[00:56] < entil> mysql is crunching cpu constantly, yes? want timeslices, want many
[00:56] < entil> apache is forking and dying at the same time
[00:57] < entil> so we have a lot of processes
[00:57] < entil> that don't die on time because the kernel ticker misses the deaths
[00:57] < entil> so the kernel sees a fuckload of tasks wanting cpu time
[00:57] < entil> and increases the load number
[00:57] < entil> (although con's "I concur" would be nice ;)