On Fri, Feb 01, 2008 at 09:30:10AM +1300, Amos Jeffries wrote: > > On Thu, Jan 31, 2008, Amos Jeffries wrote: > >> If the delay between starting and stopping is long enough and can be > >> done with X + random-time offset. Squid will cope with some helpers > >> simply stopping and resumes them. > > > > Yeah, the trouble is that you can't kill them all or Squid will die > > complaining the helpers are dying too quickly. > > That would be why I added the 'random-time' offfset. It would need to be > some random time per helper so they die spread over a period and squid has > time to restart the earliest before the last go. I tried killing the rewriters one at a time, but I didn't find this to be reliable. Squid doesn't appear to do round-robin, it looks like it hammers the first child and only queries the next one when all the previous ones are busy. I've found that it doesn't notice the rewriters dying one at a time, but tends to notice a block of them disappearing all at once. --Emil