I'm a novice (and maybe I shouldn't speak out of turn), but I wonder why you can't simply do sstandard reverse proxying, e.g., name your proxy server "original.org" in DNS, rename your slow web server "backend.org", and do a simple accel config:
http_port 80 accel defaultsite=original.org cache_peer backend.org parent 80 0 no-query originserver name=be acl all src all cache_peer_access be allow all http_access allow all Then your users would request http://original.org/index.html, which on a miss Squid would retrieve from backend.org, but on a hit, serve up directly. No? -----Original Message----- From: AJ Weber [mailto:[email protected]] Sent: Wednesday, August 04, 2010 2:07 PM To: [email protected] Subject: EXTERNAL: [squid-users] NEWBIE Q: httpd_accel_single_host? Does anyone have any config examples, tips or FAQ about simulating the "old" (pre 2.6, at least) single-host acceleration (i.e. as was done with the directive in the subject)? I have Duane Wessels' O'Reilly book here, and am trying to build a very specific server accelerator for across a slow, WAN link, but just for a single back-end host. (Chapter 15, pg 307, if you're now following-along ;) ) It being very specific, I can get the clients to request content directly from the Squid host (instead of the original), but then need Squid to know how to replace the host-portion of the URL for cache-misses to get the content. So instead of my client requesting www.original.org/index.htm it will already have determined on the client-side that it is over a slow link and instead request www.proxy.org/index.htm . Of course, the proxy is just Squid, and it should check for a cache hit and return if possible; if no cache hit, Squid has to know to send the request to www.original.org by some configuration option. If I can explain further my specific needs, I would be happy to. Please ask! Thanks in advance for any help. -AJ
