On Thu, 10 Nov 2005, Pakozdi Tibor wrote:
Let's suppose that the page generation lasts 5 seconds. At first there is no cache at the Squid reverse proxy, then come some HTTP GET requests in less than 5 seconds. Now at my configuration all of the HTTP GET requests went to the web server, individually making page generations at the web server. The required behaviour would be that only the first HTTP GET goes to the web server and the others are waiting at the Squid reverse proxy for the cache to be generated, and from the cache those requests could be served.
This is not possible out-of-the-box, but look for collapsed_forwarding at devel.squid-cache.org.
Regards Henrik
