[Twisted-web] Different behavior of twistd with and without -n

Terry Jones terry at jon.es
Fri May 25 11:24:17 EDT 2007


>>>>> "Seb" =3D=3D S=E9bastien LELONG <sebastien.lelong at dexia-securities.=
fr> writes:

Seb> FWIW, I have an application running CherryPy over Twisted, using
Seb> PyLucene.  Since PyLucene isn't thread-safe, every thread must
Seb> subclass PyLucene.PythonThread, so the underlying "java runtime
Seb> garbage collector is aware about the python thread upon their
Seb> creation"
Seb> (http://lists.osafoundation.org/pipermail/pylucene-dev/2004-June/000=
037.html).
Seb> Without this, the apps simply crashes (segfault, see also
Seb> http://chandlerproject.org/PyLucene/ThreadingInPyLucene). Maybe
Seb> there's something about that while twistd is daemonizing the process=
 ?

Hi S=E9bastien.

I read those pages at some point, but seeing as my code doesn't use threa=
ds
I never paid that much attention.

My app runs through the code where it crashes multiple times before thing=
s
die, so I'm not suffering from the kind of segfaulting you saw. From what=
 I
can tell, something is deliberately calling abort(3) when it detects a
problem. It doesn't look like a signal is received or a system call is
failing.

Thanks a lot for the suggestion. I hope I'm understanding you right - tha=
t
it's not an issue for me as I'm not using threads directly (but is Twiste=
d?
I'm not sure).

Terry



More information about the Twisted-web mailing list