Re: [squid-users] acceleration mode question

From: Matthew Shoemaker <matthew.shoemaker@dont-contact.us>
Date: Tue, 8 Aug 2006 12:53:40 -0700

Hmm interesting, no I did not build squid with that option, I didn't
know about it. And along the same lines I didn't know about the
X-Accelerator-Vary header either so I am pretty sure its not being
used. I will investigate these options now. Thanks very much.

Matthew

On 8/8/06, Robert Borkowski <rborkows@novator.com> wrote:
>
> Is your squid built with the --enable-x-accelerator-vary option? If so, do
> your webservers set the X-Accelerator-Vary header?
>
> This feature is used to store different copies of an object based on request
> header. For example, our webservers reply with:
> X-Accelerator-Vary: Accept-Encoding
>
> And squid stores the resulting object based on what type of encoding the
> client supports. Maybe you're using the above feature and the two browsers
> set the header you are varying on slightly differently.
> For example, just looking through one of my own squids I see that some
> browsers set accept-encoding to "gzip, deflate", while some set it to
> "gzip,deflate" (no space), and this results in two different objects
> for the same URL
>
>
>
> --
> Robert Borkowski
>
>
>
>
>
> On 8-Aug-06, at 1:55 PM, Matthew Shoemaker wrote:
>
> Ok, here is an update and I wonder if anyone could help me understand
> this. I tracked the problem I was having down to my use of
> mod_deflate with apache. If I have squid trying to cache the
> compressed content from the web server I get what looks like a cached
> copy of each page for each browser version that accesses it. If I
> turn compression off I get one cached page for all users with all
> browsers. I then set an apache proxy with compression on my squid
> server on port 80, it sends all requests right to squid on port 8080,
> and squid then grabs the requests on a miss from the web server. By
> doing this, squid again works as expected by feeding one cached copy
> to all users with various versions of browsers. The downside to this
> is only that the cache is now populated with the full size of the
> webpage instead of its compressed version. Not that big of a deal,
> but ideally if possible I would like to have the compressed pages
> cached, and the same page fed to all users. Any suggestions??
>
> Thanks very much for any info.
> Matthew
>
> On 8/7/06, Henrik Nordstrom <henrik@henriknordstrom.net> wrote:
> mån 2006-08-07 klockan 14:34 -0700 skrev Matthew Shoemaker:
> > I have squid 2.6stable1 up and running on suse 10.1 in acceleration
> > mode. It works fine under normal testing on my pc.. But if I open up
> > say internet explorer instead of firefox and hit the same address on
> > the cache server I get a cache miss in the access log.
>
> Many browsers forces a fresh copy on the first request. Make sure you
> navigate to the intended page via a link. Avoid bookmarks and reload
> buttons.. Also remember to clear your browser cache on at least one of
> the test stations (the first to request the test page).
>
> Regards
> Henrik
>
>
>
>
>
>
>
>
Received on Tue Aug 08 2006 - 13:53:43 MDT

This archive was generated by hypermail pre-2.1.9 : Fri Sep 01 2006 - 12:00:02 MDT