How do MaxSpareServers work in Apache?

Question

I've scoured the web but I can't find out what MaxSpareServers are in Apache MPM prefork..

The MaxSpareServers directive sets the desired maximum number of idle child server processes. An idle process is one which is not handling a request. If there are more than MaxSpareServers idle, then the parent process will kill off the excess processes.

Great, but what causes a spareserver to be created? More importantly, when does a spare server go away? I understand that minspareservers are created gradually after the server is started..

How do maxspareservers relate to maxclients?

Basically I'm at a bit of a loss on how best to configure Apache.. there's a lot of documentation out there but it isn't that clear.

Thanks, John.

Solution

If somehow there are less spare server processes than MinSpareServers, Apache will attempt to create spare server processes until they are at least MinSpareServers.

If you have load on your server, Apache will create more server processes up to MaxClients. As long as they are in use they will not be killed, but if the load drops, Apache will start killing the processes until there are at most MaxSpareServer processes left.

Comments

[John Hunt]

ok, I think I get it :)

[OZZIE]

When load drops of and stays down will it also kill processes until it reaches MinSpareServer ?

[Christopher Perrin]

I'm not sure but I think it doesn't. I believe that the lower bound is for when you set the maximum requests for a server.