Henrik, Thanks for your response!
Based on that, I would tell that the best way to accomplish controlling the size of the logs may be just cron a script that stats the last log file and, if the size is greater than the limit we want, issue a "squid -k rotate". Also, I guess it would be great not to change the cache_mem option in order to avoid taking too much time for writing the clean swap.state file. -- Bye, Fernando Maciel Souto Maior [EMAIL PROTECTED] http://www.araujo.com.br +55+31 3270-5886 LPIC-1!31908 > On Fri, 24 Oct 2003, Fernando Maior wrote: > >> 1) renames files acording to some standard > > Yes. > >> 2) stops listening to the incoming traffic > > No. > >> 3) closes the old file > > Yes. > >> 4) creates the new log file > > Yes. > > And also restarts any helper processes. > >> 5) starts listening to the incoming traffic > > No. > >> And how many time is spent in that process? > > Very little. > >> The size of the log file or memory cache >> does mean a difference in time spent on >> log rotating? > > No. > > But as it also writes out a clean swap.state the larger your disk cache is > the longer it may take before performance is normal (well.. a few seconds > longer) > > Regards > Henrik > >
