Re: [SQU] missing-http-ident ?

James Fung wrote:
>
> Hello all,
>
> Has anyone else seen this problem, and perhaps more importantly,
> does anyone know of a solution? Going to http://www.metacrawler.com (a
> popular search engine) and doing a search for "html java" will produce
> the following error :
>
> ---------------------------------------------------------------------------
> The requested URL could not be retrieved
>
> While trying to process the request:
>
> GET http://ad.doubleclick.net/adj/N1713.go2net/B23532.9;abr=!ie;sz=468x60;[
> keyword];ord=35687 HTTP/1.0
> Proxy-Connection: Keep-Alive
> User-Agent: Mozilla/4.75 [en] (X11; U; Linux 2.2.16 i686; Nav)

[...]
> At the same time this error message appears, a "missing-http-ident"
> appears in the Squid logs. It looks like there's a carriage return in the
> URL which is causing Squid grief. Is there a way to tell Squid to ignore
> the CR?

AFAIK Squid ignores CR (just another form of whitespace), but it does
not ignore NL.

NL is part of the HTTP protocol and cannot be ignored.

What can be done is to work around the issue by changing Squid to
support HTTP/0.9. However, the HTTP/0.9 support is hidden in the sources
and slightly broken for the moment...

This error is quite often seen with JavaScrip driven ad-banners and
similar things...

--
Henrik Nordstrom
Squid hacker
--
To unsubscribe, see http://www.squid-cache.org/mailing-lists.html