Thank you Amos for the always kind answer. Sorry my actual question, when client issue a "reload" request, does this request passed by squid2 and squid1 then to the original server? Thanks. On Fri, Dec 16, 2011 at 2:08 PM, Amos Jeffries <squid3@xxxxxxxxxxxxx> wrote: > On 16/12/2011 4:36 p.m., Jeff Pang wrote: >> >> Hello, >> >> if the site-arch is: >> >> original-server -> squid1 -> squid2 -> client > > > NOTE: this is a *reply* pathway. IMS is a *request*. > > >> >> When client send an IMS request to squid2, will squid2 pass it to >> squid1 then to original-server? >> >> Thanks. > > > The IMS effectively stops at squid2. Whether squid2 does a new IMS for > itself or a full fetch from squid1 depends on whether squid2 already has a > copy of that object, whether that copy is fresh or stale or newer than the > IMS date. > > All those considerations are calculated, then a request is either satisfied > by squid2 or a request created from a combination of the squid2 and client > requirements is passed to squid1. squid1 then does its own handling on the > request sent by squid2. > > Amos
