Let me put a second thought on this scenario. I don't have to worry too much now if there is a large number of time_wait connections unless I am running out of the ip local port
range (just over 28k by default). When the vm went down, I had apache keepalive turned on (now its off). If the connections on grade passback doesn't get closed properly and assuming that students would be submitting the assignments in a hectic pace (I watched the activity log for apache yesterday night and the "POST" lines are hectic indeed) wouldn't it be possible that apache (with KeepAlive on) could fork a lot more processes in a very quick time instead of relying on a handful of connections to brightspace keeping within the "KeepAlive" spirit? That would mean a lot more established connections to brightspace instead of time_wait in the current scenario and eventually would hit the 1,100 ceiling for apache processes (which it did, I have the collectd graph). Thanks Danny.