Re: refresh-parameter in a cache-hierarchy

From: Javier Puche. CSIC RedIRIS <javier.puche@dont-contact.us>
Date: Thu, 5 Feb 1998 16:15:17 +0100

Hi,

> >Is it necessary (proposed) that all configurations in a hierarchical
> >cache-topology have the same refresh_pattern rules?
>
> AFAIK, squid sends Max-age on each outgoing HTTP request, so it
should not
> be problem.
> BUT: the problem still remains if you use ICP_HIT_OBJ in your
hierarchy,
> because ICP lacks of the freshness mechanisms of HTTP.

 But it can also be the case that my cache gets an ICP_HIT (not
ICP_HIT_OBJ, we have it off) for some object and then when doing the
HTTP conection the object does not satisfy the max-age condition, and
if it is a sibling (miss_access denied) it won't fetch it for you, so
you get a "Sorry, you are not currently allowed to request:..." page.
That is what was happening between another cache and mine one and that
is the hypothesis we made, so we are synchronizing refresh_patterns to
check it.

 Correct me if I am wrong, please.

Regards,

Javier Puche.
Received on Thu Feb 05 1998 - 09:19:40 MST

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:38:47 MST