> It's possible this is because a burst of requests causes Apache to spin up > child processes to handle them, but perhaps the load-test generation slows > down at some point, Apache winds up with idle processes, and closes some > down? Is that plausible?
http://httpd.apache.org/docs/2.2/mod/mpm_common.html#minsparethreads http://httpd.apache.org/docs/2.2/mod/mpm_common.html#maxsparethreads
