00:00 0 
7f74f8021000-7f74fc000000 ---p 00000000 00:00 0 
7f74fc354000-7f74fc356000 r-xp 00000000 fd:01 3837      /usr/lib64/gconv/CP1252.so 
7f74fc356000-7f74fc555000 ---p 00002000 fd:01 3837      /usr/lib64/gconv/CP1252.so 
7f74fc555000-7f74fc556000 r--p 00001000 fd:01 3837      /usr/lib64/gconv/CP1252.so 
7f74fc556000-7f74fc557000 rw-p 00002000 fd:01 3837      /usr/lib64/gconv/CP1252.so 
7f74fc557000-7f74fc558000 r-xp 00000000 fd:01 4029      /usr/lib64/gconv/ISO8859-1.so 
7f74fc558000-7f74fc758000 ---p 00001000 fd:01 4029      /usr/lib64/gconv/ISO8859-1.so 
7f74fc758000-7f74fc759000 r--p 00001000 fd:01 4029      /usr/lib64/gconv/ISO8859-1.so 
7f74fc759000-7f74fc75a000 rw-p 00002000 fd:01 4029      /usr/lib64/gconv/ISO8859-1.so 
7f74fc75a000-7f74fc75f000 r-xp 00000000 fd:01 3803      /lib64/libnss_dns-2.12.so 
7f74fc75f000-7f74fc95e000 ---p 00005000 fd:01 3803      /lib64/libnss_dns-2.12.so 
7f74fc95e000-7f74fc95f000 r--p 00004000 fd:01 3803      /lib64/libnss_dns-2.12.so 
7f74fc95f000-7f74fc960000 rw-p 00005000 fd:01 3803      /lib64/libnss_dns-2.12.so 
7f74fc960000-7f74fc963000 r-xp 00000000 fd:01 5388      /lib64/libgpg-error.so.0.5.0 
7f74fc963000-7f74fcb62000 ---p 00003000 fd:01 5388      /lib64/libgpg-error.so.0.5.0 
7f74fcb62000-7f74fcb63000 r--p 00002000 fd:01 5388      /lib64/libgpg-error.so.0.5.0 
7f74fcb63000-7f74fcb64000 rw-p 00003000 fd:01 5388      /lib64/libgpg-error.so.0.5.0 
7f74fcb64000-7f74fcb74000 r-xp 00000000 fd:01 6697      /usr/lib64/libtasn1.so.3.1.6 
7f74fcb74000-7f74fcd73000 ---p 00010000 fd:01 6697      /usr/lib64/libtasn1.so.3.1.6 
7f74fcd73000-7f74fcd74000 rw-p 0000f000 fd:01 6697      /usr/lib64/libtasn1.so.3.1.6 
7f74fcd74000-7f74fcde6000 r-xp 00000000 fd:01 5401      /lib64/libgcrypt.so.11.5.3 
7f74fcde6000-7f74fcfe5000 ---p 00072000 fd:01 5401      /lib64/libgcrypt.so.11.5.3 
7f74fcfe5000-7f74fcfe6000 r--p 00071000 fd:01 5401      /lib64/libgcrypt.so.11.5.3 
7f74fcfe6000-7f74fcfe9000 rw-p 00072000 fd:01 5401      /lib64/libgcrypt.so.11.5.3 
7f74fcfe9000-7f74fd086000 r-xp 00000000 fd:01 13301      /usr/lib64/libgnutls.so.26.14.12 
7f74fd086000-7f74fd285000 ---p 0009d000 fd:01 13301      /usr/lib64/libgnutls.so.26.14.12 
7f74fd285000-7f74fd28c000 rw-p 0009c000 fd:01 13301      /usr/lib64/libgnutls.so.26.14.12 
7f74fd28c000-7f74fd2ee000 r-xp 00000000 fd:01 52715      /usr/lib64/libtdsodbc.so.0.0.0 
7f74fd2ee000-7f74fd4ed000 ---p 00062000 fd:01 52715      /usr/lib64/libtdsodbc.so.0.0.0 
7f74fd4ed000-7f74fd4f1000 rw-p 00061000 fd:01 52715      /usr/lib64/libtdsodbc.so.0.0.0 
7f74fd4f1000-7f74fd4f5000 r-xp 00000000 fd:05 7997029     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/python-3.5.2-0/lib/python3.5/lib-dynload/termios.so 
7f74fd4f5000-7f74fd6f5000 ---p 00004000 fd:05 7997029     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/python-3.5.2-0/lib/python3.5/lib-dynload/termios.so 
7f74fd6f5000-7f74fd6f7000 rw-p 00004000 fd:05 7997029     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/python-3.5.2-0/lib/python3.5/lib-dynload/termios.so 
7f74fd6f7000-7f74fd6fa000 r-xp 00000000 fd:05 7997070     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/python-3.5.2-0/lib/python3.5/lib-dynload/fcntl.so 
7f74fd6fa000-7f74fd8f9000 ---p 00003000 fd:05 7997070     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/python-3.5.2-0/lib/python3.5/lib-dynload/fcntl.so 
7f74fd8f9000-7f74fd8fb000 rw-p 00002000 fd:05 7997070     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/python-3.5.2-0/lib/python3.5/lib-dynload/fcntl.so 
7f74fd8fb000-7f74fd8fe000 r-xp 00000000 fd:05 10749483     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cresultproxy.so 
7f74fd8fe000-7f74fdafe000 ---p 00003000 fd:05 10749483     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cresultproxy.so 
7f74fdafe000-7f74fdaff000 rw-p 00003000 fd:05 10749483     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cresultproxy.so 
7f74fdaff000-7f74fdb00000 r-xp 00000000 fd:05 10749167     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cutils.so 
7f74fdb00000-7f74fdd00000 ---p 00001000 fd:05 10749167     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cutils.so 
7f74fdd00000-7f74fdd01000 rw-p 00001000 fd:05 10749167     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cutils.so 
7f74fdd01000-7f74fdd41000 rw-p 00000000 00:00 0 
7f74fdec1000-7f74fdf01000 rw-p 00000000 00:00 0 
7f74fdf41000-7f74fdf44000 r-xp 00000000 fd:05 10749471     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cprocessors.so 
7f74fdf44000-7f74fe143000 ---p 00003000 fd:05 10749471     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cprocessors.so 
7f74fe143000-7f74fe144000 rw-p 00002000 fd:05 10749471     /opt/cloudera/parcels/Anaconda3-4.2.0-python-3.5/pkgs/sqlalchemy-1.0.13-py35_0/lib/python3.5/site-packages/sqlalchemy/cprocessors.soAborted 

Microsoft SQL Server 데이터베이스에 연결할 때 문제가 추가되는 것으로 보입니다. dsn-less를 다음 연결 문자열과 연결합니다. 이것은 메모리 문제가 언제 발생했는지 (연결 전/도중/후에) 확실하지 않습니다 ... 문제가 다소 불규칙하게 트리거 된 것으로 보입니다.

[...] 
pyodbc.pooling=False 
oConnexion = pyodbc.connect(Driver="{FreeTDS};Server=xxx;Port=1433;TDS_Version=7.0;UID=xxx;PWD=xxx;", autocommit=True) 
oConnexion.setdecoding(pyodbc.SQL_CHAR, encoding='utf-8') 
oConnexion.setdecoding(pyodbc.SQL_WCHAR, encoding='utf-8') 
oConnexion.setdecoding(pyodbc.SQL_WMETADATA, encoding='utf-8') 
oConnexion.setencoding(encoding='utf-8') 
[...] 

구성 :
- 파이썬 3.5.2
- 아나콘다 4.2.0 (64 비트)
- 리눅스에서 레드햇 4.4.7-1가
- Pyodbc 4.0.16

Answer 1

이전 ODBC 드라이버 구성은 FreeTDS v0.91입니다. ODBC 드라이버를 FreeTDS 1.0rc5로 업그레이드하면이 문제가 해결되었습니다.

Answer 2

이 모든 것이 그때까지 잘 작동 된 이후에 발생하기 시작하는 이유

나는 ... 이제까지

확실하지 않다 너가 무언가를 바꿀 때까지 ything은 잘 작동했다. 당신은 당신이 무엇을 바꾸 었는지 알 필요가 있습니다.

서버 구성과 관련하여 알려진 문제점은 있습니까?

문제는 서버 구성에 아무것도이 없습니다. 문제는 파이썬에로드하는 것이 버그이며 힙 손상을 일으키는 것입니다.

파이썬을 Valgrind에서 실행하면 힙 손상이 어디에서 발생하는지 알 수 있습니다.