On Wed, 16 Nov 2011 14:26:17 +0200, Ghassan Gharabli wrote:
> Hello again,
>
> Sorry I replied back quickly before without noticing your rule if it
> has "/" or not and at first I didnt need to ignore "/?" because I am
> caching several websites like name.flv/?.* so now I am using :
>
> acl ExceptExt urlpath_regex -i (mp(3|4)|flv)/(\?.*)
> acl facebook dstdomain .facebook.com
> acl facebookPages urlpath_regex -i \.([jm]?htm[l]?|php)(\?.*|$)
> acl facebookPages urlpath_regex -i /(\?.*|$)
> cache deny facebook facebookPages !ExceptExt
>
> Actually , I started to see Facebook.com in cache since they changed
> to https://www.facebook.com so till now all servers that have the
> same
> settings are no longer caching facebook main page header except one
> server .. maybe one of the clients is infected with a malicious!
>
> It is only being cached when one of clients are opening facebook
> because I alredy opened facebook and it is not caching on this server
> !.
Hmm. Strange. Caching is not up to the client. All they can do is force
non-cached results to be returned. The ability to cache is granted by
the server (or not).
I'm a little suspicious it might be cached at the browser and Squid in
the middle just relaying "no change" responses. But that is a guess
without having looked at any of the transfer.
>
>> As you wish. I added that line because I noticed the front page for
>> FB you
>> wanted to non-cache has the URL path starting with the two
>> characters "/?"
>> instead of .html or .php.
>>
>
> How can I debug or trace the URL path that starts with "/?" and how
> did you notice the front page for FB including two characters "/?" ?
Experience with the HTTP URL syntax specification confirmed with some
squidclient lookups on the FB front page.
I've been bit by that kind of thing in regex patterns a few times
myself and now make a point of checking for that kind of tiny detail.
Amos
Received on Wed Nov 16 2011 - 22:42:48 MST
This archive was generated by hypermail 2.2.0 : Thu Nov 17 2011 - 12:00:02 MST