[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

Michael Hu added the comment: Could you please look into the core (attached into this bug) to figure out? This issue does not happen always. We are tight on resource to reproduce this. -- ___ Python tracker

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

STINNER Victor added the comment: By the way, you can try your application on Python 3.6 using PYTHONMALLOC=debug env var. Or retry your application on Python 2.7 recompiled using ./configure --with-pydebug. I would like to know if Python sees a buffer overflow. --

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

STINNER Victor added the comment: Ok but can you also try faulthandler to see all frames of all threads? -- ___ Python tracker ___

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

INADA Naoki added the comment: Can you check backtrace of main thread? -- ___ Python tracker ___ ___

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

Michael Hu added the comment: (gdb) py-bt #4 Frame 0x7f0ab7a2db60, for file /etc/remoting/remoting_agent.zip/Pyro4/socketutil.py, line 463, in close (self=) self.sock.shutdown(socket.SHUT_RDWR) #7 Frame 0x7f0ab0001760, for file /etc/remoting/remoting_agent.zip/Pyro4/socketutil.py, line

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

STINNER Victor added the comment: Try to get the Python traceback of thread on the crash: try the faulthandler module. -- nosy: +haypo ___ Python tracker

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

Michael Hu added the comment: Core is uploaded for python 2.7.10 to assist debugging. -- Added file: http://bugs.python.org/file45492/core_python2.7.10.gz ___ Python tracker

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

INADA Naoki added the comment: Traceback tells: #22 0x005a52db in PyInterpreterState_Clear (interp=0x95ac80) at ../Python/pystate.c:111 Py_CLEAR(interp->sysdict); While cleaning up interp->sysdict... #19 0x004d86c6 in tb_dealloc.46270 (tb=0x7f0aa1278f38) at

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

Eric V. Smith added the comment: Can you provide a short sample that causes this error? Without a way to reproduce it, there's not a lot we can do. -- nosy: +eric.smith ___ Python tracker

[issue28673] When using pyro4 with more than 15 threads, python 2.7.12 cores frequently

New submission from Michael Hu: When using pyro4 with more than 15 threads, python 2.7.12 cores frequently (>60% time) Note "v" (op in frame 1) in frame 2 is NULL which has some value in frame 3. So some other thread cleans it. === gdb === Using host libthread_db library