DragonFly kernel List (threaded) for 2008-11
DragonFly BSD
DragonFly kernel List (threaded) for 2008-11
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: [Fwd: Re: [Ticket#2008110410001492] www.shiningsilence.com reported operating system]

From: Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>
Date: Wed, 5 Nov 2008 14:12:42 -0800 (PST)

:Here's a question: is there a clear way to tell the difference between a
:FreeBSD system and a DragonFly system based on the TCP packets?  (see
:forwarded mail below for why I'm asking.)
:---------------------------- Original Message ----------------------------
:Subject: Re: [Ticket#2008110410001492] www.shiningsilence.com reported   
:operating system
:From:    "Netcraft Webmaster" <webmaster@netcraft.com>
:Date:    Wed, November 5, 2008 7:53 am
:To:      justin@shiningsilence.com
:We detect the operating system based on characteristics of the TCP packets
:received from the server when making HTTP connections. Probably the TCP/IP
:stack of DragonFly BSD is little changed from FreeBSD; it looks just like
:FreeBSD to us. If you're able to suggest any differences in the TCP stack,
:we could take a look at adding detection.
:The only thing that might be different by default is that DragonFly BSD
:apparently has random IP IDs by default (he says from just a brief skim of
:some release notes); but since that's available as an option for FreeBSD
:too, it doesn't seem like a good way to tell them apart.
: Colin Phipps
: Netcraft

    There are almost certainly differences in how we handle SACK, since
    we have a completely independant implementation of SACK then FreeBSD
    does (Jeff did ours from scratch).

    We also aggregate TCP ACKs under heavy networking loads, but that may
    not be testable over a WAN.  It really only shows up on GiGE or faster.

    I think their best bet is to characterize our SACK implementation.

					Matthew Dillon 

[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]