Burton Strauss wrote:

>OK, but that test really doesn't test anything.  If stunnel does it's thing, it's 
>transparent.
>
>The flaw could be in one of TWO places...
>
>1) ntop's implementation may not be able to handle the request as it's coming from 
>those two browsers - for example, it could be a deadlock (esp as it seems to be in 
>the second request - maybe the browser is trying to open a 2nd connection before the 
>1st is completed...)
>
>2) The browser may be sending an invalid request for https: but not http: (i.e. a 
>different path internal to the browser code)  (Since your test is http:, it wouldn't 
>test this).
>  
>
Yes, it would! I'm connecting my browser to 
"https://<IP>:<STUNNELPORT>". Stunnel is listening on that port, strips 
off the SSL shit and forwards the data to the port on which ntop is 
listening. Ergo, my browser *is* using https, and therefore using a 
possible different internal path.

Maybe I didn't explain my setup properly. It's *not* stunnel that sets 
up a SSL connection with the server (i.e. to ntop's SSL), the client 
sets up a SSL connection to stunnel (i.e. https: in the browser), which 
sets up a "normal" (i.e. http:) connection to ntop.

>-----Burton
>  
>
-- 
Mike.



_______________________________________________
Ntop mailing list
[EMAIL PROTECTED]
http://listgateway.unipi.it/mailman/listinfo/ntop

Reply via email to