Re: [squid-users] Minimal caching time squid honors

From: Thomas Ristic <thr@dont-contact.us>
Date: Sat, 18 Dec 2004 14:08:07 +0100

Am Sa, den 18.12.2004 schrieb Henrik Nordstrom um 11:20:
> On Fri, 17 Dec 2004, Thomas Ristic wrote:
>
> > Am Fr, den 17.12.2004 schrieb Henrik Nordstrom um 19:02:
> >
> >>> Is there a way to disable this with a configuration directive?
> >>
> >> No, only by modifying the source. See refresh.c
> >
> > Thanks for the hint! Does this make sense? Do I get this right?
>
> What you mean?

I meant if the way I implement the minimum_expiry_time configuration
option in the tiny patch I attached to my last mail looks sane to you.

> Not caching objects which expires in less than one minute and which can
> not be reused after that as it is impossible to ask the web server if the
> object is still valid makes perfect sense to me.

It makes sense, yes. But 60 seconds looks like a randomly defined
default to me. I believe it would be nice to keep this default but make
it configurable by the user.

> But in some extreme reverse-proxy/accelerator setups the demands may be
> different, but I tend to think that even then it is many magnitudes better
> to fix the web server to allow revalidation of cached objects.

I also think that this is most likely only relevant in accelerator
setups. In some environments it might be to complicated or costly to
find out the last modified time. So simply telling the accelerator to
keep this page for e.g. 30 seconds and then fetch a new one might still
be a good thing to do.

Regard

        Thomas
Received on Sat Dec 18 2004 - 06:08:10 MST

This archive was generated by hypermail pre-2.1.9 : Sat Jan 01 2005 - 12:00:02 MST