Closed GoogleCodeExporter closed 9 years ago
Thanks for reporting, I'll check it out!
Original comment by alobbs
on 4 Dec 2008 at 4:57
I am also having this kind of issue. I use the set up I have describes there.
http://yml-blog.blogspot.com/2009/02/serving-django-via-cherrypy-behind.html
Do you have any idea of what the pb could be ?
Thanks
--yml
Original comment by yann.ma...@gmail.com
on 1 Mar 2009 at 3:44
use cherrypy wsgi server, when do any POST - got 500
Original comment by slav0n...@gmail.com
on 6 Mar 2009 at 8:31
I'm getting this issue, too. Seems to occur very frequently.
I have Cherokee/0.99.3 and lighttpd/1.4.20 on the same box. Cherokee listens
on port 80, lighty listens on
port 81. 81 is firewalled so the rest of the world can't access it.
I'm using "HTTP Reverse Proxy" to send all *.php requests to lighty at
127.0.0.1:81. From there, FastCGI
requests go to the proper machine.
OS is FreeBSD/i386 7.0-R.
Original comment by paganlad...@gmail.com
on 12 Mar 2009 at 9:00
r2971 tries to address the issue.
I took my a while to reproduce the problem. The latest trunk revision (r2971 at
this moment) works alright for
me. Could you please guys confirm?
Thanks!
Original comment by alobbs
on 13 Mar 2009 at 9:46
Still getting it with r2973. :(
Original comment by paganlad...@gmail.com
on 13 Mar 2009 at 6:09
I try the same test again and I still cannot POST using HTTP reverse proxy. The
configuration I used is the one describes in this blog post :
http://yml-blog.blogspot.com/2009/02/serving-django-via-cherrypy-behind.html
I have tested with 0.99.4 from the ppa.
Please do not hesitate to let me know if you need more information.
Regards,
--yml
Original comment by yann.ma...@gmail.com
on 14 Mar 2009 at 1:02
paganladyserena, everything seems to work alright now.. I've spent half and
hour trying to make it fail, but I
haven't seen one single error message. How can I reproduce it?
Original comment by alobbs
on 16 Mar 2009 at 10:29
yml, I think I've found the problem you hit with Django.
Could you please test tunk (>=2976)?
Original comment by alobbs
on 16 Mar 2009 at 12:19
I have just tested with 2978 and I confirm that this issue is not fixed at that
stage. I have configure cherokee using this command :
./autogen.sh --prefix=/usr --sysconfdir=/etc --localstatedir=/var
--enable-static-module=all --enable-static --enable-shared=no --enable-beta
--enable-trace
Please do not hesitate to let me know if you need more information.
Original comment by yann.ma...@gmail.com
on 16 Mar 2009 at 7:46
Could you guys please check whether the problem is fixed in 0.99.5?
I have fixed a couple of important bugs in the reverse proxy, although I'm not
sure any of them was the bug you
hit.
Original comment by alobbs
on 18 Mar 2009 at 9:57
Hello,
I have tested with r2985 and the configuration describe above and I still get
the
error below when I try to post a form:
"""
500 Internal Server Error
Cherokee web server 0.99.4b2985 (UNIX), Port 80
"""
I am attaching to this comment the trace I have collected using this command :
"""
sudo CHEROKEE_TRACE=all cherokee-worker 2>&1 >/tmp/dump_for_alo.txt
"""
Please do not hesitate to let me know if I could provide other useful
information.
Original comment by yann.ma...@gmail.com
on 18 Mar 2009 at 1:27
Attachments:
Here is where it fails (~line 11603):
handler_proxy.c:0736 ( cherokee_handler_proxy_init): Entering phase 'send post'
post.c:0418 ( cherokee_post_walk_to_fd): errno 32
handler_proxy.c:0571 ( send_post): Sending POST fd=16, ret=-1
error 32 is 'Broken pipe'. At first look, it seems that the back-end closed the
connection.
Further investigation is need..
Original comment by alobbs
on 18 Mar 2009 at 1:43
I have just tested with 0.99.6 and I can still observe this bug. It is
impossible to
post a form without getting :
"""
500 Internal Server Error
Cherokee web server 0.99.6 (Debian GNU/Linux), Port 80
"""
Also I have observed that once I get this error page if I press "F5" and
"Resend" the
form is submited and the object is created.
Original comment by yann.ma...@gmail.com
on 26 Mar 2009 at 12:25
This issue is being quite hard to reproduce, actually..
I've been testing with the following example (find it attached), and everything
seems to work file for both
servers:
- python manage.py runcpserver port=8000
"CherryPy/3.1.1 WSGI Server" accept the POSTs just fine
- python manage.py runserver
"WSGIServer/0.1 Python/2.5.1" does as well.
The only minor issue I found is that CherryPy bound to ::1, while WSGIServer
bound to 127.0.0.1. Besides that
little difference, everything worked fine.
Guys, I'm afraid I'll need some other hint to reproduce the problem.
Original comment by alobbs
on 27 Mar 2009 at 9:31
Oops! I forgot the tarball..
Original comment by alobbs
on 27 Mar 2009 at 9:34
Attachments:
I've just fixed a couple of minor bugs in the proxy.. (>= r3020)
I don't know whether or not it has some impact on this issue.
Original comment by alobbs
on 27 Mar 2009 at 12:43
I have just tested with the version 0.99.7 and I see some amelioration. I
haven't got
any 500 error when Keep Alive is switched off. However I am getting some 500
error
when keep alive is one and if I don't fill up the form fast enought.
--yml
Original comment by yann.ma...@gmail.com
on 28 Mar 2009 at 5:44
That's because of a CherryPy bug:
http://groups.google.com/group/cherrypy-users/browse_thread/thread/e751976da1d01390
http://www.cherrypy.org/ticket/853
So I guess that we have already addressed the problem at our side.
Shall I consider this bug fixed?
Original comment by alobbs
on 30 Mar 2009 at 8:46
I think yes, as far as concern I do not see any issue related to cherokeee now.
Thank for fixing them.
--yml
Original comment by yann.ma...@gmail.com
on 30 Mar 2009 at 12:22
Thank you guys for reporting!
Without your feedback I couldn't have fixed the issues :-)
Original comment by alobbs
on 30 Mar 2009 at 4:51
Original issue reported on code.google.com by
type...@gmail.com
on 4 Dec 2008 at 4:51