Re: [squid-users] how to use refresh_pattern correct

On 29/04/2014 2:02 a.m., tile1893 wrote:
> Hi,
>
> i'm running squid on openwrt and i want squid to cache all requests which
> are made.
> I think that this is done by defining refresh_pattern in squid config.
> But in my opinion no matter how i config them, they are always be ignored by
> squid and will never be used.
>
> for example:
> refresh_pattern www 5000 100% 10000 override-expire override-lastmod
> ignore-reload ignore-no-store ignore-must-revalidate ignore-private
> ignore-auth store-stale
>
> or:
> refresh_pattern www 1200 100% 6000 override-expire
>
> But they both dont work.
> Any idea how to configure squid that it is caching every request?! Do I have
> to enable those refresh_patterns somehow?!

FYI: Caching everything is not possible. HTTP protocol requires at least
some non-cached traffic just to operate.

Now that your expectations have been lowered ...

 *correct* usage is not to have any of the override-* or ignore-*
options at all. But correct and practical are not always the same. Use
the options if you are required to, but only then.

 There is also a very large diference betwen HTTP/1.0 caching and
HTTP/1.1 caching you need to be aware of. In HTTP/1.0 there was HIT/MISS
and very little else. In HTTP/1.1 there is also revalidation (304,
REFRESH, IMS, INM) which is caching the [large] bodies of objects while
still sending the [small] headers back and forth - giving the best of
both worlds.

So tile1893...
 what version of Squid do you have?
 how are you testing it?
 what makes you think its not caching?
 how much cache space do you have?
 what are your maximum object limits?
 what order is your cache, store and object related config options?
 what traffic rate (requests per second/minute) are you serving?
 what does redbot.org say about the URLs you are trying to cache?

(Maybe more later but that should do for starters.)

Amos
Received on Mon Apr 28 2014 - 17:15:17 MDT