Hi Graham, Just following up on this thread. I replied with some details, but maybe you missed those, or maybe i missed your reply. Or maybe this isn't worth it and i should stop being lazy and just write out the VirtualHost files :) Ben
On 23 Dec 2008, at 01:13, Graham Dumpleton wrote: > > Have some further questions about what you want to do. > > Do you want a separate daemon process for each distinct Django site, > or are you happy with one really fat process which contains each > Django site in a separate sub interpreter of that process? > > How much memory does each Django site instance take up? > > How many different site instances would you have? > > Are all the site instances distinguishable by server name alone? > > Graham > > On Dec 23, 9:00 am, Graham Dumpleton <graham.dumple...@gmail.com> > wrote: >> On Dec 22, 9:44 pm, Ben Eliott <ben.dja...@googlemail.com> wrote: >> >>> Hi Graham, >>> I've finally managed to get back to the wildcard subdomains & >>> mod_wsgi >>> today. Unfortunately the discussion thread you mentioned has >>> disappeared and after a few hours i still seem to be doing a good >>> job >>> of getting nowhere. >> >> I can still access thread with no problems. >> >>> Although you mentioned using mod_rewrite to get hold of the url >>> variable, it looks like the %{SERVER} variable in mod_wsgi might >>> take >>> care of this already? >> >>> My main issue seems to be to accessing the %{SERVER} (or relevant >>> mod_rewrite) variable in the .wsgi script, to specify a particular >>> settings file. >> >>> Within a VirtualHost i have: >>> WSGIApplicationGroup %{SERVER} >>> WSGIDaemonProcess %{SERVER} ...threads etc >>> WSGIProcessGroup %{SERVER} >> >> The %{SERVER} value is only magic when used with WSGIApplicationGroup >> directive. >> >>> So this is probably hoplessly wrong also, but if you can give some >>> further pointers that would be most kind. >> >> Can you post a more complete example of your Apache configuration >> showing what you are trying to achieve. >> >> Sorry I didn't get back to you last time, it was a hectic few weeks. >> Things have settled down somewhat now, so I'll go back over your >> original post and work out again what it is you were trying to do. >> >> Graham >> >>> Thanks and Regards, >>> Ben >> >>> On 9 Dec 2008, at 10:18, Graham Dumpleton wrote: >> >>>> On Dec 9, 8:05 pm, Ben Eliott <ben.dja...@googlemail.com> wrote: >>>>> Graham, >>>>> Thank you for coming back personally to such a lowly wsgi >>>>> question! I >>>>> started reading your email and thinking the answer was 'no', then >>>>> ended up thinking 'definitely maybe'. I'll keep an eye out in case >>>>> you >>>>> post more, otherwise i'll follow those links and your directions >>>>> and >>>>> hope to report back with some progress. >> >>>> I'll definitely try and say more later when get a chance. >> >>>> Just do be aware of one thing. By using a single WSGI script file >>>> for >>>> multiple sites, you loose the ability with mod_wsgi daemon mode to >>>> touch the WSGI script file and cause a single site to be >>>> reloaded. One >>>> would normally use this as a way of reloading a single site without >>>> the need to restart the whole of Apache. When sharing the single >>>> WSGI >>>> script file across sites, touching the WSGI script file will >>>> restart >>>> all sites using that WSGI script file. If they share the code >>>> this may >>>> actually be want you want, so not a problem, but worth mentioning. >> >>>> In this arrangement, if you did want to reload one site, for >>>> example >>>> because you change its settings file, you would need to use 'ps' to >>>> identify process(es) in that daemon process group, based on what >>>> display-name option was set to, and send all those processes in >>>> that >>>> daemon process group a SIGINT using the 'kill' command. >>>> Alternatively, >>>> you would need to setup a background thread which monitored >>>> something >>>> like the distinct settings file for each site and have the process >>>> itself send a SIGINT to itself. This would be a variation on >>>> background reloader described in: >> >>>> http://code.google.com/p/modwsgi/wiki/ReloadingSourceCode#Restarting_ >>>> ... >> >>>> More later. >> >>>> Graham >> >>>>> Thanks and Regards, >>>>> Ben >> >>>>> On 9 Dec 2008, at 08:23, Graham Dumpleton wrote: >> >>>>>> On Dec 9, 6:53 pm, "ben.dja...@googlemail.com" >>>>>> <ben.dja...@googlemail.com> wrote: >>>>>>> Hi, I'm converting to the excellent mod_wsgi and wondering if >>>>>>> it's >>>>>>> possible to make a single httpd virtual host/wsgi file to manage >>>>>>> wildcard subdomains. >> >>>>>>> Basically I have an app where i'm creating a new instance for >>>>>>> each >>>>>>> client and using subdomains. So client1.example.com and >>>>>>> client2.example.com both point to the same app, but their own >>>>>>> settings.py/django instance. >> >>>>>>> So far so fine. I've been happily converting to mod_wsgi >>>>>>> daemons, >>>>>>> creating virtual hosts and independent .wsgi files for each one. >>>>>>> But >>>>>>> now just wondering whether there is some way i can make this >>>>>>> process >>>>>>> dynamic so one virtual host/.wsgi file will take care of all >>>>>>> these >>>>>>> subdomains. >> >>>>>>> I see the advice on the wsgi wiki to push domain sub- >>>>>>> directories to >>>>>>> different django instances, but i'd rather keep using the >>>>>>> subdomains >>>>>>> if possible. >> >>>>>>> It looks possible to be able to parse information about the >>>>>>> incoming >>>>>>> request in the wsgi file and push it to different settings. >>>>>>> But i'm >>>>>>> not sure what this will do in terms of spawning processes etc, >>>>>>> it >>>>>>> looks a little dangerous, or maybe this will work. Any advice >>>>>>> appreciated thanks! >> >>>>>> Start by reading recent discussion: >> >>>>>> http://groups.google.com/group/django-users/browse_frm/thread/dfd3521 >>>>>> ... >> >>>>>> I'll post more tomorrow if have time, have to do some things >>>>>> tonight >>>>>> and then out most of the day tomorrow. >> >>>>>> In short though, no support for dynamic transient daemon >>>>>> processes >>>>>> yet, ie.,: >> >>>>>> http://code.google.com/p/modwsgi/issues/detail?id=22 >> >>>>>> so, can't get away from using WSGIDaemonProcess for each >>>>>> instance at >>>>>> the moment. >> >>>>>> One can use dynamic setting of WSGIApplicationGroup via a >>>>>> variable >>>>>> set >>>>>> by mod_rewrite to select daemon process as well as set some name >>>>>> relevant to settings file. WSGI application wrapper can then be >>>>>> used >>>>>> to override DJANGO_SETTINGS_MODULE. >> >>>>>> So, information is in that post, you just need to adapt it to >>>>>> your >>>>>> situation. That is, use SERVER_NAME rather than REMOTE_USER from >>>>>> authentication as basis of selecting daemon process group. You >>>>>> could >>>>>> though skip the rewrite maps that allowed multiple levels of >>>>>> indirection and made it further dynamic in nature. >> >>>>>> Graham > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" group. To post to this group, send email to django-users@googlegroups.com To unsubscribe from this group, send email to django-users+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/django-users?hl=en -~----------~----~----~----~------~----~------~--~---