With PHP-FPM it's possible to have a single master process but the workers 
(each with it's own uid/gid and unix socket) don't start until there is a 
request, and after inactivity on those workers they are shutdown.So if I have 
500 clients, and 400 of those barely get any traffic, my memory is not being 
used with 400 processes that barely get used.
Is the same possible for uWSGI? It looks to me that if I use Emperor mode and 
have 100 vassles (each with it's own uid/gid and unix socket) then each of 
those 100 vassles will have it's own master process and those 100 processes 
ALWAYS running and consuming memory, when all I want is a single master process 
that will start up processes on demand.                                      
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi

Reply via email to