ThreadsPerChild default in mpm_winnt

[Joshua Slive <jo...@slive.ca>]

I was just trying to correct the docs for ThreadLimit and 
ThreadsPerChild to have the correct defaults for mpm_winnt, and I ran 
into a little confusion.  From a very quick reading, it appears that the 
mpm_default.h for  mpm_winnt is stolen directly from an MPM that has 
min/maxsparechild maintenance.  As far as I know, mpm_winnt has a fixed 
number of threads, so most of the defines do absolutely nothing.

It also appears that the default for ThreadsPerChild is set at 
DEFAULT_START_THREAD, which is in turn set at 5.  This may be a 
reasonable number for worker, where more processes can be forked to deal 
with load.  But it doesn't seem to make any sense for mpm_winnt.

Am I out to lunch here, or could mpm_winnt/mpm_default.h use an overhaul?

Joshua.

RE: ThreadsPerChild default in mpm_winnt

[Bill Stoddard <bi...@wstoddard.com>]

Check out the commit I just made.  Not complete but better.. 

Bill

> I was just trying to correct the docs for ThreadLimit and 
> ThreadsPerChild to have the correct defaults for mpm_winnt, and I ran 
> into a little confusion.  From a very quick reading, it appears that the 
> mpm_default.h for  mpm_winnt is stolen directly from an MPM that has 
> min/maxsparechild maintenance.  As far as I know, mpm_winnt has a fixed 
> number of threads, so most of the defines do absolutely nothing.
> 
> It also appears that the default for ThreadsPerChild is set at 
> DEFAULT_START_THREAD, which is in turn set at 5.  This may be a 
> reasonable number for worker, where more processes can be forked to deal 
> with load.  But it doesn't seem to make any sense for mpm_winnt.
> 
> Am I out to lunch here, or could mpm_winnt/mpm_default.h use an overhaul?
> 
> Joshua.
>