ons 2006-08-23 klockan 16:49 +0100 skrev Robin Bowes: > Here's a stab at the configuration: > > http_port 192.168.26.26:80 vhost > > cache_peer 192.168.0.41 parent 80 0 no-query originserver name=cache > cache_peer_domain server1 cache.example.com > cache_peer 192.168.0.42 parent 80 0 no-query originserver name=images > cache_peer_domain server1 images.example.com Looks fine. > One other thing I'm not sure about is DNS resolution. Only DNS involved is the client browser making a DNS lookup to find the server IP. This should give the IP of your Squid (or load balancer infront). > I currently have this configuration: > > client -> LB1 -> squid farm -> LB2 -> apache farm > > LB1 & LB2 are load-balancers > > So, clients access cache.example.com which externally (i.e. public IP > address) resolves to LB1. Ok. > LB1 passes the request to a machine in the squid farm (squid01,02,03) > The squid instances peer with each other and are configured as > accelerators for the apache farm via LB2 > proxy.example.com resolves to LB2 (192.168.0.41) Ok, > LB2 passes the request on to a machine in the apache farm > (proxy01,02,03) which are configured with cache.example.com as > ServerAliases in httpd.conf. Ok. > On each of the squid machines, I'm currently using this config (IP > address different per machine): > > http_port 192.168.26.26:80 vhost > cache_peer 192.168.0.41 parent 80 0 no-query originserver > > LB2 has address 192.168.0.41 > > However, I find that this only works if cache.example.com resolves > internally to 192.168.0.41. Works here without the internal DNS. Maybe you have something in http_access relying on DNS? Regards Henrik
Attachment:
signature.asc
Description: Detta =?ISO-8859-1?Q?=E4r?= en digitalt signerad meddelandedel
