https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=105725

The number of connections to make to the POD on a graceful restart
should be equal to the number of children which have ever been
started; not the number of children there *could* be.

Otherwise the parent is bound to get connect() timeouts. 

--- httpd-2.0.46/server/mpm/prefork/prefork.c.graceful
+++ httpd-2.0.46/server/mpm/prefork/prefork.c
@@ -1167,7 +1167,7 @@
 		    "Graceful restart requested, doing restart");
 
 	/* kill off the idle ones */
-        ap_mpm_pod_killpg(pod, ap_daemons_limit);
+        ap_mpm_pod_killpg(pod, ap_max_daemons_limit);
 
 	/* This is mostly for debugging... so that we know what is still
 	    * gracefully dealing with existing request.  This will break