You should use a user and group that are not apache. I usually put each project in a different user and group so I know exactly which processes and threads are running for a specific project.
Below is a link I found that may help. http://getmediacore.com/blog/turbogears-2-tg2-with-mod_wsgi-and-virtual-environments/ ~Carl On Wed, Apr 21, 2010 at 11:12 AM, noisebleed <vitorbrandao...@gmail.com> wrote: > I hope so ;) > > Updated the vhost file according to your suggestions. > > - DocumentRoot /var/www/modwsgi/gatekeeper > > - WSGIDaemonProcess gatekeeper threads=10 processes=3 > + WSGIDaemonProcess gatekeeper user=apache group=apache processes=3 > threads=10 maximum-requests=1000 > WSGIProcessGroup gatekeeper > > The docs are really good and I've checked FAQ and the configuration > docs but I'm still searching for that little magic trick that will put > this app flying... > > Vitor (noisebleed) > > In the meanwhile I've installed TG2 in the server (virtualenv), built > the quickstart project (named to GateKeeper/gatekeeper), changed > gatekeeper.wsgi to read from virtualenv but the results are the same. > Paster serve works here too... > > On 21 Abr, 15:28, Carl Nobile <carl.nob...@gmail.com> wrote: >> Vitor, >> >> The first time through is always the roughest. You'll get it working >> and the next time you'll be the expert. >> >> I see no major issues, but there are a few thing that you could change. >> >> 1st don't define DocumentRoot at all. You don't want people accessing >> your gatekeeper.wsgi hook file or anything else in there. >> 2nd you should do something like this with the WSGIDaemonProcess >> >> WSGIDaemonProcess <some process> user=<some user> group=<some group> \ >> processes=3 threads=10 >> maximum-requests=1000 >> WSGIProcessGroup <some process> >> >> Adding a user and group makes it easy to see what is running in a ps >> -ef. Also you should always have a 'maximum-requests', this will limit >> the number of request any one thread will process thus avoiding memory >> leaks. All apps leak memory so don't assume your code won't I know for >> a fact that Python itself leaks memory. >> >> Maybe Graham will find something else. He's the author of mod_wsgi so >> he is the definitive source of info on it. >> >> Also you can check the docs for mod_wsgi, Graham is very thorough documenter. >> >> http://code.google.com/p/modwsgi/w/list >> >> ~Carl >> >> On Wed, Apr 21, 2010 at 10:08 AM, noisebleed <vitorbrandao...@gmail.com> >> wrote: >> > The file is a bit long (it's using SSL) but the WSGI section is right >> > on top. I'm posting everything anyway. >> > While testing I've used a <Directory "/var/www/modwsgi/gatekeeper/ >> > public/images"> directive and I was able to browse the images folder >> > so I guess everything is fine here. Or isn't?.. >> >> > # ----- vhost begin ----- >> >> > Listen 50000 >> >> > <IfDefine SSL> >> > <IfModule mod_ssl.c> >> > ## >> > ## SSL Virtual Host Context >> > ## >> > <VirtualHost *:50000> >> >> > # General setup for the virtual host >> > ServerName gatekeeper.localhost >> > ServerAdmin ad...@localhost >> > DocumentRoot /var/www/modwsgi/gatekeeper >> >> > WSGIScriptAlias / /var/www/modwsgi/gatekeeper/gatekeeper.wsgi >> > Alias /images /var/www/gatekeeper/public/images >> > Alias /css /var/www/gatekeeper/public/css >> > Alias /files /var/www/gatekeeper/public/files >> > Alias /js /var/www/gatekeeper/public/js >> >> > WSGIDaemonProcess gatekeeper threads=10 processes=3 >> > WSGIProcessGroup gatekeeper >> > WSGIPassAuthorization On >> >> > LogLevel debug >> >> > ErrorLog /var/log/apache2/ssl_error_log >> > <IfModule mod_log_config.c> >> > TransferLog /var/log/apache2/ssl_access_log >> > </IfModule> >> >> > # SSL Engine Switch: >> > # Enable/Disable SSL for this virtual host. >> > SSLEngine on >> > SSLCipherSuite ALL:!ADH:!EXPORT56:RC4+RSA:+HIGH:+MEDIUM:+LOW: >> > +SSLv2:+EXP:+eNULL >> > SSLCertificateFile /etc/ssl/apache2/gatekeeper/server.crt >> > SSLCertificateKeyFile /etc/ssl/apache2/gatekeeper/server.pem >> > SetEnvIf User-Agent ".*MSIE.*" nokeepalive ssl-unclean- >> > shutdown >> >> > <Files ~ "\.(cgi|shtml|phtml|php?)$"> >> > SSLOptions +StdEnvVars >> > </Files> >> >> > <IfModule mod_setenvif.c> >> > SetEnvIf User-Agent ".*MSIE.*" nokeepalive ssl-unclean- >> > shutdown \ >> > downgrade-1.0 force-response-1.0 >> > </IfModule> >> >> > # Per-Server Logging: >> > # The home of a custom SSL log file. Use this when you want >> > a >> > # compact non-error SSL logfile on a virtual host basis. >> > <IfModule mod_log_config.c> >> > CustomLog /var/log/apache2/ssl_request_log \ >> > "%t %h %{SSL_PROTOCOL}x %{SSL_CIPHER}x \"%r\" %b" >> > </IfModule> >> >> > <IfModule mod_rewrite.c> >> > RewriteEngine On >> > RewriteOptions inherit >> > </IfModule> >> >> > <Directory "/var/www/gatekeeper"> >> > Order allow,deny >> > Allow from all >> > #Allow from 10.17.1.0/24 >> > </Directory> >> >> > <Directory "/var/www/modwsgi/gatekeeper"> >> > Order allow,deny >> > Allow from all >> > #Allow from 10.17.1.0/24 >> > </Directory> >> >> > </VirtualHost> >> >> > </IfModule> >> > </IfDefine> >> >> > # ----- vhost end ----- >> >> > Thanks Carl and Graham for taking your time to look at this. I'm now >> > spending more time trying to deploy the app than building it :/ >> >> > Vitor (noisebleed) >> >> > On 21 Abr, 14:41, Carl Nobile <carl.nob...@gmail.com> wrote: >> >> Vitor, >> >> >> It surely seems the problem is related to not finding a URL correctly. >> >> What does your apache config look like? You should only have a few >> >> lines as mod_wsgi doesn't really need too much in the way of >> >> configuration. >> >> >> ~Carl >> >> >> On Wed, Apr 21, 2010 at 9:08 AM, noisebleed <vitorbrandao...@gmail.com> >> >> wrote: >> >> > Hi again. >> >> >> > First, thanks for the insight about these deployment issues. >> >> >> > About your suggestion now Apache is throwing 404 Not Found instead of >> >> > 500 Internal Server Error and I see something like this in the log: >> >> > [error] 14:28:12,102 DEBUG [tg.controllers] No controller-wide >> >> > authorization at / >> >> > [error] 14:28:12,105 DEBUG [pylons.controllers.core] Merging >> >> > pylons.response headers into start_response call, status: 404 Not >> >> > Found >> >> > [error] 14:28:12,108 DEBUG [txn.-1333920912] commit >> >> > [error] 14:28:12,109 DEBUG [tg.configuration] Removing DBSession from >> >> > current thread >> >> >> > Any ideas? >> >> >> > Vitor (noisebleed) >> >> >> > On 21 Abr, 13:45, Graham Dumpleton <graham.dumple...@gmail.com> wrote: >> >> >> On 21 April 2010 22:34, noisebleed <vitorbrandao...@gmail.com> wrote: >> >> >> >> > Hi Graham. >> >> >> >> > I'm not sure what should I post about your "URL mapping" question. >> >> >> > I've searched the application and found nothing really useful. I do >> >> >> > know that TG2 is based on Pylons so I guess it uses the Pylons way. >> >> >> >> > Don't know if it helps but the same app is working with "paster >> >> >> > serve" >> >> >> > on the personal machine. To test it I'm usinghttp://localhost:8080. >> >> >> >> > Sorry for my lack of knowledge but I'm fairly new to web development >> >> >> > in Python. >> >> >> >> It isn't your fault. The problem is that 'paster serve' does >> >> >> additional setup of the process environment and application beyond >> >> >> what 'loadapp' by itself does. Given that recipes for using an >> >> >> alternate WSGI server often only talk about loadapp and not what else >> >> >> has to be done, you cant replicate the setup in the same way and this >> >> >> is why you get differences like you are encountering between paste >> >> >> server and other hosting methods. Same sort of problem occurs in other >> >> >> frameworks as well when trying to move from their builtin development >> >> >> servers to other hosting mechanism. It all really sucks and the base >> >> >> framework developers seem to not usually care much about making it >> >> >> easier by providing single object entry points that do everything that >> >> >> the internal web servers do. >> >> >> >> That said, TG setup instructions I have seen have been reasonably good >> >> >> at capturing these extra hidden steps. The question is why one of >> >> >> these steps they note, which is the app.get() isn't working. >> >> >> >> Can you perhaps try: >> >> >> >> app.get("") >> >> >> >> and see if it makes a difference. It may be something to do with >> >> >> trailing slash where something is handling single slash differently. >> >> >> >> Graham >> >> >> >> > If you have suggestion for debugging this issue I would be glad to >> >> >> > try >> >> >> > it. >> >> >> >> > Regards, >> >> >> > Vitor (noisebleed) >> >> >> >> > On 21 Abr, 03:29, Graham Dumpleton <graham.dumple...@gmail.com> >> >> >> > wrote: >> >> >> >> On 21 April 2010 00:42, noisebleed <vitorbrandao...@gmail.com> >> >> >> >> wrote: >> >> >> >> >> > Hi all. First message sent to modwsgi mailing list. >> >> >> >> >> > I've built a small web application with TurboGears2 and tested it >> >> >> >> > inside a virtualenv (on my personal machine) with "paster serve >> >> >> >> > development.ini". Works just fine. >> >> >> >> >> > Then I built an Egg and installed system-wide on my server. >> >> >> >> > Defined a >> >> >> >> > new Apache2 vhost and configured both Apache and the wsgi script >> >> >> >> > created by modwsgideploy. After some adjustments I'm stuck with >> >> >> >> > the >> >> >> >> > message in the email subject. When visiting my URL Apache shows >> >> >> >> > "500 >> >> >> >> > Internal Server Error". >> >> >> >> >> > Here goes an excerpt from Apache error log: >> >> >> >> >> > [info] Initial (No.1) HTTPS request received for child 1 (server >> >> >> >> > gatekeeper.localhost:443) >> >> >> >> > [info] mod_wsgi (pid=21618): Create interpreter >> >> >> >> > 'gatekeeper.localhost: >> >> >> >> > 50000|'. >> >> >> >> > [info] [client 192.168.1.103] mod_wsgi (pid=21618, >> >> >> >> > process='gatekeeper', application='gatekeeper.localhost:50000|'): >> >> >> >> > Loading WSGI script '/var/www/modwsgi/gatekeeper/ >> >> >> >> > gatekeeper.wsgi'. >> >> >> >> > [error] 15:54:44,649 DEBUG [pylons.configuration] Initializing >> >> >> >> > configuration, package: >> >> >> >> > 'gatekeeper' >> >> >> >> > [error] 15:54:44,651 DEBUG [pylons.configuration] Pushing process >> >> >> >> > configuration >> >> >> >> > [error] 15:54:44,681 DEBUG [pylons.configuration] Adding mako >> >> >> >> > engine >> >> >> >> > with alias None and {'mako.directories': >> >> >> >> > ['/usr/lib/python2.6/site- >> >> >> >> > packages/GateKeeper-0.1dev-py2.6.egg/gatekeeper/templates'], >> >> >> >> > 'myghty.component_root': [{'templates': '/usr/lib/python2.6/site- >> >> >> >> > packages/GateKeeper-0.1dev-py2.6.egg/gatekeeper/templates'}], >> >> >> >> > 'myghty.data_dir': '/var/www/modwsgi/gatekeeper/data/templates', >> >> >> >> > 'kid.encoding': 'utf-8', 'kid.assume_encoding': 'utf-8', >> >> ... >> >> mais informações » > > -- > You received this message because you are subscribed to the Google Groups > "modwsgi" group. > To post to this group, send email to modw...@googlegroups.com. > To unsubscribe from this group, send email to > modwsgi+unsubscr...@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/modwsgi?hl=en. > > -- ------------------------------------------------------------------------------- Carl J. Nobile (Software Engineer) carl.nob...@gmail.com ------------------------------------------------------------------------------- -- You received this message because you are subscribed to the Google Groups "modwsgi" group. To post to this group, send email to modw...@googlegroups.com. To unsubscribe from this group, send email to modwsgi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/modwsgi?hl=en.
