Date: Wed, 2 Jun 2004 19:28:38 +0300 (EEST) From: Dmitry Pryanishnikov <dmitry@atlantis.dp.ua> To: Barney Wolff <barney@databus.com> Cc: Andre Oppermann <andre@freebsd.org> Subject: Re: net.inet.ip.portrange.randomized=1 hurts Message-ID: <20040602185254.C39863@atlantis.atlantis.dp.ua> In-Reply-To: <20040602151057.GA39564@pit.databus.com> References: <20040602093940.N99493@atlantis.atlantis.dp.ua> <40BDAEEF.2AECC3F0@freebsd.org> <20040602151057.GA39564@pit.databus.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello! On Wed, 2 Jun 2004, Barney Wolff wrote: > Is the problem that the two systems have different ideas of MSL? I haven't changed default net.inet.tcp.msl: 30000 on server. Note that on client side, connection never goes to TIME WAIT, because during active FTP server side closes data connection (see RFC793, figure 6), so it goes down-left on diagram from ESTAB state through TIME WAIT state; but client side receives this first FIN and goes down-right on diagram through CLOSE WAIT and LAST-ACK states. So MSL on client side is actually N/A in this case. We have asymmetry here: client has already forgot about previous usage of data port, but server must remember about it during 2*MSL according to RFC793. Sincerely, Dmitry -- Atlantis ISP, System Administrator e-mail: dmitry@atlantis.dp.ua nic-hdl: LYNX-RIPE
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040602185254.C39863>