[mod_python] Re: child process XXXX still did not exit, sending a SIGTERM

Martin Blais blais at furius.ca
Mon Jan 2 18:20:48 EST 2006

On 1/2/06, Graham Dumpleton <grahamd at dscpl.com.au> wrote:
> Maybe mod_python has to be a bit more orderly as far as how it ends
> interpreters. Ie., maybe it should end each interpreter it has created
> explicitly using Py_EndInterpreter().
> This though may be dependent on first getting changes described in:
>   http://issues.apache.org/jira/browse/MODPYTHON-77
> incorporated. Specifically, the attachment:
>   http://issues.apache.org/jira/secure/attachment/12320489/grahamd_20051105.tar.gz
> Actually, this change may in itself fix the issue as it saves away the
> first interpreter state as "main_interpreter" and that will then be the
> interpreter state used to do the cleanup. This isn't the case at the
> moment.
> Can you apply that patch and see if things improve?

Thanks, will do!

> Beyond that, as I suggest, maybe it needs to be more orderly and
> end each interpreter besides "main_interpreter" and then finally call
> Py_Finalize().

I just found more info by fiddling just now, it seems to happen only
-- but not always, but most of the time anyway -- when I do a
redirect!  When I do not redirect I never have the problem.  When I
redirect it hangs most of the time, apart from the occasional clean

That is, my when handler returns code 302.

Will investigate further in a few hours, this is a good starting point I think.

thanks Graham.

More information about the Mod_python mailing list