You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Jim Jagielski <ji...@jaguNET.com> on 2000/11/17 21:50:41 UTC
Reaping of idle Apache child processes
In both 1.3 and 2.0, when we kill off the idle child process
in perform_idle_server_maintenance(), we simply choose the
highest numbered idle process. I think it would make sense to
choose the highest numbered and highest cpu used idle process
(basically, the highest numbered and most "used" process).
Might be a good way of handling memory leaks as well as
possibly preventing some minor thrashing in killing a recently
created, under-used process.
Comments?
--
===========================================================================
Jim Jagielski [|] jim@jaguNET.com [|] http://www.jaguNET.com/
"Casanova will have many weapons; To beat him you will
have to have more than forks and flatulence."
Re: Reaping of idle Apache child processes
Posted by Greg Stein <gs...@lyra.org>.
On Fri, Nov 17, 2000 at 03:50:41PM -0500, Jim Jagielski wrote:
> In both 1.3 and 2.0, when we kill off the idle child process
> in perform_idle_server_maintenance(), we simply choose the
> highest numbered idle process. I think it would make sense to
> choose the highest numbered and highest cpu used idle process
> (basically, the highest numbered and most "used" process).
> Might be a good way of handling memory leaks as well as
> possibly preventing some minor thrashing in killing a recently
> created, under-used process.
>
> Comments?
If you can get that information, then yah... makes total sense.
Cheers,
-g
--
Greg Stein, http://www.lyra.org/
Re: Reaping of idle Apache child processes
Posted by "Roy T. Fielding" <fi...@ebuilt.com>.
> In both 1.3 and 2.0, when we kill off the idle child process
> in perform_idle_server_maintenance(), we simply choose the
> highest numbered idle process. I think it would make sense to
> choose the highest numbered and highest cpu used idle process
> (basically, the highest numbered and most "used" process).
> Might be a good way of handling memory leaks as well as
> possibly preventing some minor thrashing in killing a recently
> created, under-used process.
OTOH, it might result in killing the process most likely to have
its innards within the fast L2 or VM cache.
*shrug*
Test it first, please.
....Roy