Re: [NSE] http-vhosts fixes

[Patrik Karlsson <patrik () cqure net>]

On Sun, Jul 29, 2012 at 11:38 AM, Patrik Karlsson <patrik () cqure net> wrote:

> On Sun, Jul 29, 2012 at 5:44 AM, David Fifield <david () bamsoftware com>wrote:
>
> > On Sat, Jul 28, 2012 at 09:19:07PM +0200, Vlatko Kosturjak wrote:
> > > Hello!
> > >
> > > Talking with David, it turned out that http-vhosts can take significant
> > time to finish. Therefore, I have
> > > reduced default list. Also, I have implemented datafiles support to it,
> > so you don't have to edit script
> > > in order to change list to try.
> > >
> > > Of course, list files should be in nselib/data/ directory (in
> > attachment).
> > > Feel free to comment or commit,
> >
> > Committed in r29399, thanks.
> >
> > It seems part of this script's problem is a long timeout, and perhaps
> > sendign packets too fast so that replies aren't received? I see a lot of
> > these delays when running with --script-trace:
> >
> > NSOCK (27.0480s) Read request from IOD #23 [86.59.30.36:80] (timeout:
> > 8000ms) EID 730
> >
> > David Fifield
> > _______________________________________________
> > Sent through the nmap-dev mailing list
> > http://cgi.insecure.org/mailman/listinfo/nmap-dev
> > Archived at http://seclists.org/nmap-dev/
>
> I'm seeing considerable speed improvements with the following patch.
> At first, running multiple threads did not provide any benefit.
> It turned out that I incorrectly assumed that disabling the cache would
> improve performance.
> What happens is that the head function still goes through parts of the
> caching code.
> To the best of my understanding, given how the mutex and states work,
> requests still run sequentially when going through that piece of code,
> making it impossible to benefit from multiple threads. This should have a
> negative impact on other scripts making use of the http library as well.
> Could someone else have a look at this and confirm or deny my suspicions?
>
> Anyway, I changed so that the script uses the generic_request which
> totally skips the cache functions.
> If anyone doesn't find a reason not to commit this, I would like to do so.
>
> Thanks,
> Patrik
>
>
> --
> Patrik Karlsson
> http://www.cqure.net
> http://twitter.com/nevdull77
I committed this change as r29452.

//Patrik
-- 
Patrik Karlsson
http://www.cqure.net
http://twitter.com/nevdull77
_______________________________________________
Sent through the nmap-dev mailing list
http://cgi.insecure.org/mailman/listinfo/nmap-dev
Archived at http://seclists.org/nmap-dev/