From owner-freebsd-hackers Wed Aug 12 15:59:30 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id PAA29349 for freebsd-hackers-outgoing; Wed, 12 Aug 1998 15:59:30 -0700 (PDT) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from alpha.xerox.com (omega.Xerox.COM [13.1.64.95]) by hub.freebsd.org (8.8.8/8.8.8) with SMTP id PAA29319 for ; Wed, 12 Aug 1998 15:59:22 -0700 (PDT) (envelope-from fenner@parc.xerox.com) Received: from crevenia.parc.xerox.com ([13.2.116.11]) by alpha.xerox.com with SMTP id <40696(1)>; Wed, 12 Aug 1998 15:58:54 PDT Received: from localhost by crevenia.parc.xerox.com with SMTP id <177515>; Wed, 12 Aug 1998 15:58:40 -0700 To: Didier Derny cc: support@yard.de, hackers@FreeBSD.ORG Subject: Re: network problem with FreeBSD/Yard (yardnetd) In-reply-to: Your message of "Wed, 12 Aug 98 02:07:23 PDT." Date: Wed, 12 Aug 1998 15:58:31 PDT From: Bill Fenner Message-Id: <98Aug12.155840pdt.177515@crevenia.parc.xerox.com> Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG In message you write: >1/ I suppose that a setsockopt(socket... TCP_NODELAY..) > would have the same result for the yard application if it was > inserted in the yard code ? It depends on exactly what the problem is. There's a common problem which is caused by sending several small packets in a row; the Nagle algorithm (and sender silly window syndrome avoidance) say that you may not send one small packet when another small packet is outstanding. If the other end delays its ack, then your small packets stay outstanding for a long time. Turning off Nagle and sender SWS avoidance (by setting TCP_NODELAY) prevents this from happening. >2/ what are the implication of setting ...ack_delayed=0 for the system ? Traffic on interactive TCP sessions (e.g. telnet, rlogin, ssh) will approximately double, since the client will send a keystroke that you typed and the server will ACK it, and then the server will send another packet with the echoed keystroke. With delayed acks, the ACK piggybacks on the echoed keystroke. >3/ What should I say to Yard to have this problem solved ? Ask them to set TCP_NODELAY? Bill To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message