[peruser] Problem with reverseproxy configuration
Samuel.Degrande at lifl.fr
Tue Nov 14 06:45:15 MST 2006
First of all, I would like to thank you a lot for this Apache MPM which
is what I really
need to protect my SVN repositories.
( and secondly, excuse me for my bad english, I'm only a poor french :-( )
However I have a problem with my configuration. Here is an extract of my
test config :
### Section 1: Global Environment
Multiplexer www www
Processor www www
Processor svn svn
### VHost : host
ServerEnvironment www www
RewriteRule ^/test-svn/(.*) http://svnhost.lifl.fr/test-svn/$1 [P]
RewriteRule ^/test-svn http://.fr/test-svn.lifl.fr/ [P]
### VHost : svnhost
ServerEnvironment svn svn
CustomLog var/logs/svnhost-access_log common
Now, here is my problem :
- access to http://host.lifl.fr/index.html is OK
- access to http://svnhost.lifl.fr/test-svn/index.html is OK
- access to http://host.lifl.fr/test-svn/index.html (which is rewritten
in http://svnhost.lifl.fr/test-svn/index.html) is not working
(the connection is closed, but nothing is returned to the web browser)
With the classical Apache MPM, everything is fine.
In svnhost-access_log, there is a "GET /test-svn/index.html HTTP1/1" 200 90
So the rewriting and the internal request to the svnhost vhost seems to
be handled correctly.
I did compile peruser with debug mode activated.
I will not copy here the whole trace which is really big, but here is
what happens :
MULTIPLEXER 0 gets a request for host.lifl.fr/...
child#0 activates child#1
PROCESSOR 1 (child#1) gets the request for host.lifl.fr/...
the last trace form child#1 is : peruser_post_read(): request for
host.lifl.fr seems to be for us
MULTIPLEXER 0 (child#0) gets a request for svnhost.lifl.fr/....
(so I guess that child#1 rewrote the first request, and launched
an internal new request)
child#0 activates child#2, and calls make_child()
child#1 calls just_die()
PROCESSOR 2 (child#2) gets the request for svnhost.lifl.fr/...
child#2 clones itself into child#3
child#0 calls make_child()
child#2 calls just_die()
WORKER 3 (child#3) calls listen_add()
child#0 gets the signal from child#2 being died
child#3 calls just_die()
and that's all, child#0 returns in standby state
I can send you some more trace infos if needed.
So, should an internal 'reverseproxy' works with the peruser module, or
is there a problem with my
Samuel Degrande LIFL - UMR8022 CNRS - INRIA Futurs - Bat M3
Phone: (33)188.8.131.52.30 USTL - Universite de Lille 1
Fax: (33)184.108.40.206.37 59655 VILLENEUVE D'ASCQ CEDEX - FRANCE
[CA certs: http://igc.services.cnrs.fr/CNRS-Standard/recherche.html ]
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 4039 bytes
Desc: S/MIME Cryptographic Signature
Url : http://www.telana.com/pipermail/peruser/attachments/20061114/39e9876a/smime.bin
More information about the Peruser