Comment 9 for bug 839569

In addition to my earlier comment and after reading other comments, I'd like to add now that I changed MaxClients for apache2( In Pre fork MPM ) on my virtual Lucid and restarted apache2. Ran killapache but this time the load on system didn't go high, and apache2 processes were limited to 3(observed using top command).

Before setting MaxClients I ran a guideline script from http://rimuhosting.com/howto/memory.jsp to find MaxClient number that I can use on my lucid. Script showed me number 3 and was surprised why so low as the default is 150 :). Anyways looking at free physical memory may be it makes sense to use 3 on system.

So I am now certain that in my case it was unhealthy configuration of apache2 that was the cause that made fully patched Lucid system die after running against kill apache script. And, even after using 1000 forks in killapache script my Lucid continued to stay in good health.