Linda W <mailto:squid-user@xxxxxxxxx> wrote on 17 June 2005 05:14 AM: > I rebuilt squid with delay pools built-in. I added an ACL for > my specific host, even for my entire subnet. I reduced the > aggregate and specific IP to 5/5 & 1/1 and noticed no delay on > loading web pages. DSL Reports measured over 1Mb download and > over 600Kb upload. Do delay pools actually work? Relevant > params: > > First tried it with 2 pools, but in trying to simplify went > down to 1 pool with my entire subnet included: > delay_pools 1 > delay_class 1 2 > acl test_subnet src 192.168.3.0/255.255.255.0 delay_access 1 > allow test_subnet delay_access 1 allow all > delay_parameters 1 5/5 1/1 # have tried ranges from 8000/8000 > ->1/1 > delay_initial_bucket_level 50 > > --- > Near as I can tell this should slow me down to about 1 > byte/second, but am not seeing any noticable delay. > > I did have 2 pools, with my test_subnet, in the restricted > pool, but when that failed, tried to simplify to 1 pool and an > absurdly low limit...but seems to have no effect. I also had a > host-specific acl, and noted in the cache log that my host > matched the host-specific acl -- but that made no difference > > Am I missing something obvious? There is no direct or routed > access > from the internal subnet to the outside. Tracing and debug > output > from the squid cache show that the client (my machine in this > case) is going through the cache....but the limits don't seem > to be enforced very well.... > > Help? > Have you tried checking up on the pools in the cachemgr.cgi? It can indicate to you if the clients are falling into the pools or not. HTH, Armin
