> I am afraid that single pool of fcgi resources is not currently possible for > multiple django sites. Maybe it would be possible if someone wrote a > specific script/wrapper for Django but, I guess, it would create more risks > (think about thread safety) than benefits.
And on your config, you got each fastcgi running under a different user too ? > Single django fcgi process on OpenBSD takes 11-12 Mb of RAM when used in > prefork mode. You will have at least 2 processes (1 manager, 1 worker) for > each site. I have not tried threaded setup though. Any reason you went for processes rather than thread ? > > However, I read on a blog from Graham > >http://www.technobabble.dk/2008/aug/25/django-mod-wsgi-perfect-match/ > > > That there is a lot of shared libs between the various fcgi processes/ > > servers, so it doesn't matter much to have unused processes/servers. > > Sorry, I didn't really catch the idea :) How could libraries be shared > between processes? As shared libraries, not sure if this is your question... > I meant that maxspare=NUMBER sets maximum number of free/idle processes. If > the load grows new processes are spawned. When the load drops unneeded > processes are killed. The problems I had under load where the total number of processes/ memory went too high, say that if you have 100 processes running django, it uses too much resources, start swapping, load skyrocketing and generally a lot of not so nice things. If I had one pool, I could put a ceiling of maxchildren=90, but I don't understand how to put that upper limit with 10 sites (and 10 different fastcgi servers) : if I put for each server maxchildren=90, it could potentially go up to 900 processes, no good. But if I put a lower limit, say 9 and a maxspare of 3, I would potentially have 9 sites doing nothing and using 30 processes, while one is busy with 10 processes running, not serving the 11th request, even so it has plenty of space to be able to have more processes. Well, might not be a problem you have with django/python and I might be looking at the wrong thing, but at least in php, I found it that putting a maxchildren low enough so you drop new requests before overloading the server avoid problems, and having one fastcgi server for all the sites works better than independent ones. What is your experience with django (mod_python shared pool of processes vs. fastcgi separated ones) ? > Valts. --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "Django users" 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/django-users?hl=en -~----------~----~----~----~------~----~------~--~---
