Re: modify time from squid (not browsed server)

Miroslaw Gawenda wrote:

> I think so this is not problem with browser.

Did not say so. It is in the protocol.

> To say more , I was think so this is but I was loose one day for
> testing and I found so My browser sends good commands but SQUID
> Cache algorithm is not good for all serwers.

True, not all servers behave correctly.

The issue with the browsers is how it behaves when you push that reload
button. When the browser is talking to a proxy it knows that there may
be a cache in between it and the origin server, and thus instructs the
proxy to ignore any cached replies.

> I think so if the SQUID can remember date and time returned from httpd
> than he can correct this problem with different algorithm for calculating
> modified date of object.
> I was suggest this in the title of my question.

Please elaborate a bit on this.

> please try (if you have enough time) this two commands :
> 1.client -m HEAD http://www.linux.org/
> 2.client -m HEAD http://www.linux.org/index.html
>
> You get two different expires date !!! but this is of course the same page.

Only if there is a cache in between (in this case Squid), one of the
pages is cached and considered fresh, and the page is modified after it
was first cached.

If this worries you a lot then use a refresh_pattern of
refresh_pattern 0 0% 0

this will cause Squid to revalidate cached objects on every request
unless the origin server provided an explicit expiry time using Expires
and/or Cache-Control: max-age.

--
Henrik Nordstrom
Squid hacker