I had a terrible moment with my old server when disk system halted to crawl (second time this week) and in couple of minutes I have load over 1200. I had limited the lsphp processes to 1000 and connections to 500 but LS just accumulated processes as people hit refresh. I had all php processes hanged. I think the problem was this: when PHP asks MySQL to deliver, and MySQL almost dies, the processes are not released/killed after set time limit as they should be. Yesterday I had above server canceled, and it and my new server shared NFS connection (old server was the NFS server). As I did not remember to kill the NFS mount in new server, the NFS connection was left hanging when old server was closed. This meant a php stats page which asked disk free with php system() command hanged (as the NFS was dead). A new page load created other process and soon I have 20 hanged lsphp processes. Interestingly enough, those processes were hanging there next day, until I killed each manually. Why did not LSWS kill them after specified timeout? A long story short: please make LSWS kill lslphp processes that go over configured time limits. Really kill them. Without need to stop and start LSWS manually (or then: add real stop/start commands in Admin). Else setting configurations do not make much sense.