Re: Crazy congestion control behavior after r33195

[David Fifield <david () bamsoftware com>]

On Wed, Jul 09, 2014 at 12:41:07PM +0200, Jacek Wielemborek wrote:
> The scan finally finished. You can see the complete log there:
>
> https://svn.nmap.org/!svn/bc/33203/nmap-exp/d33tah/uploads/r33198-complete-nmap_-p-_--unprivileged_-sT_-n_-Pn_-d4_scanme.nmap.org_8.8.8.8_stderr_to_stdout.log.lzma
>
> Short URL: https://tinyurl.com/lpyfxwk
>
> I attach a plot generated by my parse-timing.py script from
> nmap-portscan-tests. Also, groupstats show that peak active probes count
> was 133.

The shape of the cwnd curves kind of looks like response rate scaling
isn't working properly. It looks like a square root rather than a line.
It could happen if num_replies_expected isn't increased properly. Or it
could be because cc_scale reached cc_scale_max.

David
_______________________________________________
Sent through the dev mailing list
http://nmap.org/mailman/listinfo/dev
Archived at http://seclists.org/nmap-dev/