It occurs to me that it could be a mechanism to
protect from a DOS attack.  Consider the following
attack vector:  You encounter a server which requires
authentication for a resource.  You then flood the
server with POST requests with very large payloads,
requiring the server to receive the entire request
before formulating the 401 response.  With a large
enough flood, you can overwhelm the server and cause
denial of service.

    I guess a way to overcome this in the client side
would be to send a HEAD request prior to establish if
the resource is available for consumption.  If not,
the server will respond with 401 and your client can
then send the appropriate authentication credentials.

    Also, if the server is responding prematurely,
doesn't it mean that the request connection was
aborted?  And if this is the case, shouldn't the
HttpCli component detect this and stop sending?  This
still won't prevent any data currently in transit
from generatinga 402 error response when it arrives
at the server.


>------- Original Message -------
>From    : Maurizio
Lotauro[mailto:[EMAIL PROTECTED]
>Sent    : 9/17/2008 1:10:47 PM
>To      : twsocket@elists.org
>Cc      : 
>Subject : RE: Re: [twsocket] Early web server response

Yes. Using "Follow TCP Stream" of WireShark I see the
answer in the middle of
the request. Then I checked the single packet
(ordered by time) and it is
effectively so.
This happen by a customer that use IIS, then I
reproduced it on our server (that
use Apache).
Maybe a Tomcat issue?

Bye, Maurizio.

To unsubscribe or change your settings for TWSocket mailing list
please goto http://lists.elists.org/cgi-bin/mailman/listinfo/twsocket
Visit our website at http://www.overbyte.be

Reply via email to