Alvaro Lopez Ortega wrote:
> On 06-jul-09, at 17:08, Michiel van Es wrote:
> 
>>> However, whenever an old browser (without SNI support) accesses your
>>> server (let's say an IE 6) the SSL handshake will be perform using the
>>> default certificate. The problem is basically the timing: the first a
>>> SSL connection does is the handshake between server and client
>>> (sending/receiving the certs), and only when the secure connection is
>>> stabilised, the browser sends the HTTP request. The main problem is that
>>> the server does not know what vserver the client wants to access until
>>> it doesn't receive that HTTP request.
>>>
>>>> I mean: would you find it acceptable if you connect to a server but got
>>>> the wrong SSL certificate (a certificate of another server). What is 
>>>> the
>>>> use of certificates if the name not match? And how would you tell the
>>>> difference with a man in the middle attack?
>>>
>>> It is an issue, indeed.
>>
>> Are you considering to implement the *old* setup? Binding certificates
>> to an ip-adress?
> 
> 
> I did. When I wrote the target_ip plug-in I thought that it'd work.. 
> however, I missed a little detail that rendered it useless for this sort 
> of scenario.
> 
> I'd agree on implementing the old method as long as it doesn't mess the 
> code. I haven't found to way so far.. so I couldn't tell you for sure.
> 
> Antonio worked on the cryptor-libssl plug-in for a while, and he is 
> willing to check it out. Let's hope he comes up with some brilliant 
> solution! :-)

That sounds amazing and would be a real help for my situation! :-)

> 
> -- 
> Greetings, alo
> http://www.alobbs.com/

Kind regards,

Michiel

> 
_______________________________________________
Cherokee mailing list
[email protected]
http://lists.octality.com/listinfo/cherokee

Reply via email to