Closed GoogleCodeExporter closed 8 years ago
Also, when I restart Apache it shows lots of broken processes:
[Mon Apr 12 11:25:01 2010] [warn] child process 18187 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 18188 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17924 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17931 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17934 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17224 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17817 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17818 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:01 2010] [warn] child process 17950 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 18187 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 18188 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17924 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17931 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17934 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17224 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17817 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17818 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:03 2010] [warn] child process 17950 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 18187 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17924 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17931 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17934 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17224 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17817 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17818 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:05 2010] [warn] child process 17950 still did not exit,
sending a SIGTERM
[Mon Apr 12 11:25:07 2010] [error] child process 18187 still did not exit,
sending a SIGKILL
[Mon Apr 12 11:25:07 2010] [error] child process 17924 still did not exit,
sending a SIGKILL
[Mon Apr 12 11:25:07 2010] [error] child process 17931 still did not exit,
sending a SIGKILL
[Mon Apr 12 11:25:07 2010] [error] child process 17224 still did not exit,
sending a SIGKILL
[Mon Apr 12 11:25:07 2010] [error] child process 17817 still did not exit,
sending a SIGKILL
[Mon Apr 12 11:25:07 2010] [error] child process 17950 still did not exit,
sending a SIGKILL
[Mon Apr 12 11:25:08 2010] [notice] SIGHUP received. Attempting to restart
Original comment by thijs.triemstra@gmail.com
on 12 Apr 2010 at 11:28
This has also been reported for CentOS on the mailinglist:
http://groups.google.com/group/modwsgi/browse_thread/thread/ba82b2643564d2dd/1a4
3981dfdd4510b?
Original comment by thijs.triemstra@gmail.com
on 12 Apr 2010 at 11:29
The KeyError in itself should be harmless and so far hadn't been any indication
that other problems were
occurring. If the processes themselves are also hanging on shutdown, that is
more of an issue.
Short story is that they have changed the embedded behaviour of Python in
2.6.5, partly returning some
behaviour that they had fiddled with back in Python 2.5.1 which mod_wsgi had to
provide a work around for to
ensure things still worked as required. The work around may be causing an
action to occur twice in 2.6.5 now
because of their further changes. Not sure if this is causing the problems.
Cant even get Python 2.6.5 to completely work properly on MacOS X 10.5
(Leopard), but to the extent that can,
cant duplicate the issue. Still need to try MacOS X 10.6 (Snow Leopard).
Original comment by Graham.Dumpleton@gmail.com
on 12 Apr 2010 at 11:45
Original comment by Graham.Dumpleton@gmail.com
on 12 Apr 2010 at 11:45
The message is harmless. Fix added in revision 1588 of subversion trunk for
mod_wsgi 4.0 to get rid of it anyway.
If want to get rid of it in older versions of mod_wsgi, find call to
Py_Finalize() and add immediately before that:
if (!PyImport_AddModule("dummy_threading"))
PyErr_Clear();
This is enough to get Python threading module to ignore the error rather than
re raise the error causing the
message to be logged.
Will be back ported to mod_wsgi 2.X and 3.X only if bug fix versions of those
need to be released for more serious
reasons.
Note that this issue also affects Python 3.1.2 and later in 3.X series.
Original comment by Graham.Dumpleton@gmail.com
on 16 Apr 2010 at 12:23
Turns out this was a harmless message indeed, thanks for the follow up, cheers.
Original comment by thijs.triemstra@gmail.com
on 19 Apr 2010 at 6:47
Backport to mod_wsgi 3.X branch for 3.3 in revision 1605.
Original comment by Graham.Dumpleton@gmail.com
on 12 Jul 2010 at 2:03
Version 3.3 has been released.
Original comment by Graham.Dumpleton@gmail.com
on 31 Jul 2010 at 10:39
Original issue reported on code.google.com by
thijs.triemstra@gmail.com
on 12 Apr 2010 at 11:26