Hi,

 When using MPM worker with more than one child process, can anyone tell me 
whether Apache will handle connections on one child until all threads are used, 
and then pass any new connections to the next child process? This is my guess 
of how it roughly happens (since it maximises efficient use of memory). The 
reason I ask is I am trying to calculate how much memory a server running 
worker would use. With around 700 connections on a server with 5000 MaxClients 
(50 child processes with 100 threads each, with 50 servers started) 'ps' 
reports around 10 child processes using 30-100M, and the remaining 40 or so 
using about 11M, so I am making a rough guess that if every thread on a server 
was used, based on the maximum memory usage seen of one child, that this 
configuration would need about 5GB

 Regards,

 -C

Reply via email to