On 23/11/2013 3:59 a.m., davidheijkamp wrote: > Hi, > > We're currently designing our web hosting architecture and have some > difficulty figuring out the best way to use Squid as reverse proxy. <nip> > As was explained on this list, in the case of single reverse proxy with two > cache_peer entries with different origin servers, “Squid detects [when] the > server is down and stops using it until it becomes alive again”[1] > > My question is if this would be the case as well when, for example Apache_2 > goes down? Is it possible to configure the Squid servers in such a way that > Squid_0 stops sending requests to Squid_2 that are normally handled by > Apache_2? Yes the detection is done partially using HTTP request status codes. So it does not matter whether the origin is being contacted directly or through another proxy unless that proxy is supplying stale content when the origin is down. You will possibly need to disable ICP queries to the other proxies to prevent false ALIVE being detected. Amos
