2009/2/10 Ariel Mauricio Nunez Gomez <[email protected]>:
> I have a server configured like this:
>
> <VirtualHost *>
> ServerName xx.com
> DocumentRoot /home/xx/src
>
> WSGIDaemonProcess xx user=xx group=xx threads=1 processes=25
> python-path=/home/xx/lib/python2.5/site-packages
> WSGIProcessGroup xx
Hmmm, you have come from the pinax group have you?
I already warned in that group that 25 processes was a rather rediculous number.
> WSGIScriptAlias / /home/xx/deploy/bme.wsgi
> ErrorLog /var/log/apache2/error.xx.log
> LogLevel warn
> CustomLog /var/log/apache2/access.xx.log common
> ServerSignature On
> </VirtualHost>
>
> When the server is restarted it starts painfully slow until many =~ 10
> requests are made, after that speed comes to what's expected.
>
> Any ideas on how to troubleshoot that? I tested on two different servers
> (Debian Etch on a Dell Sever, and Ubuntu Intrepid on an EC2 instance) and
> experienced the same issue.
You see what you see because application is by default lazy loaded the
first time request arrives which targets it.
Things are much worse than they need to be in your case because you
have so many processes needing to load it, so first request that hits
each process will be slow.
Cut down the number of processes you need. You shouldn't need that
many even if each is single threaded, unless you have lots of long
running requests occuring.
You then want to preload the WSGI script file, so at least Python
modules loaded and ready. Django still lazy initialises itself, but
haven't worked out recipe yet for people to use to force it to early
initialise.
Thus use configuration:
WSGIImportScript /home/xx/deploy/bme.wsgi process-group=xx
application-group=%{GLOBAL}
<VirtualHost *>
ServerName xx.com
DocumentRoot /home/xx/src
WSGIDaemonProcess xx user=xx group=xx threads=1 processes=25
python-path=/home/xx/lib/python2.5/site-packages
WSGIProcessGroup xx
WSGIApplicationGroup %{GLOBAL}
WSGIScriptAlias / /home/xx/deploy/bme.wsgi
ErrorLog /var/log/apache2/error.xx.log
LogLevel warn
CustomLog /var/log/apache2/access.xx.log common
ServerSignature On
</VirtualHost>
The two changes are to add WSGIApplicationGroup directive so know
which interpreter is being used, and to add WSGIImportScript to force
script to be loaded into that process/interpreter automatically at
startup of process and not later only when request first arrives.
Right now it is a pain than WSGIImportScript has to be outside of
VirtualHost. This is fixed for mod_wsgi 3.0 and can be set more
appropriately inside of VirtualHost, athough must come after the
WSGIDaemonProcess group directive. This will actually be enforced and
so when move to mod_wsgi 3.0 in future, configuration file will need
to be updated. For this latter issue see:
http://code.google.com/p/modwsgi/issues/detail?id=110
Graham
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"modwsgi" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/modwsgi?hl=en
-~----------~----~----~----~------~----~------~--~---
