We're preparing for a transition to Mac OS X 10.5 Server, and I had a
purely theoretical question. We've run into Apache process limit
situations in the past, and I want to make sure we're prepared in
Leopard just in case!
Leopard's stock Apache 2 ships with MaxClients set to 1024 in /etc/
apache2/httpd.conf:
MaxClients 1024
Meaning that, at peak, there will be 1024 Apache processes serving
1024 possible clients.
That said, Leopard's launchd, which starts Apache 2, seems to have a
default process limit of 266 (hard limit of 532).
sh-3.2# launchctl limit
[...]
maxproc 266 532
I'm curious how this works out in reality. If launchctl limits Apache
to 266 (or 532) processes, how can it possibly serve 1024 clients?
Is launchctl smart enough to "extend" the hard limit on demand? Or do
these limits totally not relate to each other?
I apologize in advance if this is a super basic question, but thanks
for any pointers. :)
