Re: [squid-users] Squid & cache killers

 On Tue, 6 Dec 2011 20:56:11 +0200, karj wrote:
> Hi,
> I have come across a problem with a flash application that's using
> cache
> killer.
>
> For example all the request URL's ends with
> liveofday.xml_nocache=1322784085349
> liveofday.xml_nocache=1322784100431
>

 Sigh. This kind of URL is the direct end product of admin configuring
 "private", and "no-store" objects to be force-cached all over the
 network, along with ignoring "no-cache" and expiry headers. Corrupting
 the information itself in the name of speed or cost.

 This should be a huge sign that the website authors *do* know a bit
 about caching and that people absolutely should not cache this content,
 yes?

 /end rant

 Before you go any further run those URLs through the tool at redbot.org
 and see how knowledgeable the authors actually are about caching.

 NP: a file called "liveofday" implies that the content is constantly
 changing. Particularly on an XML data feed, caching it is probably NOT a
 good idea.

> Is there a way for squid to cache all the xml files ignoring the
> nocache
> numerical value.

 The fact that you call them files (with an 's') indicates that you know
 full well that they are different content. Why pretend that one file is
 exactly binary-equal to all the others that will *ever* exist at that
 URL?

> I don't want to use an external program such us a perl re-writer.

 No there is no way to do it (yet) without the external helper program.
 In fact, no way to do it without the Squid-2.7 storeurl_rewrite
 feature.

 Amos
Received on Tue Dec 06 2011 - 23:18:35 MST