On 9/5/07, Oliver Schoett <os@xxxxxx> wrote: > Did you notice any ill effects of Bug 7 > (http://www.squid-cache.org/bugs/show_bug.cgi?id=7)? This bug makes > Squid deliver resources with expiration times in the past, thus causing > the clients to revalidate the resources every time they are used. Ugh, first I've heard of this bug. So, in an httpd-"accel" setup, we're wasting more bandwidth than if we didn't use squid at all!? Can anything be done in the config to mitigate? Is the STALE state always refreshed by a IMS request? Is there a way to force a purge/re-get instead of an IMS? i.e. It'd be nice if lm-factor > percent generated an IMS, but age >max resulted in a purge and re-GET... If not, perhaps I'll write a tool to tail the log for the hottest objects, look at the headers on disk, and issue PURGE requests. Yuck. -neil
