Re: [C-icap-users] ICAP Error (403) response to IE

[<an...@un...>]

Hi Christos,

> Hi,
>
> Andy Millett wrote:
>> Hi Christos,
>>
>> Thanks for the response.
>>
>> I'd had the same thought on the 403 response myself after looking at the
>> RFC. Being that it works in Firefox and Safari I surmised it's IE a
>> fault in it's strict interpretation of the response. Still, I've changed
>> srv_clamav.c as suggested and that has worked. IE now displays as
>> Firefox and Safari. Correctly.
>
> OK thank you for reporting it. I should change the 403 response in
> c-icap cvs sources ....
>

Thanks, whats the CVS source link?

>>
>> RE: your question about integration with Blue Coat's ProxySG: It works,
>> quite well in fact. The one thing that Blue Coat's ProxyAV has over
>> c-icap is that it assigns a token to all scanned objects so that object
>> doesn't need to be scanned again if it hasn't changed when requested by
>> another user. This lowers the amount of traffic needed as the ProxySG
>> simply asks the ProxyAV if the 'token' it assigned is still valid.
>
> It is a very interesting feature. I did not considering it as important
> because you can apply the ICAP scanning  before the objects cached on
> the http proxy cache.  If an other user requests the same object will
> get it from the cache. In this case the object will not send to the ICAP
> server.
>
> How does the ProxyAV decides to not scan again an object? Does it
> examines the object http headers?

The ProxyAV, to my knowledge, receives an object from the ProxySG (if seen
for the first time) which it scans and assigns a 'token' too. The ProxySG
then cache's that object (if it isn't a virus) along with the token from
the ProxyAV. If another user requests the object which the SG has in cache
the SG asks the AV if the token is still valid (Yes, headers, ttl,
cache-expiry, etc). If it is, the SG serves the object from cache, if it
isn't the object is scanned again and a new token assigned.

This means the object doesn't get scanned every time it is requested if
the SG has it in cache. The SG simply asks if the token is still valid
thus the AV saves cpu cycles because it doesn't have to scan the object
again unless it's changed.

>
> Regards,
>     Christos
>

Regards
Andy