On 4/12/06, Bill Jacqmein <wrjacqmein@xxxxxxxxx> wrote: > Slight Off-topic but can the same configuration be done with different > ports on the same ip? Certainly, however if it's a 1-1 connection e.g. squid answers port 80 and your accel host runs on 81, you don't need the added complexity. Just define the single http_accel_host instead of using virtual. In my scenario you configure Squid to listen on a interface and pass all requests to a redirector, which is essentially a separate program that accepts the requested URL from stdin and spits out destination of the httpd_accel_host back to squid. The reason for this is squid listens on multiple interfaces and ports. The interface the request comes in on directly determines which http_accel_host I want the request to go to. While the redirection concept I get, I need to understand: a) Is it faster to have a separate instance of squid running for each interface then invoking the redirector program? b) can squid communicate to the redirector over a socket > On 4/12/06, Sketch <sketchster@xxxxxxxxx> wrote: > > On 4/11/06, Henrik Nordstrom <henrik@xxxxxxxxxxxxxxxxxxx> wrote: > > > mån 2006-04-10 klockan 17:59 -0400 skrev Sketch: > > > > > > > Not sure what host header based vhosts are, but it's just a single site on each. > > > > Gotcha. I use IP Based hosts, so from my research thus far the following is true: > > > > * set accel host to virtual, call a redirector which is a separate program, and have it rewrite the URL. > > > > My question regarding this is will we see higher performance invoking a small perl script for every request, rather then setting up a completely separate squid instance? > > > > Has anyone else treaded on this ground? Your results? > > > > Thanks! > > >
