Caching same object twice for different browsers

[Lars Jørgensen]

Hi,

Sorry for top-quoting, but stuck with Outlook and fixed settings here at the office.

Anyway, I found out that the web server does not use a Vary-header, so that wasn't the problem. After doing some manual tests and being unable the replicate the problem, I started to look harder at the two dumps below. Then it became pretty obvious.

It seems that Explorer and Firefox requests urls in a different way.

Explorer: /erez4/cache/online_master_arkiv_webbilleder_DIA_
Firefox: /erez4/cache/online%5Fmaster%5Farkiv%5Fwebbilleder%5FDIA

Of course that means diffent objects in the cache since varnish hashes on req.url. Is there a know workaround to this problem?


--
Lars

From: varnish-misc-bounces at varnish-cache.org [mailto:varnish-misc-bounces at varnish-cache.org] On Behalf Of Lars Jørgensen
Sent: Wednesday, July 13, 2011 4:09 PM
To: varnish-misc at varnish-cache.org
Subject: Caching same object twice for different browsers

Hi,

I believe this has been touched upon before, but I can't seem to find anything useful about it. I have discovered that our Varnish caches the same object twice; once for MSIE and once for Firefox. This is from varnishlog:

First Internet Explorer:

  20 RxRequest    c GET
   20 RxURL        c /erez4/cache/online_master_arkiv_webbilleder_DIA_kalender_Koncerter_diaens_mortenfranklarsen_tif_331510ad02b81dd2.jpg
   20 RxProtocol   c HTTP/1.1
   20 RxHeader     c Accept: text/html, application/xhtml+xml, */*
   20 RxHeader     c Accept-Language: da-DK
   20 RxHeader     c User-Agent: Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0)
   20 RxHeader     c Accept-Encoding: gzip, deflate
   20 RxHeader     c Host: www.kb.dk
   20 RxHeader     c Connection: Keep-Alive
   20 RxHeader     c Cache-Control: no-cache
   20 RxHeader     c Cookie: __utma=52637486.1667608039.1302159744.xxxxxxxxxx.1310564597.16; __utmz=xxxxxxxx.1307533937.4.2.utmccn=(referral)|utmcsr=intranet.kb.dk|utmcct=/ATO/Sider/Genveje.aspx|utmcmd=referral; __utmb=xxxxxx
   20 VCL_call     c recv lookup
   20 VCL_call     c hash
   20 Hash         c /erez4/cache/online_master_arkiv_webbilleder_DIA_kalender_Koncerter_diaens_mortenfranklarsen_tif_331510ad02b81dd2.jpg
   20 Hash         c www.kb.dk
   20 VCL_return   c hash
   20 Hit          c 1414499195
   20 VCL_call     c hit deliver
   20 VCL_call     c deliver deliver

And then Firefox:

   12 RxRequest    c GET
   12 RxURL        c /erez4/cache/online%5Fmaster%5Farkiv%5Fwebbilleder%5FDIA%5Fkalender%5FKoncerter%5Fdiaens%5Fmortenfranklarsen%5Ftif%5F331510ad02b81dd2.jpg
   12 RxProtocol   c HTTP/1.1
   12 RxHeader     c Host: www.kb.dk
   12 RxHeader     c User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:5.0) Gecko/20100101 Firefox/5.0
   12 RxHeader     c Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
   12 RxHeader     c Accept-Language: da,en-us;q=0.7,en;q=0.3
   12 RxHeader     c Accept-Encoding: gzip, deflate
   12 RxHeader     c Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7
   12 RxHeader     c Connection: keep-alive
   12 RxHeader     c Cookie: __switchTo5x=46; __unam=627126b-xxxxxxxxxx-90edcb2-3; __utma=52637486.497784643.xxxxxxxxxx310565486.1310565486.1; __utmb=52xxxx86; __utmc=52xxx86; __utmz=52637486.xxxxxxxxxxx.1.1.utmccn=(direct)|utmcsr=(direct)|utmcmd=(none)
   12 RxHeader     c If-Modified-Since: Wed, 13 Jul 2011 11:49:07 GMT
   12 RxHeader     c Cache-Control: max-age=0
   12 VCL_call     c recv lookup
   12 VCL_call     c hash
   12 Hash         c /erez4/cache/online%5Fmaster%5Farkiv%5Fwebbilleder%5FDIA%5Fkalender%5FKoncerter%5Fdiaens%5Fmortenfranklarsen%5Ftif%5F331510ad02b81dd2.jpg
   12 Hash         c www.kb.dk
   12 VCL_return   c hash
   12 Hit          c 1415348861
   12 VCL_call     c hit deliver
   12 VCL_call     c deliver deliver

Don't worry about the cookies, they are all stripped before going to lookup. As you can see from the lines with "Hit", the two requests for the same object hits two different objects in the cache.

I suspect I need to normalize some of the request headers, but which of them and how?


--
Lars
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.varnish-cache.org/lists/pipermail/varnish-misc/attachments/20110714/9c0135ea/attachment-0003.html>