On ons, 2007-09-05 at 12:37 -0700, Neil Harkins wrote:
> On 9/5/07, Oliver Schoett <os@sdm.de> 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?
There is an act-as-origin http_port option in Squid-2.HEAD which solves
most of this...
> 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...
The problem that Squid do not refresh the stored headers when making an
IMS, only updating it's internal freshness of the object.
Regards
Henrik
This archive was generated by hypermail pre-2.1.9 : Mon Oct 01 2007 - 12:00:02 MDT