On Fri, Jan 28, 2011 at 1:15 PM, tsuraan <[email protected]> wrote:
> Well, that's certainly timely :)  I'm not sure if the API is still up
> for discussion, but the ability to do "setTimeout 0" to disable
> timeouts entirely would be nice (most socket APIs allow this).  It's
> not a big problem to set it to some absurdly high number, but allowing
> for 0 is somewhat standard.  And maybe that does work and I missed it,
> but I didn't see any special handling of 0 in there.

Since it's so new, I'm sure he'd be willing to give adequate
consideration to reasonable requests.

My understanding is that the issue with zero timeouts has to do with
resistance to DOS attacks, specifically things like slowloris
(http://ha.ckers.org/slowloris/).  If you don't time out then I think
you're automatically vulnerable to this attack.  There may be an
argument for allowing it to be disabled though.  In our tests, the
timeout code imposes a pretty substantial performance penalty, so if
we supported disabling it, we'd also probably want to do it in such a
way that the timeout code wouldn't even be executed.  This would be a
logical thing to do to get maximum performance from servers of web
services behind firewalls that are not exposed to malicious clients.
_______________________________________________
Snap mailing list
[email protected]
http://mailman-mail5.webfaction.com/listinfo/snap

Reply via email to