Re: [squid-users] Re: Squid config for only caching files from certain domain

From: Amos Jeffries <>
Date: Thu, 23 Dec 2010 13:09:26 +1300

On 23/12/10 07:32, Michael Cole wrote:
> Hi,
> I'd like to cache files locally, but only from a small list of
> domains. This setup might be this:
> I haven't found a way to cache only certain requests (e.g.
>*), but not cache any other traffic (web
> development isn't improved with caching).

   acl someDomains dstdomain ...
   cache deny !someDomains

> How does squid know when the cached version has expired? Some requests are:
> -
> - this file is always the same data
> -
> - this file might change daily
> Would this work? Is this how Squid should be used? Is there a better way?

> How does squid know when the cached version has expired?

  The HTTP server sends Cache-Control and Expiry information about how
old the object is and when it expires.

  In this case it is probably your Squid contacting the FTP server via
FTP protocol so things are a little different...

> Would this work?

  An FTP server which supports the MTDM feature will supply Squid with
the timestamp to use as Last-Modified in HTTP.

The rest (expiry) has to be estimated. This is done by refresh_pattern
the same as if there was a web server which only provided Last-Modified.

> Is this how Squid should be used?

You will see in the default "refresh_pattern -i ftp://.* "...

This sets the min, percentage-of-past-age and max times to store the
objects. Default is to store for the same length of time as the object
was old when fetched, but no less than a day and no more than a week.

You can set our own regex patterns above the default to store the
matching URLs for longer or shorter periods as desired.


Please be using
   Current Stable Squid 2.7.STABLE9 or 3.1.9
   Beta testers wanted for
Received on Thu Dec 23 2010 - 00:09:39 MST

This archive was generated by hypermail 2.2.0 : Thu Dec 23 2010 - 12:00:04 MST