RE: www.microsoft.com looping

For all and sundry who are interested in www.microsoft.com and the
looping request problem, an ASP file is an "Active Server Page" (TM
<g>) that Microsoft IIS (Internet Information Server) uses.

ASP is more than just server side includes (SSI), but not quite CGI,
tending to be a semi-scripted page (HTML + ASP commands). It's a bit
like Visual Basic (surprise !!! all MS products seem to have gone VB
mad), but can do some more powerful, not-trivial things like hook into
ODBC databases directly to do queries. As such it should not be cached
(ala cgi-bin and ?) requests in the cache_stoplist option in squid.conf.
 

In fact, IIS/ASP has a "browser capabilities" file, so that the
resultant HTML is supposed to be compliant with the actual capabilities
of the browser. e.g. Tables support, frames, etc. It's a concept to
reduce dependence on browser-specific features by providing ASP as an
intermediate "generic browser" interface. I guess you'd call it "Write
once, view anywhere" (ala Java "Write Once Run Anywhere (TM)). ASP
sorts out what the browser can and can't do, and then modifies the
output accordingly.

The ASP script can also present different information for:

Authenticated users (Anonymous vs Basic vs NTLM Challenge Response)
People in different IP networks/subnets
User-agent
Cookies
Basically information from any HTTP header or Connection information

So, in a nutshell, I believe any ASP files should be treated as 100%
non-cacheable, all the time, PERIOD !!!

I think Duane should modify the default configuration squid.conf.in to
include ".asp" as one of the cache_stoplist (be careful about the
period so sites with asp in their name aren't affected). Maybe this
could be a bit more refined, so that URLs such as http://www.asp.com/ or
http://www.activeserver.com/pages.asp/temp.html aren't evicted, but
let's face it, these are by far in the minority of ".asp" URLs

Regards

Jason Armistead
armistej@oeca.otis.com

>----------
>From: Laurent FACQ[SMTP:facq@U-Bordeaux.FR]
>Sent: Friday, 31 October 1997 3:16
>To: squid-users@nlanr.net
>Subject: www.microsoft.com looping
>
>
>like other cache administrator, i just found the looping
>probleme on www.microsoft.com page
>
>config:
>squid 1.1.14, with special microsoft rule
>refresh_pattern/i \.microsoft\.com 1440 40% 20160
>
>
>imagine the following resquest in this order :
>
>Browser Requested URL Effect
>-----------------------------------------------------------------
>IE4 www.microsoft.com DIRECT / CACHE redirect to ie40.asp
>nonIE4 www.microsoft.com HIT / Get redirected to ie40.asp
>nonIE4 ie40.asp DIRECT / CACHE redirect to default.asp
>IE4 ie40.asp HIT / Get redirected to default.asp
>IE4 default.asp DIRECT / CACHE redirect to ie40.asp
>
>
> so you got the HIT loop :
>
> www.microsoft.com -> ie40.asp -> default.asp -> ie40.asp .....
>
> -> i've put '.asp' on hierarchy and cache stoplist...
> it seems to works fine, but the question is : is this a good
> idea ?
>
> .asp page seems to be dynamique page like cgi ?
>
> LF.
>
>
Received on Thu Oct 30 1997 - 17:48:54 MST