From owner-freebsd-stable@FreeBSD.ORG Tue Jun 24 20:14:30 2014 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1 with cipher ADH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 2E70BF03 for ; Tue, 24 Jun 2014 20:14:30 +0000 (UTC) Received: from smarthost1.sentex.ca (smarthost1.sentex.ca [IPv6:2607:f3e0:0:1::12]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "smarthost.sentex.ca", Issuer "smarthost.sentex.ca" (not verified)) by mx1.freebsd.org (Postfix) with ESMTPS id EB618287E for ; Tue, 24 Jun 2014 20:14:29 +0000 (UTC) Received: from [IPv6:2607:f3e0:0:4:f025:8813:7603:7e4a] (saphire3.sentex.ca [IPv6:2607:f3e0:0:4:f025:8813:7603:7e4a]) by smarthost1.sentex.ca (8.14.9/8.14.9) with ESMTP id s5OKEOS5091056; Tue, 24 Jun 2014 16:14:24 -0400 (EDT) (envelope-from mike@sentex.net) Message-ID: <53A9DC0D.2000705@sentex.net> Date: Tue, 24 Jun 2014 16:14:05 -0400 From: Mike Tancsa Organization: Sentex Communications User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:24.0) Gecko/20100101 Thunderbird/24.6.0 MIME-Version: 1.0 To: Rick Macklem Subject: Re: nfs slowdown (RELENG_8) References: <2091718059.2982836.1403563319620.JavaMail.root@uoguelph.ca> <53A978D1.2000207@sentex.net> <53A987AE.8060708@sentex.net> In-Reply-To: <53A987AE.8060708@sentex.net> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit X-Scanned-By: MIMEDefang 2.74 Cc: FreeBSD-STABLE Mailing List X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.18 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 24 Jun 2014 20:14:30 -0000 On 6/24/2014 10:14 AM, Mike Tancsa wrote: > > I also just tried with an igb nic and the same issue. These are i386 > boxes. I wonder if its something to do with that. (non PAE kernel) > OK, its not related to i386 as I found another box with em nics, i386 that works at full speed. One more strange issue I discovered is that there is a difference between dd and cat on the problem boxes. Normally, this would be uninteresting to me other than the fact that it seems the same underlying issue is impacting the performance of dump over NFS. e.g. on the two problem client problem boxes, if I do a simple test like # dd if=/dev/zero of=/nfsbackup/test count=200000 bs=1024 200000+0 records in 200000+0 records out 204800000 bytes transferred in 1.767594 secs (115863704 bytes/sec) which is close to gigabit speeds-- all is good. and if I do # cat /dev/zero | dd of=/nfsbackup/test count=200000 bs=1024 200000+0 records in 200000+0 records out 204800000 bytes transferred in 1.771045 secs (115637945 bytes/sec) Also, close to gigabit speeds... All is still good. But if I do # cat /dev/zero > /nfsbackup/test Its down to 25Mb/s Now, if this were consistent across all my boxes, I would not be too interested. Why is cat with a redirect so much slower, and why slower just on some boxes and not others ?!?! ---Mike -- ------------------- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, mike@sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/