Date: Thu, 9 Oct 1997 04:32:46 -0200 (EDT) From: Joao Carlos Mendes Luis <jonny@coppe.ufrj.br> To: joerg_wunsch@uriah.heep.sax.de Cc: hackers@FreeBSD.ORG, jonny@coppe.ufrj.br Subject: Re: TCP problem Message-ID: <199710090632.EAA02790@zeus.coppe.ufrj.br> In-Reply-To: <19971009081109.NI17492@uriah.heep.sax.de> from J Wunsch at "Oct 9, 97 08:11:09 am"
next in thread | previous in thread | raw e-mail | index | archive | help
#define quoting(J Wunsch) // As Joao Carlos Mendes Luis wrote: // // > // >/usr/sbin/sysctl -w net.inet.tcp.sendspace=65536 // > // >/usr/sbin/sysctl -w net.inet.tcp.recvspace=65536 // // > // This is the cause of the problem. You really don't want to change // > these, anyway. You'll see the best performance at the default of // > 16KB. // // > So, what's the good thing about RFC1323 ? // // Both, RFC 1323 and the socket buffer space size, are not related at all. 1) RFC1323 deals mostly with large windows. 2) The above sysctls control the default window size. 3) The TCP send window data must be kept in memory until receiving an ack for it. Isn't the socket buffer space where this data is kept ? What am I missing here and why the best performance is got with 16KB ? // Btw., read the other threads about TCP problems in -hackers that took // place within the last four or five days. Also read the commit message // for tcp_output.c revision 1.27 to learn what the actual problem has // been (and why it is related to your bumping the send buffer space). I've read, and now that it's been merged into RELENG_2_2 I'll recompile my system ASAP. Thanks. Jonny -- Joao Carlos Mendes Luis jonny@gta.ufrj.br +55 21 290-4698 jonny@coppe.ufrj.br Universidade Federal do Rio de Janeiro UFRJ/COPPE/CISI PGP fingerprint: 29 C0 50 B9 B6 3E 58 F2 83 5F E3 26 BF 0F EA 67
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199710090632.EAA02790>