From owner-freebsd-qa Wed Jan 9 8:17:12 2002 Delivered-To: freebsd-qa@freebsd.org Received: from rwcrmhc53.attbi.com (rwcrmhc53.attbi.com [204.127.198.39]) by hub.freebsd.org (Postfix) with ESMTP id 6C36737B400; Wed, 9 Jan 2002 08:17:07 -0800 (PST) Received: from bmah.dyndns.org ([12.233.149.189]) by rwcrmhc53.attbi.com (InterMail vM.4.01.03.27 201-229-121-127-20010626) with ESMTP id <20020109161706.BFEX24940.rwcrmhc53.attbi.com@bmah.dyndns.org>; Wed, 9 Jan 2002 16:17:06 +0000 Received: (from bmah@localhost) by bmah.dyndns.org (8.11.6/8.11.6) id g09GH5s90370; Wed, 9 Jan 2002 08:17:05 -0800 (PST) (envelope-from bmah) Message-Id: <200201091617.g09GH5s90370@bmah.dyndns.org> X-Mailer: exmh version 2.5 07/13/2001 with nmh-1.0.4 To: Peter Pentchev Cc: Robert Watson , qa@FreeBSD.ORG Subject: Re: Reduced reliability due to larger socket queue defaults for TCP In-reply-to: <20020109104509.C269@straylight.oblivion.bg> References: <20020109104509.C269@straylight.oblivion.bg> Comments: In-reply-to Peter Pentchev message dated "Wed, 09 Jan 2002 10:45:10 +0200." From: bmah@acm.org (Bruce A. Mah) Reply-To: bmah@acm.org X-Face: g~c`.{#4q0"(V*b#g[i~rXgm*w;:nMfz%_RZLma)UgGN&=j`5vXoU^@n5v4:OO)c["!w)nD/!!~e4Sj7LiT'6*wZ83454H""lb{CC%T37O!!'S$S&D}sem7I[A 2V%N&+ X-Image-Url: http://www.employees.org/~bmah/Images/bmah-cisco-small.gif X-Url: http://www.employees.org/~bmah/ Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Wed, 09 Jan 2002 08:17:05 -0800 Sender: owner-freebsd-qa@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG If memory serves me right, Peter Pentchev wrote: > On Tue, Jan 08, 2002 at 07:33:25PM -0500, Robert Watson wrote: > > The temptation here is one of two things: > > > > (1) Back out the change increasing the send-q socket buffer size as a > > default, and restore tuning(7) to recommend increasing the value, or > > > > (2) To add the following text to the release notes: > > > > In 4.5-RELEASE, default socket buffer sizes are increased to > > maximize performance on high speed networks. However, under > > some circumstances, this can dramatically increase the memory > > requirements of the network system, requiring a manual > > bumping of the kernel NMBCLUSTERS setting. This can be > > set using kern.ipc.nmbclusters. Two changes I'd make: "default socket buffer sizes" should really be "default TCP socket buffer sizes". Also, we should say that "some circumstances" includes (as an example) the case of machines serving many TCP connections. > I personally feel better about (2).. The buffer size increase really > does help with long-haul or high-latency connections. Normal users > would rarely run thousands of TCP connections; those that do would > either read the release notes, or be kindly pointed to those. People > who really do run thousands of TCP connections will hopefully read the > release notes and know what to do :) Yes, we can put this in the release notes, and there's a good chance people will read it, but we also need to document it somewhere less transitory. A new user who starts using FreeBSD 4.6-RELEASE can't be expected to go back to 4.5-RELEASE's documentation to find this little gem. > As a side note, maybe this should be mentioned in the FAQ, too. My first thought would have been tuning(7), as this question/problem isn't frequently asked (yet). But I guess the FAQ would work too. Bruce. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-qa" in the body of the message